A new-to-nature carboxylation module to improve natural and synthetic CO2 fixation

The conversion of the C2 compounds glycolate and glyoxylate into C3 metabolites is central to multiple metabolic processes, including photorespiration, fatty acid assimilation, formato- and methylotrophy, and as a junction for several synthetic CO2 fixation pathways. However, existing natural routes that convert these C2 intermediates to C3 metabolites invariably lose carbon as CO2, limiting carbon efficiency. Examples include the glyoxylate cycle and the β-hydroxyaspartate cycle, which require decarboxylation to generate C2 metabolites, and photorespiration and the glycerate pathway, which convert two glyoxylate molecules into glycerate while releasing CO2. These carbon losses are particularly detrimental in photorespiration, where crop yields can be reduced by up to 50% in hot and dry climates. Thus, synthetic routes that avoid carbon and energy loss during glycolate assimilation are expected to substantially enhance productivity. The tartronyl-CoA (TaCo) pathway was proposed as a direct, CO2-fixing route for glycolate assimilation predicted to outperform naturally evolved routes, but it remained theoretical because neither tartronyl-CoA nor the required enzymatic reactions were known in nature. A central challenge was engineering glycolyl-CoA carboxylase (GCC), the key carboxylating enzyme of the TaCo pathway. Here, the authors reconstitute the TaCo pathway in vitro by identifying and engineering the necessary enzymes, demonstrate its function, and interface it with photorespiration, ethylene glycol conversion, and a synthetic CO2 fixation cycle, aiming to improve carbon and energy efficiencies across these processes.

Natural metabolic routes for converting glycolate/glyoxylate (C2) to C3 metabolites are few and all incur CO2 loss (glyoxylate cycle; β-hydroxyaspartate cycle; photorespiration/glycerate pathway). Photorespiratory carbon loss is a major constraint on agricultural productivity, especially under heat/drought. Previous design studies proposed synthetic CO2-fixing pathways with higher carbon efficiency, including the hypothetical tartronyl-CoA (TaCo) pathway to assimilate glycolate by fixing CO2 rather than releasing it. Enzymatic precedents for key steps (for example, carboxylation of acyl-CoAs by biotin-dependent carboxylases such as propionyl-CoA carboxylase, PCC) suggested possible scaffolds for enzyme engineering, but no natural enzyme was known to catalyze glycolyl-CoA carboxylation. Prior work on acetyl-CoA synthetases (ACS) highlighted post-translational regulation by acetylation and engineering strategies to alter substrate scope. Malonyl-CoA reductase (MCR) from Chloroflexus aurantiacus provided a candidate for downstream reduction steps. Together, this context motivated engineering a new-to-nature GCC and assembling a full TaCo module to validate the theoretical pathway and benchmark it against natural and synthetic alternatives.

