The following reviews the origins and characterization of the HMEC system developed in my lab and those of co-workers. This information will be periodically updated. It includes more than you'll probably ever want to know, but hopefully, someone will find each tidbit valuable and consequently not need to query me on that subject. It also includes my personal opinions about HMEC biology and cell culture usage (section IV.F.3). I welcome feedback on how this information can be presented most usefully. Some of this information was presented in my four Newsletters from 1987-1989.

To put this information in context, my long-term goal (since 1976!) has been to develop an HMEC system that could be used to study the normal mechanisms controlling proliferation and differentiation in human cells, and to understand how these normal processes become altered as a result of immortal and malignant transformation. Guiding this work was the desire to facilitate widespread use of human epithelial cells for molecular and cellular biology studies, i.e., the hope was that HMEC would seem a reasonable alternative to fibroblasts, or tumor cell lines, or non-human cells. Therefore, I tried to develop a system that is relatively easy to use, can provide large quantities of uniform cell populations, and is relatively well-defined. I realize that "relatively easy" can be in the eyes of the beholder, and for some people HMEC will still seem difficult compared to HeLa or 3T3. While HMEC may require a little more effort, really, they are very easy to grow once you get the hang of it. What is needed is careful attention to proper tissue culture procedures, a basic understanding of the cell system, and a "feel for the organism". Cells are living creatures, with some resemblance to children - they can behave as if they have minds of their own. As Dick Ham has often said, sometimes what is most important is just to "listen to your cells". The reward is being able to use cells much closer to relevant human processes. Normal finite lifespan HMEC allow you to study growth control in cells with normal human growth control mechanisms.

In developing and promoting the use of the HMEC system, I have been influenced by the following assumptions:

(1) Prior knowledge of what constitutes normal cell behavior is necessary to determine what constitutes abnormal and deranged processes, e.g., if you want to say that something you are studying is a property of a transformed cell, you need also to look at the normal cells.

(2) Understanding normal and aberrant human epithelial cell growth control and differentiation will ultimately require examination of human epithelial cells. Non-human and non-epithelial cell studies may provide valuable information and suggest areas of research. However, the many differences which are known to exist between these cell types in culture as well as in whole body physiology indicate that only examination of the cells in question will give an accurate description of those cells' behavior. I believe this is especially true in the area of carcinogenesis, where, e.g., the major difference in control of telomerase expression between human and rodent cells results in significant differences in the transformation process (see section IV.).

(3) In a situation where whole animal experiments are not possible (i.e., with human cells), the next best option is to develop culture systems that can as accurately as possible approximate the in vivo state. I have tried to balance the goal of making the system as amenable as possible to widespread use, with the goal of trying to optimize the system to reflect in vivo biology. The result is considerably less than ideal in terms of in vivo approximation. Normal and aberrant cellular processes in vivo involve complex interactions of polarized cells within three dimensional organ systems. Single cell types growing on plastic are not that! Consequently, it's important to remember the limitations of this culture system. I believe it is very important that studies concerned with the development of culture systems that more accurately mimic in vivo cell-cell and cell-matrix interactions be well supported.

Since fostering widespread usage of HMEC has been one of my long-term goals, I have tried to make cells available to other interested investigators. I've found it helpful to talk to people individually to understand more precisely their scientific needs and goals in using HMEC. Checking through the relevant parts of this web site can provide a sense of what is available and known. While distributing cells is part of what I enjoy doing, please keep in mind that this is a non-commercial, non-official, personally-run cell bank, and I and my technician are also doing many other things. I appreciate it when shipment of cells is made as easy as possible for me.

Nomenclature notes:

"Primary" refers to cells the first time they are placed in culture (e.g., outgrowths from organoids). Cells which have been subcultured are no longer primaries and should not be described as primary culture (this is a very common error). I refer to higher passage cultures of normal finite lifespan HMEC as strains. In technical tissue culture parlance, they could be called cell lines once subcultured, but I find this usage confusing and only use "cell line" to refer to cells with indefinite growth potential (i.e., immortal). I use "Extended life" to refer to cells that grow longer than normal as a result of some abnormal in vitro exposure; e.g., chemical carcinogens or oncogenes. "Extended life" should not be used to refer to the post-selection, p16(-) HMEC strains with long-term growth since this long-term growth occurs spontaneously.

I. Tissue Derivation of HMEC Cultures

(references: Stampfer et al. 1980; Stampfer 1985; Stampfer & Yaswen, 2000)

I.A. Tissue Procurement

Index

We have obtained our human mammary cells from a variety of sources, mostly surgical discard material. What we refer to as normal HMEC is derived from reduction mammoplasty tissues. Women undergoing reduction mammoplasty operations do not have any known epithelial pathology per se (their breasts contain the same amount of epithelial cells as is present in smaller breasts, but they have much more adipose tissue). Their breast tissues do show the range of pathologies generally found in women of the same age (e.g., it may be described as containing mild to atypical hyperplasia, or fibrocystic disease). There is always the possibility that women with such large fat deposits in their breasts could have some abnormality in some aspect of their metabolism. Because large portions of the breast are removed, with minimal need for pathology evaluation, considerable quantities of cells from the same individual are made available from each reduction mammoplasty.

The other major source of tissues comes from mastectomies. Usually the amount of tumor tissue available for culture is small, due to the need for clinical evaluation of the tumor. Larger amounts of the non-tumor peripheral tissue are available. This can be particularly useful in providing matched pairs of tumor and non-tumor tissue from the same person. However, I do not consider peripheral mastectomy tissue as normal, as there is always the possibility of tumor field effects, microtumors within this tissue, field effects from some environmental conditions predisposing to tumors, and inherent genetic abnormalities. For some of the same reasons, I would not view as normal the tissues we have obtained from contralateral mastectomy - tissues removed from the breast contralateral to a tumor-bearing breast for prophylactic or cosmetic purposes.

Additional surgical tissues are obtained from benign conditions: fibroadenomas (which are not thought to be pre-malignant); fibrocystic tissues (which under some conditions could indicate an increased likelihood of tumor development); gynecomastias, which are benign hyperplasias in male breast tissue.

We also have a few samples of tissues from other conditions. We have two subcutaneous mastectomy tissues. These operations are generally performed because of extensive fibrocystic disease, and in the two samples I processed, the consistency of the tissue appeared grossly abnormal (hard and fibrous) compared to reduction mammoplasties. We have two non-tumor peripheral tissues from breasts that had sarcomas.

Another, non-surgical source of HMEC is from breast fluids. A small number of cells can be obtained from nipple secretions of around 50% of women, and larger volumes are available from lactational fluids. Our original publication in 1980 actually utilized cells from nipple secretions. Cells from milk are valuable as a source of functionally differentiated cells. We have only used these for specific purposes and do not have supplies to distribute.

I.B. Tissue Processing

(references: Stampfer et al. 1980; Stampfer 1985 - gives procedure details; Stampfer & Yaswen, 2000)

Index

Most of the surgically derived tissues are processed by gross selection of epithelial material followed by digestion for 24-72 hrs at 37°C with collagenase and hyaluronidase. This leaves nearly pure epithelial clumps (termed organoids) which can be separated from the rest of the digested material by collection on filters with pores of fixed size. The organoids can be stored frozen in liquid nitrogen (for at least 25 years - the time since I started this), permitting repeated experiments with cells from the same individual. Material in the filtrate usually contains mainly fibroblastic type cells, and is a good source of matched fibroblasts from the same individual.

The small pieces of tumor tissue are generally not structured like organoids. Digestion for 24 hrs can yield small epithelial clusters and the filtrate may contain many of the single tumor cells. This method is probably not the best available for obtaining tumor cells for culture. It is what was used for the samples that I have stored frozen. Other laboratories have developed tissue processing methods more specific for tumor tissues (see references in Stampfer & Yaswen, 2000)

Table 1 gives an idea of what and how many primary tissues we collected and processed.

Table 1: Bank of Primary HMEC Tissue

Tissue Source # Specimens Age Range Median # Ampoules

Reduction Mammoplasty 49 15-66 30

Mastectomy

carcinoma 57 29-93 5

peripheral non-tumor 43 24-87 8

contralateral 6 42-77 10

Biopsy (benign tumors) 9 13-47 5

Gynecomastia 6 17-57 9

___________________________________________________________

This represents the amount of tissues as originally collected, rather than current inventory levels. We also have the filtrate material for each specimen, from which, in most cases, fibroblast-like cells can be grown. We are reluctant to give out much primary material, since quantities are limited and we are no longer processing these tissues, but small amounts may be available if essential, particularly from the reduction mammoplasties.

II. Growth of Finite Lifespan HMEC in vitro

(references: Stampfer et al. 1980; Stampfer, 1982; Hammond et al. 1984; Stampfer 1985; Taylor-Papadimitriou et al., 1989, Stampfer & Yaswen, 1992; Brenner et al., 1998; Romanov et al., 2001; Stampfer & Yaswen, 2001; Stampfer & Yaswen, 2003)

Index

II.A. Media and Growth Capacity of Cultured HMEC

(references: Stampfer et al. 1980; Stampfer, 1982; Hammond et al. 1984; Stampfer 1985; Brenner et al., 1998; Stampfer & Yaswen, 2003)

When I started working with HMEC in 1977, I first developed the MM medium (see "Procedures" for composition of and growth of cells in MM). This medium has a 1:1 DME:F12 base, plus conditioned media from other cell lines, a variety of growth factors, and 0.5% fresh FCS. MM supports active HMEC growth for 3-5 passages, or ~15-30 population doublings (PD). The cultures initially display a mainly cobblestone morphology, but as the population becomes non-proliferative, larger, flatter senescence-associated b-galactosidase (SA-b-gal) positive cells predominate. Cultures that maintain growth beyond passage 5 display uneven proliferation. Pockets of small, actively growing cells may appear, but these cells soon become larger and less proliferative. I have also employed a number of variations on the MM theme, e.g., with and without a cAMP stimulator (cholera toxin), without the conditioned medium (designated MM4), or without particular growth factors.

While MM provided only a limited amount of cells, this was sufficient to perform many types of experiments. It also provided enough cells to begin more systematic studies on optimizing media for growth of HMEC. This work was done by Susan Hammond in Dick Ham's laboratory, the result of which was the development of the serum-free MCDB 170 medium in 1984. This has a base in which the components have been optimized for HMEC growth, plus a variety of serum-free supplements (see "Procedures" for composition of and growth of cells in MCDB 170). The only undefined element is bovine pituitary extract.

When organoids are placed in MCDB 170, there is initial active cell division for 2-3 passages of cobblestone appearing cells. These cells gradually change morphology, becoming larger, flatter, striated, with irregular edges and reduced proliferative capacity. These cells stain positive for SA-b-gal. Recent data (see next section) suggests that this growth arrest (as well as the growth arrest in MM) most resembles the M1/senescence stage of growth arrest previously described for fibroblast cells. As the larger cells cease growth, a small (i.e., a 60 mm dish seeded with 1.5 x 10⁵ cells may show 1-10 areas of active growth) number of cells with the cobblestone morphology eventually show proliferative capacity and soon dominate the culture. I called this process, whereby only a small fraction of the cells grown in MCDB 170 display long-term growth potential, self-selection, the resulting populations, post-selection, and the growth arrest selection. In retrospect, the MM-grown and MCDB170-grown HMEC prior to selection were initially called pre-selection. We now know that the post-selection cells have downregulated expression of the cyclin dependent kinase inhibitor (CKI) p16 (see below).

Self-selection can also be observed in primary cultures that are subjected to repeated partial trypsinization, a process wherein approximately 50% of the cells are removed and the remaining cells allowed to regrow. After about 10 partial trypsinizations, most of the cells remaining in the dish display the flat, striated, morphology and cease division. However, nearly every organoid patch also gives rise to areas of the growing cobblestone cells, indicating a widespread distribution of the cell type with the potential for long-term growth.

NOTE: if you are trying to take cells through the self-selection process, dishes with the large flat cells may sit there for weeks before the smaller cells become obvious. I suspect that this implies that more is happening than the outgrowth of a pre-existing p16(-) population, but we have never investigated this phenomena in depth.

NOTE: partial trypsinization is a way to obtain more good-growing secondary cultures from primary cultures than if the primaries were fully subcultured. For some reason, the cells in the primaries remain much more vigorous for a longer time period. Perhaps this is due to some heterogeneity in the primary cell population, or some extracellular matrix material. This question has always intrigued me but it has also never been investigated in depth.

Most of the normal HMEC which I make available (as well as the commercially available HMEC from Clonetics) represent these post-selection cells that display long-term growth in MCDB 170. Post-selection cells maintain growth for an additional 7-24 passages (approximately 45-100 PD in total), depending upon the individual reduction mammoplasty specimen. When net growth ceases at a second senescence block (termed agonescence, see section below), they appear flatter and more vacuolated, and stain positive for SA-b-gal, while retaining the cobblestone epithelial morphology. Post-selection HMEC are particularly useful in molecular and biochemical studies since they provide a virtually unlimited supply of uniform batches of finite lifespan human epithelial cells. Thus, experiments can be repeated using cells from both the same frozen batch, as well as from the same individual. Post-selection cells grow rapidly (doubling times of 18-24 hrs) and will grow clonally with 15-50% colony forming efficiency. However, it is important to remember that the cells with long-term growth potential represent a selected, p16(-) subpopulation of the mammary epithelial cells placed in culture (see below).

We have grown a limited number of our frozen primary organoids specimens in MCDB 170, generating large pools of frozen cells for use in our laboratory, as well as for distribution to others. We have thus far grown up cells from 12 reduction mammoplasty tissues, 8 mastectomy tissues (6 tumor tissues, 5 non-tumor, 1 contralateral), and 1 gynecomastia. Figure 1 illustrates the long-term growth potential of cells from each of these individuals.

NOTE: we have not observed a single instance of spontaneous escape from senescence in the HMEC grown under these conditions. We are not aware of any spontaneous escape from senescence in any other lab using these post-selection HMEC. In general, cells from the same individual senesced around the same passage, but there were exceptions.

The following explains how and why we kept track of this information:

Since the post-selection cells in our large freeze-downs may be derived from a small number of p16(-) cells, it was possible that all freeze-down pools were not equivalent - a few cells with some unusual quality could influence a given batch. As a consequence, we gradually (and informally) developed a nomenclature to keep track of the origins of a given cell pool. At the first level, we started using symbols to indicate every time we started a new primary organoid ampoule from the same individual. These were easy-to-write symbols with which to label the dish (e.g., heart ©, infinity ¥, birdie "v", spiral "@", etc.). These are now officially registered in our computer records as FreezeDownSymbol (FDS). Subsequently, we realized that it might be important to also keep track of cell populations coming from different pools of post-selection cells. Each selection pool can be thought of as a different substrate "batch", with the possibility that there might be batch differences. So our FDS may be followed by an indication of "selection" batch (e.g., vIP2, ©D, ¥3, @K, @L, etc.). These are the symbols present in Figure 2.1. Visually, cells from the same individual, regardless of batch, tend to have the same characteristic appearance, while we do notice interindividual morphologic differences (see Figure 2.2).

NOTE: the most common batch of cells from specimen 184 that I used to distribute was @K, which ceases net growth around passage 22, whereas most other batches from specimen 184 senesce around passages 16-18. However, we now know that this batch is heterogeneous (see next section), containing a subpopulation that senesces earlier. Consequently, there is a slowing in the growth of this 184@K batch between passages 13-15. We are now distributing the 184vIP3 batch. We have large frozen stocks at passages 7 and 8, which ceases net growth around passages 14-16.

(Click here to see figure 2.1)

Figure 2.1: Growth capacity of HMEC in MCDB 170 medium. I stopped adding information to this graph many years ago, but this gives the general idea and includes the FDS and selection batch of the cells I most frequently distribute. You can use this to see the expected passage where the cells senesce. Primary cultures obtained from reduction mammoplasties (top two rows) and mastectomies (T= tumor tissue, P= non-tumor tissue from tumor-bearing breast, C= contralateral) were initiated and subcultured with about 8-10 fold amplification per passage. Bottom horizontal lines indicate passage level of initiation of frozen ampoules. Top horizontal lines indicate passage level of no net increase in cell numbers (i.e., agonescence). Internal horizontal lines indicate that cultures were frozen and reinitiated at that passage. Same shading indicate cells derived from the same "selection". Asterisks indicates cells exposed to a cAMP stimulator during selection. For specimen 184 "v", cultures were initiated from the same primary ampoule but taken through selection with three different cAMP stimulators (cholera toxin, isoproterenol, prostaglandin E1). In a few cases, (indicated by a different shading in primary culture), the tumor cultures were grown in MM in primary culture.

[PS: the names of the symbols shown, in order of appearance, are: 161- heart, triangle, newmoon, yinyang, infinity; 184- birdie, spiral (not shown are aleph, cross, lollipop, ecology; flower); 48- silver, orange, pink (not shown are blue, tulip); 172- icecream, lollipop, diamond; 195L- teardrop, pumpkin; 186T- heartbrk, sunrise]

Other information which may be gleaned from these data:

(1) There does not appear to be any loss in viability due to multiple freeze-thaws;

(2) There does not appear to be any correlation between growth potential in culture and age of specimen donor. It is possible that some of the differences seen in growth potential could reflect interindividual differences in optimal growth requirements relative to the nutritional formulation of MCDB 170.

(Click here to see figure 2.2)

Figure 2.2 :Morphology of reduction mammoplasty derived HMEC grown in MCDB 170.

Giemsa stained cultures from

(A) 184 passage 7; (top image)

(B) 172 passage 13; (middle image)

II.B. Growth and Senescence of Cultured HMEC

(references: Brenner et al., 1998; Romanov et al., 2001; Stampfer et al., 2001; Stampfer & Yaswen, 2001; Tlsty et al., 2001; Stampfer & Yaswen, 2003; Garbe et al., in prep)

Index

II.B.1. Growth and Senescence of Pre-selection HMEC

During our studies on the growth and senescence of cultured HMEC in the 80's and 90's we and others assumed that the proliferation barrier encountered by the post-selection HMEC most closely resembled the previously described M1/senescence block for fibroblasts (see references for section II). We did not fit into the M1/M2 scheme the early growth arrest encountered by cells grown in MM, and by the majority of cells grown in MCDB 170 that ceased growth at selection. This assumption was largely based on the ongoing viability of the non-proliferative post-selection cells. On gross inspection of these cultures, vacuolated and multi-nucleated cells were observed, but little cell death was seen. Nothing resembling the large-scale cell death reported for M2/crisis was observed. Cell populations maintained with ongoing feeding, and no subculture, remained viable (metabolically active) for 1-2 years. The cells that ceased growth in MM, and at selection in MCDB 170 appeared, if anything, less viable; after several months, most of the non-proliferating cell population had sloughed off the dish.

Recent studies in Thea Tlsty's laboratory have shed new light on the nature of these two proliferation barriers encountered by the HMEC (see Charts 1 and 2 under "Brief History" for a schematic outline of overall HMEC growth, senescence, and immortalization). Based upon experiments performed in the Tlsty lab directly comparing isogenic human mammary epithelial and fibroblast cells, as well as data from our lab, the following model has emerged.

The 1^st proliferation barrier (which we are now calling stasis), appears to be mediated by stress-induced CKIs inhibiting RB inactivation. HMEC encounter this barrier after ~15-30 PD in MM and ~10-20 PD in MCDB170. We suggest that most of what has been called M1/senescence in fibroblast cultures, as well as what has been referred to as premature senescence due to “culture shock”, represents this RB-mediated barrier. The greater the stress-inducing signals, the fewer PD prior to this growth arrest.

Like the growth-arrested isogenic mammary fibroblasts, this HMEC arrest is associated with high levels of p16 and SA-b-gal expression. In growth-arrested cultures, there is little ongoing DNA synthesis, with fewer than 2% of cells incorporating bromodeoxyuridine during a four-hour pulse. There is little cell death, with only 1% Annexin-V staining. Like the fibroblasts, the HMEC retain normal diploid karyotypes, and the arrested cells display a FACS DNA profile consistent with G0/G1 arrest, with a 2N to 4N ratio of ≥ 4. When directly compared, both cell types arrested with a mean TRF of ~6 kb. Thus we suggest that what we called the selection block in MCDB170-grown HMEC is most equivalent to (though not necessarily phenotypically identical to) what has been called senescence or M1 in human fibroblasts. Unfortunately, the literature is further confused since what we termed selection in 1984 (i.e., the barrier we now think is most similar to M1, and which we are now calling stasis) was later called M0 in 1996 in the context of viral oncogene transformation; we believe that this latter nomenclature is inappropriate.

The HMEC at stasis did differ from the mammary fibroblasts in displaying fewer PD (~15-25 vs, ~40 respectively) before reaching this proliferation barrier. Human fibroblasts may experience less stress than human epithelial cells in vitro, as they routinely exhibit more PD prior to arrest. Some fibroblast strains, under some (low-stress) conditions, may even maintain growth until the second barrier (agonescence, see below). The mammary fibroblasts never showed spontaneous escape from this arrest, whereas the HMEC grown in MCDB170 do evidence a low rate of spontaneous escape during self-selection. Also, upregulation of p21 has been reported to play a role in fibroblast senescence, but was not seen in HMEC at this proliferation barrier. p53-dependent p21 expression may play a role in the stasis barrier encountered by human keratinocytes (Rheinwald et al., 2002).

We have shown that the pre-stasis HMEC grow somewhat longer in MM vs. MCDB170, accompanied by a slower rise in levels of p16 expression, and that spontaneous escape from stasis (i.e., self-selection) occurs only in the MCDB170 medium. Additionally, the rate of self-selection can be influenced by agents like cAMP stimulators; some individual specimens required a cAMP stimulator to generate post-selection cells. Therefore, our data agrees with others who have recently suggested that the nature of the culture conditions influences both how many PD occur before this block is engaged, and whether some HMEC can overcome this block.

The triggers for this stasis block have not been clearly defined. We currently favor the hypothesis that this p16-associated, RB-mediated senescence represents a cumulative response to stresses, broadly defined as conditions that are not conducive to optimal metabolic function. The term stasis was introduced to describe a stress-associated senescence (Drayton & Peters, 2002; Wright & Shay, 2002). Exposure to stress-inducing conditions may occur in cells in vivo, as evidenced by the age-related increase in p16 expression in breast tissue (Nielsen et al 1999). Stasis may be readily overcome by multiple means, consistent with an RB-mediated arrest. In HMEC, loss of p16 expression appears to be sufficient; in keratinocytes, inactivation of both the p16 and p53 pathways is required. Other alterations that target the RB/p16 pathway can also overcome this arrest, such as overexpressed cdk4 or Bm1-1, or RB inactivation. Many alterations that target RB function are seen in human cancers, suggesting that overcoming this first barrier may be essential for carcinogenesis, and that an RB-mediated, stress-associated barrier may be an in vivo as well as in vitro obstacle to immortal transformation.

Previously, telomere attrition was postulated as a key determinant of the p16-associated senescence/M1 block in both fibroblasts and epithelial cells. However: a) the mean TRF length is ~6-8 kb in the senescing HMEC at stasis, while in post-selection HMEC mean TRF can decrease to ≤ 5 kb, and b) introduction of hTERT alone is insufficient to extend the lifespan of pre-stasis HMEC populations (although rare immortal variants, with alterations in p16 expression, may occur) while hTERT uniformly immortalizes post-selection HMEC (see section IV.D). These findings suggest that telomere shortening can not be solely responsible for the p16-associated growth arrest in the pre-stasis cells. It is possible that TRF length could influence the expression of p16, or other growth inhibitors which might be playing a role in the G1 arrest expressed by cells at this barrier. It remains to be determined whether the stasis barrier has any dependence upon an intrinsic clock-like mechanism or whether p16 expression and growth arrest are primarily cumulative responses to the presence or absence of extrinsic factors.

A current theory that I entertain is that this p16-associated senescence is most closely related to what has been described as aging in model organisms, i.e., a mechanism associated with oxidative damage, stress, and dietary metabolism. In contrast, the more clock-like telomere-dependent senescence (agonescence/crisis, see below) is likely to be a much more recent evolutionary development, serving primarily as a cancer prevention mechanism for long-lived organisms.

II.B.2. Growth and Senescence of Post-selection HMEC

Index

Depending upon the individual specimen, post-selection HMEC proliferate for an additional 30-70 PD before reaching a second block to further growth. Growth slows around 3 passages prior to the passage at which no further net gains in cell number are observed, with increasing numbers of SA-b-gal positive cells displaying a larger, vacuolated, but still cobblestone epithelial morphology. The mean TRF of this growth-arrested population is ~5 kb. As mentioned above, this growth arrest is notable for its extreme stringency in culture.

Cytogenetic analysis of post-selection HMEC showed gross chromosomal abnormalities in metaphase spreads beginning 10-20 PD before the final passage. The abnormalities included abundant telomeric association and chromosome fusion and breakage events, polyploidy and aneuploidy. This onset of genomic aberrations coincided with the slow-down in proliferation rate. At the point of no net proliferation, 100% of metaphases exhibited structural abnormalities. Further analysis indicated that the post-selection HMEC at the final passage had an ~15% LI, ~20% staining with Annexin-V, arrest in both G1 and G2 (2N to 4N ratio ~1), and a substantial polyploid subpopulation. There was no net increase in cell number since continued DNA synthesis was associated with mitotic catastrophes and resulted in cell death or non-proliferative multi-nucleated cells. This second proliferation barrier has been termed agonescence.

On gross inspection, HMEC at agonescence exhibit long-term viability. Metabolically active cultures have been maintained for over a year. This overall viability contributed to the assumption that this second growth arrest was similar to the viable M1/senescence block in fibroblasts, and unlike the p53(-)/viral oncogene-induced crisis or M2 stage. However, the more recent studies from the Tlsty lab offer a new perspective. In vitro model systems of crisis have utilized cells lacking functional RB and/or p53, e.g., cells exposed to SV40 LgT or HPV16 E6/E7, or derived from Li-Fraumeni patients. In vivo, normal cells would retain p53 and RB function upon reaching senescence barriers, as do our HMEC cultures at agonescence. Similar to crisis, there are widespread chromosomal aberrations, as well as some ongoing DNA incorporation in HMEC at agonescence. However, the percentage of cells showing DNA incorporation is significantly lower than in crisis, and most cells remain viably arrested in G1 or G2. Unlike cells at crisis, the HMEC at agonescence do not spontaneously transform to immortality in culture, and there is not widespread cell death. The retention of wild type p53 may constitute the difference between the largely viable agonescence observed in post-selection HMEC, and the crisis seen in cells lacking functional p53 (see section below).

The nature of the agonescence block can account for the observed stringent senescence in cultured human epithelial cells; cells which fail to maintain a G1 or G2 arrest at agonescence will eventually die or become non-proliferative via mitotic catastrophe. Unlike an arrest based upon blocking cell cycle progression (e.g., elevated levels of CKIs), the agonescence barrier involving structural failures at mitosis can not be readily overcome and is irreversible. In contrast to the p16-associated senescence block, agonescence does appear to be largely or exclusively due to telomere attrition, and thus may serve as the telomere-based clock for limiting cell division.

We do not currently know if cells that have downregulated p16 play a normal or pathologic role in epithelial tissues. A recent study in Dr. Tlsty's lab has identified a minority population in vivo of what appear to be post-selection HMEC (cells with methylated p16 promoters). These might represent a normal subpopulation with more extensive proliferative capacity. However, the down-regulation of p16 in HMEC exposed to a carcinogen suggests that this process may occur under pathologic conditions, and such cells in vivo could be on the path of carcinogenic progression.

Given the stringency of the agonescence block, it is not surprising that overcoming agonescence, i.e., immortal transformation, may be a rate-limiting step in malignant progression. We'll discuss our hypotheses about the steps required to overcome agonescence after describing (section IV below) the agents that we have found capable of immortally transforming post-selection HMEC. In brief, we suggest that overcoming repression of hTERT expression (the catalytic subunit of the human telomerase complex) is required, and that HMEC have multiple reinforcing means of hTERT repression. Thus, two to three alterations may be required to overcome agonescence and reactivate telomerase. In cultured HMEC under no selective pressures, the likelihood that all the necessary errors would occur in the same cell, even under the conditions at agonescence where widespread genomic errors are generated, is exceedingly small.

II.C. Growth after Inhibition of p53 Function

(references: Garbe et al., in preparation)

Index

(section under construction)

Figure 2.3. Morphology of post-selection cells at agonescence (p53+) or during crisis (p53 inactivated by the GSE22)

Figure 2.4.

A: Effect of p53 inactivation on the G2 checkpoint in post-selection HMEC

B: Effect of p53 inactivation on the G1 checkpoint in post-selection HMEC

II. D. Characterization of Cultured HMEC Compared to Tissue of Origin

(references: Taylor-Papadimitriou et al.; 1989, Stampfer & Yaswen, 1992; )

Index

Since a main goal of studying human cells in vitro is to gain understanding of in vivo processes, we have considered it extremely important to characterize the HMEC grown in culture with reference to what is known about human mammary cells in the body. Unlike other organ systems, HMEC in culture (with the exception of lactational fluids and very rare surgical specimens) are not obtained from functionally differentiated tissues (i.e., pregnant, lactating, involuting). Consequently, we have not put much effort into examining these cells for features of functional differentiation. The cells we distribute, growing under standard culture conditions, do not express a-lactalbumin or b-casein. We have instead focused on the type of differentiation we termed "maturation", referring to the developmental history from a proliferative cell population to a cell with diminished reproductive capacity to a "terminally differentiated" cell no longer capable of division. We have been particularly interested in this pathway because human breast tumor cells in vivo and tumor-derived cell lines almost uniformly express the phenotype of the most mature normal HMEC in vivo.

