Co-utilization of glucose and xylose in synthetic Escherichia coli consortia for efficient 2-hydroxy-3-methylvalerate biosynthesis

Abstract

2-Hydroxy-3-methylvalerate (HMV), a kind of bioactive aliphatic 2-hydroxy acids, is a building block and a key intermediate for pharmaceuticals, including beauvericin, yet its microbial synthesis remains largely unexplored. Here, we present a synthetic biology strategy for high-titer production of HMV that couples mixed-sugar utilization with a division-of-labor microbial consortium. Co-feeding glucose and xylose synchronized substrate uptake with product formation, eliminating intermediate overflow and rerouting carbon from by-products into efficient HMV biosynthesis. To reduce the metabolic burden of the host, the HMV pathway is divided into two engineered strains: one optimized for glucose-to-intermediate conversion, the other for xylose-to-HMV completion. The best consortium, KMV-G-X, produces 2184.6 ± 111.8 mg/L HMV, comprising 82.2% of the total 2-hydroxy acids produced. Compared to mono-culture using glucose as a single substrate, this consortium exhibits less catabolic interference and enhanced HMV biosynthesis efficiency. This study shows that substrate utilization and pathway division of labor in synthetic consortium convert mixed sugars into high-value chemicals, boosting titer and robustness for scalable green production.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13068-025-02716-4.

Background

The increasing demand for sustainable and efficient production of valuable chemicals has driven the development of microbial metabolic engineering and synthetic biology [1, 2]. 2-Hydroxy acids (2-HAs) are a class of important organic acids bearing a hydroxyl group at the α-position of the carbon chain, which occur widely in natural products and metabolic pathways [3, 4]. Owing to their dual functionality of carboxylic acid and alcohol, 2-HAs not only participate in numerous biosynthetic and energy-yielding reactions but also serve as key chiral building blocks for pharmaceuticals, specialty chemicals, cosmetics and biodegradable materials [5–7]. For example, Sang Yup Lee’s group systematically rewired the metabolism of E. coli to produce high-molecular-weight poly (2-hydroxyisovalerate-co-lactate) directly from glucose at 9.6 wt% by outperforming chemical routes [8]. They further achieved one-step fermentation of glucose/xylose to 40.4 wt% PLGA whose molecular weight and thermal properties match petro-based counterparts, while an evolved PHA synthase enabled precise incorporation of functional monomers, such as 3HB, 4HB and 2HIV [9].

As one of the 2-HAs, 2-hydroxy-3-methylvalerate (HMV) is an important aliphatic hydroxy acid with significant biological activity, which can also serve as an intermediate in the synthesis of certain drugs, such as beauvericin [10]. Despite these promising attributes, the HMV biosynthesis remains extremely narrow. The only well-documented route is the “L-isoleucine-initiated” pathway, which is converted to KMV by a transaminase and then to HMV by a keto-acid reductase [11, 12]. Significantly, the feedback inhibition of branched-chain amino acids may limit the biosynthesis of HMV, making it still primarily dependent on chemical methods for production [13, 14]. Therefore, developing efficient methods for synthesizing HMV has become a research focus.

Unlike traditional mono-cultures, synthetic consortia with two or more strains with defined division of labor, form a more efficient and flexible biomanufacturing platform [15–17]. By co-culturing strains with distinct functions, the growth and production of bacterial strains are well-balanced, so complex feedstocks such as glucose-xylose mixtures or cellulose hydrolysates are efficiently converted into the final product [18–20]. Currently, synthetic consortia have been employed for biosynthesis using glucose–xylose mixtures or cellulose hydrolysates as substrates [21–23]. By designing and optimizing an E. coli-E. coli co-culture system, the efficient biosynthesis of chlorogenic acid (CGA) was achieved. Under the condition of a glucose-to-xylose ratio of 2:1, the accumulation of CGA reached 131.3 ± 7.9 mg/L [24]. In addition, in the orthogonal substrate utilization E. coli-E. coli co-culture system, the wild-type E. coli utilized only glucose, while the engineered strain E. coli LY180 utilized only xylose. At a 2:1 glucose-to-xylose ratio, this consortium converted 100 g/L total sugars into 46 g/L ethanol [25]. However, previous studies on co-culture systems have primarily focused on alleviating substrate competition or expanding the substrate spectrum, and there is still a lack of systematic exploration into how precise allocation of sugar uptake and step-by-step integration of upstream and downstream modules determines the final product titer.

In this study, we designed an HMV biosynthesis pathway based on the citramalate-derived pathway and engineered a series of E. coli-E. coli consortia that enabled the co-utilization of glucose and xylose for highly efficient HMV biosynthesis. With a rational division of substrate utilization and biosynthetic pathway, the best-performing E. coli-E. coli consortium KMV-G-X achieved an HMV titer of 2184.6 ± 111.8 mg/L, which was 11.8 times higher than that of the glucose-xylose co-utilization strain BL-P mono-culture. This work not only systematically elucidated how the pathway allocation and substrate utilization influenced HMV titer and its percentage of the 2-HAs products, but also advanced synthetic consortium fermentation strategies and provided a novel approach for the sustainable production of valuable chemicals from renewable biomass resources.

