Biochemical Journal

Research article

X-ray structural studies of the entire extracellular region of the serine/threonine kinase PrkC from Staphylococcus aureus

Alessia Ruggiero , Flavia Squeglia , Daniela Marasco , Roberta Marchetti , Antonio Molinaro , Rita Berisio

Abstract

Bacterial serine/threonine kinases modulate a wide number of cellular processes. The serine/threonine kinase PrkC from the human pathogen Staphylococcus aureus was also shown to induce germination of Bacillus subtilis spores, in response to cell wall muropeptides. The presence of muropeptides in the bacterial extracellular milieu is a strong signal that the growing conditions are promising. In the present paper, we report the X-ray structure of the entire extracellular region of PrkC from S. aureus. This structure reveals that the extracellular region of PrkC, EC-PrkC, is a linear modular structure composed of three PASTA (penicillin binding-associated and serine/threonine kinase-associated) domains and an unpredicted C-terminal domain, which presents the typical features of adhesive proteins. Using several solution techniques, we also found that EC-PrkC shows no tendency to dimerize even in the presence of high concentrations of muropeptides. X-ray structural results obtained in the present study provide molecular clues into the mechanism of muropeptide-induced PrkC activation.

  • cell wall
  • penicillin binding-associated and serine/threonine kinase-associated domain (PASTA domain)
  • peptidoglycan
  • PrkC
  • serine/threonine kinase
  • Staphylococcus aureus

INTRODUCTION

Signal transduction through reversible protein phosphorylation is a key regulatory mechanism of both prokaryotes and eukaryotes. Phosphorylation frequently occurs in response to environmental signals and is mediated by specific protein kinases. Generally, phosphorylation is coupled to dephosphorylation reactions catalysed by protein phosphatases.

Eukaryotic-type serine/threonine kinases are expressed in many prokaryotes including a broad range of pathogens. The first reported eukaryotic-type serine/threonine kinase, Pkn1 from Myxococcus xanthus, was found to be required for normal bacterial development [1]. The advance of genome sequencing has prompted the identification of similar kinases in many bacteria. These kinases are now known to regulate various cellular functions, such as biofilm formation [2], cell wall biosynthesis [3,4], cell division [3], sporulation [2,5] and stress response [6].

Previous studies have found that the eukaryotic-type serine/threonine kinase PrkC from Bacillus subtilis is also involved in bacterial exit from dormancy [7,8]. Under conditions of nutritional limitation, B. subtilis produces spores which are resistant to harsh environmental conditions and can survive in a dormant state for years [7,911]. The process of resuscitation is called, in these sporulating bacteria, germination [7,8]. Generally, growing bacteria release muropeptides into the surrounding environment, owing to cell wall peptidoglycan remodelling associated with cell growth and division [12,13]. Therefore the presence of muropeptides in the close environment of dormant spores is a clear signal that the conditions are optimal for growth. Consistently, Shah et al. [7] reported that m-Dpm (meso-diaminopimelate)-containing muropeptides are powerful germinants of B. subtilis spores. Notably, other authors have proposed independently the idea that the activating ligand of PrkC could be a component or a degradation product of the cell wall peptidoglycan [6]. These authors also showed that, once activated, PrkC is able to phosphorylate the small-ribosome-associated GTPase CpgA, the translation factor EF-Tu (elongation factor thermo-unstable) and a component of the bacterial stressosome, denoted YezB [6]. On the other hand, the possible involvement of EF-G (elongation factor G) as a substrate of PrkC is controversial [6,7,14].

