ERECTA receptor-kinases play a key role in the appropriate timing of seed germination under changing salinity.

Appropriate timing of seed germination is crucial for the survival and propagation of plants, and for crop yield, especially in environments prone to salinity or drought. However, the exact mechanisms by which seeds perceive changes in soil conditions and integrate them to trigger germination remain elusive, especially once the seeds are non-dormant. In this study, we determined that the Arabidopsis ERECTA (ER), ERECTA-LIKE1 (ERL1), and ERECTA-LIKE2 (ERL2) leucine-rich-repeat receptor-like kinases regulate seed germination and its sensitivity to changes in salt and osmotic stress levels. Loss of ER alone, or in combination with ERL1 and/or ERL2, slows down the initiation of germination and its progression to completion, or arrests it altogether under saline conditions, until better conditions return. This function is maternally controlled via the tissues surrounding the embryo, with a primary role being played by the properties of the seed coat and its mucilage. These relate to both seed-coat expansion and subsequent differentiation and to salinity-dependent interactions between the mucilage, subtending seed coat layers and seed interior in the germinating seed. Salt-hypersensitive er105, er105 erl1.2, er105 erl2.1 and triple-mutant seeds also exhibit increased sensitivity to exogenous ABA during germination, and under salinity show an enhanced up-regulation of the germination repressors and inducers of dormancy ABA-insensitive-3, ABA-insensitive-5, DELLA-encoding RGL2, and Delay-Of-Germination-1. These findings reveal a novel role of the ERECTA receptor-kinases in the sensing of conditions at the seed surface and the integration of developmental, dormancy and stress signalling pathways in seeds. They also open novel avenues for the genetic improvement of plant adaptation to changing drought and salinity patterns.

Introduction

Seed germination is a vital life-cycle transition in plants. When and under what conditions it occurs largely determine plant survival, reproductive success, yield, and ability to colonise new areas. To maximise the chances of successful completion of the life cycle and production of viable offspring, seeds have evolved mechanisms for the induction of a quiescent, dormant state during late maturation and desiccation on the mother plant. These mechanisms block the capacity of fresh seeds to germinate for a certain period of time under any combination of the same environmental conditions that would be permissive of germination in non-dormant seeds, whether hydric, gaseous, temperature, or light (Baskin and Baskin, 2004). The vast majority of higher plants, including Arabidopsis, are endowed with ‘physiological’ dormancy. The ‘primary’ dormancy acquired on the mother plant gradually declines during dry storage at ambient temperature (a process commonly referred to as ‘after-ripening’) or under natural conditions as soil temperature increases; the seed becomes increasingly capable of responding to a suite of signals in its surroundings and will eventually germinate (e.g. see Finch-Savage and Footitt, 2017, for a review). In Arabidopsis, this absolutely requires water and occurs in two temporally separate steps, namely the rupture of the testa, or seed coat (a dead tissue), and then the rupture of the endosperm by the radicle of the expanding turgid embryo (Liu ; Müller ). Embryo reactivation and the weakening of surrounding tissues are tightly coordinated through complex biochemical and hormonal pathways, with a prominent role of abscisic acid (ABA) and gibberellins (GAs) in interaction with ethylene, brassinosteroids, and reactive oxygen species (ROS) (Koornneef and Van der Veen, 1980; Steber and McCourt, 2001; Finkelstein ; Liu ). ABA inhibits germination whereas GAs promote it through regulation of inter-signalling between the seed coat, endosperm, and embryo in a feedback loop involving DELLA proteins and interactions with cell-wall remodelling enzymes (Müller ; Stamm ; Graeber ; Nonogaki, 2014).

Drought and salinity stress are two inter-related and widespread conditions in natural environments, and are major causes of germination failure, poor crop establishment, and yield loss (Boyer, 1982; Bradford, 1990; Yamaguchi and Blumwald, 2005; Finch-Savage and Leubner-Metzger, 2006; Munns and Tester, 2008). The high vulnerability of seeds to these stresses has long been recognised and yet the molecular controls remain poorly understood, apart from evidence for a deregulation of ABA–GA homeostasis and an impairment of ethylene and ROS signalling (Lopez-Molina ; Kim ; Yuan ; Yu ). Natural genetic variation in seed germination under optimal conditions, drought, or salinity has been widely documented, and numerous QTLs have been identified (e.g. Quesada ; Clerkx ; Galpaz and Reymond, 2010; Wang ; DeRose-Wilson and Gaut, 2011; Yuan ). This demonstrates the potential for genetic improvement, but also the complexity of the underlying molecular pathways. While the genetic dissection of seed dormancy has received much attention, very few genes have been demonstrated to control germination in non-dormant seeds to tune it to the prevailing soil conditions (Kim ; Ren ; Yu ). Little is known about how seeds monitor their surroundings, how this information is communicated to their inner compartments, and how the intricate communication between these compartments and the environment that is required for timely germination is modulated (Donohue ).

Receptor-like protein kinases (RLKs) at the cell plasma membrane play major roles in signal perception and transduction to downstream intra- and intercellular signalling networks. A vast array of RLKs are encoded by plant genomes (Shiu ). Among them are leucine-rich-repeat receptor-like kinases (LRR-RLKs), which form a large family of receptor proteins characterised by an extracellular receptor domain, a trans-membrane domain, and an intracellular kinase domain for signal transduction through phosphorylation cascades. The few that have been characterised provide evidence for their central functions in integrating signalling pathways associated with development, hormones, abiotic stress, and defence (Becraft, 2002; Osakabe ). Little information is available on RLKs in seeds, even though developing seeds show a high abundance of secreted peptides and studies have pointed to the importance of peptide-mediated signalling in inter-compartmental coordination during seed development (Ingram and Gutierrez-Marcos, 2015).

The Arabidopsis ERECTA gene family encodes three closely related LRR-RLKs, namely ERECTA (ER), ERECTA-like 1 (ERL1), and ERECTA-like 2 (ERL2), that are known to synergistically regulate many aspects of plant development and morphogenesis and that play prominent roles in organ shape, stomatal patterning, cell proliferation, and meristematic activity (Torii ; Shpak et al., 2004, 2005; Pillitteri ; Uchida ; Bemis ; Etchells ), as well as being involved in some pathogenic responses (Godiard ; Llorente ; Jordá ). In contrast, little is known of their function in abiotic stress responses, beyond a role in leaf heat tolerance (Shen ). We have previously reported a role of ERECTA as a major controller of water-use efficiency, under both well-watered and drought conditions (Masle ). This function appears to be broadly conserved in diverse species (Xing ; Zheng ) and is suggestive of an important adaptive role of the ERECTA family to abiotic stress. Here, we examine the function of the ERECTA family during germination, a key switch that is extremely sensitive to variations in the osmotic and ionic conditions in the soil, both of which vary widely in nature.

