Flagellin glycosylation in Paenibacillus alvei CCM 2051^T

Abstract

Flagellin glycosylation impacts, in several documented cases, the functionality of bacterial flagella. The basis of flagellin glycosylation has been studied for various Gram-negative bacteria, but less is known about flagellin glycans of Gram-positive bacteria including Paenibacillus alvei, a secondary invader of honeybee colonies diseased with European foulbrood. Paenibacillus alvei CCM 2051^T swarms vigorously on solidified culture medium, with swarming relying on functional flagella as evidenced by abolished biofilm formation of a non-motile P. alvei mutant defective in the flagellin protein Hag. Here, the glycobiology of the polar P. alvei flagella was investigated. Analysis on purified flagellin demonstrated that the 30-kDa Hag protein (PAV_2c01710) is modified with an O-linked trisaccharide comprised of one hexose and two N-acetyl-hexosamine residues, at three sites of glycosylation. Downstream of the hag gene on the bacterial chromosome, two open reading frames (PAV_2c01630, PAV_2c01640) encoding putative glycosyltransferases were shown to constitute a flagellin glycosylation island. Mutants defective in these genes exhibited altered migration in sodium dodecyl sulfate polyacrylamide gel electrophoresis as well as loss of extracellular flagella production and bacterial motility. This study reveals that flagellin glycosylation in P. alvei is pivotal to flagella formation and bacterial motility in vivo, and simultaneously identifies flagella glycosylation as a second protein O-glycosylation system in this bacterium, in addition to the well-investigated S-layer tyrosine O-glycosylation pathway.

Introduction

The bacterial flagellum plays a key role in bacterial motility and is pivotal to the adaptation of bacteria to different biological niches (Logan 2006). Flagella from different bacteria have been shown to be indispensable as virulence factors, for colonization, and biofilm formation (Twine et al. 2009; Janesch, Koerdt, et al. 2013). In recent years, an increasing number of reports on flagellar glycosylation from archaea, as well as from Gram-positive and Gram-negative bacteria has been published (reviewed by Logan 2006). However, the exact biological role of these diverse flagellin glycan structures still has to be shown (Twine et al. 2009). In comparison with Gram-negative bacteria, less is known about glycosylation of Gram-positive flagella. Among Gram-positives, the spore-forming, anaerobic bacteria of Clostridium spp. are best investigated in this regard. It has been demonstrated that the flagella of three of four Clostridium botulinum strains associated with infant botulism are modified with a 7-acetamido-5-(N-methylglutam-4-yl)-amino-3,5,7,9-tetradeoxy-D-glycero-α-D-galacto-nonulosonic acid (αLeg5GluNMe7Ac), while the flagellins from three of four strains not associated with C. botulinum infections are modified with a di-N-acetylhexuronic acid attached up to seven sites per flagellin monomer (Twine et al. 2008). Clostridium difficile is another opportunistic pathogen causing antibiotic-associated diarrhea and pseudomembranous colitis in humans. The flagellin of strain C. difficile 630 is O-glycosylated with an N-acetylhexosamine (HexNAc) residue that is modified with methylated asparagine through a phosphate linkage at up to seven sites per monomer. Inactivation of the glycosyltransferase gene CD0240 resulted in the inability of the bacterium to synthesize glycosylated flagellar filaments, with only small amounts of unmodified flagellin being produced (Twine et al. 2009). Interestingly, clinical isolates of C. difficile flagella were found to be O-glycosylated with a heterogeneous glycan containing up to five monosaccharide residues, including a HexNAc, a deoxyhexose, a methylated deoxyhexose and a heptose residue (Twine et al. 2009). There are indications that the flagellins of Clostridium tyrobutyricum and Clostridium acetobutylicum are also glycosylated (Arnold et al. 1998; Bedouet et al. 1998; Lyristis et al. 2000). Non-reductive β-elimination treatment of the C. tyrobutyricum flagellin resulted in a mass downshift and in loss of monoclonal antibody binding (Arnold et al. 1998; Bedouet et al. 1998; Logan 2006). The flagellin of C. acetobutylicum is modified with a 12-kDa compound that was sensitive to neuraminidase treatment, indicative of its relation to sialic acids (Arnold et al. 1998; Bedouet et al. 1998; Lyristis et al. 2000; Logan 2006). The Gram-positive bacterium Listeria monocytogenes is responsible for listeriosis, a foodborne infection that causes life-threatening infections in immunocompromised persons, fetuses and newborns (Schirm et al. 2004; Lemon et al. 2007). The flagellin of L. monocytogenes is glycosylated at up to six sites per monomer with a single β-O-linked GlcNAc residue (Schirm et al. 2004). Inactivation of the O-GlcNAc transferase GmaR resulted in non-motile bacteria due to the inability of flagellin expression. GmaR is a bifunctional protein indicated by its dual role as glycosyltransferase and motility anti-repressor (Shen et al. 2006). Shen et al. (2006) demonstrated that the glycosyltransferase function is independent of anti-repressor activity, since the inactivation of the glycosyltransferase gene did not result in loss of motility (Shen et al. 2006). Biofilm formation was completely abolished in non-motile mutants, while no difference in biofilm formation could be observed for mutants lacking flagellar glycosylation (Lemon et al. 2007). In thermophilic Gram-positive bacteria, O-linked flagellin glycosylation includes Geobacillus stearothermophilus (NBRC 12550) and Bacillus sp. PS3 and was derived from a positive periodic acid-Schiff (PAS) staining reaction on an sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) gel (Hayakawa et al. 2009).

Paenibacillus alvei CCM 2051^T is a Gram-positive, endospore-forming, mesophilic bacterium that has gained attention as a secondary invader of diseased honeybee colonies infected with Melissococcus pluton, the causative agent of European foulbrood (Forsgren 2010). Paenibacillus alvei has been observed to swarm vigorously on solidified standard culture medium (Kim et al. 2011; Djukic et al. 2012; Janesch, Koerdt, et al. 2013). Swarming is a flagella-driven strategy for motility utilized by a wide range of bacterial species and, in several cases, also affects bacterial biofilm formation (Cohen et al. 2000; Ingham and Ben Jacob 2008; Janesch, Koerdt, et al. 2013). In P. alvei, we have recently shown with a non-motile Δhag mutant that the flagellin protein Hag (encoded by PAV_2c01710) is essential for the biofilm life-style of the bacterium (Janesch, Koerdt, et al. 2013). We demonstrated that P. alvei possesses polar flagella that are inserted in the 2D crystalline surface (S-) layer of the bacterium (Janesch et al. 2013). The S-layer completely covers the bacterium and exposes to the environment multiple glycan chains composed of, on average, 23 [→3)-β-d-Galp-(1→)-[α-d-Glcp-(1→6)] →4)-β-d-ManpNAc-(1→] repeating units and an adaptor saccharide -[GroA-2→ OPO₂→4-β-d-ManpNAc-(1→4)]→3)-α-l-Rhap-(1→3)-α-l-Rhap-(1→3)-α-l-Rhap-(1→3)-β-d-Galp-(1→ (Messner et al. 1995; Zarschler et al. 2010; Janesch, Messner, et al. 2013). These S-layer glycans are covalently attached to distinct tyrosine residues of the S-layer protein SpaA, with the confirmed positions Tyr47 and Tyr155 (B. Janesch, C. Schäffer, M. Pabst, F. Altmann, P. Messner, unpublished data). The biosynthesis of the S-layer glycan relies on a 24.3-kb S-layer glycosylation (slg) gene cluster encoding 18 open reading frames (Zarschler et al. 2010) and the predicted glycan biosynthesis pathway is comparable with the ABC-transporter-dependent pathway of lipopolysaccharide O-antigen biosynthesis (Zarschler et al. 2010; Ristl et al. 2012). Deletion of the initiation enzyme WsfP of S-layer glycan biosynthesis in P. alvei indicated that, while S-layer glycosylation was completely abolished, other protein bands remained positive upon PAS staining for carbohydrates (B. Janesch, C. Schäffer, P. Messner, unpublished data). Thus, we hypothesized that P. alvei might possess a second protein glycosylation system besides that involved in S-layer protein glycosylation. The flagellin protein Hag might be targeted by that glycosylation system, given the multiple reports on glycosylated bacterial flagella.

