Early seed development requires the A-type ATP-binding cassette protein ABCA10

Abstract

A-type ATP-binding cassette (ABCA) proteins transport lipids and lipid-based molecules in humans, and their malfunction is associated with various inherited diseases. Although plant genomes encode many ABCA transporters, their molecular and physiological functions remain largely unknown. Seeds are rapidly developing organs that rely on the biosynthesis and transport of large quantities of lipids to generate new membranes and storage lipids. In this study, we characterized the Arabidopsis (Arabidopsis thaliana) ABCA10 transporter, which is selectively expressed in female gametophytes and early developing seeds. By 3 d after flowering (DAF), seeds from the abca10 loss-of-function mutant exhibited a smaller chalazal endosperm than those of the wild-type. By 4 DAF, their endosperm nuclei occupied a smaller area than those of the wild-type. The endosperm nuclei of the mutants also failed to distribute evenly inside the seed coat and stayed aggregated instead, possibly due to inadequate expansion of abca10 endosperm. This endosperm defect might have retarded abca10 embryo development. At 7 DAF, a substantial portion of abca10 embryos remained at the globular or earlier developmental stages, whereas wild-type embryos were at the torpedo or later stages. ABCA10 is likely involved in lipid metabolism, as ABCA10 overexpression induced the overaccumulation of triacylglycerol but did not change the carbohydrate or protein contents in seeds. In agreement, ABCA10 localized to the endoplasmic reticulum (ER), the major site of lipid biosynthesis. Our results reveal that ABCA10 plays an essential role in early seed development, possibly by transporting substrates for lipid metabolism to the ER.

Mutation in an A-type ABC transporter gene causes defects in endosperm development in young growing seeds, and its overexpression increases seed lipid content.

Introduction

ATP-binding cassette (ABC) proteins are ubiquitous across all living organisms, and their structures and functions are highly conserved (Dean et al., 2001; Kang et al., 2011; Hwang et al., 2016). Many ABCs are transporters for lipids, sterols, organic acids, ions, metals, and metalloids. ABC proteins can be classified into nine subfamilies based on their sequence similarity. In humans, members of the A-type ABC (ABCA) subfamily transport hydrophobic lipidic compounds, such as phospholipids and cholesterol. Mutations in several human ABCA genes result in lipid metabolism defects and are associated with severe genetic diseases (Stefkova et al., 2004; Tarling et al., 2013).

Plant genomes harbor many ABCA genes. However, the physiological and biochemical functions of most plant ABCA proteins remain uncharacterized, with the exception of Arabidopsis (Arabidopsis thaliana) ABCA9, which may facilitate the biosynthesis of triacylglycerol (TAG), possibly by transporting free fatty acids or acyl-Coenzyme A (acyl-CoA) to the endoplasmic reticulum (ER), the site of TAG biosynthesis, during the seed filling stage (Kim et al., 2013). The genomes of oilseed plants tend to encode more ABCA transporters than do those of plants that mainly store sugars or starch in their seeds (Yan et al., 2017). For example, oilseed plants such as Arabidopsis, lyre-leaved rock cress (Arabidopsis lyrata), and field mustard (Brassica rapa) harbor 11–12 ABCA genes in their genomes, whereas grape (Vitis vinifera) and rice (Oryza sativa) have only 5–6 ABCA genes (Yan et al., 2017). The abundance of ABCA genes in oilseed plants, therefore, suggests a potential function in lipid metabolism or accumulation.

The seed is an ideal organ to study the function of a potential lipid transporter, since it develops rapidly and stores energy in the form of lipids, placing a high demand on the lipid biosynthesis and transport machineries. Developing seeds carry out extensive lipid biosynthesis during the early stages of development to support the expansion of lipid membranes in these rapidly dividing and growing cells. At the later stages of seed development, seeds require high levels of lipid metabolism for the biosynthesis of storage lipids and to form the seed coat. Thus far, most studies on lipid metabolism and transport in embryos and endosperms have focused on the accumulation of storage lipids or unusual lipids, with the aim of increasing the production of economically valuable oils (Zhang et al., 2009; An and Suh, 2015; Tian et al., 2018), or forming protective surface coatings (Beisson et al., 2007; Coen et al., 2019).

Recent studies highlighted the importance of lipid metabolism in early embryo and endosperm development. During early embryonic development, loss of function of Δ9 stearoyl-ACP desaturases, a choline/ethanolamine kinase, or acyl-CoA-binding proteins arrests embryo development, resulting in embryonic death in Arabidopsis (Chen et al., 2010; Lin et al., 2015; Kazaz et al., 2020). During endosperm development in black nightshade (Solanum nigrum), lipid bodies form inside the endosperm cells with spatial and temporal distribution that suggests that they nourish the developing embryo (Briggs, 1993). Developing endosperms of many oleaginous species accumulate lipids with fatty acid compositions different from those of the respective embryos (Miray et al., 2021). The Arabidopsis MYB118 transcription factor is critical for the biosynthesis of omega-7 monounsaturated fatty acids in the endosperm (Troncoso-Ponce et al., 2016) and partitioning of reserves between embryo and endosperm (Barthole et al., 2014). However, most reports on lipids in endosperm have focused only on the role of lipids in fueling seed germination and postgerminative seedling establishment, and it remains poorly understood how lipid metabolism contributes to endosperm development.

In this study, we aimed to characterize the functions of Arabidopsis ABCA10, which is selectively expressed in female gametophytes and the endosperm of early developing seeds. We revealed that loss of ABCA10 causes developmental defects in the young endosperm, such as a diminished chalazal endosperm cyst and a shrunken syncytial endosperm, which are apparent as early as 3 d after flowering (DAF). Loss of ABCA10 also caused a delay in embryo growth, leading to small and shrunken mature seeds with a more drastic reduction of TAG content compared to other seed reserves. In contrast, overexpression of ABCA10 induced excess accumulation of TAG without altering seed protein or carbohydrate contents, suggesting that ABCA10 is involved in lipid metabolism. Thus, we identified a factor important for early seed development and characterized the physiological function of a plant ABCA transporter.

