The bacterial cell division protein fragment EFtsN binds to and activates the major peptidoglycan synthase PBP1b

Peptidoglycan (PG) is an essential constituent of the bacterial cell wall. During cell division, the machinery responsible for PG synthesis localizes mid-cell, at the septum, under the control of a multiprotein complex called the divisome. In Escherichia coli, septal PG synthesis and cell constriction rely on the accumulation of FtsN at the division site. Interestingly, a short sequence of FtsN (Leu75–Gln93, known as EFtsN) was shown to be essential and sufficient for its functioning in vivo, but what exactly this sequence is doing remained unknown. Here, we show that EFtsN binds specifically to the major PG synthase PBP1b and is sufficient to stimulate its biosynthetic glycosyltransferase (GTase) activity. We also report the crystal structure of PBP1b in complex with EFtsN, which demonstrates that EFtsN binds at the junction between the GTase and UB2H domains of PBP1b. Interestingly, mutations to two residues (R141A/R397A) within the EFtsN-binding pocket reduced the activation of PBP1b by FtsN but not by the lipoprotein LpoB. This mutant was unable to rescue the ΔponB-ponAts strain, which lacks PBP1b and has a thermosensitive PBP1a, at nonpermissive temperature and induced a mild cell-chaining phenotype and cell lysis. Altogether, the results show that EFtsN interacts with PBP1b and that this interaction plays a role in the activation of its GTase activity by FtsN, which may contribute to the overall septal PG synthesis and regulation during cell division.

