Evolution of a large periplasmic disk in Campylobacterota flagella enables both efficient motility and autoagglutination

Article Evolution of a large periplasmic disk in Campylobacterota flagella enables both efficient motility and autoagglutination Graphical abstract Highlights dThe C. jejuni flagellar motor has an extra-large basal disk dThe basal disk promotes filament unwrapping from the cell surface dReduction of cell-surface glycosylation is required for swimming without a basal disk Authors Eli J. Cohen, Tina Drobni c, Deborah A. Ribardo, ..., Daisuke Nakane, David R. Hendrixson, Morgan Beeby Correspondence [email protected] In brief The human pathogen Campylobacter jejuni (C. jejuni) has a larger, more complex flagellar motor than that of the model organism Salmonella typhimurium. In Cohen and Drobni c et al. the authors demonstrate that one function of the added complexity of the motor is to promote efficient directional switching while swimming and to prevent cell clumping. Cohen et al., 2024, Developmental Cell 59, 3306–3321 December 16, 2024 ª2024 The Author(s). Published by Elsevier Inc. https://doi.org/10.1016/j.devcel.2024.09.008 ll Article Evolution of a large periplasmic disk in Campylobacterota flagella enables both efficient motility and autoagglutination Eli J. Cohen, 1,6,7, *Tina Drobni c, 1,6 Deborah A. Ribardo, 2 Aoba Yoshioka, 3 Trishant Umrekar, 1 Xuefei Guo, 1 Jose-Jesus Fernandez, 4 Emma E. Brock, 5 Laurence Wilson, 5 Daisuke Nakane, 3 David R. Hendrixson, 2 and Morgan Beeby 1 1 Department of Life Sciences, Imperial College London, London SW7 2AZ, UK 2 Department of Microbiology, University of Texas Southwestern Medical Center, Dallas, TX 75390, USA 3 Department of Engineering Science, Graduate School of Informatics and Engineering, The University of Electro-Communications, Tokyo, Japan 4 Spanish National Research Council (CINN-CSIC), Health Research Institute of Asturias (ISPA), Av Hospital Universitario s/n, Oviedo 33011, Spain 5 Department of Physics, School of Physics, Engineering and Technology, University of York, York YO10 5DD, UK 6 These authors contributed equally 7 Lead contact *Correspondence: [email protected] https://doi.org/10.1016/j.devcel.2024.09.008 SUMMARY The flagellar motors of Campylobacter jejuni (C. jejuni) and related Campylobacterota (previously epsilonproteobacteria) feature 100-nm-wide periplasmic ‘‘basal disks’’ that have been implicated in scaffolding a wider ring of additional motor proteins to increase torque, but the size of these disks is excessive for a role solely in scaffolding motor proteins. Here, we show that the basal disk is a flange that braces the flagellar motor during disentanglement of its flagellar filament from interactions with the cell body and other filaments. We show that motor output is unaffected when we shrink or displace the basal disk, and suppressor mutations of debilitated motors occur in flagellar-filament or cell-surface glycosylation pathways, thus sidestepping the need for a flange to overcome the interactions between two flagellar filaments and between flagellar filaments and the cell body. Our results identify unanticipated co-dependencies in the evolution of flagellar motor structure and cell-surface properties in the Campylobacterota. INTRODUCTION Many bacteria use flagella, membrane-embedded rotary motors connected to external helical propellers, to move through their environments. 1 All flagella, across genera separated by billions of years of evolution, share the same core proteins. A ring of motor proteins called stator complexes harness ion flux to rotate a tens-of-nanometers-wide cytoplasmic ring, or C-ring. The C-ring is connected to a chassis structure in the inner membrane called the MS-ring, which, in turn, forms the hub of an 30-nm-long axial driveshaft (the rod) that spans the periplasm, an 50-nmlong universal joint for torque redirection (the hook), and the multimicron-long flagellum itself. Research has focused on the model organisms Escherichia coli and Salmonella typhimurium, which have several flagella distributed around the cylindrical sides of their rod-shaped cells. 2 When all flagella rotate counterclockwise (CCW), they form a coherent bundle that propels the cell. Re-orientation of the cell’s swimming trajectory (i.e., chemotaxis) occurs when one or more flagellar motor transiently changes rotational direction, which causes the cell to randomly re-orient or ‘‘tumble’’; chemotaxis is achieved by inhibiting tumbling if the chemical environment is becoming more favorable to bias the otherwise random walk in beneficial directions. The low-complexity architecture and randomized placement of flagella in these organisms, however, is only one paradigm of flagellation. Many species, including those from Vibrio,Pseudomonas,Bdellovibrio,Helicobacter, and Campylobacter genera, deterministically assemble motors at one or both poles. Polarly localized flagellar motors are more structurally complex than lateral flagella, featuring large periplasmic disks of unclear significance. Polar motors preclude filament bundling as a swimming style. Swimming in bipolar-flagellated species involves wrapping the leading flagellum around the cell body to exert thrust in concert with the lagging flagellum. 3–6 In previous work, we showed that Campylobacter jejuni (C. jejuni), a member of the Campylobacterota (previously epsilonproteobacteria 7 ), reorients by unwrapping its wrapped leading filament from the cell surface by switching motor rotation; this allows the previously unwrapped (lagging) 3306 Developmental Cell 59, 3306–3321, December 16, 2024 ª2024 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). ll OPEN ACCESS filament to take its place wrapped around the cell body from the other pole. The flagella thus reverse their polarity and the cell swims away in approximately the opposite direction, whereas a non-chemotactic mutant (i.e., DcheY) never unwraps its leading flagellum and is unable to re-orient its swimming direction. We also found that a non-helical C. jejuni mutant was less efficient at unwrapping its leading filament, which we attributed to increased contact between the flagellar filament and cell surface, indicating that there is an affinity between the two. C. jejuni has one of the largest and most complex flagellar motors, featuring several periplasmic disks absent in Salmonella. 8 The largest is the 100-nm-wide basal disk, which assembles from thousands of copies of the lipoprotein FlgP. 9 The basal disk is required for assembly of other periplasmic disks that form a scaffold required for incorporation of a wider ring of stator complexes into the motor. This wider ring of additional stator complexes is consistent with the C. jejuni motor producing approximately three times the torque of the E. coli and Salmonella motors. 8 The basal disk, however, is much wider than the periplasmic scaffold and stator complex ring, and it is unclear what benefit C. jejuni gains from such an apparently excessively wide disk. We speculated that the basal disk in C. jejuni has functions beyond just being the assembly platform for the wider stator complex ring. We recently discovered that attractive forces between the glycosylated flagellar filament and glycosylated cell body must be overcome during unwrapping of the leading filament. Here, we hypothesized that the large diameter of the C. jejuni basal disk is an adaptation that allows the disk to act as a flange to stabilize the high-torque C. jejuni flagellar motor during unwrapping. We present structural, genetic, and microscopic experiments consistent with this model. RESULTS Reducing disk width does not affect motor rotation We sought to isolate the contribution of disk diameter from the requirement of the disk for a functional motor. Deletion of the genes responsible for assembly of the basal disk, flgPQ, prevents stator complex scaffolding and paralyzes the flagellum. To assess the significance of disk diameter, we made a mutant (EJC168) (Table 1) that constructs narrower basal disks but still enables assembly of the stator-complex periplasmic scaffold by deleting flgPQ from the chromosome and expressing them in trans from a titratable, synthetic C. jejuni tetRA promoter system at the astA locus. 10 In the absence of inducer anhydrotetracyline (ATc), cells were non-motile (Figure S1A) and did not express detectable levels of FlgP. Increasing the concentration of ATc from the lowest concentration tested (12.5 ng/mL) to the highest tested (100 ng/mL) resulted in a corresponding increase of flgP expression (Figure 1B). Electron cryotomography of motors from cells grown at different ATc concentrations confirmed that disk size correlated with induction level (Figures 1C and 1D). Disks were absent in the absence of ATc, as were stator complexes, equivalent to flgP or flgQ deletion. 