Co-ordinate synthesis and protein localization in a bacterial organelle by the action of a penicillin-binding-protein



Organelles with specialized form and function occur in diverse bacteria. Within the Alphaproteobacteria, several species extrude thin cellular appendages known as stalks, which function in nutrient uptake, buoyancy and reproduction. Consistent with their specialization, stalks maintain a unique molecular composition compared with the cell body, but how this is achieved remains to be fully elucidated. Here we dissect the mechanism of localization of StpX, a stalk-specific protein in Caulobacter crescentus. Using a forward genetics approach, we identify a penicillin-binding-protein, PbpC, which is required for the localization of StpX in the stalk. We show that PbpC acts at the stalked cell pole to anchor StpX to rigid components of the outer membrane of the elongating stalk, concurrent with stalk synthesis. Stalk-localized StpX in turn functions in cellular responses to copper and zinc, suggesting that the stalk may contribute to metal homeostasis in Caulobacter. Together, these results identify a novel role for a penicillin-binding-protein in compartmentalizing a bacterial organelle it itself helps create, raising the possibility that cell wall-synthetic enzymes may broadly serve not only to synthesize the diverse shapes of bacteria, but also to functionalize them at the molecular level.


Though historically conceived to be simple organisms, bacteria display complex spatiotemporal patterning of cells. They precisely regulate their morphology, cytoskeletal architecture and protein localization (Gitai, 2005), and some bacteria even produce organelles of specialized form and function (Murat et al., 2010). For instance, autotrophs create proteinaceous microcompartments called carboxysomes for carbon fixation (Yeates et al., 2008), cyanobacteria contain intracellular membranes called thylakoids that specialize in photosynthesis (Nickelsen et al., 2011), and diverse magnetotactic bacteria produce vesicular invaginations called magnetosomes to orient along magnetic fields (Komeili et al., 2006). However, unlike the broadly shared primary organelles of eukaryotes, bacterial organelles have uniquely and independently evolved in different lineages for different functions. Studying the diversity of their mechanisms for compartmentalization could therefore yield insights into de novo organelle formation across cellular life. To this end, here we examine the compartmentalization of the stalk organelle in Caulobacter crescentus, an aquatic Alphaproteobacterium.

Within the Alphaproteobacteria, many aquatic species produce surface organelles called stalks, thin cylindrical extrusions of the cell body that participate in functions such as nutrient uptake, buoyancy, and reproduction (Poindexter, 1978; 2006; Wagner et al., 2006; Brown et al., 2011). Stalks vary in number, size, and positioning between different organisms, but their compartmentalization is best studied in the model organism C. crescentus (hereafter referred to as Caulobacter). Newborn Caulobacter cells are motile, stalkless and non-replicative, but undergo morphological differentiation during the onset of reproductive maturity. During differentiation, newborn swarmer cells lose their polar flagellum and pili and in their place extrude a single stalk, the tip of which anchors a surface adhesive known as the holdfast (Brown et al., 2009; Curtis et al., 2012). The stalk increases the buoyancy of cells, allows cells to extend away from attached surfaces, and increases the length of the cell available for nutrient uptake (Poindexter and Cohen-Bazire, 1964; Poindexter, 1978; Wagner et al., 2006; Klein et al., 2013). These features are thought to play a beneficial role in sustaining Caulobacter's aerobic, oligotrophic lifestyle.

Structurally, the stalk is a thin but continuous extension of the cell body's Gram-negative envelope (Poindexter and Cohen-Bazire, 1964), but several features of the stalk set it apart from the rest of the cell. First, the synthesis of the stalk proceeds through a dedicated mechanism of polar growth at the stalked cell pole, with stalk material extruded outwards from the cell body as new stalk material is synthesized at the pole (Schmidt and Stanier, 1966). Many proteins involved in peptidoglycan and outer membrane biogenesis are involved in this process. These include the primary cell wall elongase complex of PBP2, RodA and MreB (Wagner et al., 2005; Divakaruni et al., 2007), a stalked pole specific bifunctional penicillin-binding-protein (PBP) called PbpC (Kühn et al., 2009), and the BAM (β-barrel assembly machinery) complex, involved in outer membrane protein assembly (Ryan et al., 2010); the absence of any of these proteins results in reduced stalk lengths. Second, the stalk possesses unique structural features known as crossbands, proteinaceous diffusion barriers that transect the stalk at irregular intervals. Crossbands cannot be traversed by other proteins in the stalk. They therefore establish functional differences between the cell body proximal and cell body distal segments of the stalk, permitting rapid responses by the cell body to environmental fluctuations (Schlimpert et al., 2012). Third, stalks undergo dramatic regulation by phosphate levels in the environment. Cells display > 10-fold longer stalks when starved for phosphate than when under phosphate-replete conditions (Gonin et al., 2000). This observation bolsters the hypothesis that stalks are specialized in nutrient uptake. Last, despite the structural continuity between the stalk and the cell body, stalks maintain a distinct molecular composition, the primary topic of this study.

Molecular compartmentalization of the stalk from the cell body has long been hypothesized in C. crescentus. Early observations by electron microscopy showed that the stalk lacks cytoplasmic components present in the cell body, such as DNA and ribosomes (Poindexter and Cohen-Bazire, 1964). Smaller cytoplasmic proteins such as free YFP are also excluded from the stalk, even when abundantly expressed in the cell body (Schlimpert et al., 2012). Proteomic analyses further confirmed that stalks are largely devoid of cytoplasmic and inner membrane proteins, though they contain a diversity of periplasmic and outer membrane proteins involved in nutrient uptake (Ireland et al., 2002; Wagner et al., 2006). Most recently, three inner membrane proteins – StpX, CC0498 and CC1953 – were found to localize preferentially to the stalk, providing visual confirmation for the existence of dedicated mechanisms for protein compartmentalization between the stalk and the cell body (Werner et al., 2009; Hughes et al., 2010). However, the pathways for stalk-specific protein targeting are unknown. Here, we utilize the stalk-specific protein StpX as a tool to characterize one such protein-targeting mechanism in C. crescentus.

