CryoEM structure of the Vibrio cholerae Type IV competence pilus secretin PilQ

Natural transformation is the process by which bacteria take up genetic material from their environment and integrate it into their genome by homologous recombination. It represents one mode of horizontal gene transfer and contributes to the spread of traits like antibiotic resistance. In Vibrio cholerae, the Type IV competence pilus is thought to facilitate natural transformation by extending from the cell surface, binding to exogenous DNA, and retracting to thread this DNA through the outer membrane secretin, PilQ. A lack of structural information has hindered our understanding of this process, however. Here, we solved the first ever high-resolution structure of a Type IV competence pilus secretin. A functional tagged allele of VcPilQ purified from native V. cholerae cells was used to determine the cryoEM structure of the PilQ secretin in amphipol to ∼2.7 Å. This structure highlights for the first time key differences in the architecture of the Type IV competence pilus secretin from the Type II and Type III Secretin System secretins. Based on our cryoEM structure, we designed a series of mutants to interrogate the mechanism of PilQ. These experiments provide insight into the channel that DNA likely traverses to promote the spread of antibiotic resistance via horizontal gene transfer by natural transformation. We prove that it is possible to reduce pilus biogenesis and natural transformation by sealing the gate, suggesting VcPilQ as a new drug target.

proteins. Regardless, the size difference between these contaminants (<100 kDa) 144 and the PilQ multimer (~860 kDa) made it easy to distinguish PilQ from the milieu 145 in electron micrographs (Supplemental Figure 3A). 146 To confirm that full length PilQ was present in our samples (residues 30-571 after 147 the 29 amino acid N-terminal signal peptide is cleaved), gel band analysis with a 148 trypsin digest and mass spectrometry was used to analyze the multimeric PilQ in 149 amphipol from a SDS page gel (band marked with an * in Supplemental Figure  150 1D was analyzed). The results demonstrated 65% sequence coverage, with 151 fragments identified in each domain of the folded protein from residues 50 to 567 152 (Supplemental Figure 2). 153

Single particle cryoEM of the Type IV competence pilus secretin PilQ 154
Here we report the high-resolution structure of the purified Type IV competence 155 pilus secretin VcPilQ by single particle cryoEM (Figure 1) Grids with R2/2 spacing. A Titan Krios operating in a three by three pattern with 160 beam-image shift 33 was used to collect 3 movies per hole, resulting in 3,808 161 movies. The movies were motion corrected and screened for quality, which left 162 2,510 micrographs. 163 The cryoSPARC blob picker was used to identify 3,100,353 putative particle 164 location, several micrographs were inspected to adjust the threshold parameters to  Figure 3C). After the initial 2D classification runs, 66,507 particles were excluded from the dataset and used to generate a nonsense initial model 170 representing contamination. The 185,812 quality particles ('include') were used to 171 generate an initial model. A CryoSPARC heterogeneous refinement using the high 172 quality initial model and the contamination initial model was used to further clean 173 the dataset. Several rounds of 3D classification and 3D refinement were used to 174 select 100,543 particles for further analysis in Relion. We were able to recognize and model residues 160 to 571 of PilQ ( Figure 1C, 180 Supplemental Movie 1). Noting that residues 1-29 are cleaved (the N-terminal 181 signal peptide), our mass spectrometric analysis of the purified protein 182 (Supplemental Figure 2) demonstrated the presence of residues between 50 and 183 567, including the AMIN domain (residues 54-125) that is thought to interact with 184 the peptidoglycan in the periplasm 22 . While the AMIN domain was not resolved in 185 the PilQ structure, hazy density is present in the 2D classification in the region near 186 the PilQ N-terminus where we would expect the AMIN domain (marked with 187 asterisks in Figure 1B). Following the AMIN domain, residues 126-159 are 188 predicted by homology modeling to be unstructured (Supplemental Figure 7), so 189 they are also likely included in the hazy density. If that region is unstructured, it is a 190 likely source of the conformational heterogeneity observed as a hazy density in 191 bind the AMIN domain and act as a bridge between the inner and outer membrane 197 components of the Type IV pilus machinery 36 .

