Global Cellular Metabolic Rewiring Adapts Corynebacterium glutamicum to Efficient Nonnatural Xylose Utilization

ABSTRACT

Xylose, the major component of lignocellulosic biomass, cannot be naturally or efficiently utilized by most microorganisms. Xylose (co)utilization is considered a cornerstone of efficient lignocellulose-based biomanufacturing. We evolved a rapidly xylose-utilizing strain, Cev2-18-5, which showed the highest reported specific growth rate (0.357 h⁻¹) on xylose among plasmid-free Corynebacterium glutamicum strains. A genetically clear chassis strain, CGS15, was correspondingly reconstructed with an efficient glucose-xylose coutilization performance based on comparative genomic analysis and mutation reconstruction. With the introduction of a succinate-producing plasmid, the resulting strain, CGS15-SA1, can efficiently produce 97.1 g/L of succinate with an average productivity of 8.09 g/L/h by simultaneously utilizing glucose and xylose from corn stalk hydrolysate. We further revealed a novel xylose regulatory mechanism mediated by the endogenous transcription factor IpsA with global regulatory effects on C. glutamicum. A synergistic effect on carbon metabolism and energy supply, motivated by three genomic mutations (P_sod(C131T)-xylAB, P_tuf(Δ21)-araE, and ipsA^C331T), was found to endow C. glutamicum with the efficient xylose utilization and rapid growth phenotype. Overall, this work not only provides promising C. glutamicum chassis strains for a lignocellulosic biorefinery but also enriches the understanding of the xylose regulatory mechanism.

IMPORTANCE A novel xylose regulatory mechanism mediated by the transcription factor IpsA was revealed. A synergistic effect on carbon metabolism and energy supply was found to endow C. glutamicum with the efficient xylose utilization and rapid growth phenotype. The new xylose regulatory mechanism enriches the understanding of nonnatural substrate metabolism and encourages exploration new engineering targets for rapid xylose utilization. This work also provides a paradigm to understand and engineer the metabolism of nonnatural renewable substrates for sustainable biomanufacturing.

INTRODUCTION

Growing concerns over the depletion of fossil resources and associated environmental problems have motivated the development of sustainable commodity biomanufacturing from renewable resources using microorganisms (1, 2). Lignocellulosic biomass, with glucose and xylose as the major components (3), is the most abundant (~200 billion tons annually) (4) renewable resource and an ideal substrate to suit the sustainability and economics of biomanufacturing processes (5, 6). However, most microorganisms are unfortunately not evolved to naturally and/or efficiently utilize xylose (7), especially in cases with glucose existing (8). Cell factories based on natural xylose metabolism hosts, such as Escherichia coli and Bacillus subtilis, are usually limited by strong carbon catabolite repression (CCR) when cultivated from mixed pentose and hexose sugars (9). Xylose (co)utilization is therefore considered the first cornerstone of efficient lignocellulose-based biomanufacturing.

Corynebacterium glutamicum, the famous amino acid industrial workhorse with GRAS (generally regarded as safe) status (10, 11), has been extensively engineered to establish a lignocellulose-based biomanufacturing process over the past few years (12–18). In 2006, Kawaguchi et al. (19) initially enabled C. glutamicum to utilize xylose with a specific growth rate of 0.20 h⁻¹ by complementing the xylose isomerase (XI) pathway with XI (encoded by xylA gene) from E. coli. Since then, substantial metabolic engineering efforts, including separate or combinatorial engineering of the xylose transport (XT) pathway (17, 20, 21), the XI pathway (22–24), and/or the downstream pentose phosphate (PP) pathway (14, 25), have been devoted to improving the xylose (co)utilization of C. glutamicum. Unfortunately, most of these metabolic modifications were conducted based on multicopy vectors, whose instability and incompatibility not only make the strain’s capability unstable but also compete for further engineering space for targeted product production. Instead, chromosomal integration/modification of xylose-utilizing genes with optimal promoters/ribosome binding sites (RBSs)/copies is quite stable but failed to obtain a satisfactory cell growth rate on xylose (0.20 h⁻¹) (20).

Adaptive laboratory evolution (ALE), which mimics the mechanisms of natural evolution under selective pressure, is a powerful tool for strain development without the need of in-depth genetic knowledge (26, 27). With the development of data-driven technologies, systems biology approaches such as multi-omics techniques and in silico metabolic models have enabled a more comprehensive and holistic understanding of cellular metabolism (28, 29). Consequently, integrated multidisciplinary strategies such as metabolic engineering, evolutionary engineering, and systems biology have made it feasible to construct efficient cell factories, unveil nonintuitive key target genes, and elucidate underlying genetic/regulatory mechanisms (27, 30, 31). Radek et al. (32–34) obtained the xylose-utilizing C. glutamicum strain WMB2_evo with a specific growth rate of 0.26 h⁻¹ on xylose by ALE. However, its heterologous xylose catalytic pathway was introduced based on a multicopy vector, which makes the recombinant strains possibly unstable and blocks further engineering space for targeted product synthesis. More importantly, there is rare in-depth knowledge on how C. glutamicum adapted to these heterologous pentoses (xylose or arabinose) (35, 36), and the mechanism for rapid xylose metabolism is still unclear. Due to limited understanding of cellular metabolism, the development of efficient cell factories is still arduous and no ideal chassis has been achieved for a lignocellulosic biorefinery in C. glutamicum until now.

In this study, to obtain a stable, modifiable, and rapidly xylose-coutilizing chassis, ALE and metabolic engineering were conducted. The resulting strain, Cev2-18-5, showed the highest reported growth rate on xylose among C. glutamicum plasmid-free strains. A genetically clear chassis strain, CGS15 was reconstructed and applied to efficiently produce the platform compound succinate with high titer, rate, and yield (TRY) by simultaneously utilizing lignocellulosic sugar mixtures. To investigate the adaptation of the cells to engineered xylose metabolism, multiplex systems biology techniques including transcriptomics and metabolomics were conducted. The effects of three positive mutations involved in the heterologous xylose catalytic/transporter pathway and endogenous regulatory network were identified and elucidated. Notably, we revealed a novel xylose regulatory mechanism with a global effect on the regulatory and metabolic network of C. glutamicum based on the endogenous transcription factor IpsA. The new regulatory mechanism enriches the understanding of nonnatural substrate metabolism and encourages exploration of new engineering targets for rapid xylose utilization.

RESULTS AND DISCUSSION

Evolution of C. glutamicum CGS6 toward a rapid xylose (co)utilization phenotype.

A previously selected xylose utilization operon, xylAB, from Xanthomonas campestris was inserted into the chromosomal ΔackA-pta locus of CGS4 (14) (Table 1) under the control of the endogenous constitutive promoter P_sod to enable xylose utilization in strain CGS6 (see Fig. S1a in the supplemental material). Then, the CGS6 strain was subjected to two rounds of ALE on xylose to improve its general growth characteristics (Fig. 1a). The culture was periodically diluted with fresh medium to an initial optical density at 600 nm (OD₆₀₀) of 1 every 12 or 24 h (see Text S1, note 1, in the supplemental material). The evolved strains Cev1-22-10 and Cev2-18-5 were isolated from the 22nd culture of the first round and the 18th culture of the second round, respectively. As shown in Fig. 1b and Table 2, the specific growth rates of strains Cev1-22-10 (0.324 h⁻¹) and Cev2-18-5 (0.357 h⁻¹) were 1.40- and 1.64-fold higher than that of CGS6 (0.135 h⁻¹). At the same time, the evolved strains showed a significantly improved specific xylose consumption rate or maximal biomass (Table 2). Consequently, the corresponding average xylose consumption rates of strains Cev1-22-10 and Cev2-18-5 were 4.67- and 6.62-fold higher than that of CGS6. It is worth noting that the rapid cell growth of strain Cev2-18-5 was not achieved at the cost of excess substrate consumption. Instead, its biomass yield was increased from 0.302 to 0.351 g of dry cells weight (DCW)/g xylose compared with that of starting strain CGS6 (Table 2). Moreover, the catabolic repression of xylose utilization by glucose was also released in evolved strain Cev2-18-5 (Fig. 1c). Thus, C. glutamicum strain Cev2-18-5, with the highest specific growth rate on xylose among C. glutamicum plasmid-free strains (Table S2), was successfully evolved for rapid xylose (co)utilization.

TABLE 1.