Material and methods

Plant material and growth conditions

Arabidopsis thaliana Columbia (Col-0, CS1093) was used as the wild-type (WT), together with the previously described erecta family homozygous mutant lines er105, erl1-2, erl2-1, the double-mutants er105 erl1-2, er105 erl2-1, erl1-2 erl2-1, and the triple-mutant er105 erl1-2 erl2-1 (Torii ; Shpak ). For simplicity, in figures we refer to er105, erl1-2, and erl2-1 simply as er, erl1, and erl2, respectively. We also used the independent homozygous er2 (Rédei, 1962; Masle ), and novel erl1-5 (SALK_019567), and erl2-2 (SALK_015275C) mutants, and a set of double- and triple-mutants that we generated through crosses and PCR genotyping for the presence/absence of TDNA inserts and mutated alleles (primer sequences are listed in Supplementary Table S1 at JXB online). The er105 mutant was obtained by fast-neutron irradiation; it carries a large DNA fragment of unknown origin within the ER gene (At2g26330), inserted between +5 and +1056 (Torii ) and has been characterised extensively (Torii ; Shpak et al., 2004, 2005; Masle ). The er2 mutant (line CS3401 from the Nottingham Arabidopsis Stock Centre) was first identified in an X-ray irradiation mutagenesis experiment for its reduced, compact stature (Rédei, 1962). It carries a frameshift in the middle of the kinase domain, leading to a truncated gene product. The er105 and er2 mutants phenocopy each other and are fully complemented by expression of the native Col-0 ER allele (Masle ). The single erl mutants in the ERECTA-like 1 (ERL1, At5g62230) or ERECTA-like 2 (ERL2, At5g07180) genes are knock-out TDNA insertional mutants that were sourced from the Arabidopsis mutant collections and isolated by PCR genotyping. The location of the single TDNA insertion carried by these mutants was determined by sequencing, and it maps to the LRR-receptor domain in erl1-2, erl2-1 (Shpak ), and erl2-2 (SALK_015275C, Supplementary Fig. S3), or to the beginning of the kinase domain in erl1-5 (SALK_019567, Supplementary Fig. S3). Measurements of target gene expression showed that TDNA insertions totally abolished gene transcription (Shpak ; Supplementary Fig. S3). The presence of no more than a single TDNA insert in each single mutant was ascertained through segregation analysis of the TDNA allele or of antibiotic resistance in the previous segregating generation from which these homozygous mutants were isolated. Consistently, the double-mutants er-2 erl1-5, er-2 erl2-2, and triple-mutant er-2 erl1-5 erl2-2 exhibited the same morphological and developmental phenotypes as those previously described in the er105 erl1-2, er105 erl2-1, and er105 erl1-2 erl2-1 mutants, respectively. These include reduced stature, compact inflorescences, broad and blunt siliques, and, in the triple-mutant, extreme dwarfism and sterility (Shpak ; Bemis ; Supplementary Fig. S3).

As the er105 erl1.2 erl2.1 and er2 erl1-5 erl2-2 plants are sterile, the segregating progeny of er105 erl1.2+/– erl2.1 or er2 erl1-5+/– erl2-2 were used to investigate the germination of the triple-mutant seeds, and are referred to as er105 erl1.2/seg and erl2.1 and er2 erl1-5/seg erl2-2, respectively.

All the seeds in any given experiment were of the same age, and were stored together under the same conditions after being harvested from plants that were grown together in the same growth chamber, which was set at 21 °C constant temperature with a 12-h or 16-h light period depending on experiment, at 120–130 μmol quanta m−2 s−1 light intensity. To investigate the effects of parent-of-origin on seed germination and seed size, seeds were manually excised from mature siliques of the same age, produced from flowers that were tagged at fertilisation at similar positions on the primary inflorescence.

Germination assays

All assays were done using seeds stratified by moist-chilling at 4 °C to remove residual dormancy. Seeds were surface-sterilised and sown on 0.7% agar media supplemented with Hoagland’s nutrient solution [2 mM KNO3, 5 mM Ca(NO3)2.4H2O, 2 mM MgSO4.7H2O, 2 mM KH2PO4, 0.09 mM Fe-EDTA, and micronutrients) at pH 5.8, and with NaCl or KCl at the desired experimental concentrations. For germination assays under iso-osmotic conditions, seeds were placed on filter paper imbibed with solutions of either NaCl or PEG8000 dissolved in water at a range of concentrations calculated to provide the same media osmotic pressures. The osmotic pressure (π e) of the basal medium containing either NaCl or KCl was calculated using the van’t Hoff equation and verified experimentally using a VAPRO vapour pressure osmometer (Wescor Inc.). The concentrations of PEG8000 required to obtain a given π e were determined from a calibration curve of π e as a function of PEG concentration using the same vapour pressor osmometer. Seeds of the WT and all erecta mutant combinations were sown in equal numbers (≥33) in each of 3–4 plates (total n=100–120 seeds per line per treatment and experiment). After stratification at 4 °C in the dark for 2–3 d, the plates were exposed to continuous light (100–115 μmol quanta m−2 s−1) and a constant temperature of 21 °C. ‘Demucilaged’ seeds were sown immediately after removal of the mucilage (see below) and kept at 4 °C in darkness for an additional day, so as to keep the total stratification time at 48 h, as for intact control seeds.

Seeds were individually scored for both testa and endosperm rupture (germination sensu stricto) under a binocular microscope within the growth chamber, at 3–4 h intervals until all seeds on the control plates (0 mM NaCl) had germinated (30 h at most), or 1–3 times daily on the NaCl, KCl, or PEG plates as appropriate, until no change in scores were observed. Data are represented either as percentages of seeds exhibiting testa or endosperm rupture as a function of incubation time post-stratification, or as T50 values, which corresponds to the time (h) post-stratification when 50% of seeds showed testa or endosperm rupture (Bewley ).

Embryo culture

Dry seeds were pre-imbibed with water for 1–2 h, briefly rinsed twice with water to remove endosperm debris, plated on either 0 mM or 150 mM NaCl media, and then placed in the dark at 4 °C for 3 d. The plates were then transferred to the growth chamber and mature embryos were subsequently excised. Embryos were individually imaged at the time of transfer and again 72 h later using a M205 FA microscope fitted with a DFC 550 camera (both Leica). The relative embryo expansion rates over the 72-h interval were calculated from measurements of projected areas using the ImageJ software (https://imagej.nih.gov/ij/).

Staining procedures

GUS histochemical staining of seeds from proERf::GUS reporter lines (Shpak ) was performed on embryos dissected from dry and germinating seeds sampled from the 0 mM and 150 mM NaCl plates. Staining was done as described by Sessions .

For tetrazolium permeability assays (Debeaujon and Koornneef, 2000), dry seeds were incubated in the dark in an aqueous solution of 1% (w/v) tetrazolium red (2,3,5-triphenyltetrazolium chloride, Sigma-Aldrich) at 30 °C for either 4, 24, 48, 72, or 120 h and then rinsed twice with deionised water, resuspended in 95% ethanol, and quickly ground to extract formazans. The final volume was adjusted to 2 ml with 95% ethanol, followed by centrifugation at 15 000 g, and measurement of the absorbance of the supernatant at 485 nm using an Infinite M1000 Pro spectrophotometer (Tecan). Each sample was assayed with three biological replicates.

Ruthenium red staining of the mucilage was performed as described by McFarlane . Ruthenium red stains acidic pectins (Hanke and Northcote, 1975) and is widely used to stain Arabidopsis seed mucilage (Western ; Penfield ).

Toluidine Blue permeability tests were done essentially as previously described (De Giorgi ; Loubéry ). Briefly, seeds were placed on filter paper imbibed with a solution of either 5 µM ABA or 100 µM paclobutrazol (PAC) with either 0 mM or 100 mM NaCl. Large Petri dishes were used so that all genotypes could be compared within each individual dish. At 36–48 h after the start of the imbibition, seeds were transferred to a solution of the same composition but also containing Toluidine Blue (0.05%, w/v), in which they were incubated for 7 h. Embryos were then gently excised from the seeds and immediately imaged using a M205 FA stereomicroscope fitted with a DFC 550 color camera (both Leica).

Profiling of fatty acid methyl esters derived from lipids stored in the embryo

After 1 h imbibition in water, 50 mature embryos were dissected from dry seeds, with four replicates per genotype. Fatty acid methyl esters (FAMEs) were prepared by direct transesterification as described by James . The embryos were placed in 1.5-ml Reacti-Vials (ThermoFisher Scientific) fitted with Teflon-lined caps. To each vial 50 µl CHCl3 was added followed by the internal standard, heptadecanoic acid (C17:0, 15 µl, 9.66 mg in 25 ml CHCl3), and methanolic HCl (3 M, 500 µl). The samples were mixed and heated at 90 °C for 60 min, and then allowed to cool before being washed into glass tubes with CHCl3. Water (1 ml) was added to each tube and the FAMEs were extracted using hexane:chloroform (4:1 v/v, 3×1 ml). The extracts were combined and washed with water (200 µl). The organic phase was then dried with anhydrous Na2SO4, decanted, and evaporated under nitrogen. The residue was dissolved in CH2Cl2 (150 µl) and transferred to GC/MS auto-sampler vials for analysis.

