Efficient production of guanosine in Escherichia coli by combinatorial metabolic engineering

Background Guanosine is a purine nucleoside that is widely used as a raw material for food additives and pharmaceutical products. Microbial fermentation is the main production method of guanosine. However, the guanosine-producing strains possess multiple metabolic pathway interactions and complex regulatory mechanisms. The lack of strains with efficiently producing-guanosine greatly limited industrial application. Results We attempted to efficiently produce guanosine in Escherichia coli using systematic metabolic engineering. First, we overexpressed the purine synthesis pathway from Bacillus subtilis and the prs gene, and deleted three genes involved in guanosine catabolism to increase guanosine accumulation. Subsequently, we attenuated purA expression and eliminated feedback and transcription dual inhibition. Then, we modified the metabolic flux of the glycolysis and Entner-Doudoroff (ED) pathways and performed redox cofactors rebalancing. Finally, transporter engineering and enhancing the guanosine synthesis pathway further increased the guanosine titre to 134.9 mg/L. After 72 h of the fed-batch fermentation in shake-flask, the guanosine titre achieved 289.8 mg/L. Conclusions Our results reveal that the guanosine synthesis pathway was successfully optimized by combinatorial metabolic engineering, which could be applicable to the efficient synthesis of other nucleoside products. Supplementary Information The online version contains supplementary material available at 10.1186/s12934-024-02452-8.


Introduction
Guanosine and its nucleotide derivatives play essential physiological roles in nucleic acid synthesis, and possess antioxidant activity, neurotrophic and neuroprotective effects [1][2][3].Additionally, guanosine has extensive applications as a crucial precursor for some medications used in the treatment of viral infections and tumors as well as food additives [4][5][6].Guanosine production is mainly pentose phosphate pathway (PPP), and purine synthesis pathway (Fig. 1).First, the guanosine biosynthetic pathway in vivo begins with the formation of IMP and GMP.Subsequently, GMP is further converted to guanosine by phosphatase or 5'-nucleotidase.The de novo synthesis of guanosine includes 13 steps from 5′-phosphoribosyl pyrophosphate (PRPP), and precursors include bicarbonate, glycine, aspartate, glutamine, ribose-5′-phosphate (R5P), and some cofactors.In the salvage pathway, phosphoribosyltransferases can catalyze nucleobases to generate nucleotides.In B. subtilis, a gene cluster, purEKB-purC(orf )QLF-purMNH(J)-purD, constitutes the purine operon (Bspur) that contains three overlapping coding units and a single gene [7,8].Additionally, purA, guaA, and guaB are scattered across the genome as single genes.The dual regulation of the Bspur operon is mediated by transcription initiation and termination [9].The pur operon exists as single genes or small operons across the chromosome in E. coli, and its expression is affected by the PurR repressor and the addition of guanine or hypoxanthine [10].
Native guanosine synthesis in bacteria is relatively complex and regulated by the EMP, ED and PPP, several branch and degradation pathways, feedback inhibition, etc. Guanosine anabolism requires a large quantity of purine precursors and considerable energy, and accumulating high levels of guanosine is difficult.Traditionally, chemical or physical mutagenesis methods have been used to generate strains with high guanosine yields.However, it is increasingly difficult to further increase the titre of guanosine based on existing guanosine-producing strains.Microbial cell factories constructed by system metabolic engineering provide an option for increasing the guanosine production.Moreover, such cell factories have been used to boost the production of uridine, inosine and guanosine [11][12][13].Multiple engineering strategies including increasing the supply of precursor, disrupting the branch and degradation pathways, deleting the repressor PurR, overexpressing nucleoside efflux transporter PbuE, constructing an unnecessary protein-reduced chassis, and building genome-scale metabolic network models, have been applied to promote the synthesis of purine nucleosides [11,[14][15][16].E. coli was applied to synthesize up to 7.5 g/L inosine by overexpression of the desensitized prs and purF genes and knockout of the purA, purR, deoD, and purF genes [17,18].Deletion of the gsk gene encoding inosine/guanosine kinase led to increased guanosine accumulation [19].It's reported that the guanosine production in engineered E. coli and B. subtilis was 120 mg/L and 115.2 mg/L in shake flask, respectively [13,19].Overall, previous studies focused on several genetic modifications of synthesis pathways of purine nucleoside.Employing a systems metabolic engineering strategy to construct engineered strains is necessary for the effective production of guanosine.
As a model microorganism, E. coli is a promising chassis strain for guanosine synthesis.Multiple gene editing tools and synthetic biology and metabolic engineering strategies are favourable for constructing cell factories [20].CRISPR/Cas9-mediated homologous recombination is a precise gene editing method that can achieve stable targeted gene knockout or integration [21].To enhance the guanosine titre, we adopted a systematic metabolic engineering strategy to modify the chassis strain E. coli MG1655.First, overexpression of the purine operon Bspur and the prs gene was carried out to enhance the flux into guanosine synthesis.Then, the guanosine degradation pathway was blocked to reduce guanosine consumption.Subsequently, branch pathway for adenosine synthesis, feedback inhibition from repressor PurR, EMP and ED pathways were modified to drive more carbon flux into guanosine synthesis.Next, redox cofactors rebalancing, transporter engineering and enhancing the guanosine synthesis pathway were used to promote guanosine accumulation.Finally, the best strain MQ39 produced 289.8 mg/L guanosine in fed-batch fermentation in shake flasks after 72 h.The constructed cell factory of guanosine production in this study can also be served as chassis strains to synthesize other valuable purine nucleosides like adenosine and inosine.

