Strains and media

Brain heart infusion sorbitol (BHIS) medium was used for transformation of C. glutamicum. Lysogeny broth (LB) containing tryptone 10 g L⁻¹, yeast extract 5 g L⁻¹, and NaCl 10 g L⁻¹ was used for cultivation of E. coli DH5α. The modified CGIII complex medium (pH 7.4) containing tryptone 10 g L⁻¹, yeast extract 10 g L⁻¹, and 3-morpholinopropanesulfonic acid (MOPS) 21 g L⁻¹ with 20 g L⁻¹ glucose was used for precultures of C. glutamicum [31]. CGXIIA medium (pH 7.0), containing (NH₄)₂SO₄ 20 g L⁻¹, urea 5 g L⁻¹, KH₂PO₄ 1 g L⁻¹, K₂HPO₄ 1 g L⁻¹, MgSO₄·7H₂O 0.25 g L⁻¹, CaCl₂ 10 mg L⁻¹, FeSO₄·7H₂O 10 mg L⁻¹, MnSO₄·H₂O 0.1 mg L⁻¹, ZnSO₄·7H₂O 1 mg L⁻¹, CuSO₄·5H₂O 0.2 mg L⁻¹, NiCl₂·6H₂O 20 μg L⁻¹, biotin (Sangon Biotech, Shanghai, China) 0.2 mg L⁻¹, and MOPS 21 g L⁻¹ with the indicated amounts of glucose (named CGXIIA-G) or xylose (named CGXIIA-X) was used during the aerobic stage. CGXIIB medium (pH 7.0), containing (NH₄)₂SO₄ 5 g L⁻¹, urea 5 g L⁻¹, KH₂PO₄ 1 g L⁻¹, K₂HPO₄ 1 g L⁻¹, MgSO₄·7H₂O 0.25 g L⁻¹, CaCl₂ 10 mg L⁻¹, FeSO₄·7H₂O 10 mg L⁻¹, MnSO₄·H₂O 0.1 mg L⁻¹, ZnSO₄·7 H₂O 1 mg L⁻¹, CuSO₄·5 H₂O 0.2 mg L⁻¹, NiCl₂·6 H₂O 20 μg L⁻¹, and biotin 0.2 mg L⁻¹ with the indicated amounts of glucose and/or xylose was used as medium for succinate production during the anaerobic stage. Antibiotics were added to the media at the following concentrations when required: 50 μg kanamycin mL⁻¹ and 6 μg chloramphenicol mL⁻¹ for E. coli, 25 μg kanamycin mL⁻¹ and 10 μg chloramphenicol mL⁻¹ for C. glutamicum. Antibiotics and other chemicals used in this study were purchased from Sangon Biotech (Sangon Biotech, Shanghai, China). Corn stalk hydrolysate was obtained from HEBABIZ Pharmaceutical Co. Ltd., China. Before using as substrate for succinate production, the hydrolysate was sterilized at 115 °C for 10 min.

Construction of plasmids and strains

All the strains and plasmids used in this study are listed in Table 1. All the primers used in this study are listed in Additional file 1: Table S1.

For the heterologous expression of genes encoding xylose isomerase (xylA) and xylulokinase (xylB) from Escherichia coli MG1655, Paenibacillus polymyxa SC2, Streptomyces coelicolor, and Xanthomonas campestris 8004, the plasmids pX-ecoAB, pX-ppmAB, pX-scoAB, and pX-xcbAB were constructed, respectively.

For the construction of plasmid pX-ecoAB, the xylA and xylB genes were amplified from genomic DNA of E. coli MG1655 by PCR using the primer pair ecoxylAB-1/ecoxylAB-2. The resulting fragment was digested with HindIII and XbaI, and ligated into pXMJ19 cut with the same enzymes. The plasmid pX-ppmAB was constructed analogously, using XbaI and KpnI.