Mucilage extraction and analysis

Mucilage extraction was performed on 40-mg samples of dry seeds. Each sample (n=4 per genotype per experiment) was suspended in 1 ml of milliQ water, shaken at 500 rpm for 24 h at 4 °C, then vortexed for 5 s, and centrifuged at 8000 g for 3 min. A 600-µl volume of the supernatant was recovered. The seeds were then rinsed twice with 200 µl water, each time followed by vortexing, centrifuging, and recovery of 200 µl of the supernatant each time. The pooled supernatants (1 ml total volume) were flash-frozen in liquid nitrogen and immediately lyophilised. The mucilage thus recovered was weighed on a 10–6 g high-precision micro-balance. Because of the degree of genetic variation in seed size that we observed, sub-samples of a known number of seeds (at least 500) were weighed, imaged at high resolution, and analysed for size using ImageJ prior to mucilage extraction, thus allowing the average amount of mucilage per seed to be calculated. Reductions of uronic acid methyl-esters and free uronic acids in the extracted mucilage were carried out following established protocols (Kim and Carpita, 1992; Pettolino ). The reduced polysaccharides were then hydrolysed, reduced, acetylated, and subjected to GC/MS analysis as described by Peng .

Analysis of seed sodium content

Dry seeds (three biological replicates of 10 mg each per genotype and treatment) were imbibed and stratified at 4 °C in the dark in 0 mM or 150 mM NaCl for 2 d followed by 24 h at room temperature with shaking. The seeds were rinsed three times with 2 ml water, freeze-dried, weighed, and microwave-digested for 2 h in 4 ml of 20% nitric acid at 175 °C (Method 3051; US EPA, 2007). The digests were diluted to a final volume of 5 ml and sodium ions were measured using inductively coupled plasma optical emission spectrometry (Vista-Pro CCD Simultaneous ICP-OES, Varian).

Quantitative RT-PCR

Total RNA was extracted from dry, imbibed or germinating seeds using TRIzol reagent (Invitrogen). mRNA isolation and reverse-transcription were done as described by Branco and Masle (2019); primer sequences are given in Supplementary Table S1. Analyses were carried out on samples of 300 seeds, with four biological replicates per genotype, time-point, and treatment (media with 0 mM or 150 mM NaCl). The seeds were sampled from four plates where all the genotypes were present together. The expression levels were normalised to the geometric mean of the expression levels of four reference genes, namely APT1 (At1g27450), PDF2 (At1g13320), bHLH (At4g38070), and PPR (At5g55840), which were chosen on the basis of their very stable expression across tissues, developmental stages, and growth conditions, including abiotic stresses (Czechowski ). Gene expression was measured just before sowing (‘dry’ seeds), at the end of seed imbibition and stratification (germination stage I), 20 h post-stratification (stage II, testa rupture), and then 72 h post-stratification (stage III-G, when endosperm rupture had completed on control media; seeds on 150 mM NaCL that had not germinated at this time were analysed separately and are referred to as stage III-NG). Seeds were sampled within the cold room or the growth room (dry seeds and stages I–III, respectively) within 5 min from start to finish for each plate, and were immediately flash-frozen in liquid nitrogen. The experiment was repeated three times.

Statistical analysis

Statistical analysis was performed using the Statistix 9 software (Analytical Software, Tallahassee, USA). For multivariate comparisons of the mucilage composition profiles, discriminant orthogonal projected latent structure (OPLS) analysis was carried out using the SIMCA software (Umetrics, www.umetrics.com) with salinity as a quantitative variable.

Accession numbers

The accession numbers of the genes used in this study are as follows: AtER (At2g26330), AtERL1 (At5g62230), AtERL2 (At5g07180), AtPDF2 (At1g13320), AtbHLH (At4g38070), AtPPR (At5g55840).

Results

The ERECTA genes control the timing and speed of germination in response to changing salinity and osmotic conditions

Loss of function of ER/ERL had no effect on testa or endosperm rupture in 0 mM NaCl media except for er105 erl1.2 seeds, which showed a small but consistent lag in testa rupture (Fig. 1, Supplementary Fig. S1A, B) that persisted through to the next germination phase leading to radicle protrusion. Salinity delayed germination in a dose-dependent manner, as expected (Supplementary Fig. S2A, B), but with striking differences among the genotypes (Fig. 1, Supplementary Fig. S1C, D). Wild-type (WT), erl1.2, erl2.1, and erl1.2 erl2.1 seeds germinated first, ahead of er105, er105 erl2.1, er105 erl1.2, and finally er105 erl1.2/seg erl2.1 seeds, due to both delayed testa rupture and slower progression to endosperm rupture. As the ERECTA family has never previously been implicated in the control of seed germination, the experiments were repeated with an independent set of knock-out mutants carrying the er2, erl1-5, and erl2-2 null mutations (see Methods and Supplementary Fig. S3), alone or in combination, and similar results were obtained (Supplementary Fig. S4A). Taken together, these results unambiguously established that the observed genetic differences in seed germination in response to external NaCl concentrations were causally related to disruption of the ERECTA genes.

Fig. 1.

The three Arabidopsis ERECTA family members synergistically control the timing and speed of seed germination under salinity. (A, B) T50 values (h post-stratification to rupture in 50% of seeds) for testa rupture (A) and endosperm rupture (B). (C) Time interval between testa rupture and endosperm rupture. The experiment was repeated five times with different seed batches and similar results were obtained. WT, wild-type (Col-0). As the triple-mutant is sterile, the segregating progeny of er105 erl1.2+/– erl2.1 plants were used to investigate germination, and is referred to as er erl1.2/seg erl2.1 (note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure). Data are means (±SE) from n=4 plates, with 30 seeds per genotype per plate. Different letters indicate significant differences as determined by two-way ANOVA and Tukey’s HSD pair-wise tests (P<0.001).

We next examined the effects of imposing salinity stress after stratification or after germination. Similar germination kinetics and differential sensitivity to NaCl among the genotypes were observed regardless of whether seeds were subjected to salinity stress post-stratification or directly from sowing (Fig. 2A–D). However, when exposed to 150 mM NaCl after radicle protrusion, all the genotypes displayed similar sensitivity to the salinity stress (Fig. 2E). These data demonstrate a germination-specific function of the ERECTA genes in the sensing and signalling of salinity. This function requires ER but involves the three family members in a non-totally redundant manner.

Fig. 2.

Germination-specific functions of the ERECTA genes in the control of germination sensitivity to salinity. (A–D) Time-course of endosperm rupture for the wild-type (WT), er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds over a 10-d incubation period on agar media containing either 0 mM NaCl (A, C) or 150 mM NaCl (B, D) following imbibition and stratification either directly on the media (A, B) or in water prior to plating (C, D). (E) Seedling relative expansion rates on 0 mM or 150 mM NaCl media. Seeds were first germinated on NaCl-free media and then transferred to fresh 0 mM or 150 mM NaCl plates and seedling expansion was then measured over the next 72 h. Measurements of the whole-seedling projected area were made on captured images using the ImageJ software. Note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure. All data are means (±SE), n=7. Different letters indicate significant differences as determined by two-way ANOVA and Tukey’s HSD pair-wise tests (P<0.001).

ERECTA family expression during seed germination has not previously been reported. We therefore examined promoter activity in transgenic seeds expressing proERf::GUS constructs (Supplementary Fig. S5). The expression patterns of the ERECTA gene family did not appear to be influenced by salinity but they differed among members, with expression of ERL2 only seen in the cotyledons and the shoot apical meristem, while ER and ERL1 promoter activities were also detected in the hypocotyl. Measurements of transcript abundance by RT-qPCR (Fig. 3) confirmed the presence of transcripts of the ERECTA genes in dry seeds and showed a strong and early induction of ER and ERL1 expression during stratification and imbibition (germination phase I) and during the next phase (stage II) leading to testa rupture, while ERL2 had low expression. Salinity induced expression of ER, especially during germination phase III leading to radicle protrusion, but it had little influence on ERL1 or ERL2. These results support a role of the three ERECTA family members throughout germination, with specificity among them.

Fig. 3.