Strains, plasmids, and cultivation conditions
E. coli DH5α was used for vector construction, and E. coli MG1655 ΔlacI was used as the starting chassis strain to perform genomic editing.Plasmids #1 and #2 for the CRISPR/Cas9 genome editing were kindly supplied by Prof. Huo and Dr. Huang of Beijing Institute of Technology [21].Plasmid #1 consists of a p15A replication origin, a kanamycin resistance gene Kan R , the sucrose-sensitive sacB gene, a CAS9 expression system induced by L-arabinose and an isopropyl-β-D-thiogalactopyranoside (IPTG) inducible λ-Red (Gam, Beta, and Exo) recombination system.Plasmid #2 includes a temperature-sensitive pSC101 replication origin, the promoter P BAD -N20-gRNA scaffold expressing cassette for Cas9 binding, a donor DNAgeneration system serving as an editing template and the ampicillin resistance gene Amp R .The sacB gene of plasmid #1 and the pSC101 replication origin of plasmid #2 were used for plasmid curing.The strains and plasmids used in this study are listed in Table 1.E. coli DH5α and MG1655 ΔlacI were cultured in Luria-Bertani (LB) medium at 37 °C.Ampicillin (100 µg/mL) and kanamycin (50 µg/mL) were added to the cultures when needed.IPTG, L-arabinose, glucose, and sucrose were added at concentrations of 1 mM, 20 mM, 20 g/L and 20 g/L, respectively.

Construction of plasmids and recombinant strains
gRNA expression plasmids were constructed based on plasmid #2 and used for gene deletion and genomic integration.The purified two homologous arms (~ 500 bp) and the inserted expression cassette were fused by overlap extension PCR to obtain donor DNA.Subsequently, the two fragments from the plasmid #2 backbone and donor DNA were ligated with the ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China) to form the specific gRNA plasmid #2.
The recombinant strains were obtained according to a previous protocol [21].Briefly, the specific gRNA plasmid #2 was introduced into the MG1655ΔlacI strain harbouring plasmid #1for gene knockout and chromosomal integration.Colonies grown on the plates were verified by colony PCR, and the correct colonies were confirmed by DNA sequencing.Both plasmids #1 and #2 could be cured by cultivation on LB solid plate containing 2% sucrose at 37 °C for 24 h when the obtained strain was not required to further genetic modifications.The primers are listed in Supplementary Table S1.The integration expression cassettes and a flowchart of gene editing for generating the engineered strains are shown in Supplementary Figs.S1 and S2.

