Search

We investigated TGF-beta response and the expression of TGF-beta receptors in clones of MOSER colon carcinoma cells (designated MOSER II and MOSER III-10) as a function of their growth state. TGF-beta 1 response as assessed by induction of fibronectin expression was higher (3.0-fold) in exponentially growing cells than in quiescent cells. The expression of type I receptor (RI) mRNA was greater (2.5 to 3.0-fold) in exponentially growing cells than in quiescent cells. In contrast, the expression of type II receptor (RII) mRNA was marginally increased in quiescent cells relative to exponential cells. Nuclear run-off assays, and actinomycin-D treatment indicated that the increased expression of RI mRNA in exponentially growing cells was primarily due to an increase in transcription, while a marginal increase in mRNA level for RII in quiescent cells was primarily due to an increase in mRNA stability. Affinity cross-linking with 125I-labeled TGF-beta 1, showed that the exponentially growing cells displayed greater amounts of 125I TGF-beta 1 binding to RI and RII than quiescent cells, indicating that increased cell surface expression of receptors was correlated with increased response in the exponential growth state. Immunoblot analysis also indicated greater amounts of RI protein in exponential compared to quiescent cells; however, no difference in RII protein was observed in the two growth states. These data indicate that expression of the receptors responsible for TGF-beta signal transduction are differentially controlled.

Spontaneously arising, TGF beta 1-resistant colonies were isolated directly from the soft agarose plates of MOSER human colon carcinoma cells grown in the presence of TGF beta 1 but in the absence of serum. The colonies were cloned by limiting dilution and screened in a monolayer proliferation assay for sensitivity to TGF beta 1 and TGF beta 2 isoforms. Cell clones selectively sensitive or resistant to these isoforms in the growth inhibition assay displayed similar differential sensitivities to TGF beta isoforms for production of the extracellular matrix proteins laminin and fibronectin, as well as for the expression of the colon cell differentiation marker carcinoembryonic antigen. Differential receptor binding profiles for TGF beta 1 and TGF beta 2 were observed among the clones. The isolation of cell clones selectively resistant or sensitive to TGF beta isoforms as well as the identification of differential receptor binding profiles among the clones indicate the heterogeneity of TGF beta responsiveness that exists naturally in human colon tumor cells and stress the importance of defining mechanisms underlying differential responsiveness to TGF beta isoforms.

We show that integrin alpha 5 subunit expression is stimulated when human fibrosarcoma HT1080 cells are released from quiescence. The alpha 5 subunit mRNA level in quiescent HT1080 cells was increased 24 hr after their release by 10% fetal bovine serum-containing medium reaching a maximum of 2.5 fold on day 2. Similar levels of induction of cell-surface alpha 5 subunit protein as well as beta 1 subunit protein were also observed. This resulted in a significant increase of cell attachment to fibronectin. The serum stimulation also increased alpha 5 subunit promoter activity by twofold which was protein synthesis independent. Subsequent deletion of alpha 5 subunit promoter DNA showed that the cis-element responsible for the activation is located between -92 bp and the transcription start site. The promoter activity was not induced until 12 hr after the release. Comparison of the effect of a serum-free medium and a 10% fetal bovine serum-supplemented medium revealed that both the DNA synthesis and alpha 5 subunit induction were independent of exogenous growth factors. The increased integrin alpha 5 beta 1 appears to function by reducing mitogenic activity since blockade of fibronectin binding to its receptor with a RGD peptide, a monoclonal anti-fibronectin antibody, or a monoclonal anti-alpha 5 subunit antibody during the release from quiescence significantly stimulated DNA synthesis. On the other hand, stable overexpression of the alpha 5 subunit resulted in decreased DNA synthesis.

Here we describe human colon carcinoma cell clones, isolated from a transforming growth factor beta (TGF-beta)-responsive parental cell line, which display differential sensitivities to TGF-beta 1 and TGF-beta 2 isoforms. In a monolayer proliferation assay, some clones were sensitive to both isoforms (IC50 = 0.1-0.6 ng/ml; S1S2) while others were resistant to both isoforms (IC50 > 5 ng/ml; R1R2). Still other clones (R1S2) were sensitive to TGF-beta 2 (IC50 = 0.1-0.2 ng/ml), but were resistant to TGF-beta 1 (IC50 > or = 5 ng/ml). In S1S2 cells, both TGF-beta isoforms resulted in a repression of c-myc mRNA expression, a concentration-dependent increase in fibronectin levels, and an enhanced production of the colon cell differentiation marker carcinoembryonic antigen. In contrast, R1R2 cells did not display these responses, or did so only to a limited extent. In R1S2 cells, TGF-beta 2 elicited these responses, yet TGF-beta 1 was essentially without effect. Receptor cross-linking experiments indicated that TGF-beta resistance in this model system was not generally associated with a complete lack of expression of either type I or II receptors. Moreover, the R1S2 type clones were heterogeneous, although the majority of them displayed binding to type I receptors by TGF-beta 2 but not by TGF-beta 1. These data suggest that either the TGF-beta 1 and TGF-beta 2 isoforms differ with respect to their ability to interact with the type I and II classes of receptors, or the TGF-beta 1 and TGF-beta 2 isoforms can interact with distinct receptor proteins of the type I and II classes in this model system.

Gastrin is transcriptionally responsive to EGF stimulation (Merchant et al., 1991, Mol. Cell. Biol., 11:2686-2696). Consequently, we hypothesized that previously recognized gastrin autocrine loops (Hoosein et al., 1990, Exp. Cell. Res., 186:15-21), might be controlled by autocrine TGF alpha in human colon carcinoma cells. Therefore, we examined the interaction between these two autocrine growth factors in two colon carcinoma cell lines which utilize TGF alpha. The FET cell line requires exogenous TGF alpha/EGF for optimal growth and has a classical TGF alpha autocrine loop which is disrupted by TGF alpha or epidermal growth factor receptor (EGFr) antibodies. The HCT 116 cell line is not dependent on exogenous TGF alpha/EGF and exhibits a nonclassical TGF alpha autocrine loop which is not disrupted by neutralizing antibodies to either TGF alpha itself or the EGFr. Basal gastrin mRNA production is significantly higher in HCT 116 than FET as measured by RNase protection assay. In the FET cells, exogenous EGF stimulates gastrin mRNA production but not in HCT 116. When the TGF alpha autocrine loop in HCT 116 is disrupted by constitutive expression of antisense TGF alpha mRNA, the gastrin mRNA level is significantly repressed. In xenografts derived from these antisense clones, TGF alpha reverted to high expression, and the gastrin mRNA level was again increased. This interaction between the strong TGF alpha loop in HCT 116 and the gastrin autocrine loop may confer a growth advantage to these colon cells. Such interactions between growth factors may promote enhanced tumorigenicity to transformed cells with these strong, nonclassical autocrine loops.

