Medium-chain triglyceride production in Nannochloropsis via a fatty acid chain length discriminating mechanism

Abstract

Depending on their fatty acid (FA) chain length, triacylglycerols (TAGs) have distinct applications; thus, a feedstock with a genetically designed chain length is desirable to maximize process efficiency and product versatility. Here, ex vivo, in vitro, and in vivo profiling of the large set of type-2 diacylglycerol acyltransferases (NoDGAT2s) in the industrial oleaginous microalga Nannochloropsis oceanica revealed two endoplasmic reticulum-localized enzymes that can assemble medium-chain FAs (MCFAs) with 8–12 carbons into TAGs. Specifically, NoDGAT2D serves as a generalist that assembles C8-C18 FAs into TAG, whereas NoDGAT2H is a specialist that incorporates only MCFAs into TAG. Based on such specialization, stacking of NoDGAT2D with MCFA- or diacylglycerol-supplying enzymes or regulators, including rationally engineering Cuphea palustris acyl carrier protein thioesterase, Cocos nucifera lysophosphatidic acid acyltransferase, and Arabidopsis thaliana WRINKLED1, elevated the medium-chain triacylglycerol (MCT) share in total TAG 66-fold and MCT productivity 64.8-fold at the peak phase of oil production. Such functional specialization of NoDGAT2s in the chain length of substrates and products reveals a dimension of control in the cellular TAG profile, which can be exploited for producing designer oils in microalgae.

In Nannochloropsis oceanica, two chain-length discriminating diacylglycerol acyltransferases assemble medium-chain fatty acids into triacylglycerols.

Introduction

Triacylglycerol (TAG) is a main form of energy storage in cells. Each TAG molecule consists of three fatty acids (FAs) moieties that are anchored to a glycerol scaffold. Depending on the chain-length (CL) of the FAs, each TAG molecule can be classified as long-chain TAG (LCT) or a medium-chain TAG (MCT), which are distinct in application area, economical value, and market potential. For example, MCTs, a group of TAGs that possess FAs with an aliphatic tail of 8–12 carbons, are widely used in nutraceutical, personal care, and oleochemical industries, and have attracted particular attention for their health benefits (Bach and Babayan, 1982). During digestion, MCTs are converted to medium-chain FAs (MCFAs), which are transported directly in the portal venous system, as opposed to being transported as chylomicrons in the lymphatic system like LCTs (Bloom et al., 1951). The bypass of peripheral tissues, such as adipose tissue, ensures the lower susceptibility of MCT to hormone-sensitive lipase and to deposition as adipose tissues (Bach and Babayan, 1982). As a result, consumption of MCT increases energy expenditure, fat oxidation, and satiety and lowers energy, and food intake in both lean and obese individuals (Scalfi et al., 1991; Binnert et al., 1998; Krotkiewski, 2001; St-Onge et al., 2003). However, at present, pure MCT oil is manufactured by hydrolysis, filtering, and re-esterification of oils from palm and coconut crops, whose plantations however are limited to only tropical and subtropical regions (Arkcoll, 1988; Basiron, 2007; Kumar, 2011; Kinsella et al., 2017). Moreover, typically only ∼3% of the whole plant mass on a per dry weight basis is stored in the form of oil (Chisti, 2007). The low efficiency yet limited supply of land plants-based MCT has created opportunities for more efficient and environmentally friendly MCT supply.

The high photosynthetic growth potential and rich oil content of many oleaginous microalgae has led to growing interest in utilizing their biomass as feedstock for scalable production of TAGs from carbon dioxide (Hu et al., 2008; Wijffels and Barbosa, 2010; Georgianna and Mayfield, 2012). However, for microalgae, the content of MCT in TAG has been generally very low (0.01%–0.05%; Wang et al., 2021). Although overexpression of land plants thioesterases (TEs) in Dunaliella salina elevates the content of TAG-associated MCFA (up to four-fold increase in C12:0; Lin and Lee, 2017; Lin et al., 2018), whether the MCTs are actually boosted is not clear. In microalgae, like land plants, MCTs are produced by a series of biochemical reactions (Nakamura and Li-Beisson, 2016). Specifically, acyl carrier proteins (ACPs) in MCFA-ACPs are hydrolyzed by TE to form MCFAs. MCFAs are transformed into medium-chain lysophosphatidic acids (MCLAs) and then medium-chain phosphatidic acids (MCPAs) by lysophosphatidic acid acyltransferase (LPAAT). Then, MCPAs are added into medium-chain diacylglycerols (MCDs) by phosphatidic acid phosphatase (PAP). Finally, in the last yet the only committed step of TAG synthesis, MCD and an acyl-coenzyme A (CoA) produced from the de novo FA synthesis are converted into MCT (and other classes of TAG) by diacylglycerol acyltransferases (DGATs; Kennedy, 1957). Therefore, rational modulation of cellular TAG profile toward MCTs is dependent on the ability to (1) supply more MCFA substrates and (2) enhance the activity of DGATs for the MCFAs.

In Nannochloropsis oceanica, an industrial oleaginous microalga, our lipidomic analyses revealed 18 MCT species that make up 10.3% of the TAG profile (Li et al., 2014), indicating the presence of a native MCT-assembling mechanism in vivo (Li et al., 2014). Moreover, we showed that the MCFAs, such as C8:0 and C10:0 FAs, in N. oceanica can be elevated by introducing a Cuphea palustris acyl-ACP TE (CpTE; AAC49179.1; Wang et al., 2021). Notably, despite its relatively small size (30 Mb), the N. oceanica genome encodes one of the largest collection of type-2 DGAT-coding genes (11 NoDGAT2s) ever known in an organism (Supplemental Table S1) (Vieler et al., 2012; Zienkiewicz et al., 2017). Among the NoDGATs, we discovered substrate-specific labor division in LCT assembly based on “the degree of unsaturation of FA” in vivo, with NoDGAT2A, 2C, 2D preferring saturated FAs (SFAs), long-chain polyunsaturated FAs (PUFAs), and monounsaturated FAs (MUFAs), respectively (Xin et al., 2017, 2019), while NoDGAT2K and 2J are in favor of the long-chain FAs (LCFAs) of linoleic acid (LA or C18:2) and eicosadienoic acid (EPA or C20:5), respectively (Xin et al., 2019). However, whether and how the CL of TAG species is regulated in cell is not clear.

Here, ex vivo, in vitro, and in vivo profiling of all 11 candidate DGAT2s in N. oceanica reveal two endoplasmic reticulum (ER)-localized ones that exhibit the highest activity incorporating MCFAs into TAGs. In fact, NoDGAT2D serves as a generalist in FA CL as it assembles a wide range of FAs from C8 to C18 into TAGs, whereas NoDGAT2H is a specialist that incorporates only the MCFAs of C8 and C12 into TAGs. Moreover, genetic stacking of NoDGAT2D with a series of higher-plant enzymes or transcriptional factors that supply MCFAs and diacylglycerol (DAG) in N. oceanica elevates MCT content by 66-fold, at the peak of oil-production phase. These findings reveal an extraordinarily sophisticated TAG-synthetic mechanism involving at least seven DGATs acting in a hierarchical and functionally complementary manner in Nannochloropsis spp., and suggest a rational strategy to genetically enhance the production of designer TAGs with certain CL.

Results

NoDGAT2 expression in yeast reveals the MCFA-CoA preference of NoDGAT2D and 2H

To test the hypothesis that N. oceanica harbors MCFA-preferred DGATs, three MCFAs of octanoic acid (C8:0), decanoic acid (C10:0), and lauric acid (C12:0) are collectively fed to 11 yeast lines each carrying one of the 11 NoDGAT2s (Supplemental Figure S1, A and B; see “Materials and methods”). Thin-layer chromatography (TLC) of extracted total lipids (TLs) reveals the synthesis of TAG in the MCFA-fed NoDGAT2A-, 2C-, 2D-, and 2H lines, but not the others (Figure 1A). Gas chromatography–mass spectrometry (GC–MS) shows that, upon feeding the MCFAs, the NoDGAT2A-, 2C-, 2D-, and 2H-carrying yeast lines accumulate considerable level of TAG, respectively (9%–31% of TL; Figure 1B). NoDGAT2A, 2C, and 2D already show TAG-synthetic activity in yeast without any endogenous (the yeast itself contains no MCFAs; Sandager et al., 2002) or externally supplemented MCFAs (Xin et al., 2017); in contrast, for NoDGAT2H (GenBank KX867963, abbreviated as 2H), TAG synthesis is specifically associated with MCFA supplementation (Figure 1, A and B; Xin et al., 2017).

