Construction and optimization of the heterologous β-Arbutin production pathway
According to the literature 14, chorismate can be converted into β-Arbutin through three enzymes: chorismate lyase (UbiC) can transform chorismate into p-hydroxybenzoic acid (pHBA) while 4-hydroxybenzoate 1-hydroxylase (MNX1) and hydroquinone glucosyltransferase (AS) are involved in decarboxylation of pHBA to hydroquinone (HQ) and its subsequent glycosylation into β-Arbutin. Since K. phaffii naturally possesses a shikimate pathway capable of supplying chorismate, introducing this pathway artificially could enable the production of β-Arbutin in K. phaffii.
We introduced these three heterologous enzymes into K. phaffii and verified their functions (data not shown). Subsequently, all enzymes were introduced into K. phaffii, and analysis of β-Arbutin were conducted on the fermentation supernatant. As shown in the HPLC analysis chromatogram, a new peak with a retention time of 4.7 min appeared in our engineered strain, aligns with the retention time of the standard whereas the control did not (Figure S1). The titer of β-Arbutin and the biomass of the strain UA1-1 were detected in the whole fermentation period, revealing an accumulation of 1.57 g/L β-Arbutin at 120 h (Fig. 2A and B). Since the production reached a steady state after 120 h, we chose to use 120 h as the cycle for subsequent fermentation process. In addition, we also conducted fed-batch fermentation on UA1-1 and achieved an initial β-Arbutin yield of 68.02 g/L (Figure S2), demonstrating the production capability of the initial strain.
In multi-step biosynthetic pathways, the titer of the target compound can be affected by losses of intermediates, which may occur due to diffusion, oxidation and other factors. To improve the conversion rate of target compound, consecutive enzymes can be expressed as fusion with linker regions24,25. In β-Arbutin pathway, it was observed that the intermediate HQ was prone easy oxidation. During fermentation, we observed a gradual darkening of the medium, coinciding with the color of p-benzoquinone (the oxidation product of HQ). We hypothsized that this transformation could negatively affect the synthesis efficiency of β-Arbutin. Therefore, we employed substrate channels through fusion of MNX1 and AS using peptide linker GGGS. To this end, the fusion of MNX1 and AS with linker GGGS was achieved using distinct sequence orders (Fig. 3A). Fermentation was then conducted to compare the differences in β-Arbutin yield among UA1-1, UA2-1, and UA2-2. A β-Arbutin concentration of 1.88 g/L was obtained in UA2-1, resulting in a 19.7% improvement compared to UA1-1 (Fig. 3B). To validate our conclusion, we amplified the differences by supplementing the precursor substance pHBA. Specifically, 2 g/L of pHBA was added to the medium at 24 h of methanol fermentation. The β-Arbutin titer was measured at 48 h and 72 h, as shown in Fig. 3D. At 48 h, the titer of UA2-1 was 19.7% higher than UA1-1 and it remained 12.6% higher at 72 hours, demonstrating the effectiveness of the substrate channel MNX1-GGGS-AS. However, the titer of UA2-2 was 26.7% lower than UA1-1. Notably, the β-Arbutin titer of UA2-1 strain reached 3.44 g/L at 72 h, significantly surpassing the titer achieved at 120 h without pHBA added. This observation highlights that the insufficient supply of pHBA is a significant limiting factor for β-Arbutin production at present.
Engineering the shikimate pathway to improve β-Arbutin production.
The intermediate of the shikimate pathway, chorismate, is the direct precursor of pHBA. Our goal in this regulation was to maximize the metabolic flux of chorismate for the synthesis of β-Arbutin. By exploring KEGG, we identified endogenous genes involved in the biosynthesis of chorismate in K. phaffii, such as 3-deoxy-arabino-heptulonate 7-phosphate synthase (DHAPS1, PAS_chr1-4_0218; DHAPS2, PAS_chr2-1_0473; DHAPS3, PAS_chr3_0936), pentafunctional AROM polypeptide (aroM, PAS_chr3_0506) and chorismate synthase (CRS, PAS_chr2-1_0637). DHAP synthase catalyzes the condensation of E4P and PEP, representing the initial step in the shikimate pathway. Previous studies have emphasized the key role of this condensation in Y. lipolytica4. Aro4K229L is a DHAP synthase isoform derived from S. cerevisiae. Previous studies have shown that modifying specific residues can reduce the feedback inhibition of Aro426. The K229L mutation in this protein renders it insensitive to phenylalanine and tyrosine while still maintaining overall activity. This mutation was successfully utilized in S. cerevisiae to enhance the production of various aromatic products27–29. The pentafunctional peptide AroM is known to catalyze five consecutive reactions in the shikimate pathway, serving as a natural scaffold for substrate channeling in yeast30. The final step of the shikimate pathway is facilitated by the chorismate synthase CRS. Relevant studies indicate that overexpression of CRS can enhance the production of p-coumaric acid in S. cerevisiae31. To identify enzymes with superior performance, six potential genes based on the UA2-1 strain were overexpressed, and their impact on β-Arbutin productivity was assessed.