Although a great deal is known about the cellular effects of exogenous transforming growth factor-beta (TGF-beta) treatment and the effects of various exogenous agents (including TGF-beta's themselves) on TGF-beta expression, studies of cellular controls for autocrine TGF-beta expression and function have been rare. Since exogenous TGF-beta treatment blocks progression through the cell cycle, it seemed likely that autocrine TGF-beta activity would be induced by growth states in which there was little or no cell division such as confluency or quiescence. Specific TGF-beta 1 or beta 2 neutralizing antibody treatment of a colon carcinoma cell line designated CBS showed that autocrine TGF-beta activity could be demonstrated in quiescent cells but not in preconfluent cells. Studies of kinetics of TGF-beta 1 and beta 2 mRNA levels during the establishment of quiescence revealed a significant increase of both isoforms in quiescent cells. The quiescent cells also secreted three- to fourfold and four-to fivefold higher levels of total (latent plus active) TGF-beta 1 and beta 2 protein in the conditioned media than the confluent cells and preconfluent cells, respectively. There was no detectable active form of either TGF-beta isoform in the conditioned media of preconfluent cells, whereas a significant amount of active TGF-beta 1 and beta 2 was detected in the conditioned media of quiescent cells. Quantitative RNase protection assays were developed to compare the effects of cell crowding vs quiescence on TGF-beta expression. TGF-beta 1 was primarily induced by quiescence. TGF-beta 2 was induced by both quiescence and cell crowding. Increased TGF-beta 1 mRNA levels appeared to be exclusively due to an increase in stability, while increased TGF-beta 2 mRNA levels were due to increased transcription. This growth state-related induction of TGF-beta's was also observed in two other colon carcinoma cell lines. These studies show that TGF-beta 1 and beta 2 are autocrine-negative factors which can be situationally expressed by cells as a function of their growth state. Autocrine expression of the TGF-beta's in this model system appears not to affect exponentially growing cells, but rather to function by maintaining a quiescent state and/or by blocking progression through the cell cycle.

Colon carcinoma is a multistage disease. Most malignancies arise from pre-existing benign tumors. Multiple chromosomal defects affecting oncogene and tumor suppressor gene function are associated with disease progression. These aberrations result in an imbalance between the normal positive and negative growth effectors, which contribute further to disease progression. We have studied how changes in the expression of TGF alpha and TGF beta affect colon carcinoma cell behavior. Overexpression of the stimulatory factor TGF alpha in a relatively benign cell line with weak TGF alpha autocrine activity converted the cell type to an aggressive, progressed phenotype in vivo and in vitro. In contrast, disruption of TGF alpha expression by constitutive expression of TGF alpha antisense RNA in a progressed cell line with a strong, internalized autocrine loop resulted in the development of clones with decreased tumorigenicity in vitro and in vivo. Suppression of the inhibitory effects of TGF beta by constitutive expression of TGF beta antisense RNA increased the tumorigenicity of the cell lines in vitro and in vivo. None of these alterations in TGF alpha or TGF beta expression affected the doubling time of the cells. The changes in tumorigenicity were due to effects on the lag phase of growth. We conclude that TGF beta functions to maintain the cells in a quiescent state while TGF alpha drives reentry into the cell cycle. We have identified a unique cis-element that mediates TGF alpha autoregulation. The transcription factor binding this element is also involved in the cell-cycle regulation of TGF alpha expression. We hypothesize that this factor may be a convergent point TGF alpha and TGF beta interact in controlling movement into and out of quiescence.

FET cells are well differentiated human adenocarcinoma cells whose growth is partially inhibited (50-60%) by transforming growth factor-beta 1 (TGF-beta 1). In exponentially growing cultures, TGF-beta 1 induces the expression of transforming growth factor-alpha (TGF-alpha) by 3-fold. To determine whether this induction is the result of increased TGF-alpha promoter activity, FET cells were transiently transfected with a plasmid containing 2816 base pairs of the 5'-flanking region of the TGF-alpha gene linked to luciferase. Transfected FET cells treated with growth-inhibitory concentrations of TGF-beta 1 (10 ng/ml) showed up to a 10-fold increase in luciferase activity. The increase in luciferase activity was dose dependent through the normal physiological range of TGF-beta 1 (0.5-20 ng/ml), saturating at 10 ng/ml. This effect was also TGF-alpha promoter specific, inasmuch as the Rous sarcoma virus long terminal repeat used as a control remained relatively insensitive to the effects of TGF-beta 1. By using progressively smaller portions of the TGF-alpha promoter region, the TGF-beta 1-responsive element was mapped between base pairs -77 and -201 of the 5'-flanking region. TGF-beta 1 treatment also affected epidermal growth factor receptor levels. FET cells treated with TGF-beta 1 (10 ng/ml) for 48 h showed a 20% decrease in the number of epidermal growth factor receptors and a 2-fold increase in the number of high affinity epidermal growth factor receptors on their surface. These results indicate that TGF-beta 1 acts as a positive regulator of TGF-alpha transcription, and they suggest a possible mechanism by which these cells circumvent the growth-inhibitory effects of TGF-beta 1.

Previous work has shown that repression of negative autocrine transforming growth factor (TGF) beta 1 did not alter the growth rate of a human colon carcinoma cell line, but the time required for the cells to enter exponential growth from lag phase was reduced (S. P. Wu, D. Theodorescu, R. Kerbel, J. K. V. Willson, K. M. Mulder, L. E. Humphrey, and M. G. Brattain, J. Cell Biol., 116: 187-196, 1992). These results have led to the hypothesis that the tumor suppressive activity of autocrine TGF-beta 1 was directed at quiescent nondividing cells rather than actively dividing cells. In order to test this hypothesis, a weakly tumorigenic, well-differentiated human colon carcinoma cell line designated CBS, which expressed autocrine TGF-beta 1 and -beta 2 activity in quiescent cells, but not in exponential growth phase cells, was identified. This cell line was stably transfected with a full-length TGF-beta 1 antisense complementary DNA. Constitutive expression of TGF-beta 1 antisense mRNA in CBS cells resulted in repression of autocrine TGF-beta 1 and -beta 2 protein activity in quiescent cells of approximately 10-fold. TGF-beta 2 repression could have been due to interaction with TGF-beta 1 antisense mRNA, since these two isoforms have a high degree of homology, or it could have been indirectly due to TGF-beta 1 repression, since this isoform has been shown to affect transcriptional and posttranscriptional control of TGF-beta 2.(ABSTRACT TRUNCATED AT 250 WORDS)

The GEO colon carcinoma cell line is weakly tumorigenic in athymic mice and shows differentiated properties both in tissue culture and in xenografts. Proliferating monolayer cultures of GEO cells which normally require exogenous epidermal growth factor (EGF) for optimal growth displayed a marked inhibition in growth upon addition of antibodies that block binding to the EGF receptor or neutralize TGF-alpha. These results indicated that GEO cells utilize TGF-alpha in a weak autocrine loop. The availability of a weakly malignant model system in which TGF-alpha had demonstrable, but low level autocrine activity, permitted the investigation of the role of TGF-alpha in tumorigenesis by generating a stronger autocrine loop through the overexpression of the polypeptide. GEO cells were electroporated with an expression vector containing the human TGF-alpha cDNA, and stable clones were isolated that constitutively expressed the TGF-alpha cDNA in a strong autocrine loop. However, the growth rate of the parental cells in EGF-supplemented medium was the same as that of transfected cells with or without growth factor-supplemented medium. Thus, any biological changes generated by the overexpression of TGF-alpha were due to the autocrine nature of the growth mechanism rather than due to any decrease in doubling time leading to a faster growth rate. Transfected GEO cells showed an increase in anchorage-independent growth and formed tumors more readily in athymic nude mice indicating that TGF-alpha plays a role in progression of transformed properties.

