Identication of a Repressor for the Two iol Operons Required for Inositol Catabolism in Geobacillus Kaustophilus

Geobacillus kaustophilus HTA426, a thermophilic Gram-positive bacterium, feeds on inositol as its sole carbon source, and an iol gene cluster required for inositol catabolism has been postulated with reference to the iol genes in Bacillus subtilis. The iol gene cluster of G. kaustophilus comprises two tandem operons induced in the presence of inositol; however, the mechanism underlying this induction remains unclear. B. subtilis iolQ is known to be involved in the regulation of iolX encoding scyllo-inositol dehydrogenase, and its homologue in HTA426 was found two genes upstream of the first gene (gk1899) of the iol gene cluster and was termed iolQ in G. kaustophilus. When iolQ was inactivated in G. kaustophilus, not only cellular myo-inositol dehydrogenase activity due to gk1899 expression but also the transcription of the two iol operons became constitutive. IolQ was produced and purified as a C-terminal histidine (His)-tagged fusion protein in Escherichia coli and subjected to an in vitro gel electrophoresis mobility shift assay to examine its DNA-binding property. It was observed that IolQ bound to the DNA fragments containing each of the two iol promoter regions and that DNA binding was antagonized by myo-inositol. Moreover, DNase I footprinting analyses identified two tandem binding sites of IolQ within each of the iol promoter regions. By comparing the sequences of the binding sites, a consensus sequence for IolQ binding was deduced to form a palindrome of 5'-RGWAAGCGCTTSCY-3' (where R=A or G, W=A or T, S=G or C, and Y=C or T). IolQ functions as a transcriptional repressor regulating the induction of the two iol operons responding to myo-inositol.


Abstract Background
Geobacillus kaustophilus HTA426, a thermophilic Gram-positive bacterium, grows on inositol as its sole carbon source, and an iol gene cluster required for inositol catabolism has been postulated with reference to the iol genes in Bacillus subtilis. The iol gene cluster consists of two tandem operons induced in the presence of inositol; however, the mechanism underlying the induction remains unclear. B. subtilis iolQ is known to be involved in the regulation of iolX encoding a scyllo-inositol dehydrogenase, and its homolog in HTA426 was found two genes upstream of the rst gene (gk1899) of the iol gene cluster and termed as iolQ in G. kaustophilus.

Results
When iolQ was inactivated, not only the myo-inositol dehydrogenase activity in the cell due to the expression of gk1899 but also the transcription of the two iol operons became constitutive. IolQ was produced and puri ed as a C-terminal His-tag fusion in Escherichia coli and subjected to the in vitro gel mobility shift assay to examine its DNA binding property. It was observed that IolQ bound to the DNA fragments containing each of the two iol promoter regions, and its DNA binding was antagonized by myo-inositol. Moreover, DNase I footprint analyses were conducted to determine the two binding sites of IolQ within each of the iol promoter regions. By comparing the sequences of the binding sites, a consensus sequence for IolQ binding was deduced to be a palindrome of 5′-RGWAAGCGCTTSCY-3′ (where R = A or G, W = A or T, S = G or C, and Y = C or T).

Conclusion
IolQ functions as a transcriptional repressor regulating the induction of the two iol operons responding to myo-inositol.

