Polar Glycosylated and Lateral Non-Glycosylated Flagella from Aeromonas hydrophila Strain AH-1 (Serotype O11)

Polar and but not lateral flagellin proteins from Aeromonas hydrophila strain AH-1 (serotype O11) were found to be glycosylated. Top-down mass spectrometry studies of purified polar flagellins suggested the presence of a 403 Da glycan of mass. Bottom-up mass spectrometry studies showed the polar flagellin peptides to be modified with 403 Da glycans in O-linkage. The MS fragmentation pattern of this putative glycan was similar to that of pseudaminic acid derivative. Mutants lacking the biosynthesis of pseudaminic acid (pseB and pseI homologues) were unable to produce polar flagella but no changes were observed in lateral flagella by post-transcriptional regulation of the flagellin. Complementation was achieved by reintroduction of the wild-type pseB and pseI. We compared two pathogenic features (adhesion to eukaryotic cells and biofilm production) between the wild-type strain and two kinds of mutants: mutants lacking polar flagella glycosylation and lacking the O11-antigen lipopolysaccharide (LPS) but with unaltered polar flagella glycosylation. Results suggest that polar flagella glycosylation is extremely important for A. hydrophila AH-1 adhesion to Hep-2 cells and biofilm formation. In addition, we show the importance of the polar flagella glycosylation for immune stimulation of IL-8 production via toll-“like” receptor 5 (TLR5).


Introduction
Mesophilic Aeromonas spp. strains are important pathogens of humans and lower vertebrates, including amphibians, reptiles, and fish [1]. Infections produced by these strains in humans can be classified into two major groups: noninvasive disease such as gastroenteritis, and systemic illnesses [2]. Strains from Aeromonas hydrophila, A. veronii biovar veronii, or sobria are described as virulent for humans [3] and fish [4]; these strains are serologically related by their O-antigen lipopolysaccharide (LPS) (serotype O11).
This has a known chemical structure containing O-polysaccharide chains of homogeneous chain length [5]. In addition, strains express a crystalline surface array protein with a molecular weight of ca. 52,000, which forms the S layer that lies peripheral to the cell wall [6]. The strains from this serotype are the most frequently isolated from septicemia caused by mesophilic Aeromonas spp. [2].

Results
The A. hydrohila AH-1 strain belongs to serotype O11 and is able to produce an S-layer [19]. This strain is motile in liquid medium (swimming) through expression of a polar flagellum and exhibits swarming behaviors in semisolid media through expression of lateral flagella ( Figure 1). DNA probes from polar flagella region 2 of A. hydrophila AH-3 [20] allowed identification of a clone from a cosmid genomic library of A. hydrophila AH-1. This clone allowed the DNA sequence of complete region 2 [21] of AH-1 to be obtained. This DNA sequence also allowed the isolation of a non-polar flagellated mutant AH-1∆FlaB-J; this mutant was unable to swim, but able to swarm, suggesting expression of lateral flagella only. Subsequently, this mutant was used to isolate lateral flagella on this strain. DNA probes from the lateral flagella cluster of A. hydrophila AH-3 [22] allowed identification of a clone from the same cosmid genomic library of A. hydrophila AH-1 with the partial lateral flagella cluster of this strain. The partial DNA sequence of the AH-1 lateral flagella cluster allowed us to identify two lateral flagellins (LafA1 and A2) in this strain. This compares with a single lateral flagellin observed in strain AH-3 [22]. immune stimulation of IL-8 production via toll-"like" receptor 5 (TLR5) was also evaluated.

