A
Reprogramme the E. coli metabolism by engineering a functional carbon-fixation pathwayYuChen1,2
AdamBurke2
VincentChriscoli1
MengruYang1
PingChang1
TianpeiLi1
BukeZhang3
RoystonGoodacre2✉Emailroy.goodacre@liverpool.ac.uk
Lu-NingLiu1,3✉Emailluning.liu@liverpool.ac.uk
1
A
Department of Biochemistry, Cell and Systems Biology, Institute of Systems, Molecular and Integrative BiologyUniversity of LiverpoolL69 7ZBLiverpoolUnited Kingdom2Centre for Metabolomics Research, Department of Biochemistry, Cell and Systems Biology, Institute of Systems, Molecular and Integrative BiologyUniversity of LiverpoolL69 7ZBLiverpoolUnited Kingdom
3College of Marine Life Sciences and Frontiers Science Center for Deep Ocean Multispheres and Earth SystemOcean University of China266003QingdaoChina
Yu Chen1,2, Adam Burke2, Vincent Chriscoli1, Mengru Yang1, Ping Chang1, Tianpei Li1, Buke Zhang3, Royston Goodacre2*, Lu-Ning Liu1,3*
1Department of Biochemistry, Cell and Systems Biology, Institute of Systems, Molecular and Integrative Biology, University of Liverpool, Liverpool L69 7ZB, United Kingdom
2Centre for Metabolomics Research, Department of Biochemistry, Cell and Systems Biology, Institute of Systems, Molecular and Integrative Biology, University of Liverpool, Liverpool L69 7ZB, United Kingdom
3College of Marine Life Sciences and Frontiers Science Center for Deep Ocean Multispheres and Earth System, Ocean University of China, Qingdao 266003, China
*Correspondence: luning.liu@liverpool.ac.uk (L.-N.L.); roy.goodacre@liverpool.ac.uk (R.G.)
A
Abstract
Background
Rising atmospheric CO₂ levels and their impact on climate change have intensified the need for innovative carbon capture and fixation strategies. The Calvin-Benson-Bassham (CBB) cycle, a central metabolic pathway in all photoautotrophic organisms and many autotrophic bacteria, plays a pivotal role in global carbon assimilation but is limited by the low catalytic efficiency of Rubisco.
Results
Here, we engineered a complete, functional CBB cycle in Escherichia coli, by heterologously expressing up to 13 genes encoding phosphoribulokinase, α-carboxysomes, and inorganic carbon pumps. This bioengineering approach allowed E. coli to utilize atmospheric CO2 and led to upregulation of sugars such as ribose (4.94-fold) and xylitol (8.94-fold). Detailed metabolomic profiling of central carbon metabolism using gas chromatography-mass spectrometry (GC-MS) demonstrated that installation of the CBB cycle has a notable impact on the metabolic landscape of E. coli, resulting in substantial alterations in central carbon and amino acid metabolism. These findings deepen our understanding of the natural biological carbon-fixation pathway and its engineering in heterotrophic hosts. Furthermore, this work provides a versatile platform for evaluating and selecting efficient carbon-fixation modules, as well as assessing metabolic bottlenecks in engineered systems.
Conclusion
These advances offer practical guidance for rational metabolic engineering in diverse organisms for biotechnological applications, including carbon sequestration, sustainable bioproduction, and crop improvement.
Keywords
Carbon fixation
CO2-concentrating mechanisms
E. coli
metabolic profiling
synthetic biology.
Introduction
The greenhouse effect is currently leading to global warming on Earth, which in turn raises worldwide issues like climate change, ecosystem disruptions, and more frequent extreme weather events (1, 2). As CO₂ emissions continue to rise, achieving carbon reduction and neutrality has become a shared priority for people around the world. In this context, autotrophic carbon fixation pathways are crucial, as they transform CO2 into organic carbon and play a significant role in efforts to lower atmospheric carbon levels (3, 4).
The Calvin-Benson-Bassham (CBB) cycle is a natural carbon fixation pathway prevalent across diverse organisms, making a substantial contribution to global carbon assimilation (3, 5). Ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco), the most abundant protein on Earth, is the central enzyme in the CBB cycle, accounting for ~ 95% of inorganic carbon (Ci) fixation (6, 7). However, Rubisco has evolved as an insufficient enzyme with slow carboxylation rates and low specificity between CO2 and O2, leading to low CO2-fixation efficiency (7, 8, 9). To address these limitations, natural organisms have developed CO2-concentrating mechanisms (CCMs) to elevate CO2 levels around Rubisco, thereby enhancing carboxylation and suppressing oxygenation (10). In cyanobacteria and some chemoautotrophs, the CCM employs Ci import systems and specialised proteinaceous microcompartments called carboxysomes (11, 12). The Ci pumps accumulate bicarbonate within the cytoplasm (13, 14). The carboxysome encapsulates two key enzymes: Rubisco and carbonic anhydrase (CA) (15, 16). Accumulated bicarbonate can pass across the semi-permeable carboxysome shell and is then converted to CO2 by CA, generating a high concentration of CO2 around Rubisco to enhance its carboxylation activity (7, 17, 18).
Apart from natural organisms that are capable of performing carbon fixation, engineering functional carbon-fixing pathways into model heterotrophic hosts provides a powerful approach for unravelling the natural process (19, 20) and developing advanced carbon-fixing systems with potential applications in carbon capture and bioproduction of valuable chemicals (21, 22, 23, 24). The inherent pentose phosphate pathway (PPP) in heterotrophs contains almost all the enzymes and catalytic steps required for a CBB cycle, except for two key enzymes: phosphoribulokinase (Prk), which catalyzes the phosphorylation of ribulose-5-phosphate (Ru5P) to ribulose-1,5-bisphosphate (RuBP), and Rubisco, which fixes CO₂ onto RuBP to form 3-phosphoglycerate (3PG). Incorporation of these two enzymes has enabled the reconstitution of a functional CBB cycle in organisms like Escherichia coli and yeast (19, 25, 26, 27). When further supplemented with formate dehydrogenase and optimized via laboratory evolution, engineered E. coli have demonstrated the capacity to generate biomass from CO2 under high CO2 conditions (19). Moreover, Flamholz, Dugan (28) demonstrated the feasibility of engineering carbon fixation in E. coli by integrating the CBB cycle with a carboxysome-based CCM, including a Ci pump (DabA1/B1) and α-carboxysomes. Despite these advances, constructing efficient carbon-fixation pathways requires optimization of the catalytic activities of introduced components and their functional coordination in new hosts (25, 29, 30). Additionally, comprehensive metabolic profiling, in addition to phenotypic and functional analyses, is essential for examining the metabolic effects of installing the CBB cycle and CCMs and identifying bottlenecks to guide further optimization of metabolomic pathway engineering (31, 32, 33).
