Abstract
Inhibiting cell division under DNA-damaging conditions is critical for cell viability. In bacterial cells, DNA damage induces the SOS response, which inhibits cell division to prevent compounding damage while repairs are made. In the rod-shaped bacterium, Escherichia coli, the SOS-controlled inhibitor SulA blocks division at an early step, but in coccoid bacteria, such as the human pathogen Staphylococcus aureus, the link between DNA damage and cell-division inhibition remains poorly understood. To enhance our knowledge and provide unique pathways for therapeutic intervention, we studied this process in S. aureus and have identified and characterized an SOS-induced cell-division inhibitor, designated SosA. In contrast to cells lacking sosA, wildtype cells increase in size upon DNA damage due to cell-division inhibition and display a concomitant superior viability. This 77-amino-acid protein is conserved in S. aureus, with homologs being present among staphylococci, and we have provided evidence to support its predicted membrane localization. Functional studies showed that it does not interfere with early cell-division events, such as Z-ring formation, but rather interacts with known divisome components, particularly PBP1, at a later stage. Characterization of truncated SosA variants and mutant strains revealed that the extracellular C-terminus of SosA serves dual functions as a required component for cell-division inhibition while simultaneously controlling SosA levels by acting as a likely target for degradation by the uncharacterized membrane protease CtpA. Our findings provide important insights into the regulation of cell division in coccoid bacteria that in turn may foster development of new classes of antibiotics targeting this essential process.
Significance Staphylococcus aureus is a clinically relevant human pathogen and model organism for cell-division studies in spherical bacteria. We show that SosA is the DNA-damage-inducible cell-division inhibitor in S. aureus that upon expression causes cell swelling and cessation of the cell cycle at a stage prior to division plate completion. SosA appears to function by interfering with the essential membrane-associated division protein PBP1, while being negatively regulated by the membrane protease CtpA. This report represents the first comprehensive description of an endogenous cell-division inhibition mechanism in coccoid bacteria. Uncovering the molecular details of SosA function and regulation can lead to target identification for development of valuable new antibacterials against staphylococci.
Introduction
Bacteria multiply by coordinated and essential DNA replication and cell-division events, two important biological processes that are valuable targets for antimicrobial therapy (1-4). In the event of DNA damage, the SOS response is activated and ensures that cell division is delayed until the DNA is repaired. The SOS regulon is controlled by the conserved LexA repressor, which is inactivated in response to RecA, a sensor of DNA damage at stalled replication forks, binding to single-stranded DNA (5-7). The SOS response has been mostly studied in Escherichia coli, a gram-negative rod-shaped bacterium. Among the LexA-regulated genes in E. coli is sulA, which encodes a cell-division inhibitor. This inhibitor suppresses division by binding to FtsZ, which acts as a scaffold for the assembly of cell-division components at the division site (4, 8-13). In rod-shaped bacteria, inhibiting cell division leads to filamentation as a consequence of lateral peptidoglycan synthesis. This phenotype is also observed in E. coli mutants lacking the Lon protease for which SulA is a substrate, further substantiating the role of SulA in cell-division inhibition and filamentation (14, 15).
SulA is poorly conserved in species outside Enterobacteriaceae and the α-proteobacterium Caulobacter crescentus encodes another SOS-induced cell-division inhibitor SidA that does not show homology to SulA. In contrast to cytosolic SulA, SidA is a membrane-anchored, small protein, which does not interact with FtsZ but rather with later cell-division proteins (16). Hence, it appears that fundamental differences exist as to how different bacteria orchestrate regulated cell-division inhibition. For gram-positive, spherical cells such as Staphylococcus aureus, however, little is known of how SOS induction is coupled to cell division nor have specific posttranslational mechanisms for the negative regulation of cell-division inhibition been identified in gram-positive species at present.
S. aureus is a serious gram-positive human pathogen, notorious for being implicated in a wide range of infections and for being able to acquire resistance towards important antibiotic classes. It originally received its name from the grape-like clusters of coccoid cells that result from the unique cell-division process that occurs in three consecutive orthogonal planes (17-19). In this organism, we and others have previously identified lexA and noted an open reading frame (designated sosA) that was divergently transcribed from lexA and controlled by the LexA repressor and the SOS response (20-23). The location of sosA adjacent to lexA indicated that it might encode a cell-division inhibitor, as similar gene synteny has been observed for SOS-induced, cell-division inhibitors encoded by gram-positive, rod-shaped bacteria, namely Bacillus subtilis (24), Bacillus megaterium (25), Listeria monocytogenes (26, 27), Mycobacterium tuberculosis (28), and Corynebacterium glutamicum (29) (Fig. 1A). The cell-division inhibitors of these organisms display no similarity to SulA, and within the group they generally show little homology with the exceptions of YneA from L. monocytogenes and ChiZ from M. tuberculosis (28) that resemble B. subtilis YneA. Commonly, however, they have a single membrane spanning segment, a predicted extracellular C-terminus and, except for DivS from C. glutamicum, a LysM domain, which is a ubiquitous peptidoglycan binding motif (30, 31), suggesting that membrane and/or cell-wall localization is a common feature characterizing gram-positive cell-division inhibitors.
