Promoters selection and strains construction
The strain IFO0880 natively accumulates lipids at higher titers [8] and was therefore utilized for the engineering of PuA production. First, three types of plasmids based on the pZPK-PGPD1-HYG-TNOS backbone were constructed, with PgFADX expression controlled by three different promoters. The choice of promoter is directly related to the expression level of the target protein [34,35]. An appropriately selected promoter can result in higher yields of the desired product without adversely affecting cell growth. So far, several promoters, including constitutive and inducible types, have been characterized in R. toruloides. One of the promoters selected was a constitutive PPGI1 promoter of the glucose 6-phosphate isomerase. This promoter was previously shown to be four times stronger than the promoter of the glyceraldehyde 3-phosphate dehydrogenase, PGPD1, when driving the expression of HYG in cells supplemented with an increased concentration of hygromycin [36]. Since, in the effort to produce conjugated linolenic acid isomers (CLNA) in recombinant fission yeast, it was shown that a high accumulation of PuA and calendic acid influences cell growth [21,37], the second promoter selected was an inducible PNAR1 of the nitrate reductase regulated by the nitrogen source [38]. The third promoter used was PPMA1 of the plasma membrane proton-transporting ATPase, which is often employed as a constitutive promoter in the model yeast S. cerevisiae. The expression cassettes containing codon optimized PgFADX under the control of the three different selected promoters were randomly integrated into the R. toruloides IFO0880 genome using the ATMT. After selection of the stable transformants, the presence of an insertion cassette in the genomic DNA was confirmed by PCR analysis.
Screening of the transformants expressing PgFADX
Initially, we analyzed whether R. toruloides accumulates lipid storage organelles, lipid droplets, in a nitrogen-limited MedA+ medium, which is used to stimulate the formation of lipids in oleaginous yeast Y. lipolytica [39]. Lipid droplets were observed in living cells by staining with a lipid droplet-specific dye LD540. As shown in Fig. 1, the R. toruloides IFO0880 cultured in a MedA+ medium showed enlarged lipid droplets compared to cells grown in a rich YPD medium. The presence of lipid droplets of bigger sizes was detected in cells grown in MedA+ containing crude glycerol as a carbon source. Taken together, our results suggest that a MedA+ medium supplemented with glucose or crude glycerol could be a suitable medium for the heterologous production of a single-cell oil containing PuA using metabolic engineering techniques.
Secondly, twelve randomly selected PgFADX-containing transformants for each promoter were screened for their ability to produce PuA. This was because transformants obtained by the ATMT usually have the expression cassette randomly integrated into the genome, which can influence cell growth and the expression of the desired gene from the cassette [40]. As shown in Fig. 2, randomly picked PgFADX-containing transformants accumulated comparable levels of TFA per cell growth as wild type IFO0880. The most important result was the successful production of PuA in all tested engineered recombinant strains. Generally, a relatively higher content of PuA in TFA was observed for transformants expressing PgFADX under the control of the PPMA1 promoter. The use of the PPMA1 promoter enhanced the production of PuA and the best PuA-producing strain, PMA5, accumulated 8.6-fold and 11.1-fold more µg/OD PuA compared to the PuA best-producing strains of the other two promoters, PGI28, and NAR13, respectively. This result confirmed the importance of promoter choice and suitable screening of engineered strains when ATMT transformation is used to obtain transformants with random integration of the desired cassette.
