TLR-2 Recognizes Propionibacterium acnes CAMP Factor 1 from Highly Inflammatory Strains

Background Propionibacterium acnes (P. acnes) is an anaerobic, Gram-positive bacteria encountered in inflammatory acne lesions, particularly in the pilosebaceous follicle. P. acnes triggers a strong immune response involving keratinocytes, sebocytes and monocytes, the target cells during acne development. Lipoteicoic acid and peptidoglycan induce the inflammatory reaction, but no P. acnes surface protein interacting with Toll-like receptors has been identified. P. acnes surface proteins have been extracted by lithium stripping and shown to induce CXCL8 production by keratinocytes. Methodology and principal findings Far-western blotting identified two surface proteins, of 24.5- and 27.5-kDa in size, specifically recognized by TLR2. These proteins were characterized, by LC-MS/MS, as CAMP factor 1 devoid of its signal peptide sequence, as shown by N-terminal sequencing. Purified CAMP factor 1 induces CXCL8 production by activating the CXCL8 gene promoter, triggering the synthesis of CXCL8 mRNA. Antibodies against TLR2 significantly decreased the CXCL8 response. For the 27 P. acnes strains used in this study, CAMP1-TLR2 binding intensity was modulated and appeared to be strong in type IB and II strains, which produced large amounts of CXCL8, whereas most of the type IA1 and IA2 strains presented little or no CAMP1-TLR2 binding and low levels of CXCL8 production. The nucleotide sequence of CAMP factor displays a major polymorphism, defining two distinct genetic groups corresponding to CAMP factor 1 with 14 amino-acid changes from strains phylotyped II with moderate and high levels of CAMP1-TLR2 binding activity, and CAMP factor 1 containing 0, 1 or 2 amino-acid changes from strains phylotyped IA1, IA2, or IB presenting no, weak or moderate CAMP1-TLR2 binding. Conclusions Our findings indicate that CAMP factor 1 may contribute to P. acnes virulence, by amplifying the inflammation reaction through direct interaction with TLR2.


Introduction
recognition of peptidoglycan (PGN, a molecule expressed by many bacterial species), lipoproteins, and lipoteichoic acid (LTA) from Gram-positive bacteria, and of lipoarabinomannan from mycobacteria and zymosan from fungal. TLR4 acts together with CD14 and MD2 in the recognition of LPS from Gram-negative bacteria [24].
The surface proteins of P. acnes involved in the activation of TLR2 on keratinocytes and monocytes and leading to inflammatory lesions in acne have yet to be identified. The aim of this study was to identify surface proteins of P. acnes recognized by TLR2 in several strains with different inflammatory profiles.

Identification of P. acnes surface proteins recognized by TLR2
We investigated the protein recognized by TLR2, by extracting proteins in the presence of a high concentration of lithium, to ensure the selective removal of P. acnes surface proteins. We checked that all the strains used in this study were able to induce proinflammatory responses, by measuring CXCL8 production by keratinocytes, and also that the HaCaT keratinocyte cell line used in the in vitro assay could produce TLR2. Total cell lysates were analyzed by western blotting, which revealed the presence of a protein with a molecular weight of about 90-kDa in the HaCaT lysate, and in the 293T and ThP1 lysates used as positive controls (S1 Fig). The keratinocytes were then stimulated with both whole bacteria and lithium protein extract (LiE), and CXCL8 production was measured after 18 h of stimulation. Representative results are shown for the evaluation of the CXCL8 production from the 27 P. acnes strains used in this study (Fig 1). Whole P. acnes bacteria induced CXCL8 mRNA and protein production in a dose-dependent manner, at a MOI of 10 to 1000 (Fig 1A and 1B). LiE from the corresponding bacteria were tested and shown to have a similar capacity to induce CXCL8 mRNA and protein production in a dose-dependent manner (Fig 2A and 2B). Both whole bacteria and LiE of CHR (type IA 2 ) and PIE (type IB) strains, and, to a lesser extent, RON (type IA 1 ) strains, induced large amounts of CXCL8 at the lower MOI of 10 ( Fig 1B) or for a LiE protein content of 3.1 μg/ml, whereas the other strains did not (Fig 2B). These results indicate that P. acnes surface proteins induce the inflammatory reaction and are a suitable starting material for identification of the surface protein recognized by TLR2. Far-western blotting was performed on total LiE, and two protein bands recognized by TLR2, with apparent molecular masses of 24.5-and 27.5-kDa, were detected ( Fig 3B). According to the number of protein bands recognized by TLR2, and the intensity of the signal, we ranked strains as follows: (+++) for 2 strongly recognized bands (Fig 3B,  were detected in negative control experiments using only recombinant TLR2/detection substrate or biotinylated anti-TLR2 antibody/detection substrate (data not shown). We assessed the specificity of recognition by TLR2, by incubating the surface proteins with recombinant TLR4; only two faint bands were detected, at an apparent molecular weight of about 60-kDa ( Fig 3C, lanes 5, 7) in only two strains, and neither the 24.5-kDa nor the 27.5-kDa protein was recognized by TLR4. As the 24.5-kDa and 27.5-kDa proteins appeared to be recognized by TLR2 in 48% of the strains tested, we chose to focus on the characterization of these two proteins.