The fundamental processes of staphylococcal cell division have been studied spatiotemporally using super-resolution microscopy techniques (18, 32), and recent efforts combining genetic approaches have unveiled the molecular mechanism in unprecedented detail (33, 34). Here, we demonstrate that sosA encodes the SOS-inducible cell-division inhibitor in S. aureus and document its impact on cell division following treatment with DNA damaging agents. Moreover, we identify a possible mechanism for the proteolytic control of endogenous cell-division inhibition. Thus, we further our insight into these basic biological phenomena, which could lead to the development of new antimicrobials targeting the cell-division machinery in S. aureus.
Results
Conservation of sosA in staphylococci
In S. aureus, the open reading frame located adjacent to lexA (divergently transcribed) was named sosA and hypothesized to encode an inhibitor of cell division (Fig. 1A) (22). To determine the role of this gene in staphylococci and identify important characteristics of the protein to investigate further, homology studies were conducted comparing that of S. aureus to a number of other species. The 77-amino-acid-long product has homology and 25-60% amino-acid identity to proteins encoded by genes occupying the same chromosomal location in a range of staphylococci (Fig. 1B and SI Appendix, Fig. S1), while showing no clear homology to the known gram-positive cell-division inhibitors. Of note, the C-terminal part of SosA lacks a LysM domain, which is present in the cell-division inhibitors YneA and ChiZ (35, 36). The N-terminal half of SosA includes a putative transmembrane (TM) domain, and the C-terminal part is predicted to be located extracellularly (TOPCONS server [37]). Among the different staphylococcal species, a highly conserved sequence is located just C-terminally of the predicted TM domain and, although less striking, some conservation also seems to exist in the extreme C-terminus (Fig. 1B).
Genotoxic stress inhibits cell division in an SosA-dependent manner
To analyze whether SosA serves as an SOS-induced cell-division inhibitor, we deleted sosA from 8325-4, the common laboratory S. aureus strain, and JE2, a derivative of the clinically relevant USA300 lineage. Growth and survival of these mutants was monitored upon challenge with a lethal concentration of the DNA damaging agent Mitomycin C (MMC). Within a two-hour-long MMC killing experiment, wildtype cells with a functioning SosA protein showed a 10-100-fold higher survival rate than the mutant cells lacking SosA at each time point, whereas the optical densities of both cell type cultures were comparable (Fig. 2A). Flow cytometry measurements of arbitrary cell size distributions revealed that treatment with MMC led to a three- and six-fold increase in forward light scatter (FSC-A) values in 8325-4 and JE2, respectively, while the cell size distribution of the mutants lacking sosA remained essentially constant (Fig. 2B). The constant cell size of the mutants lacking sosA during DNA damage could imply that these cells, in contrast to wildtype cells, continue dividing. Indeed, this striking phenotype was confirmed by time-lapse microscopy, where wildtype cells increased in size over time, occasionally growing so large that lysis occurred, while the size of the mutant cells lacking sosA stayed the same and cell division continued (Fig. 2C and Movie S1). The phenotypes of the 8325-4 sosA mutant were abrogated in a strain complemented with a chromosomal copy of sosA expressed from its native promoter (SI Appendix, Fig. S2). Thus, the SosA protein seems to prevent or delay cell division, maintaining viability of S. aureus under DNA damaging conditions, and as a consequence of continued peptidoglycan synthesis causing cell sizes to increase.
SosA inhibits cell division
If SosA is indeed a cell-division inhibitor in S. aureus, it can be predicted that overexpression of the protein should be sufficient to inhibit growth under DNA non-damaging conditions. Hence, we expressed sosA episomally from an anhydrotetracycline (AHT)-inducible promoter and found that while the control vector does not affect cell viability in either the presence or absence of the inducer, the plating efficiency of cells with the plasmid for sosA overexpression was markedly compromised under inducing conditions (Fig. 3A). When grown in liquid culture and compared to the control, there was a 2-3-fold increase in forward light scatter values, showing that the size of the sosA-expressing cells increased (Fig. 3B). Microscopy observations over time confirmed the slight increase in cell size and the reduction in the overall number of cells upon SosA overproduction (Fig. 3C and Movie S2). Interestingly, contrary to what was observed upon sustained DNA-damaging conditions (Fig. 2B), we found that ectopic induction of SosA caused only a transient increase in cell size (compare time points 45 and 90 min to 135 min in Fig. 3B), indicating the existence of a negative regulatory mechanism.
