Trapping and Identification of Cellular Substrates of the Staphylococcus aureus ClpC Chaperone

Abstract

ClpC is an ATP-dependent Hsp100/Clp chaperone involved in protein quality control in low-GC Gram-positive bacteria. Previously, we found that ClpC affected the expression of a large number of genes, including capsule genes in Staphylococcus aureus. Here we constructed a His-tagged ClpC variant (ClpC^trap) with mutations within the Walker B motifs to identify the direct substrates of ClpC by copurification with ClpC^trap followed by gel electrophoresis combined with liquid chromatography-tandem mass spectrometry proteomics. We identified a total of 103 proteins that are potential substrates of ClpC in strain Newman. The direct protein-protein interaction of ClpC with a subset of the captured proteins was verified in a bacterial two-hybrid system. The captured proteins could be grouped into various functional categories, but most were related to proteins involved in the stress response. Several known ClpC substrates were captured, including ClpP, TrfA/MecA, ClpB, DnaK, DnaJ, GroL, RecA, and CodY, supporting the validity of our approach. Our results also revealed many new ClpC substrates, including AgrA, CcpA, RsbW, MurG, FtsA, SrtA, Rex, Atl, ClfA, and SbcC. Analysis of capsule production showed that three of the captured proteins, which were not previously known to be transcriptional regulators, did affect capsule production.

INTRODUCTION

Staphylococcus aureus is a major cause of bacterial infections, capable of causing a wide range of diseases ranging from simple skin infections to life-threatening diseases, such as endocarditis or pneumonia. Infections due to S. aureus were once limited to hospital settings and could be effectively treated with antibiotics. However, methicillin-resistant S. aureus (MRSA) with resistance to multiple antibiotics has become widespread, and highly virulent strains have now spread to the community, causing infections in normally healthy individuals. This rapid emergence of highly virulent MRSA strains in the community setting is considered one of the most surprising events in infectious diseases in recent years (1). It has been estimated that the number of deaths caused by MRSA in the United States has surpassed those caused by HIV/AIDS (2, 3).

The ability of S. aureus to cause a wide range of diseases stems from the fact that it can produce an abundance of virulence factors, including secreted toxins, enzymes, and cell surface molecules (4). The capsule is an important virulence factor in S. aureus, and it has been used for vaccine development (5, 6) and is of particular interest to our laboratory, as we have used it as a target to understand virulence gene regulation. Previously, our laboratory identified ClpC as a factor that affects capsule gene transcription, based on our screening of a transposon library (7). The fact that ClpC plays a role in transcriptional regulation is interesting because it is not a typical regulator that regulates through direct DNA binding. ClpC, which is conserved in all low-GC Gram-positive bacteria, is an ATP-dependent Hsp100/Clp chaperone of the AAA+ superfamily involved in protein quality control (8). ClpC and another Clp ATPase, ClpX, can associate with the ClpP protease to form proteolytic complexes that can affect many cellular functions, including gene regulation (8, 9). In S. aureus, very few substrates of ClpC/XP have been identified, although ClpCP has been implicated in degradation of antitoxins (10). Cohn et al. (11) showed that ClpXP, and to a lesser extent ClpCP, are involved in regulating the SOS response and thus affect expression of a subset of SOS regulon genes by degrading the LexA N-terminal domain after autocleavage of LexA.

Recently, Feng et al. (12) used a proteolytically inactive ClpP (ClpP^trap) and identified about 70 ClpP substrates in S. aureus. In addition, they used the ClpP^trap construct in clpC or clpX mutants to capture ClpXP or ClpCP substrates, respectively. In S. aureus, ClpC has been shown to affect a large number of genes and proteins based on microarray and proteomic analyses (13, 14), but how ClpC affects gene expression or protein production is largely unknown. In this study, we aimed to identify proteins that directly interact with ClpC, including those that are not destined for ClpCP proteolysis. We were especially interested in transcriptional regulators, as these could lead to a further understanding of gene regulation by ClpC. To this end, we developed a trapping method to identify ClpC substrates in S. aureus.

MATERIALS AND METHODS

Bacterial strains, culture media, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. Staphylococci were cultured in tryptic soy broth (TSB; Difco Laboratories, Detroit, MI). Escherichia coli was cultivated in Luria-Bertani broth or agar (Difco). MacConkey agar (Difco) plates containing 1% maltose were used for the bacterial two-hybrid assays. Antibiotics were added to culture media, as appropriate, at final concentrations of 10 μg/ml chloramphenicol, 3 μg/ml tetracycline, 10 μg/ml erythromycin, 50 μg/ml spectinomycin, and 100 μg/ml penicillin. Phages 52A and 80α were used for plasmid and chromosomal DNA transduction between S. aureus strains.

Table 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristic(s)	Reference or source
S. aureus strains
Newman	Wild-type CP5 strain	T. Foster
CYL6841	Newman ΔclpC::cat	14
CYL12447	Newman ΔclpC::cat(pJG4017)	This study
CYL12448	Newman ΔclpC::cat(pLL31)	This study
CYL12683	Newman ΔclpC::cat(pJG4080)	This study
CYL6620	Newman ΔsbcC	15
IK184	Newman ΔrsbUVWsigB	16
NE1922	USA300 nusG::bursa	NARSA
NE1495	USA300 murA::bursa	NARSA
NE460	USA300 atl::bursa	NARSA
E. coli strains
DH5α	Host strain for plasmids	Invitrogen
XL1 Blue	Host strain for plasmids	Stratagene
BTH101	Recipient strain for two-hybrid system	17
Plasmids
pLL31	E. coli-S. aureus shuttle vector with Pspac	14
pJG4017	pLL31-clpCtrap-His6	This study
pJG4080	pLL31-clpC-His6	This study
pKT25	Bacterial two-hybrid system vector	17
pUT18C	Bacterial two-hybrid system vector	17

Plasmid and strain construction.

