Reconstruction of the Diaminopimelic Acid Pathway to Promote L-lysine Production in Corynebacterium glutamicum

The dehydrogenase pathway and the succinylase pathway are involved in the synthesis of L-lysine in Corynebacterium glutamicum. Despite the low contribution rate to L-lysine production, the dehydrogenase pathway is favorable for its simple steps and potential to increase the production of L-lysine. The effect of ammonium (NH4+) concentration on L-lysine biosynthesis was investigated, and the results indicated that the biosynthesis of L-lysine can be promoted in a high NH4+ environment. In order to reduce the requirement of NH4+, the nitrogen source regulatory protein AmtR was knocked out, resulting in an 8.5% increase in L-lysine production (i.e., 52.3 ± 4.31 g/L). Subsequently, the dehydrogenase pathway was upregulated by blocking or weakening the tetrahydrodipicolinate succinylase (DapD)-coding gene dapD and overexpressing the ddh gene to further enhance L-lysine biosynthesis. The final strain XQ-5-W4 could produce 189 ± 8.7 g/L L-lysine with the maximum specific rate (qLys,max.) of 0.35 ± 0.05 g/(g·h) in a 5-L jar fermenter. The L-lysine titer and qLys,max achieved in this study is about 25.2% and 59.1% higher than that of the original strain without enhancement of dehydrogenase pathway, respectively. The results indicated that the dehydrogenase pathway could serve as a breakthrough point to reconstruct the diaminopimelic acid (DAP) pathway and promote L-lysine production.


Introduction
L-lysine is an essential amino acid widely used in food, animal feed, medicine, cosmetics, and other industries [1]. Methods to produce L-lysine include albuminolysis, chemical method, enzymic method, and microbial fermentation. Microbial fermentation uses renewable feedstock and produces low amounts of pollutants [2]. Therefore, L-lysine is mainly produced by microbial fermentation in the industries, with C. glutamicum and Escherichia coli being the most commonly used strains [1].
The L-lysine biosynthesis includes two pathways, the diaminopimelic acid (DAP) pathway and the α-aminoadipic acid (AAA) pathway. In the AAA pathway, L-lysine is synthesized from α-ketoglutarate and acetylcoenzyme A (acetyl-CoA) in which α-aminoadipic acid serves as an intermediate metabolite. The AAA pathway is commonly found in yeast, fungi, and some species in the domain Archaea [3,4]. In the DAP pathway, however, L-lysine is synthesized from aspartate and pyruvate in which meso-diaminopimelic acid (meso-DAP) serves as an intermediate metabolite ( Figure 1) [5]. Commonly found in archaea, algae, fungi, plants, and bacteria [6], the DAP pathway starts with the biosynthesis of L-∆ 1 -tetrahydrodipicolinate (THDPA) from L-aspartate, which is then converted into meso-DAP, and finally, L-lysine is produced with diaminopimelate deacetylase (DAPDC,  C. glutamicum is a Gram-positive bacterium isolated from soil in 1957 [17,18]. C. glutamicum is often used to produce amino acids commercially, such as L-glutamic acid, L-lysine and L-arginine [19]. Interestingly, two variants of the DAP pathway are found in C. glutamicum, the succinylase pathway, and the dehydrogenase pathway. According to Figure 1, meso-DAP is biosynthesized from THDPA in one step via the dehydrogenase pathway [20] and in four steps via the succinylase pathway, both of which involve NH 4 + . In addition, the L-lysine biosynthesis efficiency of both pathways depends on the concentration of NH 4 + in the medium [16]. L-lysine was produced mainly through the dehydrogenase pathway at first but then entirely through the succinylase pathway [16] as the concentration of NH 4 + decreases during the fermentation, resulting in the decrease in the dehydrogenase activity [16,21]. Sonntag et al. reported that 33% of L-lysine was synthesized via the dehydrogenase pathway and 66% via the succinylase pathway in C. glutamicum [22]. Although the succinylase pathway is important for increasing the titer of L-lysine, the dehydrogenase pathway has potential in improving the production intensity of L-lysine as only one step is required to biosynthesize meso-DAP. Hence the question: does upregulating the dehydrogenase pathway promote L-lysine production in C. glutamicum? In this study, the C. glutamicum XQ-5 strain was developed, which took the dehydrogenase pathway for meso-DAP biosynthesis to promote L-lysine production. The DAP pathway of C. glutamicum XQ-5 was reconstructed to upregulate the dehydrogenase pathway in L-lysine biosynthesis during the fermentation process. The strategies of this study include as follows: (1) investigating the effect of different NH 4 + concentrations on L-lysine production; (2) alleviating the nitrogen limitation to improve the efficiency of L-lysine production; (3) rationally regulating the two pathways to promote L-lysine production. As a result, a recombinant strain C. glutamicum XQ-5-W4 (i.e., C. glutamicum XQ-5-dapD W ∆amtR/pEC-ddh) derived from C. glutamicum XQ-5 was obtained, which produced 189 ± 8.7 g/L L-lysine with the maximum specific rate (q Lys,max. ) of 0.35 ± 0.05 g/(g·h) in a 5-L jar fermenter, which were 25.2% and 59.1% higher than that of the original strain C. glutamicum XQ-5, respectively.

