Genome-Based Analysis and Gene Dosage Studies Provide New Insight into 3-Hydroxy-4-Methylvalerate Biosynthesis in Ralstonia eutropha

Abstract

Recombinant Ralstonia eutropha strain PHB⁻4 expressing the broad-substrate-specificity polyhydroxyalkanoate (PHA) synthase 1 from Pseudomonas sp. strain 61-3 (PhaC1_Ps) synthesizes a PHA copolymer containing the branched side-chain unit 3-hydroxy-4-methylvalerate (3H4MV), which has a carbon backbone identical to that of leucine. Mutant strain 1F2 was derived from R. eutropha strain PHB⁻4 by chemical mutagenesis and shows higher levels of 3H4MV production than does the parent strain. In this study, to understand the mechanisms underlying the enhanced production of 3H4MV, whole-genome sequencing of strain 1F2 was performed, and the draft genome sequence was compared to that of parent strain PHB⁻4. This analysis uncovered four point mutations in the 1F2 genome. One point mutation was found in the ilvH gene at amino acid position 36 (A36T) of IlvH. ilvH encodes a subunit protein that regulates acetohydroxy acid synthase III (AHAS III). AHAS catalyzes the conversion of pyruvate to 2-acetolactate, which is the first reaction in the biosynthesis of branched amino acids such as leucine and valine. Thus, the A36T IlvH mutation may show AHAS tolerance to feedback inhibition by branched amino acids, thereby increasing carbon flux toward branched amino acid and 3H4MV biosynthesis. Furthermore, a gene dosage study and an isotope tracer study were conducted to investigate the 3H4MV biosynthesis pathway. Based on the observations in these studies, we propose a 3H4MV biosynthesis pathway in R. eutropha that involves a condensation reaction between isobutyryl coenzyme A (isobutyryl-CoA) and acetyl-CoA to form the 3H4MV carbon backbone.

INTRODUCTION

Over the past few decades, plastic materials made from petroleum have caused environmental problems such as waste plastic pollution and excess emission of carbon dioxide by incineration. Polyhydroxyalkanoates (PHAs) are bacterial polyesters accumulated as intracellular carbon and energy storage material and can be used as biomass-derived biodegradable plastic. Plastic material made from PHAs is expected to solve such environmental problems due to ecofriendliness (1, 2). Poly[(R)-3-hydroxybutyrate] [P(3HB)] is the most basic PHA, which is synthesized by PHA synthase (PhaC) derived from various sources, such as Ralstonia eutropha, Delftia acidovorans, Bacillus species, Burkholderia species, and Synechocystis species (3–7). P(3HB) is not a flexible material due to its high crystallinity and, therefore, has limited commercial application.

To date, various 3HB-based copolymers have been synthesized to improve P(3HB) properties by introducing second-monomer units such as 3-hydroxyvalerate (3HV), 3-hydroxyhexanoate (3HHx), and longer-acyl-chain monomers (8–12). Our group has previously shown that Ralstonia eutropha strain PHB⁻4 (PHA-negative strain) expressing Pseudomonas sp. strain 61-3 PHA synthase 1 (PhaC1_Ps) can biosynthesize a 3HB-based copolymer containing 3-hydroxy-4-methylvalerate (3H4MV) [P(3HB-co-3H4MV)] using sugars as the sole carbon source (13). The 3H4MV unit has a branched bulky side chain, thereby effectively decreasing the crystallinity of P(3HB) by incorporating the 3H4MV unit into the P(3HB) sequence. P(3HB-co-3H4MV) shows material properties similar to those of P(3HB-co-3HHx) (14); while P(3HB-co-3HHx) production relies on plant and vegetable oils, P(3HB-co-3H4MV) has the advantage of using widely available sugars like cellulose and hemicellulose hydrolysates for production. However, by using only sugars as the carbon source, the level of production of 3H4MV was too low to improve P(3HB) properties (15). Additionally, the mechanism of 3H4MV biosynthesis from sugars is still unclear, preventing complete metabolic engineering of 3H4MV biosynthesis in R. eutropha.

In a previous study (15), it was demonstrated that supplementation with excess leucine can increase 3H4MV production in R. eutropha strain PHB⁻4. This result prompted us to generate leucine analog-resistant mutants of the PHB⁻4 strain by chemical mutagenesis in order to examine whether 3H4MV production is increased in mutants insensitive to leucine feedback inhibition. Strain 1F2, isolated as a leucine analog-resistant mutant using 4-aza-leucine, showed slightly higher levels of 3H4MV production (0.8 mol% in the monomer fraction) than did the parent strain (0.5 mol%) grown on fructose as the sole carbon source. Furthermore, when 10 g/liter leucine was added, recombinant strain 1F2 showed an enhanced 3H4MV fraction up to 3 mol%, whereas the parent strain showed 0.9 mol% 3H4MV under the same culture conditions. Strain 1F2 would be a useful host for P(3HB-co-3H4MV) production; however, complete analysis of the mutation positions in the 1F2 genome and phenotypic characterization of this strain had not been performed. This strain might have mutations in the 3H4MV biosynthesis-related genes, which provides a clue to understanding the 3H4MV biosynthetic route in R. eutropha.

In this study, the draft genome sequences of R. eutropha strain 1F2 and its parent strain PHB⁻4 were compared to identify mutated positions. Furthermore, a gene dosage study and an isotope tracer study were conducted to understand the 3H4MV biosynthesis pathway in R. eutropha. Based on these results, we propose a 3H4MV biosynthesis pathway involving condensation between isobutyryl coenzyme A (isobutyryl-CoA) and acetyl-CoA, which revises our understanding of 3H4MV biosynthesis.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. PHA-negative R. eutropha mutant strain PHB⁻4 (DSM541, subcultured in our laboratory) (16), strain 1F2 (15), and strain KNK-DCD1 (kindly donated by Kaneka Corporation, Hyogo, Japan) were employed as host strains for PHA synthesis. Recombinant plasmid pBBR1″C1_PsAB_Re, carrying the broad-substrate-specificity PHA synthase 1 gene from Pseudomonas sp. strain 61-3 (phaC1_Ps) and a monomer supplying enzyme genes from R. eutropha (phaAB_Re), was transformed into the host strain by transconjugation (17).

