Development of an ethanol-based microbial platform for 3-hydroxypropionic acid production using engineered Pseudomonas putida KT2440

Abstract

Background

Ethanol is an attractive C2 feedstock for microbial biomanufacturing because it is directly oxidized to acetyl-CoA with favorable redox balance. 3-Hydroxypropionic acid (3-HP), a versatile platform chemical, can be synthesized via the malonyl-CoA and β-alanine pathways. An inducer-free ethanol-to-3-HP process in Pseudomonas putida KT2440 was developed by enabling constitutive expression of both pathways and deleting native 3-HP catabolism and polyhydroxyalkanoate (PHA) synthesis to redirect flux toward the target product. Each route and their co-activation were evaluated, and a genome-scale model constrained with transcriptomic data was applied to identify metabolic nodes governing pathway choice and performance.

Results

Shake-flask experiments with 1% (v/v) ethanol showed that the malonyl-CoA pathway yielded higher 3-HP titers than the β-alanine route. Co-activation of both pathways in a PHA-deficient strain improved production to 15.9 mM (1.42 g/L; 179 mg/g ethanol) while reducing acetate overflow. In a fed-batch with continuous ethanol feeding, the engineered strain reached 43.7 mM (3.92 g/L; 154 mg/g ethanol) 3-HP. Deletion of endogenous 3-HP catabolism and PHA synthesis redirected carbon flux toward the product, and additional acetyl-CoA supply was achieved by introducing a heterologous acetaldehyde dehydrogenase. Genome-scale modeling constrained with transcriptomic data revealed dominant flux routing through the glyoxylate shunt and limited oxaloacetate regeneration, explaining the advantage of the malonyl-CoA pathway and the acetate accumulation associated with β-alanine operation.

Conclusions

An inducer-free ethanol-to-3-HP platform was established in P. putida KT2440 by co-activating the malonyl-CoA and β-alanine pathways while eliminating competing sinks such as 3-HP catabolism and PHA synthesis. This approach enhanced yield, reduced acetate overflow, and systems-level analysis identified the glyoxylate shunt and oxaloacetate limitation as key control points. Overall, the study demonstrates an efficient and scalable route for 3-HP production from ethanol.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13036-025-00595-9.

Background

The growing dependence on fossil fuels has heightened concerns over greenhouse gas emissions and climate change, thereby driving the advancement of biorefineries as sustainable platforms for the production of higher-value chemicals [1]. Among available carbon feedstocks, ethanol has emerged as a versatile substrate for microbial bioconversion due to its broad availability and favorable intracellular metabolism. Ethanol also benefits from a global production infrastructure and can be manufactured from renewable feedstocks, making it relevant for sustainable bioprocessing [2]. The United States is the leading global producer of ethanol, operating over 200 commercial biorefineries with an installed annual capacity of approximately 60.64 billion liters. Although ethanol is often more expensive than glucose, advances in lignocellulosic biomass pretreatment and enzymatic hydrolysis have reduced its production cost [3].

However, most of the industrial ethanol still originates from sugar- and starch-based fermentation, and this upstream step has important implications for carbon efficiency and process economics [2, 3]. In conventional yeast fermentation, only about 2/3 of the carbon from glucose is retained in ethanol, while roughly 1/3 is released as carbon dioxide (CO₂) before the bioconversion process even begins [4]. This upstream carbon loss is critical because, although ethanol can support higher intracellular yields for acetyl-CoA–derived products, these yields are achieved only after carbon has already been lost in the sugar-to-ethanol conversion step. As a result, direct glucose utilization may provide higher overall carbon retention across the full process and can offer a more favorable feedstock cost structure [2–4]. Additional routes such as catalytic synthesis from syngas or gas-fermentation using acetogenic microorganisms are emerging, although they currently represent only a minor fraction of the global ethanol market [5–8]. Thus, the sustainability and economic relevance of ethanol depends strongly on its production route, scale, and integration with waste- or residue-based feedstocks.

Beyond feedstock considerations, ethanol offers clear metabolic advantages. Its conversion to acetyl-CoA proceeds through fewer enzymatic steps than glucose and generates two molecules of NADH per molecule oxidized, favoring biosynthesis of reduced compounds [9]. Specifically, for polyhydroxybutyrate (PHB) production, the calculated theoretical yield on ethanol is 0.935 g/g, surpassing that of acetate (0.510 g/g), glucose (0.478 g/g), and glycerol (0.467 g/g) [10]. Moreover, incorporating an acetaldehyde dehydrogenase from Dickeya zeae enables the direct conversion of acetaldehyde to acetyl-CoA, a strategy anticipated to enhance overall product yields [11]. However, these stoichiometric yields describe only the fermentation stage and not the upstream sugar-to-ethanol conversion; hence, ethanol’s metabolic benefits must be evaluated together with its upstream carbon and cost penalties.