Figure 1.

Revealing MCT-synthetic NoDGAT2s via complementation of TAG-deficient phenotype in mutant yeast H1246 by NoDGAT2s expression. A, TLs separated by TLC. FFA, free FA. B, GC–MS quantification of TAG levels extracted from transformed yeast. The total amount of TAG is normalized based on that of the TLs. C, TAG-associated MCFA profiles in the NoDGAT2-carrying yeasts after the MCFA feeding. All experiments were conducted in one experiment performed with independent triplicate samples, with one representative TLC experiment for each of the triplicates shown in Figure 1A. Here MCFAs include octanoic acid (C8:0), decanoic acid (C10:0), and lauric acid (C12:0). Data are represented as mean ± sd (n = 3 biologically independent samples). ^*Significant change (P ≤ 0.01; one-sided Student’s t test) versus the WT.

Next, we asked whether and to what degree the MCFAs are assembled into TAG by the NoDGAT2s, when MCFA substrates are available. TAG in the NoDGAT2D- and 2H-carrying yeasts contain the highest contents of MCFAs among all the 11 NoDGAT2s tested: ∼10% in 2D and 2H, versus 3% in the Saccharomyces cerevisiae DGAT2 gene of ScDGA1 (i.e. the reference; Figure 1C), suggesting their extraordinary MCT-assembling activities. Notably, besides MCFAs, NoDGAT2D (abbreviated as 2D) can also assemble LCFA (e.g. C16:1 and C18:1; Xin et al., 2017) into TAGs; on the contrary, 2H can only assemble the MCFAs (Figure 1C). Therefore, although both are capable of assembling MCFAs into TAGs, in terms of CL of FA substrates, 2D appears to be a generalist, while 2H is a specialist that is dedicated to producing only MCT.

In vitro enzymatic assays reveal MCFA-CoA specificity of NoDGAT2D and 2H

To probe their MCFA preference in vitro, a nonradiolabeled in vitro DGAT assay is employed to measure the activity and substrate specificity of 2D and 2H toward a variety of acyl-CoAs (see “Materials and methods”). The microsomal fraction of yeast H1246, which contains 2D or 2H, is used for the in vitro assay. Eight acyl-CoAs including C8:0 CoA, C10:0 CoA, C12:0 CoA, C14:0 CoA, C16:0 CoA, C18:0 CoA, C20:5 CoA, and C22:6 CoA are introduced to test the substrate preference of 2D and 2H, with prokaryotic C16:0/C18:1 DAG (the most abundant DAG in N. oceanica IMET1; Li et al., 2014) as the acyl acceptor.

Both 2D and 2H exhibit a generally stronger TAG-synthetic activity toward MCFA CoAs (i.e. C8:0 CoA, C10:0 CoA, and C12:0 CoA; Figure 2, A and B) than the longer chain FAs. Such preference is also supported by the TAG profile derived from TLC–GC–MS analysis (Figure 2C). Among the three MCFA CoA substrates, 2D exhibits the highest activity to C10:0 CoA (73.4 ± 15.7 mmol TAG·mg⁻¹·h⁻¹), while 2H appears to prefer C12:0 CoA over other MCFA CoAs (17.3 ± 1.2 mmol TAG·mg⁻¹·h⁻¹). Meanwhile, among the eight acyl CoAs, 2D exhibits the highest activity toward C18:0 CoA (up to 88.8-mmol TAG·mg⁻¹·h⁻¹), yet 2H appears to prefer C14:0 CoA (up to 27.9-mmol TAG·mg⁻¹·h⁻¹). Specifically, when C20:5 CoA or C22:6 CoA is used as donor, 2H exhibits essentially no TAG-synthetic activity (2D does show some activity), which is in sharp contrast to NoDGAT2J and 2K. However, 2D is more active to C16:0 CoA, C18:0 CoA and all MCFA CoAs than 2H, while 2H is more active to C14:0 CoA than 2D (Figure 2C). These in vitro results are consistent with the ex vivo data (Figure 1C), with both supporting the preference of 2D and 2H to MCFA CoAs over other acyl CoAs. In addition, 2D exhibits considerable activity to almost all acyl-CoAs (except EPA and diacylglycerol acyltransferase; Figure 2, A and C), but 2H only exhibits higher activity to the MCFAs of C8:0, C12:0, and C14:0 than to other FAs (Figure 2, B and C). These in vitro results further support 2D as a generalist DGAT, while 2H as a specialist one, in terms of carbon CL of their substrates and products.

Figure 2.

Profiling the substrate specificities of NoDGAT2D and 2H for acyl-CoAs in vivo. A and B, TLC analysis of lipids from in vitro enzymatic reactions of NoDGAT2D (A) and 2H (B) with various acyl-CoAs. C18:1/C16:0 DAG was used as the acyl acceptor. Experiments were conducted in triplicate, each with one representative TLC shown. Ori, original spot. C, Quantification of lipids from in vitro enzymatic reactions of NoDGAT2D and 2H by GC–MS. Ctr1, microsome; Ctr2, microsome plus 250-mM DAG. Values are presented as mean ± sd (n = 3). Different letters above the bars grouped by each acyl-CoA indicate significant difference (P ≤ 0.01), based on one-way analysis of variance and Tukey’s honest significant difference test.

Genetic manipulation of NoDGAT2D and 2H reveals their MCFA-CoA preference in vivo

To probe their MCT-synthetic roles in vivo, 2D and 2H are each knocked out via Cas9-mediating editing in N. oceanica (2Dko-1, 2Dko-2, 2Hko-1, and 2Hko-2; Supplemental Figures S1, C and D and S2, A and B), and then genetically complemented (2Dc-1, 2Dc-2, 2Hc-1, and 2Hc-2; Supplemental Figure S1, F–G; see “Materials and methods”). These transgenic lines are measured under both nitrogen-replete condition (N⁺) and nitrogen starvation for 72 h (N⁻).

Phenotypes of NoDGAT2D-transgenic lines

Under N⁺ and N⁻, significant difference in TAG content and FA composition is observed in the 2D-knockout (KO) lines, yet not in the complemented lines (as expected). Specifically, versus wild-type (WT) and under N⁺, TAG content is ∼35% lower for 2D-KO lines (Figure 3A). On the other hand, under N⁻, TAG content is 44%–69% lower for 2D-KO lines (Figure 3B). Dry weight and TL content of all 2D-modified lines are measured. Compared to WT, under N⁺, differences are observed only in the dry weight of 2Dko-1 (15% higher; Supplemental Figure S3A), and the TL content of 2Dko-2 (27% higher; Supplemental Figure S3C); however, under N⁻, no changes of dry weight and TL content are detected in them (Supplemental Figure S3, B and D). Notably, TAG-associated MCFAs (i.e. C8:0, C10:0, and C12:0) and C14:0 are 27%–99% lower in the 2D-KO lines under N⁺ (versus WT; Figure 3C). Meanwhile, under N⁻, TAG-associated C8:0, C10:0, and C12:0 are 89%–97% lower in the 2D-KO lines (versus WT; Figure 3D). As for total FA composition, no change is found under N⁺ (Supplemental Figure S4A), yet under N⁻, C18:3 increases by 61% while C20:4 decreases by 22%, compared to WT (Supplemental Figure S4B). Specifically, under N⁺, the composition of total MCFA remains unchanged among the 2D-, 2H-, 2D2H-KO lines and WT (Supplemental Figure S4C). Under N⁻, compared to WT, 2H- and 2D2H-KO results in the reduction of C8:0 by 31%–71% (Supplemental Figure S4D). Notably, the complementation lines show essentially identical phenotype to WT in N. oceanica (Figure 3, A–D). These in vivo results further pinpoint 2D as a TAG synthetase that prefers MCFA-CoAs as substrates in vivo.

