The transpeptidase activity of the essential penicillin-binding protein 2x (PBP2x) of Streptococcus pneumoniae is believed to be important for murein biosynthesis required for cell division. To study the molecular mechanism driving localization of PBP2x in live cells, we constructed a set of N-terminal GFP–PBP2x fusions under the control of a zinc-inducible promoter. The ectopic fusion protein localized at mid-cell. Cells showed no growth defects even in the absence of the genomic pbp2x, demonstrating that GFP–PBP2x is functional. Depletion of GFP–PBP2x resulted in severe morphological alterations, confirming the essentiality of PBP2x and demonstrating that PBP2x is required for cell division and not for cell elongation. A genetically or antibiotic inactivated GFP–PBP2x still localized at septal sites. Remarkably, the same was true for a GFP–PBP2x derivative containing a deletion of the central transpeptidase domain, although only in the absence of the protease/chaperone HtrA. Thus localization is independent of the catalytic transpeptidase domain but requires the C-terminal PASTA domains, identifying HtrA as targeting GFP–PBP2x derivatives. Finally, PBP2x was positioned at the septum similar to PBP1a and the PASTA domain containing StkP protein, confirming that PBP2x is a key element of the divisome complex.
Most bacteria contain a peptidoglycan (PG) layer in their cell wall, a unique, gigantic macromolecule named the sacculus (Weidel and Pelzer, 1964). It provides mechanical stability and reflects the shape of the bacterium. Cell division requires PG synthesis at septal sites, and thus the identification of components involved in PG biosynthesis and their regulation relates to a central question in microbiology: bacterial growth and division.
Cell division is governed by the localized assembly of proteins in a hierarchical manner to form a complex interactive network (reviewed in Sun and Jiang, 2011; Lovering et al., 2012; Egan and Vollmer, 2013). The key element of the division apparatus is FtsZ, a tubulin-like GTPase that assembles early at future division sites, forming a structure known as the Z-ring at mid-cell (Lutkenhaus et al., 2012). It orchestrates the recruitment of ‘early’ division proteins as part of the divisome, most of which are present in Gram-positive as well as in Gram-negative bacteria as a prerequisite for co-ordinated segregation of the chromosomes and invagination of the cytoplasmic membrane (for reviews, see Sun and Jiang, 2011; Lovering et al., 2012; Egan and Vollmer, 2013). At a later stage, immediately before constriction of the septum, other proteins associate with the divisome, including important enzymes required for septal PG synthesis such as the Escherichia coli class B penicillin-binding protein (PBP) PBP3 (FtsI) (Weiss et al., 1997). A distinct machinery, named the elongasome, ensures elongation in rod-shaped bacteria such as Bacillus subtilis and E. coli (for review, see Typas et al., 2012). It is directed by the scaffolding protein MreB, an actin-like protein that assembles on the periphery of the membrane (Den Blaauwen et al., 2008; White and Gober, 2012), and other accessory proteins, including class B PBPs, e.g. the E. coli PBP2, are involved in this process (Den Blaauwen et al., 2003; Egan and Vollmer, 2013).
Bacteria that grow as cocci or ovococci contain some features distinct from rod shaped bacteria (reviewed in Massidda et al., 2013). They lack mreB which might be the explanation for the absence of the cylindrical cell wall in these organisms. Early models predicted that spherical cocci have no peripheral machines, whereas ovococci have two systems localized at mid-cell that allow peripheral and septal PG synthesis in analogy to the lateral sidewall and septal PG synthesis in rods (Lleo et al., 1990; Sham et al., 2012). Evidence for these models has been supported by several studies (Land and Winkler, 2011; Perez-Nunez et al., 2011; Berg et al., 2013). Streptococcus pneumoniae, a major human pathogen, has become a model organism for cell division in ovococci (Massidda et al., 2013).
S. pneumoniae contains altogether six PBPs involved in PG assembly: the class A high molecular mass (hmm) PBPs 1a, 1b and 2a, two class B hmm PBPs PBP2x and 2b, and the low molecular mass (lmm) PBP3 (for review, see Hakenbeck et al., 2012). PBPs are assigned to three main classes according to their homology and function, and each bacterial species contains a distinct set of these multidomain proteins (for reviews, see Goffin and Ghuysen, 1998; Zapun et al., 2008a). Common to all PBPs is a transpeptidase/penicillin binding domain with the active site serine, the acylation site of beta-lactam antibiotics. Class A hmm PBPs contain an N-terminal transglycosylase domain, whereas the function of the N-terminal domain of class B hmm PBPs remains unknown. Lmm PBPs act mainly as D,D-carboxypeptidases rather than transpeptidases, removing the terminal D-alanine from muropeptides and thus potentially controlling the degree of murein crosslinkage.
In S. pneumoniae, PG synthesis mainly takes place at mid-cell regions, and components of both, peripheral and septal PG synthesis, are localized to the septal region (reviewed in Massidda et al., 2013). All hmm PBPs of S. pneumoniae localize to mid-cell region of dividing cells (Zapun et al., 2008b). In contrast, the lmm PBP3 appears to be evenly distributed over the entire surface (Hakenbeck et al., 1993; Morlot et al., 2004; Barendt et al., 2011). However, whereas its absence at the division septum at the start of cell division was reported earlier (Morlot et al., 2004) this could not be confirmed in a recent study (Barendt et al., 2011).
The essential S. pneumoniae PBP2b and PBP2x are homologues of E. coli PBP2 and PBP3, and are accordingly believed to be responsible for peripheral and septal PG synthesis, respectively (Zapun et al., 2008b; Land and Winkler, 2011; Perez-Nunez et al., 2011; Sham et al., 2012; Berg et al., 2013). In fact, depletion of PBP2b results in compressed cells arranged in long chains, whereas misshaped, elongated cells were caused by PBP2x depletion (Berg et al., 2013). The role of PBP1a, whose importance for PG synthesis can be deduced from its central role for the high level penicillin resistance phenotype, however, is less clear. Its participation in peripheral PG synthesis has been supported by genetic studies showing that MreCD, components of the peripheral machinery, regulate the activity or location of PBP1a at constricting septa (Land and Winkler, 2011). Recent data show separate positioning of PBP1a and PBP2x in constricting septa consistent with two separable PG synthesis machines, but a role of PBP1a in septal synthesis could not be ruled out (Land et al., 2013).
PBP2x has been in the focus of many studies since it represents a primary resistance determinant for beta-lactam antibiotics, i.e. PBP2x mutations mediate resistance and are a prerequisite for high-level resistance that requires also mutations in PBP2b and/or PBP1a in clinical isolates (reviewed in Hakenbeck et al., 2012). Moreover, its crystal structure is available at high resolution (Pares et al., 1996; Gordon et al., 2000). In addition to the central transpeptidase domain and an N-terminal extension of unknown function, it contains two C-terminal PASTA domains (for penicillin-binding protein and serine/threonine kinase associated; Yeats et al., 2002). Interestingly, the X-ray structure of an acylated PBP2x revealed two cefuroxime molecules in the protein: one covalently bound to the active site serine as expected, and another beta-lactam molecule was non-covalently associated and located in the cleft between the TP domain and the C-terminal extension (Gordon et al., 2000). It was recently shown that the PASTA domains of PBP2x are essential for the enzymatic interaction with beta-lactam antibiotics (Maurer et al., 2012).
PASTA domains are also present in the S. pneumoniae serine-threonine kinase StkP, which plays an important role in regulating septal cell wall synthesis by a yet unknown mechanism (Giefing et al., 2010; Beilharz et al., 2012; Fleurie et al., 2012). The PASTA domains of StkP and other bacterial serine-threonine kinases interact with muropeptides and beta-lactams (Shah et al., 2008; Maestro et al., 2011; Mir et al., 2011; Squeglia et al., 2011). Structural analysis of the PASTA domains of Staphylococcus aureus Stk1 confirmed their general topology as observed in PBP2x despite their weak sequence conservation, but revealed some differences that might allow the interaction with chemically distinct PG moieties (Paracuellos et al., 2010). In the S. pneumoniae StkP, PASTA domains are important for mid-cell localization with the FtsZ ring (Giefing et al., 2010; Beilharz et al., 2012). Moreover, StkP and PBP2x have recently been shown to interact with each other and to be present in the same membrane-associated complex (Morlot et al., 2013). The precise role of the PASTA domains of PBP2x, however, remains to be clarified.
