FtsEX-mediated regulation of inner membrane fusion and cell separation reveals morphogenetic plasticity in Caulobacter crescentus

During its life cycle, Caulobacter crescentus undergoes a series of coordinated shape changes, including generation of a polar stalk and reshaping of the cell envelope to produce new daughter cells through the process of cytokinesis. The mechanisms by which these morphogenetic processes are coordinated in time and space remain largely unknown. Here we demonstrate that the conserved division complex FtsEX controls both the early and late stages of cytokinesis in C. crescentus, namely initiation of constriction and final cell separation. ΔftsE cells display a striking phenotype: cells are chained, with skinny connections between cell bodies resulting from defects in inner membrane fusion and cell separation. Surprisingly, the thin connections in ΔftsE cells share morphological and molecular features with C. crescentus stalks. Our data uncover unanticipated morphogenetic plasticity in C. crescentus, with loss of FtsE causing a stalk-like program to take over at failed division sites and yield novel cell morphology. Author Summary Bacterial cell shape is genetically hardwired and is critical for fitness and, in certain cases, pathogenesis. In most bacteria, a semi-rigid structure called the cell wall surrounds the inner membrane, offering protection against cell lysis while simultaneously maintaining cell shape. A highly dynamic macromolecular structure, the cell wall undergoes extensive remodeling as bacterial cells grow and divide. We demonstrate that a broadly conserved cell division complex, FtsEX, relays signals from the cytoplasm to the cell wall to regulate key developmental shape changes in the α-proteobacterium Caulobacter crescentus. Consistent with studies in diverse bacteria, we observe strong synthetic interactions between ftsE and cell wall hydrolytic factors, suggesting that regulation of cell wall remodeling is a conserved function of FtsEX. Loss of FtsE causes morphological defects associated with both the early and late stages of division. Intriguingly, without FtsE, cells frequently fail to separate and instead elaborate a thin, tubular structure between cell bodies, a growth mode observed in other α-proteobacteria. Overall, our results highlight the plasticity of bacterial cell shape and demonstrate how altering the activity of one morphogenetic program can produce diverse morphologies resembling those of other bacteria in nature.

6 support Z-ring assembly, but constrict poorly [12]. Since FtsE interacts with FtsZ in E. 115 coli, one possibility is that FtsEX functions as a membrane anchor for FtsZ and utilizes 116 ATP binding and hydrolysis to regulate Z-ring constriction [13,6]. 117

