Deciphering the genomes of motility-deficient mutants of Vibrio alginolyticus 138-2

Bacterial cultivation and genomic DNA isolation

All Vibrio alginolyticus strains used in this study are summarized in Table S1 and the timeline in which the mutants were constructed is summarized in Fig. 1. All mutants have been isolated by focusing on the two types of flagellar systems of Vibrio, with strains defective in the Polar flagellum denoted Pof⁻ and those defective in the lateral flagella denoted Laf⁻. Strain VIO5 is a lateral flagella-deficient mutant (Pof⁺, Laf⁻) generated by EMS mutagenesis of VIK4, a spontaneous rifampicin-resistant strain obtained from the wild-type strain 138-2 (Pof⁺, Laf⁺) (Okunishi, Kawagishi & Homma, 1996). Strains YM4 and YM5 are lateral flagella-deficient mutants (Pof⁺, Laf⁻) obtained by UV treatment of strain 138-2, whereas strain YM19 is a polar flagellum-deficient mutant (Pof⁻, Laf⁺). This mutant was spontaneously isolated from YM17, a UV-treated YM5-derived strain with impaired polar and lateral flagellar formation (Kawagishi et al., 1995). Strain YM51 is a motility-deficient mutant of strain YM5, and strains NMB75, 82, 88, 93, 95, 98, 99, 102, 103, 105, 106, 111, and 116 are motility-deficient mutants of strain YM4 (Homma et al., 1996; Nishioka et al., 1998). Strains NMB136, NMB155, and KK148 are motility-deficient mutants derived from VIO5 (Kojima et al., 1999; Kusumoto et al., 2006). All Vibrio strains were cultivated in VC medium (0.5% (w/v) tryptone, 0.5% (w/v) yeast extract, 0.4% (w/v) K₂HPO₄, 3% (w/v) NaCl, and 0.2% (w/v) glucose). Genomic DNA was isolated in the late logarithmic phase using a genomic DNA isolation kit (Promega, WI, USA).

Library construction and DNA sequencing

The genomic DNA was quantified by SYBR Green fluorescence method (Leggate et al., 2006) and the purity was estimated from the absorption spectra (approximately A260/A280 ratio and A260/A230 ratio) using a NanoVue (GE Healthcare, Chicago, IL, USA). Genomic DNA libraries were constructed by the tagmentation method using the Nextera XT DNA Library prep kit (Illumina, San Diego, CA, USA) or a home-made transposase (Picelli et al., 2014), and were sequenced by the MiSeq 600PE v.3 kit or 500PEv.2 kit (Illumina, San Diego, CA, USA). All sequence data were deposited in the DRA Data bank (Accession: DRA012242–DRA012245).

De novo assembly and construction of complete V. alginolyticus strain VIO5 genome

V. alginolyticus strain VIO5 has been used in many sodium-driven polar flagellar studies (Li, Kojima & Homma, 2011). In this study, we first determined the complete genome sequence of strain VIO5. Paired-end sequencing reads (2 × 300-bp) were trimmed by quality value, and adapter sequences were removed using Trimmomatics (Bolger, Lohse & Usadel, 2014). Trimmed reads were de novo assembled with SPAdes v.3.1 (Bankevich et al., 2012). From approximately 150 assembled contigs, we selected 27 contigs larger than 3 kb whose coverage was close to the most frequent value. Primer 3 (Rozen & Skaletsky, 2000) was used to design forward primers specific to both ends of each contig, except for the repetitive contigs (the primer list can be found in Table S2). The direction and order of the long contigs were predicted by alignment with the genomic sequences of the most closely related Vibrio strain (Vibrio sp. EX25; accession numbers NC_13456 and NC_13457) using Mauve software (Darling et al., 2004). During this operation, the orientation and order of 23 of the 27 contigs were successfully determined. Two contigs formed a circular chromosome (chromosome II) and 21 contigs were linked to form four large assemblies. To determine the linkage order between these four large assemblies and the remaining four contigs that were not aligned in the EX25 genome, PCR experiments were performed using all combinations of primer sets. The amplified PCR fragments were checked for uniformity and size using 0.8% agarose gel electrophoresis and purified using a PCR fragment recovery kit (Promega, Madison, WI, USA). The recovered DNA fragments were sequenced using the MiSeq 500PE v.2 kit (Illumina, San Diego, CA, USA). Short-read sequences from each PCR fragment were assembled using SPAdes v 3.1 (Bankevich et al., 2012), and the contig with the largest size and highest coverage was adopted as the PCR fragment. For the three regions in which two large contigs appeared after assembly (The order of contigs are appeared in Table S3), it was inferred that the rRNA operons were arranged in tandem. Therefore, we designed primers in both directions in the center of the connecting contig (approximately 200 bp, contig 175 in Table S4) expected to be between these tandemly arranged rRNA operons, and amplified the two rRNA operons independently by PCR. Both the amplified fragments were independently determined and connected using a central contig. Chromosome I of strain VIO5 was completed by manually combining all contigs and PCR fragment sequences. In regions where discrepancies were found between the contig and PCR fragment sequences, the results of the PCR fragment were used preferentially for linkage because the contig ends may contain polymorphisms due to repeat sequences. Finally, sequencing reads were mapped against the full-length genome sequence using BWA (Li & Durbin, 2009) to confirm the absence of assembly errors. The Integrative Genomics Viewer (IGV) (Thorvaldsdóttir, Robinson & Mesirov, 2013) was used for map visualization.

