Virulence Potential of the Staphylococcal Adhesin CNA in Experimental Arthritis Is Determined by Its Affinity for Collagen

Abstract

Background.Staphylococcus aureusis a major cause of bacterial arthritis, which often results in severe joint damage. CNA, a collagen adhesin of S. aureus,was shown to be a virulence factor in several animal models. However, the precise molecular mechanism by which CNA contributes to virulence remains unclear. Methods. We examined the role of the collagen-binding function of CNA in a mouse model of septic arthritis by comparing the virulence of isogenic strains of S. aureusexpressing (1) wild-type CNA, (2) a truncated form of CNA (CNA35) with a higher affinity for collagen than the wild type, (3) CNA35 containing a single point mutation resulting in loss of collagen binding, (4) CNA lacking the collagen-binding domain, and (5) the collagen-binding domain of ACE (adhesin of collagen from Enterocoecus faecalis). Results and conclusions. The results provide, for the first time, direct evidence that the virulence of CNA depends on its collagen-binding ability. Collagen binding facilitated early colonization of the joints of mice. Furthermore, the virulence potential of the adhesin is determined by the adhesin's affinity for its ligand, as well as its binding kinetics. Staphylococcus aureus is the bacterium most commonly associated with septic arthritis [1], which often results in severe and irreversible joint damage and, occasionally, death. The development of S. aureus-medmtedarthritis is likely to be initiated by the successful adherence to and colonization of joint tissues by staphylococci. Earlier studies from our group demonstrated that the collagen adhesin CNA was necessary and sufficient for S. aureusto attach to cartilage in vitro [2]. Mice infected with CNA-deficient S. aureuspresented with a reduced incidence of arthritis, compared with mice infected with CNA-expressing S. aureus[3]. Furthermore, vaccination with a recombinant fragment of CNA protected mice from septic death after an infection with a lethal dose of S. aureus[4]. Studies from other groups also indicated that CNA-deficient S. aureusstrains were less virulent in their capacity to induce endocarditis [5], osteomyelitis [6], and keratitis [7]. Two recent reports indicated that the cnagene was associated more frequently with isolates from patients than with isolates from healthy adults [8, 9]. Taken together, these studies demonstrate that CNA is a virulence factor of importance in the pathogenesis of S. aureusinfections. CNA belongs to the MSCRAMM (microbial surface component recognizing adhesive matrix molecule) family of adhesins [10]. It contains from the N-terminus, a signal peptide, an A domain followed by 1–4 repeats, the cell-wall anchoring region, a transmembrane segment, and a short cytoplasmic tail. The A domain is responsible for binding to several types of collagen, whereas the repeats have no known effect on binding [10-12]. The function of the repeats is unknown, but this region is thought to form a stalk that displays the A domain on the cell surface [11, 13, 14]. Recently, CNA-like collagen-binding MSCRAMMs have been identified in other bacteria, among which are ACE (adhesin of collagen from Entero coccus faecalis)[15], Acm (adhesin of collagen from Enterococcus faecium)[16], and CNE (collagen-binding protein from Streptococcus equi)[17]. These proteins all share a domain organization similar to that of CNA, in which the A domain is responsible for their collagen-binding activities; however, the specific binding mechanisms for each collagen adhesin may be somewhat different. ACE40, a His-tag fusion protein containing the ACE A domain, bound collagen with an affinity similar to, but with kinetics different from, that of the CNA A domain [15]. Specifically, ACE40 displayed a much faster association and dissociation rate and recognized fewer binding sites in type I collagen, compared with the CNA A domain. Through genome sequencing, an open-reading frame encoding a CNA homolog was found in Clostridium perfringens[18], a gram-positive anaerobic bacterium that causes gas gangrene. In addition, a collagen-binding domain from Clostridium histolyti-cumclass I collagenase had an X-ray crystal structure [19] that was strikingly similar to that of the minimum collagen-binding domain of CNA [20], although the 2 molecules do not share significant sequence homology. It is likely that, with the elucidation of more bacterial genomes, the family of CNA-like collagen adhesins will continue to expand. Although there is a wealth of information demonstrating the importance of CNA in the pathogenesis of S. aureus, the molecular mechanisms by