A cell-free artificial anabolic pathway for direct conversion of CO₂ to ethanol†

Wanrong Dong‡ ^ab, Xiuling Ji‡ ^a, Yuhong Huang *^a, Yaju Xue ^a, Boxia Guo ^a, Dongbo Cai ^b, Shouwen Chen *^b and Suojiang Zhang *^a
aBeijing Key Laboratory of Ionic Liquids Clean Process, CAS Key Laboratory of Green Process and Engineering, State Key Laboratory of Multiphase Complex Systems, Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, China. E-mail: yhhuang@ipe.ac.cn; sjzhang@ipe.ac.cn
^bState Key Laboratory of Biocatalysis and Enzyme Engineering, Environmental Microbial Technology Center of Hubei Province, College of Life Sciences, Hubei University, Wuhan 430062, China. E-mail: chenshouwen@hubu.edu.cn

First published on 17th October 2023

---

---

Biological activation and conversion of CO₂ into high-value ethanol are a win–win strategy towards both controlling CO₂ emissions and easing energy crisis. Various natural CO₂ utilization pathways have been identified, including the Wood–Ljungdahl (WL) pathway, the 3-HP bicycle, the 3-hydroxypropionate/4-hydroxybutyrate (HP/HB) cycle, the dicarboxylate/4-hydroxybutyrate (DC/HB) cycle, the reductive glycine pathway, and the reductive tricarboxylic acid (TCA) cycle.¹ These pathways fix CO₂ or its bicarbonate carrier (HCO₃⁻) to produce the ethanol precursor acetyl coenzyme A (acetyl-CoA) via acetyl-CoA synthase (ACS)-mediated condensation of CO₂, thiamine diphosphate (ThDP)-mediated umpolung activation, pyruvate dehydrogenase-mediated pyruvate decarboxylation, or reversal of the oxidative TCA cycle.² However, there is no biosynthetic pathway precedent for the direct conversion of CO₂ to ethanol and redesigning the natural metabolic pathways to accommodate such CO₂ utilization is still challenging owing to the inherent limitations of carbon loss, ATP requirement or oxygen sensitivity.

In an effort to convert CO₂ to ethanol in a fully carbon-conserved and/or ATP-independent manner, recent research studies have first shifted focus to the construction of artificial anabolic pathways for the precursor acetyl-CoA.³ For example, the methanol condensation cycle (MCC) condensed two molecules of formaldehyde into one acetyl-CoA without carbon loss, ATP requirement and anaerobic conditions by the combination of the ribulose monophosphate pathway with the nonoxidative glycolysis pathway.⁴ Similarly, the synthetic acetyl-CoA (SACA) pathway achieved the carbon-conserved, ATP-independent and oxygen-insensitive synthesis of acetyl-CoA by combining ThDP-dependent glycolaldehyde synthase design and pathway construction.⁵ Thanks to these efforts, we can achieve a complete anabolic pathway for ethanol synthesis from CO₂ by including enzymes from CO₂ to formaldehyde and from acetyl-CoA to ethanol to extend the MCC. However, the equilibrium direction of multi-enzyme reaction remains difficult to control.⁶ All reactions are reversible except the MCC, and in particular, the CO₂ activation to formaldehyde is thermodynamically unfavorable due to the low solubility of CO₂ and the inefficiency of activation enzymes.⁷ Thus, the net concentration of ethanol is primarily a matter of maintaining the formaldehyde intermediate converted from CO₂ at a relatively high concentration.

To address this, we introduced a modularization strategy based on the enzymatic activation of CO₂, formaldehyde assimilation into acetyl-CoA, and ethanol synthesis modules to regulate the metabolic fluxes for direct conversion of CO₂ to ethanol. The efficient enzymatic CO₂ activation to formaldehyde was achieved by the high substrate binding affinity and catalytic efficiency of both formate dehydrogenase mutant (ΔPaFDH48)⁸ and formaldehyde dehydrogenase (SzFaldDH)⁹ from Paracoccus sp. MKU1 and Streptomyces zinciresistens, thereby providing an attractive starting point for the artificial CO₂ utilization pathway. This activation module can ensure sufficient formaldehyde concentrations for further conversion during the catalytic process, allowing the construction of a complete biosynthetic pathway using CO₂ as a feedstock. By assembling three functional modules, we first constructed a carbon-conserved and ATP-independent artificial ethanol pathway from CO₂, proposed as CO₂ To Ethanol (CTE) (Scheme 1). ¹³C-labeled CO₂ and gas chromatography-tandem mass spectrometry (GC-MS) revealed that modular optimization can strengthen the metabolic flux of the CTE pathway, resulting in an ethanol concentration of 0.37 mM at a conversion rate of 4.33 nmol CO₂ min⁻¹ mg⁻¹ enzymes and 0.5 mM R5P supply.