Complete genomes of V. alginolyticus 138-2, YM4, and YM19, using V. alginolyticus VIO5 as a reference strain

Next, we developed a workflow aimed at completing the genome structure of the target Vibrio strains in cases where short-read sequencing technology and very closely related reference strains were available; however, long-read sequencing technology was unavailable because of higher sequencing costs or challenges in high molecular weight DNA extraction. This workflow (for details, see Fig. S1) was applied to assemble the complete genome structures of three strains (V. alginolyticus strains 138-2, YM4, and YM19) that are closely related to strain VIO5. Paired-end sequencing reads were trimmed based on their quality using fastp v.0.20.0 (Chen et al., 2018) and used as input data (hereafter, WGS reads). WGS reads were assembled de novo using SPAdes v.3.13 (Bankevich et al., 2012) to produce error-corrected WGS contigs by two cycles of polishing with Pilon v1. 23 (Walker et al., 2014). WGS contigs of less than 1 kbp in length or contigs with abnormal coverage were excluded, and the terminal 127-bp of the remaining WGS contigs were trimmed. These contigs were designated long and normal coverage contigs. For the selection method based on the coverage count, the median of the average read counts of the contigs up to the top five in length was set as value C. Contigs whose coverage was more than twice or less than half of value C were removed as abnormal coverage contigs. Bbmap v.37.62, published by JGI (SourceForge, 2023) was used for coverage calculation. LN contigs with a 16-mer frequency greater than or equal to 2 were hard-masked using Primer3_masker (Kõressaar et al., 2018). Then, primer3 (Untergasser et al., 2012) was used to design outward primers within 1 kb at both ends of each contig. The specificity of the designed primers was checked using FastPCR (Kalendar et al., 2017) (the primer list can be found in Table S2). The workflow from short reads to LN contigs and primer design is available on GitHub: script genome_quest (https://GitHub.com/kazumaxneo/genome_quest). Minimap2 (Li, 2018) was used to align each LN contig with the complete genome sequence of the V. alginolyticus VIO5. PCR was performed using primers designed for each LN contig. Multiplexed PCR products were sequenced using MiSeq and individually assembled, as described in the previous section. LN contigs and Locally Assembled PCR fragment (LA) contigs were connected using CAP3 (Huang & Madan, 1999.). Two circular chromosomes were also identified. Finally, WGS reads were mapped to the two assembled chromosomal DNA sequences using minimap2 (Li, 2018), to correct as many errors as possible using the following variation detection tools: breseq (Deatherage & Barrick, 2014), GATK HaplotypeCaller v 3.8 (DePristo et al., 2011), minimap2 paftool (Li, 2018), and SV Quest (v1.0) (https://GitHub.com/kazumaxneo/SV-Quest).

The complete genome sequences of the four strains have been published in the database under the accession numbers AP022859–AP022866 (DDBJ).