The structure of VcPilQ demonstrates similarities to other secretins 199
In each PilQ monomer, four beta strands come together to form a beta sheet 200 ( Figure 1C). Once assembled, PilQ forms a 56-strand beta-barrel. Inside the barrel, 201 two further beta hairpins (beta strand-turn-beta strand each) form a gate. These  DNA cargo of the Type IV competence pilus is also negatively charged, so it is 216 possible that this electrostatic repulsion will help the cargo pass through the cavity, 217 rather than getting stuck. By comparison, for T2SS secretins, the charge alternates 218 in the inner cavity. 219

The N0 domain in VcPilQ agrees with previous structures 220
The N0 domain of VcPilQ (residues 160 to 227) was resolved 4 to 7 Å local 221 resolution. This allowed us to build a model based on homology to previously 222 solved structures of N0 domains (Figure 1C Figure 10E). 234

VcPilQ uses a novel helical coil to transition into the N3 domain 235
In VcPilQ, a 32 Å alpha-helix follows the N0 domain ( Figure 1C). None of the 236 T2SS or T3SS structures contain a helical coil to link N-terminal domains; instead 237 their periplasmic protein domains are linked by unstructured loops (Figure 3). This 238 difference was not anticipated, as homology models of VcPilQ based on other 239 structures more closely match the previously published T2SS and T3SS structures 240 (Supplemental Figure 7). 241 Following the end of this helix, the protein chain abruptly changes direction (~104˚ 242 angle) as the coil flows into the N3 domain ( Figure 1C). This dramatically reduces 243 the channel diameter, from 90 Å at the bottom of the N0 domain to 60 Å across the 244 N3 domain (Figure 4). In the T2SS and T3SS structures, the diameter of the 245 channel is relatively constant. 246

The putative outer membrane region of VcPilQ is thicker than T2SS secretins 247
The secretin amphipathic helix lip (AHL) is thought to be a key determinant for 248 secretin outer membrane insertion and among Type IV pilus secretins, the AHL is 249 conserved 18 . The AHL is thought to mark the lower boundary of the outer 250 membrane region of secretins 18,29,30 . The upper boundary of the outer membrane is 251 less clear since the micelle density in cryoEM reconstructions does not always 252 match the positions of aromatic residues 25 . 253 In our VcPilQ atomic model, the distances between the bottom of the AHL to the 254 top of the beta strands is about 3 nm. As seen in Figure 3C, this putative outer 255 membrane region of VcPilQ is substantially taller than the same region in the 256 previously published T2SS structures. To investigate more than just the residue 257 locations, an inverted mask based on the atomic model density of VcPilQ was 258 generated and subtracted from the empirical cryoEM density, which reflects 259 unmodeled density in the cryoEM map that is not accounted for by the atomic 260 model ( Figure 4C). This presumed micelle density (in grey) blooms around the rescue, which may be due to the fact that these cysteines are located further down in 289 the gate region making them less accessible to the reducing agent. 290

291
Here we present the first high-resolution structure of a bacterial Type IV 292 competence pilus secretin. We observed key differences in the outer membrane 293 region and the periplasmic region among the different members of the secretin 294 family. These differences identify weaknesses in relying on homology models of 295 evolutionarily related secretins, like the T2SS secretin GspD, to understand PilQ. 296

Homology modeling was insufficient to predict a Type IV Pilus Secretin 297 structure 298
Before we solved the VcPilQ structure, we used homology modeling to predict the 299 structure of VcPilQ based on its sequence and the structures of previously solved 300 secretins 42 . The top five predictions are shown Supplemental Figure 7. 301 Comparing these predictions to our now-solved structure reveal significant 302 differences, including the thick outer membrane domain of VcPilQ and the extent 303 of the alpha helix between the N0 and N3 domains (Supplemental Figure 7). 304 Additionally, homology modeling cannot predict the three-dimensional 305 arrangements and orientations of the protein domains, the number of subunits, or 306 the inner barrel diameter. These features can be hypothesized based on T2SS or T3SS secretin structures, but that strategy would neglect the clear variations in 308 selectivity and functionality across the Secretin superfamily. It is possible that 309 structural variability between secretins also indicates which regions of the protein 310 are more often repurposed by evolution to generate new function and/or selectivity. 311

