Potato tuber expression of Arabidopsis WRINKLED1 increase triacylglycerol and membrane lipids while affecting central carbohydrate metabolism

Gene construct and transformation for AtWRI1 expression in potato tubers

A synthetically produced gene corresponding to the WRINKLED1 coding region of Arabidopsis thaliana (At3 g54320) (AtWRI1) was cloned after a GBSS promoter (Solanum tuberosum) which has been characterized as tuber specific (Visser et al., 1991). Transformation was performed on potato variety Kuras, and putative transgenic shoots were selected on regeneration medium containing kanamycin. Based on PCR analysis, 45% of analysed shoots showed T‐DNA integration (results not shown) and individual shoots were selected for propagation of in vitro microtubers. Lipids were extracted from microtubers of ten individual transgenic lines and analysed for changes in content of neutral lipids in comparison with Kuras in vitro microtubers. Several lines displayed an increase in triacylglycerol (TAG) content, of which lines 8001, 8003, 8016 and 8022 were selected for more detailed studies (Figure S1). TAG content of Kuras microtubers was found to be 0.1 mg/g dw while a content of more than 20 mg/g dw (2% TAG of dry weight) could be measured for individual transgenic microtubers (Figure S2). Consistent for TAG from selected transgenic microtubers was a change in fatty acid (FA) composition with an increase in 18:2 and 18:1 at the expense of 18:3 (Figure S3). All the selected transgenic lines contained multiple inserts of the T‐DNA (Figure S4).

AtWRI1 induces oil accumulation in soil grown tubers

Cuttings of Kuras and lines 8001, 8003, 8016 and 8022 were transferred to soil in the greenhouse. Growth and development of transgenic lines were indistinguishable from control with the exception for line 8022 which appeared erect, thin in growth and with small curly leaves. Greenhouse cultivation lasted for 4 months, and sampling was performed after 2 (S2), 3 (S3) and 4 months (S4). After 4 months, the haulm was cut and watering stopped 2 weeks prior to harvesting. All transgenic lines displayed altered tuber morphology. The parental variety Kuras has rounded and slightly flattened tuber morphology, while tubers of the transgenic lines generally were slightly elongated, had deeper eyes and were rich in secondary tuber developments (Figure 1). Transgenic lines, with the exception of line 8022, on average produced more tubers than wild type but of less average weight (Figure 2). Total tuber mass production per plant showed a slight average increase with the exception for line 8022 (Figure 2). Line 8022 was excluded from further studies and generalized conclusions as it displayed aberrant growth. Dry matter content was on average less for transgenic lines compared to control but increased for both control and transgenic lines between S3 and S4 (Figure 2).

Figure 1

Phenotype of greenhouse grown tubers, (a) Kuras, (b) 8001, (c) 8003, (d) 8016. Solid arrows indicate secondary tuber developments and dashed arrows indicate deeper eyes of transgenic tubers.

Figure 2

Greenhouse tubers from Kuras and transformed lines after stage S4 (a‐c) and stage S3 compared to S4 (d). (a) Number of tubers per plant, (b) Tuber mass per plant, (c) Mass per tuber, (d) Dry matter content at stage S3 and S4. Letters distinguish significant different means according to Tukey's test at level P < 0.05.

TAG had increased up to 20‐fold, 3 mg/g dw over control tubers at the time of sampling after 4 months (Figure 3a). However, an intermediate sampling (S3), made after 3 months to represent a metabolic status of active development, showed that the TAG content decreased between S3 and S4 (results not shown). TAG composition analysis revealed a large decrease in 18:3 levels among FA contained in TAG (Figure 3b). Fatty acids associated with polar lipids increased in transgenic lines with a similar absolute amount as for TAG but with no obvious alterations in fatty acid composition (Figure 3c,d).

Figure 3

Analysis of triacylglycerols (TAG) and polar lipids (PL) of greenhouse grown tubers, (a)TAGcontent of tubers of Kuras and transgenic lines, (b) Fatty acid composition ofTAGfrom Kuras and transgenic lines, (c)PLcontent of tubers of Kuras and transgenic lines, (d) Fatty acid composition ofPLfrom Kuras and transgenic lines. Three biological replicates ±SD.

