LDs are the major location for β-carotene storage in S. cerevisiae
To expand the storage space for improving β-carotene accumulation, it is vital to identify the major storage location of β-carotene in S. cerevisiae. At present, β-carotene is believed to be stored in cell membrane or (and) LDs, but the distribution ratio is not clear [6, 23]. Herein, we quantitatively determined this value. Firstly, we visualized the number of LDs in the wild-type strain YBX-B and β-carotene synthesizing strain YBX-01 (producing 5.50 mg/g β-carotene) by transmission electron microscope (Additional file 1: Fig. S1) and confocal microscope (Fig. 1A). The results all show that more LDs are produced in strain YBX-01, indicating that the budding yeast would produce extra LDs to respond to carotenoid accumulation. Then we directly separated LDs and cell membrane components for β-carotene quantitative determination, respectively. According to the previously reported protocol [9], the cells of strain YBX-01 cultured for 72 h were disrupted, and various components were collected based on the density difference. After ultracentrifugation of the post-nuclear supernatant (PNS) at 40,000 rpm for 1 h, LDs were enriched in the upper layer of the SW40 tube, which clearly indicated that a large amount of orange β-carotene exist in LDs (Fig. 1B). The cytoplasmic components were located in the middle layer, while the total membrane components were at the bottom of the centrifuge tube. The following quantitative analysis shows that LDs contained 85.5% of β-carotene (4.70 mg/g), while the cell membrane only contained 9.4% (0.52 mg/g).
LDs mainly consist of triacylglycerol (TAG) and sterol esters (SE), in which TAG is located in the core of LDs and is the major lipid species [8]. The content of TAG in strain YBX-01 was found to reach 32.2 mg/g DCW, which is 5.37-fold greater than the wild-type strain YBX-B (Additional file 1: Fig. 2S). Correspondingly, three key genes associated with TAG synthesis, including ACC1 (encoding acetyl-CoA carboxylase), PAH1 (encoding phosphatidate phosphatase) and DGA1 (encoding diacylglycerol acyltransferase) were transcriptionally induced, especially after 24 h of the fermentation (Additional file 1: Fig. 2S). Next, we successively knocked out four key genes controlling LDs synthesis in strain YBX-01, to investigate the variation of LDs synthesis on β-carotene accumulation. In S. cerevisiae, DGA1 and LRO1 (encoding acyltransferase) are responsible for the synthesis of TAG, while ARE1 and ARE2 encoding acyl-CoA: sterol acyltransferase contribute to SE synthesis [14]. The deletions generated four strains (YBX-ld1, ld2, ld3 and ld4), respectively. The confocal microscope shows that the number of LDs in engineered strains gradually decreased, and no LDs were observed in the quadruple disruption strain YBX-ld4 (Fig. 2A). As expected, the content of β-carotene correspondingly decreased. There is only 14.8% of β-carotene left in strain YBX-ld4 (Fig. 2B). Noticeably, the strain YBX-ld1 with DGA1 disruption showed a 61.3% decrease compared to strain YBX-01, confirming that DGA1 is the major gene controlling TAG synthesis [27]. These results verified the vital role of LDs in storing β-carotene, and proved that LDs are the major storage space of β-carotene in recombinant S. cerevisiae, thus LDs were the engineer targets for increasing β-carotene accumulation.
The limitation of modulating LDs key genes for β-carotene overproduction
Genetic modulating the genes of LDs formation can promote lycopene accumulation [23]. According to this work, we engineered ACC1 in strain YBX-01 by introducing two site mutations (Ser659Ala and Ser1157Ala) through fusion PCR. ACC1 encodes acetyl-CoA carboxylase to form malonyl-CoA, the first committed and critical step in fatty acid metabolism. The obtained ACC1S659A/S1157A was linked with the strong constitutive promoter PTEF1. Then, the original promoters of PAH1 and DGA1 were substituted with inducible PGAL1 promoter, and the strain YBX-22 with overexpression of ACC1, PAH1 and DGA1 was constructed. These modifications led to increased intracellular TAG content from 32.2 to 43.5 mg/g DCW (Fig. 3A), and increased β-carotene content from 5.5 to 6.75 mg/g DCW, which is a 22.7% increase compared to the control strain (Fig. 3B). Through separating LDs, cytoplasm and membrane components, we found that the proportion of β-carotene in LDs increased from 85.5–87.6%, while the content in the cytoplasm and membrane system did not change significantly, indicating that almost all of the increased β-carotene was distributed in LDs.
