Growth of NH9 and its ability to degrade aromatic compounds
C. necator strain NH9 was grown on basal salts medium (BSM) containing 5 mM 3-CB, BA, or citric acid (CA). Strain NH9 was able to grow well with all three compounds although the growth rate was slightly lower with 3-CB than with BA and CA (Fig. 1A). High performance liquid chromatography (HPLC) analyses confirmed that both 3-CB and BA were completely degraded within 18 h of culture with NH9 (Fig. 1B). Compared with BA, 3-CB showed a slight time lag before degradation. Even after these compounds were decomposed thoroughly, the optical density at 600 nm (OD600) did not decrease quickly.
Analysis of differentially expressed genes
To identify commonly and specifically expressed genes between NH9 cells cultured with 3-CB and those cultured with BA, we conducted transcriptome analyses. Reverse-transcribed ribosomal-RNA depleted RNA samples were sequenced on the Illumina MiSeq platform (Table S1). Prior to differential expression analysis, we evaluated similarities and variations in overall gene expression datasets among the samples. The biological replicates clustered closely in multi-dimensional scaling (MDS) plot and cluster dendrogram analyses (Fig. S1), indicative of very little variation among replicates. Genes that met the criteria of log fold-change (logFC) ≥ 2 or ≤ -2 with false discovery rate (FDR) < 0.05 were considered to be significantly differentially expressed between compared pairs of samples. First we compared the transcriptome of NH9 between cells grown with 3-CB and cells grown with BA. In total, 263 genes were expressed differentially: 137 genes were upregulated and 126 genes were downregulated in the 3-CB sample compared with the BA sample. In the 3-CB sample compared with the CA sample, 591 genes were expressed differentially: 374 were upregulated and 217 were downregulated. The largest number of DEGs was in this comparison. In the BA sample compared with the CA sample, 281 genes were differentially expressed: 228 were upregulated and 53 were downregulated. All differential expression analysis results are shown in Table S2.
Genes related to degradation of aromatic compounds
The genes involved in the degradation of 3-CB and BA and the logFC differences in their transcript levels between pairs of sample groups are shown in Table 1. The transcripts per million (TPM) values of each gene are shown in Figure 2. benABCD genes (Fig. S2A) were highly expressed in both the 3-CB and BA samples compared with the CA sample (logFC values 8.0 to 8.5). This is reasonable because BenABCD enzymes presumably react with both 3-CB and BA . The chlorocatechol-degradation genes cbnABCD (Fig. S2B)  were strongly induced in the 3-CB sample (logFC values 9.4 to 9.9) but were also significantly expressed in the BA sample (logFC values 1.3 to 1.5). The genes catA (Fig. S2A), catB, catDC (Fig. S2C), and pcaIJF (Fig. S2D) encode products that participate in the degradation of catechol and 3-oxoadipate, respectively, and were expressed at almost the same levels in the 3-CB and BA samples. A Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis revealed that NH9 is able to decompose BA via another pathway, the epoxybenzoyl-CoA pathway , encoded by bclA and boxABCD genes (Fig. S2E and F). The boxABCD genes were upregulated in the BA sample compared with the CA sample. However, the transcript levels of boxABCD genes were lower than those of ben and cat.
In our previous study, analyses of the genome sequence of strain NH9 revealed genes involved in pathways that completely degrade 2-hydroxybenzoate (2-HBA), 3-hydroxybenzoate (3-HBA) (Fig. S2G), or 4-HBA (Fig. S2H) . The transcript levels of the genes that putatively degrade these aromatic compounds were determined to ascertain whether 3-CB and BA affect their expression (Fig. S3 and Table S3). The transcript levels of the genes involved in the degradation of 2-HBA or 4-HBA were not very different between the CA sample and the 3-CB and BA samples (only pobA in the BA sample was highly induced). Interestingly, the genes involved in the degradation of 3-HBA in NH9 (renamed from nag to mhb)  were significantly induced only by 3-CB.
