Highly efficient lipid production in the green alga Parachlorella kessleri: draft genome and transcriptome endorsed by whole-cell 3D ultrastructure

Background Algae have attracted attention as sustainable producers of lipid-containing biomass for food, animal feed, and for biofuels. Parachlorella kessleri, a unicellular green alga belonging to the class Trebouxiophyceae, achieves very high biomass, lipid, and starch productivity levels. However, further biotechnological exploitation has been hampered by a lack of genomic information. Results Here, we sequenced the whole genome and transcriptome, and analyzed the behavior of P. kessleri NIES-2152 under lipid production-inducing conditions. The assembly includes 13,057 protein-coding genes in a 62.5-Mbp nuclear genome. Under conditions of sulfur deprivation, lipid accumulation was correlated with the transcriptomic induction of enzymes involved in sulfur metabolism, triacylglycerol (TAG) synthesis, autophagy, and remodeling of light-harvesting complexes. Conclusions Three-dimensional transmission electron microscopy (3D-TEM) revealed extensive alterations in cellular anatomy accompanying lipid hyperaccumulation. The present 3D-TEM results, together with transcriptomic data support the finding that upregulation of TAG synthesis and autophagy are potential key mediators of the hyperaccumulation of lipids under conditions of nutrient stress. Electronic supplementary material The online version of this article (doi:10.1186/s13068-016-0424-2) contains supplementary material, which is available to authorized users.


Background
Microalgae accumulate oil as storage lipids (TAGs), and are ideal species for developing the highly productive strains that are essential for biofuel production [1][2][3]. Initial efforts to exploit microalgae for biotechnology date back to approximately 140 years ago [4]. Large-scale algal cultivation has been optimized since then, but this process was, in many cases, driven by trial and error, thus hindering the most effective utilization of algal potential [4]. Clearly, rapid development in this field can be achieved only through optimization of growth conditions guided by the results of basic research on algal physiology, morphology, and genomics.
The class Trebouxiophyceae is a major group in the green algal phylum Chlorophyta [5,6]. One of the trebouxiophycean genera, Chlorella, has been well-studied with regard to its physiology and has been exploited industrially due to its high photosynthetic growth rate and excellent biomass productivity [7][8][9]. Recently, Chlorella and the closely related genus, Parachlorella [10], have attracted attention as potential producers of triacylglycerols (TAGs) and high value-added, long-chain fatty acid feedstocks [1,11,12]. Recently, a new algal species closely related to Parachlorella kessleri, which is also a high lipid producer, was reported from the Indian Ocean [13]. Parachlorella kessleri, one of three commonly described Parachlorella species [10,14], is an organism that achieves very high biomass, lipid, and starch productivities [15,16]. Importantly, P. kessleri is one of only a few species in which lipid productivity has been assessed not only under laboratory conditions [15][16][17] but also on a semi-industrial scale in outdoor photobioreactors [18], and is considered an ideal microalgal species for biofuel production.
Given their industrially verified superior growth properties, combined with high levels of lipid productivity, these species may be sustainable sources of TAG and an alternative to petroleum-based diesel fuels in the biofuel industry [1,11]. Of the potential TAG producers, only the genome of Nannochloropsis gaditana [19], a member of the stramenopiles [20], has been sequenced. Moreover, a species from the same genus was shown to be capable of homologous recombination, which would thus allow for efficient gene targeting [21]. Although some green algae, particularly those belonging to the genera Chlorella and Parachlorella, are more productive lipid producers than Nannochloropsis [22], only few genomic sequence have been reported (Chlorella variabilis, an endosymbiont of ciliates [23], and Chlorella protothecoides [24]), and a genome sequence of autotrophic Chlorella is currently lacking, thus limiting potential improvements in growth and productivity guided by omics techniques.
Some studies have shown that lipid content in algae can be increased by nutrient depletion [1,4,[25][26][27][28]. Among the macronutrients (nitrogen, phosphate, and sulfur), nitrogen deprivation is widely used for stress experiments [26,27,[29][30][31][32][33][34]. However, such stresses lower the growth rate and productivity of the system [1,35], which is a major bottleneck for producing biofuels and byproducts on commercial scales, and therefore some studies have addressed solutions using genetic engineering [36,37]. Sulfur deprivation is an alternative stressor for the induction of starch or TAG biosynthesis [4,16,38]. Notably, a high starch content was maintained for a long period under conditions of sulfur starvation, suggesting that sulfur depletion is an effective method of enhancing starch productivity [4]. In contrast to Chlamydomonas [39], the cellular behavior and metabolic and transcriptomic responses under sulfur-depleted culture conditions are less well-characterized in Chlorella species. Here, we first report the reference genome of P. kessleri NIES-2152. Second, we also examined the transcript patterns and cellular anatomy to evaluate responses to sulfur deprivation in the induction of TAG and carbon hydrate production [4] using phenotypic assays, RNA-seq, and three-dimensional transmission electron microscope.