Accumulation and distribution of PuA in production media containing glucose
It was previously shown that the production of CLNA is a dynamic process [21,23,37]. Therefore, PuA-producing transformants were analyzed for the dynamics of PuA production in a time dependent manner. Two transformants for each promoter were selected, namely PGI26 and PGI28, NAR13 and NAR16, and PMA5 and PMA6, which express PgFADX from PPGI, PNAR1, and PPMA1 promoters, respectively. First, the growth, biomass yield, glucose consumption, and fatty acid production in the engineered strains were compared to the wild type strain IFO0880 (Fig. 3 and Table 2). All the selected recombinant strains containing randomly integrated PgFADX-expression cassettes grew comparably well, with a slight increase in OD and biomass compared to the wild type strain. Most of the glucose was consumed after 72 h cultivation. All strains accumulated, on average, 40-50% of TFA in biomass. The fatty acid profile of the wild type strain remained stable during the cultivation periods of 72 h, 120 h, and 168 h (Supplementary Table S3). A substantial difference was observed in the ratio of monounsaturated to polyunsaturated fatty acid (MUFA/PUFA) for the PuA-producing recombinant strains. The expression of PgFADX resulted in an increased relative content of oleic acid (C18:1) and decreased levels of linoleic acid (C18:2) and α-linolenic acid (C18:3). These observed changes correlated with the elevation of PuA levels in the recombinant cells. This result suggests that the PgFADX might compete for the C18:1 substrate with the activity of the endogenous FAD2 in the engineered strains, similar to what observed in recombinant Arabidopsis thaliana plants [19].
The highest relative content of PuA reached 1.3% at 72 h and 120 h in the PMA6 strain containing the PPMA1 promoter, which was much higher than in the strains with PgFADX expressed from the PPGI and PNAR1 promoters (Table 2). When productivity is taken into consideration, in the best PuA-producing strain PMA6, the total PuA reached 105.8 mg/L and 6.1 mg/g DCW at 30°C after 72 h (Table 2). Similar results were obtained with prolonged cultivation times of 120 h and 168 h, suggesting that the PuA level is stable over time in R. toruloides cells. The expression under PPGI and PNAR1 promoters showed a 10-fold and 9-fold lower production of PuA, respectively, compared to the PPMA1 promoter.
Table 2. Fatty acid accumulation in strains cultivated in MedA+ medium containing 6% glucose.
Strain
|
PgFADX promoter
|
Time
|
TFA (g/L)
|
TFA/DCW (%)
|
PuA (% of TFA)
|
PuA (mg/g DCW)
|
PuA (mg/L)
|
IFO0880
|
-
|
72 h
|
4.55 ± 0.17
|
38.81 ± 0.97
|
-
|
-
|
-
|
120 h
|
6.20 ± 0.02
|
41.80 ± 1.86
|
-
|
-
|
-
|
168 h
|
6.61 ± 0.38
|
42.22 ± 2.20
|
-
|
-
|
-
|
PGI26
|
PPGI1
|
72 h
|
7.54 ± 0.08
|
46.09 ± 3.85
|
0.13 ± 0.00
|
0.59 ± 0.03
|
9.74 ± 0.49
|
120 h
|
8.44 ± 0.62
|
45.23 ± 1.67
|
0.12 ± 0.00
|
0.55 ± 0.02
|
10.35 ± 0.71
|
168 h
|
7.26 ± 1.36
|
40.46 ± 9.80
|
0.12 ± 0.00
|
0.49 ± 0.12
|
8.70 ± 1.74
|
PGI28
|
PPGI1
|
72 h
|
6.76 ± 0.67
|
37.97 ± 1.89
|
0.10 ± 0.00
|
0.37 ± 0.00
|
6.65 ± 0.40
|
120 h
|
8.57 ± 0.71
|
45.05 ± 2.36
|
0.10 ± 0.00
|
0.43 ± 0.02
|
8.21 ± 0.66
|
168 h
|
7.43 ± 1.02
|
39.71 ± 5.44
|
0.09 ± 0.00
|
0.37 ± 0.05
|
6.90 ± 0.90
|
