Engineering of Escherichia coli for D-tagatose production from lactose and whey permeate via the tagatose-6-phosphate pathway

TL;DR

Scientists engineered E. coli to convert lactose from dairy waste into D-tagatose, a low-calorie sweetener, using a new biocatalytic method.

Contribution

A novel whole-cell biocatalytic system for D-tagatose production from lactose and whey permeate using a modular pathway in E. coli.

Findings

A 35% conversion ratio of lactose's galactose to D-tagatose was achieved in the engineered E. coli strain.

The system enables growth-coupled D-tagatose production from whey permeate without prior hydrolysis or enrichment.

Modular pathway assembly and a sugar phosphatase reduced toxic intermediates and enabled efficient tagatose synthesis.

Abstract

D-tagatose is a low-calorie natural rare sugar with significant potential in the food and pharmaceutical industries. Conventional production relies on the enzymatic isomerization of D-galactose, a process limited by an unfavorable thermodynamic equilibrium and high substrate costs. This study presents a novel whole-cell biocatalytic approach for direct production of D-tagatose from lactose, an inexpensive and abundant sugar in dairy waste streams, such as whey permeate. The main lactose permease (lacY) of Escherichia coli BL21(DE3) was deleted, creating a clean host chassis that, due to its native galactose auxotrophy, was incapable of utilizing galactose. Subsequently, genes from the tagatose-6-phosphate (T6P) pathway of Lactococcus lactis, comprising a lactose-specific phosphotransferase system (PTS), a 6-phospho-β-galactosidase, a galactose-6-phosphate isomerase and the general PTS proteins, were modularly introduced, which allowed for the functional validation of each component. The crucial final step involved the expression of a sugar phosphatase to convert the intracellular intermediate, tagatose-6-phosphate, into D-tagatose. The fully engineered strain yielded a 35% conversion ratio of the galactose moiety of lactose to tagatose. This demonstrates the repurposing of the tagatose-6-phosphate pathway in E. coli to enable direct lactose utilization and growth-coupled production of D-tagatose from both lactose and whey permeate, without prior hydrolysis or substrate enrichment. Fermentation and tagatose production using whey permeate as the sole carbon source was also demonstrated, highlighting the potential of this system for the valorization of dairy byproducts into a high-value low calorie sweetener.

Image 1

•Synthetic biology was used to engineer an E. coli auxotroph for the efficient conversion of lactose into the rare sugar D-tagatose.•The final SynTag pathway establishes a growth-coupled bioprocess.•This approach successfully valorizes low-cost dairy side-streams by utilizing whey permeate as the sole carbon source.•Modular assembly of the tagatose-6-phosphate pathway and a sugar phosphatase mitigated toxic intermediates and enabled tagatose synthesis.

Synthetic biology was used to engineer an E. coli auxotroph for the efficient conversion of lactose into the rare sugar D-tagatose.

The final SynTag pathway establishes a growth-coupled bioprocess.

This approach successfully valorizes low-cost dairy side-streams by utilizing whey permeate as the sole carbon source.

Modular assembly of the tagatose-6-phosphate pathway and a sugar phosphatase mitigated toxic intermediates and enabled tagatose synthesis.

Figures

Construction of three *E. coli* expression plasmids carrying genes from the *L. lactis* tagatose-6-phosphate pathway. A. The *lacA* and *lacB* genes, encoding the heterodimeric galactose-6-phosphate isomerase, were codon-optimized for expression in *E. coli* and cloned as separate open reading frames (ORFs) into pACYC184 using HindIII and BamHI. A T7 promoter, *lac* operator and an *E. coli* ribosome binding site between the ORFs were included for controlled co-expression. B. The lactose-specific PTS components (*lacE* and *lacF*) and 6-phospho-β-galactosidase (*lacG*) were amplified from the

Full text

Introduction

1

The global demand for low-calorie, healthy sweeteners is rising due to increased awareness of the health risks associated with high sugar and high-intensity sweetener consumption, such as obesity and type 2 diabetes (Rogers and Appleton, 2021; Suez et al., 2015). Rare sugars, which are monosaccharides that exist in small quantities in nature, have emerged as promising alternatives (Chattopadhyay et al., 2014; Smith et al., 2022). D-tagatose, a naturally occurring ketohexose, is particularly attractive as it possesses 92% of the sweetness of sucrose but with only about one-third of the calories (1.5 kcal/g) (Ahmed et al., 2022; Levin et al., 1995; Van Laar et al., 2021). In addition, it carries Generally Regarded as Safe (GRAS) status by the U.S. FDA, is non-cariogenic, has a very low glycemic index and exhibits prebiotic properties that promote gut health (Hans Bertelsen, 2001; Venema et al., 2005). Recent works demonstrated its potential role in treatment of inflammatory bowel disease (L. Wang et al., 2025) and as an anti-fungal (Corneo et al., 2021). These attributes make it an ideal ingredient in a wide range of products, including dietary foods, beverages and pharmaceuticals (Roy et al., 2018).

