Antitumor and antimetastatic activity of ribozymes targeting the messenger RNA of vascular endothelial growth factor receptors.

Chemically stabilized hammerhead ribozymes are nuclease-resistant, RNA-based oligonucleotides that selectively bind and cleave specific target RNAs. Due to their potential for specifically inhibiting gene expression, ribozymes are being investigated for therapeutic applications as well as for the elucidation of gene function. In particular, we have investigated ribozymes that target the mRNA of the vascular endothelial growth factor (VEGF) receptors because VEGF signaling is an important mediator of tumor angiogenesis and metastasis. Here we report pharmacodynamic studies testing anti-Flt-1 (VEGFR-1) and anti-KDR (VEGFR-2) ribozymes in animal models of solid tumor growth and metastasis. Ribozymes targeting either Flt-1 or KDR significantly inhibited primary tumor growth in a highly metastatic variant of Lewis lung carcinoma. However, only treatment with the anti-Flt-1 ribozyme resulted in a statistically significant and dose-dependent inhibition of lung metastasis in this model. The anti-Flt-1 ribozyme was then tested in a xenograft model of human metastatic colorectal cancer in which significant inhibition of liver metastasis was observed. Taken together, these data represent the first demonstration that synthetic ribozymes targeting VEGF receptor mRNA reduced the growth and metastasis of solid tumors in vivo.

Urokinase receptor antagonists: discovery and application to in vivo models of tumor growth.

Urokinase receptor antagonists based on the growth factor domains of both human and murine urokinase which show sub-nanomolar affinities for their homologous receptors have been expressed as recombinant proteins. Further modification of these molecules by preparing fusions with the constant region of human IgG has led to molecules with high affinities and long in vivo half-lives. Smaller peptidic inhibitors have been obtained by a combination of bacteriophage display and peptide analog synthesis. All of these molecules inhibit the binding of the growth factor domain of uPA to the uPA receptor and enhance binding of the uPA receptor to vitronectin. Protein uPA receptor antagonists were tested in an in vivo tumor model using the human breast carcinoma MDAmb231 in immunodeficient mice. Both human and murine receptor antagonists showed significant inhibition of primary tumor growth, demonstrating that in vivo, both tumor and stromal cell uPA receptor dependent plasminogen activation can modulate tumor growth.

Chemical modifications of heparin that diminish its anticoagulant but preserve its heparanase-inhibitory, angiostatic, anti-tumor and anti-metastatic properties.

Structural features of heparin potentially important for heparanase-inhibitory activity were examined by measuring the ability of heparin derivatives to affect the degradation of [3H]acetylated heparan sulphate by tumor cell heparanases. IC50 values were determined using an assay which distinguished degraded from undegraded substrate by precipitation of the latter with cetylpyridinium chloride (CPC). Removal of heparin's 2-O-sulphate and 3-O-sulphate groups enhanced heparanase-inhibitory activity (50%). Removal of its carboxyl groups slightly lowered the activity (18%), while combining the treatments abolished the activity. At least one negative charge on the iduronic acid/idose moiety, therefore, is necessary for heparanase-inhibitory activity. Replacing heparin's N-sulphate groups with N-acetyl groups reduced its activity (37%). Comparing this heparin derivative with 2,3-O-desulphated heparin, the placement of sulphate groups appears important for activity since the two structures have similar nominal linear charge density. In addition, unsubstituted uronic acids are nonessential for inhibition since their modification (periodate-oxidation/borohydride-reduction) enhanced rather than reduced heparanase-inhibitory activity. The most effective heparanase inhibitors (2,3-O-desulphated heparin, and [periodate-oxidized, borohydride-reduced] heparin) were tested in the chick chorioallantoic membrane (CAM) bioassay for anti-angiogenic activity and found to be at least as efficacious as heparin. 2,3-O-desulphated heparin also significantly decreased the tumor growth of a subcutaneous human pancreatic (Ca-Pan-2) adenocarcinoma in nude mice and prolonged the survival times of C57BL/6N mice in a B16-F10 melanoma experimental lung metastasis assay.

Extracellular annexin VI expression is associated with divalent cation-dependent endothelial cell adhesion of metastatic RAW117 large-cell lymphoma cells.