NAR13
|
PNAR1
|
72 h
|
9.10 ± 0.55
|
48.50 ± 2.58
|
0.13 ± 0.00
|
0.62 ± 0.01
|
11.62 ± 0.09
|
120 h
|
8.69 ± 0.27
|
46.81 ± 1.95
|
0.13 ± 0.01
|
0.61 ± 0.03
|
11.42 ± 0.61
|
168 h
|
8.75 ± 0.55
|
44.84 ± 2.54
|
0.13 ± 0.00
|
0.59 ± 0.05
|
11.42 ± 1.03
|
NAR16
|
PNAR1
|
72 h
|
9.27 ± 0.22
|
46.53 ± 0.24
|
0.07 ± 0.00
|
0.33 ± 0.00
|
6.59 ± 0.09
|
120 h
|
9.53 ± 0.10
|
49.86 ± 2.45
|
0.07 ± 0.00
|
0.36 ± 0.03
|
6.97 ± 0.29
|
168 h
|
8.29 ± 0.65
|
43.16 ± 2.44
|
0.07 ± 0.00
|
0.31 ± 0.03
|
6.04 ± 0.61
|
PMA5
|
PPMA1
|
72 h
|
7.35 ± 0.51
|
46.16 ± 3.71
|
0.97 ± 0.11
|
4.45 ± 0.13
|
70.91 ± 2.86
|
120 h
|
7.39 ± 0.53
|
42.09 ± 5.66
|
0.98 ± 0.01
|
4.14 ± 0.62
|
72.70 ± 6.34
|
168 h
|
6.46 ± 2.10
|
35.26 ± 6.29
|
0.96 ± 0.01
|
3.37 ± 0.58
|
61.63 ± 19.63
|
PMA6
|
PPMA1
|
72 h
|
8.02 ± 0.58
|
45.98 ± 2.65
|
1.32 ± 0.06
|
6.06 ± 0.07
|
105.77 ± 2.81
|
120 h
|
8.17 ± 0.17
|
44.05 ± 1.41
|
1.27 ± 0.03
|
5.60 ± 0.30
|
103.90 ± 4.48
|
168 h
|
7.63 ± 0.74
|
42.39 ± 5.67
|
1.25 ± 0.02
|
5.28 ± 0.80
|
95.00 ± 10.78
|
Abbreviations: DCW, dry cell weight; PuA, punicic acid; TFA, total fatty acids.
It is worth emphasizing that the PuA yield was 5.4-fold higher than the yield obtained in recombinant S. pombe over-expressing PgFADX from the strong inducible NMT1 promoter [21], and 2.9-fold higher than in the recombinant obese Y. lipolytica strain expressing PgFADX from the hybrid inducible pEYK1 4AB-coreTEF promoter [22]. It is proposed that the main limitation in PuA accumulation is the inefficient flux of PuA from phospholipids to TAG in recombinant yeasts. This hypothesis was proven in the recombinant model yeast S. cerevisiae [18] and more recently in the recombinant oleaginous yeast Y. lipolytica, which can accumulate increased levels of PuA only after significant multi-level genetic optimalization, including an improved supply of C18:2, expressing multiple copies of PgFADX, the acyl-editing, β-oxidation, and glycerol-3-phosphate synthesis pathways [23].
To sum up, our results demonstrated that the recombinant oleaginous red yeast R. toruloides, expressing PgFADX from a PPMA1 promoter, is capable of producing PuA yields comparable to those recently reported for the engineered oleaginous yeast Y. lipolytica, with comprehensive genetic refinement. The relative level of PuA in engineered R. toruloides is not high; therefore, it is expected that further optimization of the metabolic pathways, including but not limited to the carbon and C18:2 supply, enhanced PuA synthesis, and PuA channeling to TAG lipid structures, could lead to a significant increase in PuA titer.
CLNA are preferentially synthesized through biotransformation of linoleic acid esterified to phosphatidylcholine in native producers and then very efficiently channeled from membrane phospholipids to lipid storage depots, lipid droplets, in the form of TAG [19,41]. However, the precise mechanism of CLNA channeling is still not well understood and is currently under investigation. Previous studies have demonstrated a significant difference in the relative content of PuA in TAG and phosphatidylcholine lipid structures between pomegranate seeds, which naturally produce PuA, and the seeds of transgenic plants [19]. Therefore, the relative content of PuA in TAG and phospholipids was examined.
