Structural Analysis of Staphylococcus aureus Serine/Threonine Kinase PknB

Effective treatment of infections caused by the bacterium Staphylococcus aureus remains a worldwide challenge, in part due to the constant emergence of new strains that are resistant to antibiotics. The serine/threonine kinase PknB is of particular relevance to the life cycle of S. aureus as it is involved in the regulation of purine biosynthesis, autolysis, and other central metabolic processes of the bacterium. We have determined the crystal structure of the kinase domain of PknB in complex with a non-hydrolyzable analog of the substrate ATP at 3.0 Å resolution. Although the purified PknB kinase is active in solution, it crystallized in an inactive, autoinhibited state. Comparison with other bacterial kinases provides insights into the determinants of catalysis, interactions of PknB with ligands, and the pathway of activation.


Introduction
The gram-positive bacterium Staphylococcus aureus is a serious human pathogen that is responsible for an increasing number of illnesses and deaths each year [1]. The bacterium colonizes the nose and skin of humans and can cause illnesses ranging from skin infections [2] to life-threatening diseases such as endocarditis, bacteremia, pneumonia, meningitis, osteomyelitis, sepsis and the toxic shock syndrome [3,4,5,6]. Successful treatment of S. aureus infections remains a challenge as drug-resistant strains, such as methicillin-resistant and vancomycin-resistant S. aureus (MRSA and VRSA, respectively), are gaining prominence. Furthermore, the emergence of community-acquired S. aureus strains forms a rapidly emerging public health problem [7]. In order to develop new strategies to combat these bacteria, a better understanding of the organisms and the functions of its components is needed.
To overcome stressful conditions imposed by its host, S. aureus has developed various protective and offensive responses such as the sensing of environmental stimuli and the activation and inactivation of response regulators. This is generally achieved through cascades of phosphorylation reactions in the host, which in turn points to a key role of protein kinases in staphylococcal persistence. Protein kinases regulate a multitude of processes and signal transduction pathways in prokaryotes and eukaryotes [8]. A subgroup, the serine/threonine kinases (STKs), was originally thought to only be present in eukaryotic cells. However, in recent years STKs have also been identified in bacteria [9,10], and these have been classified as eukaryotic-type serine/threonine kinases [11]. While many microorganisms encode for several eukaryotictype STKs, S. aureus encodes only for one such protein, which has been termed PknB, PrkC or Stk1 by different research groups [11,12,13] and will be referred to as PknB here. PknB was originally identified through a transposon mutagenesis approach and is conserved in all S. aureus strains [14]. The kinase is composed of an N-terminal, cytosolic kinase domain, a central transmembrane domain, and three C-terminal, extracellular PASTA (penicillin-binding protein and serine/threonine kinase associated) domains (Fig. 1). PASTA domains are constructed from about 65-70 amino acids and are thought to bind betalactam compounds as well as peptidoglycans [15,16]. The number of PASTA domains present in eukaryotic-type STKs can vary. S. aureus PknB and B. subtilis PrkC have both three PASTA domains, while the PknB of M. tuberculosis contains four such repeats [11,12,17].
PknB is of particular relevance for S. aureus survival and pathogenesis as it helps to regulate purine biosynthesis, autolysis, and other central metabolic processes of the bacterium and is involved in antibiotic resistance [12,18,19]. Moreover, recent data show that PknB can also act on human cellular proteins, and that these potential targets are involved in apoptosis, immune responses, transport, and metabolism [20]. The recently discovered secretion of PknB may also help the bacterium to evade intracellular killing and facilitate its growth [20]. Proper function of PknB is important for full expression of S. aureus pathogenesis, and it is also likely that phosphorylation levels controlled by PknB are essential in controlling bacterial survival within the host [21].
Structural information on S. aureus PknB is so far limited to the three PASTA domains that constitute the extracellular portion of the protein [11,17]. Structural analyses of PknB homologs, such as PknB from M. tuberculosis [22,23,24,25,26] have provided insights into the overall fold of the cytosolic kinase domain and its interactions with ligands. However, it is well established that kinases adopt similar folds but differ in subtle ways in order to achieve their specificity. Here, we report a structural analysis of the kinase domain of S. aureus PknB in complex with a noncleavable ATP analog, adenosine 59-(b, c-imido)-triphosphate (AMP-PNP). Comparison with other bacterial STKs provides insights into the determinants of PknB catalysis, its state of activation, and its interactions with potential ligands.

