Development of tools for the genetic manipulation of Campylobacter and their application to the N-glycosylation system of Campylobacter hepaticus, an emerging pathogen of poultry

ABSTRACT Various species of campylobacters cause significant disease problems in both humans and animals. The continuing development of tools and methods for genetic and molecular manipulation of campylobacters enables the detailed study of bacterial virulence and disease pathogenesis. Campylobacter hepaticus is an emerging pathogen that causes spotty liver disease (SLD) in poultry. SLD has a significant economic and animal welfare impact as the disease results in elevated mortalities and significant decreases in egg production. Although potential virulence genes of C. hepaticus have been identified, they have not been further studied and characterized, as appropriate genetic tools and methods to transform and perform mutagenesis studies in C. hepaticus have not been available. In this study, the genetic manipulation of C. hepaticus is reported, with the development of novel plasmid vectors, methods for transformation, site-specific mutagenesis, and mutant complementation. These tools were used to delete the pglB gene, an oligosaccharyltransferase, a central enzyme of the N-glycosylation pathway, by allelic exchange. In the mutant strain, N-glycosylation was completely abolished. The tools and methods developed in this study represent innovative approaches that can be applied to further explore important virulence factors of C. hepaticus and other closely related Campylobacter species. IMPORTANCE Spotty liver disease (SLD) of layer chickens, caused by infection with Campylobacter hepaticus, is a significant economic and animal welfare burden on an important food production industry. Currently, SLD is controlled using antibiotics; however, alternative intervention methods are needed due to increased concerns associated with environmental contamination with antibiotics, and the development of antimicrobial resistance in many bacterial pathogens of humans and animals. This study has developed methods that have enabled the genetic manipulation of C. hepaticus. To validate the methods, the pglB gene was inactivated by allelic exchange to produce a C. hepaticus strain that could no longer N-glycosylate proteins. Subsequently, the mutation was complemented by reintroduction of the gene in trans, on a plasmid vector, to demonstrate that the phenotypic changes noted were caused by the mutation of the targeted gene. The tools developed enable ongoing studies to understand other virulence mechanisms of this important emerging pathogen.

and represents a significant economic and animal welfare threat to the global poultry industry (1)(2)(3)(4)(5)(6).Layer birds infected with C. hepaticus, especially free-range hens, can develop numerous hepatic lesions during the peak time of lay (6).SLD can increase flock mortality rates by up to 15% (7)(8)(9) and decrease egg production by up to 35% (10).More recently, a second novel Campylobacter species, Campylobacter bilis, was isolated from the bile of SLD-positive birds (11) and proven to be an additional cause of SLD (12).The pathogenic mechanisms of C. hepaticus and C. bilis remain largely unknown.Therefore, there are currently no effective, targeted interventions against SLD, and prevention and treatment of the disease rely on strict biosecurity measures, which are generally inadequate.The use of antibiotics can be effective (13); however, there are public health concerns associated with antibiotic resistance.In addition, tetracycline-resistant C. hepaticus strains have recently been isolated from clinical cases of SLD (13,14).Therefore, alternative methods to control SLD are needed.
Like the closely related species Campylobacter jejuni and Campylobacter coli, C. hepaticus colonizes the gastrointestinal tract of chickens (5).However, it is thought that an important mechanism of disease is the ability of C. hepaticus to traffic from the gastrointestinal tract to the gall bladder, where it is found in the bile and can induce disease in the liver (13).Transcriptomic analysis of C. hepaticus isolated from the bile of SLD positive birds has provided insights into genes, which may be of importance for survival and colonization and hence pathogenesis (14).Several genes predicted to be associated with niche adaptation and virulence were identified, including genes responsible for chemotaxis, motility, lipooligosaccharide synthesis, and metabolism (14).However, phenotypic analysis of these putative virulence mechanisms remains lacking due to inadequate genetic tools to perform mutagenesis studies.Therefore, the precise roles these genes and their products play in the development of SLD have not yet been determined.
Despite reported difficulties in cloning Campylobacter DNA and transforming Campylobacter (15)(16)(17), genetic tools, including shuttle vectors (17)(18)(19)(20)(21) and suicide vectors (22)(23)(24)(25)(26), have been paramount in understanding the mechanisms of pathogen esis in the closely related species C. jejuni, an important human pathogen (27).The successful development of vectors has primarily relied on the utilization of native plasmid DNA elements from C. jejuni or closely related species.C. jejuni tends to be more efficiently transformed with DNA isolated from the same or closely related species and, in some cases, cannot be transformed with DNA isolated from Escherichia coli (17,19,28).
Several groups have engineered E. coli-C.jejuni shuttle vectors (17)(18)(19)(20)(21), usually derivatives of the E. coli-Campylobacter shuttle vector, pILL550 (18).This shuttle vector incorporated the Campylobacter plasmid replication functions, including an origin of replication and replication proteins, from a high copy number C. coli cryptic plasmid.The transformation efficiencies of these vectors vary significantly between C. jejuni strains and between different laboratories (29).Thus, groups have characterized cryptic plasmids from C. jejuni clinical isolates to develop new expression vectors (29,30).For example, one study could not successfully transform pILL550 into several strains of Campylobacter; hence, they isolated and characterized a C. jejuni cryptic plasmid and developed a shuttle vector utilizing the cryptic plasmids replication elements (30).This has proven a reliable means to produce new E. coli-Campylobacter shuttle vectors.
The functionality of these shuttle vectors as tools to genetically complement-inacti vated chromosomal genes in C. jejuni has played crucial roles in the elucidation of key virulence factors, including the N-glycosylation of proteins.The N-glycosylation system in C. jejuni is encoded by a protein glycosylation (pgl) gene locus comprising genes responsible for the biosynthesis and transfer of a conserved heptasaccharide N-glycan onto several proteins (31)(32)(33)(34).Disruption of the N-glycosylation pathway in C. jejuni negatively impacts proteome stability and other important enzymatic and cellular pathways leading to significant changes in cell physiology (35,36).Importantly, this reduces the ability of C. jejuni to adhere to and invade human intestinal cells (37), and C. jejuni completely lacking N-glycoproteins is unable to colonize the gastrointestinal tract of its natural chicken host (35).Recently, C. hepaticus was shown to possess an N-glycosylation system (Fig. 1), much like C. jejuni, which is suspected to play key roles in host colonization, invasion, and pathogenicity (38).However, the further investigation of this system's function was hampered by the lack of genetic manipulation tools and methods for C. hepaticus.
This study aimed to develop genetic tools and methods and reports the genetic manipulation of C. hepaticus by the construction of a gene knockout mutant.E. coli-C.hepaticus shuttle vectors have been constructed and used to develop a transformation method.These newly developed methods and vectors were used to delete a gene by allelic exchange and complement the mutation.These tools were used to investigate the impacts of pglB deletion on the ability of C. hepaticus HV10 T to N-glycosylate proteins.