The Effects of Different Ammonium (NH 4 + ) Concentrations on L-lysine Production
Nitrogen is one of the essential nutrients for living cells. The nitrogen utilization is very important for the growth of bacteria. Ammonium (NH 4 + ) is the standard component of the growth medium and the preferred nitrogen source for many bacteria. Both variants of the DAP pathway in C. glutamicum require the participation of NH 4 + . Therefore, the effect of NH 4 + on L-lysine production was investigated. To do this, eight NH 4 + solutions of different concentrations (i.e., 50, 100, 200, 250, 300, 350, 400, 500 mM) were added to the culture mediums and the cell growth, L-lysine production as well as q Lys were monitored. The results showed that the growth of strain XQ-5 was significantly inhibited from high NH 4 + concentration, but the effect of low NH 4 + concentration is relatively insignificant (Figure 2a). Xu et al. observed similar results in which high NH 4 + concentration was toxic to the bacteria [23]. Interestingly, too high or too low NH 4 + concentration is not conducive to L-lysine accumulation (Figure 2b). The L-lysine yield was relatively high at NH 4 + concentrations of 250 mM, 300 mM and 350 mM, which were 42.3 ± 2.31 g/L, 48.2 ± 3.54 g/L, and 43.9 ± 3.78 g/L, respectively. Interestingly, the maximum specific rate (q Lys,max. ) was found with a 350 mM NH 4 + solution (i.e., 0.22 ± 0.03 g/(g·h)), which was 10% higher than that with a 300 mM NH 4 + solution (i.e., 0.20 ± 0.01 g/(g·h)) ( Figure 2c). These results indicated that an appropriate increase in NH 4 + supply is necessary for the effective biosynthesis of L-lysine. Previous research reported that the dehydrogenase pathway is only active at high NH 4 + concentration [22]. Therefore, increasing the concentration of NH 4 + may upregulate the dehydrogenase pathway and promote L-lysine biosynthesis. The transcription level of the DapDH-coding gene ddh was measured. As might be expected, the transcription level of ddh gene increased with the increase of NH 4 + concentration within a certain range (Figure 2d). It is safe to conclude that the increase of q Lys,max at high NH 4 + Int. J. Mol. Sci. 2021, 22, 9065 4 of 17 concentration was related to the increase in ddh gene transcription level, and upregulating the dehydrogenase pathway is conducive to promoting L-lysine biosynthesis.
at high NH4 + concentration [22]. Therefore, increasing the concentration of NH4 + may upregulate the dehydrogenase pathway and promote L-lysine biosynthesis. The transcription level of the DapDH-coding gene ddh was measured. As might be expected, the transcription level of ddh gene increased with the increase of NH4 + concentration within a certain range (Figure 2d). It is safe to conclude that the increase of qLys,max at high NH4 + concentration was related to the increase in ddh gene transcription level, and upregulating the dehydrogenase pathway is conducive to promoting L-lysine biosynthesis.

