Production of mannosylglycerate in Saccharomyces cerevisiae by metabolic engineering and bioprocess optimization

Strain construction and evaluation of gene overexpression

Physiological characterization of engineered strains

Cultivation conditions exert a great impact in the performance of producing strains. In particular, S. cerevisiae is known to accumulate ethanol as a major by-product in aerobic conditions under certain circumstances, and these different metabolic states may affect the production of the desired compound. When there is a low glucose uptake rate, S. cerevisiae exhibits a respiratory metabolism only, with no ethanol formation, while, when the glucose concentration and/or uptake rate surpasses a critical threshold value, the metabolism becomes a combination of respiration and alcoholic fermentation [26]. In this work, the engineered strains were grown under conditions that imposed a respiratory vs respiro-fermentative metabolism; the fitness and MG accumulation parameters were determined under these conditions.

Physiological characterization of engineered strains in shake flasks and batch bioreactors

The engineered strains were initially characterized in shake flasks using synthetic complete medium (SC medium) with 20 g L⁻¹ of glucose. MG01 and MG02 exhibited similar growth rates (0.24 h⁻¹ for MG01 vs 0.25 h⁻¹ for MG02), but biomass production was different: 0.13 ± 0.01 g_DCW g _glc ⁻¹ for MG01 and 0.10 ± 0.00 g_DCW g _glc ⁻¹ for MG02. Profiles of growth, glucose consumption, ethanol, glycerol and acetate production are shown in Additional file 1: Figure S1 indicates a higher production of ethanol and growth for strain MG01 but the same production of glycerol and acetate for both strains. The intracellular content of MG was determined after glucose depletion and, under these conditions, MG02 showed a 1.5-fold increase in MG production compared to the reference strain MG01. A comparison between shake flask and controlled bioreactor for both strains is depicted in Additional file 1: Table S1. S. cerevisiae GDP-mannose pool had been successfully increased before by the over-expression of the gene MPG1 (PSA1) in a multi-copy plasmid [23]. In the present work, it was possible to increase MG production by over-expressing PSA1 (MPG1) and PMI40, which re-directed part of the glycolytic flux towards the formation of GDP-mannose and consequently to MG.

In order to check whether MG accumulation could be enhanced by a controlled environment, strains MG01 and MG02 were grown in a 2-L bioreactor. In controlled batch cultivation, MG02 produced 15.86 mg g _DCW ⁻¹ of MG, which represents an increase of 2.2-fold in comparison to MG01 (Fig. 2a). In controlled batch cultivation, using a bioreactor, it is possible to maintain optimal conditions regarding pH, concentration of oxygen and temperature. Under these conditions, strain MG02 produced higher amounts of MG than in shake flasks. Moreover, MG01 and MG02 showed no detectable difference in the production of end-products ethanol, glycerol and acetate (Fig. 2b).

Physiological characterization of engineered strains in chemostat at different dilution rates

Since the relative contribution of respiratory and respiro-fermentative regimens in S. cerevisiae depends on glucose availability, it is difficult to achieve a pure respiratory regimen in controlled batch cultivations. Therefore, a clear distinction of the two metabolic states is only possible in fed-batch and chemostat cultures. In an aerobic environment and in the presence of high concentrations of glucose, S. cerevisiae has a fermentative/respiratory metabolism due to a limited respiratory capacity [26]. In addition, there is a molecular response whereby the TCA cycle is repressed and the metabolic flow towards fermentation products is increased, a phenomenon designated as the Crabtree effect [27]. This behavior happens independently of the presence of oxygen to support respiration and is seen as an adaptive response to a competitive environment, as fermentation increases S. cerevisiae’s growth rate. Nevertheless, it is possible to ensure a respiratory metabolism as long as the glucose uptake rate is kept below the critical threshold that activates fermentation.

