Comparative Proteomics-Based Identification of Genes Associated with Glycopeptide Resistance in Clinically Derived Heterogeneous Vancomycin-Intermediate Staphylococcus aureus Strains

Heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) is associated with clinical treatment failure. However, the resistance mechanism of hVISA has not been fully clarified. In the present study, comparative proteomics analysis of two pairs of isogenic vancomycin-susceptible S. aureus (VSSA) and hVISA strains isolated from two patients identified five differentially expressed proteins, IsaA, MsrA2, Asp23, GpmA, and AhpC, present in both isolate pairs. All the proteins were up-regulated in the hVISA strains. These proteins were analyzed in six pairs of isogenic VSSA and hVISA strains, and unrelated VSSA (n = 30) and hVISA (n = 24) by real-time quantitative reverse transcriptase–PCR (qRT–PCR). Of the six pairs of isogenic strains, isaA, msrA2 and ahpC were up-regulated in all six hVISA strains; whereas asp23 and gpmA were up-regulated in five hVISA strains compared with the VSSA parental strains. In the unrelated strains, statistical analyses showed that only isaA was significantly up-regulated in the hVISA strains. Analysis of the five differentially expressed proteins in 15 pairs of persistent VSSA strains by qRT–PCR showed no differences in the expression of the five genes among the persistent strains, suggesting that these genes are not associated with persistence infection. Our results indicate that increased expression of isaA may be related to hVISA resistance.


Introduction
Staphylococcus aureus can cause serious hospital-and communityacquired infections, including skin and soft tissue infections, pneumonia, bacteremia, endocarditis, and even septic shock. The high prevalence of methicillin-resistant S. aureus (MRSA) and the extensive use of vancomycin have led to the emergence of reduced vancomycin susceptibility among S. aureus strains. Although vancomycin-resistant S. aureus (VRSA) strains are rare, hVISA/VISA are common in the clinical setting, especially in persistent MRSA bacteremia and endocarditis. Our previous studies have shown that the prevalence of hVISA is 13% to 16% in large teaching hospitals in China [1]. Moreover, several studies have indicated that hVISA/VISA infections are associated with vancomycin treatment failure [2,3].
To date, no specific genetic determinants of hVISA/VISA have been universally defined, whereas VRSA strains acquire the vanA gene from Enterococcus. Several phenotypic features are characteristic of hVISA/VISA strains, among which significant cell wall thickening is a common feature associated with vancomycin resistance [4]. Compared with vancomycin-susceptible S. aureus (VSSA), hVISA produces three to five times the amount of penicillin-binding proteins (PBPs) 2 and 2'. The amounts of intracellular murein monomer precursor in hVISA are three to eight times greater than those in VSSA strains [4]. Factors such as the increased synthesis of non-amidated muropeptides and the resultant reduced peptidoglycan cross-linking contribute to the vancomycin resistance of VISA through increased affinity trapping of vancomycin [5]. In addition to thickened cell walls, hVISA/ VISA strains exhibit other phenotypic changes, including reduction in autolytic activity [6], reduced growth rate [7], resistance to lysostaphin [8], PBP changes [9], and metabolic changes [10].
Several transcriptional changes have been detected in hVISA/ VISA. DNA microarray analyses have been used to determine changes in the transcriptional profile of hVISA or VISA strains [11][12][13][14][15]. However, the protein profiles of hVISA or VISA are rarely analyzed via comparative proteomics. In a proteomics study that compared VSSA and VISA strains, several differentially expressed proteins were identified [16]. Another study identified 65 significant protein abundance changes by comparing three isogenic strains derived from a clinical VISA isolate [17].
To our knowledge, a comparative proteomics analysis of hVISA strains has not been performed to date. Here, we used comparative proteomics to analyze hVISA and VSSA strains isolated from the same patients treated with vancomycin. The differentially expressed proteins identified in our screen were validated in six pairs of isogenic hVISA and VSSA strains and unrelated hVISA (n = 24) and VSSA (n = 30) strains to identify the potential resistance mechanisms of hVISA. To further analyze the potential association of these differentially expressed genes with persistent infection, their expression was examined in 15 pairs of persistent VSSA strains. The results of our study provide insight into the molecular mechanisms underlying hVISA resistance.

Materials and Methods
The study protocol and written informed consent were approved by the Medical Ethical Committee of Peking University People's Hospital. Written informed consent was obtained from all patients at the time of enrollment.

