A link between the assembly of flagella and lipooligosaccharide of the Gram-negative bacterium

Thomas W. Cullena and M. Stephen Trenta,b,1

aSection of Molecular Genetics and Microbiology and bThe Institute of Cellular and Molecular Biology, The University of Texas at Austin, Austin, TX 78712

Edited* by Christian R.H. Raetz, Duke University, Durham, NC, and approved February 1, 2010 (received for review November 19, 2009) Campylobacter jejuni is the leading cause of acute bacterial diarrhea groups attached at the 1 and 4′ positions of the disaccharide worldwide and is implicated in development of Guillain-Barré syn- backbone (6) (Fig. 1C). drome. Two major surface features, the outer membrane lipooligo- Here we report the identification of a pEtN transferase from saccharide and flagella, are highly variable and are often targets for C. jejuni, Cj0256, that unexpectedly modifies two periplasmic tar- modification. Presumably, these modifications provide a competi- gets, a membrane lipid and a flagellar protein. Deletion of cj0256 tive advantage to the bacterium. In this work, we identify a gene results in the loss of pEtN modification of C. jejuni lipid A and encoding a phosphoethanolamine (pEtN) transferase (Cj0256) that sensitivity to CAMPs, polymyxin B. Suprisingly, cj0256 mutants serves a dual role in modifying not only the lipooligosaccharide lipid showed decreased motility and greatly reduced flagella production. anchor lipid A with pEtN, but also the flagellar rod protein FlgG. Our data indicate that Cj0256 also modifies the flagellar rod protein Generation of a mutant in C. jejuni 81–176 by interruption of FlgG (7). Modfication of FlgG is a known periplasmic post- cj0256 resulted in the absence of pEtN modifications on lipid A as translational modification of a bacterial flagellar component. Given well as FlgG. The cj0256 mutant showed a 20-fold increase in sensi- that Cj0256 is member of a large family of proteins (COG2194) tivity to the cationic antimicrobial peptide, polymyxin B, as well as a found in a number of pathogenic , periplasmic decoration of decrease in motility. Transmission EM of the cj0256 mutant revealed bacterial structures with phosphoryl substituents are likely to play an a population (approximately 95%) lacking flagella, indicating that,

important role in pathogenesis. MICROBIOLOGY without pEtN modification of FlgG, flagella production is hindered. Most intriguing, this research identifies a pEtN transferase showing Results preference for two periplasmic substrates linking membrane bio- Cj0256 Is a Lipid A pEtN Transferase. To identify the enzyme genesis and flagellar assembly. Cj0256 is a member of a large family responsible for pEtN modification of C. jejuni lipid A (Fig. 1C), we of mostly uncharacterized proteins that may play a larger role in the performed a BLASTp search using a previously characterized pEtN decoration of bacterial surface structures. transferase from (hp0022) (8), revealing a single − putative homologue, cj0256 (E-value <10 55). To confirm that cell envelope | lipid A | lipopolysaccharide | motility | antimicrobial Cj0256 functions as a lipid A pEtN transferase, we heterologously peptides expressed Cj0256 in E. coli K-12 strain W3110 (strain EC01). Cul- 32 turesofEC01wereradiolabeledwith Pi and the purified lipid A ampylobacter jejuni is a major cause of bacterial diarrhea separated using TLC. E. coli K-12 W3110 produced the Cworldwide (1). Infection with this results in sig- typical lipid A species seen in WT, 1,4´-bis-phosphate lipid A, and 1- nificant acute illness as well as serious life-threatening con- diphosphate lipid A (Fig. 2A). As a control, an E. coli K-12 PmrA fi sequences, such as Guillain-Barré syndrome (2). Like all constitutive mutant WD101 (9), modi ed with pEtN at the 1 and 4´ , C. jejuni synthesizes complex outer surface positions, was used to identify the migratory patterns of pEtN modified lipid A (Fig. 2A). Expression of Cj0256 resulted in the structures that are critical for pathogenesis. Two major surface fi fi features, the outer membrane lipooligosaccharide (LOS) and production of pEtN-modi ed lipid A species (Fig. 2A). To con rm our findings, purified lipid A from strains W3110 and EC01 was flagella, are highly variable and are often targets for modification, analyzed by MALDI-TOF MS. Analysis of WT W3110 revealed presumably to provide a competitive advantage to the bacterium. major ion peaks at m/z 1,796.9 and 1,876.9 representing 1,4´-bis- For example, the structural subunits in the flagella filament of phosphate lipid A and 1-diphosphate lipid A, respectively (Fig. 1A epsilon are glycosylated. In the case of C. jejuni, fl and Fig. S1). Analysis of EC01 revealed major ion peaks at m/z these glycosylation events are required for assembly of agellin 1,918.8 and 2,041.5 representing lipid A species bearing a single subunits. Flagella components including the rod, hook, and fi pEtN or two pEtN residues, respectively (Fig. S1). structural lament are secreted from the cytoplasm through a To determine if phosphatidylethanolamine (PtdEtN) serves as fl growing narrow channel via the agellar type III secretion system the pEtN donor for Cj0256, we used an E. coli strain (AD90) and polymerize at the distal end (3). To date, posttranslational harboring an insertion mutation in the gene encoding for phos- fi fl modi cation of agellar components has been shown to occur only phatidyl serine synthase (pss). These bacteria lack PtdEtN, and in the cytoplasm before secretion and reported only in the in this background, expression of Cj0256 had no effect on lipid A filament flagellins. synthesis (Fig. S2). Complementation of the pss mutation with Another well characterized bacterial surface structure, and the major molecule found on the surface of C. jejuni, is LOS. Like all Gram-negative bacteria, the hydrophobic anchor of LOS or Author contributions: T.W.C. and M.S.T. designed research; T.W.C. performed research; lipopolysaccharide (LPS) is lipid A. Many Gram-negative bac- T.W.C. and M.S.T. analyzed data; and T.W.C. and M.S.T. wrote the paper. teria modify their lipid A to provide protection against cationic The authors declare no conflict of interest. antimicrobial peptides (CAMPs) and to avoid detection by the *This Direct Submission article had a prearranged editor. host Toll-like receptor 4/MD2 innate immune receptor (4, 5). Freely available online through the PNAS open access option. The lipid A of C. jejuni is characterized by longer secondary acyl 1To whom correspondence should be addressed. E-mail: [email protected]. ′ ′ chains attached to the 2 and 3 positions of the molecule and by This article contains supporting information online at www.pnas.org/cgi/content/full/ the addition of phosphoethanolamine (pEtN) to the phosphate 0913451107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0913451107 PNAS Early Edition | 1of6 Downloaded by guest on September 28, 2021 Fig. 1. Structural comparison of the lipid A anchors found in E. coli (A and B) and C. jejuni (C). Dashed bonds indicate modifications of lipid A and the length of acyl chains is indicated. In WT E. coli, an additional phosphate group (in red) can be attached at the 1-position (11). Under conditions leading to activation of the transcriptional regulators PhoP and PmrA, E. coli lipid A can be further derivatized with pEtN (magenta), 4-amino-4-deoxy-L-arabionse (blue), and palmitate (black) (4, 5). In C. jejuni the glucosamine disaccharide backbone can be replaced with the analogue 2,3-diamino-2,3-dideoxy-D-glucopyranose, resulting in two additional amide linked (brown) acyl chains.