In order to characterize the distribution of PuA in individual lipid classes, the presence of conjugated double bonds in the PuA structure, which allows its detection under UV light, was utilized [42]. First, total lipid extracts of R. toruloides were separated on TLC plates to analyze the phospholipids. With this approach, PuA was mainly detected in the phosphatidylcholine (Fig. 4A), which is reported as the primary place for CLNA synthesis. Second, the total lipid extracts were loaded on a TLC plate and the plate was developed in conditions favoring the separation of neutral lipids. The majority of PuA was detected in TAG (Fig. 4B). The presence of PuA in steryl esters could not be verified due to possible interference from the signal of some sterol molecules containing conjugated double bonds. Furthermore, PuA in a pool of free fatty acids was negligible. This is in contrast with the presence of CLNA in lipid extracts from recombinant yeast strains accumulating substantial amounts of free CLNA [22,37,43].
It is worth noting that, although the relative amount of PuA in engineered R. toruloides strains is not high, the UV signal indicates that PuA is predominantly distributed in TAG lipid structures (Fig. 4A and B). This result suggests that, in engineered R. toruloides strains, the PuA is efficiently channeled from the site of synthesis (phosphatidylcholine) to the TAG lipid structures. However, this might be due to the very high ratio of TAG to phospholipids in engineered strains under tested conditions. To further analyze the distribution of PuA in lipid fractions in the two best PuA-producing strains, PMA5 and PMA6, the relative fatty acid content in phospholipids and TAG was determined (Fig. 4C and 4D). In both strains, the relative content of C18:1 increased in the lipid fractions examined compared to the wild-type strain, while the relative content of C18:3 showed an almost 4-fold decrease. The PuA contents in both strains were below 0.1% and approximately 1% in the phospholipids and TAG fractions, respectively. The high enrichment of PuA in the TAG lipid structures and, at the same time, the high yield of PuA with a minimal requirement for genome modification suggests that the red yeast R. toruloides is a more suitable oleaginous yeast for PuA production than the yeast Y. lipolytica.
Accumulation and distribution of PuA in production media containing crude glycerol
A byproduct of biodiesel production, a crude glycerol can be utilized for microbial lipid production using the yeast R. toruloides [5,44]. Converting a low-cost carbon source into a value-added product, such as PuA, could significantly reduce the upstream expenses, thereby making the overall production process more economically viable. In our initial experiments, the accumulation of enlarged lipid droplets was observed in the R. toruloides wild type IFO0880 strain, grown in a MedA+ medium supplemented with 6% crude glycerol (Fig. 1). Since the response of promoters to different carbon source may impact the final yield of the desired product [34,35] the PuA production efficiency, with the glucose replaced with crude glycerol, was evaluated.
The recombinant strains producing the highest PuA yields for all three promoters were examined for growth, TFA accumulation, PuA content and its distribution. The procedure was similar to that used with the MedA+ medium supplemented with glucose. As shown on Fig. 5A, the growth of PuA-producing R. toruloides strains was comparable to the growth of the wild type IFO0880. Compared to the glucose-containing medium (Fig. 3), the growth in crude glycerol was slightly slower, but after 120 h, the cells reached a similar OD as when grown in glucose. The biomass yield and lipid accumulation increased with cultivation time (Fig. 5B) and strains accumulated, on average, approximately 40-50% of TFA in biomass at the 120 h and 168 h time points (Table 3). This result confirms that crude glycerol is a suitable low-cost carbon source for microbial lipid production using R. toruloides. Similar to the glucose medium, the relative content of fatty acids in the wild type strain remained stable during the cultivation periods of 72 h, 120 h, and 168 h (Supplementary Table S4). However, the relative content of PUFA decreased in the MedA+ supplemented with glycerol compared to the glucose medium.
