Structural Insights into the FtsQ/FtsB/FtsL Complex, a Key Component of the Divisome

Bacterial cell division is a fundamental process that results in the physical separation of a mother cell into two daughter cells and involves a set of proteins known as the divisome. Among them, the FtsQ/FtsB/FtsL complex was known as a scaffold protein complex, but its overall structure and exact function is not precisely known. In this study, we have determined the crystal structure of the periplasmic domain of FtsQ in complex with the C-terminal fragment of FtsB, and showed that the C-terminal region of FtsB is a key binding region of FtsQ via mutational analysis in vitro and in vivo. We also obtained the solution structure of the periplasmic FtsQ/FtsB/FtsL complex by small angle X-ray scattering (SAXS), which reveals its structural organization. Interestingly, the SAXS and analytical gel filtration data showed that the FtsQ/FtsB/FtsL complex forms a 2:2:2 heterohexameric assembly in solution with the “Y” shape. Based on the model, the N-terminal directions of FtsQ and the FtsB/FtsL complex should be opposite, suggesting that the Y-shaped FtsQ/FtsB/FtsL complex might fit well into the curved membrane for membrane anchoring.

bacterial two-hybrid analysis 18 . Crystal structure of the periplasmic domains of E. coli and Yersinia enterecolitica FtsQ was solved, revealing partial structural information about the FtsQBL complex 19 . Interestingly, the periplasmic subcomplex of FtsQBL without the N-terminal membrane regions of FtsQ, FtsB, and FtsL was reconstituted in vitro by replacing the membrane parts with oppositely charged soluble coiled coils, building a structural model of a 1:1:1 FtsQBL complex in solution [20][21][22] . A mass spectrometric analysis by photo cross-linking indicated that the C-terminal regions of FtsQ and FtsB are crucial interaction hotspots 23 . FtsB and FtsL form a subcomplex in the absence of FtsQ via interactions between their transmembrane helices and membrane-proximal coiled coil regions 24 ; they are also codependent regarding their stability and divisome localization 25 . The FtsQBL complex may have a structural role as a scaffold in the assembly of the divisome since FtsB and FtsL proteins are recruited to the divisome by FtsQ and the resulting FtsQBL complex then recruits late divisome components 25 .
Despite previous intensive studies on the FtsQBL complex, the atomic details of the interface between FtsQ and FtsB were not reported, and structural organization of periplasmic domains of the FtsQBL complex was not shown yet. In this study, we have determined the crystal structure of the periplasmic domain of FtsQ in complex with C-terminal fragment of FtsB, and also solved the solution structure of the periplasmic FtsQBL complex. Based on these structures, we have proposed a 2:2:2 model of the FtsQBL complex and provided insights into its organization in cell division.
Crystal structure of the FtsQ-FtsB complex. To gain further insights into structural organization of the FtsQB complex, we determined the 2.9-Å resolution structure of FtsQ 50-276 bound to FtsB 25-103 (Fig. 2, Supplementary Fig. S3, and Table 1). The electron density map of residues 25-60 was not observed for FtsB  , while nearly all sequences of FtsQ 50-276 were found to be ordered with the exception of N-and C-terminal flexible regions (residues 50-52 and 262-276, respectively; Fig. 2A and Supplementary Fig. S3). Consistently, the electron density map of FtsB N-terminus (residues 22-63) was not observed for FtsB in the crystal structure of FtsQB complex, which was reported recently by other group 26 . One possibility regarding the lack of electron density is that the membrane-proximal helix is disordered in the crystal of FtsQB complex; however, the crystal structure of the 30 juxta-membrane amino acids of FtsB showed that it forms a canonical coiled coil 27 , thus we excluded this possibility. Another possibility is that the membrane-proximal helix is cleaved automatically during purification and crystallization, such that the remaining construct (FtsQ 50-276 -FtsB 61-103 ) crystallizes easily. Given that previous molecular dynamics simulations of the FtsQBL complex showed that the central part of FtsB around the Leu60 residue presents with a helix-break 28 , we predicted that the central part of FtsB around Leu60 is prone to proteolysis. The overall structure of FtsQ 50-276 consists of two domains, α and β, and the β-domain is responsible for FtsB 25-103 binding ( Fig. 2A). The solvent-accessible surface area buried at the interface between FtsQ and FtsB was 1330 Å 2 , approximately 20% of the monomer surface area of FtsQ. Furthermore, approximately 50% of the nonhydrogen atoms in the interface between FtsQ and FtsB are polar, as indicated by proximal isovelocity surface area calculations 29 . The FtsB sequence (residues 61-95) folds into one α-helix (α3) and one β-strand (β1; Fig. 2A). The α-helix-turn-β-strand motif begins with the α-helix (residues 64-75), followed by a kinked loop (residues 76-82), a short β-strand (residues 83-85), and another terminal kinked loop (residues 86-95; Fig. 2C); the two kinked loops contain two proline residues (Pro80 and Pro89; Fig. 2C). Interestingly, FtsQ α5 helix adopts a kinked shape by incorporating Pro228 in the middle and this shape appears to be crucial for its extensive contacts with FtsB. When Tyr precedes Pro in the amino acid sequence, it increases the tendency of the Pro residue to take on a cis conformation 30 ; however, the Pro228 residue of FtsQ adopts a trans conformation (Fig. 2C).