For strain UA2-1, it was observed that both biomass and β-Arbutin titer had not reached saturation at 120 h (Fig. 2A and B), potentially influencing the differences between strains. Compared to other engineered strains of K. phaffii under similar culture conditions, UA2-1 exhibited a biomass of less than 30 at 120 h, significantly lower than others with a biomass of 40–50 in same timeframe32,33. We hypothesized that the growth inhibition of UA2-1 could be attributed to the toxicity of p-benzoquinone, consequently impacting β-Arbutin titer. To accelerate the saturation of OD600, we adjusted the initial OD600 from 0.2 to 1.0 in our fermentation. As shown in Fig. 4A, UA3-3 showed a substantial increase in β-Arbutin production (3.58 g/L) compared to UA2-1(2.51 g/L), marking a 43.2% improvement. Among the five enzymes individually overexpressed, DHAPS3 (3.58 g/L) and DHAPS2 (3.09 g/L) increased β-Arbutin production significantly. CRS demonstrated a 12.7% improvement (2.83 g/L) over UA2-1, DHAPS1 showed an 8.5% increase (2.72 g/L) and AroM exhibited an 1.5% increase (2.54 g/L). This indicates that the condensation reaction involving DHAPS3 serves as the rate-limiting step in the shikimate pathway in K. phaffii, underscoring the importance of the condensation of PEP and E4P for production of β-Arbutin. According to the literature, the co-expression strategy of screening pathways has proven effective in enhancing the yield of target products in K. phaffii32. Based on the above results, we combined the effective genes with DHAPS3 to obtain optimal strains (Fig. 4A). As shown in Fig. 4A and B, although the β-Arbutin titer did not decrease, the overexpression of these genes imposed varying degrees of metabolic burden on the strains. These strategies did not always improve the yield in all combinations tested. However, the strain UA3-10 produced 4.32 g/L of β-Arbutin within 120 h, which is a substantial 72.1% increase compared to UA2-1.
Enrichment of the precursor PEP and E4P and UDPG pool
The de novo synthesis of β-Arbutin requires two precursors, E4P and PEP, derived from the PPP (pentose phosphate pathway) and glycolysis pathways. PEP participates in various pathways such as glycolysis and shikimate pathway. The initial step in shikimate pathway is the condensation between PEP and E4P to produce DHAP. Notably, the final step of the reaction catalyzed by AroM also utilizes PEP, indicating its continuous replenishment in shikimate pathway. Our objective is to boost PEP supply by knocking out pyruvate kinase (Prk, PAS_chr2-1_0769), overexpressing phosphoenolpyruvate carboxykinase (Pyk, PAS_FragB_0061) and the pyruvate carboxylase (Pyc, PAS_chr2-2_0024). To select the most effective strategy for increasing PEP supply, we initially selected UA2-1 as the chassis to validate these strategies (Fig. 5A). Upon overexpressing Pyk or knocking out Prk in UA2-1, the β-Arbutin titer increased to a certain extent. Specifically, it reached 3.06 g/L in UA4-3 (22.2% higher than UA2-1) and 2.78 g/L in UA4-2 (11.0% higher than UA2-1). However, overexpression of Pyc only resulted in a β-Arbutin titer of 2.31 g/L, suggesting that overexpressing Pyc does not significantly impact the metabolic flux of PEP. Therefore, we overexpressed Pyk into UA3-10 (Fig. 5A). Surprisingly, this did not lead to an increase in β-Arbutin production in UA4-3. This indicates that although overexpression of Pyk helps to increase the metabolic flux of PEP, PEP itself is not currently the main limiting factor for increasing the β-Arbutin titer.
E4P is an intermediate in the PPP. According to metabolic flux analysis, the carbon flux available for E4P in yeast is estimated to be at least an order of magnitude lower than that of PEP34. Numerous studies have also confirmed that E4P is the primary limiting substrate and bottleneck for the biosynthesis of aromatic derivatives in various microbes35–37. To augment the production of aromatic compounds in yeast27,38,39, studies have employed the strategy of overexpressing transketolase. The transketolase (Tkt, PAS_chr1-4_0150) facilitates the bypass of the oxidative phase of PPP, converting fructose-6-phosphate (F6P) and glyceraldehyde-3-phosphate (G3P) into E4P and xylulose-5-phosphate (Xu5P). Therefore, we overexpressed Tkt into UA3-10 to increase the supply of E4P. However, due to kinetic constraints inherent in the opposing reaction, the influence of tkt appears to be limited27,34,38,40,41. The titer of β-Arbutin exhibited a notable decrease (2.17 g/L), reflecting a reduction of 37.4% compared to the UA3-10 control (Fig. 5A), which may be correlated with the preference of Tkt for catalyzing the opposite reaction by consuming E4P42. This outcome aligns with findings similar to those reported by Liu43.
