Development of a Red recombinase system and antisense RNA technology in Klebsiella pneumoniae for the production of chemicals

Abstract

Klebsiella pneumoniae is a promising industrial species, however the lack of genetic engineering tools restricts its applications. Here we developed a lambda (λ) Red recombinase system and antisense RNA technology in K. pneumoniae to reshape glycerol metabolism pathways. We deleted the lactate dehydrogenase gene ldh through RecA-dependent recombination to block lactic acid synthesis. Next, the 1,3-propanediol dehydrogenase gene dhaT was replaced by an aldehyde dehydrogenase gene (aldH from E. coli) to repress 1,3-propanediol (1,3-PDO) synthesis and simultaneously convert 3-hydroxypropionaldehyde (3-HPA) to 3-hydroxypropionic acid (3-HP). Specially, we developed a Red recombinase system in K. pneumoniae, by which the enzymes related to glycerol metabolism were mutated by transformed oligos. One positive strain produced 6.39 g L⁻¹ 3-HP and 32.6 g L⁻¹ 1,3-PDO at 36 h without using any antibiotics and inducers. Sequencing results showed that the mutation occurred mainly in byproduct pathways. Finally, antisense RNA technique was applied to block the synthesis of lactic acid and acetic acid. We found that the increase of 3-HP was approximately proportional to the decrease of lactic acid and acetic acid, indicating their competition for glycerol carbon flux. Overall these results and approaches developed in this study provide basis for basic research and microbial production of 3-HP, 1,3-PDO and 2,3-butanediol in K. pneumoniae.

---

1. Introduction

Microbial production of chemicals is an alternative to chemical synthesis. Currently, Klebsiella pneumoniae is considered as a promising industrial species because it can naturally synthesize a range of economically important bulk chemicals, such as 1,3-propanediol (1,3-PDO), 2,3-butanediol (2,3-BDO) and 3-hydroxypropionic acid (3-HP). In addition, K. pneumoniae demonstrates striking biological attributes, including aggressive cell growth, a powerful ability to metabolize glycerol, and particularly the capacity to synthesize vitamin B₁₂ (ref. 1 and 2) which serves as the cofactor of glycerol dehydratase (GDHt), a key enzyme for glycerol metabolism. In K. pneumoniae, glycerol metabolism is mediated by the dha regulon which steers glycerol oxidation and reduction pathways (ESI Fig. 1†).³ Under anaerobic or micro-aerobic conditions, GDHt (encoded by dhaB cluster, GenBank No. U30903) converts glycerol to 3-hydroxypropionaldehyde (3-HPA). 3-HPA is a toxic intermediate and is catalyzed to 1,3-PDO by 1,3-propanediol dehydrogenase (PDOR, encoded by dhaT). 3-HPA can also be converted to 3-HP by aldehyde dehydrogenase (ALDH) with NAD⁺ as a cofactor.⁴ Previous endeavors on glycerol-based biosynthesis of 3-HP mainly focused on ALDH overexpression and optimization of fermentation conditions.^5,6 Although ALDH overexpression diverted glycerol carbon flux to 3-HP and approximately 50 g L⁻¹ 3-HP was generated in K. pneumoniae,⁶ further enhancing the yield of 3-HP faces obstacles. One pivotal obstacle is the lack of genetic engineering tools that can sculpt K. pneumoniae genome.

Previously, RecA and λ Red-mediated recombination techniques were applied to delete or modify the genes in K. pneumoniae,^7,8 however the efficiency was low because vector construction was limited to serial manipulation of single gene and thus was laborious and time-consuming. In particular, those techniques require multiple rounds of introducing antibiotic marker into K. pneumoniae genome because the thick capsule lowers the bacterial sensitivity to antibiotics.⁹ As a consequence, λ Red recombinase system ceases to work after several rounds of recombination because bacteria generate resistance to recombinases. To address this problem, efficient genome engineering methods are highly desirable.

Synthetic biology offers ways to reallocate metabolic flux and thereby raises the possibility of overproducing desired metabolites. Some genome engineering techniques such as transcription activator-like effector nucleases (TALEN) and CRISPR-Cas9 have been used to modify the genomes of E. coli and other model organisms.^10–13 In particular, the Red recombination-dependent multiplex automated genome engineering (MAGE) seems applicable for genome engineering of bacteria. This approach was shown to work efficiently in E. coli. For example, 5-fold increase of lycopene production in E. coli was achieved within only three days because multiple genomic locations were simultaneously altered. More attractively, the genes conferring novel phenotypes can be easily recognized because the known oligos serve as labels. It is clear that MAGE technology holds promise to overproduce desired chemicals by strategy of inverse metabolic engineering.^14–16 Since K. pneumoniae has similar genetic background with E. coli, we hypothesized that Red recombinase system should work in K. pneumoniae.

