Bacterial IgA1 proteases share the ability to cleave human IgA1 at the hinge region. Nature has developed this trait along at least five independent evolutionary lineages. To obtain further insight into the phylogeny and function of IgA1 proteases, the nucleotide sequence of the iga gene that encodes the IgA1 protease was determined from two Streptococcus mitis strains and one Gemella haemolysans strain. Heterologous expression in Escherichia coli confirmed that the genes encode human IgA1-cleaving activity. IgA1 proteases from Streptococcus and G. haemolysans shared structural features, including a motif typical for zinc-dependent metalloproteases of clan MA(E) family M26 and an N-terminal signal sequence followed by an LPXTG cell-wall-anchor motif and two putative membrane-spanning domains. In addition, they all harboured a repeat region preceding the active site of the protease. In the streptococcal IgA1 proteases, a G5 domain, which has been suggested to bind N-acetylglucosamine, was identified. Conservation of these structures in otherwise diverse proteases suggests that they are essential to the biological function of the enzyme. The phylogenetic distribution of homologous iga genes and conservation of gene order in the iga gene region in different Streptococcus species, combined with the sequence homologies, strongly suggest that the iga gene is more ancient in Streptococcus than in G. haemolysans, and therefore that the IgA1 protease gene was transferred from Streptococcus to G. haemolysans.