Differentiation of mouse erythroleukemia cells enhanced by alternatively spliced c-myb mRNA.

C-myb, the normal cellular homolog of the retroviral transforming gene v-myb, encodes a nuclear, transcriptional regulatory protein (p75c-myb). C-myb is involved in regulating normal human hematopoiesis, and inhibits dimethyl sulfoxide-induced differentiation of Friend murine erythroleukemia (F-MEL) cells. An alternately spliced c-myb mRNA encodes a truncated version of p75c-myb (mbm2) that includes the DNA binding region and nuclear localization signal present in the c-myb protein, but does not contain the transcriptional regulatory regions. Constitutive expression of mbm2, in contrast to c-myb, here resulted in enhanced differentiation of F-MEL cells. These data suggest that the c-myb protooncogene encodes alternately spliced mRNA species with opposing effects on differentiation.

Transforming potential of human c-sis nucleotide sequences encoding platelet-derived growth factor.

The nucleotide sequence of a transforming human c-sis complementary DNA shows an open reading frame 723 base pairs in length located downstream from an in-phase terminator thymine-guanine-adenine codon. Sequences within this region were identical to those previously determined for the exons of the normal human c-sis gene. Thus, the predicted transforming product, a protein of 27,281 daltons, may be the actual precursor for normal human platelet-derived growth factor chain A.

Cloning and characterization of complementary DNA for human tryptase.

The amino acid sequence of human mast cell tryptase was determined from corresponding cDNA cloned from a lambda ZAP library made with mRNA derived from a human mast cell preparation. Tryptase is the major neutral protease present in human mast cells and serves as a specific marker of mast cells by immunohistologic techniques and as a specific indicator of mast cell activation when detected in biologic fluids. Based on nucleic acid sequence, human tryptase consists of a 244-amino acid catalytic portion of 27,423 D with two putative N-linked carbohydrate binding sites and a 30-amino acid leader sequence of 3,048 D. A His74, Asp120, Ser223 catalytic triad and four cystine groups were identified by analogy to other serine proteases. Regions of amino acid sequence that are highly conserved in serine proteases, in general, were conserved in tryptase. The catalytic portion of human tryptase had an 84% amino acid sequence similarity with that of dog tryptase; their leader sequences had a 67% similarity. Asp217 in the substrate binding pocket of human tryptase is consistent with a specificity for Arg and Lys residues at the site of cleavage (P1), whereas Glu245 is consistent with the known preference of human tryptase for substrates with Arg or Lys also at P3, analogous residues also being present in dog tryptase. Asp244, which is substituted for the Gly found in dog tryptase and in most serine proteases, is present in the putative substrate binding pocket and may confer additional substrate specificity on human tryptase for basic residues. Further studies now can be designed to elucidate these structure-function relationships.

NIH/3T3 cells transfected with human tumor DNA containing activated ras oncogenes express the metastatic phenotype in nude mice.

NIH/3T3 cells transfected with DNA from malignant human tumors produced experimental and spontaneous metastases in nude mice. In contrast, parent or spontaneously transformed NIH/3T3 cells failed to metastasize. The transfected clones contained either activated c-Harvey-ras or N-ras oncogenes. A representative clone (T71-17SA2) which was used to assess selected cellular and host factors relevant to the metastatic process produced lung metastases in 100% of the NIH nude mice recipients, secreted augmented levels of type IV collagenase, and invaded human amnion basement membrane in vitro. Expression of the metastatic phenotype was not related to decreased sensitivity to natural killer cells or macrophage-mediated cytotoxicity. Analysis of the cellular DNA from the T71-17SA2 transfectant and its corresponding metastases, both of which contained activated N-ras oncogenes, revealed a twofold increase in the N-ras-specific DNA sequences in the metastatic cells. Thus, transfection with human tumor DNA containing activated ras oncogenes can induce the complete metastatic phenotype in NIH/3T3 cells by a mechanism apparently unrelated to immune cell killing.

Altered regulation of c-myc in an HL-60 differentiation resistant subclone, HL-60-1E3.

