Establishing an innovative carbohydrate metabolic pathway for efficient production of 2-keto-l-gulonic acid in Ketogulonicigenium robustum initiated by intronic promoters

Background 2-Keto-l-gulonic acid (2-KGA), the precursor of vitamin C, is currently produced by two-step fermentation. In the second step, l-sorbose is transformed into 2-KGA by the symbiosis system composed of Ketogulonicigenium vulgare and Bacillus megaterium. Due to the different nutrient requirements and the uncertain ratio of the two strains, the symbiosis system significantly limits strain improvement and fermentation optimization. Results In this study, Ketogulonicigenium robustum SPU_B003 was reported for its capability to grow well independently and to produce more 2-KGA than that of K. vulgare in a mono-culture system. The complete genome of K. robustum SPU_B003 was sequenced, and the metabolic characteristics were analyzed. Compared to the four reported K. vulgare genomes, K. robustum SPU_B003 contained more tRNAs, rRNAs, NAD and NADP biosynthetic genes, as well as regulation- and cell signaling-related genes. Moreover, the amino acid biosynthesis pathways were more complete. Two species-specific internal promoters, P1 (orf_01408 promoter) and P2 (orf_02221 promoter), were predicted and validated by detecting their initiation activity. To efficiently produce 2-KGA with decreased CO2 release, an innovative acetyl-CoA biosynthetic pathway (XFP-PTA pathway) was introduced into K. robustum SPU_B003 by expressing heterologous phosphoketolase (xfp) and phosphotransacetylase (pta) initiated by internal promoters. After gene optimization, the recombinant strain K. robustum/pBBR-P1_xfp2502-P2_pta2145 enhanced acetyl-CoA approximately 2.4-fold and increased 2-KGA production by 22.27% compared to the control strain K. robustum/pBBR1MCS-2. Accordingly, the transcriptional level of the 6-phosphogluconate dehydrogenase (pgd) and pyruvate dehydrogenase genes (pdh) decreased by 24.33 ± 6.67 and 8.67 ± 5.51%, respectively. The key genes responsible for 2-KGA biosynthesis, sorbose dehydrogenase gene (sdh) and sorbosone dehydrogenase gene (sndh), were up-regulated to different degrees in the recombinant strain. Conclusions The genome-based functional analysis of K. robustum SPU_B003 provided a new understanding of the specific metabolic characteristics. The new XFP-PTA pathway was an efficient route to enhance acetyl-CoA levels and to therefore promote 2-KGA production. Electronic supplementary material The online version of this article (10.1186/s12934-018-0932-9) contains supplementary material, which is available to authorized users.


