|Home | About | Journals | Submit | Contact Us | Français|
The Gram-positive pathogen Staphylococcus aureus secretes various proteins into its extracellular milieu. Bioinformatics analyses have indicated that most of these proteins are directed to the canonical Sec pathway, which consists of the translocation motor SecA and a membrane-embedded channel composed of the SecY, SecE, and SecG proteins. In addition, S. aureus contains an accessory Sec2 pathway involving the SecA2 and SecY2 proteins. Here, we have addressed the roles of the nonessential channel components SecG and SecY2 in the biogenesis of the extracellular proteome of S. aureus. The results show that SecG is of major importance for protein secretion by S. aureus. Specifically, the extracellular accumulation of nine abundant exoproteins and seven cell wall-bound proteins was significantly affected in an secG mutant. No secretion defects were detected for strains with a secY2 single mutation. However, deletion of secY2 exacerbated the secretion defects of secG mutants, affecting the extracellular accumulation of one additional exoprotein and one cell wall protein. Furthermore, an secG secY2 double mutant displayed a synthetic growth defect. This might relate to a slightly elevated expression of sraP, encoding the only known substrate for the Sec2 pathway, in cells lacking SecG. Additionally, the results suggest that SecY2 can interact with the Sec1 channel, which would be consistent with the presence of a single set of secE and secG genes in S. aureus.
Staphylococcus aureus is a well-represented component of the human microbiota as nasal carriage of this Gram-positive bacterium has been shown for 30 to 40% of the population (32). This organism can, however, turn into a dangerous pathogen that is able to infect almost every tissue in the human body. S. aureus has become particularly notorious for its high potential to develop resistance against commonly used antibiotics (20, 49). Accordingly, the S. aureus genome encodes an arsenal of virulence factors that can be expressed when needed at different stages of growth. These include surface proteins and invasins that are necessary for colonization of host tissues, surface-exposed factors for evasion of the immune system, exotoxins for the subversion of protective host barriers, and resistance proteins for protection against antimicrobial agents (37, 57).
Most proteinaceous virulence factors of S. aureus are synthesized as precursors with an N-terminal signal peptide to direct their transport from the cytoplasm across the membrane to an extracytoplasmic location, such as the cell wall or the extracellular milieu (38, 45). As shown for various Gram-positive bacteria, the signal peptides of S. aureus are generally longer and more hydrophobic than those of Gram-negative bacteria (38, 54). On the basis of signal peptide predictions using a variety of algorithms, it is believed that most exoproteins of S. aureus are exported to extracytoplasmic locations via the general secretory (Sec) pathway (38). This seems to involve precursor targeting to the Sec machinery via the signal recognition particle instead of the well-characterized proteobacterial chaperone SecB, which is absent from Gram-positive bacteria (16, 19, 53). The preproteins are then bound by the translocation motor protein SecA (38, 45). Through repeated cycles of ATP binding and hydrolysis, SecA pushes the protein in an unfolded state through the membrane-embedded SecYEG translocation channel (12, 30, 33, 52). Upon initiation of the translocation process, the proton motive force is thought to accelerate preprotein translocation through the Sec channel (26). Recently, the structure of the SecA/SecYEG complex from the Gram-negative bacterium Thermotoga maritima was solved at 4.5 Å resolution (58). In this structure, one SecA molecule is bound to one set of SecYEG channel proteins. The core of the Sec translocon consists of the SecA, SecY, and SecE proteins, which are essential for growth and viability of bacteria, such as Escherichia coli and Bacillus subtilis (6, 9, 22). In contrast, the channel component SecG is dispensable for growth, cell viability, and protein translocation (26, 48). Nevertheless, SecG does enhance the efficiency of preprotein translocation through the SecYE channel (26, 48). This is of particular relevance at low temperatures and in the absence of a proton motive force (17). Several studies suggest that E. coli SecG undergoes topology inversion during preprotein translocation (25, 27, 43). Even so, van der Sluis et al. reported that SecG cross-linked to SecY is fully functional despite its fixed topology (46). During or shortly after membrane translocation of a preprotein through the Sec channel, the signal peptide is removed by signal peptidase. This is a prerequisite for the release of the translocated protein from the membrane (1, 47).
