Abstract
The worldwide spread of Staphylococcus aureus strains resistant to almost all β-lactam antibiotics is of major clinical concern. β-lactams interfere with cross-linking of the bacterial cell wall, but the killing mechanism of this important class of antibiotics is not fully understood. Here we show that sub-lethal doses of β-lactams stimulate the growth of S. aureus mutants lacking the widely conserved chaperone ClpX. S. aureus clpX mutants have a severe growth defect at temperatures below 37°C, and we reasoned that a better understanding of this growth defect could provide novel insights into how β-lactam antibiotics interfere with growth of S. aureus. We demonstrate that ClpX is important for coordinating the S. aureus cell cycle, and that S. aureus cells devoid of ClpX fail to divide, or lyze spontaneously, at high frequency unless β-lactams are added to the growth medium. Super-resolution imaging revealed that clpX cells display aberrant septum synthesis, and initiate daughter cell separation prior to septum completion at 30°C, but not at 37°C. FtsZ localization and dynamics were not affected in the absence of ClpX, suggesting that ClpX affects septum formation and autolytic activation downstream of Z-ring formation. Interestingly, β-lactams restored septum synthesis and prevented premature autolytic splitting of clpX cells. Strikingly, inhibitors of wall teichoic acid (WTA) biosynthesis that work synergistically with β-lactams to kill MRSA synthesis also rescued growth of the clpX mutant, underscoring a functional link between the PBP activity and WTA biosynthesis. The finding that β -lactams can prevent lysis and restore septum synthesis of a mutant with dysregulated cell division lends support to the idea that PBPs function as coordinators of cell division and that β -lactams do not kill S. aureus simply by weakening the cell wall.
Author Summary The bacterium Staphylococcus aureus is a major cause of human disease, and the rapid spread of S. aureus strains that are resistant to almost all β-lactam antibiotics has made treatment increasingly difficult. β-lactams interfere with cross-linking of the bacterial cell wall but the killing mechanism of this important class of antibiotics is still not fully understood. Here we provide novel insight into this topic by examining a defined S. aureus mutant that has the unusual property of growing markedly better in the presence of β-lactams. Without β-lactams this mutant dies spontaneously at a high frequency due to premature separation of daughter cells during cell division. Cell death of the mutant can, however, be prevented either by exposure to β-lactam antibiotics or by inhibiting synthesis of wall teichoic acid, a major component of the cell wall in Gram-positive bacteria with a conserved role in activation of autolytic splitting of daughter cells. The finding that the detrimental effect of β-lactam antibiotics can be reversed by a mutation that affect the coordination of cell division emphasizes the idea that β-lactams do not kill S. aureus simply by weakening the cell wall but rather by interference with the coordination of cell division.
Introduction
Staphylococcus aureus is a commensal bacterium capable of causing a variety of both localized and invasive infections. Due to its ability to acquire resistance to all relevant antibiotics S. aureus remains a major clinical challenge worldwide [1]. The most challenging antimicrobial resistance issue in S. aureus has been the dissemination of methicillin-resistant S. aureus (MRSA) strains that are resistant to almost all β-lactam antibiotics, one of the safest and most widely used classes of antibiotics ever developed [2]. Early work on the mechanism of action of β-lactams culminated in the discovery that penicillin inhibits crosslinking of peptidoglycan (PG), the central component of bacterial cell walls [3]. The enzymes mediating cross-linking of peptidoglycan strands, the targets of penicillin, were therefore designated penicillin binding proteins (PBPs). The realization that penicillin inhibits PG crosslinking led to the classical model in which penicillin-mediated cell lysis is believed to occur as a consequence of a mechanically weakened cell wall incapable of withstanding high intracellular turgor [3,4]. The killing effect of β-lactam antibiotics, however, has turned out to be more complex [5-9], and may even vary between bacteria, as the organization of PG synthesis and the number of PBPs differ widely between bacterial species [10]. Spherical bacteria such as S. aureus have only one cell wall synthesis machine, and S. aureus encodes only four PBPs [11]. Notably, MRSA and other Staphylococci have obtained resistance to β-lactams by horizontal acquisition of the mecA gene encoding an alternative PBP (PBP2a) that is resistant to inhibition by most β-lactams [12,13]. PBP2a mediated resistance additionally depends on several intrinsic factors that can be targeted by specific compounds to re-sensitize MRSA to β-lactams [14-16]. As an example, inhibitors of wall teichoic acid (WTA) biosynthesis, work synergistically with β-lactams to kill MRSA both in vitro and in in vivo models of infection, thereby opening a novel paradigm for combination treatment of MRSA [16]. Indeed, a combination strategy pairing β-lactamase inhibitors with β-lactams has proven highly successful in restoring β-lactam efficacy against Gram-negative bacteria [17].
The ClpX chaperone is conserved among bacteria and organelles of eukaryotic cells [18]. The ClpX chaperone has a dual role in cells targeting proteins for degradation by the ClpP protease and, independently of ClpP, facilitating protein folding and interactions [18]. S. aureus clpX mutant exhibits a mild growth defect at 37°C that is severely exacerbated at 30°C [19,20]. This cold-sensitive growth defect of the clpX mutant is independent of ClpP, and is alleviated by loss-of-function mutations in the ltaS gene [20,21]. ltaS encodes the LtaS synthetase that is required for synthesis of lipoteichoic acid (LTA), an essential cell wall polymer of Gram-positive bacteria controlling cell division and autolytic activity [22]. Interestingly, inactivation of ClpX restored the septum placement defects of cells depleted for LTA, suggesting a link between ClpX and cell division in S. aureus [20].