In view of above information, in this study we developed λ red recombinant system in K. pneumoniae to reallocate glycerol flux towards high production of C3 compounds including 1,3-PDO, 2,3-BDO and 3-HP. To do so, lactic acid synthesis gene ldh was deleted through RecA homologous recombination. After λ Red recombinases were expressed in K. pneumoniae, the gene dhaT was replaced by gene aldH from E. coli to direct 1,3-PDO carbon flux to 3-HP. To modify glycerol metabolism and enhance 3-HP production, chemically synthesized oligos that carry wrong nucleotides were electro-transformed into K. pneumoniae. Through screening of positive clones, the highest yield strain was fermented for production of 3-HP and 1,3-PDO. Furthermore, antisense mRNAs were harnessed to clarify the relationship between byproducts formation and 3-HP production. The methods developed in this study may facilitate metabolic engineering of K. pneumoniae.

2. Materials and methods

2.1. Strains, plasmids and chemicals

Strains of E. coli Top10 and K12 were purchased from China General Microbiological Culture Collection Center (CGMCC). Strain of K. pneumoniae DSM 2026 was purchased from DSMZ GmbH, Germany. The vectors pET-28a and pUC19 were products of Novagen and New England Biolabs, respectively. The plasmid pKD46 was kindly provided by Professor Tianwei Tan from Beijing University of Chemical Technology. Primer synthesis and DNA sequencing were performed by Biomed Co., Ltd. The chemicals for enzymatic activity assay and HPLC analysis were products of Sigma. All primers and oligos used in this study were listed in ESI Table 1.†

2.2. Construction of the recombinants

To delete ldh gene (KPN_03949), its upstream and downstream homologous arms of 1000 bp each were cloned into vector pET-28a, resulting in a vector named pET-ldhUD (‘ldhUD’ indicates the homologous arms) (ESI Fig. 2†). After the vector pET-ldhUD was transformed into K. pneumoniae and cultivated in LB plate containing 50 μg mL⁻¹ kanamycin, a recombinant strain K. pneumoniaeΔldh (pET-ldhUD) was obtained. This strain was further confirmed by colony PCR and sequencing. To dispel the vector pET-ldhUD out of cell, the strain K. pneumoniaeΔldh (pET-ldhUD) was grown in LB medium containing 50 mM CaCl₂ for 3 d. The strain unable to survive in LB kanamycin plate was considered as the strain devoid of vector pET-ldhUD and named K. pneumoniaeΔldh.

To introduce λ Red recombination system into K. pneumoniaeΔldh, four fragments from vector pKD46 including three λ Red recombinases (exo, bet, gam) and one arabinose-inducible promoter were PCR amplified and subcloned into the BglII/BamHI sites on vector pET-28a. The resultant vector was named pET-Red and subsequently transformed into K. pneumoniaeΔldh, leading to recombinant strain K. pneumoniaeΔldh (pET-Red) which would express λ Red recombinases upon induction by arabinose (40 mM).

Relying on this Red recombination system, we knocked out dhaT gene and simultaneously knocked in the aldH gene (Accession No. NC_000913.3) from E. coli K12. That is, the dhaT gene was replaced by aldH gene to block 1,3-PDO synthesis and simultaneously overproduce 3-HP (ESI Fig. 2B†). Briefly, a fragment containing aldH gene, chloramphenicol resistance gene cm from pCP20, and the dhaT gene homologous arms of 40 bp each for flanking aldH and cm, were overlapped, digested by DpnI and cloned into vector pET28a, giving rise to a vector named pRED-aldH (ESI Fig. 2B†). The vector pRED-aldH was transformed into K. pneumoniaeΔldh and grown in LB medium containing kanamycin (50 μg mL⁻¹) and L-arabinose (34 μg mL⁻¹) for inducing Red recombinases. Consequently, we acquired a recombinant strain named K. pneumoniaeΔldhΔdhaT (pRED-aldH) which was further confirmed by colony PCR and sequencing. The following are PCR procedures: 94 °C for 4 min; followed by 30 cycles of 94 °C for 1 min, 55 °C for 45 s, 72 °C for X min; 72 °C for 10 min, 16 °C hold, where X depends on the length of amplified gene. All other molecular manipulation followed standard protocol.