Mass Spectrometry Analyses of Wild-Type Lateral and Polar Flagellins
Both polar and lateral flagellins were purified ( Figure 2) and their intact mass profiles analyzed using LC-MS.
Int. J. Mol. Sci. 2015, 16, page-page DNA probes from polar flagella region 2 of A. hydrophila AH-3 [20] allowed identification of a clone from a cosmid genomic library of A. hydrophila AH-1. This clone allowed the DNA sequence of complete region 2 [21] of AH-1 to be obtained. This DNA sequence also allowed the isolation of a non-polar flagellated mutant AH-1ΔFlaB-J; this mutant was unable to swim, but able to swarm, suggesting expression of lateral flagella only. Subsequently, this mutant was used to isolate lateral flagella on this strain. DNA probes from the lateral flagella cluster of A. hydrophila AH-3 [22] allowed identification of a clone from the same cosmid genomic library of A. hydrophila AH-1 with the partial lateral flagella cluster of this strain. The partial DNA sequence of the AH-1 lateral flagella cluster allowed us to identify two lateral flagellins (LafA1 and A2) in this strain. This compares with a single lateral flagellin observed in strain AH-3 [22].
The MS spectrum of a purified lateral flagellin showed a multiply charged ion envelope, as shown in Figure 3a The MS spectrum of a purified lateral flagellin showed a multiply charged ion envelope, as shown in Figure 3a, typical of protein. The spectrum was deconvoluted ( Figure 3b) and showed two protein masses of 30,402 and 30,267 Da.  These masses corresponded almost exactly to the predicted masses of the lateral flagellin proteins, LafA1 (30,401 Da) and LafA2 (30,297 Da). In addition, nLC-MS/MS analyses of tryptic digests of polar flagellins showed 42% and 45% sequence coverage for LafA1 and LafA2, respectively. Manual inspection of the MS/MS data showed no evidence of glycan-related oxonium ions in peptide spectra. Taken together, these data strongly suggested that the flagellins were not post-translationally modified with carbohydrates, as had been reported with lateral flagellins of A. hydrophila AH-3 [23].
LC-MS analysis of purified polar flagellins produced a characteristic multiply charged protein ion envelope (Figure 4a). When deconvoluted, a major mass of 34,947 Da was observed (Figure 4b).
Int. J. Mol. Sci. 2015, 16, page-page These masses corresponded almost exactly to the predicted masses of the lateral flagellin proteins, LafA1 (30,401 Da) and LafA2 (30,297 Da). In addition, nLC-MS/MS analyses of tryptic digests of polar flagellins showed 42% and 45% sequence coverage for LafA1 and LafA2, respectively. Manual inspection of the MS/MS data showed no evidence of glycan-related oxonium ions in peptide spectra. Taken together, these data strongly suggested that the flagellins were not post-translationally modified with carbohydrates, as had been reported with lateral flagellins of A. hydrophila AH-3 [23].
LC-MS analysis of purified polar flagellins produced a characteristic multiply charged protein ion envelope (Figure 4a). When deconvoluted, a major mass of 34,947 Da was observed (Figure 4b). To further investigate the nature of the protein modification, a tandem MS experiment was carried out on one multiply charged protein ion (m/z = 999.4). The resulting MS/MS spectrum was dominated by an intense ion at m/z = 404.4. An MS3 experiment on the ion observed at m/z = 404.4 produced an MS3 which showed consecutive losses of water and a methyl group from the parent ion ( Figure 5). To further investigate the nature of the protein modification, a tandem MS experiment was carried out on one multiply charged protein ion (m/z = 999.4). The resulting MS/MS spectrum was dominated by an intense ion at m/z = 404.4. An MS3 experiment on the ion observed at m/z = 404.4 produced an MS3 which showed consecutive losses of water and a methyl group from the parent ion ( Figure 5).  The MS/MS spectrum did not contain any recognizable peptide-related ions, and the fragmentation pattern strongly suggested that this moiety was a glycan. Of note, many sugar fragment ions, observed at m/z = 134.1, 162.1, 180.13, 221.19, 281.1 (denoted with an asterisk in Figure 5), were also observed in the MS/MS spectrum of the flagellin-modifying sugar Pse5Ac7Ac9Ac, found in A. caviae [24]. Other fragment ions, observed at m/z = 342.4, 355.4 and 373.2, were a single m/z unit different from fragment ions observed in the MS/MS spectrum of Pse5Ac7Ac9Ac ( Figure 4). From this data, the top-ranked plausible elemental formula was C 19 H 32 O 9 . These data, combined with the accurate mass analyses, suggest the base sugar is a pseudaminic acid-like sugar, with putative additions of two methyl groups and two molecules of water, and an unknown mass of 25 Da. Detailed structural analyses using NMR will be required to confirm this suggestion and the complete structure of this putative sugar moiety.
regions of FlaA and FlaB (except SISGIAK). The absence of asparagine residues on all sequenced glycopeptides suggests that the glycan is attached via O-linkage to serine or threonine residues. 5 this data, the top-ranked plausible elemental formula was C19H32O9. These data, combined with the accurate mass analyses, suggest the base sugar is a pseudaminic acid-like sugar, with putative additions of two methyl groups and two molecules of water, and an unknown mass of 25 Da. Detailed structural analyses using NMR will be required to confirm this suggestion and the complete structure of this putative sugar moiety.

