Engineered Passive Glucose Uptake in Pseudomonas taiwanensis VLB120 Increases Resource Efficiency for Bioproduction

Abstract

ABSTRACT

Glucose is the most abundant monosaccharide and a principal substrate in biotechnological production processes. In Pseudomonas, this sugar is either imported directly into the cytosol or first oxidised to gluconate in the periplasm. While gluconate is taken up via a proton‐driven symporter, the import of glucose is mediated by an ABC‐type transporter, and hence both require energy. In this study, we heterologously expressed the energy‐independent glucose facilitator protein (Glf) from Zymomonas mobilis to replace the native energy‐demanding glucose transport systems, thereby increasing the metabolic energy efficiency. The implementation of passive glucose uptake in engineered production strains significantly increased product titres and yields of the two different aromatic products, cinnamic acid (+10%–15%) and resveratrol (+26%; 18.1 mg/g) in batch cultures.

Abbreviations

• Gcd
• glucose dehydrogenase
• Glf
• glucose facilitator protein
• GRC3
• genome-reduced chassis strain number 3
• MC-III
• malonyl-CoA chassis strain number 3
• MC-IV
• malonyl-CoA chassis strain number 4
• MSM
• mineral salt medium

Introduction

The uptake of carbohydrates is a fundamental process of microbial life. Especially for the uptake of sugars across biological membranes, a plethora of different sugar utilisation and transport systems have evolved in bacteria (Jeckelmann and Erni 2020) in high dependency on the respective ecological niche. Biotechnological processes usually differ significantly from the microorganism's natural habitats, for example, by high concentrations of carbon sources and a lack of microbial competition. It is thus not surprising that the native sugar uptake and metabolism are not necessarily ideal for the applied bioprocess. In Pseudomonas, glucose enters the periplasm from the extracellular space via porins like one of four OprB (Wylie and Worobec 1995). It is subsequently taken up into the cytosol via ATP-binding cassette (ABC) transporter GtsABCD at the expense of ATP (del Castillo et al. 2007; Thomas and Tampé 2020) or oxidised to gluconate by periplasmic glucose dehydrogenase (GDH, Gcd) and gluconolactonases (Nerke et al. 2024). Periplasmic oxidation of glucose allows pseudomonads to shunt electrons via the pyrroloquinoline quinone (PQQ) cofactor directly into the respiratory chain (An and Moe 2016) and is considered the major route for glucose utilisation in many pseudomonads (Kohlstedt and Wittmann 2019). Periplasmic gluconate can further be oxidised to 2-ketogluconate by the gluconate 2-dehydrogenase complex (Gad; PP_3382-3384) in some pseudomonads (Kohlstedt and Wittmann 2019; Volke et al. 2023). Pseudomonas taiwanensis VLB120, a well-established bioproduction platform, lacks this second periplasmic oxidation. In contrast to GtsABCD, the gluconate transporter GntP/GntT or 2-ketogluconate transporters KguT both belong to the major facilitator superfamily (MFS) proton symporters (del Castillo et al. 2007) and are thus also energy-dependent (Lagarde 1977).

After translocation into the cytosol, glucose is phosphorylated by glucokinase (Glk) and subsequently converted by one of the three glucose 6-phosphate dehydrogenase (Zwf) isoenzymes (Volke, Olavarría, and Nikel 2021) and 6-phosphogluconolactonase (Pgl) to yield 6-phosphogluconate, which is an intermediate of the EDEMP cycle (Nikel et al. 2015) (Figure 1). Regulation of glucose uptake and sugar catabolism is controlled inter alia by the two-component system response regulator GltR-II (PP_1012 in Pseudomonas putida KT2440; PVLB_20105 in P. taiwanensis VLB120) and the repressor HexR (PP_1021; PVLB_20065) (del Castillo, Duque, and Ramos 2008; Lim et al. 2022; Udaondo et al. 2018). 2-Keto-3-deoxy-6-phosphogluconate (KDPG) is one of the effectors of HexR (Kim, Jeon, and Park 2008) whose deletion has been shown to be beneficial for the production of cis,cis-muconate in strains lacking Gcd due to the derepression of the intracellular catabolic genes (Bentley et al. 2020; Rorrer et al. 2022). In certain bioprocesses, excessive periplasmic oxidation of glucose to gluconate via gluconolactone is disadvantageous due to the acidification of the culture broth as this can inhibit microbial growth or require titration of base in pH-controlled cultivations. To avoid massive pH fluctuations or the need of excessive titration of base and acid while simultaneously enabling the application of higher glucose concentration in batch cultivations, the respective gcd gene can be deleted and, depending on the metabolic context, this may even lead to improved production, for example, muconic acid, polyhydroxyalkanoates and polyketides to name a few (Bentley et al. 2020; Poblete-Castro et al. 2013; Schwanemann et al. 2023) or avoid the unwanted bypassing conversion of alternative dimeric carbohydrates (Dvořák and de Lorenzo 2018). However, this deletion may cause detrimental effects because gluconate formation is involved in a wide regulatory network (Volke et al. 2023). Strategies to implement alternative uptake systems may further improve production in strains lacking Gcd. This would be widely applicable in many different cultivation approaches that use glucose as substrate.