In this study, we engineered a functional CBB cycle into E. coli through incorporation of a bacterial CCM, composed of α-carboxysomes and Ci pumps, along with Prk. This platform allowed us to assess systematically the activities of various Ci pumps in the engineered carbon-fixation system. Using gas chromatography-mass spectrometry (GC-MS), we conducted a comprehensive metabolic profiling analysis of the engineered E. coli strains, whereby GC-MS using methoxymation and silylation derivatisation chemistry allows central carbon and nitrogen metabolism to be readily assessed. Our results revealed distinct metabolic shifts triggered by the installation of the carboxysome-based CCM and the CBB cycle. Our findings offer valuable insights into the bioengineering of functional bacterial CCM and the CBB cycle in non-native hosts, and may inform future strategies to optimise biological carbon fixation in engineered heterotrophs to boost carbon assimilation and production of high-value chemicals.
Results and Discussion
Installation of active Ci pumps
To construct a functional CCM, we first deleted the endogenous CA-encoding gene (can) from the chromosome of E. coli to generate a high CO2-requiring E. coli mutant (ΔC), in which intracellular bicarbonate must be complemented through a higher environmental CO2 level or additional Ci pumps (13, 34) (Fig. 1a). Using this ΔC mutant, we assessed the expression and activities of four different Ci-import systems, including DabA1/B1 and DabA2/B2 from Halothiobacillus neapolitanus (35), as well as BicA and SbtA from Synechocystis sp. PCC 6803 (13, 14). Functional Ci pumps can supply alternative CO2/bicarbonate sources, which are expected to restore the growth of the ΔC mutant under ambient air conditions (13, 34), containing ~ 0.04% CO2.
A
Heterologous expression of these individual Ci importers fused with Hexahistidine tags (6xHis-tags) was verified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblot analysis (Fig. 1b, Fig. S1). Growth assays revealed that the ΔC mutant expressing DabA2/B2 or SbtA exhibited growth under ambient air conditions (Figs. 1c, 1d), indicating the robust Ci-importing activities of engineered DabA2/B2 and SbtA. Notably, the DabA2/B2-expressing strain exhibited a higher growth rate than the SbtA-expressing mutant (Figs. 1c, 1d), suggesting that DabA2/B2 mediates more efficient Ci import than SbtA in the engineered E. coli background. In contrast, mutants expressing DabA1/B1 or BicA were unable to survive under ambient air conditions (Figs. 1c, 1d), although DabA1/B1 and BicA function as Ci pumps in H. neapolitanus and Synechocystis 6803, respectively (14, 35, 36). This suggests their inadequate (or low) Ci-importing activities needed to meet the bicarbonate requirements of the ΔC mutant. These results are consistent with previous studies indicating that DabA2/B2 exhibited the highest Ci-importing activity relative to DabA1/B1 and SbtA (35), whereas engineered BicA was ineffective in restoring the growth of high CO2-requiring strains (34).Integration of metabolic modules to generate a carboxysome-based CCM and a CBB cycle
To establish a functional CBB cycle, we first generated a plasmid pCBPRK to express both α-carboxysomes and Prk in E. coli (Fig. 2a), which was confirmed by gene sequencing. As a control, we co-expressed Rubisco derived from H. neapolitanus and Prk using the plasmid pRubPRK. Expression of these components was validated by SDS-PAGE and immunoblot analysis (Fig. S2).
The growth inhibitory phenotype caused by expression of Prk alone was assessed to validate the functionality of recombinant α-carboxysomes or Rubisco. Specifically, expression of Prk alone directs the carbon metabolism pathway towards RuBP, creating a metabolic dead-end that suppresses host strain growth (Fig. S3). In contrast, a functional CBB cycle (catalysed by Rubisco) could rescue this inhibition through RuBP carboxylation to 3PG (Fig. S3) (21, 37). When grown in M9 medium supplemented with 0.2% glycerol, the E. coli variant expressing Prk only (BL:P) exhibited a substantially prolonged lag phase relative to the BL:Vb control that expresses empty pBAD33 vector (Figs. 2b, 2c). In contrast, the BL:Rp variant co-expressing Rubisco and Prk displayed enhanced growth with a shorter lag phase; co-expressing α-carboxysomes with Prk (BL:Cp) further improved cell growth (Figs. 2b, 2c). These results confirm the functionality of the heterologously expressed enzymes and highlight the significance of carboxysomes as part of the CCM in enhancing CO2 fixation. Interestingly, the strain expressing α-carboxysomes only (BL:C) exhibited extended lag phase compared to BL:Vb, which may be attributed to the metabolic burden imposed by α-carboxysome biogenesis.
To build a functional carboxysome-based CCM, we expressed DabA2/B2 in BL:Cp, along with DabA1/B1, BicA, and SbtA as alternative Ci pumps (Fig. 2b). Growth analysis revealed consistently superior performance across all carboxysome/Prk-expressing derivatives (including variants BL:CpVe, BL:CpT1, BL:CpT2, BL:CpTb, and BL:CpTs) compared to those lacking these two components (BL:Veb, BL:VbT1, BL: VbT2, BL: VbTb, and BL: VbTs) (Figs. 2b, 2d). Notably, no significant difference in growth was detected among strains expressing DabA2/B2, DabA1/B1, BicA or SbtA alongside α-carboxysomes and Prk (Fig. 2d), despite their different Ci-importing activities determined in the ΔC mutant (Fig. 1d). The results suggest that additional Ci-transport capacity cannot further enhance the engineered CBB cycle in E. coli under these conditions, potentially due to substrate saturation or competing metabolic limitations.
Metabolic landscape of the E. coli strains possessing a functional CBB cycle
A
A
To gain an understanding of how these engineering modification affects the metabolic landscape of E. coli, we performed metabolic profiling on the engineered E. coli strains expressing different combinations of metabolic modules using GC-MS (Table S1). GC-MS was chosen as methoxymation and silylation derivatisation allows central carbon and nitrogen to be probed, in addition to other metabolic pathways and networks. This approach enabled the detection of 257 metabolic features (Supplementary File 1) and the identification of 55 metabolites (Table S2): 12 to Level 1 and 43 to Level 2 of the Metabolomics Standards Initiative (38).Principal Component Analysis (PCA) score plots revealed discrete clustering distributions of the engineered strains, indicating that the expression and functions of engineered proteins/assemblies resulted in metabolic changes and fundamentally restructures E. coli host metabolism (Fig. 3a). The distance between clusters represents the degree of their metabolic variance. Cluster A contains the strains co-expressing α-carboxysomes and Prk, along with Ci pumps (including BL:CpVe, BL:CpT1, BL:CpT2, BL:CpTs, and BL:CpTb), which are distinct from the variants expressing only Ci pumps without α-carboxysomes and Prk (BL:VbT1, BL:VbT2, BL:VbTb, and BL:VbTs), which formed Cluster B. The BL:Veb variant without expression of any recombinant proteins, was distributed separately from both Clusters A and B (Figs. 2b, 3a), confirming that expression of these modules altered the metabolic landscape significantly. Cluster A exhibited maximal separation from BL:Veb, indicating that the expression of α-carboxysomes and Prk resulted in profound metabolic changes. In contrast, Cluster B exhibited remarkably closer proximity to BL:Veb than to the BL:CpVe strain expressing only α-carboxysomes and Prk, revealing that the engineered Ci pumps alone induced comparatively modest perturbations. This hierarchy aligns with functional priorities: the CBB cycle is directly involved in the inherent PPP for carbon metabolism, whereas Ci pumps improve bicarbonate availability in the cytosol to enhance carboxylation kinetics and pH homeostasis without redirecting central carbon flux.