The mammary gland consists of pseudostratified epithelia, with a basal layer resting upon a basement membrane and an apical layer facing the lumen of the ducts and alveoli. The basal layer of cells does not contact the lumen, whereas the apical layer may contact the basement membrane as well as the lumen. Apical cells display a polarized morphology, with microvilli at the luminal side. The myoepithelial cells, which contain muscle-like myofilaments, and which contract upon appropriate hormonal stimuli to cause expulsion of milk, lie in the basal layer of cells. Based upon examination of keratin expression and other marker antigens, it has been proposed for the rodent mammary gland that a stem cell population capable of differentiating into both myoepithelial cells and the apical glandular epithelial cells, also resides in the basal cell layer. The actual maturation lineage of human mammary epithelial cells in vivo has not been clearly defined. Based on the rodent mammary gland, and other epithelial tissues, it is reasonable to hypothesize that the most proliferative epithelial population in vivo lies in the basal layer, or intermediate between basal and luminal layers. Conversely, the luminal cells presumably have reduced proliferative capacity, with the most mature and least proliferative cells being those shed into the lumen (and recovered in nipple aspirations and milk fluids).

NOTE: I do not equate basal cells with myoepithelial cells, which are specialized differentiated contractile cells. Myoepithelial cells may all be basal, but all basal cells may not be myoepithelial. Unless one can demonstrate the presence of myofilaments, I do not think a cell should be referred to as myoepithelial. This is of importance since there have been references to the post-selection HMEC as "myoepithelial". As described below, since these cells express luminal epithelial markers, this appellation is incorrect.

A variety of studies from Joyce Taylor-Papamidritriou's group and others (see references at end) have defined properties which can be used to distinguish basal vs. luminal human breast cells, cells during lactation, and tumor cells. In general, mammary basal cells, similar to basal cells in stratified tissues such as the skin, express keratins 5 and 14. a-actin is present and the calmodulin-like protein (CLP) is preferentially found in the basal cell layer. A subpopulation expresses the common mesenchymal intermediate filament, vimentin. Luminal cells express the keratins 8 and 18 found in simple epithelia like the lung; keratin 19 shows variable expression and the most mature population is probably keratin 19 positive. In culture, keratin 19 expressing cells display very little proliferative potential. Expression of specific epitopes of a polymorphic epithelial mucin (PEM) is localized to luminal cells in vivo. Cells in the resting gland are weakly PEM positive, whereas cells from lactating glands may express higher levels of specific mucin epitopes. Like keratin 19, high expression of specific PEM epitopes has been correlated with a low proliferative potential in milk derived HMEC in vitro. Only a small fraction (~3-10%) of normal HMEC in vivo show detectable estrogen receptor-a, and this positive population is preferentially localized in the non-basal layer. It is not clear that the mammary gland contains cells which are terminally differentiated, such as those in the most mature layers of stratified epithelium, since even keratin 19, PEM positive cells have a limited capacity for cell division in vitro.

Unlike what one might intuitively expect, and unlike stratified epithelial tissues, breast tumor cells in vivo and tumor cell lines in vitro almost all have the phenotype of the most mature luminal cell - positive for keratins 19, 8/18, high expression of several PEM epitopes, including those found in the differentiated lactating cells, negative for keratins 5/14, and CLP. A small percentage of breast tumors have a more basal phenotype - keratin 5/14 positive. The consistency of tumor expression of keratin 19 has been utilized to locate micrometastases in lymph nodes. As normal HMEC with this luminal phenotype show little or no growth in culture, I think this tumor cell phenotype is indicative of some aspect of derangement in growth control. Most tumors also initially have high expression of estrogen receptor-a, and most are initially negative for vimentin expression, although vimentin is seen in a subset of estrogen receptor negative breast tumor cell lines and tissues. I don't know of a definitive explanation for this tumor cell phenotype, nor has the maturation state of the tumor cell precursor in vivo been clearly established. Benign proliferative tumors and some in situ carcinomas contain keratin 19 negative cells, so a keratin 8/18 positive, 19 negative cell could be the precursor of invasive breast tumors. If so, expression of the keratin 19 and high PEM phenotype in invasive tumors may be a consequence of malignant transformation.

[ER, karyotype, p53, telomerase etc, of pre-malignant cells; under construction]

In collaboration with others, we have examined the HMEC grown under our culture conditions for expression of the above phenotypic markers. Primary cultures of normal HMEC grown in MCDB 170 and early passage cultures grown in MM are heterogeneous. Some cells have the basal phenotype: keratin 5/14 positive, PEM negative, and vimentin, CLP and a-actin positive; other cells show the luminal phenotype: keratin 5/14 negative, keratin 8/18/19 positive, PEM positive; and some are in-between (e.g., keratins 5/14/8/18 positive). The cells that initially proliferate in MCDB 170 medium have the basal phenotype. However, post-selection cells begin to express some luminal markers, i. e., keratins 8 and 18 and some PEM epitopes. Expression of these luminal properties increases with continued passage in culture, such that cells at agonescence uniformly express these markers. At the same time, expression of the basal keratins 5/14, CLP, and vimentin is not lost. We have not detected keratin 19 or estrogen receptor-a in the post-selection population. Post-selection HMEC prior to the onset of agonescence show a normal karyotype.

The above results initially led us to propose that the post-selection HMEC represent a multipotent stem cell population initially present in the basal layer of the gland, since these cells show a partial differentiation towards the luminal phenotype with passage in culture. However, with the current understanding that this population has down-regulated p16 expression, the nature of this population relative to the in vivo situation is currently unclear, as discussed above. It is also possibile that culture conditions have induced some artifactual phenotypic expression. In particular, growth of cells on impermeable plastic substrates prevents the normal cell-extracellular matrix contacts and precludes the normal development of cellular polarity. Current bottom line is that we don't know if these p16(-) cells represents a normal or pathologic cellular process, or an in vitro artifact.

NOTE: The post-selection cells in MCDB 170 represent a limited phenotype. While they represent a wonderful normal human epithelial finite lifespan cell with which to study questions of growth control, for some experimental purposes, these may not be the best cell types to use. Cells grown in MM type media show the range of in vivo phenotypes. However, under standard culture conditions, they have a much more limited lifespan and we have less of them to distribute. Additionally, although they contain keratin 19 positive cells, these are not the actively dividing cells in the population.

The finite lifespan HMEC cultures have been examined for telomerase activity using the TRAP assay. We have not observed activity in the post-selection HMEC. However, in MM-grown HMEC from some individuals, a low level of telomerase activity has been detected. Similar results have been reported from other laboratories. No telomerase activity has been reported detected in normal human breast tissues in vivo, in contrast to some other epithelial tissues (e.g., colon, skin, ovary) where low levels of activity are seen in the presumed self-renewal populations. Some telomerase activity has been reported in fibroadenomas, a benign breast lesion.

III. In Vitro Transformation of HMEC Following Benzo(a)pyrene Exposure

III.A. Derivation of Extended Life Cultures

(references: Stampfer & Bartley, 1985; Stampfer & Bartley, 1988; Walen & Stampfer, 1989)

Index

One of my main goals in developing a culture system for normal HMEC was to try to transform the normal cells, so that different stages of malignant progression could be compared using cells from one individual. I wanted to use a chemical carcinogen as the agent for transformation because: (1) I wanted to induce random errors; (2) there was a lot of data indicating that polycyclic aromatic hydrocarbons (PAH) were good inducers of mammary cancer in rodents, and I wondered if the same could be true for humans; (3) chemicals seemed easier to use than radiation; (4) we performed a series of experiments indicating that HMEC were very efficient at converting PAH procarcinogens to their active form (see section VII. A .).

Three sets of experiments were performed in the early 1980's using primary cultures derived from HMEC specimen 184. The three separate original cultures had the FreezeDownSymbols: "aleph"(A), "cross"(C), and "birdie"(B). In each case, cells in at least 2 T-25 flasks were treated for 2 or 3 24 hr periods with 1-2 µg/ml of benzo(a)pyrene (BaP), (a concentration that gave 80% killing) and 2 T-25s were treated as controls. The cells were grown in MM medium with cholera toxin, in which 184 HMEC normally stop growing by passage 5. We followed the fate of the treated and control cells both in primary culture (how long growth was maintained in the primary T-25 flasks) and after subculture (the primary flasks were partially trypsinized many times, and after some trypsinizations the removed cells plated and subcultured until growth ceased).

Figure 3.1 shows the detailed (!) fate of these cells and indicates the nomenclature we used to identify Extended Life (EL) cultures and the immortally transformed cell lines. We gave a number to each of the subcultures we subcultured in experiments C (cross) and B (birdie); i.e., the B1, B2, B3 etc. that you see in figure 5. The EL cultures (i.e., treated cells that kept growing after the controls had stopped growing) derived from each of these subcultures were given the alphabetical equivalent to these numbers, e.g., B5 = Be, C2 = Cb. The immortally transformed lines that developed were given the subculture name, i.e., B5 and A1. The EL cells were notable for their heterogeneity with respect to morphology and growth potential (see Figure 3.2abcd). Growth often followed a punctuated pattern, with outgrowth (lasting 1-5 passages) of individual patches or colonies within non-growing populations.

We have extremely limited frozen stocks for some of these EL cultures. Subsequently, we grew some of these MM derived EL cultures in MCDB 170, which we found permitted growth for an additional 2-5 passages. Still, we have very limited stocks of EL cells so we do not readily distribute them. We would consider specific cases of collaboration or mutual interest, so check with me.

NOTE: we now know that all of the EL cultures tested do not express p16. In only one case, 184Aa, is this due to a detectable mutation. Thus, we surmised that while some cells can spontaneously downregulate p16 in MCDB 170, in MM, some cells will downregulate p16 after carcinogen exposure, but not spontaneously (at least, we've never observed this). Thus, the "extended life" seen in these EL carcinogen treated cells is in relation to MM grown controls. When grown in MCDB 170, EL and post-selection cells cease growth at approximately the same telomere length. We haven't looked at the karyotypes of growth arrested EL cultures, but we assume for the moment that they too undergo agonescence.

NOTE: most, though not all, of the EL cell populations had morphologies/growth patterns clearly distinct from anything in the untreated populations. I suspect this implies something about changes that have occurred, which could affect cell-cell or cell-matrix interactions

( Click here to see figure 3.1)

Figure 3.1. Growth of BaP treated specimen 184 in MM. It was conveniently fortuitous that the FDSs chosen for these three experiments started with the first three letters of the alphabet. When published, I presented them in the order A, B, C so it would look intentionally linear, but the actual order of the experiments was A, C, B and their real names are aleph, cross, birdie. The figure follows the fate of the BaP treated (T) and control (C) cells in primary culture and upon subculture. Since several dishes were plated at each subculture, and if growing, their lineages followed independently, more than one kind of growth pattern could be observed at a given passage level. In experiment 184C, cholera toxin was inadvertently omitted from the medium until 22 days after seeding (p5 of subculture C1, p4 of C2, p3 of C3).

(Click here to see figures 3.2abcd)

Figure 3.2. Morphology of EL cultures, Giemsa stained.

(A) 184C passage 8 with mixed growing and non-growing cells throughout the dish;

(B) 184C passage 6 containing two focal growing areas, one with uniform growth (shown) and one with mixed growing and non-growing cells (not-shown);

(D) 184C passage 7 with swirly thumbprint morphology. 184Aa had almost exactly the same appearance when it first showed up as a single patch in 184A passage 5;

III.B. Derivation of Cell Lines 184A1, 184B5, and 184AA4

(references: Stampfer & Bartley, 1985; Stampfer & Bartley, 1988; Walen & Stampfer, 1989; Lehman et al., 1993; Brenner & Aldaz, 1995; Sandhu et al., 1997; Stampfer et al., 1997; Stampfer & Garbe et al, 2003)

Index

Eventually, almost every EL cell from the BaP-treated 184A, B, and C experiments ceased growth. The two exceptions were the appearance of the 184A1 by passage 9 of the EL 184Aa population, and 184B5 in the passage 6 EL 184Be. 184A1 stood out as a more refractile appearing cell growing more vigorously as single cells, compared to the patchier, flatter, less vigorous 184Aa (which ceased growth by passage 11 in MM). 184B5 was a sickly looking small tight patch, somewhat more refractile than 184Be, very slow growing, that strongly caught my attention for indefinable reasons. It is curious and perhaps indicative of some underlying structure that the first time I saw both these cell populations, I was sure they were immortally transformed, and I didn't have that sense with any other cells in the many EL cultures.

Some 184A1 cells were transferred to MCDB 170 medium at passage 11 and carried continuously in that medium to passage 105. 184A1 growth was slow and heterogeneous from around passages 16-30, during the conversion process, and good uniform growth was present by around passage 40 (see more about conversion in Section IV). Early passage conditionally immortal 184A1 cells grew with minimal cell-cell contact at low densities, and showed morphologic heterogeneity, with the presence of large vacuolated cells. At higher passages, when converted to full immortality, the growth pattern showed more cell-cell association and patchy growth, with few vacuolated cells visible (see Figure 3.3bc, and discussion the of conversion process in Section IV). 184A1 was also maintained in MM up to passage 69.

184B5 was first transferred to MCDB 170 at passage 9 and grown to passage 101. G growth initially slow, gradually increased during the conversion process, with good uniform growth present by passage 30. 184B5 has a distinctive morphology, growing in tightly packed patches (see Figure 3.3d). An advantage of this is that the fate of single cells can easily be followed without needing to seed at clonal densities (the progeny of a single cell stay attached and make a colony). 184B5 maintained in MM were grown to passage 30.

(Click here to see figures 3.3abcd)

Figure 3.3. Morphology of 184A1 and 184B5, Giemsa stained. All pictures shown at the same magnification.

(A) 184 passage 9 in MCDB 170;

(B) 184A1 passage 15; note the large vacuolated cells;

(D) 184B5 passage 11 in MCDB 170.

Both 184A1 and 184B5 show a few specific clonal karyotypic aberrations, indicating their independent origins from a single cell. Some of the karyotypic abnormalities found in 184B5, e.g., 1q22 breaks and tetrasomy for 1q, are also frequently observed in cells obtained from breast tumors. Upon continued passage in culture, these two lines show some genetic drift (more so in 184B5, very little or none in 184A1), but it is relatively minimal compared to that observed in most human breast tumor cell lines. Even at passage 41, 184B5 has clearly identifiable chromosomes and a near pseudodiploid karyotype. Thus, the vast majority of the cell population would be expected to remain karyotypically stable when studied over the course of a few passages in culture, yet the presence of some genetic drift could give rise to rare variants in the cell population.

Neither 184A1 nor 184B5 expresses p16. 184A1 (like its precursor 184Aa) has mutations in the p16 alleles, while 184B5 shows methylation of the p16 promoter. Both 184A1 and 184B5 have wild type p53 by sequence analysis, and express p53-dependent functions. Like the p16(-) post-selection HMEC, the p53 in these lines is relatively stable. Both lines have normal RB, which becomes phosphorylated in response to appropriate growth stimuli. There are no known defects in other key cell cycle molecules such as p21, p27, and p15, G1-associated cyclins and cdks. p14^ARF is barely detectable in all our HMEC cultures examined.

NOTE: We have recently found that some lineages of fully immortal (post-conversion) 184A1 show defects in some p53 functions. We have not yet examined the p53 sequence in these lineages nor performed further characterization. Check back with me if you are concerned whether a 184A1 culture you may have or have used shows normal p53 functions.

We have more recently obtained a new p53(+) immortal cell line derived from EL 184Aa. 184AA4 appeared at passage 13 in MCBD 170-grown 184Aa as densely packed very slow growing patches of small flat cells. Growth remained very slow and non-uniform, with subculture only every 4-8 weeks and extremely low CFE (<0.1%), until around passages 25-28. Mass culture growth then gradually increased, with uniform good growth attained by passage 40. Early passage 184AA4 cell morphology was heterogeneous, with large flat cells mixed with areas of smaller cells with some visible mitoses. Later passage good growing 184AA4 had a typical epithelial cobblestone morphology. 184AA4 is also p16(-) and has stable p53.

Unlike 184A1 and 184B5, 184AA4 evidenced numerous karyotypic abnormalities when it was first able to be examined at passages 22-26. The numerous chromosomal aberrations included 4-9 marker chromosomes, evidence of double minutes, and many dicentrics. However, examination of the good growing fully immortal passage 56 population showed fewer aberrancies. The karyotype remained mostly near diploid, with only 3-4 of the largest marker chromosomes present, and no evidence of dicentric or double minute chromosomes. Cells with severe chromosomal aberrations might have been non-proliferative, and preferentially lost during conversion, perhaps contributing to the very low CFE of 184AA4 during that time.

While we were initially puzzled by this difference in karyotypic abnormalities between 184AA4 and 184A1 (both derived from the same EL 184Aa precursor) the uncovering of the agonescence process may provide an explanation. The difference between 184AA4 and 184A1 likely results from when the derangement occurred that permitted these cells to overcome agonescence. Agonescence can be detected around 2-4 passages prior to the cessation of net proliferation in post-selection HMEC. Based on the mean TRF length of early 184A1 (> 5 kb), and its appearance in 184Aa by passage 9, 184A1 most likely immortalized prior to encountering much, if any, agonescence induced chromosomal rearrangements. 184AA4, however, had to have immortalized in passage 12-13 184Aa cultures - a population presumably well into agonescence. Thus, the initial immortalized cell likely harbored numerous chromosomal derangements due to agonescence, and independent of the immortalization process. Once cells harbor chromosomal rearrangements, as early 184AA4 does, ongoing chromosomal fusion-breakage cycles may perpetuate genomic aberrations, even in the presence of functional p53. This has interesting implications for understanding the karyotypic abnormalities common to human breast cancers, the majority of which are p53(+) (see more in Section IV.F.2.). A cell sustaining an immortalizing event during agonescence may harbor multiple karyotypic derangements that are not due to either immortal transformation or p53 loss, while a cell that immortalizes earlier may be more genomically stable.

Although 184A1, 184AA4, and 184B5 have an indefinite lifespan (they are immortally transformed), they do not have properties associated with malignant transformation. They do not form tumors in nude mice and they do not express a sustained capacity for anchorage independent growth (AIG), although 184B5 can show a low level of random colony formation. They are also growth factor dependent for growth. Therefore, these lines can allow examination of immortal transformation without the confounding properties of malignancy associated changes. The presence of wild type p53, RB, and other cell cycle regulators, the absence of viral oncogenes, and the relative genomic stability of 184A1 and to a lesser extent 184B5, makes these immortal lines especially useful for a variety of studies. However, it's important to remember that they are immortally transformed, and we have uncovered several obligate alterations in key signal transduction pathways as a consequence of this transformation (see Section IV.A.)

Since 184A1, 184B5, and 184AA4 are cell lines of indefinite lifespan, I have unlimited supplies to distribute.

CAUTIONARY NOTE: We were remiss in our earliest tissue culture years in not routinely checking all cells for PPLO contamination. We first started routinely testing for PPLO in 1982, after experiments "A", "C" and "B" were initiated, and the cell lines 184A1 and 184B5 were being maintained in MCDB 170. These lines, as well as other normal and BaP treated EL cells growing in MCDB 170, were tested for PPLO by Hoechst stain and growth in agar broth. The results were all negative. About a year later, we took out some of our frozen BaP treated EL cells, and placed them in MM. Now our routine Hoechst stain test showed some of them to have foreign DNA, although broth growth was still negative. Transfer of the samples to MCDB 170 generally led to loss of the Hoechst stain positive material within 2 passages. This negative phenotype was retained after transfer back to MM. Two (visually equivalent Hoechst stain positive) samples were sent to Microbiological Associates for assay by agar growth, Hoechst, and strain-specific antibodies. They reported one to be completely negative and the other positive for M.hyorhinis (which doesn't grow in the usual agar broth assay). From our Hoechst stain results, it appears that 184Aa, the EL precursor of 184A1, was positive, while some, but not all of 184Be, the EL precursor of 184B5, may have been positive. We can not say whether this may, or may not, have affected any results. By the Hoechst assay, all our current HMEC are negative. It was a mistake for us not to have been testing our cells. A take-home message is to be careful to check your cells on a regular basis.

III.C. Derivation of Early Variants of 184A1 and 184B5

(references Clark et al., 1988; Stampfer & Yaswen, 1992)

Index

One of my original goals for in vitro transformation of HMEC was to obtain malignant transformants from normal HMEC. To be honest, somewhere along the line I realized that although this was important science, I myself was less than enthusiastic about creating malignant cells from normal, and the immortally transformed cells were sufficient for my scientific curiosity (or as I commented, as I got older, the question of immortality seemed more interesting than malignancy). Nonetheless, I did try (and not succeed) to obtain malignancy by exposing the 184A1 and 184B5 lines to further chemical carcinogens, in this case, the direct acting carcinogen, N-nitroso-ethyl-urea (ENU). Others exposed the lines to specific oncogenes, which could lead to cells that were AIG and/or made tumors in nude mice. As part of this work, I analyzed the cells for their nutritional requirements and isolated/developed cell variants with altered nutritional requirements. The history of these studies and the names/origin of these variants are described below. More recent studies involving retroviral infection of these lines with viral and cellular oncogenes, and hTERT, can be found in Section IV.

III.C.1. Nutritional Variants

Index

Nomenclature Note: Spontaneously occurring subpopulations isolated based on the nutritional composition of the medium were designated as A1N... and B5N.... In the case of MM-grown subpopulations, the N is followed by a (arbitrary) number; in the case of MCDB170-grown cells, the N is followed by letters indicating what factors were no longer required. Nutritional variants obtained following exposure to ENU were designated A1ZN... and B5ZN... followed by letters indicating the non-required factors. While I sometimes refer to these nutritional variants by their complete names (e.g., 184A1N4), for brevity and simplicity sake, it's OK to officially refer to them without the 184 prefix.

The first nutritional variants were isolated (in a none too systematic fashion) from 184A1 growing in MM medium. For general purposes, the only one of these to note is 184A1N4. 184A1 was seeded at passage 16 in MM minus the conditioned media and without cholera toxin. Attachment appeared poor but the few patches that were present grew fairly well. After 3 passages, there appeared to be uniform good growth and attachment. These cells were first transferred to MCDB 170 at p28. The karyology of A1N4 indicated that, unlike the pseudodiploid 184A1, A1N4 were aneuploid (near triploid) with only one additional chromosomal marker beyond the 4 seen in the parental 184A1 cells. It is therefore likely that, although not cloned, they represent a clonal population. The A1N4 were used by Robin Clark for malignant transformation with oncogenes. Some investigators that have used the A1N4 variant for past studies are still using it. I do not recommend new use of A1N4 instead of 184A1 since it is more aberrant, aneuploid, and uncharacterized.

More systematic isolation of nutritional variants was done in MCDB 170 medium. The existing literature indicates that many transformed cells show reduced nutritional requirements. As part of our initial characterizations of the immortally transformed cell lines, we first compared the requirements of 184, 184A1, and 184B5 for the individual growth factors present in MCDB 170 for short term growth, for long-term culture, and in clonal vs. mass culture (Table 2 and list below). 184A1 and 184B5 showed a few differences from each other and normal HMEC. Both were more dependent upon EGF for growth in mass culture whereas the normal cells could continue to proliferate without EGF (some of this difference is due to differences in secretion of TGFa in these cultures; see section V.). All of these HMEC showed a stringent requirement for EGF in clonal culture. 184A1 showed little effect upon removal of hydrocortisone (HC); 184B5 and 184 had greater short-term requirements. All the HMEC had a requirement for BPE for short-term growth. In the long-term experiments, removal of HC or BPE from mass cultures of normal HMEC led to cessation of growth over the course of 1 to 3 passages. Removal of insulin (I) did not prevent continued proliferation, but led to slower growth, a less healthy appearing culture, and earlier senescence. Removal of I from 184A1 and 184B5 also did not prevent continued growth. For more details of the long-term experiments with the cell lines, see the list below. These nutritional variants are available for distribution. My general conclusion from these studies is that 184A1 and 184B5 retain basically normal growth factor dependence (except for 184A1 and HC).

CAUTION: These studies were performed before we were aware of the conversion process. The passages of 184A1 and 185B5 used were still in conversion, and therefore not expressing uniform good growth. These data might be different if fully immortal cells were used.

Table 2. Growth Factor Requirements of Normal and Transformed HMEC in MCDB170

Percentage of Control Cell Growth

184 184A1 184B5

Medium MC^a CFE MC CFE MC CFE

Complete MCDB 170+IP 100 100 100 100 100 100

minus I 49 47 11 18 26 73

minus HC 36 32 84 88 18 61

minus EGF 86 2 20 0 12 0

minus BPE 15 21 21 24 16 75

_____________________________________________________________________

^aAbbreviations used: I, insulin; HC, hydrocortisone; EGF, epidermal growth factor; BPE, bovine pituitary extract; IP, isoproterenol; MC, mass culture growth; CFE, colony forming efficiency. Cells from specimen 184 (p11), and cell lines 184A1 and 184B5 (passages 17-20) were grown in complete MCDB 170 with isoproterenol. For mass culture, cells were subcultured into duplicate 35 mm dishes (5 x 10⁴ per dish) in the indicated media. When control cultures were subconfluent or just confluent, all the cultures were trypsinized and the cells counted by hemocytometer. For clonal cultures, single cells (100-1000) were seeded into triplicate 100 mm dishes. After 10-14 days, cells were stained with Giemsa and colonies greater than 30 cells counted.

List of Spontaneous Nutritional Variants:

184A1:

CAUTION: the nutritional requirement studies were done with 184A1 at passages £ 20 and the selection for variants was done with 184A1 around passages 27-32. We now know that these are non-homogeneous conditionally immortal populations. The results might be different if later passage fully immortal cells were used.

184A1NE: no EGF. For the first 2 passages, growth was slow and selective (a small number of patches). The growth rate was the same as control (+EGF) after 4 passages.

184A1NH: no HC. Slow patchy growth for first passage; growth normal after 2 passages.

184A1NI: no I. Growth was initially slowed, but less selective than -EGF; the cells looked good. Growth rates were normal within 2-4 passages.

184A1NB: no BPE. Little initial growth. Eventually a few patches grew out. After one additional passage the resultant cells grew normally.

184B5:

184B5NE: no EGF. Media first changed at passage 48. Growth initially slower and selective. It took 7 passages to select a population that looked good and had a normal growth rate. Repeated with cells at passage 36, after 3 passages of slow, selective growth a good growing population arose.

184B5NH: no HC. Media changed at passage 34. Growth initially slower but normal after 2 passages.

184B5NI: no I. Media changed at passage 35. Growth initially slowed but not as selective as -EGF. Growth rates were normal within 4-6 passages.

184B5NB: no BPE. Growth initially slow and selective but not as extreme as 184A1 -BPE. Cells didn't look good and grow normally until after 6 passages.

184B5NIB: no I or BPE. 184B5NB cells were switched at passage 42 to media without insulin. They grew initially slowly and poorly. Good patches were obvious after 2 passages and growth was normal after 3 passages.

We next examined the effect of removal of multiple growth factors to determine conditions where untreated 184A1 and 184B5 did not yield spontaneous nutritional variants. For 184A1, these were defined as removal of I and EGF, I and BPE, or EGF and BPE. For 184B5, these were defined as removal of I and EGF, or I and BPE. Populations of 184A1 and 184B5 were then tested for their ability to grow in these restrictive media after exposure to ENU concentrations that yielded 80% inhibition of colony forming efficiency (1500 µg/ml for 184A1 and 750 µg/ml for 184B5). Two T-75 flasks each of treated and control cells were exposed to ENU or solvent alone for 2 or 3 consecutive passages. Under a few conditions the ENU treated cells were capable of sustained growth whereas the untreated cell lines quickly ceased growth. The resulting growth factor independent variants may represent a further step in malignant progression. However, they did not show AIG or form tumors in nude mice.

List of ENU-induced Nutritional Variants:

184A1ZNEB: selected in MCDB170 -EGF-BPE. The treated cells had a fair amount of growth (compared to almost nothing in the controls), but most of this faded away after several passages. In one experiment, cells with patchy vigorous growth and a distinctive morphology quickly took over the population, and maintained active growth in this medium. Although not examined, these are presumably clonal.

184B5ZNEI: selected in MCDB170 -EGF-I. The treated cells showed initial widespread, morphologically heterogeneous, growth (compared to very little in the controls). Most of this growth faded after about 5 subcultures but in several cultures growth was maintained. The morphologies are not particularly distinctive and we don't know if these represent clonal cultures.

These variants are available for distribution. For more information on the non-EGF requiring variants, see Section V. Figure 5.2.

