Virulence properties of Campylobacter jejuni are enhanced by displaying a mycobacterial TlyA methylation pattern in its rRNA

Abstract Campylobacter jejuni is a bacterial pathogen that is generally acquired as a zoonotic infection from poultry and animals. Adhesion of C. jejuni to human colorectal epithelial cells is weakened after loss of its cj0588 gene. The Cj0588 protein belongs to the type I group of TlyA (TlyAI) enzymes, which 2′‐O‐methylate nucleotide C1920 in 23S rRNA. Slightly longer TlyAII versions of the methyltransferase are found in actinobacterial species including Mycobacterium tuberculosis, and methylate not only C1920 but also nucleotide C1409 in 16S rRNA. Loss of TlyA function attenuates virulence of both M. tuberculosis and C. jejuni. We show here that the traits impaired in C. jejuni null strains can be rescued by complementation not only with the original cj0588 (tlyA I ) but also with a mycobacterial tlyA II gene. There are, however, significant differences in the recombinant phenotypes. While cj0588 restores motility, biofilm formation, adhesion to and invasion of human epithelial cells and stimulation of IL‐8 production in a C. jejuni null strain, several of these properties are further enhanced by the mycobacterial tlyA II gene, in some cases to twice the original wild‐type level. These findings strongly suggest that subtle changes in rRNA modification patterns can affect protein synthesis in a manner that has serious consequences for bacterial pathogenicity.

Loss of TlyA in C. jejuni cells results in a wide range of defects including decreased ribosome subunit association, impeded motility and reduced biofilm formation, which collectively reduce virulence (Sałamaszy nska-Guz et al., 2018). In addition, sensitivity to the antibiotic capreomycin is altered. Complementation with natively folded variants of TlyA containing point mutations that abolish methyltransferase activity showed that all the physiological defects were caused by loss of rRNA methylation rather than absence of the protein itself (Sałamaszy nska- Guz et al., 2018). These findings indicate that TlyA influences the physiology and pathogenicity of C. jejuni solely through its rRNA methylation activity.
Studies on the mycobacterial TlyA II and mutant derivatives of this enzyme showed that both the C1409 and C1920 methylations contribute to capreomycin binding (Monshupanee et al., 2012). These nucleotides are respectively located on the interface of the small and large ribosomal subunits (Yusupov et al., 2001) at the extremities of the binding site for capreomycin and the related tuberactinomycin drug, viomycin (Stanley, Blaha, Grodzicki, Strickler, & Steitz, 2010). While the connection between TlyA-directed methylation and capreomycin/viomycin binding is immediately evident (Johansen et al., 2006), it remains less clear how the presence or absence of rRNA methylation would affect protein synthesis in a manner that alters pathogenic traits. Here, we address this question by equipping a tlyA-null strain of C. jejuni with either an authentic copy its own tlyA I gene (cj0588) or the mycobacterial tlyA II gene. The influence of the different methylation patterns in the C. jejuni recombinants are shown to be linked to a range of parameters including cell motility, biofilm formation, adhesion to human epithelial cells, cell invasion and the ability to elicit an innate immune response in the host cell.
2 | RESULTS 2.1 | Mycobacterial TlyA II specifically methylates two nucleotides in C. jejuni rRNAs The wild type tlyA II gene from M. smegmatis was introduced into the C. jejuni null strain 81-176Δcj0588. Screening the rRNAs from this F I G U R E 1 In vivo activity of mycobacterial TlyA II in C. jejuni. Gel autoradiograms of primer extensions on rRNA from C. jejuni strains. Extensions on 16S and 23S rRNAs from the mutant strain C. jejuni 81-176 Δcj0588 and from the same strain complemented with mycobacterial tlyA II (C. jejuni 81-176 Δcj0588::Myco_tlyA II ). Decreasing the dGTP concentrations (100, 10 and 1 μM, marked with wedges) intensifies reverse transcription termination at 16S rRNA C1409 and 23S rRNA C1920 when these nucleotides are 2 0 -O-methylated. Lanes C, U, A and G are dideoxysequencing reactions on unmodified C. jejuni rRNAs. Control lanes represent primers and reaction mixture without rRNA template recombinant by primer extension showed that expression of the mycobacterial TlyA II enzyme effectively modified nucleotide C1409 in C. jejuni 16S rRNA and C1920 in the 23S rRNA (E. coli rRNA numbering) ( Figure 1).
These modifications resulted in a concomitant increase in the sensitivity of C. jejuni to capreomycin. The MIC values for capreomycin in strains without a tlyA gene (81176Δcj0588) or with an inactivated version of the gene (81176Δcj0588::K188A) were consistently 64 μg/ml. The capreomycin MIC was lowered to 32 μg/ml by expression of the mycobacterial tlyA II gene that was the same value seen for cells expressing the original tlyA I gene cj0588 that methylates only at 23S rRNA nucleotide C1920.

