Divisome under Construction: Distinct Domains of the Small Membrane Protein FtsB Are Necessary for Interaction with Multiple Cell Division Proteins

Abstract

Cell division in bacteria requires the coordinated action of a set of proteins, the divisome, for proper constriction of the cell envelope. Multiple protein-protein interactions are required for assembly of a stable divisome. Within the Escherichia coli divisome is a conserved subcomplex of inner membrane proteins, the FtsB/FtsL/FtsQ complex, which is necessary for linking the upstream division proteins, which are predominantly cytoplasmic, with the downstream division proteins, which are predominantly periplasmic. FtsB and FtsL are small bitopic membrane proteins with predicted coiled-coil motifs, which themselves form a stable subcomplex that can recruit downstream division proteins independently of FtsQ; however, the details of how FtsB and FtsL interact together and with other proteins remain to be characterized. Despite the small size of FtsB, we identified separate interaction domains of FtsB that are required for interaction with FtsL and FtsQ. The N-terminal half of FtsB is necessary for interaction with FtsL and sufficient, when in complex with FtsL, for recruitment of downstream division proteins, while a portion of the FtsB C terminus is necessary for interaction with FtsQ. These properties of FtsB support the proposal that its main function is as part of a molecular scaffold to allow for proper formation of the divisome.

Multiprotein complexes play central roles in many important cellular functions, including protein synthesis, motility, and cell division. Specific sets of protein-protein interactions are essential for the formation and stability of these complexes. One such complex, the divisome of Escherichia coli, may have a particularly intricate array of interactions, as it contains protein components involved in the cell division process that are located in the cytoplasm, periplasm, and inner and outer membranes. Ordinarily, the complete assembly of this complex occurs only at midcell after replication and segregation of the chromosome.

The divisome of E. coli contains at least 10 proteins required for division; in the absence of any one of these proteins, cells continue to grow but fail to divide, forming filamentous cells that eventually lyse (21). In addition, there are at least 13 nonessential proteins that play a role in the division process and localize to midcell ( 3-5, 16, 19, 26, 39, 42). Studies on the divisome have revealed a dependency pathway (FtsZ→FtsA/ZipA→FtsK→FtsQ→FtsL/FtsB→FtsW→FtsI→FtsN) for the assembly of the 10 essential proteins (21, 46). In this pathway, the first protein that localizes to midcell is FtsZ, and each subsequently localized protein requires the prior localization of all upstream proteins.

However, recent studies indicate that formation of the divisome does not occur by a strictly linear set of protein-protein interactions, as might have been inferred from the localization dependency pathway. Instead, assembly of the divisomal complex may depend on interactions in which many of the proteins interact with several other proteins, and a series of individual protein-protein interactions is not sufficient to form a stable complex. Studies using bacterial two-hybrid systems have revealed such multiple interactions, although further work is necessary to determine which interactions are involved in divisome assembly (15, 31). Furthermore, we have generated conditions using the premature targeting method (see Results) in which many of the “late” cell division proteins (FtsK through FtsI, but not FtsN) are able to assemble at midcell in the absence of the upstream division proteins that are normally required for their localization (22, 23). Suppressor and overexpression studies also support a cooperative model of protein interactions in the divisome, as individual components of the divisome can be partially replaced by another division protein (2, 17, 18, 24, 25, 36, 39).

The interactions involved in divisome assembly are best characterized for FtsZ, FtsA, and ZipA, which form what is called the “early” complex, based on the timing of localization relative to the remaining cell division proteins (1, 28, 29, 32, 34, 35). However, less is known about the interactions among the late cell division proteins (FtsK, FtsQ, etc.) or the interactions between the early and late proteins. A complication to the study of interactions among the late cell division proteins is that they are all inner membrane proteins, hindering in vitro analysis.

Nevertheless, recent work in several different bacterial species identified a complex of three late division proteins. In E. coli, the three late division proteins, FtsQ, FtsB, and FtsL, coimmunoprecipitate, and two-hybrid studies have detected interactions between these proteins (7, 15, 31). In Streptococcus pneumoniae, a complex consisting of the soluble extracellular domains of DivIB (FtsQ homolog), FtsL, and DivIC (FtsB homolog) was reconstituted in vitro (33) only after forcing dimerization of FtsL and DivIC, suggesting that the missing transmembrane segments ordinarily contribute to the interaction (33). Finally, two-hybrid and three-hybrid assays revealed interactions between DivIB, FtsL, and DivIC of Bacillus subtilis (14, 37).

FtsB, FtsL, and FtsQ of E. coli and their known homologs are bitopic membrane proteins with short amino-terminal cytoplasmic domains and larger carboxyl-terminal periplasmic domains. Although all three proteins are essential for division, currently there are no defined functions for FtsB, FtsL, and FtsQ. Both FtsL and FtsB contain predicted coiled-coil motifs in their periplasmic domains, which could mediate an interaction between the two proteins (8, 20). FtsB and FtsL are codependent for their localization to midcell, and depletion of FtsB results in the disappearance of FtsL, indicating that FtsL stabilization requires FtsB (8). Similar studies of the FtsB, FtsL, and FtsQ homologs in B. subtilis revealed a set of protein stabilizations within this complex (see Results) (12-14). We have shown that the predicted leucine zipper-like motif and transmembrane domain of FtsL are important for interaction with FtsB but that the cytoplasmic domain of FtsL is dispensable for this interaction (7). Finally, FtsL and FtsB interact in the absence of FtsQ and can recruit the downstream proteins, FtsW and FtsI, to the divisomal complex (22).

