Identification of catabolite control protein A from Staphylococcus aureus as a target of silver ions† †Electronic supplementary information (ESI) available: Experimental procedures, supplementary figures and tables. See DOI: 10.1039/c7sc02251d

The catabolite control protein A (CcpA) of S. aureus has been identified as a putative target for silver ions.


Introduction
The main carbon catabolite repression (CCR) system is an important global control system of various bacteria, which allows the bacteria to adapt quickly to a preferred carbon source rst. This is usually achieved by the repression of genes whose products are involved in the catabolism of alternative, less preferred carbon sources. In Gram-positive bacteria, a highly conserved regulator, catabolite control protein A (CcpA), exerts the important catabolite repression function. 1 CcpA is usually activated by its co-regulator via the formation of a complex which recognizes the catabolite-responsive element (cre) sequences and regulates downstream gene expression. 2 Staphylococcus aureus (S. aureus), a worldwide spread human pathogen, is the leading cause of hospital-and communityacquired infections. The pathogen causes a series of human diseases ranging from minor skin infections to life-threatening sepsis. 3 In particular, the emergence of drug-resistant strains of the bacteria, such as methicillin-resistant and vancomycinresistant S. aureus, poses a huge threat to public health worldwide. 4,5 Intriguingly, S. aureus CcpA (SaCcpA) is not only involved in the regulation of carbon metabolism but also affects antibiotic resistance, biolm formation, toxin expression and even the infectivity of this bacterium, implying its critical role as an important global regulator for bacterial metabolism as well as virulence. [6][7][8][9] Recently, small molecule inhibitors targeting the S. aureus virulence regulators, SarA or MgrA, are reported to be efficacious in animal models, indicating that targeting these regulator proteins might be a promising anti-bacterial strategy. 10,11 Given the important role that CcpA played in S. aureus virulence, this transcription factor could be a feasible anti-bacterial drug target. 12 Chemical inhibition of CcpA binding to the cre DNA region could potentially diminish S. aureus virulence.
Silver ions (Ag + ) have been used as antibacterial agents for centuries. It is suggested that Ag + could bind to the thiol group (-SH) of bacterial enzymes and subsequently cause enzyme deactivation. 13 However, up to now, few Ag + protein targets have been identied and characterized. Herein, we demonstrate that SaCcpA serves as a potential target for Ag + in S. aureus. Ag + binds specically to the two cysteines of SaCcpA and abolishes its cre-binding property, which further abrogates S. aureus a-hemolysin secretion and biolm formation.

Results and discussion
It is reported that silver nanoparticles (AgNPs) could block bacterial sugar metabolism in order to be bactericidal. 14 Furthermore, recent studies demonstrated that bacterial strains with a TCA cycle genes knockout were less sensitive to Ag + treatment. 15 All of these data imply that Ag + targets the bacterial central metabolism pathway. CcpA is an important regulator that coordinates central metabolism in Gram-positive bacteria. 1, 16 We therefore investigate the possible effect of Ag + on CcpA physiological function.
Sequence alignments of 20 CcpA family proteins from different Gram-positive bacteria species revealed that SaCcpA contains two cysteine residues (Cys216 and Cys242), which are almost absent in other species (Fig. S1 †). Given that Ag + is highly thiophilic, we postulated that Ag + could bind to SaCcpA. To test this hypothesis, puried SaCcpA was incubated with 3 molar equivalents of Ag + followed by the removal of excess amounts of Ag + with a desalting column. By using a BCA assay and inductively coupled plasma mass spectrometry (ICP-MS), the stoichiometry of Ag + binding to CcpA (monomer) was determined to be 2 : 1, indicating that each CcpA monomer binds 2 molar equivalents of Ag + (Fig. S2 †). Subsequently, we examined whether the two cysteine residues are involved in the Ag + binding. We measured free thiol amounts of the CcpA protein aer premixing with different molar ratios of Ag + by Ellman's assay. As expected, the free thiols of CcpA decreased with increasing pre-mixed Ag + concentrations until the Ag + / CcpA molar ratio reached 2 : 1, conrming that the cysteines participate in Ag + binding (Fig. S3 †). The two cysteines were then individually mutated to serine. Both the ICP-MS measurement and Ellman's assay showed that the CcpA C216S and CcpA C242S mutants could bind one Ag + per monomer, while the double Cys mutant CcpA C2S had no Ag + binding capability, indicating that both cysteines are responsible for Ag + binding ( Fig. S2 and S3 †). In line with the results, isothermal titration calorimetry (ITC) data showed that wild-type (WT) CcpA binds 1.94 AE 0.02 molar equivalents of Ag + with an apparent dissociation constant (K d ) of 0.74 AE 0.03 mM. The single Cys mutant CcpA C242S binds 1.08 AE 0.03 molar equivalents of Ag + with a much lower affinity (K d ¼ 7.81 AE 0.61 mM), while the double Cys mutant CcpA 2CS had no detectable binding to Ag + ( Fig. 1 and Table S2 †).