III.C.2. Oncogene Exposed Derivatives (early studies)

(references Clark et al., 1988; Stampfer & Yaswen, 1992; Frittitta et al., 1995)

Index

Derivatives of 184A1 and 184B5 capable of AIG and tumor formation in nude mice were initially obtained with the use of oncogene containing retroviral vectors and viruses, and transfection. In the case of 184A1, A1N4, a clonal derivative with reduced nutritional requirements, was exposed by Robin Clark to the genes for SV40 large T antigen, v-H-ras, and v-mos, singly and in combination. The combination of H-ras and SV40-T led to cells (designated A1N4-TH) which formed progressively growing tumors in nude mice and showed AIG. Exposure to v-H-ras (A1N4-H) or v-mos (A1N4-M) alone led to cells that produced tumors with reduced frequency and longer latency. SV40-T alone (A1N4-T) did not yield tumorigenic cells, but did affect the growth factor requirements for anchorage dependent and independent growth. In all cases of oncogene exposure, the resultant cells were capable of proliferation in media that did not support the growth of the parental A1N4 cells. A1N4-TH has a near tetraploid karyotype, which is missing the A1N4 chromosomal marker and contains only one additional clonal chromosomal aberration relative to 184A1. Thus even the malignantly transformed derivative of 184A1, containing SV40-T and thus inactivated p53 and v-H-ras, does not show a very unstable karyotype in terms of gross chromosomal aberrations.

The 184B5 cell line was exposed to v-K-ras (designated 184B5-K) by Paul Arnstein, yielding cells which were tumorigenic in nude mice, with short latency. However, these tumors did not grow beyond approximately 5 cm diameter. Our earlier studies utilized the culture designated 184B5-KTu, which was derived from a B5-K tumor resected from a nude mouse and replaced in culture. B5-K and B5KTu do not display AIG.

Although we have some stocks of these cells available, we no longer grow or distribute these cells, as the infection was with non-defective retroviruses. If you really want oncogene exposed 184A1 or 184B5, I suggest you transduce the cells with the oncogenes of your choice. We do now have available 184A1 infected, using defective retroviral vectors, with HPV16 -E6, -E7, E1a, or SV40T, as well as other infected cultures (see section IV.A.6). Alternatively, other investigators may have already infected the cells with your oncogene of choice, and you may query for this information (see Investigator List).

184B5 has also been exposed to transfection with erbB-2, mutated erbB-2, and the insulin receptor. ErbB-2 alone (184B5-E) made the cells capable of AIG (~5-15%, large colonies), while mutated erbB-2 additionally made them capable of tumor formation in nude mice. Overexpression of the insulin receptor also made the cells capable of some AIG. We can make available our stocks of 184B5-E. The cultures we distribute, 184B5ME, were grown up from AIG colonies selected from184B5-E placed in methylcellulose,

III.D. Characterization of 184A1, 184B5, and Derivatives Compared to Tissue of Origin

(references: Stampfer & Bartley, 1985; Stampfer & Yaswen, 1992; Stampfer & Yaswen, 1993; Sanford et al. 1992; Lehman et al. 1993; Thompson et al. 1994; Sandhu et al. 1997; Brenner et al. 1998; Stampfer, Garbe et al. 2003)

Index

Differentiation/maturation markers: In general, 184A1 and 184B5 have a somewhat more mature phenotype than finite lifespan 184. However, it is important to recall that these lines were derived from cells grown in MM medium, which contains cells with phenotypes more mature than basal. We don't know the cell of origin of EL 184Aa and 184Be, the precursors to these lines. Both lines maintain some expression of keratins 5 and 14, but at significantly decreased levels, while expression of keratin 18 is increased relative to normal post-selection 184 HMEC. Both lines have barely detectable levels of vimentin. 184B5 strongly expresses the luminal and tumor associated PEM antigens, while 184A1 has some but lower expression of PEM. The tumorigenic transformants, A1N4-TH and B5-KTu have very low levels of keratin 5 and increased levels of keratin 18. While B5-KTu remains vimentin negative, the A1N4-TH cells show re-expression of vimentin. We have not been able to detect keratin 19 or estrogen receptor-a in any of these lines. Thus none of these lines fully resembles most human breast tumor cells in vivo, and the phenotypic differences between the immortal lines and MCDB 170 grown 184 HMEC could just reflect maturation states (i.e., not be related to the immortalization process).

Fibronectin represents about 10-20% of the protein secreted by normal HMEC in culture. In many transformed cells, the level of fibronectin mRNA and protein synthesis is decreased. Expression of fibronectin is greatly reduced in 184A1 and to a lesser extent in 184B5. However, we do not know if fibronectin secretion would normally be lower in HMEC with a more mature phenotype. Upregulation of fibronectin synthesis by TGFb remains normal in both cell lines (see section VI.).

Another approach we took to characterize differences between our normal and transformed HMEC cultures was to use subtractive hybridization to identify genes expressed in the normal HMEC, but downregulated in the immortal and malignantly transformed cells. This was how CLP was first isolated and identified (see section VII.B. for more), and the difference in expression of keratin 5, vimentin, and fibronectin first observed.

The normal and transformed HMEC have also been characterized with respect to both their growth patterns and their gene expression when placed on reconstituted basement membrane material derived from the Englebreth-Holm-Swarm (EHS) murine sarcoma (Matrigel), which has been shown to support increased differentiated functions of a variety of cell types. Normal HMEC are capable of forming three-dimensional structures with striking resemblance to endbuds in intact mammary gland tissue, whereas 184A1 displays only less developed structures and 184B5 forms only small clusters. The A1N4-TH cells show even less structure formation than 184A1 and the B5-KTu cells resemble 184B5. We have not examined the underlying basis for these differences, and suspect that alterations in cell-cell connections may be involved. E-Cadherin is expressed by all of these cells with the exception of the aggressively tumorigenic A1N4-TH cells.

More recently, we and others have observed the growth patterns of 184A1 and 184B5 placed within Matrigel. Tumor derived HMEC can maintain growth within Matrigel, whereas normal HMEC cease division. In this assay, our minimally deviant immortal lines behave like normal HMEC.

Malignancy associated markers: As mentioned earlier, neither 184A1 and 184B5 shows AIG or tumorigenicity, and they retain mostly normal growth factor dependence. They do differ from normal HMEC in having some karyotypic abnormalities. 184B5 has been shown to have a 10x higher rate of mutations at the HPRT locus than normal HMEC, reduced intercellular communication, and reduced DNA repair during the G2 phase. The ability of 184A1 and 184B5 to gain AIG and to be malignantly transformed when exposed to specific oncogenes also differs from normal HMEC.

No differences in expression or regulation of the RB protein have been detected in 184A1 or 184B5. p53 sequence and/or functions appear the same in 184A1, 184AA4, 184B5 and post-selection HMEC, with the exception, as previously mentioned, of some 184A1 lineages. However, we and others have shown that the p53 expressed by the post-selection, EL, and immortally transformed cultured HMEC (but not the p53 in cultured fibroblasts from the same person) is in a conformation recognized by antibodies that recognize mutant p53, and the half-life of the p53 protein is 3-4 hrs. We do not know the functional significance of the presence of this stabilized form of p53 in these cells. I do note that the presence of stable p53 correlates exactly with the cell types which do not express p16, and wonder if there is some causal basis for this association. The absence of p16 in 184A1, as in 184Aa, is due to mutations in both p16 alleles. In 184B5, as in post-selection HMEC, the p16 promoter is methylated. The loss of p16 expression in this system (which retains normal RB) may therefore facilitate immortal transformation, but is by itself insufficient. Thus these cell lines, while immortally transformed, do not express markers of invasive or malignant transformation, and are therefore useful in studying the process of immortal transformation per se, without many other potentially confounding factors. In comparison to almost all other existing immortal lines, they are valuable as minimally deviant immortally transformed cells. Our recent studies have focused on elucidating the immortalization specific changes (see below).

IV. The Conversion Process during HMEC Immortalization

(references Stampfer et al., 1997; Garbe et al., 1999; Nijjar et al., 1999; Stampfer & Yaswen, 2000, Nonet et al., 2001; Stampfer & Yaswen 2001; Yaswen & Stampfer, 2001; Stampfer et al. 2001; Stampfer, Garbe et al., 2003; Olsen et al., 2002; Hosobuchi & Stampfer, 1989; Stampfer et al., 1993)

Index

In the course of our characterization of the 184A1 and 184B5 lines, we uncovered some results that were difficult to interpret, which led us on the journey of uncovering the conversion process in HMEC immortalization. The following sections will go into some detail about the nature of conversion, what we know of the underlying molecular mechanisms, how it is affected by alteration of specific gene expression, and the possible role of a conversion process in human carcinogenesis in vivo. They will also present personal speculations about the meaning of this all. The Brief History presents a short summary description of the conversion process.

I believe there is much to be learned from all of this. Although other laboratories have yet to investigate this process, most even failing to acknowledge or reference it, and there is as yet little investigation/data to determine a possible role of conversion in vivo, I think it most likely that conversion does play an important role in human breast carcinogenesis. The phenotype of our immortally transformed and fully converted HMEC more closely resembles that of human breast cancer cells than does the phenotype of HMEC immortally transformed using viral oncogenes or hTERT, or rodent mammary cell systems. I will suggest in these sections that widespread practices in the cell and molecular biology community have led us to use as "models systems" systems that clearly do not model the human disease process, and that these practices may be seriously limiting our ability to develop useful clinical approaches. I hope that the readers of this web site will carefully consider the data (and opinions) being presented, and help spread this information in ways that can further both basic science and therapeutic options.

IV.A. Conversion of Conditionally Immortal p53(+) 184A1, 184B5, and 184AA4 to Full Immortality

Index

I summarize our results below and then provide more details.

Early passage 184A1, 184B5, and 184AA4 cells are only conditionally immortal. The cell lines, but not each individual cell, have indefinite growth potential. These early passage conditionally immortal cells express little or no telomerase activity and show no ability to maintain growth in TGFb. Telomeres continue to shorten with increasing passage. Cell populations whose mean TRF length had declined to < 3 kb exhibited slow heterogeneous growth and contained many non-proliferative cells. These cells also accumulated large quantities of the CKI p57, which we believe is responsible for the extended period of poor growth. Telomerase activity is first detected when the telomeres become critically short, mean TRF ~2.5-2 kb, and activity levels gradually increase thereafter. Around the passage levels that telomerase activity can first be detected, there begins a very gradual increase in the number of cells displaying progressively increased ability to maintain growth in TGFb (see Table 3 below). By the time the mean TRF has stabilized at > 3 kb, the cells have converted to full immortality, characterized by high levels of telomerase activity, uniform good growth in the absence or presence of TGFb and little or no expression of p57. Additionally, the fully immortal converted HMEC can maintain growth in the presence of transduced oncogenic Raf; express little or no SA-bgal activity, and may have higher levels of c-myc expression, including increased levels during G0 arrest. We have used the term "conversion" to describe this gradual process. The consistent and reproducible manifestation of conversion by repeatedly cloned cell populations, and the very gradual nature of the conversion process, suggest an epigenetic mechanism. Fully immortal converted HMEC have obligate changes in several key pathways (e,g,, Ras/Raf, c-myc, TGFb, p57, possibly p27) compared to normal finite lifespan HMEC, and are therefore most decidedly not normal or untransformed.

NOTE: This whole process of conversion and p57 expression would not be seen in cell types where adult somatic cells do not have mechanisms of stringent replicative senescence/repression of telomerase, i.e., all rodent cells. I think this raises serious questions about the use of rodent "model" systems for studies to understand human cellular immortalization and early stage carcinogenesis. If a system doesn't model, it's not a model system.

IV.A.1. Early Observations that Led to Uncovering Conversion

Index

When 184A1 and 184B5 were initially characterized in 1982-5, we observed two growth patterns with no obvious mechanistic explanations:

(1) Although both immortal lines maintained continuous growth in mass culture following their initial emergence, slow, non-uniform growth occurred during the first 20-30 passages. As mentioned above, 184B5 first appeared as a sickly slow growing patch. It continued to grow slowly, with a gradually increasing proliferative rate, until it achieved fairly rapid growth by passage 30. Visual observation of the colonial outgrowths indicated that many cells didn't grow, or that colonies stopped growth at small sizes. Early passages of 184A1 also contained many vacuolated and non-proliferative cells. Back in the 1980's, not being able to think of a mechanistic explanation for why so many individual cells of immortal cell lines were non-proliferative, I chose to ignore this problem and only gave out higher passages of these lines, where growth was more uniform (>p30 for 184A1 and >p25 for 184B5). But this bothered me.

(2) While I could pretend to ignore the above problem, the response of these cells to the pleiotropic cytokine TGFb was too odd to ignore, and it was this in-my-face puzzle that led me to start unraveling the conversion process (see more about TGFb in Section VI..).

We have not seen a single finite lifespan HMEC able to maintain growth in the continued presence of TGFb, although cells that have undergone fewer PD in culture may undergo 5-10 additional PD before complete cessation of growth. Cells closer to agonescence stopped growth within 1-2 PD. In contrast, populations of 184A1 and 184B5 that maintained growth in TGFb could be isolated. However, the pattern of resistance to TGFb-induced growth inhibition by these lines was unusual. 184A1 mass cultures exposed to TGFb at passages (p) 28-35 displayed severe growth inhibition, but a small subpopulation of cells maintained active growth. Assuming these resistant cells represented rare mutations, we attempted to obtain pure populations by clonal isolation. However, like the parental uncloned population, all four clones isolated displayed a small subpopulation of cells capable of continuous growth in TGFb. 184B5 exposed to TGFb at p26-40 maintained good growth, but most clones isolated at p13-16 were strongly growth inhibited. One particular severely inhibited clone, B5T1, repeatedly underwent an apparent "crisis" around passage 30 during which almost all the cells died. The populations derived from the few surviving cells maintained growth in TGFb. The lack of growth inhibition by TGFb was not due to loss of the ability to respond to TGFb. All 184A1 and 184B5 cultures showed morphologic alterations in the presence of TGFb, and all cells tested displayed TGFb receptors and induction by TGFb of extracellular matrix associated proteins.

In an effort to understand (1) why so many early passage cells from immortal lines failed to maintain proliferation, and (2) how clonal isolates rapidly produced cell populations heterogeneous for growth in TGFb, I particularly noted the association of TGFb resistance with an indefinite lifespan in B5T1. Since the literature at the time was starting to associate telomerase activity with an indefinite lifespan, I considered the possibility that expression of TGFb resistance and telomerase activity might be related, and that possibly both phenotypes were not initially expressed in these immortally transformed HMEC. Perhaps these cell lines were initially only "conditionally immortal", i.e., permissive for immortality but an additional step was required for them to obtain a uniform indefinite lifespan. To test this hypothesis, we proceeded to carefully characterize and ascertain possible associations among morphology, growth capacity in the absence and presence of TGFb, telomerase activity, and telomere length in 184A1 and 184B5 at different passage levels. The sections below describes these and related data; Section IV.D. presents more recent studies which now directly demonstrate a relationship between hTERT expression and subsequent acquisition of resistance to TGFb-induced growth inhibition.,

IV.A.2. Telomerase Reactivation, Mean TRF Length Stabilization, and Gain of Uniform Growth Capacity ±TGFb during Conversion to Full Immortality

Index

The conversion process with reference to mean TRF length, telomerase activity, and growth ±TGFb is illustrated for MCDB 170-grown 184A1 in Figure 4A.1 and Table 3.

(Click here to see figure 4A.1)

Figure 4A.1: Comparison of mean TRF length, telomerase activity, and growth ±TGFb in 184A1 at different passage levels.

Legend for Figure 4A.1.

Panel A: Mean TRF length; lighter shaded ovals indicate a faint signal.

Panel B: Telomerase activity, determined semi-quantitatively by comparing the levels of HMEC telomerase products generated to those generated for a constant number of 293 cells (1,000 cell equivalents). The following categories were used to designate semi-quantitative values. Note that the points are presented in a semi-log form: None = no detectable telomerase products by PhosphorImager analysis; weak = approximately 5% of telomerase activity of 293 cell control; low = approximately 10% of 293 control; medium = 25-50% of 293 control; strong = 75-100% of 293 control.

Panel C: Colony forming efficiency (CFE) and labeling index (LI) in colonies.

Mean TRF length, telomerase activity, CFE and LI were determined as described in Stampfer et al., 1997.

Early passage (11p) 184A1 has a mean TRF of ~5 kb, similar to the TRF length of agonescent post-selection HMEC. Since 184A1 had already undergone numerous PD before this assay, its mean TRF length at first emergence must have been longer. No telomerase activity is detectable at the early passages. Consistent with this, the mean TRF continues to decline with the ongoing cell proliferation. Early passage 184A1 displays uniform good growth with a high colony readily forming efficiency (CFE).

However, when grown in MCDB 170, the CFE steadily decreases with passage, with an abrupt decrease in growth around passage 16, when the mean TRF has declined to ~3 kb. In our lab, we refer to this as "hitting the wall" because of its relative abruptness, and the cells thereafter look kind of "smashed" (gross looking flat vacuolated cells). Between passages 18-30 the CFE remains low, and most colonies that are still growing contain a mixture of growing and non-proliferative cells. During this period, the mean TRF declined to £ 2 kb with a faint signal. Low telomerase activity was first detectable around passages 24-30, and increased thereafter. After passage 30, mean TRF stabilized at > 3 kb, the CFE increased, and the growth displayed by individual colonies gradually became uniform. The first detection of sustained growth in TGFb was at ~passage 28. This growth was exceedingly poor - but it was maintained. By passage 30, some cells showed OK growth in TGFb, and after passage 40, most cells showed good growth in TGFb.

The data on the conversion process in 184AA4 is very similar to that seen in 184A1 (see Figure 4C.2 and Table 3). The main difference is that 184AA4 appeared later, at passage 13, with a mean TRF already < 3kb when first examinable. It was consequently already in conversion, and as discussed previously (see section III.B.) had presumably encountered agonescence before immortalization, therefore displaying many more karyotypic abnormalities. These chromosomal errors may have contributed to the very low CFE of 184AA4 during conversion (~0.01% vs. ~1% for 184A1).

Similar overall results were seen with 184B5. However, unlike very early 184A1, early passages of 184B5 grow slowly, the mean TRF when first tested was ~3 kb, and the population already showed some heterogeneity in growth. Some cells capable of maintaining poor growth in TGFb were already present. Given this heterogeneity, we studied clonal isolates of early passage 184B5. Clones isolated at passage 15 showed a large range of growth capacity. Some didn't maintain any growth after 2-3 passages, some showed heterogeneous slow growth, and some had mixed slow and faster growth. Basically (1) clones that didn't maintain growth also showed no growth in TGFb, no or weak telomerase activity, and short mean TRFs, < 2.0-2.5 kb, with faint or very faint signals; (2) clones with slow growth behaved similar to the above description of 184A1 (from ~p20-25 onward). Repeated examination of the same clones repeatedly gave the same pattern of conversion.

One clone, B5Y16, was already heterogeneous for growth ±TGFb when first observed at p17. This extremely rapid generation of heterogeneity in a clonal isolate was further investigated by isolating subclones of B5Y16 at p20. These also unfolded the whole range of phenotypes, from (1) little or no growth, (2) 184A1-like growth, to (3) a few clones that displayed an already fully immortal converted phenotype. B5Y16 and its subclones demonstrate that a cell population obtained after less than 10 PD from one conditionally immortal cell may be widely heterogeneous. A good illustration of the inherent heterogeneity in growth response to TGFb and the gradual nature of conversion, can be seen with B5Y16G cells, a subclone (isolated at passage 20), of a clone (B5Y16, isolated at passage 15) of a clonal cell line (184B5) (see Figure 4A.2and Table 3). Although growth was slow and non-uniform when first observed at passage 21, by passage 24 rare colonies with good growth ±TGFb were present. B5Y16G was seeded at clonal densities at passage 25 and examined for growth ±TGFb. Heterogeneity was clearly visible in and among these single cell outgrowths. By p38, all B5Y16G cells gave rise to good growing colonies ±TGFb. These data with the 184B5 clones and subclones are inconsistent with a rare mutational origin of the converted phenotype.

In both 184A1 and 184B5 we have observed some instances of early conversion to full immortality (e.g., 184A1-TP; B5Y16G-bR). In 184A1 rare early converters rapidly take over the very slow growing non-converted population. In all cases, the mean TRF of early converters when initially examined was short (2.0-2.7 kb), and there was a correlation between ability to grow in TGFb and telomerase activity. These short mean TRFs indicate that newly converted cells arise from cells with critically short mean TRFs.

Table 3: Growth and LI of 184A1 and 184B5 colonies at different
passage levels in the absence or presence of TGFb

LABELING INDEX (%)

Cell Type	Pass #	TGFb	<10	10-25	26-50	>50	# Colonies
184A1	28	-	0	12	53	35	47
	32	-	12	26	17	55	95
	38	-	10	10	12	68	389
	44	-	0	2	7	91	272
	28	+	100	0	0	0	34
	35	+	59	37	4	0	83
	44	+	3	11	11	75	85

184A1-TP	28	-	0	0	0	100	17
	28	+	12	11	22	55	42

B5Y16G	25	-	6	14	44	34	50
	31	-	0	6	0	94	51
	38	-	0	0	0	100	91
	25	+	71	8	13	8	189
	31	+	20	28	21	31	102
	38	+	0	0	0	100	135

B5Y16G-bR	26	-	0	0	0	100	14
	26	+	0	10	9	81	17

Legend for Table 3: Single cells were seeded and the LI ±TGFb in colonies containing >50 cells was determined as described in Stampfer et al., 1997. 184A1-TP and B5Y16G-bR represent populations derived from isolated, early converting cells. 184A1-TP appeared in a slow growing conditionally immortal 184A1 population at passage 23, distinguishable by its much more rapid growth. B5Y16G-bR was derived from a rare colony that grew well in TGFb at p24 from the B5Y16G subclone of the B5Y16 subclone of 184B5. The data in this table also indicate that the phenotype of uniform good growth minus TGFb is acquired prior to that for good growth in the presence of TGFb. At least 45 colonies were counted to determine percentage labeling index. ND = not determined; TFTC = too few colonies to count.

(Click here to see figures 4A.2a-e)

Figures 4A.2a-e: Heterogeneity of colony growth in TGFb of subclone B5Y16G at passage 25.

1000 cells were seeded into 100 mm dishes and exposed to 5 ng/ml TGFb at15 days after seeding. Cells remained in TGFb an additional 18 days and were labeled with 3H-thymidine for the last 24 hrs. The Giemsa stained, single-cell derived colonies shown are all from the same dish.

(a) colony with no growth in TGFb;

(b) mostly flat colony with rare scattered labeled cells;

(d) colony with larger growing areas of small cobblestone cells amidst flat cells;

(e) rare large colony with uniform good growth in TGFb.

IV.A.3. Expression of the CKI p57^KIP2 in Conditionally Immortal HMEC and Loss of p57 Expression in Fully Immortal HMEC

(references: Nijjar et al., 1999; Yaswen & Stampfer 2001, Stampfer, Garbe et al., 2003)

Index

The presence of a slow growth phase in the conditionally immortal HMEC led us to examine the expression of molecules inhibitory to growth. Our first candidates were CKIs, particularly, p27^KIP1. Serendipitously, the antibody to p27 cross-reacted with p57^KIP2, and indicated that changes in p57 expression were associated with conditional immortality. I summarize the results and then present more details below.

We have examined expression of p57 mRNA (by Northern blots) and protein (by western blots). Commercially available p57 antibodies are not ideal. We have frequently seen cross-reacting proteins around the region of the p57 protein standard in situations where no p57 mRNA expression is seen. These proteins have slightly different mobilities than the protein bands seen in cells where p57 mRNA is expressed. We have found that p57 mRNA levels are generally indicative of protein levels. Consequently, in some situations, mRNA analysis is a more accurate method of p57 detection in these HMEC. Based on mRNA and protein analysis we have found:

1) No or very low levels of p57 is detected in our finite lifespan HMEC, including normal MM-grown and post-selection HMEC, and the EL cultures 184Aa and 185Be. (Figs. 4A.3ab; 4A.4a)

2) p57 is detected in G0-arrested conditionally immortal HMEC which have wild-type p53. In early passage good growing 184A1 (mean TRF > 3 kb), this p57 is downregulated in early G1 following EGF stimulation. (Figs. 4A.3ac; 4A.4b)

3) p57 is expressed in both G0-arrested cells, and cells released into cycle, in conditionally immortal p53(+) HMEC during the slow heterogeneous growth phase (mean TRF < 3 kb). (Figs. 4A.3ab; 4A.4b)

4) p57 expression in G0 and cycling populations is gradually reduced as the conditionally immortal cells gradually convert to a fully immortal phenotype. (Figs. 4A.3ab; 4A.4b)

5) Ectopic expression of p57 in good-growing early passage 184A1 recapitulates the morphology and slow heterogeneous growth seen when these cells "hit-the-wall" around passage 16.

6) p57 doesn't get expressed in cycling populations, and there is reduction of pre-existing G0 levels of p57, if the mean TRF in the conditionally immortal cells is kept above ~ 3 kb - e.g., by transduction of exogenous hTERT or induction of endogenous hTERT through inhibition of p53 function (see Fig. 4C.4).

7) The very severe growth constraint often encountered by 184A1 during passages 16-20 (Fig. 4A.3d) may be associated with loss of heterozygosity (LOH) for the maternal p57 allele, followed by transient expression from the paternal allele. (Fig. 4A.5) In 184B5, where the slow growth phase does not show such a severe "hit-the-wall" period, there is no LOH detected.

p57 belongs to the CIP/KIP family of CKIs, which includes p21, p27, and p57. These CKIs can inhibit multiple G1 kinases through binding CDK-cyclin complexes. The p57 gene has been localized to chromosome 11p15.5, a region displaying frequent allelic loss in cancers of the breast, lung, and bladder, as well as rare pediatric tumors such as Wilms' tumor. LOH and microsatellite instability at 11p15 have been associated with rapid proliferation, DNA aneuploidy, and poor prognosis in primary breast tumors. The p57 gene has been found to be imprinted with preferential expression of the maternal allele, suggesting that loss of the maternal allele by itself may severely reduce p57 expression. Germline mutations in the p57 gene have been detected in some patients with Beckwith-Wiedemann syndrome, a familial cancer-prone syndrome associated with hyperplastic growth in numerous tissues and a 1,000-fold increase in the risk of childhood tumors. Genetically engineered mice lacking the p57 gene have a variety of developmental defects consistent with Beckwith-Wiedemann syndrome, and indicate a role for p57 in control of cell proliferation and differentiation. p57 mRNA is detectable in most normal adult tissues, with highest levels in tissues consisting primarily of post-mitotic cells. In epithelia, p57 is reported to be expressed in regions of differentiated cells, but not in regions of actively dividing cells, suggesting that p57 may be up-regulated when cells exit the cell cycle and start their differentiation programs. p57 is not detectable in most immortal cell lines.

Figure 4A.3 shows p57 protein expression in G0 arrested (see Section V for methods of G0 arrest) and randomly cycling cultures of finite lifespan, conditionally immortal, and fully immortal HMEC. 184A1 conditionally immortal cells accumulate high p57 protein levels in G0 (Fig. 4A.3a) compared to the finite lifespan and fully immortal HMEC. Decreased p57 accumulation in G0 was observed around the passage levels at which 184A1 converts to the good growing, telomerase(+), TGFb resistant phenotype. In synchronized populations of early passage good growing conditionally immortal 184A1, the high level of p57 expression seen in G0 is downregulated between 4 and 12 hours following release from G0 arrest (Fig. 4A.3c); consequently, randomly cycling early passage good growing 184A1 (passages 13-15) did not show p57 protein expression (Fig. 4A.3b). Abundant p57 expression was first detected at the passage level (16p) corresponding precisely to where these cells demonstrated the onset of slow heterogeneous growth. p57 levels remained high in the randomly cycling population coincident with the period of slow heterogeneous growth (passages 16-38) and mean TRF levels < 3 kb. Similar results were obtained in conditionally and fully immortal 184B5 and 184AA4 cells, although in these lines, the conditionally immortal cells did not have a mean TRF > 3kb, and always showed some p57 in the cycling population. p57 expression was not seen in the finite lifespan or fully immortal cycling HMEC.

Click here to see figure 4A.3

These differences in p57 protein levels appear to be determined by changes in mRNA abundance. Figure 4A.4 shows p57 mRNA levels in finite lifespan, conditionally immortal, and fully immortal HMEC. Fig. 4A.4A shows that the 1.7 kb p57 transcript was detectable in G0-arrested early passage conditionally immortal 184A1 but not in normal finite lifespan or fully immortal HMEC cultures. Analysis of synchronized cell populations following release from G0 arrest (Fig. 4A.4B) shows that good growing passage 13 184A1 downregulated p57 transcript levels 4-8 hours after G0 release. In fully immortal late passage 184A1, the p57 transcript was not detectable during G0 or at any stage of the cell cycle. Thus changes in mRNA abundance correlated well with both the accumulation of p57 protein in conditionally immortal HMEC, and the absence of p57 protein in finite lifespan and fully immortal HMEC.