Sequencing of C. coli cryptic plasmids pWestmill 11, pCC311, and pCC388
To obtain replication elements that could be utilized by C. hepaticus to functionally maintain and replicate plasmid DNA, multiple C. hepaticus, C. jejuni, and C. coli strains were screened for low molecular weight plasmids (<10 kb).While no C. hepaticus and C. jejuni strains contained small plasmids (not shown), several C. coli strains contained low molecular weight plasmids (Fig. 2).
Gel extracted plasmid DNAs were sequenced and analyzed for each of these strains.Analysis of DNA sequences revealed that C. coli 311, C. coli 388, and C. coli Westmill 11 all contained at least two plasmids.Blastn analysis of the nucleotide sequence of each plasmid revealed varying levels of sequence similarity to previously identified C. coli plasmids (Table 1.) The percentage query cover, sequence identity, and accession number of previously identified C. coli plasmids were obtained by performing a blastn of sequenced and assembled plasmids presented in column 2 of the table.
Plasmid DNA sequences from each strain were further analyzed in SnapGene (Dotmatics), and the amino acid sequence of open reading frames (ORFs) were determined to identify putative replication proteins.The 2,744 bp plasmid (pCC311) from C. coli 311, the 3,303 bp plasmid (pCC388) from C. coli 388, and the 1,271 bp plasmid (pWestmill 11) from C. coli Westmill 11 were chosen for further analysis due to their small size.
The smallest of the three plasmids, pWestmill 11 isolated from C. coli Westmill 11, possessed an A+T rich region lacking tandem, iteron repeats upstream of a single coding sequence corresponding to a replication/maintenance protein, RepL (WP_002799729.1) (Fig. 3).This plasmid is 36 bp smaller than a previously identified plasmid, pCC2228-1 (40), and the same size as an unnamed plasmid (CP082876).In contrast to pCC311 and pCC388, which are likely members of the theta-replicating, repAB iteron containing plasmids, pWestmill 11 is suspected to replicate via a rolling-circle mechanism (40).
Analysis of pCC311 from C. coli 311 revealed four ORFs and an A+T rich region possessing five direct tandem repeats of 22 bp (Fig. 3), characteristic of an origin of replication (iterons nt 10-110) (41).The first ORF directly downstream of this region encodes a 26 amino acid protein with 100% identity with plasmid replication protein A (ARE81508.1).The second ORF encodes a 344 amino acid protein with 98.54% identity to a RepB family plasmid replication protein (EAL2766289.1).ORF-3 and ORF-4, further downstream, encode 78 and 107 amino acid C. coli hypothetical proteins WP_201458541.1 and WP_002806461.1,respectively.Plasmid pCC311 is a newly characterized plasmid, different from any previously reported plasmids.Interestingly, it is smaller than other theta-replicating C. coli cryptic plasmids as it lacks an ORF encoding a mobilization protein typically found in other theta-replicating C. coli cryptic plasmids, including pCCT1 (X82079) and p3384 (42), but instead encodes two hypotheti cal proteins.
Analysis of pCC388 from C. coli 388 revealed three open reading frames and an A+T rich region possessing four direct tandem repeats of 22 bp (Fig. 3).The first ORF  encodes a 345 amino acid protein with 100% identity to a replication initiation protein (ECK8531972.1).The second ORF encodes a 90 amino acid protein with 100% identity to a C. coli hypothetical protein (EAI7291999.1).The third ORF, immediately downstream of ORF-2, encodes a 424 amino acid protein with 99.53% identity to a C. coli mobilization protein (EAL7355230.1).The plasmid pCC388 is the same size as a previously identified C. coli cryptic plasmid, pCC2228-2 (40).

Construction of vectors pJBM1, pJBM2, and pJBM3
To construct plasmid pJBM1 (Fig. 4), pWestmill 11 isolated from C. coli Westmill 11 was linearized and cloned into pMW2, a plasmid that can replicate in E. coli and contains a kanamycin resistance gene [aph(3')-IIIa] that can be expressed in both Campylobacter and E. coli (18).To construct the C. hepaticus-E.coli shuttle vector pJBM2 (Fig. 4), the cryptic plasmid pCC311 carrying the predicted OriV and sequences encoding RepA, RepB, and the two hypothetical proteins was linearized and cloned into pMW2.To construct pJBM3, the cryptic plasmid pCC388 carrying the predicted OriV and sequences encoding replication initiation protein, mobilization, and the hypothetical protein was linearized and cloned into pMW2 (Fig. 4).These shuttle vectors contain multiple cloning site regions present in the pMW2 backbone (highlighted in gray) (Fig. 4).