3-Hydroxypropionic acid (3-HP) is a U.S. Department of Energy–designated biobased platform chemical derived from acetyl-CoA and serves as a precursor for acrylic acid, 1,3-propanediol, methyl acrylate, acrylamide and poly(3-hydroxypropionate). Extensive efforts have focused on engineering microbial hosts for 3-HP production [12]. Reported biosynthetic strategies include routes from glycerol or lactate as well as pathways based on malonyl-CoA or β-alanine. For instance, an engineered Escherichia coli strain using glycerol after an initial glucose phase produced 71.9 g/L 3-HP with a volumetric productivity of 1.8 g/L/h, although the overall yield was not reported [13]. This glycerol-based strategy depends on a B12-dependent glycerol dehydratase, necessitating external vitamin B12 supplementation and reducing commercial feasibility [14]. B12-independent pathways have therefore gained interest. In the malonyl-CoA pathway, acetyl-CoA is converted to malonyl-CoA and then reduced to malonate semialdehyde before forming 3-HP [15, 16]. The β-alanine pathway also proceeds through malonate semialdehyde, linking β-alanine metabolism to tricarboxylic acid cycle (TCA) intermediates [17]. The preferred pathway depends on both the host and the carbon source; for example, the β-alanine route performs well with xylose, while the malonyl-CoA route is more efficient on glucose [17, 18].

Given these considerations, Pseudomonas putida KT2440 is an attractive host for ethanol-based 3-HP production. Its pyrroloquinoline quinone (PQQ)-dependent alcohol dehydrogenase system rapidly channels ethanol into acetyl-CoA [19, 20], and the strain has been engineered to produce PHB [21], medium-chain-length polyhydroxyalkanoates (mcl-PHAs) [22], mevalonate [23], and fatty acid ethyl esters [11], using ethanol as the sole carbon source. An additional advantage is the endogenous mcl-PHA biosynthetic pathway, which can be rerouted to increase flux through the acetyl-CoA and malonyl-CoA nodes [21, 22], Recent work has also demonstrated 3-HP formation in P. putida from mixed sugars in fed-batch mode; however, those studies relied on inducible expression systems and implemented only the malonyl-CoA route [24].

In the present study, P. putida KT2440 was modified to synthesize 3-HP acid from ethanol by concurrent activation of the malonyl-CoA and β-alanine pathways, and the two routes were compared to assess their individual and combined impact on 3-HP production (Fig. 1). To improve process economy, all heterologous genes were placed under constitutive promoters, eliminating inducer requirements. Native functions involved in 3-HP degradation and polyhydroxyalkanoate formation were deleted to channel carbon flux toward the target product, and a heterologous acetaldehyde dehydrogenase was introduced to enhance formation of acetyl-CoA from ethanol. A genome-scale metabolic model (GSMM) tailored to ethanol assimilation and constrained with transcriptomic data was used to analyze intracellular flux distributions, diagnose bottlenecks and support rational optimization of the strain and process.

Fig. 1.

Engineering scheme enabling 3-HP synthesis from ethanol in P. putida KT2440. Native metabolic reactions are indicated in purple, while steps introduced or overexpressed for this study are shown in red; gene deletions are marked with crosses. Key enzymes include acetyl-CoA carboxylase subunits (accBC/dtsR1), malonyl-CoA reductase (mcr), methylmalonyl-semialdehyde dehydrogenase (mms), acetaldehyde dehydrogenase (ada), aspartate decarboxylase (panD), PHA synthases (phaC1/phaC2), PHA depolymerase (phaZ). Abbreviations: 3-HP, 3-hydroxypropionic acid; PHA, polyhydroxyalkanoate

Methods

Bacterial strains and cultivation

Table S1 and Table S2 (see Supporting Information 1) provides a detailed overview of bacterial strains and plasmids employed in this study. E. coli DH5α served as the cloning host, while P. putida KT2440 was utilized for 3-HP production. Electroporation of P. putida KT2440 was carried out with a Bio-Rad MicroPulser, following the methodology previously outlined by Luo et al. (2016) [25]. LB medium supported cloning, electroporation, and seed culture preparation, while 3-HP production was carried out in 500-ml flasks containing 50 ml of M9 minimal medium [11]. Ethanol at 1% (v/v) served as the sole carbon and energy source for P. putida strains in shake flasks.