Materials and Methods
Cloning, expression and purification DNA encoding the kinase domain of S. aureus PknB (PknB SA-KD ) (residues 1-291) of S. aureus strain 8352 (GenBank accession number BAB42315) was amplified by PCR, and NdeI and XhoI cleavage sites and an additional stop codon were introduced. After digestion with NdeI and XhoI, the PCR product was inserted into the pET28b vector (Novagen), which includes an N-terminal His 6tag followed by a thrombin cleavage site, for protein expression in E. coli strain BL21-DE3. Transformed bacteria were grown in LB medium supplemented with 30 mg/ml kanamycin at 37uC to an optical density of 0.3 at 600 nm. After lowering the temperature to 20uC, the bacteria were induced by addition of 1 mM isopropyl-bthiogalactopyranoside. After 24 hours of expression, bacteria were harvested by centrifugation and resuspended in 20 mM HEPES, 150 mM NaCl, 20 mM imidazole and 1 mM phenylmethanesulfonyl fluoride at pH 7.4. The sonified lysate was clarified by centrifugation and filtered. The solution was loaded onto a Hi-sTrap column (GE Healthcare), which was then washed with lysis buffer. The His-tagged PknB SA-KD was eluted with a linear imidazole gradient ranging from 10 to 500 mM. After reducing the imidazole concentration by alternating concentration and dilution steps, the protein was cleaved with thrombin (1 U/mg protein; GE Healthcare) for 24 h at 20uC. Cleavage was followed with a second Ni 2+ -affinity run to remove thrombin and uncleaved PknB SA-KD . The 33.1 kDa PknB SA-KD was concentrated and purified by gel filtration (Superdex 75, GE Healthcare) in 20 mM HEPES, 150 mM NaCl at pH 7.4 (Fig. S1A). SDS-PAGE confirmed the purity of the product (Fig. S1B). Mass spectrometry analysis was used to verify the identity of PknB SA-KD , and circular dichroism spectroscopy (CD-spectroscopy, Jasco J -720) confirmed that it was folded (Fig. S1C). PknB SA-KD was used at a concentration of 0.2 mg/mL in 2.5 mM HEPES, 18.75 mM NaCl at pH 7.4 for CD-spectroscopy measurements at room temperature.   The path length was 0.1 cm, and data were acquired at a scanning speed 100 nm/min with a data pitch of 0.5 nm.

Chemical cross-linking
Purified PknB SA-KD was incubated with different concentrations of glutaraldehyde for 15 min at room temperature. The reaction was stopped by adding 4xSDS-protein buffer and incubating the mixture at 95uC for 5 minutes. Identical procedures were carried out with the 22 kDa adenovirus type 21 fiber knob (Ad21), which forms trimers and served as a positive control, and with the 25kDa chymotrypsinogen A (Sigma-Aldrich), which is monomeric and served as the negative control.

Protein kinase assay
In vitro phosphorylation of 25 mg/mL PknB SA-KD was performed for the indicated time at 37uC with 1 mg/mL myelin basic protein (MBP, Sigma, Deisenhofen, Germany) and 4 mCi c 33 -ATP/ml in 20 ml reaction buffer containing 50 mM HEPES (pH 7.5), 1 mM DTT, 0.01% Brij35, 3 mM MnCl 2 and 3 mM MgCl 2 . In each case, the reaction was stopped by adding 4xSDSprotein buffer. One-dimensional gel electrophoresis was performed and finally, radioactive proteins were visualized by autoradiography using direct-exposure film.

