Lipoprotein Processing and Sorting in Helicobacter pylori

Bacterial lipoproteins have diverse roles in multiple aspects of bacterial physiology, antimicrobial resistance, and pathogenesis. Dedicated pathways direct the posttranslational lipidation and localization of lipoproteins, but there is considerable variation in these pathways among the proteobacteria. In this study, we characterized the proteins responsible for lipoprotein synthesis and localization in Helicobacter pylori, a member of the Epsilonproteobacteria that contributes to stomach cancer pathogenesis. We also provide evidence suggesting that lipidation of CagT, a component of the H. pylori Cag T4SS, is required for delivery of the H. pylori CagA oncoprotein into human gastric cells. Overall, these results constitute the first systematic analysis of H. pylori lipoprotein production and localization pathways and reveal how these processes in H. pylori differ from corresponding pathways in model proteobacteria.

Lipoprotein synthesis in Gram-negative bacteria requires multiple steps (see Fig. S1 in the supplemental material) (9,10). Newly synthesized lipoproteins are typically recognized and exported via the Sec pathway. The lipidation machinery then recognizes a short cysteine-containing amino acid sequence known as a lipobox, located near the amino terminus of the protein. The first modification is the addition of a diacylglyceride to the cysteine sulfhydryl of the preprolipoprotein, catalyzed by prolipoprotein diacylglyceryl transferase (Lgt). Amino acids preceding the cysteine are then cleaved by prolipoprotein signal peptidase (LspA, signal peptidase II), resulting in a diacylated apolipoprotein. Finally, the amino-terminal cysteine is N-acylated by apolipoprotein N-acyltransferase (Lnt) to produce the mature triacylated lipoprotein. The three lipoprotein-specific enzymes (Lgt, LspA, and Lnt) are essential for growth of the model organism, Escherichia coli. However, Wolbachia pipientis (a Wolbachia endosymbiont of Brugia malayi) lacks a lnt homolog, and mutational analyses demonstrated that lnt is not essential in Sinorhizobium meliloti (11)(12)(13) or Francisella tularensis or Neisseria gonorrhoeae (14) or in Acinetobacter species (15).
In the canonical lipoprotein sorting pathway in Gram-negative bacteria, newly synthesized lipoproteins destined for the outer membrane interact with the LolCDE complex in the inner membrane. Lipoproteins retained in the inner membrane include a "Lol avoidance signal" and do not interact with LolCDE (9). LolCDE transfers the newly synthesized lipoprotein to the periplasmic protein LolA, which then transfers the lipoprotein to the outer membrane lipoprotein LolB for insertion into the outer membrane. Additional proteins may then act to transfer a subset of outer membrane lipoproteins to the external leaflet of the outer membrane (16).
Recent studies indicate that lipoprotein sorting in some species of bacteria does not conform to the canonical Lol pathway (17,18). For example, the Alpha-and Betaproteobacteria, as well as some Gamma-, Delta-, and Epsilonproteobacteria, harbor a single lolF gene encoding a protein that has features of both LolC and LolE (14). Though lnt is typically essential in bacteria containing LolC and LolE, lnt appears to be nonessential in bacteria containing LolF. In the absence of lnt, lipoproteins are expected to be diacylated. Thus, it has been hypothesized that LolF, in contrast to LolC and LolE, can recognize both diacylated and triacylated lipoproteins for sorting to the outer membrane (14,15). As additional evidence that the canonical pathway may not be broadly applicable, lolB is not found in the Alpha-or Epsilonproteobacteria (16,19). Furthermore, although lolA and lolB are essential in wild-type (WT) E. coli, either gene may be mutated in strains deficient in Lpp or OsmB, and global transposon mutagenesis of Caulobacter crescentus suggests that lolA is not essential (20,21). Together, these results suggest that some bacterial classes utilize a lipoprotein sorting pathway distinct from the LolAB pathway (20).
