RESEARCH ARTICLE | APPLIED BIOLOGICAL SCIENCES OPEN ACCESS Conversion of CO2 into organic acids by engineered autotrophic yeast € Michael Baumschabla,b,1 , Ozge Ataa,b,1 , Bernd M. Miticb,c , Lisa Lutza,b , Thomas Gasslera,b,d , Christina Troyerc , Stephan Hanna,c , and Diethard Mattanovicha,b,2 Edited by Sang Yup Lee, Korea Advanced Institute of Science and Technology, Daejeon, Korea (South); received July 9, 2022; accepted October 13, 2022 The increase of CO2 emissions due to human activity is one of the preeminent reasons for the present climate crisis. In addition, considering the increasing demand for renew- Significance able resources, the upcycling of CO2 as a feedstock gains an extensive importance to establish CO2-neutral or CO2-negative industrial processes independent of agricultural Industrial biotechnology bears resources. Here we assess whether synthetic autotrophic Komagataella phaffii (Pichia great potential to reduce CO2 pastoris) can be used as a platform for value-added chemicals using CO2 as a feedstock emissions by producing chemicals by integrating the heterologous genes for lactic and itaconic acid synthesis. 13C labeling from renewable agricultural experiments proved that the resulting strains are able to produce organic acids via the feedstocks, however at the risk to assimilation of CO2 as a sole carbon source. Further engineering attempts to prevent compete with food production. the lactic acid consumption increased the titers to 600 mg L21, while balancing the While most industrially relevant expression of key genes and modifying screening conditions led to 2 g L21 itaconic organisms are heterotrophs there acid. Bioreactor cultivations suggest that a fine-tuning on CO2 uptake and oxygen has been recent progress to equip demand of the cells is essential to reach a higher productivity. We believe that through some with a CO2 fixation pathway further metabolic and process engineering, the resulting engineered strain can become a promising host for the production of value-added bulk chemicals by microbial assimila- leading to autotrophic growth. We tion of CO2, to support sustainability of industrial bioprocesses. have recently developed a synthetic autotrophic strain of the synthetic biology j metabolic engineering j carbon capture j organic acids j yeast industrial yeast Komagataella phaffii. Here we integrated the Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. Between 2011 and 2020 the annual average CO2 emission due to human activity pathways to lactic and itaconic exceeded 38 gigatons, of which around 22 gigatons are removed again from the atmo- acid (two chemical building blocks) sphere to terrestrial and ocean CO2 sinks. This imbalance, mostly due to combustion into this strain. Up to 2 g L1 of of fossil fuels, causes the steady increase in CO2 levels in the atmosphere which is one itaconic acid were produced from of the primary reasons for the climate crisis our planet is facing today (1). Biologically produced fuels and commodity chemicals bear the potential to counter- CO2 as the only carbon source. act this deleterious development, but the most common feedstocks used for bioproduc- This work paves the way toward tion, such as glucose, sucrose and starch rely on agricultural production and are bearing net CO2 capturing into long living the risk to threaten food security. Using autotrophic microorganisms as production chemical products based on a platforms exploits the potential of CO2 itself as an alternative carbon source. In nature, synthetic autotrophic chassis autotrophic microorganisms play a major role in CO2 fixation by fixing 200 gigatons strain. of CO2 every year (2). However, the rates of most natural microbial CO2 fixing path- ways are low: the photosynthetic efficiency in cyanobacteria is limited to 1 to 2% which makes industrial processes using photoautotrophs economically less feasible (3). Chemoautotrophs may overcome this barrier as their energy harvesting processes are more efficient than light harvesting of photoautotrophs. One well-known autotroph that is being developed for biological production of materials is Cupriavidus necator (formerly known as Ralstonia eutropha). By harvest- ing energy with a controlled Knallgas reaction these bacteria assimilate CO2 via the Calvin-Benson-Bessham (CBB) cycle. Besides naturally produced polyhydroxyalka- € Author contributions: O.A., C.T., S.H., and D.M. € noates (PHA), metabolic pathway engineering enabled the production of several other designed research; M.B., O.A., € B.M.M., L.L., and T.G. performed research; M.B., O.A., B.M.M., C.T., and S.H. chemicals (4–7). € analyzed data; and M.B., O.A., and D.M. wrote the paper. Several chemolithotrophic bacteria were demonstrated as production hosts for vari- D.M. and T.G. are co-inventors of a patent application ous chemicals such as ethanol, 2,3-butanediol, butanol, isopropanol, acetone, or isobu- disclosing organic acid production with synthetic tyric acid via natural carbon assimilation pathways (8–11). In addition to natural CO2 autotrophic yeasts. fixing microorganisms, implementation of heterologous CO2 fixing pathways to hetero- This article is a PNAS Direct Submission. trophic microorganisms like Myceliophthora thermophila provided a mixotrophic strain Copyright © 2022 the Author(s). Published by PNAS. This open access article is distributed under Creative that is able to produce malic acid with a higher yield compared to the parent strain in Commons Attribution License 4.0 (CC BY). which only the reductive tricarboxylic acid (TCA) cycle is used for the production (12). 1 € M.B. and O.A. contributed equally to this work. Natural chemoautotrophs have a large potential to convert CO2 to chemicals. How- 2 To whom correspondence may be addressed. Email: diethard.mattanovich@boku.ac.at. ever, they are often recalcitrant to genetic editing, have complex nutrient demands, or This article contains supporting information online at may require complex process technological solutions like the transfer of gaseous substrates http://www.pnas.org/lookup/suppl/doi:10.1073/pnas. and energy sources. To circumvent some of these limitations well established prokaryotic 2211827119/-/DCSupplemental. and eukaryotic production hosts have been engineered to assimilate CO2. The bacterial Published November 16, 2022. PNAS 2022 Vol. 119 No. 47 e2211827119 https://doi.org/10.1073/pnas.2211827119 1 of 10 A B E C D F Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. Fig. 1. Expression of cadA and ldhL enables organic acid production in synthetic autotrophic K. phaffii. (A–D) Schematic pathways. (A) In wild-type K. phaffii methanol is oxidized to formaldehyde (black arrow) and assimilated in the XuMP cycle (orange arrows) or dissimilated to CO2, respectively (purple arrow). (B) synthetic autotrophy in K. phaffii: the native assimilatory branch of methanol utilization was interrupted by deleting DAS1 and DAS2 (dashed gray line). AOX1 was knocked out to reduce the rate of formaldehyde formation which could be toxic to the cells. RuBisCO and PRK were integrated to complete a func- tional CBB cycle (green arrows). Additionally, two bacterial chaperones, groEL and groES, were overexpressed to assist the folding of RuBisCO. TDH3, PGK1, TKL1, TPI1 carrying each a peroxisomal targeting signal were overexpressed to assure the localization of the entire CBB cycle in peroxisomes. More details about the engineering strategy can be found in ref. (13). (C) Itaconic acid (red) and (D) lactic acid production (blue), (E) growth profiles, and (F) organic acid production profiles of the producing strains and the control. Time axis corresponds to the production phase under autotrophic conditions. At least three biological replicates were used in the screening to monitor the producing strains. Shades represent the SDs (±). 