Introduction

Microalgae have emerged as one of the most promising sources of biofuel feedstock, owing mainly to their ease of cultivation, their ability to accumulate high lipid content, and their reduced land area requirement compared with plant-based fuels. Several strains have been found to accumulate neutral lipids, in the form of triacylglycerol (TAG), at up to 50% of dry cell weight [14]. TAG could be extracted from the cells for subsequent production of biodiesel via a transesterification reaction. However, at present, the algal biofuel is still a long way from commercialization, largely due to its high production cost. To help make it possible, a large number of studies have been performed on various aspects of microalgal biodiesel production in recent years [25]. These studies have collectively led to an immense improvement in the cultivation, harvesting, dewatering, and lipid extraction processes [5]. However, there have only been a few reported achievements in the genetic improvement of algal lipid content.

Although many natural oleaginous strains of microalgae have been discovered, most of the strains with exceptional lipid content grow slowly and/or are difficult to cultivate under outdoor conditions, resulting in low biomass and lipid productivity. Therefore, improving the lipid content of robust strains and/or strains with cultivation technology already in place remains a favorable option. The genetic improvement of microalgal strains for biodiesel production has been carried out using several approaches. However, these attempts have resulted in limited success and are restricted to a few strains that are amenable to genetic transformation. For example, the straightforward metabolic engineering approach to induce the overexpression of the ACCase gene (encoding acetyl-CoA carboxylase, which catalyzes the first committed step of fatty acid synthesis) could not sufficiently increase lipid production, probably due to the complex regulation of the pathway [12]. The overexpression of the genes encoding different isoforms of acyl-CoA (diacylglycerol acyltransferase, which catalyzes the last step of TAG formation) did not lead to higher TAG production [17]. On the contrary, genetic modifications designed to inhibit pathways that compete with lipid biosynthesis appear to have met with the most success. An RNA interference approach designed to silence the gene that encodes citrate synthase, which mediates the entrance of acetyl-CoA into the tricarboxylic acid cycle, led to a Chlamydomonas reinhardtii strain that showed a near 2-fold increase in its TAG content [11]. Targeting TAG catabolism by down-regulating a lipase, via antisense RNA in Thalassiosira pseudonana, also led to a higher lipid content with no growth compromise [37].

Starch and TAG are the main storage compounds of plants and algae. Since the synthesis of starch and that of lipids share common precursors, blocking starch synthesis has been hypothesized as a way to improve TAG yield. Various starch-deficient strains of Chlamydomonas reinhardtii have been isolated or constructed to carry a mutation in one of the starch biosynthesis genes [1,26,41]. Many of these mutants were shown to accumulate a several-fold increase in TAG content, although it remains unclear whether the TAG overaccumulation in these mutants could be attributed to the starch synthesis deficiency [32]. Several starchless mutants of Scenedesmus obliquus were isolated [10], and the TAG content was significantly enhanced, from 45% per dry weight in the wild type to 57% in the most promising strain [6]. The screen used for isolating randomly mutagenized microalgal strains that are deficient in starch accumulation is straightforward and well described, and does not involve genetic transformation, provided that the algal genome is haploid (i.e., not genetically redundant). However, this approach has not yet been commonly applied in order to improve the lipid content of algal strains that are sufficiently robust for large-scale commercial cultivation.

Many algal species in the genus Dunaliella are common in marine habitats, with some exhibiting tolerance to high salinity and extreme environments. The large-scale commercial cultivation of D. salina is already in operation to produce natural β-carotene [28,33]. In addition to their rapid growth rate, the ability of Dunaliella to thrive in high-salinity culture provides a selective advantage, as high-salt concentration reduces the likelihood of contamination [30]. The present study focuses on D. tertiolecta, which is known as a potential candidate for biodiesel feedstock, due to its ability to grow rapidly in various environments, along with its high lipid accumulation [34,35]. Transcriptome data for D. tertiolecta under nitrogen-starvation conditions are also available, to facilitate metabolic engineering [31]. To further improve its TAG content, we explored the consequence of starch synthesis inhibition via random mutagenesis. Starch-deficient mutants were screened from a population of UV-mutagenized D. tertiolecta strains and characterized by their growth and biochemical content.

Materials and Methods

Strain and Growth Conditions

The D. tertiolecta strain UTEX LB999 (UTEX Culture Collection of Algae, University of Texas at Austin) culture was routinely m aintained on a h ypersaline m edium [36]. U nless otherwise indicated, cultures were grown at 25℃ under continuous illumination of approximately 50 μmol photons m-2 s-1 cool-white fluorescent light. The medium without nitrogen was prepared by substituting KNO3 with KCl. Liquid cultures were aerated by bubbling with normal air at an aeration rate of 1 vvm. For phenotypic testing on plates under high light conditions and in the presence of chemicals, log-phase cells were collected by centrifugation and the cell density was adjusted as indicated. Three microliter was spotted onto plates and the plates were kept at the indicated conditions for 10 days.

