| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Dipartimento di Biologia Università ‘Tor Vergata’, via della Ricerca Scientifica, Roma 00133, Italy
2 Laboratorio di Biologia e Patologia Molecolari del CNR di Roma, Italy
Correspondence
L. Paolozzi
Paolozzi{at}uniroma2.it
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
; Weiss, 2004; Vicente & Rico, 2006; Vicente et al., 2006).
In Escherichia coli, cell division requires the assembly of at least 15 proteins at the mid-cell (Buddelmeijer & Beckwith, 2002; Bernhardt & de Boer, 2003). These proteins belong to various functional groups that partecipate in invagination, constriction of the three envelope layers and separation of the two daughter cells. Among these proteins, 10 are essential under standard laboratory conditions and their localization occurs according to a largely linear hierarchy: FtsZ>(FtsA, ZipA)>FtsK>FtsQ>(FtsL, FtsB)>FtsW>FtsI>FtsN.
Within this hierarchy, a given protein requires all upstream proteins (to the left) to localize and is, in turn, required for the localization of proteins further downstream (to the right) (Chen & Beckwith, 2001; Buddelmeijer & Beckwith, 2002).
To initiate cell division, the GTP-binding tubulin-like FtsZ protein forms an intracellular ring at the division site localized equidistant between the two cell poles, by FtsZ monomer self-assembly (Lutkenhaus & Addinall, 1997; Lowe & Amos, 1998; for a review, see Errington et al., 2003).The Z-ring, besides its dynamic role in cytokinesis (assembly at the beginning of the division process and disassembly at the end of the division process) serves as a cytoskeletal scaffold for the recruitment of proteins of the cell division machinery (Bi & Lutkenhaus, 1991; Romberg & Levin, 2003).
Once both FtsA and ZipA are in place, the remaining proteins are recruited. These include bitopic (FtsQ, FtsL, FtsB, FtsN, FtsI) and polytopic (FtsK, FtsW) membrane proteins. The first group is characterized by a short cytoplasmic N-terminus, a single transmembrane segment and a large periplasmic domain (Guzman et al., 1992, 1997; Dai et al., 1996; Bowler & Spratt, 1989). In the second group, FtsK is a large protein, highly conserved throughout the eubacteria, with distinct roles in cell division and in chromosome segregation (Wang & Lutkenhaus, 1998) and FtsW, a member of the SEDS (shape, elongation, division and sporulation) family of proteins (Ikeda et al., 1989; Henriques et al., 1998). AmiC and EnvC are two peptidoglycan hydrolases that localize to the septal ring and play an important role in the separation of daughter cells (Bernhardt & de Boer, 2003).
The mechanism(s) governing the localization of these late proteins, and how they give rise to the assembly pathway, remain poorly understood. In fact several findings indicate that the pathway may not be as linear as first expected. Bacterial two-hybrid analysis suggests a highly interconnected network of proteins (Di Lallo et al., 2003; Karimova et al., 2005), and at least some cell division proteins assemble independently of their normal association with the divisome (Buddelmeijer & Beckwith, 2004; Goehring et al., 2005). Moreover, several of the division components, including ZipA and FtsK, can be completely bypassed via suppressor mutations or overexpression of other divisome components (Geissler et al., 2003; Geissler & Margolin, 2005).
Goehring et al. (2005) have recently developed a method to circumvent the need for pre-assembly of upstream proteins at the ring for the recruitment of downstream ring components. The first results obtained using this procedure indicated that FtsQ can coordinate the recruitment of the downstream proteins, except FtsN, in the absence of the upstream FtsA and FtsK proteins (Goehring et al., 2005). It was also observed that prematurely localized FtsQ can back-recruit FtsK even in the absence of the upstream protein FtsA. This observation suggested a direct protein–protein interaction between FtsQ and FtsK in the division complex and was the first in vivo assay to show back-recruitment of division proteins.
Understanding how all these proteins interact with each other is crucial to elucidate the molecular mechanism of their recruitment and assembly at the division site. The interaction web among FtsZ, FtsA, ZipA, FtsK, FtsQ, FtsL, FtsW, FtsI and FtsN was depicted by means of the two-hybrid assay (Di Lallo et al., 2003). The interaction between FtsB and FtsQ is described in this paper.
This interaction web is complex because some division proteins, such as FtsQ, the subject of this work, have many partners. FtsQ, besides homodimerizing, interacts with FtsI, FtsL, FtsN, FtsK, FtsW (Di Lallo et al., 2003; Karimova et al., 2005) and FtsB (this work). Interactions between FtsQ, FtsL and FtsB were recently confirmed by biochemical methods (Buddelmeijer & Beckwith, 2004), and interactions between FtsQ and FtsI, and between FtsL and FtsN, by another prokaryotic two-hybrid assay (Karimova et al., 2005).
FtsQ (Chen et al., 1999) is a protein with a short (24 aa) cytoplasmic N-terminus, a single (25 aa) transmembrane segment, and a large (227 aa) periplasmic domain (Carson et al., 1991; Guzman et al., 1997). It has a polypeptide transport-associated (POTRA) domain which is functionally associated with protein interactions or chaperone function (Sanchez-Pulido et al., 2003). In E. coli, FtsQ assembles with FtsL and FtsB into a trimeric complex before their localization (Buddelmeijer & Beckwith, 2004), while in Streptococcus pneumoniae their homologues form a transient complex during septation (Noirclerc-Savoye et al., 2005).