Expression of AtWRI1 directly or indirectly affects carbohydrate metabolism including starch

Metabolites and starch were analysed for the S3 stage, which represents tubers in an active accumulation phase. Transgenic lines showed a 10–18% decreased starch content of dry weight compared to the parental variety at this stage (Table 1).

Metabolites were extracted and investigated for differential concentration in transgenic lines compared to parental genotype. Sugars as sucrose, glucose and fructose were greatly increased in transgenic tubers (Table 1). Most prominent was the increase in glucose which amounted to up to 100‐fold higher (18.4% of dw) than in the parental genotype. Fructose was increased up to 37‐fold (0.5% of dw). TCA cycle intermediates were also affected with a fourfold increase of malic acid or decrease of citric and isocitric acid, respectively (Table 1). Amino acids as alanine and valine were found increased while asparagine was decreased (Dataset S1). The backbone for TAG assembly, glycerol‐3‐P, was also increased in the transgenic lines (Table 1).

Transcriptome analysis reveals differential gene expression of transgenic tubers

Of interest to investigate were pathways leading from sucrose to precursors for fatty acid synthesis and subsequent fatty acid modification and assembly into complex lipids. Line 8016 was selected for comparative analysis at three different time points with the parental genotype. Sequenced reads corresponding to AtWRI1 could be detected at all three stages of development in line 8016 while very low levels of AtWRI1 were detected in the parental control (results not shown). Several genes involved in late steps of plastid glycolysis were found to be up‐regulated in transgenic tubers (Table 2). Genes for plastid localized glycolysis encoding phosphoglyceromutase, enolase and pyruvate kinase were up‐regulated (12‐, 7‐ and 6‐fold, respectively, at S3). Also transcripts encoding plastidic NADP‐malic enzyme were up‐regulated (10‐fold at S3), which could provide an alternate route to supplying pyruvate. In line with these transcripts, encoding plastid dicarboxylate transporters, potentially of importance for malate transport, were up‐regulated. All transcripts encoding subunits of plastidic pyruvate dehydrogenase were up‐regulated (up to 27‐fold at S3) indicating a push towards acetyl‐CoA production. Cytosolic ATP‐citrate lyase transcripts were also found up‐regulated (up to 22‐fold at S3) (Table 3) which could provide additional supply in the direction of pyruvate via oxaloacetate and malate but also acetyl‐CoA as substrate for the mevalonate pathway. A transcript annotated as a phosphoenolpyruvate (PEP) transporter (PPT) was up‐regulated (sevenfold at S3) indicating an enforcement of plastid PEP import and subsequent plastid enzymatic reactions leading to an increased pyruvate supply for fatty acid synthesis (Table 2). The major potato tuber plastid transporter for carbohydrate import, glucose‐6‐phosphate/phosphate translocator (GPT), showed essentially unchanged transcriptional expression level while another contig representing a minor form was up‐regulated although to a low level of expression. A gene encoding a transporter (GlcT) defined as exporting glucose was up‐regulated at later stages of development.

The most highly up‐regulated transcript among genes related to metabolic cycles for substrate supply and processing was annotated as encoding a small subunit of ribulose‐1,5‐bisphosphate carboxylase/oxygenase (Rubisco), which was found having an 4000‐fold increase at S3 (Table 2). Several genes encoding products where the closest homologs in Arabidopsis are annotated as belonging to the pentose phosphate pathway as transaldolase and transketolase were also found to be up‐regulated (7‐ and 4‐fold, respectively, at S3), which could indicate an increased expression of genes for supplying additional substrates for pyruvate synthesis via hexose phosphates to ribulose‐1,5‐bisphosphate and further metabolization through late parts of glycolysis.

All essential transcripts for plastidic fatty acid synthesis were found to be up‐regulated (Table 3). Interestingly, also genes involved in fatty acid modifications and processing were found up‐regulated like transcripts annotated as stearoyl‐ACP desaturase, acyl‐ACP thioesterase and oleate desaturase. Only few genes involved in complex lipid assembly and control were found to be up‐regulated, with examples of genes as phosphatidylcholine:diacylglycerol cholinephosphotransferase (PDCT) and transcripts annotated as oleosins (fivefold and up to 35‐fold, respectively, at S3).