Peptidoglycan (PG) is an essential constituent of the bacterial cell wall. During cell division, the machinery responsible for PG synthesis localizes mid-cell, at the septum, under the control of a multiprotein complex called the divisome. In Escherichia coli, septal PG synthesis and cell constriction rely on the accumulation of FtsN at the division site. Interestingly, a short sequence of FtsN (Leu 75 -Gln 93 , known as E FtsN) was shown to be essential and sufficient for its functioning in vivo, but what exactly this sequence is doing remained unknown. Here, we show that E FtsN binds specifically to the major PG synthase PBP1b and is sufficient to stimulate its biosynthetic glycosyltransferase (GTase) activity. We also report the crystal structure of PBP1b in complex with E FtsN, which demonstrates that E FtsN binds at the junction between the GTase and UB2H domains of PBP1b. Interestingly, mutations to two residues (R141A/R397A) within the E FtsN-binding pocket reduced the activation of PBP1b by FtsN but not by the lipoprotein LpoB. This mutant was unable to rescue the DponB-ponA ts strain, which lacks PBP1b and has a thermosensitive PBP1a, at nonpermissive temperature and induced a mild cell-chaining phenotype and cell lysis. Altogether, the results show that E FtsN interacts with PBP1b and that this interaction plays a role in the activation of its GTase activity by FtsN, which may contribute to the overall septal PG synthesis and regulation during cell division.
Peptidoglycan (PG) is an essential constituent of the bacterial cell wall and a major antibacterial target; it surrounds the cytoplasmic membrane, determines the cell shape, and protects the cell from rupture under internal osmotic pressure. The PG structure consists of glycan strands made of alternating b-1,4-linked GlcNAc and N-acetylmuramic acid residues cross-linked by peptides (1). It is assembled using the lipid II (undecaprenyl-pyrophosphoryl-N-acetylmuramic acid-(pentapeptide)-GlcNAc), a cell wall precursor, by the GTase activities of the class A penicillin-binding proteins (aPBPs) and SEDS (shape, elongation, division, and sporulation) proteins and cross-linked by the transpeptidase (TPase) activities of class A and class B PBPs (bPBPs) (2)(3)(4). The GTase and TPase activities are coupled within the same aPBPs and also with their bPBPs TPase partners (e.g. PBP1a-PBP2, PBP1b-PBP3) (5). Similarly, the GTase activities of the SEDS proteins are regulated by a cognate bPBP (RodA-PBP2 and FtsW-PBP3) (3,(6)(7)(8). The activities of the PBPs were shown to be regulated within multiprotein complexes, allowing a concerted PG synthesis and sacculus enlargement in line with cell cycle progression (9)(10)(11)(12).
Bacteria generally have at least one bifunctional (GTase/ TPase) PBP of class A and one monofunctional (TPase only) PBP of class B (2,13). E. coli contains three aPBPs: PBP1a, 1b, and 1c, and two bPBPs: PBP2 and PBP3, the latter being involved in cell elongation and division, respectively (14). PBP1a and PBP1b are major PG synthases and at least one of them is required for cell viability (15). Although the two PBPs are exchangeable, they likely play specific functions during the cell cycle. PBP1a is mainly involved in cell elongation in partnership with PBP2, whereas PBP1b exhibits a preference for cell division in agreement with its enrichment at midcell during cell constriction (12,(16)(17)(18). PBP1b has a modular structure, composed of an N-terminal tail and a transmembrane anchor followed by the GTase and TPase catalytic domains separated by the regulatory UB2H domain. PBP1b was shown in vitro to form a ternary complex with FtsW and PBP3 (8), which constitutes the septal synthase subcomplex of the divisome. This complex interacts with the divisome regulatory proteins FtsBLQ, FtsN and the outer-membrane lipoprotein LpoB (9)(10)(11). FtsN is the last essential protein that localizes at the division site; its accumulation through a self-enhanced positive feedback mechanism triggers cell constriction (19). LpoB binds to the UB2H domain of PBP1b and stimulates both of its catalytic activities (20). FtsN and LpoB bind simultaneously to PBP1b and synergistically stimulate the GTase activity of PBP1b (12). The GTase activity of PBP1b is repressed by FtsBLQ (via FtsL) during divisome assembly and the presence of FtsN and/or LpoB suppresses this inhibition (9). Moreover, FtsBLQ (via FtsQ) was also shown to inhibit the TPase activity of PBP3 but not that of PBP1b (9). On the other hand, CpoB interacts with TolA, and both proteins bind to the PBP1b-LpoB complex, between the UB2H and TPase domains, thus inhibiting the TPase activity of PBP1b but not its GTase activity (12). This regulatory system blocks septal PG (sPG) synthesis catalyzed by PBP1b until the maturation of the divisome is signaled by the accumulation of FtsN, which stimulates PBP1b and counterbalances the inhibitory effect of FtsBLQ, thus contributing to sPG synthesis and cell constriction.
From a structural point of view, FtsN is a bitopic membrane protein composed of a small cytoplasmic domain, a transmembrane a-helix and a large periplasmic domain (21,22). The cytoplasmic domain of FtsN interacts with the 1C subdomain of the cytoplasmic cell division protein FtsA, which is largely responsible for the recruitment of FtsN to the divisome (23,24). The periplasmic domain of FtsN is further divided into three subdomains: a membrane-proximal portion containing three potential short a-helices, a glutamine-rich central region, and a PG-binding sporulation-related repeat (SPOR) domain at the C terminus, which binds preferentially to glycan chains devoid of stem peptides (25,26). The region located around a-helix 2 (Leu 75 -Gln 93 , E FtsN) is essential for the function of FtsN (19,27).
In this work, we have used complementary techniques including fluorescence anisotropy-binding assays, activity assays, crosslinking, and X-ray crystallography and obtained clear evidence that E FtsN region is sufficient for direct binding to PBP1b and the stimulation of its GTase activity in vitro. Furthermore, we determined the crystal structure of PBP1b in complex with E FtsN peptide and identified the binding site of E FtsN between the GTase and UB2H domains of PBP1b.