[IMAGE OMITTED. SEE PDF]

Native energy-driven glucose transport systems are often beneficial in environments with low carbohydrate concentration because they allow uptake with high affinity (Jeckelmann and Erni 2020). This can be a distinct advantage under competitive conditions with limited glucose availability. In an artificial laboratory environment, substrates are usually supplied in concentrations exceeding the physiological K_M values of the respective uptake systems by several orders of magnitude. Therefore, the native sugar transport is usually not adapted to bioprocess requirements (Jeckelmann and Erni 2020) and in the case of some pseudomonads, it leads to excessive acidification by gluconate as a carbon sequestration strategy (del Castillo et al. 2007).

An alternative glucose uptake system, specifically interesting for biotechnological applications, can be found in Zymomonas mobilis. This bacterium naturally occurs in carbohydrate-rich environments and is an established facultative anaerobic host for ethanol fermentation with higher ethanol yields than yeasts due to its use of the Entner–Doudoroff pathway and limited biomass formation (Rogers, Lee, and Tribe 1976; Wang et al. 2018; Yang et al. 2016). In Z. mobilis, glucose is taken up along the concentration gradient through a glucose facilitator (Glf) protein without the expense of energy (Snoep et al. 1994) unlike the glucose phosphotransferase system of Escherichia coli and the ABC glucose transporter of pseudomonads. The facilitated diffusion mediated by Glf_Zm enables glucose uptake at a very high specific uptake rate (Fuhrer, Fischer, and Sauer 2005) but with a low reported affinity of K_M 1–4 mM (Parker et al. 1995; Weisser et al. 1995). Xylose and fructose (K_M 39 mM) are also transported, but with much lower affinities compared to glucose (Kurgan et al. 2021; Weisser et al. 1995). Glf_Zm appears as a promising transporter for sugars in an energy-independent manner that might be favourable in bioprocesses with high metabolic energy demands caused by, for example, osmotic stress or product toxicity. In E. coli, replacing the PEP-consuming phosphotransferase system with Glf_Zm increased shikimate production (Yi et al. 2003), while the expression of Glf_Zm in P. putida enabled the use of alternative carbohydrates (Bujdoš et al. 2023).

In this study, Glf_Zm was introduced into several genome-reduced chassis strains (GRC3) of P. taiwanensis VLB120. These strains were previously tailored for improved efficiency in bioprocesses by removal of dispensable or disadvantageous cellular features (Wynands et al. 2019), which is expanded in this study by the engineering of glucose uptake. Its impact on production strains was investigated for the biosynthesis of resveratrol and cinnamate, showcasing the applicability of Glf_Zm and its benefits in Pseudomonas for the production of different molecules from glucose.

Materials and Methods/Experimental Section

Cultivation Conditions, Media, DNA Techniques

Escherichia coli and Pseudomonas taiwanensis VLB120 strains were cultured on agar plates (15 g L⁻¹) or in LB medium (10 g L⁻¹ peptone, 5 g L⁻¹ sodium chloride, and 5 g L⁻¹ yeast extract) or modified mineral salts medium (MSM) with 3-fold concentration of buffer components (3 × 22.3 mM K₂HPO₄ and 3 × 13.6 mM NaH₂PO₄) (Hartmans et al. 1989) at 37°C and 30°C, respectively. Growth experiments started from serial precultures, beginning in LB, followed by an adaptation culture in MSM and the following cultivation under experimental conditions with either 20 or 30 mM glucose. Antibiotics were added if required (50 mg L⁻¹ kanamycin sulphate; 20 mg L⁻¹ gentamicin sulphate solution; 100 mg L⁻¹ ampicillin sodium salt; tetracycline hydrochloride 30 mg L⁻¹). For biotransformation experiments for the production of resveratrol, 1 mM p-coumarate was supplemented to the medium. Plasmids were constructed using the NEBuilder HiFi DNA Assembly methodology, and knockout procedures were performed using the I-SceI system that is based on two consecutive homologous recombination events as described previously (Schwanemann et al. 2023; Wynands et al. 2018). All strains, plasmids and DNA oligonucleotides used in this study are shown in Tables S1, S2 and S3, respectively. The amino acid sequence of the glucose facilitator protein from Zymomonas mobilis (Glf_Zm) used for codon optimization corresponds to UniProt entry P21906 (Table S4). The 200 bp upstream of gtsABCD (PVLB_20095–80) was considered as the promoter region of heterologous Glf_Zm expression constructs.