Plasmid construction.

To introduce a second promoter downstream from the R. eutropha pha_Re terminator in pBBR1″C1_PsAB_Re (17), the Aeromonas caviae pha_Ac promoter (18) was amplified with EcoRV and HindIII sites from pBBR1phaPCJ_AcAB_Re (19), using PCR primer sets listed in Table S1 in the supplemental material. A 0.4-kb pha_Ac promoter fragment was digested by EcoRV and HindIII and introduced at the same sites of pBBR1″C1_PsAB_Re to yield pBBR1″C1ABP_Ac. Plasmids pBBR1″C1ABP_AcIlvH(A36T) and pBBR1″C1ABP_AcBktB (see Fig. S1 in the supplemental material) were constructed as follows. Fragments of the ilvH gene encoding an Ala-to-Thr change at position 36 [ilvH(A36T)] and bktB with their native ribosomal binding sites were amplified with HindIII and KpnI sites from 1F2 genomic DNA by using PCR primer sets (see Table S1 in the supplemental material). Next, ilvH (0.5 kb) and bktB (1.2 kb) fragments were digested by HindIII and KpnI and introduced at the same sites of pBBR1″C1ABP_Ac to yield pBBR1″C1ABP_AcIlvH(A36T) and pBBR1″C1ABP_AcBktB, respectively.

To replace the pha_Ac promoter in pBBR1″C1ABP_AcBktB with the tac promoter, the tac promoter region was amplified from pTTQ19. The 0.2-kb tac promoter fragment with EcoRV and HindIII sites was introduced into the pha_Ac promoter-removed plasmid pBBR1″C1ABP_AcBktB to yield pBBR1″C1ABP_tacBktB (see Fig. S1 in the supplemental material).

Plasmids pBBR1″C1ABP_AcLeuA and pBBR1″C1ABP_AcLeuA(G479C) (see Fig. S1 in the supplemental material) were constructed as follows. A QuickChange Multi site-directed mutagenesis kit (Agilent Technology, Santa Clara, CA) was used to introduce point mutations (GGC [Gly] 479→TGC [Cys]) in the leuA gene from Escherichia coli W3110. The primer used for mutagenesis is shown in Table S1 in the supplemental material. Next, leuA and leuA(G479C) were amplified with XhoI and KpnI sites by using the PCR primer sets shown in Table S1 in the supplemental material. The 1.5-kb fragments were digested by XhoI and KpnI and introduced into the same sites of pBBR1″C1ABP_Ac to yield pBBR1″C1ABP_AcLeuA and pBBR1″C1ABP_AcLeuA(G479C), respectively.

For the promoter assay, the green fluorescent protein (GFP) (TurboGFP) gene was amplified from the pTurboGFP-B vector (Evrogen, Moscow, Russia). The 0.7-kb TurboGFP fragment with its native ribosomal binding site was introduced into bktB-removed plasmids pBBR1″C1ABP_AcBktB and pBBR1″C1ABP_tacBktB with HindIII and KpnI sites. The resulting plasmids were termed pBBR1″C1ABP_AcGFP and pBBR1″C1ABP_tacGFP, respectively (see Fig. S1 in the supplemental material).

Genomic DNA extraction and sequencing.

To prepare cell culture for genomic DNA extraction, R. eutropha strains PHB⁻4 and 1F2 were cultivated in nutrient-rich (NR) medium (10 g Bacto tryptone, 2 g yeast extract, and 10 g meat extract per liter of distilled water). Genomic DNA extraction was performed by using Qiagen Genomic-tip 20/G and the genomic DNA buffer set (Qiagen, Hilden, Germany) according to standard procedures provided by the manufacturer.

Genomic DNA sequencing of strains PHB⁻4 and 1F2 was performed by Operon Biotechnology Co. Ltd. (Tokyo, Japan) by using the Illumina MiSeq platform (Illumina Inc., San Diego, CA) and analyzed by using MiSeq Control software 2.0.5., RTA 1.16.18, and CASAVA-1.8.2. The alignments of both strains were compared to the R. eutropha H16 whole-genome sequence, the R. eutropha H16 chromosome 1 complete genome (gi|113866031; GenBank accession number NC_008313.1), the R. eutropha H16 chromosome 2 complete sequence (gi|116693960; accession number NC_008314.1), and the R. eutropha H16 megaplasmid pHG1 complete sequence (gi|38637668; accession number NC_005241.1|), provided by the NCBI.

PHA biosynthesis.

R. eutropha strains PHB⁻4, 1F2, and KNK-DCD1 expressing PhaC1_Ps and various enzymes were cultivated in a 500-ml shaking flask (130 strokes/min) containing 100 ml MS medium (9 g Na₂HPO₄·12H₂O, 1.5 g KH₂PO₄, 0.5 g NH₄Cl, 0.2 g MgSO₄·7H₂O, and 1 ml trace elements solution per liter of distilled water) (11) containing 20 g/liter fructose with or without various concentrations of amino acids (0 to 10 g/liter) at 30°C for 72 h. In all cases, 50 mg/liter kanamycin was added to the medium to maintain the plasmid in the cell. The cultivated cells were harvested by centrifugation and washed with distilled water to remove the medium components before being lyophilized. The residual fructose concentration was measured by using a d-glucose/d-fructose F kit (Roche Diagnostics, Basel, Switzerland) according to standard procedures provided by the manufacturer.

Promoter assay using green fluorescent protein.

