Rewiring Yarrowia lipolytica toward triacetic acid lactone for materials generation

Significance

Polyketides are important molecules for both their bioactive traits and their potential as chemical building blocks. However, production of these molecules through chemistry and biocatalysts is restricted in yield and titer. Here, we demonstrate that the nonconventional yeast Yarrowia lipolytica can serve as a potent host for such production. This work provides a comprehensive evaluation of three separate pathways toward acetyl–CoA and malonyl–CoA in this host, enabling high-titer production of triacetic acid lactone. Beyond achieving unprecedented titers and appreciable yields, this production capacity allows for both purification from fermentation broth and conversion into a material using simple reaction conditions.

Abstract

Polyketides represent an extremely diverse class of secondary metabolites often explored for their bioactive traits. These molecules are also attractive building blocks for chemical catalysis and polymerization. However, the use of polyketides in larger scale chemistry applications is stymied by limited titers and yields from both microbial and chemical production. Here, we demonstrate that an oleaginous organism (specifically, Yarrowia lipolytica) can overcome such production limitations owing to a natural propensity for high flux through acetyl–CoA. By exploring three distinct metabolic engineering strategies for acetyl–CoA precursor formation, we demonstrate that a previously uncharacterized pyruvate bypass pathway supports increased production of the polyketide triacetic acid lactone (TAL). Ultimately, we establish a strain capable of producing over 35% of the theoretical conversion yield to TAL in an unoptimized tube culture. This strain also obtained an averaged maximum titer of 35.9 ± 3.9 g/L with an achieved maximum specific productivity of 0.21 ± 0.03 g/L/h in bioreactor fermentation. Additionally, we illustrate that a β-oxidation-related overexpression (PEX10) can support high TAL production and is capable of achieving over 43% of the theoretical conversion yield under nitrogen starvation in a test tube. Next, through use of this bioproduct, we demonstrate the utility of polyketides like TAL to modify commodity materials such as poly(epichlorohydrin), resulting in an increased molecular weight and shift in glass transition temperature. Collectively, these findings establish an engineering strategy enabling unprecedented production from a type III polyketide synthase as well as establish a route through O-functionalization for converting polyketides into new materials.

The growing demand for renewable chemicals and fuels has spurred great interest in using cells as biochemical factories (1). Metabolic engineering enables this goal by rewiring cells’ metabolism toward desirable chemical compounds (2–4). Among possible molecules, polyketides are an interesting class of secondary metabolites produced by microbes and plants with native roles in processes such as cellular defense and communication (5–7). While many polyketides can serve as potent antibiotics, this class of molecules also encompasses chemicals with other useful properties such as pigments, antioxidants, antifungals, and other bioactive traits (5, 8). However, the use of polyketides in more unique and nonmedical applications has been partially limited due to low natural abundance and difficult cultivation of native hosts. Specifically, polyketide-producing organisms are typically unusual plants and microbial organisms that are not well-suited for high-level industrial production (6). Synthetic production of these molecules in model host organisms has also proven quite difficult with titers and yields insufficient for industrial production (10–13). Likewise, traditional chemical synthesis of polyketides is limited by low concentrations and challenging chiral centers (14). While the scale and price-point for pharmaceuticals can tolerate plant-based sourcing of polyketides or challenging syntheses, this is not an option for any larger-scale chemistry application.

Here, we focus on the interesting, yet simple, polyketide, triacetic acid lactone (TAL) as it is derived from two common polyketide precursors, acetyl–CoA and malonyl–CoA. TAL has been demonstrated as a platform chemical that can be converted into a variety of valuable products traditionally derived from fossil fuels including sorbic acid, a common food preservative with a global demand of 100,000 t (1, 15–18). However, meeting this annual demand using the low concentrations of TAL derived from native plants like gerbera daisies (9) would require four times the quantity of global arable land. As a result, utilization of polyketides for unique industrial applications including polymers, coatings, and even commodity chemical production has not been implemented despite the promising chemical nature of these molecules. To address these limitations, previous efforts have explored microbial production of TAL. However, these efforts have been restricted to conventional organisms [like Escherichia coli (10, 19) and Saccharomyces cerevisiae (11–13)] and are limited with respect to titer only reaching 5.2 g/L with low yields (12).

