Exploring a blue-light-sensing transcription factor to double the peak productivity of oil in Nannochloropsis oceanica

Identifying NobZIP77 as a transcription factor regulating N− induced TAG synthesis

In the industrial oleaginous microalga N. oceanica IMET1, 125 TFs were identified via characteristic domains of plant TFs, and a global regulation network predicted links of 35 TFs to 2801 target genes¹¹. To pinpoint key TFs on N− induced TAG synthesis, we analyzed the transcript levels of these TFs over six timepoints (3, 4, 6, 12, 24, and 48 h) under N+ and N− conditions²³. Altogether, 32 TFs respond to N−, with eleven upregulated and fourteen downregulated (seven of them showing distinct trends of regulation among timepoints). Of them, sixteen TFs were predicted to regulate lipid-related genes (LRGs; Fig. 1A, Supplementary Data 1)¹¹, and among them, scaffold00007.g77 (or g77, Genbank ID MT273120), which is downregulated by N−, targets seven such genes that include a diacylglycerol acyltransferases (DGAT) and a few additional enzymes that are related to lipid metabolism (Supplementary Fig. 1A, Supplementary Data 1). Notably, transcripts in the LRG of g77 exhibit distinct or even opposite fold-changes, e.g., upregulation of the lipase gene g10411 yet downregulation of the DGAT gene of g6725 (Supplementary Data 1). As the activities of lipases and the DGAT are exactly opposite (oil degradation and assembly, respectively), these results indicate g77 as a negative regulator of TAG synthesis.

The g77 gene encodes a 466-aa protein harboring three conserved domains: nuclear localization signal (NLS; 209–219 aa), Per-ARNT-Sim (PAS; 323–437 aa), and basic leucine zipper (bZIP; 246–284 aa; Fig. 1B). This bZIP domain, consisting of a DNA-binding region (246–254 aa) and a leucine zipper (255–284 aa), is characterized by an asparagine and an arginine residue that are conserved among higher plants, fungi, microalgae, and animals (thus g77 was termed NobZIP77; Supplementary Fig. 2A). Consistently, homologs of the NobZIP77-bZIP domain are broadly distributed in Stramenopiles, Viridiplantae, Fungi, and Metazoa (Supplementary Fig. 2B).

To probe its in vivo role, NobZIP77 was overexpressed or knocked down in N. oceanica (Supplementary Fig. 3; “Methods”). Compared to the EV (empty-vector transformed) control, NobZIP77 transcripts in overexpression lines (NobZIP77o-1 or NobZIP77o-2) increased by 1.7–2.2-fold, while in knockdown lines (NobZIP77i-1- or NobZIP77i-2) reduced by 21.0–31.7% at 0 h, 6 h, 24 h, and 48 h under N− (Supplementary Fig. 4A). Under N+, NobZIP77 overexpression results in 9.03% and 1.23% lower growth rates (the average over the eight-day culture) than EV, while NobZIP77-knockdown lines remained unchanged (Supplementary Fig. 4B). Therefore, under such experimental conditions, NobZIP77 expression level does not seem to exert significant effects on microalgal growth.

As for TAG content, significant difference was observed under N+ (i.e., at 0 h under N−). Specifically, in NobZIP77o-1 and NobZIP77o-2, TAG content is 22.3% and 35.2% lower than EV. In contrast, in NobZIP77i-1 and NobZIP77i-2, it is 53.3% and 196.7% higher than EV, respectively (Fig. 1C and Supplementary Fig. 4C). As for TAG yield, compared to EV, NobZIP77i-1 exhibited a 31.1% higher level under N+, while NobZIP77i-2 increased 31.8–197.1% during 0–72 h under N− (Fig. 1D; Supplementary Fig. 4D). Furthermore, in NobZIP77-knockdown lines, under N+, TAG-associated C16:1 is 59.0–106.4% higher, while C18:0 and C20:5 are 31.2–35.5% and 16.8–63.0% lower than EV, respectively. In contrast, in the NobZIP77-overexpression lines, compared to EV, TAG-associated C16:1 is 19.6–34.7% lower, while C18:0 and C20:5 are 48.6–66.5% and 41.2–89.2% higher (Supplementary Fig. 4E). As for total fatty acids (TFA) content, no significant difference from EV was observed in NobZIP77 overexpression or knockdown lines (Supplementary Fig. 4F). Therefore, by increasing C16:1 while reducing C18:0 and C20:5 in TAGs, NobZIP77 negatively regulates TAG content and yield in N. oceanica, yet preserves its growth rate.