Protein crystallization and structure determination
The final solution used for crystallization contained 8 mg/mL PknB SA-KD , 5 mM MgCl 2 , 4 mM AMP-PNP (tetralithium salt hydrate), 2% (w/v) benzamidine hydrochloride and 1 mM dithiothreitol in 20 mM HEPES, 150 mM NaCl at pH 7.4. Crystals were grown with the sitting drop vapor diffusion method by mixing equal amounts of protein solution and crystallization solution (80 mM 2-(N-morpholino)ethanesulfonic acid pH 6.0, 1.3 M sodium citrate (pH 7.0), 2% (w/v) benzamidine hydrochloride and 60 mM MgCl 2 at 4uC. Crystals appeared after several days and grew to a maximum size of 150 mm diameter. Crystallization trials using mixtures of MnCl 2 and MgCl 2 or MnCl 2 could not improve the crystal quality. They belong to space group C2 and contain six kinase domains in their asymmetric unit. The crystals were mounted on a loop and flash frozen in liquid nitrogen prior to data collection.
X-ray diffraction experiments were performed at the X06SA beam line of the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland. Data extending to 3.0 Å resolution were recorded using the PILATUS detector and processed with XDS [27]. Initial phases were determined with PHASER [28] using the M. tuberculosis PknB structure as search model (Protein data bank (PDB) ID: 1O6Y [23]). The search model was modified by truncating side chains that differed in sequence from the S. aureus protein, and by removing loops and bound ligands. Molecular replacement yielded one solution containing six copies that gave rise to a sensible crystal packing. The initial model was then improved through alternating steps of model building in Coot [29] and refinement in Phenix [30]. The refinement parameters included simulated annealing, Ramachandran refinement, and non-crystallographic symmetry (NCS) restraints. Eight groups per chain were defined in the NCS refinement, excluding the most flexible loops. The fragments of chains A, B and C, which were similar to each other, were defined as NCS-linked groups, and the same was done for chains D, E and F. Electron density for AMP-PNP appeared in all six chains during the course of the refinement, allowing the incorporation of the ligand into the model. The final structure has good quality, with R work and R free values [31] of 21.49 and 24.64%, respectively. Geometric restraints for the AMP-PNP ligand were calculated using the PRODRG server [32]. Data and refinement statistics are given in Table 1. Atomic coordinates und structure factors have been deposited in the PDB (http://www.pdb.org) under the accession code 4EQM.

Activity assay
The functional activity of the kinase domain of S. aureus PknB (PknB SA-KD ) was tested by an in vitro phosphorylation assay. PknB SA-KD is able to phosphorylate other proteins such as myelin basic protein (MBP) in an efficient manner ( Fig. 2A). The target protein MBP was previously used as a surrogate substrate for activity tests of the full-length PknB [12] and of Mycobacterium tuberculosis PknB [33]. Additionally, PknB SA-KD is able to perform autophosphorylation (Fig. 2B).

Overall Structure
PknB SA-KD exhibits the typical kinase fold, with N-and C-lobes creating a central ligand binding region that serves to accommodate the substrate or its analog AMP-PNP (Fig. 3). The N-lobe comprises residues 1 to 90 and contains a six-stranded, antiparallel b-sheet (strands b0-b5) that packs against the aC-helix. The Clobe is composed of six a-helices and a small two-stranded b-sheet (strands b7-b8). Many kinases contain two additional strands, b6 and b9, which form a second b-sheet in the C-lobe. This small sheet is absent in PknB SA-KD due to the conformation of the activation segment (Fig. 4). The two lobes in PknB SA-KD are connected via the linker region (residues 87-92) and via a loop that leads from the C-terminus of the aC-helix to the b4 strand (Fig. 3A, B). Although present in the crystallized PknB SA-KD , residues 284 to 291 at the C-terminus are not well defined by electron density and could not be built. We conclude that the kinase domain of S. aureus PknB includes residues 1-282 ( Fig. 1), in contrast to the computer aided residue assignment for the kinase region in Donat et. al, 2009 [12] (residues 10-268) and the longer PknB SA-KD sequence used for crystallization. As is the case in many kinase structures, high flexibility of the activation loop (residues 160-171) results in this segment not being traceable in the electron density maps. The six PknB SA-KD structures present in the asymmetric unit (molecules A to F, respectively) differ in several surface-exposed loops as a result of non-identical crystal contacts. The molecules can be divided into two homogeneous groups; molecules A, B and C bind benzamidine in a similar location and form similar crystal contacts. However, these features are not conserved in molecules D, E and F. The electron density for molecules A, B and C is more detailed in most regions, allowing unambiguous assignment of most side chains orientations. By contrast, the electron density for chains D, E and F is less well defined, and the density for the ATP-binding site is also somewhat different in these three chains. The b-phosphates of AMP-PNP are arranged in a different orientation in chains D-F compared with chains A-C. The main chain B-factor plot ( Fig. S2) shows overall agreement of the B-factor distribution in all six chains. Residues forming a secondary structure element have significantly lower B-factors compared to residues in flexible loop regions. This flexibility is also reflected in the high overall B-factor. The B-factor differences between the chains A, B and C are small, the same is true for chains D, E and F. It is likely that variation in PknB SA-KD phosphorylation contribute to the observed differences in electron density. Unless specified otherwise, molecule A will be used to discuss the salient features of PknB SA-KD .