We previously found that cell surface molecules of approximately 70, approximately 35, approximately 32, approximately 22, and approximately 14 kDa from liver-metastatic murine RAW117 large-cell lymphoma cells bound to target liver microvessel endothelial cells. Isolation and sequencing of the approximately 35-kDa component revealed it to be annexin II, a Ca(2+)-binding molecule involved in cytoskeletal and membrane interactions. Annexin II antibodies inhibited the adhesion of RAW117 tumor cells to live or fixed liver endothelial cells, and purified tumor cell surface fractions containing the approximately 35-kDa component inhibited partially RAW117 cell-endothelial cell adhesion, suggesting a role for annexins in tumor cell-endothelial cell adhesion. In the present study we identified the 70-kDa cell surface component that binds to hepatic sinusoidal endothelial cells in a Ca(2+)-dependent manner as annexin VI. Cytofluorographic analysis indicated that annexin VI was expressed on the cell surface in slightly higher amounts on highly metastatic RAW117 cells, and it was not removable by EDTA treatment. Anti-annexin VI antibodies inhibited the adhesion of RAW117 cells to fixed or unfixed murine hepatic sinusoidal endothelial cells by approximately 40%, indicating a role for annexin VI in mediating a portion of the Ca(2+)-dependent RAW117 cell adhesion to target liver microvessel endothelial cells.

Expression of galectins on microvessel endothelial cells and their involvement in tumour cell adhesion.

Lactoside-binding lectins (galectins) with molecular weights of about 14.5 kDa (galectin-1) and 29-35 kDa (galectin-3) bind preferentially to polylactosaminoglycan-containing glycoconjugates and have been found on the surface of tumour cells and implicated in cell-cell and cell-extracellular matrix adhesion and metastasis. We have demonstrated by immunoblotting that both galectin-1 and galectin-3 are present in extracts of endothelial cells cultured from bovine aorta, rat lung, mouse lung and mouse brain microvessels, whereas mouse hepatic sinusoidal endothelial cells expressed primarily galectin-1. These galectins were also localized by indirect immunofluorescent labelling on the surface of the different endothelial cells in culture and by immunohistochemical staining in human tissues in vivo. Anti-galectin-1 antibodies inhibited the adhesion of liver-preferring murine RAW117-H10 large-cell lymphoma cells to hepatic sinusoidal endothelial cells or lung microvessel endothelial cells in vitro. The data indicate that galectin-1 is expressed on the extracellular surface of endothelial cells and can mediate in part the adhesion of RAW117-H10 cells to liver microvessel endothelial cells.

Modulation of matrilysin levels in colon carcinoma cell lines affects tumorigenicity in vivo.

The expression of the metalloproteinase matrilysin in the human colon carcinoma cell lines SW480 and SW620 correlates with the ability of the SW620 cells to invade an artificial basement membrane in vitro and metastasize to the liver following injection into the cecum of nude mice in vivo. Transfection of either wild-type or activated forms of matrilysin into the SW480 cells, which do not express endogenous matrilysin, did not reproducibly increase in vitro invasion but increased the tumorigenicity of the cells when injected into the cecum of nude mice. Antisense reduction of matrilysin levels decreased the tumorigenicity of the SW620 cells and subsequent metastasis to the liver. These results suggest that matrilysin gene expression by colon adenocarcinoma cells is not sufficient for tumor invasion and metastasis but contributes to the tumorigenicity and progression of colorectal tumors.

Anti-tumor activity of mycophenolate mofetil against human and mouse tumors in vivo.

Cultured tumor cell lines, tumor xenografts grown in athymic nude mice, and a murine experimental metastasis model were used to assess the in vitro and in vivo anti-tumor activity of the potent IMP dehydrogenase (IMPDH) inhibitor, mycophenolic acid (MPA), and its morpholinoethyl ester pro-drug, mycophenolate mofetil (MM). The growth of all the cell lines tested was inhibited by MPA in vitro, with EC50 values ranging from less than 0.1 microM to 3.9 microM. Mice were monitored for s.c. tumor outgrowth in the case of human tumor xenograft models or survival time for the murine experimental metastasis model. Treatment with MM p.o. was started 24 hr after tumor challenge or after tumors became palpable. Treatment of athymic nude mice bearing A3.01 (T-lymphoblast), Molt-4 (T-cell leukemia), CaPan-2 (pancreatic adenocarcinoma), CaLu-3 (non-small-cell lung adenocarcinoma), LS174T and T84 (colon adenocarcinoma), and Daudi (B-cell lymphoma) human tumor xenografts with MM significantly inhibited s.c. tumor growth. Treatment of BALB/c mice with MM after i.v. injection of murine RAW117-H10 lymphoma cells in an experimental metastasis assay resulted in increased survival time for treated animals. No significant inhibitory effect on s.c. tumor outgrowth was seen with MM treatment of SK-Hep-1, a human hepatic endothelioma, or Hep-3B, a liver adenocarcinoma, at any of the doses tested.