Mutagenesis and Mutant Isolation

For UV mutagenesis, D. tertiolecta cells at a density of 5 × 106 cells/ml in a glass petri dish were kept in the dark for at least 2 h, before being exposed to UV irradiation using a UV transilluminator (Model M-26; Upland, USA) for 22 min, to obtain a 1% survival rate. UV-irradiated cells were immediately wrapped in foil and allowed to recover overnight in the dark to prevent light-induced DNA repair. Cells were then spread onto agar plates and grown at 50 μmol photons m-2 s-1. Single colonies were transferred to fresh agar plates. After approximately 2 weeks, cells were transferred to agar medium without nitrogen, using a replicator. After approximately 3 weeks, when cells started to turn yellow, they were stained with iodine vapor. The C. reinhardtii sta1-1 mutant [1] growing on TAP-N plates was patched onto each plate prior to iodine staining, as a control. Cell patches that exhibited a substantially lighter color than that of the wild-type control on each plate were selected for further analysis.

Starch, Lipid, and Triacylglycerol Quantification

For preliminary characterization of the isolated mutants, mid-log phase cells in liquid medium were collected, washed once with medium without nitrogen, and inoculated into fresh medium with or without nitrogen for an additional 2 days. The cells were then harvested, kept at -80℃, and lyophilized at -50℃. Twenty to thirty milligrams of each sample was sonicated for 15 min and used for starch analysis using the Total Starch (AA/AMG) assay kit from Megazyme (Ireland), according to the manufacturer’s protocols for starch samples that also contain ᴅ-glucose.

Similar amounts of the lyophilized cells were also used for lipid extraction in chloroform:methanol (1:2 (v/v)), according to the protocol from Bligh and Dryer [2]. Total lipid was determined on a pre-weighed aluminum foil. Two-hundred micrograms of total lipid was spotted onto a 20 × 20 cm silica gel 60G F254 TLC plate (Merck, Germany). The plates were developed in hexane:diethyl ether:acetic acid (70:30:1) for 45 min and stained with iodine vapor. TAG bands were identified using standard soy oil and quantified by a densitometry method using the ImageJ program [29], in comparison with a standard curve generated from varied soy oil concentrations.

For gas chromatography (GC) analysis of the lipids produced by the sds1, sds3, and sds4 mutants, the algal strains were inoculated in 500 ml Erlenmeyer flasks containing 450 ml of liquid medium at the initial optical density at 680 nm (OD680) of 0.1 and maintained under continuous illumination at 700 μmol photons m-2 s-1. After 5 days, the cells were collected, washed once with medium without nitrogen, and transferred into fresh medium without nitrogen for an additional 2 days. The final cell density was determined by manual counting under a hemocytometer. Subsequently, the cultures were harvested by centrifugation, washed once with 1.5 M NH4HCO3, and lyophilized. The dry biomass yield was measured gravimetrically, and 40 mg of each sample was used for lipid extraction as described above. One half of the crude lipid extract was weighed to determine the total lipid yield, whereas the other half was quantified as fatty acid methyl esters (FAMEs) following a published protocol with modification [8]. First, 0.1 mg of glyceryl triheptadecanoate (Catalog No. T2151; Sigma) was added as an internal standard. Next, for purification of the TAG constituents, the crude extract was dissolved in 400 μl of chloroform, and one half of this amount was loaded into a VertiPak Si solid-phase-extraction (SPE) cartridge containing 300 mg of silica (Vertical Chromatography, Thailand) and pre-washed with 4 ml of hexane. The TAG-enriched neutral lipids were eluted by 2 ml of hexane:diethyl ether (83:17 (v/v)). The purity of the eluate was assessed by TLC in comparison with the crude lipid extract, using a two-phase mobile solvent system consisting of hexane:diethyl ether:acetic acid (70:30:1 (v/v/v)) followed by chloroform:methanol:water (65:25:4 (v/v/v)). After solvent evaporation, the purified and non-purified lipids were transesterified using 2 ml of 5% sulfuric acid in methanol under 90℃ for 120 min. After addition of 2 ml of hexane containing 0.01% (w/v) butylated hydroxytoluene, the reaction was washed with deionized water, and trace water was removed by anhydrous Na2SO4. For GC analysis using Agilent 6890 N (Agilent Technologies, USA) coupled with the Agilent 5973 Network Mass Selective Detector (Agilent Technologies), 1 μl of the derivatized FAMEs in hexane was injected into a capillary HP-Innowax column (Agilent Technologies; 30 m × 0.25 mm × 0.25 μm). The oven temperature was programed from 45℃ (1 min hold) to 180℃ at a rate of 50℃/min and increased to 260℃ at a rate of 2℃/min. Helium was used as a carrier gas with a split ratio of 1:1. The corresponding amount of each FAME was calculated by comparison of the peak area with that of methyl heptadecanoate.

