Flagellin diversity in Clostridium botulinum groups I and II: a new strategy for strain identification

Abstract

Strains of Clostridium botulinum are traditionally identified by botulinum neurotoxin type; however, identification of an additional target for typing would improve differentiation. Isolation of flagellar filaments and analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) showed that C. botulinum produced multiple flagellin proteins. Nano-liquid chromatography-tandem mass spectrometry (nLC-MS/MS) analysis of in-gel tryptic digests identified peptides in all flagellin bands that matched two homologous tandem flagellin genes identified in the C. botulinum Hall A genome. Designated flaA1 and flaA2, these open reading frames encode the major structural flagellins of C. botulinum. Colony PCR and sequencing of flaA1/A2 variable regions classified 80 environmental and clinical strains into group I or group II and clustered isolates into 12 flagellar types. Flagellar type was distinct from neurotoxin type, and epidemiologically related isolates clustered together. Sequencing a larger PCR product, obtained during amplification of flaA1/A2 from type E strain Bennett identified a second flagellin gene, flaB. LC-MS analysis confirmed that flaB encoded a large type E-specific flagellin protein, and the predicted molecular mass for FlaB matched that observed by SDS-PAGE. In contrast, the molecular mass of FlaA was 2 to 12 kDa larger than the mass predicted by the flaA1/A2 sequence of a given strain, suggesting that FlaA is posttranslationally modified. While identification of FlaB, and the observation by SDS-PAGE of different masses of the FlaA proteins, showed the flagellin proteins of C. botulinum to be diverse, the presence of the flaA1/A2 gene in all strains examined facilitates single locus sequence typing of C. botulinum using the flagellin variable region.

FIG. 1.

TEMs of negatively stained C. botulinum showing the flagellation of group I and group II strains. Although several strains did not produce flagellar filaments, the majority of strains produced several to multiple peritrichous flagella. (A) Type A strain FE9909ACS Alberta (scale bar = 2.5 μm). (B) Group I type B strain FE0507BLP (scale bar = 1.1 μm). (C) Group II type B strain KapB3 has several filaments (scale bar = 1.1 μm); inset panel, group II flagellar filament from strain GA9709EHS with basal body and hook structure attached (scale bar = 0.2 μm). (D) Type E strain Bennett (scale bar = 2.5 μm).

FIG. 2.

SDS-PAGE profiles of flagellar proteins. (A) Group I flagellins. Lane 1, Hall A; lane 2, FE9504ACG; lane 3, 17A; lane 4, FE9909ACS Alberta; lane 5, FE0303A1YO; lane 6, MUL0109ASA; lane 7, PC0101AJO; lane 8, MRB; lane 9, FE0507BLP; lane 10, PA9508B; lane 11, 920A276; lane 12, FE9904BMT; lane 13, FE9508BPD; and lane 14, Langeland. (B) Group II flagellins. Lane 1, 17B; lane 2, KapB3; lane 3, ERuss; lane 4, Bennett; lane 5, GA9709EHS; and lane 6, 610F. Molecular masses in kilodaltons are indicated by arrows. Serotypes of strains are indicated by a letter below each lane. All predominant bands were identified as flagellin by mass spectrometry, with the exception of the 30-kDa molecular mass band in panel B, lanes 1, 2, and 6 (see Table 2 and the text). The minor band at 38 kDa in lanes 1, 2, and 6 was also identified as flagellin.

FIG. 3.

Identification of putative flagellin genes in the Hall A genome sequence. (A) Arrows indicate the ORF identified as a putative flagellin with numbers indicating the location in base pairs of the ORF in the Hall A genome DNA sequence. (B) Comparison of the protein sequences of all Hall A putative flagellins and flagellin homologs in C. tetani and B. subtilis. Of the putative Hall A flagellins, CBO2730 and CBO2731 are most similar in protein sequence to the known structural subunit Hag of B. subtilis flagellin. The scale above the dendrogram is proportional to the similarity between sequences. Bootstrap values from 2,000 simulations are included adjacent to each branch.

