SnRK1 phosphorylation of FUSCA3 positively regulates embryogenesis, seed yield, and plant growth at high temperature in Arabidopsis

FUS3 phosphorylation at the SnRK1 target site regulates embryo growth rate, seed yield, and plant growth at control and elevated temperature.


Introduction
Seed production is vital to the success of higher plants and integrally important to human diet and animal feed. Seed development is a highly regulated and complex process.
Arabidopsis thaliana follows a classical pattern of double fertilization forming the zygote and endosperm. Cell division dominates early embryogenesis up until the globular stage; afterwards cell differentiation overtakes division in mid to late embryogenesis. In conjunction, seed storage reserves accumulate, dormancy is established, and desiccation tolerance is acquired, resulting in the final mature seed (Capron et al., 2009).
The LAFL proteins, LEAFY COTYLEDON 2 (LEC2), ABSCISIC ACID INSENSTIVE 3 (ABI3), and FUSCA3 (FUS3) B3 domain transcription factors, together with LEC1 and LEC1-LIKE (L1L), homologs of the NF-YB subunit of the CCAAT-binding complex, are considered master regulators of seed development. Mutations in these genes cause reduced levels of seed storage compounds, desiccation intolerance, and/or enhanced precocious germination of immature embryos, indicating their critical role in seed maturation (Jia et al., 2013;Nonogaki, 2014;Fatihi et al., 2016). Mutant analyses suggest that each gene acts during different developmental stages of seed maturation to arrest embryo growth, with fus3-3 mutants acting earliest and showing premature growth of excised embryos, from the torpedo stage onwards (Raz et al., 2001). Almost complete repression of somatic embryogenic potential is seen in lec1/2 and fus3 mutants, further demonstrating the requirement of these genes for the embryonic program (Gaj et al., 2005). In accordance, in ChIP using somatic embryo tissue, FUS3 was bound to seed-specific promoters, with the majority containing the FUS3/LEC2/ ABI3 target RY cis-motif (Reidt et al., 2000;Wang and Perry, 2013). Transcription of some of these seed-specific genes, such as those encoding storage proteins, CRUCIFERIN 3 (CRU3), LATE EMBRYOGENESIS ABUNDANT (LEA), and SEED STORAGE ALBUMIN 3 (2S3), have been demonstrated to be regulated by both FUS3 and ABI3 (Parcy et al., 1997;Nambara et al., 2000;Kagaya et al., 2005;Yamamoto et al., 2010;Roscoe et al., 2015).
Restriction of AFL function to seed development has been shown to be essential for the transition to vegetative growth and to be under epigenetic regulation (Jia et al., 2013;Fatihi et al., 2016). Ectopic expression of FUS3 during vegetative growth up-regulates seed-specific gene transcription, delays germination, growth, and flowering, and causes embryoniclike leaf development (Gazzarrini et al., 2004;Kagaya et al., 2005;Tsai and Gazzarrini, 2012b). Embryonic phenotypes are also shown in plants overexpressing the LEC genes (Lotan et al., 1998;Stone et al., 2001). Accordingly, in higher order viviparous1/abi3-like-(val), curly leaf-(clf), and pickle-(pkl) related mutants involved in chromatin remodeling, failure to suppress FUS3/LEC2/ABI3 expression during germination results in expression of embryonic traits and developmental arrest of seedlings (Jia et al., 2013;Fatihi et al., 2016).
Genetic, transcriptomic, and ChIP studies have shown that spatio-temporal expression of FUS3/LEC2/ABI3 is transcriptionally cross-regulated during seed development (Parcy et al., 1997;To et al., 2006;Mönke et al., 2012;Wang and Perry, 2013). Although extensive work has shown the importance of these genes in transcriptional regulation, little is known about how these proteins are themselves regulated during seed development. LAFL genes interact with hormone signaling and synthesis pathways at various levels to regulate seed development and inhibit germination (Jia et al., 2013). FUS3 in particular is a node in hormone crosstalk and acts as a molecular switch in the dormancy to germination transition (Gazzarrini and Tsai, 2015). FUS3 inhibits the synthesis of gibberellic acid (GA) and ethylene, and promotes that of abscisic acid (ABA) to induce dormancy while inhibiting germination and vegetative growth (Curaba et al., 2004;Gazzarrini et al., 2004;Lumba et al., 2012). ABA and GA positively and negatively regulate the FUS3 protein through a PEST degron (sequence rich in proline, glutamic acid, serine, and threonine), respectively (Gazzarrini et al., 2004;Lu et al., 2010). The FUS3 protein, which is not detected during germination at optimal temperature, accumulates during seed imbibition at high temperature due to an increase in the ABA/ GA ratio, which inhibits FUS3 degradation through inhibition of proteasome activity. Seeds overexpressing FUS3 are hypersensitive to high temperature and inhibit germination through de novo ABA synthesis (Chiu et al., 2012(Chiu et al., , 2016a. This indicates that FUS3 accumulation is under tight hormonal control and regulates both primary and secondary dormancy.