Materials and methods

Media

Luria-Bertani (LB) medium was used for the cultivation of seed cultures. LB medium comprised the following (per liter): 10 g tryptone, 5 g yeast extract, and 10 g NaCl. M9Y medium was used for 2MBA biosynthesis in E. coli. One liter of M9Y medium contained Na₂HPO₄ 6.8 g, KH₂PO₄ 3.0 g, NaCl 0.5 g, NH₄Cl 1.0 g, 5 g yeast extract, 15.7 g 3-morpholinopropanesulfonic acid, 2 mM MgSO₄, 0.1 mM CaCl₂, 2.78 mg FeSO₄, 1 mg Vitamin B1, 1 mL trace elements. For mono-culture, each liter of M9Y medium was routinely supplied with 10 g glucose and 10 g xylose. In the co-culture, the two sugars were adjusted in parallel according to the desired mass ratio (g/g), while the total sugar concentration was kept constant at 20 g/L. The working concentrations of trace elements were: 0.371 mg/L (NH₄)₆Mo₇O₂₄·4H₂O, 0.243 mg/L H₃BO₃, 0.288 mg/L ZnSO₄, 0.714 mg/L CoCl₂、0.374 mg/L CuSO₄·5H₂O, 1.583 mg/L MnCl₂. L-arabinose was added to the medium at final concentrations of 10 mM. When needed, antibiotics were supplemented into the medium to the following final concentrations: 100 μg/mL ampicillin.

Culture conditions

All cultivations in this study were performed under aerobic conditions. In the whole-cell biotransformation experiment, the seed culture was first inoculated into LB medium containing ampicillin and cultivated at 37 °C overnight. Subsequently, the seed culture was inoculated at a 1% (v/v) inoculum volume into 250 mL Erlenmeyer flasks containing 100 mL of M9Y medium with antibiotics. When the OD600 of the culture reached 0.4-0.6, L-arabinose was added to a final concentration of 10 mM, and the culture was induced at 16 °C for 24 h. After induction, the cells were collected by centrifugation and washed with PBS buffer to remove residual culture medium. Finally, the cells were resuspended in 10 mL of PBS solution, and KMV was added to a final concentration of 1 g/L, followed by further incubation at 37 °C for 24 h.

For the mono-culture, the seed culture was first cultivated overnight at 37 °C in LB medium with 100 μg/mL ampicillin. After overnight growth, seed cultures were inoculated at a 1% v/v into 250 mL Erlenmeyer flasks containing 25 mL M9Y medium with antibiotics. Fermentation was carried on at 37 °C and 200 rpm. When the OD600 of cultivation reached 0.4–0.6, inducers were added. After 72 h of cultivation, samples were taken for HPLC analysis. When needed, antibiotics were supplemented into the medium to the following final concentrations: 100 μg/mL ampicillin.

For the co-culture, the seed cultures of the upstream and downstream strains were individually cultivated overnight at 37 °C in LB medium supplemented with 100 μg/mL ampicillin. After 12 h of growth, the cell density (OD600) of each strain was measured. Once the OD600 values of the activated upstream and downstream strains were confirmed to be consistent, the initial inoculation volumes for each strain were calculated based on the desired inoculation ratio. The strains were then inoculated into 250 mL Erlenmeyer flasks containing 25 mL of M9Y medium with 100 μg/mL ampicillin at a final inoculation volume of 1% (v/v). Fermentation was conducted at 37 °C with shaking at 200 rpm. When the OD600 of the fermentation broth reached 0.4–0.6, L-arabinose was added to induce expression. After 72 h of cultivation, samples were collected for HPLC analysis.

Strains and plasmid construction

Custom DNA oligonucleotide primers were synthesized by GENEWIZ, Inc. The plasmids and strains used in this study are listed in Table 1. E. coli trans10 (TransGen Biotech Co., Ltd., Beijing, China) was adopted for cloning and plasmid propagation. E. coli BL21 (DE3) (TransGen Biotech Co., Ltd., Beijing, China) was used as a host for metabolic engineering. TransStart FastPFU DNA polymerase (TransGen Biotech Co., Ltd., Beijing, China) was used for PCR amplification of target DNA fragments as well as vector linearization. DNA gel purification, plasmid extraction kits and Gibson Assembly Cloning Kit were purchased from Vazyme (Vazyme Biotech Co., Ltd, Nanjing, China). The Gibson Assembly Cloning Kit is used for fragment assembly to obtain plasmids.

Table 1.

Plasmids and strains used in this study

Plasmids	Description	Source
pE8a	araBAD promoter, AmpR	Novagen
pPanE	pE8a carrying cimA, leuB, leuC, leuD, ilvG, ilvM, panE, AmpR	This study
pKivR	pE8a carrying kivR, AmpR	This study
pLYS	pE8a carrying LYS12, AmpR	This study
pLIKAR	pE8a carrying LIKAR, AmpR	This study
pHMV	pE8a carrying cimA, leuB, leuC, leuD, ilvG, ilvM, panE, AmpR	This study
p2KBup	pE8a carrying cimA, leuB, leuC, leuD, AmpR	This study
p2KBdown	pE8a carrying ilvG, ilvM, panE, AmpR	This study
pKMVup	pE8a carrying cimA, leuB, leuC, leuD, ilvG, ilvM, AmpR	This study
pKMVdown	pE8a carrying panE, AmpR	This study
pEcgRNA	Derived from pTargetF, ccdB	[20]
pEcCas	Derived from pCas, sacB, PrhaB-sgRNA-pMB1, pSC101	[20]
pEcΔptsG	Derived from pEcgRNA, target ptsG in E. coli BL21(DE3)	This study
pEcΔptsI	Derived from pEcgRNA, target ptsI in E. coli BL21(DE3)	This study
pEcΔglk	Derived from pEcgRNA, target glk in E. coli BL21(DE3)	This study
pEcΔxylA	Derived from pEcgRNA, target xylA in E. coli BL21(DE3)	This study
Strains
LY08	E. coli BL21(DE3) F− ompT hsdSB (rB−, mB−) gal dcm (DE3) △poxB△pta△ackA△pflB△tdcE△ilvE	[21]
BL-L	LY08 carrying pHMV	This study
BL-P	LY08 ΔptsG, carrying pHMV	This study
G-BL	LY08 ΔxylA	This study
X-BL	LY08 ΔptsI Δglk	This study
2 KB-G-Up	G-BL carrying p2KBup	This study
2 KB-G-Down	G-BL carrying p2KBdown	This study
2 KB-X-Up	X-BL carrying p2KBup	This study
2 KB-X-Down	X-BL carrying p2KBdown	This study
KMV-G-Up	G-BL carrying pKMVup	This study
KMV-G-Down	G-BL carrying pKMVdown	This study
KMV-X-Up	X-BL carrying pKMVup	This study
KMV-X-Down	X-BL carrying pKMVdown	This study