Stable acquired resistance to etoposide (VP-16) or teniposide (VM-26) in HCT116 human colon carcinoma cells and A549 human lung adenocarcinoma cells, was previously obtained by weekly 1-h exposures to either drug (B. H. Long, Natl. Cancer Inst. Monogr., 4: 123-127, 1987). The purpose of this study was to identify possible mechanisms of resistance present in these cells by using human mdr1 and topoisomerase II DNA probes, antibodies to these gene products, and P4 phage unknotting assay for topoisomerase II activities. HCT116(VP)35 cells were 9-, 7-, and 6-fold resistant to VP-16, VM-26, and Adriamycin, respectively, and showed no cross-resistance to colchicine and actinomycin D. These cells had no differences in mdr1 gene, mdr1 mRNA, or P-glycoprotein levels but displayed decreased levels of topoisomerase II mRNA and enzyme activity without any alteration of drug sensitivity displayed by the enzyme. HCT116(VM)34 cells were 5-, 7-, and 21-fold resistant to VP-16, VM-26, and Adriamycin; were cross-resistant to colchicine (7-fold) and actinomycin D (18-fold); and possessed a 9-fold increase in mdr1 mRNA and increased P-glycoprotein without evidence of mdr1 gene amplification. No alterations in topoisomerase II gene or mRNA levels, enzyme activity, or drug sensitivity were observed. A549(VP)28 and A549(VM)28 cells were 8-fold resistant to VP-16 and VM-26 and 3-fold resistant to Adriamycin. Both lines were not cross-resistant to colchicine or actinomycin D but were hypersensitive to cis-platinum. No alterations in mdr1 gene, mdr1 mRNA, or P-glycoprotein levels, but lower topoisomerase II mRNA levels and decreased enzyme activities, were observed. Of the four acquired resistant cell lines, resistance is likely related to elevated mdr1 expression in one line and to decreased topoisomerase II expression in the other three lines.

The receptor binding and cellular growth responses to exogenous epidermal growth factor (EGF) were studied using the DiFi cell line established from a familial adenomatous polyposis patient. The number of cell membrane EGF receptors on DiFi cells, as measured by competitive radioligand binding assays and Scatchard analysis of 125I-EGF binding isotherms, was calculated to be 4.8 x 10(6) receptors/cell. An acid prewash step performed prior to ligand binding assays did not reveal additional receptor numbers. A single, low-affinity receptor population was identified by Scatchard analysis, with an apparent Kd of 4.6 nM. This result was confirmed by radioligand binding studies performed in the presence and absence of the receptor-antagonist monoclonal antibody 528 IgG that binds predominantly to the low-affinity form of the EGF receptor. DiFi cells at 50-60% confluence, when exposed to 50 nM exogenous EGF, exhibited a rapid but partial (30%) reduction in their cell membrane-associated receptor, characteristic of sequestration. Exposure of DiFi cells to 50 nM EGF for longer periods of time (4 h) did not result in any further reduction in EGF-receptor number. The cellular growth response of DiFi cells to exogenous EGF was studied in monolayer cultures as well as in a soft agarose assay. Inhibition of soft agar colony formation was observed at exogenous EGF concentrations greater than 1.7 nM, and inhibition of monolayer growth occurred at EGF concentrations greater than 1 nM. In immune complex kinase assays, the DiFi receptor showed similar specific activity to that from the well-characterized A431 cell line. Additionally, phosphorylation of the receptor on tyrosine was qualitatively similar to that of A431 cells, further suggesting that the DiFi receptors identified by EGF-binding studies were biologically functional.

Previously, we reported that exponentially proliferating cultures of well-differentiated human colon carcinoma cells responded to transforming growth factor beta 1 (TGF-beta) with growth inhibition, alterations in morphology, and increased secretion of the differentiation marker, carcinoembryonic antigen. Poorly differentiated cultures were unresponsive. Here we show that TGF-beta was ineffective in repressing nutrient-stimulated mitogenesis in quiescent, poorly differentiated cells. However, in quiescent, well-differentiated cells, TGF-beta repressed the mitogenic responses to both nutrients alone (by 90%) and to nutrients plus the exogenous stimulatory factors epidermal growth factor (E), insulin (I), and transferrin (T) (by 55-65%). Thymidine incorporation experiments indicated that TGF-beta reduced both the onset and peak mitogenic response to growth factors and/or nutrients in the well-differentiated cells. Additionally, TGF-beta repressed the growth factor (E + I + T)-stimulated upregulation of expression of both c-myc and of transforming growth factor alpha (TGF-alpha) mRNAs in quiescent, well-differentiated cells. TGF-beta also elicited a rapid (t1/2 approximately 1h) down-regulation of c-myc expression in the absence of prior growth factor (E + I + T) stimulation. In contrast, TGF-beta had no effect on c-myc or TGF-alpha mRNA expression in the poorly differentiated cells. The results suggest that TGF-beta exerts rapid inhibitory effects on proliferation-associated genes in quiescent and restimulated, well-differentiated cells. Expression of these genes (c-myc and TGF-alpha) may otherwise (in the absence of TGF-beta) play roles in the cellular signaling of mitogenic responses by growth stimulatory factors in well-differentiated colon carcinoma cells.

Previously, we described a model culture system for comparing responsiveness of poorly differentiated and well-differentiated human colon carcinoma cells to exogenous growth factors. While polypeptide growth stimulators elicited an up-regulation of c-myc, as well as a mitogenic response in the well-differentiated cells, the poorly differentiated cells were insensitive to exogenous growth stimulators. We now show, by thymidine incorporation experiments and autoradiographic analysis, that transforming growth factor beta 1 (TGF-beta) abrogated the mitogenic responses to the growth factors epidermal growth factor + insulin + transferrin (IC50 = 0.8 ng/ml), as well as to nutrients (basal medium; IC50 = 0.2 ng/ml) in the well-differentiated cells. The poorly differentiated cells did not respond to TGF-beta. Moreover, TGF-beta (10 ng/ml) completely abrogated the growth factor-stimulated up-regulation of c-myc in the TGF-beta responsive, well-differentiated colon carcinoma cells. Addition of TGF-beta to the TGF-beta-responsive, well-differentiated cells, at a time after c-myc had been transiently up-regulated in response to growth stimulatory factors, resulted in a loss of responsiveness to TGF-beta. Addition of TGF-beta to these cells at increasing time periods after EIT stimulation also resulted in a loss of the TGF-beta-induced repression of c-myc. The results suggest an important role for c-myc in the mechanism of action of TGF-beta in well-differentiated human colon carcinoma cells.

The peptide gastrin has been shown to have growth stimulatory effects on normal as well as malignant gastrointestinal tissue. In this study, we have examined the possibility of autocrine growth-stimulation of cultured colon tumor cells by a gastrin-like peptide. The gastrin/CCK receptor antagonist dibutyryl cGMP inhibited the proliferation of two human colon carcinoma cell lines HCT 116 (EC50 = 1.3 mM) and CBS (EC50 = 2.5 mM) in a dose-dependent manner. Marked morphological changes were observed in HCT 116 cells after treatment with 1 mM dibutyryl cGMP. In receptor binding assays, dibutyryl cGMP competed with 125I-labeled gastrin for binding to HCT 116 cells (IC50 = 1.8 mM). Another derivative of cyclic GMP, 8-Bromo cGMP used as control due to its considerably weaker affinity for the gastrin/CCK receptor, did not compete with radiolabeled gastrin for binding to HCT 116 cells and had no effect on the morphology or proliferation in monolayer cultures of HCT 116 or CBS cells at concentrations up to 10 mM. Antigastrin/CCK antisera was also found to have dose-dependent cytostatic effects on HCT 116 and CBS cells adapted to growth in serum-free medium. The antiproliferative effect of the gastrin/CCK receptor antagonist and antigastrin/CCK antibodies suggested that a gastrin-like peptide secreted by these cells may promote growth. Radioimmunoassay of the conditioned medium of these two cell lines indicated the presence of a gastrin-like peptide (10-50 pg/10(6) cells/72 h). Northern analysis using an oligonucleotide DNA probe complementary to the nucleotide sequence coding the dodecapeptide carboxyl terminal of human gastrin showed three transcripts (0.7, 3.3, and 3.7 kb) that hybridized with the probe. These data provide, for the first time, evidence for an autocrine growth stimulatory role of a gastrin/CCK-like peptide in cultured colon tumor cells.

The effects of feeder layers of C3H/10T 1/2 cells on the growth of human and mouse cell lines were tested. Feeder layers increased colony formation by cultured cancerous cells in semisolid medium over controls grown in semisolid medium without feeder layers. Maximal colony formation was also attained at a faster rate when feeder layers were used. Plating efficiency by cancerous cells obtained directly from xenografts was increased threefold to fivefold in tissue culture when feeder cells were present as confluent monolayers.

