Improving CoQ10 productivity by strengthening glucose transmembrane of Rhodobacter sphaeroides

Background Several Rhodobacter sphaeroides have been widely applied in commercial CoQ10 production, but they have poor glucose use. Strategies for enhancing glucose use have been widely exploited in R. sphaeroides. Nevertheless, little research has focused on the role of glucose transmembrane in the improvement of production. Results There are two potential glucose transmembrane pathways in R. sphaeroides ATCC 17023: the fructose specific-phosphotransferase system (PTSFru, fruAB) and non-PTS that relied on glucokinase (glk). fruAB mutation revealed two effects on bacterial growth: inhibition at the early cultivation phase (12–24 h) and promotion since 36 h. Glucose metabolism showed a corresponding change in characteristic vs. the growth. For ΔfruAΔfruB, maximum biomass (Biomax) was increased by 44.39% and the CoQ10 content was 27.08% more than that of the WT. glk mutation caused a significant decrease in growth and glucose metabolism. Over-expressing a galactose:H+ symporter (galP) in the ΔfruAΔfruB relieved the inhibition and enhanced the growth further. Finally, a mutant with rapid growth and high CoQ10 titer was constructed (ΔfruAΔfruB/tac::galPOP) using several glucose metabolism modifications and was verified by fermentation in 1 L fermenters. Conclusions The PTSFru mutation revealed two effects on bacterial growth: inhibition at the early cultivation phase and promotion later. Additionally, biomass yield to glucose (Yb/glc) and CoQ10 synthesis can be promoted using fruAB mutation, and glk plays a key role in glucose metabolism. Strengthening glucose transmembrane via non-PTS improves the productivity of CoQ10 fermentation. Supplementary Information The online version contains supplementary material available at 10.1186/s12934-021-01695-z.


Introduction
Glucose is a common monosaccharide that is available in abundance. As glucose is cheap and easy to use for microorganisms, it can serve as an ideal source of carbon for producing high-value products, such as CoQ 10 , through microbial fermentation [1][2][3]. Studies related to microbial glucose metabolism have always been a hot topic in industrial microbiology [1,4]. As glucose metabolism pathways have been well established for many microorganisms, many novel biotechnologies, particularly metabolic engineering, synthetic biology, and systems biology, have been applied to modify intracellular metabolic pathways to enhance the glucose utilization efficiency in microorganisms [1,4,5]. Besides the functional enzymes for glucose metabolism, glucose utilization also requires a set of genes that encode specific transporters and regulators [5]. Glucose transmembrane is an important step because exogenous glucose cannot go into cells through

