Seed size critically affects grain yield of crops and hence represents a key breeding target. The development of embryo-nourishing endosperm is a key driver of seed expansion. We here report unexpected dual roles of the transcription factor EIN3 in regulating seed size. These EIN3 functions have remained largely undiscovered because they oppose each other. Capitalizing on the analysis of multiple ethylene biosynthesis mutants, we demonstrate that EIN3 represses endosperm and seed development in a pathway regulated by ethylene. We, in addition, provide evidence that EIN3-mediated synergid nucleus disintegration promotes endosperm expansion. Interestingly, synergid nucleus disintegration is not affected in various ethylene biosynthesis mutants, suggesting that this promoting function of EIN3 is independent of ethylene. Whereas the growth-inhibitory ethylene-dependent EIN3 action appears to be encoded by sporophytic tissue, the growth-promoting role of EIN3 is induced by fertilization, revealing a generation conflict that converges toward the key signaling component EIN3.

Polyploidization, the increase in genome copies, is considered a major driving force for speciation. We have recently provided the first direct in planta evidence for polyspermy induced polyploidization. Capitalizing on a novel sco1-based polyspermy assay, we here show that polyspermy can selectively polyploidize the egg cell, while rendering the genome size of the ploidy-sensitive central cell unaffected. This unprecedented result indicates that polyspermy can bypass the triploid block, which is an established postzygotic polyploidization barrier. In fact, we here show that most polyspermy-derived seeds are insensitive to the triploid block suppressor admetos. The robustness of polyspermy-derived plants is evidenced by the first transcript profiling of triparental plants and our observation that these idiosyncratic organisms segregate tetraploid offspring within a single generation. Polyspermy-derived triparental plants are thus comparable to triploids recovered from interploidy crosses. Our results expand current polyploidization concepts and have important implications for plant breeding.

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Polyploidization, the increase in genome copies, is considered a major driving force for speciation. We have recently provided the first direct in planta evidence for polyspermy induced polyploidization. Capitalizing on a novel sco1-based polyspermy assay, we here show that polyspermy can selectively polyploidize the egg cell, while rendering the genome size of the ploidy-sensitive central cell unaffected. This unprecedented result indicates that polyspermy can bypass the triploid block, which is an established postzygotic polyploidization barrier. In fact, we here show that most polyspermy-derived seeds are insensitive to the triploid block suppressor admetos. The robustness of polyspermy-derived plants is evidenced by the first transcript profiling of triparental plants and our observation that these idiosyncratic organisms segregate tetraploid offspring within a single generation. Polyspermy-derived triparental plants are thus comparable to triploids recovered from interploidy crosses. Our results expand current polyploidization concepts and have important implications for plant breeding. eLife digest Ever since Darwin published his most famous book on the theory of evolution, scientists have sought to identify the mechanisms that drive the formation of new species. This is especially true for plant biologists who have long been fascinated by the extraordinary diversity of flowering plants. Many species of flowering plant first evolved after a dramatic increase in the DNA content of an individual plant, a process termed polyploidization. Most explanations for polyploidization involve a pollen grain making sperm that mistakenly contain two sets of chromosomes rather than one. Yet, it is difficult to reconcile this explanation with an important aspect of plant reproduction – the so-called “triploid block”. Fertilization in flowering plants is more complicated than in animals. While one sperm fertilizes the egg cell to make the plant embryo, a second sperm from the same pollen grain must fertilize another cell to form the endosperm, the tissue that will nourish the embryo as it develops. This means that sperm with twice the normal number of chromosomes would affect the DNA content of both the embryo and the endosperm. Yet, an endosperm that receives extra paternal DNA typically halts the development of the seed via a process known as the triploid block, meaning it was not clear how often this process would actually result in a polyploid plant. In 2017, researchers reported