Indeed, the data presented here demonstrate that ClpX becomes critical for progression of S. aureus septum synthesis as the temperature decreases. In cells with delayed septum synthesis, autolytic splitting of daughter cells is activated prior to septum completion resulting in cell lysis, unless β-lactam antibiotics are added to the growth medium. Strikingly, inhibitors of WTA biosynthesis, similarly to mutations in ltaS specifically rescue growth of S. aureus clpX mutants, emphasizing a fundamental connection between the transpeptidase activity of PBPs and teichoic acids biosynthesis.
In conclusion, this study identifies the ClpX chaperone as an important player in S. aureus cell division at sub-optimal temperatures, and provides novel insight into the link between β -lactam antibiotics and cell division in this important pathogen.
Results
β-lactam antibiotics stimulate growth of S. aureus clpX mutants
Serendipitously, in determining the susceptibility of the S. aureus clpX mutant to oxacillin, we repeatedly observed zones of improved growth at a certain distance from the filter discs containing the antibiotic, a phenomenon that was not observed for wild-type strains (marked by arrow in Fig 1a). This observation indicated that sub-lethal concentrations of oxacillin stimulate growth of the clpX mutant. Indeed, addition of sub-lethal concentrations of oxacillin rescued the severe growth defect normally seen for S. aureus clpX mutants at 30°C (Fig 1b). To investigate if growth of S. aureus clpX mutants is generally improved by addition of β-lactam antibiotics, three S. aureus strains of clinical origin, representing both MRSA (JE2) and methicillin-sensitive S. aureus (SA564 and Newman), and the corresponding clpX deletion strains were grown in broth containing oxacillin, meropenem or cefuroxime (representing three different chemical classes of β-lactams) in various concentrations below and above the previously determined MIC values [23]. We found that the presence of β-lactam antibiotics increased the growth rate and the final yield of the clpX mutants in all strain backgrounds (Fig 1c and 1d; S1 Fig). As shown previously [23], inactivation of clpX increased the MIC values in the JE2 background (S1 fig), but not in the MSSA strain backgrounds. A wide range of β-lactam concentrations was tested, but we did not identify any concentration at which the growth rates of the wild-type strains were enhanced (S1 Fig). For comparison, we also included clpP mutants, but observed no or only a minor stimulatory effect on the growth rate of the clpP mutants in the presence of β-lactams (S1 Fig). We conclude that the ClpX dependent growth defect that is suppressed by β-lactams is caused by loss of ClpX chaperone activity, not loss of ClpXP protease activity.
Oxacillin prevents premature growth arrest and spontaneous lysis of clpX cells
Given the unusual ability of β-lactams to stimulate growth of the clpX mutant, we reasoned that a better understanding of the clpX growth defect could provide novel insights into how β-lactam antibiotics interfere with growth of S. aureus. To investigate how β-lactams improve growth of the clpX mutant, we studied the growth of single cells of the S. aureus SA564 wild type and clpX mutant in the absence or presence of oxacillin at 30°C using automated phase contrast time-lapse microscopy. Interestingly, the time-lapse experiments demonstrated that only few clpX cells started to divide and that many clpX cells lysed spontaneously or stopped dividing early on in the experiment (Fig 2a and S1 Movie); accordingly only about one third of the clpX mutant cells ended up forming a micro-colony. While addition of sub-lethal concentrations of oxacillin did not affect growth of the wild type, growth of the clpX mutant was clearly stimulated at the single-cell level (Fig 2b). In the presence of oxacillin, only very few clpX cells lysed or stopped dividing during the experiment, and the clpX mutant reached a higher number of cells than the wild type by the end of the experiment (Fig 2b and 2c). To quantify the cell generation time (time between two divisions), we tracked the fate of each individual cell in one representative micro-colony of each strain during the first 8 h of the experiment (S2 Fig). For the wild type, the average cell generation time was 66 ± 42 min without oxacillin, and 78 ± 58 min with oxacillin, with only few divisions observed after 4 h in both cases (Fig 2c, S2 Fig, and S1 Table). For the clpX mutant, the average cell generation time was 71 ± 36 (note, only 11 divisions) min in the absence of oxacillin. Interestingly, in the presence of oxacillin the generation time for clpX mutant cells decreased throughout the experiment to an average of 38 ± 23 min during the last 2 h (Fig 2c, S2 Fig, and S1 Table). Hence, oxacillin seems to stimulate growth of the clpX mutant by both shortening the generation time and preventing premature growth arrest, and lysis of the mutant.