Results

Loss of ABCA10 function results in small and shrunken seeds

An exploration of publicly available Arabidopsis transcriptome deep sequencing (RNA-seq) datasets indicated that ABCA9 and ABCA10 are highly expressed in ovules and young developing seeds (Klepikova et al., 2016; Supplemental Figure S1). ABCA9 is a putative fatty acid transporter critical for the accumulation of seed storage lipids (Kim et al., 2013). To determine the function of ABCA10 during seed development, we examined dry seeds harvested from two independent loss-of-function mutants for ABCA10, abca10-1 (SAIL_869_G08) and abca10-2 (FLAG_106D07) (Supplemental Figure S2, A and B). ABCA10 transcripts were barely detectable in mutant siliques but were clearly detected in siliques from their respective wild-types (Supplemental Figure S2, C and D). A substantial portion of abca10-1 and abca10-2 seeds was small and/or shrunken (Figure 1A); 69% of abca10-1 and 20% of abca10-2 seeds looked abnormal, in contrast to only 2% of seeds from either wild-type parent (Figure 1B). Consistent with these observations, the mean dry weights of abca10-1 and abca10-2 seeds were 68 and 89% of the wild-type values, respectively (Figure 1C). Since abca10-1 displayed more robust phenotypes than abca10-2, we mainly used abca10-1 for further analyses.

Figure 1.

Loss of ABCA10 function results in defects in seed development. A, Dry seed morphology of wild-type (Col-0 and Ws), abca10 mutants and two independent abca10-1 complemented lines (C1 and C2). Black arrows indicate small and/or shrunken seeds. Bars, 200 μm. B and C, Abnormal seed ratios (B) and weights (C) of dry seeds. Morphology and weight of 300 seeds in each line were analyzed. Error bars represent se from three biological replicates. **Significant difference of abca10 from the wild-types at P < 0.01 using Student’s t test. D–F, Contents of TAGs (D), proteins (E) and carbohydrates (sucrose + starch) (F) per dry seed. Lipids were extracted and analyzed from 300 seeds, and proteins and carbohydrates from 40 seeds. Error bars represent se from three biological replicates. Statistical analyses were carried out using one-way ANOVA with Tukey’s post-hoc test with different alphabets indicating statistically significant difference at P < 0.05.

To test whether the defects in abca10 seeds were caused by the loss of ABCA10 function, we introduced a genomic fragment encompassing the ABCA10 locus conjugated with the β-GLUCURONIDASE (GUS) coding sequence (proABCA10:ABCA10g-GUS) into abca10-1 in the Col-0 accession for complementation tests. In two independently transformed lines (C1 and C2), ABCA10 transcript levels returned to those of the wild-type (Col-0) (Supplemental Figure S2C), and seed morphology and dry weights were similar to those of the wild-type (Figure 1, A–C), indicating that the disruption of ABCA10 by the T-DNA insertion is responsible for the small and shrunken seed phenotypes observed in abca10-1. The vegetative growth of abca10-1 was comparable to that of the wild-type (Supplemental Figure S2E), consistent with the specific expression of ABCA10 in ovules and developing seeds (Supplemental Figure S1). Together, these results suggest that ABCA10 is critical for normal seed development.

Loss of ABCA10 decreases accumulation of seed storage compounds

We then examined whether the loss of ABCA10 altered the accumulation of seed storage lipids. Content of the main seed storage lipid, TAG, decreased by 56% per seed in abca10-1 compared to the wild-type Col-0 (Figure 1D and Table 1). Total fatty acid (tFA) content similarly dropped in abca10-1, resulting in a similar TAG:tFA ratio in the mutant and the wild-type (Supplemental Figure S3, A and B). The profile of fatty acids in TAGs was also altered by the loss of ABCA10 (Supplemental Figure S3C), with the relative mole percentage of linolenic acid (18:3) and oleic acid (18:1) decreasing and that of linoleic acid (18:2) increasing in abca10-1 compared to Col-0.

Table 1.

Summary of weights and contents of TAGs, proteins, and carbohydrates (sucrose + starch) of seeds from wild-type (Col-0), abca10-1 and two complemented lines

Parameter	Col-0	abca10-1	C1	C2
Seed weight (μg/seed)	15.3 ± 0.32	10.4 ± 0.20**	15.5 ± 0.35	15.0 ± 0.51
TAG weight (μg/seed)	4.14 ± 0.14	1.88 ± 0.13**	4.82 ± 0.21	4.39 ± 0.20
Protein weight (μg/seed)	4.88 ± 0.17	3.38 ± 0.12**	5.38 ± 0.19	5.33 ± 0.13
Carbohydrate weight (ng/seed)	151.2 ± 9.44	113.6 ± 7.94**	194.0 ± 8.35	191.7 ± 8.46
TAG, % of dry weight	30.4 ± 2.56	20.0 ± 1.54*	34.6 ± 1.24	32.6 ± 0.75
Protein, % of dry weight	32.1 ± 2.77	32.4 ± 2.29	34.8 ± 1.92	35.8 ± 3.07
Carbohydrate, % of dry weight	1.00 ± 0.10	1.09 ± 0.09	1.26 ± 0.12	1.29 ± 0.08

Values represent average ± standard error (se) from three biological replicates.

^* P < 0.05,

^** P < 0.01 using Student’s t test comparing the mutant to the wild-type.

In addition to lipids, abca10-1 seeds contained lower levels of proteins and carbohydrates (including sucrose and starch); the mean protein and carbohydrate contents per seed were 31% and 25% lower, respectively, in abca10-1 compared to the wild-type (Figure 1, E and F; Table 1). These results suggest that ABCA10 is involved in fundamental aspects of seed development, rather than specifically in the accumulation of seed storage lipids, as suggested for ABCA9 (Kim et al., 2013). Seeds from the complemented lines accumulated reserves to levels comparable to those of wild-type seeds, as expected (Figure 1, D–F; Supplemental Figure S3, A–C; Table 1). Taken together, these results indicate that ABCA10 is necessary for the normal accumulation of the major seed reserves: lipids, proteins, and carbohydrates.

The abca10-1 mutant exhibits defects at very early stages of seed development

We next investigated at which stages of seed development ABCA10 was expressed. ABCA10 transcript levels were highest in open flowers (0 DAF) and siliques at 1–3 DAF (Figure 2A) and gradually decreased as seed development progressed. This result was in agreement with the transcriptome data (Supplemental Figure S1).

Figure 2.

ABCA10 is specifically expressed in early developing seeds. A, Relative ABCA10 transcript levels during seed development in wild-type (Col-0). Error bars represent se from three biological replicates. Statistical analysis was carried out using one-way ANOVA with Tukey’s post-hoc test with different alphabets indicating statistically significant difference at P < 0.05. B, Developing seeds for the wild-type (Col-0) and abca10-1 throughout seed development. Black arrows indicate morphologically defective small and/or whitish seeds. Bars, 1 mm. C–H, GUS staining pattern in abca10-1 complemented with proABCA10:ABCA10-GUS in the embryo sac of ovules and the endosperm of early developing seeds. Bars, 50 μm.