which CNA acts as a virulence factor are unclear. Several questions need to be addressed. Is the collagen-binding activity and affinity of CNA responsible for its role in disease? Do the domains contribute to the potential of the MSCRAMM as a virulence factor? Can other collagen-binding MSCRAMMs replace CNA on staphylococci without loss of virulence? In the present study, we have taken advantage of S. aureus strain CYL316, which does not contain the cnagene but can be used as a host for the expression of functional wild-type (wt)and mutated CNA. In this system, we expressed 2 truncations of CNA: 1 that shows an increased affinity for collagen, and 1 that contains the non-collagen-binding domain only. We also expressed a form of CNA that did not bind collagen, because of a point mutation. Finally, we replaced the collagen-binding domain of CNA with the corresponding domain of ACE. These strains were examined in vitro for their surface display of each recombinant fragment, as well as for their collagen-binding capabilities, before they were analyzed in a mouse model of septic arthritis. The results provided, for the first time, direct evidence that CNA contributed to S. aureusvirulence by binding to collagen and that the contribution correlates with its affinity for collagen (i.e., the higher the affinity, the higher the virulence). Bacterial strains, plasmids, and culture conditions. Strains and plasmids used in the present study are listed in table 1. Escherichia colistrains were grown at 37° overnight in Lennox L broth (LB; Sigma) or on LB agar with ampicillin (50 jOtg /mL) or spectinomycin (50 jOtg /mL), when appropriate. Staphylococci were grown at 37° in tryptic soy broth (TSB; Difco) or on tryptic soy agar (TSA), with tetracycline (Tc; 3 jOtg /mL), when necessary. Plasmids and strains used in the present study. Cloning and expression of CNA35 and the CNA35 mutant. The sequence encoding amino acid residues 30–344 of CNA was amplified by use of polymerase chain reaction (PCR) from a construct containing the entire cnagene, using primer pairs CNA55 5' and CNAN2 3' (table 2). The PCR product was cloned into pQE30, as described elsewhere [21]. Oligonucleotides used in the present study. A 2-step PCR method was used to construct the CNA35 mutant, as described elsewhere [21]. Primers CNAYKL and CNAYKR contain sequences for lysine instead of the tyrosine residue at position 175 of CNA. Primer pairs CNA55 57CNAYKR and CNAYKL/CNA35R were used to amplify regions encoding amino acid residues 30–176 and 174–344 of CNA, respectively. The PCR products were used as templates for the second round of PCR amplification with primer pairs CNA55 5; and CNA35R. Large-scale expression and purification of recombinant proteins were performed as described elsewhere [22]. Analysis of the collagen binding of various CNA fragments. Surface plasmon resonance in a BIAcore 1000 system (BIAcore AB) and solid-phase binding assays were used for the analysis, as described elsewhere [22]. Construction of an S. aureus surface-display expression vector. The integration vector pCL84 [23] was modified as follows: pCL84 was digested with SaRand Pstl,and the ends were filled in with Klenow fragment and religated. A 121-bp sequence 5'to the start of the cnagene was previously shown to support the expression of cnain a heterologous host [3]. Primer pairs CNAI and CNAS were used to amplify, by use of PCR, this region plus the region encoding the signal peptide of CNA (table 2). The PCR product was cloned into pCL84, resulting in plasmid pYX102. To introduce the cell wall-anchoring motif into pYX102, the sequence encoding the last repeat to 51 nucleotides 3'to the end of the cnagene was amplified by use of PCR using primer pairs CNAB57 and PSTCNA37. The product was cloned into pYX102, resulting in plasmid pYX104. The inserted regions were confirmed by use of DNA sequence analysis. Generation of heterologous strains. Primers CNA55 5' and CNA35R were used to amplify, by use of PCR, regions encoding CNA35 and CNA35Y175K from pQE30CNA35 and pQE30CNA35Y175K, respectively. Primers ACE40L and ACE40R were used to amplify regions encoding ACE40 from an ACE40 construct [15]. The PCR products were cloned into pYX104. The ligation mixture was transformed into E. coliJM101. Trans-formants were verified by examining the DNA banding patterns by use of agarose gel electrophoresis of restriction digestions of plasmid preparations. The correct constructs were introduced into S. aureus CYL316, as described elsewhere [24]. The integration should occur in the promoter region of the lipase gene on the chromosome of CYL316, which would lead to the loss of lipase activity. This was verified by assaying