McsB Is a Protein Arginine Kinase That Phosphorylates and Inhibits the Heat-Shock Regulator CtsR

See allHide authors and affiliations

Science  05 Jun 2009:
Vol. 324, Issue 5932, pp. 1323-1327
DOI: 10.1126/science.1170088


All living organisms face a variety of environmental stresses that cause the misfolding and aggregation of proteins. To eliminate damaged proteins, cells developed highly efficient stress response and protein quality control systems. We performed a biochemical and structural analysis of the bacterial CtsR/McsB stress response. The crystal structure of the CtsR repressor, in complex with DNA, pinpointed key residues important for high-affinity binding to the promoter regions of heat-shock genes. Moreover, biochemical characterization of McsB revealed that McsB specifically phosphorylates arginine residues in the DNA binding domain of CtsR, thereby impairing its function as a repressor of stress response genes. Identification of the CtsR/McsB arginine phospho-switch expands the repertoire of possible protein modifications involved in prokaryotic and eukaryotic transcriptional regulation.

One of the most intensely studied stress-response pathways is the bacterial heat-shock response. In the Gram-positive model organism Bacillus subtilis, the heat-shock response is mediated by a complex regulatory network (1, 2) that is under control of at least four major transcriptional regulators, including the alternative sigma factor σB (3), the two-component response regulator CssR (4), and the repressors HrcA (5) and CtsR (6, 7). The latter factor, CtsR, controls the expression of genes encoding the HSP100/Clp chaperones and the protease ClpP (6, 8) that constitute the core of the bacterial protein quality control system (9, 10). CtsR is encoded by the first gene of the clpC operon that includes ctsR, mcsA, mcsB, and clpC (6). The dimeric repressor consists of an N-terminal domain with a helix-turn-helix (HTH) motif and a C-terminal domain of unknown function (11). In B. subtilis, CtsR represses transcription of the clpC heat shock operon and the clpE and clpP genes by binding specifically to a seven-nucleotide direct repeat sequence located upstream of the transcriptional start sites (7). Stress-induced transcription of the clp genes depends on the inactivation of CtsR by McsB (12). McsB shows pronounced homology to phosphagen kinases (PhKs) and has been reported to exhibit tyrosine kinase activity (12, 13). Under normal growth conditions, McsB is captured and inhibited by ClpC. However, when bacteria are exposed to stress situations, the ClpC chaperone preferentially interacts with misfolded proteins. It is assumed that the released McsB can now form a complex with CtsR, thereby displacing it from DNA and inducing the expression of heat-shock genes (14). Alternatively, the phosphorylation of CtsR by McsB may be critical for the release of the repressor from DNA (12). To clarify and delineate the precise function of CtsR and McsB in the bacterial stress response, we screened the respective proteins from various Gram-positive bacteria for recombinant production and succeeded in reconstituting the Bacillus stearothermophilus CtsR/McsB system in vitro.

To uncover how McsB modulates the repressor activity of CtsR, we performed electrophoretic mobility shift assays (EMSAs) (Fig. 1A). Addition of CtsR to the 258–base pair (bp) clpC promoter containing three ctsr half sites led to a substantial band shift caused by the formation of a CtsR4/DNA complex. Addition of McsB yielded two lower migrating bands that represent CtsR2/DNA and free DNA. The McsB-dependent release of CtsR was observed only in the presence of Mg/adenosine triphosphate (ATP), whereas addition of EDTA or phosphatase counteracted the effect of McsB. Because no protein-protein interaction could be detected by native gel analysis or size exclusion chromatography, we speculated that McsB and CtsR interact transiently and that phosphorylation of CtsR by McsB abolishes its binding to DNA. To test this hypothesis, CtsR was incubated with McsB in the presence of ATP, and subsequently, phosphorylated CtsR (CtsR-P) was separated from nonphosphorylated CtsR by heparin affinity chromatography (Fig. 1B). Mass spectrometry (MS) analysis of CtsR-P revealed two protein species with either one or two phosphate moieties per protomer (Fig. 1B). In contrast to unmodified CtsR, the isolated CtsR-P cannot bind to its target DNA, as deduced from isothermal titration calorimetry (ITC) and gel-shift experiments (Fig. 1, C and D). Removal of the phosphate group by alkaline phosphatase fully restored the DNA binding capability of CtsR. Thus, phosphorylation of CtsR by McsB is sufficient to inhibit the repressor function of CtsR.