Effect of The Upregulated Dehydrogenase Pathway on L-lysine Biosynthesis
As mentioned above, upregulating ddh gene promotes the biosynthesis of L-lysine. It is speculated that upregulating the dehydrogenase pathway may further increase L-lysine production. In order to enhance the effect of the dehydrogenase pathway on L-lysine production, the ddh gene was overexpressed to investigate whether the qLys and L-lysine production were improved. The ddh gene was ligated into the plasmid pEC-XK99E, which was then introduced into strain XQ-5 to give the target strain XQ-5/pEC-ddh (i.e., strain XQ-5-1) (Figure 3a). The growth rate, L-lysine production and qLys of strain XQ-5-1 were measured at shake flask fermentation for 72 h, and the effect of the upregulated dehydrogenase pathway on L-lysine production was investigated.

Effect of The Upregulated Dehydrogenase Pathway on L-lysine Biosynthesis
As mentioned above, upregulating ddh gene promotes the biosynthesis of L-lysine. It is speculated that upregulating the dehydrogenase pathway may further increase Llysine production. In order to enhance the effect of the dehydrogenase pathway on Llysine production, the ddh gene was overexpressed to investigate whether the q Lys and L-lysine production were improved. The ddh gene was ligated into the plasmid pEC-XK99E, which was then introduced into strain XQ-5 to give the target strain XQ-5/pEC-ddh (i.e., strain XQ-5-1) (Figure 3a). The growth rate, L-lysine production and q Lys of strain XQ-5-1 were measured at shake flask fermentation for 72 h, and the effect of the upregulated dehydrogenase pathway on L-lysine production was investigated.
According to Figure 3b, the growth of strain XQ-5-1 was not affected, remaining almost the same as the original strain. In addition, the final L-lysine production of strain XQ-5-1 was 49.3 ± 3.21 g/L, slightly increased compared with strain XQ-5 (i.e., 48.2 ± 3.54 g/L) ( Figure 3c). The results showed that the upregulated dehydrogenase pathway had little effect on L-lysine production. Similar results were also found in previous reports in which the overexpression of ddh gene on plasmid had no effect on the L-lysine production of C. glutamicum [24]. It is speculated that the intracellular NH 4 + concentration limits the increase of L-lysine production in strains with overexpressed ddh. Due to the low affinity to NH 4 + , the dehydrogenase pathway is only adoptable at high NH 4 + concentration [16]. However, it is heartening that XQ-5-1 produced the same amount of L-lysine as the original strain in a shorter time (Figure 3c). With the same NH 4 + concentration, the q Lys,max of XQ-5-1 was 5-10% higher than that of the original strain (Figures 2c and 3d). Previous research suggested that the level of extracellular NH 4 + concentration is higher than the intracellular NH 4 + concentration as the utilization of nitrogen is controlled by AmtR [25]. Then, is the increase in intracellular NH 4 + concentration beneficial to L-lysine production?
According to Figure 3b, the growth of strain XQ-5-1 was not affected, remaining almost the same as the original strain. In addition, the final L-lysine production of strain XQ-5-1 was 49.3 ± 3.21 g/L, slightly increased compared with strain XQ-5 (i.e., 48.2 ± 3.54 g/L) (Figure 3c). The results showed that the upregulated dehydrogenase pathway had little effect on L-lysine production. Similar results were also found in previous reports in which the overexpression of ddh gene on plasmid had no effect on the L-lysine production of C. glutamicum [24]. It is speculated that the intracellular NH4 + concentration limits the increase of L-lysine production in strains with overexpressed ddh. Due to the low affinity to NH4 + , the dehydrogenase pathway is only adoptable at high NH4 + concentration [16]. However, it is heartening that XQ-5-1 produced the same amount of L-lysine as the original strain in a shorter time (Figure 3c). With the same NH4 + concentration, the qLys,max of XQ-5-1 was 5-10% higher than that of the original strain (Figures 2c and 3d). Previous research suggested that the level of extracellular NH4 + concentration is higher than the intracellular NH4 + concentration as the utilization of nitrogen is controlled by AmtR [25]. Then, is the increase in intracellular NH4 + concentration beneficial to L-lysine production?