Cultivations for the production of cinnamate (Otto et al. 2019) or resveratrol were performed in a 24-square well plate system Duetz as described previously by Schwanemann et al. (2023).

Analytical Methods

Growth characterisation experiments were performed in 96-square well plates in the Growth Profiler 960 with respective calibration for conversion of ‘green values’ from pixels of a picture into OD₆₀₀ equivalents.

Determination of the optical density was performed at 600 nm with GE Healthcare Ultrospec 10 device from Fischer Scientific GmbH (Schwerte, Germany).

To determine biomass concentration after 24 h by cell dry weight (CDW) and OD₆₀₀, experiments were executed in 100 mL shake flasks with 11% filling volume, and samples of 10 mL were collected in dried and pre-weighted glass centrifuge tubes from Glaswarenfabrik Karl Hecht GmbH & Co KG (Sondheim, Germany) that were centrifuged for 20 min at 4000 g and washed with 5 mL of a 0.9% (w/v) sodium chloride solution. After discarding the supernatant, the pellets were dried at 65°C. A respective medium control was processed in parallel.

For the analysis of resveratrol, 1 mL culture broth was extracted with ethyl acetate and processed further in amber glass vials, as described in detail previously (Schwanemann et al. 2023). Cinnamate and p-coumarate were quantified from filtered culture supernatant, and all supplemented precursors and products were analysed by HPLC.

For the detection and quantification of cinnamate, p-coumarate and resveratrol, a 1260 Infinity II HPLC with a 1260 DAD WR (Agilent Technologies) and an ISAspher 100-5 C18 BDS column (Isera, Düren, Germany) was used, utilising the identical settings and gradients of 0.1% trifluoroacetic acid and acetonitrile as previously for resveratrol analysis (Schwanemann et al. 2023). Cinnamate, p-coumarate and resveratrol were measured at 245, 280, and 310 nm and eluted after 11.54, 7.13, and 9.08 min, respectively.

All experiments were executed in replicates, and significance analysis was performed using 1-way ANOVA with a confidence interval of p < 0.05.

Results and Discussion

Effect of Glf_Zm on the Growth of P. taiwanensis

The expression of Zm_glf, encoding the glucose facilitator of Z. mobilis, in different P. taiwanensis GRC3 strains was achieved either through the exchange of glucose transporter genes gtsABCD with Zm_glf or by the additional expression of Zm_glf from a neutral genomic landing pad (PVLB_06360-65) (Köbbing et al. 2024) (Figures 1B and 2). The same modifications were introduced in GRC3Δgcd and a malonyl-CoA platform strain (GRC3Δ6 MC-III) (Schwanemann et al. 2023) which are incapable of periplasmic glucose oxidation to gluconate. To elucidate the effects of these modifications, final biomass concentrations were determined by CDW to calculate biomass yields (Figure 2A) and growth rates were determined by OD₆₀₀ equivalent from Growth Profiler experiments (Figure 2B). The transporter exchange did not alter biomass yields in the GRC3 background. The additional expression of Zm_glf even reduced the biomass yield compared to the GRC3 control and the GRC3 strain with the replaced transporter. However, the difference was only significant when compared to the latter. In GRC3Δgcd strain background, a similar trend of final biomass yields was observed for additional Glf_Zm expression. The reduced biomass yield might be an effect of limited membrane space (Daddaoua et al. 2017; Nerke et al. 2024) and increased maintenance when Zm_glf is additionally expressed from the same promoter as native gtsABCD.