Outer membrane-spanning domain 312
Curiously, in some previous secretin structures, the trans-outer-membrane region 313 appears to be only 2-3 nm thick ( Figure 3C), which is much thinner than a typical 314 membrane. This has left it unclear which regions of secretin molecules are actually 315 embedded in the outer membrane, and whether any residues are exposed to the 316 extracellular surface. To investigate these questions, we used the Positioning of  organization of MxPilQ is nearly identical to that of VcPilQ 45 (Supplemental 332 Figure 11A), but it has two more AMIN domains at its N-terminus (Supplemental 333 Figure 11A). The Phyre2 protein model of MxPilQ also looks remarkably similar to the structure of VcPilQ solved here (Supplemental Figure 11B) 46   The Type IV competence pilus machinery extends and retracts the Type IV 378 competence pilus through the PilQ outer membrane pore. In our structure of 379 VcPilQ in the absence of the pilus and accessory proteins, the inner diameter of the 380 channel ranges from 25 to 108 Å (Figure 4A). 381 The V. cholerae Type IV competence pilus is a Type IVa pilus. The major pilin 382 subunit in the T4aP is typically smaller than the major pilin subunits in the T4bP 49 . T. thermophilus demonstrated the presence of narrow (4.5 nm) and wide (7 nm) pili 389 structure (in grey) 52 . The gate region with diameters of 25 Å at the lower gate and 391 36 Å in the upper gate clashes with the pilus (Figures 4A and 7D). VcPilQ was 392 solved in the absence of the pilus, so this gate conformation likely represents a 393 "closed" state in cases where the pilus is absent or fully retracted. The next 394 narrowest region in the VcPilQ channel is across the periplasmic N3 ring (~60 Å in 395 Figure 4A). Some of the previously solved T4aP are 6 to 7 nm in diameter, which 396 would be a tight fit in this region. The N3 domain is connected to the secretin 397 domain by a loop, so it could possibly expand to accommodate a 70 Å pilus. 398 We used homology modeling to predict the VcT4aP monomer structure, but 399 because Neuhaus et al. reported both wide and narrow pili assemebled by pilins 400 whose subunit structure are almost identical, we do not feel confident guessing the 401 VcT4aP diameter 53 . Regardless, the gates in VcPilQ would have to move to 402 accommodate a pilus. Consistent with this notion, our disulfide locked cysteine pair 403 mutants could not extend pili in the absence of reducing agent, which strongly 404 suggests that the conformation adopted in the presented structure (in which these 405 cysteine pairs would be in close enough proximity to disulfide bond) represents the 406 closed gate conformation and cannot accommodate a pilus fiber. Additionally, the 407 sub-tomogram averaging of the MxT4aP demonstrates a secretin conformational 408 change with and without a pilus present (Figure 7B-C) 44 . Here we report the first high-resolution structure of a Type IV competence pilus 418 secretin, V. cholerae PilQ. This protein complex facilitates DNA uptake into 419 diverse bacterial species to aid in their evolution, and thus, represents a potential 420 target for therapeutic intervention. The V. cholerae Type IV competence pilus is a 421 model system to study natural transformation in bacteria. We identify key 422 differences between VcPilQ and the previously published structures of T2SS and 423 T3SS secretins. We designed cysteine pair mutants to reversibly seal the VcPilQ 424 gate and inhibit natural transformation, which can be used as a tool to further 425 investigate the function of the Type IV competence pilus machinery in situ. We 426 suggest a structural rearrangement that would transition our closed VcPilQ into a 427 piliated state that could accommodate previously solved structures of Type IV pili. 428 We also compare our structure to previous T4aP sub-tomogram averaging results in       Only the N3 and Secretin domains were compared. RMSD was calculated with 582 MatchMaker in Chimera. The RMSD results are summarized in Table (E).   To test the impact of cysteine pair mutants on pilus biogenesis, V. cholerae strains 626 were generated akin to those described above, with the exception that the retraction 627 ATPase PilT was deleted and the strains contained a cysteine substitution mutation 628 in the major pilin that allows for competence pilus labeling as previously 629 described 15 . The full genotype of the parent strain (TND2244) was ∆lacZ::Pbad-630 10XHis-PilQ CmR, ∆pilT::TmR, ∆pilQ::TetR, ∆CTX::KanR, ∆MSHA::CarbR, 631 ∆luxO, ∆TCP::ZeoR, pilA S67C, comEA-mCherry, Ptac-tfoX. The cysteine pair 632 mutants were isogenic other than the cysteine mutations introduced into the Pbad-All strains were generated by natural transformation and cotransformation exactly 636 as previously described 55 . 637