Results and discussion
The Parachlorella genome was sequenced and 27.8-fold pyrosequencing reads were obtained (4,365,609 reads in total). Reads were assembled into 5168 contigs and a total of 400 scaffolds were acquired. The present scaffolds and contigs are from nuclear, plastid, and mitochondria genomes. The total nuclear genome size was estimated to be 62.5 megabase pairs (Mbp) and 13,057 genes were identified (Table 1, Additional file 1: Figures S1 and S2). Thirty major scaffolds cover 48.8 % of the genome. Of the annotated genes (Additional file 1: Table S1), 49.7 % of their proteins were associated with Kyoto encyclopedia of genes and genomes (KEGG) orthology numbers.
To characterize the phenotype and transcriptome under sulfur deprivation, P. kessleri was batch-cultivated under continuous light (LL) and sulfur replete and deplete (±S) conditions. Growth was determined as biomass dry weight as well as cell density. As expected, growth was restricted under sulfur-depleted conditions, in contrast to its growth in tris-acetate-phosphate (TAP) medium, where a logarithmic growth phase was clearly observed (Fig. 1a). Higher starch contents per cell were observed after 2 days in culture under conditions of sulfur deprivation (6.19-and 6.62-fold changes on days 2  . 1b), consistent with results from the previous sulfur-depletion study [38].
There was a remarkable difference in total lipid content between ±S cultures. After 5 days of cultivation, P. kessleri growing in dSTAP (sulfur-deprived TAP) reached a yield of 0.25 mg mL −1 total lipids, whereas P. kessleri in TAP gradually accumulated total lipids over a 10-day period of batch cultivation, but achieved a yield of only 0.14 mg mL −1 total lipid (Fig. 1c). This finding indicates that total lipid yield was accelerated under sulfur deprivation, in contrast to cultivation in TAP medium, so that in the 5-day-old, sulfur-deprived culture, the lipid content represented up to 50.7 % of the dry weight (Fig. 1d).
A previous report showed that lipid accumulation under nitrogen-limited conditions was tightly connected with cellular processes that are related to lipogenesis, macromolecule metabolism, and autophagy in a lipidproducing yeast [40]. Here, we focus on the autophagyrelated genes to show the transcriptome dynamics. During sulfur deprivation, some autophagy-related genes were upregulated in the early and late logarithmic-phase cultures (Fig. 1e). However, no genes were downregulated in the stationary-phase culture, except for ATG3, ATG5, and beclin 1, suggesting that autophagy and recycling of subcellular components are enhanced under sulfur-depleted conditions in the early and late logarithmic phases.
Transmission electron microscopy allows high-resolution imaging to study cellular anatomy and ultrastructure, which is inaccessible by other techniques. This is essential when studying cells in which one of the organelles is enlarged, covering other compartments and thus more or less blocking their observation by other means. Examples of such situations are cells over-producing astaxanthin [41] or lipids [42]. The main limitation of TEM is that it traditionally facilitates only two-dimensional imaging of approximately 80 nm ultrathin sections. This can be overcome by analyzing sequential sections of a single cell stacked together using three-dimensional transmission electron microscopy (3D-TEM) technology, leading to a three-dimensional image of cell ultrastructure with very high resolution. The technique was used here to study lipid hyperaccumulation within P. kessleri cells in the stress experiment.
Representative cells from three phases-logarithmic growth (control), starch-rich, and lipid-rich, were analyzed using 3D-TEM (Fig. 2, Additional files 2, 3, 4 and 5). In the logarithmic growth phase, little starch was accumulated and no lipid bodies were found ( Fig. 2a-c, Additional file 2). The chloroplast and mitochondria accounted for 38.5 and 5.9 % of the relative volume, respectively (Fig. 2j). In the starch-rich phase, many starch grains were observed in the chloroplasts ( Fig. 2d-f, Additional file 3), accounting for 9.7 % of the relative volume ( Fig. 2k), whereas lipid bodies showed less accumulation (0.4 % of the relative volume). In this phase, the relative volume of the chloroplasts was 34.3 % (Fig. 2k).
In the lipid-rich phase, the ultrastructure and volume of the subcellular components were dramatically changed ( Fig. 2g-i, Additional files 4 and 5). The most remarkable change from the starch-rich stage to the lipid-rich stage was the relative volume of chloroplasts and lipid bodies (Fig. 2i). The chloroplast was highly degenerate and its relative volume was reduced almost tenfold to 4.4 % (Fig. 2l); it was also re-located to one side of the cell periphery (Fig. 2g, Additional file 4). Concurrently, lipid bodies over-accumulated and accounted for 51.7 % of the relative volume (Fig. 2l).
Compared to the morphological changes in Chlorella sorokiniana under nutrient stress [42], P. kessleri possessed larger lipid bodies and these accounted for a considerable portion of the cell. This was accompanied by degradation of the chloroplast in the lipid-rich phase of P. kessleri growth (Fig. 2g, Additional files 4 and 5), but not in C. sorokiniana. This reflects the main differences between P. kessleri and other Chlorella species with respect to lipid production. The subcellular degradation process probably, and at least in part, involves autophagy through upregulation of ATG-related transcripts, thus enabling recycling of subcellular compartments and increased lipid accumulation.