To construct a clpC variant suitable for trapping ClpC substrates, we replaced the conserved Glu residue in each of the two Walker B domains with an Ala residue. The clpC(E280A/E618A) mutation, referred to as clpC^trap, was constructed via overlapping PCR in two steps, using the primers listed in Table 2. In the first step, one primer set (clpCtrap1 to -4) was used to construct the E280A mutation, whereas another primer set (clpCtrap5, clpCtrap9, clpCtrap10, and clpCtrap8) was used to construct the E618A mutation. In the second step, the two PCR fragments from the first step were used to construct the entire clpC^trap gene, using primers clpCtrap1 and clpCtrap8. The resulting PCR fragment, which also incorporated the His₆ tag sequence, was cloned into pLL31 (14) carrying a Tobacco etch virus protease (TEV)-Myc tag sequence (GSGGENLYFQGAYTSGEQKLISEEDLNGE) with a TTA stop codon, resulting in pJG4017, which contains the clpC^trap gene with the His₆-TEV-Myc sequence at the 3′ ends. A control plasmid, pJG4080, carrying the wild-type clpC gene and the His₆-myc tag sequence at the 3′ end, was also constructed using primers clpCtrap1 and clpCtrap11. The clones were verified by DNA sequencing. The resulting plasmids were transduced into the clpC deletion strain CYL6841 (i.e., Newman ΔclpC::cat). To construct plasmids for the two-hybrid assay, the inserts were amplified using the primers listed in Table 2 and cloned into either pUT18C or pKT25 (17). The inserts were verified by sequencing. The transposon insertion mutants used in capsule assays were constructed by phage transduction of defined bursa aurealis transposon mutations in the Nebraska transposon library obtained from the Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA). Transposon insertions were confirmed by PCR.

Table 2.

Primers used in this study

Primer use and name	Sequencea
Primers for ClpCtrap construction
ClpCtrap1	5′-GAATTCTGGAAGAGGCTTCTTATAGAT-3′
ClpCtrap2	5′-AGTATGCAACGCATCAATAAATAG-3′
ClpCtrap3	5′-CTATTTATTGATGCGTTGCATACT-3′
ClpCtrap4	5′-GGTGTCGTATGACTCTTAAGTCTTAC-3′
ClpCtrap5	5′-GTAAGACTTAAGAGTCATACGACACC-3′
ClpCtrap9	5′-TCAATTGCATCAAATAAAATTACAGAATATGGTTTAC-3′
ClpCtrap10	5′-CATATTCTGTAATTTTATTTGATGCAATTGAAAAAGCTCAT-3′
ClpCtrap8	5′-GGATCCatggtgatggtgatgatgTCCACCTGCTTGCGATGGTGTTTTAGTTTC-3′
ClpCtrap11	5′-GGATCCatggtgatggtgatgatgTCCCCATGCTTGCGATGGTGTTTTAGTTTC-3′
Primers for two-hybrid test
2h clpC F	5′-CTGCAGTCATTATTTATGTTATTTGGTAGATT-3′
2h clpC R	5′-GGATCCTGCTTGCGATGGTGTTTTAGTTTC-3′
2h ctsR F	5′-CTGCAGGTGATATACATGCACAATATGTCT-3′
2h ctsR R	5′-GGATCCGTAATAATTTATAACTGGTAACAA-3′
2h mecA F	5′-CTGCAGTGAGATGATATGAGAATAGAA-3′
2h mecA R	5′-GGATCCTTCAGTTGTCTCTGGAAAATA-3′
2h codY Fnew	5′-CTGCAGGAAAAATTCATGAGCTTATTATCT-3′
2h codY Rnew	5′-GGATCCTTTACTTTTTTCTAATTCATCTAA-3′
2h rsbW Fnew	5′-CTGCAGTCGAATAACATGCAATCTAAAGAA-3′
2h rsbW Rnew	5′-GGATCCGCTGATTTCGACTCTTTCGCCATT-3′

^aUnderlined sequences denote restriction sites. Nucleotides shown in bold are mutations made for clpC^Trap construction. Lowercase letters denote the His₆ sequence.

In vivo trapping of ClpC substrates and MS analysis.

Overnight S. aureus cultures were diluted in 200 ml of TSB with tetracycline to an optical density at 660 nm (OD₆₆₀) of 0.05 and incubated to an OD₆₆₀ of 0.3 to 0.5 (∼2 h) at 37°C with shaking at 225 rpm. Cultures were then induced with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) and incubated for 3 h to an OD₆₆₀ of 2.9 (ranging from 2.7 to 3.1). The cultures were then centrifuged at 8,000 × g for 10 min and washed with 40 ml of cold phosphate-buffered saline (PBS; pH 7.4). The pellets were resuspended in 8 ml of PBS with 1× protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN). Cell lysis was achieved by physical disruption of 1-ml portions of cell suspensions by using 0.1-mm zirconia-silica beads (Biospec, Bartlesville, OK) in a FastPrep instrument (Qbiogene, Carlsbad, CA) with six 40-s pulses at 6 m/s and 5-min incubations on ice in between pulses. The lysed cells were centrifuged at 8,000 × g for 15 min at 4°C, and supernatants were saved. The pellet was extracted twice, first with 800 μl and then with 500 μl of PBS containing 1× protease inhibitor cocktail, using the FastPrep instrument as described above, but with four 40-s pulses. The supernatants were combined after centrifugation. The pooled supernatants were then centrifuged at 18,000 × g at 4°C for 20 min and adjusted to contain 50 mM Na-phosphate, 300 mM NaCl, and 5 mM imidazole (pH 7.4). Separate HisPur cobalt resin columns (Thermo Scientific, Hudson, NH) were used for control experiments and for isolation of the His-tagged ClpC^trap-substrate complexes according to the manufacturer's instructions. The column was washed until the absorbance at 280 nm approached baseline. The column-bound proteins were eluted with 50 mM Na-phosphate and 300 mM NaCl (pH 7.4) buffer containing 150 mM imidazole. Proteins in the eluted fractions were then analyzed by using 4-to-12% SDS-PAGE gradient gels. Each selected lane of the SDS-PAGE gel was equally divided into 20 slices and subjected to in-gel trypsin digestion and mass spectrometry (MS) analysis as described previously (18). Proteins were identified by Mascot searches using thresholds of 95% protein probability, 95% peptide probability, and a minimum of two peptides per protein sequence. The spectral counts for each protein were normalized to the total counts to account for between-sample variation. The normalized spectral counts were compared using Student's t test (P ≤ 0.05) to identify proteins that were differentially enriched by ClpC^trap.

Other methods.