AMP-PNP binding
The ATP analog AMP-PNP is bound in the cleft separating the two lobes of the kinase (Fig. 3). The adenine ring projects deep inside the cleft, into a pocket that is largely hydrophobic. Two hydrogen bonds, contributed by the backbone amide of Ile90 and the backbone oxygen of Glu88, anchor the adenine ring (Fig. S3). The ribose and phosphate moieties of AMP-PNP do not make equivalent contacts in the different chains, and these contacts are therefore probably not significant. Furthermore, the c-phosphate group is not visible in the electron density maps, suggesting that the lack of contacts with PknB SA-KD increases its flexibility.

PknB SA-KD is in an inactive conformation
In order to determine whether the structure of PknB SA-KD is in an active or an inactive conformation, we have compared it with the following kinase structures: (i) PknB of M. tuberculosis, which is the first reported structure of a bacterial STK (PDB ID: 1O6Y [23] and 1MRU [24]; (ii) cAMP-dependent protein kinase A (PKA) in an open conformation (PDB ID: 1CTP [34]); (iii) a closed, active PKA structure with bound ATP (PDB ID: 1ATP [35]); and (iv) human c-Src, a tyrosine kinase in the autoinhibited conformation (PDB ID: 2SRC [36]). A structure-based sequence alignment was generated for all five kinases using ClustalW [37], Espript [38], Strap [39] (Fig. 5), and superpositions were performed by aligning residues 100-250 of the C-lobe of PknB SA-KD with the other structures. Although there is no crystal structure available for it, we also included the sequence of the B. subtilis protein kinase C (PrkC) in our analysis because of its especially high homology to the S. aureus PknB sequence and the fact that both kinases have three extracellular PASTA domains.
Our analysis highlights several highly conserved residues that are important for kinase function: the glycine-rich loop, the DFGmotif, the position of the aC-helix, and the catalytic and regulatory spines (C-spine and R-spine, respectively). Representative F obs -F calc omit electron density maps of the ATP binding regions in chains A and D are shown in Figs. S3A and S3B, respectively. F obs -F calc omit electron density maps of the DFGmotif and the inhibition helix of chain A, the AMP-PNP and its surroundings, and the aF-helix of chains A and D are shown in Fig. S4A-D. The glycine-rich loop positions the c-phosphate of ATP in an orientation that facilitates phosphoryl transfer during catalysis [40]. The DFG-motif is near the activation segment (Fig. 3). The C-and R-spines are sets of non-contiguous hydrophobic residues that line the interior of a kinase and stabilize its active, closed conformation (Fig. 6A, B) [40,41,42,43].
a. Stabilization of the phosphates. Connecting strands b1 and b2 in the N-lobe, the glycine-rich loop has a consensus sequence of GX 1 GX 2 wGX 3 V, with w denoting a hydrophobic residue (usually tyrosine or phenylalanine) and X indicating any amino acid [23,44,45]. Since it serves to stabilize the phosphates of ATP, this loop is also sometimes referred to as phosphate binding loop or P-loop [45]. In the active state of kinases, the glycine-rich loop faces the terminal phosphates of ATP, positioning the cphosphate for the transfer reaction [46,47,48]. In contrast, the glycine-rich loop faces away from ATP in open and inactive kinases such as PKA (in PDB ID: 1CTP [34]) (Figs. S3, S5). While the glycine-rich loop is strictly conserved in PKA, c-Src, and many other kinases, its consensus sequence is altered to GX 1 GGMX 2 X 3 V in PknB SA-KD and the two other bacterial kinases included in our comparison. Moreover, the X 1 residue is also a glycine in both B. subtilis PrkC and PknB SA-KD . The increased number of directly linked glycines would likely render the loop more flexible in both proteins. The comparison of the structures shows that the glycine-rich loop in PknB SA-KD is in a conformation that differs from those observed in both active and inactive kinases (see above) (Figs. S3, S5). Its ''tip'' (residues 18 to 20) is twisted upwards, facing away from the phosphates. However, its ''base'' (residues 22-25) is still in a position similar to those observed in active kinases.