Deletion of chromosomal genes

The plasmids used to delete chromosomal genes are listed in Table 1. The CRISPR/Cas9 system was used for rapid gene deletion. The plasmid pEcCas was introduced into the parental strain and incubated overnight in the LB medium with kanamycin. 10 mM L-ara was added to the system for λ-Red and pEcΔ was transferred. After recovery culture, the culture medium was seeded onto an LB agar plate containing 50 mg/L of kanamycin and 100 mg/L of spectinomycin. The target gene deletion was confirmed by colony direct PCR. The plasmids pEcCas, and pEcΔ were eliminated from the target-gene-deficient strain. All fragments inserted in the plasmids used to inactivate the respective genes were amplified using direct colony PCR using the E. coli BL21 (DE3) genomic DNA as a template.

Analytical methods

Cell growth was determined by measuring OD600 using a UV spectrophotometer. GC-MS (Trace ISQ, Thermo Fisher Scientific, Waltham, MA, USA) equipped with a TraceGOLD TG-WaxMS column (30 m × 0.25 mm × 0.25 µm, Thermo Fisher Scientific) was used to confirm HMV biosynthesis. High-performance liquid chromatography (Agilent Technologies Inc., California, USA) with the Aminex HPX-87H column (Bio-Rad Inc., CA, USA) was used for substance identification and quantification in the fermentation broths, which was employed and operated at 60 ℃ with 5 mM sulfuric acid as a mobile phase at a flow rate of 0.6 mL min⁻¹. Samples were prepared by centrifuging first at 12,000 rpm for 1 min and the supernatants were filtered with a 0.22 μm film. Glucose and xylose were examined using a refractive index detector. PYR, 2 KB, KMV, HB, HIV and HMV were analyzed by a UV detector (Agilent 1260 Infinity II, Agilent Technologies Inc., Santa Clara, CA, USA) at 215 nm.

Statistical analysis

All statistical analysis was performed using GraphPad Prism V9.5.1. A two-tailed test was used for the comparison of two groups. All experiments were performed with three replicates unless specified, and the error bars in the figure legends represent means ± s.d. When the error line is not visible, it indicates that SD is smaller than the symbol size. A p value less than 0.05 was considered statistically significant (*p < 0.05, **p < 0.01, and ***p < 0.001).

Results

De novo biosynthesis of HMV in E. coli mono-culture

Screening and selection of KARs for HMV Biosynthesis

In the HMV biosynthetic pathway, keto-acid reductases (KARs) catalyzed the conversion of KMV to the target product HMV (Fig. 1). Although various keto-acid reductases (KARs) have been reported, these enzymes displayed considerable variation in substrate specificity and catalytic efficiency. To identify the enzyme best suited for efficient HMV biosynthesis in E. coli, we selected four KARs from different micro-organisms: LYS from Saccharomyces cerevisiae [26], KivR from Beauveria bassiana [10, 27], PanE from Escherichia coli [8, 9], and LIKAR from Leuconostoc lactis [28, 29]. Each enzyme was individually overexpressed in E. coli, and whole-cell biotransformation was performed with 1 g/L 2-keto-3-methylvalerate (KMV) as the substrate. Among them, PanE exhibited the highest catalytic efficiency, producing 0.7 ± 0.0 g/L HMV (Fig. 2), which was further verified by GC–MS analysis (Fig. S1a, b). Therefore, we selected the endogenous 2-dehydropantoate 2-reductase PanE from E. coli (encoded by the panE) for catalyzing the conversion of KMV to HMV.

Fig. 1.

HMV biosynthesis based on the citramalate-derived pathway and glucose–xylose co-utilization in E. coli. Black arrows indicate the native pathways in E. coli; pink, yellow and purple arrows and shading indicate biosynthetic pathways for the generation of HB, HIV and HMV, respectively; X-marks indicate inactivated metabolic pathways. G6P: glucose 6-phosphate; PEP: phosphoenolpyruvate; X5P: xylose 5-phosphate; G3P: 3-phosphoglyceraldehyde; PYR: pyruvate; 2 KB: 2-ketobutyrate; KMV: 2-keto-3-methyl-valerate; KIV: 2-ketoisovalerate; HB: 2-hydroxybutyrate; HIV: 2-hydroxy-3-methylbutyrate; HMV: 2-hydroxy-3-methylvalerate; L-Ile: L-isoleucine; L-Val: L-Valine

Fig. 2.

HMV titer from whole-cell bioconversion by E. coli strains expressing keto-acid reductases from different micro-organisms with 1 g/L KMV supplied

HMV pathway assembly and validation in mono-culture with glucose as single substrate

To establish the complete HMV biosynthetic pathway, we first utilized the previously developed LY08 strain (ΔpoxB, Δpta, ΔackA, ΔpflB, ΔtdcE, and ΔilvE) as the parental strain. This strain has been engineered to block competing metabolic pathways, ensuring an adequate intracellular supply of pyruvate (PYR), 2-ketobutyrate (2KB), and KMV, thereby providing sufficient precursors for HMV synthesis.