8 Motors assembled at low flgPQ expression had narrower disks but were able to assemble the periplasmic scaffold and MotAB stator complexes. Increasing flgPQ expression produced correspondingly wider basal disks, with average diameters of 84, 98, and 107 nm for 25, 50, and 100 ng/mL ATc, respectively. We could discern periplasmic scaffold densities in disks as narrow as 50 nm, whereas the average wild-type (WT) motor is 105 nm but can be as wide as 130 nm. Although disk diameter correlated with the level of flgPQ induction, we found broad disk diameter distributions (Figure 1D). We therefore attempted to engineer a mutant that assembles consistently narrow basal disks while remaining motile in soft agar by alanine replacement of residues in FlgP thought to be involved in basal disk construction based on our recent in situ high-resolution structure of the C. jejuni flagellar motor. 9 We found that flgP S69A E157A K159A (flgP AAA ) assembled disks comparable in diameter to the smallest disks found in EJC168 at low induction (25 ng/mL ATc) with less size variation (Figure 1E). Based on western-blot analysis of FlgP, we conclude that FlgP levels in flgP AAA are reduced to a fraction of WT through an as-yet-unknown mechanism (Figures S1B–S1E). We found that basal disk diameters in flgP AAA cluster more tightly than in the ATc induction series and never extend beyond 80 nm in diameter (Figure 1D). We found that motor rotation was not compromised by having a smaller disk. Across all induction levels and in the flgP AAA background, both swimming velocity and filament rotation rate were comparable with WT cells (Figures 1F and S1F), demonstrating that motor rotation is independent of disk width beyond the minimal width required for stator complex incorporation. These data show that the WT disk diameter is excessive for a role in stator complex assembly alone, suggesting functions beyond rotation. Cells with small basal disks autoagglutinate faster than WT Despite small-disk motors rotating their flagella at WT speeds, we noticed that suspensions of cells grown on lower concentrations of ATc formed large clumps of cells bound together by their flagellar filaments within minutes of being applied to the sample chamber for observation, unlike suspensions of WT and high-induction cell suspensions (Video S1). This clumping, known as autoagglutination, is an adaptive behavior important for microcolony and biofilm formation during host colonization. 11,12 Autoagglutination requires a flagellar filament, and mutants with impaired or non-functional motors autoagglutinate faster than WT (Figures 2AandS2A–S2C), The increased autoagglutination of EJC168 at low induction and the flgP AAA mutant is at least partially the result of functional motors with small disks and not solely due to paralyzed motors in a proportion of the population. Discrete cells trapped in aggregates are often bound to the cell clusters by filaments with rotating motors (Video S2). Additionally, although low-induction EJC168 and flgP AAA cell suspensions have a proportion of entirely paralyzed, disk-less cells (12.5 ng ATc/mL: 9% w/disk; 25 ng ATc/mL: 35% w/disk; 50 ng ATc/mL: 75% w/disk; 100 ng ATc/mL: 88% w/disk), flgP AAA , with a higher percentage of cells with disks (44%), all of which are small, autoagglutinates faster than EJC168 induced at 25 ng/mL ATc, which has on average fewer, but larger, disks. This result suggests that wider basal disks may be needed by motors to counteract excessive autoagglutination (Figure 2B). ll OPEN ACCESS Article Developmental Cell 59, 3306–3321, December 16, 2024 3307 Table 1. List of strains used in this study Strain number Genotype Reference Notes EJC28 flaA S397C Cohen et al 5 wild type (WT) EJC168 DflgPQ DP astA ::cat-tetR_P tetA V2 DastA::flgPQ flaA S397C this study – EJC242 flgQ-6xHis flaA S397C this study – EJC243 flgP S69A E157A K159A flgQ-6xHis flaA S397C this study flgP AAA EJC244 flgP S69A E157A K159S flgQ-6xHis flaA S397C this study flgP AAS EJC112 flgP-lpp 34 flaA S397C this study – EJC113 flgP-lpp 41 flaA S397C this study – EJC114 flgP-lpp 48 flaA S397C this study – EJC115 flgP-lpp 55 flaA S397C this study – EJC137 flgP-lpp 55 this study – DRH2070 DflgP Hendrixson lab – EJC188 flgP D18–62 flaA S397C this study – –flgP-lpp 55 flgG T54N flaA S397C this study – –flgP-lpp 55 pseG I142T flaA S397C this study – –pseG I142T flaA S397C this study – EJC91 DflgPQ this study – EJC92 DflgQ this study – EJC93 DflgQ1.5 this study endpoint, motile isolate of first DflgQ evolved lineage EJC94 DflgQ2.5 this study endpoint, motile isolate of second DflgQ evolved lineage EJC95 DflgPQ1.5 this study endpoint, motile isolate of first DflgPQ evolved lineage EJC96 DflgPQ2.6 this study endpoint, motile isolate of second DflgPQ evolved lineage EJC97 DflgPQ D0661 flaA S397C this study – EJC98 DflgPQ DpglAB::aphA flaA S397C this study – EJC99 DflgPQ DkpsD flaA S397C this study – EJC100 DflgPQ D0661 DpglAB::aphA flaA S397C this study – EJC101 DflgPQ D0661 DkpsD flaA S397C this study – EJC102 DflgPQ DpglAB::aphA DkpsD flaA S397C this study – EJC103 DflgPQ D0661 DpglAB::aphA DkpsD flaA S397C this study – EJC104 DflgPQ D0661 DpglAB::aphA DkpsD DflaAB::flaA S397C this study all-FlaA filament EJC105 DflgPQ D0661 DpglAB::aphA DkpsD DflaAB::flaB S397C this study all-FlaB filament EJC106 DflgPQ fliG G305S this study – EJC107 DflgPQ DflaAB::flaB S397C/T477I this study all-FlaB T477I filament EJC108 DflgPQ DflaAB::flaB S397C/T477I fliG G305S this study – EJC109 D0661 flaA S397C this study – EJC255 DwaaF flaA S397C this study – EJC256 DwaaF flgP-lpp 55 flaA S397C this study – EJC257 DkpsM::cat flaA S397C this study – EJC258 DkpsM::cat flgP-lpp 55 flaA S397C this study – EJC259 DpglAB::aphA flgP-lpp 55 flaA S397C this study – EJC261 D0661 flgP-lpp 55 flaA S397C this study – EJC263 DkpsD flgP-lpp 55 flaA S397C this study – EJC266 D0661 DpglAB::aphA DkpsD flgP-lpp 55 flaA S397C this study – ll OPEN ACCESS Article 3308 Developmental Cell 59, 3306–3321, December 16, 2024 100 ng ATc flgPAAA 12.5 ng/mL ATc 25 ng/mL ATc 50 ng/mL ATc 100 ng/mL ATc A -ATc 12.5 ng/mL ATc 25 ng/mL ATc WT FlgP 50 ng/mL ATc 100 ng/mL ATc B C ED Stator scaffolding pflABC, E-ring Stator motAB Inner membrane Rotor (C-ring) fliGMN Basal disk flgP Outer membrane HookRod PL-ring flgHI Width (nm) WT 12.5 ng/mL ATc 25 ng/mL ATc 50 ng/mL ATc 100 ng/mL ATc T c c c T c flgPAAA FlgR WT flgPAAA 0 ms 2.5 ms 5 ms 7.5 ms 10 ms 12.5 ms 15 ms 17.5 ms 20 ms 22.5 ms F 12.5 ng/mL ATc 25 ng/mL ATc 50 ng/mL ATc 100 ng/mL ATc flgPAAA Figure 1. Engineered motors with small basal disks nevertheless incorporate stator complexes and rotate similarly to WT motors (A) The flagellar motor of Campylobacter jejuni has the same components as that of the model organisms, as well as extra embellishments such as the basal disk and stator scaffolding architecture. (legend continued on next page) ll OPEN ACCESS Article Developmental Cell 59, 3306–3321, December 16, 2024 3309 The basal disk must be in register with the P-ring for effective motility That motors with small disks may cause excessive autoagglutination suggested to us that the disk might act as a mechanically reinforcing flange to stabilize the high-torque C. jejuni motor to overcome immediate autoagglutination when near other cells. The basal disk polymerizes around the P-ring, which is believed to act with the L-ring as a non-rotating bushing to brace the rotating flagellar driveshaft. We reasoned that we could test this model by shifting the basal disk out of register with the P-ring, disrupting mechanical support of the P-ring by the basal disk, yet preserving the disk’s role in stator scaffolding. We engineered FlgP mutants to push the disk out of register with the P-ring. FlgP is a small protein with an N-terminal cysteine (C17), presumed to be lipoylated and inserted in the inner leaflet of the outer membrane (OM). A poorly conserved N-terminal 40 residues of the mature protein is likely to be a linker anchored to the OM by C17. This is supported by an 6nm unresolved gap between the OM and basal disk in the subtomogram average structure of the C. jejuni motor, a distance consistent with an 40-residue linker. To displace the disk, we inserted heptad repeats of varying length from the Salmonella lipoprotein LppA (also known as Braun’s lipoprotein) downstream of C17 (Figure 3A), followed by the native FlgP OM-linker sequence, based on previous success using Lpp to alter spacing in the periplasm. 