We then assessed the morphology of developing seeds within the siliques of the wild-type Col-0 and the abca10-1 mutant throughout seed development (4–17 DAF; Figure 2B). As early as 4 DAF, abca10-1 siliques contained many seeds that looked small and/or slender compared to the wild-type seeds; the average width and length of the abca10-1 seeds were 19% and 12% smaller than those of the wild-type, respectively (n = 214 Col-0; n = 218 abca10-1). At 7 DAF and 10 DAF, many seeds of abca10-1 were whitish and smaller than wild-type seeds; at 7 DAF, as many as 61% of seeds (n = 399) in siliques of abca10-1 were whitish and/or small, whereas 91% of seeds (n = 393) in wild-type siliques were green and even in size. By 14 and 17 DAF, many seeds in abca10-1 siliques appeared shrunken and/or darker than the wild-type seeds. These results suggest that ABCA10 is required for the early stages of seed development.

ABCA10 is expressed in ovules and early developing seeds

To further explore the function of ABCA10 during early seed development, we examined the spatial expression pattern of ABCA10 in complementation lines of abca10-1 that express ABCA10-GUS under its own promoter (Figures 1 and 2). We detected GUS signal only in ovules of open flowers and seeds at early developmental stages (Figure 2, D–H), but not in anthers or pollens (Figure 2C). In 0 DAF ovules, the GUS signal was concentrated in the female gametophyte, with a stronger signal toward the micropylar region of the embryo sac (Figure 2D). In 1 DAF ovules, the signal was further intensified in the central cell (Figure 2E), and by 2 DAF, GUS staining covered the entire young seed, though with much stronger intensity in the chalazal regions than we observed at 1 DAF (Figure 2F). From 3 DAF on, the signal remained specifically in the chalazal endosperm and became weaker over time (Figure 2, G and H). These spatiotemporal expression patterns of ABCA10 are consistent with those from recently published transcriptomic studies of ovules and developing seeds (Supplemental Figure S4; Le et al., 2010; Belmonte et al., 2013; Song et al., 2020). Thus, the data from our study and other recent studies collectively suggest that ABCA10 functions in the female gametophyte and early developing seeds, particularly in the endosperm.

ABCA10 is not essential for fertilization

The expression of ABCA10 in ovules suggested the possibility that this gene might be required for fertilization. To test this hypothesis, we counted seed sets in the wild-type Col-0 and abca10-1 mutant. Seed number per silique (Figure 3A) and silique length (Figure 3B) were not significantly different between the wild-type and abca10-1, although many abca10-1 seeds looked paler and smaller than their wild-type counterparts (Figure 3C). These results indicate that the loss of ABCA10 does not cause a defect in fertilization of the female gametophyte.

Figure 3.

abca10-1 does not affect seed setting. A and B, Comparison of seed number per silique (A) and silique length (B) of wild-type (Col-0) and abca10-1 at 10 DAF. Error bars represent standard deviation (sd, n = 10) and ns indicates no significant difference, as determined by Student’s t test. C, Representative seed sets. Bars, 1 mm.

ABCA10 is required for endosperm development

Since ABCA10 is expressed in early developing endosperm, we examined whether endosperm development was altered by the loss of ABCA10. At 3 DAF, at which time the chalazal endosperm undergoes substantial differentiation to yield various cellular organelles (Otegui et al., 2002), the chalazal endosperm of abca10-1 seeds was smaller than that of wild-type seeds (Figure 4A). The mean area of abca10-1 chalazal endosperms, measured from the median-sectioned confocal images of seeds, was 26% smaller than that of the wild-type (Figure 4B), suggesting that ABCA10 is necessary for normal chalazal endosperm development.

Figure 4.

abca10-1 mutant exhibits defects in endosperm development. A, Representative confocal images of wild-type (Col-0) and abca10-1 seeds at 3 DAF. The red signal is the autofluorescence of seeds. The chalazal endosperms are enclosed by yellow dotted lines (fluorescent image) or indicated by black arrows (brightfield image). The image was acquired by focusing the region of the chalazal endosperm and embryo from the whole developing seed and increasing the contrast to accurately observe these regions and reducing the signal from the background. Bars, 20 μm. C1 and C2, abca10-1 complemented lines. B, Mean area of chalazal endosperm of 3 DAF seeds (each n = 41). Error bars represent sd. **P < 0.01 using Student’s t test. C, Representative confocal images of seeds at 4 DAF for wild-type (Col-0), abca10-1 and complemented lines. The red signal is the autofluorescence emitted from endosperm nuclei and seed coats. The yellow dotted lines indicate the endosperm boundary, and the magenta dashed lines indicate the border of the seed cavity. Bars, 50 μm. D, Percentage of seeds at 4 DAF with a normal pattern of endosperm nuclei. Error bars represent se from three biological replicates (n = 26–33 in each replicate). E, Mean area occupied by endosperm nuclei in 4 DAF seeds. Error bars represent sd (n = 85–89). F, Glucose contents per developing seed in Col-0 and abca10-1. Error bars represent se from four biological replicates. D–F, Statistical analyses were carried out using one-way ANOVA with Tukey’s post-hoc test with different alphabets indicating statistically significant difference at P < 0.05.

Next, we characterized the morphology of the peripheral endosperm using the autofluorescence emitted from endosperm nuclei (Garcia et al., 2003). At 4 DAF, when the peripheral endosperm is still syncytial and its nuclei are evenly distributed (Brown et al., 1999), the majority of the endosperms in wild-type seeds (∼89%) looked plump, with an even distribution of syncytial nuclei (Figure 4, C and D). In contrast, only 12% of 4 DAF abca10-1 seeds were similar to those of the wild-type in their endosperm nuclei distribution, with the remaining seeds exhibiting shrunken, skewed, and distorted endosperms. In many of the abca10-1 seeds with shrunken endosperms, we noticed a gap between the endosperm and inner border of the seed coat (Figure 4C, merged image). The mean endosperm area occupied by nuclei was 56% smaller in abca10-1 mutant seeds than in the wild-type (Figure 4E). The endosperm morphology of the complemented lines was similar to that of the wild–type (Figure 4, C–E). The peripheral endosperm from the second allele, abca10-2, also appeared separated from the seed coat (Supplemental Figure S5), similar to abca10-1. These results revealed that abca10 endosperm cannot expand to fill the space enclosed by the seed coat at 4 DAF.