LDs can be cleaved by TAG lipase and SE hydrolase when cells need energy in times of scarcity. To further improve LDs formation, three genes TGL3, TGL4 and TGL5, encoding TAG lipase and SE hydrolase [28] were disrupted in strain YBX-22 resulting in strain YBX-23. The content of TAG in strain YBX-23 consequently increased from 43.5 mg/g DCW to 47.5 mg/g DCW (Fig. 3A), and the β-carotene content increased from 6.75 mg/g DCW to 6.98 mg/g DCW (Fig. 3B). A previous study indicated that the disruption of FLD1 improved lipid levels, LD clustering, and favored the formation of larger LDs [12]. FLD1 was deleted in strain YBX-23, resulting in strain YBX-24, in which TAG content was further increased to 48.3 mg/g DCW (Fig. 3A). Accordingly, β-carotene content increased to 7.37 mg/g DCW (Fig. 3B). After separation of LDs, we found that the percentage of β-carotene in LDs increased to 88.2%.
Collectively, the above results indicated that engineering the genes associated with LDs synthesis, size and degradation can lead to 50% increase of TAG content, which resulted in 34% increase in β-carotene content compared to the original strain YBX-01. This value was slightly higher than the data reported by Ma et al that showed a 25% increase of lycopene through engineering fatty acid synthesis and TAG metabolism. Interestingly, when the same strategy was applied to the wild-type strain YBX-B, 180.6 mg/g DCW of TAG was achieved which is 2.74 fold greater than that of strain YBX-24. The discrepancy is possibly due to the competition for precursor (acetyl CoA) between β-carotene and TAG biosynthesis in strain YBX-24. Obviously, the unilateral improvement of LDs formation is not conducive to β-carotene synthesis, vice versa. Another explanation might be that the promotion of genetic modification on LDs synthesis is limited, considering that LDs is a specific organelle for decreasing the cytotoxicity and will largely be synthesized only when excessive lipophilic products accumulated in cells [14]. Thus, it is essential to design a more effective strategy to address this dilemma.
The dual regulation of oleic acid for promoting LDs and metabolic pathways
In this work, we focused on OA and aimed to simultaneously promote LDs formation and drive the metabolic flux of MVA to β-carotene pathway, considering that OA has multiple beneficial physiological functions. Firstly, supplementation of OA can dramatically improve the synthesis of LDs in S. cerevisiae [14], which additionally save more acetyl-CoA for β-carotene synthesis because cells don't need de novo synthesis of UFAs. Secondly, the incorporation of UFAs into cell membrane increases the flexibility of cell membrane and relieves carotenoid-induced membrane stress [15, 19]. Thirdly, OA can bind the fatty acid regulatory (FAR) region in the upstream promoter of a specific gene to regulate its expression [7]. Thus, it is possible to mine OA-repressible promoters and substitute ERG9 original promoter to dynamically down-regulate ERG9 expression, and resulted in improved β-carotene synthesis by driving the metabolic flux towards the β-carotene biosynthetic pathway. Due to the fact that excessive OA cause lipotoxicity to cells, the optimal concentration of OA on LDs formation should be identified firstly. Thus, 0.5, 1, 2, 4 and 8 mM OA was supplemented into the medium after 12 h of fermentation when strain YBX-01 started the synthesis of β-carotene, respectively. Biomass, TAG and β-carotene content were determined. The results show that the cell growth (Additional file 1: Fig. 3S) and TAG metabolism (Fig. 4A) were improved by OA addition, especially when 2 mM (64.4 mg/g DCW) of OA was used, with 78.9% increase of TAG content achieved compared to the control. This value was 33.3% greater than the engineered strain YBX-24, confirming the high-efficient promotion of exogenous oleic acid on LDs formation. As expected, β-carotene content increased to 7.5 mg/g DCW, which is 36.4% greater than that of the control (Fig. 4B). Additionally, the membrane fluidity of cells was improved by 2.1 fold relative to the control strain through determining the value of fluorescence anisotropy, something that is consistent with our previous work [19]. Noticeably, the addition of 8 mM OA caused a large decrease in β-carotene content (4.2 mg/g DCW), with only 76.4% of the control trial, indicating that lipotoxicity occurred at this concentration. Thus, 2 mM OA was chosen in the following work.
