Stk1 is required for BlaR1-mediated broad-spectrum β-lactam resistance in epidemic-causing strains of Staphylococcus aureus

TL;DR

A new role for Stk1 in helping MRSA resist antibiotics is discovered, offering a potential new target for treating antibiotic-resistant infections.

Contribution

The novel role of Stk1 in BlaR1-mediated antibiotic resistance in MRSA is identified.

Findings

Phosphorylation of BlaR1 by Stk1 is essential for drug sensing and resistance in MRSA.

Targeting Stk1 could enhance the effectiveness of β-lactam antibiotics against MRSA.

A structural basis for Stk1 inhibition is presented for future drug development.

Abstract

Sensory induction of mecA expression plays a pivotal role in mediating broad-spectrum β-lactam resistance (BBR) of MRSA. In contemporary MRSA isolates, sensory induction of BBR originates at the membrane-localized BlaR1, which, upon detection of β-lactam drugs, triggers a signal transduction cascade that promotes mecA induction. We hereby showed that phosphorylation of BlaR1, mediated through the serine-threonine kinase, Stk1, stabilizes its membrane spanning state and localization, allowing for proper drug sensing and subsequent signal transduction events to occur, culminating in mecA-mediated BBR. Our results demonstrated that targeting Stk1 could potentiate synthetic lethality to β-lactams in the majority of naturally isolated strains of MRSA. We also presented the structural and kinetic basis for a Stk1-inhibitor complex that could enable rational design of Stk1 directed anti-MRSA therapeutics in the future. Our results reveal a unique and hitherto unknown role of the STK signaling pathway in bacterial protein stabilization in the cytosolic membrane.

Figures

b](#F1)). More importantly, >90% of ~60,000 entries for such MRSA entries that had a truncated *mecA* cluster possessed the BlaZ regulatory system, highlighting the fact that *mecA* expression in these strains is likely under the control of the BlaZ signaling system. Interestingly, a much lower percentage of entries (~70% as opposed to 90%) of ~19,000 entries for MRSA strains possessing an intact *mecA* cluster had the BlaZ regulatory system. The reason for this decreased association of the BlaZ regulatory pathway among strains that had an intact *mecA* counterpart is perhaps because the prese

c](#F2)). However, in presence of β-lactams, the Δ*stk1* [*stk1* (Wt)] strain showed increased survival as seen by enhanced growth and CFU formation compared to Δ*stk1* [E] and Δ*stk1* [*stk1* (K39G)]. These results indicated that the synthetic lethality of Δ*stk1* was indeed due to the loss of Stk1 function and that Stk1 function was necessary for the production of β-lactam resistance. To associate the Stk1 dose with the magnitude of β-lactam resistance, we performed Δ*stk1* complementation using a xylose-inducible vector, p*TX15*. Growth curves for the resultant complemented strains were sim

i](#F3)). We then treated Wt with nafcillin combined with a range of GWX concentrations (0–0.625 μg/mL) and found the strain to gain increasing susceptibility in a GWX dose-dependent manner. When co-treated with nafcillin and 0.625 μg/mL GWX, we detected a significant growth attenuation in Wt that was comparable to that displayed by nafcillin-treated Δ*stk1*, indicating that GWX rendered Wt susceptible to nafcillin. ([Fig. 3i](#F3)). Δ*stk1* had similar growth patterns when treated with only nafcillin or with a combination of nafcillin and GWX, suggesting that the phenotype displayed above was

d](#F4), Fig. S3d), as well as when exposed to nafcillin ([Fig. 4e](#F4), Fig. S3d) by immunoblotting. On the other hand, Wt complemented with *blaI* [Wt] showed efficient BlaI degradation as expected. Wt [*blaI* (N101A and F102A)] did not display growth alterations when untreated ([Fig. 4f](#F4)); however, it was significantly susceptible to nafcillin compared to Wt [*blaI* (Wt)] in growth assays, as Wt [*blaI* (Wt)] successfully underwent BlaI proteolysis ([Fig. 4e](#F4)) and thereby survived in the presence of nafcillin. We also complemented Δ*stk1* with the BlaI variants for growth assays,

e](#F5)). These findings suggest that Stk1 primarily modulates BlaR1 at the post-transcriptional level during mid-exponential growth. In the late-exponential phase, the observed increase in *blaR1* transcripts across strains likely reflects a leaky activation of the BlaR1-BlaI autoregulatory pathway^[38](#R38)^. In Δ*stk1*, impaired membrane localization of BlaR1 during mid-exponential growth may delay initiation of this autoregulatory loop, resulting in dampened *blaR1* expression later in growth. We confirmed this notion by assessing transcriptional and protein expression of BlaI. Similar to

a](#F6)). Increased survival of *C. elegans* infected with Δ*stk1* with nafcillin, or Wt with nafcillin and GWX, indicated that targeting Stk1 could be beneficial to treat *S. aureus* infections, particularly when used in combination with an established β-lactam drugs. In addition to these, our results revealed the previously unknown role of Stp1, the cognate phosphatase of the eSTK signaling pathway. Stp1 oppositely regulates BlaR1 phosphorylation and membrane localization and consequently mediates a contrasting effect to that of StK1 in mediating broad-spectrum β-lactam resistance.

Equations

Full text

Introduction

Staphylococcus aureus is a frequent colonizer of the human population and an opportunistic pathogen^1^. S. aureus is also infamous for its ability to resist treatment by important antibiotics such as β-lactams, vancomycin, and daptomycin^2^.

Eukaryotic-like (Hanks-type) serine-threonine kinase (STK) phospho-regulation pathways are universally present in bacteria and mediates important cellular processes^3^. In S. aureus and other Firmicutes, the STK kinase, Stk1 (also known as PknB or PrkC), belongs to the Penicillin-binding protein And Serine/Threonine kinase-Associated (PASTA) kinase class. Stk1 functions in concert with Stp1, its cognate phosphatase, to mediate phosphorylation and dephosphorylation of serine, threonine, or tyrosine residues of target proteins. STK signaling in S. aureus regulates numerous cellular pathways, including cell wall biosynthesis^4,5^, bacterial quiescence^6^, virulence^7^, and β-lactam resistance^8^.

β-lactam antibiotics are known for their excellent efficacy and safety, making them the most commonly prescribed class of drugs to treat infections^9^. In S. aureus, resistance to β-lactams can be detected in the form of narrow- or broad-spectrum resistance^2^. Resistance to early-generation β-lactams such as penicillin, driven by the β-lactamase PC1 and encoded by plasmid-associated blaZ^10,11^, is termed narrow-spectrum resistance. Resistance to next-generation β-lactams, such as methicillin and its newer derivatives, including advanced cephalosporins, is categorized as broad-spectrum resistance or methicillin resistance. Such resistance is usually displayed by methicillin-resistant S. aureus (MRSA) isolates and is facilitated by a low-affinity penicillin-binding protein, PBP2a, encoded by mecA (or mecC) embedded within a chromosomal cassette (SCCmec)^12^. Similar but separate inducible pathways are known to control both resistance mechanisms^13^. When expression of blaZ or mecA is unwarranted, the repressors, BlaI or MecI, respectively, suppress their transcription. BlaR1 or MecR1 are membrane-embedded sensor-inducer proteins that detect the presence of β-lactam drugs in the bacterial surroundings, which culminates in a signal transduction process that promotes degradation of BlaI or MecI to de-repress blaZ or mecA expression, respectively. The above pathways allow for inducible expression of blaZ or mecA, ensuring timely expression of the β-lactam resistance mediators when needed.