Background
There is extensive research on the biochemical pathway and the regulation of bacterial inositol catabolism in Bacillus subtilis. It has been reported that B. subtilis possesses a complete set of iol genes required for its inositol catabolism; including the iolABCDEFGHIJ operon, the iolRS operon, iolQ, iolT, iolU, iolW, and iolX [1,2,3,4,5,6].
The iolABCDEFGHIJ operon encodes the enzymes responsible for the primary pathway of inositol catabolism [7,8], whereas iolT was found to encode the major inositol transporter [4]. A repressor encoded by iolR is the major transcription factor that belongs to the DeoR family and regulates the transcription of the iolABCDEFGHIJ operon and iolT [2,4]. In the absence of inositol in the culture medium, the IolR repressor binds to the respective promoter regions to arrest the initiation of transcription. However, in the presence of inositol, one of the metabolic intermediates appearing in the catabolic pathway, 2-deoxy-d-glucuronic acid 6-phosphate produced in the IolC reaction, acts as an inducer in vivo to antagonize the repressor function of IolR, leading to the transcriptional induction of the iolABCDEFGHIJ operon and iolT [1,2,4]. The process of regulation of iol genes by IolR homologs could be conserved among a number of bacterial species, including Gram-negative bacteria. For instance, Sinorhizobium meliloti belonging to Alphaproteobacteria, was shown to possess the iol genes regulated by its IolR ortholog [9], and in Salmonella enterica, belonging to Gammaproteobacteria, its IolR ortholog was found to regulate not only the transcription of iol genes [10] but also an orphan regulator encoded by reiD involved in myo-inositol utilization [11].
On the other hand, a recent study demonstrated that in B. subtilis, an additional repressor encoded by iolQ that belongs to the LacI family regulates iolX, encoding an NAD + -dependent scyllo-inositol dehydrogenase [3], although the mechanisms underlying the regulation of iolU and iolW, either of which encodes an NADP + -dependent scyllo-inositol dehydrogenase [5,6], have not been elucidated as they appeared to be almost constitutive. Genetic evidence suggests that scyllo-inositol and myo-inositol could be the intracellular inducers for IolQ; however, both failed to antagonize its DNA binding activity in vitro [3].
Geobacillus kaustophilus HTA426 is a Gram-positive, thermophilic, and facultative anaerobic bacterium isolated from the deep-sea sediment collected from the Mariana trench in the western Paci c Ocean [12,13]. It can grow at higher temperatures from 48 °C to 74 °C, optimally at 60 °C [13]. Its entire genome was sequenced [13], and some methodologies for its genetic manipulation have also been established [14,15,16]. Furthermore, this bacterium was found to possess a gene cluster with a composition similar to the complete set of iol genes elucidated in B. subtilis [17] (Fig. 1). In fact, it has been demonstrated that G. kaustophilus metabolizes at least three inositol stereoisomers, including myo-inositol, scyllo-inositol, and d-chiro-inositol [17]. In the G. kaustophilus genome, the iol gene cluster is separated into two operons; the rst operon is 5-kb long containing 4 genes of gk1896-1899, and the second one is 12-kb long containing 10 genes of gk1885-1894, both of which are induced in parallel in the presence of inositol [17]. However, till date, no experimental study has been conducted to elucidate how these two operons are regulated for their induction.
Therefore, in the present study, we focused on the gk1901gene of G. kaustophilus homologous to the iolQ gene of B. subtilis, which is presumed to be a repressor belonging to the LacI family and located close upstream of the iol gene cluster (Fig. 1). We explored the function of gk1901, renamed here as iolQ in G. kaustophilus, encoding a repressor that bound to the two promoter regions within the iol gene cluster and was responsible for their transcriptional induction in the presence of myo-inositol.