In active kinases, the phosphates of ATP are stabilized in the cleft between the two lobes. The motif involved in binding phosphates on the C-lobe side is the DFG-motif. Protein kinases stabilize the phosphates of bound ATP with magnesium ions, which in turn are ligated to an aspartic acid in the DFG-motif (residues 151-153 in PknB SA-KD ). In active kinases, the DFG-motif also features an internal hydrogen bond between the aspartic acid and the glycine, and a second hydrogen bond between the glycine and the amide nitrogen of residue DFG+2 (Ala155 in PknB SA-KD ) [41]. The DFG-motif of PknB SA-KD lacks internal hydrogen bonds, and no magnesium ion is visible in the vicinity of Asp151. It is clearly not in an active conformation and does not stabilize the phosphate groups of AMP-PNP.
A salt bridge is located close to the DFG-motif in active kinases. To enable an active kinase conformation, the aC-helix must be oriented such that a salt bridge between the strictly conserved residues Glu58 and Lys39 can be formed. Lys39 lies in the b3strand of the N-lobe and helps to stabilize the aand b-phosphates of ATP [46,47,48]. Structural comparison with other active kinases shows that the aC-helix of PknB SA-KD is not in a closed conformation ( Fig. 4 and Fig. S4B). The helix is rotated away from the active site, and Glu58 does not form a salt bridge with Lys39.
b. The C-and R-spines. These spines stabilize the active, closed conformation of a kinase and contain residues from both lobes of the kinase. The C-spine attaches the active site to the aF-helix, connecting the two lobes via the adenine ring system. It lines the rear of the adenosine-binding pocket and stabilizes the closed, active conformation of protein kinases. In PknB SA-KD , the C-spine is formed by residues Val24, Ala37 in the N-lobe and residues Ile139, Leu140, Ile141, Leu95, Val198 and Met202 in the C-lobe (Fig. 6A, C). The C-spine of PknB SA-KD and the bound adenine ring superimpose well with those of the closed, active PKA structure (Fig. 6C).
The R-spine also serves to stabilize the active conformation of protein kinases [40,41,42,43]. In PknB SA-KD , the putative residues for the spine are Met73, Ser62, His131 and Phe152. The latter residue is part of the DFG-motif. Asp191 is stabilizing the backbone of His131, thereby anchoring the spine to the aF-helix. Ser62 lies in the aC-helix and is located four residues C-terminal to the highly conserved Glu58. Since the DFG-motif and the aChelix are not in an active conformation, the R-spine cannot be fully formed in PknB SA-KD (Fig. 6B, D).
c. Inhibition helix. In order to assume an active state, the aC-helix of PknB SA-KD would have to change its position (Fig. 4). This is however not possible in our structure because the space into which the aC-helix would have to rotate is already occupied by the activation segment located directly after the DFG-motif (Fig. 4). In active kinases the activation loop makes close contacts to the C-lobe [40] and forms the b-sheet between strands b6 and b9 (Fig. 4B). The activation segment of PknB SA-KD interacts with the N-lobe and the aC-helix. The activation segment forms a short helix directly after the DFG-motif and blocks the area for the aChelix to assume an active conformation. The putative residues of b9 are part of the inhibition helix and far away from b6, so that the b-sheet cannot be formed. The absence of the b6/b9 sheet is also a marker for an inactive conformation of PknB SA-KD (Fig. 4B). The PknB SA-KD aC-helix is stabilized by a hydrophobic interface similar to the one found in the structures of CDK2 and c-Src in their autoinhibited conformations [36,49,50]. The interface is formed by several residues in the aC-helix, strands b3 and b4, and the activation segment.