Previous studies have further indicated that the BlaI and MecI repressors, in addition to regulating their corresponding target genes (i.e., blaZ and mecA), can also cross-regulate mecA and blaZ expression owing to their high sequence homology and similar DNA recognition sequence^14,15^. This ability of cross-regulation between the two systems is exemplified in strains containing SCCmec clusters that are deficient of functionally active mecR1-mecI regulatory genes, one example being S. aureus strains that contain the type IV SCCmec cluster^16^. A prominent community- and healthcare-associated epidemic causing MRSA strain, USA300, contains the type IV SCCmec cluster that is dependent on the bla system for mecA regulation^17^ (Fig. 1a). Notably, the majority of the identified SCCmec cluster types (9 of the total identified 15 types, to date) have functionally inactive mecR1-mecI regulatory genes^18–20^. Thus, in the vast majority of MRSA strains, the plasmid encoded BlaR1-BlaI regulatory system could confer both narrow- and broad-spectrum β-lactam resistance, making it the sole regulatory pathway associated with both modes of β-lactam resistance.

While much of the mechanistic understanding of the BlaR1-BlaI signal transduction pathway remains to be elucidated, previous studies have indicated that BlaR1 is phosphorylated by Stk1, and inhibition of Stk1 leads to decreased BlaZ activity as well as narrow-spectrum β-lactam resistance^21^. The mechanism by which inhibition of Stk1 leads to decreased BlaZ activity is however unknown. Additionally, the role of Stp1, the cognate phosphatase of the STK pathway in controlling β-lactam resistance, remains undetermined.

In this study, we reveal that Stk1 functionality is essential for the successful induction of blaZ and mecA via the BlaR1-BlaI signal transduction pathway, by facilitating stabilization of the membrane inserted species of the polytopic BlaR1. Inhibition of Stk1 functionality, either by deletion of the stk1 gene or in the presence of its inhibitor GW779439X (GWX), resulted in decreased levels of membrane-localized BlaR1, subsequently leading to attenuated degradation of BlaI and compromised induction of mecA expression. Decreased induction of mecA in turn led to decreased protein levels of PBP2a and prompted synthetic lethality in bacteria to broad-spectrum β-lactam drugs. Furthermore, our results suggested that stabilization of membrane inserted BlaR1 is not only coupled with its Stk1-mediated phosphorylation status but occurs in a β-lactam-independent manner. Our results highlight the hitherto unknown role of a posttranscriptional modification of BlaR1 in facilitating proper localization into the bacterial membrane. Through performing high-throughput bioinformatics analysis and providing the structural basis of Stk1-GWX inhibition, we uncover the mechanistic basis of Stk1’s involvement in broad-spectrum β-lactam resistance. Additionally, through performing C. elegans model of infection we present Stk1 as a potential target of therapeutic development against MRSA.

Materials and Methods

Bacterial strains and plasmids

S. aureus strains were cultured in tryptic soy broth (TSB) (BD Biosciences) at 37°C with shaking at 180 rpm. E. coli cultures were grown in Luria Bertani (LB) broth at 37°C with shaking at 180 rpm unless mentioned otherwise. S. aureus carrying the constitutively expressing (pTXΔ) or a xylose-inducible (pTX15) vector were cultured in media supplemented with 12.5 μg/mL tetracycline^8,22^. Strains with the pTX15 vector were grown in TSB without dextrose. All the strains and plasmids used in this study are listed in Tables S1 and S2, respectively.

Mutant construction

Isogenic S. aureus Δstk1 and Δstp1 mutants were created using the pJB38 vector as previously described^8,22^. Briefly, a splice-overlap PCR product with at least 1 kb up- and downstream of the target gene was cloned into pJB38. The construct was transformed into the E. coli strain DH5α, electroporated into S. aureus RN4220, and then finally phage transduced into recipient S. aureus strains. Allelic replacement was performed as previously described^23^. Primers used in this study are listed in Table S3.

Growth curve assays

Growth assays were performed using the Bioscreen C Pro automated growth curve machine (Growth Curves USA), starting from an initial OD_600_ of 0.1. When needed, antibiotics nafcillin or oxacillin (Sigma), or the Stk1 inhibitor GW779439X (GWX) (MedchemExpress) were added at the start of the experiments at the required concentrations. CFU measurements were performed by sampling the Bioscreen plate at 12 h, serial diluting the cultures with phosphate-buffered saline (PBS), plating on TSA only or TSA plates containing tetracycline, and incubating overnight at 37°C. Each experiment was performed twice with 3 technical replicates to ensure reproducibility. Statistical analysis was performed using an unpaired Student’s t-test with GraphPad Prism.

Transmission Electron Microscopic analysis

Electron microscopy was carried out at the core imaging facility at the University of Maryland, Baltimore. S. aureus strains were cultured in 50 mL of TSB in flasks from a starting OD_600_ of 0.1 under the conditions specified above. After 2 h of growth, either 0 or 8 μg/mL of nafcillin was added to the flasks, which were grown for an additional 4 h. 1 mL of bacterial culture was harvested, pelleted, washed with PBS, and fixed with 2.5% glutaraldehyde in 0.1 M PIPES buffer pH 7.4. The cells were washed again, pelleted, enrobed in 2.5% low-melting-point agarose, and post-fixed with 1% osmium tetroxide, 0.25% potassium ferrocyanide in 0.1 M PIPES buffer pH 7.4 for 1 h at room temperature. Agarose blocks containing cells were washed and en bloc stained with 1% uranyl acetate in water for 60 min, then washed and dehydrated using increasing concentrations of acetone, including 30%, 50%, 70%, 90%, and 100% for 10 min at each step. Specimens were infiltrated in increasing concentrations of Araldite-Embed 812 resin (EMS, PA) following the manufacturer’s instructions and embedded in pure resin at 60°C for 24 to 48 h. Ultrathin sections of ~70 nm thickness were cut on a Leica UC6 ultramicrotome (Leica Microsystems, Inc., Bannockburn, IL), collected onto copper grids, and examined in a Tecnai T12 transmission electron microscope (Thermo Scientific) operated at 80 kV. Digital images were acquired by using an AMT bottom-mount CCD camera and AMT600 software.

Custom antibody creation

Custom polyclonal rabbit antibodies were generated by Thermo Fisher using their standard 70-day rabbit immunization protocol. Briefly, New Zealand White rabbits were immunized with synthetically generated amino acid peptides (BlaR1 Cytosolic Loop: KNEFKTYAESIMDSVLKTP; BlaI: ELNNKEIEELRDILNDISKK; Stk1: REKEETLKRFEREVHNS; Stp1: RLLVKEGTIEDHGDQLMQL; SrtA: DDYNEKTGVWEKRKIFVATEVK) over the course of 70 days. Serum was collected on days 35 and 58 and analyzed via ELISA test to verify antibody titers. A final terminal bleed was performed on day 70, and the collected serum was used for experiments.