Transformation of Campylobacter with vectors pJBM1, pJBM2, and pJBM3
To investigate the functionality of the shuttle vectors, pJBM1, pJBM2, and pJBM3 were introduced into C. hepaticus HV10 T and the respective strains (C. coli Westmill 11, C. coli 311, and C. coli 388) from which the small cryptic plasmids included in the shuttle vectors were isolated.No recoverable transformants were obtained when C. hepaticus HV10 T was transformed with pJBM1, but C. coli Westmill 11 was successfully transformed with the plasmid.Shuttle vector pJBM2 replicated and was maintained in C. hepaticus HV10 T and C. coli 311.Shuttle vector pJBM3 replicated and was maintained in C. hepaticus HV10 T and C. coli 388.Shuttle vectors pJBM2 and pJBM3 were re-isolated from C. hepaticus HV10 T , transformed into E. coli and a restriction enzyme diagnostic digest of each vector was performed (Fig. S1) to unequivocally show that each shuttle vector could be maintained and manipulated in E. coli and C. hepaticus HV10 T .In the above-described initial experiments, the C. hepaticus HV10 T type strain was shown to be transformable, but it was not known if this was a general characteristic common to other C. hepaticus strains.Therefore, the electro-competency of 11 different C. hepaticus strains were tested by transformation with the shuttle vector pJBM3 isolated from E. coli NEB 5-alpha and five with plasmid DNA isolated from C. hepaticus HV10 T .Transformation efficiencies varied considerably between C. hepaticus strains (Table 2).The strain C. hepaticus HV10 T was the most efficient in taking up plasmid DNA compared with the other 10 strains tested (Table 2).The transformation rates obtained with plasmid DNA isolated from C. hepaticus and E. coli were similar for the five strains tested with both (Table 2).Additionally, pJBM2 and pJBM3 were introduced into C. bilis VicNov18 T to determine if the shuttle vectors developed for C. hepaticus could also function in C. bilis.A diagnostic digest of pDNA isolated from C. bilis VicNov18 T transformants and a C. bilis-specific PCR (Fig. S2) confirmed both shuttle vectors were functional in C. bilis and that C. bilis was the species transformed, respectively.Thus, these vectors also serve as tools to manipulate this species.
Transformation efficiencies of >1 × 10 4 CFU/µg of pJBM3 into C. hepaticus HV10 T were obtained across six independent experiments.Thus, this most highly transformable strain was used for site-specific mutagenesis studies.

In vitro stability of shuttle vectors
To further investigate the functionality of the shuttle vectors, the stability of pJBM2 and pJBM3 replication and maintenance in the absence of antibiotic selection was quantified.After ~50 generations (five subcultures), 100% of cross-patched colonies grew on both selective and non-selective media for both plasmids.These results indicated that the plasmids remained stable through multiple generations without antibiotic selection.

Site-specific mutagenesis of pglB and complementation in trans
To determine if C. hepaticus is amenable to mutagenesis by homologous recombination, a suicide vector, pCH_pglB_SV (Fig. 5), was constructed and introduced into C. hepaticus HV10 T to delete the oligosaccharyltransferase-encoding gene, pglB.The suicide vector was introduced into C. hepaticus HV10 T in two independent transformation reactions and produced 11 and 14 kanamycin-resistant transformants, respectively.Three and six mutants, respectively, were screened from each reaction to determine if mutants had undergone a double-crossover recombination event by PCR (Fig. S3).Any mutants produced by a single-cross over recombination event were discarded, for example, M1 and M2 (Fig. S3).The suicide vector was also introduced into C. hepaticus NSW44L to determine if mutagenesis could be achieved in more than one C. hepaticus strain.Four C. hepaticus NSW44L were screened as described above (Fig. S4A).
To confirm that deletion of the pglB gene resulted in the loss of N-glycosylated proteins, the binding of soybean agglutinin (SBA) lectin to GalNAc residues of the N-linked heptasaccharide glycan of multiple C. hepaticus HV10 T ∆pglB::kan mutant whole cell lysates was investigated and compared with wild-type (WT) C. hepaticus HV10 T (Fig. 6).The same phenotypic analysis was performed to compare four C. hepaticus NSW44L∆pglB::kan mutants with WT C. hepaticus NSW44L (Fig. S4B and C).The SBA binding profiles were significantly reduced in C. hepaticus HV10 T and C. hepaticus NSW44L mutant strains compared with their respective WT strains, indicating that the deletion of pglB had disrupted the glycosylation pathway and greatly reduced the synthesis of N-glycoproteins (Fig. 6B; Fig. S6C).Furthermore, these results demonstrated that mutagenesis could be performed in more than one C. hepaticus strain.Interestingly, C. hepaticus HV10 T ∆pglB::kan mutant 7 produced a unique banding pattern where the 15 kDa band was completely absent, whereas this phenotype was not observed in the other pglB mutants screened (Fig. 6B).Thus, mutant 7 was chosen for complementation work because it showed the greatest decrease in the SBA binding profile to N-glycans present in whole cell lysates.Sequence analysis of C. hepaticus HV10 T ∆pglB::kan mutant 7 (JBEFTY000000000: Scaffold 0) showed that aph(3')-IIIa had integrated by a double crossover recombination event and deleted pglB, as depicted in Fig. 5. Sequence analysis using the raw sequence reads (NCBI SRA SRX24788994) in Snippy identified five off-site single nucleotide polymorphisms (SNPs) in the genome.Manual genome interrogation revealed in four out of five cases that the SNPs corresponded to single base deletions in homopolymer regions.Therefore, these changes are likely to be artifacts of the sequencing process known as post-homopolymer errors (43).The fifth SNP was identified as a substitution of cystine with adenine at position 549332 in the reference genome (NZ_CP031611.1),resulting in an amino acid change from arginine to leucine in a gene encoding enoyl-ACP reductase.These enzymes are responsible for the biosynthesis of bacterial fatty acids (44), and importantly, a mutation in this gene is unlikely to affect the C. hepaticus N-glycosylation system and impact the phenotype.
To complement the C. hepaticus HV10 T ∆pglB::kan mutant 7, pJBM3 was used to construct a series of vectors to restore the N-glycosylation of proteins in C. hepaticus HV10 T ∆pglB::kan mutant 7. First, pJBM4 (Fig. S5) was constructed to serve as an empty vector control for using pJBM3 as a backbone for complementation of C. hepaticus HV10 T ∆pglB::kan mutant 7.This was achieved by replacing aph(3')-IIIa with the tet(O) gene obtained from the pCJDM210L-like plasmid isolated from C. hepaticus 84B.Next, pJBM5.1, which carried a single copy of pglB under the control of the putative galE promoter (Fig. S6) was constructed and introduced into C. hepaticus HV10 T ∆pglB::kan mutant 7 but did not restore the N-glycosylation of proteins in this strain (Fig. S7).This was likely due to polar effects caused by the mutagenesis process impacting the transcription and expression of genes downstream of pglB.Attempts to complement another mutant strain (M3) with a single copy of pglB were unsuccessful (data not shown), suggesting polar effects in that mutant too.Thus, pJBM5.2,expressing the pgl locus genes including and downstream of pglB (Fig. 7), was designed and introduced into C. hepaticus HV10 T ∆pglB::kan mutant 7. Complementation of the mutant strain was assessed based on the restoration of SBA binding to N-glycans present in whole cell lysates from complemented mutants compared with WT C. hepaticus HV10 T (Fig. 8) using western blotting and an enzyme linked immunosorbent assay.
Introduction of pJBM5.2into C. hepaticus HV10 T ∆pglB::kan mutant 7 partially complemented SBA binding to N-glycans (Fig. 8) in four clones across two independently derived isogenic complements.Six distinct bands between 25 and 100 kDa were restored to WT levels in the complemented clones transformed with pJBM5.2 (Fig. 8) that were absent in the mutant strain and the mutant strain transformed with the empty vector control, pJBM4.Notably, a dominant band at 15 kDa present in C. hepaticus HV10 T , suspected to correspond to lipooligosaccharide (LOS) (38), could not be restored in the complemented mutant strain.
The results of the ELISA measurement of the SBA binding to the heptasaccharide glycan support the results observed in the lectin blot (Fig. 8A).SBA binding to the heptasaccharide present in whole cell lysates of C. hepaticus HV10 T ∆pglB::kan mutant 7 and C. hepaticus HV10 T ∆pglB::kan mutant 7(pJBM4) was greatly reduced to levels comparable with STM1 (lacks N-glycosylation) compared with WT and the complemen ted mutant strain (Fig. 8B).Notably, the levels of SBA binding were significantly higher (approximately two times) in the complemented mutant strain compared with WT.

