SmdA is a Novel Cell Morphology Determinant in Staphylococcus aureus

ABSTRACT Cell division and cell wall synthesis in staphylococci need to be precisely coordinated and controlled to allow the cell to multiply while maintaining its nearly spherical shape. The mechanisms ensuring correct placement of the division plane and synthesis of new cell wall have been studied intensively. However, hitherto unknown factors and proteins are likely to play key roles in this complex interplay. Here, we identified and investigated a protein with a major influence on cell morphology in Staphylococcus aureus. The protein, named SmdA (for staphylococcal morphology determinant A), is a membrane protein with septum-enriched localization. By CRISPRi knockdown and overexpression combined with different microscopy techniques, we demonstrated that proper levels of SmdA were necessary for cell division, including septum formation and cell splitting. We also identified conserved residues in SmdA that were critical for its functionality. Pulldown and bacterial two-hybrid interaction experiments showed that SmdA interacted with several known cell division and cell wall synthesis proteins, including penicillin-binding proteins (PBPs) and EzrA. Notably, SmdA also affected susceptibility to cell wall targeting antibiotics, particularly in methicillin-resistant S. aureus (MRSA). Together, our results showed that S. aureus was dependent on balanced amounts of membrane attached SmdA to carry out proper cell division.

IMPORTANCE Staphylococcus aureus is an important human and animal pathogen. Antibiotic resistance is a major problem in the treatment of staphylococcal infections, and cell division and cell wall synthesis factors have previously been shown to modulate susceptibility to antibiotics in this species. Here, we investigated the function of a protein named SmdA, which was identified based on its septal localization and knockdown phenotype resulting in defective cellular morphologies. We demonstrated that this protein was critical for normal cell division in S. aureus. Depletion of SmdA sensitized resistant staphylococci to b-lactam antibiotics. This work revealed a new staphylococcal cell division factor and a potential future target for narrow-spectrum antimicrobials or compounds to resensitize antibiotic-resistant staphylococcal strains.
KEYWORDS cell biology, cell division, cell wall, subcellular localization M ost bacteria are surrounded by a shape-determining cell envelope that protects against lysis and interacts with the extracellular milieu. The cell envelope of the opportunistic, Gram-positive pathogen Staphylococcus aureus consists of a thick layer of peptidoglycan (PG) along with teichoic acids (TA) and cell wall-associated surface proteins. During a bacterial cell cycle, the synthesis of PG and TA needs to be precisely regulated and coordinated with cell division, DNA replication, and chromosome segregation. Tight control of these processes is critical for staphylococcal cells to maintain their integrity and nearly spherical shape as they multiply. Therefore, they are attractive targets for antimicrobials (1). Exactly how such control is mediated in S. aureus is still not fully established, and hitherto unknown factors may be involved. Here, we described a new staphylococcal cell morphology determinant.
Staphylococcal cell division is initiated by the assembly of the Z-ring, consisting of polymerized FtsZ-proteins, that localizes to the future division septa (2). The Z-ring functions as a scaffold for cell division and cell wall synthesis proteins which together constitute the divisome (3). Cell division in S. aureus occurs in alternating orthogonal planes, meaning that the new cell division plane is always perpendicular to the previous (4). Timely and spatial control of localization of the Z-ring assembly is most likely linked with chromosome segregation and DNA replication, involving proteins such as the nucleoid occlusion factor Noc, which ensures that the cells do not establish new septa across the chromosomes (5), and CcrZ, which connects initiation of DNA replication to cell division (6). The chromosomes and chromosome segregation also contribute to establishing a physical barrier allowing the Z-ring only to be formed at an angle perpendicular to the previous division plane (4).
The Z-ring directs the synthesis of new PG in S. aureus to the septum. Synthesis of PG starts in the cytoplasm, where UDP-MurNAc-pentapeptide is first synthesized and then attached to the membrane by the enzyme MraY to form the PG precursor lipid I (7,8). A GlcNAc residue and a pentaglycine side chain are attached to produce lipid II-Gly 5 (9), which is flipped to the outer leaflet of the membrane by MurJ (10,11) where it is incorporated into the existing PG mesh by transpeptidation (TP) and transglycosylation (TG) reactions. Specifically, the shape, elongation, division, and sporulation (SEDS) proteins, FtsW and RodA with TG activity, work in pairs with monofunctional transpeptidases, the penicillin-binding proteins PBP1 and PBP3, respectively (12,13). While the PBP1-FtsW pair is essential and performs the septal cross wall synthesis, the nonessential PBP3-RodA pair is responsible for the slight elongation taking place in S. aureus. Additionally, S. aureus possesses two other PBPs: the bifunctional PBP2 with both TG and TP activity, whose role is essential in S. aureus, and the low-molecular-weight PBP4, which controls the degree of PG cross-links (14)(15)(16). Finally, MRSA strains have an additional PBP, PBP2A, a transpeptidase with low-affinity for b-lactam antibiotics (17,18). In the final step of division, PG hydrolases break covalent bonds in PG for cell wall remodeling and daughter cell splitting. The major, bifunctional autolysin Atl, together with Sle1, for which expression is regulated by the two-component system WalKR, are the primary enzymes responsible for hydrolyzing the septal PG to allow splitting of daughter cells (19)(20)(21)(22). The actual cross wall splitting is a mechanical process occurring within milliseconds (23,24).
The spatiotemporal control of cell division and PG synthesis is directly and indirectly influenced by several factors. One of these is the anionic TA polymers, the second major component of the cell wall, which are either covalently linked to the PG (wall teichoic acids [WTA]) or linked via a lipid-anchor to the plasma membrane (lipoteichoic acids [LTA]). Mutations in enzymes involved in either WTA or LTA biosynthesis result in cells of abnormal shape and lack of septum synthesis control, probably via different mechanisms (15,25,26). Furthermore, proteases, chaperones, and secretion proteins, involved in the production, folding, and/or secretion of cell cycle proteins, may also directly or indirectly affect the coordination of cell division and septum formation in S. aureus. For example, Clp-protease complexes can target both FtsZ and Sle1, and thereby have major effects on these processes (27)(28)(29).
Cell shape maintenance and control of cell division and septum formation are a complex interplay between many cellular processes where unknown key factors are yet to be discovered. Here, we identified a novel protein named SmdA (staphylococcal morphology determinant A). We showed that the level of the membrane attached SmdA protein was critical for maintaining normal division progression and morphology in S. aureus because silencing or overexpression of smdA resulted in defective cell division and septal cross wall synthesis as well as increased sensitivity toward cell wall targeting antibiotics.

RESULTS
SmdA is a conserved staphylococcal membrane protein.
