A Futile Metabolic Cycle of Fatty Acyl Coenzyme A (Acyl-CoA) Hydrolysis and Resynthesis in Corynebacterium glutamicum and Its Disruption Leading to Fatty Acid Production

The C. glutamicum genomic database (GenBank accession number BA000036) indicates the presence of the following five genes encoding putative FadD proteins (see Fig. S1A in the supplemental material): fadD1 (Cgl0284, NCgl0279), fadD5 (Cgl0400, NCgl0388), fadD4 (Cgl1198, NCgl1151), fadD15 (Cgl2296, NCgl2216), and fadD32 (Cgl2872, NCgl2774). All of these genes contain ATP/AMP and fatty acyl-CoA synthetase (FACS) motifs homologous to those in E. coli FadD (Fig. S1B) (29). To date, however, there have been no reports on their functions, except for fadD32, which is located in a chromosomal cluster with accD3 and pks (Fig. S1A) and has been assumed to be involved in the synthesis of mycolic acids by activating free fatty acids to form acyl-AMP (14, 30). On the other hand, C. glutamicum possesses two annotated tes genes, tesB (Cgl1664, NCgl1600) and tesA (Cgl2451, NCgl2365), whose functions remain to be clarified (Fig. S2A). The deduced amino acid sequence of the tesA product, unlike that of the tesB product, has an active-site motif that is homologous to the E. coli ybaW product Tes III, a long-chain acyl-CoA thioesterase (Fig. S2B) (21), suggesting the involvement of tesA in the hydrolysis of long-chain fatty acyl-CoAs.

Our first task was to identify the fadD and tes genes responsible for the transfer and release, respectively, of CoA between long-chain fatty acids and their CoA derivatives. Based on the catalytic reactions, it was reasonable to expect that the intended fadD gene negatively affects fatty acid production when amplified in a fatty acid producer while the intended tes gene plays a pivotal role in fatty acid production. Initially, to identify the types of fadD genes present among the five candidate fadD genes, their coding regions were individually cloned on a multicopy vector so as to be constitutively expressed under the control of the promoter of the endogenous gapA gene, encoding glyceraldehyde 3-phosphate dehydrogenase, to generate plasmids pCfadD1, pCfadD4, pCfadD5, pCfadD15, and pCfadD32. Each plasmid was introduced into the C. glutamicum fatty acid-producing strain WTΔfasR, and the resulting plasmid carriers were compared with the control vector carrier for fatty acid production when cultivated in minimal medium (MM) (1% glucose). As shown in Table 1, plasmids pCfadD5 and pCfadD15 brought about approximately 25% and 28% decreases in fatty acid production, respectively, while the other three plasmids had only marginal or relatively small effects on production. These data suggest that fadD5 and fadD15 play significant roles in getting free fatty acids back to their CoA derivatives. Although the pCfadD32 carrier showed a 15% decrease in production, we believed that this was probably due to the redirection of carbon into mycolic acid synthesis, considering the predicted role of FadD32, namely, activation of free fatty acids to form acyl-AMP, a precursor for mycolic acid synthesis (14, 30).

As for tes, the above-mentioned in silico analysis suggested that the gene annotated as tesA is a more likely candidate for long-chain fatty acyl-CoA hydrolysis than the other tesB gene. If so, the loss of tesA function should impair the ability of strain WTΔfasR to produce fatty acids. In fact, disruption of tesA resulted in an almost complete loss of fatty acid production ability, and this phenotype was fully complemented by the plasmid-mediated expression of tesA (Table 1). However, since the tesA-disrupted strain WTΔfasRΔtesA showed an approximately 40% lower growth level than its parent, WTΔfasR, as can be seen from the optical density (OD) (Table 1), there remained a possibility that the reduced growth might affect fatty acid production. To avoid this, we expressed the chromosomal tesA gene under the control of the myo-inositol-inducible promoter of iolT1, encoding a myo-inositol transporter in strain WTΔfasR, followed by examination for fatty acid production under the conditions with glucose and with myo-inositol. The resulting strain, WTΔfasRtesA^iol, grew well regardless of the carbon sources and exhibited a myo-inositol-dependent fatty acid-producing phenotype, while the control WTΔfasR strain produced fatty acids under both conditions (Table 1). These data strongly suggest that tesA is responsible for long-chain fatty acyl-CoA hydrolysis and is thus essential for fatty acid production.