plasmid pDD72 restored the ability of Cj0256 to decorate E. coli Cj0256 was capable of modifying E. coli lipid A. However, inacti- lipid A with pEtN (Fig. S2). These data suggest that PtdEtN vation of MsbA transport at 44 °C resulted in loss of pEtN mod- serves as the donor substrate for pEtN modification, but does not ification, indicating that Cj0256-dependent pEtN occurs in the rule out the use of other putative donors such as lyso-PtdEtN. periplasm (Fig. S3). The production of the 1-diphosphate species Modification of lipid A most often occurs following its transport of lipid A catalyzed by LpxT was lost upon inactivation of MsbA across the inner membrane by the dedicated flippase MsbA (4). In as previously shown (11) (Fig. S3). E. coli WD2, MsbA transport of lipid A across the inner membrane cj0256 C. jejuni is lost following a temperature shift to 44 °C (10). To investigate Generation of a Deletion Mutant in and Analysis of its fi Lipid A. Many pathogen-associated virulence factors in C. jejuni are the site of Cj0256-dependent modi cation, we examined pEtN 32 transfer in WD2. WD2 containing pWcj0256 were cultured at the phase variable, including its LOS (12). Therefore, P-labeled lipid A from three C. jejuni strains for which the genome sequences are permissive temperature of 30 °C until midlog, followed by a shift to available (strains 81–176, 11–168, and 81–116) and from a clinical 44 °C. Cells were pulse labeled and de novo 32P-labeled lipid A isolate (strain VLAO) were compared. The migration pattern of species separated by TLC as previously described (8). At 30 °C, lipid A species following TLC for all four strains was identical, thus indicating no variation in the lipid A domain of C. jejuni LOS. Also, a comparison of the lipid A migration to that of E. coli WD101 suggested the presence of both singly and doubly modified lipid A species (Fig. S4). To ascertain the role Cj0256 plays in lipid A modification in a C. jejuni background, deletion mutants were created in strain 81–176 by interruption of cj0256, to give strains 81–176A1 and 81–176A2. The mutation was later complemented by chromosomal insertion of a complete cj0256 gene into the arylsulfatase gene (astA) ORF as described previously (13), cre- ating strain 81–176B. The cj0256 deletion mutants revealed one major lipid A species similar to the 1,4´-bis-phosphate lipid A of E. coli W3110, devoid of pEtN modification (Fig. 2B). The lipid A modification was recov- ered by complementation in strain 81–176B showing lipid A species identical to that of WT (Fig. 2B). To confirm our findings, purified lipid A from C. jejuni strains was analyzed by MALDI-TOF MS. Analysis of WT 81–176 and complemented strain 81–176B revealed a major ion peak at m/z 1921, indicating the addition of a pEtN residue to the disaccharide backbone of hexa-acylated lipid A and the loss of a phosphate group at the 1-position (Fig. 3 A and C). The latter often occurs during MS of lipid A species (8). Analysis of cj0256 deletion mutant 81–176A1 revealed a major ion peak at m/z 1798, indicative of 1-dephosphorylated C. jejuni lipid A showing no pEtN modification (Fig. 3B). Additionally, MS indicated that Fig. 2. Cj0256 is a lipid A pEtN transferase. 32P-labeled lipid A was isolated the disaccharide backbone of C. jejuni lipid A is not composed solely from the indicated strains, separated by TLC and visualized by phosphor- C of glucosamine residues, but can be replaced with the analogue imaging. WT (W3110) and pmrA (WD101) E. coli were used as controls to 2,3-diamino-2,3-dideoxy-D-glucopyranose (Fig. 1) as previously confirm the migratory patterns of pEtN modified lipid A. “1-PP” indicates fi 1-diphosphate lipid A produced by W3110. “Double” indicates lipid A spe- reported (6). Altogether, our results con rm that Cj0256 is cies modified primarily with two pEtN groups as seen in WD101. (A) Het- responsible for catalyzing the transfer of pEtN to C. jejuni lipid A. erologous expression of cj0256 in E. coli (EC01) results in modification of lipid A with pEtN. (B) Loss of cj0256 in C. jejuni (81-176A1 and 2) results in loss of pEtN Modification Is Required for Antimicrobial Resistance in C. pEtN modification compared with WT (81–176) and complemented (81– jejuni. In the related organism H. pylori, it was demonstrated 176B) C. jejuni. that decoration of lipid A with pEtN was critical for resistance to