Natural transformation assays 638
Natural transformation assays were performed exactly as described in 15   Purification 661 Cell pellet (15 g) was resuspended in lysis buffer (

Model Building and Refinement 700
The initial model (residues 230-571) was auto-built using Buccaneer 65 . Subsequent 701 building and model adjustments were performed by hand using COOT 66 . A 702 homology model of the N0 domain (residues 160-229) was created using I-703 TASSER and manually docked using COOT 42,67,68 . Coulombic potential density for 704 residues 1-159 were not observed. The model was refined in PHENIX version 1.16-705 dev3549 using phenix.real_space_refine with the resolution set to 3 Å 69 . NCS 706 constraints were applied for the 14 subunits and were automatically detected and 707 refined. Automatically determined secondary structure restraints, rotamer restraints, 708 and Ramachandran restraints were applied as well. The quality of the model was 709 evaluated using EMRinger 70 and Molprobity 71 (Table 1). 710

Mass Spectrometry 711
After running a BioRad Stain Free gel and performing a coomassie staining, the 712 band of interest was excised with a clean razor blade. The gel piece was destained 713 with ammonium bicarbonate and reduced with DTT (50˚C, 30 minutes). Next the The gel pieces were then dehydrated. Trypsin was used to digest the protein in the 716 gel (37˚C, overnight). Peptides were extracted from the gel matrix, dried, and de 717 salted with a zip tip. 718 The in-gel-digested samples were subjected to LC-MS/MS analysis on a nanoflow 719 LC system, EASY-nLC 1200, (Thermo Fisher Scientific) coupled to a QExactive 720 HF Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) 721 equipped with a Nanospray Flex ion source. 722 Samples were directly loaded onto a C18 Aurora series column (Ion Opticks, 723 Parkville, Australia). The 25cm x 50µm ID column (1.6 µm) was heated to 45° C. 724 The peptides were separated with a 60 min gradient at a flow rate of 350 nL/min. 725 The gradient was as follows: 2-6% Solvent B (3.5 min), 6-25% B (42.5 min), and 726 25-40% B (14.5min), to 100% B (1min) and 100% B (12min). Solvent A consisted 727 of 97.8% H2O, 2% ACN, and 0.2% formic acid and solvent B consisted of 19.8% 728 H2O, 80% ACN, and 0.2% formic acid. 729 The QExactive HF Orbitrap was operated in data dependent mode. Spray voltage 730 was set to 1.8 kV, S-lens RF level at 50, and heated capillary at 275 °C. Full scan 731 resolution was set to 60,000 at m/z 200. Full scan target was 3 × 10 6 with a 732 maximum injection time of 15 ms (profile mode). Mass range was set to 300−1650 733 m/z. For data dependent MS2 scans the loop count was 12, target value was set at 1 734 × 10 5 , and intensity threshold was kept at 1 × 10 5 . Isolation width was set at 1.2 m/z 735 and a fixed first mass of 100 was used. Normalized collision energy was set at 28. 736 Peptide match was set to off, and isotope exclusion was on. Ms2 data was collected 737 in centroid mode.