I. INTRODUCTION

Adenomatosis of the colon and rectum (ACR) is an autosomal dominant genetic disease characterized by the development of hundreds to thousands of pre-malignant polyps in the colon and rectum (Bussey, 1975). Gardner's syndrome, a major subgroup within ACR, combines multiple colonic adenomas with extracolonic lesions including; multiple osteomas of the skull and mandible, multiple epidermoid cysts, and soft tissue tumors of the skin (Gardner, 1953). Unless prophylactic measures are taken (total proctocolectomy and ileostomy or ileoproctostomy followed by diathermic destruction of rectal polyps) an individual with ACR is at great risk for colorectal cancer. Although the exact molecular mechanism which renders individuals with the ACR gene to increased risk for colorectal polyps and cancer has yet to be determined, in vitro studies of biopsied tissue samples and cell cultures obtained from such individuals have been and will continue to be invaluable aids in the elucidation the genetic and molecular nature of the disease. In order to be useful in this respect, however, the results of such studies must necessarily be reproducible by independent investigators. The present study is concerned with particular aspects of two of the many in vitro abnormalities which have been reported to exist in ACR dermal fibroblasts: specifically the reported reduced serum requirement for growth; and the reported reduced or absent density-dependent inhibition of growth relative to apparently normal (control) skin fibroblasts.

Although a number of conflicting reports exist concerning the differential ability of ACR cells to grow in low serum concentrations and whether or not ACR cells have lost "density dependent regulation" of growth, there are no published reports showing a full serum dose-response comparison between ACR and normal dermal fibroblasts. Nor has any researcher determined whether or not the "saturation densities" achieved by the cell strains tested were due to nutrient exhaustion or to a true density dependent effect. These aspects as well as the effects of culture age on serum requirement were considered in this study.

No attempt was made to reexamine the numerous other abnormalities which have been reported to exist in ACR dermal fibroblasts. Rather, this study was designed to examine basically one question, i.e. whether or not there exists in vitro evidence to support the hypothesis that the increased risk for cancer in ACR is due to a constitutive production of a growth factor.

II. REVIEW OF LITERATURE

The histopathological, genetic, epidemiological, and clinical aspects of ACR have been thoroughly described in a book by Bussey (1975). A review by Schuchardt and Ponsky (1979) gives the historical aspects of the disease (especially as it relates to whether or not Gardner's syndrome (GS) is a distinct clinical entity). This thesis is concerned only with those aspects of the disease which bear a direct or indirect relationship to the in vitro studies which have attempted to find a biological marker for the ACR gene and to elucidate the nature of the gene defect. The literature reviewed herein, therefore, will not be an exhaustive review of ACR but rather will review articles which pertain directly to the experimental results which were obtained here or to the discussion which follows.

Although the hereditary nature of ACR had been known since 1882, by the mid 1970's a number of unanswered questions about the genetics of ACR still remained (McKusick, 1974). Principal among these was the question of whether or not reported differences in the number and distribution of polyps and the presence of extracolonic lesions indicated that the gene defect in the Gardner syndrome was different from the gene defect in other forms of ACR.

But as the longevity of ACR patients increased, as a result of the increasing number of prophylactic surgical operations, more and more cases with extracolonic lesions were noted (Schuchardt, 1979). By 1984 several studies had shown that up to 90% of all patients with ACR had multiple osteomas of the mandible as demonstrated by orthopantomography (see Bulow, 1984, for a list of references).

Additionally, autopsies of deceased ACR patients revealed occult neoplasias of endocrine origin (Schneider et al., 1983). These findings led some researchers to question the validity of separating ACR into two groups. Final resolution of this question will have to await the results of in vitro studies designed to search for specific biochemical lesions.

In vitro studies of ACR cells (cells obtained from individuals known to have had the disease) have in many respects paralleled the investigation of cancer cells and of carcinogenesis in general. Cells obtained from tumor tissue and cells malignantly transformed in vitro have been shown to exhibit a number of cellular, biochemical, and growth associated properties (Pitot, 1981) which are not unique to cancer cells but which have been noted to occur more often in cancer cells than in normal cells.

Some of these characteristics were thought to play important functional roles in the pathogenesis of cancer. It seemed logical, therefore, to assume that some of these characteristics might be seen in ACR cells since individuals with the ACR gene were genetically predisposed to some forms of cancer.

If an altered phenotype were found it would not only serve as a biological marker for the identification of individuals at risk for the disease but also would help elucidate the nature of the gene defect and perhaps shed some light on the carcinogenic process itself.

During the late 1970's and the early 1980's a number of similarities between cultured ACR skin fibroblasts and malignantly transformed cells were reported. Similar to transformed cells the ACR cells were said to have a lower serum requirement for growth in tissue culture (Pfeffer et al., 1976; Kopelovich, 1977a), a higher saturation density (Kopelovich, 1977a; Rasheed and Gardner, 1981), and a higher plating or colony-forming efficiency (Kopelovich, 1979; Rasheed and Gardner, 1981) than normal cells.

In addition to these growth-associated abnormalities other cancer related characteristics were reported. These included elevated levels of plasminogen activator (Kopelovich, 1979) and loss of intracellular actin cables (Kopelovich, 1977a; Kopelovich, 1977b; Kopelovich, 1980). In contrast to the more ordered parallel arrangements seen in normal skin fibroblast cultures ACR cells were said to be arranged in a more or less random fashion similar to cultured cancer cells (Pfeffer et al.,1976).

ACR cells also were reported to have increased susceptibility to transformation by certain oncogenic viruses (Pfeffer and Kopelovich, 1979; Miyaki et al., 1980; Kopelovich, 1980; Rasheed and Gardner, 1981). At this time, other than an increase in tetraploidy noted in some Gardner syndrome cells (Danes, 1976), ACR cells were considered to be karyologically normal (Pfeffer et al., 1976). A theoretical framework for cancer predisposition existed at this time for certain DNA repair-deficient autosomal recessive disorders such as xeroderma pigmentosum (see Setlow, 1978 for a review), but the reason for cancer proneness in individuals with an autosomal dominant condition such as ACR was less clear.

A possible explanation in terms of the somatic mutation theory of carcinogenesis was provided by Comings (1973). In his general theory of carcinogenesis. Comings suggested that all cells posses a number of genes capable of inducing the formation of transforming factors which release the cell from its normal growth constraints. In adult cells these genes were thought to be suppressed by diploid pairs of regulatory genes and to be tissue-specific. These genes were also thought to be temporarily activated during some stages of embryogenesis and possibly during some stage of the cell cycle in adult cells. Chemically- and radiation- induced cancers were thought to be due to a double mutation in any set of regulatory genes which would release the suppression of the transforming genes.

Hereditary tumors such as exist in ACR were thought to be the result of a germ line inheritance of one inactive regulatory gene. Subsequent somatic mutation of the other regulatory gene would then lead to cancer. This theory is in accord with the two-mutation model advanced later by Moolgavkar and Knudson (1981) to account for the age-specific incidence of cancer in the general population and in individuals with hereditary conditions such as ACR.

Within the framework of the Comings theory, the findings of altered in vitro growth characteristics and other cancer associated properties seemed logical; a partial derepression of a potential "oncogene" could perhaps account for the myriad of abnormalities reported to exist in ACR cells.

Early studies of carcinogenesis (as reviewed by Boutwell, 1964 and Boutwell, 1974) had shown that benign tumors could be induced in the skin of mice by a single application of a carcinogen (intitiation) followed by the repeated applications of certain hyperplastic agents (promotion).

The induction of hyperplasia, in and of itself, was insufficient to induce the formation of these tumors (papillomas) since only certain hyperplastic agents were effective (e.g. croton oil or its active constituents, the phorbol esters). Other non-promoting hyperplastic agents could, however, increase the tumor incidence when applied repeatedly following the repeated application of a tumor promoter.

It was also found that the papillomas would regress if tumor promoter treatment was stopped and that tumors would still be formed after a brief carcinogen exposure even if the tumor promoter treatment was delayed for long periods of time (weeks).

Since initiation was an irreversible event it was generally assumed to be the result of a mutation, whereas promotion, being reversible, was thought to be due to the derepression of the mutant gene.

The limited exposure of initiated cells to tumor promoters was thought to create dormant tumor cells which had no proliferative advantage over the surrounding normal cells unless they were propagated by the repeated application of hyperplastic agents.

This theory became known as the two stage or initiation-promotion model of cancer.

The Moolgavkar and Knudson (1981) hypothesis interpreted the initiation-promotion experiments in mouse skin in a slightly different manner. Two irreversible steps were hypothesized to exist. The first was the creation of initiated cells which could be expanded into benign tumors through the use of tumor pomoters; and the second was the conversion of these intermediate cells into malignant cells by a second irreversible step (mutation). Support for the concept of two irreversible steps in mouse skin carcinogenesis came in 1983 with the demonstration by Hennings et al. (1983) that the conversion of mouse skin papillomas into carcinomas was unaffected by tumor promoters. These researchers found that just as many carcinomas arose from solvent (control) treated papillomas as from promoter treated papillomas. This suggested that promoter induced derepression alone could not account for the occurrence of carcinomas.

The Moolgavkar and Knudson hypothesis also made some definite predictions with respect to ACR. The adenomas in ACR were considered analogous to the benign papillomas induced in mouse skin by initiation-promotion techniques with the exception that according to their theory the adenomas of ACR should be polyclonal rather than monoclonal in nature. It was suggested that this prediction could be tested though electrophoretic analyses of the glucose-6-phosphate dehydrogenase (G6PD) isozyme content of adenomas in ACR affected females who were heterozygous for the A and B forms of the enzyme. Since the alleles of the G6PD gene were located on the X- chromosomes, monoclonal tumors would posess only one form of the enzyme due to X- chromosome inactivation while both forms could be produced in tumors of polyclonal origen.

This experiment was performed by Hsu et al. (1983) who did indeed find both forms of the enzyme in adenomas obtained from three female Gardner syndrome patients who were heterozygous for the enzyme. Using similar techniques Reddy and Fialkow (1983) found that a greater proportion of the papillomas induced in mouse skin by the carcinogen-promoter regimen were of monoclonal origin than the tumors induced by repeated application of carcinogen alone. Tumors induced by carcinogen-promoter regimen also were found to regress with greater frequency than tumors induced by repeated carcinogen applications.

In the 1970's attempts were made to duplicate the two-stage carcinogenic process in vitro (Lasne et al., 1974; Mondal and Heidelberger, 1976). The proposal then arose that "the ACR cell might exist in an initiated state due to a dominant mutation" and that the "expression of the malignant phenotype would only require treatment with a promoting agent " (Kopelovich and Bias, 1980; Kopelovich et al., 1979).

After treating ACR and normal (control) skin fibroblasts with 12-O-tetradecanoyl phorbol-13-acetate (TPA), the most potent of the tumor promoting phorbol esters found in croton oil, Kopelovich and Bias (1980) found that they could isolate anchorage-independent cells from the TPA treated ACR cell cultures (which occured at a frequency of about 5 in 10,000) but not from TPA treated normal cells. The isolated anchorage-independant clones were unstable with regard to their ability to grow in soft agar, however, and the continued presence of TPA did not enrich the fraction of anchorage-independent cells in liquid culture (Kopelovich and Bias, 1980). These cells also would not form tumors when injected subcutaneously into nude (athymic) mice but did form growths when injected into the anterior chamber of the eye of the nude mouse.

Gainer et al. (1984) were able to isolate anchorage-independent clones from only one out of three TPA treated ACR cell strains. Their anchorage-independent cells, although stable with regard to anchorage-independent growth, also did not form tumors when injected subcutaneously into nude mice. It was also noted that the TPA treatment significantly increased the chromosomal aberration frequency and the incidence of tetraploidy in the treated cells.

