1

RESEARCH ARTICLE 12

Genomic Imprinting was Evolutionarily Conserved during Wheat 3Polyploidization 4

5Guanghui Yang1,3, Zhenshan Liu2,3, Lulu Gao1, Kuohai Yu1, Man Feng1, Yingyin Yao1, 6Huiru Peng1, Zhaorong Hu1, Qixin Sun1, Zhongfu Ni1, and Mingming Xin1* 7

81 State Key Laboratory for Agrobiotechnology, Key Laboratory of Crop Heterosis Utilization 9(MOE), Beijing Key Laboratory of Crop Genetic Improvement, China Agricultural University, 10Beijing, 100193, China. 112 College of Agronomy, Northwest A&F University, Yangling 712100, Shaanxi, China. 123 These authors contributed equally to this work 13*Corresponding author: Mingming Xin, mingmingxin@cau.edu.cn, Tel: 010-6273145214

15Short title: Genomic imprinting in wheat 16

17One-sentence summary: Genes controlled by imprinting were evolutionarily conserved during 18wheat polyploidization. 19

20The author responsible for distribution of materials integral to the findings presented in this 21article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) 22is: Mingming Xin (mingmingxin@cau.edu.cn) 23

24ABSTRACT 25

Genomic imprinting is an epigenetic phenomenon that causes genes to be differentially expressed 26depending on their parent-of-origin. To evaluate the evolutionary conservation of genomic 27imprinting and the effects of ploidy on this process, we investigated parent-of-origin-specific gene 28expression patterns in the endosperm of diploid (Aegilops spp.), tetraploid, and hexaploid wheat 29(Triticum spp.) at various stages of development via high-throughput transcriptome sequencing. 30We identified 91, 135, and 146 maternally or paternally expressed genes (MEGs or PEGs, 31respectively) in diploid, tetraploid, and hexaploid wheat, respectively, 52.7% of which exhibited 32dynamic expression patterns at different developmental stages. Gene ontology enrichment 33analysis suggested that MEGs and PEGs were involved in metabolic processes and 34DNA-dependent transcription, respectively. Nearly half of the imprinted genes exhibited 35conserved expression patterns during wheat hexaploidization. In addition, forty percent of the 36homeolog pairs originating from whole genome duplication were consistently maternally or 37paternally biased in the different subgenomes of hexaploid wheat. Furthermore, imprinted 38expression was found for 41.2% and 50.0% of homolog pairs that evolved by tandem duplication 39after genome duplication in tetraploid and hexaploid wheat, respectively. These results suggest 40that genomic imprinting was evolutionarily conserved between closely related Triticum and 41Aegilops species, and in the face of polyploid hybridization between species in these genera. 42

Plant Cell Advance Publication. Published on January 3, 2018, doi:10.1105/tpc.17.00837

©2018 American Society of Plant Biologists. All Rights Reserved

2

INTRODUCTION 43

During double fertilization, a phenomenon unique to flowering plants, the egg cell (1n) 44

and central cell (2n) fuse with two sperm cells (1n) to generate the diploid embryo (2n) 45

and the triploid endosperm (3n), respectively. The resulting endosperm, a functional 46

analog of the placenta in mammals, facilitates embryogenesis and supports seedling 47

growth by providing the embryo with nutrients. The endosperm tissue also interacts 48

dynamically with the embryo over the course of development by activating important 49

signaling pathways that are required for embryo development (Yang et al., 2008; Fouquet 50

et al., 2011; Costa et al., 2014; Xu et al., 2014). Therefore, proper endosperm 51

development is essential for coordinating embryo and seed growth. 52

Genomic imprinting, which refers to monoallelic gene expression in a 53

parent-of-origin-dependent manner, generally involves epigenetic regulation. In plants, 54

imprinting primarily occurs in the endosperm; however, recent studies have shown that a 55

portion of genes are also imprinted in the embryo of Arabidopsis thaliana (Raissig et al., 56

2013), rice (Oryza sativa) (Luo et al., 2011), and maize (Zea mays) (Meng et al., 2017). 57

This uni-parental transcription pattern indicates that, to some extent, parental genomes 58

might not contribute equally to the filial genome, at least for some specific loci, if not at 59

the genome-wide level (Vielle-Calzada et al., 2000; Grimanelli et al., 2005; Autran et al., 60

2011). 61

Two major hypotheses have been proposed to explain the extensive occurrence and 62

convergent evolution of genomic imprinting across flowering plants and mammals. One 63

hypothesis, the parental conflict theory, argues that paternally-derived alleles promote the 64

transport of resources from maternal tissue to the offspring to improve their fitness, 65

whereas maternally-derived alleles tend to share resources equally to balance nutrient 66

allocation among embryos (Haig and Westoby, 1989; Wilkins and Haig, 2003). The other 67

hypothesis, the maternal-offspring coadaptation model, proposes adaptive integration 68

rather than a struggle for resources between the maternal tissue and the offspring (Curley 69

et al., 2004; Wolf and Hager, 2006; Swaney et al., 2007; Keverne and Curley, 2008). 70

However, the biological relevance of the parent-of-origin expression pattern of a gene 71

remains a matter of considerable debate, and the potential effects of allele-specific 72

expression on the embryo and seedling remain ambiguous, since genomic imprinting 73

3

mainly occurs in the terminal tissue of the endosperm, which does not genetically 74

contribute to the next generation. 75

The biological implications of genomic imprinting can be inferred from the results of 76

reciprocal crosses of plants with different ploidy levels, thus providing different dosages 77

of the parental genomes. Specifically, paternal-excess crosses strongly promote seed 78

development, resulting in the production of big seeds, whereas maternal-excess crosses 79

dramatically inhibit endosperm growth, resulting in the production of small seeds (Lin, 80

1984; Scott et al., 1998). These findings indicate that the correct balance between 81

maternally- and paternally-derived genomes is responsible for proper embryo and 82

endosperm development. Furthermore, Arabidopsis plants with loss-of-function of the 83

maternally inherited alleles of the imprinted genes MEDEA (MEA) and FERTILIZATION 84