Tandem Mass Spectrometry Analyses of Proteolytic Digests of Polar Flagellin Proteins
nLC-MS/MS analyses of tryptic digests of polar flagellin preparation showed peptide sequences corresponding to FlaA and FlaB proteins giving 35% and 36% sequence coverage, respectively ( Figure 6).

Putative Pseudaminic Acid Biosynthetic Mutants
Due to the suggestion of a putative pseudaminic acid-like glycosylation of AH-1 polar flagella, genes in the biosynthetic pathway of this sugar were mutated. Using oligonucleotides 5 1 -TCCAGAAGGTTATCGCACT-3 1 and 5 1 -GATGCTGGGAGCTATTACG-3 1 with genomic DNA from strain AH-1, an internal DNA fragment (500 bp) corresponding to a pseB homologue was amplified [25]. Genome walking allowed the completion of the DNA sequence of pseB-C homologues in strain AH-1. The same approach using oligonucleotides 5 1 -CCTATACCGCTGACACC AT-3 1 and 5 1 -TCACCACTTTTTCCTGACC-3 1 was used to amplify an internal DNA fragment (679 bp) of pseI homologue [25], and completely sequence gene pseG-I homologues in this strain. Subsequently, in frame mutants AH-1∆pseB and AH-1∆pseI were constructed. Using TEM, these mutants were shown to be unable to produce polar flagellum but lateral flagella was unaffected under induced conditions by TEM (Figure 1 data shown for AH-1∆pseI). The introduction of the Aeromonas wild-type corresponding genes recovered the production of polar flagella in the mutants (data not shown).