Click here to see figure 4A.4

Altogether, these data indicate that the transformation from finite lifespan to conditional immortality in these HMEC was associated with accumulation of p57 protein during G0 arrest; however, the good growing conditionally immortal 184A1 (mean TRF > 3 kb) were able to downregulate p57 upon mitogenic stimulation and entry into G1, whereas the poorly growing conditionally immortal cells failed to downregulate p57 after release from G0, and the p57 levels remained high in the cycling populations. Conversion from poor heterogeneous to uniform good growth was associated with loss of p57 expression. A possible implication of these data that we are currently investigating is whether one, or a combination of the errors that permitted these cells to overcome agonescence and become conditionally immortal, may also lead to the expression of p57 during G0 arrest.

Of the known CDK inhibitors examined (p15, p16, p21, p27, and p57), only p57 showed this dramatic increase in abundance corresponding precisely to the slowdown in growth of conditionally immortal cells with critically shortened telomeres. We have been unsuccessful thus far in experimentally reducing or blocking p57 expression in order to directly demonstrate the role of p57 in this growth slowdown. However, we have been able to show that ectopic expression of the p57 gene into good growing passage 14 184A1 induces premature onset of slow heterogeneous growth, producing cells with a flattened, vacuolated appearance similar to uninfected184A1 at passage 17. The levels of p57 expressed and the LI's in the passage 14, p57 virus-infected and passage 17 uninfected cells were also comparable. These experiments provide direct evidence that, when expressed at levels comparable to those seen in conditionally immortal HMEC with critically short telomeres, p57 can function as a growth inhibitor in this cell system.

Exposure of early passage good growing conditionally immortal 184A1 to TGFb leads to complete cessation of growth. When G0-arrested passage 13 184A1 were exposed to TGFb upon re-entry into the cell cycle, p57 mRNA was still downregulated after release from G0. Therefore, p57 mRNA accumulation is not a general consequence of growth inhibition in these cultures. The impaired ability of fully immortal HMEC to growth arrest in TGFb is more likely to be due to changes in regulation of p27, since TGFb did prevent the downregulation of p27 protein after G0 release in these early passage 184A1, but not in the TGFb-resistant fully immortal 184A1.

We next looked to see if loss of p57 expression in fully immortal 184A1 was accompanied by genetic changes. Figure 4A.5 shows p57 LOH analysis in 184A1 mass cultures and subclones. PCR analysis of genomic DNA revealed that the proline-alanine repeat region of the p57 gene was polymorphic in finite lifespan 184 and early passage 184A1, since amplification of this region yielded two products of different size (Fig. 4A.5a). RT-PCR showed the lower band to be the allele primarily expressed in the early passage conditionally immortal cells. Genomic DNA from the passage 29 and later cells showed loss of the lower band. To determine whether this LOH represented outgrowth of a pre-existing subpopulation in the conditionally immortal cells, genomic DNA was harvested from individual colonies grown from single cells in passage 15, 16, 17, 18 and 25 cultures. Individual passage 15 clones showed partial loss of the lower band indicating that allele loss was a frequent event and not due to outgrowth of a rare cell (Fig. 4A.5b). The percentage of colonies exhibiting p57 allelic imbalance or complete LOH increased with increasing passage, indicating that allele loss occurred at different times in different cells. To reconcile the loss of the major expressed allele by passage 25 with p57 expression seen at passages 32 and 38, RT-PCR was performed using total RNA harvested at various passages. RT-PCR showed that after loss of the lower band, the upper, previously silent allele was transiently upregulated in conditional immortal 184A1 during the slow heterogeneous growth phase. After conversion to the fully immortal phenotype, the remaining allele was downregulated, but not lost. In fact, RT-PCR indicated that some p57 transcripts persisted in the 184A1 mass culture through p92 even though no p57 protein or mRNA was detected by immunoblot or Northern analysis after p38. Unlike 184A1, 184B5 did not undergo p57 LOH although it did show downregulation of p57 expression during conversion.

Click here to see figure 4A.5

The correlation between short mean TRF length and the failure to downregulate p57 expression upon exit from G0 in conditionally immortal 184A1 suggested that the shortening of telomeres below a critical length could be responsible for this failure. We were able to test this hypothesis as part of other ongoing studies on the effect of ectopic expression of hTERT and inhibition of p53 function (see Sections IV.D. and IV.C.3.).

Transduction of hTERT via retrovirus into good growing passage 12 184A1 (mean TRF > 3 kb) resulted in a rapid increase of mean TRF lengths to > 9 kb. In contrast to the control cultures, these cells never underwent the slow heterogeneous growth phase and p57 mRNA was never detected in the cycling populations. p57 levels in G0 arrested cells were reduced to undetectable levels by passage 27. Artificial lengthening of telomeres in the conditionally immortal HMEC was therefore associated with continued good growth and abrogation of p57 expression. Similarly, transduction of the p53-inhibiting GSE22 into passage 12 184A1 resulted in a rapid induction of telomerase activity, a mean TRF which did not go below 4 kb, absence of p57 expression in cycling populations and reduction of existing G0 p57 levels, and the absence of a prolonged slow heterogeneous growth phase. These data support the hypothesis that the failure to down-regulate p57 upon exit from G0 arrest is triggered by the short mean TRF length (~3 kb) attained by conditionally immortal p53(+) HMEC.

Collectively, these results suggest that p57 is expressed by certain cultured cells when growth constraints associated with agonescence have been compromised; i.e., conditionally immortal cells. The data suggest that p57 may play an important role in the observed slow heterogeneous growth of this cell population, inhibiting the conversion of conditionally immortal cells to the fully immortal state. The absence of p57 expression in almost all human cell lines is consistent with the downregulation of p57 in the fully immortal HMEC lines.

IV.A.4. Abrogation of Activated Raf-1-induced Growth Inhibition during Conversion

(reference: Olsen et al., 2002)

Index

Overexpression of activated forms of Ras and Raf can induce malignant progression in a variety of immortally transformed rodent and human cell lines, but induces cessation of growth, associated with a phenotype resembling senescence, in finite lifespan human or rodent cells. This differential effect suggests that many immortally and malignantly transformed cells have undergone a significant alteration in their response to overexpressed Raf as a consequence of transformation from the normal finite lifespan state. In vivo, the growth inhibitory effect of activated Ras and Raf in finite lifespan cells may function as a tumor suppressor mechanism to halt proliferation in the presence of abnormal signaling, while loss of this mechanism may contribute to deregulated growth. Initial reports on Ras/Raf-induced growth arrest of finite lifespan human fibroblasts and primary rodent fibroblasts suggested that this "premature senescence" proceeded through induction of p16 and/or p21. Primary rodent fibroblasts lacking either p53 or p16/p19^ARF are immortally and tumorigenically transformed in response to oncogenic Ras or MEK. Normal human fibroblasts, however, continue to exhibit growth arrest in response to oncogenic Ras when either the p53 or p16/pRB pathway is disrupted.

We first sought to determine whether finite lifespan p16(-) HMEC would still exhibit growth arrest when exposed to activated Raf-1, and if immortally transformed HMEC could escape Raf-induced growth arrest. We assayed the effect of oncogenic Raf-1 using an inducible system - a retroviral expression vector encoding a chimera consisting of the catalytic domain of human Raf-1 fused to the hormone-binding domain of the human estrogen receptor (Raf:ER). Control were infected with a retroviral vector encoding only the hormone-binding domain of the human estrogen receptor (hbER). The Raf-1 fragment used in these experiments had two adjacent tyrosine residues (Y340 and Y341 [YY] in the full-length Raf-1) mutated to aspartic acid [DD], causing increased Raf-1 activity. The Raf:ER protein possesses little or no kinase activity until induced with estrogen or its analog 4-hydroxy-tamoxifen (4-HT). I summarize the results and then present more details below.

1) Both pre-stasis and post-selection Raf:ER-HMEC are severely growth inhibited upon activation of Raf. The mechanism of this inhibition has not been determined, but it does not require p16. Growth-arrested pre-stasis HMEC showed reduced p16 levels and post-selection HMEC remained p16(-). Isogenic fibroblasts arrested with increased p16 levels. (Figs.4A.6,4A.7)

2) The growth arrest did not require functional RB or p53, was unaffected by overexpressed c-myc, and was not associated with changes in expression levels of p14, p15, p21, p27 or p57. Arrest occurred in both G1 and G2.

3) Raf:ER induction inhibited the growth of both hTERT-immortalized HMEC, as has been reported for hTERT-immortalized fibroblasts, and pre-conversion conditionally immortal HMEC. However, all HMEC populations that had undergone conversion were able to maintain growth following Raf:ER induction and were now capable of EGF-independent growth and low levels of anchorage-independent growth. Thus an alteration in cellular physiology can occur to allow HMEC to maintain growth in the presence of transduced oncogenic Raf-1 - as a result of the process of conversion, and not as a consequence of overcoming agonescence or expressing telomerase activity.

click here to see figure 4A.6

Legend for Figure 4A.6: Post-selection HMEC strain 48R was transduced at passage 9 and assayed at passage 15. The cells were transduced with hbER (control; A,B) or Raf:ER (C,D) retroviral vectors and treated with ethanol (A,C) or 100 nM 4-HT (B,D). Scale bar, 100 mm. In E), counts of cells treated with ethanol or 10-500 nM 4-HT for 6 days, or colony-forming efficiency (CFE) of cells pretreated with ethanol or 100 nM 4-HT for 6 days at passage 16, then seeded at colony-forming density and allowed to grow for two weeks, are plotted as a percentage of control (ethanol-treated) cell counts. For CFE, colonies of >50 cells were counted. (F) shows the expression of cell cycle regulators in post-selection 48R-hbER and -Raf:ER. Cells at passage 15 were treated with ethanol or 100 nM 4-HT for 48 hours, then expression of the indicated cell cycle regulators was determined by Western blot (p16, p21, p53, p27, p57, MEK phosphorylation), Northern blot (cyclin D1, p15), or quantitative RT-PCR (p14).

click here to see figure 4A.7

Legend for Figure 4A.7. Pre-stasis 48R HMEC (A) or 48R fibroblasts (B) were transduced with Raf:ER retroviruses at passage 2 or 6, respectively, then treated with ethanol or 100 nM 4-HT for 6 days. C) Cell counts and SA-b-gal staining of 48R HMEC or 48R Fb were determined after 6 days' treatment with ethanol or 100 nM 4-HT. D) Relative expression of the indicated cell cycle regulators was determined by Western blot after treatment with ethanol or 100 nM 4-HT for 48 hours. Scale bars, 100 mm.

Figures 4A.6 and 4A.7 show the changes in morphology, growth, and expression levels of cell cycle regulators, following Raf-1 induction in post-selection and pre-selection HMEC, and isogenic fibroblasts. HMEC inhibited by Raf-1 induction showed enlarged cells with an elongated, flattened, vacuolated morphology (Figs. 4A.7d&4A.8a). There was virtually no net increase in cell numbers in cells treated with 100-500 nM 4-HT for 6 days, and almost no ability to form colonies even after removal of the 4-HT (Figs. 4A.7e&4A.8c). Indicators of metabolic function (MTT) and membrane integrity (trypan blue) indicated that nearly all the cells remained viable, while TUNEL assays revealed no evidence for apoptosis. Expression of SA-b-gal, which has been correlated with Raf-induced growth arrest in human fibroblasts and astrocytes, was not increased in post-selection HMEC upon induction of Raf:ER. In the pre-stasis HMEC, the percentage of SA-b-gal-expressing cells in Raf-arrested populations was difficult to determine because the clumping together of cells prevented accurate scoring of individual cells. All cellular masses examined in Raf-arrested cultures appeared to express SA-b-gal, while uninduced control cells exhibited a combination of proliferating, SA-b-gal-negative, as well as senescent, SA-b-gal-positive cells. SA-b-gal was induced in the Raf-arrested mammary fibroblasts (Fig. 4A.7c).

Western blot analysis revealed no p16 expression in Raf-arrested post-selection HMEC. In pre-stasis HMEC, p16 protein levels decreased relative to uninduced control cells, while p16 protein was induced in Raf-arrested mammary fibroblast (Figs. 4A.6 & 4A.7). p21 protein levels remained unchanged in the induced HMEC and moderately increased in the fibroblasts. p53 protein decreased in all induced cells. Levels of expression of other CKIs, p27, p57, p15 and p14^ARF, were unchanged in the induced post-selection HMEC. These results indicate that finite lifespan HMEC possess a p16-independent mechanism for Raf-induced growth arrest, and that the arrest does not require increased expression of the CKIs whose increased expression is associated with fibroblast senescence. mRNA levels of cyclin D1, which plays a facilitating role in cell cycle progression, but which also increases during senescence were found to increase in Raf-arrested HMEC.

Experiments performed largely with primary mouse embryo fibroblasts containing targeted deletions, have led to the general hypothesis that Ras- or Raf-induced growth arrest is dependent on the function of p53 and the p16/pRB pathway. However, in human fibroblasts, ablation of p53 and p21 by viral oncogenes fails to abrogate Raf-induced arrest. We transduced Raf:ER-expressing post-selection HMEC with retroviruses encoding viral oncogenes known to inhibit p53 and/or pRB function: HPV16-E6 and -E7, adenovirus E1A, and SV40 large T antigen. Expression of these oncogenes, as well as disruption of p53 and pRB expression and/or function was verified by immunological and/or functional assays. None of these viral oncogenes was able to prevent Raf:ER-induced growth arrest. Overexpression of c-myc via retrovirus-mediated transduction was also unable to rescue post-selection 48R from arrest.

Legend for Figure 4A.8. 184A1-Raf:ER at passage 13 (pre-conversion conditionally immortal, A) or passage 71 (fully immortal, B), and 184B5-Raf:ER at passage 16 (in conversion conditionally immortal, C) or passage 48 (fully immortal, D) were treated with ethanol or 100 nM 4-HT for 6 days, then photographed, counted, or seeded at low density to determine CFE. (E) The p53(-/-) cell line 184AA2-Raf:ER at passage 57 was treated as described, then photographed or counted. Cell counts and CFE are plotted as percent of control (ethanol-treated) cells. Scale bar, 100 mm.

click here to see figure 4A.8

We next utilized our immortally transformed HMEC lines to determine whether indefinite lifespan HMEC had altered responses to oncogenic Raf, and if so, at what point in the process of attaining full immortality such changes occurred. Figure 4A.8 shows the changes in morphology and growth following Raf-1 induction in immortally transformed HMEC. Raf induction in conditionally immortal 184A1 at passage 12 produced severe growth suppression, lack of colony forming ability, and a morphology similar to Raf:ER-arrested post-selection HMEC (Fig. 4A.8a). However, when fully immortal 184A1 at passage 69 was induced, the morphology appeared similar to the uninduced controls, and some growth and colony forming was maintained, although the cell numbers were reduced (Fig. 4A.8b). Similar studies were performed in p53(+) 184B5 (Fig. 4A.8cd) and p53(-/-) 184AA2 (Fig. 4A.8e). 184B5 transduced at passage 12 (in conversion, heterogeneous) showed a heterogeneous response when Raf was activated; some cells express a flattened, elongated appearance while others express the more normal cobblestone morphology. Some cells were able to maintain growth in the presence of activated Raf, although overall growth was significantly inhibited. Fully immortal 184B5 transduced with Raf:ER at passage 46 showed less growth inhibition, and no reduction in colony forming efficiency when Raf was induced. 184AA2 maintained growth after induction of activated Raf. Altogether, these experiments indicate that the alteration(s) which allowed immortally transformed HMEC lines to gain resistance to Raf-induced growth inhibition occurred after conversion to full immortality, rather than as a consequence of overcoming agonescence.

To determine whether acquisition of telomerase activity itself could confer resistance to growth arrest by activated Raf, post-selection HMEC immortalized by transduced with hTERT at passage 11 were infected at passage 25 with the Raf:ER or hbER. Figure 4A.9 shows the changes in morphology and growth following Raf-1 induction in hTERT-transduced HMEC. Raf:ER induced cells showed growth arrest, very low CFE, and a morphology comparable to that observed in Raf-induced post-selection HMEC (Fig. 4A.9a). These results demonstrate that hTERT-induced immortalization, which bypasses agonescence and conversion, does not confer resistance to growth arrest in the presence of the inappropriate Raf:ER activity. To determine whether another aspect of conversion was necessary for acquisition of resistance to Raf-induced growth arrest, we examined conditionally immortal 184A1 transduced with hTERT prior to and during conversion. Conditionally immortal pre-conversion 184A1 transduced with hTERT at passage 12 bypass conversion and the associated period of slow heterogeneous growth (see section IV.D). However, hTERT transduction into 184A1 at passage 22, when conversion has already begun, does not affect growth, and the population gradually proceeds through conversion. The 184A1(12p)-hTERT-Raf:ER population, which had bypassed conversion, displayed morphological changes after Raf induction similar to those observed in post-selection and hTERT-immortalized HMEC, although the suppression of growth was not as severe (Fig. 4A.9b). In contrast, the 184A1(22p)-hTERT-Raf:ER induced population displayed little change in morphology and were able to maintain growth (Fig. 4A.9c).

click here to see figure 4A.9

Legend for Figure 4A.9. (A) 184-hTERT-Raf:ER at passage 27 or 29 were treated with ethanol or 100 nM 4-HT for 6 days, then photographed, counted, or seeded at low density to determine CFE. Counts and CFE are plotted as percent of control (ethanol-treated) cells. In addition, conditionally immortal 184A1 transduced with hTERT at passage 12 (B; pre-conversion) or 22 (C; in-conversion) were transduced with Raf:ER and treated with ethanol or 4-HT for 6 days, then photographed and counted. Scale bar, 100 mm.

Altogether, these data support the hypothesis that obtaining the ability to maintain growth in the presence of overexpressed oncogenic Raf results from changes that occur during conversion, rather than from achieving an indefinite lifespan simply through the acquisition of telomerase activity. The data also show that the presence of transduced hTERT did not prevent the conversion-associated alteration in response to activated Raf.

To determine whether activated Raf was capable of affecting malignancy-associated pathways in immortal HMEC, assays for growth factor and anchorage independence were performed. Finite lifespan HMEC are completely dependent upon EGF receptor signal transduction for growth, as are our fully immortal HMEC lines (see section V). EGF-independent growth is an additional level of derangement exhibited by many cancer cells. Fully immortal 184A1-Raf:ER and 184B5-Raf:ER exhibited some growth in the absence of EGF receptor signal transduction when induced. Fully immortal 184A1, which has no ability to form colonies when suspended in methylcellulose, showed a low level of colony formation when induced. These results indicate that the activities of activated Raf that contribute to malignant progression were retained in fully immortal HMEC that were not growth-inhibited.

Preliminary studies suggest that the conversion-associated change in response to Raf:ER is accompanied by altered regulation of Raf:ER expression. We first noticed that fully immortal 184A1 and 184B5 showed lower levels of the EGFP-Raf:ER fusion protein than finite lifespan or conditionally immortal HMEC. Fluorescence activated cell sorting (FACS) was used to select subpopulations expressing the highest levels of the fusion protein. The selected cells were then both re-assayed for growth in the presence of inducer and re-sorted to determine the stability of high EGFP-Raf:ER expression. Sorted cells retained the ability to grow, but several passages after sorting, EGFP-Raf:ER expression had reverted to the lower levels seen in the unsorted population. We then theorized that the converted cells were able to keep expression levels of the Raf:ER reduced to levels in a growth-stimulatory range, whereas the levels in non-converted HMEC were at a much higher cytostatic range. Post-selection Raf:ER-HMEC were then grown in very low levels of 4-HT resulting in reduced EGFP-Raf:ER levels; these cells now also exhibited EGF-independent growth.

The Raf:ER experiments are potentially important because they suggest the existence of an alternative mechanism of growth arrest in finite lifespan HMEC that does not require intact p53 or pRB function, p16 expression, or an increase in the levels of p21. Although much remains unexplained about this Raf-induced growth arrest, we are intrigued that the change in response to oncogenic Raf occurs not with overcoming agonescence or acquiring of telomerase activity, but as a consequence of conversion to full immortality. Converted cells may have an altered ability to regulate transcription.

IV.A.5. Other Changes Associated with Conversion

[under construction]

Index

IV.A.6. The effects of viral oncogenes on conversion of 184A1

(reference: Garbe et al., 1999)

Index

A gradual conversion process had not been previously reported during human epithelial cell immortalization. Most reported studies of human epithelial cell transformation in culture have used viral oncogenes and/or inactivation of the p53 gene to facilitate more consistent and efficient immortalization. The overall process of immortalization described above for our chemically transformed immortal HMEC lines did not correlate with the reported descriptions of immortal transformation following viral oncogene exposure. We wondered if the potential of these viral oncogenes to simultaneously inactivate many cellular checkpoints was producing different patterns of immortalization than would occur in the generation of minimally deviant immortal cell lines with wild type p53 and RB. We therefore examined the consequences of exposing early passage (12p) conditionally immortal 184A1 to specific viral oncogenes: HPV16 -E6, -E7, SV40T, or Ad5 E1A.

The brief summary: Exposure to HPV16-E6 resulted in near immediate conversion to the fully immortal phenotype (good growth ± TGFb, high telomerase activity, stabilized telomere length). Exposure to HPV16-E7 and SV40T greatly accelerated acquisition of some conversion phenotypes (uniform good growth, high telomerase activity, stabilized telomere length) and led to the immediate ability to maintain some growth in TGFb. Ad5 E1A caused what appeared to be massive apoptosis, though some cells survived and rapidly converted. A mutated HPV16 E6 oncogene unable to inactivate p53 was still capable of near immediate conversion of passage 14 184A1. We conclude that the multiple activities of these viral oncogenes (inactivation of p53 and RB, telomerase reactivation, and many other characterized and as yet uncharacterized additional functions) may greatly accelerate a step in HMEC immortal transformation - conversion - that might otherwise be a key rate-limiting step. The ability of these oncogenes to simultaneously inactivate many cellular checkpoints is likely responsible for their capacity to achieve reproducible immortalization of HMEC and other human cells. More details below.

Figures 4A.10A & 4A.10B and Table 4 illustrate the effect of retroviral infection of passage 12 184A1 with vectors containing the indicated viral oncogenes, or the LXSN control vector.

click here to see figure 4A.10A

click here to see figure 4A.10B

Legend for Figures 4A.10. Good growing 184A1 at passage 2 were infected with retroviral vectors containing the HPV16 E6, HPV16 E7, or SV40T genes, or control LXSN-based retrovirus. Infected cells were obtained following selection in G418 for 10 days. The infected (184A1-E6, 184A1-E7, 184A1-T) and control (184A1-LXSN) cultures were then maintained in MCDB 170 medium with periodic assays for telomerase activity, mean TRF length, and growth ± TGFb as described in Stampfer et al 1997, and the Figure 8 legend, however the scales for telomerase activity and resistance are altered compared to Fig. 4A.1.

(10A) data for 184A1-E6

(10B) data for 184A1-E7 and 184A1-T

Table 4: Growth and LI of retrovirally infected 184A1 colonies at different passage levels in the absence or presence of TGFb

LABELING INDEX (%)

Virus/Passage TGFb (-) TGFb (+)

<10 10-25 26-50 >50 <10 10-25 26-50 >50

LXSN

13 0 0 0 100 ND

15 0 1 7 92 ND

17 15 26 26 33 100 0 0 0

21 14 22 29 35 100 0 0 0

23 1 14 43 42 96 4 0 0

28 2 10 48 40 100 0 0 0

HPV16-E6

13 0 0 0 100 14 11 39 36

15 2 0 0 98 0 2 11 86

17 0 0 0 100 0 0 0 100

HPV16-E6JH26

17 0 0 0 100 0 0 26 74

HPV16-E7

14 1 0 5 94 26 35 27 12

15 0 0 11 89 15 51 30 4

17 34 24 42 0 34 38 28 0

21 5 7 11 77 22 44 29 5

25 0 12 33 55 23 33 32 1

29 0 0 0 100 6 31 42 21

SV40-T

13 0 0 0 100 12 9 31 48

14 1 0 2 97 ND

16 2 0 4 94 11 7 27 55

17 18 14 26 42 0 11 27 62

21 0 0 0 100 1 4 17 78

27 ND 0 0 0 100

Legend for Table 4: Single cells were seeded and the LI ±TGFb in colonies containing >50 cells was determined as described in Garbe et al., 1999. ND = not determined.

The 184A1-E6 culture showed high levels of telomerase activity when first assayed at passage 12, and at all passages examined thereafter. Unlike the control cultures and consistent with its expression of high telomerase activity, 184A1-E6 maintained a mean TRF of ~5 kb with continued passage. 184A1-E6 maintained uniform good growth, with no evidence of slow heterogeneous growth and was already capable of good growth in TGFb at passage13. At passage15, 184A1-E6 showed very low levels of p53 protein, presumably due to degradation of p53 by the ubiquitin-dependent proteolytic pathway.

We also tested the effects of several different mutated HPV16-E6 genes. Five mutants (HPV16 E6 Cys-63-Gly, Cys-63-Arg, Cys-63-Ser, Cys-106-Arg, and Trp-132-Arg; obtained from Vimla Band, Tufts U), with low or no binding to or degradation of the p53 protein, had no effect on 184A1 conversion, behaving like the vector alone control. Contrasting results were seen with the amino terminal HPV16 E6 mutant E6JH26 (obtained from Denise Galloway, U. Wash.), which has also been reported not to bind or target p53 for degradation and not to affect p53 transactivation, but which does bind the E6-associated protein E6-AP and has been shown to activate low levels of telomerase in finite lifespan human keratinocytes and mammary cells. 184A1-E6JH26 induced a similar near immediate conversion as wild-type E6, indicating that the E6 oncogene does not need the ability to inactivate p53 to induce a fully immortal phenotype.

Unlike 184A1-E6, 184A1-E7 showed no telomerase activity at passage 12. Low levels of activity were detected at passage 18 and increased thereafter. By passages 25-29, high levels of telomerase activity were present. 184A1-T exhibited an even more accelerated expression of telomerase activity. Very low levels could be detected at passages 12-15, and high levels were detected by passage 23. Consistent with the telomerase activity data, the mean TRF length of both 184A1-E7 and 184A1-T showed an initial decline from ~5 kb at passage 13 to faint signals of ~3.5 kb at passage 21, followed by stabilization of telomere length. Both 184A1-E7 and 184A1-T experienced some heterogeneous growth between passages 16-20 (see Table 4). 184A1-T populations showed the presence of large flat vacuolated cells similar to those observed in the 184A1 control populations during the period of slow heterogeneous growth. Many colonies with poor or non-sustained growth were visible at these passages. Uniform good growth was seen by passage 21 in 184A1-T and by passage 29 in 184A1-E7. The majority of 184A1-E7 and 184A1-T cells were able to maintain some (although not yet good uniform) growth in the presence of TGFb at the earliest passages tested, indicating that this capacity was rapidly conferred by the expression of the viral oncogene, even before detection of telomerase activity. This TGFb resistance could be related to the ability of these oncogenes to bind and inactivate p27, which has been associated with TGFb growth inhibition in our HMEC system and other cell types.

Subsequent studies (see section IV.C.) have shown that p53(-) cells rapidly convert to full immortality, showing early telomerase activity, and lack of p57 expression and the associated slow heterogeneous growth.phase. Therefore, our data on rapid telomerase reactivation following SV40-T transduction is consistent with the ability of SV40T to inactivate p53 function. Curiously, HPV16-E7 also accelerated telomerase reactivation. This result could be related to the recently described ability of HPV16-E7 to impair p53 function (Eichten et al.). Understanding how these viral oncogenes function may help elucidate the mechanisms of immortal transformation of human cells. However, we believe that systems of immortalization that employ viral oncogenes with pleiotrophic effects, some of which are still unknown, may not accurately model many aspects of human carcinogenesis in vivo. In particular, the common use of oncoproteins which inactivate p53 function has prevented understanding of the mechanisms by which p53(+) human cells overcome senescence, undergo conversion, and gain full immortality.

Previous reports had indicated that HPV16-E6 could induce low levels of telomerase activity even in finite lifespan human epithelial cells, so we examined our finite lifespan HMEC to determine whether HPV16-E6 could induce telomerase activity, and if so, whether it was dependent upon specific parameters of the HMEC population; i.e., growth conditions, age in culture, or telomere length.

