Metabolic engineering of E. coli for efficient production of glycolic acid from glucose

Biochemical Engineering Journal 103 (2015) 256–262

Contents lists available at ScienceDirect

Biochemical Engineering Journal journal homepage: www.elsevier.com/locate/bej

Regular article

Metabolic engineering of E. coli for efficient production of glycolic acid from glucose Yu Deng a,b,∗ , Yin Mao a , Xiaojuan Zhang c a b c

The Key Laboratory of Industrial Biotechnology, Ministry of Education, Jiangnan University, 1800 Lihu Road, Wuxi, Jiangsu 214122, China National Engineering Laboratory for Cereal Fermentation Technology (NELCF), Jiangnan University, 1800 Lihu Road, Wuxi, Jiangsu 214122, China School of Pharmaceutical Science, Jiangnan University, 1800 Lihu Road, Wuxi, Jiangsu 214122, China

a r t i c l e

i n f o

Article history: Received 30 July 2015 Accepted 14 August 2015 Available online 20 August 2015 Keywords: Biosynthesis Fed-batch culture Metabolite over production Recombinant DNA Chromosome integration Glycolic acid

a b s t r a c t Glycolic acid is the smallest member of the ␣-hydroxy acid family. In order to produce glycolate from glucose via the glyoxylate shunt stably, one malate synthase gene aceB in Escherichia coli BW25113 was deleted by homologous recombination; another malate synthase gene glcB was then replaced by a DNA cassette WAK harboring isocitrate lyase gene (aceA), glyoxylate reductase gene (ycdW) and isocitrate dehydrogenase kinase/phosphatase gene (aceK). The above three genes were over-expressed in the chromosome of E. coli EYX-1WAK. This strain was then transferred 20 times on M9 medium to have a mutant strain: EYX-2 with a significantly improved growth rate. The glycolate yields of EYX-2 in the shaken flasks and the 5-L bioreactor using batch fermentation strategy under 2 vvm aeration and 800 rpm stirring speed were 0.33 g/g-glucose and 0.48 g/g-glucose, respectively. The fed-batch fermentation of EYX-2 on 120 g/L glucose had the highest titer of 56.44 g/L with 0.52 g/g-glucose yield in 120 h, and this is the highest reported glycolate yield ever. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Glycolic acid (HOCH2 COOH), or glycolate, is the smallest member of the ␣-hydroxy acid family [1]. Glycolic acid has dual functionalities with both alcohol and moderately strong acid groups on a very small molecule. Its properties make glycolic acid ideal for a broad spectrum of consumer and industrial applications, including the leather industry, the oil and gas industry, the laundry and textile industry and personal care products [2]. Glycolic acid is also used together with lactic acid to produce a co-polymer (PLGA) for medical applications, e.g., drug delivery [3]. Glycolic acid market was USD 93.3 million (40 million kg) in 2011, and expected to reach USD 203 million in 2018 [1]. Glycolic acid is naturally produced by a variety of microorganisms from ethylene glycol by oxidation [2], or from glycolonitrile by hydrolyzation [4]. Chemolithotrophic iron- and sulphur oxidizing bacteria were also used for producing glycolate [5]. However, the above methods required expensive and highly pollutant precur-

sors. Recently, microorganisms were engineered to over-express glyoxylate shunt to direct carbon flux from isocitrate to glycolate. The highest titer of glycolate: 56 g/L was reported by Dischert et al. using Escherichia coli by adding ∼500 g/L glucose [6]. Koivistoinen et al. engineered Saccharomyces cerevisiae and Kluyveromyces lactis to produce glycolate at 15 g/L using d-xylose and ethanol as substrates [1]. However, the previous work used replicating plasmids to over-express genes responding to glycolate synthesis. The stability of those plasmids was not reliable for industrial purpose. In this study, to achieve glycolate synthesis from glucose via the glyoxylate shunt stably, we deleted the malate synthase gene aceB and replaced malate synthase gene glcB with WAK cassette harboring ycdW and aceA-aceK under the control of a constitutive promoter in the chromosome [7]. The adaptive evolution was used to improve the growth of the engineered strain. The final strain was characterized under optimized fermentation parameters. 2. Materials and methods

∗ Corresponding author at: National Engineering Laboratory for Cereal Fermentation Technology (NELCF), Jiangnan University, 1800 Lihu Road, Wuxi, Jiangsu 214122 China. Fax: +86 510 85918309. E-mail addresses: [email protected], [email protected] (Y. Deng). http://dx.doi.org/10.1016/j.bej.2015.08.008 1369-703X/© 2015 Elsevier B.V. All rights reserved.