Adhesion to HEp-2 Cells and Biofilm Formation
The role of polar flagella glycosylation, lateral flagella, and O11-antigen LPS in the adherence to eukaryotic cells was investigated. The adhesion of several mutants to cultured monolayers of HEp-2 cells was observed. The AH-1∆rmlB mutant lacking O11-antigen LPS [18], but with expression of either polar or lateral flagella under induced conditions by TEM (Figure 1), showed no changes in flagellin molecular weight (Figure 2). Bacterial motility, when compared to wild type, was also unaffected, as was bacterial adherence to HEp-2 cells (Table 1).
By contract, mutants ∆pseB or I that lacked expression of flagella (polar and lateral for their grown in broth) showed a minimal adhesion values with a 86% reduction compared with the wild type. The same mutants expressed lateral flagella when grown in solid media, and showed an increase in their rate of adherence to Hep-2 cells ( Table 1). The AH-1∆rmlB mutant lacking the O11-antigen LPS showed comparatively less reduction in adhesion (29%) compared with the reduced adhesion observed in the flagella mutants. In all the cases, complementation of mutants with the wild-type gene/s showed recovery of the wild-type adhesion values ( Table 1). The results obtained in previous adhesion studies prompted us to study the biofilm production from the wild type and different mutant strains in a microtiter assay ( Table 2). The wild-type strain grown on liquid media (no lateral flagella produced) showed biofilm formation ability with an average OD 570 value of 1.43. Mutants lacking flagella grown in liquid or solid media (+ or´lateral flagella) are unable to form biofilms, with values <0.1, out of the range of detection for the assay ( Table 2). The AH-1∆rmlB mutant lacking the O11-antigen LPS showed an approximately 50% reduction in the ability to produce biofilm (0.78 average value versus 1.43 for wild type). In all cases, biofilm formations of the mutants were fully rescued by the introduction of the wild-type genes ( Table 2). Mutants' strains with only the plasmid vector alone show no differences in both studies. Table 2. Biofilm values of several A. hydrophila AH-1 serotype O11 strains using the method of Pratt and Kolter as indicated in the Experimental Section. Values presented are mean˘SD from three independent experiments carried out in triplicate (n = 9).
The degree of TLR5 activation was studied through the production of IL-8 using the HEK293 cell line, which was stably transfected with TLR5. HEK293-hTLR5 cells stimulated with purified A. hydrophila AH-1polar flagellins (wild type) showed varying levels of IL-8 production in agreement with the amount of flagellin used (Figure 7). However, HEK293-hTLR5 cells were stimulated with identical amounts of A. hydrophila AH-1 polar flagellin FlaB unglycosylated monomer obtained in E. coli, and the amount of IL-8 production was reduced (60%).

Discussion
In the current study, we demonstrated that A. hydrophila strain AH-1 (serotype O11) flagella, the constitutively expressed polar flagellum but not the inducible peritrichous lateral flagella, showed modification with a putative O-linked glycan moiety. As previously published, strain AH-3 (serotype O34) showed both flagella glycosylated. Both serotypes are among the four dominant serotypes (O11, O16, O18, and O34) that are associated with gastroenteritis and septicemia in clinical studies [26]. Mass spectrometry fragmentation of the putative glycan provides some evidence that the sugar contains similarities to previously characterized pseudaminic acid-like sugars. The mass spectrometry data and putative elemental formula suggest the presence of two methyl groups and two molecules of water, and an unknown mass of 25 Da.
The altered or incomplete flagellin glycosylation resulted in altered motility phenotypes. This was also observed with Clostridium difficile, where the abolition of flagellin glycosylation resulted in the loss of motility [27]. Studies of A. hydrophila strain AH-3 showed that deletion mutants of IL-8 production of HEK293-null and HEK293-hTLR5 cells. HEK293-null and HEK293-hTLR5 cells were stimulated with five different increased amounts (from 5 to 100 ng) of purified polar flagella from A. hydrophila AH-1 (wild-type) strain. HEK293-hTLR5 cells were also stimulated with A. hydrophila AH-1 polar flagellin FlaB obtained in E. coli (non-glycosylated). Data shown are means˘SD of three independent experiments.