Proinflammatory activity of the proteins of interest
We assessed the proinflammatory properties of the 24.5-and 27.5-kDa proteins, by electrophoretic separation of these proteins from the LiE of the CHR (type IA 2 ) and PIE (type IB)  Table 1) at a MOI of 10 (A) and at MOI of 10, 100, 1000 (B). Total RNA was extracted and CXCL8 mRNA levels were determined by real-time RT-PCR and compared those of GAPDH mRNA (used as the control). The difference is expressed as a fold-change. CXCL8 production was analyzed by ELISA on culture supernatants. Control experiments were run with unstimulated cells. The data are presented as the mean ± standard deviation of three independent experiments. Statistical significance is indicated by * P 0.05, ** P 0.01, *** P 0.001, **** P 0.0001. doi:10.1371/journal.pone.0167237.g001 strains, elution from the gel and assessment of their ability to induce CXCL8 production ( Fig  4). We showed in NF-κB and CXCL8 promoter assays that these proteins were able to activate both promoters (Fig 4A). The direct stimulation of epidermal keratinocytes by these two proteins also led to the 10-to 100-fold induction of CXCL8 transcription ( Fig 4B) and an increase in the amount of CXCL8 protein to 10 pg/ml (Fig 4C). Control experiments were performed on cells alone and on cells stimulated with a solution eluted from the acrylamide gel in the same conditions; no activity was observed in either case. Both proteins of interest were able to stimulate the production of CXCL8 mRNA and protein in keratinocytes.

Characterization of the proteins of interest
After electrophoretic separation, the proteins of interest were excised from the gel and characterized by LC-MS/MS of the peptide mixture obtained after in-gel digestion. Protein sequence database searches identified the two proteins of interest as products of gene encoding a hypothetical 285-amino acid protein from the CAMP factor superfamily (Accession number gi| 488487765), covering 66% of the amino-acid sequence for the 27.5-kDa protein and 61% of the sequence for the 24.5-kDa protein ( Table 1). The sequence of the 27.5-kDa was 99%, 38%, 33%, 50% and 51% identical to the P. acnes CAMP factors 1, 2, 3, 4, and 5, respectively (Fig 5), and both proteins corresponded to the same P. acnes surface protein. The theoretical molecular mass of CAMP factor 1 is 30.411-kDa. We therefore analyzed the N-terminal sequences of the two eluted proteins (Fig 6), comparing them with the expected sequence for CAMP factor 1. The first five amino acids of the 27.5-kDa protein were APVAP, located at the putative signal sequence cleavage site starting at position 29 [35]. This sequence is consistent with the loss of 2476 Da from the N-terminus of the protein, resulting in a truncated protein with a calculated molecular mass of 27.935-kDa. For the 24.5-kDa CAMP-1, the first five amino acids were SLLDT, at position 64, corresponding to the loss of 6322 Da, resulting in a truncated protein with a calculated molecular mass of 24.089-kDa (Fig 6). Based on these findings, we concluded that the surface protein of P. acnes recognized by TLR2 was CAMP factor 1.