- Enzyme discovery and engineering for glycolyl-CoA formation: Screened 11 native/engineered enzymes for glycolate activation via CoA transferase or acyl-CoA synthetase routes. Best CoA transferase (AbfT from Clostridium aminobutyricum) showed catalytic efficiency 120 M−1 s−1 with acetyl-CoA as donor. Best ACS homologue (EryACS1 from Erythrobacter sp. NAP1) had 20 M−1 s−1 with glycolate. To mitigate in vivo acetylation-based inactivation, tested Leu641Pro (doubling specific activity but increasing Km) and produced EryACS1 in an E. coli BL21 (DE3) ΔI apatZ acetylase knockout strain, increasing catalytic efficiency to 540 M−1 s−1. Structure-guided mutagenesis at Val379 yielded Val379Ala (GCS) with apparent Km (glycolate) 13 ± 3 mM, kcat 11.1 ± 0.6 s−1, and kcat/Km 853 M−1 s−1. - Downstream reduction: Identified malonyl-CoA reductase from Chloroflexus aurantiacus (CaMCR) functioning as tartronyl-CoA reductase (TCR), converting tartronyl-CoA to glycerate in two steps with kcat 1.4 s−1 and apparent Km 26 μM. - Engineering glycolyl-CoA carboxylase (GCC): Screened four propionyl-CoA carboxylases (PCCs) for promiscuity; only PCC from Methylorubrum extorquens (MePCC) showed minimal activity on glycolyl-CoA (kcat ≈ 0.01 s−1) with high futile ATP hydrolysis (~100:1 ATP hydrolyzed per carboxylation). Obtained a 3.48-Å cryo-EM dataset and built a structural model to guide active-site redesign. Introduced substitutions Tyr143His and Asp407Ile to accommodate the substrate hydroxyl and Leu100Ser to support a hydrogen-bonding network (GCC M3), improving catalytic efficiency (>50-fold) and lowering futile ATP hydrolysis (>15-fold) vs. wild type. Performed two rounds of error-prone PCR on the carboxyltransferase subunit, screening libraries with a high-throughput droplet microfluidics TCR-coupled assay under ATP-limiting conditions and subsequent microplate kinetic screens. The best variant (GCC M5: Asp407Ile, Tyr143His, Leu100Ser, Ile450Val, Trp502Arg) achieved kcat 5.6 ± 0.3 s−1, kcat/Km 3.6 × 10^4 M−1 s−1, and >25-fold lower ATP hydrolysis than wild type. - Structural characterization: Determined a 1.96-Å cryo-EM structure of GCC M5 β-subunit core and a 3.48-Å structure of MePCC. Modelled active-site interactions, revealing a stabilizing H-bond between His143 and Asp171; identified positions of engineered substitutions relative to the active site; observed biotin in a non-catalytic ‘parking’ position. - Thermodynamic and stoichiometric analyses: Computed the max-min driving force (MDF) of the TaCo pathway using component contribution; MDF >7 kJ mol−1, similar to the CBB cycle and higher than glycolysis, indicating thermodynamically favorable operation with minimal backward flux. Conducted flux balance analysis to compare photorespiration strategies combined with CBB for net fixation of three CO2 to one 3-PGA. - In vitro reconstitutions and interfacing: • Sequential TaCo reconstitution from glycolate with 13C-bicarbonate, monitoring glycolyl-CoA, tartronyl-CoA, and glycerate by UPLC-MS/MS; ATP regeneration via phosphocreatine outperformed polyphosphate. • Photorespiration bypass: Coupled TaCo with 2-PG phosphatase and glycerate kinase; developed a malate read-out module with 13C-labeling to quantify pathway contribution under conditions mimicking 100% RuBisCO oxygenation. • Ethylene glycol conversion: Built a module using an alcohol dehydrogenase (initially FucO, later Gox0313), an aldehyde dehydrogenase (PduP) to form glycolyl-CoA, GCC, TCR, and NADH oxidase for cofactor balance; integrated ATP regeneration. • Coupling to the CETCH cycle: Combined 17 CETCH enzymes with glyoxylate reductase and TaCo; removed side reactions (avoiding formyl-CoA formation, replacing succinyl-CoA reductase to reduce glycolyl-CoA off-target activity) and optimized cofactor regeneration and ATP supply. - Library generation, screening, and assays: Constructed random mutagenesis libraries via MEGAWHOP and error-prone PCR at controlled mutation rates (0.2–3.6 mutations/kbp). Screened using droplet microfluidics and 384-well plate coupled assays (TCR-coupled NADPH readout). Stability and biotinylation assessed via circular dichroism and avidin gel-shifts. Kinetics determined from multiple independent measurements.