Levels of individual transcripts were expressed in a heat map and subsequently applied to a metabolic pathway map (Fig. 3, Additional file 6) and KEGG category analysis (Additional file 1: Figures S3-S5). Overall, the expression of genes involved in the Calvin-Benson, tricarboxylic acid (TCA), glyoxylate, and C4 dicarboxylic acid cycles showed decreased expression under sulfurdepleted conditions. Other metabolic pathways, such as fatty acid metabolism, autophagy, TAG, cysteine, and methanethiol synthesis were transcriptionally upregulated (Fig. 3). KEGG category analysis also indicated that cysteine and methionine metabolism (category 3: minor function category) were upregulated during the entire cultivation period under sulfur deprivation (Additional file 1: Figure S4).
Under sulfur deprivation, the sulfate transporter gene (10836_t) was highly upregulated (log 2 value = 6.6 in the late logarithmic phase, P < 0.0005), together with cysteine dioxygenase (6136_t) (log 2 value = 2.5 and 5.6 in the late logarithmic and stationary phases, respectively; P < 0.05 at stationary phase) (Additional file 7). This is in agreement with the induction of transcripts associated with sulfur acquisition and assimilation, synthesis of sulfurcontaining amino acids, cysteine degradation, and sulfur recycling, as shown in Chlamydomonas [39]. In addition, Beclin 1 7935_t Autophagy-related protein 4 4380_t Autophagy-related protein 5 9402_t Autophagy-related protein 3 7174_t V-type proton ATPase catalytic subunit A-like 8008_t Autophagy-related protein 12 2754_t Apg1/Unc-51-like serine-threonine kinase 6863_t Autophagy-related protein 7 892_t Phosphatidylinositol 3-kinase 7007_t Phosphoinositide-3-kinase 12392_t V-type H+-transporting ATPase subunit I 7173_t V-type H+-transporting ATPase subunit A 4345_t 5'-AMP-activated protein kinase 7175_t V-type H+-transporting ATPase subunit A 7893_t Autophagy-related protein 8 7418_t 5'-AMP-activated protein kinase 10798_t 5'-AMP-activated protein kinase 11799_t Autophagy-related protein 101-like 12432_t Os06g0267600; Autophagy-related   the gene for methionine-gamma-lyase (MGL) was highly upregulated (Fig. 3). Degradation of l-methionine is catalyzed by MGL, resulting in methanethiol production. Because methanethiol is a sulfur-containing compound, we suspect that this metabolic pathway may be used for recycling of sulfur under sulfur-limited conditions. It should be emphasized that transcripts responsive to sulfur deprivation also included genes for the lightharvesting complex of PSII (LHCB) (Additional file 7). Almost all genes encoding LHCBs were downregulated or did not show varying levels of expression between the ±S conditions; only one gene, LHCB1 (1779_t), was strongly upregulated (P < 3 × 10 −7 ). Compared to the no-stress condition, transcript levels of LHCB1 under sulfur-depleted conditions increased by 2050-and 5106-fold in the late logarithmic and stationary phases, respectively. LHCB1 (1779_t) is a homolog of LHCBM9 in Chlamydomonas (JGI accession no: 184479). In Chlamydomonas [39,43], LHCBM9 was upregulated during sulfur deprivation. In the RNA-seq analysis [39], LHCBM9 transcripts, which were barely detectable when cells were grown in +S medium, were upregulated by >1000-fold during sulfur deprivation, whereas transcript levels of the other LHCBMs declined, similarly to P. kessleri (Additional file 7). Because LHCBM9 contains at least three fewer sulfur-containing amino acids than other LHCBMs, it was suggested that exchange for LHCBM9 might enable recycling of sulfur-containing amino acids and their reallocation among other proteins with low sulfur amino acid contents. This process may represent an extreme measure for sulfur use that could extend cell viability under sulfur-depleted conditions. This notion was further supported by upregulation of some autophagy-related genes in the early and late logarithmic phases of sulfur-depleted cultures (Fig. 1e).
Considering transcriptionally upregulated metabolisms that involve hyperaccumulation of lipids, TAG synthesis (e.g., DGAT1) and autophagy emerge as candidates for the key mediator of lipid hyperaccumulation under stress conditions. The present observation based on the 3D-TEM data demonstrates that hyperaccumulation of lipids and high degradation of the chloroplast occur concurrently under stress conditions. These results illustrate that analysis based on 3D-TEM and phenotypic assays confirm the finding that the upregulation of a key mediator is involved in hyperaccumulation of lipids under nutrient stress conditions in P. kessleri.
Overall, this study supports the notion that basic research is required to accelerate development in the field of algal biotechnology. The genomic information on the industrially verified, biomass-producing algal species P. kessleri provided here will facilitate both basic and applied research not only in the field of algal-derived biofuels but will also serve as a foundation for future genetic manipulation of the TAG biosynthetic pathway (and others) in this species. Similarly, the 3D-imaging approach is a promising tool for analysis of global ultrastructural changes in whole-cells.