This study was designed to shed light on the glycobiology of the P. alvei flagellin Hag. This includes (i) the demonstration that the flagellin of P. alvei is indeed modified with an O-linked glycan at distinct sites, (ii) the identification of a putative flagellin glycosylation island in the P. alvei genome, and (iii) the confirmation of the compositional and structural difference between the flagellin glycan and the S-layer glycan by mass spectrometric analyses. (iv) In addition, our data indicate a role of flagellin glycosylation in flagellin export in vivo, but not in flagellum self-assembly as evidenced by in vitro studies.

Results

Indications of flagellin glycosylation in P. alvei

A comparison of recombinant, His₆-tagged flagellin Hag expressed in Escherichia coli BL21 (rHag_{E. coli}) with native flagellin (Hag_native) isolated from P. alvei wild-type cells flagella after SDS–PAGE gel separation revealed that one prominent protein was migrating as a PAS-positive band with an apparent molecular mass of 33–35 kDa in the flagella preparation (Hag_native), whereas rHag_{E. coli} migrated as a PAS-negative band of ∼30 kDa (Figure 1A). This value also corresponds well to the calculated molecular mass of the 277-amino acid Hag protein.

Fig. 1.

Comparison of recombinant and wild-type flagellin of P. alvei CCM 2051^T. (A) Recombinant, His₆-tagged flagellin Hag expressed in E. coli BL21 (rHag_{E. coli}, ~30 kDa) was compared with flagella isolated from P. alvei wild-type cells (Hag_native, 33–35 kDa), after separation by SDS–PAGE (12% gel) upon CBB staining (left) and PAS staining (right). (B) Anti-His western-immunoblot of recombinant, His₆-tagged flagellin Hag expressed in P. alvei CCM 2051^T wild-type strain (rHag_{P. alvei}) and in a P. alvei mutant defective in two initiation enzymes of glycan biosynthesis (rHag_{P. alvei ΔwsfPΔpglB}). Twenty microliters (1 OD₆₀₀ unit) of rHag_{P. alvei} and rHag_{P. alveiΔwsfPΔpglB} and 3 µl of rHag_{E. coli} (~60 ng) were applied to SDS–PAGE (12%). (C) Amino sequence of P. alvei flagellin. Bold amino acids indicate the sequence coverage obtained by MS analyses of tryptic peptides. Identified glycopeptides are shown in green, glycosylation sites are boxed in orange and a third potential glycopeptide is shown in yellow. (D) Sequence alignment of flagellins from P. alvei (Pa), C. botulinum (Cb), C. difficile BI-1 (Cd) and L. monocytogenes (Lm). Glycosylation sites are marked in green and occur preferentially in the central domains of the protein.

To investigate P. alvei flagellin glycosylation, His₆-tagged flagellin was expressed in P. alvei CCM 2051^T (rHag_{P. alvei}).

Western immunoblotting using an anti-His₆ antibody clearly revealed a difference of ∼3–5 kDa between rHag_{P. alvei} and rHag_{E. coli}, which supported the glycoprotein nature of Hag_native (Figure 1B). In the same Western-immunoblotting experiment, Hag expressed in a P. alvei CCM 2051^T mutant was analyzed. This mutant was defective in two initiation enzymes of glycan biosynthesis (ΔwsfPΔpglB), with WsfP being the proven initiation enzyme of S-layer protein glycosylation and PglB being annotated as a putative lipid carrier:UDP-N-acetylgalactosaminyl-transferase encoded in the P. alvei genome. The Hag protein expressed in this mutant showed the identical migration behavior as glycosylated rHag_{P. alvei} (33–35 kDa), indicating that Hag glycosylation is independent from other protein glycosylation systems of P. alvei.

Glycoproteomics analysis of the P. alvei flagellin

To identify the nature of the glycan attached to the P. alvei Hag protein a comprehensive glycoproteomics analysis was performed. The Hag_native protein band obtained after SDS–PAGE separation was digested with trypsin and the resulting (glyco)peptides were analyzed by Matrix Assisted Laser Desorption Ionization-Time of flight (MALDI-TOF-MS) peptide mass fingerprinting (PMF) as well as reversed phase nano liquid chromatography-electrospray ionizationtandem mass spectrometry (LC-ESI-IT-MS/MS). The combined data from both analyses allowed an ad hoc mapping of 81.9% of the flagellin protein sequence (Supplementary data, PMF_Detected_Peptides. xlsx). However, a number of MS2 spectra could initially not be positively correlated to the expected tryptic peptides. Further manual inspection of these spectra identified numerous peptide y and b ion sequence tags that could be positively associated to the expected tryptic peptides 135–153 and 183–205, respectively. The precursor mass of these compounds, however, showed a m/z increase of 568.2 corresponding to the addition of a trisaccharide with the composition HexHexNAc₂, which was also further supported by the MS2 data (Figure 2; Supplementary data, Glycopeptide_Fragment_Ions.xlsx). The specific glycopeptide fragments indicated that a HexNAc residue was the first monosaccharide attached to the protein backbone. The remaining mass difference corresponded to a HexNAcHex being attached to the initial HexNAc (Figure 2); however, no specific fragments were detected that could have clarified the specific sequence of the trisaccharide attached to Hag. Since all detected Hag glycopeptides exhibited an unconventional fragmentation pattern (lack of any m/z 366.1 oxonium ion), it cannot be completely excluded that the detected mass difference corresponding to a HexNAcHex disaccharide could also be caused by a different type of modification that coincidentally exhibited a similar m/z difference. This could also explain why just two putative glycosyltransferase genes were found in the vicinity of the hag gene (see next paragraph: Analysis of a putative P. alvei glycosylation island).

Fig. 2.