Binding modes of FtsQ in complex with FtsB. Although previous mass spectrometric analysis by photo cross-linking indicated that the C-terminal regions of FtsQ and FtsB are crucial interaction hotspots, the detailed interactions proposed by the mass spectrometric analysis were completely different from those observed in our crystal structure (Fig. 2). Based on the mass spectrometric analysis, it was suggested that the C-terminus of FtsB around Met77 residue forms a β-sheet-like interaction with the C-terminal strand of FtsQ around Gly255 residue 23 . In contrast to these data, our crystal structure showed that the region around Met77 residue of FtsB did not form a secondary structure, and the direction of its side chain is opposite to the expected binding region of FtsQ (Fig. 2C). In our crystal structure, the C-terminal region of FtsB (residues 61-103) extensively interacts with the β-domain of FtsQ and mainly contributes to FtsQ binding. The kinked loop between the α-helix (residues 64-75) and β-strand (residues 83-85) wraps around the Tyr248 residue of FtsQ (Fig. 2B). To stabilize the kinked loop conformation, four FtsB residues (Glu68, Arg72, Arg79, and Glu82) form a cluster with extensive salt-bridge networks (Fig. 3A). FtsB Glu68 forms salt bridges with both Arg72 (2.8 Å) and Arg79 (2.8 Å), while Arg72 forms a salt bridge with Glu82 (2.8 Å and 3.0 Å, respectively; Fig. 3A); the Arg72 NH2 atom forms one additional interaction with the Glu68 OE2 atom (2.8 Å; Fig. 3A). The salt-bridge cluster of FtsB is crucial in hydrogen bond formation (2.7 Å) between the oxygen atom of FtsQ Tyr248 and the NH1 atom of FtsB Arg72 (Fig. 3A). Furthermore, FtsB Glu69 forms a salt bridge with FtsQ Arg196 (2.3 Å; Fig. 3A). Interestingly, the β-strand of FtsB makes a continuous β-sheet with that of FtsQ by anti-parallel stacking with β12 (Fig. 2C). FtsB Phe84 forms close contacts with FtsQ Tyr248, whereas Tyr85 resides in close proximity to the hydrophobic core of the FtsQ β-domain consisting of Leu226, Leu230, Val254, and Trp256 (Fig. 2B). FtsB Leu87 also forms extensive contacts with FtsQ Leu226 and the aliphatic portion of Arg222 (Fig. 2C).
To evaluate the contributions of interfacial residues of the FtsQB complex, the interfacial conserved residues of FtsQ (Tyr248) and FtsB (Glu68, Glu69, Arg72, and Glu82) were mutated and their interactions were examined by BLI. The binding affinities between FtsQ and FtsB were measured by immobilizing wild-type (WT) or mutants of GST-e5-FtsB in complex with k5-FtsL. As expected, mutation of the interfacial residues resulted in large reductions in affinity compared with WT (K D = 0.24 μM). The K D value of the Y248A mutant could not be measured due to weak binding, while E68A reduced affinity by approximately 10-fold (Fig. 3B). The charge reversal mutations E69R, R72E, and E82R of FtsB nearly abolished binding to FtsQ (Fig. 3B). This highlights the role of FtsQ Tyr248 and the FtsB α-helix-turn-β-strand motif as affinity determinants.

FtsQ and FtsB interface in vivo.