A close homologue of PrkC exists in Staphylococcus aureus, a significant human pathogen that causes a number of infections ranging from skin infections to toxic shock syndrome, osteomyelitis and myocarditis [15,16]. PrkC from S. aureus is predicted to be a membrane protein, similar to its homologue from B. subtilis [2]. The two kinases present the same domain organizations, with an intracellular serine/threonine kinase domain, a transmembrane region and an extracellular portion that contains three domains known as PASTA (penicillin binding-associated and serine/threonine kinase-associated) domains. Such domains also exist in penicillin-binding proteins [17]. The crystal structure of the penicillin-binding protein PBP2x from Streptococcus pneumoniae, which contains two C-terminal PASTA domains, was solved in complex with cefuroxime, a β-lactam antibiotic mimicking the unlinked peptidoglycan [18,19]. In this structure, cefuxomine binds one PASTA domain [18,19], a finding which has suggested that PASTA domains of PrkC can bind muropeptides [7]. According to Shah et al. [7], B. subtilis spores germinate in response to m-Dpm-containing muropeptide, which constitutes the B. subtilis cell wall, but not in response to L-lysine-containing muropeptide. Moreover, when the PrkC from B. subtilis is replaced by PrkC from S. aureus, which is characterized by L-lysine-containing cell walls, germination is observed both in response to m-Dpm and L-lysine-containing muropeptide. This finding has shown that the source of PrkC determines the bacterial ability to respond to muropeptides and has suggested that PrkC extracellular domains exhibit specificity of muropeptide binding [7]. Despite these hypotheses, structural results are still needed to help the understanding of the function of extracellular domains.

To obtain an insight into the mechanism of regulation of PrkC activity, mediated by the kinase extracellular domains, we undertook a structural study of the extracellular portion of PrkC from S. aureus (EC-PrkC, residues 378–664, 33 kDa). The increase in S. aureus strains resistant to methicillin, a β-lactam antibiotic of the penicillin class, makes infections particularly troublesome and difficult to treat in hospitals, where patients are at greater risk of infection than the general public. The present structural study provides a valuable template for the rational design of new antibacterial molecular entities able to interfere with cellular processes regulated by PrkC.

EXPERIMENTAL

Protein expression and purification

The plasmid construct corresponding to the extracellular PrkC of S. aureus (EC-PrkC, residues 378–664) has been prepared as previously described by Shah et al. [7]. The overexpression of EC-PrkC, containing an N-terminal His6 tag was carried using Escherichia coli DH5α cells (Invitrogen), as described in [20]. Purification was performed by affinity chromatography, using a 5 ml Ni-NTA (Ni2+-nitrilotriacetate) column (GE Healthcare). A further purification step included gel filtration, on a Superdex 200 column (GE Healthcare), equilibrated in 150 mM NaCl, 20 mM Tris/HCl (pH 8.0) and 5% (v/v) glycerol. The protein, which eluted in a single peak, was concentrated using a centrifugal filter device (Millipore), and the concentration was determined using the Bradford protein assay (Bio-Rad Laboratories). Fresh concentrated protein at 10 mg·ml−1 was used for crystallization experiments.

Crystallization, data collection and processing

Crystallization trials were performed at 293 K using the hanging-drop vapour-diffusion method. Preliminary crystallization conditions were set up using a robot station for high-throughput crystallization screening (Hamilton STARlet NanoJet 8+1) and commercially available sparse-matrix kits (Index Screen and Crystal Screen kits I and II, Hampton Research). Optimization of the crystallization conditions was performed manually both using the vapour-diffusion and the macro-seeding technique. LuCl3-derivative crystals were prepared by soaking a native crystal in a solution containing 1–3 mM LuCl3, 25% (w/v) MPEG2000 [methoxypoly(ethylene glycol) 2000 kDa], 160 mM (NH4)2SO4 and 60 mM sodium acetate trihydrate buffer for 3 h at pH 4.6. A MAD (multi-wavelength anomalous diffraction) experiment was carried out on a crystal derivatized with 2 mM LuCl3 at the X12 synchrotron beamline, DORIS storage ring, DESY (Deutsches Elektronen Synchrotron; Hamburg, Germany), at 100 K. Cryoprotection of the crystals was achieved by a fast soaking in a solution containing glycerol to a final concentration of 10% (v/v). These data, extending to 3.0Å (1Å=0.1 nm) resolution, allowed us to build a part of the molecule [20]. Derivatization by overnight soaking in a solution containing 2 mM EuCl3 provided higher resolution X-ray data, at 2.2Å (Table 1). Data collection was performed in-house at 100 K using a Rigaku Micromax 007 HF generator producing Cu Kα radiation and equipped with a Saturn944 CCD (charge-coupled device) detector. The datasets were scaled and merged using the HKL2000 program package [21] (Table 1).