Our investigation of normal, hyperplastic, and neoplastic prostatic tissue during the past 2 1/2 years has produced several findings which have been published or accepted for publication. (a) Cells from hamster prostates with intense histochemically demonstrable acid phosphatase activity (HDAP) after fixation with formaldehyde which we believe to be epithelial cells can be obtained in 97.2% +/- 0.8% purity by velocity sedimentation in a previously described isokinetic density gradient; (b) similarly, cells with HDAP, many of which contain lipofuscin granules, can be obtained as 81.0% +/- 12.2% of nucleated cells from hyperplastic human prostates and as 86.4% +/- 9.4% of nucleated cells from human prostatic carcinomas; (c) more cells were obtained from human hyperplastic prostates and prostates with prostatic carcinoma per gram of tissue with the aid of Pronase than were obtained with trypsin, collagenase, or mechanical methods; (d) more cells per gram of tissue were obtained from surgically removed prostates than from prostates obtained at even very rapid autopsies, and a much larger proportion of the cells from surgically removed prostates were viable as assessed both by dye exclusion and by plating efficiency; (e) none of several substrates and inhibitors which we tested were highly specific for acid phosphatase from purified prostatic epithelial cells compared with several other kinds of purified human cells; and (f) purified hamster prostatic epithelial cells incorporate large amounts of tritiated thymidine in 72-hour cultures.

Previous work from this laboratory established a large bank of human colon carcinoma cell lines in culture and classified them with respect to growth regulatory phenotypes based on several biological and biochemical characteristics. In the present report, Northern analysis indicates that addition of the maturational agent N,N-dimethylformamide (DMF) (1.0%) to proliferating HCT 116 and MOSER cells resulted in a repression of c-myc proto-oncogene expression; retinoic acid (1.0 microM) was less effective in this regard. Repression of c-myc expression by DMF was observed in MOSER and HCT 116 cells, whether it was added to proliferating or late log phase cultures, and was associated with a corresponding reduction in cellular proliferation. The reduction in c-myc expression by DMF did not require new protein synthesis and occurred within a few minutes after its addition, resulting in a 70% reduction within approximately 2 hr. Previous work from this laboratory indicated that transforming growth factor-beta (TGF-beta) elicited alterations in MOSER cells similar to those observed following DMF treatment. The present report demonstrates that proliferating, but not late log phase, MOSER cells responded to TGF-beta with a repression of c-myc expression. Similarly, an inhibition of cellular proliferation was only observed when TGF-beta (10 ng/ml) was added to proliferating cells. Collectively, these findings indicate that repression of c-myc expression is associated with diminished cellular proliferation and the induction of a more benign phenotype in human colon carcinoma cells. Furthermore, this report is the first demonstration of a c-myc response to TGF-beta in an epithelial cell line.

Subpopulations of malignant cells from primary cultures of human colon carcinoma were characterized with respect to their response to mitomycin (MMC). Growth inhibition assays indicated values of 2.06, 0.93, and 0.33 microM for the concentration of drug giving 50% inhibition of growth for sublines HCT 116b, HCT 116, and HCT 116a, respectively. Alkaline elution of filter-bound DNA from cells exposed to MMC in vitro showed a positive correlation between the amount of DNA cross-linking and growth inhibition as a function of drug concentration. Comparable DNA cross-linking was obtained at MMC concentrations of 10 microM for HCT 116b and 5 microM and HCT 116. The cross-linking of DNA from HCT 116a cells at 5 microM MMC was approximately equal to that from HCT 116 cells at doses between 10 and 20 microM MMC. Cross-link removal as a function of time after drug removal of MMC-treated cells was also measured. There was little difference in the rates of alkaline DNA elution after drug removal between HCT 116b and HCT 116a, suggesting that the ability to repair cross-links was not responsible for the differential sensitivities of the cells to MMC. The relative sensitivities of the subpopulations to MMC were reflected in vivo by MMC treatment of nude BALB/c mice bearing xenografts of the cultured sublines.

In order to compare the effects of transforming growth factor (TGF beta) with those of the differentiation promoters N,N-dimethylformamide (DMF) and retinoic acid (RA), the antiproliferative and fibronectin-inducing activities of the three agents were examined. AKR-2B mouse embryo fibroblasts and their chemically transformed counterpart AKR-MCA cells were used as the model system. Growth in monolayer culture of both cell lines was inhibited by TGF beta (EC50 approximately 1 ng/ml), DMF (EC50 approximately 0.5%), and RA (EC50 approximately 1 microM) in a concentration-dependent manner. Time-dependent elevation in fibronectin expression was also observed with all three agents. The EC50 for growth inhibition of both cell lines by TGF beta agreed well with that obtained for stimulation of fibronectin synthesis. A 3-h exposure to TGF beta is sufficient to obtain the maximal fibronectin level observed at 48 h in AKR-2 B cells but not in AKR-MCA cells. Our results indicate that in this system the effects of TGF beta are similar to those of the chemical differentiation inducers DMF and RA. Furthermore, our data also suggest that the TGF beta signal may be processed differently by nontransformed and transformed fibroblasts.

The gastrointestinal hormone gastrin has been shown to stimulate the growth of normal colonic mucosa. To examine for a possible role of gastrin in the proliferation of cultured colon tumor cells, we have studied the effects of two gastrin receptor antagonists, proglumide and benzotript, and of antibodies to gastrin. We find that proglumide (50% effective concentration, 2 to 5 mM) and benzotript (50% effective concentration, 0.4 to 0.8 mM) inhibit the monolayer growth of six human colon cancer cell lines. Addition of exogenous gastrin abrogated the growth-inhibitory effect of proglumide. The anchorage-independent growth of colon carcinoma cells was also inhibited by the two gastrin antagonists. Also, a dose-dependent increase in carcinoembryonic antigen secretion was observed upon treatment with proglumide and benzotript in three cell lines examined. Half-maximal inhibition of labeled gastrin binding was observed at concentrations of 0.4 mM benzotript and 8.6 mM proglumide. In addition, antigastrin antiserum added to HCT 116 cells adapted to growth in serum-free medium resulted in a concentration-dependent inhibition of cellular proliferation. These data suggest that gastrin may function as an autocrine growth factor in colon carcinoma.

We have recently reported that TGF-beta induces a response similar to that of planar polar differentiation promoters in human colon carcinoma MOSER cells. N,N-Dimethylformamide and TGF-beta had similar effects on MOSER cells with respect to reversible inhibition of growth (both in monolayer culture and semisolid medium), induction of fibronectin expression and the induction of morphological alterations (Cancer Res., 47:2950-2954, 1987). Since the expression of carcinoembryonic antigen (CEA) has been reported to be modulated by planar polar compounds that promote differentiation in colon carcinomas, we addressed the issue of whether the differentiation-like effects of TGF-beta on these cells would also encompass modulation of CEA expression in the MOSER cells. The biological modulating effects of TGF-beta on extracellular matrix glycoprotein expression and the expression and secretion of cellular proteins were also studied in view of the reported modulating effects of this growth factor on untransformed, noncolonic cells. In this communication we report that TGF-beta induced the synthesis of fibronectin and laminin but not collagen IV. TGF-beta also induced CEA secretion in a dose-dependent manner. Elevated CEA secretion was detected following 48 h of TGF-beta treatment and a 16-fold increase in CEA secretion was observed following 7 days of treatment. The cells were committed to secrete CEA following one dose of TGF-beta treatment. The enhanced expression of four cellular proteins (Mr 42,000, Mr 48,000, Mr 52,000, and Mr 55,000) and the enhanced secretion of three proteins (Mr 66,000, Mr 200,000, and Mr 400,000) were also induced. Some of these protein alterations were detected as early as 6-24 h following TGF-beta treatment. It is concluded that TGF-beta modulated the production and secretion of CEA, the synthesis of fibronectin and laminin, and the expression and secretion of several cellular proteins in the colon carcinoma MOSER cells. To our knowledge, this is the first report on the modulation of CEA and laminin by TGF-beta in tissue-cultured cells, and is the first report on the modulation of cellular proteins by this growth factor in human colon carcinoma cells.