Snf1-related kinase1 (SnRK1) is a conserved eukaryotic kinase complex involved in cellular energy homeostasis. In Arabidopsis, there are three homologs of the kinase α-subunit, of which AKIN10/SnRK1α1 and AKIN11/SnRK1α2 are functional (Crozet et al., 2014). Single mutants of SnRK1α show a slight delay in flowering and lower transcript levels of darkness-induced genes, while a virally induced akin10/akin11 RNAi double mutant undergoes premature senescence and is non-viable. SnRK1 overexpression elicits transcriptome shifts to inhibit growth and promote survival (Baena-González et al., 2007;Mair et al., 2015;Nukarinen et al., 2016). In peas (Pisum sativum), an antisense SnRK1α construct driven by a seed-specific promoter results in enhancement of vivipary, reduced desiccation tolerance, and lower levels of seed storage compounds and ABA, similar to fus3/lec1/lec2/abi3 mutants, and correspondingly causes reduced transcript levels of FUS3 and LEC1 orthologs. SnRK1 knockdown lines also displayed seed abortion and delayed embryo development, suggesting that SnRK1 plays an important role during seed development (Radchuk et al., 2006(Radchuk et al., , 2010. FUS3 is phosphorylated by AKIN10/SnRK1α1 (Tsai and Gazzarrini, 2012a). Overexpression of AKIN10 delays germination, senescence, and flowering (Baena-González et al., 2007;Tsai and Gazzarrini 2012a, b;Jeong et al., 2015), which can be rescued by fus3-3, indicating that FUS3 acts downstream of AKIN10 (Tsai and Gazzarrini, 2012a). Although overexpression of AKIN10 delays FUS3 degradation, the role of FUS3 phosphorylation by AKIN10 in seeds is unknown (Tsai and Gazzarrini, 2012a). In this study, we examined the function of the SnRK1 phosphorylation sites of FUS3 (S55, S56, S57) through analysis of FUS3 phosphorylation-null (FUS3 S>A ) and phosphorylation-mimic (FUS3 S>D ) mutants. While the FUS3 S>D variant rescued fus3-3 seed maturation defects, FUS3 S>A showed increased seed abortion, an increased number of seedling showing polycotyledons [previously observed in fus3-3 (Tsai and Gazzarrini, 2012a)], delayed embryo development, and reduced seed yield. Interestingly, akin10 and akin11 mutants displayed a frequency of seed abortion similar to fus3-3. Furthermore, when fus3-3 and FUS3 S>A complemented lines were grown under elevated temperature, they displayed reduced plant vigor, seed yield, and seedling growth of the progeny. These results uncover a new function for FUS3 phosphorylation during embryogenesis and under heat stress.

Materials and methods
Plant material, growth conditions, and phenotypic analysis The AKIN10-green fluorescent protein (GFP) transgenic line, and akin10 and akin11 mutants were previously described (Bitrián et al., 2011;Tsai and Gazzarrini, 2012a;Mair et al., 2015). FUS3 phosphomutants were generated by site-directed mutagenesis of a previously described pFUS3:FUS3-eGFP/pBI construct (Gazzarrini et al., 2004) using QuikChange ® (Agilent). The non-phosphorylatable or phospho-null (pFUS3:FUS3 S>A -GFP) and phosphorylation-mimic or phospho-mimic (pFUS3:FUS3 S>D -GFP) transgenes were then introduced into the Arabidopsis Columbia fus3-3 mutant  by Agrobacterium-mediated plant transformation. Several transgenic plants were selected, and experiments were conducted using two independent homozygous lines. Arabidopsis seeds were germinated on half-strength Murashige and Skoog (MS) medium, transferred to soil at 7 d after germination, and grown under long days at 16 h light 21 °C/8 h dark 18 °C (optimal temperature regime) or 16 h light 27 °C/8 h dark 24 °C (high temperature regime).
