SUMMARY. The killing by human polymorphonuclear leukocytes of several species of bacteria, some of which were catalase positive, was examined in vitro in aerobic and anaerobic conditions. When all conditions other than the oxygen tension were identical, killing after 30 min was slightly greater in aerobic than in anaerobic conditions. However, after 60 and 120 min the difference between aerobic and anaerobic liilling was smaller, and killing was nearly complete for all strains tested. These results conflict with the common opinion that oxygen is essential for efficient killing. Minor differences in experimental conditions can greatly influence results, and may be responsible for the discrepancy between this study and some previous studies on this subject.

SUMMARY. The induction of chemotactic activity of polymorphonuclear leukocytes (PMNL) by anaerobic and aerobic bacteria alone or in combination was evaluated. Washed cells as well as the supernate of Proteus mirabilis were chemotactic for leukocytes. The supernate of cultures of two strains of Bacteroides fragilis contained small amounts of chemotactic factors. No chemotactic factors were released from the non-fragilis Bacteroides strains. The supernates of cultures of anaerobic bacteria were capable of inhibiting chemotaxis of leukocytes to the chemotactic factors of P . mirabilis. P . mirabilis and two strains of B. fragilis generated chemotactic factors in serum but none of the other Bacteroides spp. tested were able to induce serum chemotactic factors.

Summary. The virulence of Bacteroides fragilis and B. vulgatus for mice was compared in a skin-infection model. These strains were also tested for pathogenic synergy in mixed infections with Escherichia coli. Strains of B. fragilis were generally more virulent than strains of B. vulgatus and, with one exception, the effect of Bacteroides strains in mixed infections merely reflected their inherent virulence.

In this article we review our researches into the pathogenesis of mixed infections. These may conveniently be divided into in vitro and in vivo studies. In vitro we confirmed that interference with the killing of aerobes by polymorphonuclear leucocytes (PMN's) is a property of the Bacteroides strains tested and appears to depend on competition for opsonins i.e. complement factors. Further studies are in progress to define which complement factors and which bacterial structures are involved. The influence of B. fragilis on chemotaxis has also been studied. Our preliminary data suggest that B. fragilis is itself poorly chemotactic and reduces the chemoattractivity of Proteus mirabilis. This observation is surprising when we consider that abscess formation is the hall-mark of B. fragilis infections and needs clarification. In vivo we have developed a skin infection model in mice which is economical and gives reproducible and quantitative results. In this model we have demonstrated pathogenic synergy between Escherichia coli and B. fragilis. Further studies are planned to assess the role of complement and bacterial factors in this in vivo synergy.

The virulence for mice of Bacteroides fragilis is higher than that of Bacteroides vulgatus. This enhanced virulence is related to the higher growth rate of B. fragilis in broth and to the ability of B. fragilis strains to grow in serum. The failure of B. vulgatus to grow in serum could be overcome by addition of haemin and an unidentified growth factor, present in tryptone. Haemoglobin, lysed erythrocytes or iron in combination with protoporphyrin IX could replace haemin.