Despite its desirable properties, the widespread use of D-tagatose is hindered by its high production costs, which stem from its inherent scarcity in nature (Oh, 2007). Alternative chemical synthesis methods often involve harsh conditions, complex purification steps and environmental concerns (Zhao et al., 2023). The predominant commercial production route is the enzymatic conversion of D-galactose to D-tagatose using the enzyme L-arabinose isomerase (LAI) (P. Kim, 2004; Weber et al., 2025). However, the isomerization reaction catalyzed by LAI is reversible and thermodynamically constrained. At typical reaction temperatures (e.g., 30 °C), the equilibrium favors the substrate, with a galactose-to-tagatose ratio of approximately 7:3, limiting the conversion yield to around 30% and necessitating costly and complex downstream separation processes. Even at higher temperatures (60 °C), the yield rarely exceeds 40-50% (Bober and Nair, 2019; J.-H. Kim et al., 2008; Z. Wang et al., 2022). In addition, D-galactose, is typically produced by the hydrolysis of lactose, which requires additional costly enzymatic steps and subsequent purification of galactose from glucose. The cost of producing and purifying the LAI enzyme, as well as its operational stability over long-term use, adds to the overall expense of the process (Fan et al., 2025b). To address these limitations, researchers are exploring advanced bio-engineering strategies (H. Zhang et al., 2025).

Whole-cell biocatalysis, where the enzymatic reactions occur within a living, engineered microorganism, offers several advantages over in vitro enzymatic methods. These include the elimination of costly enzyme purification, in-situ regeneration of critical cofactors and potentially greater process stability (Shen et al., 2019). It was demonstrated that microbial physiology can be leveraged to shift the unfavorable galactose-tagatose equilibrium. In E. coli, deletion of galK (galactose kinase) and expression of araA (LAI) improved the conversion yield of pure galactose to tagatose to 68%, compared with 36% for purified LAI alone (J.-H. Kim et al., 2008). This strategy exploits the differential uptake and release rates of galactose and tagatose across the cell membrane, resulting in a profound improvement in tagatose yield over that achieved with purified LAI. High conversion rates of 74% from galactose to tagatose were achieved by a two-step biotransformation process in which lactose from whey powder was initially hydrolyzed to glucose and galactose using purified β-galactosidase, followed by a biotransformation step using recombinant E. coli expressing a novel LAI (G. Zhang et al., 2020). Bacillus subtilis, a GRAS organism engineered to co-express LAI and β-galactosidase, allowed for the direct conversion of lactose to tagatose reaching conversion rates of up to 45% (Wen et al., 2025; X. Zhang et al., 2021). A recent work exploring the use of metabolically engineered Lactiplantibacillus plantarum for tagatose production from lactose, involved inactivation of the galactokinase gene (galK) to block galactose metabolism, replacement of the native promoter of the β-galactosidase gene (lacLM) with a strong lactose/galactose-inducible promoter, and overexpression of an LAI gene from Lactobacillus casei. The resulting recombinant strain was capable of one-pot conversion of lactose to D-tagatose yielding 33% conversion (S. Zhang et al., 2021). In another approach, a highly favorable galactose:tagatose ratio was achieved by engineering Saccharomyces cerevisiae or B. subtilis to bypass the LAI equilibrium entirely by implementing a synthetic two-step oxidoreductive pathway (J.-J. Liu et al., 2019; G. Zhang et al., 2023). In this system, galactose is first converted to galactitol using a xylose reductase, and then to tagatose by galactitol-2-dehydrogeanse, together with careful balancing of enzyme expression to maximize flux achieving up to 55% conversion.

The use of alternative cost-effective substrates has also been explored. For example, modular-pathway engineering in E. coli, including the optimization of expression vectors, multi-copy expression of rate-limiting enzymes and deletion of bypass pathway genes (pfkA, zwf), led to enhanced D-tagatose production from maltodextrin by expressing a phosphosugar isomerase that converts fructose-6-phosphate to tagatose-6-phosphate (T6P) and a sugar phosphatase that liberates the phosphate and produces tagatose (Dai et al., 2022) as well as the use of glucose to produce galactose by reversing the Leloir pathway and coupling to the expression of LAI to produce tagatose (Love et al., 2025). C4-epimerases were also recently harnessed to convert fructose and fructose-6-phosphate to tagatose using whole-cell bioconversion (Z. Liu et al., 2025; Palur et al., 2025; H. Zhang et al., 2025). Comprehensive reviews on tagatose production have been recently published that cover the advancements in the field (Fan et al., 2025; Fan et al., 2025a; Guo et al., 2025).