To determine the thermodynamic and kinetic feasibility of the CTE pathway constructed from CO₂ activation, formaldehyde assimilation, and ethanol synthesis modules, we first calculated its Gibbs energy change for each reaction and max–min driving force (MDF) during the conversion of CO₂ to ethanol. The entire reaction from CO₂ to ethanol was thermodynamically favorable with a total Gibbs energy change (ΔrG′^m) of −25.8 kJ mol⁻¹, except for the initial CO₂ activation, which had a thermodynamically unfavorable cascade reaction (Fig. 1). The large negative values of ΔrG′^m for hexulose-6-phosphate synthase (Hps), phosphohexulose isomerase (Phi) and phosphoketolase (Fpk) (−9.7, −10.5, and −53.5 kJ mol⁻¹) in the formaldehyde assimilation module provided important driving forces for the final formation of ethanol. Meanwhile, the CTE pathway also achieved a positive MDF value of 0.21 kJ mol⁻¹ at allowable reduced nicotinamide adenine dinucleotide (NADH) concentrations (Fig. S1†), further indicating its feasibility of thermodynamic operation. Combining the results of ΔrG′^m and MDF, we concluded that the CTE pathway was theoretically functional for the biosynthesis of ethanol from CO₂in vitro.

Since the bioactivation of CO₂ to formaldehyde is thermodynamically unfavorable, there exists no complete biosynthetic pathway for multicarbon molecules from CO₂.¹⁰ The high reductive activity of FDH and FaldDH in the activation module is a key factor in supplying sufficient formaldehyde for reactions of the other two modules. However, commercially available dehydrogenases yielded only 0.013 mM formaldehyde after 5 h, which was far from efficient activation.¹¹ We thus constructed four alternative CO₂ activation modules using PaFDH or its mutant ΔPaFDH48 and SzFaldDH or the previously reported efficient FaldDH from Burkholderia multivorans (BmFaldDH) in tandem^7,12 (Fig. 2A and B). After expression and purification, the reductive activity of all candidate activation assays was investigated using CO₂ as the substrate. The activation modules containing ΔPaFDH48 displayed higher reductive activities compared to PaFDH and commercial FDH from C. boidinii (CbFDH). A significant increase in the reductive activity was only observed in the activation module containing ΔPaFDH48 and SzFaldDH acting in tandem, which was 10.68-fold higher than that achieved by the commercial dehydrogenase activation module. Thus, efficient ΔPaFDH48 and SzFaldDH were chosen to construct the CO₂ activation module.

The formaldehyde assimilation into acetyl-CoA and ethanol synthesis modules was further established by integrating the previously reported MCC^4a and the aldehyde dehydrogenase (AlDH) and alcohol dehydrogenase (ADH) to reduce acetyl-CoA.¹³ With the exception of commercial ADH, the catalytic activities of all corresponding enzymes in both modules were demonstrated in individual assays after gene synthesis and enzyme expression and purification (Fig. S2–S7†). As the initial step in the MCC, Hps irreversibly catalyzed formaldehyde and ribulose-5 phosphate (Ru5P) to hexulose-6-phosphate (H6P), which realized the formaldehyde assimilation and in turn provided the driving force for CO₂ activation.¹⁴ To find highly active Hps, we purified enzymes from three different strains including Bacillus subtilis, Methylococcus capsulatus and Bacillus methanolicus (BsHps, McHps, and BmHps)^4a,15 (Fig. 2C). The relative activity of McHps has been shown to be the highest compared to those of BmHps and the inactive BsHps, as determined by measuring generated NADPH at 340 nm (Fig. 2D). However, its low enzyme activity under the optimized buffer conditions still led to small amounts of formaldehyde assimilation. Meanwhile, we discovered and screened two Hps using peptide pattern recognition (PPR).¹⁶ Unfortunately, both purified enzymes from Methylocaldum marinum demonstrated relatively lower activity than McHps (Fig. S8†). Taken together, McHps is chosen to accomplish formaldehyde assimilation into H6P, which is isomerized to fructose-6-phosphate (F6P) by Phi. F6P can be cleaved by Fpk to acetylphosphate (AcP) and erythrose 4-phosphate (E4P). Subsequently, the generated AcP and E4P were converted to acetyl-CoA by phosphate acetyltransferase (Pta) and regenerated Ru5P via a carbon rearrangement, respectively. During this process, the Fpk concentration was optimized because the inherent kinetic trap of Fpk would inevitably accumulate E4P at the expense of decreasing AcP yield.^4a As the Fpk concentration increased, the ethanol yield in the system initially increased and then decreased (Fig. 2E). The maximum yield of ethanol reached 2.51 mM, when the Fpk concentration was 160 μg mL⁻¹. Based on this formaldehyde assimilation module, we further added another two known dehydrogenases, AlDH and ADH, which would convert acetyl-CoA into ethanol. Consequently, 1.83 mM ethanol was produced at a formaldehyde concentration of 6.0 mM (Fig. 2F). Our results indicated the feasibility of the ethanol synthesis from formaldehyde by assembling formaldehyde assimilation and ethanol synthesis modules in vitro.