that plants can, on rare occasions, generate polyploid offspring via a different route: the fertilization of one egg with two sperm rather than one. Now, Mao et al. – who include several researchers involved in the 2017 study – show that this process, termed “polyspermy”, can introduce extra copies of DNA into just the egg cell, meaning it can bypass the triploid block of the endosperm. The experiments involved a model plant called Arabidopsis, and a screen of over 55,000 seeds identified about a dozen with embryos that had three parents, one mother and two fathers. Notably, most of these three-parent embryos developed in seeds that contained endosperm with the regular number of chromosomes and hence escaped the triploid block. These new results show that polyspermy provides plants with a means to essentially sneak extra copies of DNA ‘behind the back’ of the DNA-sensitive endosperm and into the next generation. They also give new insight in how polyploidization may have shaped the evolution of flowering plants and have important implications for agriculture where the breeding of new “hybrid” crops has often been limited by incompatibilities in the endosperm. Introduction The evolutionary history of flowering plants is characterized by recurrent polyploidization events (Comai, 2005; De Bodt et al., 2005; Otto and Whitton, 2000; Van de Peer et al., 2017). Polyploids are generally assumed to arise from unreduced gametes or somatic doubling, i.e. from defects during meiosis or mitosis (Kreiner et al., 2017; Mason and Pires, 2015; Ramsey and Schemske, 1998; Sattler et al., 2016; Spoelhof et al., 2017; Tayalé and Parisod, 2013). In addition, recent in planta and in vitro assays have provided the first direct evidence that viable polyploid plants can arise from polyspermy, the fusion of one egg cell with supernumerary sperm (Nakel et al., 2017; Toda et al., 2016). In fact, this previous work indicates that a single Arabidopsis plant can generate several polyspermy-induced triploid seedlings (Nakel et al., 2017). The currently favored polyploidization scenario involves the formation of unreduced male gametes and the natural occurrence of such sperm has been reported for several species (Kreiner et al., 2017; Mason and Pires, 2015; Ramsey, 2007). Consequently, triploid plants are assumed to function as an important bridge towards polyploidization (Comai, 2005; Felber and Bever, 1997; Ramsey and Schemske, 1998) and field studies have identified both auto and allopolyploid triploids (Kyrkjeeide et al., 2019; Lee et al., 2001; Marques et al., 2018; Meng et al., 2018; Schinkel et al., 2017; Sree Rangasamy, 1972). However, the generation of triploid plants via unreduced male gametes is limited by the triploid block, which is a postzygotic hybridization barrier operating in many plants species (Dilkes et al., 2008; Köhler et al., 2010; Marks, 1966; Ramsey and Schemske, 1998; Scott et al., 1998). The triploid block is explained by the unique reproductive mode of flowering plants, which involves fertilization of two female gametes, the egg and the central cell. The required sperm cell pair is typically delivered by a single pollen tube. While the fertilized egg cell gives rise to the embryo, the fertilized central cell develops into embryo-nourishing endosperm (Russell, 1992). Fertilization involving unreduced sperm consequently not only affects the ploidy status of the egg cell but also introduces additional paternal chromosome copies to the endosperm, and it is this latter tissue, which commonly mounts the triploid block that is manifested by seed abortion (Köhler et al., 2010). In Arabidopsis thaliana, the effect of the triploid block is accession-dependent, being highly penetrant e.g. in Col-0, but less strict in Ler and C24 (Dilkes et al., 2008; Scott et al., 1998). A complete triploid block has been reported in many taxa (Ramsey and Schemske, 1998; Schinkel et al., 2017; Sekine et al., 2013; Stoute et al., 2012). In light of this fatal consequence, it has been suggested that there are ways to overcome this hybridization barrier (Köhler et al., 2010). Making use of a two-component in planta assay, we here show that polyspermy can selectively polyploidize the egg cell, while rendering the genome size of the ploidy-sensitive endosperm unaffected. By introducing the triploid block suppressor admetos, we in addition show, that this unprecedented reproductive mode bypasses the triploid block. Results Establishment of a triparental embryo detection assay Consistent with animal nomenclature, the term polyspermy is used alone when referring to egg cell polyspermy. Central cell polyspermy is specified as such. During flowering plant fertilization, both egg and central cell fuse in a coordinated