Figure 3.

Phenotypes of NoDGAT2D- and 2H-manipulated N. oceanica lines. NoDGAT2D and 2H are separately and simultaneously knocked out via CRISPR/Cas9 to produce single-KO lines 2Dko-1, 2Dko-2, 2Hko-1, 2Hko-2, and double KO lines 2D2Hko-1 and 2D2Hko-2. Moreover, NoDGAT2D is complemented back into 2Dko-1 while NoDGAT2H is complemented back into 2Hko-1 to produce complement lines 2Dc-1, 2Dc-2, 2Hc-1, and 2Hc-2. A and B, TAG content under N⁺ (A) and N⁻ (B). C and D, Profile of TAG-associated FAs under N⁺ (C) and N⁻ (D). All experiments were conducted in one experiment performed with independent triplicate samples. Fold change of the FA percentage is calculated as log₁₀[Tc(NoDGAT2-modified lines)/Tc (WT)] (Tc = FA percentage). All data had been normalized and then displayed in the heatmap. Data are represented as mean ± sd (n = 3 biologically independent samples). ^*Significant change (P ≤ 0.01; one-sided Student’s t test) versus the WT. These statistical analyses apply to all analyses in this figure.

To probe the specific role of 2D in MCT metabolism, the in vivo lipidomes of the N. oceanica lines are profiled at the resolution of individual lipid species, based on electrospray ionization mass (ESI–MS), under N⁺ and N⁻ (see “Materials and methods”; WT as control). Totally 439 lipid species from 17 classes are identified under either N⁺ or N⁻ (Supplemental Data Set S1). In terms of lipid classes, under N⁺ and compared to WT, 2D-KO results in large increase in DAG, while 21%–58% decrease in TAG, phosphatidylcholine (PC), and phosphatidylethanolamine (PE), respectively (Supplemental Figure S5A). On the other hand, under N⁻, compared to WT, 2D-KO results in 51%–151% increase in DAG and hexose-ceramide (H-Cer), while 24%–62% decrease in LPC, TAG, and PE, respectively (Supplemental Figure S5B). Notably, under N⁺ and N⁻, in all 2D-KO lines, the DAG contents increase while TAG decrease (Figure 4, A and B), consistent with the role of 2D catalyzing the DAG-to-TAG conversion. Considering the enhancement of MCT synthesis by 2D (Figure 1C), it is possible that 2D prefers not just MCFA CoAs but also MCDs as substrates. In addition, 2D is potentially also involved in PC, PE, H-Cer, and LPC metabolism, via indirect effects caused by the altered membrane metabolism or signal transduction.

Figure 4.

Profile of the lipidomic classes in the NoDGAT2D or 2H mutant lines of N. oceanica. A and B, Fold change of the FA percentage is calculated as log₁₀[Tc(NoDGAT2-modified lines)/Tc (WT)] and displayed in the heatmap. The profiles of lipid classes are presented under N⁺ (A) and N⁻ (B). C and D, Species-revolved distribution of MCFAs in the total lipidome. Results are presented as the percentage for each lipid class, determined by liquid chromatography–MS under N⁺ (C) and N⁻ (D). CoQ, coenzyme QLPG, lysophosphatidylglycerol; LPI, lysophosphatidylinositol; MGMG, monogalactosylmonoacylglycerol. Data are represented as mean ± sd (n = 3 biologically independent samples). ^*Significant change (P ≤ 0.01; one-sided Student’s t test) versus the WT. These statistical analyses apply to all analyses in this figure.

Phenotypes of NoDGAT2H-transgenic lines

Similar to 2D, TAG content, dry weight, TL content, and FA composition are all altered upon 2H-KO, and recovered upon 2H-complementation. In 2Hko-1 and 2Hko-2, versus WT, TAG content is 35%–63% lower under both N⁺ and N⁻ (Figure 3, A and B). Considering the ex vivo and in vitro activities of 2H for MCT synthesis and the low overall content of MCT in total TAG (0.01%–0.05%; Wang et al., 2021) in WT, such significant reduction of TAG content caused by the KO of 2H is intriguing. This can be explained by the substrate activities of 2H for C16-FAs and C18-FAs (Figures 1, C and 2, C), and for MUFA (2H-expressing yeasts produce a low level of MUFA-TAGs, as detected by Raman-activated flow sorting; Wang et al., 2020), in addition to MCFAs. Together, these results suggest an important role of 2H in synthesizing total TAG, not just MCT, in N. oceanica.

As for dry weight, 2H-KO leads to 21%–30% increase under N⁺ (Supplemental Figure S3A), and 26% (2Hko-2) increase under N⁻ (Supplemental Figure S3B). In addition, compared to WT, TL contents of 2Hko-1 and 2Hko-2 increase under both N⁺ and N⁻ (by 42%–65% and by 37%–45%, respectively; Supplemental Figure S3, C and D). As for TAG-associated FA composition, under N⁺, the KO results in 72%–100% reduction in TAG-associated C8:0, C10:0, and C12:0, plus 4.2- to 9.4-fold elevation in TAG-associated C16:1, C18:1, and C20:5 (Figure 3C).

On the other hand, under N⁻, TAG-associated FA composition reduces by 79%–96% in C8:0, C10:0, and C12:0 and increases by 3.1- to 8.7-fold in C16:1, C18:1, and C20:4 (Figure 3D). Meanwhile, as for total FA composition, the KO leads to 26%–39% reduction in C8:0, C20:4, and C20:5, plus ∼17% elevation in C16:0 under N⁻ (Supplemental Figure S4B). Together with the complementation results, these data confirm 2H as an MCFA-CoA-preferred DGAT. Besides TAG content, 2H also exhibits the capacity to regulate the assembly of TL, biomass, and TAG-EPA in N. oceanica.

As for lipid classes, compared to WT (1) under N⁺, 2H-KO resulted in 18%–31% decrease in TAG. Thus, besides DAG and TAG, 2H mediates H-Cer and PE metabolism, too (Supplemental Figure S5C). (2) Under N⁻, in 2H-KO lines, DAG and H-Cer increase by 24%–84%, while TAG decrease by 7%–19%. However, 2H-KO results in ∼23% increase in PE content, which is opposite to 2D counterparts (Supplemental Figure S5D). Thus, in addition to DAG and TAG, 2H mediates H-Cer and PE metabolism.

Phenotypes of 2D–2H double KO lines

Furthermore, to pinpoint the in vivo phenotype of 2D and 2H, we genetically knockout 2H via clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (CRISPR/Cas9) in 2Dko-1 (Supplemental Figures S1, E and S2, C; see “Materials and methods”) to produce double KO lines 2D2Hko-1 and 2D2Hko-2. Compared to WT, 2D2Hko-1 and 2D2Hko-2 exhibit greatly altered TAG content, FA composition, biomass assembly, and TL content. Specifically, under N⁺ and N⁻, the TAG content is on average 76% (N⁺) and 78% (N⁻) lower in 2D2Hko-1 and 2D2Hko-2 (versus WT; Figure 3, A and B). As for the dry weight, the double KO leads to ∼27% increase in 2D2Hko-1 and 2D2Hko-2 under N⁺ (versus WT; Supplemental Figure S3A). As for total content, 2D2Hko-1 and 2D2Hko-2 exhibit ∼50% increase under N⁺, plus ∼40% increase under N⁻ (versus WT; Supplemental Figure S3, C and D). In addition, for TAGs in 2D2Hko-1 and 2D2Hko-2, under N⁺, MCFAs decrease by 61%–99%, while C16:1 and C20:5 increase by 2.0– to 7.5-fold (Figure 3C). On the other hand, under N⁻, MCFAs, C14:0, and C16:0 decrease (by 58%–86%), while C16:1, C18:1, C18:2, C18:3, and C20:4 increase (by 0.7- to 9.6-fold), as compared to WT (Figure 3D). Furthermore, in the TLs of double KO lines, under N⁺, C16:0 increases by 1.4-fold, yet C20:5 are 41% lower than WT (Supplemental Figure S4A). Meanwhile, under N⁻, total C8:0 and C20:5 are ∼40% lower, while C14:0, C16:0, and C18:1 are 21%–67% higher than WT (Supplemental Figure S4B). Notably, these phenotypes due to 2D and 2H double KO are similar to those resulted from 2H rather than 2D KO. Collectively, these genetic evidences suggest that both 2D and 2H prefer to use MCFA-CoAs for TAG synthesis.

