V. gazogenes PB1 (KCTC: 12695, ATCC: 29988) was obtained from the Korean Collection for Type Cultures (KCTC). For genome extraction, V. gazogenes PB1 was cultivated as described in an earlier study (Vijay et al., 2022b). The genomic DNA was shipped, in a freeze-dried form, at room temperature for genome sequencing. Whole-genome sequencing and analysis was performed as described previously by Vijay et al., 2022b. For genomic analysis, systems and subsystems were identified using Pathosystems Resource Integration Center (PATRIC) (Davis et al., 2020) (https://patricbrc.org) in conjunction with Rapid Annotation Subsystem Technology (RAST) (Brettin et al., 2015). The genome map was generated using the Circular Genome Viewer tool of the Bacterial and Viral Bioinformatics Resource Center (BV-BRC; https://www.bv-brc.org) (Olson et al., 2023). REBASE (Roberts et al., 2015) (http://rebase.neb.com) was searched to establish the identity of candidate restriction-modification enzymes. Insertion sequences present in the genome were assessed using genome annotations and ISESCAN 1.7.2.3 (Xie and Tang, 2017). PHASTER (Arndt et al., 2016) (https://phaster.ca) was used to identify prophage sequences in the genome. Genomic islands were assessed using IslandPath-DIMOB 1.0.6 (Bertelli and Brinkman, 2018).

The genome of V. gazogenes PB1 was compared with various members of the Vibrio genus as described within the text. Progressive MAUVE (Darling et al., 2010) was used to align the genomes and to identify locally collinear blocks (LCB). Average nucleotide identity (ANI) was computed using the OrthoANIu algorithm (Yoon et al., 2017) (https://www.ezbiocloud.net/tools/ani) and the average amino acid identity (AAI) was computed using EzAAI 1.2.1 (Kim et al., 2021). The percentage of conserved proteins (POCP) was determined based on the method suggested by Qin et al., 2014. Homologous proteins in V. gazogenes PB1 required for cellulose and starch degradation were identified by performing a BLAST analysis of the proteins encoded by V. gazogenes PB1 using protein sequences from other Vibrio species obtained from UniProt (https://www.uniprot.org).

RNA was isolated from the bacterial pellets using TRIzol Reagent (Invitrogen, Waltham, MA, USA) protocols. A total of 2 µg RNA was DNase-treated and reverse transcribed to cDNA using Promega A3500 kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Approximately 50ng of cDNA was amplified using power SYBR^® Green PCR master mix (Applied Biosystems, Waltham, MA, USA) in a 40-cycle reaction carried out in a QuantStudioTM 7 Flex System (Applied Biosystems, Waltham, MA, USA). Melt curve analysis was performed. The 16S gene was used as the endogenous control and for normalisation of the expression levels of the candidate genes. RT-qPCR experiments were performed in duplicates and repeated three times for two pairs of primer sets.

CRISPR loci were predicted using CRISPRCasFinder (Couvin et al., 2018) (https://crisprcas.i2bc.paris-saclay.fr) with default parameters. CRISPR arrays with evidence level 1 or 2, which are indicative of false-positive or spurious arrays, were not considered for further analysis. All predicted loci were manually checked, and arrays located in the coding regions were discarded. To identify potential targets of CRISPR spacers, NCBI BLASTN searches of these sequences were performed. CRISPRTarget (http://bioanalysis.otago.ac.nz/CRISPRTarget/crispr_analysis.html) was used to identify potential targets of CRISPR spacers (Biswas et al., 2013). BLASTN search of the spacer sequences was performed against the GenBank phage and Refseq plasmid databases to identify matches with at least 90% identity and 90% query coverage. Genomic comparisons and gene arrangements were generated using Easyfig 2.2.5 (Sullivan et al., 2011).

For evaluating prodigiosin production in the presence of various carbon sources, cultivation was carried out in minimal media as described previously (Vijay et al., 2022b). For the quantification of prodigiosin, a 10 μl cell culture was added to 190 μl of acidified ethanol in a microplate well and mixed. The absorbance was recorded at 535nm, as described by Vijay et al., 2022b. Cell culture density was determined by measuring the scattering of light at 650nm. Viability of cells were evaluated by streaking out on growth on LB medium/agar plates supplemented with 3% (w/v).

All quantitative data represent the means ± SD of at least 3 independent culture experiments. Any significant differences between sample treatments were determined using a one-way ANOVA test followed by post-hoc analyses to identify the sample treatments which showed statistically significantly differences. A p value of less than 0.05 was considered significant.

