Engineered yeast with a CO2-fixation pathway to improve the bio-ethanol production from xylose-mixed sugars

Bio-ethanol production from lignocellulosic raw materials could serve as a sustainable potential for improving the supply of liquid fuels in face of the food-to-fuel competition and the growing energy demand. Xylose is the second abundant sugar of lignocelluloses hydrolysates, but its commercial-scale conversion to ethanol by fermentation is challenged by incomplete and inefficient utilization of xylose. Here, we use a coupled strategy of simultaneous maltose utilization and in-situ carbon dioxide (CO2) fixation to achieve efficient xylose fermentation by the engineered Saccharomyces cerevisiae. Our results showed that the introduction of CO2 as electron acceptor for nicotinamide adenine dinucleotide (NADH) oxidation increased the total ethanol productivity and yield at the expense of simultaneous maltose and xylose utilization. Our achievements present an innovative strategy using CO2 to drive and redistribute the central pathways of xylose to desirable products and demonstrate a possible breakthrough in product yield of sugars.

Scientific RepoRts | 7:43875 | DOI: 10.1038/srep43875 S. cerevisiae may lead to cofactor imbalance 22 . On the one hand, NADPH for XR is supplied by glucose-6-phosphate (G6P) metabolism through oxidative branch of PPP, but gluconeogenesis pathway for conversion of xylose to G6P is limited when S. cerevisiae grows on xylose 23,24 . On the other hand, catalysis of NAD + to NADH by XDH usually facilitates the formation of by-products such as glycerol and xylitol 14,22 . Many efforts have been devoted to relieve the cofactor imbalance. Use of the mutant XR (mXR) that exhibits higher preference for NADH 25 or the mutant XDH (mXDH) using nicotinamide adenine dinucleotide phosphate (NADP + ) 26 showed positive effects on xylose utilization. Besides, by co-expressing a heterogeneous acetate-consumption pathway 27 or a carbon dioxide (CO 2 )-fixation pathway 28 in the yeasts with XR and XDH, a decrease of by-products and an increase of ethanol from xylose were also observed since acetate acid and CO 2 could be used as electric acceptor for NADH oxidation.
Therefore, the present research constructed a PRK-Rubisco module together with a mXR-XDH module into S. cerevisiae. Then, the resulted yeast strains were applied for xylose fermentation with simultaneous utilization of maltose and CO 2 (Fig. 1). This strategy may have three major benefits: (i) The mXR-XDH module containing a mutant XR (R276H) preferring NADH, a XR, a XDH and a XK could improve the ethanol productivity and decrease the by-product accumulation. (ii) Catabolism of maltose could simultaneously supply energy and cofactors to meet the requirement for xylose metabolism. (iii) The CO 2 fixation pathway may bypass the glucose repression on xylose catabolism inside the cell. The CO 2 produced during fermentation can also form an appropriate atmosphere for Rubisco activity and a redox sink for improving the cofactor balance.

Results
Construction of xylose metabolic pathway in S. cerevisiae. We obtained a xylose-fermenting yeast strain YSX4 (Table 1) with co-expression of wild-type XR, mutant XR (R276H), XDH and XK in S. cerevisiae YS58. In previous work, the same xylose metabolic pathway has been used in an engineered S. cerevisiae DA24 29 . XR (R276H) was found to exhibit much higher preference for NADH whereas XR showed two-fold higher preference for NADPH 25 . Theoretically, co-expression of wild-type XR and mutant XR (R276H) could enable YSX4 with comparable XR activities towards NADPH and NADH.
The fermentation performance of the resulted YSX4 was evaluated under an oxygen-limited condition with xylose and glucose, or xylose and maltose as carbon source respectively (Table 2). We found that the xylose consumption rate of YSX4 (0.57 g/L/h) was similar with that of DA24 (0.53 g/L/h) on a glucose-xylose mixture, although the key enzymes involved in the xylose metabolic pathway in YSX4 were expressed by using a multi-copy plasmid, while in DA24, they were integrated in genome. Interestingly, compared with the glucose-xylose  mixture, YSX4 exhibited a higher efficiency of xylose utilization (0.70 g/L/h) when grown on xylose and maltose, which is in agreement with previous results using cellobiose and xylose as carbon source 29,30 . Evidences showed that intracellular hydrolysis of cellobiose minimized glucose repression on xylose uptake, and simultaneous fermentation of cellobiose and xylose was capable of improving ethanol productivity when compared to fermentation with xylose only 30 . For both of maltose and cellobiose, disaccharides are formed from two glucose molecules. The difference is that glucose units are joined with an α (1 → 4) bond in maltose and a β (1 → 4) bond in cellobiose respectively. Therefore, we suspected that YSX4 could also co-ferment maltose and xylose, thus consumption of xylose by YSX4 was enhanced when cells were grown on maltose-xylose mixture.