The reproducibility of the results obtained by the various researchers in the field of in vitro ACR studies was of critical importance not only for clinical and genetic counseling purposes but also for theoretical reasons since it might lead to the determination of the specific molecular lesion and establish the genetic basis for the disease. These results, however, were not always reproducible.

The finding of increased in vitro tetraploidy in Gardner syndrome cells was originally reported to be cell-specific for epithelial cells and to occur only in fibroblasts when they had been co-cultivated with epithelial tissue (Danes, 1976). Cultured cells obtained from ACR patients who were devoid of extracolonic lesions were reported not to exhibit increased in vitro tetraploidy (Danes et al., 1977; Danes, 1978). Increased tetraploidy also was observed to occur only in tissues which were subject to malignant transformation (Danes, 1978). It was subsequently shown, however, that increased tetraploidy could occur in both non-GS and GS ACR cells (Delhanty et al., 1980; Danes and Alm, 1981) and that tetraploidy could occur in the absence of epithelial cells (Delhanty et al., 1980).

Ambiguities were also reported with regard to the various in vitro growth abnormalities reported to exist in ACR skin fibroblasts. Danes et al. (1980) first reported that non-GS ACR cells could be distinguished from GS and normal cells by their reduced serum requirement and their "density-independent" growth. They, however, were unable to confirm the reported increased saturation density of ACR cells relative to normal cells. They were also unable to confirm the reported differential arrangement of ACR cells in confluent cultures. Criss-crossed and multilayered patterns were reported to occur in monolayers of both the normal and ACR cell cultures.

In a subsequent publication Danes and Alm (1981) reported that some but not all non-GS ACR dermal fibroblasts exhibited a reduced serum requirement. Rasheed and Gardner (1981), on the other hand, confirmed the reported increase in plating efficiency in high and low serum concentrations, the reported increase in saturation density and the reported higher susceptibility to viral transformation in both GS and non-GS ACR dermal fibroblasts relative to normal controls. They did, however, note some "false positives" in non-affected members of ACR families.

The variable incidence of tetraploidy and other phenotypic properties (low serum requirement and density dependent regulation of growth) was attributed to genetic heterogeneity within ACR (Danes et al., 1981). It was proposed that the influence of a modifying allele(s) to the major polyposis gene could effect the presence of specific abnormal culture phenotypes in a manner analogous to the presumed effects of such genes on the variation in the number and distribution of colonic polyps and extracolonic lesions (Danes, Alm, and Veale, 1980). However, in one study Danes and Krush (1977) could find no distinction between the incidence of tetraploidy in epithelial cells obtained from patients with "classical" and "variant" forms of Gardner syndrome. In another study an in vitro panel of biological assays on ACR skin fibroblasts failed to reveal any specific genotypic pattern within ACR (Danes and Alm, 1981).

In a more adequately controlled experiment (in terms of genetic variability) Heim (1983) compared the in vitro growth characteristics of ACR, GS, and normal skin fibroblasts within ACR families. This researcher found no difference between any of the ACR cells and the control strains in terms of saturation density, serum requirement or cloning efficiency.

More recent studies have implicated chromosomal instability as the source of cancer predispostion in ACR. Several researchers (Heim, 1985b; Miyaki et al., 1980; Rhim et al., 1980; Hori et al., 1980) have reported that ACR cells were more susceptible to the induction of chromosomal aberrations by the carcinogen N-methyl-N-nitro- nitrosoguanidine (MNNG) than normal cells. Some of these MNNG treated ACR cells were also reported to undergo morphological alterations and to acquire increased growth ability (Michiko et al., 1982; Rhim et al., 1980).

Other researchers (Gardner et al., 1982; Delhanty et al., 1983) have reported an increase in the frequency of spontaneous chromosomal abberrations in ACR cells relative to controls. Gardner et al. (1982) found both numerical and structural chromosomal aberrations with excessive random loss and gain of single chromosomes at the diploid and tetraploid levels in both lymphocyte and fibroblasts cultures. They also found a consistent heteromorphism of chromosome #2 which was tentatively identified as a deletion. Delhanty et al. (1983) found an increase in all classes of chromosomal abberation in lymphocyte, colon fibroblast, and skin fibroblast cell cultures relative to the control strains. They also noted a difficulty in establishing skin cell cultures from ACR patients and found a number of early cultures to be predominantly composed of a few cytogenetically abnormal clones.

Heim (1985a) used high resolution chromosome banding techniques to examine lymphocyte cultures obtained from ACR and normal individuals. Particular attention was paid to the reported heteromorphism in chromosome #2. No such deletion was found and all analyzed karyotypes were reported to be normal. Heim et al. (1985) also examined the constitutional C-band patterns in the peripheral lymphocytes of 78 normal individuals and the dermal fibroblasts from 23 unrelated ACR patients. In this instance a significant increase in the frequency of partial and total inversions in heterochromatin was found on chromosome #9 in the dermal fibroblasts of the ACR patients. Literature was cited which indicated that the difference in cell type could not have accounted for the difference in inversion frequency.

In two other studies, however, Heim (1985b) found no difference in the number spontaneous or mitomycin C induced gaps and breaks between lymphocyte cultures obtained from normal and ACR patients but found a significantly greater number of spontaneous and mutagen induced chromosomal aberrations in the skin fibroblasts from ACR patients relative to the control skin fibroblasts (Heim et al. 1985).

In a unique study by Parshad et al. (1983) chromatid damage after X-irradiation during the G-2 phase of the cell cycle was examined in 10 lines of dermal fibroblasts from cancer-prone individuals and 9 control cell strains. The 10 lines represented five genetic disorders and included strains obtained from individuals with Bloom syndrome, Xeroderma pigmentosum (complementation groups A, C, E, and Va), Falconi anemia, GS and non-GS ACR. The incidence of chromatid gaps after X-irradiation was found to be significantly greater in all the cancer-prone cell strains except the Xeroderma pigmentosum (XP) complementation groups A and Va in comparison to the controls. The incidence of chromatid breaks was found to be greater than the controls for all the cancer-prone strains except the XP complentation groups E and A. Since chromatid gaps and breaks were thought to represent unrepaired DNA strand breaks which could only have arisen directly or indirectly from the excision of X ray-induced DNA damage it was concluded that the cells from all of these cancer-prone individuals were deficient in some step of DNA repair (the gaps and breaks of the XP complementation group A cells were not increased by X-irradiation because in this case the lesion involved an inability to repair DNA damage which involves an endonucleolytic incision in the sugar phosphosphate backbone of DNA).

The hypothesis of a reduced capacity to repair DNA in cancer-prone and ACR individuals was later given support by the findings of Pero et al. (1983), Pero et al. (1985), and Pero et al. (1986). A reduced capacity to repair DNA was first found in the peripheral monocytes of individuals who had a hereditary predisposition to develope colon cancer, individuals diagnosed with ACR, and individuals who had had colorectal cancer but who were free of the disease at the time (due to colonic resection). The leukocytes were exposed to N-acetoxy-N-2-fluorenyl acetamide (N-AcO-2-FAA) in the presence of hydroxyurea and the level of unscheduled DNA synthesis (UDS) was measured. UDS levels were found to be depressed in both the genetically predisposed groups and the colon cancer patient group while the binding of N-AcO-2-FAA to DNA was found to be similar in all cases.

An association was then shown between the presence of both hyperplastic and adenomatous polyps and a reduced capacity for UDS (Pero, et al. 1985). In this study the leukocytes from patients with the adenomatous polyps exhibited levels of UDS similar to the controls but had a lower UDS/N-AcO-2FAA binding index because of reduced absorption of the mutagen.

In the third study by Pero et al. (1986) the rates of UDS of 13 fibroblast cell strains obtained from patients with ACR were compared with the rates of UDS of 7 control strains obtained from their unaffected relatives. The ACR strains exhibited reduced rates of UDS with and without N-AcO-2-FAA treatment. The ACR strains in this study also had elevated levels of spontaneous chromosomal aberrations, and in addition, a negative correlation between the level of spontaneous chromosomal aberration and the percent increase N-AcO-2-FAA induced UDS was found.

The studies which have implicated a reduced capacity to repair DNA as the gene defect in ACR, need of course to be replicated by independent investigators and to be followed up by more probing experiments designed to search for the specific step(s) in DNA repair which are affected. The essential question that remains, however, is how can a dominantly inherited gene function so as to inhibit DNA repair? One possibitilty is that it may alter deoxyribonucleotide metabolism. A gene which alters the balance of deoxyribonucleotide pools within a cell could have profound effects on DNA excision repair and thereby result in the accumulation of DNA strand breaks and chromosomal abnormalities (Snyder, 1984; Menth, 1984). Whether such an "oncogene" could also function to alter the growth regulatory mechanisms of a cell in the absence of genetic change remains an open question.

III. MATERIALS AND METHODS

A. Human Material

Five normal and five ACR cell strains were obtained from the American Type Culture Collection (ATCC). Two of the latter were submitted by L. Kopelovich and were used by him to identify the presumptive cellular differences between ACR and normal individuals. Three other ACR cell strains were obtained from the National Institute of General Medical Sciences (NIGMS) Human Genetic Mutant Cell Repository. Two of the eight ACR cell strains (GM 2355 and CRL 1532) were obtained from individuals with no clinical signs of Gardner's syndrome. All strains were certified as being free of mycoplasma, bacteria, and fungi by the ATCC and the NIGMS Cell Repository. The previous culture history of cell strains varied in that different split ratios were used to passage different strains. This is unlikely to affect the results of any growth studies (Hayflick, 1965). Conversion of passage numbers into population doubling levels revealed that the average culture age of the ACR strains were slightly greater than the average culture age of the control strains (Table 1). The average age of the ACR donor was slightly less than the average age of the normal donor.

B. Preparation of Media and Solutions

All media and solutions which were to come in contact with the cell cultures were prepared with deionized glass-distilled water. Media and trypsin solutions were sterilized by passage through a 0.20 micron membrane filter. Calcium/magnesium-free, phosphate- buffered saline solutions were autoclaved. Before use all media and solutions were tested for sterility using standard methods. No antibiotics were used in preparation of media. Contaminated cultures were immediately discarded. MCDB 104 (Ham and McKeehan, 1978) was provided in dry powder form by Gibco. For serum dose response experiments one liter volumes of MCDB 104 at 10 mg/ml of serum protein media were prepared, and then aliquots of these were diluted to the indicated concentrations with serum-free media under sterile conditions. Dialyzed serum (Gibco) was used to prepare the media for all growth experiments.

C. Culture Conditions and Subculture Procedures

Stock cultures were maintained in 25 square centimeter plastic tissue culture flasks (Flow laboratories) containing 4 mls of MCDB 104 with 10% (v/v) whole fetal bovine serum (Gibco) at 37 degrees centigrade in a humidified atmosphere containing 5% CO2, 5% O2, and 90% N2 (Union Carbide, Linde Division) in modular incubator chambers (Flow laboratories). Cultures used in growth experiments were incubated under the same environmental conditions as stock cultures. Sixty millimeter plastic tissue culture dishes (American Scientific) were used in all growth experiments.

Cells were harvested from monolayer cultures as follows: The growth medium was discarded and the monolayer washed with two 4 ml aliquots of calcium/magnesium-free Dulbecco's phosphate-buffered saline solution. Four mls of a HEPES-buffered saline solution containing 0.10 mg/ml crystalline trypsin (prepared as described by McKeehan et al., 1977) were then added, and the culture flask was gently rocked to insure that all parts of the monolayer were exposed to the solution. The trypsin solution was then removed and the monolayer was allowed to stand for 5 to 10 minutes at room temperature wetted only by a thin film of trypsin which remained in contact with the monolayer. One ml of culture medium was then added, and the flask was shaken to suspend the cells.