Four different sets of finite lifespan 184 HMEC were examined: (1) pre-selection cells grown in MCDB 170; this population was infected at passage 2 when still proliferative, and assayed at passage 3 when it contained mostly poorly growing cells and had a mean TRF of ~8-7 kb; (2) post-selection cells grown in MCDB 170, infected and assayed at passage 9 and passage 20; both these populations had active cell division, mean TRF ~7 kb and 5.5 kb respectively; (3) EL 184Aa, the precursor of the 184A1 line, grown in MCDB 170, infected and assayed at passage 8 and passage 13; both populations had active cell division, mean TRF ~6 kb and 5.2 kb respectively; (4) Pre-stasis 184 HMEC actively growing in the serum containing MM medium, infected at passage 3 and assayed at passage 4; mean TRF between 8-8.8 kb when assayed at passages 3-6. As expected, no telomerase activity was detected in the pre- or post-selection 184 or 184Aa without the E6 oncogene. Introduction of the E6 oncogene resulted in low telomerase activity in the post-selection 184 passage 9 and in the 184Aa passage 8 and passage 13 cells. Repeated independent infections showed no telomerase activity in the near-senescent, but still proliferating, passage 20 post-selection 184 cells, nor in the poorly growing pre-selection passage 3 cells. In contrast to the cells grown in the serum-free MCDB 170 medium, early passage 184 grown in MM showed a low level of telomerase activity, even in the absence of the E6 oncogene. This low telomerase activity was further increased to a medium level after the introduction of E6. These data may bear on previous reports indicating that certain populations of cultured HMEC may have low levels of telomerase activity. It is possible that the growth conditions, or selection for growth of specific breast epithelial cell types in culture, can influence whether this activity is present in the uninfected as well as E6 infected populations. These results show no strict correlation between the level of telomerase expression induced by E6 and the age in culture or mean TRF of the cell population. We saw no correlation between proliferative capacity and the expression of telomerase activity in the finite lifespan HMEC, as the HMEC population with the most long-term proliferative capacity, the post-selection cells, consistently shows no telomerase activity. However, a caution - these results are specific for specimen 184. Preliminary testing of HMEC from other specimens suggests that there could be some interindividual variability. We have not had the resources to check this further.

The rapid induction of telomerase activity in passage 12 184A1 by HPV16-E6, as well as its ability to induce telomerase activity in finite lifespan HMEC with mean TRF values of 5-8 kb, suggest that it acts through a mechanism other than an epigenetic response to critically short telomeres. Telomerase activity is present in both tumor cells and cells with a high self-renewal capacity. In tumor cells and immortal cell lines, telomerase activity is associated with telomeres that are generally shorter than those found in finite lifespan cells However, telomerase positive cells with high self-renewal capacity do not necessarily exhibit shorter telomeres. Possibly, the mechanism by which HPV16-E6 reactivates telomerase is more closely related to the regulation of telomerase activity in normal telomerase positive cells than to the reactivation of telomerase which occurs during malignant progression.

IV.A.7. Telomerase activity, Telomere Length, and Growth in Fully Immortal 184B5

Index

Telomerase-expressing cells, from unicellular organisms such as yeast to human tumor-derived immortal cell lines, have been shown to maintain telomere length within a set range through regulation of telomerase access/activity and the conformation and protein composition at the telomere. The large majority of such lines maintain a short mean TRF of < 7kb. The short regulated telomere lengths observed in most human tumor-derived lines is consistent with a model of cells overcoming agonescence and undergoing conversion, followed by a mechanism to regulate the resulting short telomere length. We have done just a few experiments looking at telomere length regulation in our fully immortal, converted HMEC lines.

As is seen with tumor-derived lines, assays of our converted HMEC lines indicate that they maintain a short mean TRF of ~3-7 kb. Mass populations tend to have a TRF of ~4 kb, but individually cloned subpopulations show a wider range. This was explored using the 184B5 line. Uncloned 184B5 at passage 99, and five clones isolated at passage 96 were examined (Fig. 4A.11). The range of mean TRF lengths for these clones was 2.9 - 7.0 kb. To our surprise, the clone with the shortest mean TRF, B5Y9H, showed no detectable telomerase activity when first assayed at passage 99, although all these clones exhibited good growth ± TGFb at that passage. TGFb did induce morphologic changes in all five subclones. With continued passage, B5Y9H, but not the other four clones, reproducibly showed a slowdown in growth around passage 103, and an initial total loss of proliferation at passage 105. However, after a few weeks, some B5Y9H passage 105 cells began to give rise to large outgrowths. These cells were subcultured and maintained good growth until at least passage 116. Assay for telomerase activity indicated no or very weak activity up to and including the non-proliferative passage 105 population. After the passage 105 dishes displayed the large outgrowths, telomerase activity was detectable. The mean TRF length of B5Y9H hovered around 3.0 kb prior to passage 105, and increased slightly thereafter. These data indicate that telomerase activity may cycle off and on even in converted cells, perhaps accounting for the limited range of TRF lengths. Unlike conditionally immortal cells, reactivation of telomerase in B5Y9H occurred relatively rapidly, within one passage. Additionally, these converted cells differed from the conditionally immortal in their ability to exhibit TGFb resistance in the absence of detectable telomerase activity. Their mean TRF at the point of telomerase reactivation was also longer (~3 vs. ~2 kb). It is likely that the telomeric ends of fully immortal HMEC have undergone an irreversible change into a conformation and set of telomere-associated proteins that allow assessment of telomere length in order to maintain stable, short telomeres.

click here for figure 4A.11

Figure 4A.11: Mean TRF length and telomerase activity in late passage 184B5 and 184B5 subclones at different passages. Assays were performed as described in Stampfer et al, 1997, and the Figure 8 legend. For TRF length, lighter shaded ovals indicate a faint signal.

(A) 184B5 and subclones;

(B) B5Y9H cells at different passage levels from two separate freezedowns. The first telomerase assay at passage 105 was from dishes containing vacuolated, non-growing cells. The second assay for telomerase at passage 105, and TRF values, were obtained from sister cultures which contained good growing patches.

We have not performed further investigations in this area, although we think such work would be very valuable. We believe that minimally deviant immortal lines such as 184A1 and 184B5 would be the best type of cell substrates for examining control of telomerase activity in immortally transformed human epithelial cells. These lines do not harbor the potentially many unknown and confounding errors found in tumor-derived cell lines - which such studies have thus far used. Our data on this subject of telomerase/telomere length control, although quite limited, are of particular note since what we see in fully immortal converted HMEC thus far corresponds to what is seen in human tumor-derived cells. This contrasts with what is seen in cells immortalized with hTERT.

IV.B. Immortalization of HMEC with Breast Cancer Associated Oncogenes

IV.B.1. Immortalization of HMEC with the Putative Breast Cancer Oncogene, ZNF217

(references: Nonet et al., 2001; Stampfer & Yaswen 2001)

Index

More recent studies have used another potentially pathologically relevant means, the candidate oncogene ZNF217, to immortally transform finite lifespan HMEC. ZNF217 was originally identified based on its location on chromosome 20q13.2, an amplicon common in breast cancers and associated with poor prognosis. Extra copies of this chromosomal region occur in approximately 18% of breast tumors and 40% of breast cancer cell lines. ZNF217 encodes a conserved member of the C2H2 Kruppel family of of zinc finger proteins. Members of the Kruppel family have been implicated in both neoplastic and developmental disorders.

Transduction of post-selection 184 HMEC and EL 184Aa with ZNF217 generated rare-to-frequent clonally-derived lines during the period of agonescence. The ZNF217-transduced HMEC showed no initial growth advantage over the control cultures, but continued to grow beyond the agonescence barrier. Growth was at first slow and heterogeneous, but became faster and more uniform with continued passage. After ~5-15 passages, varying among experiments, most cells were SA-b-gal negative and grew well.

Telomerase activity was not initially detectable in the ZNF217-transduced 184 and 184Aa cultures that maintained growth past agonescence, and the mean TRF length continued to decrease (Fig. 4B.1). Telomerase activity was detectable within 10 passages and then gradually increased, and mean TRF length stabilized at ~4 kb. When assayed for growth in TGFb, ZNF217-transduced 184 and 184Aa were initially completely growth-inhibited prior to and just after overcoming agonescence. With increasing passage, there was a very gradual increase in the number of cells with progressively better growth capacity in TGFb. Assay for p57 showed some expression in G0-arrested and cycling populations in the earliest passages after overcoming agonescence, when growth was slow and heterogeneous, but none in the later good-growing cultures.

click here to see figure 4B.1

Legend for Figure 4B.1. Post-selection 184 HMEC were transduced with ZNF217 or LXSN vector alone at passage 10. Assay for TRAP activity and mean TRF length showed a gradual acquisition of telomerase activity and stabilization of telomere lengths in the ZNF217-transduced population.

Southern analysis of retroviral integration sites in ZNF217-transduced HMEC growing past agonescence suggested that these cultures were rapidly overgrown by distinct clonal populations. Comparative genomic hybridization (CGH) analysis of three different immortalized cultures showed low level regional DNA-sequence copy number variations on chromosomes 1q and 8q common to all three cell lines. The region amplified on 8q included the c-myc oncogene. In addition, each line showed unique regions of high and low level DNA-sequence copy number variations. These regional copy number variation sites, some of which have also been frequently observed in breast cancer cells, may have been produced at agonescence and could contain genes that cooperate with ZNF217 in facilitating immortalization.

To determine whether loss of p53 function contributed to the immortalization of the ZNF217-transduced HMEC, p53 function was assayed by measuring p53 expression after exposure to the DNA damaging agent actinomycin D, and p53-dependent induction of GADD45 transcripts following UV irradiation. Induction of p53 similar to that in the finite lifespan cells was observed in all three ZNF217-transduced immortalized HMEC tested, and GADD45 mRNA levels were increased 4 hrs. after UV exposure in both finite lifespan 184 and ZNF217 immortalized 184Aa. pRB was also present and underwent normal cycles of phosphorylation and dephosphorylation in these cells.

Thus, ZNF217-immortalized HMEC show many similarities to the immortal HMEC lines derived by carcinogen exposure. Alterations in p53 and/or RB were not obligatory for immortalization. Most significantly, these p53(+) immortal lines underwent a conversion process before attaining a fully immortal phenotype. There was a gradual reactivation of telomerase activity and an incremental gain of TGFb resistance, and a transient expression of p57 associated with a period of slow heterogeneous growth.

IV.B.2 Immortalization of HMEC with the Breast Cancer Associated Oncogene, c-myc

[under construction]

Index

IV.C. Generation of Immortal HMEC Lacking p53 Function

(reference: Stampfer, Garbe et al., submitted)

Index

IV.C.1. Generation of p53(-/-) HMEC lines 184AA2 and 184AA3

Index

Our studies on conversion raised the question of what was the very rare event, presumably a mutation, responsible for the transformation of finite lifespan EL cultures into conditionally immortal cell lines. In our system, the data indicated that it was not changes in p53, RB, p16 or telomerase expression - the usually identified or postulated suspects. To address this question, we decided to try the Genetic Suppressor Element (GSE) strategy. GSEs are cDNA fragments that could potentially interfere with normal gene functions by either encoding antisense RNA or inhibitory protein fragments. Assuming that immortality represents a loss-of-function, the hope was that disruption of this normal function by a GSE would enable us to identify gene products lost during immortal transformation. A library of small cDNA sequences obtained from normal 184 HMEC was introduced into p12 184Aa (the EL precursor of 184A1) using high titer amphotrophic retrovirus. Three different new immortally transformed cell lines designated 184AA2, 184AA3, and 184AA4 resulted. Unfortunately, none of these lines demonstrated loss of a gene responsible for generating 184A1, and so the nature of this step is still unknown. On the up side, we did generate three new immortal HMEC lines, two of which (184AA2 and 184AA3) are p53(-/-).

184AA2 and 184AA3 were detected in MCDB170-grown 184Aa at passages 13 and 14 respectively. 184AA2 first appeared as tight patches of refractile cells with many mitoses as well as larger flatter cells. It maintained good growth thereafter, consisting mainly of tightly packed rounded cells. 184AA3 first appeared as areas of small densely packed, grossly vacuolated and extremely slow growing cells. Mass cultures began a gradual increase in growth rate by passages 16-18, with colonies becoming less densely packed with fewer grossly vacuolated cells. When seeded at very low densities to permit visualization of the growth of individual colonies, these gradual changes in growth and morphology could be observed in single cell outgrowths. By passages 20-24, uniform good growth was attained, and cell morphology changed to more rounded and refractile, without the tightly packed colonial cell growth seen previously

PCR analysis indicated that 184AA3 had a single viral insertion, but the virus had no cDNA insert. Inverse PCR to isolate the flanking genomic DNA, and sequence analysis of the recovered clones, showed that the viral insertion was near the beginning of the first intron of p53. This was confirmed by Southern analysis of genomic DNA probed with p53 sequences. Southern analysis also showed that one of the three viral inserts in 184AA2 was in the p53 gene, and that no normal p53 alleles were present in either 184AA2 or 184AA3. Western blot analysis of 184AA2 and 184AA3 showed no p53 expression even after long exposure. RT-PCR also indicated that 184AA3 has no band corresponding to a p53 transcript. These data indicate that both these HMEC lines are p53(-/-).

The functional status of p53 was determined by assaying the ability of the individual cell lines to: (1) arrest growth after exposure to the microtubule destabilizing agent colcemid; (2) arrest growth after exposure to x-irradiation; (3) accumulate GADD45 mRNA following UV exposure. Figure 4C.1 and Table 5 show the effects of colcemid and UV exposure on the cell cycle of the four immortal lines derived from EL 184Aa. Fig. 4C.1A indicates that following 96 hr of colcemid exposure, p53(+) 184A1 and 184AA4 contained DNA contents corresponding mostly to 4N, whereas p53(-/-) 184AA2 and 184AA3 contained >4N; 184AA2 and 184AA3 also showed greater ongoing DNA synthesis as evidenced by the continued incorporation of BrdU. Fig. 4C.1B indicates that 48 hrs following x-irradiation, p53(+) 184A1 and 184AA4 arrested with DNA contents largely corresponding to 2N or 4N, with few cells in S phase. The presence of some tetraploid cells in 184AA4 cultures may account for some of the 8N population seen in these assays. In contrast, p53(-/-) 184AA2 and 184AA3 cells had peaks corresponding to 4N, but also showed broad distributions of DNA contents ranging from 2N to > 4N; 184AA2 and 184AA3 also maintained DNA synthesis following x-irradiation. Additionally, 184AA2 and 184AA3 showed no GADD45 induction, whereas the p53(+) HMEC exhibited 2-5 fold GADD45 induction 4-24 hrs following UV exposure (Fig. 4C.1C).

click here to see figure 4C.1

Legend for Figure 4C.1. (B,C). FACS analysis was performed following exposure to (A) 50 ng/ml colcemid and (B) 10 Gy of ionizing radiation. (C) Northern analysis of GADD45 mRNA expression following UV exposure. as measured by in samples harvested at the indicated times following UV exposure.

Table 5. Cell cycle distribution and DNA synthesis in p53(+) and p53(-/-) HMEC lines following exposure to colcemid or x-rays.

	Control		Colcemid 96 hr		Control		X-ray 48 hr
Cell Line	% total	% BrdU+	% total	% BrdU++	% total	% BrdU+	% total	% BrdU+
184A1
<2n	0.2	0.0	9.1	0.4	0.3	0.0	0.2	0.0
2n	66.9	7.2	5.5	0.2	64.8	6.8	41.0	0.3
2n>4n	17.5	16.8	21.6	0.5	18.0	17.3	1.0	0.3
4n	13.6	8.8	50.7	1.3	14.1	9.1	52.1	1.4
>4n	1.7	1.0	13.1	2.9	2.8	1.7	5.6	1.9
Total	100.0	33.8	100.0	5.2	100.0	35.1	100.0	3.9
184AA2
<2n	1.7	0.3	2.2	0.4	1.6	0.4	0.5	0.2
2n	57.7	4.1	2.2	0.3	56.2	4.7	9.8	1.3
2n>4n	22.6	20.2	10.6	0.8	22.7	20.0	15.5	8.6
4n	13.6	6.4	4.6	0.4	14.7	8.1	45.0	7.6
>4n	4.4	2.6	80.4	21.4	4.8	3.1	29.1	18.8
Total	100.0	33.7	100.0	23.2	100.0	36.3	100.0	36.4
184AA3
<2n	0.8	0.2	2.2	0.2	0.6	0.1	0.6	0.0
2n	45.0	3.4	4.4	0.2	50.1	3.9	9.5	0.2
2n>4n	36.7	35.6	7.3	0.5	27.4	26.3	29.0	9.0
4n	16.1	10.1	34.6	3.0	20.3	11.9	54.8	9.0
>4n	1.4	1.0	51.5	13.6	1.6	0.9	6.1	2.9
Total	100.0	50.3	100.0	17.4	100.0	43.1	100.0	21.1
184AA4
<2n	0.3	0.1	0.3	0.2	6.6	0.1	0.7	0.0
2n	59.3	5.5	4.3	0.3	55.3	6.6	25.5	0.7
2n>4n	13.5	12.0	3.8	0.2	14.1	13.1	3.0	1.6
4n	18.3	10.9	69.5	1.1	18.3	10.8	53.8	2.8
>4n	8.5	4.5	22.1	3.8	5.7	3.1	17.0	4.3
Total	100.0	33.0	100.0	5.6	100.0	33.7	100.0	9.3

Table 5 Legend: Randomly cycling populations were exposed to either colcemid (50 ng/ml) or x-rays (10 Gy) for the indicated times. Cells were labeled with 10 µM BrdU during the last 4 hours of incubation, then harvested and prepared for FACS analysis. DNA content was determined by propidium iodide staining. BrdU(+) cells were identified using a fluorochrome labeled anti-BrdU antibody.

Altogether, these analyses indicate that the p53(-/-) lines lack p53 functional activities, and that the stable p53 protein present in the p53(+) HMEC lines does not show checkpoint-arresting activities in the absence of activating stimuli such as irradiation or colcemid.

The appearance of the p53(-/-) lines 184AA2 and 184AA3 suggested that total loss of p53 might facilitate immortal transformation of 184Aa. However, we have since seen that loss of p53 function alone is not sufficient to immortalize 184Aa. Transduction of 184Aa with the p53-inhibiting GSE22 produced sporadic immortal lines during the period of crisis, suggesting (as in the case of ZNF217) that an error generated by the genomic instability of crisis was necessary in addition to loss of p53 function.

NOTE! NOTE! NOTE!: 184A1, 184B5, and 184AA4 are immortally transformed cell lines. 184AA2, and 184AA3 are immortally transformed cell lines that are also p53(-/-). Normal human somatic cells are never immortal, do not express high levels of telomerase, have wild type p53 and RB, and normal karyotypes up to agonescence. Any somatic cell which is immortal, or defective in p53 or RB, and aneuploid is not normal. I consider it grossly incorrect, and a potential cause of very serious scientific error, to refer to these lines, or any immortal cell line, as "normal" or "untransformed", even if they do retain many normal properties. The defining characteristics of human cancer cells in vivo include expression of telomerase, p53 defects, aneuploidy, and defects in the RB pathway. It's REALLY not OK to refer to cells with these characteristics as normal, or "normal" (as I sometimes see). They aren't normal, so calling them normal or "normal" is simply scientifically incorrect. Would one call a mutated p53 gene "normal"? Then why call a cell that contains only mutated p53 "normal"? This is by no means a trivial or picky point. If cells which have already acquired the defining, and potentially rate-liming steps of early stage cancer (although still not invasive) are called and thought of as normal, how are we ever going to understand the processes involved in the early stages of carcinogenic progression. Or how are we ever going to know what are normal cellular growth control processes if the cells being studied are already aberrant in their growth control properties. I believe that the common failure to recognize the important distinctions between cells that are truly normal and those which are often misrepresented as normal has allowed a serious flaw to permeate much of current "mechanistic" molecular biology in the areas of cell cycle regulation, cancer, and signal transduction.

IV.C.2. Attainment of Full Immortality in p53(-) HMEC

Index

Unlike the p53(+) lines, 184AA2 and 184AA3 rapidly attained full immortality. 184AA2 displayed good initial growth and maintained good growth thereafter; 184AA3 displayed very slow initial growth, but achieved good uniform growth by passages 20-24. These initial observations indicated that these p53(-/-) lines did not undergo a very gradual conversion process. To determine to what extent they did display conversion-associated properties, 184AA2 and 184AA3 were examined at different passage levels for telomerase activity, mean TRF length, growth capacity ±TGFb (Fig. 4C.2 and Table 6), and p57 expression (Fig. 4C.3).

184AA2 already contained strong telomerase activity when assayed at passage 17. Consistent with telomerase activity being present, the mean TRF length remained at a stable value of ~4 kb from passage 17 through passage 45. 184AA2 did show the conversion-associated aspect of gradual acquisition of the ability to maintain growth in TGFb. Although some cells were capable of maintaining growth in TGFb at the earliest passages tested, uniform good growth was not present until after passage 32. In 184AA3, weak to medium telomerase activity was present at the earliest passage testable, passage 17, and strong activity was present by passage 23. The mean TRF increased from ~3.5 kb at passage 17 to a stabilized value of ~4 kb by passage 23. Some poor growth in TGFb was present in passage 19 cells, but good uniform growth in TGFb did not occur until after passage 31. Unlike the three p53(+) lines, no p57 mRNA expression was detected in either the 184AA2 or 18AA3 lines at early or late passages, in G0 or in cycling populations.

The capacity for anchorage independent growth was already present in 184AA2, but not 184AA3, at passage 19. However, both 184AA2 and 184AA3 displayed anchorage independent growth when examined at passage 50. Continued passage of 184AA3 may have selected for rare, pre-existent anchorage independent cells, or promoted the generation of cells harboring this aberration.

Altogether, these data indicate that the behavior of early passages of both p53(-/-) lines significantly differed from that displayed by early passages of the p53(+) lines. The p53(-/-) lines expressed telomerase activity more rapidly, mean TRF lengths did not decrease below 3.5 kb, and they showed early acquisition of uniform good growth potential. They differed from the p53(+) lines in their total absence of both p57 expression and an extended period of poor heterogeneous growth. Both p53(-/-) lines resembled the p53(+) lines in showing a gradual acquisition of increasing number of cells with progressively better growth capacity in TGFb.

click here to see figure 4C.2

Legend for Figure 4C.2.

Panel A: Mean TRF length; lighter shaded ovals indicate a faint signal.

Panel C: Colony forming efficiency (CFE) and labeling index (LI) in colonies.

Mean TRF length, telomerase activity, CFE and LI were determined as described in Stampfer et al., 1997.

Panel D: Representative TRAP assay for 184AA4.

Panel E: Representative TRF assays for 184AA3 and 184AA4.

click here to see figure 4C.3

Legend for Figure 4C.3. p53(+) 184A1 and 184AA4 and p53(-/-) 184AA2 and 184AA3 were examined at the indicated early and late passages for p57 mRNA expression. Cells were arrested in G0 by removal of EGF from the medium and addition of the anti-EGF receptor antibody MAb225 for 48 hrs. Random cycling cells were fed 48 and 24 hrs prior to harvest. The bottom panel shows the ethidium bromide staining.

Table 6: Growth of 184AA2, 184AA3, and 184A1-GSE22 colonies at different passage levels in the absence or presence of TGFb

LABELING INDEX (%)

Cell Line/Passage TGFb (-) TGFb (+)

<10 10-25 26-50 >50 <10 10-25 26-50 >50

184AA2

19 0 0 1 99 12 29 35 24

24 0 0 0 100 0 13 41 46

32 ND 7 12 28 64

45 0 0 0 100 0 0 0 100

184AA3

17 TFTC TFTC

19 0 0 15 85 24 38 38 0

23 0 0 2 98 15 42 30 13

31 ND 14 21 34 43

37 0 0 0 100 0 0 0 100

184A1-Babe

14 15 37 30 18 94 6 0 0

17 16 20 30 34 91 6 2 0

20 32 8 32 29 65 23 11 0

184A1-GSE22

14 17 26 25 32 24 27 25 24

17 13 11 22 54 40 18 23 19

21 4 12 10 74 39 21 18 22

25 0 0 0 100 36 16 17 32

30 0 0 0 100 12 5 21 62

Table 2 Legend: 200-10,000 single cells were seeded per 100 mm dish, and the labeling index ±TGFb in the ensuing colonies which contained >50 cells was determined as described in Materials and Methods. At least 45 colonies were counted to determine percentage labeling index. ND = not determined; TFTC = too few colonies to count.

IV.C.3. Effect of Inhibition of p53 Function in p53(+) HMEC Lines

Index

The capacity of both p53(-/-) lines to rapidly attain full immortality, relative to the p53(+) lines, and their absence of p57 expression, suggested that these properties were the result of the lack of p53 function. To directly test this possibility, p53 function was inactivated in early passage 184A1 by transduction of the p53-inactivating GSE, GSE22. The inhibition of p53 function in the 184A1-GSE22 culture was demonstrated by the failure to fully arrest after exposure to colcemid or x-irradiation in 184A1-GSE22 compared to 184A1 vector-alone control.

184A1 transduced at passage 12 with GSE22 rapidly gained full immortality. Fig. 4C.4 and Table 6 show the effects of GSE22 on telomerase activity, mean TRF length, and p57 expression. As shown in Fig. 4C.4A, 184A1-GSE22 cultures displayed intermediate levels of telomerase activity seven days after the population was infected, and strong activity was present by passage 19. As expected, the control, vector alone culture, 184A1-Babe, showed no telomerase activity at passages 12 or 16. Consistent with the TRAP assay results, quantitative RT-PCR assay showed a >10 fold increase in hTERT mRNA expression within two passages of GSE22 transduction. Thus, inactivation of p53 function allowed rapid induction of hTERT and telomerase activity in conditionally immortal telomerase(-) HMEC. As shown in Fig. 4C.4C, the mean TRF length in 184A1-GSE22 showed a modest decline from ~5 kb at passage 13 to a stabilized length of ~4 kb by passage 25, similar to the stabilized length in 184AA2 and 184AA3. The TRF signal never became faint nor declined below 4 kb.

184A1-GSE22 did not undergo a prolonged slow growth phase (Table 6), and instead showed uniform good growth by passage 25. 184A1-GSE22 assayed up to passage 30 showed a gradual increase in the capacity to maintain growth in the presence of TGFb. No p57 protein was expressed in the 184A1-GSE22 randomly cycling population at passages 14-22, in contrast to the abundant p57 at passage 14 in the 184A1-Babe control population (Fig. 4C.4D). Very low levels of p57 compared to 184A1-Babe could be seen in the G0-arrested cells. These results suggest that p57 expression in conditionally immortal HMEC may be dependent upon expression of functional p53. The absence of p57 may in turn be responsible for the absence of a prolonged conversion-associated slow growth phase.

This acceleration of the conversion process in cells lacking functional p53 may account for why the conversion process was not previously described. Most HMEC immortally transformed in vitro, as well as most immortally transformed epithelial cells from other human organ systems, have been immortalized using methods that obligately inactivate p53 function (e.g., exposure to the SV40-T or HPV-E6 oncogenes, or dominant negative p53 mutants, or use of cells derived from Li-Fraumeni patients.

click here to see figure 4C.4

Legend for Figure 4C.4. Representative (A) TRAP, (B) Quantitative RT-PCR for hTERT and (C) mean TRF results are shown for the184A1-Babe control or 184A1-GSE22 cell lines. Values at the bottom of (C) are mean TRF lengths calculated as described in Stampfer et al. submitted; (D) An immunoblot of total cell extracts probed with a specific anti-p57 antibody shows that p57 protein is downregulated in184A1-GSE22 during G0 compared to the control cells, and that p57 protein is not expressed in cycling populations.

Altogether, these data indicating that an absence of functional p53 can directly accelerate and/or circumvent the conversion process, could have implications for the clinical course of p53(-/-) tumors. Only 15-30% of primary human breast cancers have been shown to contain mutations in the p53 gene, but these cancers have a significantly worse prognosis than those containing wild type p53. Our data showing that p53 loss is directly responsible for rapid attainment of full immortality, and thus more rapid attainment of aggressive growth potential, is consistent with the in vivo data associating p53 loss with more aggressive cancer. If a conversion process occurs in breast cancer development in vivo, our results suggest a significant additional mechanism whereby p53 loss may contribute to more aggressive cancer progression.

The rapid expression of telomerase activity in 184A1 transduced with GSE22 provides direct evidence that p53 plays a role in suppression of telomerase activity. However, our preliminary data showing that introduction of GSE22 does not activate telomerase activity in post-selection 184 or EL 184Aa indicates that abrogation of p53 expression/function alone is not sufficient for expression of telomerase activity. It is possible that the overcoming of agonescence may involve induction of the potential to express telomerase activity, but where p53 remains present and functional, it may inhibit telomerase activity (see more discussion on overcoming agonescence in section IV.E.). This inhibition could occur through effects on hTERT transcriptional activation or direct association with other proteins in the telomerase complex. In the p53(+) situation, telomerase reactivation was not observed prior to the lines attaining an extremely short telomere length, mean TRF <2.5 kb, and a very faint TRF signal. However, the p53(-/-) lines did not need to attain mean TRF lengths < 3 kb prior to expressing telomerase activity, and indeed never acquired such short telomeres. p53(+) telomerase(-) conditionally immortal HMEC may require an as yet undefined alteration that occurs when the TRF reaches < 3 kb, in order to overcome a p53-imposed inhibition of telomerase expression and/or activity. A role of p53 in suppression of telomerase activity may exemplify the potential of p53 to express functional activities in the absence of conditions which lead to its activation. In this regard, the higher p53 protein levels in post- vs. pre-selection HMEC might compensate for the loss of p16 expression by increasing surveillance and suppression of genes favoring transformation, such as hTERT.

