Seed hemicelluloses tailor mucilage properties and salt tolerance

While Arabidopsis seed coat epidermal cells have become an excellent genetic system to study the biosynthesis and structural roles of various cell wall polymers, the physiological function of the secreted mucilaginous polysaccharides remains ambiguous. Seed mucilage is shaped by two distinct classes of highly substituted hemicelluloses along with cellulose and structural proteins, but their interplay has not been explored. We deciphered the functions of four distinct classes of cell wall polymers by generating a series of double mutants with defects in heteromannan, xylan, cellulose, or the arabinogalactan protein SALT-OVERLY SENSITIVE 5 (SOS5), and evaluating their impact on mucilage architecture and on seed germination during salt stress. We discovered that muci10 seeds, lacking heteromannan branches, had elevated tolerance to salt stress, while heteromannan elongation mutants exhibited reduced germination in CaCl2. In contrast, xylan made by MUCILAGE-RELATED21 (MUCI21) was found to be required for the adherence of mucilage pectin to microfibrils made by CELLULOSE SYNTHASE5 (CESA5) as well as to a SOS5-mediated network. Our results indicate that the substitution of xylan and glucomannan in seeds can fine-tune mucilage adherence and salt tolerance, respectively. The study of germinating seeds can thus provide insights into the synthesis, modification and function of complex glycans.

water. Seed germination is essential for plant establishment and is highly sensitive to salt stress.

90
In this study, we therefore explored how genes affecting different wall polymers modulate 91 mucilage properties, seed germination and early growth under salt stress (Fig. 1a).

93
Plant materials 94 Mutations were genotyped using primers listed in Table S1     Total mucilage was extracted with a ball mill, hydrolyzed, and quantified via high performance 117 anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD) as 118 previously described . The quantification of mucilage detachment via 119 HPAEC-PAD has also been described in detail (Voiniciuc, 2016). HPAEC-PAD of mucilage   were surrounded by smaller mucilage capsules (Fig. 1b,c). MUCI21, CESA5 or SOS5 were 165 epistatic to MUCI10 in terms of adherent mucilage size. While all mutants produced wild-type 166 percentages of rhamnose and galacturonic acid in total mucilage extracts (Table S2)   other genotypes (Fig. S1a,b), proportional to the RR-stained adherent capsule size (Fig. 1b,c). Each mutation also had distinct effects on S4B staining, which primarily detects cellulose 181 (Anderson et al., 2010), and all the double mutants seeds lacked the ray-like structures that were 182 observed around the wild type (Fig. 2d). Among the single mutants, only muci21 and cesa5 183 displayed clear ray-like structures (Fig. 2d), while sos5 only had more diffuse cellulose as 184 previously shown ( Fig. 2d; Griffiths et al., 2014). The impact of the different mutant 185 combinations on cellulose architecture were also supported by crystalline polymer birefringence 186 (Fig. S1c). In short, CESA5, SOS5, or MUCI21 were epistatic to MUCI10 for pectin adherence 187 (Fig. 2b,c), via partially overlapping mechanisms, and the loss of any two players severely can be dramatically reshaped when more than one structural component is modified.

191
The elongation and substitution of HM modulate salt tolerance 192 The newly generated mutant collection affecting multiple classes of wall polymers enabled us to 193 investigate the physiological consequences of altering mucilage structure. We established a novel 194 seed germination and salt stress assay using aqueous solutions in 24-well plates. Nearly all wild-195 type and mutant seeds imbibed in water germinated within 24 h post-stratification (Fig. 3a). 196 However, when placed in 150 mM CaCl 2 , few wild-type seeds germinated even after 48 h of 197 exposure to constant light. We initially hypothesized that mucilage-defective mutants might be 198 more susceptible to salt stress, but unexpectedly found that muci10 and muci10 muci21 seeds had 199 over 5-fold higher germination rate at this stage (Fig. 3a). The other mutant combinations 200 germinated like the wild type at all time points. Only muci10 and muci10 muci21 had 201 significantly longer radicles at 72 h in 150 mM CaCl 2 ( Fig. 3b and Fig. 3d), even though most 202 mutants had around a two-fold higher flotation rate compared to the wild type (Fig. 3c). The 203 enhanced germination rate and radicle growth of muci10 in 150 mM CaCl 2 was replicated in 204 multiple assays, including up to 100 seeds per well and independent growth batches ( Fig. 3e-g).