The suspension was then gently mixed by pipette to insure an even dispersion. A 0.l0 ml aliquot was then removed and the cells were counted with a hemacytometer. For stock cultures one fourth of the suspension was diluted with culture medium in a 25 square centimeter culture flask to 4 mls. For growth experiments the suspension was serially diluted to 10,000 cells per ml and then 0.10 ml (1,000 cells) was inoculated into each culture medium containing petri dish.

Periodically one ml aliquots of approximately one million cells were frozen in liquid nitrogen in MCDB 104 containing 10% (v/v) dimethylsulfoxide and 20% (v/v) whole serum in order to maintain stock cultures at low passage levels. The one milliliter aliquots were thawed and used to reinitiate stock cultures as needed. Most experiments were performed with cells within two passages (1:4 split ratio) of the passage number indicated in Table 1.

D. Procedures

1. Growth Measurements

Measurement of growth response was done by direct cell count. At the end of the growth period the culture medium was poured off gently from the culture dishes and then cells were fixed for 5 minutes with 2% (v/v) glutaraldehyde in 0.05% sodium cacodylate buffer at PH 7.6. They were then rinsed with tap water and stained for 5 minutes with 1% (wt/vol) crystal violet. Excess stain was washed off with tap water and the dishes were allowed to dry.

The number of cells per dish in most cases was determined by directly counting all of the cells within a dish. The dishes were placed under a dissecting microscope fitted with an ocular reticle and a zoom lens. The petri dishes used had squares etched on their lower surfaces. The dishes were placed over graph paper which was marked in such a way as to give location numbers to graph paper squares. The etched squares of each dish were divided by the graph paper squares into sublocations and these sublocations would be further subdivided by the squares of the reticle. The number of cells within each location or sublocation was counted with the aid of a hand-held cell counter and this number was recorded. The sum of such counts constituted the total number of cells within the dish.

When the number of cells exceeded 60,000, then cell number was determined with a compound microscope fitted with an ocular reticle. In this case the dishes were inverted and cells were counted under a 40X objective. A random field was chosen within each of the etched squares of the dish. The ocular nose-piece was rotated into place and those cells which fell within the field of the grid were counted. At least 225 and usually over 300 of such squares were counted yielding a representative sample of each portion of the dish. One field was counted for each square and squares were sampled without replacement. The total number of cells per dish was then determined by dividing the average number of cells per field by the area of the field and then multiplying this figure by the area of the dish.

2. Determination of the Effect of Cell Density on Growth Rate

The decline in growth rate with increasing cell density was estimated directly from the growth data. This was done by estimating the slope of the growth curves between the data points of the time course with the formula:

where;

R = the growth rate in generations per day                                                                                     

T = the time difference between the two data points N2 = the number of cells per dish of the higher       data point             

N1 = the number of cells per dish of the lower data point

ln 2 = the natural logarithm of 2

The cell density at the points at which the growth rates were estimated was obtained by averaging the number of cells per dish of the upper and lower points and then dividing this number by the area of the dish.

IV. RESULTS

A. Growth Measurements

1. Serum Requirement

Initially plans were made to seed 100 to 250 cells per dish in media of various serum concentrations and to measure the growth response by densitometric measurements of colony sizes as described by McKeehan et al. (1977). Several factors made this impracticable, however. First, the cloning efficiency of some of the cell strains were very low necessitating a higher seeding density. Secondly, the apparatus designed to measure the optical density of the colonies was insensitive to some of the smaller colonies. In addition to these difficulties problems were encountered with attachment of the cells to the dish surface at the various serum concentrations used.

In order to determine the serum concentration which provided optimal cell attachment, approximately 1000 cells were seeded into triplicate dishes containing MCDB 104 at serum concentrations ranging between 0 and 10.0 mg/ml serum protein and then the number of cells attaching one day post-plating was counted. The results are shown in Figure 1 and Table 2, and a statistical analysis using a square root transformation is shown in Table 3. The chart in Figure 1 would seem to indicate that best attachment occurred with 2 mg/ml of serum protein, but there was no significant difference between the number of cells attaching at 1, 2, or 4 mg/ml (Table 3). Serum concentrations above 4 mg/ml and below 1 mg/ml are seen to be inhibitory to cell attachment.

Henceforth, cells were seeded first at 2 mg/ml serum protein, and the media changed to the various serum concentrations one day post-plating, after cells had an opportunity to attach under uniform conditions.

The dose response curves to serum for 2 ACR and 2 normal cell strains using this regimen are shown in Figure 2. The growth rate data are shown in Table 4. These two ACR strains were submitted by L. Kopelovich to the ATCC and used by him to identify the presumptive cellular differences between ACR and skin fibroblasts. The determination of the serum optimum for each cell strain was done by comparing its growth rate at each serum concentration by means of a priori comparison tests as shown in Table 5. The serum optimum was defined as the serum concentration at which the cell strain grew the best. If no significant difference in growth rate between two or more of the uppermost points of the serum dose response curve was found, then the optimum was considered to be at the midpoint between the highest and lowest concentrations corresponding to these points. For growth rate ranking purposes the optimal growth rate was considered to be the average growth rate between these two points.

No assumption was made that the ACR cells would grow more rapidly than normal cells at any serum concentration so a posteriori tests were used to determine the growth rate rank for each strain at each serum concentration. The results of these a priori and a posteriori tests are summarized in chart form in Figure 3. In this chart the strains are listed according to their growth rate, the most rapidly growing strain being on top and the slowest growing strain on the bottom.

Strains are indicated by their ATCC numbers as shown in Table 1. The strains are also grouped within each column according to their growth rate rank as determined by a posteriori significance tests. The numbers to the left of each strain or group of strains indicates its growth rate rank. A line between any two strains or groups of strains indicates a significant difference in growth rate. The last column at the far right of the chart shows the growth rate rank of each strain at its respective serum optimum which is indicated to the right of the ATTC strain number.

As can be seen in Figure 2, although the two ACR strains appear to grow particularly well at high serum concentrations they exhibited similar optima with respect to the control strains. The average optimum for the ACR strains was 3.00 + 1.00 mg/ml (standard error of the mean) serum protein. For the control strains the average optimum was 2.75 + 1.25 mg/ml serum protein. The optimal growth rates of the two ACR strains (CRL 1533 and CRL 1532) were found to be greater than the optimal growth rates of the two normal strains. The optimal growth rates for all four strains differed significantly, however (Figure 3). The two ACR strains grew significantly better at 4, 6, and 8 mg/ml serum protein. At serum concentrations at and below 2 mg/ml no significant difference between the ACR and control strains could be detected (Figure 2 and Figure 3). Figure 4 shows the four curves with the data points represented in terms of a percentage of the optimal growth rate for each of the strains.

It was of interest to ascertain whether or not the shape of the serum dose response curves of normal and ACR cells differed significantly in shape so a chi-square analysis was performed on the percentages used to draw the curves in Figure 4. The results of these analyses are shown in Table 6. It can be seen that although the serum dose response curves for the ACR group was found to be significantly different from the control group, chi-square comparisons found significant differences between the shapes of all of the individual curves except for the ACR strain CRL 1533 and the normal strain CRL 1467.

In the aforementioned experiments the seeding media was removed as completely as possible leaving only a thin film of media (about 0.15 ml) before the addition of the test media. This means that the serum concentrations shown in Figure 2 are really slight underestimates of the actual serum concentrations below 2 mg/ml and are slight overestimates of the actual serum concentrations above 2 mg/ml. The cells could have been washed with media to remove the residual serum but there was concern that washing might detach cells from the growth surface in an unpredictable manner. Experiments were therefore undertaken to determined the effect of media washes on cell attachment.

Figure 5 shows the effect of various numbers of washes with serum free media on the number of cells remaining attached for two cell strains (the ACR strain CRL 1613 and the normal strain CRL 1467). A statistical analysis of the data for CRL 1467 is shown in Table 8 and an analysis of the data for CRL 1613 is shown in Table 9. It can be seen that the number of cells remaining attached after 1 and 2 washes with serum free media had no effect on cell attachment for either strain. Three and 4 washes removed a small percentage of the normal cells but had no effect on the ACR cell strain.

Table 10 and Figure 6 show the effect of 2 washes with media containing between 0 and 10 mg/ml of serum protein on the attachment of two cell strains. The statistical analyses shown in Tables 11 and 12 shows that there was no significant effect of serum protein concentration on the attachment of either of these two strains after two washes of media. Henceforth, as a further refinement, cells were washed twice with media at the serum concentration in which they were to be grown before the final addition of the test media.

Figure 7 shows the serum dose response curve for one ACR and one normal cell strain using the aforementioned procedure. The data for this Figure are in Table 13 and the accompanying statistical analyses are in Table 14. The results of a posteriori tests are shown in chart form in Figure 8. In this case the optimal growth rate for the normal strain (CRL 1467) was found to be greater than the optimal growth rate of the ACR strain (CRL 1613). The normal strain grew significantly better at 2 and 4 mg/ml serum protein, but no difference between the two strains was detected above or below these concentrations. The normal strain also exhibited a lower optimum (3.0 mg/ml) than the ACR strain (4.0 mg/ml). Although there appeared to be a difference in shape between these two curves when the growth rate data were expressed in terms of percentages of optimal growth rate (Figure 9) and analyzed by a chi-square test; the two curves were found not to differ significantly in shape. The chi-square value was found to be 9.86 with 5 degrees of freedom (0.10 > p > 0.05).

Figure 10 shows the serum dose response curves for four normal and five ACR cell strains. The conditions and procedures employed were the same as described in Figure 7 with the exception that a different serum lot was used in the preparation of the media. Figure 11 shows the same data in column chart form so that the growth rates of the ACR strains and the normal strains can be compared as groups. These nine curves are also shown in Figure 13 in area chart form with their heights linked so that the variation in the shapes of the curves can be seen more clearly. From these three Figures it is apparent that the serum dose response curves of the ACR group and the control group differ neither in shape nor in growth rate at any serum concentration. Support for this statement comes from statistical analysis. The data are seen in Table 15 with an accompanying analysis of variance and a priori tests in Table 16.

The anova Table includes data from four of the normal strains and four of the ACR strains. The data from one ACR strain (GM 3901) were not included in the analysis because of an internal heterogeneity of variances. Results of the a priori tests used to determine serum optima can be seen in Table 15 and the results of the a posteriori tests comparing the growth rates of the various strains can be seen in chart form in Figure 12. The serum dose response curves of the ACR and control groups expressed in terms of percentage of optimal growth rate are shown in Figure 14. The chi-square value obtained from an analysis of the percentages was found to be 2.20 with 6 degrees of freedom (0.90 > p > 0.50). It is apparent from these analyses that no consistent difference in growth rate between ACR and normal cells was found at any serum concentration and that the shapes of the serum dose response curves of these two groups were virtually identical. The average optimum for the ACR strains was 2.625 + 0.239 mg/ml serum protein and for the control group was 2.625 + 0.315 mg/ml serum protein.

