INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Staphylococcus aureus is a major pathogen responsible for a wide range of both acute and chronic infections (37). A key step in S. aureus infection is its ability to attach to various surfaces and colonize host tissues. For this purpose, S. aureus carries several functionally redundant surface adhesins, which have high affinity for either soluble proteins or extracellular-matrix components of the host (26, 38). These include, for instance, prothrombin (28), fibrinogen (18, 25), fibronectin (11, 33), vitronectin (14), and collagen (27), as well as other constituents.

Since attachment to host tissues is an essential step of S. aureus disease, it was assumed that interfering with bacterial adhesion could be a means to prevent infection. This possibility was investigated in staphylococcal mutants defective in one or more of these cell-wall-associated ligands (9, 18, 22). Knockout mutants were tested both in vitro for their decreased adherence to purified ligands and in vivo for their lower pathogenicity in various animal models. When compared to their parent strains, the defective mutants were always significantly less able to attach to surfaces coated with purified ligands (4, 9, 18). Unexpectedly, however, these differences did not translate into major alterations of pathogenicity in vivo. Infectivity was either moderately or not affected in animal models such as experimental mammary abscesses in mice or experimental endocarditis in rats (1, 22).

The reason of this discrepancy was unclear. However, since S. aureus carry redundant adhesins on their surface, it was conceivable that other, functionally redundant adhesins were complementing the deficient mutant, thus masking the genuine effect of the defective determinant. If true, then studying the pathogenic role of individual surface factors in the redundant staphylococcal background might be a difficult task. To circumvent this problem, we attempted to transfer and express specific staphylococcal adhesins in a surrogate bacterium lacking the staphylococcal redundant background, and we tested the recombinants for a gain in infectivity in vivo. Experiments with Streptococcus gordonii indicated that the method was feasible (P. Stutzmann, J. Entenza, P. Francioli, and P. Moreillon, Abstr. 38th Intersci. Conf. Antimicrob. Agents Chemother., abstr. B77, 1998). However, S. gordonii was not a perfect recipient because it carried pathogenic determinants of its own. Therefore, in the present experiments we refined this system by using Lactococcus lactis subsp. cremoris (6) as a recipient organism. This gram-positive bacterium is not known to carry adhesins to mammal matrix proteins and has a well-characterized genetic background. Moreover, both staphylococci and lactococci process their cell wall proteins in a similar way, using the conserved LPXTG C-terminal motif to anchor the polypeptides to the peptidoglycan (32). This condition is absolutely required if staphylococcal proteins are to be expressed on the surface of the recipient cells.

The following experiments describe the successful cloning and expression of the staphylococcal clumping factor A (clfA) in L. lactis. A new expression vector allowing the shuttling of cloned genes between Escherichia coli and L. lactis was constructed, and the functionality of the transferred clfA product was assessed by the ability of clfA-positive lactococci to clump in the presence of plasma or fibrinogen.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. L. lactis subsp. cremoris 1363 (kindly provided by A. Gruss) (7, 8) was grown at 30°C in M17 medium (Oxoid) supplemented with 0.5% glucose (GM17) either in liquid medium or on agar plates (35). S. aureus Newman (5) was grown at 37°C either in tryptic soy broth (Difco Laboratories, Detroit, Mich.) or on tryptic soy agar (Difco). E. coli XL1-Blue was grown at 37°C in Luria-Bertani medium (Difco) (31). Whenever appropriate, antibiotics were added to the media as follows: erythromycin at 5 µg/ml for L. lactis and at 500 µg/ml for E. coli and ampicillin at 50 µg/ml for E. coli. Bacterial stocks were kept at 70°C in liquid medium supplemented with 10% (vol/vol) of glycerol.

Plasmids and plasmid constructions. The plasmids and plasmids constructions used in this study are listed in Table 1. Chromosomal DNA from S. aureus Newman was prepared as described by Marmur (17). The same procedure was applied to extract DNA from L. lactis subsp. cremoris 1363, except that lactococcal wall were digested with 1 mg of lysozyme per ml instead of lysostaphin.