As a global transcription factor, CcpA binds to a couple of gene promoter regions (cre sequence), such as the pckA (encoding phosphoenol-pyruvate carboxykinase) and hla (encoding a-hemolysin) promoters. To examine the effect of Ag + binding on the CcpA's function, we investigated whether Ag + affected the CcpA-DNA binding properties in vitro. An electrophoretic mobility shi assay (EMSA) was applied to 35 nM pckA DNA probe (covers the cre sequence of the pckA gene) and a negative control proC probe with increasing concentration of CcpA (0-700 nM monomer concentration). As expected, the signicant shi of DNA was only observed for pckA but not proC ( Fig. S4a and c †). The results are consistent with a previous report that SaCcpA does not essentially require the association with phosphorylated HPr for efficient DNA binding. 8 However, the addition of gradient amounts of Ag + obviously disrupted the complex formation (Fig. 2a). The double-mutant CcpA 2CS binds pckA DNA in a similar way to the WT CcpA (Fig. S4b †). However, the CcpA 2CS mutant would not dissociate from the DNA probe even in the presence of Ag + (Fig. 2b). A similar phenomenon was observed for the hla probe ( Fig. S5 †). CcpA binding to the cre region was enhanced by phosphorylated HPr (HPr-P). 8 However, the EMSA assay demonstrates that Ag + prevented CcpA-DNA binding even in an excess amount of Hpr-P ( Fig. S6a †). Native polyacrylamide gel electrophoresis (PAGE) further conrmed that Ag + binding also disrupted the DNA-CcpA-(Hpr-P) ternary complex ( Fig. S6b †). All these results demonstrate that Ag + binding completely abolishes CcpA-DNA binding in vitro.
For further conrmation, we measured the DNA binding capabilities of the WT CcpA and CcpA 2CS mutant by BioLayer Interferometry (BLI). A Biotin-labeled pckA DNA probe was immobilized on a streptavidin sensor to enable kinetic analysis of the CcpA binding to the DNA probe. As shown in Fig. 3a, WT CcpA binds strongly to the pckA probe with a K d value of 16.1 AE 0.47 nM. While in the presence of Ag + , the binding of WT CcpA to the pckA probe is undetectable (Fig. 3b). In contrast, the binding affinities of CcpA 2CS to the pckA probe are nearly identical in the absence and presence of Ag + , with K d values of 15.1 AE 0.28 nM and 20.7 AE 0.51 nM, respectively (Fig. 3c and  d and Table S3 †). Collectively, these data demonstrate that Ag + binds to the two Cys residues of CcpA, and Ag + binding disrupts its DNA binding capability.