Effect of Gene amtR Deletion on NH4 + Utilization and L-lysine Synthesis
As mentioned above, L-lysine production showed no significant increase with the overexpression of ddh, possibly due to the low intracellular NH4 + concentration. To address this limitation, the regulatory proteins controlling the utilization of NH4 + should be inactivated. In C. glutamicum, the expression of genes responding to nitrogen utilization is controlled by the TetR-type regulator AmtR [25]. In contrast with most TetR-type regu-

Effect of Gene amtR Deletion on NH 4 + Utilization and L-lysine Synthesis
As mentioned above, L-lysine production showed no significant increase with the overexpression of ddh, possibly due to the low intracellular NH 4 + concentration. To address this limitation, the regulatory proteins controlling the utilization of NH 4 + should be inactivated. In C. glutamicum, the expression of genes responding to nitrogen utilization is controlled by the TetR-type regulator AmtR [25]. In contrast with most TetR-type regulators, the dissociation of AmtR from its target promoters is triggered by complex formation of the PII-type signal transduction protein GlnK rather than by the binding of a low-molecular mass ligand [26]. The transcriptional regulator AmtR responses to changes in nitrogen levels, thus at least 35 genes involved in nitrogen utilization and metabolism needed to be regulated [27]. In order to increase the concentration of NH 4 + , the AmtR-coding gene amtR was deleted to give the target strain C. glutamicum XQ-5-∆amtR (i.e., XQ-5-2) (Figure 4a).
Interestingly, the q Lys,max of strain XQ-5-2 at 250 mM NH 4 + (i.e., 0.19 ± 0.02 g/(g·h)) was similar as the strain XQ-5 at 300 mM NH 4 + (i.e., 0.20 ± 0.01 g/(g·h)). Moreover, the q Lys,max of strain XQ-5-2 was almost the same at 300 mM (i.e., 0.22 ± 0.03 g/(g·h)) and 350 mM NH 4 + (i.e., 0.23 ± 0.04 g/(g·h)) ( Figure 4b). In addition, the highest L-lysine production (i.e., 52.3 ± 4.31 g/L) of strain XQ-5-2 was obtained at 300 mM NH 4 + , which was 8.5% higher than that of strain XQ-5 (i.e., 48.2 ± 3.54 g/L) (Figure 4d). Previous results also indicated that the deletion of the amtR gene increased the yield of L-lysine [28]. These results indicated that the alleviation of nitrogen restriction increased the intracellular NH 4 + concentration. Subsequently, the plasmid pEC-XK99E-ddh was introduced into strain XQ-5-2 to give the target strain XQ-5-2/pEC-ddh (i.e., strain XQ-5-3). As expected, the growth of strain XQ-5-2 was inhibited from high NH 4 + concentration (Figure 4c). It should be noted that the L-lysine yield of strain XQ-5-3 was 53.8 ± 3.98 g/L, which is similar to that of strain XQ-5-2 (i.e., 52.3 ± 4.31 g/L) (Figure 4d). These results indicated that the overexpression of ddh gene did not significantly increase the L-lysine production of strain XQ-5-2, possibly due to the fact that the dehydrogenase pathway is still not the dominant pathway for L-lysine production of strain XQ-5-3. According to Figure 1, the dehydrogenase pathway is a reversible reaction. In addition, previous reports pointed out that DapDH is highly specific to meso-DAP, while L,L-DAP and D,D-DAP are competitive inhibitors [29]. Theoretically, L,L-DAP, an intermediate in the succinylase pathway, inhibits the reverse reaction of DapDH [29]. Thus, the succinylase pathway was upregulated to supply L,L-DAP during the overexpression of ddh. In order to confirm this conjecture, the transcription levels of ddh and dapD were measured in strain XQ-5-3. The results showed that the transcription levels of ddh and dapD increased with the increase of NH 4 + concentration within a certain range (Figure 4e). The transcription level of ddh was 82% higher at high NH 4 + concentration (i.e., 500 mM) than that at low NH 4 + concentration (i.e., 50 mM). Similarly, the transcription level of dapD was 92% higher at high NH 4 + concentration (i.e., 500 mM) than that at low NH 4 + concentration (i.e., 50 mM). Moreover, it is reported that the expression level of dapD gene increases after the deletion of amtR gene [28]. These results indicated that the upregulated dapD expression level comes with the upregulated ddh expression level. However, DapD (tetrahydrodipicolinate succinylase, E.C. 2.3.1.117) has a high affinity with the substrate THDPA [15], making the succinylase pathway the main pathway for biosynthesizing L-lysine rather than the dehydrogenase pathway in spite of the overexpression of ddh.

