Metabolic engineering of Escherichia coli strains for the in vivo phosphorylase-mediated synthesis of disaccharides

Abstract

Background

Enzymatic synthesis of rare disaccharides by reverse-phosphorolysis is a potentially sustainable route to produce high-value glycosides for human health and nutrition. In vivo synthesis using microorganisms expressing recombinant glycoside phosphorylases is an attractive strategy, as it may overcome key limitations associated with conventional chemical and enzymatic approaches. However, successful implementation requires dedicated metabolic chassis enabling efficient precursor supply and carbon flux redirection. In Escherichia coli, inactivation of the phosphotransferase system (PTS) combined with deletion of pfkA constitutes a promising strategy to decouple sugar uptake from their concomitant phosphorylation, promote the accumulation of key sugar phosphates, and enable reverse biosynthetic pathways for oligosaccharide production.

Results

We first established β-1,2-mannobiose production as a proof of concept. The expression of heterologous permease GalP restored the growth on mannose in a PTS⁻ background and allowed mannose into the reverse biosynthetic pathway. Deletion of pfkA, which promotes intracellular accumulation of glucose 1-phosphate (G1P) and mannose 1-phosphate (M1P), combined with the expression of a β-1,2-mannoside-phosphorylase, enabled the direct synthesis of β-1,2-mannobiose from mannose. The pfkA deletion significantly increased product titers (> 0.6 g·L⁻¹) and yields (up to 9% g/g mannose), reflecting an efficient redistribution of carbon fluxes toward disaccharide formation. Furthermore, mixed-substrate cultures using glycerol as the carbon and energy source, together with additional metabolic engineering, enabled partial growth–production decoupling and redirected mannose utilization primarily toward product synthesis, achieving yields of up to 60%. The same metabolic chassis was further employed for the production of β-1,4-mannobiose and laminaribiose through the expression of a β-mannoside-phosphorylase and a laminaribiose-phosphorylase, using mannose and glucose as substrates, respectively.

Conclusion

Here, we report the metabolic engineering of Escherichia coli for in vivo production of oligosaccharides with different glycosidic linkages from hexose sugars. These results demonstrate the modularity and efficiency of the proposed platform for fermentative production of non-conventional oligosaccharides and expand the scope of metabolic engineering strategies for glycoside biosynthesis.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12934-026-03051-5.

Introduction

Oligosaccharides are a highly diverse class of biomolecules consisting of linear or ramified chains of glycosyl moieties or glycosyl derivatives. Oligosaccharides play a multitude of biological roles in cellular processes, the modulation of immune response and host-microbe interactions [1], and have thus found a broad range of applications in the food (functional foods, prebiotics or as additives) and health (as drugs, drug carriers or vaccine intermediates) industries [2–7], with ongoing scientific and economic interest in improving their production. Oligosaccharides can be produced enzymatically and chemo-enzymatically by exploiting the natural diversity and selectivity of enzymes to enlarge the panel of glycan structures that can be produced, avoiding multiple chemical synthesis steps [8–11]. Four main types of enzymes have been used to efficiently catalyze oligosaccharide production: glycoside hydrolases (GHs) and glyco-synthases, transglycosylases (TGs), glycosyltransferases (GTs) and glycoside phosphorylases (GPs) [12]. GPs are reversible enzymes that catalyze both the breakdown of glycosidic linkages in the presence of phosphate (in what is commonly referred to as the phosphorolysis reaction) and the synthesis of glycosides from a glycosyl-phosphate and an acceptor (the reverse phosphorolysis reaction) [13]. High-yield glycoside synthesis using GPs has been successfully achieved in vitro, as exemplified by the production of 2-O-α-D-glucosyl glycerol [14]. However, regardless of the classes and mechanisms involved, in vitro use of glycoside synthesizing enzymes remains constrained by enzyme production and purification costs, as well as by losses in enzymatic activity during storage and under reaction conditions. These limitations can partly be alleviated through whole-cell bioprocesses, as illustrated by the particularly efficient GP-based production systems developed in Lactobacillus paracasei [15] and Corynebacterium glutamicum [16] for the synthesis of 2-O-α-D-glucosyl glycerol. Nevertheless, in both in vitro and whole-cell approaches, substrate costs remain a major limitation for large-scale glycoside production. Specifically, glycosyltransferase-based approaches require expensive nucleotide-sugars and glycosynthase-based methods require chemically modified glycosyl donors that are difficult to produce with high yields at a reasonable cost. Sucrose synthases (SuSy) have been extensively used over the past decades in combination with GTs to enable efficient in vitro glycoside synthesis. In these so-called SuSy/GT cascades, a wide range of glycosides can be produced at reduced cost, as the NDP-sugars required by GTs are continuously regenerated from inexpensive sucrose and NDP via sucrose synthase [17].

Besides, in vivo synthesis using microorganisms that express recombinant glycoside synthesizing enzymes has therefore attracted attention for its potential to circumvent substrate cost limitations. This approach is based on the ability of the engineered cells to produce the precursor substrates (activated sugars) from affordable carbon sources as well as the catalysts and the targeted products, leading to a more economically sustainable process. Microbial production has proved particularly popular in recent years for human milk oligosaccharide production, with GTs as biocatalysts [18, 19]. The use of other catalysts has been far more limited however; for instance, GPs have only been used in vivo alongside GTs [20], in the phosphorolysis direction, to increase the pool of sugar phosphates through the phosphorolysis of low-cost oligosaccharides, the sugar phosphates being themselves transformed into nucleotide sugars, the substrates of GTs. Nonetheless, in vivo reverse-phosphorolysis has been reported in Mycobacterium smegmatis strains, which can use native GPs for the synthesis of energy storage glycosides such as glycogen from sugar phosphates [21]. Recently, Tian et al. [22] reported the production of manno-oligosaccharides in a mannose fermentation system using an engineered strain of Corynebacterium glutamicum, achieving approximatively 1.7 g/L mannobiose and 5.4 g/L mannotriose from 20 g/L mannose.

To date however, oligosaccharide synthesis through GP-based reverse-phosphorolysis has yet to have been achieved with E. coli, the preeminent host for oligosaccharide production using GTs [23]. One explanation for this is the low intracellular availability of the substrates in E. coli. For oligosaccharide synthesis indeed, GPs require sugar phosphates as donors and non-phosphorylated sugars as acceptors, and while the former are abundant in E. coli cells, non-phosphorylated monosaccharides are lacking in native strains because most carbohydrates are internalized by the phosphoenolpyruvate–carbohydrate phosphotransferase system (PTS) [24], preventing the use of GPs in the reverse-phosphorolysis direction.

