In vitro analysis of multistage carcinogenesis.

Several key events in the multistep process of neoplastic transformation of rat tracheal epithelium (RTE) are described. Whether tracheal epithelium is exposed in vivo to carcinogenic agents or whether primary tracheal epithelial cells are exposed in vitro to carcinogens, initiated stem cells can be detected soon after the exposure by their ability to grow under selective conditions in culture. These initiated stem cells differ fundamentally from normal stem cells in their response to factors normally constraining proliferation and self-renewal. Thus, disruption of inhibitory control mechanisms of stem cell replication appears to be the first event in RTE cell transformation. While the probability of self-renewal (PSR) is clearly increased in initiated stem cells, most of the descendants derived from such stem cells differentiate and become terminal and do not express transformed characteristics. Progression from the first to the second stage of RTE cell transformation, the stage of the immortal growth variant (IGV), is characterized by loss of responsiveness to the growth-restraining effects of retinoic acid. In the third stage of neoplastic transformation, the stage during which neoplastic growth variants (NGV) appear, a growth factor receptor gene is inappropriately expressed in some of the transformants. Thus, it appears that loss of growth-restraining mechanisms may be an early event, and activation of a growth stimulatory mechanism a late event, in neoplastic transformation of RTE cells. ImagesFIGURE 3.FIGURE 12.FIGURE 13.


Introduction
The combined epidemiological, clinical, and experimental evidence suggests that the development of cancer occurs in discrete stages. According to current models of carcinogenesis such as the one presented in Figure 1 (1), carcinogen exposure does not instantly transform the affected target cells to cancer cells, rather, it causes heritable changes in a small proportion of stem cells, converting them to a preneoplastic state (internediate cells in Fig. 1). In the second stage of carcinogenesis, clonal expansion of the preneoplastic stem cell variants occurs. This can be brought about by exogenous promoting agents (such as phenobarbital, dioxins, or phorbol esters), by endogenous promoters (e.g., prolactin in mamma r carcinogenesis), or it may *Laboratory of Pulmonary Pathobiology, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709. t Terminology used: stem cell: a cell with self-replicating ability; we consider clonogenic cells to be stem cells. Initiated stem cell: the stem cell that has been heritably altered by exposure to a carcinogen and as a result of that exposure produces progeny with a greater than normal probability of becoming neoplastic. We consider the founder cell of the enhanced growth variant colony (EGV colony) to be an initiated stem cell. Preneoplastic cells: any cell whether appearing early or late during the multistage process of neoplastic transformation that has a greater than normal probability of giving rise to neoplastic offspring.  phenotypes to appear (e.g., invasiveness, metastatic ability, hormone independence). This model of careinogenesis is, as one might imagine, not uncontested. However, in principle, this model is supported by a great wealth of experimental data and by studies on heritable neoplastic diseases in humans such as familial retinoblastoma (2). For the purposes of this presentation, it may serve to provide the overall perspective and the hypothetical framework for the discussion of our own investigations. Our interest is primarily focused on the compartment of intermediate preneoplastic transfonnants (Fig. 1). What experimental evidence do we have that preneoplastic stem cells exist? Can they be isolated? How do they differ in their behavior from normal stem cells? What is the molecular basis of their phenotype? What is involved in the conversion of the preneoplastic to the neoplastic cell variant, a process we call progression, and what factors can accelerate or inhibit that progression? Preneoplastic cells that have not yet acquired malignant characteristics presumably are present only in small numbers. Thus, it may be possible to prevent the progression of preneoplastic cells to malignancy or to eliminate them from the host altogether through chemopreventive measures. The studies to be discussed are concerned with the following topics: the detection, isolation, and quantification of early, preneoplastic stem cells in rat tracheal epithelium; manifestations of aberrant growth control in these early transfornants and their abnormal self-renewal capacity; manifestations of progression, namely, the loss of responsiveness to negative growth regulators; and altered gene expression in late stages of neoplastic transformation. The examples used in this review to illustrate the manifestations of altered growth control of RTE cell transformants are taken from several in vitro transformation studies; however, transformants isolated from tracheal epithelium exposed in vivo to various types of carcinogens show the same growth alterations and the same sequence of phenotypic changes during progression from the preneoplastic to the neoplastic stages of transformation.