Next, we overexpressed citramalate synthase (cimA) from Methanococcus jannaschii, along with the endogenous 3-isopropylmalate dehydratase (leuCD) and 3-isopropylmalate dehydrogenase (leuB). These enzymes catalyzed the conversion of PYR into 2KB, which was then further converted to KMV through the native branched-chain amino acid biosynthetic pathway [16, 30]. To further enhance this process, we overexpressed the rate-limiting enzyme acetolactate synthase II (ilvGM). Finally, 2-dehydropantoate 2-reductase, encoded by panE, was also overexpressed to reduce KMV to HMV, generating the overproduction strain BL-L.

The fermentation performance of strain BL-L was evaluated in a medium containing 20 g/L glucose. Glucose, a preferred carbon source for E. coli, was rapidly depleted between 12 and 24 h (Fig. 3a), accompanied by a pronounced accumulation of PYR. Nevertheless, as the fermentation progressed, PYR was not efficiently channeled into the HMV biosynthetic pathway. There were still 2986.5 ± 55.0 mg/L PYR at the end of the fermentation (Fig. 3b). Concomitantly, the intermediates 2KB and KMV accumulated to maximal titers of 97.9 ± 11.4 (Fig. 3c) and 1518.5 ± 132.1 mg/L (Fig. 3d), respectively. This indicated that, although overexpression of the relevant genes boosts intracellular formation of these keto acids, their subsequent conversion to HMV constituted a clear bottleneck. The final HMV titer reached only 189.1 ± 13.9 mg/L (Fig. 3e).

Fig. 3.

De novo biosynthesis of HMV in mono-culture of E. coli using glucose as the single substrate. a-e Titers of a glucose, b PYR, c 2 KB, d KMV, and e HB, HIV and HMV in mono-culture of BL-L. f Percentages of HB, HIV, and HMV among the three products in mono-culture with glucose as the single substrate

In addition, PanE exhibited broad substrate specificity, quantitatively converted 2 KB and KIV to HB (370.3 ± 44.8 mg/L, Fig. 3e) and HIV (378.3 ± 52.1 mg/L, Fig. 3e), respectively. Consequently, HMV accounted for merely 20% of the total 2-hydroxy acid titer (Fig. 3f). This phenomenon occurs, because when glucose served as the single substrate for HMV biosynthesis, its rapid uptake accelerated PYR synthesis in the cell. The surplus PYR, instead of being promptly redirected toward HMV production, overflowed into the HB and HIV pathways, resulting in higher titers of these by-products than HMV.

Co-utilization of glucose-xylose in mono-culture to enhance HMV titer and percentage

To reduce the biosynthesis of by-products HIV and HB and increase the HMV titer, we planned to adopt a glucose-xylose dual-substrate strategy to avoid the accumulation of intracellular PYR caused by the rapid consumption of substrates, thereby weakening the overflow of the carbon source to the byproduct pathway.

The fermentation performance of strain BL-L was evaluated in a medium containing 10 g/L glucose and 10 g/L xylose. Results showed that glucose was completely depleted within 24 h (Fig. 4a). However, owing to carbon catabolite repression (CCR), xylose utilization was delayed until the glucose was exhausted (Fig. 4b). Halving the initial glucose concentration reduced the peak pyruvate level to 246.7 ± 55.3 mg/L (Fig. 4c). Notably, intracellular 2 KB (35.6 ± 6.5 mg/L, Fig. 4d) and KMV (1283.6 ± 20.4 mg/L, Fig. 4e) still accumulated, indicating that the bottleneck of precursor conversion still existed. At the end of fermentation, BL-L produced 498.3 ± 59.2 mg/L HMV (Fig. 4f), 2.64-fold higher than the titer obtained with glucose alone, raising its percentage of the total 2-hydroxy acid titer from 20.2% to 50.4% (Fig. 4g). Thus, the glucose-xylose dual-substrate strategy delayed rapid carbon exhaustion, alleviated intracellular metabolite imbalance and overflow, and markedly increased both the titer and the percentage of HMV.

Fig. 4.

De novo biosynthesis of HMV in BL-L mono-culture using glucose and xylose as dual substrate. a-f Titers of a glucose, b xylose, c PYR, d 2 KB, e KMV, and f HB, HIV and HMV in mono-culture of BL-L. g Percentages of HB, HIV, and HMV among the three products in mono-culture with glucose and xylose as dual substrate

To relieve carbon catabolite repression (CCR) and promote the synergistic co-utilization of glucose and xylose, we deleted ptsG (encoding the PTS glucose-specific permease) in BL-L, generating strain BL-P. As anticipated, BL-P simultaneously consumed glucose and xylose (Fig. 5a, b). However, the ptsG deletion disrupted intracellular metabolism, causing PYR to accumulate sharply to 1160.4 ± 83.2 mg/L (Fig. 5c) and leading to a surge in the byproduct HB to 583.7 ± 22.5 mg/L (Fig. 5f), while the titer of HMV decreased from 498.3 ± 59.2 mg/L to 186.0 ± 7.8 mg/L (Fig. 5f). Although deleting ptsG effectively lifted CCR and markedly improved glucose–xylose co-utilization, it simultaneously impaired HMV synthesis, shrinking its share of the total 2-hydroxy acids from 50.4% to 16.1% (Fig. 5g). These results underscored that, in mono-culture, micro-organisms faced a dual constraint of CCR and metabolic bottlenecks, keeping HMV biosynthesis at a low level during mixed-sugar fermentation.

Fig. 5.