2.3. Modifying K. pneumoniae genome through chemically synthetic oligos

Oligos were designed in view of the enzymes related to glycerol metabolism, including those participating in signal transduction, quorum sensing, acid tolerance, and the formation of byproducts such as porins and lipopolysaccharide^17,18 (ESI Fig. 1†). Each chemically synthesized oligo was 59 bp in length including internal 5 bp wrong sequence and two homologous arms of 27 bp each at the ends (ESI Fig. 2A; Table 1†). The oligos were diluted to 50 mM and 5 μL oligos were mixed with 100 μL competent K. pneumoniae. The mixture was subjected to electric shock at 2.5 kV and incubated in 20 mL fresh LB medium for 3 h recovery. After 4 rounds of this process, the cell pellet was plated onto LB plate. The aggressive colonies were individually cultivated in 4 mL fermentation medium (Wang et al., 2013 Curr Microbiol). The high-yield strain was subjected to next round of electro-transformation and screening, and the highest yield strain was investigated for the mutation sites. The recombinant K. pneumoniae strain was grown in medium till OD₆₀₀ of 0.2. After incubation at 4 °C for 30 min, cells were washed with cold ultrapure water, and mixed with oligos.

2.4. Enzyme activity assay

Strains were grown in 50 mL shake flasks each containing 20 mL medium, and rotated at 140g for 12 h. To measure ALDH activity, 10 mL cells were centrifuged at 10 000g for 10 min and washed by 5 mL phosphate buffered saline (PBS) (pH 7.0). Phenylmethanesulfonyl fluoride (PMSF) (10 mg mL⁻¹) was used to inhibit protease activity. The cells were sonicated and the resulting solution was centrifuged at 16 000g for 15 min. Protein concentration was determined using Bradford reagent (BioRad). 200 μL cell-free fermentation broth was added into 5 mL Bradford solution. After 5 min reaction, absorbance was measured by spectrophotometer at 595 nm wavelength. Another 200 μL cell-free solution was added into 2 mL centrifugation tube containing 50 mM PBS solution, 10 mM propionaldehyde and 0.2 mM NAD⁺. The mixture was incubated at 37 °C for 5 min. The amount of NADH was determined by measuring the increase of absorbance at 340 nm wavelength in a spectrophotometer. One unit of ALDH activity was defined as the amount of enzyme used to generate 1 μmol NADH per minute. PDOR activity assay followed the method for ALDH. One unit of enzyme activity was defined as the amount of enzyme used to produce 1 μmol NAD⁺ in one minute. For GDHt activity assay, 200 μL cell free extract was added into the mixture that contained 50 mM PBS solution, 30 mM glycerol, 15 μmol coenzyme B₁₂ and 5 mM tryptophan. After 5 min reaction, the absorption was measured at 560 nm. One unit of GDHt activity was defined as the amount of enzyme required to generate 1 mmol 3-HPA per min.

2.5. Shake flask and bioreactor cultivation of the recombinants

The recombinant strains were grown in LB medium containing the following components per liter: 5 g yeast extract, 10 g NaCl, 10 g peptone, and 50 mg kanamycin. 1% of overnight culture was inoculated to the medium without antibiotics. To maintain microaerobic conditions, flasks were plugged with a cotton stopper and incubated in an orbital incubator shaker at 170g and 37 °C. Samples were collected every 3 h to measure biomass, residual glycerol and 3-HP concentration.

Fed-batch cultivation of the strains was carried out in a 5 L bioreactor (Baoxing, China) containing 3 L fermentation medium. The experiments were conducted in glycerol fed-batch mode, and the fermentation conditions were according to the previously reported method.⁶ The strain was pre-cultivated in 100 mL fermentation medium overnight at 37 °C and then added into the bioreactor. The agitation speed was 400g and the air was supplied at 1.5 vvm. Alternatively, the agitation speed was set at 300g and 200g when air was supplied at 1.0 vvm and 0.5 vvm, respectively. The temperature was 37 °C and pH value was maintained at 7.0 by addition of 5 M NaOH. Residual glycerol was maintained at 25 g L⁻¹. Dissolved oxygen (DO) was monitored automatically. Samples were taken out every 3 h to examine cell concentration, residual glycerol, and metabolites. During the entire fermentation process, no antibiotic was needed because no vectors were included in recombinant strains.