A transcript with homology to LEC1 was found up‐regulated (more than 1000‐fold at S3), which could be connected to the up‐regulation of transcripts downstream of fatty acid synthesis (Table 3).

No transcripts related to β‐oxidation or starch degradation were found to be up‐regulated except for a contig annotated as cytosolic invertase although the final expression levels were still in a low range in comparison with other genes.

Structural studies show presence of oil droplets and changes of starch granules

The cell structure of the tubers was studied under light microscope and transmission electron microscope (TEM) to examine and compare storage cells of the wild‐type and transgenic tubers. Transgenic material displayed smaller and irregularly shaped starch granules compared to the wild type (Figure 4a,c). Oil droplets were detected along the cell walls, which could not be found in the wild type (Figure 4b,d). Furthermore, impact on membranes due to increased amounts of membrane lipids was noted. TEM micrographs revealed clear membrane invaginations (abbreviated as CWT) in the transgenic line (Figure 5).

Figure 4

Light micrographs of Kuras (a,b) and transgenic potato line 8016 (c,d). (a,c) Overview staining withMAS(Triple staining Methylene blue—Azur A—Safranin O) and (b, d) Oil specific staining, Sudan black. Arrows indicate oil droplets, s—starch granule, v—vacuole, cw—cell wall, n—nuclei. Scale bars 10 μm.

Figure 5

Transmission electron micrographs of potato tubers. (a) Kuras, scale bar 1 μm, (b) 8016 scale bar 1 μm, (c) 8016 scale bar 500 nm. S—starch granule, cw—cell wall, cwt—cell wall thickness, o—oil droplet, m—mitochondria, p—peroxisome.

Expression of AtWRI1 in field grown tubers affects tuber yield, size and phenotype

In a field trial, Kuras, 8001, 8003, 8016 and 8022, were grown in four rows of 13–15 plants per row. Lines 8001, 8003 and 8016 could not be distinguished from the parental variety. No differences in emergence, plant morphology, haulm height and size, maturation, wilting or susceptibility to diseases could be observed. The tuber fresh weight yield at maturity was found to be higher for two of the transgenic lines, 8001 and 8016 with 13 and 27 percentage increase, respectively, compared to the parental variety, while line 8003 had 28 percentage points lower yield (Table 4).

For all transgenic lines, the average tuber size decreased and the fraction of tubers smaller than 55 mm increased from 35 to approximately 60% by weight of the harvested tubers for line 8001 and 8016 and more than 90% by weight for line 8003 (Table 4).

Harvested tubers from the transgenic lines showed similar phenotypic alterations as for greenhouse grown tubers with elongated shape and deeper eyes, but the characters were more pronounced (Figure S5).

Dry matter content of smaller and larger fraction of tubers from transgenic lines was similar or slightly lower compared to that of Kuras (Figure 6).

Figure 6

Dry matter content of field grown tubers size fractionated into <55 mm and >55 mm. Three biological replicates ±SD. Letters distinguish significant different means according to Tukey's test at level P < 0.05.

TAG and polar lipid levels and compositions are affected in field grown tubers

TAG was found to accumulate in both smaller and larger tubers of field grown plants. Line 8003 had the highest TAG content with a 38‐fold increase in fraction of smaller tubers and a 125‐fold (0.8% of dw) increase in fraction with larger tubers (Figure 7). Diacylglycerol (DAG) levels were also higher in tubers of line 8003 with an eightfold and 10‐fold increase in respective tuber fractions over control tubers (Figure 7).

Figure 7

Analysis ofTAGandDAGfrom field grown tubers, (a)TAGcontent of tuber of Kuras and transgenic lines, (b)DAGcontent of tubers of Kuras and transgenic lines, (c)TAGcomposition of Kuras and transgenic lines, (d)DAGcomposition of Kuras and transgenic lines. Three biological replicates ±SD.

TAG composition was altered in the transgenic lines, and the most prominent change was related to higher TAG content. Fraction of 18:2 was increased from approximately 40% to around 64% of total FA in TAG, while the fraction of 18:3 was decreased. A very similar pattern could be observed for DAG composition with increases for 18:2 and decreases for 18:3 FA.

