Dynamic control over feedback regulatory mechanisms improves NADPH flux and xylitol biosynthesis in engineered E. coli

Abstract

We report improved NADPH flux and xylitol biosynthesis in engineered E. coli. Xylitol is produced from xylose via an NADPH dependent reductase. We utilize 2-stage dynamic metabolic control to compare two approaches to optimize xylitol biosynthesis, a stoichiometric approach, wherein competitive fluxes are decreased, and a regulatory approach wherein the levels of key regulatory metabolites are reduced. The stoichiometric and regulatory approaches lead to a 20-fold and 90-fold improvement in xylitol production, respectively. Strains with reduced levels of enoyl-ACP reductase and glucose-6-phosphate dehydrogenase, led to altered metabolite pools resulting in the activation of the membrane bound transhydrogenase and an NADPH generation pathway, consisting of pyruvate ferredoxin oxidoreductase coupled with NADPH dependent ferredoxin reductase, leading to increased NADPH fluxes, despite a reduction in NADPH pools. These strains produced titers of 200 g/L of xylitol from xylose at 86% of theoretical yield in instrumented bioreactors. We expect dynamic control over the regulation of the membrane bound transhydrogenase as well as NADPH production through pyruvate ferredoxin oxidoreductase to broadly enable improved NADPH dependent bioconversions or production via NADPH dependent metabolic pathways.

Keywords: Dynamic metabolic control; Metabolic dysregulation; NADPH; Xylitol.

Figure 1:

a) Time course of two stage dynamic metabolic control upon phosphate depletion. Biomass levels accumulate and consume a limiting nutrient (in this case inorganic phosphate), which when depleted triggers entry into a productive stationary phase, levels of key enzymes are dynamically reduced with synthetic metabolic valves (red). b & c) Synthetic metabolic valves utilizing CRISPRi based gene silencing and/or controlled proteolysis. b) Array of silencing guides can be used to silencing target multiple genes of interest (GOI). This involves the inducible expression of one or many guide RNAs as well as expression of the modified native Cascade system wherein the cas3 nuclease is deleted. The gRNA/Cascade complex binds to target sequences in the promoter region and silences transcription. c) C-terminal DAS+4 tags are added to enzymes of interest (EOI) through chromosomal modification, they can be inducibly degraded by the clpXP protease in the presence of an inducible sspB chaperone. d) Dynamic control over protein levels in E. coli using inducible proteolysis and CRISPRi silencing in glucose minimal media. As cells grow phosphate is depleted, cells “turn off” mCherry and “turn on” GFPuv. Shaded areas represent one standard deviation from the mean, n=3. e) The relative impact of proteolysis and gene silencing alone and in combination on mCherry degradation, f) mCherry decays rates.

Figure 2:

Dynamic control over mCherry levels in E. coli using inducible proteolysis and CRISPRi silencing in xylose minimal media microfermentations. Cells are grown to mid-exponential phase, washed, resuspend normalized to OD(600nm) ~ 1 in phosphate free minimal media with xylose as a sole carbon source. After 24 hrs (post phosphate depletion) GFPuv and mCherry levels are measured. The relative impact of proteolysis and silencing alone and in combination on final mCherry levels is reported.

Figure 3:

The design of engineered strains and plasmids used in this study. a) An overview of xylitol production and the location of metabolic valves in central metabolism, including stoichiometric valves (blue area, as shown in c) and regulatory valves (green area, as shown in d). Xylitol is produced from xylose by a xylose reductase (XyrA). Valves comprise inducible proteolysis and/or silencing of 5 enzymes: citrate synthase (GltA) , xylose isomerase (XylA), glucose-6-phosphate dehydrogenase (Zwf), enoyl-ACP reductase (FabI) and soluble transhydrogenase (UdhA). The membrane bound transhydrogenase (PntAB) is also shown. b). Schematics of the low phosphate inducible xylose reductase (XyrA) expression plasmid (left) and pCASCADE silencing plasmid (right). Red regions indicate spacers targeting promoters, and black regions repeats. c & d) An overview of promoters being silenced and chromosomal modifications introducing degron tags (red triangles) to key enzymes. pCASCADE plasmids can silence one or more of the target promoters (red lines). c) Stoichiometric valves: xylose isomerase (xylA) and soluble transhydrogenase (udhA), which directly compete with xylitol production. d) Additional regulatory valves: citrate synthase (gltA), glucose-6-phosphate dehydrogenase (zwf), enoyl-ACP reductase (fabI), which control flux through the tricarboxylic acid cycle, pentose phosphate pathway and fatty acid biosynthesis, respectively. Abbreviations: xylE: xylose permease, xylFGH: xylose ABC transporter, PPP: pentose phosphate pathway, PDH: pyruvate dehydrogenase multienzyme complex, TCA: tricarboxylic acid, G6P: glucose-6-phosphate, 6-PGL: 6-phosphogluconolactone, 6PG: 6-phosphogluconate, GA3P: glyceraldehyde-3-phosphate, PEP: phosphoenolpyruvate, OAA: oxaloacetic acid, X5P: xylulose-5-phosphate.

Figure 4:

Dynamic control over the levels of the central metabolic enzymes. The impact of silencing and proteolytic degradation on enzyme levels of a) XylA (xylose isomerase), b) UdhA (soluble transhydrogenase), c) GltA (citrate synthase), and d) Zwf (glucose-6-phosphate dehydrogenase). e) The impact of proteolytic degradation on FabI (enoyl-ACP reductase) levels. In the case of XylA and UdhA (a&b) enzyme activity in lysate was used to quantify protein levels. In the case of GltA, Zwf & FabI (c-e), genes were tagged with a C-terminal sfGFP (+/− a degron tag) and quantified via an ELISA. All assays were performed 24 hour post induction by phosphate depletion in microfermentations.