2. Effect of Culture Age on Serum Requirement Growth Curves

In order to determine the extent to which culture age could affect the serum dose response curve of a human dermal fibroblast cell strain, one strain was chosen beforehand to be grown to near the end of its in vitro life span. For comparison, one flask of cells was frozen down early in the strain's proliferative life span and then thawed when the test was to take place so that the experiment could be done using the same batch of serum. Figure 15 shows the time course in the growth of this strain fitted to the equation of Shall and Stein (1979). The arrows indicate the points in the proliferative life span of the strain at which it was tested for serum requirement. Figure 16 shows a comparison between two determinations of the strain's serum requirement performed very early and very late in the strains proliferative life span (population doubling levels 14.5 and 49.7 respectively). The same batch of serum was used for both determinations. Figure 17 shows the shape of the two curves relative to their respective optima. A posteriori statistical tests showed that the growth rate of the strain at population doubling level (PDL) 49.7 was significantly less than at PDL 14.5 at all concentrations of serum. Examination of the a priori tests shown in Table 18 (data are shown in Table 17) revealed that the optimum for the strain at PDL 14.5 was 3.5 mg/ml serum protein and 3.0 mg/ml serum protein at PDL 49.7. Chi-square analysis revealed that the serum dose response curves at these two population doubling levels were significantly different in shape. The chi-square value was found to be 28.26 with 7 degrees of freedom (p < 0.005).

3. Growth Curves

The growth curves of three normal and four ACR cell strains are shown in Figure 18 . The data used to draw these curves are from Table 19 and the statistical analyses used to determine significant differences between data points of individual strains are in Tables 20 through 26. The cells were seeded at 1000 cells per dish in 4 mls of MCDB 104 containing 2 mg/ml serum protein. The cultures were fed every three days for periods lasting from about two to about four weeks.

Under this regimen some of the strains grew poorly. The Gardner strain GM 3223 (Table 24) and the normal strain CRL 1505 (Table 23) appeared to have spurts in growth and the normal strain 1471 had an unusually long lag period which lasted for about 12 days (Table 20).

Doubling rates were determined between the data points of the growth curves as described in Materials and Methods. These are shown in Figure 19 for two normal and three ACR strains for cell densities ranging from about 4 to 215 cells per square millimeter. Declining growth rates of all of the cell strains were associated with increasing cell density and there appeared to be no consistent differences in the relative magnitude of the assumed inhibitory effect of cell densities on growth rate between normal and ACR cells.

Three attempts were made to "rescue" the cells from their declining growth rate by refeeding cultures with 8 mls of media instead of 4 mls. These attempts were ineffectual (Figure 20 and Table 22, Table 24, and Table 25). When seeded at higher cell densities in MCDB 104 containing 10% (v/v) fetal bovine serum, much higher densities were achieved. When the cells were seeded into flasks which were then inclined at an angle so that the cells attached to only the lower half of the vessel, the cells appeared to grow to the same density as cells seeded in the conventional manner despite the fact that the former had one half the cell to media volume ratio of the latter.

This point is uncertain, however, since it was not known at what point the cells were in their respective growth curves when they were counted. Furthermore, the cell densities in half-flask seeded cultures were greater toward the edge of the monolayer. This may have been due to the shearing forces which occurred during media renewal which may have caused a retraction of the monolayer toward its edge and/or to due to a diffusion boundary effect. In any event the latter experiment was only preliminary and was neither critical nor definitive.

B. LIST OF TABLES

C. LIST OF FIGURES

V. DISCUSSION

According to the Coming's hypothesis the gene defect in ACR involves a partial derepression of a gene which is responsible for the production of a tissue specific growth factor. Gardner et al. (1980) proposed the production of a more generalized growth factor to account for the extracolonic stigmata of the Gardner syndrome.

Since it now appears that there is very little clinical or in vitro experimental evidence to justify the separation of ACR into two groups we may generalize this concept and call it the "growth factor hypothesis." Implicit in the growth factor hypothesis is the assumption that more than one growth factor mutation is required to change a normal cell into a cancer cell and that the ACR cell has sustained one of these mutations.

It may be further assumed from carcinogen experiments on mouse skin that "initiated" cells may not ordinarily express their tumorigenic potential in the absence of tumor promoting agents. Cultured ACR cells have been likened to the "initiated" cells of carcinogen treated mouse skin because of their susceptibility to transformation. On the basis of the reported low serum requirement and the reported high saturation density of ACR cells it has been implied that such "initiated" cells would occupy an intermediate position in growth regulation between a normal cell and a fully transformed cancer cell (Kopelovich, 1982 a; Kopelovich, 1982 b).

Since no one has yet isolated "initiated" cells, however, we can not be sure of how they would behave in cell culture. The growth factor hypothesis might explain the reported high saturation density and low serum requirement of ACR cells in terms of the in vitro production of a hypothetical growth factor. If such a growth factor were constitutively produced by cultured ACR cells (i.e. in the absence of tumor promoters and/or carcinogens) then it should be manifest in a comparative analysis of the serum dose response growth curves of normal and ACR cells.

In their efforts to develop an improved medium for the cultivation of normal human diploid fibroblasts McKeehan et al. (1980) and McKeehan and McKeehan (1981) found that they could apply the principles of Michaelis-Menten kinetics to analyze the nutrient and serum protein requirements of human cells. They found that the growth rate of normal human fibroblasts responded to increasing concentrations of nutrient in a hyperbolic manner similar to the response of an enzyme to increasing concentrations of its substrate. Growth rate could be described by the formula: Y = AX/(B+X); where X = the nutrient concentration, Y = the growth rate, A = the maximum growth rate analogous to the V max constant of the Henri-Michaelis-Menten equation, and B is analogous to the Michaelis constant or Km.

Under the conditions they described, the growth rate response of human lung fibroblasts to a purified fraction of serum was biphasic and exhibited higher values for A and B at a threshold serum concentration. A similar effect was noted for the growth rate response of human foreskin fibroblasts to epidermal growth factor. Under the conditions and procedures described in this study, however, the growth rate response of both normal and ACR dermal fibroblasts was neither biphasic nor continuously hyperbolic. Rather, there appeared to be an initial hyperbolic response followed by an approximately linear or slightly curvilinear decline in growth rate with increasing serum concentration. As can be seen in Figure 13 virtually all of the growth curves appear to be parabolic and most appear to be slightly skewed to the right.

As an initial attempt to account for the shape of these curves, we might hypothesize that the growth rate response to serum is essentially hyperbolic but there also exists an inhibitory effect (designated as CX) which is dependent on the serum concentration. Subtracting this effect from the hyperbolic equation we obtain Y = (AX/(B+X)) - CX which is the formula of a parabolic curve which is skewed to the right. The formula is useful in that it allows us to dissect the shape of the serum dose response curve into constants which might have real meaning. The constant A, for instance, may be expected to vary from strain to strain depending on the efficiency by which the cell's receptors for serum growth hormones induce cell division once saturated (analogous to the turnover number of an enzyme in enzyme kinetic theory). This constant might also be affected by the number of hormone receptors present within or on a cell and thus be affected by the previous cultural conditions and by the culture age of the cell strain. The constant B might represent the inherent affinity of a cell's receptors for growth hormones in a manner analogous by which the Michaelis constant (Km) measures the strength of an enzyme-substrate complex. The constant C represents an inhibitory effect present at all serum concentrations should also be taken into consideration when evaluating the relative responsiveness of a cell strain to changes in serum concentration.

The curves generated by this formula were found to fit the data reasonably well. Undoubtedly, deviations from the expected curves mean that the situation is more complex than envisaged, particularly when one considers that in varying serum concentration one is varying several hormones and attachment factors simultaneously. Nevertheless, the fact that the general shapes of the theoretical and empirical curves are quite similar, suggests that the essential assumptions of the hypothesis have validity.

The usefulness of the formula in analyzing serum dose response curves is illustrated by Figures 21 and 22. The graph on the left in Figure 21 shows the serum dose response data of the normal cell strain CRL 1467 (also shown in Figure Figure 2) fitted to the growth rate equation Y = (7.09 X/(0.80 + X)) - 0.72 X.

It can be seen that the fit is reasonably good except for the point at zero mg/ml serum protein. This is understandable since the data shown were taken from the experiment where media washing was not done. Thus a residual amount of serum protein was left at the 0 mg/ml point.

A better fit was obtained when it was assumed that the actual serum concentration began at 0.265 mg/ml and was increased by an additional 0.265 mg/ml at each serum concentration shown. The growth rate response could then be described by the equation shown in the graph to the right in Figure 21 in which the factor X + 0.265 was substituted for X in the equation. Note that in this graph the entire curve as well as the optimum has shifted to the left. Note, also that the curve retains its shape. A similar situation might also occur if a strain produced its own growth factor assuming such a factor was produced at equal rates at all serum concentrations and that the factor had both growth stimulatory and growth inhibitory effects.

In any case it seems reasonable to assume if a cell strain became capable of producing a serum constituent which it required for growth it would then exhibit a lower serum optimum and perhaps acquire the ability to grow for longer periods of time and at more rapid rates in the presence of little or no serum. If cultured ACR cells constitutively produced a growth hormone one might expect to find a lower serum optimum and a higher growth rate at low serum concentrations than normal. This was not found to be the case. Both the ACR and the normal control cell strains grew best at very similar if not identical serum concentrations and no difference in growth rate at low serum concentrations (0 and 1 mg/ml serum protein) could be detected between the normal and ACR groups (Figures 2, 3, 7, 8, 10, 11, and 12).

An alternative hypothesis might be that cultured ACR cells could produce a greater number of hormone receptors than normal (increasing the A constant in the equation) or produce unique hormone receptors which are more efficient in inducing cell division than normal (which would also increase the A constant) or which have a greater affinity for some growth factor (increasing the B constant).

Figure 22 shows the effects of various values of the three constants on the shapes and heights of the hypothetical serum dose response curve. The separate graphs were drawn so as to make the top of the curves the same height so that changes in curve shape may be seen clearly. The three graphs on the top from left to right show that as A increases the height of the curve increases and the optimum shifts to the right while the shape of the curve broadens. The middle three graphs show the effect of various values of the B constant.

It can be seen that as B decreases (i.e. as the affinity of the growth factor receptor for growth hormone increases) the curve increases in height and the optimum shifts to the left while the shape of the curve becomes increasingly skewed to the right. The bottom row of graphs in Figure 22 shows the effect of various values of the inhibitory constant C. It can be seen that as C decreases the height of the curve increases shifting the optimum to the right and the shape of the curve approaches that of a hyperbola. Thus, in the hypothetical serum dose response curve all three constants affect the growth rate, the position of the optimum and the shape of the curve.

Figure 23 shows the average growth rate of 4 normal cell strains at various concentrations of serum protein fitted to a serum dose response curve generated by the proposed equation. Figure 24 shows the average growth rate of 4 ACR cell strains fitted to a similarly derived curve. As can be seen these two curves have very similar shapes and their constants are also very similar. It is unlikely that the slight differences in the constants A and B in the two equations are significant since the growth rates of the strains of the ACR group and the normal group are not significantly different at any given serum concentration (Figure 12) and the shapes of the empirical curves are virtually identical (Figure 14).

In contrast to the similarity between these two curves the serum dose response curve of the normal cell strain CRL 1467 and the curve obtained from the same strain when it was 35 population doublings older differed significantly in shape (Figure 17) and growth rate (Figure 16). Figure 25 shows the serum dose response data fitted to equation generated curves.

It can be seen here that there is only a small difference between these two curves with respect to the constant B but there appears to be over a 50% decrease in the values of the A and C constants as a result of the increase in culture age.

The proportional decrease in the value of the C constant with respect to the A constant is intriguing. Perhaps the inhibitory effect of serum requires the consumption of growth factor either due to the simultaneous absorption of an inhibitor and growth factor or because the growth factor itself is the inhibitor. In any event it appears that the decrease in the value of C suggests that growth rate of senescent cells is less inhibited by increases in serum concentration than that of young cells. This is in agreement with the findings of Ohno (1979) who found that the growth rate of senescent cells were less inhibited at high concentrations of serum than the growth rate of young cells.