In Brassicaceae, the early seed size increase is temporally associated with an increase in hexose content (Baud et al., 2002; Morley-Smith et al., 2008). The hexose accumulated in the central endosperm vacuole may act as an osmolyte, with osmotic pressure contributing to the expansion of the coenocytic endosperm (Beauzamy et al., 2016). To examine whether abca10 alters the accumulation of hexoses during early seed development, we quantified glucose levels in developing seeds at 3 and 4 DAF, at which stage hexose contents start to increase rapidly (Baud et al., 2002). Wild-type seeds experienced a 1.7-fold increase in glucose content from 3 to 4 DAF (Figure 4F). However, abca10-1 seeds experienced only a modest and nonsignificant rise in glucose content over the same time period, leading to abca10-1 seeds with 45% less glucose per seed compared to wild-type seeds at 4 DAF. These data suggest that accumulation of hexoses compromised in abca10 seeds at the stage of coenocytic endosperm expansion.

ABCA10 is required for embryo development

Next, we examined the development of abca10-1 embryos in 7 DAF seeds. At this age, embryos typically grow rapidly and reach the torpedo or linear cotyledon stages (Belmonte et al., 2013). In 7 DAF siliques, >95% of wild-type embryos, but only 14% of abca10-1 embryos, were in the torpedo or linear cotyledon stages (Figure 5). About 44% of abca10-1 embryos remained in the early heart stage and 17% of embryos were still at the globular or earlier stages. The complemented lines showed the same progression of embryo development as the wild-type, with about 90% of seeds containing torpedo or later stage embryos. This analysis indicates that ABCA10 is required for proper embryo development.

Figure 5.

The abca10-1 mutant exhibits a delay in embryo development. Distribution of embryo stages in 7 DAF siliques of wild-type (Col-0), abca10-1 and the complemented lines (C1 and C2). Ten siliques from each genotype were analyzed in each replicate. Error bars represent se from three biological replicates. **Significant difference of abca10-1 from Col-0 at P < 0.01 using Student’s t test.

ABCA10 localizes to the ER membrane

We next investigated the subcellular localization of ABCA10 using transgenic plants accumulating a fusion between ABCA10 and superfolder green fluorescent protein (ABCA10-sGFP). ABCA10-sGFP seemed to be functionally comparable to endogenous ABCA10, as ABCA10-sGFP transgene expression driven by the ABCA10 promoter fully complemented the mature seed phenotype of the abca10-1 mutant (Supplemental Figure S6, A–C).

In root cells of Arabidopsis plants that co-express ABCA10-sGFP and the ER reporter Nodulin 26-like Intrinsic membrane Protein 1;1 (NIP1;1) fused to mCherry (Geldner et al., 2009) under the control of the 35S promoter, sGFP fluorescence co-localized with that of mCherry in cytosolic net-like structures (Figure 6A), indicating that ABCA10 is an ER protein. In Nicotiana benthamiana leaf epidermal cells transiently co-expressing ABCA10-sGFP and another ER marker, BINDING PROTEIN1 (BiP1), fused to tagRFP (Benghezal et al., 2000), sGFP and tagRFP fluorescence also co-localized in net-like structures (Figure 6B), supporting this conclusion. These observations indicate that ABCA10 functions at the ER membrane.

Figure 6.

ABCA10 localizes to the ER. A, Representative confocal images of primary root cells of transgenic Arabidopsis seedlings co-expressing pro35S:ABCA10-sGFP and the ER marker construct proUBQ10:NIP1;1-mCherry. Bar, 10 μm. B, Representative confocal images of N. benthamiana cells co-expressing pro35S:ABCA10-sGFP and the ER marker construct pro35S:BiP1-tagRFP. Bar, 20 μm.

Overexpression of ABCA10 increases total seed oil yield per plant

Next, we tested whether ABCA10 overexpression increased seed oil yield, as ABCA9 overexpression does (Kim et al., 2013). We generated six independent pro35S:ABCA10-sGFP lines. In these lines, the transgene expression was at least 28-fold higher than the native gene expression in 1 DAF siliques (Figure 7A). ABCA10-sGFP expression was also substantial in embryos at the later seed maturation phase, when the native gene was barely expressed (Figure 2A; Supplemental Figure S7).

Figure 7.

Overexpression of ABCA10-sGFP increases lipid yield but not the contents of proteins or carbohydrates. A, Relative ABCA10 transcript levels in 1 d after flowering siliques of wild-type (Col-0) and ABCA10-sGFP overexpression lines (OX) . Mean values (±se) from three biological replicates were normalized against that of wild-type. B, Mean dry seed weight of each line. Error bars represent se from four biological replicates. C, Mean surface area quantified from dry seed images. Error bars represent se from three biological replicates with 20 seeds analyzed per replicate. D–F, Contents of TAGs (D), proteins (E), and carbohydrates (sucrose + starch) (F) per dry seed. Lipids were extracted and analyzed from 300 seeds and proteins and carbohydrates from 40 seeds. Error bars represent se from four biological replicates. **P < 0.01, *P < 0.05 as determined by Student’s t test. ns indicates no significant difference between Col-0 and overexpression lines.

ABCA10 overexpression lines produced similar numbers of siliques per plant and seeds per silique as the wild-type (Supplemental Figure S8, A and B). However, the mean seed dry weight rose by up to 25% in five of the six ABCA10 overexpression lines (Figure 7B) and the mean seed surface area rose by up to 29% in the same five overexpression lines (Figure 7C). Likewise, mean TAG contents per seed increased in the five ABCA10 overexpression lines by up to 47% compared to that of the wild-type (Figure 7D), without any significant difference in fatty acid composition of TAGs (Supplemental Figure S8C). In contrast, the mean contents of proteins and carbohydrates per seed in the ABCA10 overexpression lines were comparable to those of the wild-type (Figure 7, E and F). Therefore, the seed dry weight increments in ABCA10 overexpression lines were due specifically to higher TAG contents. Taken together, these results indicate that higher ABCA10 protein levels specifically induce seed TAG accumulation.