Analysis of Growth and Photosynthetic Oxygen Evolution Rate

Inoculum cells were cultured in 5 ml of liquid medium for 2 days and then transferred to 50 ml of liquid medium to give an initial density at OD680 of 0.05. The cultures were continuously illuminated at 700 μmol photons m-2 s-1, and the culture density was monitored over a course of 6 days. For oxygen evolution measurements, cells were grown at 50 μmol photons m-2 s-1 to mid-log phase and the density was adjusted to OD680 of 0.1. Cells were dark-adapted for 15 min prior to measurements with a polarographic Clark-type oxygen electrode, OX1LP Dissolved Oxygen Package (Qubit Systems, Canada), at 23℃.

Pigment Quantification

Cells were grown in liquid medium at 700 μmol photons m-2 s-1 to a density of 1–2 × 106 cells/ml. Cells were collected, washed once with medium without nitrogen, and inoculated into fresh medium—with or without nitrogen—for an additional 2 days. Four milliliters of culture was harvested and used for pigment extraction in 80% acetone, and the analysis of carotenoid content was performed as previously described [24].

RNA Extraction and Real-Time RT-PCR Analysis

Cells were grown in liquid medium to mid-log phase before being collected, washed once with medium without nitrogen, and inoculated into fresh medium—with or without nitrogen—at a density of 2 × 106 cells/ml for an additional 2 days. Total RNA was extracted using the GENEzol TriRNA Pure Kit (Geneaid, Taiwan) and the concentration was determined spectrophotometrically. Genomic DNA was eliminated by treatment with RQ1 RNase-free DNase (Promega, USA) according to the manufacturer’s protocol. Total RNA (1.5 μg) was used for cDNA synthesis using MMuLv reverse transcriptase (Biotechrabbit, Germany). Specific primers targeting genes involved in starch synthesis and metabolism were designed based on RNA-seq and annotation data [31]. The primer sequences and PCR product sizes are shown in Table 1.

Table 1.List of primers used in this study.

Real-time PCR was performed on a Mastercycler ep realplex 4 (Eppendorf, Germany) using QPCR Green Master Mix (Biotechrabbit, Germany). Each reaction contained 30 ng of total RNA, 5 pmol of each primer, and 10 μl of QPCR Green Master Mix, in a final volume of 20 μl. The PCR cycle was 94℃ for 5 min, followed by 40 cycles of 94℃ for 30 sec, 69℃ for 30 sec, and 72℃ for 30 sec. For each sample, the PCR was carried out in triplicates. A melting curve program was employed to confirm the presence of only one specific PCR product. Relative gene expression was calculated by the 2-ΔΔCT method using RACK1 as a reference gene.

Results

Screening for Starch-Deficient Mutants

Four putative starch-deficient strains (sds) were identified from a screen of 20,016 UV-mutagenized D. tertiolecta strains. Whereas the parental wild-type strain turned dark purple after iodine-vapor staining, the sds1 and sds3 mutants appeared as light brown to purple and the sds2 and sds4 mutants remained yellowish (Fig. 1A). To confirm that the mutants had a lower starch accumulation, their starch content was quantified in cells grown in nitrogen-replete (N-replete) and nitrogen-depleted (N-depleted) liquid media. In the N-replete medium, there was no significant difference in the starch contents between the mutants and the wild type, as all strains showed around 2-6% starch per dry weight (Fig. 1B). In the N-depleted medium, however, the starch content in the wild type increased to 21% per dry weight, whereas this number ranged from 6% to 16% in the mutants: a significantly lower starch content. The starch contents also correlated well with the color intensity after iodine staining, as the sds2 and sds4 mutants exhibited the lowest starch content: 8% and 6%, respectively. These results confirm that these D. tertiolecta strains were deficient in starch accumulation and the iodine-staining method could be used in rapid screening for such mutants.