Open Access
Microbial Cell Factories free diffusion and must rely on a transporter to cross the cell membrane [6]. Recently, researchers have realized the importance of glucose transport efficiency during microbial fermentation and focus on microbial sugar transmembrane studies [6][7][8]. So far, the sugar transmembrane mechanisms of many industrial microbes are still unknown, limiting metabolic modification of sugar transmembrane in these microbes.
Microorganisms depend on more than one system to transport exogenous glucose; the glucose transmembrane mechanisms for E. coli have been widely investigated [6,7]. E. coli can use two pathways for glucose transmembrane: phosphoenolpyruvate (PEP): carbohydrate phosphotransferase system (PTS) and non-PTS [6]. The non-PTS include the ATP binding cassette (ABC) system and the major facilitator superfamily (MFS) system. PTS Glc is a multiprotein phosphorelay system that accompanies the import and simultaneous phosphorylation of carbohydrates. Since the discovery of the PTS Glc in E. coli, it has existed in many other bacteria [9]. It has been confirmed that the PTS Glc is primarily composed of enzymes including IICB Glc /IIA Glc (EIIs), HPr, and enzyme I (EI). EI and HPr, the two sugar-nonspecific protein constituents of the PTS, are soluble cytoplasmic proteins participating in the transport of all PTS carbohydrates [6]. EIIs are sugar-specific transporters connecting the common PEP/EI/HPr phosphoryl transfer pathway. PTS Glc is considered an effective way to use glucose because only one phosphoenolpyruvate is coupled with the translocation phosphorylation of glucose when forming an ATP. In contrast, use of glucose through the ABC transporter requires extra ATP for glucose phosphorylation in the carbohydrate kinase reaction [6,10]. Therefore, PTS Glc is a preferred channel for transferring exogenous glucose into cells in industrial bacteria, such as E. coli, Bacillus subtilis, and Corynebacterium glutamicum [4,11,12]. E. coli activates the non-PTS system for glucose transmembrane when exogenous glucose concentration is low (< 1 mM), or PTS Glc function is defective [7]. Some bacteria lacking PTS Glc , such as Pseudomonas putida, utilize the ABC system to transfer exogenous glucose into cells [9].
R. sphaeroides has received significant attention because of its wide biotechnological applications, such as its ability to synthesize a high content of CoQ 10 , carotenoids, and isoprenoids as a source of pharmaceutical materials [13][14][15][16]. CoQ 10 is an oil-soluble quinone that has a decaprenyl side chain. So far, it has been widely used in functional food and cosmetics industries because of its antioxidant function. Researchers have recently found that CoQ 10 can regulate several genes that play an important role in cholesterol metabolism, inflammatory responses, or both [14]. Moreover, it is beneficial to patients with cardiovascular diseases, hypertension, and Parkinson's disease [17]. Compared with the methods of animal and plant extraction and chemosynthesis, the production of CoQ 10 by microbial fermentation is lowcost, safe, and efficient [16]. Additionally, R. sphaeroides, a CoQ 10 producer with high contents of CoQ 10 , has been widely used in the industrial production of CoQ 10 . For the commercial production of CoQ 10 with microbial fermentation, glucose acts as a major carbon source. Various strategies for guiding the metabolic flux toward CoQ 10 biosynthesis have been exploited for R. sphaeroides [16]. Nevertheless, presently, there is little research on improving CoQ 10 production by modifying the glucose transmembrane. Although R. sphaeroides has a wide spectrum of carbon source utilization, it had a low glucose consumption rate than E. coli [18,19]. Fuhrer et al. reported that the glucose uptake rate of R. sphaeroides was 1.8 ± 0.1 mmol/g dry cells weight (DCW)/h, which was only approximately 23.07% as that of E. coil [19]. Therefore, the glucose transmembrane process of R. sphaeroides was a bottle-neck step for glucose metabolism, which may be a new way to further promote the productivity of CoQ 10 .
Considering the importance of glucose transmembrane for glucose metabolism and poor glucose utilization efficiency of R. sphaeroides, the first potential pathway of glucose transmembrane of R. sphaeroides were analyzed in this work. Later, a deep study was conducted to show the function of these pathways on glucose metabolism. Finally, we evaluated the influence of glucose transmembrane on CoQ 10 synthesis efficiency and optimized CoQ 10 fermentation by R. sphaeroides by modifying glucose transmembrane. Moreover, R. sphaeroides ATCC 17023 is a paradigmatic organism among isolated R. sphaeroides strains with clear genetic background and mature genetic manipulation tools that we chose as a research object in this study [20].