Helicobacter pylori is a Gram-negative bacterium, classified among the Epsilonproteobacteria, that colonizes the gastric mucosa of humans (22)(23)(24)(25). Colonization with H. pylori induces gastric mucosal inflammation and is associated with an increased risk for peptic ulcer disease, gastric adenocarcinoma, and gastric lymphoma (26)(27)(28)(29). The H. pylori cag pathogenicity island (cag PAI) encodes a secreted effector protein (CagA) and a type IV secretion system (Cag T4SS) that delivers CagA into human gastric cells (30,31). Individuals colonized with H. pylori strains harboring the cag PAI have a higher risk of gastric cancer or peptic ulcer disease than individuals colonized with cag PAInegative strains or H. pylori-negative individuals (32). The H. pylori genome is predicted to encode approximately 20 lipoproteins, but these predictions are largely based on informatics-driven identification of putative lipoboxes (short peptide motifs containing the cysteine that becomes lipidated) (33)(34)(35). Experimental analysis of putative H. pylori lipoproteins has been hindered by an inability to label lipoproteins in H. pylori using 3 H-palmitate, perhaps because H. pylori lacks proteins involved in long-chain fatty acid transport and catabolism (36)(37)(38), and by the failure of globomycin to inhibit signal peptide cleavage by H. pylori LspA (36,39). Thus, characterization of H. pylori lipoproteins by recombinant expression in E. coli, coupled with 3 H-palmitate incorporation or globomycin-mediated inhibition of LspA or both, has been undertaken for only a small number of H. pylori lipoproteins (36,37,39,40). H. pylori lipoproteins are currently being investigated as antigens for potential inclusion in H. pylori vaccines (41)(42)(43)(44)(45)(46)(47) and are believed to have important functions in bacterial adhesion to mammalian cells and colonization of the stomach (37,39,48), altering cell migration and signaling (36,49) and stimulating gamma interferon (IFN-␥) production (50) and natural transformation competence (51).
H. pylori homologs of the enzymes involved in posttranslational modification of lipoproteins (Lgt, Lsp, and Lnt) have been proposed, as have homologs of LolA and LolF (33,(52)(53)(54). However, it remains unclear whether these enzymes are essential for H. pylori growth. One study reported that lspA is essential whereas lolA is nonessential (54). Conversely, a microarray-based analysis of a transposon mutant library in H. pylori detected transposon insertions in lgt, lspA, lnt, and lolF but no insertions in lolA (55). A systematic experimental analysis of lipoprotein synthesis and localization pathways in H. pylori has not yet been performed.
In the present report, we present a comprehensive analysis of H. pylori lipoprotein synthesis and localization pathways. We report that the genes whose products are predicted to mediate posttranslational modification of H. pylori lipoproteins are able to complement E. coli mutants in which the corresponding E. coli genes were conditionally regulated. We constructed gene knockouts or conditional mutants in H. pylori and show that lgt, lspA, lolA, and lolF are essential for H. pylori growth whereas lnt is dispensable. We also provide experimental evidence that CagT (a component of the H. pylori Cag T4SS and a putative VirB7 homolog) undergoes amino-terminal modifications consistent with lipidation. By analyzing a lnt mutant, we show that there is little if any alteration of Cag T4SS activity when H. pylori lipoproteins (including CagT) were diacylated rather than triacylated, and by analyzing mutants in which the CagT lipobox was disrupted, we show that lipidation of CagT is essential for CagT stability and Cag T4SS activity.

RESULTS
Lipoprotein synthesis: functional complementation of conditionally lethal E. coli mutant strains. H. pylori genomes are predicted to contain homologs of the canonical genes lgt, lspA, and lnt (genes hp0955, hp0074, and hp0180, respectively, in sequenced strain 26695) required for the posttranslational modification of lipoproteins in E. coli (Fig. 1). The predicted H. pylori proteins share a relatively low level of amino acid sequence identity with characterized homologs found in E. coli or Pseudomonas aeruginosa (about 20 to 40% amino acid identity). However, each of the H. pylori proteins is predicted to adopt a three-dimensional fold similar to that of respective counterparts in other bacteria, and critical residues in each enzyme are conserved (Fig. 1). Conserved residues in H. pylori 26695 Lgt include R155 and E163 (corresponding to E. coli Lgt R143 and E151), predicted to bind phosphatidyl glycerol, the HGGL motif (residues 115 to 118 of H. pylori Lgt) which may bind to the peptide substrate, and an H-bond network consisting of R155, R232, E236, and R239 (R143, R239, E243, and R246 in E. coli Lgt) predicted to catalyze the transfer of diacylglycerol to the preprolipoprotein (56)(57)(58). H. pylori 26695 LspA includes the catalytic dyad D114 and D131 (D124 and D143 in Pseudomonas aeruginosa LspA) and 10 of 12 additional residues strictly conserved in LspA proteins from 485 organisms (the remainder of the 12 residues, G108 and A109 in P. aeruginosa LspA, are replaced by A98 and G99 in H. pylori LspA) (59). Finally, conserved residues in H. pylori 26695 Lnt include a catalytic triad comprised of E242, K296, and C349 (E267, K335, and C387 in E. coli Lnt) as well as residues Q207, K210, F211, N371, and W374 (Q233, K236, W237, N412, and W415 in E. coli Lnt) that form a pocket surrounding the site where the phosphate head group of the phospholipid donor is predicted to bind (60)(61)(62).