To identify novel proteins potentially involved in cell cycle and morphology control in S. aureus, we performed a combined depletion and subcellular localization analysis of essential staphylococcal proteins with no annotated functions (30)(31)(32). From this, we identified a septum-enriched protein (SAOUHSC_01908, named SmdA), whose knockdown resulted in cells of variable sizes that formed large clusters (see below). The function of SmdA in S. aureus was investigated further here.
SmdA is a protein of 302 amino acids which is fully conserved in species within the Staphylococcaceae family (Fig. S1). The smdA gene is monocistronic and located .100 bp away from the neighboring genes (SAOUHSC_01907; unknown function, and metK; S-adenosylmethionine synthase). The protein had a predicted N-terminal transmembrane helix and a C-terminal cytoplasmic part with partial homology to a so-called nuclease-related domain (NERD) (PF08375, E = 1.08Â10 25 ) (Fig. 1A). The domain was named based on distant similarities to endonucleases and was found in bacterial, archaeal, and plant proteins (33). However, the functional role of NERD in bacteria has to our knowledge never been studied. SmdA is essential in transposon mutagenesis studies in S. aureus (30,32). Growth analysis of SmdA CRISPRi knockdown (SmdA down ) in S. aureus SH1000 in a rich medium at 37°C resulted in a reduction in growth compared to the control strain when spotted on agar plates (Fig. 1B). The growth rate in the liquid medium was only slightly affected when CFU and the optical density at 600 nm (OD 600 ) were determined at different time points in the same culture during growth (Fig. S2A). By RT-PCR we verified that the expression of smdA was fully knocked down by the CRISPRi system in S. aureus SH1000 (Fig. S2B). Because the SmdA down strains were still viable, we attempted to construct deletion mutants of smdA by allelic replacement with a spectinomycin resistance cassette using the pMAD-vector (34). However, we were not able to obtain the deletion mutant in S. aureus SH1000. Similarly, for S. aureus strains NCTC8325-4, HG001, and the MRSA strain COL, SmdA down resulted in reduced growth on agar plates, with HG001 being most affected (Fig. 1B), but we were not able to obtain any deletion mutants in these strains. We, therefore, used the CRISPRi system in the different S. aureus strains to study the phenotypes of SmdA further.
Depletion and overexpression of SmdA result in cells with highly aberrant cell shapes. During the initial analysis, we observed that SmdA down in S. aureus SH1000 resulted in clusters containing cells of variable sizes. The knockdown experiment was repeated, and exponentially growing cells were stained with fluorescent vancomycin (VanFL, binds to noncross-linked stem peptides throughout the cell wall). SmdA down indeed resulted in severe phenotypic defects (Fig. 1C): increased cell clustering and a large fraction of cells with multiple septa (20.1% for SmdA down as opposed to 1.6% for the control strain; Fig. 1D) and abnormal, nonspherical morphology (Fig. 1E). The SmdA down cells were also significantly larger than the control strain (Fig. 1F).
Transmission and scanning electron microscopy (TEM and SEM) were used to obtain more detailed images of the defects in morphology and septal placement found in SmdA down cells. Strikingly, SmdA depleted S. aureus SH1000 displayed highly aberrant septum formation ( Fig. 2A). In addition to lysis, cells with several nonperpendicular or parallel septa were frequently observed, resulting in cells, or small cell clusters, with aberrant morphologies (Fig. 2A). This was also evident from the SEM micrographs, which showed clustered cells with various morphologies (Fig. 2B). Similar phenotypes from TEM and SEM analyses were observed for the NCTC8325-4, HG001, and COL strains, with the HG001 strain being more affected by SmdA down than the other two ( Fig. S3 and S4).
Because reduced levels of SmdA led to defects in cell division and morphology of S. aureus, we next wondered whether overexpression of SmdA would affect the cells. An ectopic copy of smdA under the control of an IPTG-inducible promoter in the plasmid  pLOW (35) was expressed in S. aureus NCTC8325-4. For SmdA down , the cells were stained with VanFL ( Fig. S5A). Although less evident than for SmdA down , overexpression of SmdA also resulted in clusters of cells with several septa per cell as defined by VanFL staining (Fig. S5A) (3.9%, n = 181 for SmdA overexpression, compared to 20.1% for SmdA down , and 1.6% for the control). Together, these knockdown and overexpression experiments demonstrate that septum formation and splitting in S. aureus were dependent on proper levels of SmdA.
Depletion of SmdA results in increased sensitivity toward antimicrobials targeting cell wall synthesis. Given the negative effect SmdA depletion had on cell morphology and division, we reasoned that reduced expression of SmdA could influence the sensitivity of S. aureus to cell wall targeting antibiotics. To test this, SmdA down strains were treated with PBP-targeting b-lactams (oxacillin, cefotaxime, cefoxitin, and imipenem), the glycopeptide vancomycin (blocking cell wall synthesis by targeting the terminal D-Ala-D-Ala on the stem peptides of nascent PG (36)), tunicamycin (targeting TarO and MraY, enzymes involved in the early stages of WTA and PG synthesis, respectively (37)), targocil (targeting the WTA exporter TarG (38)) and Congo red (inhibitor of the LTA biosynthesis enzyme LtaS (39)) ( Table 1). Two antibiotics with alternative targets, tetracycline (targeting protein synthesis) and ciprofloxacin (a quinolone targeting DNA synthesis), were also included.
For the methicillin-susceptible S. aureus (MSSA) strain SH1000, a 2-fold reduction in MIC, compared to the control, was observed for oxacillin and cefoxitin in SmdA down . More notably, in the MRSA strain COL, SmdA down sensitized the COL strain toward all b-lactams with a 2 to 8-fold reduction in MIC compared to the control (Table 1). SmdA depletion did not seem to significantly influence vancomycin susceptibility. Strikingly, however, we observed that SmdA down cells became highly sensitive toward tunicamycin, with a 64-fold reduction in MIC compared to the control for S. aureus SH1000 and .4-fold reduction for COL (40,41). We therefore also tested the tunicamycin sensitivity of the NCTC8325-4 and HG001 SmdA down strains, which also showed increased susceptibility toward tunicamycin (HG001; 125-fold reduced MIC and NCTC8325-4; 4-fold reduced). While no difference in MIC was observed for targocil, the SmdA down strains also displayed increased sensitivity toward Congo red. Finally, SmdA depletion in S. aureus SH1000 led to a 2 to 4-fold reduction in MIC against tetracycline compared to the control. However, depletion of SmdA in COL did not change its sensitivity toward tetracycline or ciprofloxacin (Table 1).