Additionally, the ionic nature of MG might be unbalancing cell homeostasis and therefore reducing cell fitness for MG02, as 15.86 mgMG g _DCW ⁻¹ represents an intracellular concentration of approximately 25 mM, which is a substantial value for an organism that intrinsically accumulates non-charged osmolytes, like glycerol and trehalose [1, 30]. Increasing concentrations of MG produces an acidification of the cytoplasm (pKa of MG is 3.35) which can compromise growth [31, 32] and impair essential metabolic functions [33]. S. cerevisiae cultivated in aerobic conditions with glucose as carbon source, switches from respiratory-fermentative growth to mainly fermentative growth in the presence of high concentrations of acetic acid, that shares with MG the characteristics of a weak acid. Under these conditions, a significant decrease in the specific rate of O₂ consumption was observed while high level of CO₂ were maintained [36], as observed for the MG02 strain. Additionally, the metabolic shift to a fermentative behavior (higher RQ) observed in MG02 has been described before for other inhibiting compounds such as furfural, a compound known to inhibit glycolytic enzymes, react with cellular constituents and compromise the respiratory activity [34, 35], although the mechanisms of action are probably different from weak acids.

Additionally, steady-state MG02 cells cultivated under a dilution rate of 0.1 h⁻¹ showed a profile of ethanol production that is largely influenced by the concentration of ethanol produced during growth in the batch mode. Chemostat experiments were initiated with different ethanol concentrations, which apparently produced different results for the ethanol concentration at steady-state, explaining the high standard deviation associated with ethanol production observed in Table 2. However, the rate of ethanol decrease from the beginning of the chemostat until steady-state is very similar for both experiments (see Additional file 1: Figure S2). The fact that the initial ethanol concentration is not fully washed from the bioreactor during the continuous cultivation at low dilution rates may be another indication that both strains are not in full respiratory regimen [37]. However, the influence of initial ethanol concentration in continuous cultures in the respiratory capacity has not been, to our knowledge, previously reported and remains unexplained.

To evaluate the impact of increasing the dilution rate, we performed chemostat experiments at dilution rates close to the maximum (washout) values (0.15 h⁻¹ for MG02 and 0.2 h⁻¹ for MG01). The difference between MG01 and MG02 in MG production observed at 0.1 h⁻¹ is kept, while the productivity of both strains in chemostats at 0.1 h⁻¹ dilution rate is lower when compared to chemostats performed close to maximal dilution rates (Additional file 1: Table S2). Curiously, MG yield on biomass is not influenced by the variation on the dilution rate, either for MG01 or MG02, although for MG02 the MG yield on substrate is larger for the lower dilution rate.

In order to understand the role of different metabolic states in the production of MG and to gain insight into the respiratory capacity, cells from MG02 were cultivated in a chemostat at dilution rate of 0.05 h⁻¹ to allow for fully respiratory mode. A comparison of the three dilution rates for this strain shows that, as expected, the biomass yield on glucose decreases with high dilution rates (Fig. 3a). In aerobic glucose-limited chemostat cultures, it is well-known that when yeast cells are cultivated at high rates (close to the maximal growth rate), the flux from glycolysis is re-directed to the production of by-products (e.g. ethanol) and respiration is kept at lower levels. However, when the dilution rate is reduced, respiration becomes the major source of energy through TCA cycle and consequently, the production of biomass increases [37]. Also, by gradually decreasing the dilution rate in MG02, we verified a decrease in the production rate of MG (Fig. 3b). However, there is no correlation between the dilution rate and MG yield on glucose (Fig. 3c) or even production on biomass (remains the same for dilutions 0.15 and 0.1 h⁻¹ (Additional file 1: Table S2) and decreases to 8.21 mg_MG g _DCW ⁻¹ at dilution 0.05 h⁻¹). Despite the lack of correlation with dilution, MG seems to reach a peak of production in terms of yield per substrate at 0.1 h⁻¹ dilution rate (Fig. 3c). In light of these results, we can conclude that MG production does not respond linearly to a variation on the growth rate; however, as cells are growing more rapidly, MG productivity is higher at higher dilution rates. Another important result is the variation of RQ values with the dilution rates examined. Coherently with previously published results [29], RQ decreases at lower dilutions and strain MG02 is virtually at a fully respiratory state at dilution 0.05 h⁻¹ (Fig. 3d), corroborated by the absence of glycerol production and decrease on acetate/ethanol yields which dropped to 0.1 g g _glc ⁻¹ .