INDEPENDENT SEED2 (FIS2), which encode components of the Polycomb Repressive 85

Complex 2 (PRC2), exhibit over-proliferation of endosperm after fertilization. This 86

finding suggests that MEA and FIS2 restrain seed growth (Chaudhury et al., 1997; 87

Grossniklaus et al., 1998; Kiyosue et al., 1999; Luo et al., 1999). However, other studies 88

have argued that the expression level rather than the imprinting pattern of a gene is likely 89

indispensable for phenotypic variation, as plants with loss-of-function of MULTICOPY 90

SUPPRESSOR OF IRA1 (MSI1), which encodes a non-imprinted subunit of PRC2, 91

exhibit the same phenotypes as those with mutations in MEA and FIS2 (Kohler et al., 92

2003; Guitton and Berger, 2005; Leroy et al., 2007). In addition, emerging evidence 93

suggests that paternally expressed genes are involved in establishing postzygotic 94

hybridization barriers in Arabidopsis, as downregulating the expression of the paternally 95

imprinted genes ADMETOS (ADM), SU(VAR)3-9, HOMOLOG 7 (SUVH7), 96

PATERNALLY EXPRESSED IMPRINTED GENE 2 (PEG2), and PEG9 can partially 97

rescue triploid seed development (Kradolfer et al., 2013; Wolff et al., 2015). In maize, the 98

maternal expression of Meg1 in basal endosperm transfer cells was directly shown to be 99

functionally relevant for seed development and growth. This study not only verified the 100

necessity and sufficiency of Meg1 in regulating transfer cell differentiation, but also 101

demonstrated the importance of imprinted gene expression in controlling seed size, as 102

revealed through the development of transgenic lines with reduced expression, ectopic 103

expression, and non-imprinted expression of Meg1 (Costa et al., 2014). 104

4

RNA-sequencing (RNA-seq) analyses have identified hundreds of imprinted genes in 105

Arabidopsis, maize, rice, castor bean (Ricinus communis), and sorghum (Sorghum bicolor; 106

Gehring et al., 2011; Hsieh et al., 2011; Luo et al., 2011; Waters et al., 2011; Wolff et al., 107

2011; Zhang et al., 2011; Waters et al., 2013; Xin et al., 2013; Pignatta et al., 2014; Xu et 108

al., 2014; Zhang et al., 2016). However, the amount of overlap among imprinted genes of 109

various plant species is limited (Waters et al., 2013; Pires and Grossniklaus, 2014;110

Hatorangan et al., 2016). For example, only 14% of MEGs and 29% of PEGs in C. 111

rubella were commonly imprinted in Arabidopsis (Hatorangan et al., 2016). Subsequent 112

study indicated that genes controlled by imprinting are highly conserved between A. 113

lyrata and A. thaliana (Klosinska et al., 2016). In addition, the consistently imprinted 114

expression of two paralogous maize genes, Fertilization-independent endosperm 1 (Fie1) 115

and Fie2, suggests that parent-of-origin-dependent allelic expression can be maintained 116

during tetraploidization or gene duplication events (Danilevskaya et al., 2003). 117

Hexaploid wheat (AABBDD, Triticum aestivum) is a typical allopolyploid species 118

with three distinct subgenomes that has undergone two separate allopolyploidization 119

events. The first event involved a cross between Triticum urartu (AA genome) and an 120

unidentified species (BB genome) 0.36 to 0.50 million years ago (MYA) (Dvořák, 1976; 121

Huang et al., 2002; Dvorak and Akhunov, 2005; Pont and Salse, 2017). The resulting 122

tetraploid wheat species, Triticum turgidum (AABB), then hybridized with Aegilops 123

tauschii (DD genome) to generate hexaploid wheat (AABBDD) ~10,000 years ago 124

(Kihara, 1944; McFadden and Sears, 1946; Dvorak et al., 1998; Huang et al., 2002). 125

These wheat species provide an excellent model system for studying the genomics of 126

polyploid plants. 127

In the current study, we performed genome-wide identification of imprinted genes in 128

wheat species with different ploidy levels using reciprocal endosperm at different 129

developmental stages. We then analyzed the conservation of genomic imprinting among 130

diploid, tetraploid, and hexaploid wheat species, as well as their homeologous genes. Our 131

findings demonstrate that parent-of-origin-dependent allelic expression was 132

evolutionarily conserved during wheat polyploidization. 133

5

RESULTS 134

Transcriptome Sequencing, Data Processing, and SNP Calling 135

To assess the allelic expression patterns of genes in diploid (DD), tetraploid (AABB), and 136

hexaploid wheat (AABBDD) endosperm, we performed deep transcriptome sequencing 137

of reciprocally crossed developing endosperm from wheat species with different ploidy 138

levels, including diploid wheat (DD; Y177 [Y]×RM220 [R] and R×Y at 15 and 20 days 139

after pollination, DAP), tetraploid wheat (AABB; Jinying8 [J]×SCAUP [S] and S×J at 15 140

and 20 DAP), and hexaploid wheat (AABBDD; Doumai [D] and Keyi5214 [K] and K×D 141

at 15, 20, and 25 DAP), together with their respective parental lines. In total, we obtained 142

3539.3 million paired-end reads, with an average of ~111.1, ~149.2, and ~242.2 million 143

reads per sample which, on average, covered 14,550, 27,983, and 40,489 144

endosperm-expressed genes (FPKM > 1) in diploid, tetraploid, and hexaploid wheat, 145

respectively. To identify high-quality single-nucleotide polymorphisms (SNPs) between 146

the parental lines, we mapped the corresponding RNA-seq reads of the parental lines to 147

the wheat reference genome (TGACv1 [Chinese Spring, CS]) using Bowtie 2 (v2.2.9; 148

Langmead and Salzberg, 2012); ~33.7% (diploid wheat, 31.5–35.8% mapped to the 149