As the pronounced metabolic shifts induced by the biosynthesis of both α-carboxysomes and Prk could potentially mask subtler differences among strains expressing various Ci pumps within the same PCA clusters (Fig. 3a), we subdivided our experimental samples to resolve these effects better. Specifically, we analyzed two separate groups: one comprising all strains harboring both α-carboxysomes and Prk (Cluster A), and another including strains lacking α-carboxysomes and Prk (Cluster B and BL:Veb). We then applied PCA and principal component-discriminant function analysis (PC-DFA) to each group independently (Figs. S4, S5). In PC-DFA, one uses an a priori class structure, where the aim of PC-DFA is to minimise the variance within a group and maximise variation between groups. This was performed on the biological replicates, so it is semi-supervised in nature. PC-DFA was validated by projecting samples (n = 1 biological replicate) not used to calibrate the model into the resulting PC-DFA ordination plot from the training data (n = 6 biological replicates). Our results showed no significant differences in metabolic profiles among strains expressing different Ci pumps, consistent with the findings on growth profiles of E. coli expressing various Ci pumps (Fig. 2d) and PCA score plots for all the strains (Fig. 3a). This suggests a largely similar metabolic influence of the engineered Ci pumps under these conditions. The lack of differentiation may reflect the limited bicarbonate requirement in wild-type E. coli, where endogenous CA activity may be sufficient to provide bicarbonate for cellular needs (39). Although engineered Ci pumps can facilitate Ci import and elevate the cytosolic bicarbonate levels, excess bicarbonate may suppress their own activities (40). In contrast, the ΔC mutant without endogenous CA exhibited high bicarbonate demand, maximizing the activities of Ci pumps and thus unmasking functional differences among Ci pumps (Fig. 1d). These findings highlight the role of endogenous CA in modulating the functionality of the installed CBB cycle in E. coli.
Enhanced sugar accumulation in the E. coli host induced by the CBB cycle
A
A
Since no substantial differences in growth and metabolic profiles were determined among strains expressing various Ci pumps, we focused on DabA2/B2 as the representative Ci pump for subsequent analysis of the corresponding variants (BL:Veb, BL:VbT2, BL:CpVe, BL:CpT2). Analysis of variance (ANOVA) revealed that 134 out of all 257 detected metabolite features showed statistically significant (false discovery rate (FDR) corrected P-value threshold < 0.05) differences among the engineered E. coli variants (Fig. 3b, Table S3). From these features, we annotated 30 metabolites (Table S4). To identify and quantify the metabolites that were specifically affected by the introduction of α-carboxysomes, Prk, and DabA2/B2, we performed pairwise comparisons of the variants with the 257 features using t-tests coupled with fold change (FC) analysis. The statistical significance was determined using FDR corrected P-value with threshold of < 0.05, and Log2(FC) > 1 (upregulated) or < -1 (downregulated). Comparing with the negative control with only empty plasmids (BL:Veb), fourteen metabolites were upregulated in the presence of both α-carboxysomes and Prk, including ribose (4.94-fold), xylitol (8.94-fold), specific amino acids (e.g. aspartate: 2.30-fold, and cystathionine: 76.55-fold), and a dipeptide, cysteinylglycine (32.85-fold) (Figs. 4a, 4d, Fig. S6, Table S4). Integration of the CBB cycle provides an additional carbon assimilation route via the pentose phosphate pathway, channelling Ru5P to 3PG through RuBP and Rubisco-mediated CO2 fixation (Fig. 5). These metabolic changes were consistent with the enhanced growth determined for strains expressing α-carboxysomes and Prk (Fig. 2c), suggesting an accelerated carbon flux.In the global metabolic pathways of E. coli (Fig. 5), ribose and xylitol are synthesised from ribose 5-phosphate (R5P) and xylulose 5-phosphate (xylulose-5P), respectively. These compounds serve as precursors for Ru5P generation, which is subsequently converted to RuBP for carbon fixation via Rubisco. The upregulation of these two five-carbon sugars might be attributed to the coordination of the CBB cycle and the PPP and a rerouting of carbon flux between the two pathways (Fig. 4a), which could potentially maintain R5P homeostasis (Fig. 3b) and modulating carbon fixation by tuning substrate availability (26). Additionally, the activity of carboxysome-encapsulated CA is regulated by RuBP (41) which may suppress the conversion of bicarbonate to CO2, and eventually modulate the efficacy of Ci pumps in facilitating carbon fixation and cell growth (Fig. 2c).
Our analysis revealed that expression of DabA2/B2 alone led to increased levels of 14 metabolites (including lactate, pyruvate, cis-aconitate, 2-oxoglutarate) and downregulation of ribose (0.03-fold) and nicotinate (0.25-fold) (Figs. 4b, 4d, Fig. S6, Table S4). The upregulation of lactate (25.79-fold) suggests ongoing fermentation processes and excess NADH in E. coli (42), as lactate production serves to reoxidize NADH during glycolysis (43). Upregulation of cis-aconitate (2.56-fold) and 2-oxoglutarate (4.39-fold) may result from feedback inhibition in the tricarboxylic acid (TCA) cycle, where high NADH levels inhibit their conversion to fumarate, a process that also generates NADH and CO2 (Fig. 5). This may account for the increased levels of cis-aconitate and 2-oxoglutarate observed in strains expressing DabA2/B2 alone. Functional DabA2/B2 increases the cytosolic bicarbonate pool and intracellular CO2 levels, creating a catalytically favourable environment to stimulate fermentative metabolism.
Notably, the incorporation of the CBB cycle into DabA2/B2-expressing strains led to a substantial downregulation of metabolites that were upregulated with DabA2/B2 alone (Figs. 4c, 4d; Table S4). The lactate content was reduced (0.01-fold) in strains expressing α-carboxysomes and Prk (Fig. 4c, d), consistent with enhanced NADH consumption by the Rubisco-catalysed CBB cycle (44). The NADH consumption may also alleviate feedback inhibition, resulting in lower levels of cis-aconitate (0.24-fold) (Fig. 4c, d; Table S4). Consistently, a decrease in 2-oxoglutarate (Fig. 4d) and an increase in fumarate (Fig. S6) were observed, indicating a broader metabolic shift resulted from the engineered CBB cycle.