Several pathogens, including Streptococcus gordonii, Streptococcus pneumoniae, Bacillus anthracis, Bacillus cereus, and S. aureus, contain a second set of chromosomal secA and secY genes named secA2 and secY2, respectively (39). Comparison of the amino acid sequences of the SecY1 and SecY2 proteins shows that their similarity is low (about 20% identity) and that the conserved regions are mainly restricted to the membrane-spanning domains. It has been shown for S. gordonii that the transport of at least one protein is dependent on the presence of SecA2 and SecY2. This protein, GspB, is a large cell surface glycoprotein that is involved in platelet binding (4). The protein contains an unusually long N-terminal signal peptide of 90 amino acids, large serine-rich repeats, and a C-terminal LPXTG motif for covalent cell wall binding. The gspB gene is located in a gene cluster with the secA2 and secY2 genes. Two other genes in this cluster encode the glycosylation proteins GftA and GftB, which seem to be necessary for stabilization of pre-GspB. Furthermore, the asp4 and asp5 genes in the secA2 secY2 gene cluster show similarity to secE and secG, and they are important for GspB export by S. gordonii (44). Despite this similarity, SecE and SecG cannot complement for the absence of Asp4 and Asp5, respectively. The secA2-secY2 gene cluster is also present in S. aureus, but homologues of the asp4 and asp5 genes are lacking. This seems to suggest that SecA2 and SecY2 of S. aureus share the SecE and SecG proteins with SecA1 and SecY1. The sraP gene in the secA2-secY2 gene cluster of S. aureus encodes a protein with features similar to those described for GspB. Siboo and colleagues (41) have shown that SraP is glycosylated and capable of binding to platelets. Importantly, the disruption of sraP resulted in a decreased ability to initiate infective endocarditis in a rabbit model. Consistent with the findings in S. gordonii, SraP export was shown to depend on SecA2/SecY2 (40). However, it has remained unclear whether other S. aureus proteins are also translocated across the membrane in an SecA2/SecY2-dependent manner.
The present studies were aimed at defining the roles of two Sec channel components, SecG and SecY2, in the biogenesis of the S. aureus exoproteome. The results show that secG and secY2 are not essential for growth and viability of S. aureus. While the absence of SecY2 by itself had no detectable effect, the absence of SecG had a profound impact on the composition of the exoproteome of S. aureus. Various extracellular proteins were present in decreased amounts in the growth medium of secG mutant strains, which is consistent with impaired Sec channel function. However, a few proteins were present in increased amounts. Furthermore, the absence of secG caused a serious decrease in the amounts of the cell wall-bound Sbi protein. Most notable, a secG secY2 double mutant strain displayed synthetic growth and secretion defects.
All strains used in this study are listed in Table Table1.1. Unless stated otherwise, E. coli strains were grown in Luria-Bertani broth (LB). S. aureus strains were grown at 37°C in tryptic soy broth (TSB) or B medium (1% peptone, 0.5% yeast extract, 0.5% NaCl, 0.1% K2HPO4, 0.1% glucose), under vigorous shaking, or on trypic soy agar (TSA) plates or B plates. If appropriate, medium for E. coli was supplemented with 100 μg/ml ampicillin or 100 μg/ml erythromycin, and medium for S. aureus was supplemented with 5 μg/ml erythromycin, 5 μg/ml tetracycline, or 20 μg/ml kanamycin. To monitor β-galactosidase activity in cells of E. coli and S. aureus, 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) was added to the plates at a final concentration of 80 μg/ml.
Mutants of S. aureus were constructed using the temperature-sensitive plasmid pMAD (2) and previously described procedures (23). Primers (Table (Table2)2) were designed using the genome sequence of S. aureus NCTC8325 (http://www.ncbi.nlm.nih.gov/nuccore/NC_007795). All mutant strains were checked by isolation of genomic DNA using a GenElute Bacterial Genomic DNA Kit (Sigma) and PCR with specific primers.
