Rubisco, the enzyme that catalyzes the carboxylation of RuBP in CBB cycle, is one of the most abundant enzymes on the earth [12]. Given its central role in CO₂ fixation, Rubisco has been intensively studied. As a carboxylase, it is widely recognized that Rubisco has low catalytic efficiency, with an average turnover number (k_cat,c) of 4.15 s⁻¹, compared with the k_cat values of approximately 80 s⁻¹ of typical central metabolic enzymes [49]. In addition to the poor catalytic efficiency, Rubisco also catalyzes a non-productive oxidative side reaction, producing a toxic by-product 2-phosphoglycolate, which must be recycled through the phosphoglycolate salvage process mentioned above, thus leading to energy and carbon loss [24]. Furthermore, there exists a proposed tradeoff between substrate specificity and carboxylase activity [30,50]. In other words, Rubisco with higher catalytic efficiency has less differentiation between CO₂ and O₂, resulting in more oxidation reactions and more energy input for the phosphoglycolate salvage process. Most Rubiscos function in atmosphere with O₂-abundant and CO₂-scarce condition (21% O₂, 0.04% CO₂, corresponding to 0.012 mM CO₂ and 0.26 mM bicarbonate in water (pH 7.4, 20 °C)), the carboxylation efficiency of which further decreases [29]. For example, the net carboxylation rate of Rubisco from the cyanobacterium Synechococcus elongatus PCC 7942 is estimated to be approximately 4% of its apparent k_cat,c (14 s⁻¹) under atmospheric conditions [50]. To date, four forms of Rubiscos have been found in nature [51,52]. Forms I to III are bona fide Rubiscos that catalyze the carboxylation or oxygenation of RuBP, while form IV Rubiscos termed Rubisco-like proteins (RLPs) are not involved in carbon fixation. Form I Rubiscos are widely distributed in plants, algae, cyanobacteria, and proteobacteria that rely on CBB cycle for carbon fixation. Form I Rubiscos are composed of eight large (L) and eight small (S) subunits organized in an L₈S₈ structure, representing the most structurally complex Rubiscos [51]. Small subunits were found to impact the conformation of the catalytic sites of large subunits dimers, and affect catalytic efficiency and substrate specificity [53]. A recently discovered clade of form I Rubiscos lacking small subunits exhibits a relatively low k_cat,c, and average K_m,c compared to form I enzymes [54]. In addition, certain form I Rubiscos are associated with CO₂ concentrating mechanisms to improve carbon fixation efficiency (see below) [13]. Form II Rubiscos are mainly found in proteobacteria, archaea, and dinoflagellate algae, and also participate in CBB cycle [52]. Unlike form I Rubiscos, form II Rubiscos only consist of large subunits that form an L₂ or L₆ complex, exhibiting a simpler structure [51,55]. As a result, form II Rubiscos generally have a higher catalytic rate (k_cat,c about 7 s⁻¹) and lower substrate specificity (K_m,c about 85 μM) compared with those of form I Rubiscos (K_m,c less than 20 μM, k_cat,c about 3 s⁻¹) [50]. Form II Rubiscos typically exist in CO₂-rich and/or anaerobic environments [21]. Form III Rubiscos are mainly found in archaea and participate in the ribonucleosides metabolism of pentose biphosphate pathway rather than the CBB cycle [52,56]. These archaea lack a complete pentose phosphate pathway, and the ribose moieties of nucleosides are important carbon sources and are linked to central metabolism through carboxylation, followed by cleavage catalyzed by form III enzymes. Form IV Rubiscos are not real Rubiscos and do not catalyze the key reaction of carboxylation, but share sequence similarity with Rubiscos despite lacking the conserved amino acid residues essential for carboxylation [13,51,57]. The catalytic mechanism of RLPs is similar to that of Rubiscos via the formation of an enolate intermediate, suggesting a divergent evolution.