E FtsN is sufficient for the stimulation of the GTase activity of PBP1b
The stimulation of PBP1b in vitro by FtsN is well-established (5,28). In vivo, a small periplasmic domain of FtsN ( E FtsN: Leu 75 -Gln 93 ) (Fig. 1A) is required and sufficient for function in cell division (27), but the molecular mechanism of PBP1b activation remains largely unknown. The first indication that E FtsN was involved in PBP1b activation originates from the in vitro study of the FtsN mutant W83L. This residue change not only impairs FtsN activity in vivo (27) but is less able to stimulate PBP1b activity and to suppress FtsBLQ-mediated inhibition of PBP1b GTase activity in vitro (9). In addition, the interaction between FtsN W83L mutant and PBP1b was reduced compared with the WT protein. Based on these observations, we prepared the synthetic peptide ( E FtsN: Leu 75 -Gln 93 ) to test its effect on PBP1b. Interestingly, although a higher concentration than full-length FtsN protein was required (50 mM versus ;1 mM), presumably because of the lack of the transmembrane segment (see below), the peptide was able to stimulate the GTase activity of PBP1b (Fig. 1B) but has no effect on that of PBP1a (Fig. 1C), confirming that E FtsN directly contributes to PBP1b activation. This result is consistent with in vivo data showing that E FtsN (GFP fusion) was functional only when overexpressed (19,27). When the activity of the TPase domain of PBP1b was analyzed using S2d (a mimic of the D-Ala-D-Ala of the natural substrate) as substrate, the addition of FtsN had no effect (Fig. 1D), indicating that the protein only modulates the GTase activity directly but not the TPase activity of PBP1b.
It was previously shown, using different variants of FtsN (28), that multiple interaction sites contribute to the binding with PBP1b (28). The soluble form (sFtsN, D1-57), lacking the cytoplasmic tail and the transmembrane (TM) anchor, displayed significantly reduced stimulatory effect on the PBP1b activity (29). To compare the in vitro activity of the membrane-bound proximal region containing the three potential a-helices with the full-length protein, we have prepared a truncated variant of FtsN (FtsN 130 encompassing residues Met 1 to Arg 130 ) (Fig. 1, A and F). This variant was found to bind PBP1b and to stimulate its GTase activity to a lesser extent compared with full-length protein but was much more efficient in PBP1b activation than the E FtsN peptide (Fig. 1, E and F). Altogether, these data indicate that the N-terminal region, including the TM segment, is important for high-affinity binding and efficient activation of PBP1b. In addition, the central Gln-rich region and SPOR domain seem to contribute to the optimal activation of PBP1b. E FtsN interacts specifically with PBP1b but not with PBP3, FtsW-PBP3, or FtsBLQ To measure the binding between E FtsN and PBP1b or other divisome proteins, a fluorescent peptide (FITC-Lys 69 -Gln 93 ) containing a FITC fluorophore attached at the N terminus of the minimal sequence (Leu 75 -Gln 93 ) via six additional amino acids of FtsN (to avoid interference with binding) was prepared and used as a probe to develop a fluorescence anisotropy (FA) assay. Interaction studies with PBP1b show a high increase of the FA signal, and fitting of the graph allowed the determination of a dissociation constant k d of 8.1 6 1.7 mM ( Fig. 2A). In contrast, no significant change in the FA signal was observed with PBP1a, PBP3, FtsW-PBP3, and FtsBLQ complexes (Fig. 2, A and B). This result is consistent with direct and specific interaction of E FtsN with PBP1b and suggests that E FtsN alone is not sufficient for direct interaction with FtsW-PBP3 or FtsBLQ as previously suggested (27). This result is consistent with the absence of strong interaction between FtsN and FtsBLQ and with the fact that FtsN has no effect on the activity of PBP3 (9).
To further characterize the interaction between E FtsN and PBP1b, we performed cross-linking experiments using the FITC-Lys 69 -Gln 93 peptide to facilitate the visualization of the adduct. The heterobifunctional cross-linking agent 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC), a zero-length cross-linker that couples carboxyl groups to primary amines, efficiently coupled PBP1b to the fluorescent peptide (Fig. 2C). SDS-PAGE, followed either by Coomassie staining or fluorescence labeling, revealed a band slightly higher (PBP1b 1 peptide) than PBP1b alone labeled with fluorescent ampicillin. The band intensity increases when the concentration of the crosslinker was increased. This result further confirms that E FtsN interacts with PBP1b.