manner with a single sperm each (Hamamura et al., 2011; Kawashima and Berger, 2011). In order to address whether during polyspermy egg cell fertilization is still coupled to the fertilization of the central cell, we aimed at analyzing endosperm in developing seeds that contain polyspermy-derived embryos. To ease the screening process, we established a novel polyspermy-detection assay termed HIPODSCO1, which can efficiently and unambiguously detect the rare event of egg cell polyspermy already in developing seeds. HIPODSCO1 capitalizes on the pale green appearance of developing seeds defective for the gene SNOWY COTYLEDON 1 (SCO1) (Ruppel and Hangarter, 2007) (Figure 1—figure supplement 1) and a bipartite SCO1 complementation system, provided by two different pollen donors (Figure 1A). Pollen donor one contains the synthetic GAL4 transcription factor under the control of the RPS5a promoter. Pollen donor two contains a functional copy of tdTOMATO tagged SCO1 under the control of the GAL4 responsive UAS enhancer sequence. Seeds that contain a monospermy-derived embryo inherit an incomplete complementation system and will consequently be rendered pale green due to the lack of functional SCO1. By contrast, combinations of both constructs, which can only result from polyspermy, will give rise to green seeds with a positive tdTOMATO fluorescence signal (Figure 1B). It should be noted that this assay only detects polyspermy if the two sperm are derived from different pollen donors. Monopaternal polyspermy, which only delivers a single HIPODSCO1 component, does not rescue seed color. This scenario is expected to account for 50% of all polyspermy events, and escapes detection. Figure 1 with 1 supplement see all Download asset Open asset Establishment of a detection assay for polyspermy-derived embryos. (A) Illustration of HIPODSCO1. The assay is based on the UAS-GAL4 two-component system whereby a synthetic transcription factor mGAL4 expressed under the control of the ubiquitous RPS5a promoter activates the tdTOMATO-tagged SCO1 gene. These two components were combined with the sco1 mutant to generate pollen donor 1 and 2 (PD1 and PD2), respectively. (B) Pollen of PD1 and PD2 (blue, yellow) are applied to the stigma of a sco1 gynoecium (green). Gamete fusion involving two sperm from two different pollen donors leads to transactivation of the SCO1 gene resulting in dark green seeds and fluorescence of tdTOMATO in the embryo, while monospermy-derived seeds remain pale green with no fluorescence. (C) Silique and seed analysis of sco1 mutants containing only pRPS5a::mGAL4-VP16, (upper panel), only pUAS::SCO1-tdTOMATO (middle panel), and both pRPS5a::mGAL4-VP16 and pUAS::SCO1-tdTOMATO (lower panel). Scale bars, 500 μm and 100 μm in left and right panel, respectively. To test the system, we compared seed color of sco1 mutant plants expressing either one of the constructs with seeds containing both, the GAL4 activator and the UAS reporter line. This experiment confirmed that only the presence of both constructs complemented the defect resulting in dark green seeds, which exhibited a tdTOMATO signal (Figure 1C). Polyspermy can selectively polyploidize the egg cell The novel HIPODSCO1 assay enabled us to screen for seeds that contain polyspermy-derived embryos at an advanced seed developmental stage. We processed a total of 56,493 seeds seven days after pollination (DAP) and identified 10 normally developed seeds with a change in color (Figure 2A, Figure 2—figure supplement 1A). To determine whether the candidate embryos were indeed of triparental origin, we microscopically inspected the developing seeds and found that all 10 embryos exhibited a tdTOMATO signal (Figure 2B, Figure 2—figure supplement 1A). This implies that the embryo inherited two rather than one paternal copy. To identify a corresponding shift in embryo ploidy, we carried out a chromosome spread assay. Chromosome counts are technically challenging when performed on subfractions of individual seeds and some chromosomes escape detection. However, comprehensive controls and the fact that parental chromosome contributions are quantal in nature make the assay robust and reliable. Notably all embryos showed a triploid profile (Figure 2C, Figure 2—figure supplement 1B). This finding is comparable to the results obtained from triploid embryos segregated from an interploidy cross between diploid and tetraploid plants and contrasts with the diploid profile detected in embryos recovered from a regular cross involving haploid gametes (Figure 2C). To substantiate this result we introduced a GFP-tagged centromere-localized CENH3 reporter into PD1 (De Storme et al., 2016). In this complementary experiment we screened 10,774 seeds by HIPODSCO1 