2.1. Strains and plasmids The strains and plasmids are shown in Table 1. E. coli BW25113, which was the strain used for chromosome modification with

Y. Deng et al. / Biochemical Engineering Journal 103 (2015) 256–262 Table 1 Strains and plasmids used in this study. Strains and plasmids Strains E. coli BW25113 BW25113pGLY-1 EXY-1 EXY-1-pGLY-1 EYX-1-glcBEYX-1-glcb-1pGLY-1 EYX-1WAK EYX-2 hhh Plasmids pGLY-1 pGLYPRBSaceAK pGLY- PRBS pGLY-PRBSFRT pGLY-WAK pKD46

pKD4 pCP20

Relevant genotype

Reference

F-, DE(araD-araB) 567, lacZ4787(del)::rrnB-3, LAM-, rph-1, DE(rhaD-rhaB) 568, hsdR514 BW25113 carring pGLY-1

[8]

aceB (BW25113) EXY-1 with plasmid pGLY-1 aceB, glcB (BW25113) EXY-1-glcB- with plasmid pGLY-1

This study This study This study This study

aceB, glcB::WAK cassette Evolved strain of EYX-1WAK

This study This study

This study

pCOLADuet-1harboring ycdW and aceA-aceK from E. coli MG1655 pGLY-1 harboring PRBSaceAK cassette

[7]

pGLY- PRBSaceAK harboring PRBS cassette pGLY- PRBS harboring FRTterminator cassette pGLY-1 harboring WAK cassette Temperature sensitive replication (repA101ts); encodes lambda Red genes (exo, bet, gam); native terminator (tL3) after exo gene; arabinose-inducible promoter for expression (ParaB); encodes araC for repression of ParaB promoter; Ampicillin resistant. This is a template plasmids for frt-flanked kan cassette pCP20 has the yeast Flp recombinase gene, FLP, chloramphenicol and ampicillin resistant genes, and temperature sensitive replication

This study

This study

This study This study [8]

[8] [8]

the help of pKD46 and pCP20 [8]. The plasmid pGLY-1 for overexpressing ycdW, aceA-aceK, was a gift from Dr. Prather’s group of MIT [7]. pGLY-1 was transformed to BW25113 strain forming BW25113-pGLY-1. All the primers used in this study are shown in Table S1. 2.2. Culturing conditions The E. coli strains used for transformation and chromosome modifications were cultured in Luria Broth (LB) medium. The fermentation medium was M9 minimal medium with the addition of glucose and NH4 Cl. The first round of strain tests were done in the 500 mL shaken flasks with addition of CaCO3 . The fermentations were done in the 5-L bioreactor (Eppendorf Bioflo 310). The pre-cultures were grown in the shaken flasks for overnight and 10% culture was inoculated into the bioreactor and then grew for 24–120 h. Cultures were carried out at 37 ◦ C. The pH was adjusted at pH 6.8 by addition of base (NH4 OH 7.5% W/W). The volumetric oxygen transfer coefficient (KL a) was described previously [9]. The fermentation was in discontinuous fed-batch mode, with a feed stock solution of concentrated glucose [10]. 2.3. Construction of EYX-1 and EYX-1-glcb-strains The general strategy for engineering E. coli to accumulate glycolate is shown in Fig. 1A. The protocol of deleting aceB was described previously [8]. The brief process is: the aceB deletion cassette was

257

amplified from pKD4 plasmid [8] by PCR using primers aceB-KO-F and aceB-KO-R flanked by 39-nt DNA oligos homologous to the upstream and downstream of aceB gene, respectively. The aceB deletion cassette was transformed to E. coli BW25113 with pKD46 by electroporation. The homologous recombination was to replace aceB gene by the aceB deletion cassette, and the antibiotic marker was deleted by FLP recombinase expressed on plasmid pCP20 [11]. The deletion of aceB was verified by PCR using primers aceB-veriF and aceB-veri-R. The resulting strain was designated as EYX-1. In EYX-1, glcB was deleted by using the same method as deleting aceB, forming a new strain EYX-1-glcb-. The primers used for glcB deletion are shown in Table S1.