In general, the amount of PuA increased with the cultivation time (Table 3). In comparison to the glucose medium, the PuA yield significantly increased in strains with PgFADX expressed from the PNAR1 promoter when cultivated on a low-cost crude glycerol medium. The highest relative level of PuA was 0.9% of TFA at 72 h in the PMA5 strain containing the PPMA1 promoter, which was again much higher than in the strains with PgFADX expressed from the PPGI and PNAR1 promoters. When productivity is taken into consideration, the best PuA-producing strain, PMA5, achieved a total PuA yield of 72.8 mg/L and 3.6 mg/g DCW at 30°C after 168 h in shake flask conditions. The level of PuA obtained is approximately two-fold higher than that has been reported for the engineered S. pombe [21], and Y. lipolytica [22] strains. However, it is about one-third lower than the yields reported in a recent study on recombinant Y. lipolytica, which achieved higher yields through significant optimization of the metabolic pathways grown with glucose as a carbon source in shake flask conditions [23]. Nevertheless, it should be noted that, considering the upstream costs for the carbon source, the yield obtained from low-cost crude glycerol is more economically viable, despite the lower PuA levels achieved.
Table 3. Fatty acid accumulation in strains cultivated in MedA+ medium containing 6% crude glycerol.
Strain
|
PgFADX promoter
|
Time
|
TFA (g/L)
|
TFA/DCW (%)
|
PuA (% of TFA)
|
PuA (mg/g DCW)
|
PuA (mg/L)
|
IFO0880
|
-
|
72 h
|
5.26 ± 0.15
|
39.40 ± 0.12
|
-
|
-
|
-
|
120 h
|
9.19 ± 0.21
|
48.73 ± 0.79
|
-
|
-
|
-
|
168 h
|
11.71 ± 1.12
|
52.57 ± 4.33
|
-
|
-
|
-
|
PGI26
|
PPGI1
|
72 h
|
4.92 ± 0.08
|
39.11 ± 0.35
|
0.04 ± 0.00
|
0.17 ± 0.00
|
2.10 ± 0.01
|
120 h
|
8.43 ± 0.97
|
46.38 ± 0.50
|
0.03 ± 0.00
|
0.15 ± 0.00
|
2.76 ± 0.29
|
168 h
|
8.84 ± 1.32
|
49.67 ± 2.83
|
0.03 ± 0.00
|
0.16 ± 0.01
|
2.93 ± 0.38
|
PGI28
|
PPGI1
|
72 h
|
4.85 ± 0.35
|
38.53 ± 0.06
|
0.04 ± 0.01
|
0.14 ± 0.01
|
1.73 ± 0.01
|
120 h
|
8.31 ± 0.74
|
43.18 ± 1.89
|
0.03 ± 0.01
|
0.11 ± 0.01
|
2.14 ± 0.09
|
168 h
|
9.82 ± 2.71
|
49.70 ± 4.23
|
0.03 ± 0.01
|
0.12 ± 0.01
|
2.44 ± 0.57
|
NAR13
|
PNAR1
|
72 h
|
3.00 ± 0.03
|
31.65 ± 1.50
|
0.14 ± 0.01
|
0.43 ± 0.02
|
4.05 ± 0.03
|
120 h
|
6.98 ± 1.12
|
41.43 ± 2.18
|
0.14 ± 0.01
|
0.58 ± 0.08
|
9.70 ± 0.67
|
168 h
|
8.20 ± 1.99
|
50.20 ± 4.98
|
0.26 ± 0.04
|
1.26 ± 0.03
|
20.44 ± 2.48
|
NAR16
|
PNAR1
|
72 h
|
3.16 ± 0.03
|
33.71 ± 0.90
|
0.09 ± 0.01
|
0.29 ± 0.04
|
2.68 ± 0.25
|
120 h
|
7.88 ± 1.68
|
42.08 ± 0.67
|
0.13 ± 0.01
|
0.53 ± 0.01
|
9.91 ± 1.87
|
168 h
|
8.79 ± 1.60
|
48.90 ± 1.10
|
0.29 ± 0.02
|
1.40 ± 0.07
|
24.98 ± 2.69
|
PMA5
|
PPMA1
|
72 h
|
5.22 ± 0.25
|
37.33 ± 1.14
|
0.89 ± 0.08
|
3.33 ± 0.21
|
46.54 ± 2.13
|
120 h
|
9.06 ± 0.97
|
48.95 ± 1.56
|
0.69 ± 0.06
|
3.35 ± 0.19
|
61.74 ± 1.18
|
168 h
|
10.86 ± 1.96
|
52.83 ± 1.64
|
0.68 ± 0.06
|
3.56 ± 0.19
|
72.81 ± 7.12
|
PMA6
|
PPMA1
|
72 h
|
5.17 ± 0.11
|
34.88 ± 0.12
|
0.75 ± 0.00
|
2.63 ± 0.01
|
38.94 ± 0.86
|
120 h
|
8.46 ± 0.13
|
41.64 ± 2.68
|
0.60 ± 0.00
|
2.48 ± 0.16
|
50.43 ± 0.75
|
168 h
|
10.27 ± 0.89
|
47.96 ± 2.99
|
0.61 ± 0.01
|