Blocking the Succinylase Pathway to Upregulate the Dehydrogenase Pathway
Based on the above results, increasing the expression of ddh gene in the dehydrogenase pathway showed no significant effect on L-lysine production, possibly due to the fact that the expression level of dapD is increased with the overexpression of ddh. In order to upregulate the dehydrogenase pathway, the succinylase pathway was blocked. The succinylase pathway involves enzymes such as DapC (succinyl-amino-ketopimelate-coding gene dapC), DapE (N-succinyl-diaminopimelate desuccinylase-coding gene dapE), and DapF (diaminopimelate epimerase-coding gene dapF), respectively [30][31][32]. Previous research indicated that the genes dapC and dapE are dispensable for L-lysine overproduction in shake-flask cultures [33,34], whereas the genes dapF and dapD are indispensable for the succinylase pathway [15,33]. DapD is the first key enzyme in the succinylase pathway, thus DapD was inactivated to block the succinylase pathway, hence the DapD-deficient strain XQ-5-∆dapD (i.e., strain XQ-5-4). As a control, the DapDH-deficient strain XQ-5-∆ddh (i.e., strain XQ-5-5) was also constructed. The cell morphology of the three strains (i.e., a:strain XQ-5, b:strain XQ-5-4, c:strain XQ-5-5) were observed with FESEM, and the results indicated almost no change in cell morphology compared to the original strain (Figure 5a-c). It is well-known that meso-DAP connects the glycan backbone on the cell wall of many bacteria to give them shape and rigid structure [8]. These results showed that meso-DAP synthesized by either of these two pathways had met the needs of cell structure.
In order to investigate the L-lysine production of different strains under the different NH 4 + concentrations, three NH 4 + solutions of different concentrations (i.e., 250 mM, 300 mM, and 350 mM) were used in the test. According to Figure 5d, L-lysine production of strain XQ-5-4 decreased significantly, especially at low NH 4 + concentration (i.e., 17.6 ± 5.11 g/L), indicating that dapD gene is essential for the succinylase pathway and L-lysine production. By contrast, the L-lysine production of strain XQ-5-5 (i.e., 39.3 ± 4.11 g/L) decreased slightly. These results have once again proven that the succinylase pathway is the main pathway for L-lysine production rather than the dehydrogenase pathway [22]. It should be noted that the L-lysine production of strain XQ-5-4 increased at high NH 4 + concentration (Figure 5d), possibly due to the fact that the one-step dehydrogenase pathway compensated the L-lysine production at high NH 4 + concentration when the succinylase pathway was downregulated [34]. Among the recombinant strains, strain XQ-5-∆dapD∆amtR/pEC-ddh (i.e., strain XQ-5-8) accumulated the highest L-lysine production (i.e., 41.9 ± 4.57 g/L) if not counting the strains with succinylase pathway (Figure 5d). It is worth noting that the overexpression of ddh gene in strain XQ-5-6 increased the L-lysine production (300 mM) from 29.3 ± 3.49 g/L (strain XQ-5-6) to 41.9 ± 4.57 g/L (strain XQ-5-8). Interestingly, although the L-lysine yield of strain XQ-5-8 (i.e., 41.9 ± 4.57 g/L) was lower than that of strain XQ-5-3 (i.e., 53.8 ± 3.98 g/L), the q Lys,max. of strain XQ-5-8 (i.e., 0.30 ± 0.04 g/(g·h)) was 20% higher than that of strain XQ-5-3 (i.e., 0.25 ± 0.03 g/(g·h)) (Figure 5e), possibly due to the fact that the L-lysine precursor (i.e., meso-DAP) was biosynthesized in one step rather than four steps [35] since the dehydrogenase pathway is the only pathway for L-lysine production of strain XQ-5-8. These results also showed that the dehydrogenase pathway has potential to increase L-lysine production. Taken together, these results indicated that blocking the succinylase pathway is beneficial to upregulating the dehydrogenase pathway, thus improving the q Lys,max . However, the L-lysine production in strains with blocked the succinylase pathway decreased. Two reasons were proposed: (1) More NH 4 + were required to produce L-lysine in the dehydrogenase pathway. Previous researches indicated that C. glutamicum strain with the dehydrogenase pathway alone cannot produce L-lysine at low NH 4 + concentration [36] due to the low affinity of the dehydrogenase pathway to NH 4 + in the substrate [16]; (2) the dehydrogenase pathway is a reversible process. L,L-DAP cannot be synthesized while blocking the succinylase pathway, thus relieving the competitive inhibition of the reverse reaction [29]. These proposed reasons have been validated by examining the concentration of the by-products (Table 1). As might be expected, strain XQ-5-8 had the highest by-product concentration at high NH 4 + concentration ( Table 1). All of the above mentioned results indicated that blocking the succinylase pathway is beneficial to improve the production intensify of L-lysine, but unhelpful for increasing the yield of L-lysine.
drogenase pathway to NH4 + in the substrate [16]; (2) the dehydrogenase pathway is a reversible process. L,L-DAP cannot be synthesized while blocking the succinylase pathway, thus relieving the competitive inhibition of the reverse reaction [29]. These proposed reasons have been validated by examining the concentration of the by-products ( Table 1). As might be expected, strain XQ-5-8 had the highest by-product concentration at high NH4 + concentration ( Table 1). All of the above mentioned results indicated that blocking the succinylase pathway is beneficial to improve the production intensify of L-lysine, but unhelpful for increasing the yield of L-lysine.  Table 1. By-products of Corynebacterium glutamicum strains XQ-5-1, XQ-5-3, and XQ-5-8.   Table 1. By-products of Corynebacterium glutamicum strains XQ-5-1, XQ-5-3, and XQ-5-8.

