A shift of carbon and nitrogen metabolism by N starvation. Growth maximized after 4 days of culture (Fig. 1a) as the medium nitrate concentration decreased to a minimum (Fig. 1b). The rate of photosynthetic oxygen evolution (Fig. 1c) and the activity of PSII (Fig. 1d,e) remained high at day 4 with an increase in sugar content up to 55% (dry weight) (Fig. 1f) dominated in glucose (46% dry weight) (Fig. 1g) and xylose (9% dry weight) (Fig. 1h), followed by 21% (dry weight) lipid accumulation after 7 days (Fig. 1i). The intensity of iodine-stained starchy carbohydrates increased 4 days after culture while the Nile Red fluorescence increased after 7 days (Fig. 1j), demonstrating that carbohydrate reserves accumulated in the form of starch and later lipid bodies had appeared.
Starch accumulation by N starvation was attributed to the stimulation of gluconeogenesis and carbohydrate reserves biosynthesis, revealed by the comparative analysis of RNA-seq transcriptome profiles between 2-day (less starch content and sufficient N) and 4-day (significant starch accumulation and initial N starvation) cultures and qPCR assay (Supplementary Tables 1–6). The results of GO (Supplementary Fig. 1) and KEGG analysis on transcripts with significant change (log2(FC) ≥ or ≤ 1, P < 0.05) demonstrated not only enhanced gluconeogenesis but also a concomitant decrease in glycolysis and TCA cycle after 4 days (Fig. 2a). As compared to the glucose kinase (GLK), phosphofructokinase (PFK), and pyruvate kinase (PK) for glycolysis, the gluconeogenesis pathway uses four unique enzymes, pyruvate carboxylase (PC), phosphoenolpyruvate carboxykinase (PEPCK), fructose-1,6-biphosphatase (FBP), and glucose-6-phosphatase (GP) (Fig. 2a). The transcript abundance of GLK and PFK remained unchanged (Supplementary Table 7a) and that of PK slightly increased (Fig. 2b), while those of PC (Fig. 2c), PEPCK (Fig. 2d), and FBP (Fig. 2e) showed a significant increase at day 4, followed by a drop to the level near the day 2 sample after 7 days. However, the enzymes commonly used for both glycolysis and gluconeogenesis pathways appeared a different expression pattern in transcription. The transcript abundances of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and enolase (ENO) increased and those of fructose-1,6-bisphosphatase (FBP), triosephosphate isomerase (TPI), phosphoglycerate kinase (PGK), phosphoglycerate mutase (GPM), and phosphoglucose isomerase (PGI) remained unchanged or slightly decreased without statistical difference after 4 days (Supplementary Table 7a). Meanwhile, the TCA cycle was downregulated reflected by a decrease in the transcript abundances of 2-oxoglutarate dehydrogenase (OGDH), malate dehydrogenase (MDH), citrate synthase (CS), aconitate hydratase (ACO), succinate CoA ligase (SCS), and succinate dehydrogenase (SDH) and constant expression of isocitrate dehydrogenase (IDH) after 4 days (Supplementary Table 7b). Taken together, enhanced gluconeogenesis for glucose synthesis occurred when N became limited. Then, the pathways from glucose towards sucrose/starch were also enhanced after 4 days (Fig. 2f). Although the expression of HK and glucose-1-P adenyltransferase (APG), which was responsible for glucose to α-D-glucose-6-P and α-D-glucose-1,6-P to ADP-Glucose, was downregulated after 4 days, the expression of GLK, phosphoglucomutase (PGM), and glucose-1-phosphate uridylyltransferase (UGP) remained constant (Supplementary Table 8). GLK was expressed in Chromochloris zofingiensis when glucose concentration was high12. Hence, instead of HK acting as a critical enzyme for glycolysis and sucrose synthesis in plants13, the constant expression of GLK (Fig. 2g) reflects the opening of C flow from glucose to sucrose/starch in Desmodesmus cells during initial N starvation. Next, the conversion of UDP-Glucose to sucrose/starch was promoted, and this was demonstrated by increased transcript abundances of sucrose synthase-like partial (SUS) (Fig. 2h), starch synthase (SS) (Fig. 2i), and starch-branching enzyme (SBE) (Fig. 2j) after 4 days.
Evidently, in coordination with reduced glycolysis/TCA, enhanced gluconeogenesis and sucrose/starch biosynthesis, C derived from photosynthetic carbohydrate substrates flows to storage carbohydrates in the form of starch during early N limitation.