3PG: 3-phosphoglycerate, AcCoA: acetyl-coenzyme A, AOX1 and AOX2: alcohol oxidase 1 and 2, cadA: cis-aconitate decarboxylase, CBB cycle: Calvin-Benson-Bassham cycle, CISAc: cytosolic cis-aconitate, CISAm: mitochondrial cis-aconitate, DAS1 and DAS2: dihydroxyacetone synthase 1 and 2, DHA: dihydroxyacetone, FAL: formaldehyde, G3P: glyceraldehyde 3-phosphate, ITA: itaconic acid, LA: lactic acid, ldhL: L-lactate dehydrogenase, MeOH: methanol, mttA: mitochondrial tricarboxylic acid transporter, NAD+/NADH: nicotinamide adenine dinucleotide, PRK: phosphoribulokinase, PYR: pyruvate, RuBP: ribulose 1,5-bisphosphate, RuBisCO: ribulose 1,5-bisphosphate carboxylase/oxygenase, Xu5P: xylulose 5-phosphate, XuMP cycle: xylulose monophosphate cycle. workhorse Escherichia coli and the yeast Komagataella phaffii annual polymer production of more than 300 million tons, a (Pichia pastoris) were provided with the CBB cycle, enabling volume that denotes a significant impact on the global CO2 them to assimilate CO2 by using formate or methanol, respec- balance. tively, as energy sources. Both engineered microorganisms can Here, we set out to evaluate if the autotrophic K. phaffii grow sustainably with CO2 as carbon source (13, 14). Concep- strain can be used as a platform for organic acid production. tually formate and methanol are regarded as sustainable feed- Synthetic autotrophy was introduced to K. phaffii by converting stocks for biotechnology when they are derived from CO2 by the native peroxisomal methanol assimilation pathway, the hydrogenation or electrochemical reduction (15). xylulose monophosphate (XuMP) cycle, into the CBB cycle (13). To make an impact on the global CO2 household such auto- To achieve that, the formaldehyde assimilating enzyme dihy- trophic processes need to convert CO2 into bulk products. droxyacetone synthase (DAS) was replaced by a bacterial ribulose Besides ethanol, short chain organic acids are the second largest 1,5-bisphosphate carboxylase/oxygenase (RuBisCO), and phos- group of chemicals manufactured by industrial biotechnology. phoribulose kinase (PRK) from spinach was added to supply The market of biologically produced organic acids is expected ribulose bisphosphate from XuMP precursors. Four yeast glyco- to reach more than $36 billion by 2026 (16). The annual bio- lytic enzymes were targeted to peroxisomes to close the CBB production of some of the key organic acids (citric, acetic, lac- cycle to glyceraldehyde 3-phosphate (G3P), both as an intermedi- tic, succinic acid) is more than 12 million metric tons (17–20). ate and the product of the assimilation cycle (Fig. 1 A and B). In Recently, itaconic acid has also gained attention as a promising the present work, using modular synthetic biology tools, we chemical building block with an estimated market increase to implemented the genes for lactic and itaconic acid synthesis plus 170 kilotons per year and $260 million in 2025 (21). Lactic accessory genes into the K. phaffii genome and demonstrate that and itaconic acid are feedstocks for polymer production so that the autotrophic K. phaffii strain is capable of producing organic they, and other biobased commodity chemicals compete for the acids solely from CO2 as carbon source. 2 of 10 https://doi.org/10.1073/pnas.2211827119 pnas.org Results A B Production of Organic Acids in Synthetic Autotrophic K. phaffii. Itaconic acid is produced through the decarboxylation of cis- aconitate into itaconic acid by cis-aconitate decarboxylase encoded by cadA in Aspergillus terreus (Fig. 1C). The cadA gene was expressed under a strong methanol inducible promoter (PAOX1) coupling product formation to CO2 assimilation via the syn- thetic CBB cycle which is controlled by methanol-induced pro- moters (13). As autotrophic growth of the strains is slow we used a two-phase screening protocol, first growing the cells on glycerol for biomass formation for 22 to 24 h. In the second phase, cells were transferred to the production medium with an initial OD600 of 3 to 4 where air is enriched with 5% CO2 as the only carbon source, and 1% methanol is added for produc- tion of NADH and ATP. Calculated specific growth and pro- Fig. 2. Fine-tuning the balance of expression of cadA and mttA increases duction rates and yields of the screened strains are given in SI itaconic acid production. (A) Growth profiles, (B) itaconic acid production profiles of the itaconic acid producing strain and the control. Time axis cor- Appendix, Table S1. We screened the producing clones for 200 h responds to the production phase under autotrophic conditions. At least under autotrophic conditions in comparison to the control strain three biological replicates were used in the screening to monitor the pro- without cadA (Fig. 1E). Expression of cadA did not affect growth, ducing strains. Shades represent the SDs (±). and the specific growth rates through most of the cultivation period were about 0.007 h1 while the control strain had a spe- was nearly restored (almost 80% of the cadA strain was achieved) cific growth rate of 0.006 h1. The expression of a single gene, and the final titers were improved. Using a medium strength pro- cadA, was sufficient for the cells to produce itaconic acid up to moter resulted in the highest titer (approximately. 780 mg L1) 530 mg L1 at the end of the production phase with a specific and highest specific productivity (3.2 mg g1 h1). Therefore, production rate of 2.3 mg g1 h1 (Fig. 1F). Using a faster grow- this strain was selected for further experiments. ing synthetic autotrophic strain carrying a mutation in the PRK gene (22) did not improve the final titers, but even led to lower Deletion of CYB2 Prevents Lactic Acid Consumption. Under specific production rates (SI Appendix, Fig. S1). The control strain autotrophic conditions the cells face severe carbon limiting con- that lacks cadA did not produce any itaconic acid during the culti- ditions. Since lactic acid can be efficiently converted back to Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. vation as wild-type K. phaffii does not harbor necessary genes for pyruvate it could serve as a potential carbon source which itaconic acid synthesis. would reduce the final lactic acid yield. Indeed, lactic acid con- L-lactic acid is produced via the reduction of pyruvate to sumption was seen in preliminary experiments (SI Appendix, L-lactic acid catalyzed by L-lactate dehydrogenase encoded by Fig. S2). To overcome this bottleneck, the CYB2 gene was ldhL in Lactobacillus plantarum. To enable lactic acid production knocked out in the lactic acid production strain. CYB2 encodes in K. phaffii the ldhL gene was expressed under the control of for a L-lactate cytochrome c oxidoreductase located in the mito- the strong methanol inducible promoter of the AOX1 gene (Fig. chondrial intermembrane space. Deletion of this gene abolished 1D). The resulting strain was able to produce up to 200 mg L1 growth on L-lactate in Saccharomyces cerevisiae (26, 27). To eval- after 250 h of production time (Fig. 1F) with a specific produc- uate the effect of the deletion in K phaffii we performed cultiva- tion rate of 0.85 mg g1 h1. Growth was slightly reduced tions using lactate as the sole carbon source. The ldhL strain was compared to the control strain with a specific growth rate of able to grow on L-lactic acid with a maximum growth rate of 0.005 h1 (Fig. 1E). K. phaffii control strains that do not have around 0.01 h1 (Fig. 3A). Lactate concentration was reduced a heterologous lactate dehydrogenase do not produce any from 8.5 g L1 to 4.0 g L1 throughout the cultivation of this detectable lactic acid. Balancing the Coexpression of cadA and mttA Improves A B Production of Itaconic Acid. Itaconic acid metabolism requires two key genes: cis-aconitate decarboxylase (cadA) and a mito- chondrial tricarboxylic acid transporter (mttA) to shuttle the substrate, cis-aconitate, to the cytosol where cadA is localized. Coexpression of mttA with cadA was found to increase the ita- conic acid production titers in Aspergillus niger (23, 24). To investigate the effect of the coexpression of mttA, we used three different promoters with different strengths: PFDH1 as a strong, PPOR1 as a medium and PPDC1 