DISCUSSION
This study reports the successful genetic manipulation of C. hepaticus by site-specific mutagenesis via allelic exchange.The construction of E. coli-C.hepaticus shuttle vector plasmids enabled the development of a DNA electro-transformation method, which was effective for introducing plasmid DNA into C. hepaticus and C. bilis and was of  S1.
sufficient efficiency to recover double recombination events (knockout mutants) after transformation with a non-replicating suicide vector.
A series of E. coli-C.hepaticus shuttle vectors have been developed utilizing the replication machinery from two small cryptic plasmids from two different C. coli isolates.The cryptic plasmids used in the current study are theta-replicating, iteron-contain ing plasmids that belong to the same incompatibility group as previously described Campylobacter shuttle vectors (18,20,30).The cryptic plasmid pCC311 is at least 456 bp smaller than other previously characterized C. coli theta-replicating cryptic plasmids.The size of shuttle vectors pJBM2 and pJBM3 could be further reduced by excision of the ampR gene, which is non-functional in Campylobacter.This could reduce the size of the shuttle vector by approximately 1,000 bp.Another putative E. coli-C.hepaticus shuttle vector, pJBM1, utilizing a 1,271 bp cryptic plasmid previously characterized and suspected to replicate by a rolling-circle mechanism (40), was constructed but could not replicate in C. hepaticus.This is likely due to the repL gene being insufficient for plasmid replication and maintenance in C. hepaticus HV10 T .
The shuttle vector plasmids pJBM2 and pJBM3 were functional in C. coli 311 and C. coli 388, respectively, and both in C. bilis VicNov18 T , demonstrating their versatility and use for gene expression in other Campylobacter species.The plasmids were also stably maintained without antibiotic selection in C. hepaticus HV10 T .This was not surprising as Miller and others (45) previously demonstrated >95% plasmid stability of C. jejuni shuttle vectors developed using cryptic plasmid elements as replication machinery after a similar number of generations without antibiotics.The number of generations in the current study (50) was chosen due to the need to grow C. hepaticus on solid-growth medium plates rather than in liquid culture.Assessment of plasmid stability could not be achieved using liquid culture due to the fastidious nature of C. hepaticus.The stability of pJBM2 and pJBM3 highlights their usefulness as tools for the genetic complementation of C. hepaticus HV10 T mutant strains for future in vivo studies.
Transformation efficiencies varied between C. hepaticus strains, which was expected as this has been observed for C. jejuni, where some strains are highly electrocompetent and others cannot be transformed with plasmid DNA by electroporation (16,17,28,30).To our surprise, transformation efficiencies of C. hepaticus transformed with plasmid DNA isolated from E. coli did not differ significantly from C. hepaticus transformed with plasmid DNA isolated from itself.The higher number of transformants observed in C. hepaticus strains transformed with plasmid DNA isolated from E. coli compared with plasmid DNA isolated from C. hepaticus may be due to higher quality plasmid DNA preparations from E. coli.Previous studies have shown that the source of plasmid DNA has a significant impact on the efficiency of transformation whereby most C. jejuni strains cannot be transformed with plasmid DNA isolated from E. coli, or efficiencies of transformation are reduced by four orders of magnitude due to a DNA restriction/modi fication system responsible for restricting heterologous DNA uptake (17,28).Holt and others showed that C. jejuni NCTC 11168 possesses a gene, cj1051c (cjeI), encoding a type II restriction and methyl transferase enzyme, which significantly decreases transforma tion efficiencies with plasmid DNA (46).Furthermore, a mutation in cjeI permitted transformation with plasmids isolated from an E. coli host (46).In contrast, the current study suggests no DNA restriction barrier between C. hepaticus and E. coli.A blastp search of cjeI restriction-modification enzyme against C. hepaticus HV10 T produced no significant sequence similarity, suggesting this enzyme is absent in C. hepaticus HV10 T , which may modestly explain this phenomenon.
Next, the feasibility of gene deletion and replacement via homologous recombination for the mutagenesis of C. hepaticus HV10 T and C. hepaticus NSW44L was demonstrated by producing deletion mutants of pglB.It was recently shown that C. hepaticus possesses an N-glycosylation system that modifies proteins with a heptasaccharide glycan (38).Most of the glycosylated proteins identified were highly conserved glycoproteins, also present in C. jejuni, that play crucial roles in host colonization (47,48).Some glycoproteins were found to be unique to C. hepaticus, which may contribute to its ability to cause SLD.Mutations in C. jejuni pgl genes have profound adverse effects on proteome stability, growth, cellular function, its virulence in humans, and host colonization (47)(48)(49).Thus, pglB, the oligosaccharyltransferase that is essential for the N-glycosylation of proteins in C. jejuni (31,50), was chosen as a target for gene deletion in C. hepaticus HV10 T .