2.6. Unraveling the relationship between 3-HP production and byproducts formation

To clarify the relationship between 3-HP production and byproducts formation, antisense RNA technique was employed because antisense mRNA represses gene expression at post-transcription level.^19–21 To do so, antisense mRNA was generated by cloning target genes inversely into vector pET-pk.²² The antisense mRNA would be reverse-complemented to mRNA. As a result, an mRNA dimer was formed and the translation was thus blocked. Since there exist a total of four lactic acid synthesis genes in K. pneumoniae genome, the conserved DNA fragment of 300 bp was PCR amplified and used for generating mRNA dimer. Although there have two acetic acid synthesis pathways, only pyruvate dehydrogenase was chosen to be blocked by antisense RNA because this enzyme directly catalyzes pyruvate to acetic acid, while other acetic acid synthesis pathways involve multi-enzyme reactions. The 300 bp of pyruvate dehydrogenase coding gene poxB was also reversely cloned into vector pET-pk. The two antisense mRNA modules for repressing lactic acid and acetic acid synthesis genes, together with the aldH gene from E. coli K12 were overlapped and cloned into vector for catalyzing 3-HPA to 3-HP.

2.7. Analytical methods

Cell concentrations were measured by using microplate reader at 600 nm with 200 μL fermentation broth added in a cuvette. The metabolites 3-HP, lactic acid and acetic acid were determined by high performance liquid chromatography (HPLC) system (Shimadzu, Kyoto, Japan) equipped with a C₁₈ column and a SPD-20A UV detector at 210 nm. The column temperature was 25 °C, and mobile phase was 0.05% phosphoric acid at 0.8 mL min⁻¹. 1,3-PD was quantitatively analyzed by HPLC (Shimazu, Japan) equipped with a column of Aminex HPX-87H Ion Exclusion particles (300 mm × 7.8 mm, Bio-Rad, Hercules, CA, USA) using a differential refractive index detector. The column was maintained at 65 °C and mobile phase was 5 mM sulfuric acid (in Milli-Q water) at 0.6 mL min⁻¹. Residual glycerol concentration was measured every 3 h by a titration method with NaIO₄ (for control of glycerol). All samples were filtered through 0.22 μm membrane filter.

3. Results and discussion

3.1. Characterization of the recombinants

For biosynthesis of 3-HP in K. pneumoniae, lactic acid is one of major byproducts.²³ Lactic acid not only competes with 3-HP for glycerol carbon flux, it also entangles downstream separation. In fact, it is extremely difficult to separate lactic acid from 3-HP because they are isomers. To address this problem, in this study we knocked out the lactic acid synthesis gene through RecA recombination system. Restriction enzyme digestion and DNA sequencing showed that the donor vector pET-ldhUD was correctly constructed. Colony PCR confirmed the ldh mutant strain named K. pneumoniaeΔldh (pET-ldhUD). To dispel the vector pET-ldhUD out of cells, K. pneumoniaeΔldh (pET-ldhUD) was grown in medium containing CaCl₂ but devoid of kanamycin for 3 days. The strains unable to survive in LB kanamycin plate were considered as the ldh mutant strain K. pneumoniaeΔldh. To develop Red recombinase system in K. pneumoniaeΔldh, Red recombinase vector pET-Red was transformed into K. pneumoniaeΔldh, giving rise to recombinant strain K. pneumoniaeΔldh (pET-Red). To block 1,3-PDO synthesis and simultaneously overproduce 3-HP, the dhaT gene was replaced by aldH gene through Red recombinase system. The resulting strain was named K. pneumoniaeΔldhΔdhaT (pET-RED). Colony PCR and sequencing confirmed this recombinant strain.

3.2. Reshaping glycerol pathways with oligos and enzyme activity assays

After iterative transformation of oligos into K. pneumoniaeΔldhΔdhaT (pET-RED), a total of 201 positive strains were generated. We found that only 10% strains produced 3-HP up to 0.6 g L⁻¹. Of these strains, a high yield strain named Kp3 produced more lactic acid compared with other strains (Fig. 1). This might be ascribed to anaerobic conditions that facilitated lactic acid synthesis. In fact, biosynthesis of lactic acid consumes NADH and simultaneously generates NAD⁺ which serves as the cofactor of ALDH and thus benefits 3-HP production.²⁴ Another reason for high production of 3-HP might be energy generation under anaerobic conditions. The formation of lactic acid indicated the generation of ATP from EMP pathway, which might benefit 3-HP production. Collectively, there exists an interdependence between 3-HP production, cofactor recycling and energy provision.

Fig. 1 Three-dimensional plot demonstrating metabolite concentrations of strains in shake flask cultivation. Each point represents one strain and the corresponding concentrations of 3-hydroxypropionic acid, lactic acid and acetic acid.