Despite its high number of interactions with the other division proteins and its role in localization of some of them, the exact role of FtsQ is still unknown. It has been proposed that FtsQ provides a structural link between early- and late-assembling divisomal components (Di Lallo et al., 2003) and could regulate the assembly of these division proteins at the division site and the activity of the peptidoglycan assembly machinery within the divisome (Piette et al., 2004). In addition, it was recently proposed that the FtsQ POTRA domain might function as a chaperone that specifically recognizes secretion- or assembly-competent forms of these polypeptides, suggesting that FtsQ could have a role in controlling the correct assembly of the divisome.
The aim of this work was not only to identify the FtsQ domains involved in the interaction with the other partner proteins but also to investigate the biological significance of these interactions in order to elucidate their assembly and participation in the divisome construction by also exploiting the possible capacity of orthologous proteins to complement an E. coli ftsQ null mutant.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The antibiotics used (Sigma) were ampicillin (50 µg ml–1), chloramphenicol (34 µg ml–1) and kanamycin (30 µg ml–1). Synthetic oligonucleotides used in this work are listed in Table 1.
|
. Recombinant plasmids were constructed by cloning the genes (or gene fragments) of interest in the SalI and BamHI restriction sites of pcIP22 and pcI434 vectors. The DNA of the gene of interest was obtained by PCR amplification using specific oligonucleotides carrying compatible restriction sites at the ends. All the constructs were controlled by DNA sequencing.
|
-Galactosidase assay and two-hybrid assay.-Galactosidase activity was assayed as described by Miller (1972). The two-hybrid assay was performed on bacterial cultures grown to OD600 0.5 at 34 °C in LB medium supplemented with 1x10–4 M IPTG, as described previously (Di Lallo et al., 2001).
General microbiological and recombinant DNA techniques.
Standard microbiological techniques were as described by Miller (1972). Standard procedures were used for small-scale plasmid preparations, endonuclease digestion, ligation, agarose gel electrophoresis, the elution of DNA fragments from agarose and bacterial transformation (Sambrook et al., 1989). PCR was carried out using the Taq DNA polymerase kit (Promega), according to the recommendations of the manufacturer.
Construction of ftsQD8.
In order to obtain a mutated FtsQ protein in which 8 aa, in the region between 147 and 171 bp of the gene sequence, were deleted and replaced by 2 aa derived by religation of the XbaI site, two ftsQ fragments (1–147 bp and 148–831 bp) were amplified with oligos 1 and 10, and 6 and 19, respectively. After purification and digestion with XbaI, the two DNA fragments were joined together with T4 DNA ligase and then reamplified with oligos 1–19. The PCR fragment obtained was purified, digested with SalI and BglII and inserted into plasmid pcIp22 digested with SalI and BamHI. The construct was then confirmed by DNA sequencing.
Construction of Q1Q2 plasmid.
For the construction of plasmid pcIP22-ftsQ1Q2, the ftsQ gene was divided into two fragments, Q1 and Q2 (Fig. 1). The Q1 fragment (1–408) was amplified by PCR using primers 1 and 16; the resulting fragment contains the SalI site at the 5' end and the PstI and NotI sites at the 3' end. The Q2 fragment (409–831 bp), amplified with primers 8 and 20, contains the NotI and XhoI sites at the 5' end and the BglII site at the 3' end. After digestion with NotI, the two fragments were joined together with T4 DNA ligase and reamplified with the external primers 1 and 20.
|
For the construction of plasmid pcIP22-ftsQ2, containing only the Q2 fragment, the sequence ftsN1–159-Q2 was amplified from pcIP22-ftsQ1Q2 with primers 25 and 20, digested with PstI and BglII and cloned into the vector pcIP22 previously digested with the same enzymes. The constructs were confirmed by DNA sequencing.
Construction of GFP and GST derivatives for co-immunoprecipitation studies.
The division proteins fused with GFP (green fluorescent protein of Aequorea victoria) were obtained by in-frame cloning of the PCR-produced DNA fragment of the corresponding gene downstream of the GFP gene in the plasmid pTTQ18-GFP (V. D'Ulisse and others, unpublished). Analogously, the GST (glutathione S-transferase of Schistosoma japonicum) derivatives were obtained by cloning the gene of interest downstream of the GST gene in the plasmid pBad33-GST (V. D'Ulisse and others, unpublished).
Preparation of extracts for co-immunoprecipitation experiments.
Cultures of E. coli DH5
harbouring the recombinant plasmids containing the genes of interest were fused in-frame to GST and GFP or to the 
repressor cI and GFP, respectively. Cultures were grown in LB supplemented with the appropriate antibiotics to OD600 0.3–0.4. The expression of the tagged genes was induced by adding 1x10–4 M IPTG (for the GFP or cI-fused genes) and 0.2 % arabinose (for the GST-fused gene) and the culture was grown for 4 h. The culture was then centrifuged, washed and resuspended (1/300, v/v) in lysis buffer (EDTA 1 mM pH 8, HEPES 25 mM pH 7.6, 0.1 mg lysozyme ml–1) and cooled on ice for 30 min. Membrane fractions were then prepared as described by Buddelmeijer & Beckwith (2004). Soluble fraction was prepared by centrifuging the lysed sample at 50 000 g for 45 min and recovering the supernatant.
Co-immunoprecipitation studies.
Immunoprecipitation experiments were similar to those described by Duong & Wickner (1997). Protein concentration was determinated by the Bradford method.