Naturally biotin-auxotrophic C. glutamicum can grow on glucose even under biotin-free conditions, provided that oleic acid is used to supplement the medium (31). This is because exogenous oleic acid is activated to oleoyl-CoA by FadD and is then utilized for membrane lipid biosynthesis. Accordingly, based on the above-mentioned results, it is reasonable to expect that the loss of fadD5 and/or fadD15 functions would result in impaired growth even in the presence of oleic acid under biotin-free conditions. To verify this, we constructed fadD5- and fadD15-disrupted strains, designated WTΔfadD5 and WTΔfadD15, respectively, and their double disruptant, designated WTΔfadD5&15, from the wild-type strain. As shown in Fig. 2A, the wild-type strain (WT) grew normally under biotin-free and oleate-supplemented conditions, while strains WTΔfadD5 and WTΔfadD15 exhibited the expected phenotypes of retarded growth under the same culture conditions. In the case of strain WTΔfadD5&15, no growth was observed, although the presence of biotin in the culture restored the growth. We further confirmed that the inability of strain WTΔfadD5&15 to utilize exogenous oleate was complemented, although only partially, by the plasmid-mediated expression of fadD5 or fadD15 (Fig. 2B). These results fortify our earlier conclusion that fadD5 and fadD15 play a role in converting long-chain fatty acids to their CoA derivatives.

Although the data discussed above indicate the essential role of the genomic tesA gene in fatty acid production, its physiological function remains to be clarified. Thus, we examined the phenotype of deficiency in tesA under the wild-type background. As shown in Fig. 3A, the tesA-disrupted strain WTΔtesA showed impaired growth on glucose, and this phenotype was almost fully recovered under the conditions of supplementation with either oleate or palmitate. To further confirm this, we expressed chromosomal tesA under the control of the myo-inositol-inducible promoter of iolT1. When the resulting strain, WTtesA^iol, was cultivated on 1% glucose or 0.5% glucose plus 0.5% myo-inositol, it exhibited myo-inositol-dependent growth (Fig. 3B). These results indicate that the C. glutamicum wild type requires the tesA function for normal growth on glucose and its deficiency causes the requirement for the free fatty acid oleate or palmitate.

The results so far described suggest that fadD5, fadD15, and tesA are all expressed in wild-type cells grown on glucose. However, if TesA and FadDs operate simultaneously, a futile cycle that leads to the hydrolysis of ATP as the net effect would result (Fig. 1), which would be unreasonable. Therefore, we examined whether the coexpression actually occurred in wild-type cells by measuring the enzymatic activities and transcript levels. Initially, we determined the enzymatic activities using soluble fractions prepared from cells in the late exponential phase of growth on glucose. As for FadD, although significant activity was detected in wild-type cells (11.0 ± 0.4 mU/mg of protein), double disruption of fadD5 and fadD15 resulted in activity being reduced to a marginal level (1.6 ± 0.9 mU/mg). The fadD5- and fadD15-disrupted strains showed moderate levels of activity (5.5 ± 0.3 mU/mg and 7.1 ± 0.4 mU/mg, respectively). On the other hand, a relatively high level of Tes activity was detected in wild-type cells (276 ± 53 mU/mg), in contrast to a trace level in the tesA-disrupted strain.

Next, we investigated the transcript levels of fadD5, fadD15, and tesA in the wild-type strain and its derived fadD5-, fadD15-, fadD5- and fadD15-, and tesA-disrupted strains during growth on glucose (Fig. S3). The data for the three genes in each disruptant are presented as relative to values obtained for the corresponding genes in the wild-type strain. The transcript level of each gene was shown to fall to a negligible level as a consequence of the disruption of the corresponding gene, indicating that the three genes are all expressed in wild-type cells grown on glucose.

This series of data not only reconfirmed our earlier conclusion that the intended Tes and FadD activities are specified by tesA and the two fadD genes (fadD5 and fadD15), respectively, but also verified that the two opposing reactions operate simultaneously in wild-type cells during growth on glucose.