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.0913451107 Cullen and Trent Downloaded by guest on September 28, 2021 3.6 μg/mL; Table S1). Additionally, PMB resistance (MIC of 15.0 ± 2.0 μg/mL) was restored in strain 81–176C, which was cre- ated by chromosomal complementation of 81–176A1 with E. coli eptA. EptA was previously shown to be involved in the modification of E. coli lipid A with pEtN (15). These results confirm that the addition of pEtN to the lipid A backbone in C. jejuni provides a 20- fold increase in resistance to PMB and is likely important for resistance to CAMPs found within the intestinal mucosa.

Cj0256 Is Required for Efficient Motility and Flagella Production. In C. jejuni, the flagellum or flagella motility is required for effective adherence and invasion into intestinal epithelial cells for the col- onization of chickens and humans, demonstrating the importance of this virulence factor (16–18). Interestingly, a previous screen of C. jejuni transposon mutants identified Cj0256 as an unknown protein involved in motility (19). To test the role Cj0256 plays in motility, we used soft agar assays. Strain 81–176A1 (Δcj0256) revealed a decrease in motility to approximately 50% relative to 81–176 or 81–176B (Δcj0256, cj0256+; Fig. 4 A and B). To explain the loss of motility, bacterial strains were examined by TEM. Strains 81–176 and 81–176B were indistinguishable from each other, showing normal C. jejuni morphology, spiral or curved rods with bipolar flagella. Unexpectedly, strain 81–176A1 revealed a population (approximately 95%) lacking flagella but otherwise unremarkable with regard to cell morphology (Fig. 4C and Table S2). Surprisingly, motility was not recovered in strain 81–176C (Δcj0256, eptA+), although E. coli EptA was capable of restoring Fig. 3. MALDI-TOF MS analysis of C. jejuni lipid A. Analysis of of lipid A from polymyxin resistance. The latter suggested that the observed MICROBIOLOGY – – – strains 81 176, 81 176A1, and 81 176B revealed major ion peaks of m/z motility phenotype was not the result of changes in the LOS lipid 1,921.2, 1,798.3, and 1,921.1, respectively, indicating the addition of a pEtN fl residue to the disaccharide backbone in strains with an active copy of cj0256 A anchor, but rather an unrelated role played by Cj0256 in agella and the loss of a phosphate group at the 1-position in all strains. The latter production (Fig. 4B). often occurs during MS of lipid A species. To confirm that our motility phenotype was not the result of phase variation, cj0256 was deleted in the following backgrounds: 81–116, VLAO, and 11–168. All cj0256 deletion mutants showed polymyxin B (PMB), a positively charged lipopeptide that binds to reduced motility relative to its respective parent, similar to that the phosphate groups of lipid A, killing Gram-negative bacteria in found in 81–176A1, confirming our previous results and ruling out a manner similar to CAMPs of the innate immune system (14). random phase variation as the cause (Fig. 4B and Fig. S5). Strain 81–176A1 (Δcj0256) showed a striking decrease in resist- ance to PMB (MIC of 0.8 ± 0.2 μg/mL) compared with WT (MIC Cj0256 Modifies FlgG with pEtN. Discovery of an unexpected motility of 17.3 ± 3.3 μg/mL) and cj0256 complemented strain (17.6 ± phenotype and the loss of flagella production raised unanswered

Fig. 4. Cj0256 is required for efficient motility and flagella production. (A and B) Quantitative and qualitative analysis of motility of indicated C. jejuni strains was determined using semisolid MH agar. (C) Transmission EM of select C. jejuni strains.

