Heterogeneous diacylglycerol acyltransferase expression enhances lipids and PUFA in Chlorella species

Algae have been explored for renewable energy, nutraceuticals, and value‐added products. However, low lipid yield is a significant impediment to its commercial viability. Genetic engineering can improve the fatty acid profile of algae without compromising its growth. This study introduced the diacylglycerol acyltransferase (BnDGAT) gene from Brassica napus into Chlorella sorokiniana‐I, a fast‐growing and thermotolerant natural strain isolated from wastewater, which increased its intracellular lipid accumulation. Hygromycin‐resistant cells were selected, and enhanced green florescence protein fluorescence was used to distinguish pure transgenic cell lines from mixed cultures. Compared to the wild type, BnDGAT expression in transgenic C. sorokiniana‐I caused a threefold increase in non‐polar lipid and a twofold increase in polyunsaturated fatty acids. Nile red staining reaffirmed the presence of higher intracellular lipid bodies in transgenic cells. There was a substantial alteration in the fatty acid profile of transgenic alga expressing BnDGAT. The non‐essential omega 9 (C18: 1) fatty acid decreased (5%–7% from 18%), while alpha‐linolenic acid, an essential omega 3 fatty acid (C18: 3), was increased (23%–24% from 11%). This study substantiates a valuable strategy for enhancing essential omega‐3 fatty acids and neutral lipids to improve its nutritional value for animal feed. The increased lipid productivity should reduce the cost of producing fatty acid methyl esters (FAME). Improved FAME quality should address the clouding issues in cold regions.


| INTRODUCTION
Microalgae are diverse photosynthetic organisms that thrive across many ecosystems. They are in the spotlight due to their use in nutraceuticals and other value-added products. Biofuels are being explored as a solution to the expected 40% rise in demand for fossil fuels by 2040 (Saad et al., 2019). Microalgae, the third-generation biofuel, circumvents problems of first-generation (biofuel from crop plants) and second-generation (biofuel from non-edible lignocellulosic biomass) feedstocks such as excessive use of land, water, and pesticides. Fourth-generation biofuels include genetically engineered microalgae for enhanced biomass, lipids, or high-value products. Because microalgae can grow in waste/brackish/marine water, they have distinct advantages over first-and second-generation biofuels, including no competition with food crops for cultivable area and fresh water. Microalgae have a faster growth rate and higher CO 2 mitigation capacity than land plants, resulting in higher biomass turnover. Moreover, microalgae farming requires far less area than plant agriculture (Aratboni et al., 2019). However, integrated approach is required to bring down the overall cost of biofuel production by improving the biomass, lipid productivity, and value-added by-products.
Being photosynthetic, microalgae can efficiently convert sunlight and atmospheric CO 2 into important organic compounds such as proteins, carbohydrates, and lipids. During photosynthesis, microalgal cells accumulate nonpolar lipids like triacylglycerol (TAG) intracellularly, serving as energy storage molecules. Like higher plants, algal fatty acid biosynthesis occurs in the chloroplast, producing 16-18C carbon products. The major product of fatty acid synthase (FAS) is palmitic acid, except for the elongation of palmitic acid and the desaturation of stearic acid, which take place in the chloroplast. Other changes (elongation, desaturation, hydroxylation, and epoxidation) occur mainly in the endoplasmic reticulum (Fatiha, 2020). The proportion of non-polar lipids and the polar membrane lipids varies from species to species, though the proportion of polyunsaturated fatty acids (PUFAs) is always more (Harwood, 2019). The PUFAs are essential for forming mitochondrial supercomplexes (Althoff et al., 2011), and treating Parkinson's and Alzheimer's diseases (Yates et al., 2014). Omega-6 linoleic acid (18:2n-6; linoleic acid [LA]) and omega-3 linolenic acid (18:3n-3; αlinolenic acid [ALA]) are essential for human nutrition and have to be supplied exclusively through diet, as all mammals lack the enzyme to introduce double bonds in fatty acids beyond carbons 9 and 10 (Berg et al., 2007). Because of its importance to human health, the animal feed industries are also interested in omega-3 fatty acid-rich feedstocks for producing enriched food for human consumption. The microalgae are the major source of LA and ALA along with chia, canola, flaxseed, and soybean.
DGAT catalyzes the final and committed step in the Kennedy pathway in plants and microalgae for TAG biosynthesis. Generally, there are three types of DGATs: DGAT1 and DGAT2 are membrane bound, while DGAT3 is a soluble acetyltransferase . Membranebound, non-homologous DGATs 1 and 2 are the major contributors of TAG biosynthesis in plant-developing seeds and microalgae. In plants, DGAT1 is responsible for oil accumulation in seeds like Brassica napus (Rahman et al., 2013) and safflower (Weselake et al., 1993). DGAT2 is important for the incorporation of unusual fatty acids in plants like Tung tree (Shockey et al., 2006), castor (Kroon et al., 2006), and ironweed (Li et al., 2010) and plays a minor role in oil accumulation in seeds. However, in microalgae, only one or two copies of DGAT1 and multiple copies of DGAT2 were observed to contribute to TAG production (Liu & Benning, 2013;Mao et al., 2019). The physiological roles of different types of DGAT enzymes remain unclear in microalgae (Xu et al., 2020). Our group has previously reported double the amount of lipid accumulation compared to WT on expression of DGAT from B. napus into the model species Chlamydomonas reinhardtii (Ahmad et al., 2015). This work has been further extended to C. sorokiniana-I, a potential candidate for biofuel and nutritional values.
Chlorella species have potential biotechnological applications in feed and food supplements due to their rapid growth and high content of carotenoids, lipid, and carbohydrates. It has a thick cell wall that help in protecting its nutrient value (Azaman et al., 2017). Chlorella sorokiniana strains are known for high growth, temperature tolerance (up to 42°C), wastewater remediation , and potential uses in the health industry (Azaman et al., 2017). The natural strain C. sorokiniana-I was isolated from a local urban wastewater pond, which is remarkably useful for the remediation of wastewater (Nawkarkar et al., 2019). In this study, we have genetically transformed C. sorokiniana-I with BnDGAT and enhanced green florescence protein (eGFP), which was instrumental in differentiating the genetically modified (GM) algal cells from wild type (WT) under a fluorescence microscope. Transgene integration was confirmed using polymerase chain reaction (PCR). Copy number was calculated using absolute quantification by real time polymerase chain reaction, and further validated by Southern blot. The protein expression of BnDGAT was confirmed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) gel and western blot analysis. The lipid content was enhanced to threefold higher in GM cells than WT. The fatty acid profile was significantly altered; especially ALA (an essential omega-3 fatty acid) was observed to double in the transgenic cells.

