Increasing the productivity of oilseed crops is an important challenge for plant breeders and biotechnologists. To date, attempts to increase oil production in seeds via metabolic pathway engineering have focused on boosting synthetic capacity. However, in the tissues of many organisms, it is well established that oil levels are determined by both anabolism and catabolism. Indeed, the oil content of rapeseed (Brassica napus L.) has been reported to decline by approximately 10% in the final stage of development, as the seeds desiccate. Here, we show that RNAi suppression of the SUGAR-DEPENDENT1 triacylglycerol lipase gene family during seed development results in up to an 8% gain in oil yield on either a seed, plant or unit area basis in the greenhouse, with very little adverse impact on seed vigour. Suppression of lipolysis could therefore constitute a new method for enhancing oil yield in oilseed crops.
Vegetable oils (triacylglycerols) are an important global commodity. They form a significant part of the nutritional intake of humans and livestock, serve as feedstock for the chemical industry and also provide a renewable source of energy in the form of biofuels (Durrett et al., 2008; Dyer et al., 2008). World production has increased by more than 50% in the past decade and now is in excess of 145 million metric tons per year (http://www.fas.usda.gov/oilseeds/). Improving the yield of oil crops is therefore widely acknowledged to be an important agronomic goal (Lu et al., 2011), and conventional breeding, combined with better crop management, continues to make incremental improvements (Weselake et al., 2009). Alternatively, one transgenic approach to achieve this objective is to boost the metabolic flux of carbon to oil within the developing seed (Weselake et al., 2009). A number of studies have established that seed oil content can be increased by the overexpression of individual oil biosynthetic enzymes (Jako et al., 2001; Roesler et al., 1997; Vigeolas et al., 2007; Zou et al., 1997), or transcriptional ‘master’ regulators that govern the expression of multiple enzymes (Cernac and Benning, 2004; Shen et al., 2010).
Interestingly, in many eukaryotic tissues, the content of oil is governed by the dynamic balance between both synthesis and breakdown and a deficiency in triacylglycerol hydrolysis has been shown to cause an increase in oil deposition (Grönke et al., 2005; Kurat et al., 2006; Zimmermann et al., 2004). The role of oil catabolism in determining the content of oil-rich plant tissues such as those of seeds has received little attention, although there is a literature to suggest that catabolic pathways are active during seed development (Graham, 2008). In Arabidopsis thaliana, Brassica napus, Crambe abyssinica and Nicotiana tabacum, oil content peaks late in the maturation phase, but data in several studies suggest that it then drops as the seed desiccates on the plant (Baud et al., 2002; Gurr et al., 1972; Molina et al., 2008; Murphy and Cummins, 1989; Tomlinson et al., 2004). Chia et al. (2005) investigated this phenomenon in B. napus (oilseed rape) and showed that the amount of oil in seeds falls by more than 10% over the course of desiccation. Oilseed rape is the third most important oil crop, after palm (Elaeis guineensis) and soybean (Glycine max), and supplies more than 23 million metric tons of oil, with an estimated commodity value in excess of $30 billion (http://www.fas.usda.gov/oilseeds). Even a gain in oil yield of approximately 10% would add substantial value to the crop. Furthermore, seed desiccation occurs after the supply of nutrients from the mother plant has effectively ceased (Chia et al., 2005). Because maternal resources are already committed to the seed, prevention of oil loss during desiccation may be considered very likely to enhance the oil yield in the field.
We previously identified a small family of triacylglycerol lipase genes in Arabidopsis, consisting of SUGAR-DEPENDENT1 (SDP1) and SDP1-LIKE (SDP1L), which are responsible for initiating oil breakdown in the seeds following germination (Eastmond, 2006; Kelly et al., 2011). SDP1 and SDP1L are members of an unorthodox group of lipases that are related to patatin from potato (Solanum tuberosum), but contain a divergent active site (Scherer et al., 2010). Well-characterized examples include Homo sapiens adipose triglyceride lipase (Zimmermann et al., 2004), Drosophila melanogaster Brummer (Grönke et al., 2005) and Saccharomyces cerevisiae triacylglycerol lipase 3, 4 and 5 (Athenstaedt and Daum, 2005; Kurat et al., 2006). SDP1 appears to be expressed in all Arabidopsis tissues, but most strongly during seed maturation, prompting us to question whether genes from this family might be responsible for oil turnover in developing seeds, as well as during post-germinative growth (Eastmond, 2006). Here, we investigated this question directly in oilseed rape using a transgenic approach.