Cullen and Trent PNAS Early Edition | 3of6 Downloaded by guest on September 28, 2021 questionsabouttheroleplayedbyCj0256inC. jejuni. Identification of the PptA, a protein involved in the decoration of the Neisseria gonorrheae Type IV pili with phosphoryl substituents that is struc- turally related to enzymes involved in pEtN modification of LPS, raised the possibility of phosphoryl modification of bacterial pro- teins (20). Therefore, we reasoned that C. jejuni use Cj0256 to modify a flagellar structural component with pEtN promoting flagella assembly. Following an exhaustive search of the literature, we found a published report detailing a whole-genome screen of C. jejuni using two-hybrid arrays looking for protein–protein inter- actions involving known flagellar proteins (21). The screen identi- fied Cj0256 as a motility protein showing direct interaction with FlgG (21). To test for FlgG modification, we created a chromoso- mal in-frame C-terminal histidine-tagged (His6) flgG in strains 81– + 176 and 81–176A1, creating 81–176D (flgG-His6 )and81–176E + (Δcj0256, flgG-His6 ), respectively, giving us the ability to easily purify FlgG from C. jejuni. The His6 fusion to FlgG did not alter motility when compared to its wild type parent (Fig. 4B). Cultures of 81–176D and 81–176E 32 were labeled with Pi and FlgG-purified via affinity chromatog- raphy followed by SDS/PAGE. FlgG-His6 was easily purified from both strains and detectable by Western blotting (Fig. 5A). However, fi 32 fi only FlgG-His6 puri ed from 81 to 176D revealed P-modi ed Fig. 6. ESI-MS confirmation of pEtN modification of FlgG in C. jejuni back- 32 fi – protein (Fig. 5A). P-modi cation of FlgG-His6 was absent in 81 ground. C. jejuni FlgG-His purified via affinity chromatography was subjected fi 6 176E, implicating Cj0256 in phosphoryl modi cation of FlgG (Fig. to ESI-MS for mass analysis and are shown as deconvoluted spectra. (A)FlgG-His6 5A). To confirm this finding and identify the type of modification, purified from strain 81–176D reveals a major ion peak at m/z 28,670.3. (B)FlgG- fi – FlgG purified from both strains 81–176D and 81–176E was sub- His6 puri ed from strain 81 176E reveals a major ion peak at m/z 28,546.9, a jected to electrospray ionization (ESI) MS. Strain 81–176D difference of 123.4 Da, the approximate mass of a single pEtN. deconvoluted spectra revealed a major ion peak at m/z 28,670 (Fig. 6A), whereas strain 81–176E deconvoluted spectra revealed a flgG-His or flgF-His along with cj0256 or eptA. The plasmids were major ion peak at m/z 28,546.9 (Fig. 6B). The resulting mass dif- 6 6 transformed into strain W3110 creating strains EC06 (flgF-His +, ference of 123.4 Da is the predicted size of a single pEtN residue, 6 cj0256+), EC08 (flgG-His +, cj0256+), and EC09 (flgG-His +, thus confirming the Cj0256-dependent modification of FlgG. 6 6 eptA+). Briefly, the strains were labeled with 32P and the proteins In C. jejuni, the distal portion of the flagella rod is made of two i separated by SDS/PAGE followed by Western blotting (Fig. 5B). major structural proteins, FlgF (annotated as FlgG2) and FlgG 32P-labeled Flg proteins were visualized by exposure to a phos- (22). We reasoned that, if FlgG is modified with pEtN, then per- phorimaging screen and only strain EC08 (flgG-His +, cj0256+) haps other structural components, such as the structurally similar 6 revealed the addition of phosphoryl substituents. The other two protein FlgF, are modified with pEtN. Furthermore, in light of our test strains, EC06 (flgF-His +, cj0256+) and EC09 (flgG-His +, current findings, we believed it necessary to test if other identified 6 6 eptA+), showed no modification of Flg proteins. lipid A pEtN transferases could modify FlgG. To test this theory, Protein FlgG-His was later purified by affinity chromatography we generated plasmid constructs capable of expressing C. jejuni 6 from strain EC08 and a control strain not expressing cj0256, EC07 + (flgG-His6 ), and subjected to ESI-MS. The deconvoluted spectra of EC08 revealed the addition of pEtN to FlgG; this modification was absent in EC07 (Fig. S6). These results confirm that Cj0256 is responsible for catalyzing the transfer of pEtN to FlgG and not the rod flagellar component FlgF. Furthermore, E. coli EptA was not capable of modifying C. jejuni FlgG, illustrating an unexpected promiscuous activity for Cj0256 in catalyzing not only lipid A modification with pEtN, but also FlgG. A model for Cj0256- dependent pEtN modification is shown in Fig. 7. Discussion Biogenesis of bacterial surface glycans and flagella assembly are two of the most dynamic processes observed in prokaryotes. The outer leaflet of the outer membrane in all medically important pathogenic Gram-negative bacteria is composed of LOS or LPS that is anch- ored to the outer membrane by lipid A. Not surprisingly, great variability is seen in the composition of this outer membrane com- ponent when comparing different (4, 5). The lipid A of C. jejuni differs from E. coli K-12 by increases in acyl-chain length fi 32 fi Fig. 5. Cj0256 modi es FlgG with pEtN. P-labeled FlgG-His6 was puri ed via and acyl-chain linkage, and by modification of the lipid A phosphate fi af nity chromatography from C.jejuni, Westernblottedusing anti-Hisantibody, groups with pEtN (6). Our data indicates that the Campylobacter and visualized by phosphorimaging. 32P-labeled cell free extracts of E. coli strains were used directly for all analyses. (A)Purified FlgG from C. jejuni strains reveal protein Cj0256 is responsible for pEtN addition to C. jejuni lipid A pEtN modification in strains with an active copy of cj0256.(B)Coheterologous and provides resistance to antimicrobial peptides. Furthermore, the expression of Cj0256 with FlgG-His or FlgF-His in E. coli reveals the addition of lipid A domain of C. jejuni’s LOS is not variable, but constant, as 6 6 fi pEtN to only FlgG. Coheterologous expression of eptA with FlgG-His6 reveals illustrated by examination of the lipid A pro les from multiple no pEtN modification. isolates (Fig. S4).