View this table:
Table 1 Data collection and refinement statistics

Values in parentheses are for the highest-resolution shell (2.28–2.20Å)

Structure determination

Initial phasing was performed on MAD X-ray data using the AutoRickshaw platform using the anomalous signal from the Lu3+ ions [22]. Better phasing was achieved using in-house SAD (single-wavelength anomalous diffraction) data, collected on a crystal soaked overnight in a solution containing 2 mM EuCl3. These data provided structural factor phases to 2.2Å resolution. Both SHELXD [23] and SOLVE [24] identified five Eu3+ ions. Phases, improved by phase extension and density modification by RESOLVE [24] and wARP [25], allowed us to trace nearly the entire molecule structure. Crystallographic refinement was carried out against 95% of the measured data using the CCP4 program suite [26]. The remaining 5% of the observed data, which was randomly selected, was used in Rfree calculations to monitor the progress of refinement. Structures were validated using the program PROCHECK [27].

Bioinformatic analysis

A set of tools was used to analyse the structural characteristics of PrkC. The database Pfam was searched to assign the domain organization of the enzyme. BLAST searches were performed to look for possible structures with significant sequence identity. Transmembrane regions were searched using TMPRED and TMHMM [28,29].

MALS (multiple-angle light-scattering) experiments

Purified EC-PrkC was analysed by SEC (size-exclusion chromatography) coupled to a DAWN MALS instrument (Wyatt Technology) and an Optilab™ rEX (Wyatt Technology). Sample (600 μg) was loaded a S200 10/30 column, equilibrated in 50 mM Tris/HCl (pH 8.0) and 0.15 M NaCl. A constant flow rate of 0.5 ml/min was applied. The online measurement of the intensity of the Rayleigh scattering as a function of the angle as well as the differential refractive index of the eluting peak in SEC was used to determine the weight-average molecular mass (Mw) eluted proteins, using the Astra 5.3.4.14 software (Wyatt Technologies). To check whether the oligomerization state of EC-PkrC changed upon binding of muropeptides, 600 μg of sample was incubated overnight with 10-molar excess of a mixture of natural muropeptides at room temperature (22 °C). The protein/muropeptide mixture was then analysed by SEC–MALS under the same conditions used for the protein alone.

In batch mode, samples were prepared at room temperature in 50 mM Tris/HCl and 150 mM NaCl buffer (pH 8.0). A stock solution of EC-PrkC was filtered through a 0.02 μm Millex syringe-driven filter unit (Millipore). After a further measure of protein concentration, samples were prepared at concentrations ranging between 2.5 and 5.0 mg/ml. All measurements were registered in triplicate for an acquisition time of 2 min. The Rh (hydrodynamic radius) of the scattering molecules was derived, using the ASTRA software, from the diffusion coefficient from the Einstein–Stokes equation.

SPR (surface plasmon resonance) experiments

Real-time binding assays were performed on a Biacore 3000 SPR instrument using two different sensorchips. Covalent immobilization of EC-PrkC on a CM5 sensorchip was carried out with EDC [N-ethyl-N′-(3-dimethylaminopropyl)carbodi-imide]/NHS (N-hydroxysuccinimide) amine coupling chemistry in 10 mM acetate buffer (pH 4.0). In parallel, the N-terminal His6 tag of EC-PrkC was immobilized on an NTA (nitrilotriacetate) sensorchip (flow rate 5 μl/min, injection time 7 min). In the covalent CM5 immobilization, residual reactive groups were deactivated by treatment with 1 M ethanolamine hydrochloride (pH 8.5). The reference channel was prepared by activating with EDC/NHS and deactivating with ethanolamine.

SPR experiments were performed with two different immobilization levels: 2800 and 1000 RU (resonance units) in CM5 covalent immobilization and 1063 and 364 RU in NTA His6-tag capturing. A stock EC-PrkC solution was diluted in running buffer [HBS (Hepes-buffered saline: 10 mM Hepes, 150 mM NaCl and 3 mM EDTA, pH 7.4)] and injected as analyte (flow rate 20 μl/min, injection volume 90 μl) at various concentrations in the range 1.0–100 μM. In parallel, EC-PrkC was pre-incubated with a mixture of natural muropeptides (Table 2), in the concentration range 0–100 equivalents. As a control experiment, the muropeptide mixture (Table 2) was also injected as analyte.