To experimentally evaluate whether these H. pylori proteins function as predicted, we sought to complement E. coli mutant strains in which the expression of genes required for posttranslational modification of lipoproteins is conditionally regulated. In these mutant strains, the expression of lgt, lspA, or lnt is under the control of an arabinose-inducible promoter (58,63,64). These strains grew on media supplemented with arabinose but did not grow on media supplemented with glucose ( Fig. 2),   (58,63,64). Since expression of lgt, lsp, and lnt is essential for bacterial growth, these strains do not grow on medium containing glucose but do grow on medium containing arabinose. Introduction of plasmids containing the homologous H. pylori genes under the control of the lacUV5 or Trc promoter supported growth on glucose. Vector-only control plasmids (C) did not support growth on glucose. Results are representative of 3 experiments.
indicating that the regulated enzymes are essential for growth. We introduced plasmids encoding the predicted H. pylori homologs under the control of the lacUV5 or trc promoter into the E. coli strains in which expression of lgt, lspA, or lnt is controlled by the arabinose-inducible promoter. In contrast to the parental strains, the strains containing these plasmids grew on media supplemented with glucose ( Fig. 2). These results indicate that H. pylori lgt, lspA, and lnt homologs can functionally complement conditionally lethal E. coli lgt, lspA, and lnt mutant strains, respectively.
Lipoprotein synthesis: construction and analysis of H. pylori mutant strains. We next sought to determine whether the lgt, lspA, or lnt genes are essential in H. pylori. H. pylori strain 26695 was transformed with DNA constructs designed to disrupt lgt, lspA, or lnt by insertion of a gene conferring antibiotic resistance. Derivatives in which lnt was disrupted were readily isolated (e.g., VM211), indicating that lnt is not essential for H. pylori growth. In contrast, we were unable to isolate strains in which lgt or lspA was disrupted, despite repeated attempts.
As a complementary approach for assessing the essentiality of these genes, we constructed derivatives of H. pylori strain 26695 in which lgt or lspA was placed under the control of a TetR-regulated promoter (65). We then compared the growth of the resulting strains, VM176 and VM183, in the absence or presence of anhydrotetracycline (ATc), which results in repressed or derepressed expression of the gene of interest, respectively. Results indicated that strains containing TetR-regulated copies of lgt or lspA grew in the presence but not in the absence of ATc, indicating that lgt and lspA are essential for H. pylori growth (Fig. 3).
Lipoproteins represent pathogen-associated molecular patterns that can be recognized by TLR2 heterodimers on host cells. Triacylated lipoproteins are recognized by TLR1/TLR2 heterodimers, whereas diacylated lipoproteins are recognized by TLR2/TLR6 heterodimers. Upon binding of lipoproteins to TLR2 heterodimers (8), host cells respond by activation of NFB-responsive promoters, including promoters for genes encoding proinflammatory cytokines such as interleukin-8 (IL-8). To compare the TLR2-activating properties of wild-type (WT) and lnt mutant H. pylori, we used HEK293 cell lines stably transfected to express human TLR1/TLR2 or mouse TLR2/TLR6 heterodimers. Addition of lipoprotein-enriched protein extracts from WT H. pylori to TLR2-expressing cells elicited responses similar to the responses induced by the triacylated control peptide Pam3CSK4 (Fig. 4). In contrast, addition of lipoproteinenriched protein extracts prepared from lnt mutant VM211 elicited responses similar to the responses induced by the diacylated control peptide Pam2CSK4 (Fig. 4). These results are consistent with the presence of triacylated lipoproteins in WT H. pylori and the presence of diacylated lipoproteins in lnt-deficient H. pylori.