We then investigated how overexpressed ABCA10 increased the storage lipid accumulation at the later maturation phase, despite native ABCA10 primarily functioning during the very early phase of seed development. Overexpression of ABCA10 at the very early development phase (Figure 7A) might have promoted ABCA10-mediated endosperm development, ultimately producing bigger and heavier seeds, since the extent of the initial endosperm expansion is closely related with the final seed size (Garcia et al., 2003; Beauzamy et al., 2016). Alternatively, overexpression of ABCA10-sGFP, which resulted in artificially high levels of ABCA10-sGFP in embryos at the later maturation phase (Supplemental Figure S7), might have promoted TAG accumulation in embryos at the maturation phase. To test these possibilities, we quantified the size of the endosperms of ABCA10 overexpression lines at 4 DAF (Supplemental Figure S8D). The 2D areas occupied by endosperm nuclei of ABCA10 overexpression lines did not differ from that of the wild-type (Supplemental Figure S8D). These results suggested that the increased seed weight and dry seed surface area of ABCA10 overexpression lines (Figure 7, B and C) was not due to a change during the initial endosperm expansion stage, but more likely due to TAG accumulation promoted by ABCA10-sGFP in embryos at the maturation phase.

Discussion

In this study, we characterized the function of an Arabidopsis ABCA transporter during early seed development. ABCA10 was specifically expressed in female gametophytes and in endosperms of early developing seeds (Figure 2). At the stage of rapid endosperm expansion, ABCA10 expression was concentrated in the chalazal endosperm (Figure 2, F–H; Supplemental Figure S4, B and C). Loss of ABCA10 delayed development of the endosperm, starting from 3 DAF (Figure 4), and caused abnormal development of young seeds, which was apparent from 4 DAF onward (Figure 2B). These data indicated that ABCA10 functions in endosperm development in young seeds. No other ABC transporters have been revealed to function during such an early stage of seed development. The developmental defect of abca10 occurs very early and is likely to affect all subsequent development, including the accumulation of seed storage reserves. ABCA10 may also be functional, albeit at a much lower level, in later stages of seed development, contributing to storage TAG accumulation.

The endosperm at the chalazal end is likely the primary region of ABCA10 function because its expression is relatively concentrated there (Figure 2). Many reports suggest the importance of the chalazal area in seed development. In many Brassicaceae seeds, the chalazal endosperm has a complex cyst structure; the dome-shaped apical part of the cyst is enriched in numerous organelles, including polynuclei, densely packed ER, stacks of Golgi body, plastids, mitochondria, and vesicles, while the basal part exhibits numerous projections that penetrate the underlying maternal tissue, forming a haustorium-like structure (Nguyen et al., 2000). This structure, together with its location above the nutrient unloading zone of the seed coat, were the bases of the suggestion that the chalazal endosperm acts as a conduit to transfer maternal nutrients to developing seeds (Nguyen et al., 2000; Brown et al., 2004). Supporting this possibility, the cutin barrier separating the seed coat from the zygotic tissues is absent at the chalazal end (Coen et al., 2019). Furthermore, minerals accumulate in the ERs and vacuoles of the chalazal endosperm immediately before the embryo needs those minerals, supporting the idea that the chalazal endosperm functions as a nutrition conduit (Otegui et al., 2002). In addition to nutrients, the chalazal endosperm provides developmental signals for seeds. In a previous transcriptomic study (Belmonte et al., 2013), the biosynthetic genes of gibberellic acid, abscisic acid, and cytokinin were predominantly expressed in the chalazal endosperm. More recently, it was revealed that TERMINAL FLOWER1 synthesized in the chalazal endosperm moves to the syncytial peripheral endosperm to mediate endosperm cellularization (Zhang et al., 2020). Therefore, it is reasonable to suggest that the failure of normal chalazal endosperm development underlies the defective seed phenotypes of abca10. Further studies are necessary to understand the detailed physiological function of ABCA10 in the transport of nutrients and/or developmental cues to other parts of the developing seed. For instance, ultrastructural analysis of abca10 chalazal endosperm, metabolite analysis of different seed sub-regions, and transcriptome comparison between the wild-type and abca10 seed sub-regions may be helpful in addressing this issue.

We observed that abca10-1 seeds at 4 DAF are defective in endosperm expansion and exhibit aggregation of endosperm nuclei (Figure 4, C–E; Supplemental Figure S5). Endosperm expansion is temporally coincident with the accumulation of hexoses, which led to the idea that endosperm expansion is driven by an elevation of turgor pressure of the endosperm central vacuole (Beauzamy et al., 2016) caused by the accumulation of hexose (Morley-Smith et al., 2008). We observed that abca10-1 seeds at this stage (3–4 DAF) accumulate less glucose than wild-type seeds (Figure 4F). It is tempting to speculate that low glucose levels might have failed to generate sufficient water potential to drive the influx of water necessary for normal endosperm expansion. Alternatively, the reduced glucose level might have resulted from the reduced size of abca10 seeds (Figure 2B). We prefer the former interpretation because ABCA10 is primarily expressed in the chalazal endosperm (Figure 2), abca10 seeds were small in the chalazal endosperm area (Figure 4), and the chalazal endosperm is likely important for nutrition transfer from the mother tissue (Nguyen et al., 2000; Brown et al., 2004), as discussed above. In short, our data, together with previous studies, support the idea that the developmentally defective chalazal endosperm could not fully support the transfer of sugar from the maternal tissues to the growing abca10 zygotes.

abca10 reduced endosperm growth much more than seed coat development; a gap was apparent between the seed coat and endosperm (Figure 4C; Supplemental Figure S5A). In contrast, growth of both the endosperm and seed coat was reduced in iku2 and mini3, mutants of sporophytically acting HAIKU2 (IKU2) and MINISEED3 (MINI3; Luo et al., 2005). IKU2 and MINI3 determine the final seed size by controlling the timing of endosperm cellularization. Precocious syncytial endosperm cellularization caused premature arrest of endosperm size increase and thus limited the final seed size of the mutants. Seed coat development is activated by auxin synthesized from the fertilized central cell (Figueiredo et al., 2016). Also, seed coat development can be induced by some unknown compounds inside the pollen tube even without fertilization (Kasahara et al., 2016). The signals that trigger seed coat development immediately after fertilization might be normal in abca10, increasing the seed cavity for zygote expansion. It appears, however, that abca10 could not expand its endosperm to fill the seed cavity. This defect might have hampered the seed size increase stimulated by the interaction of the expanding endosperm and the seed coat (Garcia et al., 2005; Creff et al., 2015), resulting in smaller seeds as early as 4 DAF (Figure 2B).