2.4. Construction of WAK cassette for integration of genes into the chromosome All sequences of the synthetic cassette and primers were shown in Table S1. pGLY-1 was used for over-expressing glycolate pathway genes and its backbone was pCOLADuet-1. In order to replace glcB gene in EYX-1 strain, a DNA cassette WAK was constructed to replace glcB in the chromosome. The order of genes in the WAK cassette was (5 →3 ): constitutive promoter P(Lac) IQ (tggtacaaaacctttcgcggtatggcatgatagcgcc), ribsome binding site (RBS), aceA-aceK, P(Lac) IQ promoter, ribsome binding site (RBS), ycdW, terminator, FRT sequence, kanamycin resistant gene region and FRT sequence (Fig. 1B). The procedures of constructing WAK cassette are shown below: The first step was to insert P(Lac) IQ and ribosome binding site (RBS) to the upstream of aceA-aceK on pGLY-1 plasmid. P(Lac) IQ fused with ribosome binding site (RBS) was synthesized and flanked by AciI and NcoI restriction sites on 5 and 3 ends, respectively, and the above cassette was designated as PRBSaceAK(Table S1). PRBSaceAK cassette and plasmid pGLY-1 were digested by AciI and NcoI, and the larger piece of pGLY-1 was isolated from agarose gel. The digested PRBSaceAK and the above larger piece were mixed and ligated by T4 ligase, and then were transformed to E. coli DH5␣ strain for screening by electroporation. The insertion of PRBSaceAK on the upstream of aceA-aceK of pGLY-1 was verified by PCR using primers aceAKF and aceAKR (Table S1) and the plasmid with the right structure was designated as pGLY-PRBSaceAK. Second, in order to insert P(Lac) IQ and RBS on the upstream of ycdW on pGLY-PRBSaceAK, a DNA cassette PRBSycdW containing P(Lac) IQ fused with RBS was synthesized, and there were BsrGI and NdeI restriction sits at 5 and 3 ends of this cassette. The rest of the steps were the same as insertion of PRBSaceAK cassette described above. The insertion of PRBSaceAK on the upstream of aceA-aceK was verified by PCR using primers ycdWF and ycdWR (Table S1) and the resulting plasmid was designated as pGLY-PRBS. Third, in order to attach Kanamycin resistant gene region as well as FRT sequences to the end of ycdW on pGLY-PRBS, pGLY-PRBS was digested by XhoI and XmnI. An FRTterminator cassette including XhoI restriction site, terminator, FRT sequence and XmnI restriction site was synthesized. FRTterminator cassette was digested by XhoI and XmnI. The digested FRTterminator and pGLY-PRBS was ligated by T4 ligase to form pGLY-PRBS-FRT plasmid. The insertion of FRTterminator was analyzed by PCR using primers FRTTF and FRTTR (Table S1). To add FRT sequence to the other end of Kanamycin resistant gene, pGLY-PRBS-FRT was digested by NheI. A FRT sequence flanked by two 39-nt sequences homologous to 5 and 3’ ends of the digested plasmid pGLY-PRBS-FRT was synthesized (FRTKan cassette). The FRTKan cassette was assembled with linear pGLY-PRBS-FRT plasmid using Gibson Assembly [12], forming pGLY-WAK plasmid. WAK cassette was amplified from pGLY-WAK plasmid by PCR using primers WAKF and WAKR, which

258

Y. Deng et al. / Biochemical Engineering Journal 103 (2015) 256–262

Fig. 1. Strategy of over-producing glycolate in E. coli BW25113. (A) Glycolate synthesis pathway. Malate synthase gene aceB was deleted in this strain. Isocitrate lyase gene (aceA), glyoxylate reductase gene (ycdW) and isocitrate dehydrogenase phosphatase/isocitrate dehydrogenase kinase gene (aceK) were over-expressed to redirect isocitrate to glycolate in E. coli EYX-2 strain by replacing another malate synthase gene glcB. (B) Flowchart of constructing EYX-2 strain. Step 1: WAK cassette including aceA-aceK, ycdW was transformed to EYX-1 strain (aceB deletion strain). glcB was replaced by WAK cassette. Step 2: the Kanamycin resistant gene was deleted by FLP recombinase. Step 3: aceA-aceK, ycdW were expressed and glycolate was accumulated. Glu: glucose; G3P: glyceraldehyde 3-phosphate; PEP: phosphoenolpyruvate; PYR: pyruvate; OAA: oxaloacetate; MAL: malate; CIT: citrate; ICI: isocitrate; AKG: ␣Ketoglutarate; SUCC: succinyl-CoA; SUC: succinate; FUM: fumarate; GLYX: glyoxylate. red open cross: the pathway deactivated. red open dotted cross: the pathway was down-regulated.

were flanked by 39-nt sequences homologous to the upstream and downstream of glcB gene.