Crystal structure of PBP1b in complex with E FtsN
Having strong evidence of the binding of E FtsN to PBP1b, we performed co-crystallization assays between PBP1b (positions 58-804), in both the presence and the absence of moenomycin A, which helps in the crystallization of PBP1b, and purified E FtsN (Leu 75 -Gln 93 ) peptide to identify the binding site of this domain on PBP1b. Crystals, which were only obtained in the presence of moenomycin A, belong to the space group P2 1 2 1 2 (Table 1), similar to those of previously reported for PBP1b structures (30,31). The diffraction data were anisotropic, and therefore additional processing was carried out with Staraniso (32). The three axes of the ellipsoid used to cutoff the resolution were 2.3, 4.3, and 2.3 Å. The structure was solved to 2.4 Å resolution by molecular replacement and showed an additional  electron density between the GTase and the UB2H domains (Fig. 3, A-D). Because of its initial low quality, an attempt to place a helix in both directions and an elongated conformation of the main chain were initially tested. The best solution was selected based on the R free and the improvement of the electron density after refinement. The side-chain attribution was done using the secondary structure prediction of E FtsN, with Pro 79 and Ser 91 , respectively, capping the N-and C-terminal ends of the helix. An attempt at shifting the sequence in both directions on the helix did not lead to improved R free or electron density. This 13-amino acid helix mostly interacts with the GTase domain and runs approximately parallel to its last a12 helix. Three hydrogen bonds are observed with PBP1b, between the side chain of Trp 83 and the main-chain carbonyl of Gln 384 (a11a12 loop, GTase domain), between the main-chain carbonyl of Leu 89 and side-chain hydroxyl of Thr 140 (a1b2 loop, UB2H domain), and between the Glu 90 side chain and the Arg 141 side chain (a1b2 loop, UB2H), as well as the Leu 344 main chain (a9a10 loop, GTase) (Fig. 3, C-F). In addition, Tyr 85 , Ile 86 , and Leu 89 from E FtsN form a hydrophobic cluster with Leu 224 (a2), Leu 344 (a9a10 loop), Ile 386 , Leu 390 , and Leu 394 (a12) from the GTase domain (Fig. 3, C-F).
Compared with other PBP1b structures available, no major conformation change was observed upon E FtsN binding with a RMSD of 0.85 Å over 685 common a carbons (PDB code 5HLA) (Fig. 3B). The differences are however larger in the GTase domain (RMSD 1.0 Å) than in the TPase (RMSD 0.56 Å) and the UB2H (RMSD 0.38 Å) domains. In the structure of the PBP1b-E FtsN complex, the GTase domain is also characterized by overall poorer electron densities compared with the rest of the protein as well as, compared with the GTase domains of other structures available. Of note, the crystals were only obtained in the presence of moenomycin A; therefore it is possible that this ligand may prevent the conformational change induced by the peptide. Nevertheless, we observed a significant destabilization of the GTase domain. This is materialized by a higher mean B factor for a carbons of the GTase domain (125 Å 2 ) compared with other PBP1b structures (75 Å 2 for 5HLA), whereas the TPase and UB2H domains are characterized by lower and more similar values (50 and 48 Å 2 , respectively, versus 37 and 46 Å 2 in the 5HLA structure). This could indicate that the conformational change responsible for the increased GTase activity may indeed be hindered by either the binding of moenomycin A required for crystallization and/or the crystal packing, which strongly traps the GTase domain in an inactive state, whereas the peptide attempts to displace it to an active state.

Mutations in the E FtsN-binding pocket reduce activation of PBP1b by FtsN and induce cell elongation and lysis
To confirm the importance of the E FtsN-binding site in PBP1b activation and the attempt to establish a link between this site and cell division, we addressed the roles of residues Thr 140 and Arg 141 from the UB2H domain side of the binding cavity that interact with E FtsN via their side chains and the residue Arg 397 from the GTase domain side of this cavity, which participate in the same positively charged residues cluster as Arg 141 and could contribute to the interaction with E90 of E FtsN according to other PBP1b structures. All these residues were modified to Ala, and single or double mutations were introduced in the ponB gene. The activity of the mutants was first evaluated by complementation experiments using E. coli EJ801 (DponB-ponA ts ) (33) as a host strain (results in 10 g/liter LB medium or low salt 0.5 g/liter LB were comparable). The single mutants T140A, R141A, and R397A and the double mutants T140A/R141A and T140A/R397A were able to restore the growth of the strain at the nonpermissive temperature (42°C) (Fig. 4A), and the cells exhibited normal phenotype. In contrast, the double mutant PBP1b R141A-R397A was unable to rescue E. coli EJ801 at 42°C (Fig. 4A). This double mutant was stable at 42°C, and its GTase and Bocillin-binding activities were not affected after 1 h of incubation at this temperature (Fig. S1). The observation of the cells after 1 h at the nonpermissive temperature showed a mild elongated phenotype (L: 6.14 6 1.84 mM versus 2.17 6 0.46 mM) before lysis (Fig. 4, B-D). Moreover, although the purified PBP1b R141A/R397A mutant showed a minor or no difference in the GTase activity compared with the WT protein, its activation by FtsN was reduced by 2-fold (Fig. 4E). Interestingly, the activation of PBP1b R141A/R397A by LpoB remained unchanged (Fig. S2). These results indicate that, as a consequence of the mutations in the E FtsN-binding site, the activation of PBP1b by FtsN was reduced, causing defects in sPG integrity and cell division. This strongly confirms the link between PBP1b and FtsN during cell division and demonstrates the role of E FtsN in the activation of its GTase activity.