For the construction of plasmid pX-xcbAB, the xylA and xylB genes were amplified from genomic DNA of X. campestris by PCR using the primer pairs xcbxylA-1/xcbxylA-2 and xcbxylB-1/xcbxylB-2, respectively. The two PCR products were spliced using the primer pair xcbxylA1/xcbxylB2, and the resulting fused PCR product was digested with HindIII and XbaI, and ligated into pXMJ19 cut with the same enzymes. The plasmid pX-scoAB was constructed analogously, using PstI and KpnI.

The replacement of the native promoter of tal by the strong sod promoter in the recombinant strain SAZ3 was achieved via a two-step homologous recombination procedure using the suicide vector pDsacB. The flanking sequences of the tal gene were amplified from genomic DNA of C. glutamicum using primer pairs PSL-1/PSL-2 and PSL-3/PSL-4. The sod promoter region (192 bp upstream of the sod gene) was amplified from chromosomal DNA of C. glutamicum using the primer pair P_sod-1/P_sod-2. The plasmid pD-PSL was assembled by circular polymerase extension cloning (CPEC) [32], using the overlap of the flanking sequences of the tal gene, P_sod and pDsacB cut with XbaI. The resulting plasmid pD-PSL was used to transform SAZ3 by electroporation. The procedure of promoter replacement was carried out as described previously [14]. Subsequently, the native promoter of tkt was analogously replaced by the strong sod promoter.

For the construction of the integrating plasmid pD-ldh, the fragment ldhF upstream of the ldh gene was first amplified from genomic DNA of C. glutamicum 13032 by PCR using the primer pair ldhF1/ldhF2. The PCR product was digested with EcoRI and SalI and ligated into pDsacB cut with the same enzymes. Similarly, the fragment ldhB downstream of the ldh gene was ligated into the plasmid containing ldhF cut with SalI/HindIII.

For the construction of plasmid pEP-P_tufaraE, the tuf promoter region (200 bp upstream of the tuf gene) was first amplified from chromosomal DNA of wild-type C. glutamicum using the primer pair P_tuf-1/P_tuf-2. The PCR product was digested with EcoRI and SalI and ligated into pEP2 cut with the same enzymes. Then, the 1.7-kb DNA fragment of the araE gene was amplified from chromosomal DNA of B. subtilis 168 by PCR using the primer pair araE-1/araE-2. The PCR product was digested with XbaI and SalI and ligated into pEP-P_tuf cut with the same enzymes.

Chromosomal integration of the xylose transporter gene araE was achieved using the plasmid pD-P_tufaraE. The DNA fragment P_tuf-araE was amplified from pEP-P_tufaraE, in which the araE gene is under the control of the constitutive tuf promoter. The resulting plasmid pD-P_tufaraE was integrated into the genome of CGS2 by electroporation.

Culture conditions

For the precultures of C. glutamicum, single colonies were grown in 5 mL of BHIS medium [33] at 30 °C and 220 rpm overnight, after which the entire resulting culture was used to inoculate 50 mL of modified CGIII medium supplemented with 10 g L⁻¹ of glucose, and grown to an OD₆₀₀ of 10.

For aerobic growth, cells were washed twice with CGXIIB medium and used to inoculate 50 ml of CGXIIA medium with 20 g L⁻¹ of xylose or glucose to an initial OD₆₀₀ = 0.8, and cultured in 500 mL shake flasks at 30 °C and 220 rpm on a rotary shaker.

A two-stage process was performed for succinate production. When the cultures reached an OD₆₀₀ of 20, the cells were harvested by centrifugation (5 000×g, 4 °C, 10 min), washed with CGXIIB medium. An appropriate amount of washed cells was suspended in 25 mL CGXIIB medium with the indicated sugars and 200 mM sodium bicarbonate, and the cell suspension was cultured in 50 mL serum bottles at 30 °C and 220 rpm on a rotary shaker. To prevent acidification, 30 g L⁻¹ of magnesium carbonate hydroxide (4MgCO₃·Mg(OH)₂·5H₂O) was added as a buffering agent. For induction, isopropyl β-d-1-thiogalactopyranoside (IPTG; Sangon Biotech, China) was added to a final concentration of 1 mM. All the fermentation experiments were performed in triplicate.