Bacterial Isolates
A clinical VSSA (CN9) strain with a vancomycin MIC of 0.5 mg/mL and teicoplanin MIC of 2 mg/mL, and hVISA (CN10) with a vancomycin MIC of 2 mg/mL and teicoplanin MIC of 8 mg/mL were isolated in 2008 from the purulent sputum of an 84-year-old man who had MRSA pneumonia with a 20-year history of emphysema. Strain CN9 (VSSA), isolated after approximately 1 year of hospitalization, was the parental (pretherapy) vancomycin-susceptible isolate; strain CN10 was the hVISA organism recovered during vancomycin treatment. Another clinical VSSA (CN3) strain with vancomycin MIC of 0.5 mg/mL and teicoplanin MIC of 1 mg/mL, and hVISA (CN4) with vancomycin MIC of 0.5 mg/mL and teicoplanin MIC of 2 mg/mL were isolated in 2008 from a 90-year-old woman who had MRSA bacteremia with a 15-year history of diabetes mellitus. Strain CN3 (VSSA), isolated after approximately 1 month of hospitalization, was the parental (pretherapy) vancomycin-susceptible isolate; strain CN4 was the hVISA organism recovered during vancomycin treatment. The above two pairs of VSSA and hVISA strains were selected for comparative proteomics analysis.
Six pairs of VSSA and hVISA strains, and unrelated 24 hVISA and 30 VSSA strains from our previous study were selected to validate the proteomics results [1]. Fifteen pairs of persistent VSSA strains were selected to determine whether the differentially expressed genes were associated with persistent infection. Pairs of bacterial strains with the same genetic background were isolated from the same patient (Table 1). All hVISA strains were confirmed by the population analysis profile (PAP)-area under the curve (AUC) (PAP-AUC) method. All the strains used in the study were tested from frozen stocks and were not frozen and thawed multiple times. Each strain was stored at 280uC in three separate tubes.

PAP-AUC Method
PAP-AUC was determined as described previously [1]. Briefly, 50 mL of a 0.5 McFarland standard suspension at dilutions of 10 23 and 10 26 was inoculated onto brain heart infusion (BHI) agar plates containing 0, 0.5, 1.0, 2.0, 2.5, 4.0, and 8.0 mg/mL of vancomycin. After 48 h of incubation at 35uC, the colonies were counted and the log CFU/mL was plotted against vancomycin concentration. The ratio of the AUC of the test isolate to the AUC of S. aureus Mu3 was calculated and interpreted as follows: for VSSA, a ratio of ,0.9; for hVISA, a ratio of 0.9 to 1.3; and for VISA, a ratio of $1.3. S. aureus ATCC 29213 was used as the reference VSSA strain.

Molecular Typing Methods
All isolates were analyzed by SCCmec typing, spa typing, MLST typing, and PFGE. The SCCmec types were determined by the multiplex PCR strategy developed by Kondo et al. [18]. The spa typing was performed as described previously [19]. Purified spa PCR products were sequenced, and short sequence repeats were assigned by using the spa database website (http://www.ridom. de/spaserver). MLST was carried out as described previously [20]. The sequences of the PCR products were compared with the existing sequences available on the MLST website (http://saureus. mlst.net) for S. aureus. DNA extraction and SmaI restriction were performed as described previously [21]. The PFGE patterns were visually examined and interpreted according to the criteria of Tenover et al. [22].

Protein Sample Preparation
Overnight cultures of VSSA and hVISA strains were diluted at 1/100 in BHI broth and harvested at similar culture densities (exponential phase, OD 600 nm = 0.5). The samples were centrifuged at 7,000 g for 10 min to collect the deposits. The deposits were then washed in 50 mM PBS three times and incubated in 220 mL of 20 mM Tris-HCl, pH 7.5; 50 mL of 1 mg/ mL lysostaphin; 4 mL of protease inhibitor cocktail; and 6 mL of DNase for 30 min at 37uC. Subsequently, 1.5 mL of 2D lysis buffer (100 mL acetone, 20 mM DTT, 10%TCA) was added, and the samples were vortexed and frozen at -20uC for 2 h. Samples were centrifuged at maximum speed in a microcentrifuge for 2 min to remove insoluble materials, and protein was quantitated using the 2D Quant Kit (GE Healthcare, Arizona, USA).