Discussion
The mechanisms of E. coli septal PG synthesis and regulation are complex and involve several trans-envelope factors, with
On the other hand, biochemical data have shown that FtsN interacts with the major PG synthase PBP1b and stimulates its GTase activity (5,28). Using different techniques including FAbinding assay, activity assay, cross-linking, X-ray crystallography, mutagenesis, and complementation, we provide evidence that E FtsN region is sufficient for direct binding to PBP1b and the partial stimulation of its GTase activity. Importantly, the crystal structure of the PBP1b-E FtsN complex allowed the identification of the binding site of E FtsN being located between the GTase and UB2H domains of PBP1b. The GTase domain accounts for most of the interaction with E FtsN, but two residues (Thr 140 and Arg 141 ) in the a1b2 loop of the UB2H domain pointing toward the GTase domain are also involved in the binding. This is consistent with the activation of the GTase but not the TPase activity of PBP1b by FtsN. The binding site of E FtsN is distinct from that of LpoB, which binds the exposed surface of UB2H domain (20,41). This is supported by the fact that FtsN and LpoB bind simultaneously to PBP1b and that the stimulation of the GTase activity of PBP1b by FtsN was synergistic with the stimulation by LpoB (5). Furthermore, the structure reveals that the conserved residues Trp 83 , Tyr 85 , and Leu 89 of E FtsN, which were shown to be important for the in vivo function of the protein (27), together with Ile 86 and Glu 90 are directly involved in the interaction with PBP1b. Interestingly, the mutant FtsN W83L was shown to exhibit reduced binding and activation of PBP1b (9). The double mutation R141A/R397A in the E FtsN-binding pocket of PBP1b reduces the activation of PBP1b by FtsN but not by LpoB. In addition, this mutant was unable to rescue DponB-ponA ts strain at nonpermissive temperature and induced a mild cell-chaining phenotype and cell lysis, indicating sPG synthesis and division defects. These data suggest that the observed chaining phenotype is the result of the partial suppression of the E FtsN-binding site. Altogether, the present results reveal that PBP1b, the major class A PBP, is one of the binding targets of E FtsN and that E FtsN is an important determinant for the stimulation of the GTase activity of PBP1b by FtsN. This could contribute to sPG synthesis regulation during cell division. However, our assays could not detect direct interaction between E FtsN and FtsW-PBP3 or FtsBLQ, as previously proposed (27), but this does not exclude that such an interaction occurs with one of these proteins when E FtsN is displayed in the context of the full-length FtsN or an alternative membrane-bound variant.
Overall, our results bridge the gap between the existing biochemical and in vivo data (9,19,27,28,42) and provide additional information to the general model of regulation of cell division in E. coli. FtsN seems to play multiple and intricate roles (regulation of sPG synthesis and late hydrolysis) (19,42) that are coordinated with events on both sides of the cytoplasmic membrane, starting early in the cell cycle via its interaction with FtsA in the cytoplasm (22,(43)(44)(45)(46) and the gradual accumulation throughout the cell cycle (47), using a self-enhanced positive feedback mechanism (19). During divisome assembly, the GTase activity of PBP1b (as well as the TPase activity of PBP3) is kept inactive by the FtsBLQ complex until FtsN reaches a critical threshold, which enables it to relieve this inhibition, probably by disrupting the interaction with FtsBLQ, and to restore PBP1b activity (via direct interaction with E FtsN) (9). This contribute to the global sPG synthesis, probably in