Analytical techniques

Extracellular organic acids and pentoses were measured by HPLC [34]. Glucose was monitored using an SBA sensor machine (Institute of Microbiology, Shandong, China). Growth was determined by measuring the optical density at 600 nm (OD₆₀₀). SDS-polyacrylamide gel electrophoretic (PAGE) analysis of intracellular proteins was conducted as previously described [35]. The specific growth rate was calculated as: (lnW2 − lnW1)/(t2 − t1), where W1 and W2 are biomass at times t1 and t2. The sugar consumption rate was calculated as: C_t1/t1, where C_t1 is the concentration of total consumed sugar at time t1. Codon usage analysis was based on the Graphical codon usage analyzer v. 2.0 (http://gcua.schoedl.de/index.html).

Real-time quantitative PCR (RT-qPCR)

RT-qPCR was conducted as described previously [15]. After 3 h of anaerobic fermentation, the cells were harvested for RNA isolation. The 16S rRNA gene was used as an internal reference for normalization. The primers used for RT-qPCR are listed in Additional file 1: Table S1. Samples were analyzed as previously described [31].

Selection of optimal xylose isomerases and xylulokinases for heterologous expression

Heterologous expression of the XI pathway from E. coli endowed C. glutamicum with the ability to grow on xylose [19, 23]. However, its xylose utilization capabilities were still unsatisfactory. Hence, selection of a better candidate pathway was carried out in this study. The xylA and xylB genes from E. coli MG1655, P. polymyxa SC2, S. coelicolor, and X. campestris 8004 (designations and sources are listed in Table 2) were cloned into the IPTG-inducible expression vector pXMJ19. The resulting plasmids pX-ecoAB, pX-ppmAB, pX-scoAB, and pX-xcbAB, were individually introduced into C. glutamicum ATCC 13032 along with pXMJ19 to generate the corresponding recombinant strains I-eco, I-ppm, I-sco, I-xcb, and I-pXMJ19, respectively. The growth and xylose consumption of the strains are shown in Fig. 2a.

Gene	Source	Gene symbol	Protein size (kDa)	Codon usage
				Rarely used codons	Very rarely used codons
xylA	E. coli	b3565	47.5	7	1
	P. polymyxa	PPSC2_c4572	52.2	7	2
	S. coelicolor	SCO1169	41.8	23	0
	X. campestris	XC_2477	49.2	5	0
xylB	E. coli	b3564	52.3	29	2
	P. polymyxa	PPSC2_c4571	54.8	17	2
	S. coelicolor	SCO1170	51.9	26	0
	X. campestris	XC_2478	54.8	5	0

‘Very rarely used codons’ and ‘Rarely used codons’ indicate codons represented at less than 10% or 20% in the codon usage table of C. glutamicum (http://www.kazusa.or.jp/codon/cgi-bin/showcodon.cgi?species=196627&aa=1&style=N)

Owing to the lack of xylose isomerase, the control strain I-pXMJ19 could not grow on xylose, as expected. By contrast, all the recombinant strains with the XI pathway were able to grow on xylose as sole carbon source. Among them, strain I-xcb exhibited a markedly faster maximum specific growth rate (0.224 ± 0.003 h⁻¹) and average xylose consumption rate (0.44 g L⁻¹ h⁻¹, calculated at 30 h) than the other three strains. This observation was in agreement with a previous report, in which the overexpression of xylA from X. campestris and the endogenous xylB endowed C. glutamicum with a fast growth rate (0.199 ± 0.009 h⁻¹) [23]. SDS-PAGE analysis of intracellular proteins showed that strain I-xcb had the highest protein expression of xylose isomerase and xylulokinase (Fig. 2b). Furthermore, codon usage analysis revealed that xylAB from X. campestris has the most similar codon usage to that of C. glutamicum (Table 2), which might be the main reason for the observed high protein expression level.