To delete the secG or secY2 gene, primer pairs with the designations F1/R1 and F2/R2 were used for PCR amplification of the respective upstream and downstream regions (each, ~500 bp) and their fusion with a 21-bp linker. The fused flanking regions were cloned in pMAD, and the resulting plasmids were used to delete the chromosomal secG or secY2 genes of S. aureus RN4220. To delete the secG or secY2 genes from the S. aureus SH1000 genome, the respective pMAD constructs were transferred from the RN4220 strain to the SH1000 strain by transduction with phage 85 (29).
To create the spa sbi double mutant of S. aureus Newman, the sbi gene was deleted from a spa mutant strain kindly provided by T. Foster (31). For this purpose, the kanamycin resistance marker encoded by pDG783 was introduced between the sbi flanking regions via PCR with the primer pairs sbi-F1/sbi-R1, sbi-F2/sbi-R2, and kan-F1/kan-R1. The obtained ~2,000-bp fragment was ligated into pMAD, and the resulting plasmid was used to transform competent S. aureus Newman spa cells. Blue colonies were selected on TSA plates with erythromycin and kanamycin, and the spa sbi double mutant was subsequently identified following the previously described protocol (23).
For complementation studies, the secG or secY2 gene was cloned into plasmid pCN51 (11). Expression of genes cloned in this plasmid is directed by a cadmium-inducible promoter. Primer pairs with the F3/R3 designation (Table (Table2)2) were used to amplify the secG or secY2 gene. These primers contain an EcoRI restriction site at the 5′ end and an SalI restriction site at the 3′ end of the amplified gene. PCR products were purified using a PCR Purification Kit (Roche) and ligated into a TOPO vector (Invitrogen). The resulting constructs were then cut with EcoRI and SalI, and the secG or secY2 gene (284 and 1,233 bp, respectively) was isolated from an agarose gel and ligated into pCN51 cut with EcoRI and SalI. This resulted in the secG- and secY2-pCN51 plasmids. Competent S. aureus RN4220 ΔsecG, ΔsecY2, or ΔsecG ΔsecY2 cells were transformed with these plasmids by electroporation, and colonies were selected on TSA plates containing erythromycin. The plasmids were then transferred to S. aureus SH1000 by transduction as described above.
Extracellular proteins from 100 ml of culture supernatant were precipitated, washed, dried, and resolved as described previously (56). The protein concentration was determined using Roti-Nanoquant (Carl Roth GmbH & Co., Karlsruhe, Germany). Preparative two-dimensional (2-D) PAGE was performed by using the immobilized pH gradient technique (5, 13). The protein samples (350 μg) were separated on immobilized pH gradient strips (Amersham Pharmacia Biotech, Piscataway, NJ) with a linear pH gradient from 3 to 10. The resulting protein gels were stained with colloidal Coomassie blue G-250G (10) and scanned with a light scanner. Each experiment was performed at least three times.
For identification of proteins by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS), Coomassie-stained protein spots were excised from gels using a spot cutter (Proteome Work) with a picker head of 2 mm and transferred into 96-well microtiter plates. Digestion with trypsin and subsequent spotting of peptide solutions onto the MALDI targets were performed automatically in an Ettan Spot Handling Workstation (GE-Healthcare, Little Chalfont, United Kingdom) using a modified standard protocol. MALDI-TOF MS analyses of spotted peptide solutions were carried out on a Proteome-Analyzer 4700/4800 (Applied Biosystems, Foster City, CA) as described previously (13). MALDI with tandem TOF (TOF-TOF) analysis was performed for the three highest peaks of the TOF spectrum as described previously (13, 51). Database searches were performed using the GPS explorer software, version 3.6 (build 329), with organism-specific databases.
By using the MASCOT search engine, version 184.108.40.206. (Matrix Science, London, United Kingdom), the combined MS and tandem MS (MS/MS) peak lists for each protein spot were searched against a database containing protein sequences derived from the genome sequences of S. aureus NCTC8325. Search parameters were as described previously (51). For comparison of protein spot volumes, the Delta 2D software package was used (Decodon GmbH Germany). The induction ratio of mutant to parental strain was calculated for each spot (normalized intensity of a spot on the mutant image/normalized intensity of the corresponding spot on the parental image). The significance of spot volume differences of 2-fold or higher was assessed by the a Student's t test (α of <0.05; Delta 2D statistics table).