Blocking of CAMP-1 / TLR-2 interaction
We investigated the ability of a blocking agent to outcompete the interaction of CAMP factor 1 with TLR2. Anti-TLR2 antibodies have been shown to block CXCL8 production by keratinocytes [18,21]. The stimulation of keratinocytes with eluted CAMP-1 protein after their prior treatment with anti-TLR2 antibodies resulted in significantly lower levels of NF-κB (P = 0.008) and CXCL8 (P = 0.0012) promoter activation (Fig 7A and 7B), and, consequently, significantly lower levels of CXCL8 mRNA (P = 0.00047) ( Fig 7C) and CXCL8 protein production in keratinocytes (P = 0.0024) (Fig 7D). No such effect was observed when this experiment was TLR2 binding to P. acnes surface proteins. P. acnes surface proteins were extracted from a 5-day culture bacterial pellet and separated by electrophoresis in a 4-12% NuPAGE LDS BisTris gel (50 μg) with detection by Coomassie blue staining (A). Separated proteins were transferred onto nitrocellulose membranes, which were incubated with recombinant TLR2 (B) and TLR4 (C) (0.1 μg/ml). TLR binding activity was detected with specific biotinylated antibodies against TLR2 and TLR4, respectively, as described in the Materials and Methods. Lane 1 contains the molecular mass markers. Lanes 2 to 7 contain proteins from strains 6919, RON, CHR, GUE, TRI, and PIE, respectively. Arrows indicate the positions of the 24.5-and 27.5-kDa bands of interest.  performed with the IgG isotype control. Thus, CAMP factor 1 isolated from P. acnes interacts with TLR2, triggering a downstream signaling pathway leading to CXCL8 production.
Comparison of phylotype, TLR2 binding activity and CXCL8 production Using a previously described rapid multiplex PCR assay [36], we phylotyped all the isolates. We identified four different phylotypes: IA 1 (15 strains), IA 2 (2 strains), IB (9 strains) and II (2 Coomassie blue staining. Proteins of interest were excised from the gel and eluted as described in the Materials and Methods. (A) HaCaT cells were transfected with CXCL8 (-173 bp) (light gray bar) and NF-κB (dark gray bar) inserted into a construct upstream from the luciferase gene, for 24 h, after which, an internal control (the Renilla luciferase expression plasmid) was added to the transfection mixture. Cells were stimulated by incubation with the eluted proteins (10 μg) for 24 h at 37˚C. Relative NF-κB and CXCL8 promoter activities were determined as the ratio of firefly and Renilla luciferase activities. HaCaT cells were stimulated with the eluted proteins (20 μg) for 24 h at 37˚C and CXCL8 mRNA levels were evaluated by RT-qPCR (B), and CXCL8 protein levels were assessed by ELISA (C). Data are presented as the mean ± standard deviation of two independent experiments. Statistical significance is indicated by * P 0.05, ** P 0.01, *** P 0.001, **** P 0.0001. We analyzed the recognition of CAMP factor 1 by TLR2 and found that most of the strains with strong TLR2 binding activity were of types IB and II (Table 2). We then measured CXCL8 production by keratinocytes stimulated with the 27 strains. We found that the P. acnes strains tested could be separated into two groups, one with low to moderate levels of CXCL8 production (93 to 400 pg/ml) and the other with high levels of CXCL8 production (>400 pg/ml) ( Table 2). Most of the strains producing high level of CXCL8 were  of types IB and II. A comparison of phylotype with ability to induce inflammation for the 27 P. acnes strains studied showed that most of the strains with high levels of CAMP1-TLR2 binding activity belonged to types IB and II, whereas the type IA 1 and IA 2 strains produced CAMP factor 1 molecules that were only moderately strongly recognized, if at all, by TLR2 (Table 2).

CAMP-1 nucleotide and peptide sequences analysis
CAMP1-TLR2 binding profile differed between P. acnes strains (Fig 3). We therefore investigated the nucleotide and peptide sequences of the 27 P. acnes strains. The 858 bp corresponding to the sequence of CAMP factor 1 reported elsewhere [35] was sequenced and compared with the sequence obtained for CAMP factor 1 reference strain NCTC 737 (ATCC 6919). Sequences were then classified into four groups according to the number of nucleotide polymorphisms (Fig 8, Table 2). The first group corresponded to CAMP factor 1 genes with no nucleotide polymorphism ( Fig 8A). The second group corresponded to CAMP factor 1 genes with 1 nucleotide polymorphism, either A31>G or G361>T, corresponding to amino-acid changes at positions I11>V and V121>L, respectively (Fig 8B and 8C; Table 2). A third group consisted of CAMP factor 1 genes displaying two single nucleotide polymorphisms from the Proteins of interest were excised from the gel, eluted and pooled as described in the Materials and Methods. HaCaT cells were treated for 2 h with human anti-TLR2 (dashed bar) or with goat anti-IgGa antibodies (dark gray bar) and transfected by incubation for 24 h with NF-κB transcription factor (A) and CXCL8 (-173 bp) (B) inserted into a construct upstream from the luciferase gene, after which, an internal control (the Renilla luciferase expression plasmid) was added to the transfection mixture. Cell were stimulated with eluted CAMP factor 1 (50 μg/ml) for 24 h at 37˚C. Relative NF-κB and CXCL8 promoter activities were determined as the ratio of firefly and Renilla luciferase activities. Pretreated HaCaT cells were stimulated by incubation with eluted CAMP factor 1 (50 μg/ml) for 24 h at 37˚C, CXCL8 mRNA levels were determined by RT-qPCR (C), and CXCL8 production was measured by ELISA (D). Control experiments were performed with cells alone (dark bar) and with stimulated cells not previously treated with CAMP factor 1 (light gray bar). Data are presented as the mean ± standard deviation of two independent experiments. Statistical significance is indicated by * P 0.05, ** P 0.01, *** P 0.001, **** P 0.0001.