The retardation of early embryo development in abca10-1 (Figure 5) might also be caused by the loss of ABCA10 function in the embryo itself. At 7 DAF, 17% of embryos within abca10-1 seeds remained at the globular or earlier stages, whereas most wild-type embryos had reached the torpedo or later stages (Figure 5). Embryo development until the early heart stage is metabolically independent from the endosperm. The major maternally derived sucrose import pathway into developing embryo is symplastic, via the suspensor, until the early heart stage (Stadler et al., 2005; Morley-Smith et al., 2008; Lafon-Placette and Kohler, 2014). Furthermore, no symplastic connections exist between the embryo and endosperm at these stages in rapeseed (Brassica napus) (Morley-Smith et al., 2008). Overall, the contribution of the endosperm to embryo development until the early heart stage remains unclear, and thus we cannot exclude the possibility that the autonomous function of ABCA10 in the very young embryo contributes to embryo development until the heart stage. However, it is clear that embryo development depends largely on the endosperm at later stages of seed development (Lafon-Placette and Kohler, 2014; Chen et al., 2015). Thus, we suggest that the defective development of abca10 endosperm contributes to delayed embryo development and, consequently, to the abnormal appearance of some abca10 seeds.

A critical question remains regarding the biochemical mechanism by which ABCA10 functions in development. We suggest that ABCA10 is involved in eukaryotic lipid metabolism, based on the following evidence. First, among major seed reserves, storage lipids were most dramatically reduced by the loss of ABCA10 (Figure 1, D–F; Table 1). Second, in contrast, overexpression of ABCA10-sGFP produced a higher seed oil yield without altering the amounts of other storage compounds such as proteins and carbohydrates (Figure 7, D–F). Third, ABCA10-sGFP localized to the ER, a major site of lipid synthesis (Figure 6). Fourth, many members of the ABCA sub-family transport lipid compounds. Arabidopsis ABCA9, a homolog of ABCA10, is involved in the uptake of fatty acids for TAG biosynthesis in the ER (Kim et al., 2013), and human ABCA transporters are involved in the efflux of sterols and/or trans-leaflet flip-flop of phospholipids (Tarling et al., 2013). Early endosperm and embryo development is an active process that requires large quantities of lipids as building blocks. This process demands the biosynthesis of high amounts of phospholipids at the ER and their efficient transport to other cellular membranes, as evidenced by the embryonic lethality of loss-of-function mutations in several genes involved in lipid metabolism (Lin et al., 2015; Yunus et al., 2016). ABCA10 expression might facilitate the supply of the lipid raw materials to the rapidly developing organ and structures. ABCA10 could increase lipid synthesis by facilitating influx of fatty acids and acyl-CoA to the ER, as is the case with ABCA9 (Kim et al., 2013). As a nonexclusive alternative, ABCA10 may be involved in flip-flopping of phospholipids between ER membrane leaflets, as was described for human ABCA transporters (Tarling et al., 2013). Further studies are needed to solve exactly how ABCA10 contributes to overall ER-controlled lipid metabolism.

The overexpressed ABCA10 increased the storage lipid accumulation and seed weight (Figure 7) probably by promoting lipid metabolism in embryos at the seed filling stage. To test the alternative possibility that improved endosperm development increases embryo size and consequently seed size in ABCA10 overexpression lines, we measured the endosperm area of developing seeds of ABCA10 overexpression lines. However, the overexpression lines did not show any significant enhancement of endosperm size compared to the wild-type (Supplemental Figure S8D), making it unlikely that ABCA10 overexpression enhanced endosperm size.

ABCA10 resembles ABCA9 in multiple aspects (this study and Kim et al., 2013). Loss of function of ABCA10 produced small and shrunken seeds with reduced lipid content (Figure 1D; Supplemental Figure S3), as did the loss of ABCA9. Overexpression of either ABCA9 (Kim et al., 2013) or ABCA10 (Figure 7) under the control of the 35S promoter increased seed weights by enhancing storage lipid accumulation. Their encoded proteins are both localized to the ER. Recent transcriptome data revealed that ABCA10 and ABCA9 exhibit very similar spatiotemporal expression patterns (Supplemental Figure S1; Le et al., 2010; Klepikova et al., 2016). Therefore, both ABCA10 and ABCA9 likely play essential roles in similar eukaryotic lipid metabolic pathways. ABCA9 function is most prominent in seed storage lipid accumulation (Kim et al., 2013), and its role during early seed development has not been carefully analyzed. We speculate that ABCA10 function is not redundant with that of ABCA9, because single loss-of-function mutations of either ABCA10 or ABCA9 produced obvious seed development defects. Furthermore, ABCA10 and ABCA9 belong to distinct subgroups of the ABCA subfamily (Kang et al., 2011) and share ˂30% amino acid sequence identity. Future structural and biochemical studies of these ABC transporters will be necessary to determine whether they transport different substrates.

Plant seed storage lipids are an important source of nutrition and energy. Manipulation of genes involved in lipid transport can be a useful engineering method to increase seed oil yield. In this study, constitutive ectopic overexpression of ABCA10 increased TAG accumulation up to 47% in Arabidopsis seeds. Overexpression of FATTY ACID EXPORT (FAX) genes (Tian et al., 2018; Li et al., 2020) and ABCA9 (Kim et al., 2013) also increased seed oil yield. The combined heterologous expression of Arabidopsis FAX1 and ABCA9 enhanced seed oil content and seed yield in false flax (Camelina sativa; Cai et al., 2021). Thus, genetic manipulation of ABCA10 in combination with the aforementioned lipid transporters and other factors that enhance seed oil yield (An and Suh, 2015; Lee et al., 2017) might provide an efficient method to increase seed oil yield.

In summary, ABCA10 is an ER-localized protein necessary for normal endosperm and embryo development during the early stages of seed development that probably transports lipid-related compounds, thereby contributing to lipid metabolism. ABCA10 is essential for normal development of the chalazal endosperm and expansion of the coenocytic endosperm, which is necessary for embryo development and for achieving normal seed size (Figure 8). Thus, this work reveals a factor critical for early endosperm development, providing a deeper understanding of the intricate process of early seed development and the role of an ABCA protein.

Figure 8.

Model for ABCA10 function during early seed development of Arabidopsis. ABCA10 contributes to chalazal endosperm development, endosperm expansion and even distribution of endosperm nuclei, which are necessary steps to achieve full seed size and normal seed morphology. Color legend: green, embryo; orange, peripheral endosperm nuclei; magenta, chalazal endosperm; gray, micropylar endosperm; yellow, central cell; light green, seed coat. Bars, 100 μm.

Materials and methods

Plant materials and growth conditions

Arabidopsis (A. thaliana) seeds were surface-sterilized, stratified, and sown on half-strength Murashige and Skoog (MS) medium supplemented with 1% (w/v) sucrose and 2.5 mM MES, pH 5.7. Two-week-old seedlings were transferred to soil and grown in 22°C/18°C temperature cycles in a 16-h-light/8-h-dark photoperiod.