In order to evaluate promoter activities, R. eutropha strain PHB⁻4 with a plasmid carrying the GFP gene under the control of either the pha promoter from A. caviae (P_Ac) or the tac promoter (P_tac) was cultivated in 1.7 ml MS medium containing 20 g/liter fructose for 24 h at 30°C. Cell growth was determined by measuring the optical density at 600 nm (OD₆₀₀). Fluorescence of GFP was measured by using an Mx3005 QPCR system (Agilent Technology, Santa Clara, CA) with a 6-carboxyfluorescein (FAM)/SYBR green I filter (excitation at 492 nm and emission at 516 nm). The fluorescence intensity was normalized by the OD₆₀₀.

Amino acid concentration assay.

R. eutropha strain PHB⁻4 and strain 1F2 were inoculated into 500-ml shake flasks containing 100 ml MS medium supplemented with 20 g/liter fructose and cultivated for 24 h at 30°C (130 strokes/min) to evaluate amino acid production. After 24 h, cells were removed by centrifugation, and the supernatant was stored at −20°C for amino acid concentration assays. The concentration of each amino acid was determined by high-performance liquid chromatography (HPLC) according to a method described previously (20). R. eutropha strain PHB⁻4 harboring pBBR1″C1_PsAB_Re and strain 1F2 harboring pBBR1″C1ABP_tacBktB were inoculated into 500-ml shake flasks containing 100 ml MS medium supplemented with 20 g/liter fructose and 0 to 10 g/liter leucine and cultivated for 72 h at 30°C (130 strokes/min). The residual leucine concentration after cultivation was also determined as described above.

PHA analysis.

PHA contents and composition were determined by gas chromatography (GC) (GC14B; Shimadzu, Kyoto, Japan) with a flame ionization detector and by GC-mass spectrometry (GC-MS) (GCMS-QC 2010; Shimadzu). Approximately 30 mg lyophilized cells was methanolyzed in the presence of 15% sulfuric acid before analysis (11). The resulting 3HA methyl esters were analyzed by GC-MS as described previously by Tanadchangsaeng et al. (13).

RESULTS

Genomic DNA analysis of strain 1F2.

Because R. eutropha strain 1F2 was generated by chemical mutagenesis using N-methyl-N′-nitro-N-nitrosoguanidine (NTG) (15), a number of mutations were expected to be randomly introduced into the genome. To determine the point mutation positions, genomic DNA sequencing of R. eutropha strains PHB⁻4 and 1F2 was performed, and the obtained sequences were compared. Briefly, the single nucleotide substitutions found in strain 1F2 relative to its parent strain PHB⁻4 are shown in Table 1 (detailed results are listed in Tables S2 and S3 in the supplemental material). From genomic DNA analysis, four mutations were found to have been introduced into the 1F2 strain genome: positions 518612, 1133514, and 3065189 in chromosome 1 and position 418411 in chromosome 2. Position 518612 contained a mutation in a region between the H16_A0492 and H16_A0493 genes, and the mutations at positions 1133514, 3065189, and 418411 were found in the coding sequence of the acetohydroxy acid synthase III (AHAS III) regulator subunit (IlvH) gene (H16_A1036), the type IV pilus assembly protein gene (H16_A2839), and the pimeloyl-CoA synthetase gene (H16_B0370), respectively.

TABLE 1.

Point mutation locations in R. eutropha strain 1F2 compared with its parent strain, PHB⁻4

Chromosome	Position	Mutation (strain PHB−4→strain 1F2)	Locus tag or regiona	Mutated genea
1	518612	G→C	Intergenic region between H16_A0492 and H16_A0493
	1133514	GCA (Ala-36)→ACA (Thr)	H16_A1036	Acetohydroxy acid synthase III regulator subunit (IlvH) gene
	3065189	CTC (Leu-117)→TTC (Phe)	H16_A2839	Type IV pilus assembly protein gene
2	418411	TAC (Tyr-377)→TTC (Phe)	H16_B0370	Pimeloyl-CoA synthetase gene

^aAccording to the KEGG database.

The 1F2 strain was able to resist higher concentrations of the leucine analog than its parent strain PHB⁻4 (15); however, in contrast to our expectation, no mutations were found in the leucine biosynthesis operon (leuABCD). Of four mutations found in strain 1F2, we focused on the mutation of Ala-36 to Thr (A36T) in IlvH, which, together with catalytic subunit IlvB, constitutes a regulatory subunit of AHAS (21). AHAS catalyzes the conversion of pyruvate to 2-acetolactate, the first reaction of branched amino acid biosynthesis, as shown in Fig. 1, and IlvH plays a critical role in end product feedback inhibition by branched amino acids such as valine and leucine. Previously reported observations (15) suggested that the 3H4MV biosynthesis pathway is partly shared with that of branched amino acids. Thus, the IlvH(A36T) mutation was the most likely source of the acquired resistance to feedback inhibition by branched amino acids, thereby increasing metabolic flux toward branched amino acid biosynthesis concomitant with 3H4MV biosynthesis.

FIG 1.

Putative biosynthesis pathways of branched amino acids and endpoint feedback inhibition in R. eutropha. IlvBH, acetohydroxy acid synthase III; IlvC, ketol-acid reductoisomerase; IlvD, dihydroxy acid dehydratase; IlvE, branched-chain amino acid aminotransferase; LeuA, isopropylmalate synthase; LeuCD, isopropylmalate isomerase; LeuB, 3-isopropylmalate dehydrogenase. ⦵, inhibition; ⊕, relief of inhibition.

Assay of leucine and valine in culture supernatants.

E. coli IlvH is known to be extremely sensitive to feedback inhibition by valine (21). It was shown previously for the E. coli K-12 strain that mutagenesis of Ala-36 to Val (A36V) in IlvH resulted in the strain being less sensitive to feedback inhibition by valine (22). The 1F2 strain has the amino acid substitution (A36T) at the same position in IlvH. In addition, the amino acid sequence homology between IlvH from R. eutropha H16 and that from E. coli K-12 (MG1655) was relatively high (60% identical and 79% positive) by BLAST analysis (see Fig. S2 in the supplemental material). Thus, IlvH(A36T) found in strain 1F2 was expected to be less sensitive to feedback inhibition, which might lead to branched amino acid overproduction.