2.5. Integration of ycdW, aceA-aceK into the chromosome All the sequences of the synthetic cassettes and primers were shown in Table S1. There were two 39-nt sequences flanked to 5 end and 3 end of the WAK cassette, which were homologous to the upstream and downstream of glcB gene. The WAK cassette was transformed to EYX-1 strain for homologous recombination with the help of pKD46 plasmid. After incubating the colonies overnight, the colony PCR (forward primer: glcbVF and reverse primer glcbVR, in Table S1) were used to screen the mutants with the right WAK cassette. More than 100 reactions of PCR were done and 12 can-

didates were found having the WAK cassette. Subsequently, the chromosome of them were isolated and sequenced to verify ycdW, aceA and aceK in the chromosome. The kanamycin resistant gene was cut by FLP recombinase with expression of pCP20 [11]. The final strain was named EYX-1WAK.

2.6. Adaptive evolution of the EYX-1WAK strain EYX-1WAK strain was grown on M9 medium. The strain was transferred to the fresh medium every 24 h, when each culture was harvested for testing the existence of the cassette, cell density and metabolites. Transfer was done by subculturing at a final dilution of 1:100 every day. The chromosome DNA was isolated from each culture and amplified by PCR targeting the WAK cassette using

Y. Deng et al. / Biochemical Engineering Journal 103 (2015) 256–262

the primers WAKF and WAKR (Table 1S). The PCR products were verified by electrophoresis and re-sequencing. The transfer was stopped once the cell growth was stable for 3 days. The cell culture was diluted and spread on the plates to form single colonies. The single colonies were inoculated into the shaken flasks for testing the growth rate, glycolate yield and existence of the WAK cassette. The final strain was designated as EYX-2. 2.7. Metabolite analysis by HPLC The metabolites were analyzed on a Dionex HPLC equipped with an Aminex HPX-87 ion exchange column (Biorad), with 0.005 M H2 SO4 . Refractive Index Detector was used. 2.8. Enzyme activity assays The cells of E. coli strains on the mid-log phase were harvested. The cells were spin down by centrifugation (10,000 g for 3 min). The cell pellets were washed three times by distilled deionized water. The washed cells were then subjected to the sonicator treatment: 50% strength for 2 min on ice. The cell lysates were centrifuged (12,000 × g for 5 min) and the supernatant was harvested for enzyme assays. The specific activity was defined as 1 U/mg = ␮mole/min/mg-protein. 2.8.1. Malate synthase activity assay The general process of measuring malate synthase activity was described previously [13,14]. Malate synthase was assayed at room temperature in 96-well plates. Typically, 50 ␮l of 20 mM Tricine-HCl (pH 7.5) containing 5 mM MgCl2 , 0.8 mM EDTA, 2 mM glyoxylate and 2 mM acetyl-CoA were mixed with 50 ␮l supernatant of the cell lysate in Tricine-HCl (pH 7.5), 5 mM MgCl2 and 0.8 mM EDTA buffer and incubated at room temperature for 30 min. Protein concentrations were determined by the method of Bradford [15]. The enzyme-catalyzed reaction was stopped by adding DTNB to a final concentration of 2 mM in Tris–HCl (pH 8.0). The amount of CoA-SH released was measured by titrating the free thiol groups with the DTNB and measuring the change in absorbance at 412 nm [13]. 2.8.2. Isocitrate dehydrogenase activity assay Isocitrate dehydrogenase activity was measured by using the Assay Kits from Gen Way Biotech, Inc. The protocols were according to the manual of the manufacturer. 2.9. Quantification of mRNA transcript levels The transcript levels for the mRNAs corresponding to ycdW and aceA were measured using a RT-qPCR assay. The methods of doing RT-PCR were described previously [7]. 3. Results 3.1. Deactivation of malate synthase 3.1.1. Deletion of aceB In the glyoxylate pathway, malate synthase makes the net synthesis of one molecule of malate from two molecules of acetate [14] (Fig. 1A). There were two genes glcB and aceB encoding malate synthase. The aceB gene was deleted in the BW25113 strain by homologous recombination, and the resulting strain was named as EYX-1. The activity of malate synthase of this strain is shown in Table 2. Although the malate synthase activity was decreased, it was still high enough to convert glyoxylate to malate. In order to direct carbon flux through the glyoxylate shunt to glycolate, the pGLY-1 plasmid was transformed to the EYX-1 strain and BW25113 strain,