Discussion
In the current study, we demonstrated that A. hydrophila strain AH-1 (serotype O11) flagella, the constitutively expressed polar flagellum but not the inducible peritrichous lateral flagella, showed modification with a putative O-linked glycan moiety. As previously published, strain AH-3 (serotype O34) showed both flagella glycosylated. Both serotypes are among the four dominant serotypes (O11, O16, O18, and O34) that are associated with gastroenteritis and septicemia in clinical studies [26]. Mass spectrometry fragmentation of the putative glycan provides some evidence that the sugar contains similarities to previously characterized pseudaminic acid-like sugars. The mass spectrometry data and putative elemental formula suggest the presence of two methyl groups and two molecules of water, and an unknown mass of 25 Da.
The altered or incomplete flagellin glycosylation resulted in altered motility phenotypes. This was also observed with Clostridium difficile, where the abolition of flagellin glycosylation resulted in the loss of motility [27]. Studies of A. hydrophila strain AH-3 showed that deletion mutants of pseudaminic acid biosynthesis abolish polar and lateral flagellum formation [23]. However, deletion mutants of pseudaminic acid biosynthesis in strain AH-1 (serotype O11) abolish polar flagellum formation but not lateral flagella. Because the presence of the putative pseudaminic acid seems to be a requirement for flagellin export and flagella formation in Aeromonas [23], this point is in agreement with our results that lateral flagella in this strain AH-1 are not glycosylated.
AH-1∆rmlB mutant is unable to produce O11-antigen LPS [18]; however, it is able to produce either polar or lateral flagella under induced conditions by TEM studies. No interaction between the flagella O-glycosylation and O-antigen LPS biosynthetic pathway was found in A. hydrophila. This is in contrast with studies of Pseudomonas aeruginosa, which suggested involvement of the O-antigen biosynthesis in O-glycosylation [28]. In addition, O-antigen LPS and flagella have been described as important molecules for bacterial adherence or biofilm formation in Aeromonas [17].
Results from the current study suggest that flagella is a more important adhesion factor than O-antigen LPS, as shown by adhesion to Hep-2 cells. Furthermore, the polar flagellum seems to be a determinant factor, with the lateral flagella unable to fully compensate for lack of expression of polar flagellin. This is the first study to show that polar flagellum glycosylation in Aeromonas is a determinant factor in adherence to eukaryotic cells. The results obtained in biofilm formation studies are more marked, with polar flagellum a requirement for biofilm formation, with mutants deficient in polar flagellin expression unable to form biofilms. Then, if there is any compensation from the lateral flagella to the loss of polar flagellum, it is unable to achieve the degree for biofilm formation. The O-antigen LPS mutant, able to produce either polar or lateral flagellin, was observed to form biofilms, but in a reduced capacity.
Toll-like receptors (TLRs) are major components of innate immunity. TLR5 is involved in recognizing bacterial flagellin and, after binding, it triggers the induction of pro-inflammatory cytokines such as IL-8 by the myeloid differentiation primary response gene 88 (MyD88)-dependent signaling pathway [29]. The results obtained indicate that A. hydrophila AH-1 polar flagellin glycosylation is important for and quantitatively related to immune stimulation of IL-8 production via TLR5. A clear reduction in IL-8 production was observed when polar non-glycosylated flagellin monomers (FlaB) were expressed in E. coli. Fish infected with pathogenic bacteria showed a significant enhanced IL-8 expression in the blood and intestine [30]. It is tempting to speculate the possible role of A. hydrophila AH-1 polar flagella glycosylation in cell invasion as observed for different A. hydrophila wild-type strains [17].
This study supports the existence of a complex mechanism of flagella glycosylation in different A. hydrophila strains. We show for the first time the presence of inducible lateral flagella, either with lateral flagellins glycosylated or not, in different A. hydrophila strains. We also show that different A. hydrophila strains used different glycans for polar flagellin glycosylation. Furthermore, we continue to shed light on the biological role of flagellum glycosylation, showing their implication in flagellum production, motility, adhesion to eukaryotic cells, biofilm formation, and TLR5 activation.
The bacterial strains and plasmids used in this study are listed in Table 3.

Flagella Purification
A. hydrophila AH-1 was grown in TSB for the polar flagellum purification and in TSA for the isolation of lateral flagella, and the flagella purified as previously described [40]. Purified flagella were analyzed by SDS-PAGE or by glycosylation chemical studies.

Transmission Electron Microscopy (TEM)
TEM was performed on Formvar-coated grids and negative stained with a 2% solution of uranyl acetate pH 4.1.

