A B S T R A C T The effect of penicillin treatment of Streptococcus sanguis in vitro, on subsequent bacterial density in the bloodstream and on cardiac valves in the rabbit model of endocarditis was studied. As experimental tools for this study, isogenic pairs of S. sanguis differing in resistance to streptomycin or rifampin were prepared by genetic transformation. Rabbits with traumatized heart valves received an intravenous inoculation of penicillin treated (1 ,g/ml) and untreated S. sanguis, each marked by resistance to either streptomycin or rifampin. The number of penicillin-treated and untreated bacteria attached to the valvular surfaces was determined by differential counting on streptomycin or rifampin containing media. Penicillin pretreatment reduced cardiac valve colonization 5 min after inoculation ("adherence ratio" X 108 was 4.11 for the control and 3.66 for the penicillin-treated bacteria, P < 0.001). The results were not due to differences in serum killing or bacterial densities in the bloodstream. There was no difference in valvular bacterial densities 24 h after bacterial inoculation (adherence ratio X 108, 7.26 untreated vs. 6.34 penicillin-pretreated, P > 0.10).In vitro experiments were performed using plateletfibrin surfaces to test the possibility that penicillininduced loss of lipoteichoic acid was responsible for decreased streptococcal adherence. Pretreatment of S. sanguis cultures with inhibitory concentrations of penicillin or with antiserum against lipoteichoic acid and precoating of the platelet-fibrin surfaces with lipoteichoic acid, all caused reduction in bacterial adherence.The findings are interpreted as support for the role of lipoteichoic acid as an adhesin in S. sanguis interactions with particular host tissue surfaces.

The response of tolerant Streptococcus sanguis and nontolerant Streptococcus mitis infections to penicillin therapy was compared in the rabbit model of endocarditis. The minimal inhibitory and bactericidal concentrations of penicillin were 0.1 and 0.1 μg/ml, respectively, for S. mitis and 0.05 and 6.2 μg/ml, respectively, for S. sanguis . Time-kill studies done in vitro with penicillin concentrations of 2 and 20 μg/ml demonstrated minimal killing of the tolerant strain, with a 3 log difference in survival between the two strains after 24 and 48 h. Both strains produced endocarditis with comparable bacterial densities on the valvular vegetations. Rabbits were treated with procaine penicillin G in two dosage regimens, 80,000 or 5,000 U/kg given every 8 h. There was no difference between bacterial densities in valvular vegetations removed from rabbits infected with either strain after 2, 4, or 6 days of treatment with the high-dose regimen (serum penicillin concentration at 0.5 h, 9.4 μg/ml), despite the fact that serum bactericidal activity against the tolerant strain at 0.5 h was minimal. With the low-dose penicillin regimen (serum concentration at 0.5 h, 2.5 μg/ml), therapy was significantly less effective in the tolerant group only after 6 days of treatment. Similar results were obtained when penicillin was administered in low and high doses to prevent infection. In this animal model of infection, penicillin tolerance was associated with a diminished response to penicillin therapy only when the dose was severely restricted. In the high-dose regimen, there was no difference in the response to penicillin therapy between animals infected with either strain, despite the presence of only minimal serum bactericidal activity in the rabbits infected with the tolerant strain.