Furthermore, 2D–2H-double KO results in increase of DAG and H-Cer, while decrease of TAG under N⁻ (Figure 4B;Supplemental Figure S5E). Moreover, the profiles of MCFA-containing lipid species reveal that MCDs greatly increase under N⁻, while MCTs are substantially reduced under N⁺ and N⁻ in all KO lines, especially in 2D–2H-double KO lines (Figure 4, C and D). Specifically, DAG (10:0/20:5), DAG (12:0/20:4), and DAG (12:0/20:5) are, respectively, elevated in all KO lines, under both N⁺ and N⁻. On the other hand, compared to WT, TAG (12:0/14:0/20:5) and TAG (12:0/20:4/20:5) are reduced in all KO lines. In addition, TAG (12:0/20:5/20:5) decreases by 53%–67% in the 2H- and 2D–2H-double KO lines, respectively (Supplemental Data Set S1). These lipidomic data thus confirm the roles of 2D and 2H in MCFA-DAG acylation.

Subcellular localization of NoDGAT2D and 2H reveals MCT-synthetic mechanism in N. oceanica

As 2D and 2H prefer MCFA-CoAs, we next probed the spatial features of MCT metabolism. Based on TMHMM prediction of transmembrane helices (Moller et al., 2001), subcellular localization cassettes are designed to encode part of the amino acids of 2D (1–398 aa) and 2H (1–114 aa) fused to the N-terminus of green fluorescent protein (GFP, as a reporter). These cassettes are then transformed into N. oceanica under the control of the strong, nearly constitutive NO20G00500 promoter (supplemental Figure S1, H and I) in N. oceanica (Gong et al., 2020). Confocal scanning laser microscopy reveals that green fluorescence is absent in WT cells (Figure 5A), but present in the cytoplasm of gfp-transformed line (Figure 5B). Meanwhile, transformants containing the cassette of gfp:NoVCP1 (a protein which had been confirmed to target to chloroplast stroma; Moog et al., 2015) exhibit a close spatial proximity with the autofluorescence of plastid (Figure 5C). On the other hand, GFP signals from 2D:gfp and 2H:gfp (Figure 5, D and E) represent a more discrete, network-like structure, which does not merge with the autofluorescence of the plastid, yet is more concentrated than the cytosolic GFP fluorescence pattern. Moreover, staining via ER Tracker (excited at 559 nm and detected at 590–620 nm) reveals ER-specific localization of 2D- and 2H-GFP product (Figure 5, G and H), in contrast to the control line (Figure 5F). Thus, 2D and 2H are located in the ER of N. oceanica.

Figure 5.

NoDGAT2D- and 2H-mediated MCT synthesis in the ER of N. oceanica. A, A WT N. oceanica cell that shows baseline plastid autofluorescence in the GFP channel. B–E, Nannochloropsis oceanica cells that express GFP without any N-terminal fusion (B), a signal:GFP fusion protein that targets chloroplast stroma (C), a NoDGAT2D:GFP fusion protein (D), or a NoDGAT2H:GFP fusion protein (E). F–H, ER Tracker labeling using WT (F), NoDGAT2D:gfp (G), or NoDGAT2H:gfp line (H). TML: transmission light; PAF: plastid autofluorescence; scale bar: 2 μm. The parameters were consistently set up for all fluorescent assays in this figure.

Taken together, the substrate preference and specialized subcellular spatial localization of 2D and 2H unveil an in vivo mechanism of MCT synthesis in N. oceanica. In addition to the LCT-assembly line mediated by NoDGAT2A, 2C, 2D, 2J, and 2K, respectively, there is an ER-localized assembly line dedicated for MCTs that are mediated by NoDGAT2H (Figure 6). One junction of these two assembly lines is NoDGAT2D, which contributes to the synthesis of both MCTs and LCTs. Intriguingly, NoDGAT2D and 2H are both thought to be derived from the heterotrophic eukaryotic secondary host (Xin et al., 2017), suggesting that the shared feature of MCT-synthetic activity between 2D and 2H may have an ancient origin. Collectively, MCTs are pooled into various forms of “TAG sink” (e.g. lipid droplets) and together underpin the rich content and diverse profile of TAGs in N. oceanica.

Figure 6.

A mechanistic model of NoDGAT2s-mediated TAG synthesis in N. oceanica. The first dimension of control for the profile is FA degree of unsaturation: (1) at the FA-class level (dark yellow ovals), NoDGAT2A, 2C, and 2D prefers SFA, PUFA, and MUFA as substrate, respectively and (2) at the FA-species level, NoDGAT2B, 2K, and 2J prefer hexadecenoic acid, LA, and EPA, respectively. The second dimension of control is FA carbon CL: (1) a generalist in FA CL, NoDGAT2D, assembles C8–C18 FAs into DAG and (2) a specialist, NoDGAT2H, incorporates only the medium-chain FAs of C8 and C12 into DAG. Not all intermediates or reactions are displayed. Arrows indicate catalytic steps in the pathway.

Producing MCTs in N. oceanica by gene stacking of TAG-synthetic pathway genes from diverse origins

To drive MCT overproduction, we started by overexpressing 2D and 2H, respectively, in N. oceanica (Supplemental Figure S1, F and G; see “Materials and methods”). However, compared to WT, MCT exhibits only minimal increase (up to 0.02% under N⁺ and 0.4% under N⁻) in 2D- and 2H-overexpression lines (Supplemental Figure S6). This can be due to the deficiency of DGAT substrates (such as MCFAs, MCPAs, and MCDs) in N. oceanica. To test this hypothesis, a gene-stacking strategy where combinatorial expression can elevate MCT production in N. oceanica is designed for elevated MCT production (Figure 7A). Besides DGATs, FA synthetase, TE, glycerol-3-phosphate acyltransferase (GPAT), phospholipid:diacylglycerol acyltransferase (PDAT), LPAAT, TAG-lipase (TGL), and lipid-related transcription factor (LTF), are also implicated in MCT production in N. oceanica (Figure 6). Specifical, ACPs in MCFA-ACPs are hydrolyzed by TE to form MCFAs. MCFAs are transformed into MCLAs and then MCPAs by LPAAT. After that, MCPAs are added into MCDs by PAP. Finally, MCDs are further acylated to form MCTs by DGATs. In addition, LTFs regulate all these enzymes to equilibrate cellular MCT synthesis (Figure 7A).

Figure 7.

Producing MCTs and MCLs by gene stacking in N. oceanica. A, Graphical summary of the simplified MCT assembly pathway that depicts the genes that have been used in combination for overproduction of oils. DGAT, diacylglyceryl transferase. B, The gene-stacking cassettes for transforming N. oceanica. mCpTE, rationally engineering CpTE, MT085746;LPAAT, Q42670.1; AtWR1, NP_001325502.1. C–F, MCT contents (C), MCT productivity (D), MCL contents (E), or percentage of MCFA in TLs (F) in the gene-stacking lines and in WT at 72 h under N⁻. MCL, medium-chain lipid. Here MCFAs/MCT/MCLs include octanoic acid (C8:0), decanoic acid (C10:0), and lauric acid (C12:0). Data are represented as mean ± sd (n = 3 biologically independent samples). ^*Significant change (P ≤ 0.01; one-sided Student’s t test) versus the WT. The statistical analyses apply to all analyses in this figure.