Rapeseed oil content declines during seed desiccation
To investigate whether SDP1 triacylglycerol lipase family genes play a role in oil turnover during oilseed rape seed development, we chose to use a commercial open-pollinated spring variety called Kumily (Lantmännen SW Seed AB) as an experimental system. To characterize the pattern of oil accumulation over the course of seed development in this variety, plants were grown under both greenhouse and field conditions. Seeds were harvested from siliques on the primary raceme and their developmental stage was scored morphologically (Figure 1a) using a scale similar to that described previously by Chia et al. (2005). This scale includes seeds at early, early–mid and mid–late phases in oil accumulation, at the onset of desiccation and at maturity. The total fatty acid content of Kumily seeds (as determined by gas chromatography) increased up until the onset of desiccation, when the content reached approximately 2.3 mg of total fatty acids/seed. It then declined significantly by maturity [P < 0.05, least significant difference (LSD) test], equating to a net loss of approximately 10% in the greenhouse and approximately 9% in the field (Figure 1b).
RNAi strategy suppresses SDP1 family during seed desiccation
In order to suppress SDP1 triacylglycerol lipase family function in Kumily seeds, three cDNAs were identified in oilseed rape (GenBank accession numbers GN078290, GN078281 and GN078283), which are >80% identical to Arabidopsis SDP1 (At5g04040) and SDP1L (At3g57140) at the nucleotide level. A very highly conserved 300-bp region of the 3′-open reading frame of GN078283 was then chosen to create an SDP1 family RNA interference (RNAi) construct (Wesley et al., 2001) driven by the USP promoter from Vicia faba, which is active throughout embryo maturation (Bäumlein et al., 1991) (Figure S1). The construct was transformed into Kumily via Agrobacterium-mediated transformation (Moloney et al., 1989), and approximately 30 low copy number primary transformants were selected by quantitative PCR (qPCR; Bubner and Baldwin, 2004). Based on preliminary analysis of seed from the primary transformants (Zank et al., 2008), five lines (24AS, 31AS, 92AS, 62AS and 72AS) were taken to the third generation (T3) and homozygous plants were identified by segregation analysis. SDP1 family transcript levels were measured in desiccating embryos of these plants (Figure 1a) using qPCR, and they were repressed by between 10- and 30-fold relative to either wild type (WT) or a transformed empty vector control (EVC; Figure 2).
SDP1 RNAi enhances seed oil content
To determine whether the oil content of the seeds was affected by SDP1 RNAi, 12 homozygous T3 plants of each genotype were grown in the greenhouse in a randomized block design. Seeds harvested from each plant were then dried and weighed, and their oil and protein contents were assayed using near-infrared reflectance spectroscopy (Hom et al., 2007; Table 1). The seed oil content (as a % of seed weight) was significantly increased in all five SDP1 RNAi lines when compared to WT or EVC (P < 0.001, LSD test), and the gain ranged between 3% and 8%. In contrast, seed protein content was significantly reduced (P < 0.05, LSD test) in all five, but by a comparatively small amount (<4%). The total weight of seeds produced by the plants was taken as a measure of seed yield and did not differ significantly between genotypes (P = 0.928, F-test). However, the oil yield, calculated by multiplying the total weight of seeds from each plant by the % oil content, was significantly different (P < 0.05, LSD test) in three of the SDP1 RNAi lines. In these lines (62AS, 72AS and 92AS), which also had the highest % oil content and the lowest SDP1 transcript levels (Figure 2), there was a gain in oil yield of between 7% and 9%.
Table 1. RNAi-mediated suppression of SDP1 during Kumily seed maturation increases seed oil content per plant
Oil content (%)
Protein content (%)
Seed yield (g/plant)
Oil yield (g/plant)
Protein yield (g/plant)
Values are the mean ± SE of measurements on seeds from individual plants (n = 12) of each genotype grown in the greenhouse.
The superscript letters denote a statistically significant difference from WT (LSD test), where significant differences between genotypes were detected (F-test).
P < 0.05,
P < 0.01,
P < 0.001.
WT, wild type; EVC, empty vector control; AS, independent homozygous T3 SDP1 RNAi line; F, F-statistic on df in subscript; SED, SE of the difference between means on df in subscript; LSD, least significant difference.