13–15 These mutants were motile in soft agar, but motility decreased as more heptads were added (Figures S3A– S3C). We chose the mutant that formed the smallest swarm in soft agar but was nevertheless still motile, flgP-lpp 55 , for further analysis. Subtomogram averaging confirmed that the flgP-lpp 55 motor had a basal disk that had been shifted out of register with the P-ring by 7–8 nm, consistent with the 8.4 nm expected by insertion of a 55-residue ahelix (Figure 3B). In this mutant, the disk encircles the proximal rod instead of the P-ring. Curiously, the disk appears to self-assemble non-specifically around whichever axial component it is in register with, as the inner radius of the flgP-lpp 55 disk (Figure 3B red dashed line) was narrower than the WT disk, matching the decreased width of the structure around which it was assembling (i.e., the rod rather than the wider P-ring). Although flgP-lpp 55 forms small-diameter swarms in soft agar (Figure 3C), the stator scaffolding architecture in the flgP-lpp 55 motor was indistinguishable from the WT motor, and three-dimensional (3D) holographic tracking microscopy of the flgP-lpp 55 mutant swimming in viscous media revealed a subset of the population swimming at WT velocity (white arrowheads in Figure 3D), indicating that motors with basal disks displaced from around the P-ring could still function as well as WT. Together, these results demonstrate that the poor motility of the flgP-lpp 55 mutant in soft agar is not due to impairment of motor rotation. To determine how displacement of the basal disk reduced motility in soft agar despite the presence of motile cells in liquid media, we labeled the flagellar filament with fluorescent dye and recorded swimming by high-speed video fluorescence microscopy. As with our small-disk mutants, and consistent with our holographic tracking, flagellar rotation rate in the flgP-lpp 55 mutant was comparable with WT, but the mutant was incapable of unwrapping the leading flagellum from the cell body during motor reversals (Video S3, top). Consequently, the flgP-lpp 55 mutant exhibited a stuttering motility characterized by short runs interrupted by brief pauses with no net change in swimming trajectory (Figure 3E), as well as doubly wrapped cells (Video S4), phenotypes that were reminiscent of the defective-unwrapping phenotype we observed by high-speed fluorescence video in non-helical mutants of C. jejuni in one of our previous studies. 5 This is in contrast to WT cells, which have a swimming style referred to as darting motility, whereby chemotaxing cells change swimming trajectory during motor reversals due to the wrapping and unwrapping of the opposed flagellar filaments (Video S3, bottom), and where we have never seen doubly wrapped cells. Thus, the phenotype of the flgP-lpp 55 mutant in soft agar reflects the inability of the filament to be pulled from the cell body during motor switching across the entire population. We used low-magnification darkfield microscopy and C. jejuni’s tendency to migrate toward regions of higher oxygen (a.k.a. aerotaxis) to visualize how the failure to unwrap in the flgP-lpp 55 background manifested at the population level, as in our previous work. 5 This showed that the failure of flgP-lpp 55 to unwrap upon motor reversal results in population-level failure to swarm toward regions of higher oxygen content, unlike WT populations (Video S5). To understand how this impairment would affect motility in environments like those of the mucous-filled gastric crypts of a host’s digestive tract, we observed fluorescently labeled cells swimming through viscous media in a microfluidic device with confined 1 mm channels, allowing only one cell through at a time. Although individual flgP-lpp 55 cells were able to traverse the device at the same speed as WT cells (Figures 3F and 3G), further confirming our findings from subtomogram averaging and 3D holographic cell tracking that the function of the (B) Increasing the level of ATc in the growth medium corresponded to an increased level of flgP expression. FlgR antisera was used as an internal loading control. (C and D) (C) Slices of individual tomograms (top) and subtomogram averages (bottom) (scale bars, 20 nm) of motors at different flgP expression levels and (dashed white lines indicate width measurement, stators [MotB] are incorporated into the motor even under low-induction conditions [red arrowheads]). (D) Measurements of basal disk widths of individual motors in electron cryo-tomograms (disk widths are significantly different between WT, flgP AAA , and EJC168 at 25 and 50 ng/mL ATc, one-way ANOVA and two-tailed t tests, p%0.0001–0.0385, bars represent mean ± SEM). In the WT, all the motors imaged possessed basal disks, whereas in EJC168, lower concentrations of ATc corresponded to a lower proportion of motors with disks. Disk-less motors were excluded from the analysis in (D). (E) The flgP S69A E157A K159A (flgP AAA ) mutant constructed small-diameter basal disks, and the distribution of basal-disk diameters clustered more tightly than EJC168 at all induction levels and never extended beyond 80 nm (left: slice through a single tomogram; right: subtomogram average of FlgP AAA motor [scale bars, 20 nm]). (F) Still frames of high-speed video of fluorescently labeled WT and flgP AAA cells (scale bars, 1 mm). The position of filaments relative to the cell body (white arrowheads) can be used to approximate rotation rate. For both the WT and flgP AAA ,rotation rate was found to be 100 Hz (1 revolution/10 ms). See also Videos S1 and S2. ll OPEN ACCESS Article 3310 Developmental Cell 59, 3306–3321, December 16, 2024 torque-generating motor core is unaffected by displacing the disk, the displaced-disk flgP-lpp 55 mutant was unable to reverse direction when encountering obstacles (i.e., immobilized cells) in the channel, leading to C. jejuni pileups (Video S6). In contrast, WT cells reversed direction upon encountering an obstacle and exited the device (Figure S3D; Video S7). We next inoculated day-old chicks with identical numbers of WT, DflgP or flgP-lpp 55 cells and enumerated the number of colony-forming units (CFUs) in the ceca of the chicks at 1 week post inoculation. As predicted by our microfluidic device results, we found that the flgP-lpp 55 mutant colonized the chicken cecum as poorly as the non-motile DflgP control strain (Figure 3H). These results show that, in addition to the basal disk’s role in stator scaffolding, flanging of the motor by the basal disk is an important adaptive trait for C. jejuni in its native environments. We also produced a mutant in which the disk was moved closer to the OM, and out of register with the P-ring, by deletion of the OM-linker while retaining C17 (flgP D18–62 ). In soft agar, flgP D18–62 produced a swarm 30% that of WT (Figure S3E). In contrast to flgP-lpp 55 , however, the subtomogram average of the flgP D18–62 motor had poorly resolved stator scaffolds, suggesting that the motility defect in this background is likely due to disruption of motor assembly and consistent with FlgP being too far from the inner membrane to template formation of the periplasmic scaffold (Figure S3F). We conclude that a wide disk is required for overcoming filament-filament (self-other) interactions during autoagglutination, while a correctly positioned disk is required to act as a flange to overcome filament-cell (self-self) interactions during unwrapping. Mutations that restore motility to a displaced-disk mutant suggest a link between filament unwrapping and filament glycosylation Suppressing mutations in the flgP-lpp 55 background that restored near-WT levels of motility arise following 36–48 h of incubation in soft agar, appearing as flares emanating from the original, poor-motility swarm. Reasoning that the identity of the suppressor mutations would provide further information regarding the function of the basal disk, we isolated two independent revertants, performed whole-genome sequencing, and found two paths to suppress the flgP-lpp 55 motility defect: restoration of the P-ring/basal disk register (i.e., restoration of flanging), or decreased O-glycosylation of the flagellar filament (Figure S4A). The first motile revertant of flgP-lpp 55 acquired a point mutation in the distal rod gene, flgG T54N . In WT flagellar motors, the distal rod only grows long enough for a single P-ring and a single L-ring before it contacts the OM and stops polymerizing, but alleles in flgG have been isolated in Salmonella that allow the distal rod to continue polymerizing once it reaches the OM. 16 These so-called flgG* alleles often arise in an N-terminal region of FlgG known as the Dc domain, encompassing residues 30– 70 of the protein. 17–19 Similarly, our flgG T54N allele allows the distal rod to grow longer than usual to accommodate two P-rings around the distal rod (Figure 4A); thus, the additional space imposed by the Lpp 55 insertion is compensated for by an additional P-ring on flgG T54N , thus restoring register of the first P-ring to the basal disk while the second P-ring templates assembly of the L-ring for correct OM-penetration and hook/filament assembly. In addition to suppressing the motility defect of the flgP-lpp 55 , the flgG T54N allele also suppressed the moderate autoagglutination defect of the flgP-lpp 55 mutant (Figure 4D), indicating that flanging of the P-ring by the basal disk is also important for preventing excessive autoagglutination. The second revertant was intriguing because the suppressing mutation, pseG I142T , affects the flagellar filament rather than the motor. PseG is a uridine-diphosphate-sugar (UDP-sugar) hydrolase involved in the synthesis of pseudaminic acid (PseAc), 20,21 the O-linked sugar that decorates the flagellar filament at 19 serine and threonine residues of each flagellin monomer in C. jejuni 81– 176. 