Strains used in this study

Strain(s)	Relevant characteristics	Reference
CGS4	ATCC 13032 Δldh Δpta-ackA Δpqo Δcat Psod-pyc Psod-ppc Psod-tal Psod-tkt ldhA::Ptuf-araE
CGS6	CGS4 pta-ackA::Psod-xylAB	This study
Cev1-1 to Cev1-22	Xylose-adapted strains evolved from strain CGS6	This study
Cev1-22-7, -10, and -16	Three clones of the evolved strain Cev1-22	This study
Cev2-1 to Cev2-18	Xylose-adapted strains evolved from strain Cev1-22-10	This study
Cev2-18-5	A single clone of evolved strain Cev2-18	This study
CGS7	CGS6 Psod(C131T)-xylAB	This study
CGS8	CGS6 Ptuf(Δ21)-araE	This study
CGS9	CGS6 NCgl2418535T536	This study
CGS10	CGS6 ipsAC331T	This study
CGS11	CGS7 Ptuf(Δ21)-araE	This study
CGS12	CGS7 NCgl2418535T536	This study
CGS13	CGS7 ipsAC331T	This study
CGS14	CGS11 NCgl2418535T536	This study
CGS15	CGS11 ipsAC331T	This study
CGS16	CGS15 NCgl2418535T536	This study
CGS17	CGS10 Ptuf(Δ21)-araE	This study
Cev2-18-5R	Cev2-18-5 Ptuf-araE	This study
CGS6-SA1	CGS6 containing pEC-pycsucE	This study
CGS15-SA1	CGS15 containing pEC-pycsucE	This study

FIG 1.

Evolution and reconstruction of C. glutamicum for rapid xylose metabolism. (a) Growth reevaluation (OD₆₀₀ at 24 h) of all the evolved strains on xylose without isolation. (b) Cell growth (filled circles) and xylose consumption (open squares) of CGS6 (red), Cev1-22-10 (blue), and Cev2-18-5 (orange). (c) Cell growth (filled triangles) and sugar consumption (open circles) of strains CGS6 and Cev2-18-5 in CGXII minimal medium supplemented with 10 g/L xylose (solid line) and 10 g/L glucose (dashed line). (d) Cell growth (filled circles) and xylose consumption (open squares) of CGS7 (red), CGS11 (blue), and CGS15 (orange). (e) Specific growth rate, maximal biomass, and underlying mutations of the reconstructed strains.

TABLE 2.

Cell growth and xylose consumption of evolved and reconstructed strains^a

Strain	Specific growth rate (h−1)	Maximal biomass (g DCW/L)	Biomass yield (0 to 20 h, g DCW/g xylose)	Specific xylose consumption rate [0 to 20 h, g/g DCW/h]	Avg xylose consumption rate (0 to 20 h, g/L/h)
CGS6	0.135 ± 0.002	7.150 ± 0.154	0.302 ± 0.011	0.266 ± 0.009	0.127 ± 0.007
Cev1-22-10	0.324 ± 0.008	7.229 ± 0.231	0.309 ± 0.016	0.312 ± 0.015	0.720 ± 0.013
Cev2-18-5	0.357 ± 0.001	8.304 ± 0.306	0.351 ± 0.006	0.277 ± 0.004	0.968 ± 0.006
CGS7	0.262 ± 0.001	8.421 ± 0.117	0.310 ± 0.039	0.312 ± 0.039	0.509 ± 0.065
CGS11	0.311 ± 0.003	4.738 ± 0.097	0.201 ± 0.014	0.473 ± 0.031	0.621 ± 0.007
CGS15	0.341 ± 0.002	8.567 ± 0.176	0.310 ± 0.022	0.316 ± 0.022	0.901 ± 0.066

^aValues are the averages and standard deviations of results of three independent cultures.

Genome-scale mutation reconstruction bridged the excellent xylose (co)utilization phenotype with the underlying genotype.

In order to investigate the genetic basis for the rapid nonnatural xylose utilization metabolism, we sequenced the genomes of starting strain CGS6 and evolved strains Cev1-22-10 and Cev2-18-5. Three mutations (P_sod(C131T)-xylAB, P_tuf(Δ21)-araE, and NCgl2418^535T536) were found in evolved strain Cev1-22-10, and one more mutation (ipsA^C331T) was identified in evolved strain Cev2-18-5 (Table 3; see Fig. S2 and S3 and Text S1, note 2, in the supplemental material). The P_sod(C131T)-xylAB mutation and P_tuf(Δ21)-araE were located in front of the heterologous xylAB operon and xylose transporter gene araE. The NCgl2418^535T536 mutation was a frameshift mutation (535T536) in the endogenous putative transposase gene NCgl2418. The ipsA^C331T mutation was a single base replacement (C331T) in gene ipsA encoding an endogenous LacI type regulator.

TABLE 3.

Primary mutations identified by whole-genome resequencing of the evolved strains

Mutation	Mutation location	Mutation typea	Source
Psod(C131T)-xylAB	In front of the heterologous xylose utilization operon	Nucleotide replacement, C131T in sod promoter	Cev1-22-10 and Cev2-18-5
Ptuf(Δ21)-araE	In front of the heterologous xylose transporter	Nucleotide deletion, 21 bp around the 5′-UTR	Cev1-22-10 and Cev2-18-5
NCgl2418535T536	Endogenous, transposase	Nucleotide frameshift insertion, 535T536	Cev1-22-10 and Cev2-18-5
ipsAC331T (NCgl2538C331T)	Endogenous, transcription factor	Nucleotide replacement, C331T; amino acid replacement, P111S	Cev2-18-5

^aDetailed descriptions of the mutations are provided in Fig. S2 in the supplemental material. UTR, untranslated region.

To identify the effects of each mutation on the xylose utilization phenotype, the mutations were reconstructed in strain CGS6. First, the three mutations (P_sod(C131T)-xylAB, P_tuf(Δ21)-araE, and NCgl2418^535T536) that emerged during the first round of ALE were individually introduced into CGS6 to generate strains CGS7, CGS8, and CGS9, respectively. As shown in Fig. 1e, strain CGS7 exhibited a significant 94% increase in the specific growth rate (0.262 versus 0.135 h⁻¹), which indicated that P_sod(C131T)-xylAB is a beneficial mutation. In contrast, the individual introduction of P_tuf(Δ21)-araE or NCgl2418^535T536 caused an obvious decrease in cell growth (0.104 or 0.113 versus 0.135 h⁻¹).

Then, P_tuf(Δ21)-araE and NCgl2418^535T536 were introduced into strain CGS7 to test their combined effects on cell growth together with the P_sod(C131T)-xylAB mutation, generating strains CGS11 and CGS12, respectively. Interestingly, the P_tuf(Δ21)-araE mutation, which appeared to be an unfavorable mutation in CGS8, had a synergistic effect with the P_sod(C131T)-xylAB mutation on improving the specific growth rate in strain CGS11 (0.311 versus 0.262 h⁻¹). Moreover, after recovery of this mutation to the wild-type form (P_tuf-araE) in strain Cev2-18-5, the growth rate of resulting strain Cev2-18-5R decreased dramatically (Fig. 1e), which confirmed that P_tuf(Δ21)-araE is also a beneficial mutation for accelerated xylose metabolism. However, the NCgl2418^535T536 mutation also inhibited cell growth in strain CGS12, and when it was introduced into strain CGS11, resulting strain CGS14 showed no difference in specific growth rate (Fig. 1e), suggesting that this mutation is not relevant to the xylose utilization phenotype.

Subsequently, the ipsA^C331T mutation that emerged in the second round of ALE was further introduced into strain CGS11. Resulting strain CGS15 showed a significant improvement in cell growth (0.341 versus 0.311 h⁻¹). Moreover, the introduction of ipsA^C331T together with P_sod(C131T)-xylAB or P_tuf(Δ21)-araE improved either the specific growth rate (CGS13 versus CGS7, 0.271 versus 0.262 h⁻¹) or maximal biomass (CGS17 versus CGS8, 5.121 versus 2.542 g DCW/L) (Fig. 1e), which confirmed that ipsA^C331T is a beneficial mutation. In contrast, when only the ipsA^C331T mutation was introduced into strain CGS6, resulting strain CGS10 showed a decrease in the specific growth rate (0.109 versus 0.135 h⁻¹), indicating that the ipsA^C331T mutation had a synergistic effect on xylose metabolism when combined with P_sod(C131T)-xylAB and/or P_tuf(Δ21)-araE. In addition, the NCgl2418^535T536 mutation was also introduced into CGS15, but resulting strain CGS16 exhibited almost the same cell growth as that of strain CGS15 (Fig. 1e), which confirmed that the NCgl2418^535T536 mutation is nonessential for rapid xylose metabolism. Finally, the rapidly xylose-utilizing C. glutamicum chassis strain CGS15 with a clear genetic background was reconstructed, which almost recovered the excellent rapid growth (0.341 versus 0.357 h⁻¹) on pure xylose of evolved strain Cev2-18-5 (Fig. 1d; Table 2). The analysis of phenotypic stability showed that the growth characteristics of CGS15 were unchanged after more than 10 transfers in minimal medium (Fig. S1d and e), confirming that it is a stable chassis.