This study explored an alternative strategy by harnessing the naturally occurring T6P pathway. In bacteria, galactose can be metabolized via two main routes (Holden et al., 2003; Iskandar et al., 2019; Pessione, 2012; Zeng et al., 2010) (Fig. 1). The common and highly conserved Leloir pathway converts free galactose into glucose-6-phosphate for entry into glycolysis (Holden et al., 2003). The T6P pathway, found in organisms like Lactococcus lactis, is primarily involved in metabolizing the galactose moiety of lactose. In this pathway, lactose is first transported into the cell and simultaneously phosphorylated by a lactose-specific phosphotransferase system (PTS) (De Vos et al., 1990; Honeyman and Curtiss, 1993). The resulting lactose-6-phosphate is hydrolyzed to glucose and galactose-6-phosphate (G6P) (Honeyman and Curtiss, 1993; Wiesmann et al., 1995). G6P is then isomerized to T6P before it is phosphorylated to tagatose-1,6-bisphosphate and converted to glyceraldehyde-3-phosphate (GADP) and dihydroxyacetone phosphate (DHAP) that can enter glycolysis (Shakeri-Garakani et al., 2004). In this work, this pathway was repurposed for D-tagatose production by introducing part of the T6P pathway into E. coli and adding a final enzymatic step to remove the phosphate group from T6P. This strategy offers several theoretical advantages because the initial phosphorylation of the sugar by the PTS system effectively traps it inside the cell, and the final dephosphorylation step is essentially irreversible, providing a strong thermodynamic pull towards product formation and bypassing the equilibrium limitations of the LAI enzyme. As a proof-of-concept, the native lactose and galactose metabolic pathways were first deleted, and expression of T6P pathway genes was engineered using a modular plasmid-based approach, which enabled validation of the function of each component.Fig. 1Bacterial galactose utilization pathways. Bacteria use two primary metabolic routes to convert galactose into glycolytic intermediates. The Leloir pathway (left) processes free galactose, transforming it through a series of enzymatic steps into glucose-6-phosphate, a direct entry point into glycolysis. The tagatose 6-phosphate pathway (right), commonly found in lactic acid bacteria, primarily metabolizes the galactose moiety of lactose. It converts galactose-6-phosphate into triose phosphates, which are also key intermediates in glycolysis. Green enzymes belong to the Leloir pathway, blue enzymes belong to the Tag-6-p pathway and black enzymes are common to both. LacY – lactose permease, LacZ - β-galactosidase, GalM - aldose 1-epimerase, GalK- galactokinase, GalT - galactose-1-phosphate uridylyltransferase, Pgm - phosphoglucomutase, Glk - glucokinase, Pgi - glucose-6-phosphate isomerase, Pfk - phosphofructokinase, Fba - Fructose-bisphosphate aldolase, PTS - phosphoenolpyruvate-dependent sugar phosphotransferase system, LacE - PTS system lactose-specific EIICB, LacF - PTS system lactose-specific EIIA, LacG - 6-phospho-beta-galactosidase, LacAB - galactose-6-phosphate isomerase, LacC - tagatose-6-phosphate kinase, LacD - tagatose 1,6-diphosphate aldolase.Fig. 1

Materials and methods

2

Bacterial strains, plasmids, and growth conditions

2.1

E. coli BL21(DE3) was used as the parent strain for all genetic manipulations. Plasmids pACYC184 (Chang and Cohen, 1978), pRSFDuet-1 and pETDuet-1 (Novagen, USA), with compatible origins of replication and different antibiotic resistance markers, were used for gene cloning and expression. Strains were cultured in lysogeny broth (LB) medium for cloning and M9 minimal medium for tagatose production experiments. To create 1X M9 minimal medium, a stock solution of 5x M9 salts was diluted to its working concentration (1X), which contained the following concentrations of base salts: Na_2_HPO_4_⋅7H_2_O 12.8 g/L, KH_2_PO_4_ 3.0 g/L, NaCl 0.5 g/L, and NH_4_Cl 1.0 g/L as well as 2 mM MgSO_4_, 0.1 mM CaCl_2_. The medium was supplemented with various carbon sources (glucose, galactose, lactose, or whey permeate) at the indicated concentration for growth and production assays.

Sweet whey for experimental use was sourced from a small local cheese manufacturer and collected immediately after the production of mozzarella cheese, a process that yields sweet whey. The whey was defatted by subjecting it to two sequential centrifugations at 15,000×g. Subsequently, the whey was sterilized by filtration through a 0.2 μm pore-size membrane filter, aliquoted and stored at −20 °C until further use.