Fed-batch fermentation
For guanosine fermentation, fresh single colonies of engineered strains were inoculated in 10 mL of liquid LBG medium overnight at 37 °C and 220 rpm.Then, the seed cultures of 3 mL were transferred into 27 mL of LBG (LB plus 2% glucose) medium in a 500-mL shake flask and cultured for 72 h at 37 °C and 220 rpm.Fed-batch fermentation was almost the same as the batch method.During the fed-batch process, the pH was kept at approximately 7.0 by adding ammonia with a microinjector, and phenol red was used as the pH indicator.Glucose solution (30%) was added for fed-batch culture when the glucose is depleted.

Substrate and product analysis
Optical density of the engineered strains was measured by the ultraviolet spectrophotometer at 600 nm.The glucose concentration was analysed by the 3,5-dinitrosalicylic acid (DNS) method.Cells were collected and centrifuged at 12,000 rpm for 10 min.The resultant supernatants were filtered through 0.22 μm syringe filters for the next analysis.The guanosine and hypoxanthine concentrations were quantified by HPLC (Agilent 1200) using a RD-C18 5 μm column (Zhongpu scientific, China) using the mobile phase methanol/water mixture (15:85 v/v) with a flow rate of 1 mL/min.The detection temperature and wavelength were set to 25 °C and 254 nm, respectively.

Statistical analysis
All experiments were performed with four independent cultures.Statistical significance was analysed by a twotailed Student's t-test.p value < 0.05 (*) was defined as statistical difference; p value < 0.01 (**) was defined as significant; p value < 0.001 (***) was defined as highly significant; ns was defined as no significant difference.

Increasing the metabolic flux and precursor PRPP supply
To increase the metabolic flux of guanosine synthesis, we sought to introduce the purine synthesis pathway of B. subtilis 168 into chromosome of E. coli.However, the complete purine synthesis pathway Bspur operon contains 11 genes (~ 12 kb) without purR (Fig. 2a and b).It is quite difficult for chromosomal integration expression for large DNA fragments by the CRISPR/Cas9mediated homologous recombination.With increasing length of the integrated fragment, the recombination efficiency significantly decreases, which is not conducive to obtaining positive clones.Therefore, the Bspur operon was divided into four parts, each 1000-5000 bp.These parts were controlled by the P trc promoter and different synthetic terminators to generate expression cassettes (Fig. 2b).Four expression cassettes were further flanked by upstream and downstream homologous arms to form the donor DNA.Meanwhile, we deleted the lacI gene to eliminate regulation of the LacI repressor on the P trc promoter, and the resultant strain MG1655 ΔlacI was used as the starting chassis.After integration of the donor DNA into different loci located on the chromosome, #2 Amp R , gRNA expression vector [21] the engineered E. coli strain MQ4 overexpressing the Bspur operon was obtained.The results showed that MG1655 ΔlacI strain did not produce inosine, while the MQ4 strain achieved the highest inosine production at 7.3 mg/L after 24 h of fermentation, indicating that Bspur operon integration is beneficial for increasing the metabolic flux of purine synthesis in E. coli (Fig. 2d).In addition, MG1655 ΔlacI and MQ4 strains did not produce guanosine (data not shown).MG1655 ΔlacI and MQ4 accumulated 8.1 mg/L and 10.9 mg/L of hypoxanthine after 24 h of fermentation, respectively, and the amount of hypoxanthine still showed an increasing trend.The final OD 600 of MQ4 was very similar to that of MG1655 ΔlacI.
PRPP is a crucial precursor for purine and pyrimidine nucleoside synthesis in vivo.PRPP is generated from R5P and ATP, which is catalysed by PRPP synthase (Figs. 1  and 3a).PRPP synthase is regulated by feedback inhibition via ADP, and mutated PRPP synthase can relieve nucleotide inhibition [17].The prs gene from E. coli, the mutated Ecprs D128A gene and the Baprs gene from Bacillus amyloliquefaciens were integrated into the aslA-glmZ locus of MQ4 (Fig. 3b), respectively.Guanosine and adenosine were not detected by HPLC in MQ5, MQ6, and MQ7.The production of inosine in MQ5, MQ6, and MQ7 was 7.8 mg/L, 8.8 mg/L, and 10.6 mg/L, respectively, which increased by 6.8%, 20.5%, and 45.2% compared to the MQ4 strain.In addition, MQ5, MQ6 and MQ7 strains accumulated 11.6 mg/L, 13.9 mg/L, and 16.0 mg/L of hypoxanthine, respectively (Fig. 3c).These results implied that although the flux of inosine synthesis was further enhanced by the overexpression of the prs gene, the guanosine concentration was possibly too low to be monitored by HPLC.In the de novo synthesis pathway of purine nucleosides, IMP requires a one-step reaction to generate inosine, while the conversion of IMP to guanosine and adenosine requires a three-step reaction.We speculate that IMP may be used to preferentially synthesize inosine, and the metabolic flux of the inosine synthesis pathway is higher than that of guanosine.In addition, guanosine is possibly degraded in E. coli.