View this table:
Table 2 Composition of the muropeptide mixture based on HPLC elution profiles and MS data

Bacterial growth and purification of peptidoglycan

Dried cells of E. coli containing m-Dpm-type peptidoglycan were extracted as described elsewhere [30]. Briefly, cells were suspended in ice-cold water, added dropwise to boiling 8% (w/v) SDS and boiled for 30 min. After cooling to room temperature, the SDS-insoluble material was collected by centrifugation at 9800 g for 30 min. The pellets were washed several times with water until no SDS could be detected. High-molecular-mass glycogen and covalently bound lipoprotein were removed by treatment with α-amylase and trypsin. Further sequential washes with 8 M LiCl, 0.1M EDTA and acetone were performed sequentially.

Preparation of muropeptides, GC (gas chromatography)–MS, HPLC and LC (liquid chromatography)–MS analyses

The isolated peptidoglycan was degraded with muramidase mutanolysin from Streptomyces globisporus ATCC 21553 (Sigma–Aldrich) at 37 °C overnight. The enzyme reaction was stopped by boiling (5 min), and insoluble contaminants were removed by centrifugation at 9800 g for 30 min.

The muropeptides generated were dissolved in 0.5 M sodium borate buffer (pH 9.0), and solid sodium borohydride was added immediately. After incubation for 30 min at room temperature, excess borohydride was destroyed with 2 M HCl. Finally, the samples were adjusted to pH 3–4 with TFA (trifluoroacetic acid). Reduced muropeptides were fractionated by reverse-phase HPLC [30]. Briefly, the muropeptides were eluted with a linear gradient (run time 40 min) from 0 to 17.5% acetonitrile. Detection was by A206 measurements, whereas identification was achieved by ESI+ (ESI is electrospray ionization) MS and ESI–MS spectra in 2% formic acid/methanol (1:1, v/v) and water/methanol (1:1, v/v) respectively (Agilent 1100MSD) as described in [31].

RESULTS

PrkC from S. aureus is a 664-residue-long multidomain protein. Using sequence analysis tools, the PrkC sequence is predicted to embed a serine/threonine kinase domain (residues 1–270), a region of unknown structure and function (residues 271–377) that includes a transmembrane helix (residues 349–373) and an extracellular region, denoted EC-PrkC (residues 378–664). We determined the crystal structure of EC-PrkC, which is predicted to contain three successive PASTA domains (Figure 1A). We overexpressed this fragment of PkrC in E. coli and obtained a soluble form that was amenable for crystallization. Since a sole methionine residue is present in the sequence of EC-PrkC, we tackled structure factor phasing by preparing derivatives of EC-PrkC crystals with several heavy metals. The best results were obtained with lanthanides. Indeed, MAD data collection using a LuCl3-derivatized crystal was performed at DESY. Diffraction data, at 2.9Å resolution, allowed us to trace a significant part of the molecule [20]. Higher diffraction data (2.2Å resolution) were collected in-house on a EuCl3-derivatized crystal. Using the SAD method, these data produced a readily interpretable electron density throughout the entire structure. The final model contained 285 residues, five Eu3+ ions and 125 water molecules. Data collection, refinement and model statistics are summarized in Table 1.

Figure 1 The modular structure of EC-PrkC

(A) Domain prediction of PrkC, according to the Pfam database. IC-PrkC and EC-PrkC are the intracellular and extracellular regions of PrkC respectively. (B) Ribbon representation of EC-PrkC. Helices and β-strands are shown in orange and blue respectively. (C) Schematic diagram of canonical Ig domain topology. The grey strand in the topology sketch (strand β*) is lacking in PrkC-Ig. (D) Schematic diagram of PASTA domain topology.