In summary, installation of the CBB cycle and carboxysome-based CCM restructured the metabolic landscape, leading to the upregulation of several sugars and intermediates that could complement the enhanced growth profile associated with the CBB cycle. The increased abundance of ribose and xylitol suggests altered carbon flux exiting the PPP, highlighting a potential strategy to direct carbon flux through the engineered CBB cycle by targeted suppression of the enzymes converting xylulose-5P and R5P to xylitol and ribose, respectively.
Network-level metabolic shifts induced by the CBB cycle and carboxysome-based CCM
Interestingly, in addition to the 55 annotated metabolites, several features among the 202 detected metabolites also exhibited variations in response to the engineering modifications. To assess in detail the metabolic impact of introducing the CBB cycle alongside a carboxysome-based CCM, we performed correlation analysis on all 257 metabolites (Supplementary File 1) and identified four distinct correlation clusters (Fig. 6). Cluster I includes 22 highly positively correlated features, including ribose, xylitol, and gentisate, which were upregulated in the presence of the carboxysome and Prk (Fig. 4a). Cluster II, the largest group, comprises 153 positively correlated features such as ethanolamine, aspartate, glucose-6-phosphate (glucose-6P), deoxythymidine 5’-diphosphate (dTDP), and 14 other annotated metabolites that were significantly changed by the incorporation of the Prk, carboxysome and DabA2/B2. Among them, 10 were remarkably upregulated in response to the installation of the CBB cycle (Figs. 4a, 6). Cluster III contains 49 metabolites which were positively correlated. Seven of them, including lactate, pyruvate, and cis-aconitate, were significantly upregulated in the strain expressing only DabA2/B2 (Fig. 4b). Cluster IV showed a negative correlation of the metabolites in Cluster I and Cluster III. This is in line with the discussion above, where the incorporation of the CBB cycle relief the fermentation processes by consuming cytosolic CO2 and NADH.
A
Pathway analysis using the 30 alternated metabolites also revealed an extended impact on 47 metabolic pathways (Fig. S7, Table S5). Among these, six pathways in the host showed significant alterations (FDR corrected P-value < 0.05), including: (1) alanine, aspartate and glutamate metabolism, (2) arginine biosynthesis, (3) TCA cycle, (4) glyoxylate and dicarboxylate metabolism, (5) starch and sucrose metabolism, and (6) pyruvate metabolism.Collectively, both metabolic profiling and correlation analysis demonstrate that incorporating a functional CBB cycle with a carboxysome-based CCM into E. coli reprogrammed the metabolic landscape of the host cells.
Conclusions
In this work, we engineered a complete, functional CBB cycle in E. coli, by incorporating α-carboxysomes and Ci pumps derived from H. neapolitanus, as well as cyanobacterial Prk, to enhance carbon assimilation in heterotrophic organisms. Among the four tested Ci pumps, DabA2/B2 exhibited the highest Ci-import activity. The installation of α-carboxysomes improved carbon fixation in the engineered strains. Comprehensive metabolic profiling revealed distinct metabolic adaptations in response to the engineered CBB cycle and CCM system. This allowed us to identify the upregulation of xylitol and ribose as potential metabolic bottlenecks amenable to future engineering. While a deeper understanding of these metabolic shifts is essential for elucidating the mechanistic basis for the observed phenotypes, our findings provide valuable guidance for the rational optimization of artificial carbon-fixation modules using synthetic biology and lay the framework for targeted metabolic engineering to be leveraged for diverse biotechnological applications, such as carbon capture and storage, crop improvement, and the production of biofuels and green chemicals.
Materials and Methods
Bacteria and cultivation conditions
A
The E. coli strain BL21(DE3) was utilised as a platform for expressing the recombinant proteins and integrating the CBB cycle. Cultures were grown aerobically in lysogeny broth (LB) medium (10 g L− 1 tryptone (Apollo Scientific, BIT1332), 5 g L− 1 yeast extract (Millipore, 1.03753), and 10 g L− 1 NaCl (Fisher Scientific, 10735921)) or on LB agar (LB with 1.5% (w:v) agar (Fisher Scientific, 10548030)) plates. The high CO2-dependent mutant strain was maintained under 10% CO2 conditions in a CO2 incubator (MCO-18AC, PHCbi, USA). Antibiotics were added to the culture medium or agar plates at final concentrations of 50 µg mL− 1 chloramphenicol (Sigma-Aldrich, C0378) and 50 µg mL− 1 kanamycin (Melford, K22000), as required. Protein expression was induced with appropriate inducers at final concentration of 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG; Melford, 2946-I56000) or 1 mM arabinose (Melford, A51000). Minimal medium (M9; 47.8 mM Na2HPO4, 22.0 mM KH2PO4, 18.7 mM NH4Cl, 8.7 mM NaCl, 0.010 mM CaCl2, 0.100 mM MgSO4, 0.005 mM MnCl2, 0.012 mM ZnCl2, 0.003 mM CuCl2, 0.003 mM CoCl2, 0.002 mM Na2MoO4), containing glycerol as the sole carbon source, was supplemented with the appropriate antibiotics and inducers to illustrate the growth profile of the variants.Construction of expression plasmids for recombinant proteins
Genes encoding DabA1/B1 and DabA2/B2 were cloned from H. neapolitanus (NC_013422.1) (35), while genes encoding BicA and SbtA were cloned from Synechocystis sp. PCC 6803 (NC_000911.1). These genes were individually inserted into linearized vector pET28a, generating the corresponding expression plasmids (Fig. 1a). Hexahistidine tags (6xHis-tag) were attached to the recombinant proteins for validation of their expression through immunoblot analysis.
The α-carboxysome expression plasmid (pCB) (Fig. 2a) was acquired from a previous study (20). In brief, all ten genes encoding the components of α-carboxysomes from the chemolithoautotrophic proteobacterium H. neapolitanus (45), namely, cbbL, cbbS, csoS2, csoSCA, csoS4A/B, and csoS1A/B/C/D, were amplified from the pHnCBS1D (Plasmid #52065, Addgene, UK), and assembled into the pBAD33 vector amplicon (46).
The prk gene was first cloned from Synechococcus elongatus PCC 7942 (GenBank: CP000100.1), and assembled into the pBAD33 backbone with a 6xHis-tag at its N-terminus. Then, it was amplified, along with the 6xHis-tag, and inserted downstream of csoS1D on pCB, resulting in the co-expression plasmid pCBPRK. All eleven genes, comprising ten for α-carboxysomes and one for Prk, were regulated by the araBAD promoter in this construct. A Rubisco and Prk co-expression plasmid (pRubPRK) was generated by removing the genes for the α-carboxysome shell components and the encapsulated CA from pCBPRK (Fig. 2a).