118
In this study, we were originally motivated to characterize FtsEX as a novel membrane 119 anchor for FtsZ in C. crescentus since FtsE is one of the first proteins recruited to the 120 nascent division site and is important for efficient cell separation and Z-ring assembly 121 and/or stability [14,15]. Considering the conserved function of FtsEX as a modulator of 122 cell wall remodeling, we asked whether FtsEX, in addition to promoting Z-ring structure, 123 regulates cell wall cleavage in C. crescentus. We find that ftsE has strong synthetic cell 124 separation defects with cell wall hydrolytic factors. Interestingly, however, deleting ftsE 125 produces chains of cell bodies connected by thin, tube-like connections that contain all 126 layers of the cell envelope. This is in stark contrast to the thick, uncleaved septa and 127 To begin to address the role of FtsEX in C. crescentus, we first attempted to make ftsE 140 and ftsX deletion strains. Although ftsE is annotated as essential [16,14], we successfully 141 made several independent ∆ ftsE clones [15]. ftsX is also annotated as essential [16], but 142 unlike ftsE, we were unable to make an ftsX deletion, depletion, or overexpression strain, 143 suggesting that C. crescentus cells are highly sensitive to changes in FtsX levels. To 144 understand the role of the FtsEX complex in C. crescentus morphogenesis, we focused on 145 characterizing the ftsE mutant in detail. slowly than WT [15]. Transmission electron microscopy (TEM) offered us better 149 resolution of cells lacking FtsE (Fig 1). ∆ ftsE cell bodies were heterogeneous in length, 150 but overall appeared elongated compared to WT, which suggests a delay or inefficiency 151 in the initiation of constriction. The thin connections between ∆ ftsE cell bodies were also 152 heterogeneous in length, with some extending hundreds of nanometers, and had 153 dimensions qualitatively similar to those observed for stalks ( Fig 1B) clusters of puncta instead of focused Z-rings (Fig 2) [15]. These data suggest that FtsE 162 may regulate early Z-ring structure and/or assembly. Consequently, we tested if ftsE 163 interacted genetically with the positive Z-ring regulator zapA, which, like FtsE, is also 164 recruited early to midcell by FtsZ in C. crescentus [17,14]. ∆ zapA∆ftsE cells displayed 165 mild synthetic growth and cell length defects, but had severely disrupted, diffuse Z-ring 166 structures (Fig 2). We also observed that ∆ zapA∆ftsE cells were very sensitive to even 167 slight increases in FtsZ levels and were noticeably filamentous after only one hour of 168 ftsZ-cfp expression ( Fig 2D) FtsZ. Overexpression of either ftsE or ftsEX caused dramatic filamentation, and 176 overexpression of ftsE alone also caused ectopic poles to form (Fig 3). After four hours of 177 FtsE overproduction, instead of Z-rings, FtsZ-CFP formed discrete puncta along the 178 length of the filamentous cells ( Fig 3A). Interestingly, when we overproduced FtsEX, 179 FtsZ-CFP localized in a drastically different pattern, as multiple wide bands ( Fig 3A). C. 180 crescentus Z-ring positioning is in part dictated by a negative regulator of FtsZ assembly 181 called MipZ, which forms a complex near the origin of replication [18]. After the polar 182 origin region is duplicated, the second copy is quickly transported to the opposite cell 183 9 pole. Bipolar MipZ thereby directs Z-ring assembly at midcell by inhibiting FtsZ 184 polymerization at the poles [18]. MipZ-YFP localized at the poles and as fairly regularly 185 spaced puncta in cells overproducing FtsE or FtsEX, but its localization was more diffuse 186 in cells overproducing FtsEX (Fig 3B). We interpret the MipZ localization data as 187 evidence that chromosomal replication and segregation still occur in cells overproducing 188 containing protein with a signal peptide, two N-terminal CC domains, and a C-terminal 1 1 LytM domain (Fig S1A). We hypothesized that C. crescentus FtsEX-LdpF-AmiC may 229 function in an activation pathway analogous to E. coli FtsEX-EnvC-AmiA/B. 230

231
We first adopted a genetic approach to investigate the role of FtsEX in cell wall 232 hydrolysis during division. In E. coli, FtsEX is required for EnvC's localization at 233 midcell [9]. However, LdpF-mCherry is diffuse, and we did not observe differences in its 234 localization between WT and ∆ ftsE cells (Fig S1B). chaining defects, which is consistent with them acting in a common pathway (Fig 4). 243 244 E. coli cells lacking EnvC depend on NlpD for cell separation: simultaneous inactivation 245 of either EnvC or FtsEX and NlpD results in severe chaining, supporting the hypothesis 246 that FtsEX activates EnvC's ability to promote septal PG cleavage [22]. In C. crescentus, 247 the LytM protein most closely related to NlpD is DipM, which has similar domain 248 organization to NlpD but differs in that it is not associated with the outer membrane.  (Fig 1), electron cryotomography (ECT) allowed us to dissect the exact stages at 288 which these cells are blocked during division (Fig 6). To capture the cell envelope 289 organization at the skinny cell-cell connections, we imaged cells lacking both FtsE and 290 AmiC since loss of AmiC exacerbates the ∆ ftsE chaining phenotype (Fig 4). In five out 291 of six tomograms, we could identify with certainty the presence of all layers of the cell 292 envelope in the skinny connections between chained cell bodies (Fig 6). Four of these 293 had obvious cytoplasmic volume between the unfused inner membranes. In at least one 294 example, however, the inner membranes were closely stacked on top of each other, but 295 not fused (Fig 6). 296 1 4 In WT C. crescentus, the final stages of inner membrane fission are rapid, and the 298 smallest diameter for inner membrane connections that have been captured by ECT are 299 ~60 nm [31]. In cells lacking FtsE and AmiC with fairly uniform connections (Fig 6B,  300 D), we observed inner membrane diameters ranging from ~12 to 60 nm. Others were 301 more variable and had intermittent bulging, with inner membrane diameters ranging from 302 ~20 to 300 nm within a single cell-cell connection (Figs 6C, S3). This organization of the 303 cell envelope is strikingly different from other mutants deficient in cell wall hydrolysis. 304

