grain size and number1 negatively regulates the …56 dephosphorylation of osmpk6 to coordinate the...

1

RESEARCH ARTICLE

GRAIN SIZE AND NUMBER1 Negatively Regulates the

OsMKKK10-OsMKK4-OsMPK6 Cascade to Coordinate the

Trade-off between Grain Number per Panicle and Grain Size in

Rice

Tao Guoa,b, Ke Chena,b, Nai-Qian Donga,b, Chuan-Lin Shia,b, Wang-Wei Yea, Ji-Ping Gaoa, Jun-Xiang Shana,*, Hong-Xuan Lina,b,c,*

aNational Key Laboratory of Plant Molecular Genetics, CAS Centre for Excellence

in Molecular Plant Sciences and Collaborative Innovation Center of Genetics & Development, Shanghai Institute of Plant Physiology & Ecology, Shanghai Institute for Biological Sciences, Chinese Academic of Sciences, Shanghai 200032, China bUniversity of the Chinese Academy of Sciences, Beijing 100049, China cSchool of Life Science and Technology, ShanghaiTech University, Shanghai, 201210, China *Corresponding Authors: [email protected] and [email protected]

Short title: GSN1 coordinates grain number and grain size

One-sentence summary: The GRAIN SIZE AND NUMBER1–mitogen-activated protein kinase module coordinates the trade-off between grain number and size in rice by integrating localized cell differentiation and proliferation.

The authors responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) are: Hong-Xuan Lin ([email protected]) and Jun-Xiang Shan ([email protected]).

ABSTRACT Grain number and size are interactive agronomic traits that determine grain yield. However, the molecular mechanisms responsible for coordinating the trade-off between these traits remain elusive. Here, we characterized the rice (Oryza sativa) grain size and number 1 (gsn1) mutant, which has larger grains but sparser panicles than the wild type due to disordered localized cell differentiation and proliferation. GSN1 encodes the mitogen-activated protein kinase phosphatase OsMKP1, a

Plant Cell Advance Publication. Published on March 27, 2018, doi:10.1105/tpc.17.00959

©2018 American Society of Plant Biologists. All Rights Reserved

2

dual-specificity phosphatase of unknown function. Reduced expression of GSN1 resulted in larger and fewer grains, whereas increased expression resulted in more grains but reduced grain size. GSN1 directly interacts with and inactivates the mitogen-activated protein kinase OsMPK6 via dephosphorylation. Consistent with this finding, the suppression of mitogen-activated protein kinase genes OsMPK6, OsMKK4, and OsMKKK10 separately resulted in denser panicles and smaller grains, which rescued the mutant gsn1 phenotypes. Therefore, OsMKKK10-OsMKK4-OsMPK6 participates in panicle morphogenesis and acts on a common pathway in rice. We confirmed that GSN1 is a negative regulator of the OsMKKK10-OsMKK4-OsMPK6 cascade that determines panicle architecture. The GSN1-MAPK module coordinates the trade-off between grain number and grain size by integrating localized cell differentiation and proliferation. These findings provide important insights into the developmental plasticity of the panicle and a potential means to improve crop yields.

INTRODUCTION 1

Rice (Oryza sativa), the number one staple cereal crop worldwide, feeds more than 2

half of the global population every day. Yield in rice depends on tiller number, grain 3

weight, and the number of grains per panicle. Grain number per panicle is 4

determined by the number of primary and secondary branches, and grain weight is 5

determined by grain size and the degree of filling (Xing and Zhang, 2010). The 6

genetic and molecular bases of these agronomic traits have been extensively studied 7

via mapping of quantitative trait loci combined with mutant identification using 8

multiple genetic resources. Considerable progress has been made in understanding 9

the molecular mechanisms of rice panicle morphogenesis and reproductive organ 10

development, which are controlled by diverse proteins and unknown signaling 11

pathways (Zhang and Yuan, 2014; Zuo and Li, 2014). In particular, several genes 12

associated with grain number (Ashikari et al., 2005; Kurakawa et al., 2007; Huang et 13

al., 2009; Oikawa and Kyozuka, 2009; Wu et al., 2016; Huo et al., 2017) and grain 14

size (Song et al., 2007; Mao et al., 2010; Li et al., 2011; Qi et al., 2012; Hu et al., 15

2015; Si et al., 2016) have been identified and characterized in rice; these genes 16

strongly influence cell differentiation and either cell proliferation or cell expansion, 17

depending on phytohormones or other unidentified signaling molecules. Rice has a 18

determinate inflorescence, in which the meristems differentiate into primary branch 19

3

meristems attached to the central rachis, which then form several secondary branch 20

meristems (Zhang and Yuan, 2014). The spikelet meristems form directly on primary 21

and secondary branches, which determines the final number of grains per panicle. 22

The differentiated spikelets develop and produce terminal floral organs with palea 23

and lemma, which determine grain size. These spatiotemporal programmed cellular 24

processes contribute interactively to rice panicle architecture. In many plant species, 25

there is a negative evolutionary correlation between seed number and seed size 26

(Sadras, 2007). Nevertheless, the explicit molecular mechanisms coordinating the 27

trade-off between grain number per panicle and grain size in rice remain largely 28

unknown. 29

30

Mitogen-activated protein kinase (MAPK) cascades are highly conserved, ubiquitous 31

signaling pathways in eukaryotes (Widmann et al., 1999) that relay and amplify 32

signals via a set of three types of activated protein kinases: MAPK kinase kinase 33

(MKKK, or MEKK), MAPK kinase (MKK, or MEK) and MAPK (Rodriguez et al., 34

2010). MKKK receives signals from the topmost receptor or sensor to activate MKK, 35

and the activated MKK then activates MAPK. This cascade depends on sequential 36

phosphorylation and leads to the altered activities of substrates through 37

phosphorylation, which regulates cell proliferation and differentiation and the 38

coordination of responses to environmental inputs (Xu and Zhang, 2015). Plant 39

MAPK signaling modules play essential roles in multiple processes, including 40

immunity, stress responses, and developmental programs (Wang et al., 2007; 41

Lampard et al., 2008; Kong et al., 2012; Meng et al., 2012; Meng and Zhang, 2013; 42

Xu and Zhang, 2015; Wang et al., 2017). Components of the rice MAPK cascades, 43

OsMKK4 and OsMAPK6, have recently been characterized, both of which influence 44

grain size, depending on cell proliferation (Duan et al., 2014; Liu et al., 2015). 45

However, the negative regulators of MAPK cascades in crops have not yet been 46

identified. 47

48

To investigate the genetic and molecular mechanisms underlying the trade-off 49

4

between grain number per panicle and grain size in rice, we screened an ethyl 50

methanesulfonate (EMS) mutant library for presumed panicle phenotypes. Here, we 51

report the isolation and characterization of a rice mutant, grain size and number 1 52

(gsn1), that displays increased grain size and reduced grain number per panicle 53

compared to the wild type. We demonstrate that GSN1 acts as a negative regulator of 54

the OsMKKK10-OsMKK4-OsMPK6 cascade by inducing specific 55

dephosphorylation of OsMPK6 to coordinate the trade-off between grain number per 56

panicle and grain size. Thus, our findings establish a molecular understanding of 57

trade-offs in rice panicle development and provide a potential opportunity to 58

improve crop yields. 59

60

RESULTS 61

gsn1 has larger grain size and reduced grain number 62

To investigate the genetic association between grain size and grain number, we 63

treated the mutagenic elite indica rice variety Fengaizhan-1 (FAZ1) with EMS and 64

isolated the gsn1 mutant, which displays distinctly larger grain size but reduced grain 65

number per panicle and larger floral organs than the wild type without additional 66

phenotypes (Figures 1A to 1H). The average grain length, width and 1000-grain 67

weight in the gsn1 mutant were markedly increased, but grain number per panicle 68

and setting percentage were dramatically reduced compared with wild-type FAZ1 69

(Figures 1I to 1M and Supplemental Figures 1A, 1B and 1E), which resulted in 70

reduced grain yield per plant (Figure 1N). The reduced numbers of both primary and 71

secondary branches, rather than panicle size, directly determined the total spikelet 72

number in the gsn1 mutant (Figures 1O to 1Q). Taken together, these results imply 73

that GSN1 specifically influences panicle and spikelet development in rice. 74

75

Map-based cloning of GSN1 76

We isolated the GSN1 gene via map-based cloning using an F2 population derived 77

from a cross between the gsn1 mutant line and the japonica rice variety, 78

5

Zhonghua-11 (ZH11). The GSN1 locus was initially mapped to the short arm of 79

chromosome 5 between the marker loci G05220 and G05591 using 682 homozygous 80

F2 plants and was then fine-mapped to a 17.5-kilobase (kb) region between marker 81

loci G05333 and G05340 that contains two open reading frames (ORFs), 82

LOC_Os05g02490 and LOC_Os05g02500 (Figure 2A). Comparison of the genomic 83

DNA sequences of the FAZ1 and gsn1 alleles revealed that the allele from gsn1 has a 84

single C to T nucleotide mutation at position 437 in the candidate gene 85

LOC_Os05g02500, which is predicted to encode a mitogen-activated protein kinase 86

phosphatase OsMKP1, the dual-specificity phosphatase with a PTPc/DSPc (Protein 87

Tyrosine Phosphatase catalytic/Dual Specific Phosphatase catalytic) domain and a 88

GEL (Gelsolin) domain (Figure 2A). Protein sequence alignment showed that this 89

mutation directly leads to a serine to phenylalanine change at amino acid S146, 90

which is a conserved residue, and that GSN1 shares high amino acid sequence 91

identity with its homologs in monocots and dicots (Figure 2B). These results 92

indicate that LOC_Os05g02500 (OsMKP1) represents the only candidate gene for 93

GSN1. 94

95

To further confirm the identity of the candidate gene, we performed a genetic 96

complementation test in which a DNA fragment from the wild type FAZ1 containing 97

the putative promoter region, the entire ORF, and the 3’untranslated region (UTR) of 98

GSN1 was introduced into the gsn1 mutant via Agrobacterium tumefaciens-mediated 99

transformation. All positive transgenic lines harboring the full-length GSN1 100

transgene showed completely wild-type phenotypes with respect to grain number per 101

panicle and grain size (Figures 2C, 2D and Supplemental Figures 2A and 2D to 102

2F). Knockdown of GSN1 using an artificial microRNA in FAZ1, which reduced the 103

levels of endogenous gene expression, produced plants with phenotypes similar to 104

those of the gsn1 mutant (Figures 2E to 2J and Supplemental Figures 2B, 2G and 105

2J), but this is in contrast to the phenotypes of GSN1-overexpression lines in FAZ1, 106

which have reduced grain size and increased grain number per panicle (Figures 2G, 107

2H, 2K, 2L and Supplemental Figures 2C, 2H and 2K). In addition, we developed 108

6

a nearly isogenic line, NIL(gsn1), with a very small chromosomal region carrying 109

the gsn1 allele in the japonica variety 95-22 genetic background from the BC6F2 110

generation, which displayed larger grain size and reduced grain number per panicle 111

like the gsn1 mutant (Supplemental Figure 3). These results demonstrate that GSN1 112

is responsible for the alterations in grain size and grain number per panicle found in 113

the gsn1 mutant. 114

115

GSN1 is a negative regulator of grain size but a positive regulator of grain 116

number 117

To further confirm and estimate the genetic effect of GSN1, we performed a wide 118

array of transgenic experiments using the japonica variety ZH11. The knockdown of 119

GSN1 using an artificial microRNA in ZH11 resulted in significantly increased grain 120

size and reduced grain number per panicle (Supplemental Figures 4A and 4C to 121

4F), which is consistent with the phenotypes of transgenic lines produced by 122

CRISPR/cas9 genome editing, with distinct grain size and grain number per panicle, 123

as well as very low fertility (Supplemental Figures 4B, 4G to 4J and 1C, 1D, 1F). 124

However, overexpression lines produced using the GSN1 alleles from FAZ1 and 125

ZH11 exhibited increased grain number per panicle and reduced grain size, which 126

are completely the opposite panicle phenotypes compared with the gsn1 allele 127

GSN1S146F that resembles the GSN1RNAi lines expressing an artificial microRNA 128