Transcripts of Arabidopsis ER, ERL1 and ERL2 genes are present in mature dry seeds and de novo transcription is activated early during germination. Gene expression was measured in dry seeds and in germinating seeds at the end of the stratification period (stage I), then 20 h later (stage II, testa rupture), and then after an additional 52 h (stage III-G, when endosperm rupture had completed on control media). Seeds on 150 mM NaCl that had not germinated by stage III were sampled and analysed separately, and are labelled as III-NG. Data are means (±SE) from n=4 pooled samples of 300 seeds per genotype and treatment. The experiment was repeated three times with similar results. Different letters indicate significant differences as determined using two-way ANOVA and Tukey’s HSD pair-wise tests (P<0.001).

Salinity induces both osmotic and ionic stress (Munns and Tester, 2008). To investigate the contributions of these two components, we next examined germination responses to polyethylene glycol (PEG)8000, a high molecular weight non-permeating osmoticum that mimics drought-induced osmotic stress, in the salinity hyper-sensitive mutants er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1. Under equivalent external osmotic conditions, seed germination was significantly less inhibited by PEG than by NaCl. The effect of PEG was innocuous up to an osmotic pressure in the medium (π e) of 0.5 MPa (Fig. 4A). When PEG was present at higher concentrations (π e=0.74–0.99 MPa; equivalent to 150 mM and 200 mM NaCl, respectively), germination was slowed down, and to a greater extent in the double- and triple-mutants than the WT. However, the delay in germination was mild, of the order of 1 d. By Day 3 post-stratification, even under 0.99 MPa with PEG, germination was either complete (WT, er105, er105 erl1.2, and er105 erl2.1 seeds) or nearly complete (90%, er105 erl1.2/seg erl2.1 seeds), (Fig. 4A, B), which was in contrast to the strong inhibition observed at the same π e with 200 mM NaCl (Fig. 4C, Supplementary Fig. S2A, B). Such severe inhibition was only observed at much higher PEG concentrations, and even then some seeds still germinated (Fig. 4B). Taken together, these data indicate that within the germination-permissive range of NaCl concentrations, the ERECTA genes modulated the germination sensitivity to salinity mostly via interactions with the ionic effects of NaCl, but are also involved in the control of germination sensitivity to osmotic and hyperosmotic stress.

Fig. 4.

The Arabidopsis ERECTA family regulates seed germination sensitivity to salinity mostly via interactions with ionic effects, but is also involved in the control of germination under osmotic stress. (A) Percentage of seeds with endosperm rupture for the wild-type (WT) and the salt-hypersensitive mutants er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 at different times post-stratification as a function of the osmotic potential of the media (π e), which was varied using PEG8000 at concentrations ranging from 0–171 g l−1. Note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure. Data are means, n=3 plates. (B) Germination response over an extended range of PEG concentrations, in an independent experiment with a different seed batch. Data are means of n=3 plates, and show the percentage of seeds exhibiting endosperm rupture at 6 d post-stratification. (C) Kinetics of seed germination under 0.99 MPa π e induced by NaCl, using the same seed batch as in (B). The arrow indicates germination scores on day 6 when at least 90% of seeds had germinated under the same osmotic conditions induced by PEG, as shown in (B). Data are means of n=3 plates, with 30 seeds per plate and per genotype. The experiments were repeated three times with similar results.

The NaCl-hypersensitive mutants also exhibited increased sensitivity to KCl, but to a much lower extent under equivalent osmotic conditions (Fig. 5). This indicates that the function of the ER genes in seed germination under salinity was predominantly related to effects of the sodium ion.

Fig. 5.

Seed germination in the salt-hypersensitive Arabidopsis mutants er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 is more sensitive to external NaCl than KCl under iso-osmotic conditions. Time-courses of germination under two different osmotic potentials (π e) are shown, induced by supplementation of the media with either NaCl or KCl. Data are means (±SE) of n=3 plates, with 30 seeds per plate and per genotype. The experiments were carried out twice with similar results. Note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure.

It was notable that while all or the vast majority of WT, erl1.2, erl2.1, and erl1.2 erl2.1 seeds plated on NaCl medium eventually germinated (90–100%, similar to salt-free media), a significant proportion of er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds failed to do so, even after a lengthy incubation period (Fig. 6, Supplementary Fig. S1). Among those, the majority (up to 70%) did not even exhibit testa rupture. To determine whether they were damaged or dead seeds, we transferred them to NaCl-free media. Most then germinated readily, within 20–25 h (Fig. 6), bringing the final percentage of germinated seeds to similar levels as observed for control seeds that had not been exposed to salt. Failure to germinate on saline media was therefore not due to irreversible cellular damage and loss of seed viability, but rather to a slower or halted progression of the germination process. Consistent with their maintained viability and fast germination upon release from salinity stress, seeds with arrested germination on saline media showed similar expression levels of the ERECTA genes to germinated seeds (ERL1 and ERL2) or even higher (ER) (Fig. 3, comparison of III-NG to III-G).

Fig. 6.

Seed germination in the salt-hypersensitive Arabidopsis mutants er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 readily resumes upon removal of the stress. Time-course of seed germination on media containing 150 mM NaCl (0–490 h) followed by transfer to NaCl-free media (arrow). The experiment was repeated three times with similar results. Data are means (±SE) of n=4 plates, with 30 seeds per genotype per plate. Different letters indicate significant differences as determined by one-way ANOVA and Tukey’s HSD pair-wise tests (P<0.05; NS, not significant). Note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure.

The ERECTA genes affect the regulation of seed germination by ABA and GA

Salinity and osmotic stress promote ABA signalling and biosynthesis during germination (Seo ; Piskurewicz ; Yuan ), and ABA is a strong inhibitor of germination. We therefore compared the germination kinetics of WT seeds and ERECTA-family mutants on media supplemented with ABA. ABA treatment consistently had a mild delaying effect on testa rupture, and this was most pronounced for er105 erl1.2 seeds (Fig. 7A). ABA strongly inhibited endosperm rupture in a genotype-dependent manner (Fig. 7B). Germination of er105 erl1.2 was the most sensitive to ABA, and was slower than for WT seeds even in the 1 µM ABA range. Under higher concentrations, seeds of the other two salt-hypersensitive mutants, er105 erl2.1 and er105 erl1.2/seg erl.12, but not er105, also showed differences from the WT, with enhanced ABA sensitivity; interestingly, this was also the case for seeds of erl1.2 erl2.1, which is not hypersensitive to salinity (Fig. 7B). These data indicate the involvement of both ABA-dependent and ABA-independent pathways in the ERECTA family-mediated sensitivity of seed germination to salinity.

Fig. 7.

The Arabidopsis ERECTA genes interact with the sensitivity of seed germination to exogenous ABA and with the expression of major ABA and GA signalling genes. Wild-type (WT) and the salt-hypersensitive Arabidopsis mutants er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 were examined (note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure). (A) Germination response to exogenous ABA application as indicated by T50 values (h post-stratification to rupture in 50% of seeds) for the testa (A) and the endosperm (B). Data are means (±SE) of n=3 plates, with 30 seeds of each genotype per plate. The experiment was repeated three times with similar results. (C) Expression of ABI3, ABI5, RGL2, and DOG1 genes in dry seeds and seeds sampled at the end of stratification (stage I), then 20 h later (stage II, testa rupture), and then after an additional 52 h (stage III-G, when endosperm rupture had completed on control media). Seeds on 150 mM NaCl that had not germinated by stage III were sampled and analysed separately, and are labelled as III-NG. Different letters indicate significant differences as determined using two-way ANOVA and Tukey’s HSD pair-wise tests (P<0.05).