In this paper, we use several approaches to determine which domains of FtsB are involved in interactions with FtsL and FtsQ, as well as with other division proteins. We identify a portion of the FtsB C terminus involved in interaction with FtsQ, but not with FtsL or the downstream division protein FtsI, and we find that the predicted coiled-coil motif and transmembrane domain of FtsB are necessary for interaction with FtsL. Together our results provide the first detailed analysis of the domains of FtsB involved in interactions with other division proteins, strengthening the picture of an intricate array of interactions in this multiprotein complex.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are listed in Table 1. All experiments were performed with NZ medium (9). The following antibiotics were added when appropriate at the indicated final concentrations: ampicillin, 25 μg/ml (chromosome); chloramphenicol, 30 μg/ml (high-copy plasmid) or 10 μg/ml (low-copy plasmid); kanamycin, 40 μg/ml; spectinomycin, 100 μg/ml (plasmid) or 50 μg/ml (chromosome); tetracycline, 15 μg/ml. Medium was supplemented with 0.2% l-arabinose or 0.2% d-glucose to induce or repress expression of genes regulated by the P_BAD promoter. Isopropyl-β-d-thiogalactoside (IPTG) was added at the appropriate concentrations as indicated.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant genetic marker(s) or feature(s)a	Source or reference
E. coli strains
MC4100	F−araD139 ΔlacU169 relA1 rpsL150 thi mot flb5301 deoC7 rbsR	Laboratory collection
JOE309	MC4100 araD+	9
JOE417	JOE309 ftsQE14::kan/pJC10 (pBAD33-ftsQ)	10
NB946	JOE309 ΔftsB::kan Δ(λattL-lom)::bla araC PBAD-ftsB	8
NB976	NB946 attφ80::pJC115(P207-gfp-ftsI)	8
NB977	NB946 attφ80::pJC116(P207-gfp-ftsL)	8
NB980	NB946 attφ80::pJC118(P207-gfp-ftsQ)	8
MDG201	NB946 attφ80::pJC117(P207-gfp-ftsN)	This study
MDG238	JOE309 leu::Tn10 ftsQE14::kan ΔftsB::kan Δ(λattL-lom)::bla lacIqP206-ftsB/pJC10 (pBAD33-ftsQ)	This study
HSC074	JOE309 ΔftsN::kan/pBAD33-ftsN	H. S. Chung
MDG277	JOE309 ΔftsL::kan Δ(λattL-lom)::bla araC PBAD-ftsL	This study
Plasmids
pNG162	pSC101 ori, Specr, lacIq, P204	22
pIP2	pSC101 ori, Cmr, lacIq, P204-gfp	I. Petrovska
pDSW206	pBR322 ori, Ampr, lacIq, P206	47
pET-24b(+)	T7 lac promoter, C-terminal His tag, Kmr	Novagen
pMDG2	pSC101 ori, Cmr, lacIq, P99a	This study
pMDG3	pMDG2-zapA-Fcyto-ftsB	This study
pMDG4	pMDG2-zapA-Fcyto-ftsB Y85stop	This study
pMDG6	pMDG2-zapA-Fcyto-ftsB A55stop	This study
pMDG7	pNG162-flag3-ftsB	This study
pMDG8	pNG162-ftsB-flag3	This study
pMDG9	pNG162-ftsB45stop	This study
pMDG10	pNG162-ftsB45-flag3	This study
pMDG11	pNG162-ftsB50stop	This study
pMDG12	pNG162-ftsB55stop	This study
pMDG13	pNG162-ftsB55-flag3	This study
pMDG14	pNG162-ftsB60stop	This study
pMDG15	pNG162-ftsB65stop	This study
pMDG16	pNG162-ftsB70stop	This study
pMDG17	pNG162-ftsB75stop	This study
pMDG18	pNG162-ftsB80stop	This study
pMDG19	pNG162-ftsB85stop	This study
pMDG20	pNG162-ftsB85-flag3	This study
pMDG21	pNG162-ftsB90stop	This study
pMDG22	pNG162-ftsB95stop	This study
pMDG23	pNG162-ffb	This study
pMDG24	pNG162-ffb-flag3	This study
pMDG25	pDSW206-ftsB	This study
pMDG26	pIP2-ftsB	This study
pMDG27	pIP2-ftsB85stop	This study
pMDG28	pET-24b(+)-ftsBperi-his6	This study

^aori, origin; Spec^r, spectinomycin resistance; Cm^r, chloramphenicol resistance; Amp^r, ampicillin resistance; Km^r, kanamycin resistance.

Standard laboratory techniques were used for DNA cloning and analysis, PCR, electroporation, transformation, and P1 transduction (40). Individual bacterial strains and plasmids were constructed as described below. Enzymes for DNA manipulation were from New England Biolabs.

Strain construction.

Strain MDG201 was constructed by P1 transducing the integrate at the attφ80 site in strain JOE653 [JOE309 attφ80::pJC117(P₂₀₇-gfp-ftsN)] (J. Chen, unpublished strain) into strain NB946 and selecting for growth on NZ plates with spectinomycin.

The ftsB and ftsQ double depletion strain, MDG238, was constructed as follows. Plasmid pMDG25 (pDSW206-ftsB) was integrated into the chromosome of strain JOE309 at the λatt site using λInCh (6), and growth was selected for on plates with low ampicillin (25 μg/ml). The resulting strain was transduced with the ΔftsB::kan allele from strain NB946 and selected for growth on plates with kanamycin (40 μg/ml) and IPTG (20 μM), generating strain MDG236 [JOE309 ΔftsB::kan Δ(λattL-lom)::bla lacI^q P₂₀₆-ftsB]. Plasmid pJC10 (pBAD33-ftsQ) was transformed into strain MDG236, and resulting transformants were transduced with a P1 lysate from strain IP27 [JOE417 leu::Tn10 Δ(λattL-lom)::bla lacI^q P₂₀₆-ftsQ] (I. Petrovska), selected for tetracycline resistance (leu::Tn10) and screened for arabinose-dependent growth to identify clones with the chromosomally encoded ftsQ gene knocked out (ftsQE14::kan).

Strain MDG277 was generated by first P1 transducing the P_BAD-regulated ftsL gene from strain NB811 [JOE309 ftsL::TnphoAL81ΔIS50R (Kan^r) Δ(λattL-lom)::bla P_BAD-ftsL] (7) into strain JOE309 and selecting for growth on low ampicillin (25 μg/ml). The resulting transductants were then transduced with the ΔftsL::kan allele of strain WM2240 (DY329 ΔftsL::kan/pWM1845) (18) and selected for growth on NZ plates with kanamycin.

Plasmid construction.

Oligonucleotide primers used for the construction of all plasmids are listed in Table S1 in the supplemental material. Plasmid pMDG2 was created by excising the NsiI-HindIII fragment of lacI^q P₂₀₄ from pNG162 and ligating the fragment into the same sites of pBAD44, replacing araC P_BAD. The pBAD44 plasmid is a low-copy plasmid derived from pSC101 that contains a chloramphenicol resistance gene.

Plasmid pMDG3 was constructed in the following manner. The EcoRI site at the 5′ end of zapA from pNG166 (22) was eliminated according to the QuikChange method (Stratagene) using oligonucleotides ZapA EcoRI E2E For and ZapA EcoRI E2E Rev. The resulting plasmid was digested with EcoRI, and the annealed product of MalF cyto NoEco For and MalF cyto Eco Rev, which encodes the first cytoplasmic domain of malF, was ligated between zapA and ftsB. The resulting plasmid was digested with NsiI and HindIII to excise the lacI^q P₂₀₄-zapA-malF_cyto-ftsB sequence, and the fragment was ligated into the same sites of pBAD44, replacing araC P_BAD.

Plasmids pMDG4, -5, and -6 were constructed by PCR amplification of ftsB truncations from pMDG3 using oligonucleotide YgbQ 5EcoRIgfp for all reactions and oligonucleotides 3′ XbaI FtsB Y85stop, 3′ XbaI FtsB P80stop, and 3′ XbaI FtsB A55stop, respectively. The resulting PCR products were digested with EcoRI and XbaI and then ligated into a purified digest (EcoRI and XbaI) of pMDG3, lacking wild-type ftsB.

Plasmid pMDG7 was constructed by PCR amplification of ftsB from pNB10 (8) using oligonucleotides 5′ NcoI FtsB full start and pBADrev. The resulting PCR product was digested with NcoI and XbaI and then ligated into pNG162 digested with the same enzymes.

Plasmid pMDG8 was constructed by PCR amplification of ftsB-flag₃ from pDSW204-ftsB-flag₃ (N. Buddelmeijer) using oligonucleotides pTrcFor and pTrcRev. The resulting PCR product was digested with EcoRI and HindIII and then ligated into pNG162 digested with the same enzymes.

