To establish well-characterized cellular reagents for the study of colon carcinoma, we have examined 19 human colorectal carcinoma cell lines with regard to morphology, ultrastructure, expression of tumor-associated antigens, proliferative capacity in vitro, anchorage-independent growth, oncogene expression, tumorigenicity and malignant potential. Cell lines examined were cultured under identical conditions, and in vitro and in vivo analyses were performed in parallel on replicate cultures. Three classes of colorectal cell lines were defined according to their tumorigenicity in nude mice. Class-1 lines formed rapidly progressing tumors in nearly all mice at an inoculum of 10(6) cells. Cell lines belonging to class-2 were less tumorigenic, producing tumors later and at a slower growth rate. Class-3 lines were non-tumorigenic under all experimental conditions tested. By Northern analysis, the oncogenes c-myc, H-ras, K-ras, N-ras, myb, fos and p53 were expressed in nearly all cell lines examined. In contrast, transcripts for abl, src and ros were not detected. The best in vitro predictor of tumorigenicity was colony formation in soft agar. There was no detectable correlation between tumorigenicity and metastatic potential, doubling time in vitro, production of tumor-associated markers, xenograft histology or expression of specific oncogenes.

To investigate the role of oncogene activation in the pathogenesis of malignant tumors, we have studied the tumorigenic and metastatic properties of NIH/3T3 secondary transfectants (designated A51) containing an activated c-Haras-) gene derived from the human T24 bladder carcinoma cell line and compared them Tumors are classically defined as benign or malignant (1). Benign tumors are noninvasive growths that do not spread to distant organs. Unless located in a functionally vital site (e.g., brain), they pose little threat to the patient and usually can be removed surgically. In contrast, malignant neoplasms are readily invasive, metastasize to organs throughout the body, and eventually kill their host (1). The biochemical events that distinguish malignant from benign neoplasms remain unidentified. Work in several laboratories has implicated oncogene activation in the expression of tumorigenicity but the role of oncogenes in the pathogenesis of malignant versus benign tumors has received little attention. The NIH/3T3 cell line, although immortalized in vitro, is reportedly nontumorigenic and is widely used as a recipient cell line for detecting transforming gene sequences isolated from tumorigenic cell lines or tissues (2-6). For example, upon transfection with the activated ras gene, NIH/3T3 cells form foci in tissue culture, exhibit anchorage independence, and, in those experiments in which in vivo studies have been conducted, produce tumors in nude mice (6-10). However, the behavior of the tumors formed by transfected NIH/3T3 cells has not been rigorously evaluated and it remains unclear whether oncogene activation is an event associated strictly with the pathogenesis of benign neoplasms or whether activation is also an essential feature for expression of metastatic properties. We report that transfection with the activated c-Ha-ras-l gene accelerates the tumorigenicity and enhances the metastatic potential of NIH/3T3 Tumorigenicity and Metastasis. Tumorigenicity and spontaneous metastatic potential were assayed by inoculating mice with different cell doses in the footpad (i.m.) or the supraclavicular region (s.c.). Tumor size was monitored at the supraclavicular site every 2-3 days by caliper measurement. For studies on experimental metastasis, different numbers of cells were injected into the tail vein of nude mice. At autopsy the major organs of all animals were examined both grossly and histologically for evidence of metastases. Single sections were prepared from each organ except the lung, in which case multiple sections were examined.Detection of Activated c-Ha-ras Oncogene and Human Alu Sequences. For preparation of DNA (12), cell monolayers established from primary tumors or metastatic foci were dispersed into phosphate-buffered saline (Pi/NaCl), pelleted, rinsed, resuspended in 10 mM Tris HCl, pH 8.0/0.35 M NaCl/1 mM EDTA, lysed in 0.5% NaDodSO4, and treated for 4-12 hr with Pronase (0.1 mg/ml) at 37°C. DNA was extracted with phenol, ethanol precipitated, and dissolved in 10 mM Tris-HCl, pH ...

The eradication of established metastases in patients with malignant tumors is the single most important objective in clinical oncology. The current panel of antineoplastic agents discovered through random and semiempirical screening procedures has proven largely ineffective in treating disseminated disease and there is a clear and urgent need for more efficient antimetastatic drugs. Unfortunately, although progress has been made in examining the biology of metastatic spread, our understanding of the pharmacology, biochemistry and molecular genetics of this process is meager and insufficient to provide a rational foundation for the design of mechanism-based antineoplastic agents. Faced on the one hand with the failure of existing drugs to control metastatic spread and on the other with a dearth of alternative pharmacological approaches, the prospect of offering significantly improved therapy to the cancer patient of the 1990's is poor. The challenge of the coming decade lies in obtaining better insights into the molecular mechanisms of metastasis and using this information to identify pharmacological opportunities to curtail the proliferation of secondary tumor growths. As a first step toward this goal we need to define more rigorously what constitutes a therapeutic target in malignant disease and what steps in the pathogenesis of cancer metastasis represent the gravest risk to the patient and thus are most eligible for direct pharmacological intervention. In addressing these issues and developing future strategies for antimetastatic drugs, Paget's 100 year-old 'seed and soil' hypothesis continues to offer a useful conceptual framework for analysis of metastatic behavior. Although Paget's proposal has been validated by a century of clinical observation, efforts to define the 'seed and soil' theory in molecular terms have not been attempted. With the advent of more efficient methodologies for culturing human normal and neoplastic cells coupled with the availability of microanalytical technologies it now becomes possible to investigate and identify the complementary biochemical components of the tumor cell 'seed' and organ 'soil' that combine to encourage the proliferation of metastases. With this information the design of specific pharmacological strategies to uncouple the 'seed and soil' relationship may emerge as a potential therapeutic approach for antagonizing the growth of disseminated malignant tumors.

The role of DNA methylation in the generation of tumor cell variants with altered growth behavior has been investigated. Cultures of the clonally heterogeneous B16 melanoma cell line and a clonal population (B16-CL) derived from it were treated with the DNA hypomethylating agent, 5-azacytidine (5-Aza-CR). The tumorigenic and metastatic properties of (sub)clones isolated from these cultures before and after drug treatment were assayed by injection via multiple routes into syngeneic C57BL/6 mice using a range of cell doses. The rate of tumor growth was monitored following intrafootpad (i.f.p.) injection and the tumor incidence was calculated from the frequency of tumor formation at i.f.p. and supraclavicular subcutaneous (s.c.) sites. Formation of both spontaneous (i.f.p., s.c. inoculations) and experimental (intravenous (i.v.) inoculation) metastatic potential was also investigated. The most consistent effect of 5-Aza-CR was the introduction of heterogeneity with respect to the tumorigenic phenotype. The effect of 5-Aza-CR treatment on metastatic behavior was variable. The majority of tumor cell variants that arose following 5-Aza-CR treatment displayed decreased malignant potential and reduced DNA methylation levels relative to untreated control cells, but the correlation was not absolute. The decreases in DNA methylation levels induced by 5-Aza-CR were unstable and began to rebound within 1 week of drug treatment. The results of the current study indicate that although 5-Aza-CR can introduce significant shifts in the malignant properties of treated cells, the direction and magnitude of the induced alterations are not predictable and are influenced by a variety of experimental parameters including the starting tumor cell population, route of tumor cell inoculation, and the drug treatment protocol. In addition, because DNA methylation levels can rebound rapidly (days) it is difficult to correlate changes in this parameter with the observed alterations in malignancy, which can only be assessed in long-term biological assays (weeks).

No abstract