CS_D genome), ~50.0% (tetraploid wheat, 46.3–53.5% mapped to the CS_A and B 150

genome), and ~49.9% (hexaploid wheat, 42.0–59.5% mapped to the CS_A, B, and D 151

genome) of uniquely mapped reads were retained for subsequent analysis (Supplemental 152

Table 1). We then performed SNP calling using Samtools (v1.4; Li et al., 2009) and 153

BCFtools (v1.4; Li, 2011). 154

The accuracy of SNP identification will be reduced in polyploid wheat due to the 155

widespread presence of homeologs, which might cause ambiguous mapping. Therefore, 156

to improve the reliability of the SNPs in polyploid wheat, we only considered RNA-seq 157

reads that were uniquely mapped to the A, B, or D subgenome under the condition that 158

reference sequence information was available for all three homeologous loci (see 159

Methods). Ultimately, we identified 7,109 (mapped to the CS_D genome), 14,995 160

(mapped to the CS_A and B genome), and 13,085 (mapped to the CS_A, B, and D 161

genome) high-confidence SNPs located in 3,485, 4,612, and 5,063 genes in diploid, 162

tetraploid, and hexaploid wheat species, respectively. 163

Next, we aligned the uniquely mapped RNA-seq reads from reciprocal crosses to the 164

6

reference sequence with SNP information to distinguish their parental origin. To 165

determine whether the ratio of allele-specific reads properly reflected the ratio of 166

maternal-to-paternal transcripts, we calculated the correlation coefficient between each 167

pair (~0.91 on average) after excluding genes with fewer than 10 reads. This analysis 168

indicated that the allelic expression patterns inferred from parent-specific reads were 169

proportional to total gene expression. By plotting paternal vs. maternal expression for all 170

retained genes, we found that the majority of SNP-containing genes exhibited the 171

expected ratio of 2m:1p, whereas approximately 9.5–31.1% exhibited allele-biased 172

expression patterns, i.e., the ratio of maternally- to paternally-derived reads significantly 173

deviated from 2:1 in both reciprocal crosses (𝜒2 test, FDR-adjusted p value ≤ 0.05). 174

175

Genome-wide Survey of Imprinted Genes in Hybrid Endosperm from Diploid, 176

Tetraploid, and Hexaploid Wheat Species 177

To identify high-confidence imprinted genes in wheat species with different ploidy levels, 178

we performed a 𝜒2 test to determine whether these selected genes possessed 179

parent-specific expression patterns in both reciprocal crosses. At a significance level of p 180

value = 0.05, 434 genes (mapped to the CS_D genome) were determined to have 181

maternally or paternally preferred expression patterns in reciprocally crossed diploid 182

endosperm during at least one stage of development. Correspondingly, 1,307 (mapped to 183

the CS_A and B genome) and 1,630 (mapped to the CS_A, B, and D genome) genes had 184

either maternally or paternally preferred expression patterns in tetraploid and hexaploid 185

wheat, respectively (Figure 1 and Supplemental Figure 1). To determine the 186

parent-of-origin expression status of allele-specific genes, we further filtered the 187

candidate imprinted genes, with maternally and paternally expressed genes (MEGs and 188

PEGs, respectively) defined as genes that had 90% maternal reads or 70% paternal reads 189

among all SNP-associated reads in both reciprocal crosses (with a minimum of 10 190

SNP-associated reads per cross and a FDR-adjusted p value of 0.05), respectively. We 191

calculated the proportion of maternal/paternal reads separately for each reciprocal cross 192

based on the criterion that the values for both crosses had to be above the threshold (see 193

Methods). Using this more stringent, ratio-based criterion, we finally identified 372 194

imprinted genes (Figure 2), with 91 (62 MEGs and 29 PEGs), 135 (90 MEGs and 45 195

7

PEGs), and 146 (94 MEGs and 52 PEGs) in diploid, tetraploid, and hexaploid wheat 196

species, respectively (Figure 2A, 2B, Supplemental Data Set 1 and 2). Of these 372 197

imprinted genes, 176 genes (47.3%) exhibited consistent imprinted expression patterns 198

across all developmental stages (40 MEGs and 14 PEGs for diploid wheat, 50 MEGs and 199

21 PEGs for tetraploid wheat, and 35 MEGs and 16 PEGs for hexaploid wheat). The 200

remaining genes were considered to be imprinted in a stage-specific manner (30 MEGs 201

and 7 PEGs for diploid wheat, 39 MEGs and 25 PEGs for tetraploid wheat, and 59 MEGs 202

and 36 PEGs for hexaploid wheat) (Figure 2B, Supplemental Data Set 1), indicating 203

that a portion of imprinted wheat genes exhibited dynamic expression patterns during 204

endosperm development. Further investigation revealed that all of these stage-specific 205

imprinted genes were also expressed during other developmental stages, but their parental 206

alleles exhibited biallelic expression patterns. Thus, the major cause of stage-specific 207

imprinting is not a lack of expression, but can instead be attributed to biallelic expression 208

patterns. 209

Next, we investigated whether the imprinted genes exhibit tissue-specific expression 210

patterns in wheat (Figure 3). By examining previously published datasets 211

(http://www.plexdb.org/plex.php?database=Wheat) (Schreiber et al., 2009), we found 212

that both PEGs and MEGs were more abundantly expressed in endosperm than in other 213

tissues, and MEGs appeared to exhibit more endosperm-specific expression compared 214

with PEGs (the fold-change of gene expression [endosperm vs. other tissues] was 4.14 215

and 2.58 for MEGs and PEGs, respectively) (Figure 3A). In addition, MEGs and PEGs 216

were expressed at higher levels in developing endosperm than non-imprinted genes 217

(Figure 3B). An analysis of previously published laser capture microdissection data 218

revealed that MEGs were more highly expressed in all endosperm compartments 219

compared to PEGs (Pfeifer et al., 2014) (Supplemental Figure 2). Gene ontology (GO) 220

analysis showed that MEGs were enriched in the categories “regulation of nitrogen 221

compound metabolic process” (GO:0051171), “regulation of macromolecule biosynthetic 222

process” (GO:0010556), and “regulation of primary metabolic process” (GO:0080090). 223