Functional expression of CO 2 -fixation pathway in S. cerevisiae. As a heterotrophic microorganism, S. cerevisiae intracellularly converts ribulose-5-phosphate (Ru5P) to glycerate-3-phosphate (G3P) through the non-oxidative branch of PPP without consumption of CO 2 . In contrast, many autotrophic microbes achieve this conversion by a two-step reaction using ribulose bisphosphate carboxylase-oxygenase (Rubisco) and phosphoribulokinase (PRK) 31,32 . PRK catalyzes Ru5P to ribulose-1,5-bisphosphate (Ru1,5BP), while Rubisco converts Ru1,5BP to G3P with CO 2 fixation. There are four known forms of Rubisco in nature 33 . Form-I Rubisco (L 8 S 8 ) is composed of eight large subunits and eight small subunits, of which the small subunits are able to concentrate surrounding CO 2 to improve the reactivity of the large subunits. Form-II (L 8 ) only contains eight large subunits. In previous works, the gene of cbbM from Rhodospirillum rubrum or Thiobacillus denitrificans coding for Form-II Rubisco and the gene of sPRK from Spinacia oleracea coding for PRK have already been successfully expressed in S. cerevisiae 28,34 . Interestingly, it was found that functional expression of cbbM in yeast depended on co-expression of a chaperone (GroEL-GroES) from Escherichia coli. In this work, we engineered S. cerevisiae YS58 with a Form-II based CO 2 fixation system by co-expression of cbbM, sPRK, and GroEL-GroES, resulting in a strain named YSC111. Similarly, in another yeast strain called YSC222, cbbL1-cbbS1 (Form-I Rubisco gene) and cfxP1 (PRK gene) from Ralstonia eutropha H16 35 constructed a Form-I based CO 2 fixation system. To ensure the functionality of Form-I Rubisco, an endogenous chaperone (Hsp60-HSP10) of S. cerevisiae YS58 was simultaneously over-expressed.
To test whether the CO 2 -fixation pathway was workable in the engineered S. cerevisiae, iodoacetate (IA) was used to inhibit the activity of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), thus repressing the cellular glycolysis. After addition of IA to 0.19 g/L (~0.001 mol/L), the ethanol productivity of YSC000 or YSC110 decreased sharply (Fig. 2a,b), but no negative effects on YSC111were observed (Fig. 2c). YSC111 also showed similar growth profiles with or without IA addition (Fig. 2d). The ethanol yields of YSC222 and YSC111 were slightly affected by IA, but the ethanol yields of YSC220 and YSC110 lacking of chaperones decreased by 27% (Table S1), indicating that chaperones were essential for functional expression of both Form-I and Form-II Rubisco. We also found that the performance of the engineered CO 2 -fixation pathway would be enhanced with increase of glucose concentrations (Fig. S1), which was probably due to more surrounding CO 2 present.
Then, YSX4C111 and YSX4C222 were applied in the medium containing 70 g/L maltose and 40 g/L xylose. As shown in Table 2, the xylose consumption rates of YSX4C111 and YSX4C222 arrived to 0.97 and 1.1 g/L/h respectively. The total sugar consumption rate and the ethanol yield of YSX4C222 reached 3.1 g/L/h and 0.47 g/g sugars, which was 63% and 15% higher than YSX4C000. YSX4C222 cell extract also showed the highest carboxylation activity (Fig. S2). These findings suggested that the Form-I based CO 2 fixation system was preferred in facilitation of the xylose metabolism.
Relative quantification of heterotrophic CO 2 fixation. To evaluate the strength of the CO 2 flux in YSX4C111 and YSX4C222, a reported metabolic flux index, MFI h-CO2 36 , was employed to indicate the metabolic flux ratio between the CO 2 -fixation and the PPP-based pathways. The G3P generated by these two pathways was differentiated by using 13 C-labeled CO 2 and unlabeled sugars to obtain the MFI h-CO2 value.