Variation analyses in the mutant strains

Genomic libraries were constructed and sequenced by MiSeq using paired-end sequencing (2 × 300 bp) for 17 strains generated by EMS mutagenesis and one strain, YM51, derived from YM5. Using the genome of V. alginolyticus 138-2 as a reference, all variants (SNVs, indels, RNVs, and LSVs) present in the three strains were extracted using Equation (Deatherage & Barrick, 2014) and other tools, as described in the previous section.

Core gene phylogenetic tree inference

Each mutant genome was created using the gdtools APPLY command from the output of the Breseq variant calling against the 138-2 genome sequence (Deatherage & Barrick, 2014). Core genome alignment of the 138-2 sequence and the derived sequence of the mutant strain was performed using Parsnp (https://GitHub.com/marbl/parsnp). Unreliable alignment blocks were excluded based on the Parsnp criteria. A phylogenetic tree was manually constructed based on the number of variations.

Genome structures of V. alginolyticus strain 138-2 and derived mutant strains

We determined the complete genome structure of V. alginolyticus strain 138-2 and three mutant strains: VIO5, YM4, and YM19. These strains have been widely used for the functional analysis of polar flagella of the genus Vibrio (Kawagishi et al., 1995). Similar to the genomes reported for other Vibrio spp., the genomic DNA consists of two circular chromosomes and does not harbor any plasmids (Okada et al., 2005). The genome size was 5,185,395 bp for strains 138-2 and VIO5 and 5,185,324 bp for strains YM4 and YM19. DFAST annotation predicted 4,601 protein-coding sequences (CDSs) for strain VIO5, 4,602 for 138-2, and 4,603 for YM4 and YM19. Thirty-seven rRNA genes (twelve 16S rRNA, twelve 23S rRNA, and thirteen 5S rRNA) and one hundred and sixteen tRNA genes were assigned to all four strains (detailed genomic information is provided in Table S5).

Variation sites in V. alginolyticus strains VIO5, YM4, and YM19

V. alginolyticus strain VIO5 is a lateral flagellar-deficient mutant that arises from EMS mutagenesis of VIK4 (rifampicin-resistant), which results from a spontaneous mutation in the parent strain 138-2 (Fig. 1). The VIO5 strain has four SNVs compared to its parent strain 138-2, three on chromosome 1 and one on chromosome 2 (Table 1). The rifampicin-resistant phenotype of VIO5 can be attributed to a mutation in chromosome I (position 3,206,619) of the VIO5 genome (Table 1). This mutation leads to a Q513L amino acid substitution in the RNA polymerase beta subunit, which has been reported as a causal SNP of rifampicin resistance in E. coli (Campbell et al., 2001). Two other SNVs in chromosome 1 are on the hslO gene and Vag1382_04350 gene, which are predicted to encode heat shock 33 kDa chaperonin and glutamate synthase, respectively. Neither of these two genes is thought to be involved in the lateral flagellar deficient phenotype of VIO5. A mutation on chromosome II (position 1,534,464) introduced a stop codon at codon 64 in motY2 (TGG→TGA). Given that this is the sole mutation observed in chromosome II of VIO5, and considering that all lateral flagellar genes are present on chromosome II, the lateral flagellar deficiency was attributed to a mutation in the motY2 gene.

V. alginolyticus strain YM4, a lateral flagellar-deficient mutant, and strain YM19, a polar flagellar-deficient mutant, were generated through UV mutagenesis of strain 138-2 (Fig. 1). Thirteen mutations were common between YM4 and YM19 (Table 1), suggesting that these mutations accumulated during the early stages of UV mutagenesis. Since YM4 and YM19 exhibit different phenotypes for the two types of flagella, a mutation on chromosome II (position 255,362), exclusive to YM4, resulting in a serine substitution at Gly53 in the lateral flagellar P-ring protein FlgI, is the likely cause of the YM4 lateral flagellar-deficient phenotype. Similarly, a mutation on chromosome I (positions 2,299,407), found only in YM19, introduced a stop codon at codon 295 (CAG→TAG) in flhA, which encodes the polar flagellar export apparatus protein and presumably resulted in a truncated FlhA and the polar flagellar-deficient phenotype of YM19.