EC-PrkC has the shape of a golf club

The crystal structure of EC-PrkC revealed that it consists of four consecutive domains. The four domains are arranged sequentially in a golf-club shape, such that only neighbouring domains interact with each other (Figure 1B). Three of the four domains are, as predicted, PASTA domains (Figure 1A). The structure shows that the three PASTA domains display a linear and regular organization. Indeed, each domain exhibits a two-fold symmetry with respect to its neighbouring domains (Figure 1B). In this organization, the sole α-helix of each domain is alternatively located on the two sides of the golf club (Figure 1B). Interestingly, the structure reveals the existence of a fourth domain, at the C-terminal end of the molecule, not predicted by searches in the Pfam database [32]. Furthermore, sequence analyses against the PDB do not identify any significant homologue for this domain.

Three predicted PASTA domains

The present study provides the highest-resolution study of PASTA domains, which are arranged in βαββββ motifs (Figures 1B and 1D). Despite moderate sequence identity between EC-PrkC PASTA domains (from 21 to 27%, Table 3), the structures of these domains are well conserved.

View this table:
Table 3 Structural comparison, as computed using DALI, of PrkC PASTA domains (PASTA1, PASTA2 and PASTA3) with those of PBP2x (the two domains are denoted PBP2x-1 and PBP2x-2) and of PknB (the four domains are denoted MTB-P1–MTB-P4)

The three rows in each comparison represent Z-score, RMSD (Å) and sequence identity (%) respectively.

PASTA domains are constituents of the penicillin-binding PBP2x protein from S. pneumoniae [18]. The crystal structure of PBP2x shows that its two PASTA domains form a compact structure with a pronounced interdomain bending (Figure 2). In this arrangement, each PASTA domain is involved in interactions with both the other PASTA domain and with the dimerization and transpeptidase domains of PBP2x [18]. In contrast, in the structure of EC-PrkC, the three PASTA domains are arranged linearly, with a limited number of interdomain interactions. This finding suggests a high flexibility of the EC-PrkC.

Figure 2 Comparison of EC-PrkC with PBP2x

(A) Superposition of the PASTA domain 1 of PBP2x (PBP2x-1, orange, PDB code 1QME) with the PASTA domain 2 of PrkC (purple). (B) Enlargement of the insertion loop of PrkC PASTA2 and hydrogen-bond interactions mediated by Ser508. (C) Top view of PBP2x structure. The ribbon and surface representations of PASTA domains are coloured orange, whereas representations of the rest of the molecule are shown in green.

The structural comparison of PrkC PASTA domains with those of PBP2x shows a significant conservation of the PASTA fold, despite the low sequence identities, ranging from 5 to 28%, between domains (Table 3). All secondary-structure elements are conserved, although β-strands β4 and β5 do not appear in lower-resolution PBP2x structures [33]. Despite an overall conservation of the fold, however, EC-PrkC PASTA domains present a six-residue insertion between β-strands β2 and β3 (YSDKYP, YNNQAP and YSDDID for PASTA1, PASTA2 and PASTA3 respectively). Furthermore, an SXG (Ser-Xaa-Gly) motif, which does not exist in PBP2x, occurs at the C-terminal end of each PASTA domain. This sequence motif is conserved in many homologous bacterial proteins (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/435/bj4350033add.htm). In each PASTA domain of the EC-PrkC crystal structure, the serine side chain (Ser508 in PASTA2; Figure 2B) of the SXG motif establishes a hydrogen bond with the backbone nitrogen of glycine (Gly510 in PASTA2; Figure 2B) and with the carbonyl oxygen of a neighbouring residue (Ala480 in PASTA2; Figure 2B). These interactions may play a role in keeping a linear arrangement of PASTA domains. Conversely, a contribution to compact head-to-tail arrangement of PBP2x PASTA domains probably comes from their tertiary structure interactions with the rest of the protein structure (Figure 2C).