HL-60 cells treated with phorbol dibutyrate (PDBu) differentiate into cells which functionally and morphologically resemble macrophages (G. Rovera, D. Santoli, and D. Damskey, Proc. Natl. Acad. Sci. USA, 75: 2779-2783, 1979; E. Huberman and M.F. Callahan, Proc. Natl. Acad. Sci. USA, 76:1293-1298, 1979). This differentiation involves modulation of the expression of several cellular oncogenes. However, the significance of the temporal relationships between differentiation events and specific oncogene expression are not known. Others have reported that transcriptional down regulation of c-myc occurs early in the differentiation of HL-60 cells (R.D. Dalla-Favera et al., Haematol. Blood Transfusion, 28: 247-253, 1983; L.E. Grosso and H.C. Pitot, Biochem. Biophys. Res. Commun., 119: 473-480, 1984). To determine the significance of the regulation of c-myc during HL-60 maturation, we performed parallel PDBu induction studies analyzing the kinetics of expression of c-myc, cell cycle frequency distribution, cytotoxic effector activity, and clonogenic potential in HL-60 cells and in a partial-differentiation resistant HL-60 clone (HL-60-1E3) (J. A. Leftwich, P. Carlson, B. Adelman, and R. E. Hall, Cancer Res., 47: 1319-1324, 1987). We report that PDBu stimulation results in early c-myc transcriptional down regulation in the HL-60-1E3 clone cells with the same kinetics as has been previously reported for HL-60 parental cells (R. D. Dalla-Favera et al., Haematol. Blood Transfusion, 28: 247-253, 1983; L. E. Grosso and H. C. Pitot, Biochem. Biophys. Res. Commun., 119: 473-480, 1984). However, reexpression of c-myc occurs 15 hours postinduction in HL-60-1E3 but not parental cells. This reexpression is maintained through 30 h of stimulation and correlates with a lack of terminal commitment as assessed by an increase in clonogenic potential and the inability of these cells to acquire cytotoxic function. Sequential stimulation of HL-60-1E3 cells with DMSO and PDBu overcomes the block in macrophage differentiation (J. A. Leftwich, P. Carlson, B. Adelman, and R. E. Hall, Cancer Res., 47; 1319-1324, 1987). Such treatment results in a transient reexpression of c-myc at 15 h after PDBu treatment, and the complete downregulation of c-myc 24 h postinduction. These data suggest that the reported early decrease in c-myc transcripts following PDBu stimulus in HL-60 cells is not sufficient to commit these cells to macrophage-like terminal differentiation. Late regulation of c-myc gene expression may be an important additional component of the regulatory mechanisms which allow HL-60 cells to complete this program.

Microsatellite deletions in the c-myb transcriptional attenuator region associated with over-expression in colon tumour cell lines.

In the hemopoietic system c-myb expression is required for proliferation of immature cells and its down-regulation is required for differentiation. In colonic mucosa c-myb expression occurs at levels comparable to immature hemopoietic cells. Inhibition of c-myb expression in colon cell lines, using anti-sense oligonucleotides, indicates that c-myb expression is required for proliferation. However, the mechanism of c-myb regulation during colon cell differentiation has not been explored. Using the LIM1215 and CaCo-2 colon carcinoma cell lines induced to differentiate with sodium butyrate, we demonstrate that c-myb mRNA is down-regulated as an early event in differentiation by a mechanism involving transcriptional attenuation in intron 1. By analogy with procaryotic and eucaryotic genes, transcriptional attenuation probably occurs in a region containing nineteen consecutive thymidine residues. Computer prediction of the secondary structure of the nascent mRNA chain encoded by this region suggests a strong potential for stem-loop formation. Sequence analysis of several colon tumour cell lines reveals mutations in this region that may disrupt transcriptional attenuation and result in the increased c-myb expression observed in colon tumours and tumour cell lines.

Identification of a second promoter in the human c-myb proto-oncogene.

We have previously described an alternatively spliced cDNA clone of the human c-myb proto-oncogene which has been shown to enhance the differentiation of Friend murine erythroleukemia cells. This clone, pMbm-2, contains unique 5' sequences which replace exon 1. The human c-myb intron 1 was sequenced to determine the exact position of this unique sequence and to further characterize the role of intron 1 in the regulation of the human c-myb gene. Here we report that intron 1 of c-myb is highly conserved between human and mouse throughout the intron, while only those sequences directly adjacent to exons 1 and 2 are conserved between human and chicken. The unique sequence of pMbm-2 was located directly adjacent to exon 2, suggesting that it arose as a product of alternative transcription initiation within intron 1. RNAase protection analysis was used to map a cluster of transcription start sites at the 5' end of exon 2. Levels of messages utilizing these start sites are proportional to those arising from the primary promoter. Functional characterization of this region revealed that this region can function as a promoter. Deletion studies have revealed the presence of negative and positive regulatory elements within this region which are utilized with different efficiencies in different cell lines. These studies suggest that cis or trans factors acting in this region may serve a dual function in both attenuation and transcription initiation.