Discussion
The aim of this study was to characterize the P. acnes surface protein recognized by TLR2. We used keratinocytes in our in vitro assay, because these cells are the first line of defense against external aggressions. They contribute to the innate immune response by producing chemotactic factors that attract leukocytes and neutrophils to the site of infection [25]. P. acnes has been shown to trigger the production of proinflammatory molecules by monocytes [28, 33,], and the massive secretion of IL-1α, TNF-α [26], IL-8 [37], and reactive oxygen species from keratinocytes [29]. In this study, we used the HaCaT cell line, corresponding to spontaneously transformed human adult skin keratinocytes. These cells can differentiate [38], express Toll-like receptors [39] and produce numerous cytokines in response to bacterial stimulation [27]. We found that all strains induced CXCL8 production in vitro in a dose-dependent manner, although the amount of CXCL8 produced differed between strains. This finding is consistent with previous studies describing differences in the modulation of the immune response between different P. acnes strains [12,21]. We assessed the ability of surface proteins to induce CXCL8 production in vitro. All P. acnes crude surface protein extracts were able to activate the synthesis of CXCL8 mRNA and protein by keratinocytes, and were therefore considered suitable for further analysis. The overproduction of CXCL8 observed with whole bacteria was reproduced with the corresponding protein extract, consistent with the presence of specific molecular material involved in CXCL8 induction. However, overall CXCL8 levels were lower for protein extracts than for whole bacteria, suggesting a partial loss of proteins during the extraction process. Using Far-western blot analysis, we characterized two surface proteins recognized by TLR2 with apparent molecular masses of 24.5-and 27.5-kDa; no bands were detected in the control experiments. We assessed the specificity of recognition, by testing the same protein extracts with recombinant TLR4. No cross-reactivity was observed, suggesting that the interaction with TLR2 was specific. For confirmation of the relevance of the 24.5-and 28.5-kDa proteins as PAMP candidates, we evaluated the ability of these proteins, isolated from electrophoresis gels, to induce inflammation in keratinocytes. Both proteins activated the CXCL8 and NF-κB promoters, and induced the production of CXCL8 mRNA and protein.
These results are consistent with these two proteins being good candidates for triggering an inflammatory response. These proteins were then subjected to proteomic analysis and characterized as P. acnes CAMP factor 1. CAMP is the acronym for the Christie-Atkins-Munch-Petersen reaction, a co-hemolytic reaction of erythrocytes involving synergy between the CAMP protein from group B streptococci and the sphingomyelinase (SMase) from S. aureus [40]. CAMP factors have been identified in several bacterial species, including Streptococcus agalactiae [41], and Pasteurella haemolytica [42]. Our results are consistent with those of previous studies reporting a co-hemolytic reaction involving P. acnes [43], and the identification of nucleotide sequences for P. acnes CAMP factors 1-5 in whole genomes and the detection of the corresponding proteins in protein extracts [35, 44,].
In this study, two proteins were identified as binding to TLR2, and these two proteins matched two proteins stained with Coomassie blue on the gel. A previous study detected only one protein, at about 28-kDa, with a monoclonal antibody against CAMP factor 1 and staining [35]. This discrepancy may be accounted for by the detection methods (antibody versus TLR2 binding), growth media (BHI versus RCM), and protein extraction methods (sonication/detergent versus lithium) used. According to the peptide sequence of CAMP factor 1 from the reference strain NCTC 737, which contains 285 amino-acid residues, the calculated molecular mass of the protein is 30.411-kDa including the signal sequence; the removal of the signal sequence would result in a protein of 27.656-kDa [35]. In terms of precision, we considered the two proteins at 28-kDa [35] and 27.5-kDa to correspond to CAMP factor 1 without its signal sequence. This hypothesis was confirmed by N-terminal sequencing of the 27.5-kDa protein, which yielded a sequence of APVAP. However, for the 24.5-kDa protein, which was also identified as CAMP factor 1, the N-terminal sequencing obtained was SLLDT. No other putative cleavage site has ever been described and additional studies will be required to determine whether this truncated CAMP factor 1 protein originates from P. acnes itself, is produced artificially during the extraction process or results from effects of the growth medium on protein production, as shown in a previous study [14]. Moreover, although CXCL8 induction experiments have shown that both the full-length and truncated forms of CAMP factor 1 are active, it is not yet possible to draw any firm conclusions about which of these forms is most active.
There is currently no clear evidence concerning the fate of P. acnes CAMP factors: secreted and/or associated with the cell surface. Previous studies of CAMP factors in S. agalactiae showed these factors to be both present on the bacterial surface and secreted into the culture medium [45]. In this study, we extracted surface proteins from washed bacterial pellets by lithium stripping. Our results are consistent with the CAMP factor 1 protein being associated with the surface but not firmly bound to it, as it can be removed from whole bacteria by trypsin cleavage [46]. Previous studies have reported the presence of P. acnes CAMP factors in both the external medium and on the cell surface, but with considerable differences between strains [22,35]. However, growth phase conditions have been shown to influence the expression of P. acnes genes [47]. In our study, we systematically used the same growth anaerobic conditions, until stationary phase was reached (OD > 0.6), on reinforced clostridial media. CAMP factors have no C-terminal LPXTG anchoring domain. Further investigations will therefore be necessary to determine how CAMP factor 1 is produced and processed at the surface of P. acnes and secreted into the external medium.
Nothing is currently known about the recognition of CAMP proteins by TLRs. CAMP factors can bind to several proteins, including immunoglobulins, via their Fc-binding region [48], and GPI-anchored proteins, via their glycan moieties [49]. TLR2 recognition preferentially involves lipoproteins expressed at the bacterial surface, but several members of the nolipoprotein pore-forming toxin family have been shown to interact with TLRs, consistent with broad specificity for TLR2-mediated recognition [50]. Further studies are underway to identify the parts of the protein involved in this binding.