The aim of this study was to investigate the localization of various GFP–PBP2x derivatives in living cells, and to identify domains that are critical for positioning at the septum with emphasis on the role of the PASTA domains in this process. Phenotypes associated with depletion of PBP2x in the cell are presented, and the timely resolution of its localization compared with PBP1a and StkP was investigated. Since several derivatives of GFP–PBP2x appeared to be degraded we furthermore analysed whether the cell wall associated serine protease/chaperone HtrA is responsible for this process.
Construction of a GFP–PBP2x fusion protein
In order to visualize the localization pattern of PBP2x in living S. pneumoniae cells, an N-terminal GFP–PBP2x fusion was constructed using the vector pJWV25 (Fig. 1A). This vector contains the fast folding GFP+ gene which is under control of a zinc-inducible promoter PZn (Eberhardt et al., 2009) and cannot replicate in S. pneumoniae. It contains a tetracycline resistance cassette and the desired gfp+ fusion construct is flanked by spr0564 and spr0565 (bgaA) sequences to ensure integration at the bgaA locus under tetracycline selection. It should be noted that the concentration of zinc present in the C+Y medium does not suffice to induce the promoter; expression of GFP–PBP2x requires the addition of ZnCl2 to the growth medium (Fig. 1B).
The PBP2x gene lacking the first ATG codon was fused to the linker placed 3′ of gfp+, resulting in plasmid pFP09. Thus GFP should be located in the cytoplasm, while the main part of PBP2x is extracytoplasmic (Fig. 1A). pFP09 was transformed into the unencapsulated S. pneumoniae R6 strain and tetracycline resistant transformants were readily obtained. These transformants thus are merodiploid and contain two copies of pbp2x: the genomic PBP2x gene which is expressed constitutively (our own unpublished experiments), and the zinc-inducible ectopic gfp+–pbp2x. Control PCR and DNA sequencing analysis confirmed the correct integration and DNA sequence of pbp2x; one transformant (DKL03) was used for all further experiments.
The GFP–PBP2x fusion protein is active in beta-lactam binding and localizes to division sites
For expression of GFP–PBP2x, 0.15 mM ZnCl2 was used throughout, based on previous experiences for the expression of membrane and cytoplasmic proteins involved in choline utilization (Eberhardt et al., 2009); also this concentration has been used recently for pbp2x expression in a similar construct (Morlot et al., 2013). Under these conditions, the amount of the fusion protein was present in slightly higher amounts (approximately 135%) compared with wild type PBP2x when tested by Western blot using anti-PBP2x antiserum (Fig. S1). With 0.05 or 0.10 mM ZnCl2, the fusion protein was present in lesser amounts (60–85%), and with 0.20 mM ZnCl2 the amount was clearly higher (> 150%) (Fig. S1).
When expression of the GFP–PBP2x fusion protein in strain DKL03 was induced, no effects on cellular growth or cell morphology were detected (data not shown). The presence and penicillin-binding activity of GFP–PBP2x in the cells was tested by Western blotting using anti-PBP2x and anti-GFP antibodies, and BocillinTMFL labelling (for details see Experimental procedures). DKL03 cells were grown with and without ZnCl2. In both samples, the typical PBP-profile was detected, and the GFP–PBP2x fusion protein was present only in cells grown in the presence of ZnCl2 (Fig. 1B). This documents that the fusion protein is functional in terms of beta-lactam binding. Western blot analysis of cell lysates using purified anti-PBP2x and anti-GFP antibodies confirmed presence and predicted size of the GFP–PBP2x fusion protein upon induction (110.6 kDa), whereas no GFP–PBP2x band was visible in lysates of the negative control grown in the absence of ZnCl2 (Fig. 1B). No major lower molecular weight peptides were detected with anti-GFP antibodies, showing that the expressed fusion protein was not significantly degraded in the cell.
We then determined the localization of GFP–PBP2x fusion protein at different stages of the cell cycle in exponentially growing cells (Fig. 1C). Cells were assigned to three different classes according to the progression of the cell cycle: predivisional single cells or oval cells without apparent division septum (Fig. 1C, columns 1 and 2), early divisional cells showing a slight constriction in the middle (Fig. 1C, columns 3 and 4), and bacteria with a clear invagination of the division septum (late divisional cells; Fig. 1C, column 5). In predivisional cells (48%), the fluorescent signal was localized at mid-cell, first seen as two dots representing septal ring structure, progressing to a central line corresponding to the growing division septum. In the early divisional cells (34%) the fusion protein was present at the septum. In a small percentage of cells (4%), the GFP–PBP2x signal was already found at the mid-cell of daughter cells while still present at the old division septum (Fig. 1C, column 5). In the remaining cells (14%), mixed patterns were observed which were not further specified (not shown). This pattern was observed independent on the concentration of ZnCl2 in the growth medium (0.05–0.20 mM), showing that localization is independent on the amount of PBP2x (not shown).
Previously, it was reported that the capsule of S. pneumoniae influences cell shape and cell division (Barendt et al., 2009). To test whether or not the capsule influences localization of PBP2x, we also examined the fluorescence signal across lateral and septal sites in the encapsulated strain D39 versus R6. As shown in Fig. S2, GFP–PBP2x clearly localized at septal sites in both cases. However, the lateral signal was much stronger in the R6 background compared with that in D39 (Fig. S2). Differences between D39 and R6 in respect to cell morphology and cell wall composition have been described, but the basis for this difference is not clear yet (Barendt et al., 2009). R6 has monomeric stem peptides with Ala-Ser or Ala-Ala attached to L-Lys which are not found back in D39, and it is possible that this has an impact also on localization of PBP2x. Taken together, these data are in agreement with early immunofluorescence results in that PBP2x localizes at the division septum (Morlot et al., 2003). Most importantly, they indicate that the transpeptidase/penicillin-binding domain of the GFP–PBP2x fusion protein is functional.
A conditional mutant of pbp2x
In order to investigate the role of PBP2x in cell morphology, a mutant was constructed containing a conditional pbp2x. It was shown already by Eberhardt et al. (2009) that the Zn2+ inducible system can be efficiently used for fine control of gfp fusion gene expression and for protein depletion experiments in S. pneumoniae. Since the experiments outlined above documented the functionality of the GFP–PBP2x fusion protein, we used strain DKL03 containing the ectopic gfp–pbp2x under the control of the zinc inducible promoter and replaced the genomic pbp2x with a spectinomycin resistance cassette using the plasmid p2xKO (see Experimental procedures) so that only GFP–PBP2x can be produced. Transformants were selected on agar plates containing spectinomycin and ZnCl2 at a final concentration of 0.15 mM and obtained at high frequency but only in the presence of Zn2+. One transformant DKL031 was used for all further experiments.
Functionality of the GFP–PBP2x fusion protein in the absence of the genomic wild type copy was confirmed by the facts that DKL031 did not grow on agar plates lacking Zn2+, whereas in the presence of 0.15 mM ZnCl2 in solid media the strain grew well (Fig. S3). Furthermore, generation time and cell morphology of DKL031 in liquid medium with ZnCl2 was indistinguishable from that of the parental strain DKL03.
GFP–PBP2x was monitored in DKL031 using cells grown at 30°C to late exponential growth phase [N = 70–80; given in nephelometry units (N)]. Binding to BocillinTMFL, expression (Fig. S4A) and localization pattern of the GFP–PBP2x fusion protein (Figs S4B and S5) in DKL031 were identical compared with the parental strain DKL03. It should be noted that a very thin band corresponding to GFP–PBP2x protein was detected in cell lysates prepared from cultures grown in the absence of ZnCl2. This probably represents some residual GFP–PBP2x originating from the starting culture grown in liquid medium with ZnCl2, suggesting slow depletion of the zinc pool or of PBP2x.
These results show that the GFP–PBP2x fusion is fully functional and can replace the wild type PBP2x.