In this work, we explore the unique application of an oleaginous, nonconventional yeast (Yarrowia lipolytica) based on its potential for high flux through the key polyketide precursors, acetyl–CoA and malonyl–CoA. By investigating three distinct pathways toward CoA precursor formation along with targets hypothesized to enhance β-oxidation, we demonstrate the utility of a previously uncharacterized pyruvate bypass pathway for significantly increasing TAL production. After subsequent optimization, our final strain achieved over 35% of the theoretical conversion yield to TAL in unoptimized tube culture and achieved a maximum observed titer of 35.9 ± 3.9 g/L in bioreactor operation. We demonstrate that a higher-yield strain (43% of theoretical conversion yield in a test tube) is possible by overexpressing a β-oxidation–related target. Finally, we demonstrate the chemical opportunities gained by high polyketide titers for novel materials modification by O-functionalization of biosourced TAL with commodity poly(epichlorohydrin) to tune and upgrade thermal properties of the parent material. This work both establishes a host organism for polyketide overproduction and demonstrates the potential utility of polyketides for materials synthesis and modification.

Results and Discussion

Y. lipolytica Can Support TAL Production.

High lipid flux in oleaginous organisms like Y. lipolytica suggests a strong potential for these organisms to produce alternative acetyl–CoA-derived products like polyketides. Moreover, Y. lipolytica exhibits sufficient tolerance to many chemicals (20) including TAL to concentrations approaching the soluble limit (SI Appendix, Fig. S1). With these two features in place, we first established heterologous TAL production in Y. lipolytica through expression of the codon-optimized Gerbera hybrida 2-pyrone synthase gene, g2ps1. While this initial strain produced TAL (SI Appendix, Fig. S13), further amplifying the gene copy number to four enabled 2.1 g/L production in tube fermentations (defined medium including CSM, YNB, and glucose). This strain, named YT, was selected as the starting point for further metabolic engineering work. Additional studies of previously characterized mutants of g2ps1 established in E. coli (10) were also tested but produced lower titers than the wild-type allele (SI Appendix, Fig. S2).

A Previously Uncharacterized Y. lipolytica Pyruvate Bypass Pathway Improves TAL Production.

To increase metabolic flux through acetyl–CoA and malonyl–CoA, we investigated three independent metabolic engineering strategies (Fig. 1). First, we explored the citrate route—a pathway that has been extensively studied for its capacity to increase lipid production in Y. lipolytica (21–23). Overexpression of ACC1, which codes for the enzyme that converts acetyl–CoA to malonyl–CoA, has also been shown to promote lipid production (24, 25). When the pathway genes (ACL1, ACL2, and AMPD) were concurrently overexpressed, TAL production was significantly reduced and only marginally improved with the addition of ACC1 (Fig. 2A).

Fig. 1.

Strain engineering scheme to evaluate native Yarrowia lipolytica pathway potential for the production of TAL. Four overall schemes were tested in this work. Illustrated here are the three anabolic pathways targeted for overexpression in this work: the citrate route (shown in blue), the pyruvate dehydrogenase complex (green), and the pyruvate bypass pathway (purple). Additionally, shown in red are two potential β-oxidation up-regulation targets. These color schemes are maintained in future figures to enable continuity and rapid identification.

Fig. 2.

Difference of means plots demonstrating the effect of overexpressing acetyl–CoA production pathways. TAL titers were measured following 96-h tube fermentations in defined media and presented as the increase in titer over the YT parental strain (the color scheme used in Fig. 1 has been retained). Error bars represent the SE of n ≥ 2. Significance was tested using Dunnett’s test, *P < 0.05, **P < 0.01, ***P < 0.001. (A) The effect of sequential overexpression of genes involved in the citrate pathway. (B) The effect of sequential overexpression of pyruvate dehydrogenase complex genes. (C) The effect of sequential overexpression of pyruvate bypass pathway genes in a full combinatorial fashion; i.e., every combination of five potential acetylaldehyde dehydrogenase genes and two pyruvate decarboxylase genes was tested.

Second, we explored the pyruvate dehydrogenase (PDH) complex pathway [located in the mitochondria in S. cerevisiae (26)]. In contrast to the citrate route, this pathway has not been extensively studied in Y. lipolytica, but would theoretically enable a direct path to convert pyruvate to acetyl–CoA. The related alpha-ketoglutarate dehydrogenase complex, which shares one subunit with the PDH complex, has been previously overexpressed to promote alpha-ketoglutaric acid production (27), suggesting a similar strategy may be successful here. To test this approach, we established a coordinated overexpression of the different subunits for this complex (encoded by PDA1, PDE2, PDE3, and PDB1). By combining this pathway with ACC1 overexpression, overall TAL production was significantly improved by 23%, achieving 2.5 g/L (Fig. 2B).