Fig. 1.Mutant screening and starch measurement. (A) Appearance of cells after iodine staining. Different densities of cell cultures were spotted onto agar medium without nitrogen for 2 weeks before staining. (B) Starch content of each strain. Values are from three biological experiments. Error bars are standard deviations. An asterisk (*) indicates a significant difference calculated by one-way ANOVA (p < 0.05) between the mutant and the wild type. Light bars, + N; dark bars, - N.

Determination of Total Lipid, TAG, and Carotenoid Contents

To determine whether the sds mutants accumulated a higher amount of lipid, the total lipid content was quantified gravimetrically in cells grown under the N-replete and N-depleted conditions. In general, the total lipid contents of all strains were slightly lower when grown in the N-depleted medium than in the N-replete medium (Fig. 2). Under both conditions, however, the lipid contents of the mutants were not significantly different from those of the wild type, except sds3 under N-replete condition.

Fig. 2.Lipid content of wild-type and mutant D. tertiolecta strains under N-replete and N-depleted conditions. Values are from three biological replicates. Error bars are standard deviations. The asterisk (*) indicates a significant difference calculated by one-way ANOVA (p < 0.05) between the mutant and the wild type. Light bars, + N; dark bars, - N.

Because both starch and TAG serve as the carbon and energy storage compounds in algae, and their biosynthesis pathways share common precursors, the blockade of starch biosynthesis may have a more prominent effect on the TAG content than it does on other classes of lipid. To test this possibility, equal amounts of the lipid extracts were separated in TLC to determine the amount of TAG. From the TLC plate, it was clear that the TAG constituents were enriched to a much greater extent in the lipid extracts from the sds1, sds3, and sds4 mutants under the N-depleted condition, compared with the wild type (Fig. 3A). However, there was no detectable TAG in the lipid extract from sds2. Quantification of the TAG spots confirmed that the three mutants had a significantly improved TAG content, at an increase of approximately 1.5-2 folds compared with the wild type (Fig. 3B).

Fig. 3.TAG content of wild-type and mutant D. tertiolecta strains under N-replete and N-depleted conditions. (A) Thin-layer chromatography (TLC) showing TAG from cells grown under N-replete and N-depleted conditions. (B) TAG content quantified from the TLC chromatogram. Values are from three biological replicates. Error bars are standard deviations. An asterisk (*) indicates a significant difference calculated by one-way ANOVA (p < 0.05) between the mutant and the wild type.

Carotenoids are high-value products of microalgae that have a close relationship with TAG formation [18]. The synthesis of TAG and that of carotenoids are coupled, since carotenoids are esterified with TAG and sequestered in lipid bodies, which are distributed to the cell periphery to protect cells from oxidative damage caused by excess light [14]. In D. bardawil, which overproduces β-carotene under stress conditions, the inhibition of lipid synthesis also blocked the overproduction of β-carotene, indicating that the lipid droplet is used as a sink for carotenoids [27]. Furthermore, the starchless mutant of S. obliquus has also been suggested to contain a higher carotenoid level than the wild type in nitrogen-depleted conditions [6]. Therefore, it might be possible that the high TAG content in starchless mutants may also lead to carotenoid overaccumulation. To investigate the effect of high TAG content on carotenoid level, cells grown at 700 μmol photons m-2 s-1 under both N-replete and 2 days of N-depleted conditions were used for carotenoid measurements. The results showed that all starch-deficient strains exhibited similar levels of carotenoids as the wild type (Fig. 4). Moreover, nitrogen depletion did not cause a significant change in total carotenoid level, except in sds4 whose carotenoid level slightly decreased.

Fig. 4.Total carotenoid content of wild-type and mutant D. tertiolecta strains under N-replete and N-depleted conditions. Values are from three biological replicates. Error bars are standard deviations. The asterisk (*) indicates a significant difference calculated by one-way ANOVA (p < 0.05) between the mutant and the wild type. Light bars, + N; dark bars, - N.

Growth and Photosynthesis

Some of the algal starch-deficient strains, such as the C. reinhardtii sta6 mutant, are reported to have growth impairments and an increased susceptibility to photoinhibition [22,23]. Reduced growth and photosynthesis efficiency are important factors to consider in commercial outdoor cultivation. To evaluate whether the mutations in these D. tertiolecta starch-deficient mutants might lead to the impairment of growth and photosynthetic characteristics, growth performance was monitored over the course of 6-days’ cultivation under moderate light intensity, at 700 μmol photons m-2 s-1. Because the sds2 mutant did not show an improved TAG content, it was not included in this and further studies. On the first day, sds1 and sds3 exhibited slightly slower growth when compared with the wild type (Fig. 5A). From the second day onward, only sds3 showed a lower biomass compared with the other strains. Nevertheless, all strains entered a stationary phase with no significant difference among them, suggesting that the mutations did not cause significant aberrations to the growth of these algal strains in this condition. Additionally, the photosynthetic performance—as determined by light-saturation curves of photosynthesis—followed a similar trend to the growth kinetics (Fig. 5B). Although the mutants seemed to have lower average values for each data point, there were no statistically significant differences between any of the strains at any of the tested light intensities.

