Enabling malic acid production from corn-stover hydrolysate in Lipomyces starkeyi via metabolic engineering and bioprocess optimization

Abstract

Background: Lipomyces starkeyi is an oleaginous yeast with a native metabolism well-suited for production of lipids and biofuels from complex lignocellulosic and waste feedstocks. Recent advances in genetic engineering tools have facilitated the development of L. starkeyi into a microbial chassis for biofuel and chemical production. However, the feasibility of redirecting L. starkeyi lipid flux away from lipids and towards other products remains relatively unexplored. Here, we engineer the native metabolism to produce malic acid by introducing the reductive TCA pathway and a C4-dicarboxylic acid transporter to the yeast.

Results: Heterogeneous expression of two genes, the Aspergillus oryzae malate transporter and malate dehydrogenase, enabled L. starkeyi malic acid production. Overexpression of a third gene, the native pyruvate carboxylase, allowed titers to reach approximately 10 g/L during shaking flasks cultivations, with production of malic acid inhibited at pH values less than 4. Corn-stover hydrolysates were found to be well-tolerated, and controlled bioreactor fermentations on the real hydrolysate produced 26.5 g/L of malic acid. Proteomic, transcriptomic and metabolomic data from real and mock hydrolysate fermentations indicated increased levels of a S. cerevisiae hsp9/hsp12 homolog (proteinID: 101453), glutathione dependent formaldehyde dehydrogenases (proteinIDs: 2047, 278215), oxidoreductases, and expression of efflux pumps and permeases during growth on the real hydrolysate. Simultaneously, machine learning based medium optimization improved production dynamics by 18% on mock hydrolysate and revealed lower tolerance to boron (a trace element included in the standard cultivation medium) than other yeasts.

Conclusions: Together, this work demonstrated the ability to produce organic acids in L. starkeyi with minimal byproducts. The fermentation characterization and omic analyses provide a rich dataset for understanding L. starkeyi physiology and metabolic response to growth in hydrolysates. Identified upregulated genes and proteins provide potential targets for overexpression for improving growth and tolerance to concentrated hydrolysates, as well as valuable information for future L. starkeyi engineering work.

Keywords: Lipomyces starkeyi; Machine learning medium optimization; Malic acid production; Oleaginous yeast.

Fig. 1

Engineered malic acid biosynthetic pathway and production comparison among mutants. (A). The engineered rTCA pathway and the C4-dicarboxylic acid transporter in L. starkeyi. (B). Malic acid titers among different engineered L. starkeyi strains grown in glucose containing minimal media. Abbreviations: aoMDH, A. oryzae malate dehydrogenase; aoMT, A. oryzae malate transporter; anPYC, A. niger pyruvate carboxylase; lsPYC, L. starkeyi pyruvate carboxylase. Error bars represent standard deviations (n = 8 for WT and n = 12 for transgenic isolates constructs)

Fig. 2

Malic acid production dynamics in buffered mock hydrolysates. (A). 0.5 L bioreactor production at various pH setpoints. Dashed lines indicate glucose and solid lines indicate malic acid. (B). Malic acid time dynamics and (C). sugar consumption time dynamics in buffered shake flask conditions using mock hydrolysate. Dashed lines indicate glucose and solid lines indicate xylose. (D). Malic acid titers of the lsPYC + and anPYC strains in buffered shaking flask conditions. A long lag phase was observed in the growth due to a starting OD₆₀₀ of 0.01. Error bars represent standard deviations (n = 6 for B, n = 5 for C up to 72 h, n = 3 after 72 h)

Fig. 3

Comparison of malic acid production and sugar consumption on mock and DDR hydrolysate in 0.5 L bioreactors. Malic acid titers, DCW, and glucose and xylose consumption in (A). Mock (GX– glucose & xylose) and low gravity DDR hydrolysates over 70 h and, (B). high gravity DDR hydrolysate from 70 h to over 320 h. The high gravity DDR had 3x the starting sugar concentration as the low gravity DDR. The * represents timepoints at which samples were collected for omic analyses. Error bars represent standard deviation (n = 2)

Fig. 4

PCA of omic data collected from the mock and DDR hydrolysate bioreactor cultivations. (A). Transcriptomic data with the outlier removed, (B). Proteomic data, and (C). Metabolomic data. The circles indicate samples collected from the same fermentation timepoints in both the mock and DDR hydrolysates. Blue - early phase, Green - mid phase, Red - late phase, Yellow - high gravity DDR fermentation. GX represented the glucose-xylose mock hydrolysate and DDR represented the real hydrolysate. Dashed lines indicate different fermentation regimes noted in the figure panels. Legend applies to all panels

Fig. 5

ART medium optimization results. (A). Malic acid titers over the course of three DBTL cycles. (B). The percentage of media designs displaying growth across the cycles. (C). Dynamic sampling of malic acid production from the highest titer conditions identified in (A). Blue dotted lines represent the two highest titer conditions from the experiment. Error bars represent standard deviation (n = 3)

Fig. 6

Tolerances to trace elements and metal media components. Biomass DCW plotted versus the trace element concentration for each trace element varied in the media. Y-axes indicate the obtained DCW (g/L) and the X-axes represent the specified component concentrations. Black solid lines indicate the DCW threshold for growth/no growth in a condition. The majority of medium components had growth across their concentration range except for boric acid indicated with the red solid line delimitating point of inhibitory effect. Data points represent the average of triplicate flasks. Error bars have been omitted for clarity