The diversity of Rubiscos with different subunits, structural complexity, and kinetic parameters provides evidence for the evolution of Rubiscos [13]. It is believed that over a long period of evolution, Rubiscos have optimized substrate specificity and enzymatic activity for the hosts’ niche [64]. Rubiscos are not independent enzymes and require a series of proteins and enzymes to assist Rubiscos to function properly [65]. The natural evolution process of Rubiscos provides important clues for improving Rubiscos and CBB cycle. Considering the central role in CO₂ fixation, there have been continuous efforts to improve the catalytic performance of Rubiscos. The latest research progress in improving the efficiency of Rubiscos for carboxylation is summarized below.

The most direct way to enhance the carboxylation of CBB cycle is to introduce highly active Rubiscos. For example, a Rubisco with higher catalytic efficiency (k_cat,c value about 14 s⁻¹) from Synechococcus species was introduced into the facultative chemoautotroph Cupriavidus necator H16 in combination with the expression of endogenous GroES/EL chaperones, leading to an increased growth rate and biomass production of the engineered strain growing autotrophically [66]. Moreover, simultaneous overexpression of RcbX chaperone and hydrogenases resulted in a 93.4% increase in final biomass and a 99.7% increase in polyhydroxybutyrate (PHB) production. This result demonstrated that efficient Rubisco, along with appropriate CO₂ supply, can sustain rapid growth and carbon fixation of autotrophic microorganisms.

Since effective Rubiscos have a significant effect on the growth rate of autotrophic microorganisms, screening Rubiscos with a high carboxylation rate is important to improve CO₂ fixation efficiency. Through systematical mining of genomic information, the most catalytically efficient Rubisco was identified from Gallionella genus [55]. The Rubisco from Gallionella belongs to form II and has a k_cat,c of 22 s⁻¹ and a relatively low affinity for CO₂ (K_m,c = 276 μM). Importantly, the Gallionella Rubisco displays as a homodimer, and the simple structure may be beneficial for heterologous expression. Another form II Rubisco with a relatively high k_cat,c of 16.4 s⁻¹ was discovered from the endosymbiont of a deep-sea tubeworm Riftia pachyptila (RPE Rubisco) [58]. RPE Rubisco can be readily expressed in E. coli BL21(DE3) without additional expression of chaperones and showed high carboxylation efficiency. However, the low specificity for CO₂ (K_m,c = 172.4 μM) prevented the RPE Rubisco from functioning properly under ambient atmospheric conditions. Collectively, these results suggested that more catalytically efficient Rubiscos can be found in form II Rubiscos, although they are less specific and require an elevated CO₂ concentration.

The oxygenation catalyzed by Rubisco produces by-products, which are usually recycled by carbon loss photorespiration. It is an alternative strategy to recycle the by-products through non-carbon loss or carbon fixation processes, by which CO₂ fixation efficiency of CBB cycle will be improved. The synthetic malyl-CoA-glycerate (MCG) cycle, relying on PEP carboxylase and malyl-CoA lyase to catalyze the cleavage of malyl-CoA to acetyl-CoA and glyoxylate, has been shown to enhance bicarbonate assimilation in cyanobacteria by approximately two-fold [67]. The PEP carboxylase is one of the most active oxygen-tolerant carboxylases with specific activity up to 150 μmol/min/mg, enabling fast carboxylation. Notably, glyoxylate is assimilated to C3 metabolites via tartronate semialdehyde without net carbon loss in the MCG pathway. Meanwhile, in the MCG cycle, the C3 metabolites produced in the CBB cycle were converted to acetyl-CoA through additional carbon fixation catalyzed by PEP carboxylase, which favors the generation of acetyl-CoA from CBB cycle products.

Through rational design, high-throughput microfluidics, and microplate screening, a novel glycolyl-CoA carboxylase with a k_cat,c of 5.6 s⁻¹ was designed recently based on the propionyl-CoA carboxylase from Methylorubrum extorquens [68]. The engineered enzyme displays comparable catalytic properties to natural carboxylases in CO₂ fixation. By the catalysis of the engineered glycolyl-CoA synthetase, a tartronyl-CoA reductase from Chloroflexus aurantiacus, and a semialdehyde reductase, glyoxylate produced in the oxygenation of Rubisco can be converted to glycerate in the so-called tartronyl-CoA pathway [68]. In the synthetic pathway, an additional carboxylation is catalyzed by the engineered glycolyl-CoA carboxylase, thus connecting the by-products of CBB cycle with the central carbon metabolism. Notably, when using the tartronyl-CoA pathway to recycle 2-phosphoglycolate, additional CO₂ is fixed rather than released, thus increasing the carbon efficiency of 2-phosphoglycolate recovery from 75% to 150% and decreasing ATP consumption by 21%. Moreover, the new pathway can be easily integrated into the CBB cycle for by-product recycling by the expression of three additional enzymes.