Luciferase assay. To test the functionality of the lactococcal P23 and P59 promoters in the expression vector pOri23 and pOri59 (Table 1), a luciferase (luc) gene was inserted downstream of these sequences, and luc expression was determined by measuring light emission by the microorganisms at various times of growth. In brief, 10-ml cultures of L. lactis and E. coli harboring the appropriate plasmids were incubated at 30 and 37°C, respectively, and monitored for growth with a spectrophotometer (Sequoia-Turner, Montainville, Calif.) at a wavelength of 620 nm (optical density at 620 nm). At several time points, 100-µl portions of cultures were removed and transferred into tubes containing 250 µl of 0.1 M sodium citrate (pH 5.5). A total of 50 µl of 1 mM beetle luciferine (Promega Corporation) diluted in 25 mM glycylglycine and 15 mM MgSO₄ (final concentrations) was automatically added by the autoinjector of the luminometer (Turner designs, Sunnyvale, Calif.). After 1 s of equilibration, light was measured during a 10-s period. Bacteria carrying empty pOri23 and pOri59 plasmids (Table 1) were used as negative control, and background was measured by light emission of sterile medium. All of the experiments were performed in triplicate.

SDS-PAGE and Western Blots. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed by standard procedures using 6.5% acrylamide gels (13). Cell-wall-associated proteins were extracted from L. lactis and S. aureus as follows. Whole cells from 500-ml portions of exponential cultures were harvested by centrifugation (5,000 × g, 10 min, 4°C), washed twice with ice-cold 0.9 M NaCl, and resuspended in 5 ml of digestion buffer (phosphate-buffered saline [PBS], 1.1 M sucrose, 1 mM CaCl₂, 0.5 mM MgCl₂) in the presence of 100 µg of RNase (Sigma Chemicals, St. Louis, Mo.) per ml, 50 µg of DNase (Sigma) per ml, and a cocktail of antiproteases (Complete; Roche Molecular). S. aureus peptidoglycan was solubilized with 100 µg of lysostaphin (Sigma) per ml and L. lactis peptidoglycan with 5 mg of lysozyme (Sigma) per ml plus 100 µg of mutanolysin (Sigma) per ml. The digestion reactions were allowed to proceed for 30 min at 37°C in a shaking incubator at 120 rpm before the mixtures were chilled and nonsoluble material was removed by centrifugation for 30 min at 10,000 × g. The supernatants (containing the solubilized walls) were recovered, and sucrose was removed by overnight dialysis against PBS at 4°C. The dialyzed material was allowed to equilibrate at room temperature. SDS (1% final concentration) was added to the samples before they were concentrated 10-fold using an Ultrafree-15 centrifugal filter device (Millipore Corporation, Bedford, Mass.). Protein concentrations were then determined using the BCA Method (Pierce, Rockford, Ill.). Ten-microgram protein samples of each preparation were separated by electrophoresis and transferred overnight (at 4°C) onto Immobilon-P polyvinylidene difluoride (PVDF) membrane (Millipore) using a Mini-Trans-Blot electrophoretic transfer cell (Bio-Rad Laboratories, Hercules, Calif.) and a constant voltage of 35 V. The membranes were proceeded and incubated with a 1:500 dilution of anti-ClfA antibodies (kindly provided by T. Foster, Dublin, Ireland, and P. Vaudaux, Geneva, Switzerland) for 4 h as described elsewhere (18). After repeated rinsing in PBS- 0.2% Tween, immunoblots were incubated for 1 h with a 1:5,000 dilution of goat anti-rabbit antibodies coupled with alkaline phosphatase (Pierce). Bands were revealed by the BCIP (5-bromo-4-chloro-3-indol-1-phosphate-p-toluidine)-nitroblue tetrazolium method (AP Development Reagent; Bio-Rad).