Strains, media and growth conditions
All the strains used in this study were summarized in Table 1. R. sphaeroides ATCC 17023 (wild type, WT) was used as the parental strain in this work. The mutant strains were constructed in this background. E. coli JM109 was used as a plasmid host, and E. coli S17-1 was used to conjugate DNA into R. sphaeroides. R. sphaeroides ATCC 17023 and mutant strains were routinely cultivated at 32℃ in medium A (3 g/L glucose, 2 g/L NaCl, 8 g/L yeast extract, 0.256 g/L MgSO 4 ·7H 2 O,1.3 g/L KH 2 PO 4 , 15 μg/L biotin, 1 mg/L thiamine hydrochloride, 1 mg/L nicotinic acid, pH 7.2) as seed (exponentially phase cells, OD 600 ≥ 2). R. sphaeroides cultures were incubated at 32℃ in modified Sistrom's minimal medium (MSMM) lacking succinate and with glucose as an alternative carbon source (6.5 g/L) [18]. Before inoculation the seeds of these strains were washed with sterilized PBS (pH 7.2), and then resuspended with the same solution to adjust the cell density equal (OD 600 1). Antibiotics were added into the medium A and MSMM when necessary. E. coli JM109 and E. coli S17-1 were grown at 37 °C in Luria-Bertani (LB) medium with antibiotics when necessary. The concentrations of antibiotics and chemicals used in the construction of plasmids and recombinant strains were as follows: 25 μg/mL kanamycin and 150 μg/mL K 2 TeO 3 for R. sphaeroides, and 100 μg/mL kanamycin for E. coli strains. Lab-scale fermentation of CoQ 10 was carried out with medium B containing 40 g/L glucose, 6.3 g/L MgSO4, 4 g/L corn steep liquor, 3 g/L sodium glutamate, 3 g/L (NH4) 2 SO 4 , 3 g/L KH 2 PO 4 , 2 g/L CaCO 3 , 1 mg/L nicotinic acid, 1 mg/L thiamine hydrochloride, and 15 μg/L biotin, supplemented with 25 μg/mL kanamycin.

Plasmids construction
Plasmids used in this study were listed in Table 1, and the related primers and restriction enzymes were presented in Additional file 6: Table S1. pK18mobsacB::glk-L-R was constructed with splicing by overlap extension (SOE) PCR. Briefly, an upstream and downstream fragment of glk was amplified with primer pair glk-L-F/glk-L-R and glk-R-F/glk-R-R using R. sphaeroides ATCC 17025 genomic DNA as template, respectively. The two fragments were joined by SOE PCR to generate a 1,560-bp fragment, which was digested and ligated into the PstI/SphI sites of pK18mobsacB to obtain pK18mobsacB::glk-L-R. The details of experiment were presented in Additional file 1: Fig. S1. An upstream and downstream fragment of fruA was amplified with primer pair fruA-L-F/fruA-L-R and fruA-R-F/fruA -R-R. The two fragments were joined by SOE PCR to generate a 1,819-bp fragment, which was digested and ligated into the SalI/EcoRI sites of pK18mobsacB to obtain pK18mobsacB::furA-L-R. The construction of pK18mobsacB::fruB-L-R obeyed to the same procedure of pK18mobsacB::furA-L-R construction. The two fragments were amplified with primer pair fruB-L-F/fruB-L-R and fruB-R-F/fruA-R-R, and then were joined by SOE PCR to generate a 1399-bp fragment. Afterwards, the fragment was ligated into the EcoRI/SphI sites of pK18mobsacB to obtain pK18mobsacB::furB-L-R. The details of the two plasmids construction were shown in Additional file 2: Fig. S2. The wide host-range conjugative plasmids pBBR1MCS-2 and pBBR1MCS-2::tac were used to construct the over-expression vectors. For the construction of pBBR1MCS-2::tac::galP, the entire open reading frame of galP gene was amplified by PCR with primer pair galP op -F/galP op -R using E. coil K-12 substr. MG1655 genomic DNA as template. After digestion with XbaI/SacI, it was ligated into pBBR1MCS-2::tac to obtain pBBR1MCS-2::tac::galP. The details of pBBR1MCS-2::tac::galP construction were presented in Additional  Fig. S3. Simultaneously, pBBR1MCS-2::galP was obtained following the same method of pBBR1MCS-2::tac::galP construction.