To evaluate branched amino acid production by strains PHB⁻4 and 1F2, these strains were cultivated in MS medium supplemented with 20 g/liter fructose. The amino acids present in the culture supernatant were derivatized and subsequently analyzed by HPLC (Fig. 2). Strain PHB⁻4 produced a small amount of valine (2.6 mg/liter), whereas strain 1F2 produced much more valine (32.4 mg/liter) and leucine (4.9 mg/liter). This suggested that feedback inhibition by branched amino acids in strain 1F2 was less sensitive, possibly as a result of the A36T mutation in IlvH.

FIG 2.

Amino acids produced by R. eutropha strains PHB⁻4 and 1F2. Cultures were grown in MS medium containing fructose (20 g/liter) at 30°C for 24 h. Concentrations of valine and leucine in culture supernatants were determined by HPLC. The results are the averages of data from three independent cultivations. ND, not detected.

Effect of ilvH(A36T) dosage on the 3H4MV fraction.

To investigate the gene dosage effect of ilvH(A36T) on 3H4MV biosynthesis, R. eutropha recombinant strains PHB⁻4, 1F2, and KNK-DCD1 harboring pBBR1″C1ABP_AcIlvH(A36T) were cultivated in MS medium containing 20 g/liter fructose (the results are listed in Table 2). For all strains, no significant effect of ilvH(A36T) dosage on cell growth and PHA content was observed. No significant dosage effect on 3H4MV production (from 9.5 to 9.7 mg/liter) was observed for recombinant strain 1F2, whereas slight increases in 3H4MV production, from 5.9 to 8.9 mg/liter and from 7.6 to 10 mg/liter, were observed for recombinant strains PHB⁻4 and KNK-DCD1, respectively. Wild-type IlvH may have a dominant effect in these strains, replacing IlvH(A36T) in the AHAS complex, resulting in a relief of feedback inhibition and a slight increase in 3H4MV production.

TABLE 2.

Gene dosage effect on PHA production in R. eutropha recombinants grown on fructose as the sole carbon sourceⁱ

Host strain	Dosed gene	Promoter for dosed gene	Avg dry cell wt (g/liter) ± SD	Avg PHA content (wt%) ± SD	Avg PHA composition (mol%)			Avg level of 3H4MV production (mg/liter) ± SD
					3HB	3HV	3H4MV
PHB−4	Controla		1.62 ± 0.04	53 ± 1	99.1	0.4	0.5	5.9 ± 0.2
	ilvH(A36T)b	PAcg	1.64 ± 0.07	54 ± 3	98.8	0.4	0.8	8.9 ± 0.1
	leuAc	PAc	1.33 ± 0.02	45 ± 2	99.1	0.4	0.5	3.7 ± 0.3
	leuA(G479C)d	PAc	1.47 ± 0.05	50 ± 1	99.2	0.4	0.4	3.9 ± 0.2
	bktBe	PAc	1.83 ± 0.03	58 ± 1	98.8	0.5	0.7	9.9 ± 0.3
	bktBf	Ptach	1.70 ± 0.08	61 ± 1	98.6	0.6	0.8	11 ± 1
1F2	Control		1.65 ± 0.01	54 ± 1	97.7	1.5	0.8	9.5 ± 0.3
	ilvH(A36T)	PAc	1.78 ± 0.01	59 ± 2	98.6	0.7	0.7	9.7 ± 0.3
	leuA	PAc	1.51 ± 0.02	49 ± 1	97.8	1.3	0.9	8.8 ± 0.2
	leuA(G479C)	PAc	1.50 ± 0.06	53 ± 1	99.1	0.5	0.4	4.2 ± 0.2
	bktB	PAc	1.64 ± 0.02	51 ± 1	96.2	2.3	1.5	17 ± 1
	bktB	Ptac	1.48 ± 0.06	55 ± 2	95.9	2.4	1.7	19 ± 2
KNK-DCD1	Control		1.80 ± 0.05	64 ± 1	99.2	0.3	0.5	7.6 ± 0.4
	ilvH(A36T)	PAc	1.75 ± 0.01	57 ± 2	98.6	0.6	0.8	10 ± 1
	leuA	PAc	1.38 ± 0.07	46 ± 3	99.2	0.3	0.5	3.9 ± 0.9
	leuA(G479C)	PAc	1.49 ± 0.02	51 ± 1	99.2	0.4	0.4	4.1 ± 0.1
	bktB	PAc	1.86 ± 0.04	61 ± 1	98.7	0.5	0.8	11 ± 1
	bktB	Ptac	1.61 ± 0.09	55 ± 3	98.6	0.5	0.9	10 ± 2

^apBBR1″C1_PsAB_Re.

^bpBBR1″C1ABP_AcIlvH(A36T).

^cpBBR1″C1ABP_AcLeuA.

^dpBBR1″C1ABP_AcLeuA(G479C).

^epBBR1″C1ABP_AcBktB.

^fpBBR1″C1ABP_tacBktB.

^gpha promoter from A. caviae.

^htac promoter.

ⁱCells were cultured in MS medium containing fructose (20 g/liter) at 30°C for 72 h. The results are the averages of data from three independent cultivations (the standard deviation for PHA composition was <5% of the mean).

Effect of leuA(G479C) dosage on the 3H4MV fraction.

LeuA is a subunit of 2-isopropylmalate synthase, which is the first enzyme of the leucine biosynthesis pathway and sensitive to feedback inhibition by leucine (Fig. 1). An E. coli K-12 mutant with a LeuA Gly-479-to-Cys (G479C) substitution was reported to be tolerant to feedback inhibition by leucine (23). To pull carbon flux toward leucine biosynthesis in R. eutropha for 3H4MV enhancement, gene dosage of leuA and leuA(G479C) was examined. Recombinant strains PHB⁻4, 1F2, and KNK-DCD1 expressing leuA or leuA(G479C) were cultured on fructose as the sole carbon source, and the results are listed in Table 2. The level of 3H4MV production of all strains was decreased by increased dosages of leuA and leuA(G479C). Particularly, recombinant strain 1F2 expressing leuA(G479C) showed a significant decrease in the level of 3H4MV production, from 9.5 to 4.2 mg/liter. The carbon flux from 2-ketoisovalerate to the leucine biosynthesis pathway might be increased by the gene dosage; contrary to our expectation, 3H4MV production was not enhanced. The enhancement of leucine biosynthesis had a negative effect on the production of 3H4MV, implying that leucine and its biosynthetic intermediates might not be direct precursors for 3H4MV synthesis.