The serum dose response curves in Ohno's study showed a linear increase in growth rate with increasing serum concentration followed by a rapid decline at high serum concentrations beyond the optimum. In contrast the growth rate curves shown in this study exhibited an initial hyperbolic response followed by a gradual curvilinear decline. Ohno's study also showed an increasing optimum with increasing culture age. This is consistent with a decreasing value of C (Figure 22, bottom three graphs). There also was a continual reduction in growth rate at all serum concentrations with increasing culture age. This is consistent with a reduction in the value of the A constant (Figure 22, top three graphs). A reduction in the A constant shifts the optimum to the left while a reduction of the C constant shifts the optimum to the right.

The net effect of a reduction of both of these constants on the optimum depends on the relative magnitude of the changes in A and C providing the B constant remains the unchanged. But since there was a slight reduction in the value of the B constant and a slightly greater proportional reduction in the A constant relative to the C constant the net effect was to shift the optimum slightly to the left (Figure 22).

The foregoing discussion points out the complexity of interpreting the meaning of the shapes of serum dose response curves and the usefulness of the formula in their analysis. A low serum optimum necessarily implies a low serum requirement but tells us nothing about the biological processes involved in determining the optimum. The ability to grow for extended periods of time or at rapid rates in low serum media has also be used in the determination of a cell strain's "serum requirement" but this tells us little about the nature of the growth response. As shown in the bottom three graphs of Figure 22 rapid growth in low serum might just as well be due to a low value of the inhibitory constant C as to a high value of A or low value of B.

Furthermore, given a hyperbolic growth rate response to increasing serum concentrations, a very small change in the potency of serum may cause a large change in the growth response. Thus, designating a strain as "low serum positive" or "low serum negative" as has often been the case in ACR studies (see Review of Literature) may be a tricky matter at best.

The decline in serum responsiveness of the normal cell strain CRL 1467 with increased culture age would be interpreted by this equation as due to a loss in responsiveness to mitogens primarily due to a loss of mitogen receptors and/or the reduction of receptor induced cell division efficiency. This conclusion is supported by studies with WI 38 cells. Phillips et al. (1984) have reported that these cells undergo a progressive loss of growth responsiveness with increasing culture age to several serum mitogens including epidermal growth factor, transferrin, insulin, dexamethasone, and platelet-derived growth factor.

They also found that the concentration of epidermal growth, transferrin, insulin, and dexamethasone required to elicit a maximum growth response did not change as a function of culture age. The optimum for platelet derived growth factor may have increased with culture age but this point was uncertain. The loss in mitogen responsiveness to epidermal growth factor could not have been due to a loss in the number of receptors since Philips et al., (1983) has shown that the number of specific epidermal growth factor binding sites per cell actually increased with culture age though there was no change in the number of specific binding sites per unit membrane surface area (WI 38 cells become larger as they age). These researchers also reported a slight decrease in the dissociation constant for the receptor-epidermal growth factor complex in senescent WI 38 cells.This is interesting considering the slight decrease in the value of the B constant reported in the present study for the senescent cells of the normal cell strain CRL 1467 (Figure 25).

In contrast to the reported increase in the number of epidermal growth factor receptors per cell in senescent WI 38 cells, Rosner (1981) found a 40% reduction in the number of dexamethasone receptors per cell in senescent WI 38 cells. They also reported reduced nuclear translocation of the dexamethasone-receptor complex and no change in the dexamethasone-receptor dissociation constant.

A complete or partial loss of density dependent inhibition of growth has been said to characterize ACR cells (Danes, 1980; Kopelovich et al., 1979). Yet if density dependent inhibition of growth were completely lost one would expect that ACR cells would grow continuously until essential nutrients were depleted and the cell culture was lost.

Figure 19 shows the declining growth rate of 2 normal and 3 ACR cell strains with increasing cell density as computed from the growth curve data shown in Figure 18. As can be seen the decline in growth rate with increasing cell density of the control strains can not be said to be consistently greater than that of the ACR strains.

The decline in growth rate for any given strain may have been due to either nutrient depletion or due to a density dependent effect. That the latter alternative is true for the two ACR cell strains CRL 1610 and GM 3223 and for the control strain CRL 1467 is supported by the fact that refeeding each of these strains with 8 mls instead of 4 mls of medium toward the end of their growth periods did not significantly affect the decline in their growth rates (Figure 20). This point remains uncertain, however, since in each case there was an increase in cell number after the 8 ml refeeding (though not judged to be statistically significant; Tables 22, 24, and 25) and since the growth rates of these strains had already started to decline when the 8 ml refeeding schedule had begun.

This study has attempted to perform a more refined comparative analysis of the serum requirements of normal and ACR skin fibroblasts and to examine some of the possible reasons for the lack of reproducibility of the reported altered in vitro growth characteristics of ACR cells. An attempt will now be made to place the present findings in context with other ACR studies, to examine some of the difficulties in interpreting the results of such studies, and to point out some possible areas of future research.

Several of the past studies which have compared the in vitro growth characteristics of ACR cells with normal cells have been performed with Minimal Essential Medium (MEM). The growth of human skin fibroblasts in this medium requires a sizable supply of whole serum (15% volume per volume). This is probably because MEM lacks many of the essential nutrients and metabolic intermediates which are required for the growth of these cells.

The poor growth of cells plated in MEM with low concentrations of serum or at low cell density may, therefore, be due to inadequate supply of nutrients rather than to an inadequate supply of serum mitogens. Likewise the inhibition of growth of fibroblast cultures at high cell density could also be due to a nutrient limitation.

Balk (1980) has argued that the density dependent inhibition of growth which occurs in fibroblast cultures is an in vitro artifact which is unrelated to the normal regulatory mechanisms which control the growth of these cells in vivo. A number of researchers have assumed that substances which relieve density dependent inhibition of growth represent true mitogens. Balk (1980), however, has has pointed out that even the addition of precipitates of calcium phosphate and barium phosphate or the addition of particulates such as kaolin, aluminum oxide, and polystyrene beads can cause proliferative activity in density-inhibited cell cultures. This he proposed was due to a disturbance in the diffusion boundary layer of the cell cultures (the diffusion boundary layer is a stagnant layer of media just above the flat surface of a cell confluent culture).

It has been known for some time that density dependent inhibition of growth in fibroblast cultures may be overcome by media agitation (Kruse and Miedema, 1965; Stoker, 1973; Stoker and Piggott, 1974; Froehlich and Anastassiades 1975) and by increasing the concentration of serum in the media (Todaro et al., 1965; Holley, 1968; Dulbecco, 1970; Clark et al., 1970; Dulbecco and Elkington, 1973; Holley, 1974).

On theoretical grounds Maroudas (1974) concluded that the limiting factor in density dependent inhibition of growth is likely to be a consumable substance present at low concentration in serum but the concentration of low molecular weight nutrients (Holley, 1975), the release of growth inhibitory substances (Harel et al., 1978; Steck et al., 1979; Strobel-Stevens and Lacey, 1981), and the degree of cell spreading and adhesiveness to the surface substratum (Folkman and Moscona,1978) may also play important roles.

Thus, perhaps the most serious difficulty in past in vitro ACR work has been the lack of a precise definition of the phenomenon under study. For instance, If a high saturation density were found in a particular ACR strain do we ascribe this reported high saturation density of a reduced requirement for nutrients? If so, then which nutrient(s) are needed in lessor quantity? Or is the increased saturation density due to a reduced requirement for serum? If so, then which component in serum is responsible for this reduced requirement? If the component is found then how is it utilized? Does the cell strain produce it or is the requirement of this component some how bypassed. If a specific biochemical lesion is found in one ACR pedigree then do other pedigrees exhibit the same or similar lesions? These are the kinds of questions which should have been asked but which few studies have addressed themselves to.

The choice of MCDB 104 in these experiments was not an arbitrary one. This medium has been developed over a period of years through a sequential process of optimization of the growth response of human lung fibroblasts to various concentrations of individual nutrients (McKeehan et al., 1977; McKeehan and Ham, 1977). Through the use of this process and other refined cell culture techniques, these researchers have been able to reduce the amount of serum protein required for human fibroblast growth to microgram quantities (McKeehan et al., 1977; Ham and McKeehan, 1978). The growth response curves shown here are therefore more likely to represent a true mitogenic response to serum growth hormones than to a nutritive response due to an inadequate nutrient supply.

The saturation density of human cells is greater in conventional media than in MCDB 104 but probably this is not due to any inadequacies of the medium. Rather, this could be due to the higher concentrations of nutrients in conventional preparations. Human fibroblasts have been shown to grow much better at low density in MCDB 104 than in conventional media. In conventional media preparations, transformed cells have often been noted to require much less serum than normal cells (Holley, 1974; Holley, 1975; Clark et al. 1970 Smith et al., 1971; Scher and Todaro, 1971) but few studies have shown that this reduced requirement is due to a reduced requirement for any particular mitogenic component in serum. In polylysine coated dishes in MCBD 105 (a medium which differs only slightly from MCBD 104) McKeehan and McKeehan (1981) found that human lung fibroblasts and their SV40-transformed counterparts exhibited similar serum protein growth requirements. The transformed cells were found to exhibit a reduced requirement for many different nutrients, however.

The progressive loss in responsiveness of human cells to serum mitogens points out another difficulty in interpreting the results of previous ACR studies. It has been assumed from the early studies on cellular senescence (Hayflick, 1961; Hayflick, 1965) that the in vitro life span of human cells can be divided into three distinct phases: Establishment of the cell culture (phase 1), a period of logarithmic growth potential (phase 2), and a period of declining growth potential or "cellular senescence" (phase 3).

The progressive loss of growth responsiveness to serum (Ohno, 1979) and to purified serum mitogens (Phillips et al. 1984) suggests another possibility, which could be that growth potential is lost continuously from the very beginning of the cell culture as suggested by the theory of Shall and Stein (1979). Their theory predicts that an increasing fraction of newly born cells will become non-dividers with each generation in cell culture. This fraction is determined by only one constant (Gamma) which is unique for each cell strain and for each set of environmental conditions. With each generation the probability of mortalization or reproductive sterility (Pm) increases according to the hyperbolic equation Pm = t/(Gamma + t); where t= the number of generations.

The graph on the left in Figure 15 shows the cumulative population doubling values of the normal cell strain CRL 1467 fitted to a computer generated curve based on the Shall and Stein equation. The data points in the graph on the right in the same figure shows the growth rate of the strain in doublings per day at various generations and the theoretical growth rate curve as predicted by the theoretical cumulative curve.

The growth rate curve shows a continuous decline from the very beginning of the life span of the cell strain. Hence, if saturation density were to remain constant throughout the life span of a cell strain then, this would necessarily imply a continual increase in density dependent inhibition of growth. Conversely, if density dependent inhibition of growth remains constant with increasing culture age, then saturation density must necessarily decline with increasing culture age.

The cumulative population doublings curve was based on the theoretical assumption that an increasing fraction of newly born cells became reproductively sterile with each generation. Thus, the cloning efficiency this strain would also be expected to decline continuously with increasing culture age. That cloning efficiency does in fact decline continuously has been shown to be the case by Merz and Ross (1969), Smith and Hayflick (1974), and Smith et al. (1978).

The effect of culture age on the variation of in vitro growth characteristics of human cell strains may not be an insignificant problem particularly if these strains are established from one or a few clones. For instance if a cell culture of 1,000,000 cells was established from a single cell that strain will have already undergone 20 population doublings by the time the strain has become established. A very wide variation in doubling potential in cells of human fibroblast cell cultures has been reported (Merz and Ross, 1973; Smith and Hayflick, 1974; Smith and Whitney, 1980). Even at relatively low population doubling levels only 50% of the cells are capable of more than 8 additional population doublings (Smith and Hayflick, 1974).