Previous studies demonstrated that the binding of nonphysiological metal ions to proteins usually caused protein aggregation and dysfunction. 17,18 To further investigate the  mechanism of the Ag + -induced loss of DNA-binding capability of CcpA, we examined the oligomerization states of CcpA before and aer Ag + binding using size-exclusion chromatography (SEC). In the absence of Ag + , WT CcpA eluted at 9.6 ml with a molecular weight (M w ) of 64.9 kDa, corresponding to a dimeric form (Fig. 4a). With increasing amounts of pre-incubated Ag + , the intensities of the dimeric peak of CcpA decreased, whereas a new peak appeared at 8.7 ml with a M w of 159.7 kDa, indicative of a tetrameric form of CcpA. This result implied that CcpA forms a dimer of dimers aer Ag + binding. A similar phenomenon was observed for the single Cys mutants of CcpA, CcpA C216S and CcpA C242S , which also formed a tetramer aer incubation with Ag + (Fig. S7 †). In contrast, the majority of the double mutant CcpA 2CS eluted at exactly the same volume as WT CcpA, even in the presence of 2 molar equivalents of Ag + , owing to the loss of the Ag + binding capability of the protein (Fig. 4b).
In line with the SEC results, native PAGE shows that Ag + binding slowed down the migration rates of WT CcpA, and single mutants CcpA C216S and CcpA C242S in the native PAGE, which is indicative of the formation of a higher molecular weight oligomer upon Ag + binding. However, Ag + had no effect on the migration rate of the CcpA double mutant, CcpA 2CS (Fig. S8 †). Taken together, the binding of Ag + to CcpA induces its tetramerization, which is possibly attributable to the loss of DNA binding capability.
Next, we investigated whether CcpA binds Ag + in vivo using the cellular thermal shi assay (CETSA), a method based on the change in protein thermal stability upon ligand binding for studies of the target engagement of drug candidates in a cellular condition. 19,20 As shown in Fig. 5a, supplementations of 10 mM Ag + to the bacterial culture resulted in the apparent aggregation temperature (T agg ) of the intracellular WT CcpA shiing from 49.5 C to 45.4 C, indicating that Ag + binds to CcpA in vivo and such a binding destabilizes the protein. A similar result was obtained when using puried CcpA protein (Fig. S9a †). Since Ag + binds to the two Cys residues of CcpA, it is plausible to hypothesize that Ag + would not change the thermal stability of the double-cysteine mutant CcpA 2CS due to the loss of Ag + binding sites. To verify this, a CcpA gene mutant of S. aureus, the Newman strain, was constructed, in which the WT CcpA gene was replaced by a double cysteine mutant gene CcpA 2CS (denoted as S. aureus ccpA::ccpA 2CS ) and a similar CETSA was performed with the mutant strain. As expected, Ag + treatment did not alter the intracellular CcpA thermal stability in the CcpA mutant strain (Fig. 5b). Similarly, the puried CcpA double mutant CcpA 2CS protein had the same thermal denaturation curves in the absence and presence of Ag + (Fig. S9b †). Collectively, we demonstrated that Ag + binds to CcpA intracellularly via its two Cys residues.
CcpA is the major gene regulator of central metabolism in S. aureus and the CcpA gene knockout was found to retard bacterial growth. 9 Since Ag + binds to CcpA and abolishes its   function in S. aureus, this prompted us to investigate whether Ag + affects S. aureus growth. We examined the bacterial growth of both the WT S. aureus and ccpA::ccpA 2CS mutant strains upon supplementation of 30 mM Ag + into the cultures at the exponential-growth phase (OD 600 ¼ 0.6). As shown in Fig. 6a, both the WT and ccpA::ccpA 2CS mutant of S. aureus display nearly identical growth curves in the absence of Ag + . Both cultures exhibit typical S-shaped growth curves and enter a stationary-growth phase aer around 250 min, indicating that ccpA::ccpA 2CS did not affect S. aureus growth signicantly. In contrast, the growth rates of both the WT and ccpA::ccpA 2CS mutant of S. aureus were remarkably inhibited aer the addition of Ag + until 300 min. However, a clearly different behavior was observed for the WT and mutant cultures aer 300 min, with the WT culture displaying a slower growth, yielding cell density which signicantly lagged behind that of the ccpA::ccpA 2CS mutant culture. Aer 700 min, the mutant culture with the addition of Ag + reached almost the same OD 600 value as the control group, whereas the WT culture with the addition of Ag + only reached approximately 60% of the OD 600 value of the control group. The results indicated that the S. aureus ccpA::ccpA 2CS mutant is less sensitive to Ag + than the WT. In line with this, the IC 50 values of Ag + for the WT and ccpA::ccpA 2CS mutant of S. aureus were calculated to be 79.8 AE 1.1 mM and 94.1 AE 0.8 mM, respectively (Fig. S10 †).