StpX is an inner membrane protein with an N-terminal Sec-type signal peptide followed by a periplasmic domain, a single transmembrane domain, and a C-terminal cytoplasmic domain. Domain truncation analysis of StpX revealed that its periplasmic domain is both necessary and sufficient for stalk-specific targeting; membrane anchored GFP fused to the periplasmic domain of StpX alone could localize to the stalk (Hughes et al., 2010). These results demonstrate that the events responsible for the stalk localization of StpX occur in the periplasm. However, the cellular factors involved in this process remained unknown. StpX was able to localize in the absence of both the crossband diffusion barriers of the stalk and the polar developmental regulators DivJ and PleC (Hughes et al., 2010; Schlimpert et al., 2012).

In this study, we perform a forward genetic screen to identify genes involved in StpX's stalk localization. We identify the stalked-pole PBP PbpC as required for StpX localization; in PbpC's absence, StpX mislocalizes throughout the cell. We further show that PbpC acts co-ordinately with stalk synthesis at the stalked pole to immobilize StpX within the elongating stalk. PbpC does this by directly or indirectly anchoring StpX to the outer membrane, possibly in concert with the outer membrane BAM complex lipoprotein BamE. Finally, we show that StpX plays a role in cellular homeostasis of divalent transition metal ions, suggesting a possible new function for the stalk in C. crescentus.


A PBP at the stalked pole is required for stalk-specific localization of StpX

To characterize the cellular mechanism for stalk-specific localization of StpX, we applied a forward genetic screen to identify mutants of Caulobacter that mislocalize StpX-GFP to the cell body. Cells expressing StpX-GFP from its native chromosomal locus were mutagenized using the Mariner transposon, and individual mutants were isolated and screened for patterns of StpX-GFP localization using fluorescence microscopy. Of ∼ 2300 mutants tested, we obtained three mutants that mislocalized StpX-GFP to the cell body. All three mutations independently mapped to the gene pbpC, whose clean deletion from the genome caused StpX-GFP mislocalization (Fig. 1A and B).

Figure 1.

The penicillin-binding-protein PbpC is required for stalk localization of StpX.

A. A schematic of PbpC representing the sites of transposon insertion (red arrowheads) in three independent mutants that mislocalized StpX-GFP in our screen. Cyt, cytoplasmic localization domain; TM, transmembrane domain; TG, transglycosylase domain; TP, transpeptidase domain.

B. StpX-GFP localization in WT (YB5042), ΔpbpC (YB4075), ΔbacAB (YB6741) and ΔbacAB ΔpbpC (YB6746) cells. Scale bar: 2 μm.

PbpC is a predicted Class A bifunctional PBP that contributes to stalk elongation; in the absence of PbpC, cells produce stalks that are ∼ 25% shorter than wild-type cells (Kühn et al., 2009). It is localized at the stalked cell pole by proteins known as bactofilins, BacA and BacB, which themselves localize at the stalked pole and serve as recruitment factors for other proteins (Kühn et al., 2009). To test whether the bactofilins play a role in StpX localization, we visualized StpX-GFP in a ΔbacAB mutant. Indeed, we found that the ΔbacAB mutant had a partial StpX mislocalization phenotype, with elevated levels of StpX-GFP in the cell body compared with wild-type cells (Fig. 1B). When single deletions of bacA and bacB were tested for StpX-GFP localization, we found that the deletion of bacA was responsible for the ΔbacAB phenotype, whereas ΔbacB resembled wild-type (Fig. S1A). Correspondingly, the stalked-pole localization of PbpC was disrupted in ΔbacAB and ΔbacA cells, but not in ΔbacB cells (Fig. S1B). These results are consistent with the previous finding that BacA is present in higher copy numbers and plays a more dominant role in the cell than BacB (Kühn et al., 2009).

Importantly, the phenotype of the ΔbacAB mutant was not as severe as that of the ΔpbpC mutant; ΔbacAB cells displayed some StpX-GFP enrichment in the stalk, unlike ΔpbpC cells (Fig. 1B). Furthermore, the triple ΔbacAB ΔpbpC mutant was identical to ΔpbpC in its StpX mislocalization phenotype, indicating that pbpC is epistatic to the bactofilins (Fig. 1B). Thus, the bactofilins play an indirect role in StpX localization by targeting PbpC to the stalked pole, improving PbpC's efficiency in localizing StpX to the stalk.

Intriguingly, we observed that PbpC's retention at the stalked pole depended on the presence of StpX in the cell. In cells containing StpX, Venus-PbpC displayed a strong polar focus at the stalked pole, with only a small amount dispersed within the stalk (Fig. 2A). But in cells lacking StpX, Venus-PbpC no longer localized at the stalked pole and instead was present at elevated levels throughout the stalk in a patchy distribution (Fig. 2A). Thus, it appears that functional interactions between StpX and PbpC occur at the stalked pole, simultaneously localizing StpX within the stalk and maintaining PbpC at the stalk-cell body junction.

Figure 2.

StpX localization is concurrent with stalk synthesis at the stalked pole.

A. Venus-PbpC localization in ΔpbpC (StpX+, YB2104) and ΔpbpC ΔstpX (StpX, YB6799). Black arrowheads point to foci of PbpC at the stalked pole.

B. Cells expressing StpX-GFP under a xylose-inducible promoter (YB6782) were labelled with the outer membrane stain TRSE following a first phase of stalk elongation under non-inducing conditions, then grown for a second period of stalk elongation under inducing conditions. Scale bars: 2 μm.