(Supplemental Figures 5A to 5J and 6A to 6J). Thus, we hypothesize that GSN1 is 129

both a negative regulator of grain size and a positive regulator of grain number per 130

panicle and that the larger grains with high levels of exogenous gsn1 allele 131

expression resulted from dominant negative effects. Interestingly, only the repression 132

of GSN1 led to greatly reduced setting percentage similar to that of the gsn1 mutant 133

compared with the wild type, which implies that GSN1 is required for both fertility 134

and final grain number in rice (Supplemental Figures 1G to 1L). These results 135

suggest that GSN1 plays an interactive role in the trade-off between grain number 136

per panicle and grain size. 137

138

7

GSN1 influences cell proliferation instead of cell elongation to control organ 139

size 140

Given that the gsn1 mutant has dramatically enlarged grains and floral organs, we 141

examined the cell number and cell size of this line. We compared cross-sections of 142

the central parts of the spikelet hulls in FAZ1 and gsn1 before heading to investigate 143

the cellular basis of the increased organ size (Figures 3A to 3D). Our observations 144

revealed that the outer parenchyma cells in gsn1 significantly increased in number 145

compared to wild type, but the average cell length in gsn1 was comparable to that in 146

the wild type (Figure 3E). Since the mutant has large floral organs, we also 147

examined the areas and cell numbers of the vascular bundles of the anther 148

connectivum (which connects four different pollen sacs in FAZ1 and gsn1), finding 149

that there were distinguishable differences between lines (Figure 3F). Furthermore, 150

we examined the epidermal cells of mature grains and found that the average lengths 151

of the outer epidermal cells of the spikelet hulls were not significantly different 152

between gsn1 and FAZ1 (Figures 3G and 3I). Similarly, the average length of the 153

inner epidermal cells of gsn1 spikelet hulls was comparable to that in FAZ1 (Figures154

3H and 3J). Together, these results indicate that increased cell number, rather than 155

cell size, contributes to the enlarged organs in the gsn1 mutant. 156

157

Because of the similar life cycles and heading days of FAZ1 and gsn1, we 158

hypothesized that an increased rate of cell division was responsible for the larger 159

spikelet hulls of the mutant. Therefore, we analyzed the cell division rate in the 160

spikelets of younger panicles by flow cytometry, finding that the percentage of G2/M 161

phase cells with higher 4c DNA content was distinctly elevated in gsn1 but that the 162

percentage of G1 phase cells with 2c DNA content decreased after new cell cycles 163

were initiated (Figures 3K, 3L). Hence, we conclude that cell division is more rapid 164

in gsn1 during spikelet development compared to the wild type. Consistent with the 165

role of GSN1 in the regulation of cell division, the expression levels of cell 166

cycle-related genes and genes encoding several grain size regulators that influence 167

cell proliferation were significantly up-regulated in the spikelets of gsn1 compared to 168

8

FAZ1 (Supplemental Figure 7). Furthermore, the epidermal cell length of the 169

spikelet hulls in different transgenic lines, including lines with repression or 170

overexpression of GSN1, showed no obvious differences, but the cell number at the 171

longitudinal axis increased dramatically in the repression lines and decreased in the 172

overexpression lines (Supplemental Figures 8 and 9). In summary, these data show 173

that GSN1 influences cell proliferation instead of cell elongation to control floral 174

organs size. 175

176

GSN1 encodes a dual-specificity phosphatase that is localized to the cytoplasm 177

According to the Rice Genome Annotation Project and Phytozome databases, the 178

GSN1 gene (LOC_Os05g02500) is predicted to encode a dual-specificity 179

phosphatase that contains PTPc/DSPc and GEL domains. Phosphatases are pivotal 180

negative regulator of MAPK cascades, and studies in many model organisms have 181

revealed that the dephosphorylation of MAPKs can be implemented via PTPs 182

(protein tyrosine phosphatases), PSTPs (protein serine-threonine phosphatases) and 183

DSPs (dual-specificity phosphatases) in vivo (Bartels et al., 2010). The MAPK 184

phosphatases belong to a subgroup of the DSP family whose members specifically 185

dephosphorylate both phosphotyrosine and phosphoserine/threonine residues within 186

the activation motif of MAPKs (Luan, 2003). To determine whether the mutation of 187

GSN1 in gsn1 abolishes its phosphatase activity, we expressed recombinant 188

MBP-GSN1 and the mutated version MBP-GSN1S146F in E. coli and purified the 189

fusion proteins for in vitro assays. Initially, the PNPP (p-nitrophenyl phosphate) was 190

used as an artificial substrate to assay the phosphatase activity, but there was no 191

significant activity in the recombinant GSN1 proteins, as determined based on 192

absorbance at 405 nm (Figure 4A). Previous studies have indicated that DSPs 193

usually prefer bulky polycyclic aryl phosphates and display weak enzyme activity 194

toward simple aryl phosphates, such as PNPP, which likely contain shallower active 195

site pockets (Yoo et al., 2004; Lee et al., 2008). Therefore, we used the bulky 196

polycyclic aryl phosphate OMFP (3-o-methylfluorescein phosphate) as the substrate 197

to assay the phosphatase activity of GSN1. We calculated the amount of 198

9

3-o-methylfluorescein (OMF) released by dephosphorylation based on the 199

absorbance at 477 nm, which was used to characterize the enzymatic properties of 200

GSN1 and to determine its kinetic parameters (Figures 4B and 4C). The 201

phosphatase activities of both wild-type GSN1 and the mutated version GSN1S146F 202

determined using OMFP were markedly different, in that the mutated GSN1S146F 203

version is inactive in dephosphorylating OMFP (Figure 4B). These results imply 204

that the conserved serine at position 146 is vital for the phosphatase activity of 205

GSN1. 206

207

We also investigated the subcellular localization of GSN1 in Nicotiana benthamiana 208

leaf cells and rice protoplasts. Fluorescence from GSN1-GFP (green fluorescent 209

protein) fusion protein was observed at the cell periphery and in the cytoplasm of 210

Nicotiana benthamiana leaf epidermal cells compared with the ubiquitous 211

distribution of signals in the GFP positive control (Figure 4D). In agreement with 212

this finding, fluorescence from GSN1-YFP (yellow fluorescent protein) fusion 213

protein was mainly localized to the cytoplasm in rice protoplasts (Figure 4E). We 214

also examined the expression pattern of GSN1 using qRT-PCR and in transgenic 215

plants expressing GUS fusion protein driven by the GSN1 promoter. We found that 216

GSN1 was widely expressed in various vegetative and reproductive phase organs at a 217

relatively high level in young panicles and spikelet hulls, which demonstrates that 218

GSN1 plays a role in panicle and spikelet development (Figures 4F, 4G and 219

Supplemental Figures 10A to 10H). 220

221

GSN1 interacts with and inactivates OsMPK6 via specific dephosphorylation 222

Using a yeast two-hybrid assay, we found that GSN1 interacts strongly with 223

OsMPK6 (LOC_Os06g06090), as does OsMPK1 (Reyna and Yang, 2006) (Figure 224

5A). To investigate the direct physical interaction between GSN1 and OsMPK6 in 225

planta, we performed a BiFC (biomolecular fluorescence complementation) assay to 226

stabilize the protein-protein interaction after recognition of the N-terminal and 227

C-terminal parts of a split YFP. We produced DNA constructs in which GSN1 and 228

10

OsMPK6 were fused separately to both the N-terminal (nYFP) and C-terminal 229

fragments of YFP (cYFP). The nYFP-GSN1 fusion protein interacted with 230

cYFP-OsMPK6, and cYFP-GSN1 interacted with nYFP-OsMPK6, as expected, but 231

no interaction was detected between nYFP-GSN1 and the negative control, 232

cYFP-OsHAL3 (Su et al., 2016) (Figure 5B), suggesting that GSN1 associates with 233

OsMPK6 in vivo. 234

235

We further confirmed the interaction between GSN1 and OsMPK6 using Co-IP 236

(co-immunoprecipitation) assays in which GSN1-Flag and OsMPK6-Myc fusion 237

protein constructs driven by the 35S promoter were co-expressed in Nicotiana 238

benthamiana leaves. The proteins were immunoprecipitated with Flag beads and 239

detected by immunoblot analysis using anti-Flag and anti-Myc antibodies. 240

OsMPK6-Myc co-immunoprecipitated with GSN1-Flag instead of the negative 241

control. These results further confirm that GSN1 interacts with OsMPK6 in vivo 242

(Figure 5C). We then performed phosphorylation and dephosphorylation assays to 243

determine whether GSN1 can dephosphorylate and inactivate OsMPK6 in vitro. 244

GST-OsMPK6 protein expressed in E. coli and purified as substrate was primarily 245

phosphorylated and activated by a constitutively active version of OsMKK4 246

(LOC_Os02g54600) known as MBP-OsMKK4CA that carries the T238D and S244D 247

mutations (Yang et al., 2001). We then used the activated GST-OsMPK6 in a 248

dephosphorylation reaction with MBP-GSN1 and MBP-GSN1S146F. The use of the 249

phospho-p44/42 antibody specifically showed that GST-OsMPK6 can be activated in 250

the T-E-Y motif of the activation loop by MBP-OsMKK4CA, following inactivation 251

by MBP-GSN1 rather than MBP-GSN1S146F, which both exhibited a dosage effect 252

(Figure 5D). We further characterized the T-E-Y motif of the activation loop in both 253

the activated and inactivated forms of GST-OsMPK6 by mass spectrometry 254

(Supplemental Figures 11A and 11B). Consistent with the results of the in vitro 255

phosphorylation and dephosphorylation assays, the phosphorylation levels of 256

OsMPK6 obviously increased in the younger panicles of gsn1 and the repression 257

lines of GSN1 but markedly decreased in the overexpression lines (Supplemental 258

11

Figure 12). These results confirm that GSN1 can interact with and inactivate 259

OsMPK6 via dephosphorylation in vitro and in vivo. 260

261

GSN1 acts as a negative regulator of the MAPK cascade in coordinating the 262

trade-off between grain number and grain size 263

Given that GSN1 associates with and dephosphorylates OsMPK6 and that OsMKK4 264

can interact with and activate OsMPK6 (Figure 5D and Supplemental Figure 11C), 265

both of which are thought to act on the same cascade influencing rice grain size that 266

depends on cell proliferation (Duan et al., 2014; Liu et al., 2015), we attempted to 267

identify the components of the MAPK cascade that participate in panicle and 268

spikelet development in rice, as well as the genetic association between GSN1 and 269

the MAPK signaling pathway. We therefore performed genome editing using 270

CRISPR/Cas9 in the wild-type FAZ1 background. Unfortunately, we were not able 271

to isolate the related loss-of-function mutants for OsMPK6 due to embryo lethality, 272

which is consistent with the results of a previous study (Yi et al., 2016); instead, we 273

used gene knockdown mutants produced based on RNA interference (RNAi). We 274

found that the OsMPK6RNAi single mutant, which showed reduced expression of 275

OsMPK6, had reduced grain size and increased grain number per panicle compared 276

to the wild-type FAZ1, strongly suggesting that OsMPK6 is a positive regulator of 277

grain size and is required for panicle and spikelet development (Figures 6A, 6B and278

6G). In agreement with this notion, the down-regulation of OsMPK6 expression in 279

the gsn1/OsMPK6RNAi double mutant partially rescued the grain size and grain 280

number per panicle phenotypes of gsn1, which indicates that the larger grains and 281

sparser panicles in the gsn1 mutant rely on OsMPK6 and that there is a conclusive 282

genetic association between GSN1 and OsMPK6 (Figures 6A, 6B and 6G and283

Supplemental Figures 13A, 13B and 14A). Furthermore, plants of the OsMKK4 284

null mutant produced by CRISPR/Cas9 gene editing, OsMKK4CRISPR, exhibited 285

smaller grains and denser panicles than the wild type, which are very similar to the 286

phenotypes observed in the small grain1 mutants (Duan et al., 2014). In addition, the 287

double mutant gsn1/OsMKK4CRISPR perfectly rescued the mutant gsn1 phenotypes by 288