SosA is a late-stage cell-division inhibitor
To pinpoint at which step of the cell-division process the SosA-mediated inhibition is taking place, we employed fluorescence microscopy to monitor the cellular localization of FtsZ and EzrA, crucial cell-division proteins in S. aureus, upon the induction of SosA overproduction. Neither the sosA overexpression plasmid in non-inducing conditions (Fig. 4A and B) nor the control vector in non-inducing and inducing conditions (SI Appendix, Fig. S3A and B) changed the cell morphology and protein localization. Contrasting with the cell-division inhibitor SulA in E. coli, sosA overexpression caused cell swelling but did not affect the localization of the cell-division initiator FtsZ (Fig. 4A) or the placement of the membrane-bound division protein EzrA (Fig. 4B), which is an early recruited protein connecting the cytoplasmic division components with the peptidoglycan-synthesizing membrane-bound complex (38).
To further explore the cellular consequences of SosA-mediated cell-division inhibition, we examined the phenotypes of JE2/pSosA cells that were labeled for five minutes with HADA, marking regions of nascent peptidoglycan synthesis, and grown in the absence or presence of inducer (SI Appendix, Fig S3C). While in the absence of AHT, 47% and 23% of cells, respectively, showed a ring (incomplete septum) or line (complete septum) of nascent peptidoglycan synthesis, there was accumulation (89%) of cells with incomplete septa and a severe drop in cells with complete septa (1%) when sosA was overexpressed (Fig. 4C). Additionally, 1% of cells had an ‘hourglass’ shape after incubation with AHT, indicative of non-productive, premature splitting taking place prior to septum completion. Next, whole cell walls of those cells that overproduced SosA were labeled with a fluorescent NHS ester and examined by 3D-structured illumination microscopy (3D-SIM). Microscopy visualization revealed that most cells had only signs of septation — a ‘piecrust’ (18) (SI Appendix, Fig. S3Ei) — and showed that in an hourglass-like cell, there was a gap in the septal peptidoglycan (SI Appendix, Fig. S3Eii). The fact that SosA does not affect localization of early cell-division proteins, FtsZ and EzrA, and the cell population becomes synchronized to a particular stage of the cell cycle (phase P2 [32]) suggests that SosA does not hinder the initiation of septum formation but blocks the progression of septum completion — a phenotype similar to S. aureus DivIB-depleted cells (39). Although DivIB has been shown to be dispensable for FtsZ and EzrA localization to the septum and piecrust formation, its absence results in the inhibition of septum progression and completion and in the mislocalization of GpsB (39), a late cell-division protein (39, 40). Again, under non-inducing conditions, SosA did not alter cell morphology and GpsB-GFP localization (Fig. 4D), and the control vector did not have any effect on the presence and absence of AHT (SI Appendix, Fig. S3D). Interestingly, while causing visible cell enlargement in this reporter strain, SosA overproduction also did not impair the recruitment of GpsB to the mid-cell (Fig. 4D), indicating that SosA acts on (a) later cell-division component(s) than GpsB.
SosA interacts with PBP1
No direct interaction has been observed between the gram-positive cell-division inhibitors and FtsZ, the scaffold for cell-division components assembly at the division site. Though poorly characterized mechanistically, the gram-positive inhibitors may suppress cell division at later stages, potentially by interacting with FtsI/FtsQ (for M. tuberculosis ChiZ)(35) or via delocalization of FtsL and/or DivIC (for B. subtilis YneA) (41). By employing a bacterial two-hybrid system approach that reports protein-protein interactions, we sought to identify possible in vivo interaction candidates for SosA based on cell-division proteins from S. aureus (38). This analysis indicated that PBP1 (penicillin-binding protein 1) and possibly also DivIC and PBP3 are likely targets for the activity of SosA (Fig. 5A). PBP1, which is a homolog of E. coli FtsI (42), is an essential component of the divisome that, in S. aureus, is considered to play a role in septum formation and cell separation, with only the latter activity relying on transpeptidase (TP) functionality of the enzyme (43). If PBP1 is indeed a target for SosA, it seemed possible that ectopic expression of pbpA (encoding PBP1) would mitigate the effect of concomitant sosA overexpression by increasing the level of target protein. Intriguingly, overproduction of PBP1 along with SosA manifested as a synergistically lethal phenotype that lead to considerable growth impairment even at low inducer concentrations, whereas the effects of SosA or PBP1 overproduction alone were negligible (Fig. 5B). While this does not suggest a simple titration of SosA by PBP1, wherein excess PBP1 is able to alleviate the inhibitor activity of SosA, it does support the evidence from the two-hybrid system that SosA is mechanistically linked to PBP1. To further explore this, we found that SosA was also synergistically lethal when co-expressed with a mutant variant of PBP1 lacking TP activity (SI Appendix, Fig. S4). This indicated that SosA works to obstruct septum formation progression and closure in S. aureus through an interaction with PBP1 and with a mechanism independent of the PBP1 TP activity.