Total RNA from S. aureus RN4220 was isolated using the acid-phenol method (14). Digoxigenin (DIG)-labeled RNA probes were prepared by in vitro transcription with T7 RNA polymerase using a DIG-RNA labeling mixture (Roche, Indianapolis, IN) and appropriate PCR fragments as templates. The PCR fragments were generated by using the respective oligonucleotides (Table (Table2)2) and chromosomal DNA of S. aureus RN4220 isolated with the chromosomal DNA isolation kit (Promega, Madison, WI) according to the manufacturer's recommendations. Reverse primers contain the T7 RNA polymerase recognition sequence at the 5′ end. Northern blot and slot blot analyses were performed as described previously (50, 57). Before hybridization, each RNA blot was stained with methylene blue in order to check the RNA amount blotted onto the membrane. Only blots showing equal amounts of 16S and 23S rRNAs for each sample loaded onto the respective gels were used for hybridization experiments. The hybridization signals of the Northern blots were detected with a Lumi-Imager (Roche, Indianapolis, IN) and analyzed with the software package LumiAnalyst (Roche, Indianapolis, IN). Slot blot signal detection was performed with the Intas ChemoCam system and analyzed with LabImage1D software (Intas Science Imaging Instruments GmbH, Göttingen, Germany). In slot blot experiments, the induction ratios were calculated by dividing the volumes obtained for the different RNA samples by the volume of the signals of the exponentially grown RN4220 parental strain. An internal RNA standard was spotted onto each membrane to correct for intermembrane variations.
Overnight cultures were diluted to an optical density at 540 nm (OD540) of 0.05 and grown in 25 ml of TSB under vigorous shaking. For complementation of mutant strains with pCN51-based plasmids, CdSO4 was added after 3 h of growth to a final concentration of 0.25 μM. Samples were taken after 6 h of growth and separated into growth medium and whole-cell and noncovalently cell wall-bound protein fractions. Cells were separated from the growth medium by centrifugation of 1 ml of the culture. The proteins in the growth medium were precipitated with 250 μl of 50% trichloroacetic acid (TCA), washed with acetone, and dissolved in 100 μl of loading buffer (Invitrogen). Cells were resuspended in 300 μl of loading buffer (Invitrogen) and disrupted with glass beads using a Precellys 24 bead-beating homogenizer (Bertin Technologies). From the same culture 20 ml was used for the extraction of noncovalently bound cell wall proteins using KSCN. Cells were collected by centrifugation, washed with PBS, and incubated for 10 min with 1 M KSCN on ice. After centrifugation the noncovalently cell wall-bound proteins were precipitated from the supernatant fraction with TCA, washed with acetone, and dissolved in 100 μl of loading buffer (Invitrogen). Upon addition of reducing agent (Invitrogen), the samples were incubated at 95°C. Proteins were separated by SDS-PAGE using precast NuPage gels (Invitrogen) and subsequently blotted onto a nitrocellulose membrane (Protran; Schleicher and Schuell). The presence of a cytoplasmic marker protein (TrxA), a lipoprotein (DsbA), and several cell wall-associated proteins (Sle1, Aly, ClfA, and IsaA) or extracellular proteins (Sle1, Aly, IsaA, and SspB) was monitored by immunodetection with specific polyclonal antibodies raised in mice or rabbits. Bound primary antibodies were visualized using fluorescent IgG secondary antibodies (IRDye 800 CW goat anti-mouse/anti-rabbit; LiCor Biosciences). Membranes were scanned for fluorescence at 800 nm using an Odyssey Infrared Imaging System (LiCor Biosciences).