Lipoprotein localization genes: construction of H. pylori mutant strains. Of the five genes whose products direct lipoprotein localization in E. coli (lolABCDE), H. pylori appears to lack lolB (16,19), and a homolog of lolD has not been identified. One of numerous predicted ABC transporter-like ATP-binding proteins likely encodes LolD (53). H. pylori is predicted to encode a homolog of lolA (HP0785) as well as a single homolog of lolC or lolE (which has been termed lolF [HP0787]) to direct localization of newly Helicobacter pylori Lipoproteins ® synthesized lipoproteins (14). We sought to determine whether the lolA or lolF genes are essential in H. pylori. Repeated attempts to isolate strains in which either lolA or lolF was disrupted by insertion of a gene conferring antibiotic resistance were unsuccessful. As an alternative approach for assessing the essentiality of these genes, we constructed derivatives of strain 26695 in which lolA or lolF was placed under the control of a TetR-regulated promoter (65). The resulting strains, VM189 and VM191, were able to grow in the presence of ATc, but growth was arrested when the bacteria were cultured on media lacking ATc (Fig. 3). These results indicate that lolA and lolF are essential in H. pylori.
H. pylori Cag T4SS activity and posttranslational lipidation. Secretion of the oncoprotein CagA and other effector molecules by H. pylori is mediated through a type IV secretion system (Cag T4SS) (66). Several of the H. pylori proteins required for T4SS activity exhibit sequence relatedness to components of T4SSs in other bacterial species. These include CagT, a putative homolog of the lipoprotein VirB7 (67). CagT (32 kDa) is much larger in size than VirB7 proteins from most bacterial species (for example, Agrobacterium tumefaciens VirB7 is 6 kDa), and comparisons of CagT with putative VirB7 homologs showed a very low level of amino acid sequence relatedness (31). To evaluate whether CagT undergoes N-terminal modifications consistent with lipidation, we introduced a gene encoding CagT with a DDK epitope between amino acids 26 and 27 into an H. pylori strain harboring a deletion of the endogenous cagT gene (Fig. 5A). The resulting strain, BV357, expressed CagT-DDK and retained Cag T4SS function ( Fig. 5B and C). We next deleted lnt from the CagT-DDK strain, resulting in strain VM207. Protein extracts were prepared from BV357 and VM207, and the extracts were treated with enterokinase (which cleaves at the C-terminal end of the DDK epitope) (Fig. 5A). Immunoblotting the resulting protein preparations with anti-DDK antibody revealed the presence of DDK-reactive peptides in both BV357 and VM207, each with an apparent molecular mass somewhat higher than expected (predicted molecular weights for the triacylated and diacylated peptides are 2.5 and 2.2 kDa, respectively). Anomalous SDS-PAGE migration of lipoproteins is a common phenomenon (68)(69)(70)(71). The DDK-reactive peptide from lnt mutant strain VM207 has a lower molecular mass than the peptide from strain BV357, consistent with a failure to add the third acyl chain in the absence of lnt (Fig. 5). These results are consistent with the presence of an Lnt-dependent N-terminal posttranslational modification of CagT.
We next analyzed Cag T4SS function in wild-type H. pylori 26695 and the lnt mutant strain (VM211). Immunoblotting demonstrated that the steady-state levels of CagT were similar in 26695 and VM211 (Fig. 6A). Cocultured with AGS cells, both 26695 and VM211 were able to induce IL-8 production (a phenotype dependent on Cag T4SS activity), whereas a control strain lacking the cag PAI was not (Fig. 6B). Furthermore, both 26695 and VM211 were able to translocate the effector protein CagA into AGS cells, where the protein then became phosphorylated (Fig. 6C). These results suggest that there is little if any alteration of Cag T4SS activity when H. pylori lipoproteins (including CagT) are diacylated rather than triacylated.