UDPG plays a crucial role in the glucose transfer of β-Arbutin, representing the final step in its synthesis. In order to ensure sufficient availability of UDPG, we overexpressed the phosphoglucomutase (Pgm, PAS_chr1-4_0264) and UDP-glucose pyrophosphorylase (UGP, PAS_chr1-3_0122) in UA3-10, aiming to enhance the conversion of G6P to UDPG. However, as shown in Fig. 5A, the β-Arbutin titer exhibited unfavorable change. This suggests that UDPG may not serve as a limiting factor in the production of β-Arbutin in UA3-10.
Optimizing methanol concentration to enhance the yield of β-Arbutin
Methanol serves as the primary carbon source and inducer in the fermentation of most K. phaffii. Elevated methanol concentrations can result in the overproduction of formaldehyde, posing a risk of cell toxicity and disrupting normal growth and metabolism. Conversely, insufficient methanol levels can hinder characterization of the strain’s production capacity. Therefore, precise regulation of methanol concentration during fermentation is crucial to promote optimal cell growth, metabolism, and production yield in K. phaffii. Various studies have investigated the impact of different methanol concentrations on the synthesis of recombinant proteins in methylotrophic yeast cells44. Consequently, it becomes essential to optimize methanol concentration for β-Arbutin production. To assess the growth and production capabilities of UA3-10 under varying methanol concentrations (1%, 1.5%, 2%, 2.5% and 3%), growth curves and β-Arbutin titers were measured (Fig. 6A and B). The results indicated that the maximum β-Arbutin titer was achieved at 2% methanol concentration. Specifically, at this concentration, strain UA3-10 produced 6.32 g/L of β-Arbutin, which is a 46.3% increase compared to 1% concentration. The biomass gradually increased with the rise in methanol concentration from 1–2%. However, beyond 2% methanol concentration, the β-Arbutin titer showed no further increase. Moreover, at 3% methanol concentration, both biomass and β-Arbutin titers sharply decreased, potentially attributed to the accumulation of methanol and formaldehyde8,45. These findings highlight the importance of maintaining an appropriate methanol concentration for optimal β-Arbutin synthesis in K. phaffii. While increasing methanol supply at a suitable concentration enhances these processes, exceeding a certain threshold negatively affects growth and product synthesis adversely, and can even lead to cell death. It is noteworthy that methanol tolerance and bioconversion represent ongoing challenges in methylotrophic yeast biotransformation8, necessitating further research.
Overexpression of heterologous PHK pathway to increase β-Arbutin production
In order to identify the current bottleneck limiting β-Arbutin production, we overexpressed UbiC in the UA3-10 strain. Increasing precursor supply based on this approach may further enhance β-Arbutin production. Therefore, we further introduced the heterologous PHK (phosphoketolase) pathway to construct strain UA8-1.
As shown in Fig. 7, the production of β-Arbutin in UA8-1 showed an initial increase followed by a decrease, reaching the highest yield at 72h, but only 1.25 g/L. The titer of β-Arbutin in UA8-1 is much lower than the control. Similar to increasing E4P supply by overexpress Tkt, the titer of β-Arbutin also suffered a significant decrease. The oxidative PPP is a major source of reductive power. Therefore, we speculate that the reason for the decrease in β-Arbutin production may be due to the introduction of the PHK pathway, which interferes with the generation of NADPH in the oxidative pentose phosphate pathway, thereby limiting the production of β-Arbutin. Therefore, using fructose-6-phosphate as the substrate to increase the supply of E4P may not be an appropriate strategy for enhancing β-Arbutin production in K. phaffii. How to increase the supply of E4P without disrupting the normal reductive power supply of K. phaffii will be a question worth considering.
Scale-up production of β-Arbutin
We conducted scale-up fermentation of UA3-10 in a 5 L fermenter and monitored the biomass and β-Arbutin titer throughout the fermentation process (Fig. 8). To evaluate the performance of the optimal strain UA3-10, high-density fermentation was carried out in a 5 L fermenter, employing a fed-batch strategy. The two-stage fermentation process precisely controlled the carbon source for both cell growth and product accumulation. Upon culturing the strain for 18.6 h, glycerol in the BSM medium was depleted, initiating the glycerol fed-batch fermentation phase with a controlled flow rate at 5–10 g/h. After 32 h of cultivation, the biomass of the strain reached 278, prompting a 1-hour pause in feeding. Following this starvation period, the methanol induction phase commenced. Initially, methanol was supplied at a flow rate of 5 g/h, with hourly interruptions in feeding to assess the methanol utilization ability of the strain. Monitoring DO changes within a minute helped prevent excessive methanol accumulation. As the strain adapted to the methanol flow rate, it was gradually increased, reaching a maximum flow rate of 20.7 g/h in this fermentation cycle. With the initiation of methanol feeding, there was a notable surge in β-Arbutin accumulation, while the increase in biomass slowed down. Fermentation was terminated after 132 h of induction, with the maximum biomass reaching 384 at 164 h. The peak production of β-Arbutin reached 128.6 g/L at 140 h, which was 1.89 times that of strain UA1-1 (68.02 g/L). This achievement represents the highest titer reported using K. phaffii, surpassing other strains documented in the literature.