Effect of bktB dosage on the 3H4MV fraction.

Another possible route for 3H4MV biosynthesis is the formation of the 3H4MV carbon backbone via a condensation reaction. BktB is a broad-substrate-specificity 3-ketothiolase, which catalyzes a condensation reaction between isobutyryl-CoA and acetyl-CoA to form the 3H4MV precursor (24). Here, we studied the feasibility of increasing 3H4MV production by increasing the bktB dosage. Recombinant strain 1F2 harboring pBBR1″C1ABP_AcBktB showed a significant increase in 3H4MV production by increasing the bktB dosage (Table 2): 3H4MV production reached 17 mg/liter on fructose as the sole carbon source. In addition to valine synthesis, the concentration of isobutyryl-CoA, a valine degradation intermediate, might be enhanced in strain 1F2; therefore, 3H4MV synthesis was stimulated in the BktB-overexpressing strain. This result implied that 3H4MV was derived via a BktB-catalyzed condensation reaction in R. eutropha.

To further increase 3H4MV production, the promoter for bktB expression was changed to a stronger one. Fukui et al. (25) showed previously that among the various promoters tested, the tac promoter has the strongest activity in R. eutropha. Hence, the pha promoter from A. caviae (P_Ac) was replaced with the tac promoter (P_tac), yielding a new plasmid, pBBR1″C1ABP_tacBktB. Recombinant strains PHB⁻4, 1F2, and KNK-DCD1 harboring this plasmid were cultured, and the results are listed in Table 2. Although the tac promoter is known to have strong activity, 3H4MV production (from 17 to 19 mg/liter) in recombinant 1F2 was not significantly increased by promoter replacement. To understand this result, a GFP assay for promoter activity in R. eutropha was performed (Table 3). This experiment showed that the GFP fluorescence intensities of P_Ac and P_tac were 64- and 89-fold higher than that of the negative-control plasmid (pBBR1″C1ABP_AcBktB), revealing that P_Ac functioned as well as P_tac in R. eutropha.

TABLE 3.

GFP fluorescence of R. eutropha strain PHB⁻4 under the control of the pha promoter and the tac promoter^d

Plasmid	Promoter for GFP expression	Avg GFP fluorescence/OD600 (103) ± SD
pBBR1″C1ABPAcBktBa		1.4 ± 0.4
pBBR1″C1ABPAcGFPb	PAc	90 ± 3
pBBR1″C1ABPtacGFPc	Ptac	125 ± 3

^aControl plasmid.

^bGFP gene placed under the control of the A. caviae pha promoter.

^cGFP gene placed under the control of the tac promoter.

^dR. eutropha PHB⁻4 harboring each plasmid was cultured in MS medium containing fructose (20 g/liter) at 30°C for 24 h. The results are the averages of data from three independent cultivations.

PHA biosynthesis with leucine supplementation.

Our previous study (15) demonstrated that supplementation of excess leucine can increase 3H4MV production in a PHA copolymer synthesized by recombinant strains PHB⁻4 and 1F2. We examined whether BktB-overexpressing strain 1F2 (strain 1F2/pBBR1″C1ABP_tacBktB) shows the same trend by adding excess leucine. Figure 3 shows the results, together with that for the other strain tested. The strains showed similar trends in cell growth and PHA contents. The BktB-overexpressing strains showed increased 3H4MV production with the addition of excess leucine. The level of 3H4MV production was 150 ± 4 mg/liter; as a result, the 3H4MV fraction reached 4.8 mol% (Fig. 3), the highest fraction so far recorded with leucine supplementation.

FIG 3.

PHA production by R. eutropha recombinants in the presence of fructose (20 g/liter) and various concentrations of l-leucine (0 to 10 g/liter). Recombinant strain PHB⁻4/pBBR1″C1_PsAB_Re (A) and recombinant strain 1F2/pBBR1″C1ABP_tacBktB (B) were cultured for 72 h at 30°C. Data for dry cell weight (triangles), PHA (circles), and the 3H4MV fraction in PHA (squares) are the averages of data from three independent cultivations.

To gain insight into the role of leucine in 3H4MV biosynthesis, the amino acid contents of culture supernatants were assayed. The amino acids present in the supernatant of recombinant strain PHB⁻4 and BktB-overexpressing strain 1F2 cultured with 0 to 10 g/liter leucine were derivatized and subsequently detected by HPLC (Fig. 4). The detected leucine concentration at the end of cultivation was 0.2 g/liter, meaning that the added leucine was almost consumed by the cells within 72 h of cultivation. Host-produced valine was detected at 0.34 and 1.35 g/liter in the supernatant of the BktB-overexpressing 1F2 strain with supplementation with 7 and 10 g/liter leucine, respectively. A small amount of valine (0.10 g/liter) was also detected in the supernatant of recombinant strain PHB⁻4, smaller than the amount detected with BktB-overexpressing strain 1F2. Because BktB-overexpressing strain 1F2 showed significant increases in the levels of both 3H4MV and valine production, we speculate that 3H4MV biosynthesis is coordinated with valine metabolism.

FIG 4.

Concentrations of amino acids and residual fructose in culture supernatants of R. eutropha recombinants. Recombinant strain PHB⁻4/pBBR1″C1_PsAB_Re (A) and recombinant strain 1F2/pBBR1″C1ABP_tacBktB (B) were cultured for 72 h at 30°C in the presence of fructose (20 g/liter) and various concentrations of l-leucine (0 to 10 g/liter). Data for residual fructose (circles), residual leucine (open triangles), and host-produced valine (closed triangles) are the averages of data from three independent cultivations.