| Strain and growth conditions
The algal strain C. sorokiniana-I was isolated from Neela-Hauz Lake in New Delhi and identified using 18S and 16S rDNA genetic markers (Nawkarkar et al., 2019). Stocks were cryopreserved in liquid nitrogen. For experimental purposes, the strain was grown in tris-acetatephosphate (TAP) liquid medium as defined by Gorman and Levine (1965) or on the same TAP medium solidified with 1.2% agar. The culture was maintained at 25°C under solid-state lamps with a light intensity of 6000 lux (for 16:8 h light and dark) on solid plates. The cultures were cultivated using TAP medium in 250 mL conical flasks on an orbital shaker at 150 rpm for 15 days to calculate biomass, total lipid (TL), and fatty acid methyl ester (FAME).

| Construction of vector
The expression vector for algae pchlamy_1 (4283 bp) was procured from Life Technologies, Invitrogen, USA. The vector has an aph7 gene driven by βtubulin promoter that provides resistance against hygromycin. The DGAT gene (DNA sequence with optimized codons is given in Box S2) synthesized based on sequence from B. napus (GenBank accession no. AF155224.1), was placed under the fusion promoter Hsp70-RbcS2 for its expression. Also, eGFP (accession no. JN596101) was introduced under the 35SDE promoter as the reporter gene. The complete synthetic DNA cassette (DGAT-eGFP) containing DGAT-6XHIS-Tag-KDEL-NOS-PolyA-35Sde-eGFP was synthesized by Bio Basic, Inc. Canada. The BnDGAT gene had been fused with a KDEL sequence (Lys-Asp-Glu-Leu) that ensured the expressed protein remained in the ER membrane by targeting it to COPI-coated vesicles for retrograde transport back to the ER from Golgi (Newstead & Barr, 2020). The cassette was cloned into Xba1 and Not1 restriction sites in the MCS in the pChlamy_1 vector. The final vector was named "pAlgae-DGAT-eGFP" ( Figure S1a) and was transformed into Escherichia coli DH5α cells. Positive clones were selected on LB plates containing 100 μg/mL of ampicillin. The positive clones were maintained as glycerol stocks, and cultures were scaled up to obtain the plasmid in bulk.

| Antibiotic sensitivity assay and genetic transformation of C. sorokiniana-I
An antibiotic sensitivity assay was conducted to determine the lethal dose of hygromycin for C. sorokiniana-I prior to genetic transformation using the Gene Gun. About 1 × 10 6 cells were plated on solidified TAP agar plates containing varying concentrations of hygromycin. The concentration above the lethal dose was used to select positive transformed cells. The transformation was carried out using the Biolistics Gene Gun (PDS-1000/He™ System, BioRad). To make a thick lawn, cells were collected after 24-36 h of inoculation (0.4 OD 750 ), in 50 mL tube and centrifuged at 1137 g for 10 min. Cell pellets were spread on solid TAP agar plates. This lawn was subjected to bombardment with the gold microcarrier particles (Seashell Technology) coated with linearized plasmid DNA (digested with PvuI restriction enzyme; Figure S1b). Different parameters of the biolistic particle delivery system were optimized, such as rupture discs of 1100, 1350, and 1800 psi, and distance was varied from 6 to 12 cm at a vacuum of 26 inches of Hg (data not shown). The plates were kept in the dark overnight. The lawn was scraped off using a sterile spatula and re-suspended in sterile water. The suspended cells were plated on a selection medium containing 150 μg/mL hygromycin at a density of 1 × 10 6 cells. The selection plates were incubated in light:dark conditions (16:8 h) at 25°C for 2 weeks until the appearance of green colonies.