12

mimicking the grain size and grain number per panicle of FAZ1 (Figures 6C, 6D289

and 6H and Supplemental Figures 13C, 13D and 14B). Interestingly, 290

overexpression of the constitutively active version of OsMKK4 in FAZ1 resulted in 291

plants with increased grain size and reduced grain number per panicle 292

(Supplemental Figure 15). These results suggest that OsMKK4 is a positive 293

regulator of grain size but a negative regulator of grain number per panicle and that 294

the phenotypes of the gsn1 mutant that result from disordered cell proliferation and 295

differentiation largely depend on OsMKK4. Considering the association of the 296

phenotypes and protein characterization results between OsMPK6 and OsMKK4, we 297

hypothesized that the OsMKK4-OsMPK6 module, which participates in panicle and 298

spikelet development, may be inactivated by GSN1. 299

300

In Arabidopsis thaliana, the proposed complete YODA 301

(YDA)-MKK4/MKK5-MPK3/MPK6 cascade plays a critical role in specifying 302

inflorescence architecture by promoting localized cell proliferation; the yda mutant 303

has a tight, compact inflorescence, and YDA is an upstream regulator of 304

MKK4/MKK5 during inflorescence development (Meng et al., 2012). The rice 305

homolog of YDA has not yet been identified. We asked whether the rice homolog of 306

the Arabidopsis YDA gene might coordinate the trade-off between grain size and 307

grain number per panicle in rice and act upstream of the OsMKK4-OsMPK6 module. 308

We searched the Phytozome database and found that OsMKKK10 309

(LOC_Os04g47240) (Rao et al., 2010) shares high sequence similarity with 310

Arabidopsis YDA and has an unknown function in rice. Therefore, we generated 311

OsMKKK10 single mutants using CRISPR/Cas9, from which the OsMKKK10CRISPR 312

null mutant was isolated and characterized. In line with our hypothesis, the 313

loss-of-function mutant OsMKKK10CRISPR had smaller grains but denser panicles 314

with more grains per panicle than the wild type, which is analogous to the 315

OsMKK4CRISPR mutant, suggesting that OsMKKK10 plays a conserved role in plant 316

inflorescence development and is responsible for rice grain number per panicle and 317

grain size and may act upstream of OsMKK4 (Figures 6E, 6F and 6I and318

13

Supplemental Figures 13E, 13F and 14C). We also generated the 319

gsn1/OsMKKK10CRISPR double mutant and we found that it rescued the gsn1 320

phenotypes, suggesting that the phenotypes observed in the gsn1 mutant are also 321

mostly dependent on OsMKKK10 in rice and that this gene likely participates in 322

panicle and spikelet formation in the combined OsMKK4-OsMPK6 module 323

regulated by GSN1 (Figures 6E, 6F and 6I and Supplemental Figures 13E, 13F324

and 14C). 325

326

These results indicate that the reduced function of OsMPK6 and the loss of function 327

of OsMKK4 and OsMKKK10 can all rescue the gsn1 panicle and spikelet phenotypes 328

to different extents and together contribute to panicle and spikelet development, 329

which is consistent with the strong expression of OsMPK6, OsMKK4 and 330

OsMKKK10 in young panicles and spikelet hulls (Supplemental Figures 10I to331

10K). Finally, we assayed the phosphorylation level of OsMPK6 in young panicles 332

in different transgenic plants and found that suppression of OsMPK6, OsMKK4, and 333

OsMKKK10 expression rescued the phosphorylation level of OsMPK6 in gsn1 334

(Supplemental Figure 16). These results suggest that the phosphorylation status of 335

OsMPK6, which is precisely regulated by the GSN1-MAPK module, is pivotal for 336

panicle development in rice. In summary, we hypothesize that OsMPK6, OsMKK4, 337

and OsMKKK10 act in a common signaling pathway and that the proposed 338

OsMKKK10-OsMKK4-OsMPK6 cascade, which is inactivated by GSN1, is 339

responsible for specifying the trade-off between grain number per panicle and grain 340

size. This trade-off results from the coordination between cell differentiation in the 341

panicle and localized cell proliferation in the spikelet. 342

343

DISCUSSION 344

Both grain number per panicle and grain size are essential agronomic traits in rice 345

that strongly affect grain yield. It is well known that a panicle with larger grains 346

always has reduced grain number and that spatiotemporally, the determination of 347

panicle identity and spikelet number per panicle occurs prior to the complete 348

14

development of the spikelet. This is true because of the constraint of intrinsic 349

programs, although a larger panicle could simultaneously have both larger grains and 350

grain number (Hu et al., 2015; Si et al., 2016) or have both fixed grain size and more 351

grains when properly modified (Ashikari et al., 2005; Wu et al., 2016; Huo et al., 352

2017). Several genes involved in determining grain number and grain size have been 353

characterized, which always exhibit a preference and separate these two traits. 354

Nevertheless, the inherent correlation between grain size and grain number per 355

panicle tends to be ignored, although these two traits have different and interactive 356

developmental processes. It is well known that there is a negative correlation 357

between seed number and seed size in many plant species, which plays a central 358

evolutionary role (Sadras, 2007). However, the genetic basis for coordinating the 359

trade-off between grain number per panicle and grain size in rice remains largely 360

unknown. 361

362

In this study, we first determined that GSN1 is a negative regulator of grain size but a 363

positive regulator of grain number per panicle, which specifically links these two 364

traits through a conserved MAPK cascade. GSN1 encodes a dual-specificity 365

phosphatase that can inactivate OsMPK6 by dephosphorylation, and the 366

phosphorylation level of OsMPK6 is excessively enhanced in gsn1 (Figure 5 and 367

Supplemental Figures 12 and 16). GSN1 is localized to the cytoplasm and is 368

mainly expressed in younger panicles and spikelets (Figure 4). Reduced expression 369

of GSN1 resulted in larger but fewer grains per panicle, whereas its up-regulation 370

increased grain number per panicle but reduced grain size (Figure 2 and 371

Supplemental Figure 2). Molecular evidence revealed that the expression of most 372

of the downstream cell division-related genes was largely increased in the gsn1 373

mutant compared to FAZ1, whereas genes responsible for primordium differentiation 374

and grain number formation, such as DENSE AND ERECT PANICLE1 (DEP1), 375

Grain number 1a/CYTOKININ OXIDASE2 (Gn1a/OsCKX2), and LAX PANICLE1 376

(LAX1), were either significantly up- or down-regulated in the gsn1 mutant 377

(Supplemental Figure 7). Interestingly, loss of function of OsMKKK10 or OsMKK4, 378

15

and reduced expression of OsMPK6, rescued the larger grain size and reduced grain 379

number phenotypes in gsn1 (Figure 6 and Supplemental Figures 13 and 14), 380

supporting the hypothesis that GSN1 negatively regulates the 381

OsMKKK10-OsMKK4-OsMPK6 cascade. Although the proposed 382

YDA-MKK4/MKK5-MPK3/MPK6 cascade in Arabidopsis has been shown to 383

participate in inflorescence development (Meng et al., 2012), we found that the rice 384

OsMKKK10-OsMKK4-OsMPK6 cascade, which is inactivated by GSN1, plays a 385

distinct role in specifying the trade-off between grain number per panicle and grain 386

size. 387

388

Taken together, we proposed a working model for the role of the GSN1-MAPK 389

module in the trade-off between grain number per panicle and grain size. After 390

sensing panicle cell-specific communication signaling molecules (e.g., peptide 391

ligands), unknown receptors, including receptor-like protein kinases, activate the 392

OsMKKK10-OsMKK4-OsMPK6 cascade directly or through unknown mediators. 393

The activated OsMKKK10-OsMKK4-OsMPK6 cascade sequentially phosphorylates 394

its unidentified downstream substrates, which ultimately determine cell 395

differentiation in the panicle primordia and cell proliferation in the spikelet and 396

confer grain number per panicle and grain size at the young panicle developmental 397

stage. The spatiotemporally activated GSN1 protein acts like a molecular brake to 398

negatively regulate the OsMKKK10-OsMKK4-OsMPK6 cascade and precisely 399

inactivates OsMPK6, which then represses the phosphorylation of substrates. Hence, 400

the signaling output of the integrated GSN1-MAPK module coordinates the trade-off 401

between grain number per panicle and grain size, ultimately determining panicle 402

architecture in rice (Figure 7). 403

404

Mitogen-activated protein kinase cascades are highly conversed, pivotal signaling 405

pathways in eukaryotes. Numerous studies have demonstrated the multiple roles of 406

diverse, specific combinations of MAPK cascades in regulating cell growth or 407

environmental stress responses (Widmann et al., 1999). Genes for MAPK cascade 408

16

components are abundant in plants genomes and can form many distinct 409

MKKK-MKK-MPK combinations. However, studies of plant MAPKs have mainly 410

focused on their functions in immunity and stress responses, although MAPKs play 411

essential roles in multiple developmental programs. For example, the perception of 412

flagellin pathogen-associated molecular patterns (PAMPs) by receptors leads to the 413

activation of the MEKK1-MKK1/MKK2-MPK4 cascade, which represses cell death 414

and immune responses in Arabidopsis (Rodriguez et al., 2010; Kong et al., 2012). In 415

addition, in rice, the perception of chitin signals at the cell surface by the PAMP 416

receptor OsCERK1 triggers the rapid activation of MAPK cascade required for 417

resistance to rice blast fungus (Wang et al., 2017). While the 418

YODA-MKK4/MKK5-MPK3/MPK6 cascade also regulates stomatal development 419

and patterning by coordinating cell differentiation in Arabidopsis (Wang et al., 2007; 420

Lampard et al., 2008), the complete MAPK components and module that confer the 421

specific developmental programs in rice are still largely unknown. In this study, the 422

discovery of the proposed OsMKKK10-OsMKK4-OsMPK6 cascade, which 423

coordinates the trade-off between seed number and seed size, provides important 424

insights into plant inflorescence development. Furthermore, studies aimed at 425

identifying negative regulators of MAPK cascades specifying plant developmental 426

programs have rarely been performed. Here, we demonstrated that GSN1 encodes a 427

dual-specificity phosphatase that is a negative regulator of MAPK that specifically 428

dephosphorylates OsMPK6 and inactivates the OsMKKK10-OsMKK4-OsMPK6 429

cascade in rice to regulate the trade-off between grain number per panicle and grain 430

size. Reversible phosphorylation is an important regulatory mechanism that controls 431

the activities of some proteins and determines the intensity and duration of MAPK 432

signaling activation. Dual-specificity phosphatases are crucial negative regulators of 433

MAPK cascades, and they specifically dephosphorylate both phosphotyrosine and 434

phosphoserine/threonine residues within the activation motifs of MAPKs to trigger 435

the inactivation of MAPK cascades. Notably, we found that GSN1 also interacts with 436

and dephosphorylates OsMPK3 (LOC_Os03g17700), as does its homolog, OsMPK6, 437

implying that GSN1 associates with other MAPKs to regulate diverse processes 438

17

during rice development (Supplemental Figure 17). Therefore, the discovery of the 439

proposed rice GSN1-MAPK module provides important insights into the precise 440

regulation of MAPK cascades via a dual-specificity phosphatase to remodel plant 441

inflorescence development. 442

443

The essential phytohormones brassinosteroid (BR) and cytokinin play vital roles in 444

numerous developmental programs in plants. The small grain 1 and dwarf and small445

grains 1 rice mutants are relatively insensitive to BR and share several similar 446

phenotypes with BR mutants, such as small grains and erect panicles and leaves, 447

indicating that both OsMPK6 and OsMKK4 influence BR homeostasis and signaling 448

(Duan et al., 2014; Liu et al., 2015). Consistent with this notion, we found that the 449

gsn1 mutant was hypersensitive to BR and that the expression levels of BR-related 450

genes, including BR biosynthesis and signal transduction genes, were significantly 451

elevated in the gsn1 mutant, although the contents of BR and its precursor in 452

younger panicles were essentially unchanged (Supplemental Figure 18). These 453

results suggest that the loss-of-function of GSN1 can trigger the BR response and 454

that the maintenance of BR homeostasis is required for the functioning of the 455