To determine whether an oil yield increase could be reproduced on a per unit area basis, small plots of WT and 72AS plants were grown in the greenhouse at a density of 120 plants m2, which simulates field conditions (Table 2). The yield of seeds per m2 plot was not significantly different between WT and 72AS (P = 0.636, F-test). However, the oil yield per m2 was significantly greater (P = 0.022, LSD test), increasing by approximately 9%. The seed yield per unit area that was obtained in the greenhouse (approximately 270 g m2) was within the range obtained for the same variety grown in UK field trials in 2010 (http://www.hgca.com). To investigate when, during seed development, SDP1 RNAi exerts an effect on seed oil content, a time-course experiment was performed (Figure 3). A significant difference in seed total fatty acid content was only detected between WT and SDP1 RNAi lines at maturity (P < 0.05, LSD test), with 72AS and 92AS having 7%–8% greater fatty acid content at this stage. These data suggest that SDP1 RNAi mainly impacts oil turnover during seed desiccation, although a small effect in the earlier phase of maturation cannot be discounted (Figure 3). The WT (Kumily) is a ‘canola’ or ‘00’ variety with characteristically high oleic acid and very low erucic acid contents in its seed oil. The analysis of fatty acid composition using gas chromatography confirmed that SDP1 RNAi had no effect on this trait in mature seeds (Table S1), and no obvious differences in fatty acid composition were detected throughout seed maturation.
Table 2. RNAi-mediated suppression of SDP1 during Kumily seed maturation increases oil yield per unit area
Oil content (%)
Protein content (%)
Seed yield (g m2)
Oil yield (g m2)
Protein yield (g m2)
Values are the mean ± SE of measurements on seeds from 1-m2 plots (n = 6) of each genotype grown in the greenhouse.
The superscript letters denote a statistically significant difference from WT (LSD test), where significant differences between genotypes were detected (F-test).
P < 0.05,
P < 0.01,
P < 0.001.
WT, wild type; AS, independent homozygous T3 SDP1 RNAi line; F, F-statistic on df in subscript; SED, SE of the difference between means on df in subscript.
Using SDP1 RNAi as a strategy to boost oil yield is beneficial so long as seed vigour is not compromised, because slow and uneven seedling establishment is known to adversely affect crop yield (Finch-Savage et al., 2010). SDP1 is necessary to support normal seedling establishment in Arabidopsis, although growth inhibition is only apparent when oil breakdown is quite substantially restricted (Eastmond, 2006; Kelly et al., 2011). The RNAi strategy was designed to suppress SDP1 family gene function during seed maturation, but not following germination. The USP promoter is highly active throughout embryo maturation (Bäumlein et al., 1991). However, like many other ‘seed-specific’ promoters, a low level of activity has been observed in some other tissues (Bäumlein et al., 1991; Zakharov et al., 2004). The analysis of total fatty acid breakdown following seed germination suggested a very slight initial delay in SDP1 RNAi lines versus WT (P < 0.05, LSD test), but no significant difference in peak rate between genotypes (P = 0.892, F-test; Table 3). Studies on seed vigour in small-seeded crop species have consistently identified germination rate and initial root and shoot growth rate to be good predictors of performance in the field (Finch-Savage et al., 2010). No significant difference (P < 0.05, F-tests) was observed between WT and SDP1 RNAi lines for any of these three traits when they were assayed under laboratory conditions (Table 3). Finally, homozygous T3 SDP1 RNAi seed batches were stored at room temperature and humidity for up to 4 years and only those that were more than 2 years old exhibited a significant reduction in germination rate versus WT (P < 0.05, LSD test), and this difference remained comparatively small (Table 4).
Table 3. Suppression of SDP1 during Kumily seed maturation does not impair the rate of fatty acid breakdown (FAB) following germination or seed vigour
Rate of FAB (μg/day)
FAB at 24 h (%)
Shoot growth (mm/h)
Root growth (mm/h)
Seed vigour assays were performed as described by Finch-Savage et al. (2010), and the values are the mean ± SE on measurements taken on ten seed from six plants of each genotype (n = 6). Measurements of total fatty acid contents were also taken on pools of ten seeds from six plants of each genotype (n = 6), and the percentage of fatty acid breakdown (FAB) after 24 h and the peak rate of FAB between 48 and 72 h were determined.
Table 4. Suppression of SDP1 during Kumily seed maturation only affects the germination rate after prolonged storage
Time to 50% germination (h)
1 year old
2 years old
3 years old
4 years old
Germination assays were performed on seed batches harvested between 1 and 4 years ago using the method described by Finch-Savage et al. (2010), and the values are the mean ± SE on measurements taken on 60 seed from six plants of each genotype (n = 6).