Background
2-Keto-l-gulonic acid , the precursor of vitamin C, is currently produced by two-step fermentation from d-sorbitol. The first step is the generation of l-sorbose from d-sorbitol using Gluconobacter oxydans, and then l-sorbose is transformed into 2-KGA in the second step by the symbiosis system composed of Ketogulonicigenium vulgare and Bacillus megaterium. In the symbiosis system, B. megaterium is generally used as a companion strain that generates metabolites to assist the growth of K. vulgare and the accumulation of 2-KGA [1]. However, due to the different nutrient requirements and the uncertain optimal ratio of K. vulgare and B. megaterium, the co-culture system strongly limits the strain improvement and the fermentation optimization. Therefore, a mono-culture of K. vulgare is considered to be more cost-effective and manageable in the 2-KGA fermentation system. Nevertheless, the available industrial strain K. vulgare exhibited poor growth ability and lower 2-KGA productivity under mono-cultured condition [2]. Thus, to replace the coculture fermentation system with one robust strain of Ketogulonicigenium could be revolutionary in the vitamin C industry. The alternatives for the two-step fermentation process are described in Fig. 1.
With the rapid development of high-throughput DNA sequencing technology, the whole genomes of several strains of K. vulgare have been completely sequenced and annotated in recent years [3][4][5][6], providing new perspectives to carry out detailed investigations on strain improvement. According to the genome annotation and metabolic network analysis, Liu et al. [7] found that K. vulgare were deficient in some key genes in the de novo biosynthesis of glycine, histidine, lysine, threonine, proline, isoleucine, methionine and leucine. Therefore, it was found that reconstruction of the biosynthetic pathways of threonine in K. vulgare Hkv604 increased 2-KGA production by 25.13%, and the fermentation period Fig. 1 The alternatives of the two-step fermentation process of vitamin C. a The old version of the two-step fermentation process. b The new version of the two-step fermentation process decreased by 28.57% when co-cultured with Bacillus endophyticus [8]. The defects in the metabolism of carbohydrates, amino acids and vitamins were the main reasons for the poor growth of K. vulgare Hbe602; however, the up-regulation of the tricarboxylic acid (TCA) cycle, as well as the metabolism of amino acids and nucleotide facilitated K. vulgare better growth [3].
Acetyl-CoA, as a central metabolic intermediate, plays important roles in cellular processes involved in the TCA cycle, lipid synthesis, amino acids metabolism and posttranslational acetylation [9,10]. Its main function is to deliver the acetyl group to the TCA cycle to be oxidized to CO 2 and H 2 O, as well as energy in the form of ATP and GTP. The intermediates generated in the TCA cycle can be used as precursors for the biosynthesis of amino acids and lipids. In addition, the central position in metabolism endows acetyl-CoA with an important regulatory role. Acetyl-CoA serves as a substrate for the synthesis of lysine acetyltransferases (KATs), which transfer the acetyl group to lysines in histones and other proteins [11]. The acetylation of histones at the genes important for growth culminates in the expression of growth genes and a commitment to cell growth and division [9,12]. To meet the cellular requirement, a variety of routes to synthesize acetyl-CoA have been formed, such as the decarboxylation of pyruvate, the catabolism of branched amino acids and the β-oxidation of fatty acids [13]. In the synthesis of acetyl-CoA from carbohydrate catabolism, the hexose is converted into d-glucose 6-phosphate (G6P) by phosphorylation and sequentially converted to d-ribulose 5-phosphate (Ru5P) by 6-phosphogluconate dehydrogenase (pgd) in the oxidative part of the pentose phosphate pathway (PPP). In the Embden-Meyerhof-Parnas (EMP) pathway, G6P is broken down to form pyruvate followed by the oxidative decarboxylation by pyruvate dehydrogenase (pdh). The byproduct of the two decarboxylation reactions is carbon dioxide, which is the main reason for the inefficiency of carbon utilization.
Recently, genetic engineering has been widely used in strain improvement in 2-KGA fermentation [8,22,23]. The promoter is a basic factor, playing a significant role in gene expression. The E. coli_tufB promoter, identified by An and Friesen [24], was successfully used in the conversion of d-sorbitol to 2-KGA in G. oxydans_G624 [25]. Shi et al. [26] isolated and identified the promoter gHp0169 from G. oxydans DSM 2003, which exhibited stronger activity than the E. coli_tufB promoter in the initiation of gene expression, and resulted in a twofold increase in the specific molar yield of 2-keto-d-gluconic acid. Besides the promoter of the sorbose dehydrogenase gene (Psdh) and the promoter of the sorbosone dehydrogenase gene (Psndh) in K. vulgare DSM 4025 [27], no other speciesspecific promoter in the Ketogulonicigenium sp. has been reported. Thereby, developing a series of well-characterized promoters is a necessary prerequisite to fine-tuning heterogeneous gene expression in Ketogulonicigenium sp.
In this study, a unique strain of K. robustum SPU_B003 with good growth ability and 2-KGA producing capacity in a mono-culture system, was reported. To fully understand the genome characteristics and to provide a comprehensive understanding of the genetic background for gene expression research, the whole genome of K. robustum SPU_B003 was sequenced. Based on the genome information, two putative promoters, P1 (orf_01408 promoter) and P2 (orf_02221 promoter), were predicted and identified by initiating the heterologous gfp gene expression in K. robustum SPU_B003. Compared to the promoter E. coli_tufB, the two promoters exhibited stronger activity, especially promoter P1, when initiating gfp expression.
To investigate the effect of oversupply of acetyl-CoA on cell growth and 2-KGA production, a higher efficiency acetyl-CoA biosynthetic pathway with lower carbon loss was constructed by heterologously expressing phosphoketolase (xfp) and phosphotransacetylase (pta) in K. robustum SPU_B003. The engineered strain K. robustum/pBBR-P1_xfp2502-P2_pta2145 achieved the highest acetyl-CoA production, and reached a 22.27% increase in 2-KGA production. Accordingly, the transcriptional level of the genes pgd and pdh were decreased by 24.33 ± 6.67 and 8.67 ± 5.51%, respectively, and the transcriptional level of the genes sorbose dehydrogenase (sdh) and sorbosone dehydrogenase (sndh) were up-regulated at different levels. This may be because the XFP-PTA pathway is helpful in promoting the TCA cycle and generating α-ketoglutarate, the precursor of glutamate. Glutamate, as a building block, contributes to the biosynthesis of pqqA, the rate-determining step in pyrroloquinoline quinone (PQQ) biosynthesis. So, presumably, it is helpful to regulate electron transfer in the l-sorbose oxidation process and then affects the transcriptional levels of sdh and sndh.