Construction of the E. coli lactose auxotroph

2.2

All primers used in the study are listed in the supplementary information (Table S1). The lacY gene in E. coli BL21(DE3) was deleted using the lambda red recombinase procedure (Datsenko and Wanner, 2000), with the pkD3 plasmid carrying the chloramphenicol resistance cassette serving as a template for PCR reactions. The deletion was verified using the nearby locus-specific primers lacY_ver_for and lacY_ver_rev (Table S1) with the respective primers c1 or c2 (Datsenko and Wanner, 2000). Afterwards, the cassette was removed using the FLP recombinase expressed from pCP20.

Plasmid construction

2.3

The sequence for the genes for the T6P pathway were sourced from Lactococcus lactis subsp. cremoris SK11 (ATCC BAA493) and were cloned into three separate compatible plasmids. The mini-operon containing lacA and lacB, each preceded by a strong E. coli ribosome binding site (RBS) and driven by a T7–Lac promoter, were codon-optimized for expression in E. coli, produced synthetically (GenScript, USA) and cloned into pACYC184 using HindIII and BamHI (pACYC-lacAB, GenBank# PX645022). A gene encoding a broad-specificity sugar phosphatase (GenBank accession WP_012940213.1) from Archaeoglobus profundus (Dai et al., 2020; Y. Kim et al., 2004) was synthesized (GenScript) and cloned into the first MCS of pETDuet-1 using NcoI and BamHI. The HPr and EI encoding for the general PTS components, were also codon-optimized for expression in E. coli, produced synthetically (GenScript, USA) and cloned into the second MCS of pETDuet-1 using NdeI and XhoI (pETDuet-tpp-EI_HPr, GenBank# PX645023). The lacE and lacF genes encoding for the lactose PTS components, and the lacG gene encoding for the 6-phospho-β-galactosidase, were amplified from the pSK11L plasmid and cloned into the second MCS of pRSFDuet-1 (pRSFDuet_FEG, GenBank# PX645024) using RF cloning with primers lacFEG_RF_f and lacFEG_RF_r (Table S1). The copy numbers of the plasmids used were pRSFDuet-1 – 100, pETDuet-1 – 40, pACYC184 – 15 (Chang and Cohen, 1978; Wu et al., 2017).

Pathway assembly and validation

2.4

The constructed plasmids were used to transform the E. coli lactose auxotroph in various combinations. A single colony of each strain was used to inoculate M9 medium containing 0.2% lactose, 0.01% glucose and appropriate antibiotics. Cultures were grown at 30 °C, in a shaker. Cell growth was monitored by measuring the optical density at 600 nm (OD_600_).

Sugar analysis by TLC and LC-MS

2.5

At the specified time points of cultivation, supernatants were collected by centrifugation. Sugar consumption and production were qualitatively analyzed by thin-layer chromatography (TLC). A volume of 1 μL of each sample and 1 μL of 10 mg/mL standards (lactose, glucose, galactose and tagatose) was applied onto TLC silica gel 60 plates (Sigma-Aldrich). A running solution composed of butanol/ethanol/deionized water (5:3:2, v/v/v) was used as the mobile phase. The plates were allowed to run for 3 h. Sugars were detected using a coloring solution consisting of 0.3% (w/v) N-(1-naphthyl) ethylenediamine and 5% (v/v) sulfuric acid in methanol (all from Sigma-Aldrich). Color development was achieved by heating the TLC plates in an oven at 150 °C, for 2 min, until spots appeared. For quantitative analysis, supernatants were filtered and analyzed by high-performance liquid chromatography (HPLC) coupled with mass spectrometry (MS) (Waters). Separation was performed under hydrophilic interaction with an X-bridge amide column (3.5 μm 4.6 × 250 mm, Waters). The mobile phase was acetonitrile and water (75:25, v/v) containing 0.1% formic acid. Pure sugars were used to generate calibration curves, and the analysis was performed by monitoring the ion at m/z 179.57. Mass spectrometry measurements were performed in negative ESI mode using a heated electrospray ionization source connected to a Waters ZQ mass detector. The capillary voltage was set to −2000 V and the cone voltage to 18 V. The source temperature was adjusted to 120 °C, with a desolvation temperature of 300 °C. Nitrogen was used as the nebulizing and drying gas, with a desolvation gas flow of 500 L/h and a cone gas flow of 50 L/h. All samples were assayed in triplicates. Time-course fermentations in minimal medium and whey permeate were conducted in shake flasks at 30 ^ᵒ^C. Samples were taken periodically to measure cell growth (OD_600_) and for sugar analysis by TLC and HPLC.