The C-terminal part of SosA is functionally essential and possibly required for autoregulation
The C-terminal part of SosA contains segments that appear highly conserved in staphylococci (Fig. 1B), and it was predicted in silico to be extracellular. First, we confirmed its extracellular location by constructing PhoA-LacZ reporter fusions (44), whereby a fusion of the reporter chimera to the C-terminus of SosA conferred a phosphatase-positive, β-galatosidase-negative phenotype in E. coli, indicative of translocation across the membrane (SI Appendix, Fig. S5). Next, aiming at a functional characterization of its role in SosA activity, we constructed a series of C-terminally truncated variants by the successive elimination of ten amino acids (AAs) (Fig. 6A), resulting in four variants of SosA lacking between 10 and 40 of the C-terminal AAs. Strikingly, we observed that truncation of the 10 C-terminal AAs (and to some extent also removal of 20 C-terminal AAs) generated a hyper-potent inhibitor that gave rise to a severely compromised plating efficiency even at low inducer concentrations (Fig. 6B). Time-lapse microscopy revealed that cells producing SosAd10 (SosA missing the last 10 AAs) had a dramatic phenotype, with prominent cell swelling and a reduction of the cell-cycle rate (Fig. 6C and Movie S3). The phenotypes from the plating assay correlated with the cell size measurements by flow cytometry (SI Appendix, Fig. S6), which indicated that cells expressing the 10 AA truncated protein were stalled in the non-dividing state that resulted in abnormal cell size, whereas those expressing full-length SosA were more transiently halted or delayed in cell division (SI Appendix, Fig. S6 and Movie S3). In contrast, deletion of 30 or 40 AAs from the C-terminal part eliminated the ability of SosA to inhibit cell division, as SosAd30 and SosAd40 gave comparable plating efficiency results to the control (Fig. 6B). This prompted us to investigate the relevance of the highly conserved residues located in the membrane-proximal part of the SosA C-terminal part by performing alanine substitutions within the protein. To this end, we used the hyper-active SosAd10 variant as a scaffold and found that while point mutations at AA positions 37 or 38 (E and Q) had no effect, point mutations at positions 40 or 41 (Y and E) partially inactivated and the mutations at positions 44 and 45 (D and H) completely abolished the cell-division inhibitory activity of the protein (SI Appendix, Fig. S7). Importantly, the inactive variants SosAd40 and SosAd10(44A) are presumably localized to the membrane, as are SosA and SosAd10, when evaluated by the PhoA-LacZ translocation assay (SI Appendix, Fig. S5). We conclude that conserved residues in the membrane-proximal segment of the extracellular C-terminal part of SosA are essential for inhibiting cell division, while truncation at the extreme C-terminus of the protein augments activity.
SosA is regulated by the CtpA protease
We reasoned that one explanation for the severe impact of the 10-AA-truncated SosA on cell division could be that the extreme C-terminus of SosA serves as an endogenous signal for proteolytic degradation, and its removal allowed the mutant to escape degradation and accumulate. We therefore searched the Nebraska Transposon Mutant Library (45) for S. aureus protease mutants, specifically focusing on proteases considered to be membrane-localized. As a potential candidate, we identified a mutant in the carboxyl-terminal protease A (ctpA). The S. aureus CtpA protein is located at the cell-membrane/cell-wall fraction and is involved in stress tolerance and virulence in a mouse model of infection (46). Experiments involving the overproduction of the wildtype SosA protein in a ctpA mutant background (inactivated in ctpA) completely abolished cell viability, similarly to what was observed when the 10-AA-truncated SosA protein was overproduced in wildtype S. aureus cells (Fig. 7A). In agreement with this result, we found that the co-overproduction of SosA and CtpA in the ctpA mutant background alleviated the detrimental effect of SosA and that the degree of complementation depended on the relative expressional levels of the two proteins, such that at higher expression level of SosA only partial complementation was observed by concomitant CtpA expression (SI Appendix, Fig. S8). Furthermore, substantiating the negative regulatory role played by CtpA with respect to SosA, we noted that the ctpA mutant was moderately more sensitive to MMC (measured as a minimum inhibitory concentration) than wildtype S. aureus, but only when also encoding a functional sosA gene (Fig. 7B). Additionally, upon MMC-induced DNA damage, higher levels of the SosA protein accumulated in the ctpA mutant when compared to the wildtype (Fig. 7C). Altogether, based on these results, we propose that CtpA proteolytic activity acting on the extracellular C-terminus of SosA relieves SosA-mediated cell-division inhibition and allows growth to resume once DNA stress has ceased.