Third, we investigated the pyruvate bypass pathway, which converts pyruvate to acetaldehyde through pyruvate decarboxylase (PDC), then to acetate through acetylaldehyde dehydrogenase (ALD), and finally to acetyl–CoA via acetyl–CoA synthetase (ACS) (26, 28). Previous work has targeted this pathway using heterologous enzymes (29); however, the function and potential of the native Y. lipolytica pyruvate bypass pathway has not been previously explored. While a single ACS gene had been previously characterized in Y. lipolytica (27), two PDC homologs (arbitrarily named PDC1 and PDC2) and five potential ALD homologs were identified based on previous yeast homology studies (30). Next, we established a full combinatorial assembly of this pathway in the YT strain background. Unlike the previous two approaches, ACC1 overexpression did not consistently increase production for all combinations tested. Nevertheless, four genetic combinations emerged as the top TAL-producing strains including ACS1, ALD5, PDC2, ACC1 (64.7% improvement over YT), ACS1, ALD3, PDC1, ACC1 (61.1% improvement), ACS1, ALD2, PDC2, ACC1 (31.7% improvement), and ACS1, ALD3, PDC2, ACC1 (17.9% improvement) (Fig. 2C). The top strain from this effort (YT- ACS1, ALD5, PDC2, ACC1) produced 2.8 g/L of TAL in tube fermentations, equivalent to 30.4% of the theoretical yield.

Modification of β-oxidation Can Likewise Improve TAL Production.

As an alternative (and potentially complementary) approach to increase acetyl–CoA pools, we targeted participants in the β-oxidation pathway for overexpression. Specifically, we evaluated the transcription factor Por1 [reported in other hosts to increase polyketide formation (31)] and the peroxisomal matrix protein Pex10. When overexpressed in the YT background, POR1 had no effect on TAL production, whereas PEX10 overexpression increased TAL titer by 22% (2.4 g/L) (Fig. 3), suggesting β-oxidation up-regulation as a strategy for acetyl–CoA recycling if it cannot be shuttled away from lipid synthesis effectively. This result is intriguing as Pex10p is not directly involved in the catalytic conversion of fatty acids to acetyl–CoA and thus provides an area for further biochemical study.

Fig. 3.

Difference of means plot demonstrating the effect of overexpressing β-oxidation targets. TAL titers were measured following 96-h tube fermentations in defined media and presented as the increase in titer over the YT parental strain. Error bars represent the SE of n ≥ 2. Significance was tested using Dunnett’s test, **P < 0.01.

Strain Engineering Effectively Diverted Flux from Lipids to TAL.

To evaluate the true efficacy of these four rewiring strategies, both lipid and TAL production were evaluated from tube fermentations. We observed a clear, inverse correlation between TAL titer and lipid titer with an R² of 0.89 (Fig. 4). Collectively, these results demonstrate that the PDH complex, pyruvate bypass, and PEX10 overexpressions can divert acetyl–CoA from lipids into TAL, whereas the citrate pathway is strongly coupled to lipid formation.

Fig. 4.

Lipid production as a function of TAL production. Following defined media tube fermentation of YT and strains containing the overexpressions outlined in Fig. 1, average TAL titer and average total lipids were assessed. An inverse correlation between TAL titer and lipid titer was observed, R² = 0.89. Error bars represent the SD of n = 3.

Culture Conditions Differentially Alter TAL Production.

Y. lipolytica is highly responsive to environmental factors, and thus we evaluated a series of conditions for increased TAL production. First, we evaluated the impact of nitrogen starvation, a strategy commonly used in Y. lipolytica to induce lipid formation (22). Under these conditions, the impact was varied across the various rewiring schemes (SI Appendix, Fig. S3A) with the most significant improvement observed in the PEX10 overexpression strain (Fig. 5A). When grown in C20N2 media, this strain achieved a greater than twofold increase over the YT strain under normal conditions (reaching 4.1 g/L in a test tube, 43.4% theoretical yield). Further nitrogen limitation did not improve TAL production as a result of the expense to growth.

Fig. 5.

Impact of nitrogen starvation and acetate spike on TAL production. TAL production under different conditions was assessed following tube fermentation in defined media. Error bars represent the SD of n = 3. Statistical significance was determined by a Dunnett’s test; each new condition was compared with the relevant control (C20N5 in the case of nitrogen limitation and glucose in the case of feeding spikes), **P < 0.01, ***P < 0.001. (A) Nitrogen limitation enhances the effect of gene overexpressions related to β-oxidation. (B) Feeding assay demonstrating the effect of adding 10 g/L carbon molar equivalent of glucose as a feeding spike 24 h into standard tube fermentation.