SmdA has no major effects on the TA biosynthetic pathways. The increased sensitivity to tunicamycin and Congo red in SmdA down strains, prompted us to study whether there were any major alterations in the TA in these cells, although it was also noted that no changes in sensitivity were observed for the WTA export inhibitor targocil (Table 1). Notably, the SmdA down strain displayed morphologies reminiscent of what has previously been reported for cells depleted of TA, i.e., with larger cell sizes, cells with irregular septum formations, and reduced splitting (26,37,38,42). It has previously been shown that there was a synthetic lethal relationship between the WTA and LTA biosynthetic pathways (42,43), and it could therefore be hypothesized that hypersensitivity to tunicamycin could result from deficient LTA biosynthesis in the smdA down mutants, Using an anti-LTA antibody, we compared the quantity and lengths of LTA in the SmdA down and control strains for SH1000, NCTC8325-4, HG001 and COL (Fig. S6A). For HG001, the strain with the most severe division defects, we observed a reduction in LTA amounts in SmdA down and this can contribute to the severe phenotype of this strain. However, no consistent changes in LTA amounts or lengths were observed between the SmdA depletions and the controls in the four strains. Furthermore, we could not detect any LTA release into the growth medium in the depletion strains, indicating that the stability of LTA (44) was intact (Fig. S6B). We, therefore, conclude that SmdA does not have any consistent effect on LTA synthesis across strains. WTA has been shown to protect cells from the LTA-inhibitor Congo red. Without WTA, cells became hypersensitive toward Congo red (MIC of ,4 mg/mL for tarO deletion mutants and .1024 mg/mL for wild-type cells) (45). Although to a much lesser degree, SmdA depletion strains were also more sensitive to Congo red compared to the controls (Table 1). We, therefore, investigated whether WTA could be disturbed in a SmdA down strain. Cells without WTA have previously been shown to lack the dark, electron-dense layer observed in TEM images of crosswalls of S. aureus wild-type cells ( Fig. S6C) (37,46). TEM images of TarO down and SmdA down showed that TarO down , as expected (37,46), lacked this dark, high-density layer, while it was still present in the SmdA down strain, suggesting that WTA was still produced (Fig. S6C). We also performed Fourier transform infrared spectroscopy (FTIR), which has been used before to detect differences in the composition of WTA due to variable glycosylation patterns (44,47). As expected, changes in the FTIR spectra were evident in the polysaccharide region (1200 cm 21 to 800 cm 21 ) for the TarO down strain compared to the control, with the most significant differences recorded for the peaks at 1076 cm 21 , 1048 cm 21 , 1033 cm 21 , and 1000 to 970 cm 21 , representing aand b-glycosidic bonds in WTA (47). However, no changes were observed between SmdA down and the control (Fig. S6D). Together, these results suggested that SmdA has no major effect on the WTA biosynthetic pathway.
SmdA is important in several stages of staphylococcal cell division. To identify potential protein interaction partners of SmdA, we next performed a protein pulldown experiment using GFP-trapping with a chromosomal smdA-m(sf)gfp fusion strain. Interestingly, the major staphylococcal autolysin Atl, as well as the bifunctional PBP2, were identified along with 12 other proteins (Table S1). The experiment was repeated in a strain with a plasmid-based expression of SmdA-m(sf)GFP, and this setup resulted in an extended list of proteins that were pulled down, probably due to the elevated expression of SmdA-m(sf)GFP. Selecting the proteins for which at least 10 unique peptides were detected and with a fold change of .2 compared to the control, resulting in a list of 57 proteins (Table S1). Several proteins with activity involved in protein folding, secretion, and/or degradation (e.g., FtsH, PrsA, SpsB, ClpB, ClpC, SecD) were identified in this assay, in addition to the penicillin-binding proteins PBP1, PBP2, PBP3, and the early division protein EzrA. All the 14 proteins identified in the initial experiment were also pulled down in the second experiment.
Bacterial two-hybrid analyses of the SmdA-PBP1-3 and SmdA-EzrA were performed to see whether these interactions could be reproduced in a heterologous system. The proteins were fused either N-or C-terminally to the domains of adenylate cyclase, an enzyme that catalyzes the production of cyclic AMP (cAMP) and eventually induction of b-galactosidase production when brought in proximity by the interaction between the target proteins. Indeed, SmdA interacted with PBP2, as well as PBP1, PBP3, and EzrA in the two-hybrid assays (Fig. 3A). By expressing a version of SmdA without its Nterminal membrane domain (SmdADTMH), we also showed that the observed interactions between SmdA and the PBPs in this assay occurred via the transmembrane segment, while the interaction with EzrA was retained for SmdADTMH. The latter suggests that EzrA interacted with the intracellular part of SmdA (Fig. 1A). Furthermore, we also showed that SmdA could self-interact, and this interaction was dependent on the transmembrane helix (Fig. S7).
The observed interactions between SmdA and PBPs may suggest that SmdA is somehow important for proper regulation and/or localization of PBPs in S. aureus ( Fig. 1 and 2). The VanFL approach used above (Fig. 1C) labeled non-crosslinked PG throughout the cell surface. Therefore, to determine more precisely the sites of active PG synthesis in S. aureus NCTC8325-4 SmdA down and SmdA overexpression cells, we pulse-labeled the cells for 90 sec with the fluorescent D-amino acid 7-hydroxycoumarin carbonyl amino-D-alanine (HADA), a molecule that is integrated into PG by the specific activity of transpeptidases (48). In the control cells, the expected midcell-localized band was observed (Fig. 3B). However, in the SmdA down strain, we observed a more diffuse HADA signal with PG synthesis often occurring at several sites within the cell, forming patterns visible as crosses or Y-shapes (Fig. 3C). Similar observations were made in the cells overexpressing SmdA, although with a lower frequency ( Fig. 3D and E). Thus, the site of active transpeptidation, and likely at least one of the PBPs, were mislocalized when the levels of SmdA were altered.
Moreover, we also studied whether SmdA could affect the localization of EzrA by knocking down smdA in a strain expressing a chromosomal ezrA-gfp fusion (Fig. 3F). Indeed, abnormal localization patterns for EzrA-GFP were observed with similar frequency as the abnormal HADA labeling (Fig. 3G). These results suggest that depletion of SmdA, directly or indirectly, influenced the localization of both early and late divisome proteins in S. aureus.