22–24 Glycosylation of the flagellar filament in C. jejuni is required for both filament assembly as well as autoagglutination, but only three of the 19 glycosylable flagellin residues are critical for filament assembly and motility, and a pseG knockout is nonmotile. 25,26 We therefore reasoned that substitution of a non-polar isoleucine residue adjacent to the substrate-binding site of PseG with a threonine, decreases, but must not abolish, its enzymatic Figure 2. Low FlgP expression and mutation of FlgP enhance autoagglutination (A) A flagellar filament is required for autoagglutination, and a functional motor prevents excessive autoagglutination. The sedimentation rates (i.e., autoagglutination) of a suspension of WT cells paralyzed with CCCP or a paralyzed mutant (DmotB::cat) are significantly faster than the non-treated cell suspension (two-tailed t test, p= 0.0062). (B) Low FlgP expression in EJC168 corresponded to a faster sedimentation rate. The flgP AAA mutant sedimented significantly faster than EJC168 induced with 25 ng/mL ATc (two-way ANOVA, p%0.0001, error bars are mean ± SEM). See also Videos S1 and S2. ll OPEN ACCESS Article Developmental Cell 59, 3306–3321, December 16, 2024 3311 AB CD E FGH Figure 3. Displacement of the basal disk from the P-ring reduces filament unwrapping (A) A 55-residue segment from Salmonella LppA (also known as Lpp or Braun’s lipoprotein) was inserted after C17 in FlgP to make FlgP-Lpp 55 . (B) Subtomogram averaging of the flgP-lpp 55 mutant’s motor revealed that the basal disk had been displaced from the P-ring by 7 nm (P-ring: white dashed line, first ring of basal disk: red dashed line) and had a narrower first ring of FlgP subunits but that stator recruitment was not impacted (red circles) (scale bars, 20 nm). (legend continued on next page) ll OPEN ACCESS Article 3312 Developmental Cell 59, 3306–3321, December 16, 2024 activity and results in filaments with reduced or altered O-glycosylation. This interpretation is supported by the observation that while the pseG I142 allele in an otherwise WT background has no effect on swarm diameter in soft agar or flagellar filament length (Figures 4B, 4C, and S5A), the flagellin in a pseG I142T background migrates differently by isoelectric-focusing gel electrophoresis, indicating it has altered glycosylation (Figures S5B and S5C). Furthermore, the pseG I142T allele significantly reduces autoagglutination in both the WT and flgP-lpp 55 backgrounds (Figure 4D) (two-way ANOVA, p%0.0001, error bars represent SEM), in agreement with previous work showing that eight of 19 glycosylable flagellin residues are important for autoagglutination. 22 When we observed the flgP-lpp 55 pseG I142T double mutant by high-speed video fluorescence microscopy, we found that the double mutant unwrapped its filament from the cell body during directional reversals at rates comparable with WT, in contrast to flgP-lpp 55 (Figure 4E). Furthermore, in both aerotaxis assays and microfluidic experiments the flgP-lpp 55 pseG I142 mutant exhibited near-WT behavior (Figure 4F). Consistent with our previous study showing interaction between the flagellar filament and cell body, 5 these data implicate O-glycosylation of the flagellar filament in filament-cell body interactions in addition to filament-filament autoagglutination and filament assembly. We generated knockouts of the capsular polysaccharide (CPS) genes kpsM and kpsD, 27 as well as a gene involved in synthesis of lipooligosaccharide (LOS), waaF, 28 to try to identify what the filament interacts with on the cell surface (Figures S5D and S5E). Whereas deletion of CPS reduced motility in both the WT and flgP-lpp 55 background in soft agar, deletion of waaF had a small but significant suppressing effect on the motility of flgPlpp 55 in soft agar (p= 0.016, two-tailed t test). The waaF knockout also had an unusual phenotype in soft agar in which the main swarm is surrounded by a larger, ghostly swarm, suggesting that loss of LOS may potentiate suppression of the flgP-lpp 55 mutant. We speculate that the filament-cell-surface interaction is a complex interaction between multiple glycan moieties on the cell surface, but this will require further investigation. Swimming without a basal disk requires deglycosylation of the cell surface Although deletion of flgPQ produces a non-motile phenotype in motility agar, observing this mutant by fluorescence microscopy revealed occasional cells with rotating flagella. Thus, even in the absence of the basal disk, stator complexes can still be inefficiently recruited to the motor. Given that we were able to isolate suppressors of disk displacement, we speculated that we might be able to isolate a suppressor strain of a wholesale flgPQ deletion that would shed more light on the role of the basal disk. We selected for suppression of the DflgPQ motility defect through prolonged incubation in motility agar, as previous attempts over smaller time frames (48–72 h) had been unsuccessful in isolating motile revertants. We independently inoculated two colonies each of DflgPQ and DflgQ mutants in motility agar and incubated the plates for 4–6 days. Although each colony had a non-motile phenotype after 2 days of incubation, all four isolates had speckles emanating from the site of inoculation within 5 days. The speckled phenotype occurs when the majority of cells in motility agar are non-motile, with occasional cells possessing a functional flagellum. Upon division, these motile cells deposit non-motile daughter cells that seed colonies of non-motile descendants. 29 As this process occurs around the site of inoculation, the swarm takes on a speckled, or ‘‘bushy,’’ phenotype. After 4–6 days of incubation, the edge of each bushy swarm was picked from the agar, single-colony purified, stored at 80C, and also used to inoculate a fresh motility plate. This process was repeated four to five times for each lineage, at which point all lineages had evolved a non-bushy swarm phenotype in soft agar (Figure 5A), indicating that the majority of cells in the population are swimming. We then performed wholegenome sequencing of each endpoint isolate for each of the four lineages to determine the mutations required for motility in the absence of the basal disk. We predicted that disk-less motility would require mutations in flagellar genes, specifically periplasmic scaffold genes and/or the stator complex genes motAB, as such mutations might enable stable incorporation of stator complexes despite the lack of FlgP. To our surprise, however, the theme across all four evolved lineages was a similar constellation of spontaneous mutations in genes involved, or implicated, in decorating the cell surface with polysaccharides (Table 2). Each lineage had mutations in the pgl operon, responsible for N-glycosylation of a diverse cohort of periplasmic and surface-exposed proteins. 30–33 A further two lineages had spontaneously acquired mutations in kps genes, which are responsible for CPS biogenesis. 27 Additionally, mutations in a gene predicted to function as a polysaccharide deacetylase, cjj_81-176_0661 (hereafter referred to as 0661) were present in all four evolved lineages. Each lineage also had evidence of phase variation in kps-associated sugar transferase and CPS-modification genes. 34,35 In addition to mutations in genes involved or implicated in decoration of the cell surface with sugars, two of the isolates had single-nucleotide polymorphisms (SNPs) in the gene encoding the cytoplasmic rotor, fliG, 36 and one isolate had a second flagellar mutation at the fla locus: DflaA::flaB T477I , in which the WT flagellin locus encoding both the major flagellin flaA and (C and D) (C) Following overnight incubation in soft agar, the flgP-lpp 55 mutant formed small-diameter swarms in motility agar compared with WT, despite (D) the presence of a population of cells swimming at WT velocity when cell suspensions were observed by 3D-holographic-tracking microscopy (arrowheads). (E) Kymographs generated from high-speed fluorescence video show that the flgP-lpp 55 mutant is incapable of unwrapping its filament from the cell body during motor reversals and does not change swimming direction, as opposed to the characteristic darting motility of WT C. jejuni where unwrapping leads to a reversal of swimming direction. Single asterisk: unwrapping of leading filament and wrapping of lagging filament; double asterisk: failure to unwrap (scale bars,1mm). (F and G) The velocity of individual flgP-lpp 55 cells traversing a microfluidic device with 1-mm-wide channels was found to be identical to WT (two-tailed t test, p= 0.212, bars represent mean ± SEM), confirming our observations from holographic tracking, although flgP-lpp 55 cells fail to reverse upon encountering obstacles. (H) The flgP-lpp 55 mutant has a significant host colonization defect relative to WT, colonizing the chicken cecum as poorly as a non-motile DflgP mutant (one-way ANOVA, p%0.0001, bars represent mean CFU/g cecal content). See also Videos S3,S4,S5,S6, and S7. ll OPEN ACCESS Article Developmental Cell 59, 3306–3321, December 16, 2024 3313 assemble wider stator rings using a conserved protein scaffold. Proc. Natl. Acad. Sci. USA 113, E1917–E1926. 