Selection for enhanced adhesion to microvessel endothelial cells or resistance to interferon-gamma modulates the metastatic potential of murine RAW117 large-cell lymphoma cells.

Poorly liver metastatic large-cell lymphoma RAW117-P cells were sequentially selected in vitro for increased adhesion to murine hepatic sinusoidal endothelial cells. After three or four sequential selections, the selected sublines showed increased rates of adhesion to target hepatic microvessel endothelial cells and increased formation of experimental metastases in the liver. However, the endothelial cell adhesion-selected RAW117 sublines were generally unstable and gradually lost their enhanced adhesive and metastatic properties during passage in culture. Highly metastatic, liver-selected RAW117-H10 large-cell lymphoma cells were more resistant to the cytostatic effects of interferon-gamma (IFN-gamma) than poorly metastatic unselected parental RAW117-P cells. When tested for their sensitivity to IFN-gamma, the endothelial cell adhesion variants were significantly more resistant than the unselected RAW117-P cells, but after a 72-h treatment with IFN-gamma, the in vitro-selected cells lost their enhanced endothelial cell adhesion characteristics, their potential to colonize the liver, and their ability to grow when injected at subcutaneous or intramuscular sites. In contrast, the metastatic potential of similarly treated RAW117-P cells was unaffected by IFN-gamma during a 72-h treatment. Sequential selection of RAW117-P cells for increased resistance to IFN-gamma in vitro resulted in variant lines that were refractory to the growth-inhibiting effects of IFN-gamma, and these IFN-gamma-selected variants were also less adhesive to liver microvessel endothelial cells. The IFN-gamma-selected variants also lost their experimental metastatic potentials completely and their tumorigenicities at sites of subcutaneous or intramuscular injection. Cytofluorographic analysis indicated reduced cell surface expression of H-2Kd antigen and fibronectin receptor on the selected variant cells but no change in cell surface mu heavy chain immunoglobulin. The unselected and selected RAW117 lines had similar sensitivities to natural killer (NK) cell-mediated cytolysis, indicating that the in vivo differences were probably not due to differences in NK cell-mediated cytolysis. The results suggest that selection for adhesion to organ microvessel endothelial cells or sequential exposure to certain cytokines can affect the adhesive, growth and metastatic properties of RAW117 cells without modifying their responses to NK cells.

Extracellular annexin II is associated with divalent cation-dependent tumor cell-endothelial cell adhesion of metastatic RAW117 large-cell lymphoma cells.

Using fixed microvessel endothelial cell monolayers the molecules involved in the adhesion of liver-preferring murine RAW117 large cell lymphoma cells to murine liver-derived microvessel endothelial cells were identified by affinity isolation. Detergent lysates obtained from poorly (P) or highly (H10) liver-metastatic cells inhibited RAW117-H10 cell adhesion to hepatic sinusoidal endothelial (HSE) cell monolayers. Allowing detergent lysates of cell surface-labeled RAW117 cells to bind to fixed HSE cell monolayers and eluting the bound components indicated that several tumor cell surface molecules (approximately 70, approximately 35, approximately 32, approximately 22, and approximately 14 kDa) might be involved in RAW117 cell-HSE cell adhesion. The approximately 35 kDa component was cation dependent in its binding to target HSE cells. Increasing detergent concentration had no effect on binding of the approximately 35 kDa component to HSE cell monolayers, whereas treatment with 0.5 M NaCl resulted in its selective elution from HSE cells. Incubation of the HSE cell monolayers with detergent lysates from cell surface-labeled RAW117-H10 cells resulted in selective depletion of the approximately 35 kDa component, suggesting that the binding is saturable. This divalent cation-dependent molecule is one of the major tumor cell surface components bound by several types of endothelial cells and murine hepatocytes, whereas there was poor binding of this component to unfixed or fixed human red blood cells. The purified, partially (approximately 40%) sequenced molecule had amino acid sequence identity with murine but not bovine annexin II, indicating that it was not bound from the bovine serum used to grow RAW117 cells. Using antibodies specific for annexin II flow cytometry indicated equivalent amounts of annexin II are expressed on RAW117 cell surfaces in the absence or presence of excess EDTA, whereas annexin I was only found in low amounts on the surfaces of RAW117 cells. Annexin II antibodies inhibited by approximately 40-50% the adhesion of RAW117 tumor cells to live or fixed endothelial cells, and purified tumor cell surface fractions containing the approximately 35 kDa component partially inhibited (approximately 35%) RAW117 cell-HSE cell adhesion. The data indicate that annexin II is expressed on the extracellular surface of RAW117 cells, and cell surface-annexin II mediates a portion of the Ca(2+)-dependent RAW117 cell adhesion to liver microvessel endothelial cells.