as a weak promoter (25). We found that balancing the coexpression of cadA and mttA is cru- cial to increase itaconic acid productivity (Fig. 2). Using the strong FDH1 promoter severely impaired growth and the final itaconic acid titers were only around 190 mg L1 at the end of the cultivation. As the cells did not grow, a higher specific pro- Fig. 3. Deletion of CYB2 prevents lactic acid consumption. (A) Growth pro- ductivity is expected, but the final titer was 4.4-fold lower than files, (B) lactic acid consumption profiles of the lactic acid producing strain, with overexpression of cadA alone leading to 1.4-fold lower spe- the lactic acid producing cyb2Δ strain, and the cyb2Δ strain. Time axis cor- responds to the cultivation time on media containing lactic acid under non- cific productivity (1.3 mg g1 h1). However, when we used a autotrophic conditions. At least three biological replicates were used in the medium strength (PPOR1) or a weak promoter (PPDC1), growth screening to monitor the producing strains. Shades represent the SDs (±). PNAS 2022 Vol. 119 No. 47 e2211827119 https://doi.org/10.1073/pnas.2211827119 3 of 10 strain (Fig. 3B) resulting in an average specific consumption rate degree of our samples. We took samples to monitor the tran- of 28 mg g1 h1. A nonproducing control strain harboring the sient changes of the isotopologue patterns and 13C labeling CYB2 knockout (cyb2Δ) was not able to grow on L-lactate nor degree (according to the equation denoted in the SI Appendix, to consume lactic acid. The production strain harboring the Supplementary Methods) of the products (Fig. 5 B and C and SI CYB2 knockout (ldhL cyb2Δ) could barely grow with a maxi- Appendix, Fig. S4) on day 3 (time where we start to measure mum growth rate of around 0.002 h1. This strain was able to significant amounts of organic acid), day 5 and day 8. In addi- consume small amounts of lactic acid with a reduction in lactic tion to the producing strain, we also used a RuBisCO knockout acid concentration from 8.5 g L1 to 7.2 g L1. The specific lac- producer strain as a control that should not be able to fix CO2 tic acid consumption rate was reduced nearly threefold compared due to the interrupted CBB cycle. Still these strains produced to the nonknockout strain resulting in 11 mg g1 h1. low amounts of itaconic or lactic acid, presumably from reserve In the next step, the capabilities of these strains to produce metabolites or residual intracellular metabolite pools, however lactic acid were evaluated under autotrophic conditions. As providing enough material to evaluate the isotopologue patterns expected, the control strain harboring the CYB2 knockout did of these products. not produce any lactate (Fig. 4). The production strain harboring SI Appendix, Fig. S5 shows how the 13C labeling pattern of the CYB2 knockout resulted in an increased lactic acid produc- itaconic and lactic acid changed during the production. As tion reaching a final titer of around 350 mg L1 with a specific expected, the 12C contents of both organic acids were increased production rate of 1.3 mg g1 h1. Growth of both knockout throughout the cultivation. For itaconic acid, the 13C labeling strains was slightly higher compared to the nonknockout strain degree decreased from 35% (day 3) to 18% (day 8) (Fig. 5B). with the nonproduction strain resulting in fastest growth (Fig. The low amounts of itaconic acid produced by the delta 4A). By transforming the genetic setup into a faster growing RuBisCO producer strain (∼0.4 mmol L1) resulted in a high 13 version of the synthetic autotrophic strain harboring a muta- C labeling degree of 75% at the end of the cultivation. tion in the PRK gene (22) we could not further enhance pro- For lactic acid, the 13C labeling degree dropped to only 10% duction of lactic acid. The nonknockout strain was only at the end of the cultivation. Lactic acid is derived from pyru- producing 50 mg L1. In this strain background the knockout vate which is much closer to the entry of the CBB cycle into of CYB2 could increase the titer up to fourfold reaching up to the central carbon metabolism than cis-aconitate, the precursor 200 mg L1 (SI Appendix, Fig. S3). for itaconic acid. Therefore, more of the unlabeled carbon com- Interestingly all strains harboring the CYB2 knockout pro- ing from the CBB cycle will appear at early time points in duced glycolic acid. Under autotrophic conditions the non- lactic acid than in itaconic acid, thus a higher 12C content of lactic acid producing knockout strain was able to produce up lactic acid is to be expected in the transient 12C enrichment to 200 mg L1 and the lactic acid strain accumulated 150 mg phase observed here. Additionally, the 12C pattern of the lactic Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. L1 after 250 h of production phase (Fig. 4C). acid produced by the delta RuBisCO producer strain differs sig- nificantly from the itaconic acid producer: The 13C labeling Reverse 13C Labeling Proves That Carbon in Lactic and degree of the produced lactic acid (around 0.4 mmol L1) was Itaconic Acid is Incorporated From the Captured CO2. After 68% at the end of the cultivation (Fig. 5C). Natural CO2 fixa- demonstrating that the CO2 capturing K. phaffii can produce tion pathways, like anaplerotic reactions, might have contributed organic acids, we wanted to prove that the carbon in the to the incorporation of 12CO2 into the produced organic acids. organic acids is incorporated from the captured CO2 via the To calculate the expected 13C labeling degree, we built a CBB cycle. We used a reverse labeling approach (Fig. 5A). We metabolic model including CO2 capture, glycolysis, anaplerotic first fully labeled the biomass by cultivating the strains for at reaction from pyruvate to oxaloacetate and TCA cycle reac- least 10 generations on 13C glycerol as the only carbon source. tions, including itaconic or lactic acid metabolism genes and The resulting 13C content measured using EA-IRMS was 97 ± calculated the theoretical isotopologue distributions of the 2% for the lactic acid and 94 ± 3% for the itaconic acid pro- products using isotopomer network compartmental analysis ducing strains, respectively. [INCA; (28)] using the secretion rates of itaconic and lactic After 13C enrichment, natC CO2 (CO2 containing carbon acid as constraints (SI Appendix, Table S2). Expected labeling with a natural isotope distribution of 1.1% 13C) was supple- degrees were calculated on day 8 as 23% for the itaconic and mented as the C source, and 13C MeOH was used for induc- 18% for the lactic acid producer strains, respectively, which is tion and energy harvest. As shown by Gassler et al. (13) 12C equivalent to a 77% or 82% CO2 incorporation. Experimen- can only be derived from CO2 in this experimental setup with tally derived CO2 incorporation was 6% and 10% higher than our strains. Therefore, we expect a drop in the 13C labeling predicted for itaconic and lactic acid, respectively, which may A B C Fig. 4. Deletion of CYB2 improves lactic acid production in synthetic autotrophic K. phaffii. (A) Growth profiles, (B) lactic acid production profiles, (C) glycolic acid production profiles of the lactic acid producing strain and the control. Time axis corresponds to the production phase under autotrophic conditions. At least three biological replicates were used in the screening to monitor the producing strains. Shades represent the SDs (±). 