By contrast, the significantly enriched GO categories for PEGs were “RNA biosynthetic 224

process” (GO:0032774) and “transcription, DNA-dependent” (GO:0006351). The high 225

expression levels of MEGs, together with their roles in regulating nutrient biosynthesis 226

8

and metabolism based on GO analysis, suggest that they play important roles in 227

regulating nutrient accumulation in wheat endosperm. This notion is consistent with the 228

finding that MEGs are rapidly upregulated in maize endosperm at the filling stage (Xin et 229

al., 2013), which partially supports the parental conflict theory, i.e., that MEGs tend to 230

control the growth of their offspring by limiting nutrient allocation to their offspring 231

(Haig and Westoby, 1989; Wilkins and Haig, 2003). 232

233

Experimental Validation of Imprinted Genes in Diploid, Tetraploid, and Hexaploid 234

Wheat Species 235

To validate the bioinformatically identified imprinted genes, we performed RT-PCR 236

followed by cleaved amplified polymorphic sequence (CAPS) assays or sequencing. We 237

examined ten diploid imprinted candidate genes (six MEGs and four PEGs), among 238

which six were sequenced, seven were cleaved with SNP-sensitive restriction enzymes, 239

and three (TRIAE_CS42_5DS_TGACv1_457221_AA1483900,240

TRIAE_CS42_1DL_TGACv1_062392_AA0213710, and 241

TRIAE_CS42_3DS_TGACv1_272480_AA0921380) were confirmed by both methods 242

(Supplemental Figure 3 and 4). Consistent with the RNA-seq data, all ten putative 243

imprinted genes showed the expected parent-of-origin expression patterns, which exactly 244

coincided with the imprinting predictions at different developmental stages. For example, 245

TRIAE_CS42_7DL_TGACv1_602576_AA1961380 was predicted to be a PEG at 15 DAP 246

and 20 DAP in diploid wheat, although with a few maternal reads detected in its 20-DAP 247

endosperm of the Y×R cross, and CAPS analysis confirmed these expression patterns at 248

both developmental stages (Supplemental Figure 3). In addition, although 249

TRIAE_CS42_3DS_TGACv1_272480_AA0921380 was identified as a PEG according to 250

our ratio-based criteria, a few maternal reads appeared in the Y×R cross at 15 and 20 251

DAP but not in the reciprocal cross; both the CAPS assay and sequencing confirmed our 252

observation (Supplemental Figure 3 and 4, Supplemental Data Set 1). 253

Although it is more difficult to experimentally confirm imprinted genes in tetraploid 254

and hexaploid wheat than in diploid wheat due to the presence of homeologous genes, but 255

we overcame this challenge by using homeolog-specific primer pairs. In tetraploid 256

reciprocal crosses, six putative imprinted genes (five MEGs and one PEG) showing 257

9

parent-of-origin expression status were confirmed via CAPS or sequencing 258

(Supplemental Figure 3 and 4). Of these six genes, 259

TRIAE_CS42_4BS_TGACv1_329466_AA1101720 was predicted to be maternally 260

expressed only in 15-DAP endosperm based on a 90% ratio cutoff, and consistently, 261

paternally-derived reads were abundant at 20 DAP according to sequencing validation 262

(Supplemental Figure 4 and Supplemental Data Set 1). For hexaploid wheat, five 263

candidate imprinted genes were verified in 15-, 20-, and 25-DAP reciprocally crossed 264

endosperm, including four MEGs and one PEG. These five candidates were considered to 265

be consistently imprinted at 15, 20, and 25 DAP based on the RNA-seq data, and the 266

CAPS experiment revealed a clear maternally or paternally imprinted pattern at all three 267

developmental stages in both reciprocal crosses. Together, the experimental validation of 268

imprinted genes in diploid, tetraploid, and hexaploid wheat confirmed the efficiency of 269

our strategy for identifying genomic imprinting in polyploid plants. We also performed 270

sequencing analysis of two other genes 271

(TRIAE_CS42_1DL_TGACv1_061444_AA0195450 and 272

TRIAE_CS42_5AS_TGACv1_393185_AA1269500) that showed imprinted expression 273

patterns only in one biological replicate due to limited read counts in the other replicate. 274

The maternal or paternal expression patterns of these two genes were clearly confirmed 275

in developing wheat endosperm (Supplemental Figure 5), indicating that a subset of 276

imprinted genes were not identified in wheat endosperm at the current sequencing depth. 277

278

Imprinted Wheat Genes were Evolutionarily Conserved during Polyploidization 279

Genes controlled by genomic imprinting are poorly conserved between Arabidopsis and 280

monocots as well as between rice and maize, possibly because this biased expression 281

pattern is partially dependent on the presence of transposable elements (Luo et al., 2011; 282

Waters et al., 2011; Waters et al., 2013; Rodrigues and Zilberman, 2015). The paternally 283

expressed auxin biosynthesis-related genes YUCCAs and Tryptophan aminotransferase 284

related 1 (TAR1) are rare examples of conserved imprinted genes present in rice, maize, 285

and Arabidopsis endosperm (Hsieh et al., 2011; Luo et al., 2011; Zhang et al., 2011; 286

Chen et al., 2017). In addition, only 14% of MEGs and 29% of PEGs in C. rubella were 287

commonly imprinted in Arabidopsis (Hatorangan et al., 2016). However, 50% of 288

10

imprinted genes were subsequently found to be conserved between A. lyrata and A. 289

thaliana, which diverged approximately 13 MYA (Klosinska et al., 2016). The paralogs 290

of 10 imprinted genes (resulting from the recent whole genome duplication) also exhibit 291

parent-of-origin-dependent allelic expression patterns in maize endosperm (Waters et al., 292

2013), which prompted us to investigate the conservation of imprinted genes during the 293

polyploidization of wheat. 294

Hexaploid wheat has undergone two separate allopolyploidization events and has 295

arisen from the convergence of three diploid ancestors (AA, BB, and DD). To investigate 296

the conservation of imprinted genes during wheat evolutionary history, we only 297

considered genes with high sequence identity (> 90%, E-value < 1e-10) and syntenic 298

chromosome regions among different wheat species. For example, if gene X in position 299