Identification of glycopeptides derived from P. alvei flagellin. Top: the b and y ions allowed the annotation of the peptide sequence ¹³⁵LLNAASTVIFQVGANSGEK¹⁵³ and its glycosylation with a single trisaccharide HexHexNAc₂. The site of glycosylation could be narrowed down to be either Ser140 or Thr141, respectively. Bottom: the triply charged precursor was identified as peptide ¹⁸³LNATAASNLTAIQTAIDAVSGIR²⁰⁵. The fragment spectra allowed a clear assignment of Thr192 being glycosylated with the HexHexNAc₂ trisaccharide.

In peptide 183–205 (¹⁸³LNATAASNLTAIQTAIDAVSGIR²⁰⁵), Thr192 could be identified as being glycosylated, whereas on glycopeptide 135–153 (¹³⁵LLNAASTVIFQVGANSGEK¹⁵³), the exact site could not be unambiguously determined, but narrowed down to either Ser140 or Thr141 (Figure 1C). With these glycopeptides positively identified, the overall sequence coverage of Hag _native increased to 96% (Figure 1C, bold-labeled sequences, Supplementary data, PMF_Detected_Peptides.xlsx). A signal corresponding to a potential third glycopeptide (⁹⁰MSELMVQGANEVLTTTDAK¹⁰⁸) carrying a similar trisaccharide could be detected, but unfortunately the MS/MS spectra did not deliver sufficient y and b ions to unambiguously identify the peptide sequence (Supplementary data, PMF_Detected_Peptides.xlsx). Nevertheless, the oxonium ions clearly indicated the glycopeptide nature of the triply charged precursor m/z 880.36 and the resulting peptide signal corresponded well to the doubly methionine-oxidized form of this peptide, making it very likely that this signal corresponds to the glycosylated peptide 90–108.

Sequence alignment of the flagellins from P. alvei (Pa), C. botulinum (Cb), C. difficile BI-1 (Cd) and L. monocytogenes (Lm) revealed that the glycosylation sites identified in P. alvei Hag (Ser140/Thr141, and Thr192) are located within the central region of Hag (D2–D3 domains). This is in good agreement with the situation found in other Gram-positive bacteria (Figure 1D).

Analysis of a putative P. alvei flagellin glycosylation island

Database comparison of different flagellin glycosylation islands of Gram-positive bacteria carrying glycans without nonulosonic acids revealed that, in all cases, glycosyltransferases involved in flagellin glycosylation are located in close proximity to the flagellin structural gene on the bacterial chromosome (Figure 3A).

Fig. 3.

Comparison of flagellin glycosylation islands in Gram-positive bacteria. (A) Flagellin glycosylation islands of P. alvei CCM 2051^T, Clostridium difficile 630, Clostridium difficile BI-1, L. monocytogenes EGD-e, Butyrivibrio fibrisolvens 16/4 and C. tyrobutyricum showing the chromosomal organization of the flagellin and glycosyltransferase genes important to flagellin glycosylation. (B) Schematic drawing of the L. monocytogenes Lmo0688 enzyme and the P. alvei PAV_2c01630 and PAV_2c01640 proteins. The predicted family-2 glycosyltransferase domains are shown in red, the predicted β4-glucosyltransferase domains are shown in yellow, and the TPR domains are shown in blue. (C) Sequence alignment of Lmo0688, PAV_2c01630 and PAV_2c01640. Amino acid residues identical for all three proteins are blocked in blue, residues shared only between two of the proteins are blocked in green. The DxD glycosyltransferase motif is indicated above the alignment.

Closer inspection of the P. alvei genome sequence showed that two predicted glycosyltransferase genes — PAV2c_01630 and PAV2c_01640 — are indeed located in close proximity upstream of the hag gene coding for P. alvei flagellin (Figure 3A). PAV2c_01630 and PAV2c_01640 encode proteins of 628 and 363 amino acids, respectively, equaling apparent molecular masses of 73 and 42 kDa. Related proteins according to NCBI entries are summarized in Table I. Bio-informatics analysis of PAV2c_01630 and PAV2c_01640 predicted the presence of a glycosyltransferase family-2 domain and tetratricopeptide (TPR) domains comparable with the O-GlcNAc transferase Lmo0688 of L. monocytogenes (Figure 3B). Partial sequence alignment of Lmo0688 and the P. alvei glycosyltransferases PAV2c_01630 and PAV2c_01640 showed that all three proteins possess the DxD motif that constitutes the active site in family-2 glycosyltransferases (Shen et al. 2006) (Figure 3C).

Table I.

Proteins encoded in a putative glycosylation island of P. alvei. The nomenclature of the NCBI database for P. alvei DSM 29 was used for the ORFs

Gene	Size of product (amino acids)	Related proteins
		Protein and function	Bacterium	Identity/positives (%)	Sequence ID
PAV_2c01630	628	Glycosyl transferase family 2	Paenibacillus lactis	40/61	WP_007132294
		Glycosyl transferase family 2	Bacillus sp. 10403023	41/64	WP_010676910
		Glycosyl transferase family 2	Paenibacillus polymyxa SC2	36/57	YP_003949087
		Glycosyl transferase family 2	Thermoanaerobacterium xylanolyticum LX-11	36/59	YP_004470480
PAV_2c01640	363	Glycosyl transferase family 2	Paenibacillus lactis	36/62	WP_007132295
		Glycosyl transferase family 2	Paenibacillus vortex	37/58	WP_006213045
		Glycosyl transferase	Clostridium carboxidovorans	33/56	WP_007059572
		Glycosyl transferase family	C. botulinum A str. ATCC f3502	34/53	YP_001255225

To confirm the involvement of PAV2c_01630 and PAV2c_01640 in P. alvei Hag_native flagellin glycosylation, knockout mutants of the respective genes were generated based on bacterial mobile group II intron-mediated gene disruption (Zarschler et al. 2009). To analyze putative changes in Hag glycosylation, His₆-tagged Hag was expressed in the knockout strains P. alvei PAV2c_01630::Ll.LtrB (P. alvei ΔPAV2c_01630) and P. alvei PAV2c_01640::Ll.LtrB (P. alvei ΔPAV2c_01640), and any glycosylation-induced SDS–PAGE shifts of rHag_{ΔPAV2c_01630} and rHag_{ΔPAV2c_01640} were compared with that of purified, fully glycosylated rHag_{P. alvei} and non-glycosylated, recombinant Hag from E. coli rHag_{E. coli} by Western immunoblotting (Figure 4A). Both recombinant flagellins from the glycosyltransferase mutants — rHag_{ΔPAV2c_01630} and rHag_{ΔPAV2c_01640} — migrated at apparent molecular masses between those of HagP. alvei (33–35 kDa) and rHag_{E. coli} (∼30 kDa), with rHag_{ΔPAV2c_01640} showing a higher apparent molecular mass than rHag_{ΔPAV2c_01630}. These data were clearly indicative of both glycosyltransferases directly affecting the glycosylation of flagellin, and, furthermore, of a sequential order of P. alvei PAV2c_01630 and PAV2c_01640 action.

Fig. 4.