Fed-batch fermentation was performed in a 5-L fermenter containing 1.8 L of culture broth, using M9 minimal medium as the basal medium. Inoculation was performed using seed cultures prepared from single colonies, which were grown overnight in LB medium. The fermentation pH was maintained using phosphoric acid (H₃PO₄) and ammonium hydroxide (NH₄OH) as acid and base, respectively. The bioreactor was aerated at 2 vvm, and DO levels were regulated at 10–20% via automated control of impeller agitation, which varied from 300 to 700 rpm. At the start of fermentation, 1% (v/v) ethanol was added to the culture, corresponding to 18 mL. Ethanol feeding was initiated at 30 h and continued at a constant rate until the end of the cultivation at 96 h. The feed consisted of pure ethanol and supplied a total of 40 mL over the course of the process at an average rate of 0.61 ml/h. Samples were withdrawn at regular intervals for metabolite analysis.

Plasmid construction and genome editing

Macrogen (Seoul, Korea) manufactured primers listed in Table S3 (see Supporting Information 1). Malonyl-CoA reductase (mcr) gene was PCR-amplified from the genomic DNA of Chloroflexus aurantiacus and partitioned into functional domains as outlined by Liu et al. (2013) [26]. From Corynebacterium glutamicum ATCC 13,032, the acc gene coding acetyl-CoA carboxylase and the panD gene coding for aspartate decarboxylase were PCR-amplified. Genes encoding 3-hydroxypropionate dehydrogenase and β-alanine transaminase (pp0656) were amplified from E. coli and P. putida, respectively. Additionally, a codon-optimized version of the ada gene from D. zeae was produced by Integrated DNA Technologies (Daejeon, Korea). PCR amplification was performed using LAMP Pfu DNA polymerase (BioFACT, Korea), and plasmids were assembled via Gibson assembly (New England Biolabs, USA).

Gene deletions in P. putida KT2440 were generated using the suicide plasmid pCM433 with 0.8–1.0 kb homology arms and the sacB marker. Constructs were introduced by electroporation, and kanamycin-resistant transformants were isolated on LB agar supplemented with kanamycin (50 µg/mL). Transformants were subsequently subjected to sucrose counter-selection (25% w/v) to obtain recombinants that had resolved the integrated plasmid through double-crossover recombination. The resulting colonies were analyzed by colony PCR, and those carrying the expected chromosomal modification were designated as deletion mutants.

Reconstruction of GSMM of P. putida KT2440 growth on ethanol and integration with transcriptomic data

The P. putida KT2440 GSMM iJN1463 [27] was obtained from the BiGG Models repository [28]. Simulations were run in the COBRA environment (COBRA Toolbox and COBRApy) [29, 30], after adapting exchange and transport bounds to represent M9 minimal medium with ethanol as the sole carbon substrate. RNA-seq data [21] were incorporated to generate condition-specific flux distributions using an expression-constrained workflow following the iStrat procedure previously applied to Synechococcus sp. PCC7002 [31]. In the model, ethanol oxidation through the PQQ-dependent alcohol dehydrogenase pathway was activated, with the ethanol uptake rate constrained to 9.54 mmol g/DCW/h, while glucose uptake was restricted to zero. Downstream, acetaldehyde oxidation was channeled through the acetaldehyde dehydrogenase reactions ALDD2x and ALDD2y, with the alternative reaction ACALD disabled by setting its bounds to zero. These GSMM reaction identifiers correspond to the acetaldehyde dehydrogenase (ada) step depicted in the pathway schematic. The resulting ethanol-configured model is referred to as iJN1463-E.

Analytical methods

Samples were taken at defined intervals. OD₆₀₀ was measured with a UV–visible spectrophotometer. Metabolite concentrations in the culture supernatants were determined by HPLC (Aminex HPX-87 H, 5 mM H₂SO₄, 65 °C, 0.6 mL/min) equipped with a refractive index detector (Agilent 1260).