All animal studies were approved by the Animal Care and Experimentation Committee of the district government of Lower Franconia, Germany, and conformed to University of Würzburg guidelines. Female BALB/c mice (16 to 18 g; Charles River, Sulzfeld, Germany) were housed in polypropylene cages and received food and water ad libitum. S. aureus isolates were cultured for 18 h in B medium, washed three times with sterile phosphate-buffered saline (PBS), and suspended in sterile PBS to 1.0 × 108 CFU/100 μl. As a control, selected dilutions were plated on B agar. Mice were inoculated with 100 μl of S. aureus via the tail vein. Control mice were treated with sterile PBS. For each strain, eight mice were used. Three days after challenge, kidneys and livers were aseptically harvested and homogenized in 3 ml of PBS using Dispomix (Bio-Budget Technologies Gmbh, Krefeld, Germany). Serial dilutions of the organ homogenates were cultured on mannitol salt-phenol red agar plates for at least 48 h at 37°C. The numbers of CFU were calculated as the number of CFU/organ. The statistical significance of bacterial load was determined using Mann-Whitney tests.
To investigate the roles of SecG and SecY2 in the biogenesis of the S. aureus exoproteome, the respective genes were completely deleted from the chromosome of S. aureus strain RN4220. This resulted in the single mutant ΔsecG and ΔsecY2 strains and the double mutant ΔsecG ΔsecY2. Next, cells of these mutants were grown in TSB medium until they reached the stationary phase (Fig. (Fig.1;1; not shown for the ΔsecY2 strain). All three mutants displayed exponential growth rates similar to the rate of the parental strain. However, the secG secY2 double mutant entered the stationary phase at a lower optical density (OD540 of 8) than the parental strain and the ΔsecG mutant (OD540 of 15). Since the amounts of most exoproteins of S. aureus increase mainly in the stationary growth phase at high cell densities (37, 56), extracellular proteins for 2-D PAGE analyses were collected from the supernatant of cell cultures that had reached stationary phase (Fig. (Fig.11 and and2).2). Comparison of the exoproteomes of the secG mutant and its parental strain revealed that 11 proteins with Sec-type signal peptides and type I signal peptidase cleavage sites (i.e., SAOUHSC_00094, SdrD, Sle1, Geh, Hlb, HlY, HlgB, HlgC, Plc, SAOUHSC_02241, and SAOUHSC_02979) were present in significantly decreased amounts when SecG was absent from the cells. This was also true for the secreted moiety of the polytopic membrane protein YfnI, which is processed by signal peptidase I as was previously shown for the YfnI homologue of B. subtilis (1). In contrast, the amounts of three other exoproteins (i.e., IsaA, Spa, and SsaA) were considerably increased due to the secG deletion (Fig. (Fig.2A;2A; Table Table3).3). These effects of the secG mutation were fully compensated when secG was ectopically expressed from plasmid secG-pCN51 (Fig. (Fig.2C).2C). Northern blot analyses revealed similar transcript levels for geh, hlb, and spa in the secG mutant and the parental strain RN4220. This shows that the changes in the amounts of the respective exoproteins in the secG mutant were not caused by decreased transcription of the corresponding genes (Fig. (Fig.33).
Deletion of the secY2 gene encoding a channel component of the accessory Sec system in S. aureus did not affect the extracellular protein pattern (data not shown). However, the deletion of both secG and secY2 caused additional changes in the extracellular proteome compared to the secG single mutant (Fig. (Fig.2B).2B). Specifically, one additional exoprotein was identified in decreased amounts (i.e., LipA), and one additional exoprotein (i.e., LytM) was identified in increased amounts (Table (Table3).3). Furthermore, proteins such as IsaA, Spa, and SsaA were secreted in larger amounts not only by the secG mutant but also by the secG secY2 double mutant. This effect was significantly exacerbated for IsaA and SsaA in the secG secY2 double mutant. It is interesting that IsaA, LytM, Spa, and SsaA represent cell surface-associated proteins (34, 37, 42). In contrast, most proteins that were secreted in reduced amounts in the secG or secG secY2 mutant are secretory proteins without retention signals, except for SAOUHSC_00094 (Table (Table3).3). Importantly, the secretion and growth defects of the secG secY2 mutant strain could also be fully reversed by ectopic expression of secG from plasmid secG-pCN51, and the synthetic effects of the secG and secY2 mutations could be reversed by plasmid secY2-pCN51 (data not shown).