C. sorokiniana-I with eGFP and PCR
Transformed cells were observed for eGFP expression using a fluorescent microscope (Nikon TS Eclipse) and compared with the untransformed WT cells (Figure 1a; Sheen et al., 1995). The microalgal cells (1 mL) were washed twice with sterile water and re-suspended in 20-30 μL of deionized water. Ten microliters of the cells was mounted on glycerol and observed under the microscope at the blue filter (457, 477, and 488 nm) to check for e-GFP expression.
The eGFP-positive colonies on the selection plates were screened for the BnDGAT gene by PCR using genespecific primer pairs: F-DGAT/R-DGAT (Table S5). For PCR, the colonies were directly dissolved in 10 μL of TE buffer (Tris 100 mM, ethylenediamine tetraacetic acid [EDTA] 1 mm, pH 8), boiled for 12 minutes, and cooled (Wan et al., 2011). Five microliters of this colony (template) was added to the PCR mix containing 5-picomole of each primer, 10 mM-dNTP, and 1 unit of Taq polymerase (G-BioSciences). The PCR amplification was carried out for 35 cycles of denaturation for 30 s at 95°C, annealing for 30 s at 58°C, and extension for 1 min at 72°C, after an initial denaturation for 5 min at 95°C and final extension of amplicon for 10 min at 72°C. The PCR product of amplified DGAT was visualized on 1% agarose gel (Figure 1b). The positive colonies were inoculated into liquid TAP medium for scale up to 10 mL volume.

| Non-polar lipid estimation by Nile red cell assay and staining
Fresh algal culture (200 μL at OD 750 0.3-1.0) aliquots were added to a 96-well microtiter plate, followed by 50 μL of 15 μg/mL of Nile red in dimethyl sulfoxide (DMSO; Higgins et al., 2014). The working solution of 15 μg/mL of Nile red was prepared by diluting a μg/mL stock solution of Nile red in DMSO. The microtiter plate was incubated at room temperature in the dark for 10 min, and thereafter fluorescence was recorded at 530 nm (excitation) and 575 nm (emission) with a 570 nm cutoff using a Spectramax M3 multimode spectrophotometer (Molecular Devices). The fluorescence was obtained by subtracting the value of the unstained cell from that of stained cells and normalized by dividing with the OD 750 of culture ( Figure 1c).
To visualize the intracellular lipid bodies, Nile red (9-diethylamino-5H-benzo [alpha] phenoxazine-5-one) Maximum lipid accumulation was observed in the GM lines D1, L20 and M29 compared to WT and other GM algal cell lines. (d) The maximum non-polar lipid accumulation was observed as golden colored, intracellular globules in the GM lines D1, L20, and M29 compared to WT. was used to stain the algal cells. The microalgal cells (1 mL) were briefly centrifuged (1137 g for 3 min), and the resultant pellet was suspended in 1 mL of 20% DMSO. The mixture was agitated gently by vortex, briefly centrifuged (1137 g for 3 min), and the pellet was re-suspended in 1 mL of water. The Nile red stock solution (5 μL of 1 mg/mL), prepared in DMSO was added to algal cells and incubated in the dark for 10 min. The Nile red-stained cells were visualized under the fluorescent microscope (Nikon Model 50i) at 530 nm excitation and 575 emission wavelengths ( Figure 1d; Chen et al., 2009).

| Determination of copy number by real time quantification with real-time PCR and confirmation by Southern blot
The copy number of the integrated transgene was determined using absolute quantification with real-time PCR (qPCR, quantitative real time PCR). The qPCR was carried out using a set of primers and probes (IDT) specific for the transgene BnDGAT (Table S5). The standard curve was prepared using plasmid pAlgae-DGAT-eGFP. The plasmid was linearized using a "zero cutter" restriction enzyme HindIII for the BnDGAT gene. It is advised to linearize the plasmid DNA before qPCR to avoid overestimating copy numbers due to supercoiled confirmations (Shirima et al., 2017). Plasmid copy numbers were determined using the following equation: where 6.022 × 10 23 = Avogadro's constant. MW = molecular weight of the plasmid, which is calculated as: Using the above Equation (1), number of plasmid copies per μL was calculated, and serial dilutions were made to produce dilution series from 1 × 10 0 to 1 × 10 10 plasmid copies per μL.
TaqMan real-time qPCR assays were performed using BnDGAT-specific primers and probes (Adams et al., 2013). The qPCRs contained 1× PCR Master Mix (iTaq Universal Probe supermix, BioRad), 250 nM each forward and reverse primers, and 100 nM probe (Integrated DNA Technologies). A total qPCR mixture of 20 μL containing PCR grade water and 1 μL of linearized plasmid was used to prepare the BnDGAT standard curve. The qPCR was run on a BioRad real-time PCR instrument (CFX96 Touch Real-Time PCR Detection System, BioRad). Each reaction was run in triplicate, and the thermocycling profile was as follows: 5 min initial denaturation at 95°C and 40 cycles of 95°C for 40 s and 57°C for 40 s with fluorescent data collection during the 57°C step. Data acquisition was done by CFX MaestroTM software, BioRad (Figure 2a,b). A sample with water in place of the linearized plasmid was used as the no-template control with PCR. Genomic DNA from GM and WT algal samples was digested with enzymes XbaI and HindIII for 16 h and diluted to 20 ng concentrations. In each qPCR, 1 μL of these dilutions was used as a template.
Total genomic DNA from the GM lines of C. sorokiniana-I (D1, L20, and M29) as well as the WT was isolated by the cetyl trimethyl ammonium bromide method (Doyle & Doyle, 1987). For Southern blot analysis, probe labeling and hybridization were carried out using DIG high prime DNA labeling and detection starter kit I according to the manufacturer's instructions (Roche). The probe was prepared using PCR amplicon of ~900 bp of Aph7 (Figure 2c), amplified from the vector "pChlamy_1 BnDGAT-eGFP" using Aph7-specific primers (Table S5). The PCR product was purified using the GeneJet PCR purification Kit (Thermo Scientific), and subsequently labeled with DIGlabeling kit (Roche) according to the manufacturer's instructions. For random transgene integrations, 15 μg of genomic DNA from GM and WT algal lines was digested with XbaI and HindIII and separated by electrophoresis on 0.8% agarose gel for 5-6 h. The digested DNA was transferred to a positively charged nylon membrane (Roche) and hybridized with Aph7 probe at 52°C. DIG-labeled Aph7 probe was detected using NBT/BICP reagent after incubating with anti-digoxigenin-AP.