The findings that tesA and the two fadD genes were coexpressed in wild-type cells during growth on glucose suggest the formation of a cyclic metabolic route between long-chain fatty acids and their CoA derivatives. If so, and considering that this organism naturally lacks the β-oxidation pathway involving fatty acid degradation, it is likely that blockage of the cycle at the FadD step would provoke cells to accumulate fatty acids which are normally synthesized by this organism, namely, oleic acid, palmitic acid, and stearic acid. Furthermore, the potential to accumulate the fatty acids should be cancelled by the additional disruption of tesA. Based on these assumptions, we examined whether the disruption of fadD5 and/or fadD15 causes fatty acid production in wild-type cells during cultivation in MM with 1% glucose. As shown in Fig. 4A, significant amounts of fatty acids were found to accumulate in the cultures of the fadD5-disrupted and fadD15-disrupted strains WTΔfadD5 and WTΔfadD15, respectively, whereas the control wild type produced no detectable fatty acids. When both fadD5 and fadD15 were disrupted in wild-type cells, the resulting double disruptant, WTΔfadD5&15, exhibited synergistically increased production under the same conditions.

Next, we examined whether tesA played a key role in fatty acid production by strain WTΔfadD5&15. As shown in Fig. 4B, the disruption of tesA in strain WTΔfadD5&15 led to an almost complete loss of fatty acid production ability concomitantly with a loss of Tes activity, while plasmid-mediated amplification of tesA in the triple disruptant WTΔfadD5&15ΔtesA resulted in dramatically increased production with increased Tes activity. It is noteworthy that when tesA was amplified in the wild-type strain, fatty acid production was not observed (data not shown).

To further confirm that fatty acid production by strain WTΔfadD5&15 occurred during growth on glucose and not after cells entered the stationary phase, we examined its fermentation profile using the typical fatty acid producer WTΔfasR as a control. As shown in Fig. 5A, fatty acid production by strain WTΔfadD5&15 occurred mainly in the late exponential phase of growth on glucose and was stopped after glucose was consumed, which was almost the same profile as that for the control strain WTΔfasR (Fig. 5B).

These results reinforce our conclusions that TesA and the two FadDs normally form a cycle between long-chain fatty acids and their CoA derivatives (Fig. 1) and that overproduction of fatty acids results when the cycle is blocked at the FadD step.

As mentioned above, engineering of the metabolic cycle consisting of TesA and FadDs, specifically, a combination of disrupted fadD5 and fadD15 and amplified tesA (referred to as the TesFad method), allows C. glutamicum to produce a significant amount of fatty acids, even on a wild-type background (Fig. 4B). The TesFad method differs in production mechanism from the general method, namely, deregulation of fatty acid synthesis, and is expected to be a new strategy for fatty acid production in C. glutamicum. To substantiate the method, we applied it to our best fatty acid producer, PCCA-3 (14). This producer was developed by assembling four positive mutations (fasR20, fasA63^up, fasA2623, and accD3^A433T) in the wild-type genome, and it has the capability of producing approximately 600 mg/liter of fatty acids, which consist mainly of oleic acid (449 mg/liter), palmitic acid (133 mg/liter), and stearic acid (40 mg/liter), in flask cultivation with 1% glucose (Fig. 6). Using this strain as a host, we examined the effects of the TesFad method on fatty acid production. As shown in Fig. 6, disruption of the two fadD genes alone (designated strain PCCA-3ΔfadD5&15) or plasmid-mediated amplification of tesA alone [designated strain PCCA-3(pCtesA)] had relatively small effects on production, but their combination [resulting in strain PCCA-3ΔfadD5&15(pCtesA)] resulted in significantly increased production that achieved a titer of 1,071 mg/liter with a conversion yield of approximately 10% on glucose. These data demonstrate that the effects of the TesFad method are not offset by the performance that strain PCCA-3 has already acquired, and the method could therefore be a useful addition to strain improvement for fatty acid production in this organism. It should be noted that, despite a large increase in the total amount of fatty acids, the fatty acid composition of oleic acid, palmitic acid, and stearic acid remained substantially unchanged when the TesFad method was applied (Fig. 6). It is also noteworthy that no significant by-production of glutamic acid was observed in the engineered strain, suggesting that the engineering would not affect the cell surface structure enough to elicit glutamic acid production.

All strains used in this study are derivatives of the C. glutamicum wild-type strain ATCC 13032. The fatty acid producer WTΔfasR (10) was derived from ATCC 13032 through the in-frame deletion of fasR, which encodes a fatty acid biosynthesis repressor protein. The high-fatty-acid producer PCCA-3 (14) was developed by the so-called “genome breeding” approach (36), in which four specific mutations (fasR20, fasA63^up, fasA2623, and accD3^A433T) were assembled in the ATCC 13032 genome. The former three mutations contribute to the deregulation of fatty acid biosynthesis, and the latter one is assumed to impair the function of the gene product AccD3 to diminish mycolic acid biosynthesis. E. coli DH5α was used as a host for DNA manipulation.