To determine whether lipidation of CagT is required for Cag T4SS function, we constructed H. pylori strains expressing mutant forms of cagT in which the putative  CagT lipobox (a peptide motif within the signal peptide containing an essential cysteine as the site of lipidation) was disrupted (by introducing a C21S mutation, strain BV218) or restored (strain BV260) (Fig. 7). Analyses of steady-state levels of CagT by immunoblotting revealed that disruption of the putative lipobox led to destabilization of CagT whereas introduction of an intact lipobox after the C21S mutation (CagT1) resulted in wild-type levels of CagT (Fig. 7). Consistent with these findings, the C21S mutation abolished the ability of H. pylori to induce IL-8 expression in AGS gastric epithelial cells (Fig. 7).
One possible explanation for the results shown in Fig. 7 is that the CagT-C21S mutant protein may lack a functional signal peptide and therefore may not be properly secreted. Analysis of CagT by SignalP-5.0 predicts signal peptide cleavage by LspA between Ala20 and Cys21 (Fig. 8A) (72). In contrast, analysis of CagT-C21S predicts no signal peptide cleavage (data not shown). Therefore, as a complementary approach, we constructed an H. pylori strain, VM253, expressing a mutant protein (CagT2) in which the 20-amino-acid signal peptide of CagT was replaced with the 33-amino-acid signal peptide from the secreted protein VacA (Fig. 8A and B). This mutant CagT also includes a DDK epitope between amino acids 26 and 27. Both wild-type CagT and the CagT2 chimera are expected to undergo signal peptide cleavage immediately preceding a Cys residue (C21 in WT CagT, C34 in CagT2) ( Fig. 8A and B). However, the mutant CagT2 is not expected to undergo lipidation due to the absence of a lipobox (Fig. 8B). Immunoblotting of protein extracts from strain VM253 revealed that exchange of the native CagT signal peptide with the signal peptide from VacA led to destabilization of CagT (Fig. 8C). Correspondingly, strain VM253, expressing the mutant protein (CagT2), was defective in inducing IL-8 expression in AGS gastric epithelial cells (Fig. 8D). Therefore, the combined results of Fig. 7 and Fig. 8 indicate that an intact CagT lipobox is required for CagT stability and Cag T4SS activity.

DISCUSSION
Genes HP0955, HP0074, and HP0180 in H. pylori strain 26695 are annotated as homologs of lgt, lspA, and lnt, respectively (33,(52)(53)(54). Although there is very limited primary amino acid sequence similarity between the encoded H. pylori proteins and more thoroughly characterized homologs from other species, multiple amino acids predicted to be required for functional activity are conserved (Fig. 1). In the present study, we report that H. pylori HP0955, HP0074, and HP0180 are indeed functional homologs of lgt, lspA, and lnt, based on the ability of each gene to complement conditionally lethal E. coli mutant strains.
Results of previous studies have provided conflicting evidence on whether genes involved in lipoprotein synthesis and sorting in H. pylori are essential (54,55). The comprehensive analyses conducted in the present study indicated that the lgt and lspA genes encoding lipoprotein synthetic enzymes as well as the genes whose products direct lipoprotein localization, lolA and lolF, are essential in H. pylori. In contrast, lnt is not essential. The latter result is consistent with recent studies suggesting that lnt is not essential in bacterial species containing lolF in place of lolC and lolE (14,15). The affinity of the LolCDE complex is much higher toward triacylated lipoproteins than diacylated lipoproteins, and this likely explains the essential nature of lnt in species containing lolC and lolE (73). In contrast, the nonessential nature of lnt in species containing lolF suggests that the affinity of LolF toward diacylated lipoproteins (the product of Lgt action) is sufficient to direct them to LolA for sorting.