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.0913451107 Cullen and Trent Downloaded by guest on September 28, 2021 Fig. 7. Proposed model for pEtN modifications catalyzed by Cj0256. The model illustrates the bifunctional nature of Cj0256 in the periplasmic modification of lipid A and the flagellar rod of C. jejuni with pEtN. The model indicates that PtdEtN serves as the pEtN donor resulting in the production of diacylglycerol. Lyso-PtdEtN could also serve as a substrate for pEtN decoration. The TMHMM program (34) was used to predict the membrane topology of Cj0256. Organization of the flagellum is based on what is described for (35) and what has been proposed for C. jejuni (36).

Flagella production is a metabolically costly endeavor for a conveniently located phospholipid donor, PtdEtN, to membrane- growing bacterial population, but necessary for adaptation to the associated components (28, 29). Phosphoryl substituents are

surrounding environment. Flagella assembly is tightly regulated increasingly being recognized as important membrane associated MICROBIOLOGY and has been well characterized in organisms such as Salmonella structural components (20). A well documented example is mod- enterica serovar Typhimurium; however, the process is highly vari- ification of various LPS components with pEtN by CptA (28), able and much less is known about the regulatory and structural EptB (30), and EptA (15, 29) in E. coli and Salmonella typhimu- fl characteristics of C. jejuni agella (22). Flagellins from many polarly rium. All three proteins presumably use PtdEtN as a substrate fl agellated bacteria such as Campylobacter and Helicobacter spp. donor but show specificity in the pEtN recipient. Here, we dem- are glycosylated (23). The function of posttranslational glycan onstrate that Cj0256 requires PtdEtN for lipid A modification in modification of flagellin subunits is not fully understood, but with- fi ε whole cells (Fig. S2). A BLAST search of the genomes of E. coli out these modi cations, members of the -proteobacteria (e.g., and S. typhimurium reveals multiple proteins homologous to pEtN C. jejuni ) show motility defects and in some cases are unable to transferases, whereas C. jejuni contains only one, Cj0256. Our assemble flagella (24). Here we identify an unexpected flagellar research shows that Cj0256 functions to modify both lipid A and modification occurring on the rod assembly in C. jejuni.Purified FlgG. Complementation of a cj0256 deletion mutant with cj0256 FlgG from cj0256 deletion strains reveals a change in mass of fi approximately 123.4 Da compared with WT, representing a single or eptA reveals, in both cases, recovery of pEtN-modi ed lipid A. fi Only complementation with cj0256 showed recovery of motility, pEtN modi cation. This illustrates the bifunctional nature of fi Cj0256, linking flagella assembly and LOS biogenesis. Deletion of suggesting that Cj0256 diverged from other identi ed pEtN Cj0256 results in decreased motility and an approximately 95% transferases, acquiring a bifunctional promiscuous role in pEtN fi reduction in flagella production, suggesting that modification of modi cation, perhaps in response to the specialization of C. jejuni FlgG with pEtN may impart structural stability between the FlgG to a specific environmental niche and a “slimming down” of C. subunits. Similarly, it has been proposed that glycosylation of fila- jejuni’s genomic size. An LOS core sugar of C. jejuni is modified by ment proteins may provide additional stability to the flagellar fila- the addition of pEtN (31) similar to that of S. typhimurium (28). ment (25, 26). Still, it is not possible to rule out that Cj0256 modifies Considering Cj0256s identified promiscuous pEtN transferase other targets that play a role in flagellar assembly. activity, a third role in modification of core LOS is possible. The Flagellin glycosylation in C. jejuni occurs on the cytoplasmic side LOS from the cj0256 mutant showed no differences compared of the inner membrane before export through the flagella export with WT by SDS/PAGE analysis (Fig. S7); however, this does not apparatus (27). The active site of Cj0256 is localized on the peri- rule out possible changes in phosphorylation patterns. plasmic side of the inner membrane (Fig. S3), demonstrating that This research identifies an unexpected link between the assem- modification of FlgG occurs during subunit assembly representing a blies of two of the most important surface-associated virulence posttranslational periplasmic modification of a flagellar compo- factors. We have identified a promiscuous pEtN transferase show- nent. Interestingly, proteins involved in