Bacterial two-hybrid experiments were done using the bacterial adenylate cyclase two-hybrid system (BACTH) as described previously (17). Capsule assays were performed using cultures grown in TSB without glucose, essentially as described previously (14). Western analyses were carried out using anti-His antibody (Abcam, Cambridge, MA), anti-RecA antibody (Abcam), and anti-CodY antibody (generated by 21st Century Biochemicals, Marlboro, MA). Protein stability tests for CodY and RecA were carried out as described previously (12).

RESULTS AND DISCUSSION

Construction of ClpC^trap in S. aureus.

The Clp ATPases of the AAA+ superfamily are closely related chaperones with one or two highly conserved AAA domains that consist of Walker A and Walker B motifs, which are involved in nucleotide binding and hydrolysis, respectively (19). To capture the ClpC substrates, we constructed a ClpC^trap variant based on the methods used in the study reported by Weibezahn et al. (20), in which an E. coli ClpB variant, with mutations in the Walker B motifs of both AAA domains, was able to form stable complexes with its substrates. ClpB and ClpC are both members of the class 1 HSP100/Clp family of ATPases, with two AAA domains (21). A comparison of the Walker B motifs of E. coli ClpB with those of S. aureus ClpB and ClpC showed highly conserved residues, including the Glu residues involved in ATP hydrolysis (Fig. 1). To construct a ClpC^trap variant, the Glu residues in the Walker B motifs of both AAA domains in ClpC were changed to Ala (E280A/E618A). We also engineered a His₆ tag at the C-terminal end for purification purposes. The clpC^trap construct was expressed in the clpC-null background of S. aureus Newman under the control of an IPTG-inducible Pspac promoter. The expression and binding of the His-tagged ClpC^trap protein to cobalt affinity columns were validated by Western blotting using anti-His antibody (data not shown). A similar plasmid carrying a His-tagged, wild-type ClpC protein was also constructed. The plasmid was able to complement a clpC mutant strain to the wild-type level of capsule production, indicating that the tag did not alter the activity of ClpC (data not shown).

Fig 1.

Amino acid sequence alignment of E. coli ClpB and S. aureus ClpB and ClpC proteins. The conserved sequences within the Walker B motifs are shown in bold. The conserved active site glutamic acid residues involved in ATP hydrolysis were changed to alanine residues in ClpC^trap (arrows).

In vivo trapping of ClpC substrates.

To carry out in vivo trapping, we performed two independent trapping experiments using strains CYL12447 carrying ClpC^trap-His and CYL12448 carrying the empty vector. Protein extracts were obtained and loaded onto cobalt affinity columns as described in Materials and Methods. The bound proteins were eluted with buffer containing 150 mM imidazole. This concentration of imidazole was chosen based on the results of a pilot study, wherein we eluted proteins using a step gradient of an increasing imidazole concentration (data not shown). The eluted proteins were analyzed in 4-to-12% SDS-PAGE gradient gels (Fig. 2). As expected, considerably more protein bands were found in the ClpC^trap strain than the strain with the empty vector. However, some protein bands were also found in the strain with the empty vector, suggesting that these proteins interact with the cobalt resin nonspecifically. The gel lanes were subjected to gel electrophoresis combined with liquid chromatography-tandem MS (GeLC-MS/MS) analysis, and the proteins were identified by Mascot searches as described in Materials and Methods. A total of 465 and 634 proteins were identified from the two ClpC^trap samples, respectively, and 382 and 521 were identified from the two control samples (see Table S1 in the supplemental material). After comparing the means of the normalized spectral counts between the ClpC^trap and the control samples via Student's t test, we identified a total of 91 proteins for which there was at least a 2.0-fold difference in counts between the two samples. However, we also found that 8 proteins were preferentially captured in the empty vector samples. It is unclear why some of the proteins were preferentially captured in the vector control samples, where there was no ClpC^trap-His. One possibility is that the unbound resin may allow some native proteins, for example, those with a metal binding capability, to bind better in the absence of a His-tagged protein (22). The proteins that were significantly enriched by ClpC^trap by at least 2-fold are shown in Table 3. Table 3 also includes an additional 12 proteins (indicated by asterisks) that were found in both samples of ClpC^trap but were absent in both control vector samples, implying that they specifically interact with ClpC. Because these proteins were exclusively captured by ClpC^trap, we considered these proteins to be potential ClpC substrates, despite the fact that their statistical P values were above 0.05.

Fig 2.

Proteins trapped by ClpC^trap. Proteins eluted from cobalt affinity columns were separated on 4-to-12% SDS-PAGE gradient gels and stained with Coomassie blue. Gel lanes were sliced and subjected to in-gel trypsin digestion, and proteins were identified by GeLC-MS/MS analyses. The left-most lane shows molecular mass standards, with sizes indicated in kDa.

Table 3.