GSN1-MAPK module, which might confer plasticity in grain size in the rice panicle. 456

457

Interestingly, attenuation of the cytokinin metabolic enzyme Gn1a/OsCKX2 causes 458

the accumulation of cytokinin and enhances cell proliferation in the meristem, 459

resulting in a significant increase in grain number (Ashikari et al., 2005). We found 460

that Gn1a/OsCKX2 was significantly up-regulated in gsn1 compared to FAZ1 during 461

young panicle developmental, but the expression of the cytokinin-activating gene 462

LOG was dramatically reduced in this mutant (Kurakawa et al., 2007) 463

(Supplemental Figure 7B). In agreement with this finding, the levels of several 464

cytokinins and their biosynthetic intermediates were significantly reduced in the 465

mutant (Supplemental Figure 19). Thus, we hypothesize that cytokinin activity is 466

substantially attenuated in the gsn1 panicle meristem, which contributes to its 467

reduced grain number per panicle. These findings point to a potential association 468

18

between the GSN1-MAPK module and phytohormone signaling in the determination 469

of panicle architecture plasticity in rice. 470

471

Although an evolutionary understanding of the trade-off and plasticity between seed 472

number and seed size is crucial for inflorescence remodeling and the improvement of 473

grain yield, little is known about the molecular mechanisms behind the plasticity 474

determining final inflorescence architecture in plants. Therefore, this study 475

represents an initial step in dissecting the mechanism by which the GSN1-MAPK 476

module coordinates the trade-off between grain number per panicle and grain size. 477

Further identification of the components upstream and downstream of the 478

GSN1-MAPK module will ultimately reveal the molecular mechanisms underlying 479

the coordination of localized cell differentiation and proliferation, which contributes 480

to the trade-off between grain number per panicle and grain size in rice. 481

482

METHODS 483

Plant materials and growth conditions 484

The gsn1 mutant was isolated from EMS (ethyl methanesulfonate)-treated seeds of 485

the elite indica rice (Oryza sativa) variety, Fengaizhan-1 (FAZ1). The M1 gsn1 486

mutant was crossed back to FAZ1 to produce M2 plants in an isogenic background. 487

All rice plants were cultivated in experimental fields in Shanghai and Hainan, China 488

under natural growth conditions. 489

490

Map-based cloning of GSN1 491

The gsn1 mutant was crossed with the japonica variety, Zhonghua 11 (ZH11), and F1 492

plants were self-pollinated to generate an F2 mapping population. To fine-map the 493

GSN1 locus, new molecular markers were developed. GSN1 was mapped to a 494

17.5-kb region on the short arm of chromosome 5, and DNA fragments from this 495

region were amplified from genomic DNA from both gsn1 and the wild type for 496

further sequencing. The PCR primer sets used for gene amplification are shown in 497

Supplemental Data Set 1. 498

19

499

Plasmid construction and plant transformation 500

To produce the complementation construct pCAMBIA1300-gGSN1, the full-length 501

genomic sequence of GSN1 was amplified from the wild type and cloned into the 502

plant binary vector pCAMBIA1300. A 2.5-kb DNA fragment upstream of the GSN1 503

start codon was amplified from FAZ1 and cloned into the pCAMBIA1300-GUSplus 504

vector to generate the plasmid ProGSN1:GUS. The artificial microRNA oligo 505

sequences used for GSN1 silencing were designed as previously described 506

(Warthmann et al., 2008). To generate the overexpression constructs, the full-length 507

coding sequence of GSN1 was amplified from both FAZ1 and gsn1 and cloned into 508

plant binary vectors pCAMBIA1306 and pCAMBIA1301 under the control of the 509

CaMV 35S promoter and the ubiquitin (UBI) promoter, respectively. To generate the 510

OsMPK6 RNAi lines, an OsMPK6 DNA fragment from the FAZ1 wild-type line was 511

amplified and cloned into pTCK303 to obtain the OsMPK6 RNAi construct. To 512

generate constitutively active OsMKK4CA, the full-length coding sequence of 513

OsMKK4 harboring the T238D and S244D mutations was cloned into 514

pCAMBIA1301. The gene editing constructs for GSN1, OsMKKK10, and OsMKK4 515

via CRISPR/Cas9 were designed as previously described (Ma et al., 2015). 516

Agrobacterium tumefaciens-mediated transformation of rice with strain EHA105 was 517

performed as previously described (Hiei et al., 1994). The DNA constructs used in 518

this study were generated using NEBuilder HiFi DNA Assembly Master Mix (NEB). 519

All constructs were confirmed by sequencing. The PCR primer sets are given in 520


522

RNA extraction and qRT-PCR 523

Total RNA was extracted from various plant tissues using TRIZOL Reagent 524

(Invitrogen). Reverse transcription was performed using ReverTra Ace qPCR RT 525

Master Mix with gDNA Remover (Toyobo) from 500 ng total RNA. Quantitative 526

RT-PCR analysis was performed on the ABI 7300 Real Time PCR System with Fast 527

Start Universal SYBR Green Master Mix with ROX (Roche), and the data were 528

20

analyzed using the 2-ΔΔCT method. The UBQ5 gene was used as the internal reference 529

to normalize gene expression data. All analyses were repeated at least three times. 530

PCR primer sets for gene amplification are given in Supplemental Data Set 1. 531

532

GUS staining 533

GUS staining of ProGSN1:GUS transgenic plants was performed as described 534

previously (Duan et al., 2014). The samples were stained in GUS staining buffer 535

overnight at 37°C. The samples were then cleared in 75% ethanol to remove 536

chlorophyll. 537

538

Histochemical analysis 539

Plant materials were fixed in FAA (50% ethanol, 5% glacial acetic acid, 5% 540

formaldehyde) overnight at 4°C and dehydrated in a graded alcohol series. After 541

fixing with xylene, the samples were embedded in Paraplast (Sigma-Aldrich) and 542

sliced into 8-μm thick sections with a rotary microtome (Leica). The prepared tissue 543

sections were stained with safranin and observed under a light microscope (Leica). 544

545

Scanning electron microscopy 546

Spikelet hulls harvested just before heading were fixed in FAA (50% ethanol, 5% 547

glacial acetic acid, 5% formaldehyde) overnight at 4°C and dehydrated in a graded 548

alcohol series. The samples were then dried in a critical point drier (Leica), gold 549

sputter coated, and observed under a scanning electron microscope (Hitachi). Cell 550

size and cell number were calculated using Image J software after scanning. 551

552

Nucleus isolation and assessment of ploidy 553

Isolation of cell nuclei and assessment of ploidy were performed as described 554

previously (Qi et al., 2012). Spikelet hulls of younger panicles were selected, soaked 555

separately in nuclear isolation and staining solution (Beckman), and chopped with a 556

sharp blade. After filtering the slurry through a 40-μm nylon filter, the suspension of 557

nuclei was loaded into a Beckman Moflo for flow cytometric analysis, and the 558

21

ploidy of approximate 10,000 nuclei was recorded for each test. The numbers of 559

diploid and tetraploid nuclei were recorded, and the relative proportions of G1, S, 560

and G2/M phase cells were calculated using FCS Express 4 software. 561

562

Subcellular localization of GSN1 563

To determine the subcellular localization of GSN1, the Pro35S:GSN1-GFP DNA 564

constructs were introduced into Agrobacterium tumefaciens strain GV3101 and 565

infiltrated into the leaves of Nicotiana benthamiana plants. The GSN1 fragment was 566

inserted into the pA7-YFP plasmid and transformed into rice protoplasts. GFP 567

fluorescence in leaf epidermal cells and YFP fluorescence in rice protoplasts were 568

detected with an LSM 880 confocal laser-scanning microscope (Zeiss). The 569

sequences of the PCR primers used for vector construction are given in 570


572

Yeast two-hybrid assays 573

Yeast two-hybrid assays were performed with the Y2H Gold-Gal4 system (Clontech). 574

DNA fragments containing the GSN1, OsMPK6, OsMPK3, and OsMKK4 genes were 575

inserted into the pGBKT7 and pGADT7 vectors to form the bait and prey constructs, 576

respectively. The bait and prey constructs were transformed into yeast strain Y2H 577

Gold according to the manufacturer’s instructions (Clontech). The yeast cells were 578

cultured on SD/-Trp-Leu-His or SD/-Trp-Leu-His-Ade medium containing X-ɑ-gal 579

at 30 °C in the dark for three days. SD/-Trp-Leu-His is yeast culture medium without 580

tryptophan, leucine, and histidine. SD/-Trp-Leu-His-Ade is culture medium without 581

tryptophan, leucine, histidine, and adenine. The PCR primers used for Y2H are given 582

in Supplemental Data Set 1. 583

584

Bimolecular fluorescence complementation (BiFC) assays 585

For the BiFC assays, the DNA fragments containing the GSN1, OsMPK6, OsMPK3586

and OsHAL3 genes were cloned into pCAMBIA1300S-YN and 587

pCAMBIA1300S-YC to form the nYFP-protein and cYFP-protein constructs, 588

22

respectively. Leaves of five-week-old Nicotiana benthamiana plants were 589

co-infiltrated with Agrobacterium tumefaciens strain GV3101 carrying the two 590

constructs. The plants were grown in the dark for 48 hours after infiltration, and 591

BiFC fluorescence or protein localization fluorescence signals were then observed 592

using an LSM 880 confocal laser-scanning microscope (Zeiss). PCR primers used 593

for BiFC are given in Supplemental Data Set 1. 594

595

Co-immunoprecipitation 596

The DNA fragment carrying GSN1 was cloned into the pCAMBIA1306-Flag (3x) 597

plasmid to produce the Pro35S:GSN1-Flag vector. The OsMPK6 DNA fragment was 598

cloned into pCAMBIA1301-Myc (7x)-His (6x) to generate the 599

Pro35S:OsMPK6-Myc vector. The Pro35S:GSN1-Flag and Pro35S:OsMPK6-Myc 600

constructs in Agrobacterium tumefaciens strain GV3101 were transiently 601

co-expressed in Nicotiana benthamiana leaf cells following agroinfiltration. Protein 602

extraction and immunoblot analyses were performed as previously described (Lee et 603

al., 2012). The immunoblot (WB for western blot) assays were performed using 604

anti-Flag (Cell Signaling Technology, 14793) and anti-Myc (Cell Signaling 605

Technology, 2276) antibodies at 1:1000 dilution. The PCR primers used for Co-IP 606

are given in Supplemental Data Set 1. 607

608

Protein expression 609

The coding sequences of GSN1, OsMKK4, and their mutated versions were inserted 610

into the pMAL-c5x vector. The coding regions of OsMPK6 and OsMPK3 were 611

cloned into the pGEX-4T-2 vector. The recombinant pMAL-c5x and pGEX-4T-2 612

vectors were then transformed into Rossetta-gami (DE3) pLysS competent cells, and 613

cultured to produce fusion proteins. Amylose resin (NEB) and glutathione-agarose 614

4B (GE Healthcare) were used for fusion protein purification. The sequences of PCR 615

primers used for protein expression are given in Supplemental Data Set 1. 616

617

Phosphatase activity assay 618

23

Phosphatase activity assays were performed as described previously (Lee et al., 619

2008). To analyze the activity of the dual-specificity phosphatase, OMFP 620

(Sigma-Aldrich) and PNPP (Thermo Fisher) were used as substrates. The 621

phosphatase activity of GSN1 or GSN1S146F at various enzyme concentrations (0, 10, 622