Measurement of bacterial clumping and adherence to solid-phase fibrinogen. Functional expression of staphylococcal clumping factor was determined using previously described methods (19) modified as follows. Bacteria from overnight cultures were centrifuged, washed twice with ice-cold 0.9 M NaCl, and resuspended in a 1/10 volume of the same solution. (i) For qualitative cell clumping, 20 µl of rabbit plasma (BioMérieux, Marcy L'Etoile, France) was mixed with 20 µl of washed bacteria on a glass slide, and clumping was assessed visually by immediate bacterial aggregation on the slide. (ii) Cell clumping was measured quantitatively in microtiter trays. Serial twofold dilutions of a 1-mg/ml solution of fibrinogen (Sigma) were made in PBS, and 100 µl was incubated with 20 µl of washed bacteria at 4°C. The reciprocal of the highest dilution of fibrinogen showing clumping after 24 h was recorded as the titer. Adherence to solid-phase fibrinogen was tested using a previously described method (39).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of the E. coli-L. lactis shuttle-expression vectors. Two vectors (pOri23 and pOri59 in Table 1) were constructed using the following two-step strategy. First, the new shuttle vector pOri253 (Table 1 and Fig. 1B) was designed to replicate in both E. coli and L. lactis. The basic skeleton of pOri253 was the previously described L. lactis plasmid pIL253 (34) (Fig. 1A), carrying an ermAM macrolide-lincosamide-streptogramin resistance gene. ermAM can confer erythromycin resistance (Ery^r) to both lactococci and E. coli (16). However, pIL253 does not replicate autonomously in E. coli. To circumvent this problem, pIL253 was equipped with the high-copy-number oriColE1 replicon of pBluescript (Stratagene). oriColE1 was amplified by PCR using the primers presented in Table 2, and inserted at the XbaI restriction site of pIL253. The resulting shuttle vector pOri253 (Table 1 and Fig. 1B) could successfully replicate in both E. coli and L. lactis.

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Restriction map of the shuttle-expression vectors generated and utilized in this study (Table 1). The vectors were based on the original lactococcal plasmid pIL253 (A). This plasmid was previously reported as capable of replicating at high copy number and maintaining large pieces of DNA in lacotococci (34). pIL253 was used to construct the E. coli-L. lactis shuttle vector pOri253 (B) by inserting the OriColE1 replicon at the XbaI restriction site of its MCS. The shuttle vector pOri253 was further modified by inserting either of the lactococcal promoters P23 or P59 at the EcoRI/BamHI end of the MCS, resulting in the two lactococcal expression vectors pOri23 and pOri59 (C). Finally, these two shuttle expression vectors were used to subclone either the luciferase reporter gene or the staphylococcal clumping factor (clfA) gene to be expressed in lactococci (C). ermAM is the Eryr determinant of the original pIL253 plasmid, and repD and repE are the genes responsible for replication in lactococci.

Functional characterization of the expression vectors with luciferase. To select the most efficient promoter, we used a promoterless firefly luciferase (luc) determinant as a reporter gene. luc was amplified by PCR from pFW5-luc and cloned along with its ribosome-binding site downstream of either the P23 or P59 promoters, at the BamHI/PstI site of plasmids pOri23 and pOri59. The resulting pOri23-luc and pOri59-luc plasmids (Fig. 1D) were transformed into E. coli and L. lactis, and light emission was determined in the recipients.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Bacterial growth and luciferase expression of L. lactis carrying the shuttle-expression pOri23 equipped with a luciferase reporter gene. Bacterial growth is expressed by the culture colony counts (CFU/milliliter). Luciferase expression is reported in relative light units (R.L.U.), as given by the luminometer. Luciferase expression was proportional to the number of viable counts over the whole growth phase. In stationary phase, a slight decrease (ca. 20%) in light expression was observed as compared to colony counts. Each datum point represents the mean ± the standard deviation of at least three individual determinations.