After alleviating CCR, de novo biosynthesis of HMV in BL-P mono-culture using glucose and xylose as dual substrate. a-f Titers of a glucose, b xylose, c PYR, d 2 KB, e KMV, and f HB, HIV and HMV in mono-culture of BL-P. g Percentages of HB, HIV, and HMV among the three products in mono-culture with glucose and xylose as dual-substrate

Substrate division in E. coli consortia for co-utilization of glucose and xylose

In mono-culture, the glucose-xylose mixture can raise HMV productivity, yet the potential remains largely locked by the twin constraints of CCR and metabolic bottlenecks. Conversely, engineered consortia built on “division of labor” bypasses CCR-driven conflicts at the system level, promising superior co-utilization efficiency and synthetic performance. Guided by this insight, we propose a “functional division with orthogonal complementarity” strategy: the “glucose-only” and “xylose-only” catabolic modules are divided into separate strains, eliminating network chaos and enabling highly efficient, targeted HMV synthesis.

Following this rationale, we implemented to engineer an E. coli-E. coli consortium with explicit division of substrate preference to promote simultaneous co-utilization of glucose and xylose, and improve overall biosynthetic performance. To construct a glucose-specific chassis strain G-BL, we knocked out the xylA gene, which encodes xylose isomerase, an enzyme that isomerizes D-xylose to D-xylulose and serves as a key initial enzyme in the xylose metabolic pathway (Fig. 6a) [31, 32]. Disruption of xylA abolished the capacity of strain G-BL to metabolize xylose, thereby ensuring exclusive glucose utilization.

Fig. 6.

Construction and performance validation of strains with single-substrate utilization preference. a Strain G-BL with the xylA gene knocked out can utilize only glucose in glucose–xylose mixed medium. The strain X-BL with the ptsI and glk genes knocked out can utilize only xylose in glucose–xylose mixed medium. b Growth of strains G-BL and X-BL in glucose-xylose mixed medium. c, d Substrate utilization by strains c X-BL and d G-BL in glucosexylose mixed medium

Conversely, to generate a xylose-specific chassis strain X-BL, we sequentially knocked out genes ptsI and glk (Fig. 6a). The ptsI gene encoded enzyme I of the phosphotransferase system (PTS), a crucial component for glucose uptake and metabolism in E. coli [33]. The deletion of ptsI can block the transport and phosphorylation of glucose via the PTS. In addition, the glk gene encoded glucokinase, which phosphorylates glucose to glucose-6-phosphate, enabling its entry into the glycolytic pathway [34]. The double knockout of ptsI and glk should effectively abolish glucose metabolism in strain X-BL.

To evaluate the substrate specificity, strains X-BL and G-BL were individually cultured in the mixed-sugar medium. The results indicated that strains G-BL and X-BL exhibited distinct substrate preferences while maintaining different growth profiles (Fig. 6b). Specifically, strain X-BL preferentially utilized xylose, which was completely consumed after 24 h, and no glucose utilization was observed (Fig. 6c). In contrast, G-BL rapidly consumed glucose, depleting it within approximately 12 h, and exhibited no detectable xylose utilization throughout the fermentation period (Fig. 6d). These findings confirmed that the engineered strains exhibited orthogonal substrate selectivity, with X-BL and G-BL exclusively metabolizing xylose and glucose, in a mixed-sugar environment, respectively.

Optimization of substrate allocation in E. coli consortia for enhanced HMV synthesis

To further investigate whether the E. coli-E. coli consortium composed of strains G-BL and X-BL could cooperatively synthesize HMV in a glucose–xylose medium, we separated the overall biosynthetic pathway at the 2KB node which served as a key metabolic intermediate. In this design, the upstream module was responsible for converting PYR to 2KB, while the downstream module carried out the conversion of 2KB to KMV and ultimately to HMV. To validate the functional division of substrate metabolism, we constructed two distinct consortia by assigning glucose and xylose utilization to either the upstream or downstream strain, respectively.

Consortium 2KB-X-G: xylose-consuming upstream and glucose-consuming downstream with 2KB as allocation node

The upstream strain X-2KB-Up was constructed by introducing the upstream pathway plasmid into the xylose-specific strain X-BL (Fig. S2a), whereas the downstream strain G-2KB-Down was generated by introducing the downstream pathway into the glucose-specific strain G-BL (Fig. S3a). To assess the cooperative performance of the consortium 2KB-X-G, the strains X-2KB-Up and G-2KB-Down were co-cultured at a 1:1 inoculation ratio in the mixed-sugar medium with 10 g/L glucose and 10 g/L xylose (Fig. 7a).

Fig. 7.

Consortium 2KB-X-G with 2KB as a single node for HMV biosynthesis. a Schematic of consortium 2KB-X-G composed of upstream strain X-2KB-Up and downstream strain G-2KB-Down for HMV production. Upstream strain X-2KB-Up, with poxB, pta, ackA, pflB, tdcE, ilvE, ptsI, and glk deletion and cimA, leuB, leuC, and leuD overexpression, was able to biosynthesize excess 2KB, providing precursors for downstream strain G-2KB-Down to biosynthesize KMV and HMV. Downstream strain G-2KB-Down, with poxB, pta, ackA, pflB, tdcE, ilvE and xylA deletion and ilvGM, panE overexpression, specialized in KMV and HMV biosynthesis. b Titers of the target product HMV and by-products HB and HIV in consortium 2KB-X-G and BL-P mono-culture. c Percentage of the target product HMV and by-products HB and HIV in consortium 2KB-X-G

Despite the division of labor, the xylose utilization efficiency of consortium 2KB-X-G was suboptimal, with 6.4 ± 0.2 g/L xylose remaining after fermentation (Fig. S4b). This inefficiency may be attributed to a growth lag in the upstream strain X-2KB-Up (Fig. S2b). Nevertheless, the downstream strain G-2KB-Down effectively reduced KMV accumulation to 104.0 ± 17.4 mg/L (Fig. S4d), and produced 349.0 ± 20.5 mg/L HMV (Fig. 7b).