Two-Dimensional Gel Electrophoresis (2DE)
2DE was performed as described previously [17]. Samples were run in triplicate. In the first dimension, 500 mg of protein was run on 24 cm Immobiline DryStrips (GE Healthcare) at a pH range of 4 to 7 on an IPGphorII IEF system (GE Healthcare) as recommended by the manufacturer. Strips were equilibrated in equilibration buffer (50 mM Tris-Cl, pH 8.8, 6 M urea, 30% glycerol, 2% SDS, and 0.25% trace of bromophenol blue) containing 10 mg/mL of DTT for 15 min followed by incubation in the same buffer containing 40 mg/mL of iodoacetamide for 15 min. The strips were then applied to 12.5% self-made acrylamide gels using 0.5% agarose in standard Tris-glycine electrophoresis buffer.
Second-dimension sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed in a Protean II (Bio-Rad, Hercules, CA, USA) at 40 mA/gel and 15uC until the tracking dye ran-off the gel. Proteins were visualized by Coomassie brilliant blue (CBB) staining. Gels were fixed in 20% TCA for 1 h, stained in 0.1% CBB for 2 h, destained in 40% ethanol and 10% acetic acid for 2630 min, intensified overnight in 1% acetic acid, and then washed in deionized water for 30 min. Gels were imaged using ImageScanner (GE Healthcare) and images were analyzed by PDQuest 6.0.

Mass Spectrometry for 2D Gel Protein Identification
Gel plugs containing differentially expressed proteins were excised using a ProXcision robot (Perkin Elmer Inc., Wellesley, MA, US) and subjected to matrix-assisted laser desorption ionization-time of flight/time of flight (MALDI-TOF/TOF) analysis (Bruker Daltonics, Leipzig, GER). Gel plugs were placed in 96-well plates, and then washed with 25 mM NH 4 HCO 3 (pH 8.0). Gel plugs were pre-frozen at -80uC for 1 h, and then digested with trypsin (Promega, WI, USA). After extraction from the gel into 50% acetonitrile/0.1% formic acid, peptides were lyophilized in a speed vacuum and resuspended in 10 mL of 0.1% formic acid solution. Peptide MS/MS spectra were obtained by MALDI-TOF/TOF (Bruker Daltonics). The resulting MS/MS spectra were interpreted using Mascot and searched against eubacterial proteins in the National Center for Biotechnology Information protein database. Results showing MASCOT score $75 and confidence level $95% were considered reliable [23].

Real-Time Quantitative Reverse Transcriptase-PCR (qRT-PCR)
Total RNA was extracted from each sample using the RNeasy Kit (Qiagen, CA, USA). Complementary DNA (cDNA) was generated from total RNA using a random primer hexamer. Gene-specific primers were designed using Primer Express 3.0 (Applied Biosystems) and are shown in Table 2. Samples were run in triplicate and quantified by qRT-PCR following the protocol for SYBR H Premix Ex Taq TM II (TaKaRa, Tokyo, Japan). The mixture was incubated at 95uC for 30 s, and then cycled at 95uC for 5 s and at 60uC for 20 s 40 times using the LightCyclerH 480 (Roche, Mannheim, GER). Amplification efficiencies were validated and normalized to the expression of the 16 S rRNA gene as a standard. The quantities of the target and standard genes were calculated according to a standard curve.

Genotypic Characterization of the Isolate Set
Two pairs of isogenic VSSA and hVISA isolates that belonged to SCCmecIII-ST239-spa t030 were selected for the comparative proteomics analysis to minimize variation unrelated to the vancomycin resistance phenotype. Six pairs of isogenic VSSA and hVISA strains, and unrelated VSSA (n = 30) and hVISA (n = 24) strains were selected to validate the differentially expressed genes identified by comparative proteomics screening. Fifteen pairs of persistent VSSA isolates were selected to determine whether the differentially expressed genes were associated with persistent infection.

Relative Expression of the five Differentially Expressed Proteins in Clinical VSSA and hVISA Isolates
The expression of the five differentially expressed proteins was assessed in six pairs of isogenic VSSA and hVISA strains by qRT-PCR to validate the accuracy of the results of comparative proteomics. The results showed that isaA, msrA2, and ahpC were up-regulated in all six hVISA strains, whereas asp23 and gpmA were up-regulated in five hVISA strains compared with the VSSA parental strain. The asp23 and gpmA genes were not up-regulated in the CN2/CN1 pair (Table 4). Statistical analysis showed that the expression of these genes, except for asp23, was significantly up-regulated in the hVISA strains. The expression of these five genes was also evaluated in unrelated VSSA (n = 30) and hVISA (n = 24) strains by qRT-PCR, which showed that only isaA was significantly up-regulated in the hVISA strains ( Figure 1).
To determine whether the differentially expressed genes were associated with persistent infection, their expression was assessed in 15 pairs of persistent VSSA strains by qRT-PCR. The results showed no significant differences in the expression level of the five genes among the 15 pairs of persistent VSSA strains (Table 5).