Discussion
Based on the results reported here, we propose the following model for regulation of cell division in the prolate spheroid bacterium S. aureus under DNA-damaging conditions (Fig. 8). During normal growth, the expression of sosA is repressed by the strong LexA repressor, but LexA is inactivated through autocleavage under SOS conditions, which results in the expression of sosA. It appears that the N-terminal autocleavage product of LexA needs to be further proteolytically processed for sosA expression to be fully induced (22), thus adding another layer of expressional control, suggesting tight regulation of the inhibitory protein. Upon derepression of LexA-regulated genes, SosA accumulates to levels that inhibit cell division without causing FtsZ delocalization nor preventing the assembly of downstream components, EzrA and GpsB, which are representatives of early and late cell division. The inhibitor appears to affect the progression and completion of septum formation via an interaction with PBP1. As a consequence of SosA activity, the cells appear synchronized at the stage of septum formation initiation, while an ongoing off-septal peptidoglycan synthesis leads to cell enlargement (SI Appendix, Fig. S3C). Once conditions are favorable for continued division, sosA expression is repressed, the membrane protease CtpA is involved in the removal or inactivation of SosA activity, and cell division resumes (Fig. 8).
The model illuminates several remaining unanswered questions. It will be particularly important to determine exactly how SosA affects the function of PBP1. In this context, it should be noted that our data fit well with the finding that the genetic depletion of PBP1 causes cell enlargement and incomplete septum formation (42). At present we cannot readily explain the peculiar finding that SosA and PBP1 seem to act in a synergistically lethal manner (Fig. 5B), though it is possible that the two proteins together constitute a signal to halt septum formation beyond a certain point.
Alternatively, the synergistic effects could depend on stoichiometry and physical accessibility in the sense that PBP1 might engage in a non-productive complex with SosA, more efficiently preventing residual and functional PBP1 access to the site. We keep in mind that SosA may also interact with PBP3 and DivIC (Fig. 5A), and these relationships should also be investigated. Interactions between the S. aureus divisome components constitute a complex web, and in the bacterial two-hybrid study, PBP1 interacted with DivIC, which in turn interacted with PBP3 (38). Thus, the inhibitory activity of SosA could be multifactorial.
An analysis of the AA sequence of SosA indicated that the N-terminal part contained a TM domain while the C-terminal was located extracellularly. The hypothesis that the TM domain promotes localization to the cellular membrane was supported by PhoA-LacZ fusion to SosA and by various point and deletion mutants of the fusion protein (SI Appendix, Fig. S5). The abrogation of activity by the 40-AA-truncation as well as by a single alanine substitution (D44A) suggests that this segment is essential for correct localization to the septum or, perhaps more interestingly, that the actual motif for cell-division inhibition is localized at the exterior of the membrane. If PBP1 interference indeed lies within this segment of SosA, it could provide a novel paradigm for interference with penicillin-binding proteins that could supplement the transpeptidase-inhibitory activity exerted by β-lactam antibiotics.
Endogenous cell-division inhibition must be stringently regulated, as demonstrated by pioneering work showing that SulA of E. coli is degraded by the Lon protease (15). Similarly, YneA from B. subtilis was reported to be regulated by proteolysis (36), while in B. megaterium, the temporal expression of a YneA homologue has been suggested to be due to mRNA instability (25). We demonstrated that a minor truncation of the C-terminus of SosA yields a hyper-potent cell-division inhibitor (Fig. 6), which led us to speculate that the extreme C-terminus serves as a signal for proteolysis. We identified a CtpA protease-deficient strain in which SosA appeared to accumulate and had an increased detrimental effect compared to the wildtype background. Bacterial carboxy-terminal proteases are poorly characterized, particularly with respect to their substrate preferences. Due to this protein’s involvement in stress tolerance and virulence in S. aureus (46) and its predicted carboxy-terminal activity, we tested its effect in conjunction with SosA and propose that CtpA specifically regulates SOS-induced cell-division inhibition in staphylococci, likely via recognition of the C-terminus of SosA. Biochemical evidence for such a direct activity will be needed, and at present we cannot exclude an indirect effect. Interestingly, a point mutation in B. subtilis YneA at its extreme C-terminus generated a stabilized yet functional variant of the cell-division inhibitor (36), suggesting that YneA and SosA may be degraded by similar mechanisms. Of note, carboxy-terminal proteases are ubiquitous in bacterial genera, and homologs of CtpA are found in other gram-positive species, such as B. subtilis and L. monocytogenes. Hence, the CtpA-like proteases could represent functional homologs of the E. coli Lon protease, allowing negative regulation of endogenous cell-division inhibition more broadly amongst gram-positives.
