Biosynthesis of cannabinoid precursor olivetolic acid in genetically engineered Yarrowia lipolytica

Screening enzymes to improve pool size of hexanoyl-CoA

Pool size of hexanoyl-CoA was proven to be a rate-limiting factor in both E. coli and S. cerevisiae^17,18. An acyl activating enzyme encoded by CsAAE1 from Cannabis sativa was characterized to catalyze the formation of hexanoyl-CoA from hexanoic acid, and the expression of CsAAE1 has been shown to successfully increase the titer of OLA by twofold in S. cerevisiae¹⁷. Hexanoic acid is toxic and the dosage of hexanoic acid (HA) should be optimized to minimize its negative impact on cell fitness¹⁸. Thus, CsAAE1 was co-expressed with CsOLS and CsOAC in Y. lipolytica supplemented with 0.5 mM, 1 mM, and 2 mM of HA. We found that OLA production was declined (Fig. 3a). The yeast grows slowly with HA supplementation, confirming the negative effect of HA on cell growth (Fig. 3b). To mitigate this effect, we delayed HA supplementation to 48 h. OLA production with 0.5 mM HA supplementation slightly increased to 0.13 mg/L (Fig. 3a), and the yeast growth fitness remains unchanged (Fig. 3c). Thus, the supplementation of 0.5 mM hexanoic acid at 48 h was used for further strain screening.

Nevertheless, codon-optimized CsAAE1 showed lower catalytic capacity in Y. lipolytica than in S. cerevisiae¹⁷. In order to optimize the conversion of HA to hexanoyl-CoA, we screened a panel of six enzymes that have been annotated with acyl-CoA synthetase activity. CsAAE3 encoding a peroxisomal enzyme can accept a variety of fatty acid as substrates including HA. We also removed the C-terminal peroxisome targeting sequence 1 (PTS1) and overexpressed the codon-optimized CsAAE3 in Y. lipolytica^19–21. Both the short-chain and long-chain fatty acyl-CoA synthetases, EcFadK and EcFadD from E. coli, have been shown to exhibit a catalytic activity on C6–C8 fatty acids¹⁸. The medium-chain fatty acyl-CoA synthetase ScFAA2 from S. cerevisiae and the Y. lipolytica native fatty acyl-CoA synthetase encoded by ylFAA1 (YALI0D17864g) were also tested in this study. Pseudomona sp, known as a superior environmental degrader for various carbon sources. The LvaE gene encoding the short-chain acyl-CoA synthetase of Pseudomonas putida KT2440 (PpLvaE, PP_2795 with uniport ID Q88J54), which is reported to active with C4-C6 carboxylic acids, was also selected²². Among these chosen hexanoyl-CoA synthetases, we found the overexpression of PpLvaE with CsOLS and CsOAC (strain YL110) resulted in the highest OLA production, an eightfold increase in OLA titer (1.07 mg/L) (Fig. 4a). P. putida has been reported to exhibit better solvent tolerance and grow under a number of harsh or hydrophobic conditions. The overexpression of CsAAE1 in S. cerevisiae only showed a twofold increase in OLA titer¹⁷. By contrast, PpLvaE outperforms other candidate enzymes in converting HA to hexanoyl-CoA. To the best of our knowledge, this is the first report that harnessed PpLvaE to overcome a rate-limiting step in OLA and cannabinoids pathway. Future enzymology study or directed evolution of PpLvaE may improve its catalytic function. Apparently hexanoic acid toxicity poses severe inhibition to cell growth, and the hexanoyl-CoA ligase needs to be further improved. An evolutionary approach to improve hexanoic acid tolerance and selection of variants of hexanoyl-CoA ligase will be important to solve this challenge. Alternative approach is to harness the endogenous β-oxidation pathway and target OA pathway into the peroxisome.

Fig. 4

Boosting hexanoyl-CoA and malonyl-CoA to improve olivetolic acid (OLA) production.

a OLA titers in engineered strains by expressing different hexanoyl-CoA synthetases for optimizing the conversion of hexanoic acid to hexanoyl-CoA. b OLA titers in engineered strains for enhancing the availability of malonyl-CoA. Error bars represent standard deviation of three biological replicates (n = 3).

Boosting malonyl-CoA to improve olivetolic acid production

Malonyl-CoA has been reported as the rate-limiting precursors for polyketides synthesis²³. Next, we used two strategies to improve malonyl-CoA supply. In the first strategy, two endogenous genes ylDGA1 (YALI0E32769g) and ylDGA2 (YALI0D07986g) encoding diacylglycerol acyltransferases were knocked out, in such way, we could minimize the acyl-CoA flux flowing into the TAG pathway which competes with OA pathway for malonyl-CoA^24,25. When the OA pathway (CsOLS–CsOAC–PpLvaE) was transformed into the strain YL111 with ylDGA2 knockout and YL112 strain with ylDGA1 and ylDGA2 double knockout, the resulting strains YL113 and YL114 showed a dramatic decline in OLA production (Fig. 4b). We speculate that that the increased acyl-CoAs, or acyl-ACPs in the DGA1/DGA2 knockout strains may strongly feedback inhibit the activity of the acetyl-CoA carboxylase^26,27, which has been reported in both prokaryotic and eukaryotic cells.