| Protein extraction and Western blot analysis
The transformed and non-transformed C. sorokiniana-I cultures, grown for 6-7 days, were concentrated to 15 OD 750 by centrifugation (1137 g, 10 min) and ground in liquid nitrogen. Crushed cells were suspended in 2 mL of Hurkman protein extraction buffer (0.7 M sucrose, 0.5 M Tris, 30 mM HCl, 50 mM EDTA, 0.1 M KCl 2 mM phenylmethylsulfonyl fluoride; Hurkman & Tanaka, 1986) and sonicated briefly (10 s on/ 10 s off for 10 min × 3) on ice. The mixture was centrifuged at 3032 g for 50 min at 4°C. The supernatant, containing soluble fraction of proteins, was collected in a separate tube. Fifty microliters of 8 M urea was added to the remaining pellet to obtain the membrane fraction of proteins. Both the soluble and membrane fractions were run on 10% SDS PAGE gel and stained with Coomassie Brilliant Blue G-250 (0.12% w/v) in O-phosphoric acid (10% v/v), ammonium sulphate (10% w/v), and methanol (20% v/v; Candiano et al., 2004). (1) Purified monoclonal anti-His antibodies, raised in mouse (IgG1, immunoglobulin G1) were used to detect the recombinant proteins carrying His-tags at the C-terminus of BnDGAT in the transformed algal cells. Thirty micrograms of the soluble and membrane proteins was resolved by 10% SDS-PAGE gel and blotted onto Hybond-C membrane (GE Healthcare). Membrane was hybridized with monoclonal anti-His epitope (Sigma) raised in mouse. The membrane and Anti-His antibodies at dilution 1:3000 was incubated for 1 h at room temperature. The blot was further exposed to an appropriate secondary antibody (at dilution of 1:3000) raised in goat (Sigma) coupled with horseradish peroxidase. To observe the signal (10×His) fused at C-terminus of the DGAT, blot was developed using DAB (3,3′-diaminobenzidine) and hydrogen peroxide (Figure 3a).

| Purification of BnDAGT2 protein by affinity chromatography
Partial purification of membrane-bound BnDGAT from C. sorokiniana-I was carried out by affinity chromatography using nickel-nitrilotriacetic acid (Ni-NTA) agarose (ThermoFisher Scientific). The standardized buffer used was urea buffer containing 8 M urea, 0.1 M NaH 2 PO 4 , 0.1 M tris (pH 8.0), and 300 mM sodium chloride (NaCl). About 15 OD 750 of GM and WT cells were collected on day 6 of the culture and harvested by centrifugation at 4000 rpm for 10 min. The pellet was crushed in liquid nitrogen and re-suspended in 2 mL urea buffer with 10 mM imidazole. The suspension was sonicated briefly (10 s on/15 s off for 10 min) on ice and centrifuged at 4548 g for 30 min at 15°C. The supernatant was collected in a fresh tube and the process was repeated with 2 mL of urea buffer and 10 mM imidazole. The collected supernatant was passed through a 0.4-μm syringe filter and mixed with 1 mL Ni-NTA Resin, pre-equilibrated with urea buffer. The resin/lysate mixture was incubated for 1 h on a rotatory shaker at room temperature. After incubation, the resin/lysate mixture was transferred to a gravity flow column and allowed to settle and the flow-through (FT) fraction was collected. The column was first washed with 4 mL (1 mL × 4) of wash buffer 1 (urea buffer + 20% ethanol + 20 mM imidazole) and then with 4 mL of wash buffer 2 (urea buffer + 20 mM imidazole). Finally, the protein was eluted in 2 mL (500 μL × 4) of elution buffer containing 250 mM imidazole in urea buffer. All the FT fractions were collected and visualized on 12.5% SDS PAGE. The eluted protein fractions from the GM algal cell lines and WT were concentrated using 30 kDa molecular weight cut-off spin columns (Vivaspin 500, GE Healthcare) and electrophoresed on 12.5% SDS-PAGE gel (Figure 3b). The protein bands were excised and digested by trypsin (Shevchenko et al., 2006). The resultant fragments were analyzed by liquid chromatography-mass F I G U R E 2 (a-c) Copy number determination of integrated BnDGAT gene in genetically modified (GM) Chlorella sorokiniana-I compared to the WT. (a, b) Transgenic cell lines D1, L20, and M29 of C. sorokiniana-I were studied using the quantitative real-time PCR. Using the standard curve, Cq values of WT and transgenic lines were plotted to obtain number of transgene copies per 20 ng of DNA. (c) Southern blot analysis of GM C. sorokiniana-I cells lines D1, L20, and M29 and WT genomic DNA. The aph6-probe (0.9 kb) was amplified using plasmid DNA "pChlamy_1 BnDGAT-eGFP" with gene-specific primers. Hybridization of aph6 probe was observed with transgenic genomic DNA of D1, L20, and M29. One major signal was recorded in lane M29, two in lanes D1 and L20, respectively, while no signal was observed in WT (untransformed algal cells). The high lipid producing GM cell line (M29) was reconfirmed by Southern blot, yielding a single hybridization band with aphA6 probe. No signal was recorded in WT. spectroscopy (LC-MS)/MS to match its identity with BnDGAT protein sequence.