To further investigate the outcome of oligo-mediated genomic modification, we measured the activities of three key enzymes in K. pneumoniaeΔldhΔdhaT, which was named Kp3. The cell-free extract showed enhanced GDHt, ALDH and PDOR activities compared with the reference strain K. pneumoniae (Table 1). For the strain Kpac, the replacement of dhaT by aldH resulted in enhanced ALDH and GDHt activities, however, the PDOR activity significantly declined due to the deletion of dhaT gene. Compared with the strain Kpac, the oligo-directed mutant strain Kp3 showed slight decrease in the activities of GDHt and ALDH, however the PDOR activity increased. Since YqhD (KPN_03431) is an isoenzyme of PDOR, the enhancement of PDOR activity in strain Kp3 was presumably due to the expression of yqhD. Overall, oligo-directed genome modification led to enhanced activities of key enzymes.

Table 1 The ALDH, GDHt and PDOR enzyme activities of strains^a

Strains	K. pneumoniae	Kpac	Kp3
a ALDH: aldehyde dehydrogenase; GDHt: glycerol dehydratase; PDOR: 1,3-propanediol dehydrogenase; Kpac: the recombinant strain K. pneumoniae (pRED-aldH)ΔdhaTΔldh; Kp3: the oligo-mediated mutant strain K. pneumoniae (pRED-aldH)ΔdhaTΔldh. One asterisk (*) indicates P value < 0.05, while two asterisks (**) indicate P value < 0.01. The experiments were performed in triplicate.
ALDH	3.20 ± 0.10	4.04 ± 0.04**	3.83 ± 0.08*
GDHt	7.28 ± 0.17	12.96 ± 0.27**	11.17 ± 0.16**
PDOR	3.24 ± 0.19	1.79 ± 0.07**	3.94 ± 0.15**

---

3.3. Sequencing result and analysis

To determine whether Red recombinase system worked in K. pneumoniae, the oligo-mediated targeted genes were sequenced. Results showed that pyruvate dehydrogenase gene poxB (KPN_00904) and lactate dehydrogenase gene ldh (KPN_01632) were mutated by oligos, indicating that this Red recombination system was successfully developed in K. pneumoniae and the genome could be further sculpted towards overproduction of C₃ compounds including 3-HP, 1,3-PDO and 2,3-BDO. Unlike E. coli, K. pneumoniae naturally synthesizes coenzyme B₁₂, which is the cofactor of GDHt, a key enzyme for glycerol utilization. In addition, K. pneumoniae shows remarkable capacity to metabolize glycerol. Early studies showed that nearly all glycerol was consumed, and lactic acid and acetic acid accounted for a substantial portion of glycerol flux.⁶ In this study, although ldh and poxB genes were deleted, lactic acid and acetic acid were still produced. This is because of multiple synthesis pathways of lactic acid and acetic acid. For instance, there are four L-lactate dehydrogenases and two D-lactate dehydrogenases genes in K. pneumoniae. In fact, the production of lactic acid and acetic acid is responsible for energy provision and cofactor balance.

3.4. Shake flask cultivation of the recombinant strains

Given the enhanced enzyme activities of strain Kp3 by oligo-mediated gene editing, shake flask cultivation was performed to investigate glycerol metabolism. As shown in Fig. 2A, both the strains Kp3 and Kpac grew slower relative to wild type. Presumably, this was due to deletion of dhaT gene, which led to shortage of NAD⁺. Interestingly, although Kp3 grew slower than wild type K. pneumoniae, it grew faster than Kpac (Fig. 2A). Evidently, this was ascribed to oligo-mediated genetic modification, which in turn affected glycerol metabolism and cell growth (Fig. 2A and B). As expected, cell growth was roughly proportional to glycerol consumption. As shown in Fig. 2B, most glycerol was consumed in 24 h by wild type K. pneumoniae and only 2 g L⁻¹ glycerol was left in the medium. By contrast, the strain Kpac consumed only 5 g L⁻¹ glycerol in 24 h and residual glycerol was 35 g L⁻¹ (Fig. 2B). Clearly, the lack of dhaT gene in the strain Kpac significantly reduced glycerol consumption. Compared with Kpac, the strain Kp3 consumed approximately 20 g L⁻¹ glycerol in 24 h, although it was much lower than wild type K. pneumoniae (38 g L⁻¹). We ascribed this result to oligo-mediated genomic modification which elevated the PDOR activity and accelerated glycerol consumption in strain Kp3 (Table 1). As for 3-HP production, Kp3 produced 0.6 g L⁻¹ 3-HP in 15 h, while the strain Kpac yielded only 0.2 g L⁻¹ 3-HP (Fig. 2D). Clearly, this was also attributed to oligo-mediated genomic modification. Another interesting finding in this study is that, although ldh gene was knocked out, the strains Kp3 and Kpac failed to completely abolish the production of produced lactic acid. As shown in Fig. 2C, the strain Kp3 produced 1.9 g L⁻¹ lactic acid in 15 h, while the wild type K. pneumoniae produced only 0.8 g L⁻¹ lactic acid. This can be explained by the existence of other lactic acid pathways. As for acetic acid production, both Kpac and Kp3 produced much less acetic acid compared with wild type K. pneumoniae (Fig. 2E), indicating that acetic acid synthesis was largely blocked.