| Motility and rRNA methylation
Loss or inactivation of its natural tlyA I gene results in decreased C. jejuni motility. Under the microaerobic conditions at 37 C F I G U R E 2 Motility of the C. jejuni strains. Agar plates with (a) WT-wild type strain, (b) Δcj0588, (c) Δcj0588::cj0588, (d) Δcj0588:: cj0588K188A and (e) Δcj0588::Myco_tlyA II strains of C. jejuni grown for 48 hr. The morphologies of the corresponding stains (including flagella) were visualised by Field Emission Scanning Electron Microscopy (FESEM). Strain motility is summarised in the histogram, where values represent the means ± SEM of three independent experiments measuring distances migrated over 48 hr. There was no significant difference for migration of the WT compared to Δcj0588::M.smeg_tlyA II , whereas significant differences were observed for Δcj0588 versus Δcj0588::M.smeg_tlyA II (p < .05), and for Δcj0588 versus Δcj0588::cj0588 (p < .05) employed here, C. jejuni null strains exhibit only one-third of the mobility of the wild-type ( Figure 2). Complementation with an active copy of cj0588 goes some way to restoring mobility to wild-type levels. Similarly, introduction of the mycobacterial tlyA II gene into the null strain partially rescues the cell's motility ( Figure 2). F I G U R E 3 Images of biofilm structures produced by C. jejuni (a) WT-wild type strain, (b) Δcj0588, (c) Δcj0588::cj0588, (d) Δcj0588:: cj0588K188A and (e) Δcj0588::Myco_tlyA II strains visualised by confocal laser microscopy. Relative biofilm depths are color-coded as shown 2.3 | Biofilm formation is influenced by the rRNA methylation pattern Biofilms were visualised using confocal laser microscopy producing three-dimensional images of these structures (Figures 3 and 4). Inactivation of the cj0588 gene reduces the cell's ability to form biofilms, and this effect is rescued by introduction of a functional copy of this gene. The null strain formed a thin biofilm that failed cover the whole surface and reached a depth of only 3.9 μm in its thickest region. Complementation with an active cj0588 gene fully restored biofilm density to 5.9 μm, comparable to that of the wild-type strain (5.2 μm). Surprisingly, transformation of the null strain with the mycobacterial tlyA II gene (forming the 81176Δcj0588::Myco_tlyA II strain) not only rescued the phenotype but supported uniform biofilm formation at a density of 7.5 μm, significantly surpassing that the wild-type strain ( Figure 4).