Conclusions
We sequenced and analyzed the whole genome of P. kessleri, a high-biomass lipid-rich green alga related to Chlorella. Transcriptomic analysis suggested that lipid accumulation under conditions of sulfur depletion is associated not only with the induction of sulfur metabolism but also TAG synthesis, light-harvesting complexes, and autophagy. Moreover, the metabolic changes under ±sulfur conditions, autophagy-like changes in cell anatomy following lipid accumulation, were visualized by 3D-TEM ultrastructural analysis. The Parachlorella genome information provided by this study will facilitate basic research and further analysis of applied phycology, in addition to potential genetic manipulation, of the TAG biosynthesis pathway in this industrially verified biomass-producing species.

Growth conditions and DNA and RNA extraction
Parachlorella kessleri NIES-2152 was obtained from the National Institute for Environmental Studies in Tsukuba, Japan. For genomic DNA extraction, P. kessleri cells were grown at 20 ℃ under 70 µmol photons m −2 s −1 for 12 h:12 h = light (L):dark (D) cycle at 20 ℃ in TAP medium (Additional file 8: Table S2). Cells from 2-week-old cultures were harvested by centrifugation at 2500g for 10 min at room temperature (r.t.), and the resulting pellet (10 g of wet weight) was immediately frozen with liquid nitrogen. DNA extraction was performed using the DNeasy Plant Maxi Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. For RNA extraction, cells were grown in TAP or dSTAP (Additional file 8: Table S3) medium in a 500 ml flask (Iwaki, Tokyo, Japan) under 100 µmol photons m −2 s −1 for 12 h:12 h = light (L):dark (D) cycle at 20 ℃. Cells from 2-week-old cultures were harvested by centrifugation at 2500g for 10 min at r.t., and the resulting pellet was immediately frozen with liquid nitrogen. Total RNA extraction was performed using the Sepasol ® -RNA I Super G kit (Nacalai Tesque, Kyoto, Japan) followed by poly (A) mRNA purification with Dynabeads Oligo (dT) (Life Technologies, Carlsbad, USA) in cultures grown under conditions of ±sulfur in the early (2 days old) and late (4-5 days old) logarithmic phases and stationary phase (7-8 days old). In this experiment, 12.5 mL of the Sepasol ® -RNA I Super G for every 0.5 g of cells were used in each experimental group. The purification was performed in three rounds. In the first round, 150 µg of total RNA were used for mRNA purification using 1 mg of Dynabeads Olig (dT) according to the manufacturer's protocol with modifications. Finally, 500 ng of mRNA were obtained.

Genomic and mRNA sequences and annotation
The genomic sequence of P. kessleri NIES-2152 was determined using 454 pyrosequencing for single-end (SE) and paired-end (PE, 8 kb-span library) data. We generated 3,561,169 reads by SE and 804,440 reads by PE, which provided 27.8-fold coverage of the genome. The assembly of the obtained sequence data resulted in the generation of 400 scaffolds using Newbler version 2.8 (Roche, Branford, CT, USA). The total scaffold length was 2,653,566 bp with a G + C content of 58.4 %. Proteinencoding regions were predicted using the GeneMark-ES [44]. We annotated 13,057 predicted genes using the BLASTP program [45] (e-value cut-off of 1E−5) with the NCBI-nr and KEGG databases.