In this study, we tackled the problem of limited precursor availability in E. coli using a metabolic engineering strategy focused on oligosaccharide production via GPs. Central to this strategy is the decoupling of sugar import from phosphorylation through the replacement of the native PTS (which internalizes the carbon source and monosaccharide acceptors) by a permease-based transport mechanism. Further deletions in central metabolism lead to the accumulation of the sugar phosphate intermediates required to produce oligosaccharides. Further engineering allowed the monosaccharide acceptor flux to be channeled almost completely toward oligosaccharide synthesis, using glycerol as carbon and energy source. As proof of concept of this approach, we report the production of various oligosaccharides of interest for nutrition and health, in particular β-1,2-mannobiose, an antigenic motif of the pathogenic yeast Candida albicans which is very difficult to chemically synthesize cost-effectively at large-scale [25], hindering the development of candidiasis vaccines.

Materials and methods

Chemicals

β-1,4-mannobiose and laminaribiose were purchased from Biosynth (Bratislava, Slovakia). β-1,2-mannobiose was purchased from Omicron Biochemicals (South Bend, IN, USA). LC-MS grade solvents (methanol, acetonitrile) were obtained from Instrumentation Consommables et Service (ICS, Lapeyrousse-Fossat, France).

Strain construction

Strain MDO, constructed from strain ZLKA, was provided by Dr Éric Samain (CERMAV, Grenoble, France). Strain ZLKA is a derivative of Escherichia coli K12 strain DH1 (endA1 recA1 gyrA96 thi-1 glnV44 relA1 hsdR17), constructed by disrupting genes lacZ, nanKETA, and lacA as previously described [26]. Strain MDO involves three additional mutations (melA, wcaJ, mdoH) and insertion of a Plac promoter upstream of gmd (encoding for GDP-mannose 4,6-dehydratase) (Samain laboratory, unpublished research). As a result, the manB gene for phosphomannomutase, which is located downstream from the gmd gene, is inducible by ITPG. The complete genotype of the parental MDO strain is: endA1 recA1 gyrA96 thi-1 glnV44 relA1 hsdR17 lacZ- wcaF::Plac nanKETA lacA melA wcaJ mdoH.

Selected genes (ptsG, manXYZ, and pfkA) were deleted using the PKO3 gene knock-out protocol [27]. PKO3 plasmid was obtained from the Church laboratory (Harvard Medical School). Briefly, flanking regions upstream and downstream of genes selected for deletion were amplified by PCR and cloned together into the BamHI SalI sites of the suicide vector pKO3. For ptsG, the upstream fragment was cloned with primers P1 and P2 and the downstream fragment with primers P3 and P4 (see Supplementary Table 1), resulting in the deletion of a 351 kb DNA segment between nucleotides 568 and 919 of ptsG. For manXYZ, the upstream fragment was cloned with primers P5 and P6 and the downstream fragment with primers P7 and P8, resulting in the deletion of a 1,219 kb segment between nucleotides 332 of manX and 522 of manY. For pfkA, the upstream fragment was cloned with primers P9 and P10 and the downstream fragment with primers P11 and P12, resulting in deletion of a 396 kb DNA segment between nucleotides 146 and 542 of pfkA.

The genes maa and manA were deleted using the Datsenko and Wanner protocol [28]. Briefly, a DNA cassette containing the gene for kanamycin resistance and flanked by Flippase Recognition Target (FRT) sites was amplified using the plasmid pKD4 as template, with 50 bp of the 5’ and the 3’ ends of the gene at the extremities. For maa and manA, the primer couples were P13/P14 and P15/P16, respectively.

The resulting strains were transformed with plasmid pKD46 at 30 °C on LB agar plates supplemented with ampicillin at 50 µg/mL. Transformants carrying pKD46 plasmid were grown in 5-mL LB cultures with ampicillin and L-arabinose (0.2% w/v) at 30 °C to an optical density at 600 nm (OD₆₀₀) of 0.6 and then made electrocompetent by concentrating 100-fold and washing three times with ice-cold 10% glycerol. Electroporation was done using a Cell-Porator device with a voltage booster and 0.1 cm chambers according to manufacturer instructions, with 50 µL of cells and 200 ng of PCR product. Shocked cells were added to 1 mL of SOC medium, incubated for 2 h at 37 °C, and then spread onto agar to select kanamycin transformants. Once colonies lacking the gene had been identified, the cassette was removed by transforming one of the positive colonies with the plasmid pCP20. This plasmid is an ampicillin resistance plasmid with temperature-sensitive replication and thermal induction of FLP synthesis. KmR mutants were transformed with pCP20, and ampicillin-resistant transformants were selected at 30 °C, after which a few were colony-purified once, non-selectively, at 43 °C and then tested for loss of all antibiotic resistances.

All PCR reactions were performed with NEB phusion DNA polymerase according to manufacturer instructions, adapting cycles to primers’ melting temperatures and fragment lengths.

Plasmid construction

Cloning of monosaccharide transporter

A 1,471 kb DNA fragment containing the sequence of the galP gene was amplified by PCR using the genomic DNA of Escherichia coli K12 as a template and primers P17 and P18. The amplified fragment was then cloned into the SalI and XbaI sites of the pWKS130 expression vector to form pWKS-GalP [29].

Cloning of GPs

The genes encoding the Teth514-1788 protein from Thermoanaerobacter sp. strain X-514 (Teth514-1788, GenBank accession number ABY93073.1 [30]), a β-1,2-mannoside-phosphorylase belonging to the GH130 family in the CAZy classification (http://www.cazy.org/) [31], and the gene encoding the ACL0729 protein, a laminaribiose-phosphorylase from Acholeplasma laidlawii strain PG-8 A (ACL0729, GenBank accession number ABX81345.1 [32]), belonging to the GH94 family in the CAZy classification (http://www.cazy.org/) [31], were synthesized and cloned into the pBAD HisA vector by GeneCust (GeneCust, Boynes, France) between the NcoI and XhoI restriction sites.

The gene encoding the Uhgb_MP β-1,4-mannoside-phosphorylase from an uncultured human gut bacterium, belonging to the GH130 family (Uhgb_MP, GenBank accession number ADD61463.1 [33]), was cloned into the pBAD HisA vector using the primers P19 and P20.

Cloning was performed according to manufacturer’s instructions (Invitrogen Corp., Carlsbad, CA, USA). All plasmid constructs were verified by sequencing, and protein expression profiles were assessed by SDS-PAGE electrophoresis using Any kD gels from Bio-Rad as follows. Cell pellets were lysed in 5 mM Tris-HCl buffer containing lysozyme (Euromedex, ref. 5934-C) at 0.5 mg·mL⁻¹ and DNase (NEB, ref. M0303L) at 50 µL (100 U) per 100 mL of 5 mM Tris-HCl buffer. All pellets were resuspended in buffer to an OD₆₀₀ of 45, incubated for 30 min at 37 °C, and then frozen at − 20 °C. Samples were thawed and centrifuged for 25 min at 5,000 g. Fifteen microliters of each supernatant were mixed with 5 µL of blue loading dye (NEB, ref. 50994905), incubated for 10 min at 95 °C, and then loaded onto the gel. Electrophoresis was carried out for 35 min at 100 V in an electrophoresis tank to confirm that the proteins were indeed produced in the transformed PFKA1 strain and had the expected molecular masses.