Regulation of PBP1b by FtsN
coordination with the FtsW-PBP3 complex (3,8,37). sPG synthesis then recruits the amidases that hydrolyze sPG and trigger cell constriction (42); this creates a high-affinity binding substrate (denuded PG) of the SPOR domain, allowing the recruitment of more FtsN, and accelerates the constriction process (25,26).
In E. coli, cell wall constriction depends mainly on sPG synthesis and not on FtsZ treadmilling (40,48). PBP1b was shown to play important role in sPG synthesis complementary to that of FtsW and PBP3 (37,40) and to exhibit two populations: fastand slow-moving fractions (49). The slow-moving fraction represents the enzymes catalytically active in PG synthesis. FtsN also exhibits dynamic behavior, and its localization was shown to be spatially separated from constricting FtsZ-ring (50). Because FtsN, a specific divisome protein, binds and activates PBP1b, it may play a role in the transition of PBP1b from the fast and low-activity state to the slow-moving processive enzyme at the division site.

Plasmid construction
The complete list of the plasmids used in this study is given in Table S1. All point mutations were introduced using the Q5 site-directed mutagenesis kit (New England Biolabs). The primers used in this study are shown in Table S2; they were purchased from Eurogentec (Angleur, Belgium). Details for plasmid constructions are described in the supporting information.

Expression and purification of proteins
These proteins were purified as previously described: PBP1bg (54), FtsN and FtsBLQ (9), PBP3 and FtsW-PBP3 (8), PBP1a (55), and LpoB (41). Further details for protein purifications that are not described elsewhere are provided in the supporting information. Of note, all the proteins are stable, PBP3 binds fluorescent ampicillin and hydrolyzes the substrate analog S2d (9), and the FtsW-PBP3 complex was shown to bind lipid II (56).

FA assay
FA experiments were performed to determine the binding affinity of the E FtsN peptide, consisting of the FITC-Lys 69 -Gln 93 used as probe, to PBP1bg. Serial dilutions of the proteins in their specific buffer were prepared in 384-well plates, and the probe was added at 100 nM final concentration in a final volume of 30 ml. The mixtures were incubated for 30 min at 25°C, and the FA signals were recorded using an Infinite® F Plex (Tecan, Männedorf, Switzerland) equipped with polarization filter with excitation wavelength at 485 nm and emission at 535 nm. FA values were calculated using the equations FA = (I || 2 G·I \ )/I || 1 2G·I \ ), where I || is the fluorescence intensity of emitted light parallel to excitation, I \ is the fluorescence intensity of emitted light perpendicular to excitation, and G is the correction factor that correct for instrument bias. The G factor is experimentally determined using the probe alone. For K d determination, the fluorescence anisotropy data were analyzed by nonlinear curve fitting using GraphPad Prism 6.0 software as described (56).

Continuous fluorescence GTase assay
The GTase activity assays with dansyl-lipid II as substrate were performed in a medium binding black 96-well microplate (Greiner Bio One) as described (57). The samples contained 10 mM dansyl-lipid II, 50 mM HEPES-NaOH, pH 7.5, 200 mM NaCl, 10 mM CaCl 2 , 0.085% of decyl-PEG, 20% of DMSO, and 1 unit of N-acetylmuramidase of Streptomyces globisporus (Sigma). The E FtsN peptide Leu 75 -Gln 93 was solubilized in 50 mM Tris-HCl, pH 8.0, and used at 50 and 125 mM concentrations. The proteins FtsN and FtsN 130 were used at 0.8 mM. The reactions were initiated by the addition of 30 nM PBP1bg (100 nM for PBP1a) and monitored by following the fluorescence decrease over 20 min at 30°C using an Infinite 200 PRO microplate reader (Tecan, Männedorf, Switzerland) with excitation wavelength at 340 nm and emission at 520 nm.

Effect of FtsN on the hydrolysis of S2d by PBP1b
The activity of the TPase domain of PBP1b was measured in the presence of S2d (analog of the peptide moiety) as a mimic of donor substrates as previously described (58,59). The assay was performed in a UV-Star 96-well microplate (Greiner Bio One) at 37°C in the presence of 50 mM phosphate buffer, pH 7.0, 2.0 mM S2d, 3.2 mM 4,49-dithiodipyridine, and 1.2 mM PBP1b. The absorbance at 330 nm was monitored with an Infinite M200 Pro microplate reader (Tecan, Männedorf, Switzerland). FtsN was used at 2 mM to test its effect on the PBP1b activity. The experiments were repeated three times with reproducible results.