The beads (Protein A-Sepharose) were divided into aliquots of 50–100 µl in microcentrifuge tubes; to each tube, 10 µl GFP mouse monoclonal antibodies (1 : 100) (Santa Cruz Biotechnology) was added. The samples were incubated for 15–60 min at room temperature and gently mixed on a suitable shaker. After centrifugation, samples were washed with 1 ml washing buffer (20 mM HEPES buffer, pH 7.5, containing 150 mM NaCl, 0.1 % Triton X-100 and 10 %, v/v, glycerol), then 0.1–1 ml cell lysate was added to each tube.
Samples were incubated from 90 min to overnight at 4 °C, with gentle mixing on a suitable shaker. Immunoprecipitated complexes were collected by centrifugation at 3000 g for 2 min at 4 °C.
Western blotting.
Electrophoresis and immunoblotting were performed as described by Laemmli (1970) and Towbin et al. (1979), respectively. The nitrocellulose membranes were probed with cI rabbit polyclonal antibodies (1 : 6000) or with GST mouse monoclonal antibodies (1 : 1000), depending on the specific cI- or GST-fused protein used in the experiment, and detected with ECF reagents (Amersham Pharmacia).
In some experiments, anti-GST antibodies (Santa Cruz Biotechnology) were used instead of anti-cI since the latter were no longer commercially available. Antiserum against FtsQ (MV J11), kindly provided by M. Vicente (CSIC, Madrid, Spain), was pre-adsorbed onto VIP210, according to Dopazo et al., (1992), and then processed as described by Sherman & Goldberg (1992) for final steps of the purification.
Complementation of strain JOE170.
Strain JOE170/pQ and its derivatives transformed with various plasmids were grown in LB medium supplemented with 0.2 % L-arabinose and 50 µg ampicillin ml–1. After overnight incubation at 37 °C, cells were centrifuged, and resuspended in SM buffer for 1 h. After this time of starvation, bacteria were plated on LB plates supplemented or not with arabinose in order to determine c.f.u. The efficiency of plating (e.o.p.) corresponds to the ratio between the bacterial titre on LB medium and that on LB medium supplemented with arabinose.
In the case of the growth kinetics, bacteria grown to OD600 0.2 in LB supplemented with arabinose were centrifuged, washed twice and resuspendend in SM for 1 h. After starvation, bacterial cultures were diluted in prewarmed LB containing 1x10–4 M IPTG to a final concentration of about 1x104 cells ml–1 and grown at 37 °C. Over time, aliquots were withdrawn and plated on LB supplemented with arabinose.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
). To study protein–protein interactions, we used the THA, because it is the easiest and quickest method, especially when the number of interactions to be tested is high (for a review see Legrain & Selig, 2000).
To date, interactions between the whole FtsQ and the other cell division partners have been confirmed by another THA, called BACTH (Karimova et al., 2005), but only the interactions FtsQ–FtsL and FtsQ–FtsB were confirmed by biochemical experiments (Buddelmeijer & Beckwith, 2004). Therefore, to confirm and strengthen the TP and THA data (Di Lallo et al., 2003; Karimova et al., 2005) we analysed the interactions between FtsQ and the other partner proteins, namely FtsI, FtsN, FtsK and FtsW, by co-immunoprecipitation experiments.
Study of interactions between FtsQ and the other cell division partners by co-immunoprecipitation
Since we did not have at our disposal specific antibodies against these proteins, the assays were performed on the division proteins fused with tags (the repressor cI, GST or GFP), for which commercial antibodies are available. As a first step, we tested the specificity of these antibodies and our results showed that there was no cross-hybridization, i.e. the anti-cI, anti-GST and anti-GFP antibodies only recognize the cI-, GST- and GFP-fused protein, respectively (Fig. 2A, lanes 3 and 4; Fig. 2C, lane 1).
|
shows the results of the FtsQ–FtsW, FtsQ–FtsN, FtsQ–FtsI and FtsQ–FtsK co-immunoprecipitation experiments. As expected from the TP and THA, in every case an interaction between the pairs of proteins was observed.
Identification of FtsQ interaction domains by means of the THA
To identify the FtsQ domains involved in the interactions with the other cell division proteins, we constructed recombinant plasmids harbouring partial deletions of the ftsQ gene encoding truncated forms of the protein and swap constructs, where the FtsQ cytoplasmic and/or MSS (membrane-spanning segment) domains were substituted with that of other proteins (Guzman et al., 1997). These constructs were used in the prokaryotic THA (Di Lallo et al., 2001). The rationale is that if two chimeric proteins formed by the N-terminal portion of phage 434 cI repressor fused in-frame with the division protein X (cI434-X), and the N-terminal portion of phage P22 cI repressor fused in-frame with the division protein Y (cIP22-Y), form the heterodimer cI434-X/cIP22-Y by interaction of their C-terminal domains, a fully functional repressor will be formed. This repressor will be able to shut down the expression of the lacZ reporter gene under the control of the 434/P22 hybrid promoter/operator region.
In our case, X is constituted by various fragments of the ftsQ gene to form the recombinant plasmids (pcI434-X), and Y by the various division genes whose products interact with FtsQ, namely FtsB, FtsI, FtsK, FtsL, FtsN and FtsW, to form the recombinant plasmids (pcIP22-Y). Pairs of recombinant plasmids encoding these chimeric repressors were co-transformed into the recipient strain R721 carrying the reporter gene lacZ under the control of the hybrid promoter/operator 434-P22. -Galactosidase synthesis, which is constitutive in the strain without plasmids, is repressed in the presence of the two plasmids only if the two chimeric repressors interact.