PHA biosynthesis with labeled leucine.

To determine whether leucine is converted to 3H4MV while keeping its C₆ backbone, PHA biosynthesis was carried out by using stable isotope-labeled leucine. To this end, recombinant strain 1F2 harboring pBBR1″C1_PsAB_Re was cultured in MS medium containing 20 g/liter fructose and 10 g/liter leucine with a 1-¹³C or 2-¹³C label, followed by GC-MS analysis of the purified PHA (the copolymer composition is shown in Table S4 in the supplemental material).

Table 4 lists the isotope ratios (m/z 104) of the base peak ion fragments (m/z 103 [HO⁺=CHCH₂CO₂CH₃, generated by cleavage of the side chain from 3HA methyl ester]) that were observed in the 3HA units synthesized by recombinant strain 1F2 supplemented with 10 g/liter labeled leucine. When nonlabeled leucine was supplied as a control, the isotope ratio of the 3H4MV base peak ion fragment was 3.9%, representing the natural ¹³C abundance. Assuming that 3H4MV is derived from leucine while keeping its carbon backbone, supplementation with [1-¹³C]leucine should dramatically increase the isotope ratio (m/z 104) in 3H4MV. However, this experiment showed a slight increase in the isotope ratio for 3H4MV (5.8%), but it was not specific to 3H4MV. Other 3HAs also showed an increased isotope ratio (from 4.5 to 11% for 3HB and from 1.7 to 4.6% for 3HV). In the case of [2-¹³C]leucine, the isotope ratio in each 3HA unit was further increased compared to those of [1-¹³C]leucine and nonlabeled leucine; nevertheless, the isotope ratio (m/z 104) detected in 3H4MV was still small (12%), and the increase was also not specific to 3H4MV. This result indicates that 3H4MV is not directly derived from labeled leucine.

TABLE 4.

Ratio of isotope peaks detected by GC-MS analysis of PHA synthesized with supplementation with 10 g/liter labeled leucine or valine mixture^a

Supplemented amino acid	Supplemented amino acid concn(s) (g/liter) (labeled position)	Ratio of isotope fragments (m/z 104) in base peak ion fragment (%)b
		3HB	3HV	3H4MV
Leucine	10 (nonlabeled)	4.5	1.7	3.9
	10 (1-13C labeled)	11	4.6	5.8
	10 (2-13C labeled)	30	18	12
Valine	10 (nonlabeled)	4.5	3.4	4.4
	2.5 (1-13C labeled) + 7.5 (nonlabeled)	5.3	3.5	5.0
	2.5 (2-13C labeled) + 7.5 (nonlabeled)	9.3	7.9	27

^aR. eutropha 1F2 harboring pBBR1″C1_PsAB_Re was cultured in MS medium containing fructose (20 g/liter) with 10 g/liter labeled or nonlabeled leucine or valine at 30°C for 72 h. When 10 g/liter valine was added, 1 g/liter leucine and 1 g/liter isoleucine were also added to the medium to help cell growth.

^bm/z 103 (HO⁺=CHCH₂CO₂CH₃, generated by cleavage of the side chain from 3HA methyl ester) is the base peak of 3HA methyl esters. The isotope ratio was calculated from peak intensities of m/z 103, 104, and 105.

PHA biosynthesis with valine supplementation.

A high concentration of valine may inhibit valine, leucine, and isoleucine biosynthesis; therefore, R. eutropha is not able to grow in the presence of 10 g/liter valine (15). On the other hand, host-produced valine was detected in the culture supernatants of strains PHB⁻4 and 1F2 (Fig. 2). Here, we aimed to elucidate the metabolic relationship between valine and 3H4MV. To this end, recombinant strains PHB⁻4 and 1F2 were cultivated in MS medium containing fructose (20 g/liter) with or without the addition of valine (10 g/liter). Leucine and isoleucine were also added to the medium at 1 g/liter each to help cell growth. The results are shown in Table 5. Without valine addition, the level of 3H4MV production of each strain was in the range of 4.7 to 53 mg/liter. In contrast, the addition of 10 g/liter valine significantly enhanced 3H4MV production to 162 mg/liter, with the 3H4MV fraction reaching 4.1 mol%. These results demonstrated that the presence of valine in the culture enhanced 3H4MV biosynthesis, probably due to an increased provision of intermediates for 3H4MV synthesis by valine degradation.

TABLE 5.

PHA production by R. eutropha recombinants grown with valine supplementation^d

Host strain	bktB dosage	Valine concn added (g/liter)c	Avg dry cell wt (g/liter) ± SD	Avg PHA content (wt%) ± SD	PHA composition (mol%)			Avg level of 3H4MV production (mg/liter) ± SD
					3HB	3HV	3H4MV
PHB−4	Nonea	0	5.57 ± 0.06	64 ± 2	99.5	0.4	0.1	4.7 ± 0.1
		10	6.24 ± 0.25	32 ± 3	92.4	5.6	2.0	52 ± 7
1F2	Nonea	0	5.63 ± 0.43	65 ± 4	96.3	2.9	0.7	34 ± 5
		10	6.40 ± 0.39	40 ± 2	92.0	5.2	2.8	92 ± 11
	Overexpressedb	0	5.54 ± 0.18	65 ± 2	94.0	4.9	1.1	53 ± 6
		10	7.24 ± 0.09	42 ± 1	88.5	7.4	4.1	162 ± 6

^apBBR1″C1_PsAB_Re.

^bpBBR1″C1ABP_tacBktB.

^cLeucine (1 g/liter) and isoleucine (1 g/liter) were added to support cell growth.

^dCells were cultured in MS medium containing fructose (20 g/liter) with branched amino acid supplementation at 30°C for 72 h. The results are the averages of data from three independent cultivations (the standard deviation for PHA composition was <5% of the mean).

PHA biosynthesis with a mixture of labeled valines.