The earlier acquisition of TGFb resistance in the p53(-/-) lines relative to the p53(+) lines is likely a consequence of their earlier expression of telomerase activity. We have shown (see section IV.D. below) that transduction of hTERT can confer gradual TGFb resistance, over 10-20 passages, to telomerase(-) early passage conditionally immortal 184A1. In the three p53(+) lines, most cells showed good uniform growth in TGFb by around 15 passages following the first detection of telomerase activity. In the two p53(-/-) lines, most cells showed good growth in TGFb around 15 passages following the initial isolation of these lines. 184A1-GSE22 also showed a gradual increase in TGFb resistance over the 13 passages in which it was assayed. Altogether, these data indicate that both p53(+) and p53(-/-) newly emerged immortal HMEC lines can gradually gain resistance to TGFb growth arrest after acquiring telomerase activity.

IV.D. Effects of ectopic expression of hTERT in finite lifespan and conditionally immortal HMEC

(references: Stampfer et al., 2001;Hosobuchi and Stampfer, 1989; Stampfer et al., 1993a; Nijjar et al., 1999)

Index

Initial published reports on hTERT-immortalized human cells showed no alterations in growth control in response to serum deprivation, high cell density, specific pharmacological inhibitors, or oncogenic Ras, nor were gross chromosome instability, anchorage-independent growth or tumorigenicity reported. These data suggested that hTERT-induced immortalization does not affect normal cell behavior. More recent data from our lab, and other labs (using a variety of cell systems) now suggest that hTERT can have biological effects distinct from telomere length maintenance.

Since we had observed the gradual development of resistance to TGFb growth inhibition following detection of telomerase activity in all our immortally transformed lines, as well as a tight correlation between good growth in TGFb and expression of telomerase activity in individual conditionally immortal clones showing rapid conversion, we tested the hypothesis that expression of telomerase activity in p16(-) HMEC could induce the ability to maintain growth in TGFb. hTERT was transduced into both pre-stasis and post-selection finite lifespan HMEC, and into conditionally immortal 184A1 both pre-conversion and in-conversion. Our results indicate that the expression of exogenously introduced hTERT can induce resistance to TGFb inhibition in HMEC lacking expression of p16.

hTERT transduction into three different post-selection HMEC at differing passage levels led to telomerase activity within one passage after infection, and a rapid increase in mean TRF to ~10-12 kb (Fig. 4D.1a). All the hTERT-exposed post-selection HMEC have maintained rapid continuous growth. Similar to other reports, the hTERT-transduced cells exhibited no anchorage-independent growth, and remained severely growth inhibited when exposed to oncogenic Raf-1. Assay for growth in TGFb indicated that these post-selection HMEC rapidly gained TGFb growth resistance (Fig. 4D.2a).

Pre-stasis p16(+) HMEC transduced with hTERT at passage 3 did not become immortal nor acquire TGFb resistance. Transduced and control cells senesced similarly around passage 5. However, one hTERT-transduced culture dish did give rise to a continuously growing population. With increasing passage, this population displayed decreasing levels of p16 expression (Fig. 4D.3a&b). By passage 20, almost all cells were p16(-), and the earliest indications of the ability to maintain growth in TGFb were detectable (Fig. 4D.3c). Subsequently, there was a gradual increase in the number of cells with progressively better growth capacity in TGFb.

hTERT transduction into good-growing, pre-conversion conditionally immortal 184A1 at passage 12 produced rapid telomere elongation (Fig. 4D.1b). Consistent with the prevention of telomere erosion to < 3 kb, hTERT transduction eliminated elevated p57 protein expression as well as the associated slow heterogeneous growth phase. hTERT also conferred TGFb resistance, well before it would have been acquired as part of the conversion process (Fig. 4D.2b). Similar to the kinetics observed in our immortally transformed HMEC lines following endogenous reactivation of telomerase activity, TGFb resistance was acquired gradually over 10-20 passages.

hTERT transduction of conditionally immortal 184A1 at passage 22, which had already begun the conversion process (mean TRF < 2.5 kb, poor heterogeneous growth, elevated p57 levels) also resulted in rapid telomere elongation (Fig. 4D.1c). However, in this case there was no significant effect relative to control cultures on the existing levels of p57, nor on growth capacity in the absence of TGFb. hTERT again conferred TGFb resistance gradually over 10-20 passages (Fig. 4D.2b). Thus, the acquisition of good growth in TGFb following hTERT expression is distinct from attainment of good growth in the absence of TGFb.

To determine what functions of the ectopic hTERT were required for induction of TGFb resistance, two different hTERT mutants were introduced into post-selection 184 HMEC at passage 12. One mutant contains inactivating amino acid substitutions in the reverse transcriptase domain, and the other is a wild type hTERT with a carboxyl-terminal HA epitope tag that inhibits in vivo telomerase activity. Neither mutant induced TGFb resistance in recipient cells, indicating that hTERT must be catalytically active and capable of telomere maintenance to confer TGFb resistance.

click here to see figure 4D.1

Legend for Figure 4D.1. Ectopic hTERT causes telomeres to lengthen in normal and conditionally immortal HMEC, while telomeres become critically short in immortal HMEC undergoing conversion. (A) Post-selection 184 HMEC; 184-hTERT transduced at passage 11 shows rapid telomere lengthening while 184-LXSN telomeres continue to shorten to a mean TRF of ~5 kb at senescence. Both (B), conditional immortal 184A1 transduced at passage 12, and (C), conditional immortal 184A1 transduced at passage 22 with hTERT show rapid telomere lengthening, while 184A1-LSXN undergoing conversion show continued telomere erosion to faint critically short mean TRF of ~2 kb, followed by stabilization.

click here to see figure 4D.2

Legend for Figure 4D.2. hTERT expression induces TGFb resistance in (A) post-selection and (B) conditionally immortal HMEC. Closed arrows indicate passage of infection; open arrows indicate passage when telomerase activity was first detected in the uninfected lines.

click here to see figure 4D.3

Legend for Figure 4D.3. Rare hTERT immortalization of pre-selection 184 HMEC is associated with gradual downregulation of p16 and acquisition of TGFb resistance. (A) Decreasing p16 expression (brown precipitate) in individual cells with increasing passage in the rare pre-stasis HMEC that gained immortality after hTERT transduction at passage 3 (hTERT-184(3p)). p16(-) post-selection HMEC transduced with hTERT (184-TERT(11p)) are shown for comparison. Bar = 100µm. (B) Western blot analysis showing decreasing p16 expression with increasing passage in hTERT-184(3p). (C) Increasing ability to grow in TGFb correlates with decreasing p16 expression in hTERT-184(3p). Growth was assayed by seeding 5x10⁴ cells in 35-mm dishes for 1-8 days, then labeling and visualizing by autoradiography; LI was determined from 5 separate fields at the indicated days post-seeding.

The mechanism responsible for hTERT induction of TGFb resistance remains to be elucidated. Our data indicate that there is no correlation between telomere length and TGFb resistance. The incremental acquisition of TGFb resistance in conditionally immortal HMEC suggests that the effect of hTERT is likely to be indirect, possibly involving cumulative changes in chromatin structure and/or soluble factors. hTERT expression might indirectly change the abundance, modification, and/or spatial arrangements of signalling molecules involved in TGFb growth inhibition. Although the hTERT-transduced HMEC were no longer sensitive to TGFb-induced growth inhibition, like our other finite lifespan and immortal HMEC, they were still capable of responding to TGFb with differentiated functions.

Gaining the ability to maintain growth in TGFb is common to malignant progression in human carcinomas. In some cases, this gain can be attributed to mutations in the TGFb signal transduction pathway, but in most instances, including most breast cancers, no such mutations are seen. These data can be explained by an obligate gain of TGFb resistance with telomerase reactivation. However, since loss of TGFb inhibition may provide a strong growth advantage during carcinogenesis, mutational mechanisms of TGFb resistance may also be selected for prior to the gradual acquisition of telomerase-induced TGFb resistance

The ability of hTERT to induce TGFb resistance suggests that immortality could be more than a passive facilitator of malignant progression. However, although TGFb resistance may be a tumor promoting property for immortal epithelial cells, it is not a malignant property per se, since normal mesenchymal cells may be TGFb responsive but not growth inhibited by TGFb. Furthermore, hTERT-induced immortalization did not produce other phenotypes characteristic of malignancy (e.g., anchorage-independent growth), or of the full immortality resulting from overcoming agonescence and undergoing conversion (e.g., growth in the presence of oncogenic Raf-1). hTERT-induced indefinite lifespan HMEC did not express p57 nor the associated period of slow heterogeneous growth, and most importantly, never underwent an extended period with critically short telomeres. hTERT transduction may therefore generate the least deviant indefinite lifespan human cells. However, some changes in signal transduction do occur, such as the altered responsiveness to TGFb and possibly other alterations currently under investigation. Consequently, cells immortalized by hTERT are not totally equivalent to normal finite lifespan HMEC.

Conversely, hTERT-immortalized cells may not provide the best model of cancer cells. Ectopic hTERT produces long telomeres, while most human cancers and tumor-derived cell lines contain short telomeres. Moreover, hTERT immortalization bypasses agonescence and conversion. Overcoming agonescence requires cells to sustain as yet undefined error(s) that most likely occur only after very short telomeres and chromosomal aberrations have been generated. Conditionally immortal cells and cells with extremely short telomeres might possess unique properties that are vulnerable to therapeutic interventions. Methods of HMEC immortalization that don't include overcoming agonescence and undergoing conversion, or don't produce cells with critically short telomeres, may preclude testing of therapeutic interventions that target these potentially rate-limiting steps in immortalization and tumor progression.

IV.E. Overcoming Agonescence and Genomic Instability

[under construction]

Index

IV.F. Speculations about Immortalization and the Conversion Process

Index

My approach to science, reflected in my development of this HMEC model system, differs from most current cell and molecular biologists. I've always been most interested in being able to grasp a view of the "Big Picture". This means to me observing and measuring at a level of resolution that may allow the whole of some process (e.g., HMEC senescence and immortalization) to be seen - albeit at a resolution where individual details may not be in focus. Then, the goal is to gradually bring into more detailed focus those areas which the overview indicates to be of most relevance to understanding the overall process, and to see the ways in which the many individual components of that process may interact in linear pathways and in very complex, interactive networks. Thus my goal of developing a relatively easy to use HMEC system - so that many individual investigators can contribute to filling in the details of the behavior of this one cell type.

In the following sections, I present some of my "Big Picture" speculations and opinions about immortalization and conversion, and what it all means. The questions that I find interesting are:

(1) What are the underlying mechanisms?

(2) Does a conversion like process occur during in vivo carcinogenesis?

(3) Can these studies lead to something clinically useful?

(4) What can this tell us about scientific approaches?

IV.F.1. Speculations about mechanisms

Index

Inappropriate growth regulation is a hallmark of malignancy, and the failure to stop growth by undergoing senescence is a defining difference between cells in normal vs. tumor tissues. Unlike cells from normal breast tissues, some tumor-derived cells can express an immortal potential in culture. Underlying this difference is the expression of telomerase activity in the vast majority of breast cancers but not in normal breast tissues. Most investigators would now agree that the ability to maintain proliferation beyond the limits set by replicative senescence is essential for cells to acquire the multiple errors that permit development of invasive and metastatic tumors. Yet, despite many years of investigation by many scientists, the precise pathways by which finite lifespan human epithelial cells acquire an indefinite lifespan have not been defined.

Based upon our studies with HMEC, in the context of the enormous amount of information that has been reported in the past years, we have now ventured to propose a comprehensive overview of the steps involved in HMEC senescence and immortalization (see Brief History for a summary, and sections II.B-C for more details).

Overcoming a stress-induced, RB-mediated stasis barrier appears to be a common requirement for malignant progression, as evidence of some disruption in this pathway is found in most human carcinomas (e.g., RB inactivation, loss of p16 expression, loss of p53-dependent p21 expression, overexpressed cyclin D1, derangements in expression of other cyclins, overexpressed cdk4 or mutated cdk4 unable to bind p16, overexpressed Bmi-1 or Id-1, inhibitors of the ets transcription factors that are activators of p16, derangements in ets expression). These data strongly suggest that the stasis barrier may be an in vivo as well as in vitro obstacle to immortal transformation. In human breast cancer, loss of RB is not the most common derangement in this pathway (<10%), while alterations in p16 expression, overexpressed D1, and alterations in expression of other cyclins and ets factors are commonly seen.

We are hypothesizing that the main rate-limiting step in immortal transformation may be attaining the errors necessary for telomerase reactivation/telomere stabilization. Consequently, from the "Big Picture" viewpoint, we think one of the most critical unanswered questions is: What turns on telomerase? Further, we believe that the HMEC system we've developed may provide one of the best means of addressing this question. This is because:

1) Short lived animals such as rodents do not express stringent repression of telomerase activity. Thus turning on telomerase is not rate-limiting for immortalization/malignancy. Stringent repression of telomerase activity appears to have developed as a tumor-suppressor mechanism in long-lived animals (short-lived animals have more pressing problems than carcinogenesis). We therefore think that rodents may be an inappropriate system to model this particular crucial step, although studies with mTerc- mice have been valuable in pointing out the importance of telomerase reactivation in malignant progression.

2) In vitro model systems that immortalize human cells using ectopic hTERT expression can not be used to study what is responsible for endogenous telomerase reactivation during malignant progression (section IV.D).

3) We have been able to immortalize post-stasis HMEC (post-selection and EL) with combinations of molecular derangements observed in human breast cancer. These immortalized lines are now being used to investigate the underlying mechanisms.

Our data have indicated that (thus far) we can not immortalize post-selection HMEC using only one known cancer-associated error. We suggest that the extremely stringent telomere-length based senescence seen in human cells is at least partially due to the necessity of attaining multiple errors for telomerase reactivation/telomere stabilization. In cells with none of these errors (even post-stasis) the probability that one cell will accumulate all the necessary errors for telomerase reactivation is exceedingly small, even in the presence of the genomic instability of agonescence/crisis - since the telomere dysfunction is reducing proliferative potential and viability. However, if one error pre-disposing to telomerase reactivation is already present in the cells as they approach agonescence/crisis, then the ensuing genomic instability may give rise to a complimentary error(s) that leads to telomerase reactivation/telomere stabilization. This situation can be most clearly seen in our ZNF217 immortalized lines (see section IV.B.1, and chart IV in Brief History), as well as in the vastly different efficiency with which c-myc immortalizes EL 184Aa vs. post-selection 184. Presumably, 184Aa harbors an immortalization predisposing defect not present in post-selection HMEC.

As a working model, telomerase reactivation may require: (1) an error to make the hTERT gene accessible for transcription; (2) an error to promote transcription of the hTERT gene; (3) loss of p53-mediated repression of hTERT expression via loss of p53 or an unknown mechanism that relieves this repression as part of conversion, when the mean TRF declines to < 2.5 kb. We have shown that some of the derangements seen in breast cancer (e.g., overexpressed ZNF217, overexpressed c-myc, and loss of p53 function) can contribute to overcoming the telomere-length based barrier. We therefore suggest that there may be a set off breast cancer associated errors that are correlated with overcoming this senescence barrier, as there is a set that correlates with overcoming the stasis barrier.

Additionally, our data (see section IV.E.) suggests that overcoming agonescence is associated with some form of telomere protection, i.e., gross genomic instability is no longer present. The mechanism underlying this observation, and its relationship to telomerase activation, has not yet been ascertained.

Our p53(+) HMEC that have overcome all senescence barriers still need to undergo the conversion process. We have not yet defined the precise mechanisms that contribute to this conditionally immortal phenotype, nor the ways in which the constraints on conditionally immortal HMEC are relieved during conversion. Our realization that loss of p53 function will relatively quickly alleviate the conditional immortal constraints (e.g., repression of telomerase activity and expression of p57) has affected our hypotheses about possible mechanisms.

1) The p53(+) conversion situation (70-85% of breast cancers are reported to have wild-type p53).

We are still in the process of determining the cause and effect relationships among the various phenotypes expressed in conditionally immortal cells and during conversion. We have shown that the acquisition of resistance to TGFb-induced growth inhibition can be attributed to the expression of telomerase activity, but we do not understand the nature of the p53 imposed repression of telomerase, nor how this is relieved as telomeres become exceedingly short. p53 could exert repression by inhibition of transcription, through protein-protein interactions, and/or effects on chromatin conformation and telomerase accessibility. The loss of telomere sequences may then alter chromatin conformation, telomere molecular composition/organization, and/or gene expression, to effect the observed changes. Additionally, it is possible that when telomeres get so short, the telomere protection mechanism that was preventing gross genomic instability may partially break down, leading to sister chromatid fusions and chromosomal rearrangements in individual cell lineages, and subsequent altered gene expression.

The p53-dependent p57 mediated growth inhibition can not be strictly correlated with telomerase activity or TRF length, as we have seen (section IV.D.) that ectopic expression of hTERT, once conversion begins, does not affect the existing p57 expression. However, TRF length (< 3 kb) does appear to be the trigger responsible for the failure to downregulate p57 expression upon exit from G0 into G1 in cells grown with all required growth factors. Again, changes in chromatin structure or telomere protein composition/organization may be affecting gene expression. Drawing from yeast models, it is possible that (1) decreasing telomere length results in heterochromatin changes, which produces altered gene expression; (2) decreasing telomere lengths results in release of telomere associated proteins which then bind elsewhere, producing altered gene expression.

One of the most unusual aspects of conversion is the short time required to generate extensive heterogeneity from repeatedly cloned populations. This property is most easily demonstrated by the rapid generation of heterogeneity in growth capacity in response to TGFb. A possible underlying cause could be generation of heterogeneity in telomere length of individual chromosomes of individual cells at each population doubling. Differences in either telomere length of specific chromosomes, or the overall level of remaining telomeric repeats, could result in different gene expression among recently cloned populations.

We theorize that the relative levels of expression of several interacting molecules determine whether individual conditional immortal cells complete the conversion process or lose proliferative ability. External conditions may also influence an individual cell's fate; preliminary studies indicated that culture conditions (e.g., the presence of serum) influenced the efficacy with which conditionally immortal cells converted. Thus the mechanisms underlying conversion are likely to depend upon quantitative interactions of multiple cellular components, each of whose levels may vary over a continuous range. Such complex interactions may be difficult to precisely determine in a system with multiple undefined variables.

2) Immortalization with non-functional p53

The only aspect of conversion, as we first described it in p53(+) lines, that clearly remains unchanged where p53 function is absent is the acquisition of TGFb-induced growth inhibition. However, this is not surprising given that this resistance is a consequence of the reactivation of telomerase, which occurs in both situations.

However, we can not say that full telomerase reactivation in the p53(-) situation occurs immediately. In our p53(-/-) lines, and where p53 function has been abrogated by GSE22 or SV40T, some telomerase activity is present as soon as assayed, but activity rises gradually over several passages. Possibly, feedback mechanisms exist that contribute to a gradual vs. immediate full reactivation of telomerase or there may be preferential growth of cells with higher levels of telomerase. In any case, the rapid presence of telomerase activity post-crisis, or following abrogation of p53 function, suggests that the event(s) which allowed agonescence/crisis to be overcome may involve, in addition to the stabilization of telomere ends, the potential for hTERT expression and telomerase activity. It will be of interest to determine how these two properties may be related.

Once cells are fully converted, they appear to have controls on telomere length similar to what has been described in yeast and the examined immortal human tumor cell lines. We postulate that malignant transformation requires additional errors providing positive growth advantages and invasive capacity. Given the relative ease with which our immortally transformed HMEC can be made anchorage independent or tumorigenic (exposure to 1 or 2 known oncogenes), vs. the extreme rarity of generating immortal HMEC lines, and the extensive period of poor

heterogeneous growth in the p53(+) conditionally immortal HMEC, we suggest that it may be the acquisition of unlimited proliferative potential which provides the rate-limiting step in malignant transformation.

IV.F.2. Speculations about conversion in vivo

Index

I consider one of the most interesting question to be whether a conversion process occurs during carcinogenic progression in vivo. There is currently no definitive data relevant to this, so my speculations relate to the presence of in vivo data consistent with our in vitro model.

There have been few other studies of cultured HMEC that address the question of conversion. This is not surprising, given that most in vitro transformed lines lack p53 function. There is a recent report of in vitro transformed HMEC (using a chemical carcinogen) consistent with the occurrence of conversion where p53 function is present (Li et al., 2002). I've found one report about the existence of continued telomere shortening for around 30 passages post-establishment prior to stabilization of telomere length in two immortally transformed cell lines from breast cancer pleural effusions (Rogalla et al., 1994), and I've heard from other investigators that establishment of some human tumor cells lines in culture commonly involves a long period of poor growth.

There are several aspects of malignant progression and tumor cells in vivo consistent with our models for overcoming agonescence/crisis and undergoing conversion:

1) Cells in most human tumor tissues have telomere lengths shorter than normal tissues, and most tumor derived cell lines have short regulated telomere lengths, similar to what we have observed in our fully immortalized HMEC lines. Moreover, recent studies (Meeker et al., 2002) have shown that short telomere lengths are found in the precursor lesions of cancer, e.g., DCIS and benign prostatic hypertrophy. These data are all what we would predict if tumor cells attain immortality and reactivate telomerase as a consequence of errors generated by telomere dysfunction during agonescence/crisis, and/or a need to attain extremely short telomeres for telomerase reactivation in the presence of functional p53 (as is the case for most breast and prostate cancers).

Additionally, telomerase activity is detected early in breast cancer progression, at the stage of DCIS, and recent studies have suggested heterogeneity in telomerase activity within tumors. These data are consistent with overcoming agonescence/crisis in DCIS, and the process of conversion continuing in primary carcinomas.

2) Malignant progression in vivo is a gradual process, with generation of extensive heterogeneity. The gradual process of conversion, with its generation of extensive heterogeneity, more closely models this process compared to the more rapid transformation to immortality often seen in viral oncogene mediated immortal transformation. It's possible that a gradual conversion process in vivo could, at least partially, account for the slow, heterogeneous growth seen in many primary carcinomas (particularly ones that are p53+) prior to the appearance of more aggressive, invasive tumors.

A significant difference between our model system in vitro and what may occur in vivo is that we are selecting only for acquisition of indefinite proliferative potential. In vivo, under conditions such as DCIS, an extended period of conversion could provide a continuous pool of slowly dividing cells able to accumulate errors which both provide a selective growth advantage and promote malignant behavior (e.g., growth factor independence, vascularization, genomic instability). Conversion to full immortality might not even be necessary for a tumor to become malignant and metastatic; an extended period of conditional immortality could be sufficient. Our data indicate that conditionally immortal cells can undergo a very large number of PD before becoming fully converted. We have also seen that there can be stochastic emergence of rare, more aggressively growing fully converted cells.

3) p53(-) breast tumors have a significantly worse prognosis than p53(+). Our data showing that loss of p53 function produces more aggressively growing immortal cells more rapidly is consistent with these data, and with observed loss of p53 occurring mostly at the stage of DCIS.

4) Most breast cancers exhibit aneuploidy and genomic instability, although most are also p53(+). As described above (section IV.E.), the generation of genomic instability/bridge-fusion-breakage cycles due to telomere dysfunction prior to immortalization could account for these observations. Further, cells which attained all the errors necessary for immortalization prior to the onset of extensive telomere dysfunction could account for the smaller percentage of near-diploid breast cancers.

How can any of this be translated into improvements in detection, prevention, prognostic information, or treatment of breast cancer? I don't know, and I'm not even ready to speculate on the website. I do believe that uncovering novel processes in human epithelial cell transformation may well open the door to novel methods for clinical intervention. And, I am concerned that potential therapeutic targets will be overlooked if these novel processes are ignored, and researchers use only models in which conversion doesn't exist, or that are p53(-) (e.g., viral oncogene immortalization, all rodent immortalization, immortalization with hTERT). I therefore welcome input and collaboration with others on this most important topic.

IV.F.3. Speculations and opinions on how all this relates to approaches to scientific questions

Index

CAUTION: The following presents my strong editorial opinions.

Pick up any good molecular and cell biology journal and look at the abstracts. Many make very generic conclusions about how "cells" work, e.g., cell cycle and signal transduction mechanisms, without ever identifying in the abstract what "cells" are being studied. Some never identify the "cells" in the introduction. For some papers, you have to look carefully in the Methods section to find out what cells were used. My random sampling notes that in a large percentage of these cases the "cells" being referred to are some variant of 3T3 or HeLa. Does most of the scientific community really believe that all cells are equivalent, that an immortally transformed p53-/-, aneuploid rodent fibroblast cell (3T3) or an immortally transformed functionally p53-/-. RB-/-, HPV positive, aneuploid cervical carcinoma derived cell (HeLa) will accurately reflect the cell cycle, signal transduction, etc., mechanisms of normal finite lifespan human cells? Or any normal cell? Sweeping generalizations are being presented as "truths" based on studies of cells grossly deranged in many known (and unknown) processes of growth control. Cancer cells are defined by having deranged growth control. We even have distinguished journals publishing papers that say that an immortally transformed functionally p53-/- and RB-/-, SV40T positive, aneuploid human mammary cell line, which upon continued passage can demonstrate AIG and even tumorigenicity (HBL100) are "normal diploid" cells, and can be used as "normal" controls for tumor cells. This is incontrovertibly serious scientific error. And I observe almost no notice or concern about this in the scientific community.

One cell type obviously does not accurately represent the behavior all of cell types. We are humans and not yeast; our bodies function based on very complex interactions of many different differentiated cell types - which have different histories, functions, gene expression, and fates. Our research and that of others shows significant differences in key cell cycle regulators and senescence mechanisms between normal epithelial and fibroblast cells, even from the same person's breast (e.g., p21, p16, p53, responses to TGFb, PDGF, cAMP, HPV16). Our research shows obligate differences in cell cycle regulation between finite lifespan and immortal cells (e.g., expression and regulation of some CKIs and response to Raf-1 and TGFb). Our research and others show significant differences in telomerase regulation between human and rodent cells.

Generalizations from one cell type to another can also be misleading at a subtler level than the presence or absence of specific molecular derangements. Many areas of science and biology are coming to appreciate that whole systems function not so much by simple linear pathways as by complex interacting networks, e.g., see Pfeifer, Nature 400:213'99; "despite our desire for cells to think in linear terms, they refuse to do so... (we) will have to adjust to the new millennium, in which intricate networks integrate many inputs to generate the complex output that is cell behavior."

So even if two different cell types contain a similar complement of key "nodal points" (molecules), the actual pathways utilized by those two cells could be quite different, due to how that cell's network integrates inputs to yield outputs. The actual way a cell wiring system is utilized under physiological circumstances is also unlikely to be mimicked by examining the effects of non-physiological forced overexpression of one or multiple genes under no normal cellular control.

Yet in practice, many biologists act as if finite lifespan and immortal, p53+ and p53-, epithelial and fibroblast, rodent and human cells, etc., are all functionally equivalent. The behavior of molecules ectopically overexpressed is used to "define" mechanisms, with the implicit assumption that what is being observed is the same as actual physiology. How have we come to such an illogical, patently incorrect situation, and what can be done about it?

Let's say you're a good molecular biologist studying growth factor molecules, epidermal growth factor in particular. Would you write papers that didn't distinguish, that treated as functionally equivalent, EGF and TGFa, amphiregulin, or VGF? EGF and the FGF or PDGF family, CSF, NGF? human EGF and the Drosophila EGF-like homologue spitz?, wild type EGF and an EGF molecule with significant amino acid substitutions? prepro-EGF and active EGF? So, as a cell biologist studying human carcinogenesis, I'm baffled as to why good molecular biologists write papers and give talks that don't distinguish, that treat as functionally equivalent: mammary, liver, colon, and keratinocyte cells; epithelial, fibroblast, macrophage, and neuronal cells; cells with wild type vs. mutant p53; human, mouse, rat, and chicken cells; finite lifespan, immortal, viral oncogene transformed, and tumor derived cells. The most logical explanation I've come up with is that the exigencies of funding have led many investigators to keep doing the next step in the same systems they've been working with, since most funding agencies are not known to fund actual proposals, and it could be scientific career suicide to do work that required more long-term system development.

In my opinion, one thing that could be done about this situation is to have journals require all papers to clearly identify the source of the cells being used, and all information potentially relevant for the conclusions drawn, e.g., finite lifespan or immortal, cell type and species, status of p53 and RB, tumorigenic properties, presence of viral oncogenes, etc. These are very significant variables, and need to be identified just as one would identify other variables which may significantly influence the results. Generic conclusions about "how cells work" based on studies with one or a limited number of cell types should be disallowed.

This current, to me highly illogical situation of acting as if all mammalian cell types are functionally equivalent, has led me to consider how our approaches to scientific questions influence the kinds of data we obtain, how we interpret that data, and how we value different kinds of data.

My point of view is well reflected in a quote from Dan Mazia in the ASCB newsletter: "There are many paths in the advancement of science, but the giant leaps in our Science of the Cell have been made by seeing. First we see and then we interpret and only then do we pursue mechanisms and theories. The gift of the great microscopist is the ability to think with the eyes and see with the brain".