| Adhesion and invasion of the C. jejuni strains on Caco-2 cells
After inactivation of TlyA-directed methylation, the capacity of C. jejuni to adhere to the surface and to invade Caco-2 human colon epithelial cells was reduced to less than half that of the wild-type ( Figure 5). Both adhesion and internalisation were restored (to 93 and 103% wild-type levels, respectively) by complementation of the null F I G U R E 4 Analysis of the Figure 3 images showing the biofilm density produced by C. jejuni (a) WT-wild type strain, (b) Δcj0588, (c) Δcj0588::cj0588, (d) Δcj0588::cj0588K188A and (e) Δcj0588:: Myco_tlyA II strains. (f) Histogram summarising the biofilm data, colorcoded as in Figure 2. Experiments were carried out in triplicate and representative images are shown here. p < .001 for WT versus Δcj0588; p < .005 for WT versus Δcj0588::M.smeg_tlyA II F I G U R E 5 (a) Adhesion onto, and (b) invasion into Caco-2 cells by C. jejuni strains. Values represent means ± SEM of three independent experiments. Adhesion p < .05 for WT versus Δcj0588 and Δcj0588:: Myco_tlyA II ; invasion p < .005 for WT versus Δcj0588 and Δcj0588:: Myco_tlyA II strain with cj0588. This effect was more marked after complementation with the mycobacterial gene where the C. jejuni tlyA II recombinants became about one-third more adept than the wild-type at sticking to Caco-2 cells enabling the pathogen to invade the epithelial cells 25% more effectively ( Figure 5). The bacterial strains' ability to enter the Caco-2 cells was roughly proportional to their adhesive properties and thus the invasive index, which is the proportion of the surface-adhered bacteria that actually enter the eukaryotic cell, was fairly constant (i.e., varied less than 25%) for the different C. jejuni recombinants.
Subsequent to invasion, the virulence of the C. jejuni attack was inferred from the IL-8 response within the Caco-2 epithelial cell line.
Consistent with the adhesion/invasion data, the mildest reaction was seen with the cj0588-deletion strain and the inactive K188A variant, where IL-8 levels were barely above background ( Figure 6). The wildtype and complemented strains with an active cj0588 gene produced a clearer response, approximately doubling the amount of IL-8. The highest level of IL-8 was observed with C. jejuni expressing the mycobacterial tlyA II gene, reflecting the augmented adhesion/invasion properties of this strain in the Caco-2 cell model.

| Influence of the rRNA methylation pattern on bacterial survival in macrophages
Another important aspect of C. jejuni virulence is its ability to survive within host cells, and this was tested here using a RAW264.7 macrophage model system. While C. jejuni strains expressing functional TlyA proteins attached to and invaded the macrophages significantly more avidly than null strains, their survival within macrophages was not improved (Table 2), and all the C. jejuni strains within macrophages were killed by 48 hr.