Second, a series of supplements (SI Appendix) and different carbon sources were tested as spikes or sole carbon sources along with glucose as a control (SI Appendix, Fig. S4). The largest gains in TAL titer were realized through providing an acetate spike when the pyruvate bypass pathway was overexpressed. Under these conditions, the top strain from the pyruvate bypass pathway produced 4.9 g/L TAL in a test tube (representing over 35% of the theoretical conversion yield calculated from both glucose and acetate fed) (Fig. 5B). Analysis of acetate consumption indicates this result is not simply due to acetate acting as a carbon source for TAL production, as a substantial portion of the fed acetate still remains at the end of the fermentation (SI Appendix, Fig. S5). This result suggests a more regulatory or redox-related impact on metabolism. Further to this point, in S. cerevisiae, acetate feeding has been shown to induce changes to metabolism mediated through protein acetylation (32, 33), a possible mechanism to be explored here. Additionally, acetylaldehyde dehydrogenase activity assays suggest pathway engineering (especially with ALD5) resulted in altered redox cofactor usage favoring NAD+ over NADP+ (SI Appendix, Fig. S6). Thus, an overall redox and regulatory mechanism may explain the almost twofold increase in TAL production observed in this study. Intriguingly, this improvement was not seen under nitrogen-limited conditions (SI Appendix, Fig. S3B), suggesting that these two strategies (acetate feed and nitrogen starvation) are not compatible.

Bioreactor Cultivation Boosts Overall TAL Titer.

Fermentation of the pyruvate bypass overexpression strain (YT- ACS1, ALD5, PDC2, ACC1) with acetate spiking was scaled up to the 3-L bioreactor scale (YP, 18% glucose, 13.7 g/L acetate spike). After optimization, fermentation resulted in production of 35.9 ± 3.9 g/L of TAL (Fig. 6A). In this scale-up, we achieve a substantially improved productivity over the previous batch cultures, upward of a fourfold increase to a glucose-phase maximum specific productivity of 0.21 ± 0.03 g/L/h (SI Appendix, Table S4). Although a long fermentation time was necessary, this timescale is comparable to a previously published TAL study (12). The overall productivity reported here [0.12 g/L/h (SI Appendix, Table S4)] improves upon the previous report in S. cerevisiae by greater than sixfold (12). As this production level is well outside the soluble range of TAL, substantial in situ precipitation occurred, an attractive feature for industrial production, but a unique source of sampling error. Likewise, we observed a diauxic shift from glucose to (produced) citrate utilization (Fig. 6A) that leads to an increased production as cell viability stagnates and even decreases throughout the process (Fig. 6B).

Fig. 6.

Bioreactor cultivation of pyruvate bypass overexpression strain. YT-ACS1, ALD5, PDC2, ACC1 was fermented in a 3-L bioreactor with YP media, 180 g/L glucose, and a 13.7 g/L sodium acetate spike at 36 h. This figure demonstrates a representative run with a duplicate presented in SI Appendix, Fig. S7. (A) Concentrations of TAL, citrate, and glucose were determined from three independent samples taken at each time point; error bars represent SD of n = 3. (B) Viable cell count was determined by plating different dilutions of sample; error bars represent SD of n ≥ 2.

Biosourced TAL Can Be Incorporated into a Polymer Through O-Functionalization.