The cytosolic proteins and phosphoproteins of a mitomycin C-sensitive cell line were examined as progressively greater doses of mitomycin C were administered over a period of 44 weeks. Resistance of the human colon carcinoma cell line increased from a 50% inhibitory concentration of 1 to 6 microM over this time period. Changes in cytosolic protein patterns included increases in the amounts of three proteins with molecular weight X 10(-3)/apparent isoelectric point (Mr/pl) values of 56/6.2, 37/7.3, and 27/6.1. Analysis of in vitro 32P-labeled phosphoprotein patterns revealed reductions in the amounts of four proteins with Mr/pl values of 42/6.3, 40/6.7, 31/6.3, and 25/6.1. One increase was detected in a phosphoprotein with a Mr/pl value of 33/6.1. These changes in cytosolic components paralleled the development of resistance to mitomycin C and may reflect changes in the clonal composition of the cell line as it becomes progressively more resistant to mitomycin C or changes in critical proteins or enzymes involved in the activation or biotransformation of the drug.

Tumorigenic cell lines were established in culture from three transplantable mouse colonic carcinomas designated CT 26, CT36, and CT 51. The cultured lines were characterized for the retention of the biological characteristics of the parental lines. All three cultured lines retained the ability to form tumors in vivo. Serially transplanted parental lines CT 26 and CT 51 grew at a faster rate than did CT 36 and showed a greater propensity for the formation of lung metastases. Similar characteristics were exhibited by the tumors formed from the injection of cultured cells. The cultured cell lines were also evaluated with respect to a number of in vitro markers for cancer. Cultured CT 26 and CT 51 cells formed tumors at lower inocula than did CT 36. CT 26 and CT 51 showed anchorage-independent growth and lack of contact inhibition, while CT 36 grew as a strict monolayer and did not form colonies in 0.27% agarose. CT 26 had the highest saturation density of the cell lines when grown in media supplemented with either 10 or 2.5% fetal bovine serum, while CT 51 had the lowest saturation density under these conditions. The varying degrees of malignancy exhibited by the three cell lines and the overall retention of the biological characteristics of the parental lines by the cultured lines suggest that the cultured cells (without the contaminating stromal elements present in the serially transplanted lines) will provide suitable material for the investigation of the molecular bases of these malignant characteristics.

We demonstrate the differential sensitivity of poorly differentiated and well differentiated human colon carcinoma cells to nutrients alone or to nutrients and polypeptide growth factors under completely serum-free conditions. 3H-Thymidine incorporation into trichloroacetic acid precipitable material and autoradiographic analysis indicated that nutrient replenishment alone was sufficient to initiate DNA synthesis in quiescent poorly differentiated cells, whereas defined polypeptide growth factors produced no additional effect. In contrast, well differentiated cells were mitogenically stimulated to a much greater extent by growth factors (epidermal growth factor + insulin + transferrin), than by nutrient replenishment alone. Expression of the c-myc protooncogene was increased approximately 5-fold after growth factor addition to the well differentiated cells. Maximal expression of c-myc occurred at 4 h post stimulation. In contrast, nutrients resulted in only a slight up-regulation of c-myc (1.8-fold) at approximately 90 min after addition. Addition of nutrients and/or growth factors to the poorly differentiated colon carcinoma cells resulted in an initial decline in c-myc expression (90 min), presumably due to removal of endogenous growth stimulators. Expression of c-myc returned to baseline levels by 24 h after additions. The results indicate that differential sensitivity to polypeptide growth factors is related to differentiation status in this model system and suggest that the insensitivity of poorly differentiated cells to exogenous growth factors may be due to a greater production of autocrine growth stimulators.

A population of cells with increased resistance to 5-fluorouracil was isolated from cultures of the human colonic carcinoma cell line HT29. The resistant cells (HTFU) showed an altered morphology by light and electron microscopy and demonstrated contact inhibition in vitro. DNA assays and chromosome counts showed that HT29 cultures exhibit both hyper- and hypoaneuploidy, while HTFU cultures appear exclusively hypoaneuploid. One year after the isolation of HTFU, both cell lines showed equal sensitivity to 5-fluorodeoxyuridine while HTFU cells retained comparative insensitivity to 5-fluorouracil. Carcinoembryonic antigen production was not demonstrated in pre- or postconfluent cultures of HTFU, although carcinoembryonic antigen was present in both cells and media of HT29 cultures. Growth in semisolid medium was demonstrated for both cultures; however, HTFU showed a lower plating efficiency than did HT29. Tumors were observed in all of the nude mice given injections of HT29 or HTFU cells. Tumors formed from HTFU cells were smaller, and frequently the primary site receded after 6 to 8 weeks; Although in vitro tests suggested a reduced tumorigenic potential for HTFU cells, metastasis was observed only in mice given injections of cells from the HTFU line.

Three subpopulations of malignant cells were isolated from a primary cell culture of a single human colonic carcinoma. The variant cells were established as cell lines designated HCT 116, HCT 116a, and HCT 116b, respectively. In vitro characterizations of the variant lines included growth in 0.5% agarose and growth on confluent layers of mouse fibroblasts. HCT 116a showed the highest colony formation in agarose and on confluent fibroblasts, while colony formation by HCT 116 was higher than that of HCT 116b in both of these systems. All of the variant lines were tumorigenic in athymic nude mice given injections of 10 x 10(6) cells, but the time between inoculation and tumor development (latency period) was approximately 10 times longer for HCT 116b as for HCT 116a and 8 times longer than for HCT 116. HCT 116b was not tumorigenic at an inoculum of 5 x 10(6) cells, while both HCT 116 and 116a were tumorgenic at this level. However, HCT 116a was clearly more tumorigenic than was HCT 116 on the basis of the number of animals developing tumors at inoculate of both 10 x 10(6) and 5 x 10(6) cells and on the basis of their differences in latency periods. While all the cell lines had near diploid numbers of chromosomes, each line showed a distinct histological pattern when grown as xenografts in athymic nude mice.

A polypeptide which inhibits the growth of human carcinoma cells has been characterized from Novikoff rat ascites fluid. This tumor inhibitory factor co-purified with transforming growth factor activity through acid/ethanol extraction and Bio-Gel chromatography. The two activities were completely separated by reverse phase HPLC. The tumor inhibitory factor is heat stable and requires disulfide bonds for bioactivity. This factor inhibited the anchorage independent growth of the more differentiated human colon carcinoma cell lines but did not affect the less differentiated carcinoma cells. The presence of stimulatory and inhibitory activities in the same extracts suggests that the relative concentrations of these factors may be important in the control of cell growth.