Proteins that interact with ClpC^trap identified by GeLC-MS/MS

Functional group and gene	ORFa	Identified protein	Mass (kDa)	IDb	Normalized count (expt 1/2)c		P value	Fold change
					ClpCtrap	Vector
Regulatory functions
ccpA	1629	Catabolite control protein A	36	A6QHR9	8.0/8.7	0.00/1.5	0.006	11.3
codY	1165	GTP-sensing transcriptional pleiotropic repressor CodY	29	A6QGF5	19.0/28.3	11.0/7.9	0.050	2.5
agrA	1946	Staphylococcal accessory gene regulator A	28	A6QIN6	31.0/31.8	10.0/14.2	0.006	2.6
rsbW	1971	Serine-protein kinase RsbW	18	A6QIR1	27.0/23.1	11.0/8.3	0.011	2.6
rex	1953	Redox-sensing transcriptional repressor Rex	24	A6QIP3	5.0/4.6	0.0/1.0	0.007	9.8
hprK	0728	HPr kinase/phosphorylase	34	A6QF68	3.0/4.0	0.0/0.0	0.011	infd
Protein fate
clpC	0487	Clp protease, ATP binding subunit ClpC	91	A6QEH7	5028.0/3263.6	49.0/0.0	0.021	169.2
clpB	0845	Clp protease, ATP binding subunit ClpB	98	A6QFI5	363.0/266.8	112.0/177.1	0.050	2.2
clpP	0736	ATP-dependent Clp protease proteolytic subunit	22	A6QF76	29.0/19.6	10.0/5.4	0.043	3.2
dnaK	1483	Chaperone protein DnaK	66	A6QHC3	158.0/181.9	33.0/29.0	0.004	5.5
dnaJ	1482	Chaperone protein DnaJ	42	A6QHC2	34.0/46.2	6.0/3.9	0.015	8.1
groL	1937	60-kDa chaperonin	58	A6QIM7	50.0/60.1	11.0/16.2	0.009	4.0
trfA	0868*	Adaptor protein MecA/TrfA	28	A6QFK8	1.0/4.6	0.0/0.0	0.130	inf
—e	0495*	Conserved hypothetical protein	20	A6QEI5	3.0/0.6	0.0/0.0	0.139	inf
—	1187*	Conserved hypothetical protein	49	A6QGH7	3.0/1.2	0.0/0.0	0.077	inf
Biosynthesis of cofactors, prosthetic groups, and carriers
lipA	0796	Lipoyl synthase	35	A6QFD6	21.0/16.2	8.0/6.9	0.023	2.5
ispD1	0185	2-C-methyl-d-erythritol 4-phosphate cytidylyltransferase 1	27	A6QDM5	8.0/6.4	0.0/0.0	0.006	inf
sufC	0785	FeS assembly ATPase SufC	28	A6QFC5	37.0/43.9	8.0/8.3	0.006	3.6
sufD	0786	FeS assembly protein SufD	49	A6QFC6	15.0/18.5	4.0/7.9	0.027	2.5
folD	0932	Bifunctional protein FolD	31	A6QFS2	6.0/7.5	0.0/0.5	0.007	27.5
hemE	1725	Uroporphyrinogen decarboxylase	39	A6QI15	2.0/2.9	0.0/1.0	0.049	5.0
hemL	1756	Glutamate-1-semialdehyde 2,1-aminomutase 2	47	A6QI46	2.0/3.5	0.0/0.0	0.032	inf
thiD	0543	Phosphomethylpyrimidine kinase	30	A6QEN3	10.0/15.6	0.0/3.9	0.043	6.5
Amino acid metabolism
sbnB	0061	Ornithine cyclodeaminase	38	A6QDA1	4.0/3.5	0.0/1.5	0.031	5.1
—	1421	2-Oxoisovalerate dehydrogenase, E2 component	47	A6QH61	7.0/6.9	4.0/1.5	0.039	2.5
—	1422	2-Oxoisovalerate dehydrogenase, E1 component	36	A6QH62	6.0/7.5	0.0/0.5	0.007	27.5
—	0475	Cysteine synthase	33	A6QEG5	6.0/8.1	0.0/0.5	0.012	28.7
—	1871	Aspartate transaminase	48	A6QIG1	3.0/2.3	0.0/0.5	0.015	10.8
Cell envelope
atl	0922	Bifunctional autolysin	137	A6QFR2	117.0/106.8	55.0/59.9	0.005	2.0
murG	2028	UDP-glucose diacylglycerol glucosyltransferase	45	A6QGX0	6.0/10.4	0.0/1.0	0.038	16.7
femB	1287*	Methicillin resistance expression factor FemB	50	A6QGS7	3.0/6.4	0.0/0.0	0.054	inf
srtA	2426	Sortase A, peptide LPXTG peptidoglycan transferase	24	A6QK16	3.0/2.9	1.0/1.0	0.000	3.0
clfA	0756	Clumping factor A	97	Q53653	10.0/9.8	0.0/4.9	0.047	4.0
—	0369	Putative lipoprotein	24	A6QE59	4.0/4.6	1.0/2.0	0.019	2.9
—	2356	Putative lipoprotein	17	A6QJU6	8.0/5.8	2.0/1.5	0.023	4.0
Cellular processes
ahpF	0371	Alkyl hydroperoxide reductase subunit F	55	A6QE61	8.0/11.6	0.0/1.5	0.021	13.3
—	1639	Propeptide, PepSY, and peptidase M4	12	A6QHS9	3.0/2.3	0.0/0.0	0.008	inf
Cell division
ftsA	1095	Cell division protein FtsA	53	A6QG85	59.0/51.4	22.0/28.5	0.013	2.2
divIVA	1102*	Putative uncharacterized protein	24	A6QG92	8.0/3.5	0.0/0.0	0.064	inf
DNA metabolism
dnaN	0002	DNA polymerase III subunit beta	42	A6QD42	40.0/25.4	4.0/9.3	0.039	4.9
recA	1194	Protein RecA	38	A6QGI4	58.0/86.6	16.0/12.8	0.028	5.0
recN	1425	DNA repair protein RecN	64	A6QH65	7.0/6.4	0.0/2.0	0.016	6.8
sbcC	1258	Nuclease SbcCD subunit C	117	A6QGP8	9.0/8.7	0.0/3.4	0.027	5.1
mutS	1204	DNA mismatch repair protein MutS	100	A6QGJ4	8.0/7.5	2.0/4.4	0.033	2.4
Central intermediary metabolism
glmS	2056	Glucosamine-fructose-6-phosphate aminotransferase, isomerizing	66	A6QIZ6	84.0/56.0	17.0/20.1	0.034	3.8
—	0584	Hydrolase	31	A6QES4	7.0/4.0	0.0/0.0	0.032	inf
—	0973	Inositol-1-monophosphatase family protein	30	A6QFW3	6.0/6.4	0.0/2.0	0.017	6.3
—	2375	NAD-dependent epimerase/dehydratase	25	A6QJW5	2.0/1.2	0.0/0.0	0.032	inf
—	2434	Conserved hypothetical protein	37	A6QK24	2.0/3.5	0.0/0.5	0.042	11.1
Energy metabolism
citC	1587	Isocitrate dehydrogenase [NADP]	46	A6QHM7	5.0/5.8	0.0/0.0	0.003	inf