Plasmids pMDG9, -11, -12, -14 to -19, -21, and -22 were cloned by PCR amplification of ftsB truncations from pMDG8 using oligonucleotide pTrcFor for all reactions and oligonucleotides 3′XbaI FtsB K45stop, 3′XbaI FtsB N50stop, 3′ XbaI FtsB A55stop, 3′ XbaI FtsB L60stop, 3′ XbaI FtsB E65stop, 3′ XbaI FtsB R70stop, 3′ XbaI FtsB L75stop, 3′ XbaI FtsB P80stop, 3′ XbaI FtsB Y85stop, 3′ XbaI FtsB D90stop, and 3′ XbaI FtsB A95stop, respectively. The resulting PCR products were digested with EcoRI and HindIII and then ligated into pNG162 digested with the same enzymes.

Plasmids pMDG10, -13, and -20 were constructed by PCR amplification of ftsB truncations from pMDG8 using oligonucleotide pTrcFor for all reactions and oligonucleotides 3′ XbaI FtsB K45flag, 3′ XbaI FtsB A55flag, and 3′ XbaI FtsB Y85flag, respectively. The resulting PCR products were digested with EcoRI and HindIII and then ligated into a purified digest (EcoRI and XbaI) of pMDG8, lacking wild-type ftsB. pMDG8 was isolated from a dam mutant strain to prevent methylation of the XbaI site.

Plasmid pMDG23 was constructed by PCR amplification of the first cytoplasmic and transmembrane domains of malF from pKT25-malF (G. Karimova) using oligonucleotides 5′ NcoI FF primer and 3′ FFB primer overlap and PCR amplification of the periplasmic domain of ftsB from pMDG7 using oligonucleotides 5′ FFB overlap and 3′ XbaI FtsB full. The resulting PCR products were annealed using overlapping sequence, and the full-length ffb construct was then PCR amplified using oligonucleotides 5′ NcoI FF primer and 3′ XbaI FtsB full. The resulting ffb PCR product was digested with NcoI and XbaI and then ligated into pNG162 digested with the same restriction enzymes.

Plasmid pMDG24 was constructed by PCR amplification of ffb from pMDG23 using oligonucleotides 5′ NcoI FF primer and ftsBXbaImyc. The resulting PCR product was digested with NcoI and XbaI and then ligated into a purified digest of pMDG20 with the same enzymes, in which ftsB85stop was removed.

Plasmid pMDG25 was constructed by PCR amplification of ftsB from pMDG7 using oligonucleotides 5′ NcoI FtsB full start and pTrcRev. The resulting PCR product was digested with NcoI and XbaI and then ligated into pDSW206 digested with the same enzymes.

Plasmid pMDG26 was constructed by PCR amplification of ftsB from pDSW207-ftsB (N. Buddelmeijer) using oligonucleotides YgbQ 5EcoRIgfp and pTrcRev. The resulting PCR product was digested with EcoRI and HindIII and then ligated into pIP2 digested with the same enzymes.

Plasmid pMDG27 was constructed by PCR amplification of ftsB85stop from pMDG26 using oligonucleotides YgbQ 5EcoRIgfp and 3′ XbaI FtsB Y85stop. The resulting PCR product was digested with EcoRI and XbaI and then ligated into pIP2 digested with the same enzymes.

Plasmid pMDG28 was constructed by PCR amplification of the periplasmic encoding domain of ftsB from pNB10 (8) using oligonucleotides Peri FtsB NdeI and YgbQ XhoI His 3′. The resulting PCR product was digested with NdeI and XhoI and then ligated into pET-24b(+) digested with the same enzymes.

Generation and affinity purification of polyclonal antibodies against FtsB.

Polyclonal antibodies were raised against the periplasmic domain of FtsB using a C-terminally His-tagged version (FtsB_peri-His₆) as follows. Plasmid pMDG28 was transformed into E. coli strain BL21(DE3) for protein production. Expression from pMDG28 was induced with 1 mM IPTG for 2 h. FtsB_peri-His₆ from these cells was soluble and present in the cytoplasmic fraction and affinity purified with a HiTrap chelating HP column (GE Healthcare) using an AKTAprime fast-performance liquid chromatography system. The final preparation of FtsB_peri-His₆ was >95% pure. Purified FtsB_peri-His₆ was sent to Covance Research Products for generation of anti-FtsB antibody in rabbits.

Nonspecific and/or cross-reactive antibodies present in the anti-FtsB_peri-His₆ serum were removed by adsorption to an acetone purification of FtsB-depleted cells. Depletion of FtsB in strain NB946 was carried out as described here but was performed in a larger volume of NZ medium (∼150 ml). An acetone powder of cells depleted of FtsB was prepared as described previously (30). The resulting acetone powder was added to a final concentration of 1% to anti-FtsB_peri-His₆ serum with 10% glycerol and incubated at 4°C with mixing for 3 h. After incubation, the acetone powder was removed by centrifugation, and the supernatant was collected to be used as anti-FtsB antibody.

Depletion strain experiments.

Strains were grown overnight at 37°C in 5 ml of NZ medium supplemented with arabinose and appropriate antibiotics, then diluted 1:100 into fresh medium containing arabinose and antibiotics, and grown to an optical density at 600 nm (OD₆₀₀) of ∼0.3. An aliquot of the culture was washed once with NZ medium lacking sugar, diluted 1:100 or 1:200 (strain HSC074) into 5 ml of prewarmed NZ medium (37°C) supplemented with arabinose or glucose and the appropriate antibiotics, and grown at 37°C. Under these conditions, the time required to deplete the cells of FtsL, FtsB, and FtsQ is approximately 3 h, while the time to deplete the cells of FtsN is approximately 4 h. In experiments to test the FtsB truncations or FFB construct, the same conditions were used as described above, but 20 μM IPTG was added at the beginning of the 3-h depletion to induce the FtsB constructs. Similarly, the experiment in Fig. 2 was performed as described above for the FtsL depletion strain but with IPTG added at the beginning of the 3-h depletion to induce the FLAG-tagged FtsB constructs.

FIG. 2.

The C terminus of FtsB is cleaved in the absence of FtsL. Western blot analysis was performed on trichloroacetic acid-precipitated samples grown in the presence of arabinose (+) to induce expression of the P_BAD-regulated ftsL or in the presence of glucose (−) to repress expression of ftsL. Cells were induced with 60 μM IPTG for lanes 6 and 7 or with 5 μM IPTG for lanes 4 and 5. Samples were separated on 15% acrylamide gels for the anti-FLAG (α-FLAG), anti-FtsB, and anti-FtsL blots and on a 10% acrylamide gel for the anti-FtsQ blot. An equivalent of 0.05 OD₆₀₀ unit was loaded per lane. Note that in the anti-FLAG blot, the faster-migrating bands in lanes 3 and 4 appear to be degradation products of FtsB-FLAG that result from overexpression of this construct and are not specific to depletion of FtsL from cells. The strain used was MDG277 (FtsL depletion strain) (lanes 1 to 6). The plasmids used were pNG162 (empty vector [−]) (lanes 1 and 2), pMDG8 (C-terminally FLAG-tagged FtsB [C-term]) (lanes 3 and 4), and pMDG7 (N-terminally FLAG-tagged FtsB [N-term]) (lanes 5 and 6).