| Transesterification of lipid into FAMEs
The lipid was extracted from dried algal biomass of both transformed and untransformed C. sorokiniana-I by Bligh and Dyer (1959) method, with modifications as follows. The dried algal biomass (100 mg) was ground in 2:1 chloroform/methanol, and then added to 0.9 mL of distilled water. After overnight shaking, an additional 1 mL of chloroform and 0.9 mL of distilled water were added to the mixture and shaken for about 4 h. The layers were allowed to separate. The bottom chloroform layer was aspirated and filtered through Whatman No. 1 filter paper into a pre-weighed glass vial. Chloroform was evaporated, and TL was estimated by subtracting the weight of the empty vial from that of the vial containing the lipid. The TAG was separated from TLs using thin-layer chromatography (TLC). The solvents used were hexane:diethyl ether:acetic acid in the ratio of 70:30:1 (Chungjatupornchai et al., 2019) and glycerol trioleate (SRL Chemicals) used as a standard reference. TAG was visualized after staining with iodine ( Figure S4).
The lipid, transesterified into FAME was analyzed by gas chromatography (GC). For transesterification, 10 mg lipid was mixed with 2 mL of hexane and 200 μL of 2 M methanolic KOH as a catalyst. The mixture was vigorously mixed for 5 min. The upper clear supernatant (hexane) part was collected for FAME analysis. Quantification of FAME was carried out using GC (Agilent GC) equipped with Omega Wax 250 column (30 m × 0.25 mm × 0.25 μm) and flame ionization detector. The operating conditions were as follows: split ratio 1:10, injection volume 1 μL, nitrogen carrier gas with constant linear velocity 33.9 cm/s, H 2 at 40 mL/min, air at 400 mL/min, makeup gas (nitrogen) at 30 mL/min; injector temperature of 270°C, detector temperature of 280°C, oven temperature started at 140°C for 5 min and increased at the rate of 4°C/min to 240°C, and hold time of 20 min at 240°C. Heptadecanoic acid (C17:0) was used as an internal standard. The fatty acids were identified by comparison of chromatogram with mass spectral library (NIST). The response factor was calculated using a known concentration of heptadecanoic acid methyl ester (C17 FAME) run externally with the samples (Capoun & Krykorkova, 2020). The FAME content was calculated as mg per gram of dry cell weight (DCW) as described (Breuer et al., 2013). Critical biodiesel properties such as cetane number (CN), saponification value (SV), iodine values (IV), higher heating values (HHV), kinematic viscosity (KV), and density of biodiesel were calculated from the FAME composition of WT and transgenic C. sorokiniana-I as per our previous study (Nawkarkar et al., 2019).

| Lab scale semi-continuous cultivation
Semi-continuous cultivation refers to a method of culturing where partial periodic harvesting is followed by immediate restoration of the culture volume (Benvenuti et al., 2016). Both the WT and GM line M29 of C. sorokiniana-I were grown in 2 L cylindrical reactor vessels, in half strength TAP medium, aerated by introducing sterile air from the bottom of the vessel. A culture volume of 1.5 L was maintained for 32 days, and periodic harvests were made at an interval of 2 days after 4 days of initial biomass accumulation; for the biomass and TL calculations. Half the culture volume (750 mL) was harvested, initially after 4 days and later every second day. The culture was topped up with the same volume of fresh medium. DCW and TL were expressed as the productivity (mg/L/day) of the harvested aliquots (Figure 4).

| Statistical analysis
All experiments were conducted in triplicates, and the data are reported as the mean ± standard deviation (±SD). The results were compared across treatments using twoway ANOVA (post hoc Tukey multiple comparisons test) with a significance level of 95% (p ≤ 0.05).

| Genetic transformation
The antibiotic hygromycin was found most effective at a concentration above 120 μg/mL ( Figure S2) on C.
sorokiniana-I when supplemented in the ½ strength TAP medium. The concentration just above the lethal dose was used for the selection of positive transformed cells. Primarily, transformed colonies were selected on TAP selection medium containing 150 μg/mL hygromycin, in 16 h light: 8 h dark conditions at 25°C for 2 weeks, until the appearance of green colonies. For secondary selection, the putatively transformed colonies were patched in grid form on the TAP selection medium and tested by PCR using gene-specific primers. The maximum number of stable GM colonies (6.7%-6.9%) was recovered at the following parameters: rupture disc of 1350 psi at a 9 and 12 cm distance between the algal lawn and microcarrier (Table S1).

sorokiniana-I
The GM cells express eGFP as indicated by green fluorescence at 488-nm wavelength. In contrast, untransformed cells showed red-colored auto-fluorescence due to chlorophyll at the same wavelength, distinguishing between WT and transgenic cell lines. The lustrous green fluorescent transgenic cell lines, D1, L20, M13, M29, R8, R12, and Q16, were segregated from WT cells (Figure 1a). With multiple subcultures on antibiotic selection medium, the eGFP fluorescence was also instrumental in establishing the pure transgenic lines. F I G U R E 4 Comparison of biomass productivities (DCW) of Chlorella sorokiniana-I WT and GM M29 strain in semi-continuous cultivation system. The difference in the biomass productivities of WT and GM was found negligible after regular harvesting the biomass at interval of 2 days.

| Transgene integration analysis by PCR
PCR confirmed that the transgenic cell lines of C. sorokiniana-I exhibiting eGFP fluorescence contained the gene of interest. The integrated transgene (BnDGAT) was amplified by PCR, yielding an amplicon of ~600 bp. All the GM lines (D1, L20, M13, M23, M29, Q19, Q24, R8, R12, and R22) emitting green fluorescence were observed positive for the integration of "pChlamy_1 BnDGAT-eGFP" cassette into the GM algal cells (Figure 1b). The PCR-positive cell lines were further scaled up and analyzed for lipid accumulation and BnDGAT expression in transformed cell lines.