Figure 5:

a) A stoichiometric approach to improving xylitol production using dynamic control, wherein enzyme levels of competitive pathways are dynamically reduced to redirect flux to the desired product. a) In this case xylose isomerase (XylA) and the soluble transhydrogenase (UdhA) were targeted for dynamic control. b-d) Specific xylitol production in strains engineered for dynamic control over levels of b) xylose isomerase (XylA), c) soluble transhydrogenase (UdhA) and d) the combined control over xylose isomerase soluble transhydrogenase. ev: empty vector, x: xylA promoter gRNA, u: udhA promoter gRNA. All results were obtained from xylose minimal media microfermentations.

Figure 6:

a) An overview of the regulatory approach to xylitol production and the location of metabolic valves in central metabolism and our model of optimal NADPH flux. Xylitol is produced from xylose by a xylose reductase (XyrA). Valves comprise inducible proteolysis and/or silencing of 5 enzymes: citrate synthase (GltA) , xylose isomerase (XylA), glucose-6-phosphate dehydrogenase (Zwf), enoyl-ACP reductase (FabI) and soluble transhydrogenase (UdhA). The membrane bound transhydrogenase (PntAB, encoded by pntAB), pyruvate-ferredoxin/flavodoxin oxidoreductase (Pfo, encoded by the ydbK gene) and flavodoxin reductase (Fpr, encoded by fpr), are also shown. SoxRS regulon, which is sensitive to oxidant levels and NADPH pools, can activate the expression of Pfo and Fpr. b) Specific xylitol production (g/L-OD_600nm) in microfermentations as a function of silencing and or proteolysis. c) P-values for the data in b), comparing each strain to the no-valve control using a Welch’s t-test. d) a rank order plot of the data from the b). A post hoc Dunnett test shows combinations that differ from the DLF_Z0025-Empty vector control significantly at p-value < 0.05, which are indicated as green bars. Abbreviations: Fd: ferredoxin. Silencing: ev: empty vector, g2: gltAp2 promoter, z: zwf promoter, x: xylA promoter, u: udhA promoter. Proteolysis: F: fabI-DAS+4, G: gltA-DAS+4, Z: zwf-DAS+4, U: udha_DAS+4, X: xylA-DAS+4. All results were obtained from microfermentations.

Figure 7 :

Identification of pathways responsible for NADPH and xylitol production in the “FZ” valve strain a) the impact of deletions of ydbK and fpr on specific xylitol production, b) the impact of pntAB overexpression on xylitol production. (c-d) “FZ” valve strains further modified for dynamic control over c) GltA levels and d) UdhA levels. ev: empty vector, z: zwf promoter gRNA, g2: gltAp2 promoter gRNA, u: udhA promoter gRNA. All results were obtained from microfermentations. (e-f) Stoichiometric flux models of e) cellular growth and f) stationary phase xylitol production in “FZ” valve strains. Pathway flux is relative to xylose uptake rates. During growth the majority of flux is through the pentose phosphate pathway (PPP), pyruvate dehydrogenase multienzyme complex (PDH) with minimal flux through the pentose membrane bound transhydrogenase. Upon dynamic control, a 4-fold increase in membrane bound transhydrogenase flux is accompanied by increased flux through Pfo (encoded by ydbK) and Fpr. Abbreviations: G6P: glucose-6-phosphate, 6-PGL: 6-phosphogluconolactone, 6PG: 6-phosphogluconate, GA3P: glyceraldehyde-3-phosphate , OAA: oxaloacetic acid.

Figure 8:

a) Stationary phase NADPH pools and pfo/ydbK promoter reporter GFP levels in engineered strains. NADPH pools and fluorescence were measured 24 hours post phosphate depletion in microfermenations. b) A conceptual model of 2-stage NADPH production in our engineered system. Glucose-6-phosphate dehydrogenase (encoded by the zwf gene) is normally responsible for the biosynthesis of a majority of NADPH. This irreversible reaction drives an NADPH set point, in which the SoxRS oxidative stress response is OFF (gray area). Dynamic reduction in Zwf levels reduces NADPH pools activating the SoxRS response, which in turn activates expression of Pyruvate ferredoxin oxidoreductase (Pfo, encoded by the ydbK gene) and NADPH dependent ferredoxin reductase (Fpr). Together Pfo and Fpr (operating in reverse) constitute a new pathway to generate NADPH as well as allow for continued pyruvate oxidation and generation of acetyl-CoA for entry into the tricarboxylic acid cycle (TCA cycle). NADPH flux is further enhanced by reducing fatty acid biosynthesis whose products inhibit the membrane bound transhydrogenase ( encoded by the pntAB genes). Activated PntAB uses the proton motive force to convert NADH from the TCA cycle to NADPH. NADPH can be used for bioconversions such as for xylitol production.

Figure 9:

Xylitol production in minimal media fed batch fermentations in instrumented bioreactors by a) the control strain expressing xylose reductase (DLF_Z0025, pCASCADE-ev, pHCKan-xyrA), b) the “FZ” valve strain (DLF_Z0025-fabI-DAS+4-zwf-DAS+4, pCASCADE-z, pHCKan-xyrA) c) the “FZ” valve strain also overexpressing the membrane bound transhydrogenase pntAB (DLF_Z0025-fabI-DAS+4-zwf-DAS+4, pCASCADE-z, pHCKan-xyrA, pCDF-pntAB). Biomass (black) and xylitol (blue) are given as a function of time. For b) & c) x’s and triangles represent the measured values of two duplicate runs.