3.1 Debottlenecking mevalonate pathway for squalene production
In yeast, squalene was primarily synthesized from the mevalonate (MVA) pathway (Fig. 1). Staring with acetyl-CoA condensation, yeast uses a number of critical enzymes to synthesize squalene, including acetoacetyl-CoA thiolase (Erg 10, YALI0B08536g), HMG-CoA reductase (YALI0E04807g), mevalonate kinase (Erg 12, YALI0B16038g), phosphomevalonate kinase (Erg 8, YALI0E06193g), mevalonate pyrophosphate decarboxylase (MVD1, YALI0F05632g), farnesyl pyrophosphate synthase (Erg20, YALI0E05753g), geranyl pyrophosphate synthase (YALI0D17050g) and squalene synthase (SQS1, YALI0A10076g). Genome annotation indicates thatY. lipolytica contains the complete mevalonate pathway (Fig. 1). In MVA pathway, HMG-CoA reductase was reported as the rate-limiting metabolic step in squalene accumulation (RODWELL, NORDSTROM, & MITSCHELEN, 1976). In addition, there was almost no squalene accumulated by native Y. lipolytica due to the quick consumption of squalene by downstream ergosterol synthase. After we overexpressed the endogenous squalene synthase gene (SQS ), squalene production was increased to 17.25 mg/L at 120 h with chemically-defined complete synthetic media (CSM-leu) in test tube. With this as a starting strain, we investigated the effect of three HMG-CoA reductases (encoded by HMG) on squalene production. The three HMGs were derived from Saccharomyces cerevisiae , Silicibacter pomeroyi and Y. lipolytica . Truncated form of HMG-CoA reductase devoid of N-terminal membrane targeting signal has been proven to be effective in improving isoprenoid production in Saccharomyces cerevisiae (encoded by SctHMG ) (Polakowski, Stahl, & Lang, 1998; Thompson, Kwak, & Jin, 2014). When co-expressed with endogenous SQS, the strain with the truncated HMG1 (SctHMG) led to squalene production at 83.76 mg/L (Fig. 2A), indicating that overexpression of HMG-CoA reductase was beneficial for squalene production. To test whether other sources of HMG-CoA reductase could display better functions, we co-expressed SpHMG from Silicibacter pomeroyi and endogenous ylHMG with SQS, respectively. A low yield of squalene (9.24 mg/L) was produced in the strain expressingSpHMG . This result was consistent with previous findings that HMG from Silicibacter pomeroyi was highly specific for NADH (Meadows et al., 2016) and this bacterial-derived enzymes could not be directly translated to yeast system. When endogenous ylHMG was co-expressed with SQS (strain HLYaliS01 ), the engineered strain yielded 121.31 mg/L squalene at 120 h in test tube, demonstrating the potential of using Y. lipolytica as a platform to synthesize various terpenes. We also tested the truncated form of ylHMGsequence (YALI0E04807p), of which the first 495 nucleotides that encode the 165 amino acid N-terminal domain responsible for membrane localization (ER targeting) were removed. The remaining C-terminal residues containing the catalytic domain and an NADPH-binding region (Gao et al., 2017) were overexpressed. We then overexpressed the truncated ylHMG (t495ylHMG ) to compare how the variation of ylHMG may improve squalene synthesis. Contrary to our hypothesis, removal of the N-terminal 495bp of ylHMG exhibits adverse effect on both squalene production and cell growth (Fig. 2A), indicating that the N-terminal membrane-binding domain plays a critical role in squalene synthesis.
In addition to the overexpression of the endogenous SQS and ylHMGgenes, we also tested whether the expression of other genes in the MVA pathway would improve squalene production, including ylErg8 encoding phosphomevalonate kinase, ylErg10 encoding acetoacetyl-CoA thiolase, Erg12 encoding mevalonate kinase, and ylErg20 encoding farnesyl pyrophosphate synthetase; ylGPS encoding geranyl pyrophosphate synthase. And ylErg8, ylErg10, Erg12 and ylErg20 from S. cerevisiae were also overexpressed to compare how the variation of these genes may enhance squalene synthesis. As shown in Fig. 2B, co-overexpression of ylErg8, ylErg10, Erg12 could not further improve squalene synthesis, regardless of the source of the gene. Among all of these combinations (Fig. 2B), the highest squalene production was obtained for the strain in which ylGPS and SQS-ylHMG1 were overexpressed, with titer of 107.08 mg/L and a specific production of 36.24 mg/g DCW, which is still lower than the strain only expressing SQS-ylHMG1. These results indicate that sequential overexpression of the genes involved in the MVA pathway could not further improve the carbon flux toward squalene, possibly due to the stringent regulation of MVA pathway at multiple nodes, including ergosterol-mediated feedback inhibition or SREBP-related transcriptional repression.