In view of the different profiles of aerobic and anaerobic metabolism, the xylose utilization of strains I-eco, I-ppm, I-sco, and I-xcb was also evaluated under anaerobic conditions (Fig. 2c). Similar with the results obtained under aerobic conditions, strain I-xcb exhibited a faster xylose consumption rate (1.80 g L⁻¹ h⁻¹, calculated at 12 h) than the other three strains (1.62 g L⁻¹ h⁻¹ for strain I-eco, 0.57 g L⁻¹ h⁻¹ for strain I-ppm and 0.65 g L⁻¹ h⁻¹ for strain I-sco).

It has been proved that overexpression of pyruvate carboxylase (pyc), citrate synthase (gltA), and succinate exporter (sucE) could efficiently improve succinate production in C. glutamicum [15], and therefore, plasmid pEpycgltAsucE (overexpressing pyc, gltA, and sucE genes) was introduced into SAZ3 (ΔldhAΔptaΔpqoΔcatP_sod-ppcP_sod-pyc) along with the plasmid pX-xcbAB. The resulting strain CGS1 was further tested for succinate production.

Effect of the carbon source used during the two-stage fermentation on succinate production

The carbon source influences gene expression of metabolic networks and, consequently, affects succinate production during the fermentation. It has been proven that the succinate yield from xylose is significantly higher than from glucose (25% versus 14%) in C. glutamicum under anaerobic conditions [19], suggesting that xylose might be a better substrate for succinate production. However, it has been demonstrated that the supply of ATP from xylose metabolism is insufficient during anaerobic succinate production in E. coli [7], and glucose was recommended as a co-substrate to alleviate the observed ATP shortage [36]. Therefore, the effect of carbon source used during the anaerobic phase on succinate production was evaluated in CGS1.

It is notable that the xylose consumption rate of strain CGS1 was 50.5% lower than its glucose consumption rate during the second stage (1.39 versus 2.75 g L⁻¹ h⁻¹, Table 3) when the seed cultures were prepared using glucose as substrate. Thus, to improve the xylose consumption rate under anaerobic conditions, xylose was used as substrate for the seed culture in CGXIIA-X medium, and its effect on succinate production and xylose utilization was evaluated. Perhaps, unsurprisingly, the xylose consumption rate of CGS1 in the second stage (1.95 g L⁻¹ h⁻¹) was improved by 40.3% when xylose was used for the seed culture, and a slight improvement of succinate yield was also observed (0.93 versus 0.90 g g⁻¹). There was no obvious difference in the accumulation of α-ketoglutarate and pyruvate in fermentations conducted with seed cultures from CGXIIA-G and CGXIIA-X medium. However, CGS1 collected from CGXIIA-X medium produced a 1.5-fold higher acetate concentration (1.14 g L⁻¹) than when collected from CGXIIA-G medium (0.45 g L⁻¹), which might be due to the aforementioned ATP shortage observed in aerobic seed cultures with xylose as carbon source.

To fully evaluate the effects of the carbon source, seed cultures from CGXIIA-X medium were also used for succinate production in the anaerobic stage with glucose as substrate. Compared to succinate production by seed cultures from CGXIIA-G medium, the glucose consumption rate of CGS1 collected from CGXIIA-X medium was decreased by 15.3% (2.33 versus 2.75 g L⁻¹ h⁻¹, Table 3). However, both the succinate titer (28.1 versus 24.6 g L⁻¹) and yield (0.93 versus 0.81 g g⁻¹) were significantly improved.

As these results demonstrate, an effect of carbon source was inevitable during succinate production under oxygen deprivation, and xylose was preferable to glucose both during the aerobic and the anaerobic fermentation phase. As shown in Fig. 3b, c, CGS1 was able to produce succinate with a good yield (0.93 g g⁻¹ d-xylose, 83% of the theoretical yield), while its productivity (1.70 g L⁻¹ h⁻¹ at 8 h) was still not satisfactory. Hence, further genetic manipulations were carried out to improve the xylose utilization rate and succinate productivity.