Y of the B subgenome in hexaploid wheat was imprinted, we only examined the 300

expression patterns of its homolog in position Y of the B subgenome in tetraploid wheat 301

and not its homolog in position Y of the A subgenome of tetraploid wheat or the D 302

subgenome of diploid wheat, and vice versa. We examined the imprinted expression 303

patterns of individual MEGs and PEGs in wheat species with different ploidy levels 304

(Figure 4). Of the 135 imprinted genes (59 in the A subgenome and 76 in the B 305

subgenome) in tetraploid reciprocal crosses (AABB), 27 genes (13 in the A subgenome 306

and 14 in the B subgenome) were found to possess SNPs between parental lines of 307

hexaploid wheat. Among these, 15 genes (55.6%; 4 in the A subgenome and 11 in the B 308

subgenome) showed conserved parent-of-origin expression patterns based on our criteria, 309

including 10 MEGs and 5 PEGs (Figure 4A, Supplemental Data Set 3). Similarly, of 310

the 91 imprinted genes in diploid reciprocal crosses, eight candidates happened to possess 311

SNPs between D and K, and three (37.5%) genes were also imprinted in hexaploid wheat 312

(Supplemental Data Set 3). No overlap of imprinted genes between diploid wheat (DD) 313

and tetraploid wheat (AABB) was observed, since these species have different 314

subgenomes. Thus, in total, 18 genes (18/35, 51.4%) exhibited conserved 315

parent-of-origin-dependent expression patterns between diploid and hexaploid wheat or 316

between tetraploid and hexaploid wheat when considering both sequence identity and 317

syntenic chromosome regions. Interestingly, at a significance level of FDR-adjusted 318

p=0.05, 62.9% (22/35) of genes showed conserved maternally or paternally preferred 319

11

expression patterns during wheat hexaploidization (Figure 4A). For example, 320

TRIAE_CS42_2BS_TGACv1_147693_AA0486590 met the requirement of 70% paternal 321

reads for PEGs in both reciprocal crosses of hexaploid wheat, but it was excluded from 322

the imprinted gene sets due to the limited number of SNP-associated reads, whereas 323

TRIAE_CS42_6AS_TGACv1_487181_AA1568730 exhibited significant maternally biased 324

expression at 15, 20, and 25 DAP in the reciprocal crosses of hexaploid wheat based on 325

the 𝜒2 test (p value < 0.05), but it did not pass the cutoff criterion of 90% maternal reads 326

in one cross. 327

Polyploid wheat has experienced one or two rounds of allopolyploidization events, 328

resulting in thousands of homeologs due to whole-genome duplication. Thus, we next 329

investigated whether the parent-of-origin-dependent expression pattern is conserved 330

among these homeologous genes. Interestingly, we successfully distinguished the 331

parental origins of reads for 25 pairs of homeologous genes in reciprocally crossed 332

hexaploid wheat, as supported by SNP information. Among these gene pairs, 10 pairs 333

simultaneously exhibited imprinted expression patterns, accounting for 40% of the total, 334

whereas the proportion for tetraploid wheat was 23.1% (3/13) (Figure 4B, C and D, 335

Supplemental Data Set 4). In addition, although SNP information might have been 336

unavailable for the homeologs of imprinted genes in one wheat species, it might have 337

been available for other wheat species for a subset of homeologs. In total, 37 pairs of 338

such homeologs simultaneously exhibited imprinted expression patterns in two wheat 339

species (Supplemental Data Set 5). 340

In addition to homeologs resulting from whole genome duplication, many homologs 341

evolved via tandem duplication after whole genome duplication, such as Fie1 and Fie2 in 342

maize, which both exhibit imprinted expression patterns in developing endosperm, 343

although in a stage-specific manner for Fie2 (Dickinson et al., 2012). As expected, we 344

identified 34 and 20 homologous pairs (identity > 90%, E-value < 1e-10) of imprinted 345

genes containing SNP information in tetraploid and hexaploid wheat, respectively, 14 and 346

10 of which also showed conserved imprinted expression patterns in the respective hybrid 347

endosperm (Supplemental Data Set 6). For example, the homologous genes 348

TRIAE_CS42_6AS_TGACv1_487181_AA1568730 and 349

TRIAE_CS42_6AS_TGACv1_485362_AA1544140, which are located on the short arm of 350

12

chromosome 6A in tetraploid wheat, encode an F-box domain-containing protein, and 351

both exhibited paternally biased expression patterns in J and S reciprocal endosperm. 352

Furthermore, TRIAE_CS42_3B_TGACv1_220627_AA0712550, 353

TRIAE_CS42_3B_TGACv1_220627_AA0712570, 354

TRIAE_CS42_3B_TGACv1_220627_AA0712610, and 355

TRIAE_CS42_3B_TGACv1_220627_AA0712620 are located close to each other on 356

chromosome 3B in hexaploid wheat and encode a putative E3 ubiquitin-protein ligase. 357

Supported by SNP information, we found that all of these genes showed maternally 358

preferred expression patterns in K and D reciprocal crosses. In conclusion, our results 359

suggest that the expression patterns of imprinted genes were largely conserved 360

throughout the evolutionary history of wheat. 361

13

DISCUSSION 362

Genomic Imprinting Occurs Extensively in Wheat Species of Various Ploidy Levels 363

Hexaploid wheat (AABBDD; Triticum aestivum) has undergone two allopolyploidization 364

episodes during its evolutionary history, including tetraploidization and hexaploidization, 365

which involved the hybridization of Triticum urartu (AA), an unidentified species (BB), 366

and Aegilops tauschii (DD) (Kihara, 1944; McFadden and Sears, 1946; Dvorak et al., 367

1998; Huang et al., 2002). Thus, it is difficult to identify imprinted genes in polyploid 368

wheat due to the widespread presence of homeologs resulting from genome duplication. 369

