Previous studies have indicated that an autostimulatory transforming growth factor was required for the optimal growth of SW-13 adrenal carcinoma cells in soft agar. The production of SW-13 colony-stimulating activity by other human malignant cell lines of both epithelial and mesenchymal origin has been demonstrated. Evidence was presented indicating that the stimulating activity detected in crude acid-ethanol extracts was an acid- and heat-stable polypeptide requiring disulfide bonds for full activity. This activity was detected more frequently in tumors and human cancer cells in culture of epithelial origin than of mesenchymal origin and in a variety of nonneoplastic tissues. In the present study, this activity, termed epithelial transforming growth factor (TGFe) because of its ability to stimulate soft agar growth of certain epithelial cells, was partially purified from bovine kidney. Fourfold purification of the kidney acid-ethanol extract with 50% maximal growth-stimulatory activity of 10 micrograms was achieved using molecular sieve chromatography where TGFe eluted with an apparent molecular weight of 20,000-25,000. The next purification step, molecular sieve high performance liquid chromatography, yielded a 50% maximal growth-stimulatory activity of 50 ng and an 800-fold purification from the initial acid-ethanol extract. TGFe eluted in the Mr 11,000 range. Reversed phase high performance liquid chromatography with a C18 column was then used, yielding a single or double peak of SW-13 colony-stimulating activity at 30-35% acetonitrile. The degree of purification was 11,000-fold with a 50% maximal growth-stimulatory activity of 3.5 ng. Analysis of the peak on 12.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed a major and sometimes single band with a molecular weight of 23,000-25,000. Extraction of protein from the polyacrylamide gel demonstrated that only the Mr 23,000-25,000 band stimulated soft agar growth of SW-13 cells. The biological activity of the partially purified TGFe was found to differ from other known growth factors with regard to its ability to stimulate soft agar growth of SW-13 cells with the exception of basic fibroblast growth factor (FGF). The acid lability of FGF, the different molecular weights of these two growth factors, the lack of stimulation of soft agar growth of A431 cells, and the lack of binding of TGFe to FGF receptors indicated that TGFe was not related to basic FGF. Partially purified TGFe was also found to stimulate soft agar growth of two squamous cell carcinoma lines, A431 and D562, and the mouse embryo-derived AKR-2B cells.(ABSTRACT TRUNCATED AT 400 WORDS)