Media and growth conditions

Escherichia coli strains C for DNA manipulation and strain construction were grown at 37 °C or 42° or 30 °C in LB media (10 g·L^− 1 tryptone, 5 g·L^− 1 yeast extract, 10 g·L^− 1 NaCl) as indicated. All other growth strains were grown in minimal medium M9. M9 salts: Na₂HPO₄ 12H₂O 17.4 g·L^− 1, KH₂PO₄ 3.02 g·L^− 1, NaCl 0.51 g·L^− 1, NH₄Cl 2.04 g·L^− 1. Trace metals salts: Na₂EDTA 2 H₂O 15 mg·L^− 1, ZnSO₄ 7 H₂O 4.5 mg·L^− 1, CoCl₂ 6H₂O 0.3 m g·L^− 1, MnCl₂ 4H₂O 1 mg·L^− 1, H₃BO₃ 1 mg·L^− 1, Na₂MoO₄ 2 H₂O 0.4 mg·L^− 1, FeSO₄ 7 H₂O 3 mg·L^− 1, CuSO₄ 5 H₂O 0.3 mg·L^− 1. MgSO₄ 0.5 g·L^− 1 CaCl₂ 4.38 mg·L^− 1, Thiamine hypochloride 0.1 g·L^− 1.

E. coli strains were grown in M9 minimal medium complemented with the desired carbon sources. Cultures were performed in duplicate or triplicate in shake-flasks at 37 °C and 200 rpm, using a working volume of 250 mL containing 3 g·L⁻¹ of selected carbon source. Growth was monitored by measuring the OD₆₀₀ using a Genesys 6 spectrophotometer (Thermo, USA). Kanamycin and ampicillin were added at a final concentration of 50 µg·mL^− 1 to ensure propagation of plasmids. Isopropyl β-D-1-thiogalactopyranosid (IPTG - final concentration 60 µM) and arabinose (final concentration 1 mM) were added to the media to induce protein expression. For bioreactor experiments, the concentrations of M9 salts were changed to: KH₂PO₄, 3.02 g·L^− 1; NaCl, 0.51 g·L^− 1; NH₄Cl, 2.04 g·L^− 1; (NH₄)₂SO₄, 5 g·L^− 1. All the other components were as described above. Bioreactor cultures were performed in 500 ml of M9 medium complemented with desired carbon source. Parameters were set and monitored using a Multifors bioreactor system (Infors, Switzerland). The percentages of CO₂ and O₂ in the gas output were determined using a Dycor ProLine Process Mass Spectrometer (Ametek, DE, USA).

Metabolomics analysis

Sampling and metabolite extraction

Samples (100 µL) of an overnight culture of liquid LB with the desired strains were used to inoculate 50 mL baffled shake flasks containing 10 mL of M9 medium supplemented with 3 g·L^{− 1} of glucose. The flasks were incubated for 24 h at 37 °C with orbital shaking at 220 rpm. Cells were harvested by centrifugation (Sigma, Seelze, Germany) for 10 min at 2,000 g at room temperature, washed with five times diluted M9 and used to inoculate 50 mL baffled shake flasks at OD₆₀₀ = 0.1 containing 50 mL of diluted M9 supplemented with 3 g·L^{− 1} of glucose and incubated at 37 °C at 220 rpm. Cell growth was monitored by measuring the OD₆₀₀ with a cell density meter (Ultrospect 10, Amersham BioSciences). Sampling was performed using the fast filtration method [25]. When cultures reached the exponential phase, (OD₆₀₀ of 1–1.5) 120 µL of broth was harvested by vacuum filtration (Sartolon polyamide 0.2 μm). Cells were washed with 1 ml of medium with five times reduced concentrations of phosphate and sulfate salts and of carbon source. All the other components of the medium were unchanged in the washing solution. The filters were rapidly placed in liquid nitrogen and stored at − 80 °C until further metabolite extraction. For metabolite extraction, the filters were incubated in closed glass tubes containing 5 ml of a precooled acetonitrile/methanol/H2O (4:4:2) solution and 60 µL of fully ¹³C-labeled cell extract were added to each tube as internal standard. The tubes were incubated for 1 h at − 20 °C, the filters were removed, and the supernatants were evaporated under vacuum (SC110A SpeedVac Plus, ThermoSavant, Waltham, MA, USA) for 12 h and then stored at − 80 °C until further treatment.

IC-ESI-HRMS analysis of central metabolites

Samples were dissolved in 120 µL of mQ water and then centrifuged at 10,000 g at 4 °C for 10 min to remove cell debris. The supernatant was then analyzed by ion chromatography (Thermo Scientific Dionex ICS-5000 + system, Dionex, Sunnyvale, CA, USA) mass spectrometry, on LTQ Orbitrap device (Thermo Fisher Scientific, Waltham, MA, USA) equipped with an electrospray ionization probe. The ion chromatography method was adapted from Kiefer et al. [34]. The KOH gradient was modified as follows: 0 min, 0.5 mM; 1 min, 0.5 mM; 9.5 min, 4.1 mM; 12.5 min, 30 mM; 24 min, 50 mM; 36 min, 60 mM; 36.1 min, 90 mM; 43 min, 90 mM; 43.5 min, 0.5 mM; 48 min, 0.5 mM. The gradients were all linear. A Dionex anionic electrolytically regenerated suppressor (300–2 mm) was used for background suppression. The device was operated at a constant electrolysis current of 87 mA, in external-water mode with ultrapure water, and regenerant was delivered using an external AXP pump at a flow rate of 1 mL/min. The volume of the injected sample was 35 µL in all cases. Mass spectrometry analysis was performed in the negative FTMS mode at a resolution of 30,000 in fullscan mode, with the following source parameters: capillary temperature, 350 °C; source heater temperature, 300 °C; sheath gas flow rate, 50 a.u. (arbitrary units); auxiliary gas flow rate, 5 a.u; S-Lens RF level, 60%; and ion spray voltage, 3.5 kV. The data were acquired using the Xcalibur software and processed using the software TraceFinder 3.1 (Thermo Fisher Scientific, Waltham, MA, USA). 6-phosphogluconate (6PG) fructose 6-phosphate (F6P), fructose 1,6-bisphosphate (FBP), Glucose 6-phosphate (G6P), combined pools of glucose 1-phosphate and mannose 1-phosphate (G1P/M1P), glucosamine 6-phosphate (GlcN-6P), N-Acetylglucosamine 1-phosphate (GlcNAc-1-p), combined pools of xylulose 5-phosphate, ribose 5-phosphate, ribulose 5-phosphate (P5P), phosphoenolpyruvate (PEP) and sedoheptulose 7-phosphate (S7P) were injected as pure standards, and the isotope dilution mass spectrometry (IDMS) method was used to ensure accurate quantification [35]. The absolute intracellular concentration for each metabolites in the different strains was calculated from these calibration curves, assuming a cell volume of 1.77 ml/gDW [36].

