Enhancing the capability of Klebsiella pneumoniae to produce 1, 3‐propanediol by overexpression and regulation through CRISPR‐dCas9

Summary Klebsiella pneumoniae is a common strain of bacterial fermentation to produce 1, 3‐propanediol (1, 3‐PDO). In general, the production of 1, 3‐PDO by wild‐type K. pneumoniae is relatively low. Therefore, a new gene manipulation of K. pneumoniae was developed to improve the production of 1, 3‐PDO by overexpressing in the reduction pathway and attenuating the by‐products in the oxidation pathway. Firstly, dhaB and/or dhaT were overexpressed in the reduction pathway. Considering the cost of IPTG, the constitutive promoter P32 was selected to express the key gene. By comparing K.P. pET28a‐P32‐dhaT with the original strain, the production of 1, 3‐PDO was increased by 19.7%, from 12.97 to 15.53 g l−1 (in a 250 ml shaker flask). Secondly, three lldD and budC regulatory sites were selected in the by‐product pathway, respectively, using the CRISPR‐dCas9 system, and the optimal regulatory sites were selected following the 1, 3‐PDO production. As a result, the 1, 3‐PDO production by K.P. L1‐pRH2521 and K.P. B3‐pRH2521 reached up to 19.16 and 18.74 g l−1, which was increased by 47.7% and 44.5% respectively. Overexpressing dhaT and inhibiting expression of lldD and budC were combined to further enhance the ability of K. pneumoniae to produce 1, 3‐PDO. The 1, 3‐PDO production by K.P. L1‐B3‐PRH2521‐P32‐dhaT reached 57.85 g l−1 in a 7.5 l fermentation tank (with Na+ neutralizer), which is higher than that of the original strain. This is the first time that the 1, 3‐PDO production was improved in K. pneumoniae by overexpressing the key gene and attenuating by‐product synthesis in the CRISPR‐dCas9 system. This study reports an efficient approach to regulate the expression of genes in K. pneumoniae to increase the 1, 3‐PDO production, and such a strategy may be useful to modify other strains to produce valuable chemicals.

Oxidation and reduction reactions are the two metabolic pathways for glycerol as the only carbon source in K. pneumoniae (Cheng et al., 2004). Glycerol dehydrase (GDHt) and 1, 3-propanediol oxidoreductase (PDOR) are the key enzymes in the reduction reaction of K. pneumoniae to produce 1, 3-PDO (Cheng et al., 2006). The gene dhaB encodes GDHt and dhaT encodes PDOR. In the reduction pathway, glycerol is first converted to 3hydroxypropanal (3-HPA) and then reduced to 1, 3-PDO by consuming nicotinamide adenine dinucleotide (NADH) (Chen et al., 2005).
In the oxidative pathway, adenosine triphosphate (ATP) and nicotinamide adenine denucleotide (NADH), which are required for cell growth, are produced, along with a series of by-products, such as acetate, ethanol, lactate, 2, 3-butanediol (2, 3-BDO) and so on (Kumar and Park, 2018). This research focuses on the regulation of lactate and 2, 3-BDO to ultimately improve the production of 1, 3-PDO under the previous study (Wang et al., 1800). ldhA and lldD encode L-lactate dehydrogenase (Aguilera et al., 2008;Fu et al., 2016), while budA, budB and budC are genes in the entire butanediol (BDO) pathway (Kumar et al., 2016). Knockout of genes in the lactate pathway and the entire BDO pathways ultimately increases the production of 1, 3-PDO. However, completely deleting budO (whole-bud operon) may result in heavy carbon metabolic traffic at pyruvate nodes, thereby inhibiting GDHt (Kumar et al., 2016). And deletion of ldhA could not have a significant improvement in the production of 1, 3-PDO . Therefore, in this study, the expression levels of lactate dehydrogenase and 2, 3-BDO dehydrogenase were reduced by disturbance attenuation, which could not only reduce the consumption of NADH, but also minimize the generation of by-products, and eventually improve the 1, 3-PDO production.
In this research, the 1, 3-PDO production by K. pneumoniae was investigated by means of overexpression and regulation (as shown in Fig. 1). On the one hand, the plasmids pET28a-P32-dhaB, pET28a-P32-dhaT and pET28a-P32-dhaB-dhaT were constructed, and the influence of overexpressing dhaB and/or dhaT on the 1, 3-PDO production was studied. On the other hand, plasmids L1~L3-pRH2521, B1~B3-pRH2521 and L1-B3-pRH2521 were constructed to explore the effect of lldD and budC on enhancing the production of 1, 3-PDO in the oxidation pathway. Finally, the plasmid L1-B3-pRH2521-P32-dhaT was constructed, and the 1, 3-PDO production by K.P. L1-B3-pRH2521-P32-dhaT was increased by 26.1% compared with the original strain. These results will provide ideas on how to improve the 1, 3-PDO production by K. pneumoniae and also give new insights into the genetic modification of other strains.