Immunoblotting

S. aureus strains were cultured in 50 mL of TSB in flasks from a starting OD_600_ of 0.1 under conditions specified above. Following appropriate treatment with antibiotics and/or the GWX inhibitor, cells were harvested, washed with PBS, then resuspended in Immunoblotting Lysis Buffer (50 mM Tris pH 7.5, 150 mM NaCl, 2mM EDTA, 1 mM Sodium Pyrophosphate, 1X cOmplete Mini^™^ protease inhibitor cocktail (Roche)). Cells were lysed mechanically using a FastPrep-24 Classic bead-beating machine (MP Biomedicals) with 4 cycles of speed 6.5 m/s for 45 s to obtain whole cell lysates. Membrane fractions from cell lysates were isolated via ultracentrifugation using a Sorvall WX Ultra 80 centrifuge (Thermo Fisher Scientific) at 66,000 Xg for 1 h at 4°C. The supernatant (cytosol fraction) was carefully removed and saved while the pellet was resuspended with Immunoblotting Lysis Buffer. Protein quantification was carried out using a Pierce BCA protein assay kit (Thermo Fisher), following which 20 μg of protein samples were separated on SDS-PAGE gel using the appropriate polyacrylamide gel concentration (BlaI: 15%; PBP2a, BlaR1, Stk1, Stp1: 10%). SDS-PAGE gels were transferred onto a low-fluorescence polyvinylidene difluoride (PVDF) membrane (Sigma Millipore) using a semi-dry transfer buffer (48 mM Tris base, 39 mM glycine, 20% methanol) and a semi-dry transfer unit (Bio-Rad) set at 20 V, 350 mA for 90 min. The blots were first blocked using Intercept (TBS) Blocking Solution (Li-Cor) for 1 h at room temperature. Primary antibody incubation was carried out overnight at 4°C by diluting the antibody in Intercept (TBS) containing 0.2% Tween-20 at appropriate concentrations (Rabbit anti-BlaI: 1:500; Rabbit anti-BlaR1: 1:500; Rabbit anti-PBP2a [Product #130–10178, RayBiotech]: 1:500; Rabbit anti-Stk1: 1:1000; Rabbit anti-Stp1: 1:2000; Rabbit anti-SrtA: 1:1000; Mouse anti-Spa [Product#P2921-.2ML, Sigma]: 1:1000; Mouse anti-P-Tyrosine [Product #96215, Cell Signaling]: 1:1000; Mouse anti-His-Tag [Product # MA1–21315, Invitrogen]: 1:1000). For duplex staining, the blot was washed with TBST and probed for an additional 2 h at 4°C with a second primary antibody. Next, after washing the blots with TBST, blots were incubated for at least 1 h at room temperature with a secondary antibody at appropriate concentrations (anti-Rabbit 700nm: 1:15,000 or anti-Mouse 800nm: 1:10,000) in Intercept (TBS) containing 0.2% Tween-20 and 0.01% SDS. Loading controls were included by probing for Sortase A by cutting the blot in half and incubating the section with anti-SrtA primary antibody, and/or including a Coomassie-stained gel for each immunoblot. For Coomassie staining, gels were incubated in Coomassie stain (0.25% Coomassie Brilliant Blue, 40% methanol, 10% acetic acid in water) for 2 h and de-stained using De-stain solution (50% methanol, 10% acetic acid in water). Imaging was performed using the Azure 600 imaging machine (Azure Biosystems). Blots showing quantitative differences were performed thrice, and densitometry analysis was performed using ImageJ. Statistical analysis was carried out using an unpaired Student’s t-test on GraphPad Prism. Blots showing qualitative differences were performed twice.

Densitometry analysis was performed using ImageJ. Briefly, the Western blot image was imported into ImageJ and each lane was highlighted using the rectangle tool. A pixel darkness histogram was generated from the aforementioned rectangles. The histograms were further gated on the bottom and sides of each band peak. The bottom gate follows the slope of the background signal of the PVDF membrane. The side gates are vertical gates at the point of the band peak where the tangent to the curve is at a 45-degree angle (to allow for standardization of the area under the peak when bands are close together). The area under the peak was calculated after gating. Statistical analysis was carried out using an unpaired Student’s t-test on GraphPad Prism.

Bacterial total RNA isolation and qRT-PCR analysis

S. aureus strains were cultured in 50 mL of TSB in flasks from a starting OD_600_ of 0.1 under the conditions specified above. Following appropriate treatment with antibiotics and/or the GWX inhibitor, approximately 5 × 10^9^ bacterial cells were harvested and washed with RNase-free water. Cells were resuspended in RLT Buffer from the RNeasy Mini Kit (Qiagen) and supplemented with 0.1% 2-mercaptoethanol. Cells were mechanically lysed using the FastPrep-24 Classic bead beating machine (MP Biomedicals) with 2 cycles of speed 6.0 m/s for 45 s. After lysis, total RNA was isolated using the RNeasy Mini Kit (Qiagen), and cDNA was synthesized as described previously^24^. Quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) was performed in a 7500 Real-Time PCR System (Applied Biosystems) using SYBR Green PCR Master Mix (Thermo Fisher) with gyrB as a housekeeping control. Primers used in qRT-PCR are listed in Table S4. Each experiment was performed twice to ensure reproducibility. Data analysis was performed using Microsoft Excel and statistical analysis was performed using an unpaired Student’s t-test on GraphPad Prism.

Bioinformatics analysis

To analyze the distribution of key antibiotic resistance genes in S. aureus strains, all publicly available S. aureus genome assemblies deposited into the National Center for Biotechnology Information (NCBI) database^25^ were downloaded and combined to build a custom BLAST nucleotide database using the makeblastdb application from the BLAST+ suite (version 2.15.0+)^26^. The full-length nucleotide sequences of the target genes of interest (stp1: Genbank: CP000730.1, locus tag: USA300HOU_1156; stk1: Genbank: CP000730.1, locus tag: USA300HOU_1157; mecA: Genbank: CP000730.1, locus tag: USA300HOU_0031; mecI: Genbank: BA000018.3, locus tag: SA_RS00345; blaZ: Genbank: CP000730.1, locus tag: USA300HOU_pUSA300HOUMR0011; and blaI: Genbank: CP000730.1, locus tag: USA300HOU_pUSA300HOUMR0013) were queried against the custom S. aureus BLAST database using blastn^26^ with the -max_target_seqs option. In the case of a genome accession having multiple hits per query, only the single top-scoring hit was chosen, prioritizing alignments with the greatest percent identity and alignment length. Hits were filtered using the following criteria to minimize false positives: (i) an alignment length fraction ≥ 0.95 (i.e., ≥ 95% of the gene length aligned) and (ii) percent nucleotide identity ≥ 95%. The alignment length fraction was calculated as the aligned region divided by the full length of the query gene. A total of 127894 S. aureus genome accessions were used in this analysis.

The protein multiple sequence alignment of the kinase domain of S. aureus strain Mu50 Stk1 (Genbank: WP_000579563.1; identical to Stk1 from USA300) against the kinase domains of various Hanks-type kinases was performed using Clustal Omega. The aligned sequences include B. subtilis strain 168 (PrkC, Genbank: NP_389459.1), M. tuberculosis strain H37Rv (PknB, Genbank: NP_214528.1), H. sapiens (ABL1, Uniprot: P00519), and E. faecalis (IreK, GenBank: WP_002387383).

