Role of Leucine Zipper Motifs in Association of the Escherichia coli Cell Division Proteins FtsL and FtsB

Abstract

FtsL and FtsB are two inner-membrane proteins that are essential constituents of the cell division apparatus of Escherichia coli. In this study, we demonstrate that the leucine zipper-like (LZ) motifs, located in the periplasmic domain of FtsL and FtsB, are required for an optimal interaction between these two essential proteins.

TEXT

The FtsB, FtsL, and FtsQ proteins play a key role in the assembly of the cell division machinery in both Gram-negative and Gram-positive bacteria (1, 14, 30). They form a trimeric complex in the inner membrane that is centrally located in the sequential recruitment pathway of the divisomal proteins in Escherichia coli (5, 15). This complex connects the upstream cell division components, which are mainly cytosolic and assembled around the FtsZ ring, with the downstream partners that are mainly membrane-associated or periplasmic proteins (8, 9, 14, 30). FtsL and FtsB can form a subcomplex independently of FtsQ (6, 15, 18, 21) and are mutually dependent for their stabilization: FtsL requires FtsB to be stable whereas FtsB is partially degraded when FtsL is depleted (7, 18). Noticeably, a regulated proteolytic degradation of these proteins might be involved in the control of cell division (4, 31). It was previously shown that FtsL and FtsB interaction is mediated mainly by their transmembrane domains and the membrane-proximal portion of their periplasmic domains, while the C-terminal portion of their periplasmic domains is involved in their interaction with FtsQ (5, 16–18, 29). Both FtsL and FtsB proteins contain, in their periplasmic part, a leucine zipper (LZ) motif present in all orthologous proteins, although these proteins are usually poorly conserved (7, 13, 17). The LZ motif was originally suspected to mediate the FtsL-FtsB heterodimerization (5, 7, 28), although it was shown that in Bacillus subtilis, FtsL variants modified in the heptad repeat were functional (27). More recently, Gonzalez and Beckwith reported that in E. coli, a truncated form of FtsB, lacking about half of the LZ, was still able to associate with FtsL (18), thus questioning the precise contribution of this motif in the interaction between these molecules. Here we reexamined this issue by analyzing FtsL and FtsB variants modified in their LZ motifs. The functionality of these variants was assessed by their ability to complement the absence of wild-type copies of the corresponding proteins. Their association capacities were tested by bacterial two-hybrid (BACTH) and coimmunoprecipitation techniques.

Modification of the leucine zipper motifs of FtsB and FtsL affects their interaction.

We generated FtsB and FtsL variants, called FtsB_M and FtsL_M, in which the key leucines L46, L53, L60, and L67 of FtsB and L63, L70, L77, and L84 of FtsL were replaced with alanines (see the supplemental material). These modifications should destabilize the hydrophobic interface formed between the two LZ motifs without affecting the overall alpha-helical structure of the coiled coil, thus avoiding a dramatic alteration of the proteins that could lead to their instability and degradation (7, 18). The impact of these modifications was assessed by using the BACTH assay that has been previously used to analyze interactions between Fts proteins (2, 10, 11, 21, 23). In this assay, the proteins of interest are fused to two complementary fragments, T25 and T18, from the adenylate cyclase of Bordetella pertussis and expressed in an E. coli cya strain. Upon interaction between the two proteins of interest, the fused T25 and T18 fragments reconstitute a chimeric adenylate cyclase that produces cyclic AMP (cAMP), which in turn activates β-galactosidase (β-Gal) synthesis (22). Genes encoding the two modified variants, FtsL_M and FtsB_M, as well as the wild-type counterparts, were fused to the C terminus of either T25 or T18 and expressed from the BACTH vector pKT25 or pUT18C, respectively, under the control of isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible plac promoters (see the supplemental material) (24).

To probe the functionality of the hybrid proteins, in vivo complementation assays were carried out in an E. coli FtsL or FtsB depletion strain (NB988 and CR14, respectively [26]), in which the corresponding chromosomal gene is inactivated, while a complementing ftsL or ftsB allele is expressed from a pBAD plasmid (19) (see the supplemental material) by arabinose induction. Figure 1 shows that expression of T18-FtsL_M or T18-FtsB_M fusion restored the viability of the depleted strain NB988 or CR14—in the absence of the wild-type FtsL or FtsB—with the same efficiency as the wild-type T18-FtsL or T18-FtsB fusion, indicating that they were all functional. Further experiments using green fluorescent protein (GFP) fusions showed that both variants, FtsL_M and FtsB_M, could be recruited to the divisome (data not shown), suggesting that the Leu-to-Ala changes in the LZ motif of FtsL or FtsB had no drastic effect on the protein structures and functions, in agreement with an earlier report (27).