Residual -galactosidase activity was evaluated for each strain and compared to that of the parental strain without plasmids. Each value reported in Figs 3 and 4 is the mean of five independent determinations. The relevance of the values has been discussed previously (Di Lallo et al., 2003). Standard deviations (not reported) were of the order of 5 % compared to the mean values of the -galactosidase activities obtained in the various determinations. In this paper, we assumed that residual -galactosidase synthesis under 50 % is indicative of repression and hence of protein interaction, whereas for synthesis above 50 %, the interaction is uncertain or negligible. The same results were obtained when the genes encoding the proteins under investigation were cloned in the reciprocal vectors, i. e. cI434–X/cIP22–Y and cI434–Y/cIP22–X (data not shown).
|
|
) were used to study the interaction of this protein with itself and the other cell division proteins. The results, reported in Figs 3 and 4, show that: (i) the cytoplasmic domain deletion does not affect the ability of FtsQ to interact with itself and FtsN, FtsI, FtsL, FtsB or FtsW (Fig. 3, row c) but results in loss of FtsK interaction; (ii) no interaction is observed when both the cytoplasmic and transmembrane domains are deleted (Fig. 3, row b), although this truncated protein is normally produced, as shown by Western blot experiments performed with anti-FtsQ antibodies (data not shown); (iii) all these interactions, again with the exception of FtsK, are restored with the swap construct in which the FtsQ MSS domain is substituted with that of a protein involved in a process totally unrelated to cell division such as the membrane protein MalF, which is required for maltose transport in E. coli (Froshauer & Beckwith, 1984; Hofnung, 1974) (Fig. 4, row a), (iv) the presence of a cytoplasmic domain is essential for FtsK interaction, but the domain itself is not specific since the FtsQ cytoplasmic domain can be substituted with that of FtsN and the interaction is conserved (Fig. 4, row c). It is important to note that FtsN does not interact with FtsK (Di Lallo et al., 2003). Moreover, since it has been reported that FtsN can interact with FtsA (Corbin et al., 2004) we studied the interactions of the swap construct NNQ with FtsA. As shown in Fig. 4(A), no FtsA interaction was evident (rows c, d, e) whereas normal interactions were observed with FtsQ, FtsN, FtsI, FtsK and FTsB (rows c and d). From these results we can conclude that all the interactions of FtsQ both with itself and with the other cell division proteins are localized in the periplasmic domain of the protein although the MSS domain is necessary.
Overall, all these results agree with the conclusions of Dai et al. (1996) showing that the FtsQ periplasmic domain is necessary for the division process. In addition, constructs in which the cytoplasmic domain was intact but the FtsQ MSS was replaced were still functional. Even the construct in which the MalF MSS replaced the FtsQ MSS complemented the null mutation. Thus, the specific sequence of the MSS of FtsQ is not required for its function and may act only as a membrane-anchoring sequence (Guzman et al., 1997). Lastly, the need of the MSS domain for FtsQ interactions is in agreement with the fact that this protein is post-translationally translocated (Scotti et al., 1999) and is not able to fold correctly in the cytoplasm.
Identification of the FtsQ periplasmic subdomains involved in interactions with FtsB, FtsI, FtsL, FtsN, FtsK and FtsW
FtsQ interaction with itself and other bitopic proteins, FtsB,FtsI, FtsL, FtsN.
Progressive deletions of the FtsQ periplasmic domain indicate that the last 43 aa are essential for FtsL interaction. Indeed, the residual -galactosidase activity of 23 %, suggesting an interaction between FtsQ and FtsL, increased to 77 % when the last 43 residues of FtsQ were deleted (Fig. 3, rows c and d), in agreement with the results of Karimova et al. (2005). Analogously, the interaction of FtsB can be localized between residues 136 and 202 (Fig. 3, rows e and f).
These data are also consistent with the observation of Chen et al. (2002) that an ochre mutation interfering with translation of the last 29 aa is lethal and with the hypothesis proposed by Buddelmeijer et al. (1998) that the C-terminus of FtsQ might define a part of the protein involved in the interactions with FtsL and FtsB. This same region shows up as surface exposed in structure predictions for FtsQ, consistent with its possible role in protein–protein interactions.
Surprisingly, the interaction of FtsQ with itself and with FtsN and FtsI seems to be restricted to 8 aa. As shown in Fig. 3, progressive deletions of the ftsQ sequence from the C-terminus to residue 57 still maintain the interaction ability, which is simultaneously lost for all three proteins with the fragment containing residues 1–49. The interaction between the whole FtsQ and the FtsQ fragment 1–57 was also confirmed by co-immunoprecipitation experiments using the GFP-FtsQ and the cI-FtsQ1–57 fusion proteins as described above. The results reported in Fig. 2(C) are in agreement with those obtained with the THA. To elucidate whether other sites of FtsQ were involved in the interaction of the protein with itself, FtsN and FtsI, we constructed a mutated FtsQ protein (FtsQ8D) in which these 8 aa were deleted and replaced by 2 aa derived by religation of the XbaI site, as described in Methods. This deletion mutant retained the interaction ability of the wild-type FtsQ protein (Fig. 3, row m), whereas the FtsQ fragment 1–72 containing the 8D mutation was unable to interact with the FtsQ partners (Fig. 3, row o). Also in this case, the production of the truncated protein was verified by Western blot experiments with anti-FtsQ antibodies (data not shown). Taken together, these two results indicate that the 8 aa are effectively involved since their deletion affects the interaction ability, but also suggest that at least two sites are involved in interaction with FtsQ, FtsN and FtsI. The first site is located in the N-terminal part of the periplasmic domain between residues 49 and 57, and the second is located between residues 202 and 234 (Fig. 4, rows e and d), although Karimova et al. (2005) observed that a deletion of the last 30 aa of FtsQ (FtsQ1–246) abolished the capacity of the protein to dimerize as well to associate with all tested proteins. The fact that FtsB but not FtsA is able to interact with both the constructs indicates that in both cases the truncated proteins are produced and the FtsN cytoplasmic and transmembrane fragments present do not interfere with the results obtained.