Finally, we leveraged the newfound bulk-availability of polyketides such as TAL to demonstrate their utility in the modification of polymer properties. TAL serves a dual role as both a polymer modifier but also a functional adduct for later chemical derivatization owing to its lactone and unsaturated functionalities (34). TAL was rapidly extracted and purified from fermentation broth and ultimately used for the modification of structure and properties of commodity poly(epichlorohydrin). This was achieved through heat and an activating organic base 1,8-Diazabicyclo(5.4.0)undec-7-ene (DBU) (Fig. 7A) (34, 35). In this process, the displacement of chloride through the O-functionalization of TAL was evident spectroscopically, chromatographically, and thermally through the use of NMR spectroscopy (Fig. 7B and SI Appendix, Fig. S8), size exclusion chromatography (SEC), and differential scanning calorimetry (DSC) (Table 1), respectively. The amount of TAL incorporated along the poly(epichlorohydrin) backbone was tuned stoichiometrically from 16 to 83% by mole resulting in poly[(epichlorohydrin)-co-(epoxy triacetic acid lactone)] or PETAL. We measured the compositionally dependent changes in the glassy amorphous solid based on molecular weight and glass transition temperature of PETAL. Molecular weight was seen to increase proportionately to the amount of TAL incorporated as measured by size exclusion chromatography with multiangle light-scattering detection to yield absolute number-average molecular weights (M_n) (Table 1). While poly(epichlorohydrin) exhibits a native glass transition temperature (T_g) of −30 °C, this value was markedly increased and strongly dependent on TAL incorporation with the highest TAL composition (83% by mole) exhibiting a T_g of 70 °C. The final product can be formed into a film and is seen to exhibit a unique hue and relative transparency (Fig. 7C). In particular, the orange color of the polymer is due to the native color of covalently bound TAL, as no other characteristic spectroscopic differences were observed via UV-Vis spectroscopy (SI Appendix, Fig. S9). It should be noted that the reaction rate obtained with this purified, biosourced TAL was comparable to a reaction conducted with commercially sourced TAL. Moreover, the biosourced PETAL structure and properties fall exactly along the trend observed using commercially sourced TAL, again supporting that there is no difference (SI Appendix, Fig. S10). Finally, the residual lactone and unsaturated functionality from TAL repeat units offer a unique future strategy toward further modification of material properties.

Fig. 7.

Production of poly[(epichlorohydrin)-co(epoxy triacetic acid lactone)] or PETAL. (A) Reaction scheme to create PETAL. (B) H NMR characterization of PETAL which shows distinct new peaks and shifts from the start molecules. (C) Photo of PETAL pressed into a film.

Table 1.

Characteristics of copolymers in comparison with PECH

Monomer pairs	Monomer feed* ECH:TAL:DBU	Polymer composition† ECH:TAL	Mn‡ (g/mol)	Tg§ (°C)
PECH		1: 0	19,700	−30
P(ECH-TALcom)	1: 1.5: 0.75	1: 1	23,700	30
PECH		1:0	7,200	−30
P(ECH-TALcom)	1: 1.5: 1	1: 5	9,400	70
P(ECH-TALbio)	1: 0.5: 1	6: 1	7,900	−11

Polymer characteristics were measured for PETAL.

^*Determined by gravimetry.

^†Determined by ¹H NMR spectroscopy.

^‡Number-average molecular weight determined by size exclusion chromatography in chloroform using light-scattering and differential refractometer detectors.

^§Thermal properties determined by differential scanning calorimetry.

Methods

A full Materials and Methods section is provided in the SI Appendix.

Strain Engineering and Analysis.

Strains used in this study were constructed in the wild-type Y. lipolytica strain, PO1f, through random genomic integration employing a sequential overexpression strategy. TAL concentrations were assessed via reverse-phase HPLC following 96-h tube fermentations at 28 °C in defined media containing 2% glucose. Bioreactor fermentations were performed at the 3-L scale at pH 6.5 in YP media with 18% glucose and a 13.7 g/L sodium acetate spike at 36 h.

Materials Generation.

Biosourced TAL was purified from fermentation broth through an ethyl acetate/acetic acid extraction. Commodity poly(epichlorohydrin) was functionalized with TAL using the activating organic base 1,8-Diazabicyclo(5.4.0)undec-7-ene and the resulting material characterized by NMR, differential scanning calorimetry, UV-Vis spectroscopy, and size exclusion chromatography.

Conclusions

In summary, this work demonstrates the use of an oleaginous organism for high-level production of an acetyl–CoA and malonyl–CoA-derived polyketide. Moreover, we establish a previously uncharacterized pyruvate bypass pathway as superior for rewiring CoA flux from lipid biosynthesis and into high-level TAL production reaching a titer of 35.9 ± 3.85 g/L in a bioreactor and overall yield of 0.164 g/g in a tube. Additional engineering efforts related to the β-oxidation pathway increased yields to 0.203 g/g. The achieved titer far exceeds previous efforts in the field with conventional organisms (a summary of the achieved titers and yields in this work is provided in SI Appendix, Table S4). This high-level production enabled rapid purification and conversion into a unique polymer with favorable molecular weight and glass transition temperature. This work and resulting strain provides a path forward for microbial production of other acetyl–CoA and malonyl–CoA-derived polyketides for novel applications such as polymers and chemical conversion.