Aberrant septum synthesis and premature splitting of clpX daughter cells
To further investigate the clpX phenotypes, we studied the morphology of wild-type and clpX mutant cells by transmission electron microscopy (TEM) and scanning electron microscopy (SEM) after growth at 30°C. As reported before, cells lacking ClpX are significantly smaller than wild-type cells and have a thickened cell wall (Fig 3a) [23]. Consistent with the spontaneous cell lysis observed in the time-lapse microscopy approximately 10% of the clpX mutant cells grown at 30°C appeared as lysed ghost cells in the TEM images (S3 Fig). Interestingly, these ghost cells had a characteristic appearance in which the cell wall was ripped apart at the tips of ingrowing, still incomplete, septa (see examples in Fig 3c), indicating that these cells underwent lysis while in the process of daughter-cell splitting. To divide, S. aureus builds a septal cross wall generating two hemispherical daughter cells connected through a narrow peripheral ring [24,25]. Resolution of this peripheral wall ring leads to rapid splitting of daughter cells, in a process designated as “popping” [25]. Popping normally occurs only in cells with closed septa and consistent with this notion, daughter cell splitting was not observed in wild-type cells with incomplete septa (S3 Fig). In contrast, a fraction (quantified below) of the clpX mutant cells seemed to have initiated splitting of daughter cells despite having incomplete septa (see examples Fig 3d and 3e, and S3 Fig). In TEM images, initiation of splitting appeared as small invaginations at the external edge of the septum in cells with incomplete septa, but occasionally, a complete splitting of the ingrowing septum and elongation of cells were observed (Fig 3d and 3e, middle panels). In SEM images, the stage of septum ingrowth could not be monitored. Strikingly, however, it was possible to visualize clpX cells in the process of splitting despite displaying a non-closed septal cross-wall (seen as a hole; Fig 3c right panel), or clpX mutant cells in the process of splitting while still being connected by an undivided cytoplasm (Fig 3d right panel). Taken together, these findings strongly suggest that in the absence of ClpX, the system controlling the onset of autolytic separation of daughter cells becomes dysregulated, and that premature splitting of clpX cells with incomplete septa result in cell lysis.
Some clpX cells also displayed septa extending asymmetrically towards the cell center, and in extreme cases extending inwards only from one side (+/- premature split; Fig 3d and 3e). This contrasts with wild-type S. aureus cells whose septa always extended symmetrically inwards from the edge of the cell wall (Fig 3a and S Fig 3). Finally, in some clpX cells, unordered membranous material reminiscent of mesosome-like structures [26] were observed at the site of septum ingrowth (Fig 3e and 3f). The latter phenotypes suggest that ClpX contributes to coordinating septum formation in S. aureus.
The ClpX chaperone becomes critical for septum completion at 30°C
To quantify these phenotypes and to observe overall differences in progression of the cell cycle between wild-type and clpX cells, we performed Super-Resolution Structured Illumination Microscopy (SR-SIM) on cells stained with the membrane dye Nile red, and scored cells according to the stage of septum ingrowth as described by Monteiro et al. [24]; see Fig 4a for example images. To enumerate cells with incomplete septa that show signs of premature splitting, cells were additionally stained with fluorescently modified vancomycin (Van-FL), which labels the entire cell wall (cell periphery and septum), or with a green fluorescent derivative of wheat germ agglutinin WGA-488 that only labels the peripheral wall [24,27]. To estimate the number of lysed cells, DNA was stained with the blue dye Hoechst 3334. In this analysis, no significant differences in the distribution of cells in the different phases were observed for wild-type and clpX cells grown at 37°C (Fig 4a). At 30°C, however, significantly fewer clpX cells displayed a complete septum (4% as opposed to 15% of wild-type cells; P < 0.001). Moreover, while the fraction of cells that were in the process of building a septum (phase 2) was similar in wild-type and clpX cells at both temperatures, a more detailed analysis of the phase 2 cells revealed striking differences (Fig 4): consistent with the TEM analysis, a substantial number of clpX cells with incomplete septa showed signs of premature daughter cell splitting (20% of phase 2 cells), or had asymmetrical septum ingrowth (7% of phase 2 cells) when cells were grown at 30°C. None of these phenotypes were observed in wild-type cells at any temperature. While asymmetrical septum ingrowth was not observed in clpX cells grown at 37°C, premature splitting cells could be observed, however, at a lower frequency (Fig 4b). Furthermore, when subdividing phase 2 cells into two subclasses based on the extent of septum ingrowth, the proportion of clpX cells displaying early septum ingrowth (defined as cells with less than 15% septum ingrowth; see examples in Fig 4b) was significantly (P < 0.001) higher after growth at 30°C compared to 37°C, and when compared to the wild-type. For wild-type cells, an equal fraction of cells displayed early septum ingrowth at 30°C and 37°C. Finally, SR-SIM confirmed that the fraction of lysed clpX cells increased significantly when the temperature was decreased (2% at 37°C, and 16% at 30°C, P < 0.001). In comparison, the proportion of lysed wild-type cells was estimated to be below 2% at both temperatures.
In conclusion, the proportion of clpX mutant cells displaying a complete septum or late septum ingrowth was significantly reduced at 30°C, while the proportion of clpX cells displaying early septum ingrowth and aberrant septum was significantly increased at 30°C. Thus, the microscopy analyses suggest that ClpX chaperone activity becomes critical for the ability of S. aureus to complete the division septum as the temperature decreases.
