Phosphate Starvation Triggers Production and Secretion of an Extracellular Lipoprotein in Caulobacter crescentus

Life in oligotrophic environments necessitates quick adaptive responses to a sudden lack of nutrients. Secretion of specific degradative enzymes into the extracellular medium is a means to mobilize the required nutrient from nearby sources. The aquatic bacterium Caulobacter crescentus must often face changes in its environment such as phosphate limitation. Evidence reported in this paper indicates that under phosphate starvation, C. crescentus produces a membrane surface-anchored lipoprotein named ElpS subsequently released into the extracellular medium. A complete set of 12 genes encoding a type II secretion system (T2SS) is located adjacent to the elpS locus in the C. crescentus genome. Deletion of this T2SS impairs release of ElpS in the environment, which surprisingly remains present at the cell surface, indicating that the T2SS is not involved in the translocation of ElpS to the outer membrane but rather in its release. Accordingly, treatment with protease inhibitors prevents release of ElpS in the extracellular medium suggesting that ElpS secretion relies on a T2SS-secreted protease. Finally, secretion of ElpS is associated with an increase in alkaline phosphatase activity in culture supernatants, suggesting a role of the secreted protein in inorganic phosphate mobilization. In conlusion, we have shown that upon phosphate starvation, C. crescentus produces an outer membrane bound lipoprotein, ElpS, which is further cleaved and released in the extracellular medium in a T2SS-dependent manner. Our data suggest that ElpS is associated with an alkaline phosphatase activity, thereby allowing the bacterium to gather inorganic phosphates from a poor environment.


Introduction
Caulobacter species live in all kinds of aquatic environments where nutrients are often scarce. While some nutrients are directly usable by the bacteria, other organic substrates must be partially degraded or modified before being assimilated by the cell as a nutrient source. These processes often rely on degradative enzymes that are secreted at the cell surface or in the extracellular environment depending on specialized secretion systems. The Type II secretion system (T2SS) or main terminal branch of General secretion pathway (Gsp) is well known for its role in virulence of pathogenic species but is also involved in environmental adaptation. A T2SS is present in at least 16 environmental non-pathogenic bacteria [1] where it promotes secretion of enzymes such as lipases in Acinetobacter calcoceticus and Pseudomonas alcaligenes [2,3] or chitinase in Escherichia coli K12 [4]. The T2SS apparatus is composed of at least 12 proteins, and mediates a twostep translocation of secreted proteins across the cell envelope. Briefly, unfolded exoproteins are first exported by the Sec or Tat pathways into the periplasm where they undergo maturation. They are then translocated across the outer membrane through secretin pores [5,6,7,8]. The T2SS was first described in Klebsiella oxytoca where it triggers export across the outer membrane of the starch-hydrolysing lipoprotein pullulanase [9]. Pullulanase is a surface anchored lipoprotein partially released in the extracellular environment through the formation of micelles [10]. However, the extracellular release of bacterial surface lipoproteins could occur differently. Recent studies have thus shown that the autotransporter NalP of Neisseria meningitides was responsible for the proteolytic cleavage of lactoferrin-binding protein LbpB from the cell surface [11].
Although T2SS-encoding genes are generally well conserved and display a similar genetic organization in many bacterial species, secreted proteins have a wide variety of functions or enzymatic activities [4,12,13].
Phosphorus, which can be limiting for Caulobacter survival, is present in soil and water in organic and inorganic forms. Inorganic phosphate (P i ) is the preferential source and is readily taken up by the cell when available. Since nutrient-poor environments can also lack P i , many bacterial species have developed mechanisms to use organic phosphates as a source of phosphorus [14]. While some organic phosphates such as dNTPs may be directly incorporated by the cells, others may require extracellular P i release before being taken up [15]. Under phosphate starvation, C. crescentus strikingly increases stalk length, thereby optimizing the surface/ volume ratio and in turn facilitating phosphate uptake [16,17]. This process depends on Pst and PhoB [18], which both belong to the pho regulon best characterized in Escherichia coli. The pho regulon [19] also encompasses phoA, encoding an alkaline phosphatase required for the utilization of organic phosphates [20]. C. crescentus harbors the pho regulon but lacks a phoA homolog.
In the present study, we report the ability of C. crescentus to trigger production and T2SS-dependent secretion of a lipoprotein, ElpS, under phosphate starvation conditions. We show that ElpS is an extracellular outer membrane-anchored protein subsequently released in the environment in a T2SS-dependent manner. Functional data suggest that ElpS is involved in phosphate mobilization and stimulates extracellular alkaline phosphatase activity in low-phosphate medium.