However, it should be noted that the by-product HIV reached 1452.1 ± 20.1 mg/L (Fig. 7b), accounting for 80.5% of the total 2-HAs (Fig. 7c). This HIV accumulation likely resulted from disrupted acetate pathways in strain G-BL, leading to elevated intracellular PYR levels. The excessive PYR was redirected to KIV via the enhanced branched-chain amino acid pathway, thereby causing the overproduction of HIV in the consortium (Fig. S3d).

Consortium 2KB-G-X: glucose-consuming upstream and xylose-consuming downstream with 2KB as allocation node

In synthetic consortia, substrate allocation between upstream and downstream strains profoundly influences metabolic flux distribution and product formation. To further investigate the impact of substrate division on metabolic efficiency, we constructed an alternative consortium 2KB-G-X, in which the upstream strain utilized glucose, while the downstream strain utilized xylose.

The upstream strain G-2KB-Up was constructed by introducing the upstream pathway plasmid into the glucose-specific strain G-BL (Fig. S5a), whereas the downstream strain X-2KB-Down was generated by introducing the downstream pathway into the xylose-specific strain X-BL (Fig. S6a). Subsequently, the upstream strain G-2KB-Up and the downstream strain X-2KB-Down were co-cultured in the mixed-sugar medium with 10 g/L glucose and 10 g/L xylose at an initial inoculation ratio of 1:1, which was designated as consortium 2KB-G-X (Fig. 8a).

Fig. 8.

Consortium 2KB-G-X with 2 KB as a single node for HMV biosynthesis. a Schematic of consortium 2KB-G-X composed of two strains G-2KB-Up and X-2KB-Down for HMV production. Upstream strain G-2KB-Up, with poxB, pta, ackA, pflB, tdcE, ilvE and xylA deletion and cimA, leuB, leuC, and leuD overexpression, was able to biosynthesize excess 2KB, providing precursors for downstream strain X-2KB-Down to biosynthesize KMV and HMV. Downstream strain X-2KB-Down, with poxB, pta, ackA, pflB, tdcE, ilvE, ptsI and glk deletion and ilvGM, panE overexpression, specialized in KMV and HMV biosynthesis. b Titers of the target product HMV and by-products HB and HIV in consortium 2KB-G-X and BL-P mono-culture. c Percentage of the target product HMV and by-products HB and HIV in consortium 2KB-G-X

During fermentation, glucose consumption was significantly faster than that of xylose (Fig. S7a, b), which could be explained by the slower growth of strain X-2KB-Down and its limited metabolic capacity for xylose (Fig. S6b, c). This imbalance in carbon source utilization potentially reduced the efficiency of overall pathway progression and product formation. Nevertheless, the overall substrate utilization efficiency of this system was higher than that of the consortium 2KB-X-G, as evidenced by the reduced residual xylose concentration after fermentation (Fig. S7b).

Notably, HMV titer in the co-culture reached 375.7 ± 6.5 mg/L (Fig. 8b), which was twice that of BL-P mono-culture. The percentage of HMV among the total was also improved to 83.9% (Fig. 8c), indicating more efficient carbon flux redirection. However, high KMV accumulation was still observed at the end of fermentation, reaching 621.8 ± 77.2 mg/L (Fig. S7d). These results suggested that assigning the tasks of both KMV synthesis and HMV production solely to the downstream strain may have imposed excessive metabolic burden on strain X-2KB-Down, thereby limiting the overall biosynthesis efficiency.

These results collectively underscored that the assignment of glucose to the upstream strain and xylose to the downstream strain (as in 2KB-G-X) was the more favorable strategy for efficient co-utilization and product biosynthesis. Moreover, while strengthening 2KB conversion to KMV, it also shifted metabolic flux toward the by-product HIV due to PanE poor substrate specificity. Therefore, optimizing both substrate utilization and pathway allocation is essential for maximizing target product biosynthesis in co-culture systems.

Rational pathway allocation enhances HMV titer and selectivity

As demonstrated in the previous sections, the rational division of substrate utilization in a consortium significantly increased the percentage of the HMV among the three 2-hydroxy acids. However, allocating both the conversion of 2KB to KMV and KMV to HMV in the downstream strain faced two key limitations: (1) the poor substrate specificity of PanE led to the undesired conversion of 2KB to the byproduct HB, thereby reducing flux toward KMV and HMV and (2) the co-expression of multiple pathway modules increased the metabolic burden on the host. To address these issues, we redefined the pathway division by selecting KMV as a node. In this design, the upstream module was responsible for converting PYR to KMV, while the downstream module carried out only the conversion of KMV to HMV, thereby minimizing competition for precursors and improving HMV titer and selectivity.

Consortium KMV-G-X: glucose-consuming upstream and xylose-consuming downstream with KMV as allocation node

Based on prior findings that glucose was more suitable for the upstream strain and xylose for the downstream strain, we constructed strain G-KMV-Up by overexpressing the upstream pathway (cimA, leuB, leuC, leuD, ilvG, and ilvM) into the glucose-specific strain G-BL (Fig. S8a), and strain X-KMV-Down by introducing the downstream pathway (panE) into the xylose-specific strain X-BL (Fig. S9a). The consortium KMV-G-X was assembled by co-culturing these two strains at an initial 1:1 inoculation ratio in a mixed-sugar medium (Fig. 9a).

Fig. 9.