pycA	0979	Pyruvate carboxylase	129	A6QFW9	5.0/4.6	0.0/0.0	0.001	inf
gapA	0741	Glyceraldehyde 3-phosphate dehydrogenase 1	36	A6QF81	17.0/22.5	0.0/0.0	0.009	inf
sucB	1325	Dihydrolipoamide succinyltransferase E2 component of 2-oxoglutarate dehydrogenase complex	47	A6QGW5	2.0/4.0	0.0/0.0	0.049	inf
sucC	1155	Succinyl coenzyme A ligase (ADP-forming) subunit beta	42	A6QGE5	9.0/12.7	0.0/0.0	0.014	inf
zwf	1412	Glucose-6-phosphate 1-dehydrogenase	57	A6QH52	15.0/9.2	3.0/2.0	0.041	4.9
glk	1451	Glucokinase	35	A6QH91	5.0/4.0	0.0/0.0	0.006	inf
pgk	0742	Phosphoglycerate kinase	43	A6QF82	8.0/6.9	3.0/0.0	0.032	5.0
ackA	1605	Acetate kinase	44	A6QHP5	46.0/38.7	14.0/10.3	0.009	3.5
fbaA	2029	Fructose-bisphosphate aldolase	31	A6QIW9	1.0/1.7	0.0/0.0	0.032	inf
—	2210	Formate dehydrogenase homolog	111	A6QJF0	7.0/6.9	4.0/1.5	0.039	2.5
—	1672*	Transaldolase	26	A6QHW2	3.0/0.6	0.0/0.0	0.139	inf
Fatty acid and phospholipid metabolism
accC	1431	Biotin carboxylase subunit of acetyl-CoA carboxylase	50	A6QH71	9.0/6.4	0.0/2.0	0.028	7.8
gpsA	1383	Glycerol-3-phosphate dehydrogenase	36	A6QH23	3.0/4.6	0.0/0.0	0.021	inf
Protein synthesis
rsmH	1089	rRNA small subunit methyltransferase H	36	A6QG79	5.0/4.0	0.0/1.5	0.025	6.1
rplA	0500	50S ribosomal protein L1	25	A6QEJ0	18.0/18.5	7.0/6.4	0.001	2.7
rpsA	1385	30S ribosomal protein S1	43	A6QH25	29.0/20.8	13.0/9.8	0.046	2.2
rpsP	1148	30S ribosomal protein S16	10	A6QGD8	8.0/6.4	4.0/2.0	0.043	2.4
rpsD	1613	30S ribosomal protein S4	23	A6QHQ3	65.0/50.2	22.0/33.9	0.044	2.1
gatA	1838	Glutamyl-tRNA(Gln) amidotransferase subunit A	53	A6QIC8	40.0/65.3	6.0/16.2	0.046	4.7
gatB	1837	Aspartyl/glutamyl-tRNA(Asn/Gln) amidotransferase subunit B	54	A6QIC7	42.0/27.1	4.0/11.3	0.041	4.5
tyrS	1622	Tyrosine-tRNA ligase	48	A6QHR2	2.0/1.7	0.0/0.5	0.014	7.6
ileS	1103*	Isoleucyl-tRNA ligase	48	A6QG93	1.0/3.5	0.0/0.0	0.106	inf
—	0721	Sigma 54 modulation protein	22	A6QF61	31.0/26.6	15.0/13.7	0.012	2.0
—	1408	SpoU rRNA methylase family protein	27	A6QG38	3.0/3.5	0.0/1.0	0.019	6.6
Purine, pyrimidine, nucleoside, nucleotides, purine ribonucleotide synthesis
guaB	0380	Inosine-5′-monophosphate dehydrogenase	53	A6QE70	42.0/27.1	4.0/11.3	0.041	4.5
prs	0463	Ribose-phosphate pyrophosphokinase	35	A6QEF3	44.0/40.4	4.0/15.2	0.016	4.4
pyrH	1168	Uridylate kinase	26	A6QGF8	7.0/5.8	1.0/1.0	0.006	6.4
pyrR	1109	Bifunctional protein PyrR	20	A6QG99	15.0/25.4	0.0/4.9	0.045	8.2
nrdE	0700	Ribonucleoside-diphosphate reductase	82	A6QF40	55.0/56.0	27.0/28.5	<0.001	2.0
—	0284	Conserved hypothetical protein	21	A6QDX4	3.0/2.9	0.0/0.0	<0.001	inf
Transcription
sigA	1464	RNA polymerase sigma factor	42	A6QHA4	14.0/9.8	0.0/2.0	0.021	12.1
rpoA	2126	DNA-directed RNA polymerase subunit alpha	35	A6QJ66	13.0/7.5	0.0/0.0	0.032	inf
nusA	1176	Transcription termination-antitermination factor	44	A6QGG6	14.0/12.1	2.0/7.4	0.049	2.8
nusG	0498	Transcription antitermination protein NusG	21	P0C1S3	9.0/6.9	4.0/1.5	0.043	2.9
Transport and binding proteins
—	0581	Iron compound ABC transporter, iron compound binding protein	33	A6QES1	2.0/2.3	0.0/0.0	0.003	inf
—	2312	Amino acid ABC transporter, permease protein	26	A6QJQ2	2.0/2.3	0.0/0.0	0.003	inf
—	0705	Ferrichrome ABC transporter lipoprotein	38	A6QF45	10.0/9.8	4.0/3.9	<0.001	2.5
—	0954	Conserved hypothetical protein	24	A6QFU4	3.0/2.3	0.0/0.0	0.008	3.00
Unknown
—	0272	Putative uncharacterized protein	21	A6QDW2	6.0/4.6	0.0/0.0	0.008	inf
—	0632	Putative uncharacterized protein	24	A6QEX2	7.0/6.4	1.0/2.9	0.022	3.4
—	0737	Putative uncharacterized protein	34	A6QF77	4.0/3.5	0.0/1.5	0.031	5.1
—	0976	Putative uncharacterized protein	19	A6QFW6	7.0/4.6	1.0/1.5	0.032	4.7
—	1265*	Putative uncharacterized protein	11	A6QGQ5	3.0/1.2	0.0/0.0	0.077	inf
—	1381	Putative uncharacterized protein	22	A6QH21	2.0/1.7	0.0/0.0	0.003	inf
—	1730	Putative uncharacterized protein	13	A6QI20	7.0/5.8	1.0/2.0	0.012	4.3
—	1820*	Putative uncharacterized protein	10	A6QIB0	5.0/1.2	0.0/0.0	0.125	inf
—	2067	ATP binding Mrp/Nbp35 family protein	38	A6QJ07	3.0/5.8	0.0/0.0	0.044	inf
—	2002*	Putative uncharacterized protein	19	A6QIU2	1.0/2.9	0.0/0.0	0.088	inf
—	2201*	Dehydrogenase family protein	41	A6QJE1	25.0/6.4	0.0/0.0	0.117	inf
—	2209	Putative uncharacterized protein	17	A6QJE9	2.0/2.3	0.0/0.0	0.003	inf
—	2405	Putative uncharacterized protein	27	A6QJZ5	1.0/1.2	0.0/0.0	0.003	inf
—	2468	Acetyltransferase, GNAT family protein	19	A6QK58	3.0/2.3	0.0/0.0	0.008	inf
—	2511*	Putative uncharacterized protein	22	A6QKA1	4.0/1.7	0.0/0.0	0.064	inf