| DISCUSSION
The natural version of the TlyA methyltransferase found in C. jejuni is a type I (TlyA I ) enzyme encoded by the cj0588 gene, and stoichiometrically methylates the 2´-O-ribose of 23S rRNA nucleotide C1920 (Sałamaszy nska- Guz et al., 2018). Type II (TlyA II ) variants, found in some Gram-positive bacteria including the Actinobacteria, methylate not only at nucleotide C1920 but also at 16S rRNA nucleotide C1409 (Johansen et al., 2006;Monshupanee et al., 2012). Both these nucleotides are effectively modified in the natural M. smegmatis host (Monshupanee et al., 2012), and the mycobacterial tlyA II gene product retains the same specificity when transferred to and expressed from the C. jejuni chromosome (Figure 1).
We have previously shown that a possession of an active TlyA I (Cj0588) methyltransferase is a prerequisite for C. jejuni to function  (Hyatt et al., 1994), while loss of the enzyme's function in H. pylori lowers adhesion to human gastric adenocarcinoma (AGS) cells and prevents colonisation of the gastric mucosa (Martino et al., 2001;Zhang et al., 2002). The TlyA II variant of this enzyme promotes survival of M. tuberculosis in macrophages (Rahman et al., 2015) and aids the binding of capreomycin to ribosomes (Maus et al., 2005). Several other endogenous rRNA methylations (reviewed in Purta, O'Connor, Bujnicki, & Douthwaite, 2009) have also been noted to promote ribosome-antibiotic interactions within bacterial pathogens (LaMarre, Howden, & Mankin, 2011;Sergeeva, Bogdanov, & Sergiev, 2015). In the present study, we demonstrate that the defective virulence traits exhibited in C. jejuni tlyA null strains can be rescued by a mycobacterial tlyA II ortholog, and that some phenotypic traits of the recombinant strain are distinctly different from the original wild-type and recombinants rescued with the original cj0588 (tlyA I ) gene.
The lower mobility of the C. jejuni null strain was restored to roughly the same extent by the wild-type C. jejuni tlyA I gene and the mycobacterial tlyA II gene ( Figure 2). However, the recombinants differed in their capacity to form biofilms. Complementation with tlyA I re-establishes wild-type levels, while cells transformed with tlyA II form significantly denser biofilms (Figures 3 and 4) M. tuberculosis cells lacking tlyA II become more susceptible to autophagy, and animals infected with this mutant strain exhibit increased immune response, reduced bacillary load and improved survival rates than when infected with wild-type bacilli (Rahman et al., 2015). Our findings here suggest that the mycobacterial tlyA II gene supports a comparable set of virulence traits in C. jejuni.
The ability of C. jejuni to attach to and invade human epithelial cells is central to its pathogenicity and was notably impaired by loss of TlyA-directed methylation ( Figure 5). Restoring TlyA I methylation by complementation of C. jejuni with cj0588 rescued its adhesion to and invasion of Caco-2 cells. Surprisingly, these features were not only rescued by the mycobacterial tlyA II but this recombinant clung to and entered the epithelial cells significantly more effectively than the original wild-type strain C. jejuni ( Figure 5).
When under attack by pathogenic bacteria, epithelial cells secrete chemotactic mediators (Eckmann, Kagnoff, & Fierer, 1993) and consistent with this, C. jejuni induces human-derived epithelial cell lines to release pro-inflammatory chemokines including the interleukin, IL-8 (Hickey, Baqar, Bourgeois, Ewing, & Guerry, 1999;Watson & Galan, 2005). In related studies of Campylobacter invasion, cytolethal distending toxin, outer membrane vesicles and the flagella activate the host cell's toll-like receptors to elicit secretion of IL-8 (Zheng, Meng, Zhao, Singh, & Song, 2008). The absence of tlyA activity is shown here to reduce the ability of C. jejuni to trigger the IL-8 innate immune response in Caco-2 cells ( Figure 6). Induction of IL-8 was restored by complementing null strains with an active tlyA gene, where the tlyA II gene produced the most marked stimulation increasing IL-8 production to twice that with wild-type C. jejuni.
These observations raise a number of questions. First, why a single ribose methylation (at 23S rRNA nucleotide C1920) would be a prerequisite for successful infection by C. jejuni and how an additional methylation (at 16S rRNA nucleotide C1409) would further improve its capacity to infect. The two TlyA II methylations are located approximately 20 Å apart on opposite sides of the ribosomal subunit interface and lie adjacent to the capreomycin binding site (Johansen et al., 2006). Both methylations have been shown to contribute individually to drug binding (Monshupanee et al., 2012). From the crystal structure of capreomycin-bound ribosomes (Stanley et al., 2010), the methylations are slightly too far apart to make contact with the drug. However, they are nevertheless positioned where they might lubricate the relative rotational movement of the subunits during translation (Yusupov et al., 2001), a process where one of the pivoted subunit conformations is favoured for drug binding (Ermolenko et al., 2007).
Each of the methylations thus contributes to ribosome function, and our working hypothesis (presently being tested) is that changes in the methylation pattern subtly skew the relative synthesis rates of different proteins in the bacterium, with this being ultimately reflected in altered virulence properties.
Another, and potentially more important, question is whether it would make a difference in the real world if C. jejuni were to attain both methylations through changes in its own tlyA gene or via transfer of a tlyA II ortholog from another bacterium. The ability of C. jejuni to adhere to epithelial cells is dependent on having a functional tlyA gene and is enhanced with a tlyA II -type gene, and these adhesive properties determine the degree of cell invasion ( Figure 5).
An additional aspect to be taken into consideration is that C. jejuni pathogenicity depends on its ability to survive subsequent to phagocytosis. On the one hand, the RAW 264.7 macrophage data (Table 2) show that significantly fewer C. jejuni survive when they lack an active cj0588 gene and that complementing the cells with an active copy of cj0588 or the mycobacterial tlyA II gene restores their initial survival rates. This observation is consistent with the role of tlyA II mentioned above, where it supports the survival within macrophages of its authentic host, M. tuberculosis (Rahman et al., 2015).
However, when extending the time frame of observations past the initial phagocytotic event, we find that the C. jejuni-tlyA II cells are no more resilient after 12 hours (and in fact appear slightly more frail) than strains expressing cj0588 (Table 2). In this case, the wild-type

| Bacterial strains
The C. jejuni strains used in this study (Table 1) were grown under microaerobic conditions (BD GasPak EZ CO2 sachets, Becton Dickinson) at 37 C on brain-heart infusion (BHI) agar containing 5% (v/v) sheep blood, and in some cases supplemented with chloramphenicol at 20 μg/ml and/or kanamycin at 30 μg/ml.