For the ftsB and ftsQ double depletion experiment (strain MDG238), the following protocol was followed. The strain was grown overnight in NZ medium containing arabinose, 20 μM IPTG, and chloramphenicol at 37°C, diluted 1:100 into fresh NZ medium containing arabinose, 20 μM IPTG, and chloramphenicol, and grown at 37°C to an OD₆₀₀ of ∼0.3. An aliquot of the culture was washed once with NZ medium lacking sugar and diluted 1:100 into 5 ml of prewarmed NZ medium containing one of the following: (i) arabinose, 20 μM IPTG, and chloramphenicol; (ii) arabinose and chloramphenicol; (iii) glucose, 20 μM IPTG, and chloramphenicol; and (iv) glucose and chloramphenicol. All cultures were grown at 37°C for 150 min.

Membrane preparation and coimmunoprecipitation experiments.

Coimmunoprecipitation experiments were performed in depletion strains as described above but in 50-ml culture volumes. Membrane preparations for the coimmunoprecipitation experiments were performed as previously described (37). Protein concentrations for the n-dodecyl-β-d-maltopyranoside (DDM)-solubilized membrane fractions were determined by BCA assay (Pierce). Immunoprecipitation experiments were performed with 200 μg of membrane fraction in a volume of 400 μl incubated with 5 μl of anti-FtsL or anti-FtsQ antibodies or with 20 μl of anti-FLAG M2 affinity gel (Sigma-Aldrich) at 4°C with mixing for 1 h. IgGSorb (The Enzyme Center), which was washed three times with buffer I (10 mM Tris-HCl, 150 mM NaCl [pH 7.4]) with a final concentration of 0.1% DDM, was added to the anti-FtsL and anti-FtsQ immunoprecipitation samples and incubated at 4°C with mixing for 30 min. All samples were then washed three times with buffer I plus 0.1% DDM and resuspended in 60 μl of sample buffer (New England Biolabs).

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blot analysis.

Cell samples for Western blot analysis were prepared by either resuspension of pelleted cells in 1× sample buffer (red loading buffer; New England Biolabs) or by precipitation of proteins with trichloroacetic acid. Samples to be analyzed were separated by Tris-glycine sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred by semidry blot electrotransfer according to the manufacturer's protocol (Bio-Rad) onto Immobilon-P polyvinylidene fluoride membranes (Millipore). Anti-FtsL (lab collection; 1:5,000 dilution), anti-FtsQ (lab collection; 1:1,000 dilution), anti-FtsB (lab collection; 1:5,000 dilution), anti-FtsN (lab collection; 1:2,000 dilution), anti-green fluorescent protein (anti-GFP) (1:500 dilution), anti-ZapA (lab collection; 1:500), and anti-FLAG (M2) (Sigma; 1:2,000 dilution) were used for immunoblotting. Secondary anti-mouse and anti-rabbit antibodies conjugated to horseradish peroxidase were used with the ECL Plus detection reagents according to the manufacturer (Amersham) for immunodetection.

Premature targeting experiments.

Strains were grown overnight at 30°C in 5 ml of NZ medium supplemented with arabinose and appropriate antibiotics, then diluted 1:100 into fresh medium containing arabinose and antibiotics, and grown to an OD₆₀₀ of ∼0.3. A sample of the culture was washed once with NZ medium lacking sugar, diluted 1:100 into 5 ml of prewarmed NZ medium (30°C) supplemented with glucose and the appropriate antibiotics, and then grown at 30°C. Cells were depleted of the complementing FtsB for 210 min, at which time IPTG was added to the cultures at a final concentration of 20 μM for the final 30 min of growth.

Microscopy.

After cultures were grown for the appropriate time, samples were fixed (11) and mounted onto 1% agarose cushions as previously described (45). Cells were examined using an Axioskop II (Zeiss) microscope equipped with a 100× plan-Apochromat oil immersion lens and a 100-W mercury lamp. ProImages were captured using an Orca-100 charge-coupled-device camera (Hamamatsu Photonics) and then were analyzed with the software Openlab (Improvision). The final processing of images for presentation was done using Adobe Photoshop.

RESULTS

The carboxy terminus of FtsB is degraded in the absence of either FtsL or FtsQ.

Analysis of the FtsL/ FtsB/FtsQ complex in both E. coli and B. subtilis indicates that some of the components of the complex become unstable in the absence of other components (8, 12-14). For instance, depletion of E. coli FtsB from cells results in the disappearance of FtsL (Fig. 1, lane 4) (8). Here we ask whether FtsB shows a similar instability in the absence of either FtsL or FtsQ.

FIG. 1.

FtsB stability depends on the presence of FtsL and FtsQ. Western blot analysis was performed on whole-cell samples grown in the presence of arabinose (+) to induce expression of the P_BAD-regulated gene or in the presence of glucose (−) to repress expression. Samples were separated on 15% acrylamide gels for the anti-FtsL (α-FtsL) and anti-FtsB blots and on 10% acrylamide gels for the anti-FtsQ and anti-FtsN blots. An equivalent of 0.05 OD₆₀₀ units was loaded per lane. The two anti-FtsB blots in the top of the figure are the same blot, but the top blot is exposed for a longer period of time to show the presence of an FtsB degradation product. The strains used were JOE309 (wild type [wt]) (lanes 1 and 2), NB946 (FtsB) (lanes 3 and 4), MDG277 (FtsL) (lanes 5 and 6), JOE417 (FtsQ) (lanes 7 and 8), and HSC074 (FtsN) (lanes 9 and 10).

To study FtsB stability, we used depletion strains that are inactivated for a particular cell division gene but express a complementing version of that cell division protein from the arabinose-inducible P_BAD promoter (27). In the absence of arabinose and with the addition of glucose, the basal level of expression of these proteins from the P_BAD promoter is undetectable by Western blotting. In the following experiments, unless stated otherwise, the expression of proteins occurs from their native chromosomal loci, with the exception of the P_BAD-regulated genes.

Depletion of FtsL resulted in a significant decrease in the levels of full-length FtsB and the appearance of an apparent degradation product running at a position below that of full-length FtsB (Fig. 1, lane 6). These findings, in combination with previously published work, indicate that FtsB and FtsL are codependent for stabilization (8). There was no observable change in FtsQ levels under the same FtsL depletion conditions. However, as with depletion of FtsL, depletion of FtsQ resulted in a significant decrease of full-length FtsB levels and the appearance of a degradation product (Fig. 1, lane 8).

To determine whether the degradation of FtsB occurs from the N or C terminus of FtsB, we constructed versions of ftsB in which a FLAG epitope was fused to one terminus or to the other terminus and determined whether the degradation product of FtsB was detected with anti-FLAG antibody. The N- and C-terminally FLAG-tagged FtsB constructs, called FLAG-FtsB and FtsB-FLAG, respectively, both complemented a strain depleted of FtsB when expressed from a low-copy plasmid, pNG162, under IPTG control (data not shown) (22). These constructs were then expressed in FtsL and FtsQ depletion strains that contained a wild-type copy of ftsB at the normal chromosomal locus.

When we expressed the C-terminally FLAG-tagged FtsB, which tags the periplasmic domain of FtsB, the degradation product of FtsB was not detected with anti-FLAG antibody upon depletion of FtsL or FtsQ (Fig. 2, lane 5, and data not shown). Under these conditions, we observed reduced levels of FtsB-FLAG and an increased amount of the degradation product detected by anti-FtsB antibody, which results from the degradation of FtsB-FLAG and the chromosomally encoded FtsB. In contrast, when we expressed N-terminally FLAG-tagged FtsB, three degradation products of FtsB, which result from degradation of the periplasmic C-terminal domain, were detected with anti-FLAG antibody (Fig. 2, lane 6). These results indicate that the periplasmic C-terminal domain of FtsB is degraded upon depletion of FtsL or FtsQ.