Strains construction
Δglk and ΔfruAΔfruB were constructed as in-frame markerless deletion of almost the entire open reading frames by a two-step recombination strategy, which was based on diparental conjugation as previously described [14]. In short, E. coli S-17 bearing the target plasmid was used as the donor strain, and R. sphaeroides ATCC 17023 was used as the recipient strain. The donor/ recipient ratio was 1:7 for conjugation and the colonies resistant to 150 μg/mL K 2 TeO 3 and 25 μg/mL kanamycin were picked and cultivated for plasmid extraction and sequencing verification. For the construction of Δglk, the pK18mobsacB::glk-L-R was transformed into R. sphaeroides ATCC 17023 by diparental conjugation. Clones obtained from single homologous recombination event were selected on kanamycin/K 2 TeO 3 supplemented medium A agar plates, and then grown for 48-72 h in medium without antibiotic, followed by serial dilution and plating onto SMM plates containing 10% (w/w) sucrose to select for a second crossover event. Finally, sucrose-resistant and kanamycin-sensitive colonies were selected as potential positive transformants and verified by PCR using glk-P1/glk-P4 primers. To obtain the mutant ΔfruAΔfruB, ΔfruA was firstly constructed, and then the gene fruB was knocked out of the genome of ΔfruA further. The method was similar to that of the construction of Δglk. The ΔfruAΔfruB was verified by primer pair fruA-P1/fruA-P4 and fruB-P1/fruB-P4. For the construction of ΔfruAΔfruB/tac::galP OP , pBBR1MCS-2::tac::galP was transformed into ΔfruAΔfruB by diparental conjugation, and selected on medium A supplemented with kanamycin (25 μg/mL) and K 2 TeO 3 (150 μg/mL). Afterwards, the resistant clones were selected and verified. Also, other gene over-expression mutants were constructed following the same method. Primers used for mutant strains verification were listed in Additional file 7: Table S2.

RNA extraction and RT-qPCR assay
Different growth-phase cells (1 × 10 7 ) cultured in MSMM medium were harvested by centrifugation at 8000×g for 3 min at 4 ℃. Total RNA was extracted from R. sphaeroides using a Total RNA Extraction Kit and purified as described by the procedures. Synthesis of cDNA was performed by the reverse transcription reaction according to the HiFiScript cDNA Synthesis Kit instructions. Cham Q Universal SYBR qPCR master mix (Vazyme, Nanjing, China; Q711-02) and the ABI First Step Quantitative Polymerase Chain Reaction System (Applied Biosystems, San Mateo, CA, USA) were used to perform quantitative polymerase chain reactions under the following reaction conditions: 95 °C for 3 min, followed by 40 cycles of 95 °C for 10 s, 56 °C for 30 s, and 72 °C for 30 s. As an internal standard control, the relative abundance of 16S rRNA was used to standardize the results. All assays were performed in triplicate. Primers used in RT-qPCR were listed in Additional file 8: Table S3. The result of RT-qPCR was calculated based on the following formula: where G i/A is the gene of interest in sample A, G c/A is the gene of internal control in sample A, G i/B is the gene of interest in sample B and G c/B gene of internal control in sample B.

CoQ 10 fermentation in Lab-scale bioreactors
For the lab-scale fermentation of CoQ 10 , three 1-L quadruple fermentation tanks (Bench-Top Bioreactor Multifors 2 type, INFORS HT, Switzerland) were used. The operation procedure was based on the method of Zhang et al. [13]. Seeds of R. sphaeroides were prepared as above shaking-flask cultivation. Subsequently, 16 mL of seeds were inoculated into tank with 0.8 L sterilizing medium B supplemented with 25 μg/mL kanamycin, and then, cultured at 32 °C for 96 h. The pH was maintained at 7.0. The aeration and agitation protocol were 1 vvm and 400 rpm during the whole fermentation. Bacterial growth (OD 600 ), glucose concentration and CoQ 10 titer were detected each 12 h.

Analytical methods
Cultured broth was fetched each 12 h for biomass determination at OD 600 by spectrophotometer (MAPADA INSTRUMENTSUV-1800, China) and calculated using a calibration curve which indicated the relationship between OD 600 and dry cell weight (DCW) (1OD 600 approximately equaled to 0.40 g DCW/L). In this study, 100 mL of cultivated broth were centrifugated to obtain cell pellets, and then used to get dry cells through lyopilization for DCW measuring. Residual glucose was detected by a SBA-40 Biosensor (Biology Institute of Shandong Academy of Sciences, China). pH was measured at the starting and end of cultivation by a Sev-enCompact ™ pH meter S220 (METTLER TOLEDO, China). The method of extraction and quantification of CoQ 10 was according to Zhang et al. [13].