| Analysis of non-polar lipid accumulation in GM lines of C. sorokiniana-I
Since the enzyme DGAT catalyzes the last and committed step in the Kennedy Pathway by transferring the acyl moiety from an acyl-CoA to diacylglycerol (DAG) forming triacyl glycerol (TAG), the expression of heterogeneous BnDGAT leads to higher accumulation of TAGs inside the cell. This intracellular TAG was estimated fluorometrically and visualized as lipid globules using the Nile red stain. Among different screened GM cell lines, three lines, namely D1, L20, and M29, have accumulated much higher lipids than WT and other GM lines (Figure 1c,d).

| Analysis of copy number in GM lines of C. sorokiniana-I by quantitative real-time PCR and Southern blot
The copy number of the integrated transgene in the GM lines of C. sorokiniana-I expressing BnDGAT (D1, L20, and M29) was determined by quantitative real-time PCR (qPCR). Based on the standard curve, Cq values of WT and GM lines (D1, L20, and M29) were plotted to obtain the number of transgene copies per 20 ng of DNA. These values were normalized by the number of genome (C. sorokiniana-I) present in 20 ng genomic DNA to give the absolute copy number of transgenes integrated per genome. As per the calculations, the GM M29 line was observed to have a single copy of transgene integration (Table S2).
The outcome of copy number of integrated transgenes in C. sorokiniana-I cell lines D1, L20, and M29 using qPCR was reaffirmed by Southern blot analysis, as mentioned in the Materials and Methods. One prominent band was detected in GM M29, two in L20 and D1, respectively (Figure 2c). The single transgene copy integration in M29 line was reaffirmed by Southern blot that yielded the same result as the qPCR (Figure 2c). There was no hybridization of aphA6 probe with WT genomic DNA (Figure 2c).

| Analysis of DGAT expression in transformed cell lines of C. sorokiniana-I
On the Western blot, there were feeble protein signals in the soluble fraction, while very strong protein signals were observed in the membrane fraction of GM lines D1, L20, and M29 (Figure 3a). The corresponding samples were stained with 0.25% Coomassie Blue R-250 (Figure 3a). However, BnDGAT signals were not evident in GM line D1 due to the sample's low total protein content.
After partial purification using Ni-NTA resin, the different elutes (Elute 2 & 3) from GM cell lines D1, L20, and M29 showed the presence of desired protein in the range of 50-70 kDa on 12.5% SDS-PAGE gels (Figure 3b). The eluted and trypsinized protein band from the M29 GM cell line was analyzed using LC-MS/MS, to confirm its identity. The M29 GM cell line matched with the amino acid residues of BnDGAT, yielding two long matching peptides of sizes 15 and 20 amino acids (Table S3). The total protein coverage was about 11.27%. The two peptide sequences matched perfectly with BnDGAT protein as shown in Box S1, using red and bold font.

| Lipid and FAME accumulation in transformed C. sorokiniana-I
TL content doubled in all three GM lines (D1, L20, and M29), whereas the FAME content was threefold higher in GM lines than in WT (Table 1). The TL includes polar (phospho-and glycolipids) and non-polar lipids (TAG). In the GM line, non-polar lipid accumulation has been T A B L E 1 Total lipid and fatty acid methyl esters (FAME/ Biodiesel) analysis in GM cell lines D1, L20, and M29 compared to wildtype (WT) cells of Chlorella sorokiniana-I using 100 mg dry cell weight (DCW) after 15 days of growth. more than WT due to BnDGAT expression. The threefold increase in FAME between GM and WT can be explained by higher non-polar lipid accumulation in GM than other polar lipids. TAG separated from TLs using TLC and determined qualitatively using iodine staining, was observed more than the WT ( Figure S4). The FAME produced from GM cell lines was observed to be rich in PUFA ( Figure S3). Particularly, 7,10,13-hexadecatrienoic acid methyl ester (C16: 3) was increased by 64% in D1, 74% in L20, and 43% in M29 GM lines compared to WT ( Table 2). The linolenic acid methyl ester (C18: 3), an essential fatty acid, increased by 109% in D1, 116% in L20, and 108% in M29 GM line, respectively, compared to WT (Table 2). However, there was about 57%-70% reduction in elaidic/oleic acid methyl ester (C18: 1) in all three GM cell lines compared to WT (Table 2). There was no significant alteration in the saturated fatty acid content of the transgenic strains compared to the WT. The quality of biodiesel produced from C. sorokiniana-I (WT and GM) was analyzed in compliance with the quality standards of the European (EN) and American Society for Testing and Materials (ASTM; Table S6). The physical properties of biodiesel, such as density, KV, CN, IV, SV, and HHV, were calculated based on the FAME content. CN, IV, HHVs, and density of biodiesel produced by WT and GM C. sorokiniana-I were found to be in accordance with the range specified by EN and ASTM. The KV of the FAME produced by the GM lines was observed to be higher than that of the WT and within the range specified by ASTM and EN.

| Comparison of growth and lipid accumulation in WT and GM C. sorokiniana-I on semi-continuous cultivation
The dry cell weight (DCW) was determined for every periodic harvest of WT and GM M29 line of C. sorokiniana-I cultures cultivated in the semi-continuous mode for 32 days. The biomass and lipid productivity were calculated as mg biomass or mg lipid produced per liter of medium per day (mg/L/day; Figure 4). The average biomass productivity calculated for WT C. sorokiniana-I was 549 mg/L/day, whereas for GM M29 it was 450 mg/L/day; which was 18% less than the WT. However, the average lipid productivity calculated for WT C. sorokiniana-I was 56.3 mg/L/day compared to GM M29 it was 98.9 mg/L/ day, which was 75.6% higher than the WT.