The VIO5 strain generated by EMS mutagenesis had only four SNVs with no insertion or deletion mutations, but the YM4 and YM19 strains created by UV mutagenesis had two deletion mutations and five repeat number variation mutations, in addition to seven and six SNVs for YM4 and YM19, respectively (Table 1).

Variation sites in other motility-deficient mutant strains

Vibrio strains NMB136, NMB155, and KK148 were generated from strain VIO5, whereas strains NMB75, NMB82, NMB88, NMB93, NMB95, NMB98, NMB99, NMB102, NMB103, NMB105, NMB106, NMB111, and NMB116 were generated from strain YM4. All 16 strains were generated through EMS mutagenesis and screened as mutants that could not form a swimming ring on a soft agar plate; motility was observed under dark-field microscopy (Homma et al., 1996). In Table S6 summarizes all detected mutation sites in the genome of each strain compared with the 138-2 strain genome. The mutations were categorized into three types: single nucleotide variations (SNVs), short insertions/deletions (indels), and short tandem repeat number variations (RNVs). Considering these variations, the number of SNVs, indels, and RNVs detected in each strain was used to create a pedigree for the strain (Fig. 2). Because NMB136, NMB155, and KK148 were generated from the VIO5 strain at different times, these three strains carried completely independent mutations. Conversely, the 14 NMB strains generated from YM4 almost simultaneously carried nine common mutations (five SNVs, two indels, and two RNVs) in addition to various unique mutations ranging from two to 54. Considering the individual strains analyzed, the mutations they carried were counted independently, resulting in an average of 20.6 ± 12.7 mutations (mean ± standard deviation).

Putatively responsible variations for motility-deficient mutant strains

The motility-deficient mutants analyzed in this study can be classified into three types: those with no or incomplete flagella (Fla⁻ type), those with chemotaxis problems (Che⁻ type), and those with an increased number of polar flagella (Pof^m type). Another type, the Mot⁻ type, which has abnormalities in the rotation apparatus, was absent in the analyzed mutants. These three types of motility-deficient mutants had 10–30 variation sites, but all had mutations in a known flagella-related gene (Figs. 3A and 3B). The Fla⁻ type mutants, NMB103 and NMB116, had mutations in the flgL gene, leading to the observation of only the hook structure without visible flagellar filaments, which explains their flagella-deficient phenotype. Che⁻ type mutants included NMB75, NMB82, NMB88, NMB91, NMB93, NMB95, NMB98, NMB99, NMB102, NMB105, NMB106, NMB111, and NMB136. NMB82 and NMB105 harbored mutations in the cheA gene, NMB91 and NMB98 harbored mutations in the zomB gene, NMB93 and NMB136 harbored mutations in the cheY gene, and NMB88, NMB95, NMB99, NMB102, and NMB106 harbored mutations in the fliM gene. These genes are part of the gene cluster responsible for the chemotactic response in Vibrio, especially the change in flagellar rotation (Fig. 4). Pof^m type mutants included the KK148 and NMB155 strains. The KK148 strain harbored a mutation in the flhG gene, whereas the NMB155 strain harbored a mutation in the fliM gene.