Electrospray Liquid Chromatography Mass Spectrometry Analysis of Intact Flagellins
Mass spectrometry studies of intact flagellin proteins were carried out using 50 µL aliquots of protein-containing sample as described previously, with some modification [23]. The purified flagellin preparations were injected onto a protein microtrap (Michrom Bioresources Inc., Auburn, CA, USA) connected to a gradient HPLC pump (Agilent 1100 HPLC, Agilent Technologies, Santa Clara, CA, USA). Solvent A was 0.1% formic acid in HPLC-grade water (Fisher, Waltham, MA, USA) and solvent B was 0.1% formic acid in acetonitrile. An HPLC gradient of 5%-60% solvent B (1 mL/min) over 60 min was used to resolve the protein mixture. A pre-column splitter was used to direct 35 µL/min of the HPLC column eluate into the electrospray interface of the QTOF2 (Waters, Milford, MA, USA), allowing real-time monitoring of ion elution profiles. Intact masses of proteins were calculated by spectral deconvolution, using MaxEnt (Waters, Beverly, MA, USA). For each protein, front-end collision-induced dissociation (feCID) was performed by increasing the cone voltage from 45 to 85 V and glycan-associated oxonium ions were observed. Tandem MS of prominent ions were also performed. In addition, multiply charged protein ions were selected for MS/MS. The collision energy was increased from X to Y incrementally and labile-, protein-, and glycan-associated fragment ions were observed.

Solution Enzymatic Digests and Bottom-Up Mass Spectrometry Analysis of Glycopeptides
In preliminary experiments, tryptic digests of intact proteins were performed, with 50 to 200 µg of pure flagellin prepartion in solution digested with trypsin (Promega, Madison, WI, USA) at a ratio of 30:1 (protein/enzyme, w/w) in 50 mM ammonium bicarbonate at 37˝C overnight. Protein digests were analyzed by nano-liquid chromatography MS/MS (nLC-MS/MS) using either a Q-TOF Ultima hybrid quadrupole time-of-flight MS (Waters) or an LTQ XL orbitrap MS (Thermo Fisher Scientific, Ottawa, ON, Canada) coupled to a nanoAcquity ultrahigh-pressure liquid chromatography system (Waters). MS/MS spectra were acquired automatically on doubly, triply, and quadruply charged ions in collision-induced dissociation (CID) mode for initial glycopeptide identification. Peak lists were automatically generated by PROTEINLYNX software (Waters) with the following parameters: smoothing-four channels, two smooths, Savitzky-Golay mode; centroid-minimum peak width at half height of four channels, centroid top 80%. Tryptic peptides were analyzed by nano-LC-MS/MS, and spectra were searched against the National Centre for Biotechnology nonredundant database and an in-house database of sequenced flagellin proteins using MASCOT 2.0.6 (Matrix Science, London, UK). A peptide score of 30 and above for a top-ranked hit was taken as positive identification, with each MS/MS spectrum verified by manual inspection. MS/MS spectra not identified by MASCOT were de novo sequenced. Proteinase K digests were analyzed up nanoliquid chromatography MS/MS, using a Q-TOF Ultima, coupled to a nanoAcquity ultrahigh-pressure liquid chromatography system (Waters, Milford, MA, USA). The column setup and gradients did not deviate from those reported in our other studies [23]. MS/MS spectra were acquired automatically on doubly, triply, and quadruply charged ions in collision-induced dissociation (CID) mode for initial glycopeptide identification. High resolution multi-stage mass spectrometry studies of glycan moieties were performed in high resolution (100,000) mode on the LTQ XL Orbitrap Mass spectrometer (Fisher Scientific, Waltham, MA, USA). All proteinaseK MS/MS spectra were de novo sequenced, no database searching was employed in peptide sequence assignment.

Adherence Assay to HEp-2 Cell
The adherence assay was conducted in triplicate as previously described [40].

Biofilm Formation
Quantitative biofilm formation was performed in a microtiter plate as described previously [40], by adapting the method of Pratt and Kotler [41].

Statistical Analysis
Results are expressed together with the standard deviations (SD) from several experiments and Student's t-test was used to compare mean values. Differences were considered significant when p-values were <0.05.