The inhibitory effect of ABA on germination is antagonised by GAs (Koornneef ; Holdsworth ; Weitbrecht ; Liu ). Rather than the absolute levels, the balance of ABA/GA is key to the commitment of seeds to germinate. The DELLA RGL2 protein plays a pivotal role in the cross-talk between ABA and GA signalling in the imbibed seed. RGL2 acts as the main GA signalling repressor in germinating seeds, through activation of a number of transcriptional regulators, including ABI3 and ABI5 that are central effectors of ABA signalling, establishment of dormancy, and repression of seed germination (Lopez-Molina et al., 2001, 2002; Lee ; Piskurewicz et al., 2008, 2009; Liu ). ABI3 and ABI5 are also involved in the regulation of early seedling arrest of growth under water stress in Arabidopsis (Lopez-Molina et al., 2001, 2002) and in the reversible inhibition of germination under salinity in its halophytic relative Eutrema salsugineum (Kazachkova ). To better understand the interaction of the ERECTA genes with ABA regulation of seed germination, we examined the expression of ABI3, ABI5, and RGL2 in WT and er105 erl1.2/seg erl2.1 seeds. We also examined DELAY OF GERMINATION1 (DOG1), a pivotal seed dormancy gene that genetically interacts with ABI3 and with a central type 2C protein phosphatase of the ABA signalling pathway during germination, and that also regulates ABI5 expression (Dekkers ; Née ; Nishimura ). Constitutive expression levels were similar in the WT and mutant seeds (Fig. 7C). Salinity consistently caused an up-regulation of expression, and this was stronger in er105 erl1.2/seg erl2.1 than in the WT. This indicated that the salinity signalling cascade mediated by the ERECTA genes interacted with the ABA–GA signalling network of germination and dormancy. We also examined the expression of the ABA biosynthesis genes ABA2 and NCDE4 and the GA biosynthesis genes GA3OX1 and GAOX2, and found that none of them showed a differential response to salinity between the mutants and the WT (Supplementary Fig. S6).

The role of the ERECTA genes in seed germination partly overlaps with a role in determining seed size and is primarily maternally controlled

Seed germination occurs when the pressure exerted by the turgid expanding radicle of the embryo overcomes the mechanical resistance of the surrounding testa and micropylar endosperm (Linkies ; Nonogaki, 2014). As mature embryos of the er105 erl1.2 erl2.1 mutant have smaller cotyledons (Uchida ), we reasoned that reduced growth potential could be a factor in the observed delay in the emergence of their radicle under salinity and osmotic stress, and possibly that of the other salt-hypersensitive mutants. We first measured seed size as a surrogate for embryo size, since the Arabidopsis embryo occupies most of the seed volume. The seeds of all the salt-hypersensitive genotypes (i.e. er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1) were significantly smaller than those of the WT, erl1.2, and erl2.1 (Fig. 8A). Remarkably, however, the seeds of the non-salt-hypersensitive mutant erl1.2 erl2.1 were larger than those of the WT. These data revealed a novel function of the ERECTA genes in the determination of seed size, with specificity among them. Moreover, they suggested a link between the function of the ERECTA gene family in germination sensitivity to salinity and its influence on seed size. However, the fact that erl1.2 erl2.1 seeds germinated simultaneously with those of the WT in the presence or absence of salt despite their significant difference in size indicated that the link was not absolute.

Fig. 8.

The function of Arabidopsis ERECTA genes in seed germination sensitivity to salinity is maternally controlled and shows partial overlap with a function in the determination of seed size. The wild-type (WT) and the mutants er105, erl1.2, erl2.1, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 were examined (note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure). (A) Seed projected area. Data are means (±SE) of n≥400 seeds per genotype from 11 siliques. Different letters indicate significant differences as determined using one-way ANOVA and Tukey’s HSD pair-wise tests (P<0.001. (B) Relative expansion rate (mm2 mm−2 d−1) of excised mature embryos over a 72-h incubation period on either 0 mM or 150 mM NaCl media (n=7). Different letters indicate significant differences as determined using two-way ANOVA and Tukey’s HSD pair-wise tests (P<0.001. (C) Time-course of germination for WT and er erl1.2 seeds, and F1 seeds generated from their reciprocal crosses. Similar results were obtained from crosses between WT and er erl2.1 flowers (data not shown). (D) Size of F1 seeds from reciprocal crosses between WT and er erl1.2+/– erl2.1 flowers (n=86–143 seeds per cross). Different letters indicate significant differences as determined using one-way ANOVA and Tukey’s HSD pair-wise tests (P<0.001). Crosses in (C, D) were made between flowers at similar positions on the main inflorescence and seeds were harvested at the same time, 3 weeks after crossing.

We next considered the possibility of developmental defects in the smaller, salinity hypersensitive mutant seeds. As expected, homozygous er105 erl1.2 erl2.1 segregants displayed reduced, rounder cotyledons and a broader shoot apical meristem, as previously reported (Uchida ). However, their hypocotyls and embryonic roots were similar to the WT in length, number, and the size of constitutive cells (Supplementary Fig. S7).

Seeds reserves are essential for successful germination and are mostly stored in cotyledons in Arabidopsis. Smaller seeds and cotyledons suggest less reserves, which could be responsible for hypersensitivity to salinity and osmotic stress. To examine this, we quantified fatty acid methyl esters (FAMEs) derived from embryo lipids, which constitute the major fraction of Arabidopsis seed reserves (Penfield ; Lionen and Schwender, 2009). There were no significant differences across the range of genotypes except for er105 erl1.2/seg erl2.1 seeds (15% decrease) and hence, apart from this one genotype, no correlation with germination sensitivity to salt (Supplementary Fig. S8A). The relative proportions of FAME species were also similar across genotypes (Supplementary Fig. S8B, C). Taken together, these results indicated that the delayed or arrested germination of er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds on saline media was not likely to be due to reduced embryo size and growth potential per se.

Germination involves complex communication between the embryo, seed coat, and the intermediate endosperm, which is a single-cell layer in the Arabidopsis seed. We therefore next considered a potential role of the ERECTA genes on seed germination via effects on the tissues surrounding the embryo. We took advantage of the different contributions of the maternal and paternal genomes to the genetic make-up of the three seed compartments (seed coat ♀ ♀, endosperm ♀ ♀ ♂, embryo ♀ ♂) and performed reciprocal crosses between the WT and the salt-hypersensitive er105 erl1.2 and er105 erl2.1 mutants. These generated F1 seeds with the same embryo genotype but with either WT or mutant seed coat, and with predominantly WT or mutant endosperm. The two groups of F1 seeds germinated synchronously on NaCl-free media, but showed significantly different kinetics when subjected to salinity stress (Fig. 8C). Remarkably, for each cross, F1 seed germination occurred synchronously with seeds of the maternal parent. This demonstrates that the function of the ERECTA genes in the regulation of germination sensitivity to salinity was primarily maternally controlled and mediated by the tissues surrounding the embryo, in particular the seed coat. Supporting this, when excised from their covering layers, ‘naked’ er105 erl1.2, er105 erl2.1, and er105 erl1.2 erl2.1 mature embryos grew at similar rates to WT embryos, whether cultured with or without salt (Fig. 8B). F1 seeds also clustered with their maternal parent with respect to seed size (Fig. 8D), showing that the ERECTA genes effect on seed size was also of maternal origin, and strengthening the case for overlap of the control of seed size and germination response to salinity by the ERECTA genes.

The regulation of germination mediated by the ERECTA genes involves the seed-coat mucilage

We examined which properties of the seed coat might be controlled by the ERECTA genes in response to salinity. To examine seed-coat permeability, seeds were incubated in tetrazolium red, a cationic dye classically used to detect seed-coat defects and abnormal permeability (Wharton, 1955; Molina ). Similar staining and tetrazolium salt reduction rates were observed across the genotypes except for significant increases in er105 erl1.2 and to a lesser extent in erl1.2 seeds (Fig. 9A), which was suggestive of increased seed-coat permeability or NADPH-dependent reductase activity in these two mutants. We next measured seed sodium contents after 24 h stratification with or without salt and found no differences between the genotypes (Fig. 9B). A thick cuticle lining the outer side of the endosperm has been described in mature Arabidopsis seeds, which constitutes a barrier protecting the inner living tissues of the seed (De Giorgi ; Loubéry ). Furthermore, this cuticle is maternally inherited, similar to the differential salt sensitivity of germination that we observed. We therefore examined the possibility that its permeability may have differed between the WT and salt-hypersensitive er105 erl1.2, er105 erl2.1, and triple-mutant endosperms by using Toluidine Blue permeability tests. Embryos were excised from seeds incubated either in the presence of paclobutrazol (PAC) to inhibit both endosperm and testa rupture or in the presence of ABA to allow only testa rupture and hence expose the endosperm cuticle. Following incubation with Toluidine Blue, no staining was apparent in WT embryos from seeds treated with either PAC or ABA (Supplementary Fig. S9A), which was consistent with previous reports (De Giorgi ; Loubéry ). This was also the case for embryos from the mutants, apart from er105 erl1.2 treated with ABA (i.e. exhibiting testa rupture), which frequently showed some faint staining in parts of the superficial cell layers of the hypocotyl and in the very tip of cotyledons. To test for possible interactions between salinity and genotype on endosperm cuticle permeability, the assays were repeated in the presence of 100 mM NaCl (Supplementary Fig. S9B). Similar results were obtained, with no sign of staining in PAC-treated WT or mutant seeds, and no marked staining in ABA-treated seeds. Overall, these results provide no indication for a role of differential seed-coat or endosperm permeability in the dramatic differences in germination sensitivity to NaCl between the WT and the three mutants examined.