The effect of two inhibitors of polyamine biosynthesis, difluoromethylornithine and dicyclohexylammonium sulfate, on the transformed fibroblastic cell line AKR-MCA and its parental counterpart AKR-2B was investigated. Treatment of monolayer AKR-MCA cells with either agent results in morphological changes akin to AKR-2B; the cells appear to be flattened with a polygonal shape. The ability of the inhibitors to alter the phenotype is lost when the cells are cocultured with polyamines. More specifically, putrescine and spermidine abrogate the effects of difluoromethylornithine while only spermidine is effective in reversing the dicyclohexylammonium sulfate induced phenomenon. Further evidence that these enzyme inhibitors are reversing the transformed state of AKR-MCA is obtained from soft agarose experiments. AKR-MCA cells will generate colonies only in the absence of either difluoromethylornithine or dicyclohexylammonium sulfate. Polyamine levels were determined in parental AKR-2B and AKR-MCA cells. The levels of putrescine and spermine were similar in both cell types. In contrast, significantly more (P less than or equal to 0.01) spermidine was expressed by the malignant line [7.3 +/- 0.8 (SD) nmol/10(6) cells] when compared with the untransformed AKR-2B (5.4 +/- 0.8 nmol/10(6)) cells. Intracellular putrescine and spermidine were sensitive to difluoromethylornithine, dicyclohexylammonium sulfate, and dimethylformamide, a planar, polar solvent which has been reported to "normalize" the transformed phenotype. AKR-MCA treated with difluoromethylornithine or dimethylformamide manifested time dependent reductions in both polyamines which preceded morphological changes. Dicyclohexylammonium sulfate similarly caused a 70% reduction in spermidine, but in contrast to the other agents there was a marked accumulation of putrescine. These data concur with the established molecular actions of the two enzyme inhibitors as blockers of ornithine decarboxylase (difluoromethylornithine) and spermidine synthase (dicyclohexylammonium sulfate). The normalizing capacity of dimethylformamide was not compromised by cotreatment with putrescine or spermidine. Both difluoromethylornithine and dicyclohexylammonium sulfate inhibited the growth of monolayer AKR-2B and AKR-MCA. In view of the well documented cytostatic effects of polyamine inhibitors, it is suggested that a decrease in growth by these agents triggers a more normal phenotype in AKR-MCA cells.

The effects of vasoactive intestinal polypeptide (VIP) and dibutyryl cyclic adenosine 3':5'monophosphate (dbcAMP) on two human colon carcinoma cell lines, HCT 116 and GEO, were investigated. VIP and dbcAMP inhibited the growth of both cell lines in monolayer culture in a dose-dependent manner. Within 6 h of treatment with 1 mM dbcAMP or 0.3 microM VIP, numerous mucin-like droplets were secreted by GEO cells. VIP and dbcAMP also increased carcinoembryonic antigen (CEA) secretion. In both cell lines, a 9-fold increase in conditioned medium CEA levels was observed at 1 mM dbcAMP and a 2.6-fold increase at 1.5 microM VIP. Time- and concentration-dependent evaluation in cAMP levels were elicited by VIP in the two cell lines. Immunocytochemical studies for cell-surface glycoprotein detection in GEO cells showed that VIP induced a morphological and functional organization of mucin-secreting cells. These results indicate that VIP and dbcAMP have antiproliferative and strong differentiation-promoting effects in colon cancer cells. This is the first report of VIP-induced mucin secretion in colon tumor cells.

The nuclear protein and phosphoprotein profiles from 3 subpopulations of human colonic carcinoma cells which expressed different levels of neoplastic properties were characterized by two-dimensional electrophoresis. The silver stained nuclear protein profiles were found to be remarkably similar among the subpopulations. However, 2 types of nuclear proteins were found to be selectively modified by phosphorylation/dephosphorylation reactions. The dephosphorylation of type I and the phosphorylation of type II nuclear proteins were found to be associated with the HCT 116a subpopulation which expressed a high level of neoplastic properties. Conversely, the phosphorylation of type I and the dephosphorylation of type II nuclear proteins were found to be associated with the HCT 116b subpopulation, which expressed a low level of neoplastic properties. The HCT 116 subpopulation, which expressed an intermediate level of neoplastic properties, was found to possess an intermediate phosphoprotein profile relative to that of the other two subpopulations. Selective modification of cellular proteins by phosphorylation/dephosphorylation reactions may be involved in the generation of tumor cell diversity and heterogeneity.

We report the continuous growth maintenance of untransformed and chemically transformed fibroblasts (AKR-2B, AKR-MCA cells) in low concentrations of serum (0.1% FBS). The cell lines established (AKR-0.1F, MCA-0.1F) proliferated at rates comparable to cells maintained under high serum conditions (10% FBS). Complete removal of serum from the cells did not induce quiescence. The MCA-0.1F cells were more similar to the untransformed AKR-2B fibroblasts in their morphology, saturation density, inability to form colonies under anchorage-independent conditions, steady-state level of c-myc expression, and kinetics of induction of c-myc in response to specific growth factors. This report demonstrates the utility of this cell line as a nonquiescent model system for investigating growth factor-specific effects in serum-free, cycling cells. Addition of transforming growth factor-beta (TGF-beta) (5 ng/ml) to proliferating MCA-0.1F cells, in the absence of any serum, induced a multilayered growth pattern at confluency, similar to that of AKR-MCA cells maintained in 10% FBS. Other growth factors tested did not elicit this effect. The induction of this growth pattern by TGF-beta was associated with a sustained induction of the c-myc proto-oncogene at confluency, but not with a restoration of anchorage-independent growth. The data suggest that TGF-beta may play a role in the up-regulation of c-myc at confluency previously described for AKR-MCA cells maintained in 10% serum.

A sub-line with increased metastatic ability was previously isolated from an established human colonic carcinoma cell line [Kimball & Brattain (1980) Cancer Res. 40, 1574-1579]. The separation and characterization of the isoenzymes of N-acetyl-beta-D-hexosaminidase from each cell line are reported. The parental cell line contained A and B isoenzymes. The sub-line lacked the A-isoenzyme activity and contained an atypical B isoenzyme that was thermolabile, susceptible to alkylation and of lower molecular weight.

The effects of retinoic acid on a transformed mouse embryo fibroblast cell line (AKR-MCA) were examined. Treatment with retinoic acid restored a non-transformed phenotype to this transformed cell line in a dose dependent manner. Retinoic acid (RA) treated AKR-MCA cells showed a non-transformed morphology, a slower growth rate, and did not grow with anchorage independence. A 38,000 Da protein was phosphorylated to a high degree in the AKR-MCA transformed cell line compared to the non-transformed AKR-2B cell line. RA treatment greatly reduced the level of phosphorylation of this protein in AKR-MCA cells. Growth arrested AKR-MCA cells showed a mitogenic response to nutrient replenishment, but not to epidermal growth factor (EGF). Treatment of AKR-MCA cells with RA restored their ability to respond to EGF while the response to nutrient replenishment was lost. This pattern of growth control was similar to that of the non-transformed AKR-2B cells.

Twenty human colon carcinoma cell lines were studied for their ability to develop some of the characteristics of the normal intestinal epithelium, e.g., epithelial polarity, presence of the actin-binding protein villin, or the occurrence of an enterocytic differentiation either when cultured under standard conditions, as for Caco-2 cells, or when grown in a glucose-free medium, as for HT-29 cells. Except for the regular presence of villin, which can be considered a marker of the colonic origin of the cells, the cell lines of this study could be classified into four types with regard to their differentiation characteristics. In type 1 (only one cell line, i.e., Caco-2) the cells undergo spontaneously an enterocytic differentiation characterized by a polarization of the cell layer with the formation of domes and the presence of an apical brush border the membrane of which is endowed with hydrolases such as sucrase-isomaltase, lactase, amino-peptidase N, dipeptidylpeptidase IV and alkaline phosphatase. In type 2 (three cell lines: HT-29, HCT-EB, and HCT-GEO) the cells are undifferentiated when grown in the presence of glucose but undergo an enterocytic differentiation when grown in the absence of glucose. In type 3 (eight cell lines: HCT-GLY, HCT-FET, HCT-FRI, HCT-CBS, HCT-ALA, Co-115, HRT-18, and SW-1116) the cells are organized into a polarized monolayer with the formation of domes but without any enterocytic differentiation characteristics, whatever the culture conditions. In type 4 (eight cell lines: HCT-116a, HCT-R, HCT-RCA, HCT-Moser, HCT-8R, SW-480, LS-174T, and Vaco-9P) the cells are organized into a multilayer without any feature of epithelial polarity or enterocytic differentiation, whatever the culture conditions.