The pglB gene was successfully mutated in C. hepaticus HV10 T and C. hepaticus NSW44L via allelic exchange using a suicide vector lacking a contra-selective marker.The strain NSW44L was chosen in addition to the C. hepaticus type strain (HV10 T ) as this strain was previously shown to be significantly more virulent compared with the type strain and another C. hepaticus strain (51).The ability to perform mutagenesis in a highly virulent strain is desirable so that the phenotypic effects of knockouts of various genes encoding potential virulence factors can be assessed in animal studies.Homologous DNA regions of approximately 1,000 bp were used in the suicide vector developed to induce allelic exchange, as increasing the average length of homologous DNA regions on suicide vectors has been shown to improve recombination efficiency in C. jejuni (17).Phenotypic analysis of SBA binding profiles to whole cell lysates of C. hepaticus pglB mutants using a lectin, SBA, which binds to the terminal GalNac residues of the heptasaccharide present on glycoproteins (38,52), showed that the pglB gene is critical for generating N-glycoproteins in C. hepaticus HV10 T and C. hepaticus NSW44L.The SBA reactive bands present in the lectin blots of whole cell lysates of the mutant 7 strain (Fig. 6 and 8) are likely due to non-specific binding of the lectin.This is further supported by the quantitative measurement of SBA binding to the heptasaccharide, which demonstrated that S. Typhimurium STM1 levels were comparable with the empty vector control mutant strain.S. Typhimurium lacks an N-glycosylation system, indicating that the signals seen are likely due to non-specific reactions and, more importantly, suggests that deletion of pglB completely abolished the production of N-glycoproteins in C. hepaticus HV10 T .Additional unexpected reactivity of SBA lectin to whole cell extracts of other Campylobacter species has previously been attributed to non-specific interac tions (32).These data confirm the essential role of pglB in the glycosylation reactions as the suspected oligosaccharyltransferase.
To demonstrate the utility of the newly developed shuttle vectors, the C. hepaticus HV10 T ∆pglB::kan mutant 7 was complemented in trans to restore glycosylation activity.Initial attempts to complement by expressing only the native pglB gene under the control of the C. hepaticus HV10 T galE putative promoter on pJBM3 were unsuccessful.It was hypothesized that this failure to complement was likely due to polar effects caused by the deletion of pglB.A previous study showed that deletion of pglB in C. jejuni resulted in decreased abundance of other essential enzymes encoded by pgl genes downstream of pglB, including PglF, which is responsible for the biosynthesis of bacillosamine, a rate-limiting step for the N-glycosylation pathway (36).The abundance of enzymes encoded by genes upstream of pglB was unaffected (36), supporting the rationale for the construction of a complementation vector containing pglB and the pgl genes downstream of pglB.The pglB gene and all genes downstream of pglB in the pgl locus, excluding pglG, were cloned into pJBM5 under the control of the galE promoter to counteract any of these polar effects.The gene, pglG, was excluded from the vector to reduce the size of the complementation vector as a previous study had shown that mutagenesis of pglG had no impact on SBA reactivity to C. jejuni whole cell lysates and likely plays no role in glycan biosynthesis (39).The complementation vector, pJBM5.2,partially restored the production of N-glycosylated proteins in the pglB mutant strain.Interestingly, quantitative analysis of SBA binding to the heptasaccharide glycan in whole-cell lysates revealed that the relative abundance of the heptasaccharide in the complemented strain was significantly greater than WT levels.It is hypothesized that this was due to the presence of multiple DNA copies (in trans on the multi-copy number plasmid) of the galE promoter and all genes downstream of pglB, resulting in higher than WT levels of gene expression.For example, the enzymes involved in the early stages of heptasaccharide synthesis, including PglA, responsible for adding the first GalNAc residue of the heptasaccharide, may be expressed at amounts greater than WT, thus elevating the levels of heptasaccharide produced, which may not be bound to protein but could be present as additional free oligosaccharide.
Notably, the SBA lectin blot phenotype could not be completely reverted to WT levels in the complemented strain.A dominant band present on the SBA lectin blot at 15 kDa for the WT strain has previously been shown to be non-proteinaceous and likely corresponds to C. hepaticus HV10 T LOS (38), which could not be complemented in the pglB mutant strain.The reasons why this band was impacted during mutagenesis and why it could not be complemented in this particular mutant strain remain unknown.A previous study demonstrated pglB mutagenesis in C. jejuni results in a significant decrease in isoprenyl transferase, UppS, a crucial enzyme involved in the synthesis of undecaprenyl diphosphate, which is an essential lipid carrier that plays key roles in the biosynthesis of LOS in C. jejuni (35).This would explain the loss of SBA binding to the 15 kDa band in the C. hepaticus HV10 T ∆pglB::kan mutant 7 strain.Therefore, the expression of UppS to increase the abundance of Und-P may be needed to restore the synthesis of LOS and SBA binding to the 15 kDa band.The disruption of LOS produc tion may have implications on cell physiology and morphology, which warrants further investigation.
In summary, the newly developed capability to produce targeted mutations in C. hepaticus has provided the first direct evidence of pglB involvement in the N-glycosyla tion of C. hepaticus HV10 T proteins.To confirm that N-glycosylation in C. hepaticus HV10 T is an essential pathogenic determinant of SLD, future research should utilize the tools, methods, pglB mutant strains, and complemented strains developed in this study to investigate if C. hepaticus lacking pglB can cause SLD in the poultry model of disease.The tools developed could also be used to improve our understanding of the direct and indirect effects of disrupting the N-glycosylation system on proteome stability and other critical cellular pathways linked to virulence of C. hepaticus and other Campylobacter species.The new genetic manipulation technologies developed in this study will aid in the study of C. hepaticus virulence and SLD pathogenesis and provide a pathway for the development of effective intervention strategies against SLD that can reduce the economic and animal welfare impact of SLD.