Amino acid and protein contents also increased as initial N limitation (4 days) mainly due to enhanced N pool through the stimulation of N uptake and assimilation (Fig. 3a), and then decreased after 7 days (Supplementary Table 9). At day 4, the transcript abundances of the genes encoding nitrate/nitrile/ammonium uptake (high-affinity nitrate transporter, nitrate transporter (NRT1, NRT3, and NRT4), nitrite transporter (NIT1 and NIT2), ammonium transporter (AMT2, AMT3)) and primary N assimilation (nitrate reductase (NR2 and NR3), nitrite reductase (NiR1), ferredoxin-dependent glutamate synthase (GS) (Fig. 3b), glutamine synthetase (GOGAT) (Fig. 3c), and glutamine amidotransferase (GAT)). In contrast to the upregulation of the GS/GOGAT cycle that operates when ammonium is present in the growth medium at low levels, the expression of glutamate dehydrogenase (GDH) induced under high ammonium concentration14 decreased after 4 days (Fig. 3d). Thus, both high-affinity N uptake and GS/GOGAT-mediated ammonium assimilation accelerated when N was limited. Additionally, the upregulation of the urea cycle and polyamine biosynthesis, demonstrated by increased transcript abundances of arginase (ARG), ornithine carbomoytranslase (OTC) (Fig. 3f), argininosuccinate synthase (ASS), argininosuccinate lyase (ASL) (Fig. 3g), ornithine decarboxylase (ODC), and N-acetylglutamate synthase (AGS) after 4 days (Supplementary Table 10), indicate the involvement of enzymes in biosynthesis for enhanced amino acid concentrations. Overall, these results reflect an enlarged N pool for amino acid and protein synthesis during initial N starvation.
Here, we have an interesting finding in Desmodesmus sp. CNW-N that the contents of amino acids abundant in alanine and aspartate increased concomitantly as carbohydrate accumulated at the initial N starvation period, followed by a decline at day 7 together with increased fatty acid accumulation. In Desmodesmus sp. CNW-N, C was primarily stored as carbohydrates (50–70%), and next was amino acids/protein or lipid (15–20%) after long-term N starvation. A similar trend was found in Desmodesmus sp. and Scenedesmus obliquus upon exposure to different light intensities15. The accumulation of starch is an immediate response to stressful conditions in these microalgae, while lipid is the long-term energy substance. The preference to store carbon as starch at the start of N starvation may become energetically more favorable for synthesis compared with making TAG16. The synthesis of a TAG molecule (C55H98O6; C16:0, C18:1, and C18:3) requires 6.3 ATP and 2.9 NADPH but only 4.2 ATP and 2 NADPH for starch synthesis17. Therefore, starch accumulation is convenient during initial N starvation in Desmodesmus sp. CNW-N. The response was to channel more C to starch, as can be seen from synchronized inhibition of the glycolysis/TCA cycle and upregulation of both gluconeogenesis and starch biosynthesis via PC, PEPCK, and FBP for the formation of UDP-glucose as the building block for the SS-mediated synthesis of linear sugar polymers (amylose) and α-(1,6) branches in amylopectin through SBE. The formation of ADP-glucose catalyzed by AGP is a limiting step in starch biosynthesis18. Chlamydomonas reinhardtii mutant sta6, encoding AGP, cannot synthesize ADP-glucose, and starch does not accumulate19. Currently, when AGP is downregulated, and UGP expression remains constant, we can hypothesize that UDP rather than ADP acts as a carrier to transfer glucose for starch synthesis. On the verge of N deficiency, inorganic nitrogen transporters and primary N assimilation into glutamate and glutamine were stimulated and then re-distributed to other molecules via the amino-transferases system to synthesize other amino acids/proteins or other nitrogen-containing compounds to cope with N limitation.
Alanine catabolism for lipid biosynthesis after prolonged N starvation. The content of fatty acids increased from 5.8% on day 2 to around 20% after prolonged N starvation (7 days), and the lipid droplets emitting yellow fluorescence appeared in a small amount after 4 days and reached a significant level on day 7 (Fig. 2). Although starch and membrane can be re-directed to TAG biosynthesis20,21,22,23,24,25, our RNA-seq data found that the genes associated with these processes, for example, starch phosphorylases and phospholipases responsible for starch and phospholipid degradation, respectively, were not transcriptionally affected over the culture period (Supplementary Table 11a). Moreover, the transcript abundances of the enzymes functioning for lipid degradation, like N-acyl phosphatidylethanolamine phospholipase and lysophospholipase were not affected (Supplementary Table 11a).