[IMAGE OMITTED. SEE PDF]

Upon expression of Zm_glf in strain GRC3Δ6 MC-III, which was metabolically engineered for increased malonyl-CoA supply including a gcd deficiency (Schwanemann et al. 2023), no significant changes were determined regarding biomass yield, but a reduced growth rate was observed. A trend to yield a high final biomass for the exchanged transporter strain was noted, although the respective control reached a slightly lower final biomass than the GRC3 control strains. That might not be surprising given that strain MC-III was engineered for bioproduction.

When comparing growth rates of the respective strains (Figures 2B, S5) the deletion of Gcd alone did not decrease the maximal growth rate, but the malonyl-CoA platform strain, GRC3Δ6 MC-III, has a 24% decreased growth rate (0.36 h⁻¹) compared to the GRC3 (0.47 h⁻¹). Strains with intact periplasmic gluconate formation showed no impact on their growth rate compared to the parental strain. However, strains with deleted Gcd and GtsABCD replaced by Glf_Zm, in which glucose is taken up solely via the heterologous Glf_Zm, showed a severe growth rate reduction of approximately half. The additional expression of Glf_Zm decreased the rate by 23% or 19% for strains lacking only Gcd (GRC3Δgcd) or the malonyl-CoA platform strain (GRC3Δ6 MC-III), respectively.

In general, it can be concluded that the glucose uptake systems in P. taiwanensis VLB120 can be replaced by Glf_Zm with little effect on biomass yield, but the growth rate is significantly affected in engineered strains that rely solely on glucose import through the Glf_Zm transporter. Since Zm_glf evolved in a different host and under different metabolic conditions at high carbohydrate concentrations, it is likely unbalanced, interfering with the host's cell envelope and expression capabilities, which could be addressed and optimised by adaptive laboratory evolution (ALE). However, the effects of glucose uptake modifications on growth were only tested on non-producing strains. Exponentially growing Pseudomonas, especially genome-reduced GRC3 with higher energy efficiency (Wynands et al. 2019), is not energy limited (Zobel et al. 2017). Deep engineering for the conversion of glucose to products typically causes growth rate reductions and higher energy demands, and here the use of Glf_Zm might be more beneficial.

Effect of Glf_Zm on Cinnamate Formation by Engineered P. taiwanensis

To test the effect of Glf_Zm in P. taiwanensis VLB120 aromatics production strains, a high-yield cinnamate producer was equipped with the alternative glucose uptake system. Strain P. taiwanensis GRC3Δ8ΔpykA-tap (Otto et al. 2019) (here called GRC3 PHE) was the basis for the phenylalanine platform strain. When equipped with a phenylalanine ammonia-lyase (AtPAL2) from Arabidopsis thaliana at the genomic Tn7 integration site (attTn7::P_14f-AtPAL2), this strain produces cinnamate from glucose with a yield of 23% C-mol C-mol⁻¹ (Otto et al. 2019).

Cinnamate biosynthesis was evaluated in engineered strains featuring either native or modified glucose transport, as well as with and without periplasmic oxidation of glucose to gluconate (Figure 3). In strains with exchanged glucose uptake system, titres were significantly improved by 5% to 4.3 ± 0.03 mM. Further, a 15% increase to 4.7 ± 0.03 mM was observed in the production strain with additional expression of Zm_glf. In this strain, the expression driven by the P_gts promoter may compete with native GtsABCD expression and gluconate uptake. Deletion of gcd slightly reduced final OD₆₀₀ and had a minor but significant negative effect on cinnamate titres (4.1 ± 0.04 mM for GRC3 PHE, 4.0 ± 0.03 mM GRC3 PHE Δgcd). This relatively minor effect on production would likely be offset in scaled-up batch cultures through the avoidance of transient acid formation. Without periplasmic Gcd, both expression strategies for Glf_Zm improved cinnamate titres to approximately 4.4 mM from 30 mM glucose compared to strain GRC3 PHE Δgcd, which constitutes a significant 10% improvement in production.