Thus, all the mutations were classified, and three beneficial mutations were identified. Specifically, the reconstructed strains harboring the P_sod(C131T)-xylAB mutation showed obvious growth advantages (CGS7 versus CGS6, CGS11 versus CGS8, and CGS14 versus CGS9) (Fig. 1e). However, the P_tuf(Δ21)-araE mutation must be combined with P_sod(C131T)-xylAB to be beneficial for cell growth (CGS11 versus CGS8) (Fig. 1e). Moreover, the second-generation mutation ipsA^C331T showed positive effects on the growth phenotype only when combined with the first-generation beneficial mutations (CGS13 versus CGS7, CGS17 versus CGS8, and CGS15 versus CGS11) (Fig. 1e). This sequential synergistic contribution of beneficial mutations to the excellent growth phenotype of CGS15 indicated that the adaptation of the cells to xylose metabolism was systemic and progressive.

The excellent glucose-xylose coutilization chassis enabled efficient succinate production from lignocellulosic sugars.

C. glutamicum is a facultative anaerobe that stops growing but remains metabolically active under anaerobic conditions (37). Under such constraints, energy and carbon are channeled toward product formation rather than biomass (14, 37, 38). Thus, anaerobic xylose (co)utilization by chassis strain CGS15 was also evaluated. As expected, the average xylose consumption rate (0 to 12 h) was enhanced by 13% in strain CGS15 compared with that in CGS6 (Fig. 2a). In addition, xylose coutilization (0 to 4 h) was also enhanced (Fig. 2b), with increases in both the xylose-to-glucose ratio (0.33 versus 0.17) and total sugar consumption rate (4.42 versus 4.01 g/L/h). Therefore, the beneficial mutations that emerged during the aerobic ALE process also promoted the anaerobic xylose metabolism.

FIG 2.

Evaluation of xylose (co)utilization and succinate production by the reconstructed strains under anaerobic conditions. (a) Xylose consumption of CGS6 (red) and CGS15 (blue) in modified mineral salts medium containing 30 g/L xylose. (b) Sugar consumption of CGS6 (red) and CGS15 (blue) in modified mineral salts medium containing 15 g/L xylose (solid line) and 15 g/L glucose (dashed line). (c and d) Succinate production (blue) and sugar consumption (red) of CGS6-SA1 (c) and CGS15-SA1 (d) in modified mineral salts medium containing 60 g/L xylose and 60 g/L glucose. (e) Succinate production (blue) and sugar consumption (red) of CGS15-SA1 in lignocellulose hydrolysate containing 38.1 g/L xylose and 79.4 g/L glucose.

To evaluate the potential of the reconstructed chassis for application in industrial biomanufacturing processes, CGS6 and CGS15 were further engineered to efficiently produce succinate, one of the most important biobased platform chemicals (1). Plasmid pEC-pycsucE, overexpressing the Pyc^T132A pyruvate carboxylase mutant (39) and the SucE succinate exporter (40), was introduced into CGS6 and CGS15 to generate the succinate producer strains CGS6-SA1 and CGS15-SA1, respectively. A two-phase fermentation was carried out for succinate production in 50-mL serum bottles, and the residual sugar and product concentrations were determined. As shown in Fig. 2c and d, although both CGS6-SA1 and CGS15-SA1 could simultaneously consume glucose and xylose, CGS15-SA1 showed an obviously increased coutilization capacity (0 to 8 h) in terms of both the ratio of xylose to glucose (0.78 versus 0.47) and the total sugar consumption rate (13.46 versus 10.11 g/L/h). In addition, the overall succinate yield of CGS15-SA1 was 22% higher than that of CGS6-SA1 (0.82 versus 0.67 g/g total sugar) (Fig. 2c and d), suggesting that enhanced glucose-xylose mixture utilization can also promote succinate production. A titer of 100.52 g/L succinate was obtained from a pure glucose-xylose mixture at 12 h using strain CGS15-SA1, corresponding to an average productivity of 8.40 g/L/h. Finally, the succinate production of strain CGS15-SA1 was evaluated using corn stalk hydrolysates containing high concentrations of glucose and xylose as substrates. A succinate titer of 97.1 g/L was produced, with a yield of 0.83 g/g total sugars and an average productivity of 8.09 g/L/h (Fig. 2e). Moreover, after three batches of repeated fermentation, CGS15-SA1 still maintain 74% of the average productivity (Text S1, note 3, and Table S3), highlighting the long-term stability of CGS15-SA1 as a biocatalyst.

CGS15 unexpectedly exhibited excellent coutilization of xylose and glucose at a ratio of about 0.5:1 (Table S4), approaching the most common xylose/glucose ratio in lignocellulosic biomass (14). After introduction of just one succinate production vector into the chassis, resulting strain CGS15-SA1 achieved the high average succinate productivity of 8.09 g/L/h with a comparable succinate titer and yield (Table S5), which highlights CGS15 as a promising chassis for the lignocellulosic biorefinery. To elucidate the adaptation of CGS15 to nonnatural xylose and its underlying metabolic mechanism, three newly identified positive mutations (P_sod(C131T)-xylAB, P_tuf(Δ21)-araE, and ipsA^C331T) were further characterized.

Coordinated expression of the xylose isomerase pathway and xylose transporter released carbon catabolite repression for xylose in C. glutamicum.

To remove the obstacles for rapid xylose metabolism, inefficient xylose transport and xylose catabolism (35, 41), starting strain CGS6 first evolved two mutations (P_sod(C131T)-xylAB and P_tuf(Δ21)-araE) in the regulatory sequences (promoter or 5′-untranslated region [UTR]) of the corresponding pathway genes. Xylose isomerase activity analysis and green fluorescent protein (GFP) expression analysis revealed that both mutations can significantly increase the expression of genes under their controls (Text S1, note 4, and Fig. S4). It is worth mentioning that the P_sod(C131T) mutant is an ultrahigh efficiency promoter with at least ~10-times-higher GFP intensity (Fig. S4b and c) than that obtained using the native P_sod promoter. This is likely caused by a newly emerged −35 box resulting from the C131T mutation (Fig. S4d).

However, the sole expression of the membrane protein AraE appeared to cause an overly strong metabolic burden on pure xylose (CGS8 versus CGS6) (Fig. 1e). The cell growth rate and average xylose consumption rate could be enhanced only when accompanied by xylAB expression (CGS11 versus CGS6), suggesting that xylose transport and catabolism had a synergistic effect on xylose metabolism. Interestingly, when cultured on a glucose-xylose mixture, strains CGS8 and CGS11 with the evolved P_tuf(Δ21)-araE mutation achieved simultaneous utilization of glucose and xylose (Fig. 3c and d). In contrast, parental strains CGS6 and CGS7 showed obvious diauxic growth curves, and xylose utilization occurred only when glucose was almost exhausted (Fig. 3a and b). Seen from the comparison of sugar mixture utilization (Fig. 3; Table S4), the optimization of expression of xylose transporter and XI pathway genes was critical for the release of CCR in C. glutamicum.

FIG 3.

Evaluation of the CCR effect in reconstructed strains using glucose-xylose mixtures. Growth (filled symbols, solid line), xylose consumption (open symbols, solid line), and glucose consumption (open symbols, dashed line) of CGS6 (a), CGS7 (b), CGS8 (c), CGS11 (d), CGS15 (e), and CGS16 (f) in CGXII mineral medium containing 10 g/L glucose and 10 g/L xylose. Error bars indicate the standard deviations of results from three independent replicates.

It was worth noting that the P_tuf(Δ21)-araE mutation caused obvious decreases in biomass synthesis in strains CGS8 and CGS11 on xylose and a xylose-glucose mixture (Table 2; Table S4). However, when they were cultured on glucose, there were no obvious growth differences among these reconstructed strains (Fig. S5). Moreover, CGS8 and CGS11 showed a significantly elongated cell morphology compared with that of CGS6 (Fig. 4), suggesting that the seemingly contradictory phenotype produced by the P_tuf(Δ21)-araE mutation is more complex than a simple metabolic burden.

FIG 4.

Scanning electron microscopy images of CGS6, CGS7, CGS8, CGS11, and CGS15.

Emerging a novel regulatory mechanism mediated by the transcription factor IpsA using xylose as the effector molecule.

The endogenous transcription factor IpsA was reported to act as a LacI-type transcriptional regulator involved in triggering inositol-derived cell wall biogenesis during growth on glucose. As reported, ΔipsA cells showed chain-like structures and a reduced growth rate on glucose (42). Interestingly, a similar elongated cell morphology was observed in CGS8 and CGS11, and the IpsA^P111S variant almost completely recovered the normal cell shape in CGS15 (Text S1, note 5; Fig. 4). The IpsA^P111S variant also recovered the biomass accumulation on either pure xylose or a xylose-glucose mixture (Fig. 1e and 3e and f). Therefore, we hypothesized that there may be an unknown xylose regulatory mechanism mediated by IpsA, whose effector molecule might be xylose and/or its derivatives.