Because amino acids like BCAAs are also the C source for lipid biosynthesis in microalgae9, it was speculated that amino acids may contribute to lipid biosynthesis in Desmodesmus sp. CNW-N. But, the gene encoding for BCKDH that is responsible for BCAAs catabolism to Acetyl-CoA was not detected in the RNA-seq database (Supplementary Table 2). However, current evidence suggests that aspartate and alanine under catabolic control for pyruvate generation and OAA were required to synthesize Acetyl-CoA (the starting material for fatty acid biosynthesis). The transcript abundances and enzyme activities of aspartate oxidase (AO) (Fig. 4a,g) and alanine aminotransferase (ALT) (Fig. 4c,g), as well as the transcript abundances of PEPCK (Fig. 2a, Supplementary Table 6) and pyruvate kinase (PK) (Fig. 4b) significantly increased 4 days after culture. The transcript abundance of pyruvate dehydrogenases (PDHs) that convert pyruvate to acetyl-CoA remained constant (Supplementary Table 10b). These results reflect a sequential conversion from aspartate and alanine to pyruvate to acetyl-CoA generation (Fig. 4k).
Next, acetyl-CoA carboxylase (ACC) catalyzed the conversion of acetyl-CoA to malonyl-CoA. Here, ACC’s activity and transcript abundance increased linearly 4 days after culture (Fig. 4d,h). The malonyl-CoA/ACP transacylase (MAT) was downregulated in the subsequent steps responsible for forming acyl-ACP after 4 days, followed by a recovery after 7 days (1.39±0.12 fold assayed by qPCR assay). Simultaneously, 3-oxoacyl-ACP synthase I/II (KASI/II) triggered the Claisen-type condensation reaction of a malonyl-acyl carrier protein (ACP) with acyl-ACP, and 3-oxoacyl-ACP synthase III that initiated this process by explicitly using acetyl-CoA over acyl-CoA was detected but a change in transcript abundance was not observed (Supplementary Table 2). Then, the sequential steps were catalyzed by 3-oxoacyl-ACP reductase (KAR), which had 8 genes, one upregulated (CNW_00005695|m.37349), one downregulated (CNW_00004947|m.31371), 3-hydroxy acyl-CoA dehydratase (HD) that was not detected, and enoyl-ACP reductase (ENR) that did not have a change in transcription level (Supplementary Table 11b).
Following fatty acid biosynthesis, TAG formation started with the lycerol-3-phosphate acyltransferase (GPAT)-mediated acylation on the sn-1 site of glycerol-3-phosphate (G3P) in the formation of lysophosphatidic acid (LPA), followed by the conversion of LAP to phosphatidic acid (PA) by lysophosphatidic acid acyltransferase (LPAT) and then the formation of diacylglycerol (DAG) through dephosphorylation of PA by phosphatidic acid phosphatase (PAP). Subsequently, the transfer of acyl group to DAG is catalyzed by diacylglycerol acyltransferase (DGAT)21. We found that glycerol-3-phosphate dehydrogenase (G3PDH1) was upregulated for G3P synthesis while GPAT was downregulated 4 days after culture but increased after 7 days (Supplementary Table 10b, 2.65±0.31 fold increment assayed by qPCR assay). Subsequently, the transcript abundances of oleoyl-acyl carrier protein thioesterase chloroplastic-like (FATA1 and FATA4) increased (Fig. 4e), while that of lysophosphatidate acyltransferase (LPAT) remained unchanged, and that of phosphatidic acid phosphatase (PAP1) responsible for the reversion of DAG to PA increased after 4 days and then recovered after 7 days (Supplementary Table 11b, 1.22±0.09 fold assayed by qPCR assay). At the last step, diacylglycerol acyltransferase/O-acyltransferase (DGAT1, DGAT2, and DGAT3) increased transcriptionally during culture (Supplementary Table 2 and DGAT1 in Fig. 4f) with an increase in DGAT activity at day 7 (Fig. 4i,j). Taken together, G3PDH, GPAT, FATA, and DGAT are responsible for enhanced TAG biosynthesis in the acyl-CoA-dependent pathway (the Kennedy pathway) in Desmodesmus sp. CNW-N after prolonged N starvation. The above results suggest that the pathway from aspartate and alanine to TAG can be stimulated after prolonged N starvation.