and recovered three green seeds containing tdTOMATO positive embryos (Figure 2—figure supplement 1A). In all seeds we detected between 11 and 15 GFP foci indicative of triploid embryos (Figure 2E; Figure 2—source data 1). Together, the analysis confirms the triparental origin of embryos in seeds with dark green color, establishing HIPODSCO1 as a powerful novel tool to identify polyspermy-derived embryos already in developing seeds. In order to determine whether egg cell polyspermy is concomitant with central cell polyspermy, we assessed the ploidy of the endosperm in developing seeds containing polyspermy-induced triparental embryos. The central cell of many flowering plants, including Arabidopsis thaliana, is homodiploid and generates a triploid nurturing tissue after sperm fusion. In fact, we detected between 12 and 15 chromosomes in the endosperm of seeds recovered from a cross involving diploid plants. By contrast, more than 15 chromosomes are detected in control interploidy crosses between diploid female and tetraploid male (Figure 2D). Remarkably, in the 10 developing seeds containing triparental embryos we detected between 11 and 15 chromosomes, which is characteristic of a triploid endosperm (Figure 2D, Figure 2—figure supplement 1C). The result was substantiated by a complementary experiment involving the recombinant CENH3-GFP reporter, which detected a triploid profile in the endosperm of three analyzed seeds (Figure 2F; Figure 2—source data 1). Notably, at this advanced seed developmental stage, we recovered one abnormal seed from the HIPODSCO1 assay containing an underdeveloped triploid heart-stage embryo and tetraploid endosperm (Figure 2—figure supplement 1D), characteristic of triploid block-induced seed abortion (Dilkes et al., 2008; Kradolfer et al., 2013; Scott et al., 1998). Figure 2 with 1 supplement see all Download asset Open asset HIPODSCO1 identifies developing seeds harboring polyspermy-derived embryos. (A and B) Bright- and fluorescence light images of seeds from different crosses seven days after pollination (DAP). Upper panel (HIPODSCO1): sco1 × pRPS5a::mGAL4-VP16/+ sco1 (PD1) × pUAS::SCO1-tdTOMATO/+ sco1 (PD2); middle panel: pUAS::SCO1-tdTOMATO/+ sco1 × pRPS5a::mGAL4-VP16/+ sco1; lower panel: sco1 × sco1. Asterisk indicates polyspermy-induced complementation of a sco1 seed. (C and D) DAPI-stained chromosome spreads of embryo (EM) (C) and endosperm (EN) (D) resulting from different crosses. Left panel: HIPODSCO1-rescued embryo segregating from cross between sco1 × pRPS5a::mGAL4-VP16/+ sco1 (PD1) × pUAS::SCO1-tdTOMATO/+ sco1 (PD2); middle panel: pUAS::SCO1-tdTOMATO/+ sco1 × pRPS5a::mGAL4-VP16/+ sco1; right panel: sco1 × wild type (4n). (E and F) Chromosome counting through centromere-targeted CENH3-GFP of embryo (E) and endosperm (F) resulting from different crosses. Left panel: rescued embryo segregating from cross between sco1 × pRPS5a::mGAL4-VP16/+ p35S::CENH3-GFP sco1 (PD1 with CENH3-GFP) × pUAS::SCO1-tdTOMATO/+ sco1 (PD2); middle panel: pUAS::SCO1-tdTOMATO/+ sco1 × pRPS5a::mGAL4-VP16/+ p35S::CENH3-GFP sco1 ; right panel: wild type (4n) × pRPS5a::mGAL4-VP16/+ p35S::CENH3-GFP sco1 . The numbers in parenthesis indicate the average counted chromosomes from all analyzed cells, from left to right, (C) n = 11, 11, 9, (D) n = 20, 6, 9, (E), n = 82, 57, 78, (F), n = 12, 26, 6. Scale bars, 200 μm (A), 100 μm (B), 1 μm (C–E). Figure 2—source data 1 Chromosome counting through centromere-targeted CENH3-GFP in embryo and endosperm resulting from different crosses. https://cdn.elifesciences.org/articles/52976/elife-52976-fig2-data1-v1.xlsx Download elife-52976-fig2-data1-v1.xlsx Together our data indicate that egg cell polyspermy can occur independent of central cell polyspermy. Such selective polyploidization of the egg cell implies that polyspermy has the potential to bypass the triploid block. Most polyspermy-induced polyploidization events are insensitive to the triploid block suppressor admetos To further substantiate our findings, we established a functional assay to determine the potential of polyspermy in bypassing this reproductive barrier. It was previously shown that mutations in the paternally expressed imprinted gene ADMETOS (ADM) suppress the triploid block in Arabidopsis thaliana (Kradolfer et al., 2013). In fact, interploidy crosses between diploid and tetraploid plants lead to a 15.6 fold increase in fertile triploid seeds when the tetraploid pollen donor segregated the adm-1 allele (Figure 3A–C, Figure 3—source data 1). If polyspermy would equally trigger the triploid block, we would expect a similar increase in polyspermy frequencies when using adm-1 segregating pollen donors. In order to identify polyspermy-derived seedlings, we made use