Characterization of P. alvei PAV_2c01630 and PAV_2c01640 knockouts by western-blot, motility assay, and TEM. (A). A clear downshift of Hag from P. alvei ΔPAV_2c01630 (rHag_{P. alveiΔ1630,} lane 1) and ΔPAV_2c01640 (rHag_{P. alveiΔ1640,} lane 2) compared with glycosylated rHag _{P. alvei} from the wild-type strain (lane 3) is visible on a Western-immunoblot using anti-His₆-antibody. Ten microliters [0.5 optical density (OD) units] of the wild type and 20 µl of ΔPAV_2c01630 (rHag_{P. alveiΔ1630}) and ΔPAV_2c01640 (rHag_{P. alveiΔ1640}) (1.0 OD unit), each, were applied to SDS–PAGE (12%). As a control, rHag_{E. coli} (~60 ng) was used revealing its apparent molecular mass (~30 kDa) below that of the mutants. (B) P. alvei ΔPAV_2c01630 (Δ1630) and ΔPAV_2c01640 (Δ1640) cells lose their ability to swarm on 1% LB-agar plates when compared with the wild type (middle picture). Swarming of the mutants is restored upon complementation with the respective endogenous protein (Δ1630_comp, Δ1640_comp). The pictures represent one of three independent experiments. (C) TEM images of P. alvei ΔPAV_2c01630, ΔPAV_2c01640, wild-type cells and the complemented strains Δ1630_comp and Δ1640_comp. Flagella are clearly visible on P. alvei wild type, Δ1630_comp and Δ1640_comp cells but not on the glycosyltransferase mutant cells. The upper lane shows the cells with (wild type, Δ1630_comp and Δ1640_comp) or without (Δ1630 and Δ1640) associated flagella. The lower lane shows a detailed view of many broken flagella lying around for the wild type, Δ1630_comp and Δ1640_comp, while no such structures could be found for Δ1630 and Δ1640.

Paenibacillus alvei ΔPAV2c_01630 and ΔPAV2c_01640 cells lose the ability to swarm on LB-agar plates

Based on our previous observation that non-functional P. alvei flagella result in loss of swarming ability of the bacterium on LB-agar plates (Janesch, Koerdt, et al. 2013), we were interested to see whether deletion of either of the two glycosyltransferases from the flagellin glycosylation island would result in a change of motility of the P. alvei cells. Indeed, swarming on semi-soft (1%) LB-agar plates was completely abolished for both P. alvei ΔPAV2c_01630 and P. alvei ΔPAV2c_01640 after 16 h of incubation. In contrast, the P. alvei wildtype bacterium produced a complex branching pattern characteristic of motile P. alvei cells when cultivated on semi-soft LB-agar plates (Figure 4B).

Reconstitution of P. alvei ΔPAV2c_01630 and P. alvei ΔPAV2c_01640 was done by plasmid-based expression of the respective gene, which verified that loss of swarming is indeed due to hypoglycosylation of Hag flagellin (Figure 4B).

Knockout of PAV2c_01630 and PAV2c_01640 glycosyltransferases affects flagellation of P. alvei

To visualize the flagellation status, negatively stained cells of the glycosyltransferase mutants P. alvei ΔPAV2c_01630 and P. alvei ΔPAV2c_01640 were examined by transmission electron microscopy and compared with P. alvei wild-type cells. No flagella could be detected on either of the glycosyltransferase mutant cells, whereas polar flagella were abundantly present on P. alvei wild-type cells and on the complemented strains P. alvei ΔPAV2c_01630_comp and P. alvei ΔPAV2c_01640_comp. In the vicinity of the wild-type and the complemented cells, many long flagella could be found. For comparison, only small pieces of flagella could be seen for the mutant cells, but none of these pieces was found associated with the cells, indicative of a possible release from the cytoplasm of lysed cells (Figure 4C).

Paenibacillus alvei ΔPAV2c_01630 and ΔPAV2c_01640 do not assemble extracellular flagella in vivo

In an attempt to shear off the flagella with a commercial blender from cells of the glycosyltransferase mutants P. alvei ΔPAV2c_01630 and P. alvei ΔPAV2c_01640, no flagellin protein could be obtained according to a Coomassie Brilliant Blue G250 (CBB)-stained SDS–PAGE evidence, while, under the same conditions, prominent bands representing Hag_native could be isolated from P. alvei wild-type, P. alvei ΔPAV2c_01630_comp and P. alvei ΔPAV2c_01640_comp cells (Figure 5A). Tryptic peptide maps of the SDS–PAGE-separated dominant protein bands (Supplementary data, Figure S1) verify the presence of the P. alvei flagellin in the flagella isolations of P. alvei wild type, P. alvei ΔPAV2c_01630_comp and P. alvei ΔPAV2c_01640_comp.

Fig. 5.

Analysis of P. alvei ΔPAV_2c01630 and ΔPAV_2c01640 in comparison with P. alvei wild type by SDS–PAGE and western immunoblotting. (A) No flagella could be enriched from ΔPAV_2c01630 and ΔPAV_2c01640 cells (Hag_{P. alvei Δ1630}, Hag_{P. alvei Δ1640}) while flagella isolated from P. alvei wild type (Hag _native), Δ1630_comp (Hag_{P. alvei Δ1630comp}) and Δ1640_comp (Hag_{P. alvei Δ1640comp}) cells are clearly present as a prominent band on an SDS–PAGE gel (glycosylation of this band is shown in Figure 1A). (B) Anti-His₆ western immunoblotting of recombinant Hag proteins revealed that flagellin can be found in the pellet fraction of ΔPAV_2c01630 (rHag_{P. alvei Δ1630}) and ΔPAV_2c01640 (rHag_{P. alvei Δ1640}) cells, but not in the corresponding supernatant fractions. Twenty microliters (1.0 OD unit) of the pellets and 20 µL of trichloroacetic acid precipitated supernatant, each, were applied to the gels (14%). In contrast, both the pellet and the supernatant fraction of P. alvei wild-type cells contained rHagP. alvei. It is evident that in P. alvei ΔPAV_2c01630 and ΔPAV_2c01640 cells, a smaller amount of rHag was expressed compared with wild-type cells. As a control, rHag_{E. coli} (~30 ng) was used revealing its apparent molecular mass (~30 kDa) below that of the mutants.

To further specify our analysis, the pellet (P) and the supernatant (S) fractions of recombinant Hag proteins were subjected to Western immunoblotting using an anti-His₆ antibody (Figure 5B). In all cases, rHag could be detected in the pellet fraction, although in the mutants in a lesser amount than in the wild type. Comparing the migration behavior of the recombinant flagellins, it was evident that all flagellins are glycosylated, since they migrated clearly >30 kDa—which would correspond to non-glycosylated flagellin (compare with Figure 4A). Only for the P. alvei wild-type strain, rHag_{P. alvei} was also found in the supernatant, which indicated its location at the cell surface. rHag from the P. alvei ΔPAV2c_01630 and ΔPAV2c_01640 mutants, however, seemed to be detained inside the cells as could be deduced from its recovery in the pellet fraction (Figure 5B).