Weakening the dapD Gene Makes the Metabolic Flux of the Two Pathways Reach the Best Balance
As mentioned above, L-lysine production decreased significantly while blocking the succinylase pathway (Figure 5d) and the amount of by-products increased because of the lack of L,L-DAP (Table 1). To address this problem, the key enzyme DapD in the succinylase pathway was weakened to balance the flux between the succinylase pathway and the dehydrogenase pathway.

Fed-Batch Fermentation of C. glutamicum XQ-5-W4
The production performance of strain XQ-5-W4 was investigated in a fed-batch process. As a comparison, fed-batch fermentation of the original strain XQ-5 was also conducted. Figure 7 shows the time profiles of fed-batch fermentations in a 5-L jar fermenter. Fed-batch fermentation of XQ-5-W4 resulted in 189 ± 8.7 g/L of L-lysine with a q Lys,max of 0.35 ± 0.05 g/(g·h) (Figure 7b,c). However, fed-batch fermentation of XQ-5 resulted in 151 ± 9.3 g/L of L-lysine with a q Lys,max of 0.22 ± 0.02 g/(g·h) (Figure 7a,c). Consistent with the results of production intensity of L-lysine in shake flasks, the q Lys,max of strain XQ-5-W4 was higher than that of strain XQ-5 (0.35 ± 0.05 g/(g·h) vs. 0.22 ± 0.02 g/(g·h)) ( Figure 7c). In addition, the L-lysine yield of strain XQ-5-W4 stabilized faster, about six hours earlier than that of strain XQ-5. However, the dry weight of strain XQ-5-W4 was 38.9 ± 5.12 g/L, 13.9% lower than that of the original strain (i.e., 45.2 ± 7.64 g/L) (Figure 7a,b). Previous research also found the similar result in which the biomass decreased while enhancing the yield of the target products [23]. Taken together, these results demonstrated that the final strain XQ-5-W4 shows an efficient L-lysine production under fed-batch fermentation, making it a very promising platform for L-lysine production.
In addition, the L-lysine yield of strain XQ-5-W4 stabilized faster, about six hours earlier than that of strain XQ-5. However, the dry weight of strain XQ-5-W4 was 38.9 ± 5.12 g/L, 13.9% lower than that of the original strain (i.e., 45.2 ± 7.64 g/L) (Figure 7a,b). Previous research also found the similar result in which the biomass decreased while enhancing the yield of the target products [23]. Taken together, these results demonstrated that the final strain XQ-5-W4 shows an efficient L-lysine production under fed-batch fermentation, making it a very promising platform for L-lysine production.