In this study, we performed genome-wide identification of imprinted genes in 370

reciprocally crossed endosperm from diploid, tetraploid, and hexaploid wheat, and 371

detected 91, 135, and 146 imprinted genes in reciprocal endosperm, respectively, 372

including 246 MEGs and 126 PEGs. We validated 21 out of 23 imprinted genes by 373

RT-PCR followed by CAPS or sequencing, suggesting that our strategy was highly 374

effective for identifying imprinted genes in polyploid plants. In addition, we 375

experimentally confirmed the parent-of-origin expression patterns of two genes 376

(TRIAE_CS42_1DL_TGACv1_061444_AA0195450 and 377

TRIAE_CS42_5AS_TGACv1_393185_AA1269500), which were identified as imprinted 378

genes in only one biological replicate (Supplemental Figure 5, Supplemental Data Set 379

2), indicating that our criteria for identifying imprinted genes might have been too 380

stringent for a subset of genes expressed in the endosperm. 381

The imprinted genes identified in this study are unevenly distributed on the 382

chromosomes, with the D subgenome containing the smallest number of imprinted genes 383

in hexaploid wheat (55, 69, and 19 imprinted genes for the A, B, and D subgenome, 384

respectively). This is likely due to the reduced genetic diversity of the D subgenome 385

compared with the A and B subgenome that arose during hexaploid wheat evolution and 386

domestication, since we found less SNP information in the D subgenome (1,533) than in 387

the A (4,863) and B subgenome (6,197). This observation is consistent with the previous 388

finding that the D subgenome has the lowest nucleotide diversity among the three 389

subgenomes in hexaploid wheat (Akhunov et al., 2010). Accordingly, the proportion of 390

genes that could be evaluated for imprinting was 15.7%, 16.8%, and 5.1%, respectively. 391

Therefore, the reduced SNP information in the D subgenome among wheat lines used in 392

14

the present study probably led us to underestimate the number of imprinted genes in the 393

D subgenome. In addition, only 125 MEGs and 51 PEGs exhibited a persistent imprinted 394

expression pattern during endosperm development (Figure 2B), and a large proportion 395

(52.7%) of genes showed stage-specific imprinting patterns due to their biallelic 396

expression during other developmental stages. These results are consistent with findings 397

for rice, maize, Arabidopsis, castor bean, and sorghum (Gehring et al., 2011; Hsieh et al., 398

2011; Luo et al., 2011; Waters et al., 2011; Wolff et al., 2011; Zhang et al., 2011; Waters 399

et al., 2013; Xin et al., 2013; Pignatta et al., 2014; Xu et al., 2014; Zhang et al., 2016). 400

Since we only considered reads that mapped to specific subgenomes and due to the 401

dynamic nature of genomic imprinting as well as the limited availability of SNP 402

information, it is reasonable to assume that we underestimated the number of imprinted 403

genes in wheat endosperm. Nevertheless, this study, which provides a genome-wide 404

survey of imprinted genes in various wheat species involving the use of high-throughput 405

RNA-seq analysis, indicates that genomic imprinting is widespread among wheat species. 406

407

Imprinted Genes were Evolutionarily Conserved during Wheat 408

Hexapolyploidization 409

Triticum and Aegilops species provide an ideal system for studying polyploid genome 410

evolution, because hexaploid wheat has extant diploid and tetraploid progenitors with 411

well-established phylogenetic relationships (Levy and Feldman, 2004; Feldman and Levy, 412

2005). In a comprehensive study of genomic imprinting in maize, ten pairs of 413

homologous genes exhibited conserved maternally biased expression patterns in 414

endosperm, suggesting that genomic imprinting might have been maintained during the 415

various genome duplication events (Waters et al., 2013), this prompted us to investigate 416

the conservation of genomic imprinting among wheat species with various ploidy levels. 417

We found that the parent-of-origin expression patterns are evolutionarily conserved 418

among wheat species with different ploidy levels, as 51.4% (18/35) of imprinted genes in 419

diploid or tetraploid wheat were also imprinted in hexaploid wheat when considering the 420

criteria of both sequence identity and syntenic chromosome region (Figure 4A and 421

Supplemental Data Set 3). These imprinted genes, which are conserved among wheat 422

species, might play crucial roles in regulating seed development, among which, 423

15

TRIAE_CS42_2BS_TGACv1_147972_AA0489480, a homolog of Arabidopsis DA 424

(LARGE IN CHINESE) 1, was identified as a PEG in both hexaploid and tetraploid wheat 425

endosperm at 15 DAP. Interestingly, one amino acid change in DA1 (arginine-to-lysine at 426

position 358) dramatically increases seed size by extending the duration of proliferative 427

growth, suggesting that the imprinted DA1 gene might help regulate seed development 428

(Li et al., 2008). Notably, TRIAE_CS42_5BL_TGACv1_410079_AA1366740 encodes a 429

component of PRC2 in wheat (homolog to EMBRYONIC FLOWER 2 in Arabidopsis 430

and rice) and exhibits maternally biased expression patterns in both tetraploid and 431

hexaploid wheat (Supplemental Data Set 3), indicating parent-of-origin effects of PRC2 432

are conserved during seed development among wheat, rice, maize, and Arabidopsis, 433

although the imprinted PRC2 genes might be various between each other. In addition, 50 434

pairs of homeologs (13 in one wheat species and 37 in another wheat species) exhibited 435

conserved genomic imprinting in reciprocally crossed endosperm (Supplemental Data 436

Set 4 and 5). 437

There is relatively little overlap between imprinted genes in Arabidopsis vs. 438

monocots and rice vs. maize, indicating that parent-of-origin expression patterns tend to 439

vary during evolutionary history (Waters et al., 2013; Pires and Grossniklaus, 2014), with 440

the exception of paternally expressed YUCCA genes and TAR1 in rice, maize, and 441