In this study, four phylotypes were identified. Most of the acne-(4/6) and skin-related strains (3/4) belonged to type IA 1 , whereas the strains isolated from bone, joint and soft tissue infections mostly belonged to phylotypes IA 2 , IB and II. This result is consistent with previous findings [23]. We found that the intensity of TLR2 binding to CAMP factor 1 differed between the strains tested. Indeed, 59% (16/27) of the strains displaying little or no TLR2-CAMP1 binding activity were phylotyped IA 1 and IA 2 , whereas strains with a strong binding activity were typed IB and II. Moreover, the levels of CXCL8 production induced by P. acnes in keratinocytes seemed to be higher for the IB and II phylotypes, and lower for the IAs phylotype. This difference in CAMP1-TLR2 binding activity may reflect two phenomena: 1) differential CAMP1 gene expression and, 2) CAMP1 sequence variation affecting binding activity. Differences in CAMP factor 1 gene expression between P. acnes strains have been described before, with CAMP1 genes most strongly expressed in types IB and II, but analyses of the Shine-Dalgarno sequence shed no light on the reasons for this [35]. Many bacteria regulate virulence factor expression as a function of growth phase. Indeed, P. acnes produces most of its secreted virulence-associated factors during the exponential growth phase [47], and we cannot exclude the possibility of a role for regulatory proteins in modulating CAMP factor 1 production, as previously reported for the major carbon catabolite repressor protein CcpA, which regulates expression of the CAMP gene in Streptococcus pyogenes by interacting directly with its promoter [51]. We first investigated possible variations of the CAMP factor 1 sequences from the 27 strains, and detected considerable genetic polymorphism. This result is consistent with previous studies reporting differences in CAMP factor nucleotide and amino-acid sequences between P. acnes phylotypes [18]. We found that CAMP factor 1 was genetically distant from the other CAMP factors, but more related to CAMP factor 5, as previously described [35]. CAMP factor 1 sequences formed two main groups, according to the level of sequence polymorphism. The first group (A) was characterized by a highly polymorphic CAMP-1 sequence, with 14 amino-acid changes relative to the reference sequence, isolated from type II strains with moderate and high levels of TLR2 binding activity and with strong CXCL8 induction. This group included one strain (A11) isolated from a patient with severe acne. The other group (B) was more heterogeneous, and included CAMP sequences displaying 0, 1 or 2 amino-acid changes relative to the reference sequence. Interestingly, the group with no sequence polymorphism (B3) corresponded to strains inducing low or moderate levels of CXCL8 production, producing a CAMP factor 1 not recognized by TLR2; all these strains were isolated from patients with moderate acne. Otherwise, the groups presenting 1 or 2 sequence polymorphisms (B1, B2) corresponded to strains with a CAMP factor 1 recognized by TLR2 and inducing moderate or high levels of CXCL8 production. These results suggest that CAMP factor 1 recognition by TLR2 may be associated with the presence of polymorphisms in its sequence. Given that we have characterized a truncated CAMP-1 protein, it would be logical to rule out the I11>V amino-acid change as being of relevance. The I107>V polymorphism seems to be associated with low levels of TLR2 binding activity, whereas the E74>K polymorphism is present in CAMP-1 sequences from strains with high levels of TLR2 binding activity. Two strains did not seem to match this pattern (38862 and 22795) and may not be the best representative strains for their groups, because previous studies have shown that neither the type IB strain KPA171202 nor the type II strain NCTC 10390 produces CAMP factor 1 [35, 52,]. Furthermore, A30 was the only IA 1 type strain tested to display moderate TLR2 binding and moderate CXCL8 induction. This strain was isolated from a patient with severe acne and appears to be genetically separate from the other groups. However, the production of small or moderate amounts of CXCL8 by keratinocytes in response to IA P. acnes strains indicates that other P. acnes surface molecules are involved in this process. Indeed, as in other Gram-positive bacteria, peptidoglycan is a major constituent of the outer envelope in P. acnes and it promotes CXCL8-related inflammation through the TLR2 pathway [28, 53,]. Overall, these results suggest that specific sequence polymorphisms may play an important role in determining the strength of CAMP1-TLR2 binding. However, this approach has limitations and further investigations with mutant strains lacking CAMP1 expression, and with recombinant forms of CAMP1 will be required for better evaluations of the intensity and the specificity of TLR2 recognition.
Several components of Gram-positive bacteria, such as peptidoglycan, lipoteichoic acid, and lipoproteins, have been shown to induce inflammation by interacting with TLR2. However, LPS-TLR4 interaction seems to induce a stronger inflammation response than PGN-TLR2 interaction [37], and the P. acnes inflammatory mediators able to induce reactions resembling those induced by LPS have not yet been identified. P. acnes was long considered a bacterial contaminant, but several studies have provided evidence to suggest that P. acnes should be considered an opportunistic pathogen producing several proteins potentially involved in the degradation of molecules present in host tissues, together with cell wall-associated surface proteins that may interact with the host, triggering inflammation [23, 54,]. CAMP factors are among the proteins thought to be involved in P. acnes pathogenesis [55,56]. These factors are present in both commensal and pathogenic strains and their classification as "host adaptation factors" potentially helping the bacteria to survive in this highly competitive niche has been suggested, as S. aureus bacteria have been reported to use P. acnes CAMP factors to enhance their virulence [23,57,]. CAMP factor 1 is expressed by P. acnes in both case and control samples [58], but the injection of recombinant CAMP into mouse ear induces local inflammation [55]. CAMP factor 2 knockout partially alters co-hemolytic activity, whereas CAMP factor 4 knockout has no effect, suggesting redundancy between these factors. Moreover, the stimulation of HaCaT keratinocytes with a ΔCAMP2/4 knockout mutant revealed no differences in initial inflammatory response, whereas expression of the neutral SMase 3 gene was deregulated in the ΔCAMP4 mutant [52]. Moreover, the interaction of P. acnes CAMP factors with SMase suggests that CAMP factors may be involved in P. acnesrelated pathogenesis [56]. In conclusion, we suggest that CAMP factor 1 may contribute to P. acnes virulence by amplifying the inflammation reaction through direct interaction with TLR2.