4 of 10 https://doi.org/10.1073/pnas.2211827119 pnas.org A 13 nat Producer strain C C Cultivation in YNB Inoculation main culture: Sampling for Evaluation of 13 ΔRuBisCO strain with 13C glycerol for C MeOH for induction GC-TOFMS isotopologue fractions nat labeling the cells C CO2 for growth Analysis of products and calculation of 13 and production C labeling degree B C Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. Fig. 5. Reverse labeling confirms that carbon in lactic and itaconic acid is incorporated from the captured CO2. (A) Experimental design of the reverse labeling approach. After labeling the cells on 13C glycerol, we inoculated them into YNB and supplemented them with natC CO2 (CO2 with natural isotope distribution) and 13C MeOH. Samples were taken at different time points (day 3, day 5 and day 8) for the analysis with GC-TOFMS to monitor the change in the13C labeling pattern of the produced organic acids. (B) 13C labeling degree at different timepoints of the produced itaconic acid using the production strain on the left and the nongrower control. (C) 13C labeling degree at different timepoints of the produced lactic acid using the production strain on the left and the nongrower con- trol. Error bars indicate the SD of 3 biological replicates. Detailed isotopologue distribution of the carbon atoms can be found in SI Appendix, Fig. S5. be due to the contribution of natural CO2 fixing reactions For lactic acid production, we obtained a different produc- which were not included in the metabolic model. Additionally, tion profile. Increasing the initial OD did not increase the final the higher gap between the measured and predicted values for titer. We screened different starting OD values in the range lactic acid producing strains could be due to the degradation from OD600 = 4 to 16 (SI Appendix, Fig. S7C). In the begin- rate of lactic acid which overlays the synthesis rate. ning of the production phase higher OD led to slightly higher Complementary to the reverse labeling experiment we tested titers but later got outperformed by the lowest starting OD. if the production depends on the supply of CO2. After switch- Higher biomass cultures showed that lactic acid was consumed ing the cells from 5 to ambient CO2 concentration production again toward the end of the cultivation, indicating an increase and growth was nearly stopped (SI Appendix, Fig. S6). of substrate limitations with increased biomass. With the lowest starting OD, 600 mg L1 lactic acid were produced, compared Modification of Process Conditions Further Enhances the to 160 mg L1 lactic acid for the highest starting OD (Fig. Production of Itaconic Acid but not of Lactic Acid. We sought 6C). Interestingly, the glycolate production profile was differ- to modify the process conditions in the shake flask to see their ent. OD600 = 8 to 16 resulted in very similar production pro- effect on the production and screened the cells at 10% CO2 files ending up with 170 mg L1 glycolate. OD600 = 4 ended supply and with different initial OD600 values (between 4 and up with a slightly lower glycolate titer of 120 mg L1 (SI 20). For itaconic acid production, we observed a positive corre- Appendix, Fig. S7E). To overcome some possible limitations lation between higher initial ODs and production at 5% CO2 with higher starting ODs we doubled the CO2 concentration supply. The higher the initial OD, the higher final titers were from 5 to 10%. Growth was very similar compared to our stan- obtained, reaching more than 1 g L1 (OD600 = 20) (Fig. 6A). dard CO2 concentrations (SI Appendix, Fig. S7D). Increased However, specific growth and production rates decreased with CO2 supply was able to increase the production for cultures increasing initial OD. Increasing the CO2 supply did not help starting with OD600 = 12 and 16. OD600 = 8 ended up with to increase the specific growth rates which stayed also the same similar titers compared to 5% but without the peak after for all different initial ODs. However, a significant increase in 140 h of cultivation time. Interestingly OD600 = 4 resulted in the final titers was obtained (again in positive correlation with decreased lactic acid production with elevated CO2 concentra- increasing ODs) when 10% CO2 was supplied. With a start tions (Fig. 6D). The glycolate production profile was similar OD600 = 20 at 10% CO2 almost 2 g L1 itaconic acid (Fig. but with a slight reduction in titers at 10% CO2 resulting in 6B) could be reached at the end of the production phase with a 100 mg L1 for OD600 = 8 to 16 and 70 mg L1 for specific productivity of 2.7 mg g1 h1. OD600 = 4 respectively (SI Appendix, Fig. S7F). PNAS 2022 Vol. 119 No. 47 e2211827119 https://doi.org/10.1073/pnas.2211827119 5 of 10 A B C D Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. Fig. 6. Modification of process conditions shows that each product has its own requirements for an optimum bioprocess. (A, B) Itaconic acid production profiles and (C, D) lactic acid production profiles at 5% CO2 and 10% CO2 supply, respectively, of producing clones and the control. Time axis corresponds to the production phase under autotrophic conditions. At least two biological replicates were used in the screening to monitor the producing strains. Shades represent the SDs (±). Bioreactor Cultivations: Upscaling Is Not Straightforward. (and keeping the DO concentration at 8%). Reducing the oxy- Based on the screening results with different process conditions, gen concentration led to nearly doubling of the growth rates of we proceeded with bioreactor cultivations to investigate the pro- the lactic acid producing strains (Fig. 7C). Similar as for itaconic duction capacity of the producing strains at a larger scale. acid, the oxygen concentration also influenced the lactic acid Depending on the initial screenings, we designed different biore- production of the cells. In contrast to itaconic acid, high oxy- actor cultivation setups. For itaconic acid production, after a gen levels were beneficial for production. At ambient oxygen batch phase with glycerol to reach an OD600 of 20 to 30 (corre- levels up to 260 mg L1 were produced with a specific produc- sponding to 4 to 6 g L1 dry cell weight (DCW)), cells were fed tivity of 0.76 mg g1 h1(Fig. 7D). Lowering the oxygen with 10% CO2 as the carbon source and feeding with 1% supply reduced the final titer to 130 mg L1 and the specific MeOH every 2 d. We tested two different dissolved oxygen productivity to 0.22 mg g1 h1. A similar pattern was (DO) concentrations: we either kept the DO above 20%, or at observed for the production of glycolic acid in this strain. At 8% by decreasing the oxygen concentration in the inlet gas. Simi- high oxygen levels up to 150 mg L1 and at low levels up to lar to the screenings in shake flasks, the cells did not grow and the 40 mg L1 were produced (Fig. 7E). biomass remained at 4 to 6 g L1 until the end of the cultivation. The DO concentration in the medium has an effect on the itaconic Discussion acid production: at lower DO concentration, higher itaconic acid concentrations could be achieved than at a higher DO. However, We have demonstrated previously that the yeast K. phaffii can we could not achieve as high itaconic acid concentrations as we be engineered to assimilate CO2 instead of methanol as its only could reach in the shake flask: the highest titer was 530 mg L1 carbon source (Fig. 1 A and B) (13). Here we show that this with a specific productivity of 0.74 mg g1 h1 (Fig. 7B). synthetic chemoautotrophic K. phaffii can be used as a platform The initial screenings of different process conditions (Fig. 6 C to produce organic acids from CO2. and D) indicated that a low biomass concentration was better for Expression of a cytosolic cis-aconitate decarboxylase from lactic acid production. Therefore, we performed bioreactor culti- A. terreus, the key gene that encodes the enzyme for the decar- vations starting with a low biomass concentration (OD600 = 2) boxylation reaction of cis-aconitate to itaconic acid, enabled and tested two different oxygen concentrations with 21% in the K. phaffii to reach more than 500 mg L1 itaconic acid final inlet air (uncontrolled DO concentration) and 5% in the inlet air titers (Fig. 1F). Coexpression of a mitochondrial cis-aconitic 6 of 10 https://doi.org/10.1073/pnas.2211827119 pnas.org A B C D E Fig. 7. Bioreactor cultivations show each process should be designed based on the specific requirements of the target product. (A) Growth and (B) itaconic acid production profile at constant DO concentrations 8% and 20%. (C) Growth, (D) lactic, and (E) glycolic acid production profile