To verify the catabolism of alanine, the most abundant amino acid, to lipid accumulation as N starvation was prolonged, several ALT inhibitors were used. Treatment with amino-oxyacetate (AOA), an inhibitor of ALT activity26, at day 6 from 10, 30, to 60 µM decreased ALT activity (Fig. 5a) and increased alanine contents (Fig. 5b) along with a decline in fatty acid contents (Fig. 5c) in a dose-dependent manner; however, carbohydrate contents were not affected (Fig. 5d). A similar result was also observed by exposure to 25 or 50 µM L-cycloserine (CS), another ALT inhibitor (Fig. 5e–h).
For the carbohydrate-rich Desmodesmus, the improvement of carbohydrate production is conducive for enhancing low-cost algae-based bioethanol industrialization. According to the current metabolic analysis, the switch of the carbon skeleton after ammonia removal from alanine or aspartate to carbohydrates instead of fatty acids provides a potential route for improving carbohydrate accumulation. Because of the ACC induction after prolonged culture, the inhibition of C flux to fatty acid biosynthesis through downregulation of ACC expression was applied to test the conversion of the alanine C skeleton to carbohydrates. Treatment with clethodim ((RS)-2-[(E)-1-[(E)-3-chloroallyloxyimino] propyl]-5-[2-(ethylthio) propyl]-3-hydroxycyclohex-2-enone), a cyclohexanedione-based herbicide targeting ACC27, at a concentration of 10 or 30 mg⋅l− 1, 6 days after culture, inhibited an increase in ACC activity (Fig. 5i) and fatty acid contents (Fig. 5j). This was accompanied by carbohydrate accumulation up to 60.52% and 65.27%, respectively, which was higher than the 55.31% in clethodim-free control cells (Fig. 5k).
For several novel carbohydrate-rich microalgae isolated from Taiwan including Chlorococcum humicola TM-51, Desmodesmus sp. TM-30, Chlamydomonas sp. TM-1, Dunaliella sp. TM-Tainan-5, and Scenedesmus obliquus TM-12, dominated with alanine as the main amino acid (10–16% d. wt.) can accumulate a large amount of carbohydrates in the range from 36–51% d. wt. 4 days after culture near the N starvation period, reflected by the absence of nitrate in BG-11 medium (Fig. 5). The combined treatment of ALT and ACC inhibitors on other isolated species and some commercial microalgae (Tetraselmis sp. TM-8, Chlorella vulgaris, Nannochloropsis onceania CCMP1779 and N. oceanica IMET) whose total amino acid was not dominated in alanine did not show enhanced carbohydrate accumulation during N starvation in BG-11 medium or natural seawater containing BG-11 medium (data not shown). Following the classification of core Chlorophyta28, the isolated microalgae that are able to accumulate carbohydrate through redirecting the catabolism of alanine-to-lipid are the freshwater unicellular green algae belonging to Chlorophyceae but not for Trebouxiophyceae (Chlorella vulgaris), Prasinophyceae (Ostreococcus sp. TM-2), and Chlorodendrophyceae (Tetraselmis sp. TM-8), as well as Orchrophyta (Nannochloropsis spp.) (Fig. 5r). The metabolic networks in alanine synthesis and catabolism for fatty acid biosynthesis are restricted to Chlorophyceae. It needs more work, however.
To further clarify the contribution of alanine catabolism to TAG accumulation, the ALT-overexpressing (ALT-OE) lines were generated under the control of. The over-expression of ALT was driven by a Chlorella vulgaris N Deficiency Inducible (CvNDI1) promoter (Fig. 6a). We obtained two ALT-OE lines (ALT-OE-6, ALT-OE-17 (Fig. 6b), which showed a higher ALT transcript abundance and enzyme activity as compared with the wild-type (Fig. 6c). After 7 days of culture, there was a more significant decrease in alanine contents in ALT-OE-6 and ALT-OE-17 as compared to wild type and vector control (Fig. 6d) and meanwhile, an increase of fatty acid contents by 47–71% in ALT-OE lines (Fig. 6e) without further carbohydrate increment as compared to wild type and vector control (Fig. 6f). The accumulation of fatty acids in ALT-OE lines was inhibited in the presence of 30 mg⋅l− 1 clethodim (Fig. 6e) accompanied with a remarkable increase of carbohydrate content up to 73.15% and 75.32% in ALT-OE-6 and ALT-OE-17, respectively (Fig. 6f). It was similar to the recent study that total carbohydrate content of a CO2 tolerant Chlorella sp. AE10 can reach up to 77.6% d. wt. by two-stage process32. However, our culture system using the chemical treatment of a ALT-overexpressing microalga is a low-cost process.