The ino1 gene, which was reported to be directly regulated by IpsA (42), the LacI type of transcriptional regulator, was used as a targeted DNA sequence for the formation of complexes with purified IpsA/IpsA^P111S protein (Text S1, note 6). An electrophoretic mobility shift assay (EMSA) showed that IpsA^P111S exhibited a higher affinity for the ino1 promoter than IpsA. Xylose was also found to act as an effector molecule of IpsA/IpsA^P111S (Fig. S6) that could dissociate the IpsA/IpsA^P111S-DNA complexes in vitro. Further surface plasmon resonance (SPR) analysis showed that the DNA probe binds to IpsA and IpsA^P111S proteins with equilibrium dissociation constant (K_D) values of 802 nM and 58 nM, respectively (Text S1, note 7; Fig. 5a and b), indicating a 12.8-fold-higher DNA binding affinity of the IpsA^P111S variant. Moreover, the IpsA^P111S-DNA complex showed a significantly improved resistance to the effector molecule compared to that of the IpsA-DNA complex (Fig. 5c and d). The molecular dynamics (MD) simulations showed that IpsA^P111S had a larger opening amplitude of the two structural domains than IpsA (22.3 versus 10.2 degrees) (Text S1, note 8; Fig. 5g), which likely led to a reduction of protein-DNA deformation after effector binding, thereby stabilizing the protein-DNA complex (Fig. 5c and d).

FIG 5.

Molecular mechanism of the IpsA-mediated xylose regulatory effect. (a and b) BIAcore diagrams of ino1 promoter fragment bound to a series of concentrations of IpsA (a) and IpsA^P111S (b) without xylose. (c and d) BIAcore diagram of 2.5 μM IpsA (c) and IpsA^P111S (d) protein bound to the same sensor chip surface with an immobilized ino1 promoter fragment in the presence of increasing xylose concentrations. (e) An electrophoretic mobility shift assay (EMSA) was performed to verify the binding of IpsA and IpsA^P111S to the promoter region of ino1 with xylose. (f) Schematic diagram of the hypothetical IpsA-mediated xylose regulatory mechanism. Green arrows indicate the regulatory effect of IpsA/IpsA^P111S on the targeted DNA. Red arrows indicate the dissociating effect of xylose on IpsA/IpsA^P111S-DNA complexes. Dashed lines indicate a weakened interaction between protein and DNA or protein and effector molecule. (g) Structural conformation of IpsA and IpsA^P111S after 1,000 ns of MD simulation.

Thus, a novel xylose regulatory mechanism mediated by transcription factor IpsA was identified in C. glutamicum (Fig. 5f). Specifically, when xylose was bound to IpsA, the binding affinity of IpsA to its cognate DNA sequence was obviously reduced (Fig. 5c and d; Fig. S6). Probably due to the similar attenuation effect on the interaction of IpsA and its regulated genes, CGS8 and CGS11 showed cell morphologies similar to that of the ΔipsA strain. This rewiring finally resulted in a seemingly contradictory phenotype consisting of a higher specific sugar consumption rate and a lower biomass yield on xylose and a glucose-xylose mixture in CGS11 (Table 2; Table S4). Notably, the endogenous transcription factor IpsA variant (IpsA^P111S) showed higher DNA binding affinity to IpsA-regulated DNA and more resistance to effector molecules (Fig. 5a to d). This evolved molecular mechanism probably rewired the cellular regulatory network in CGS15 and consequently recovered the biomass synthesis affected by the IpsA-mediated xylose regulatory effect.

Known xylose regulatory mechanisms, such as the mechanism of XylR in E. coli (43–45), usually manifests as specific activation of the xylose utilization operon (45–47). However, the newly identified xylose regulatory mechanism discovered in this study showed a global regulatory impact. This xylose-responsive repression effect on the ino1 promoter by wild-type IpsA may facilitate future development of novel xylose-repressed biosensors for synthetic biology (48). More importantly, xylose, once thought to be innocuous to cells, unexpectedly disturbed the cellular metabolic network to a certain degree. This discovery indicates that there may also exist a “regulatory impact” of a nonnatural substrate, beyond its noticeable chemical toxicity and specific catalytic pathways (49, 50), as a nonintuitive target for further understanding/engineering of nonnatural substrate metabolism. However, this growth-impaired xylose regulatory mechanism can be recovered simply by one amino acid substitution in IpsA due to the significantly improved DNA-binding affinity as well as resistance to xylose, offering a novel target for engineering xylose metabolism in C. glutamicum.

Rebalancing cellular carbon metabolism and energy supply through global regulatory rewiring.

Nonnatural substrates as well as their requisite heterologous utilization pathways and/or import proteins are rarely compatible with the natural endogenous metabolic network (46, 50, 51). Therefore, introduction of a heterologous xylose utilization pathway and transporter merely enabled a poor xylose-(co)utilizing capacity in strain CGS6 (Fig. 1c and 3a). In order to investigate cellular regulatory and metabolic adaptations responding to the seemingly contradictory phenotype caused by the P_tuf(Δ21)-araE mutation and elucidate how the IpsA^P111S variant significantly improved cellular biomass synthesis, the global differences in transcriptional levels and intracellular metabolites of strains CGS7, CGS11, and CGS15 were analyzed (Text S1, note 9, and Fig. S7 and S8). A total of 76 metabolites involved in core carbon and energy metabolism were detected, and their relative abundances were compared. An orthogonal partial least-squares discriminant analysis (O₂PLS-DA) was carried out for the classification of high-dimensional omics data as shown in score plots (Fig. S7a and S8a). The metabolomic/transcriptomic samples from each group are well clustered, indicating good repeatability and suitability for further analysis.

The number of significantly differentially expressed genes (fold change ≥ 2; adjusted P value [P_adj] ≤ 0.05) for each of the comparisons was shown in Fig. S7b. Specifically, 313 genes were upregulated and 394 were downregulated in CGS11 compared with CGS7 (Fig. S7c). Similarly, 253 genes were upregulated and 274 were downregulated in CGS15 compared with CGS11 (Fig. S7d). Cluster analysis of differentially expressed genes showed a generally reversed trend of transcription regulation in CGS11 versus CGS7 and in CGS15 versus CGS11 (Fig. S7e). The metabolite loading scatterplots showed that the model terms of CGS7, CGS11, and CGS15 were significantly different from each other (Fig. S8b), which was consistent with the differently regulated global metabolic networks of CGS7, CGS11, and CGS15. Detailed relative changes of gene transcription levels and intracellular metabolite concentrations are described as follows.

(i) Fine adjustment of central carbon metabolism enhanced xylose metabolism. The cooverexpression of the xylose transporter and xylose isomerase pathway efficiently channeled xylose into the central carbon metabolism (Fig. 6a). Metabolomic analysis showed that the abundances of node metabolites of central carbon metabolism, such as those of α-d-fructose 6-phosphate (F6P)/α-d-glucose 6-phosphate (G6P) (102% higher) and phosphoenolpyruvate (PEP) (102% higher), were significantly increased in CGS11 (Fig. 6c and d). The general transcriptional levels of genes in the central metabolic network, such those of as pgi (73% lower) and sucD (78% lower), were decreased (Fig. 6a and b). Notably, this change failed to enhance the supply of other metabolites synthesized from these node metabolites, such as malonyl coenzyme A (malonyl-CoA) (93% lower) (Fig. 6c) for fatty acid precursor synthesis and amino acids for protein synthesis, such as l-serine (55% lower), l-valine (52% lower), and l-threonine (67% lower) (Fig. 7c and d). In addition to the shortage of relevant precursors, another probable reason could be the downregulated transcriptional levels of genes related to the synthesis of malonyl-CoA and major amino acids such as dtsR1 (82% lower), serA (67% lower), leuD (74% lower), hom (52% lower), and thrC (81% lower) (Fig. 6b and 7a and b). Although these intracellular metabolites were decreased, they might still be sufficient for rapid growth in CGS11 in exponential growth phase. Instead, the transcriptional levels of many genes (such as acn, sdhAB, and fumC) in the tricarboxylic acid (TCA) cycle were upregulated in CGS11 compared with those in CGS7 (Fig. 5). The abundances of intermediate metabolites of the TCA cycle, such as a-ketoglutarate (AKG) (45% higher) and l-malate (MAL) (54% higher), were increased in CGS11 (Fig. 6c and d). Extracellular by-product analysis showed that overflow metabolism occurred in CGS11 with obvious accumulation of AKG (Fig. S9). The increased intracellular and extracellular metabolites and genes in the TCA cycle suggest that more carbon fluxes might be led into the TCA cycle. However, the intracellular relative concentrations of α-ketoglutarate-derived amino acids such as l-glutamate and l-proline were decreased (Fig. 7c). This suggests that the increased carbon fluxes of the TCA cycle might lead to oxidation catabolism to obtain enough energy to support a high growth rate, which means greater carbon loss and consequently results in a lower biomass yield.