SosA may also prove to be a useful tool for scientists conducting studies into bacterial cell division, which have recently benefitted from the development of genetic tools and highly sophisticated microscopic imaging technologies. However, most often bacteria grow as an unsynchronized population with individual cells in different phases of cell division, making it challenging to order the temporal events that occur during cell-cycle progression, DNA replication, and division. Essentially, SosA is involved in halting the S. aureus cell cycle at a specific morphological stage, e.g. following the piecrust formation, and we anticipate that the exploitation of this function may constitute a useful tool for the generation of synchronized staphylococcal cells for experimental purposes.
Apart from its basic biological interest, the bacterial cell-division process may constitute an unexhausted source of novel therapeutic targets (1,2). We believe that SosA could be a natural scaffold, and the study of this protein can provide new therapeutically accessible targets that would interfere with an essential process in S. aureus and other staphylococci.
Materials and Methods
Bacterial strains and plasmids used in this work
The bacterial strains and plasmids used in this study are listed in SI Appendix, Table S1 and Table S2, respectively. E. coli strains were grown in Luria-Bertani medium (Oxoid) or on LB agar (Oxoid). S. aureus strains were grown in Tryptic Soy Broth (TSB) or on Tryptic Soy Agar (TSA) (both Oxoid). Ampicillin (100 µg/ml), chloramphenicol (10 µg/ml), or erythromycin (5 µg/ml) were added when appropriate. Anhydrotetracycline (AHT) (Sigma-Aldrich) and Isopropyl β-D-1-thiogalactopyranoside (IPTG) (Thermo Scientific) were used as inducers for protein expression.
Construction of strains and plasmids
All oligonucleotides used are listed in SI Appendix, Table S3. S. aureus 8325-4ΔsosA is a clean deletion of sosA (SAOUHSC_01334) in strain 8325-4 obtained by allelic replacement using the temperature-sensitive shuttle vector pIMAY (47). 1 kb regions up- and downstream of sosA were PCR amplified (Phusion Hot Start II DNA Polymerase, Thermo Scientific) using primer pairs Up-sosA_fw-KpnI/Up-sosA_rev and Dw-sosA_fw/Dw-sosA_rev-SacI, respectively, and subsequently joined in a spliced overhang PCR using Up-sosA_fw-KpnI/Dw-sosA_rev-SacI. The resulting deletion fragment was cloned into pIMAY via KpnI/SacI, generating pIMAY-ΔsosA, purified from E. coli DC10B and transformed into S. aureus subsequently maintained at 28°C. Chromosomal integration of the plasmid was performed at 37°C under selective conditions (chloramphenicol) followed by passage at 28°C without antibiotic selection and final plating on TSA containing 500 ng/ml of AHT for plasmid counterselection. Colonies were replica-plated to select for sensitivity towards chloramphenicol and successful allelic exchange were screened for by PCR amplification using primer pairs Ctrl_dsosA_F/Ctrl_dsosA_R positioned outside sosA and Fwd_MCS/Rev_MCS targeting the vector, respectively. The allelic replacement procedure was identical for S. aureus JE2 to create JE2ΔsosA.
The chromosomal transposon insertion mutation in ctpA conferring erythromycin resistance was obtained from the Nebraska Transposon Mutant Library (NTML, NE847) (45) and was moved by transduction (phage Φ11) to S. aureus JE2, JE2ΔsosA, S. aureus 8325-4 and 8325-4ΔsosA, resulting in JE2-ctpA, JE2ΔsosA-ctpA, 8325-4-ctpA and 8325-4ΔsosA-ctpA, respectively. JE2-ctpA(−erm), an erythromycin-sensitive derivative, was generated by elimination of the transposon-encoded ermB in JE2-ctpA by allelic exchange using the pTnT vector using temperature-mediated chromosomal integration and secY-mediated counterselection as described before (48).
All plasmid constructs mentioned below were cloned in E. coli IM08B (49) from where they were transformed into S. aureus strains. For complementation of the sosA knockout, the sosA gene with its native promoter was PCR amplified from strain 8325-4 using primers Up-sosA-promo_SalI and Dw-sosA_EcoRI and cloned into designated restriction sites in plasmid pJC1112 (50). The resulting plasmid (pJC1112-sosA) was transformed into the integration-proficient strain RN9011, creating RN9011-sosA-compl. when selected with erythromycin. The chromosomally integrated plasmid was transduced into 8325-4ΔsosA, resulting in 8325-4ΔsosA-compl. For sosA expression the sosA gene including its predicted ribosomal binding site was cloned into the BglII and EcoRI sites of pRAB12-lacZ (51) using primers Up-sosA_BglII/Dw-sosA_EcoRI, at the same time eliminating the lacZ reporter gene. The resulting plasmid was called pSosA. Generation of plasmids encoding C-terminally truncated variants of SosA was obtained by PCR amplification of DNA fragments using Up-sosA_BglII and downstream primers Dw-sosA(d10)_EcoRI to Dw-sosA(d40)_EcoRI, all equipped with premature stop codons. The PCR products were ligated into pRAB12-lacZ using BglII and EcoRI cut sites to create pSosAd10, pSosAd20, pSosAd30 and pSosAd40. Single and double amino acid substitutions (alanine) in the SosAd10 protein were obtained by cloning commercially synthesized DNA fragments (Twist Bioscience) into pRAB12-lacZ using the same restriction sites. Plasmids for pbpA or ctpA expression, pPBP1 and pCtpA, were constructed by cloning the pbpA and ctpA genes behind the Pspac promoter in pSK9067 (52) using primer pairs pbpA_F-SalI/pbpA_R-EcoRI and ctpA_F-SalI/ctpA_R-EcoRI, respectively. A plasmid for pbpA* expression was constructed by use of the same primer pair but using pMAD-PBP1* (53) as a template. For membrane topology analysis using pKTop (44), sosA, sosAd10, sosAd10(44A) and sosAd40 were PCR amplified from respective expression constructs using primer SosA_F-BamHI and primers SosA_R-KpnI, SosAd10_R-KpnI or SosAd40_R-KpnI, respectively, and cloned in frame in front of the phoA-lacZ chimeric gene using BamHI/KpnI restriction sites.