Fig. 1.

Functional assays of the Leu-to-Ala variants of FtsB and FtsL. The modified genes ftsL_M and ftsB_M were chemically synthesized by Geneart AG (Regensburg, Germany) and subcloned into pKT25 and pUT18C as described in the supplemental material. (A) Complementation assays of the T18 fusion proteins were performed in the FtsB- or FtsL-depleted strain (CR14 or NB988 [26], respectively) as described in the supplemental material, by spotting 5 μl of each overnight preculture diluted 10⁻¹ to 10⁻⁶ on NZY solid medium containing IPTG (100 μM) and either arabinose (ARA) to induce expression of the pBAD-regulated gene or glucose (GLU) to repress its expression. Plates were then incubated for 24 h at 37°C. (B and C) Western blotting assays were performed as previously described (26) on total extracts from cells grown in the presence of arabinose or glucose after 1 h of induction of the T18 fusion construct with 100 μM IPTG. The Western blots were probed with anti-FtsL or anti-FtsB antibodies (26) or with an anti-T18 monoclonal antibody (see the supplemental material). The strains used are the FtsB-depleted strain CR14 or the FtsL-depleted strain NB988 harboring pUT18C derivatives expressing the indicated hybrid proteins. Note that the T18 fusions are significantly less abundant in the cells grown in the presence of glucose, probably because the plac promoter that drives the transcription of the T18 hybrids is under catabolite repression (24).

The BATCH assay was then performed to characterize the interaction properties of the different hybrids. The various T18 and T25 fusions were coexpressed in the E. coli cya strain DHM1 (21), and the interaction efficiencies were quantified by measuring β-Gal activities. Associations of the different fusions with FtsQ were tested as a control (5, 18). As shown in Fig. 2, the mutations within the LZ motif of both FtsL and FtsB had no significant effect on their interactions with FtsQ, indicating that the overall structure of FtsL_M and FtsB_M C-terminal domains had not been altered by the Leu-to-Ala replacements. However, the hybrid protein T25-FtsB_M interacted less efficiently with T18-FtsL than did the wild-type T25-FtsB fusion, as suggested by the lower level of β-Gal activity. Similarly, the interaction efficiency of T25-FtsB with T18-FtsL_M was about 40% lower than that measured with T18-FtsL. Importantly, cells coexpressing T18-FtsL_M and T25-FtsB_M exhibited an even lower β-Gal activity, corresponding to about 25% of that measured in DHM1/T18-FtsL/T25-FtsB (Fig. 2). Thus, the Leu-to-Ala replacement within the LZ motifs of both FtsL and FtsB weakened their association, indicating that the heptad leucine repeats in both proteins contribute to the stabilization of the FtsL-FtsB heterodimer.

Fig. 2.

Interactions of the FtsB and FtsL LZ variants studied by BACTH assay and coimmunoprecipitation. (A and B) β-Galactosidase activities in the E. coli DHM1 strain coexpressing T18 and T25 fusion proteins were measured on liquid cultures as described in the supplemental material. Results are expressed as relative units with standard deviations (SD) indicated in parentheses. Western blotting assays were performed as described previously (23) on total cell extracts from the same cultures used for the β-galactosidase assays using anti-T18 or anti-T25 antibodies (see the supplemental material) (B). The asterisk indicates an unspecific band. (C) Coimmunoprecipitations (IP, lanes 3 to 8) were performed as described in the supplemental material with anti-Flag M2 affinity beads (Sigma) on solubilized membrane protein extracts from the E. coli FtsB-depleted strain CR14 (26) coexpressing the indicated GFP fusions (GFP-FtsL or GFP-FtsL_M, expressed from plasmid pNG162) and Flag fusion proteins (FtsB_flag3 or FtsB_Mflag3, expressed from plasmid pDSW204 integrated onto the chromosome at the λatt site) (see the supplemental material). The Western blot was probed with anti-GFP antibodies. Lane 1 is a total extract (TE) of strain CR14 coexpressing GFP-FtsL and FtsB_flag3. Lane 2 corresponds to a control IP performed with strain CR14 coexpressing GFP-FtsL and the Flag3 epitope only (i.e., not fused to FtsB). The asterisk indicates an unspecific band.