In the second strategy, the endogenous gene ylACC1 (YALI0C11407g) encoding acetyl-CoA carboxylase was overexpressed to facilitate the conversion of acetyl-CoA to malonyl-CoA²⁸. To minimize spliceosomal or transcriptional level regulation, the two internal introns of the native gene ylACC1 were removed. The engineered strain YL115 with overexpression of ylACC1 produced 1.9 mg/L of OLA, with a 1.8-fold improvement (Fig. 4b). Additionally, snf1 (AMP kinase) has been reported to phosphorylate and inhibit ACC activity^29,30. Recently, a snf1-resistant version of ACC mutant (ScACC1^S659A,S1157A) has been reported to boost intracellular malonyl-CoA and improve 3-hydroxyl propionate or fatty acids production in S. cerevisiae³¹. Thus, a SNF1-resistant version of ylACC1^S667A,S1178A was engineered and incorporated into strain YL116. In Y. lipolytica, we did find the SNF1 homolog protein (YALI0D02101). However, OLA production in strain YL116 failed to improve further (Fig. 4b), possibly due to some hidden or underexplored regulations in Y. lipolytica.

Effect of pH on olivetolic acid production

During fermentation, we found that the culture pH dramatically dropped to below 3.5. This strong acidity may negatively affect membrane permeability and strain performance^28,32. We next sought to control the medium pH by using either PBS buffer or CaCO₃³². Supplementation of 20 g/L CaCO₃ maintained stable pH and increased OLA titer by threefold, reaching 5.86 mg/L at 96 h, whereas PBS failed to improve OLA production (Fig. 5a). We speculate that PBS buffer may shift the cell metabolism or change membrane permeability³³, which negatively impact cell performance.

Shake flask cultivations and pH control

For performing shake flask cultivations, seed culture was carried out in the shaking tube with 2 mL seed culture medium at 30 °C and 250 r.p.m. for 48 h. Then, 0.6 mL of seed culture was inoculated into the 250 mL flask containing 30 mL of fermentation medium and grown under the conditions of 30 °C and 250 r.p.m. for 96 h. One milliliter of cell suspension was sampled every 24 h for OD₆₀₀ and desired metabolite measurement.

Seed culture medium used in this study included the yeast complete synthetic media regular media (CSM, containing glucose 20.0 g/L, yeast nitrogen base without ammonium sulfate 1.7 g/L, ammonium sulfate 5.0 g/L, and CSM-Leu 0.74 g/L) and complex medium (YPD, containing glucose 20.0 g/L, yeast extract 10.0 g/L, and peptone 20.0 g/L). Fermentation medium used in this study contained the yeast complete synthetic media regular media (CSM, containing glucose 40.0 g/L, yeast nitrogen base without ammonium sulfate 1.7 g/L, ammonium sulfate 1.1 g/L, and CSM-Leu 0.74 g/L) and complex medium (YPD, containing glucose 40.0 g/L, yeast extract 10.0 g/L, and peptone 20.0 g/L). To control the pH, 20 mM phosphate buffer saline (PBS, Na₂HPO₄–NaH₂PO₄) or 20 g/L CaCO₃ was used, respectively.

Yeast transformation and screening of high-producing strains

The standard protocols of Y. lipolytica transformation by the lithium acetate method were described as previously reported^16,41. In brief, one milliliter cell was harvested during the exponential growth phase (16–24 h) from 2 mL YPD medium (yeast extract 10 g/L, peptone 20 g/L, and glucose 20 g/L) in the 14-mL shake tube, and washed twice with sterile 100 mM phosphate buffer (pH 7.0). Freshly cultivated yeast colony lawns picked from overnight-grown YPD plates could also be used for genetic transformation. Then, cells were resuspended in 105 μL transformation solution, containing 90 μL 50% PEG4000, 5 μL lithium acetate (2 M), 5 μL boiled single stand DNA (salmon sperm, denatured) and 5 μL DNA products (including 200–500 ng of plasmids, linearized plasmids or DNA fragments), and incubated at 39 °C for 1 h, then spread on selected plates. It should be noted that the transformation mixtures needed to be vortexed for 15 s every 15 min during the process of 39 °C incubation. The selected markers, including leucine, uracil and hygromycin, were used in this study. All engineered strains after genetic transformation were undergone PCR screening using the GoTaq Green PCR kits, and the strain with the correct gene fragment was selected to perform shake flask cultivation. For shaking tube cultivations, 100 μL seed cultures were inoculated into 5 mL fermentation media in a 50 mL tube.

Single-gene and multi-genes expression vectors construction

In this work, the YaliBrick plasmid pYLXP’ was used as the expression vector⁴². The process of plasmid constructions followed the YaliBrick gene assembly platforms¹⁶. In brief, recombinant plasmids of pYLXP’-xx (a single gene) were built by Gibson Assembly of linearized pYLXP’ (digested by SnaBI and KpnI) and the appropriate PCR-amplified or synthetic DNA fragments. Multi-genes expression plasmids were constructed based on restriction enzyme subcloning with the isocaudamers AvrII and NheI. All genes were respectively expressed by the TEF promoter with intron sequence and XPR2 terminator, and the modified DNA fragments and plasmids were sequence-verified by Sanger sequencing (Quintarabio).