FIG. 4.

nLC-MS/MS analysis matched peptides from the tryptic digests of the expressed flagellin proteins to FlaA1 (Hall A) or FlaB (Bennett). (A) Protein sequence for Hall A FlaA1/CBO2731. Peptides in bold and underlined were identified following MS/MS of various flagellin proteins from different strains of group I and group II C. botulinum. At least one peptide matching the Hall A FlaA1 sequence was identified in all flagellin proteins examined. Amino acids 142 and 145 are indicated in bold with italics: these residues differ between FlaA1 and FlaA2 protein sequences (E₁₄₂→ K₁₄₂ and K₁₄₅→ D₁₄₅, respectively). (B) FlaB from type E strain Bennett. Peptides identified by MS/MS of FlaB are in bold and underlined. While some peptides (i.e., AGDDAAGLAISEK) were present in both FlaA1/A2 and FlaB, a number of peptides matched only the sequence predicted for FlaB. For MASCOT scores and the number of peptides from each putative flagellin protein matching Hall A FlaA1 or FlaB (when applicable), see Table 2.

FIG. 5.

Dendrogram analysis of protein sequences for known flagellins from different members of the genus Clostridium. Protein sequences for C. botulinum FlaA1/A2 from strain Hall A and strain Bennett and FlaB from strain Bennett were compared by generating a dendrogram using the UPGMA and nearest-neighbor-joining methods in Bionumerics 4.5 with a large maximum gap size of 98 to accommodate flagellins with large differences in molecular mass. Flagellins indicated with an asterisk have been documented in the literature as expressed proteins. In the case of ORF pairs, where it was not specified which was expressed, both ORFs were given an asterisk. Gray shaded boxes show subclusters containing C. botulinum flagellins. C. botulinum Hall A FlaA1 and FlaA2 sequences were obtained from the Hall A genome project at www.sanger.ac.uk/Projects/C_botulinum. For all other flagellin sequences, GenBank accession numbers are as follows: C. botulinum Bennett FlaA1 (DQ845000), FlaA2 (DQ845001), and FlaB (DQ658239.1); C. acetobutylicum ATCC 824 (genome strain) FlaC (AAC16553.1), CAC1634 (AAK79601.1), CAC1555 (AAK79522.1), CAC2211 (AAK80168.1), and CAC2167 (AAK80125.1); C. beijerinckii NCIMB 8052 CbeiDRAFT4550 (ZP_00907552.1) and CbeiDRAFT4535 (ZP_00907537.1); C. chauvoei FliC (BAB13814.1) and FliB (C) (BAB87728.1); C. difficile FliC (AAF09167.1); C. haemolyticum FliA (H) (BAB87738.1); C. novyi FliA (B) (BAB87737.1), FliB (A) (BAB87735.1), and FliA (A) (BAB87734.1); C. septicum FliA (S) (BAB87729.1), FliC (S) (BAB87732.1), and FliB (S) (BAB87730.1); C. tetani E88 (genome strain) CTC01679 (AAO36215.1), CTC01691 (AAO36226.1), and CTC01715 (AAO36250.1); C. thermocellum CtheDRAFT 2324 (ZP_00504695.1) and CtheDRAFT 2323 (ZP_00504694.1); C. tyrobutyricum Fla (CAB44444.1); and B. subtilis 168 genome strain Hag (NP_391416.1). The scale above the dendrogram is proportional to the similarities between sequences. Bootstrap values from 2,000 simulations are included adjacent to each branch.

FIG. 6.

Dendrogram analysis of the DNA sequence of the flaVR. A dendrogram was constructed for the flaVR DNA sequences from 80 C. botulinum strains using the UPGMA with default settings followed by a global alignment and the nearest-neighbor-joining method incorporating the Kimura 2P parameter. The designation and BoNT serotype for each strain are listed to the right of the tree. Light gray bars flank flaVR sequences clustering into group I or group II. Dark gray bars flank sequences of assigned flaVR types within group I or group II. The scale above the dendrogram is proportional to evolutionary time, due to incorporation of the Kimura 2P parameter for nucleotide sequences. Bootstrap values from 2,000 simulations are included adjacent to each branch.