2.92 ± 0.10
|
62.44 ± 3.56
|
Abbreviations: DCW, dry cell weight; PuA, punicic acid; TFA, total fatty acids.
To examine the distribution of PuA in individual lipid classes, the total lipids were separated on TLC plates and the detection of PuA under UV light was used, as described above. Similarly, as with the glucose-containing medium, the signal for PuA was mainly detected in phosphatidylcholine (Fig. 6A) and TAG lipid structures (Fig. 6B). The signal for free PuA was negligible. Next, the relative contents of PuA in phospholipid and TAG fractions in the two best PuA-producing strains, PMA5 and PMA6, was analyzed (Fig. 6C and 6D). The PuA content in both strains was approximately 0.2% and 0.6% in the phospholipids and TAG fractions, respectively. The enrichment of PuA in the single-cell oil, which are comprised of TAG lipid structures, was not as high as in the glucose medium. However, since the TAG are predominant lipid structures in R. toruloides, grown under nitrogen-limited conditions, with glycerol as the carbon source, the majority of PuA is stored in TAG lipid structures. Our results confirmed that the medium with a low-cost carbon source is suitable for the high production of PuA by the recombinant red yeast R. toruloides.
The price of media components, mainly carbon and nitrogen sources, is a critical factor in any biotechnological process [44,45]. R. toruloides exhibits greater versatility than Y. lipolytica in assimilating low-cost carbon substrates, including crude glycerol, lignocellulosic hydrolysates, and molasses [2]. This versatility significantly enhances its potential use in various biotechnological applications. Moreover, R. toruloides can accumulate more lipids from crude glycerol than from pure glycerol, without being negatively affected by the impurities present in crude glycerol [46]. An additional advantage of R. toruloides is its natural production of carotenoids, which act as antioxidants and are widely used in food, pharmaceuticals, and cosmetics [3,4]. It has been demonstrated that carotenoids can effectively inhibit lipid peroxidation [47]. Therefore, carotenoids, which naturally co-purify with lipids during the lipid extraction from engineered R. toruloides cells (data not shown), could prevent PuA oxidation of PuA-enriched single cell oil.
Another notable difference between the two oleaginous yeasts is that R. toruloides naturally synthesizes C18:3, unlike Y. lipolytica. It is well documented that the level of unsaturation in PUFA correlates with increased membrane fluidity, elasticity, and flexibility [48]. Therefore, C18:3 has a greater impact on membrane properties compared to C18:2. It can be hypothesized that, due to this difference, R. toruloides efficiently channels C18:3 from phospholipids to TAG lipid structures to maintain membrane properties. The presence of conjugated double bonds in PuA further impacts the cellular membrane properties. Considering these factors, along with the high accumulation of PuA in engineered R. toruloides cells, it is evident that R. toruloides may have even greater biotechnological potential for PuA production than Y. lipolytica.