Furthermore, the major autolysin Atl was also pulled down with SmdA. Atl is a secreted multidomain enzyme, which is processed to an acetylmuramyl-L-alanine amidase and b-N-acetylglucosaminidase, involved in septal cross wall splitting resulting in daughter cell separation (19). SmdA down cells frequently displayed increased clustering (Fig. 2, Fig. S4). This indicated reduced cross wall splitting, a phenotype also observed in Datl mutants and WalKR depleted cells (22,49). WalKR is the two-component regulatory system controlling the expression of atl and other cell wall hydrolase encoding genes (22). To assess the reduced cross wall splitting phenotype in more detail, we performed Triton X-100-induced autolysis assays on the cultures. Indeed, reduced autolysis was observed in SmdA down cells, demonstrating reduced autolytic activity (Fig. 4A). The lysostaphin sensitivity was also reduced (Fig. 4B), suggesting alterations in the cell wall affecting the lytic properties of this enzyme. It should be noted that the resistance toward Triton X-100-and lysostaphin-induced autolysis in the SmdA down strain was reduced compared to the control strain, where walR and all the regulated autolysins were knocked down. It should also be noted that Triton-X-induced autolysis was not severely altered upon overexpression of SmdA (Fig. S5B).
SmdA localizes at the septal region after FtsZ. The SmdA-m(sf)GFP fusion protein displayed a septum-enriched signal when expressed ectopically from the low copy- number plasmid pLOW (Fig. 5A). To analyze the SmdA localization with native expression levels, a chromosomally integrated version of smdA-m(sf)gfp was made in which SmdA was expressed with m(sf)GFP fused to its C terminus. Localization analyses in these cells further confirmed the septum-enriched localization of the fusion protein ( Fig. 5A) with an average septum/periphery fluorescence signal ratio of 3.4 (n = 52, see Materials and Methods). To visualize the localization of SmdA relative to the cell division process, we used FtsZ as a marker. Expression of ftsZ-fusion genes in the smdA-m(sf)gfp chromosomal fusion strain, however, only resulted in strains with extremely poor growth, suggesting that the cells did not tolerate such double-labeling. Instead, we created a double-labeled strain in which SmdA-mYFP and FtsZ-mKate2 were coexpressed from plasmids, while the native smdA and ftsZ genes were still present on the chromosome. Structured illumination microscopy (SIM) analysis showed that SmdA-mYFP was localized around in the membrane when the Z-ring was formed (Fig. 5B) and that a septum-enriched localization occurred as the septal cross wall was being synthesized. This was further confirmed by stimulated emission depletion (STED) microscopy analysis (Fig. 5C). STED imaging also revealed that there was no apparent colocalization of SmdA-mYFP and FtsZ-mKate2 in newborn cells before FtsZ-constriction initiates. Combined, the localization of SmdA was reminiscent of the localization of PBPs in S. aureus (11) but appeared to localize to the septal area after the early divisome proteins such as EzrA (50,51).
The SmdA protein was anchored to the membrane by a single transmembrane helix (Fig. 1A). To verify its significance for the septal localization, we ectopically expressed a version of SmdA without the transmembrane helix (SmdADTMH). The N-terminally truncated SmdADTMH-m(sf)GFP localized to the cytoplasm of the cells (Fig. 5A). This showed that, while the transmembrane helix is critical for protein-protein interactions and subcellular localization of SmdA, the interaction observed between EzrA and SmdADTMH (and full-length SmdA) (Fig. 3A) is probably not involved in determining the localization of this protein.
During these latter experiments, we noticed that SmdADTMH-m(sf)GFP overexpression led to cells with obvious morphology defects. Overexpression of SmdADTMH (without the GFP-tag) (Fig. 6A) strikingly resulted in a more extreme phenotype than overexpression of full-length SmdA (Fig. 3D, Fig. S5). The cells were often inflated or with a bean-shaped appearance (Fig. 6A). HADA pulse-labeling of these cells revealed a highly abnormal PG incorporation pattern, with the fluorescent signal forming both condensed clumps and Y-shapes in different directions in almost 50% of the cells (Fig. 6A and B). This demonstrats that membrane attachment is critical for the localization and function of SmdA.
Conserved amino acids in the predicted NERD domain were critical for the function of SmdA. The functional importance of the NERD domain ( Fig. 1A) has, to our knowledge, neither been studied nor verified previously in any bacterial protein.
As mentioned, we observed that overexpression of SmdADTMH led to severe phenotypic changes in the staphylococcal morphology (Fig. 6A). We decided to use this as a tool to gain insight into whether the predicted NERD domain was important for the function of SmdA. Multiple sequence alignment of SmdA proteins from different staphylococcal species (Fig. S1) revealed conserved residues in this domain, and sitedirected mutagenesis was used to create two versions of SmdADTMH containing mutations in the NERD domain: one in which H145 was changed to Ala (mut1) and a second containing the mutations R150A and T151A (mut2). Furthermore, we also observed from the alignment that the C terminus of SmdA was highly conserved (Fig. S1), and another C-terminally mutated variant was created (mut3; F280A, H281A). The three-dimensional structure of SmdA, predicted by AlphaFold (52), suggested that the N-terminal part of the protein folded into a long a-helix and that the NERD domain was located within the C-terminal structured part of the protein (Fig. 6C [II]). The mutated SmdADTMH versions were overexpressed in a wild-type background. Expression of these mutated variants partially or fully abolished the phenotypic defects observed when overexpressing the nonmutated SmdADTMH version. HADA staining of mut1 (H145A) showed septum placement and PG synthesis comparable to wildtype cells, while mut2 (R150A, T151A) and mut3 (F280A, H281A) also reduced the functionality of the protein (Fig. 6A and B). Furthermore, the growth inhibition upon overexpression of SmdADTMH (Fig. 6D) was abolished when the mutated variants were expressed under the same conditions. This strongly suggests that the given residues in both the predicted NERD and the conserved C terminus are important for the function of SmdA.

DISCUSSION
The results presented here provide the first functional insights into the conserved staphylococcal protein SAOUHSC_01908, here named SmdA. SmdA, a protein conserved within the Staphylococcaceae family, is critical for maintaining cell morphology and for proper progression of cell division in different S. aureus strains (including MRSA and MSSA). Unbalanced levels, as well as loss of proper subcellular localization, of SmdA, resulted in severe cell morphology defects due to mislocalized cell division, uncontrolled cell wall synthesis, and lack of proper cross wall splitting.