The reduction in full-length FtsB levels is specific to depletion of FtsL or FtsQ, as we observed no change in FtsB levels in cells depleted of other cell division proteins, e.g., FtsK, FtsW, FtsI, or FtsN (data not shown and Fig. 1, lane 10). Conversely, depletion of one component of the FtsB/FtsL/FtsQ complex results in no change in the levels of FtsN or FtsI (Fig. 1 and data not shown). It should be noted that in experiments with FtsQ depletion, we observed variability in the level of FtsL, with either no or some change in the level of FtsL. Reduced FtsL levels do not appear to directly result from depletion of FtsQ but instead depend on the levels of FtsB (see results below).

FtsL interacts with residual FtsB and the breakdown FtsB product.

The finding that, in general, the levels of FtsL are not diminished in cells depleted of FtsQ, despite the reduction in full-length FtsB and appearance of breakdown products, could be explained if the breakdown products of FtsB are still capable of interacting with and stabilizing FtsL. To test for an interaction between FtsL and FtsB and its breakdown products, we performed coimmunoprecipitation experiments using an FtsQ depletion strain in which FtsB and FtsL are expressed from their native chromosomal loci. As shown previously, anti-FtsL antibodies coimmunoprecipitated FtsL and FtsB in cells expressing FtsQ (Fig. 3, lane 1) (7). In the absence of FtsQ, anti-FtsL antibodies coimmunoprecipitated two bands reacting with anti-FtsB antibodies, corresponding to full-length FtsB and a breakdown product of FtsB (Fig. 3, lane 2). This result indicates that FtsL and FtsB can interact in the absence of FtsQ. Moreover, a portion of the FtsB C terminus is dispensable for the interaction with FtsL.

FIG. 3.

FtsL interacts with FtsB and an FtsB degradation product in the absence of FtsQ. Western blot analysis was performed on coimmunoprecipitated samples from DDM-solubilized membrane fractions of strain JOE417 (FtsQ depletion strain) immunoprecipitated (IP) with anti-FtsL (α-FtsL). Samples were separated on a 20% acrylamide gel for the anti-FtsB blot and a 15% acrylamide gel for the anti-FtsL blot. Cells were grown in the presence of arabinose (+) to induce expression of the P_BAD-regulated ftsQ or in the presence of glucose (−) to repress expression of ftsQ.

A previous conclusion that FtsL and FtsB could not interact in the absence of FtsQ can be explained by the degradation of the C-terminal FLAG tag used for coimmunoprecipitation and immunodetection of FtsB in those experiments (7). Thus, our results explain the earlier findings and are consistent with those from the premature targeting system that showed FtsB and FtsL can interact in the absence of FtsQ (22).

The FtsB C terminus is necessary for interaction with FtsQ.

Our results indicate that a portion of the FtsB C terminus is dispensable for interaction with FtsL and that proteolysis of the C terminus of FtsB occurs in cells depleted of FtsQ or FtsL. This proteolysis may take place because FtsQ ordinarily binds to and protects the C terminus of FtsB in an FtsL-dependent manner, since FtsB localization to midcell depends on the presence of FtsL. To better define the amount of the FtsB C terminus necessary for interactions with FtsL and FtsQ, we constructed a series of C-terminal truncations of FtsB by introducing the FLAG epitope tag at regular intervals in the ftsB sequence and used the FLAG epitope for coimmunoprecipitation experiments and Western blot analysis. In addition, we introduced stop codons at the same positions in FtsB to determine whether the FLAG tag might alter the interactions and complementation ability of the FtsB truncations. We expressed the FtsB truncations from a low-copy plasmid, pNG162, in an FtsB depletion strain, NB946, to test complementation ability and FtsL stabilization and to perform coimmunoprecipitation experiments with FtsL and FtsQ.

The results from the analysis of the C-terminal truncations of FtsB revealed several properties of the FtsB C terminus (Table 2). The last 13 amino acids of FtsB, after Asp90 (D90), are dispensable for complementation of an FtsB depletion strain, and consistent with this finding, coimmunoprecipitation experiments showed that FtsB Ala95stop (A95) still forms a complex with FtsL and FtsQ. However, an FtsB protein truncated at Tyr85 (Y85) did not complement an FtsB depletion strain, despite being expressed at a level equal to those of wild-type FtsB and stabilizing FtsL. Coimmunoprecipitation experiments showed that FtsB Y85stop also interacts with FtsL, but not with FtsQ. As an alternative method for assaying an interaction between FtsQ and FtsB Y85stop, we fused FtsB Y85stop to GFP and assayed localization in cells depleted of complementing FtsB. When we expressed a GFP fusion to FtsB Y85stop in an FtsB depletion strain, the resulting filamentous cells had no observable fluorescent bands at potential division sites (data not shown), even though Western blot analysis indicated that sufficient levels of GFP-FtsB Y85stop were present to observe septal localization (data not shown). This set of results indicates that a portion of the FtsB C terminus is necessary for the interaction with FtsQ. Note that the FtsB Y85-FLAG construct, in contrast to the lack of complementation by FtsB Y85stop, weakly complements an FtsB depletion strain and results in a “wrinkled” colony morphology, which is an indication of a severe but not lethal cell division defect (data not shown) (24). Examination of cells from these wrinkled colonies revealed long filamentous cells (data not shown). The defect in division appears to be related to a reduced interaction with FtsQ, as FtsB Y85-FLAG does not coimmunoprecipitate with FtsQ as efficiently as FtsB-FLAG does, even though it does coimmunoprecipitate with FtsL (data not shown).

TABLE 2.

C-terminal truncations of FtsB

FtsB constructa	Complementationb	FtsL levelc	FtsL interactiond	FtsQ interactiond	FtsI recruitmente
Wild type	+	+	+	+	+
Vector	−	−	NA	NA	−
K45stop	−	−	−	−	ND
N50stop	−	+	ND	ND	ND
A55stop	−	+	+	−	+/−
L60stop	−	+	ND	ND	ND
E65stop	−	+	ND	ND	ND
R70stop	−	+	ND	ND	ND
L75stop	−	+	ND	ND	ND
P80stop	−	+	ND	ND	+
Y85stop	−	+	+	−	+
D90stop	+	+	ND	ND	ND
A95stop	+	+	+	+	ND

^aFtsB constructs were generated by placing a stop codon in the amino acid position listed. The full-length FtsB is 103 amino acids long.

^bFtsB constructs were tested for complementation by expression in an FtsB depletion strain (NB946). Both the FLAG-tagged and non-FLAG-tagged constructs were tested for complementation, and similar results were obtained, with the exception of the FtsB Y85 construct as indicated in Results. Complementation ability was tested by streaking bacteria on NZ plates with glucose and various levels of IPTG and allowing the bacteria to grow overnight at 37°C. Symbols: +, growth; −, no growth.

^cFtsL levels were determined by Western blot analysis in cells expressing the given FtsB construct and depleted of the complementing FtsB (NB946). Symbols: +, FtsL present; −, absence of detectable FtsL.