I too believe that careful seeing is the important first step for novel scientific exploration. Thus, I see that science advances by first carefully observing nature, then discerning the patterns behind the observed phenomena, describing and measuring these patterns, and finally looking for the mechanisms which give rise to the observed patterns. The image that conveys for me this logical order is also the microscope. If one wishes to observe a structure at high resolution, one doesn't first use the highest power objective available. First the lower resolution picture is brought into focus, and then higher resolution images are gained via a stepwise approach. In this manner one has a sense of the larger context in which the observed detailed structure lies. In contrast, focusing on the detail of an area at high resolution, before ascertaining the location and function of that area within the larger context, may elucidate patterns of great beauty and elegance. These lovely patterns, however, may not provide much predictive power about the behavior and organization of the larger scale structure of interest.

So, what I see, is that many molecular biologists seem to believe that the most important thing is to explain "mechanisms" at the highest available resolution. Looking at larger scale pictures at lower resolution is commonly dismissed as "phenomenology", unworthy of serious attention or funding. In the absence of stepping back to grasp the more encompassing picture, more and more detailed studies are being done using the same cell culture technology previously employed. This cell culture technology may have been the state-of-the-art 10, 20, 30 years ago, but is now very dated. Much better options are currently available, and even better options could be developed if resources were expended in that direction. In contrast, most molecular biologists would consider it silly to even think of using 10, 20, 30 year old molecular techniques, when much better options are available. And they would consider expending resources to improve molecular technologies reasonable and valuable. I think this attitude is detrimental to the advancement of science. I believe it has led to some very elegant studies being done in cell systems that are basically uninterpretable if the question of interest is understanding normal human cellular physiology, and the derangements which transform normal human cells into malignant. How can one possibly know if results seen in deranged cells reflect normal human unless the studies have already been done on normal human cells. I see no good reason for most of the continued use of cell technology like 3T3 and HeLa for studies on topics like cell cycle regulation and signal transduction. Much better options are available, although they may take a little more effort to start.

Yes, it is somewhat more difficult to use normal human cells, and there is no easy way to do in vitro - in vivo correlations, or things like knockout experiments (although the advent of siRNA may make such studies more feasible). Sometimes experiments need so many cells with sufficient PD that immortal lines will be required. There will still be a need for model systems that model. Nonetheless, I believe that if there were the collective agreement, bolstered by funding support, to put more resources into development of cell culture systems that can optimize modeling of in vivo reality (including ones that allow study of cell-cell and cell-matrix interactions) this could be accomplished in a reasonable time frame. I truly believe that most scientists are motivated to discover the true laws of nature and to better the human condition. While there is no way to know for sure the eventual significance of one's research on human health and well-being, I think we can greatly improve our odds by avoiding obvious anachronisms like using HeLa to study cell cycle control. If you must use immortal cells, minimally deviant lines like 184A1, 184B5, and MCF10A are available.

One more personal point of view. In biological systems, complexity and productivity are enhanced by diversity. So too, I believe the scientific endeavor is enriched by including those with a diversity of scientific approaches, and impoverished if we are all required to conform to a majority mold which does not fit. Since some of my style and skills differ from what I perceive as the accepted and valued current norm, I am sensitive to how minority viewpoints and skills get excluded. I survived many years of being told that what I was doing (trying to develop improved human cell culture technology to better address questions of human cell physiology and carcinogenesis) was just phenomenology, "gardening", even "witchcraft". Valid, although minority approaches can illuminate things that others miss, can create resources that others can't. While I think it appropriate for most scientists to follow the "high resolution - small field" approach, I believe we all benefit if our scientific culture (including funding and publishing organs) has the tolerance limits both to openly support some who follow the "low resolution - large field" approach and to direct "high resolution - small field" studies to areas that appear most promising based upon the "larger field" work.

V. Synchronization of HMEC Cultures and Role of EGF Receptor Signal Transduction

(references: Stampfer, Pan et al. 1993; Stampfer and Yaswen, 1993; Bates et al., 1990)

Index

One of my long-term goals in developing the HMEC system was to ask questions related to the role of specific positive and negative growth factors in controlling gene expression, cell cycle progression, and proliferation of normal and transformed cells. I assumed that these studies would require synchronized cell populations; therefore, I looked for a method to synchronize the HMEC. Additionally, I wanted to find a method that would not involve use of metabolic inhibitors or general starvation, and thus potentially not be cytotoxic or stress inducing.

We had previously shown a stringent requirement for EGF receptor (EGFR) ligands (e.g., EGF/TGFa) for clonal growth of the HMEC, although the finite lifespan HMEC could grow in mass culture without addition of exogenous EGF. Further study demonstrated that the mass culture growth without exogenous EGF was due to an autocrine loop resulting from endogenous production of EGF-like ligands such as TGFa and amphiregulin. Blockage of EGFR signal transduction with an antibody to the EGFR prevented growth. Although 184A1 and 184B5 synthesized similar amounts of TGFa as normal 184 HMEC, they failed to secrete this protein, and thus required addition of EGF to the medium for growth (see figure 5.2).

The above data suggested that blockage of EGFR signal transduction might provide a method to reversibly arrest these HMEC, and this possibility was examined in detail. After growth in medium with EGF to midconfluence, HMEC cultures were maintained for 48 hrs in medium without EGF and containing monoclonal antibody (MAb) 225 to the EGFR. Cells were then refed with medium containing 5 times normal EGF. During exposure to MAb 225, the HMEC acquired a less refractile morphology with increased cell-cell contact, decreased motility, and few mitoses. After re-exposure to EGF, the typical cobblestone epithelial morphology and many mitoses were visible by 24 hrs. Post-selection HMEC could be maintained for at least 18 days in EGF deficient medium plus MAb 225 and still regain a normal cobblestone appearance with many mitoses after EGF re-exposure, suggesting that the growth inhibited cells were arrested in a viable, non-cytotoxic state (I refer this state as sleeping/hibernating). We have not systematically tested 184A1 and 184B5 for long-term viability in minus EGF medium. Although they enter quiescence after 48 hrs without EGF, differences in the G0 state between finite lifespan and fully immortal HMEC may mean that the immortal cells differ from the normal HMEC in their ability to remain viable in "hibernation".

Protein and DNA synthesis in post-selection 184 and 184B5 were assayed by incorporation of 14C-leucine and 3H-thymidine (Figures 5.1a & 5.1b). Protein synthesis remained depressed as long as the antibody was present and increased rapidly following re-exposure to EGF. DNA synthesis decreased 12 hr after antibody addition, and was sharply decreased by 24 hr. DNA synthesis resumed only 10 hr after EGF re-exposure and then increased sharply to a peak around 18 hr. We have examined cells from reduction mammoplasty specimens 48 and 161, and found similar results. For specimen 48, DNA synthesis following restimulation with EGF began and peaked about two hours earlier, and there was greater synchrony exiting S phase, suggesting that good synchrony could be maintained into the next cell cycle.

Click here to see figures 5.1a & 5.1b

Figure 5.1. Effects of blockage of EGF receptor signal transduction on DNA and protein synthesis by post-selection 184 HMEC and 184B5. (A) Cells from specimen 184 were grown in 35mm dishes in complete MCDB 170 until midconfluence. Treated cultures were then exposed to MCDB 170 minus EGF plus 8 ug/ml MAb 225 for 49 hrs, while control cultures received complete MCDB 170. After 49 hrs, all dishes were washed once with PBS and refed. Treated cultures were refed with either complete MCDB 170 containing 25 ng/ml EGF (triangles) or maintained in MCDB 170 minus EGF plus 8 ug/ml MAb225 (squares). Control cultures (circles) were refed with complete MCDB 170 containing 25 ng/ml EGF. Cells were exposed to a 2 hr pulse of 5 uCi 3H-thymidine (closed symbols) and 80 nCi 14C-leucine (open symbols) in 1.5 ml of medium for 1 hr before and after the indicated times. Total acid-insoluble counts were then determined and are presented on a per dish basis. (B) Cell from 184B5 were treated as for 184 with two differences: no MAb 225 was used and the cells were kept -EGF for 48 hrs.

Click here to see figure 5.2

Figure 5.2. TGFa production and secretion, and effects of EGF on growth and DNA synthesis of post-selection, immortally transformed, and EGF independent variant HMEC.

To determine the effect of EGF on growth rates, 0.5 x 10-5 cells were seeded into 35mm dishes in either complete MCDB 170, MCDB 170 minus EGF, or MCDB 170 minus EGF containing 6µg/ml of MAb 225. The number of attached cells was determined 16-24 hr later. When control cultures (complete MCDB 170) were just confluent, all cell cultures were trypsinized and cell numbers determined by Coulter Counter. To determine the effect of EGF on DNA synthesis, midconfluent cultures that had been grown in complete medium were switched to the indicated medium for 24 hr. For the last 2 hr, cells were exposed to 4 µCi 3H-thymidine in 1.5 ml. Acid precipitable counts were determined by scintillation counting. To determine TGFa synthesis and secretion, cells were grown in 60mm dishes until subconfluence. 24 hr conditioned medium was then removed, and the cells harvested and frozen. Radioimmunoassays of medium and cells were performed by Robert Coffey, Vanderbilt University. These data indicate that 184A1 and 184B5 synthesized amounts of TGFa similar to normal 184 HMEC, but failed to secrete this protein. The ability of the variant lines to maintain growth in the absence of EGF did not appear to be due to increased TGFa secretion. Although A1NE and B5NE could continue to grow in the absence of EGF, their rate of growth was decreased, and they were still sensitive to MAb 225 induced inhibition of growth and DNA synthesis. It is possible that increased synthesis of another ligand for the EGF receptor, such as amphiregulin, may be responsible for their altered growth properties. The ENU induced variant, A1ZNEB also grew faster in the presence of EGF, but it was no longer sensitive to MAb 225 inhibition. Additionally, it showed significantly increased levels of cell-associated TGFa protein. One possible explanation for this phenotype is internal or membrane associated stimulation of the EGF receptor. The ENU induced variant B5ZNEI showed no difference in growth rate with or without EGF, but was still partially sensitive to MAb 225 inhibition.

Post-selection 184 HMEC were tested for how long a period of EGF re-exposure was needed following arrest to allow cells to subsequently enter S phase, with the result that a 1 hr exposure was sufficient to allow the majority of cells capable of cycling to later enter S (Figure 5.3). Thus EGF seems necessary just to get the cells into cycle (a competence rather than a progression factor). All the other growth factors were present in the medium, so we can not say if any of them specifically function as progression factors.

Click here to see figure 5.3

Figure 5.3. Effects of varying lengths of EGF exposure on subsequent DNA synthesis. Post-selection 184 HMEC were grown in 35mm dishes in complete MCDB 170 until midconfluence. Cultures were then exposed to MCDB 170 minus EGF plus 6 ug/ml MAb 225 for 48hr, washed once with PBS, and refed with either complete MCDB 170 containing 25 ng/ml EGF or MCDB 170 minus EGF plus 6 ug/ml MAb 225. At 1,2, 4, and 8 hrs after this refeeding, cells that had been refed with EGF were washed once with PBS and then refed with MCDB 170 minus EGF plus 6ug/ml MAb 225. For 1 hr before and after the indicated times, cells were exposed to 4 uCi 3H-thymidine in 1.5ml of medium. Total acid-insoluble counts were then determined and are presented on a per dish basis. Points represent the average of triplicate dishes.

The above results suggested that HMEC restimulated with EGF following the growth arrest were exiting a G0 state and entering S phase in a highly synchronous fashion. Examination of early response gene expression supported this conclusion. Expression of c-myc, c-jun and c-fos was readily detectable in post-selection cycling HMEC cells, but decreased during growth arrest (Figure 5.4ab). High levels of mRNA for all these genes were observed at 1 hr following re-exposure to EGF. 184B5 differed from normal HMEC in not showing any decrease in expression of the early response genes during the G0 arrest, while 184A1 at passage 31 showed a partial decrease. We subsequently have seen that this difference correlates with conversion to full immortality. Both late passage 184A1 and 184B5 do not downregulate c-myc mRNA or protein during the G0 state. Synthesis of TGFa mRNA, which was also inhibited in the presence of MAb225, was detected by 2 hr after EGF re-exposure. Some mRNA species, such as for keratin 5 and CLP, continued to be expressed during the growth arrest.

Click here to see figure 5.4a

Click here to see figure 5.4b

Figures 5.4ab. Effects of blockage and restimulation of EGF receptor signal transduction on mRNA expression of post-selection 184 HNEC and immortal 184B5. HMEC. Cells were grown in 100mm dishes in complete MCDB 170 until midconfluence. Treated cultures were then exposed to MCDB 170 minus EGF plus 8 ug/ml MAb 225 for 48hr, washed once with PBS, and refed with complete MCDB 170 containing 25 ng/ml EGF. Cells were harvested for RNA isolation at the indicated times. Control cultures were maintained in complete MCDB 170, and harvested 24hr after feeding. 10 ug of total cellular RNA was fractionated on 1.3% agarose/formaldehyde gels and transferred to nylon filters. The filters were sequentially probed with cDNA to: Fig. 5.4a: (A) c-myc; (B) c-fos; (C) c-jun; (D) MGSA; (E) TGFa; (F) histone 3.2; (G) NB-1; (H) keratin 5. (I) shows total RNA in the original gel stained with ethidium bromide.

Fig. 5.4b: A first filter was sequentially probed with cDNA to (A) c-myc; (B) c-fos; (C) c-jun; (D) MGSA; (E) shows total RNA in the original gel stained with ethidium bromide. The second filter was probed with (F) TGFa; (G) histone 3.2; (H) NB-1; (H) shows total RNA in the original gel stained with ethidium bromide.

Studies done largely with growth arrested fibroblast cells have defined a G0 state characterized by low metabolic activity, a rapid increase in levels of mRNAs for certain early response genes upon release from the growth arrest, and an increase of 6-7 hr in the time required to begin DNA synthesis following release from growth arrest,

relative to continuously cycling cells. These properties are all observed with the HMEC growth arrested by EGFR blockage, suggesting that this is a G0 arrest. Thus, blockage of EGFR signal transduction is sufficient by itself to cause finite lifespan and immortally transformed HMEC to enter a G0-like resting state. Unlike fibroblast cells from the same specimens, the levels of c-myc mRNA in HMEC remained high throughout the cell cycle, while c-fos and c-jun mRNA, though showing cycle dependent fluctuation, were readily detectable in the cycling epithelial cell populations.

Exit from a G0 state is not the same as passage from M into G1. In experiments with 184B5, we examined the mRNA synthesis into the second cycle, and could see an absence of the strong burst of early response gene expression as cells entered G1 without a G0 exit (Figure 5.4b). Levels of these mRNA species did increase, and it could be seen that the rise in c-fos expression preceded that for c-myc and c-jun. To study the cycle without a G0 arrest, one can follow the cells into the second cycle, or use a method for G1 arrest developed by Keysomarsi et al., 1990, which utilizes Lovastatin, an inhibitor of 3-hydroxy-3-methylglutaryl-coenzyme A reductase.

VI. TGFb Effects on Finite and Immortally Transformed HMEC

(references: Hosobuchi & Stampfer, 1989; Stampfer, Yaswen et al. 1993; Slingerland et al. 1994; Sandhu et al. 1997; Stampfer et al. 1997; Stampfer et al., 2001)

Index

TGFb is a pleotropic cytokine known to be a potent growth inhibitor of normal epithelial cells in vitro and in vivo. It modulates several key physiologic processes such as wound healing, differentiation, and tissue morphogenesis and remodeling, and is also thought to be involved in carcinogenic progression. Varying degrees of resistance to TGFb induced growth inhibition are seen in human carcinoma cells, and this loss of negative growth regulation may contribute to tumor development. While in some instances, resistance correlates with loss of functional type I and II TGFb receptors, most resistant tumor lines express normal numbers of apparently functional receptors.

I initially examined the effects of TGFb on HMEC to determine if it could be a good method for cell synchronization. This was clearly not the case. However, in the course of these studies I found that growth responses to TGFb were one of the clearest differences between the finite lifespan and immortal cells, and got hooked on trying to understand why (see Section III.).

My first difficulty with TGFb was simply getting consistent results on the extent and speed of growth inhibition of post-selection HMEC from the same individual. This mystery was partially solved when I controlled for passage level and selection batch. The effect of TGFb on some individual specimens depended upon age in vitro. While every finite lifespan HMEC that we have tested is ultimately growth inhibited by TGFb, earlier passage cells may undergo 8 or more PD before full arrest, whereas later passage cells stop growth in 1-2 PD, and with lower TGFb concentrations (see figure 6.1). Finite Lifespan HMEC show distinctive morphologic changes in the presence of TGFb, characterized by an elongated, flattened appearance. The growth inhibition was only partially reversible, the extent of reversibility decreasing with age in vitro, and was relatively asynchronous (see figure 6.2a). The cells were not in a resting state, since 14C-leucine incorporation indicated that protein synthesis was stimulated even as growth was inhibited (see figure 6.2a). The growth arrest is in mid to late G1; TGFb added 10 hrs following G0 exit was not growth inhibitory.

Click here to see figure 6.1

Figure 6.1. Effect of TGFb on growth of post-selection and immortally transformed HMEC.

Cultures were seeded into triplicate 35 mm dishes (4-5 x 10-4 cells/dish) in the indicated concentration of human recombinant TGFb1. The number of attached cells was determined 4-16 hr later. When control cultures were subconfluent or just confluent, all cell cultures were trypsinized and cell numbers determined by Coulter Counter. The attached cell number was subtracted from the final cell count. Data are presented as percentage cell number of the TGFb exposed cells relative to the non-TGFb treated controls. The b-resistant 184B5 were assayed at passage 26; B5T1 passage16 represents a very b-sensitive clone, isolated at passage 15, which ceased almost all growth by passage 30. The b-resistant B5T1 passage 35 represent populations derived from the few surviving cells. 184A1 was assayed at passage 39; the A1L5-S represent one of four clones isolated from 184A1 at passages 29-34, and was assayed at passage 32. All 4 clones gave results similar to the uncloned population when assayed within 11 passages from isolation.

NOTE: When I refer to the finite lifespan cells as TGFb sensitive, this means that no cell (as in zero) has been capable of maintaining growth in the presence of TGFb. This is very different from what may be called 'TGFb sensitive" breast tumor cell lines, where what it referred to is often a reduction in cell number or growth rate. When I refer to immortally transformed cells as TGFb resistant, this means that the cells can maintain growth indefinitely in the presence of TGFb. However, there may still be some reduction in growth rate or cells which don't maintain growth (i.e., similar to what may be called TGFb sensitive in tumor cell lines).

Another mystery which I never solved (or published, since I didn't know what to make of it) was stumbling across the observation that addition of MAb 225 to post-selection HMEC growth arrested by TGFb led to synchronous entry into S phase, within 3 hrs, of the cell population that was reversibly inhibited (see figures 5.2ab). This seemed to imply that some function of the EGFR was required to maintain TGFß growth arrest in late G1.

Click here to see figures 6.2a & 6.2b

Figure 6.2. Effect of addition of MAb 225 on cells arrested in late G1 by TGFb.

(A) Post-selection 184 passage 12 were seeded in triplicate 35mm dishes in MCDB 170 until midconfluence. Treated cultures (squares) were then exposed to 5ng/ml TGFb for 48 hr. All dishes were then washed once with PBS+0.1% BSA. The TGFb treated cultures were refed with either complete MCDB 170+BSA with no TGFb (triangles), MCDB 170+BSA with TGFb plus 6 µg/ml MAb225 (diamonds), or maintained in MCDB 170+BSA+5ng/ml TGFb (squares). Control cultures (circles) were refed with complete MCDB 170+BSA. Cell labeling was as described in figure 5.1. Results are presented on a per dish basis. Cells exposed to continuous TGFb had about 1/2 the cell number as control cultures after 48 hr, and about 1/3 after 72 hr

(B) Post-selection 184 passage 13 were seeded into triplicate 35 mm dishes and grown in MCDB 170 until sparse- midconfluent. Treated cultures were then exposed to 20 ng/ml TGFb for 48 hr. All dishes were then washed 1X and refed as indicated. For 1 hr before and after the indicated times, cells were exposed to 5 µc 3H-thymidine in 1.5 ml. Total acid-insoluble counts were determined and presented on a per dish basis. Cells exposed to continuous TGFb had 1/3 the cell number as control dishes after 72 hr of TGFb exposure. TAb 1 is an antibody to TGFa. It's effect was generally similar to that of MAb225, indicating that the observed result with MAb225 is not due to an agonist effect; however, since EGFR ligands in addition to TGFa may be present, TAb 1 did not have as stringent an effect as MAb225.

Examination of 184A1 and 184B5 for their responses to TGFb is what really caught my attention. Whereas I had never seen a single finite lifespan HMEC maintain growth in TGFb, the immortally transformed HMEC lines could give rise to populations that maintained growth indefinitely in the presence of TGFb. The data illustrating these initially puzzling results that led to the studies described in Section III. A. is shown in figure 6.1.

Although growth responses to TGFb varied among the finite and immortalized HMEC, all of these HMEC showed a similar profile of TGFb1 receptors and all expressed specialized responses to TGFb including strong induction of: mRNA and/or protein for extracellular matrix associated proteins such as fibronectin, collagen IV, and laminin; the proteases, type IV collagenase and urokinase type plasminogen activator; the protease inhibitor, plasminogen activator inhibitor 1. The level of overall protein synthesis, especially secreted proteins, was increased following TGFb exposure even where cell growth was inhibited. These results indicated that the effects of TGFb on HMEC proliferation could be dissociated from its effects on specialized responses likely to play a role in glandular remodeling, homeostasis and/or wound healing. It should be noted that fibroblast cells can show similar specialized responses to TGFb in the absence of growth inhibition. Fibroblasts from specimen 184 show a slight growth stimulation in TGFb.

We next examined the effects of TGFb on cell cycle related genes and proteins. Finite lifespan and immortal 184A1 and 184B5 HMEC populations were analyzed after exit from G0 following release from EGFR blockage. The following summarizes some of our results (some of which we have never gotten around to publishing).

1) RB is normally phosphorylated in these HMEC 5-10 hr after G0 release. Exposure to TGFb prevented this phosphorylation in the TGFb growth inhibited finite lifespan and conditionally immortal cells, but not in the TGFb resistant fully immortal cells.

2) In finite lifespan and conditionally immortal HMEC, levels of c-myc, c-fos, and c-jun mRNA, and c-myc protein (c-fos and c-jun proteins were not examined) are low during G0 arrest. In contrast, fully immortal 184A1 and 184B5 did not show downregulated expression of these molecules during G0. TGFb had no effect in any of the HMEC (growth inhibited or not) on the normally high level expression of these molecules seen l hr after G0 release (see fig. 5.4). In the presence of TGFb, post-selection 184 and all 184B5 HMEC examined (conditional and fully immortal) showed decreased myc levels ~5 hr after G0 release, while levels in all 184A1 examined were not significantly affected. Thus, contrary to reports in other systems, we did not see any TGFb effects on myc expression that correlated with extent of growth inhibition. This difference may be due to our use of cells that are all responsive to TGFb, in contrast to studies where the cells were not only resistant to TGFb growth inhibition, but also non-responsive. Because TGFb can affect more than one cellular pathway, these results caution against assuming that any differences observed between responsive and non-responsive cells are necessarily contributing to the growth inhibition pathway.

3) In general, cyclins D1, D2, E, A, and B are expressed during the cell cycle with kinetics similar to reports in other cell types. Exposure to TGFb reduces expression of cyclin A mRNA and protein in the growth inhibited cells, and has little if any effect on D1, D2, E, or cdk2 bound E, with the exception of reduction in cyclin E mRNA. Cdk2 and cdk4 protein levels are constant throughout G0 and the cell cycle, and are unaffected by TGFb in any of the HMEC.

The above studies had not detected major TGFb induced effects on expression of cell cycle associated molecules to account for the failure to phosphorylate RB in the growth inhibited populations. In collaboration with Joyce Slingerland, U.Toronto (now U. of Miami), this led us to examine the activities responsible for RB phosphorylation, namely the cyclin/cdk complexes D/cdk2,4,6, E/cdk2, and A/cdk2, and further, the molecules capable of inhibiting this activity - the CKI.

Cyclin D1-cdk4 (or D1-cdk6) complex formation is normally detected 6-18 hr following G0 release. In the presence of TGFb, the growth inhibited cells failed to show this complex formation, or D-associated kinase activity, whereas these complexes were unaffected by TGFb in the resistant cells. Assays for cyclin E and A associated kinase activity also showed significant inhibition in the TGFb sensitive cells upon exposure to TGFb. This decreased activity was due to the presence the CKI p27. p27 inhibitory activity is normally highest in G0, and then decreases as the cells progress through G1. In TGFb exposed normal HMEC, this activity remains elevated throughout G1. We also observed that the fully immortal cells had significantly less CKI activity in G0, independent of TGFb exposure.

Further studies examined the role of the CKI p15. In all examined HMEC, p15 mRNA is expressed in G0 and decreases as cells progress into G1. Consistent with the report that TGFb increased p15 mRNA levels, we observed that TGFb caused p15 mRNA expression to remain at the G0 level even after exit into G1. However (like the studies above with c-myc), this was true for both the growth inhibited and resistant cells, indicating that this effect is not correlated with the role of TGFb in growth inhibition. We did observe that p15 protein levels accumulated in TGFb exposed growth inhibited, but not resistant HMEC. Measurement of p15 protein half-life showed it to be more stable in the TGFb exposed growth inhibited cells (~34 vs. 8.5 hr). We hypothesize that the observed increased binding of p15 to cdk4 and cdk6 in these cells stabilizes the p15 protein. This binding correlated with loss of D1, p27 and p21 associated cdk4/6 and increased p27 association with E/cdk2 complexes.

We did not detect any inherent differences in the p15 protein in the inhibited vs. resistant cells. p15 isolated from either could displace in vitro D1, p27 and p21 from cdk4 complexes isolated from inhibited cells. However, neither p15 could disrupt these preformed complexes from resistant cell lysates. It is therefore most likely that the p15 differences in the inhibited vs. resistant cells are secondary to a change in the D1/cdk4-6/p27/p21 complexes which alters the ability of p15 to bind the cdk. It is also possible that an alteration in p27 increases its affinity for cdk2 vs. cdk4-6.

VII. Other Properties of HMEC System

Index

Of the variety of studies we and others have done on these HMEC cultures, I want to include two topics in this review that, at various times, have been the subject of extensive investigation, and may be relevant to work of others.

VII.A. Metabolism of Chemical Carcinogens

(references: Stampfer et al., 1981; Bartley at al., 1982; Bartley and Stampfer, 1985, Leadon et al., 1988)

Index

Before we began the studies on BaP transformation of normal HMEC, we had performed extensive studies on the capacity of human mammary epithelial and fibroblastic cells to metabolize the polycyclic aromatic hydrocarbon (PAH) class of procarcinogens. PAH like BaP require a series of metabolic steps for conversion of the inactive procarcinogen into the active, ultimate carcinogenic form, the diol-epoxide, which is capable of forming bulky adducts with DNA. Since the extent and pattern of BaP metabolism can vary greatly among species, as well as among different individuals and cell type within one specie, we examined the rate and path of BaP metabolism in cells from many individuals.

Our results indicated that HMEC are extremely active in metabolizing BaP through the pathways that lead to the 7,8-diol-9,10-epoxide, which can form adducts with the deoxyguanosine of DNA. In contrast, the same concentration of BaP given to fibroblast cells from the same person's breast tissue yielded a much slower rate of metabolism, mainly through pathways that do not lead to the diol-epoxide, and a much lower production of DNA adducts. At the time these studies were performed, it was surprising to observe this high degree of PAH metabolism by non-liver tissues. Whether this has any bearing on in vivo transformation of HMEC is still unknown. Of possible note is the fact that the breast consists largely of adipose tissue, in which the lipid soluble PAH can concentrate. Comparisons of BaP metabolite products from 22 different specimen donors showed around a 5-fold range in values. Thus, this is a situation where individual variability needs to be considered. Additionally, we found that culture conditions can significantly influence the metabolites formed. The pattern of metabolites could vary with medium used (MM vs. MCDB 170), passage level pre- or post-selection MCDB 170 cells) and use of sub-optimal culture conditions (confluent or overly acidic cell cultures).

We also showed that the damage resulting from BaP metabolism may be due to oxidative damage as well as bulky adduct formation. We found that the lethal effect of BaP correlated with the extent of thymine glycol formation (a measure of oxidative damage) rather than bulky adduct formation, and could be reduced by agents which protect against oxidative damage, such as superoxide dismutase.

VII.B. Calmodulin-Like Protein

(references, Yaswen et al., 1990; Yaswen et al., 1992; Edman et al., 1995)

Index

One approach we took to characterize the differences between our normal and transformed HMEC cultures was to use subtractive hybridization to identify genes that are expressed in the normal HMEC and are downregulated in the immortal and malignantly transformed cells. Subtractive hybridization was performed between post-selection 184 cell cDNA and both 184B5 and B5KTu cell mRNA. In addition to identifying fibronectin, keratin 5, and vimentin, a 350 base pair cDNA fragment was isolated which initially showed no similarity to any sequence reported in GenBank. This cDNA hybridized specifically to a 1.4 kb mRNA, designated NB-1, which was expressed in normal HMEC, but was downregulated (184B5) or undetectable (184A1) in the transformed cell lines. Sequence analysis of a full length NB-1 clone revealed a 447 bp open reading frame with extensive similarity (70%, 71%, and 80%) at the nucleic acid level to the three known human genes coding for the ubiquitous calcium binding protein, calmodulin. The similarity between the translated amino acid sequence of NB-1 and human calmodulin was 85% over the length of the entire protein.