Oxacillin restores the cell cycle of clpX cells
To further examine how β-lactams improve growth of the clpX mutant, we performed SR-SIM analysis on oxacillin treated wild-type and clpX mutant cells grown at 30°C, as described above (Fig 4). Interestingly, sub-lethal concentrations of oxacillin significantly increased the fraction of phase 3 cells (closed septum): from 15 to 31% in the wild-type (P < 0.001), and from 4 to 14% in the clpX mutant (P < 0.001). Moreover, oxacillin significantly decreased the fraction of clpX cells (phase 2) that had initiated cell separation prior to septum completion from 20% to 2%, and in line with this observation, almost no lysed clpX mutant cells were observed (Fig 4b). Hence, oxacillin increases the fraction of cells with complete division septa in both the wild-type and the clpX backgrounds, and prevents premature splitting of clpX cells with incomplete division septa. In contrast, asymmetrical ingrowth of septa is still readily observed in oxacillin treated clpX mutant cells (Fig 4a and b).
These conclusions were supported when the oxacillin treated SA564 wild-type and clpX cells were analyzed by TEM (S4 Fig). TEM also showed that oxacillin prevented formation of mesosome-like structures in clpX cells (S4 Fig). However oxacillin treatment conferred a number of well described morphological changes that were shared by wild-type and clpX cells including thickening and misplacement of septa, a more fuzzy surface, and blurring of the electron-dense septal mid-zone (S4 Fig) [7,27,28]. Finally, many cells were present in pairs that have remained partly attached at the midline following septum completion (S4 Fig). Hence, both the SR-SIM and the TEM images support that oxacillin, even in concentrations well below the MIC value, prolongs phase 3 and delays splitting of the septum.
Oxacillin antagonizes the septum progression defects conferred by inactivation of ClpX
To directly assess the impact of ClpX and oxacillin on progression of septal PG synthesis, we used an established metabolic labeling method with fluorescent D-amino acids (FDAAs) to visualize regions of new PG insertion [24,29,30]. PG synthesis was followed at 30°C and 37°C by sequentially labeling cells with FDAAs of different colors, thereby creating a virtual time-lapse image of PG synthesis [24,29,30]. Cells were first pulse-labeled for 10 min with green nitrobenzofurazan-amino-D-alanine (NADA), followed by a 10-min pulse with the blue hydroxycoumarin-amino-D-alanine (HADA). Labeled cells were imaged by SR-SIM, and progression of PG synthesis was scored in 300 randomly picked wild-type and clpX mutant cells grown in the absence or presence of oxacillin (Fig 5.,note that NADA is displayed in red). In the absence of oxacillin, PG synthesis proceeded from phase 1 (no septa, PG synthesis takes place in the lateral wall) to phase 2 (septal PG synthesis progresses inwards), and finally phase 3 (closed septum, PG synthesis occurs in both septum and the lateral wall) in > 95% of wild-type cells, as described in [24,25] (see Fig 5a). When the clpX mutant was grown at 37°C, PG synthesis followed the wild-type paradigm (S5 Fig). In contrast, when the clpX mutant was grown at 30°C, the septal PG synthesis progressed abnormally in a substantial fraction of phase 2 cells, as 22 ± 3 % of the clpX cells that had initiated septum formation in the first period of labeling (NADA) did not continue septum synthesis in the second period of labeling (HADA). Instead, the HADA signal co-localized with the NADA signal in the early septum ingrowth, and additionally, a peripheral HADA signal was visible (see examples in Fig 5a indicated by grey asterisks). Because other clpX cells displaying NADA labeling in an early septal ingrowth were indeed capable of septum progression and septum closure (green asterisks in Fig 5a), the septal PG synthesis rate does not seem to be generally reduced in the clpX mutant. Instead, the co-localization of the NADA and HADA in an early septum ingrowth may reflect stalling of septum synthesis in a subpopulation of clpX cells. Interestingly, in the presence of a sub-lethal concentration of oxacillin the fraction of clpX cells displaying co-localization of NADA and HADA at the early-septum ingrowth was reduced to 6 ± 2 % (Fig 5b).
FDAAs only incorporate into newly synthesized PG and therefore premature splitting initiating from the peripheral wall cannot be detected with this approach [30]. However, splitting of newly synthesized, still incomplete, septum was observed (pink asterisks in Fig 5a), and while this phenotype was not observed in wild-type cells, this phenotype was displayed in about 20 ± 2 % of the clpX cells (phase 2 cells) grown in the absence of oxacillin. In the presence of oxacillin, only 8 ± 2 % of clpX cells showed splitting of newly synthesized still incomplete septa (see example in Fig 5b). In wild-type cells grown in the presence of oxacillin, NADA- and HADA signals more often co-localized in the entire septal plane (examples depicted in Fig 5b), supporting that wild-type cells grown with oxacillin spend longer time in phase 3. We conclude that at temperatures below the optimum, the ClpX chaperone activity becomes important for S. aureus septal PG synthesis to proceed beyond the point of septum initiation, and that oxacillin antagonizes the septum progression defects conferred by inactivation of ClpX.
Oxacillin promotes septal PG synthesis in clpX cells with premature split
The results presented so far suggest that oxacillin improves growth of an S. aureus clpX mutant by stimulating septal PG synthesis and inhibiting premature splitting and lysis of daughter cells. To investigate septal PG synthesis in cells with premature splitting, we randomly picked 50 clpX cells grown at 30°C that had initiated septum formation during incubation with NADA, and that displayed the characteristic morphology of premature splitting, and assessed where HADA was incorporated in these cells. Interestingly, only very few clpX cells displaying premature septum split continued synthesizing septum (Fig 5c-i); instead HADA was incorporated at the cell periphery (Fig 5c-ii). In a few cells no HADA signal was detected at all (Fig 5c-iii). Similar results were obtained if the order of labeling was reversed (data not shown). Hence, septal PG synthesis seems to stop and instead become dispersed to the peripheral wall in clpX cells displaying splitting of a yet incomplete septum. Remarkably, in oxacillin treated cells, septum synthesis progressed normally in most cells with premature split (40 ± 1 of 50 cells, P < 0.001, Fig 5c). Taken together, this analysis demonstrates that oxacillin antagonizes the arrest of septum synthesis observed in clpX cells with premature septal split.