Results
Inactivation of iolQ (gk1901) rendered myo-inositol dehydrogenase constitutive In B. subtilis, iolR encodes the major transcription regulator (repressor) for the iolABCDEFGHIJ operon and iolT [1,2,4]. We attempted to identify a counterpart of iolR within the genome of G. kaustophilus HTA426 using the conventional homology search tool BLASTP [18]. The best candidate was found to be gk1840, which is predicted to encode a transcriptional regulator, belonging to the DeoR family, annotated as a fructose operon transcriptional repressor, and sharing a homology with iolR [Bit-score = 102 bits (253); Evalue = 1e -26]. In contrast, a recent study showed that B. subtilis iolX is regulated by another repressor encoded by iolQ [3]. The best candidate for an iolQ counterpart in G. kaustophilus was gk1901, which encodes a transcriptional regulator of the LacI family and shares a much higher homology with B. subtilis iolQ [Bit-Score = 353 bits (905); E-value = 5e-121]. Furthermore, gk1901 was located close (only two genes upstream) to the rst gene of the iol gene cluster (gk1899). Therefore, we focused on gk1901 as the possible candidate for the regulator of the iol genes in G. kaustophilus and created a mutant strain YS202, in which the entire coding region of gk1901 was deleted by replacement with a kanamycinresistant gene ( Table 1). It is known that the NAD + -dependent myo-inositol dehydrogenase is encoded by gk1899, which is the rst gene of the iol gene cluster and transcribed from a promoter located its own upstream to be induced in the presence of inositol in the growth medium [17]. In the parental strain MK72 (a derivative of HTA426 lacking both functional pyrF and pyrR, which had previously been constructed for the counter selection system [15]) grown in the minimal medium containing 0.1% casamino acids as the carbon source supplemented with and without 10 mM myo-inositol, the activity of NAD + -dependent myo-inositol dehydrogenase was repressed in the absence of myo-inositol but induced in its presence; in contrast, the activity became completely constitutive in YS202 lacking gk1901 (Table 2). These results suggested that gk1901 is involved in the induction mechanism of the iol genes, including gk1899. Consequently, we renamed gk1901 as iolQ of G. kaustophilus hereafter. *Bacterial strains were grown in the liquid minimal medium containing both 0.1 g/ml casamino acids and 1 µg/ml uracil and additionally supplemented with the carbon sources as indicated (each 10 mM). Values are mean ± SD of three independent measurements.
On the other hand, the enzyme activities that were induced in MK72 and constitutive in YS202 were not repressed in the presence of additional glucose in the medium (Table 2), a nding that was consistent with the previous report that the iol genes in G. kaustophilus were not under catabolite repression [17].
iolQ encodes a regulator involved in the transcriptional induction of the two iol operons of G. kaustophilus The constitutive activity of the NAD + -dependent myo-inositol dehydrogenase in YS202 lacking iolQ suggested that the transcription of the iol genes in G. kaustophilus is regulated by iolQ. YS202 and its parental strain MK72 were grown in the presence and absence of myo-inositol, and their total RNAs were extracted and then subjected to northern blot analyses (Fig. 2). The internal short stretches of the coding regions of gk1899 and gk1894 were used as probes to detect the transcripts of 5-and 12-kb operons [17], respectively. In MK72, as previously found in HTA426, the speci c transcripts of the two operons were detected only in the presence of myo-inositol [17], whereas in YS202, both were completely constitutive irrespective of the presence or absence of myo-inositol (Fig. 2). Our results clearly indicated that iolQ was involved in the transcriptional regulation, especially induction responding to myo-inositol, of the two iol operons.