Bacterial strains, media, and growth conditions
The bacterial strains and plasmids used in this study are described in Table 3.The Campylobacter strains were grown on Brucella agar (Amyl Media) with 5% horse blood (HBA) plates and incubated at 42°C under microaerophilic conditions (85% N 2 , 10% CO 2 , and 5% O 2 ) using CampyGen gas packs (Oxoid) for 24-72 h.E. coli NEB 5-alpha (New England Biolabs) was used for cloning and plasmid propagation.E. coli and S. Typhi murium were grown in Luria-Bertani (LB) (BD Difco) broth or medium at 37°C over night.When appropriate, HBA and LB were supplemented with antibiotics; kanamycin (30 µg/mL for Campylobacter and 50 µg/mL for E. coli), ampicillin (100 µg/mL) for E. coli, and tetracycline (10 µg/mL) for E. coli and Campylobacter.

Plasmid and genomic DNA isolation and preparation
Plasmid DNA was isolated from C. coli and C. hepaticus strains grown on HBA plates for 24 h and then harvested by flooding plates with 2 mL Brucella broth and gently scraping cells off with a plastic spreader bar.The cells were transferred to a tube, centrifuged, and the cell pellet collected for DNA extraction.Plasmid DNA isolated from E. coli was obtained from 5 mL overnight cultures.All plasmid preparations were performed using Monarch Plasmid Miniprep Kit (NEB) according to manufacturer's instructions.Genomic DNA was isolated from C. hepaticus HV10 T WT and mutant strains, grown for 24 h using a Monarch Genomic DNA Purification Kit (NEB) according to the manufactur er's instructions.Plasmid and genomic DNA was quantified using either a Qubit 1× dsDNA Quantitation, high sensitivity, or Broad range assay kit (Invitrogen).The quality of plasmid and genomic DNA was determined by agarose gel electrophoresis and analyz ing A260/280 absorbance ratios using a NanoDrop spectrophotometer (Thermo Fisher).

Sequence analysis of C. coli cryptic plasmids and C. hepaticus pglB mutant strains
Plasmid DNA and genomic DNA sequencing libraries were prepared using the Illumina DNA Library Prep (M) Tagmentation (24 samples, IPB) kit according to the manufacturer's protocol.Libraries were sequenced on a MiSeq machine (Illumina) using v3 reagents with 2 × 300 bp paired end reads according to the manufacturer's instructions.Sequencing reads were analyzed by FastQC version 0.11.9 (55), and any reads with quality scores of <Q30 were trimmed using Trimmomatic version 0.36.6 (56).Plasmidomes were assembled using plasmidSPAdes Galaxy Version 3.15.4+galaxy2(57), or untreated reads were assembled with Unicycler v0.5.0 (58).All software used for analysis was accessed via Galaxy Australia (59).Assembled plasmid sequences were first used as queries in blastn (60) searches to identify significant sequence similarities to other previously reported Campylobacter plasmids.Sequences were uploaded to SnapGene version 7.2.(Dotmatics) and manually annotated.The amino acid sequence of ORFs present on the plasmids was determined by blastx to identify plasmid replicons.Plasmid sequences were manually analyzed to identify any A+T rich regions containing direct tandem repeats characteristic of a replication origin.
The C. hepaticus HV10 T ∆pglB::kan mutant genomes were assembled using the A5-MiSeq pipeline version 20150522 (61).Snippy version 4.6.0,accessed via Galaxy Australia was used to identify SNPs in the paired-end reads of mutant strains aligned against the reference genome (GenBank accession no.CP031611.1).Any SNPs identified by Snippy were manually interrogated by blasting assembled mutant strain contigs against C. hepaticus HV10 T reference genome to confirm the presence of any SNPs.