We observed that knockdown of SmdA increased the sensitivity of S. aureus to several cell wall-targeting antibiotics, including b-lactams and tunicamycin (Table 1). Many of the frequently used antibiotics, including b-lactams, target cell wall synthesis, but the rise of both methicillin-resistant and vancomycin-resistant S. aureus (MRSA and VRSA) have made S. aureus infections more difficult to treat and novel anti-staphylococcal targets and strategies are needed. Targeting cell cycle proteins have been shown to resensitize resistant staphylococci toward existing antibiotics due to synergistic effects (53)(54)(55). For example, the inactivation of some proteins involved in WTA synthesis sensitizes MRSA to b-lactams (37,56). The same was also the case for a diversity of factors contributing to cell division and cell wall biogenesis, including FtsZ, FtsA, PBP4, and most proteins involved in the PG synthesis pathway (55). The secretion-associated proteins SecDF and the chaperones PrsA and HtrA1 have also been tightly linked to b-lactam sensitivity (57,58), and recently, inactivation of the autolytic cell wall amidase Sle1 (21), or the membrane proteins AuxA and AuxB (44), was shown to result in increased sensitivity to b-lactams. Mechanisms of resensitization in these cases vary and may be a result of a weakened cell wall and inactivation or mislocalization of PG synthesis, including the key resistance determinant PBP2A. The large number of factors affecting b-lactam sensitivity reflects the tight links between different processes involved in cell wall biosynthesis, and most likely SmdA affects several steps in the cell division process ( Fig. 1 and 3). The exact mechanism by which SmdA affects these processes remains to be determined. Our results suggested that cell division and PG synthesis, but not teichoic acid biosynthesis, was affected. This is supported by observations showing mislocalization of cell division and PG synthesis in SmdA down cells ( Fig. 1 and 3). Furthermore, SmdA down cells displayed hypersensitivity toward tunicamycin (targeting enzymes involved both in PG and WTA synthesis) but not toward targocil (targeting the WTA exporter only) (38). SmdA may influence the localization of cell division and cell wall synthesis by direct protein-protein interactions. This is supported by the SmdA-EzrA and SmdA-PBP1-3 interactions, combined with the abnormal localization of EzrA-GFP and HADA incorporation in SmdA down cells (Fig. 3). The division defects observed in SmdA down cells ( Fig. 1 and 3) are reminiscent of previous studies of ezrA null mutants, strains with point mutations in FtsZ, or cells exposed to Z-ring inhibitors (50,51,59,60). However, the contribution of the identified interactions to the observed phenotypes must be investigated further, particularly because the subcellular localization of SmdA-m(sf)GFP does not appear to overlap the Z-ring throughout the cell cycle. The SmdA localization, which depends on the transmembrane helix, is instead more similar to the reported localization of PBPs (Fig. 5) (11,50,51).
Lack of cross wall splitting is another prominent feature of the SmdA depleted cells ( Fig. 2 and 4). After septum formation, cells remain attached, forming clusters of misshaped cells. Indeed, the autolytic activity was reduced in the SmdA down cells (Fig. 4). The major autolysin Atl was also pulled down together with SmdA-m(sf)GFP as one of the major hits (Table S1). Atl is an extracellular processed protein whose amidase and glucosaminidase domains together process PG in the septum to allow splitting. It is possible that SmdA somehow influences the export or processing of Atl. In this context, it is interesting to note that proteases, foldases, and chaperones (FtsH, PrsA, SpsB, ClpC, ClpB, HtrA1) were also pulled down with SmdA (Table S1). It remains to be determined whether these interactions have any functional relevance. However, perturbation of such pathways may have major consequences for cell division and septal placement because they are important for the proper folding and secretion of key proteins involved in such processes. This has, for example, previously been demonstrated for the chaperone ClpX, which is critical for the coordination of autolysins and cell division pro- teins (29). Involvement in such a mechanism could also explain the pleiotropic phenotypes observed in SmdA down cells, affecting several stages in the cell division process. A part of SmdA displays limited similarity to the so-called nuclease-related domain (NERD). This domain, which is found in a broad range of bacterial species and some archaea and plants, was initially identified on a virulence-plasmid from Bacillus anthracis (33) and was later suggested to belong to a superfamily of phosphodiesterases (61). Most NERD proteins are single-domain proteins, but they are also occasionally found together with a kinase-, pseudokinase, or helicase domain (62). The actual function of the NERD, to our knowledge, has not been studied experimentally. The results here showed that conserved residues in the predicted NERD were important for the functionality of SmdA, thus demonstrating for the first time a functional role of this domain (Fig. 6). Although highly speculative at this point, it would be interesting if a cell division factor such as SmdA had DNA-interacting capabilities and thus forming a potential link between the membrane-associated cell division proteins and the nucleoid. It should also be noted that several of the most conserved residues of the NERD domain (33) are not found in SmdA, suggesting a functional diversity among proteins harboring homology to this domain and implying that the NERD of SmdA may have functions unrelated to the nucleoid.
It is also interesting to note that the effect of SmdA knockdown was different between strains. For example, SmdA down resulted in a more severe phenotype in HG001 compared to NCTC8325-4 and SH1000 (Fig. 1, Fig. S3 and S4), although these strains derive from the same parent (63). While HG001 contains prophages, NCTC8325-4 and SH1000 have been cured of these, and it could, for example, be speculated that the stress imposed by knockdown of SmdA somehow results in prophage induction in HG001 and a more severe phenotype. Understanding these and other strain differences could also help to further pinpoint the exact function of SmdA.
The staphylococcal genome encodes hundreds of essential proteins, which all represent potential target sites for antimicrobials. To be able to fully exploit the antibiotic target repertoire, it is critical to understand how essential proteins are involved and linked between different cell cycle processes. In this work, we identified and characterized SmdA as a novel factor essential for cell morphology and cell division in S. aureus. Based on the results presented here, future research should aim at pinpointing the molecular mechanism by which SmdA affects different stages of the cell division process. Finally, because SmdA depletion results in increased sensitivity to b-lactams, it may as such also be a possible future target for combatting b-lactam resistance by resensitizing MRSA to these antibiotics.

MATERIALS AND METHODS
Bacterial strains, growth conditions, and transformations. Escherichia coli IM08B (64), E. coli XL1-Blue, and E. coli BTH101 were grown in lysogeny broth (LB) at 30 to 37°C with shaking or on lysogeny agar (LA) plates at 30 to 37°C, with 100 mg/mL ampicillin and/or 50 mg/mL kanamycin for selection. S. aureus SH1000, NCTC8325-4, HG001, and COL were grown in brain heart infusion (BHI) broth or tryptic soy broth (TSB) at 37°C with shaking or on BHI plates at 37°C. For selection, 10 mg/mL of chloramphenicol, 5 mg/mL of erythromycin, 15 mg/mL neomycin or 100 mg/mL spectinomycin were added. For induction of gene expression, 50 or 300 mM isopropyl b-D-1-thiogalactopyranoside (IPTG) or 2.5 ng/mL anhydrotetracycline (ATc) were added to the bacterial cultures.
A standard heat shock protocol was used for the transformation of E. coli IM08B. Isolated plasmids from E. coli were transformed into S. aureus by electroporation. Preparation of electrocompetent S. aureus cells and electroporation were performed according to Löfblom et al. (65). Bacterial strains used in this study are listed in Table S2.