StpX localization is coupled to stalk synthesis

The stalked pole is the site of stalk synthesis in Caulobacter. The stalk undergoes polar growth at this junction, with older stalk material extruded away from the pole as new stalk material is synthesized (Schmidt and Stanier, 1966). Given PbpC's localization at the stalked pole and its involvement in stalk elongation (loss of pbpC causes a 25% reduction in stalk lengths) (Kühn et al., 2009), we hypothesized that StpX localization might be coupled to stalk synthesis. To test this hypothesis, we used a strain in which StpX-GFP expression was driven by a xylose-inducible promoter at the chromosomal xylX locus. Cells were first grown in the absence of xylose to allow an initial phase of stalk elongation in the absence of StpX-GFP expression. Then, existing cell/stalk material was labelled using Texas Red succinimidyl ester (TRSE), an amine-reactive dye that non-specifically labels surface-exposed proteins (Brown et al., 2012). Finally cells were subjected to a second phase of stalk elongation in the presence of xylose, and therefore StpX-GFP expression (Fig. 2B, schematic). Surface-exposed proteins in Caulobacter include the S-layer, which is a crystalline array of hexamers of the protein RsaA surrounding the entire cell (Smit et al., 1981; 1992; Gilchrist et al., 1992), as well as outer membrane proteins, many of which are anchored to the peptidoglycan and therefore do not diffuse. Old stalk material labelled with TRSE therefore retains the dye and appears red even as the stalk continues growing from its base, whereas new stalk material lacks TRSE labelling. If StpX localization occurs independently of stalk synthesis, we should expect StpX-GFP expressed in the second stalk elongation phase to be present in both old and new stalk material, whereas if the two processes are coupled, StpX-GFP should be present exclusively in the new stalk material (Fig. 2B, schematic). The results of our experiment showed perfect complementarity in the patterns of TRSE and StpX-GFP; StpX localized exclusively to new stalk material as inferred by the absence of TRSE labelling (Fig. 2B and S2, left). This demonstrates that StpX localizes only to the stalk material that is synthesized during its time of expression. Conversely, when StpX-GFP was expressed during the first phase of stalk elongation instead of the second, the StpX-GFP and TRSE signals were colocalized (Fig. S2, right), consistent with StpX localization occurring during stalk synthesis.

The proline-rich periplasmic tail of PbpC is required for StpX localization

PbpC is a large, 733 amino acid PBP homologous to Escherichia coli's PBP1A. It has a short, N-terminal cytoplasmic domain, a single transmembrane domain, and an extensive periplasmic domain featuring both transglycosylase and transpeptidase domains for peptidoglycan synthesis (Fig. 1A). To gain insight into how PbpC localizes StpX to the stalk, we constructed mutations targeting different features of PbpC to test their effects on StpX localization (Figs 3A and S3). We made these mutations in a venus-pbpC gene integrated at the chromosomal xylX locus, which served as the sole copy of pbpC in a ΔpbpC background. This allowed us to confirm using fluorescence microscopy and Western blotting that the mutant proteins were properly expressed and localized in the cell (Fig. 3B, Venus panels, and Fig. S4A). StpX-CFP (Fig. S4B) was expressed from its native locus to allow simultaneous detection of both proteins.

Figure 3.

The periplasmic tail of PbpC, but not its enzymatic activity, is required for StpX localization.

A. Schematic of the Venus-PbpC fusion and its mutant variants used for domain analysis. V, Venus; Cyt, cytoplasmic localization domain; TM, transmembrane domain; TG, transglycosylase domain; TP, transpeptidase domain.

B. Phase, Venus, and CFP micrographs and fluorescence overlays from cells expressing StpX-CFP and mutant variants of Venus-PbpC (WT, YB6767; Cyt, YB6832; Per, YB6831; Tail, YB6793; Enz, YB6780). Grey arrowheads indicate localization of Venus-PbpC at the stalked pole. Scale bar: 2 μm.

WT Venus-PbpC is functional for StpX localization. In ΔpbpC cells expressing WT Venus-PbpC, StpX-CFP exhibited stalk-specific localization and WT Venus-PbpC exhibited focal localization at the stalked cell pole as expected (Fig. 3B, top row). Next, we analysed a version of Venus-PbpC lacking its cytoplasmic domain, Cyt Venus-PbpC (Figs 3A and S3). The cytoplasmic domain of PbpC is its localization domain, interacting with the bactofilins to achieve stalked pole localization (Kühn et al., 2009). Consequently, the truncated Cyt Venus-PbpC protein was uniformly distributed throughout the cell body and stalk, and no longer maintained focal localization at the stalked pole (Fig. 3B, second row). StpX-CFP in these cells had a partial mislocalization phenotype identical to what was seen in ΔbacAB cells in which PbpC is mislocalized, with elevated levels of StpX-CFP in the cell body compared with cells expressing WT Venus-PbpC (Fig. 3B, second row).

Dramatically, Per Venus-PbpC (Figs 3A and S3) lacking the periplasmic domain of PbpC recapitulated the pbpC null phenotype. StpX-CFP was almost entirely retained in the cell body in these cells (Fig. 3B, third row). This requirement of the periplasmic domain of PbpC for localizing StpX is consistent with the previous finding that the periplasmic domain of StpX serves as its own localization determinant (Hughes et al., 2010). Furthermore, proline rich motifs occur in both the periplasmic domain of StpX as well as a divergent 65 amino acid stretch of the periplasmic tail of PbpC (Fig. S3), with similar motifs implicated in localization interactions between the bactofilins and PbpC (Kühn et al., 2009). We therefore constructed a truncation of PbpC that removed its proline-rich tail while leaving its enzymatic domains intact, to generate Tail Venus-PbpC (Figs 3A and S3). The phenotype of this mutant was identical to Per Venus-PbpC lacking the entire periplasmic domain (Fig. 3B, fourth row). Thus, the proline rich tail of PbpC appears to be important for localizing StpX to the stalk. Additionally, both inactive variants Per Venus-PbpC and Tail Venus-PbpC failed to localize at the stalked pole, confirming our previous observation that functional interactions between PbpC and StpX are required to maintain PbpC's localization at this site (Fig. 3B).

Finally, we carried out site-directed mutagenesis of the predicted active site residues of the transpeptidase and transglycosylase domains of PbpC to determine whether the enzymatic activity of PbpC plays a role in its ability to localize StpX. Enz Venus-PbpC (Figs 3A and S3) carries two missense mutations – E171Q and S439A – equivalent to mutations in E. coli's PBP1B that abolish its enzymatic activities (Terrak et al., 1999). However, this mutant displayed a wild-type phenotype for StpX localization. It localized StpX-CFP to the stalk and underwent stalked pole localization like WT Venus-PbpC (Fig. 3B, bottom row). This finding indicates that PbpC plays a novel, non-enzymatic function in the process of StpX localization.

PbpC is one of five high molecular weight bifunctional PBPs in Caulobacter. By constructing Venus fusions to each of the other PBPs (CC0252, CC1516, CC1875 and CC3570), we found that only the PBP1B homologue CC0252 also localized to the stalked pole under conditions of stalk elongation (Fig. S5A). However, its deletion had no effect on StpX-GFP localization (Fig. S5B). This result indicates that the role of PbpC in StpX localization is not shared by other high molecular weight PBPs present at the stalked pole.