Generation of the high CO2-requiring mutant
The endogenous CA encoding gene, can, was eliminated from the chromosome of BL21(DE3) using the I-Scel mediated genomic editing method (47, 48), generating the high CO2-requiring mutant for evaluating the activity of Ci pumps. Briefly, two homologous arms (~ 500 bp each) flanking the can gene were amplified from the wild-type BL21(DE3) and inserted into the helper plasmid pEMG between two I-Scel cleavage sites, creating the plasmid pEMG-Δcan. This plasmid was then transformed into the helper strain S17-1λpir vis electroporation for selection and reproduction, followed by conjugation with the target strain BL21(DE3). After recombining the helper plasmid into the chromosome of BL21(DE3), the target gene, along with the recombined helper plasmid, was cleaved from the chromosome by the I-SceI endonuclease, resulting in the mutant lacking the can gene (ΔC). The desired ΔC mutant was selected and confirmed using PCR analysis and growth profiling. The helper plasmids were later eliminated from the mutant by cultivating it in an antibiotic-free medium.
Growth profile measurement
Growth curves were measured using a growth profiler (GP960, EnzyScreen, Netherlands). Seeding cells were harvested and washed after overnight incubation in LB medium supplemented with the appropriate antibiotics and inducers. The initial optical density at 600 nm (OD600) of the culture was adjusted to 0.1 absorption unit (a.u.), and 250 µL of the culture was inoculated into a 96-well plate, with 6 replicates (or 5 if specified) for each colony. The plate was incubated at 37°C under ambient air conditions with continuous shaking at 225 revolutions per minute (rpm).
GC-MS metabolic profiling
Samples for metabolic profiling were prepared following a method adapted from a previous study (49). Cultures of the variants, with 7 replicates each, were harvested at mid-log phase and quenched immediately by mixing 15 mL of culture with 30 mL of cold (-48°C) aqueous methanol (methanol:water, 3:2). The cells were immediately pelleted via centrifugation at -9°C, 4,800 g for 10 min, and then stored at -80°C. Intracellular metabolites were extracted from the cell pellets using 1 mL extraction solution (methanol:water, 4:1, -48°C) through three flash-freeze and thawing cycles. After removing cell debris via centrifugation at 21,000 g for 10 min at -9°C, the supernatant was collected and normalized based on the OD600 measured at the time of sample collection. Pooled quality control (QC) samples were created by mixing portions from each sample (50). All samples were spiked with a mixture of deuterated chemicals as internal standards, and the solvents were removed using a vacuum concentrator (Vacufuge Vacuum concentrator 5301, Eppendorf, USA) for a minimum of 4 h to pellet the metabolites.
The extracted metabolites were derivatized using a two-step derivatization approach adapted from Begley, Francis-McIntyre (51). Methoxylamine hydrochloride (A0415450, Acros Organics) was dissolved in dry pyridine (20 mg mL− 1), and 50 µL was added to each dried metabolite pellet. The mixtures were then incubated at 65°C for 40 min. Afterwards, 50 µL of N-Methyl-N-trimethylsilyltrifluoroacetamide (MSTFA; TS-48913, Thermo Scientific, USA) was added to each sample, followed by another 40 min incubation at 65°C. A solution containing five alkane chemicals, used as retention indexes, was spiked into each sample before centrifugation at 13,500 g for 15 min (52). Subsequently, 100 µL of the supernatant, avoiding any solid residues, was transferred to GC vials (5190–9590, Agilent, USA) and ready for GC-MS analysis.
GC-MS analysis was carried out using an Agilent GC-MS system (8890 GC system tended with 7250 GC-qToF-MS, Agilent, USA) and an HP-5ms (30 m, 0.25 mm, 0.25 µm) capillary column (Agilent, USA), following methods adapted from a previous study (Dunn et al., 2011). One microliter of the sample was injected into the inlet (280°C) and processed to the column under a split ratio of 20:1. Helium gas was used as the carrier gas, maintained at a constant flow rate of 1 mL min− 1. The oven temperature was initially set to 70°C and held for 4 min, then ramped to 300°C at a rate of 20°C per minute and held for an additional 4 min. Following each injection, the oven temperature was reduced from 300°C to 70°C in preparation for the next sample. The mass spectrometer operated in electron ionisation (EI) mode with an electron energy of 70 eV. The detector acquired signals across a m/z range of 45–600 at a rate of 20 Hz. The raw data was converted to ‘*.mzML’ format using the MSConvert (53), and then deconvoluted using MS-DIAL (54). The Golm Metabolome Database (GMD) (55) and a combined database from the MS-DIAL project (Library with all records using Kovats retention index) (52, 54, 56) were employed to annotate the metabolic results. Standard chemicals were also utilised to provide a level I annotation for some metabolites, following the Metabolomics Standards Initiative Guidelines (38).
Statistical analysis
The peak intensity of the metabolite features was logarithmically (log10) transformed and then autoscaled, where each feature is first mean-centred and then divided by its standard deviation. Principal component analysis (PCA) and principal component-discriminant function analysis (PC-DFA) (57, 58, 59) were performed on this pre-processed dataset using MATLAB (MatLabR2023b, MathWorks). Analysis of Variance (ANOVA) and correlation analysis were conducted using MetaboAnalyst (60, 61). In order to compensate for multiple testing and reduce the FDR, Benjamini-Hochburg procedure (62) adjusted P-values with threshold of < 0.05 were used to determine the statistical significance. Pearson correlation analysis was employed to assess the distance in correlation patterns among metabolites.
Pathway analysis
Pathway analysis was carried out in MetaboAnalyst (60, 61), with E. coli K-12 MG1655 (KEGG) as the reference organism. To visualise the connections between the identified metabolites, a pathway map was generated based on the metabolic pathways of BL21(DE3) in KEGG (63). The streptomycin biosynthesis pathway was removed from the result since only a few shared reactions can be fulfilled in E. coli BL21(DE3), though it was identified as significantly affected.
List of abbreviations
CO2
Carbon dioxide
CBB cycle
Calvin-Benson-Bassham cycle
GC-MS
Gas chromatography-mass spectrometry
Rubisco
Ribulose 1,5-bisphosphate carboxylase/oxygenase
Ci
Inorganic carbon
CCM
CO2-concentrating mechanism
CA
Carbonic anhydrase
PPP
Pentose phosphate pathway
Prk
Phosphoribulokinase
Ru5P
Ribulose-5-phosphate
RuBP
Ribulose-1,5-bisphosphate
3PG
3-Phosphoglycerate
6xHis-tags
Hexahistidine tags
SDS-PAGE
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis
PCA
Principal Component Analysis
PC-DFA
Principal component-discriminant function analysis
ANOVA
Analysis of variance
FDR
False discovery rate
FC
Fold change
R5P
Ribose 5-phosphate
Xylulose-5P
Xylulose 5-phosphate
TCA
Tricarboxylic acid
Glucose-6P
Glucose-6-phosphate
dTDP
Deoxythymidine 5’-diphosphate
IPTG
Isopropyl β-D-1-thiogalactopyranoside
a.u.
