Recently, Novak et al. (1998, Mol Microbiol 29: 1285±1296) reported their investigation on the phenomenon of penicillin tolerance in Streptococcus pneumoniae. A library of mutants in pneumococcal surface proteins was screened for the ability to survive in the presence of 10 ́ the minimum inhibitory concentration of antibiotic. A mutant harbouring an insertion in the known gene psaA was isolated among 10 candidate tolerance mutants. Inactivation of psaA was previously shown to result in reduced virulence of S. pneumoniae (as judged by intranasal or intraperitoneal challenge of mice) and in reduced adherence to A549 cells (type II pneumocytes), leading to the suggestion that PsaA was an adhesin (Berry and Paton, 1996, Infect Immun 64: 5255±5262). This gene is part of the psa locus (Fig. 1) that encodes an ATP-binding cassette (ABC) permease belonging to cluster 9, a family of ABC metal permeases (Dintilhac et al., 1997, Mol Microbiol 25: 727±740). Novak et al. (1998, Mol Microbiol 29: 1285±1296) reported that psa mutants displayed pleiotropic phenotypes: (i) reduced sensitivity to the lytic and killing effects of penicillin; (ii) growth in chains of 40±50 (psaC ) to 200±300 (psaD ) cells; (iii) autolysis defect and loss of sensitivity to low concentrations of deoxycholate (DOC), a species characteristic trait; (iv) absence of LytA, the major autolytic amidase; (v) almost complete loss of choline-binding proteins (ChBPs) (psaC and psaD ) and absence of CbpA; (vi) loss of transformability (except psaA); and (vii) manganese (Mn) requirement for growth in a chemically de®ned medium. Because penicillin tolerance was ®rst associated with an autolysis defect (Tomasz et al., 1970, Nature 227: 138± 140), the absence of LytA (phenotype iv) could itself explain phenotypes i and iii. Dysregulation of lytA could not be investigated because, according to Novak et al. (1998, Mol Microbiol 29: 1285±1296), the dif®culty in lysing psa mutant cells prohibited Northern analysis, although lysates of the psa mutants could be obtained for immunoblot analysis of LytA and of RecA and for Southern con®rmation of the psa mutations. Nevertheless, because expression of the lytA gene has been shown to be driven by three different promoters, including Pb which is the recA basal promoter (Mortier-BarrieÁre et al., 1998, Mol Microbiol 27: 159±170), and because wild-type levels of RecA were detected in the psa mutants (Novak et al., 1998, Mol Microbiol 29: 1285±1296), it seems dif®cult to account for the complete absence of LytA on the basis of altered expression. On the other hand, phenotypes i±iv are reminiscent of alterations observed after the replacement of choline (Ch) by ethanolamine (EA) in the cell wall of pneumococcus (Tomasz, 1968, Proc Natl Acad Sci USA 59: 86±93). Similar phenotypes were also displayed by Ch-independent mutants of S. pneumoniae (Severin et al., 1997, Microb Drug Res 3: 391±400; Yother et al., 1998, J Bacteriol 180: 2093±2101). S. pneumoniae has a nutritional requirement for Ch that is incorporated by covalent bonds into the cell wall teichoic acids (TA) and in the membrane-bound lipoteichoic acid (LTA). Ch residues bound to TA (ChTA) were shown to be absolutely required for LytA activity (Holtje and Tomasz, 1975; J Biol Chem 250: 6072±6076). The action of LytA has long been thought to be restricted to pneumococcal cell walls because of this requirement. However, recent reports suggest that ChTA is required Molecular Microbiology (1999) 32(4), 881±891