C. crescentus genome harbours a complete set of T2SS genes
Although T2SS has been largely studied in c-proteobacteria it appears that many a-proteobacteria possess T2SS genes (e.g., Bradyrhizobium, Gluconacetobacter, Mesorhizobium) [21]. In silico analysis of the C. crescentus (CB15N) genome reveals the presence of an entire set of T2SS genes, gspC (cc0172) to gspN (cc0184) annotated in alphabetical order (Fig. 1). This gene cluster organization is well conserved in species producing a T2SS apparatus. An additional open reading frame (ORF), gspO, homologous to the prepilin peptidase required for processing of Type IV secretion system prepilin, is present downstream of the gspN gene but in the opposite orientation, indicating that gspO is probably transcribed independently of the rest of the T2SS genes. Interestingly, two ORFs, cc0170 and cc0171, encoding a hypothetical protein and a TonB-dependent receptor, respectively, are found immediately upstream of the T2SS genes ( Fig. 1). cc0170 is conserved among a large number (122 Blastp hits) of bacteria living in a wild variety of environments such as Shewanella species living in ocean or Bacteroides species found in the gut. This high degree of conservation implies that cc0170 probably plays an important role in bacterial survival in different environments. Sequence analysis reveals that cc0170 is homologous to putative lipoproteins or secreted proteins, indicating that this gene likely encodes an extracellular protein. Interestingly, the tandem organization of cc0170 and T2SS genes found in C. crescentus is also conserved in others species such as the plant pathogen Xhanthomonas campestris and the marine species Sphingomonas alaskensis and Hyphomonas neptunium in a similar genetic organization (Fig. 1), suggesting that cc0170 encodes the first characterized T2SS substrate in C. crescentus.
T2SS genes are expressed in poor and rich media T2SS gene expression was assessed by measuring gspE and gspL mRNA levels in CB15N cells grown in rich culture medium (PYE). gspE and gspL encode a putative cytoplasmic ATPase and an inner membrane protein promoting secretion pore opening, respectively. gspE and gspL mRNA levels are represented as cycle threshold (CT) ratios calculated from qRT-PCR raw data (see Methods).
Both gspE and gspL are expressed in rich medium, since their CT is significantly decreased in CB15N relative to CB15N DgspC-N in which the whole operon was knocked out ( Fig. 2A). In order to generate more biologically relevant conditions, cells were grown in 1/5X PYE, hereafter referred to as poor medium. Under these conditions, gspE and gspL expression undergoes a 1.7 fold increase (Fig. 2B). To confirm these observations at the protein level, we constructed a strain carrying a mgfp-gspL fusion as the unique chromosomal copy and analyzed mGFP-GspL production by anti-GFP immunoblotting. mGFP-GspL is detected in rich and poor media at similar levels ( Fig. 2C).