Depletion of GFP–PBP2x results in severe growth and morphological defects
With the functional zinc-inducible GFP–PBP2x fusion in hand, we were now in a unique position to study the effects of the absence of the essential PBP2x protein on cell growth, cell division and cell shape. GFP–PBP2x was depleted in DKL031 during growth in liquid medium without the addition of ZnCl2 and followed by phase-contrast microscopy and Western blotting. Cells of an exponentially growing starting culture in medium plus ZnCl2 were carefully washed and resuspended in the same volume of C+Y medium without additional ZnCl2 (see Experimental procedures) and diluted 1:20 in the same medium. Under these conditions, cells grew with the same generation time as the control culture with ZnCl2, confirming slow depletion of GFP–PBP2x (Fig. 2A). However, slight morphological changes were detectable after five generations (180 min) at the end of the exponential growth phase, where cells became more and more elongated: the cell length increased from 1.04 μm ± 0.15 (R6 strain; > 500 cells counted, ± indicates the standard deviation) to 1.23 μm ± 0.14 in DKL031. During the next hour cells elongated even more to 1.39 μm ± 0.33 and the cell shape deteriorated continuously, resulting in highly heterogeneous morphologies, including cells with bulges, ballooned shaped, elongated and ghost cells (Fig. 2B). Western blotting showed that already after 120 min the amount of the fusion protein was approximately 25% compared with control cells (Fig. 2C) confirming efficient depletion of GFP–PBP2x.
To enforce complete depletion of GFP–PBP2x, cells were again diluted 1:20 at the end of the exponential growth phase, and divided into two samples. Cells continued to grow only for approximately one generation, then they became more elongated and finally lysed (Fig. 2A and B). The addition of ZnCl2 to one sample at the time when growth defects became apparent resulted in complete rescue of the cells: after a lag phase of 1.5 h they resumed growth and after 3.5 h cell morphology showed signs of normalization (Fig. 2B). Nevertheless, after four generations in the presence of inducer, the cell morphology was still not completely restored.
To assess the importance of PBP2x for pneumococcal PG synthesis, DKL031 cells were also stained with fluorescently labelled vancomycin (Van-FL) which targets the D-Ala-D-Ala moieties of the newly synthetized PG (Daniel and Errington, 2003). As expected, wild type R6 cells showed a mid-cell staining pattern (Fig. 2D, time point 3). Surprisingly, the DKL031 cells showed Van-FL signals at the septum only in the first two samples after depletion of GFP–PBP2x (N = 40 and N = 70), and at later time points fluorescence was very low (Fig. 2D, time point 6) or not detectable any more in the microscope (Fig. 2D, time point 7). In parallel, the culture of DKL031 to which Van-FL had not been added, was used to determine GFP–PBP2x localization. The GFP–PBP2x fluorescence signal was not detectable by microscopy in the zinc-depleted culture even at the first time point at the N = 40, confirming that the fluorescence signal seen in the Van-FL treated cells was only due to this compound.
Taken together the results clearly show that PBP2x is required for the formation of the division septum, and that without PBP2x neither cell division nor proper morphology can be maintained while the availability of uncross-linked muropentapeptides apparently discontinued when targeted by the fluorescently labelled vancomycin derivative.
PBP2x localization does not require a functional transpeptidase domain
To determine whether an active transpeptidase domain is required for localization of the GFP–PBP2x fusion protein, the active site serine S337 was mutated to A337 in the GFP fusion protein (see Supporting information). Such constructs were readily obtained (DKL13; R6-background); no impact on generation time was observed. Also the morphology was investigated in 200 cells each of DKL13 and R6 grown to N = 120. DKL13 showed the same length and width (0.96 μm ± 0.08 and 0.75 μm ± 0.07) compared with the parental strain R6 (0.95 μm ± 0.07 and 0.74 μm ± 0.07). This indicates that an enzymatic inactive copy of GFP–PBP2x is well tolerated in presence of wild type PBP2x. In this context one should note that cell size of the control R6 strain varies slightly between different experiments due to slight day-to-day variations of the growth medium due to its complexity.
To confirm that GFP–PBP2xS337A is enzymatically inactive, binding to BocillinTMFL was examined by fluorography. As shown in the Fig. 3A, no signal was obtained in DKL13 at the position of the GFP–PBP2xS337A fusion protein, whereas it was clearly visible in the control strain DKL03. In both, DKL13 and the control strain, full-length GFP–PBP2x fusion proteins were detected on Western blots using anti-PBP2x antibodies in equal amounts in addition to the wild type PBP2x, confirming the production of the inactive mutant protein (Fig. 3A).
Fluorescence microscopy showed that the GFP–PBP2xS337A fusion protein was localized at the division site with the same distribution pattern as the parental strain DKL03 (Fig. 3B). We cannot exclude the possibility that the GFP–PBP2xS337A fusion protein localizes to mid-cell by a protein-protein interaction with wild type PBP2x or with another divisome protein, but these experiments at least are the first indications that a functional transpeptidase domain is not required for PBP2x localization.
Inactivation of the transpeptidase function can also be achieved by treatment with beta-lactam antibiotics. We chose cefotaxime (CTX) since PBP2x is the primary CTX resistance determinant while PBP2b, which is part of the peripheral PG synthesis machinery (Berg et al., 2013), is not targeted by this antibiotic; moreover CTX induces a tolerant response and the cells do not lyse for many hours (Hakenbeck et al., 1987) (Fig. S6; for details see Experimental procedures). Various concentrations of the antibiotic were added to growing cultures; after one hour growth slowed down (Fig. S6A). Saturation of PBPs by CTX was followed by secondary labelling with BocillinTMFL. As shown in Fig. S6B, already after 5 min of treatment with 0.04 μg ml−1 of CTX (2× MIC), bands corresponding to GFP–PBP2x and PBP3 were not visible on the gel, and at 10× MIC and higher all PBPs except PBP2b could not be labelled with BocillinTMFL and thus were inhibited by CTX (Fig. S6B). The localization of GFP–PBP2x was examined after one and two hours.
After 1 h at concentrations up to 10× MIC, approximately 80% of the cells showed septal localization of GFP–PBP2x (Fig. S6C). Even at 30× MIC, this was still true for 60% of the cells (Table S3). Elongated cells with multiple septa were also observed, as much as 8.8% at 2× MIC (Table S3), and an example is shown in Fig. S6C. When DKL031 cultures were treated with 20× MIC of oxacillin (1.6 μg ml−1), which targets PBP2b as well, the fluorescence signal still localized at the septum to 37% of the cells (data not shown). This shows that septal localization of PBP2x is independent not only of its function, but also of other functional PBPs as well.
Although these experiments clearly show that the function of PBP2x is not required for septal localization, they do not exclude the possibility that the presence of the protein rather than its function is essential. Therefore, we tried to delete the native pbp2x gene in DKL13 using the plasmid p2xKO. However, the transformation efficiency was very low indicating that the deletion of pbp2x in this genetic background is not possible. Sequencing of the few transformants obtained verified deletion of the wild type pbp2x, but in the GFP–PBP2x A337 was reverted to S337. This result thus confirms that the transpeptidase activity of PBP2x is essential for normal growth of S. pneumoniae.
The PASTA domains of PBP2x are required for proper localization
In addition to beta-lactam binding, the PASTA domains of PBP2x have been implicated to play a role in the interaction with PG (Yeats et al., 2002; Maestro et al., 2011) and thus might mediate the localization of PBP2x at the septum. Furthermore, it was recently shown that the extracellular PASTA domains of the eukaryotic-type serine-threonine kinase StkP govern its localization to division sites by recognition of uncross-linked PG (Beilharz et al., 2012; Fleurie et al., 2012). This begs the question whether septal localization of PBP2x is driven by its PASTA domains. Therefore, a set of N-terminal GFP fusions to PBP2x containing various deletions were constructed (Fig. 4A). In DKL12, the C-terminal part including both PASTA domains but not the linker region are deleted (aa 632–750; GFP–PBP2xOP); in DKL15 the PASTA domains (aa 617–750) are fused to the N-terminal domain by deleting the central transpeptidase domain (aa 266–616; GFP–PBP2xOT), in DKL17 the N-terminal domain was deleted (aa 71–249; GFP–PBP2xON) and DKL19 contains only the two PASTA domains by deletion of the N-terminal domain as well as the penicillin-binding domain (aa 56–616; GFP–PBP2x-NP) (Fig. 4A).