259

forming EYX-1-pGLY-1 strain and BW25113-pGLY-1, respectively. In Fig. 2, the EYX-1-pGLY-1 strain was able to produce 0.88 g/L of glycolate from 8 g/L glucose (yield: 0.11 g/g-glucose) on M9 medium, and the yield of this strain was 35.38% higher than that of the BW25113-pGLY-1 strain (0.65 g/L from 8 g/L glucose). 3.1.2. Integration of glycolate synthesis pathway genes in the chromosome of E. coli The glcB gene also encodes malate synthase. The initial idea was to delete glcB directly in EYX-1 strain to deactivate malate synthase, and then over-express plasmid pGLY-1. The glcB gene was deleted to form EYX-1-glcb-strain. After 24 h, there was not plasmid found during the fermentation (no plasmid was isolated from cell cultures) and the strain was growing very slowly. In order to stabilize the expressions of aceA-aceK and ycdW and improve the growth rate of the engineered strain, a DNA cassette WAK harboring aceA-aceK and ycdW genes was transformed to EYX-1 to replace glcB gene (Fig. 1B). In the WAK cassette, aceA-aceK genes were in the same operon under the control of a constitutive promoter P(Lac) IQ, and ycdW was on the other site under the control of P(Lac) IQ, followed by a terminator and kanamycin resistant gene (Kan). Kan was flanked by two FRT sequences, recognized by FLP recombinase. The WAK cassette was flanked by two 39-nt nucleotides homologous to the upstream and downstream of glcB gene. The WAK cassette replaced glcB gene by homologous recombination with the help of -Red system [16]. The kanamycin resistant gene was deleted by FLP recombinase targeting FRT sequences with expression of pCP20 plasmid, and the final strain was designated as EYX-1WAK with very slow growth on M9 medium. In order to enhance the cell growth, EYX-1WAK was subject to the adaptive evolution. The strain was transferred 20 times in the shaken flasks on M9 medium every day and the WAK cassette was determined to exist in the chromosome before every transfer. The evolution stopped when the OD kept the same for 3 days. After screening the single colonies, EYX-2 strain was picked for further study. The relative transcript levels of ycdW and aceA were studied in strains BW25113-pLGY-1, EYX-1-pGLY-1, EYX-1-glcB-pGLY-1 and EYX-2 (Fig. 2C and D). BW25113-pGLY-1 had the highest ycdW and aceA transcript levels (0.653 and 0.544, respectively). EYX-2 with ycdW and aceA integrated into the chromosome had the lowest ycdW and aceA transcript levels (0.488 and 0.379, respectively). The production of glycolate of EYX-2 is shown in Fig. 2A. EYX2 was cultured on M9 medium with 8 g/L glucose and the titer of glycolate was 2.66 g/L. The glycolate yield in EYX-2 (0.33 gglycolate/g-glucose) was much higher than the rest of strains involved in this study, and was the highest among all reported glycolate yields [1,7,10], although it was much lower than the theoretical yield of glycolate (0.85 g-glycolate/g-glucose). 3.2. Enzyme activities Isocitrate dehydrogenase kinase/phosphatase (aceK) is a bifunctional enzyme which can phosphorylate or dephosphorylate isocitrate dehydrogenase (IDH) on a specific serine residue [17]. There is a regulatory mechanism which enables bacteria to bypass the TCA cycle via the glyoxylate shunt in response to the source of carbon [18]. AceK was in the aceBAK operon of E. coli and the transcription of this operon was repressed by a regulator IciR, when E. coli was grown on the glucose or rich medium [18,19]. In order to decrease the isocitrate dehydrogenase activity, redirecting more carbon flux from isocitrate to glycolate, aceK should be over-expressed. However, if the expression of aceK was too high, IDH could be fully deactivated, leading to the block of TCA cycle. The specific activities of isocitrate dehydrogenase of BW25113, EYX-1, EYX-1WAK and EYX2 strains were 0.389, 0.336, 0.057 and 0.115 U/mg, respectively (Table 2). In the EYX-1WAK strain, the

260

Y. Deng et al. / Biochemical Engineering Journal 103 (2015) 256–262

Table 2 The specific malate synthase and isocitrate dehydrogenase activities of the E. coli strains. Strain

Specific malate synthase activity (U/mg)

Specific isocitrate dehydrogenase activity (U/mg)

Relevant genotype

BW25113 EYX-1 EYX-1-glcBEYX-2

0.132 (0.019) 0.036 (0.004) NA NA

0.389 (0.029) 0.336 (0.014) 0.057 (0.004) 0.115 (0.02)