Notably, among the two overexpression lines (also among the knockdown lines), the expression pattern of NobZIP77 is consistent, but not TAG content or yield. This can be due to the unpredictable effect of integration of overexpression cassette or silencing of the RNAi constructs. To settle this concern, NobZIP77 was knocked out in N. oceanica (Supplementary Fig. 5A; “Methods”). Under N+, knockout lines showed essentially no effects on microalgal growth (Supplementary Fig. 6A). However, significant difference in TAG content was observed in NobZIP77-knockout lines. Specifically, versus the wild type (WT) and under N−, TAG content was 178.1% and 51.0% higher at 0 h and 24 h for NobZIP77ko-1, while 160.0% and 49.5% higher at 0 h and 24 h for NobZIP77ko-2 (Fig. 1E and Supplementary Fig. 6B). For TAG yield, NobZIP77 knockout resulted in 61.6% and 63.6% increase at 0 h, respectively (52.7% and 44.4% at 24 h; under N−; Fig. 1F and Supplementary Fig. 6C). Notably, TAG-associated C18:0 and C18:2 are 26.0–32.4% and 59.0–59.5% lower in the NobZIP77-knockout lines under N+ (versus WT; Supplementary Fig. 6D). As for TFA content, no difference was found versus WT via knockout, under either N− or N+ (Supplementary Fig. 6E). These results indicated that the knockout lines behave similarly as silencing lines. To further pinpoint its in vivo role, NobZIP77 was genetically complemented in its knockout lines (Fig. 1D and Supplementary Fig. 5; “Methods”). Importantly, these complementation lines showed essentially identical phenotype to WT in N. oceanica. These results firmly pinpoint NobZIP77 as a TF that inhibits N− induced TAG synthesis.

NoDGAT2B is a key target of NobZIP77

To identify the binding sites of NobZIP77, a phylogenomic footprinting pipeline was developed via MERCED²⁴, revealing seven LRGs (Supplementary Data 1). These LRGs encode multiple TAG/FA metabolism-related enzymes, including DGAT, phosphatidic acid phosphohydrolase (PAP), fatty acid elongase (FAE), and lipase. To validate such regulatory links, the transcript levels of all seven LRGs were quantified by RT-qPCR in the NobZIP77 overproduction, knockdown, and knockout lines (Fig. 2A; Supplementary Fig. 7; Supplementary Fig. 8). Notably, NoDGAT2B (g6725; Genbank ID KX867957), a member of the N. oceanica type-2 DGAT family^21,22, is the only LRG that exhibits a temporal transcription pattern precisely opposite to NobZIP77 (consistent with the transcriptome data; Fig. 2A; Supplementary Fig. 7; Supplementary Data 1). The NoDGAT2B transcript is 50.4% higher in NobZIP77ko-1, 128.9% higher in NobZIP77ko-2, 127.5% higher in NobZIP77i-2, while 26.0% lower in NobZIP77o-2 (average fold change over 0 h, 6 h, 24 h, and 48 h under N−; versus EV; Fig. 2A; Supplementary Fig. 8), which suggests the repression of NoDGAT2B by NobZIP77. Such a trend of transcript fold change is also shared by the PAP of g2171 and the lipase of g4187 at the early phase of N− (Supplementary Fig. 8). Thus, NoDGAT2B is one key target of NobZIP77.

To probe this hypothesis and test whether NobZIP77 directly interacts with the promoter of NoDGAT2B (pNoDGAT2B; and of additional predicted target genes or PTGs), electrophoretic mobility shift assays (EMSAs) were used to identify specific band shifts of ACGT-harboring promoter probes (the binding sites of bZIP-type TF genes²⁵). Recombinant maltose-binding protein (MBP)—NobZIP77 was expressed in Escherichia coli and purified for EMSA (“Methods”). ACGT-harboring probes of the NobZIP77-PTG promoters [g3857 (TAG-lipase), g2171 (PAP), g10041 (lipase), NoDGAT2B, and g5641 (FAE)] result in mobility shift, indicating the direct binding between the NobZIP77-PTG promoters and NobZIP77 (Supplementary Fig. 9). Notably, the degree of mobility shift is greatly reduced by adding unlabeled fragment of NobZIP77-PTG promoter, yet no binding between NobZIP77 and the negative controls of g4867 (hsp70) and g9477 (actin) was detected (Fig. 2B), which supports the specificity of binding between NobZIP77 and pNoDGAT2B.

