Here, we analyzed and compared the genomes of the three propionibacterial species known to colonize the human skin. Species-specific gene clusters were identified in each genome that encode traits for colonization and host-interaction. Applying high-resolution microscopy and proteomic approaches we could verify the production of these surface-associated functions.
P. avidum was found to be surrounded by an EPS-like meshwork. A gene cluster that encodes proteins involved in the biosynthesis and modification of EPS was identified. The cluster encodes several homologs for enzymes involved in LPS and EPS biosynthesis (RfbA, RfbB, RfbD, RfaG, ExoU, NeuA, NeuB) as well as a number of glycosyl transferases with unknown specificities. RfbA, B, D are found in Lactococcus lactis, and required for dTDP-rhamnose biosynthesis, which is an important precursor of rhamnose-containing exopolysaccharides . The genes neuA and neuB are found in the LPS biosynthesis gene clusters of several Gram-negative species. NeuA and NeuB have been shown to be important in polysialic acid capsule biosynthesis . The P. avidum EPS gene cluster lacks a gene for a flippase, indicating that the EPS structure is formed on the outside of the cell. Interestingly, the cluster also contains genes involved in trehalose biosynthesis (TreY-TreZ pathway). The disaccharide trehalose can protect cells from environmental stresses such as low water availability . 20 of the 35 genes have a homolog in R. mucilaginosa that produces a mucilaginous capsular material . Like P. avidum, R. muciloginosa is occasionally isolated from disease sites, thus regarded as an opportunistic pathogen, for instance involved in prosthetic device infections. Well-studied EPS in other bacteria, such as in Pseudomonas aeruginosa, have several roles in pathogenicity; EPS contributes to biofilm formation, adherence to surfaces and host cells, evasion of phagocytosis, and elicitation of immune response [32–34]. We hypothesize that the EPS structure of P. avidum could have a role in biofilm formation, and thereby contribute to its pathogenicity by leading to persistent infections that cannot be cleared by the immune system. This might explain why P. avidum is in particular recognized in abscess formation after surgical intervention [11–13, 22]. It should be noted that several studies reported the presence of a cell wall-associated polysaccharide of P. acnes that can partially be extracted by phenol extraction [35–37]. Such a cell wall polysaccharide was further described as a lipidated macroamphiphile; this lipoglycan cell envelope component of P. acnes was found to have a lipid anchor and a polysaccharide moiety containing mannose, glucose and galactose, and probably diaminohexuronic acid . We strongly suspect that the lipoglycan of P. acnes is distinct from the EPS of P. avidum. We tested different strains of P. acnes grown under different conditions and could not detect any EPS-like structure by EM analyses (data not shown). To our knowledge no study so far could visualize an EPS-like meshwork on P. acnes cells. Moreover, the identified putative EPS biosynthesis genes of P. avidum are absent from the genomes of P. acnes (and P. granulosum).
We found that P. granulosum possesses pili/fimbriae-like appendices and pilin subunits were identified among cell surface-exposed proteins of P. granulosum. We determined two gene clusters encoding pilin subunits in direct vicinity to genes encoding sortases. Such clustering of genes for sortase and pilin subunits has been reported for a number of Gram-positive bacteria, including related actinobacteria such as Corynebacterium diphtheriae that produces three distinct pilus structures, SpaA-, SpaD- and SpaH-type pili . In corynebacteria pilins are covalently polymerized and the formed pilus is anchored to the bacterial cell wall; these steps are catalyzed by pilin-specific and housekeeping sortases, respectively. It has been shown that minor pilins (SpaB/SpaC) represent the major adhesins of corynebacteria [27, 39]. Thus, we hypothesize that pili of P. granulosum could have a role in adhesion to human skin tissue and colonization. They might also have a role in forming a multispecies biofilm, since P. granulosum and P. acnes are often detected together within sebaceous follicles. The anchorage of the base of the pilus to the cell wall is usually mediated by a housekeeping sortase . A likely candidate for this housekeeping sortase was identified among the surface-associated proteins: H641_09423 (strain DSM20700), a protein with a sortase E domain. A homolog exists in P. acnes KPA (PPA0777) and in P. avidum ATCC25577 (HMPREF9153_2132). These sortases likely catalyze the anchoring of other LPXTG-motif proteins of the three propionibacterial species to their cell walls. A genome search revealed that P. avidum contains 12, P. acnes 15, and P. granulosum 18 proteins (including 7 putative pilin subunits) with a C-terminal LPXTG motif. Most of these LPXTG-motif proteins have no or little similarity to known proteins, exceptions are proteins with nucleotidase or phosphoesterase domains. Only few of these LPXTG-motif proteins have been identified in the surfome. That could either indicate that they were not or weakly expressed under the applied growth conditions (liquid culture, complex broth), or were not accessible for trypsin cleavage.