Non-glycosylated, recombinant P. alvei Hag expressed in E. coli retains its self-assembly capability in vitro

To investigate whether the self-assembly capability of Hag flagellin proteins into flagellar filaments is dependent on the glycosylation, recombinant, non-glycosylated flagellin was purified from E. coli (compare with Figure 1). Immediately after Ni²⁺-affinity chromatography, rHag_{E. coli} was dialyzed against MQ-water to promote in vitro self-assembly of flagellins. Transmission electron microscopy of a negatively stained preparation of this solution revealed that rHag_{E. coli} retains its self-assembly capability into long flagellar filaments of d ∼10 nm, comparable with those isolated from P. alvei wild-type cells, indicating that glycosylation is not a prerequisite for self-assembly of the flagellin monomers into flagellar filaments (Figure 6).

Fig. 6.

TEM images of rHag protein from E. coli and isolated flagella from P. alvei wild-type cells. Recombinant, non-glycosylated, Hag protein is fully capable of self-assembling in vitro into flagellar filaments (left), which are comparable with those sheared off from P. alvei wild-type cells (right).

Discussion

This study provides the first evidence that P. alvei CCM 2051^T glycosylates its polar flagella. While O-glycan modification of the P. alvei S-layer protein SpaA at tyrosine residues, representing the most abundant protein of this bacterium, has been described in detail previously (Zarschler et al. 2010), we show here that glycosylation of P. alvei Hag flagellin is independent of S-layer protein glycosylation and relies on a separate glycosylation island.

His-tagged flagellin expressed and purified from P. alvei was shown by nanoLC-ESI-IT-MS/MS to be O-glycosylated with a putative HexHexNAc₂ trisaccharide at up to three serine or threonine sites per monomer. The location of these residues within the protein conforms with that of the analyzed flagellin proteins from C. botulinum, C. difficile BI-1, and L. monocytogenes, where glycosylation is restricted to the central variable region of the flagellins which represents the surface exposed D2 and D3 domains of assembled flagella (Hayakawa and Ishizuka 2012).

While flagellin glycosylation has mainly been associated with pathogenic bacteria, there are several reports providing indirect evidence of flagellin glycosylation in non-pathogenic strains, including Azospirillum brasilense Sp7 (Moens et al. 1995), Caulobacter crescentus CB15 (Faulds-Pain et al. 2011), Spirochaeta aurantia (Brahamsha and Greenberg 1988), Shewanella oneidensis (Wu et al. 2011), marine magnetotactic ovoid bacterium MO-1 (Zhang et al. 2012), Thermus thermophilus HB8 (Papaneophytou et al. 2012), C. tyrobutyricum (Bedouet et al. 1998), C. acetobutylicum (Lyristis et al. 2000), G. stearothermophilus (Hayakawa et al. 2009) and Bacillus sp. PS3 (Hayakawa et al. 2009). The only detailed structural characterization of a flagellin glycan from a non-pathogenic bacterium reported to date is that from Burkholderia thailandensis (Scott et al. 2011), supporting the importance of flagellin glycosylation under diverse physiological settings.

Like for other flagellated Gram-positive bacteria carrying glycans without nonulosonic acids (and, thus, unlikely to be pathogenicity-related) glycosyltransferases (PAV2c_01630 and PAV2c_01640) important for the modification of the P. alvei flagellin could be found in close proximity to the flagellin structural gene (compare with Figure 3A). In a partial multiple sequence alignment of P. alvei PAV2c_01630 and PAV2c_01640 with L. monocytogenes Lmo0688, a characteristic DxD motif that constitutes the active site could be identified for either of the P. alvei glycosyltransferases (Shen et al. 2006), pinpointing a similar role of the enzymes in P. alvei flagellin glycosylation (compare with Figure 3C).

Based on MS/MS data and Western immunoblotting of recombinant Hag expressed in the different P. alvei backgrounds (wild-type, ΔPAV2c_01630 and ΔPAV2c_01640 glycosyltransferase mutants; compare with Figure 4A), PAV2c_01630 is likely to be involved in the transfer of one or even both HexNAc moieties and PAV2c_01640 is likely to be involved in the transfer of the hexose (Hex) moiety, onto Hag. This interpretation is supported by the observed step-wise upshift in apparent molecular mass from ∼30 kDa of rHag_{E. coli} over rHag _{ΔPAV2c_01630} and rHag _{ΔPAV2c_01640} to fully glycosylated rHag_{P. alvei} at 33–35 kDa. If PAV2c_01630 transfers only one Hex-NAc residue, there must be one remaining glycosyltransferase left, which we were not able to identify so far and that is not located in close proximity to the hag gene. If PAV2c_01630 is responsible for the transfer of both HexNAc moieties, rHag _{ΔPAV2c_01630} should be non-glycosylated, but there is still a remaining upshift in apparent molecular mass compared with rHag_{E. coli} which would mean that the non-glycosylated protein migrates differently in E. coli and P. alvei for reasons we do not know. Detailed mass spectrometry (MS) analysis of the Hag glycans produced by P. alvei ΔPAV2c_01630 and ΔPAV2c_01640 was rendered impossible because of the low constitutive expression level of the recombinant hypo-glycosylated flagellins in the P. alvei mutants.

Considering the need of functional flagella for the swarming phenotype of P. alvei CCM 2051^T (Kim et al. 2011; Djukic et al. 2012; Janesch, Koerdt, et al. 2013,) as well as the reported effect of glycosylation on proper flagella assembly (Thibault et al. 2001; Goon et al. 2003; Ewing et al. 2009; Twine et al. 2009; Parker et al. 2014), we were interested to analyze this situation in P. alvei. Indeed, P. alvei mutants lacking either of the glycosyltransferase from the flagellin glycosylation island (i.e., PAV2c_01630 or PAV2c_01640) were neither able to swarm (Figure 4B) nor assemble their flagella in vivo (Figure 4C).

The biological role of glycosylation in flagellar assembly is still not fully understood. It could be required for the efficient secretion of the flagellin proteins through the basal body apparatus but also for stability of flagellin subunit–subunit interactions within the flagellar filament (Shen et al. 2006). Loss of flagella glycosylation does not necessarily results in loss of filament formation. Scott et al. reported that inactivation of a gene from the LPS O-antigen biosynthesis cluster, which is also involved in flagella glycosylation, abolished flagella glycosylation and motility of Burkholderia pseudomallei but the observed defects were not due to the loss of assembled flagella (Scott et al. 2011). Furthermore, Shen et al. showed that the inactivation of the glycosyltransferase GmaR from L. monocytogenes had no effect on flagella-driven motility. In contrast to other organisms, the secretion and stability of L. monocytogenes flagellin was not dependent on glycosylation (Shen et al. 2006). Also for Pseudomonas aeruginosa and Pseudomonas syringae, flagella glycosylation is not required for assembly, but in these cases, the glycan has been implicated in virulence (Arora et al. 2005;Taguchi et al. 2006).