Bacterial culture and protein extraction
We used 27 strains of P. acnes: strain 6919 from the American Type Culture Collection (Manassas, VA), 6 strains from acne lesions (3 from patients with severe acne and 3 from patients with moderate acne, collected at the Dermatology Department of Nantes Hospital), and 21 strains isolated from various infections (11 bone and joint infections, 1 case of pleuritis, 1 of pericarditis, 2 of pneumonia, 2 cutaneous infections, 1 case of bacteremia, and 3 ascitic fluid infections) ( Table 3). P. acnes strains were grown under anaerobic conditions in liquid reinforced clostridial medium (RCM) at 37˚C for five days, until they reached stationary phase. Typically, cultures were carried out in 100 to 1000 ml of RCM and bacteria were harvested by centrifugation at 5 000 x g for 10 minutes at room temperature. Pellets were washed in 20 to 200 ml of PBS and centrifuged again. The total surface protein extract was obtained by suspending the bacterial pellet in 20 to 200 ml of 1 M LiCl and incubating at 4˚C for 18 h with gentle mixing. The bacterial suspension was centrifuged at 5 000 x g for 20 minutes to remove the bacteria, and the supernatant, containing the surface proteins, was concentrated by precipitation with ammonium sulfate at 80% saturation for 18 h at 4˚C. The precipitated proteins were recovered by centrifugation at 15 000 x g for 30 minutes at 4˚C, resuspended in 20 ml of PBS and extensively dialyzed against PBS. Protein concentration was determined by the Lowry method, using BSA as the standard, as previously described [59].