As illustrated in Fig. S3, we assumed that a mole of glyceraldehyde-3-phosphate (GADP) was metabolized to G3P, while b mole of 13 CO 2 from NaH 13 CO 3 and 12 CO 2 from sugars was fixed to G3P within a given time period. Moreover, we assumed that c mole of unlabeled G3P and d mole of 13 C-G3P were channeled into the downstream metabolism. Fermentation experiments were performed using 10 g/L maltose and 20 g/L xylose with 100 mM NaH 13 CO 3 and cell density of OD 600 = ~0.1. The amounts of 13C-labeled and unlabeled G3P were measured by LC-MS/MS. S. cerevisiae YS58 was cultivated in a medium free of any carbon isotope to determine the ratio of 13 C-G3P to the unlabeled G3P as the basal isotopic level, which was about 3.1%. The actually detected molar amount of 13 C-G3P (y) can be calculated by Eq. (1), while the actually detected unlabeled G3P (x) can be calculated by Eq. (2).
Under a metabolic steady-state, the relationship of d, c, x and y is shown in Eq. (3).
Therefore, the value of MFI h−CO2 can be calculated by Eq. (4).
h CO2 As shown in Fig. 4a, 13 C-G3P amount in YSX4C000 did not change during fermentation. In contrast, a significant increase of 13 C-G3P was observed in YSX4C111 and YSX4C222. These results demonstrated that CO 2 was able to be incorporated into G3P through the engineered CO 2 -fixation pathways. The MFI h-CO2 values of these strains at different times were calculated to evaluate the relative CO 2 flux (Fig. 4b), and the results indicated that over 8% of Ru5P from xylose was converted to G3P through the CO 2 -fixation pathway in YSX4C111 or YSX4C222. More interestingly, the higher levels of 13 C-G3P and MFI h-CO2 in YSX4C222 than in YSXC111 suggested that the Form-I based system was more efficient than Form-II in S. cerevisiae.  Determination of the CO 2 -fixation rate in the engineered yeasts. The carbon balance of wild-type S. cerevisiae 37 indicated that 96% carbon of the consumed sugars (C consumed sugars ) was incorporated into biomass (C biomass ), CO 2 , ethanol (C ethanol ), xylitol (C xylitol ), glycerol (C glycerol ) and Acetate (C acetate ). Therefore, the carbon of CO 2 fixed by YSX4C111 or YSX4C222 (C fixed CO2 ) can be calculated using Eq. (5). C CO2 with biomass and C CO2 with ethanol represent the CO 2 from cell respiration and from ethanol biosynthesis respectively. We assume that C CO2 with biomass is in direct proportion to C biomass . Therefore, Eq. (5) can be transformed into Eq. (6).
fixed CO2 biomass e thanol glycerol xylitol a cetate consumed sugars The carbon balance of YSX4C000 is shown as Eq. (7). Fermentation data of YSX4C000, YSX4C111 or YSX4C222 on 70 g/L maltose and 40 g/L xylose are summarized in Table 3. The results showed that our yeasts were able to fix CO 2 at a rate of 336.6-436.3 mg CO 2 /L/h, significantly exceeding the natural or the engineered microbes (5.8 to 147.0 mg CO 2 /L/h) in previous reports 36 (Table S2).

Discussion
Currently, xylose fermentation with the engineered S. cerevisiae is inferior to glucose fermentation by the wild-type S. cerevisiae. Besides cofactor imbalance of the engineered XR-XDH pathway, supply of NADPH was limited because expression of the enzymes involved in gluconeogenesis in S. cerevisiae cannot be upregulated on xylose. Some rate-limiting enzymes in the non-oxidative branch of PPP need to be over-expressed to enhance the xylose downstream metabolic flux. Here, we engineered S. cerevisiae to achieve efficient bio-ethanol production from maltose and xylose with CO 2 fixation. NADH could be reduced through the mXR and CO 2 -fixation pathway, while the NADPH for the wild-type XR and the adenosine triphosphate (ATP) for activating Ru5P to Ru1,5BP used in Rubisco reaction were provided by co-fermentation of maltose and xylose.
Xylose can enter the cell by facilitated diffusion through some yeast hexose transporters (HXTs) 38 . For engineered XR-XDH-carrying yeasts, intracellular xylose conversion is only slightly affected by catabolism of intracellular glucose, but xylose transport is strongly inhibited by extracellular glucose 39 . The engineered xylose transporters by non-rational mutagenesis or rational design have been used to improve simultaneous utilization of xylose and glucose 40,41 . However, the specific xylose transporters without glucose inhibition have still not been developed 42 . On the other hand, introduction of a cellodextrin transport (CDT-1) from Neurospora crassa into  the yeast with intracellular β -glucosidase (GH1-1) achieved simultaneous utilization of cellobiose and xylose and showed synergistic effects due to sufficient NADPH supply 29,30 . But in yeast, β -glucosidase (GH1-1) can convert cellobiose to cellodextrins (cellotriose, cellotetraose) by its transglycosylation activity, while cellodextrin transport (CDT-1) can transport cellodextrins and glucose into the medium. This problem resulted in low ethanol yield and productivity from cellobiose and xylose. In this study, we employed another disaccharide, maltose, which can be utilized as efficiently as glucose by the wild-type yeast, to facilitate xylose metabolism and achieved the synergistic effects.