Fig. 9.

The Arabidopsis ERECTA genes are involved in the control of seed-coat permeability and mucilage composition, and play a salinity-dependent role in the regulation of germination speed. The wild-type (WT) and the mutants er105, erl1.2, erl2.1, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 were examined (note that for simplicity, er105, erl1-2, and erl2-1 are abbreviated to er, erl1, and erl2, respectively, in the figure). (A) Seed-coat permeability to Tetrazolium Red. Data are means (±SE) of n=4 replicates of 100 seeds each. Significant differences were determined using two-way ANOVA and Scheffe’s post hoc test (*P<0.05; **P<0.01). (B) Seed sodium content of seeds at 24 h post-stratification on 0 mM or 150 mM NaCl media. Data are means (±SE) of n=3 pools of 10 mg mature seeds each. Different letters indicate significant differences as determined using two-way ANOVA and Tukey’s HSD pair-wise tests; P=0.42 for genotype effect under control conditions and P=0.39 under salt treatment. (C) Correlations between mass of water-soluble mucilage per seed and seed size. Data for mucilage are means of n=4 seed samples of 40 mg per genotype and data for weight are means of n=5 pools of a known number of seeds (20–40) corresponding to seeds from five siliques of same age. Regression lines: 0 mM NaCl, y=36.6x–0.20, r2 =0.84; 150 mM NaCl, y=36.3x+0.002, r2 =0.81. The experiment was repeated three times with similar results. Similar results were also obtained with size expressed as area (data not shown). (D) Relationship between ratios of galacturonic acid/galactose (GalUA/Gal) and rhamnose/xylose (Rhm/Xyl). Different letters next to data points indicate significant differences in GalUA/Gal as determined using one-way ANOVA and Tukey’s post-hoc tests, compared to all unlabelled data points (P<0.05). The difference in Rhm/Xyl between er erl1.2/seg erl2.1 and the WT is significant at P=0.08. (E) T50 values (h post-stratification to rupture in 50% of seeds) for testa rupture (TeR) and endosperm rupture (EnR) for intact seeds and seeds with the outer water-soluble mucilage removed (‘demucilaged’). Data are means of n=3 plates, with 30 seeds per genotype per plate. Labelled points highlight genotypes where removal of the mucilage significantly advanced germination on 150 mM NaCl media. The 1:1 line represents no effect of mucilage removal. (F) Expression of TCH3 in the WT and er erl1.2/seg erl2.1 in dry and imbibed seeds during the three germination phases Data are means (±SE) of n=4 samples of 300 seeds each per genotype and treatment.

During seed-coat differentiation on the mother plant, the specialised epidermal cells secrete mucilage polysaccharides that line their inner walls and build a central volcano-shaped columella (Beeckman ; Western ; Haughn and Western, 2012). The desiccated, highly hydrophilic mucilage rapidly swells upon hydration and ruptures the enclosing outer primary wall, wrapping the seed in a gelatinous capsule traversed by cellulosic rays radiating from the columella. Mutant seeds affected in mucilage synthesis or extrusion have been reported to be more sensitive to low water potential during germination (Penfield ; Yang ). This prompted us to next examine mucilage release during imbibition by the WT and mutant seeds in our study. We collected the loosely adhering mucilage that can easily be detached from the seed surface, as opposed to the inner fraction that is bound to the cell wall. Large variation was observed in the amounts recovered between the genotypes, but that correlated with the genetic variation in seed size (Fig. 9C). Salinity caused large increases in mucilage extrusion, but to a similar extent in all genotypes and so the relationship with seed size was unchanged. Staining with ruthenium red, which binds to pectins, consistently showed a thicker and often darker halo of mucilage under saline conditions, but with no indication of variation among genotypes within the saline and control treatments (Supplementary Fig. S10A, B).

Previous studies have suggested the importance for germination of the physico-chemical properties of the mucilage and of its attachment to the seed, rather than simply the amount present (Rautengarten ; Saez-Aguayo ). We therefore analysed mucilage composition. The expected sugars were detected, mostly rhamnose (Rhm) and galacturonic acid (GalUA) derived from rhamnogalacturonans type I (RG I), the major pectin of Arabidopsis seed mucilage (Macquet ; Arsovsky et al., 2009), and low amounts of other neutral and acidic sugars derived from RG I side-chains (Supplementary Fig. S10C). When analysed individually, these sugars showed no significant variations among genotypes; however, examination of compositional profiles by multivariate analysis suggested genetic variation in the relative abundance of backbone sugars (Rhm and GalUA) and some side-chain sugars (Xyl and Gal), which lead us to compare their ratios across the full range of genotypes (Fig. 9D). This revealed dramatically increased GalUA/Gal ratios in the mucilage of er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds compared to that of WT and other mutant seeds (P=0.027) and, apart from erl1.2 erl2.1, a trend for higher rhamnose to xylose ratios in mutant seeds, especially of the er105 erl1.2 and er105 erl1.2/seg erl2.1 genotypes (P=0.08). These results suggest that the ERECTA genes play a role in the control of mucilage composition and architecture via interactions with the mechanisms controlling the abundance of carboxyl sites (i.e. the potential sites for pectin cross-linking) and perhaps also pectin branching. Moreover, they indicate a link between this role and the function in the regulation of seed germination.

To examine this in more detail, we took an indirect, holistic approach, and compared the germination kinetics of intact seeds and ‘demucilaged’ seeds from which we removed the shell of loosely adherent mucilage extruded during imbibition. Demucilaged seeds consistently germinated more slowly than intact seeds on salt-free media (Fig. 9E), as is common. This was also the case under saline conditions for WT, erl1.2, erl2.1, and erl1.2 erl2.1 seeds, but remarkably, mucilage removal had the opposite effect in er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds, where the delay in germination relative to the WT was 66% to 74% shorter than observed for intact seeds (Fig. 9E), dropping from 39 h to 13 h for er105 erl2.1, 68 h to 23 h for er105 erl1.2, and 153 h to 39 h for er105 erl1.2/seg erl2.1. This was due to faster progression from testa rupture to endosperm rupture. These results demonstrate a critical role of the water-soluble mucilage in mediating the function of the ERECTA genes in controlling the germination response to salinity.

Although they appear as distinct layers in the imbibed seed, the mucilage and epidermal cell walls are tightly bound. The suberised seed coat and underlying endosperm constitute a mechanically strong barrier that needs to be weakened to enable radicle emergence. The micropylar endosperm that surrounds the radicle tip is thought to be the major source of mechanical resistance to radicle protrusion (Linkies ; Dekkers ). Endosperm weakening is effected by cell wall-modifying enzymes in interaction with ROS and hormonal signals from the embryo, especially GA (Finch-Savage and Leubner-Metzger, 2006; Müller ; Penfield ). We therefore hypothesised that the importance of the mucilage and seed coat in mediating delayed or arrested germination in the er105, er105 erl1.2, er105 erl2.1, and er105 el1.2/seg erl2.1 mutants on saline media could in part be related to ERECTA family-dependent differences in the mechanical properties of the endosperm and seed coat. The Arabidopsis seed is too small for direct measurement of the forces involved in testa and endosperm rupture, as is possible in other species (Linkies ), leading us instead to examine the expression of the Arabidopsis TOUCH (TCH) gene TCH3, which encodes a calmodulin-like protein and is greatly up-regulated in response to a range of mechanical signals in other tissues (Braam and Davis, 1990). Comparison of TCH3 expression in the WT and er105 erl1.2/seg erl2.1 seeds showed the presence of transcripts at similar, low levels in dry seeds (Fig. 9F). Imbibition triggered de novo transcription of TCH3 on 150-mM NaCl media in both the WT and mutant seeds, consistent with the known role of calcium in salinity signalling (Munns and Tester, 2008). Induction was significantly enhanced in er105 erl1.2/seg erl2.1 seeds and was transient, preceding testa rupture and then disappearing. These results are suggestive of enhanced mechanical constraints imposed on er105 erl1.2/seg erl2.1 embryos before endosperm rupture compared with the WT. On salt-free media, de novo transcription of TCH3 did not occur before the final phase of germination, and again it was enhanced in er105 erl1.2/seg erl2.1 seeds compared to the WT.