Analysis of surface conservation
In order to identify conserved features and compare them with homologous proteins, the PknB SA-KD sequence was aligned with   (Fig. S6). Blue indicates highly conserved residues (100-91% conservation in dark blue, 87-70% conservation in blue, and 66-54% light blue). Residues that are highly conserved in most kinases but are different in PknB SA-KD are colored in orange (83-70% conservation in orange, 66-54% conservation in light orange). In the right panel, the benzamidine bound to PknB SA-KD in three of the six chains of the asymmetric unit is shown as a stick model. The benzamidine is not visible in the other two panels. doi:10.1371/journal.pone.0039136.g007 representatives of several bacterial STKs (Fig. S6). For our analysis, we selected the bacterial STKs recently analyzed by Pereira et al. [10]. All analyzed kinases exhibit strong conservation in the prototypical regions required for catalytic function, such as the DFG-, HRD-and SPE-motifs and the glycine-rich loop. In order to depict the location and distribution of conserved residues, we mapped them onto the surface of PknB SA-KD (Fig. 7). As expected, residues important for the catalytic function of the kinase are highly conserved (highlighted blue in Fig. 7). These residues cluster in the ATP-binding site, the glycine-rich loop, and the DFG-, HRD-and SPE-motifs. However, a small number of residues are highly conserved in most other kinases but differ from the consensus sequence in PknB (highlighted orange in Fig. 7). The remaining surface of PknB SA-KD is remarkably devoid of conserved residues.

Discussion
We report here the crystal structure of the catalytically competent kinase region of S. aureus PknB. In its physiological context, the kinase is attached to the bacterial cell wall via a membrane anchor, and it phosphorylates substrates in response to stimuli that engage the extracellular PASTA domains. Activity assays demonstrate that purified PknB SA-KD is able to phosphorylate substrates efficiently. However, an analysis of structural parameters that define the active states of protein kinases clearly demonstrates that PknB SA-KD has crystallized in an inactive conformation. Although it does bind the ATP analog AMP-PNP, this substrate is not bound in a conformation that would enable catalysis. The AMP-PNP triphosphate moiety is not contacted by either a magnesium ion or residues from the glycine-rich loop or the DFG-motif. As PknB SA-KD is catalytically active in solution, it is conceivable that it exists in different conformations, corresponding to active and inactive states, in solution, perhaps owing to different states of phosphorylation. Hence, crystallization likely selected the inactive state.

Dimer formation
An attractive scenario for PknB activation could be based on dimer formation, and dimerization has in fact been implicated in the regulation of the activity of the M. tuberculosis PknB kinase domain. In that case, the kinase domain forms dimers that are stabilized by a salt bridge between Arg9 in one monomer and Asp75 of another monomer [22,51]. Although PknB SA-KD consistently eluted as a monomer in gel filtration experiments in solution (Fig. S1A), it is conceivable that low-affinity dimerization of PknB SA-KD could occur at higher concentration. This hypothesis was evaluated by chemical crosslinking experiments using glutaraldehyde (Fig. S7). No crosslinked dimer of PknB SA-KD was obtained under any of the tested conditions, while a control protein known to form trimers could be successfully cross-linked under identical conditions. Additionally, inspection of the crystal packing can sometimes provide clues about the possible existence of oligomers. We find that three of the six PknB SA-KD molecules present in the crystals form nearly identical dimers with their symmetry mates (A-C9, B-B9, C-A9), and the arrangement of these putative dimers resembles the M. tuberculosis PknB dimer (Fig. 8). It is therefore conceivable that activation of S. aureus PknB also involves dimerization. We note, however, that the putative PknB SA-KD dimer interface contains benzamidine, a compound that was present in the crystallization solution and that was required to obtain good-quality crystals. The observed dimer may therefore be a crystallization artifact. Moreover, the three remaining molecules in the crystals (chains D, E, F) do not form similar dimers.
It is of course possible that dimerization of the kinase region is linked to its phosphorylation status, and that the copies of PknB SA-KD present in the crystals have different states of phosphorylation, thus impacting their dimerization properties.
Another, perhaps more likely, scenario could link dimerization to the extracellular PASTA domains, which could alter their association state in response to signal binding [11,17].