FIG 6.

Changes of the central carbon metabolism in response to different mutations. (a) Diagram of the central carbon metabolism of C. glutamicum with modules partitioned into the xylose isomerase pathway (XI; pink background), pentose phosphate pathway (PPP; yellow background), Embden-Meyerhof-Parnas pathway (EMP; blue background), and tricarboxylic acid cycle (TCA; orange background). (b and c) Heat maps (log₂ fold change) showing the changes of gene transcription levels (b) and relative intracellular metabolite concentrations (c) in CGS7, CGS11, and CGS15. #, metabolites that could not be distinguished by LC-MS, such as F6P and G6P. (d) Comparison (log₂ fold change) of representative metabolites between CGS11 and CGS7 or CGS15 and CGS11. Single and double asterisks indicate P values of <0.05 and <0.01, respectively. Error bars indicate the standard deviations of results from three independent replicates.

FIG 7.

Changes of the amino acid metabolism in response to different mutations. (a to c) Heat maps (log₂ fold value) showing the changes of the transcription levels of genes related to amino acid metabolism (a and b) and relative intracellular amino acid concentrations (c) in CGS7, CGS11, and CGS15. (d) Comparison (log₂ fold change) of representative amino acids. Single and double asterisks indicate P values of <0.05 and <0.01, respectively. Error bars indicate the standard deviations of results from three independent replicates.

To reasonably regulate the central metabolic network to redistribute the imported xylose, the first-generation strain Cev1-22-10 subsequently evolved the transcription factor mutation ipsA^C331T. The transcriptional levels of genes encoding transporters (ptsHGSIF and araE) in CGS15 were increased compared with that in CGS11 (Fig. S10), suggesting that ipsA^C331T could upregulate the expression of transporters and further enhance the uptake of the carbon source. The relative concentrations of major node metabolites of the central metabolism were obviously decreased in CGS15 compared with those in CGS11 (Fig. 6c and d). These included d-erythrose 4-phosphate (E4P) (40% lower) and S7P sedoheptulose 7-phosphate (S7P) (31% lower) in the pentose phosphate pathway (PPP), as well as F6P/G6P (67% lower) and PEP (43% lower) in the Embden-Meyerhof-Parnas (EMP) pathway, but also citrate/isocitrate (32% lower) and succinate (26% lower) in the TCA cycle. These changes may explain the decreased specific xylose consumption rate in CGS15 (0.316 versus 0.473 g/g DCW/h, Table 2). At the same time, malonyl-CoA and major amino acid synthesis genes, such as dtsR1 (421% higher), serA (190% higher), leuD (142% higher), and hom (124% higher), were obviously upregulated (Fig. 6b and 7a and b). Consistently, the synthesis of related metabolites, including malonyl-CoA (110% higher), l-serine (31% higher), l-valine (134% higher), l-aspartate (67% higher), l-threonine (108% higher), and l-glutamate (26% higher), was significantly reinforced (Fig. 7c and d). This indicated that the metabolites that were previously stacked in the central carbon metabolism of CGS11 had been successfully rerouted into the downstream metabolic network of CGS15. Thus, the disordered carbon metabolism caused by the surging substrate import was reorganized to support the higher biomass yield (0.310 versus 0.201 g DCW/g xylose) and maximal biomass (8.567 versus 4.738 g DCW/L) (Table 2).

(ii) Reinforced bioenergetic efficiency rebalanced the energy supply with rapid cell growth. The general transcriptional levels of genes involved in oxidative phosphorylation (Fig. 8a), as well as its directly linked TCA cycle and NADH generation reactions (Fig. 6b), were enhanced in CGS11 compared with CGS7. However, the relative NADH/NAD⁺ ratio was deceased from 0.26 to 0.11 (Fig. 8b), and the relative intracellular ATP content and ATP/ADP ratio at 12 h were reduced by 42% and 64%, respectively (Fig. 8c). This can be explained by a decrease in the efficiency of the electron transport chain (ETC), which is mainly dictated by the energy coupling efficiency of terminal oxidases (Fig. 8f). As shown in Fig. 8a, the transcriptional levels of the ctaCDEF and qcrABC operons (encoding the cytochrome bc₁c and cytochrome aa₃ complex with a holistic H⁺/O ratio of 6) (52) were downregulated ~2-fold in CGS11 compared to CGS7 at 12 h, while the cydAB operon (encoding cytochrome bd oxidase with a H⁺/O ratio of 2) (52) was upregulated by ~50%. Moreover, substrate-level phosphorylation (pfkA, pyk, pyk2, and sucCD) was also downregulated in CGS11. A decrease in the energy supply was reported to increase the demand for fluxes related to catabolism rather than biomass formation (53). Consistently, the intracellular amino acid levels were generally reduced, while metabolites of the EMP pathway or TCA cycle were obviously accumulated in CGS11 (Fig. 7c and 6c). Although this cellular response enabled a high specific xylose consumption rate to support the rapid cell growth of CGS11, the biomass yield was drastically decreased by 35% compared with that of CGS7 (Table 2).

FIG 8.

Changes of the energy metabolism in response to different mutations. (a) Heat maps (log₂ fold change) showing the changes in the transcription levels of genes related to energy metabolism in CGS7, CGS11, and CGS15 at 12 h. (b and d) Relative NADH contents and NADH/NAD⁺ ratios at 12 h (b) and 20 h (d). (c and e) Relative ATP contents and ATP/ADP ratios at 12 h (c) and 20 h (e). (f) Diagram of oxidative phosphorylation in C. glutamicum. (g) Transcription-level changes of genes related to the electron transport chain at 20 h according to RT-PCR. Error bars indicate the standard deviations of results from three independent replicates.

To overcome the low bioenergetic efficiency, the cells rewired their energy metabolism by evolving the transcription factor mutation ipsA^C331T. The transcription levels of ctaCDEF and qcrABC both increased by about 50% in CGS15 compared with CGS11. Concomitantly, the cydAB operon was downregulated by ~50%. Furthermore, substrate-level phosphorylation was also enhanced in CGS15, with 1.03- and 4.44-fold transcriptional upregulation of sucC and sucD, respectively, at 12 h compared with that of CGS11 (Fig. 8a). The ATP/ADP ratio was consequently increased by 16%, from 0.32 to 0.37 (Fig. 8c). This reinforced supply of energy was more clearly observed in the late log phase (20 h), with significantly upregulated respiratory chain genes in CGS15 compared to those in CGS11 (Fig. 8g). As a result, the relative ATP content and ATP/ADP ratio of CGS15 at 20 h were increased by 46% and 70%, respectively, compared to those of CGS11 (Fig. 8e). The enhanced bioenergetic efficiency meant that a part of the carbon fluxes that were once flowing into catabolism to maintain the cellular energy demand could be saved, thereby recovering biomass synthesis. The rebalanced energy supply eventually contributed to the rapid cell growth as well as efficient xylose utilization in CGS15 (Table 2).

(iii) A synergistic effect on carbon metabolism and energy supply endowed the evolved strain with efficient substrate utilization and a fast growth phenotype. Following the identification of the xylose regulatory mechanism mediated by the transcription factor IpsA and the summary of global regulatory rewiring in the reconstructed strain, a systemic and progressive adaptation of the cells to heterologous xylose metabolism was revealed (Fig. 9). First, the synergistic optimization of the expression of the heterologous xylose isomerase pathway and the xylose transporter successfully channeled carbon fluxes into the central metabolism of CGS11 (Fig. 6c) but failed to efficiently guide them into the energy metabolism or the supply of other biomass monomers (Fig. 7c and 8c). The surge of nonnatural xylose imports unexpectedly results in a global regulatory impact of the transcription factor IpsA (Fig. 5f) on the endogenous metabolic network (Fig. 6b, 7a and b, and 8a), which led to an imbalance between carbon and energy metabolism in CGS11 (Fig. 6 to 8; Fig. S11). The generally similar trend of transcription level change of “CGS11 versus CGS7” and “ΔipsA strain versus the wild type” (42) (Fig. S12) might explain why the phenotypes of CGS8 and CGS11 are similar to that of the ΔipsA strain. Then, the cellular regulatory network was rewired to merge the enhanced heterologous xylose metabolism with the endogenous carbon and energy metabolism. Specifically, the emergence of ipsA^C331T (encoding the IpsA^P111S variant) into CGS11 guided excess carbon fluxes away from the central metabolism toward the amino acid synthesis and energy supply (Fig. 7 and 8; Fig. S11). The likely molecular mechanism is an increase in the opening amplitude of the two structural domains when the 111st hydrophobic and structurally rigid proline is replaced with hydrophilic serine (Fig. 5g). This results in the preservation of the opening conformation despite the binding of the effector, thus allowing the protein-DNA binding state to be effectively maintained and leading to a higher binding affinity to IpsA-regulated DNA elements and higher resistance to effector molecules (Fig. 5a to d; Fig. S7). Finally, the recoordination between the carbon and energy metabolism endowed CGS15 with a more comprehensive adaptation to the xylose metabolism, which not only finally enabled rapid xylose utilization and cell growth (Fig. 1c) but also released the CCR effect of the naturally preferred glucose (Fig. 3).