On establishing individual modules for the conversion of CO₂ into ethanol, we constructed the entire pathway by assembling three modules and achieved a detectable ethanol concentration of 0.10 mM. Since CO₂ and initially added 0.5 mM ribose 5-phosphate (R5P) were employed as feedstock, ¹³C-labeled experiments to detect the carbon flux were thus performed to verify the actual source of carbon in ethanol. By comparing with “unlabeled” controls, we tested the ethanol synthesis from the CTE pathway using ¹³C-labeled CO₂ and HCO₃⁻, respectively. GC-MS showed that the main fragmentation patterns (i.e. [M − 1]⁺ and [M]⁺) of ethanol (molecular weight = 46) achieved in the control group were 45 and 46 ions, corresponding to unlabeled carbon in ethanol (Fig. 3A and S9A†). In contrast, another 47 ions were detected by ¹³C-labeled CO₂ along with 45 and 46 ions, while both 47 and 48 ions were detected by ¹³C-labeled HCO₃⁻ (Fig. 3B and S9B†). The emerging ions were designated as single labeled [2-¹³C]-ethanol and fully labeled [1,2-¹³C]-ethanol. These undetectable 48 ions from ¹³C-labeled CO₂ led to speculation that the low formaldehyde concentration from the CO₂ activation module may not be able to supply multiple passes of CO₂ conversion. Our results confirmed that some of the carbons in ethanol came from CO₂, verifying the full functionality of ethanol synthesis from CO₂ in CTE.

We further proposed a carbon flux network to understand the metabolic logic of the CTE pathway. As shown in Fig. 3C, the ¹³C-labeled CO₂ was first converted to ¹³C-labeled formaldehyde through a cascade reaction. Then the resulting ¹³C-labeled formaldehyde and Ru5P, which was isomerized from the initial added R5P, entered the formaldehyde assimilation reaction and successively produced ¹³C-labeled H6P and F6P by Hps and Phi. Finally, ¹³C-labeled F6P was cleaved to E4P and ¹³C-labeled AcP, which was further used to produce single labeled [2-¹³C]-ethanol (black line). Meanwhile, ¹³C-labeled F6P and E4P were used to regenerate unlabeled R5P and synthesize ¹³C-labeled xylulose 5-phosphate (X5P), which was then isomerized to ¹³C-labeled Ru5P. Similarly, the ¹³C-labeled Ru5P and the newly generated ¹³C-labeled formaldehyde would also enter the following formaldehyde assimilation and ethanol synthesis reactions to produce fully labeled [1,2-¹³C]-ethanol (orange lines). In addition, some fully labeled [1,2-¹³C]-ethanol can be formed from ¹³C-labeled formaldehyde and double ¹³C-labeled Ru5P produced from multiple passes (violet line). Therefore, we concluded that a single pass of CO₂ conversion in CTE mainly synthesized single labeled [2-¹³C]-ethanol, while [1,2-¹³C]-ethanol can be synthesized by the assistance of the Ru5P regeneration cycle.

After validating the full function for ethanol synthesis from CO₂ in CTE, we proceeded towards improving this pathway by addressing the bottleneck of the inadequate supply of formaldehyde. Due to the unfavorably reversible cascade reaction conditions for CO₂ activation, we met the need for relatively high formaldehyde concentrations in the formaldehyde assimilation module with a two-step enzymatic cascade reaction in CTE. In the first hour, CO₂ was first activated to formaldehyde by ΔPaFDH48 and SzFaldDH in tandem, and the resulting formaldehyde with the precursor Ru5P was irreversibly converted to H6P and then to F6P by Hps and Phi. In the following 1 h, the accumulated F6P was further cleaved to AcP and E4P by supplementing the remaining 7 core enzymes and CoA, which were used to produce ethanol and for carbon rearrangement, respectively (Fig. 4A). The final concentration of ethanol reached 0.30 mM, which was almost 3-fold higher than that achieved using the one-pot strategy (0.10 mM) (Fig. 4B). Such yield improvement demonstrates the effectiveness of the two-step reaction.