Furthermore, microscopic observation showed that lipid accumulation in the form of TAG (Nile red staining) can be inhibited in the presence of clethodim accompanied with an increase in carbohydrate accumulation (Iodine staining) (Fig. 6g). The compositions of enhanced accumulated carbohydrate were dominant in glucose (91.35%) with xylose as a minor one (8.65%) (data not shown). The biomass of ALT-OE 17 was suitable for ethanol production with high conversion rate of approximately 98% (Fig. 6h).
ALT and ACC coordination of the switch of alanine catabolism to carbohydrate accumulation
In microalgae, several strategies for modifications made to metabolic pathways via genetic manipulation trigger an enhancement or an inhibition of targeted metabolite production. For example, lipid production can be enhanced by altering storage carbohydrate metabolism through starch degradation for providing the C skeleton for lipid synthesis is widely studied in microalgae, such as Chlorella sorokiniana29 and Chlamydomonas sp. JSC44. The starchless C. reinhardtii mutant sta6, which is defective in AGP, showed around 10-fold increase in TAG content compared with the wild type. Because the expression of genes responsible for carbohydrate degradation was not upregulated in Desmodesmus sp. CNW-N, the other routes are responsible for lipid synthesis following long-term N starvation16. Carbon and nitrogen assimilation and distribution pathways enhance lipid biosynthesis in Chlorella under N deficiency30. Our results showed that C was re-distributed from certain amino acids to lipid when N was insufficient for a longer time. The switch of the C skeleton for amino acids and proteins synthesis could serve as C and energy for TAG synthesis31. Catabolism of amino acids like BCAAs generates acetyl-CoA for fatty acid synthesis in P. tricornutum7, a TAG producer D. tertiolecta mutant G11_78, and C. reinhardtii9. In the present study, BCAA catabolism enzymes showed less expression or remained constant, and BCKDH was not discovered (Supplementary Table 2). A degradation of protein after N starvation reflected by a decrease in protein contents (Supplementary Table 9) and an increase of many protease gene expression (Supplementary Table 12), and meanwhile, AO, ALT, and PEPCK were induced with concomitantly enhanced G3PDH, GPAT, FATA, and DGAT, this suggests that alanine and aspartate contributed to TAG biosynthesis. The inhibition of TAG accumulation by ALT inhibitors, AOA and CS, and ALT mutants, and the enhancement of fatty acid biosynthesis by exogenously applied alanine provide support.
Because Desmodesmus ALT-OE lines without and with 30 mg⋅l− 1 clethodim treatment on day 6 showed significant carbohydrate accumulation after prolonged N starvation without remarkable growth inhibition (Table 1), they were applied to test whether they were suitable for bioethanol production as compared to wild type. Using Z. mobilis and other degrading enzymes, the ethanol conversion rate reached 97.62–99.31% for these algal cells (Table 1). Although biomass production was slightly reduced, the ethanol production for the ALT-OE-6, and ALT-OE-17 was 735.76, and 755.21 g⋅Kg− 1, in which it was higher than 645.76 and 658.97 g⋅Kg− 1 for the wild type and the wild type treated with clethodim, respectively (Table 1). In combination with the blockage of ACC-mediated lipid biosynthesis, the metabolic engineering Desmodesmus ALT-OE lines are a suitable feedstock for bioethanol production.
The inhibition of fatty acid biosynthesis by the ACC inhibitor, clethodim, improved N starvation-induced carbohydrate accumulation. Thus, C from alanine flowed to fatty acids but not carbohydrates during prolonged N starvation because of a pull of C flux to TAG caused by the elevated activity of enzymes involved in TAG biosynthesis. Through metabolic engineering of ALT together with the blockage of lipid biosynthesis via ACC inhibitor, the overexpression of ALT driven the a CvNDI1 promoter exhibits a more significant accumulation of fermentative carbohydrates (glucose as the abundant component) up to 75.32% d. wt. with 755.21 g⋅Kg− 1 ethanol production for ALT-OE-17 line. It is significantly higher than 520 g⋅Kg− 1 for Chlorococcum humicola33 and 235 g⋅Kg− 1 for C. reinhardtii UTEX 901.
The carbohydrate-enriched biomass of several microalgae with the treatment with ACC inhibition and alanine as abundant amino acid is a good feedstock for bioethanol production. For successful commercialization, fermentative carbohydrate hyper-accumulation under N limitation conditions can be achieved by metabolic engineering approaches with the aid of chemical-mediated inhibition of byproduct synthesis (lipid), such as genetic engineering on boosting carbon flux from alanine towards carbohydrates. The carbohydrate-rich biomass was further used as feedstock for bioethanol fermentation.