Protein expression and purification

The kinase domain (residues 1–291) of stk1 (Stk1 KD) from S. aureus strain Mu50 (ATCC 700699) was cloned into the expression vector pET28b (+) using primers listed in Table S3 and transformed into E. coli BL21 (DE3) as described previously^27^. The vector includes an N-terminal 10X His-tag followed by a thrombin cleavage site. Overnight cultures of BL21 (DE3) were sub-cultured at a starting dilution of 1:100 and grown in Terrific Broth (TB) media supplemented with 50 μg/mL kanamycin under conditions mentioned above until they reached an OD_600_ of 0.6. 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added to the media to induce protein expression and the culture was incubated for 16 h at 23°C and 180 rpm. Cells were then harvested and pelleted at 6,200 × g for 15 min, following which the pellet was resuspended in 4 mL/gm in Lysis Buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 20 mM imidazole, 0.1X cOmplete Mini Protease Inhibitor Cocktail (Roche)). Cell lysis was performed by doing two passes through a CF1 Cell Disruptor (Constant Systems Ltd.) at a pressure of 30 kPSI. The insoluble fraction was pelleted via centrifugation at 40,000 g for 45 min at 4°C. The supernatant was filtered through a 45 μm filter and loaded onto a gravity Ni-NTA column (Thermo Scientific HisPur Ni-NTA resin) pre-equilibrated with Buffer A (20 mM HEPES pH 7.5, 150 mM NaCl, 20 mM imidazole). The resin was washed with 10X column volume (CV) of Buffer A, followed by 10X CV of Buffer B (20 mM HEPES pH 7.5, 150 mM NaCl, 50 mM imidazole). The protein was then eluted with 10X CV of Elution Buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 300 mM imidazole). Eluted fractions were pooled and dialyzed into Buffer C (20 mM HEPES pH 7.5, 150 mM NaCl) using a 3.5 kDa MWCO membrane (Repligen). The dialyzed proteins were incubated with a 1:1000 molar ratio of bovine α-thrombin (HTI) overnight at 4°C. Subsequently, the 10X His-tag cleaved proteins were passed over the Ni-NTA resin once again to remove the His10-tag and any uncleaved protein. The remaining protein was injected into a Superdex 200 10/300 GL size exclusion column (GE Life Sciences) equilibrated with Buffer C. Fractions containing the thrombin-cleaved Stk1 KD were concentrated using a 10 kDa MWCO Amicon Ultra centrifugal filter (Millipore Sigma).

Crystallization and X-ray crystallographic analysis of STK1 with GWX inhibitor

A solution of 9 mg/mL of purified Stk1 KD, 5 mM MgCl_2_, 2% (w/v) benzamidine hydrochloride, 1 mM dithiothreitol, 20 mM HEPES pH 7.5, 150 mM NaCl, and 2 mM of Stk1 inhibitor GWX was drop-wise added onto 24-well Cryschem plates (Hampton Research) at 22°C using the sitting drop vapor diffusion method. The drop that yielded the best diffracting crystals contained 1 μL of protein solution mixed with 1 μL of reservoir solution (80 mM 2-(N-morpholino)-ethanesulfonic acid pH 6.0, 1.2 M sodium citrate pH 7.0, 2% (w/v) benzamidine hydrochloride, and 30 mM MgCl_2_), optimized relative to an earlier report^4^. Crystals were grown at 4°C and began to appear after 1–2 days, achieving a maximal size of up to 400 μm in approximately one week. To improve the size and quality of the crystals, a pinprick hole was applied to the sitting drop well to allow for controlled evaporation. The crystals belong to space group P4_3_2_1_2 with a single Stk1 KD chain in the asymmetric unit. Crystals were looped and flash frozen in liquid nitrogen using the high sodium citrate concentration of the mother liquor as a cryoprotectant prior to data collection.

X-ray diffraction experiments were performed at the Canadian Light Source (CLS) at the University of Saskatchewan in Saskatoon, Saskatchewan, Canada. Data was collected at 100 K on the CMCF-BM (08B1) beam line with a Pilatus3 S 6M X-ray detector using MxDC data collection software. Diffraction data were processed using XDS^28^, and data reduction was performed with Aimless^29^ through the CCP4 software^30^. Data were corrected for anisotropic diffraction through elliptic truncation using the STARANISO server^31^. Phasing was carried out with molecular replacement using PDB 4EQM, chain A as the search model for Phaser-MR^32^ as part of the Phenix^33^ software package. The initial model was iterated upon through alternating steps of model building in Coot^34^ and refinement in Phenix. Model validation was carried out using MolProbity^35^. Final model and refinement statistics are provided in Table S5. Coordinates have been deposited to the Protein Data Bank with PDB code 9YMT. Figures were created using PyMOL^36^.

Stk1 KD activity assay and inhibition by GWX

Kinase activity assays were performed using the ADP-Quest assay kit (DiscoverX)^37^. The assay was programmed to run in a BioTek Synergy H1 microplate reader for 10 min at 25°C with excitation and emission wavelengths of 530 nm and 590 nm, respectively. Black polypropylene 96-well microplates (Greiner Bio-One) were used for assays with a final volume of 100 μL. 2-fold serial dilutions of GWX were incubated with STK1 for 5 min to allow for binding before combining with the remaining reactants. Measurements were performed with final concentrations of 100 nM STK1, 100 μM ATP, 1X Buffer A (20 mM HEPES pH 7.5, 150 mM NaCl, 20 mM imidazole), 1X Buffer B (20 mM HEPES pH 7.5, 150 mM NaCl, 50 mM imidazole), 2% DMSO, and 2000 nM-4 nM GWX.

Data analysis was performed using Microsoft Excel and a kinetics analysis was performed using non-linear regression, variable Hill slope in GraphPad Prism. Initial reaction rates were determined from the linear part of the progress curves. Inhibition of Stk1 at each inhibitor concentration was determined by comparing the rates of each compound against the “maximum rate” of uninhibited Stk1 (with 2% DMSO) after subtracting the background.

$$[eqn]$$

IC_50_ values and errors were derived from the non-linear regression analysis of these % inhibition curves using GraphPad Prism.

Isothermal Titration Calorimetry binding data of Stk1 KD and GWX inhibitor

Isothermal titration calorimetry (ITC) measurements were performed using a MicroCal PEAQ-ITC instrument (Malvern) at 25°C in high-feedback mode with a stirring speed of 750 rpm for sixteen 2 μL injections. GWX powder was solubilized to a concentration of 200–400 μM in Buffer D (20 mM HEPES pH 7.5, 150 mM NaCl, 2% DMSO) and then titrated into 13–20 μM purified Stk1 KD in a 200 μL volume to measure the heat of interaction. A control titration of GWX into Buffer D was integrated, and the resulting heat of dilution of the titrant was subtracted from the integrated heats of interaction to correct for it. The obtained data were fitted with a one-binding-site model to determine Kd, ΔH, ΔS, and the number of binding sites (n) using MicroCal PEAQ-ITC Analysis Software v1.41 (Malvern). Experiments and controls were performed in triplicate, and error analysis was performed using the Analysis Software package.

Caenorhabditis elegans killing assay

Infection of C. elegans DH26 was performed as described previously^24^. Briefly, after age synchronization, 15 L4 young adult worms were incubated with 1.5 × 10^5^ S. aureus in each well of a 96-well flat plate. Bacteria were suspended in 100 μL Liquid Assay media (80% M9 buffer, 20% TSB, 10 μg/mL cholesterol, and 7.5 μg/mL nalidixic acid). Nafcillin and/ or GWX were added at the indicated concentrations. The worms were incubated at 26°C, and the survival rate was recorded at 72 h after the addition of bacteria.