T2SS triggers ElpS secretion in poor medium
The potential role of the T2SS was first determined by comparing protein secretion profiles between CB15N and CB15N DgspC-N in rich, poor and minimal media. Briefly, extracellular proteins were precipitated, separated on polyacrylamide gel and stained with silver nitrate. C. crescentus secretes a large amount of proteins in all experimental conditions but with different patterns (data not shown). Remarkably, one protein of about 50 kDa is strongly produced and secreted from CB15N cells grown in poor medium (Fig. 3A). The corresponding band is absent in CB15N DgspC-N culture supernatant, suggesting that its secretion relies on T2SS. This protein was identified by mass spectrometry as Cc0170, hereafter named ElpS for Extracellular lipoprotein produced under phosphate Starvation (Fig. 3B). As a lysis control, the cytoplasmic response regulator DivK was detected in cellular fractions but not in supernatants, indicating that secretion profiles indeed reflect protein secretion (Fig. 3C). Furthermore, qRT-PCR performed on CB15N grown in rich and poor media reveals a 3.5 fold upregulation of elpS in poor medium Genes are shown as boxes with arrowheads to indicate their predicted transcriptional orientation. Genes encoding T2SS apparatus are named C to O. Sequences and alignments were obtained from the CMR website (http://cmr.jcvi.org). doi:10.1371/journal.pone.0014198.g001 (Fig. 3D). The induction of elpS in poor medium was confirmed by a b-galactosidase assay performed on a strain carrying the lacZ reporter gene from E. coli K12 under the control of the elpS promoter (Fig. 3E). Moreover, no significant b-galactosidase activity was found in CB15N grown in poor and rich media confirming that the observed b-galactosidase activity corresponds to a specific activation of the elpS promoter. In silico analysis of ElpS sequence (http://www.cbs.dtu.dk) highlights a conserved lipoprotein signal at the N-terminus (Fig. 3B). The lipoprotein signal was predicted using LipoP algorithm based on the Junker et al. method [22]. In particular, ElpS sequence is predicted to harbor a cleavage site for the Signal Peptidase II specific for lipoproteins maturation between aminoacid 19 and 20 with a score of 31.235 against a cutoff of 15.771. The authors also refers to a lipobox found at position -3 to +1 from the modified cystein with the following consensus sequence L-(A,S)-(G,A)-C. Accordingly, ElpS bears a lipobox at position 17-20 corresponding to a LAGC sequence. Nevertheless no other functional domains were found. Overexpression of elpS-3flag from the strong lac promoter on the low-copy pMR10 vector is toxic for cells lacking T2SS apparatus as evidenced by DivK release into the supernatant (data not shown), suggesting that a functional interaction may exist between ElpS and the T2SS apparatus.
Interestingly, ElpS-3Flag, expressed from the chromosome under its own promoter, is only produced in poor medium (Fig. 4A). No ElpS-3Flag could be detected in cellular extracts or supernatants from bacteria grown in rich or minimal media. This observation is in accordance with the transcriptional data previously presented showing that elpS is strongly upregulated in poor medium. ElpS-3flag accumulates in CB15N DgspC-N cells and is not released into the culture supernatant (Fig. 4A), indicating that the T2SS proteins are required for ElpS secretion. The absence of cytosolic DivK and MreB in the extracellular fraction from both cell lines rules out cell lysis (Fig. 4B).
Extracellular ElpS-3Flag is shorter in size than its cellular counterpart. This can be explained by a cleavage of the signal peptide by the Sec machinery during the export, which is generally observed with T2SS secreted proteins [23].
ElpS is specifically produced and secreted under phosphate starvation C. crescentus does not produce any ElpS-3Flag when grown in minimal medium, which is surprising if we consider that minimal medium has lower nutrient content than poor medium. This observation implies that poor medium is devoid of a molecule repressing ElpS-3Flag synthesis, whereas this molecule is not limiting in minimal medium. The only nutrients more abundant in minimal medium than in poor medium are phosphate and ammonium salts (M2 salts), and FeSO 4 . Complementation of poor medium with M2 salts completely inhibited the production and the secretion of ElpS-3Flag, while 1 mM FeSO 4 had no effect (Fig. 5A). Accordingly, elpS is not upregulated in iron-depleted medium (data not shown). In order to test which M2 salts component specifically represses ElpS synthesis, cells carrying elpS-3flag were grown in poor medium supplemented with phosphate or ammonium salts. The addition of NH 4 Cl had no effect on ElpS-3Flag level whereas phosphate salts (Na 2 HPO 4 and KH 2 PO 4 ) significantly reduced the production of ElpS-3Flag in both CB15N and CB15N DgspC-N strains (Fig. 5B). Accordingly, depletion of phosphate salts in minimal medium induces production of ElpS-3Flag in both strains (Fig. 5C). These results were confirmed using media containing a decrease range of phosphate concentrations. Briefly, ElpS-3Flag production is repressed in rich medium and in balanced phosphate HIGG but not in low phosphate HIGG medium (Fig. 5D). As expected, ElpS-3Flag is also secreted in low phosphate HIGG when T2SS is present (Fig. 5E). Taken together, these results demonstrate that production and secretion of ElpS are specifically upregulated under phosphate starvation.