To verify the in vitro binding of the periplasmic regions of FtsB and FtsQ, we performed a bacterial two-hybrid assay based on a cyclic AMP signaling cascade 31 . As N-terminus of T18 or T25 fused Fts protein was toxic for E. coli and has a certain degree of instability problem, we constructed C-terminus-fused Fts proteins (T25-FtsQ, T18-FtsB) 18 . E. coli BTH101 reporter strain was co-transformed with a pair of indicated plasmids and the binding efficiency was determined by β-galactosidase assay (Fig. 4A). As shown in Fig. 4A, WT FtsB, and WT FtsQ interacted with each other. In similar with the BLI assay, FtsB (E68A) can bind to WT FtsQ, while the binding affinity was reduced approximately two-fold compared with WT FtsB.
is the intensity of reflection hkl, Σ hkl is the sum over all reflections, and Σ i is the sum over i measurements of reflection hkl. c R = Σ hkl ||F obs | − |F calc ||/Σ hkl |F obs |, where R free was calculated for a randomly chosen 5% of reflections, which were not used for structure refinement and R work was calculated for the remaining. The mutated proteins except for the FtsB (E68A) diminished their ability to interact with WT FtsQ. It is consistent with in vitro binding results, indicating that the interfacial residues of FtsQ (Y248) and FtsB (E69, R72, and E82) play an important role in the formation of the FtsQB complex. The interaction between FtsQ and FtsB is required for recruiting downstream Fts proteins. When FtsQ or FtsB is depleted or introduced with deleterious mutation, it leads to filamentation and eventually cells death. To evaluate the in vivo function of FtsQB interfacial residues, we conducted a viability assay of cells expressing WT or mutated proteins with peptide nucleic acids (PNAs). A previous study showed that peptide-conjugated PNAs (PPNAs) targeting essential genes had bactericidal effects on E. coli cells [32][33][34] . Ribosome binding site is known to be a key PNA target site for inhibiting translation 32 . As the tight binding of PNA to mRNA interferes with ribosome function, protein translation is prevented 33 . Based on these principles, we designed four antisense PPNAs that are complementary to the translation initiation sites of ftsB and ftsQ ( Supplementary Fig. S4). The growth of the E. coli MG1655 strain was measured to examine the inhibitory effects of antisense PPNA1, PPNA2, PPNA3, and PPNA4. ftsB-targeting PPNA1 and ftsQ-targeting PPNA3 strongly inhibited bacterial growth and filamentation of E. coli was observed ( Supplementary Figs S4 and S5). Thus, we utilized PPNA1 and PPNA3 for further experiments. We constructed inducible plasmids expressing WT FtsB, WT FtsQ, FtsB mutants (E68A, E69R, R72E, and E82R), and FtsQ mutant (Y248A). As the used PPNAs have no complementary sequences to the recombinant plasmids, they do not interfere with the protein expression in plasmids. Strains expressing WT or mutated proteins showed no significant differences in colony-forming units (CFU) at 3 h without PPNA. Cells expressing the WT proteins (FtsB or FtsQ) in the plasmid recovered bacterial growth when treated with PPNA, but the cell expressing the mutated protein did not recover growth except for the FtsB (E68A) (Fig. 4C). Also, filamentation was observed in bacteria expressing mutated proteins lacking the ability to recover growth when treated with PPNA ( Supplementary Fig. S5). To exclude a possibility that the protein stability could affect the function of the mutated protein, we also determined in vivo protein stability by blocking de novo protein synthesis. The plasmid encoding FtsB or FtsQ fused with His tag at the C-terminus was introduced into E. coli MG1655,

Oligomeric states of FtsQB and FtsQBL in solution.