Butanol-extractable and detergent-solubilized cell surface components from murine large cell lymphoma cells associated with adhesion to organ microvessel endothelial cells.

Metastatic variant cell lines of the murine RAW117 large cell lymphoma were used to study the cell surface components involved in syngeneic tumor cell/microvessel endothelial cell interactions. Poorly liver-metastatic parental RAW117-P cell line adhered to murine hepatic sinusoidal endothelial cell monolayers at significantly lower rates than the liver-selected, highly liver-metastatic RAW117-H10 line and both cell lines were poorly adherent to lung microvessel and bovine aorta endothelial cells. Viable, 2% 1-butanol-treated RAW117-H10 tumor cells formed fewer liver tumor nodules in experimental metastasis assays than untreated H10 cells and were significantly less adherent to murine hepatic sinusoidal endothelial cell monolayers. When 2% 1-butanol extracts of metabolically labeled or CHAPS detergent lysates of cell surface-labeled tumor cells were analyzed for their ability to bind to fixed microvessel endothelial cell monolayers, quantitative differences were found in the extractable tumor cell surface components that bound to the different organ-derived microvessel endothelial cells. Cell surface components (1-butanol extractable), of Mr approximately 85,000-90,000 and approximately 37,000-40,000 bound to hepatic sinusoidal endothelial cell monolayers to a greater extent than to murine lung microvessel endothelial or bovine aortic endothelial cell monolayers, whereas tumor cell surface components of Mr approximately 45,000, approximately 33,000, and approximately 25,000 bound similarly to endothelial cells regardless of origin. The results suggest but do not prove that tumor cell/endothelial cell adhesion involves multiple tumor cell surface components, some of which commonly bind to various endothelial cells and others for which binding may be dictated by the tissue origin and type of endothelial cell. Particular RAW117 butanol-extractable cell membrane components were associated with tumor cell-endothelial cell adhesion, and these components could be responsible, in part, for the preferential adhesion of RAW117 cells to liver sinusoidal endothelial cells and metastasis to liver.

Microvascular endothelial cell heterogeneity: interactions with leukocytes and tumor cells.

Endothelium constitutes a highly specialized organ that lines the vascular system and lymphatic channels. This organ is a complex network of arteries, veins, and microvessels that differ in size, structure, and function. The unique and strategic location imposes functional demands on the endothelium that are far greater than just being a passive barrier. Endothelial cells have the ability to differentiate both in structure and function in response to the needs of diverse tissue environments, making this organ extremely heterogeneous. Although vascular endothelial cells share certain common properties, they differ in regard to structure, antigenic and cell surface determinants, adhesion molecules, and metabolic function. The unique cell surface profiles expressed by endothelial cells in different tissue locations can be recognized by specific populations of circulating leukocytes or tumor cells, which contribute to their arrest and invasion patterns. This article attempts to review our current understanding of endothelial cell heterogeneity and its significance to patterns of leukocyte and tumor cell trafficking.

Adhesive, invasive, and growth properties of selected metastatic variants of a murine large-cell lymphoma.