None of the PBP2x fusion constructs could be detected as BocillinTMFL complexes, including GFP–PBP2x-OP (not shown), confirming that the transpeptidase/penicillin-binding domain requires the PASTA domains for beta-lactam binding as shown previously (Maurer et al., 2012). Western blot analysis with anti-PBP2x antibodies demonstrated the presence of PBP2x fusion products in DKL12, DKL15 and DKL17 albeit at significantly lower amounts compared with DKL03 (Fig. 4B). Western blotting with anti-GFP antibodies detected peptides with the size of GFP indicating substantial degradation (Fig. 4B). It should be noted that full length GFP–PBP2xOP was perfectly detectable with anti-GFP antibodies but less so with our polyclonal anti-PBP2x antibodies, suggesting that major immunogenic epitopes are located on the PASTA domains. In DKL19, no band of the predicted size of the fusion protein was detected with anti-PBP2x (not shown) or with anti-GFP antibodies (Fig. 4B).
The fluorescence signal was dispersed throughout most cells (DKL12, DKL17 and DKL15) (Fig. 4C). Interestingly, a low number of DKL15 cells (∼ 2%) showed a fluorescence signal at the division septum, and an even higher number in case of DKL17 where 12.6% of the cells showed septal or equatorial localization (see arrows in Fig. 4C). This indicates the possibility that the PASTA domains which are present in these PBP2x-derivatives might indeed be responsible for septal localization.
GFP–PBP2x derivatives are targets for HtrA
Since degradation of the GFP fusion proteins was apparent, a protease might be responsible for this process. A potential candidate is the cell wall associated serine protease/chaperone HtrA, especially since it localizes to the septal region similar to PBP2x (Tsui et al., 2011). Therefore, htrA was deleted in DKL12, DKL15, DKL17, DKL19 and DKL031. Strikingly, upon induction with 0.15 mM ZnCl2, the fusion proteins were present in higher amounts in DKL12ΔhtrA, DKL15ΔhtrA and DKL17ΔhtrA compared with DKL12, DKL15 and DKL17, respectively, and degradation products were substantially reduced especially in the case of DKL15ΔhtrA (Fig. 5A). No fusion product was visible in case of DKL19ΔhtrA (not shown), but the pattern on Western blots with anti-GFP antibodies was identical to that of the parental DKL19 strain (see Fig. 4B), suggesting that degradation and cleavage of the GFP fusion protein occurred intracellularly and independent on HtrA.
Fluorescence microscopy of DKL12ΔhtrA cells showed no localization but a distribution in irregular patches throughout the cells (Fig. 5B). In case of DKL17 no significant difference was noted if the HtrA gene was deleted or not in terms of septal localization (17% versus 13%); however fluorescence was confined to irregular, large patches on the surface of the cells (Fig. 5B) in contrast to cytoplasmic localization in the presence of HtrA (see Fig. 4C). Most importantly, however, was the fact that 66% of DKL15ΔhtrA cells showed fluorescence signals at the septum (Fig. 5B and C) in contrast to 2% in DKL15 (Fig. 4C). This clearly indicates that the PASTA domains govern localization of PBP2x-derivatives to the division zone independent on the presence of the transpeptidase domain.
In Bacillus subtilis, htrA is regulated by the two component system (TCS) CssRS together with htrB, both of which encode cell envelope-associated quality control proteases HtrA and HtrB responsible for degradation of misfolded proteins (Hyyrylainen et al., 2001; Noone et al., 2012). Secretion stress induces the expression of HtrA and HtrB genes via activation of the CssRS system. The S. pneumoniae HtrA gene is regulated by the TCS CiaRH (Halfmann et al., 2007b). Therefore we tested whether also in S. pneumoniae HtrA is induced and thus present in higher amounts in cells expressing GFP–PBP2x fusion proteins lacking various domains, using Western blot analysis and anti-HtrA antibodies. However, in none of the strains DKL12, DKL15, DKL17 or DKL19, an increase of HtrA was detected in the presence of ZnCl2 compared with growth in medium without of ZnCl2 (data not shown), i.e. expression of the GFP–PBP2x fusions did not result in stimulation of the CiaRH system. In this context it should be noted that already under these conditions which are routinely used in our laboratory the CiaR regulon is highly expressed (Halfmann et al., 2011). Together with the fact that the fusion proteins fluorescence (and in some cases still bind BocillinTMFL, see below), these data suggest the absence of general protein misfolding and subsequent activation of the secretion stress response by expression of GPF-PBP2x fusion proteins.
C-terminal deletions of GFP–PBP2x
It has recently been shown that deletion of the last 40 amino acids of PASTA2 strongly affects beta-lactam binding at the active site whereas deletion of 30 aa does not (Maurer et al., 2012). Thus, the presence of PASTA1 plus the α-helix within PASTA2 appears to be critical for antibiotic binding. To investigate whether similar structural requirements exist for PBP2x localization, we constructed two additional fusion proteins and deleted the last 30 (aa 721–750; GFP–PBP2x720) and the last 40 residues (aa 711–750; GFP–PBP2x710) resulting in strains DKL20 and DKL21 respectively (Fig. S7). The HtrA gene was deleted in both strains since septal localization could be severely affected by the presence of HtrA as shown above.
BocillinTMFL binding and expression of GFP fusion proteins were tested as described before. As expected, GFP–PBP2x710 was not able to bind BocillinTMFL (Fig. 6A) whereas in strains DKL20 and DKL20ΔhtrA a band corresponding to GFP–PBP2x720 was detected. Western blot analysis with anti-PBP2x antibodies showed that only small amounts of GFP–PBP2x720 were present in the cells, corresponding with the low BocillinTMFL signal relative to that of the control DKL03. Interestingly, with anti-PBP2x antibodies degradation products of truncated fusion proteins were detected in htrA deletion strains that were inactive in terms of BocillinTMFL binding (Fig. 6B). They were not seen on Western blots developed with anti-GFP antibodies in agreement with an extracellular cleavage site. Moreover, in htrA deletion derivatives higher amounts of fusion proteins were clearly visible on Western blots with anti-GFP antibodies in both, DKL21 and DKL20, and degradation products were substantially reduced (Fig. 6C). At the same time, deletion of htrA strongly affected localization of both GFP–PBP2x derivatives. The percentage of cells showing fluorescence signals at septal sites increased in case of DKL21 from 3% with htrA to 23% in the htrA deletion mutant, and in case of DKL20 from 12% (htrA) to 28.5% (ΔhtrA) (Fig. 6D). We did not observe any septal localization in strain DKL12 (no PASTA domains) (Fig. 4C). In other words, the C-terminal PASTA domain is necessary for septal localization, but not sufficient for full localization and other PBP2x functions such as BocillinTMFL binding.
PBP2x – a substrate for HtrA?
The previous experiments documented that HtrA acts upon GFP–PBP2x fusion proteins lacking various domains. Therefore we asked the question whether HtrA also targets full length GFP–PBP2x or wild type PBP2x. First, a deletion of htrA was constructed in DKL031 (Δpbp2x, PZn–gfp–pbp2x). Western blot analysis using an anti-GFP antibody (see Fig. 5A) as well as anti-PBP2x antibodies (data not shown) showed no major differences between DKL031 and DKL031ΔhtrA. Since growth is a more sensitive measure of the amount of an essential protein in cells, depletion assays were performed (for details see Experimental procedures) (Fig. 7A). Both strains grew equally well in the absence of inducer indistinguishable from the wild type strain R6 for approximately 5 h and then entered stationary phase. However, after a second dilution step, DKL031ΔhtrA reached a higher cell density than DKL031, i.e. GFP–PBP2x was less depleted in the absence of HtrA (Fig. 7A). This suggests that HtrA also targets the functional full length GFP–PBP2x fusion.
In order to clarify whether also wild type, untagged PBP2x is an HtrA substrate, a strain was constructed where pbp2x expression is only driven from the zinc-inducible promoter PZn and the genomic PBP2x gene was replaced by a spectinomycin resistance cassette resulting in the conditional mutant DKL41 (see Experimental procedures). Subsequently, htrA was deleted in DKL41 resulting in strain DKL41ΔhtrA.
First, standard conditions (0.15 mM ZnCl2) were used for DKL41 and DKL41ΔhtrA as in the depletion experiment with DKL031 described above. During the first period of depletion, no difference in growth was noted (Fig. 7B). Also after the second dilution step, growth stopped and cells lysed identically in both cultures. Thus, the presence or absence of HtrA had no effect on the amount of wild type PBP2x.
The difference between DKL031 and DKL41 and their derivatives was even more pronounced if less ZnCl2 was used during initial growth (0.09 mM), i.e. conditions where less PBP2x or GFP–PBP2x was present. DKL031 cells grew significantly slower already during the first round of depletion and to a lower cell density compared with DKL031ΔhtrA, whereas again no difference was apparent between DKL41 and DKL41ΔhtrA (Fig. S8).