Strain NH9 has genes related to anthranilate degradation on chromosome 1 (designated as and1 or andAc1Ad1Ab1Aa1) (Fig. S2I). The products of those genes exhibit 43.9% to 73.3% amino acid sequence identities with the corresponding subunits of AndAcAdAbAa from Burkholderia cepacia DBO1, which is regulated by an AraC/XylS-type transcriptional regulator  (Fig. S4A). The presence of the complete set of genes for the initial degradation of anthranilate, together with andR encoding an AraC/XylS-type transcriptional regulator located upstream of the and1 genes, suggests that this gene cluster is functional. Like the mhb genes above, and1 was induced by 3-CB to a transcript level 8-fold that in the BA and CA samples (Fig. S3 and Table S3). Chromosome 2 also harbors putative and genes (designated as and2 or andAc2Ad2Ab2Aa2) (Fig. S2J) and their transcript levels were significantly higher in the 3-CB sample than in the BA sample (Table S4). However, their amino acid sequence identities with the corresponding subunit of AndAcAdAbAa from B. cepacia DBO1 were found to be lower than 45% (Fig. S4A). Also, the putative transcriptional regulator located close to the degradation genes was a member of the MarR family, rather than being an AraC/XylS-type regulator. Therefore, it is difficult to speculate whether and2 genes are involved in anthranilate degradation or not.
The KEGG BRITE functional classification of strain NH9 revealed that 348 genes encode proteins with transporting functions (“Transporters,” ko02000). Of these 348 genes, those that were upregulated (logFC ≥ 2 and FDR < 0.05) by 3-CB and/or BA encoded eight major facilitator superfamily (MFS) transporters and 12 sets of ATP-binding cassette (ABC) transporters. This analysis identified the transporters induced by 3-CB and/or BA (Table 2).
Of the eight MFS transporter genes mentioned above, BJN34_12320, BJN34_18155, and BJN34_30890 had higher transcript levels in the 3-CB sample than in the BA sample, and BJN34_32125 showed the opposite result. The logFC values of the other four genes were not significantly different between 3-CB vs. CA and BA vs. CA. A BLASTP analysis was performed to compare the amino acid sequences of the eight transporters of NH9 with those that have been experimentally verified or functionally analyzed (Table S5). The products of BJN34_18155 and BJN34_32125 exhibited more than 70% amino acid sequence identities with BenP (a 3-CB transporter) . The products of BJN34_30890 and BJN34_33870 showed moderate identities (>50%) with MhbT (a 3-HBA transporter)  and PcaK (a 4-HBA transporter) [37, 38], respectively (Table 2). Phylogenetic analysis of the eight MFS transporters of NH9 together with other known MFS transporters confirmed the close relationships of the four transporters mentioned above with their counterparts in other species, and grouped them in the aromatic acid:H+ symporter (AAHS) family of MFS (Fig. 3). Three other transporters (products of BJN34_11715, BJN34_12320, and BJN34_26825) belonged to the anion/cation symporter (ACS) family and the product of BJN34_20520 belonged to the metabolite:H+ symporter (MHS) family. We then explored the genes surrounding the eight MFS transporter-encoding genes in the NH9 genome, and found that BJN34_32125, BJN34_30890, BJN34_33870, and BJN34_18155 were located next to clusters of genes related to the degradation of BA, 3-HBA, 4-HBA, and anthranilate, respectively (Fig. S2F, G, H, and I). No clusters of genes involved in degradation of aromatic compounds were located around the genes encoding the other four MFS transporters.
Our results showed that 3-CB and BA induced many genes encoding ABC transporters in NH9 (Table 2). The logFC values of most ABC transporter genes were similar between 3-CB vs. CA and BA vs. CA. However, BJN34_29445 to BJN34_29465 were clearly overexpressed in the 3-CB sample, suggesting that these genes were induced specifically by 3-CB. These gene products showed 27.7% to 40.3% amino acid sequence identities with Pca proteins, which are involved in 3,4-dihydroxybenzoate transport . Transporters in other families were also identified in the BLASTP analysis (Table S6). Although a few genes (e.g., BJN34_08680 and BJN34_26835) were differentially expressed in response to both 3-CB and BA, most genes did not show significant changes in their transcript levels, or were downregulated, in either the 3-CB or BA samples compared with the CA sample.
Stress response genes were upregulated when NH9 cells were cultured with 3-CB and BA (Table S4). Four genes encoding molecular chaperones, dnaK (BJN34_09490), groEL (BJN34_09495), groES (BJN34_09500), and clpB (BJN34_11475) were significantly upregulated more than 2-fold by both 3-CB and BA compared with CA. hslV (BJN34_00915), hslU (BJN34_00920), grpE (BJN34_06000), and dnaK (BJN34_16500) were induced only by BA (FDR < 0.05). We also searched for the genes in strain NH9 corresponding to the aromatics stress response genes identified in the previous study  in the KEGG database, and their expression patterns are summarized in Table S4 (categorized as “Benzoate stress response genes”). Only the genes encoding the phosphate transporter PstBACS (BJN34_13095 to BJN34_13110) and superoxide oxidase (SOO) (BJN34_16665) were induced by 3-CB and BA, respectively.