Results

mecA expression is mediated by the BlaR1-BlaI pathway in a majority of MRSA isolates

The MecI-MecR1 regulatory pathway is recognized to be impaired in a majority of known SCCmec types^12^. To better determine the predominance of MRSA where mecA was regulated via the BlaR1-BlaI pathway, we assessed all available S. aureus genomic entries on NCBI, including those for which only shotgun sequences were available. Of the 127,894 S. aureus entries found, 61.34% represented MRSA, as they were positive for mecA. A majority (85.19%) of these MRSA entries were positive for blaZ and blaI, which served as a surrogate to determine the presence of the Bla regulatory pathway (Fig. 1b). Interestingly, a significant proportion of the MRSA entries (75.90%) did not possess the mecI gene. Moreover, most of the entries devoid of mecI were positive for the blaZ and blaI genes (90.27%), signifying that the MecI-MecR1 pathway was impaired in a significant proportion of MRSA, and that mecA expression was under the control of the BlaR1-BlaI regulon in these entries (Fig. 1b). The above analysis demonstrated that BlaR1-BlaI-mediated mecA expression played a vital role in mediating broad-spectrum β-lactam resistance among a majority of the natural MRSA isolates sequenced to date. A prior study identified a key step in the activation of β-lactam resistance was the phosphorylation of the sensor-inducer protein BlaR1, following β-lactam detection^21,38^. This phosphorylation of BlaR1 has been attributed to Stk1, the kinase counterpart of the eSTK1 signaling pathway. Notably, we found stk1 and stp1, the two components of the eSTK1 pathway to be conserved in 99.71% of our 127894 analyzed S. aureus entries (Fig. 1b). In addition to S. aureus, the stk1 and stp1 genes are also conserved in several staphylococcal species^39^ as a bicistronic operon^7,40^. These analyses together suggest that the eSTK1 signaling pathway is highly conserved and plays an important role in β-lactam resistance.

Deletion of stk1 induces synthetic lethality to broad-spectrum β-lactam drugs

In order to determine the role of the stk1-stp1 operon on β-lactam resistance in MRSA, we deleted stk1 or stp1 from SF8300, a clinically isolated USA300 strain^2^ that contains an impaired MecR1-MecI regulatory pathway (type IV SCCmec) and has mecA controlled by the BlaR1-BlaI regulatory pathway. Following successful phenotypic validation of the deletion mutants (Figs. 1c, d and Figs. S1a, b), we performed growth assays for Wt (wild type), Δstk1, and Δstp1 in the absence or presence of nafcillin or oxacillin, two representative broad-spectrum, clinically relevant β-lactam drugs^41–43^. Growth assay without β-lactams (untreated) resulted in similar growth curves and bacterial CFUs for the three isogenic strains, suggesting that deletion of either of the genes did not cause a growth defect when evaluated in rich culture media (Figs. 1e). On the contrary, β-lactam treatment resulted in significantly attenuated bacterial growth and CFU formation for Δstk1 when compared to Wt and Δstp1, indicating that the loss of Stk1 led to β-lactam susceptibility (Fig. 1e). To determine the reproducibility of these results we deleted Δstk1 and Δstp1 in MW2, a clinically isolated, β-lactamase positive, USA400 strain^2^ with type IV SCCmec, and found a similar growth attenuation in Δstk1 treated with β-lactams when compared to Wt and Δstp1 (Fig. S1c). These findings suggested that the deletion of stk1 resulted in synthetic lethality to broad-spectrum β-lactams. We evaluated the basis of β-lactam susceptibility in Δstk1 by performing electron microscopy and detected substantial cell lysis in Δstk1 samples treated with nafcillin, compared to Wt and Δstp1 (Fig. 1f, Fig S1d). When untreated, no significant morphological differences were detected amongst the isogenic strains, demonstrating that the altered growth phenotype of Δstk1 was due to cell lysis caused by β-lactam exposure, and not due to factors such as growth arrest or small colony formation. Survival patterns and morphological characteristics of Δstp1 in the presence of β-lactams were similar to or greater than those seen in Wt for both SF8300 and MW2, demonstrating that loss of stp1 was favorable for β-lactam resistance (Figs. 1e–f, Fig. S1c-d).

To associate Stk1 function with the susceptible phenotypes of Δstk1, we complemented the mutant with either an empty vector (E), wild-type stk1 (Wt), or a mutant stk1 (K39G), using a constitutively expressing vector, pTXΔ (Figs. 2a–b, S1e-f). The K39G mutation renders Stk1 incapable of auto-phosphorylation, as well as target protein phosphorylation, making it enzymatically inactive^40^. When untreated, we detected comparable growth and CFUs for all complemented mutants, suggesting that neither the loss nor the constitutive expression of Stk1 had adverse effects on bacterial growth (Fig. 2c). However, in presence of β-lactams, the Δstk1 [stk1 (Wt)] strain showed increased survival as seen by enhanced growth and CFU formation compared to Δstk1 [E] and Δstk1 [stk1 (K39G)]. These results indicated that the synthetic lethality of Δstk1 was indeed due to the loss of Stk1 function and that Stk1 function was necessary for the production of β-lactam resistance. To associate the Stk1 dose with the magnitude of β-lactam resistance, we performed Δstk1 complementation using a xylose-inducible vector, pTX15. Growth curves for the resultant complemented strains were similar when untreated, with or without xylose induction (Fig. 2d). Upon nafcillin exposure, Δstk1 [stk1 (Wt)] remained susceptible at 0% xylose, but gained resistance as seen by enhanced growth when induced with increasing concentrations of xylose. At the highest xylose concentration of 0.2%, the growth curve of Δstk1 [stk1 (Wt)] was comparable to that displayed by Wt [E], indicating complete loss of susceptibility. Δstk1 [stk1 (K39G)] remained susceptible even at 0.2% xylose, substantiating that Stk1 function played a crucial role in mediating broad-spectrum β-lactam resistance (Fig. 2d).

We verified the phenotype identified in Δstp1 (Figs. 1e–f) by complementing it with an empty vector (E), wild-type stp1 (Wt), or a functionally inactive mutant of stp1 (G40A)^44^ using the constitutively expressing vector, pTXΔ (Figs. 2a–b, S1e-f). All strains had similar growth patterns when untreated (Fig. 2e). In presence of β-lactams, Δstp1 [E] and Δstp1 [stp1 (G40A)] remained resistant to nafcillin and oxacillin, while the presence of a functional Stp1 in Δstp1 [stp1 (Wt)] led to significant decrease in growth, indicating β-lactam susceptibility. These findings confirmed that the enhanced β-lactam survival seen in Δstp1 (Figs. 1e–f) was due to loss of Stp1 function, and indicated that Stk1 and Stp1 affected β-lactam resistance in a contrasting manner.