To mine OA-repressible promoters, the transcriptome analysis was adopted to compare mRNA levels in YBX-01 after treated with 2 mM of OA for 2 h and 12 h relative to the control without addition (Additional file 1: Table 4S). Five genes (OLE1, MGA2, IZH1, IZH4 and YDR274C) were further selected as potential candidates since their mRNA levels were greatly decreased during the fermentation based on quantitative real-time PCR (qRT-PCR) analysis (Additional file 1: Fig. 4S). OLE1 encodes Delta (9) fatty acid desaturase and is involved in UFAs synthesis, and MGA2 encoding endoplasmic reticulum membrane protein, participates in the transcriptional regulation of OLE1 gene as a transcription factor [16]. IZH1 and IZH4 encode membrane proteins and associated with zinc ion homeostasis, which are transcriptionally down-regulated when cells are treated with UFAs [19]. YDR274C encodes putative protein of unknown function. We cloned their promoters and replaced the original ERG9 promoter in strain YBX-01, respectively, and generated five strains, including YBX-01-IZH1, YBX-01-IZH4, YBX-01-MGA2, YBX-01-YDR274C and YBX-01-OLE1. The reference strain YBX-01 and the five engineered strains were then examination of cell growth and β-carotene synthesis after 2 mM oleic acid was added.
As shown in Fig. 5A, except for YBX-01-IZH4 and YBX-01-YDR274C, the resulting strains showed a similar growth pattern to that of the reference strain YBX-01, indicating that the modulation of ERG9 expression controlled by IZH1, MGA2 and OLE1 promoters dose not interfere with cell growth. Except for strain YBX-01-OLE1, the other four strains showed increase in β-carotene production (Fig. 5B), especially strain YBX-01-IZH1 that exhibited the highest increase, the content and yield increased to 9.88 mg/g DCW and 102 mg/L, which is 31.7% and 32.6% higher than strain YBX-01. The decrease of β-carotene in YBX-01-OLE1 was ascribed to induced transcription of ERG9, which led to activation of ergosterol synthesis, as later confirmed. The content of β-carotene in YBX-01-IZH4 and YBX-01-YDR274C was comparable with YBX-01-IZH1, but due to the decreased biomass, the production of β-carotene was only 81.6% and 80.2% of YBX-01-IZH1, respectively.
To verify the effect of OA-repressible promoters at molecular level, the expressions of ERG9 in the strains was determined and compared with the parent strain YBX-01, in which, the expressions remained unchanged after 14 h (Fig. 6A). In strain YBX-01-IZH1 and YBX-01-IZH4, ERG9 expression was low until the end of the fermentation. ERG9 mRNA level under the control of PMGA2, although slightly lower, was comparable to that of strain YBX-01. In strain YBX-01-YDR274C, ERG9 was down-regulated at first, but slowly increased, and no significant differences were observed after 60 h compared to strain YBX-01. Unexpectedly, ERG9 expression in YBX-01-OLE1 showed a sharp increase after OA addition, and was 2.7 times higher than that of strain YBX-01 at 36 h, after which it decreased to the comparable level of YBX-01 until the end of the fermentation, implying that ergosterol synthesis was activated, which could explain the rise of ergosterol and the descend of β-carotene content. However, the detailed mechanism needs to be further explored.