Responsible genes for lateral flagellar-deficient mutants

In each of the three lateral flagellar-deficient strains (VIO5, YM4, and YM51), the reasons for the flagellar-deficient phenotype were different. In the VIO5 strain, the lateral flagellar deficiency was attributed to a mutation in the motY2 gene. Intriguingly, a single mutation in a structural gene can result in the complete loss of flagellar gene expression. The motY2 gene was one of the earliest members to be expressed in the lateral flagellar expression hierarchy (Stewart & McCarter, 2003) and was positioned at the head of the operon (Fig. 3B). Therefore, a mutation leading to a premature stop codon in the motY2 might significantly impact the translation of downstream lafK genes, a σ⁵⁴-dependent regulator required for the expression of class 2 genes of lateral flagella (Stewart & McCarter, 2003). In the strain YM4, a mutation in the flgI gene may be strongly related to lateral flagellar deficiency. This gene product FlgI is the major component protein that forms the P ring located in the periplasmic region, and when P ring formation is incomplete, protein transport that constitutes hook and flagellin fibers is impaired and normal flagellin formation is inhibited. In the P-ring protein FlgI of E. coli, a point mutation such as G21C causes a complete loss of motility (Hizukuri et al., 2008), so it is not surprising that a mutation in FlgI(G51S) of strain YM4 suppresses expression of lateral flagellar. It has also been shown that in Salmonella when the anti-sigma factor FlgM is not expelled by the protein transport hook-basal body apparatus, the sigma factor FliA does not function (Hughes et al., 1993; Kutsukake, 1994) and transcription of the class 3 genes involved in flagellar formation does not occur (Aldridge et al., 2006). It remains to be elucidated to what extent lateral flagellar formation occurs in the strain YM4. In the strain YM51, a frame-shift mutation in the flhA2 gene (Fig. 3B) may be strongly related to lateral flagellar deficiency. YM51 was originally isolated from YM5 (Fig. 1) and exhibits a Fla⁻ type phenotype. In the flhA gene region, seven consecutive G bases were found to have increased by one, causing truncation of the FlhA protein to approximately 100 amino acids. In Paenibacillus glucanolyticus, it was reported that swarming was suppressed by reversible hotspots that reduced the number of eight consecutive A bases to seven and that the strain could easily revert to a swarm-competent state (Hefetz et al., 2023). Thus, there appear to be two types of lateral flagella-deficient Vibrio mutants: a strong phenotype with almost irreversible mutations, such as VIO5 and YM4, and a leaky phenotype that is relatively easy to revert, such as YM51. A causal mutation in YM5’s lateral flagellar-deficient phenotype may be the same as in strain YM51.

Responsible genes for polar flagellar-deficient mutants

Polar flagellar-deficient mutants can be divided into three categories based on microscopic observations of movement: Che⁻ mutants, in which the direction of flagellar rotation is dysregulated; Pof^m mutants, in which the number of polar flagella is increased; and Fla⁻ mutants, in which have no flagella at all.

Among the Che⁻ type mutants, the fliM gene, identified as the causal gene for many mutations, appears to be involved in flagellar rotation control and the regulation of flagellar numbers (Homma et al., 2022) (Figs. 3A and 4). This suggests versatile roles of the fliM gene in governing flagellar expression. The NMB75 shows to smoothly swim with reduced response to phenol (Homma et al., 1996). The NMB75 strain harbored a mutation in the cheR gene, resulting in a leaky phenotype due to partial signal transmission by CheA, partially affected by the deficiency of CheR activity in methylating chemoreceptors. Strain NMB88, NMB95, NMB99, NMB106 and NMB136 swim smoothly without much tumbling by locking the direction of flagellar rotation to CCW. Strain NMB102, on the other hand, has its flagella locked in the CW direction of rotation and swims constantly backward. NMB136 has a defective mutation of CheY (nonsense mutation of W57), which may be the reason for the observed CCW-locked flagellar motion. NMB88, NMB95, NMB99, and NMB106 all have mutations in FliM, possibly weakening its interaction with phosphorylated CheY, which is required for tumbling. On the other hand, in strain NMB102, R49P, one of the two mutations in FliM, has been shown to be important for the CW-locked phenotype, in which structural changes in FliM itself result in a CheY-independent rotational motion fixed in the CW direction (Takekawa et al., 2021b). NMB111 showed weakly reduced swimming ability due to mutations in the flhG gene, which regulates flagellar number, and flagella were rarely observed with the FlhG(D171A) mutation (Ono et al., 2015). The reduced swimming ability of NMB111 may be due to the reduced flagellar number caused by the FlhG(D171N) mutation. Thus, the NMB111 strain may be included in the Fla⁻ type mutants.

In two Pof^m type mutants, the KK148 strain harbored a mutation (Q109*) in the flhG gene and a defective flhG gene product has been shown to form multiple polar flagella (Kusumoto et al., 2006). Whereas the NMB155 strain harbored a mutation (E9K) in the fliM gene, and it has been shown that the FliM(E9K) mutation changes flagellar numbers (Homma et al., 2022). Based on the evidence, the fact that the fliM gene, which has been implicated in chemotaxis, also plays a significant role in regulating flagellar number in the Che⁻ type mutants is highly intriguing.