In summary, these results document that the wild type untagged PBP2x is not a target of HtrA, whereas GFP–PBP2x fusion constructs are. However, only when the fusion proteins contained deletions as described above, degradation products could be detected by Western blot analysis; with full length GFP–PBP2x, some impact on stability or degradation became apparent only in depletion assays.
A GFP–PBP1a fusion protein is functional
PBP2x and PBP1a are localized at the septum in immunostaining experiments like all other PBPs (Morlot et al., 2003; Zapun et al., 2008b). Using GFP fusion proteins it should now be possible to determine not only spatially but also the timely localization of both proteins. Therefore, first the gene encoding the GFP–PBP1a fusion protein was introduced in the bgaA locus and tested similar to the gfp–pbp2x derivatives.
A modified version of vector pJWV25 was used. Since PBP1a is not essential, the plasmid was introduced in the pbp1a-deletion strain DKL2. Transformants were obtained at high frequency, and the transformant DKL23 was used for all further experiments. Western blot analysis confirmed that DKL23 produced a GFP–PBP1a fusion protein with the predicted size, which was active in the beta-lactam binding assay, and the amount of GFP–PBP1a was similar compared with the wild type untagged protein (Fig. 8A and B). Moreover, the deletion of pbp1a in DKL23 as well as in the parental strain DKL2 was verified in these assays. Fluorescence microscopy showed that GFP–PBP1a is localized at the division sites in both the R6 and D39 genetic backgrounds, in a similar fashion as GFP–PBP2x (Figs 8C and S9). The number of R6 cells showing GFP–PBP1a location at septal sites (78% of 272 cells counted) was not significantly different from the number obtained with GFP–PBP2x (86%; see Fig. 1C).
However, although no effect on generation time was noticed in DKL23 (Δpbp1a, PZn–gfp–pbp1a) compared with R6, cells appeared to be slightly smaller. Since it has been reported recently that Δpbp1a D39 mutant cells have smaller diameters compared with the D39 parent (Land and Winkler, 2011), this phenomenon was tested in more detail. Strain DKL23 was grown in the presence and absence of ZnCl2 to mid-exponential growth phase and cells were examined by phase-contrast light microscopy; 200 cells were evaluated for each strain under different growth conditions. Cell length and width of R6 (1.00 μm ± 0.05 and 0.93 μm ± 0.03) were identical to the values obtained with DKL23 grown in the presence of ZnCl2 (1.03 μm ± 0.047 and 0.91 ± 0.04) (Table S2). In contrast, cells of DKL23 grown in the absence of ZnCl2 were significantly smaller (0.91 μm ± 0.09 and 0.73 μm ± 0.08). D39 (Avery et al., 1944) is the progenitor of strain R6 (Smith and Guild, 1979) and differs in several genes as detected by genome sequencing (Lanie et al., 2007). The results here show that pbp1a deletion leads to the same phenotype in both the encapsulated D39 and unencapsulated R6 strain, i.e. smaller cells.
PBP2x and PBP1a colocalize and arrive to mid-cell relatively early during the cell cycle
Septal localization of PBP1a and PBP2x in S. pneumoniae was demonstrated by immunofluorescence before (Zapun et al., 2008b) and now also using functional GFP fusions (Figs 1 and 8). Furthermore, it has been suggested that PBP1a and PBP2x of S. pneumoniae interact with each other (Zerfaß et al., 2009). To establish whether S. pneumoniae PBP2x and PBP1a colocalize, a merodiploid double labelled strain was constructed (KB1–71) utilizing the zinc-inducible GFP–PBP2x construct and a zinc-inducible red fluorescent protein (RFP) fused to PBP1a inserted near the scpB locus (see Supporting information). Line scan analysis of fluorescent signals showed perfect colocalization for 24 out of 31 cells (Fig. S9). In order to determine when they arrive at the cell division sites relative to the cell cycle, we performed time-lapse microscopy. For this experiment, cells were first grown to mid-exponential phase and then transferred to a microscope slide containing medium with 1.5% agarose and 0.1 mM ZnCl2. Images were acquired every 8 min. As shown in Fig. 9A, Fig. S10A and Movie S1, GFP–PBP2x and RFP-PBP1a colocalize at cell septa, and throughout cell division they shift simultaneously from old to new septa. As can be observed from the separate channels, in most of the cells there is a colocalization between GFP–PBP2x and RFP-PBP1a. However, not all signals are equally strong in the overlay panel (thus not resulting in only yellow localizations). This is likely due to heterogeneity in single cell expression strength from the PZn promoter.
Since we showed previously that StkP arrives at mid-cell relatively early within the cell cycle but slightly after FtsA, and dependent on uncrosslinked PG (Beilharz et al., 2012), we colocalized RFP-PBP1a with GFP–StkP. As shown in Fig. 9B, Fig. S10B and Movie S2, PBP1a and StkP colocalized throughout cell division. These results are in line with previous results that have shown that PBP1a and StkP arrive at division sites after FtsZ, but before the late division protein DivIVA (Morlot et al., 2003; Giefing et al., 2010; Beilharz et al., 2012; Fleurie et al., 2012). Together, these results show that PBP2x and PBP1a both colocalize and arrive at mid-cell in a similar fashion to the PASTA domain containing StkP protein.
Cell division represents one of the fundamental processes of life, and many aspects of the complex protein machinery responsible for chromosome segregation and septum formation have been revealed in bacteria (for review, see Daniel and Errington, 2003; Thanbichler, 2010). GFP-tagging has widely enhanced our understanding of protein organization in live cells, and the discovery of the FtsZ ring as a prerequisite for septum formation, i.e. localized synthesis of cell wall material, represents a milestone in this research field (for review, see Adams and Errington, 2009). The mechanisms underlying cell elongation and cell division required for the typical ovococcoid shape of S. pneumoniae, however, remain relatively poorly understood. Here, we focused on the major S. pneumoniae PG transpeptidase, PBP2x.
Making use of different genetic, biochemical and cell biological assays we made three major contributions. First, we unambiguously show that PBP2x is the key PBP of S. pneumoniae responsible for septal cell wall synthesis and cell division (Figs 1 and 2). This finding supports the model that two separate biosynthetic activities exist for septal and peripheral cell wall synthesis in ovococci (Higgins and Shockman, 1970; Lleo et al., 1990; Perez-Nunez et al., 2011). Second, we show that localization of PBP2x does not rely on its transpeptidase activity but depends on its PASTA domains. Since PASTA domains recognize and bind uncrosslinked PG (Maestro et al., 2011), this finding strongly suggests that localization of PBP2x is dependent on the localization of its substrate. In line with this, immunofluorescence showed that PBP2x delocalizes in the absence of the D,D-carboxypeptidase PBP3 (Morlot et al., 2004), i.e. under conditions where PG composition is extensively altered (Severin et al., 1992). Recently, it was shown that PBP2A and PbpH of B. subtilis, PBPs responsible for cell elongation, delocalize from mid-cell upon the addition of the LipidII-sequestering lantibiotic nisin (Hasper et al., 2006), supporting the idea that substrate availability also guides PG synthesis by these PBPs (Lages et al., 2013). Thirdly, we have now identified HtrA as a protease targeting GFP–PBP2x derivatives (Figs 5 and 6). Interestingly, htrA is part of the CiaR/H two-component regulatory system which is involved in several stresses including cell wall stress, and the system affects β-lactam susceptibility and autolysis (Mascher et al., 2006). We could not detect degradation of wild type PBP2x by HtrA and only very little of GFP–PBP2x whereas substantial degradation was observed for GFP–PBP2x derivatives containing various deletions. This suggests that these truncated GFP fusions are partly misfolded and thereby targeted by HtrA. Nevertheless, it is tempting to speculate that expression of HtrA via regulation by CiaR ensures the integrity of PBP2x by degrading misfolded derivatives of the protein.