Intracellular ElpS is a membrane lipoprotein
In silico analysis reveals a conserved lipoprotein sequence in the ElpS N-terminal domain. Precursors of lipoproteins are generally translocated as non-modified proteins through the inner membrane into the periplasm where lipidic groups are then added [24]. In order to verify the status of ElpS as a lipoprotein, CB15N, CB15N elpS::elpS-3flag, CB15N DgspC-N elpS::elpS-3flag and CB15N DelpS strains were grown in poor medium containing 3 H-palmitic acid. Thereby, palmitoylated proteins will contain the isotope. Anti-Flag immunoprecipitates from cellular extracts or culture supernatants were subjected to SDS-PAGE and analysed by autoradiography. Cellular ElpS-3Flag appears to be palmitoylated irrespective of T2SS presence (Fig. 6A), whereas no signal is observed in the supernatant fraction. Accordingly, N-terminal sequencing performed on the secreted isoform of ElpS-3Flag indicates that the protein is cleaved at position 38 downstream the predicted fattyacylated cysteine (Fig. 6B). These observations suggest a loss or a cleavage of the lipidic group when ElpS-3Flag is released in the extracellular medium. Lipoproteins are generally thought to be membrane-associated. Subcellular fractionation of CB15N elp-S::elpS-3flag and CB15N DgspC-N elpS::elpS-3flag grown in poor medium confirmed the presence of ElpS-3Flag in the membrane fraction only, irrespective of T2SS presence (Fig. 6C). Moreover, additional fractionation assays showed that ElpS-3Flag is associated with the outer-membrane fraction and is never detected in the inner-membrane and cytoplasmic fractions (Fig. 6D).

ElpS is a membrane surface-anchored protein subsequently released in the extracellular medium
To see whether ElpS is surface-exposed, ElpS-3Flag-producing CB15N elpS::elpS-3flag and CB15N DgspC-N elpS::elpS-3flag cells were treated with proteinase K to degrade membrane surfaceexposed proteins. As shown in Fig. 7A, whereas the level of the cytoplasmic protein MreB remains stable, ElpS-3Flag signal entirely disappears in total extracts obtained from both proteinase K-treated cell lines, indicating that ElpS-3Flag is located at the external face of the outer membrane independently of T2SS. Similar treatment was performed on lysed cells without previous inactivation of proteinase K. In this case, proteinase K treatment led to a complete loss of cellular proteins including MreB (data not shown). Interestingly, treatment of CB15N elpS::elpS-3flag grown in poor medium with proteases inhibitors prevents ElpS release in the extracellular medium (Fig. 7B). Taken together, these data suggest that ElpS is independently sorted to bacterial surface and is subsequently cleaved by a T2SS-secreted protease and released in the environment.

ElpS is associated with an alkaline phosphatase activity
Since ElpS is specifically synthesized in response to phosphate starvation, it may promote release of P i from organophosphate sources. Among the potential enzymatic activities responsible for P i release from various substrates, alkaline phosphatase activity was measured in 100-fold concentrated culture supernatants using pnitrophenyl phosphate as a substrate. The level of alkaline phosphatase activity remains unchanged in CB15N DelpS. In contrast, overexpression of elpS from pMR10 under the control of the lac promoter in CB15N DelpS leads to a significant increase of the alkaline phosphatase activity (Fig. 8A). These results suggest that other alkaline phosphatases masks or compensates a loss of phosphatase activity in the absence of ElpS synthesis. The absence of alkaline phosphatase activity in re-folded ElpS purified under denaturating conditions from E. coli and in extracellular immunoprecipitated ElpS-3Flag fractions from C. crescentus grown in poor medium suggests that ElpS is not an alkaline phosphatase itself but rather an activator of another still unknown phosphatase. The absence of the cytoplasmic MreB in the culture supernatants rules out any potential cell lysis that might explain the increase in alkaline phosphatase activity in the elpSoverexpressing cells supernatant (Fig. 8B).