The open question remains if glycosylation is necessary for efficient secretion of flagellin proteins through the basal body apparatus. Recently, Parker et al. showed that glycosylation is not required for flagellin export in Aeromonas caviae but is essential for filament assembly, since non-glycosylated flagellin was still secreted (Parker et al. 2014). In our study, we could not detect secreted recombinant hypo-glycosylated Hag flagellin in the supernatant of the P. alvei glycosyltransferase mutants ΔPAV2c_01630 and ΔPAV2c_01640, and also no flagellar filaments were formed as cellular appendages of the mutants in vivo. Thus, it is conceivable to assume that the hypoglycosylated flagellin proteins are trapped within the cytoplasm of the mutant P. alvei cells because of failure of proper transport to the cell surface. Another reason might concern analytical limitations, considering that in the P. alvei glycosyltransferase mutants only reduced amounts of recombinant Hag were produced, which implicates less protein being secreted that, then, might escape detection from the extracellular fraction as postulated by others (Parker et al. 2014). The question why less flagellin protein is produced in the P. alvei mutants when the glycosylation is incomplete needs to be further investigated. Since we could show in an in vitro experiment with recombinant non-glycosylated flagellin (rHag_{E. coli}) that self-assembly of the Hag flagellin into flagellar filaments proceeds independently of its glycosylation, we assume that glycosylation is rather important for either the stability of the flagellin protein or for the transport of the protein to the cell surface and not for the self-assembly capability of the protein per se.

Materials and methods

Bacterial strains and growth conditions

Paenibacillus alvei CCM 2051^T was obtained from the Czech Collection of Microorganisms (CCM; Brno, Czech Republic) and was grown at 37°C and 200 rpm in Luria-Bertani (LB) broth or on LB agar plates supplemented with 10 µg/mL chloramphenicol (Cm), when appropriate. E. coli DH5α cells (Invitrogen) and E. coli BL21 (DE3) cells (In-vitrogen) were cultivated at 37°C and 200 rpm in LB medium supplemented with 30 µg/mL chloramphenicol (Cm) and 50 µg/mL kanamycin (Km), respectively. All strains used in this study are listed in Table II.

Table II.

Bacterial strains and plasmids used in this study

Strain or plasmid	Genotype and/or relevant characteristics	Source
P. alvei CCM 2051T	Wild-type isolate; Kmr	Czech Collection of Microorganisms (CCM)
Escherichia coli DH5α	F−φ80dlacZ M15 (lacZYA-argF)U169 deoR recA1 endA1 hsdR17 (rK−mK−) phoA supE44 thi-1 gyrA96 relA1−	Invitrogen
Escherichia coli BL21 (DE)	F−, ompT, hsdS (rB−mB−), gal, dcm (DE3)	Invitrogen
P. alvei CCM 2051T ΔwsfP	P. alvei CCM 2051T carrying a targetron insertion at the wsfP locus	Zarschler et al. (2009)
P. alvei CCM 2051T ΔwsfPΔpglB	P. alvei CCM 2051T carrying a targetron insertions at the wsfP and the pglB locus	This study
P. alvei CCM 2051T ΔPAV_2c01630	P. alvei CCM 2051T carrying a targetron insertion at the PAV_2c01630 locus	This study
P. alvei CCM 2051T ΔPAV_2c01640	P. alvei CCM 2051T carrying a targetron insertion at the PAV_2c01640 locus	This study
P. alvei CCM 2051T ΔPAV_2c01630 comp	P. alvei CCM 2051T carrying a targetron insertion at the PAV_2c01630 locus and the pEXALV_P(1640)_1630 plasmid; Cmr	This study
P. alvei CCM 2051T ΔPAV_2c01640 comp	P. alvei CCM 2051T carrying a targetron insertion at the PAV_2c01640 locus and the pEXALV_P(1640)_1640 plasmid; Cmr	This study
P. alvei CCM 2051T (pEXALV_hag_6His)	P. alvei CCM 2051T wild-type isolate carrying the pEXALV_hag_6His plasmid; Cmr	This study
P. alvei CCM 2051T ΔPAV_2c01630 (pEXALV_hag_6His)	P. alvei CCM 2051T ΔPAV_2c01630 carrying the pEXALV_hag_6His plasmid; Cmr	This study
P. alvei CCM 2051T ΔPAV_2c01640 (pEXALV_hag_6His)	P. alvei CCM 2051T ΔPAV_2c01640 carrying the pEXALV_hag_6His plasmid; Cmr	This study
pEXALV	pNW33N carrying the sgsE S-layer gene promoter of G. stearothermophilus NRS 2004/3a; Cmr	Zarschler et al. (2009)
pET28a	Expression vector with a His6-tag, Kanr	Novagen
pTT_wsfA243	pTT_wsfP1176 targeted for insertion at position 243/244 from the initial ATG of wsfA	Zarschler et al. (2010)
pTT_PAV2c_01630_1143	pTT_wsfA243 targeted for insertion at position 1143/1144 from the initial ATG of PAV2c_01630	This study
pTT_PAV2c_01640_273	pTT_wsfA243 targeted for insertion at position 273/274 from the initial ATG of PAV2c_01640	This study
pET28a_hag_6His	pET28a carrying the His6-tagged hag gene of P. alvei CCM 2051T; Kanr	This study
pEXALV_hag_6HIS	pEXALV carrying the his-tagged hag gene of P. alvei CCM 2051T	This study
pEXALV_P(1640)_1630	pEXALV without P(SgsE) carrying the PAV_2c01630 gene and the promoter of PAV_2c01640 of P. alvei CCM 2051T	This study
pNW33N_P(1640)_1640	pNW33N carrying the promoter and the PAV_2c01640 gene of P. alvei CCM 2051T	This study

General molecular methods

All enzymes were purchased from Thermo Scientific. Genomic DNA of P. alvei CCM 2051^T was isolated by using a Genomic Tip 100 kit (Qiagen) according to the manufacturer’s instructions, except that cells were broken by repeated freezing and thawing cycles (10 times) (Zarschler et al. 2009). The GeneJET™ Gel Extraction Kit (Fermentas) was used to purify DNA fragments from agarose gels and to purify digested plasmids and oligonucleotides. Plasmid DNA from transformed cells was isolated with the GeneJET™ Plasmid Miniprep kit (Fermentas). Agarose gel electrophoresis was performed as described elsewhere (Sambrook et al. 1989). Primers for PCR and DNA sequencing were purchased from Invitrogen (Table III). PCR was performed using the Phusion^® High-Fidelity DNA Polymerase (Fermentas) and the thermal cycler My CyclerTM (Bio-Rad). Transformation of chemically competent E. coli DH5α cells and E. coli BL21 (DE3) cells was done according to the manufacturer’s protocol (Invitrogen). Transformation of P. alvei CCM 2051^T wild type, P. alvei CCM 2051^T ΔwsfP, P. alvei CCM 2051^T ΔPAV2c_01630 and P. alvei CCM 2051^T ΔPAV2c_01640 cells is described elsewhere (Zarschler et al. 2009). Transformants were screened by PCR using RedTaq ReadyMix PCR mix (Sigma-Aldrich), and recombinant clones were analyzed by restriction mapping and confirmed by sequencing (Microsynth).