Continuous mode chemostat presents several advantages in comparison with batch cultures and it is a valuable approach to explore the best conditions for industrial production. One advantage, which is exploited in this work, is the evaluation under controlled conditions of physiological parameters based on growth rates [38]. The cultivation of S. cerevisiae in respiro-fermentative metabolic mode holds some issues as it is accompanied by the accumulation of by-products and low biomass yield. Additionally, ethanol production might interfere with the productivity of compounds by provoking toxicity [39]. For this reason, industrial bioprocesses usually use a dilution rate of 0.1 h⁻¹ or lower [38] as is the case of resveratrol production [40]. Nonetheless, the respiro-fermentative metabolism favors the production of certain compounds. In a recent study, the cultivation conditions for ethylene production were optimized and were found to be coupled with the growth rate [41]. MG is produced from GDP-mannose and 3-phosphoglycerate (3PG), both derived from glycolysis. As the respiro-fermentative metabolism of S. cerevisiae implies a higher glycolytic flux than respiration, it is reasonable to speculate that MG production would increase with higher growth rates in chemostat. Additionally, it has been described that PSA1 is an essential gene linked to the progression of cell cycle, since efficient glycosylation of proteins is a key-point for cell division [42]. Even more, PSA1 transcript levels increase with higher growth rates [43], which may favor the synthesis of GDP-mannose and ultimately the synthesis of MG. In the present work, this correlation between MG production and growth rates was not observed. The fact that MG production was not dependent on growth rate may indicate that there is a bottleneck downstream of GDP-mannose, i.e., in the production of MG which is catalyzed by the bifunctional enzyme mannosylglycerate synthase/phosphatase.

DNA manipulation

Saccharomyces cerevisiae genomic DNA for colony PCR and gene amplification was prepared as described by Lõoke et al. [44]. Primers were purchased from Metabion (Germany). Gene amplification and colony PCR were performed using, respectively, Phusion polymerase and DreamTaq DNA polymerase (both enzymes from Thermoscientific, USA). PCR reactions were performed in a Thermal Cycler from Bio-Rad (USA), following the protocols recommended by the manufacturers for each polymerase. Plasmid extraction, PCR product purification, and DNA gel extraction were carried out with Zymo Research kits (USA). All restriction enzymes were obtained in fast-digestion format from Thermoscientific (USA). The plasmid pDES, containing the gene mgsD that codes for mannosyl-3-phosphoglycerate synthase/phosphatase from D. mccartyi under the control of the ENO2 promoter from enolase II, was kindly provided by Milton S. da Costa (University of Coimbra, Portugal) [17]. S. cerevisiae transformation was performed using the lithium acetate method as described by Gietz et al. [45]. Escherichia coli DH5α was used for plasmid isolation and maintenance following the competence and transformation procedures developed by Dower et al. [46].

Construction of engineered strains

The plasmid pSP-GM was kindly provided by Prof. Jens Nielsen’s group (Chalmers University of Technology, Sweden) [25]. This plasmid was used to express the mannose-6-phosphate isomerase (PMI40) and GDP-mannose pyrophosphorylase (PSA1) encoding genes from S. cerevisiae. The gene PMI40 was amplified using genomic DNA isolated from S. cerevisiae strain CENPK2-1C with the primers pair 5′-CCGCGGCCGCAAAAAAATGTCCAACAAGCTGTTCAGG-3′ and 5′-CCGAGCTCCTAATTTGGTTCCACAAAGGC-3′. This PCR product was digested with SacI/NotI and ligated into pSP-GM between promoter of TEF1 and terminator of ADH1 using a T4 ligase (Thermoscientific, USA), yielding pSP01. Next, the gene PSA1 was amplified using again S. cerevisiae strain CENPK2-1C genomic DNA with primers pair 5′-GGCCCGGGAAAAAAATGAAAGGTTTAATTTTAGTCGG-3′ and 5′-CCAAGCTTTCACATAATAATAGCTTCCTTTGG-3′. This PCR product was digested with HindIII/XmaI and ligated into pSP01 using a T4 ligase between promoter of PGK and terminator of CYC1, resulting in plasmid pSP02. Colonies harboring pSP01 and pSP02 were identified by performing colony PCR with the primers used in the amplification. Correct constructions of plasmids were confirmed by restriction analysis and DNA sequencing (GATC Biotech, Germany).