In order to construct JE2 strains producing a fluorescent fusion of FtsZ, EzrA or GpsB in the presence of the sosA expression plasmid, JE2 pSosA was transduced with a lysate from SH4665 (SH1000 pCQ11-FtsZ-eYFP), JGL227 (SH1000 ezrA-gfp+) or JGL228 (SH1000 gpsB-gfp+), resulting in SJF4694 (JE2 pRAB12-lacZ pCQ11-FtsZ-eYFP), SJF4696 (JE2 pRAB12-lacZ ezrA-gfp+) and SJF4699 (JE2 pRAB12-lacZ gpsB -gfp+), respectively. Control strains SJF4693 (JE2 pRAB12-lacZ pCQ11-FtsZ-eYFP), SJF4696 (JE2 pRAB12-lacZ ezrA-gfp+) and SJF4699 (JE2 pRAB12-lacZ gpsB-gfp+) were constructed by a phage transduction of JE2 pRAB12-lacZ with lysates from SH4665 (SH1000 pCQ11-FtsZ-eYFP), JGL227 (SH1000 ezrA-gfp+) or JGL228 (SH1000 gpsB-gfp+), respectively. To induce FtsZ-eYFP production in SJF4693 and SJF4694, cells were grown in the presence of 50 μM IPTG.
In order to screen for interaction of SosA with cell-division/cell wall synthesis proteins of S. aureus, sosA was amplified from S. aureus SH1000 genomic DNA using primers ALB133 and ALB134, and ligated into pUT18C using BamHI/EcoRI cut sites, resulting in pT18-SosA. All fusion constructs to T25 and control vectors were generated previously (38, 54-56, SI Appendix, Table S2).
Determination of OD600 and CFU after MMC treatment
Strains were grown overnight on TSA plates and used for inoculating TSB and allowed to grow to OD600 = 0.05 when Mitomycin C (MMC from Streptomyces caespitosus, Sigma-Aldrich) was added. Cell density was monitored by OD600 measurements at intervals onwards and samples were withdrawn to determine culture colony forming units by serial dilution in 0.9% w/v NaCl and plating on TSA.
Flow cytometry
Cell size distributions of cultures were arbitrarily quantified by flow cytometry using the forward scatter signal (FSC-A) acquired on a BD Accuri™ C6 flow cytometer (BD Biosciences). Cell samples were diluted in 0.9% w/v NaCl to an approximate density of 106 cells/ml and sonicated briefly to allow acquisition of scatter signal from single cells predominantly. Sonication was performed with a Bandelin sonopuls HD2070/UW2070 apparatus (Bandelin electronics, Germany) fitted with the MS 73 probe. Ten pulses of 500 msec were given at 50% power. All flow cytometry experiments were independently repeated at least twice with similar results.
Plating assays
Spot dilution was used to evaluate plating efficiency of strains carrying various plasmid constructs. Strains were grown exponentially until an approximate OD600 of 0.5 under plasmid selective conditions. Strains were then ten-fold serially diluted in 0.9% w/v NaCl and positioned as 10 µl spots on TSA containing selective antibiotics with/without inducer at indicated concentrations and incubated at 37°C overnight. All spot plating assays were independently repeated with similar results.
Labelling S. aureus with HADA
Cells grown to mid-exponential phase (OD600 ~ 0.5) were incubated with 500 µM HADA at 37°C for 5 min. Cells were then washed by centrifugation and resuspension in PBS.