Table III.

Oligonucleotide primers used for PCR amplification reactions

Primers	Sequence (5′ → 3′)a
PglB_107|108s-IBS	AAAAAAGCTTATAATTATCCTTAGGGCTCGCCCGTGTGCGCCCAGATAGGGTG
PglB_107|108s-EBS1d	CAGATTGTACAAATGTGGTGATAACAGATAAGTCGCCCGTACTAACTTACCTTTCTTTGT
PglB_107|108s-EBS2	TGAACGCAAGTTTCTAATTTCGATTAGCCCTCGATAGAGGAAAGTGTCT
PaPglB_for	ATGAAGCGATTACTGGATATTGTGG
PaPglB_rev	CTATCCGCTTATTTTGCGTTCC
PAV_2c1630_1143|1144s-IBS	AAAAAAGCTTATAATTATCCTTACTGGACTTGCTGGTGCGCCCAGATAGGGTG
PAV_2c1630_1143|1144s-EBS1d	CAGATTGTACAAATGTGGTGATAACAGATAAGTCTTGCTGCTTAACTTACCTTTCTTTGT
PAV_2c1630_1143|1144s-EBS2	TGAACGCAAGTTTCTAATTTCGGTTTCCAGTCGATAGAGGAAAGTGTCT
PAV_2c1630_KO_for	ATGCCGATGAGTATTTGGACGAATCG
PAV_2c1630_KO_rev	CCTCAACATAATTCAGTAACAATGGAT
PAV_2c01640_273|274s-IBS	AAAAAAGCTTATAATTATCCTTACTAGACCATCGCGTGCGCCCAGATAGGGTG
PAV_2c01640_273|274s-EBS1d	CAGATTGTACAAATGTGGTGATAACAGATAAGTCCATCGCACTAACTTACCTTTCTTTGT
PAV_2c01640_273|274s-EBS2	TGAACGCAAGTTTCTAATTTCGGTTTCTAGTCGATAGAGGAAAGTGTCT
PAV_2c01640_KO_for	ATGAAATCTTTAATTTCCGCTTGC
PAV_2c01640_KO_rev	GGCAGGCAATAATGGAAATCTTGC
1630_XbaI_for	aaTCTAGATGCTTAAGTAAAGGATACGGTGTCTAGTTTTCTAGGCGCCGTATCTTTTATAAAAGGGATGTGGATAATGACACGTCCCAAGATCTC
1630_KpnI_rev	aatcaGGTACCTTACTTCATATTCAACCATTCGG
1640_HindIII_for	aaAAGCTTTGCTTAAGTAAAGGATACGGTGTCTAGTTTTCTAGGCGCCGTATCTTTTATAAAAGGGATGTGGATAATGAAATCTTTAATTTCCGCTTG
1640_KpnI_rev	aatcaGGTACCTTATCGTAACAAGAGCAATCGTTCCATAGC
PaFlag_SphI_for	aatcaGCATGCGTATTAATCACAATATCAGCTC
PaFlag_KpnI_6His_rev	aatcaGGTACCTTAATGGTGATGGTGATGGTGACGGAGCAATTGCAGTACTCCTTGAGG
PaFlag_NcoI_for	aatcaCCATGGGTATTAATCACAATATCAGCTC
PaFlag_XhoI_6His_rev	aatcaCTCGAGTTAATGGTGATGGTGATGGTGACGGAGCAATTGCAGTACTCCTTGAGG

^aArtificial restriction sites are underlined. Lowercase letters indicate artificially introduced bases to improve restriction enzyme cutting.

Isolation of flagella

Isolation of flagella was done according to Montie and Stover (1983). Briefly, 5 mL LB media was inoculated with cells from an LB-swarming plate. After overnight incubation at 37°C, this pre-culture was transferred into 500 mL of LB and incubated again overnight at 37°C. Centrifugation was done at 5000 × g for 15 min at 4°C. The pellet was resuspended in 0.01 M PBS buffer, pH 7.5 (100 mL buffer per 6 g of wet cells). The suspension was blended with a commercial blender (Spar) for 45 s to shear off flagella and subsequently centrifuged at 5000 × g for 30 min at 4°C. The supernatant was further processed and centrifuged at 16 000 × g for 15 min at 4°C, followed by another centrifugation step of the new supernatant at 40 000 × g for 3 h at 4°C. The remaining pellet was resuspended in 1 mL of MilliQ (MQ) water.

Protein expression and purification

Paenibacillus alvei flagellin protein (rHag) was produced as His₆-tagged construct for detection and purification purposes in E. coli BL21 (DE) (rHag_{E. coli}) cells as well as in P. alvei CCM 2051^T (rHag_{P. alvei}). For recombinant expression in E. coli BL21 (DE), cells rHag was produced as a C-terminal His-Tag fusion by PCR, using the primers PaFlag_NcoI_for/PaFlag_XhoI_6His_rev. Amplification was done using P. alvei CCM 2051^T genomic DNA as a template. The His₆-tagged hag amplification product was digested with NcoI/XhoI and cloned into NcoI/XhoI-linearized pET28a(+) (Novagen). Escherichia coli BL21 (DE3) cells were transformed with the corresponding plasmid pET28a_hag_6His. Freshly transformed cells were grown in LB medium (Sambrook et al. 1989) to the mid exponential growth phase (OD₆₀₀ ∼0.6), protein expression was induced with a final concentration of 0.5 mM isopropyl-β-d-thiogalactopyranosid and cultures were grown for additional 4 h at 37°C and 200 rpm. Cells were harvested by centrifugation (4500 × g, 30 min, 4°C).

For homologous expression in P. alvei CCM 2051^T cells, rHag was produced as a C-terminal His-Tag fusion by PCR, using the primers PaFlag_SphI_for/PaFlag_KpnI_6His_rev. The His₆-tagged hag amplification product was digested with SphI/KpnI and inserted into the linearized vector pEXALV digested with the same restriction enzymes. The corresponding plasmid was named pEXALV_hag_6His and transformation of P. alvei CCM 2051^T wild-type cells was performed as described by Zarschler et al. (2009).

His₆-tagged flagellin expressed in E. coli BL21 (DE) and in P. alvei CCM 2051^T cells, respectively, was purified under denaturing conditions using Ni-NTA Agarose (Qiagen) according to the manufacturer’s protocol. Purified protein was dialyzed against MQ water.

Gene knockouts

Disruption of the predicted lipid carrier:UDP-N-acetylgalactosaminyl-transferase gene pglB (PAV_1c00540), and the glycosyltransferase genes PAV_2c01630 and PAV_2c01640 located in close proximity to the P. alvei CCM 2051^T flagella gene hag (PAV_2c01710) was performed as described previously (Zarschler et al. 2009). The Ll.LtrB targetron of pTT_wsfA243 was retargeted prior to transformation into P. alvei CCM 2051^T. Identification of potential insertion sites and design of PCR primers for the modification of the intron RNA was accomplished by a computer algorithm (www.Sigma-Aldrich. com/Targetronaccess). The retargeted Ll.LtrB targetron was subsequently digested with HindIII and BsrGI and ligated into pTT_wsfA243 (Zarschler et al. 2010) digested with the same restriction enzymes, thereby replacing the wsfA targetron.