In order to analyse the physiological role of the FtsQ region between residues 49 and 57, we transformed the plasmid carrying the FtsQ8D mutant under plac control into strain JOE170/pQ and studied its ability to complement an ftsQ null mutant in conditions of depletion of the wild-type FtsQ protein carried by the pQ plasmid. The e.o.p. of strain JOE170/pQ, which is 1 in the presence of arabinose (inducer of the expression of the wild-type ftsQ gene), was about 1.2x10–6 without arabinose (Table 3). When this strain, harbouring the plasmid carrying the FtsQ8D mutant, was plated in the presence of IPTG but without arabinose, the e.o.p. was 3.3x10–2 (mean of three independent experiments). This reduction of about 100-fold suggests that the region between residues 49 and 57 participates in FtsQ functionality but is not essential. This result was also confirmed by the study of the kinetics of growth of strain JOE170/pQ, harbouring the plasmid carrying the FtsQ8D mutant, in the absence of arabinose (Fig. 5). The generation time of this strain is in the order of 60 min, i.e. about double that observed upon expression of the wild-type ftsQ gene.
|
|
, rows c, d, e).
Fig. 3 shows that the FtsQ fragment 1–136 is able to interact with FtsW. This fragment includes the region called POTRA that is a conserved domain in the FtsQ family and a class of -barrel outer-membrane proteins (Sanchez-Pulido et al., 2003) identified in the -domain of DivIB/FtsQ (Robson & King, 2006). The fact that a deletion of the POTRA region results in loss of interaction (fragment 1–105) suggests that this domain could have a role in FtsW interaction.
Since the FtsQ periplasmic domain is essential for FtsW interaction, one could imagine that, on the other hand, the FtsW periplasmic spans are involved too. For this reason, we constructed two FtsW deletions, determined on the basis of the predicted topology model of the E. coli protein (Gerard et al., 2002; Pastoret et al., 2004), containing two spans (residues 1–181, containing spans 67–75 and 130–140) and only one span (residues 1–75, containing the span 67–75), respectively. The ability of these fragments to interact with FtsQ was then studied. As shown in Fig. 4(B), the smallest FtsW fragment able to interact with FtsQ is that containing just the first span.
Is it possible to divide the FtsQ periplasmic domain into functional subdomains?
It is possible to assume that the FtsQ domain architecture is similar to that of the orthologous protein DivIB of the Gram-positive thermophile Geobacillus stearothermophilus, whose 3D structure was recently resolved (Robson & King 2006). The extracytoplasmic region of DivIB comprises three discrete domains: (residues 47–116), (residues 117–230) and 
(residues 231–261) from the membrane-proximal N-terminus to the C-terminus. The - and -domains are structurally autonomous whereas the -domain is proteolytically sensitive and presumably unstructured in the absence of other divisomal proteins.
In this section the term ‘domains' will be used to identify the FtsQ periplasmic regions involved in protein–protein interactions, which will thereafter be compared with that deriving from the 3D structure of DivIB/FtsQ.
According to the above results, various sites of the FtsQ periplasmic domain are involved in its interactions with other division proteins. In addition, from the point of view of protein–protein interaction, this domain can be subdivided into three parts: (i) proximal to the transmembrane segment, residues 49–136, specific for FtsQ, FtsN, FtsI and FtsW; (ii) central, residues 136–202, where the FtsB interaction is localized; and (iii) C-terminal, residues 202–277, for FtsK, FtsL and for FtsQ, FtsN and FtsI (Fig. 6).
|
): the first fragment (called Q1) extends from residues 1 to 136, involved in FtsQ, FtsI, FtsN and FtsW interactions, and the second (called Q2) extends from 136 to 277, in which are localized the FtsB interaction site and the second site for FtsQ, FtsI, FtsN and FtsL interactions. Upstream of the coding sequence of Q2, the FtsN cytoplasmic and MSS domains were inserted, as described in Methods, to allow the peptides to anchor to the cell membrane, avoiding the possibility of recombination with the analogous domains of FtsQ, which could rearrange the plasmid sequence. These two domains do not interact with the other division proteins (Fig. 4, row f). If complementation occurs, the biological activity of this construct is verified.