BMY 25282, a newly designed analogue of mitomycin C (MMC), was assessed for its non-cross-resistant cytotoxic and biochemical action against MMC-resistant human colon carcinoma cells. The analogue has an amidine substituted at position 7 of MMC and has a more efficient intracellular activation to its active species than MMC. In this study we demonstrated that BMY 25282 can overcome MMC resistance in a series of previously described human colon carcinoma cells resistant to MMC (Cancer Res., 44: 5880, 1984). The non-cross-resistance of the analogue in the model was confirmed in vivo by treating tumor xenograft-bearing athymic mice with equitoxic doses of MMC or BMY 25282. We further investigated the formation of interstrand DNA cross-link (IDC) formation by BMY 25282 and MMC. MMC-sensitive cells contained 3 to 8 times as many IDCs as resistant colon carcinoma cells, while no significant differences in IDCs were found between the MMC-sensitive or -resistant cells incubated with BMY 25282. When MMC-sensitive or -resistant cells were exposed to the 70% inhibition concentration of either MMC or BMY 25282, no differences were seen with respect to IDC formation. These studies demonstrate that BMY 25282 is able to overcome MMC resistance in a series of human colon carcinoma cells and that IDC formation in the MMC-sensitive or -resistant cells parallels cytotoxicity for both MMC and the analogue.

Cellular phosphoprotein profiles from normal mouse embryonic fibroblast AKR-2B cells were compared to those of their permanently, chemically transformed malignant counterparts AKR-MCA cells, and AKR-2B cells reversibly transformed by transforming growth factor (AKR-TGF). Similar 32P-phosphorylation profiles were observed for both the AKR-TGF and AKR-MCA cells which were distinct from that of the normal AKR-2B cells. Dimethylformamide (DMF)-induced differentiation of the AKR-MCA cells resulted in a restoration of the normal AKR-2B phosphorylation profile to the malignant AKR-MCA cells.

Numerous investigators have found several substrates and inhibitors to be particularly suited for the demonstration of acid phosphatase of prostatic origin. There has been much controversy over the specificity or lack of specificity of several substrates and inhibitors. We have investigated acid phosphatase activities obtained from several kinds of purified cells. None of the substrates or inhibitors which we studied permitted us to discriminate "prostatic" acid phosphatase from acid phosphatase activities obtained from other kinds of cells.

Molecular alterations of cellular proteins were assessed in three previously described subpopulations of human colon carcinoma cells that were originally isolated from a single human primary colon carcinoma. The subpopulations designated HCT 116b, HCT 116, and HCT 116a exhibited significant differences in their expression of several biological properties. Immunoautoradiographic analysis of membrane components, fractionated by SDS-PAGE and electrophoretically transferred onto nitrocellulose sheets, distinguished the most aggressive tumor cell subpopulation from the intermediate and least aggressive subpopulations. An elevated expression of antigens of molecular weight in the range of 116-120 kd, 92 kd, 55-66 kd, and 31 kd was found to be associated with the most aggressive subpopulation of HCT 116a cells. The binding pattern of RCA I and WGA lectins to membrane components, fractionated by SDS-PAGE and fixed on nitrocellulose, also distinguished the three subpopulations of cells. Elevated bindings of RCA I to 80-92 kd components of HCT 116b cells and elevated bindings of WGA to 120 kd components of HCT 116a cells distinguished the three subpopulations from one another. Analysis of cellular phosphoprotein profiles, following labeling of the cells with 32Pi in the presence of phosphate-free medium, further distinguished the most aggressive subpopulation from the intermediate and least aggressive subpopulations. High level of phosphorylation of 120 kd and 250 kd nuclear components, 30 and 90 kd cytosolic components, and low level of phosphorylation of lower molecular weight membrane components were found to be associated with the most aggressive subpopulation. It is concluded that differences in the expression of membrane antigens, differences in the glycosylation of membrane components, and the selective phosphorylation and/or dephosphorylation of cellular proteins exist in subpopulations of intratumoral colonic carcinoma cells with different biological properties. These biochemical alterations of cellular proteins may play an important role in the generation of phenotypic diversity and heterogeneity of malignant cells.

Previous work indicated that transforming growth factor-beta (TGF-beta) elicits proliferation-inhibitory effects in the human colon carcinoma cell line MOSER. This paper describes the isolation and characterization of spontaneously arising subclones from this TGF-beta-sensitive parental line which were relatively refractory to the inhibitory effects of TGF-beta. While the parental cell line responded to TGF-beta with an inhibition of cellular proliferation in monolayer culture and in soft agarose, an increase in extracellular fibronectin, and a down-regulation of c-myc protooncogene expression, these responses were absent or attenuated in the sublines. However, the resistant clones retained the ability to specifically bind TGF-beta. N,N-Dimethylformamide and retinoic acid, two other agents associated with induction of a partial differentiation-like response in the MOSER parental cells (similar to that elicited by TGF-beta), inhibited the monolayer proliferation of both the parental cells and the TGF-beta-resistant sublines. Thus, the refractoriness observed in the isolated clones was relatively specific for TGF-beta.

A human colon carcinoma cell line resistant to mitomycin C (MMC) was obtained by repeated exposure of a previously described sensitive parental line, HCT 116, to MMC in vitro. Xenografts grown from the MMC-resistant phenotype were not inhibited in MMC-treated animals, while MMC treatment produced growth inhibition in parental cell xenografts. The MMC-resistant phenotype exhibited a greater amount of a Mr 148,000 cell surface protein than did the parental line. The increase in this Mr 148,000 cell surface protein correlated positively with the degree of MMC resistance. Alkaline elution of filter-bound DNA from resistant cells exposed to MMC in vitro showed a decrease in DNA cross-link formation such that a 10-fold higher MMC concentration was required to produce similar cross-link formation in the resistant cell as compared to the parental cell. The development of MMC resistance was not associated with in vitro cross-resistance to other natural product cytotoxic drugs. This model for resistance to MMC will be useful in future studies to define the mechanisms for MMC action and resistance in human colon carcinoma cells.

In this study we have employed a model system comprising three groups of colon carcinoma cell lines to examine the growth-inhibitory effects of two molecular forms of transforming growth factor-beta (TGF-beta), TGF-beta 1 and TGF-beta 2. Aggressive, poorly differentiated colon carcinoma cells of group I did not respond to growth inhibitory effects of TGF-beta 1 or TGF-beta 2, while less aggressive, well-differentiated cells of group III displayed marked sensitivity to both TGF-beta 1 and TGF-beta 2 in monolayer culture as well as in soft agarose. One moderately well-differentiated cell line from group II which has intermediate growth characteristics failed to respond to TGF-beta 1 or TGF-beta 2, but the growth of two other cell lines in this group was inhibited. TGF-beta 1 and TGF-beta 2 were equally potent, 50% growth inhibition for responsive cell lines being observed at a concentration of 1 ng/ml (40 pM). Antiproliferative effects of TGF-beta 1 and TGF-beta 2 in responsive cell lines of groups II and III were associated with morphological alterations and enhanced, concentration-dependent secretion of carcinoembryonic antigen. Radiolabeled TGF-beta 1 bound to all three groups of colon carcinoma cells with high affinity (Kd between 42 and 64 pM). These data indicate for the first time a strong correlation between the degree of differentiation of colon carcinoma cell lines and sensitivity to the antiproliferative and differentiation-promoting effects of TGF-beta 1 and TGF-beta 2.