For germination and seedling establishment assays, three sets of 50 seeds were cold stratified and germinated for 7 d. Bolting time was counted when the primary inflorescence reached 2 cm in length. An average of 15-20 plants per genotype was used. Experiments were repeated twice, and one is shown. Total seed production per silique and seed abortion were determined by counting the number of funiculi and seeds in eight siliques per genotype. Embryo development was scored by the morphological shape of dissected embryos from four siliques per genotype. Growth arrest was scored by growth of dissected embryos on 0.4% water agar after 5 d under constant light at 22 °C. Seed yield was measured from the total weight of dry seeds of eight plants per genotype grown in two pots.

Microscopy
Light microscopy images were taken using an SMZ1500 dissecting microscope (Nikon). Confocal images were taken using an LSM510 (Zeiss) with a 488 nm excitation laser with a 515-535 nm bandpass filter for GFP emission (green channel) and a 595 nm high-pass filter (red channel). Plant tissue was directly mounted on glass slides in 10% glycerol.
Gene expression analysis RNA was extracted from ground frozen plant material using the RNeasy Plant Mini Kit (Qiagen), followed by reverse transcription using GoScript™ (Promega). Quantative PCR was performed using SsoFast™ EvaGreen ® (BioRad) with the CFX Connect™ real-time PCR detection system (BioRad). Primers used are listed in Supplementary Table S1 at JXB online.

AKIN10 is expressed during embryogenesis and phosphorylates FUS3
It has been previously shown that FUS3 is phosphorylated by AKIN10 on serine residues 55, 56, and 57 in an in-gel kinase assay using cell extract from seedlings (Tsai and Gazzarrini, 2012a). However, the FUS3 protein has only been detected during embryogenesis using a FUS3-GFP reporter, although mRNA is also found post-embryonically (Gazzarrini et al., 2004;Lu et al., 2010;Tsai and Gazzarrini, 2012a). To test whether AKIN10 may phosphorylate FUS3 in vivo, we first examined the expression pattern of AKIN10 during seed development. In transcriptomic studies, FUS3 and AKIN10 transcripts have been detected throughout embryo development, and as early as the elongated zygote stage (Le et al., 2010;Xiang et al., 2011). Using the AKIN10-GFP line previously described (Bitrián et al., 2011), the AKIN10 protein was found to be expressed in all cells from the heart to mature embryo stages, which was also confirmed by western blots (Fig. 1A, B; Supplementary Fig. S1). Thus, the AKIN10-GFP expression pattern during embryogenesis overlaps with that of FUS3-GFP previously published (Gazzarrini et al., 2004). Furthermore, in-gel kinase assays show that AKIN10 from cell extracts of siliques corresponding to early embryo stages can phosphorylate FUS3 ( Fig. 1C; Supplementary  Fig. S2), suggesting that AKIN10 interaction with and phosphorylation of FUS3 is biologically relevant and may play a role during early to mid embryogenesis.

FUS3 phosphorylation at SnRK1 sites positively regulates expression of FUS3 target genes
To elucidate the function of FUS3 phosphorylation by SnRK1, we mutated serine residues 55, 56, and 57 of the pFUS3:FUS3-GFP construct (Gazzarrini et al., 2004) to alanine and aspartic acid to generate phosphorylation-null (FUS3 S>A ) and potential phosphorylation-mimic (FUS3 S>D ) constructs driven by the FUS3 promoter, respectively. Substitution of serine with aspartic acid has been shown to mimic phosphorylation in several proteins including transcription factors (Maréchal et al., 1999;Kawamoto et al., 2015;Zorzatto et al., 2015). Both constructs were introduced into the fus3-3 background and rescued the desiccation intolerance phenotype of fus3-3 seeds ( Fig. 2A). In the embryo, FUS3 S>A -GFP and FUS3 S>D -GFP were predominately localized to the nucleus, similar to wild-type FUS3-GFP, indicating that phosphorylation was not required for FUS3 subcellular localization and both mutant proteins were functional (Fig. 2B).