These considerations make data regarding the frequency with which skin biopsies yield fibroblast cultures and the time required for the establishment of these cultures particularly relevant. Unfortunately such details are often omitted in ACR studies.

Of the possible reasons for the inconsistent findings between different laboratories with regard to the reported increased in vitro growth ability of ACR dermal fibroblasts only a few will be mentioned here. One possible explanation is that the ACR gene might be closely linked to genes which do affect growth. In this respect it is of considerable interest to note that increased growth ability in ACR cells has been reported more often in papers in which it was stated that most of the poorer growing control strains used were obtained from individuals who were unrelated to ACR family members. In contrast, in the study by Heim (1983) only cell strains obtained from non-affected family members were used as controls and no growth differences were noted.

Another possible explanation is that the methods of obtaining and processing skin biopsies from ACR patients and normal individuals could have influenced the results of these studies. In the study by Heim (1983) skin biopsies were obtained from the most superficial portion of the dermis; the papillary dermis. The standard procedure of obtaining skin biopsies samples a full thickness of skin which includes a fair portion of the reticular dermis. Harper (1979) found that fibroblasts obtained from the papillary dermis grow more vigorously than fibroblasts obtained from the deeper reticular layer of the dermis. Harper's study also showed that reticular fibroblasts have a shorter in vitro life span than papillary fibroblasts.

As noted previously Delhanty et al. (1983) experienced difficulty in establishing skin fibroblast cultures from ACR patients but not from normal controls. They also reported that a number of ACR cultures were composed of a few cytogenetically abnormal clones. The seemingly greater physiological age of reticular fibroblasts suggests that they may be closer to the "senescence phase 3" of Hayflick than papillary fibroblasts. During this period of time in the life span of a cell strain the strain will undergo a crisis in which it will either become senescent and die out or will be "transformed" into an immortal cell line. Chromosomal abnormalities often accompany such a transition. Perhaps a similar transition takes place early on during the establishment of an ACR cell strain when it is established from a full thickness of skin.

In this regard it should be noted that the growth kinetics of human fibroblasts in cell culture have been likened to a diffentiating system in which there is an attentuation of the growth of serial clones, with the continuous selection for more vigorous stem cells (Martin et al., 1974; Bell et al., 1978).

Supporting this notion is the finding that cells from all phases in the lifespan of a human cell strain are heterogenous with regard to proliferative potential. "Young" cells with high proliferative potential predominate during the "log" phase and "old" cells with low proliferative potential predominate during the "senescent phase" (Smith and Hayflick, 1974). The distribution of proliferative potentials of cells obtained from cloned cells has also found to be bimodal with both "young" and "old" cells (Martin et al. 1974; Smith and Whitney, 1980).

The demonstrated susceptibility of ACR cells to certain carcinogenic agents, the reported chromosomal instability, and the reported reduced capacity for DNA repair (see Review of Literature for a list of references) also suggests that biopsies obtained from individuals with ACR might more frequently give rise to the establishment of abnormal cell strains than biopsies obtained from normal individuals.

In this respect it is of interest to note that in this study the two strains submitted by L. Kopelovich exhibited more rapid growth rates than the two control strains not at low concentrations of serum but at high concentrations of serum (Figure 2). One strain in particular (CRL 1532) grew rapidly at very high serum concentrations despite the presence of an apparent toxic effect. At high serum concentrations the cells of this strain exhibited irregular appearances and were surrounded by cytoplasmic fragments.

In continuation of this line of reasoning, perhaps the most interesting aspect of the study by Danes and Alm (1981) is its ambiguity. Six of the ACR strains which they studied exhibited a lower than normal serum requirement and a loss of density dependent inhibition of growth but were found to be normal with regard to tetraploidy. Six other ACR strains which they studied exhibited both a higher than normal incidence of tetraploidy and a loss of density dependent inhibition of growth but were normal with regard to serum requirement. Although only 1 out of 7 Gardner strains studied was found to be normal with regard to tetraploidy, all 7 of these strains were found to be normal with regard to serum requirement and density dependent inhibition of growth.

What may be instructive at this point is to consider a series of revealing experiments performed by Harry Rubin (Rubin et al., 1983; Rubin, 1984; Rubin et al., 1984). Rubin isolated a spontaneously transformed clone from a line of BALB/3T3 cells. Using this clone he initiated 5 subclones from randomly chosen cells. Each of these subclones differed in colony morphology, cloning efficiency and colony diameter in soft agar, and in the rate of tumor formation in nude mice. Subsubclones isolated from these 5 subclones also differed from each other in cloning efficiency and colony diameter in soft agar. With weekly passages in cell culture the differences in colony morphology and growth in soft agar of the 5 primary subclones gradually decreased. The cells from one of the tumors intiated from a subclone which grew well in soft agar initially, grew poorly in soft agar but after 2 months in culture the tumor derived cells grew as well as the parent clone.

These experiments by Rubin suggest that a tremendous capacity for variation and adaptation exists in some transformed cells. Perhaps a similar capacity exists in some cultured ACR cells albeit to a lessor degree.

There are other possible explanations for the inconsistent reports of abnormal in vitro growth characteristics in ACR cells. One might be that there was variable "penetrance" of the ACR gene because of the presence of modifying alleles or to other modifying genes. Alternatively, several different mutant genes might produce the same ACR phenotype in vivo while producing different phenotypes in vitro. It would be challenging task to design experiments and statistical tests in order to ascertain the relative importance of each of these variables.

The Comings hypothesis with regard to the gene defect in the ACR has not been ruled out by this or other studies since it is possible that the expression of the ACR gene may require the presence of tumor promoters. Many researchers believe that the growth of mammalian cells is controlled by mechanisms which regulate the onset of DNA synthesis. In this respect the findings of a reduced capacity for DNA repair in a dominantly inherent condition such as ACR may be significant. Perhaps the gene defect in ACR involves an alteration in a growth regulatory pathway which produces a change in the relative concentrations (pool balance or ratio) of the 4 deoxyribonucleotides. Such a gene defect could also cause faulty DNA repair and result in chromsomal aberrations (Snyder, 1984; Menth, 1984).

Alternatively, unrepaired DNA strand breaks could persist due to error prone repair. This might reduce the requirement of the ACR cell for external mitogens because the continuous induction of DNA repair synthesis might provide some of the enzymes and/or other anabolic products which are also required for replicative DNA synthesis.

Although the results of this study offers no support for the concept of an inherent increased growth capacity in ACR cells there is evidence which suggests that the colonic epithelial tissue within individuals with ACR exits in a hyperplastic state possibly due to the presence of bacterial carcinogens and/or endogenous tumor promoters. Luk and Baylin (1984) have found that the normal appearing flat mucosa as well as the non- dysplastic and dysplastic polyps from individuals with ACR contain elevated levels of ornithine decarboxylase activity relative to the colonic tissue of normal individuals.

Onithine decarboxylase is a rate limiting enzyme in the synthesis of polyamines which are required for the normal growth and differentiation of a variety of tissues (Heby, 1981; Pegg and McCann, 1982). The induction of ornithine decarboxylase activity may also be required for the formation of papillomas in mouse skin (Obrien, 1976; Obrien et al., 1975; Weeks et al., 1982). The potent tumor promoter TPA which induces ornithine decarboxylase activity in mouse skin has also been found to stimulate DNA synthesis in the colonic epithelial cells of individuals with ACR but not in the colonic epithelial cells of normal individuals (Friedman et al., 1984). These researchers found that tissue specimens obtained from most benign tubular adenomas from normal individuals and from normal appearing ACR colonic epithelium respond to TPA by the induction of DNA synthesis while tissue specimens obtained from more advanced tumors (villous tumors, tumors with dysplasia, and adenocarcinomas) respond to TPA by changes in cellular morphology and by the production of plasminogen activator.

Certain kinds of plasminogen activator have been implicated in the ability of carcinomas to invade and destroy surrounding tissues (Dano et al., 1985). It also should be noted that there appears to be an inverse correlation between the DNA repair capacity of human cells and their ability to produce plasminogen activator in response to ultraviolet radiation (Ben-Ishai et al., 1984; Misken and Ben-Ishai, 1981).

Tumor promoters were originally thought not to affect DNA and to exert their effect solely through the derepression of genes in carcinogen treated cells (Boutwell, 1974; Trosko et al., 1975). More recent evidence has shown, however, that tumor promoters can induce the formation of clastogenic (chromosome breaking) factors (Emerit and Cerutti, 1981) and that at least part of the ability of tumor promoters to induce the formation of ornithine decarboxylase in some cells may be related to the formation of active oxygen species (superoxide radicals, hydroxyl radicals, singlet oxygen, hydrogen peroxide) and lipid hydroperoxides (Friedman and Cerutti, 1983). Anti-inflammatory drugs which are thought to inhibit this process (Marx, 1983) have been shown to be effective in inducing the regression of polyps in patients with ACR (Waddell and Loughry, 1983).

VI. SUMMARY

According to the Comings hypothesis, the increased risk for cancer in individuals with dominantly inherited diseases such as ACR results from the presence of a mutation in one of two regulatory genes which normally suppress the expression of a potential oncogene and that the tissue specific partial derepression of such a gene results in the production of a transforming factor capable of releasing the cells of the affected tissues from their normal growth constraints. The early findings of increased growth ability and other cellular abnormalities in ACR dermal fibroblasts supported this view and suggested that the production of the transforming growth factor by the cultured cells was constitutive (i.e. did not require the presence of tumor promoters and/or carcinogens).

The results of these early studies, however, were not consistently confirmed by subsequent researchers. The present study examined this question by comparing the growth of 8 ACR and 4 normal control dermal fibroblast cell strains in the culture medium MCDB 104 at various concentrations of dialyzed serum protein. Since the nutrients in MCDB 104 have been "optimized" to provide for the nutritional requirements of human fibroblasts in minimal concentrations of serum protein it was reasoned that in vitro growth control experiments conducted in this medium might be more representative of the regulatory mechanisms which control the growth of human dermal fibroblasts in vivo than in vitro growth control experiments conducted in minimal media.

In this study both the normal and ACR strains grew at similar rates in MCDB 104 supplemented with low serum concentrations of serum protein. Both the normal and ACR strains also exhibited similar serum optima. Two of the ACR strains used in this study had been previously used to identify the presumptive cellular differences between ACR gene carriers and normal individuals. These 2 cell strains differed from 2 control strains in growth rate at high serum concentrations but not in growth rate at low serum concentrations. Comparative analyses of the serum dose response data obtained from the 4 other ACR strains and 4 normal control strains revealed no significant differences in average serum dose response curve shapes, average serum optima or average growth rates at any serum concentration between the ACR and control group. The growth rates of both the 2 normal and 3 ACR strains declined with increasing cell densities to a similar extent. Feeding some of the cultures with 8 mls of media instead of 4 mls failed to halt the decline in growth rate with increasing cell densities suggesting that the decline in growth rates was due to a density-dependent effect and not to nutrient exhaustion.

An equation was developed to analyze the shape of the serum dose response curves. The equation is: Y = (AX/(X + B)) - CX. In this equation Y = growth rate, X = serum protein concentration, and the values of A, B, and C are constants. Equation generated curves were used to analyze the serum dose response data of 4 normal and 4 ACR strains. One normal strain was analyzed similarly at two different population doubling levels. Reductions in the constants A, B, and C were found in the normal cell strain which had undergone an additional 35 doublings in vitro. The A, B, and C constants for the 4 normal and 4 ACR cell strains were found to be similar.