NMR analysis

Exocellular metabolites were identified and quantified by nuclear magnetic resonance (NMR). Broth samples were collected at different times and centrifuged. The supernatants (500 µL) corresponding to the culture medium, were mixed with 100 µL of D₂O containing 2.35 g·L^{− 1} of TSPd4 (3-(trimethylsilyl)-1-propanesulfonic acid-tetra deuterated) as internal compound. Proton NMR spectra were recorded on an Avance 500 MHz NMR spectrometer equipped with a 5 mM BBI probe head (Bruker, Rheinstatten, Germany). Spectra processing and metabolite quantification were performed using Topspin 3.1 (Bruker, Rheinstatten, Germany). The concentrations of extracellular metabolites were determined using TSP as internal standard [37].

HPAEC-PAD analysis

For carbohydrate analysis using high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD), culture supernatants (1 mL) were heated at 95 °C for 10 min, centrifuged for 10 min at 15,000 g, and the resulting supernatants were passed through a 0.22 μm membrane filter. Filtered samples were diluted 25-fold in ultrapure water prior to chromatographic analysis. Carbohydrate and related compounds were separated on a Dionex CarboPac PA100 column (2 × 250 mm; Thermo Fisher Scientific) using a 0.25 mL∙min⁻¹ elution gradient with the following eluents: A, H₂O; B, 150 mM NaOH; and C, 150 mM NaOH supplemented with 500 mM sodium acetate. The gradient program was: 0–3 min, 50% A/50% B; 3–6 min, 100% B; 6–12 min, 50% B/50% C; 12–15 min, 5% B/95% C; followed by re-equilibration at 50% A/50% B from 15 to 23 min. Detection was carried out using a Dionex ED40 electrochemical detector equipped with a gold working electrode and an Ag/AgCl reference electrode.

Results and discussion

Metabolic engineering strategy for mannobiose bioproduction

The strategy devised to enable in vivo reverse-phosphorolysis by GPs for mannobiose production in E. coli is presented in Fig. 1. To increase the intracellular availability of unphosphorylated mannose, the native PTS, responsible for coupled uptake and phosphorylation of glucose and mannose, was inactivated by deleting the ptsG and manXYZ genes. However, since this severely impairs growth on glucose and mannose [38, 39], the galP gene encoding galactose permease was overexpressed from a plasmid. Importantly, GalP facilitates the import of both mannose [40] and glucose [38] in their unphosphorylated forms, allowing cytosolic accumulation of free hexoses. Mannose is subsequently phosphorylated to mannose 6-phosphate (M6P), presumably by glucokinase (glk) [41]. From this branch point, M6P can either be converted to fructose-6-phosphate (F6P) by mannose 6-phosphate isomerase (ManA), thereby entering central carbon metabolism, or be converted directly to mannose 1-phosphate (M1P) by phosphomannomutase (ManB), generating the activated donor substrate required for reverse-phosphorolysis. To further favor M1P accumulation, the concentrations of phosphorylated intermediates were altered by deleting the phosphofructokinase gene pfkA, thereby modulating intracellular sugar phosphate pools. The assumption underlying this channeling of the carbon flux toward mannobiose biosynthesis is that the disaccharide product is efficiently exported, minimizing feedback inhibition and enabling sustained production.

Fig. 1.

Metabolic engineering of E. coli for the synthesis of mannobiose by glycoside phosphorylases. The strategy involves several gene deletions (indicated in red) and overexpressions (indicated in green). The relative concentrations of intracellular metabolites expected to change upon pfkA deletion are indicated by green arrows (for increases) or red arrows (decreases). GalP: galactose permease; PTS: phosphotranferase system; manXYZ: mannose specific PTS enzyme II; ptsG: glucose specific PTS enzyme IIBC; glk: glucokinase; manA: phosphomannose isomerase; pgi: phosphoglucose isomerase; manB (cpsG): phosphomannomutase; pfkA: phosphofructokinase I; zwf: glucose 6-phosphate dehydrogenase; cpsB: mannose 1-phosphate guanylyltransferase; GP: glycoside phosphorylase (mannoside phosphorylase); M6P: mannose 6-phosphate; F6P: fructose 6-phosphate; G6P: glucose 6-phosphate; M1P: mannose 1-phosphate; FBP: fructose 1:6-biphosphate: GDP-Man: GDP-mannose; CCM: central carbon metabolism. * The mannobiose export system is unknown

Monosaccharide import through galactose permease in PTS mutant strains

Our approach to mannoside production requires the import of unphosphorylated mannose into the cytoplasm to serve as a substrate (acceptor) for the recombinant mannoside-phosphorylases expressed from plasmids in the chassis strains (Fig. 1). We therefore inactivated the PTS, which normally imports and phosphorylates mannose. In E. coli however, while the mannose PTS is the primary route for mannose uptake, the glucose PTS can also (less efficiently) transport mannose [42]. We therefore used the MDO strain [43] (Table 1), based on E. coli K-12 strain DH1, previously engineered for oligosaccharide production. This strain has deletions in several galactosidase-encoding genes (lacZ, lacA, melA) and in genes (nanATEK, wcaJ, opgH) involved in the biosynthesis of non-essential oligosaccharides that could interfere with mannoside production. It also has a Plac promoter upstream of the gmd gene, enabling IPTG-inducible overexpression of downstream genes, including manB (cpsG), which encodes a phosphomannomutase responsible for the reversible conversion of mannose 6-phosphate (M6P) into mannose 1-phosphate (M1P).

Table 1.

List of strains and plasmids used in this study

Strains	Description	Origin
MDO	Escherichia coli K-12 (DH1) ΔlacZ wcaF::Plac ΔnanATEK ΔlacA ΔmelA ΔwcaJ ΔmdoH	Samain laboratory
MGX	Derived from MDO. ΔptsG ΔmanXYZ	This Study
PFKA	Derived from MGX. ΔpfkA	This Study
PFMA	Derived from PFKA. Δmaa	This Study
PFMM	Derived from PFKA2. ΔmanA	This Study
Plasmids
Pko3	Suicide vector for knock-out generation	[27]
Pko3-ptsg	Suicide vector for ptsG deletion	This Study
Pko3-manxy	Suicide vector for manXY deletion	This Study
Pko3-pfka	Suicide vector for pfkA deletion	This Study
Pwks	medium copy plasmid carrying Plac promoter	[29]
Pwks-galp	medium copy plasmid expressing galactose permease	This Study
pKD4	Medium copy plasmid containing the KAN cassette for the Datsenko protocol	[28]
Pbad-his-Uhgb_MP	medium copy plasmid expressing the β-1,4-mannoside-phosphorylase Uhgb_MP	This Study
Pbad-his-teth	medium copy plasmid expressing the β-1,2-mannoside-phosphorylase Teth514_1788	This Study
Pbad-his- ACL0729	medium copy plasmid expressing the laminaribiose-phosphorylase ACL0729	This Study

It is well established that PTS transporters are the main routes for glucose and mannose uptake in E. coli. These systems mediate the simultaneous import and phosphorylation of sugars, using phosphoenolpyruvate (PEP) as the phosphate donor. To promote the uptake of unphosphorylated mannose, we deleted ptsG and manXYZ, which respectively encode the IIB subunits of glucose and mannose specific PTS enzyme II, in the MGX strain (Table 1). However, PTS deletion severely impairs growth, despite the presence of alternative, less efficient uptake systems [44]. To evaluate the impact of these deletions, we compared the growth of parental MDO and MGX strains in M9 minimal medium supplemented with either glucose or mannose as sole carbon source (Table 2).