Interestingly, fewer FUS3 S>A lines were recovered compared with FUS3 S>D , suggesting that phospho-null mutations provide weaker rescue. This may be due to lower mRNA levels of FUS3 S>A compared with FUS3 and FUS3 S>D transgenes (Fig. 2C). Given that FUS3 binds to its own promoter and increases its own expression (Parcy et al., 1997;To et al., 2006), these data suggest that FUS3 transcriptional activity may be increased by phosphorylation. To test this hypothesis, we investigated FUS3 transcriptional target gene expression. ChIP studies have shown that FUS3 binds to the promoters of genes that are highly expressed during seed maturation (Wang and Perry 2013), including those encoding 12S storage protein CRU2 (At1G03880), 2S storage protein 2S3 (At4G27160), fatty acid elongase FAE1/KCS18 (AT4G34520), and scorpion toxin proteinase/trypsin inhibitor SCT (At1g47540). These genes are expressed throughout embryogenesis (Toufighi et al., 2005) and their transcript levels were strongly reduced in fus3-3; thus, they were good markers for FUS3 transcriptional activity ( Fig. 2D; Yamamoto et al., 2010). Transcript measurements taken at the bent cotyledon stage [10 days after pollination (DAP)] show that the FUS3 S>A line has lower transcript levels of these genes compared with the FUS3 and FUS3 S>D lines (Fig. 2D). However, the levels of storage proteins in dry seeds did not appear to be affected in FUS3 S>A lines (Fig. 2E). Thus, although phosphorylation modulates FUS3 transcriptional activity, it did not appear to be required for seed maturation processes such as desiccation tolerance and storage protein accumulation under normal growth conditions.

FUS3 phosphorylation is required for early embryo development and positively impacts seed yield
Previous work has shown that silencing and overexpression of AKIN10 cause dramatic phenotypes during vegetative and reproductive development, including alteration of growth rate, senescence, and fertility (Baena-González et al., 2007;Tsai and Gazzarrini, 2012a, b). This suggests that phosphorylation of FUS3 by SnRK1 could affect vegetative and reproductive development. When grown under long days at 21 °C/18 °C, FUS3 S>D lines consistently displayed a delay in flowering (Fig. 3A), similar to plants overexpressing FUS3 and AKIN10 (Gazzarrini et al., 2004;Baena-González et al., 2007;Lu et al., 2010;Tsai and Gazzarrini, 2012b). Given that neither FUS3 nor FUS3phosphomutant proteins were expressed during vegetative development ( Supplementary Fig. S3), embryonic FUS3 S>D could affect the regulation of genes that control flowering time and are also expressed in the embryo, such as the floral repressor FLOWERING LOCUS C or other regulators of flowering that are expressed in the embryo (Sheldon et al., 2008).
In contrast, FUS3 S>A plants appeared less vigorous and their seed yield was reduced by ~50% compared with the wild type and fus3-3 (Fig. 3B). To understand the cause of decreased seed yield of the FUS3 S>A plants, the seed set per silique and quality of seeds produced were examined. When grown under long days, 25-35% of the FUS3 S>A seeds were aborted ( Fig. 3C; Table 1). Aborted seeds were randomly distributed within a silique and among siliques. Surprisingly, seed abortion was also displayed by fus3-3, but at a much lower frequency, ~8% (Table 1). To test if the higher number of aborted seeds of FUS3 S>A was due to a zygotic or gametic effect, we crossed FUS3 S>A with FUS3 and analyzed the F 1 progeny of reciprocal crosses. When FUS3 S>A was the pollen donor, seed abortion was comparable with fus3-3, at 7%, indicating that maternal FUS3 and paternal FUS3 S>A did not rescue the fus3-3 seed abortion phenotype. However, in the reciprocal cross with maternal FUS3 S>A and paternal FUS3, seed abortion was increased to 23-39%, with a higher frequency of white aborted seeds due to unfertilized ovules (Fig. 3D, E; Table 2). These results show that the lower seed Total seeds produced from eight siliques including aborted seeds were counted. Seed abortion included both fertilized and unfertilized ovules. Plants were grown under long days at 21 °C/18 °C. yield of FUS3 S>A is due in part to an increase in seed abortion due to a maternal effect of FUS3 S>A . Lastly, akin10 and akin11 mutants showed a rate of seed abortion similar to that of fus3-3 (Table 1), suggesting that FUS3 and AKIN10 may interact to regulate early embryo development processes.