Discussion

The mechanisms by which seeds monitor conditions in their immediate surroundings in order to optimise the timing of the initiation and completion of germination are mostly unknown. In this study, we demonstrated that the Arabidopsis ERECTA family acts to control the timing of seed germination according to external salinity and osmotic levels (Figs 1, 4). Loss of function of ER, or of ER and its paralogs, slows down germination or even prevents it under increasing salinity and osmotic stress, although seed viability is not compromised as germination readily resumes upon the return of favourable conditions (Fig. 6). The sensing of changing salinity levels mediated by the ERECTA gene family involves interactions with the ABA–GA signalling network of germination and dormancy, and is primarily controlled by the endosperm and testa surrounding the embryo, with a critical but not exclusive role of the testa and its mucilage (Figs 8, 9). These findings reveal previously unsuspected regulators of the interactions between the seed and its environment, and a novel function of the three ER family receptor-like kinases in controlling these interactions, together with cryptic genetic variation in seed germination.

The ERECTA gene family regulates germination in saline conditions via maternally controlled effects on the tissues surrounding the embryo

The seed coat derives from the maternal ovule integuments, which expand and undergo profound developmental and biochemical transformations following fertilisation, resulting in a highly differentiated, impermeable and mechanically strong tissue (Beeckman ; Western ). Mucilage is secreted and deposited in its outer, epidermal layer concomitantly with embryo morphogenesis, following the cessation of integument expansion (reviewed by Haughn and Western, 2012; North ), but its physiological roles have remained elusive. Apart from anchoring the imbibed seed to its physical substrate, the gelatinous mucilage is generally thought to facilitate germination, especially under osmotic stress, through sequestering water and keeping the seed hydrated (Penfield ; Arsovskiet al., 2010; Yang ). However, several studies have suggested that mucilage can also inhibit germination under unsuitable conditions, perhaps through limiting water and oxygen diffusion to the embryo (Western, 2012, and references therein). Our current study has shed some light on the poorly understood genetic control of the context-dependent role of the seed mucilage in germination, and revealed that the ERECTA genes are key players. We observed a promoting role of the mucilage on the speed of germination in WT, erl1.2, erl2.1, and erl1.2 erl2.1 seeds, under both saline and non-saline conditions (Fig. 9E). However, in the salt-hypersensitive er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds this role was only expressed under non-saline conditions: under salinity, it was lost (in er105 erl1.2, er105 erl2.1) or even reversed (in the triple-mutant). These results link, for the first time, the seed mucilage and the ERECTA genes in the regulation of seed germination sensitivity and response to environmental variations at the seed surface.

By what mechanisms could the ERECTA genes control the salinity-dependent properties of the seed mucilage that regulate the germination process? The mucilage is like a pectin-rich secondary cell wall (Haughn and Western, 2012). The degrees of pectin branching and cross-linking (with calcium ions in particular) are known to greatly influence their hydrophilicity, adsorption to cellulose microfibrils, and partitioning between loose outer mucilage and adherent inner mucilage (Willats ; North ; Ralet ). It is also well established that the small monovalent Na+ ions have the general capacity to easily displace the larger divalent Ca2+ ions that cross-link the carboxyl residues of adjacent pectin molecules (Fry, 1986; Willats ; Ghanem ), thus leading to looser, more hydrophilic mucilage upon imbibition with saline compared to salt-free water, and also to greater abundance of mucilage (Fig. 9C; Ghanem ) due to increased release of pectin molecules from the cellulose matrix. The enrichment in uronic acids in the seed mucilage of er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 potentially increases the sites for Ca+2–Na+ exchange, and the trend of reduced xylose content relative to backbone rhamnose is suggestive of altered branching (Fig. 9D). We propose that this has the potential to significantly modify the swelling properties of the mucilage and of the subtending walls, and to change their osmotic potential, conformation, and rigidity upon imbibition with a saline or high osmolarity solution (Willats ; Ghanem ; Ralet ). It may also significantly modify the rearrangement of mucilage- and wall components that occurs as pectin molecules are released (Rautengarten ), and it perhaps also affects Ca2+ influx to the adjoining inner endosperm and embryo. Thus, the increase in uronic acids and decrease in xylose/rhamnose ratio would modify the overall chemical and mechanical interactions between the seed environment, seed coat, and interior compartments, and thereby the perception and early signaling events of salinity.

The

The mutant seeds with enhanced sensitivity to salt and hyperosmotic stress during germination were smaller than those of the WT (Fig. 8A). Seed size in Arabidopsis is controlled by complex interactions of zygotic and maternal factors, and by signalling between the integuments and endosperm (Garcia ; Luo ; Day ; Dilkes ; Zhou ; Wang ; Jiang ). Our reciprocal crosses showed that the variation in final seed size among the mutants and WT is of maternal origin (Fig. 8D). Final seed size is reached early in seed development, through a first phase of active cell proliferation in both the integuments and the endosperm, triggered by fertilisation, followed by a period of mostly cell expansion. Expansion ceases 5–6 d post-anthesis, concomitantly with the endosperm switching from syncitial development to cellularisation (Garcia ) and with the start of starch and mucilage synthesis. Variation in maximum cell elongation appears to be the main driver of maternal variations in seed size, such as observed here, through a so-called ‘compensatory’ growth mechanism (Garcia, 2005). The ER gene has been implicated in a compensatory mechanism between cell number and size in Arabidopsis leaves (Ferjani ), and comparison of the seed epidermis in Ler and Columbia accessions suggests ‘compensation’ may take place in the seed integuments too (Garcia ). Interestingly, the progression and completion of integument growth during ovule development has previously been reported to require a minimum level of ERECTA family signalling (Pillitteri ), which was ascribed to a role of ERECTA genes in cell proliferation activity through interactions with cell-cycle regulators. However, the final cell number in the mature ovule was unchanged, making it unlikely that the reduced size of the seed cavity and less-expanded seed coat that we observed in the er105, er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds (Fig. 8A) were pre-determined prior to fertilisation. Moreover, no ovule integument growth defects were reported by Pillitteri other than in the triple mutant er105 erl1.2 erl2.1+/–. In the present work, we found that loss of ER alone was sufficient to cause reduced seed size, and further loss of ERL1 or ERL2 had only a small additional effect. Furthermore, when it occurred in an ER background, loss of ERL1 and ERL2 instead caused an increase of seed size beyond that observed in the WT (Fig. 8A). This supports the idea that partly different mechanisms are involved in the ERECTA-mediated control of seed size and germination sensitivity to salinity. Given the role of the ERECTA family in the composition of the mucilage (Fig. 9D) and reported increases in uronic acids and cellulose in leaves of two er mutants (Sánchez-Rodriguez ), an intriguing hypothesis is that the ERECTA family may regulate cell wall formation and assembly, not only during mucilage and secondary cell wall deposition but also prior to that, during seed-coat enlargement and formation of the seed cavity. This would provide a unifying explanation for the link that we found between the ERECTA family-mediated regulation of seed germination sensitivity to salinity, salinity-dependent mucilage properties in germinating seeds, and seed size.