Multiple mechanisms of regulation of the human c-myb gene during myelomonocytic differentiation.

Alterations in expression of c-myb can have profound effects on the growth and differentiation of hematopoietic cells. Thus, it is important to understand the mechanisms by which c-myb is regulated during hematopoietic cell differentiation. Previous studies pertaining to the regulation of c-myb have been carried out with the avian and murine forms of the gene; the current studies were designed to determine the mechanisms of regulation of the human form of c-myb. Transcriptional analysis by nuclear run-on assays revealed that an attenuation of transcription was the means of primary regulation during retinoic acid- and vitamin D3-induced differentiation of HL-60 cells, while other mechanisms in addition to attenuation were active during dimethyl sulfoxide (DMSO)- and phorbol ester-induced differentiation. Densitometric analysis of the changes in c-myb transcription caused by phorbol ester suggested that the c-myb promoter may be down-regulated during phorbol ester-induced differentiation of HL-60. Additional studies exhibited post-transcriptional regulation by phorbol ester. DMSO was also shown to regulate c-myb at the post-transcriptional level. Interestingly, the post-transcriptional regulation of c-myb by DMSO required continuous transcription. This requirement was shared for c-myc but not ornithine decarboxylase expression. The transcriptional dependency of c-myb post-transcriptional regulation did not equate to translational dependency, thus a novel post-transcriptional regulatory mechanism may control c-myb gene expression. The multiple levels of regulation of c-myb suggest the importance of proper control for hematopoietic cell differentiation.

Alternative splicing of the human c-myb gene.

Two cDNA clones of the human c-myb gene have been isolated from a CCRF-CEM leukemia cell cDNA library and sequenced in their entirety. These sequences, when compared with those previously reported for the human c-myb gene, reveal an alternative splicing process that generates at least four forms of the c-myb message. Three of these forms co-migrate on Northern blots and are co-expressed in several human hematopoietic cell types. Data on sequence comparisons with mouse and chicken homologues of c-myb coupled with oligonucleotide hybridization to genomic clones of the human c-myb gene indicate that this alternative splicing process utilizes three closely spaced splice donor sites and two unique exons present between viral defined exons 5 and 6. In one clone, the alternative splicing would generate a predicted myb protein with a three amino acid deletion in the region involved in transcription activation. In the other clone, incorporation of a new exon leads to introduction of a translation stop codon leading to loss of the entire carboxy terminus of the protein. This includes loss of a portion of the region involved in transcription activation as well as a separate highly conserved domain. The effect of these changes on protein function is currently unknown.

Src can regulate carboxy terminal interactions with AFAP-110, which influence self-association, cell localization and actin filament integrity.

The SH2 and SH3 binding partner AFAP-110 is a tyrosine phosphorylated substrate of Src. AFAP-110 has been hypothesized to link Src to actin filaments, which may contribute to the effects of Src upon actin filament integrity. However, it has been unclear what effect activated Src (Src527F) has upon AFAP-110 structure or function and whether AFAP-110 plays a role in actin filament integrity. We report here that the carboxy terminal 127 amino acids of AFAP-110 are comprised of an alpha-helical region that contains a leucine zipper motif. This indicated the potential of AFAP-110 to self-associate. Expression of the carboxy terminus as a fusion protein (GST-cterm) will permit affinity absorption of cellular AFAP-110. The integrity of the alpha-helical leucine zipper motif in GST-cterm is required for affinity absorption, but binding is not due to a classical leucine zipper interaction. Co-expression of Src527F, unlike cSrc, will abrogate affinity absorption of AFAP-110 with GST-cterm. These data indicate that Src527F has affected a change in the carboxy terminal structure that renders AFAP-110 unavailable for affinity absorption. Superose chromatography demonstrate that AFAP-110 will fractionate as a monomer or multimer, indicating AFAP-110 can be detected in a self-associated form in cell lysates. Co-expression of Src527F resulted in AFAP-110 fractionating with a molecular weight that predicts only a multimeric population. Deletional mutagenesis also indicate a biological role for the carboxy terminus in cellular localization and actin filament integrity. Deletion of the entire carboxy terminal alpha-helix (84 amino acids) will not permit AFAP-110 to efficiently colocalize with actin filaments or the cell membrane. Deletion of only the leucine zipper region of the carboxy terminal alpha-helix (44 amino acids) from AFAP-110 (AFAPAdeltazip) demonstrate that both AFAPdeltalzip and actin filaments are repositioned into rosette-like structures, similar to the effects of Src527F, while co-expression of AFAP-110 with cSrc will not affect actin filaments. These data indicate that AFAP-110 can play an important role in modulating actin filament integrity through carboxy terminal interactions that can be affected by Src527F.