Strain maintenance and cultivation media

Selection and maintenance of plasmids in E. coli was performed in LB medium containing 10 g L⁻¹ of peptone, 10 g L⁻¹ of NaCl, 5 g L⁻¹ of yeast extract, and supplemented with 100 mg L⁻¹ of ampicillin. The solid LB medium also included 20 g L⁻¹ of agar. All cultivations of E. coli were carried out at 37 °C and 200 rpm of agitation.

Saccharomyces cerevisiae strain CENPK2-1C (MATa ura3-52 his3-Δ1 leu2-3,112 trp1-289, MAL2-8c SUC2) was cultivated in YPD medium, containing 10 g L⁻¹ of yeast extract, 20 g L⁻¹ of peptone, and 20 g L⁻¹ of glucose, at 30 °C and 160 rpm. Recombinant strains were selected on Synthetic complete (SC) medium containing 6.7 g L⁻¹ of yeast nitrogen base without amino acids from Difco (USA), 20 g L⁻¹ of glucose and 20 g L⁻¹ of agar. When necessary, histidine (20 mg L⁻¹), tryptophan (20 mg L⁻¹), uracil (20 mg L⁻¹) and leucine (30 mg L⁻¹), were added to cover auxotrophies. All cultivations of S. cerevisiae were performed at 30 °C with 160 rpm agitation.

The SC medium used for shake flasks, batch in controlled bioreactors, and continuous cultivation contained 6.7 g L⁻¹ of yeast nitrogen base without amino acids from Difco (USA) supplemented with 20 g L⁻¹ of glucose. For strain MG01, SC medium was supplemented with histidine (20 mg L⁻¹), tryptophan (20 mg L⁻¹), and uracil (20 mg L⁻¹), while for strain MG02 it was complemented only with histidine and tryptophan, in the same concentrations [48]. For shake flask cultivations, MG01 and MG02 were pre-grown in 50 mL of SC medium at 30 °C and 160 rpm. Then, cells were used to inoculate 500 mL shake flasks containing 100 mL of SC medium with an initial OD₆₀₀ of 0.1. Growth was followed using a spectrophotometer V-560 from Jasco (Japan); cells were harvested at the end of the exponential phase (24 h upon inoculation).

Batch cultivations were performed as following: firstly, cells were grown overnight in shake flasks from a single colony. Each fermenter was inoculated at an initial OD₆₀₀ of 0.1. The batch fermentations were performed in a BIOSTAT-B Plus system (Sartorius, Germany) coupled with a 2-L culture vessel. The operating conditions are the following: the fermentation volume was 0.8 L, the temperature was maintained at 30 °C, the airflow was 1 vvm, the pH was kept at 5.5 controlled by addition of 2 M NaOH, and the dissolved O₂ was kept above 30% of saturation by feedback control of the stirring speed from 200 rpm up to 600 rpm. Concentrations of O₂ and CO₂ in the exhaust gas were monitored by Bluesens gas analyzers (Germany) and RQ represents the ratio between CO₂ produced and O₂ consumed during fermentation. Biological samples were collected at the end of the exponential phase, when CO₂ production dropped drastically.

Continuous cultivations were also carried out in a Sartorius BIOSTAT-B Plus system coupled with a 2-L culture vessel with a constant working volume of 0.8 L, with the same controlled variables and set-points as before. The medium described above was used to feed the bioreactor at dilution rates of 0.05, 0.1, 0.15 and 0.2 h⁻¹. The volume was kept at 0.8 L by controlling the level of broth inside the vessel. To guarantee that the culture reached a steady-state mode, at least 5 volumes of medium were allowed to pass through the culture whenever the dilution rate was changed. Samples were collected when O₂ consumption and CO₂ production were constant.