glycosylation of C. jejuni ing preference for two substrates, a membrane lipid and a structural flagellin subunits are shown to localize to the poles of the bacterium protein. In light of our current findings, more research is needed to fl where the growing agella subunits are exported, suggesting that elucidate the role played by this family of mostly uncharacterized Cj0256 may be localized to the poles in a similar manner (27). A transferases in the assembly of bacterial surface structures small subpopulation (approximately 5%) retained the ability to and pathogenesis. assemble mature flagella in the absence of an active copy of Cj0256, suggesting a possible second site suppressor mutation. Perhaps Materials and Methods FlgF, a structurally related rod protein can play a similar role in Bacterial Strains and Growth. A complete list of bacterial strains can be found distal rod assembly, compensating for the loss of properly assembled in Table S3. E. coli strains were grown routinely at 37 °C in Luria–Bertani (LB) FlgG. However, this merits further investigation. broth or on LB agar. C. jejuni strains were grown routinely at 37 °C in Members of COG2194 and other distantly related proteins (e.g., Mueller–Hinton (MH) broth, on MH agar, or on tryptic-soy agar supple- PptA) have been shown to catalyze the transfer of pEtN from a mented with 5% blood under microaerophilic conditions.

Cullen and Trent PNAS Early Edition | 5of6 Downloaded by guest on September 28, 2021 Construction of cj0256 Deletion Mutants. The cj0256 gene and 1,000-bp inverse PCR (primers: 13, 14) engineered to add an in-frame His6 coding flanking sequence upstream and downstream were amplified by PCR (primers: sequence before the stop codon of flgG and insert the restriction site NdeI to 1, 2) from 81 to 176 genomic DNA. The amplicon was digested with KpnI and both ends of the amplicon. A aph3 cassette, obtained by PCR (primers: 15, SacI and inserted into vector pBluescript II SK(+) creating pBcj0256. The vector 16) from cloning vector pRY107 (32), was inserted into the NdeI sites on the pBcj0256 was then used as template for an inverse PCR (primers: 3, 4) engi- R inverse PCR amplicon creating pFlgGHISKO:Kan . The resulting vector, neered to remove 1211 bps from the center of cj0256 and insert two restriction pFlgGHISKO:KanR, was transformed into C. jejuni strain 81–176 allowing for sites, XbaI and XhoI. A chloramphenicol resistance cassette (cam), obtained by expression of C. jejuni FlgG-His . PCR (primers: 5, 6) from cloning vector pRY111 (32), was inserted into the XbaI 6 and XhoI sites on the inverse PCR amplicon creating pBcj0256KO:CamR. The 32 resulting vector, pBcj0256KO:CamR, was transformed into C. jejuni strains by Visualization of Pi-Labeled Flagellar Components. E. coli (50 mL) and C. jejuni μ 32 natural transformation, and resistant colonies were selected on blood agar (200 mL) cultures were grown in media supplemented with 1.5 Ci/mL Pi. 32 plates containing 10 μg/mL of chloramphenicol. Chromosomally expressed FlgG-His6 from P-labeled C. jejuni was con- Complementation of the 81–176 cj0256 mutant (81–176A1) was achieved centrated for further analysis from approximately 8.0 mg of cell-free extracts by insertion of WT cj0256 or eptA into the arylsulfatase gene atsA (13). (CFEs) using the ProFound Pull-Down PolyHis Protein:Protein Interaction Kit R Vector pGEMatsAKO:Kan , a gift from S.A. Thompson (Medical College of (Thermo Scientific) in the presence of 6 M urea. For E. coli, no further pro- Georgia, Augusta, GA), containing atsA interrupted with a kanamycin cessing of the CFE sample was required as C. jejuni flagella components were resistance cassette (aph3) on original vector pGEM-T Easy (32, 33) was expressed from inducible plasmids. Protein samples (approximately 2 μgof digested with AgeI. AgeI cuts the vector in a noncoding region of the aph3 purified FlgG or 15 μgofE. coli CFE) were resolved by SDS/PAGE (NuPAGE 4– cassette, upstream from its promotor. The cj0256 and eptA genes plus 100 12% Bis-Tris Gels; Invitrogen), western blotted using anti–polyHis-alkaline bp upstream sequence were amplified by PCR (primers: 7–10) from 81 to 176 and W3110 genomic DNA, respectively, and inserted into the AgeI cut site. phosphatase antibody (Sigma), and analyzed by phosphorimaging. A more The resulting vectors pAtsAKO::cj0256:KanR and pAtsAKO::epta:KanR were detailed description of each step can be found in SI Materials and Methods. used to transform 81–176A1 for complementation studies. Kanamycin- resistant colonies were screened for loss of AstA activity as previously Other Methods. Methods describing motility assays, lipid A isolation and