Strains, plasmids and genes
All the strains and plasmids used in this work were described in Table 1. K. robustum SPU_B003 was isolated from a soil sample (E104°19′17.2″/N23°31′48.10″) and kept in our laboratory and in Microbial Resource Center of Shenyang Pharmaceutical University by cryopreservation and lyophilization. The E. coli WM3064 [28] was used as a donor strain in conjugation with K. robustum SPU_B003. The Bifidobacterium animals subsp. lactis was exchanged from the Bioresource Collection and Research Center (BCRC). The plasmid pBBR1MCS-2 [29] was used as an expression vector in K. robustum SPU_B003.

Genome sequence, annotation and metabolic pathway analysis
The complete genome sequence of K. robustum SPU_ B003 was determined by the Chinese National Human Genome Center (Shanghai, China) using 454 singleend sequencing technology. A total of 123,540 reads with 20.49-fold coverage was generated. The reads were assembled into 29 large contigs (> 500 nucleotides) by using the 454 Newbler assembler (454 Life Sciences, Branford, CT). The relationship of contigs was determined by ContigScap [30], and the gaps between contigs were filled by PCR amplification followed by DNA sequencing.
The genome analysis was performed by Rapid Annotation using Subsystem Technology (RAST) [31]. The coding regions were distinguished by Gene Locator and Interpolated Markov ModelER (GLIMMER) [32]. The function annotation of open reading frames (ORFs) was carried out by using the Basic Local Alignment Search Tool (BLAST) [33] against the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Japan) [34]. Both tRNA and rRNA genes were identified by tRNAscan-SE [35] and RNAmmer [36], respectively. The NAD and NADP related genes were predicted by RAST. The metabolic networks were constructed using the KEGG database.
The nucleotide sequences immediately upstream of the ORF were subjected to BPROM [37] to predict promoters as well as the position of the − 10 and − 35 boxes.

Genetic manipulation
The gfp gene was cloned from the pX551-gfp plasmid and was used as a reporter. The promoter E. coli_tufB was cloned from E. coli K12 by PCR according to the sequence reported by An and Friesen [24]. The fusion of the predicted promoter P1 sequence and the gfp fragment was made by overlap extension PCR, followed by insertion into the expression vector pBBR1MCS-2 to obtain the recombinant plasmid pBBR-P1_gfp. The same methods were used to obtain the plasmid pBBR-P2_gfp and pBBR-PtufB_gfp.
The gene xfp with its original upstream promoter were amplified from B. animals, and pta with its original promoter were from E. coli K12. The PCR fragments were digested with restriction enzyme and ligated into pBBR1MCS-2 to produce pBBR-xfp-pta. The original promoters of xfp and pta were replaced by promoter P1 and P2 of K. robustum SPU_B003, respectively, herein to construct the plasmid pBBR-P1_xfp-P2_pta. Due to the codon preference of K. robustum SPU_B003, the genetic bases of heterologous xfp and pta were optimized to xfp2502 and pta2145 and synthesized by Sangon Biotech (Shanghai Co., Ltd). The codon frequency table and the optimized parameters were listed in Additional file 1: Tables S1, S2. Thus, another plasmid pBBR-P1_xfp2502-P2_pta2145 was built. Afterwards, the plasmids were transferred from E. coli WM3064 to K. robustum SPU_ B003 by conjugation.
For the orthogonality test, the promoters P1 and P2 were assembled with xfp2502 and pta2145 (4 combinations) on pBBR1MCS-2. The orthogonality test was conducted by the same method described above. All the primers and DNA sequences used in this work were listed in Additional file 1: Tables S3, S4.