at 21% and 5% inlet oxygen concentrations. Time axis corresponds to the production phase under autotrophic conditions. Bioreactor cultivations for itaconic acid production were done in duplicates and shades represent the SD (±). For lactic acid results of each fermentation run are shown separately. acid transporter gene, mttA, was shown to be beneficial for interesting aspect of the deletion of CYB2 is the increased gly- itaconic acid production by creating a pull effect from mito- colic acid production in the synthetic autotrophic K. phaffii chondria to the cytosol and increasing the availability of cis- strain. This is probably linked to the oxygenation reaction of Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. aconitate for CadA (29, 30). Our results demonstrate that the RuBisCO enzyme. The products of this reaction are one balancing the expression levels of cadA and mttA is essential to molecule each of 3-phosphoglycerate and 2-phosphoglycolate, reach higher productivities (31) and that the strengths of the which can be afterward dephosphorylated to glycolate. This promoters used for expression of cadA and mttA do not have to hypothesis correlates with a decrease in glycolic acid secretion be necessarily at the same level: using a weaker promoter for when the CO2 concentration was increased (SI Appendix, Fig. coexpression of mttA than the one used for cadA led to S7) and oxygen concentration was reduced (Fig. 7E), condi- increased itaconic acid concentrations and specific productivity, tions which should limit the oxygenation reaction of RuBisCO. reaching more than 700 mg L1 final titers (Fig. 2). Reverse labeling experiments provided a further proof of the In addition to itaconic acid, we also showed that lactic acid incorporation of captured CO2 into the organic acids produced. can be produced successfully with the synthetic autotrophic With natC CO2 as the carbon source we observed a decreasing 13 K. phaffii strain. Integration of a bacterial lactate dehydrogenase C labeling degree over time. After 8 d, the 13C labeling degree gene was sufficient to produce lactic acid titers of 200 mg L1 had dropped to 18% and 10% for itaconic and lactic acid, (Fig. 1F). However, the produced lactic acid also served as a respectively, indicating that the carbon is indeed derived from potential carbon source (Fig. 3) during the growth in the shake CO2 in the production phase. The faster reverse labeling flask screenings as the cells were facing limited cultivation con- obtained with lactic acid can be explained by being metabolically ditions while growing on CO2 as the sole C source. Therefore, closely (and linearly) linked to the CBB cycle. For itaconic acid our first engineering target was the reduction of lactic acid con- the intermediate TCA cycle pools first need to be enriched with sumption. With a single knockout of CYB2, which encodes for label before complete enrichment of the product is obtained. a L-lactate cytochrome c oxidoreductase located in the mito- We also observed some incorporation of 12C atoms into the chondrial intermembrane space, growth on lactate was nearly organic acids by the nongrowing strains. The 13C labeling fully abolished. The specific lactic acid consumption rate degree had dropped to 75% for itaconic acid and 68% for lac- decreased threefold in the final production strain expressing tic acid respectively at the end of the cultivation of the non- ldhL and harboring the CYB2 knockout. Similar to S. cerevisiae growing strains. The unlabeled starting biomass (∼5%) might Cyb2 (26) it is also the only enzyme responsible for lactic acid have caused an increase in the 12C content of the produced oxidation in wildtype K. phaffii. The small amounts of lactic organic acids. In addition, natural CO2-fixing pathways might acid consumption which was observed in the production strain have also contributed to the incorporation of 12CO2 into the could be explained by the heterologous lactate dehydrogenase produced organic acids, thereby increasing the 12C content. performing the reverse reaction. The oxidation of lactate by One apparent pathway is the anaplerotic reaction, carboxylating LdhL is unfavorable compared to the reduction with a KM for pyruvate into oxaloacetate by the native pyruvate carboxylase. lactate more than 20× higher compared to pyruvate, neverthe- Another possible explanation for the itaconic acid producing less the results indicate that there is a certain degree of lactate strains could be the reverse TCA cycle. To assess the possibili- oxidation catalyzed by the LdhL (32). The decrease in lactate ties, we modeled the theoretical isotopologue distribution of oxidation in the cyb2Δ strain is probably the main reason for the products using the metabolic model described above (SI the increased lactic acid titers in our final production strain. An Appendix, Table S2) where either forward or reverse TCA cycle PNAS 2022 Vol. 119 No. 47 e2211827119 https://doi.org/10.1073/pnas.2211827119 7 of 10 reactions are active. However, we could not verify whether the compared to those where products can potentially serve as a car- reverse TCA cycle is active or not. The incorporation of CO2 bon source (like lactic acid), especially when assimilation of the through natural metabolic pathways still remains to be resolved. carbon source is limited and/or energy intensive (47). Therefore, K. phaffii is one of the suitable platforms for the production target products have to be carefully evaluated for potential indus- of recombinant proteins, enzymes or chemicals (33). To assess trial application of such a system. the feasibility to produce chemicals with the synthetic chemo- The titers we reached in shake flask cultivations (up to 2 g L1) autotrophic K. phaffii, we investigated also process conditions. are still far from industrially feasible processes, but through further Each product has its own requirements, and the process needs metabolic and process engineering steps, the synthetic CO2 assimi- to be optimized accordingly. Increasing the initial biomass did lating K. phaffii can be a platform for production of value-added not increase lactic acid titers, while starting with a higher bio- chemicals and form a chassis for a CO2-neutral or even CO2- mass increased the final titers for itaconic acid (at a cost of negative bioprocesses. lower specific production rates). One reason is that unlike itac- onate, lactate can be used as a carbon source by K. phaffii. Even Materials and Methods though the knockout of CYB2 blocked the lactic acid assimila- tion, a small portion could be still consumed in strains express- Construction of the Strains. The synthetic autotrophic K. phaffii strain ing ldhL (Fig. 3). In contrast to lactic acid production, itaconic described by Gassler et al. (13) was used as the host to build the organic acid acid production was significantly improved when the CO2 con- producing strains. Cis-aconitate decarboxylase (cadA) and the mitochondrial centration was increased from 5 to 10%, showing that the CO2 carrier protein mttA from Aspergillus terreus (24) were expressed under metha- availability at high concentration is one of the key steps to nol inducible promoters. For the lactic acid producing strains the ldhL gene of increase productivity. This led to an almost 2-fold increase in Lactobacillus plantarum was expressed under the control of the methanol induc- the final titers reaching ∼2 g L1 itaconic acid at the end of ible AOX1 promoter. The deletion of CYB2 was done using CRISPR-Cas9 and a the process. However, considering a well-known bioprocess like homologous donor DNA repair template deleting the gene. All the plasmid con- citric acid production which yields more than 200 g L1, the structions and transformations were performed by Golden Gate Assembly (GGA) organic acid titers we reached in a shake flask are still far from and the CRISPR-Cas9 system described in Prielhofer et al. (25) and Gassler et al. a feasible industrial process (23). So far, the highest itaconic (48). Integration to the correct loci and the correct gene deletion were verified acid titers in bioreactors reached by a native producer and an by PCR. A list of the strains used in this study is given in Table 1. engineered yeast is 160 and 55 g L1, respectively (34, 35). We Shake Flask Cultivations. A two-phase screening protocol was used in this could produce up to 600 mg L1 of lactic acid using our auto- study. First, a preculture in YPG (yeast extract 10 g L1, soy peptone 20 g L1, tropic K. phaffii strain which is already close to the titers glycerol 20 g L1) was conducted for 22 to 24 h until enough biomass was reached in cyanobacteria of ∼1 g L1 (36, 37) but still far away Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. reached. Then, the preculture volume required for the inoculation was washed from industrially relevant titers of far over 100 g L1 (38). twice and cells were transferred into the main culture with the target starting However, upscaling the process to bioreactors was not straight- OD600 (4, 8, 10, 12, 16, or 20) for the cultivation under autotrophic conditions forward. Using the best process conditions for organic acid pro- where cells were cultivated in 100 mL narrow-neck flasks at 30 °C with 5% or duction at different oxygen availability, we could not obtain as 10% constant CO2 supply in buffered YNB (3.4 g L1, pH 6) supplemented high titers as in the shake flask screenings. Nevertheless, it is with 10 g L1 (NH4)2SO4 as the nitrogen source. Cells were induced with 0.5% apparent that a fine balancing between the carboxylation and (vol/vol) methanol at the start of the main culture, and further adjusted to 1% oxygenation reaction of RuBisCO enzyme is crucial. Oxygen is (vol/vol) methanol from there on until the end of the cultivations. Cell growth required for the oxidation reaction of methanol to harvest energy (OD600) was monitored throughout the cultivation and extracellular metabolite and an important factor for the production of itaconic acid concentrations (methanol, lactic acid, itaconic acid, and glycolic acid) were mea- (29, 39), but also acts as a competitor for RuBisCO decreasing sured by high-performance liquid chromatography (HPLC) and culture volume the CO2 fixing rates (40) which eventually leads to lower was corrected for evaporation by the addition of water. organic acid production. Therefore, maintaining the balance Cultivations performed with lactic acid as sole carbon source were performed between the carbon dioxide and oxygen transfer rates is of great by using 100 mL narrow-neck flasks at 25 °C in buffered YNB (3.4 g L1, pH 6) importance here to fine-tune the anabolic reactions and energy supplemented with 10 g L1 (NH4)2SO4 as the nitrogen source and 10 g L1 requirements of the cell to reduce by-product formation and lactic acid as carbon source. Cells were inoculated with the target starting OD600 reach higher productivities. One approach here could be replac- of 4 and samples for optical density and HPLC (for remaining lactic acid concen- ing alcohol oxidase by a methanol dehydrogenase to decrease tration) were taken regularly. the oxygen demand and eliminate the oxygenation side reac- HPLC Measurements. A Biorad Aminex HPX-87H HPLC column (300 × tion. Other process conditions including pH, temperature, and methanol feeding regime known to be affecting lactic or ita- 7.8 mm) was used for the HPLC measurements. H2SO4 at a concentration of 4 mmol L1 was used as mobile phase, with a 0.6 mL min1 flow rate at 60 °C. conic acid production or the production capacity of K. phaffii Itaconic acid was measured with a photodiode array detector (SPD-M20A, Shi- should also be optimized (41–44). Additionally, redirecting the madzu) at 254 nm. Lactic acid, glycolic acid, glycerol, and methanol concentra- flux toward target product through metabolic engineering cou- tions were measured with a refraction index detector (RID-10A, Shimadzu). After pled with a model-based approach (45) might help us to deter- centrifugation, the supernatant of each sample was mixed with H2SO4 (c = mine the bottlenecks in the metabolism and increase the 40 mmol L1) resulting in a final concentration of 4 mM. Samples were vor- productivity. texed and centrifuged at full speed (16,100 g) for 5 min at room temperature. We have demonstrated that synthetic autotrophic K. phaffii After centrifugation, they were filtered using a 0.22 μm filter into the vials for is able to produce organic acids by CO2 assimilation. In a the HPLC analysis. CO2-based process, organic acids are suitable chemicals (46) to produce, as they have intermediate oxidation states which make Bioreactor Cultivations. Bioreactor cultivations were performed using 1.4 L them feasible products by using an oxidized substrate, like DASGIP reactors (Eppendorf). Cultivations were conducted using YNB media sup- CO2. Our results support the metabolic engineering concept plemented with 10 g L1 (NH4)2SO4 as the nitrogen source and buffered using that pathways containing an irreversible step (like decarboxyl- 100 mmol L1 phosphate buffer at pH 6 at 30 °C. pH was controlled using ation of cis-aconitate to itaconic acid) can run more efficiently 2 mol L1 NaOH. Dissolved oxygen concentration was controlled by adjusting 8 of 10 https://doi.org/10.1073/pnas.2211827119 pnas.org Table 1. K. phaffii strains used in this study Short Name Genotype References Control CBS7435 aox1Δ::pAOX1_TDH3 + Gassler et al. (13) pFDH1_PRK + pALD4_PGK1 das1Δ::pDAS1_RuBisCO + pPDC1_groEL + pRPP1B_groES das2Δ::pDAS2_TLK1 + pRPS2_TPI1 cadA Control + pAOX1_cadA This study cadA + mttAstrong Control + pAOX1_cadA + pFDH1_mttA This study cadA + mttAmedium Control + pAOX1_cadA + pPOR1_mttA This study cadA + mttAweak Control + pAOX1_cadA + pPDC1_mttA This study ldhL Control + pAOX1_ldhL This study ldhL cyb2Δ Control + pAOX1_ldhL cyb2Δ This study Δ RuBisCO ita Control ΔRuBisCO + pAOX1_cadA + This study pPOR1_mttA Δ RuBisCO lac Control ΔRuBisCO + pAOX1_ldhL cyb2Δ This study Control (eng strain) Control (PRK 5 C > G) Gassler et al. (22) ldhL (eng strain) Control (eng strain) + pAOX1_ldhL This study ldhL cyb2Δ (eng strain) Control (eng strain) + pAOX1_ldhL cyb2Δ This study cadA (eng strain) Control (eng strain) + pAOX1_cadA This study the stirrer speed and inlet gas flow whereby 200 rpm and 6 sL h1 were the coupled time of flight mass spectrometer). Itaconic and lactic acid concentrations minimal setpoints. (SI Appendix, Fig. S4) were determined by HPLC as described above. Three bio- The itaconic acid fermentations were performed as follows: bioreactors were logical replicates were used for isotopologue distribution analysis. inoculated from a YPG preculture to an OD of 1 and a glycerol batch in YNB 13 Elemental Analysis–Isotope-Ratio Mass Spectrometry for C Biomass media was performed to reach an end biomass of approximately OD 20 and 30, 13 Content. To verify C labeling of the biomass a fraction of the cells was har- respectively, based on a yield on glycerol of 0.5. After the batch-end cultures vested after 13C glycerol cultivation and centrifuged at 4 °C for 5 min at 4,000 g. were fed with 0.5% (vol/vol) methanol and supplied with 10% CO2 in the inlet Biomass corresponding to ∼0.5 mg DCW was washed once with filter sterilized gas. After the first sample (∼16 h) methanol concentration was adjusted to Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. 0.1 mol L1 HCl and twice with sterile distilled water. The 13C/12C ratio was mea- 1% (vol/vol). From this time on samples were taken daily including OD, DCW, sured with an isotope-ratio mass spectrometer coupled to an elemental analyzer. and HPLC samples. After each sampling the methanol concentration was Elemental analysis–isotope-ratio mass spectrometry (EA–IRMS) measurements adjusted to 1% (vol/vol). The lactic acid fermentation was performed as follows: were carried out by Imprint Analytics GmbH, Austria as described in Gassler et al. bioreactors were inoculated from a YPG preculture to an OD of 2. The autotrophic cultivation was performed from beginning on using 5% CO2 in the inlet gas and (13). Samples were kept at 20 °C until analysis. 