Positive selection in penicillin-binding proteins 1a, 2b, and 2x from Streptococcus pneumoniae and its correlation with amoxicillin resistance development

The efficacy of β-lactam antibiotics in Streptococcus pneumoniae has been compromised because of the development of altered penicillin-binding proteins (PBPs), however, this has been less so for amoxicillin than for penicillin. Recently, there have been a number of important methods developed to detect molecular adaptation in protein coding genes. The purpose of this study is to employ modern molecular selection approaches to predict sites under positive selection pressure in PBPs, derived from a large international S. pneumoniae collection of amoxicillin resistant and susceptible isolates, and encompassing a comparative data set of 354 pbp1a, 335 pbp2b, and 389 pbp2x gene sequences. A correspondence discriminant analysis (CDA) of positively selected pbp sites and amoxicillin MIC (minimum inhibitory concentration) values is then used to detect sites under positive selection pressure that are important in discriminating different amoxicillin MICs. Molecular adaptation was evident throughout PBP2X, with numerous positively selected sites in both the transpeptidase (TP) and C-terminal domains, strongly correlated with discriminating amoxicillin MICs. In the case of PBP1A positive selection was present in the glycosyltransfer (GT), TP and C-terminal domains. Sites within the TP domain tended to be correlated with the discrimination of low from intermediate MICs, whereas sites within the C-terminal tail, with a discrimination of intermediate from fully resistant. Most of the positively selected sites within PBP2B were in the N-terminal domain and were not correlated with amoxicillin MICs, however, several sites taken from the literature for the TP domain were strongly associated with discriminating high from intermediate level amoxicillin resistance. Many of the positively selected sites could be directly associated with functional inferences based on the crystal structures of these proteins. Our results suggest that clinical emphasis on TP domain sequences of these proteins may result in missing information relevant to antibiotic resistance development.