the lipase activity of the transformants [3]. The presence of the integrated gene was verified by use of PCR. Extraction of S. aureus cell wall-anchored proteins and Western blot analysis. Lysostaphin was used to extract cell wall-anchored proteins from S. aureusstrains, as described elsewhere [25]. Extracts from equal amounts of S. aureuscells were analyzed by use of Western blot analysis. Attachment of S. aureus strains to collagen. Various concentrations of log-phase S. aureuscells were incubated with immobilized bovine type I collagen (10 ji ¿g/well), as described elsewhere [25]. Attached bacteria were fixed with 25% formaldehyde and stained with 0.5% crystal violet. After washing, 100 jOtL of 10% acetic acid was added, and absorbance at 590 nm was recorded. Determination of the growth kinetics of S. aureus CYL strains. Individual colonies were used to inoculate 2 mL of fresh TSB-Tc and were cultured overnight at 37°. The cell cultures were adjusted to an optical density at 600 nm (OD600) of ∼ 1.0, and 1 mL of each suspension was then used to inoculate 20 mL of fresh TSB-Tc. OD600 was measured at every hour after inoculation. Examination of S. aureus recombinant strains in a septic arthritis mouse model. Pathogen-free, 4–6-week-old BALB/ c female mice (Harlan Sprague-Dawley) were used. Single colonies of S. aureusstrains were used to inoculate 2 mL of TSB. After shaking overnight at 37°, 1 mL of the overnight culture was inoculated into 20 mL of fresh TSB. Bacteria were harvested when the culture reached an OD600 of ∼0.4. The cell suspensions were further quantified by use of a BBL CrystalSpec Nephe-lometer (BD Biosciences) and adjusted to the corresponding colony-forming units per milliliter. Each mouse was injected intravenously in the tail vein with 0.5 mL of bacterial suspension. TSB alone was injected into mice as a control and did not show any effects. Mice were assessed over the course of 4 weeks, at days 3, 7, 14, and 28 after infection. Individual weights, appearance, and pathogenic lesions were documented. Mouse limbs were visually assessed for symptoms of arthritis by an observer with no knowledge of the infection status of the mice. Histopathological examination. BALB/c female mice were injected with the different S. aureus strains at a dose of 2 X 107cfu/mouse (15 mice/group). At day 28 after injection, both hind limbs of mice were removed and histologically examined and scored for arthritis by an observer with no knowledge of the infection status of the mice, as described elsewhere [26]. Tissue colonization of infected mice. Mice (3-7/group) that were injected with 2 X of S. at various and and days after were from as described elsewhere Bacteria were grown on at 37°, and the of colonies was analysis. The in of arthritis and were analyzed by use of the for bacterial and arthritis were assessed by use of the in were assessed by use of the were for by use of the Generation of CNA35 and a CNA35 and collagen-binding analysis. Several recombinant proteins containing different of the CNA A domain were expressed in E. of the which the of the collagen-binding A domain of CNA a higher affinity for collagen than did the A domain by both surface plasmon resonance for CNA35 was ∼ than that for CNA55 In a study, CNA35 and CNA55 each other for binding to collagen, that the binding was not the increased affinity for collagen by CNA35 was to of the collagen molecules specific organization of CNA and its recombinant fragments. CNA contains a signal peptide collagen-binding A region repeats and and cell wall-anchoring region, a transmembrane segment, and a cytoplasmic and are His-tag fusion recombinant The in the point mutation. for the heterologous expression and display of CNA on the surface of Staphylococcus promoter region of peptide sequences of of and to bovine type I collagen. plasmon resonance of and over a for CNA35 and CNA55 over immobilized collagen. bovine type I collagen were incubated with concentrations of CNA35 and then with a mouse and a were by use of the method with the model [22]. The for CNA35 and CNA55 were and respectively. of collagen-binding CNA35 and by use of acid residue in CNA was previously shown to be for collagen binding of the collagen-binding domain of CNA In the present study, we mutated to a lysine residue in the of to The mutated recombinant protein collagen-binding by both and The for was not Generation and of S. aureus strains and mutated aureus strain does not contain the gene for does not bind collagen. This strain was used as a host to display various CNA which were then examined in a mouse model of S. arthritis which was previously contains the cnagene [3]. regions in CNA35 and were cloned