Improvement of xylose utilization and succinate production by the expression of tkt, tal, and a xylose transporter

It was reported that overexpression of tkt (encoding transketolase) and tal (encoding transaldolase) increased the flux from the non-oxidative PPP into the glycolytic pathway, and improved the rate of xylose assimilation [39, 40]. Recently, Jo et al. [28] demonstrated that overexpression of tal in C. glutamicum could efficiently improve xylose utilization and succinate production during an aerobic fermentation process. Hence, to further facilitate xylose utilization, the strong constitutive promoter sod was inserted in front of tkt and tal in strain SAZ3, yielding strain CGS2. The plasmid pX-xcbAB was introduced into CGS2 along with pEpycgltAsucE to generate CGS3. Under anaerobic conditions, the transcriptional levels of tkt and tal in CGS3 were increased 13.08- and 5.77-fold, respectively, compared with strain CGS1.

As shown in Fig. 3a, CGS3 had an obvious growth advantage over CGS1 during aerobic xylose fermentation, which was consistent with a previous report [28]. During the anaerobic fermentation, overexpression of tkt and tal also improved the xylose consumption rate by 12.3% (2.19 versus 1.95 g L⁻¹ h⁻¹ at 8 h). Consequently, the succinate productivity increased from 1.70 to 1.85 g L⁻¹ h⁻¹. In addition, the yield of succinate was also slightly improved from 0.93 to 0.98 g g⁻¹, reaching 88% of the theoretical yield.

Engineering of sugar transport was a driving force for efficient bioconversion of sugar mixtures derived from lignocellulose [41]. It is notable that the consumption rates of pentose sugars in C. glutamicum decreased appreciably at low sugar substrate concentrations [19, 42], which would encumber the whole fermentation process and cause a low production rate of succinate. To address this, the pentose transporter gene araE from Bacillus subtilis [43] was integrated into the chromosome of strain CGS2 at the ldh locus, yielding the strain CGS4. Subsequently, the plasmid pX-xcbAB was introduced into CGS4 along with pEpycgltAsucE to generate CGS5.

As expected, CGS5 showed a further improvement in growth under aerobic conditions (Fig. 3a). During the anaerobic fermentation, the yield of succinate remained at 0.98 g g⁻¹, and the xylose consumption rate was slightly improved from 2.19 to 2.31 g L⁻¹ h⁻¹. Notably, the succinate production rate was significantly improved from 1.85 to 2.28 g L⁻¹ h⁻¹. The improvement of the succinate production rate was relatively higher than that of the xylose consumption rate. This was consistent with a previous report in which the recombinant strain CRX-araE showed 4.2-fold higher lactate productivity with a 2.7-fold higher xylose consumption rate after introducing araE from C. glutamicum ATCC 31831 [24].

Succinate production from a mixture of glucose and xylose

A high sugar concentration can lower the operating costs of industrial processes by reducing the time required for feeding and the risk of contamination. To test the strain’s ability to produce succinate at high sugar concentrations, we used CGXIIB medium supplemented with 81.3 g L⁻¹ glucose and 40.3 g L⁻¹ xylose to culture CGS5 in the second stage. The ratio of glucose to xylose (2:1) was consistent with the glucose/xylose ratio of the most common lignocellulosic hydrolysate. As shown in Fig. 4a, CGS5 simultaneously and completely consumed glucose and xylose without significant carbon catabolite repression (CCR) under anaerobic conditions. 100.2 g L⁻¹ succinate was accumulated within 23 h, with a productivity of 4.35 g L⁻¹ h⁻¹. Okino et al. [13] observed that the succinate yield (about 1.45 mol/mol glucose) of C. glutamicum ΔldhA-pCRA717 was not affected by the glucose concentration (from 20 to about 280 mM). Notably, when the glucose concentration was subsequently increased to 420 mM (75.6 g L⁻¹), the succinate yield was decreased by about 15.2% to 1.23 mol/mol. Our results showed that the yield of succinate from a high-concentration sugar mixture (121.6 g L⁻¹ total sugars) was 18.0% lower than that from a low-concentration mixture (31.7 g L⁻¹ total sugars) with the same glucose/xylose ratio (0.82 versus 1.00 g/g total sugars), which was consistent with the aforementioned report [13]. However, the loss of carbon yield can be attributed to the soaring accumulation of α-ketoglutarate, since its titer increased from 1.24 to 16.2 g L⁻¹ and its yield increased from 0.04 to 0.13 g g⁻¹ total sugars, which means that the activity of the oxidative branch of the TCA cycle was also drastically increased. Because of the similar fermentation conditions, the residual oxygen in serum bottles could not answer for the variation of TCA activity. Even under strict anaerobic condition, a small but statistically significant flux through the oxidative part of the TCA cycle in wild-type C. glutamicum has been observed by bubbling CO₂ instead of argon [44], which means that the activity of oxidative TCA arm does exist and can be activated under anaerobic conditions. This does make sense in view of the overall redox balance [44], but the reason for the variation of TCA activity is still unknown. In addition, acetate increased to 2.3 g L⁻¹, but the acetate yield decreased, which suggested that the ATP demand was actually reduced at the higher sugar concentration. The concentration of fumarate reached 0.55 g L⁻¹, while the concentrations of other organic acids were below 0.1 g L⁻¹, which was negligible. However, the succinate yield of 0.82 g g⁻¹ was still acceptable and was obtained alongside a titer of 100.2 g L⁻¹.