Polar lipids were found to be increased in transgenic lines, and their fatty acid composition showed the same trend in the shift as for TAG although not as dramatically (Figure 8).

Figure 8

Analysis of polar lipids from field grown tubers, (a) polar lipid content of tubers of Kuras and transgenic lines, (b) Fatty acid composition of polar lipids from Kuras and transgenic lines. Three biological replicates ±SD.

Increase in tuber lipids is associated with altered starch content and composition in field grown tubers

Starch content and starch structure in tubers of the transgenic lines showed alterations in both accumulation and composition. Increased levels of tuber lipids led to a progressive decrease in starch content in almost all tuber fractions. The exception was line 8001, which showed an unaffected starch content. Starch content in both 8003 and 8016 small tuber fractions was 81% compared to wild type and only 68 and 74% compared to parental genotype in the large tuber fraction, respectively (Figure 9). The higher tuber yield in both 8001 and 8016 compensated for the low starch content, and the total starch yield per plant ended up at the same magnitude as the parental genotype (Figure 9). All transgenic lines had a higher ratio of amylopectin in the starch; however, the increase was not directly correlated to the oil content of the tubers (Figure 9). Compared to Kuras, the starch granules of 8016 were smaller and more irregular in shape (Figure 10).

Figure 9

Starch content, amylose content and starch yield of field grown transgenic lines; 8001, 8003 and 8016 and parental variety Kuras. (a) Starch content (% of fresh weight), (b) Amylose ratio (% of defatted starch), (c) Average starch yield per plant (kg per plant). a‐c n = 3 ± SD.

Figure 10

Light micrographs of starch granules of parental variety Kuras stained with iodine (a‐c) and transgenic line 8016 (d‐f). Scale bars 200 μm (a,d), 100 μm (b,e) and 50 μm (c,f).

Starch‐bound neutral lipids increased in the transgenic lines

Potato starch has a very low amount of lipids bound to the starch. In the transgenic lines, the total content of neutral lipids bound to starch was drastically higher, ranging between 2 and 24‐fold increase compared to parental genotype starch (Table 5). However, the lipid content varied a lot between different tubers from the same line. The average starch‐bound lipid content of lines 8003 and 8016 was in the same range as commercially available maize starch, for which levels of 0.4% have been published (Vasanthan and Hoover, 1992).

Gene construct

A synthetic gene corresponding to AtWRINKLED1 (Cernac and Benning, 2004) was ordered (Eurofins/MWG, Ebersberg, Germany) and cloned after a 990 bp Solanum tuberosum GBSS promoter fragment (Hofvander et al., 2004) in a pBIN19 (Bevan, 1984) derivative binary vector. After controlling the integrity of the plasmid, it was transformed into Agrobacterium tumefaciens, strain AGL0 (Lazo et al., 1991) via direct transformation (An et al., 1989).

Plant material and growth conditions

Leaf tissue of potato cultivar Kuras was transformed as previously described (Andersson et al., 2003), but with 50 mg/L kanamycin as selection agent. Regenerated shoots were analysed for the presence of nptII and the absence of Agrobacterium virG sequence using REDExtract‐N‐Amp Plant PCR Kit (Sigma‐Aldrich, St Louis, MO). Primers; nptIIf 5'‐CCTGTCATCTCACCTTGCTC‐3′, nptIIr 5'‐AGTCCCGCTCAGAAGAACTC‐3', VirG‐ 5′‐GCCGGGGCGAGACCATAGG‐3′ and VirG+ 5′‐ CGCACGCGCAAGGCAACC‐3′(Sigma‐Aldrich).

Cuttings were propagated to produce microtubers (4.4 g/L MS medium, 2.5 mg/L kinetin, 0.5 mg/L abscisic acid (ABA), 8% sucrose and 200 mg/L cefotaxime at 22°C) in dark.