Strains, Growth Medium, and Culture Conditions
L-lysine producing strain C. glutamicum XQ-5 derived from the wild-type strain C. glutamicum ATCC 13032 after multiple rounds of random mutagenesis. Strains and plasmids used in this study are listed in Table 4. Oligonucleotides used in this study are listed in Supplementary Material (Table S2). The L-lysine high-producing strain C. glutamicum

Strains, Growth Medium, and Culture Conditions
L-lysine producing strain C. glutamicum XQ-5 derived from the wild-type strain C. glutamicum ATCC 13032 after multiple rounds of random mutagenesis. Strains and plasmids used in this study are listed in Table 4. Oligonucleotides used in this study are listed in Supplementary Material (Table S2). The L-lysine high-producing strain C. glutamicum XQ-5 was derived from the wild-type strain C. glutamicum 13032 [41]. E. coli grew in Luria-Bertani (LB) medium at 37 • C. C. glutamicum grew in LB-glucose (LBG) medium at 30 • C [42]. EPO medium and LB-Brain Heart Infusion-Sorbitol (LBHIS) medium were used to construct the recombinant bacteria [43]. In addition, a 50 µg/mL kanamycin solution was used to build the plasmids and a 25 µg/mL kanamycin solution was used to screen the recombinant strains. A 1 mMol/L IPTG solution was used to induce gene overexpression. Samples were taken from the shake flasks or fermenters every four hours.
Fed-batch fermentation was carried out in a 5-L jar fermenter (BLBio-5GJ-2-H, Bailun Bi-Technology Co. Ltd., Shanghai, China). The fermentation medium contained (per liter): 70 g glucose, 20 g corn steep liquor, 2 g KH 2 SO 4 , 50 g beet molasses, 40 g (NH 4 ) 2 SO 4 , 1.5 g MgSO 4 .7H 2 0, 0.03 g FeSO 4 , 0.02 g MnSO 4 , 0.03 g glycine betaine, 600 ug biotin, 300 ug thiamine-HCI, and 2 mL antifoam. Ammonia was used to control pH 7.0 and provide nitrogen source for bacteria. The relative dissolved oxygen was controlled at 20-30% by stirring speed and ventilation. The temperature is maintained at 30 • C by jacket cooling. OD 600 , residual sugar concentration and L-Lysine concentration were determined every 4 h during fermentation. The prepared feed solution [44] was used to control the glucose concentration at about 5 g/L by adjusting the feeding rate. L-lysine concentration was determined as lysine·HCl in duplicates.

Analytical Methods
The cell concentration after 25-fold dilution was measured at OD 600 using a spectrophotometer. The correlation coefficient between the dry cell weight (DCW) and OD 600 was 0.32 (1 OD 600 = 0.32g DCW). After sample dilution of 100 times, glucose and L-lysine concentrations were measured with an SBA-40E immobilized enzyme biosensor. The concentration of the by-products was measured with high performance liquid chromatography (HPLC) [44]. Cell morphology was observed via field emission scanning electron microscopy (FESEM). Cells of C. glutamicum in the mid-log phase were collected by centrifugation and rinsed three times in physiological saline (pH 7.0). Bacterial cells were spread onto a small silicon platelet and air dried under room temperature, followed by in-situ fixation with a 2.5% glutaraldehyde solution in a 0.15 M sodium phosphate buffer (pH7.4) for 10 min. The samples were coated with gold and transferred to FESEM (SU8220, Hitachi, Japan) for observation at an accelerating voltage of 3 kV.