Results

Knocking out 3-HP degradation pathway in P. putida KT2440

The genetic determinants responsible for 3-HP catabolism in P. putida KT2440 were investigated based on prior work by Hanko et al. (2017) [32]. The mmsA (PP_4667) and mmsB (PP_4666) genes form a two-gene operon encoding methylmalonyl-semialdehyde dehydrogenase and 3-hydroxyisobutyrate dehydrogenase, respectively. In contrast, 3-hpdh (PP_0056) is located in a separate operon and encodes a choline/3-hydroxypropionate dehydrogenase. A third homologous gene, PP_0488, is annotated as a NAD-dependent dehydrogenase belonging to the medium-chain dehydrogenase/reductase family and shares sequence similarity with known 3-HP dehydrogenases [32, 33]. In Paracoccus denitrificans, 3-HP exposure was shown to activate transcription of catabolic genes [34], suggesting a broad regulatory network. The genome of P. putida KT2440 encodes four annotated dehydrogenases with potential relevance to 3-HP conversion: mmsA (PP_4667), mmsB (PP_4666), PP 0488, and 3-hpdh (PP_0056). These four loci were therefore selected as deletion targets to systematically evaluate their contributions to 3-HP degradation in P. putida KT2440 (Fig. 2A).

Fig. 2.

Disruption of the 3-HP degradation pathway in P. putida KT2440. (A) Schematic representation of gene deletions constructed in this study, including mmsA/mmsB (mms operon), PP_0488, and 3-hpdh (PP_0056). (B) 3-HP degradation profiles of whole-cell suspensions from mutant P. putida strains. NC, negative control (medium without cells); KT-WT; KT-Δmms; KT-ΔmmsΔ488; KT-ΔmmsΔ488Δ56

Elimination of mmsA and mmsB in the single-deletion strain WT-Δmms did not abolish 3-HP degradation, and the degradation pattern closely resembled that of the wild type P. putida KT2440 (KT-WT) (Fig. 2B). Introducing an additional deletion of the PP_0488 gene produced the double-deletion strain WT-ΔmmsΔ488, which exhibited a slower 3-HP degradation rate relative to WT and WT-Δmms, although approximately 10 mM 3-HP was still consumed within 96 h. Finally, deletion of the third homolog, 3-hpdh (PP 0056), in this background generated the triple-deletion strain WT-ΔmmsΔ488Δ56, which completely lost the ability to degrade 3-HP for at least 144 h. Because this strain was entirely unable to catabolize 3-HP, WT-ΔmmsΔ488Δ56 was selected as the chassis for subsequent ethanol-based 3-HP biosynthesis.

3-HP production from ethanol in recombinant P. putida expressing the β-alanine and malonyl-CoA routes

The malonyl-CoA pathway was enabled in P. putida by introducing mcr from C. aurantiacus, encoding malonyl-CoA reductase [35]. The recombinant strain KT-mcr, harboring bb-mcr plasmid (pBBR1-MCS2 derived expression plasmid), produced 2.7 mM 3-HP after 96 h on 1% (v/v) ethanol, corresponding to a yield of 31 mg 3HP/g ethanol (Fig. 3A). In parallel, the β-alanine pathway was established by expressing panD (aspartate decarboxylase), panM (activator of panD), ydfG (3-hydroxy acid dehydrogenase), and pp0596 (malonate semialdehyde dehydrogenase) from the plasmid 89-pyp9 (a pAWP89-derived expression plasmid) [36, 37]. The resulting strain KT-pyp accumulated 1.6 mM 3-HP (Fig. 3B) together with 88 mM acetate, indicating strong overflow metabolism. Under identical conditions, the malonyl-CoA route yielded 1.66-fold more 3-HP than the β-alanine route.

Fig. 3.

3-HP production from the recombinant strains of P. putida. 3-HP production from (A) KT-mcr, (B) KT-pyp, (C) KT-macc, (D) KT-pada, (E) KT-macc-pada. Red, green and blue lines indicate OD_600, acetate and 3-HP, respectively

To address the acetyl-CoA-to-malonyl-CoA bottleneck, accBC and dtsR1 (also referred to as accD1) from C. glutamicum were co-expressed, providing the biotin carboxylase, biotin-carboxyl-carrier protein, and carboxyltransferase β-subunit of acetyl-CoA carboxylase (ACC). The corresponding strain KT-macc reached 4.83 mM (436 mg/L) 3-HP from ethanol (Fig. 3C), a 1.8-fold increase compared with KT-mcr. KT-macc also exhibited higher OD₆₀₀ values and accumulated more acetate probably due to the enhanced intracellular carbon redistribution induced by increased malonyl-CoA demand. Prior studies have shown that elevating malonyl-CoA synthesis through ACC overexpression increases flux through malonyl-CoA-dependent pathways and can redistribute acetyl-CoA between central metabolism and product formation, thereby influencing both biomass and by-product profiles [35, 38].