PbpC immobilizes StpX within the stalk by anchoring it to a non-diffusible component of the outer membrane

To characterize the novel mechanism by which PbpC localizes StpX to the stalk, we considered an important feature of StpX localization – its restricted mobility in the stalk. Previous experiments testing the diffusion of StpX-GFP showed that properly localized StpX in the stalk is immobile and therefore cannot diffuse back into the cell body, whereas mislocalized Per StpX-GFP is unrestrictedly mobile in the cell (Hughes et al., 2010). This led us to ask whether PbpC is involved in immobilizing StpX within the stalk. To answer this question, we compared the mobility of StpX-GFP in WT and ΔpbpC cells using fluorescence recovery after photobleaching (FRAP) experiments. In WT cells, when a region of the stalk was bleached, there was no recovery of fluorescence in the bleached region over a 40 s period, despite the presence of StpX-GFP in adjacent stalk material (Fig. 4A, top row, and Fig. S5C). This confirms the known immobility of localized StpX in the stalk (Hughes et al., 2010). In contrast, when StpX-GFP was bleached in the stalks of ΔpbpC cells, fluorescence was recovered over the 40 s period, with StpX-GFP diffusing from the cell body into the stalk (Fig. 4A, bottom row, and Fig. S5C). This result together with our other findings argues that PbpC immobilizes StpX within the elongating stalk concurrent with stalk synthesis at the stalked pole, causing retention of StpX within the elongating stalk organelle.

Figure 4.

PbpC immobilizes StpX in the stalk by anchoring it to the rigid outer membrane.

A. StpX-GFP in WT (YB5042) and ΔpbpC (YB4075) cells was bleached in the region indicated by the yellow rectangle, followed by immediate acquisition of a post-bleach image. A post-recovery image was obtained 40 s after the bleaching event. The orange outline in WT panels represents the cell body.

B. WT cells expressing StpX-GFP and StpB-mCherry (YB5043) were introduced into a microfluidic chamber and subjected to Tris-EDTA treatment. Fluorescence was monitored as the outer membrane lost its integrity over a 42 min period. A representative sequence from time-lapse microscopy of these cells is shown. Scale bars: 2 μm.

Next, we wanted to understand how StpX becomes immobilized within the stalk during its localization by PbpC. A likely explanation is that StpX might be anchored to rigid components of the cell envelope – either the peptidoglycan or the outer membrane – which are simultaneously synthesized during stalk elongation. StpX does not contain any known peptidoglycan binding motifs, but it has an extensive 347 amino acid periplasmic domain that is rich in prolines. In fact, 48 proline residues occur in the periplasmic domain of StpX, second only to alanine in abundance, an unusual feature for most proteins. In inner membrane proteins, most notably TonB, it has been shown that proline rich periplasmic motifs form long, relatively inflexible envelope spanning segments that enable transenvelope interactions between the inner and outer membranes (Köhler et al., 2010). Therefore, we predicted that the periplasmic domain of StpX might similarly traverse the cell envelope to interact with components of the outer membrane.

To test whether interactions with the outer membrane are required for StpX's immobility and stalk localization, we examined the effect of outer membrane disruption on StpX localization in a strain expressing StpX-GFP and StpB-mCherry. StpB-mCherry is a periplasmic protein that is dispersed in the periplasm of the cell body and is also a constituent of crossbands, proteinaceous diffusion barriers that compartmentalize the stalk (Schlimpert et al., 2012). Cells expressing these proteins were introduced into a microfluidic chamber, and Tris-EDTA was flowed over them to disrupt their outer membranes during time-lapse microscopy. Under these conditions, the outer membranes of cells experienced gradual disruption until they were completely breached by ∼ 30 min after initiation of Tris-EDTA treatment, as indicated by the loss of StpB-mCherry from the periplasm of the cell body (Fig. 4B). Coincident with the final stages of outer membrane disruption, StpX-GFP in the stalk segment between the cell body and its nearest crossband diffused out of the stalk and into the cell body, showing that immobility and localization of StpX require an intact outer membrane (Fig. 4B). This experiment was repeated identically in cells expressing StpX-GFP and cytoplasmic DsRed to confirm that the observed behaviour of StpX-GFP was not due to clipping of the GFP tag into the cytoplasm. Indeed, the behaviour of StpX-GFP in the cell body did not coincide with the behaviour of cytoplasmic DsRed. StpX-GFP remained associated with the cell even after cytoplasmic DsRed was lost (Fig. S6A), indicating that StpX-GFP diffusing from the stalk into the cell body was still associated with the inner membrane. Finally, 6–12 min after completion of outer membrane disruption in these experiments, all cell body fluorescence was lost (Figs 4B and S6A). This was likely due to cellular autolysis as a result of EDTA treatment (Leduc et al., 1982), whereas the tips of stalks remained fluorescent as they are slower to lyse than the cell body (Schmidt and Stanier, 1966). When water was flown over cells under similar conditions as a control, StpX-GFP localization was not impacted (Fig. S6B).

The requirement of outer membrane integrity for the immobilization and retention of StpX in the stalk supports the hypothesis that StpX is anchored to an outer membrane protein. To identify outer membrane factors that are involved in this anchoring process, we performed co-immunoprecipitation using an anti-StpX antiserum in lysates of WT and ΔstpX cells, followed by mass spectrometry of proteins that were pulled down. Seven proteins were pulled down in WT but not in ΔstpX cells (Table S1). Notable among these was BamE, an outer membrane lipoprotein. BamE is a member of the BAM complex, which is responsible for assembly of outer membrane beta barrel proteins in Gram-negative bacteria (Knowles et al., 2009). In Caulobacter, BamE also plays a role in stalk elongation (Ryan et al., 2010), making it a good candidate for involvement in StpX's stalk-anchoring. To test the role of BamE in StpX localization, we examined StpX-GFP fluorescence in WT and ΔbamE cells. We found that ΔbamE cells were indeed compromised in StpX localization, exhibiting a 4.5-fold elevation in StpX-GFP levels in the cell body compared with WT cells (Fig. 5). Nevertheless, ΔbamE cells exhibited partial enrichment of StpX-GFP in the stalk, indicating that StpX must be anchored to some other outer membrane protein in the stalk. Furthermore, the cellular distribution of BamE revealed that it is primarily in the cell body, with lower levels in the stalk (Fig. S7A). These findings, together with the previous report that many outer membrane beta barrel proteins are reduced in their levels of expression in the ΔbamE mutant (Ryan et al., 2010), suggest that BamE likely plays an indirect role in StpX localization by facilitating the assembly into the stalk of some other outer membrane protein that ultimately anchors StpX.