FUS3 phosphorylation positively regulates embryo growth rate
Upon closer inspection, many FUS3 S>A siliques also showed delayed embryo development. While almost all wild-type embryos reached the late bent cotyledon stage at 11 DAP, <50% of the FUS3 S>A embryos were at a comparable stage (Fig. 4). Most FUS3 S>A embryos were closer to the walking stick stage and 10-15% were still at the torpedo or earlier stages. In contrast, the developmental progression of FUS3 S>D embryos was not significantly different from that of the wild type. Unlike FUS3 S>D , FUS3 S>A did not rescue and instead enhanced the fus3-3 cotyledon defects, such as polycotyledons (1.5%), previously reported (Table 3; Tsai and Gazzarrini, 2012a). Together, these results indicate that FUS3 phosphorylation promotes embryo growth and is required for proper cotyledon development.
In agreement with the literature, fus3-3 embryos showed precocious vegetative growth, as shown by growth of the radicle , and all FUS3 phosphomutants rescued this phenotype (Fig. 4A). Previous work has shown that fus3-3 mutants fail to suppress growth of excised embryos (Raz et al., 2001). Therefore, we asked if the same growth suppression mechanism caused the difference in embryo development arrest or delay. Using the same growth assay on torpedo stage embryos (7 DAP), we observed no significant difference in the ability to complement the growth suppression defect of fus3-3 by both FUS3 variants (Fig. 5A,  B). Accordingly, expression of GA3ox2, a GA biosynthetic gene, is derepressed in fus3-3 but not in the FUS3 phosphomutant embryos ( Fig. 5D; Curaba et al., 2004;Gazzarrini et al., 2004;Wang and Perry, 2013). A lower level of asynchronous embryo development was also observed in the early embryo stage of fus3-3 (Fig. 5C). Overall, these data indicate Total seeds produced including aborted seeds from crossed plants were counted after mature seeds began browning. Aborted seeds were classified as unfertilized ovules (white) and aborted embryos (brown) based on coat color and morphology. Plants were grown under long days at 21 °C /18 °C. that FUS3 phosphorylation positively regulates the embryo growth rate, but is not required for suppression of precocious germination.

FUS3 phosphorylation is essential for plant growth and seed development at high temperature
The FUS3 protein is not detected post-embryonically under optimal growth conditions; however, it accumulates in seeds imbibed at supraoptimal temperature to inhibit germination (Chiu et al., 2012). We examined whether FUS3 phosphorylation by SnRK1, a major regulator of the stress response, plays a role during growth at elevated temperature. When grown under long days at 27/24 °C, the overall growth was negatively impacted in fus3-3 and FUS3 S>A plants, which showed smaller rosettes, weaker inflorescences that were unable to stand upright, and reduced seed yield compared with the wild type (Fig. 6A, B). FUS3 S>A line 3 also produced shriveled seeds with a low germination rate, poor progeny seedling establishment, and yellowing of the cotyledons at 27 °C, similar to fus3-3 (Fig. 6A, C). The mRNA level of FUS3 S>A of line 3 was strongly reduced, explaining the lack of fus3-3 rescue of this line at 27 °C (Fig. 6D). When plants were grown at elevated temperature, both FUS3 S>A and FUS3 S>D variants caused delayed embryo development and increased seed abortion compared with the wild type and fus3-3 ( Fig. 7; Table 4). This suggests that while aspartic acid may mimic phosphorylation under control conditions, effective similarity is reduced by elevated temperature. Alternatively, constitutive phosphorylation (S>D) may not be as effective as regulated phosphorylation. Altogether, these data indicate an essential role for FUS3 phosphorylation during growth  at high temperature. Given that neither FUS3 nor its phosphomutant variants were detected during vegetative growth at high temperature ( Supplementary Fig. S4), these results also suggest that FUS3 phosphorylation in the mother plant integrates temperature cues to regulate growth and survival of the progeny indirectly.

Discussion
The LAFL transcription factors are master regulators of the seed maturation program. Through alteration of the transcriptome, they co-ordinate the accumulation of seed storage compounds, the establishment of dormancy, and acquisition of desiccation tolerance, while suppressing germination (Jia et al., 2013;Sreenivasulu and Wobus 2013;Fatihi et al., 2016). Compared with their role in transcription, there is less known about upstream regulators and modulators of their functions. Previously, we have shown that FUS3 interacts with and is phosphorylated by the SnRK1α1/AKIN10 kinase on three serine residues in the conserved B2 domain (Tsai and Gazzarrini, 2012a). However, the biological role of this post-translational modification was not clear. Here we found that AKIN10 is expressed in embryos, from the triangular to mature stages, and phosphorylates FUS3 during early to mid stages of embryogenesis. Using FUS3 phosphomutants at the SnRK1 target sites, we show that FUS3 phosphorylation plays a crucial role during embryogenesis, by regulating embryo growth rate and seed development under optimal and stress conditions. Impairment of FUS3 phosphorylation negatively affects FUS3 transcriptional activity, increases seed abortion, delays embryo development, and ultimately results in a large decrease in seed yield. The importance of FUS3 phosphorylation is more evident when plants are grown at high temperature, as even germination and vegetative growth of the next generation is greatly impacted. This suggests that SnRK1 and FUS3 may integrate endogenous signals (energy level) and environmental cues to regulate seed development and ensure survival of the next generation.