Table 2.

Growth rates (µ) and specific uptake rates (qs) for the studied E. coli strains grown on glucose and mannose

Strains	Glucose		Mannose
	µ (h− 1)	qs (mmol.gcdw−1.h− 1)	µ (h− 1)	qs (mmol.gcdw−1.h− 1)
MDO	0.47 ± 0.04	6.14 ± 1.02	0.26 ± 0.01	2.05 ± 0.09
MGX	0.15 ± 0.01	1.11 ± 0.20	0.10 ± 0.01	1.31 ± 0.07
MGX1	0.26 ± 0.02	5.27 ± 1.34	0.20 ± 0.01	1.90 ± 0.09
PFKA1	0.20 ± 0.01	2.73 ± 0.43	0.14 ± 0.01	1.09 ± 0.06

As expected, the MGX strain grew roughly three times more slowly on both sugars. To restore sugar uptake and growth, we introduced a plasmid (pWKS) expressing the galP gene, encoding E. coli galactose permease under the control of the Plac promoter, into the MGX strain, generating MGX1. Consistent with previous reports, induction of recombinant GalP partially restored glucose uptake and growth [38]: on glucose (mannose), the growth rate of the MGX1 strain was roughly 55% (75%) of the MDO strain’s (Table 2). Despite the efficient restoration of sugar uptake and growth by GalP in the PTS⁻ background, further improvements in substrate uptake efficiency and overall strain performance may be achieved through the exploration of alternative or optimized sugar transporters.

Deletion of phosphofructokinase A leads to accumulation of intracellular hexose-phosphate

Inactivation of the PTS leads to the accumulation of unphosphorylated monosaccharides in the cytoplasm, making them available as substrates for GPs involved in reverse-phosphorolysis. To further elevate intracellular levels of hexose 1-phosphates, the glycosyl donor substrates required by GPs, we deleted the pfkA gene, which encodes the glycolytic enzyme phosphofructokinase A. This enzyme catalyzes the conversion of fructose 6-phosphate (F6P) to fructose 1,6-bisphosphate (FBP). Deletion of pfkA is known to result in the intracellular accumulation of glucose 6-phosphate (G6P) and F6P [45], which should in turn increase the availability of glucose 1-phosphate (G1P) and mannose 1-phosphate (M1P) for GP-mediated synthesis.

Restoration of growth in the PFKA strain via GalP expression

The PFKA strain, generated by deleting pfkA from the MGX1 strain, showed no detectable growth after three days in minimal medium with glucose or mannose as sole carbon source. Introduction of a plasmid (pWKS) expressing the galactose permease gene (galP), yielding the PFKA1 strain, restored glucose uptake and partially recovered growth in minimal medium with glucose (growth rate: 0.20 ± 0.01 h⁻¹, less than half that of the MDO strain; see Table 2). On mannose, PFKA1 had a growth rate of 0.14 ± 0.01 h⁻¹, indicating viability, albeit with slower growth.

Overall, it can be deduced from Table 2 that sugar uptake rates and growth do not vary proportionally across the different strains, suggesting that carbon utilization efficiency is not solely governed by substrate uptake, but is likely shaped by the intracellular metabolic state. This was further investigated through intracellular metabolite concentration profiling, which also aimed to assess the potential accumulation of phosphorylated sugar intermediates.

Intracellular metabolite profiles of the MDO, MGX1, and PFKA1 strains on glucose and mannose

The results of metabolomic analyses of MDO, MGX1, and PFKA1 strains grown on either glucose or mannose as sole carbon source are presented in Fig. 2.

Fig. 2.

Absolute intracellular concentrations of phosphorylated sugars in strains grown with glucose or mannose as the sole carbon sources. Values are expressed in µmol·gDW^− 1 and represent the mean of two independent experiments; standard deviations are indicated as error bars. Data from glucose- and mannose-based cultures are shown in blue and green boxes, respectively. In each box, the left bar corresponds to the MDO strain, the middle bar to strain MGX1, and the right bar to strain PFKA1. GalP: galactose permease; PfkA: phosphofructokinase I; Metabolites: GLC: glucose; Man: mannose; G1P: glucose 1-phosphate; M6P: mannose 6-phosphate; F6P: fructose 6-phosphate; G6P: glucose 6-phosphate; M1P: mannose 1-phosphate; F1:6BP: fructose 1:6-bisphosphate; 6-PGNL: 6-phosphogluconolactone: 6-PG: 6-phosphogluconate; P5P: pentose 5-phosphate; E4P: erythrose 4-phosphate; S7P: sedoheptulose 7-phosphate; tkt-C2: C2 fragment bound to transketolase ; tal-C3: C3 fragment bound to transaldolase; G3P: glyceraldehyde 3-phosphate; PEP: phosphoenolpyruvate; GlcN-6P: glucosamine 6-phosphate; GlcNac-6P: N-acetylglucosamine 6-phosphate

On glucose (Fig. 2, blue box), the intracellular metabolite profiles of MGX1 were similar to those of MDO, as expected, since deletion of the PTS transporter in MGX1 is not expected to significantly impact central carbon metabolism. In contrast, the profiles of PFKA1 deviated significantly from those of the two other strains. Most notably, the intracellular concentration of fructose-1,6-bisphosphate (FBP) was twofold lower than in MDO and fivefold lower than in MGX1. This is consistent with deletion of the pfkA gene, since FBP is the direct product of the phosphofructokinase-catalyzed reaction. Correspondingly, concentrations of the upstream glycolytic intermediates glucose 6-phosphate (G6P) and fructose 6-phosphate (F6P) levels were approximately 10- and 8-fold higher than in MDO, respectively. Increases were also observed for other hexose phosphates, including a 20-fold increase in the concentration of mannose 6-phosphate (M6P), and a strongly elevated combined pool of glucose 1-phosphate (G1P) and mannose 1-phosphate (M1P), which are not singularly distinguishable by ion chromatography. Increases were also observed for other hexose phosphates, including a 20-fold increase in the concentration of mannose 6-phosphate (M6P), and a strongly elevated combined pool of glucose 1-phosphate (G1P) and mannose 1-phosphate (M1P), which are not singularly distinguishable by ion chromatography. The maximal accumulation of G1P and M1P may be constrained by endogenous E. coli phosphatases such as YihX and YidA [46]; however, these enzymes were not deleted, as unphosphorylated mannose is also required as an acceptor for the reverse-phosphorolysis reaction.