Here, we have shown that suppressing oil turnover during rapeseed maturation using an SDP1 RNAi strategy can result in up to a approximately 8% increase in oil yield on a per seed, per plant or a per unit planted area basis, in the greenhouse, with relatively little adverse impact on seed vigour. Multisite field trials would ultimately be necessary to fully validate the commercial potential of this approach, including assessing the impact of certain agricultural practices such as swathing and desiccant application. However, our data suggest that this technology is likely to deliver small, but consistent, oil yield improvements in the field. SDP1 RNAi primarily blocks oil loss during seed desiccation. This loss occurs in greenhouse and field conditions and takes place after the nutrient supply from the mother plant has effectively ceased (Baud et al., 2002; Chia et al., 2005).
The increase in percentage oil content observed in SDP1 RNAi seeds is accompanied by a decrease in percentage protein content, although this decrease is not quantitatively proportional (Table 1). The protein content of rapeseed is important because the protein meal, generated as a by-product of oil extraction, is sold as feed for livestock. However, oil is by far the most valuable constituent of the seed and has a commodity value that is more than 4.5 times greater than the meal (http://www.fas.usda.gov/oilseeds/). In many countries, farmers receive a bonus at market, based on the percentage oil content of their crop. For example, if the results we present in Table 2 translated to the field, then a farmer in the UK would gain an additional approximately 4.5% of the contract price in oil bonus, with no deduction for the marginal difference in the protein content of the seed meal (http://www.hgca.com). This equates to a gain of approximately £17/ton of seed at current market prices.
It is important to consider that SDP1 RNAi might also be compatible with other transgenic approaches that are designed to increase oil biosynthetic capacity in the seed, such as overexpression of oil biosynthetic enzymes (Jako et al., 2001; Roesler et al., 1997; Zou et al., 1997) or their transcriptional activators (Cernac and Benning, 2004; Shen et al., 2010). In this case, an additive effect on oil yield might be achieved. Several such approaches to increase biosynthetic capacity have already been reported to increase rapeseed oil content by between 5% and 20% per seed, depending on the mechanism and study (Weselake et al., 2009). Although there have only been a few cases where it has been demonstrated that this translates into an oil yield increase per unit area, as would be required in the field (Weselake et al., 2009).
The precise role of oil turnover during rapeseed desiccation is not fully understood. It may simply result from some overlap in the developmental programmes that govern late seed maturation and post-germinative seedling development. Alternatively, Chia et al. (2005) proposed that it provides a carbon source to support metabolic activity during desiccation, promoting continued synthesis of seed storage proteins and late embryogenesis–abundant proteins. Radiolabel feeding experiments do indicate that the oil is used to fuel respiration, and to a lesser extent amino acid synthesis, but not gluconeogenesis (Chia et al., 2005). Although our analysis of SDP1 RNAi lines suggests that the physiological consequences of impairing oil breakdown in developing seeds are only slight, we did observe a small reduction in seed longevity.
Given that SDP1 RNAi seeds appear to gain oil content partly at the expense of protein, it is possible that a metabolic readjustment occurs in the desiccating seeds, whereby protein and amino acids provide a carbon source to compensate for the relative inaccessibility of oil. There is already evidence that this occurs during early post-germinative growth in Arabidopsis mutants that are deficient in the glyoxylate cycle (Cornah et al., 2004). Finally, there is some evidence to suggest that oil turnover may actually be a widespread occurrence in desiccating seeds from various species (Chia et al., 2005), and therefore, the suppression of SDP1 family genes might have a broader application in other oilseed crops.
Plant growth conditions and experimental design
For the analysis of seed from greenhouse-grown plants, wild type and transgenic Brassica napus (L.) cv. Kumily were planted either individually in 3-L pots or in plots of approximately 120 in 1 × 1 × 0.3 m trays containing Scotts Levington M2 compost and grown to seed. Two neighbouring 20-m2 greenhouse compartments with air conditioning (18 °C day/12 °C night) and 16 h of supplemental illumination (approximately 300 μmol/m2/s at canopy height) were used. Individual plants were arranged into a 12-block, one-way randomized design with one plant of each genotype per block and randomized within each block. Six blocks were placed in each compartment. Plots of each genotype were arranged alternately with six in each compartment. The position of blocks and plots was rotated within each compartment every week. For the analysis of seed from field-grown plants, the material was harvested from within a 1-m2 area. For all the experiments performed on developing seeds, the material was harvested from siliques situated around the middle of the primary raceme. Seed vigour assays were performed as described by Finch-Savage et al. (2010). Seeds were applied to vertically orientated wet filter paper and placed in the dark at 15 °C, and time to germination, 15 mm shoot length and 40 mm root length were scored visually. There were ten seeds per replicate and six replicates per genotype. The initial rate of shoot and root growth was calculated by subtracting the time to germination. Seeds/seedlings were also harvested at 24-h periods to measure the total fatty acid content and determine the peak rate of fatty acid breakdown.