Specifically, co-expression of Umbellularia californica TE (Q41635.1), Cocos nucifera LPAAT (CnLPAAT, Q42670.1), Arabidopsis thaliana WRINKLED1 (AtWR1, NP_001325502.1), and Elaeis guineensis DGAT1 ( XP_010924968.1) results in over 50% MCFAs in total TAG-FAs in Nicotiana benthamiana, compared to <10% MCFAs of control (Reynolds et al., 2017). Notably, AtWR1 can regulate the expression of key genes in the glycolytic and FA biosynthetic pathways, and thus promote TAG accumulation in many land plants (Fei et al., 2020), yet its sequence homologs are absent in the N. oceanica genome (Gong et al., 2020). To probe the effect of AtWR1 in this microalga, we overexpressed AtWR1 in N. oceanica. The results suggest that AtWR1-overexpression can significantly elevate TAG content, but fails to change FA composition, under both N⁺ and N⁻ (Supplemental Figure S7). Thus, AtWR1 can potentially be combined with those enzymes with MCFA specificity (such as 2D and 2H) to enhance MCT productivity in N. oceanica. Moreover, we showed that introducing CpTE can boost C8:0 and C10:0 FAs, while expression of mCpTE (MT085746; which is designed to change substrate specificity from C8:0 and C10:0 to C12:0) can elevate C12:0 FA (Wang et al., 2021). Therefore, we stacked CpTE, mCpTE, CnLPAAT, AtWR1, 2D, and 2H in N. oceanica, by generating two groups of genetically engineered lines: (1) Group I, which consists of CpTE-CnLPAAT (182-1), mCpTE-CnLPAAT (182-2), 2D (182-3), and 2H (182-4), is designed to test separately the effects of the “substrate” modules (i.e. CpTE, mCpTE, and CnLPAAT) and the “product” modules (i.e. 2D and 2H) (Figure 7B;Supplemental Figures S1, F and G and S8, A and B). (2) Group II, which consists of CpTE-CnLPAAT-2D (182-5), CpTE-CnLPAAT-2H (182-6), mCpTE-CnLPAAT-2D (182-7), and mCpTE-CnLPAAT-2H (182-8), is designed to test the effects of combined substrate and product modules (Figure 7B;Supplemental Figure S8, C–F). (3) Group III, which consists of CpTE-CnLPAAT-2D-AtWR1 (182-9), CpTE-CnLPAAT-2H-AtWR1 (182-10), mCpTE-CnLPAAT-2D-AtWR1 (182-11), and mCpTE-CnLPAAT-2H-AtWR1 (182-12), is designed to finally elevate MCT production in N. oceanica (Figure 7B;Supplemental Figure S8, G–J). Together, phenotypes of 36 lines are profiled (three independently validated lines for each of the combinations; see “Materials and methods”).

Under N⁺ and N⁻, a huge variation in MCT content, TAG content, level of MCFA in TLs (MCLs), and MCT productivity is observed in most of the lines. Specifically, versus WT and under N⁺, (1) biomass dry weights are 9%–38% higher in the 182-10, 182-11, and 182-12 mutants, while remain unchanged for the other lines (Supplemental Figure S9); (2) TAG contents are 8%–465% higher in all the transgenic lines (Supplemental Figure S10); (3) MCT contents are up to 98-fold higher in transgenic lines (Supplemental Figure S11); (4) MCT productivity is up to 93-fold higher in mutants, except the 182-2 lines which remain unchanged (Supplemental Figure S12); (5) TL contents are up to 125% higher in transgenic lines, except the 182-1 and 182-4 lines which are unchanged (Supplemental Figure S13); (6) MCL contents are up to 1.6-fold higher in all the transgenic lines (Supplemental Figure S14); (7) MCL productivities increase by up to 125% in mutants, except the 182-5 and 182-12 lines which remain unchanged (Supplemental Figure S15).

Under N⁻, compared to WT, among the 30 lines of 182-3 to 182-12, (1) biomass dry weights do not change in all transgenic lines, except 182-2-1 and 182-4-1 which are ∼8% lower than WT (Supplemental Figure S9); (2) TAG contents are up to 1.2-fold higher in mutants, except 182-2 and 182-6 lines which remain unchanged (Supplemental Figure S10); (3) MCT contents are up to 67-fold higher in transgenic lines (Figure 7C;Supplemental Figure S11); (4) MCT productivity is up to 93-fold higher in mutants, except 182-1 and 182-4 which are unchanged (Figure 7D;Supplemental Figure S12); (5) TL contents are 0.7%–43% higher in all mutants, except 182-3 which remains unchanged (Supplemental Figure S13); and (6) MCL contents are up to 3.9-fold higher in mutants, except 182-5 which decreases by ∼12% (Figure 7E;Supplemental Figure S14). In addition, the percentages of MCFA in TLs increase by 1.7– to 2.9-fold in the 182-9, 182-10, 182-11, and 182-12 lines (Figure 7F); (7) MCL productivity increases by up to 125% in mutants, except 182-3 which remains unchanged (Supplemental Figure S15).

Notably, among all the gene-stacking lines, 182-11 (mCpTE + CnLPAAT + 2D + AtWR1) stands out in most of the oil-related performance metrices. Specifically, under N⁺ and N⁻, respectively, when compared to WT (for the three independent transformants of 182-11-1, 182-11-2, and 182-11-3), (1) the average growth rates are unchanged (Supplemental Figure S16); (2) TAG contents are up to 4.7-fold higher (Supplemental Figure S9); (3) MCT contents are up to 116-fold higher (Figure 7C;Supplemental Figures S10 and S17); (4) MCT productivity is up to 93-fold higher (Figure 7D;Supplemental Figure S12); (5) TL contents are up to 29% higher (Supplemental Figure S13); (6) MCL contents are up to 7.5-fold higher (Figure 7E;Supplemental Figure S14); (7) MCL productivity is up to 29% higher (Supplemental Figure S15). Furthermore, 182-11-1 stands out among the three lines under N⁻ (the peak oil-productivity phase): its 2.7  ± 0.5 mg L⁻¹·day MCT productivity represents 64.8-fold elevation versus the WT (Figure 7D).

Discussion

Oleaginous industrial microalgae are potential feedstock for efficient and sustainable supply of MCT. Here in N. oceanica, ex vivo, in vitro, and in vivo profiling reveals two ER-localized NoDGAT2s that can both assemble MCFAs into MCT. Such substrate specificity is then exploited to elevate MCT contents by 66-fold, via genetic stack of NoDGAT2D with mCpTE, CnLPAAT, and AtWR1.

The discovery of substrate-specific labor division based on FA carbon CL among DGATs in vivo suggests a previously unknown dimension of control for MCT assembly in the cell. It provides a logical explanation for the large dose of DGAT2s in this oleaginous microalga (which represents perhaps the densest genomic landscape of such genes in an organism: 12 putative DGAT2s (7 confirmed with TAG-synthetic activity so far) in a 30-Mb genome, versus just 4 in Zea mays and 2 in Homo sapiens). More importantly, a dual-dimension, hierarchical TAG-synthetic mechanism with an unprecedented degree of sophistication is revealed. Specifically, the first dimension of control for the TAG profile is FA degree of unsaturation, which consists of two hierarchies (Figure 6): (1) at the FA-class level, NoDGAT2A, 2C, and 2D prefer SFA, PUFA, and MUFA as substrate, respectively and (2) at the FA-species level, NoDGAT2K and 2J prefer the individual PUFA species of LA and eicosapentaenoic acid (EPA), respectively (Xin et al., 2017, 2019), while the lipid-droplet-localized NoDGAT2B appears to favor C16:1 (hexadecenoic acid) as substrate (Zhang et al., 2022). The second dimension of control, as revealed here, is FA carbon CL, which is also hierarchical: (1) a generalist in FA CL, NoDGAT2D, assembles C8–C18 FAs into DAG and (2) a specialist, NoDGAT2H, incorporates only the MCFAs of C8 and C12 into DAG.