Absorption unit
OD600
Optical density at 600 nm
rpm
Revolutions per minute
QC
Quality control
PRPP
5-Phosphoribosyl diphosphate
Declarations
Ethics approval and consent to participate
Not applicable.
Consent for publication
The publication has been approved by all coauthors.
A
Data Availability
Data is provided within the manuscript or supplementary information files.
Competing interests
The authors declare that they have no competing interests.
A
Funding
This work was supported by the National Key R&D Program of China (2021YFA0909600, 2023YFA0914600), the Natural Environment Research Council (NE/Z00019X/1), Research England (UK) Partnership and Innovation fund (175945), Leverhulme Trust (RPG-2021-286), the Biotechnology and Biological Sciences Research Council (BB/Y01135X/1, BB/Y008308/1, BB/V009729/1), the Royal Society (URF\R\180030), and a University of Liverpool-China Scholarship Council PhD studentship (to Y.C.).
A
Author Contribution
L.-N.L. and R.G. designed and supervised the project. Y.C. designed and conducted the experiments, data acquisition and analysis. A.B. assisted with GC-MS data acquisition and data analysis. V.C. assisted with sample collection and thin-section electron microscope imaging. M.Y., P.C., T.L., and B.Z. assisted with mutant construction. Y.C., L.-N.L. and R.G. wrote the manuscript with all authors’ contributions.
Acknowledgements
We thank Gregory F. Dykes from the Liverpool Biomedical Electron Microscopy Unit for technical assistance and provision for microscopic imaging.
Electronic Supplementary Material
Below is the link to the electronic supplementary material
References
1.
Suggitt AJ, Wheatley CJ, Aucott P, Beale CM, Fox R, Hill JK, et al. Linking climate warming and land conversion to species' range changes across Great Britain. Nat Commun. 2023;14(1):6759.
2.
Abbass K, Qasim MZ, Song H, Murshed M, Mahmood H, Younis I. A review of the global climate change impacts, adaptation, and sustainable mitigation measures. Environ Sci Pollut Res Int. 2022;29(28):42539–59.
3.
Berg IA. Ecological Aspects of the Distribution of Different Autotrophic CO2 Fixation Pathways. Appl Environ Microbiol. 2011;77(6):1925–36.
4.
Gong FY, Zhu HW, Zhang YP, Li Y. Biological carbon fixation: From natural to synthetic. J CO2 Utilization. 2018;28:221–7.
5.
Biel K, Fomina I. Benson-Bassham-Calvin cycle contribution to the organic life on our planet. Photosynthetica. 2015;53(2):161–7.
6.
Raven JA. Contributions of anoxygenic and oxygenic phototrophy and chemolithotrophy to carbon and oxygen fluxes in aquatic environments. Aquat Microb Ecol. 2009;56:177–92.
7.
Liu LN. Advances in the bacterial organelles for CO2 fixation. Trends Microbiol. 2022;30(6):567–80.
8.
Davidi D, Shamshoum M, Guo Z, Bar-On YM, Prywes N, Oz A, et al. Highly active rubiscos discovered by systematic interrogation of natural sequence diversity. EMBO J. 2020;39(18):e104081.
9.
Flamholz AI, Prywes N, Moran U, Davidi D, Bar-On YM, Oltrogge LM, et al. Revisiting Trade-offs between Rubisco Kinetic Parameters. Biochemistry. 2019;58(31):3365–76.
10.
Wang Y, Stessman DJ, Spalding MH. The CO2 concentrating mechanism and photosynthetic carbon assimilation in limiting CO2: how Chlamydomonas works against the gradient. Plant J. 2015;82(3):429–48.
11.
Pandey P, Gupta R, Tiwari S, Gupta A, Mondal S, Sinha RP, et al. In: Mishra AK, Singh SS, editors. Carbon concentrating mechanism in cyanobacteria: necessity and evolution. Cyanobacteria: Academic; 2024. pp. 57–67.
12.
Sui N, Huang F, Liu LN. Photosynthesis in Phytoplankton: Insights from the Newly Discovered Biological Inorganic Carbon Pumps. Mol Plant. 2020;13(7):949–51.
13.
Fang S, Huang X, Zhang X, Zhang M, Hao Y, Guo H, et al. Molecular mechanism underlying transport and allosteric inhibition of bicarbonate transporter SbtA. Proc Natl Acad Sci U S A. 2021;118(22):e2101632118.
14.
Wang C, Sun B, Zhang X, Huang X, Zhang M, Guo H, et al. Structural mechanism of the active bicarbonate transporter from cyanobacteria. Nat Plants. 2019;5(11):1184–93.
15.
Sun Y, Sheng Y, Ni T, Ge X, Sarsby J, Brownridge PJ et al. Rubisco packaging and stoichiometric composition of the native β-carboxysome in Synechococcus elongatus PCC7942. Plant physiology. 2024:in press.
16.
Sun Y, Harman VM, Johnson JR, Brownridge PJ, Chen T, Dykes GF, et al. Decoding the Absolute Stoichiometric Composition and Structural Plasticity of alpha-Carboxysomes. mBio. 2022;13(2):e0362921.
17.
Huffine CA, Zhao R, Tang YJ, Cameron JC. Role of carboxysomes in cyanobacterial CO(2) assimilation: CO(2) concentrating mechanisms and metabolon implications. Environ Microbiol. 2023;25(2):219–28.
18.
Faulkner M, Szabó I, Weetman SL, Sicard F, Huber RG, Bond PJ, et al. Molecular simulations unravel the molecular principles that mediate selective permeability of carboxysome shell protein. Sci Rep. 2020;10(1):17501.
19.
Gleizer S, Ben-Nissan R, Bar-On YM, Antonovsky N, Noor E, Zohar Y, et al. Conversion of Escherichia coli to Generate All Biomass Carbon from CO(2). Cell. 2019;179(6):1255–e6312.
20.
Chen T, Fang Y, Jiang Q, Dykes GF, Lin Y, Price GD, et al. Incorporation of Functional Rubisco Activases into Engineered Carboxysomes to Enhance Carbon Fixation. ACS Synth Biol. 2022;11(1):154–61.
21.
Zhou Y, Whitney S. Directed Evolution of an Improved Rubisco; In Vitro Analyses to Decipher Fact from Fiction. Int J Mol Sci. 2019;20(20):5019.
22.
Wilson RH, Whitney SM. Improving CO2 Fixation by Enhancing Rubisco Performance. In: Alcalde M, editor. Directed Enzyme Evolution: Advances and Applications. Cham: Springer International Publishing; 2017. pp. 101–26.
23.