Consortium KMV-G-X with KMV as a single node for HMV biosynthesis. a Schematic of consortium KMV-G-X composed of two strains G-KMV-Up and X-KMV-Down for HMV production. Upstream strain G-KMV-Up, with poxB, pta, ackA, pflB, tdcE, ilvE, and xylA deletion and cimA, leuB, leuC, leuD, ilvG, and ilvM overexpression, was able to biosynthesize excess KMV, providing precursors for downstream strain X-KMV-Down to biosynthesize HMV. Downstream strain X-KMV-Down, with poxB, pta, ackA, pflB, tdcE, ilvE, pstI, and glk deletion and panE overexpression, specialized in HMV biosynthesis. b Titers of the target product HMV and by-products HB and HIV in consortium KMV-G-X and BL-P mono-culture. c Percentages of the target product HMV and by-products HB and HIV in consortium KMV-G-X

Substrate utilization analysis showed that glucose and xylose were completely consumed within 72 h (Fig. S10a, b). During the first 24 h, corresponding to rapid glucose consumption, the production of intermediates 2KB and KMV markedly increased (Fig. S10c, d). After glucose depletion, the rates of 2KB and KMV synthesis dropped below the rate of KMV conversion to HMV, resulting in a decline in intermediate accumulation. At the end of fermentation, the KMV concentration was only 50.7 ± 7.9 mg/L (Fig. S10d), significantly lower than that observed in consortium 2KB-G-X.

In terms of HMV production, the sufficient KMV supply by the upstream strain G-KMV-Up ensured a sustained HMV biosynthesis throughout fermentation (Fig. S8e). The final HMV titer reached 2184.6 ± 111.8 mg/L (Fig. 9b), the highest among all consortia tested. Moreover, the percentage of HMV was as high as 82.2% (Fig. 9c). These results demonstrated that the consortium KMV-G-X with KMV as the key node effectively channeled metabolic flux toward the desired product, resulting in highly efficient HMV biosynthesis.

Consortium KMV-X-G: xylose-consuming upstream and glucose-consuming downstream with KMV as allocation node

To further evaluate whether the consortium KMV-G-X was the best for HMV biosynthesis, we also constructed the consortium KMV-X-G by reversing the substrate preferences of the two strains (Fig. 10a). However, a lower substrate utilization was observed compared with the consortium KMV-G-X, with 3.5 ± 0.4 g/L of xylose remaining accumulated (Fig. S14b). This situation of xylose remaining and glucose depletion was attributed to the fast substrate utilization rate of the downstream strain G-KMV-Down (Fig. S12c) and the slow substrate utilization rate of the upstream strain X-KMV-Up (Fig. S11c). HMV titer reached only 636.4 ± 7.9 mg/L (Fig. 10b), with HMV percentage just 39.2% of the total 2-hydroxy acids (Fig. 10c). This was primarily attributed to suboptimal use of xylose by the upstream strain. Since xylose was not the optimal substrate for 2KB and KMV biosynthesis, the upstream strain X-KMV-Up produced only 709.6 ± 5.7 mg/L of KMV during mono-culture (Fig. S11e). However, compared to strain X-2KB-Up, the KMV titer in strain X-KMV-Up mono-culture increased by 44.0%. This suggested that using KMV as a pathway allocation node enhanced KMV biosynthesis in the upstream strain, thereby improving HMV biosynthesis in the consortium.

Fig. 10.

Consortium KMV-X-G with KMV as single node for HMV biosynthesis. a Schematic of consortium KMV-X-G composed of two strains X-KMV-Up and G-KMV-Down for HMV production. Upstream strain X-KMV-Up, with poxB, pta, ackA, pflB, tdcE, ilvE, ptsI, and glk deletion and cimA, leuB, leuC, leuD, ilvG, and ilvM overexpression, was able to biosynthesize excess KMV, providing precursors for downstream strain G-KMV-Down to biosynthesize HMV. Downstream strain X-KMV-Down, with poxB, pta, ackA, pflB, tdcE, ilvE, and xylA deletion and panE overexpression, specialized in HMV biosynthesis. b Titers of the target product HMV and by-products HB and HIV in consortium KMV-X-G and BL-P mono-culture. c Percentages of the target product HMV and by-products HB and HIV in consortium KMV-X-G

Carbon source and inoculation optimization in consortium KMV-G-X for robust HMV production

Modulating the glucose-to-xylose ratio and the initial inoculation ratio is a key strategy for improving the biosynthetic efficiency of synthetic consortia. Adjusting the glucose-to-xylose ratio helps balance the metabolic rates between different strains, ensuring that each strain meets its specific metabolic needs. This balance promotes cooperative interactions, minimizes the accumulation of undesired by-products, and enhances both the stability and overall productivity of the system. In addition, fine-tuning the initial inoculation ratio could directly regulate the synergistic relationships among strains, thereby improving the adaptability and robustness of the consortium in fluctuating environments.

Comparative analysis of four consortia (2KB-G-X, 2KB-X-G, KMV-G-X, and KMV-X-G) revealed that the consortium KMV-G-X exhibited the highest synthetic capability under the same culture conditions. The total titer of three 2-hydroxy acids (HB, HIV, and HMV) reached 2659.3 mg/L, with HMV achieving the highest titer of 2184.6 ± 111.8 mg/L, representing 82.2% of the total products (Fig. 11a). Therefore, this consortium KMV-G-X was selected for further optimization to maximize HMV production.

Fig. 11.