Analysis of the surfome data of P. granulosum further revealed the presence of several cytosolic proteins, including ribosomal proteins and those involved in core metabolic functions (methylmalonyl-CoA:pyruvate transcarboxylase 12S subunit; two methylmalonyl-CoA mutases; fumarate hydratase class II; succinyl-CoA ligase). That indicates that P. granulosum seems to be more sensitive to trypsin treatment or lyse earlier than P. avidum and P. acnes.
P. acnes has, unlike P. granulosum and P. avidum, no obvious surface appendices. However, P. acnes is by far the most prevalent bacterium in sebaceous follicles of the face and back [2, 7]. Thus, this species must have evolved a different strategy to adhere to and colonize human tissues. Surface proteins could act as powerful adhesins. Indeed, the dermatan-sulphate adhesins DsA1 and DsA2 have been identified and partially characterized in P. acnes[18, 19]. These were not found in the surfome of the strain KPA, most likely because the respective genes are phase variable, but DsA1 and DsA2 are present on the surface of the type Ia strain 266 (data not shown). In addition, the surfome data revealed an abundance of lipoproteins with RlpA (rare lipoprotein A) domains on the surface of P. acnes. The bacterium specifically produces PPA2175 on the surface; it contains a SH3 and a peptidoglycan-binding domain (Figure 6). Bacterial lipoproteins have diverse functions; they play roles in a wide range of physiological processes. They can also function as ligands of the innate immunity host cell receptor Toll-like receptor 2 (TLR2), thus triggering an innate immune reaction . It has been reported that TLR2 was sufficient for NF-kappaB activation in response to P. acnes and activation of TLR2 resulted in an inflammatory cytokine response, which is thought to be of crucial importance in acne vulgaris [41, 42]. The TLR2 ligand of P. acnes is so far unknown. We speculate that one or all of the surface-exposed RlpA-domain lipoproteins of P. acnes are TLR2 ligands. These lipoproteins are abundantly produced on the surface of P. acnes and they are not covered or protected from host cell contact by other surface structures, such as EPS or pili in P. avidum and P. granulosum, respectively. It will be interesting to investigate if P. avidum and P. granulosum are also able to trigger TLR2 responses, or if this is specific to P. acnes.
The colonization of human skin by P. acnes can be achieved by other strategies, such as factors that allow successful competition with other bacteria, including P. avidum and P. granulosum. Successful competition might include the efficient acquisition of nutrients from host components. In this respect, only P. acnes expressed surface-attached endoglycoceramidases, which might hydrolyze gangliosides on host cell membranes . Another specific feature of P. acnes is the presence and the production of surface-exposed CAMP factors 1 and 2. It has been shown that CAMP factor 2 has properties of a co-hemolysin [17, 24, 25, 44]. Moreover, inhibition of CAMP2 by neutralizing antibodies efficiently attenuated P. acnes-induced inflammation in the mouse ear model , suggesting that CAMP2, and probably the other ones as well, are virulence factors of P. acnes. The corresponding camp1 and camp2 genes are absent in the genomes of P. granulosum and P.avidum. CAMP factors have been partially characterized in streptococcal species as co-hemolsyins and pore-forming toxins . They are involved in the CAMP reaction, the lysis of sheep erythrocytes by the synergistic action of the sphingomyelinase C from S. aureus and CAMP factor from Group B Streptococcus strains . The sphingomyelinase initially hydrolyzes sphingomyelin to ceramide (and phosphocholine) on the erythrocyte membrane, which renders the erythrocytes susceptible to the lytic activity of CAMP factor. It was recently shown that CAMP factor 2 of P. acnes can act as an exotoxin, exhibiting cytotoxic activity on host cells . The study of Nakatsuji et al. further suggests that CAMP factor 2 acts together with host acid sphingomyelinase to amplify bacterial virulence, thus supporting the degradation and invasion of host cells. The gene for CAMP factor 2 is located within a small gene cluster that also contains genes encoding sialidases and a sialic acid transporter. This cluster seems to be inserted into the P. acnes genome (Figure 2). It is tempting to suggest a functional connection of these factors as host-interacting and/or virulence traits. One possible scenario is that sialidases act directly on host cell membrane exposed gangliosides, thus releasing terminal sialic acid residues that are taken up by the sialic acid transporter and used as energy source. The remaining ceramide moiety could be a binding site for CAMP factor 2, in analogy to the CAMP reaction. Another component in this scenario could be surface-associated endoglycoceramidases of P. acnes that is predicted to hydrolyze gangliosides on host cell membranes into ceramides and oligosaccharides.
Important questions remain to be answered, in particular regarding the host tissue and host cell interactions of these three species. Although all three species are colonizing human skin, it is not known if these species actually compete at those sites or have adapted to occupy unique niches through species-specific host interactions. The different surface properties of the three species suggest that they have different colonization strategies that could be host cell or tissue-specific at skin and non-skin sites.