Next, we tested the in vivo binding between NobZIP77 and pNoDGAT2B, by chromatin immunoprecipitation (ChIP) qPCR assays in N. oceanica. A GFP gene fused immediately downstream of the full-length NobZIP77 cDNA was transformed into N. oceanica and positive expression in transgenic lines (OE1) was confirmed with laser-scanning confocal microscopy (Supplementary Fig. 10). Under N+, OE1 and WT cells (cell density of 2 × 10⁷ cells/mL) were subject to chromatin extraction and immunoprecipitation with anti-GFP antibody (“Methods”). The ChIP products were analyzed by qPCR and the fold change of pNoDGAT2B amount calculated via 2^−ΔΔCT (β-actin as internal reference). Notably, the amount of pNoDGAT2B in OE1 is 2.2-fold of that in WT (Fig. 2C), supporting the direct binding of NobZIP77 to the ACGT-harboring pNoDGAT2B sequence in vivo.

Furthermore, to probe the functional consequence of the binding between NobZIP77 and pNoDGAT2B, transient NoDGAT2B transcription level was measured via a reporter-effector system in Arabidopsis mesophyll protoplast (TEAMP; “Methods”)²⁶. Arabidopsis mesophyll protoplasts were employed in transient analysis of promoter activity in vivo, and β-glucuronidase (GUS) used as reporter for interaction with effector. Specifically, the 900 bp pNoDGAT2B was linked to the GUS reporter gene to create the reporter construct (pNoDGAT2B promoter-:GUS), and the full-length NobZIP77 coding sequence (CDS) was ligated downstream of the 35S promoter to generate the effector construct (CaMV35S promoter-:NobZIP77). The two constructs were co-transfected into A. thaliana leaf protoplasts and the GUS activity measured (“Methods”). These ex vivo experiments revealed that relative GUS activity of the reporter-effector co-transfected protoplasts was 31.9% lower than the protoplasts transfected with the reporter construct alone, indicating an inhibitory effect on NoDGAT2B transcription due to the binding of NobZIP77 to pNoDGAT2B (Fig. 2D).

To test whether the NobZIP77-pNoDGAT2B binding inhibits NoDGAT2B transcription in vivo, a GFP reporter system was designed for quantifying pNoDGAT2B-driven gene expression (“Methods”). To avoid potential interference of the inhibition by the endogenous NobZIP77, this experiment was run in a NobZIP77-knockout line of N. oceanica (NobZIP77ko1). A control vector pXJ545 was constructed with a core cassette of pNoDGAT2B–GFP-Ttub-Ptub-ble-Tvcp (Supplementary Fig. 3F) and a reporter vector pXJ546 designed with a core cassette of pNoDGAT2B–GFP-Ttub-Ptub-NobZIP77-TpsbA-Ptub-ble-Tvcp (Supplementary Fig. 3G). In all NobZIP77ko1 lines transformed with pXJ545 and pXJ546, both GFP transcription level and fluorescence intensity were measured. In NobZIP77-p2B (NobZIP77-pNoDGAT2B-transformed N. oceanica) and versus p2B (pNoDGAT2B-transformed line), the GFP transcript level was reduced by 69.7% under N+ and 58.8% under N− (Fig. 2E), whereas fluorescence intensity of GFP was reduced by 55.8% (under N+; Fig. 2F and Supplementary Fig. 11). These results support the inhibitory effect of NobZIP77-pNoDGAT2B binding on NoDGAT2B transcription in vivo. Collectively, these in vitro, ex vivo, and in vivo evidence suggest that, in N. oceanica, NobZIP77 inhibits NoDGAT2B expression by binding to its promoter via the ACGT core sequence.