IolQ bound to DNA fragments containing each of the two iol promoter regions in vitro
The northern blot analyses showed that iolQ was involved in the repression of the two operons of iol genes in the absence of myo-inositol. Therefore, we examined whether IolQ could bind to the promoter regions of the two iol operons. For this purpose, iolQ was cloned into pET30a(+) to be expressed as a Cterminal His-tag fusion (IolQ-his) and puri ed in Escherichia coli BL21(DE3). The production and puri cation of IolQ-his (approximately 35 kDa) were con rmed (Fig. 3), after which IolQ-his was subjected to gel mobility shift analyses to check whether it binds to the upstream sequences of the two iol operons in vitro as follows.
The transcriptional start point of the former 5-kb-long operon containing gk1896-1899 was previously de ned at 136-bp upstream from the start codon of gk1899 [17], and the corresponding − 35 and − 10 regions were deduced to be the Pgk1899 region (Fig. 4A). To determine the transcriptional start point of the latter 12-kb-long operon for gk1885-1894, we conducted a 5′-rapid ampli cation of cDNA ends (5′-RACE) analysis using the total RNA sample prepared from the strain MK72 grown in the presence of myoinositol (Fig. 2). Based on the results, we could identify a transcriptional start point at 210-bp upstream from the start codon of gk1894 (Fig. 4B), and the corresponding − 35 and − 10 regions were found to be the Pgk1894 region.
For gel mobility shift analyses, the following three DNA fragments were prepared: the Pgk1894 fragment containing the stretch from − 400 to − 100 of the gk1894 start codon, the Pgk1899 fragment containing the stretch from − 250 to + 50 of the gk1899 start codon, and the negative control fragment corresponding an internal coding sequence of gk1894 containing + 959 to + 1109 from the start codon. A xed amount of DNA fragments was reacted with increasing concentrations of IolQ-his in vitro and loaded onto the polyacrylamide gel. Both the Pgk1894 and Pgk1899 fragments exhibited an obvious gel mobility shift, forming DNA-protein complexes in a dose-dependent manner with the increase in the concentration of IolQ-his, whereas the negative control fragment did not exhibit such a result (Fig. 5). On the other hand, when one of myo-inositol, scyllo-inositol, scyllo-inosose, and ribose was added to the DNA-protein reaction mixture, the gel mobility shifts of the Pgk1894 and Pgk1899 fragments were reduced partially but signi cantly only in the presence of myo-inositol (Fig. 5). These results indicated that IolQ-his bound to each of the Pgk1894 and Pgk1899 fragments speci cally and that myo-inositol could antagonize the interaction between each of the DNA fragments and IolQ-his.
Two IolQ binding sites within each of the iol promoter regions All the above-described results suggested that IolQ could be a transcriptional repressor of the two iol operons that are induced in the presence of myo-inositol. We conducted DNase I footprint analyses to de ne the IolQ binding sites within the two promoter regions. The DNA fragments corresponding to each of the promoter regions were prepared using the respective 5′-6-carboxy uorescein (6-FAM)-labeled primers and reacted with IolQ-his under the conditions similar to those used in the gel mobility shift analyses. The DNA-protein complexes were treated with DNase I, and the digested DNA was subjected to fragment size analysis to determine the regions protected from DNase I by the bound IolQ-his (Fig. 6). The results suggested that there could be two protected regions within each of the two iol promoter regions (Figs. 4 and 6, the nucleotide sequences of upper and lower strands of the protected regions are shown in red), although the protection in the lower strand of the Pgk1894 fragment was less evident than that in the others due to unknown reason (Fig. 6D). The protected regions could be considered as the IolQ binding sites, and a comparison of the nucleotide sequences among the four binding sites allowed us to deduce a conserved palindromic sequence of 5′-RGWAAGCGCTTSCY-3′ (where R = A or G, W = A or T, K = G or C, and Y = C or T) recognized by IolQ.
Each of the two iol promoter regions contained two IolQ binding sites, and four additional DNA fragments were prepared to contain each one of the IolQ binding sites as follows: for the Pgk1899 region, fragment A1, containing the stretch from − 250 to − 50 of the gk1899 start codon, and fragment A2, from − 108 to + 114, whereas for the Pgk1894 region, fragment B1, containing the stretch from − 460 to − 284 of the gk1894 start codon, and fragment B2, from − 304 to − 100 (Fig. 7). The four fragments were subjected to another set of gel mobility shift assays to con rm whether IolQ bound to any of them, although fragment A2 failed to form a clear protein-DNA complex (Fig. 7). For either promoter region, it was suggested that IolQ exhibited higher a nities to the binding sites adjacent to the respective transcriptional start points, contained in fragments A1 and B2, than to the other binding sites in fragments A2 and B1. In addition, myo-inositol could antagonize the binding of IolQ to all fragments.