Genetic modifications. For all cloning, DNA fragments were amplified from S. aureus SH1000 genomic DNA, and cloning was performed with restriction digestion (New England Biolabs [NEB]) and subsequent ligation using T4 DNA Ligase (NEB) unless otherwise stated. Ligation mixtures were transformed into E. coli IM08B and sequence-verified plasmids were transformed into S. aureus. All strains and primers used in this study are listed in Table S2 and Table S3, respectively. Fluorescent fusion constructs. (i) Construction of pLOW-smdA-m(sf)gfp. Monomeric superfolder GFP, m(sf)gfp, was initially fused to ftsZ in the plasmid pLOW (35). m(sf)gfp was amplified from the plasmid pMK17 (66) using forward primer im1_linker_FP_F_BamHI (annealing to the linker sequence) and reverse primer im2_m(sf)gfp_R_NotI_EcoRI. The amplified fragment encodes a linker sequence N-terminally of the m(sf)gfp gene. The fragment was digested with BamHI and EcoRI and subsequently ligated into the corresponding sites of the plasmid pLOW-ftsZ-gfp to replace the original gfp-gene with the link-er_m(sf)gfp sequence. SAOUSHSC_01908 was then amplified with primers im77_SA1908_F_SalI_RBS and im78_SA1908_R_BamHI, the product was digested with SalI and BamHI and ligated into the corresponding sites of pLOW-ftsZ-m(sf)gfp to create pLOW-smdA-m(sf)gfp.
(ii) Construction of pMAD-smdA-m(sf)gfp_aad9 for chromosomal integration. To tag the chromosomal smdA, the plasmid pMAD-smdA-flag_aad9 was constructed initially. The insert in this plasmid (smdA-flag_aad9) was assembled by overlap extension PCR (primers im147-im152) with the flag-tag sequence inserted by primer overhangs and cloned into pMAD (34) using restriction enzymes MluI and BamHI and T4 DNA Ligase (NEB). The insert was designed so that the flag-tag could be removed using NotI and SpeI. The m(sf)gfp sequence was amplified from template pMK17 (66) using the primers im153 and im154. The PCR product was digested with NotI and SpeI and subsequently ligated with NotI-and SpeI-digested pMAD-smdA-flag_aad9. The ligation mix was transformed to E. coli IM08B, and the resulting plasmid, pMAD-smdA-m(sf)gfp_aad9, was verified by sequencing and transformed into electrocompetent S. aureus SH1000. Integration of smdA-m(sf)gfp in the native locus of smdA using the temperature-sensitive pMAD-system was carried out as previously described (34).
(iii) Construction of pHC-ftsZ-mKate2. ftsZ-mKate2 was amplified from the plasmid pLOW-ftsZ-mKate2. pLOW-ftsZ-mKate2 was made in a similar manner as described for pLOW-ftsZ-m(sf)gfp but with reverse primer im5_mKate_R_NotI_EcoRI and genomic DNA derived from strain MK119 (67) as the template for amplification of mKate2. ftsZ-mKate2 was then amplified with primers USHC109 and USHC148, generating a product with SalI and MluI as an overhang. The PCR product and the plasmid pSK9065 (68) were digested with SalI and MluI and thereafter ligated.
(iv) Construction of pLOW-smdA-mYFP. This plasmid was constructed by using pLOW-ftsZ-mYFP as starting point, made similarly as described for pLOW-ftsZ-m(sf)gfp but with reverse primer im3_cfp_myfp_R_NotI_EcoRI and plasmid pMK20 (lab collection) as the template for amplifying mYFP. The plasmids pLOW-smdA-m(sf)gfp (described above) and pLOW-ftsZ-mYFP were digested with SalI and BamHI and subsequently ligated, resulting in pLOW-smdA-mYFP.
CRISPRi constructs. (i) Construction of pCG248-sgRNA(smdA). For gene knockdowns, the CRISPR interference (CRISPRi) system developed by Stamsås et al. (69) was used. In this system, dcas9 is placed downstream of an IPTG-inducible promoter in the plasmid pLOW-dcas9. A second plasmid is carrying the single guide RNA (pCG248-sgRNA[x], where x represents the targeted gene), which constitutively expresses the sgRNA, including the 20 nt base-pairing regions specific for the gene to be knocked down. The pLOW-dcas9 plasmid contains an erythromycin resistance gene and the pCG248-sgRNA(x) plasmid is a chloramphenicol resistance gene. The gene-specific 20 nt sequences were replaced in the pCG248-sgRNA(x) plasmids using an inverse PCR approach as described earlier (69) using primers mk299 and mk323.
(ii) Construction of pLOW-dCas9_aad9. To make the CRISPRi system compatible with the MRSA strain COL (which is intrinsically erythromycin resistant), the erythromycin resistance gene ermC in pLOW-dcas9 was replaced with aad9, encoding a spectinomycin resistance cassette. The primers im183 and im184 were used to amplify the entire pLOW-dcas9 plasmid, except the ermC gene. The aad9 gene was amplified with primers im185 and im186 using pCN55 (70) as the template, where im185 contained the sequence of im183 as overhang, and im186 the sequence of im184 as an overhang. Thus, the two fragments had overlapping sequences and were fused using NEBuilder HiFi DNA Assembly (NEB). The construct was transformed to E. coli IM08B, and the plasmid was verified by PCR and sequencing. For CRISPRi in S. aureus COL, the pLOW-dcas9_aad9 plasmid was used along with pCG248-sgRNA(x).
Construction of plasmids used for overexpression studies. (i) Construction of pLOW-smdA and mutated versions. smdA was amplified from S. aureus genomic DNA using primers im77_SA1908_F_SalI_RBS and mk517_1908_R_NotI. The fragment was digested with SalI and NotI and ligated into the corresponding sites of plasmid pLOW-dcas9 (69) to produce the plasmid pLOW-smdA, with IPTG-inducible overexpression of smdA. pLOW-smdADTMH was constructed in a similar manner, except that primer, mk518_1908_F_RBS_SalI was used instead of im77 to remove the 29 N-terminal amino acids of SmdA, predicted to encode the TMH and extracellular part. Site-directed mutagenesis in the two plasmids was performed by a two-step overlap extension PCR approach, where the mutations were introduced in the primers. The primers mk519 and mk520 were used for introducing mutation H145A (mut1), mut2 (R150A, R151A) was made with primers mk521 and mk522, and mut3 (F280A, H281A) with primers mk529 and mk530.