Previously, contradictory results were reported for the oligomeric state of the FtsQ protein. Dimerization of FtsQ via its transmembrane region (or C-terminal 30 residues) was observed in a bacterial two-hybrid experiment 18,35 . In contrast, a direct experiment for determining the oligomeric state of FtsQ in solution was performed using analytical ultracentrifugation indicating that FtsQ is a monomer 19 . However, we cannot exclude the possibility that the transmembrane region of FtsQ affects the oligomerization state of FtsQ, since the analytical ultracentrifugation experiment was performed using the periplasmic region of FtsQ 19 . Even though we also used the periplasmic regions of FtsQ, FtsQB, and FtsQBL, we hypothesized that their oligomeric states in solution depends on the concentration and that further oligomerization may occur at high concentrations. To analyze the concentration-dependent oligomeric states of FtsQ, FtsQB, or FtsQBL in solution, analytical gel filtration was performed using a Superdex 200 (10/300 GL) column and the experimental Stokes radii (R H ) were compared with those calculated from structure models ( Supplementary  Fig. S7). Structure models of FtsQ, FtsQB, and FtsQBL for calculating R H values were generated based on our crystal structure of FtsQB and SAXS structure of FtsQBL as described below. The invisible segments of the N-and C-termini of FtsQ, FtsB, and FtsL were modeled as flexible loops (Supplementary Fig. S7  mainly exists as a monomer in solution (Fig. 5). In contrast, for the FtsQB complex, the apparent R H (4.33 nm) was closer to that of the 2:2 FtsQB heterotetramer (4.87 nm) at high concentrations (138 μM), whereas the apparent R H (3.12 nm) was closer to that of the 1:1 FtsQB heterodimer (3.49 nm) at low concentrations (8 μM ; Fig. 5).
To further support the 2:2 FtsQB heterotetramer model, the molecular weight of FtsQB at high concentrations (138 μM) was measured by size-exclusion chromatography with multi-angle light scattering (SEC-MALS). Indeed, the molecular mass of FtsQB was consistent with a 2:2 complex of FtsQ and FtsB (Fig. 1C). These results suggest that FtsB mediates dimerization of FtsQB since FtsQ exists as a monomer at high and low concentrations.
For the FtsQBL complex, heterohexameric and heterotrimeric forms were eluted separately at the gel filtration during purification and they were collected separately. Each form was stable as indicated by SEC-MALS (Fig. 1C), and further used for the analysis of analytical gel filtration. At high and low concentrations of the FtsQBL complex (6.8-20 μM and 101-200 μM, respectively), the apparent R H values of heterohexameric FtsQBL were closer to that of the 2:2:2 FtsQBL model (5.59 nm), while the R H values of heterotrimeric FtsQBL were closer to that of the 1:1:1 FtsQBL model (4.28 nm) (Fig. 5). This suggests that both heterohexameric and heterotrimeric forms of FtsQBL exist in a stable form regardless of its concentrations.
Structural organization of the FtsQB tetramer. We next examined structural organization of the FtsQB heterotetramer in solution. When we evaluated crystal packing of the FtsQB complex, a potential FtsQB heterotetramer was generated through crystallographic symmetry operations and exhibited an elongated structure with overall dimensions of 30 × 40 × 150 Å, such that the FtsQB heterodimer dimerized in an antiparallel head-to-head fashion via the N-terminal α-domains of FtsQ subunits ( Supplementary Fig. S7C). Although we speculated that FtsQB forms a larger oligomer via FtsB based on analytical gel filtration data, we could not rule out the possibility that the FtsQB heterotetramer in solution was formed by the same interface observed in the crystal contact. Thus, we examined if the crystallographic interface is involved in forming the FtsQB heterotetramer. At the dimer interface of FtsQ, strictly conserved Leu57 and Phe87 residues form a hydrophobic core at the interface ( Supplementary Fig. S2A); therefore, we hypothesized that the FtsQB L57R or F87A mutant exists as a heterodimer regardless of its concentration if Leu57 and Phe87 are crucial residues for the assembly of the FtsQB heterotetramer. However, both L57R and F87A mutants assembled into a 2:2 heterotetramer only at high concentrations ( Supplementary Fig. S8), suggesting that the observed FtsQB heterotetramer in solution was not formed via the N-terminal α-domains of FtsQ subunits. We next hypothesized that the coiled coil region of FtsB is responsible for the dimerization of FtsQB heterodimers. In this case, we predicted that a helix-breaking L46P mutant of FtsQB inhibits dimerization of FtsQB heterodimers. Indeed, some portions of FtsQB L46P mutant assemble into 1:1 heterodimers, even at high concentrations ( Supplementary Fig. S8). These results suggest that the coiled coil region of FtsB is crucial for the assembly of the FtsQB heterotetramer in solution. In the 2:2 FtsQB heterotetramer model, the N-terminal directions of both FtsB subunits are oriented in the same direction, and thus the transmembrane region of FtsB may also dimerize to facilitate homodimerization of full-length FtsB. This may explain why higher-order oligomerization of FtsQB was observed only at high concentrations. Consistent with our findings, a previous study reported that  Fig. S8). In agreement with this, a recent study using FRET analysis and computational methods also showed that the FtsBL complex forms a tetramer 36 .