Figure 5.

The outer membrane lipoprotein BamE promotes StpX localization.

A. Phase and fluorescence micrographs of StpX-GFP expressed in WT (YB5042) and ΔbamE (YB7323) cells. Scale bar: 2 μm.

B. Quantification of StpX-GFP fluorescence in the cell bodies of the strains in panel A, analysing > 250 cells for each strain. Error bars represent the standard error of the mean. RFU, relative fluorescence units. *P < 0.0001.

The specific identity of StpX's outer membrane target remains unknown. One possibility is that it is the rigid, proteinaceous S-layer that surrounds the outer membrane of the Caulobacter cell. The S-layer is a repeating array of the protein RsaA, which in turn is tethered to the lipopolysaccharide of the outer membrane (Smit et al., 1992; Walker et al., 1994). To test the S-layer's involvement in anchoring StpX to the stalk, we examined the localization of StpX in a mutant lacking the protein RsaA. However, ΔrsaA cells localized StpX into the stalk identically to wild type cells (Fig. S7B).

StpX is required for cell tolerance to zinc toxicity

The existence of dedicated mechanisms for the localization of StpX in the stalk raises questions regarding its function within the organelle. However, StpX is a protein of unknown function, without sequence similarity to any known proteins or protein domains in existing databases. But visual inspection of the primary sequence of StpX offers some clues to its possible function. Its most striking feature is the abundance of histidines in the cytoplasmic domain, a 200 amino acid stretch of which contains 37% histidines, higher than any other amino acid (Fig. S8). A similarly histidine-rich polypeptide is found in Helicobacter pylori, the protein Hpn, which consists of 47% histidines (Gilbert et al., 1995). H. pylori Hpn is a potent binder of Ni2+, Cu2+ and Zn2+, and can contribute to both Ni2+ storage and detoxification depending on exogenous levels of the metal (Ge et al., 2006; Seshadri et al., 2007). In addition to histidine, the cytoplasmic domain of StpX is also enriched in aspartates (Fig. S8). Similar histidine and aspartate-rich motifs are seen in the high-affinity Zn2+ uptake protein ZnuA of E. coli and Neisseria gonorrhoeae (Patzer and Hantke, 1998; Chen and Morse, 2001), further suggesting a metal-binding role for the cytoplasmic domain of StpX. Therefore, we hypothesized that StpX may play a role in metal homeostasis in Caulobacter.

To test whether variations exist between WT and ΔstpX cells in intracellular levels of different metal ions, we performed inductively coupled plasma mass spectrometry (ICP-MS), a technique that permits quantitative analysis of numerous transition metal ions (Jacobsen et al., 2011). Comparing lysates from WT and ΔstpX cells grown in rich medium, we found that zinc, manganese, and iron levels were comparable between the two strains, while nickel was undetectable (Fig. 6A). However, WT cells showed a > 3-fold elevation in copper levels compared with ΔstpX cells (Fig. 6A). This finding suggested that StpX may be involved in the uptake or storage of copper in cells. We therefore predicted that ΔstpX cells might be compromised for growth when limited for copper. To test this, we compared the growth rates of the two strains in a minimal medium that lacked Cu2+. In the absence of Cu2+, WT cells grew slightly faster than ΔstpX cells, though the difference vanished when the medium was supplemented with trace amounts of Cu2+ (Fig. S9). Thus, StpX appears to play a role in efficient storage and utilization of Cu2+ when it is limited in the environment.

Figure 6.

StpX plays a role in metal homeostasis.

A. ICP-MS analysis of different metal ions in lysates of WT (YB127) and ΔstpX (YB5231) cells. Error bars represent the standard error of the mean. *P < 0.02.

B. Recovery of WT and ΔstpX cells subjected to Zn2+ treatment for 0 to 20 h. Error bars represent the standard error of the mean.

C. Recovery of WT, ΔstpX and Δper stpX (YB5229) cells subjected to 2 mM Zn2+ treatment for 20 h. Error bars represent the standard error of the mean. ND, viability not detected.

D. Quantification of cell body zinquin fluorescence from WT and ΔstpX cells treated in the absence or presence of 1 mM Zn2+. For each condition, measurements were averaged from > 150 cells. *P < 0.0001. Representative WT and ΔstpX cells showing zinquin fluorescence, quantified in Panel D, are shown without any added Zn2+ (E) or with 1 mM Zn2+ treatment for 5 h (F). Scale bars: 2 μm.

Additionally, we found an even greater difference between WT and ΔstpX cells in their response to zinc toxicity. When cultures of the two strains were treated with 1 or 2 mM Zn2+ for between 0 and 20 h, the viability of ΔstpX cells dropped several orders of magnitude below that of WT cells, and at 20 h of 2 mM Zn2+ treatment, ΔstpX cells had lost detectable viability (< 100 cfu ml−1; cfu, colony-forming units), whereas WT cells still recovered at > 104 cfu ml−1 (Fig. 6B). Furthermore, the stalk-specific localization of StpX appeared to optimize cellular tolerance to Zn2+. When we compared the viability of WT and ΔstpX cells to Δper stpX cells expressing a truncated, mislocalized mutant of StpX lacking its periplasmic domain, cells expressing mislocalized Per StpX were compromised in their Zn2+ resistance compared with WT, though not as severely as ΔstpX cells (Fig. 6C). Taken together, these results imply that StpX plays a role in protecting cells from Zn2+ toxicity and that its localization in the stalk optimizes its ability to do so.