Li YJ, Wang MM, Chen YW, Wang M, Fan LH, Tan TW. Engineered yeast with a CO(2)-fixation pathway to improve the bio-ethanol production from xylose-mixed sugars. Sci Rep. 2017;7(1):43875.
24.
Tong T, Chen X, Tang K, Ma W, Gao C, Song W, et al. A new-to-nature photosynthesis system enhances utilization of one-carbon substrates in Escherichia coli. Nat Commun. 2025;16(1):145.
25.
Antonovsky N, Gleizer S, Noor E, Zohar Y, Herz E, Barenholz U, et al. Sugar Synthesis from CO2 in Escherichia coli. Cell. 2016;166(1):115–25.
26.
Li YH, Ou-Yang FY, Yang CH, Li SY. The coupling of glycolysis and the Rubisco-based pathway through the non-oxidative pentose phosphate pathway to achieve low carbon dioxide emission fermentation. Bioresour Technol. 2015;187:189–97.
27.
Rin Kim S, Kim SJ, Kim SK, Seo SO, Park S, Shin J, et al. Yeast metabolic engineering for carbon dioxide fixation and its application. Bioresour Technol. 2022;346:126349.
28.
Flamholz AI, Dugan E, Blikstad C, Gleizer S, Ben-Nissan R, Amram S, et al. Functional reconstitution of a bacterial CO2 concentrating mechanism in Escherichia coli. Elife. 2020;9:e59882.
29.
Prywes N, Phillips NR, Oltrogge LM, Lindner S, Taylor-Kearney LJ, Tsai YC, et al. A map of the rubisco biochemical landscape. Nature. 2025;638(8051):823–8.
30.
Carneiro S, Ferreira EC, Rocha I. Metabolic responses to recombinant bioprocesses in Escherichia coli. J Biotechnol. 2013;164(3):396–408.
31.
Volk MJ, Tran VG, Tan SI, Mishra S, Fatma Z, Boob A, et al. Metabolic Engineering: Methodologies and Applications. Chem Rev. 2023;123(9):5521–70.
32.
Kumar RR, Prasad S. Metabolic engineering of bacteria. Indian J Microbiol. 2011;51(3):403–9.
33.
Fiehn O. Metabolomics – the link between genotypes and phenotypes. Plant Mol Biol. 2002;48(1/2):155–71.
34.
Du J, Forster B, Rourke L, Howitt SM, Price GD. Characterisation of cyanobacterial bicarbonate transporters in E. coli shows that SbtA homologs are functional in this heterologous expression system. PLoS ONE. 2014;9(12):e115905.
35.
Desmarais JJ, Flamholz AI, Blikstad C, Dugan EJ, Laughlin TG, Oltrogge LM, et al. DABs are inorganic carbon pumps found throughout prokaryotic phyla. Nat Microbiol. 2019;4(12):2204–15.
36.
Price GD, Woodger FJ, Badger MR, Howitt SM, Tucker L. Identification of a SulP-type bicarbonate transporter in marine cyanobacteria. Proc Natl Acad Sci U S A. 2004;101(52):18228–33.
37.
Parikh MR, Greene DN, Woods KK, Matsumura I. Directed evolution of RuBisCO hypermorphs through genetic selection in engineered E. coli. Protein engineering, design & selection: PEDS. 2006;19(3):113-9.
38.
Sumner LW, Amberg A, Barrett D, Beale MH, Beger R, Daykin CA et al. Proposed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics: Official journal of the Metabolomic Society. 2007;3(3):211 – 21.
39.
Merlin C, Masters M, McAteer S, Coulson A. Why is carbonic anhydrase essential to Escherichia coli? J Bacteriol. 2003;185(21):6415–24.
40.
Price GD. Inorganic carbon transporters of the cyanobacterial CO2 concentrating mechanism. Photosynth Res. 2011;109(1–3):47–57.
41.
Pulsford SB, Outram MA, Forster B, Rhodes T, Williams SJ, Badger MR, et al. Cyanobacterial alpha-carboxysome carbonic anhydrase is allosterically regulated by the Rubisco substrate RuBP. Sci Adv. 2024;10(19):eadk7283.
42.
Vijayakumar J, Thangavelu V, Rajendran A. Recent Trends in the Production, Purification and Application of Lactic Acid. Chem Biochem Eng Q. 2008;22(2):245–64.
43.
Liu H, Kang J, Qi Q, Chen G. Production of lactate in Escherichia coli by redox regulation genetically and physiologically. Appl Biochem Biotechnol. 2011;164(2):162–9.
44.
Guadalupe-Medina V, Wisselink HW, Luttik MAH, de Hulster E, Daran JM, Pronk JT, et al. Carbon dioxide fixation by Calvin-Cycle enzymes improves ethanol yield in yeast. Biotechnol Biofuels. 2013;6:12.
45.
Bonacci W, Teng PK, Afonso B, Niederholtmeyer H, Grob P, Silver PA, et al. Modularity of a carbon-fixing protein organelle. Proc Natl Acad Sci U S A. 2012;109(2):478–83.
46.
Guzman LM, Belin D, Carson MJ, Beckwith J. Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. J Bacteriol. 1995;177(14):4121–30.
47.
Martinez-Garcia E, de Lorenzo V. Engineering multiple genomic deletions in Gram-negative bacteria: analysis of the multi-resistant antibiotic profile of Pseudomonas putida KT2440. Environ Microbiol. 2011;13(10):2702–16.
48.
Flannagan RS, Linn T, Valvano MA. A system for the construction of targeted unmarked gene deletions in the genus Burkholderia. Environ Microbiol. 2008;10(6):1652–60.
49.
Valle A, de la Calle ME, Muhamadali H, Hollywood KA, Xu Y, Lloyd JR, et al. Metabolomics of Escherichia coli for Disclosing Novel Metabolic Engineering Strategies for Enhancing Hydrogen and Ethanol Production. Int J Mol Sci. 2023;24(14):17.
50.
Broadhurst D, Goodacre R, Reinke SN, Kuligowski J, Wilson ID, Lewis MR, et al. Guidelines and considerations for the use of system suitability and quality control samples in mass spectrometry assays applied in untargeted clinical metabolomic studies. Metabolomics: Official J Metabolomic Soc. 2018;14(6):72.
51.
Begley P, Francis-McIntyre S, Dunn WB, Broadhurst DI, Halsall A, Tseng A, et al. Development and Performance of a Gas Chromatography – Time-of-Flight Mass Spectrometry Analysis for Large-Scale Nontargeted Metabolomic Studies of Human Serum. Anal Chem. 2009;81(16):7038–46.
52.
Kovats E, Giddings JC, Keller R. Gas chromalographic characterization of organic substances in the retention index system. Adv Chromatogr. 1965;1:229–47.
53.
Chambers MC, Maclean B, Burke R, Amodei D, Ruderman DL, Neumann S, et al. A cross-platform toolkit for mass spectrometry and proteomics. Nat Biotech. 2012;30(10):918–20.