The 346-aa NoDGAT2B is encoded via two exons and contains two transmembrane domains. We have validated the TAG-synthetic function of five NoDGAT2s via ex vivo assays in Saccharomyces cerevisiae strain H1246, a TAG-deficient quadruple knockout mutant^21,22, yet TAG was not detected in NoDGAT2B-containing H1246, probably due to the absence in yeast of proper subcellular structure or cofactor for TAG synthesis. Meanwhile, in vitro assay of NoDGAT2B was unavailable due to the failure of microsome acquirement, thus we turned to the direct characterization of NoDGAT2B function in vivo. NoDGAT2B was overexpressed or knocked down (via RNAi) in N. oceanica (“Methods”). Compared to EV, NoDGAT2B transcripts increased by 4.8–8.7-fold in its overexpression lines 2Bo1 and 2Bo2, while reduced by 41.9–59.0% in the knockdown lines of 2Bi1 and 2Bi2, under N− (Supplementary Fig. 12A). Moreover, under N+ and at the early phase of N−, TAG content is 18.7–71.9% and 23.2–83.9% higher in 2Bo1 and 2Bo2, respectively, while 19.1–64.3% and 23.6–48.9% lower for 2Bi1 and 2Bi2 (versus EV; Fig. 2G, Supplementary Fig. 12B, C). Notably, such difference in TAG content versus EV disappears at the late phase of N− (after 48 h), which indicates the role of other NoDGAT2s in this process. Furthermore, in the NoDGAT2B-overpression lines, under N+, TAG-associated C16:1 is 29.3% and 30.0% higher, while C18:0 and C20:5 are 52.9–53.0% and 29.6–29.8% lower than EV. On the opposite, in the NoDGAT2B-knockdown lines, TAG-associated C16:1 is 32.3% and 45.1% lower, while C18:0 and C20:5 are 82.1–172.0% and 49.9–106.5% higher than EV (Supplementary Fig. 12D). Thus, the phenotypes of NobZIP77 and NoDGAT2B manipulation are opposite, further supporting the negative regulation of NoDGAT2B by NobZIP77.

To further pinpoint its in vivo role, NoDGAT2B was knocked out via CRISPR/Cas9 (Supplementary Fig. 13) and complemented in N. oceanica (Supplementary Fig. 3; “Methods”). No effects on growth were observed (Supplementary Fig. 14A), yet TAG content changed upon knockout and recovered upon complementation. Specifically, under N−, TAG content is 30.6–82.6% lower in 2Bko-1, and 27.8–78.2% lower for 2Bko-2 (versus WT; Fig. 2H; Supplementary Fig. 14B). As for TAG yield, NoDGAT2B knockout led to 26.8–56.5% and 29.5–58.4% reduction, respectively, under N− (Supplementary Fig. 14C). Moreover, the knockout resulted in 2.4–3.8% and 51.5–60.1% lower TAG-associated C16:0 and C16:1, plus 42.4–48.8%, 403.0–571.9%, 124.2–135.2% and 124.3–189.6% higher TAG-associated C18:0, C18:3, C20:4 and C20:5 (under N+, Supplementary Fig. 14D). As for TFA content, no changes were found (Supplementary Fig. 14E). These results are consistent with the NoDGAT2B knockdown and overexpression experiments.

Moreover, to prove that differential regulation of DGAT2B is relevant for the lipid accumulation phenotypes of the NobZIP77-knockout and overexpression lines, we genetically knocked out NoDGAT2B via CRISPR/Cas9 in the NobZIP77-knockout line of NobZIP77-ko1 (Supplementary Fig. 15; “Methods”). As compared to WT, the resulted double-knockout lines of NobZIP77-2B-ko1 and NobZIP77-2B-ko2 both exhibit no change in growth rate (Supplementary Fig. 16A), yet significant decrease in both TAG content and TAG yield. Specifically, at 0 h and 24 h under N−, the TAG content is 28.1% and 41.3% lower in NobZIP77-2B-ko1 (19.4% and 32.6% lower for NobZIP77-2B-ko2; versus WT; Fig. 2I and Supplementary Fig. 16B). As for the TAG yield, at 0 h and 24 h under N−, the double knockout led to 27.6% and 46.7% reduction in NobZIP77-2B-ko1 (21.8% and 42.6% in NobZIP77-2B-ko2; versus WT; Supplementary Fig. 16C). In addition, for NobZIP77-2B-ko1 and NobZIP77-2B-ko2, under N+, TAG-associated C16:1 is 16.6 and 29.3% lower, yet TAG-associated C18:0, C18:3, C20:4 and C20:5 are 29.6 and 40.7%, 58.3 and 105.3%, 67.9 and 111.0%, and 34.5 and 55.9% higher, respectively (Supplementary Fig. 16D). As for TFA content, no changes were found in either double-knockout lines (except for 48 h under N−; Supplementary Fig. 16E). These TAG-related phenotypes due to the NobZIP77 and DGAT2B double-knockout are like those resulted from NoDGAT2B knockout (Supplementary Figs. 14 and 16), while opposite to those from NobZIP77 knockout (Supplementary Figs. 6 and 16).