| DISCUSSION
Several genomics, transcriptomics, and proteomic studies have suggested modifying the Kennedy pathway for enhanced TAG production in photosynthetic organisms (Arora et al., 2018;Radakovits et al., 2010). The overexpression of the LPAAT gene in C. reinhardtii led to 20% increase in oil content (Yamaoka et al., 2016). Wang et al. (2018) reported 67.5% increase in lipid production in transgenic C. reinhardtii on the introduction of the LPAAT gene from B. napus and GPD1 gene from S. cerevisiae. The T A B L E 2 Relative percentage of fatty acid methyl ester (FAME) composition in GM D1, L20, and M29 compared with wildtype (WT) of Chlorella sorokiniana-I after 15 days of growth. There was significant increment in polyunsaturated fatty acids C16:3 and C18:3 in the GM cell line D1, L20, and M29 comparative to WT marked in bold. overexpression of heterogeneous (yeast) GPAT, LPAAT, and DGAT genes in Chlorella minutissima UTEX 2219 has doubled lipid content (Hsieh et al., 2012). In another study by Muñoz et al. (2019), overexpression of LPAAT, GPAT, and DGAT genes from Acutodesmus obliquus in Nanochloropsis oleoabundans has led to 1.3-and 1.4-fold increase in total fatty acid and triglycerols, respectively. In microalgae, lipid biosynthesis involves specific genes; their products are governed by rate-limiting steps and poorly understood feedback regulations. This notion was supported by the overexpression of three type-2 DGAT genes in C. reinhardtii, which neither caused surplus intracellular TAG accumulation nor altered the fatty acid profiles during nitrogen or sulfur stress (La Russa et al., 2012). Therefore, overexpressing a homologous gene for lipid biosynthesis seemed unlikely to increase the lipid yield in C. sorokiniana-I. However, such inhibition may not be effective if a heterogeneous gene is introduced into microalgae because of slight variation at the enzyme's active sites. Our results supported the aforementioned notion as BnDGAT expression led to twofold higher accumulation of TAG in the transformed C. reinhardtii (Ahmad et al., 2015). Similarly, the expression of heterologous type 2 DGATT of Chlamydomonas increased the lipid content in leaves of Arabidopsis thaliana (Sanjaya et al., 2013). In a mutant Saccharomyces cerevisiae that was defective in TAG biosynthesis, around ninefold lipid increment was observed after introducing Chlamydomonas DGTT2 (Hung et al., 2013). Also, complete lipid body formation restoration in a TAG-deficient mutant of S. cerevisiae was observed after introducing type 2 PtDGAT2B genes of diatom Phaeodactylum tricornutum (Gong et al., 2013).