Stk1 is required for efficient induction of mecA upon broad-spectrum β-lactam exposure

Since efficient induction of mecA expression is required for the manifestation of broad-spectrum β-lactam resistance, we next investigated whether Stk1 played any role in mecA expression by quantifying the increase in mecA transcripts and PBP2a expression in response to β-lactam exposure (Figs. 3a–e). qRT-PCR analysis revealed that exposure to nafcillin (Fig. 3a) or oxacillin (Fig. 3b) led to a significantly attenuated induction of mecA transcription in Δstk1 when compared to Wt and Δstp1 (Figs. 3a–b). To validate these at the protein level, we performed immunoblotting to determine the expression of PBP2a among the isogenic strains (Figs. 3c–e). When untreated, PBP2a expression was minimal but comparable among all three strains (Figs. 3c, S2a). Nafcillin (Fig. 3d, S2a) or oxacillin (Fig. 3e, S2a) exposure led PBP2a induction in all three strains, but Δstk1 induced PBP2a significantly less when compared to Wt and Δstp1. Thus, findings of immunoblotting assays were consistent with those of qRT-PCR, demonstrating that during β-lactam exposure, Δstk1 had a dampened mecA induction that resulted in inadequate PBP2a expression and β-lactam susceptibility.

Chemical inhibition of Stk1 function in wild-type bacteria produces sensitivity to broad-spectrum β-lactams by attenuating mecA induction

Since the above findings suggested that loss of Stk1 function was favorable for sensitization of MRSA to broad-spectrum β-lactams (Figs. 1e, 2c), we used GW779439X (GWX), an established kinase inhibitor, to inhibit Stk1 function in Wt^45^. We validated the in vitro activity of GWX to inhibit the S. aureus kinase domain (Stk1^1−291^) through the commercially available ADP-QuestTM assay kit, which monitors kinase-induced ADP production via fluorescence^46^. The ability of the GWX inhibitor to abrogate kinase activity was tested using this assay, which produced an in vitro IC50 value of 188.5 nM^46^ (Fig. 3f). Next, we used isothermal titration calorimetry (ITC) to determine the thermodynamic parameters associated with the binding of GWX to Stk1^1−291^ with a calculated K_d_ of 21.1nM ± 11.1nM (Fig. S2b) highlighting the potent nature of GWX. To investigate the molecular basis of Stk1 inhibition by GWX in further detail, we co-crystallized the protein-ligand complex and determined its structure through X-ray diffraction methods to a 2.1 Å resolution (Table S5). Stk1^1−291^ adopts a Hanks-type kinase fold, a highly conserved fold common to many prokaryotic and eukaryotic kinases^39,47^. This fold has a characteristic bi-lobal feature with an N-terminal lobe (N-lobe) comprised of a 5-stranded β-sheet connected primarily through β-hairpins and a regulatory α-helix known as the C-helix, while the C-terminal lobe (C-lobe) is comprised of primarily helices and loops (Figs. 3g–h, Figs S2c-d). The Stk1-GWX inhibitor complex structure superposes closely with that of the previously solved Stk1 structure bound to the non-cleavable substrate analog AMP-PNP (PDB ID: 4EQM^27^). Using PyMol’s align function, model 4EQM was compared to our structure across all residues with a final RMSD of 0.608 Å after 5 cycles, indicating little conformational difference in the kinase domain upon binding of GWX. The N- and C-lobes of the kinase domain create an extensive active site cleft at their interface which accommodate the ATP and protein substrates upon which it acts. In our inhibited structure, this cleft is occupied instead by the substrate analog GWX ligand. In the crystallographic structure determined here, residues 161–170 and 284–291 remain unmodelled due to disorder in the corresponding regions of the electron density maps, and likely as a result of the high flexibility of the so-termed activation loop and the C-terminus, respectively. The activation loop is one of the regulatory elements of the kinase active site, participating in protein-protein interactions, regulating substrate specificity and containing conserved serine/threonine residues used in autophosphorylation/transphosphorylation. These regions were also lacking clear electron density and were presumed dynamic in the prior AMP-PNP structures, 4EQM. In our structure of the kinase domain, Asp 151 of the DFG motif is pushed away from the active site cleft into the “DFG-out,” conformation where it can no longer participate in the co-ordination of magnesium ions as part of its role in catalysis^48,49^. Instead, the preceding Phe150 lies in the adenine pocket and forms a pi-stacking interaction with the pyrazolopyridine moiety of the GWX ligand. The αC-helix of Stk1^1−291^ also adopts the inactive conformer and is positioned such that the conserved Glu 58 in the C-Helix is blocked by the activation loop, preventing it from making a salt-bridge with Lys 39, a key electrostatic determinant of kinase activation that we have mutated for a phosphorylation-deficient strain in other aspects of this study^50,51^ (Figs. 2c and 3g–h). (Refer to supplementary text for more results).

After determining the molecular basis of GWX-mediated Stk1 inhibition, the only conserved hanks-type kinase in S aureus, we examined the phenotypic effect of using the inhibitor in combination with β-lactams in Wt and Δstk1. We first carried out a growth assay with GWX alone and found that the inhibitor did not cause any phenotypic alterations, as both Wt and Δstk1 had similar growth patterns without and with GWX (Fig. 3i). We then treated Wt with nafcillin combined with a range of GWX concentrations (0–0.625 μg/mL) and found the strain to gain increasing susceptibility in a GWX dose-dependent manner. When co-treated with nafcillin and 0.625 μg/mL GWX, we detected a significant growth attenuation in Wt that was comparable to that displayed by nafcillin-treated Δstk1, indicating that GWX rendered Wt susceptible to nafcillin. (Fig. 3i). Δstk1 had similar growth patterns when treated with only nafcillin or with a combination of nafcillin and GWX, suggesting that the phenotype displayed above was specifically due to GWX-mediated inhibition of Stk1 function. To determine if the susceptibility caused by nafcillin and GWX in Wt mimicked the effect of stk1 deletion, we performed qRT-PCR and immunoblotting to quantify mecA and PBP2a expression (Figs. 3j–k, S2e). While Wt exposed to nafcillin displayed mecA induction (Fig. 3j) and increased PBP2a production (Fig. 3k), exposure to nafcillin combined with GWX caused a significant reduction in levels of mecA and PBP2a. In alignment with findings of growth curve analysis (Fig. 3i), diminished levels of mecA and PBP2a in Wt were comparable to those seen in nafcillin-treated-Δstk1, confirming that inhibition of Stk1 by GWX mimicked the effect of stk1 deletion. Since β-lactamase activity was reported to be attenuated in cultures induced by CBAP (2-(2’-carboxyphenyl)-benzoyl-6-aminopenicillanic acid) combined with kinase inhibitors^21^, we explored the effect of nafcillin and GWX on blaZ induction using qRT-PCR and observed similar results to those seen by mecA (Fig. S2f).

Stk1 is required for efficient proteolysis of BlaI upon broad-spectrum β-lactam exposure