Cell culture and stimulation
The human keratinocyte cell line HaCaT (provided by Dr N.E. Fuseining, Heidelberg, Germany) was grown in Dulbecco's modified Eagle's medium-Glutamax-I (DMEM) (Invitrogen, Cergy Pontoise, France). The ThP1 immortalized human monocytic cell line was grown in Roswell Park Memorial Institute 1640 Medium-Glutamax-I (RPMI). DMEM and RPMI were supplemented with 10% heat-inactivated fetal calf serum (Invitrogen, Cergy Pontoise, France), and an antibiotic/antimycotic solution (10 U/ml pencillin, 10 μg/ml streptomycin, 0.25 μg/ml amphotericin; Invitrogen) and cultures were incubated at 37˚C, under a humidified atmosphere containing 5% CO 2 . For stimulation experiments, HaCaT cell were incubated with a P. acnes suspension adjusted to the appropriate concentration (by absorbance measurement at 620 nm) or with lithium protein extract (LiE) in buffer solution, for desired period of time, at 37˚C, under an atmosphere containing 5% CO 2 .

DNA extraction and phylotyping
Bacterial strains were grown for 5 days at 37˚C under anaerobic conditions and DNA was extracted with the E.Z.N.A. 1Bacterial DNA kit, in accordance with the manufacturer's recommendations (Omega Bio-tek, Norcross, USA). DNA concentration was determined by measuring absorbance at 260 nm (Nanodrop 2000 ThermoScientific). The values obtained were in the range of 30 to 220 ng/μl, and the concentration of the solution was adjusted as required. Phylotyping was performed by multiplex touchdown PCR, as previously described [3], and the amplification reaction was carried out with a mixture containing 0.2 U Taq DNA polymerase, 1X PCR buffer, 50 mM MgCl 2 , 5 or 10 mM dNTP, with an initial denaturation at 94˚C for 60 s followed by 14 cycles of 94˚C for 30 s, 66˚C for 30 s, and 72˚C for 60s; then 11 cycles of 94˚C for 30 s, 62˚C for 30 s and 72˚C for 60 s. The mixture was then subjected to a final elongation step consisting of heating at 72˚C for 10 minutes. The genes amplified were the 16S rRNA (expressed by all P. acnes phylogroups), ATPase (type IA 1 , IA 2 , IC), sodA (types IA 2 , IB), toxin (Fic family) (type IC), atpD (type II), and recA (type III) genes, with the primers described elsewhere [35]. The PCR products (5 μl) were analyzed by electrophoresis in a 1.5% agarose gel and the amplified nucleotide bands were visualized by UV transillumination of the ethidium bromide-stained gel, with comparison to a 50-bp DNA ladder (Invitrogen, Carlsbad, California, USA).

Western blotting
HaCaT and ThP1 cells were washed in sterile cold PBS and lysed in RIPA buffer containing 50 mM Tris, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS (pH 8.0). Total extracts were centrifuged at 14,000 x g for 15 minutes at 4˚C and the supernatant was used as total protein extract. Protein samples (15 μg), and lysates of human TLR2-transfected 293T cells were fractionated by electrophoresis in a NuPAGE Novex 4-12% Bis-Tris gel (1 mm, 12 wells, Invitrogen, UK) and transferred onto nitrocellulose membranes. Membranes were saturated by incubation for 1 h at room temperature in TBS/T buffer and incubated at 4˚C for 18 h with goat polyclonal IgG antibodies against human TLR2 (C-19), diluted to 0.5 μg/ml in TBS supplemented with 5% BSA, and 0.1% Tween-20 (all antibodies and 293T lysate were purchased from Santa Cruz Biotechnology, Inc., Santa Cruz, Calif., USA). The membranes were washed and bound antibodies were detected by incubation with donkey polyclonal antibodies against goat IgG coupled to peroxidase, diluted 1:2000 in saturation buffer, for 1 h at room temperature. Unbound material was removed by washing and peroxidase activity was detected in a chemiluminescence assay (Advansta Corp., Menio Park, Calif. USA).

Protein elution
Proteins were fractionated by electrophoresis in denaturing conditions (SDS-PAGE, gel 13 x 13 cm) and detected by Coomassie blue. Proteins from a second gel were transferred onto nitrocellulose membranes and analyzed by Far-western blotting, as described above. Proteins recognized by TLR2 were matched to the stained bands, and excised from the gel. Excised protein bands were incubated for 18 h at 37˚C in the presence of elution buffer (1% SDS in PBS), with stirring. The excess of SDS was removed by precipitation of the proteins in cold acetone (vol/vol) and centrifugation for 30 minutes at 14 000 x g and 4˚C. Protein pellets were suspended in PBS and protein concentration was determined by spectroscopy at 280 nm (NanoDrop).