Results

3

The T6P pathway was exploited to produce tagatose, as an alternative to the more common Leloir pathway for galactose metabolism (Fig. 1) (Honeyman and Curtiss, 1993). A prerequisite for this work was to create a host strain incapable of metabolizing lactose or galactose through its native pathways, and thereby prevent the diversion of substrate and ensure that any observed lactose metabolism is due solely to the engineered pathway. To this end, E. coli BL21(DE3), which naturally lacks the genes of the Leloir pathway (Daegelen et al., 2009), was subjected to targeted deletion of the lactose permease gene (lacY) (Fig. 2A). To validate the phenotype of the knockout strain, its growth was compared to the wild-type strain in M9 minimal medium with glucose, galactose or lactose as carbon sources (Fig. 2B–C). The wild-type strain grew on glucose, was able to grow on lactose (by using only the glucose moiety of lactose), but failed to grow on galactose (Fig. 2B). In contrast, the engineered auxotroph grew on glucose only and showed no growth on galactose or lactose, confirming the successful deletion of the native catabolic pathways and establishing a suitable genetic background for subsequent engineering efforts (Fig. 2C).Fig. 2Construction and validation of an E. coli BL21(DE3) lactose auxotroph. A. Schematic representation of the tagatose production pathway, achieved by deletion of galM, galK, galT, galE (preventing galactose metabolism) and lacY (eliminating lactose uptake) in E. coli BL21(DE3) to generate the lactose auxotroph. B. Growth validation of E. coli BL21(DE3) (wild-type) and **C.**E. coli BL21(DE3) ΔlacY strains in M9 minimal medium supplemented with glucose, galactose, or lactose as the sole carbon source. The absence of growth on lactose and galactose for the ΔlacY strain confirms its glucose and lactose auxotrophy.Fig. 2

The heterologous biosynthetic pathway (synthetic tagatose pathway – SynTag) derived from pSK11L was reconstructed in E. coli using a modular three-plasmid system. The eight pathway genes were distributed among the pACYC184, pRSFDuet-1 and pETDuet-1 plasmids to enable modularity while maintaining plasmid compatibility (Fig. 3). The pACYC-lacAB vector carried the codon-optimized lacA and lacB genes encoding the G6P isomerase subunits, assembled as a synthetic mini-operon under a T7-Lac promoter with individual RBSs. The lactose PTS transporter (lacE, lacF) and 6-phospho-β-galactosidase (lacG) were amplified from pSK11L and cloned into pRSFDuet-1 to yield pRSFDuet-lacFEG. The general PTS components (HPr, EI) and a broad-specificity sugar phosphatase (tpp) were introduced into pETDuet-1, resulting in pETDuet-tpp-EI_HPr. Each construct was sequence-verified and PCR-confirmed to possess the expected inserts. The use of three plasmids with compatible replication origins and antibiotic markers permitted simultaneous maintenance and inducible expression of all pathway enzymes. The modular design also facilitated troubleshooting and optimization of individual pathway modules during subsequent characterization (Fig. 4).Fig. 3. Construction of three E. coli expression plasmids carrying genes from the L. lactis tagatose-6-phosphate pathway. A. The lacA and lacB genes, encoding the heterodimeric galactose-6-phosphate isomerase, were codon-optimized for expression in E. coli and cloned as separate open reading frames (ORFs) into pACYC184 using HindIII and BamHI. A T7 promoter, lac operator and an E. coli ribosome binding site between the ORFs were included for controlled co-expression. B. The lactose-specific PTS components (lacE and lacF) and 6-phospho-β-galactosidase (lacG) were amplified from the pSK11L plasmid and cloned into the second multiple cloning site of pRSFDuet-1 using Gibson assembly. C. A sugar phosphatase was cloned into the first multiple cloning site of pETDuet-1 using NcoI and BamHI. Separately, the PTS-related genes from L. lactis, HPr and EI, were cloned into the second multiple cloning site of pETDuet-1 using NdeI and XhoI.Fig. 3. Fig. 4The SynTag pathway for lactose utilization and tagatose production in E. coli. A. According to the newly assembled pathway, lactose is transported into the cell by a lactose-specific phosphotransferase system (PTS) involving enzyme I (EI) and the phosphocarrier protein (HPr), with phosphoenolpyruvate (PEP) serving as the phosphate donor. Intracellular lactose is then converted into glucose and galactose-6-phosphate (G6P) by a 6-phospho-β-D-galactosidase (pbg). Glucose enters glycolysis (indicated by a broken arrow). G6P is converted to tagatose-6-phosphate by a G6P isomerase (gpi) and dephosphorylated to tagatose by a tagatose-6-phosphate phosphatase (tpp). Tagatose is then likely exported across the membrane by the sugar exporter SetA. B. To assess the impact of various gene combinations on lactose metabolism and tagatose production, different sets of genes were systematically introduced into engineered E. coli and the resulting cellular performance was evaluated. Growth was monitored by optical density (OD) in M9 medium with lactose, and sugar consumption/production was analyzed from spent medium, using thin layer chromatography (TLC). **0.**E. coli ΔlacY. **I.**E. coli ΔlacY + LacFEG. **II.**E. coli ΔlacY + LacAB. **III.**E. coli ΔlacY + HPr, EI and sugar phosphatase. **IV.**E. coli ΔlacY + LacAB, LacFEG, HPr, EI and sugar phosphatase.Fig. 4