described (13). analysis, large-scale purification of C. jejuni FlgG-His6, protein MS, primer sequences (Table S4), and general cloning techniques are described in SI C. jejuni Construction of Chromosomal FlgG-His6 Mutants. The coding sequence Materials and Methods. 500 bp upstream and downstream of the flgG stop codon was amplified by PCR (primers: 11, 12) from 81 to 176 genomic DNA. The amplicon was ACKNOWLEDGMENTS. We thank Rasika M. Harshey for helpful conversa- digested with KpnI and SacI, and inserted into vector pBluescript II SK(+) tions. This work was supported by National Institutes of Health Grants creating pBflgGHis. The vector pBflgGHis was then used as template for an AI064184 and AI76322 (to M.S.T.).

1. Samuel MC, et al.; Emerging Infections Program FoodNet Working Group (2004) 19. Golden NJ, Acheson DW (2002) Identification of motility and autoagglutination Epidemiology of sporadic Campylobacter infection in the United States and declining Campylobacter jejuni mutants by random transposon mutagenesis. Infect Immun 70: trend in incidence, FoodNet 1996-1999. Clin Infect Dis 38 (Suppl 3):S165–S174. 1761–1771. 2. Ang CW, et al. (2002) Structure of Campylobacter jejuni lipopolysaccharides 20. Naessan CL, et al. (2008) Genetic and functional analyses of PptA, a phospho-form determines antiganglioside specificity and clinical features of Guillain-Barré and transferase targeting type IV pili in . J Bacteriol 190:387–400. Miller Fisher patients. Infect Immun 70:1202–1208. 21. Rajagopala SV, et al. (2007) The protein network of bacterial motility. Mol Syst Biol 3: 3. Minamino T, Imada K, Namba K (2008) Mechanisms of type III protein export for 128. bacterial flagellar assembly. Mol Biosyst 4:1105–1115. 22. Hendrixson DR (2008) Regulation of flagellar gene expression and assembly. 4. Trent MS, Stead CM, Tran AX, Hankins JV (2006) Diversity of endotoxin and its impact Campylobacter, eds Nachamkin I, Szymanski CM, Blaser MJ (ASM Press, Washington, on pathogenesis. J Endotoxin Res 12:205–223. DC), 3rd Ed, pp 545–558. 5. Raetz CR, Reynolds CM, Trent MS, Bishop RE (2007) Lipid A modification systems in 23. Logan SM (2006) Flagellar glycosylation - a new component of the motility gram-negative bacteria. Annu Rev Biochem 76:295–329. repertoire? Microbiology 152:1249–1262. 6. Moran AP, et al. (1991) Structural analysis of the lipid A component of Campylobacter 24. Guerry P (2007) Campylobacter flagella: not just for motility. Trends Microbiol 15: jejuni CCUG 10936 ( O:2) lipopolysaccharide. Description of a lipid A 456–461. containing a hybrid backbone of 2-amino-2-deoxy-D-glucose and 2,3-diamino-2,3- 25. Goon S, et al. (2006) A sigma28-regulated nonflagella gene contributes to virulence dideoxy-D-glucose. Eur J Biochem 198:459–469. of Campylobacter jejuni 81-176. Infect Immun 74:769–772. 7. Homma M, Kutsukake K, Hasebe M, Iino T, Macnab RM (1990) FlgB, FlgC, FlgF and 26. Schirm M, et al. (2003) Structural, genetic and functional characterization of the FlgG. A family of structurally related proteins in the flagellar basal body of Salmonella flagellin glycosylation process in Helicobacter pylori. Mol Microbiol 48:1579–1592. typhimurium. J Mol Biol 211:465–477. 27. Ewing CP, Andreishcheva E, Guerry P (2009) Functional characterization of flagellin 8. Tran AX, et al. (2004) Periplasmic cleavage and modification of the 1-phosphate glycosylation in Campylobacter jejuni 81-176. J Bacteriol 191:7086–7093. group of Helicobacter pylori lipid A. J Biol Chem 279:55780–55791. 