To test whether the synthetic effects of the secG secY2 double mutation might relate to jamming of the SecYE translocation channel by SraP, the only known substrate for the Sec2 pathway, we tried to construct an secG secY2 sraP triple mutant. Unfortunately, despite several attempts, we did not manage to obtain this triple mutant for reasons that have so far remained obscure. To obtain further insights into the expression of sraP under the conditions tested, we performed Northern blotting and slot blot experiments with RNA extracted from the secG single mutant, the secG secY2 double mutant, and the parental strain RN4220. These experiments revealed that sraP expression is highest in the transient phase between the exponential and stationary growth phases (Fig. (Fig.4).4). Furthermore, the deletion of secG reproducibly triggered a 2-fold elevation in the sraP transcript level during the transient phase. This moderate but reproducible effect was observed both in the secG single mutant and in the secG secY2 double mutant, which argues to some extent against the possible jamming of SecYE by SraP, at least when SecY2 is still present in the cells.
Western blotting experiments were performed to investigate whether particular protein export defects of the secG and secY2 mutants had remained unnoticed in the proteomic analyses. These analyses included secreted proteins in the growth medium (Sle1, Aly, IsaA, and SspB), noncovalently attached cell wall proteins (Sle1, Aly, and IsaA), a covalently attached cell wall protein (ClfA), a lipoprotein (DsbA), and a cytoplasmic marker protein (TrxA) in S. aureus strains RN4220 and S. aureus SH1000. For most tested proteins no differences were detectable between the secG and/or secY2 mutant strains and their parental strain. However, these analyses showed that a band of ~50 kDa, which was cross-reactive with all tested sera, had disappeared from the fraction of noncovalently bound cell wall proteins of the secG mutant. It is known that proteins, such as protein A (36) and Sbi (55), have IgG-binding properties. To investigate whether the missing band would relate to protein A or Sbi, protein fractions from an spa mutant and an spa sbi double mutant were included in the Western blotting analyses. As shown in Fig. Fig.5A,5A, the band of ~50 kDa that was missing from the noncovalently bound cell wall proteins in the secG mutant was also missing from these proteins in the spa sbi double mutant but not from those in the spa single mutant (only the results for S. aureus SH1000 are shown, but essentially the same results were obtained for S. aureus RN4220). Taken together, these findings show that Sbi is noncovalently bound to the cell wall of S. aureus strains RN4220 and SH1000 and that SecG is required for export of Sbi from the cytoplasm to the cell wall. As was the case for the secreted S. aureus proteins detected by proteomics, Sbi export to the cell wall was not affected by the absence of SecY2 (Fig. (Fig.5B).5B). Finally, it is noteworthy that Sbi is detectable only among the noncovalently bound cell wall proteins of S. aureus strains RN4220 and SH1000, whereas it is detectable both in a cell wall-bound and a secreted state in S. aureus Newman.
To test whether the deletion of secG and/or secY2 would affect the virulence of S. aureus SH1000, a mouse infection model was used. The results revealed no significant differences in virulence of the ΔsecG, ΔsecY2, or ΔsecG ΔsecY2 strains compared to the parental strain SH1000 (Fig. (Fig.6).6). This shows that SecG and SecY2 have no important roles in the virulence of strain SH1000 in the context of the mouse infection model used.
The extracellular and surface-associated proteins of bacterial pathogens, such as S. aureus, represent an important reservoir of virulence factors (38, 39, 57). Accordingly, protein export mechanisms will contribute to the virulence of these organisms. While protein export has been well characterized in model organisms, such as E. coli and B. subtilis, relatively few functional studies have addressed the protein export pathways of S. aureus. Notably, the Sec pathway is generally regarded as the main pathway for protein export, but, to date, this has not been verified experimentally in S. aureus. Therefore, the present studies were aimed at assessing the role of the Sec pathway in establishing the extracellular proteome of S. aureus. We focused attention on the nonessential channel component SecG as this allowed a facile coassessment of the nonessential accessory Sec channel component SecY2. Our results show that the extracellular accumulation of proteins is affected to different extents by the absence of SecG: some proteins are present in reduced amounts, some are not affected, and some are present in elevated amounts. Furthermore, the effects of the absence of SecG are exacerbated by deletion of SecY2, suggesting that SecY2 directly or indirectly influences the functionality of the general Sec pathway. This is all the more remarkable since the absence of SecY2 by itself had no detectable effects on the composition of the extracellular proteome of S. aureus.