FtsZ localization and dynamics are not affected in the clpX mutant
ClpX from diverse bacteria interacts directly with FtsZ suggesting that the ClpX chaperone has a conserved role in assisting assembly/disassembly of the FtsZ polymer [31-34]. We therefore reasoned that ClpX may regulate septum progression by interfering with FtsZ dynamics. To study localization and dynamics of FtsZ, a plasmid expressing eYFP-tagged derivative of FtsZ from an IPTG-inducible promoter [35] was introduced into S. aureus wild-type and clpX mutant cells, and time-lapse fluorescence microscopy was performed on cells growing on a semi-solid matrix at 30°C. In both wild-type and clpX mutant cells, Z-ring dynamics progressed predictably throughout the cell cycle (S2 Movie and Fig 6a): in newly divided cells, the Z ring has the same diameter as the cell until the ring starts to reduce in diameter and eventually closes (as described in [36]). Following closure, FtsZ undergoes a period of highly dynamic re-distribution, before the Z-ring cycle starts over again in newly divided cells. Hence, FtsZ dynamics appear not to be affected by lack of ClpX activity. Next, we imaged the relative localization of FtsZ and PG synthesis by sequentially labeling PG synthesis with FDAAs as described above, except that tetramethylrhodamine 3-amino–d-alanine (TADA, red signal) was used instead of NADA to avoid overlap with the yellow eYFP signal. In both wild-type and clpX cells, the eYFP signal localized ahead of septal PG synthesis in all phase 2 cells (Fig 6b and overview in S6 Fig). Specifically, FtsZ also localized ahead of the FDAA signal in clpX cells having HADA and TADA signal co-localizing in an early septum in growth (Fig 6b). Strikingly, the FtsZ signal maintained its septal localization in clpX cells with premature split and arrest of septal PG-synthesis (see example in Fig 6b). As also shown above, PG incorporation in such cells takes place in the peripheral wall. Hence, our data supports the idea that FtsZ dynamics is not impeded in cells lacking ClpX.
Inhibitors of WTA biosynthesis rescue growth of the clpX mutant
Finally, we asked if the ability to rescue growth of an S. aureus clpX mutant is specific for the β-lactam class of antibiotics (S7 Fig). The compounds assessed were either antibiotics with completely different targets, or compounds inhibiting various steps in the cell envelope synthesis pathway. Interestingly, only tunicamycin and tarocin A1, two well characterized inhibitors of WTA biosynthesis that work synergistically with β-lactams to kill S. aureus [14,16], stimulated growth of the clpX mutant (Fig 7 and S7 Fig). Strikingly, tunicamycin was almost as efficient as oxacillin in stimulating growth of the clpX (Fig 7c). In contrast, other late stage inhibitors of PG synthesis, such as vancomycin, did not stimulate growth of the clpX mutant, even though this antibiotic, similarly to β-lactams, prevents PG cross-linking (S7 Fig). Lysostaphin, which breaks already formed cross-bridges, also had no stimulatory effect on growth of the clpX mutant (S7 Fig). These findings indicate that reduced cross-linking per se does not alleviate the growth defect of the clpX mutant and that the growth defect of S. aureus clpX mutants is specifically rescued by tunicamycin, tarocin A1, and β-lactam antibiotics underscoring a functional link between the PBP activity and WTA biosynthesis.
Discussion
Assembly of the bacterial cell division machinery is a highly coordinated process with proteins recruited to the division site in a specific order and depending on the timely interaction between a large number of proteins [37]. Here, we show that the widely conserved ClpX chaperone plays an important role in staphylococcal cell division at 30°C but not at 37°C. In wild-type S. aureus cells, splitting of daughter cells is not initiated prior to septum closure. In contrast, a substantial fraction of clpX cells displaying incomplete septa had initiated splitting of daughter cells indicating that the system responsible for coordinating autolytic splitting with septum completion has become dysregulated. In clpX cells displaying the premature splitting phenotype, septal PG synthesis was not continued, and instead became dispersed to the peripheral wall demonstrating that clpX cells with premature split are unable to finalize the septum. The detrimental character of this defect likely prevents cells from undergoing further divisions, explaining why a large proportion of clpX cells are non-dividing and end up lysing. In support hereof, TEM pictures show that most clpX ghost cells were in the process of splitting despite having an incomplete septum. This is likely due to turgor pressure forces breaking the tip of the ingrowing septum where the cell wall is thin and mechanically weak [38]. Hence, we assume that premature splitting is the underlying cause for the high rate of spontaneous lysis observed among clpX cells.