Fig. 5.Growth under 700 μmol photons m-2 s-1 (A) and oxygen evolution rates (B) of wild-type and mutant D. tertiolecta strains. Values are from three biological replicates. Error bars in (A) are standard deviations. Error bars in (B) are standard errors. The plus sign (+) and an asterisk (*) indicate a significant difference calculated by one-way ANOVA (p < 0.05) between sds1 and the wild type, and between sds3 and the wild type, respectively. Diamonds, wild type; squares, sds1; triangles: sds3; circles: sds4.

Under adverse conditions, both starch and TAG are believed to accumulate as sinks for excess electrons, which could lead to harmful ROS formation. For example, the lack of starch in the C. reinhardtii sta6 mutant resulted in a higher sensitivity to photoinhibition [16]. Similarly, the pgd1 mutant of C. reinhardtii, which failed to accumulate the normal amount of TAG, also accumulated more TBARS than the wild type under nitrogen-starvation conditions, implying a higher amount of ROS was produced [21]. In outdoor cultivation, oxidative stress is unavoidable owing to uncontrolled light intensity. Nitrogen starvation has also been shown to induce oxidative stress upon illumination [42]. Therefore, to examine the effect of reduced starch synthesis on growth under oxidative stress conditions, cells were subjected to high light intensity at 1,200 μmol photons m-2 s-1 or to low light intensity at 50 μmol photons m-2 s-1, in the presence of 5 mM H2O2 or the presence of 5 μM Rose Bengal (a singlet oxygen generator). High light intensity did promote the growth of all strains, except sds3, which proved very sensitive to high light, as it failed to grow at a lower cell density (Fig. 6). Under the presence of H2O2, sds4 seemed to grow slightly slower than the other strains. On the other hand, all stains grew as well as the wild type in the presence of Rose Bengal.

Fig. 6.Growth phenotype of wild type and mutant D. tertiolecta strains under high light (1,200 μmol photons m-2 s-1), 5 mM H2O2, and 5 μM Rose Bengal.

Expression Profile of Starch Synthesis Genes

To evaluate whether the mutations in the sds mutants affect the expression of genes involved in starch biosynthesis and metabolism, quantitative RT-PCR was employed to monitor the transcript abundances of the genes involved. Genes encoding phophoglucomutase, small and large subunits of ADPglucose pyrophophorylase, several starch synthases, branching enzymes, isoamylase, and starch phosphorylase were identified in the D. tertiolecta RNA-seq and annotation data [31]. Gene expression profiling of all genes would provide a better understanding of the nature of these mutants. Nevertheless, partly due to short sequence length and the presence of multiple members in each gene family, no suitable primers for RT-PCR analysis of some genes were obtained despite several attempts. In the end, five genes were successfully analyzed by quantitative RT-PCR (Figs. 7A and 7B). The steps where these genes may be involved in the starch biosynthesis and metabolism are shown in Fig. 7C. Generally, expression levels of the genes tested were either unchanged or downregulated in the mutants compared with the wild type. The sds1 mutant showed a reduced level of transcripts encoding starch phosphorylase and starch synthase in N-replete and N-depleted condition, respectively. Similarly, sds3 also exhibited a decrease in expression of the gene encoding starch phophophorylase in N-replete cultivation. Under nitrogen starvation, however, this mutant had a reduced abundance of transcripts encoding the large subunit of ADPglucose pyrophophorylase and a branching enzyme. In both growth conditions, sds4 exhibited differential expression of transcripts encoding the large subunit of ADPglucose pyrophophorylase, starch synthase, and a branching enzyme. There was no differential expression detected in one of the branching enzymes (BE-1) in any of the mutants compared with that of the wild type both in N-replete and N-depleted conditions.

Fig. 7.Relative transcript levels of genes involved in starch biosynthesis and metabolism in D. tertiolecta cells. Under N-replete condition (A) and N-depleted condition (B). Error bars are standard errors (n = 3). An asterisk (*) indicates a significant difference calculated by one-tailed t-test (p < 0.05) between the mutant and the wild type. AGPase-LS, ADPglucose pyrophosphorylase-large subunit; SS, soluble starch synthase; BE-1, branching enzyme-1; BE-2, branching enzyme-2; SP, starch phosphorylase. (C) Simplified pathway of starch synthesis. Glucose-6-P, glucose-6-phosphate; Glucose-1-P, glucose-1-phosphate; ADP-glucose, adenosine diphosphase-glucose; PGM, phosphoglucomutase; AGPase, ADP-glucose pyrophosphorylase; SS, starch synthase; BE, branching enzyme; ISA, isoamylase; SP, starch phosphorylase. Dashed arrows indicate steps where several enzymes and intermediates are involved.