Detection, Isolation and Quantification of Early Preneoplastic Transformants
Carcinogen exposure causes cellular and molecular changes that, months and years later, result in the development of tumors. To investigate the clonal expansion and progression of the putative initiated stem cells during this long latency period, one must be able to isolate the stem cells and study their descendants over many cell generations. More than 10 years ago, studies in our laboratory (3,4) led to the discovery that tracheas ofrats exposed to carcinogens contain a small proportion ofcells exhibiting a fundamental derangement in growth control. This becomes evident when the cells from such tracheas are isolated and cultured. Epithelial cells ob-FIGURE 6. Effect of carcinogen dose and time after the end of exposure on the rate of appearance of advanced preneoplastic cell variants. Tracheas were exposed in vivo to various amounts of 7,12-dimethylbenz[a]anthracene for a period of 2 weeks. At differrent times after the end of exposure 10 to 20 tracheas were removed and the number of tracheas containing transformed stem cells giving rise to anchorage independent growth variants as well as the relative frequency of such transformed stem cells (i.e., AiGV/EGV %) was determined (3,4).
tained from tracheas of normal rats undergo approximately 10 population doublings in primary culture (Fig. 2); thereafter, they senesce and become terminal. In contrast, cells that have been exposed to carcinogens either in vivo or in vitro proliferate indefinitely. If the cells are seeded at clonal densities ( Fig. 3), one can observe many small colonies around 8 to 10 days after plating; however, subsequently, most of these senesce and only a few (0.5-2.0%) continue to expand and can be scored as "enhanced growth variant (EGV) clones" several weeks later (5). This is regarded as the first stage of RTE cell transformation (Fig. 4). These enhanced growth variant colonies can be isolated and subcultured. Approximately one-half of these clones become permanent cell lines, i.e., they undergo immortalization (IGV = immortal growth variants) and between 30 and 50% of the immortalized clones become neoplastic (NGV = neoplastic growth variants) after 5 to 20 passages. The anchorage-independent growth variant (AiGV), which is frequently observed when the tracheal epithelium is initiated in vivo, is not an obligatory preneoplastic phenotype. Thus we have identified three different stages of neoplastic transformation in cells isolated from rat tracheas exposed in vivo to carcinogens (EGV, IGV, NGV), and we can say with confidence that the early EG-variant clones are preneoplastic because they have an increased probability of becoming neoplastic (Fig. 5). Preneoplastic stem cells with EGV characteristics can be detected in the tracheal epithelium within a few days after in vivo carcinogen exposure. Obviously it is of great interest to learn about the fate of this cell compartment during the tumor latency period. Following a single carcinogen exposure, which causes an approximate 20% tumor incidence at about 24 months, the preneoplastic stem cell population expands (Fig. 6). As a function of time, the number of tracheas containing transformants increases and the number of transformants per trachea also increases. The growth ofthe transformned stem cell pool is carcinogen-dose dependent. It is important to note that the expansion of preneoplastic cells in the tracheal epithelium occurs in the absence of any exogenous stimulus and many months before tumors arise.