Besides regulation of carbon catabolite repression, CcpA also exerts a critical role in S. aureus virulence factor secretion and biolm formation. 6,7,9 It is reasonable to postulate that Ag + would interfere with these physiological processes of S. aureus by targeting CcpA. Therefore, the effect of Ag + on the virulence factor secretion and biolm formation in both the WT and ccpA::ccpA 2CS mutant S. aureus strains was investigated. The promoter region of the hla gene (encoding a-hemolysin) in S. aureus contains the cre sequence that could be recognized by CcpA. Previous studies demonstrated that the hla transcription level was markedly down-regulated in the ccpA knockout S. aureus strain. 21 Indeed, secreted a-hemolysin and rabbit erythrocyte lysis activity of a S. aureus mutant strain with the CcpA gene knockout (denoted as S. aureus DccpA) is almost undetectable. On the other hand, the S. aureus ccpA::ccpA 2CS mutant exhibited a signicant decrease on secreted a-hemolysin and retained approximately 50% erythrocyte lysis activity compared to the WT strain, implying that the two Cys residues have a potential role in the regulation of a-hemolysin expression (Fig. S11 †). Upon supplementation of 20 mM Ag + , a westernblot showed a signicant decrease in secreted a-hemolysin in WT S. aureus. While in the ccpA::ccpA 2CS mutant strain, Ag + caused no obvious change in the secreted a-hemolysin. Consistently, the rabbit erythrocyte lysis activity of WT S. aureus decreased dramatically by 60% aer Ag + treatment. In contrast, the treatment of Ag + led to much smaller decrease on the lysis activity of the ccpA::ccpA 2CS mutant strain, which still exhibited 85% activity compared to the control group (Fig. 6b and c). The results are consistent with qPCR data, which demonstrated that attenuation of the transcription level of CcpA regulated genes (pckA and hla) was much higher in the WT strain than that in the ccpA::ccpA 2CS mutant strain (Fig. S12 †). Similarly, bacterial biolm formation is inhibited by Ag + to a lesser extent in the ccpA::ccpA 2CS mutant than in the WT strain (Fig. S13 †). It is noteworthy that regulation of a-hemolysin expression and biolm formation in S. aureus are complicated. For example, a-hemolysin expression is affected by multiple regulatory networks, including the global regulators SarA and MgrA. 22,23 It is possible that these regulatory networks were also perturbed upon Ag + treatment, which could at least partially explain the decrease of erythrocyte lysis activity observed in the ccpA::ccpA 2CS mutant strain upon Ag + treatment. Nevertheless, the evident discrepancy observed between the ccpA::ccpA 2CS mutant and WT Fig. 6 (a) The inhibition effect of Ag + on the bacterial growth of the wild-type S. aureus and ccpA::ccpA 2CS mutant strains. The OD 600 was recorded at 20 min intervals, and 30 mM Ag + was added when OD 600 reached 0.6. The secreted a-hemolysin in the supernatant was normalized to OD 600 . (b) Rabbit erythrocyte lysis activities of S. aureus and the S. aureus ccpA::ccpA 2CS mutant with or without 20 mM Ag + . Ag + was added at the beginning of the bacterial culture. All experiments were performed in triplicate. Results are shown as mean AE sd. The haemolytic activities of the wild-type strain without Ag + treatment are used as a control and the mean value is set at 100%. The activities in other groups are normalized to the control. The statistical difference is determined by the two-tailed Student's t-test.