Quantification of MG, biomass and fermentation end products

For each sampling, 50 mL of broth were removed from the culture and centrifuged at 3600×g for 10 min to separate cells from the medium. Next, the supernatant was filtered through a membrane filter with a pore size of 0.22 μm into HPLC vials and stored at − 20 °C until further analyzed for glucose, glycerol, acetate, and ethanol. MG is an intracellular compound and was extracted from the cell pellet by adding 5 mL of water, 5 mL of methanol and 10 mL of chloroform. This mixture was vigorously shaken with the help of a vortex and then centrifuged at 12,000×g for 30 min at 4 °C. The top part, corresponding to the aqueous phase, was carefully transferred to a new tube and the process of centrifugation was repeated after the addition of 5 mL of water. Once the aqueous phase was separated, samples were evaporated using a Savant™ SPD131DDA SpeedVac (Thermoscientific, USA). Next, the evaporated samples were dissolved in 1 mL of water and transferred to an HPLC vial. Analysis was performed in an HPLC apparatus from Jasco (Japan) model LC-NetII/ADC equipped with UV-2075 Plus and RI-2031 Plus detectors, also from Jasco. The samples were analyzed using an Aminex HPX-87H column from Bio-Rad, which was kept at 50 °C, and a solution of 0.01 M of H₂SO₄ was used as the mobile phase with a flow rate of 0.3 mL min⁻¹. Quantitative analysis of glucose, MG, glycerol, acetate and ethanol was performed by comparison with a mixture of standards with known concentrations of each metabolite. MG standards were obtained from cell biomass of Rhodothermus marinus as previously described [49]. Calibration curves were prepared using the peak areas of the RI detector for glucose, glycerol, acetate and ethanol and of the UV detector at 210 nm for MG. The cell dry weight was determined by filtering 5 mL of culture broth through a 0.22 μm pore Millipore filter and washing once with 5 mL of water. Filters were then dried for 10 min at 150 W in a microwave oven and weighted using an analytical balance.

Quantification of enzymatic activities

For the determination of mannose-6-phosphate isomerase and GDP-mannose pyrophosphorylase activities, strains WT and MG02 were grown in shake flasks until mid-exponential phase. Cells were harvested by centrifugation and re-suspended in water with a cocktail of protein inhibitors (1×), following manufacturer’s instructions (Roche, USA). Next, cells were broken with three pulses of vortexing (4.0 m s⁻¹ for 60 s) using a Fastprep 24 system (MP, USA), keeping cells on ice for 5 min between each pulse. Supernatants were obtained by centrifugation at 13,500×g for 5 min and used to determine enzymatic activities. Total protein was determined using the Bradford method [50]. The spectrophotometric methods used to determine the enzymatic activities have been described by Bergmeyer [51]. Briefly, for the mannose-6-phosphate isomerase activity, the assay mixture (final volume of 1 mL) contained: 50 mM Tris–HCl (pH 7), 5 mM mannose-6-phosphate, 8 mM MgCl₂, 2.8 mM NADP, 0.27 U glucose-6-phosphate dehydrogenase, 0.27 U phosphoglucose isomerase, and 0.25 mg of cell extract. Reactions were started by the addition of mannose-6-phosphate and absorbance was measured at 340 nm for 2 min. For the GDP-mannose pyrophosphorylase activity, the assay mixture (final volume of 1 mL) contained: 50 mM of Tris–HCl (pH 7), 0.5 mM NaF, 10 mM MgCI₂, 0.1 mM ADP, 1 mM glucose, 1 mM NADP, 2 mM PPi, 5.5 mM GDP-mannose, 1 U each of nucleoside kinase, hexokinase, and glucose-6-P dehydrogenase, and 0.25 mg of cell extract. The reaction was started by the addition of GDP-mannose, and the formation of NADPH was monitored by recording the absorbance at 340 nm for 2 min. One unit corresponds to the conversion of 1 µmol of substrate per minute.