Construction of C. glutamicum Recombinant Strains
Restriction endonucleases and the DNA Ligase were used to construct the plasmids. In this study, the plasmid pEC-XK99E was used for gene overexpression in C. glutamicum. The suicide plasmid pK18mobsacB was used for gene knockout in C. glutamicum. Firstly, the constructed plasmid was electroporated into C. glutamicum, and then the positive transformants were screened with a 25 µg/mL kanamycin solution in LBH medium. The final positive transformants were obtained by eliminating the plasmids according to the sucrose lethal principle. The deletions in the chromosome were verified by PCR analysis.

Real-Time PCR
In order to analyze RNA, cells in exponential phase were collected during shake flask fermentation for mRNA isolation. RNA was extracted with an RNAiso Plus reagent (Takara, Dalian, China). The cDNA was synthesized with RevertAid TM First Strand cDNA synthesis kit (Fermentas, Shanghai, China). The Ct values of the 16S rDNA gene and those of the ddh and dapD genes were obtained by RT-qPCR using a Bio-Rad CFX96 Touch Real-Time PCR Detection System (Bio-Rad Hercules, CA, USA) with SYBR Premix Ex Taq TM II (Takara, Dalian, China). The primer sequences used for RT-qPCR and RT-PCR are shown in Table S3. Each sample was analyzed in triplicate.

Preparation of Crude Extracts and Enzyme Assays
Crude enzyme solution was prepared to measure the activities of DapDH and DAPD. The preparation method was based on a previous report [45]. Enzyme activity was analyzed in triplicate.
DapD activity was measured by the formation of free coenzyme A (CoA) at 412 nm. The reaction mixture contained 0.1 M Tris-HCI (pH 8.0), 0.5 mM DTNB, 0.2 mM succinyl-CoA, 5 mM 2-aminopimelate, and crude enzymes extract. One unit is defined as the amount of enzyme which catalyzes the formation of 1 µmol CoA per minute [32].

Conclusions
For the first time, DAP pathway was reconstructed to optimize L-lysine production in C. glutamicum, which demonstrated that the dehydrogenase pathway is promising for promoting L-lysine production. In C. glutamicum, both the dehydrogenase pathway and the succinylase pathway are involved in the production of L-lysine, but the relative proportion of each pathway on L-lysine biosynthesis is different because of the different demand for NH 4 + concentration [16]. The proportion of the dehydrogenase pathway on L-lysine production increased when increasing the NH 4 + concentration (Figure 2d). Since the L-lysine biosynthesis in the dehydrogenase pathway has less steps, the strain showed the highest q Lys,max at high NH 4 + concentration (Figure 2c). The similar results were also found in the AmtR-deficient strain (Figure 4b), as NH 4 + was efficiently transferred into the cell during inactivation of AmtR [25]. The L-lysine yield and q Lys,max in strain XQ-5-2 reached 52.3 ± 4.31 g/L and 0.22 ± 0.03 g/(g·h), which were 8.5% and 10% higher than that of the original strain XQ-5, respectively. Although the dehydrogenase pathway is promising in promoting L-lysine production, redirecting the flux into the dehydrogenase pathway while blocking the succinylase pathway is counterproductive to L-lysine production ( Figure 5d). Fortunately, this problem can be overcome by weakening the succinylase pathway (Figure 6b). The target strain C. glutamicum XQ-5-W4 produced 58.5 ± 5.43 g/L L-lysine with a q Lys,max. (i.e., 0.31 ± 0.04 g/(g·h)) in shake-flask fermentation, which were 21.4% and 55% higher than that of strain XQ-5. In addition, fed-batch fermentation of strain XQ-5-W4 resulted in 189 ± 8.7 g/L of L-lysine with a q Lys,max of 0.35 ± 0.05 g/(g·h). These results indicated that the reconstruction of DAP pathway to switch the flux in the variants of DAP pathway has great potential to promote L-lysine production in C. glutamicum.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.