Notably, NoDGAT2D, 2E, 2F, 2H, 2J, and 2K were proposed as originated from a heterotrophic secondary host, NoDGAT2C from a red algae, while NoDGAT2A, 2B, 2G, and 2I from a green algae (Wang et al., 2014). Thus, the distinct carbon CL specificity of NoDGAT2D and 2H seems to have survived a long history of evolution, as a result of functional differentiation among putative paralogs in an ancestral genome. Collectively, the dual-dimension, hierarchical mechanism in the last step of TAG assembly that involves at least 7 NoDGAT2s, together with the over 20 FA species present, apparently results in the highly diverse TAG profile (175 TAG species) of N. oceanica. To the best of our knowledge, such a degree of complexity and sophistication in TAG synthesis in vivo has not been reported in any other organisms.

Overproduction of MCT via genetic engineering has attracted great attention. In Camelina sativa, introduction of a DGAT and a LPAT from Cuphea sp. results in a four-fold increase of C10:0-TAG in seed oil (Iskandarov et al., 2017). In Yarrowia lipolytica, expression of a Clostridium perfringens ACP-TE elevated MCT to 32% of total TAG (Rutter et al., 2015). However, bacterial and fungal production systems are capital intensive and their operational cost can be expensive, while land plants suffer from relatively slow growth and low TAG productivity (TAG stored only in seeds). To enhance the MCT production in N. oceanica, the coupling of MCFA-ACP-preferred TEs and MCFA-CoA-preferred NoDGAT2s is designed to provide both pushing and pulling powers to drive MCT synthesis in N. oceanica. Considering that Tes can be designed via targeted mutagenesis to shift their enzymatic substrate specificity toward the MCFAs of a particular carbon length, it is foreseeable that the MCFA substrate specificity of DGATs can also be modulated, for example, by matching between DGAT and TE, the TAG synthesized can be enriched with particular MCFA species.

Despite the successful overproduction and enrichment of lauric acid (C12:0) in MCT in N. oceanica here, there are opportunities for further enhancement of MCT production. (1) Octanoic acid (C8:0) and decanoic acid (C10:0) are abundant intermediates in N. oceanica (Figure 7E), but incorporation of MCFA into TAG has so far been limited to C12:0 FA (Figure 7C). Thus DGATs with specificity for C8:0 or C10:0 FA should be mined to promote their incorporation into TAG to further elevate MCT content. (2) To further enrich MCT content, additional genetic manipulation such as halting MCT degradation, controlling FA elongation, and increasing total TAG content should be considered (Wang et al., 2021). For example, the stacking mid-chain-specific KASIV-coding genes can enhance the activity of mid-chain TEs (Lin et al., 2018). (3) MCT productivity can also be boosted by elevating algal biomass productivity, for example, via the engineered carbon fixation apparatus (Wei et al., 2017, 2019) or via the Blue-Light-Induced Oleaginousness strategy (Zhang et al., 2022). Such efforts should further enhance cellular MCT production, and accelerate the pace of engineering industrial microalgae into efficient and sustainable feedstock of “designer oils.”

Materials and methods

Strains and growth conditions

Saccharomyces cerevisiae strain H1246 with KO of DGA1, LRO1, ARE1, and ARE2 (Sandager et al., 2002) is maintained on YPD plates (1% [w/v] yeast extract, 2% [w/v] peptone, and 2% [w/v] glucose) solidified with 2% (w/v) agar. Nannochloropsis oceanica strain IMET1 is cultivated and nitrogen deficiency induced as previously described (Moustafa et al., 2009; Li et al., 2014; Jia et al., 2015). To determine MCFA substrate preferences among NoDGAT2s, supplementation of synthetic glucose cultures with octanoic acid (C8:0), decanoic acid (C10:0), and lauric acid (C12:0) is carried out with 90 μM of the appropriate FA in the presence of 0.1% (v/v) dimethyl sulfoxide. Cells are incubated for 20–22 h at 30°C and 150 rpm, harvested by centrifugation.

Unless specified, the microalga is cultivated and induced via nitrogen depletion as previously described (Moustafa et al., 2009; Li et al., 2014; Jia et al., 2015). It is cultured in modified f/2 liquid medium containing 35-g L⁻¹ sea salt, 1,000-mg L⁻¹ NaNO₃, 66.6-mg L⁻¹ NaH₂PO₄·H₂O, 3.65-mg L⁻¹ FeCl₃·6H₂O, 4.37-mg L⁻¹ Na₂EDTA·2H₂O, 0.0196-mg L⁻¹ CuSO₄·5H₂O, 0.0126-mg L⁻¹ Na₂MoO₄·2H₂O, 0.044-mg L⁻¹ ZnSO₄·7H₂O, 0.0109-mg L⁻¹ CoCl₂·6H₂O, 0.036-mg L⁻¹ MnCl₂·4H₂O, 5-µg L⁻¹ VB₁₂, 5-µg L⁻¹ biotin, and 0.1-mg L⁻¹ thiamine HCl. Cells are cultivated in liquid cultures under continuous light (∼50 ± 5 µmol photons m⁻² s⁻¹) at 25°C. For induction of oil production via nitrogen deficiency, the cells are harvested by centrifugation (3,500 g for 5 min) and then resuspended in the N⁻ medium (modified f/2 medium yet without NaNO₃). Cell growth is determined based on biomass dry weight, cell density, and optical density (OD₇₅₀).

To characterize the NoDGAT2D and 2H transgenic lines, mid-logarithmic phase algal cells (OD₇₅₀ of 2.6) are collected for validating successful transformants via polymerase chain reaction (PCR) amplification and Sanger sequencing (Supplemental Table S2). Positive lines are then cultured for further measurements. For phenotyping, the transgenic lines (plus the WT control) are grown to OD₇₅₀ of 4.5 ± 0.5 for 5 days, and then the content and FA profiles of both TAG and TL are tracked under N⁺ and N⁻.

Construction of vectors

Total N. oceanica RNA is extracted from the cells using Trizol reagents (Invitrogen, Carlsbad, CA, USA). cDNA is synthesized with PrimeScript RT reagent Kit (Takara, Kyoto, Japan) and used as a template for PCR. The primers used are listed in Supplemental Table S2 online. The amplified PCR products are digested with KpnI and EcoRI for NoDGAT2D and 2H. The products are then subcloned into pYES2 vector (Invitrogen) to form pXJ404 (Supplemental Figure S1A) and pXJ408 (Supplemental Figure S1B) for expression in the yeast S. cerevisiae (more details below). As a positive control in yeast expression assays, the yeast DGA1 gene encoding DGAT2 is cloned, in a manner similar to NoDGAT2s, to form pXJ412.

To construct the CRISPR/Cas9 vectors for NoDGAT2D and 2H KO, two DNA fragments with the hammerhead ribozyme and gRNA target sequence (Supplemental Table S2) are designed and formed a primer dimer via annealing. 5’-TGTGTGGAC ACGCACACAGG-3’ (207–226 bp for NoDGAT2D) and 5’-CCGCTCGGTCTTTCCGACGA-3’ (189–208 bp for NoDGAT2H) are used as the targeted sequence with a PAM sequence of GGG or AGG. The primer dimer is ligated to the BspQI-digested pNOC-ARS-CRISPR-v2 vector to form the KO vector pXJ515 (Supplemental Figure S1C) and pXJ522 (Supplemental Figure S1D). The Hyg marker is replaced by BleR for zeocin resistance. This vector expresses Cas9 and gRNA via the bidirectional Ribi promoter (using the CS and LDSP terminators, respectively). For NoDGAT2D–2H double KO, NoDGAT2H primer dimer is ligated to the BspQI-digested pNOC-ARS-CRISPR-v2 vector to form pXJ523 (Supplemental Figure S1E).