into the surface-display integration vector (table 1 and resulting as well as the integration vector without the surface-display and were introduced into CYL316, to strains and (table Western blot of of equal amounts of log-phase of the 4 strains, as well as indicated that the different of CNA were displayed on the surface of these strains and were equal in The in and not bound and different of protein A that bound Western blot of cell-wall of recombinant Staphylococcus S. aureuscells were with to extract cell wall-anchored the recombinant fragments. with an and with with a and In which displays the region corresponding to CNA35 and a repeat region, the for collagen which displays the CNA, to collagen at an which contains the point did not attach to collagen, that the point mutation was sufficient to collagen binding and of S. control strains and did not attach to collagen. These results are with from using the corresponding recombinant Collagen binding of Staphylococcus aureuscells various collagen-binding S. aureuscells were incubated with immobilized bovine type I collagen for 1 at in the of bacteria were fixed with 25% formaldehyde and stained with 0.5% crystal as described in and Methods. and and The A domain of the collagen adhesin ACE was cloned into and introduced into S. aureus CYL316, to strain (table 1 and blot of the from that the ACE A domain was to the cell also to collagen We measured the in vitro growth of the isogenic S. aureus strains. the strains and at the rate not effect on the by S. aureus strains and To the virulence potential S. aureus strains and we injected mice with 4 X 2 X 1 X or X the tail vein (15 At were in both the and the arthritis and mice. the mice infected with X by day 3 after infection compared with 2 mice that in the group all mice infected with 4 X of by day 3, there were no arthritis for this However, after the of the mice infected with arthritis by day 28 At X no significant were in the However, by day 3, of the mice 2 X of arthritis. day of the mice in this group had arthritis In of the mice of arthritis the day of mice 1 X of had arthritis, compared with of mice infected with the dose of in of arthritis were in mice infected with 0.5 X of the These that the collagen-binding activity of CNA the adhesin's virulence potential in mouse of both septic arthritis and septic death. of the virulence Staphylococcus and at different BALB/c mice were infected with different of and and were for rate and of arthritis, for to 28 infected with all mice infected with at 4 by day 3, there were no arthritis for this and infected with was by use of of S. aureus in tissues of infected mice. We that collagen-binding CNA the and colonization of joint tissues by S. this bacteria were cultured from joints taken from mice infected with 2 X of or at various after and quantified At no significant were in the of bacteria in mice infected with or mice/group). However, and after mice infected with had more bacteria in their joints than did mice infected with or A similar was at day These results that the collagen-binding function of CNA is for the colonization of the mouse joints early in the course of although CNA may not be in the of the Bacterial in the joints in the early of infection. Mice were infected with 2 X of Staphylococcus were at and and days after infection. were and for and was by use of the Analysis of arthritis in mice infected with S. aureus strains with different collagen-binding To the virulence potential of CNA was by the affinity of MSCRAMMs for collagen, we compared the of arthritis in mice infected with with that of mice infected with which collagen with affinity than does day of mice in the group had arthritis, more than in the group that higher collagen-binding affinity in a higher incidence of arthritis. the of arthritis in mice as did that the domains did not a role in arthritis in this model. The incidence of arthritis among mice infected with strain was similar to that among mice infected with the vector control strain that was the collagen-binding of CNA that to increased virulence of S. in mice infected with isogenic Staphylococcus aureusstrains different collagen-binding BALB/c mice were infected with 2 X of S. aureuscells the tail vein and were assessed for symptoms of arthritis by an observer with no knowledge of the infection status of the mice. were from 3 and and and and was using at day 28 by use of To a heterologous collagen adhesin replace CNA and contribute to the virulence of S. aureus in this we infected mice with day of mice arthritis similar to the incidence of arthritis in the group and higher than that for the non-collagen-binding control groups the control ACE40, a heterologous