To provide convincing evidence that 3H4MV was synthesized from valine as a main precursor, cultivation with labeled valine was carried out. Recombinant strain 1F2 harboring pBBR1″C1_PsAB_Re was cultivated in MS medium containing 20 g/liter fructose and 10 g/liter of a valine mixture (2.5 g/liter [1-¹³C]valine or [2-¹³C]valine and 7.5 g/liter nonlabeled valine). To help the cell growth, the medium was supplemented with 1 g/liter of leucine and 1 g/liter isoleucine (the composition of the 3H4MV copolymer is shown in Table S4 in the supplemental material). The observed isotope ratios (m/z 104) of 3HAs are listed in Table 4. The isotope ratio of the 3H4MV base peak ion fragment was 4.4% when the medium was supplemented with nonlabeled valine as a control. In the case of supplementation with the [1-¹³C]valine mixture, the isotope ratio of 3H4MV was slightly increased to 5.0%; however, it was not specific to 3H4MV (from 4.5 to 5.3% for 3HB and from 3.4 to 3.5% for 3HV), as also observed for [1-¹³C]leucine supplementation. On the other hand, the isotope ratio of 3H4MV was increased to 27% when the medium was supplemented with the [2-¹³C]valine mixture. This result strongly suggests that valine was converted to 3H4MV as a main precursor.

DISCUSSION

Incorporation of the 3H4MV unit into the sequence of P(3HB) improves the material properties of P(3HB), making it soft and flexible. However, in R. eutropha, the fraction of the 3H4MV unit that can be incorporated into P(3HB-co-3H4MV) is only 0.5 mol% when fructose is fed as the sole carbon source. To achieve 3H4MV enhancement, some structurally related fatty acid 3H4MV precursors, such as 4-methylvalerate (4MV) and 4-methylpentenoate (4MPE), are needed (13); however, these precursors are toxic for bacterial cell growth. Therefore, uncovering the 3H4MV biosynthesis pathway in R. eutropha is necessary for metabolic engineering to achieve increased fractions of the 3H4MV unit in PHA copolymers without using 3H4MV precursors.

Previously, we found that 3H4MV production could be slightly increased by feeding an excess amount of leucine to recombinant R. eutropha PHB⁻4 (15). This finding led us to hypothesize that the 3H4MV unit is synthesized from leucine or its biosynthetic intermediates, while keeping its C₆ backbone. Our hypothesis was supported by the fact that the leucine analog-resistant mutant strain 1F2, which is tolerant of leucine feedback inhibition, produced a larger amount of 3H4MV. Based on these facts, we presumed that the 3H4MV biosynthesis pathway in R. eutropha might be involved in the conversion of leucine to 4MV and/or 4MPE by functionally relevant enzymes such as those found in the leucine degradation pathway of an obligate anaerobe, Clostridium difficile (currently designated Peptoclostridium difficile). Indeed, by employing these enzymes from C. difficile, the 3H4MV biosynthesis pathway was successfully constructed in recombinant E. coli; however, the key enzyme 2-hydroxy-4-methylvaleryl-CoA dehydratase (HadIBC) was particularly sensitive to oxygen (20). As for R. eutropha, 3H4MV is produced without any oxygen limitation, and no HadIBC gene homologs were found in the genome by BLASTP analysis. Taken together, these observations imply the possibility of a different 3H4MV synthetic route. To gain a further understanding of the 3H4MV biosynthesis pathway in R. eutropha, we conducted a genome-based analysis of strain 1F2, which is thought to have mutations in the 3H4MV biosynthesis-related genes (15).

The genome of R. eutropha strain H16 was previously sequenced by a genome project (26, 27), but strain PHB⁻4, a PHA-negative mutant derived from strain H16 (16), and its daughter strain 1F2 were not. Therefore, the genome of strain PHB⁻4 was first sequenced in this study. The comparison of draft genome sequences of strains H16 and PHB⁻4 revealed that strain PHB⁻4 has 81 single nucleotide substitutions and 80 inserted or deleted nucleotides (see Table S3 in the supplemental material). No mutations were found in megaplasmid pHG1. The inability to produce PHA in strain PHB⁻4 is attributed to the G-to-A mutation at position 1557672 (TGG→TAG, a nonsense mutation in phaC1_Re) in chromosome 1, thus confirming the causative mutation, as described in previous studies (28, 29). Recently, Raberg et al. (29) conducted a proteome analysis of strain PHB⁻4 and identified 20 upregulated proteins compared to those in strain H16. Due to the inability to accumulate PHA, strain PHB⁻4 is known to excrete pyruvate (30). The upregulated proteins were involved mostly in the reduction of pyruvate and acetyl-CoA pool sizes (29). In our genome analysis, however, all upregulated proteins in strain PHB⁻4 showed no mutation in the corresponding genes. Therefore, these upregulated proteins are thought to result from the maintenance of metabolic balance in these cells.

The mutations found in strain PHB⁻4 would not be involved in 3H4MV enhancement because both recombinant strains PHB⁻4 and KNK-DCD1 (phaC1_Re-deleted strain of strain H16) produced 3H4MV at similar levels (Table 2). In strain 1F2, we identified four additional mutations relative to strain PHB⁻4, at positions 518612, 1133514, and 3065189 in chromosome 1 and at position 418411 in chromosome 2. Of the four mutations found in strain 1F2, only the mutation at position 1133514 (amino acid substitution of A36T), which resides in the ilvH gene encoding a regulator subunit of AHAS, is of relevance to branched amino acid biosynthesis. It was considered that the 1F2 strain mutation in the ilvH gene is responsible for the increased-3H4MV phenotype.

Lu et al. (31) reported previously that various single mutations at the N-terminal region of R. eutropha IlvH reduced feedback inhibition by valine. In addition, Mendel et al. (22) showed that the A36V mutation in E. coli IlvH resulted in the strain being less sensitive to feedback inhibition by valine. The 1F2 strain also had a mutation at this position (A36T mutation) in IlvH and showed increased production of 3H4MV, valine, and leucine (Table 2 and Fig. 2). The A36T mutation in IlvH would increase tolerance to feedback inhibition by valine, thus shifting carbon flux toward valine and leucine and concomitantly inducing 3H4MV production.