The function of different combinations of the engineered modules was sequentially tested by assessing growth and sugar conversion of the E. coli lactose auxotroph transformed with the aforementioned plasmids and grown on lactose-containing medium (Fig. 4B). The control strain (E. coli ΔlacY) and strains containing only partial sets of genes (e.g., only LacAB, only LacFEG, or only HPr/EI) showed no significant growth on lactose, as they lacked a complete pathway for its uptake and conversion. A strain containing both the PTS uptake system (LacFEG) and the isomerase (LacAB) showed minimal growth with no significant lactose consumption (Fig. S1E). This suggests that while lactose was imported and converted to tagatose-6-phosphate, accumulation of the phosphorylated intermediate was likely inhibitory to the cells. Strains containing the phosphatase and the PTS genes showed lactose utilization but no galactose or tagatose accumulation (Fig. S1G). The essentiality of the complete pathway was demonstrated in the strain containing all three plasmids (E. coli SynTag strain) and, subsequently expressing all components of the pathway, i.e., PTS uptake (LacFEG, HPr, EI), isomerization (LacAB) and dephosphorylation (tpp) (Fig. 4), which showed growth on lactose as well as galactose and tagatose accumulation (Fig. 4B).

To assess the impact of cultivation conditions on strain performance, we tested the effect of temperature and metal ions on growth and tagatose production in M9 minimal medium containing 2 mM MgCl_2_ and 0.1 mM CaCl_2_. Cultivation at 30 °C supported robust growth and tagatose production, whereas at 37 °C no growth or tagatose formation was observed (Fig. S2). At 25 °C, cells displayed slower growth accompanied by reduced tagatose titers relative to 30 °C (Fig. S2). We next supplemented M9 medium with either 1 mM MnCl_2_ or 1 mM ZnCl_2_ and found no significant change in growth or tagatose production compared to the standard formulation (Fig. S3). Based on these results, all subsequent experiments, were carried out in M9 minimal medium at 30 °C.

LC-MS was then used to separate and directly quantify tagatose from the growth medium (Fig. 5). The chromatogram revealed a significant peak corresponding to tagatose, confirming the qualitative TLC results. Using a calibration curve generated from a pure tagatose standard, the final tagatose titer was determined to be 0.35 g/L from an initial 2 g/L of lactose. This corresponds to a yield of 0.175 g/L tagatose per 1 g/L of lactose consumed and a 35% conversion ratio from the galactose moiety of lactose.Fig. 5LC-MS quantification of sugars during cultivation of the E. coli SynTag strain on lactose. An HPLC system equipped with an XBridge Amide column and coupled to a mass spectrometry detector (LC–MS) was used for the separation and detection of carbohydrates present in the culture supernatants of the E. coli SynTag strain. Samples collected after 0, 24, 29, 48, and 72 h of cultivation in M9 minimal medium containing lactose were analyzed to monitor sugar composition in the extracellular fraction. Quantification of individual sugars was performed by comparing the peak areas of ions with m/z 179.57 to calibration curves generated from known concentrations of commercial standards of lactose, galactose, glucose, and tagatose.Fig. 5

A time-course analysis was performed to investigate the relationship between cell growth, lactose consumption and formation of galactose and tagatose by E. coli SynTag grown in a shake flask containing M9 minimal medium with lactose and glucose as the carbon source (Fig. 6). Cell growth (OD_600_) followed a typical bacteria growth curve, reached a stationary phase after approximately 48 h. The lactose concentration in the medium decreased steadily as the cells grew and was entirely depleted after 48 h, indicating that substrate consumption was growth-associated. Concurrently, the concentration of tagatose and galactose in the medium increased. Both, tagatose and galactose reached their maximal concentration as the culture entered the stationary phase. This profile confirms that D-tagatose is produced and exported by the engineered cells throughout their growth cycle.Fig. 6Time-course of lactose metabolism and tagatose biosynthesis by E. coli SynTag strain. The E. coli SynTag strain was cultivated in M9 minimal medium containing 0.2% (w/v) lactose as the carbon source. A. Sugars visualization using thin-layer chromatography analysis of culture supernatants collected after 0-168 h of culture and pure standards of lactose, galactose, glucose and tagatose. (B) Growth and sugar consumption profiles obtained from time-course sampling. Optical density at 600 nm (OD_600_) was monitored in parallel with HPLC quantification of extracellular lactose, galactose, and tagatose.Fig. 6