Blocking the degradation pathway
Guanosine can be degraded by multiple cellular enzymes, including inosine/guanosine kinase encoded by gsk gene, two guanosine phosphorylases encoded by deoD and ppnP genes, and three nucleoside hydrolases encoded by rihA, rihB, and rihC genes [19,22] (Fig. 4a).Notably, phosphorylase DeoD can also degrade inosine and adenosine into hypoxanthine and adenine, respectively.These enzymes were selected to be blocked in E. coli.The results showed that MQ9 (deoD deletion) produced 13.1 mg/L guanosine after 72 h of fermentation (Fig. 4b).After further knockout of ppnP and gsk, the titre of guanosine in MQ14 and MQ15 reached 16.2 and 23.2 mg/L, respectively.However, the production of guanosine in the MQ16 (rihA deletion), MQ17 (rihB deletion) and MQ18 (rihC deletion) strains was obviously decreased, indicating that the deletion of rihA, rihB and rihC is unfavourable for guanosine accumulation.Subsequently, MQ19 (triple deletion of deoD, ppnP and gsk) accumulated 41.5 mg/L guanosine after 72 h fermentation, an increase of 78.9% compared with MQ15 (Fig. 4b).These results suggested that disrupting the degradation pathway is beneficial for the accumulation of guanosine in E. coli.

Down regulating purA expression and removing feedback and transcription dual inhibition
The conversion of IMP to adenosine was catalysed by adenylosuccinate synthetase (ADSS), which is a competitive pathway for guanosine production (Fig. 5a).Given that adenosine is important for cell growth, the adenosine synthesis pathway was attenuated to direct more carbon flux towards guanosine synthesis.The ssrA degradation tag (DAS + 4) was added to the C-terminus of ADSS encoded by purA gene to induce cytoplasmic degradation of ADSS [23].Additionally, the native purA was replaced with BspurA P242N to lower the flux towards adenosine branch from IMP.The guanosine titres of MQ20 (purA disruption), MQ21 (BspurA P242N ) and MQ23 (purA-ssrA) were 56.3, 57.7 and 50.0 mg/L, respectively (Fig. 5c).However, the final OD 600 of the MQ20 and MQ21 strains was 1.9, which was significantly reduced by 70.3% compared with the MQ19 strain.The biomass of MQ23 strain is similar to that of MQ19 strain, and the guanosine production increased by 20.4%.Considering that OD 600 is an important factor for strain engineering and fermentation, the MQ23 strain was selected for subsequent genetic modification.These results suggested that the purA gene is indispensable for cell growth and that its deletion can lead to severe growth deficiency.
The PurR repressor can inhibit the transcription of the pur operon and the prs gene (Fig. 5b).We sought to delete the purR gene to remove the repression regulation and enhance the intracellular concentration of PRPP.The results showed that MQ24 (purR deletion) accumulated 62.4 mg/L guanosine, an increase of 24.8% compared with MQ23 (Fig. 5c).