The key experiment that allowed us to draw the above mentioned conclusions was the construction of an inducible GFP–PBP2x fusion that could be used to image, for the first time, the localization and dynamics of PBP2x in live cells. The GFP–PBP2x fusion protein was functional and able to bind beta-lactams (Fig. 1), and it was possible to delete the native copy of the PBP2x gene in the presence of the ectopic gfp–pbp2x copy (strain DKL031). DKL031 was indistinguishable from the parental R6 strain in the presence of inducer in terms of cell growth and morphology. GFP–PBP2x localizes at the division septa in exponentially growing pneumococcal cells (Fig. 1) independent of the presence of the genomic pbp2x copy (Fig. S4), or of the capsule and genetic background (Fig. S2). This localization pattern is in agreement with results reported previously using immunofluorescence microscopy and specific anti-PBP2x antisera in cells after a complex fixation and staining procedure (Morlot et al., 2003; Zapun et al., 2008b). Meanwhile, similar constructs have been reported and will be discussed below (Berg et al., 2013; Land et al., 2013; Morlot et al., 2013).
Experimental evidence for the vital function of PBP2x could be obtained using the conditional mutant DKL031 where the GFP-tagged PBP2x is under the control of a zinc-inducible promoter. Upon shifting the cells to a medium without zinc, they continued growing for more than five generations before growth came to a halt, followed by cell lysis. Morphological defects became apparent after approximately four generations, accompanied by the disappearance of newly synthesized PG at mid-cell (Fig. 2B and C). First, elongated cells became apparent whose shape changed during the time of depletion into large lemons shortly before the onset of lysis (see Fig. 2B). Such round cells with pointed ends have been reported also for cells exposed to low doses of piperacillin (Williamson et al., 1980) which targets PBP2x primarily (Laible and Hakenbeck, 1987), and recently in similar PBP2x depletion experiments (Berg et al., 2013). On the other hand, elongated cells were observed upon methicillin addition at concentrations that inhibitis PBP2x but only inhibits marginally PBP3 and other PBPs (Land et al., 2013). Given the fact that only a certain fraction of the cells show such morphological alterations, and that morphology changes over time and depends on the antibiotic concentration used (see Table S3), it is impossible to compare experiments from different references. Nevertheless, all these reports are in agreement with PBP2x carrying out septal cell wall synthesis. PBP2b on the other hand is likely responsible for peripheral cell wall synthesis, and depletion of PBP2b resulted in compressed cells which were arranged in chains (Berg et al., 2013).
Somewhat puzzling although interesting is the fact that D-Ala-D-Ala containing muropeptides apparently disappear upon depletion of GFP–PBP2x as monitored by Van-FL (see Fig. 2D). One explanation could be that the activity of PBP3, a D,D-carboxypeptidase, is responsible for this phenomenon. However, no changes in the muropeptide pattern in PG isolated from PBP2x-depleted cells were apparent (Berg et al., 2013). Another possibility could be that uncross-linked muropentapeptides are masked by some component and thus cannot be targeted by the vancomycin-derivative any more when PBP2x is depleted, or that due to the defect in cell wall synthesis the cells are stressed and shut down the cytoplasmic synthesis of lipid-2 (and pentapeptide) detected by the fluorescent vancomycin. We cannot distinguish between the muropentapeptides in the lipid-II moiety versus those incorporated in nascent PG by transglycosylase activity at this point, and future studies are required to solve this puzzle.
Taken together, our data verified the assumptions that PBP2x is an essential protein which were based on the failure to isolate deletion mutants (Kell et al., 1993). Interestingly, an inactive GFP–PBP2x derivative containing a mutation in the active S337A (strain DKL13) localized like wild type GFP–PBP2x (Fig. 3B) as did GFP–PBP2x-beta lactam complexes (Fig. S6C and Table S3), showing that an active transpeptidase domain is not required for proper localization. Nevertheless, deletion of the genomic wild type gene was not possible in DKL13, documenting that it is the transpeptidase activity of PBP2x that is responsible for its essentiality and not a function as scaffold for other divisome proteins.
The role of the C-terminal PASTA domains of PBP2x has remained somewhat enigmatic. Curiously, the X-ray structure of an acylated PBP2x revealed two cefuroxime molecules in the protein: in addition to the one covalently bound to the active site serine, another antibiotic molecule was non-covalently associated and located in the cleft between the TP domain and the C-terminal extension (Gordon et al., 2000). It has therefore been hypothesized that the PASTA domains interact with muropeptides, the substrate molecules of PBPs (Yeats et al., 2002). Whereas this has been verified for several bacterial serine-threonine kinases including StkP of S. pneumoniae (Shah et al., 2008; Maestro et al., 2011; Mir et al., 2011), no experimental evidence has been obtained so far to confirm this conclusion for PBP2x. The importance of the PASTA domains of PBP2x for beta-lactam binding has been shown recently, where deletion of the last 50 amino acid residues resulted in a complete loss of beta-lactam binding activity, whereas a 40 aa deletion reduced binding only slightly (Maurer et al., 2012). We now constructed other PBP2x derivatives in order to investigate whether the PASTA domains of PBP2x are also responsible for its localization as has been documented for the PASTA domain of the S. pneumoniae StkP (Giefing et al., 2010; Beilharz et al., 2012; Fleurie et al., 2012). Four constructs were analysed: DKL17 where the N-terminal domain was deleted, DKL15 lacking the transpeptidase domain, DKL12 with deletion of the PASTA domains, in DKL19 the PASTA domain were fused to the transmembrane region via a linker to GFP (see Fig. 4A). However, such constructs appeared unstable and degraded rapidly. Only in case of DKL17, a relatively high number of cells (12.6%) showed fluorescence signals at septal sites (Fig. 4C). However, upon deletion of the protease/chaperone HtrA degradation products were significantly reduced, and striking changes in localization pattern were observed especially in case of DKL15: fluorescence at septal sites shifted from 2% in the presence of HtrA to 66% without HtrA (Fig. 5B and C). This implies that HtrA is responsible for degradation of the DKL15 GFP–PBP2x-OP, and moreover that it is the PASTA domains that are responsible for localization of PBP2x. Additional constructs revealed that apparently the PASTA1 domain is sufficient for septal localization at least in some cells, but again only in the absence of HtrA the PBP2x derivatives localized at the equators and only in approximately one fourth of the cells (see Fig. 6).
The HtrA family of serine proteases plays a key role in quality control of surface proteins (Clausen et al., 2011). S. pneumoniae HtrA localizes to the equators and septa of most dividing cells and this coincides in growing cells with the localization of the translocase subunits SecA and SecY (Tsui et al., 2011). The authors suggested that the Sec translocase directs the PG synthesizing machinery to regions where it is required during cell division, and thus HtrA plays an important role in quality control of proteins exported by the Sec translocase (Tsui et al., 2011). Evidence has been obtained that HtrA targets the competence stimulating peptide CSP (Cassone et al., 2012), and that it is involved in the expression of bacteriocins as well (Kochan and Dawid, 2013). Our data now shows that GFP-tagged PBP2x variants serve as HtrA substrates, while no evidence has been obtained that native PBP2x is a bona fide target for HtrA under the conditions used here (see Fig. 7).
In addition to the spatial distribution of PBP2x, the timing in relation to two other proteins of the divisome complex was studied: PBP1a and StkP. It was shown by immunofluorescence microscopy before that PBP1a localizes to division sites (Morlot et al., 2003), similar to all other hmm PBPs (Zapun et al., 2008b). An interaction between PBP2x and PBP1a has been suggested on the basis of the phenotypic expression of resistance, where only certain variants of the two proteins resulted in high level resistance, whereas other combinations did not (Zerfaß et al., 2009). The GFP–PBP1a fusion protein appeared to be functional similar to the PBP2x derivatives in terms of beta-lactam binding and localization (Fig. 8). The cells tolerated depletion of PBP1a since deletion mutants of pbp1a are viable (Hoskins et al., 1999; Paik et al., 1999), but were smaller in size, similar to data obtained with the encapsulated ancestor strain of R6, the serotype 2 D39 strain (Land and Winkler, 2011). Both GFP–PBP2x and RFP-PBP1a colocalized at division sites in most cells (Figs 9A and S9A). Interestingly, recent results documented that although PBP2x and PBP1a share a similar overall pattern of localization at septal sites, they occupy different positions in constricting division septa suggesting that they function in different PG synthesizing machines (Land et al., 2013).