S. aureus upon Ag + treatment conrms that CcpA indeed serves as one of the targets of Ag + in vivo.
All of the results demonstrated that the WT S. aureus strain was more sensitive to Ag + treatment than the ccpA::ccpA 2CS mutant strain due to Ag + binding to the two cysteine residues. To further conrm this, the severity of a S. aureus infection was compared between the WT and ccpA::ccpA 2CS mutant in a murine model. The details are described in the ESI. † In brief, groups of female BALB/c mice were inoculated with S. aureus WT or ccpA::ccpA 2CS mutant strains to develop abscesses on the skin. Twice-daily treatment of AgNO 3 with different concentrations (20 mg ml À1 and 100 mg ml À1 ) was applied onto the abscesses. The skin abscesses were excised 64 h post-infection, homogenized and serially diluted for CFU quantication. As shown in Fig. 7, the control groups infected with the WT or ccpA::ccpA 2CS mutant had similar viable bacterial counts, with log CFU mean values of 8.4 and 8.1 respectively, indicating that double Cys mutation does not signicantly perturb S. aureus viability in a murine model. A low dosage of AgNO 3 treatment (20 mg ml À1 ) did not change the bacterial loads in both the WT and ccpA::ccpA 2CS mutant infected groups. However, a signicant difference was observed when the two infected groups were treated with a high dosage of AgNO 3 (100 mg ml À1 ). In the WT S. aureus infected group, the viable bacterial counts dropped signicantly compared to the control group. In contrast, the bacteria counts were almost the same in the ccpA::ccpA 2CS mutant infected groups upon a high dosage Ag + treatment, conrming that WT S. aureus is more sensitive to Ag + than the ccpA::ccpA 2CS mutant. Intriguingly, a signicant difference on the dermonecrosis of the skin abscess was observed for the mice infected with two different bacterial strains (Fig. S14 †). The mice infected with wild-type S. aureus had much more severe dermonecrosis than those infected with the ccpA::ccpA 2CS mutant strain. Given that a-hemolysin is the major contributor to necrotic lesions, 24 the observation is consistent with the results that the ccpA::ccpA 2CS mutant had a lower a-hemolysin level than the wild-type strain.

Conclusions
CcpA is one of the important global DNA regulators of Grampositive bacteria. Recent transcriptome and proteome analyses revealed that CcpA has broad effects on gene expression in S. aureus, even in the absence of glucose. 25 Particularly, a CcpA gene knockout abrogates biolm formation and virulence factor expression in S. aureus, which remarkably decreases bacterial pathogenesis. We show clearly that Ag + binds to CcpA via the two Cys residues both in vitro and in vivo, leading to the disruption of protein functions, thus attenuating bacterial growth, bacterial toxin expression and biolm formation. Importantly, we demonstrated that WT S. aureus was more sensitive to Ag + treatment than the ccpA::ccpA 2CS mutant in a murine skin infection model. The results herein conrm SaCcpA as an intracellular target for Ag + . It should be noted that metal-based drugs are usually multi-targeted. 26 Although it is commonly believed that the antimicrobial activity of silver is due to its interaction with thiol groups in enzymes and proteins, other cellular components are likely to be involved. [27][28][29] Therefore, identication of Ag-binding proteins at a proteome-wide scale may allow extensive exploration of silver targets to advance our understanding on the bactericidal effects of silver. [30][31][32] However, the physiological function of the two cysteine residues in SaCcpA remains unclear. It has been reported previously that several DNA regulators of S. aureus use the cysteine-based oxidation sensing pathway for regulatory functions. [33][34][35] Whether the two cysteines in SaCcpA are also involved in oxidative sensing may warrant further studies.

Live subject statement
The animal studies strictly followed the recommendations in "Guide for the Care and Use of Laboratory Animals" published by the National Institutes of Health. The protocols were approved by the Committee on the Use of Live Animals in Teaching and Research (CULATR) and the University of Hong Kong (Permit no. 4008-16).

Conflicts of interest
The authors declare that there is no conict of interest. Fig. 7 Murine skin infection model (a) to investigate the sensitivity of S. aureus strains to Ag + treatment. The bacterial load of local abscesses induced by wild-type S. aureus (b) or the ccpA::ccpA 2CS mutant (c) was enumerated in control or AgNO 3 treatment groups (20 mg ml À1 and 100 mg ml À1 ). The log CFU values are presented as the mean AE sd. The statistical difference is determined by the Mann-Whitney U test.