Results
Overexpressing the dhaB and/or dhaT in the reduction pathway In the process of glycerol metabolism, GDHt is the rate limiting enzyme, which converts glycerol to 3-HPA, and then is catalysed by PDOR to 1, 3-PDO (Sankaranarayanan et al., 2017). Therefore, some plasmids without the isopropyl-beta-D-thiogalactopyranoside (IPTG) as promoter was constructed to transfer into K. pneumoniae to study the production of 1, 3-PDO. Figure 2A is a comparison of the 1, 3-PDO productions by K. pneumoniae ATCC 15380, K.P. pET28a-P32-dhaB, K.P. pET28a-P32-dhaT and K.P. pET28a-P32-dhaB-dhaT in 250 ml shaker. The 1, 3-PDO production was 12.97, 8.35, 15.52 and 12.25 g l À1 respectively. In contrast, only overexpressing dhaB was found to reduce the 1,3-PDO production. In this case, the plasmid pET28a-P32-dhaT was discovered to be the most effective. In order to further study the effect of K. pneumoniae ATCC 15380, K.P. pET28a-P32-dhaT and K.P. pET28a-P32-dhaB-dhaT on the field of 1, 3-PDO, experiments were conducted in a 7.5 L fermenter (Fig. 2B). The 1, 3-PDO production by K.P. pET28a-P32-dhaT was found to be the highest at 47.71 g l À1 , which was only slightly higher than that of the original strain (45.86 g l À1 ). During the entire fermentation process, the concentration of the glycerol as substrate was maintained at 20 g l À1 . The overlapped chromatograms of glycerol and 1, 3-PDO during fermentation, including the structures are shown in Fig. 2C. The characteristic peaks at~15.25 and~20.23 min represented the glycerol and 1, 3-PDO respectively.
Attenuating the expression of the lldD and budC in the oxidative pathway According to off-target effects and specific quality, three sgRNA were designed for lldD and budC respectively.
To further compare the 1, 3-PDO production after regulating lldD and budC, a fermentation experiment was carried out in a 7.5 l fermenter. According to Fig. 3C, it can be found that the 1, 3-PDO production by B3, L1 and L1-B3 are 50.04, 48.54 and 52.85 g l À1 respectively. The 1, 3-PDO production of L1-B3 was increased by 15.2% compared with the original strain. Overexpressing dhaT, or inhibiting the expression of budC and lldD, respectively, did not significantly increase the production of 1, 3-PDO by K. pneumoniae. Therefore, simultaneous overexpression and inhibition of K. pneumoniae should be considered to improve the ability of K. pneumoniae to produce 1, 3-PDO. Figures 2D and 3C show the overlapped chromatograms of lactate and 2, 3-BDO in zymotic fluid. The peaks located at~8.43 and~21.11 min could attach to the lactate and 2, 3-BDO. Through the spectrum, the concentration of lactate and 2, 3-BDO could be obtained.