Discussion
Although the clinical importance of hVISA strains has been well-established, the resistance mechanism of hVISA remains unclear. In the present study, the potential mechanism of low-level vancomycin resistance was assessed in two pairs of isogenic VSSA and hVISA isolates by comparative proteomics analysis, which identified five differentially expressed proteins that were upregulated in both hVISA strains (Table 3).
Among the identified proteins, AphC, Asp23, and MsrA2 are involved in defense mechanisms. AhpC directly reduces organic hydroperoxides to their dithiol forms [24]. The hVISA/VISA strains showed thickened cell walls, increased peptidoglycan crosslinking, and a high positive charge [25], which could have caused changes in oxidation and osmotic pressure inside and outside the cell, and further induced AhpC expression. The stress response gene asp23, which encodes the Asp23 protein, is a possible target gene of the key global regulator, SigB [26]. Asp23 has a key role in the alkaline pH tolerance of S. aureus [27]. A previous microarray-based study also showed that Asp23 is upregulated in the hVISA strain [28]. The results of our comparative proteomics analysis showed that MsrA2 was enhanced in both hVISA strains. MsrA2, which catalyzes the reversible oxidationreduction of methionine sulfoxide to methionine, has a key function as a repair enzyme for proteins inactivated by oxidation. S. aureus possesses three MsrA enzymes (MsrA1, MsrA2, MsrA3) [29]. The msrA gene is a highly induced member gene of the cell wall stress stimulon (CWSS), which can be induced by cell wallactive antibiotics, such as oxacillin and vancomycin. The upregulation of msrA can lead to an increased rate of peptidoglycan biosynthesis, which results in cell wall thickening [11]. In addition, Msr proteins regulate virulence in several bacteria [29,30]. In a cDNA microarray study [11], msrA2 was over-expressed in VISA strains, which coincided with our results. The study also demonstrated that msrA2 contributes to vancomycin resistance by gene knockout and trans-complementation assay [11]. In addition, cell morphology experiments showed that msrA2 overexpression increases the cell wall thickness of S. aureus [11].
Collectively, these observations are consistent with a previous report showing that vancomycin affects the expression of CWSSassociated genes [12]. Another differentially expressed protein, GpmA, functions in cellular metabolism. GpmA catalyzes the interconversion of 2phosphoglycerate and 3-phosphoglycerate and is therefore involved in the glycolytic pathway [31]. As a key enzyme in glycolysis and energy metabolism, GpmA is a potential target for novel antibiotics [31]. This study is the first to report that GpmA is up-regulated in hVISA.
IsaA, which is involved in cell wall biogenesis, was also overexpressed in both hVISA isolates, as shown in our comparative proteomics results. IsaA cleaves peptidoglycan and thus plays a significant role in peptidoglycan turnover, cell wall crosslinking, and cell division [32]. Therefore, IsaA over-expression could be associated with the thickened cell walls of hVISA strains, which may be related to hVISA resistance. Another comparative proteomics study found that IsaA is up-regulated in the VISA strain Mu50, which is similar to our result [16]. The lack of RNAIII can lead to the over-expression of IsaA [33]. Several studies have indicated that VISA is characterized by agr dysfunction or RNAIII down-regulation [6,34,35]. A cDNA microarray study showed that IsaA is up-regulated in VRSA strains [36]. Therefore, the isaA gene may have an important function in S. aureus resistance to vancomycin.
To validate the accuracy of the results of our comparative proteomics analysis, 6 pairs of isogenic VSSA and hVISA strains isolated from the same patient, unrelated VSSA (n = 30) and hVISA (n = 24) strains, and 15 pairs of persistent VSSA strains were selected for confirmation by qRT-PCR. Analysis of the isogenic strains showed that isaA, msrA2, gpmA, and ahpC were significantly up-regulated in most of the hVISA strains compared with the VSSA strains, which was partly consistent with the results of comparative proteomics. However, only isaA was significantly up-regulated in hVISA strains compared with the unrelated VSSA strains. Therefore, the over-expression of isaA may be related to hVISA resistance. Analysis of the 15 pairs of persistent VSSA strains showed no differences in the expression of the identified genes, which indicates that these genes are not associated with persistent infection.
The present study has several limitations. First, the functionality of the identified genes could not be assigned in the absence of gene knockout experiments or further studies. Furthermore, the gene expression changes observed may be a consequence of vancomycin resistance and not causal of this phenotype. For example, these changes may be necessary to compensate for increased cell wall thickness or a consequence of reduced growth rate.
In summary, five differentially expressed proteins, IsaA, MsrA2, Asp23, GpmA, and AphC, were identified in two pairs of isogenic VSSA and hVISA strains via comparative proteomics analysis. The results of qRT-PCR showed that the isaA gene was significantly up-regulated in most of the clinical hVISA isolates, suggesting a relation between increased expression of isaA and hVISA resistance.