Competent cells of P. alvei CCM 2051^T ΔwsfP, P. alvei CCM 2051^T ΔPAV2c_01630 or P. alvei CCM 2051^T ΔPAV2c_01640 were transformed by the plasmid pEXALV_hag_6His (Zarschler et al. 2009). To complement the knockouts, ΔPAV2c_01630 and ΔPAV2c_01640 under the native promoter (P(1640)) was amplified from P. alvei CCM 2051^T genomic DNA using the primers 1630_XbaI_for/ 1630_KpnI_rev and 1640_HindIII_for/ 1640_KpnI_rev, respectively. The amplification products were XbaI/KpnI and HindIII/KpnI-digested and inserted into the linearized vectors pNW33N (XbaI/KpnI) and pEXALV (HindIII/KpnI), respectively. The resulting plasmid was named pNW33N_P(1640)_1630 and pEXALV_P(1640)_1640 (Table II) and were transformed into P. alvei CCM 2051^T ΔPAV2c_01630 and ΔPAV2c_01640 cells, respectively (Zarschler et al. 2009).

General and analytical methods

SDS–PAGE was done according to a standard protocol (Laemmli 1970) using a Protean II electrophoresis apparatus (Bio-Rad). Protein bands were stained by CBB. Carbohydrates were stained with periodic acid-Schiff (PAS) reagent (Hart et al. 2003). Western blotting of proteins onto a polyvinylidene difluoride membrane (Bio-Rad) was done using a Mini Trans-Blot cell (Bio-Rad). Detection of the His₆-tag was done with the Li-Cor Odyssey infrared imaging system using anti-His₆ mouse antibody (Invitrogen) in combination with goat anti-mouse IgG-IRDye 800CW conjugate (Li-Cor). Precipitation of proteins trichloroacetic acid is described elsewhere (Sanchez 2001).

Swimming and swarming of P. alvei cells

To test the P. alvei CCM 2051^T wild type, the glycosyltransferase mutants (ΔPAV_2c01630 and ΔPAV_2c01640) and the complemented knockout strains (ΔPAV_2c01630_comp and ΔPAV_2c01640_comp) for their ability to swim or swarm on LB-agar plates, cells were grown to the exponential growth phase (OD₆₀₀ ∼1.0) and 5 µL of each culture was spotted on 1% (semi-solid agar) LB-agar plates (pH 7.0). All P. alvei cells were cultivated at 37°C for 16 h. Pictures were taken using a SPImager (S&P Robotics).

Transmission electron microscopy

To identify potential differences in flagellation of P. alvei CCM 2051^T wild-type and the two glycosyltransferase mutants, cells were grown overnight in LB medium. In addition, the capability of purified recombinant, non-glycosylated Hag from E. coli (rHag_{E. coli}) to self-assemble in MQ-water into flagellar filaments was analyzed by transmission electron microscopy (TEM) and compared with the native flagella isolated from P. alvei wild-type cells.

Thirty microliters of these bacterial suspensions were applied to Formvar- and carbon-coated 300-mesh copper grids (Agar Scientific) that were rendered hydrophilic upon glow discharge using a Pelco easiGlow apparatus (Ted Pella). The grids were incubated for 10 min face down on the cell suspensions. Samples were fixed with 2.5% glutaraldehyde, washed three times with MQ-water, and stained with 1% uranyl acetate solution, pH 4.2, for 40 s (Messner et al. 1986). Samples were investigated using a Tecnai G² 20 Twin TEM (FEI), operating at 120 kV. Pictures were taken with an FEI Eagle 4 K CCD camera (4096 × 4096 pixels).

Mass spectrometry

For the identification of glycopeptides of the Hag flagellin protein of P. alvei, an in-gel trypsin digest of the respective band from multiple lanes on an SDS–PAGE gel was performed as described recently (Posch et al. 2011). (Glyco)peptides were extracted from the gel pieces and subsequently analyzed by MS analysis as described previously (Kolarich et al. 2012).

Two different approaches were used for peptide identification. PMF was performed on an Autoflex Extreme MALDI-TOF-MS (Bruker Daltonics, Bremen, Germany) in positive mode using α-cyano-4-hydroxycinnamic acid as a matrix. In depth, glycoproteomics analyses were done by reversed-phase nanoLC-ESI-IT-MS/MS using C18 reversed-phase chromatography (precolumn: Dionex PepMap100 100 µm (ID) × 2 cm; C18, 5 µm, 100 Å P/N 164564; analytical column: Dionex Acclaim PepMap RSLC 75 µm (ID) × 15 cm; C18, 2 µm, 100 Å; P/N 164534) coupled to a Bruker amaZon Speed ETD ion trap using the following LC conditions: gradient from 2 to 30% solvent B (acetonitrile containing 0.1% formic acid) over 35 min (solvent A: 0.1% formic acid in water); signals were detected over a range from 350 to 1800 m/z, with an SPS target mass set to 900 m/z. The three most intense signals were selected for CID (MS/MS scan range 100–2200 m/z). All ions were detected as protonated signals in positive mode. More details about the instrument settings are given in Supplementary data, PMF_Detected_Peptides.xlsx and Glycopeptide_Fragment_Ions.xlsx according to the MIAPE guidelines (Supplementary data, LCMS_Settings.xlsx) (Kolarich et al. 2013).

The acquired MALDI-TOF-MS and ion trap MS2 data were analyzed using Data Analysis 4.1 and ProteinScape 3.1 (both Bruker Daltonics), MASCOT server 2.3 (MatrixScience) and furthermore the online tool FindPept on the ExPASy server was used (http://web.expasy.org/findpept/) (Wilkins et al. 1999; Gattiker et al. 2002).

A custom P. alvei protein database containing 17 901 entries was retrieved from UniProt (http://www.uniprot.org/uniprot/?query=paenibacillus+alvei) on 12 February 2014 and used for MASCOT protein searches.

The MASCOT search parameters were also modified to include HexHexNAc₂ as a variable modification present on Ser/Thr residues to identify any glycopeptides as well as the putative amino acids attachment sites.

Supplementary data

Supplementary data for this article are available online at http://glycob.oxfordjournals.org/.

Abbreviations

Hex

hexose

HexNAc

N-acetylhexosamine

LB

Luria-Bertani

LC-ESI-MS/MS

liquid chromatography-electrospray ionization-tandem mass spectrometry

MALDI-TOF

Matrix Assisted Laser Desorption Ionization-Time of flight

MQ

MilliQ

MS

mass spectrometry

OD

optical density

PAS

periodic acid-Schiff

PMF

peptide mass fingerprinting

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

TEM

transmission electron microscopy

TPR

tetratricopeptide