Functional expression of the S. aureus clumping factor A gene in lactococci. The clfA gene of S. aureus Newman was inserted at the EcoRI/SalI site of pOri253, using a strategy similar to that described above for luc. The primers for clfA amplification are described in Table 2. The recombinant plasmid pOri23-clfA (Fig. 1D) was then transformed into L. lactis cremoris 1363, and the expression of the clfA product was determined both by Western blot and by bacterial clumping in the presence of rabbit plasma.

View larger version (52K): [in this window] [in a new window]   FIG. 3.   Western blots of cell wall extracts of S. aureus Newman (Newman), L. lactis carrying the empty vector (pLI253), and L. lactis carrying the shuttle-expression vector pOri23-clfA, containing a copy of the staphylococcal clfA gene [ClfA (+)]. Bacteria in the late exponential phase were recovered, and their cell walls were isolated and solubilized as described in Materials and Methods. Solubilized walls were subjected to SDS-PAGE separation, transferred to PVDF membranes, and probed with anti-ClfA F(ad')2 antibody fragments. For each sample, similar quantities (10 µg in 50 µl) of proteins were loaded on the gels. A band of ca. 130 kDa (arrow), corresponding the molecular mass of ClfA (18), appeared in wall extracts of both S. aureus Newman and L. lactis containing pOri23-clfA. In contrast, no band was observed in the wall extract of lactococci containing the empty vector. The double band in the S. aureus Newman extract might be due to interference of incompletely digested peptidoglycan moieties with migration in the electric field.

View larger version (53K): [in this window] [in a new window]   FIG. 4.   Binding of S. aureus Newman and ClfA-positive and negative lactococci to solid-phase fibrinogen. Microtiter plates were coated with serial twofold dilutions of a 1-mg/ml solution of fibrinogen (left to right), followed by incubation either with no cells (rows A) or with washed bacteria (rows B). Bound cells were then stained with crystal violet as described (39). Both S. aureus and ClfA-positive lactococci adhered to solid-phase fibrinogen in a similar way. In contrast, ClfA negative lactococci (L. lactis pIL253) did not.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Infection of host tissues by pathogenic microorganisms, such as S. aureus, is a multistep process beginning with the attachment of bacteria to specific-host-soluble or extracellular-matrix components. Cell-wall-associated factors responsible for this adhesion have long been considered as potential anti-infective targets and were studied in mutants of S. aureus defective in these specific derminants (1, 12, 22). However, this approach was limited because S. aureus harbors redundant surface adhesins that might overcome and compensate for the function of the missing or mutated factor.

To circumvent this problem, we sought to express S. aureus pathogenic factors in a less-virulent organism and eventually to test for a gain in infectivity related to this specific determinant. We hypothesized that if expression of a specific factor in lactococci resulted in a gain of function, e.g., increased infectivity, then this very factor is also likely to be important for staphylococcal pathogenesis, even though staphylococcal knockout mutants are not defective in infection models. Thus, we designed original molecular tools to express staphylococcal surface adhesins in lactococci. This recipient organism was chosen both because it is not known to carry adhesins to human tissues (6) and because it shares with S. aureus a common system to process and anchor cell-wall-associated proteins (32). Indeed, expression of heterologous surface proteins by lactococci was previously reported with other genes (29), thus supporting the rationale of the present approach. However, lactococci are not convenient organisms for "routine" molecular cloning and cannot easily be made competent for DNA transformation (10, 29). Therefore, a novel E. coli-L. lactis shuttle vector (pOri253 in Table 1) was designed, thus greatly facilitating the cloning of staphylococcal genes by using standard molecular techniques available for E. coli. The new shuttle vector was purposely based on the lactococcal plasmid pIL253, previously described as a high-copy-number lactococcal vector that could maintain large DNA fragments in this organism (34). Introduction of the oriColE1 replicon into pIL253 allowed this plasmid to replicate autonomously in E. coli. Moreover, the E. coli transformants could express the plasmid ermAM determinant and thus be selected for Ery^r (16).