For the β-alanine route, acetyl-CoA flux was enhanced by introducing ada from D. zeae, encoding acetaldehyde dehydrogenase, into 89-pyp9 to generate 89-pyp9-ada. The resulting strain KT-pada produced 2.32 mM (209 mg/L) 3-HP (Fig. 3D), outperforming KT-pyp but remaining about twofold below KT-macc. KT-pada accumulated 88 mM acetate but achieved a final OD₆₀₀ of 4.3, higher than KT-pyp, KT-mcr, and KT-macc, suggesting that ada expression primarily diverted flux toward biomass rather than 3-HP. Finally, to test simultaneous pathway activation, bb-mcr-acc and 89-pyp9-ada were co-introduced into KT-TM to generate KT-macc-pada. Cultivation on 1% (v/v) ethanol resulted in 11.5 mM 3-HP (1.02 g/L; Fig. 3E) with a yield of 129 mg/g ethanol. Acetate transiently accumulated to 22 mM at 48 h but decreased to near-background by 72 h. Co-activation of both pathways therefore enhanced 3-HP titers, alleviated metabolic bottlenecks, and reduced acetate overflow compared with single-pathway strains. In KT-macc-pada, acetate accumulated to approximately 20 mM at 48 h and was almost completely depleted between 48 and 72 h. During the same period, 3-HP increased from about 7–8 mM to more than 10 mM, whereas OD₆₀₀ slightly decreased, indicating that acetate was re-assimilated mainly as a carbon source for 3-HP formation rather than for further biomass growth.

Reconstruction of GSMM and multi-omics-based hybrid flux balance analysis

GSMMs provide a structured, mechanistic representation of cellular metabolism and enable hypothesis-driven analysis of flux redistribution under defined genetic or environmental perturbations. Although only a limited number of reactions required modification to configure P. putida KT2440 for ethanol utilization, integration of transcriptomic data allowed the ethanol-specific model iJN1463-E to generate physiologically informed flux distributions that cannot be extracted from pathway maps alone. The curated KT2440 model iJN1463 reported by Nogales et al. (2020) [27], served as the foundation; ethanol uptake, PQQ-dependent alcohol dehydrogenase activity, and ethanol-specific exchange boundaries were adjusted to represent growth in M9–ethanol medium. RNA-seq datasets obtained from KT-WT cultivated in 1% (v/v) ethanol (Supporting Information 2) were incorporated using regularized multi-level FBA [39], with biomass synthesis as the primary objective and ATP maintenance as a secondary constraint.

Figure 4 presents the resulting flux distribution for ethanol-grown KT-WT. A dominant fraction of carbon flux entered the TCA cycle through the glyoxylate shunt, where isocitrate lyase converted isocitrate into glyoxylate and succinate at 7.86 mmol g/DCW/h. Succinate was subsequently metabolized to fumarate and then to malate. In contrast, flux through malate synthase was markedly lower (1.57 mmol g/DCW/h). When combined with the fumarate-to-malate step, total flux through malate dehydrogenase reached 7.48 mmol g/DCW/h. Downstream, conversion of malate and oxaloacetate toward pyruvate or phosphoenolpyruvate (PEP) proceeded at relatively minor rates, with the malic enzyme operating at 0.51 mmol g/DCW/h. By comparison, gluconeogenic flux from PEP to 1,3-phosphoglycerate was elevated, reaching 2.03 mmol g/DCW/h.

Fig. 4.

Regularized flux distribution for P. putida KT2440 cultivated on 1% (v/v) ethanol, generated by integrating transcriptomic data into the ethanol-adapted GSMM iJN1463-E. Arrow widths represent relative flux magnitudes, and black arrows indicate negligible flux

This raised the question of how carbon flux from glyoxylate is distributed. Flux analysis revealed that glyoxylate was directed toward tartronate semialdehyde and subsequently to glycerate via glyoxylate carboxyligase and tartronate semialdehyde reductase, respectively, with a flux of 3.14 mmol g/DCW/h. Glycerate was then phosphorylated to 3-phosphoglycerate, which was converted to 2-phosphoglycerate through glycolysis and further to 1,3-bisphosphoglycerate along the gluconeogenic route. This flux distribution aligns with previous reports describing glyoxylate-derived glycerate as a key intermediate during ethylene glycol metabolism in P. putida KT2440 [40, 41]. Collectively, these findings suggest that glycerate derived from glyoxylate represents the major entry point for gluconeogenesis during ethanol-based growth of P. putida KT2440.