20, 30, 40, 50, 60, 80, and 100 μg) was measured in 200 μl reactions containing 623

phosphatase buffer (50 mM Tris-HCl pH 8.0, 15 mM NaCl, 1 mM EDTA) and 500 624

μM OMFP or PNPP, and the mixtures were incubated at 30 °C for 1 h. Also, a series 625

of OMFP concentrations (50, 100, 200, 300, 400, 500, 600, 700, 800, and 1000 μM) 626

was prepared to analyze enzyme activity kinetics with the same amount of 627

phosphatase at different time points (10, 30, 45, 60, 90, 120, 150, 180, 210, 240, 270 628

and 300 min). The amounts of dephosphorylated OMFP and PNPP were recorded 629

based on the absorbance at 477 nm and 405 nm, respectively after the 630

dephosphorylation reaction. The kinetic parameters Km and Vmax for GSN1 were 631

determined using OMFP as substrate and calculated using GraphPad Prism 5 632

software. 633

634

In vitro dephosphorylation assays 635

The purified fusion proteins MBP-GSN1, MBP-GSN1S146F, MBP-OsMKK4CA, 636

GST-OsMPK6 and GST-OsMPK3 were used in the dephosphorylation assays. 637

Purified MBP-OsMKK4CA was mixed with recombinant GST-OsMPK6 and 638

GST-OsMPK3 at a 1:3 ratio in kinase buffer (25 mM Tris-HCl PH7.5, 5 mM 639

β-glycerolphosphate, 2 mM dithiothreitol, 10 mM MgCl2, 200 μM ATP) and 640

incubated at 30 °C for 30 min. The reaction mixture was then desalted to separate 641

free ATP. To dephosphorylate the fusion proteins GST-OsMPK6 and GST-OsMPK3, 642

the recombinant MBP-GSN1 and MBP-GSN1S146F proteins were mixed separately 643

with phosphorylated GST-OsMPK6 and GST-OsMPK3 protein at a 1:4 ratio and 644

incubated at 30 °C for 30 min. The reaction was terminated by the addition of 645

concentrated SDS-PAGE sample buffer, followed by boiling for 5 min. 646

Phosphorylated GST-OsMPK6 and GST-OsMPK3 were visualized by immunoblot 647

24

analysis using the polyclonal antibody phospho-p44/42 at 1:2000 dilution (Cell 648

Signaling Technology, 9101), which specifically recognizes the dual-phosphorylated 649

T-E-Y motif of the phosphorylated GST-OsMPK6 and GST-OsMPK3. The loading 650

control was probed using anti-AtMPK6 (Sigma, A7104) and anti-AtMPK3 (Sigma, 651

M8318) antibodies against Arabidopsis MPK6 and MPK3, respectively at 1:2000 652

dilution. 653

654

Mass spectrometry analysis 655

The proteins from the phosphorylation and dephosphorylation assays were reduced 656

with 10 mM DTT and alkylated with 50 mM iodoacetamide. In-solution digestion 657

was carried out using sequencing-grade modified trypsin (Promega) at 37°C 658

overnight. The tryptic peptides were acidated with a final concentration of 0.5-1% 659

trifluoroacetic acid. For mass spectrometry analysis, the peptides were separated via 660

a 90 min gradient elution at a flow rate 0.22 μl/min with the EASY-nLC 1000 HPLC 661

system (Thermo Scientific), which was directly interfaced with a Q Exactive mass 662

spectrometer (Thermo Scientific). The mobile phase consisted of 0.1% formic acid, 663

and the mobile phase consisted of acetonitrile with 0.1% formic acid. The Q 664

Exactive mass spectrometer was operated in the data-dependent acquisition mode 665

using Xcalibur 2.2 SP1 software. The MS/MS spectra from each run were searched 666

against the target sequence using the Sequest HT and phosphoRS 3.0 modules in 667

Proteome Discoverer software PD1.4 (Thermo Scientific), which followed these 668

criteria: full tryptic specificity was required; two missing cleavages were allowed; 669

carbamidomethylation were set to fixed modifications; the oxidation was set to 670

dynamic modification; precursor mass tolerances were set to 10 ppm; and the 671

fragment mass tolerance was set to 0.02 Da. The peptide false discovery rate 672

(FDR)<1% was set using percolator in PD1.4. 673

674

Accession numbers 675

Sequence data from this article can be found in the GenBank/EMBL libraries under 676

the following accession numbers: GSN1, LOC_Os05g02500; OsMPK6, 677

25

LOC_Os06g06090; OsMPK3, LOC_Os03g17700; OsMKK4, LOC_Os02g54600; 678

and OsMKKK10, LOC_Os04g47240. 679

680

Supplemental Data 681

Supplemental Figure 1. GSN1 is required for fertility in rice.682

Supplemental Figure 2. Panicle phenotypes in transgenic RNA interference 683

GSN1RNAi lines and ProUBI:GSN1 overexpression lines.684

Supplemental Figure 3. Phenotypic characterization of 95-22 and the nearly 685

isogenic line for gsn1, NIL(gsn1).686

Supplemental Figure 4. Suppression of GSN1 in ZH11 results in larger grains and 687

sparser panicles. 688

Supplemental Figure 5. GSN1S146F has a dominant negative effect on panicle 689

development in ZH11. 690

Supplemental Figure 6. Overexpression of GSN1 in ZH11 results in smaller grains 691

and denser panicles.692

Supplemental Figure 7. Expression of cell cycle-related genes in young panicles 693

(0.5-2 cm) of FAZ1 and gsn1.694

Supplemental Figure 8. Overexpression of GSN1 decreases cell proliferation rather 695

than cell size to control grain length in FAZ1.696

Supplemental Figure 9. Repression of GSN1 increases cell proliferation rather than 697

cell expansion to control grain length in ZH11.698

Supplemental Figure 10. GUS staining showing GSN1 expression in different 699

tissues of ProGSN1:GUS transgenic rice plants.700

Supplemental Figure 11. GSN1 specifically interacts with and inactivates OsMPK6 701

and not OsMKK4, but OsMPK6 interacts with OsMKK4.702

Supplemental Figure 12. GSN1 dephosphorylates OsMPK6 in vivo.703

Supplemental Figure 13. The reduced expression of OsMPK6 or the 704

loss-of-function of OsMKK4 or OsMKKK10 rescues the gsn1 phenotypes.705

Supplemental Figure 14. Confirmation of the OsMPK6RNAi, OsMKK4CRISPR, and 706

OsMKKK10CRISPR CRISPR/Cas9 transgenic lines. 707

26

Supplemental Figure 15. Overexpression of the constitutively active version of708

OsMKK4, ProUBI:OsMKK4CA, specifically increases grain size in rice.709

Supplemental Figure 16. Suppression of OsMPK6, OsMKK4, and OsMKKK10 710

rescues the phosphorylation level of OsMPK6 in gsn1.711

Supplemental Figure 17. GSN1 interacts with and dephosphorylates OsMPK3.712

Supplemental Figure 18. The gsn1 mutant is hypersensitive to BR, and GSN1 713

influences BR signaling. 714

Supplemental Figure 19. GSN1 is associated with the maintenance of cytokinin 715

homeostasis during panicle development in rice.716

Supplemental Data Set 1. Primers used in this study. 717

718

Acknowledgements 719

We thank Min Shi (Institute of Plant Physiology and Ecology, SIBS, CAS) for 720

technical support with the transgenic assay. We thank Xiaoshu Gao, Xiaoyan Gao, 721

Zhiping Zhang, Jiqin Li and Wenfang Zhao (Institute of Plant Physiology and 722

Ecology, SIBS, CAS) for technical support. We thank Professor Jian Zhang (China 723

National Rice Research Institute) and Professor Kang Chong (Institute of Botany, 724

CAS) for donating the plasmids. This work was supported by grants from the 725

Chinese Academy of Sciences (XDA08010102, QYZDY-SSW-SMC023, 726

XDPB0404), National Natural Science Foundation of China (31788103), the 727

Ministry of Science and Technology of China (2016YFD0100902) and 728

CAS-Croucher Funding Scheme for Joint Laboratories. 729

730

Author Contributions 731

HXL, JXS and TG designed the study. TG, KC, NQD, CLS, WWY, JPG, JXS, and 732

HXL performed the experiments. TG, and HXL analyzed the data and wrote the 733

paper. 734

735

736

References 737

27

Ashikari, M., Sakakibara, H., Lin, S., Yamamoto, T., Takashi, T., Nishimura, A., 738

Angeles, E.R., Qian, Q., Kitano, H., and Matsuoka, M. (2005). Cytokinin 739

oxidase regulates rice grain production. Science 309, 741-745. 740

Bartels, S., Gonzalez Besteiro, M.A., Lang, D., and Ulm, R. (2010). Emerging 741

functions for plant MAP kinase phosphatases. Trends Plant Sci 15, 322-329. 742

Duan, P., Rao, Y., Zeng, D., Yang, Y., Xu, R., Zhang, B., Dong, G., Qian, Q., and 743

Li, Y. (2014). SMALL GRAIN 1, which encodes a mitogen-activated protein 744

kinase kinase 4, influences grain size in rice. Plant J 77, 547-557. 745

Hiei, Y., Ohta, S., Komari, T., and Kumashiro, T. (1994). Efficient transformation 746

of rice (Oryza sativa L.) mediated by Agrobacterium and sequence analysis 747

of the boundaries of the T-DNA. Plant J 6, 271-282. 748

Hu, J., Wang, Y., Fang, Y., Zeng, L., Xu, J., Yu, H., Shi, Z., Pan, J., Zhang, D., 749

Kang, S., Zhu, L., Dong, G., Guo, L., Zeng, D., Zhang, G., Xie, L., Xiong, 750

G., Li, J., and Qian, Q. (2015). A Rare Allele of GS2 Enhances Grain Size 751

and Grain Yield in Rice. Mol Plant 8, 1455-1465. 752

Huang, X., Qian, Q., Liu, Z., Sun, H., He, S., Luo, D., Xia, G., Chu, C., Li, J., 753

and Fu, X. (2009). Natural variation at the DEP1 locus enhances grain yield 754

in rice. Nat Genet 41, 494-497. 755

Huo, X., Wu, S., Zhu, Z., Liu, F., Fu, Y., Cai, H., Sun, X., Gu, P., Xie, D., Tan, L., 756

and Sun, C. (2017). NOG1 increases grain production in rice. Nat Commun 757

8, 1497. 758

Kong, Q., Qu, N., Gao, M., Zhang, Z., Ding, X., Yang, F., Li, Y., Dong, O.X., 759

Chen, S., Li, X., and Zhang, Y. (2012). The MEKK1-MKK1/MKK2-MPK4 760

kinase cascade negatively regulates immunity mediated by a 761

mitogen-activated protein kinase kinase kinase in Arabidopsis. Plant Cell 24, 762

2225-2236. 763

Kurakawa, T., Ueda, N., Maekawa, M., Kobayashi, K., Kojima, M., Nagato, Y., 764

Sakakibara, H., and Kyozuka, J. (2007). Direct control of shoot meristem 765

activity by a cytokinin-activating enzyme. Nature 445, 652-655. 766

Lampard, G.R., Macalister, C.A., and Bergmann, D.C. (2008). Arabidopsis 767

28

stomatal initiation is controlled by MAPK-mediated regulation of the bHLH 768

SPEECHLESS. Science 322, 1113-1116. 769

Lee, J.S., Kuroha, T., Hnilova, M., Khatayevich, D., Kanaoka, M.M., McAbee, 770

J.M., Sarikaya, M., Tamerler, C., and Torii, K.U. (2012). Direct 771

interaction of ligand-receptor pairs specifying stomatal patterning. Genes 772

Dev 26, 126-136. 773

Lee, K., Song, E.H., Kim, H.S., Yoo, J.H., Han, H.J., Jung, M.S., Lee, S.M., Kim, 774

K.E., Kim, M.C., Cho, M.J., and Chung, W.S. (2008). Regulation of 775

MAPK phosphatase 1 (AtMKP1) by calmodulin in Arabidopsis. J Biol Chem 776

283, 23581-23588. 777

Li, Y., Fan, C., Xing, Y., Jiang, Y., Luo, L., Sun, L., Shao, D., Xu, C., Li, X., Xiao, 778

J., He, Y., and Zhang, Q. (2011). Natural variation in GS5 plays an 779

important role in regulating grain size and yield in rice. Nat Genet 43, 780

1266-1269. 781

Liu, S., Hua, L., Dong, S., Chen, H., Zhu, X., Jiang, J., Zhang, F., Li, Y., Fang, 782