Gene knockout

A marker-free gene knockout method based on Cre-lox recombination system was used as previously reported^16,43. For performing gene knockout, the upstream and downstream sequences (both 1000 bp) flanking the deletion targets were PCR-amplified. These two fragments, the loxP–Ura/Hyr–loxP cassette (digested from plasmid pYLXP’-loxP-Ura/Hyr by AvrII and salI), and the gel-purified plasmid backbone of pYLXP’(linearized by AvrII and salI) were joined by Gibson Assembly, giving the knockout plasmids pYLXP’-loxP–Ura/Hyr–xx (xx is the deletion target). Next, the knockout plasmids were sequence-verified by Quintarabio. Then, the gene knockout cassettes were PCR-amplified from the knockout plasmids pYLXP’-loxP–Ura/Hyr–xx, and further transformed into Y. lipolytica. The positive transformants were determined by colony PCR. Knockout strains were built on top of the Ku70-deficient strains (Po1f background). Subsequently, plasmid pYLXP’-Cre was introduced into the positive transformants and promoted the recombination of loxP sites, which recycle the selected marker. Finally, the intracellular plasmid pYLXP’-Cre was evicted by incubation at 30^oC in YPD media for 48 h. Here, Ura is the uracil marker, and Hyr is hygromycin marker.

Genomic integration of desired genes

In this work, genomic integration of desired genes was performed in two different ways: site-specific genomic integration plasmids or application of pBR docking platform by linearizing the plasmid pYLXP’ with digested enzyme NotI. We constructed two genomic integration plasmids pURLA and pURLB, corresponding to the Ku70 and YALI0C05907g (encoding a hypothetical protein conserved in the Yarrowia clade) genomic sites, respectively. The procedure of using these two plasmids was similar as that of gene knockout protocol. The method of constructing integration plasmids was described in previous work. The application of pBR docking platform was achieved by linearizing the plasmid pYLXP’ with NotI restriction enzyme digestion. All the genes manipulated in this study are one copy.

HPLC quantification and LC–MS characterization of olivetolic acid

Cell densities were monitored by measuring the optical density at 600 nm (OD₆₀₀). The concentrations of olivetolic acid were measured by high-performance liquid chromatography (HPLC) through Agilent HPLC 1220. In detail, olivetolic acid was measured at 270 nm under 40 °C (column oven temperature) with a mobile phase containing 60% (v/v) methanol in water at a flow rate of 0.4 mL/min equipped with a ZORBAX Eclipse Plus C18 column (4.6 × 100 mm, 3.5 μm, Agilent) and the VWD detector.

To quantify the concentration of olivetolic acid, 0.5 mL whole cell sample with both cell pellet and liquid culture was taken. Subsequently, samples were treated with 2 U/OD zymolyase (2 h, 30 °C with shaking at 1000 r.p.m.), and then cell suspensions were added with 20% (w/v) glass beads (0.5 mm) and the cells were grinded with hand-powered electrical motor (VWR). Subsequently, the crude extracts were mixed with an equal volume of ethyl acetate (v/v), followed by vortex at room temperature for 2 h. Organic and inorganic layers were separated by centrifugation at 12,000 r.p.m. for 10 min. Samples were extracted three times. The combined organic layers were evaporated in a vacuum oven (50 °C) and the remainders were resuspended in 0.5 mL 100% methanol. Then, 100 μL of the sample were gently transferred into a HPLC vial insert and 5 μL were injected into HPLC for OLA quantification. Under this condition, the retention time for OLA is 10.8 min.

Olivetolic acids standards and samples were also characterized by the Perkin Elmer QSight LX50 UHPLC (ultra-high-performance liquid chromatography) with the QSight 210 Mass Spectrometer under negative ESI (electrospray ionization) mode. The Agilent Zorbax Eclipse XDB-C18 2.1 × 50 mm, 1.8 μm (P/N 981757-902) column was used for the UHPLC system. Under the UHPLC-MS system, the retention time for olivetolic acid is 1.8 min and the characteristic mass/charge (m/z) ratio for OLA is 223.5 and 179.5 (decarboxylated-CO₂ fragment). A detailed UHPLC-MS report for OLA could be found in the Supplementary note 2.

Statistics and reproducibility

General data analysis (means and standard deviation) was performed primarily by GraphPad Prism 9.4.0. All experiments were performed with biological triplicates and values were expressed as means ± standard errors.

This project is supported by the Bill & Melinda Gates Foundation Award (OPP1188443) and National Science Foundation (CBET-1805139). We thank Dr. Joshua Wilhide and Dr. LaCourse to help run the LC–MS characterization of OLA at the UMBC Molecular Characterization and Analysis Complex center (UMBC chemistry facility) and help prepare the LC–MS report.