Parent strain aceB aceB, glcB aceB, glcB::WAK cassette

expression of aceK into the chromosome led to the strong inhibition of isocitrate dehydrogenase activity, which was decreased from 0.336 U/mg in EYX-1 to 0.057 U/mg in EYX-1WAK (83.03 % reduction). Due to the significant reduction of isocitrate dehydrogenase activity in the EYX-1WAK strain, the TCA cycle was blocked, causing the poor growth of EYX-1WAK strain. In order to increase the growth rate, a mutant strain EYX-2 was selected via adaptive evolution, whose specific isocitrate dehydrogenase activity was 0.115 U/mg, much higher than the one in EYX-1WAK strain (0.057 U/mg). The malate synthase activity was shown on Table 2. In EYX-1 strain, the specific malate synthase activity cannot be neglected, although it was significantly decreased (0.036 U/mg) compared with the parent strain BW25113 (0.132 U/mg). After the deletion of both aceB and glcB genes, the activity of malate synthase was not detectable.

3.3. Fermentation of glycolate in the bioreactor 3.3.1. Effect of aeration rates and stirring speeds on glycolate production Because the deletions of two genes made EYX-2 grow much more slowly than BW25113, the initial glucose was 19.03 g/L, consumed within 24 h. Although glycolate was produced by glyoxylate shunt, the TCA cycle was still extremely important for cell growth aerobically. In order to balance the strength of TCA cycle and glycolate synthesis, the well-controlled aerated-fermentations were studied. In Fig. 3A, at low aeration rate (2 vvm and 1 vvm) with the high stirring speed (800 rpm), EYX-2 could produce almost the same amount of glycolic acid as the one at 4vvm aeration with 400 rpm stirring speed, for example, at 2 vvm and 800 rpm, the yield of glycolate was 0.48 g/g-glucose and at 1vvm and 800 rpm, the yield

Fig. 2. Characterization of different E. coli strains for glycolic acid production. (A) maximum glycolate titer from 8 g/L glucose; (B) Maximum OD on 8 g/L glucose; (C) ycdW relative transcript levels; (D) aceA relative transcript Levels. The engineered E. coli strains with the expressions of ycdW and aceA-aceK were cultured in the shaken flasks on M9 medium to test the production ability of glycolate as well as cell growth.

Y. Deng et al. / Biochemical Engineering Journal 103 (2015) 256–262

261

Fig. 3. Batch fermentation of glycolate by EYX-2 strain on glucose. (A) optimization of the fermentations under different aeration rates and stirring speed; B: the fermentation results of the EYX-2 strain with the highest glycolate yield. The EYX-2 strain was cultured in a 5-L fermentor on M9 medium to test the fermentation characteristics of the strain.

4. Discussion

Fig. 4. Fed-batch fermentation of glycolate in EYX-2 strain on the high concentration of glucose.

of glycolate was 0.47 g/g-glucose. These three conditions had the same KL a, around 16.8 h−1 . The details of the batch-fermentation are shown in Fig. 3B. The maximum glycolate titer was 8.96 g/L around 24 h. In addition to glycolate, acetate was another major metabolite excreted during fermentation. 3.3.2. Production of glycolate by fed-batch fermentation The accumulation of glycolate and acetate could be harmful to the cell growth and inhibit metabolisms in the batch fermentation. High concentration of glucose can also impair the cell growth and glycolate yield due the osmosis stress and end-product inhibitions. In order to avoid the inhibitions, the fed-batch fermentation was used in this study (Fig. 4). The initial glucose concentration was 32.2 g/L, and after 30 h, the concentrated glucose solution was gradually added into the M9 medium at a constant rate until the final concentration reached ∼30 g/L again. The above feeding strategy was repeated twice after the first feeding. The total glucose added into the bioreactor was 125.6 g/L, among which 108.3 g/L glucose was consumed within 120 h. The maximum titer of glycolate was 56.44 g/L with the yield 0.52 g /g-glucose, which was higher than those in the shaken flask (0.33 g/g-glucose) and the batch fermentation (0.47 g-glycolate/glucose). The maximum OD (600 nm) was 19.26, and the production of glycolate was growth-associated. Although the yield of glycolate by EYX-2 was still lower than the theoretical yield: 0.85 g-glycolate/g-glucose, 0.52 g/g-glucose was the highest reported glycolate yield.