The observation that the secretion of a wide range of proteins was affected by the absence of SecG is consistent with the fact that all of these proteins contain Sec-type signal peptides. On the other hand, this finding is remarkable since studies of other organisms, such as E. coli (26) and B. subtilis (48), have shown that deletion of secG had fairly moderate effects on protein secretion in vivo. In B. subtilis, a phenotype of the secG mutation was observed only under conditions of high overproduction of secretory proteins (48). Clearly, our present data show that SecG is more important for Sec-dependent protein secretion in S. aureus than in B. subtilis or E. coli. Importantly, the transcription of genes for three proteins (Geh, Hlb, and Spa) that were affected in major ways by the absence of SecG was not changed, and all observed effects of the secG mutation could be reversed by ectopic expression of secG. This suggests that the observed changes in the exoproteome composition of the S. aureus secG mutant strain relate to changes in the translocation efficiency of proteins through the Sec channel rather than to regulatory responses at the gene expression level. This could be due to altered recognition of the respective signal peptides or mature proteins by the SecG-less Sec channel or combinations thereof. However, some indirect effects, for example, at the level of translation of exported proteins, posttranslocational folding, proteolysis, or cell wall binding of proteins like IsaA, LytM, Spa, and SsaA, can currently not be excluded, especially since no proteins were found to accumulate inside the secG mutant cells (data not shown). It remains to be shown why the extracellular accumulation of particular proteins is affected by the absence of SecG while that of other proteins remains unaffected.
Unexpectedly, our studies revealed that export of the IgG-binding protein Sbi to the cell wall was almost completely blocked in secG mutant strains. The reason that this export defect was not detected by 2-D PAGE relates to the fact that Sbi is predominantly cell wall bound in the tested S. aureus strains under the experimental conditions used. It has been proposed previously that Sbi would remain cell wall attached through a proline-rich wall-binding domain and electrostatic interactions (55). Nevertheless, Burman and colleagues showed that Sbi is extracellular, and they suggested that cell surface-bound Sbi might be disadvantageous for the bacterium due to its role in modulating the complement system (8). On the other hand, cell surface localization of Sbi would be appropriate for interference with the adaptive immune system through IgG binding (3). Irrespective of these previously reported findings, our Western blotting analyses show that Sbi is noncovalently bound to the cell wall, not only in S. aureus SH1000 and S. aureus RN4220 but also in S. aureus Newman. However, consistent with the findings of Burman et al., Sbi was also detected in the growth medium of S. aureus Newman, which indicates that the location of Sbi in the cell wall or extracellular milieu may differ for different S. aureus strains. In case of the Newman strain, the release of Sbi into the growth medium could be due to the fact that this strain produces Sbi and several other cell wall-bound proteins at increased levels compared to the RN4220 and SH1000 strains (35). Conceivably, this increased production of wall-bound proteins might lead to a saturation of available cell wall binding sites for Sbi.
Remarkably, the absence of SecG was shown to impact the relative amounts of various extracellular proteins while effects of the absence of SecG were detected for only one cell wall-associated protein, namely, Sbi. We do not believe that these differences in the numbers of identified proteins relate to the method that was used to monitor effects of the absence of SecG. Specifically, the analysis of proteins secreted by secG mutant strains via regular one-dimensional (1-D) SDS-PAGE already revealed major differences in the composition of the exoproteome (data not shown). It was for this reason that we initiated our 2-D PAGE analyses to identify the affected proteins. On the other hand, a 1-D SDS-PAGE analysis of cell wall-associated proteins did not reveal any major differences, and this was in fact the reason why we investigated potentially wall-associated proteins by Western blotting. Furthermore, we have no evidence from the different studies that we performed that the time point at which the sampling was done during the stationary phase had any major influence on the outcomes of our analyses.