Complementation experiments were performed as described in Methods. Aliquots of bacterial cultures of strain JOE170/pQ harbouring pcIP22-ftsQ1Q2 grown in LB medium in the presence of arabinose to about 2x108 cells ml–1 were centrifuged, washed and resupended in medium without arabinose. After 1 h, Q1Q2 expression was induced by addition of 1x10–4 M IPTG, and, over time, aliquots were plated on arabinose-supplemented medium. As shown in Fig. 5(C), during the first 3 h, the strain harbouring pcIP22-ftsQ1Q2 grew with a generation time of about 40 min (i.e. similar to that of the parental strain in the presence of arabinose), whereas JOE170/pQ without the plasmid was unable to grow. The Q1 and Q2 fragments, cloned separately on the same vectors, showed a different behaviour: Q1 allowed transient bacterial growth, with a generation time of about 70 min for the first 2 h, whereas JOE170/pQ cells containing Q2 alone died with a kinetics comparable with that of the parental strain without plasmid
The above observations were confirmed by the results of the experiments in which the e.o.p. of the strains in the presence and in absence of arabinose in the medium was evaluated (Table 3). The value of 1.2x10–6 shown by JOE170/pQ increased to 2.7x10–3 in the presence of plasmid pcIP22-ftsQ1Q2. Lastly, to exclude the possibility that the complementation could be due to the presence of a residual number of wild-type FtsQ molecules able to form multimers with the truncated form of the protein, we evaluated the amount of native FtsQ present in depleted cells. Western blot analysis of bacterial extracts obtained in the same conditions as used for complementation experiments showed that FtsQ was absent after 1 h of depletion (Fig. 7).
|
Complementation of an ftsQ null mutant with the orthologue DivIB of S. pneumoniae
It has been shown that FtsQ forms an in vivo complex with FtsL and FtsB (Buddelmeijer & Beckwith, 2004). Since the S. pneumoniae DivIB (FtsQ) shows the same properties, forming a trimer with the S. pneumoniae orthologous proteins FtsL and DivIC (FtsB) (Noirclerc-Savoye et al., 2005), we asked whether DivIB could interact with the E. coli proteins involved in this trimer formation. In other words, whether the interacting domains were somehow conserved during evolution, supporting the hypothesis of a ‘minimal common divisome’ whose specific characteristics were thereafter differentiated by evolution.
Results of interaction experiments between DivIB and the E. coli FtsQ partners, and of interactions between FtsQ and S. pneumoniae FtsL and DivIC (FtsB), are reported in Table 4. As far as the trimer formation is concerned, only the interaction DivIB–FtsQ is conserved. In addition, DivIB was able to interact with the E. coli FtsI protein. These results were confirmed by the co-immunoprecipitation experiments reported in Fig. 2(B). Surprisingly, in spite of its inability to interact with most of the FtsQ partners, DivIB was able to sustain, at least partially, the growth of the E. coli ftsQ null mutant (Fig. 5): the wild-type divIB gene, cloned in a plasmid under plac promoter control, was inserted by transformation into strain JOE170/pQ and its ability to sustain the growth of the strain in conditions of FtsQ depletion was tested. The e.o.p. of strain JOE170/pQ with and without arabinose in the medium, which is of the order of 10–6, clearly increases in the presence of DivIB and in an IPTG concentration-dependent manner (data not shown), reaching about 10–1 with 1x10–3 M IPTG (Table 3). In the absence of arabinose, the colonies were smaller (diameter about 2–3 times less) than those grown with arabinose. In addition, this rescue was also observed in broth, where the filamenous phenotype of strain JOE170/pQ, observed after 3 h of growth without arabinose, was lost in the presence of the DivIB plasmid. Indeed, in this case the bacteria were heterogeneous in size, with cells of normal size and cells showing some filamentation (data not shown).
|
), the stability of FtsQ is the same in the absence and in the presence of DivIB. The observed complementation suggests that DivIB could localize at the E. coli Z-ring level without interacting with the other proteins that normally interact with FtsQ. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
; Weiss, 2004; Vicente et al., 2006). Even if all of these proteins are essential, the role and biochemical activities of many of them remain to be discovered. It is known that some of them (FtsB, FtsQ, FtsL, FtsN, FtsI) are bitopic and some (FtsK and FtsW) are polytopic membrane proteins.
Genetic data, such as suppression of thermosensitive mutants in cell division proteins by means of overexpression of another division protein (Dai et al., 1993, 1996; Draper et al., 1998), as well as two-hybrid screens (Di Lallo et al., 2003; Karimova et al., 2005) and biochemical techniques (Buddelmeijer & Beckwith, 2004) indicate that these proteins have a complex web of interaction. FtsQ is a well-documented example. By a bacterial THA, it was suggested that FtsQ interacts with FtsI, FtsL and FtsN (Di Lallo et al., 2003). This conclusion was recently confirmed by Karimova et al. (2005) using a different THA. In addition, in each case by THA, it was shown that FtsQ interacts with FtsK and FtsW (Di Lallo et al., 2003) and FtsB (this work). Interactions between FtsQ, FtsL and FtsB were also confirmed by biochemical methods (Buddelmeijer & Beckwith, 2004).
Two aims formed the basis of this work: (i) to determine the FtsQ domains involved in interactions with the other division proteins; and (ii) to investigate the physiological role of the interactions described between FtsQ and its cell division partners, i.e. are all the identified interactions essential for septum construction?
FtsQ domains involved in the interactions with other division proteins
The structure of DivIB, an FtsQ orthologue (Robson & King, 2006), predicts that the extracytoplasmic region of DivIB comprises three discrete domains: (residues 47–116), (residues 117–230) and (residues 231–261) from the membrane-proximal N-terminus to the C-terminus. The - and -domains are structurally autonomous whereas the -domain is proteolytically sensitive and presumably unstructured in the absence of other divisomal proteins.