E. coli SynTag cultured in M9 medium with whey permeate, a lactose-rich byproduct of the dairy industry (Fig. 7), successfully utilized the lactose in the whey permeate for both growth and tagatose production. The strain successfully grew across all tested concentrations of the whey permeate (12.5%, 25%, 50%, and 100%) and the final tagatose titer increased with higher concentrations of whey. Direct tagatose production from whey permeate demonstrated that the engineered bacterial strain is capable of converting a complex industrial waste stream into a high-value product, paving the way for a more economical and sustainable D-tagatose production process.Fig. 7Time-course analysis of lactose metabolism and tagatose biosynthesis by E. coli SynTag using whey permeate. The E. coli SynTag strain was cultivated in M9 minimal medium supplemented with 12.5% (v/v) whey permeate as the carbon source. A. Thin-layer chromatography analysis of culture supernatants collected after 0-240 h, alongside pure standards of lactose, galactose, glucose and tagatose. B. Growth and sugar utilization profiles during cultivation in whey permeate-containing medium. Optical density at 600 nm (OD_600_).Fig. 7

Discussion

4

In this study, an E. coli strain was engineered to produce the rare sugar D-tagatose directly from lactose. This was achieved by introducing a new pathway composed of genes from the T6P pathway from L. lactis and a sugar phosphatase, establishing a whole-cell biocatalytic platform that can produce D-tagatose directly from lactose and whey water permeate. This approach can offer several key advantages over conventional LAI-based enzymatic production methods. First, it circumvents the unfavorable thermodynamic equilibrium that limits LAI conversion yields. Advanced strategies using oxidoreductase reactions have been developed to create a thermodynamic pull toward the product achieves of up to 63% conversion (J.-J. Liu et al., 2019; G. Zhang et al., 2023). Similarly, the T6P pathway system presents substantial opportunities for improvement through optimization of enzyme expression levels, cofactor regeneration, membrane transport engineering, and fermentation conditions. The modular nature of the pathway and the irreversible thermodynamic pull of the final dephosphorylation step provide a strong foundation for further enhancement toward industrially viable conversion efficiencies. By coupling lactose uptake to phosphorylation via the PTS in a strain that cannot metabolize galactose, metabolic flux is directed into the cell and committed to the pathway by phosphorylation of the galactose moiety. The final, irreversible dephosphorylation of the high-energy T6P intermediate by a sugar phosphatase ensures pull of the reaction toward D-tagatose, bypassing the equilibrium limitations of simple isomerization. The system also enabled direct utilization of lactose without prior hydrolysis, simplifying upstream processing. This E. coli system utilizes glucose from lactose-6-phosphate hydrolysis for growth while G6P is directly shunted into the tagatose-production pathway.

The final conversion of the galactose moiety in lactose to D-tagatose in the SynTag strain was 35%, which is lower than the highest yields reported for optimized whole-cell and oxidoreductive systems using pure galactose substrates (50–74%) or immobilized LAI reactors reaching titers above 100–200 g/L from highly concentrated feedstocks (J.-J. Liu et al., 2019; Weber et al., 2025; Wen et al., 2025; G. Zhang et al., 2020, 2023). In contrast to such benchmark studies, our process operates directly on lactose or whey permeate in a growth coupled manner without prior hydrolysis or enrichment. Given that lactose concentration in whey water is typically around 50 g/L (Guimarães et al., 2010), the theoretical maximum D-tagatose titer is limited to approximately 25 g/L. Under these substrate and process constraints, the 35% conversion obtained here demonstrates that the tagatose-6-phosphate pathway can approach and, in some cases, match LAI-based performance while using a low-cost, minimally processed dairy side stream as both carbon source and feedstock. The key novelty of this work lies in repurposing the native T6P route in a galactose-auxotrophic E. coli chassis and coupling PTS-driven lactose phosphorylation to an irreversible dephosphorylation step, thereby providing a whole-cell biocatalyst that simultaneously overcomes LAI thermodynamic limitations and enables direct tagatose production from whey permeate.

The successful use of whey permeate as feedstock for E. coli growth and tagatose production is an important aspect of the presented system (Fig. 7). Whey is a major byproduct of the dairy industry, and its disposal represents both an environmental challenge and a lost economic opportunity (Zandona et al., 2021). The ability of E. coli SynTag to convert the lactose in this waste stream into a high-value food ingredient suggests a possible path toward a circular bioeconomy. The titers achieved are promising for optimization and scale-up.