Fluorescence intensity determination
The recombinant K. robustum SPU_B003 was cultured in seed medium for 24 h, harvested at 6000 g for 10 min, and washed twice with PBS buffer. The whole cell fluorescence (RFU/OD 600 , the relative fluorescence unit divided by the corresponding cell density) was determined according to Xu et al. [38] by a fluorescence microplate reader (Infinite M200 Pro, Tecan, Mannedorf, Zurich, Switzerland). The excitation and emission wavelength were 485 and 535 nm, respectively. The cell density was determined at 600 nm by a microplate reader. Meanwhile, the cells were photographed by using a fluorescence microscope (Olympus BX53F, Tokyo, Japan).

Enzyme activity assays
The phosphoketolase activity was detected spectrophotometrically at 505 nm by the formation of ferric acetyl hydroxamate [39]. One unit of phosphoketolase activity was defined as the amount of protein forming 1 μmol of acetyl phosphate (acetyl-P) per 1 min from fructose 6-phosphate. Specific activity was defined as units per 1 mg of protein. Protein concentration was determined according to the method described by Bradford with bovine serum albumin as standard [40].
The phosphotransacetylase activity was measured by the CoA-dependent arsenolytic decomposition of acetyl-P according to the procedure of Suzuki [41]. One unit of phosphotransacetylase was defined as the amount of protein extract that catalyzed the decomposition of 1 μmol of acetyl-P per minute.

Intracellular acetyl-CoA quantification
Intracellular acetyl-CoA was measured by using an Acetyl-CoA Assay Kit (Suzhou Comin Biotechnology Co., Ltd). Malate dehydrogenase (MDH) catalyzed malic acid and NAD to oxaloacetic acid and NADH. Then, citrate synthase (CS) catalyzed acetyl-CoA and oxaloacetic acid to form citric acid and CoA. Through the coupled reaction of MDH and CS, the content of acetyl-CoA was proportional to the formation rate of NADH. Therefore, the concentration of acetyl-CoA was determined by measuring the rate of absorbance rise at 340 nm. The sample was normalized by protein concentration [40].

Quantitative real time PCR
The cells were collected at 24 h of fermentation by centrifugation at 10,000g for 10 min at 4 °C. The total RNA was isolated by using the TriZol reagent (Vazyme Biotech Co., Ltd), and then extracted with chloroform. The RNA was precipitated with isopropanol and washed with 75% ethanol. After dissolving in RNase-free water, the sample concentration was quantified by a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA). Quantitative real time PCR was conducted with GoTaq qPCR Master Mix [Promega (Beijing) Biotech Co., Ltd] on Stratagene Mx3000P (Agilent Technologies, Inc, USA). The 16S rRNA gene was used as an internal standard. The primers and DNA sequences were listed in Additional file 1: Tables S3, S4.

Cell density determination
The cell density was measured spectrophotometrically at 600 nm after dissolving CaCO 3 with 100 mM HCl.