The adhesive, invasive, and growth properties of parental murine large-cell lymphoma cells of low metastatic potential (RAW117-P) were compared to in-vivo-selected sublines of high metastatic potential to liver (RAW117-H10) or lung (RAW117-L17). Using small (approximately 0.5 mm3) pieces of syngeneic organ tissue (lung, liver, kidney) we found that RAW117-L17 cells selectively attached to and invaded lung tissue, whereas RAW117-H10 cells preferentially attached to and invaded liver tissue. We measured adhesion to microvessel endothelial cells established from syngeneic lung and liver and found that the RAW117-L17 cells bound to lung microvessel endothelial cells at significantly higher rates than the other lines, and RAW117-H10 and -L17 cells attached to hepatic sinusoidal endothelial cells at significantly faster rates than RAW117-P cells. Such organ specificity of adhesion was not found at the level of the subendothelial matrix, and the rates of adhesion of RAW117 cells to subendothelial matrix were lower than to endothelial cells. RAW117 cells of low or high metastatic potential bound to immobilized extracellular matrix components, such as fibronectin, at high rats, but adhesion to laminin or collagen IV was minimal. Previous studies indicated that RAW117 lines could proliferate in vitro in certain organ-conditioned media under limiting serum conditions. We therefore examined the ability of a purified paracrine lung growth factor (LDGF-1) to stimulate growth of RAW117 cells in limiting serum-containing medium. The high lung-colonizing L17 line was stimulated to proliferate by LDGF-1 at faster rates than the other lines. The data support Paget's hypothesis that the organ specificity of tumor metastasis is determined by specific tumor cell and host properties.

Correlation of inhibition of adhesion of large cell lymphoma and hepatic sinusoidal endothelial cells by RGD-containing peptide polymers with metastatic potential: role of integrin-dependent and -independent adhesion mechanisms.

Murine RAW117 large-cell lymphoma cells that show organ preferences of metastatic colonization were selected. We examined the role of adhesive systems in determining the organ preference of metastasis using cell lines of low (RAW117-P) and high (RAW117-H10) liver-metastatic potential. Highly metastatic H10 cells adhered at higher rates than low metastatic P cells to target organ microvessel endothelial cells, and these interactions were partially inhibited by RGD-containing polymers but not by small peptides such as GRGDS or GRGES. The most effective polymers, such as (GRGDS)4 and GRGDS(GRGES)2GRGDS, significantly inhibited H10 cell adhesion but had less effect on P cell adhesion to target liver sinusoidal endothelial cell monolayers or on P cell or H10 cell adhesion to bovine aortic endothelial cell monolayers. The (GRGDS)4 polymer reduced the rate of H10 liver sinusoidal endothelial cell adhesion to that of P cells in the absence of inhibitors, suggesting that the quantitative difference in adhesion of H10 cells versus P cells to liver sinusoidal endothelial cells may have been due to integrin-like molecules. Other RGD-containing polymers, such as (GRGES)2(GRGDS)2, GRGES(GRGDS)2GRGES, or (GRGES)4, were less effective, suggesting that the secondary structure of the polymers may be an important consideration. A peptide from the B1 chain of laminin (YIGSR) or its homopolymer, (YIGSR)4, had no effect on endothelial cell adhesion, consistent with the lack of differential laminin adhesion seen with various RAW117 cell lines. The results suggest that integrin-related molecules may play a role in the organ specificity of endothelial cell adhesion seen with RAW117 tumor cells.

Cell surface biochemical and metastatic properties of Lens culinaris hemagglutinin-binding variants of a murine large cell lymphoma.

Using a low metastatic potential parental (P) line of the murine large cell lymphoma RAW117 and a highly metastatic in vivo-selected liver-colonizing subline (H10), we examined the relationship between cell surface glycoprotein expression and metastasis. The highly metastatic H10 cells showed loss of the major RNA tumor virus envelope glycoprotein gp70 and increased expression of a concanavalin A and Lens culinaris hemagglutinin (LcH)-binding glycoprotein of Mr approximately 15,000 (gp150) by lectin affinity chromatography and 125I-lectin staining of isolated RAW117 glycoproteins. When the amounts of cell surface LcH-binding components were determined on P and H10 cells, the mean amount of cell-bound LcH on H10 cells was significantly greater than on P cells. RAW117-P cells were sorted for low (PLcH-low) or high (PLcH-high) LcH binding using a fluorescence-activated cell sorter, or for binding to immobilized LcH, and the resulting cell sublines were analyzed for their metastatic properties by intravenous injection into BALB/c mice. The parental P cells formed few liver tumor nodules (median 0; range 0-8), as did the PLcH-low cells (median 0; range 0), whereas the high LcH-adherent P cells and the cells sorted for increased LcH binding, PLcH-high, were highly metastatic to the liver (median 200; range 156 to 200+). Analyses of gp150 and gp70 contents indicated higher amounts of gp150 but lower quantities of gp70 on PLcH-high cells than on PLcH-low or P cells. The results suggest that the amounts of cell surface gp150 and gp70 are important in determining the metastatic properties of RAW117 cells.