E. coli cells lacking EnvC and NlpD or all four LytM-domain containing factors 305
complete inner membrane fusion and cytoplasmic compartmentalization, but struggle to 306 constrict their outer membrane due to a layer of intact PG between adjacent chained cells 307 [22]. Similarly,  (Figs 1, 6). In addition to sharing approximate widths 319 (~12-300 nm inner membrane diameter for the skinny connections; ~20-40 nm inner 1 5 membrane diameter for the stalks) and cell envelope organization, we occasionally 321 observed electron dense structures that spanned the short axis of the cell envelope of the 322 thin connections (Fig 6). These structures were reminiscent of stalk cross-bands, 323 multiprotein assemblies that transect the stalk at regular intervals and function as 324 diffusion barriers to compartmentalize stalk and cell body periplasmic and membrane 325 proteins (Figs 6A,C, D, S3) [33]. but was not enriched at the extended constriction sites (Fig. 7D). Consequently, the 360 proteinaceous, envelope-spanning discs observed in the ftsE mutant by ECT (Fig 6) may 361 not, in fact, be cross-bands or may differ in molecular composition from stalk cross-362 bands. We conclude that the skinny connections share numerous morphological and 363 molecular similarities with stalks, but the two structures are not physically or 364 biochemically identical. 365 1 7

Discussion 367
The role of FtsEX in synchronizing PG remodeling with cell division appears to be 368 conserved amongst distantly related bacterial species such as E. coli, S. pneumoniae, and 369 M. tuberculosis, although the downstream adaptor or enzyme targets vary [7-11]. We 370 provide evidence that this paradigm also extends to the α -proteobacterium C. crescentus. 371 Our data indicate that FtsE is important for initial Z-ring assembly and regulates Z-ring 372 structure in a manner dependent on its stoichiometry with FtsX (Figs 2, 3, 8A). Different 373 levels of FtsE or FtsEX not only affect FtsZ localization, but also FtsZ function, namely 374 its ability to localize incorporation of new cell wall material (Fig 3). Additionally, our 375 data implicate FtsEX in a cell wall metabolic pathway involving LdpF and an 376 unidentified downstream cell wall factor regulated by LdpF (Fig 4, 8B). Thus C. 377 crescentus FtsEX, similar to what has been proposed in E. coli, may synchronize PG 378 remodeling with Z-ring constriction during division [9]. ECT of the skinny connections 379 in ∆ ftsE revealed a cell envelope architecture remarkably distinct from E. coli hydrolase 380 mutants, however, and an overall morphology that was strikingly stalk-like (Fig 6). connected only by a small tubular structure [31]. Out of thousands of cells Judd and 1 8 colleagues examined in their study, only five displayed inner membranes with diameters 390 less than ~100 nm and the smallest inner membrane connection was 60 nm in diameter 391 [31]. ECT of cells lacking FtsE and AmiC with fairly uniform connections showed inner 392 membrane diameters ranging from ~12 to 60 nm (Fig 6). Thus, the majority of cells 393 lacking FtsE and AmiC have inner membrane diameters that fall well below the lowest 394 threshold reported for inner membrane diameters at any stage of WT cell division. 395 Furthermore, WT cells spend a short amount of time in these late, transitional stages, 396 perhaps only a few seconds, and membrane topology changes very rapidly [31].  are not, in actuality, ectopic stalks forming at failed division sites: the spatial pattern of 454 PG incorporation is distinct from stalks, the diameters are not as homogeneous as for 455 stalks, and StpB does not localize at the skinny connections (Figs S3, 6, 7). We instead 456 favor the hypothesis that when  terminus, were purified as described previously with minor changes [15]. Rosetta cells 546 containing the constructs were grown in 1 L of LB at 30°C to an OD 600 of 0.4 and then 547 induced with 1 mM IPTG for 4 h. Cells were collected by centrifugation at 6000 x g at 548 4°C for 10 minutes and resuspended in 40 ml Column Buffer A (CBA: 50 mM Tris-HCl 549 pH 8.0, 300 mM NaCl, 10% glycerol, 20 mM imidazole) per 1 L of culture. Cells were 2 5 snap-frozen in liquid nitrogen and stored at -80°C until use. Pellets were thawed at 37°C 551 and lysozyme was added to 1 µg/ml and MgCl 2 to 2.5 mM. Cell suspensions were left on 552 ice for 45 minutes, then sonicated and centrifuged for 30 minutes at 15,000 x g at 4°C. 553 The protein supernatant was filtered and loaded onto a HisTrap FF 1ml column (GE Life 554 Sciences) pre-equilibrated with CBA. The protein was eluted with 30% Column Buffer 555 B (same as CBA except with 1M imidazole). The protein fractions were combined and 556 His 6 -Ulp1 (SUMO protease) was added (1:500 Ulp1:protein molar ratio). The protease 557 and protein fractions were dialyzed overnight at 4°C into CBA. Cleaved protein was run 558 over the same HisTrap FF 1mL column equilibrated in CBA and the flow-through was 559 collected. Flow-through fractions were dialyzed overnight at 4°C into Storage Buffer (50 560 mM HEPES-NaOH pH 7.2, 150 mM NaCl, 10% glycerol). Dialyzed protein was then 561 concentrated (if needed), snap-frozen in liquid-nitrogen, and stored at -80°C. 562