Fig. 1

Phosphorylation of CtsR impedes DNA binding. (A) EMSA analysis of the DNA binding capability of CtsR in the presence of McsB. CtsR was incubated with a clpC promoter fragment, McsB (+, 2 μM; ++, 8 μM), EDTA, and phosphatase (P-ase), as indicated. The promoter fragment, which was visualized by ethidium bromide staining of the native polyacrylamide gel, was either bound to one (CtsR2/DNA) or two (CtsR4/DNA) CtsR dimers. (B) Schematic presentation of the separation of CtsR-P from CtsR and McsB by heparin chromatography (left) and deconvoluted MS spectra of CtsR (average mass of 19047.2 daltons) and CtsR-P (19127.2 and 19207.1 daltons for mono- and di-phosphorylated isoforms, respectively) (right). (C) ITC analysis of CtsR2/DNA complex formation. The 26-bp DNA duplex containing the ctsr box was injected into the sample cell containing either CtsR or CtsR-P (inset). The area under each peak was integrated and plotted against the molar ratio DNA/CtsR inside the sample cell. Thermodynamic values of CtsR/DNA complex formation are Kd = 22.2 ± 3.0 nM and n = 0.53 (reflecting the stoichiometry of bound DNA per CtsR protomer), whereas DNA binding of CtsR-P could not be detected by ITC. (D) EMSA analysis of the DNA binding capability of CtsR and CtsR-P, before and after phosphatase treatment.

To understand how phosphorylation of CtsR affects DNA binding, we determined the crystal structure of CtsR bound to a 26-bp DNA derivative of the clpC promoter (table S1). The CtsR2/DNA structure revealed that the CtsR protomer is composed of two distinct domains: (i) an N-terminal DNA binding domain that adopts the winged HTH fold (residues 2 to 72) and (ii) a C-terminal dimerization domain (residues 79 to 153) that consists of four α helices organized in a four-helix bundle (Fig. 2A). The DNA reading heads of the major and minor groove comprise the recognition helix of the HTH motif and the extended β-hairpin wing, respectively. Key residues for recognizing and binding the ctsr consensus sequence are indicated in Fig. 2A and fig. S3. After obtaining a molecular model of the CtsR/DNA complex, we used MS to pinpoint individual phosphorylation sites. Our initial analyses of “in-solution” and “in-gel” digested CtsR-P were not successful; thus, we attempted to sequence mono-phosphorylated CtsR in a “top-down” MS experiment (Fig. 2B). Purified CtsR-P was directly infused into the mass spectrometer and fragmented by different techniques including electron-capture dissociation (ECD), collisionally activated dissociation (CAD), and infrared multiphoton dissociation (IRMPD). Mapping of the resulting modified protein fragments to the CtsR amino acid sequence revealed that the phosphorylation sites reside in the winged HTH domain. Furthermore, the broad distribution of modified fragments pointed to the existence of product isoforms with different phosphorylation sites. The highest probability for a phosphorylation event was observed for the region Tyr55 to Asp82, making up the β-hairpin wing (Fig. 2A). A lower, albeit still substantial, number of modified fragments matched the N-terminal segment from Ser18 to Tyr55.

Fig. 2

Identification of arginine phosphorylation sites of CtsR. (A) Ribbon diagram showing the CtsR dimer (gray, with labeled domains) bound to the DNA direct repeat motif (green). The identified CtsR phosphopeptide I57VESKpRGGGGYIRIM71, which constitutes the β-hairpin of the winged HTH domain penetrating the DNA minor groove, is highlighted in orange. The lower panel illustrates the binding mode of Arg62 (orange), the main phosphorylation site, at the floor of the DNA minor groove (green). (B) Phosphosite mapping with top-down MS. The mono-phosphorylated isoform of full-length CtsR was sequenced by three different fragmentation techniques. The blue line represents the average (ave) value of the three experimental setups and refers to the number of fragments additionally identified in CtsR-P, relative to unmodified CtsR. The residue with the highest phosphorylation score was Arg62. (C) ECD-MS/MS spectrum of the major phosphopeptide I57 VESKpRGGGGYIRIM71 obtained after chymotryptic cleavage of phosphorylated CtsR. Individual fragments are labeled according to the c- or z-ion nomenclature. The characteristic mass difference of the phosphorylated Arg62 is highlighted, and the threefold charged precursor ion is marked with an asterisk. m/z, mass/charge ratio.