The architecture of FtsQ is expected to be very similar based on sequence homology with DivIB. The study of the progressive deletions in the ftsQ gene and domain-swap constructs suggests that the FtsQ self-interaction as well as its interactions with the other cell division proteins are localized in the periplasmic domain of the protein, in which it is possible to identify three regions comparable to that described by the 3D structure of DivIB (Fig. 6). However, from data presented in Figs 3 and 4, six interaction subdomains rather than three can be identified. Domain I (residues 1–21) contains the FtsK interaction site; domain II (residues 21–57) contains one of the two sites involved in FtsQ, FtsI and FtsN interactions; domain III (residues 105–136) is responsible for the FtsW interaction. In domain IV (residues 136–202), the FtsB interaction is localized, while in domain V (residues 202–234) the FtsK interaction site is located together with that of FtsQ, FtsI and FtsN. Lastly, in domain VI (residues 234–277) we find the interaction site for FtsL.
This suggests that within the structured regions , and found in DivIB smaller subdomains are probably involved in specific interactions.
Interestingly, two protein fragments are implicated in FtsQ, FtsI and FtsN interactions for which each of these sites is sufficient, since the FtsQ mutants carrying the fragment spanning between residues 1–57 as well as that between 137 and 234 are still able to interact with FtsQ itself, FtsI and FtsN. The role of residues 49–57 was also proved by biochemical experiments showing co-immunoprecipitation of the FtsQ fragment 1–57 with both FtsI and FtsQ. The importance of this region is also deduced by the observation that the strain harbouring a recombinant plasmid with the FtsQ protein deleted for residues 49–57 showed a 100-fold reduction in its ability to complement a bacterial strain depleted for FtsQ compared to the strain with the plasmid encoding the wild-type protein.
The analysis of the distribution of FtsQ sites of interactions with the other division proteins raises the hypothesis of its having a modular structure. We asked whether the subdomains described above could have a physiological role by testing if cells expressing peptides corresponding to these subdomains could complement the FtsQ depletion. This complementation was studied with two fragments. The first, called Q1, contains residues 1–136 and the second, Q2, spans from 136 to 277.
Our results indicate that the presence of plasmid Q1Q2 transiently complements the FtsQ depletion in strain JOE170/pQ, suggesting that the two peptides are able to localize independently at the Z-ring level and interact with the other division partnerships. The different behaviour shown by the two fragments as far as the complementation ability is concerned could be explained by a different turnover rate of the two peptides.
Robson & King (2006) proposed that the -domain corresponds to the POTRA domain that was predicted on the basis of bioinformatic analyses to be present in DivIB/FtsQ and exists as an autonomously folded protein domain. It was suggested that the POTRA domain might function as a chaperone that specifically recognizes secretion or assembly-competent forms of polypeptides (Sanchez-Pulido et al., 2003). In addition, the -domain could serve as a chaperone for its cognate divisomal partner FtsL. DivIB forms a ternary complex with FtsL and DivIC (Goehring & Beckwith, 2005; Buddelmeijer & Beckwith, 2004; Noirclerc-Savoye et al., 2005). These latter two proteins are unstructured and intrinsically unstable in vitro in the absence of DivIB (Robson & King, 2006), and FtsL is rapidly degraded in a divIB null strain at the non-permissive temperature (Daniel & Errington, 2000).
From these elements, we can deduce that Q1 roughly corresponds to the -domain and Q2 to domains and . According to the model of Robson & King (2006), Q1, besides its direct interaction with the other proteins, could also act as a chaperone. This activity could protect FtsL and/or other proteins from degradation. The Q2 fragment alone, without this chaperone Q1 activity, is therefore unable to complement since it cannot prevent the protein degradation. Future work should address this hypothesis.
Heterologous complementation: a new approach for in vivo studies of protein–protein interaction
One of the main problems in studying protein–protein interactions, identified by both in vivo genetic and in vitro biochemical methods, is to understand their biological role. It is possible that when many proteins form a complex, only a few interactions are essential. The role of the others may be to stabilize the structure in which they take part. In addition, it is possible that some interactions are only due to the protein's proximity to the complex without either determining its formation or having a role in it.
An approach to this problem could be to determine if it is possible in vivo to substitute the single components of a multiprotein complex with orthologous proteins from other bacteria, while retaining cell viability. In this case, the orthologous protein should be able to interact with the other proteins of the complex. In fact, it could be hypothesized that if a protein is conserved, as orthologue, the interactions with the partner proteins should also be conserved in the course of evolution. The loss of some interactions would thus suggest their ‘secondary’ role.
We showed that the S. pneumoniae DivIB can functionally substitute, at least partially, for the orthologous FtsQ protein of E. coli, although it interacts only with FtsQ and FtsI and not with the other FtsQ partners. This result was unexpected for two reasons. The first is that, although interspecies interactions amongst division proteins, such as FtsA and FtsZ, have been observed (Ma et al., 1997), these occur when the species are quite closely related and this is not the case for E. coli and S. pneumoniae. The second is that, although both FtsQ and DivIB form a trimer with FtsB (DivIC in the case of S. pneumoniae) and FtsL, DivIB is unable to interact with the E. coli orthologue, in spite of its ability to suppress the FtsQ depletion.
The finding that an E. coli ftsQ null mutant can be complemented by S. pneumoniae DivIB suggests a possible new strategy for investigating the biological significance of the protein–protein interactions identified by the various methods. We can therefore argue that only the homodimeric and the FtsI interactions are essential. Interactions with FtsN, FtsK, FtsB, FtsL and FtsW could be only secondary and contribute to stabilizing the complex.