StkP has been shown to be recruited to division sites after the formation of the FtsZ ring (Giefing et al., 2010) and after FtsA, but before DivIVA (Beilharz et al., 2012) similar to PBP1a (Morlot et al., 2003). Time-lapse experiments with RFP-PBP1a and GFP–PBP2x or GFP–StkP now revealed that in fact all three proteins associate with septal sites approximately simultaneously: PBP2x and PBP1a colocalized, as did PBP1a and StkP (Figs 8 and 9). This is in agreement with recent results obtained by Morlot et al. that suggested that both SktP and PBP2x are present in the same membrane-associated complex in S. pneumoniae (Morlot et al., 2013). The authors also reported that localization of PBP2x at septal sites was dependent on the presence of StkP based on weak interactions of PBP2x with the PASTA domains of StkP (Morlot et al., 2013). Clearly, further experiments are required to unravel the complexity of apparent interactions of PBP2x with other components at septal sites. It will also be interesting to see whether StkP controls the activity of PBP1a and PBP2x.
Bacterial strains, plasmids and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. All S. pneumoniae strains in this work are derivatives of S. pneumoniae R6, a non-encapsulated derivative of the Rockefeller University strain R36A (Ottolenghi and Hotchkiss, 1962), or the encapsulated D39 strain (Avery et al., 1944). Bacteria were grown at 37°C without aeration in C medium (Lacks and Hotchkiss, 1960) supplemented with 0.1% yeast extract (C+Y medium) or on D-agar plates (Alloing et al., 1996) supplemented with 3% defibrinated sheep blood. For microscopic examination, cells were grown at 30°C to ensure complete folding of GFP as recommended (Eberhardt et al., 2009). Growth in liquid culture was monitored by nephelometry and is given in nephelometry units (N). For induction of PZn present in pJWV25 derived strains, ZnCl2 was added to liquid medium and agar plates at a final concentration of 0.15 mM. For particular depletion experiments the concentration of ZnCl2 0.09 mM was used.
pJWV25 derivative, SpeI restriction site was converted into AgeI
Escherichia coli strain DH5α was used as a host for cloning and propagation of plasmids. E. coli strains were grown aerobically at 37°C either in LB medium or on LB agar plates (Sambrook et al., 1989). Growth of E. coli was followed by measuring of the optical density at 600 nm using a spectrophotometer.
Transformation of S. pneumoniae strains was carried out as described previously (Mascher et al., 2006) and transformation efficiency was determined using chromosomal DNA of S. pneumoniae AmiA9 conferring resistance to streptomycin (Salles et al., 1992). Transformation efficiency was calculated as the percentage of transformants (CFU) compared with CFU on control plates without selective antibiotic. Antibiotic resistance genes used for chromosomal integrations in S. pneumoniae were selected either with 200 μg ml−1 kanamycin (Kan), 200 μg ml−1 streptomycin (Str), 2.5 μg ml−1 tetracyclin (Tet) or 20 μg ml−1 spectinomycin (Spc). E. coli DH5α was transformed by using chemically competent cells (Sambrook et al., 1989) and transformants were selected in the presence of 100 μg ml−1 ampicillin or tetracycline 20 μg ml−1.
DNA manipulations and oligonucleotides
All DNA manipulations were performed using standard methods (Sambrook et al., 1989). Chromosomal DNA was isolated from S. pneumoniae as described earlier (Laible et al., 1989) and plasmid purification from E. coli was done according to the manufacturer's protocol using the QIAprep Spin Miniprep kit (Qiagen). Restriction enzymes and T4 DNA ligase were purchased from Fermentas or New England Biolabs and used as described by the manufacturer. PCR products and DNA recovered after restriction endonuclease digestions were purified using a JETquick DNA Purification kit (Genomed). DNA fragments from agarose gels were purified by using innuPREP Gel Extraction Kit (Analytik Jena). PCRs were performed using either Goldstar Red Taq polymerase (Eurogentec), DreamTaqTM-Polymerase (Fermentas), Pfu-Polymerase (New England BioLabs) or iProof high fidelity DNA polymerase (Bio-Rad Laboratories) according to the manufacturer's instructions. The oligonucleotides used in this study are listed in Table S1 and were obtained from Eurofins MWG Operon. The inserts of recombinant plasmids were sequenced to confirm the insertion and subsequently used to transform pneumococcal strains. Plasmid integration into pneumococcal chromosome was verified by PCR amplification and sequencing. The sequences were analysed by using CloneManager Suite 7and Chromas Lite 2.01 software.
Construction and verification of mutants and plasmids
Construction of pJWV25 based plasmids
To construct the plasmid pFP09 carrying the gfp+ variant (Scholz et al., 2000) fused to S. pneumoniae PBP2x gene under the control of the zinc-inducible czcD promoter (PZn), a PCR with the primers pbp2x_gfp_f and pbp2x_gfp_r was performed, using iProof high fidelity DNA polymerase and genomic DNA of R6 strain as a template. The amplified pbp2x fragment was digested with NotI and SpeI and subcloned into the same sites in the vector pJWV25 (Eberhardt et al., 2009), generating the plasmid pFP09. Derivatives of pFP09 used for the construction of various GFP–PBP2x derivatives are constructed as outlined in the Supporting information.
To construct plasmid pFP13 which expresses pbp2x from the Zn2+-inducible promoter PZn, the plasmid pFP092 was digested with SphI and SnaBI and the resulting 9092 bp fragment purified from the agarose gel. Two fragments were amplified using the primer pair Pz2x-pcr1-f/Pz2x-pcr1-r and Pz2x-pcr2-f/Pz2x-pcr2-r from plasmid pFP092. These two fragments were used as templates in a fusion PCR with primers Pz2x-pcr1-f/ Pz2x-pcr2-r. The resulting PCR product was digested SphI and SnaBI, ligated with 9092 bp fragment and transformed into E. coli DH5α with ampicillin selection. Transformants were screened by colony PCR using primers Pzn2x-screen-f and Pzn2x-screen-r.
To generate gfp fusion of pbp1a, the cloning site SpeI (ACTAGT) in the vector pJWV25 was altered into AgeI (ACCGGT) since pbp1a contains a SpeI restriction site in the coding region. The restriction site AgeI was introduced by QuickChange SDM using primers pJWV-f and pJWV-r and plasmid pJWV25 DNA as template. The amplified PCR product was purified, subsequently treated with DpnI to digest plasmid DNA, and transformed into E. coli. The altered restriction site was verified by sequencing and the plasmid named pFP11.
Construction of S. pneumoniae strains
For construction of the gfp+–pbp2x fusion in the S. pneumoniae R6 background, pbp2x was amplified using primers pbp2x_gfp_f and pbp2x_gfp_r and genomic DNA of the R6 strain as a template and iProof high fidelity DNA polymerase. The pbp2x fragment was purified from agarose gel followed by digestion with NotI and SpeI and ligation into the corresponding sites of plasmid pJWV25. The plasmid was transformed into S. pneumoniae R6 strain under selection on D-agar blood plates containing tetracycline. Correct integration into the bgaA region by double crossover was confirmed by PCR amplification with the oligonucleotides bga_check_F and bga_check_R as well as DNA sequencing. To exclude the possibility that any mutation occurred in the pbp2x gene, the gfp fusion with pbp2x was amplified and sequenced. One transformant, DKL03, which contains the wild type R6 pbp2x was used for all further experiments.
The construction of strains expressing various GFP–PBP2x derivatives is described in the supplement.
R6 pbp1a::ahpIII (DKL2) was created by replacing pbp1a with the kanamycin resistance gene aphIII. The PCR fragment of 2689 bp comprising the kanamycin resistance gene as well as the flanking regions of pbp1a gene for homologous recombination was amplified using primer pair 1a_for/1a_rev using chromosomal DNA of the strain R6pbp2xT388GΔpbp1a::aphIII described previously (Zerfaß, 2011). The purified PCR fragment was introduced into R6 by transformation and selection with kanamycin. Correct deletion of pbp1a was verified by PCR using primer pair KO_1a_for/ KO_1a_rev and sequencing. The generated strain was named DKL2. Briefly, DNA fragments corresponding to the upstream and downstream regions of pbp1a gene were amplified by PCR from R6 strain using primer pairs 1a_for/1a-seq_rev and 1a-seq_for/1a_rev. The primers 1a-KanR_for and 1a-KanR_rev were used to amplify aphIII from the Janus cassette (Sung et al., 2001); iProof high fidelity DNA polymerase was used throughout. The three PCR fragments were purified and joined by overlapping PCR using primer pair 1a_for and 1a_rev, the resulting (2689 bp) purified from agarose and used for transformation of S. pneumoniae R6pbp2xT388G strain. The correct integration into pbp1a locus was verified using primer pair KO-1a_for and KO-1a_rev.