- Created a new-to-nature CO2-fixing enzyme, glycolyl-CoA carboxylase (GCC), by rational design and directed evolution. Final variant GCC M5 exhibited kcat 5.6 ± 0.3 s−1, apparent Km 0.15 ± 0.03 mM for glycolyl-CoA, and catalytic efficiency 3.6 × 10^4 M−1 s−1—an improvement of ~10^3-fold over the MePCC wild type—with >25-fold reduction in futile ATP hydrolysis. - Glycolyl-CoA synthetase (GCS; EryACS1 V379A) achieved kcat 11.1 ± 0.6 s−1, apparent Km 13 ± 3 mM, kcat/Km 853 M−1 s−1. Producing EryACS1 in an acetylase knockout strain raised catalytic efficiency to 540 M−1 s−1 (pre-V379A), surpassing prior engineered ACS with glycolate. - Tartronyl-CoA reductase (TCR; CaMCR) converted tartronyl-CoA to glycerate with kcat 1.4 s−1 and apparent Km 26 μM (kcat/Km 5.4 × 10^4 M−1 s−1). - Verified atomic-level active-site redesign with a 1.96-Å cryo-EM structure of GCC M5; clarified roles of engineered residues (His143–Asp171 H-bond; positions of Ile450Val and Trp502Arg) and observed biotin at a ‘parking’ site. - TaCo pathway thermodynamics: MDF >7 kJ mol−1, indicating favorable driving forces comparable to the CBB cycle and superior to glycolysis. - In vitro TaCo operation: • Sequential reconstitution from glycolate and 13C-bicarbonate yielded glycerate at 27 ± 1 nmol min−1 mg−1 total protein. • Photorespiration bypass: Malate read-out showed a 130 μM (33%) increase in malate attributable to TaCo; isotopic labeling confirmed double 13C incorporation from 2-PG via TaCo; TaCo produces physiologically relevant (R)-glycerate (accepted by E. coli GlxK). • Ethylene glycol conversion: Optimized system produced 485 μM glycerate at 2.1 nmol min−1 mg−1 (fivefold rate increase over initial setup). • Coupling with CETCH: After addressing side reactions and optimizing cofactors, produced 331 μM glycerate at 4.8 nmol min−1 mg−1 TaCo enzymes, comparable to standalone CETCH CO2 fixation rates (~5 nmol min−1 mg−1). - Systems-level impacts (calculations): Replacing natural photorespiration with TaCo can increase carbon efficiency from 75% to 150% and reduce ATP by 21% and reducing equivalents by 29% for net production of one 3-PGA from three CO2. GCC/TaCo predicted to increase carbon efficiency across interfaced processes by up to 150% while lowering theoretical energy demand.

The study addresses the challenge of limited natural carboxylases by engineering a new-to-nature CO2-fixing enzyme (GCC) and assembling a complete tartronyl-CoA (TaCo) module that converts glycolate (C2) into glycerate (C3) with net CO2 fixation. By starting from a promiscuous PCC scaffold, structure-guided mutations and directed evolution shifted specificity and boosted catalytic performance to levels comparable with natural biotin-dependent acyl-CoA carboxylases, while markedly reducing futile ATP hydrolysis. The atomic-resolution cryo-EM structure confirms and rationalizes the effectiveness of active-site redesign. Integrating GCS, GCC, and TCR yields a thermodynamically favorable pathway with strong driving forces, enabling efficient in vitro operation and seamless interfacing with key metabolic contexts: a carbon-positive photorespiration bypass, valorization of ethylene glycol to a central metabolite, and augmentation of a synthetic CO2 fixation cycle (CETCH). Stoichiometric analyses predict substantial gains in carbon efficiency (up to 150%) and reduced energy demands versus natural photorespiration and other bypasses, highlighting the broader relevance to biotechnology and agriculture. The work demonstrates that expanding enzyme repertoires beyond natural reactions can unlock more efficient metabolic solutions and provides a blueprint for building improved carbon assimilation modules.

This work constructs and validates a new-to-nature, CO2-dependent tartronyl-CoA pathway for glycolate assimilation, centered on an engineered glycolyl-CoA carboxylase (GCC) with catalytic properties akin to natural acyl-CoA carboxylases. Together with engineered GCS and identified TCR, the module operates efficiently in vitro, interfaces with photorespiration, ethylene glycol conversion, and the CETCH cycle, and is predicted to substantially improve carbon yield and reduce energy requirements. The atomic-resolution structure corroborates the design principles that enabled the activity shift. Future directions include further directed evolution to enhance flux, eliminate residual futile ATP hydrolysis in GCC, refine cofactor and energy regeneration, and in vivo implementation to assess performance in cellular contexts and crops. This approach exemplifies how designing theoretically feasible but non-natural reactions can expand metabolic solution spaces and improve sustainability in biotechnology and agriculture.

- Residual futile ATP hydrolysis by GCC persists (though greatly reduced), potentially lowering energy efficiency. - The natural-substrate (propionyl-CoA) catalytic efficiency of GCC M5 decreased about twofold relative to wild type, indicating trade-offs in reprogramming specificity. - Cryo-EM structures did not capture glycolyl-CoA bound in the active site (only free CoA observed), likely due to thioester hydrolysis; α subunits appeared flexible, with unresolved biotin carboxylase domains. - Results are demonstrated in vitro; in vivo performance, regulation, metabolite channeling, and cellular burdens remain to be validated. - Some modules required careful avoidance of side reactions (for example, formyl-CoA formation; off-target reduction of glycolyl-CoA), indicating system-level optimization needs for broader deployment.