It is known that the down-regulation of ERG9 would interfere with the metabolic balance of MVA pathway, and result in the accumulation of intermediate metabolites, such as farnesol, which is a disadvantage for cell growth and product synthesis due to its toxicity [35]. The contents of downstream and intermediate metabolites in different strains was determined, including squalene and ergosterol (Fig. 6B), and farnesol (Fig. 6C). Except for YBX-01-OLE, the contents of squalene and ergosterol in engineered strains were lower than strain YBX-01. Especially in strains YBX-01-IZH1 and YBX-01-IZH4, 39.3% and 48.3%, and 33.6% and 31.8% reduction observed, respectively. This confirmed the efficiency of PIZH1 and PIZH4 on down-regulating ERG9 expression. Concerning the intermediate metabolite farnesol, similar to previous work results, farnesol was not detected in reference strain YBX-01, while 0.29 and 0.26 mg/g DCW of farnesol were produced in strains YBX-01-IZH1 and YBX-01-IZH4, respectively, indicating that this intermediate metabolite cannot be effectively consumed by inducible biosynthetic pathway of β-carotene. No farnesol accumulated in strain YBX-01-OLE1, corresponding to the highest amount of squalene and esgosterol. These results collectively indicated that PIZH1 is a suitable promoter capable to dynamically down-regulate ERG9 expression, which can drive the metabolic flux towards carotenoid pathway to improve β-carotene synthesis.
The introduction of constitutive metabolic pathways to balance the utilization of farnesol
To further improve β-carotene production, it is vital to facilitate further conversion of precursor farnesol to β-carotene. The accumulation of farnesol in strain YBX-01-IZH1 is possibly due to the imbalance in expression time between ERG9 down-regulation and β-carotene biosynthesis. In this work, in order to reduce the toxicity of carotenoid on cell growth, we introduced an inducible GAL promoter to drive β-carotene biosynthesis [35]. This strategy guarantees that β-carotene will start to synthesize after 12 h when glucose is exhausted, and achieve the separation of the cell growth stage from the product accumulation stage. However, using this strategy, before the synthesis of β-carotene, the engineered strain already synthesized farnesol.
To address this issue, we introduced another β-carotene synthesis pathway driven by glucose and medium-strength promoter PYK1 and TPI1, aimed to initiate β-carotene biosynthesis before 12 h during which glucose is available in the medium. Additionally, to reduce the loss of farnesol by other side pathway, we knocked out DPP1 and LPP1 genes in YBX-01-IZH, involved in the hydrolysis of isoprene pyrophosphate [11], resulting in strain YBX-41. The biomass, squalene, farnesol and β-carotene content in YBX-01-IZH1 and YBX-41 were determined after 72 h of cultivation (Table 2). The results show that the biomass of YBX-41was slightly affected compared to YBX-01-IZH1, and squalene content was further decreased to 1.2 mg/DCW from 1.42 mg/DCW, while farnesol was no longer detectable in strain YBX-41. Correspondingly, β-carotene content and yield increased to 11.4 mg/DCW and 142 mg/L, which is 15.6% and 13.6% greater than those of strain YBX-01-IZH1. This verified that the introduction of extra constitutive synthesis pathway facilitates the complete consumption of farnesol, which resulted in additional increase of β-carotene production with the introduction of inducible synthesis pathway. It is reasonable to believe that by using other metabolic engineering strategies, such as supply of more precursors, energy and reducing equivalents, could further increase the production of β-carotene.
Table 2
The biomass, β-carotene and metabolites content in strains YBX-01-IZH1 and YBX-41
Strains | Biomass (OD600) | Squalene (mg/g DCW) | Farnesol (mg/g DCW) | β-Carotene content (mg/g DCW) | β-Carotene yield (mg/L) |
YBX-IZH1 | 31.20±0.13 | 1.41±0.11 | 0.29±0.03 | 9.88±0.10 | 125±2.96 |
YBX-41 | 30.30±0.27 | 1.20±0.08 | N.D. | 11.42±0.21 | 142±1.73 |
N.D. represents not detected. |