The model also revealed metabolic constraints that are directly relevant to 3-HP biosynthesis. The dominant flux through the glyoxylate shunt together with the limited activity of malate synthase identified a bottleneck in oxaloacetate regeneration which restricts the β-alanine pathway. In contrast the malonyl-CoA pathway bypasses this limitation. The model predicted substantial diversion of acetyl-CoA and malonyl-CoA into PHA synthesis which identifies the PHA cycle as a major competing sink for these precursors under ethanol growth conditions. These model-derived predictions informed the subsequent engineering strategy by indicating (i) the malonyl-CoA pathway as the more favorable production route, (ii) the need to delete PHA synthase genes to mitigate the predicted acetyl-CoA drain, and (iii) the benefit of introducing accBC/dtsR1 and ada to overcome model-identified limitations in malonyl-CoA and acetyl-CoA supply. Experimental validation of these predictions confirms that the model provides qualitative yet mechanistically informative guidance for rational strain engineering.

Enhancing the 3-HP production by knocking out PHA synthase

P. putida KT2440 naturally accumulates mcl-PHAs as carbon and energy reserves. PHA polymerization is catalyzed by the synthases phaC1 and phaC2, which convert (R)-3-hydroxyacyl-CoA intermediates into polymer chains, while depolymerization is mediated by the intracellular depolymerase phaZ, which hydrolyzes PHA granules to release monomers for carbon and energy mobilization. Because malonyl-CoA is a critical precursor for both fatty acid and PHA biosynthesis, disruption of PHA formation was expected to increase malonyl-CoA availability for 3-HP production. To test this, phaC1, phaZ, and phaC2 were deleted in the WT-ΔmmsΔ488Δ56 background to generate the mutant strain KT-TM-ΔCZ. Introduction of bb-mcr-acc and 89-pyp9-ada into KT-TM-ΔCZ yielded TM-ΔCZ-macc and TM-ΔCZ-pada, respectively. In 1% (v/v) ethanol, TM-ΔCZ-pada achieved a higher final OD₆₀₀ (4.5) than KT-pada, but no increase in 3-HP production was observed (Fig. 5A). Acetate levels were similar to those in KT-pada and KT-pyp, indicating that the additional carbon flux from PHA deletion was redirected toward biomass rather than entering the β-alanine pathway.

Fig. 5.

3-HP production in the PHA deficient strain of P. putida KT2440. 3-HP production from (A) TM-ΔCZ-pada, (B) TM-ΔCZ-macc, (C) TM-ΔCZ-macc-pada. Red, green and blue lines indicate OD_600, acetate and 3-HP, respectively

By contrast, TM-ΔCZ-macc accumulated 10.31 mM (928.48 mg/L) 3-HP (Fig. 5B), a 2.13-fold increase over KT-macc. In the TM-ΔCZ-macc strain, acetate reached approximately 20–22 mM by 48 h and was almost completely depleted by 72 h. During the same interval, 3-HP increased from about 6 mM to more than 10 mM, while OD₆₀₀ showed a slight decrease. These data indicate that acetate re-assimilation contributed directly to 3-HP formation during the late phase rather than supporting further biomass growth. Finally, co-activation of both pathways in TM-ΔCZ-macc-pada yielded 15.9 mM (1.42 g/L) 3-HP with a yield of 179 mg/g ethanol, while acetate levels were reduced to 2.5 mM (Fig. 5C).

Ethanol-based fed-batch fermentation for 3-HP biosynthesis

Fed-batch fermentation was conducted with the TM-ΔCZ-macc-pada strain harboring the engineered 3-HP biosynthetic pathways, using ethanol as the sole carbon and energy source (Fig. 6). During the first 24 h, cell growth was minimal, with OD₆₀₀ reaching only 1.17, suggesting a prolonged lag phase. From 24 to 48 h, rapid growth occurred, with OD₆₀₀ rising to 13.8, coinciding with active 3-HP production that reached ~ 21 mM. Between 48 and 72 h, growth slowed, peaking at OD₆₀₀ 16.81, while 3-HP levels continued to increase, reaching 43.72 mM (3.92 g/L) by 72 h. After 72 h, OD₆₀₀ declined, indicating reduced viable biomass, while 3-HP concentrations stabilized at ~ 43.72 mM. Acetate levels remained low early in the process but increased sharply after 60 h, reaching 48 mM at 72 h and 67 mM by 96 h.

Fig. 6.

Growth profile and metabolite production during fed-batch fermentation of engineered P. putida using ethanol as the sole carbon source. OD₆₀₀ (red), 3-HP (blue), and acetate (green) concentrations were monitored over a 96-hour period

A total of 3.22% (v/v) ethanol was supplied during the fermentation, resulting in an overall 3-HP yield of 154 mg/g of ethanol. After 72 h 3-HP concentrations remained essentially unchanged while OD₆₀₀ decreased. This decrease in biomass coincided with the onset of acetate accumulation which increased from 48 mM at 72 h to 67 mM at 96 h. The stabilization of 3-HP despite continued ethanol availability indicates that production ceased as biomass viability declined. The overall yield in fed-batch fermentation was lower than that obtained in flask cultures, reflecting the more inhibitory environment created by the combined presence of elevated 3-HP, ethanol and acetate during prolonged cultivation.