Western blot analysis of T18 (Fig. 2A) and T25 (Fig. 2B) fusion stability revealed that the protein amounts were in good correlation with the levels of measured β-Gal activities. Indeed, the hybrid proteins were unstable in the absence of an interacting partner, consistent with the fact that FtsL and FtsB costabilize each other (18), but were much more stable when coexpressed with FtsQ.

The two-hybrid interaction data were further confirmed by coimmunoprecipitation. For this, FtsL or FtsL_M fused to GFP (on the low-copy-number plasmid pNG162 [15]) was coexpressed in the FtsB depletion strain CR14 with FtsB or FtsB_M tagged with a Flag epitope (26). These variants (FtsB_flag3 or FtsB_Mflag3) were expressed from a pDSW204 plasmid integrated into the chromosome (3, 32). Immunoprecipitations were carried out on bacterial membrane extracts using anti-Flag M2 affinity beads, and immunoprecipitated complexes were then probed by Western blotting with anti-GFP antibodies. As shown in Fig. 2C, the GFP-FtsL fusion was efficiently immunoprecipitated by both the FtsB_flag3 and FtsB_Mflag3 polypeptides (Fig. 2C, lanes 5 and 7). In contrast, the GFP-FtsL_M fusion was immunoprecipitated with a lower efficiency by the wild-type FtsB_flag3 and, more importantly, GFP-FtsL_M was not immunoprecipitated by FtsB_Mflag3 (Fig. 2C, lanes 6 and 8). These results demonstrated that the Leu-to-Ala modifications in the LZ of both FtsL and FtsB significantly decreased their association, thus corroborating the conclusion of the two-hybrid assays.

Swapping of the leucine zipper domain of FtsL and FtsB with the analogous domain from Haemophilus influenzae FtsL and FtsB orthologs.

To further delineate the polypeptide regions that mediate the specificity of association between FtsL and FtsB, we applied a “domain swapping” approach (12, 20). For that, we selected the orthologous FtsL and FtsB proteins from H. influenzae, FtsLhinf and FtsBhinf. FtsLhinf shares about 30% amino acid sequence identity with E. coli FtsL, while FtsBhinf is about 40% identical with E. coli FtsB (Fig. 3 A). Ghigo and Beckwith (12) previously reported that FtsLhinf could not complement the E. coli FtsL null mutant and did not localize to the septum. However, replacement of the E. coli FtsL LZ domain with the corresponding sequence of FtsLhinf resulted in a hybrid protein, LLhinfL, that was able to complement the absence of FtsL, although less efficiently than the native FtsL protein (12).

Fig. 3.

BACTH analysis of FtsBhinf, FtsLhinf, and LZ swap variants. (A) The amino acid sequence alignment of FtsB and FtsBhinf is shown with single-letter abbreviations. The predicted coiled-coil domain of FtsB and FtsBhinf is framed in gray, whereas the leucine zipper motifs are boxed in gray. The proposed cytoplasmic (Cyto), transmembrane (TM), and periplasmic domains of each protein are underlined. An alignment of FtsL and FtsLhinf is available in reference 12. (B) Schematic view of the domains of FtsB, FtsBhinf, FtsL, FtsLhinf, and the obtained swap proteins BBhinfB and LLhinfL. Alignments of the E. coli and H. influenzae leucine zipper domains (LeuZip) swapped are shown with single-letter abbreviations. The seven residues of each periodic repeat are referred to by the letters a to g. The d position is occupied by a conserved leucine shown in bold. Conserved residues are indicated between the two sequences, and plus signs indicate conservative substitutions. (C) BACTH analysis of T18 and T25 fusion proteins in the E. coli DHM1 strain (21). The genes encoding FtsLhinf and FtsBhinf were amplified by PCR from H. influenzae chromosomal DNA, and the swap constructs LLhinfL and BBhinfB were constructed by overlapping PCR, cloned into the pKT25 and pUT18C vectors, and transformed into DHM1 (see the supplemental material). Relative β-galactosidase activities in the E. coli DHM1 strain coexpressing T18 and T25 fusion proteins were measured as described in the supplemental material. The levels of the β-galactosidase activity in the control cells expressing nonfused T25 and T18 fragments were below 1 relative unit.