In the canonical sorting pathway, lipoproteins destined for localization to the outer membrane are transferred from the inner membrane LolCDE complex to the periplasmic LolA protein. LolA then transfers the lipoproteins to LolB anchored in the outer membrane (9). Previous studies have concluded that LolA receives lipoproteins from LolCDE but cannot insert lipoproteins into membranes (74)(75)(76)(77)(78), and similar studies have concluded that LolB can insert lipoproteins into membranes but cannot accept lipoproteins from LolCDE (9,78,79). There are several possible explanations for the essentiality of lolA in H. pylori and the lack of a recognizable H. pylori lolB (and for the similar apparent absence of lolB in the Alphaand other Epsilonproteobacteria [16,19]), including the following: (i) outer membrane Helicobacter pylori Lipoproteins ® lipoproteins are not essential in H. pylori; (ii) H. pylori LolA, unlike E. coli LolA, can insert lipoproteins into membranes; (iii) there is an uncharacterized protein in H. pylori with a function analogous to that of LolB; and/or (iv) H. pylori possesses a LolAB-bypass system, as has recently been suggested in E. coli (20).
H. pylori strains lacking CagT (either through cagT knockout mutation or with repressed cagT gene expression) are deficient in Cag T4SS activity (65,80,81), and we now present experimental evidence indicating that CagT undergoes N-terminal modification consistent with lipidation. Lipidation of CagT is necessary for CagT stability and therefore for Cag T4SS activity. Comparisons of Cag T4SS activities in wild-type and mutant H. pylori in which lipidation of CagT is disrupted suggest that CagT functions similarly whether it is triacylated or diacylated. However, disruption of CagT lipidation appears to destabilize CagT, most likely due to defects in secretion or membrane anchoring leading to degradation of the mislocalized protein.
Previous studies demonstrated that CagT is essential for Cag T4SS assembly and function (65,80,81). Analyses of the Cag T4SS structure by electron cryotomography (cryo-electron tomography [cryo-ET]) revealed that outer membrane core complexes from cagT mutant strains were highly variable in structure, generally consisting of a central ring but lacking the peripheral densities seen in intact complexes (82). A reduced number of core complexes were visualized by cryo-ET in a cagT mutant strain compared to a WT strain, and a reduced number of core complexes were purified from the cagT mutant strain (82,83); this might be attributable to reduced stability or impaired localization of the complexes in a cagT mutant strain. Single-particle cryoelectron microscopy analysis of Cag T4SS outer membrane core complexes suggested that the lipidated amino-terminal end of CagT is positioned to interact with the H. pylori outer membrane (Fig. 9) (31). Together, these results suggest that lipidated CagT helps stabilize the core complex and direct it or anchor it to the outer membrane.
In summary, this report provides a systemic experimental analysis of lipoprotein synthesis and localization pathways in H. pylori, a member of the Epsilonproteobacteria. The lipoprotein synthetic enzymes Lgt and LspA are essential, implying that one or more lipoproteins are essential for H. pylori. Lnt (which performs the final enzymatic step in the synthesis of triacylated lipoproteins in Gram-negative bacteria) is nonessential in H. pylori. This is consistent with previous reports indicating that Lnt is not required by certain members of the Alpha-, Beta-, and Gammaproteobacteria (11)(12)(13)(14)(15). The nonessential nature of Lnt in H. pylori is consistent with current understanding that Lnt is not required by bacteria (such as H. pylori) that produce LolF rather than LolC and LolE (14,15). Further differentiating lipoprotein sorting in H. pylori from model proteobacteria is the observation that H. pylori, like other members of the Epsilonproteobacteria as well as the Alphaproteobacteria, appears to lack LolB (16,19). The current study showed that lipidation of CagT is essential for Cag T4SS function, but we did not detect any substantial difference in T4SS activity dependent on whether CagT was diacylated or triacylated. In future studies, it will be important to determine what fitness advantage is provided by the nonessential lnt, to determine how CagT and other lipoproteins are directed to the H. pylori outer membrane in the absence of a lolB homolog, and to define the structural basis by which a lipid moiety in CagT helps direct or anchor the T4SS outer membrane core complex to the outer membrane.

MATERIALS AND METHODS
Bacterial strains and culture conditions. Strains and plasmids used in this study are listed in Table 1. H. pylori strain 26695 was grown on either Trypticase soy agar (TSA) plates containing 5% sheep blood or in bisulfite-free Brucella broth containing 10% fetal bovine serum (BB-FBS) at 37°C in room air supplemented with 5% CO 2 . H. pylori mutant strains were selected using metronidazole (15 g ml Ϫ1 ), kanamycin (12.5 g ml Ϫ1 ), or chloramphenicol (2.5 g ml Ϫ1 ). Anhydrotetracycline (ATc) was added as indicated at 100 ng ml Ϫ1 (65). Plasmids were maintained using E. coli strain DH5␣ or strain XL-10 Gold. E. coli strains were grown on Luria-Bertani (LB) agar plates or in LB broth containing ampicillin (100 g ml Ϫ1 ), chloramphenicol (25 g ml Ϫ1 ), or kanamycin (25 g ml Ϫ1 ).