The YM51 strain, similar to NMB103 and NMB116, showed only a hook structure without visible flagellar filaments, indicating that the assembly of the polar flagellar filament was impaired (Nishioka et al., 1998). However, no mutations were found in flagellar structure genes, including the flgL gene, but a mutation was detected in the rpoN gene, which has been reported to play an important role in polar flagella formation (Kawagishi et al., 1997). Although the YM14 strain used for cloning the rpoN gene was not included in the current genome analysis, it is highly plausible that YM51 has a mutation similar to that of YM14.

Comparison of two flagellar systems

Since the two entire flagellar systems, the polar and lateral flagellar systems, are homologous to each other, it is very interesting to examine the similarity of each flagellar gene and whether there are other paralog genes with similar functions in the genome. Therefore, for the 104 flagellar-related genes that appeared in Fig. 3, we examined the paralog genes in the genome of strain 138-2 by amino acid homology and compared the corresponding genes in the polar and lateral flagella, which are summarized in Table 2. Many paralog genes were detected in the genome for the genes constituting the Che protein group of the signal transduction system (CheA, CheY, CheV, CheW, etc.), however, there were no functionary paralogous genes for flagellar-motility system that showed full-length homology, but limited to a few regions such as histidine kinase domain or response regulator domain(A list of paralogs of flagellar-related genes found in the genome is provided in Table S7). This is inferred from the fact that only the cheY gene mutation located in the flagellar gene cluster region (Fig. 3) resulted in a motility-deficient phenotype. The flagellar-related genes are divided into two groups: genes encoding proteins that form the flagellar structure and genes encoding proteins that regulate the expression of flagellar genes. In both cases, the paralogs of the two flagellar systems were the most closely related genes in the genome, and the only multiplicated genes were flagellin genes encoding the polar flagella (A list of amino acid identities among the seven flagellar proteins is provided in Table S8). In the two flagellar systems, a certain degree of amino acid identity was observed between the proteins constructing the hook, L-ring, P-ring, rod, MS-ring, C-ring, and T3SS (type three secretion system) (Table 2). However, proteins related to the regulation of flagellar gene expression, proteins constituting the T-ring and H-ring, and proteins constituting the Stator-Motor showed very low amino acid identity, and in many cases, no homologous protein was found in the lateral flagellum (Table 2). If the two flagellar motor systems are somewhat similar, it is possible that homologous proteins, especially those constituting the T-ring and H-ring, are located elsewhere on the chromosome. Genome analysis of a large number of motility deficient mutants using a strain deficient in the polar flagellum (YM19) will most likely reveal genes involved in lateral flagellar formation.

The potential of powerful forward genetics as a comprehensive analysis of systems

Through genomic analysis of 18 mutant strains selected by a combination of EMS mutagenesis and screening for motility-deficient phenotypes, it was found that these strains contained 3 to 75 gene mutations in addition to the one or two gene mutations presumed to be strongly related to the phenotype (motility). Gene manipulation to selectively disrupt (or introduce mutations in) specific genes is necessary to determine the gene responsible for the phenotype, and in Vibrio alginolyticus, several genes have been identified as flagellar-related genes by mutagenesis (Homma et al., 2022; Kitaoka et al., 2013; Takekawa et al., 2021a, 2021b). In this study, six genes involved in the regulatory system of flagellar rotation were enriched (i.e., several different types of mutations were concentrated in the same gene) in the analysis of only 18 strains (Fig. 4), suggesting the possibility of comprehensively extracting genes in the entire system by increasing the number of mutant strains analyzed. Since most of the 18 strains were selected for the Che⁻ phenotype (Homma et al., 1996), it is possible that this selection pressure narrowed down the number of genes to a few genes out of approximately 50 genes involved in a polar flagellum formation. Thus, genome analyses of a large number of completely random, independent motility-deficient mutants could extract additional genes involved in the flagellar system (if the genes are not essential for growth). This means that even for microorganisms that cannot be genetically manipulated, a sufficient amount of mutant strain analysis, using a well-designed selection pressure, could efficiently detect the various phenotypic components that build a living system.