Potential glucose transmembrane pathways for R. sphaeroides ATCC 17023
By retrieving the NCBI database, the only integral PTS, fructose-specific PTS (PTS Fru ), was found in the genome of R. sphaeroides ATCC 17023, which is encoded by the gene cluster fruAB (RSP_1788 and RSP_1786). fruB encodes EI and HPr, the two sugar-nonspecific protein constituents of the PTS, and fruA encodes the sugarspecific transporter. It was reported that PTS Fru encoded by fruAB simultaneously had a function of glucose transmembrane in some E. coil strains [6]. Additionally, fruAB in R. sphaeroides may have a similar function as that of the abovementioned E. coil strains. Glucose transported into cells with non-PTS must be phosphorylated before subsequent metabolism. Although the non-PTS-type glucose-specific transporter in R. sphaeroides ATCC 17023 has not been identified, the enzyme glucokinase (glk, RSP_2875), which plays a role in glucose phosphorylation, exists in the genome [5,8]. Additionally, the glucokinase activity had been determined in some R. sphaeroides strains when cultured with glucose as the sole carbon source [21]. Considering the above information, R. sphaeroides ATCC 17023 should possess the non-PTS. R. sphaeroides ATCC 17023 metabolizes glucose exclusively with the Entner-Doudoroff pathway (ED) under aerobic and anaerobic conditions because of the lack of phosphofructokinase in the Embden-Meyerhof-Parnas pathway (EMP) [18]. According to the abovementioned analysis, metabolic networks that contain the glucose transmembrane and catabolism were constructed and the result is depicted in Fig. 1.

Influence of PTS and non-PTS on bacterial growth and glucose metabolism
To clarify whether PTS (fruAB), non-PTS, or both of R. sphaeroides ATCC 17023 influences cellular glucose metabolism, two mutants (ΔfruAΔfruB and Δglk) were constructed using an in-frame markerless deletion method. The ΔfruAΔfruB was a mutant with double knock out of fruA and fruB. Considering that no non-PTS type glucose transporter has been identified in R. sphaeroides presently, glk was knocked out to study the function of non-PTS in glucose metabolism. Subsequently, these mutants' growth and glucose consumption were studied under aerobic incubation using glucose as the sole carbon source (Fig. 2). All mutant strains showed a lag phase at the beginning of cultivation (between 0 and 12 h), which was similar to that of the R. sphaeroides ATCC 17023 (WT) (Fig. 2a). Evident variations were observed among these strains since then. The WT went into an exponential growth phase and displayed a rapid growth rate than others between 12 and 24 h. Furthermore, the ΔfruAΔfruB went into an exponential growth phase though the growth rate was slower than the WT; however, Δglk still showed a slow growth rate. After that, the growth rate of the WT became slower from 36 to 72 h, going into a decline phase at 72 h. ΔfruAΔfruB showed a faster growth rate between 24 and 36 h, and then the rate gradually slowed. The stationary growth phase was observed at approximately 60 h, and the decline phase appeared at 72 h for ΔfruAΔfruB. In contrast, Δglk continually kept a slow growth status between 24 and 36 h and went into a long stationary growth phase till the end of the experiment. Interestingly, the Biomax obtained by ΔfruAΔfruB was 3.22 ± 0.04 g DCW/L, which was much higher than that of the WT (Bio max was 2.23 ± 0.07 g DCW/L) ( Table 2). Although the Δglk showed a typical bacterial growth process, both the growth rate and Bio max were much weaker than the WT during the whole culture process that the Bio max was 0.82 ± 0.01 g DCW/L achieved by the Δglk. Additionally, glucose concentration was determined simultaneously (Fig. 2b). At the beginning of cultivation (0-12 h), the strains showed slow glucose consumption rates (r glc ) that fit the characteristics of the lag phase. During the culture time between 12 and 24 h, the WT and ΔfruAΔfruB sped up glucose consumption, though ΔfruAΔfruB had a little slower consumption rate than the WT. The result could explain the reason why ΔfruAΔfruB grew slower than the WT. Afterward, the residual glucose concentration in the group with ΔfruAΔfruB was less than the WT. Finally, ΔfruAΔfruB exhausted the glucose within 72 h, which was 12 h earlier than the WT. Compared with the WT, Δglk showed a low ability on glucose consumption during the entire process. After incubation for 96 h, there was still 3.73 ± 0.21 g/L residual glucose in the medium. The r glc of Δglk was only 0.026 ± 0.001 g/L/h, whereas the r glc of the ΔfruAΔfruB could get to 0.086 ± 0.002 g/L/h, which was approximately 1.18 times that of the WT (Table 2). Besides the r glc , the Y b/glc of the ΔfruAΔfruB was also promoted, approximately 43.4% higher than that of WT. Additionally, we constructed the mutant strains, ΔfruA and ΔfruB. The results revealed that the two mutants showed similar growth and glucose metabolism status as those of the ΔfruAΔfruB (unpublished data). Summarily, glk mutation seriously inhibited the growth and glucose metabolism of R. sphaeroides. Considering that glk is a vital gene involved in non-PTS, we speculated that the non-PTS played a major role in transporting glucose for R. sphaeroides ATCC 17023 during the entire process. However, deleting fruAB also influenced bacterial growth and glucose metabolism. The depressing effect of the fruAB mutation on growth appeared at the early incubation phase (12-24 h), whereas it showed a promotion effect on growth at the later phase (24-72 h). The influence of fruAB mutation on growth could be indirectly explained by the status of glucose metabolism. Therefore, we supposed that both the PTS and non-PTS had influence on glucose metabolism at the early culture phase. The relative transcriptional level of the fruA and glk in WT during cultivation was determined by RT-qPCR to verify the hypothesis further. The result was depicted in Fig. 2c, both fruA and glk showed an increasing tendency in the early culture phase (12-24 h). After that, the transcriptional level of fruA showed a decreased tendency since 36 h, whereas the glk still kept increasing from 36 to 48 h. Additionally, ΔfruAΔfruBΔglk was obtained through deleting the gene glk on the basis of ΔfruAΔfruB. Comparing with Δglk, both growth and glucose utilization were inhibited further at the early culture phase (Additional file 4: Fig. S4). To sum up, the results illustrate that both non-PTS and PTS have a function on glucose metabolism during the early phase, and the non-PTS played an important role in glucose metabolism.