X., and Chen, F. (2015). OsMAPK6, a mitogen-activated protein kinase, 783

influences rice grain size and biomass production. Plant J 84, 672-681. 784

Luan, S. (2003). Protein phosphatases in plants. Annu Rev Plant Biol 54, 63-92. 785

Ma, X., Zhang, Q., Zhu, Q., Liu, W., Chen, Y., Qiu, R., Wang, B., Yang, Z., Li, 786

H., Lin, Y., Xie, Y., Shen, R., Chen, S., Wang, Z., Chen, Y., Guo, J., Chen, 787

L., Zhao, X., Dong, Z., and Liu, Y.G. (2015). A Robust CRISPR/Cas9 788

System for Convenient, High-Efficiency Multiplex Genome Editing in 789

Monocot and Dicot Plants. Mol Plant 8, 1274-1284. 790

Mao, H., Sun, S., Yao, J., Wang, C., Yu, S., Xu, C., Li, X., and Zhang, Q. (2010). 791

Linking differential domain functions of the GS3 protein to natural variation 792

of grain size in rice. Proc Natl Acad Sci U S A 107, 19579-19584. 793

Meng, X., and Zhang, S. (2013). MAPK cascades in plant disease resistance 794

signaling. Annu Rev Phytopathol 51, 245-266. 795

Meng, X., Wang, H., He, Y., Liu, Y., Walker, J.C., Torii, K.U., and Zhang, S. 796

(2012). A MAPK cascade downstream of ERECTA receptor-like protein 797

29

kinase regulates Arabidopsis inflorescence architecture by promoting 798

localized cell proliferation. Plant Cell 24, 4948-4960. 799

Oikawa, T., and Kyozuka, J. (2009). Two-Step Regulation of LAX PANICLE1 800

Protein Accumulation in Axillary Meristem Formation in Rice. Plant Cell 21, 801

1095-1108. 802

Qi, P., Lin, Y.S., Song, X.J., Shen, J.B., Huang, W., Shan, J.X., Zhu, M.Z., Jiang, 803

L., Gao, J.P., and Lin, H.X. (2012). The novel quantitative trait locus GL3.1 804

controls rice grain size and yield by regulating Cyclin-T1;3. Cell Res 22, 805

1666-1680. 806

Rao, K.P., Richa, T., Kumar, K., Raghuram, B., and Sinha, A.K. (2010). In silico 807

analysis reveals 75 members of mitogen-activated protein kinase kinase 808

kinase gene family in rice. DNA Res 17, 139-153. 809

Reyna, N.S., and Yang, Y. (2006). Molecular analysis of the rice MAP kinase gene 810

family in relation to Magnaporthe grisea infection. Mol Plant Microbe 811

Interact 19, 530-540. 812

Rodriguez, M.C., Petersen, M., and Mundy, J. (2010). Mitogen-activated protein 813

kinase signaling in plants. Annu Rev Plant Biol 61, 621-649. 814

Sadras, V.O. (2007). Evolutionary aspects of the trade-off between seed size and 815

number in crops. Field Crops Research 100, 125-138. 816

Si, L., Chen, J., Huang, X., Gong, H., Luo, J., Hou, Q., Zhou, T., Lu, T., Zhu, J., 817

Shangguan, Y., Chen, E., Gong, C., Zhao, Q., Jing, Y., Zhao, Y., Li, Y., 818

Cui, L., Fan, D., Lu, Y., Weng, Q., Wang, Y., Zhan, Q., Liu, K., Wei, X., 819

An, K., An, G., and Han, B. (2016). OsSPL13 controls grain size in 820

cultivated rice. Nat Genet 48, 447-456. 821

Song, X.J., Huang, W., Shi, M., Zhu, M.Z., and Lin, H.X. (2007). A QTL for rice 822

grain width and weight encodes a previously unknown RING-type E3 823

ubiquitin ligase. Nat Genet 39, 623-630. 824

Su, L., Shan, J.X., Gao, J.P., and Lin, H.X. (2016). OsHAL3, a Blue 825

Light-Responsive Protein, Interacts with the Floral Regulator Hd1 to Activate 826

Flowering in Rice. Mol Plant 9, 233-244. 827

30

Wang, C., Wang, G., Zhang, C., Zhu, P., Dai, H., Yu, N., He, Z., Xu, L., and 828

Wang, E. (2017). OsCERK1-Mediated Chitin Perception and Immune 829

Signaling Requires Receptor-like Cytoplasmic Kinase 185 to Activate an 830

MAPK Cascade in Rice. Mol Plant 10, 619-633. 831

Wang, H., Ngwenyama, N., Liu, Y., Walker, J.C., and Zhang, S. (2007). Stomatal 832

development and patterning are regulated by environmentally responsive 833

mitogen-activated protein kinases in Arabidopsis. Plant Cell 19, 63-73. 834

Warthmann, N., Chen, H., Ossowski, S., Weigel, D., and Herve, P. (2008). Highly 835

specific gene silencing by artificial miRNAs in rice. PLoS One 3, e1829. 836

Widmann, C., Gibson, S., Jarpe, M.B., and Johnson, G.L. (1999). 837

Mitogen-activated protein kinase: conservation of a three-kinase module 838

from yeast to human. Physiol Rev 79, 143-180. 839

Wu, Y., Wang, Y., Mi, X.F., Shan, J.X., Li, X.M., Xu, J.L., and Lin, H.X. (2016). 840

The QTL GNP1 Encodes GA20ox1, Which Increases Grain Number and 841

Yield by Increasing Cytokinin Activity in Rice Panicle Meristems. PLoS 842

Genet 12, e1006386. 843

Xing, Y., and Zhang, Q. (2010). Genetic and molecular bases of rice yield. Annu 844

Rev Plant Biol 61, 421-442. 845

Xu, J., and Zhang, S. (2015). Mitogen-activated protein kinase cascades in 846

signaling plant growth and development. Trends Plant Sci 20, 56-64. 847

Yang, K.Y., Liu, Y., and Zhang, S. (2001). Activation of a mitogen-activated 848

protein kinase pathway is involved in disease resistance in tobacco. Proc Natl 849

Acad Sci U S A 98, 741-746. 850

Yi, J., Lee, Y.S., Lee, D.Y., Cho, M.H., Jeon, J.S., and An, G. (2016). OsMPK6 851

plays a critical role in cell differentiation during early embryogenesis in 852

Oryza sativa. J Exp Bot 67, 2425-2437. 853

Yoo, J.H., Cheong, M.S., Park, C.Y., Moon, B.C., Kim, M.C., Kang, Y.H., Park, 854

H.C., Choi, M.S., Lee, J.H., Jung, W.Y., Yoon, H.W., Chung, W.S., Lim, 855

C.O., Lee, S.Y., and Cho, M.J. (2004). Regulation of the dual specificity 856

protein phosphatase, DsPTP1, through interactions with calmodulin. J Biol 857

31

Chem 279, 848-858. 858

Zhang, D., and Yuan, Z. (2014). Molecular control of grass inflorescence 859

development. Annu Rev Plant Biol 65, 553-578. 860

Zuo, J., and Li, J. (2014). Molecular genetic dissection of quantitative trait loci 861

regulating rice grain size. Annu Rev Genet 48, 99-118. 862

Parsed Citations

Ashikari, M., Sakakibara, H., Lin, S., Yamamoto, T., Takashi, T., Nishimura, A., Angeles, E.R., Qian, Q., Kitano, H., and Matsuoka, M.(2005). Cytokinin oxidase regulates rice grain production. Science 309, 741-745.

Pubmed: Author and TitleCrossRef: Author and TitleGoogle Scholar: Author Only Title Only Author and Title

Bartels, S., Gonzalez Besteiro, M.A., Lang, D., and Ulm, R. (2010). Emerging functions for plant MAP kinase phosphatases. Trends PlantSci 15, 322-329.


Duan, P., Rao, Y., Zeng, D., Yang, Y., Xu, R., Zhang, B., Dong, G., Qian, Q., and Li, Y. (2014). SMALL GRAIN 1, which encodes a mitogen-activated protein kinase kinase 4, influences grain size in rice. Plant J 77, 547-557.


Hiei, Y., Ohta, S., Komari, T., and Kumashiro, T. (1994). Efficient transformation of rice (Oryza sativa L.) mediated by Agrobacterium andsequence analysis of the boundaries of the T-DNA. Plant J 6, 271-282.


Hu, J., Wang, Y., Fang, Y., Zeng, L., Xu, J., Yu, H., Shi, Z., Pan, J., Zhang, D., Kang, S., Zhu, L., Dong, G., Guo, L., Zeng, D., Zhang, G., Xie,L., Xiong, G., Li, J., and Qian, Q. (2015). A Rare Allele of GS2 Enhances Grain Size and Grain Yield in Rice. Mol Plant 8, 1455-1465.


Huang, X., Qian, Q., Liu, Z., Sun, H., He, S., Luo, D., Xia, G., Chu, C., Li, J., and Fu, X. (2009). Natural variation at the DEP1 locusenhances grain yield in rice. Nat Genet 41, 494-497.


Huo, X., Wu, S., Zhu, Z., Liu, F., Fu, Y., Cai, H., Sun, X., Gu, P., Xie, D., Tan, L., and Sun, C. (2017). NOG1 increases grain production inrice. Nat Commun 8, 1497.


Kong, Q., Qu, N., Gao, M., Zhang, Z., Ding, X., Yang, F., Li, Y., Dong, O.X., Chen, S., Li, X., and Zhang, Y. (2012). The MEKK1-MKK1/MKK2-MPK4 kinase cascade negatively regulates immunity mediated by a mitogen-activated protein kinase kinase kinase inArabidopsis. Plant Cell 24, 2225-2236.


Kurakawa, T., Ueda, N., Maekawa, M., Kobayashi, K., Kojima, M., Nagato, Y., Sakakibara, H., and Kyozuka, J. (2007). Direct control ofshoot meristem activity by a cytokinin-activating enzyme. Nature 445, 652-655.


Lampard, G.R., Macalister, C.A., and Bergmann, D.C. (2008). Arabidopsis stomatal initiation is controlled by MAPK-mediated regulationof the bHLH SPEECHLESS. Science 322, 1113-1116.


Lee, J.S., Kuroha, T., Hnilova, M., Khatayevich, D., Kanaoka, M.M., McAbee, J.M., Sarikaya, M., Tamerler, C., and Torii, K.U. (2012).Direct interaction of ligand-receptor pairs specifying stomatal patterning. Genes Dev 26, 126-136.


Lee, K., Song, E.H., Kim, H.S., Yoo, J.H., Han, H.J., Jung, M.S., Lee, S.M., Kim, K.E., Kim, M.C., Cho, M.J., and Chung, W.S. (2008).Regulation of MAPK phosphatase 1 (AtMKP1) by calmodulin in Arabidopsis. J Biol Chem 283, 23581-23588.


Li, Y., Fan, C., Xing, Y., Jiang, Y., Luo, L., Sun, L., Shao, D., Xu, C., Li, X., Xiao, J., He, Y., and Zhang, Q. (2011). Natural variation in GS5plays an important role in regulating grain size and yield in rice. Nat Genet 43, 1266-1269.


Liu, S., Hua, L., Dong, S., Chen, H., Zhu, X., Jiang, J., Zhang, F., Li, Y., Fang, X., and Chen, F. (2015). OsMAPK6, a mitogen-activatedprotein kinase, influences rice grain size and biomass production. Plant J 84, 672-681.


Luan, S. (2003). Protein phosphatases in plants. Annu Rev Plant Biol 54, 63-92.


Ma, X., Zhang, Q., Zhu, Q., Liu, W., Chen, Y., Qiu, R., Wang, B., Yang, Z., Li, H., Lin, Y., Xie, Y., Shen, R., Chen, S., Wang, Z., Chen, Y.,Guo, J., Chen, L., Zhao, X., Dong, Z., and Liu, Y.G. (2015). A Robust CRISPR/Cas9 System for Convenient, High-Efficiency MultiplexGenome Editing in Monocot and Dicot Plants. Mol Plant 8, 1274-1284.