9. Drobni c, T., Cohen, E.J., Alzheimer, M., Froschauer, K., Svensson, S., Singh, N., Garg, S.G., Henderson, L., Umrekar, T., Nans, A., et al. (2023). Molecular model of a bacterial flagellar motor in situ reveals a ‘‘parts-list’’ of protein adaptations to increase torque. Preprint at bioRxiv. https://doi.org/10.1101/2023.09.08.556779. 10. Cohen, E.J., Quek, R.T., and Beeby, M. (2019). A tetRA-based promoter system for the generation of conditional knockouts in Campylobacter jejuni. Preprint at bioRxiv. https://doi.org/10.1101/616649. 11. Misawa, N., and Blaser, M.J. (2000). Detection and characterization of autoagglutination activity by Campylobacter jejuni. Infect. Immun. 68, 6168–6175. 12. Trunk, T., Khalil, H.S., and Leo, J.C. (2018). Bacterial autoaggregation. AIMS Microbiol. 4, 140–164. 13. Cohen, E.J., Ferreira, J.L., Ladinsky, M.S., Beeby, M., and Hughes, K.T. (2017). Nanoscale-length control of the flagellar driveshaft requires hitting the tethered outer membrane. Science 356, 197–200. 14. Asmar, A.T., Ferreira, J.L., Cohen, E.J., Cho, S.H., Beeby, M., Hughes, K.T., and Collet, J.F. (2017). Communication across the bacterial cell envelope depends on the size of the periplasm. PLoS Biol. 15, e2004303. 15. Mathelie ´-Guinlet, M., Asmar, A.T., Collet, J.-F., and Dufre ˆne, Y.F. (2020). Lipoprotein Lpp regulates the mechanical properties of the E. coli cell envelope. Nat. Commun. 11, 1789. 16. Chevance, F.F.V., Takahashi, N., Karlinsey, J.E., Gnerer, J., Hirano, T., Samudrala, R., Aizawa, S.I., and Hughes, K.T. (2007). The mechanism of outer membrane penetration by the eubacterial flagellum and implications for spirochete evolution. Genes Dev. 21, 2326–2335. https://doi.org/10. 1101/gad.1571607. 17. Moriya, N., Minamino, T., Ferris, H.U., Morimoto, Y.V., Ashihara, M., Kato, T., and Namba, K. (2013). Role of the Dc domain of the bacterial hook protein FlgE in hook assembly and function. Biophysics 9, 63–72. 18. Matsunami, H., Barker, C.S., Yoon, Y.-H., Wolf, M., and Samatey, F.A. (2016). Complete structure of the bacterial flagellar hook reveals extensive set of stabilizing interactions. Nat. Commun. 7, 13425. 19. Horva ´th, P., Kato, T., Miyata, T., and Namba, K. (2019). Structure of salmonella flagellar hook reveals intermolecular domain interactions for the universal joint function. Biomolecules 9, 462. 20. Liu, F., and Tanner, M.E. (2006). PseG of pseudaminic acid biosynthesis: a UDP-sugar hydrolase as a masked glycosyltransferase. J. Biol. Chem. 281, 20902–20909. 21. Rangarajan, E.S., Proteau, A., Cui, Q., Logan, S.M., Potetinova, Z., Whitfield, D., Purisima, E.O., Cygler, M., Matte, A., Sulea, T., et al. (2009). Structural and Functional Analysis of Campylobacter jejuni PseG: a UDP-sugar hydrolase from the Pseudaminic Acid Biosynthetic pathway. J. Biol. Chem. 284, 20989–21000. 22. Ewing, C.P., Andreishcheva, E., and Guerry, P. (2009). Functional characterization of flagellin glycosylation in Campylobacter jejuni 81–176. J. Bacteriol. 191, 7086–7093. 23. Doig, P., Kinsella, N., Guerry, P., and Trust, T.J. (1996). Characterization of a post-translational modification of Campylobacter flagellin: identification of a sero-specific glycosyl moiety. Mol. Microbiol. 19, 379–387. 24. Thibault, P., Logan, S.M., Kelly, J.F., Brisson, J.R., Ewing, C.P., Trust, T.J., and Guerry, P. (2001). Identification of the carbohydrate moieties and glycosylation motifs in Campylobacter jejuni flagellin. J. Biol. Chem. 276, 34862–34870. 25. McNally, D.J., Hui, J.P.M., Aubry, A.J., Mui, K.K.K., Guerry, P., Brisson, J.R., Logan, S.M., and Soo, E.C. (2006). Functional characterization of the flagellar glycosylation locus in Campylobacter jejuni 81–176 using a focused metabolomics approach. J. Biol. Chem. 281, 18489–18498. 26. Guerry, P., Ewing, C.P., Schirm, M., Lorenzo, M., Kelly, J., Pattarini, D., Majam, G., Thibault, P., and Logan, S. (2006). Changes in flagellin glycosylation affect Campylobacter autoagglutination and virulence. Mol. Microbiol. 60, 299–311. 27. Karlyshev, A.V., Linton, D., Gregson, N.A., Lastovica, A.J., and Wren, B.W. (2000). Genetic and biochemical evidence of a Campylobacter jejuni capsular polysaccharide that accounts for Penner serotype specificity. Mol. Microbiol. 35, 529–541. 28. Oldfield, N.J., Moran, A.P., Millar, L.A., Prendergast, M.M., and Ketley, J.M. (2002). Characterization of the Campylobacter jejuni heptosyltransferase II gene, waaF, provides genetic evidence that extracellular polysaccharide is lipid A core independent. J. Bacteriol. 184, 2100–2107. 29. Cohen, E.J., and Hughes, K.T. (2014). Rod-to-hook transition for extracellular flagellum assembly is catalyzed by the L-ring-dependent rod scaffold removal. J. Bacteriol. 196, 2387–2395. 30. Szymanski, C.M., Yao, R., Ewing, C.P., Trust, T.J., and Guerry, P. (1999). Evidence for a system of general protein glycosylation in Campylobacter jejuni. Mol. Microbiol. 32, 1022–1030. 31. Kelly, J., Jarrell, H., Millar, L., Tessier, L., Fiori, L.M., Lau, P.C., Allan, B., and Szymanski, C.M. (2006). Biosynthesis of the N-linked glycan in Campylobacter jejuni and addition onto protein through block transfer. J. Bacteriol. 188, 2427–2434. 32. Cain, J.A., Dale, A.L., Sumer-Bayraktar, Z., Solis, N., and Cordwell, S.J. (2020). Identifying the targets and functions of N-linked protein glycosylation in Campylobacter jejuni. Mol. Omics 16, 287–304. 33. Scott, N.E., Parker, B.L., Connolly, A.M., Paulech, J., Edwards, A.V.G., Crossett, B., Falconer, L., Kolarich, D., Djordjevic, S.P., Højrup, P., et al. (2011). Simultaneous glycan-peptide characterization using hydrophilic interaction chromatography and parallel fragmentation by CID, higher energy collisional dissociation, and electron transfer dissociation MS applied to the N-linked glycoproteome of Campylobacter jejuni. Mol. Cell. Proteomics 10. M000031-MMCP201. 34. Bacon, D.J., Szymanski, C.M., Burr, D.H., Silver, R.P., Alm, R.A., and Guerry, P. (2001). A phase-variable capsule is involved in virulence of Campylobacter jejuni 81–176. Mol. Microbiol. 40, 769–777. 35. van Alphen, L.B., Wenzel, C.Q., Richards, M.R., Fodor, C., Ashmus, R.A., Stahl, M., Karlyshev, A.V., Wren, B.W., Stintzi, A., Miller, W.G., et al. (2014). Biological roles of the O-methyl phosphoramidate capsule modification in Campylobacter jejuni. PLoS One 9, e87051. 36. Paul, K., Gonzalez-Bonet, G., Bilwes, A.M., Crane, B.R., and Blair, D. (2011). Architecture of the flagellar rotor. EMBO J. 30, 2962–2971. 37. Johnson, J.G., Livny, J., and DiRita, V.J. (2014). High-throughput sequencing of Campylobacter jejuni insertion mutant libraries reveals mapA as a fitness factor for chicken colonization. J. Bacteriol. 196, 1958–1967. 38. Iwata, T., Watanabe, A., Kusumoto, M., and Akiba, M. (2016). Peptidoglycan acetylation of Campylobacter jejuni is essential for maintaining cell wall integrity and colonization in chicken intestines. Appl. Environ. Microbiol. 82, 6284–6290. 39. Roujeinikova, A. (2008). Crystal structure of the cell wall anchor domain of MotB, a stator component of the bacterial flagellar motor: implications for peptidoglycan recognition. Proc. Natl. Acad. Sci. USA 105, 10348–10353. 40. Szymanski, C.M., Logan, S.M., Linton, D., and Wren, B.W. (2003). Campylobacter – a tale of two protein glycosylation systems. Trends Microbiol. 11, 233–238. 41. Karlyshev, A.V., Ketley, J.M., and Wren, B.W. (2005). The Campylobacter jejuni glycome. FEMS Microbiol. Rev. 29, 377–390. 42. Cain, J.A., Dale, A.L., Niewold, P., Klare, W.P., Man, L., White, M.Y., Scott, N.E., and Cordwell, S.J. (2019). Proteomics reveals multiple phenotypes associated with N-linked glycosylation in Campylobacter jejuni. Mol. Cell. Proteomics 18, 715–734. 43. Duma, J., Nothaft, H., Weaver, D., Fodor, C., Beadle, B., Linton, D., Benoit, S.L., Scott, N.E., Maier, R.J., and Szymanski, C.M. (2020). Influence of protein glycosylation on Campylobacter fetus physiology. Front. Microbiol. 11, 1191. 44. Alemka, A., Nothaft, H., Zheng, J., and Szymanski, C.M. (2013). N-glycosylation of Campylobacter jejuni Surface proteins promotes bacterial fitness. Infect. Immun. 81, 1674–1682. ll OPEN ACCESS Article 3320 Developmental Cell 59, 3306–3321, December 16, 2024 45. Abouelhadid, S., Raynes, J., Bui, T., Cuccui, J., and Wren, B.W. (2020). Characterization of posttranslationally modified multidrug efflux pumps reveals an unexpected link between glycosylation and antimicrobial resistance. mBio 11, e02604–e02620. 46. Abouelhadid, S., North, S.J., Hitchen, P., Vohra, P., Chintoan-Uta, C., Stevens, M., Dell, A., Cuccui, J., and Wren, B.W. (2019). Quantitative analyses reveal novel roles for N-glycosylation in a major enteric bacterial pathogen. mBio 10, e00297-19. 