Fig. 2 Time courses of the recombinant K. pneumoniae strains in shake flask cultivation. (A) Cell concentration; (B) residual glycerol; (C) lactic acid; (D) 3-hydroxypropionic acid; (E) acetic acid. K. pneumoniae: wild type K. pneumoniae; Kpac: the recombinant K. pneumoniae strain where the ldh gene was deleted and the dhaT gene was replaced by aldH. Kp3: the mutant Kpac by oligos.

3.5. Bioreactor cultivation of recombinants

To overproduce 3-HP with the strain Kp3, fermentation was performed in a 5 L bioreactor with maintenance of pH value at 7.0 by supplying NaOH. We first investigated the effect of air supply on cell metabolism. Results showed that air supply in the first 12 h had no significant impacts on cell metabolism (ESI Fig. 3†). To reduce energy consumption, fermentation process was divided into two stages. In the first 15 h, air supply was 0.4 vvm and agitation speed was 200g, while from 15 h to 36 h they were adjusted to 1.5 vvm and 400g, respectively. In this two-phase fermentation, dissolved oxygen was not necessary, however micro-aerobic conditions were maintained to benefit 3-HP production. Consequently, the strain Kp3 produced 6.39 g L⁻¹ 3-HP and 32.69 g L⁻¹ 1,3-PDO in 36 h (Fig. 3), and 100 g L⁻¹ glycerol was consumed. The total glycerol conversion ratio was 0.465 mol mol⁻¹ in 36 h.

Fig. 3 Fed-batch cultivation of the strain Kp3 in a 5 L bioreactor. Kp3: the mutant Kpac strain by oligos. Kpac: the recombinant K. pneumoniae strain in which the ldh gene was deleted and the dhaT gene was replaced by aldH.

3.6. Effect of byproducts formation on 3-HP production

To unravel the influence of byproducts formation on 3-HP production, antisense mRNAs were recruited to block the translation of lactic acid and acetic acid synthesis genes (ldh and poxB, respectively) (Table 2, Fig. 4). The strain K. pneumoniae (pET-anti-ldh) that harbored the antisense fragment of ldh produced only 0.27 g L⁻¹ lactic acid. By contrast, the reference strain K. pneumoniae (pET-pk) that harbored the empty vector pET-pk yielded 0.99 g L⁻¹ lactic acid. Similarly, the strain Kp (pET-anti-poxB) that carried the antisense fragment of poxB yielded 0.81 g L⁻¹ acetic acid, while the reference strain K. pneumoniae (pET-pk) produced 2.25 g L⁻¹ acetic acid. Clearly, the two antisense mRNAs largely repressed the translation of ldh and poxB and thereby repressed the formation of lactic acid and acetic acid (Table 2, Fig. 4). To simultaneously block the synthesis of lactic acid and acetic acid, two antisense fragments were fused and cloned into vector, resulting in the recombinant vector pET-anti-ldh-anti-poxB and subsequently the recombinant strain K. pneumoniae (pET-anti-ldh-anti-poxB). As expected, compared to control strain, double repression of the two genes decreased the production of lactic acid and acetic acid by 80% and 50%, respectively.

Table 2 Effects of antisense RNA modules on metabolism of recombinant K. pneumoniae strains^a

Strains	Kp (pET-pk)	Kp (pET-anti-ldh)	Kp (pET-anti-poxB)	Kp (pET-anti-ldh-anti-poxB)
a 3-HP: 3-hydroxypropionic acid; Kp (pET-pk): the recombinant K. pneumoniae strain harboring vector pET-pk, pk indicates the native promoter of dhaB genes; Kp (pET-anti-ldh): the recombinant K. pneumoniae strain expressing antisense fragment of ldh gene; Kp (pET-anti-poxB): the recombinant K. pneumoniae strain expressing antisense fragment of poxB gene; Kp (pET-anti-ldh-anti-poxB): the recombinant K. pneumoniae strain expressing the antisense fragments of both ldh and poxB genes. One asterisk (*) indicates P value < 0.05, while two asterisks (**) indicate P value < 0.01. The experiments were performed in triplicate.
Biomass (OD600)	1.42 ± 0.05	1.43 ± 0.04	1.47 ± 0.04	1.49 ± 0.10
Residual glycerol (g L^−1)	4.2 ± 0.25	11.2 ± 1.49*	17.5 ± 1.78**	23.30 ± 1.82**
3-HP (g L^−1)	0.85 ± 0.06	0.82 ± 0.04	0.83 ± 0.04	0.62 ± 0.04**
Lactic acid (g L^−1)	0.99 ± 0.08	0.27 ± 0.02**	0.78 ± 0.06	0.24 ± 0.03**
Acetic acid (g L^−1)	2.25 ± 0.11	2.06 ± 0.03	0.81 ± 0.06**	1.11 ± 0.30**