205
To evaluate the basis of the observed salt tolerance, we assayed the effects of the muci10 206 mutation in additional stress conditions. The use of 150 mM NaCl also reduced the rate of seed 207 germination, but radicles that protruded from NaCl-treated seeds failed to further elongate 208 compared to the CaCl 2 treatment (Fig. S2). Nevertheless, muci10 and muci10 muci21 germinated 209 faster than wild type in both salt treatments (Fig. 3a and Fig. S3a). All seeds sunk in water within 210 the stratification period (Fig. S3b), but a significant proportion of certain seeds (only muci21 in 211 8 NaCl, and most mutants in CaCl 2 ) continued to float in the salt solutions (Fig. S3c). When 212 subjected to ionic (150 mM CaCl 2 or MgCl 2 ) or purely osmotic stress (PEG 4000 or sorbitol) of 213 equivalent pressure, the germination rate of muci10 seeds was significantly higher than wild type 214 only in calcium salt stress (Fig. 4a). Once protruded from the seed coat, muci10 radicles 215 elongated significantly faster than wild type in both CaCl 2 and sorbitol treatments ( Fig. 4b;   216 despite 3-fold difference in sample sizes), while the magnesium and PEG solutions showed demonstrated that muci10 enhances growth in calcium stress during radicle emergence as well as 221 subsequent elongation (Fig. S3d,e). 222 We then investigated how mucilage removal impacts salt tolerance, by extracting seed coat 223 polysaccharides using a ball mill prior to stratification. With or without mucilage, CaCl 2 -treated 224 muci10 seeds germinated faster than wild-type (Fig. 4c). Mucilage β -glucans continue to 225 encapsulate wild-type seeds at 72 h post-stratification (Fig. 4d), but were absent from de-226 mucilaged wild-type seeds and from HM-deficient muci10 seeds (regardless of treatment).

227
Despite not altering the germination rates of after-ripened seeds, the mucilage extraction  Table S4), suggesting that additional CSLAs elongate HM in the same tissues.

234
Using microarray data, we found that the transcription of CSLA2, CSLA3, CSLA9 along with 235 CSLA7 and CSLA11 (to a lesser extent) increased during germination relative to dry seeds (Fig.   236 S4d). Compared to the wild type, we found that the csla2-1 csla3-2 csla9-1 triple mutant 237 (abbreviated as csla239), reported to have glucomannan-deficient stems (Goubet et al., 2009), 238 had significantly lower germination (Fig. 4f) and smaller radicles (Fig. 4g)  In summary, we found that the biosynthesis of two substituted hemicelluloses in the seed 245 coat epidermis can be uncoupled and that HM and xylan have largely independent functions. HM 246 substituted by MUCI10 is responsible for controlling pectin density, supporting cellulose 247 synthesis and modulating seed tolerance to salt stress. In contrast, MUCI21, CESA5 and SOS5 248 are all epistatic to MUCI10 for pectin adherence to the seed surface, via partially overlapping 249 means (Fig. 1a). Since muci21, cesa5 and sos5 had additive effects (Fig. 1, Fig. 2, and Fig. S1, to the seed surface than previously thought (see remnants of rays in Fig. 2d and Fig. S1c).

257
In addition to gaining insight into the genetic regulation of mucilage properties, we 258 discovered that HM structure modulates seed germination in CaCl 2 solutions, and to a lesser 259 extent in other ionic/osmotic conditions. Ca 2+ ions can cross-link unesterified mucilage pectin 260 and all the generated double mutants had elevated flotation compared to the wild type. However, 261 only the muci10 mutation promoted germination in CaCl 2 , while the csla239 triple mutant 262 reduced it. Consistent with these effects, MUCI10 and other HM biosynthetic genes were up-263 regulated during seed germination ( Fig. 1d and Fig. S4d), while MUCI21 was not. Since CESA5 264 was also expressed in germinating seeds (Fig. 1d) Table S1 Insertional mutants and genotyping primers used in this study. 500 Table S2 Monosaccharide composition of total mucilage extracted from seeds. 501 Table S3 Detachment of mucilage components after gentle shaking. 502 Table S4 Cell wall composition of csla2 mutant germinated seeds. 503 Table S5 Cell wall composition of muci10 and csla239 germinated seeds. 504