Compounds
Suitable regulatory sequences were identified to enhance the expression of heterogeneous gene and choose a heterologous gene for lipid biosynthesis. The hybrid promoter HSP70A fused with RBCS2 has been reported to increase the expression of transgenes in C. reinhardtii when compared with HSP70A promoter alone (Schroda et al., 2000). GFP expressed threefold higher under the HSP70A-RBCS2 promoter (Wu et al., 2008). The hybrid promoter used in this study has resulted in significant expression of the BnDGAT gene and accumulated threefold higher lipid than WT (Figure 1c,d). Maximum lipid was accumulated in line M29, which displayed single copy transgene integration (Figure 2c).
Western blot analysis of transgenic lines (D1, L20, and M29) has confirmed that BnDGAT protein was localized in the membrane but not in the soluble fraction ( Figure 3a). It is also supported by previous reports that the DGAT protein belongs to a family of transmembrane (TM) proteins and is localized in the endoplasmic reticulum (Caldo et al., 2015). BnDGAT protein expressed in the transgenic C. sorokiniana-I was further affirmed by partially purifying the protein with affinity chromatography (Ni-NTA agarose) and analyzed by mass spectrometry (MS). The desired BnDGAT protein was digested with trypsin, and Mass fingerprinting using LC MS/MS performed its subsequent identification. A unique identification of purified protein requires only 10%-20% of the protein to be covered by tryptic digest (McMahon, 2005). These peptide sequences of digested protein matched (11%) with the amino acid residues of BnDGAT (see Box S1). Since the heterogeneously expressed BnDGAT is a membrane-bound protein, it was observed at ~51 kDa even though the molecular size of the desired protein band, calculated based on the amino acid sequence, is around 39.6 kDa. This may be because the membrane proteins show "gel shifts" on the SDS-PAGE gels, a common phenomenon for membrane-bound proteins. Peptide segments with abundant hydrophobic residues, as found in the helical TM regions of membrane proteins, are expected to be embedded within the SDS micelle interior, thereby affecting the protein mobility on the SDS-PAGE (Rath et al., 2009).
An alteration of fatty acid composition was observed with a considerable enhancement in the proportion of PUFAs in the GM lines (D1, L20, and M29) except C18:1, which was lower than WT. The ALA (18:3) content was increased twofold in GM lines than WT (Table 2). Such an alteration in FAME profile, that is an increase in relative abundance in linolenic acid (C18:3) has been reported earlier by Muñoz et al. (2019). Also, introduction of DGAT from Echium pitardii has led to a significant (8%-30% of total fatty acid) increase in the PUFA content in the transgenic Tetraselmis chui than the WT, which contains 5% PUFA (Úbeda-Mínguez et al., 2017). The overexpression of BnDGAT in C. reinhardtii has increased the PUFA about 12% while saturated fatty acids like palmitic and stearic (C16:0 and C18:0) were reduced by 7% (Ahmad et al., 2015). Also, in yeast transformed with PtDGAT2B of diatom showed alteration in TAG proportion of unsaturated C16 and C18 fatty acids (Gong et al., 2013). Muñoz et al. (2019) reported increase in C18:2 in the non-polar (TAG) fraction in the transformed lines of Neochloris oleoabundans. The DGAT1 from C. reinhardtii, when cloned into P. tricornutum, increased 16C and 18C saturated fatty acids (Li-Beisson et al., 2019). Total protein and carbohydrate content was estimated from the dry biomass upon completion of the growth cycle after 15 days (data not shown). No significant change was observed in the protein content of the WT and GM lines. However, the carbohydrate content was found to have decreased in the GM lines as compared to the WT strain (Table S4). The decrease in carbohydrate content was due to the diversion of carbon flux to lipid production in the transformed strain as observed in the diatom P. tricornutum upon introduction of DGAT by Zhang et al. (2021).
Microalgae strains with high lipid and biomass are need of the hour for an economically viable production of biodiesel and other value added products. In the reported literature, there is a conflicting property about algae like high biomass with low lipid accumulation or high lipid accumulation with low biomass (Shokravi et al., 2020). In our study, transgenic C. sorokiniana-I cell lines and WT has accumulated higher lipid under nitrogen and sulfur deprivation but with extreme loss of biomass and growth. Thereafter, cell cultures turned yellowish by the end of the growth cycle. Previous works on nitrogen starvation stress on transformed cells for lipid accumulation had shown the biomass loss in Botrycoccus species when exposed to nitrogendeficient conditions (Yeesang & Cheirsilp, 2011). There have been reports regarding decreased growth of Scenedesmus obliquus under nitrogen-deficient conditions (Gouveia & Oliveira, 2009) and decreased biomass in Nannochloropsis sp. under low nitrate concentration (Hu & Gao, 2006). Previous studies have also shown that genetic modifications of algae, especially overexpression of lipid pathway genes, have reduced growth in the transgenic lines (Muñoz et al., 2019). We have demonstrated that this problem can be circumvented by semicontinuous cultivation method (Figure 4). In this mode, a specific amount of culture is harvested periodically, and the remaining culture is topped up to its original volume and allowed to grow for a definite time interval. Since the culture is not completely harvested, it ensures a periodic and increasing amount of product yields than the batch cultures (Amini et al., 2020). In our study, average biomass productivity in GM line M29 was observed 18% less than WT, which is not a substantial loss compared to 75.6% increased lipid productivity in GM line M29 than WT. The semi-continuous mode of cultivation can also be translated into large-scale, outdoor culturing systems expecting similar results.
TAGs extracted from microalgal biomass are converted to FAME or biodiesel via chemical hydrolysis and transesterification. The degree of saturation/unsaturation of the fatty acids determines the quality of the FAME or biodiesel (Folayan et al., 2019). For instance, the density of biodiesel tends to increase with degree of unsaturation. In general, biodiesel has greater KV and heating values than petroleum fuels. CN, which is a measure of fuel's ignition quality, is marginally higher for biodiesel than petroleum fuels due to presence of long-chain hydrocarbon groups (Saxena et al., 2013). Due to its high PUFA content, biodiesel derived from algal oil is not appropriate for direct combustion in sensitive engines. However, they can be effectively blended with regular diesel at 5% (B5), 7% (B10), or 20% (B20). In addition, PUFA extraction by solvent extraction or hydrocracking of algal oil is considered viable options for making biodiesel with oxidation stability similar to biodiesel from oil crops (Sarpal et al., 2019).
PUFA is also emerging as a significant dietary supplement derived from microalgae in addition to biofuel utilization. Dietary ALA from microalgae has been perceived as an alternative source of omega-3 FA due to its role as a precursor for DHA (Charles et al., 2019). The transgenic C. sorokiniana-I lines accumulated higher ALA content than WT. Chlorella sorokiniana species has also been favored for dietary supplement for human beings (Barone et al., 2018); but due to genetic modifications, it may be considered unsuited for direct human consumption. However, extracted oil from this strain can be added to edible oils because of its high PUFA content.
In conclusion, the transgenic C. sorokiniana-I demonstrated three times enhanced non-polar lipid content after the expression heterogeneous DGAT gene, making it suitable as biodiesel feedstock. Growth of transgenic strain under semi-continuous mode has circumvented the problem of low biomass yield in batch culture, thereby generating good biomass containing high lipid. The transgenic C. sorokiniana-I has high essential fatty acid content that can be explored for nutraceutical purposes. Chlorella sorokiniana species are known for their adaptability to environmental conditions and dominance in the outdoor culture systems. Therefore, using GM algal species with DAGT can potentially reduce the commercial costs of algae-based products like biofuels or nutraceuticals.

AUTHOR CONTRIBUTIONS
Prachi Nawkarkar carried out all the major experiments of the present study. Vikas U. Kapase assisted in basic study and adaptation of strain in lab conditions. Sarika Chaudhary helped in the analysis of DGAT protein. Sachin Kajla (Tata Steel) actively participated in research discussions and data analysis. Shashi Kumar (ICGEB) conceived the research idea and supervised all experimentations in this study. Prachi Nawkarkar took the lead in writing the manuscript, edited by Shashi Kumar. All authors provided the feedback about the research, analysis, and manuscript preparation and writing.