Discussion
In this paper, we demonstrate that phosphate starvation triggers production and T2SS-dependent release in the extracellular medium of a membrane-bound lipoprotein, ElpS. The constitutive production of a T2SS apparatus in C. crescentus, irrespective of environmental conditions, suggests that T2SS-dependent secretion of ElpS is controlled at another level, such as the transcriptional level since elpS expression is strongly induced under phosphate starvation. Accordingly, in Aeromonas hydrophila and Vibrio cholerae, constitutive expression of exe and eps, which encode T2SS components, is balanced by the regulated expression of genes encoding the cognate secreted product [21]. Similarly, the growthphase-dependent secretion of AspA protease in Aeromonas hydrophila mainly relies on aspA transcriptional modulation [25]. Constitutive expression of a secretion apparatus combined with a tight regulation of secreted substrate production may offer an energetic advantage and a rapid adaptation to changing environments.
The abundance of lipoproteins in bacteria [22,26,27,28] suggests the existence of many membrane-associated activities. Yet, the functions of most of them are still unknown. Lipoproteins are bound to membrane via a fatty-acyl group linked to a conserved cysteine and are sorted via the Lol system to the inner or outer membranes based on the identity of the amino acid following the cysteine [29,30]. In particular, the presence of an aspartate following the conserved cysteine is considered as an inner membrane retention signal. Some atypical lipoproteins can also contains particular retention signals after the cystein such as CKVE/CGGG/CGDD/CQGS found in 5% of P. aeruginosa lipoproteins [31]. Other lipoproteins are conversely targeted to the outer membrane. In some cases, lipoproteins must also reach the external side of the outer membrane. Similarly to Klebsiella oxytoca pullulanase [9], ElpS is a membrane surface-anchored lipoprotein subsequently released in the environment. It is not always clear how lipoproteins are transported to the bacterial envelope [32,33]. For instance, the export of pullulanase occurs through T2SS apparatus [34] whereas Pseudomonas aeruginosa HxcQ liposecretin is self-piloted to the outer membrane [35]. Unlike pullulanase, ElpS does not exhibit any known inner membrane retention signal and is consequently supposed to be addressed to the outer membrane. However, contrary to a non-secreted variant of pullulanase that accumulates in the periplasm [36], ElpS crosses the outer membrane in a T2SS-independent manner. Our results suggest that ElpS release in the environment is mediated by another T2SS-secreted protein. Indeed, secretion of ElpS in the environment is likely due to the proteolytic cleavage of ElpS by a T2SSsecreted protease. The absence of radiolabeling with 3 H-palmitate of the extracellular ElpS isoform indicates that the exoprotein is deacylated when secreted, contrary to pullulanase [10]. Moreover, N-terminal sequencing of the secreted isoform of ElpS indicates that the protein is probably cleaved just before the asparagine 38 located downstream the fatty-acylated cysteine. Surprisingly, all detectable cellular ElpS is sensitive to proteinase K treatment, which led us to the idea that ElpS is primarily and transiently located in the outer membrane before secretion into the medium. Consistent with this hypothesis, enteropathogenic E. coli is able to accumulate an outer membrane-bound isoform of its major enterotoxin, collicin A, prior to release in its host cell [37]. Therefore, we can speculate that the release of ElpS relies on a specific signal such as an increase in population density, biofilm initiation or a drastic limiting threshold in inorganic phosphate concentration.
Phosphate is an important limiting nutrient in C. crescentus environment. C. crescentus is able to considerably increase stalk length when cells undergo phosphate starvation [16], thereby optimizing nutrient exchange [38]. Interestingly, stalk extracts harvested from C. crescentus grown in low phosphate HIGG medium contain many outer membrane proteins including Cc0171, a TonB-dependent receptor, and ElpS [39]. This particular protein content implies that the stalk must be an essential organelle for nutrient uptake and suggests that ElpS and Cc0171 play a role in the adaptation to nutrient-deficient media. The co-occurrence of elpS and cc0171 homologs in genomes probably reflects their functional interaction. Interestingly, RT PCR analysis show that, contrary to elpS, cc0171 is constitutively induced in rich and poor media suggesting that both genes are independently regulated (data not shown). TonB-dependent receptors are generally sensors of extracellular signals [40], and Cc0171 could be involved in the detection of phosphate starvation and the establishment of an adaptive response through the induction of elpS. Nevertheless, deletion of cc0171 does not suppress the production of ElpS-3Flag in poor medium (data not shown). We also assume that Cc0171 does not act on elpS induction under phosphate starvation. On the contrary, the TonBdependent receptor may be involved in phosphate uptake through the activation of ElpS. However, no significant change in intracellular phosphate concentration is detected when cells are switched from rich to poor medium which could be due to the low sensitivity of the method used (data not shown). ElpS is specifically produced under phosphate starvation and is associated with an increased secreted alkaline phosphatase activity when overproduced, Together, our data suggest that ElpS participates in phosphate mobilization.