Many of the proteins for which the absence of SecG affects extracellular amounts are considered to be important virulence factors of S. aureus. These proteins are involved in host colonization (e.g., the serine-aspartic acid repeat proteins SdrC and SdrD), invasion of host tissues (e.g., hemolysins and leukocidins), cell wall turnover (LytM), and evasion of the immune system (Spa and Sbi). The altered amounts of these proteins suggest that S. aureus strains depleted of SecG might perhaps be less virulent. However, in the mouse infection model used, no changes in virulence of the S. aureus SH1000 secG, secY2, or secG secY2 mutant strains could be detected. This implies that the presence or absence of SecG or SecY2 is not critical for the virulence of S. aureus SH1000, at least under the conditions tested in the mouse infection model used. Clearly, this does not rule out the possibility that such mutants are attenuated in virulence in other infection models that have not yet been tested.
Since we were unable to detect secretion defects for secY2 single mutant strains, our studies confirm that only very few proteins are translocated across the membrane in an SecA2/SecY2-dependent manner, as has previously been suggested by Siboo et al. (40). Furthermore, we did not detect differences in the export of glycosylated proteins by the secY2 mutants (data not shown), which is in line with the suggestion that glycosylated proteins are not strictly dependent on the accessory Sec pathway for export (40). It was therefore quite surprising that the secY2 mutation exacerbated the secretion defect of the S. aureus secG mutant. In fact, the secretion of two additional proteins was found to be affected in the secG secY2 double mutant. Moreover, a synthetic growth defect was observed for this double mutant. At this stage, it is possible that both the growth defect and the secretion defects are consequences of an impaired Sec channel function. In addition, the exacerbated secretion defects may relate to SecYE jamming by SraP, which is the only known SecA2/SecY2 substrate. As shown by Northern blotting analyses, the deletion of secG somehow triggers a 2-fold elevation in sraP transcript level during the transition between exponential and postexponential growth in not only the secG single mutant but also the secG secY2 double mutant. This argues to some extent against the jamming of SecYE by SraP, at least when SecY2 is still present in the cells. In the absence of SecG and SecY2, indeed, jamming of SecYE by the overexpressed SraP may occur in the transient phase. On the other hand, sraP expression seems relatively low in the stationary phase during which we harvested the extracellular proteins for proteomics analyses, which would suggest that any jamming effects of SraP are relatively slight in this growth phase. Unfortunately, we have so far not been able to assess the possibility of SecYE jamming by SraP directly because we were unable to obtain an secG secY2 sraP triple mutant. Notably, it is also possible that the exacerbated secretion defects are, to some extent, a secondary consequence of the growth defect of the double mutant. Irrespective of their primary cause, these synthetic effects of the secG and secY2 mutations suggest that the regular Sec channel can somehow interact with the Sec2 channel. Whether this means that mixed Sec channels with both SecY and SecY2 exist remains to be determined. However, this possibility would be consistent with the observation that S. aureus lacks a second set of secE and secG genes. It would thus be important to focus future research activities in this area on possible interactions between the regular Sec channel components and SecY2.
We thank W. Baas and M. ten Brinke for technical assistance, S. Dubrac for providing the pCN51 plasmid, T. Foster for the spa mutant of S. aureus Newman, I. Siboo and P. Sullam for advice, Decodon GmbH (Greifswald, Germany) for providing Delta2D software, and T. Msadek and other colleagues from the StaphDynamics and AntiStaph programs for advice and stimulating discussions.
M.J.J.B.S., T.W., M.M.V.D.K.-P., T.B., T.S., K.O., M.H., H.A., S.E., and J.M.V.D. were in part supported by the CEU projects LSHM-CT-2006-019064, LSHG-CT-2006-037469, and PITN-GA-2008-215524, the Top Institute Pharma project T4-213, and DFG research grants GK840/3-00, SFB/TR34, and FOR585.
Published ahead of print on 14 May 2010.