Heterotrophic Form-I and Form-II based CO 2 fixation systems were successfully expressed in S. cerevisiae. We observed that the engineered yeast YSX4C111 and YSX4C222 grew better and exhibited higher ethanol productivity than YSX4C000. The results suggest that additional ATP consumption by implementing the CO 2 -fixation pathway had a less negative effect on cell growth than the beneficial effect that the pathway might bring during co-fermentation of xylose and maltose. With increase of sugar concentrations in fermentation, the engineered yeasts expressing CO 2 fixation pathway showed better ethanol productivity. The possible reason was more CO 2 production from sugars increased environmental CO 2 and HCO 3 − concentrations so that improved the carboxylation activity of Rubisco under oxygen-limited condition. The same effects were also observed by supplementation of external CO 2 34 . The activity of Rubisco could further be optimized through expressing heterogeneous carbonic anhydrase (CA) 36 , which catalyzes the reversible conversion of CO 2 and HCO 3 − , or constructing a microcompartment for Rubisco 43 , similar to the semi-permeable carboxysome in cyanobacteria.
At present, low efficiency of photosynthetic processes and electro-synthetic processes using photoautotrophic and chemoautotrophic microbes are limited in their growths and productivities 32,44 . Differing from autotrophic microbes, heterotrophic microbes such as S. cerevisiae, which has very short doubling time, have been widely applied in industrial production of bio-chemicals. The CO 2 fixation at the expense of sugars in heterotrophic S. cerevisiae presented high CO 2 -fixation rates and product yields. The construction of a functional CO 2 -fixation pathway for efficient ethanol production from xylose and maltose in S. cerevisiae would create a foundation for other biofuels and chemicals production from lignocellulosic biomass.
Scientific RepoRts | 7:43875 | DOI: 10.1038/srep43875 inoculated into 200 mL YP medium with appropriate sugars (maltose, glucose or xylose) in 500 mL flasks. All fermentation experiments were carried out at 30 °C and 200 rpm under oxygen-limited conditions. The initial cell density was adjusted to an OD 600 (optical density at 600 nm) of ~0.5. Rubisco activity was determined through Rubisco Carboxylation Activity Assay Kit (GENMED, Shanghai) by UV-visible Spectrophotometer SU-2000.
IA inhibition assay. The cells of YSC000, YSC110, YSC111 and YSC222 were inoculated into 200 mL YP medium with 70 g/L glucose in 500 mL flasks. The initial cell density was adjusted to an OD 600 of ~0.1. IA was added to the media at 8 h. Ethanol concentration and OD 600 were detected in whole fermentation process.
Analytical methods. Cell growth was monitored using UV-visible Spectrophotometer SU-2000 (OnLab Instruments). Maltose, glucose, xylose, xylitol, glycerol, acetate and ethanol were quantified by high performance liquid chromatography (Agilent 1200 Series HPLC system) equipped with a refractive index detector (Shimadzu, Japan) and an Bio-Rad Aminex HPX-87H organic acid analysis column (7.8 × 300 mm) which was maintained at 50 °C and used 0.05 mM sulfuric acids as mobile phase. The sample injection volume was 10 μ L and the flow rate was 0.6 mL/min. Metabolites was detected by liquid chromatography-mass spectrometry/mass spectrometry system (Agilent 6460 series LC-MS/MS system) with Agilent XDC18 column (5 uM, 150 mm × 4.6 mm) 36 . Di-n-butylammonium acetate (DBAA) and NaH 13 CO 3 were purchased from Sigma-Aldrich. Methanol was purchased from Fisher Scientific. The mobile phase was the mixture of solution A (water with 5 mM DBAA) and solution B (methanol with 5 mM DBAA) at the gradient shown in Table S6. The flow rate was 0.6 mL/min. The injection volume was 50 μ L and the column temperature was 40 °C. The negative ion and selected multiple reactions monitoring (MRM) mode were used for MS detection. All experiments were conducted at least in triplicate, and the error bars in the figures denote the standard deviation from the means of independent experiments.