SummaryMost clinical isolates of Streptococcus pneumoniae consist of heterogeneous populations of at least two colony phenotypes, opaque and transparent, selected for in the bloodstream and nasopharynx, respectively. Microarray analysis revealed 24 orfs that demonstrated differences in expression greater than twofold between variants of independent strains. Twenty-one of these showed increased expression in the transparent variants, including 11 predicted to be involved in sugar metabolism. A single genomic region contains seven of these loci including the gene that encodes the neuraminidase, NanA. In contrast to previous studies, there was no contribution of NanA to adherence of S. pneumoniae to epithelial cells or colonization in an animal model. However, we observed NanA-dependent desialylation of human airway components that bind to the organism and may mediate bacterial clearance. Targets of desialylation included human lactoferrin, secretory component, and IgA2 that were shown to be present on the surface of the pneumococcus in vivo during pneumococcal pneumonia. The efficiency of desialylation was increased in the transparent variants and enhanced for host proteins binding to the surface of S. pneumoniae .Because deglycosylation affects the function of many host proteins, NanA may contribute to a proteaseindependent mechanism to modify bound targets and facilitate enhanced survival of the bacterium.

IgA, the major class of Ig in secretions, classically functions by interfering with microbial attachment to host tissues. Many mucosal pathogens, including Streptococcus pneumoniae, express an IgA1 protease that may circumvent the protective effects of this Ig subclass. Because these proteases are specific for human IgA1, we generated human mAbs to the major surface antigen of the pneumococcus, its capsular polysaccharide, and tested their effect in a colonization model of bacterial adherence to respiratory epithelial cells in culture. Rather than inhibiting adherence, typespecific IgA1 markedly enhanced bacterial attachment to host cells, but only when cleaved by IgA1 protease. Neither antibodies of protease-insensitive subclasses (IgA2 and IgG) nor those directed against heterologous capsules had such activity. The adherencepromoting properties of cleaved antibodies correlated with the cationic characteristics of their variable segments, suggesting that bound Fab fragments may neutralize the inhibitory effect of negatively charged capsules on adhesive interaction with host cells. Coating of pneumococci with anticapsular polysaccharide antibody unmasked the bacterial phosphorylcholine ligand, allowing for increased adherence mediated by binding to the platelet activating factor receptor on epithelial cells. In addition, our findings provide evidence for a novel function of bacterial IgA1 proteases. These enzymes may enable pathogens to subvert the antigen specificity of the humoral immune response to facilitate adhesive interactions and persistence on the mucosal surface.

Although dogs have been proposed as carriers of extraintestinal pathogenic Escherichia coli (ExPEC) with infectious potential for humans, presumed host species-specific differences between canine and human ExPEC strains have cast doubt on this hypothesis. The recent discovery that allele III of papG (the P fimbrial adhesin gene) predominates among human cystitis isolates and confers an adherence phenotype resembling that of canine ExPEC prompted the present reevaluation of the canine-human ExPEC connection. Sixteen paired pappositive urine and rectal E. coli isolates from dogs with urinary tract infection were studied. papG (adhesin) and papA (pilin) allele type, agglutination phenotypes, virulence factor genotypes, and randomly amplified polymorphic DNA and pulsed-field gel electrophoresis fingerprints were analyzed and compared with those of human ExPEC controls. The 16 canine strains contained predominantly papG allele III. Agglutination phenotypes segregated strictly according to papG allele status and were homogeneous among strains with the same papG allele profile irrespective of their human versus canine origin. Canine and human PapG variant III peptide sequences were highly homologous, without host species-specific differences. The most prevalent canine papA allele was F48, a novel variant recently identified among human urosepsis isolates. In addition to pap, human ExPEC-associated virulence genes detected among the canine strains included sfa/focDE, sfaS, fyuA, hlyA, cnf1, cdtB, kpsMT-II and -III, rfc, traT, ompT, and a marker for a pathogenicity-associated island from archetypal human ExPEC strain CFT073. Molecular fingerprinting confirmed the fecal origin of all but one canine urine isolate and showed one pair of O6 canine urine and fecal isolates to be extremely similar to an O6 human urosepsis isolate with which they shared all other genotypic and phenotypic characteristics analyzed. These data demonstrate that canine ExPEC strains are similar to, and in some instances essentially indistinguishable from, human ExPEC strains, which implicates dogs and their feces as potential reservoirs of E. coli with infectious potential for humans.