Using Northern and PCR analysis, NB-1 mRNA has been thus far found only in normal epithelial cells and tissues from human breast, prostate, cervix, and skin. It has not been found in normal epithelial cells other than those from stratified or pseudo-stratified tissues. It was not detectable in non-epithelial cells and tissues, nor in any of the mammary epithelial tumor cell lines which we have examined. Human breast cells obtained from lactational fluids were also negative for NB-1 expression by PCR analysis.

Expression of NB-1 mRNA is not significantly decreased when cells are growth arrested by exposure to anti-EGF receptor antibodies or in senescing cells where proliferation is minimal. It is increased in cells growth arrested by TGFb and reduced when HMEC are grown on reconstituted extracellular matrix material.

Using antisera which displayed a strong preference for NB-1 protein over calmodulin., the level of endogenous NB-1 protein in 184 HMEC was approximately 100-200 ng/106 cells, a level similar to the estimated level of calmodulin in other cultured cell lines. The relative abundance of the 16 kD NB-1 protein (named calmodulin like protein, or CLP) reflected relative NB-1 mRNA levels in various cell types. In contrast, levels of calmodulin protein were nearly constant in the same cell extracts.

Using indirect immunoflourescence, CLP was shown to be present diffusely throughout the cytoplasm and, to varying degrees, in the nuclei of 184 interphase cells. uring mitosis, CLP was particularly bright in regions around mitotic spindles. In 184B5, CLP expression was heterogeneous both among different cells and within

individual cells; no significant CLP immunofluorescence was observed in 184A1. In surgical specimens from histologically normal breast tissues, CLP staining was strong in the majority of basal cells from small ducts. Luminal cells in the small ducts showed some staining, although not as intense. In larger ducts, staining was mainly confined to the basal cells and was generally less intense than in the small ducts. In all cases, distribution of the protein appeared uniformly intracellular. No staining was evident in basement membrane or stromal areas. In contrast to normal breast tissue, sections from six infiltrating ductal breast carcinomas were consistently negative for CLP expression. Serial sections of the normal and tumor tissues all showed abundant calmodulin expression.

CLP distribution has also been examined in other stratified and pseudo-stratified epithelial tissue sections. In normal prostate, nearly all the epithelial cells were stained to a similar degree. In normal cervix and skin, no staining was observed in the basal cell layer. In the cervix, suprabasal cells were intensely CLP positive, with the degree of staining diminishing in the more distant upper layers. In the skin, the intensity of staining increased from the suprabasal layer until the stratum corneum, which itself was not stained. Thus, in the four different tissues examined, CLP showed distinct patterns of expression. These results suggest that the role of CLP may be defined by the differentiated state of the cells where it is expressed.

An unusual feature of CLP genomic DNA is the absence of introns, whereas all vertebrate calmodulin genes studied to date contain five similarly placed introns. It is possible that CLP may be a rare example of an expressed retroposon.

External calcium concentration has been shown to affect the proliferative potential and differentiated states of some cultured epithelial cells, including keratinocytes and mammary epithelial cells. In normal keratinocytes, increasing calcium concentrations can lead to cessation of proliferation and expression of markers of terminal differentiation, and loss of response to the calcium induced differentiation signal has been shown to correlate with the early stages of transformation. The downregulation of CLP expression observed after in vitro and in vivo transformation of HMEC may reflect a consequence of, or a requirement of the transformed state. Possibly, a particular state of differentiation is required for transformation to occur, or the transformed state may be incompatible with high expression of CLP.

VIII. Information on HMEC Computer Records, Mailing Sheets, and Distribution

Index

In order to accurately record both the many varieties of HMEC being generated and used in my lab, and the cell cultures distributed to others, it became acutely necessary to develop appropriate record keeping practices, e.g.,

(1) A complex relational database (4^th Dimension for the Mac) for recording frozen cell culture inventories and information.

(2) A simple database (Panorama) for recording cell cultures distributed to other laboratories.

(3) Mechanisms to disseminate informational material on cell usage.

(4) Standardized record keeping formats for my lab. More details are presented below.

VIII.A. Cell Inventory Database

Index

When I first starting collecting and freezing cell material in the days before personal computers, records were maintained on index cards on rolodexes. This was obviously not an ideal format, particularly as the number and complexity of cell types increased. We acquired a PC in 1985 and I worked with a programmer to develop a suitable inventory database using DBase. Since I’m not familiar with computer programming, DBase on a PC was not an optimal situation for me. As soon as the Macintosh II and the 4th Dimension database were available in 1987, the existing program and records were transferred and refined. I worked closely with a computer programmer to design the 4th Dimension program, which is flexible enough to allow me to make adjustments in layout, data organization, and ways to select and present data. This Database has subsequently undergone two major updates and improvements.

The database consists of two main related files. One file (Inventory) records the complete identity of each frozen cell batch (identified by Specimen ID, Cell, Tissue, Passage #, and, where relevant, also Type, Subtype, FreezeDown Symbol, Selection, Transduced Genes, Viral Vector, Promoter, Growth Medium), the number of ampoules of that batch which were made and which remain, the location of that batch of ampoules in the freezer, and spaces for comments and experimental use. Whenever cells are frozen, a test ampoule is included. When the test (or the first ampoule of that batch) is removed it is scored for viability, health, and the number of days it takes to reach confluence. This information is then entered into the Inventory file. Unfortunately, not all batches have been tested, so sometimes I end up sending cells whose viability has not been ascertained. All freeze downs are also tested for mycoplasma contamination by Hoechst staining, and the results of this testing are entered into the Inventory file. This program allows for easy selection and sorting on any field.

The second file (Location) records the location of each individual ampoule. When an ampoule is removed, it records and files the information on date, purpose for removal, and who removed them. It also updates the Inventory file records to indicate the reduced number of ampoules remaining.

In both files, all ampoules are identified by a 5 digit code, the FreezeDownNumber (FDN) you see on the Mailing sheets I send. The purpose of this code is to encapsulate all of the information above into 5 digits that can be easily written on an ampoule and stored in a computer. IT DOES NOT BY ITSELF IDENTIFY THE CELLS. PLEASE, DO NOT REFER TO THE CELLS BY THIS CODE! I have no idea what cells you're talking about without checking the database. These code numbers have even crept into publications - a truly confusing situation. The cells should be referred to by their Specimen Identification, Type, and Subtype (sent to you on the mailing sheets, see below).

VIII.B. Cell Distribution Database

Index

I use simple databases I created in Panorama to keep track of cells sent to other investigators, (Recipients file) and to keep investigator addresses and information. The Recipients file generates the mailing sheets that go along with the distributed HMEC. These databases allow for easy selection on all fields. Figure 8.1 gives an example of these mailing sheets and explanations of the categories.

Figure 8.1 Example of a HMEC Mailing sheet.

VIII.C. Distribution of Cells and Information about our HMEC System

Index

I first tried to disseminate HMEC information through mailed Newsletters to those on my Investigator list. Basically, I wanted to ensure that a certain level of information about the cells was given to each investigator. Additionally, I thought it might be helpful for all those using the same cell system to be aware of what others labs were doing. Output of the Newsletters was rather sporadic (they’re posted also on the web site, see Newsletters). One of the goals of this website was to replace the Newsletters. Thus, this review is intended to provide the basic information (and more, much more!) about the HMEC system, and the Procedures section provides the methods for use of the cells.

Given the large numbers and types of cells available, I prefer to talk individually with each person desiring cell cultures to determine what is most appropriate for their needs. If you want cells, you will be asked to send a brief (1 page) letter describing your planned experiments, and indicating that:

(1) you will keep me informed of results or major changes in planned experiments;

(2) you will not give the cells to others without my permission.

(3) Provide a FedEx number or equivalent to charge the costs of shipping the cells, as well as appropriate FedEx address and telephone number. For shipments outside the US, you will need to ensure that all proper customs forms and delivery arrangements are made.

Let me know in you want to be on the Investigators List.

Contact our Technology Transfer Department to obtain the approporiate MTA forms from the University of California for you and your institution to sign and return (legal forms).

We prefer to send out frozen cell cultures; these are placed in dry ice and send overnight. We can also send live cells at room temperature if absolutely necessary, but we seriously discourage this alternative as it is much more work for us. A mailing sheet will be sent with the cells (see above). We have not tested every freezedown (particularly the more obscure) so occasionally we will be sending cells that have not been tested for viability. I will note this on the mailing sheet. Otherwise, the cells that are being sent are known to be capable of good growth under the appropriate culture conditions.

General Cell Culture Reminders:

Taking care of normal human epithelial cells in culture bears some resemblance to taking care of children - the cells may behave by their own logic and timing, which may not coincide with that of the care provider. To optimize the accuracy and consistency of experimental results, the cells’ needs have to come first. Some of the important things to remember for these HMEC:

(1) pH must be carefully controlled. The color of the pH indicator should be around salmon-orange. Yellow indicates too acid conditions; in my experience cells left at such acidity become irreversibly sick. The HMEC quickly acidify the culture medium, particularly when near confluent. We change the medium every 2 days (3 on

weekends) and refeed a culture 24 hrs before subculture or experimental usage. Your results may differ if you use cells that are acidic or haven't been fed in a while.

(2) The cells do not stay healthily once they become confluent. They should be subcultured when subconfluent or just confluent. We use subconfluent cultures for most biochemical and molecular studies. Your results may differ if you use confluent (non-proliferating) cultures.

(3) Some cell biology changes as a function of age in culture. It is best to repeat experiments using cells at around the same passage level, with the same life expectancy. This is also a good practice for the cell lines, which may change over extended periods of time in culture. Your results may differ if you use cells at very different passage levels.

(4) It's really helpful to look at the cells frequently, to become familiar with how they appear under different circumstances. An enormous amount of useful information can be gleaned simply by careful visual observation. If something doesn't look right, it probably isn't, and should be investigated immediately.

ABBREVIATIONS USED:

Index

AIG: anchorage independent growth

BaP: benzo(a)pyrene

CFE: colony forming efficiancy

CKI: cyclin dependent kinase inhibitor

CLP: calmodulin like protein

EGF: epidermal growth factor

EGFR: epidermal growth factor receptor

ENU: N-nitroso-ethyl-urea

GSE: genetic suppressor element

HMEC: human mammary epithelial cells

HPV: human papilloma virus

LI: labeling index

LOH: loss of heterozygosity

p: passage

PAH: polycyclic aromatic hydrocarbon

PD: population doublings

PEM: polymorphic epithelial mucin

RB: retinoblastoma

TRF: terminal restriction fragment

REFERENCES (from our lab and collaborations)
Index

Bartley, JC, Bartholomew, JC, Stampfer, MR, Metabolism of Benzo(a)pyrene in Human Mammary Epithelial and Fibroblast Cells: Metabolite Pattern and DNA Adduct Formation. J. Cell. Biochem. 18:135-148, 1982.

Bartley, JC, Stampfer, MR, Factors influencing benzo(a)pyrene metabolism in human mammary epithelial cells. Carcinogenesis 6:1017-1022, 1985.

Bates, SE, Valverius, E, Ennis, BW, Bronzert, DA, Sheridan, JP, Stampfer, M, Mendelsohn, J, Lippman,
ME, Dickson, RB, Expression of the TGFa/EGF receptor pathway in normal human breast epithelial cells. Endocrin. 126: 596-607, 1990.

Brenner, AJ, Stampfer, MR, Aldaz, M, Increased p16 expression with first senescence arrest in human mammary epithelial cells and extended growth capacity with inactivation, Oncogene, 17:199-205, 1998.

Clark, R, Stampfer, M, Milley, B, O'Rourke, E, Walen, K, Kriegler, M, Kopplin, J, McCormick, F, Transformation of human mammmary epithelial cells by oncogenic retroviruses. Cancer Res. 48:4689-4694, 1988.

Dairkee, SH, Deng, G., Stampfer, MR, Waldman, FM, Smith, HS, Selective cell culture of primary breast carcinoma, Cancer Res. 55:2516-1519, 1995.

Frittitta, L, Vigneri, R, Stampfer, MR, Goldfine, ID, Insulin receptor overexpression in 184B5 human mammary epithelial cells induces a ligand-dependent transformed phenotype, J. Cell. Biochem. 57:666-669, 1995.

Garbe, J, Wong, M, Wigington, D, Yaswen, P, Stampfer, MR, Viral oncogenes accelerate conversion to immortality of cultured conditionally immortal human mammary epithelial cells, Oncogene, 18:2169-2180, 1999.

Garbe, JC, Holst, C, Tlsty, T, Yaswen, P, and Stampfer, MR, Inactivation of p53 function turns to telomere-based proliferation barrier from agonescence into crisis, in preparation.

Hammond, SL, Ham, RG, and Stampfer, MR, Serum-free growth of human mammary epithelial cells: Rapid clonal growth in defined medium and extended serial passage with pituitary extract. PNAS (USA) 81:5435-5439, 1984.

Hosobuchi, M, Stampfer, MR, Effects of Transforming Growth Factor-b on growth of human mammary epithelial cells. In Vitro 25:705-713, 1989.

Leadon, SA, Stampfer, MR, Bartley, J, Production of thymine glycols during the metabolism of benzo(a)-pyrene by human mammary epithelial cells. Proc. Natl. Acad. Sci. (USA), 85:4365-4368, 1988.

Lehman T, Modali R, Boukamp P, Stanek, J., Bennett, WP, Welsh, JA, Metcalf, RA, Stampfer, MR, Fusenig, N, Rogan, EM, Reddel, R, Harris, CC, p53 mutations in human immortalized epithelial cell lines. Carcinogenesis 14, 833-839, 1993.

Nijjar, T, Wigington, D, Garbe, JC, Waha, A, Stampfer, MR, Yaswen, P, p57KIP2 expression and loss of heterozygosity during immortal conversion of cultured human mammary epithelial cells, Cancer Res,:59, 5112-5118, 1999.

Nonet, GH, Stampfer, M., Chin, .,Gray, JW, Collins, CC, and Yaswen, P, The ZNF217 Gene amplified in breast cancers promotes immortalization of human mammary epithelial cells, Cancer Res. 61: 1250-1254, 2001.

Olsen CL, Gardie, B, Yaswen, P, Stampfer, MR, Raf-1-induced growth arrest in human mammary epithelial cells is p16-independent and is overcome in immortal cells during conversion, Oncogene 21:6328-6339 2002.

Romanov, SR, Kozakiewicz, K, Holst, CR, Stampfer, MR, Haupt, LM, and Tlsty, TD, Normal human mammary epithelial cells spontaneously escape senescence and acquire genomic changes, Nature, 409:633-637, 2001.

Sandhu, C, Garbe, J, Bhattacharya, N, Daksis, J, Pan, C-H, Yaswen, P, Koh, J, Slingerland, JM, Stampfer, MR, TGF-b increases p15INK4B protein and p15INK4B/cdk4 complexes and prevents cyclin D1/cdk4 association in human mammary epithelial cell, Mol. Cell Biol. 17:2458-2467, 1997.

Slingerland, JM, Hengst, L, Pan, C-H., Alexander, D, Stampfer, MR, Reed, SI, A novel inhibitor of cyclin/cdk activity detected in TGF-b arrested cell, Molecular Cell Biology 14: 3683-3694, 1994.

Stampfer, MR, Hallowes, R, and Hackett, AJ, Growth of normal human mammary epithelial cells in culture. In Vitro 16:415-425, 1980.

Stampfer, M.R., Vlodavsky, I., Smith, H.S., Ford, R., Becker, F.F., and Riggs, J., Fibronectin production by human mammary cells. J. Natl. Cancer Inst. 67:253-261, 1981.

Stampfer, MR, Bartholomew, JC, Smith, HS, and Bartley, J, Metabolism of benzo(a)pyrene by human mammary epithelial cells: Toxicity and DNA adduct formation. PNAS. (USA) 78:6251-6255, 1981.

Stampfer, MR, Cholera toxin stimulation of human mammary epithelial cells in culture. In Vitro 18:531-537, 1982.

Stampfer, M, Bartley, JC, Induction of transformation and continuous cell lines from normal human mammary epithelial cells after exposure to benzo(a)pyrene. PNAS. (USA) 82:2394-2398, 1985.

Stampfer, M.R., Isolation and Growth of Human Mammary Epithelial Cells. J. Tissue Culture Methods 9:107-116, 1985.

Stampfer, MR, Taylor-Papadimitriou, J, and Yaswen, P, Culture of human mammary epithelial cells, in Culture of Epithelial Cells, 2nd Edition, Ed. I. Freshney, Wiley-Liss, Somerset, NJ. pp95-135, 2002.

Stampfer, MR, Yaswen, P, Factors influencing growth and differentiation of normal and transformed human mammary epithelial cells in cultures. In: G.E. Milo, B.C. Castro, C.F. Shuler (eds.), Transformation of Human Epithelial Cells: Molecular and Oncogenetic Mechanisms, 117-140. CRC Press, 1992.

Stampfer, MR, Yaswen, P, Alhadeff, M, Hosoda, J, TGFb induction of extracellular matrix associated proteins in normal and transformed human mammary epithelial cells in culture is independent of growth effects. J. Cell. Physiol. 155, 210-221, 1993.

Stampfer, MR, Pan, C-H, Hosoda, J, Bartholomew, J, Mendelsohn, J, Yaswen, P, Blockage of EGF receptor signal transduction causes reversible arrest of normal and immortal human mammary epithelial cells with synchronous reentry into the cell cycle. Exp. Cell Res. 208: 175-188, 1993.

Stampfer, MR and Yaswen, P, Culture Models of Human Mammary Epithelial Cell Transformation, J. Mam. Gland Bio. Neo. 5:365-378, 2000.

Stampfer, MR, Bodnar, A, Garbe, J, Wong, M, Pan, A, Villeponteau, B, Yaswen, P, Gradual phenotypic conversion associated with immortalization of cultured human mammary epithelial cells, Mol. Biol. Cell 8:2391-2405, 1997.

Stampfer, MR, Garbe, J, Levine, G, Lichtsteiner, S, Vasserot, AP, and Yaswen, P, hTERT expression can induce resistance to TGFb growth inhibition in p16INK4A(-) human mammary epithelial cells, Proc. Natl. Acad. Sci. (USA), 2001.

Stampfer, MR, Garbe, JC, Nijjar, T, Wigington, D, Swisshelm, K, and Yaswen, P, Loss of p53 function accelerates acquisition of telomerase activity in indefinite lifespan human mammary epithelial cell lines, Oncogene, in press, 2003.

Stampfer, MR and Yaswen, P, Immortal transformation and telomerase reactivation of human mammary epithelial cells in culture, in: Advances in Cell Aging and Gerontology, Volume 8: Aging and Disease, editors M.P. Mattson and T. Pandita, Elsevier, Amsterdam, pp103-130, 2001.

Stampfer, MR and Yaswen, P, Human epithelial cell immortalization as a step in carcinogenesis, Cancer Letters, in press.

Taylor-Papadimitriou, J, Stampfer, M, Bartek, J, Lewis, A, Boshell, M, Lane, E., Leigh, IM, Keratin expression in human mammary epithelial cells cultured from normal and malignant tissue: Relation to in vivo phenotypes and influence of medium. J. Cell Sci. 94:403-413,1989.

Tlsty, TD, Romanov, SR, Kozakiewicz, BK, Holst, C R, Haupt, LM, Crawford, YG, Loss of chromosomal integrity in human mammary epithelial cells subsequent to escape from senescence. J. Mammary Gland Biol Neoplasia, 6: 235-243, 2001.

Thompson, EW, Torri, J, Sabol, M, Sommers, CL, Byers, S, Paik, S, Martin, GR, Lippman, ME, Valverius, E, Stampfer, MR, Dickson, RB, Oncogene-induced basement membrane invasiveness in human mammary epithelial cells, Clin. Exp. Metast. 12:181-194, 1994.

Walen, KH, Stampfer, MR, Chromosome analyses of human mammary epithelial cells (HMEC) at stages of chemically-induced transformation progression to immortality. Cancer Genet. Cytogen.. 37:249-261, 1989.

Yaswen, P, Smoll, A, Hosoda, J, Parry, G, Stampfer, MR, Protein product of a human intronless calmodulin-like gene shows tissue-specific expression and reduced abundance in transformed cells. Cell Growth & Diff., 3: 335-345, 1992.

Yaswen, P, Smoll, A, Peehl, D, Trask, DK, Sager, R, and Stampfer, MR, Downregulation of a novel calmodulin-related gene during transformation of human mammary epithelial cells. Proc. Natl. Acad. Sci. USA, 87:7360-7364, 1990.

Yaswen, P, Stampfer, MR, Ghosh, K, Cohen, JS, Effects of sequence of thioated oligonucleotides on cultured human mammary epithelial cells. Antisense Research and Development 3, 67-77, 1993.

Yaswen, P, and Stampfer, MR, Epigenetic changes accompanying human mammary epithelial cell immortalization, J. Mam. Gland Bio. Neo. 6: , 2001.

Yaswen, P and Stampfer, MR,  Molecular changes accompanying senescence and immortalization of cultured human mammary epithelial cells, Int. J. Biochem. Cell Biol. 34:1382-1394 2002.

References for Sections

Section II. B.

M.A. Dickson, W.C. Hahn, Y. Ino, V. Ronfard, J.Y. Wu, R.A. Weinberg, D.N. Louis, F.P. Li, J.G. Rheinwald, Human keratinocytes that expess hTERT and also bypass a p16^INK4a-enforced mechanism that limits lifespan become immortal yet retain normal growh and differentiation characteristics. Mol. Cell Biol. 20, 1436-1447, 2000.

J.G. Rheinwald, W.C. Hahn, M.R. Ramsey, J.Y. Wu, Z. Guo, H. Tsao, M. De Luca, C. Catricala, K.M. O'Toole, A two-stage, p16^INK4a-and p53-dependent keratinocyte senescence mechanism that limitsreplicative potential independent of telomere status. Mol. Cell Biol. 22, 5157-5172, 2002.

S. Drayton G. Peters, Immortalisation and transformation revisited. Curr. Op. Gen. Dev. 12 (2002) 98-104. W.E. Wright J.W. Shay, Historical claims and current interpretations of replicative aging. Nat. Biotech. 20, 682-688, 2002.

S.A. Foster D.A. Galloway, Human papillomavirus type 16 E7 alleviates a proliferative block in early passage human mammary epithelial cells. Oncogene 12, 1773-1779, 1996.

G.P. Nielsen, A.O. Stemmer-Rachamimov, J. Shaw, J.E. Roy, J. Koh, D.N. Louis, Immunohistochemical survey of p16^INK4A expression in normal human adult and infant tissues. Lab. Invest. 79, 1137-1143, 1999.

Section II. D.

Chang, S. E., and Taylor-Papadimitriou, J., Modulation of phenotype in cultures of human milk epithelial cells and its relation to the expression of a membrane antigen, Cell Diff. 12:143, 1983.

Bartek, J., Taylor-Papadimitriou, J., Miller, N., and Millis, R., Pattern of expression of keratin 19 as detected with monoclonal antibodies to human breast tumors and tissues, Int. J. Cancer 36:299, 1985.

Bartek, J., Durban, E. M., Hallowes, R. C., and Taylor-Papadimitriou, J., A subclass of luminal epithelial cells in the human mammary gland, defined by antibodies to cytokeratins, J. Cell Sci. 75:17, 1985.

Taylor-Papadimitriou, J., Millis, R., Burchell, J., Nash, R., Pang, L., and Gilbert, J., Patterns of reaction of monoclonal antibodies HMFG-1 and -2 with benign breast tissues and breast carcinomas, J. Exp. Pathol. 2:247, 1986.

Guelstein, V. I., Tchypysheva, T. A., Ermilova, V. D., Litvinova, L. V., Troyanovsky, S. M., and Bannikov, G. A., Monoclonal antibody mapping of keratins 8 and 17 and of vimentin in normal human mammary gland, benign tumors, dysplasias and breast cancer, Int. J. Cancer 42:147, 1988.

Rudland, P. S., and Hughes, C. M., Immunocytochemical identification of cell types in human mammary gland: Variations in cellular markers are dependent on glandular topography and differentiation, J. Histochem. Cytochem. 37:1087, 1989.

Petersen, O. W., Hoyer, P. E., and van Deurs, B., Frequency and distribution of estrogen receptor-positive cells in normal, non-lactating human breast tissue, Cancer Res. 47:5748, 1987.

Ricketts, D., Turnbull, L., Ryall, G., Bakhshi, R., Rawson, N. S. B., Gazet, J.-C., Nolan, C., and Coombes, R. C., Estrogen and progesterone receptors in the normal human breast, Cancer Res. 51:1817, 1991.

Section III.

Pierce, J. H., Arnstein, P., DiMarco, E., Artrip, J., Kraus, M. H., Lonardo, F., DiFiore, P. P., Aaronson, S. A., Oncogenic potential of erbB-2 in human mammary epithelial cells. Oncogene 6:1189-1194, 1991.

Zhai Y-F, Beittenmiller H, Wang B, et al. Increased expression of specific protein tyrosine phosphatases in human breast epithelial cells neoplastically transformed by the neu oncogene. Cancer Res. 53:2272-2278, 1993.

Eldridge SR and Gould MN, Comparison of spontaneous mutagenesis in early-passage human mammary cells from normal and malignant tissues. International J. Cancer 50:321-324, 1992.

Eldridge SR, Martens TW, Sattler CA and Gould MN, Association of decreased intercellular communication with the immortal but not the tumorigenic phenotype in human mammary epithelial cells. Cancer Res. 49:4326-4331, 1989.

Delmolino, L., Band, H., Band, V., Expression and stability of p53 protein in normal human mammary epithelial cells. Carcinogeneisis 14:827-832, 1993.

Band, V., Dalal, S., Delmolino, L., Androphy, E. J., Enhanced degradation of p53 protein in HPV-6 and BPV-1 E6-immortlaized human mammary epithelial cells. EMBO J. 12:1847-1852, 1993.

Section IV.

Meeker, A. K., Hicks, J. L., E.A., P., March, G. E., Bennett, C. J., Delannoy, M. J., and De Marzo, A. M. Telomere shortening is an early somatic DNA alteration in human prostate tumorigenesis. Cancer Res., 62: 6405-6409, 2002.

Section V.

K. Keyomarsi, Sandoval, L., Band, V., Pardee, A.B. Synchronization of tumor and normal cells from G1 to multiple cell cycles by lovastatin. Cancer Res. 51: 3602-3609, 1991.

Section VII.A.

A. Tischler, G. Levine, J.C. Bartley, Metabolism of benzo(a)pyrene and benzo(a)pyrene-7,8-dihydrodiol in human mammary epithelial cells: feedback inhibition by 7-hydroxybenzo(a)pyrene. Carcinogenesis 12: 1539-1543, 1991.

Section VII.B.

Edman, C.F., George, S.E., Means, A.R., Schulman, H., and Yaswen, P. Selective activation and inhibition of calmodulin dependent enzymes by a calmodulin-like protein found in human epithelial cells. Eur. J. Biochem. 226: 725-730, 1994.

Harris, E., Yaswen, P., and Thorner J. Gain-of-function mutations in a human calmodulin-like protein identify residues critical for calmodulin action in yeast. Mol. Gen. Genetics 247:137-147, 1995.

II.C. Growth after Inhibition of p53 Function

II. D. Characterization of Cultured HMEC Compared to Tissue of Origin

III. In Vitro Transformation of HMEC Following Benzo(a)pyrene Exposure

III.A. Derivation of Extended Life Cultures

III.B. Derivation of Cell Lines 184A1, 184B5, and 184AA4

III.C. Derivation of Early Variants of 184A1 and 184B5

Table 2. Growth Factor Requirements of Normal and Transformed HMEC in MCDB170

III.D. Characterization of 184A1, 184B5, and Derivatives Compared to Tissue of Origin

IV.A.4. Abrogation of Activated Raf-1-induced Growth Inhibition during Conversion

IV.A.5. Other Changes Associated with Conversion

IV.C.1. Generation of p53(-/-) HMEC lines 184AA2 and 184AA3

Cell Line

184A1

184AA2

184AA3

184AA4

IV.C.2. Attainment of Full Immortality in p53(-) HMEC

IV.C.3. Effect of Inhibition of p53 Function in p53(+) HMEC Lines

IV.D. Effects of ectopic expression of hTERT in finite lifespan and conditionally immortal HMEC

IV.E. Overcoming Agonescence and Genomic Instability

IV.F.1. Speculations about mechanisms

V. Synchronization of HMEC Cultures and Role of EGF Receptor Signal Transduction

VI. TGFb Effects on Finite and Immortally Transformed HMEC

VII. Other Properties of HMEC System

VII.A. Metabolism of Chemical Carcinogens

VII.B. Calmodulin-Like Protein

VIII. Information on HMEC Computer Records, Mailing Sheets, and Distribution

VIII.A. Cell Inventory Database

VIII.B. Cell Distribution Database

mrstampfer@lbl.gov | Index | back