47. Maue, A.C., Mohawk, K.L., Giles, D.K., Poly, F., Ewing, C.P., Jiao, Y., Lee, G., Ma, Z., Monteiro, M.A., Hill, C.L., et al. (2013). The polysaccharide capsule of Campylobacter jejuni modulates the host immune response. Infect. Immun. 81, 665–672. 48. Coward, C., Grant, A.J., Swift, C., Philp, J., Towler, R., Heydarian, M., Frost, J.A., and Maskell, D.J. (2006). Phase-variable surface structures are required for infection of Campylobacter jejuni by bacteriophages. Appl. Environ. Microbiol. 72, 4638–4647. 49. Guerry, P. (2007). Campylobacter flagella: not just for motility. Trends Microbiol. 15, 456–461. 50. Merino, S., and Toma ´s, J.M. (2014). Gram-negative flagella glycosylation. Int. J. Mol. Sci. 15, 2840–2857. 51. Tram, G., Day, C.J., and Korolik, V. (2020). Bridging the gap: A role for Campylobacterjejuni biofilms. Microorganisms 8, 452. 52. Howard, S.L., Jagannathan, A., Soo, E.C., Hui, J.P.M., Aubry, A.J., Ahmed, I., Karlyshev, A., Kelly, J.F., Jones, M.A., Stevens, M.P., et al. (2009). Campylobacter jejuni glycosylation island important in cell charge, legionaminic acid biosynthesis, and colonization of chickens. Infect. Immun. 77, 2544–2556. 53. K€ uhn, M.J., Schmidt, F.K., Eckhardt, B., and Thormann, K.M. (2017). Bacteria exploit a polymorphic instability of the flagellar filament to escape from traps. Proc. Natl. Acad. Sci. USA 114, 6340–6345. 54. Kinosita, Y., Kikuchi, Y., Mikami, N., Nakane, D., and Nishizaka, T. (2018). Unforeseen swimming and gliding mode of an insect gut symbiont, Burkholderia sp. RPE64, with wrapping of the flagella around its cell body. ISME J. 12, 838–848. 55. Hendrixson, D.R., Akerley, B.J., and DiRita, V.J. (2001). Transposon mutagenesis of Campylobacter jejuni identifies a bipartite energy taxis system required for motility. Mol. Microbiol. 40, 214–224. 56. Beauchamp, J.M., Leveque, R.M., Dawid, S., and DiRita, V.J. (2017). Methylation-dependent DNA discrimination in natural transformation of Campylobacter jejuni. Proc. Natl. Acad. Sci. USA 114, E8053–E8061. 57. Wilson, L., and Zhang, R. (2012). 3D Localization of weak scatterers in digital holographic microscopy using Rayleigh-Sommerfeld back-propagation. Opt. Express 20, 16735–16744. 58. K€ uhn, M.J., Schmidt, F.K., Farthing, N.E., Rossmann, F.M., Helm, B., Wilson, L.G., Eckhardt, B., and Thormann, K.M. (2018). Spatial arrangement of several flagellins within bacterial flagella improves motility in different environments. Nat. Commun. 9, 5369. 59. Thornton, K.L., Butler, J.K., Davis, S.J., Baxter, B.K., and Wilson, L.G. (2020). Haloarchaea swim slowly for optimal chemotactic efficiency in low nutrient environments. Nat. Commun. 11, 4453. 60. Weibel, D.B., DiLuzio, W.R., and Whitesides, G.M. (2007). Microfabrication meets microbiology. Nat. Rev. Microbiol. 5, 209–218. 61. Agulleiro, J.I., and Fernandez, J.J. (2011). Fast tomographic reconstruction on multicore computers. Bioinform. Oxf. Engl. 27, 582–583. 62. Agulleiro, J.-I., and Fernandez, J.-J. (2015). Tomo3D 2.0—exploitation of advanced vector extensions (AVX) for 3D reconstruction. J. Struct. Biol. 189, 147–152. 63. Tegunov, D., and Cramer, P. (2019). Real-time cryo-electron microscopy data preprocessing with Warp. Nat. Methods 16, 1146–1152. 64. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682. 65. Suloway, C., Pulokas, J., Fellmann, D., Cheng, A., Guerra, F., Quispe, J., Stagg, S., Potter, C.S., and Carragher, B. (2005). Automated molecular microscopy: the new Leginon system. J. Struct. Biol. 151, 41–60. 66. Cheng, A., Negro, C., Bruhn, J.F., Rice, W.J., Dallakyan, S., Eng, E.T., Waterman, D.G., Potter, C.S., and Carragher, B. (2021). Leginon: new features and applications. Protein Sci. 30, 136–150. 67. Castan ˜o-Dı ´ez, D., Kudryashev, M., Arheit, M., and Stahlberg, H. (2012). Dynamo: A flexible, user-friendly development tool for subtomogram averaging of cryo-EM data in high-performance computing environments. J. Struct. Biol. 178, 139–151. 68. Kremer, J.R., Mastronarde, D.N., and McIntosh, J.R. (1996). Computer visualization of three-dimensional image data using IMOD. J. Struct. Biol. 116, 71–76. ll OPEN ACCESS Article Developmental Cell 59, 3306–3321, December 16, 2024 3321 STAR+METHODS KEY RESOURCES TABLE EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Infection of avian hosts Specific pathogen-free fertilized eggs from white leghorn chickens were procured from Charles River SPAFAS. White leghorn chickens are an outbred strain. REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Rabbit polyclonal anti-FlgP antisera Laboratory of David R. Hendrixson N/A Rabbit polyclonal anti-FlgR Laboratory of David R. Hendrixson N/A Goat anti-rabbit HRP-conjugated antibody Enzo Life Sciences Cat#ADI-SAB-300-J Bacterial and virus strains See Table 1 in the main text for a list of bacterial strains used in this study. This paper N/A Chemicals, peptides, and recombinant proteins Q5 DNA polymerase New England Biolabs Cat#M0491L EcoRI methyltransferase New England Biolabs Cat#M0211S FM 4-64 dye Thermo Fisher Scientific/Invitrogen Cat#T13320 DyLight 488 maleimide dye Thermo Scientific Cat#46602 Anhydrotetracycline HCl Sigma-Aldrich Cat#37919 Methylcellulose 4000 cP Sigma-Aldrich Cat#M0512 Critical commercial assays Clarity Western ECL Substrate Bio-Rad Laboratories Ltd. Cat#1705061 Novex pH 3-7 IEF Protein Gels, 1.0 mm, 10-well Thermo Fisher Scientific/Invitrogen Cat#EC6645BOX Deposited data Subtomogram average of EJC168 12.5 ng/mL ATc This paper Electron Microscopy Database: EMD-17769 Subtomogram average of EJC168 25 ng/mL ATc This paper Electron Microscopy Database: EMD-17770 Subtomogram average of EJC168 50 ng/mL ATc This paper Electron Microscopy Database: EMD-17771 Subtomogram average of EJC168 100 ng/mL ATc This paper Electron Microscopy Database: EMD-17772 Subtomogram average of flgP AAA This paper Electron Microscopy Database: EMD-17773 Subtomogram average of flgP D18-62 This paper Electron Microscopy Database: EMD-17774 Subtomogram average of flgP-lpp 55 This paper Electron Microscopy Database: EMD-18274 Subtomogram average of DflgPQ This paper Electron Microscopy Database: EMD-17775 Subtomogram average of DflgPQ D0661 DkpsD DpglAB::aphA This paper Electron Microscopy Database: EMD-17776 Experimental models: Organisms/strains Fertilized eggs from White Leghorn chickens Charles River SPAFAS N/A Oligonucleotides See Table S1 for a list of oligonucleotides used for strain construction in this study. Integrated DNA Technologies (IDT) N/A Software and algorithms ImageJ/FIJI v2.1.0/1.53c Schindelin et al. 64 N/A Leginon 3.6 Suloway et al. 65 ; Cheng et al. 66 N/A Dynamo 1.1.532 Castan ˜o-Dı ´ez et al. 67 N/A IMOD 4.11 Kremer et al. 68 N/A Geneious Prime 2021.0.3 Biomatters, New Zealand N/A Tomo3D 2 Agulleiro and Fernandez 61,62 N/A ll OPEN ACCESS Article e1 Developmental Cell 59, 3306–3321.e1–e5, December 16, 2024 Age and development of avian hosts Fertilized eggs from white leghorn chickens were incubated for 21 days at 37.5 C for 21 d with appropriate humidity and rotation in a Digital Sportsman model 1502 incubator (Georgia Quail Farms). All subsequent infections with Campylobacter jejuni only involved chickens within 24 h of hatch. Undeveloped chicks in eggs were not used in any infection. Housing conditions of avian hosts Within 24 h of hatch, groups of at least six chicks were placed into brooders and given water and generic chick feed ad libitum. Heat was provided with a heat lamp. Due to these conditions, the chickens developed a normal gastrointestinal flora. Considerations of sex and gender as a biological variable As chicks upon hatching are difficult to sex type, both male and female chicks were included in all studies and randomly assigned to control and experimental groups. There is no known association of sex and gender of outbred white leghorn chickens with colonization by Campylobacter jejuni. Authorization of in vivo vertebrate usage All use of animals in experimentation has been approved by IACUC at the University of Texas Southwestern Medical Center. METHOD DETAILS Cultivation of C. jejuni All strains of C. jejuni used in this study are derivatives of DRH212 (rpsL K88R ), a streptomycin-resistant isolate of C. jejuni strain 81176. 