[IMAGE OMITTED. SEE PDF]

Effect of Glf_Zm on Resveratrol Formation in P. taiwanensis

High-yield cinnamate production poses a high drain on central metabolites with a relatively non-toxic product. In contrast, secondary metabolites are typically produced at lower yields and titres, but their production can pose a higher stress on the cell in terms of metabolic burden. To differentiate between these effects, a malonyl-CoA platform strain (GRC3Δ6 MC-III) was evaluated for its ability to produce resveratrol from glucose and p-coumarate (Figure 4) with implemented Zm_glf modifications. Resveratrol production was enabled by equipping the strain with the corresponding stilbene production module (attTn7::FRT-P_14f-his.AhSTS-Sc4CL^A294G). The GRC3Δ6 MC-III control with the stilbene module produced 77.6 mg L⁻¹ (0.34 mM) resveratrol from 30 mM glucose and 1 mM p-coumarate with 0.72 mM remaining precursor. This resveratrol production is in a similar range to those previously reported (Schwanemann et al. 2023). By exchanging the GtsABCD glucose transporter with Glf_Zm, resveratrol titres were increased to 97.7 mg L⁻¹ (0.43 mM), which represents a 26% improvement and a yield of 18.1 mg_resveratrol g_glucose⁻¹. Since this strain background already features a gcd knockout, this constitutes a complete replacement of glucose uptake by Glf_Zm. In contrast to previous reports (Braga et al. 2018), no product loss was observed (Figure S6), with only 0.54 mM p-coumarate remaining. The additional expression of Zm_glf in a GtsABCD⁺ background reduced overall biomass and resveratrol titre (Figure 4).

[IMAGE OMITTED. SEE PDF]

Consequently, GRC3Δ6 MC-III ΔgtsABCD::Zm_glf with stilbene production module was identified as a better platform for resveratrol production from p-coumarate than GRC3Δ6 MC-III in spite of its reduced growth rate (Figure 2) and thus strain P. taiwanensis GRC3Δ6MC-III ΔgtsABCD::Zm_glf is hereafter called GRC3Δ6MC-IV (generation No. 4 of the malonyl-CoA platform strain).

Conclusion

In this study, we investigated the impact of modified glucose uptake by a passive facilitator in P. taiwanensis VLB120 GRC3 and derivatives thereof. Final biomass yields of GRC3 and Gcd-deficient variants were not affected significantly when gtsABCD was replaced by Glf_Zm. Additional expression of Glf_Zm did not necessarily lead to improved growth on glucose or production, indicating that glucose uptake is not limited by transport per se or that membrane stress from the additional transport protein outweighed any potential benefit. In the absence of production, the resource-efficient passive glucose uptake came at the cost of reduced growth rates when both native glucose uptake and periplasmic oxidation were eliminated. However, this replacement boosted bioproduction of value-added aromatic molecules from different primary pathways in strains that already were highly optimised, highlighting the strategy's applicability, especially under demanding biotechnological conditions. The enhancement was observed not only in the synthesis of cinnamate (10% increase), a shikimate pathway-derived product, but also in resveratrol production (26% increase), which is derived from malonyl-CoA and supplemented p-coumarate. Hereby Glf_Zm can demonstrate its benefits in the absence of Gcd when the used host relies on cytosolic glucose metabolism, which is used in several bioprocess setups to ensure pH stability. However, the implementation of Glf_Zm in production strains with native glucose oxidation can be improved as well, as demonstrated for cinnamate formation. This straightforward engineering of the strain could boost future efforts in optimising Pseudomonas for production applications.

Author Contributions

Tobias Schwanemann: conceptualization, methodology, validation, formal analysis, investigation, funding acquisition, writing – original draft, visualization. Nicolas Krink: conceptualization, writing – review and editing, supervision. Pablo I. Nikel: conceptualization, resources. Benedikt Wynands: validation, writing – review and editing, supervision. Nick Wierckx: conceptualization, validation, resources, writing – review and editing, supervision, funding acquisition, project administration.

Acknowledgements

This work was supported by the German Federal Environmental Foundation (DBU) [PhD Scholarship 20019/638-32], the German Academic Exchange Service [DAAD Scholarship 57556281], the German Federal Ministry of Education and Research (BMBF) with the project NO-STRESS [FKZ 031B0852A], and the European Research Council via the project PROSPER (grant number 101044949). P.I.N. acknowledges funding from the Novo Nordisk Foundation through grant numbers NNF20CC0035580, NNF18OC0034818, and NNF21OC0067996. Open Access funding enabled and organized by Projekt DEAL.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

An, R., and L. A. Moe. 2016. “Regulation of Pyrroloquinoline Quinone‐Dependent Glucose Dehydrogenase Activity in the Model Rhizosphere‐Dwelling Bacterium Pseudomonas putida KT2440.” Applied and Environmental Microbiology 82, no. 16: 4955–4964. [DOI: https://dx.doi.org/10.1128/AEM.00813-16].