To construct the vectors for NoDGAT2D and 2H complementation and overexpression in N. oceanica, NoDGAT2D and 2H cDNAs are amplified (Supplemental Table S2) and subcloned into pXJ470 vector (containing Hyg marker for hygromycin resistance) separately, forming pXJ551 (containing the NoDGAT2D fragment flanked by XhoI and EcoRV sites, Supplemental Figure S1F) and pXJ425 (containing the NoDGAT2H fragment flanked by XhoI and EcoRI sites, Supplemental Figure S1G), respectively.

To construct vectors for GFP fusions, NoDGAT2D and 2H sequences are amplified from cDNA by PCR (Supplemental Table S2). Based on TMHMM (Moller et al., 2001), NoDGAT2D (1–398 aa) and 2H (1–114 aa) are used as signal peptides. The PCR-amplified gfp gene is inserted between KpnI and BamHI sites in the transformation vector pXJ53 (Wang et al., 2016), which carries an endogenous VCP1 promoter and the bleR gene as a selection marker. Amplified NoDGATs are then inserted into the pXJ53 vector in frame with GFP to produce pXJ53-2D (Supplemental Figure S1H) and pXJ53-2H (Supplemental Figure S1I). In addition, a known chloroplast stroma targeting marker (Moog et al., 2015) is incorporated into the pXJ53 vector as a reference. After sequencing confirmation, these plasmids are then transformed into N. oceanica.

Since CnLPAAT (GenBank Q42670.1), AtWR1 (GenBank NP_001325502.1), CpTE (GenBank AAC49179.1), and mCpTE (GenBank MT085746) are found to reinforce MCT synthesis (Reynolds et al., 2017; Wang et al., 2021), these genes are synthesized according to the codon usage of N. oceanica. For gene stacking, CnLPAAT-subcloned plasmid pXJ182 is used as a backbone. First, amplified CpTE or mCpTE is ligated to PstI-digested pXJ182-LAPPT to produce pXJ182-1 (Supplemental Figure S8A) and pXJ182-2 (Supplemental Figure S8B). Second, amplified NoDGAT2D or 2H is ligated to NdeI-digested pXJ182-1 or pXJ182-2, respectively, to form pXJ182-5 (Supplemental Figure S8C), pXJ182-6 (Supplemental Figure S8D), pXJ182-7 (Supplemental Figure S8E), and pXJ182-8 (Supplemental Figure S8F). At last, amplified AtWR1 is ligated to EcoRI-digested pXJ182-5/pXJ182-6/pXJ182-7/pXJ182-8 to form pXJ182-10 (Supplemental Figure S8G), pXJ182-11 (Supplemental Figure S8H), pXJ182-12 (Supplemental Figure S8I), and pXJ182-13 (Supplemental Figure S8J). The primers for cassette amplification were listed in Supplemental Table S2.

Transformation of yeasts and N. oceanica

Yeast mutant H1246 is transformed with an expression vector (pYES2.0) harboring coding sequence for NoDGAT2s using the lithium acetate procedure (Gietz and Schiestl, 2007), In addition, the empty vector pYES2.0 and the expression vector harboring the yeast DGA1 are transformed into the mutant strain as negative and positive controls, respectively. Transformants are then selected by growth on synthetic glucose medium (2% [w/v] glucose and 0.67% [w/v] yeast nitrogen base without amino acids) containing appropriate auxotrophic supplements.

Nuclear transformation of N. oceanica is performed for linearized vectors (only Cas9 vector is transformed as circular) using the high-voltage (11,000 V cm⁻¹) electroporation method (Wang et al., 2016). Mid-logarithmic-phase algal cells (OD₇₅₀ = 2.6) are collected for validation of successful transformants via PCR amplification (Supplemental Table S2). For NoDGAT2D- and 2H-CRISPR/Cas9, 12 PCR-positive monoclones are identified via 3730 sequencing. Among NoDGAT2D–CRISPR/Cas9-derived lines, two types of KO events are identified, including ‘TGTGTGGACACGCACACTAGG’ (2Dko-1) and ‘TGTGTGGACACGCA CA–GG’ (2Dko-2) (Supplemental Figure S2A). Among the NoDGAT2H-CRISPR/Cas9 lines, two types of KO are identified, including ‘CCGCTCGGTCTTTCCGATCGA’ (2Hko-1), and ’CCGCTC GGTCTTTCCGAACGA’ (2Hko-2) (Supplemental Figure S2B). Complemented strains for the NoDGAT2D and 2H KO lines are produced via transforming pXJ551 (supplemental Figure S1F) into 2Dko-1, while transforming pXJ552 (Supplemental Figure S1G) into 2Hko-1, respectively. NoDGAT2D–2H double KO lines are produced by transforming pXJ523 (Supplemental Figure S1E) into 2Dko-1 to produce double KO lines 2D2Hko-1 and 2D2Hko-2. Both 2D2Hko-1 and 2D2Hko-2 include modified sequences of ‘TGTGTGGACACGCACACTAGG’ for NoDGAT2D and ’CCGCTCGGTCTTTCCGAACGA’ for NoDGAT2H (Supplemental Figure S2C). For complemented and double KO transformation, both 300-μg mL⁻¹ hygromycin and 2.5-μg mL⁻¹ zeocin (final concentration) are used for screening.

Yeast microsome preparation and nonradiolabeled DGAT in vitro assay

Yeast microsome preparation and in vitro assay are conducted as described (Liu et al., 2017). Microsome fraction alone and microsome fraction with DAG are used as controls, and the background levels of TAG are subtracted from the data for DGAT activity analysis. The acyl CoAs (ammonium salt) tested include octanoyl CoA (C8:0 CoA), decanoyl CoA (C10:0 CoA), lauroyl CoA (C12:0 CoA), myristoyl CoA (C14:0 CoA), palmitoyl CoA (C16:0 CoA), stearoyl CoA (C18:0 CoA), 5z,8z,11z,14z,17z-eicosapentaenoyl CoA (C20:5 CoA), and docosahexaenoyl CoA (C22:6 CoA), while the 1-palmitoyl-2-oleoyl-sn-glycerol (16:0/18:1-DAG) is used as the acyl acceptor. All lipid reagents are purchased from Avanti.

Lipid isolation and quantification via TLC, GC–MS, and LC–MS

TLs are extracted from dried samples using chloroform:methanol (2:1 [v/v]) with 100-mM internal control of tri13:0 TAG and separated on a silica TLC plate using a mixture of solvents consisting of petroleum ether, ethyl ether, and acetic acid (70:30:1, by volume). To quantify the amount of TAG accumulated in N. oceanica strains that express the NoDGAT2D or 2H constructs, TAG bands are scraped from the TLC plate. FA methyl esters (FAMEs) are prepared by acid-catalyzed transmethylation of the TAG bands and then analyzed by GC–MS as described (Zhang et al., 2003). Mixed analytical standards of FAMEs and pentadecane are used as external and internal standard, respectively. The amounts of TAG and the profiles of TAG-associated FA are calculated based on the results derived from GC–MS. The chemicals used as standards are from Sigma, USA.

Lipidomic analysis is performed with an Agilent 6460 triple quadruple electrospray ionization mass spectrometer equipped with an Agilent 1260 high-performance liquid chromatograph. To achieve lipidome-wide quantitative analysis, algal lipid extracts containing the known amounts of internal standards are separated by two liquid chromatography gradients and detected in the positive and negative mode of ESI/MS, separately. Coenzymes, simple Glc series (Hex1Cers), lysophosphatidylcholines (LPCs), and PEs are analyzed as [M + H]⁺ at the positive mode of MS, while DAGs and TAG are detected as [M + NH₄]⁺; lysophosphatidylethanolamines (LPEs), (O-acyl)-1-hydroxy FAs (OAHFAs), PEs, phosphatidylglycerols (PGs), and phosphatidylinositols (PIs) are analyzed in the form of [M-H]⁻, ceramides (Cers), monogalactosyldiacylglycerols (MGDGs) and PCs are detected as [M+CH₃COO]⁻. Precursor ion and neutral loss scanning modes are employed to identify lipid species for a given class according to previously described methods. Product ion mode is used to resolve the acyl groups of each lipid species.