Sialoglycoproteins of murine RAW117 large cell lymphoma/lymphosarcoma sublines of various metastatic colonization properties.

A metastatic model for large-cell lymphoma/lymphosarcoma has been developed by sequential selection in vivo of the murine RAW117 cell line for enhanced liver metastasis or in vitro for loss of lectin-binding properties. The metastatic variants obtained from such selections show alterations in cell surface lectin-binding components, such as the wheat germ agglutinin (WGA)-reactive sialoglycoproteins. Detergent lysates from RAW117 cells were analyzed by polyacrylamide gel electrophoresis (PAGE) followed by reaction with 125I-labeled WGA. The [125I]WGA became bound to a diffuse band of Mr 120 000-200 000 in the gels that overlapped with the major sialoglycoprotein band revealed by the periodate-sodium borotritide labeling. However, the [125I]WGA reactivity diminished when gels were pretreated with mild acid to remove sialic acid in situ. The binding of [125I]WGA to the glycoprotein(s) was greater in the high liver-colonizing RAW117-H10 subline than in the parental RAW117-P line. Another lectin with different saccharide specificity, Ricinus communis agglutinin I (RCAI), became bound to a similar class of sialoglycoproteins, as well as to glycoproteins of lower Mr, but only when the gels were pretreated with mild acid to remove sialic acid. These differences in the relative RCAI-binding intensities after chemical removal of sialic acid were similar to those seen with WGA and indicate that differences in WGA reactivity of this class of sialoglycoproteins were not due to increased sialylation of the carbohydrate chains. Sialic acid was removed from RAW117 cells by neuraminidase treatment, and lysates were analysed for [125I]RCAI reactivity after electrophoresis. The migration of the glycoproteins was not affected by neuraminidase, indicating that the diffuseness of the major sialoglycoprotein band was not due to differences in sialylation. [125I]WGA reactivity to the sialoglycoprotein components, before and after Smith degradation in situ, strongly suggests that the oligosaccharide back-bones are highly branched and asparagine-linked. Only the high Mr portion of the diffuse sialoglycoprotein band was stained with peanut agglutinin (PNA) after in situ removal of sialic acid. To determine whether the expression of the sialoglycoprotein was causally related to liver metastasis, the amounts of sialoglycoproteins in RAW117 cells obtained by in vitro selection for increased or decreased metastasis were examined. Binding of [125I]WGA to intact cells and affinity chromatography of vectorially radiolabeled cell surface proteins on WGA-agarose were performed, and the results indicated that the in vitro selected high liver-colonizing RAW117 variants possesses high WGA r

Differential expression of metastasis-associated cell surface glycoproteins and mRNA in a murine large cell lymphoma.

A metastatic variant cell subline of the Abelson virus-transformed murine large lymphoma/lymphosarcoma RAW117 has been selected in vivo ten times for liver colonization. Highly metastatic subline RAW117-H10 forms greater than 200 times as many gross surface liver tumor nodules as the parental line RAW117-P. Analysis of cellular proteins and glycoproteins indicates reduced expression of murine Moloney leukemia virus-associated p15, p30, and gp70, and increased expression of a sialoglycoprotein, gp150, in the highly metastatic H10 cells. Northern analyses of oncogene expression suggested that mRNA of various oncogenes was expressed equally or not expressed in the RAW117 cells of differing metastatic potential. Differential gene expression was examined using a cDNA library of 17,600 clones established from poly A+ mRNA isolated from H10 cells. The cDNA library was screened by the colony hybridization technique using probes made from both RAW117-P and -H10 cells. Approximately 99.5% of these cDNA clones were expressed identically in P and H10 cells. Of the few differentially expressed cDNA clones (approx. 150/17,600), one-half of these were identified as Moloney leukemia virus sequences in a separate probing with a radiolabeled Moloney leukemia virus probe. The remainder of the differentially expressed mRNA detected by colony hybridization of the cDNA library were expressed at higher levels (approx. 1/6) or lower levels (approx. 1/3) in the highly metastatic H10 cells.