Complementation of E. coli strains with H. pylori lgt, lspA, and lnt. E. coli strains PAP9403, YX238, and KA349 are conditionally lethal strains in which lgt, lspA, and lnt (respectively) are expressed under the control of arabinose-inducible promoters (58,63,64). Expression of the relevant gene in these strains is stimulated by the presence of arabinose and inhibited by the addition of glucose. The lgt, lspA, and lnt genes from H. pylori 26695 were subjected to PCR amplification and cloned into pBbA5a-RFP (pBbA5ared fluorescent protein), pBbE1c-RFP, and pBbA5c-RFP (Addgene), using EcoRI and BglII restriction sites, to yield pBV449, pMM690, and pMM691, respectively. These plasmids then were transformed into the relevant E. coli mutant strains. E. coli strains were cultured with isopropyl ␤-D-1-thiogalactopyranoside (IPTG) (Roche) (0.1 mM) and either arabinose (0.2%) or glucose (0.2%). Construction of recombinant H. pylori strains. H. pylori strain 26695 was transformed with synthetic constructs (GenScript) designed to delete lgt, lspA, lnt, lolA, or lolF. For lgt, the construct included a deletion of lgt (HP0955) and rdxA (HP0954); deletion of rdxA confers metronidazole resistance (84). For lspA, lnt, lolA, and lolF, the genes of interest were replaced with an antibiotic resistance determinant. We succeeded only in obtaining H. pylori transformants in which lnt was deleted, resulting in strain VM211.
To place lgt under TetR regulation, tetR linked to a kanamycin resistance determinant was inserted into the ureA locus of strain 26695 to yield strain VM165. Then, a synthetic construct (GenScript) in which a chloramphenicol resistance determinant and 3 copies of tetO linked to the promoter of cagUT (65) placed upstream of HP0955 (and changing the ATG start codon to a TTG start codon [85]) was introduced into the intergenic region between HP0630 and HP0631 to yield strain HP173. Finally, the endogenous lgt locus was deleted using the lgt deletion construct described above to yield strain VM176.
To place lspA, lolA, or lolF under TetR regulation, strain VM124 (which contains TetR in the ureA locus) (65) was transformed with synthetic constructs in which a kanamycin resistance determinant and 3 copies of tetO linked to the promoter of cagUT (65) were placed upstream of lspA, lolA, and lolF to yield strains VM183, VM189, and VM191, respectively. The start codons for lolA (ATG) and lspA (GTG) were changed to TTG (85); lolF naturally begins with a TTG start codon.
To delete cagT from H. pylori, a DNA sequence containing cagT along with 500 bp of flanking sequences was amplified from H. pylori and ligated into pGEMT-Easy (Promega) to yield pBV173. Using inverse PCR, cagT was excised to produce pBV175. H. pylori 26695ΔrdxA (cagT::cat/rdxA) (80) was then transformed with pBV175, and metronidazole-resistant transformants were selected to yield H. pylori BV199, which contains an unmarked ΔcagT mutation. The cagT mutation in BV199 was complemented in cis by transformation with pBV342 (ureA::cagT-CAT) to produce strain BV321.
To generate an H. pylori strain expressing DDK-tagged CagT, pBV173 was mutagenized to insert DDK epitope-encoding sequences after the codon for lysine 26 of cagT, yielding pBV334. Plasmid pBV334 was then used to transform H. pylori strain 26695 ΔrdxA cagT::cat/rdxA (80), and metronidazole-resistant transformants were selected to yield H. pylori strain BV357. The lgt locus was deleted from strain BV357 using the lgt deletion construct described above to yield strain VM207.