Lipid Production Potential and Fatty Acid Profile

Since the previous results showed that the sds1, sds3, and sds4 mutants could tolerate light intensity up to 700 μmol photons m-2 s-1 and their TAG yield increased in response to nitrogen depletion, the lipid production potential and fatty acid composition of the algal strains grown under this condition were reevaluated and quantified in more detail. The results showed that growth performance of the wild type and the mutants were similar in terms of final biomass concentration and cell density (Fig. 8A). Gravimetric quantification of the total lipid extracted from the wild type and the mutants also showed no significant difference, ranging from 20.0% to 23.2% (w/w) lipid per dry algal biomass or 2.3 to 3.0 μg lipid per 106 cells (Fig. 8B). Since the crude lipid extract from algae usually contains several classes of non-acyl lipid, such as carotenoids, which cannot be converted into fatty acid esters, the total acyl lipid content was further determined by transesterification of the total lipid with methanol. GC analysis of the FAMEs showed that the wild type produced 2.3% (w/w) acyl lipid per dry algal biomass or 0.27 μg acyl lipid per 106 cells. In contrast, the amounts of acyl lipid produced by the mutants were increased by 39-52% per dry weight, or 48-63% per cell (Fig. 8C). To quantify the TAG content, nonpolar lipids were further purified from the crude lipid extracts using a silica-SPE cartridge. Preliminary experiments were conducted to determine an appropriate volume of the eluent (Fig. S1) and assess the purity of the eluates (Fig. S2). The SPE-purified nonpolar lipids were subsequently transesterified into FAMEs. GC analysis of FAMEs from TAG showed that the wild type produced 1.2% (w/w) TAG per dry algal biomass or 0.14 μg TAG per 106 cells. In contrast, the amounts of TAG produced by the mutants were increased by 42-92% per dry weight, or 43-100% per cell (Fig. 8D).

Fig. 8.Growth and lipid production potential of the D. tertiolecta wild type and starch-deficient mutants. The algal strains were cultured under 700 μmol photons m-2 s-1 in N-replete medium for 5 days and transferred to N-depleted medium for 2 days before the harvested cells were analyzed for growth yield (A), total lipid content (B), total fatty acid content (C), and fatty acid content of TAG (D). Values are from three biological replicates. Error bars are standard errors. An asterisk (*) indicates a significant difference calculated by one-tailed t-test (p < 0.05) between the mutant and the wild type. Light bars, biomass concentration or calculated per biomass; dark bars, cell density or calculated per cell.

Comparison of fatty acid constituents of total lipid (Fig. 9A) and TAG (Fig. 9B) of the wild type and the mutants showed that there was no substantial difference, except for a slight increase of oleic acid (C18:1) and a decrease in linolenic acid (C18:3) in the mutants compared with the wild type.

Fig. 9.Fatty acid profile of total lipid (A) or TAG (B) from the D. tertiolecta wild type and starch-deficient mutants. The algal strains were cultured under 700 μmol photons m-2 s-1 in N-replete medium for 5 days and transferred to N-depleted medium for 2 days before lipid extraction. Relative amounts of major fatty acids are shown. Values are from three biological replicates. Error bars are standard errors.

Discussion

Several microalgal species have established themselves as candidates for biodiesel feedstock. In addition to a rapid growth rate and high oil content, ease of large-scale cultivation and co-production of high-value products are among the major considerations. These latter attributes are particularly important characteristics of Dunaliella. In fact, D. salina has been cultivated at a commercial scale to produce β-carotene [34]. Although the TAG content of Dunaliella is relatively low compared with other algal biodiesel candidates [8], our work has demonstrated that it can be genetically improved. This technique may also be applicable to other algal species that have all the important characteristics but possess a relatively low oil content.