N-Acetyl-beta-D-hexosaminidase isolated from normal rat colon was compared to that obtained from a transplantable rat colonic carcinoma. Levels of total hexosaminidase from purified epithelial cells of normal colon were similar to those from purified malignant cells from the transplantable tumor. Cultured malignant cells had significantly higher levels of activity than did freshly purified tumor cells. Isoelectric focusing of hexosaminidase from normal rat colon indicated approximately equal amounts of A (pI 5.0 ) and B (pI 8.1) isoenzyme activity. The B isoenzyme (normal cell) was more stable to heat inactivation than was the A isoenzyme and had significantly higher activity at low pH's. In contrast, the B isoenzyme from the tumor was relatively unstable to heat and low pH.

Phosphoproteins from cytosol preparations of methylcholanthrene-transformed AKR mouse (AKR-MCA) cells were compared to those of their untransformed counterparts, AKR-2B cells, by two-dimensional electrophoresis following an in vitro 32P phosphorylation procedure using endogenous kinases and substrates. Five proteins were phosphorylated in the AKR-MCA cells which were not observed in the AKR-2B cells, while six proteins were phosphorylated in the untransformed cells which were not observed in the malignant cells. Treatment of AKR-MCA cells with 1% N,N-dimethylformamide induced the reversion of the malignant cells to a phenotype similar to that of untransformed AKR-2B cells (S. Chakrabarty et al., Cancer Res., 44: 2181, 1984). Treatment of AKR-MCA cells with dimethyl formamide resulted in the restoration of five of the AKR-2B-associated phosphorylations and abolished 2 of the AKR-MCA-associated phosphorylations. AKR-2B cells have been shown to respond to transforming growth factors with reversible phenotypic transformation (R. F. Tucker et al., Cancer Res., 43: 1581, 1983). Transforming growth factor treatment of AKR-2B cells induced all five of the AKR-MCA-associated phosphoproteins and the loss of all six of the AKR-2B phosphoproteins. Epidermal growth factor treatment of AKR-2B cells resulted in the phosphorylation of several proteins which were not observed in either AKR-MCA or untreated AKR-2B cells. Some, but not all, of the AKR-2B-associated phosphorylations were also observed in epidermal growth factor-treated cells. The results of these studies demonstrated qualitative and/or quantitative changes in cytosolic protein kinase-phosphatase activities between transformed and normal AKR-2B cells. Treatment of AKR-MCA cells with dimethylformamide resulted in the restoration of some of the normal AKR-2B cell-associated protein kinase-phosphatase activities.

The functional relationship between membrane/cell surface expression of fibronectin and laminin and transformation/differentiation was examined in an AKR mouse fibroblastic cell model. This model consisted of the untransformed AKR-2B cells, their chemically transformed counterpart (AKR-MCA cells) and the chemically differentiated form of the AKR-MCA cells. The transformed AKR-MCA cells were found to express more surface laminin and less fibronectin than the untransformed AKR-2B cells. The transformed AKR-MCA cells were slower to attach and spread on both plastic and type IV collagen-coated dishes in comparison to the AKR-2B cells. However, a higher percentage of the AKR-MCA cells ultimately attached and spread on the type IV collagen-coated dishes. The induction of differentiation in the AKR-MCA cells by N,N-dimethylformamide (DMF) restored fibronectin to the surface of the AKR-MCA cells but reduced laminin expression only slightly. The DMF-treated AKR-MCA cells resembled the AKR-2B cells in that they rapidly attached and spread on plastic dishes and dishes coated with type IV collagen. They also resembled the AKR-MCA cells in that a high proportion ultimately attached and spread on the collagen-coated dishes.

Carcinoembryonic antigen (CEA) was isolated from a human tumor with 0.02 M sodium phosphate containing 0.14 M NaCl (pH 7.0) (saline) and further studied after treatment with perchloric acid or 8 M urea. Preparations CEA obtained from saline homogenates and both methods of treatment were characterized by isoelectric focusing and gel filtration. Perchloric acid treatment resulted in an approximate 10-fold decrease in protein and a significant loss of CEA as compared to the saline- and urea-treated material. Isoelectric focusing revealed that the resultant CEA subpopulations were dependent upon the method of isolation. Urea- and saline-treated material showed complex isoelectric patterns that were quantitatively dissimilar. Perchloric acid-treated material showed a comparatively simple isoelectric pattern that was not significantly affected by electrofocusing in the presence of urea. Gel filtration on ACA 34 of the CEA obtained from each method of isolation resulted in two peaks of activity. The first peak corresponded to the void volume of the column, and the second peak coeluted with commercially available purified 125I-labeled CEA. Centrifugation of the peaks obtained resulted in a significantly greater loss of CEA from the void peak of each isolation procedure. The amount of CEA lost from the void peaks following centrifugation differed with each method of isolation and suggested the presence of aggregates.

We have recently characterized the growth-inhibitory and cellular responses [carcinoembryonic antigen (CEA) secretion, protein secretion, protein expression, fibronectin and laminin synthesis] of the human colon carcinoma MOSER cell line to transforming growth factor-beta (TGF-beta) (Cancer Res., 47: 2950, 1987; 48: 4059, 1988). We have also recently isolated a subline (MOSER R2) from the parental MOSER cells which, unlike the parental line, is relatively resistant to the growth-inhibitory effect of TGF-beta (Biochem. Biophys. Res. Commun., 150: 711, 1988). We now report on the characterization of the cellular responses of this resistant MOSER R2 subline to TGF-beta and compare its responses to that of the highly growth-inhibition-sensitive MOSER cell line. In view of the reported relationship between CEA expression and differentiation in colon cancer and the ability of colon-derived substrata material to modulate the phenotypic properties of colon cancer cells, additional characterization and direct comparison of the effects of TGF-beta on the two cell lines were also performed with respect to (a) cellular expression of CEA and CEA cross-reactive glycoproteins; and (b) colon-derived substrata material. Unlike the growth-inhibition-sensitive MOSER cells, TGF-beta had no effects on fibronectin/laminin synthesis nor on the cellular morphology of the resistant MOSER R2 cells. TGF-beta was also unable to modulate protein secretion and deposition of substrata material by these cells. However, several other responses of the resistant cells to TGF-beta were found to be similar to that of the sensitive MOSER cells. These responses include: (a) a prolonged and stable secretion of CEA; (b) a prolonged and stable induction of elevated cellular expression of CEA and CEA cross-reactive glycoproteins; and (c) enhancement of the expression of three cellular proteins with molecular weights corresponding to 52,000, 48,000, and 42,000. We further report that the differences observed in the responses to TGF-beta in the two cell lines were not due to differences in TGF-beta binding or other receptor parameters such as the expression of distinct TGF-beta receptor subspecies.

Treatment of the transformed mouse embryo fibroblast cell line AKR-MCA with 1% N,N-dimethylformamide (DMF) resulted in the restoration of a nontransformed phenotype in these cells. In order to determine if an increase in growth inhibitory peptides might be responsible for these changes in growth properties of the DMF-treated AKR-MCA cells we examined the serum-free conditioned medium for its ability to inhibit the anchorage-independent growth of a human colon carcinoma cell line. The extracellular levels of inhibitory activity were two-fold higher in conditioned medium derived from AKR-MCA cells than in AKR-MCA cells grown in 1% DMF (AKR-MCA/DMF). Fractionation of the crude conditioned medium indicated the presence of an Mr 20,000 inhibitory fraction in AKR-MCA/DMF conditioned medium which was reduced in AKR-MCA cells. This Mr 20,000 inhibitory activity was acid and heat stable and sensitive to dithiothreitol and trypsin. In addition to inhibiting the growth of a human colon carcinoma cell line this protein induced colony formation in AKR-2B cells and competed for binding to the transforming growth factor beta (TGF-beta) receptor. Therefore, this Mr 20,000 inhibitory polypeptide induced by DMF is probably TGF-beta. TGF-beta was also shown to inhibit the growth of AKR-MCA cells in monolayer culture.