NobZIP77 as a blue-light sensor that releases the NoDGAT2B promoter under blue light

NobZIP77 harbors a conserved light–oxygen–voltage (LOV) domain, which belongs to the PAS superfamily that specifically senses blue light (BL) by a noncovalently bounding to flavin cofactor [FMN (flavin mononucleotide); Supplementary Fig. 17]²⁷. For NobZIP77-LOV, a series of key residues are conserved among its homologs in higher plants, fungi, and microalgae (Supplementary Fig. 18A, B). Together, the features of the bZIP and LOV domains and the NCR-Q-V-IR-N-N-F-Q conserved residues (Supplementary Fig. 18C) support NobZIP77 as an AUREO1-subfamily member in AUREOCHROME (AUREO), which is a type of BL receptors in stramenopiles²⁸, and raise the hypothesis that it regulates TAG assembly in response to BL. Moreover, its characteristic order of the sensor and effector domains, with the sensor domain (LOV) at the C-terminus and the effector domain (bZIP) at the N-terminus, strongly suggests NobZIP77 as an aureochrome²⁸.

To probe the effect of BL on NobZIP77-NoDGAT2B interaction, EMSA was performed in the dark, or under continuous illumination of white light (WL) or BL (Supplementary Fig. 19; “Methods”). The shifting of NobZIP77-pNoDGAT2B band was observed starting at the NobZIP77 concentration of 0.012 mM under the dark, at 1.5 mM under BL, and at 0.75 mM under WL (Fig. 3A). Thus, the binding between NobZIP77 and pNoDGAT2B, which shuts down NoDGAT2B-mediated TAG assembly, can be relieved by BL (and to a lesser degree, by WL).

To probe the in vivo effect of BL on the NobZIP77-NoDGAT2B interaction, multiple light wavelengths were tested in cultivating N. oceanica WT and NobZIP77ko-1 (Fig. 3B). For WT, under BL (but not green light (GL)), the NoDGAT2B transcript is 1.6- and 13.3-folds higher than under the dark (for both N+ and N−; quantified via qRT-PCR; Fig. 3C). Intriguingly, for NobZIP77ko1, no change of NoDGAT2B transcript was detected under either BL or GL versus the dark (Fig. 3E). Therefore, BL but not GL can specifically induce NobZIP77-inhibited NoDGAT2B expression.

To probe the consequence of such BL-specific induction, the TAG content of N. oceanica quantified via single-cell Raman spectra (SCRS²⁹; “Methods”). In WT, under N−, TAG content is 65.89% higher under BL, 25.34% higher under GL and 3.82% lower in the dark (versus N+, Fig. 3D), suggesting BL/GL (but not the dark) promotes TAG production under N−, with BL exerting much stronger stimulatory effect than GL. Moreover, in NobZIP77ko-1, TAG content is reduced by 11.78% under BL plus N−, revealing a temporal lag of TAG content in response to BL under N− versus WT (p < 0.05; Fig. 3D–F), despite a similar light-quality-responsive pattern, i.e., stronger stimulation of TAG synthesis by BL than by GL (for NobZIP77ko-1, under N−, TAG content is 19.69% and 49.83% higher under GL and BL, respectively, versus the dark; Fig. 3F). Therefore, NobZIP77 mediates N− induced TAG synthesis via its sensing of BL.

A working model of stress-induced TAG synthesis mediated by NobZIP77 in N. oceanica

To probe how the interactions among BL, NoZIP77, and TAG synthesis take place, we probed the subcellular locations of NobZIP77 and NoDGAT2B by fusing the GFP-coding gene (gfp) immediately downstream of the full-length NobZIP77 or NoDGAT2B cDNA and transforming them into N. oceanica, respectively. The NobZIP77:gfp fusion protein is colocalized with the DAPI-stained nuclei, indicating nucleus targeting (Fig. 4A), while the NoDGAT2B:gfp fusion protein is colocalized with the Nile red-stained lipid droplet (LD), suggesting LD targeting (Fig. 4B). The nucleus is physically surrounded by the plastid³⁰ (Fig. 4A, B), which as the dominating sink of chlorophyll a (N. oceanica does not harbor chlorophyll b) can shield nucleus from BL by absorbing the 420–663 nm wavelength³¹. Thus, when chlorophyll a dramatically declines under N− (Fig. 4C), the nucleus would be exposed to more BL, which via the LOV domain of NobZIP77 triggers release of this TF (via allosteric effects such as de-dimerization³²) from pNoDGAT2B.