2-KGA concentration determination
The concentration of 2-KGA in fermentation broth was measured by the high-performance liquid chromatography (HPLC) using an amino column. The mobile phase used was acetonitrile-KH 2 PO 4 (5%/95%, v/v) with a flow rate of 0.6 mL/min.
Compared with the published genome, species-specific characteristics of K. robustum SPU_B003 were presented in Table 2. The numbers of tRNAs (61) and rRNA (18) in K. robustum SPU_B003 are more than those in the four reported K. vulgare. In the dividing cell, the amount of tRNAs determines codon translation efficiency [45]. rRNAs are the key players in the ribosome which carries out protein synthesis. Therefore, the abundance of tRNAs and rRNAs can contribute to the biosynthesis of protein in K. robustum SPU_B003. The number of NAD and NADP biosynthetic genes (10) is more than K. vulgare WSH-001 (8), SKV (8) and Hbe602 (8). K. robustum SPU_B003 has abundant regulation and cell signaling system related genes (32) in response to changes in their external environment. In addition, unlike K. vulgare WSH-001 [7], K. robustum SPU_B003 has a more complete amino acids biosynthetic pathway except for histidine, alanine and asparagine (Fig. 2). All these factors probably facilitated the ability of K. robustum SPU_B003 to grow independently.
According to the gene annotation and metabolic network analysis, carbohydrate metabolism pathways were studied, showing the presence of the genes encoding all the enzymes of PPP, Entner-Doudoroff (ED) pathway and TCA cycle, and the deficiency of the gene encoding 6-phosphofructokinase (EC 2.7.1.11) (Fig. 3). Therefore, the PPP pathway and ED pathway were the main central carbohydrate metabolism routes to link up with the TCA cycle. Considering the low efficiency (2/9 carbon loss) in the synthesis of acetyl-CoA in the conventional pathway, a new acetyl-CoA biosynthetic pathway was established by expressing heterologous phosphoketolase (xfp) and phosphotransacetylase (pta) in K. robustum SPU_B003 (Fig. 3). In the new pathway, l-sorbose was converted to F6P after a series of reactions, and F6P was directly broken down to acetyl-P and E4P by phosphoketolase (XFP). Then, E4P molecules regenerated F6P by carbon rearrangement. After the reactions catalyzed by transketolase and transaldolase, F6P and E4P were broken down to X5P and d-ribose 5-phosphate (R5P). X5P was broken down to acetyl-P and G3P. Finally, the acetyl group was transferred to coenzyme A catalyzed by phosphotransacetylase (PTA) to produce acetyl-CoA. Therefore, theoretically, 1 mol of l-sorbose can produce 3 mol of acetyl-CoA without CO 2 production, which greatly improves the utilization rate of l-sorbose and the content of intracellular acetyl-CoA. The improvement of acetyl-CoA content helps to promote the TCA cycle, and then generates various intermediates for amino acid biosynthesis and energy to meet biological requirements. In addition, acetyl-CoA as a signal initiates the cellular growth program by promoting the acetylation of histones specifically at growth genes.

Promoter analysis
The upstream sequences of orf_01408 and orf_02221, which encode 50 s ribosomal protein L13 and DNA-directed RNA polymerase beta subunit, were predicted as promoter P1 and P2 by BPROM, separately. The spacing between the putative − 10 and − 35 regions were 13 bp for P1 and 15 bp for P2, which met the general pattern of most bacterial promoters.

Promoter activity identification
The activities of promoter P1 and P2 were compared with E. coli_tufB, a commonly used promoter, by measuring the gfp fluorescence intensity (RFU/OD) in K. robustum SPU_B003. As shown in Fig. 4
The specific activity of PTA exhibited a similar trend to XFP (Fig. 5b). Strain K. robustum/pBBR-P1_xfp2502-P2_pta2145 (6.83 ± 0.35 U/mg) exhibited the best performance in the three engineered strains, which illustrated that the optimized gene pta2145 was expressed in K. robustum SPU_B003 with high efficiency and that promoter P2 was stronger than the original promoter of pta. Therefore, the foreign genes could well exhibit enzymatic activity initiated by intronic promoters in K. robustum SPU_B003.

Intracellular acetyl-CoA quantification
To verify the introduced XFP-PTA pathway was functional, the level of its downstream product, acetyl-CoA, was determined by the coupled reaction of MDH and CS, in which the content of acetyl-CoA was proportional to the formation rate of NADH. As shown in Fig. 5c, the acetyl-CoA level in strain K. robustum/ pBBR-P1_xfp2502-P2_pta2145 (224.10 ± 11.14 nmol/ mg prot) increased by approximately 2.4folds compared to K. robustum/pBBR1MCS-2 (94.88 ± 7.00 nmol/mg prot) and also higher than strain K. robustum/pBBR-xfp-pta (123.20 ± 8.44 nmol/ mg prot) and strain K. robustum/pBBR-P1_xfp-P2_pta (140.10 ± 13.64 nmol/mg prot). This indicated that the introduced xfp/xfp2502 and pta/pta2145 were functional in the biosynthetic process of acetyl-CoA and that the optimized genes, xfp2502 and pta2145, could further increase intracellular acetyl-CoA levels. The carbon flow through the new XFP-PTA pathway was successfully constructed in K. robustum SPU_B003.