To generate an H. pylori strain expressing the 33-amino-acid signal peptide from H. pylori VacA fused to CagT at Cys21 and including a DDK epitope inserted after amino acid 26 of CagT, a chimeric sequence including a chloramphenicol resistance determinant was synthesized by GenScript and cloned into plasmid pAD1. The resulting plasmid, pMM692, was used to transform H. pylori strain BV199, and chloramphenicol-resistant colonies were selected. The resulting strain was designated VM253.
Analysis of CagT lipidation in H. pylori. To evaluate N-terminal modifications of CagT in H. pylori, protein extracts were prepared from strains BV357 and VM207 (described above). H. pylori strains were cultured on TSA plates containing 5% sheep blood for 24 h. Bacteria were resuspended in phosphatebuffered saline (PBS), pelleted, and washed in PBS. Bacterial pellets were resuspended in PBS containing 0.5% SDS. Insoluble material was pelleted at 21,000 ϫ g for 5 min. Proteins in the soluble fraction were recovered using methanol-chloroform and solubilized in 0.1% SDS (86). Protein concentrations were determined by microBCA assay (Pierce), and 50 g of each extract was incubated with 2 units of enterokinase (EK Max; ThermoFisher) at 37°C for 4 h. Samples then were separated by Tricine-SDS-PAGE using a 16%T/3%C gel containing 6 M urea (87) and transferred to polyvinylidene difluoride (PVDF). The PVDF membrane was developed using mouse monoclonal anti-DDK antibody (clone M2; Sigma) followed by anti-mouse horseradish peroxidase (HRP) (Promega) and chemiluminescence detection (Pierce).
TLR2 activation assay. Protein extracts enriched in lipoproteins were prepared using Triton X-114 (TX114) (88). H. pylori strains were cultured in broth for 20 h. Cultures were pelleted and washed in PBS, and the bacterial pellets were resuspended in PBS containing 2% TX114 and incubated at 4°C with continuous mixing for 24 h. Insoluble material was pelleted, and the protein extract was warmed and pelleted at 37°C to promote separation of TX114 and aqueous phases. The upper, aqueous phase was discarded, and the lower, detergent phase was extracted two times with the starting volume of PBS. Proteins in the final detergent phase were precipitated with 4ϫ volumes of Ϫ20°C acetone. The protein pellets were resuspended in PBS, and the protein concentrations were determined by microbicinchoninic acid (BCA) assay (Pierce).
Assays of H. pylori Cag T4SS activity. The ability of H. pylori to induce production of IL-8 when cocultured with gastric cells was determined as previously described (84). AGS gastric epithelial cells (ATCC CRL-1739) were grown in RPMI medium containing 25 mM HEPES and 10% FBS and were inoculated with H. pylori (multiplicity of infection [MOI] of 100) from liquid cultures that had been grown overnight. Cocultures were incubated for 4 h at 37°C with 5% CO 2 . Supernatant was collected, and IL-8 content was determined using an IL-8 ELISA (GenScript), according to the manufacturer's specifications.
The ability of H. pylori to translocate CagA under conditions of coculture with gastric cells was determined as previously described (89,90). H. pylori cells were cocultured with AGS cells for 7 h at 37°C with 5% CO 2 . Monolayers were washed in PBS and lysed in NP-40 lysis buffer containing Complete Mini protease inhibitors (Roche) and Phos-Stop phosphatase inhibitors (Roche). CagA translocation was assessed based on immunoblotting cell lysates with anti-CagA polyclonal antibody (Santa Cruz) and anti-phospho-tyrosine monoclonal antibody (pY-99; Santa Cruz).
Immunoblotting. Unless otherwise indicated, protein lysates were resolved by SDS-PAGE and proteins were transferred to nitrocellulose membranes. Membranes were blocked using PBS containing 0.1% Tween and 2% nonfat dry milk. Proteins were detected by incubating the membrane with primary antisera (diluted 1:5,000 to 1:10,000), followed by horseradish peroxidase-conjugated secondary antibody. Rabbit antiserum to H. pylori CagT and antiserum to HspB (a GroEL heat shock protein homolog) have been described previously (80,91). Anti-CagT antiserum was preabsorbed to H. pylori BV199 (26695 ΔrdxA ΔcagT) prior to immunoblotting. Signals were detected by the use of an enhanced chemiluminescence (ECL) methodology.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, PDF file, 0.1 MB.