Estrogen receptor expression and function in long-term estrogen-deprived human breast cancer cells.

Hormone-dependent breast cancer responds to primary therapies that block estrogen production or action, but tumor regrowth often occurs 12-18 months later. Additional hormonal treatments that further reduce estrogen synthesis or more effectively block its action cause additional remissions, but the mechanisms responsible for these secondary responses are not well understood. As a working hypothesis, we postulated that primary hormonal therapy induces adaptive changes, resulting in enhanced estrogen receptor (ER) expression and target gene activation and, further, that secondary treatment modalities interfere with these receptor-mediated transcriptional pathways. To test this hypothesis, we used an MCF-7 breast cancer model system involving deprivation of estradiol in culture for a prolonged period. These long-term estradiol-deprived (LTED) cells adapt by acquiring the ability to regrow in the absence of added estradiol. The experimental paradigm involved the comparison of wild-type cells with LTED cells. As endpoints, we directly assessed ER expression at the messenger RNA-, protein-, and ligand-binding levels and ER functionality by quantitating reporter gene activation and expression of endogenous estrogen target gene messenger RNA, as well as ER coactivator levels. Our data demonstrated an adaptive increase in ER expression and in basal ER functionality, as assessed by read-out of three different transfected reporters in LTED, as opposed to wild-type MCF-7 cells. Increased reporter gene read-out was dramatically inhibited by the pure antiestrogen ICI 182,780. As verification that endogenous (as well as transfected) estrogen target genes had enhanced transcription, we found that the basal levels of c-myb and c-myc message were substantially increased in LTED cells and could be inhibited by antiestrogen. Interestingly, the levels of c-myb and c-myc message in the LTED cells seemed to be increased out of proportion to the degree of ER reporter gene activation and were similar to those in wild-type cells maximally stimulated with estradiol. In addition, not all estrogen-responsive genes were activated, because transforming growth factor-alpha message level was not increased in LTED cells. Up-regulation of the steroid receptor coactivator SRC-1 did not seem to mediate the process of enhanced ER-induced transcription. Considering these observations together, we suggest that long-term estradiol deprivation causes adaptive processes that not only involve up-regulation of the ER but also influence the specificity and magnitude of activation of estrogen-responsive genes.

Increased expression and characterization of two distinct folate binding proteins in murine erythroleukemia cells.

We previously identified two membrane-bound folate binding proteins, FBP1 and FBP2, in murine L1210 leukemia cells. We now report on the development of two variant murine erythroleukemia cell lines that were used for direct comparison and biochemical characterization of the two murine folate binding proteins. Based on the results of northern analysis and the mobilities of affinity-labeled proteins on polyacrylamide gels, these cell lines exhibit specific up-regulated expression of FBP1 or FBP2. The affinities of the folate binding proteins for various (anti)folates were determined based upon the ability of the compounds to inhibiting of [3H]folic acid. The two proteins exhibited considerably different affinities and stereospecificities and, in general, FBP2 consistently bound each test compound with lesser affinity than FBP1. Both proteins displayed greatest affinity for folic acid, 5-methyltetrahydrofolate, and the antifolates CB3717 and 5,10-dideazatetrahydrofolate (DDATHF). Conversely, the proteins exhibited poor affinity for the dihydrofolate reductase inhibitors methotrexate and aminopterin. For 5-formyltetrahydrofolate, FBP1 had high affinity for the (6S) diastereoisomer, whereas FBP2 showed preference for the non-physiologic (6R) diasterceoisomer. The binding properties of FBP1 and FBP2 overexpressed in these cell lines closely paralleled those of their respective human homologs. These lines provide a model system in which to examine the biochemical characteristics of the individual folate binding proteins without the potential problems associated with expression of proteins in dissimilar cell lines.

Characterization of a rearrangement in the c-MYB promoter in the acute lymphoblastic leukemia cell line CCRF-CEM.