Comparative synthesis capabilities of the four synthetic consortia and modulation of the ratios of glucose:xylose and initial inoculum in KMV-G-X. a Total titers of HB, HIV, and HMV as well as their respective percentages in the four consortia. b Titers of HMV in the consortium KMV-G-X under different glucose-to-xylose ratios in the medium and different initial inoculation ratios

To systematically assess the effects of carbon source composition and inoculation strategy, five glucose-to-xylose mass ratios (4:1, 3:2, 1:1, 2:3, and 1:4, g/g) and seven upstream-to-downstream inoculation ratios (9:1, 6:1, 3:1, 1:1, 1:3, 1:6, and 1:9, v/v) were tested in an orthogonal design. The results demonstrated that these parameters significantly influenced the biosynthesis of key intermediates (2KB and KMV), which in turn affected the final HMV titers.

Notably, under a glucose-to-xylose ratio of 1:1 and an upstream-to-downstream inoculation ratio of 1:1, the consortium achieved its highest HMV titer of 2184.6 ± 111.8 mg/L, which was 11.7 times higher than that of the BL-P monoculture (Fig. 11b). Under these optimal conditions, the concentrations of 2 KB and KMV were low, at 44.5 ± 11.9 mg/L and 126.0 ± 20.3 mg/L, respectively, indicating that most of the metabolic flux was directed toward HMV synthesis (Fig. S14a, b). Furthermore, under a fixed glucose-to-xylose ratio, varying the initial inoculation ratio from 1:1 to 1:9 led to minimal fluctuation in HMV production. This observation highlighted the intrinsic robustness of the KMV-G-X consortium in maintaining high biosynthetic performance across a broad range of culture conditions.

Discussion

In this study, we engineered a synthetic biology platform to enable the efficient biosynthesis of 2-hydroxy-3-methylvalerate (HMV) through the co-utilization of glucose and xylose. Compared with glucose-only fermentation, the introduction of xylose into the substrate mixture significantly improved the balance between substrate uptake and product formation. This co-substrate strategy mitigated the carbon overflow typically observed at the pyruvate node, a common bottleneck when glucose is rapidly consumed. By alleviating this metabolic congestion, we effectively reduced the synthesis of by-products, such as 2-hydroxybutyrate (HB) and 2-hydroxyisovalerate (HIV), while simultaneously enhancing HMV titers and its relative proportion among the 2-hydroxyacid products. We further proposed that the strategy using two substrates with distinct uptake rates as co-substrates to balance substrate utilization with biosynthetic flux, was broadly applicable to any product whose biosynthetic pathway is directly coupled to the central carbon flux initiated by glucose and xylose.

Furthermore, recognizing that simultaneous utilization of mixed sugars by a single strain often leads to trade-offs between growth and production, we designed a microbial consortium based on the principle of division of labor. In this system, one strain was engineered to specialize in glucose metabolism, while the other was optimized for xylose utilization. The biosynthetic pathway for HMV was strategically split and reconstituted across the two strains to maximize metabolic efficiency and minimize pathway interference. This modular approach not only reduced the metabolic burden on individual strains but also improved overall system robustness and productivity.

The optimal consortium, designated KMV-G-X, achieved an HMV titer of 2184.6 ± 111.8 mg/L, 11.8-fold higher than that of the mono-culture control (BL-P) under the same glucose-xylose co-utilization conditions. Importantly, this increase in HMV production was accompanied by a significant reduction in the relative abundance of undesired by-products, highlighting the effectiveness of both substrate and pathway partitioning. This titer is still markedly higher than the maximum 2-HIV titer (≈1.75 g/L) previously reported for Klebsiella pneumoniae in shake–flask cultures, in which the biosynthetic route closely resembles the HMV pathway described here [35]. Consequently, the consortium constructed in this study demonstrates superior efficiency.

Beyond improving product yields, this work demonstrates that multi-substrate strategies can expand the range of renewable feedstocks usable in microbial bioprocesses. By diversifying the carbon source portfolio, we not only enhance the sustainability of the bioprocess but also introduce metabolic flexibility that can be leveraged to fine-tune the kinetics of substrate consumption and product synthesis. This is particularly relevant for industrial applications, where substrate availability and cost are critical factors.

Moreover, synthetic microbial consortia offer a powerful alternative to traditional monocultures, especially in the context of mixed-sugar utilization. They provide a scalable and tunable framework for overcoming intrinsic limitations of single-strain systems, such as metabolic burden, regulatory conflicts, and inefficient co-substrate uptake. Our findings underscore the potential of engineered microbial communities as a versatile platform for the production of high-value chemicals.

In conclusion, this study establishes a foundation for the scalable and economically viable production of HMV and related compounds through the rational design of microbial consortia and the strategic utilization of mixed substrates. It also contributes to the broader goal of developing greener, more sustainable biomanufacturing processes by aligning metabolic engineering with principles of systems biology and ecological cooperation.

Abbreviations

PYR

Pyruvate

2KB

2-Ketobutyrate

KMV

2-Keto-3-methylvalerate

HB

2-Hydroxybutyrate

HIV

2-Hydroxyisovalerate

HMV

2-Hydroxy-3-methylvalerate

CCR

Carbon catabolite repression

CGA

Chlorogenic acid

G6P

Glucose 6-phosphate

PEP

Phosphoenolpyruvate

X5P

Xylulose 5-phosphate

G3P

3-Phosphoglyceraldehyde

KIV

2-Ketoisovalerate

L-Ile

L-isoleucine

L-Val

L-Valine

Author contributions

Yu Liu performed the experiments, collected and analyzed the data, and wrote the manuscript with input from all authors. Shaojie Wang conceived and designed the study, contributed to the writing and editing of the manuscript. Haijia Su conceptualized the study, conceived and designed the research, and edited the manuscript.

Funding

The National Natural Science Foundation of China, 22308018, 22578021, Beijing Natural Science Foundation, 2252017, Fundamental Research Funds for the Central Universities, buctrc202341, JD2537

Data availability

All data supporting the findings of this study are available within the article and its Supplementary Information file or available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.