Importantly, cells devoid of ClpX contain elevated levels of the two major autolysins associated with separation of S. aureus daughter cells, Sle1 and Atl [20,21,39,40]. Therefore, premature splitting of clpX cells could simply be a consequence of excess autolysins. However, whilst the elevated levels of SleI and Atl may contribute to the premature splitting and spontaneous lysis of clpX cells, we believe that additional factors are in play. This is based on the findings that i) premature splitting and lysis of clpX cells is more frequent at 30°C than at 37°C, whereas autolysin levels are elevated at both temperatures [20,21,39,40]; and, ii) the clpX phenotypes described here are not shared by a clpP mutant, although clpP mutant cells also contain elevated Sle1 levels [20,21,23,41]. Taken together these findings indicate that the high levels of autolysins are more detrimental to clpX cells growing at 30°C, suggesting that ClpX in addition to controlling the levels of autolysins also affects their activation via a temperature dependent pathway. As stalling of early septum synthesis in clpX cells was observed only at the lower temperature, we speculate that stalling of septum synthesis contributes to premature activation of autolysins in the clpX mutant as depicted in the working model shown in Fig 8. In this model, S. aureus depends on ClpX chaperone activity for transforming an early stage divisome complex into a late stage divisome complex at 30°C but not at 37°C. At both temperatures, the high levels of autolysins in the clpX mutant will make the mutant more prone to initiate daughter cell separation before septum completion. However, stalling of the divisome by an unknown mechanism exacerbates the risk of premature split at 30°C. In support hereof, premature split was also observed in clpX cells grown at 37°C, however, at this temperature septal progression seems to proceed fast enough to enable completion of the septum, as outlined in Fig 8. Notably, FtsZ localization and dynamics were not affected in the absence of ClpX, suggesting that ClpX affects septum formation and autolytic activation downstream of Z-ring formation. Intriguingly, S. aureus cytokinesis was recently proposed to occur in two-steps: an initial FtsZ dependent slow step that may drive the initial membrane invagination, and a second faster step driven by PG synthesis and recruitment of late division proteins such as MurJ [36]. Hence, we speculate that ClpX promotes septum formation at 30°C by assisting recruitment of MurJ or other central components of the late divisome complex.
Because mis-coordination in activation of autolytic enzymes may have fatal consequences for the cell, regulatory checkpoints that coordinate the autolytic system with septum completion likely exist, however, little is known about these mechanisms. Remarkably, the growth and lysis defect of the clpX mutant was alleviated by sub-lethal concentrations of β-lactam antibiotics. This intriguing finding is to our knowledge the first example of β-lactam antibiotics increasing the growth-rate and preventing spontaneous lysis of a bacterial mutant. The presented data show that oxacillin simultaneously rescues septum synthesis, and prevents premature splitting, mesosome formation, and spontaneous lysis of the clpX mutant, lending further support to a linkage between these phenotypes. The ability of sub-lethal concentrations of β-lactam antibiotics to suppress spontaneous lysis of clpX mutant cells was surprising, as the lethal activity of β-lactam antibiotics is believed to stem from the loss of wall integrity accompanied by cell lysis [5,9,42]. Here, we observed that oxacillin treatment of both S. aureus wild-type and clpX mutant cells increased the fraction of cells displaying a complete division septum, supporting previous findings that β-lactams delay autolytic splitting of daughter cells [7,28]. Moreover, the sequential PG staining experiments showed that late septal FDAA signals often overlap in wild-type cells grown in the presence of oxacillin, indicating that β-lactams also prolong PG synthesis in the completed septum (which may be a consequence of delayed autolytic cell splitting). Consistent with these findings, β-lactams treated S. aureus cells display characteristic thickened septum in TEM images [7,28]. One possible scenario to explain how β-lactams rescue clpX cells is therefore that oxacillin prevents spontaneous lysis of the clpX mutants by suppressing activation of autolytic enzymes and by stimulating late septal PG synthesis (Fig 8). Vice versa, the earlier onset of autolytic activity mediated by inactivation of ClpX, may counteract the delay in autolytic splitting of daughter cells observed in oxacillin treated wild-type cells, thereby explaining the reduced cell generation time of clpX cells compared to wild-type cells in the presence of oxacillin.
In support of a central role of autolysins in the clpX phenotypes, we previously showed that the fast-growing suppressor mutants arising when clpX cells are grown at 30°C have loss-of-function mutations in the ltaS gene encoding the LtaS synthetase that is essential for synthesis of LTA, another key regulator of autolytic activity [20]. Similarly, WTA seems to impede splitting of the septal cross wall and control Atl localization [14,43]. Strikingly, two inhibitors of WTA synthesis, tunicamycin and tarocin A1, were the only other compounds that similarly to β-lactams rescued growth of the clpX mutant. In our working model, we therefore propose that inhibition of WTA stimulates growth of the clpX mutant by impeding premature split of clpX cells (Fig 8).
In conclusion, we have shown that S. aureus cell division is temperature sensitive, and that at 30°C, the ClpX chaperone serves an important function in coordinating initiation of daughter cell separation with septum completion. When ClpX is absent, cell division frequently has a fatal outcome because septal PG synthesis stalls and cell separation is initiated prior to completion of the septum. Interestingly, these defects were prevented by binding of β-lactam antibiotics to the PBP transpeptidase activity domain, indicating that this final stage in PG biosynthesis plays a role in coordinating septum synthesis and activation of autolytic splitting of daughter cells. Consistent with this hypothesis, the transpeptidase site of PBP1 was proposed to take part in a checkpoint-type mechanism functioning at the end of each round of cell division to ensure that autolytic splitting of daughter cells can only take place upon successful completion of septum synthesis [44,45]. Our work therefore supports the idea that in this clinically important bacterium, the effect of β-lactam antibiotics is tightly linked to coordination of cell division.