^dFtsB construct interaction with FtsL or FtsQ was determined by coimmunoprecipitation (co-IP) in cells depleted of the complementing FtsB (strain NB946). Only the FLAG-tagged versions of FtsB K45 and A55 were used for co-IP experiments, while both the FLAG-tagged and non-FLAG-tagged versions were used for co-IP experiments with FtsB Y85 and A95. Co-IPs were performed with anti-FtsL, anti-FtsQ, and anti-FLAG antibodies. The specificity of these antibodies for co-IP experiments has been previously analyzed (7). Symbols: +, interaction between the tested proteins; −, little to no interaction between the tested proteins. Abbreviations: NA, not applicable; ND, not determined.

^eThe ability of the FtsB construct to recruit FtsI was determined using the premature recruitment assay. The given FtsB construct was fused to ZapA and tested for the recruitment of GFP-FtsI in an FtsB depletion strain (Fig. 4). Symbols: +, recruitment of FtsI; +/−, partial recruitment of FtsI; −, no recruitment of FtsI. Abbreviation: ND, not determined.

We tested additional truncations of FtsB to determine what portion of the C terminus is needed for interacting with and stabilizing FtsL. We determined that FtsB C terminally truncated up to Asn50 (N50) contains sufficient sequence of FtsB to stabilize FtsL. However, when truncated to Lys45 (K45), FtsB no longer stabilized FtsL. The ability of FtsB C-terminal truncations to stabilize FtsL correlates with formation of an FtsB/FtsL complex, as assessed in coimmunoprecipitation experiments. For example, FtsB A55-FLAG coimmunoprecipitated with and stabilized FtsL, whereas FtsB K45-FLAG, which itself is stable, did not coimmunoprecipitate with or stabilize FtsL. Moreover, FtsB A55-FLAG and K45-FLAG, as expected from the FtsB Y85stop construct, did not coimmunoprecipitate with FtsQ. These results indicate that a major portion of the C terminus of FtsB, including most of the leucine zipper-like motif, is not necessary for interaction with FtsL.

While we have noted that there is a codependency between FtsB and FtsL for stability, we nonetheless observed sufficient levels of FtsB K45FLAG, despite the absence of FtsL (data not shown). The stability of FtsB K45FLAG could result from the deletion of sequences that are recognized by the protease(s) and normally leads to reduced levels of FtsB.

Dispensability of the FtsB C terminus for recruitment of FtsI.

FtsB and FtsL can recruit the downstream division proteins, FtsW and FtsI, in the absence of FtsQ (22). Since the noncomplementing C-terminal truncations of FtsB do not localize to midcell, we were unable to directly address whether the C-terminal end of FtsB includes domains involved in recruitment of downstream cell division proteins. To study this interaction, we used the premature targeting method to force the localization of truncated versions of FtsB to potential division sites and examined the recruitment of other cell division proteins fused to GFP. This is done by fusing the ZapA protein to the cytoplasmic N-terminal domain of the FtsB truncations. ZapA is a nonessential cytoplasmic protein that binds to FtsZ; proteins fused to ZapA localize to the FtsZ ring (22, 23, 26). We expressed the IPTG-regulated ZapA fusions from a low-copy plasmid, pMDG2, in an FtsB depletion strain. We performed all premature targeting experiments in cells depleted of FtsB to prevent wild-type FtsB from affecting the interpretation of the experiments. The GFP fusions to FtsQ, FtsL, FtsI, or FtsN, used to determine whether the ZapA-FtsB fusions could recruit other proteins, are IPTG regulated and expressed from the phage φ80 attachment site.

We expressed GFP-FtsQ as a positive control for septal localization, since the localization of FtsQ is not dependent on the presence of FtsB (Fig. 4). In the empty vector samples, in addition to all samples tested, GFP-FtsQ localized to potential division sites in the filamenting cells. Note that ZapA fused to wild-type FtsB complements the FtsB-depleted cells, resulting in cells the size of wild-type cells. Western blot analysis revealed that the majority of FtsB detected when expressing ZapA-FtsB is not the full-length ZapA-FtsB fusion but instead appears to be breakdown products of this fusion (data not shown). Thus, we are unable to determine which FtsB product complements the FtsB-depleted cells.

FIG. 4.

The FtsB C terminus is dispensable for recruitment of the downstream cell division protein FtsI. Premature targeting analysis and microscopy of FtsB variants were performed as described in Materials and Methods in cells depleted of the complementing FtsB. The FtsB variants were fused to ZapA to force localization, and the ability to recruit various cell division proteins was examined with GFP fusions to the given protein. Note that 40 μM IPTG was used to induce GFP-FtsI in the cells expressing ZapA-FtsB A55stop; all other cells were induced with 20 μM IPTG. Representative fluorescence micrographs are shown, with white arrowheads indicating the positions of some septal bands. GFP-FtsN fluorescent foci are present in the ZapA-FtsB Y85stop sample and represent active division sites, but no GFP-FtsN fluorescent bands are observed in the filament portion. The strains used were NB980 (GFP-FtsQ), NB977 (GFP-FtsL), NB976 (GFP-FtsI), and MDG201 (GFP-FtsN), and the plasmids used were pMDG2 (empty vector), pMDG3 (ZapA-FtsB wild type [wt]), pMDG4 (ZapA-FtsB Y85stop), and pMDG6 (ZapA-FtsB A55stop).

In the empty vector control samples, we did not observe septal localization for the GFP fusions to FtsL, FtsI, or FtsN in FtsB-depleted cells, as would be expected from the localization dependency pathway (Fig. 4). In contrast, the ZapA fusion to full-length FtsB recruited GFP fusions to FtsL, FtsI, and FtsN to potential division sites (for comments on FtsN recruitment, see below). The ZapA fusions to both FtsB Y85stop and FtsB P80 stop recruited GFP fusions to FtsL and FtsI to potential division sites but not GFP-FtsN (Fig. 4 and data not shown). These results suggest that the C-terminal portion of FtsB, which is important for interaction with FtsQ, is not necessary for the recruitment of GFP-FtsI or presumably FtsW.

An analysis of FtsB A55stop with the premature targeting system resulted in positive but less-intense localization signals for GFP fusions to FtsL and FtsI, while again, GFP-FtsN did not localize (Fig. 4). The lower-intensity localization signals may be due to the reduced levels of FtsL that are indicated by Western blot analysis (data not shown). The lower FtsL levels, encoded from the native chromosomal locus, are probably a result of the protocol for premature targeting experiments, which involves depletion of the complementing FtsB over several hours and induction of the ZapA and GFP fusions during the final 30 min of growth (22). The late induction of the ZapA-FtsB A55stop fusion could result in less stabilization of FtsL but did not result in altered FtsL levels for ZapA fusions to FtsB Y85stop or FtsB P80stop. These results could indicate that there is some difference in the ability of the FtsB truncations to interact with and stabilize FtsL. Nonetheless, the results with the ZapA-FtsB A55stop fusion show that there is still some capacity to recruit FtsI, suggesting that much of the periplasmic domain of FtsB is not necessary for recruiting FtsI and presumably FtsW. We did not attempt premature targeting experiments with FtsB K45stop, since it is unable to stabilize or interact with FtsL, and we have found that a portion of FtsL is required to recruit the downstream cell division proteins, FtsW and FtsI (M. D. Gonzalez and J. Beckwith, unpublished results).