2-KGA production assay
To assess the effect of the new metabolic pathway on cell growth and 2-KGA production, all strains were cultivated in fermentation medium containing 40 g/L l-sorbose in flasks. The concentration of 2-KGA in fermentation broth was measured by HPLC using acetonitrile-KH 2 PO 4 (5%/95%, v/v) as mobile phase with a flow rate of 0.6 mL/min. The production of 2-KGA and biomass were described in Fig. 6a, b. The 2-KGA production of wild type strain K. robustum SPU_B003 reached 33.68 ± 2.17 g/L with a conversion rate of 78.13 ± 5.03% and a biomass OD 600 of 1.15 ± 0.043 at 84 h. However, the control strain K. robustum/pBBR1MCS-2 exhibited a slight decrease in 2-KGA yield (32.42 ± 0.96 g/L, conversion rate 75.21 ± 2.23%) and in biomass (OD 600 1.10 ± 0.042) at the end of the fermentation. These indicated that the metabolic burden caused by the blank plasmid had little inhibitory effect on cell growth and 2-KGA production.
By introducing the XFP-PTA pathway into K. robustum SPU_B003, the intracellular acetyl-CoA pool was increased, and the cell biomass and the conversion rate of 2-KGA were enhanced (Fig. 6d). The large pool of acetyl-CoA could contribute to the formation of higher levels of intermediates in the TCA cycle, including malate, citric acid, α-ketoglutarate, succinic acid and fumaric acid, which could facilitate the growth of Ketogulonicigenium sp. in the process of 2-KGA production [3,47]. Moreover, the strong carbon flux via the TCA cycle increased the energy supply for biological activities as well as intermediates for the biosynthesis of amino acids. For example, α-ketoglutarate as an intermediate in the TCA cycle, is used to synthesize glutamate, glutamine, proline and arginine. Subsequently, glutamine is used as a major source of cellular nitrogen participates in the de novo biosynthesis of nucleotides and other nonessential amino acids. Furthermore, oxaloacetate is an important precursor for the biosynthesis of threonine, lysine and aspartate. Additionally, glycine, serine, proline, threonine and isoleucine were the key amino acids affecting cell growth and 2-KGA production [7,8,48]. Additionally, acetyl-CoA as a carbon-source rheostat, initiates the cellular growth by promoting the acetylation of histones at the growth genes [9]. So, presumably, the high concentration of acetyl-CoA can up-regulate the TCA cycle, consequently stimulate the relevant amino acid biosynthesis, and promote the acetylation of histones at growth genes, then facilitate cell growth and 2-KGA production.
Although the XFP-PTA pathway can theoretically achieve a 100% carbon yield to the desirable acetyl-CoA, this occurs at the expense of reducing equivalents in PPP and ATP in the EMP pathway. The abundance of NAD and NADP biosynthetic genes probably contributed to overcome the insufficiency of reducing power. In addition, the up-regulated TCA cycle could compensate for the consumption of ATP during biological activity.

Effect of the established XFP-PTA metabolic pathway on the related gene transcription
As shown in Fig. 7a, the genes xfp/xfp2502 and pta/ pta2145 were not transcribed in the wild type K. robustum SPU_B003 or control strain K. robustum/ pBBR1MSC-2, because there were no foreign genes in the two strains. While the genes xfp/xfp2502 and pta/ pta2145 were both transcribed in the recombinant strains, and the transcriptional levels in K. robustum/ pBBR-P1_xfp2502-P2_pta2145 and K. robustum/pBBR-P1_xfp-P2_pta were higher than those in K. robustum/ pBBR-xfp-pta. This indicated that promoters P1 and P2 were stronger than the original promoters of heterologous xfp and pta in initiating gene expression in K. robustum SPU_B003.
To further investigate the carbon distribution in the new XFP-PTA pathway, the transcriptional levels of decarboxylation related genes were detected by RT-qPCR. As shown in Fig. 7b, the transcriptional level of pgd directly decreased by 24.33 ± 6.67% in K. robustum/pBBR-P1_xfp2502-P2_pta2145 compared to the control strain, while that of the pdh decreased slightly (8.67 ± 5.51%). l-Sorbose can be metabolized to acetyl-CoA directly through the XFP-PTA pathway, or to pyruvate through the PPP pathway and the ED pathway, and further converted to acetyl-CoA by pyruvate dehydrogenase (PDH). The widespread source of pyruvate makes it difficult to control the transcriptional level of pdh, leading to a slight decrease in transcription. Whereas 6PG as the substrate of PGD is only produced in the PPP pathway. Therefore, changes in the carbon flow through the PPP pathway can easily lead to the change in the transcriptional level of pgd. All in all, the down-regulation of pdh and pgd means a change occurrence in carbon flux and that the CO 2 release has been reduced to some extent.
To investigate the effects of the XFP-PTA pathway on the key genes responsible for the biosynthesis of 2-KGA, the transcriptional levels of sdh and sndh were determined by RT-qPCR. Among the five key genes, sdh-1 (orf_02251), sdh-2 (orf_02271), sdh-3 (orf_p00164) and sndh-2 (orf_01127) exhibited higher transcriptional levels than the control strain, while sndh-1 (orf_00349) decreased to some extent (Fig. 7c). One explanation for this phenomenon was that the genes of sdh played more important roles in the biosynthesis of 2-KGA. This corroborated a previous report showing that the 2-KGA concentration exhibited a positive linear relationship with SDH activity and quantity of Ketogulonicigenium sp. [46]. SDH and SNDH are the key enzymes responsible for the biosynthesis of 2-KGA, both of which are membrane-bound dehydrogenases with PQQ as cofactor. The PQQ synthesis genes in K. vulgare were annotated as the gene cluster pqqABCDE. pqqA, a rate-determining step in PQQ biosynthesis, encodes a 24-amino-acid polypeptide bearing tyrosine and glutamate as building blocks [49]. Because the XFP-PTA pathway is helpful in promoting the TCA cycle and generating α-ketoglutarate, the precursor of glutamate, this route is presumably conducive to the biosynthesis of PQQ, and then affects the PQQ-regulated electron transfer in the l-sorbose oxidation process, finally affecting the transcription of sdh and sndh.