Isolation of abca10 mutants

Two independent T-DNA insertion lines, abca10-1 (SAIL_869_G08) and abca10-2 (FLAG_106D07), were obtained from the Nottingham Arabidopsis Stock Center (http://arabidopsis.info/BasicForm) and Versailles Resource Center (http://publiclines.versailles.inra.fr/tdna/index), respectively. Homozygous mutants were identified by polymerase chain reaction (PCR analysis of genomic DNA using T-DNA-specific primers and gene-specific primers. Absence of ABCA10 transcripts was confirmed by reverse transcription-quantitative PCR (RT-qPCR) or end-point RT-PCR. Primer information is available in Supplemental Table S1.

Generation of transgenic lines

Complemented lines were generated by introducing an ABCA10 genomic fragment (6,088 bp), including the 5'−upstream sequence (1,977 bp) and the entire ABCA10 coding region without the stop codon (4,111 bp), cloned upstream of and in-frame with the coding sequence of either GUS or sGFP, into abca10-1 plants by Agrobacterium (Agrobacterium tumefaciens)-mediated transformation. The GUS fusion construct used the pGWB3 vector, while the sGFP fusion construct used the pGWB4 vector (Nakagawa et al., 2007).

For ABCA10 overexpression, a clone corresponding to the ABCA10 genomic coding region without the stop codon was recombined in the pGWB5 vector to bring ABCA10 in-frame with the coding sequence of GFP. The resulting construct was then introduced into Col-0 plants by Agrobacterium-mediated transformation. Overexpression lines with detectable GFP fluorescence in developing embryos were selected for further analyses.

Quantification of transcript levels

A Ribospin Plant RNA extraction kit (GeneAll, South Korea) and Ribospin Seed/Fruit extraction kit (GeneAll, South Korea) were used to isolate total RNAs from siliques at 6 DAF or earlier and after 6 DAF, respectively. Relative ABCA10 transcript levels were quantified by RT-qPCR and normalized against that of the ubiquitin-conjugating enzyme gene UBC21 (Dekkers et al., 2012). Three sets of siliques from three different plants were used in one biological replicate.

Seed lipid analysis

For the seed lipid analysis, two sets of 300 seeds from individual plants were utilized in each biological replicate. A total of 300 dry seeds were collected and weighed, and 50-μL 2-mM triheptadecanoin (C17:0 TAG) was added as an internal standard for quantification. The samples were incubated in 1-mL boiling 2-propanol (with 0.01% (w/v) butylated hydroxytoluene) at 85°C for 5 min. After cooling, 2-mL chloroform was added, the seeds were homogenized and centrifuged at 1,300 g for 15 min at room temperature, and the supernatant was collected. The pellets were re-extracted with 2-mL chloroform and 1-mL methanol, and after centrifugation at 1,300 g for 15 min at room temperature, the supernatant was combined with the first one. The combined supernatant was mixed vigorously with 1.2 mL 0.9% (w/v) KCl by shaking, and after centrifugation at 1,300 g for 15 min at room temperature, the lower lipid layer was recovered into a new preweighed glass tube. The solvent was evaporated under nitrogen gas. The dried lipid pellet was then weighed and dissolved in chloroform at a concentration of 10 mg/mL for subsequent thin layer chromatography (TLC).

To isolate TAG, 0.8-mg total lipids were spotted onto a TLC plate and separated using a solvent mixture of hexane/diethyl ether/acetic acid (80:30:1, v/v/v). Neutral lipid spots were visualized by primuline spraying. The TAG spot was scraped off the plate and heated in 1 mL 2.5% (v/v) sulfuric acid in methanol at 90°C for 45 min to transform lipids into fatty acid methyl esters. To isolate tFAs, TLC was omitted and the extracted lipids were directly boiled with the sulfuric acid in methanol. After cooling, 1-mL distilled water and 2-mL hexane were added, and after centrifugation at 1,300 g for 15 min at room temperature, the upper lipid layer was recovered to isolate fatty acid methyl esters. The samples were dried, re-dissolved in several drops of hexane, and analyzed by gas chromatography with flame ionization detection (GC-2010; Shimadzu, Kyoto, Japan), equipped with an HP-INNOWax capillary column (30 m × 0.25 mm, 0.25-μm film thickness; Agilent Technologies, Santa Clara, CA, USA).

Analysis of seed proteins and carbohydrates

For the seed protein and carbohydrate analysis, three sets of 40 seeds from individual plants were utilized in each biological replicate. Total proteins were extracted from 40 dry seeds as previously described (Focks and Benning, 1998) and quantified using a Pierce BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA). Seed carbohydrates were analyzed as described previously (Focks and Benning, 1998) with some modifications. Forty dry seeds and about 200 developing seeds were used to analyze carbohydrate contents. To measure glucose content in developing seeds, young siliques were opened under a dissecting microscope. The number of seeds in the siliques was then counted, and seeds were dissected out from the siliques. Seeds were homogenized in 80% (v/v) ethanol and incubated at 70°C for 90 min. After centrifugation at 16,000 g for 5 min at room temperature, the supernatant was collected. The pellet was re-extracted in 400 μL 80% (v/v) ethanol two more times, and the supernatants were combined with the first one. The solvent was evaporated at room temperature using a vacuum dryer. The residue was dissolved in 100-μL water and used for sucrose or glucose quantification. The remaining insoluble pellet was suspended in 200-μL 0.2-M KOH and then incubated at 95°C for 1 h. After the addition of 35-μL 1-M acetic acid and centrifugation for 5 min at 16,000 g at room temperature, the supernatant was used for starch quantification. Sucrose and starch contents were measured using a sucrose colorimetric/fluorometric assay kit and starch colorimetric/fluorometric assay kit (BioVision, Milpitas, CA, USA), respectively. Glucose was quantified using the glucose and sucrose colorimetric/fluorometric assay kit (BioVision, USA).

Observation of developing seeds in siliques

Siliques at various developmental stages were harvested and fixed onto a glass slide with a piece of double-sided adhesive tape. Siliques were then slit open with a syringe needle, and the seeds were observed under an SZX12 dissecting microscope (Olympus, Tokyo, Japan).

Surface area measurement of dry seeds

Images of dry seeds were taken using a dissecting microscope. 2D seed surface area was quantified from the collected images using ImageJ.