Reconstruction of KEGG pathway map and mRNA expression analysis
The RNA-seq reads were mapped to the predicted genes using Newbler v.2.8. Comparative expression analyses were performed using standardized reads per kilobase of exon per million mapped sequence reads (RPKM) values. Functional annotation descriptions were assigned by BLASTP [45] with the KEGG database (e-value cutoff of 1E−10). The metabolic network was reconstructed using KEGG mapper (http://www.genome.jp/kegg/) with the KO numbers as objects shown in Additional file 6. The resulting KEGG map was redrawn manually using Adobe Illustrator v. 16.0.4 (Adobe Systems). An abbreviated gene names are given in Additional file 6. A heat map of the autophagy-related transcriptomes was generated using the heatmap2 function from the Gplots package v. 2.16.0 in the R statistical software v. 3.1.0 (http://www.R-project.org/). Statistical testing for gene expression was performed in R with DESeq [46] using the no replicate method. The P values are provided in Additional file 9. We performed KEGG category analysis in which different sequences were treated as different genes.

Phenotypic assays for biomass and lipid production
Pre-cultures were grown under 100 µmol photons m −2 s −1 for 12 h:12 h = L:D cycle at 23 ℃. Cells of 4-dayold cultures were centrifuged at 1500g for 5 min at r.t. and collected as a pellet. The pellet was re-suspended and inoculated into 500 mL of TAP or dSTAP medium (the initial concentration of the culture was ~7 × 10 6 cells mL −1 ). The batch cultures were grown under continuous illumination (100 µmol photons m −2 s −1 ) at 21-23 °C and agitated with a magnetic stirrer (MGM-66, Shibata, Tokyo, Japan) at 100-150 rpm.
Cells were counted using a particle counter (CDA-1000, Sysmex, Kobe, Japan). A general linear model was fit to values of the cell numbers in each condition for growth curves. For dry weight determination, an aliquot of cell culture was sampled into a pre-weighed sampling tube and centrifuged at 6000g for 5 min at r.t. The supernatant was then removed and the cell pellet was dried for at least 3 h to a constant weight at 105 ℃. The sampling tube was weighed using a precision analytical balance (NewClassic MS, Mettler Toledo, MD). For total lipid extraction and measurements, we used a previously described method [16]. Briefly, total lipids were extracted using methyl-tertbutyl ether (MTBE) [47], and the weight of total lipids was measured gravimetrically using a precision analytical balance. Starch content was quantified using the Lugol staining method as described [48]. Growth curves were estimated using a general linear model (glm function), and other curves (except for starch assay) were fitted using the ksmooth function in R statistical software v. 3.1.0.

3D-TEM analysis
We observed three representative stages: control (4 days old in TAP under LD), starch-rich [6 days old in dSTAP under light/dark cycle (LD)], and lipid-rich (6 days old in dSTAP under LL). Cells at each stage were pre-fixed for 2 h with 2.5 % glutaraldehyde, post-fixed with 1 % OsO 4 for 2 h at r.t., and then rinsed with 0.05 M sodium cacodylate buffer (pH 7.2). The fixed cells were then dehydrated using a graded ethanol series, incubated in ethanol:acetone = 1:1 and finally suspended in 100 % acetone at r.t. The dehydrated samples were infiltrated with increasing concentrations of Spurr's resin [49] in acetone and finally with 100 % Spurr's resin. Ultrathin serial sections were cut on a Reichert Ultracut S ultra-microtome (Leica, Vienna, Austria) using a diamond knife. Serial sections were mounted on copper grids coated with polyvinyl formvar films and stained in 3 % aqueous uranyl acetate and lead citrate [50]. The sections were observed at 100 kV using an H-7650 transmission electron microscope (Hitachi High Technologies, Tokyo, Japan). 3D-TEM imaging followed a previously described method [41]. Briefly, contours of each subcellular element (e.g., nucleus, chloroplast, lipid body) were traced manually. After binarization of the traced subcellular elements, 3D images were reconstructed using TRI/3D SRFIII software (Ratoc System Engineering, Tokyo, Japan). Voxel-based volumetric analyses were performed using the TRI/3D SRFIII system and data were presented as the means of two representative cells.