Redistributing the metabolic flux of the EMP and ED pathways and balancing redox cofactors
Glucose-6-phosphate (G6P) is an important intermediate of guanosine production as well as a component of the EMP and ED pathways.The biosynthetic pathway of purine competes for metabolic flux with the glycolysis and ED pathways (Fig. 6a).Nevertheless, the disruption of glycolysis and the ED pathway may lead to metabolic imbalance and sever growth defect.The carbon flux between the EMP and purine synthesis pathways should be efficiently distributed to achieve higher guanosine production.In glycolysis, the Pfk enzyme catalyzes the conversion of fructose-6-phosphate (F6P) to fructose-1,6-diphosphate (FBP).Pfk has two isoenzyme forms, and 6-phosphofructokinase I (Pfk-I, encoded by pfkA) possesses more than 90% of Pfk enzymatic activity.Fbpase, including Fbpase I and Fbpase II, can hydrolyse FBP to F6P (Fig. 6a).Compared with Fbpase I (encoded by fbp), Fbpase II, encoded by glpX, exhibits reduced sensitivity towards feedback inhibition of G6P [24,25].In the ED pathway, phosphogluconate dehydratase encoded by edd and 2-keto-3-deoxygluconate-6-phosphate aldolase encoded by eda convert 6-phosphogluconate (6-PG) into pyruvate and glyceraldehyde-3-phosphate.
To enhance flux into the PPP towards the production of guanosine, we deleted the pfkA gene and overexpressed glpX and BsglpX using the weak synthetic promoter P J23116 to adjust the carbon flux of the glycolysis and ED pathways.The guanosine titres of MQ25 (pfkA deletion), MQ26 (pfkA deletion plus glpX overexpression) and MQ27 (pfkA deletion plus BsglpX overexpression) reached 75.6, 78.6 and 90.7 mg/L, respectively (Fig. 6c).Compared to the MQ24 strain, the production of guanosine was increased by 21.2%, 26.0% and 45.4%, respectively.The ED pathway was also interrupted by double knockout of the edd and eda genes.The resultant MQ28 strain accumulated 100.9 mg/L guanosine, an increase of 11.2% compared with MQ27 (Fig. 6c).Engineering the EMP and ED pathways displayed a synergistic effect for guanosine accumulation.These results proved that attenuating the flux of the EMP and ED pathways is a useful strategy for enhancing guanosine synthesis.
Although adjusting glycolysis and the ED pathway can improve guanosine production, it might result in cofactor imbalance.The inactivation of pfkA, edd, and eda and overexpression of glpX can redirect more carbon flux into the PPP, which can lead to excess NADPH and NADH deficiency [26].To restore cofactor balance, soluble pyridine nucleotide transhydrogenase SthA (UdhA), encoded by sthA, was introduced into the genome to enhance the conversion of NADPH to NADH, and the membranebound transhydrogenase encoded by the pntAB gene was deleted to decrease the conversion of NADH to NADPH (Fig. 6b).The guanosine titres of MQ29 (pntAB deletion), MQ30 (sthA overexpression) and MQ31 (pntAB deletion and sthA overexpression) were 88.3, 82.3 and 96.3 mg/L, respectively (Fig. 6c).Redox cofactor rebalancing failed to further improve the production of guanosine, which suggested that the NADH and NADPH levels in MQ28 might be suitable for guanosine synthesis.

Transporter engineering for guanosine accumulation
Excessive accumulation of nucleosides leads to metabolic burden on the cell and causes product-mediated feedback inhibition.Previous studies illustrated that the NupG protein of E. coli can transport extracellular nucleosides into cells, and knockout of the nupG and nupC genes can increase the uridine and cytidine titres [11,27,28].Meanwhile, overexpression of the nucleoside efflux transporters PbuE and NepI notably increased inosine secretion [29].Transporters are essential for the hyperproduction of target products.To investigate the effects of these transporters on guanosine synthesis, we deleted the nupG gene in MQ28 to obtain strain MQ32.The inactivation of the nupG gene led to a slight increase in the guanosine titre (Fig. 7).Subsequently, overexpression of nepI and BspbuE in MQ32 generated the MQ33 and MQ34 strains, respectively.The guanosine titre of MQ33 was 123.6 mg/L, which was 16.4% higher than that obtained from MQ32.Blockage and overexpression of nucleoside transporters can prevent the absorption of exogenous guanosine and facilitate the extracellular outflow of guanosine [30,31].The results demonstrated that engineering transporters can enhance the performance of chassis strains by facilitating efflux and decreasing the accumulation of the desired intracellular products.