Starch accumulation has been shown as an initial response to nitrogen starvation in a number of algal species, including Dunaliella [22,38,40,43], as it is believed to operate as a short-term energy storage mechanism under adverse conditions. The degradation of starch, which usually occurs when the duration of nitrogen starvation is prolonged, has the possible function of supporting TAG synthesis as a long-term energy storage mechanism [6,13,20,43]. Impediment in starch synthesis would redirect carbon metabolites toward TAG synthesis, at least during the early response to nitrogen starvation. However, this relationship between the starch and TAG metabolic pathways had previously been shown only in a number of C. reinhardtii starchless mutants and in a few algal species. The results shown in this study further supported this hypothesis and provided another clear demonstration that such mutants could be quite easily obtained through random mutagenesis and mutant screen, which do not involve transgenic modification. All of the mutants presented in this study were confirmed to have significant reduction in starch content, but some starch remained present; hence, we referred to these mutants as starch-deficient strains. The reduction in starch content could be observed only under N-depleted condition (Fig. 1B) possibly because the basal starch content under N-replete condition was too low to detect a further reduction, or because the genes affected in these mutants may have a role in starch metabolism predominantly under the N-depleted condition. Consistently, the difference in TAG content between the wild type and the mutants could not be observed when the cells were grown in N-replete medium, as neither the wild type nor the mutants accumulated TAG (Fig. 3A). Under the N-depleted condition, three of the isolated mutants showed a significant and/or substantial increase in TAG content, ranging from 0.4- to 2-fold elevation in TAG accumulation (Figs. 3B and 8D).

The gravimetric quantification of crude lipid extracts (Figs. 2 and 8B) showed little or no difference between the wild type and the mutants, whereas the GC analysis of derivatized FAMEs (Fig. 8C) showed the mutants had a higher lipid content. This may be explained by impurity of the crude lipid extracted using Bligh and Dryer’s solvent, which typically includes additional cellular components, such as pigments, that out-weighed the acyl lipid content and masked the differences among the algal strains. Derivatization of the lipid into FAMEs allows more specific quantification of the acyl lipid, which is the primary substrate for biodiesel production [7]. SPE purification of the crude lipid extract followed by FAME derivatization allows assessment of TAG (Fig. S2). Based on the GC analysis of FAMEs, the total acyl lipid and TAG content of wild-type D. tertiolecta cell were slightly lower than those reported by Breuer et al. [8]. The lower lipid yields in our study may be caused by discrepancies in culture conditions, such as no additional CO2 supply and shorter duration of nitrogen starvation. Based on mass balance, the elevation in total acyl lipid content of the sds mutants over the wild type is mostly due to their increased TAG pool, suggesting that the metabolic redirection from reduced starch biosynthesis is used toward TAG production rather than synthesis of other classes for acyl lipid. Comparative profiling of FAMEs derived from the total acyl lipid of D. tertiolecta grown under N-depleted condition showed that the most abundant fatty acid was C16:0, followed by C18:3, C18:1, C18:2, C16:4, and C18:0, respectively (Fig. 9A). Similar results were reported by Breuer et al. [8] and Lee et al. [19], although in these studies C18:3 appeared more abundant than C16:0. These major fatty acids were also the main constituents of TAG (Fig. 9B). Interestingly, all three mutants possessed a higher content of C18:1 in TAG than the wild type. Thus, it might be possible that C18:1 is preferentially involved in TAG synthesis, for example as a main acyl substrate, during the metabolic redirection in the sds mutants. Consistent with this hypothesis, a lipidomic study of D. tertiolecta grown under nitrogen starvation showed a marked increase of diacylglyceryltrimethylhomoserine carrying C18:1, which was further hypothesized to function as an acyl donor in the biosynthesis of TAG under this condition [15]. However, this hypothesis remains to be tested.

The four isolated starch-deficient mutants exhibited varying degrees of starch deficiency and elevation of lipid content. In addition, the gain in TAG in each sds mutant did not strictly correlate with the loss of starch content, implying that the characteristics of starch synthesis inhibition—and consequently, the metabolic flux modifications—in these mutants are distinct. The fact that no TAG was detected in sds2 suggested that the mutation might affect an upstream metabolic or regulatory process, leading to the inhibition of both starch and TAG synthesis. Real-time RT-PCR analysis showed that different sets of genes exhibited altered expression in the three sds mutants. One explanation is that these mutants carried different mutations and thus were affected at different steps of starch synthesis and metabolism, resulting in varying degrees of starch deficiency and accumulation of different metabolic intermediates. Some intermediates might also function as signaling molecules that could affect expression of multiple genes in the pathway. In fact, it has been suggested that in the C. reinhardtii sta6 mutant, which cannot convert glucose-1-phosphate to ADP-glucose, glucose-1-phosphate might be converted to trehalose, a molecule that might have a role in signaling [3,4]. In addition, the recently described starch-deficient mutants of Chlorella sorokiniana did not show increased accumulation of TAG [39]. These mutants were defective in isoamylase and starch phosphorylase, which are involved in the formation of starch granule from glucan chains [39]. It was suggested that the regulation of carbon partitioning between starch and TAG formation might actually be linked to glucan synthesis rather than the availability of carbon precursors [39]. Therefore, the steps where the starch synthesis is blocked might largely contribute to differences in phenotypes observed in starch-deficient mutants.