of the previously established HIPOD assay that works analogous to the HIPODSCO1 system introduced above, but positively selects triparental seedlings on the basis of herbicide resistance (Nakel et al., 2017). A total of 116,279 and 113,777 seeds were harvested from three independent HIPOD experiments using either adm-1 or wild-type segregating pollen donors, respectively. Out of these, 47 herbicide resistant seedlings segregated from pollen donor with adm-1 background while 27 were recovered from wild type (Figure 3D–F, Figure 3—source data 1 ). This corresponds to an almost two fold increase in the adm-1 segregating approach (Figure 3C), which is more than eight times lower than the effect observed in the interploidy cross. Figure 3 Download asset Open asset Polyspermy-induced polyploidization is partially insensitive of adm-mediated triploid block repression. (A and B) Mature seed (A) and corresponding 9 day old seedling (B) from one silique of wild-type pollinated with diploid wild-type pollen and diploid adm-1 pollen. (C) Effect of the triploid block repressor adm on interploidy cross- recovered triploids (2n × 4n) and polyspermy- derived triploids. Shown is the ratio of triploid plants recovered from two crosses involving either adm or wild-type pollen donors (adm/WT). The data are means ± SEM (n = 3 experiments). (D) Herbicide-treated offspring of triparental triploid (TT) plants recovered from HIPOD with (TTWT) or without (TTadm) ADMETOS segregating pollen donors. Lower panel, herbicide-sensitive offspring of biparental diploid wild type (BD). (E and F) YFP fluorescence (E) and flow-cytometric analysis (F) of TTWT, TTadm and BD plants corresponding to the categories shown in (D). Scale bars, 1 mm (A), 5 mm (B, D), 50 μm (E). Figure 3—source data 1 Comparison of viable triploids recovered from polyspermy and interploidy crosses (2n × 4n). https://cdn.elifesciences.org/articles/52976/elife-52976-fig3-data1-v1.xlsx Download elife-52976-fig3-data1-v1.xlsx Previous results suggested that the egg cell block is stricter than the central cell block (Grossniklaus, 2017; Scott et al., 2008) and fertilization of the two female gametes during monospermy has been shown to occur in a coordinated fashion (Kawashima and Berger, 2011; Hamamura et al., 2011). Our unprecedented finding that most polyspermy-derived embryos develop in the presence of a monospermy-derived endosperm show that polyspermy enables selective polyploidization of the egg cell and concomitant bypassing the triploid block. Polyspermy-derived triparental plants are comparable to triploids generated by interploidy crosses The transcript profile of triploid plants has been characterized previously and remarkably few differences were found with respect to their cognate diploid controls (Hou et al., 2018). Polyspermy-induced plants differ from triploids derived from interploidy crosses as they inherit two sperm cytoplasms and, as shown in this work, mostly develop in a seed characterized by identical ploidies in embryo and endosperm. Given their special mode of origin, we aimed to compare the transcript profile of polyspermy-induced triparental triploids (TT) with that of biparental triploids (BT). In addition, we compared the transcriptome profile of BT plants with that of biparental diploids (BD) to identify ploidy-dependent changes in the transcriptional landscape. We used ein3 mutants as pollen acceptor as they have previously been shown to attract supernumerary pollen tubes (Völz et al., 2013). We performed RNAseq on five plants 18 days after sowing (DAS) each from BD, BT, and TT (Figure 4—figure supplement 1A). We chose this early state as we expected potential differences to become established during seed development, which differs in the three settings. In order to assess the quality of the transcriptome data, and to identify potential transcriptome-wide differences between the mRNA profiles of the three groups of plants, we performed a hierarchical clustering, a two-dimensional principal component analysis (PCA), and a two-dimensional multidimensional scaling (MDS) analysis of the fifteen 21,450-dimensional normalized expression profiles. We found that Spearman’s correlation coefficient c was greater than 0.94 for all of the 105 pairs of profiles (Figure 4—figure supplement 1B). Notably, the sub-trees of the dendrogram obtained by hierarchical average linkage clustering did not correspond to the five biological replicates from the same genotype (Figure 4A), and the 15 samples showed a high overlap in the PCA (Figure 4B) and MDS (Figure 4—figure supplement 1C) plots. These results not only reflect the high quality of the transcriptome data but also suggest a high similarity of the 15 transcriptome profiles