collagen adhesin that collagen with an affinity to that of CNA, was to the potential of S. aureusto a similar to that of S. aureusexpressing the in mice infected with different S. aureus strains. The of loss an infection correlates with the of the We examined the in mice (15 infected with the different S. aureusstrains at days 3, 7, 14, and 28 At days 3 and 7, were in mice infected with strains and compared with mice infected with the 2 non-collagen-binding strains, which at mice infected with strain significant loss at day 3 but their by day than mice infected with strains and these mice to their in weeks, compared with for the 2 mice infected with strain the that strain the most damage to mice the early of infection. of mice infected with isogenic Staphylococcus aureusstrains with different collagen-binding S. aureuscells X were injected into the tail of mice (15 mice/group). were at days 3, 7, 14, and 28 after infection and were compared with the before infection. The were for each and The for different was by use of At day 3, for the collagen-binding strains the non-collagen-binding strains. At day 7, for the non-collagen-binding strains. At day 14, for the non-collagen-binding strains. At day for the non-collagen-binding strains. Histopathological of mice. that the of the limbs were hind hind limbs from a were removed for (table 3 and The arthritis the results by as described to of arthritis, when all the limbs were taken into a higher was in mice infected with strain than in mice infected with strain groups had higher than mice infected with non-collagen-binding strains (table The arthritis for mice infected with was higher than that for mice infected with non-collagen-binding strains, but this was not the limbs were taken into similar were (table These results the from and further that the of S. aureusto attach to increased both the and of arthritis. Histopathological of the hind limbs of mice infected with different Staphylococcus Histopathological of the joints of mice. BALB/c mice were infected with of Staphylococcus aureus At day 28 after both hind limbs were removed and histologically of joint from mice with arthritis of 2 3 and 4 are A of no arthritis. joints with a of 1 were found in this S. aureus a of cell wall-anchored proteins of which as adhesins for matrix proteins The importance of these MSCRAMMs and as virulence in S. been demonstrated in various studies In the present study, we have confirmed that CNA, a of a family of collagen adhesins in gram-positive bacteria, is a virulence factor in septic arthritis. We also have determined the mechanism of of the MSCRAMM in infection. the of S. aureusto the joint is not on the presence of a collagen-binding of bacteria were from joints after the mice S. or not the strain a collagen-binding However, after mice infected with a collagen-binding S. had more bacteria in their joints than did mice infected with non-collagen-binding strains. These were of shown in a model of endocarditis [5], as well as from a study comparing S. aureusstrains that did or did not that S. the of mice when injected The adhesins the bacteria to containing their corresponding ligand, and, as a these bacteria to studies of S. aureus strains that did or did not CNA did not the in virulence to the of the strain to bind collagen. we have for the first time, that the virulence potential of CNA is to the adhesin's affinity for collagen. of a CNA domain that does not contain any known activity or CNA35 that contains a point mutation in S. aureusstrains with the virulence potential as S. aureusstrains that do not contain the In addition, of the results of mice infected with and indicated that there is a the affinity for collagen and the virulence of the higher the affinity, the higher the Furthermore, the of the collagen adhesin does not to be S. the collagen adhesin also increased the virulence of the staphylococci in the mouse model of septic arthritis. However, is to that the mice infected with to the as mice infected with and but much more than the 2 ACE40 collagen with a much faster dissociation rate than does CNA, which lead to of the bacteria in that the mechanism of binding the virulence potential of an of CNA is not to a bacterial host into a in the mouse model of septic arthritis. In we have expressed CNA in Staphylococcus staphylococci used in the These collagen bacteria did not cause septic arthritis in this model when injected into BALB/c mice at a dose of 1 X virulence in to CNA are to cause The of these is unclear. S. aureus is an to from other strains or as well as to be an in It is likely that can on their of virulence for adhesins or in the corresponding the to the host more We and for