While this work was in preparation, the solution NMR structure of three overlapping sequences of PknB from Mycobacterium tuberculosis, each embedding two PASTA domains, and the small-angle X-ray scattering study of the entire extracellular PknB portion (four PASTA domains) appeared [34]. Unlike PBP2x, the organization of PASTA domains in the PknB structure is nearly linear [18,34]. However, the overall RMSD (root mean square deviation) calculated on Cα atoms after superposition of the three PASTA domains of EC-PrkC with those of PknB is as high as 9.5Å (after superposition with PknB PASTA domains 2, 3 and 4) and 14.4Å (after superposition with PknB PASTA domains 1, 2 and 3). This is mainly due to different arrangements of PASTA domains in PrkC and PknB (Figure 3). Indeed, when single PASTA domains of EC-PrkC and PknB are superposed, a strong conservation of the domain fold is observed, even in cases when sequence identities are as low as 2–5% (Table 3). On the basis of the observation that the two PASTA domains of PBP2x interact with each other (Figure 2C) [18], the authors proposed that extracellular regions of PknB dimerize to activate the kinase, since only dimerization would bring two PASTA domains in close contact [34]. However, it should be noted that the two PASTA domains of PBP2x are oriented in opposite directions (Figure 2A). Therefore similar interactions would produce PknB dimers that would not allow the catalytic domains to interact (see the Discussion).

Figure 3 Comparison of EC-PrkC with PknB

(A) Ribbon representation of the X-ray structure of EC-PrkC. (B) Ribbon representation of PASTA domain arrangement in PknB, as derived by small-angle X-ray scattering studies; PDB code 1KUI. (C) Superposition of PknB PASTA domain 1 (light grey) with PrkC PASTA domain 2 (dark grey).

PrkC C-terminal domain: an Ig-like domain with a missing strand

The PrkC C-terminal domain (residues 577–664) exhibits an all-β structure, which contains six β-strands arranged in a β-sandwich (Figure 1C). Sequence alignment analyses do not identify any structure in the PDB with significant sequence identity. However, a DALI [35] search of the PDB reveals that PrkC C-terminal domain structurally resembles a set of Ig domains (Table 4). The canonical structure of Ig domains is formed by a three-stranded and a four-stranded β-sheet packed in β-sandwich arrangement [36]. Like canonical Ig domains, the PrkC Ig domain (PrkC-Ig) presents a hydrophobic core formed by both β-sheets, whereas residues pointing to the solvent are mainly hydrophilic; this results in alternating polar/non-polar sequence patterns of the β-strands. The topology of PrkC-Ig can be classified as belonging to the s-type Ig domains [36]. However, superposition of PrkC-Ig to those identified with DALI (Table 4) shows that this domain lacks the N-terminal strand (β-strand β* of the canonical Ig domain; Figure 1C). The sole identified protein that shares a similar feature is a type III domain of human fibronectin (PDB code 2H41; Table 4). In EC-PrkC, the N-terminal lacking strand of PrkC-Ig forms the linker with the PASTA3 domain (Figure 1). Different from typical Ig folds, where hydrophobic residues in the β* strand pack against the hydrophobic core, the PrkC linker is highly hydrophilic. The volume occupied by hydrophobic residues of the β* strand in canonical Ig domains is filled by bulky hydrophilic residues of β-strands β1 and β6 in EC-PrkC (results not shown).

View this table:
Table 4 Structural comparison of the PrkC Ig domain with those, identified using DALI, with the highest structural similarity (Z-score>5)

It is worth noting that Ig domains similar to PrkC-Ig are usually involved in cell–cell interactions and cell signalling [37]. Their adhesive properties are exploited both by bacterial [38,39] and eukaryotic cells [40], and are typically due to the exposure of an anomalously large number of backbone β-strand hydrogen-bond donors and acceptors [41]. This is particularly true when Ig folds are not complete and lack one β-strand (as in PrkC-Ig; Figure 1C), since this feature leads in extreme cases to Ig polymerization [42,43]. These properties of PrkC-Ig suggest that this domain may also play a role in peptidoglycan binding, in a fashion similar to the E. coli adhesin PapG with host glycolipids [44,45]. Consistently, we noted that B. subtilis PrkC, which shares similar properties in inducing bacterial sporulation, also contains a similar domain at its C-terminus.

EC-PrkC oligomerization state in the crystal state and in solution

Analysis of the packing interactions between symmetry-related molecules of EC-PrkC crystal structure, using the software PISA [46], found no strong protein interfaces that would anticipate the existence of an oligomeric form of the protein. Given the importance of dimerization in the activation of several kinases [37,47], we decided to check it using several techniques and experimental conditions.