Discussion
The majority of genes involved in inositol catabolism in bacterial systems have been demonstrated under the regulation of the repressor IolR belonging to DeoR family [1,2,4,9,10]. However, in B. subtilis, a recent research showed that the LacI family transcriptional repressor IolQ regulates the gene iolX for scyllo-inositol dehydrogenase [3]. Therefore, we focused on the iolQ gene of G. kaustophilus (a homolog of iolQ of B. subtilis, designated formerly as gk1901 in databases), which is located two genes upstream of the iol gene cluster comprising two operons; the rst operon is 5-kb long containing 4 genes of gk1896-1899, and the second one is 12-kb long containing 10 genes of gk1885-1894, as we speculated that iolQ might be involved in the transcriptional regulation of the two iol operons. The NAD + -dependent myoinositol dehydrogenase encoded by gk1899 (the rst gene of the former 5-kb-long iol operon) is known to be induced only in the presence of inositol in the culture medium [17]. We observed that the inactivation of iolQ rendered not only the NAD + -dependent myo-inositol dehydrogenase ( Table 2) but also the transcription of the two iol operons constitutive (Fig. 2). Furthermore, the gel mobility shift analyses revealed that IolQ bound to the promoter regions of both the iol operons, and its DNA binding activity was antagonized in the presence of myo-inositol (Fig. 5). All these results indicated that in G. kaustophilus, iolQ encoded a transcriptional repressor regulating the induction of the two iol operons responding to myo-inositol.
G. kaustophilus can grow not only on myo-inositol but also on other inositol stereoisomers, including scyllo-inositol and d-chiro-inositol [17], suggesting that these inositol isomers would also induce the iol operons. In B. subtilis, iolW and iolI, which are the respective homologs of gk1898 and iolI of G. kaustophilus, are required to metabolize scyllo-inositol and d-chiro-inositol, respectively [19,20] (Fig. 1). Moreover, studies have shown that these genes together with iolG, the counterpart of gk1899 of G. kaustophilus, are also involved in interconversion among myo-inositol, scyllo-inositol, and d-chiro-inositol [19,20]. In the present study, we demonstrated that the two iol operons of G. kaustophilus are under the regulation of IolQ and that the DNA binding activity of IolQ was antagonized exclusively by myo-inositol in vitro. In G. kaustophilus, gk1898 and gk1899 are contained in the former 5-kb operon, whereas iolI is the third gene of the latter 12-kb operon. The transcriptional repression by IolQ might allow some basal expression of iol genes (Table 2), and this may allow gk1898, iolI, and gk1899 to function to convert scyllo-inositol and d-chiro-inositol into myo-inositol in the cell to induce the iol operons. On the other hand, the inositol transporters in G. kaustophilus have not been identi ed yet; however, some genes included in the two iol operons, such as gk1893, 1894, and 1895 possibly composing an ABC transporter and gk1885 encoding a member of the major facilitator superfamily [13], might serve as myo-inositol transporter(s) with broader speci city to uptake any of these inositol isomers. The broader speci city in the inositol transporters was reported for that in Caulobacter crescentus that transports not only myoinositol but also ribose [21]. Although ribose might be an additional substrate of the putative inositol transporters, it is unlikely that ribose acts an inducer in G. kaustophilus because the gel mobility shift assays showed that IolQ binding to the DNA fragments of the iol promoter regions was not antagonized by ribose (Fig. 5), and to our knowledge, there is no report describing any pathway involved in converting ribose into myo-inositol.
The transcription of iol genes in a number of bacterial species, including B. subtilis, Clostridium perfringens, Lactobacillus casei, Salmonella enterica, and Sinorhizobium meliloti, is repressed in the presence of glucose and thus under carbon catabolic repression [10,22,23,24,25,26]. Catabolite repression in Gram-positive bacteria such as B. subtilis involves transcriptional repression through CcpA/P-Ser-HPr binding to the speci c chromosomal sites with cre sequence [26]. As gk2810 was identi ed within the G. kaustophilus genome to be highly homologous to the ccpA gene of B. subtilis, it was considered that the carbon catabolite repression might function in G. kaustophilus in a similar manner as that in B. subtilis. However, the iol genes in G. kaustophilus were not subjected to catabolite repression at all (Table 2) [17]. This nding implied that the inositol catabolism in G. kaustophilus might have a physiological signi cance other than being a simple strategy to utilize this carbon source, and a similar nding has been discussed with respect to another Gram-positive bacterium, Corynebacterium glutamicum [27,28]. As already mentioned, G. kaustophilus HTA426 was isolated from the deep-sea sediment collected from the Mariana trench in the western Paci c Ocean [12,13]. It has been reported that inositols are often found on the deep sea oor [29] and could be considered as osmotic regulators (osmolytes) [30,31]. Deep-sea organisms are known to accumulate osmolytes to increase their intracellular osmotic pressure to adapt to the speci c environment in the deep-sea [32]. Moreover, scylloinositol is particularly known as a chemical chaperone that interferes in the aggregation of denatured proteins such as amyloid beta accumulating in the brain resulting in the development of Alzheimer's disease [33,34]. Through the possible interconversion among inositol stereoisomers [17,20], some scylloinositol might be produced in G. kaustophilus to prevent the aggregation of proteins denatured under lower temperatures and high pressure in the deep-sea. The iol genes in G. kaustophilus might have been excluded from the general control of carbon catabolite repression to be responsible for such a speci c cellular function.
Each of the two iol promoter regions (Pgk1899 and Pgk1894) had a pair of IolQ binding sites that were found to share a palindromic consensus sequence of 5′-RGWAAGCGCTTSCY-3′ (Fig. 5). Within the genome sequence of G. kaustophilus, the consensus sequence can be found in the intergenic regions preceding gk0016 (a hypothetical protein), gk0786 (a transposase), gk1017 (a hypothetical protein), gk2450 (a 50S ribosomal protein L33), gk2733 (a hypothetical protein), and gk3430 (a glycyl-tRNA synthetase) [13]. IolQ might have the ability to bind to these sites, as IolQ-his could bind to the DNA fragments containing only one of the binding sites of the iol promoter regions (Fig. 7). However, the presence of pairwise IolQ binding sites is speci c to the two iol promoter regions, suggesting that the closely located two binding sites are required to establish the transcriptional repression via IolQ binding. In fact, IolQ belongs to the transcription factors of the LacI family, which is represented by LacI of E. coli forming a dimeric structure to bind to the two operator sites of the lactose operon [35], and the two DNAbound LacI dimers associate into a tetramer to form a DNA loop structure [36]. It has been observed that RNA polymerase cannot bind to the looped-out promoter due to steric hindrance, resulting in tighter repression [37]. Results of our DNase I footprint analyses revealed that the DNA containing the Pgk1899 region was not looped-out via IolQ binding, whereas that containing the Pgk1894 region could be (Fig. 5). In fact, it has been observed that the transcription driven by Pgk1899 is less tightly repressed in the absence of myo-inositol than that driven by Pgk1894 [38].
In B. subtilis, IolQ was found to act as a transcriptional repressor of iolX, encoding the NAD + -dependent scyllo-inositol dehydrogenase [5]. B. subtilis IolQ binds to the two sites with a consensus sequence of 5′-AGAAARCGCTTKCKCAAA-3′ (where K = G or T) in the iolX promoter region [5]. scyllo-Inositol and myoinositol might possibly be the intracellular inducers; however, neither of them antagonized its DNA binding activity in vitro [5]. In contrast, in our study, we have clearly demonstrated that G. kaustophilus IolQ bound to the four binding sites with a palindromic consensus sequence of 5′-RGWAAGCGCTTSCY-3′ and its DNA binding activity was antagonized by myo-inositol. These two repressor proteins share a certain similarity in their amino acid sequences and the consensus sequences required for their binding, and in addition, both are functionally related based on their involvement in the regulation of the genes required for inositol catabolism. Further investigation of the similarities and differences between the two IolQs may yield some implications for understanding the evolution of the regulatory genes with related functions.