Despite the frequent description of 6q- structural abnormalities in human leukemias and lymphomas, rearrangements of the c-MYB locus have not been detected. We have detected a rearrangement in the c-MYB proto-oncogene in the cell line CCRF-CEM, an immature T-cell leukemia cell line which is not 6q-. Due to this rearrangement, a large portion of the c-MYB promoter conserved between the human and murine c-MYB genes is lost. The rearranged locus, which we have designated MRR (MYB rearranged region), has been cloned and mapped to chromosome 6. Field inversion gel electrophoresis (FIGE) studies reveal that the MRR sequence is linked to the c-MYB locus, suggesting that the rearrangement is due to a submicroscopic deletion. The rearrangement appears to have no effect on c-MYB promoter activity as analyzed in CCRF-CEM cells. The normal locus of the MRR sequence has been cloned from a human placental genomic library. Partial sequence analysis of this clone reveals that a portion of the DNA lost in the rearrangement shows a high degree of homology to a member of the myc family of oncogenes. Thus the characterization of this rearrangement has yielded a new set of probes for the study of chromosome 6q abnormalities in human leukemias and lymphomas and provides the first evidence for potential involvement of the c-MYB locus itself in submicroscopic deletions within chromosome 6.

Use of phorbol-12,13-dibutyrate as a mitogen in the cytogenetic analysis of tumors with low mitotic indexes.

Solid tumors, particularly those involving the colon, breast, and lung, are the most common tumors in humans. However, many technical difficulties exist in obtaining analyzable chromosomes from these tumors, including the inability to stimulate cell division. Phorbol-12,13-dibutyrate (PDBu) is a tumor promoter that activates a variety of cellular responses, including proliferation. Using flow cytometry, we have demonstrated that PDBu acts as a mitogen in primary cultures of colon tumor cells. Based on these results, we developed a short-term culture technique that greatly improves the yield of analyzable metaphases from colon tumors. Stimulated cultures consistently contained at least ten times more metaphases than unstimulated cultures, and chromosome morphology was improved. By modifying this technique with the addition of the calcium ionophore A23187, we have successfully obtained analyzable chromosomes from the peripheral blood of normal individuals, chronic lymphocytic leukemia patients, and a nodular small cell lymphoma patient. These results demonstrate that mitogenic stimulation by PDBu is a valuable technique in the cytogenetic analysis of colon tumors. By using PDBu alone or in combination with other agents, this technique may also be applicable to many other tumors that are difficult to karyotype because of an inability to obtain mitoses.

Cytogenetic and molecular genetic studies of a patient with atypical lymphoid hyperplasia.

We have karyotyped cells from a lymph node of a patient with atypical lymphoid hyperplasia. Among other clonal chromosomal abnormalities, a t(2;19) translocation was observed with breakpoints at 2p11.2 and 19q13. The genes for transforming growth factor alpha and beta have been mapped to 2p11-p13 and 19q13, respectively, but Southern blot analysis did not reveal any alteration in the structure of these genes. Similarly, the kappa immunoglobulin gene, which maps to 2p11-p12 was not rearranged. In addition, Southern blot analysis using immunoglobulin and T-cell receptor genes as probes, did not demonstrate any clonality of either B or T cells. We propose that this patient represents an early, polyclonal stage of atypical hyperplasia. The chromosome changes observed may have been one of the etiologic factors causing this disorder.

Establishing paternity using minisatellite DNA probes when the putative father is unavailable for testing.

A paternity case involving a putative father who had died a few years earlier in an automobile accident was referred to the laboratory for testing. The child and his mother, the deceased's parents, and nine of the deceased's siblings were available for analysis. As previously reported, paternity testing using red blood cell groups, human leukocyte antigens (HLA), red blood cell enzymes, serum proteins, and immunoglobulin allotypes gave a cumulative paternity index of 43,300 and a combined probability of paternity equal to 99.998%. RFLP analysis using Hinf I and Sau 3A single digests and the minisatellite deoxyribonucleic acid (DNA) probes 15.1.11.4 and 6.3 showed no exclusion of paternity and gave nearly conclusive evidence that the putative father was the biological father of the child.

T-small cell disorders.

