A comparative genome analysis identifies distinct sorting pathways in gram-positive bacteria

Abstract

Surface proteins in gram-positive bacteria are frequently required for virulence, and many are attached to the cell wall by sortase enzymes. Bacteria frequently encode more than one sortase enzyme and an even larger number of potential sortase substrates that possess an LPXTG-type cell wall sorting signal. In order to elucidate the sorting pathways present in gram-positive bacteria, we performed a comparative analysis of 72 sequenced microbial genomes. We show that sortase enzymes can be partitioned into five distinct subfamilies based upon their primary sequences and that most of their substrates can be predicted by making a few conservative assumptions. Most bacteria encode sortases from two or more subfamilies, which are predicted to function nonredundantly in sorting proteins to the cell surface. Only approximately 20% of sortase-related proteins are most closely related to the well-characterized Staphylococcus aureus SrtA protein, but nonetheless, these proteins are responsible for anchoring the majority of surface proteins in gram-positive bacteria. In contrast, most sortase-like proteins are predicted to play a more specialized role, with each anchoring far fewer proteins that contain unusual sequence motifs. The functional sortase-substrate linkage predictions are available online (http://www.doe-mbi.ucla.edu/Services/Sortase/) in a searchable database.

FIG. 1.

(A) Diagram illustrating a CWS-containing protein. It is composed of an N-terminal signal peptide and a C-terminal sorting signal that has a conserved LPXTG motif followed by a hydrophobic stretch of amino acids and positively charged residues at the C terminus. (B) A flowchart illustrating how functional linkages between sortase homologs and CWS-containing proteins were established. Three methods were used to make “simple” predictions: (i) single sortase, genome only contains a single sortase; (ii) single cluster, genome contains a gene cluster with a single sortase and a single CWS-containing protein; (iii) SrtA-SrtB, genome contains only SrtA and SrtB homologs. Two additional methods extended the initial predictions. With substrate sequence homology, a new functional link was made when a CWS-containing protein in one genome shared significant sequence homology (≥30%) to a previously assigned CWS-containing protein in another genome and both organisms encoded closely related sortases (see the text). With unique sorting signals, additional CWS-containing proteins were linked to their cognate sortase by examination of their sorting signal motif (subfamily-4 and subfamily-5 enzymes process the motifs LPXTA and LAXTG, respectively).

FIG. 2.

Sorting signals categorized by subfamily type. The figure shows the position-specific frequency of amino acids within the sorting signals of different types of sortases. The one-letter symbol for the amino acid residue is given for each position in the six-residue motif. The font size of each letter is proportional to the frequency with which an amino acid occurs. If an amino acid appears in fewer than 8% of the substrates, then the letter does not appear in the figure. When one type of amino acid is completely conserved at a particular position of the sorting signal motif or when one type of amino acid occurs in more than 92% of the CWS-containing proteins, then only one letter is present in a position. When no amino acid type is predominant in a given position of the motif, then the amino acid types found in the motif are given in brackets.

FIG. 3.

(A) Pie chart showing the distribution of sortase homologs in gram-positive bacteria. A total of 176 sortase homologs were identified in gram-positive bacteria: 42 SrtA sortases, 17 SrtB sortases, 54 subfamily-3 sortases, 13 subfamily-4 sortases, and 14 subfamily-5 sortases. (B) Pie chart showing the fraction of CWS-containing proteins that are anchored by different types of sortases. A total of 203 CWS-containing proteins were not assigned to a specific sortase. However, several lines of evidence suggest that these remaining proteins are processed by members of the SrtA subfamily (94 of the remaining 203). First, several of the unassigned substrates contain the N-terminal motif YSIRK (5). Second, they frequently contain the sequence LPXTG followed by an acidic amino acid, which appears to be an expanded recognition motif for SrtA-type enzymes (Fig. 2). Finally, most are not genomically adjacent to another sortase, similar to nearly all SrtA substrates.