Plasmid pCS299P (37), a C. glutamicum-E. coli shuttle vector, was used to clone the PCR products. Plasmid pESB30 (37), which is nonreplicative in C. glutamicum, is a vector for gene replacement in C. glutamicum. Plasmids pCfadD1, pCfadD4, pCfadD5, pCfadD15, and pCfadD32 were constructed so that fadD1 (Cgl0284, NCgl0279), fadD4 (Cgl1198, NCgl1151), fadD5 (Cgl0400, NCgl0388), fadD15 (Cgl2296, NCgl2216), and fadD32 (Cgl2872, NCgl2774) were constitutively expressed under the control of the promoter of the endogenous gapA gene. For the construction of pCfadD15, the open reading frame (ORF) of fadD15 was amplified using primers fadD15sdFusF and fadD15FusR, with wild-type ATCC 13032 genomic DNA as the template. On the other hand, the genomic region comprising the gapA promoter was amplified using primers PgapAKpBgF and PgapAfadD15sdFusR so that the ribosome-binding site (RBS) sequence for gapA was altered to the consensus RBS sequence proposed for C. glutamicum (38). Similarly, the genomic region comprising the gapA terminator was amplified using primers fadD15TTgapAFusF and TTgapAKpR. These three fragments were fused by PCR, digested with KpnI, and then ligated to KpnI-digested pCS299P to yield pCfadD15.

Plasmids for the other four fadD genes were constructed by replacing the fadD15 ORF on pCfadD15 with the ORF of the corresponding genes, as follows. Plasmid pCfadD15 was linearized by inverse PCR using primers InVer-PgapAsdR and InVer-TTgapAF so as to completely remove only the fadD15 ORF (for convenience, the resulting linear plasmid is referred to here as fragment A). On the other hand, the ORFs of the target genes were amplified using the primer pairs InFu-fadD1F and InFu-fadD1R for fadD1, InFu-fadD4F and InFu-fadD4R for fadD4, InFu-fadD5F and InFu-fadD5R for fadD5, and InFu-fadD32F and InFu-fadD32R for fadD32. The amplified ORFs were individually cloned into fragment A using an In-Fusion HD cloning kit (Clontech Laboratories, Inc., Mountain View, CA, USA) to yield pCfadD1, pCfadD4, pCfadD5, and pCfadD32. For fadD4 on pCfadD4, the native rare start codon GTG was modified to ATG.

Plasmid pCtesA, which contains the tesA gene (Cgl2451, NCgl2365), was constructed as follows. The genomic region comprising tesA and its native promoter (from −1 to −123 bp upstream of tesA) was amplified using primers tesAup120FBamHI and tesAdown70RBamHI. The resulting fragment was digested with BamHI and then ligated to BamHI-digested pCS299P to yield pCtesA.

The sequences of the primers used in this study are listed in Table 2. All primers were designed based on the genomic sequence of C. glutamicum (BA000036) (39), which is publicly available at http://www.genome.jp/kegg/genes.html.

Complete BY medium and minimal medium (MM) were used as basal media for the growth of C. glutamicum strains (40). Solid plates were made by the addition of Bacto agar (Difco) to 1.5%. For preparation of MM containing sodium oleate or sodium palmitate, the fatty acid sodium salt was separately autoclaved and then mixed with a magnesium sulfate solution and a solution containing other components to prevent insolubilization of the fatty acid. For cultivation of plasmid carriers, kanamycin was added at a final concentration of 10 mg/liter. For growth of E. coli, Luria-Bertani broth or agar was used.

Standard protocols (41) were used for the construction, purification, and analysis of plasmid DNA and for the transformation of E. coli. The extraction of C. glutamicum chromosomal DNA and transformation of C. glutamicum by electroporation were carried out as described previously (40). PCR was performed using a DNA thermal cycler (GeneAmp PCR system 9700; Applied Biosystems, Foster City, CA, USA) using Phusion High-Fidelity DNA polymerase (New England Biolabs, Ipswich, MA, USA). Sequencing to confirm the nucleotide sequences of relevant DNA regions was performed using an ABI PRISM 377 DNA sequencer from Applied Biosystems, with an ABI PRISM BigDye Terminator cycle sequencing kit (Applied Biosystems). The subsequent electrophoresis analysis was carried out using Pageset SQC-5ALN 377 (Toyobo, Osaka, Japan).