Conclusions
The iolQ gene of G. kaustophilus (formerly known as gk1901) is located two genes upstream of the iol gene cluster comprising two tandem operons. The NAD + -dependent myo-inositol dehydrogenase encoded by gk1899 (the rst gene of the iol cluster) is known to be induced only in the presence of inositol in the culture medium. Our results showed that the inactivation of iolQ rendered not only the NAD + -dependent myo-inositol dehydrogenase but also the transcription of the two iol operons constitutive. Furthermore, the gel mobility shift analyses demonstrated that IolQ bound to the promoter regions of both the iol operons and its DNA binding activity was antagonized in the presence of myo-inositol. All these results indicated that in G. kaustophilus, iolQ encoded a transcriptional repressor regulating the induction of the two iol operons responding to myo-inositol. We also conducted DNase I footprint analyses to determine the two binding sites of IolQ within each of the iol promoter regions. By comparing the sequences of the binding sites, we deduced that the consensus sequence required for IolQ binding was a palindrome of 5′-RGWAAGCGCTTSCY-3′ (where R = A or G, W = A or T, S = G or C, and Y = C or T). Bacterial strains, plasmids, primers, and growth conditions The strains and plasmids used in this study are shown in Table 1, and the primers are depicted in Table 3 10.0 mM K-MOPS (pH 8.0), and carbon sources (10 mM myoinositol, 10 mM glucose, and 0.1% casamino acids). When required, 1 or 100 µg/ml uracil and 50 µg/ml 5-uoroorotic acid (5-FOA) were added.