To generate a gfp+-pbp1a fusion in DKL2, pbp1a was amplified using the primer pair Pbp1a_gfp_f/Pbp1a_gfp_r and genomic DNA of strain R6 as template. The amplified 1922 bp fragment was cleaved with AgeI and NotI and ligated into the corresponding sites of plasmid pFP11. The ligation mixture was used to transform S. pneumoniae DKL2 strain, and transformants were selected with tetracycline. The correct integration into bgaA region was confirmed by PCR and the resulting strain was named DKL23.
To delete htrA gene in different strains, a PCR cassette replacing htrA with a kanamycin resistance gene aphIII, was amplified from RKL557 (Schnorpfeil et al., 2013) using the primers htrA_ko_1 and N17. Integration of that fragment into the S. pneumoniae genome was selected by 200 μg ml−1 kanamycin. The correct integration was verified by PCR analysis using the primers htrA_ko_6/spo0J_1 and DNA sequencing.
Construction of D39 derivatives expressing various GFP or RFP-PBP derivatives is described in the Supporting information.
Construction of pbp2x conditional mutant
The pbp2x gene was deleted in the DKL03 and DKL4 genetic background with the help of plasmid p2xKO containing the spectinomycin resistance gene aad9 (LeBlanc et al., 1991) as well as the flanking regions of pbp2x. A 1106 bp upstream fragment covering part of yllC plus ftsL was amplified by PCR using the primers KO2x_pcr1_f and KO2x_pcr1_r and iProof DNA polymerase, the 947 bp downstream fragment including the 5′-region of mraY was amplified by PCR using the primers KO2x_pcr3_f and KO2x_pcr3_r. The upstream and downstream fragments were amplified from S. pneumoniae R6 genomic DNA. A 797 bp fragment containing spectinomycin resistance gene aad9 was amplified from S. pneumoniae RCR1 (Mascher et al., 2003) genomic DNA using the primers KO2x_pcr2_f and KO2x_pcr2_r. The three PCR fragments were purified and joined by overlapping PCR using primers KO2x_f and KO2x_r. The resulting PCR fragment was purified from agarose gel and ligated to plasmid SmaI site of plasmid pUC19 (Fermentas). The constructed plasmid was confirmed by sequencing and named p2xKO.
The plasmid p2xKO was transformed to competent S. pneumoniae DKL03 and DKL4 cells and transformants selected with spectinomycin and ZnCl2 (0.15 mM). The correct integration was verified by PCR using primer pair KO2xR6_kont-f/KO2xR6_kontr_r and sequencing. The conditional mutants of pbp2x were named DKL031 and DKL41 respectively. PCR analysis and sequencing in three transformants each verified that the native pbp2x gene was deleted.
Growth of pbp2x conditional mutants and depletion assays
Pneumococci were grown up to a cell density of N = 70 at 30°C in 6 ml C+Y medium supplemented with 0.15 mM ZnCl2. In order to remove ZnCl2 the cells were twice washed by centrifugation (5 min at room temperature and 6600 g) with warm C+Y medium. The cell sediment was resuspended in 6 ml C+Y medium and diluted 1:20, followed by incubation at 30°C. When cell density reached 70 N, the cells were diluted 1:20 into pre-warmed C+Y medium. ZnCl2 was added to the medium at a final concentration of 0.15 mM as soon as the nephelometer units of the conditional mutant strain stopped increasing. Samples were taken at regular intervals during growth for analyses of the morphology by phase-contrast microscopy. The depletion experiments with the strain DKL41 were monitored in C+Y medium supplemented with 0.15 mM ZnCl2 or 0.09 mM ZnCl2.
Visualization of PBPs with BocillinTMFL and Western blot analysis
Cells from exponentially growing cultures were collected at a cell density of N = 70–80 by centrifugation. Preparation of samples, PBP labelling of cell lysates with BocillinTMFL and separation of proteins by SDS-PAGE was carried out essentially as described previously (Maurer et al., 2008). Briefly, cells were resuspended and lysed in 20 mM sodium phosphate buffer, pH 7.2, 0.2% (w/v) Triton X-100. The volume was adjusted so that 7.5 μl of cell suspension or one sample corresponds to 1.5 ml of culture of N = 20. Proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) with a 5% stacking gel and 7.5% resolving gel (acrylamide : bisacrylamide = 30:0.8). Bocillin-PBP complexes were visualized by fluorography with the FluorImager 595 (Molecular Dynamics) at 488 nm. After SDS-PAGE, the proteins were transferred to a PVDF membrane (Roche Diagnostics) and probed with affinity purified polyclonal anti-PBP2x antibody (Maurer et al., 2008), anti-PBP1a antiserum (Hakenbeck et al., 1991) respectively anti-GFP-antibody (rabbit IgG fraction (polycloal, Invitrogen) followed by incubation with alkaline phosphatase-conjugated goat anti-rabbit immunoglobulin G (Sigma-Aldrich) and staining with 4-nitrobluetetrazolium chloride and 5-bromo-4-chloro-3-indolylphosphate (Roche). The presence of equal amounts of protein in the samples was verified on separate SDS-polyacrylamide gels by Coomassie blue staining. To support the visual results immunoblots were quantified using the program ImageJ (NIH).
Effect of antibiotics on GFP–PBP2x localization
The strain DKL031 was grown up to a cell density of N = 70 at 30°C in C+Y medium supplemented with 0.15 mM ZnCl2. At this time point cultures were diluted 1:20 and divided into different aliquots that later either received antibiotics or not. Different concentrations of antibiotics were added when the cell density reached 30 N. The concentrations of antibiotics were calculated according the MIC of S. pneumoniae R6 strain (MIC for cefotaxime 0.02 μg ml−1 and for oxacillin 0.08 μg ml−1). The cells were incubated for 5 min, 60 or 120 min in the presence or absence of antibiotics. Then the samples were taken for fluorescent microscopy or for labelling of PBP's in cells with BocillinTMFL (30 μM final concentration). Growth of the cultures was followed by nephelometry (N, Nephelo units).
Cells grown in liquid medium at 30°C were analysed by phase contrast or epi-fluorescence microscopy using an Eclipse E600 (Nikon) microscope and 100× NA 1.4 oil immersion objective; pictures were taken with a DXM1200C camera (Nikon). Fluorescence signals of GFP were visualized using the Epi-FL filterblock B-2E/C (EX: 465–495, DM: 505, BA: 515–555; Nikon) using identical fluorescence intensity. Typical exposure times were between 1 and 2 s.
Van-FL staining on unfixed cells was performed as recently described (Beilharz et al., 2012). Briefly, a 1:1 mixture of Bodipy® FL Vancomycin (Invitrogen) with Vancomycin (Sigma) was added to 25 μl of a S. pneumoniae DKL031 cell culture to a final concentration of 0.1 μg ml−1, and incubated for 5–10 min at room temperature in the dark before examination by fluorescence microscopy.
The images were analysed, adjusted und cropped and determination of the cell size was performed with the Nikon Imaging software Nis-Elements BR (version 3.2).
Time-lapse microscopy was performed as described before (de Jong et al., 2011). In brief, a IX71 Microscope (Olympus) equipped with a CoolSNAP HQ2 camera (Princeton Instruments), 300W Xenon Light Source, 100× objective, GFP filterset (Chroma, excitation at 470/40 nm, emission 525/50 nm), mCherry filterset (Chroma, excitation at 572/35 nm, emission 632/60 nm), assembled by Imsol (UK) was used. Images were deconvolved prior to analysis using Softworx (Deltavision). Image analysis was carried out using ImageJ.
The authors would like to thank Nathalie Heß for providing us with the pFP093 plasmid, Ilka Zerfaß with the strain R6pbp2xT388GΔpbp1a::aphIII and Kerstin Breckner with the strain DKL017. We thank Leiv Håvarstein for communicating results prior to publication and Reinhold Brückner for helpful discussions. This work was supported by a grant from the Deutsche Forschungsgemeinschaft Ha 1011/11-1 and a fellowship from the Stiftung Alfried-Krupp Kolleg Greifswald to R.H., and the work in the lab of JWV was supported by a Veni fellowship from NWO-ALW. The authors have no conflict of interest to declare.