Combining regulation and overexpression
If gene dhaT was overexpressed or lldD and budC were regulated separately, the increase of 1, 3-PDO production was limited. In order to further improve the production of 1, 3-PDO, it is necessary to overexpress dhaT in the oxidation pathway and simultaneously attenuate lldD and budC in the reduction pathway. By comparing the Compare the qRT-PCR of 2, 3-BDO and lactate respectively in the 250 ml shaker for the K. pneumoniae ATCC15380, B1, B2, B3, L1, L2 and L3. C. The comparison of the yield of 1, 3-PDO in the 7.5 L fermentation tank for the K. pneumoniae ATCC15380, B3, L1 and B3+L1 (K.P. B3-L1-pRH2521). (Solid represents 1, 3-PDO and hollow represents glycerol.) D. The overlapped chromatograms of lactate in zymotic fluid, including the structures. The data represent the mean values of three independent biological replicates, and the error bars represent the standard deviations. production of 1, 3-PDO in Fig. 4A, it is found that the 1, 3-PDO production by K.P. L1-B3-pRH2521-P32-dhaT reaches the highest, with the production of 57.85 g l À1 , which is 26.1% higher than that of the original strain, confirming that the simultaneous overexpression and inhibition of K. pneumoniae by CRISPR-dCas9 can improve the ability of producing 1, 3-PDO. Figure 4B shows the cell growth (OD 600 ) of the original strain, K.P. pET28a-P32-dhaT, K.P. B3-L1-pRH2521 and K.P. B3-L1-pRH2521-P32-dhaT in 7.5 L fermenter at 40 h, which is positively correlated with the production of 1, 3-PDO. The higher the 1, 3-PDO production, the greater the OD 600 of the strain. Among them, the 1, 3-PDO production by K.P. B3-L1-pRH2521-P32-dhaT reached the highest OD 600 (6.02), which was a 31.15% increase compared with the OD 600 value of the original strain (4.59).