Discussion

Ethanol has proven to be an efficient and redox-balanced substrate for acetyl-CoA generation, making it particularly suitable for producing reduced chemicals such as 3-HP. P. putida KT2440, with its robust ethanol oxidation system and stress tolerance, provides an effective microbial chassis for channeling ethanol into value-added metabolites. This study demonstrates how a combination of metabolic engineering and systems-level modeling can expand the utility of P. putida toward scalable C2-based bioprocesses.

The genetic dissection of 3-HP catabolism in P. putida KT2440 demonstrated that multiple enzymes contribute to degradation, with partly redundant roles. Deletion of mmsAB alone did not abolish 3-HP consumption, confirming compensation by other loci. In the WT-ΔmmsΔ488 mutant, residual degradation was only observed above 20 mM 3-HP, suggesting a concentration-dependent induction mechanism, a behavior also reported in P. denitrificans [34]. Complete abolition of degradation in WT-ΔmmsΔ488Δ56 confirmed that simultaneous deletion of mmsAB and both 3-hpdh homologs was necessary to fully block the pathway, thereby providing a stable chassis in which 3-HP accumulation was no longer counteracted by native metabolism.

Comparison of the malonyl-CoA and β-alanine pathways confirmed that both routes can convert ethanol into 3-HP, but with differing efficiencies. The malonyl-CoA route outperformed the β-alanine pathway (2.7 mM vs. 1.6 mM), and the latter was accompanied by significant acetate accumulation (88 mM), consistent with overflow metabolism. Enhancing acetyl-CoA carboxylation with accBC and dtsR1 from C. glutamicum raised 3-HP titers to 4.83 mM, underscoring malonyl-CoA supply as a bottleneck. In the β-alanine pathway, ada expression increased 3-HP to 2.32 mM but diverted flux primarily toward biomass, and acetate accumulation persisted. By contrast, co-activation of both pathways (KT-macc-pada) synergistically raised 3-HP to 11.5 mM (1.02 g/L) with a yield of 129 mg/g ethanol. Notably, acetate concentrations steadily decreased in this strain, demonstrating productive re-assimilation and redistribution of carbon toward 3-HP formation rather than toward biomass or maintenance metabolism.

Deletion of PHA biosynthesis genes further strengthened carbon channeling. In TM-ΔCZ-macc, 3-HP titers increased more than two-fold compared with KT-macc, demonstrating that PHA formation competes directly with malonyl-CoA-derived 3-HP. In contrast, TM-ΔCZ-pada did not improve titers, suggesting that flux redistribution in this background favored biomass rather than product formation. The co-activation of both pathways in TM-ΔCZ-macc-pada resulted in the highest 3-HP titer and yield, while simultaneously lowering acetate levels. For comparison, previous work achieved 1.2 g/L 3-HP in resting E. coli cells via the malonyl-CoA route [42], whereas the actively growing P. putida in this study reached 1.42 g/L under ethanol-fed conditions. Moreover, the yields were comparable to mixed glucose/xylose fermentations in P. putida [22]. Distinct from earlier efforts that relied on inducible promoters [22]; the use of constitutive promoters here simplifies process operation and reduces production cost.

The GSMM provided a mechanistic explanation for the contrasting performance of the two engineered pathways. The strong dependence on the glyoxylate shunt together with very low malate synthase activity imposes a constraint on oxaloacetate regeneration [21] and this constraint limits the β-alanine route. Persistent acetate formation in strains expressing the β-alanine pathway is consistent with this mechanism because acetyl-CoA cannot enter the TCA cycle efficiently when oxaloacetate is limiting. The malonyl-CoA route does not require oxaloacetate and therefore aligns with the higher titers observed experimentally. The model also predicted a substantial drain of acetyl-CoA and malonyl-CoA into PHA biosynthesis and this behavior explains why deletion of phaC1, phaZ and phaC2 enhanced production in the malonyl-CoA background. The consistency between model predictions and experimental performance indicates that ethanol metabolism inherently favors the malonyl-CoA pathway and that relief of oxaloacetate limitation and PHA diversion is essential for optimal 3-HP biosynthesis. These findings highlight the value of integrating metabolic modeling with genetic engineering for rational pathway design in P. putida KT2440.