Enhancing the non-PTS pathway to promote cellular glucose metabolism
According to the above study, blocking the non-PTS inhibited the glucose metabolism of R. sphaeroides ATCC 17023. Whether over-expressing the non-PTS-type glucose transporter helps in improving glucose catabolism. In this section, the galactose:H + symporter (galP) from E. coil K-12 substr. MG1655 was selected for the study. First, three mutants, ΔfruAΔfruB/bp, ΔfruAΔfruB/galP OP , and ΔfruAΔfruB/tac::galP OP , were constructed with the over-expression vectors, pBBR1MCS-2 and pBBR1MCS-2::tac. The ΔfruAΔfruB/bp was directly introduced to the blank plasmid in ΔfruAΔfruB. The ΔfruAΔfruB/galP OP was introduced to the plasmid, only harboring the gene, galP. The ΔfruAΔfruB/tac::galP OP is inserted with a strong promoter tac before the gene galP based on the pBBR1MCS-2::tac. Subsequently, these mutant strains were separately cultivated with glucose as the carbon source, and the biomass and glucose concentration was determined every 12 h. The ΔfruAΔfruB/bp showed almost no difference from that of ΔfruAΔfruB in growth and glucose metabolism (Fig. 3). This means that the plasmid introduction did not influence bacterial growth and glucose metabolism. Compared with ΔfruAΔfruB/bp, the growth rate of ΔfruAΔfruB/galP OP was increased at the early phase (12-24 h), but the Table 2 Growth and glucose consumption assay of the R. sphaeroides mutant strains and WT * Bio max the maximum biomass; ** r glc the average glucose consumption rate; *** Y b/glc biomass yield to glucose consumption vs. the Bio max . Statistics analysis was performed based on one-way ANOVA method and the data in the same column with the same letters (a-c) meant no significant difference (P ≤ 0.05)