In summary, we showed that early preneoplastic cell variants can be isolated from carcinogen-exposed tracheal epithelium of rats and can be quantitated in cell culture; the number of these early transformants increases as a function of carcinogen dose and as a function of time after exposure; and their clonal descendants undergo secondary changes, most importantly, immortalization and neoplastic transformation. Other laboratories have observed similar findings: cells with characteristics of EG-variants have been isolated from a number of organs in rats (6) and from initiated mouse skin (7). transformation is a stem cell disorder. If this hypothesis is correct, we should be able to detect abnormal stem cell growth kinetics during early preneoplastic stages of neoplastic transformation. Before we describe the proliferation kinetics of EG-variant clones, it may be useful to briefly review a simple model of cell replication in differentiating tissues (Fig. 7). This model (8)  rying out the main differentiated functions of the respective tissue. In normal steady-state epithelia, most of the stem cells are believed to remain in Go and thus are noncycling. When they enter cell cycle, their probability for self-renewal (PSR) is 50%. During nonsteady-state conditions such as regenerative growth or neoplastic disease, the PSR increases and the stem cell compartment increases in size. In cell culture, one determines the size of the stem cell compartment by means of clonogenic assays that measure the proportion of clone-forming cells in a cell population. Typically, during the growth of normal primary rat tracheal cells (Fig. 8A), the number of clonogenic cells, or stem cells, in the population amounts to only a few percent during the early logarithmic growth phase, and as the clonal cell density increases and the cultures approach plateau of growth, the stem cell compartment decreases rapidly in size. Simultaneously the cultures produce terminally differentiated cells. Ultimately cell replication ceases and the cultures senesce. This dynamic relationship between clonogenic cells, committed cells, and terminally differentiated cells is fundamentally altered in transformed cell populations. To emphasize this change we have chosen to illustrate the growth dynamics of a preneoplastic, immortalized growth variant (IGV) (Fig. 8B); the growth rate is similar to that ofnormal cells during the logarithmic growth phase. However, subsequently, two crucial differences are noted: growth continues in spite of the high cell density being reached, and the clonogenic cell compartment does not decrease in response to high cell density (the same phenomenon is seen in early trans-  formants, EGV clones). It is clear that the regulation of the size of the stem cell compartment in these transformants is altered. It is important to note, however, that even in the immortal growth variant, only 10% of the cells are clonogenic, i.e., a substantial proportion of cells in these IG-variant cultures differentiate and become terninal (9,10). Thus, the transformed phenotype is transmitted to only a fraction of the descendants of the transformed stem cell. In the early transformants, namely the EG-variants, the stem cell compartment is even smaller (Fig. 9). Most of the transformed clones have S 1% cells with clonogenic potential, and at most 10 to 20%. Autoradiographic studies have shown that the growth fraction, i.e., all cells able to proliferate, is not more than -30%, and a large proportion of cells shows ultrastructural evidence of terminal differentiation (9). However, as the EG-variant clones progress and acquire unlimited growth capacity (i.e., become immortal) (Fig. 10), the fraction of EGV-stem cells increases disproportionately within each clone (10). The significance of this steady increase in the size of the preneoplastic stem cell compartment seems obvious: the larger the pool of proliferating stem cells, the greater the chance for a new cell variant to arise. This brings us to the third topic of our discussion, namely, the manifestations of progression.