An additional task in generating these molecular tools was to ensure gene expression. For this purpose, two previously described lactococcal promoters (P23 and P59) (36) were amplified and inserted upstream of the plasmid's MCS. Both promoters were previously used to overexpress streptococcal proteins in L. lactis (29). In the present experiments, both promoters drove gene expression, but with very different efficiencies. Indeed, the plasmid carrying promoter P23 (pOri23) expressed luciferase constitutively at a level up to 10,000 times greater than the plasmid containing promoter P59 (pOri59). In contrast, no luciferase activity was detected in E. coli, probably because the promoter sequences required for RNA polymerase recognition in this organism are different from those in lactococci (21, 24).

The fact that pOri23 ensured constitutive gene expression might have some advantages for the purpose of the present work. Indeed, in most bacteria pathogenic factors are tightly regulated using two-component regulatory systems (2). Factors are expressed when useful for the bacteria and downregulated when unnecessary or deleterious. Thus, the pathogenic contribution of a given factor might be mistakenly overemphasized or underestimated depending on whether the gene is expressed or not. Hence, expressing these factors constitutively in a surrogate bacterium might help discriminate between their intrinsic contribution when taken alone and the importance of their regulation in the staphylococcal context. Therefore, although not quite representative of the staphylococcal physiology, the present system should provide relevant complementary information.

A salient example of such a discrepancy was provided by recent experiments with S. aureus alpha-hemolysin (hla) mutants. Inactivation of hla decreased S. aureus infectivity in rabbits with experimental endocarditis. Paradoxically, however, overexpression of hla on a multicopy plasmid also decreased infectivity in the model. In this case, overexpression of hla was presumably deleterious to the bacterium via the microbicidal proteins released by alpha-hemolysin-activated platelets (3). Therefore, the same alpha-hemolysin could either increase or decrease infectivity depending on the experimental conditions.

Most interestingly, this apparent paradox finds its explanation in gene regulation. In wild-type S. aureus, hla is only expressed in the stationary phase but not during early bacterial growth (23). By that time, it is likely that bacteria are already embedded in the vegetation fibrin meshwork and remote from direct attack by platelet microbicidal proteins (PMPs). In contrast, organisms overexpressing alpha-hemolysin at an earlier stage of infection might be more directly exposed to the deleterious effect of PMPs (30). This emphasized the importance of obtaining results on constitutively expressed pathogenic genes, in order to understand better the implication of gene regulation in the infection process. Therefore, although not quite representative of the staphylococcal physiology, the present system should provide relevant complementary information.

The last requirement for establishing such a model in lactococci was to test whether the new pOri23 vector could indeed drive the expression of functional staphylococcal gene products in the L. lactis background. The S. aureus clumping factor A determinant was chosen as a model because it is a well-studied adhesin anchored to the staphylococcal wall (18). ClfA binds to fibrinogen and is responsible for bacterial clumping in the presence of plasma, thus providing a convenient in vitro functional assay (19). Moreover, ClfA was shown to mediate bacterial attachment to cardiac vegetations and significantly, yet marginally, contributed to virulence in rats with experimental endocarditis (22). Thus, ClfA provided (i) a test for the attachment of the protein to the cell wall, (ii) a method to assess its functionality, and (iii) a precedent for experimental infection.

The present results supported the two first issues, indicating that the ClfA protein could both be properly located in the cell wall fraction and mediate attachment to soluble and solid-phase fibrinogen. This will be important in assessing the gain of function conferred by ClfA and other genes in more sophisticated in vivo infectivity experiments, such as experimental endocarditis, that are planned for the future. Indeed, preliminary experiments indicated the validity of the concept. Thus, these new tools set the basis for a new strategy to study single staphylococcal pathogenic genes. It should help complement both classical knockout mutagenesis in staphylococci and modern means such as in vivo expression technology (15) and signature-tagged mutagenesis (20).