One hypothesis for StpX-mediated Zn2+ tolerance is that the protein might act as a metal sink that sequesters Zn2+ into the stalk, away from DNA, ribosomes, and other crucial machinery of the cell body that might undergo oxidative damage. Or alternatively, StpX might function as part of an efflux system that rapidly releases Zn2+ back into the extracellular environment to prevent its excessive accumulation in the cell body. To distinguish between these scenarios, we grew WT and ΔstpX cells in the presence of zinquin ethyl ester, a small molecule that is taken up by cells and hydrolysed to form zinquin acid (Ammendola et al., 2007). Both zinquin ethyl ester and zinquin acid are zinc-specific fluorescent sensors that can detect zinc-bound proteins in the cell (Nowakowski and Petering, 2011). Cultures pre-incubated with zinquin ethyl ester were treated in the presence or absence of 1 mM Zn2+ and imaged after 5 h for zinquin fluorescence to examine their intracellular Zn2+ levels. In the absence of exogenous Zn2+, zinquin fluorescence was almost undetectable in WT and ΔstpX cells (Fig. 6D and E). But when cells were treated with 1 mM Zn2+, we detected a > 2-fold elevation in zinquin fluorescence in ΔstpX cells compared with WT (Fig. 6C and F). Importantly, zinquin fluorescence in Zn2+-treated cells was abundant in cell bodies but was undetected in stalks (Fig. 6F). This pattern of Zn2+ distribution was corroborated by STEM/EDS analysis (scanning transmission electron microscopy with energy dispersive X-ray spectroscopy) of WT cells treated with Zn2+ (Fig. S10). These data argue against the hypothesis that StpX acts as a sink to sequester Zn2+ in the stalk, away from the cell body. Instead, our results support a model wherein StpX might be involved in the efflux of Zn2+ out of the stalk, allowing WT cells to maintain a lower intracellular concentration of the heavy metal, thereby promoting survival.


Bacterial organelles take on a variety of forms, but their defining feature is their molecular compartmentalization from the rest of the cell, allowing for functional specialization. The Caulobacter stalk maintains strong differences from the cell body in its molecular composition. It is mostly devoid of inner membrane and cytoplasmic proteins, contains good representation of outer membrane and periplasmic proteins (Ireland et al., 2002), and displays selective enrichment of a few known proteins such as StpX (Werner et al., 2009; Hughes et al., 2010). The targeting of these stalk-specific proteins has important implications both for discovering the mechanisms of subcellular organization in stalked bacteria, as well as for understanding the varied functions of the stalk organelle.

Here we characterized the mechanism of localization of the stalk-specific protein StpX in C. crescentus. Previous work on the intrinsic localization determinants of StpX demonstrated that its periplasmic domain was responsible for its stalk-specific localization (Hughes et al., 2010). Here, we focused on identifying additional cellular factors involved in the process. In doing so, we have identified a class A, bifunctional PBP called PbpC as essential for StpX's stalk-specific localization. Furthermore, we have identified a non-enzymatic function for PbpC in this process, the first such description of a PBP that we are aware of.

To localize StpX, PbpC anchors it into the elongating stalk organelle concurrent with stalk synthesis. This is evident from the localization of StpX exclusively in the stalk material that is synthesized during its time of expression. Consistently, functional interactions between StpX and PbpC occur at the stalked cell pole, the site of stalk synthesis in Caulobacter. Several lines of evidence demonstrate this. First, PbpC localizes at the stalked cell pole and disruption of its localization either by deletion of its localization domain or by the loss of its bactofilin scaffold results in increased StpX mislocalization. Second, functional interactions between PbpC and StpX appear to maintain PbpC's own stalked pole localization; PbpC is mislocalized from the stalked pole when StpX is absent from the cell, and mutant variants of PbpC that are inactive in localizing StpX are also unable to achieve stalked pole localization.

What is the nature of the functional interactions between PbpC and StpX at the stalked pole? Properly localized StpX is immobile, likely through anchoring of its periplasmic domain to some rigid element within the stalk (Hughes et al., 2010). PbpC is required for StpX to establish these anchoring interactions, because in its absence, StpX is unrestrictedly mobile. One hypothesis is that PbpC, being a peptidoglycan-synthase, somehow tethers StpX to the elongating stalk peptidoglycan. However, the enzymatic activity of PbpC is dispensable for its role in localizing StpX, and StpX lacks any known peptidoglycan binding domains. Instead, StpX appears to be anchored by interactions with some rigid component of the outer membrane of the stalk. When the outer membrane is disrupted, StpX is mobilized from the stalk back into the cell body. Furthermore, the outer membrane lipoprotein BamE of the BAM complex is an interacting partner of StpX, identified by Co-IP/mass spectrometry. In the absence of BamE, cells show increased StpX mislocalization, suggesting that StpX is anchored to BamE and/or other outer membrane proteins that depend on the BAM complex for their assembly. These lines of evidence taken together suggest that PbpC at the stalked pole facilitates interactions between StpX and the outer membrane, causing StpX retention in the stalk.

Interestingly, a divergent, proline-rich, 65 amino acid stretch of the periplasmic tail of PbpC was required for StpX localization. The secondary structure of this region of PbpC was predicted to be a coiled coil using I-TASSER, a 3D-structure prediction tool that models proteins based on known crystal structures of related proteins (in this case, other class A bifunctional PBPs) (Zhang, 2008; Roy et al., 2010). Coiled coil regions of proteins are often implicated in protein-protein interactions. However, we were unable to detect direct interactions between StpX and PbpC. PbpC was not pulled down in our Co-IP/mass spectrometry analysis of interacting partners of StpX, nor did we see interactions between these two proteins in a bacterial two-hybrid assay (data not shown). It is possible that the periplasmic tail of PbpC instead interacts with BamE and/or another outer membrane target to facilitate their interactions with StpX. Such transenvelope interactions between a PBP and an outer membrane protein would not be without precedent. The primary bifunctional PBPs of E. coli, Pbp1a and Pbp1b, both interact with outer membrane lipoproteins called the Lpo proteins (Paradis-Bleau et al., 2010; Typas et al., 2010). In these interactions, the outer membrane Lpo proteins stimulate the peptidoglycan-synthetic activity of their associated PBPs, using an interaction domain between the transglycosylase and transpeptidase domains of the PBPs (Paradis-Bleau et al., 2010; Typas et al., 2010). The converse might occur in the case of PbpC, where interactions between PbpC's periplasmic tail and some outer membrane protein serve to stimulate that outer membrane protein's binding to StpX, co-ordinate with stalk synthesis. Future work characterizing interactions between PbpC, BamE and the outer membrane proteins present at the stalked pole during stalk-synthesis will be required to characterize the details of PbpC's non-enzymatic role in StpX localization.