The low concentration of CO₂ in the atmosphere is not sufficient to sustain rapid carboxylation reactions of Rubiscos, especially those with less specificity. Carbon concentration mechanisms (CCM) exist in many photo-/chemo-autotrophic microorganisms and plants, providing Rubiscos with CO₂-rich conditions that promote carboxylation and inhibit deleterious oxygenation of RuBP [69]. Carbonic anhydrases (CAs) can generate CO₂ in cell by catalyzing the hydrolysis of bicarbonate and are key to the autotrophic growth of microorganisms [70,71]. Similarly, a biogenic polyamine has been reported to capture atmospheric CO₂, forming carbamates and enriching CO₂ for Rubiscos [72]. Furthermore, CO₂ is concentrated through a biophysical CCM in Synechococcus species. A bicarbonate uptake system is coupled with a virus-like proteinaceous shell named carboxysome that encapsulates Rubiscos, CAs, and other accessory proteins in a confined microcompartment, which can exclude O₂ and provide a CO₂-rich condition for Rubiscos to promote carboxylation and inhibit deleterious side reaction [73]. In addition, a biochemical CCM which relies on PEP carboxylase was found in C₄ plants. PEP carboxylase catalyzes CO₂ to malate, which is then oxidatively decarboxylated to concentrate CO₂ for Rubiscos [74]. As a stable carbon carrier, malate is suitable for long-distance transportation in multicellular plants. It can be speculated that transplantation of CCM from C₄ to C₃ plants such as rice is expected to increase photosynthesis and grain yield [75,76].

A functional carboxysome was recently constructed in a Rubisco-dependent E. coli strain. An engineered E. coli strain harboring form I Rubisco from Halothiobacillus neapolitanus and a prk gene from Synechococcus elongatus PCC 7942 was constructed, which can grow autotrophically with 10% CO₂. The engineered E. coli was able to grow in ambient atmosphere when the CCM of H. neapolitanus was co-expressed [77]. It wasproved that the large subunit of Rubisco determines whether the Rubisco complex can be encapsulated in the H. neapolitanus carboxysome [78]. The CsoS2 protein wasproposed as a scaffold protein that acts as an interaction hub to encapsulate Rubisco of H. neapolitanus [79]. These results suggested that the interaction between Rubisco and scaffold protein is essential for the formation of functional carboxysome, and foreign Rubiscos might not be incorporated into the carboxysome. Nonetheless, expression of a functional carboxysome in E. coli demonstrates the potential to increase carboxylation efficiency in microorganisms without CCM.

Increasing carboxylation capacity of Rubisco is key to improving photosynthesis, as it is the main entry for inorganic carbon into biosphere. In addition to strategies described above, the approaches such as direct evolution of Rubiscos, engineering large or small subunits of Rubisco, and changing expression and activation of Rubiscos can also improve CO₂ fixation [80].

Biotin-dependent carboxylases are an ancient group of carboxylases that carboxylate a variety of substrates, including acetyl-CoA, propionyl-CoA, pyruvate, and 2-oxoglutarate. These carboxylases play an important role in many essential metabolic pathways, including synthesis and degradation of fatty acids, degradation of certain amino acids, anaplerosis of TCA cycle intermediates, gluconeogenesis, and autotrophic fixation of CO₂ [81]. Acetyl-CoA carboxylase, propionyl-CoA carboxylase, pyruvate carboxylase, and 2-oxoglutarate carboxylase are involved in four of the above discussed six CO₂ fixation pathways. These enzymes are composed of at least three functional components: biotin carboxyl carrier protein (BCCP), biotin carboxylase (BC) domain, and carboxyl transferase (CT) domain, and share a common catalytic mechanism to catalyze a two-step reaction [82]. The BC domain catalyzes the ATP-dependent carboxylation of the biotin moiety of BCCP using bicarbonate as a carbon source, followed by the transfer of the carboxyl group from the carboxylated biotin at BCCP domain to the substrates determined by the CT domain.