Mao, H., Sun, S., Yao, J., Wang, C., Yu, S., Xu, C., Li, X., and Zhang, Q. (2010). Linking differential domain functions of the GS3 protein tonatural variation of grain size in rice. Proc Natl Acad Sci U S A 107, 19579-19584.


Meng, X., and Zhang, S. (2013). MAPK cascades in plant disease resistance signaling. Annu Rev Phytopathol 51, 245-266.


Meng, X., Wang, H., He, Y., Liu, Y., Walker, J.C., Torii, K.U., and Zhang, S. (2012). A MAPK cascade downstream of ERECTA receptor-likeprotein kinase regulates Arabidopsis inflorescence architecture by promoting localized cell proliferation. Plant Cell 24, 4948-4960.


Oikawa, T., and Kyozuka, J. (2009). Two-Step Regulation of LAX PANICLE1 Protein Accumulation in Axillary Meristem Formation in Rice.Plant Cell 21, 1095-1108.


Qi, P., Lin, Y.S., Song, X.J., Shen, J.B., Huang, W., Shan, J.X., Zhu, M.Z., Jiang, L., Gao, J.P., and Lin, H.X. (2012). The novel quantitativetrait locus GL3.1 controls rice grain size and yield by regulating Cyclin-T1;3. Cell Res 22, 1666-1680.


Rao, K.P., Richa, T., Kumar, K., Raghuram, B., and Sinha, A.K. (2010). In silico analysis reveals 75 members of mitogen-activated proteinkinase kinase kinase gene family in rice. DNA Res 17, 139-153.


Reyna, N.S., and Yang, Y. (2006). Molecular analysis of the rice MAP kinase gene family in relation to Magnaporthe grisea infection. MolPlant Microbe Interact 19, 530-540.


Rodriguez, M.C., Petersen, M., and Mundy, J. (2010). Mitogen-activated protein kinase signaling in plants. Annu Rev Plant Biol 61, 621-649.


Sadras, V.O. (2007). Evolutionary aspects of the trade-off between seed size and number in crops. Field Crops Research 100, 125-138.


Si, L., Chen, J., Huang, X., Gong, H., Luo, J., Hou, Q., Zhou, T., Lu, T., Zhu, J., Shangguan, Y., Chen, E., Gong, C., Zhao, Q., Jing, Y.,Zhao, Y., Li, Y., Cui, L., Fan, D., Lu, Y., Weng, Q., Wang, Y., Zhan, Q., Liu, K., Wei, X., An, K., An, G., and Han, B. (2016). OsSPL13 controlsgrain size in cultivated rice. Nat Genet 48, 447-456.


Song, X.J., Huang, W., Shi, M., Zhu, M.Z., and Lin, H.X. (2007). A QTL for rice grain width and weight encodes a previously unknownRING-type E3 ubiquitin ligase. Nat Genet 39, 623-630.


Su, L., Shan, J.X., Gao, J.P., and Lin, H.X. (2016). OsHAL3, a Blue Light-Responsive Protein, Interacts with the Floral Regulator Hd1 toActivate Flowering in Rice. Mol Plant 9, 233-244.


Wang, C., Wang, G., Zhang, C., Zhu, P., Dai, H., Yu, N., He, Z., Xu, L., and Wang, E. (2017). OsCERK1-Mediated Chitin Perception andImmune Signaling Requires Receptor-like Cytoplasmic Kinase 185 to Activate an MAPK Cascade in Rice. Mol Plant 10, 619-633.


Wang, H., Ngwenyama, N., Liu, Y., Walker, J.C., and Zhang, S. (2007). Stomatal development and patterning are regulated byenvironmentally responsive mitogen-activated protein kinases in Arabidopsis. Plant Cell 19, 63-73.


Warthmann, N., Chen, H., Ossowski, S., Weigel, D., and Herve, P. (2008). Highly specific gene silencing by artificial miRNAs in rice.PLoS One 3, e1829.


Widmann, C., Gibson, S., Jarpe, M.B., and Johnson, G.L. (1999). Mitogen-activated protein kinase: conservation of a three-kinasemodule from yeast to human. Physiol Rev 79, 143-180.


Wu, Y., Wang, Y., Mi, X.F., Shan, J.X., Li, X.M., Xu, J.L., and Lin, H.X. (2016). The QTL GNP1 Encodes GA20ox1, Which Increases GrainNumber and Yield by Increasing Cytokinin Activity in Rice Panicle Meristems. PLoS Genet 12, e1006386.


Xing, Y., and Zhang, Q. (2010). Genetic and molecular bases of rice yield. Annu Rev Plant Biol 61, 421-442.


Xu, J., and Zhang, S. (2015). Mitogen-activated protein kinase cascades in signaling plant growth and development. Trends Plant Sci20, 56-64.


Yang, K.Y., Liu, Y., and Zhang, S. (2001). Activation of a mitogen-activated protein kinase pathway is involved in disease resistance intobacco. Proc Natl Acad Sci U S A 98, 741-746.


Yi, J., Lee, Y.S., Lee, D.Y., Cho, M.H., Jeon, J.S., and An, G. (2016). OsMPK6 plays a critical role in cell differentiation during earlyembryogenesis in Oryza sativa. J Exp Bot 67, 2425-2437.


Yoo, J.H., Cheong, M.S., Park, C.Y., Moon, B.C., Kim, M.C., Kang, Y.H., Park, H.C., Choi, M.S., Lee, J.H., Jung, W.Y., Yoon, H.W., Chung,W.S., Lim, C.O., Lee, S.Y., and Cho, M.J. (2004). Regulation of the dual specificity protein phosphatase, DsPTP1, through interactionswith calmodulin. J Biol Chem 279, 848-858.


Zhang, D., and Yuan, Z. (2014). Molecular control of grass inflorescence development. Annu Rev Plant Biol 65, 553-578.


Zuo, J., and Li, J. (2014). Molecular genetic dissection of quantitative trait loci regulating rice grain size. Annu Rev Genet 48, 99-118.


0

20

40

60

80

100

120

Plan

t hei

ght (

cm)

FAZ1 gsn1

G

0

5

10

15

20Ti

ller n

umbe

r

FAZ1 gsn1

H

0

2

4

6

8

10

12

Gra

in le

ngth

(mm

) FAZ1 gsn1

I **

0

0.5

1

1.5

2

2.5

3

Gra

in w

idth

(mm

)

FAZ1 gsn1

J **

02468

10121416

Num

ber o

f prim

ary

b

ranc

hes

01020304050607080

Num

ber o

f sec

onda

ry

bra

nche

s

FAZ1 gsn1 FAZ1 gsn1

O P

**

**

FAZ1 gsn1 0

5

10

15

20

25

30

0

5

10

15

20

25

30

0

50

100

150

200

250

300

350

0

0.2

0.4

0.6

0.8

1

1000

-gra

in w

eigh

t (g)

Spik

elet

num

ber

Setti

ng p

erce

ntag

e

FAZ1 gsn1 FAZ1 gsn1 FAZ1 gsn1

K L M **

**

**

05

10152025303540

Yiel

d pe

r pla

nt (g

)

FAZ1 gsn1

N

**

FAZ1 gsn1

FAZ1 gsn1

FAZ1 gsn1

A B

C

D

E F

FAZ1 gsn1 FAZ1 gsn1

FAZ1 gsn1

Pani

cle

leng

th (c

m)

Q

Figure 1. Characterization of the gsn1 mutant, which exhibits increased grain size and reduced grain number per panicle. (A) Plant architecture of wild-type FAZ1 and mutant gsn1 plants at the reproductive stage. Scale bar, 10 cm. (B) Mature paddy rice grains from FAZ1 and gsn1. Scale bar, 1 mm. (C) Brown rice grains from FAZ1 and gsn1. Scale bar, 1 mm. (D) Mature panicles from FAZ1 and gsn1 rice plants. Scale bar, 5 cm. (E) Pistils of FAZ1 and gsn1 flowers. Scale bar, 1 mm. (F) Stamens of FAZ1 and gsn1 flowers. Scale bar, 2 mm. (G-Q) Comparisons between FAZ1 and gsn1 for average plant height (n=20 plants) (G), average tiller number (n=20 plants) (H), average grain length (n=20 plants) (I), average grain width (n=20 plants) (J), 1000-grain weight (n=30 plants) (K), average spikelet number per panicle (n=20 plants) (L), setting percentage (n=20 plants) (M), average yield per plant (n=20 plants) (N), average number of primary branches (n=20 plants) (O), average number of secondary branches (n=20 plants) (P), average panicle length (n=20 plants) (Q). Values (G-Q) are given as the mean ± SD. **P < 0.01 compared with the wild type using Student’s t-test.

Figure 2. Map-based cloning of GSN1 and genetic complementation. (A) The GSN1 locus was initially mapped to the short arm of chromosome 5 between the loci defined by markers G05220 and G05591 and then delimited to a 17.5-kb region between marker loci G05333 and G05340 that contained two ORFs. The numbers beneath the marker positions indicate the number of recombinants. Predicted protein structure of the first GSN1 transcript; the red box indicates the PTPc/DSPc domain and the green box indicates the GEL domain. The start codon (ATG) and the stop codon (TGA) are also indicated. A single nucleotide mutation from C to T in GSN1 results in a serine to phenylalanine change at S146 in the predicted PTPc/DSPc domain. (B) Amino acid sequence alignment of the partially conserved PTPc/DSPc domain in six monocot and six dicot species. Identical and similar residues are shown in colored boxes. The asterisk indicates the site of the S146F mutation in the PTPc/DSPc domain at the highly conserved serine residue at position 146 in gsn1. (C) Comparison of panicles between FAZ1, gsn1, and the T2 complementation transgenic lines gGSN1Com #1, and gGSN1Com #2 harboring the full-length GSN1 gene in the mutant gsn1 background. Scale bar, 5 cm. (D) Mature grains of FAZ1, gsn1, gGSN1Com #1, and gGSN1Com #2. Scale bar, 1 mm. (E) Mature grains of FAZ1 and the two GSN1RNAi transgenic lines. Scale bar, 1 mm. (F) Panicles of FAZ1 and the RNAi transgenic lines GSN1RNAi #1 and GSN1RNAi #2 in FAZ1. Scale bar, 5 cm. (G) Panicles of FAZ1 and the overexpression lines ProUBI:GSN1 #1 and ProUBI:GSN1 #2 in FAZ1. Scale bar, 5 cm. (H) Mature grains of FAZ1 and the T2 overexpression transgenic lines ProUBI:GSN1 #1 and ProUBI:GSN1 #2 in FAZ1. Scale bar, 1 mm. (I-J) Comparisons of average grain length (I) and spikelet number per panicle (J) between FAZ1 and the GSN1RNAi lines (n=10 plants). (K-L) Comparisons of average grain length (K) and spikelet number per panicle (L) between FAZ1 and the ProUBI:GSN1 lines (n=15 plants). Values (I-L) are given as the mean ± SD. **P < 0.01 compared with the wild type using Student’s t-test.