A minority (less than 2%) of all lymphoproliferative disorders are derived from small T cells. These include T-cell prolymphocytic leukemia, T-cell granular lymphocytic leukemia, and mycosis fungoides/Sézary syndrome. With the possible exception of early-stage, skin-localized mycosis fungoides, all are considered incurable, although palliation can be achieved with radiation therapy, chemotherapy, biologic therapy, and combinations of these modalities. Of these disorders, mycosis fungoides is the most common; it follows an indolent, though gradually progressive, course that spans years. The T-cell prolymphocytic leukemias, in contrast, are generally refractory to treatment, with a median survival of typically less than 1 year. Although effective therapy remains elusive in most cases, the development of nucleosides as a class of chemotherapeutic agents and biologics, including interferon, monoclonal antibodies, and vitamin A derivatives, offers new hope for at least more effective palliation of these progressive lymphoproliferative disorders. However, rapid improvement in the treatment of these disorders remains hampered by the rarity of these individual entities. More rapid progress in treatment depends on national and international cooperation to accrue patients for definitive trials of sufficient size to evaluate new treatment options quickly.

Somatic rearrangement of the c-myc oncogene in primary human diffuse large-cell lymphoma.

Chromosome translocations involving 8q24, the band to which c-myc has been mapped (Dalla-Favera et al., 1982), are a uniform finding in Burkitt's lymphoma (Bernheim et al., 1981). However, in only a minority of the tumors is the rearrangement of the c-myc locus sufficiently close to the gene to be detected with currently available probes (Dalla-Favera et al., 1983). Approximately 25% of diffuse large-cell lymphomas have also been reported to have translocations involving 8q24 (Mitelman, 1985), but there have been no reports of c-myc rearrangements in this form of non-Hodgkin's lymphoma. We have examined the structure of the c-myc locus in primary tumor tissue of 10 cases of diffuse large-cell lymphoma. In one patient, Southern blot analysis revealed additional c-myc fragments in the tumor DNA but not in the germ-line DNA. Southern blot analysis using probes from both the heavy- and light-chain immunoglobin loci showed that the myc rearrangement was unlikely to involve the immunoglobulin loci in this patient.

Altered expression of cell cycle related genes including c-myb in a subclone of HL-60 that exhibits reversible differentiation.

A differentiation resistant subclone of HL-60, DMSOr, was removed from the selective pressure of dimethyl sulfoxide and characterized with a new stable phenotype of reversible differentiation. DMSOr cells, when treated with 1.25% dimethyl sulfoxide, differentiated in a manner similar to the parental HL-60 with respect to morphological changes, increase in superoxide production, and withdrawal from cell cycle. Upon removal of the dimethyl sulfoxide at points normally associated with commitment to terminal differentiation, DMSOr reverted to the immature phenotype. This demonstrates an uncoupling of the morphological, functional, and antiproliferative effects of differentiation from commitment to terminal differentiation. Associated with the reversible phenotype of DMSOr was an altered expression of the c-myb oncogene. In HL-60, c-myb expression was down-regulated by 72 h and completely diminished by 144 h. Northern blot analysis of DMSOr demonstrated greater levels of expression of c-myb at 72 and 144 h. Similar results were shown with histone H4, cdc2 kinase, and, to a lesser extent, ornithine decarboxylase. The c-myb related gene B-myb did not show altered regulation during differentiation. The results suggest that altered expression of genes that control cell cycle may be critical for the reversible phenotype of DMSOr.

Malignant transformation of NIH 3T3 fibroblasts by human c-sis is dependent upon the level of oncogene expression.

High-level expression of the c-sis oncogene, which encodes the beta chain of platelet-derived growth factor, transforms immortalized rodent fibroblasts in vitro to a malignant phenotype. c-sis gene expression has been demonstrated in a variety of human tumors, although generally at levels much lower than those shown to transform cells in vitro. We examined the effect of lower levels of c-sis expression on the phenotype of NIH 3T3 fibroblasts. Clones with various levels of c-sis expression were generated by transfecting NIH 3T3 cells with a plasmid that expressed the human c-sis cDNA and the TN5 neomycin-resistance gene. G418-resistant clones, which expressed the c-sis cDNA, were selected and characterized. Alterations in the phenotype of the clones that expressed c-sis ranged from increased growth in soft agar to malignant tumor formation in nude and syngeneic mice. Increased levels of c-sis cDNA expression correlated with the acquisition of features of transformation in a dose-dependent manner and altered the cellular phenotype in a manner consistent with the progression of cells towards malignancy. These data support a model in which low levels of sis gene expression in tumors contribute to the acquisition of some features of transformation but require complementation by other genes or factors to produce a fully malignant phenotype.