The biotin-dependent carboxylases have several advantages over Rubiscos. Firstly, biotin-dependent carboxylases exhibit a higher carboxylation rate than Rubiscos (Table 1). In addition, O₂ does not interfere with the carboxylation by biotin-dependent carboxylases, and no oxygenation reaction would occur to impair the carboxylation efficiency. Moreover, considering substrate availability for biotin-dependent carboxylases, bicarbonate is more available than CO₂ in water, which is important for maintaining rapid carboxylation.

Efficient biotin-dependent carboxylases are promising for CO₂ fixation and production of valuable products. Recently, simultaneous -carbon fixation and succinate production was achieved in a recombinant E. coli through partially introducing 3HP bi-cycle [83]. The succinate synthetic pathway is centered on acetyl-CoA carboxylase and propionyl-CoA carboxylase, the carboxyltransferase domain of which was mutated. Through directed evolution, the k_cat,c value of the mutant enzyme is 13.3 s⁻¹, and the overall catalytic efficiency is improved by 94 times based on k_cat,c/K_m,c. As a result, the highest production of succinate reached 2.66 g/L with a CO₂ fixation rate of 0.94 mmol/L/h. However, the carboxylation and the subsequent reduction step require a substantial amount of ATP and NADPH (one succinate from acetyl-CoA requires 3 ATP and 3 NADPH), which will restrict the succinate yield [83,84]. In addition to the incomplete 3HP bi-cycle, pyruvate carboxylase, which plays an important role in anaplerotic generation of TCA intermediates, is also used for the production of dicarboxylate. Overexpression of pyruvate carboxylase from Rhizopus oryzae (RoPYC) was reported to enhance fumarate production in Saccharomyces cerevisiae, and further enhancement was achieved by mutation P474N in RoPYC, resulting in a 14% increase in PYC activity, and the final fumaric acid yield reached 314.5 mg/L [85]. Furthermore, the R458P mutation, which affects the allosteric and biotin carboxylation domain in RoPYC, was reported to further increase fumaric acid production with a maximal titer of 465.5 mg/L [86].

A unique member of the biotin-dependent carboxylases family, 2-oxoglutarate carboxylase (OGC), is predominantly identified in thermophilic bacteria from the rTCA cycle-dependent phylum Aquificae [36]. OGC catalyzes the carboxylation of 2-oxoglutarate to oxalosuccinate, which is further reduced to isocitrate. However, this process is accomplished by a single enzyme isocitrate dehydrogenase (ICDH) in the conventional rTCA cycle. Interestingly, ICDH normally catalyzes the oxidative decarboxylation of isocitrate in the oxidative TCA cycle. The reaction catalyzed by ICDH is characterized by reversibility and high specific activity. The recombinant ICDH from Chlorobium limicola showed a carboxylation rate of 27.2 μmol/min/mg at pH 7 and a decarboxylation rate of 160 μmol/min/mg at pH 9 [87,88]. Therefore, the carboxylation and decarboxylation reactions can be controlled by changing pH, showing great application prospects in CO₂ fixation and storage process.

Bifunctional carbon monoxide dehydrogenase/acetyl-CoA synthase (CODH/ACS) is a key enzyme of the rAC pathway in acetogens and methanogens. These enzymes typically consist of two functional subunits, CODH and ACS, which are associated with a heterotetrameric complex [89]. The CODH subunit catalyzes the reversible reduction of CO₂ to CO, which is delivered to the active sites of ACS through an internal gas channel, preventing the leak of toxic CO and confining it within CODH/ACS complex [89,90]. The ACS subunit catalyzes the reversible condensation of CO, a methyl moiety bound to a corrinoid iron-sulfur protein and coenzyme-A to acetyl-CoA. Some CODH/ACS variants have been reported to contain unsealed channels for CO oxidation to generate reducing equivalents [91]. The reduction of CO₂ is quite thermodynamically unfavorable, with a standard redox potential as low as –520mV, requiring highly reduced ferredoxin as an electron donor [29,92]. Therefore, this enzyme is mainly restricted to anaerobic environments.