| Minimal inhibitory concentration (MIC) determination
Overnight cultures of the C. jejuni strains were diluted to a turbidity of 0.5 McFarland standard, and 3 μl were plated onto BHI agar plates with two-fold increases in the capreomycin concentration. The MIC values are the lowest concentration of antibiotic at which no growth was observed after incubation under microaerobic conditions for 48 hr at 37 C.

| Motility
The motility of C. jejuni cells was assessed by adding 3 μl of culture (OD 600 0.5) onto BHI with 0.25% agar. Plates were left to dry and were incubated under microaerobic conditions for 48 hr at 37 C before measuring cell migration.

| Biofilm assays
Three-dimension confocal microscope images of biofilms were produced from C. jejuni grown on glass slides (Millicell EZ, Milllipore). Strains were diluted in BHI broth to OD 600 0.05 before aliquoting to the Millicell dishes and incubating at 37 C for 48 hr. Broth was removed, and biofilms were washed twice with water and dried at 55 C for 15 min before staining with acridine orange solution (1 μg/ml) for 30 min and rinsing twice with PBS. Biofilms were visualised at an excitation wavelength of 490 nm using a Leica white laser scanning confocal microscope (Leica TCS SP8-WWL) with a 63× oil-immersion lens. Three-dimensional T A B L E 1 C. jejuni strains used in this study images were created from Z-stacks images collected from top down to obtain an overall view of the biofilm volume and converted to TIFF files with depth-coding using LAS X software (Leica Microsystems).

| Field Emission Scanning Electron Microscopy
Visualisation of cell morphology was carried out using field emission scanning electron microscopy (FESEM). C. jejuni cells were grown on Columbia agar plates for 24 hr before harvesting and suspending in 5 ml BHI broth (at OD600 of 0.05) and cultivating for 48 hr at 37 C in 5% CO 2 on glass cover slides. Cells were fixed for 24 hr in 0.

| Adhesion and invasion assays
Caco-2 epithelial cells, derived from a human colonic carcinoma, were seeded into a 24-well tissue culture dishes and grown overnight at 37 C to a cell density of 10 5 cells per well in Eagle's minimum essential medium containing Earle's salts, 2 mM L-glutamine, 10% fetal bovine serum, 0.1 mM nonessential amino acids and 1 mM sodium pyruvate in a 5% CO 2 (CO 2 incubator, Thermo Scientific).
The C. jejuni strains were added into the wells at a multiplicity of infection (MOI) of one hundred bacteria to one epithelial cell and incubated for 2 hr to allow adhesion and invasion of the Caco-2 cells.
The Caco-2 monolayers were washed three times with PBS to remove unattached bacteria. A portion of the Caco-2 cells was then lysed with 0.1% Triton X-100 to estimate the total complement of bacterial cells.
The remaining Caco-2 cells were incubated for a further 2 hr in modified minimal essential medium with 100 μg gentamicin ml −1 to kill extracellular bacteria, while retaining viable internalised bacteria. Bacteria adhering to and internalised by the Caco-2 cells were tallied by serial dilution in phosphate-buffered saline (PBS) and plating on BHI agar.
4.9 | Survival assay C. jejuni survival was quantified in RAW 264.7 macrophages cultured in RPMI medium with 10% fetal bovine serum at 37 C in 5% CO 2 atmosphere (CO 2 incubator, Thermo Scientific). Tissue culture trays (24-well) were seeded with 2 × 10 5 macrophages per ml and incubated for 24 hr prior to inoculating with C. jejuni at an approximate MOI of 100.
Infected macrophage monolayers were incubated for 2 hr before killing extracellular bacteria as described above. Surviving bacteria were monitored (as above) at 3, 6, 12, 24 and 48 hr post-infection.

| Innate immune response in epithelial cells
Production of interleukin of IL-8 by Caco-2 cells was taken as an indicator of the extent to which by C. jejuni strains provoked an innate immune response. Caco-2 cells were seeded in 24-well plates and infected with bacteria as described above. Cell supernatants were assayed after one day using a human IL-8 ELISA kit (Merck). Optical densities were measure with a microplate reader (Epoch spectrophotometer, BioTek Instruments) and normalised relative to negative controls (no C. jejuni cells) using the instrument supplier's software.