The antitumor effect of two strains of Propionibacterium acnes (PAI and PAII) and chemically derived fractions from the whole bacterial cell were studied using a murine ovarian teratocarcinoma (MOT) model. When injected i.p. in high doses (700 to 1400 micrograms/mouse), both strains produce survival of a significant proportion of tumor-bearing mice (30 to 90%). On a weight to weight basis, however, PAI was significantly more effective than PAII. PAI and PAII were extracted using pyridine, which yielded four fractions, i.e., pyridine-extracted strains PAI and PAII (PA-PEI and PA-PEII, respectively) which are composed of the cell wall material extracted by pyridine, and the residues of PA-PEI and PA-PEII (PA-RI and PA-RII, respectively) which are composed of the residue material following the chemical extraction. The chemical composition of PA-PEI was different from that of PA-PEII (the latter had proportionately three times as many carbohydrates and one-third of the protein content of the former) and so were their antitumor properties in the MOT model. PA-PEI had markedly reduced antitumor effect when compared to the untreated cell on a per weight basis. Furthermore, curability was only seen when using a high dose (1400 micrograms/mouse). By contrast, the cell wall components extracted by pyridine from PAII (PA-PEII) had powerful antitumor effects, i.e., greater than 50% of mice given 1400-micrograms injections survived. The material contained in PA-PEII was further fractionated on the basis of its organic solubility in chloroform:methanol solvent. The water-soluble and solvent-insoluble fractions retained most of the antitumor effects of PA-PEII, while the water-insoluble and solvent-soluble fractions were only moderately effective, suggesting that the active moiety(ies) was associated with the nonlipid components of this fraction. Both residue fractions (PA-RI and PA-RII) were as effective on a per weight basis in controlling the growth of 10(5) tumor inoculum as were whole untreated cells. However, periodate oxidation of PA-RI resulted in complete loss of its antitumor effects. When surviving mice that had no evidence of tumor persistence following a tumor challenge (10(5) MOT cells) and i.p. treatment with PA were subsequently rechallenged with 10(4) tumor cells, survival was significantly prolonged, as compared to tumor-challenged (10(4) MOT) naive mice. In addition, 10 to 20% of these rechallenged mice had complete eradication of the tumor inoculum (no evidence of disease for greater than 120 days).(ABSTRACT TRUNCATED AT 400 WORDS)