To identify individual CtsR phosphorylation sites, we established a modified protocol for sample preparation and MS analysis (15). Most importantly, we implemented ECD and CAD fragmentation in two parallel MS/MS experiments. Only the ECD MS/MS spectrum of the phosphorylated CtsR peptide I57VESKpRGGGGYIRIM71 (16) allowed the unambiguous identification of Arg62 as the site of modification (Fig. 2C). Both c- and z-fragment ion series unveiled a fragment of 236.067 daltons, reflecting the addition of a phosphate moiety (79.966 daltons) to an Arg residue (156.101 daltons). Moreover, CAD MS/MS of the I57VESKpRGGGGYIRIM71 phospho-peptide resulted in a discrete mass shift of 98 daltons, indicating the loss of phosphoric acid (fig. S1). This fragmentation behavior argues against a tyrosine kinase activity of McsB because phospho-tyrosine is stable upon CAD fragmentation (17). Further MS analysis led to the identification of two additional phosphorylation sites, Arg28 and Arg49 (fig. S2). Consistent with the results of the top-down approach, these amino acids are located within the winged HTH domain. Moreover, all Arg residues are strictly conserved in the CtsR protein family and play a crucial role in DNA binding, as predicted by our crystal structure. Arg62 is a residue within the β wing and deeply invades the minor groove of the DNA duplex. In addition to undergoing extensive van der Waals contacts, the guanidinium group of Arg62 forms hydrogen bonds with the DNA backbone and with one of the thymine pyrimidine carbonyls (Fig. 2A). Similarly, in the major groove of the CtsR consensus site, Arg28 and Arg49 bind to purine bases and coordinate the sugar-phosphate backbone, respectively (fig. S3).

To explore the functional relevance of the identified phosphosites, we conducted a mutational analysis of full-length CtsR by introducing various Arg-to-Lys mutations. Mutating the target sites in position 28, 49, and 62 (3RK) did not completely abolish, but did substantially reduce the phosphorylation of CtsR by McsB (Fig. 3A). Moreover, a mutant protein (8RK), in which the eight Arg residues located in the DNA binding region were replaced by Lys residues, was completely unsusceptible to McsB modification. Reintroduction of Arg62 (7RK) markedly restored the phosphorylation potential. To study the direct effect of CtsR phosphorylation on DNA binding, we replaced Arg62 by a phosphomimicking Glu residue. EMSA experiments clearly demonstrated that the Arg62 → Glu62 (R62E) mutant lost its capability to bind DNA (Fig. 3B), thus corroborating our finding that phosphorylation of CtsR alone is sufficient to inhibit its repressor activity. Conversely, replacing Arg62 by Lys62 did not alter the DNA binding ability of CtsR in band-shift assays. To test which state of CtsR is targeted by McsB, we incubated the kinase with DNA-bound and -unbound CtsR. Following the interaction with DNA over time revealed that McsB preferentially phosphorylates free CtsR, thereby preventing DNA complex formation (fig. S4). We conclude that the selective introduction of a negatively charged phosphate moiety functions as a molecular switch regulating DNA binding. Whereas the unphosphorylated CtsR binds with high affinity to its DNA consensus site and inhibits transcription of downstream genes, the McsB-phosphorylated CtsR repressor is not able to bind to DNA, thus allowing heat-shock gene expression.

Fig. 3

Characterization of McsB-mediated arginine phosphorylation. (A) Phosphorylation level of CtsR Arg mutants analyzed by electrospray ionization–MS. The Arg-to-Lys mutants are 28/49/62 (3RK), 28/36/49/54/62/69/114/125 (8RK), and 28/36/49/54/69/114/125 (7RK). (B) DNA binding ability of different CtsR mutants in gel-shift assays. (C) Peptide phosphorylation assay (schematically shown in the inset). (Left) Matrix-assisted laser desorption/ionization–time-of-flight spectra of selected peptides after incubation with McsB. Non-phosphorylated and phosphorylated peptides are marked in gray and green, respectively. (Right) Effect of the exchange of Arg to other potential phospho-acceptor sites (shown in red) on the phosphorylation efficiency.