Gel-filtration profiles of freshly purified EC-PrkC showed the existence of a species with a molecular mass of approx. 66 kDa, compatible with a dimeric arrangement of the protein (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/435/bj4350033add.htm). However, estimates of molecular mass by gel filtration depend on the protein shape and abnormal values are calculated for non-globular elongated molecules.

We therefore used analytical SEC, coupled with MALS. MALS is a powerful tool to characterize the Mw of compounds showing anomalous elution profiles in SEC. The online measurement of the intensity of the Rayleigh scattering as a function of the angle as well as the differential refractive index of the eluting peak in SEC was used to determine Mw. This analysis produced an Mw value of 31060±620 Da, which corresponds to a monomeric organization of the molecule (Figure 4). In parallel, SEC–MALS experiments were carried out after overnight incubation of EC-PrkC with a mixture of natural muropeptides (Table 2). The results, which provided an Mw value of 31120±311 Da, clearly showed that the protein retains its monomeric state also in the presence of a large excess of muropeptides (Figure 4).

Figure 4 Analytical SEC–MALS of EC-PrkC

Rayleigh ratios and molecular masses (left-hand and right-hand y-axes respectively) are plotted against the elution time. The black curve corresponds to EC-PrkC in 0.15 M NaCl and 50 mM Tris/HCl (pH 8.0). The grey curve was measured in the same buffer after overnight incubation of EC-PrkC with 10-molar excess muropeptides. In both experiments, Mw values correspond to the monomeric state of the protein.

Furthermore, to check whether the molecular size distribution of EC-PrkC changes with increasing protein concentration, we also performed DLS (dynamic light scattering). This analysis was carried out at 25 °C using protein concentrations ranging from 1.0 to 6.0 mg/ml. Under all conditions, a population of 2.8±0.3 nm Rh particles dominated, and no significant variations in Rh values were observed as a function of concentration. Similarly, we measured the molecule Rh after its incubation with a 4-molar excess of natural muropeptides mixture (Table 2). As a result, we observed no significant variation in Rh. Therefore high concentrations do not induce EC-PrkC dimerization, also after incubation with muropeptides.

The possible formation of a complex deriving from self-recognition and the effect of muropeptides on this process were also evaluated using SPR. Real-time binding assays were performed on a Biacore 3000 SPR instrument. EC-PrkC was immobilized using two different methods to ensure different experimental conditions and protein orientations on the chip: on a CM5 sensorchip by covalent EDC–NHS amine coupling and on an NTA sensorchip, which immobilizes the protein N-terminal His6 tag on NTA-chelated nickel of the chip. Furthermore, to investigate the effect of the ligand concentration on potential protein–protein interactions, we used two immobilization levels (see the Experimental section).

In both experiments, EC-PrkC was injected as analyte at concentrations ranging from 1.0 to 100 μM. The possible effect of muropeptides on protein–protein interactions was evaluated by injecting EC-PrkC after pre-incubation with the muropeptide mixture (Table 2) from 0 to 100 equivalents. As a result, no significant RU variation was observed in all SPR assays (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/435/bj4350033add.htm). These results confirmed those obtained by SEC–MALS and DLS that EC-PrkC has no tendency to self-aggregate and that muropeptides do not induce EC-PrkC dimerization.

DISCUSSION

In the present study, we performed X-ray structural characterization of the entire extracellular region of the serine/threonine kinase PrkC from S. aureus, EC-PrkC. This kinase has been shown to signal peptidoglycan-induced revival of B. subtilis from dormancy, similar to B. subtilis endogenous PrkC [7]. The mechanism of peptidoglycan-induced activation of PrkC is, however, so far unknown.

We have shown that EC-PrkC adopts an elongated structure composed of four domains, three PASTA domains and a C-terminal domain, which was not predicted by searches in the Pfam database (Figure 1A). This latter domain, denoted PrkC-Ig, structurally resembles an Ig fold, despite the low sequence identity with Ig folds in the PDB. It is worth noting that PrkC from S. aureus and from B. subtilis share the same domain architecture and a reasonable sequence identity (approx. 27%). A domain similar to S. aureus PrkC-Ig exists also in PrkC from B. subtilis, as suggested by the significantly high sequence identity (30%) between the C-terminal regions of the two homologous proteins. It is therefore not surprising that PrkC from the two bacteria induce similar responses in terms of germination of B. subtilis spores [7].