Strain Time (h) Glucose metabolism (g/L)
Bio max * (g DCW/L) r glc ** (g/L/h) Y b/glc *** (g/g)  growth status was the same as that of ΔfruAΔfruB/bp between 24 and 48 h (Fig. 3a). From 48 h, the biomass achieved by ΔfruAΔfruB/galP OP was higher than that of ΔfruAΔfruB/bp though the growth trends were similar. The higher biomass achieved by ΔfruAΔfruB/galP OP could be explained by the faster glucose consumption rate than the ΔfruAΔfruB/bp during this period.
The result also suggested that the over-expression of galP could improve cellular glucose metabolism. For ΔfruAΔfruB/tac::galP OP , the growth improved further than ΔfruAΔfruB/galP OP at the early phase (12-24 h), and then, it still kept a fast growth status than others until the time glucose was nearly exhausted. Additionally, the biomass quantity achieved was higher than that of ΔfruAΔfruB/galP OP . The Bio max was 4.01 ± 0.15 g DCW/L achieved by ΔfruAΔfruB/tac::galP OP , which was the highest value among these strains. For glucose metabolism, ΔfruAΔfruB/tac::galP OP exhausted the glucose in the medium within 60 h, and the r glc reached 0.107 ± 0.003 g/L/h (Table 3). Furthermore, glk was overexpressed in ΔfruAΔfruB (ΔfruAΔfruB/tac::glk). However, both the growth and glucose metabolism decreased compared with ΔfruAΔfruB/bp (Additional file 5: Fig. S5). Additionally, the result suggested that the original glk transcriptional level was fitting for glucose metabolism. Maybe, over-expression of the glk produced excessive glucose-6-P, which is toxic to cells.

Improving CoQ 10 productivity of R. sphaeroides
The CoQ 10 content of these mutants was determined, and the result is presented in Table 4. Compared with the WT, Δglk, ΔfruAΔfruB, and ΔfruAΔfruB/tac::galP OP synthesized a low content of CoQ 10 when incubated for 24 h; especially, the CoQ 10 content of Δglk was 3.12 ± 0.04 mg/g DCW. After that, CoQ 10 content of ΔfruAΔfruB and ΔfruAΔfruB/tac::galP OP was increased after 48 h, whereas the CoQ 10 content of the WT and Δglk showed a slight reduction. As incubation proceeded (48-96 h), the CoQ 10 content of the WT and Δglk stopped reducing and increased. Simultaneously, ΔfruAΔfruB and ΔfruAΔfruB/tac::galP OP increased in the CoQ 10 content. Finally, the CoQ 10 content of ΔfruAΔfruB and ΔfruAΔfruB/tac::galP OP reached 5.02 ± 0.18 and 5.11 ± 0.14 mg/g DCW, respectively. The maximum CoQ 10 content of ΔfruAΔfruB/tac::galP OP was increased by 29.4% than the WT. It can be proposed that the mutation of fruAB improved Y b/glc and bacterial r glc but also enhanced the CoQ 10 synthesis of R. sphaeroides. Moreover, strengthening glucose transportation by overexpressing galP showed little help to strengthen CoQ 10 synthesis.
Although galP over-expression in ΔfruAΔfruB played no role in promoting the CoQ 10 synthesis ability of R. sphaeroides, the strategy can promote glucose metabolism rate, which shortens the fermentation time. The inactivation of fruAB improved Y b/glc and the bacterial CoQ 10 synthetic ability. Considering the advantages of the two strategies, ΔfruAΔfruB/tac::galP OP was applied to CoQ 10 fermentation in a lab-scale tank (1 L), evaluating whether the CoQ 10 fermentation is improved. ΔfruAΔfruB/tac::galP OP showed an evident improvement in growth compared with the WT/bp during the Table 3 Growth and glucose metabolism of R. sphaeroides strains * Bio max the maximum biomass; ** r glc the average glucose consumption rate; *** Y bio/glc biomass yield to glucose consumption versus the Bio max . Statistics analysis was performed based on one-way ANOVA method and the data in the same column with the same letters (a-d) meant no significant difference (P ≤ 0.05)