Phase-contrast and fluorescence microscopy analysis. For induction of plasmid-encoded fluorescent fusions, exponentially growing cultures were diluted to OD 600 0.05 in BHI medium with 50 mM IPTG and incubated for 2 h before microscopy. For CRISPRi-knockdown and overexpression experiments, cultures of OD 600 = 0.4 were diluted 250-fold in medium with 300 mM IPTG and grown until OD 600 was approximately 0.4. For labeling of the cell wall, BODIPY TM FL vancomycin (VanFL) (Invitrogen) was added. To label newly synthesized peptidoglycan, HADA (van Nieuwenhze group, Indiana University) was added actively growing cultures (OD 600 approximately 0.4) at a final concentration of 250 mM. The cells were incubated with HADA at 37°C for 90 sec and then put on ice. The cultures were pelleted by centrifugation at 10 000 Â g at 4°C for 1 min, washed with cold 1 Â PBS, pH 7.4 and finally resuspended in 25 mL PBS. Cells were subsequently added to agarose pads and microscopy was performed on a Zeiss AxioObserver with ZEN Blue software. An ORCAFlash4.0 V2 Digital complementary metal-oxide-semiconductor (CMOS) camera (Hamamatsu Photonics) was used to capture images through a 100 Â PC objective. HPX 120 Illuminator (Zeiss) was used as a light source for fluorescence microscopy.
All microscopy analyses were repeated at least two times. Images were processed and prepared for publication using Fiji (71). For analysis of cell roundness, cell area, and fluorescent labeling patterns, MicrobeJ (72) was used to determine cell outlines. Outlines were corrected manually, when necessary. Cell area and roundness as a measure of morphology were calculated and plotted using MicrobeJ. After cell outline detection, categorization of cells into normal or abnormal labeling patterns was done manually using images from independent experiments. The septum/periphery fluorescence signal ratio was measured as described before (73) in cells with a full septal signal of SmdA-GFP. A fluoresce ratio between 2.5 and 3.5 is typical for septum-enriched proteins (11).
Structured illumination microscopy (SIM) and stimulated emission depletion (STED) microscopy. Super-resolution SIM imaging was performed using a Zeiss ELYRA PS.1 microscope equipped with a 100 Â 1.46 NA alpha plan apochromat oil immersion objective and a pco.edge sCMOS camera. Fluorescence images were acquired sequentially using 200 to 300 ms exposure times per image, for a total of 15 images per SIM reconstruction. All imaging was performed at room temperature (;23°C). Raw data were reconstructed using the SIM algorithms in ZEN 2011 SP7 software (black edition, Carl Zeiss). Brightfield images were captured using widefield imaging mode. Images had a final pixel size of 25 nm.
Gated STED (gSTED) images were acquired on a Leica TCS SP8 STED 3Â system, using an HC PL Apo 100 Â oil immersion objective with NA 1.40. Fluorophores were excited using a white excitation laser operated at 509 nm for mYFP and 563 nm for mKate2. A STED depletion laser line was operated at 592 nm and 775 nm for mYFP and mKate2, respectively, using a detection time delay of 0.8 to 1.6 ns for both fluorophores. The total depletion laser intensity was in the order of 20 to 40 MW cm 22 for all STED imaging. The final pixel size was 13 nm, and the scanning speed was 400 Hz. The pinhole size was set to 0.9 AU.
Images were processed and analyzed using Fiji (71). Line scans were analyzed using the Plot Profile function in Fiji, using a line width of 1.5. Fluorescence intensities were normalized to the highest value for each channel.
Scanning and transmission electron microscopy analysis. Overnight cultures were diluted to approximately OD 600 = 0.1 in BHI. When OD 600 reached 0.4, the cultures were diluted at 1:250. Antibiotics and IPTG were added when appropriate. The cultures (10 mL) were grown to OD 600 = 0.3 and 1 volume of fixation solution, containing 5% (wt/vol) glutaraldehyde and 4% (wt/vol) paraformaldehyde in 1ÂPBS, pH 7.4, was added. The tubes were carefully inverted a few times and incubated for 1 h at room temperature before being placed at 4°C overnight. The following day, the cultures were centrifuged at 5000 Â g, and the pellets were washed three times with PBS. Further preparations of samples to be analyzed with TEM were performed as described before (69).
Samples for SEM were, after washing with PBS, dehydrated with EtOH, essentially in the same manner as for sample preparations for TEM (69). The samples were subjected to critical point drying by exchanging the EtOH with CO 2 . The samples were then coated with a conductive layer of Au-Pd before being analyzed in a Zeiss EVO50 EP Scanning electron microscope. Images were analyzed and prepared using Fiji (71).
Growth assays. For examining growth on solid medium, overnight cultures were diluted 1:250 in BHI containing 300 mM IPTG, unless otherwise specified. When reaching the exponential-phase, OD 600 was adjusted to 0.3 for all samples. A 10-fold dilution series were made for all strains, and 2 mL of each dilution were spotted on BHI agar containing proper antibiotics and 300 mM IPTG. The plates were incubated at 37°C for approximately 16 h, and pictures of the plates were captured in a Gel Doc TM XR 1 Imager (Bio-Rad).
For measurement of growth in liquid cultures, cells were at OD 600 0.4, were diluted 1:250 in a medium containing 300 mM IPTG. Every hour for 5 h, OD 600 was measured spectrophotometrically using Genesys 30 (Thermo Scientific) and dilutions of the cultures were plated for CFU counting.
MIC assays. The experiments were set up in 96-well microtiter plates with a total volume of 300 mL. A 2-fold dilution series of the antibiotics were prepared in BHI containing selective antibiotics and IPTG when appropriate. The overnight cultures were diluted 1:1000 in BHI containing 300 mg/mL IPTG for induction. The cells were grown at 37°C, and the plate was shaken for 5 sec before measurements of OD 600 were taken every 10th min throughout the experiment, using either a Synergy TM H1 Hybrid Multi-Mode Reader (BioTek Instruments) or a Hidex Sense (Hidex Oy). The experiments were repeated at least two times with the same results.
Detection of lipoteichoic acid (LTA) by Western blotting. Detection of LTA was performed by Western blotting, and sample preparations were done according to descriptions found in Hesser et al. (26). Cells were grown in TSB medium. The samples were separated on a 4 to 20% gradient Mini Protean TGX acrylamide gel (Bio-Rad) and subsequently transferred to a polyvinylidene difluoride (PVDF) membrane by semidry electroblotting. The membrane was blocked for 1 h in 5% (wt/vol) skim milk in PBST and placed overnight at 4°C. Next, the membrane was incubated for 1 h with a-LTA (Hycult) 1:4000 in PBST, washed three times (10 min each) with PBST before incubation for 1 h with a-Mouse IgG HRP Conjugate (Promega) secondary antibody (1:10 000 in PBST). The membrane was again washed three times and LTA bands were visualized by using SuperSignal TM West Pico PLUS Chemiluminescent substrate (Thermo Fisher Scientific) in an Azure Imager c400 (Azure Biosystems).