ERECTA family-mediated salt signalling in germinating seeds involves a complex regulatory network

The Arabidopsis seed coat is in immediate contact with the single-cell layer endosperm, which itself is in direct contact with the embryo. Although it is less well-documented than in humans, there are demonstrated cases of the ability of plant membrane receptors or mechano-sensitive channels to monitor cell wall integrity, physical interactions between the membrane and wall, wall deformation, and wall rheology (Hamann, 2012; Monshausen and Haswell, 2013; Hamilton ; Haswell and Verslues, 2015). The ERECTA family proteins belong to the class XIII of leucine-rich repeats receptor-kinases. Interestingly, among its other four members, this class includes FEI1 and FEI2 (Shiu ) that have been shown to interact with the SOS5 arabinogalactan protein to influence cellulose production and pectin assembly in the Arabidopsis seed mucilage, as well as mucilage adherence (Harpaz-Saad ; Griffiths ). In addition, based on analysis of disease resistance in two er mutants, the ER protein has been suspected of interacting with wall-associated kinases (WAKs) during defence against some pathogens via effects on cell wall composition (Sánchez-Rodríguez ). WAKs are known to be tightly bound to pectins, especially galacturonic acids, in a Ca2+-dependent manner (Wagner and Kohorn, 2001; Decreux and Messiaen, 2005), and several WAK/WAK-Like proteins have been implicated in responses to mineral ions, including Na+ (Sivaguru ; Hou ; de Lorenzo ), and to osmotic stress (SOS6/AtCSLD5; Zhu ), through unknown mechanisms. Modifications of mucilage and bound cell walls mediated by the ERECTA family may thus be perceived and signalled to the seed interior by the ERECTA proteins themselves either directly or through modified interactions with cell wall-associated proteins, osmo-sensors, or mechano-sensors (Dekkers ; Nonogaki, 2014). The induction of TCH3 in the er105 erl1.2/seg erl2.1 seeds supports this hypothesis (Fig. 9F).

It would be intriguing to unravel the downstream cascade. The salt-hypersensitive er105 erl1.2, er105 erl2.1, and er105 erl1.2/seg erl2.1 seeds showed enhanced sensitivity to exogenous ABA, and enhanced up-regulation of ABI3, ABI5, and RGL2 under saline conditions compared to the WT (Fig. 7). These genes are emerging as important mediators of salinity and osmotic stress and as controllers of ABA–GA homeostasis in imbibed seeds. ABA synthesised in the endosperm and released to the embryo activates the abundance and activity of the ABI3 and ABI5 transcription factors, and triggers an auto-feedback loop that maintains RGL2 mRNA at high levels and represses cell-wall modifying enzymes (Giraudat ; Finkelstein and Lynch, 2000; Lopez-Molina et al., 2001, 2002; Lee et al., 2002, 2010; Piskurewicz et al., 2008, 2009; Kang ). Our data indicate that the regulation of germination sensitivity to changing salinity levels mediated by the ERECTA genes interferes with that signalling loop.

A well-documented adaptive mechanism that seeds have evolved to withstand unfavourable conditions such as high temperatures, cold, osmotic or salinity stress, and to maintain embryo viability is secondary dormancy (Bewley, 1997), a reversible, transient quiescent state induced and released in adaptation to fluctuating environmental conditions (Koornneef ; Giraudat ; Léon-Kloosterziel ; Finch-Savage and Leubner-Metzger, 2006; Lefebvre ; Weitbrecht ; Ibarra ). ABI3, ABI5, and RGL2 are prominent players in the regulation of secondary dormancy and increased sensitivity to ABA, and up-regulation of ABI3, ABI5 and RGL2 has been reported during early growth arrest in newly germinated Arabidopsis seedlings under water stress and salinity (Lopez-Molina et al., 2001, 2002). Here, we found that loss of various combinations of ERECTA genes sensitised seed germination to salinity and frequently arrested it (Fig. 6). This arrest was reversible, as germination readily resumed upon stress release and progressed to completion as quickly as in seeds that were never exposed to stress. Arrested seeds showed up-regulation of DOG1 (Fig. 7), a major controller of the coat- and endosperm-mediated dormancy that occurs in Arabidopsis seeds. DOG1 interacts with GA and ABA signalling upstream of ABI5, and appears to be an agent of environmental adaptation of germination among Arabidopsis accessions (Dekkers ; Graeber, 2014; Née ; Nishimura ). Taken as a whole, these observations suggest that the ERECTA family interacts with the molecular controls of dormancy to appropriately cue and pace germination. While promotion of fast germination under stress may be seen as desirable, it can also expose the newly germinated seedling to the risk of death should the adverse conditions persist or worsen as the embryo becomes directly exposed to the external environment with all its reserves already depleted. In such circumstances, delay or arrest of germination may be a useful protective strategy to maximise the chances of survival by temporarily safeguarding the embryo against such a fate. In that context, the environment-dependent function of the ERECTA family with regards to germination speed and temporary arrest would perform a vital adaptive function. Interestingly, the loss of ER and ERL1 and/or ERL2 caused the arrest of germination in absolutely all seeds within a cohort only under extremely severe stress (~200 mM NaCl) (Supplementary Fig. S2). Under milder stress, some seeds did germinate at the same time as the WT, some showed increasing delays, and others were arrested until the stress was released, thus demonstrating a mixed response that might balance the risks of death against loss of fitness or ability to complete the life cycle in time.

There is increasing agreement that the endosperm plays a prominent role in the regulation of dormancy and germination in Arabidopsis seeds, in particular through its role in modulating the ABA–GA balance that is central to the control of these processes. The endosperm is a predominantly maternal tissue that is in direct contact with the seed coat, and it is of utmost importance in the determination of seed size (Garcia ) and in the de novo synthesis of ABA and its transport to the embryo (Lee ; Kang ). It is also an important site of interactions between hormones, ROS, and cell-wall remodelling enzymes that weaken the mechanical resistance to radicle protrusion (Lee ; Dekkers ; Nonogaki, 2014). According to our permeability tests, there was no apparent association between the differential germination sensitivity to salinity among the genotypes and the variations in endosperm permeability, apart from possibly the er105 erl1.2 mutant (Supplementary Fig. S9). However, this requires further detailed examination, as more broadly does the role of the ERECTA family on the biochemical and mechanical properties of the endosperm, and on the upstream perception events of changes in the seed environment.

Conclusions

Plants need to be endowed with a ‘surveillance’ system for the perception of external environmental cues, their transduction to internal compartments, and their integration with developmental pathways. This study has demonstrated a key role of the ERECTA family in the integrative network in seeds that controls the most critical decision in the life cycle, namely when to initiate a new plant. Given the evolutionary conservation of the ERECTA receptor-kinases across a broad range of plant species, and an increasing interest in mucilage as a model for cell wall studies and as an important adaptive feature, our findings open new avenues for determining the mechanisms that seeds have evolved to control germination and to tune it to local conditions in order to maximise the chances of survival.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Germination dynamics of the wild-type and ERECTA mutants in response to NaCl.

Fig. S2. Loss of ER alone or in combination with ERL1 and ERL2 sensitises seed germination to salinity in a dose-dependent manner.

Fig. S3. Phenotypes of selected erecta mutants.

Fig. S4. Increased sensitivity of seed germination to salinity in the er2, er2 erl1-5, er2 erl2-2, and er2 erl1-5 erl2-2 mutants.

Fig. S5. Time-course of promoter activity of ERECTA genes in mature dry seeds and germinating seeds.

Fig. S6. Expression levels of ABA and GA biosynthesis genes in response to salinity in WT and er105 erl1.2/seg erl2.1 seeds.

Fig. S7. Mature er105 erl1.2 erl2.1 embryos exhibit similar radicle size and patterning to WT embryos.

Fig. S8. Relative abundance of total fatty acid methyl-esters and relative proportions of individual species in embryos of the WT and ERECTA family mutants at full seed maturity.

Fig. S9. Toluidine Blue tests for seed-coat and endosperm cuticle permeability in response to salinity in the WT and salt-hypersensitive er105 erl1.2, er105 erl2.1, and er105 erl1.2 erl2.1 seeds.

Fig. S10. Characteristics of the seed mucilage in the WT and erecta mutants.

Table S1. List of primers used for genotyping and RT-qPCR.

Click here for additional data file.