Glycolic acid (glycolate) has dual functionalities with both alcohol and moderately strong acid functional groups on a very small molecule [2]. Glycolic acid could be biologically converted from ethylene glycol by alcohol oxidation and aldehyde oxidation. Wei et al. reported that by using Gluconobacter oxydans DSM2003, more than 70 g/L of ethylene glycol was converted to glycolic acid [20]. However, compared to glucose, ethylene glycol was much more expensive and it prevented the application of bioconversion method for producing glycolate in the industry. Before this study, there were several studies focusing on the production of glycolic acid by microorganisms from renewable, relatively cheap substrates such as glucose [1,10,21,22]. The genes encoding enzymes (aceA and ycdW) for producing glycolate from isocitrate were existing in E. coli. However, they were silent in most of time [7]. Although glycolate was found to be excreted in E. coli with over-expressions of aceA and ycdW genes on plasmid pGLY-1 [7]. The unstability of the plasmid pGLY-1 in E. coli during the fermentation was not appropriate for industrial production. To solve the above problem, instead of using plasmids to express genes, we integrated aceA-aceK and ycdW genes into the chromosome of E. coli by replacing glcB gene, forming EYX-1WAK strain. This strain grew very slowly, because of the imbalanced carbon flux distributions between glyoxylate shunt and TCA cycle, which were controlled by isocitrate lyase and isocitrate dehydrogenase, respectively. To improve the cell growth, EYX-1WAK strain was evolved on M9 medium and a mutant strain EYX-2 with a significantly higher cell growth was obtained. The isocitrate dehydrogenase activity of EYX-2 strain was much higher than its parent strain EYX-1WAK, but significantly lower than EYX-1 and BW25113. The low activity of isocitrate dehydrogenase resulted to the weakness of TCA cycle, causing the slow growth of EYX-2 compared to the parent strain. The yield of glycolate by EYX-2 in fed-batch fermentation was much higher than batch fermentation because the stress caused by glycolate was partially eliminated by dilution. In order to avoid salt stress to E. coli, caused by the massive sodium and potassium from base solutions, ammonia was used to control pH during the fermentation rather than NaOH or KOH. In addition to that, ammonia was also a nitrogen source, which could balance the carbon/nitrogen ratio to direct most of the carbon flux to glycolate.

262

Y. Deng et al. / Biochemical Engineering Journal 103 (2015) 256–262

Dischert, et al. reported their methods engineering E. coli to produce glycolic acid [6]. They first over-expressed the aceA and ycdW genes to direct carbon from isocitrate to glycolate, and then attenuated icd gene, whose enzyme responsible for converting isocitrate to ␣-ketoglutarate and the last one was the deletion of the genes consuming glycolate. The strain they made had very low yield (0.29 g/g-glucose) with very weak growth. Then, they engineered the strain to accumulate biomass at high temperatures and produce glycolic acid at low temperatures [21]. The yield of the resulting strain AG1385 was 0.38 g/g-glucose with 51.3 g/L of titer. The EYX-2 strain in this study had a much higher yield: 0.52 g/gglucose with the titer of 56.44 g/L. There is still room to optimize the fermentation conditions to increase the titer of glycolic acid in the future. The most important advantage of this strain is that it could produce decent amount of glycolate with relatively fast growth rate. Compared to E. coli AG1385, there was no need to change the temperatures during the fermentation by using EYX-2 and it could reduce a lot of cost of heating and equipment for industry. With E. coli, the fermentation conditions are limited due to the growth of neutral pH. The benefit with many fungal species is their tolerance towards low pH conditions. Koivistoinen et al. engineered yeasts to produce glycolate from various substrates [1]. However, the titer of glycolate from their methods was up to 15 g/L, much low than the one reported in this study. Acknowledgements This work was supported by the grant of the Key Laboratory of Industrial Biotechnology, Ministry of Education (KLIB-KF201403). We thank Dr. Kristala Prather of the department of chemical engineering in Massachusetts Institute of Technology generously gave us plasmid pGLY-1. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bej.2015.08.008. References [1] O.M. Koivistoinen, J. Kuivanen, D. Barth, H. Turkia, J.P. Pitkanen, M. Penttila, P. Richard, Glycolic acid production in the engineered yeasts Saccharomyces cerevisiae and Kluyveromyces lactis, Microb. Cell Fact. 12 (2013) 82–97. [2] M. Kataoka, M. Sasaki, A.R. Hidalgo, M. Nakano, S. Shimizu, Glycolic acid production using ethylene glycol-oxidizing microorganisms, Biosci. Biotechnol. Biochem. 65 (2001) 2265–2270.