FIG 9.

Schematic diagram of the mechanism of rapid xylose metabolism in CGS15. Heat maps (log₂ fold change) show the changes of intracellular metabolite levels related to the core carbon metabolism in CGS15 compared with those in CGS11. The bold solid arrows and dotted arrows indicate metabolic reactions catalyzed by genes whose transcriptional levels are upregulated and downregulated, respectively, in CGS15 compared with CGS11.

Conclusions.

Considering all these steps together, a stable, genetically clear, modifiable, and rapid xylose-(co)utilizing C. glutamicum chassis was constructed and the adaptation of this chassis to nonnatural xylose metabolism was systematically elucidated. The identification and characterization of three positive mutations (P_sod(C131T)-xylAB, P_tuf(Δ21)-araE, and ipsA^C331T) not only provided elements for further engineering cellular metabolism in C. glutamicum but also reproduced the rebalancing of cellular carbon metabolism and energy supply for rapid xylose metabolism. For the first time, a global regulatory and metabolic effect responding to evolved xylose metabolism was uncovered, which enriches the understanding and future engineering targets of xylose metabolism.

MATERIALS AND METHODS

Reagents, strains, and media.

Plasmids were extracted using the AxyPrep plasmid miniprep kit (Axygen, USA). Yeast extract and tryptone were purchased from Sangon Biotech (Shanghai, China). Brain heart infusion (BHI) broth was purchased from Hopebio (Qingdao, China). Antibiotics and other reagents such as xylose and glucose were from Sangon Biotech (Shanghai, China).

All strains and plasmids used in this study are listed in Table 1 and in Table S1 in the supplemental material. C. glutamicum strain CGS4 (14), with deletions of ldhA, pqo, cat, and pta-ackA, replacements of the native promoters of pyc, ppc, tkt, and tal with the P_sod promoter, and chromosomal integration of the xylose transporter araE from Bacillus subtilis under the control of the constitutive endogenous tuf promoter at the ΔldhA locus (Δldh Δpta-ackA Δpqo Δcat P_sod-pyc P_sod-ppc P_sod-tal P_sod-tkt ldhA::P_tuf-araE), was used as the host strain. Escherichia coli DH5α was used for plasmid construction and was grown aerobically at 37°C in Luria-Bertani (LB) medium containing (per L) 10 g tryptone, 5 g yeast extract, and 10 g NaCl. The modified CGIII complex medium (pH 7.4), containing (per L) 10 g tryptone, 10 g yeast extract, and 21 g 3-morpholinopropanesulfonic acid (MOPS), was used for precultures of C. glutamicum. Aerobic growth of C. glutamicum was conducted in CGXII medium (pH 7.0) containing the following (per L): 20 g (NH₄)₂SO₄, 5 g urea, 1 g KH₂PO₄, 1 g K₂HPO₄, 0.25 g MgSO₄·7H₂O, 10 mg CaCl₂, 10 mg FeSO₄·7H₂O, 0.1 mgMnSO₄·H₂O, 1 mg ZnSO₄·7H₂O, 0.2 mg CuSO₄·5H₂O, 20 μg NiCl₂·6H₂O, 0.2 mg biotin, and 21 g MOPS, with the indicated amounts of sugars. Anaerobic cultivation of C. glutamicum was conducted in modified mineral salts medium containing the following (per L): 9 g NaCl, 5 g urea, 10 mg FeSO₄·7H₂O, 0.1 mg MnSO₄·H₂O, 1 mg ZnSO₄·7H₂O, 0.2 mg CuSO₄·5H₂O, 20 μg NiCl₂·6H₂O, 0.2 mg biotin, and the indicated amounts of sugars. Antibiotics were added to the media at the following concentrations when required: 50 μg/mL kanamycin for E. coli and 25 μg/mL kanamycin for C. glutamicum. Corn stalk hydrolysate was obtained using the commercial cellulolytic enzyme reagent Cellic CTec2 (54) from HebaBiz Pharmaceutical Co. Ltd., China. Before use as a substrate for succinate production, the hydrolysate was sterilized at 115°C for 10 min.

Culture conditions.

The strains were stored at −80°C and revived by growth on BHI agar plates at 30°C. Single colonies were transferred into 5 mL of BHI medium in a tube and grown at 30°C and 220 rpm overnight, after which 0.5 mL of the resulting seed culture was used to inoculate 50 mL of modified CGIII medium supplemented with 20 g/L glucose in a 500-mL shake flask and grown to an OD₆₀₀ of about 10. If not otherwise specified, seed cultures in CGIII medium were washed twice with CGXII medium for aerobic growth characterization of C. glutamicum and used to inoculate 50 mL of CGXII medium with 20 g/L xylose in a 500-mL shake flask to an initial OD₆₀₀ of 0.3, followed by cultivation at 30°C and 220 rpm.

To evaluate anaerobic sugar utilization, when the CGXII cultures reached an OD₆₀₀ of 20, the cells were harvested by centrifugation (5 000 × g, 4°C, 10 min), washed with modified mineral salts medium, and resuspended to an OD₆₀₀ of 30 in modified mineral salts medium with the indicated sugars and 200 mM sodium bicarbonate. The cells were then cultured in 50-mL serum bottles at 37°C and 220 rpm on a rotary shaker. To prevent acidification, 60 g/L of magnesium carbonate hydroxide [4MgCO₃·Mg(OH)₂·5H₂O] was added as a buffering agent.

For anaerobic succinate production, when the CGXII cultures reached an OD₆₀₀ of 20, the cells were harvested by centrifugation (5,000 × g, 4°C, 10 min), washed with modified mineral salts medium, and resuspended to an OD₆₀₀ of 150 in modified mineral salts medium with the indicated sugars and 300 mM sodium bicarbonate or in corn stalk hydrolysate with 300 mM sodium bicarbonate. The cells were then cultured in 50-mL serum bottles at 30°C and 220 rpm on a rotary shaker. To prevent acidification, 120 g/L of magnesium carbonate hydroxide [4MgCO₃·Mg(OH)₂·5H₂O] was added as a buffering agent. To induce the expression of heterologous genes in a multicopy vector, isopropyl β-d-1-thiogalactopyranoside (IPTG; Sangon Biotech, China) was added to a final concentration of 2 mM.

Construction of plasmids and strains.

All the primers used in this study are listed in Table 4. The genetic modification of C. glutamicum was achieved via a two-step homologous recombination procedure using the suicide vector pD-sacB. To construct strain CGS6, the plasmid pD-pta-ackA-P_sod-xylAB was constructed as follows: the sequences of the P_sod promoter and xylAB operon were amplified from genomic DNA of C. glutamicum ATCC 13032 and the pX-xcbAB plasmid (14) by PCR using the primer pairs pD-xylAB-1/pD-xylAB-2 and pD-xylAB-3/pD-xylAB-4, respectively, and fused by PCR. The resulting product was digested with BamHI and SalI and ligated between the corresponding sites of pD-pta-ackA.

TABLE 4.