Strains, plasmids and growth conditions
All strains of E. coli and C. crescentus were derivatives of DH10B and CB15N strains, respectively. DNA manipulations were performed according to standard techniques. Restriction enzymes were purchased from Roche, and primers were purchased from Eurogentec. Strains and plasmids are summarized in Table S1 [47] and their mode of construction is described in the Text S1. E. coli strains were grown in Luria Bertani broth medium (LB broth, Difco) supplemented with 100 mg/ml ampicillin or 50 mg/ml kanamycin when necessary. C. crescentus strains were grown in Peptone Yeast extract (PYE) [41], 1/5X PYE (5-fold diluted PYE except 1 mM MgSO 4 and 0.5 mM CaCl 2 +0.2% glucose), M2G [42], in balanced phosphate (200 mM) or low phosphate (30 mM) Hutner baseimidazole-buffered-glucose-glutamate HIGG at 30uC [43].

RNA preparation
Total RNAs were extracted from CB15N, CB15N DgspC-N and CB15N DelpS (all cultured in duplicate) as follows: 45 ml culture  (OD 660 nm = 0.5 at 30uC) were centrifuged for 15 min at 3500 rpm. Bacterial pellets were resuspended in 100 ml 10% SDS and 20 ml 20 mg/ml proteinase K and incubated for 1 h at 37uC with shaking. Five ml of TRIzolH Reagent (Invitrogen) were added and suspensions were vigorously shaken. After a 10 min incubation at 65uC, 1 ml chloroform was added to the suspensions and the mixtures were shaken and incubated for 5-10 min at room temperature. Samples were centrifuged for 15 min at 14,000 rpm at 4uC, 2.5 ml propan-2-ol were added to the aqueous phases and samples were stored overnight at -20uC. After centrifugation for 30 min at 14,000 rpm at 4uC, pellets were washed with 75% RNase-free ethanol. Supernatants were discarded and pellets were dried for 15 min at room temperature. Total RNA samples were resuspended in 100 ml RNase-free H 2 O, incubated for 10 min at 55uC and stored at 280uC. The integrity of the RNA and the absence of DNA were checked by gel electrophoresis. RNA was quantified with a NanoDrop (ND-1000, Thermo Fisher Scientific).

Quantitative real-time RT-PCR
RNA samples were obtained using DNA-free kit (Ambion) and reverse-transcription was performed with SuperScript II Reverse Transcriptase (Invitrogen). Primers listed in Table S2 were designed with PrimerExpressH 2.0 (Applied Biosystems). PCR products ranged from 80 to 100 bp. Real-time PCR reactions were performed with SYBR Green Mix (Applied Biosystems) in an Applied Biosystems Step One Plus real-time PCR instrument. Ratios were calculated using the CT method for each primer. CT (Cycle Threshold) was defined as the number of cycles required for the fluorescent signal to cross a defined threshold of signal (i.e., up to background level). CT levels were thus inversely proportional to the amount of target nucleic acid in the sample. CTs b-galactosidase assays C. crescentus strains were grown in PYE and 1/5X PYE media for 6 h at 30uC. Cells were harvested by centrifugation for 10 min at  6000 rpm and resuspended in 1/2 volume of Z-Buffer (16.1 g/l Na 2 HPO 4 .7H 2 O; 5.5 g/l NaH 2 PO 4 .7H 2 O; 0.75 g/l KCl; 0.246 g/l MgSO 4 .7H 2 O; 0.0005% NP40) and 1/185 volume of b-mercaptoethanol. Cell lysis was obtained by adding 1/10 volume of glass beads and vortexing 6 cycles of 30 seconds spaced with 30 seconds cycles on ice. Membrane and beads were separated from cytoplasmic fractions by centrifugation for 10 min at 6000 rpm. Supernatants containing soluble proteins were collected, mixed with 1/5 volume of 4 mg/ml ortho-Nitrophenyl-b-galactoside (ONPG) and incubated for 1 min to 1 h at room temperature. ONPG degradation was monitored at OD 420 nm by formation of an yellow product ortho-Nitrophenyl (ONP) and stopped by the addition of 0.3 M Na 2 CO 3 . Specific b-galactosidase activity was determined in nmol of degraded ONPG (e 420 nm = 4500 M 21 ) per min per mg of total protein. Total protein concentrations were determined using Bradford assay (Bio-Rad Protein Assay).
Preparation of cellular extracts and precipitated supernatants under denaturing conditions C. crescentus was grown in PYE, 1/5X PYE, M2G, balanced or low phosphate HIGG to OD 660 nm = 0.5 at 30uC. Culture supernatants were isolated by a 10 min centrifugation at 3500 rpm. Corresponding supernatants were filtered on 0.22 mM filters and proteins were precipitated using a pyrogallol red-molybdate-methanol (PRMM) protocol ,Caldwell, 2004 #5.. Protein pellets were resuspended in 1/150 volume of SDS-PAGE loading buffer (1 M Tris-HCl pH 6.8; 2% SDS; 10% glycerol; 16% b-mercaptoethanol; 0.06% bromophenol blue). Cellular extracts were obtained by resupending bacterial pellets in SDS-PAGE loading buffer to a final corresponding OD 660 nm = 2. All protein samples were boiled for 5 min and stored at 220uC.