In addition, the PFKA1 strain had elevated levels of several pentose phosphate pathway (PPP) intermediates, such as 6-phosphogluconate (6-PG), pentose 5-phosphates (P5P; including ribose 5-phosphate and ribulose 5-phosphate), and in particular, sedoheptulose 7-phosphate (S7P), whose concentration was 10–20-fold higher than in MDO and MGX1. These results point to a broad re-routing of carbon fluxes. Indeed, previous studies have shown that pfkA deletion induces a metabolic shift from glycolysis toward the PPP [47], which explains the observed accumulation of PPP-related metabolites.

Similar trends in metabolite concentrations were observed when the same strains were grown on mannose as sole carbon source (Fig. 2, green box). However, a major exception was the pronounced accumulation of M6P in both the MDO and MGX1 strains, as expected, given that M6P is the first phosphorylated intermediate in mannose metabolism. In the PFKA1 strain, FBP levels were approximately tenfold lower than in MDO, consistent with the altered metabolic flux associated with the absence of PfkA. No general increase in PPP intermediates was observed however, with the exception of S7P, whose elevation remains unexplained at this stage. Notably, the accumulation of hexose phosphates in the PFKA1 strain was much less pronounced during growth on mannose than during growth on glucose. While M6P and M1P levels remained similar, G6P and F6P were slightly higher than in the MDO strain (F6P, 0.88 ± 0.11 vs. 0.64 ± 0.05; G6P, 1.75 ± 0.21 vs. 1.13 ± 0.34). Surprisingly, intracellular concentrations of several metabolites, including the G1P/M1P pool, were lower in MGX1 than in MDO, despite manB overexpression. Under mannose conditions therefore, PFKA1 accumulated higher levels of these intermediates than MGX1.

In summary, these results show that PFKA1 has an intracellular metabolite profile that differs from those of MDO and MGX1, with a strong accumulation of hexose phosphates and reduced FBP levels on glucose. These differences are less pronounced on mannose, where only FBP remains consistently low. Given MGX1’s unexpectedly lower metabolite levels on mannose, the increased accumulation of GP substrates under mannose conditions evidenced here in PFKA1 make it the most promising candidate for mannooligosaccharide production.

Strain PFKA1 produces β-1,2-mannobiose in vivo by GP-mediated synthesis

β-1,2-mannobiose production was evaluated in the double mutant PFKA1 strain lacking both PTS and PfkA. This disaccharide can be synthesized by the glycoside phosphorylase Teth514_1788 [30], a member of the GH130 family of carbohydrate-active enzymes (CAZy database, http://www.cazy.org/ [48]). The PFKA1 strain was transformed with plasmid pBAD harboring Teth514_1788 under arabinose-inducible control. The resulting strain, PFKA1-Teth514_1788, was cultivated along with the parental PFKA1 strain as a control in baffled shake flasks containing minimal medium with mannose as the sole carbon source. NMR spectra of the culture supernatants are shown in Fig. 3.

Fig. 3.

A Chemical structure of β-1,2-mannobiose, showing the anomeric protons of the reducing (ManA) and the non-reducing (ManB) mannosyl moieties (highlighted in blue), corresponding to the signals observed in the NMR spectra in (B). B Selected regions of the 1D ¹H NMR spectra of β-1,2-mannobiose corresponding to the anomeric proton signals. C HPAEC-PAD chromatograms showing that the β-1,2-mannobiose produced in the culture supernatant co-eluted with a β-1,2-mannobiose standard. Bright green lines, β-1,2-mannobiose standard; red, culture supernatant of PFKA1-Teth514_1788; blue, culture supernatant of PFKA1-control

Two well-defined peaks were detected in the spectra of the culture supernatants of the PFKA1-Teth514_1788 strain that were absent from the control samples (Fig. 3B), and whose intensity increased over the course of cultivation (data not shown), suggesting progressive secretion of the product into the medium. These signals correspond to those observed in the NMR spectrum of β-1,2-mannobiose, as previously annotated [49]. Based on their chemical shifts, the two peaks at 5.31 ppm and 4.79 ppm correspond to the α-anomeric proton (H1α) of the reducing mannosyl moiety (ManA) and to the anomeric proton of the non-reducing mannosyl moiety (ManB) respectively (Fig. 3A). These chemical shift assignments were further corroborated by high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD), which confirmed the presence of β-1,2-mannobiose in the culture supernatant through elution at the same retention times as the standard (Fig. 3C).

GP-based production of β-1,4-mannobiose and laminaribiose

The versatility of our engineered strain was tested through the production of two additional oligosaccharides, β-1,4-mannobiose and laminaribiose. For β-1,4-mannobiose, the PFKA1 strain was transformed with the plasmid pBAD-UhgbMP, encoding a β-1,4-mannoside-phosphorylase [33]. Presence of β-1,4-mannobiose in the culture supernatant was confirmed by NMR and HPAEC-PAD, with a concentration of 1.58 mM (0.54 g·L^{− 1}), and characteristic signals matching those of a commercial standard (Figure S1), demonstrating efficient production and extracellular accumulation of the disaccharide.

Laminaribiose synthesis was catalyzed by the laminaribiose-phosphorylase ACL0729 using glucose and G1P as substrates [32]. Figure S2B presents an expanded view of the spectral regions spanning 5.28–5.20 ppm and 4.76–4.67 ppm, where distinct signals are observed exclusively in the supernatant of the E. coli PFKA1-pBAD-ACL0729 strain and in a commercial laminaribiose standard, but not in the supernatant of the untransformed PFKA control strain. Spiking the PFKA1-pBAD-ACL0729 supernatant with commercial laminaribiose resulted in a clear increase in the same signals, confirming the presence of the compound. These NMR results were validated by HPAEC-PAD, which confirmed the presence of laminaribiose in the culture supernatant at a concentration of 0.33 mM (0.11 g·L^{− 1}) (Figure S2C).

These results confirm that the proposed PFKA chassis strain is compatible with multiple GPs from different CAZy families and can efficiently synthesize and secrete diverse β-linked oligosaccharides in vivo, highlighting its potential as a platform for tailored oligosaccharide biosynthesis.