Materials and Methods
Bacterial strains and growth conditions
Strains used in this study are listed in S2 Table. S. aureus strains were grown in tryptic soy broth media (TSB; Oxoid) under vigorous agitation at 200 rpm at 37°C. In most experiments, 20 ml of medium was inoculated in 200-ml flasks to allow efficient aeration of the medium. For solid medium, 1.5% agar was added to make TSA plates. Erythromycin (7.5 μg ml-1) was added as required. Upon receipt of the low-passage isolate SA564, the strain was cultured once and stored frozen at -80°C. In all experiments, we used bacterial strains freshly streaked from the frozen stocks on TSA plates with antibiotics added as required and incubated overnight at 37°C. The growth was followed by measuring the optical densities at 600 nm. The starting OD was always below 0.05. When inoculating S. aureus clpX deletion strains, care was taken to avoid visibly larger colonies containing potential suppressor mutants [20]. To reduce the selection pressure for suppressor mutants in broth cultures, strains were upon inoculation first grown at 37°C for four generations (OD600 ~0.1-0.2) before shifting to 30°C. S. aureus JE2-derived strains were obtained from the Network of Antimicrobial Resistance in Staphylococcus aureus (NARSA) program (supported under NIAID/NIH contract HHSN272200700055C.
Growth calculations
Growth of S. aureus strains was assessed by measuring optical density (OD600) of cultures grown in 96-well microplates (for end-point OD) or in a Bioscreen C instrument (for growth rates). Overnight cultures were grown in TSB at 37°C. For end-point ODs, overnight cultures were diluted 1:200 in TSB and grown to exponential phase (OD600 0.1) and then diluted 1:10,000 in 200 μl TSB in 96-well format, and incubated 24 h at 30°C with shaking. For growth in the Bioscreen C instrument overnight cultures were diluted in 300 μl TSB to an OD600 of approx. 0.001, and OD600 was measured every 5 min with 20 seconds of shaking before each measurement. All exponential growth rates were determined by growing the relevant strains in a Bioscreen C instrument as described above. The growth rates were automatically calculated as described before [20]. In short, OD600 values were log-transformed and linear regressions were determined for each data point in the OD600 interval from 0.02 to 0.12 based on a window containing 15 data points. The exponential growth rate was identified as the maximal slope of the linear regressions. The standard error of the mean was calculated using values from different biological replicates.
Disc diffusion assays
S. aureus strains were inoculated on TSA plates and incubated at 37°C overnight. The next day, a bacterial suspension was adjusted to a 0.5 McFarland (Sensititre® nephelometer and the Sensititre® McFarland Standard) and streaked on MHA. The plates were allowed to dry prior to the addition of 1 μg oxacillin discs (Oxoid) and incubated at 37°C for 48 hours.
Time-lapse microscopy
S. aureus wild-type (SA564 or 8325-4/pCQ11ftsZ::eYFP) and clpX mutant (SA564ΔclpX or 8325-4ΔclpX/pCQ11ftsZ::eYFP) were grown overnight in TSB medium at 37°C and cultures were diluted 100 times in fresh TSB medium and grown until an OD600 of 0.1. Cells were washed once in fresh TSB medium and spotted onto a TSB-polyacrylamide (10%) slide incubated with TSB medium supplemented when appropriate with 0.008 μg ml-1 oxacillin or with 100 μM IPTG. Acrylamide pads were placed inside a Gene frame (Thermo Fisher Scientific) and sealed with a cover glass as described before [46].
Phase contrast images acquisition was performed using a DV Elite microscope (GE healthcare) with a sCMOS (PCO) camera with a 100x oil-immersion objective. Images were acquired with 200 ms exposure time every 2 or 4 minutes for at least 4 h at 30 °C using Softworx (Applied Precision) software. Images were analyzed using Fiji (http://fiji.sc). Each experiment was performed at least in triplicate. 69 micro colonies were imaged for SA564; 63 micro colonies for SA564 + oxacillin; 43 micro colonies for SA564 clpX and 36 micro colonies for SA564 clpX + oxacillin. Cell tracking was performed using the TrackMate plugin (https://imagej.net/TrackMate) in Fiji, and the output was converted to Newick format for lineage tree plotting and calculation of cell generation times using a custom R script (available at http://github.com/ktbaek/beta-lactams-clpx).
Time-lapse images of FtsZ-eYFP were acquired using a Leica DMi8 microscope with a sCMOS DFC9000 (Leica) camera with a 100x oil-immersion objective and a Spectra X (Lumencor) illumination module. Fluorescent images were acquired every 4 min with 400 ms exposure using a YFP filter cube (Chroma, excitation 492-514 nm, dichroic 520 nm, emission 520-550 nm). Images were processed using LAS X (Leica) and signal was deconvolved using Huygens (SVI) software.