Although the A36T mutation in IlvH would explain the valine tolerance, it does not account for the leucine analog tolerance of strain 1F2. An interesting observation to explain the underlying mechanism was reported previously by Wiegel and Schlegel (32), who demonstrated that valine has antagonist activity for R. eutropha LeuA in the presence of leucine. Wiegel and Schlegel showed that LeuA activity was decreased to 14% in the presence of 0.05 mM leucine, while this activity was recovered to 65.5% by the addition of 1 mM valine. In this study, we observed that the 1F2 strain overproduced valine, thereby possibly allowing recovery of LeuA activity even in the presence of the leucine analog. This is one potential explanation for why strain 1F2 acquired leucine analog tolerance by the A36T mutation in IlvH.

The isotope tracer study using 1-¹³C-labeled leucine and valine showed that the ¹³C abundance was increased for all 3HA units (Table 4). In the bacterial leucine and valine degradation pathway (Fig. 5), the first reaction is deamination, followed by decarboxylation of leucine's first carbon (C-1). R. eutropha is a facultative chemolithoautotrophic bacterium able to grow with hydrogen and carbon dioxide under aerobic conditions; Shimizu et al. (33) showed previously that carbon dioxide assimilation is potentially functional even under heterotrophic growth conditions. Therefore, the increase in the ¹³C abundance in 3HA units observed here may be attributed to the assimilation of ¹³CO₂ derived from labeled leucine by decarboxylation. Similarly, 2-¹³C-labeled leucine was also degraded into ¹³CO₂ and some metabolites that were incorporated into all 3HA units, but not specifically the 3H4MV unit, when [2-¹³C]leucine was added. Furthermore, we performed an isotope tracer study using [5,5,5-D₃]leucine labeled with three deuteriums on the branched side chain (see Table S4 in the supplemental material). GC-MS analysis of the silylated 3HA methyl esters revealed that deuterium was not specifically located in the side chain of 3H4MV (see Table S5 in the supplemental material), implying that the branched structure of 3H4MV was not derived from that of leucine. From these observations, the carbon backbone of leucine would not be kept in 3H4MV biosynthesis from leucine as the substrate.

FIG 5.

Possible 3H4MV biosynthesis pathway from sugars in R. eutropha strain 1F2. IlvBH* has the A36T mutation in the IlvH subunit and is thought to be less sensitive to feedback inhibition by branched amino acids. BktB, 3-ketothiolase; PhaB, acetoacetyl-CoA reductase; PhaC, PHA synthase. See Fig. 1 for other enzymes.

Apart from leucine metabolism, BktB overexpression increased the 3H4MV fraction in strain 1F2 (Table 2). Because BktB catalyzes the condensation reaction between C_n acyl-CoA and acetyl-CoA (C₂) to form C_{n + 2} acyl-CoA, it is reasonable to assume that the 3H4MV biosynthesis pathway involves the condensation reaction of acetyl-CoA and isobutyryl-CoA, which has a branched C₄ acyl moiety. Isobutyryl-CoA is known as a metabolite of the valine degradation pathway (34, 35). Also, in isotope trace studies, the isotope ratio of 3H4MV was significantly increased when the medium was supplemented with a [2-¹³C]valine mixture, because [2-¹³C]valine was converted to isobutyryl-CoA, keeping labeled carbon (Fig. 5). This fact has highlighted that the valine degradation pathway makes a channeling route to the 3H4MV biosynthesis pathway.

Based on these observations, we propose the 3H4MV biosynthesis pathway in R. eutropha (Fig. 5). The isobutyryl-CoA derived from valine degradation and acetyl-CoA would constitute the 3H4MV carbon backbone through a condensation reaction catalyzed by BktB. Following reduction by PhaB, 3H4MV-CoA is finally polymerized by PHA synthase with the release of free CoA. Because 3H4MV production is increased with increasing amounts of supplemented leucine, we previously speculated that leucine functions as a direct precursor of 3H4MV (15). However, the observations in this study led us to revise the 3H4MV biosynthesis pathway into a new pathway involving the condensation reaction between isobutyryl-CoA and acetyl-CoA. Based on the revised pathway, the role of supplemented leucine during 3H4MV synthesis can be explained as follows: leucine addition enhanced 3H4MV synthesis because the leucine biosynthesis pathway is subjected to feedback inhibition by added leucine. As a result, the flux from 2-ketoisovalerate to leucine decreased, while the flux from 2-ketoisovalerate to valine and isobutyryl-CoA increased. In addition, leucine was degraded into acetyl-CoA, and its pool size would be enlarged in the cells. Thus, 3H4MV synthesis was promoted due to the enhanced supply of isobutyryl-CoA and acetyl-CoA.

Our goal is to synthesize P(3HB-co-3H4MV) with >5 mol% 3H4MV using sugars as the sole carbon source. At present, the 3H4MV fraction achieved by using fructose is 1.7 mol% (Table 2), which is still lower than the desired level. Additionally, 2.4 mol% 3HV was incorporated concomitant with 3H4MV. Several studies (36–39) have indicated that valine is an effective precursor for 3HV synthesis because propionyl-CoA is generated by valine degradation with a release of valine's first and second carbons (C-1 and C-2) as carbon dioxide (40), and the 3HV carbon backbone is formed by the BktB-catalyzed condensation of propionyl-CoA and acetyl-CoA. As discussed in this study, isobutyryl-CoA is also a product of valine degradation and an important intermediate for 3H4MV synthesis. Isobutyryl-CoA is a more upstream intermediate than propionyl-CoA in the valine degradation pathway; therefore, it may be possible to further increase the 3H4MV fraction by weakening the downstream pathway from isobutyryl-CoA in valine degradation in order to pull isobutyryl-CoA toward 3H4MV synthesis. To achieve the goal, we have undertaken further studies.