Construction of E. coli-C. hepaticus shuttle vectors
Vector pJBM1; a 1,413 bp XhoI fragment was amplified, using primers 1 and primer 2, from pWestmill 11.The PCR product was cloned into the XhoI site of pMW2, which had been linearized with XhoI (NEB) and treated with Quick CIP (NEB).The vector was introduced into E. coli JM109 (Promega) by heat shock according to manufacturer's instructions.Shuttle vectors pJBM2, pJBM3, and pJBM4 were constructed by Gibson assembly (62).Gibson assembly primers (Table S2) were designed using SnapGene (Dotmatics).Briefly, to construct shuttle vectors pJBM2 and pJBM3, pMW2 and the cryptic plasmids pCC311 and pCC388, containing the origin of replication and different replicons, were linearized by PCR using Q5 High-Fidelity DNA polymerase (NEB) and the respective primer pairs (Table S2).All PCR reactions for cloning were analyzed by agarose gel (0.8%-1%) electrophoresis and either cleaned up using a Monarch PCR & DNA Cleanup Kit (NEB) or the amplicons were extracted from the gel using a Monarch DNA Gel Extraction Kit (NEB) and quantified using a Qubit 1× dsDNA Quantitation, high sensitivity assay kit (Invitrogen).Vector fragments were then assembled using The assembled vectors were transformed into E. coli.Transformants were selected on LB kan plates; pJBM1, pJBM2, and pJBM3 were reisolated from clones, and diagnostic digests were performed (Fig. S1).To construct the shuttle vector pJBM4, pJBM3 was used as a backbone and linearized to remove the kanamycin gene [aph(3')-IIIa], which was replaced with tet(O) and its putative promoter, obtained from pCJDM210L like plasmid, native to C. hepaticus 84B.These amplicons were obtained by PCR using Q5 High-Fidelity DNA polymerase (NEB) and the respective primer pairs (Table S2.) and analyzed as described above.The assembled vector was transformed into E. coli as above and transformants selected on LB tet plates.Shuttle vectors pJBM2, pJBM3, and pJBM4 were reisolated from E. coli cells and sequenced as described above.

Transformation of Campylobacter hepaticus
Electrocompetent cells were prepared as described in (28) with some modifications.Briefly, frozen stocks of C. hepaticus strains were grown for 72 h at 42°C under microaer ophilic conditions.Cells were subcultured onto two separate HBA plates and grown for an additional 20-24 h.Plates were chilled in the fridge for 15 min, and each plate was flooded with 2 mL of ice-cold wash buffer (272 mM sucrose, 15% glycerol), and cells were harvested by gentle scraping.Cells were pelleted at 5,000 × g for 10 min at 4°C and resuspended in 1 mL of ice-cold wash buffer.Cells were pelleted and washed twice in ice-cold wash buffer.Each cell pellet was resuspended in 125 µL of wash buffer, resulting in a cell density of approximately 10 10 -10 11 CFU/mL as determined by plating serial dilutions of cells.Electroporation cuvettes (2 mm) were cooled on ice.Plasmid DNA was added to 50 µL of electrocompetent cells and incubated for 10 min on ice.Samples were electroporated using the parameters 2.5 kV, 200 ohms, 25 uF producing a time constant of 4.6-4.8ms.Cuvettes were immediately flushed with 100 µL of Super Optimal broth with Catabolite repression, 2% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl 2 , 10 mM MgSO 4 , and 20 mM glucose (S.O.C) medium (Invitrogen) and recovered on non-selective HBA plates for 6 h at 42°C under microaerophilic conditions.Cells were harvested from recovery plates by flooding with 1 mL Brucella broth and pelleted at 5,000 × g for 3 min.Cells were resuspended in 100 µL of Brucella broth and spread plated onto HBA supplemented with the appropriate antibiotic.Cells were grown for 4-7 days under microaerophilic conditions.All other Campylobacter species used in this study were transformed following the same procedure.

In vitro plasmid stability assay
To measure the stability of the shuttle vectors, C. hepaticus HV10 T transformed with pJBM2 and pJBM3 were resuscitated, streaked, and grown on HBA kan plates under microaerophilic conditions at 42°C for 72 h to ensure plasmid carriage.Individual colonies were picked and resuspended in 200 µL of Brucella broth, and the OD 600 was measured using a UV-spectrophotometer (Eppendorf ).After incubation, 100 µL of these bacterial suspensions were spread-plated onto non-selective HBA plates and grown under microaerophilic conditions at 42°C for 48 h.Cells were harvested using a 10 µL loop, resuspended in 200 µL of Brucella broth, and the OD 600 was measured.The change in OD 600 was used to determine the number of generations/doublings the cultures had undergone.After 50 generations, serial dilutions of cultures were performed in triplicates, and 100 µL of the appropriate dilutions were plated and grown for 72 h on non-selective HBA plates as described above.A total of 100 colonies from each replicate were cross-patched onto selective and non-selective HBA plates and grown for 24 h to determine the percentage of kanamycin-resistant CFU among the total number of CFU.The assay was performed in three independent biological replicates using technical triplicates.

Site-specific mutagenesis and complementation of pglB in C. hepaticus HV10 T
Construction of C. hepaticus HV10 T ∆pglB::kan mutants was achieved by allelic exchange using the suicide vector, pCH_pglB_SV, constructed using a four-fragment Gibson assembly.The flanking regions on the left (1,143 bp) and right side (1,160 bp) of the pglB gene were amplified using Q5 High-Fidelity DNA polymerase and the oligonucleotide primer pairs 17/18 and 21/22, respectively.These regions were cloned on their respective sides of the aph(3')-IIIa gene amplified using primer pair 19/20 and assembled into the linearized pMW2 backbone using primer pair 15/16.The assembled vector was electroporated into E. coli and colonies selected on LB kan plates.Plasmid DNA was reisolated from E. coli transformants, and large quantities of the plasmid were prepared using a Qiagen Plasmid Midi Kit and 3.7 µg of the final suicide vector was electrotrans formed into C. hepaticus HV10 T and C. hepaticus NSW44L as described above.
Transformants were screened for Integration of the aph(3')-IIIa gene into the pglB locus by double homologous recombination using pgl locus and kan-specific primer pairs 31/32 and 33/34 (Table S2) and sequence analysis.The pglB complementation vectors pJBM5.1 (Fig. S6) and pJBM5.2(Fig. 7) were constructed using a two and three-fragment Gibson assembly, respectively.To construct pJBM5.1 the putative galE promoter was cloned into the shuttle vector pJBM5.The galE promoter was predicted using BPROM (63) and amplified using primer pair primer 37/38.The shuttle vector pJBM5 was linearized using the primer pair 35/36 and the fragments assembled.To construct pJBM5.2, the putative galE promoter, and a large DNA fragment carrying pglB, pglA, pglC, pglD, pglE, and pglF were cloned into the shuttle vector pJBM5.The putative galE promoter was amplified using primer pair 41/42.The pgl genes were amplified using primer pair 43/44.The shuttle vector pJBM5 was linearized using the primer pair primer 39/40, and the fragments were assembled.The assembled vectors were transformed into E. coli as per manufacturer's instructions and colonies selected on LB tet plates.The complementation vectors pJBM5.1 and pJBM5.2 were introduced into C. hepaticus HV10 T ∆pglB::kan mutant 7 strain by electroporation, and tetracycline-resistant transform ants were selected.