Arabidopsis endosperm (Hsieh et al., 2011; Luo et al., 2011; Zhang et al., 2011; Chen et 442

al., 2017). In addition, the proportion of commonly imprinted genes is also limited 443

between C. rubella and Arabidopsis (Hatorangan et al., 2016). However, extensive 444

conservation of imprinted expression patterns was revealed between A. lyrata and A. 445

thaliana (Klosinska et al., 2016). We compared the conservation of imprinted genes 446

among diploid wheat, tetraploid wheat, hexaploid wheat, maize, rice, sorghum, castor 447

bean, and Arabidopsis (sequence identity > 50%, E-value < 1e-10). As shown in 448

Supplemental Data Set 7, we found 52 homologous pairs (20 between hexaploid and 449

tetraploid wheat, 16 between hexaploid and diploid wheat, and 16 between diploid and 450

tetraploid wheat) with conserved imprinted expression patterns between two wheat 451

species. Furthermore, the overlap of genomic imprinting among diploid, tetraploid, and 452

hexaploid wheat was the most significant among the plant species examined according to 453

Fisher’s exact test. We also detected statistically significant overlaps in imprinted genes 454

16

between wheat species and maize, as well as between wheat and rice, but not between 455

wheat and Arabidopsis (Supplemental Figure 6, Supplemental Table 2 and 456

Supplemental Data Set 7). This finding is not unexpected, since these three wheat 457

species are closely related, and hexaploid wheat appeared only approximately ~10,000 458

years ago due to the hybridization of diploid and tetraploid wheat (Feldman, 1995). 459

Furthermore, monocots branched off from dicots 140–150 MYA, whereas wheat, rice, 460

and maize diverged from a common ancestor ~40 MYA (Gill et al., 2004).In summary, 461

our analyses indicated that the degree of between-species overlap of genes exhibiting 462

parent-of-origin expression biases is correlated with their phylogenetic relationship. 463

464

Interplay among Subgenomes Might Influence Genomic Imprinting in Hexaploid 465

Wheat 466

It is thought that homeologs make unequal contributions to total gene expression levels in 467

polyploid wheat and that gene expression is regulated in a complex manner during grain 468

development, possibly due to crosstalk between genomes during polyploidization 469

(Akhunova et al., 2010; Chague et al., 2010; Leach et al., 2014; Liu et al., 2015; Han et 470

al., 2016). In this study, a major proportion of homeologous gene pairs (65.8%) indeed 471

exhibited divergent expression patterns in terms of genomic imprinting in polyploid 472

wheat. Genomic imprinting is a contributing factor to the divergence in expression 473

patterns of duplicated genes due to the silencing of one allele in a 474

parent-of-origin-specific manner (Qiu et al., 2014). We also found that the silencing of 475

the parental allele varied among homologs after polyploidization. For example, the 476

maternal allele of TRIAE_CS42_3DL_TGACv1_249702_AA0854790 was preferentially 477

expressed in diploid wheat endosperm, whereas its homolog, 478

TRIAE_CS42_1AL_TGACv1_000533_AA0014130, exhibited paternally biased 479

expression after polyploidization in hexaploid wheat. In addition, many MEGs and PEGs 480

arose after the hexaploidization event, e.g., 22 imprinted genes in hexaploid wheat 481

exhibited biallelic expression patterns in diploid or tetraploid wheat (Figure 4). These 482

findings indicate that parent-of-origin gene expression is more prevalent in hexaploid 483

wheat than in diploid and tetraploid wheat and that interplay among subgenomes might 484

play a role in regulating genomic imprinting, a topic that merits further investigation. 485

17

METHODS 486

Plant Materials 487

Hexaploid wheat (Triticum aestivum, AABBDD) cultivars Keyi5214 (K) and Doumai (D) 488

and tetraploid wheat (Triticum turgidum, AABB) cultivars SCAUP (S) and Jinying8 (J), 489

as well as diploid goatgrass (Aegilops tauschii, DD) lines Y177 (Y) and RM220 (R), 490

were sown in a field at China Agricultural University, Beijing, China. Reciprocal crosses 491

and self-pollination were performed as follows: spikelets at the base and very top of the 492

spike, as well as florets from the central part of the spike, were removed before anthesis, 493

and the top of the florets was cut off and bagged. Pollination was performed 1–2 days 494

later using the appropriate pollen. Endosperm tissues were collected from at least three 495

different ears to create three biological replicates at 15, 20, and 25 DAP; the endosperm 496

tissues were isolated by hand dissection and immediately frozen in liquid nitrogen. 497

498

RNA Extraction 499

Total RNA was extracted from 60 (20 samples × 3 replicates) plant samples using the 500

SDS-phenol method (Shirzadegan et al., 1991) with some modifications. Endosperm 501

tissue (~0.5 g) was ground to a fine powder in liquid nitrogen and mixed with 6 mL of 502

buffer containing 1% SDS, 50 mM Tris-HCl (pH 8.0), 150 mM LiCl, 5 mM EDTA, and 503

10 mM DTT. The sample was combined with 6 mL phenol: chloroform (5:1, pH 4.5, 504

Ambion AM9720) and incubated on ice for 5 min. The mixture was centrifuged at 5,000 505

rpm for 10 min at 4°C and the aqueous phase was transferred to a new tube. These steps 506

were repeated using phenol:chloroform (1:1), followed by chloroform alone. The RNA 507

was then precipitated with 2.5 M LiCl at 4°C overnight, washed with ice-cold 2 M LiCl, 508

dissolved in TE, mixed with a 1/9 volume of 3 M sodium acetate (pH 5.2) and 2.5 509

volumes of ethanol, and incubated at -80°C for at least 4 h, after which the RNA was 510

pelleted by centrifugation at 14,000 rpm for 15 min at 4°C. After rinsing with 75% 511

ethanol and air-drying, the RNA was dissolved in diethylpyrocarbonate-treated water. 512

DNA was removed with TURBO DNase I (Ambion), and the RNA was purified using an 513

RNeasy column (Qiagen). 514

515

Illumina Sequencing 516

18

The RNA samples were sent to Berry Genomics Co., Ltd. 517

(http://www.berrygenomics.com/) for mRNA library construction and deep sequencing 518

using the Illumina HiSeq2000 platform. Before library construction, the quality of the 519