SAXS structure of FtsQBL complex.
To determine the overall conformation of the FtsQBL complex in solution, SAXS data were collected. The 1:1:1 or 2:2:2 FtsQBL complex was used to determine the ab initio structure of the molecular envelope at low and high concentrations, respectively ( Fig. 6A and Supplementary Fig. S9). Interestingly, the overall envelope of the FtsQBL 2:2:2 complex showed an elongated "Y" shape ( Fig. 6A). Even though the envelope did not show sufficient details to position the individual domains of FtsQ, FtsB, and FtsL, we observed that the large envelope with a concave surface fit reasonably well with the shape of FtsQ (Fig. 6A).
To incorporate the two FtsQ molecules into the large envelope with a concave surface, the N-terminal directions of FtsQ and FtsBL should be opposite. This was unexpected as the previous models formed by other groups possessed N-terminal transmembrane helices oriented in the same direction 20 . The crystal structure of the FtsQ 50-276 / FtsB 25-103 complex was successfully incorporated into the molecular envelope ( Fig. 6A and Supplementary  Fig. S9), indicating that the overall conformation of the FtsQ 50-276 -FtsB 25-103 complex is quite rigid in solution. The additional envelope was assigned to the positions of the heterotetrameric coiled coil and FtsBL e5-and k5-coils (Fig. 6A). It was recently suggested that the FtsBL subcomplex is a 2:2 heterotetramer with an uninterrupted FtsL helix linking the transmembrane and periplasmic coiled coil 36 ; therefore, the Y-shaped FtsQBL may fit well into the curved membrane for membrane anchoring (Fig. 6B). Due to the low resolution of the SAXS envelope, it was difficult to clearly define the structural assembly of FtsQBL; therefore, further structural determination is required. Although we observed 2:2 FtsQB heterotetramer and 2:2:2 FtsQBL heterohexamer formation in vitro, whether these forms exist in vivo remains unclear and further studies are required to reveal the physiological stoichiometry of the divisome. However, it should be noted that FtsQBL was previously suggested to form a 2:2:2 complex at the divisome 37 . Although additional studies are needed to fully understand structural organization of FtsQBL in divisome assembly, our data provide a foundation for further investigation.

Conclusion
Divisome proteins are essential for bacterial viability and its assembly and disassembly of the divisome occurs for each round of cell division in a predominantly hierarchical order 3 . Among the divisome proteins the FtsQBL complex was suggested to be a scaffold protein in the assembly of the divisome 25 . The FtsQBL complex is essential for bacterial viability 4 ; thus, exploiting it would be useful in discovering antibiotics that curb bacterial growth.
Despite previous biochemical studies on the FtsQBL complex, the atomic details of crucial interaction hotspots between FtsQ and FtsB were not reported. In this study, we have determined the crystal structure of the periplasmic domain of FtsQ in complex with C-terminal fragment of FtsB, showing that the C-terminal region of FtsB is a key binding region of FtsQ. The identification of crucial residues in protein-protein interactions between FtsQ and FtsB would be useful not only to provide insights into mechanisms of recognition between them, but also to indicate the regions to be targeted with inhibitors. We also solved the solution structure of the periplasmic FtsQBL complex by SAXS analysis. It should be noted that the roles of transmembrane helices of the FtsQBL complex should be explored in future studies because the periplasmic subcomplex of FtsQBL without the N-terminal membrane regions of FtsQ, FtsB, and FtsL was reconstituted in vitro in this study. However, we could propose a 2:2:2 Y-shaped model of the FtsQBL complex based on our SAXS structure, providing insights into its organization on the curved membrane for membrane anchoring.

