Archaeosortases and exosortases are widely distributed systems linking membrane transit with posttranslational modification

Abstract

Multiple new prokaryotic C-terminal protein-sorting signals were found that reprise the tripartite architecture shared by LPXTG and PEP-CTERM: motif, TM helix, basic cluster. Defining hidden Markov models were constructed for all. PGF-CTERM occurs in 29 archaeal species, some of which have more than 50 proteins that share the domain. PGF-CTERM proteins include the major cell surface protein in Halobacterium, a glycoprotein with a partially characterized diphytanylglyceryl phosphate linkage near its C terminus. Comparative genomics identifies a distant exosortase homolog, designated archaeosortase A (ArtA), as the likely protein-processing enzyme for PGF-CTERM. Proteomics suggests that the PGF-CTERM region is removed. Additional systems include VPXXXP-CTERM/archeaosortase B in two of the same archaea and PEF-CTERM/archaeosortase C in four others. Bacterial exosortases often fall into subfamilies that partner with very different cohorts of extracellular polymeric substance biosynthesis proteins; several species have multiple systems. Variant systems include the VPDSG-CTERM/exosortase C system unique to certain members of the phylum Verrucomicrobia, VPLPA-CTERM/exosortase D in several alpha- and deltaproteobacterial species, and a dedicated (single-target) VPEID-CTERM/exosortase E system in alphaproteobacteria. Exosortase-related families XrtF in the class Flavobacteria and XrtG in Gram-positive bacteria mark distinctive conserved gene neighborhoods. A picture emerges of an ancient and now well-differentiated superfamily of deeply membrane-embedded protein-processing enzymes. Their target proteins are destined to transit cellular membranes during their biosynthesis, during which most undergo additional posttranslational modifications such as glycosylation.

Fig 1

Sequence logos for C-terminal protein-sorting domains associated with exosortase/archaeosortase family proteins. The putative sorting signals shown are as follows: panel A, PGF-CTERM (TIGR04126), cognate sequence for archaeosortase A; panel B, VPXXXP-CTERM (TIGR04143), cognate sequence for archaeosortase B; panel C, PEF-CTERM (TIGR03024), cognate sequence for archaeosortase C; panel D, VPDSG-CTERM (TIGR04151), cognate sequence for exosortase C; panel E, VPLPA-CTERM (TIGR04152), cognate sequence for exosortase D; panel F, VPEID-CTERM (TIGR04161), cognate sequence for exosortase E/CAAX prenyl protease; panel G, Firmcu-CTERM domain (TIGR04145), candidate cognate sequence for exosortase G.

Fig 2

Gene neighborhoods of archaeosortases. Genes encoding members of the archaeosortase/exosortase family are shown in red. Genes for their putative substrates with the corresponding C-terminal TM domains are shown in purple. (A) Tandem organization of the major cell surface glycoprotein of H. marismortui with its putative processing enzyme, archaeosortase A. A similar gene neighborhood, with a GTP-binding protein (yellow), a conserved hypothetical gene (black), a glyoxylase homolog, and artA, occurs in both Halomicrobium mukohataei DSM 12286 (with the S-layer protein gene adjacent to artA) and Haloquadratum walsbyi DSM 16790. (B) Three tandem genes from M. mahii DSM 5219 code for two VPXXP-CTERM proteins and their putative cognate sorting enzyme, archaeosortase B. An artB gene and a VPXXP-CTERM gene are also in tandem in M. evestigatum Z-7303. (C) The artC gene of M. mazei Go1 sits between a nine-gene predicted exopolysaccharide locus and a pair of PEF-CTERM-containing putative targets. (D) The invariant gene neighborhood of PIP-CTERM/archaeosortase D systems consists of the lone putative target, a very small protein with an archaeal type III signal peptide, and artD. Several surrounding uncharacterized proteins appear also to be part of the conserved gene neighborhood.

Fig 3

Peptides observed by mass spectrometry analysis. For M. hungatei protein YP_503687 (top) and M. barkeri protein YP_305280 (bottom), peptides assigned by mass spectrometry are shown as mapped to the respective full-length precursor sequences. For YP_503687, the 809 observed peptides cover nearly every residue between the N-terminal signal peptide and the final 70 residues. M. barkeri protein YP_305280, shown in the lower picture, had 104 observed peptides, although none overlapping the final 89 residues. An alignment of the respective C-terminal regions lacking peptide coverage is shown between the two proteomic coverage graphics. The threonine-rich regions are shown in olive, with Thr (T) residues in black boldface. The PGF-CTERM regions are in red.

Fig 4

Model of archaeosortase A processing of the halobacterial major cell surface glycoprotein. Panel A shows the archaeosortase (green) as a multiple-membrane-spanning protein with the presumed active-site Cys oriented toward the extracellular face. The putative target protein is shown with its signal peptide already removed and with the helix of the C-terminal PGF-CTERM domain also spanning the membrane (orange). Panel B shows physical association of the PGF-CTERM helix with the bundle of archaeosortase A TM helices. The active-site residues are positioned to cleave the glycoprotein. At this point, several posttranslational modifications, such as glycosylations, may already have happened. The exact target site is unknown. Panel C shows the target protein, after cleavage, transiently attached to the conserved Cys residue of the archaeosortase/exosortase family, following the mechanism of sortase and the model of exosortase. Release of the glycoprotein is shown by transfer onto diphytanylglyceryl phosphate (transpeptidation), although another possibility is transfer to water (hydrolysis), with the lipid moiety being attached elsewhere. Meanwhile, the removed C-terminal peptide is headed for degradation. Panel D shows the glycoprotein after its interaction with archaeosortase, able to undergo (further) glycosylation and entry into the S-layer in mature form.