Goehring et al. showed by premature targeting analysis that in the absence of functional FtsA or FtsQ, ZapA fusions to FtsQ, FtsL, FtsB, FtsW, or FtsI do not recruit GFP-FtsN fusions to division sites (22, 23). One interpretation of these results is that GFP-FtsN is unable to localize in the absence of FtsA or FtsQ, even though FtsI is present, because recruitment of FtsN requires interaction with several of the proteins present in the complex. However, in our premature targeting experiments with the FtsB truncations described here, performed in an FtsB depletion strain, all division proteins upstream of FtsN are present, but GFP-FtsN was not recruited (Fig. 4). These results could indicate either that the C-terminal domain of FtsB is directly or indirectly necessary for recruitment of FtsN or that ZapA-formed division complexes are not identical to the normal complexes and, therefore, are not competent for recruitment of FtsN. Attempts to differentiate between these interpretations were unsuccessful, so it remains unclear why GFP-FtsN is not recruited in the premature targeting experiments.

The N terminus of FtsB is important for interaction with FtsL and FtsQ.

To examine the role of the FtsB N-terminal domain, we constructed a version of FtsB (FFB) in which the first cytoplasmic and transmembrane domains of the unrelated membrane protein MalF replace those of FtsB. We expressed the IPTG-regulated FFB from a low-copy vector, pNG162, in an FtsB depletion strain.

Western blot analysis indicated that the levels of FFB were comparable to those seen for FtsB in wild-type cells. However, when expressed in cells depleted of FtsB, FFB did not complement for growth and did not stabilize FtsL, as determined by Western blot analysis (Table 3). FFB also did not coimmunoprecipitate with FtsL or FtsQ. These results indicate that FFB does not interact strongly with either FtsL or FtsQ. As an alternative method for assaying the ability of FFB to interact with FtsQ, we analyzed the localization of a GFP fusion to FFB in cells depleted of FtsB but observed no septal localization (data not shown). Western blot analysis of GFP-FFB revealed that sufficient levels of the fusion were expressed to observe septal bands if they had been present (data not shown). These results suggest that the cytoplasmic domain, transmembrane domain, or both, of FtsB are necessary for formation of a stable FtsQ/FtsB/FtsL complex.

TABLE 3.

Analysis of the N-terminal domain of FtsB

FtsB construct	Complementationa	FtsL levelb	FtsL interactionc	FtsQ interactionc
Vector	−	−	NA	NA
Wild type	+	+	+	+
FFB	−	−	−	−

^aFFB was tested for complementation by expression in an FtsB depletion strain (NB946). Both the FLAG-tagged and non-FLAG-tagged versions of FFB were tested for complementation, and similar results were obtained. Complementation ability was tested by streaking bacteria on NZ plates with glucose and various levels of IPTG and allowing the bacteria to grow overnight at 37°C. Symbols: +, growth; −, no growth.

^bFtsL levels were determined by Western blot analysis. Symbols: +, presence of FtsL; −, absence of detectable FtsL.

^cFFB interaction with FtsL or FtsQ was determined by coimmunoprecipitation (co-IP). Only the FLAG-tagged version of FFB was tested for co-IP with FtsL and FtsQ, by co-IP with anti-FLAG, anti-FtsL and anti-FtsQ antibodies. Symbols: +, interaction between the tested proteins; −, no interaction between the tested proteins. NA, not applicable.

FtsQ does not have a negative effect on FtsL levels in the absence of FtsB.

In B. subtilis, it appears that the instability of the FtsB homolog DivIC, observed in the absence of the FtsL homolog, depends on the presence of the FtsQ homolog, DivIB, suggesting that DivIB has a negative effect on DivIC stability (14). Given the B. subtilis results, we could explain the instability of FtsL in the absence of FtsB by a negative effect of FtsQ on FtsL. Additionally, while we showed that FtsL can coimmunoprecipitate with full-length FtsB and a breakdown product of FtsB, it does not show that this interaction is necessary for stabilization of FtsL. To ask whether FtsL stability in an FtsQ depletion strain depends on the presence of FtsB and its breakdown products, we created a strain with ftsB and ftsQ deleted from their native chromosomal loci and complemented with an IPTG-inducible version of ftsB and an arabinose-inducible version of ftsQ (MDG238). By having the complementing versions of ftsB and ftsQ under different regulation systems within the double deletion strain, we could reduce expression of one gene while maintaining expression of the other gene or reduce the expression of both at once.

Using the double depletion strain, we observed that FtsL levels were reduced to a similar level in the absence of FtsB or in the absence of both FtsB and FtsQ (Fig. 5, lanes 2 and 4). These results indicate that wild-type levels of FtsL are dependent only on the presence of FtsB and do not appear to be negatively affected by the presence of FtsQ. The reduction in FtsL levels that occurred when depleting both FtsB and FtsQ was specific and not a general effect on cell division proteins, as we observed no change in the levels of FtsN.

FIG. 5.

FtsL stability depends on the presence of FtsB and is not negatively regulated by FtsQ. Western blot analysis was performed on whole-cell samples separated on 15% acrylamide gels for the anti-FtsL (α-FtsL) and anti-FtsB blots and on 10% acrylamide gels for the anti-FtsQ and anti-FtsN blots. Expression of ftsB was induced with 20 μM IPTG (+) (lanes 1 and 3) or repressed with glucose (−) (lanes 2 and 4), while expression of ftsQ was induced with arabinose (+) (lanes 1 and 2) or repressed with glucose (−) (lanes 3 and 4). Strains JOE309 (wild type [wt]) (lane 5) and MDG238 (double depletion strain) (lanes 1 to 4) were used.

While low levels of FtsB were observed upon depletion of FtsB or FtsQ, a reduction in FtsL levels is observed only with depletion of FtsB. We suspect that this difference in FtsL levels results from the presence of FtsB C-terminal degradation products in the FtsQ-depleted cells, which can still interact with and stabilize FtsL. We did not readily observe any FtsB degradation products with anti-FtsB antibody in these FtsQ-depleted cells, but this could result from the lower levels of FtsB in these cells relative to wild-type cells (Fig. 5, lane 1 versus lane 5). Nonetheless, we still expect there to be FtsB degradation products under these conditions.

DISCUSSION

In E. coli, the three cell division proteins FtsB, FtsL, and FtsQ form a complex that is essential for the recruitment of other proteins to midcell. Bioinformatic analysis suggests that homologs of these three proteins are found in many bacteria, including B. subtilis and S. pneumoniae, where evidence for similar complexes has been presented (14, 33, 43). In spite of evidence for this trimeric complex, premature targeting experiments show that E. coli FtsL and FtsB can function as a subcomplex in the absence of FtsQ, capable of recruiting the downstream division proteins FtsW and FtsI (22). This subcomplex appears to be conserved, as a strong interaction between the FtsB and FtsL homologs of B. subtilis can be observed when these two proteins are expressed in E. coli in the absence of other B. subtilis division proteins (37). In this study, we focused on FtsB, one protein of the FtsB/FtsL/FtsQ complex.