To verify our finding that McsB is a protein arginine kinase, we established an in vitro phosphorylation assay (Fig. 3C) using synthetic oligopeptides that resembled the CtsR sequence (residues 61 to 73). To avoid side effects during sample preparation that would preclude quantification of the phosphorylation reaction, we replaced one potential oxidation site (Met71) and one arginine (Arg69), yielding the 13-residue model substrate K61RGGGGYIKIIKV73. Systematic incorporation of potential phosphorylation sites (Tyr, Ser, Thr, His, Asp, and Lys) in position 62 revealed that only peptides with an Arg moiety are modified by McsB (Fig. 3C). Moreover, modification of the guanidinium group of Arg62 by asymmetric dimethylation prevented McsB-mediated modification. Additionally, we analyzed the purified phospho-peptide K61RGGGGYIKIIKV73 by 31P nuclear magnetic resonance (NMR) spectroscopy (fig. S5). The chemical shift of about –2.4 parts per million (ppm) fits well to the measured NMR spectrum of free phospho-arginine (–3.0 ppm) (18), suggesting that the phosphate is attached via a phosphoramidate N-P linkage. Corresponding spectra of O-P linked phosphor compounds (as, for example, phospho-tyrosine, -serine, and -threonine) exhibit markedly higher chemical shifts of ~0.7 to 4.0 ppm (18). Thus, McsB is a protein kinase that acts exclusively on Arg residues, phosphorylating one of the amine nitrogens of the guanidinium group.

Phosphorylation of the free amino acid l-arginine by eukaryotic PhKs yields a chemically labile compound (19). We studied the CtsR/McsB system of a thermophilic organism living at ~55°C and thus explored the thermostability of a phosphorylated Arg residue present in a peptide context. For this purpose, we phosphorylated the K61RGGGGYIKIIKV73 peptide with McsB, incubated the purified phospho-peptide at different temperatures, and quantified the stability of the phosphorylation signal by high-performance liquid chromatography–MS analysis. The results clearly showed that peptide arginine phosphorylation is surprisingly stable up to 60°C. Dephosphorylation of the phospho-peptide occurred only at 95°C, with a half life t1/2 of ~130 min (fig. S6). Therefore, phosphorylation of protein arginine residues should represent a relevant biological signal.

Sequence analysis indicated that the McsB protein arginine kinase exhibits no substantial homology to known Ser, Thr, Tyr, or His kinases. However, the catalytic domain of McsB is highly homologous to the catalytic domain of PhKs (12, 13), which are involved in maintaining energy homeostasis but not in intermolecular signaling (20). Mutational analyses revealed that McsB and PhKs use a common mechanism to phosphorylate the terminal guanidinium group of substrates (fig. S7) (14). However, in contrast to PhKs where substrate specificity is primarily determined by the N-terminal domain, McsB harbors a distinct C-terminal domain that may re-direct the substrate specificity from free Arg to protein-incorporated Arg residues.

McsB appears to be the founding member of a new class of protein kinases acting specifically on Arg residues. It should be noted that protein arginine phosphorylation has been reported previously (21). Remarkably, histone H3 was identified as a potential eukaryotic target (22), implying that Arg phosphorylation activity might be relevant for epigenetic regulation. However, these analyses failed to identify the corresponding kinase and obtained only indirect evidence for Arg modification. The thorough characterization of a protein arginine kinase presented in this work should provide the experimental tools to directly address the impact of Arg phosphorylation in prokaryotic and eukaryotic signaling pathways.

Supporting Online Material

Materials and Methods

Figs. S1 to S8

Table S1


  • * These authors contributed equally to the work.

References and Notes

  1. Materials and methods are available as supporting material on Science Online. A detailed description of the MS approach is provided.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  3. We thank D. Fiegen and D. Reinert (Boehringer Ingelheim) and the staff at Swiss Light Source for assistance with collecting synchrotron data, A. Carrieri for his contribution with the kinase assays, L. Becker for the NMR analysis of phosphopeptides, T. Krojer and D. Hellerschmied for their support in the structural analysis of CtsR, and C. Stingl and M. Mazanek for assisting MS analysis. The Research Institute of Molecular Pathology is funded by Boehringer Ingelheim. J.F., S.S., E.C., and T.C. were supported by Wiener Wissenschafts, Forschungs und Technologiefonds; A.S. by the Christian-Doppler-Society; K.T. by the Deutsche Forschungsgemeinschaft; and T.C. by the European Molecular Biology Organization Young Investigator Program. This work was further supported by the Austrian Proteomics Platform (GEN-AU). The Protein Data Bank accession number for the CtsR2/DNA complex is 3H0D.
View Abstract

Stay Connected to Science

Navigate This Article