Labelling S. aureus with NHS ester
Labelling with NHS ester was performed as described before (33). Briefly, cells grown to mid-exponential phase (OD600 ~0.5) were collected by centrifugation and growth medium was discarded. Cells were resuspended in PBS containing 8 μg ml-1 Alexa Fluor 647 NHS ester (Invitrogen) and incubated at room temperature for 5 min. Cells were washed by centrifugation and resuspension in PBS.
Fixing
Cells were fixed by incubation in 4% (w/v) paraformaldehyde at room temperature for 30 min.
Widefield Epifluorescence Microscopy
Fixed cells were dried onto a poly-L-Lysine coated slide and mounted in PBS. Imaging was performed using either a Nikon Ti Inverted microscope fitted with a Lumencor Spectra X light engine or a v4 DeltaVision OMX 3D-SIM system (Applied Precision, GE Healthcare, Issaquah, USA).
3D Structured Illumination Microscopy
A high precision cover slip (High-precision, No.1.5H, 22×22mm, 170±5 µm, Marienfeld) was cleaned by sonicating in 1 M KOH for 15 min at room temperature. The coverslip was washed with water and incubated in 0.01% (w/v) poly-L-lysine solution (Sigma) for 30 min at room temperature. The coverslip was rinsed with water and dried with nitrogen. Fixed cells were dried onto the poly-L-Lysine coated cover slip and mounted on a slide with SlowFade Diamond (Invitrogen). 3D SIM visualisation was performed using a v4 DeltaVision OMX 3D-SIM system (Applied Precision, GE Healthcare, Issaquah, USA) equipped with a Plan Apo 60x, 1.42 NA oil objective, using 1.514 immersion oil, a 642 nm laser and a standard excitation/emission filter set (683/40). For each z-section, a sample was imaged in 5 phase shifts and 3 angles. The z-sections were 0.125 nm in depth. Raw data were reconstructed with the Softworx software (GE Healthcare) using OTFs optimized for the specific wavelength and oil used.
Time-lapse microscopy
Strains were grown overnight in TSB medium at 37°C, then diluted 100 times in fresh TSB and grown until OD = 0.1. Cells were washed once with fresh TSB and spotted onto TSB-acrylamide (10%) pads previously incubated for 2h in TSB medium supplemented, when appropriate, with 0.04 µg/mL mitomycin C (MMC) or 200 ng/mL anhydrotetracycline (AHT). Pads were placed into a Gene frame (Thermo Fisher Scientific) and sealed with a cover glass. Phase-contrast images were acquired on a DV Elite microscope (GE healthcare) equipped with a sCMOS (PCO) camera and a 100x oil-immersion objective. Images were acquired with 200 ms exposure time every 4 minutes for at least 6 h at 37°C using Softworx (Applied Precision) software. Images were analyzed using Fiji (http://fiji.sc).
Western blot analysis
A custom peptide (EZBiolab) comprising AAs 35-77 of SosA was conjugated, through NH2 (N-terminus) and NH2 of lysines (K) to a KLH carrier protein using glutaraldehyde and used for antibody production in rabbits (Covalab). Bacterial cells were pelleted and frozen immediately at −80°C before being lysed in TE-buffer by bead-beating (Fastprep-24™, MP Biomedicals). The protein concentration of the lysates was measured using the Qubit™ Protein Assay Kit (ThermoFisher Scientific) and were normalized to equal amounts of protein, separated on NuPAGE® 4-12% Bis-Tris gels using MES buffer and the XCell sure-lock mini-cell system (ThermoFisher Scientific), transferred to a polyvinylidene difluoride membrane(PVDF) and probed with primary antibody diluted 1:500. Bound antibody was detected with the WesternBreeze® Chemiluminescent Kit, anti-rabbit according to the instructions from the manufacturer (ThermoFisher Scientific).
Bacterial two-hybrid assay
Interactions between SosA and proteins involved in cell division and cell wall synthesis of S. aureus were tested using E.coli BTH101 (∆cyaA) cotransformed with pT18-SosA and a plasmid carrying a fusion of T25 with a cell-division or cell wall synthesis protein on minimal medium as previously described (38).
SI Appendix: 3 supplementary tables, 8 supplementary figures, 3 videos
Acknowledgements
This work was supported by grants from the Danish Council for Independent Research – Technology and Production (1337-00129 and 1335-00772) to M.S.B., by the Danish National Research Foundation (DNRF120) to H.I., and by the Medical Research Council (MR/K015753/1 to the Wolfson Light Microscopy Facility at the University of Sheffield and MR/N002679/1). We would like to thank Prof Simon Jones (University of Sheffield) for the HADA dye.M.S.B., K.W., P.K., M.T.C., S.J.F., and H.I. conceived and designed the study. Experiments were performed by M.S.B., K.W., C.G., and A.L.B. M.S.B. K.W., C.G., G.L., D.F., J.-W.V., S.J.F., and H.I. contributed to analysis of data and drafting of the manuscript. All authors read and approved the final manuscript.