Strain
Time ( (Fig. 4a). The Bio max of ΔfruAΔfruB/tac::galP OP was harvested at 72 h of fermentation, which was 24 h earlier than the WT/bp. Moreover, the value of the Bio max reached 17.24 ± 0.97 g DCW/L, which was promoted by approximately 16% higher than that of the WT/bp (14.85 ± 0.57 g DCW/L). Simultaneously, the glucose concentration in the medium was almost exhausted after 72 h for ΔfruAΔfruB/tac::galP OP (< 5 g/L), whereas there was more than 10-g/L residual glucose residual for the WT/bp. In the aspect of CoQ 10 synthesis (Fig. 4b), the yield gradually increased as the fermentation proceeded for both strains. At 48 h incubation, the yield of ΔfruAΔfruB/tac::galP OP showed a higher level than that of the WT/bp, and the phenomenon lasted to the end. The maximum CoQ 10 titer of ΔfruAΔfruB/tac::galP OP reached 78.14 ± 2.31 mg/L, which was approximately 49.76% higher than that of the WT/bp. Moreover, ΔfruAΔfruB/tac::galP OP achieved the maximum CoQ 10 titer at 72 h, which was 24 h earlier than the WT/bp.

Discussion
For many bacteria, PTS Glc is the first-selected pathway to transport exogenous glucose [6,22]. After that, some other sugar-specific-PTSs, such as the PTS Fru (fruAB), have been identified with the same function as that of the PTS Glc [6,23]. In R. sphaeroides ATCC 17023, no PTS Glc -encoding genes are identified presently, but it has a PTS Fru encoding gene cluster (fruAB). It is revealed that glucose metabolism and bacterial growth were influenced by the mutation of the fruAB in R. sphaeroides ATCC 17023 (Fig. 2). Interestingly, the result of fruAB mutation revealed two effects on bacterial growth during the whole culture process. It showed an inhibition effect at the early cultivation phase (12-24 h) and displayed a promoting effect at the late phase at 36 h. Finally, the Bio max received by ΔfruAΔfruB was much higher than that of the WT. The glucose metabolism of ΔfruAΔfruB also displayed a fit change characteristic vs. that of the growth change.
Additionally, RT-qPCR assay revealed that the transcription level of the fruA in the WT kept a relatively high level at the early cultivation phase (12-24 h) and then decreased at 36 h. With the comprehensive analysis of the results, we supposed that the PTS Fru in R. sphaeroides ATCC 17023 majorly joined in the glucose metabolism at the early cultivation phase. The PTS Glc is considered an effective way to utilize glucose because only one PEP is coupled with the translocation-phosphorylation of PTS carbohydrates when forming one ATP [6,22]. In contrast, glucose utilization through a non-PTS transporter requires extra ATP to phosphate a carbohydrate molecule in the carbohydrate kinase reaction. Regarding energy consumption, bacteria synthesize cellular skeleton materials at the early cultivation phase, requiring a large amount of energy; thus, the PTS type system may be a good choice for saving energy at the early growth phase. As cultivation continued, the function of fruAB was weakened. The result might be due to the energy production ability of the bacteria, which is not as a limiting factor for cell growth after the lag phase. The high biomass achieved by ΔfruAΔfruB meant that more carbon fluxed to cellular assimilation metabolism. For the native glucose utilization pathway in E. coli, half of the PEP produced is used for glucose uptake and phosphorylation. PEP is an essential precursor for synthesizing many chemicals, such as succinate, malate, and aromatic compounds. In this sense, fruAB mutation might reduce PEP catabolism, which helps to synthesize cytoskeleton substances after the lag growth phase. The similar phenomenon is also found in the mutation of PTS Ntr in P. putida, which is due to the enhancement of anabolism [16]. The phenomenon obtained from fruAB mutation is interesting though the mechanism is unclear. In future studies,