Cross-linking experiments
PBP1bg (10 mM) was incubated with a 10-fold excess of FITC-E FtsN peptide and two conditions of the heterobifunctional cross-linker EDC (protein:EDC molar ratio 1:1000 and 1:2000) in buffer containing 50 mM HEPES-NaOH, pH 7.5, 0.3 mM NaCl, 0.7% CHAPS. The mixture was incubated for 2 h at room temperature and analyzed by SDS-PAGE followed by fluorescence imaging using a Typhoon Trio1 imager and Image Quant TL software (GE Healthcare). The PBP1b control (2 mM) was incubated with fluorescent ampicillin 10 mM for 30 min at 37°C prior to SDS-PAGE.
Crystallization, data collection, and structure determination Crystallization was carried out using the hanging drop vapor diffusion method at 20°C. The PBP1b (Lys 58 -Ser 804 ) concentration was 20 mg/ml in the buffer solution 20 mM Tris-HCl, pH 8.0, 0.3 M NaCl, and 4.5 mM DM and contained 1:1 molar ratio of moenomycin A and a synthetic peptide corresponding to E FtsN (LPPKPEERWRYIKELESRQ) in a 3:1 molar ratio. The crystals were grown in drops made of 2 ml of protein solution and 2 ml of precipitant solution containing 0.1 M (NH4) 2 SO 4 , 0.3 M sodium formate, 0.1 M Tris-HCl, pH 7.8, 3% (w/v) lowmolecular-weight poly-g-glutamic acids, and 20% (v/v) PEG 550 monomethyl ether. The cryoprotectant solution was made of 22% (w/v) PEG 6000 and 30% (v/v) PEG 400. The diffraction data were collected on the Proxima 1 beamline of the Soleil Synchrotron (Paris-Saclay). The data were indexed, integrated, and scaled using XDS and reached a 2.4 Å resolution (60). Because of the anisotropy of the data, additional processing was carried out with Staraniso (32). The three axes of the ellipsoid used to cutoff the resolution were 2.3, 4.3, and 2.3 Å. The crystal used belongs to the orthorhombic P2 1 2 1 2 space group. The structure was solved by molecular replacement with Phaser (61) using the PBP1b structure of PBD code 3VMA as search model (30). The refinement and model building cycles were respectively performed with buster (BUSTER version 2.10.2, Global Phasing Ltd., Cambridge, UK) and Coot (62). A summary of the relevant statistics of the data collection and refinement is given in Table 1. The figures were prepared using PyMOL (PyMOL molecular graphics system, version 1.7.4.3 Enhanced for Mac OS X, Schrödinger, LLC). The coordinates and structure factors of the structure have been deposited in the Protein Data Bank under code 6YN0.

Complementation assay
To test the activity of the PBP1b mutants in vivo, we tested their ability to complement E. coli strain EJ801, which lacks PBP1b and has a thermosensitive PBP1a (33). EJ801 cells were transformed with the corresponding plasmids and grown in LB at 30°C with 50 mg/ml kanamycin. The preculture was then diluted in 30 ml of fresh LB (10 g/ml NaCl) or LB with 0.5 g/ml NaCl medium to an A 600nm value of 0.04 and grown at 30°C until A 600nm = 0.1-0.3, and the absorbance of the culture was monitored for 4 h at 42°C.

Microscopy and image analysis
Samples were taken from the complementation cultures (42°C) at different time (1-h intervals), and the cells were fixed as described (63). Photographs were taken with a cooled AxioCam MRm (Zeiss) mounted on a Zeiss Axio Imager. Z1 microscope, and the images were acquired in phase-contrast mode using the AxioVision Rel. 4.5 (Zeiss) software. The cell morphologies analysis were determined using ImageJ software (RRID:SCR_003070) running under plugin ObjectJ (RRID:SCR_019190).

Data availability
All data relevant to this work are contained within this article and the associated supporting information. The structural data are available in RCSB Protein Data Bank under the accession number: 6YN0.