0.5% methanol. After the first sample methanol concentration was increased to GC-TOFMS Isotopologue Distribution Analysis of Lactic Acid and 1%. Samples were taking daily including OD and HPLC measurements. The Itaconic Acid in the Culture Supernatant. As the workflow published by methanol concentration was adjusted to 1% after 2 or 3 d. Mairinger et al. (49), which is used routinely in our laboratory for the determina- Overall specific growth rate and production rates of the producer strains in tion of 13C labeling patterns, includes neither lactic nor itaconic acid as target the shake flasks and bioreactor cultivations were calculated according to Eqs. 1 analytes, a GC-EI-TOFMS method was developed for the selective determination and 2: of the 13C labeling patterns of the two analytes of interest in culture superna-  tants. More specifically, derivatization and separation strategies had to be estab- lnX2 X1 μ= [1] lished. In brief, 0.5 mL of the supernatants were filtered using 10 kDa spin filters t2  t1 (Millipore Amicon Ultra). Aliquots of 9 to 120 μL were dried in 1.5 mL chroma- qP = μ  YP=X [2] tography crimp vials supplied with 400 μL flat bottom inserts, resulting in μ is the specific growth rate (h1), X is the cell concentration (g L1) at t (h) of on-column amounts of 10 to 12 pmol for itaconic acid and 20 to 22 pmol for the cultivation, qP is the specific production rate (mg g1 h1), YP/X is the prod- lactic acid (estimation of aliquot volume needed was based on HPLC data uct yield (mg g1 h1). (SI Appendix, Fig. S4). Immediately after drying the samples were stored on a For the nongrowing strains, production rates were calculated according to Eq. 3: cooled rack (7 °C) of the Gerstel MPS2 sample preparation robot until just in time derivatization and injection into the GC-(Q)TOFMS. For automated tert- P2  P1 qP = X X1 [3] butyldimethylsilyl derivatization, 40 μL of water free pyridine (Sigma Aldrich) 2  ðt2  t1 Þ and 40 μL of N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide with 1% tert- 2 butyldimethylchlorosilane (Sigma Aldrich) were added and the samples were qP is the specific production rate (mg g1 h1), P is the product concentration incubated at 85 °C for 60 min. 1 μL of derivatized sample was injected into an (mg L1) at t (h) of the cultivation, X is the cell concentration (g L1) at t (h) of the Ultra Inert straight split liner of an Agilent split/splitless injector in split mode cultivation. (250 °C, septum purge flow 3 mL min1). The split ratio was set to 1:5, 1:10 or 13 C Labeling. The incorporation of carbon from CO2 into the produced organic 1:50 targeting the desired on-column amount, hence depending on analyte con- acids was determined with a reverse 13C labeling approach. To fully label bio- centration and aliquot volume. The gas chromatographic system (Agilent 7890B mass with 13C, cells were grown on YNB with 13C glycerol until the entire glyc- GC, Agilent Technologies Inc.) was equipped with 1) a nonpolar deactivated pre- erol was depleted (>10 generations). After washing, cells were transferred into column [3 m × 0.25 μm inner diameter (i.d.), Phenomenex], followed by 2) the YNB and induced with 13C labeled methanol as described above. The cells were analytical column [HP-5MS UI (5% phenylpolysiloxane, 30 m × 0.25 mm × fed with 5% natCO2 (CO2 with a natural isotope distribution) for growth and pro- 0.25 μm, Agilent] and 3) a nonpolar deactivated restrictor column (3 m × duction. Samples were taken at Day 3, Day 5, and Day 8. After harvesting, cells 0.18 μm i.d., Phenomenex). All columns were connected with purged ultimate were centrifuged at 4,000 g for 5 min at 4 °C and supernatants were kept at unions, (Agilent Technologies Inc.). The carrier gas was helium (purity >99.999%) 20 °C until further analysis with GC-TOFMS (gas chromatography with a at a constant flow of 1.2 mL min1 for the precolumn, 1.3 mL min1 for the PNAS 2022 Vol. 119 No. 47 e2211827119 https://doi.org/10.1073/pnas.2211827119 9 of 10 analytical column and 1.5 mL min1 for the post column. For the separation the Data, Materials, and Software Availability. All study data are included in following GC-temperature program was employed (total run time 29.3 min): the article and/or supporting information. 95 °C for 1 min, 10 °C min1 to 135 °C, 3 °C min1 to 170 °C, 30 °C min1 to ACKNOWLEDGMENTS. This work was supported by the Federal Ministry for 310 °C, hold for 8 min. The transfer line temperature was set to 280 °C. Digital and Economic Affairs, the Federal Ministry for Climate Action, Environ- As mass analyzer, an Agilent Technologies 7200B Q-TOF mass spectrometer ment, Energy Mobility, Innovation and Technology, the Styrian Business Promo- (Agilent Technologies Inc.) was used in TOF mode with electron ionization (70 eV, tion Agency SFG, the Standortagentur Tirol, the Government of Lower Austria 230 °C). The solvent delay was set to 5.5 min and the ion source was turned off and ZIT - Technology Agency of the City of Vienna through the COMET Funding during elution of the highly concentrated phosphate derivative (17 to 19 min). Program managed by FFG. We thank the Austrian Science Fund for support to Retention times of itaconic and lactic acid, mass/charge ratios evaluated for 13 D.M., S.H., M.B. and B.M.M. (FWF W1224, Doctoral Program on Biomolecular C isotopologue distribution analysis and the respective extraction windows of € (FWF M2891). O.A.€ Technology of Proteins (BioToP)), as well as support to O.A. the fragments are listed in SI Appendix, Table S3. and D.M. were additionally supported by the VIVALDI project which has received For peak integration, the software MassHunter Workstation, Quantitative funding from the European Union’s Horizon 2020 research and innovation pro- Analysis (Version 10.1, Agilent Technologies Inc.) was used in centroid and pro- gramme under grant agreement No 101000441. This project was supported by file mode, using a symmetric extraction window of ± 50 ppm. The data inter- EQ-BOKU VIBT GmbH and the BOKU Core Facility Mass Spectrometry. preted in this paper are those obtained by evaluation of the fragment [M-CH3] + in centroid mode (for further information on method optimization see SI Appendix). To receive the isotopologue distribution of the carbon backbone only, Author affiliations: aAustrian Centre of Industrial Biotechnology (ACIB), Vienna, 1190, the peak areas were corrected for naturally distributed heavy stable isotopes Austria; bDepartment of Biotechnology, Institute of Microbiology and Microbial using the ICT correction toolbox (50). This program corrects for Si, S, O, N and H Biotechnology, University of Natural Resources and Life Sciences (BOKU), Vienna, 1190, Austria; cDepartment of Chemistry, Institute of Analytical Chemistry, University of isotopes of the derivatized molecule as well as the C isotopes stemming from Natural Resources and Life Sciences (BOKU), Vienna, 1190, Austria; and dPresent derivatization (i.e., all C atoms apart from backbone C atoms). address: Institute of Microbiology, ETH Zurich, Zurich, 8093, Switzerland 1. P. Friedlingstein et al., Global carbon budget 2021. Earth Syst. Sci. Data 14, 1917–2005 (2022). 26. B. Guiard, Structure, expression and regulation of a nuclear gene encoding a mitochondrial 2. C. B. Field, M. J. Behrenfeld, J. T. Randerson, P. Falkowski, Primary production of the biosphere: protein: The yeast L(+)-lactate cytochrome c oxidoreductase (cytochrome b2). EMBO J. 4, Integrating terrestrial and oceanic components. Science (80-.) 281, 237–240 (1998). 3265–3272 (1985). 3. H. Kirst, C. Formighieri, A. Melis, Maximizing photosynthetic efficiency and culture productivity in 27. M. K. M. Engqvist et al., Glycolate oxidase 3, a glycolate oxidase homolog of yeast L-lactate cyanobacteria upon minimizing the phycobilisome light-harvesting antenna size. Biochim. cytochrome c oxidoreductase, supports L-lactate oxidation in roots of Arabidopsis. Plant Physiol. Biophys. Acta 1837, 1653–1664 (2014). 