Taken together, we propose an in vivo mechanistic model of N− induced TAG synthesis in N. oceanica (Fig. 4D). NobZIP77 normally (i.e., N+) represses NoDGAT2B transcription by directly binding to the latter’s promoter; yet under N−, the plastid is no longer able to shield the NobZIP77-located nucleus from BL (due to the degradation of chlorophyll a in plastid), and the resulted exposure of NobZIP77 to BL reduces its binding to pNoDGAT2B and elevates expression of the enzyme for TAG assembly preferably from C16:1. Moreover, the level of NobZIP77 transcript is downregulated under N− (Fig. 1A), which further facilitates expression of its downstream TAG-synthetic genes. Although additional mechanisms might be present, this model can explain how both BL exposure and NobZIP77 downregulation eventually lead to elevated expression of the enzyme that assemblies TAG preferably from C16:1.

A rational approach to elevate the peak productivity of oil via blue light

The above discoveries inspired us to devise a strategy for superior TAG production, by simultaneously tuning four key factors: light quality (WL/ BL), NobZIP77 expression (WT/NobZIP77ko-1), nitrogen supply (N+/N−), and time of TAG harvest (“Methods”). For the light-quality factor, WL and BL were adopted in wild-type cultivation. Compared to WL, (1) under BL and N+, TAG content increased by 33.7% at 24 h (Supplementary Fig. 20A). (2) under BL and N−, TAG content increased by 6.9–42.9% from 12 h to 168 h (the highest improvement at 72 h), and TAG yield increased by 3.6–48.3% from 6 h to 96 h (the highest improvement at 48 h; Fig. 5A). Meanwhile, microalgal growth rate remained unchanged under BL, versus WL (Supplementary Fig. 20A–C). Therefore, consistent with existing results (Fig. 3D), BL is a key factor in TAG improvement without inhibiting growth, especially under N−.

For the NobZIP77-expression factor, under WL and N+, compared to WT, TAG content of NobZIP77ko-1 increased by 72.5–225.7% from 24 h to 264 h (the highest improvement at 264 h; Supplementary Fig. 20B), and TAG yield of NobZIP77ko-1 increased by 60.9–184.5% from 24 h to 264 h (the highest improvement at 216 h) (Fig. 5B). Meanwhile, no slowdown in microalgal growth was apparent. Under N−, the TAG content, TAG yield, and growth of NobZIP77ko-1 were all equivalent to WT (Supplementary Fig. 20B). Therefore, knocking out NobZIP77 can greatly promote TAG synthesis yet without inhibiting growth (under N+), consistent with the above findings (Fig. 3F).

We next tested the link between BL and NobZIP77-knockout, by comparing NobZIP77ko-1 under BL (BL-KO) to WT under WL (WL-WT) (Supplementary Fig. 20C). Under N+, the TAG content of BL-KO increased by 83.6% and 98.4% at 24 h and 144 h, and the TAG yield increased by 73.4% and 50.5% at 144 h and 216 h. Under N−, the TAG content of BL-KO increased by 17.2–35.7% from 72 h to 120 h (the highest improvement at 72 h), and the TAG yield increased by 35.3% and 21.1% at 72 h and 96 h (Fig. 5C). Notably, the growth rate of BL-KO remained equivalent to WL-WT, under both N+ and N−. Moreover, versus BL-WT or WL-KO, the TAG content, the TAG yield, and growth rate of BL-KO exhibited no change (under N+ and N−). Since BL and NobZIP77-knockout can relieve the repressive effect of NobZIP77 and activate TAG synthesis, yet NobZIP77-knockout eliminates such an effect of BL (i.e., they are not additive) under N+, BL and NobZIP77 are likely orthogonal in this oleaginous signaling (Fig. 4D).

Finally, the observations that elevation of TAG by BL under N− yet by NobZIP77-knockout under N+ allowed us to propose a BL-induced oil-production (BLIO) strategy (Fig. 6A). Specifically, the NobZIP77ko-1 line was cultivated under WL for nine days (corresponding to peak yield of TAG under N+; Fig. 5B) and then with BL for another seven days (i.e., when N in medium is gradually naturally depleted). As a control, WT was cultivated for 16 days under WL (Fig. 6A). Microalgal growth rate for BLIO is equivalent to the control (Supplementary Fig. 21), yet the TAG yield is elevated in a rather consistent manner along the cultivation course (e.g., increases by 103.5% on Day 16; Fig. 6B). Moreover, the TAG productivity of BLIO doubles that of the control (on average 91.1% higher; Fig. 6C). Therefore, via both genetic and process engineering, we have established a rational approach that doubles the peak productivity of oil in N. oceanica.