Discussion
As reported (Zhang and Xiu, 2009), glycerol disproportionation involves two parallel pathways: reduction pathway and oxidation pathway in the metabolic process of producing 1, 3-PDO by K. pneumoniae ( Fig. 1 and Table 1). The reaction of each pathway can be seen from Table 1. The reduction pathway consists of two enzymes: the first enzyme DhaB removes a water molecule from glycerol to produce 3-HPA; The second enzyme DhaT reduces 3-HPA to 1, 3-PDO. In the oxidation pathway, glycerol generates dihydroxyacetone (DHA) by the glycerol dehydrogenase. Then dihydroxyacetone phosphate (DHAP) is generated by the dihydroxyacetone kinase, and the pyruvate is generated through a series of enzyme reactions, accompanied by some by-products (e.g. lactate, 2, 3-BDO, acetate, ethanol, isobutanol and so on). These by-products inhibit the glycerol transfer into 1, 3-PDO, which significantly reduces the yield of 1, 3-PDO. In addition, they are toxic to cell growth at high concentrations. In the lactate synthetic pathway, pyruvate is catalysed to lactate by Llactate dehydrogenase (lldD) (Aguilera et al., 2008;Fu et al., 2016). In the BDO synthetic pathway, (S)-2acetolactate is generated by pyruvate under the catalysis of acetolactate synthase. In the next step, (S)-2acetolactate is converted to acetoin under the catalysis of acetolactate decarboxylase. Finally, NADH is used as the reductant to reduce the compound to 2,3-BDO through butanediol dehydrogenase (budC) (Blomqvist et al., 1993;Wood et al., 2005;Celinska and Grajek, 2009). In this study, the synthesis of lactate and BDO was investigated. Therefore, to improve the 1, 3-PDO production by K. pneumoniae, overexpressing dhaT in the reduction pathway followed by attenuating the synthetic pathway of 2, 3-BDO and lactate in the by-product pathway were required in this work. Since pET28a requires induction with IPTG to start, the development of 1, 3-PDO industrial production processes may be discouraged by expensive IPTG. Therefore, the subsequent experiments in this study were conducted without adding IPTG. The T7 promoter in pET28a was replaced by P32, hence reducing the cost of the whole process. At the same time, according to literature, P32 promoter is conducive to improving production (Minami and Shah, 2021). The fermentation experiment showed that K.P. pET28a-P32-dhaT was the best strain to produce 1,3-PDO by overexpressing dhaT. This result may be attributed to the accumulation of 3-HPA, an intermediate metabolite that is toxic to the strain, and detrimental to cell growth and reduces the 1, 3-PDO production. Therefore, the higher expression of dhaB affected the 1, 3-PDO production, which was consistent with the report of Zhao et al. (2009aZhao et al. ( ,2009b and Oh (2013). If dhaT was overexpressed alone, the 1, 3-PDO production by the modified strain was only slightly higher than that of the original strain. This may be due to the fact that NADH needs to be consumed during the transformation of 3-HPA into 1, 3-PDO in the reduction pathway, and the overexpression of dhaT will lead to insufficient NADH in the reduction reaction. Thus, it is necessary to regulate the expression of some genes from the by-product pathway, such as lldD and budC to reduce the amount of NADH required to produce by-products, thus further increasing the production of 1, 3-PDO.
The knockout methods were adopted in traditional methods (Kumar et al., 2016;Li et al., 2016), while the increase of 1, 3-PDO production by knocking out of the lactate synthesis pathway was not remarkable. Direct knockout of genes in the by-product pathway may lead to an increase in other by-products. Additionally, completely inactivating all of the genes involved in the formation of by-products is unreasonable due to the control of central carbon metabolic fluxes (Bro et al., 2006;Celinska, 2010). In fact, the diversity of by-products reflects the delicate redox equilibrium of cells, so it is important to regulate the formation of by-product while maintaining the overall redox equilibrium to improve glycerol utilization and 1, 3-PDO yield. Therefore, it is necessary to consider the use of the regulatory method to reduce the generation of by-products lactate and 2, 3-BDO. Moreover, the regulatory method may reduce the consumption of NADH in the synthesis of lactate and 2, 3-BDO, which will supply more NADH to DhaT in the reduction pathway, and subsequently enhance the production of 1, 3-PDO. lldD and budC are mainly involved in the oxidation pathway in which the by-products (lactate and 2, 3-BDO) during the synthesis of 1, 3-PDO are NADH-dependent. Thus, regulating the competitive NADH pathway can increase the concentration of NADH in the cell and reduce the concentration of the by-product, providing NADH to the reduction pathway to ultimately improve the production of 1, 3-PDO. Weakening expression of budC is beneficial to reduce the production of 2, 3-BDO and affect the cell metabolism. By comparing the expression levels of lldD in L1, L2 and L3, and budC among B1, B2 and B3, it was found that L1 and B3 had the lowest expression levels, indicating that the regulatory effects of these two sgRNAs were the best. As a result, the combination of L1 and B3 might perform better. The results showed that the production of 1, 3-PDO by K.P. L1-B3-pRH2521 was higher than that of the original strain. Compared with the complete deletion of lactate and the whole 2, 3-BDO pathway would not inhibit GDHt because of the massive carbon metabolism of pyruvate nodes, thus affecting the production of 1, 3-PDO (Kumar et al., 2016).
By combining overexpression and regulation with CRISPR-dCas9, it could be found that the 1, 3-PDO production in the presence of K.P. L1-B3-pRH2521-P32-dhaT was the highest (57.85 g l À1 ), suggesting that the inhibiting the expression of lldD and budC in pathway of lactate and 2, 3-BDO might reduce the consumption of NADH and enhance the conversion of 3-HPA to 1, 3-PDO.
Simultaneous multiple gene modification is more beneficial to 1,3-PDO production than traditional single-gene modification methods (e.g., single overexpression (Zheng et al., 2006;Hao et al., 2008;Zhao et al., 2009aZhao et al., ,2009bMa et al., 2010;Zhu et al., 2015), knockout (Horng et al., 2010;Guo et al., 2013), and regulation (Wei et al., 2014;Hirokawa et al., 2017;Lee et al., 2018Lee et al., , 2019. In summary, this is the first time the dhaT has been overexpressed in K. pneumoniae, and the expression of lldD and budC was simultaneously inhibited to enhance the performance of the strain in the CRISPR-dCas9 system. This method is relatively simple and efficient and only requires to transfer two plasmids into bacteria simultaneously to achieve the effects of the overexpression and attenuation. Compared with the original (45.86 g l À1 ), the production of 1, 3-PDO by K.P. L1-B3-pRH2521-P32-dhaT was successfully increased up to 57.85 g l À1 . In K. pneumoniae, the overexpression and inhibition of the gene expression not only does not affect the NADH balance in the metabolic process, but also enhancing the final 1, 3-PDO production. Therefore, this method may have certain advantages over the NADH imbalance in metabolic process compared with the traditional knockout, and finally decreases the yield of target products. There are many genes in K. pneumoniae that need to be further studied, such as dhaD, dhaK, ldhA, budA, ldhA, poxB, pta, ackA, gltA, arcA and so on. These genes can be overexpressed and regulated in combination to promote the production of 1, 3-PDO by CRISPR-dCas9. These findings provide a reference for the direction of the genetic modification for the target products to promote the productive ability of other strains.