Nitrogen is used as nebulizing gas (at 0.3 Bar) and a dry gas (4 L, 21 min at 200°C). The spray capillary voltage is 3,700 V for the negative ion mode and 4,200 V for the positive ion mode. Internal standards, including Cer (d17:1/24:0), palmitic acid, octadecanoic acid, PC (19:0/19:0), PC(15:0/15:0), LPI (17:1), PE (15:0/15:0), PG (15:0/15:0), PG (15:0/15:0), SM (d18:1/17:0), SM (d18:1/12:0), daturic acid, DAG (14:0/14:0), LPC (19:0), LPE (14:0), TAG (15:0/15:0/15:0), and TAG (17:0/17:0/17:0), are added to the lipid extract prior to MS analysis. Lipid extracts are separated at 40°C on a ZOBAX SBC18 column (1.8 μm, 2.13 150 mm; Agilent, Santa Clara, CA, USA) and an Extend C18 column (1.8 μm, 2.13 150 mm; Agilent) for positive- and negative-mode MS analysis, respectively. For positive mode, the mobile phases are methanol:acetonitrile:water (19:19:2) (A) and isoproponal (B) containing 0.1% formic acid and 10-mM ammonium acetate; the liquid chromatography (LC) gradients are as follows: 0 min, 90% A and 10% B; 5 min, 90% A and 10% B; 25 min, 60% A and 40% B; 60 min, 45% A and 55% B; 66 min, 45% A and 55% B; and 68 min, 90% A and 10% B. For negative mode, the mobile phases are 85% methanol (A) and isopropanol containing 0.025% NH₄OH; the LC gradients are as follows: 0 min, 95% A and 5% B; 15 min, 85% A and 15% B, 22 min, 45% A and 55% B; 42 min, 45% A and 55% B; and 44 min, 95% A and 5% B. The flow rate is 0.2 mL/min. Multiple reaction monitoring scanning mode is used for MS detection: PC ([M + H]⁺→m/z 184), DGTS ([M + H]⁺→m/z 236), PE ([M + H]⁺→ [M + H-141]⁺), PG ([M - H]2→ m/z 153), and SQDG ([M - H]2→ m/z 225, PI ([M - H]+→m/z 241). ForMGDG, DGDG, and TAG, single-stage MS scanning mode is employed for detection of [M + NH₄]⁺. For quantification, calibration standards of each lipid class are titrated relative to a constant amount of internal standard. Relative intensity ratios are plotted against their molar concentration ratios to establish standard curves. The calibration standards included PC 18:1/18:1, DGTS 16:0/16:0, PE 18:0/18:1, MGDG 16:3/18:3, DGDG, PI 18:1/18:1, PG 18:0/18:1 (all from Avanti Polar Lipids); SQDG 16:0/18:3 (Indofine Chemical); and TAG 16:1/16:1/16:1, TAG 16:0/18:1/16:0, TAG 18:1/16:0/18:1, and TAG 18:1/18:1/18:1 (all from Sigma-Aldrich, St Louis, MO, USA).

Fluorescence microscopy

Localization of the various DGAT:GFP fusion proteins (described above) in N. oceanica is carried out with a laser-scanning confocal microscope, FluoView FV1000 (Olympus, Tokyo, Japan). The wavelength of laser1 was set as 488 nm for GFP observation, and the transmissivity was set as 50.7%. Fluorescence of GFP was excited at 488 nm and detected at a band width of 500–530 nm. The intensity (HV) was around 630–650, and the gain was 1 and the offset was 80–90. The wavelength of laser2 was set as 559 nm for chlorophyll autofluorescence observation, and the transmissivity was set as 57%. The autofluorescence was excited at 559 nm and detected at a band width of 650–750 nm. The intensity (HV) was around 570–600 (the intensity was set around 380–400 during the staining experiment), and the gain was 1 and the offset was 60–80.

To validate the ER-localization of NoDGAT2D and 2H, the NoDGAT2:gfp-transformed lines and the WT strain are stained with 0.01 μM (final concentration) ER-Tracker Red (BODIPY TR Glibenclamide) for 20 min at 37°C and then twice-washed with f/2 liquid medium. To be specific, the laser1 was still set as 488 nm for GFP observation and laser2 was set as 559 nm for both chlorophyll autofluorescence and fluorescence of ER-Tracker. Fluorescence of ER-tracker is excited at 559 nm and detected at 590–620 nm. The intensity (HV) was around 470–500, and the gain was 1 and the offset was 70–80.

Statistical analyses

All experiments were in triplicates, with results presented as mean ± standard deviation (SD). Statistical analyses were performed using Graphpad Prism version 5 (GraphPad, La Jolla, CA, USA). The P-values were calculated via one-way analysis of variance.

Accession numbers

The GenBank Accession IDs of validated full-length sequences of the 11 NoDGAT2s are: 2A (KX867956), 2B (KX867957), 2C (KX867958), 2D (KX867959), 2E (KX867960), 2F (KX867961), 2G (KX867962), 2H (KX867963), 2I (KX867964), 2J (KX867965), and 2K (KX867955). Additional information on N. oceanica IMET1 genome annotation is available at http://nandesyn.single-cell.cn.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Map of vectors for genetic manipulation of NoDGAT2D and 2H.

Supplemental Figure S2. Genome sequences of the editing sites in the NoDGAT2-KO lines and WT of N. oceanica.

Supplemental Figure S3. Phenotypes of the NoDGAT2D- and 2H-manipulated N. oceanica lines.

Supplemental Figure S4. Profile of total FAs in the NoDGAT2D- and 2H-manipulated N. oceanica lines.

Supplemental Figure S5. Comparison of lipid classes in the NoDGAT2D- and 2H-manipulated N. oceanica lines.

Supplemental Figure S6. MCT contents in the NoDGAT2D- and 2H-overexpression lines and WT of N. oceanica.

Supplemental Figure S7. Comparison of oil accumulation between AtWR1-overexpressing N. oceanica lines and WT under N⁺ and N⁻.

Supplemental Figure S8. Expression vector cassettes for MCT production in the transgenic N. oceanica lines.

Supplemental Figure S9. Comparison of biomass dry weight among the N. oceanica gene-stacking lines and WT under N⁺ and N⁻.

Supplemental Figure S10. Comparison of TAG contents among the multi-gene-stacking lines and WT of N. oceanica under N⁺ and N⁻.

Supplemental Figure S11. Comparison of TAG-associated FA profiles among the multi-gene-stacking lines and WT of N. oceanica under N⁺ and N⁻.

Supplemental Figure S12. Comparison of MCT productivity among the N. oceanica multi-gene-stacking lines and WT under N⁺ and N⁻.

Supplemental Figure S13. Comparison of TL contents among the multi-gene-stacking lines and WT of N. oceanica under N⁺ and N⁻.

Supplemental Figure S14. Comparison of total FA profiles among the multi-gene-stacking lines and WT of N. oceanica under N⁺ and N⁻.

Supplemental Figure S15. Comparison of MCL productivity among the multi-gene-stacking lines and WT of N. oceanica under N⁺ and N⁻.

Supplemental Figure S16. Growth kinetics among the three N. oceanica 182-11 lines and WT type.

Supplemental Figure S17. Gas chromatograph-mass spectrogram of TAG-derived FAMEs from the three N. oceanica 182-11 lines under N+.

Supplemental Table S1. Sequence features of the 11 NoDGAT2s in N. oceanica strain IMET1.

Supplemental Table S2. Nucleotide sequences of the primers used in this study.

Supplemental Data Set S1. Lipidomic data from NoDGAT2D- and 2H-manipulated lines and the WT of N. oceanica.

 

J.X. and Y.X. designed research. Q.W., N.L., and X.D. generated and screened transgenic lines of Nannochloropsis oceanica. C.S. performed the protein localization assay. Y.X. conducted the GC–MS assay. C.H., X.S., and G.X. conducted the lipidomic assay. J.X. and Y.X. analyzed data and wrote the article.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is Jian Xu (xujian@qibebt.ac.cn).