Another possible explanation could be that the local FtsQ competes with DivIB for the interactions. Only the FtsI–DivIB interaction could be strong enough to be seen. In contrast, in the complementation experiment, the local FtsQ is absent and DivIB may well interact with more partners. This hypothesis can be eliminated since it is well known that the copy number of native FtsQ coded by the bacterial chromosome is low whereas the number of FtsI and FtsN copies is high, due to the plasmid localization of the genes under investigation. In addition the same result is obtained when studying the interaction between FtsQ and the S. pneumoniae division proteins.
Lastly, there is also the possibility that the S. pneumoniae protein could be degraded in E. coli. Also this hypothesis can be ruled out since we have shown that the S. pneumoniae division proteins normally interact with DivIB in E. coli (V. D'Ulisse and others, unpublished).
A function for FtsQ
FtsQ is arguably the most enigmatic divisomal protein. Discovered 25 years ago, its role in bacterial cell division remains unclear. It has been proposed that FtsQ provides a structural link between early- and late-assembling divisomal components (Di Lallo et al., 2003). A function in regulation of divisome assembly was suggested by Piette et al. (2004) and deduced by the fact that this assembly is very sensitive to FtsQ overexpression in the presence of thermosensitive ftsZ, ftsA or ftsI mutations (Dai & Lutkenhaus, 1992). In addition, recent evidence suggests that the FtsQ orthologue DivIB in Bacillus subtilis might also play a role in linking chromosome segregation to asymmetric cell division during sporulation (Real et al., 2005).
The finding that DivIB/FtsQ belong to a class of -barrel outer-membrane proteins involved in transport or assembly of polypeptides (Sanchez-Pulido et al., 2003) provided new elements to hypothesize a new function for FtsQ. Robson & King (2006) suggested that the POTRA domain might function as a chaperone that specifically recognizes secretion or assembly-competent forms of these polypeptides. These facts allow one to hypothesize a role for FtsQ of control of correct divisome assembly. Our data agree with both the scaffold and the assembly control role of FtsQ. (i) FtsQ interacts with at least six division proteins, but many of these interactions may not be essential, as suggested by the heterologous interactions between S. pneumoniae DivIB and the E. coli FtsQ interacting proteins; (ii) the whole FtsQ protein does not seem necessary for its biological activity, since the protein divided into two fragments is also able to partially sustain the growth of an ftsQ null mutant; and (iii) such complementation is also observable with the orthologous DivIB protein, which shows only 16 % homology with FtsQ (Massidda et al., 1998).
These data could indicate that FtsQ behaves like the central element of a puzzle. In such a complex, the central element should also be able to work when broken in two parts, since each part could be kept bound to the other by its joining with the other pieces, i.e. the various proteins interacting with FtsQ. Moreover, the complementation experiments, which showed that Q1 alone works much like Q1Q2 whereas Q2 is only partially active, accord with the FtsQ chaperone role in the divisome assembly control.
In conclusion, from this paper the idea arises that the complementation studies of null mutants with orthologous proteins (as for the complementation of an E. coli ftsQ null mutant with the S. pneumoniae DivIB protein, described here) could represent a new strategy for investigating the biological significance of the interactions identified with current methods for studying protein–protein interactions. With this strategy it should be possible to distinguish between the essential (or ‘primary’) and the dispensable (or ‘secondary’) interactions.
|
ACKNOWLEDGEMENTS |
|---|
Edited by: W. Margolin
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bi, E. F. & Lutkenhaus, J. (1991). FtsZ ring structure associated with division in Escherichia coli. Nature 354, 161–164.[CrossRef][Medline]
Bowler, L. D. & Spratt, B. G. (1989). Membrane topology of penicillin-binding protein 3 of Escherichia coli. Mol Microbiol 3, 1277–1286.[CrossRef][Medline]
Buddelmeijer, N. & Beckwith, J. (2002). Assembly of cell division proteins at the E. coli cell center. Curr Opin Microbiol 5, 553–557.[CrossRef][Medline]
Buddelmeijer, N. & Beckwith, J. (2004). A complex of the Escherichia coli cell division proteins FtsL, FtsB and FtsQ forms independently of its localization to the septal region. Mol Microbiol 52, 1315–1327.[CrossRef][Medline]
Buddelmeijer, N., Aarsman, M. E., Kolk, A. H., Vicente, M. & Nanninga, N. (1998). Localization of cell division protein FtsQ by immunofluorescence microscopy in dividing and nondividing cells of Escherichia coli. J Bacteriol 180, 6107–6116.
Carson, M. J., Barondess, J. & Beckwith, J. (1991). The FtsQ protein of Escherichia coli: membrane topology, abundance, and cell division phenotypes due to overproduction and insertion mutations. J Bacteriol 173, 2187–2195.
Chen, J. C. & Beckwith, J. (2001). FtsQ, FtsL and FtsI require FtsK, but not FtsN, for co-localization with FtsZ during Escherichia coli cell division. Mol Microbiol 42, 395–413.[CrossRef][Medline]
Chen, J. C., Weiss, D. S., Ghigo, J. M. & Beckwith, J. (1999). Septal localization of FtsQ, an essential cell division protein in Escherichia coli. J Bacteriol 181, 521–530.
Chen, J. C., Minev, M. & Beckwith, J. (2002). Analysis of ftsQ mutant alleles in Escherichia coli: complementation, septal localization, and recruitment of downstream cell division proteins. J Bacteriol 184, 695–705.
) and absence (
) of arabinose. (B) Growth of JOE170/pQ (
) and its derivative carrying plasmid pcI22ftsQD8 (
) or pcI22Q2 (
) in the absence of arabinose. (D) Growth of JOE170/pQ (