We next determined whether Stk1 affected the signal transduction step preceding mecA induction, namely the proteolysis of the transcriptional repressor BlaI^14,52^ through immunoblotting analysis (Figs. 4a–c). Following the validation of a custom-generated BlaI antibody (Figs. S3a-b), when untreated, we detected BlaI in similar amounts in our three isogenic strains, suggesting that deletion of stk1 or stp1 did not intrinsically alter its abundance (Figs. 4a, S3c). Quantification of BlaI following exposure to nafcillin (Figs. 4b, S3c) or oxacillin (Figs. 4c, S3c) revealed that BlaI was highly diminished in the Wt and Δstp1 samples, but was largely intact in Δstk1. Detection of a significant amount of intact BlaI in Δstk1 signified insufficient proteolysis, resulting in weakened mecA de-repression and broad-spectrum β-lactam susceptibility seen above (Fig. 3). To associate insufficient BlaI cleavage with β-lactam susceptibility, we complemented Wt with an empty vector [E], wild-type blaI [Wt], or a non-cleavable mutant blaI [N101A and F102A] and assessed their response to nafcillin (Figs. 4d–f). As the amino acid substitutions in Wt [blaI (N101A and F102A)] prevented proteolysis of BlaI by the sensor-inducer BlaR1^53^, we detected intact BlaI in the strain when left untreated (Fig. 4d, Fig. S3d), as well as when exposed to nafcillin (Fig. 4e, Fig. S3d) by immunoblotting. On the other hand, Wt complemented with blaI [Wt] showed efficient BlaI degradation as expected. Wt [blaI (N101A and F102A)] did not display growth alterations when untreated (Fig. 4f); however, it was significantly susceptible to nafcillin compared to Wt [blaI (Wt)] in growth assays, as Wt [blaI (Wt)] successfully underwent BlaI proteolysis (Fig. 4e) and thereby survived in the presence of nafcillin. We also complemented Δstk1 with the BlaI variants for growth assays, and found that Δstk1 [blaI (Wt)] and Δstk1 [blaI (N101A and F102A)] were both susceptible to nafcillin (Fig. 4f), suggesting that Stk1 function, as well as efficient BlaI proteolysis, were both crucial for β-lactam resistance. Furthermore, we found that nafcillin-treated Wt, Δstk1, and Δstp1 samples had similar levels of blaI induction (Fig. 4g), indicating that the BlaI degradation deficiency in Δstk1 was post-transcriptional in nature.

Stk1 is required for optimal BlaR1 presence in the bacterial cell membrane

BlaR1 is the polytopic membrane protein and sensor-inducer receptor that mediates the cleavage of BlaI^14,54^. To determine if attenuated BlaI proteolysis in Δstk1 was due to altered amounts of BlaR1 prior to the detection of β-lactams, we performed immunoblotting assays with membrane fractions of our isogenic strains (Figs. 5a–b) following validation of a custom-generated BlaR1 antibody targeting the cytosolic protease helix (Figs. S4a-d). We detected significantly reduced BlaR1 in Δstk1 compared to Wt at both the mid- (Figs. 5a, S4e) and late-exponential (Figs. 5b, S4f) growth phases, indicating that Stk1 is important for BlaR1 presence in the cell membrane. To further attribute the BlaR1 membrane localization defect to the loss of Stk1 function, we complemented Δstk1 with an empty vector [E], stk1 [Wt], or with stk1 [K39G] and observed that BlaR1 remained severely attenuated in the membrane fractions of Δstk1 [E] and Δstk1 [stk1 (K39G)]. Only the complementation with a functional Stk1 in Δstk1 [stk1 (Wt)] enabled normal membrane levels of BlaR1 (Figs. 5c, S5a), verifying that Stk1 promoted normal localization of BlaR1 into the membrane. Further, we used GWX to inhibit Stk1 in Wt and observed a drastic decrease in BlaR1 levels in the cell membrane (Figs. 5d, S5b), substantiating that Stk1 function was vital for BlaR1 localization. In Δstp1, BlaR1 levels were similar to that of Wt particularly at the late-exponential phase (Fig. 5b). We complemented Δstp1 with an empty vector [E], stp1 [Wt], and stp1 [G40A] and saw that Δstp1 [stp1 (Wt)] led to significant attenuation of BlaR1, whereas the absence of a functional Stp1 in Δstp1 [E] and Δstp1 [stp1 (G40A)] was favorable for efficient BlaR1 localization in the membrane (Fig. 5c). Stk1 and Stp1 thus had contrasting effects on BlaR1, as Stk1 promoted BlaR1 localization in the membrane, while Stp1 impeded it (Figs. 5a–d). Remarkably, these alterations in membrane-anchored BlaR1 levels were detected in the absence of β-lactam exposure, suggesting that the impact of Stk1 and Stp1 functions on BlaR1 was independent of β-lactams.

Since our findings showed that BlaR1 was more severely attenuated in Δstk1 in late- and in mid-exponential phase (Figs. 5a–b), we next questioned whether it was due to an altered expression of blaR1. On performing qRT-PCR, we observed that at the mid-exponential growth phase, blaR1 was present in low but similar amounts in Wt, Δstk1, and Δstp1. At the late-exponential phase, blaR1 transcripts were more abundant in all three strains compared to mid-exponential phase, but it was significantly attenuated in Δstk1 compared to Wt and Δstp1 (Fig. 5e). These findings suggest that Stk1 primarily modulates BlaR1 at the post-transcriptional level during mid-exponential growth. In the late-exponential phase, the observed increase in blaR1 transcripts across strains likely reflects a leaky activation of the BlaR1-BlaI autoregulatory pathway^38^. In Δstk1, impaired membrane localization of BlaR1 during mid-exponential growth may delay initiation of this autoregulatory loop, resulting in dampened blaR1 expression later in growth. We confirmed this notion by assessing transcriptional and protein expression of BlaI. Similar to blaR1, blaI also displayed an overall increase in abundance in transcripts at the late-exponential phase (Fig. S6a). In alignment with BlaR1, the abundance of blaI transcripts, as well as BlaI protein, was attenuated in Δstk1 (Fig. S6a-c).

Stk1-mediated phosphorylation is required for facilitating membrane localization of BlaR1

BlaR1 is reported to sense β-lactams in the bacterial environment and undergo phosphorylation by Stk1^21^, triggering subsequent signaling events leading to the induction of broad-spectrum β-lactam resistance. Since BlaR1 has been suggested to undergo phosphorylation at least at one tyrosine residue^21^, we assessed the phosphorylation state of BlaR1 inserted into the membrane by probing it for phospho-tyrosine (P-tyrosine). We detected BlaR1 in Wt and Δstp1 when probed for P-tyrosine, demonstrating that the protein was phosphorylated at tyrosine residues (Figs. 5f, S5c). Further, in alignment with previous results (Figs. 5a–b), the significant reduction of phosphorylated BlaR1 in the membrane of Δstk1 suggested that phosphorylation of BlaR1 was vital for protein membrane localization (Fig. 5f). Treatment of the Wt with the kinase inhibitor, GWX, led to a reduction in membrane-localized BlaR1, which was accompanied by a corresponding decrease in its phosphorylation (Fig. 5d). These findings also revealed that phosphorylation of BlaR1 occurred independent of β-lactam exposure. These results together reiterated that Stk1-mediated phosphorylation was crucial for BlaR1 membrane localization.

We had difficulty being able to detect the reported autocleavage of BlaR1 following detection of β-lactam due to limited sensitivity of our Western blot assays. At the time that BlaI is being degraded (i.e., at 10 min after β-lactam induction), no discernable BlaR1 degradation fragment can be detected. Only until much later at 20 min after induction that enough BlaR1 degradation fragments are able to accumulate to be detectable via our assay, at which we are not able to detect a difference between the isogenic strains. The lack of difference seen at the later time point could be due to the induction of additional BlaR1 and the degradation fragments from said newly synthesized BlaR1 proteins.