To detect changes in biological function, we conducted Gene Ontology (GO) enrichment analysis by the parametric analysis of gene set enrichment (PAGE) method based on logFC values. The comparisons of 3-CB vs. CA, BA vs. CA, and 3-CB vs. BA detected enrichment of 22, 22, and 15 GO terms, respectively, with FDR < 0.05 (Fig. 4 and Table S7). The GO terms “ferric iron binding” (GO:0008199), “metal ion binding” (GO:0046872), and “2 iron, 2 sulfur cluster binding” (GO:0051537) were significantly upregulated only in the 3-CB sample. On the contrary, the GO terms “nucleotide binding” (GO:0000166) and “peptidyl-prolyl cis-trans isomerase activity” (GO:0003755) were significantly downregulated only in the 3-CB sample. In the BA sample specifically, the GO terms “peptide transport” (GO:0015833) and “bacterial-type flagellum-dependent cell motility” (GO:0071973) were significantly upregulated and “GTPase activity” (GO:0003924), “porin activity” (GO:0015288), and “oxidoreductase activity, acting on the CH-OH group of donors, NAD or NADP as acceptor” (GO:0016616) were significantly downregulated. Interestingly, “chemotaxis” (GO:0006935), “signal transduction” (GO:0007165), and “bacterial-type flagellum-dependent cell motility” (GO:0071973) were downregulated in the 3-CB sample compared with the BA sample. The trends in the variations of the other GO terms listed in Fig. 4 and Table S7 were similar between 3-CB vs. CA and BA vs. CA.
The induction or repression of genes in the “signal transduction,” “chemotaxis,” and “bacterial-type flagellum-dependent cell motility” categories in response to 3-CB, BA, and CA is summarized in Table S8. The 72 genes in the “signal transduction” category mainly encoded proteins related to bacterial chemotaxis and a histidine kinase. Crucially, this category included 12 genes encoding methyl-accepting chemotaxis proteins (MCPs), which play key roles in sensing extracellular signals [41, 42]. Eight of 12 MCP genes were DEGs in the 3-CB vs. BA comparison, and were downregulated in the 3-CB sample. Because three of these eight genes (BJN34_09575, BJN34_21800, and BJN34_32190) were upregulated more than 2-fold with FDR < 0.05 by BA compared with CA, it is likely that their products detect BA or related chemicals as ligands. One MCP gene (BJN34_24350) was significantly upregulated more than 16-fold by both 3-CB and BA compared with CA, indicating that it responded to 3-CB and BA or their related chemicals. In the “chemotaxis” category, many genes were classified as “signal transduction.” Seven of 16 genes were DEGs in the 3-CB vs. BA comparison, and six of them were upregulated more than 2-fold (FDR < 0.05) by BA compared with CA. These six genes encoded CheABDVW proteins and a MCP. Of the 14 genes in the “bacterial-type flagellum-dependent cell motility” category, 11 were upregulated more than 2-fold (FDR < 0.05) by BA compared with CA. These genes encoded proteins comprising the flagellum: the hook, hook-filament junction, distal rod, proximal rod, L ring, P ring, and a part of the C ring. Our data indicated that the genes encoding MCP, Che, and components of the flagellum in NH9 were upregulated by BA and downregulated or not affected by 3-CB. This was predicted to result in differences in cell motility or chemotaxis functions of NH9 cells between 3-CB and BA.
Chemotactic response toward aromatic compounds
To determine whether the transcriptional responses of chemotaxis genes corresponded to actual differences in chemotaxis behavior towards 3-CB and BA, we performed semi-solid agar plate assays (Fig. 5). The formation of a concentric ring was a positive response, as it was indicative of the accumulation of bacterial cells encircling the attractant. NH9 cells formed clear migrating rings around casamino acids (positive control) and BA within 3 and 6 h, respectively (Fig. 5A and B). In contrast, NH9 cells formed a migrating ring only weakly around 3-CB after 14 h (Fig. 5C). There was no ring around BSM without any carbon source (negative control) (Fig. 5D). These results confirmed that strain NH9 has a strong chemotactic response towards BA but a weak response towards 3-CB.