Conclusions
In this study, a robust strain K. robustum SPU_B003, showing good growth characteristics and improved 2-KGA productivity, was reported. The complete Fig. 7 Transcriptional level of relevant genes in K. robustum/pBBR-P1_xfp2502-P2_pta2145. a Reverse transcription PCR (RT-PCR) analysis of the transcriptional information of xfp/xfp2502 and pta/pta2145. b RT-qPCR analysis of the relative transcriptional level of pgd and pdh in fermentation system. c RT-qPCR analysis of the relative transcriptional level of sdh and sndh. Comparison was performed with control strain K. robustum/ pBBR1MCS-2. Data represent the mean ± SD of 3 replicates. *, ** and *** Significant difference at p < 0.05, 0.01 and 0.001, respectively genome of K. robustum SPU_B003 was sequenced, and the genome-based functional analysis was helpful to further understand the species-specific metabolic characteristics. Compared with K. vulgare, K. robustum SPU_B003 contained more tRNAs, rRNAs, NAD and NADP biosynthetic genes, as well as regulationand cell signaling-related genes. At the same time, the amino acids biosynthesis pathways were more complete. In addition, two internal gene promoters were identified, and their strength was validated by detecting the initiation activity. To further enhance 2-KGA production, an innovative acetyl-CoA biosynthetic pathway was constructed by expressing phosphoketolase and phosphotransacetylase initiated by a species-specific promoter. After introducing the XFP-PTA pathway, strain K. robustum/pBBR-P1_xfp2502-P2_pta2145 achieved the highest biomass and 2-KGA production (39.64 ± 2.84 g/L), which was significantly higher than that of K. vulgare Hbe602 (5.16 ± 0.34 g/L) and K. vulgare Hkv604 (5.73 ± 0.04 g/L) and was also increased by 22.27% compared to the control strain K. robustum/ pBBR1MCS-2 (32.42 ± 0.96 g/L). Meanwhile, the transcriptional levels of pgd and pdh were decreased, and sdh and sndh were increased at different levels. These results suggested that the CO 2 release was decreased and that the oxidation ability was enhanced in recombinant strain. Authors' contributions CYW, YL and YXZ designed the study. CYW carried out the molecular manipulation and fermentation studies and drafted the manuscript. YL carried out the genome annotation and metabolic pathway analysis. ZWG and LCL assisted in the fermentation process. MYZ and TYZ assisted in the revision of the manuscript. CFW and YXZ revised the manuscript. All authors read and approved the final manuscript.

Additional file
Additional file 1: Table S1. Codon frequency of K. robustum SPU_B003. Table S2. Optimization parameters. Table S3. Primers used in this study. Table S4. DNA sequences used in this study. Table S5. Growth characteristics of K. robustum SPU_B003. Figure S1. Phylogenetic analysis of K. robustum SPU_B003 with other species. Figure S2. Orthogonality test of the promoters and heterologous genes.