Preparation of cellular extracts and supernatants under non-denaturing conditions
C. crescentus was grown in 1/5X PYE to OD 660 nm = 0.5 at 30uC except when notified. Culture supernatants were 100 foldconcentrated on Amicon (AmiconH Ultra-10k Millipore). Cellular extracts were obtained by resuspending bacterial pellets in PBS (140 mM NaCl; 2 mM KCl; 10 mM Na 2 HPO 4 ; 2 mM KH 2 PO 4 ) to a final OD 660 nm = 2 and sonicating 6 times for 30 s. Protein samples were stored at 220uC.

Silver nitrate staining and immunoblots of protein extracts
When prepared under non-denaturing conditions, samples were first mixed with 2X SDS-PAGE loading buffer and boiled for 5 min. Proteins were then separated on 12% polyacrylamide gels and detected by silver nitrate staining or transferred to a PVDF membrane for immunoblotting. For silver nitrate staining, gels were first fixed for 20 min in 50% methanol, 5% acetic acid, then washed for 10 min in 50% methanol and for 2 h in H 2 O. Gels were incubated in sensitizing solution containing 0.02% sodium thiosulfate and washed again twice for 1 min in H 2 O. Proteins were revealed by incubating the gels another 20 min in 0.1% silver nitrate followed by 2 washes for 1 min in H 2 O and one last incubation in 0.04% formalin, 2% sodium carbonate. The staining reaction was stopped in 5% acetic acid. ElpS-3Flag, DivK, MreB, mGFP-GspL, FliF and CcrM were immunodetected with 1: Protein radiolabeling with 3 H-palmitic acid PYE precultures at OD 660 nm = 0.1 were diluted in 1/5X PYE containing 50 mCi/ml 3 H-palmitic acid and grown for 6 h at 30uC. Cellular extracts and supernatants were prepared under non-denaturing conditions as above. ElpS-3Flag was immunoprecipitated with 1:1000 anti-Flag antibodies (Monoclonal anti-FlagH M2, Sigma-Aldrich) and Protein A -SepharoseH 4B (Invitrogen). Immunoprecipitates were resuspended in 100 ml SDS-PAGE loading buffer and concentrated supernatants were mixed with 2X SDS-PAGE loading buffer. Proteins were boiled for 5 min and 10 ml of each sample were separated on 12% polyacrylamide gels. Radioactive signal was enhanced using fluorography [46]. ElpS-3Flag was immunodetected with rabbit anti-Flag antibodies.

N-terminal sequencing
Proteins extracts were obtained from 10000-fold concentrated supernatants on Amicon from CB15N elpS::elpS-3flag and CB15N DelpS grown for 6 h in 1/5X PYE, separated on 12% polyacrylamide gel and transferred to PVDF membrane. Proteins were detected by Coomassie blue staining using the Pierce Coomassie Plus Protein Assay Reagent and Kit (Thermo Scientific) and ElpS-3Flag was immunodetected with anti-Flag antibodies. The band corresponding to ElpS-3Flag was cut through the simultaneous analysis of total proteins profiles and immunoblot. The N-terminal end sequence of ElpS was obtained from Pick-n-Post (http://www.pick-n-post.com).