With the successful improvement of the CTE pathway with the two-step reaction, we intended to convert CO₂ into ethanol over a 2.5 h time course. Starting with 12 mL min⁻¹ of CO₂ and 1 h of F6P accumulation, the CTE pathway using ΔPaFDH48 and SzFaldDH achieved the highest ethanol concentration of 0.37 mM in the following 30 min (Fig. 4C). The ethanol synthesis rate reached 4.33 nmol CO₂ min⁻¹ mg⁻¹ enzymes, which is 1.37-fold higher than those of PaFDH and BmFaldDH. After 30 min, the ethanol concentration decreased slightly, which may be due to the instability of intermediates like acetyl-CoA and AcP.^4a Our result demonstrated the feasibility of continuous ethanol synthesis from CO₂ in the CTE pathway.

Conclusions

In summary, a carbon-conserved and ATP-independent anabolic pathway for direct conversion of CO₂ to ethanol was presented. Thanks to the modularization strategy based on the enzymatic activation of CO₂, formaldehyde assimilation into acetyl-CoA, and ethanol synthesis modules, this complete biosynthetic pathway achieves not only the efficient activation of CO₂ into formaldehyde but also the carbon rearrangement with high carbon and energy efficiency from formaldehyde. As a result, CO₂ was converted into ethanol with a concentration of 0.37 mM at a conversion rate of 4.33 nmol CO₂ min⁻¹ mg⁻¹ enzymes. Our study presents a green and attractive avenue for the direct conversion of CO₂ into versatile biofuels, providing a feasible strategy and valuable inspiration for constructing a complete biosynthetic pathway from CO₂.

Author contributions

Y. Huang, S. Chen and S. Zhang supervised the project. Y. Huang conceived the project and performed the PPR prediction. W. Dong performed the experimental work. X. Ji discussed and wrote the manuscript. Y. Xue and B. Guo worked on the purification and protein quantification of key enzymes for CO₂ activation. D. Cai discussed the manuscript. X. Ji, W. Dong and Y. Huang revised the manuscript with support from all the authors.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Key Research and Development Program (2021YFC2104200), the “Transformational Technologies for Clean Energy and Demonstration” Strategic Priority Research Program of the Chinese Academy of Sciences (XDA21000000), and the CAS Project for Young Scientists in Basic Research (YSBR-072). This work was also partially supported by the Open Funding Project of the State Key Laboratory of Biocatalysis and Enzyme Engineering (SKLBEE2021001).