Fed-batch fermentation confirmed the industrial relevance of this chassis, with engineered strains reaching 3.92 g/L 3-HP. However, stabilization of 3-HP after 72 h coincided with declining biomass, suggesting cytotoxicity due to acidification, membrane destabilization, and accumulation of inhibitory by-products [43, 44]. Ethanol toxicity likely exacerbated these effects, both through membrane perturbation and formation of acetaldehyde and acetate [11, 45]. Notably, acetate rose sharply after 60 h (48 mM at 72 h; 67 mM at 96 h), consistent with overflow metabolism or feedback inhibition in stressed cells. Similar late-stage acetate accumulation has been reported in stressed E. coli and Pseudomonas cultures [46, 47]. The overall fed-batch yield of 154 mg/g ethanol was lower than in flask cultures, where a higher yield of 179 mg/g ethanol was achieved under non-inhibitory conditions.

Although the titers in this study remain below the best-performing microbial platforms reported to date, which include more than 80 g/L 3-HP from glycerol or glucose in engineered E. coli and yeast systems [13, 37] and 48.2 g/L 3-HP from methanol in an optimized methylotrophic host [48], these high-productivity processes rely on carbon sources other than ethanol and commonly require tightly regulated induction systems and multistage fed-batch strategies (Table S4, see supporting information 1). In contrast, the present study demonstrates ethanol-to-3-HP conversion in a solvent-tolerant, actively growing P. putida chassis using a fully constitutive and inducer-free expression system. The fed-batch yield of 154 mg/g ethanol corresponds to approximately 8.4% of the theoretical maximum (1.834 g/g ethanol) [10], while the shake-flask yield of 179 mg/g ethanol reaches about 9.8% of this theoretical limit. These results indicate the need for strain and process improvements to alleviate organic acid and solvent stress during ethanol fed-batch cultivation. Strategies such as ethanol-sensor-based feeding [49] and DO-based ethanol supply [50] tuned to metabolic demand may prevent excess carbon input and reduce inhibitor buildup. Additional yield improvements can be enabled through promoter and RBS strength optimization to balance expression of accBC, dtsR1, mcr, and ada, cofactor-balancing strategies, including transhydrogenase engineering or modulation of the oxidative pentose phosphate pathway to increase NAD(P)H availability, elimination of residual competing sinks in acetyl-CoA or malonyl-CoA metabolism; and tolerance engineering or adaptive laboratory evolution to mitigate ethanol and organic acid stress. These strategies have been shown to substantially increase malonyl-CoA–derived bioproduction efficiencies in multiple hosts and represent actionable routes for elevating both titer and yield in ethanol-fed processes [51–53].

Taken together, these findings underscore ethanol’s effectiveness as a C2 feedstock and P. putida KT2440’s potential as an ethanol-based production host. By combining pathway deletions, dual-route activation, and genome-scale modeling, this study advances the development of inducer-free 3-HP production platforms. Future work should address tolerance through adaptive evolution, implement dynamic control to minimize by-product formation, and test scale-up in continuous fermentation modes to further establish industrial feasibility.

Conclusions

This study demonstrates the successful engineering of P. putida KT2440 for efficient 3-HP production from ethanol as the sole carbon source. Constitutive expression of the malonyl-CoA and β-alanine pathways enabled robust, inducer-free biosynthesis, while deletion of native 3-HP degradation and PHA synthesis pathways redirected carbon flux toward the target product. Introduction of a heterologous acetaldehyde dehydrogenase further enhanced acetyl-CoA availability, strengthening ethanol assimilation. Integration of a genome-scale metabolic model with transcriptomic data revealed the glyoxylate shunt as a central route for carbon distribution, with limited oxaloacetate flux explaining the relative efficiency of the malonyl-CoA pathway. Collectively, these findings validate P. putida KT2440 as a promising microbial chassis for ethanol-based 3-HP biosynthesis and highlight its potential for scalable, inducer-free bioprocesses using renewable C2 substrates.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Arslan Sarwar: Conceptualization, Investigation, Writing–original draft. Linh Thanh Nguyen: Conceptualization, Writing - Review & Editing. Eun Yeol Lee: Conceptualization, Writing - Review & Editing, Supervision, Funding acquisition, Project administration.

Funding

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2024-00466473). This work was supported by the Technology Innovation Program (Industrial Strategic Technology Development Program) (RS-2023-00265608 & 1415188462) funded by the Ministry of Trade, Industry and Energy (MOTIE, South Korea).

Data availability

The datasets used and/or analyzed in the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.