The FtsLhinf and FtsBhinf coding regions were PCR amplified and cloned into the pKT25 and pUT18C vectors (see the supplemental material). We also analyzed the recombinant protein LLhinfL, in which the LZ region of E. coli FtsL was replaced with the corresponding sequence of FtsLhinf (12). Similarly, a BBhinfB variant was constructed by replacing the LZ region of FtsB with the corresponding region of H. influenzae FtsBhinf (Fig. 3B). The resulting T18 hybrid proteins were first characterized by functional complementation assays in E. coli depletion strains lacking either FtsL or FtsB. We found that T18-LLhinfL and, more surprisingly, T18-FtsLhinf could partially restore the growth of the E. coli FtsL-depleted strain NB988 (Fig. 1A), in variance from the previous results of Ghigo and Beckwith (12). This may be due to a higher expression level of the T18-Lhinf protein achieved with the high-copy-number vector pUT18C (pUC19 origin) while, in the earlier study, FtsLhinf was expressed from a low-copy-number plasmid (pSC101 origin) (12). Similarly, the T18-FtsBhinf fusion and T18-BBhinfB swap construct were also able to complement the lack of FtsB (Fig. 1A). Altogether, these data indicated that, when produced from a high-copy-number vector, the wild-type FtsB and FtsL from H. influenzae as well as the BBhinfB and LLhinfL variants were functional in E. coli.

Interactions between these hybrid proteins were then characterized by BACTH assays. As shown in Fig. 3C, FtsLhinf and FtsBhinf heterodimerized, although less efficiently than their E. coli orthologs (≈9 β-Gal units versus ≈22 units, respectively). Interaction of FtsLhinf with E. coli FtsB was reduced (≈6 units), while E. coli FtsL did not associate with FtsBhinf (<2 units), revealing an intraspecies selectivity of association between these two components of the divisome. Replacing the LZ motif of FtsL with the corresponding region of FtsLhinf restored the association of the hybrid protein T18-LLhinfL with T25-FtsBhinf to a level similar to that found for the FtsLhinf/FtsBhinf heterodimerization (9.5 β-Gal units versus 9 units, respectively). The converse replacement of the LZ of FtsB with the corresponding region of FtsBhinf (T25-BBhinfB) did not restore association with T18-FtsLhinf (2.3 units) but significantly enhanced the interaction with the LLhinfL (T18-LLhinfL) variant compared to the wild-type E. coli FtsL (T18-FtsL) (12.5 versus 6.7 units). This confirmed the intraspecies selectivity of association of the LZ motifs of FtsL and FtsB.

In control experiments, the different hybrids were assayed for interaction with E. coli FtsQ protein. FtsLhinf and LLhinfL both associated efficiently with FtsQ, while FtsQ interaction with FtsBhinf was reduced compared to E. coli FtsB or BBhinfB. As the C-terminal end of FtsBhinf diverges largely from that of E. coli FtsB (Fig. 3A), these results support the idea that the C-terminal extremity of FtsB is involved in FtsQ binding as recently proposed (18). Altogether, the present data suggest that the selective heterodimerization of the FtsL and FtsB proteins in each species is partly specified by their LZ motifs.

Concluding remarks.

Our present mutational analysis of the key leucine residues from the LZ motifs of FtsL and FtsB revealed the contribution of the hydrophobic interface between these motifs to the stable association of the FtsL and FtsB proteins. The study of chimeric derivatives, LLhinfL and BBhinfB, obtained by replacing the LZ motifs of E. coli FtsL and FtsB with their corresponding motifs from the H. influenzae counterparts, further confirmed that the LZ motifs are important determinants of the species-specific heterodimerization of these two cell division components. Our results therefore demonstrate that the LZ motifs of FtsL and FtsB are required for an optimal interaction between these two essential proteins. Yet, additional regions of these proteins, especially their transmembrane segments, are clearly contributing to the stabilization of the FtsL/FtsB complex (5, 18). Furthermore, the complexity and redundancy of interactions between the divisomal proteins may also directly influence in vivo the stability of individual components, as illustrated here by the fact that FtsQ could efficiently stabilize the FtsB_M or FtsL_M variant.