The genomic DNA of V. gazogenes PB1 was used to generate Illumina and PacBio sequencing libraries. After quality control and filtering of raw reads, de novo genome assembly of V. gazogenes PB1 was performed using Unicycler version 0.4.9 by combining PacBio and Illumina sequencing data. This resulted in two circular contigs ( Table S1 ). Chromosome 1 is 3.5 Mbp in size with 45.3% GC content (NCBI Accession: CP092587.1/NZ_CP092587.1), and chromosome 2 is 1.2 Mbp in size with 45.1% GC content (NCBI Accession: CP092588.1/NZ_CP092588.1). Based on NCBI PGAP annotations, the genome encodes 4178 genes, of which 3988 are protein-coding, and 76 are pseudogenes. The genome also has 114 RNA-encoding genes, which includes 87 tRNAs, 22 rRNAs and 5 non-coding RNAs ( Figure 1 ). Assessment of the genome using BUSCO version 5.2.2 based on the vibrionales_odb10 dataset indicated the presence of 1378 out of a core of 1445 Vibrionales genes (95.4%); 1275 were single-copy and 3 were duplicated BUSCOs. Annotated genes were grouped into subsystems in PATRIC using RAST. The most abundant subsystems were metabolism (92), protein processing (44), energy (32), stress response, defence and virulence (36), membrane transport (22), DNA processing (21), and cellular processes (18) ( Figure S1 ).

To evaluate its potential for pathogenicity, the genome of V. gazogenes PB1 was analysed in PATRIC for virulence factors. A total of 55 genes were identified with virulence potential ( Table S2 ). Based on PATRIC, these were related to adherence, secretion, regulation, motility, and metabolism and included dksA, fur, ompR, rpoE, rpoS, phoU, fleN and flgP. Genes relating to quorum sensing (luxS), a type VI secretion system (T6SS) (hcp2, vipA, vipB) antimicrobial resistance (catB, qnrB), and efflux systems (emrAB-tolC, macA/macB/tolC) were also encoded in the genome ( Table S3 ). An inspection of the genome for potential restriction endonucleases (REase) identified the annotated CDS “MvaI/BcnI restriction endonuclease family protein” which is a putative Type II restriction enzyme with potential specificity for the motif CCWGG. Though methylation (M) and DNA sequence-recognition (S) subunits of a putative methyltransferase (MTase) were identified on chromosome 1, a required restriction enzyme was not identified to complete a Type I REase (Loenen et al., 2014). Based on analysis by ISESCAN and the genome annotations, 48 transposases were identified and annotated including IS110, IS200/IS605, IS256, IS3, IS30, IS630, ISNCY family transposases as well as tnpA, DDE-type integrase/transposase/recombinase and Mu transposase C-terminal domain-containing protein. PHASTER identified four prophage regions in the V. gazogenes PB1 genome, of which two regions are intact, one region is incomplete, and one region is questionable ( Table S4 ). The two intact regions have proteins with the highest number of hits from Vibrio phage VP882 (NCBI RefSeq accession: NC_009016) located on chromosome 1, and Vibrio_12B12 (NCBI RefSeq accession: NC_021070) located on chromosome 2. Genomic islands (GIs), which are indicative of horizontal gene transfer, were assessed using IslandPath-DIMOB. Six GIs were located on chromosome 1, and four on chromosome 2 ( Table S5 ).

The most similar genome was that of Vibrio genus ATCC 43942 with average nucleotide and average amino acid identities of 94.22% and 96.66%, respectively ( Table S6 ). A comparison of the two proteomes in PATRIC showed that around 6% of the proteins shared 100% sequence identity, while 83.6% shared 90-99.99% identity ( Figure S2 ). The percentage of conserved proteins (POCP) between the two genomes was computed to be 85.46%. Chromosome 1 of V. gazogenes PB1 is 3.52 Mbp and is comparable to the 3.47 Mbp chromosome 1 of V. gazogenes ATCC 43942. Alignment of chromosome 1 of V. gazogenes PB1 and V. gazogenes ATCC 43942 using progressiveMauve indicated that the genome orientation is largely preserved apart from two shorts regions of lengths which have been swapped and inverted ( Figure S2 ). Chromosome 2 of V. gazogenes PB1 is 1.2 Mbp while chromosome 2 of V. gazogenes ATCC 43942 is 1.3 Mbp long. The genomic rearrangement in the two species includes the inversion of a 400 kbp region in chromosome 2 ( Figure S3 ). The region absent from V. gazogenes PB1 was found to encode for several transposases, toxins (toxin HipA, zeta toxin), endonuclease and several hypothetical genes ( Table S7 ).