For the chromosomal deletions of fadD5, fadD15, and tesA, plasmids pCΔfadD5, pCΔfadD15, and pCΔtesA, which contained the corresponding genes with internal deletions, were used to replace the wild-type chromosomal genes with the deleted genes. For the construction of pCΔfadD5, the 5′ and 3′ regions of fadD5 were amplified using the primer pair In-FufadD5up760Fw and fadD5delFusRev and the primer pair fadD5delFusFw and In-FufadD5down740Rev, respectively. These two fragments were fused by PCR and then cloned into linearized pESB30, which was obtained by inverse PCR using primers P9L and P10L, with an In-Fusion HD cloning kit. The resulting plasmid, pCΔfadD5, carried the in-frame-deleted fadD5 gene, which was shortened from 1,707 to 120 bp. Similarly, for the construction of pCΔfadD15, the 5′ and 3′ regions of fadD15 were amplified using the primer pair In-FufadD15up750Fw and fadD15delFusRev and the primer pair fadD15delFusFw and In-FufadD15down700Rev, respectively. These two fragments were fused by PCR and then cloned into linearized pESB30, which was obtained by inverse PCR using primers P9L and P10L. The resulting plasmid, pCΔfadD15, carried the in-frame-deleted fadD15 gene, which was shortened from 1,848 to 180 bp. For the construction of pCΔtesA, the 5′ and 3′ regions of tesA were amplified using the primer pair tesAup630FBamHI and tesAdelFusR and the primer pair tesAdelFusF and tesAdown770RBamHI, respectively. These two fragments were fused by PCR, digested with BamHI, and then ligated to BamHI-digested pESB30 to yield pCΔtesA. This plasmid carried the in-frame deleted tesA gene, which was shortened from 468 to 117 bp. The defined chromosomal deletion of the individual gene was accomplished using each plasmid via two recombination events as described previously (42).

For the construction of the myo-inositol-dependent tesA-expressing strains WTtesA^iol and WTΔfasRtesA^iol, the tesA-disrupted strains WTΔtesA and WTΔfasRΔtesA, respectively, were used as host strains to replace the chromosomal iolT1 gene (Cgl0181, NCgl0178), which is expressed under the control of its native myo-inositol-inducible promoter (43), with the tesA gene. For this gene replacement, plasmid pCPiolT1-tesA was constructed as follows. The upstream and downstream regions of the iolT1 gene ORF (for convenience, these regions are referred to here as fragments B and C, respectively) were amplified by pairs of primers (InFu-iolT1up450F–iolT1-tesAFusR and tesA-iolT1downFusF–InFu-iolT1down740R, respectively). The tesA gene was amplified using primers tesAFusForf and tesAFusRorf. Fragment B, the tesA gene, and fragment C were fused stepwise using PCR. The resulting 1.7-kb fragment was cloned using an In-Fusion HD cloning kit into linearized pESB30, which was obtained by inverse PCR using primers P9L and P10L, to yield plasmid pCPiolT1-tesA. The defined chromosomal replacement of iolT1 with tesA in strains WTΔtesA and WTΔfasRΔtesA resulted in strains WTtesA^iol and WTΔfasRtesA^iol, respectively. Although strains WTtesA^iol and WTΔfasRtesA^iol lack the iolT1 gene, both strains can use myo-inositol as a carbon source due to the existence of an additional transporter encoded by iolT2 (Cgl3058, NCgl2953).

A 3-ml sample of the seed culture grown in BY medium to the mid-exponential phase at 30°C was inoculated into a 300-ml baffled Erlenmeyer flask containing 30 ml of MM (1% glucose or 1% myo-inositol), followed by cultivation at 30°C using a rotary shaker at 200 rpm. After glucose or myo-inositol was consumed, total lipids, including free fatty acids, were extracted from the culture supernatant as described previously (14). In the case of the high-fatty-acid producers PCCA-3 and its derivatives shown in Fig. 6, total lipids were extracted from the culture broth containing cells, because some of the fatty acids produced were likely to be insolubilized in the broth. The extracted lipids were subject to quantitative determination of free fatty acids by gas chromatography as described previously (14).