Strengthening the guanosine synthesis pathway
The reduction reaction from GMP to IMP is catalyzed by GMP reductase encoded by guaC gene, and overexpression of guaA and guaB genes is beneficial to guanosine synthesis.To favor guanosine production, we integrated the guaAB gene into ykgH-betA locus and guaC locus to obtain strains MQ38 and MQ39.These two strains accumulated 127.0 mg/L and 134.9 mg/L guanosine, respectively (Fig. 8).These results illustrated that enhancing the synthesis pathway of guanosine is conducive to guanosine accumulation.

Guanosine production in shake flasks
The strain MQ39 was grown in shake flasks to evaluate the performance of guanosine production under fedbatch fermentation.The cells grew rapidly from the initiation of fermentation, and the OD 600 reached 11.7 at 72 h (Fig. 9).The production of guanosine gradually increased in the whole fermentation process, representing a cell growth-independent production trend.The guanosine titre reached 289.8 mg/L with a yield of 9.89 mg/g glucose after fermentation for 72 h.

Discussion
Guanosine and other purine nucleosides are crucial for various cellular physiological processes and have broad applications in the fields of antitumor/antiviral medication as well as food additives.Nevertheless, microbial production of guanosine is still a large challenge because of multiple metabolic pathway interactions and complex regulatory mechanisms.Here, we successfully constructed a microbial cell factory for efficient guanosine production by combinational metabolic engineering strategies.The engineered strains can produce 289.8 mg/L guanosine after 72 h of fermentation.
Low metabolic flux and insufficient precursor supply are the main limiting factors for synthesizing target products in microbial hosts [32].Overexpression of homologous or heterologous metabolic pathways is an effective method to enhance the metabolic flux for target product production [33].Previous methods have mainly focused on plasmid expression.The advantage of plasmid expression is that more copies of the target genes possibly lead to high expression levels.However, it is necessary to add amino acids or antibiotics to prevent plasmid loss, and the continuous addition of antibiotics may increase the cost of fermentation.In addition, a large recombinant plasmid may bring about a metabolic burden on the strain and result in growth delay, which might be unsuitable for constructing pyrimidine-producing strains [11].Plasmid loss is also easily observed in fed-bath cultivation, which results in ineffective consumption of carbon sources and reduced product titre [26].Integration of the desired metabolic pathway into the genome is a stable overexpression strategy for multiple genes [34].Chromosomal integration of multiple genes possesses several advantages, such as achieving stable gene expression, avoiding vector capacity limitations and plasmid incompatibility.For example, the pyrimidine biosynthetic operon from B. subtilis was introduced into the yghX locus of E. coli with the help of two exogenous protospacer and protospacer-adjacent motifs and two corresponding gRNA plasmids [11].We divided the Bspur operon into four parts and integrated them into different neutral sites using the CRISPR/Cas9-mediated homologous recombination method.The introduced Bspur operon was used to enhance the enzymatic activities of purine nucleotide synthesis pathway.However, we could not detect the accumulation of guanosine in MQ4 (Fig. 2).In addition, the reaction of R5P to PRPP is considered a rate-limiting step for purine and pyrimidine nucleoside synthesis.prs overexpression also did not lead to detectable guanosine accumulation.Unexpectedly, the titre of inosine was significantly increased by overexpression of the pur operon and prs, indicating that high purine metabolic flux is directed towards inosine synthesis in cells (Fig. 3).Subsequently, the accumulation of guanosine was observed by disrupting degradation-related genes (∆deoD, ∆ppnP and ∆gsk), and MQ19 obtained 41.5 mg/L guanosine (Fig. 4).These results indicated the necessity of blocking product catabolism pathways for the target products, which was in accord with previous reports [11,35].Notably, because guanosine synthesis involves multiple precursors, enhancing the supply of other precursors, such as glutamine and aspartate, should be taken into consideration in the future.
Metabolic flux redistribution is an effective strategy to enhance chemical synthesis.In this study, downregulating the flux of adenosine synthesis and the EMP and ED pathways drove more metabolic flux into guanosine biosynthesis, which illustrated a synergistic effect on guanosine production (Figs. 5 and 6).Optimization of central carbon metabolism can enhance the availability of precursors, thereby improving the titre, productivity and yield of the biosynthetic target compounds and enhancing the performance of engineered strains.This strategy has been used to enhance the titre of other chemicals, including nucleosides, terpenoids and fatty acid derivatives, amino acids, organic acids and natural products [15,[36][37][38][39][40].For example, the titre of inosine was significantly improved by overexpressing the key enzyme Zwf and blocking two essential backflow nodes including the purine synthesis pathway towards the PPP, and from the PPP to the glycolysis [15].A previous study illustrated that deleting the pfkA, edd and eda genes can result in an approximately 11-fold increase in riboflavin titre in shake flasks [41].The fbp gene overexpression also led to a shift in metabolic flux from the glycolysis to the PPP and an increase in riboflavin titre [26].Alternatively, we can finetune metabolic pathway through promoter replacement with different strengths, inducible promoters and quorum sensing system to dynamically regulate key metabolic nodes to increase guanosine production.Further engineering should be focused on balancing biomass and products by controlling the carbon flux of glucose into the glycolysis.