Primers used in this study

Plasmid/primer	Sequence (5′–3′)a
pD-pta-ackA-Psod-xylAB
pD-xylAB-1	CGTCGCTCGAGCATATGTAGCTGCCAATTATTCCGGG
pD-xylAB-2	CGATGAAAACGGTGTTGCTCATGGGTAAAAAATCCTTTCGTAGG
pD-xylAB-3	CTACGAAAGGATTTTTTACCCATGAGCAACACCGTTTTCATCGGCG
pD-xylAB-4	AGTCCTTAAGGTCGACAGAGTTTGTAGAAACGCAAAAAG
pD-RM-Psod(C131T)-xylAB
pD-RM-xylAB-F	GACCGTCGGATCCTAATGGGGGGTGAAGAGCTGTAAAGTACC
pD-RM-xylAB-R	TGGTTAAGCTTTAGCGCGGGTGCGAGAACAGGTTGGCGGT
pD-RM-Ptuf(Δ21)-araE
pD-RM-araE-F	CATTCGGATCCCCGGAACTAGCTCTGCAATGAC
pD-RM-araE-R	TTGAGAAGCTTGTAAAGAGCTGATATAACGAAGAT
pD-RM-NCgl2418535T536
pD-RM-2418-F	TCTGCGGATCCCTCTACCGGGCTATTACCGCGGGTG
pD-RM-2418-R	TCCCTCAAGCTTACACCTTCAGGATCAAGGTTCACTCC
pD-RM-ipsAC331T
pD-RM-ipsA-F	TACCGGGATCCCCATCGCCGCGAAACTAGGTATCTCAAGGAC
pD-RM-ipsA-R	CGGTTAAGCTTAACGCCGAATGCCAGTGCATCGACGGTAC
pEC-Psod-gfp/pEC-Psod(C131T)-gfp
Psod-gfp-1	CTGACGAGCTCTAGCTGCCAATTATTCCGGGCTT
Psod-gfp-2	AAGTTCTTCTCCTTTACGCATGGGTAAAAAATCCTTTCGTAG
Psod-gfp-3	GAAAGGATTTTTTACCCATGCGTAAAGGAGAAG
Psod-gfp-4	ATGACTCTAGATTATTTGTATAGTTCATCCATGCCAT
pEC-Ptuf-gfp/pEC-Ptuf(Δ21)-gfp
Ptuf-gfp-1	CTGTAGAGCTCTGGCCGTTACCCTGCGAATGTCC
Ptuf-gfp-2	CAGTGAAAAGTTCTTCTCCTTTACGCATTGTATGTCCTCCTGGACTTCG
Ptuf-gfp-3	CCAGGAGGACATACAATGCGTAAAGGAGAAGAACTT
Ptuf-gfp-4	ATGACTCTAGATTATTTGTATAGTTCATCCATGCCAT
Ptuf(Δ21)-gfp-2	CGAAAGTCGTAGCCACCACGAAGTCCATGCGTAAAGGAGAAGAACTTTT
Ptuf(Δ21)-gfp-3	AAGTTCTTCTCCTTTACGCATGGACTTCGTGGTGGCTACGACTTTC
pEC-pycsucE
pycsucE-1	ACATGTCTAGAATGTCGACTCACACATCTTC
pycsucE-2	CCAGCCTTCTTCGCGGCGGCTACCGCGCGAGACTTAT
pycsucE-3	CGGTAGCCGCCGCGAAGAAGGCTGGTCT
pycsucE-4	CTGCAACAGCGGTCTTGGTGGTTGTCCTCCTTTTTAGGAAACGACGACGATC
pycsucE-5	CTTGATCGTCGTCGTTTCCTAAAAAGGAGGACAACCACCAAGACCGCTGTTGCAGT
pycsucE-6	CAGTGCCTGCAGGTGCGCTTAAGGGGTCAATGCTAG
pET28a (+)-ipsA/ipsAC331T
pET28a-ipsA-F	CAATGGAGCTCATGATTATGGGTAGGAAACAAC
pET28a-ipsA-R	TGGATAAGCTTCTAGATTGGCGCAACCGTGGAACCTG
ino1 promoter fragment
ino1-1	CAGGGTACGGACATACCATTC
ino1-2	CGTTCCAAAATGTGGGGATTCC
ino1-a-1	AATGGGGGTAATGCTTGATCGATCAATTGA
ino1-a-2	TCAATTGATCGATCAAGCATTACCCCCATT

^aThe underlined nucleotides indicate the restriction sites for the appropriate enzymes.

For the reconstruction of strains with rapid xylose metabolism, the plasmid pD-RM-P_sod(C131T)-xylAB was constructed. The flanking sequences of the four mutations were amplified from genomic DNA of Cev2-18-5 using primer pair pD-RM-xylAB-F/pD-RM-xylAB-R. The PCR product was digested with BamHI and HindIII and ligated into pD-sacB digested with the same enzymes. The plasmids pD-RM-P_tuf(Δ21)-araE, pD-RM-ipsA^C331T, and pD-RM-NCgl2418^535T536 were constructed analogously, using BamHI and HindIII.

The plasmid pEC-XK99E was used for plasmid-based gene expression. The plasmid pEC-P_sod-gfp was constructed for overexpression of the gfp gene (encoding green fluorescent protein). The sequences of the P_sod promoter and the gfp gene were amplified from genomic DNA of C. glutamicum ATCC 13032 and the pXMG plasmid by PCR using the primer pairs P_sod-gfp-1/P_sod-gfp-2 and P_sod-gfp-3/P_sod-gfp-4, respectively, and fused by PCR. The resulting product was digested with SacI and XbaI and ligated between the corresponding sites of pEC-XK99E. The pEC-P_sod(C131T)-gfp, pEC-P_tuf-gfp, and pEC-P_tuf(Δ21)-gfp plasmids were also constructed the same way, using SacI and XbaI. The pEC-pycsucE plasmid was constructed for overexpression of pyc^T132A and the sucE gene. pyc^T132A and the sucE gene were amplified from the genome of C. glutamicum ATCC 13032 using the primer pairs pycsucE-1/pycsucE-2, pycsucE-3/pycsucE-4, and pycsucE-5/pycsucE-6 and fused by PCR. The mutant site of T132A was introduced into primers pycsucE-2 and pycsucE-3, and the pyc^T132A gene was generated after PCR fusing. The resulting product was digested with XbaI and SbfI and ligated between the corresponding sites of pEC-XK99E.

The plasmid pET28a (+)-ipsA/ipsA^C331T was used for IpsA and IpsA^P111S expression. The sequence of ipsA or ipsA^C331T was amplified from genomic DNA of C. glutamicum ATCC 13032 or Cev2-18-5 by PCR using primer pair pET28a-ipsA-F/pET28a-ipsA-R. The resulting product was digested with SacI and HindIII and ligated between the corresponding sites of pET28a (+).

Genome resequencing.

The genomic DNA of strains CGS6, Cev1-22-10, and Cev2-18-5 was extracted using a genomic DNA kit (Qiagen, Beijing, China) from cells that were grown to the mid-exponential phase at 30°C and 220 rpm. Next-generation sequencing library samples were prepared in accordance with the manufacturer’s protocol of the NEBNext ultra DNA library prep kit from Illumina (CA, USA). Sequencing was carried out using a 2 × 150 paired-end (PE) configuration. For quality control, Cutadapt (v1.9.1) was used to remove the sequences of adaptors, PCR primers, content of N bases more than 10%, and bases with a quality lower than 20. The C. glutamicum ATCC 13032 genome (GenBank accession number NC_003450.3) and the genetically modified background of CGS6 (Table 1) were used as references. After comparative analysis of the genomes, the identified mutations were verified using Sanger sequencing. The resequencing data have been deposited in an online database under BioProject accession no. PRJNA763329.

Transcriptome analysis.

Transcriptome sequencing (RNA-seq) was performed on samples from CGS7, CGS11, and CGS15 in three independent replicates for each. The total RNA of each sample was extracted using an RNeasy minikit (Qiagen, Beijing, China) with cells grown to the mid-exponential phase at 30°C and 220 rpm. Total RNA was qualified and quantified using an Agilent 2100 bioanalyzer (Agilent, Palo Alto, CA, USA), a NanoDrop spectrophotometer (Thermo, Waltham, MA, USA), and 1% agarose gel electrophoresis. Next-generation sequencing library samples were prepared according to the manufacturer’s protocol. Then, libraries with different indices were multiplexed and loaded onto an Illumina HiSeq instrument according to the manufacturer’s instructions (Illumina, San Diego, CA, USA). Sequencing was carried out using a 2 × 150 PE configuration; image analysis and base calling were conducted on the HiSeq instrument using HiSeq Control software (HCS), OLB, and GAPipeline-1.6 (Illumina, San Diego, CA, USA). For quality control, pass filter data of fastq format were processed by Cutadapt (v1.9.1) to remove technical sequences, including adapters, PCR primers, or fragments thereof, and bases with a quality lower than 20. The sequences were processed and analyzed by Genewiz (Suzhou, China). RNA-seq data were assembled and analyzed by comparison with the translated regions of the annotated reference genome (GCA_000011325.1) using HTSeq (v0.6.1p1). The estimation of fold changes and other statistical analyses were performed using DESeq2. KEGG and GO enrichment analyses were carried out using in-house scripts and GOseq (v1.34.1). Heat maps (log₂ fold change) showing the changes in gene transcription levels were constructed using TBtools (https://github.com/CJ-Chen/TBtools) (55). The transcription data have been deposited in the GEO online database under accession no. GSE184402.

Metabolome analysis.