Greenhouse trial was performed from mid of August to mid of December, with 5–10 pots per line, in 7.5 L pots at 16 h day length, 18/15°C day/night temperature, supplementary light intensity of approximately 200 μmol/s/m² photons and 60% relative humidity. Generally, three tubers were pooled to yield one biological replicate. Tubers were stored at 8°C for 4 months and used as seed potato for field trial. About 52 or 60 seed tubers per line were planted and grown in field close to Kristianstad (Skåne, Sweden) between 14th of May and 17th of October. The plants were grown in plots of 15 or 13 plants per row in four rows. The distance between the plants was 32 cm, and the distance between the rows was 75 cm. Line plots harvested after senescence as one sample and size fractionated. Generally, three tubers were pooled to yield one biological replicate.

Determination of copy number by Southern blotting

Genomic DNA was isolated using Illustra DNA Extraction kit Phytopure (GE Healthcare, Buckinghamshire, UK). About 10 μg DNA was digested with EcoRV and separated on a 1% agarose gel and blotted to a Hybond N+ membrane (GE Healthcare). A 498 bp npt II PCR product amplified with primers 5′‐ CCTGTCATCTCACCTTGCTC ‐3′ and 5′‐AGTCCCGCTCAGAAGAACTC ‐3′ and labelled with PCR DIG Probe synthesis kit (Roche Diagnostics GmbH, Mannheim, Germany) was used as probe and hybridized to the membrane o/n. CDP‐Star, ready‐to‐use (Roche Diagnostics GmbH, Mannheim, Germany) was used for detection of the hybridized probe and exposed to an X‐ray film.

Lipid analyses

Tuber samples were freeze‐dried (Coolsafe, Scanvac, Lynge, Denmark). Total lipids were extracted (Bligh and Dyer, 1959), and the CHCl₃ extracts were subsequently separated (aliquots corresponding to 100 mg fw) using thin layer chromatography on TLC Silica gel 60 (Merck, Darmstadt, Germany). For neutral lipids, 70:30:1 heptane:diethylether:HAc was used as a mobile phase and for polar lipids 95:15:10:3 CHCl₃:MeOH:HAc:H₂O. Fatty acid methyl esters were quantified using gas chromatography as previously described (Leonova et al., 2008).

Quantitative and qualitative starch determination

Dry matter was determined after 3 days of freeze drying (Coolsafe, Scanvac, Lynge, Denmark). Homogenized freeze‐dried tuber tissue was used for measurement of total starch content using a Total starch kit (Megazyme, Bray, Co. Wicklow, Ireland). Samples were washed with 80% ethanol to remove glucose present in the sample. For calculation of starch content, a standard curve was used based on a maize starch sample supplied in the kit.

Starch was isolated and purified from fresh tubers (Larsson et al., 1996). Amylose content of isolated starch was measured according to a DMSO and iodine‐based colorimetric analysis method (Andersson et al., 2006).

Starch‐bound lipids were extracted from purified starch using a water‐saturated n‐propanol extraction method (Vasanthan and Hoover, 1992).

Extraction, determination and quantification of metabolites

For the metabolite measurements by GC‐MS analysis, 20 mg of flash‐frozen and lyophilized potato tuber material was ground to a fine powder using a shaking mill and glass beads (5 mm). The polar fraction was extracted and derivatized with 30 μL methoxyamine hydrochloride and 60 μL N‐methyl‐N‐(trimethylsilyl) trifluoroacetamide (MSTFA) as previously described (Bellaire et al., 2014) to transform the metabolites into their methoxyimino (MEOX)‐ and trimethylsilyl (TMS)‐ derivatives. allo‐Inositol was used as an internal standard. The samples were analysed on an Agilent 5973 Network mass selective detector connected to a Agilent 6890 gas chromatograph equipped with a capillary HP5‐MS column (30 m × 0.25 mm; 0.25 μm coating thickness; J&W Scientific, Agilent, Santa Clara, CA). Helium was used as carrier gas (1 mL/min). The inlet temperature was set to 230°C, and the temperature gradient applied was 50°C for 2 min, 50–330°C at 5 K/min 330°C for 2 min. Electron energy of 70 eV, an ion source temperature of 230°C and a transfer line temperature of 330°C were used. Spectra were recorded in the range of 71–600 da/e.

Masses used for quantification of the individual metabolites are depicted in Dataset S1. For absolute quantification, all compounds were quantified against the standard using a calibration curve performed with pure substances based on a total of 12 measurements for each substance.