Manifestations of Progression in Early Preneoplastic Clones: Loss of Responsiveness to Retinoic Acid
One can view neoplastic growth behavior as a consequence of abnormal growth stimulation (e.g., autocrine growth stimulation) or as a result of diminished growth restraints. There is in fact evidence for both mechanisms operating in neoplastic transfornation. We are especially interested in exploring the possible breakdown of negative growth control mechanisms as a causal event in early stages of transformation. One key growth and differentiation regulator ofnormal RTE cells in vivo is vitamin A and related retinoids. Withdrawal of vitamin A from the diet results in increased cell proliferation and squamous metaplasia of the tracheal epithelium (11). Thus, retinoic acid (RA) can be regarded as a physiological, negative growth regulator. It has also been shown to inhibit the development of skin, mammary, and bladder tumors in mice and rats (12). In our studies we found that RA inhibits growth of normal RTE cells and inhibits the development of RTE cell transformants when the cultures are treated as early as 1 day or as late as 20 days after carcinogen exposure (13). What is perhaps more interesting is that the sensitivity to RA of clonogenic cells isolated from transformed colonies (EG-variant colonies) decreases markedly with time (Fig. 11). The proliferation of cells isolated from 3-week-old transformants is readily inhib-  colony formation by 50%, is 0.1 nM); however, clonogenie cells from 9-and 12-week-old transformants become increasingly resistant (IC50 > 10 nM). We believe that this loss of RA sensitivity represents an important step in the progression of the early EG-variant to become an IG-variant, which escapes the growth regulatory function of RA. It is noteworthy that this change occurs concomitant with an increase in the size of the transformed stem cell pool described earlier. The biochemical and molecular mechanisms underlying these changes need to be elucidated. In summary, we would like to emphasize the following points: transformed stem cells produce a large proportion of nontransformed, terminal offspring; in transformed clones the stem cell pool size progressively and selectively increases with time, which is probably linked to immortalization of the clone; and the transformed stem cell progressively loses responsiveness to negative growth controlling factors such as high clonal cell density and retinoids.

Expression of Oncogenes during Neoplastic Conversion
The RTE cell system is well suited to examine the possible role of oncogenes during various stages of neoplastic transformation. We have begun to examine the expression of oncogenes in neoplastic RTE cell transformants. Five, independent, tumor-derived clones were tested with a battery of oncogene probes (Table  1). Of the panel of 15 oncogenes whose level of expression was measured in the tumor-derived lines relative to normal RTE cells, only two, H-ras and fms, showed a significant elevation in the transformed cells. H-ras elevation was threefold that observed in normal cells, and fms was elevated 5-to 20-fold in three of the five cell lines examined. The fms-related gene observed in these tumorigenic cell lines was expressed as an RNA transcript of 9.5 kb in length (Fig. 12).
The human fms gene encodes the receptor for macrophage colony-stimulating factor, or CSF-1 (15). The CSF-1 receptor/fms gene is expressed as a 4.0 kb transcript in humans and mice. To determine the size of the CSF-1 receptor/fms mRNA in rats, we isolated poly A' RNA from rat alveolar macrophages, which express the CSF-1 receptor. We found that the CSF-1 receptor/fms gene was expressed as a 3.8 kb transcript in these cells (Fig. 13), comparable to the transcript size observed in humans and mice. Thus, the gene expressed as a 9.5 kb transcript in the epithelial cell lines is related to, but distinct from, the CSF-1 receptor/fms gene. We speculate that this gene encodes a related growth factor receptor, possibly for another hematopoietic growth factor.
Because epithelial cells retain in their normal repertoire the capacity to synthesize and secrete various hematopoietic growth factors such as granulocyte colony-stimulating factor, if the fms related gene encodes the receptor for one of these growth factors, the possibility of autocrine transformation of cells expressing the fms related gene exists. This could be an important clue to the mechanisms underlying the neoplastic behavior of these cells. Conceivably, the growth factor responsible for the unregulated growth of these cells may be secreted by the neoplastic cells themselves.
Thus, at the early stages of RTE cell transformation, loss of responsiveness to a negative growth regulator, namely, retinoic acid, may be a key factor in the progression from enhanced growth variant to immortal growth variants; enhanced expression ofa growth factor receptor gene may be a crucial mechanism in the expression of the neoplastic phenotype. Obviously, much work needs to be done before we know what role the fmsrelated gene plays in the mechanism of transfomation of RTE cells. Studies to find the ligand for this putative growth factor receptor gene and determine at what stage of neoplastic transformation the gene is first expressed are currently underway.
Our findings suggest that at least two distinct mechanisms may be involved in the multistage process of neoplastic transformation of rat tracheal epithelial cells: loss of responsiveness to physiological growth restraining factors in the early stages of transformation, and activation of growth stimulatory mechanisms resulting from inappropriate expression of growth factor receptors in late stages of transformation.