Our discovery of a new mode of protein localization in the stalk makes it increasingly evident that the Caulobacter cell has multiple pathways for subcellular compartmentalization. Previous work has already described diffusion barriers within the stalk called crossbands. Crossbands are multi-subunit proteinaceous structures that assemble once per cell cycle in the Caulobacter stalk, segmenting it into distinct subcompartments between which proteins cannot diffuse (Schlimpert et al., 2012). Crossbands serve to functionally segregate the cell body distal part of the stalk from the rest of the cell, since changes in protein expression in the cell body are conveyed only to the newest stalk material that is continuous with the cell body, prior to the first crossband. In doing so, crossbands allow cells to rapidly respond to environmental fluctuations by concentrating newly synthesized proteins to the cell body and its proximal stalk (Schlimpert et al., 2012). However, PbpC-dependent and crossband-dependent stalk compartmentalization still do not complete the picture of stalk-specific protein compartmentalization in Caulobacter. The two other hypothetical proteins CC0498 and CC1953 which exhibit enrichment in the stalk (Werner et al., 2009) are not influenced in their localization by the absence of either crossbands or PbpC (data not shown), implying that additional mechanisms exist for compartmentalizing the stalk from the cell body in C. crescentus.

Finally, the presence of dedicated mechanisms for the molecular compartmentalization of the stalk from the cell body raises questions regarding the functional specialization of the stalk. The stalk is already known to function in buoyancy, nutrient uptake, and extension of cells from attached surfaces (Poindexter, 1978; Wagner et al., 2006; Klein et al., 2013). Learning the functions of stalk-localized proteins will continue to inform our understanding of the multiple specializations of the stalk. Here, we show that StpX functions in the uptake/retention of copper while also promoting the tolerance of cells to zinc toxicity. Under elevated levels of zinc, the presence of StpX allows a 50% reduction in intracellular accumulation of zinc, possibly through efflux of the metal out of cells. In freshwater lakes, sediment concentrations of zinc can exceed two hundred times what is seen in the Caulobacter cell by ICP-MS (∼ 80 μg g−1 protein in cells compared with ∼ 17 300 μg g−1 in sediment) (Wentsel et al., 1977). Especially under conditions of environmental acidification, dissolved levels of zinc can undergo significant increase (Schindler et al., 1980), particularly in proximity to zinc-rich sediments. StpX might provide a survival advantage to Caulobacter cells under such ecological conditions, where the stalk's role as a nutrient-uptake device may be a double-edged sword. Increased stalk length may facilitate nutrient uptake, but will also increase the accumulation of toxic substances from the environment. By functionalizing the stalk to immediately rid the cell of toxic substances, the cell could establish a balance between uptake and efflux through the stalk, in order to maintain cellular homeostasis. How StpX can at the same time act as a copper storage/scavenger and a zinc exporter is not clear. One possibility is that StpX may only be responsible for one of the two functions, either reducing zinc or increasing copper levels, in the actual ecological context. Alternately, StpX may be capable of interacting with two or more chaperones/efflux systems that influence its bias towards one function or the other depending on the cell's needs.

In summary, our study identifies a novel, non-enzymatic role for a PBP in functionalizing a specialized bacterial structure that it itself helps create. Since PBPs undergo strict spatiotemporal regulation in their localization to generate specific growth patterns and cellular morphologies in bacteria, it is an interesting notion that they may be coopted for functionalizing cellular subdomains by co-ordinating protein localization concurrent with structural synthesis.

Experimental procedures

Bacterial strains, plasmids, oligonucleotides, and growth conditions

All C. crescentus strains used in this study were generated in the NA1000 background and grown at 30°C. Strains were maintained on peptone yeast extract (PYE) plates (Poindexter, 1964) supplemented with antibiotics as necessary (kanamycin 20 μg ml−1, spectinomycin 50 μg ml−1, nalidixic acid 20 μg ml−1, gentamicin 5 μg ml−1). For microscopy, cells were grown to saturation (∼ 36 h) in Hutner base–imidazole-buffered–glucose (HIGG) medium (Poindexter, 1978) containing 30 μM phosphate (phosphate-limited) and with 0.03% (w/v) xylose for induction of Venus-PbpC and its variants. For analysis of metals, cells were grown in liquid PYE medium or M5G, as specified in the procedures for those specific experiments. A detailed list of oligonucleotides, strains, plasmids, and their methods of construction are included in the Supporting information (Tables S2–S5).


Standard fluorescence imaging was done on an upright Nikon 90i with a Plan Apo 100×/1.40 Oil Ph3 DM objective, with 51017 CFP/YFP and 83700 DAPI/FITC/Texas Red filter cubes from Chroma, and a Photometrics Cascade 1K EMCCD camera. Time-lapse imaging was done on an inverted Nikon Ti-E, using a Plan Apo 60×/1.40 Oil Ph3 DM objective, with a GFP/DsRed filter cube, and an Andor iXon3 DU885 EM CCD camera. Images were captured using NIS Elements (Nikon, Melville, NY) or MetaMorph (Molecular Devices, Sunnyvale, CA). Fluorescence quantification was done with FIJI using phase contrast images to identify cell boundaries and generate binary masks, which were then applied to their corresponding fluorescence images to quantify the fluorescent signal from individual cells.

Genetic screen for StpX mislocalization

NA1000 cells expressing StpX-GFP from the native chromosomal locus (YB5042) were conjugated overnight with E. coli containing the Mariner transposon on a plasmid (YB2028). Caulobacter mutants with transposon insertions were selected based on kanamycin resistance, and using Caulobacter's native nalidixic acid resistance to counter-select against E. coli donors. Individual colonies were picked into liquid PYE in 96 well plates and grown overnight to saturation. These stock plates were then used to inoculate fresh 96 well plates containing HIGG medium with 30 μM phosphate, which were grown for 48 h prior to imaging. Image acquisition was done using pedestal slides (Werner et al., 2009), implementing a macro within MetaMorph to automate data collection across the 48 samples on each slide. Mutants that mislocalized StpX-GFP were restreaked from their frozen stocks, and their transposon insertion sites were mapped using touchdown PCR (Levano-Garcia et al., 2005).