169, 1042–1061 (2015). 4. C. Bi et al., Development of a broad-host synthetic biology toolbox for Ralstonia eutropha and its 28. J. D. Young, INCA: A computational platform for isotopically non-stationary metabolic flux analysis. application to engineering hydrocarbon biofuel production. Microb. Cell Fact. 12, 107 (2013). Bioinformatics 30, 1333–1335 (2014). 5. C. Liu, B. C. Colon, M. Ziesack, P. A. Silver, D. G. Nocera, Water splitting–biosynthetic system with 29. A. Li et al., Reduced by-product formation and modified oxygen availability improve itaconic acid CO2 reduction efficiencies exceeding photosynthesis. Science (80-.) 352, 1210–1213 (2016). production in Aspergillus niger. Appl. Microbiol. Biotechnol. 97, 3901–3911 (2013). 6. J. Lu, C. J. Brigham, C. S. Gai, A. J. Sinskey, Studies on the production of branched-chain alcohols 30. L. van der Straat et al., Expression of the Aspergillus terreus itaconic acid biosynthesis cluster in in engineered Ralstonia eutropha. Appl. Microbiol. Biotechnol. 96, 283–297 (2012). Aspergillus niger. Microb. Cell Fact. 13, 11 (2014). Downloaded from https://www.pnas.org by 158.109.129.22 on November 17, 2022 from IP address 158.109.129.22. 7. Z. Liu, K. Wang, Y. Chen, T. Tan, J. Nielsen, Third-generation biorefineries as the means to produce 31. A. H. Hossain et al., Rewiring a secondary metabolite pathway towards itaconic acid production in fuels and chemicals from CO2. Nat. Catal. 3, 274–288 (2020). Aspergillus niger. Microb. Cell Fact. 15, 130 (2016). 8. F. E. Liew et al., Carbon-negative production of acetone and isopropanol by gas fermentation at 32. A. Feldman-Salit et al., Regulation of the activity of lactate dehydrogenases from four lactic acid industrial pilot scale. Nat. Biotechnol. 40, 335–344 (2022). bacteria. J. Biol. Chem. 288, 21295–21306 (2013). 9. B. Schiel-Bengelsdorf, P. D€urre, Pathway engineering and synthetic biology using acetogens. 33. B. Gasser, D. Mattanovich, A yeast for all seasons—Is Pichia pastoris a suitable chassis organism for FEBS Lett. 586, 2191–2198 (2012). future bioproduction? FEMS Microbiol. Lett. 365, 1–4 (2018). 10. J. R. Phillips et al., Butanol and hexanol production in Clostridium carboxidivorans syngas 34. S. Krull, A. Hevekerl, A. Kuenz, U. Pr€uße, Process development of itaconic acid production by a fermentation: Medium development and culture techniques. Bioresour. Technol. 190, 114–121 natural wild type strain of Aspergillus terreus to reach industrially relevant final titers. (2015). Appl. Microbiol. Biotechnol. 101, 4063–4072 (2017). 11. H. N. Abubackar, A. Fernandez-Naveira, M. C. Veiga, C. Kennes, Impact of cyclic pH shifts on carbon 35. L. Rong et al., Engineering Yarrowia lipolytica to produce itaconic acid from waste cooking oil. monoxide fermentation to ethanol by Clostridium autoethanogenum. Fuel 178, 56–62 (2016). Front. Bioeng. Biotechnol. 10, 888869 (2022). 12. J. Li et al., Coordination of consolidated bioprocessing technology and carbon dioxide fixation to 36. S. A. Angermayr et al., Exploring metabolic engineering design principles for the produce malic acid directly from plant biomass in Myceliophthora thermophila. Biotechnol. photosynthetic production of lactic acid by Synechocystis sp. PCC6803. Biotechnol. Biofuels 7, Biofuels 14, 186 (2021). 99 (2014). 13. T. Gassler et al., The industrial yeast Pichia pastoris is converted from a heterotroph into an 37. A. M. Varman, Y. Yu, L. You, Y. J. Tang, Photoautotrophic production of D-lactic acid in an autotroph capable of growth on CO2. Nat. Biotechnol. 38, 210–216 (2020). engineered cyanobacterium. Microb. Cell Fact. 12, 117 (2013). 14. S. Gleizer et al., Conversion of Escherichia coli to generate all biomass carbon from CO2. Cell 179, 38. J. Y. Song et al., Introduction of a bacterial acetyl-CoA synthesis pathway improves lactic acid 1255–1263.e12 (2019). production in Saccharomyces cerevisiae. Metab. Eng. 35, 38–45 (2016). 15. B. Rego de Vasconcelos, J. M. Lavoie, Recent advances in power-to-X technology for the production 39. A. Li et al., A clone-based transcriptomics approach for the identification of genes relevant for of fuels and chemicals. Front Chem. 7, 392 (2019). itaconic acid production in Aspergillus. Fungal Genet. Biol. 48, 602–611 (2011). 16. Organic Acids Market To Reach USD 36.86 Billion By 2026 j Reports And Data. https://www. 40. I. Andersson, A. Backlund, Structure and function of Rubisco. Plant Physiol. Biochem. 46, 275–291 globenewswire.com/en/news-release/2019/08/22/1905650/0/en/Organic-Acids-Market-To-Reach- (2008). USD-36-86-Billion-By-2026-Reports-And-Data.html (April 14, 2022). 41. A. Hevekerl, A. Kuenz, K. D. Vorlop, Influence of the pH on the itaconic acid production with 17. Bio-succinic Acid Market Worth $272.4 Million By 2030. https://www.grandviewresearch.com/ Aspergillus terreus. Appl. Microbiol. Biotechnol. 98, 10005–10012 (2014). press-release/bio-succinic-acid-market# (April 14, 2022). 42. Z. Yang, Z. Zhang, Engineering strategies for enhanced production of protein and bio-products in 18. Market volume of lactic acid worldwide from 2015 to 2021, with a forecast for 2022 to 2029. Pichia pastoris: A review. Biotechnol. Adv. 36, 182–195 (2018). https://www.statista.com/statistics/1310495/lactic-acid-market-volume-worldwide/ (31 October 43. B. C. Saha, Emerging biotechnologies for production of itaconic acid and its applications as a 2022). platform chemical. J. Ind. Microbiol. Biotechnol. 44, 303–315 (2017). 19. Global Acetic Acid Market to Reach 11.85 Million Tons by 2026. https://www. 44. M. Sauer, D. Porro, D. Mattanovich, P. Branduardi, 16 years research on lactic acid production with expertmarketresearch.com/pressrelease/global-acetic-acid-market (April 14, 2022). yeast—Ready for the market? Biotechnol. Genet. Eng. Rev. 27, 229–256 (2010). 20. Citric Acid Market Size, Share, Analysis, Report 2022-2027. https://www.expertmarketresearch. 45. B. J. Harder, K. Bettenbrock, S. Klamt, Model-based metabolic engineering enables high yield com/reports/citric-acid-market (April 14, 2022). itaconic acid production by Escherichia coli. Metab. Eng. 38, 29–37 (2016). 21. J. C. De Carvalho, A. I. Magalh~aes, C. R. Soccol, Biobased itaconic acid market and research 46. M. G. Steiger, D. Mattanovich, M. Sauer, Microbial organic acid production as carbon dioxide sink. trends-is it really a promising chemical? Chim. Oggi/Chemistry Today 36, 56–58 (2018). FEMS Microbiol. Lett. 364, 1–4 (2017). 22. T. Gassler, M. Baumschabl, J. Sallaberger, M. Egermeier, D. Mattanovich, Adaptive laboratory 47. D. Porro, P. Branduardi, M. Sauer, D. Mattanovich, Old obstacles and new horizons for microbial evolution and reverse engineering enhances autotrophic growth in Pichia pastoris. Metab. Eng. chemical production. Curr. Opin. Biotechnol. 30, 101–106 (2014). 69, 112–121 (2022). 48. T. Gassler, L. Heistinger, D. Mattanovich, B. Gasser, R. Prielhofer, CRISPR/Cas9-mediated 23. M. G. Steiger, M. L. Blumhoff, D. Mattanovich, M. Sauer, Biochemistry of microbial itaconic acid homology-directed genome editing in Pichia pastoris. Methods Mol. Biol. 1923, 211–225 production. Front. Microbiol. 4, 23 (2013). (2019). 24. M. G. Steiger, P. J. Punt, A. F. J. Ram, D. Mattanovich, M. Sauer, Characterizing MttA as a 49. T. Mairinger et al., Gas chromatography-quadrupole time-of-flight mass spectrometry-based mitochondrial cis-aconitic acid transporter by metabolic engineering. Metab. Eng. 35, 95–104 determination of isotopologue and tandem mass isotopomer fractions of primary metabolites for (2016). (13)C-metabolic flux analysis. Anal. Chem. 87, 11792–11802 (2015). 25. R. Prielhofer et al., GoldenPiCS: A Golden Gate-derived modular cloning system for applied 50. C. Jungreuthmayer, S. Neubauer, T. Mairinger, J. Zanghellini, S. Hann, ICT: Isotope correction synthetic biology in the yeast Pichia pastoris. BMC Syst. Biol. 11, 123 (2017). toolbox. Bioinformatics 32, 154–156 (2016). 10 of 10 https://doi.org/10.1073/pnas.2211827119 pnas.org