Cell transfection and luciferase reporter gene assay
HaCaT cells were cultured in 24-well plates (Corning Costar) with DMEM containing 0.1% FCS without antibiotics. They were then transfected with reporter plasmids containing the CXCL8 promoter (-173 bp) or the NF-κB promoter (5 units) upstream from the firefly luciferase gene in JetPEI TM (Polypolus Transfection, Illkirch, France) [60]. Transfection efficiency was normalized by co-transfecting the cells with 10 ng of pSRa-Renilla luciferase expression vector. The cells were incubated for 24 h, and were left unstimulated or were stimulated with LiE or proteins of interest eluted from the gel after electrophoretic separation. After 18 h of incubation at 37˚C, cells were washed with sterile cold PBS and scraped directly into the lysis buffer provided by the manufacturer (Promega Corp., Madison, WS, USA). Firefly and Renilla luciferase activities were measured on a MicroLumat Plus LB 96V Luminometer (Bertold Technologies, Bas Wild-Bad, Germany). Results are expressed as a fold-increase relative to the basal activity measured in unstimulated transfected cells.

RNA isolation and RT-qPCR
HaCaT cells were grown in six-well plates (Corning Costar) for 24 h with 10% FCS in DMEM, and then for 24 h in 0.1% FCS DMEM. HaCaT cells were then incubated for 5 h at 37˚C under an atmosphere containing 5% CO 2 and stimulated with whole bacteria (MOI 10), LiE (50 μg/ ml), or proteins of interest eluted after electrophoretic separation (50 μg/ml). After stimulation, cells were washed twice with cold sterile PBS and scraped into lysis buffer, according to the manufacturer's instructions (Macherey-Nagel). Total RNA was isolated with the NucleoS-pin1 RNA II kit (Machery-Nagel), according to the manufacturer's instructions. RNA concentration was determined at 260 nm (NanoDrop). It ranged from 110 to 225 ng/μl, and was adjusted to desired concentration. RT-qPCR was performed with the iTaq Universal Sybr Green one-Step kit (BioRad) containing reverse transcriptase (RT) and Taq polymerase. Complementary DNA was generated from 330 ng of total RNA by RT for 10 minutes at 50˚C, in a final volume of 10 μl. Standard amplification for quantitative PCR was carried out in the same test tube containing the reaction mixture, and the PCR conditions were as follows: 95˚C for 60 s, followed by 40 cycles of 95˚C for 15s, 68˚C for 60s, and 72˚C for 15s, followed by a final stage to obtain a melting curve for 65 to 95˚C, at 0.1˚C/s over 60 s on a LightCycler Nano (Roche). PCR results were calculated as a fold-increase relative to GADPH gene expression. Threshold cycles (Ct) were determined for the genes studied, from the amplification curves. The amount of RNA in stimulated cells relative to control cells was calculated by the 2ΔCt method. We used the TCTTGGCAGCCTTCCTGATT and TTTCGTGTTGGCGCAGTGT primers for IL-8 and the GCCACATCGCTCAGACAC and GCCCAATACGACCAAATCC primers for GAPDH, which was used as a control.

N-terminal sequencing
For N-terminal determination, 20 μg of protein mixture was subjected to electrophoresis and the bands of interest were excised and incubated in a buffer for passive extraction of the protein from the acrylamide gel. After overnight incubation with shaking, the solution was eluted on a ProSorb Filter (Applied Biosystems) to fix the protein on a PVDF disc. The PVDF disc was submitted to microsequencing by Edman degradation (8 cycles of pulsed-liquid chemistry), with a 494 automated protein sequencer (Applied Biosystems).

CAMP-1 PCR amplification and sequencing
For amplification of the nucleotide sequence of CAMP factor 1 from the 27 P. acnes strains, we performed PCR with the C1-F and C1-R primers, amplifying a 946 bp fragment, as previously described [35]. Briefly, standard amplification was carried out in a final volume of 25 μl con-

Phylogenetic analysis
Phylogenetic analyses were performed to assess the relationships between the P. acnes CAMP factor 1 sequences obtained in this study and other known CAMP sequences. These analyses were carried out with the maximum likelihood method. Multiple sequence alignments were obtained with Mafft version 7 [61] and exported into Seaview software [62], for input into JModelTest [63] and ProtTest [64] to identify an appropriate substitution model for phylogenetic analysis with the Bayesian information criterion. The best model for nucleotides was HYK85 and the best model for proteins was WAG, both without a gamma distribution. The phylogenetic analysis was performed with PhyML3.0 [65], implemented in Seaview software [16]. Sequences were randomized by 500 bootstrapping resamplings, and statistical analyses were performed on 100 datasets for each analysis.

Nucleotide sequence accession numbers
All the sequences determined in this study have been deposited in the GenBank database under the accession numbers listed in Table 4.

Statistical analysis
The statistical significance of differences between data from experimental groups was analyzed in paired Student's t-tests. P < 0.05 was considered significant. Statistical significance is indicated by Ã (P 0.05), ÃÃ (P 0.01), ÃÃÃ (P 0.001) and ÃÃÃÃ (P 0.0001), respectively.