VII. APPENDIX

Iterative Procedure Used to Fit Serum Dose Response Data to Hypothetical Equation Generated Curves:

Finding a curve of best fit to the serum dose response data using the hypothetical growth rate equation

involved the use of an iterative procedure. The method that was used to obtain the curve of best fit was as follows:

First, advantage was taken of some of the important features of the curve. In particular the fact that the slope of the curve at the optimum must be zero. Thus, at the serum concentration which promotes an optimal growth rate, dY/dX = 0.

Differentiating equation (1) it can be shown that

Solving for X (which is designated here as Xo for the serum concentration which promotes an optimal growth rate) it can be shown that

Also note that the ratio of A divided by C is

At Y = 0 the serum concentration at which growth rate is extinguished (which is designated here as Xe) is

It can be see from this formula that the ratio of A divided by C is

Thus,

Solving for B we have

Thus, knowing Xo and Xe one can find the constant B and from B one can determine the ratio of A/C. Now at any given data point (X,Y) one can estimate the value of C with the formula:

Since each data point will probably estimate a different value for C an average value can be obtained with an estimated standard error. The value of Xo can be determined empirically from the data as described in Results. The value of Xe can be estimated initially through the use of a straight edged ruler fitted to the data points which lie to the right of the optimum.

By varying the value of Xe different values of C can be obtained with their corresponding standard errors. The values of Xe and C are then chosen which give the minimum error in the estimation of the value of C. This can be done quite readily with the aid of a digital computer.

When the empirical serum dose response curve intersects the growth rate axis above 0 doublings per day at the 0 mg/ml serum protein then the residual amount of serum protein remaining attached to the dish surfaces can not be ignored and a correction must be made*. This is done by assuming that the actual serum concentration is greater by a factor of D than the apparent serum concentration shown in the ordinate of the graph.

After the constants A, B, and C have been determined by the method given above then the actual serum concentrations can be obtained. This is done by estimating the serum concentration (D) at which growth rate equals the empirically determined growth rate at 0 mg/ml serum protein. A new growth curve is then computed from the actual serum concentration by substituting the actual serum concentration Z for X in the equation. Thus, the equation

(where X = the apparent serum concentration and Z = the actual serum concentration = X + D) is used to plot the abscissae against the X ordinate.

*COMMENT: The residual amount of serum protein remaining attached to the dish surfaces is due to the need to attach cells first in media at a concentration of 2 mg/ml serum protein. A thin film of protein is require for firm adhesion of the cells to the dish surface. This film is apparently not removed by media washes.

VIII. LITERATURE CITED

Balk, S. D. (1980) Precipitates and particulates cause proliferative activity of density- inhibited cultured cells by disturbing the diffusion boundary layer: An artifact superimposed upon an artifact? Life Sciences 27: 1917-1920.

Ben-Ishai, R., R. Sharon, M. Rothman, and R. Miskin (1984) DNA repair and induction of plasminogen activator in human fetal cells treated with UV light. Carcinogenesis 5: 357- 362.

Boutwell, R. K. (1964) Some biological aspects of skin carcinogenesis. Prog. Exp. Tumor Res. 4: 207-250.

Boutwell, R. K. (1974) The function and mechanism of promoters of carcinogenesis. Crit. Rev. Toxicol. 2: 419-443.

Bulow, S., J. O. Sondergaard, I. Witt, E. Larsen and G. Tetens (1984) Mandibular osteomas in familial polyposis coli. Dis. Colon Rectum 27: 105-108.

Bussey, H. J. R. (1975) Familial polypsis coli. Family studies, histopathology, differential diagnosis, and results of treatment. Baltimore: Johns Hopkins University Press.

Clarke, G., M. Stoker, A. Ludlow, and V. Thornton (1970) Requirement of serum for DNA synthesis in BHK 21 cells: Effects of density, suspension, and virus transformation. Nature (London) 227: 798-801.

Comings, D. E. (1973) A general theory of carcinogenesis. Proc. Natl. Acad. Sci. (USA) 70: 3324-3328.

Danes, B. S. (1976) Increased tetraploidy: Cell-specific for the Gardner gene in the cultured cell. Cancer 38: 1983-1988.

Danes, B. S. (1978) Increased in vitro tetraploidy tissue specific within the heritable colorectal cancer syndromes with polyposis coli. Cancer 41: 2330-2334.

Danes, B. S., and T. Alm (1981) In vitro evidence of genetic heterogeneity within the heritable colon cancer syndromes with polyposis coli. Scan. Journ. of Gastro. 16: 421- 427.

Danes, B. S., T. Alm, and A. M. O. Veale (1980) Modifying alleles in the heritable colorectal cancer syndromes with polyps. In Winawer, sl, Schottenfeld, D. & Sherloch, P. (eds.) Colorectal cancer: Prevention, Epidemiology, and Screening. Raven Press, New York. 73-81.

Danes, B. S., S. Bulow, and L. B. Svendsen. (1980) Hereditary colon cancer syndromes: an in vitro study. Clinical Genetics 18: 128-136.

Danes, B. S., A. J. Krush, E. J. Gardner (1977) Is Gardner syndrome a distinct genetic disorder? Lancet ii: 925.

Danes, B. S., and A. J. Krush (1977) The Gardner syndrome: A family study in cell culture. J. Natl. Cancer Inst. 58: 771-775.

Dano, K., P. A. Andreasen, J. Grondahl-Hansen, P. Kristensen, L. S. Nielsen, and L. Skriver (1985) Plasminogen activators, tissue degradation, and cancer. Advances in Cancer Research. 44: 139-266.

Delhanty, J. D., M. B. Davis, H. J. R. Bussey, and B. C. Morson (1980) Tetraploidy fibrblasts and familial polyposis coli. Lancet 1: 1365.

Delhanty, J. D. A., M. B. Davis, and J. Wood. (1983) Chromosome instability in lymphocytes, fibroblasts, and colon epithelial-like cells from patients with famililial polyposis coli. Cancer genet. Cytogenet., 8: 27-50.

Dulbecco, R. and J. Elkington (1973) Conditions limiting multiplication of fibroblastic and epithelial cells in dense cultures. Nature 246: 197-199.

Dulbecco, R. (1970) Topoinhibition and serum requirement of transformed cells. Nature 227: 802-806.

Emerit, I. and P. A. Cerutti (1981) Tumor promoter phorbol-12-myristate-13-acetate induces chromosomal damage via indirect action. Nature 293: 144-149.

Emerit, I. and P. A. Cerutti (1982) Tumor promoter phorbol-12-myristate-13-acetate induces a clastogenic factor in human lymphocytes. Proc. Natl. Acad. Sci. (USA) 79: 7509- 7513.

Folkman, J. and A. Moscona (1978) Role of cell shape in growth control. Nature 273: 345-349.

Friedman, J. and P. Cerutti (1983) The induction of ornithine decarboxylase by phorbol- 12-myristate-13-acetate or by serum is inhibited by antioxdants. Carcinogenesis 4: 1425- 1427.

Friedman, E., S. Gillin and M. Lipkin. (1984) 12-O-tetradecanoylphorbol-13-acetate stimulation of DNA synthesis in cultured preneoplastic familial polyposis colonic epthelial cells but not in normal colonic epthelial cells. Cancer Res. 44: 4078-4086.

Friedman, E., C. Urmacher, and S. Winawer (1984) A model for human colon carcinoma evolution based on the differential response of cultured preneoplastic, premalignant, and malignant cells to 12-O-tetradecanoylphorbol-13-acetate. Cancer Res. 44: 1568-1578.

Froehlich, J. E. and T. P. Anastassiades (1975) Possible limitation of growth in human fibroblast cultures by diffusion. J. Cell. Physiol. 86: 567-579.

Gainer, H. St. C., S. Schor, and A. R. Kinsella (1984) Susceptibility of skin fibroblasts from individuals genetically predisposed to cancer to transformation by the tumor promoter 12-O-tetradecanoylphorbol-13-acetate. Int. J. Cancer 34: 349-357.

Gardner, E. J. and R. C. Richards. (1953) Multiple cutaneous and subcutaneous lesions occurring simultaneously with hereditary polyposis and osteomatosis. Am. J. Hum. Genet. 5: 139-147.

Gardner, E. J., S. W. Rogers, and S. Woodward. (1982) Numerical and structural chromosome aberrations in cultured lypmphocytes and cutaneous fibroblasts of patients with multiple adenomas of the colorectum. Cancer 49: 1413-1419.

Ham, R. G. and W. L. McKeehan (1978) Development of improved media and cultural conditions for clonal growth of normal diploid cells. In Vitro. 14: 11-22.

Harper, R. A. and G. Grove (1979) Human skin fibroblasts derived from papillary and reticular dermis: Differences in growth potential in vitro. Science 204: 526-527.

Harel, L., M. Jullien, and M. De Monti (1978) Diffusible Factor(s) controlling density inhibition of 3T3 cell growth: A new approach. J. Cell. Physiol. 96: 327-332.

Hayflick, L. (1965) The limited in vitro lifetime of human diploid cell strains. Exptl. Cell Res. 37: 614-636.

Hayflick, L. and P. S. Moorhead (1961) The serial cultivation of human diploid cell strains. Exptl. Cell Res. 25: 585-621.

Heby, O. (1981) Role of polyamines in the control of cell proliferation and differentiation. Differentiation 19: 1-20.

Heim, S., R. Berger, A. Bernheim, and F. Mitelman. (1985) Constitutional C-band pattern in patients with adenomatosis of the colon and rectum. Cancer Genetics and Cytogenetics. 18: 31-35.

Heim, S. (1983) In vitro growth characteristics of skin fibroblasts from patients with adenomatosis of the colon and rectum and their relatives. Clin. Genetics., 23: 41-48.

Heim, S. (1985 a) Normal high resolution karyotypes in patients with adenomatosis of the colon and rectum. Hereditas 102: 171-175.

Heim, S. (1985 b) Spontaneous and mitomycin C induced structural chromosome aberrations in lymphocyte cultures from patients with adenomatosis of the colon and rectum. Hereditas 103: 235-238.

Heim, S., S. G. Johansen, A. M. Kolnig, and B. Strombeck. (1985) Increased levels of spontaneous and mutagen induced chromosome aberrations in skin fibroblasts from patients with adenomatosis of the colon and rectum. Cancer Genet. Cytogenet. 17: 333- 346.

Hennings, H. R., R. Shores, R. Wemk, M. L. Spangler, E. F. Spangler, R. Talrone, and S. H. Yuspa (1983) Malignant conversion of mouse skin tumors is increased by tumour intitiators and unaffected by tumor promoters. Nature (London) 304: 67-71.

Holley, R. W. (1975) Control of growth of mammalian cells in cell culture. Nature 258: 487-490.

Holley, R. W. (1974) Serum factors and growth control. In: Control of proliferation in animal cells. B. Clarkson and R. Baserga, ed. Cold Spring Harbor Press, New York. 13- 18.

Holley, R. W. and J. Kierman. (1968) "Contact inhibition" of cell division of 3T3 cells. Proc. Natl. Acad. Sci. (USA) 60: 300-304.

Hori, T., M. Murata, and J. Utsunomiya. (1980) Chromosome aberrations induced by N- methyl-N-nitroguanidine in cultured skin fibroblasts from patients with adenomatosis coli. Gann. 71: 628-636.

Hsu, S. H. (1983) Multiclonal origen of polyps in Gardner syndrome. Science 221: 951- 953.

Kopelovich, L. (1977) Phenotypic markers in human skin fibroblasts as possible diagnostic indices of hereditary adenomatosis of the colon and rectum. Cancer 40: 2534- 2541.