The same procedure as in (44) was carried out for possible detection of LTA released to the medium. The supernatants, after harvesting cells during sample preparations, were kept. Supernatant samples were centrifuged for 16 000 Â g for 10 min, and 75 mL was mixed with 25 mL 4Â SDS-PAGE sample buffer. These samples were boiled for 30 min and applied on the 4 to 20% Mini Protean TGX acrylamide gel. Thereafter, the same immunodetection procedure as described above was followed.
Fourier-transform infrared spectroscopy (FTIR) analysis. Cultures of the strains IM313 (HG001, CRISPRi[control]), IM312 (HG001, CRISPRi[smdA]) and IM357 (HG001, CRISPRi[tarO]) were initially pregrown in BHI to exponential-phase, back diluted to OD 600 = 0.05 and induced with 300 mM IPTG. The bacterial cells (1 mL) were harvested at OD 600 = 0.4 by centrifugation at 5000 Â g, 4°C, for 3 min. The pelleted cells were kept at 220°C before further processing. Pellets were resuspended in 40 mL 0.1% (wt/ vol) NaCl, and 10 mL of the suspensions were added to an IR-light-transparent silicon 384-well microplate (Bruker Optic, Germany), with three technical replicates for each sample. The plates were left to dry at room temperature for approximately 2 h. FTIR spectra were recorded in transmission mode using a high-throughput screening extension (HTS-XT) unit coupled to a Vertex 70 FTIR spectrometer (Bruker Optik GmbH, Leipzig, Germany). Spectra were recorded in the region 4000 to 500 cm 21 , with a spectral resolution of 6 cm 21 , a digital spacing of 1.928 cm 21 , and an aperture of 5 mm. For each spectrum, 64 scans were averaged. The OPUS software (Bruker Optik GmbH, Leipzig, Germany) was used for data acquisition and instrument control. The obtained spectra were processed by taking second derivatives and extended multiplicative signal correction (EMSC) preprocessing in Unscrambler X version 11 (CAMO Analytics, Oslo, Norway). The results presented are averaged spectra from 3 biological replicates (each with 3 technical replicates) for the region with wavelengths between 1200 cm 21 and 800 cm 21 .
GFP-trap and liquid chromatography with tandem mass spectrometry (LC-MS/MS). Cultures (S. aureus SH1000 wild-type, IM308 SH1000 smdA-m(sf)gfp, IM104 SH1000 pLOW-smdA-m(sf)gfp and IM164 SH1000 pLOW-smdA-flag) were pregrown in BHI to exponential-phase, back diluted to OD 600 = 0.05 and induced if necessary. When reaching OD 600 at 0.4, 80 mL of each culture was harvested by centrifugation at 4000 Â g, 4°C for 3 min. Supernatants were decanted and pellets resuspended in cold TBS before transfer to 1.5 mL microcentrifuge tubes. Centrifugation was repeated for 1 min, and pelleted cells were stored at 280°C before further use.
For GFP-trap, cells were resuspended in 1 mL cold buffer containing 10 mM Tris pH 7.5, 150 mM NaCl, 1 mM PMSF, 6 mg/mL RNase and 6 mg/mL DNase. Suspensions were transferred to 2 mL lysing matrix B tubes (MP Biomedicals) containing 0.8 g # 106 mm glass beads (Sigma-Aldrich) and subjected for mechanical lysis by agitation in a FastPrep-24 TM (MP Biomedicals) for 3 Â 30 sec at 6.5 m/s, with 1 min pause on ice between the runs. Tubes were centrifuged at 5000 Â g, 4°C for 10 min, and supernatants were transferred to new tubes. Concentrations were determined by measuring Abs280 using NanoDrop TM 2000 (Thermo Fisher Scientific), where a small amount of the samples were added to a final concentration of 1% (wt/vol) SDS before measurements. GFP-Trap beads (25 mL per sample) (Chromotek) were washed three times with 500 mL ice-cold Dilution/Wash buffer (10 mM Tris pH 7.5, 150 mM NaCl, 0.5 mM EDTA) and centrifuged at 2500 Â g, 4°C for 5 min. Lysates were diluted in Dilution/Wash buffer to a final concentration of 1 mg in a total volume of 500 mL, before being transferred to GFP-Trap beads. Samples were placed in a Bio RS-24 Multi rotor (Biosan) at 4°C for 1 h. The samples were then centrifuged at 2500 Â g, 4°C for 5 min, the supernatant removed, and beads washed three times with Dilution/Wash buffer. During the last washing step, solutions were transferred to new tubes, and after centrifugation and removal of the supernatant, beads were resuspended in 50 mL 5% (wt/vol) SDS, 50 mM Tris pH 7.6. Tubes were incubated at 95°C for 5 min and centrifuged at maximum speed for 30 sec. After standing at the bench for a few minutes,30 to 50 mL were transferred into new tubes. Samples were kept at 220°C and heated for 2 min at 95°C before the sample preparation method Suspension trapping (STrap), conducted as described by Zougman et al. (75).
The peptide samples were analyzed by coupling a nano UPLC (nanoElute, Bruker) to a trapped ion mobility spectrometry/quadrupole time of flight mass spectrometer (timsTOF Pro, Bruker). The peptides were separated by an Aurora Series 1.6 mm C18 reverse-phase 25 cm Â 75 mm analytical column with nanoZero and CaptiveSpray Insert (IonOpticks, Australia). The flow rate was set to 400 nL/min and the peptides were separated using a gradient from 2% to 95% acetonitrile solution (in 0.1% [vol/vol] formic acid) over 120 min. The timsTOF Pro was run in positive ion data-dependent acquisition PASEF mode, with a mass range at 100 to 1700 m/z. The acquired spectra were analyzed against an S. aureus NCTC8325 proteome database.
Bacterial two-hybrid (BACTH) assays. Plasmid construction, and procedure for the BACTH assays, were conducted in the same manner as previously described (69), and primers used are listed in Table S3. Briefly, gene fusions of selected genes, to the T18 or T25 domains of adenylate cyclase form Bordetella pertussis, were made by restriction cutting and ligation in the plasmid vectors pKT25, pKNT25, pUT18, or pUT18C (Euromedex). E. coli XL1-Blue cells were used for transformation, and plasmids were verified by sequencing before BACTH assays (76) were set up according to the manufacturer (Euromedex). Cotransformation of plasmids containing fusion-genes of opposite domains, that is T25 in one plasmid and T18 in the other were done in E. coli BTH101 with 50 mg/mL kanamycin and 100 mg/ mL ampicillin as selection markers. Five random colonies were picked, grown in liquid LB to visible growth, and spotted on LA plates containing 40 mg/mL X-gal and 0.5 mM IPTG, in addition to the selection markers. Plates were incubated dark at 30°C for 20 to 48 h before being inspected, and blue colonies are an indication of positive interaction between tested genes. Presented results are representative of at least six independent replicates.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.