While the selected production pathway aimed to optimize tagatose production by circumventing the equilibrium barrier, E. coli SynTag still produced significant amounts of galactose alongside tagatose. The galactose accumulation can be attributed to the lack of substrate specificity of the phosphatase component and an imbalance in expression levels of pathway components, particularly the T6P isomerase (LacAB) and the downstream phosphatase. Similar to the isomerization between galactose and tagatose, the equilibrium between G6P and T6P is strongly shifted toward G6P (approximately 9:1 ratio under physiological conditions (Bissett et al., 1980)). However, the irreversible dephosphorylation of T6P by the phosphatase creates a thermodynamic pull that shifts equilibrium toward T6P production, as evidenced by the approximately 2:1 concentration of galactose and tagatose obtained in our system (Fig. 6).

The systematic, module-by-module assembly of the pathway (Fig. 4) was critical to the proof-of-concept. This approach, similar to that used in other metabolic engineering projects, enabled verification of the function of each part of the complex eight-gene pathway and determination that the final phosphatase step is essential for relieving toxic accumulation of phosphorylated intermediates and for enabling productive flux. The modular nature of the SynTag pathway facilitates pathway reconstruction and debugging in a controlled, stepwise manner.

Future work will focus on improving pathway specificity and reducing galactose byproduct formation, which can be achieved using two complementary strategies. First, fine-tuning the expression ratio of isomerase vs. phosphatase can enhance flux from G6P toward T6P while reducing spurious galactose production. This can be accomplished through promoter engineering, RBS tuning, or adjustment of gene and plasmids copy numbers. Second, identifying or engineering more selective phosphatases would directly address substrate promiscuity. This can be achieved through protein engineering of the enzymes to enhance specificity toward T6P, or by screening genomic and metagenomic databases for natural phosphatase variants with higher catalytic preference for T6P. Additionally, inclusion of specific sugar efflux transporters could lower intracellular tagatose and galactose levels and thereby help alleviate toxicity associated with phosphosugar accumulation(J. Y. Liu et al., 1999).

A major challenge lies in strain stability, as the current system relies on three separate plasmids, which impose significant metabolic burden and compromise genetic stability during extended fermentations. While antibiotics were used here as selective pressure to maintain plasmid retention, this strategy is costly, unsustainable at scale and raises regulatory concerns for food-related applications. A long-term solution would involve integrating the entire pathway into the E. coli chromosome to generate a stable, plasmid-free production strain.

Once synthesized intracellularly, D-tagatose must be exported efficiently to the growth medium to avoid feedback inhibition and facilitate downstream recovery. Overexpression of native or heterologous sugar efflux pumps such as SetA could enhance transport (J. Y. Liu et al., 1999), reduce intracellular stress and improve overall productivity. Additionally, efficient separation and purification of D-tagatose from complex fermentation broth represents a non-trivial challenge that will require tailored recovery and crystallization strategies for successful industrial application.

In conclusion, this study presents the engineering of E. coli SynTag for the whole-cell biocatalytic production of the rare sugar D-tagatose directly from lactose. This was achieved by reconstructing the T6P pathway from L. lactis using a modular, three-plasmid system. This novel strategy is advantageous because it circumvents the unfavorable thermodynamic equilibrium that limits conventional LAI-based production methods. By coupling lactose uptake to phosphorylation via the PTS and employing the final irreversible dephosphorylation of T6P by a sugar phosphatase, metabolic flux is directed toward D-tagatose formation. The fully engineered strain demonstrated a 35% conversion ratio from the galactose moiety of lactose to tagatose. Furthermore, the system proved capable of utilizing the abundant whey permeate dairy waste stream as the sole carbon source for tagatose production, highlighting its potential for the valorization of dairy byproducts into a high-value sweetener. Future research will concentrate on optimizing pathway specificity to reduce galactose byproduct formation by tuning enzyme expression ratios or engineering more selective phosphatases. Ultimately, integrating the eight-gene pathway into the E. coli chromosome will be necessary to achieve the strain stability required for sustainable and economical industrial application.

Author contributions

A.A. designed and conducted experiments and analyzed the data. R.B.A. conducted experiments and analyzed the data. N.G. designed experiments and analyzed the data. S.M. conducted analytical experiments and analyzed the data. I.Y. conceived the research, obtained funding, designed experiments, analyzed the data, and wrote the manuscript.

Declaration of competing interest

All authors declare that they are inventors on a patent related to the bacterial production of D-tagatose described in this manuscript.