Previously, we showed that SnRK1 phosphorylation positively regulates FUS3, and that FUS3 acts downstream of AKIN10 in Arabidopsis (Tsai and Gazzarrini, 2012a). However, AKIN10 was previously found to be expressed only in the ovule, but not in the embryo where FUS3 mainly functions (Bitrián et al., 2011). Here we show that AKIN10 is indeed expressed in the embryo, and that FUS3 and AKIN10 protein expression patterns partially overlap during embryogenesis. This is in agreement with the expression of the regulatory subunit SnRK1βγ in embryos (Bitrián et al., 2011). Furthermore, FUS3 is phosphorylated by AKIN10 in cell extracts of early to mid embryos, thus their interaction is biologically relevant. In pea seeds, SnRK1α knockdown lines are impaired in maturation and show phenotypes similar to those displayed by fus3 and abi3 mutants, including precocious germination, reduced ABA and seed storage levels, and reduced desiccation tolerance (Radchuk et al., 2006(Radchuk et al., , 2010. This suggests that SnRK1 plays a prominent role during seed development and maturation, and probably acts through regulation of AFL functions. To understand the role of FUS3 phosphorylation by SnRK1, we transformed fus3-3 with FUS3:FUS3 S>D -GFP phospho-mimic and FUS3:FUS3 S>A -GFP phospho-null constructs. Our expression studies indicate that down-regulated gene transcription in fus3-3 mutants is fully restored in  FUS3 S>D , but only partially restored in FUS3 S>A embryos, suggesting that FUS3 transcriptional activity is increased by phosphorylation. Furthermore, FUS3 S>A transgene expression was reduced while that of FUS3 S>D was increased compared with FUS3. This may be due to reduced FUS3 S>A and slightly increased FUS3 S>D ability, respectively, to regulate the FUS3 promoter positively, as previously shown for FUS3 (Parcy et al., 1997;To et al., 2006;Wang and Perry, 2013). However, FUS3 phosphorylation by SnRK1 did not appear to be required for seed maturation at normal temperature. The dramatic fus3-3 phenotypes, including reduced dormancy, low levels of seed storage compounds, desiccation intolerance, and ectopic trichomes on the cotyledons (Bäumlein et al., 1994;Keith et al., 1994;Meinke et al., 1994), were rescued by expressing either FUS3 S>A or FUS3 S>D transgenes. Altogether, these data suggest that SnRK1 phosphorylation positively regulates FUS3 expression and transcriptional activity, but does not play a prominent role during seed maturation. It is possible that the full rescue of fus3-3 seed maturation phenotypes in FUS3 S>A lines may be due to the action of other LAFL genes. Expression of FUS3 and its target genes is induced by other B3 domain proteins through binding of the RY motif, which may compensate for reduced FUS3 S>A transcriptional activity (Jia et al., 2013). Upon closer inspection of siliques from plants grown under optimal temperatures, we found that ~8% of the fus3-3 seeds were aborted and, surprisingly, FUS3 S>A lines showed an increase in the frequency of aborted seeds (~30%) due to a maternal FUS3 S>A effect. Furthermore, FUS3 S>A lines also showed delayed embryo development, with 10-15% of the seeds still at the heart to torpedo stages when all wild-type embryos already reached the mature stage. This resulted in a reduction of seed yield by ~50% in FUS3 S>A plamts. Thus, it appears that in FUS3 S>A siliques only half of the seeds were able to develop fully and reach the mature stage, and these embryos did not show maturation defects. Notably, akin10 and to a lesser extent akin11 mutants showed frequencies of seed abortion close to that of fus3-3, and similar phenotypes were observed during seed development of antisense SnRK1α lines (with 30-50% reduction in SnRK1 activity) in pea, where 10% seed abortion and delayed embryo development was also reported (Radchuk et al., 2006). Interestingly, FUS3 phosphorylation affects the vegetative development and flowering time of the progeny, despite the lack of FUS3 protein expression post-embryonically, suggesting a link between seed development and flowering time. While FUS3 S>D rescued all fus3-3 embryonic phenotypes, the FUS3 S>D progeny flowered later than the wild type at control temperature, similarly to ML1:FUS3 plants that overexpress FUS3 post-embryonically (Gazzarrini et al., 2004). During seed development, FUS3 S>D may alter the regulation of flowering-related genes such as FLC, whose expression in the embryo was shown to be required for its repression of flowering (Sheldon et al., 2008), or DELAY OF GERMINATION1 (DOG1), which controls dormancy and flowering time through miRNA 156/172 (Huo et al., 2016). DOG1 and miRNA 156/172 are FUS3 targets based on ChIP studies (Wang and Perry, 2013).