---

Fig. 4 Reduced levels of lactic acid and acetic acid by antisense RNA technology. Kp(pET-aldh): the recombinant K. pneumoniae strain expressing the aldh gene from E. coli. Kp(pET-anti-ldh-anti-poxB-aldh): the recombinant K. pneumoniae strain coexpressing two antisense modules and aldH gene.

To overproduce 3-HP, the aldH gene was expressed under the control of pk promoter, the native promoter of dhaB gene cluster. The alhH expression cassette was cloned into the downstream of the two antisense modules. The entire expression vector was transformed into K. pneumoniae, resulting in a recombinant strain named K. pneumoniae (pET-anti-ldh-anti-poxB-aldH). This recombinant strain was micro-aerobically fermented and the strain K. pneumoniae (pET-aldh) was used as the control. Results showed that the antisense modules had no significant impacts on cell growth, except for a slight decrease in glycerol consumption. However, lactic acid production was significantly reduced (Fig. 4A). During the entire fermentation process, the strain K. pneumoniae (pET-anti-aldh) produced less lactic acid compared with the reference strain K. pneumoniae (pET-aldh). In 24 h, the strains K. pneumoniae (pET-aldh) and K. pneumoniae (pET-anti-ldh-anti-poxB-aldh) produced lactic acid of 1.91 g L⁻¹ and 1.46 g L⁻¹, respectively. Although K. pneumoniae (pET-anti-ldh-anti-poxB-aldh) produced more acetic acid relative to the reference strain in the mid-phase of fermentation, however, approximately 50% acetic acid disappeared at 24 h (Fig. 4B). More importantly, the strain K. pneumoniae (pET-anti-ldh-poxB-aldh) yielded 1.29 g L⁻¹ 3-HP at 24 h, whereas the reference strain K. pneumoniae (pET-aldh) yielded only 0.76 g L⁻¹ 3-HP. Evidently, the enhanced production of 3-HP in K. pneumoniae (pET-anti-ldh-poxB-aldh) was ascribed to antisense modules which repressed the synthesis of lactic acid and acetic acid. In view of above results, we conclude that there has a competition between 3-HP production and byproducts formation for glycerol carbon flux. So far, antisense RNA technology has been recruited to engineer industrial strain²⁵ and regulate gene expression at a genomic scale.²⁶ Future work may focus on simultaneous repression or activation of multiple genes using CRISPR-Cas9 technology based on genome-scale modeling of K. pneumoniae.²⁷

4. Conclusion

To reduce the lactic acid production in K. pneumoniae, we knocked out its synthesis gene ldh by RecA homologous recombination. Next, we developed Red recombinase system in K. pneumoniae, by which the dhaT gene was replaced by aldH gene to block 1,3-PDO formation and simultaneously overproduce 3-HP. To reshape K. pneumoniae genome, chemically synthesized oligos were electro-transformed into K. pneumoniae. One mutant strain showed enhanced production of 3-HP. Sequencing results revealed that the mutation occurred in lactic acid and acetic acid synthesis pathways. In a 5 L bioreactor, this mutant strain (Kp3) produced 6.39 g L⁻¹ 3-HP and 32.6 g L⁻¹ 1,3-PDO without using any antibiotics and inducers. To clarify the relationship between 3-HP production and byproducts formation, antisense mRNA technique was employed. Results showed that the decrease of lactic acid and acetic acid levels was synchronized with the increase of 3-HP production, implying the competition between byproducts and 3-HP for the glycerol carbon flux. Overall these results provide valuable insights into glycerol metabolism in K. pneumoniae. We believe that these newly developed techniques including Red-mediated recombinant system and antisense RNA technology will facilitate basic research and also advance the metabolic engineering of K. pneumoniae for production of 3-HP, 1,3-PDO, 2,3-BDO, or beyond.