GUS staining of developing seeds

Young siliques were slit open with a needle, infiltrated in GUS staining solution (100-mM NaPO₄, pH 7.0, 3-mM potassium ferricyanide, 3-mM potassium ferrocyanide, 10 mM EDTA, 0.1% Triton X-100, and 2-mM 5-bromo-4-chloro-3-indolyl β-d-glucuronide) in a vacuum for 30 min at room temperature and then incubated at 37°C for 10 h in the dark. Chlorophyll was cleared from siliques by incubation in ethanol:acetic acid (6:1, v/v) overnight. Cleared siliques were then mounted in VISIKOL clearing agent (Phytosys LLC, Hampton, NJ, USA) and observed for GUS staining under an Eclipse TE2000-U microscope (Nikon, Japan) with 200× magnification. The images were acquired and further processed using Nikon ACT-1 (version 2.70) software.

Observation of embryo stages in developing seeds

Siliques of 7 DAF plants were slit open with needle and cleared in 0.2-M NaOH solution in 1% (v/v) sodium dodecyl sulfate (SDS) at 37°C for 3 h. After washing three times with water, siliques were cleared for 10 min in a 2% (v/v) bleach solution and washed five times with water. Cleared siliques were mounted in liquid half-strength MS medium and silique valves were dissected out using needles and forceps. The developmental stages of embryos were determined under an Eclipse TE2000-U microscope with 200× magnification operated by Nikon ACT-1 software.

Observation of endosperm development

To observe chalazal endosperm, developing seeds at 3 DAF were fixed and cleared as described previously (Rabiger and Drews, 2013) with some modifications. Siliques were slit and fixed in phosphate-buffered saline (PBS, pH 7.4) with 4% (v/v) glutaraldehyde for 3 h at room temperature under vacuum. Fixed samples were dehydrated in a graded ethanol series for 15 min each: 1× PBS 10% ethanol, 1× PBS 20% ethanol, 1× PBS 40% ethanol, 0.5× PBS 60% ethanol, and 80% ethanol. The samples were then incubated in 95% ethanol overnight at 4°C and in 100% ethanol for 30 min twice. The samples were cleared in benzyl benzoate:benzyl alcohol (2:1 v/v) for 20 min, rinsed and mounted in immersion oil, and the seeds were dissected out of the siliques with needles. The autofluorescence signals emitted from the seeds were observed using confocal laser scanning microscopy (CLSM) with 400× magnification.

To observe endosperm nuclei, developing seeds at 4 DAF were isolated and mounted in liquid half-strength MS medium, and the autofluorescence from endosperm nuclei was observed using CLSM with 200× magnification.

Subcellular localization of ABCA10

The pGWB5:ABCA10 construct was introduced into Arabidopsis plants stably accumulating the ER marker NIP1;1-mCherry (Geldner et al., 2009). The tip of the primary roots of 7-d-old seedlings was cut and mounted in liquid half-strength MS medium. The fluorescence of ABCA10-sGFP and NIP1;1-mCherry was captured by CLSM with 600× magnification.

For localization in N. benthamiana epidermal cells, Arabidopsis BiP1 cDNA was introduced into the pGWB560 binary vector. Nicotiana benthamiana leaves were co-infiltrated with mixtures of Agrobacterium cultures bearing either the pGWB5:ABCA10 or the pGWB560:BiP1 constructs. After 48 h, infiltrated leaves were cut and mounted in liquid half-strength MS medium. The fluorescence of ABCA10-sGFP and BiP1-tagRFP from epidermal cells was captured by CLSM with 400× magnification.

Confocal microscopy

A FV1000 confocal microscope (Olympus, Japan), operated using Olympus FV10-ASW (version 04.02.01.20) software, was used to acquire fluorescence signals. Images were processed using Olympus FV31S-SW software. The autofluorescence of chalazal endosperm was observed by excitation with 543 nm laser and detection at 550–600 nm. The autofluorescence signals of endosperm nuclei were observed by excitation with 633-nm laser and detection with a long-pass detection barrier filter (BA650IF). The areas of the chalazal endosperm and areas occupied by endosperm nuclei were measured using ImageJ (https://imagej.nih.gov/ij/). The signals of ABCA10-sGFP were observed by excitation with 488-nm laser and detection at 500–530 nm. The signals of NIP1;1-mCherry and BiP1-tagRFP were observed by excitation with 543-nm laser and detection at 550–600 nm. The gain value of confocal observations of chalazal endosperm and peripheral endosperm of developing seeds was set as 3 and that of confocal observations of subcellular localizations and ABCA10-sGFP signal in either 0 DAF seed or 10 DAF embryo were set as 2.

Statistical analysis

GraphPad Prism (version 9.0.2) program was used for statistical analyses in this study. Student’s t test and Welch’s multiple t test were utilized to determine statistical difference between wild-type and an independent transgenic line: double asterisks, single asterisk and ns indicate P < 0.01, P < 0.05 and no significant difference, respectively. One-way analysis of variance (ANOVA) with Tukey’s post-hoc test was used to determine significant differences between more than two independent lines, with different alphabets indicating significant difference.

Accession numbers

Arabidopsis genes mentioned in this study are identified in The Arabidopsis Information Resource (https://www.arabidopsis.org) under the following accession numbers: ABCA10 (At5g61740), UBC21 (At5g25760), NIP1;1 (At4g19030), and BiP1 (At5g28540).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. ABCA10 is primarily expressed in ovules and young developing seeds.

Supplemental Figure S2. Identification of abca10 mutants.

Supplemental Figure S3. Characterization of lipid contents and fatty acid composition in dry seeds of abca10-1.

Supplemental Figure S4. Expression patterns of ABCA10 in ovules or early developing seeds.

Supplemental Figure S5. abca10-2 exhibits an abnormal syncytial endosperm.

Supplemental Figure S6. ABCA10-sGFP is as functional as native ABCA10.

Supplemental Figure S7. Expression of ABCA10-sGFP in embryos at maturation phase (10 DAF) from pro35S:ABCA10-sGFP lines.

Supplemental Figure S8. ABCA10-sGFP overexpression does not alter seed number per plant, endosperm development of young seed, or fatty acid composition of dry seed TAGs.

Supplemental Table S1. Primers used for the experiments.

 

S.S., C.C., Y.Y., Y.L., and J.H. designed the research. S.S., C.C., Q.Y., and Y.Y. performed the experiments. S.S. and C.C. generated the constructs for transgenic lines. S.S. and Y.Y. analyzed seed lipids. S.S., Y.L., and J.H. wrote the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Jae-Ung Hwang (jaeung.hwang@postech.ac.kr).