[3] S. Fredenberg, M. Wahlgren, M. Reslow, A. Axelsson, The mechanisms of drug release in poly(lactic-co-glycolic acid)-based drug delivery systems—a review, Int. J. Pharm. 415 (2011) 34–52. [4] Y.C. He, J.H. Xu, J.H. Su, L. Zhou, Bioproduction of glycolic acid from glycolonitrile with a new bacterial isolate of Alcaligenes sp. ECU0401, Appl. Biochem. Biotechnol. 160 (2010) 1428–1440. [5] I. Nancucheo, D.B. Johnson, Production of glycolic acid by chemolithotrophic iron- and sulfur-oxidizing bacteria and its role in delineating and sustaining acidophilic sulfide mineral-oxidizing consortia, Appl. Environ. Microb. 76 (2010) 461–467. [6] W. Dischert, R. Figge, P. Soucaille, Use of inducible promoters in the production of glycolic acid, Google Patents, 2011. [7] H. Dhamankar, Y. Tarasova, C.H. Martin, K.L.J. Prather, Engineering E-coli for the biosynthesis of 3-hydroxy-gamma-butyrolactone (3HBL) and 3,4-dihydroxybutyric acid (3,4-DHBA) as value-added chemicals from glucose as a sole carbon source, Metab. Eng. 25 (2014) 72–81. [8] K.A. Datsenko, B.L. Wanner, One-step inactivation of chromosomal genes in E. coli K-12 using PCR products, Proc. Natl. Acad. Sci. U. S. A. 97 (2000) 6640–6645. [9] G. Rao, Dynamic measurement of the volumetric oxygen transfer coefficient in fermentation systems, Biotechnol. Bioeng. 104 (2009) 841–853. [10] W. Dischert, C. Colomb, P. Soucaille, Fermentation process for producing glycolic acid, Google Patents, 2011. [11] P.P. Cherepanov, W. Wackernagel, Gene disruption in E. coli: TcR and KmR cassettes with the option of Flp-catalyzed excision of the antibiotic-resistance determinant, Gene 158 (1995) 9–14. [12] D.G. Gibson, L. Young, R.Y. Chuang, J.C. Venter, C.A. Hutchison, H.O. Smith, Enzymatic assembly of DNA molecules up to several hundred kilobases, Nat. Methods 6 (2009) 343–345. [13] C.V. Smith, C.C. Huang, A. Miczak, D.G. Russell, J.C. Sacchettini, K. Honer zu Bentrup, Biochemical and structural studies of malate synthase from Mycobacterium tuberculosis, J. Biol. Chem. 278 (2003) 1735–1743. [14] M. Chan, T.S. Sim, Malate synthase from Streptomyces clavuligerus NRRL3585: cloning, molecular characterization and its control by acetate, Microbiology 144 (Pt 11) (1998) 3229–3237. [15] M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding, Anal. Biochem. 72 (1976) 248–254. [16] Y. Deng, S.S. Fong, Metabolic engineering of Thermobifida fusca for direct aerobic bioconversion of untreated lignocellulosic biomass to 1-propanol, Metab. Eng. 13 (2011) 570–577. [17] S.P. Miller, E.J. Karschnia, T.P. Ikeda, D.C. LaPorte, Isocitrate dehydrogenase kinase/phosphatase. Kinetic characteristics of the wild-type and two mutant proteins, J. Biol. Chem. 271 (1996) 19124–19128. [18] J. Zheng, Z. Jia, Structure of the bifunctional isocitrate dehydrogenase kinase/phosphatase, Nature 465 (2010) 961–965. [19] A.J. Cozzone, M. El-Mansi, Control of isocitrate dehydrogenase catalytic activity by protein phosphorylation in E. coli, J. Mol. Microbiol. Biotechnol. 9 (2005) 132–146. [20] G. Wei, X. Yang, T. Gan, W. Zhou, J. Lin, D. Wei, High cell density fermentation of Gluconobacter oxydans DSM 2003 for glycolic acid production, J. Ind. Microbiol. Biotechnol. 36 (2009) 1029–1034. [21] W. Dischert, R. Figge, P. Soucaille, Use of inducible promoters in the production of glycolic acid, Google Patents, 2013. [22] C.H. Martin, H. Dhamankar, H.C. Tseng, M.J. Sheppard, C.R. Reisch, K.L. Prather, A platform pathway for production of 3-hydroxyacids provides a biosynthetic route to 3-hydroxy-gamma-butyrolactone, Nat. Commun. 4 (2013) 1414–1422.