Construction Of Plasmids
pGKE25diolQ was prepared to construct the strain YS202 as described here. A 1.0-kb stretch corresponding to the upstream region of iolQ in the G. kaustophilus HTA426 chromosome was ampli ed by PCR using the primer pair of gk1901-UF/gk1901-UR (Table 3). Similarly, another 1.0-kb stretch corresponding to the downstream region of iolQ was ampli ed using gk1901-DF/gk1901-DR (Table 3). pGKE25 [39] was linearized by inverse PCR using pGKE25-ER and pGKE25-HF as primers ( Table 3). The PCR fragments of the upstream and downstream regions of iolQ and the linearized pGKE25 were incubated with Gibson Assembly Master Mix (New England Bio) at 50 °C for 15 min and transformed into E. coli DH5α (Table 1) to yield the recombinant plasmid pGKE25diolQ, whose accurate construction was con rmed by DNA sequencing. pETiolQhisGK was constructed as described subsequently, which was designed to express IolQ-his and purify the gene product as a C-terminal His-tag fusion in E. coli. The coding region of iolQ was ampli ed by PCR from the chromosome of HTA426 using the primer pair of gk1901-F/gk1901-R ( Table 3). The PCR fragment and the DNA of pET30a(+) previously cleaved using NdeI and EcoRI were mixed and incubated for 15 min at 50 °C with Gibson Assembly Master Mix to produce pETiolQhisGK, whose accurate construction was con rmed by DNA sequencing.

Construction Of Strains
G. kaustophilus YS202 is a derivative of MK72 [15], which was constructed to introduce an in-frame deletion of iolQ as described here. The recombinant plasmid pGKE25diolQ was introduced into E. coli BR408 (Table 1) and then transferred into MK72 by conjugation as described previously [15] to form colonies on the minimal plate medium containing 0.1% casamino acids and 1% glucose without uracil at 60 °C. One of the resulting uracil non-demanding transconjugants was proliferated once in LB medium at 60 °C, and an aliquot of the culture was inoculated and allowed to grow in the minimal medium containing 0.1% casamino acids, 1% glucose, 1 µg/ml uracil, and 50 µg/ml 5-FOA for 24 h. From this culture, colonies were formed on the minimal medium plates containing 0.1% casamino acids, 1% glucose, 1 µg/ml uracil, and 50 µg/ml 5-FOA, and one of them was selected as the strain YS202, whose accurate construction was con rmed by DNA sequencing.

Inositol Dehydrogenase Assay
Strains of G. kaustophilus were grown aerobically in the liquid minimal medium containing 0.1% casamino acids and 1 µg/ml uracil and additionally supplemented with or without 10 mM myo-inositol. The bacterial cells were harvested by centrifugation and washed once in a buffer prepared using 1 M NaCl and 50 mM Tris-HCl (pH 8.0). The cells were suspended in 100 µl of 100 mM Tris-HCl (pH 8.0) and transferred into a microtube containing 5 µl of 10 mg/ml lysozyme in 100 mM Tris-HCl (pH 8.0). The cells were incubated at 37 °C, after which they were disrupted completely by Bioruptor UCD-250 (Cosmo Bio, Tokyo, Japan). After centrifugation, the supernatant was stored as the enzyme solution. Next, the activity of the NAD + -dependent myo-inositol dehydrogenase in the enzyme solution was measured spectrophotometrically as described previously [17].

Rna Sample Preparation
Strains of G. kaustophilus were grown aerobically in the liquid minimal medium containing 0.1% casamino acids and 1 µg/ml uracil and additionally supplemented with or without 10 mM myo-inositol. The bacterial cells were collected and disrupted to extract the total RNA as described previously [17].