A 0.05-ml sample of the first-seed culture grown in BY medium to the mid-exponential phase was inoculated into 5 ml of biotin-free MM and cultivated for 20 h to deplete biotin in the culture. The resulting second-seed culture was harvested, washed three times with saline, and resuspended in 5 ml of biotin-free MM. The main culture was started by inoculating 0.05 ml of the biotin-depleted second-seed culture into 5 ml of biotin-free MM supplemented with the indicated concentrations of biotin or sodium oleate (Fig. 2). All liquid cultures were performed at 30°C in L-type test tubes on a Monod shaker at 48 strokes/min.

For strain WTΔtesA, a 0.05-ml sample of the seed culture grown in BY medium supplemented with 1% glucose was inoculated into 5 ml of MM (1% glucose) supplemented with the indicated concentrations of sodium oleate or sodium palmitate (Fig. 3A). For strain WTtesA^iol, a 0.05-ml sample of the seed culture grown in BY medium supplemented with 1% glucose or 0.5% glucose plus 0.5% myo-inositol was inoculated into 5 ml of MM containing 1% glucose or 0.5% glucose plus 0.5% myo-inositol. All liquid cultures were performed at 30°C in L-type test tubes on a Monod shaker at 48 strokes/min.

Cells grown at 30°C to the late exponential phase in a 300-ml baffled Erlenmeyer flask containing 30 ml of MM (1% glucose) were collected by centrifugation at 10,000 × g for 10 min and washed twice with 50 mM Tris-HCl buffer (pH 8.0) and 150 mM 3-(N-morpholino)propanesulfonic acid (MOPS)-NaOH (pH 7.2) for Tes and FadD assays, respectively. The cells were suspended in the corresponding buffer and sonicated on ice for 10 min using a UD-200 ultrasonic disruptor (Tomy Seiko Co., Ltd., Tokyo, Japan). For the Tes assay, cell debris was removed by centrifugation at 10,000 × g for 10 min, and the supernatant was further ultracentrifuged at 100,000 × g for 90 min using an Optima TL ultracentrifuge (Beckman Coulter, Inc., Fullerton, CA, USA). Using the resulting supernatant, the Tes activity was spectrophotometrically measured at 30°C by the methods described by Barnes (44) with the assay mixture consisting of 50 mM Tris-HCl (pH 8.0), 100 μM 5,5′-dithiobis(2-nitrobenzoic acid), 0.008% bovine serum albumin, 14 μM palmitoyl-CoA, and enzyme solution. One unit of Tes activity was defined as 1 μmol of acyl-CoA cleaved per min.

For the FadD assay, the cell debris after sonication was removed by centrifugation at 10,000 × g for 10 min, and the resulting supernatant was used for the FadD assay. The activity was basically measured according to the enzyme-coupled assay described by Ichihara and Shibasaki (45). The assay involved the conversion of substrate (5 mM palmitic acid) to acyl-CoA by the endogenous FadD activity at 30°C in 150 mM MOPS-NaOH (pH 7.2) containing 0.5 mM CoA, 4.5 mM ATP, 12 mM MgCl₂, 1 mM dithiothreitol, 2 μM FAD, 1% methanol, 0.55 mM Triton X-100, 10 U/ml of acyl-CoA oxidase, and 10,000 U/ml of catalase. Acyl-CoA formed from the fatty acid and CoA by FadD was dehydrogenated by acyl-CoA oxidase and then converted into formaldehyde in the presence of methanol by catalase. The concentration of formaldehyde that originated from acyl-CoA was spectrophotometrically measured at time intervals by the colorimetric method (45). One unit of FadD activity is defined as 1 μmol of acyl-CoA formed per min. Protein content was determined using a Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, CA, USA). Data are mean values and standard deviations from triplicate assays.

Extraction of RNAs from C. glutamicum strains and subsequent purification were performed according to the methods described previously (46). cDNA synthesis was performed with 300 ng of RNA using the methods described by Kind et al. (47). Quantitative PCR (qPCR) analysis was performed using the method described by Katayama et al. (48). The gene expression levels were standardized to the constitutive level of 16S rRNA expression and calculated by the comparative cycle threshold method (49).

Bacterial growth was monitored by measuring the optical density at 660 nm (OD₆₆₀) of the culture broth using a Miniphoto 518 R spectrophotometer (Taitec, Saitama, Japan). Concentrations of glucose, myo-inositol, and protein were determined using their respective assay kits, as described previously (50).