RBB labeled sacculi preparation 564
Sacculi were prepared from strain EG865 as described in [23]. C. crescentus cells were 565 grown in 1 L of PYE at 30°C, collected at an OD 600 of 0.6 by centrifugation at 6,000 x g 566 for 10 minutes, and resuspended in 10 ml of PBS. The cell suspension was added drop 567 wise to 80 ml of boiling 4% sodium dodecyl sulfate (SDS) solution. Cells were boiled 568 and mixed for 30 minutes and then incubated overnight at room temperature. Sacculi 569 were then pelleted by ultra-centrifugation at ~80,000 x g for 60 minutes at 25°C. Pelleted 570 sacculi were then washed four times with ultra-pure water and resuspended in 1 ml of 571 PBS and 20 µl of 10 mg/ml amylase and incubated at 30°C overnight. The next day, 572 sacculi were pelleted at ~400,000 x g for 15 minutes at room temperature, washed three 573 2 6 times with ultra-pure water, and resuspended in 1 ml of water. The sacculi suspension 574 was labeled with 0.4 ml of 0.2 M remazol-brilliant blue (RBB), 0.3 ml 5 M NaOH, and 575 4.1 ml of water, and incubated at 30°C overnight. The labeled solution was neutralized 576 with 0.4 ml of 5 M HCl and 0.75 ml of 10X PBS. Labeled sacculi were pelleted at 577 16,000 x g for 20 minutes at room temperature. The pellet was washed with water until 578 the supernatant was clear. Blue-labelled sacculi were resuspended in 1 ml of 0.2% azide, 579 incubated at 65°C for 3 hours, and then stored at 4°C. 580 581

Dye-release assay 582
The dye release assay was adapted from [23]. Briefly, 10 µl of RBB-labeled sacculi was 583 incubated at 30°C for 3 hours with AmiC, LdpF variants, DipM variants, or FtsX ECL 584 singly or in combination. All proteins were used at 4 µM. Total reaction volumes were 585 brought to 100 µl with PBS. Lysozyme (4 µM) was used as a positive control. After 3 586 hours of incubation, reactions were heat inactivated at 95°C for 10 minutes and 587 centrifuged for 20 minutes at 16,000 x g. Supernatants were collected and the absorbance 588 was measured at OD 595 . 589