Cell fractionation assays
Cells (400 ml) were grown to OD 660 nm = 0.5 in 1/5 X PYE at 30uC. Cells were harvested by centrifugation at 4uC, washed in 50 ml CBB buffer (20 mM Tris-Cl pH 8.0, 25 mM NaCl, 5 mM EDTA, 3.6 mM b-mercaptoethanol), and resuspended in 10 ml CBB buffer with a protease inhibitor cocktail (Complete EDTA free, Roche). Cells were passed twice through a French pressure cell (,20-25,000 psi) and the lysate was cleared by centrifugation at 20,0006 g for 15 min at 4uC. The cleared lysate was then centrifuged at 150,0006 g for 2 h at 4uC, and a sample of the supernatant was taken for analysis. The pellet was resuspended in 1 ml CBB buffer and spun at 105,0006g for 1 h. The supernatant was discarded and the washed pellet was resuspended in 1 ml CBB with 0.5% SDS and used for analysis. The protein content of the cleared lysate was measured using a Bradford assay and was used to normalize load volumes between samples. Within a fractionation sample, equivalent volumes were loaded on gels. Samples were mixed with 5X SDS-PAGE loading buffer, boiled for 5 min, separated on 12% polyacrylamide gels, transferred to PVDF membranes, and blotted with anti-Flag, anti-FliF and anti-CcrM antibodies.

Inner and outer membranes preparation
Cells (300 ml) were grown to OD 660 nm = 0.5 in 1/5 X PYE at 30uC. Cells were harvested by centrifugation at 4uC, washed 3 times in 50 ml 50 mM ammonium bicarbonate (AmBic) pH 8 and resuspended in 5 ml AmBic. Cells were sonicated at maximal intensity 20 times for 5 s and kept on ice between each cycle. The lysate was cleared by centrifugation for 10 min at 12,0006 g at 4uC. The cleared lysate was then centrifuged for 40 min at 100,0006 g at 4uC, and a sample of the supernatant was taken for analysis (cytoplasmic fraction). The pellet was resuspended in 1 ml 1% sodium lauryl sarkonisate and spun for 40 min at 100,0006 g. The supernatant corresponding to solubilized inner membrane proteins was taken for analysis and the pellet was resuspended in 1 ml 2.5 M NaBr and incubated for 30 min at 4uC and then ultracentrifuged for 40 min at 100,0006 g. The outer membrane pellet was resuspended in 1 ml 100 mM Na 2 CO 3 and spun again for 40 min at 100,000 g 6. The OM proteins was resuspended in SDS-PAGE loading buffer. Within a fractionation sample, equivalent volumes were loaded on gels. Samples were mixed with 2X SDS-PAGE loading buffer, boiled for 5 min, separated on 12% polyacrylamide gels, transferred to nitrocellulose membrane, and blotted with anti-Flag antibodies.

Proteinase K sensitivity assays
Cells were grown to OD 660 nm = 0.5 in 1/5X PYE at 30uC. Cultures were treated or not with 2 mg/ml proteinase K for 30 min at 37uC. Cells were harvested by centrifugation for 10 min at 3500 rmp, the supernatant was discarded and the pellet washed in 1 volume of M2 salts (10 mM NH 4 Cl, 6 mM Na 2 HPO 4 and 4 mM KH 2 PO 4 ). Cells were finally resuspended in 1/4 volume of M2 salt containing 10 mM phenylmethanesulphonylfluoride (PMSF). Samples were mixed with 2X SDS-PAGE loading buffer, separated on a 12% polyacrylamide gel and transferred to nitrocellulose membrane. ElpS-3Flag and MreB were immunodetected using anti-Flag and anti-MreB antibodies respectively.

Anti-proteases treatment on cells producing ElpS-3Flag
CB15N elpS::elpS-3flag was grown in 1/5X PYE for 4.5 h at 30uC. Cells were then harvested by centrifugation for 10 min at 3500 rpm and resuspended in an equal volume of fresh 1/5X PYE containing or not a cocktail of protease inhibitors (Complete, EDTA-free; Roche) and incubated for 90 min at 30uC. ElpS-3Flag was immunodetected using anti-Flag antibodies in cellular extracts and precipitated supernatant from resulting samples.

Alkaline phosphatase assays on concentrated supernatants
Alkaline phosphatase assays were performed on 100 foldconcentrated supernatants obtained under non-denaturating conditions from cells grown for 4 h in 1/5X PYE. Specific alkaline phosphatase activity was measured by degradation of 1 mg/ml p-Nitrophenyl Phosphate (PNPP) in 100 mM pH 7.9 Tris-HCl, 2 mM MgCl2. P-Nitrophenol formation induced by PNPP cleavage was monitored at OD420 nm. Alkaline phosphatase activity was calculated in nmol of degradated PNPP (e420 nm = 18000 M-1)/min/mg of total protein. Total protein concentrations were determined using micro-Bradford assays (Bio-Rad Protein Assay).