Acetogens, which depend on CODH/ACS and the rAC pathway for energy generation and biomass accumulation, produce acetate as the end product by using CO₂, H_2, and toxic CO. The rAC pathway is one of the most energy-efficient carbon fixation pathways, and it is estimated that acetogens produce over 10¹³ kg acetate annually, demonstrating its potential for biomanufacturing [41]. Various acetogens have been engineered for the production of valuable products including acetate, ethanol, butanol, 2,3-butanediol, and acetoin, from syngas composed of CO₂, H₂, and CO [93,94,95,96].

In a recent study, the tolerance of Eubacterium limosum ATCC 8486 to CO increased up to 80% through adaptive evolution. The resulting strain displayed a faster growth rate and produced 42.28 mM/gCDW acetate under 40% CO condition [96]. Moreover, this strain was engineered to produce acetoin with a production of 19.6 mM/gCDW, a 1.34–fold increase compared with the parental strain. An engineered E. limosum KIST612 overexpressing CODH, ACS, and a coenzyme CooC2 exhibited a 3.1-fold increase in the rate of CO oxidation and a 1.4-fold increase in acetate production to 13.9 g/gCDW acetate [97]. The results suggest that acetogens are promising for biosynthesis by utilizing CO and CO₂ [98].

Three of the six natural carbon fixation pathways use enzymes from 2-oxoacid (2-keto acid):ferredoxin oxidoreductase (OFOR) family to catalyze the reductive carboxylation. Two members of OFOR enzymes, PROF and OGOR, are involved in the carboxylation of acetyl-CoA and succinyl-CoA to pyruvate and 2-oxoglutarate, respectively. Ferredoxins are a group of small proteins containing iron-sulfur clusters with mid-potentials between −400 mv and −500 mV. They are required as the electron donor to drive the thermodynamically and energetically challenging reductive carboxylation of acetyl-CoA and succinyl-CoA, the standard redox potentials of which are about −500 mV [29,99]. OFOR reactions are typically reversible and can operate in both fixation and release of CO₂, mainly depending on the redox potential of co-factors [100,101].

Diverse oligomeric states of OFOR enzymes have been discovered, including homodimers, heterotetramers, and heterohexamers [102,103]. OFOR enzymes share a similar domain arrangement despite different domains constitution and contain up to seven different domains, several of which are conserved for substrate and cofactor binding, and housing iron-sulfur cluster(s) for electron transfer [104]. Notably, most OFOR enzymes are highly oxygen-sensitive and rapidly and irreversibly inactivated upon exposure to air, constraining the corresponding autotrophic pathways in a strictly anaerobic or microaerobic environment [30]. In addition, to drive the thermodynamically and energetically challenging process, ferredoxins need to be kept highly reduced by sodium (or proton) motive force driven oxidoreductases or electron bifurcation mechanism [29,99].

The oxygen-tolerant PFOR from the extremophilic crenarchaeon Sulfolobus acidocaldarius was proposed to catalyze the decarboxylation of pyruvate. Interestingly, the PROF exhibited carboxylase activity in vitro when coupled with low-potential ferredoxins or methylviologen (with standard redox potentials of less than −500 mV and −446 mV, respectively) as electron donors [104]. Besides using ferredoxin or methylviologen as electron donors, cadmium sulfide nanorods were reported to act as electron donors to drive the reductive carboxylation of succinyl-CoA to 2-oxoglutarate catalyzed by OGOR from Magnetococcus marinus MC-1 [62]. Notably, cadmium sulfide nanorods can be reduced by photo-excitation and exhibited a standard redox potential of ≤−700 mV. Generally, OFOR from the three anaerobic carbon fixation pathways, rTCA, rAcCoA, and DC/HB, require fully reduced ferredoxin to catalyze reductive carboxylation. Even though reduced ferredoxin is difficult to achieve outside native cells, artificial electron donors can be used as an alternative in vitro, expanding their application for biosynthesis.