Electron microscopy and image analysis
Overnight cultures grown at 37°C were diluted 1:200 into 40 ml of fresh TSB and grown at 30°C or 37°C to an OD600 of 0.5. Bacteria (SA564, 8325-4, and JE2 and the clpX mutant derived here from) from a 10-ml culture aliquot were collected by centrifugation at 8,000 x g, and the cell pellets were suspended in fixation solution (2.5% glutaraldehyde in 0.1 M cacodylate buffer [pH 7.4]) and incubated overnight at 4°C. The fixed cells were further treated with 2% osmium tetroxide, followed by 0.25% uranyl acetate for contrast enhancement. The pellets were dehydrated in increasing concentrations of ethanol, followed by pure propylene oxide, and then embedded in Epon resin. Thin sections for electron microscopy were stained with lead citrate and observed in a Philips CM100 BioTWIN transmission electron microscope fitted with an Olympus Veleta camera with a resolution of 2,048 by 2,048 pixels. For quantitative analysis, the images were acquired in an unbiased fashion by using the multiple image alignment function in the ITEM software (Olympus). Sample processing and microscopy were performed at the Core Facility for Integrated Microscopy (CFIM), Faculty of Health and Medical Sciences, University of Copenhagen.
Scanning electron microscopy (SEM)
Exponentially growing S. aureus SA564 were collected by centrifugation, fixed in 2% glutaraldehyde in 0.05 M sodium phosphate buffer, pH 7.4 and sedimented on coverslips for 1 week at 4?°C. The cells were washed three times with sodium cacodylate buffer and progressively dehydrated by immersion in a graded series of ethanol (50–100%). Cells were subsequently mounted on stubs using colloidal silver as an adhesive, and sputter coated with gold (Leica Coater ACE 200) before imaging with a FEI Quanta 3D scanning electron microscope operated at an accelerating voltage of 2 kV. Sample preparation and SEM imaging was performed at CFIM.
SR-SIM analysis
SR-SIM was performed with an Elyra PS.1 microscope (Zeiss) using a Plan-Apochromat 63x/1.4 oil DIC M27 objective and a Pco.edge 5.5 camera. Images were acquired with five grid rotation and reconstructed using ZEN software (black edition, 2012, version 8.1.0.484) based on a structured illumination algorithm, using synthetic, channel specific optical transfer functions and noise filter settings ranging from -6 to -8. Laser specifications can be seen in S3 Table. SR-SIM was performed at CFIM.
Analysis of the cell cycle
To address progression of the cell cycle, exponential cultures of S. aureus were incubated for 5 min at room temperature with the membrane dye Nile Red, the cell wall dye WGA-488 or Van-Fl and the DNA dye Hoechst 3334 (S4 Table). Samples were placed on an agarose pad and visualized by SR-SIM as described above. 300 cells were scored according to the stage of septum ingrowth: no septum (phase 1), incomplete septum (phase 2), or non-separated cells with complete septum (phase 3). Dead cells were scored based on Hoechst staining: lysed cells, as cells where DNA had leaked out of the cell and anucleated cells as cells devoid of Hoechst staining. The analysis was performed on two biological replicates.
Additionally, 200 cells were scored according to the state of septum ingrowth (cells with less than 15% septum ingrowth were scored as “early”, while cells with more than 15% septum ingrowth were scored as “late”) and whether the ingrowth was asymmetrical or showed signs of premature splitting. The latter was based on staining with Van-FL. This analysis was performed on two biological replicates.
Analysis of progression of PG synthesis
To evaluate localization of PG synthesis, exponential cultures of S. aureus (SA564 or 8325-4) were pulse labeled with FDAAs; cells were initially incubated 10 minutes with NADA, washed in PBS and resuspended in TSB. The cells were then incubated 10 minutes with HADA, washed with PBS, placed on an agarose pad and visualized by SR-SIM. This experiment was conducted in three biological replicates including a staining in reverse order and one using the red TADA as a replacement for NADA. Analysis on the progression of PG synthesis was performed on 300 cells for each biological replicate with similar results.
To investigate the progression of septal PG synthesis in clpX mutant cells displaying premature split, HADA incorporation was assessed in 50 cells (in each of three biological replicate) that had initiated septum formation during the initial labeling and displayed the characteristic morphology of premature splitting were selected randomly. PG synthesis was followed by assessing.
FtsZ localization in SR-SIM
In order to assess FtsZ relative to the active PG synthesis, S. aureus (8325-4) wild-type and clpX mutant transformed with pCQ11 expressing an eYFP-tagged derivative of FtsZ from an IPTG-inducible promoter were analyzed using sequentially labeling with FDAAs as described above (incubation with TADA for 10 minutes followed by HADA for 10 minutes). Cells were grown at 30°C in the presence of 50 μM IPTG (at higher IPTG concentrations cell division defects were observed in the wild-type strain).
Statistical analysis
Statistical analysis was done using R statistical software. Student’s t-test was used to assess significant differences in growth in the absence or presence of a tested antibiotic. The Chi-squared test of independence was used to determine if there was a significant relationship between the proportion of cells assign to each of the three phases or relevant phenotypes under the tested condition (number of cells in the relevant phase or phenotype/the total number of cells). A value P < 0.05 was considered significant.
Acknowledgments
We would like to thank Professor Simon Foster (University of Sheffield) for the generous gift of FDAA’s and the FtsZ-eYFP fusion plasmid. We would like to thank Ewa Kuninska (University of Copenhagen) for excellent technical assistance and the staff at the Core Facility for Integrated Microscopy (University of Copenhagen) for their enthusiastic assistance in doing SEM, TEM and SR-SIM.