Synergistic effect of PTS and pfkA deletions on β-1,2-mannobiose production

After demonstrating mannobiose production in strain PFKA1 as proof of concept, we next aimed to assess the individual and combined contributions of the underlying genetic modifications by comparing β-1,2-mannobiose production in different strains. The pBAD-Teth514_1788 plasmid was introduced into strains MDO, MGX1, and PFKA1, which were then cultivated in bioreactors under previously described conditions using 10 g·L^− 1 mannose as the sole carbon source. NMR analyses of the supernatants at the end of cultivation are summarized in Table 3.

Table 3.

β-1,2-mannobiose conversion yields from mannose in engineered strains under different culture conditions

Strain	Genetic features	Culture conditions	Yielda
MDO- Teth514-1788	Reference strain	Bioreactor with mannoseb	ND
MGX1- Teth514-1788	PTS- with recombinant galP overexpressed; manB induced	Bioreactor with mannoseb	1.9 ± 1.2
PFKA1- Teth514-1788	PTS- with galP overexpressed; pfka deletion; manB induced	Bioreactor with mannoseb	8.5 ± 0.4
PFMA1- Teth514-1788	PFMA with galP overexpressed	Bioreactor with mannoseb	8.7
PFMA1- Teth514-1788	PFMA with galP overexpressed	Bioreactor with mannose + glycerolc	16.0 ± 1.1
PFMM1- Teth514-1788	PFMM with galP overexpressed	Shake-flask with mannose + glycerold	60.5 ± 12.0

ND: Not determined

^aGrams of produced β-1,2-mannobiose per 100 g of consumed mannose, as determined by NMR spectroscopy

^bMannose, 10 g∙L^− 1

^cInitial condition: mannose, 7 g∙L^− 1; glycerol, 7 g∙L^− 1. Second condition: mannose, 2.5 g∙L^− 1; glycerol, 12 g∙L^− 1

^dMannose, 1 g∙L^− 1; glycerol, 6 g∙L^− 1

No β-1,2-mannobiose was detected in the supernatant of the MDO strain, indicating that in the presence of an intact PTS, mannose is internalized exclusively in its phosphorylated form, precluding the reverse-phosphorylation reaction required for mannobiose synthesis.

Strain MGX1, lacking the PTS, produced a detectable amount of mannobiose (0.4 ± 0.2 mM; 0.13 ± 0.07 g∙L^− 1) (Figure S3). This result highlights the pivotal role of PTS deletion in enabling the import of unphosphorylated mannose into the cytoplasm—a prerequisite for the synthetic mannobiose production. Furthermore, overexpression of the manB gene in MGX1 may result in increased intracellular availability of M1P, resulting from manB-driven conversion of M6P to M1P.

The PFKA1 strain, featuring both PTS and pfkA deletions along with manB overexpression, had the highest β-1,2-mannobiose titer (1.8 mM ± 0.1 mM; 0.62 ± 0.03 g·L⁻¹) and a conversion yield of 8.5 ± 0.4%, an almost four-fold improvement over strain MGX1 (Figure S3). This increase is likely due to elevated intracellular concentrations of M1P resulting from the redirected carbon flux caused by pfkA deletion.

In summary, the fact that strains with an intact PTS system fail to produce detectable amounts of mannobiose show that PTS deletion is crucial for β-1,2-mannobiose biosynthesis. Two additional engineering effects, enhanced mannose transport and increased intracellular M1P availability, further boost β-1,2-mannobiose yields in the PFKA1 strain.

Further chassis improvement

Elimination of an acetylated sugar contaminant via maa gene deletion

The NMR spectra of supernatants from strain PFKA1 cultivated in minimal medium contained an un unexpected peak, which could not be assigned to any known sugar standards. Preliminary spectral analysis suggested that the unknown compound was an acetylated monosaccharide. Given the broad substrate specificity of maltose O-acetyltransferase (encoded by maa), it was hypothesized that this enzyme catalyzes the acetylation of internalized monosaccharides with acetyl-CoA as an acetyl donor—a reaction that is presumably facilitated by the unique metabolic configuration of the PFKA1 strain, whose defective PTS system allows the uptake of unphosphorylated sugars into the cytosol.

The maa gene was therefore knocked out, both to confirm the origin of this acetylated contaminant and to prevent a potentially competitive side reaction diverting precursors from β-1,2-mannobiose synthesis. Two strains, PFKA1 and its isogenic ∆maa derivative (PFMA1), both harboring the plasmid pBAD-Teth514_1788, were cultivated in M9 minimal medium in shake-flasks supplemented with 6 g·L⁻¹ mannose. The extracellular metabolite profiles were analyzed by NMR (Figure S4).

Deletion of maa led to the complete disappearance from the culture supernatant of the acetylated sugar contaminant, which we tentatively identified as 6-O-acetyl-mannose (6-O-AcMan). However, this deletion did not result in a measurable increase in β-1,2-mannobiose yield compared to the PFKA1 strain (see Table 3).

Improved mannose conversion through glycerol co-feeding and partial relief of carbon catabolite repression

Maximizing the conversion efficiency of mannose is particularly important given its relatively high cost as a carbon source. In the proof-of-concept configuration, mannose serves both as a carbon source for cellular growth and as a substrate for β-1,2-mannobiose biosynthesis, which intrinsically constrains the maximum achievable yield. However, the ability of PTS-deficient E. coli strains to co-utilize substrates offers a potential means to decouple biomass formation from product synthesis. As well as mediating sugar uptake, the PTS plays a key regulatory role in carbon catabolite repression (CCR), and its disruption enables simultaneous assimilation of multiple carbon sources. Glycerol was selected as a co-substrate due to its efficient uptake kinetics in the PTS-deficient background and its low cost (relative to mannose). To evaluate this strategy, a first bioreactor cultivation was carried out using strain PFMA1 harboring the pBAD-Teth514_1788 plasmid in minimal medium supplemented with equimolar concentrations of mannose and glycerol (7 g·L⁻¹ each) (Figure S5A).

As anticipated, glycerol was rapidly consumed within the first 25 h of cultivation, whereas mannose uptake was incomplete, with approximately 75% of the initial concentration remaining unutilized. Importantly though, and as expected, mannose was co-consumed, presumably because of the alleviation of CCR in this strain. The final concentration of β-1,2-mannobiose was 0.70 mM (0.23 g·L⁻¹); however, the mannose-to-β-1,2-mannobiose conversion yield calculated on the basis of mannose consumed, was significantly improved, reaching 16.8% compared with 9% in the mannose-only setup. These results were corroborated in a second bioreactor experiment in which the glycerol concentration was doubled and the mannose concentration reduced to one-third of its initial level (Figure S5B). Under these conditions, mannose utilization was also incomplete, with approximately 50% of the initial concentration remaining unutilized. However, the mannose-to-β-1,2-mannobiose conversion yield was similar to that observed previously, reaching 15.2%. Overall, under mannose–glycerol co-feeding conditions, the mannose-to-β-1,2-mannobiose conversion yield reached 16.0 ± 1.1% (see Table 3).