Conclusions
In summary, guanosine production in E. coli engineered strains was progressively increased by combinatorial metabolic engineering strategies, which included overexpression of the Bspur operon and prs gene, blockage of the guanosine degradation pathway, downregulation of purA expression, elimination of feedback and transcription dual inhibition, redirection of the metabolic flux of the EMP and ED pathways, redox cofactor rebalancing and engineering of transporters and strengthening the guanosine synthesis.Ultimately, the final optimized strain MQ39 produced 289.8 mg/L guanosine.Combinatorial metabolic engineering strategies would be beneficial to further engineer E. coli to act as an excellent chassis strain for industrial guanosine production in the future.

Fig. 2
Fig. 2 Overexpressing the purine synthesis operon from B. subtilis.(a) The native metabolic pathway of purine biosynthesis in E. coli; (b) Structural gene diagram of the purine operon in B. subtilis; (c) Integrated expression cassettes of the purine operon; (d) Production of hypoxanthine and inosine by overexpressing the purine synthesis operon.All data represent the mean ± s.d.(n = 4 biologically independent samples).Error bars were analysed by Student's t test (two-sample, two-tailed; ** p value < 0.01, *** p value < 0.001)

Fig. 3
Fig. 3 Overexpressing the prs gene for guanosine production.(a) Schematic of the PRPP synthesis pathway in E. coli; (b) Integrated expression cassettes of the prs gene; (c) Production of hypoxanthine and inosine by overexpressing the prs gene.All data represent the mean ± s.d.(n = 4 biologically independent samples).Error bars were analysed by Student's t test (two-sample, two-tailed; ** p value < 0.01, *** p value < 0.001)

Fig. 4 Fig. 5
Fig. 4 Disruption of guanosine catabolism-related genes.(a) The related pathways of guanosine catabolism in E. coli.(b) Production of guanosine by deleting guanosine catabolism-related genes.All data represent the mean ± s.d.(n = 4 biologically independent samples).Error bars were analysed by Student's t test (two-sample, two-tailed; *** p value < 0.001)

Fig. 6
Fig. 6 Redistributing the metabolic flux of EMP and ED and redox cofactor rebalancing for the accumulation of guanosine.(a) Schematic of the EMP and ED pathways; (b) Schematic of redox cofactor rebalancing; (c) Increased guanosine production by downregulating the metabolic flux of EMP and ED and redox cofactor rebalancing.All data represent the mean ± s.d.(n = 4 biologically independent samples).Error bars were analysed by Student's t test (twosample, two-tailed; ns represents no significant difference)

Fig. 8
Fig. 8 The effect of engineering the guanosine synthesis pathway.(a) Schematic of guanosine synthesis pathway; (b) Effects of strengthening the guanosine synthesis pathway on guanosine accumulation.All data represent the mean ± s.d.(n = 4 biologically independent samples).Error bars were analysed by Student's t test (two-sample, two-tailed; *p value < 0.05)

Fig. 9
Fig. 9 Assessment of guanosine production using the strain MQ39.Fed-batch fermentation was performed in shake flasks.Four biological replicates were performed, and the error bars indicate the standard deviation

Table 1
Strains and plasmids used in this study