Kopelovich, L., L. M. Pfeffer, and N. Bias (1979) Growth characteristics of human skin fibroblasts in vitro. A simple experimental approach for the identification of hereditary adenomatosis of the colon and rectum. Cancer 43:218-223.

Kopelovich, L. (1982 a) Hereditary adenomatosis of the colon and rectum: Relavence to cancer promotion and cancer control in humans. Cancer Gen. Cytogenet. 5: 333-351.

Kopelovich, L. (1982 b) Genetic predisposition to cancer in man: In vitro studies. Int. Rev. Cyt. 77: 63-87.

Kruse, P. F., Jr. and E. Miedema (1965) Production and characterization of multiple- layered populations of animal cells J. Cell Biol. 27:273-279.

Kuehl, F. A., J. L. Humes, T. W. Egan, E. A. Ham, G. C. Beverbridge, and C. G. Van Arman (1977) Role of endoperoxide PGG2 in the inflammatory process. Nature 265:170- 173.

Lasne, C., A. Gentil, and I. Chouroulinkov (1974) Two-stage malignant transformation of rat fibroblasts in tissue culture. Nature 247: 490-491.

Luk, G. D. and S. B. Baylin (1984) Ornithine decarboxylase as a biological marker in familial colonic polyposis. New Engl. J. Med. 311: 80-93.

Marx, J. L. (1983) Do tumor promoters affect DNA after all? Science 219: 158-159.

McKeehan, W. L. and R. G. Ham (1976) Methods for reducing the serum requirement for growth in vitro of nontransformed diploid fibroblasts. Joint WHO/IABS symposium on the standardization of cell substrates for the production of virus vaccines, Geneva, Develop. Biol. Stand. 37: 97-108.

McKeehan, W. L. and K. A. McKeehan (1981) Extracellular regulation of fibroblast multiplication: A direct kinetic approach to analysis of role of low molecular weight nutrients and serum growth factors. J. of supramol. Struct. 15: 83-110.

McKeehan, W. L., K. A. McKeehan and D. Calkin (1980) Extracellular Regulation of fibroblast multiplication. Quantitative differences in nutrient and serum factor requirements for multiplication of normal and SV40 virus-transformed human lung cells. J. Biol. Chem. 256: 2973- 2978.

McKeehan, W. L., K. A. McKeehan, S. L. Hammon, and R. G. Ham (1977) Improved growth of human diploid fibroblasts at low concentrations of serum protein. In Vitro 13: 399-416.

McKusick, V. A. (1974) Digestive Diseases 19: 954-958.

Maroudas N. G. (1974) Short-range diffusion gradients Cell 3:217-219.

Menth, M. (1984) The genetic consequences of nucleotide precursor pool imbalance in mammalian cells. Muta. Res., 126: 107-112.

Merz, G. S. and Ross J. D. (1969) Viability of human diploid cells as a function of in vitro age. J. Cell. Physiol. 74: 219- 222.

Merz, G. S. and Ross J. D. (1973) Clone size variation in the human diploid cell strain, WI-38. J. Cell. Physiol. 82: 75-80.

Miskin, R. and R. Ben-Ishai (1981) Induction of plasminogen activator by UV light in normal and xeroderma pigmentosum fibroblasts. Proc. Natl. Acad. Sci. (USA) 78: 6236- 6240.

Miyaki, M., N. Akamatsui, M. Rokutanda, T. Ono, H. Yoshikura, M. Saski, A. Tonomura, and J. Utsunomiya (1980) Increased sensitivity of fibroblasts of skin from patients with adenomatosis coli and peutz-Jeghers' syndrome to transformation by murine sarcoma virus. Gann 71: 797-803.

Mondal, S. and C. Heidelberger (1976) Transformation of C3H/10T1/2 CL8 mouse embryo fibroblasts by ultraviolet irradiation and a phorbol ester. Nature 26: 710-711.

Moolgavkar, S. H. and A. G. Knudson. (1981) Mutation and cancer: A model for human carcinogenesis. Jour. Natl. Cancer Inst. 66: 1037-1052.

Obrien, T. G., R. C. Simsiman, and R. K. Boutwell (1975) Induction of the polyamine- biosynthetic enzymes in mouse epidermis and their specificity for tumor promotion. Cancer Res. 35: 2426-2433.

Obrien, T. G. (1976) The induction of ornithine decarboxylase as an early possibly obligatory, event in mouse carcinogenesis. Cancer Res. 36: 2644-2653.

Ohno, T. (1979) Strict Relationship between dialyzed serum concentration and cellular life span in vitro. Mech. of Ageing and Devel. 11: 179-183.

Parashad, R., K. K. Sanford, and G. M. Jones (1983) Chromatid damage after G2 phase X-irradiation of cells from cancer-prone individuals implicates deficiency in DNA repair. Proc. Natl. Acad. Sci. (USA) 80: 5612-5616.

Paul, D., M. Henahan, and S. Walter (1974) Changes in growth control and growth requirements associated with neoplastic transformation in vitro J. Natl. Cancer Inst. 53: 1499- 1503.

Pegg, A. E. and P. P. McCann (1982) Polyamine metabolism and function. Cell Physiol. 12: C212-C221.

Pero, R. W., M. Ritchie, S. J. Winawer, M. Markowitz, and D. G. Miller (1985) Analysis of unscheduled DNA synthesis in mononuclear leukocytes from patients with colorectal polyps. Cancer Res. 45: 3388-3391.

Pero, R. W., S. Heim, and C. Bryngelsson (1986) Lower rates of thymidine incorporation into DNA of skin fibroblasts from patients with adenomatosis of the colon and rectum. Carcinogenesis. 7: 541-545.

Pero, R. W., D. G. Miller, M. Lipkin, M, Markowitz, S. I. Winawer, W. Enker, and R. Good (1983) A reduced capacity for DNA synthesis in patients with or genetically predisposed to colorectal cancer. J. Natl. Cancer Inst. 70: 867-975.

Pfeffer, L. and L. Kopelovich (1977) Differential genetic susceptibility of cultured human fibroblasts to transformation by Kirsten murine sarcoma virus. Cell 10: 313-320.

Pfeffer, L., L. M. Lipkin, O. Stutman, and L. Kopelovich (1976) Growth abnormalities of cultured human skin fibroblasts derived from individuals with hereditary adenomatosis of the colon and rectum. J. Cell. Physiol. 89: 29-38.

Phillips, P. D., K. Kazuhiko, and V. J. Cristofalo (1984) Progressive loss of the proliferative response of senscing WI-38 cells to platelet-derived growth factor, epidermal growth factor, insulin, transferrin, and dexamethasone. J. of Gerontol. 39: 11- 17.

Phillips, P. D., E. Kuhnle, and V. J. Cristofalo (1983) 125 I-EGF binding ability is stable throughout the replicative lifespan at WI-38 cells. J. Cell. Physiol. 114: 311-316.

Pitot, H. C. (1981) Fundamentals of oncology. Marcel Dekker, Inc., New York, New York. pp. 145-158.

Rasheed, S. and M. B. Gardner (1981) Growth properties and susceptibility to viral transformation of skin fibroblasts from individuals at high risk of colorectal cancer. J. Natl. Cancer Inst. 66: 43-49.

Reddy, A. L. and P. J. Fialkow (1983) Papillomas induced by intiation-promotion differ from those induced by carcinogen alone. Nature 304: 69-71.

Rhim, J. S., Huebner, R. J., Arnstein, and L. Kopelovich (1980) Chemical transformation of cultured human skin fibroblasts derived from individuals with hereditary adenomatosis of the colon and rectum. Int. J. Cancer 26: 565-569.

Rosner, B. A. and V. J. Cristofalo (1981) Changes in specific dexamethasone binding during aging in WI-38 cells. Endocrinology 108: 1965-1971.

Rubin, H. (1983) Adaptive changes in spontaneously transformed Balb/3T3 cells during tumor formation and subsequent cultivation. JNCI 72: 375-381.

Rubin, H. (1984) Early origen and pervasiveness of cellular heterogeneity in some malignant transformations. Proc. Natl. Acad. Sci. (USA) 81: 5121-5125.

Rubin, H., P. Arnstein, and B. M. Chu (1984) High frequency variation and population drift in a newly transformed clone of Balb/3T3 cells. Cancer Res. 44: 5242-5248.

Setlow R. B. (1978) Repair deficient human disorders and cancer. Nature 291: 713-717.

Scher, C. and G. Todaro (1971) Selective growth of human neoplastic cells in medium lacking serum growth factor. Exptl. Cell Res. 68: 479-481.

Schneider, N. R., A. L. Cubilla, and R. S. K. Chaganti (1983) Association of endocrine neoplasia with multiple polyposis of the colon. Cancer 51: 1171-1175.

Schuchardt, W. A. and Ponsky, J. L. (1979) Familial polyposis and Gardner's syndrome. Surg., Gyn., and Obstet. 148: 97-103.

Shall, S. and W. D. Stein (1979) A mortalization theory for the control of cell proliferation and for the origen of immortal cell lines. J. Theor. Biol. 76: 219-231.

Smith, H., C. Scher and G. Todaro (1971) Induction of cell division in medium lacking serum growth factor by SV40. Virology 44: 359-370.

Smith, J. R. and L. Hayflick (1974) Variation in the life-span of clones derived from human diploid cell strains. J. Cell Biol. 62: 48-53.

Smith, J. R., O. M. Pereira-Smith, and E. J. Schneider (1978) Colony size distributions as a measure of in vivo and in vitro aging. Proc. Natl. Acad. Sci. (USA) 75: 1353-1356.

Smith, J. R. and R. G. Whitney (1980) Intraclonal variation in proliferative potential of human diploid fibroblasts: Stochastic mechanism for cellular aging. Science 207: 82- 84.

Snyder, R. D. (1984) The role of deoxynucleotide triphosphate pools in the inhibition of DNA-excision repair and replication in human cells by hydroxyurea. Mutat. Res.131:163- 172.

Steck, P. A., P. G. Voss, and J. L. Wang (1979) Growth control in cultured 3T3 fibroblasts. Assays of cell proliferation and demonstration of a growth inhibitory activity. J. Cell Biol. 83: 562-575.

Stoker, M. G. P. (1973) Role of diffusion boundery layer in contact inhibition of growth. Nature 246: 200-203.

Stoker, M. and D. Piggott (1974) Shaking 3T3 cells: Further studies on diffusion boundary effects. Cell 3: 207-215.

Strobel-Stevens, J. D. and J. C. Lacey (1981) Further evidence for an inhibitor of proliferation elaborated by normal human fibroblasts in culture: partial characterization of the inhibitor. J. Cell. Physiol. 106: 201-207.

Todaro, G., G. Lazar, and H. Green (1965) The initiation of cell division in a contact- inhibited mammalian cell line. J. Cell. Comp. Physiol. 66: 325-333.

Trosko, J. E., J. D. Yager Jr., G. T. Bowden, and Fred R. Butcher (1975) The effects of several croton oil constituents on two types of DNA repair and cyclic nucleotide levels in mammalian cells in vitro. Chem.-Biol. Interactions 11: 191-205.

Waddel, W. R. and R. W. Loughry. (1983) Sulindac for polyposis of the colon. J. Surg. Oncol. 24: 83-88.

Weeks, C. E., A. L. Herrmann, F. R. Nelson, and T. J. Slaga (1982) Alpha- difluoromethylornithine, an irreversible inhibitor of ornithine decarboxylase, inhibits tumor promotor-induced polyamine accumulation and carcinogenesis in mouse skin. Proc. Natl. Acad. Sci. 79: 6028-6032.