^aORF numbers are based on strain Newman. ORF numbers followed by an asterisk indicate proteins with a P value of >0.05 but that were specifically captured by ClpC^trap.

^bThe ID number is the UniProt accession number.

^cTwo biological replicates were used to identify substrates of ClpC. The spectral counts were normalized based on total counts.

^dinf, the fold change could not be accurately estimated due to no detection in the negative-control samples.

^e—, a gene name has not yet been designated.

Potential ClpC substrates.

As shown in Table 3, proteins in various functional categories were captured by ClpC^trap, suggesting that ClpC is involved in many different cellular processes. Collectively, proteins involved in determining protein fate, including ClpB, DnaK, DnaJ, GroL, and ClpP, are among the most abundant proteins trapped by ClpC^trap. The trapping of ClpP was expected, since ClpC can form a proteolytic complex with ClpP to degrade misfolded proteins, especially under stress conditions. ClpB, which does not associate with a protease, has been shown to associate with Hsp70 chaperones (DnaK, DnaJ, and GrpE [referred to as KJE]) to disaggregate or refold aggregated proteins under stress or nonstress conditions (23, 24). GroEL, which also does not partner with a protease either, does interact with GroES to properly fold certain proteins (25). Capture of these chaperones in large amounts by ClpC suggests that ClpC interacts with these chaperones in S. aureus. Since ClpC has not been shown to associate with ClpB/KJE or GroESL for protein folding activities, we speculate that ClpC may escort these chaperones for ClpCP degradation when in excess. Indeed, ClpB, DnaK, and DnaJ have been identified as the ClpCP substrates (12). The E. coli ClpA protein, an equivalent to S. aureus ClpC, has been shown to participate in self-degradation by ClpAP (26, 27). Thus, it is possible that ClpCP could be responsible for the turnover of other chaperones in S. aureus.

The Bacillus subtilis ClpC has a low intrinsic ATPase activity and appears to rely on adaptor proteins for chaperone activity (9). Three adaptors have been shown to interact with ClpC and modulate its function. These adaptors are also substrates for degradation by ClpCP when not involved in substrate delivery. Homologs of two of these adaptors were also found in S. aureus. MecA is the first characterized adaptor protein found in B. subtilis, and it enables the recognition of specific substrates by ClpC, including ComK, as well as misfolded or aggregated proteins (28, 29). ComK, which is involved in competence regulation, is normally bound by the MecA-ClpC complex and destined for degradation by ClpCP. Upon competence induction, ComS is produced and displaces ComK in the ComK-MecA-ClpC ternary complex, which allows ComK to activate the transcription of competence genes (30). We found TrfA, the MecA homolog in S. aureus (based on the terminology used in reference 12), was present in the ClpC^trap samples but absent in the controls (Table 3), suggesting that ClpC interacts with this adaptor in S. aureus. Another well-characterized ClpC adaptor in B. subtilis is McsB, whose kinase activity is inhibited by interaction with ClpC (31, 32). McsB is released from the McsB/ClpC complex by unfolded proteins, which allows McsB to autophosphorylate in the presence of McsA. Phosphorylated McsB then interacts with CtsR and acts as an adaptor of ClpC, leading to degradation of CtsR by the ClpCP proteolytic complex. Degradation of CtsR results in upregulation of genes involved in stress tolerance, including clpC, clpB, and clpE (32). In this study, we found that McsB was enriched 3.2-fold by ClpC^trap and CtsR was enriched by 4.0-fold. However, the statistical significance of enrichment in either case was slightly above the cutoff level (P = 0.064 and 0.107 for McsB and CtsR, respectively). Nonetheless, based on the conservation of these proteins between B. subtilis and S. aureus and the high level of enrichment by ClpC^trap found in this study, we suggest that McsB and CtsR interact with ClpC in S. aureus similarly as in B. subtilis.

Several proteins with regulatory function (CcpA, CodY, AgrA, RsbW, Rex, and HprK) were identified in this study, including those that are directly involved in transcriptional regulation, suggesting that ClpC can exert its regulatory function by modulating these regulators. In particular, AgrA (by affecting expression of the Agr quorum-sensing effector, RNAIII), RsbW (by controlling the activity of sigma factor SigB), and CodY affect a large number of genes, including virulence genes (33–35). Interaction of ClpC with these regulators may account for previous findings in which deletion of clpC affected transcription of many genes, including those involved in virulence (13, 14). It is interesting that all of these regulators identified here are involved in some form of stress-related response, although to different extents and in response to different forms of stress. Specifically, CcpA, CodY, and HprK are involved in nutrition sensing (36–39), AgrA is involved in cell density quorum sensing (40), Rex is involved in redox stress sensing (41), and RsbW (by modulating SigB) is involved in stationary-phase gene expression (42). Thus, ClpC may regulate different stress responses by interacting with these regulators and modulating their activities.

Enzymes involved in cell wall synthesis, cell division, and DNA repair are important targets for regulation when bacteria encounter stress. MurAA, which is involved in the first step of peptidoglycan synthesis, has been shown to be a substrate of ClpCP in B. subtilis (43). FtsZ, a cell division protein, has been shown to be a target of ClpXP protease in E. coli (44, 45). In S. aureus, FtsZ and RecA were both identified as potential substrates of ClpCP (12). Here, we found several proteins that were enriched by ClpC^trap and that are involved in cell wall metabolism (MurG and FemB), cell division (FtsA and DivIVA), and DNA repair (DnaN, RecA, RecN, SbcC, and MutS). In addition, we found MurA (a homolog of MurAA) was enriched by 5.9-fold (P = 0.052). The fact that MurAA is a substrate of ClpCP in B. subtilis suggests that MurA is a likely substrate of ClpC in S. aureus. MurA was not captured by ClpP^trap of S. aureus, however (12). Similarly, FtsZ was enriched by 2.1-fold (P = 0.146), suggesting that FtsZ may also interact with ClpC. In support of this, Feng et al. (12) reported that FtsZ was a potential substrate of ClpCP in strain 8325-4. In the same study, however, FtsZ was not captured by ClpP^trap in strain Newman. Together, these results suggest that ClpC may interact with FtsZ and MurA, but further study is required to conclusively demonstrate their interactions.