compared. Figure 4 with 4 supplements see all Download asset Open asset Characterization of triparental triploid plants. (A) Dendrogram from hierarchical clustering of the 3 × 5 biological replicates of biparental diploid (BD), biparental triploid (BT), and triparental triploid (TT) recovered from polyspermy plant samples. The 15 × 15 matrix of Spearman’s correlation coefficients c was computed from the fifteen regularized log transformed expression profiles. The dendrogram was computed by average linkage clustering (UPGMA) of the 15 × 15 dissimilarity matrix with elements 1 – c. (B) Two-dimensional principle component analysis (PCA) of the fifteen 21,450-dimensional regularized log transformed expression profiles. The points represent the biological replicates, while the shapes resulting from connecting biological replicates from the same genotype highlight the similarity between the transcriptome profiles. (C) Distribution of ploidy level in the aneuploidy swarms produced by a triparental triploid plant. 2.0 in DNA index represents near diploids, 3.0, near triploids, and 4.0, near tetraploids. The gray areas indicate the intermediate ploidies. Each orange triangle represents an F2 plant derived from a triparental triploid. (D) Flow cytometric analysis of tetraploid progeny plants in F3 generation. 2n and 4n represent diploid and tetraploid controls. (E) Analysis of flowering time window of triparental triploid plants recovered from a three accession cross (TT3), Col-0, Ler and C24 during bolting stage. Scale bar, 1 cm. The black bold line represents the flowering period of different accessions. The three red bold lines bordered by the gray dashed lines label the day gaps between flowering termination of the parents and flowering initiation in the TT3. We first addressed, if there were genes with ploidy-specific expression changes by comparing transcript profiles of BD and BT. This comparison did not yield a single gene with a statistically significant differential expression. This result is in support of previous transcriptome profiling approaches that have uncovered remarkably few changes in plants with different ploidy (Pignatta et al., 2010; Riddle et al., 2010; Stupar et al., 2007; Wang et al., 2006; Yu et al., 2010). We next compared the transcript profiles of BT and TT in order to identify specific expression changes potentially associated with polyspermy. Interestingly, also this approach did not yield genes with statistically significant differential expression. The similarity in the overall transcriptional profiles between TT and BT plants is reflected by strong similarities in various life-history traits, including flower organ size, cell size, and even fertility (Figure 4—figure supplement 2). Please note that the normalized expression data presented here does not allow any conclusions on alterations in transcriptome size, i.e. changes affecting the total number of transcripts per cell (Coate and Doyle, 2015). However, the data suggests that Arabidopsis responds in a transcriptionally balanced fashion to the inheritance of supernumerary genomes and seed homoploidy. Along these lines, also the ability to generate tetraploid offspring within a single generation, a parameter that has been described previously for interploidy cross-induced triploids was maintained: To assess whether polyspermy-derived triploids can segregate stable polyploid offspring, we harvested the seeds of polyspermy-derived triparental plants and propagated them in two successive generations. The progeny of polyspermy-derived triploids segregates a complex swarm of karyotypes, similar to what has previously been described for interploidy crosses (Henry et al., 2005). On the basis of flow cytometric analysis, we grouped the plants into five different categories: near-diploids, 2n-3n, near-triploids, 3n-4n and near-tetraploids (Figure 4C, Figure 4—figure supplement 3A). Already in the F2 generation 5 out of 109 plants were found to fall into the near-tetraploids category, while 22 plants segregated a diploid-like profile (Figure 4C, Figure 4—figure supplement 3B). To determine whether any of the high-ploidy plants represented a genuine tetraploid, we collected seeds and determined the ploidy of 20 offspring per individual F2 plant. Flow cytometric analysis revealed that 3 out of 5 near-tetraploid F2 plants segregated exclusively plants that exhibit a ploidy profile characteristic to 4n plants (Figure 4D). This result was confirmed in the F4 generation, which, again, revealed a homogenous tetraploid ploidy profile. Together these results show that polyspermy-derived triploid plants have the potential to generate stable tetraploid and diploid offspring within a single generation. Polyspermy-derived three