Methods
Protein expression and purification. Constructs comprising residues 50-276 of FtsQ, residues 25-103 of FtsB, and residues 64-121 of FtsL were cloned into the pET21a (Novagen, Madison, WI, USA), pGST2 38 , and pRSF-duet (Novagen) vectors, respectively. To prepare FtsB-FtsL complexes, the sequence encoding the e5-coil and k5-coil was inserted into the N-terminus of FtsB  and FtsL 64-121 , respectively 20,22 . The pET21a, pGST2, and pRSF-duet vectors contained a C-terminal six-histidine tag, an N-terminal GST tag, and an N-terminal six-histidine tag, respectively (Supplementary Table S1). FtsQ 50-276 was expressed in E. coli BL21 (DE3) cells induced with 0.5 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG). For cell lysis, cell pellets were resuspended in buffer A (20 mM Tris-HCl pH 7.9, 500 mM NaCl, and 5 mM imidazole) containing 1 mM phenylmethylsulfonyl fluoride. Cells were lysed with a microfluidizer (Microfluidics, Westwood, MA, USA) and then lysed cells were centrifuged at 4600 × g (Vision V506CA rotor) for 30 min at 277 K to pellet the cell debris; the supernatant was applied to a Ni-affinity column (GE Healthcare, Little Chalfont, UK) pre-equilibrated with buffer A. The eluate was further purified by gel filtration on a HiLoad 16/60 Superdex 75 column (GE Healthcare) that had been pre-equilibrated with buffer B (20 mM Tris-HCl pH 8.0 and 200 mM NaCl).   were resuspended together in buffer B and mixed in a 1:1 ratio. Cells were lysed with a microfluidizer and the lysed cells were centrifuged at 4600 × g for 30 min at 277 K to pellet the cell debris; the supernatant was then applied to a glutathione-Sepharose column (GE Healthcare) pre-equilibrated with buffer B. Proteins were eluted with buffer B containing 15 mM reduced glutathione. The eluates were desalted into buffer B, cleaved by TEV protease, and re-passed through a glutathione-Sepharose column pre-equilibrated with buffer B to remove the GST tag. The eluates were further purified by gel filtration on a HiLoad 16/60 Superdex 200 column (GE Healthcare) that had been pre-equilibrated with buffer B. For the FtsQBL complex, cell pellets of FtsQ 50-276 and the e5-FtsB 25-103 /k5-FtsL 64-121 complex were resuspended together in buffer A and mixed in a 1:1 ratio. The FtsQBL complex was purified using the same method of purifying the FtsQB complex, except the supernatant was applied to a Ni-affinity column pre-equilibrated with buffer A before using the glutathione-Sepharose column. Interestingly, two major peaks were eluted separately after size-exclusion chromatography, which corresponds to the 2:2:2 and 1:1:1 complexes of FtsQBL, respectively.
Structure determination and refinement. The structure of the FtsQ 50-276 /FtsB 25-103 complex was solved by the molecular replacement method using the monomer model (chain A) of FtsQ from E. coli (PDB ID: 2VH1) 19 . A cross-rotational search followed by a translational search was performed using the Phaser program 40 . Subsequent manual model building was carried out using the COOT program 40 and restrained refinement was performed using the REFMAC5 program 41 . Several rounds of model building, simulated annealing, positional refinement, and individual B-factor refinement were performed. Table 1 lists the refinement statistics. The crystallographic asymmetric unit of the FtsQ 50-276 /FtsB 25-103 complex contained one heterodimer. The refined model included 21 water molecules and 100% of the residues were in the most allowed region of the Ramachandran plot. No electron density was observed for residues 25-60 in FtsB.
Size exclusion chromatography with multi-angle light scattering. SEC-MALS experiments were performed using an FPLC system (GE Healthcare) connected to a Wyatt MiniDAWN TREOS MALS instrument and Wyatt Optilab rEX differential refractometer (Santa Barbara, CA, USA). A Superdex 200 10/300 GL (GE Healthcare) gel-filtration column pre-equilibrated with buffer B was normalized using ovalbumin protein.
Proteins were injected (3-10 mg/mL) at a flow rate of 0.4 mL/min. Data were analyzed using the Zimm model for fitting static light-scattering data and graphed using EASI graph with a UV peak in ASTRA V software (Wyatt). Plasmid construction. We employed an isothermal assembly with DNA fragments, including linearized plasmid and PCR products. Preparation of fragments for isothermal assembly was performed by PCR and target sequences were amplified to overlap the adjacent fragment (~30 base pairs). The prepared fragments were assembled by incubation at 50 °C for 1 h in reaction buffer (25% PEG8000, 500 mM Tris-HCl pH 7.5, 50 mM DTT, 1 mM of dNTPs, and 5 mM NAD) with several enzymes (T5 exonuclease, Phusion polymerase, and Tag DNA ligase).