Experimental procedures
Strains, microorganisms and cultivations K. pneumoniae ATCC 15380 was purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). The microorganisms and cultivations were provided in the Appendix S1.

Analytical methods and instrumentation
The concentrations of 1, 3-PDO, glycerol and 2, 3-BDO were measured with an HPX-87H column (300 9 7.8 mm 2 ) (Bio-Rad, Palo AHO, CA, USA) with a differential refractive index detector (SFD GmbH, Schambeck, Germany) (as shown in Fig. 2C). 5 mM aqueous H 2 SO 4 solution was used as a mobile phase with a flow rate of 0.5 ml min À1 at a working temperature at 65°C.
The concentration of lactate was measured with an MP C18 column (250 mm 9 4.6 mm, 5 µm; Agilent) by a liquid chromatography on a 1100 series instrument (Agilent, Santa Clara, CA, USA) (as shown in Fig. 3D). 5 mM H 2 SO 4 (with 5% acetonitrile) solution was used as a mobile phase with a flow rate of 0.5 ml min À1 and a working temperature at 35°C.

Plasmids, primers and sequences
The plasmids used in this study are listed in Table 2. The primers used for gene deletion, complementation and overexpression are shown in Table 3. The sequences of the devices and gene fragments are shown in Table 4.

Transformation and screening
Take plasmid pRH2521 as an example. The constructed plasmid pRH2521 was transformed into E. coli DH5a. Colonies were selected on LB plates containing hygromycin (50 µg ml À1 ). PCR and sequencing were used to verify the accuracy of plasmid. Finally, plasmid pRH2521 was extracted from E. coli DH5a using the plasmid extraction kit (Biomed). Then, the extracted pRH2521 and pRH2502 (kanamycin resistant to dCas9 expression) were electrically transferred into K. pneumoniae, respectively, and colonies were selected on LB plates containing hygromycin (50 µg ml À1 ) and kanamycin (25 µg ml À1 ). Additionally, the plasmids were transferred to K. pneumoniae by electroporation to verify the accuracy of plasmids. Then, the plates were incubated at 37°C for 12-24 h. The stability of the plasmid was the ratio of the percentage of colonies on the antibiotic agar plates over those on the plates without antibiotics. To induce sgRNA and dCas9 expression, anhydrotetracycline was added to the cultures to achieve a final concentration of 200 ng ml À1 at 6 h (Jha et al., 2020). The plasmid pET28a (kanamycin 50 µg ml À1 ) was also constructed as described above. The recombinant plasmids were transformed into E. coli to obtain transformants and transformed into K. pneumoniae by electroporation.

Quantitative Real-Time PCR (qRT-PCR) analysis
The transcriptional levels of budC and lldD in the K.P. ATCC15380, L1~L3 and B1~B3 were measured by qRT-PCR. The total RNA of the bacteria was extracted by using the RNA prepPure culture bacterial total RNA extraction kit (Tiangen) according to the operation instructions, and treated with DNase I to remove the residual DNA fragments. The RNA samples were quantified by spectrophotometer at 260 and 280 nm, and purified by 1% formaldehyde agarose gel electrophoresis. cDNA synthesis was completed by using the fast quant cDNA first strand synthesis kit (Tiangen) according to the operation instructions. The PCR reaction was designed according to the K. pneumoniae NCTC 418 genomic sequence at Gen-Bank in NCBI database. The synthetic primers of the target gene were designed by primer design software (Primer Premier 5.0). qRT-PCR was performed with the Real Master Mix (SYBRGreen) kit (Tiangen) according to the operation instructions using the K. pneumoniae 16S rRNA gene as the internal standard control.
Relative quantitation 2 -DDCT method was used to analyse the qRT-PCR data. To calculate the C T value obtained from the experiment, and the relative differences of gene transcription were obtained (Wang et al., 2021a(Wang et al., ,2021b.