Northern Blot Analysis
The RNA samples were subjected to northern blot analyses using DIG-labeled RNA probes speci c for gk1894 and gk1899, as described previously [3]. The RNA probes were prepared as follows: the DNA fragments corresponding to the part of the gk1894and gk1899-coding regions were ampli ed by PCR using the DNA of the strain HTA426 as a template and the primer pairs of n-gk1894-F/n-1894-R and n-gk1899-F/n-1899-R, respectively (Table 3) to introduce a T7 RNA polymerase promoter sequence in the tail. The PCR product was used as the template for in vitro transcription using a DIG RNA labeling kit (SP6/T7) (Roche Diagnostics, Basel, Switzerland) to produce the DIG-labeled RNA probes. Cellular RNAs were separated by gel electrophoresis, transferred to a positively charged nylon membrane (Roche Diagnostics), hybridized using the DIG-labeled probes, and then detected using a DIG luminescence detection kit (Roche Diagnostics).

5′-RACE
5′-RACE was performed to identify a transcriptional start point using the 5′-Full RACE Core Set (Takara Bio, Kusatsu, Japan) according to the protocol provided by the supplier. Brie y, the rst strand of cDNA was synthesized from the RNA sample prepared from the cells of MK72 grown in the liquid minimal medium containing 0.1% casamino acids, 1 µg/ml uracil, and 10 mM myo-inositol by reversetranscription reaction using a 5′-end phosphorylated RT primer (Table 3) and then treated with RNase H to liberate the single-stranded cDNA. The single-stranded cDNA was cyclized by T4 RNA ligase and subjected to the two-step nested inverse PCR using two PCR pairs of A1/S1 and A2/S2 (Table 3) successively. The resulting PCR fragments were cloned into the T vector pMD20 (Takara Bio) and sequenced to identify the transcriptional start point.
Puri cation Of Iolq-his E. coli BL21(DE3) carrying pETiolQhisGK was grown aerobically in LB liquid medium containing kanamycin, and then 1 mM IPTG was added to the growing culture to induce the expression of IolQ-his. Bacterial cells were harvested, suspended in a buffer prepared using 50 mM phosphate buffer (pH 8.0), 20% glycerol, and 0.5 M NaCl, and then disrupted completely by Bioruptor UCD-250 (Cosmo Bio). After centrifugation, the supernatant was subjected to histidine tag fusion protein puri cation using TALON®Metal a nity resins (Takara Bio) according to the supplier's instructions. The eluted fractions were subjected to SDS-PAGE, and the purity of IolQ-his was con rmed.

Gel Mobility Shift Assay
The gel mobility shift assay was performed fundamentally based on the protocol described previously [40]. DNA fragments were prepared by PCR ampli cation from the DNA of HTA426. For preparation of the Pgk1894 fragment, the primer pair of EMSA-DF/EMSA-DR was used; for the Pgk1899 fragment, the primer pair of EMSA-UF/EMSA-UR was used; and for the negative control fragment corresponding to an internal coding sequence of gk1894, the primer pair of EMSA-IF/EMSA-IR was used ( Table 3). The Pgk1894 fragment was divided into fragments B1 and B2 to separate the two IolQ binding sites, and similarly, the Pgk1899 fragment was divided into fragments A1 and A2. To prepare the fragments A1, A2, B1, and B2, the primer pairs EMSA-UF/EMSA-5UR, EMSA-5DF/EMSA-5DR, EMSA-12UF/EMSA-12UR, and EMSA-12DF/EMSA-DR were used, respectively (Table 3)

Dnase I Footprint Analysis
DNase I footprint analysis was conducted essentially as described previously [3]. PCR procedures were used to amplify 5′-6-FAM-labeled DNA fragments containing the Pgk1899 region from the DNA of the strain HTA426 using the speci c primers of [6-FAM]EMSA-UF/EMSA-UR and EMSA-UF/[6-FAM]EMSA-UR for labeling the sense and antisense strands, respectively (Table 3). Similarly, 5′-6-FAM-labeled DNA fragments containing the Pgk1894 region were ampli ed using the speci c primer pairs of [ Availability of data and materials The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.
Positions of IolQ-DNA complex (C), free DNA (F), and negative control fragment (NC) are indicated by arrowheads. Figure 6 DNase I foot prints of IolQ on Pgk1899 (panels A and B) and Pgk1894 (C and D). For clarity, each of the panels has two sets of chromatograms, which are the two divisions of a continuous chromatogram; the left is the former part and the right is the latter.