Many proteins involved in energy metabolisms were captured, including several enzymes involved in the tricarboxylic acid (TCA) cycle (PycA, CitC, SucB, and SucC) and glycolysis (GapA, Glk, Pgk, and FbaA). Numerous other proteins involved in various metabolic pathways were also identified as ClpC substrates. These include those involved in protein translation and synthesis, nucleic acid synthesis, transcription, transport, fatty acid synthesis, cofactor synthesis, and central metabolism (Table 3). These results are consistent with previous reports that ClpC affects cellular metabolism, including the TCA cycle, nucleotide metabolism, and the stringent response (13, 46). Not surprisingly, most of the identified proteins can be related to the stress response, consistent with the general biological role of ClpC. Interestingly, we noted that two cell surface proteins, ClfA and Atl, were also captured in our assay. The interaction of ClpC with surface proteins has not been previously reported and was unanticipated. However, it seems likely that ClpC may interact with the precursors of these proteins in the cytoplasm, prior to their export to the cell surface. It is also of interest that SrtA, the sortase protein responsible for anchoring most surface proteins on cell wall, including ClfA, was also captured.

Interaction of ClpC and its substrates.

To verify that the trapped proteins identified in the GeLC-MS/MS analysis interact with ClpC in vivo, we employed a bacterial two-hybrid system based on the reconstitution of Bordetella pertussis adenylate cyclase activity in E. coli (17). We chose four proteins, MecA, CtsR, CodY, and RsbW, to test for direct interactions with ClpC. As indicated above, both MecA and CtsR have been shown to interact with ClpC in B. subtilis. As expected, both proteins were found to interact with ClpC (Fig. 3). These results confirmed that ClpC interacts with MecA and CtsR in S. aureus, similarly to B. subtilis. It is interesting that although we found enrichment of CtsR by ClpC^trap, the data were not statistically significant compared to the empty vector control.

Fig 3.

Demonstration of protein-protein interactions, by use of an E. coli bacterial two-hybrid system. Red appearance of the colonies on MacConkey agar supplemented with 1% maltose indicates a positive interaction of proteins fused to T18 and T25 fragments of Bordetella pertussis adenylate cyclase, respectively. ClpC was expressed from pUT18C (T18-ClpC), whereas putative ClpC substrate proteins were expressed from pKT25 (T25-CtsR, T25-RsbW, T25-MecA, and T25-CodY). A positive control with leucine zipper domains (T18-Zip/T25-Zip) and negative controls (T18/T25 and T18-ClpC/T25) are also shown. Two proteins (NWMN_2258 and NWMN_1135 with 0.97- and 1.07-fold changes, respectively) were also included to show lack of interaction of ClpC with proteins that were not specifically trapped.

CodY, which is highly conserved in low-GC Gram-positive bacteria, is a transcriptional regulator involved in metabolic stress (37). Here, we found that CodY was enriched by ClpC^trap by 2.5-fold. Using the two-hybrid system, we showed that CodY could indeed interact with ClpC (Fig. 3). Interestingly, CodY was also trapped by ClpP^trap in both the clpC and clpX mutant background, indicating that it is likely a substrate of ClpCP and/or ClpXP. The stability of CodY, however, was not affected by clpP deletion, suggesting that although CodY is bound, it is not degraded by ClpCP or ClpXP (12). To test whether ClpC is involved in the stability of CodY, we performed Western blot analyses with uninduced and mupirocin-induced cultures (mupirocin induces the stringent response that releases CodY from DNA binding [47]). Our results, which are consistent with those of Feng et al. (12), showed that ClpC, similar to ClpP, did not affect CodY stability either with or without mupirocin induction (Fig. 4A). Taken together, these results indicate that CodY is not regulated by ClpCP proteolysis, nor is its stability affected by ClpC. What role then does ClpC play by binding to CodY? One possibility is that ClpC modulates CodY activity without affecting the stability of the protein. For example, ClpC binding to CodY could affect CodY binding to branched-chain amino acids or to DNA. Interestingly, we previously reported that ClpC repressed CodY at the transcriptional level (14). Thus, ClpC appears to affect CodY at both the transcriptional and posttranscriptional levels. CodY has been shown to be negatively autoregulated in some low-GC bacteria (48, 49). It is therefore possible that the interaction of ClpC and CodY could result in transcriptional autorepression of codY. However, no CodY binding site has been found upstream of the codY gene in S. aureus (35), suggesting that CodY is not autoregulated. Thus, more complicated mechanisms than those described above are likely involved, and further in-depth studies are needed to understand how ClpC affects CodY.

Fig 4.

Protein stability test of CodY and RecA in Newman and Newman ΔclpC strains by Western blotting. A portion of the gel stained with Coomassie blue was used as a loading control below each blot. Western blots (n = 3) were quantified by densitometry, normalized to one of the stained protein bands, and analyzed by a one-way analysis of variance. (A) CodY stability was tested after inducing the stringent response by adding mupirocin to mid-log-phase cultures at time zero (T = 0*). Chloramphenicol was added 30 min after mupirocin addition. Cells were harvested immediately after chloramphenicol addition (labeled as 0) and 0.5, 1, 2, and 3 h thereafter. A codY-deleted strain and a codY-complemented strain, denoted − and +, respectively, were also included as controls. There was no significant difference between results at the different time points. (B) RecA stability was tested after SOS induction by adding mitomycin C at T = 0*. Chloramphenicol was added 15 min after mitomycin C addition. Cells were harvested as described for panel A. The RecA level was significantly lower (*, P < 0.05) in preinduced samples (T = 0*) than at other time points for both strains.