References

1. (a) Z. Liu, K. Wang, Y. Chen, T. Tan and J. Nielsen, Nat. Catal., 2020, 3, 274–288 CrossRef CAS ; (b) M. Kopke, C. Held, S. Hujer, H. Liesegang, A. Wiezer, A. Wollherr, A. Ehrenreich, W. Liebl, G. Gottschalk and P. Durre, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 13087–13092 CrossRef CAS PubMed ; (c) F. Gong, Z. Cai and Y. Li, Sci. China: Life Sci., 2016, 59, 1106–1114 CrossRef CAS PubMed ; (d) R. Gupta, A. Mishra, Y. Thirupathaiah and A. K. Chandel, Biomass Convers. Biorefin., 2022 DOI:10.1007/s13399-022-02552-8 .
2. S. Bierbaumer, M. Nattermann, L. Schulz, R. Zschoche, T. J. Erb, C. K. Winkler, M. Tinzl and S. M. Glueck, Chem. Rev., 2023, 123, 5702–5754 CrossRef CAS PubMed .
3. (a) I. W. Bogorad, T. S. Lin and J. C. Liao, Nature, 2013, 502, 693–697 CrossRef CAS PubMed ; (b) T. Schwander, L. Schada von Borzyskowski, S. Burgener, N. S. Cortina and T. J. Erb, Science, 2016, 354, 900–904 CrossRef CAS PubMed ; (c) L. Xiao, G. Liu, F. Gong, H. Zhu, Y. Zhang, Z. Cai and Y. Li, ACS Catal., 2021, 12, 799–808 CrossRef .
4. (a) I. W. Bogorad, C. T. Chen, M. K. Theisen, T. Y. Wu, A. R. Schlenz, A. T. Lam and J. C. Liao, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 15928–15933 CrossRef CAS PubMed ; (b) J. Wang, Z. Chen, X. Deng, Q. Yuan and H. Ma, Bioengineering, 2023, 10, 415 CrossRef CAS PubMed ; (c) X. Yang, Q. Yuan, H. Luo, F. Li, Y. Mao, X. Zhao, J. Du, P. Li, X. Ju, Y. Zheng, Y. Chen, Y. Liu, H. Jiang, Y. Yao, H. Ma and Y. Ma, Metab. Eng., 2019, 56, 142–153 CrossRef CAS PubMed .
5. X. Lu, Y. Liu, Y. Yang, S. Wang, Q. Wang, X. Wang, Z. Yan, J. Cheng, C. Liu, X. Yang, H. Luo, S. Yang, J. Gou, L. Ye, L. Lu, Z. Zhang, Y. Guo, Y. Nie, J. Lin, S. Li, C. Tian, T. Cai, B. Zhuo, H. Ma, W. Wang, Y. Ma, Y. Liu, Y. Li and H. Jiang, Nat. Commun., 2019, 10, 1378 CrossRef PubMed .
6. W. Liu, Y. Hou, B. Hou and Z. Zhao, Chin. J. Chem. Eng., 2014, 22, 1328–1332 CrossRef CAS .
7. X. Ji, Y. Xue, Z. Li, Y. Liu, L. Liu, P. K. Busk, L. Lange, Y. Huang and S. Zhang, Green Chem., 2021, 23, 6990–7000 RSC .
8. Y. Huang, Y. Xue, X. Ji and S. Zhang, CN2022115986319, 2022 .
9. Y. Huang, X. Ji, Y. Xue, B. Guo and S. Zhang, CN2022115994885, 2022 .
10. (a) T. Zheng, M. Zhang, L. Wu, S. Guo, X. Liu, J. Zhao, W. Xue, J. Li, C. Liu, X. Li, Q. Jiang, J. Bao, J. Zeng, T. Yu and C. Xia, Nat. Catal., 2022, 5, 388–396 CrossRef CAS ; (b) T. Cai, H. Sun, J. Qiao, L. Zhu, F. Zhang, J. Zhang, Z. Tang, X. Wei, J. Yang, Q. Yuan, W. Wang, X. Yang, H. Chu, Q. Wang, C. You, H. Ma, Y. Sun, Y. Li, C. Li, H. Jiang, Q. Wang and Y. Ma, Science, 2021, 373, 1523–1527 CrossRef CAS PubMed ; (c) J. Zhang, D. Liu, Y. Liu, H. Chu, J. Bai, J. Cheng, H. Zhao, S. Fu, H. Liu, Y. Fu, Y. Ma and H. Jiang, Green Chem., 2023, 25, 3247–3255 RSC ; (d) S. Desmons, K. Grayson-Steel, N. Nunez-Dallos, L. Vendier, J. Hurtado, P. Clapes, R. Faure, C. Dumon and S. Bontemps, J. Am. Chem. Soc., 2021, 143, 16274–16283 CrossRef CAS PubMed .
11. M. Guo, F. Gu, L. Meng, Q. Liao, Z. Meng and W. Liu, Sep. Purif. Technol., 2022, 286, 120480 CrossRef CAS .
12. R. K. Singh, R. Singh, D. Sivakumar, S. Kondaveeti, T. Kim, J. Li, B. H. Sung, B.-K. Cho, D. R. Kim, S. C. Kim, V. C. Kalia, Y.-H. P. J. Zhang, H. Zhao, Y. C. Kang and J.-K. Lee, ACS Catal., 2018, 8, 11085–11093 CrossRef CAS .
13. Q. Wang, C. Sha, H. Wang, K. Ma, J. Wiegle, A. E. Abomohra and W. Shao, Sci. Rep., 2021, 11, 1050 CrossRef CAS PubMed .
14. I. Orita, N. Sakamoto, N. Kato, H. Yurimoto and Y. Sakai, Appl. Microbiol. Biotechnol., 2007, 76, 439–445 CrossRef CAS PubMed .
15. J. E. N. Muller, F. Meyer, B. Litsanov, P. Kiefer, E. Potthoff, S. Heux, W. J. Quax, V. F. Wendisch, T. Brautaset, J. C. Portais and J. A. Vorholt, Metab. Eng., 2015, 28, 190–201 CrossRef CAS PubMed .
16. (a) P. K. Busk and L. Lange, Appl. Environ. Microbiol., 2013, 79, 3380–3391 CrossRef CAS PubMed ; (b) P. K. Busk, B. Pilgaard, M. J. Lezyk, A. S. Meyer and L. Lange, BMC Bioinf., 2017, 18, 214 CrossRef CAS PubMed ; (c) Y. Xue, Y. Zhao, X. Ji, J. Yao, P. K. Busk, L. Lange, Y. Huang and S. Zhang, Green Chem., 2020, 22, 8656–8668 RSC .

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3gc03159d

‡ These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2023