When plants were grown at high temperature, plant growth was accelerated and seed yield was reduced in all genotypes. All phenotypes displayed by FUS3 S>A lines were enhanced when plants were grown at high temperature, in particular, both fus3-3 and FUS3 S>A plants were less vigorous and unable to stand upright, and the progeny seedlings grew poorly compared with the wild type. This indicates that phosphorylation of FUS3 by SnRK1 is critical during heat stress and affects the survival of the next generation. Possibly, specific transcriptional activities may be altered if phosphorylation of FUS3 changes the nature of ternary transcription complexes with LEC1/NF-YB, which increase the transcriptional activity of ABI3/LEC2 on the OLE1 promoter . Alternatively, this may arise from novel protein-protein interactions. Thus, FUS3 phosphorylation in the mother plant integrates temperature cues to regulate growth and survival of the progeny. Interestingly, progeny dormancy is regulated by maternal temperature through activation of the florigen Flowering Locus T in siliques (Chen et al., 2014;Penfield and MacGregor, 2017). It will be interesting to determine if maternal temperature affects both progeny dormancy and flowering time through similar pathways, thus passing environmental memory across generations to optimize plant growth.
The mechanism of how prevention of FUS3 phosphorylation causes seed abortion and delayed embryo development is not known. However, one possibility is that FUS3-mediated regulation of cell division is dependent on energy levels, which fluctuate in response to growth conditions and are signaled through SnRK1. FUS3 was previously shown to regulate cell division negatively; ectopic cell division was observed in the fus3-3 embryo, while FUS3 overexpression represses cell cycling (Raz et al., 2001;Gazzarrini et al., 2004). FUS3 directly binds to the promoter of cycD3;1, and cycD activity suppressors KIP-RELATED PROTEIN (KRP) 4 and 6, which have reduced expression in the fus3-3 mutant (Yamamoto et al., 2010;Wang and Perry, 2013). Higher order mutants of the cyclin-D3 (cycD3) cell cycle regulator subfamily also display seed abortion and delayed embryo development phenotypes shown by FUS3 S>A lines (Collins et al., 2012). Thus, FUS3 phosphorylation may control the rate of cell cycling to modulate the rate of embryo development and transition to maturation. Recently, a mutant affected in a novel nuclearlocalized protein, red1, also displayed seed abortion, delayed embryo development, and desiccation intolerance phenotypes described in this study (Du and Wang, 2016). However, the function of this gene is currently unknown.
In conclusion, this work uncovered a novel and essential role for FUS3 phosphorylation by SnRK1 during early embryogenesis at optimal and especially at high temperatures. Lack of phosphorylation impairs FUS3 function and results in unfertilized ovules, aborted seeds, and delayed embryo growth, suggesting metabolic and signaling impairment. Seed abortion, reduced seed yield, and overall plant growth were further impacted when plants were grown at elevated temperature, suggesting that phosphorylation is essential for FUS3 function under stress, possibly integrating energy and stress levels through SnRK1 phosphorylation to ensure the survival of the next generation. Future work will focus on elucidating the molecular mechanisms underlying these processes.

Supplementary data
Supplementary data are available at JXB online. Table S1. List of primers used in this research. Fig. S1. AKIN10 protein expression levels during seed development. Fig. S2. In-gel kinase assays. Fig. S3. FUS3-GFP and its phospho-variants are undetectable during early vegetative development at control temperature. Fig. S4. FUS3-GFP and its phospho-variants are undetectable during early vegetative development under elevated temperature.