Efficient succinate production from lignocellulosic hydrolysate

Under oxygen deprivation, C. glutamicum exhibits high tolerance to general lignocellulose-derived fermentation inhibitors [45]. However, inhibitors such as furfural, 5-hydroxymethylfurfural and vanillic had an inhibitory effect on glucose consumption and succinate accumulation under anaerobic conditions, and their mixtures collocated as the actual diluted acid hydrolysate of corn cobs showed a more toxic effect on succinate production [46]. To avoid this problem, we used corn stalk hydrolysate with much less inhibitors, which was prepared using the most widely used commercial cellulolytic enzyme reagent Cellic^® CTec2 [47]. In the second stage, CGXIIB medium with diluted hydrolysate (71.0 g L⁻¹ glucose and 30.1 g L⁻¹ xylose) was used for succinate production.

There were no significant differences in the sugar utilization pattern between the fermentations with mixed sugars (Fig. 4a) and the hydrolysate (Fig. 4b), but the consumption rate of total sugars decreased by 16.8% (4.40 versus 5.29 g L⁻¹ h⁻¹). However, to our surprise, 98.6 g L⁻¹ succinate was accumulated with a yield of 0.98 g/g total sugars, which was 16.3% higher than the yield of 0.82 g/g total sugars generated from the mixed sugars. The volumetric productivity of succinate from lignocellulosic hydrolysate was, therefore, comparable to that from sugar mixture (4.29 versus 4.35 g L⁻¹ h⁻¹). Likewise, 11.6 g L⁻¹ of α-ketoglutarate and 2.4 g L⁻¹ of acetate were accumulated as major by-products. Although the titer of α-ketoglutarate was reduced, the yield of α-ketoglutarate was not significantly decreased (0.11 versus 0.13 g/g total sugars). The concentration of fumarate was 0.15 g L⁻¹ and that of other organic acids was less than 0.1 g L⁻¹. Other sugars (not completely hydrolyzed disaccharides and other monosaccharides) generated during enzymatic hydrolysis might be one reason for the increased succinate yield, although they were not identified. The main reason could be the presence of citric acid/sodium citrate in the medium (the citric acid concentration in CGXIIB was about 22.1 g L⁻¹), which was used as the enzyme buffer during the preparation of corn stalk hydrolysate. After cultivation for 22.5 h, 12.4 g L⁻¹ of citric acid (64.5 mM) were consumed for succinate production, which simulated a higher succinate yield from total sugar. As shown in Fig. 1, the carbon flux supposed to flow to the oxidative branch of the TCA cycle could be saved by consuming citrate to produce α-ketoglutarate and NADPH. The succinate yield based on the total sugars was, therefore, increased; however, to be more accurate, the succinate yield was recalculated as 0.87 g/g total substrates, based on the consumed glucose, xylose, and citrate.