Construct preparation and transformation
In order to suppress storage oil hydrolysis in developing Kumily embryos, an RNAi construct was generated (Zank et al., 2008). Three cDNAs were identified in B. napus (GenBank accession numbers GN078290, GN078281 and GN078283) that are >80% identical to Arabidopsis SDP1 (At5g04040) and SDP1L (At3g57140) at the nucleotide level, and a highly conserved 300-bp region of the 3′-open reading frame of GN078283 was selected for RNAi (Figure S1a). This region was fused in direct and reverse-complement orientation to the ends of a linker sequence from Physcomitrella patens. The RNAi construct was then fused with the USP promoter from Vicia faba, which drives seed-specific expression (Bäumlein et al., 1991), and with the OCS terminator from Agrobacterium tumefaciens. The expression construct was cloned into the vector pENTR-A and used in a Gateway® reaction with an empty pENTR-B and pENTR-C and a pSUN2-based destination vector (Cheng et al., 2010) to create the binary plant transformation vector (Figure S1b) B. napus cv. Kumily was transformed using the method described by Moloney et al. (1989) except that imidazolinone resistance was used for positive selection in tissue culture. Low copy number lines were identified by performing quantitative PCR on genomic DNA (Bubner and Baldwin, 2004) using primers to the RNAi construct, as described below for transcript measurements. An empty vector control (EVC) line with a single T-DNA insertion (confirmed by Southern blot) was used as a reference.
DNase-treated total RNA was isolated from B. napus embryos using the RNeasy kit from Qiagen with the modifications described by Eastmond (2006). The synthesis of single-stranded cDNA was performed using SuperScript II RNase H-reverse transcriptase from Invitrogen. Quantitative PCR was performed with an ICycler (Bio-Rad) using qPCR Mastermix Plus (Eurogentec), according to the manufacturer's instructions, and the data were analysed with ICycler IQ5 software. The primer pairs were SDP1 (5′-GTCCTCTTCTGCAAATCAATGCT-3′ and 5′-GAGGAACCAGTGGAGGAGGAA-3′), RNAi (5′-CCATCTCCCTCTCAGGTATATCTATCTG-3′ and 5′-CATAGGCGTCTCGCATATCTCA-3′) and ACTIN2 (5′-ACGAGCTACCTGACGGACAAG-3′ and 5′-GAGCGACGGCTGGAAGAGTA-3′).
A Buchi NIRFlex N-500 Solids Near-Infrared Reflectance Spectrometer (NIRS) with vial add-on (BUCHI UK Ltd, Oldham, UK) was used for seed oil and protein determination in conjunction with the INGOT Rapeseed calibration for vial format (Aunir Ltd, Towcester, UK). For each individual plant, three 15 × 45 mm glass vials of seeds were measured and the mean values used for subsequent calculations. In the case of plots, twenty glass vials of seeds were measured and the mean values used for subsequent calculations. The accuracy of the NIRS method for determining B. napus seed oil content was supported by a strong correlation (r > 0.95) with data obtained using an MQC23 bench top NMR machine (Oxford Instruments, Oxford, UK). For the determination of fatty acid content and composition, gas chromatographic analysis of fatty acid methyl esters was carried out using the method described by Barker et al. (2007) with the following modifications. For each sample, ten or twenty seeds or seedlings were frozen in liquid nitrogen and freeze-dried, before being homogenized and incubated together with 5 mg of tripentadecanoin standard in 5 mL of 1 N methanolic HCl at 85 °C for 12 h.
We wish to thank BASF Plant Science GmbH (Limburgerhof, Germany) for cloning SDP1 homologues in B. napus, performing transformation and providing the T1 transgenic material used in this study. We are also grateful to the Horticultural Services staff at the University of Warwick for plant husbandry and Dr Christian Craddock and Dr Nicolette Adams for their assistance in harvesting and threshing seeds. This work was supported by the UK Biotechnology and Biological Sciences Research Council through grant BB/E022197/1. PJE is the named inventor for patent US8093452.