Discussion

Resistance to broad-spectrum β-lactams in MRSA remains a global public health problem^37^. While various last-resort antibiotics are currently available to treat MRSA, owing to β-lactams’ high efficacy and safety, efforts to re-sensitize MRSA to the antibiotic class are ongoing. Previous studies have highlighted the importance of BlaR1, the membrane-anchored, de facto sensory inducer of narrow-spectrum (BlaZ-mediated) β-lactam resistance, which, in addition, also controls broad-spectrum (mecA-mediated) β-lactam resistance in USA300 background strains of MRSA^53,55^. USA300 strains possess an intact, fully functional BlaZ regulatory system but bear a truncated mecA counterpart. Owing to a highly similar basis of regulation between the BlaZ and mecA regulatory systems, the expression of mecA in USA300 strains, in the absence of its cognate regulatory elements, is controlled via the BlaR1-BlaI pathway^14^. It is postulated that such uncanonical control of mecA expression enables the bacteria to mount a far superior and more effective broad-spectrum β-lactam resistance than its canonical counterpart (i.e., through MecR1-MecI)^56^. Since the blaZ and mecA regulatory systems in S. aureus are plasmid and genome (SCCmec) encoded, respectively, control of mecA expression in this manner exemplifies regulation of a core-genome element via a pathway that is plasmid encoded.

Even though the majority of the SCCmec clusters identified to date possess a truncated mecA cluster^12^, closely similar to that described above, the prevalence of such an uncanonical mode of control of mecA expression among natural MRSA strains is unclear. It is particularly so, since it is unknown how many strains with a truncated mecA cluster also possess the BlaZ regulatory system, a prerequisite for such an uncanonical mode of broad-spectrum β-lactam resistance to occur. To determine this, we analyzed all the genomic data entries of S. aureus strains available in NCBI and examined them for the presence of genes mediating blaZ and mecA regulatory pathways. Our results indicated that >75% of the ~78,000 MRSA genomic entries analyzed had a truncated mecA cluster (Fig. 1b). More importantly, >90% of ~60,000 entries for such MRSA entries that had a truncated mecA cluster possessed the BlaZ regulatory system, highlighting the fact that mecA expression in these strains is likely under the control of the BlaZ signaling system. Interestingly, a much lower percentage of entries (~70% as opposed to 90%) of ~19,000 entries for MRSA strains possessing an intact mecA cluster had the BlaZ regulatory system. The reason for this decreased association of the BlaZ regulatory pathway among strains that had an intact mecA counterpart is perhaps because the presence of both intact BlaZ and mecA regulatory systems is unable to mount an effective response to produce β-lactam resistance^56^. Furthermore, it is likely that broad-spectrum β-lactam resistance in certain MRSA strains with both intact blaZ and mecA regulatory systems is also controlled by the BlaR1-BlaI pathway. A previous study showing BlaR1-BlaI pathway-mediated control of mecA expression in a common hospital-associated MRSA (CC5/ST5) strain with type II SCCmec provided an example of such a control mechanism^14^. These results underscore that among the majority of natural MRSA isolates, inducible expression of broad-spectrum β-lactam resistance is controlled by BlaZ regulatory elements.

BlaR1 plays a pivotal role in the signal transduction process that enables a timely, cost-effective response to β-lactam drugs for the production of drug resistance. Although a full understanding of the mechanistic principles of the signaling pathway is currently lacking, previous studies have shown that the sensory domain of BlaR1, upon recognition of β-lactam drugs in the bacterial surroundings, undergoes conformational changes that propagate through the membrane helices to the cytosolic facing protease domain, triggering cleavage of the BlaI repressor substrate, which in turn enables de-repression of blaZ expression and production of drug resistance^14,53,54,57^. Recent studies have indicated that, of its many functions, the eSTK signaling pathway can regulate broad-spectrum β-lactam resistance in S. aureus. The kinase component of the pathway, Stk1, was previously associated with sensing β-lactams in the bacterial environment, phosphorylation of BlaR1, and modulation of BlaZ-mediated narrow-spectrum β-lactam resistance in S. aureus^21,56^. It was also shown to be involved in controlling broad-spectrum β-lactam resistance in USA300 background strains through modulation of mecA expression^55^. The basis through which eSTK exerted such roles in the bacteria was, however, unknown. In this study, we demonstrated that Stk1-mediated phosphorylation of BlaR1 is essential for optimal presence of BlaR1 in the bacterial membrane, allowing it to perform its transmembrane signaling and regulatory role. Consequently, a loss of Stk1 function severely impaired BlaR1 levels in the cell membrane, which in turn resulted in attenuated proteolytic cleavage of BlaI and mecA (as well as blaZ) induction during β-lactam assault. Through the usage of robust genetic, biochemical, and phenotypic assays in determining these results, we have also shown that both the function and dose of Stk1 are important for the manifestation of broad-spectrum β-lactam resistance. Our results demonstrated that Stk1-mediated BlaR1 phosphorylation likely occurs independent of β-lactam treatment and that membrane levels of BlaR1 are coupled to its phosphorylation status.

Our immunoblotting assays suggested that Stk1 is likely a key mediator of BlaR1 phosphorylation. We provided results that support the chemical inhibition of Stk1 by GWX potentiates synthetic lethality to broad-spectrum β-lactams in MRSA strains lacking a functional MecI-MecR1 and provided the structural basis for the potent inhibition of Stk1 by GWX. We substantiated our in vitro data through an in vivo infection assay using C. elegans. Our results showed that worms infected with Wt had increased survival when treated with nafcillin and GWX, compared to worms treated with nafcillin only (Fig. 6a). Increased survival of C. elegans infected with Δstk1 with nafcillin, or Wt with nafcillin and GWX, indicated that targeting Stk1 could be beneficial to treat S. aureus infections, particularly when used in combination with an established β-lactam drugs. In addition to these, our results revealed the previously unknown role of Stp1, the cognate phosphatase of the eSTK signaling pathway. Stp1 oppositely regulates BlaR1 phosphorylation and membrane localization and consequently mediates a contrasting effect to that of StK1 in mediating broad-spectrum β-lactam resistance.

Outlook and limitations:

In this study, we demonstrated that Stk1 mediated phosphorylation of BlaR1 is crucial for the induction of broad-spectrum β-lactam resistance, and inhibition of Stk1 can re-sensitize MRSA lacking a functional MecI-MecR1 to broad-spectrum β-lactams (Fig. 6b). Our analysis suggested that an approach that uses an effective Stk1 inhibitor as an adjuvant to produce synthetic lethality to broad-spectrum β-lactams could be widely applicable to majority of the natural MRSA strains. The high-resolution structure of Stk1-GWX inhibitory complex solved in this study will allow for the design of derivatives that can be rationally modified to serve as more specific targets. GWX was originally designed to target human CDK4 and was chosen for this study for its aqueous solubility and relative potency compared to the other inhibitors screened.

We recognize that this study also has several limitations, which would need to be addressed in the future. Firstly, BlaR1, being a polytopic transmembrane protein, is likely inserted into the bacterial membrane co-translationally via the SecYEG section machinery and the associated insertase, YidC^58^. Once inserted and folded into the membrane, it can be subjected to modification by Stk1, which makes the cytosolic Zinc metalloprotease domain of BlaR1 the likely target of phosphorylation. Currently, the the sites at which BlaR1 phosphorylation/s occurs are unknown. Furthermore, the exact cause for the decreased abundance/localization of BlaR1 in the Δstk1 also would need to be sorted out. Finally, the role played by Stp1 in β-lactam resistance needs to be resolved in further detail. Particularly, whether it solely dephosphorylates Stk1 or can it also dephosphorylate BlaR1 to mediate its function is unknown.

Supplementary Files

This is a list of supplementary files associated with this preprint. Click to download.

Supplementarytext.pdf

Supplementaryfigures.pdf