It has been reported that the lack of starch in the C. reinhardtii BAFJ5 mutant causes an immense reduction in photosynthesis efficiency, leading to susceptibility to high light intensity and growth retardation, particularly under N-depleted condition [16,22]. Metabolomic study revealed that the loss of starch biosynthesis in this mutant led to a slower rate of NADPH re-oxidation, resulting in redox imbalance and ultimately photoinhibition [16]. In contrast to the reported high light sensitivity of the C. reinhardtii starch-deficient mutant, our study showed that only sds3 among the isolated D. tertiolecta mutants was severely sensitive to high light intensity at 1,200 μmol photons m-2 s-1, suggesting that a decrease of starch content is not sufficient to cause this sensitivity. One possible explanation for this contradiction is that our mutants retained a substantial level of starch synthesis capability, which might still be enough to maintain a normal cellular redox state. However, S. obliquus slm1, with only 1.2% starch remained, also did not show adverse effects on photosynthesis compared with the wild type, even under N-depleted condition [6], implying that there might be other mechanisms that could compensate for the lack of starch as an energy carbon sink. An accumulation of TAG under stress conditions is also suggested to be a protective mechanism, protecting the cells from the damage caused by reactive oxygen species [14,21]. The relative importance of these storage compounds in alleviating cellular damage from oxidative stress should be further explored. Nevertheless, our work and the work on S. obliquus demonstrate the feasibility that a starch-deficient mutant can be isolated without associated growth retardation or high light sensitivity.

The net improvement of TAG content in starch-deficient mutants varies from species to species. For example, the starchless C. reinhardtii BAFJ5 mutant was reported to have 10-fold elevation of TAG [22]. On the other hand, the S. obliquus slm1 mutant only increased its TAG content from 45% in the wild type to 57% in the mutant [6]. These variations in the fold increase of TAG might partly be explained by the differences in basal starch and TAG contents produced by the wild-type strains and the increased level of starch synthesis that normally occurs upon stress conditions. However, the degree of TAG improvement may be compounded by additional factors. Both C. reinhardtii and D. tertiolecta favor starch as their main storage compound [8,22,31]. In the case of C. reinhardtii, the starch content increases from 2-6% to over 40% during its initial response to nitrogen starvation, and thus the blockade in starch synthesis resulted in large TAG improvement as in the BAFJ5 mutant [1,22]. In contrast, our study showed that the starch content of wild-type D. tertiolecta increased approximately 3-fold, from 5% to 21%, yet the TAG content of the sds mutants increased by only 1-2-fold. The difference may be partly explained by the substantial amounts of starch remained in our mutants. Nevertheless, it is also possible that the starch and TAG metabolic pathways are not as connected in D. tertiolecta as in C. reinhardtii. In fact, Breuer et al. [8] showed that unlike the other algae tested, D. tertiolecta cells did not degrade their starch under prolonged exposure to nitrogen starvation, implying that D. tertiolecta may not depend on the stored starch for TAG biosynthesis. Furthermore, in D. tertiolecta subjected to nitrogen starvation, while TAG was accumulated, the total acyl lipid content did not increase, suggesting that its TAG biosynthesis is coupled with the breakdown of preformed acyl lipid [8]. Consistent with this hypothesis, using 14C isotope labeling, Davidi et al. [9] showed that in D. bardawil, the TAG pool stored in cytoplasmic lipid bodies is synthesized mostly through de novo fatty acid synthesis, which occurs for a brief period of time following nitrogen starvation, whereas TAG in the chloroplastic plastoglobuli continues to be produced using the precursors from membrane lipid and cytoplasmic TAG.

Carotenoid levels have been found to increase in response to nitrogen starvation in carotenogenic species, such as D. salina and D. bardawil [9,18]. In D. tertiolecta, carotenoid overaccumulation is typically not substantial in cells under nitrogen starvation condition, although it has been observed in cells grown in an N-replete medium at stationary phase [15]. In contrast to the S. obliquus starch-deficient slm1, which contained more carotenoid than the wild type [6], our results showed that the inhibition of starch synthesis in D. tertiolecta did not result in carotenoid overaccumulation. As carotenoids have a high commercial value, the fact that the carotenoid content in starch-deficient mutants did not decrease under TAG-accumulating conditions indicates that carotenoids could be used as potential co-products for reducing the cost of biodiesel production from this algal species.