Structural analysis of tuber tissue

For fixation and plastic embedding, tubers with an approximate weight of 10 g were harvested from pots grown in greenhouse. Tissue samples were taken from six different locations in the tuber, and orientation of the samples in relation to the stolon was noted. The tissue was immediately aldehyde‐fixed, post‐fixed in osmium tetroxide and embedded in fresh Spurr's resin (Low viscosity kit; Ted Pella, Redding, CA) (Turesson et al., 2010).

For transmission electron microscopy, ultrathin sections were cut from Spurr‐embedded blocks and handled as in Turesson et al. (2010). A JEM‐1010 electron microscope (JEOL, Tokyo, Japan) with an accelerating voltage of 60 kV was used to examine the sections.

For staining of light micrographs, an overview was obtained by staining 1‐μm‐thin sections of Spurr‐embedded tuber tissue with MAS (Triple staining methylene blue‐azur A‐safranin O), visualising proteins, lipids and starch (Heneen et al., 2008; Turesson et al., 2010; Warmke and Lee, 1976). Sudan Black was used for staining oil and lipids as described earlier (Heneen et al., 2008; O'Brien and McCully, 1981; Turesson et al., 2010). The stained and air‐dried sections were mounted with Biomount (British Biocell, Cardiff, UK) and studied in a light microscope (Leica Microsystems, Wetzlar, Germany).

For starch staining of homogenized tissue, Lugol's solution (Scharlau, Sentmenat, Spain) and glycerol (1:1) was added to purified starch granules and studied under a light microscope (Leica Microsystems, Wetzlar, Germany).

Transcriptome analysis

Tuber tissue from line 8016 and Kuras was sampled from greenhouse grown plants at three different time points, 2, 3 and 4 months after planting (S2, S3 and S4, respectively), and immediately flash‐frozen in liquid nitrogen. Tissue from three tubers grown in individual pots was pooled prior to total RNA extraction using Plant RNA Reagent (Invitrogen, Life technologies Ltd, Waltham, MA) according to the manufacturer's instructions. This was applied in triplicates as biological replicates for all three stages. Quality and quantity of the RNA samples were measured on a NanoDrop (NanoDrop^™ 1000 Spectrophotometer, Thermo Scientific, Waltham, MA) and using electrophoresis on a 1.2% E‐gel (Invitrogen, Life Technologies Ltd).

Library construction and sequencing using Illumina HiSeq technology was performed by GATC Biotech (Constance, Germany).

All the prepared samples were sequenced through Illumina sequencing platform HiSeq 2000 as unpaired‐end reads with 50‐bp read length. Reads were trimmed by removing adapter sequences, and low quality sequences were removed. Assembly was carried out with high‐quality reads using Trinity version trinityrnaseq_r2014‐07‐17 with default parameters (Haas et al., 2013).

To assess the quality of the transcriptome assembly, reads were mapped back to the assembled transcripts using the Bowtie2 aligner (Langmead and Salzberg, 2012). The mapping results were visualized using Integrated Genomics Viewer version 2.3.2 (IGV) (Thorvaldsdottir et al., 2013). The completeness of the transcriptome assembly was estimated with CEGMA version 2.4.010312 software (Parra et al., 2007). CEGMA analysis was performed on default parameters.

High‐quality‐assembled transcriptome was annotated with BLAST2GO (Conesa et al., 2005). BLASTX was used against the NCBI nonredundant (nr) protein database using an e‐value cut‐off of 1E‐5. InterProScan was used to identify the conserved domains/motif for each transcript.

RSEM version 1.2.15 was used for abundance estimation of the assembled transcripts using the default parameters (Li and Dewey, 2011). The relative measure of transcript abundance was TPM (transcripts per million) and FPKM (fragments per kilobase of transcript per million mapped reads). The relative level of all assembled transcripts can be found in Dataset S4. EdgeR Bioconductor (Robinson et al., 2010) was used in identification and analysis of differentially expressed genes and transcripts through Trinity versiontrinityrnaseq_r20140717 on default settings. Data on all transcripts determined as differentially expressed can be found in Dataset S3.