Acknowledgements

This work was supported by grants from National High Technology Research and Development Program (863 Program) (No. 2015AA021003), National Basic Research Program of China (973 Program) (No. 2012CB725200), National Natural Science Foundation of China (No. 21276014, 21476011), Fundamental Research Funds for the Central Universities (YS1407), Innovation Capability Enhancement Program for Universities Governed by Beijing Municipal Commission of Education (No. PXM2015_014209_000010), and Project from Beijing Key Laboratory of Bioactive Substances and Functional Foods in Beijing Union University (ZK70201406).

References

1. K. Menzel, A. Zeng and W. Deckwer, Enzym. Microb. Tech., 1997, 20, 82–86.
2. E. Johnson and E. Lin, J. Bacteriol., 1987, 169, 2050–2054.
3. R. Forge and E. Lin, J. Bacteriol., 1982, 151, 591–599.
4. Y. Ko, S. Ashok, S. Zhou, V. Kumar and S. Park, Process Biochem., 2012, 47, 1135–1143.
5. Y. Huang, Z. Li, K. Shimizu and Q. Ye, Bioresour. Technol., 2012, 103, 251–359.
6. Y. Huang, Z. Li, K. Shimizu and Q. Ye, Bioresour. Technol., 2013, 128, 505–512.
7. D. Wei, M. Wang, J. Shi and J. Hao, J. Ind. Microbiol. Biotechnol., 2012, 39, 1219–1226.
8. S. Ashok, S. Raj, C. Rathnasingh and S. Park, Appl. Microbiol. Biotechnol., 2011, 90, 1253–1265.
9. K. Amako, Y. Meno and A. Takade, J. Bacteriol., 1988, 170, 4960–4962.
10. H. Wang, F. Isaacs, P. Carr, Z. Sun, G. Xu, C. Forest and G. Church, Nature, 2009, 460, 894–898.
11. F. Isaacs, P. Carr, H. Wang, M. Lajoie, B. Sterling, L. Karaal, A. Tolonen, T. Gianoulis, D. Goodman, N. Reppas, C. Emig, D. Bang, S. Hwang, M. Jewett, J. Jacobson and G. Church, Science, 2011, 333, 348–353.
12. J. Hauschild, B. Petersen, Y. Santiago, A. Queisser, J. Carnwath, A. Lucas-Hahn, L. Zhang, X. Meng, P. Gregory, R. Schwinzer, G. Cost and H. Niemann, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 12013–12017.
13. L. Cong, F. Ran, D. Cox, S. Lin, R. Barretto, N. Habib, P. Hsu, X. Wu, W. Jiang, L. Marraffini and F. Zhang, Science, 2013, 339, 819–823.
14. C. Bro and J. Nielsen, Metab. Eng., 2004, 6, 204–211.
15. Y. Jin, H. Alper, Y. Yang and G. Stephanopoulos, Appl. Environ. Microbiol., 2005, 71, 8249–8256.
16. J. Bailey, A. Sburlati, V. Hatzimanikatis, K. Lee, W. Renner and P. Tsai, Biotechnol. Bioeng., 2002, 79, 568–579.
17. R. Podschun and U. Ullmann, Clin. Microbiol. Rev., 1998, 11, 589–603.
18. A. Clements, D. Tull, A. Jenney, J. Farn, S. Kim, R. Bishop, J. McPhee, R. Hancock, E. Hartland, M. Pearse, O. Wijburg, D. Jackson, M. McConville and R. Strugnell, J. Biol. Chem., 2007, 282, 15569–15577.
19. N. Delihas and S. Forest, J. Mol. Biol., 2001, 313, 1–12.
20. L. Waters and G. Storz, Cell, 2009, 136, 615–628.
21. J. George and H. Wolfgang, Microbiol. Mol. Biol. Rev., 2011, 75, 286–300.
22. M. Wang, L. Fan and T. Tan, RSC Adv., 2014, 4, 57791–57798.
23. Y. Li, M. Su, X. Ge and P. Tian, Biotechnol. Lett., 2013, 35(10), 1609–1615.
24. Y. Li, L. Liu and P. Tian, Biotechnol. Lett., 2014, 36(11), 2215–2221.
25. Y. Yang, Y. Lin, L. Li, R. J. Linhardt and Y. Yan, Metab. Eng., 2015, 29, 217–226.
26. J. A. Brophy and C. A. Voigt, Mol. Syst. Biol., 2016, 12(1), 854.
27. P. Xu, S. Ranganathan, Z. Fowler, C. Maranas and M. A. G. Koffas, Metab. Eng., 2011, 13(5), 578–587.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra12511e

---