Specifically, we defined the domains of FtsB that are involved in the set of interactions within the complex and in recruitment of other proteins to the divisome. We draw the following conclusions about the locations of these domains. (i) We show that a carboxy-terminal domain of FtsB is essential for its interaction with FtsQ. When the region covering the last 18 amino acids (amino acids 85 to 103) is deleted from FtsB, the protein has greatly reduced affinity for FtsQ but retains the wild-type affinity for FtsL. In contrast, when only 13 amino acids are deleted from the carboxy terminus, FtsB behaves like the wild-type protein does. (ii) The membrane-proximal portion of the FtsB periplasmic domain is required for interaction with FtsL. When the region from amino acids 45 to 103 is truncated, affinity for and stabilization of FtsL are no longer detectable, while an FtsB containing up to amino acid 50 is still able to stabilize FtsL. (iii) An FtsB containing up to amino acid 55 is able to recruit FtsI and presumably FtsW. (iv) Finally, our results show that the amino terminus of FtsB, most likely the transmembrane segment, is essential for the interaction with FtsL.

Since FtsB forms a complex with FtsL, we cannot distinguish whether the domains of FtsB we have defined for interaction with FtsQ and recruitment of downstream proteins reflect direct or indirect interactions. These interactions may take place only on the structure formed by the assembly of FtsL and FtsB into a complex. Nevertheless, given the small size of FtsB (103 amino acids), it appears that much, if not all, of the protein is contributing to interactions with other proteins (Fig. 6). On this basis, we consider it unlikely that the FtsB/FtsL complex has a function other than being part of a scaffold for the recruitment of other proteins to the divisome. FtsQ, which is part of this subcomplex in the divisome, may have some additional activity in cell division, as its periplasmic domain is much larger than that of either FtsB or FtsL.

FIG. 6.

A model for the interaction domains of FtsB. Within the topological model of FtsB are several features that are discussed in this paper. The amino acids of the cytoplasmic domain (residues 1 to 3), transmembrane domain (residues 4 to 22), and periplasmic domain (residues 23 to 103) are the same as previously described (8). The positions of the truncated residues are indicated by their respective number, as is the predicted position of the leucine zipper-like domain (residues 46 to 67) (leucines in gray) and coiled-coil domain (residues 29 to 67). The positions of the FtsL interaction domain (residues 4 to 50), the FtsB/FtsL interaction domain with downstream division proteins (residues 4 to 55), and the FtsQ interaction domain (residues 85 to 90) of FtsB are also shown. The FtsQ interaction domain of FtsB probably includes flanking amino acids in addition to those indicated in the figure, as mentioned in the Discussion.

On the basis of our results, we present a model for the interaction domains of FtsB, shown in Fig. 6. FtsB interacts with and stabilizes FtsL via its transmembrane and predicted coiled-coil domains. This complex, which does not require most of the FtsB periplasmic domain, is sufficient for recruitment of the downstream division protein FtsW. Finally, the interaction of the FtsB/FtsL complex with FtsQ requires a portion of the FtsB C terminus. The crystal structure of E. coli FtsQ was recently described, and it was noted that several mutations in FtsQ that disrupt recruitment of the FtsB/FtsL complex map to a surface-exposed region near the C terminus of FtsQ (44). Thus, assembly of the FtsB/FtsL/FtsQ complex may be mediated by their C-terminal domains.

It has been proposed that the predicted coiled-coil domains of FtsB and FtsL mediate their interaction (8). The predicted coiled-coil domain of FtsB encompasses a significant portion of the periplasmic domain, spanning residues 29 to 67 and containing a leucine zipper-like motif (residues 46 to 67). Yet, the largest truncations of the FtsB periplasmic domain that still interact with and stabilize FtsL are missing the predicted leucine-zipper-like motif. An example is the FtsB N55stop construct which is missing two of the four leucines. This truncated protein is still predicted to form coiled-coils according to the COILS prediction program (Gonzalez and Beckwith, unpublished). However, when truncated to K45, the predicted coiled-coil motif of FtsB is eliminated. This loss of potential for coiled-coil structure is correlated with the loss of ability of FtsB truncated at K45 to interact with or stabilize FtsL.

For FtsL, previous results have suggested that its leucine zipper-like motif is important for interaction of FtsL with FtsB. When the leucine zipper-like motif of FtsL was swapped with the corresponding segment from the FtsL homolog of Haemophilus influenzae, interaction with FtsB was drastically reduced (7). Given our results that the predicted coiled-coil domains play an important role in the interaction between E. coli FtsB and FtsL and since predicted coiled-coil domains appear to be conserved in FtsL and FtsB homologs, it is reasonable to conclude that these regions are generally involved in interaction of the homologs.

Our results also suggest that the formation of the FtsB/FtsL complex requires the transmembrane segment of FtsB. When the short cytoplasmic and transmembrane domains of FtsB were replaced with those of the unrelated protein MalF, the interaction with and stabilization of FtsL were eliminated. In complementary fashion, replacement of the FtsL transmembrane domain with a MalF domain abrogated coimmunoprecipitation of FtsB (7). Furthermore, the cytoplasmic domain of FtsL is dispensable for interaction with FtsB. Given the small predicted size of the FtsB cytoplasmic domain, it seems likely that it is the transmembrane domains and not the cytoplasmic domains of FtsB and FtsL that are necessary for interaction (7). This interpretation is consistent with other results that suggest a role for the transmembrane segments of these proteins in other organisms (S. pneumoniae, B. subtilis, and Bacillus stearothermophilus) (33, 38, 43).

While the interaction of FtsB with FtsL required the N-terminal portion of the FtsB periplasmic domain, the interaction of FtsB with FtsQ requires a portion of the FtsB C terminus, separate from the predicted coiled-coil motif required for interaction with FtsL. We found that FtsB truncated at D90 still complements an FtsB depletion strain and presumably still interacts with FtsQ; however, truncation of FtsB at Y85 eliminated both properties. This finding indicates that there are amino acids downstream of Y85 required for FtsQ interaction but does not eliminate the possibility that the actual interaction region also extends upstream of Y85. The latter suggestion is strengthened by the finding that replacement of residues 85 to 88 of FtsB with nonconservative amino acids, while maintaining the flanking sequence, resulted in a minor defect with a mixed population of cells of various lengths (Gonzalez and Beckwith, unpublished). On the basis of our results, we suggest that while the region of FtsB fromY85 to D90 is required for interaction with FtsQ, the flanking sequence to this region is also involved in the interaction with FtsQ.

Although the FtsB truncations identified a C-terminal portion required for interaction with FtsQ, this interaction could be direct or indirect. We found, however, that even though the C terminus of FtsB is necessary for interaction with FtsQ, this domain by itself in the FFB swap construct is not sufficient for interaction with FtsQ. One interpretation is that the N-terminal domain of FtsB is necessary for interaction with FtsQ. This explanation of this result seems unlikely, given that the cytoplasmic and transmembrane domains of FtsQ, while recently shown to play a role in FtsQ localization, are not necessary for complementation of an FtsQ depletion strain and presumably are not needed for interaction with FtsL/FtsB (11, 24, 41). The more likely explanation, therefore, is that a complex of FtsB/FtsL is required for interaction with FtsQ, and since FFB does not stabilize or interact with FtsL, FFB is unable to interact with FtsQ. Given that FtsB and FtsL can interact in the absence of FtsQ, it is reasonable to speculate that the FtsB/FtsL complex forms first and that this complex can then interact with FtsQ.