RsbW is an anti-sigma factor that binds to and inhibits the activity of SigB under nonstressed conditions. Upon stress induction, RsbU dephosphorylates RsbV, leading to the formation of an RsbVW complex and thereby relieving inhibition of SigB by RsbW (50). In this study, we found that RsbW was trapped by ClpC^trap (Table 3), suggesting that ClpC could affect SigB activity by binding to RsbW, thereby increasing transcription of SigB-dependent genes. As shown in Fig. 3, we validated the RsbW-ClpC interaction in the two-hybrid system. Interestingly, neither SigB nor any of its regulatory proteins were captured by ClpP^trap (12). These results suggest that ClpC can affect SigB activity independently from ClpP, supporting the contention that ClpC may exert its regulatory function by binding and sequestering target proteins rather than by ClpCP-directed proteolysis.

RecA is another protein identified by the ClpP^trap method as a potential ClpCP substrate. It was further shown that RecA was very stable, but that ClpP-mediated proteolysis of RecA could be observed 3 h after SOS induction. Although this effect was slight, the results argued that ClpCP plays a role in poststress rebalancing of RecA (12). Because we also captured RecA with our ClpC^trap method, we expected that ClpC would have an effect on RecA stability following SOS induction. Therefore, we carried out experiments similar to those described by Feng et al. (12) to determine whether ClpC is involved in RecA degradation. We did not detect any apparent difference in RecA stability between the wild-type and clpC mutant strains despite repeated efforts (Fig. 4B). The apparent discrepancy between the two studies may be due to strain differences; however, it is possible that RecA could be degraded by ClpXP rather than ClpCP.

Effect of ClpC substrates on the capsule.

Among the transcriptional regulators identified by ClpC^trap in this study, AgrA and RsbW/SigB have been shown to activate capsule gene expression, whereas CodY and CcpA have been shown to repress capsule gene expression (14, 36, 51, 52). These regulators could therefore function downstream of ClpC to regulate the capsule genes. In addition, SbcC, which is likely involved in DNA repair and has been shown to repress capsule genes (15), was captured by ClpC^trap, indicating that ClpC could also affect the capsule through SbcC. As expected, deletion of sbcC in strain Newman increased capsule production, whereas deletion of the rsbUVWsigB operon reduced capsule production (Fig. 5). To determine whether any new ClpC substrates identified in this study had an effect on capsule production, we selected three substrates, Atl, MurA, and NusG, that do not have a known effect on the capsule and inactivated the genes encoding these proteins in strain Newman. We found all three genes had an effect on capsule production (Fig. 5). Because murA is located upstream of a closely linked gene, we also performed a complementation experiment to rule out potential polar effects. Our results confirmed that MurA was involved in capsule production (data not shown). A complementation experiment was not done with an atl or nusG mutant, since neither gene is likely to affect its downstream genes, based on genetic organization. These results were surprising, as these genes are not likely to be directly involved in capsule synthesis. Atl is a murein hydrolase that is involved in cell separation following cell division (53). MurA is a primary enzyme involved in catalyzing the first committed step in peptidoglycan biosynthesis (54). Because the capsule is anchored to the cell wall (55), it is conceivable that genes involved in cell wall synthesis, like murA and atl, could affect capsule production. NusG is a general transcription factor that binds to and affects transcription by RNA polymerase, primarily by affecting transcript elongation (56). The finding here that the nusG::bursa mutation reduced the capsule suggests that NusG may be required for transcription of the full-length cap operon. In some bacteria, NusG paralogs (termed NusG^SP [57]) can selectively promote antitermination of operons, including capsule operons (58, 59). Because S. aureus capsule genes are transcribed as a long (∼17-kb) transcript (60), it seems likely that NusG may promote production of full-length transcripts through its antitermination activity. However, further studies are needed to test this possibility. Taken together, our results suggest that ClpC cannot only regulate capsule through transcriptional regulators like AgrA, CodY, and SigB at the transcriptional level but also can affect capsule production through nonregulatory proteins, like MurA and Atl, most probably at the posttranscriptional level. Our findings therefore highlight the complexity of capsule regulation and regulation of other virulence factors in S. aureus.

Fig 5.

Effect of bursa aurealis transposon and deletion mutations on capsule production. Capsules isolated were serially diluted (3-fold) and analyzed by immunoblotting.

Conclusion.

In this study, we constructed a ClpC^trap variant to capture substrates that interact with ClpC and identified more than 100 potential substrates. The captured substrates included those that have previously been shown to directly interact with ClpC in low-GC Gram-positive bacteria as well as many novel ClpC substrates. To further verify the proteomic results, we employed a bacterial two-hybrid system and demonstrated that four of the captured proteins were capable of interacting with ClpC in E. coli. Recently, Feng et al. (12) employed the ClpP^trap method in a clpX mutant of strain 8325-4 to identify ClpCP substrates that are targeted for proteolysis. Of the 31 proteins identified by them, 10 were also identified in our study, including Prs, GuaB, NrdE, RecA, CodY, DnaJ, DnaK, ClpB, ClpC, and GlmS. Our results are in good agreement with their report, although a number of proteins were not captured by ClpC^trap in our study. These discrepancies could be due to differences in the strains used or to differences in growth conditions between the studies. However, since we employed a more sensitive GeLC-MS/MS method than the matrix-assisted laser desorption ionization–time of flight method employed by Feng et al., it is possible that some of the substrates identified in this study are proteolytic substrates of ClpCP that were not detected by Feng et al. Moreover, it seems more likely that many proteins that we identified may interact with ClpC without being delivered to the ClpCP protease complex for degradation, suggesting that ClpC could have a biological impact independent of ClpP. Little is known regarding how ClpC, ClpA, and ClpX affect gene expression, other than by association with their proteolytic partner, ClpP. This is true even in highly developed model bacteria, such as E. coli and B. subtilis (8, 9). In S. aureus, ClpX has been shown to affect protein A independently of ClpP, but the mechanism is unknown (61, 62). In our laboratory, we found that ClpC could affect capsule and other virulence factors by a ClpP-independent mechanism (unpublished data). Thus, unraveling how the ClpC ATPase affects its substrates independently of ClpP will undoubtedly lead to further understanding of the regulatory function of ClpC and its role in pathogenesis.