28. Tamayo R, et al. (2005) Identification of cptA, a PmrA-regulated locus required for 9. Trent MS, et al. (2001) Accumulation of a polyisoprene-linked amino sugar in polymyxin- phosphoethanolamine modification of the Salmonella enterica serovar typhimurium resistant Salmonella typhimurium and : structural characterization and lipopolysaccharide core. J Bacteriol 187:3391–3399. transfer to lipid A in the periplasm. JBiolChem276:43132–43144. 29. Lee H, Hsu FF, Turk J, Groisman EA (2004) The PmrA-regulated pmrC gene mediates 10. Doerrler WT, Gibbons HS, Raetz CR (2004) MsbA-dependent translocation of lipids phosphoethanolamine modification of lipid A and polymyxin resistance in Salmonella across the inner membrane of Escherichia coli. J Biol Chem 279:45102–45109. enterica. J Bacteriol 186:4124–4133. 11. Touzé T, Tran AX, Hankins JV, Mengin-Lecreulx D, Trent MS (2008) Periplasmic 30. Reynolds CM, Kalb SR, Cotter RJ, Raetz CR (2005) A phosphoethanolamine transferase phosphorylation of lipid A is linked to the synthesis of undecaprenyl phosphate. Mol specific for the outer 3-deoxy-D-manno-octulosonic acid residue of Escherichia coli Microbiol 67:264–277. lipopolysaccharide. Identification of the eptB gene and Ca2+ hypersensitivity of an 12. Gilbert M, et al. (2002) The genetic bases for the variation in the lipo-oligosaccharide eptB deletion mutant. J Biol Chem 280:21202–21211. of the mucosal pathogen, Campylobacter jejuni. Biosynthesis of sialylated ganglioside 31. Aspinall GO, Lynch CM, Pang H, Shaver RT, Moran AP (1995) Chemical structures of mimics in the core oligosaccharide. J Biol Chem 277:327–337. the core region of Campylobacter jejuni O:3 lipopolysaccharide and an associated 13. Yao R, Guerry P (1996) Molecular cloning and site-specific mutagenesis of a gene polysaccharide. Eur J Biochem 231:570–578. involved in arylsulfatase production in Campylobacter jejuni. J Bacteriol 178: 32. Yao R, Alm RA, Trust TJ, Guerry P (1993) Construction of new Campylobacter cloning 3335–3338. vectors and a new mutational cat cassette. Gene 130:127–130. 14. Tran AX, et al. (2006) The lipid A 1-phosphatase of Helicobacter pylori is required for 33. Guerry P, et al. (2006) Changes in flagellin glycosylation affect Campylobacter resistance to the antimicrobial peptide polymyxin. J Bacteriol 188:4531–4541. autoagglutination and virulence. Mol Microbiol 60:299–311. 15. Trent MS, Raetz CRH (2002) Cloning of EptA, the lipid A phosphoethanolamine 34. Sonnhammer EL, von Heijne G, Krogh A (1998) A hidden Markov model for predicting transferase associated with polymyxin resistance. J Endotoxin Res 8:159. transmembrane helices in protein sequences. Proceedings of the Sixth International 16. Wassenaar TM, van der Zeijst BA, Ayling R, Newell DG (1993) Colonization of chicks Conference on Intelligent Systems for Molecular Biology,edsGlasgowJ,etal.(AAAIPress, by motility mutants of Campylobacter jejuni demonstrates the importance of flagellin Menlo Park, CA), pp 175–182. A expression. J Gen Microbiol 139:1171–1175. 35. Chevance FF, Hughes KT (2008) Coordinating assembly of a bacterial macromolecular 17. Black RE, Levine MM, Clements ML, Hughes TP, Blaser MJ (1988) Experimental machine. Nat Rev Microbiol 6:455–465. Campylobacter jejuni infection in humans. J Infect Dis 157:472–479. 36. Larson CL, Christensen JE, Pacheco SA, Minnich SA, Konkel ME (2008) Campylobacter 18. Hendrixson DR, DiRita VJ (2004) Identification of Campylobacter jejuni genes involved jejuni secretes proteins via the flagellar type III secretion system that contribute to in commensal colonization of the chick gastrointestinal tract. Mol Microbiol 52: host cell invasion and gastroenteritis. Campylobacter, eds Nachamkin I, Szymanski CM, 471–484. Blaser MJ (ASM Press, Washington, DC), 3rd Ed, pp 315–332.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.0913451107 Cullen and Trent Downloaded by guest on September 28, 2021