The kinase domain of PrkC from S. aureus is homologous with that of the serine/threonine kinase PknB from M. tuberculosis (37% sequence identity). Similar to PknB [48] and to PrkC from B. subtilis [5], the kinase domain of PrkC from S. aureus undergoes autophosphorylation [49]. Furthermore, the X-ray structure of the kinase domain of PknB suggests a model in which a structural and functionally asymmetric ‘front-to-front’ association occurs. This dimerization mode leads to the phosphorylation of serine and threonine residues located in the kinase-activation loop [48]. Therefore the most obvious mechanism to explain the muropeptide-driven activation of PrkC involves the muropeptide-induced dimerization of EC-PrkC that leads to the formation of the asymmetric active PrkC dimer.

Dimerization is a typical event that activates kinases, since it permits autophosphorylation. However, the structural basis of this phenomenon may be different. Indeed, dimerization can either be intrinsic, although stabilized by the activating ligand, or entirely induced by the activating ligand [37]. Our data demonstrate that EC-PrkC displays no intrinsic tendency to dimerize (Figure 4) and that the sole muropeptide fails to induce dimerization even in the presence of a large muropeptide excess (Figure 4 and Supplementary Figure S3). It is worth noting that the extracellular region and, to a lesser extent, the transmembrane region of PrkC from B. subtilis are able to form dimers in vivo [2]. Therefore our data point to a more complex dimerization mechanism, which may require the contribution of the transmembrane domain [2] and may be only indirectly modulated by muropeptides.

Furthermore, the dimerization of the extracellular PrkC observed in vivo [2] may involve a third molecule, yet unknown, in a fashion similar to that observed for the FGFR (fibroblast growth factor receptor) [50]. Indeed, dimerization of FGFR involves three Ig-like domains (domains D1–D3) and requires the simultaneous participation of both a FGFR and heparin, through the formation of a ternary complex [37,50]. It cannot be excluded, however, as observed previously for other kinase receptors [51], that the binding of the sole muropeptide molecule is not sufficient to induce dimerization, but it stabilizes a conformation of the extracellular region that favours the dimerization of the entire molecule. Finally, our work provides a high-resolution structure of the extracellular, and therefore accessible to drugs, region of PrkC. As such, they will be an important target for the development of novel antibacterial molecular entities able to interfere with processes involving PrkC.

AUTHOR CONTRIBUTION

Alessio Ruggiero and Flavia Squeglia carried out recombinant protein production, crystallization and data collection and SLS experiments Daniela Marasco carried out SPR experiments, Roberta Marchetti carried out muropeptide preparation, and Rita Berisio conceived the study, wrote the paper and carried out structure solution, refinement and analysis.

FUNDING

We acknowledge COST (European Cooperation in the field of Scientific and Technical Research) action BM1003, ‘Microbial cell surface determinants of virulence as targets for new therapeutics in Cystic Fibrosis’.

Footnotes

  • The structural co-ordinates reported for Staphylococcus aureus PrkC will appear in the PDB under accession code 3PY9.

Abbreviations: DESY, Deutsches Elektronen Synchrotron; DLS, dynamic light scattering; EC-PrkC, extracellular region of PrkC; EDC, N-ethyl-N′-(3-dimethylaminopropyl)carbodi-imide; ESI, electrospray ionization; FGFR, fibroblast growth factor receptor; MAD, multi-wavelength anomalous diffraction; MALS, multiple-angle light-scattering; m-Dpm, meso-diaminopimelate; NHS, N-hydroxysuccinimide; Ni-NTA, Ni2+-nitrilotriacetate; NTA, nitrilotriacetate; PASTA, penicillin binding-associated and serine/threonine kinase-associated; RMSD, root mean square deviation; RU, resonance units; SAD, single-wavelength anomalous diffraction; SEC, size-exclusion chromatography; SPR, surface plasmon resonance

References

View Abstract