Samples for metabolome analysis were prepared as described previously (56). A sample comprising 3 mL of a C. glutamicum culture grown for 12 h was quenched by quickly injecting it into 15 mL of 40% methanol that was prechilled to −20°C and quickly mixing it by vortexing (within 1 s). Three parallel samples were taken within 1 min. The mixture was centrifuged (14,000 × g, 2 min, 0°C), and the supernatant was removed. Then, the quenched cells were resuspended in methanol that was prechilled to −20°C and centrifuged at 14,000 × g for 2 min at 4°C, after which the supernatant was collected into a precooled tube and stored at −20°C. The cell pellets were resuspended in 2 mL acidic acetonitrile-water (1:1, vol/vol, with 0.1% formic acid) that was prechilled to −20°C and placed in an ice water bath for 15 min. The second mixture was centrifuged at 14,000 × g for 5 min at 4°C. Then, the cell pellets were resuspended with 100°C ethanol-water (3:1, vol/vol) and processed in a boiling water bath for 15 min. The third mixture was centrifuged at 14,000 × g for 5 min at 4°C. The first, second, and third extraction supernatants were mixed and centrifuged at 14,000 × g for 5 min at 4°C. The supernatant was then collected, freeze-dried, and stored at −80°C.

An ultrahigh-performance liquid chromatography (UHPLC) LC-30A system (Shimadzu, Kyoto, Japan) equipped with a TripleTOF 6600 mass spectrometer (Sciex, Framingham, MA, USA) was used for LC-tandem mass spectrometry (MS/MS) analysis. The samples were separated using an Xbridge BEH amide column (100 mm by 2.1 mm, 1.7-μm inner diameter) (Waters, Milford, MA, USA). Solvents were composed of water-acetonitrile-ammonium acetate-ammonium hydroxide (solvent A, 100%/0%/25 mM/10 mM; solvent B, 0%/100%/0 mM/0 mM). The LC method was as follows: 0 to 2 min, 95% B; 2 to 18 min, 95 to 65% B; 18 to 20 min, 65 to 40% B: 20 to 22 min, 40% B; 20 to 20.1 min, 40 to 95% B; and 22.1 to 30 min, 95% B. A flow rate of 0.3 mL/min was employed. The MS parameters were as follows: electrospray ionization (ESI) source; negative mode; ion voltage, 4,500 V; declustering potential, 80 V; source temperature, 550°C; curtain gas, 35 lb/in²; nebulizer gas, 55 lb/in²; and heater gas, 55 lb/in². Each scan cycle contained one time-of-flight (TOF) MS survey scan and 15 MS/MS scans. The mass ranges were m/z 63 to 1,000 for TOF MS and m/z 30 to 1,000 for MS/MS. Acquisition of MS/MS spectra was controlled by the information-dependent acquisition (IDA) function of the Analyst TF 1.7 software (Sciex, Framingham, MA, USA) with dynamic background subtraction on. Mass accuracy was calibrated by an automated calibrant delivery system (Sciex, Framingham, MA, USA) interfaced with the second inlet of the DuoSpray source. Calibration was performed for every five samples. Data analysis for LC-MS/MS was performed with PeakView 2.0 software (Sciex, Framingham, MA, USA). Metabolite identification was carried out by comparing their retention times, accurate masses, and MS/MS spectra with those of authentic standards. The relative content of metabolites, including NADH, NAD⁺, ATP, and ADP, was normalized to cell density (DCW = 0.25 × OD₆₀₀ [g/L])(53). Each metabolomic experiment was repeated four times. Heat maps (log₂ fold change) showing the changes of relative intracellular metabolite concentrations were constructed using TBtools (https://github.com/CJ-Chen/TBtools) (55).

RT-PCR.

Total RNA was reverse transcribed into cDNA by Genewiz (Suzhou, China). The transcriptional levels of different genes were identified by fluorescence real-time quantitative RT-PCR using the cDNA of the indicated strains cultured at 220 rpm and 30°C in CGXII medium with 20 g/L xylose until the exponential growth phase was reached. The fluorescence RT-PCR was performed using the TransStart top green qPCR superMix (TransGen, Beijing, China) on a LightCycler 480 device (Roche, Basel, Switzerland). The transcriptional level of the 16S rRNA was used as an internal reference. The relative expression of the target gene was normalized to the endogenous reference gene (16S rRNA). Fold change was determined by the comparative threshold cycle (C_T) method.

Analytical techniques.

Extracellular organic acids and pentoses were measured by HPLC as described previously (14). Glucose was monitored using an SBA sensor instrument (Institute of Microbiology, Shandong, China). Growth was monitored by measuring the OD₆₀₀ using a conventional spectrophotometer (TU-1901; PUXI, Beijing, China). The specific growth rate was calculated using the formula d(lnW₂ – lnW₁)/d(t₂ – t₁), where W₁ and W₂ are the biomass values at times t₁ and t₂. The average xylose consumption rate (q) was calculated using the formula q = (C₂ – C₁)/(t₂ – t₁), where C₁ and C₂ are the xylose concentration values at times t₁ and t₂. The specific sugar consumption rate was calculated using the formula 2q/(W₁ + W₂), where q is the average xylose consumption rate at times t₁ and t₂, and W₁ and W₂ are the biomass values at times t₁ and t₂. Expression of GFP was measured using a Varioskan LUX fluorescence microplate reader (Thermo Fisher Scientific, CA, USA) in black 96-well plates with excitation and emission wavelengths of 488 and 530 nm, respectively. The GFP fluorescence was normalized to the cell density.

IpsA/IpsA^P111S expression and purification.

E. coli BL21(DE3) containing pET28a(+)-ipsA/ipsA^C331T was cultured in 5 mL of LB medium containing 50 μg/mL kanamycin at 37°C and 220 rpm overnight. Then, the culture was used to inoculate 200 mL of LB medium in 1-L flasks at a ratio of 1% (vol/vol). To induce plasmid expression, 0.5 mM IPTG (final concentration) was added when the culture OD₆₀₀ reached 0.6 to 0.8, and the cultivation temperature was set and maintained at 16°C for 16 to 18 h. Recombinant cells were harvested by centrifugation at 4°C, 4,000 × g, for 40 min and subsequently resuspended in 20 mL phosphate buffer (50 mM, pH 7.5) containing 150 mM NaCl. The cell pellets were lysed using a high-pressure homogenizer at 4°C. After centrifugation at 4°C, 4,000 × g, for 40 min, the crude extract was purified using a Ni-nitrilotriacetic His-binding column and concentrated as described previously (57). The purified proteins were stored at −80°C with 10% glycerol. The purity of proteins was examined by 12% SDS-PAGE, and the concentrations of proteins were measured using a Bradford assay kit (CWBiotech, Beijing, China).

SPR analysis.

Before surface plasmon resonance (SPR) analysis, the IpsA/IpsA^P111S stored at −80°C was further purified in an ÄKTA purifier 10 system (GE Healthcare) by gel filtration on a Superdex200 10/300 GL column (GE Healthcare, Madison, WI, USA) in running buffer A, consisting of 50 mM Tris-HCl (pH 8), 250 mM NaCl, and 1 mM dithiothreitol (DTT), at a flow rate of 0.5 mL/min. Every 0.5 mL of the running buffer containing the purified protein fractions was collected when the ÄKTA UV₂₈₀ value rose above 5. The collected supernatant was immediately analyzed by SDS-PAGE and the Bradford assay kit (CWBiotech, Beijing, China). Fractions with relatively high protein purity and concentration were used for subsequent SPR analysis.

The interaction of double-stranded DNA (dsDNA) fragments with the transcription factor (IpsA and IpsA^P111S) was monitored by SPR experiments carried out at 25°C using a BIAcore T200 system (Cytiva, Uppsala, Sweden). To capture the biotinylated dsDNA fragments, the series S sensor chip (Cytiva, Uppsala, Sweden) was first immobilized with streptavidin in all the flow cells by amine-coupling chemistry, and then different dsDNA fragments were separately captured in the corresponding flow cells at a response unit level of about 100, except for the negative reference channel. The size exclusion chromatography (SEC)-purified IpsA and IpsA^P111S were diluted using a 2-fold gradient from 5 μM with running buffer A. During each cycle, the protein solution was injected for 120 s at a flow rate of 40 μL/min and dissociation was monitored for 90 s. The chip surface was regenerated with 0.1% SDS between each cycle. Raw data were processed and analyzed using BIAcore T200 Evaluation software. The apparent equilibrium dissociation constants (K_D) for each DNA-protein interaction were calculated by fitting each set of binding curves to the single binding site model of kinetics.

MD simulations.

The three-dimensional (3D) complex structure of IpsA (PDB ID 3H5T) was extracted from the Protein Data Bank (PDB; https://www.rcsb.org), and the mutant protein structure was constructed based on its wild-type structure using Schrödinger 2018 (58). Then, 1,000-ns molecular dynamics (MD) simulations were performed on both systems using Amber 2018 software (59) with the ff14SB forcefield. The complete simulation methodology used in this work is available in Text S1, note 8, in the supplemental material. The molecular dynamics trajectories were collected for conformation analysis.