Inhibition helix
Although it represents an inactive conformation of the kinase, the PknB SA-KD crystal structure nevertheless provides insights into a pathway of activation. In order to activate PknB SA-KD, the aChelix would need to significantly change its location by rotating into the binding site. Such a rotation, however, is not possible in our crystal structure because the space that would accommodate the rotated aC-helix is already occupied by the activation segment, which forms a short helix directly following the DFG-motif ( Fig. 4 and Fig. S4A). The activation loop of active kinases makes close contacts with the C-lobe, whereas in PknB SA-KD it interacts with the N-lobe and the aC-helix. The activation segment would have to be displaced to allow for the formation of an active PknB conformation. This could be triggered by autophosphorylation in the activation segment. Thr164 could play a key role in this conformational change. Structural alignments with inactive c-Src kinase reveal a similar orientation of the activation segment, with an autoinhibition helix blocking the inward rotation of the aC-helix [36,52] (Fig. 4). In this case, c-Src can be activated through the phosphorylation of residue Tyr416, which is the residue equivalent to Thr164 in PknB.
A dissolving of the inactive state of the activation segment could conceivably trigger a downward movement of the glycine-rich loop to stabilize the band c-phosphates of the ATP. Due to weak electron density as a result of flexibility, it is not possible to model the activation segment. However, difference electron density in this region suggests that parts of the activation loop lie next to the glycine-rich loop, preventing interaction of the glycine-rich loop with phosphates.
In conclusion, our structural study provides improved understanding of the function of eukaryotic like serine/threonine kinases in bacteria at the molecular level. Future work will aim at the identification of substrates of PknB, the molecular mechanisms of substrate selection and the role of autophosphorylation for the activity of the kinase. In particular, the role of ligand-dependent dimerization of extracellular PASTA domains for activation of PknB remains to be clarified. Moreover, recent work provided evidence that PknB is embedded in the tight regulatory network controlling virulence of S. aureus. The structural information presented in this study may serve as a basis for further investigations of the molecular mechanisms determining pathogenesis of the major human pathogen S. aureus.  Figure S4 Stereo views of representative omit maps of PknB SA-KD . All panels show F obs -F calc omit electron density maps contoured at 1.0 s. In panel A the omit map for the DFG-motif and the inhibition helix is shown. Panel B shows the phosphate binding region. Panels C and D show the omit map for the aFhelix for chain A in panel C and chain D in panel D. Maps were drawn with radii of 3Å (panels A, C and D) and 8Å (panel B) around the depicted coordinates. The larger radius for panel B was chosen to show that no extra density that would account for a magnesium ion exists in the vicinity of the AMP-PNP ligand. (TIF) Figure S5 Orientation of the glycine-rich loop. All kinases were aligned with the C-lobe of PknB SA-KD (residues 100-250). The PknB SA-KD structure is drawn in orange. The closed PKA structure is shown in green (PDB ID: 1ATP [35]), and the open PKA structure is shown in red (PDB ID: 1CTP [34]). The kinase domain of c-Src (PDB ID: 2SRC [36]) is shown in purple. (TIF) Figure S6 Alignment of selected kinases and analysis of conservation. The kinases were selected according to [10]. Five kinases were omitted due to lack of DFG-, SPE-, HRD-motif or the N-lobe. The color code is identical to that used in Fig 7. Blue indicates highly conserved residues (100-91% conservation in dark blue 88-74% in blue and 69-54% light blue). Residues highly conserved but different in PknB SA-KD are colored in orange (88-71% in orange, 69-54% in light orange). The selected kinases are