54.
Tsugawa H, Cajka T, Kind T, Ma Y, Higgins B, Ikeda K, et al. MS-DIAL: data-independent MS/MS deconvolution for comprehensive metabolome analysis. Nat Methods. 2015;12(6):523–6.
55.
Hummel J, Strehmel N, Selbig J, Walther D, Kopka J. Decision tree supported substructure prediction of metabolites from GC-MS profiles. Metabolomics: Official J Metabolomic Soc. 2010;6(2):322–33.
56.
MS-DIAL metabolomic database [Science webpage]. 2024 [cited 2024 Sep 2nd ]. Available from: http://prime.psc.riken.jp/compms/msdial/main.html#MSP
57.
Alsberg BK, Jensen VR, Børve KJ. Use of multivariate methods in the analysis of calculated reaction pathways. J Comput Chem. 1996;17(10):1197–216.
58.
Goodacre R, Timmins ÉM, Burton R, Kaderbhai N, Woodward AM, Kell DB, et al. Rapid identification of urinary tract infection bacteria using hyperspectral whole-organism fingerprinting and artificial neural networks. Microbiology. 1998;144(5):1157–70.
59.
Gromski PS, Muhamadali H, Ellis DI, Xu Y, Correa E, Turner ML, et al. A tutorial review: Metabolomics and partial least squares-discriminant analysis–a marriage of convenience or a shotgun wedding. Anal Chim Acta. 2015;879:10–23.
60.
Pang Z, Lu Y, Zhou G, Hui F, Xu L, Viau C, et al. MetaboAnalyst 6.0: towards a unified platform for metabolomics data processing, analysis and interpretation. Nucleic Acids Res. 2024;52(W1):W398–406.
61.
Ewald JD, Zhou G, Lu Y, Kolic J, Ellis C, Johnson JD, et al. Web-based multi-omics integration using the Analyst software suite. Nat Protoc. 2024;19(5):1467–97.
62.
Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J Roy Stat Soc: Ser B (Methodol). 1995;57(1):289–300.
63.
KEGG. Metabolic pathways - Escherichia coli BL21(DE. 3) [Database]. 2024 [cited 2024 Oct 10th ]. Available from: https://www.kegg.jp/pathway/ebl01100
A
Fig. 1
Evaluation of the activity of Bicarbonate transporters. (a) Demonstrating the high CO2 dependency of the ΔC mutant. (b) shows the plasmids constructed for expressing Ci pumps in E. coli. Abbreviations in brackets were used to indicate the expression plasmids in the variants (e.g. BL:T1 represents the BL21(DE3) strains harbouring pDabA1/B1). (c) shows the strains with different Ci pumps or an empty plasmid incorporated. ‘N’ signifies the absence of the can gene, while 'Y' indicates its presence. ‘+’ symbols indicate the presence of corresponding Ci pumps or empty plasmid. (d) Growth curves of ΔC mutant with the presence of different Ci pumps or empty pET28a; LB medium supplemented with 50 µg mL− 1 kanamycin, and 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) was used; a.u.: absorption unit. The solid lines represent the average OD600 from 6 replicate inoculations, and the transparent shade along the lines with a similar colour represents the standard deviation.
A
Fig. 2
Growth profile of the strains expressing functional recombinant proteins. (a) Expression plasmids constructed for α-carboxysomes expression (pCB), Prk expression (pPRK), α-carboxysomes and Prk co-expression (pCBPRK), and Rubisco and Prk co-expression (pRubPRK). Abbreviations in brackets are used to indicate the expression plasmids in the variants. (b) Strains that possess corresponding components or empty plasmids are labelled using abbreviations given in the plasmid map and indicated by grey boxes in the table. ‘Free Rubisco’ refers to Rubisco expressed alone, without other α-carboxysome components, whereas α-carboxysome represents the carboxysome including the Rubisco and CA encapsulated. (c) Growth curves of E. coli with the presence of different components of the CBB cycle indicate the functionality of these components. The solid lines represent the average OD600 from 6 replicate inoculations, and the transparent shade along the lines with a similar colour represents the standard deviation. M9 medium supplemented with 0.2% glycerol, 50 µg mL-1 chloramphenicol, and 1 mM arabinose was used for the growth curves. (d) Growth curves of variants harbouring various Ci pumps along with α-carboxysomes and Prk, or the empty plasmid pBAD33 (:Vb). M9 medium supplemented with 0.4% glycerol, 50 µg mL-1 chloramphenicol, 50 µg mL-1 kanamycin, 1 mM arabinose, and 0.5 mM IPTG was used for the growth curves.
A
Fig. 3
PCA and ANOVA illustrate the variances in response to the introduced components of the CBB cycle and the carboxysome-based CCM. (a) PCA score plot with all ten variants demonstrates a clear separation between variants harbouring the CBB cycle and those that lack it. The percentage numbers in the axis legends indicate the total variance explained by that PC. Ellipses show the 95% confidence. (b) ANOVA identified metabolites (or features) that were changed in response to the introduced components for the CBB cycle and CCM, FDR-corrected P-value threshold < 0.05.
A
Fig. 4
Identification of differential metabolites in response to the presence of the CBB cycle and DabA2/B2. Volcano plots of the paired variants, including (a) BL:CpVe v.s. BL:Veb, (b) BL:VbT2 v.s. BL:Veb, and (c) BL:CpT2 v.s. BL:VbT2. The annotated metabolites were black cycled; blue dots, downregulated (P-value < 0.05, Log2(FC) < -1); red dots, upregulated (P-value < 0.05, Log2(FC) > 1); yellow dots, non-significantly changed (P-value < 0.05, -1 < Log2(FC) < 1); Black and grey dots, non-significantly changed, (P-value > 0.05). (d) Boxplots represent the relative concentration of some annotated metabolites in the variants. The box range gives the interquartile range (IQR: 25% and 75%) of the data set; bars show the data within the range of 1.5 x IQR, and those data excessed this range were defined as outliers (stars). The line within the box indicates the median and the empty square (‘□’) represents the mean value.
A
Fig. 5
Pathways map adapted from KEGG metabolic pathways (ebl01100) of E. coli BL21(DE3) illustrates the connections between the affected metabolites and pathways. According to ANOVA results, metabolites that were significantly changed between the variants are shown in
A
Fig. 6
Heatmap of the correlation analysis of all 257 metabolite features. The x- and y-axes indicate all 257 metabolites (features), and the colour of the x-y cross indicates the Pearson’s correlation coefficient of two metabolites (-1 to 1). Four correlation clusters, marked as ‘I’, ‘II’, ‘III’, and ‘IV’, are revealed. Only labels of the 30 annotated and influenced metabolites were shown on the heatmap. The four correlation clusters were given by the dash triangles or square. Pearson’s correlation was employed for the analysis. See also Supplementary File 1.