Whole cell lysates for ELISA and western blotting
Freshly grown cells grown under microaerophilic conditions at 42°C for 24-48 h were harvested and washed twice in phosphate buffered saline (PBS).Cells were pelleted by centrifugation at 5,000 × g for 5 min, and cell pellets were resuspended in 97.5 µL PBS, 2.5 µL lysozyme (100 mg/mL), and 100 µL of tissue lysis buffer (Monarch DNA Gel Extraction Kit, NEB).Samples were vortexed and heated at 37°C for 5 min.Samples underwent centrifugation at 5,000 × g for 5 min to remove any insoluble material, and the supernatants were collected.Protein quantification was determined using a Qubit Protein Assay kit: Q33211 (Invitrogen).Protein samples were stored at −20°C.Lectin blotting was performed using SBA (Vector Laboratories) as described in (38).

Quantification of SBA binding to N-glycans in pglB mutant strains
A 96-well plate ELISA assay was developed to quantify and compare the abundance of heptasaccharide glycan in C. hepaticus HV10 T WT with C. hepaticus HV10 T pglB mutant strain and the C. hepaticus HV10 T complemented mutant strain.A 96-well High Bind Microplate (Corning) was coated with 50 µL of whole cell lysates containing 200 µg/mL% protein diluted in 0.5 M carbonate buffer from C. hepaticus HV10 T , C. hepaticus HV10 T ∆pglB::kan M7, C. hepaticus HV10 T ∆pglB::kan M7(pJBM4), C. hepaticus HV10 T ∆pglB::kan M7(pJBM5.2),STM1 (negative control), and 0.5 M carbonate buffer as a blank control.The plate was incubated at RT for 1.5 h.Unbound protein was removed, and the plate was blocked at 4°C overnight with 200 µL of 5% BSA.After discarding the blocking solution, the plate was probed with 100 µL of SBA (2 µg/mL) diluted in PBS-T (Tween 20 ) (0.1%) with 1% BSA and incubated for 1.5 h at RT.The lectin solution was removed, and wells were washed three times with 300 µL of PBS-T; 100 µL of Streptavidin-HRP (1:7,000) (Invitrogen) in PBS-T with 1% BSA was added to wells, and the plate was incubated at RT for 1 h.The secondary solution was discarded, and wells were washed three times with 300 µL of PBS-T.Any remaining solution was removed from wells, and the plate was developed with 50 µL of 3,3′,5,5′-tetramethylbenzidine substrate.Plates were incubated at RT in the dark for 15 min, and 50 µL of 1 M hydro chloric acid was added to each well to stop the reaction.The optical density at 450 nm was measured using a plate reader with background corrections to each sample.The assay was performed in technical quadruplicates.

Statistical analysis
One-way analysis of variances with Tukey's post hoc test was used to compare SBA binding with the heptasaccharide glycan present in whole cell lysates across the different C. hepaticus HV10 T strains.Statistical analysis was performed using GraphPad Prism version 10.1.2.324 (Dotmatics).

FIG 3
FIG 3 Plasmid maps of the sequenced C. coli Westmill 11 (green ORF), C. coli 311 (orange ORFs), and C. coli 388 (black ORFs) cryptic plasmids used to construct the shuttle vectors pJBM1, pJBM2, and pJBM3, respectively.The location and orientation of predicted replication elements (Rep), the origin of replication (OriV), and genes encoding mobilization proteins (Mob) and hypothetical proteins (HP) are indicated.Direct tandem repeats are represented by gray boxes positioned under the feature labeled OriV.

FIG 4
FIG 4 Construction of the shuttle vectors pJBM1, pJBM2, and pJBM3.Genes encoding replication proteins, hypothetical protein (HP), and mobilization protein (Mob), and the origin of replications obtained from pWestmill 11, pCC311, and pCC388 are highlighted in green, orange, and black, respectively.The remaining sequence of the vectors contains the E. coli origin of replication (ori), the kanamycin [aph(3')-IIIa], ampicillin-resistance (ampR) genes, and multiple cloning sites (MCS) regions from pMW2.The plasmid maps have been generated from the sequenced shuttle vectors.

FIG 5
FIG 5 Schematic diagram of the construction of pglB mutants by homologous recombination.Relevant chromosomal genes are shown.The diagram illustrates the process by which the wild-type (WT) pglB gene from C. hepaticus HV10 T was deleted by replacement with the aph(3')-IIIa on pCH_pglB_SV by a double crossover recombination event utilizing 1,103 and 1,120 bp homologous left and right flanking regions (blue) of pglB, respectively.The primers used to screen for double crossover events are provided in purple on the C. hepaticus HV10 T ∆pglB::kan chromosome.

TABLE 2
Efficiency of electro-transformation of C. hepaticus and C. bilis strains with shuttle vector pJBM3 isolated from E. coli NEB 5-alpha or C. hepaticus HV10 T a C.

hepaticus strain Maximum transformation efficiency (CFU/μg) using pDNA isolated from:
a Values represent the maximum transformation efficiencies obtained from at least two independent experiments.b Nt, not tested.

TABLE 3
Bacterial strains and plasmids used in this study

TABLE 3
Bacterial strains and plasmids used in this study(Continued)