A

FAZ1 gsn1 gGSN1Com #1 #2

C

B

#1 #2

E

D

#1 #2

FAZ1 GSN1RNAi #1 #2

F

FAZ1 ProUBI::GSN1 #1 #2

G

ProUBI::GSN1 #1 #2

H

ProUBI::GSN1

0

2

4

6

8

10

Gra

in le

ngth

(mm

)

#1 #2

Spik

elet

num

ber

#1 #2 ProUBI::GSN1

0

100

200

300

400

**

**

**

**

0

2

4

6

8

10

Gra

in le

ngth

(mm

)

#1 #2 GSN1RNAi

Spik

elet

num

ber

**

**

I J

#1 #2 GSN1RNAi

0

100

200

300

**

**

K L

K

FAZ1 gsn1

0

250

500

750

1000

1250

1500

Area

of v

ascu

lar b

undl

e of

co

nnec

tivum

(μm

2 )

0

200

400

600

800

1000

Cell

num

ber

FAZ1 gsn1

E **

0

2

4

6

8

10

12

Cell

leng

th (μm

)

FAZ1 gsn1 FAZ1 gsn1 0

20

40

60

80

100

120

Cell

num

ber

FAZ1 gsn1

**

**

F

FAZ1 gsn1

FAZ1 gsn1

0

20

40

60

80

100Tu

berc

le b

i-pea

k di

stan

ce (μ

m)

FAZ1 gsn1

I

Epi

derm

al c

ell l

engt

h (μm

)

0

20

40

60

80

100

120

FAZ1 gsn1

J

A B C D

FAZ1

gsn1

FAZ1

gsn1

FAZ1

gsn1

FAZ1 gsn1

G

H

0

0.2

0.4

0.6

0.8

1

1.2

G1 S G2/M

FAZ1 gsn1

L

*

*

Cell

cycl

e ph

ase

dist

ribut

ion

of

youn

ger p

anic

le c

ells

Cell

num

ber (

coun

ts)

DNA content (FL6-Area)

Cell

num

ber (

coun

ts)

DNA content (FL6-Area)

2C

4C

2C

4C

Figure 3. GSN1 influences cell proliferation rather than cell elongation to control organ size. (A) Spikelet hulls just before heading in FAZ1 and gsn1. The dotted line indicates the position of the cross-section shown in B. Scale bar, 1 mm. (B) Cross-sections of FAZ1 and gsn1 spikelet hulls. Scale bar, 300 μm. (C) Magnified views of the spikelet hull cross-sections boxed in B. Arrows show the outer parenchymal cell layer. Scale bar, 50 μm. (D) Cross-sections of FAZ1 and gsn1 anthers. Arrows show the vascular bundle cells of the connectivum. Scale bar, 20 μm. (E) Comparisons of cell numbers and average cell length in the outer parenchymal cell layers of FAZ1 and gsn1 spikelet hulls (n=10 views).

(F) Comparisons of the average area of the vascular bundles in the connectivum and cell numbers in the connectivum in FAZ1 and gsn1 (n=10 views). (G-H) Scanning electron micrographs of the outer (G) and inner (H) surfaces of FAZ1 and gsn1 spikelet hulls. Scale bars, 100 μm and 50 μm, respectively. (I) Comparison of average tubercle bi-peak distance on the outer surfaces of the spikelet hulls between FAZ1 and gsn1 (n=10 grains). (J) Comparison of inner epidermal cell length in FAZ1 and gsn1 spikelet hulls (n=10 grains). (K) Flow cytometry analysis of young panicle (1-3 cm) cells in FAZ1 and gsn1. The tall and short peaks show 2c and 4c nuclei, respectively. (L) Percentage comparison of the distribution of cells in different phases of the cell cycle in young spikelet cells (n=10 pooled tissues, and three younger panicles from different plants per pool). The G1, S and G2/M phases are shown in colored boxes. All numerical values are given as the mean ± SD. *P < 0.05; **P < 0.01 compared with the wild type using Student’s t-test.

0

2

4

6

8

10

12

14

16

18F

Rela

tive

expr

essi

on le

vel

of G

SN1

G

Rela

tive

expr

essi

on le

vel

of G

SN1

0

0.5

1

1.5

2

2.5

3

3.5

0

0.2

0.4

0.6

0.8

1

1.2

OM

FP A

477

nm

** B

0

0.02

0.04

0.06

0.08

0.1

PNPP

A40

5 nm

A

E

Pro35S:GSN1-YFP

DIC Merged

chloroplast

Pro35S:GSN1-GFP

Pro35S:GFP

D

Figure 4. GSN1 encodes a dual-specificity phosphatase with dephosphorylation activity that localizes to the cytoplasm. (A-B) GSN1 specifically dephosphorylates the bulky polycyclic aryl phosphate OMFP but not PNPP. The enzymatic activity of GSN1 from FAZ1 and gsn1 was examined using MBP fusion proteins expressed in E. coli using the substrates PNPP (A) and OMFP (B). Dephosphorylated OMFP and PNPP levels were determined based on absorbance at 477 nm and 405 nm after the dephosphorylation reactions were complete (n=10 independent dephosphorylation reactions). MBP protein was used as the negative control. Values are given as the mean ± SD. **P < 0.01 compared with the wild type and the negative control using Student’s t-test. (C) Enzymatic characterization of GSN1 activity. The kinetic parameters Km and Vmax for GSN1 using OMFP as a substrate were calculated using GraphPad Prism 5 software. (D) Subcellular localization of GSN1 in leaf epidermal cells of Nicotiana benthamiana. The Pro35S:GFP construct was used as the positive control and was distributed ubiquitously. Scale bar, 20 μm. (E) GSN1 protein was shown to target the cytoplasm by transient expression analysis of GSN1-YFP in rice protoplasts. DIC, differential interference contrast. Scale bar, 2 μm. (F-G) Relative expression levels of GSN1 in various organs in the vegetative (F) and reproductive phase (G) determined by qRT-PCR (n=3 pooled tissues, three plants per pool). Values are given as the mean ± SD. The UBQ5 gene was used as an internal reference to normalize gene expression data.

** C

pGBKT7-GSN1+pGADT7

pGBKT7-53+pGADT7-T

pGBKT7-Lam+pGADT7-T

pGBKT7+pGADT7-GSN1

pGBKT7-OsMPK6+pGADT7

pGBKT7+pGADT7-OsMPK6

pGBKT7-GSN1+pGADT7-OsMPK6

pGBKT7-OsMPK6+pGADT7-GSN1

1 10-1 10-2 1 10-1 10-2

SD/-Trp-Leu-His SD/-Trp-Leu-His-Ade A

nYFP-GSN1+cYFP-OsMPK6

nYFP-OsMPK6+cYFP-GSN1

nYFP-GSN1+cYFP-OsHAL3

B

YFP DIC Merged

Figure 5. GSN1 interacts with and inactivates OsMPK6 via specific dephosphorylation. (A) Yeast two-hybrid assays indicate that GSN1 interacts with OsMPK6. The yeast cells were cultured on SD/-Trp-Leu-His (medium without tryptophan, leucine, and histidine) or SD/-Trp-Leu-His-Ade (medium without tryptophan, leucine, histidine, and adenine) containing X-ɑ-gal. pGBKT7 and pGADT7 are the bait and prey vectors, respectively. (B) GSN1 associates with OsMPK6, as shown by BiFC assays in Nicotiana benthamiana leaf cells. GSN1 and OsMPK6 were both fused to the N-terminal fragment of YFP (nYFP) and the C-terminal fragment of YFP (cYFP), nYFP-GSN1 and cYFP-OsMPK6 or cYFP-GSN1 and nYFP-OsMPK6 were then co-expressed in Nicotiana benthamiana leaves. The nYFP-GSN1 and cYFP-OsHAL3 fusion proteins were used as negative controls. (C) Co-immunoprecipitation assays indicate that GSN1 interacts with OsMPK6 in planta. Pro35S:GSN1-Flag and Pro35S:OsMPK6-Myc fusions were co-expressed in Nicotiana benthamiana leaves. Proteins were extracted (Input) and immunoprecipitated (IP) with Flag beads. The immunoblot (WB for western blot) assays were performed using anti-Flag and anti-Myc antibodies.

C D

(D) GSN1 dephosphorylates OsMPK6 in vitro. The MBP-GSN1, MBP-GSN1S146F, MBP-OsMKK4CA, and GST-OsMPK6 proteins were expressed in E. coli and purified using MBP and GST beads, respectively. The in vitro phosphorylation and dephosphorylation reactions were performed using the purified proteins. OsMPK6 phosphorylation was detected using the polyclonal antibody phospho-p44/42. The loading control was blotted using the anti-AtMPK6 antibody against Arabidopsis MPK6.

A B

D

F

02468

101214

02468

101214

0

2

4

6

8

10

12

14

Gra

in le

ngth

(mm

)

G

Gra

in le

ngth

(mm

)

Gra

in le

ngth

(mm

)

H I **

** **

**

**

**

e E

C

Figure 6. GSN1 is a negative regulator of the MAPK cascade signaling pathway in rice. (A-B) Panicle (A) and grain (B) morphology in FAZ1, gsn1, OsMPK6RNAi, and gsn1/OsMPK6RNAi double mutant plants. Scale bars, 5 cm and 1 mm, respectively. (C-D) Panicle (C) and grain (D) morphology in FAZ1, gsn1, OsMKK4CRISPR, and the gsn1/OsMKK4CRISPR double mutant. Scale bars, 5 cm and 1 mm, respectively. (E-F) Panicle (E) and grain (F) morphology in FAZ1, gsn1, OsMKKK10CRISPR, and the gsn1/OsMKKK10CRISPR double mutant. Scale bars, 5 cm and 1 mm, respectively. (G) Comparison of average grain length in FAZ1, gsn1, OsMPK6RNAi, and the gsn1/OsMPK6RNAi double mutant (n=15 plants). (H) Comparison of average grain length in FAZ1, gsn1, OsMKK4CRISPR, and the gsn1/OsMKK4CRISPR double mutant (n=15 plants). (I) Comparison of average grain length in FAZ1, gsn1, OsMKKK10CRISPR, and the gsn1/OsMKKK10CRISPR double mutant (n=15 plants). Values (G-I) are given as the mean ± SD. **P < 0.01 compared with FAZ1 and gsn1 using Student’s t-test.

GSN1 GSN1S146F

GSN1 gsn1 A B

Figure 7. Proposed working model of the role of the GSN1-MAPK module in determining grain number per panicle and grain size in rice. (A-B) In young panicles at a specific developmental stage, after sensing panicle cell-specific communication signaling molecules, unknown receptor-like protein kinases activate the OsMKKK10-OsMKK4-OsMPK6 cascade through mediators or scaffold proteins. This cascade sequentially phosphorylates downstream, unidentified substrates that determine cell differentiation in the panicle primordia or cell proliferation in the spikelet. (A) Spatiotemporally activated GSN1 protein acts as a molecular brake and precisely dephosphorylates OsMPK6 and then inactivates the OsMKKK10-OsMKK4-OsMPK6 cascade, which represses the phosphorylation of downstream substrates. (B) The inactivation of GSN1 (GSN1S146F) results in direct, durative activation of the OsMKKK10-OsMKK4-OsMPK6 cascade and excessive amplification of downstream signaling, which attenuates cell differentiation in the panicle primordia but increases cell proliferation in the spikelet. Hence, the signaling output of the integrated GSN1-MAPK module coordinates the trade-off between grain number per panicle and grain size, ultimately contributing to final panicle architecture in rice. Arrows with solid lines show associations that are supported by genetic or biochemical evidence. Arrows with dashed lines show the presumptive associations. Double lines represent durative activation signaling. Question marks indicate unidentified or unknown signaling components.

DOI 10.1105/tpc.17.00959; originally published online March 27, 2018;Plant Cell

Hong-Xuan LinTao Guo, Ke Chen, Nai-Qian Dong, Chuan-Lin Shi, Wang-Wei Ye, Ji-Ping Gao, Jun-Xiang Shan and

to Coordinate the Trade-off between Grain Number per Panicle and Grain Size in RiceGRAIN SIZE AND NUMBER1 Negatively Regulates the OsMKKK10-OsMKK4-OsMPK6 Cascade

This information is current as of March 2, 2020

Supplemental Data /content/suppl/2018/03/27/tpc.17.00959.DC1.html

Permissions https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X

eTOCs http://www.plantcell.org/cgi/alerts/ctmain

Sign up for eTOCs at:

CiteTrack Alerts http://www.plantcell.org/cgi/alerts/ctmain

Sign up for CiteTrack Alerts at:

Subscription Information http://www.aspb.org/publications/subscriptions.cfm

is available at:Plant Physiology and The Plant CellSubscription Information for

ADVANCING THE SCIENCE OF PLANT BIOLOGY © American Society of Plant Biologists

https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X

http://www.plantcell.org/cgi/alerts/ctmain

http://www.plantcell.org/cgi/alerts/ctmain

http://www.aspb.org/publications/subscriptions.cfm

grain size and number1 negatively regulates the …56 dephosphorylation of osmpk6 to coordinate the...

Documents