Induction of StpX-GFP during stalk synthesis

NA1000 cells expressing StpX-GFP under a xylose-inducible promoter (YB6824) were grown overnight to late exponential phase in HIGG medium (30 μM phosphate) containing either 0.3% (w/v) glucose or 0.3% (w/v) xylose. Three millilitres of each culture was washed twice in bicarbonate buffer (0.1 M NaHCO3, pH 8), and the pellets were resuspended in 1 ml of bicarbonate buffer containing 10 μg ml−1 TRSE and incubated for 5 min at room temperature. After labelling, cells were pelleted and washed twice in HIGG medium (30 μM phosphate) and finally resuspended in 3 ml HIGG supplemented with either 0.3% (w/v) xylose or 0.3% (w/v) glucose, switching sugars from what the cells were previously exposed to. The cultures were grown for 24 h to allow additional stalk elongation prior to imaging.

Photobleaching analysis of StpX mobility

WT or ΔpbpC cells expressing StpX-GFP (YB5042 or YB4075) were grown to saturation in HIGG medium containing 30 μM phosphate and mounted on 1% agarose pads (w/v in water). Photobleaching was done using a Leica TCS SP5 scanning confocal microscope equipped with a HCX PL APO Lambda Blue 63× 1.4 Oil objective (Leica Microsystems, Bannockburn, IL). A 488 nm argon laser was used for GFP excitation, and emission was measured between 500 and 600 nm. A pre-bleach image was acquired at 4% of maximum laser power. Then, indicated regions of the stalk were bleached using 100% laser power, followed by immediate acquisition of a post-bleach image and a 40 s recovery image, both using 4% laser power. Quantification of fluorescence from two independent experiments in each strain was done by drawing a pixel-wide line along the bleached region of the stalk and quantifying mean fluorescence in their pre-bleach, post-bleach, and recovery frames.

Outer membrane disruption

Cells coexpressing StpX-GFP and StpB-mCherry (YB5043) or StpX-GFP and cytoplasmic DsRed (YB4070) were grown to saturation in HIGG medium. Cells were washed twice in water using centrifugation at 10 000 g for 5 min to eliminate solutes from the medium. Then they were resuspended in water and loaded into the imaging chambers of four independent channels in a bacterial microfluidic plate using an ONIX perfusion system (CellASIC, Hayward, CA). Phase contrast and fluorescence images were acquired from fields in each microfluidic channel using a Nikon Ti-E microscope. Water was initially flowed over cells for 6 min, and then either 10 μM ethylenediaminetetraacetic acid (EDTA) in 50 mM Tris pH 7.8 or water as a control was flowed over different chambers while imaging was continued for an hour.

ICP-MS analysis

Six independent cultures of WT (YB127) and ΔstpX (YB5231) were grown to early saturation in PYE, their ODs normalized, and subcultured 10-fold into 3 ml of fresh PYE medium in a 12 well plate. Plates were incubated for 5 h at 30°C. Then, 1.5 ml of each culture was washed 3 times in ICP-MS Buffer (0.1 M NaHCO3 pH 8, filtered after treatment with 10 g l−1 of Amberlite CG-120 cation exchange resin (Mallinckrodt, St Louis, MO) to wash excess metal ions from the samples. Pellets were then resuspended in 300 μl ICP-MS Buffer containing 0.1% (v/v) Triton X-100 and 1 μl ml−1 Ready Lyse lysozyme (Epicentre, Madison, WI). The samples were incubated at room temperature for 1.5 h until lysis was complete. Fifty microlitres of the lysates was removed for protein quantification using the Pierce BCA Protein Assay kit (Thermo, Rockford, IL). Two hundred microlitres of the lysates was added to 2.8 ml of 0.1% (v/v) HNO3 for mass spectrometry.

ICP-mass spectrometry analysis was performed using a PerkinElmer ELAN DRCII ICP-MS as described previously (Jacobsen et al., 2011). Germanium at 50 p.p.b. was added as an internal standard using an EzyFit glass mixing chamber. For each sample, metal quantities measured by ICP-MS in μg were normalized to the protein concentration of that sample.

Zinc toxicity analysis

WT (YB127), ΔstpX (YB5231) or Δper stpX (YB5229) cells were grown to exponential phase in PYE medium, their optical densities were normalized, and they were allowed to grow again until early saturation. Cells were subcultured 10-fold into 3 ml of fresh PYE medium in 12 well plates, supplemented with 0, 1 or 2 mM zinc chloride (Sigma-Aldrich, St Louis, MO). The plates were incubated at 30°C without shaking. At 0, 5 or 20 h, 100 μl were drawn from each sample and serial dilutions were made and spotted onto PYE plates to calculate the number of cfu per ml of original sample.

To measure the relative amount of Zn2+ in WT and ΔstpX cells under these conditions, we used the zinc-specific fluorescent reporter zinquin ethyl ester (Sigma-Aldrich, St Louis, MO) (Ammendola et al., 2007). WT and ΔstpX cells were grown to exponential phase in PYE, their optical densities were normalized, and they were allowed to grow to early saturation in the presence of 50 μM zinquin ethyl ester. They were then subcultured 10-fold into 3 ml of fresh PYE + 50 μM zinquin ethyl ester, with or without 1 mM Zn2+, and incubated at 30°C in a 12 well plate without shaking. After 5 h, 1 ml of each sample was withdrawn, washed twice in fresh PYE, and imaged for zinquin fluorescence in the DAPI channel.


We thank members of the Brun lab for critical reading of this manuscript. We thank Martin Thanbichler and Kathleen Ryan for strains and plasmids, Patrick Viollier for providing us antibodies against StpX, Adrien Ducret for his help with image analysis, and Sapna Mehta, Ellen Quardokus and Evelyn Toh for assistance with strain construction. Photobleaching analysis was performed using instrumentation at the IUB Light Microscopy Imaging Center. Proteomic analysis was done using instrumentation at the IUB Laboratory for Biological Mass Spectrometry. DNA sequencing was performed with help from the Indiana Molecular Biology Institute, Bloomington. STEM/EDS was performed with help from David Morgan using instrumentation at the IUB Electron Microscopy Center. This work was supported by NIH grants GM051986 to Y. V. B. and GM042569 to D. P. G. The authors of this study declare no conflicts of interest.