55 For all experiments, cultures were grown at 37C on 1.4% Mueller-Hinton agar supplemented with 10 mg/mL Trimethoprim (MHT agar) plus other antibiotics as needed. Soft agar for motility assays used 0.35% MHT agar. Antibiotics were added as needed at the following concentrations: kanamycin, 50 mg/mL; chloramphenicol, 12.5 mg/mL; Streptomycin, 200 mg/mL and 2 mg/mL; Anhydrotetracycline HCl, 0.0125-0.2 mg/mL. For experiments at Imperial College London and UT Southwestern, cultures were grown microaerobically (85% N 2 , 10% CO 2 ,5% O 2 ) in trigas incubators. Thermo-Fisher Campygen (University of York) and Mitsubish (UEC) Gas-generating sachets were used at University of York and University of Electrocommunications, respectively. Cell suspensions for optical microscopy experiments, tomography, and autoagglutination assays were prepared by seeding a small amount of culture from a -80C master stock on a fresh MHT plate. Following 20-24 hours incubation, overnight growth of the inoculum was spread on another fresh MHT plate and incubated overnight. In the morning, fresh growth was gently washed off the plate into MH broth by pipetting. Soft agar motility assays For soft agar motility assays, a small amount of fresh overnight growth from an MHT plate was poked into 0.4% agar MHT using a toothpick and incubated microaerobically for 18-48 hours. All swarm assays were performed in triplicate For each of the four lineages in the DflgQ/DflgPQ evolution experiment, a single colony at day 0 was poked into 0.4% agar MHT and incubated microaerobically for 4-5 days, at which point the periphery or any flares from the initial point of inoculation were picked with a toothpick and poked a single time into a fresh 0.4% agar MHT plate. This was repeated 4-5 times, at which point all four of the lineages had evolved from a non-motile phenotype to a non-bushy mot+ phenotype. As our trigas incubator maintains 100% humidity at all times, evaporation from the agar during the extended incubation periods of the evolution experiment was presumed to be negligible. Genetic manipulation of C. jejuni All mutations generated in this study are chromosomally integrated at their native loci, unless otherwise stated. In-frame markerless gene deletions were constructed by leaving several Nand C-terminal residues of the targeted gene intact in order to minimize polarity on flanking genes. For the DkpsD allele, the deletion covers from I24 to V520. For the Cjj_81176_0661 clean deletion, residues F19 to K322 were removed. For the DflgPQ strain, the deletion covered from codon 17 of flgP codon 129 of flgQ, and the DflgQ deletion spanned from codon 56 to 129 of flqQ. The pglAB::aphA replaces codon 24 of pglB to codon 360 of pglA with aphA, conferring kanamycin resistance. Strain construction was performed as previously described. Briefly, an aphA-rpsL WT cassette with 500-1000 bp of flanking homology to the targeted gene was introduced by natural transformation, selecting for kanamycin resistance (Km R ) and screening for streptomycin sensitivity (Sm S ) on 200 mg/mL streptomycin (the rpsL WT allele is dominant to rpsL K88R in the merodiploid). Counterselection for loss of the aphA-rpsL WT cassette was accomplished by transformation of the Km R Sm S intermediate strain with a fragment of DNA encoding the desired mutation, selecting for Sm R on 2 mg/mL streptomycin and screening for Km S .Sm R Km S transformants were single colony purified, sequenced and stored at -80C. For all transformations, linear DNA was generated by SOE PCR as described previously. ecoRI sites were added to both ends of each fragment for methylation in order to increase transformation efficiency. 56 Typically 1-2 mg of methylated DNA was transformed for the initial Km R selection, and 5-10 mg for the Sm S counterselection. ll OPEN ACCESS Article Developmental Cell 59, 3306–3321.e1–e5, December 16, 2024 e2 Western blotting Western blotting was performed using whole-cell lysates. Cells were washed off plates into MH broth and the O.D. 600 adjusted to 0.5 prior to boiling in 2x SDS Laemmli buffer. 15 uL of each sample was run on either 12% or 4-20% Novex wedgewell tris-glycine polyacrylamide gels. Separated proteins were transferred to 0.2 mm nitrocellulose membrane using an iBlot 2 transfer apparatus. Secondary antibodies were HRP-conjugated and imaging was performed using Clarity Western ECL substrate (Bio-rad) and a Chemidoc imaging system (Bio-rad) Western blotting for FlgP was performed using anti-FlgP antisera raised in rabbit. Western blotting for the 6xHis-tag was performed using either HRP-conjugated primary antibody (Sigma, rabbit) or non-HRP-conjugated antisera (rabbit). Western blotting for FlgR was performed using anti-FlgR antisera raised in rabbit. Filament depolymerization for blotting To characterize differences between the flagellar filaments in the flgP-lpp 55 and flgP-lpp 55 pseG I142T mutants, overnight cultures of each strain were washed off solid MHT agar to an O.D. 600 of 1.0. 1 mL of cell suspension was added to 100 mL of MHT broth supplemented with 10 mM MgSO 4 ,7H 2 O and grown microaerobically with shaking to early-mid log phase (O.D. 600 0.4). Cells were pelleted and washed 2x with PBS before being suspended in a pH 2.2 solution of 200 mM glycine and icubated with gentle agitation at room temperature for 30 minutes to depolymerize the flagellar filaments. Following depolymerization, the cell suspensions adjusted to pH 7 with 2M NaOH. Following neutralization, cells were pelleted by centrifuging at 20k x g for 20 minutes. 15 mL of each supernatant were then concentrated 100 fold with a centrifugal filter unit (Amicon Ultra Centrifugal Filter unit, 30kDa MWCO). For Coomassie blots, concentrated supernatants were mixed 1:1 with 2xSDS loading buffer with 5% b-mercaptoethanol and briefly boiled. 10 mL of each sample was run on a 12% continuous Tris-Glycine SDS-PAGE gel (Novex Wedgewell) and stained/destained according to manufacturer’s recommendations (ProteinArk Quick Coomassie). For IEF gels, the manufacturer’s recommended protocol was followed (ThermoFisher Scientific Novex IEF gels, pH 3-7). One gel was run at 100 V for one hour, followed by 1.5 hours at 200 V. A second, identical gel was run for 1 hour at 100 V, 1 hour at 200 V and 30 minutes at 300 V. Protein bands were fixed via 30 minute incubation in 12% trichloroacetic acid, followed by Coomassie staining (ProteinArk Quick Coomassie) Autoagglutination assays For autoagglutination assays, fresh growth was washed from plates to an O.D. 600 of 1.0 in 90:10 PBS:MH broth. 1 mL of cell suspensions were pipetted into disposable 1.25 cm polystyrene cuvettes and left to sit at ambient temperature (21-23C) for 24 hours. O.D. 600 measurements were taken with a benchtop cell density meter (Amersham Biosciences Ultrospec 10) every hour for 6-8 hours, as well as a final reading at 24 hours. For O.D. 600 reading, samples were not collected from the top portion of the cell suspensions. Rather, measurements were taken directly from the original 1 mL cell suspensions in their cuvettes. 3D-holographic microscopy Holographic cell tracking was performed on an inverted microscope as previously described. 57 In brief, sample chambers measuring 20 ✕5✕0.3 mm 3 were constructed from glass slides and coverslips. These chambers were loaded with cell suspensions diluted to a concentration of approximately 3 ✕10 6 cells/ml. The standard condenser assembly in the microscope was replaced with a holder for a single-mode optical fibre directed along the optical axis of the microscope. A fibre coupled laser with a wavelength of l=642 nm and an optical power at the sample of 3 mW/cm 2 was used to illuminate the sample. The sample was imaged using a 20✕magnification objective lens onto a camera with pixel size of 14 um, leading to a spatial sampling frequency of 1.422 pixels/mm. Images were acquired at 100 Hz with a 3 ms exposure time. Background correction was performed by creating an image from the median pixel value at each (x,y) location, then dividing the pixel value in each frame by its corresponding value in the median image. We used RayleighSommerfeld back-propagation to create a stack of refocused images from each frame of the raw video and segmented the corresponding 2D image stack by finding places in which the axial intensity gradient lay above a certain (manually-determined) threshold. These locations are candidates for cell positions. We then linked the coordinates in subsequent frames into cell tracks, 58 which were subjected to further analysis. Tracks shorter than 0.4 seconds were discarded. These were typically the result of cells entering and leaving the field of view. We calculated the mean-squared displacement (MSD, ) for each cell, 59 and fitted the first second of data with the function . The exponent indicates the nature of the cell’s motion and takes values between 1 (diffusive motion) and 2 (purely straight-line swimming). These values are plotted against the cells’ root-mean-squared displacement after 1 second (obtained by extrapolation for short tracks) in Figure 2. High-speed fluorescence microscopy High-speed videos were recorded as described previously. 