Western blot analysis.
Cells were sampled at the indicated time points and harvested. Pellets were immediately frozen in liquid nitrogen and stored at −20 °C until further analysis. Next, 1× sample buffer was added to the frozen pellet and sonicated. Whole-cell fractions were separated by molecular weight using SDS-PAGE, followed by transfer to a polyvinylidene difluoride membrane. The membrane was blocked with 5% non-fat dry milk and probed with an anti-Histidine antibody ( Analysis of protein stability. E. coli strain encoding FtsB-His or FtsQ-His protein was grown aerobically at 37 °C for 3 h in LB medium supplemented with 0.1 mM IPTG. Subsequently, the culture was harvested and resuspended the cell pellet in fresh LB medium containing chloramphenicol (0.2 mg/mL) to block de novo protein synthesis. After adding chloramphenicol, the cells were subjected to western blot analysis at the indicated time points.
PNA knockdown experiment. PNAs used in this study were commercially synthesized by Panagene (Daejeon, Korea). All PNAs were covalently conjugated to the cell-penetrating peptide (KFF) 3 K (in which K is lysine and F is phenylalanine) at the carbon terminus with an O linker. To determine the bacterial susceptibility to PNA, bacteria harboring pUHE21-2-lacI q encoding WT or mutated proteins were grown aerobically at 37 °C for 3 h in LB medium supplemented with 0.1 mM IPTG. Subsequently, the cultures were diluted to 5 × 10 4 CFU/ mL and incubated with 12.5 μM anti-ftsB PPNA1 or anti-FtsQ PPNA3 at 37 °C. After incubation for 3 h, the CFUs were determined on LB agar plates.
Solution SAXS measurements. SAXS measurements were carried out using the 4C SAXS II beamline of Pohang Light Source II with 3 GeV power at the Pohang University of Science and Technology (POSTECH), Korea. Light source from the In-vacuum Undulator 20 (IVU20: 1.4 m length, 20 mm period) of the Pohang Light Source II storage ring was focused with a vertical focusing toroidal mirror coated with rhodium and monochromatized with a Si (111) double crystal monochromator, yielding an X-ray beam wavelength of 0.734 Å. The X-ray beam size at the sample stage was 0.1 (V) × 0.3 (H) mm 2 . A two-dimensional charge-coupled detector (Mar USA, Inc., Evanston, IL, USA) was employed. Sample-to-detector distances of 4.00 and 1.00 m were used for SAXS. The magnitude of the scattering vector, q = (4π/λ) sin θ, was 0.09 nm −1 < q < 6.00 nm −1 , where 2θ is the scattering angle and λ is the wavelength of the X-ray beam source. The scattering angle was calibrated with a polystyrene-b-polyethylene-b-polybutadiene-b-polystyrene block copolymer standard. We used a quartz capillary with an outside diameter of 1.5 mm and wall thickness of 0.01 mm as the solution sample cell. All scattering measurements were carried out at 277 K using an FP50-HL refrigerated circulator (JULABO, Seelbach, Germany). The SAXS data were collected in ten successive frames of 3 s each to monitor radiation damage. Measurements of FtsQBL protein solutions were carried out over a small concentration range of 1.0-3.0 mg/ mL. Each 2D SAXS pattern was radially averaged from the beam center and normalized to the transmitted X-ray beam intensity, which was monitored with a scintillation counter placed behind the sample. Scattering of specific buffer solutions was used as experimental background. Structural parameters obtained from the SAXS data are shown in Supplementary Table S3. The R g (radius of gyration) values were estimated from the scattering data by Guinier analysis 43 . The pair distance distribution p(r) function was obtained through the indirect Fourier transform method using the program GNOM 44 .
Construction of 3D structural models. To reconstruct molecular shapes, the ab initio shape determination program DAMMIF 45 was used. For each model reconstruction, ten independent models were selected and averaged aligned using the program DAMAVER 46 . SAXS curves were calculated from the atomic models using the program CRYSOL 47 . To compare overall shapes and dimensions, ribbon diagrams of the atomic crystal models were superimposed onto the reconstructed dummy atom models using the program SUPCOMB 48 . Atomic coordinates and structure factors for the FtsQ 50-276 /FtsB 25-103 complex have been deposited in the Protein Data Bank (PDB ID code 5Z2W).