Decoupling growth and production to increase mannose conversion into β-1,2-mannobiose

These preliminary results support the feasibility of decoupling growth and production by co-substrate utilization. Nonetheless, there remains substantial room for optimization, as these data indicate that approximatively 20% of the internalized mannose was directed toward β-1,2-mannobiose synthesis, with the remainder (> 80%) still funneled into central carbon metabolism.

To further decouple cellular growth from β-1,2-mannobiose biosynthesis we deleted the manA gene, generating strain PFMM1. As illustrated in Fig. 4, deletion of manA creates a “closed circuit” by preventing the conversion of M6P into F6P, thereby channeling M6P preferentially towards β-1,2-mannobiose synthesis. The potential flux diversion from M1P to GDP-mannose via M1P guanylyl-transferase (encoded by CpsB) is likely negligible. In the PFMM1 strain, deletion of wcaJ abolishes colanic acid biosynthesis, for which GDP-mannose is an essential precursor. Cultivations were carried out in M9 minimal medium under standard conditions in shake-flasks with glycerol (6 g·L^− 1) and mannose (1 g·L^− 1) as carbon sources.

Fig. 4.

β-1,2-mannobiose production via co-substrate metabolism enabled by metabolic uncoupling of growth and product synthesis through the deletion of manA. GalP: galactose permease; manA: phosphomannose isomerase; GP: glycoside phosphorylase; M6P: mannose 6-phosphate; F6P: fructose 6-phosphate; M1P: mannose 1-phosphate; GDP-Man: GDP-mannose; G3P: glycerol 3-phosphate; Pyr: pyruvate; ATP: adenosine triphosphate; Gly: glycerol; OxPhos: oxidative phosphorylation

The β-1,2-mannobiose concentration at the end of cultivation was 0.84 ± 0.07 mM (0.29 ± 0.02 g·L^− 1 ), corresponding to a mannose conversion rate, calculated on the basis of mannose consumed, of 60.5 ± 12%, substantially higher than in the manA-positive strain (16%) (Table 3). Cultivating this PFMM1 strain in a bioreactor led to a final β-1,2-mannobiose titer of 1.21 mM (0.41 g·L^− 1) and a conversion rate of 58%, consistent with flask-scale experiments, and demonstrating the robustness and scalability of the process (data not shown). A current limitation is the substantial residual mannose, which is not fully internalized as the fermentation terminates once glycerol is depleted (Figure S5). Developing a cultivation strategy that enhances mannose uptake and conversion would further improve chassis performance.

Further improvement of process scalability could also be achieved through chromosomal integration of the galP and GP-encoding genes, which are currently plasmid-borne, thereby enhancing strain stability and reducing the dependence on plasmid maintenance during large-scale cultivation. In addition, the implementation of dynamic control strategies (e.g., inducible or growth-phase–dependent regulation of pfkA and manA expression), as an alternative to the constitutive knockouts used in this study, could provide an additional means to better balance growth and production performance.

Conclusion

In this study, we engineered an E. coli strain for the biosynthesis of disaccharides via GPs, leveraging a dual metabolic blockade strategy: inactivation of the PTS, allowing import of non-phosphorylated monosaccharides, and deletion of the pfkA gene, leading to intracellular accumulation of the phosphorylated sugar precursors (M1P, G1P) required for synthesis.

Our results show that mannose uptake in PTS⁻ strains is restored by heterologous expression of the galP permease, enabling partial growth on mannose. Introduction of the ΔpfkA mutation resulted in marked accumulation of intracellular phosphohexoses, particularly under glucose conditions, but also to a lesser extent under mannose, establishing strain PFKA1 as a favorable metabolic chassis for oligosaccharide biosynthesis.

Through expression of glycoside phosphorylases from distinct CAZy families (GH 130 and GH 94), we achieved in vivo production of two mannobioses (β-1,2 and β-1,4-mannobiose) and laminaribiose, thus demonstrating the modularity and versatility of the PFKA chassis for tailored oligosaccharide synthesis. The combination of PTS deletion, manB overexpression, and pfkA knockout had a pronounced synergistic effect, leading to a mannose-to-β-1,2-mannobiose conversion efficiency of approximatively 9% in the PFKA1 strain, far higher than measured in the corresponding pfkA-positive strain (2%).

We also observed the accumulation of an acetylated form of mannose in the culture supernatant, which disappeared upon deletion of the maa gene. This gene encodes maltose O-acetyltransferase, an enzyme known for its broad sugar substrate specificity [50]. To our knowledge, the accumulation of acetylated mannose in culture supernatants has never previously been reported, and likely results from the import of unphosphorylated mannose under PTS-deficient conditions.

Finally, we demonstrated that decoupling biomass formation from product synthesis through co-substrate feeding is an effective strategy to improve mannose-to-β-1,2-mannobiose conversion in engineered E. coli. The PTS-deficient background enables simultaneous assimilation of glycerol and mannose, with glycerol supporting cellular energy demands and mannose being redirected toward disaccharide biosynthesis. While co-feeding only modestly improved titers, the conversion yield nearly doubled compared with mannose-only cultures, highlighting the benefit of alleviating CCR. Further enhancement was achieved by deleting manA, thereby blocking the competing metabolic branch from M6P to central carbon metabolism. This modification significantly increased mannose utilization in β-1,2-mannobiose synthesis, with conversion rates of nearly 60% in both shake-flask and bioreactor cultures. Collectively, these results validate the effectiveness of growth-production decoupling strategies and metabolic channeling to optimize the biosynthesis of oligosaccharides in microbial systems. Metabolic flux analysis would further strengthen this framework by providing detailed insights into carbon partitioning and supporting continued strain and process optimization.

Overall, our approach illustrates the power of rational E. coli metabolic engineering combined with GP-based catalysis for the sustainable biosynthesis of costly sugars, with potential applications in prebiotics and in the cosmetic and pharmaceutical industries.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

PT: Conceptualization, Methodology, Formal analysis, Investigation, Writing - Original Draft. JD: Conceptualization, Methodology, Formal analysis, Investigation. LT: Methodology. GPV: Conceptualization, Validation, Writing - Review & Editing, Supervision, Funding acquisition. FL: Conceptualization, Validation, Writing- Review and Editing, Visualization, Supervision, Project administration, Funding acquisition. All authors have read and approved the final version of the manuscript.

Funding

This work was supported by the Agence Nationale de la Recherche [grant number ANR-16-CE20-0006-03] and by Toulouse Tech Transfer, a technology transfer acceleration company, with the mission to transfer and commercialize innovative technologies developed by the research laboratories of the west Occitanie region (http://www.toulouse-tech-transfer.com).

Data availability

The data that support the findings of this study will be made available on request.

Declarations

Competing interests

PT, JD, GPV, and FL declare competing financial interest related to the primary production process which is covered by a patent application filed under the PCT(WO/2021/229185) and licensed to Sweetech. JD is a co-founder of Sweetech SAS. LT declares no competing interests.