accession hybrids are reproductively isolated from their parental lines We previously combined three distinct Arabidopsis accessions in a three parent cross (Nakel et al., 2017) and noticed that the resulting triploid hybrids initiate flowering later than their parents, an effect which was previously described also for two-accession hybrids (Groszmann et al., 2014; Moore and Lukens, 2011) (Figure 4—figure supplement 4A). To address whether this phenotype has the potential to reduce gene flow, we compared flowering time between the parental line and triparental three accession hybrids, henceforth referred to as TT3. We found that flowering is induced more than one month later in TT3 than in the parental lines, and the flowering period of TT3 is completely isolated from the parents (Figure 4E). Under our plant growth conditions, Ler and Col-0 start to flower 24.4 ± 0.7 and 26.6 ± 1.0 DAS, and flowering terminates 44.9 ± 1.2 and 47.1 ± 0.9 DAS, respectively. This corresponds to a flowering period of around 20 days. The C24 flowering window is comparable but flower initiation is delayed by five days (31.8 ± 1.2 DAS). By contrast, TT3 plants initiate flowering only after 64.6 ± 5.4 days, which is around 20, 17 and 15 days after the respective parental lines have terminated their flowering phase (Figure 4E; Figure 4—figure supplement 4B). Even though these data are obtained under optimized growth condition, the results suggest that polyspermy-derived triploid three accession hybrids are reproductively isolated from the parental plants in the first generation. Discussion The triploid block is an established and widely distributed postzygotic hybridization barrier. In light of its fatal consequence, it has been suggested that there are ways to overcome this hybridization barrier (Köhler et al., 2010). We here established a novel polyspermy detection assay that allows to identify and characterize developing embryos resulting from supernumerary sperm fusion. With this tool we were able to show that most polyspermy-derived plants develop from seeds resulting from selective egg cell polyploidization. In those seeds, supernumerary paternal copies are only transmitted to the embryo, thereby bypassing the triploid block of the endosperm. Our results expand previous polyploidization concepts, which state that the increase in genome copies is caused by infrequent meiotic or mitotic defects. In fact, the currently favored route towards polyploid plants involves unreduced male gametes; however, this scenario introduces supernumerary paternal copies also to the endosperm, which is not tolerated in many plants resulting in seed abortion (Dilkes et al., 2008; Ramsey and Schemske, 1998; Scott et al., 1998; Stoute et al., 2012). In fact, this endosperm-related triploid block is considered a means of reproductive isolation (Köhler et al., 2010; Ramsey and Schemske, 1998). Plant polyploidization via polyspermy, by contrast, often affects the embryo-derived seed fraction only and hence has the potential to bypass the triploid block. It will be a challenge for the future to determine whether and to what extent polyspermy is relevant in nature and contributed to the evolution of polyploid plants. From an evolutionary and agricultur

This Article contained errors in Fig. 3 that were brought to our attention by the authors during the production process but, inadvertently, were not corrected before publication. The tick marks on the y-axis in panels b, f, and k, and the median line in the box-and-whisker plot for biparental diploid plants (BP) in panel i were shifted downwards by up to 2 mm. This has now been corrected in both the PDF and HTML versions of the Article

It is considered an inviolable principle that sexually reproducing organisms have no more than two parents and fertilization of an egg by multiple sperm (polyspermy) is lethal in many eukaryotes. In flowering plants polyspermy has remained a hypothetical concept, due to the lack of tools to unambiguously identify and trace this event. We established a highthroughput polyspermy detection assay, which uncovered that supernumerary sperm fusion does occur in planta and can generate viable polyploid offspring. Moreover, polyspermy can give rise to seedlings with one mother and two fathers, challenging the bi-organismal concept of parentage. The polyspermy derived triploids are taller and produce bigger organs than plants resulting from a regular monospermic fertilization. In addition, we demonstrate the hybridization potential of polyspermy by instantly combining three different Arabidopsis accessions in one zygote. Our results provide direct evidence for polyspermy as a route towards polyploidy, which is considered a major plant speciation mechanism.