The genome of V. gazogenes PB1 were found to harbour 3 CRISPR-Cas operons and 4 CRISPR arrays without cas genes. Three CRISPR arrays were identified on chromosome 1, while chromosome 2 has 3 CRISPR-Cas operons and one CRISPR-array without cas genes. Based on CRISPR-Cas classification, all the CRISPR-Cas operons identified belong to Class 1 CRISPR system. Chromosome 2 has cas3, cas8, cse2, cas7, cas5e, cas6, cas1 and cas2 genes followed by the CRISPR array. Therefore, it belongs to the type I-E CRISPR-Cas system ( Figure 2A ). Type III-B and type III-D CRISPR system for RNA degradation were also identified on chromosome 2 ( Table S8 ). Genomic comparisons, of the homologous regions between V. gazogenes PB1 and V. gazogenes ATCC 43942, and gene arrangements were constructed using Easyfig ( Figures 2A, B ). The Type III-B system contains the cmr6, cmr5, cmr4, cmr3, cmr2, cas10, csm3, cas1 and cas2 genes followed by the CRISPR array while the type III-D CRISPR system has the cas10, csm3, csm3, cas1 and cas2 genes with the CRISPR array located upstream of cas10 ( Figures 2B, C ). In the type I-E CRISPR system, an additional gene between the repeats, which encodes a hypothetical protein, was annotated between the CRISPR array and cas3 gene ( Figure 2A ). In type III-B system of V. gazogenes PB1, the presence of a reverse transcriptase (RT) domain in cas1 proteins was also observed ( Figure 2B ).

BLASTN analysis identified spacer sequences which were identical to phage and plasmid sequences, and are potential targets of the CRISPR-Cas immune systems. CRISPR arrays in V. gazogenes PB1 contained 64 spacers and two of these spacer sequences were identical to sequences of Vibrio phage vB VpaM MAR, and V. parahaemolyticus strain 20130629002S01 plasmid pVPGX2 ( Table S9 ). The canonical type I-E CRISPR-Cas system was identified with 12 spacers, and the type III-B and III-D were identified with 6 and 8 spacers respectively.

Several genes encoding for enzymes capable of breaking down the polysaccharides cellulose and starch were identified within the V. gazogenes PB1 genome ( Figure 3A ). These genes were annotated under different protein names by the NCBI’s PGAP annotation pipeline: endoglucanase as a ‘glycoside hydrolase family 9’ protein (NCBI Accession: USP15540.1), exoglucanase as a ‘glycoside hydrolase family 6’ protein (NCBI Accession: USP14760.1), β-glucosidase as a ‘β-glucosidase’ (NCBI Accession: USP15545.1), α-amylase as a ‘starch-binding protein’ (NCBI Accession: USP15696.1) and α-glucosidase as an ‘α-glucosidase’ (NCBI Accession: USP13629.1). The presence of these genes suggested that V. gazogenes PB1 is capable of degrading starch and cellulose to its constituent glucose monomers. To confirm if this was the case, V. gazogenes PB1 cells were streaked out on plates containing NaCl/minimal media/agar solid media supplemented with 1% (w/v) of starch and cellulose ( Figure 3B ). Additional plates containing glucose and another without a supplemented carbon source were included as controls ( Figure 3B ). The appearance of V. gazogenes PB1 cells on plates containing cellulose and starch confirmed that V. gazogenes PB1 was capable of metabolising cellulose and starch for growth. Additionally, RT-qPCR was performed to evaluate the transcript levels of the genes identified for starch and cellulose degradation based on the PGAP annotation pipeline ( Figure 3C , Table S10 ). The significant differences (as much as 50-fold) between the transcript levels of the genes encoding for the α-amylase (USP15696.1), α-glucosidase (USP13629.1) and the exoglucanase (USP14760.1) genes under the different experimental conditions indicated that these genes are most likely involved in starch and cellulose metabolism ( Figures 3C , S4 ; Table S11 ). Contrary to our expectation, the transcript levels of the isoamylase (USP16020.1) and β-glucosidase (USP15545.1) genes were not upregulated suggesting that these genes have no role to play in cellulose or starch metabolism ( Figures 3C , S4 ). The endo-1,4-β-glucanase gene (USP15540.1) could not be amplified whatsoever. Given that the RT-qPCR protocol was carried out with two primer sets, the failure to amplify this particular transcript could have been due to its high susceptibility to degradation in vivo or the presence of RNA secondary structures that inhibited reverse transcriptase enzyme activity.

We noted that the growth of V. gazogenes PB1 in the presence of starch and cellulose was associated with the production of the secondary metabolite, prodigiosin ( Figure 3C ). A previous study had reported on the growth of PB1 in the presence of starch and carboxymethylcellulose though not for cellulose (Harwood, 1978). Moreover, the quantitative levels of prodigiosin in the presence of these carbon sources were not reported. Therefore, we set up cultivation conditions to monitor the levels of prodigiosin in the presence of starch and cellulose. We used laboratory-grade (>99% purity) starch and cellulose prepared in minimal media to firmly establish whether starch and/or cellulose can serve as the sole carbon source for prodigiosin production in V. gazogenes PB1 ( Figure 4 ). For the first 2 days of cultivation, prodigiosin titres reached as high as 50mg/Litre. Under the same conditions and with prolonged incubation, a similar level could be attained in the presence of 1% (w/v) cellulose. An increase in the starting cellulose content to 2% (w/v) resulted in two-fold higher prodigiosin titres. In contrast, the titres in the presence of 1% (w/v) starch, however, reached approximately 30 mg/liter indicating that starch may be the less effective carbon source for prodigiosin production compared to cellulose.