5 Briefly, specimen chambers were prepared by adhering a 24 mm x 40 mm coverslip to a 18 mm x 18 mm coverslip using porous double-sided tape (Nichiban (size 02)). Following pipetting of sample into specimen chambers, chambers were sealed with clear nail lacquer to reduce drift. We used DyLight 488-conjugated maleimide dye (Thermo Fisher) to label flagellar filaments. Cell bodies were labelled using FM 4-64 dye (Life Technologies). Labelling was carried out by using PBS to wash cells off MH plates to an O.D. 600 of 1, washing cells 1x in PBS by pelleting cells at 10,000 x g and resuspension in fresh PBS, to which Dylight 488-conjugated maleimide was added and ll OPEN ACCESS Article e3 Developmental Cell 59, 3306–3321.e1–e5, December 16, 2024 incubated at 37C for 20 minutes before addition of FM 4-64, immediately after which cells were pelleted again and resuspended in fresh MH + 0.5% methylcellulose. Unless otherwise stated, all movies were captured at 400 frames per second, and cell suspensions were MH broth supplemented with methylcellulose (4000 cP, Sigma Aldrich) to a final concentration of 0.5%. Movies were captured with an inverted microscope (IX83, Olympus), equipped with an objective lens (UPLXAPO1003OPH, N.A. 1.45, Olympus), dichroic mirrors (Di01-R488, Semrock), dual-view imaging system with optical filters (FF560-FDi01, FF03-535/50 and BLP01-568R, Semrock), a CMOS camera (Zyla 4.2, Andor), and an optical table (ASD-1510T, JVI). Projection of the image to the camera was made at 0.065 mm per pixel. The focal position of the sample was kept at the focal position by a Z-drift compensation module (ZDC, Olympus). A blue laser beam (OBIS488-20, Coherent) was introduced into the microscope, and the resultant fluorescent images were acquired by imaging software (Solis, Andor) as 16-bit images under 2.5-ms intervals. Aerotaxis assays Aerotaxis assays were performed as described previously. 5 Briefly, specimen chambers were prepared by adhering a 24 mm x 40 mm coverslip to a 18 mm x 18 mm coverslip using porous double-sided tape (Nichiban (size 02)). Cell suspensions were adjusted to an O.D. 600 of 1 and pipetted into a sample chamber. Due to the speed at which populations of WT cells will aerotax, sample chambers were not sealed with clear nail lacquer. Recording was started prior to the addition of samples for the same reason. Movies were recorded at 3 frames per second using a darkfield microscope (IX83, Olympus) equipped with an objective lens (CPLFLN103PH, N.A. 0.3, Olympus), darkfield condenser (U-DCD, Olympus), and a CMOS camera (Zyla 4.2, Andor) and an optical table (ASD-1510T, JVI). Projection of the image to the camera was made at 0.65 mm per pixel. Sequential images of cells were acquired by the imaging software (Solis; Andor) as 16-bit images with the CMOS camera. Kymographs were generated in ImageJ version 1.48. The height of the sequential images was resized to one pixel and aligned vertically so that the y-axis represents time. Microfluidic experiments Microfluidic devices with confined 1 mm channels were fabricated using standard photolithography and soft lithography methods as described previously. 60 Briefly, polydimethylsiloxane (PDMS, Sylgard 184, Dow), a two-part silicone elastomer, was cast over a photolithography master and cured at room temperature for 48 h. A piece of PDMS was cut out using a scalpel and used as a microfluidic device. Cell suspensions with MH broth containing 0.5% methylcellulose, were dropped onto a glass slide and then covered with the microfluidic device casting from the top. Movies were captured with an inverted microscope (IX73, Olympus), equipped with an objective lens (UPLXAPO1003OPH, N.A. 1.45, Olympus), a filter set (GFP-4050B, Semrock), mercury lamp (U-HGLGPS, Olympus), a CMOS camera (DMK33UX174, Imaging Source), and an optical table (HAX-0806, JVI). Projection of the image to the camera was made at 0.058 mm per pixel. Sequential images were acquired by the imaging software (Solis, Andor) as 16-bit images under 25-ms intervals. Negative stain TEM For measuring flagellar filament lengths, overnight cultures were gently washed from MHT plates into PBS and washed 1x with PBS, followed by fixation with 1% EM grade glutaraldehyde in PBS on ice for 10 minutes. Fixed cell suspensions were centrifuged and resuspended in water and kept on ice. Samples were applied to glow-discharged carbon-coated copper grids (Agar Sciences) and stained with 2% uranyl acetate. Images were collected on a FEI T12 TEM at an acceleration voltage of 120 kV at a nominal magnification of 6,500x. Filament lengths were measured by hand in FIJI. Electron cryotomography and subtomogram averaging Strains to be imaged for subtomogram averaging were washed off plates and concentrated to an O.D. 600 of 10-20 and mixed with 10 nm gold fiducial markers (Sigma-Aldrich) in 5% BSA. Samples were applied to Quantifoil R2/2 grids and plunge frozen in liquid ethane using a Vitrobot (FEI). Imaging was performed on a Thermo-Fisher Glacios 200 kV electron microscope equipped with a Falcon 4 direct electron detector and Selectris energy filter. Tomograms were reconstructed using a combination of IMOD 4.11.8 for fiducial modeling and Tomo3D for SIRT tomographic reconstruction. 61,62 To enhance the contrast of tomograms for display of unaveraged motors and to measure disk diameters, tomograms were CTF-deconvoluted as first described by (Tegunov and Cramer 2019) 63 but with CTF deconvolution performed in 2-D on the tilt series prior to 3-D tomographic reconstruction. In short, the procedure restores the magnitude of the low-resolution components that are attenuated by the CTF while removing the noisy components at medium and high resolution, which results in an overall contrast improvement. This code is available in version 2.2 of Tomo3D. For subtomogram averaging, particles were picked using 3dmod from the IMOD suite and imported into Dynamo 1.1.532 for subtomogram averaging. We imposed C17 symmetry for averaging based upon established prominent symmetry of the periplasmic structures of the C. jejuni motor. Subtomogram averages of motors in this work have been deposited with the Electron Microscopy Data Bank (EMDB) under the following accession numbers: EJC168 12.5 ng/mL ATc: EMD-17769; EJC168 25 ng/mL ATc: EMD-17770; EJC168 50 ng/mL ATc: EMD-17771; EJC168 100ng/mL ATc: EMD-17772; flgP AAA : EMD-17773; flgP D18-62 : EMD-17774; flgP-lpp 55 : EMD-18274; DflgPQ: EMD-17775; DflgPQ D0661 DkpsD DpglAB::aphA: 17776. ll OPEN ACCESS Article Developmental Cell 59, 3306–3321.e1–e5, December 16, 2024 e4 Chicken colonization assays Chick colonization assays. The ability of WT C. jejuni 81–176 rpsL K88R and isogenic mutants to colonize chicks after oral inoculation was determined as previously described (32). Briefly, fertilized chicken eggs (SPAFAS) were incubated for 21 days at 37.5 C with appropriate humidity and rotation in a Digital Sportsman model 1502 incubator (Georgia Quail Farms Manufacturing Company). One day after hatch, chicks were orally inoculated with 100 mL of phosphate buffered saline (PBS) containing approximately 180240 CFU WT or mutant strains. Strains were prepared for infection after 16 h growth at 37 C under microaerobic conditions on MH agar by suspending C. jejuni strains in MH broth. Dilution series in PBS were performed to achieve the appropriate inoculum for oral gavage of chicks. Dilutions of the inoculum were plated on MH agar to assess the number of bacteria in each inoculum. At 7 days post-infection, chicks were sacrificed, the cecal contents were removed and suspended in PBS, and serial dilutions were plated on MH agar containing trimethoprim and cefoperazone. Following 72 h of growth at 37 C in microaerobic conditions, bacteria were counted to determine CFU per gram of organ content. Recovered colonies were analyzed by colony PCR to verify that WT and mutant strains were isolated from respectively infected chicks. Whole genome sequencing Whole genome sequencing was performed by Source Biosciences (U.K.). Genomes were assembled and analysed using the software package Geneious Prime 2021.0.3 (Biomatters, New Zealand). The paired reads provided by Source Biosciences were imported into Geneious and trimmed using BBDuk, removing adapters and low-quality reads. Whole genome sequencing reads of parental strains flgP-lpp 56 and DflgPQ were mapped to a C. jejuni reference genome NC_008787. These assembled genomes were then used as reference genomes against which suppressor genomes were assembled and analysed. We used the Geneious variant finder to find mutations in each sequenced suppressor genome relative to its parental reference genome, characterise mutation frequency and its possible effect on codon and amino acid changes. Phylogenetics A phylogenetic tree of the Cjj_81176_0661 family was determined using the sequences allocated to the PFAM PF04748 family as downloaded on 17th May 2022 and performing a multiple sequence alignment using RAxML with a Jones-Taylor-Thornton (JTT) model of amino acid substitution rates with a discrete gamma distribution. QUANTIFICATION AND STATISTICAL ANALYSIS Quantification and statistical analyses were carried out using GraphPad Prism (v10.1.1). For comparison of two groups, e.g. comparing swimming speed of two mutants, a two-tailed t-test was performed and difference were considered significant if p=<0.05. For comparisons of >2 groups, 1or 2-way ANOVA was used for statistical analysis. Number of replicates, p-values and SEM are provided in figures and figure legends. ll OPEN ACCESS Article e5 Developmental Cell 59, 3306–3321.e1–e5, December 16, 2024