RNA samples was examined using an Agilent 2100 Bioanalyzer. High-quality mRNA 520

from three biological replicates per sample was sequenced. FastQC software (v0.11.5; 521

http://www.bioinformatics.babraham.ac.uk) was used to examine the sequencing quality 522

of the reads in each sample (Andrews, 2010). Then, raw data were processed using 523

Trimmomatic (v0.36; http://www.usadellab.org/cms/index.php?page=trimmomatic) to 524

trim adaptor sequence and low-quality end (Bolger et al., 2014), and only high-quality 525

reads were retained for further analysis. In total, 125.5, 400.4, and 430.1 Gb of 526

high-quality RNA-Seq data were generated from parental and reciprocally crossed 527

endosperm from diploid, tetraploid, and hexaploid wheat, respectively. Because the 528

sequencing depth of the first replicate was not as high as that of the other two, data from 529

the first two replicates were combined, and three sets of sequenced transcriptomes were 530

considered to represent two biological replicates. The correlation coefficient of two 531

biological replicates was 0.977–0.998. The biological replicates were treated 532

independently and imprinted genes were identified separately and then compared with 533

each other; only overlapping candidates were considered to represent imprinted genes. 534

The RNA-seq reads used in this study were deposited in the National Center for 535

Biotechnology Information Short Read Archive under accession number SRP075528. 536

537

SNP Calling and Identification of Imprinted Genes 538

RNA-seq data from each parent (K and D for hexaploid wheat, J and S for tetraploid 539

wheat, Y and R for diploid goatgrass) were used for SNP identification. High-quality 540

reads were mapped to the reference gene sequence (TGACv1) using Bowtie2 (v2.2.9; 541

Langmead and Salzberg, 2012) with the parameters “--end-to-end --reorder --score-min L, 542

-0.6,-0.3 -L 15”. To improve credibility, only reads that uniquely mapped to one 543

subgenome with no more than two mismatches were considered. SNP calling was then 544

performed using the mpileup function of Samtools (v1.4; Li et al., 2009) and the call 545

function of BCFtools (v1.4; Li, 2011). SNPs supported by ≥10 reads, ≥95% of the total 546

SNP-site mapped reads, and a genotype-likelihood of ≥ 95% in each parent were 547

19

identified as SNPs between parents and used for subsequent allele-specific expression 548

analysis in hybrids. 549

RNA-seq reads from reciprocal crosses of both biological replicates were mapped to 550

the reference genes separately using Bowtie2 (v2.2.9; Langmead and Salzberg, 2012), 551

and only uniquely mapped reads with no more than two mismatches were retained. 552

SNP-containing reads originating from different parents were then distinguished based on 553

maternal and paternal SNPs identified in the previous step and counted using customized 554

Perl scripts. Genes with a ratio deviating from 2m:1p (Chi-Square Goodness-of-Fit Test, 555

FDR-adjusted p value < 0.05) and ≥90% of total SNP-containing reads that were 556

maternally-derived or ≥70% that were paternally-derived in two reciprocal crosses of 557

both biological replicates (with a minimum of 10 SNP-associated reads per cross) were 558

identified as imprinted genes. 559

560

Cleaved Amplified Polymorphic Sequence (CAPS) Assay 561

RNA samples from reciprocal crosses of wheat species with different ploidy levels were 562

independently prepared to validate imprinted gene expression patterns using CAPS 563

assays, as previously described (Konieczny and Ausubel, 1993), or by sequencing. 564

RT-PCR was performed using the gene- or homeolog-specific primers listed in 565

Supplemental Table 3. The amplification products were digested with the restriction 566

enzymes listed in Supplemental Table 3. 567

568

Accession Numbers 569

The RNA-seq reads used in this study were deposited in the National Center for 570

Biotechnology Information Short Read Archive under accession number SRP075528. 571

572

Supplemental Data 573

Supplemental Figure 1. Parental expression ratio plot for endosperm-expressed genes at 574

different developmental stages for each reciprocal cross in diploid, tetraploid, and 575

hexaploid wheat species. 576

Supplemental Figure 2. The expression levels of MEGs are higher than those of PEGs 577

in the aleurone layer and transfer cells of 20- and 30-DAP wheat endosperm. 578

20

Supplemental Figure 3. Experimental validation of the 13 imprinted genes in diploid, 579

tetraploid, and hexaploid wheat by CAPS assays. 580

Supplemental Figure 4. Experimental validation of 11 imprinted genes in diploid, 581

tetraploid, and hexaploid wheat by sequencing. 582

Supplemental Figure 5. Experimental validation of candidate imprinted genes identified 583

in only one biological replicate of diploid and hexaploid wheat by sequencing. 584

Supplemental Figure 6. Conservation of imprinted genes among different species. 585

Supplemental Table 1. Summary of RNA-Seq data and reads mapping results. 586

Supplemental Table 2. Overlaps between wheat imprinted genes and those of maize, rice, 587

Arabidopsis, sorghum, and castor bean. 588

Supplemental Table 3. Primers and enzymes used for sequencing and CAPS assays. 589

Supplemental Data Set 1. Allele-specific expression of the 91, 135, and 146 imprinted 590

genes in diploid, tetraploid, and hexaploid wheat, respectively. 591

Supplemental Data Set 2. Allele-specific expression of imprinted candidate genes from 592

each replicate in diploid, tetraploid, and hexaploid wheat, respectively. 593

Supplemental Data Set 3. Parent-of-origin expression of imprinted genes is highly 594

conserved between diploid/tetraploid and hexaploid wheat. 595

Supplemental Data Set 4. Thirteen pairs of homeologs show similar imprinted 596

expression patterns in tetraploid and hexaploid wheat. 597

Supplemental Data Set 5. Expression patterns of homeologs of imprinted genes in 598

different wheat species. 599

Supplemental Data Set 6. Imprinted homologous pairs resulting from tandem 600

duplication after polyploidization in tetraploid and hexaploid wheat. 601

Supplemental Data Set 7. Conserved imprinted genes in various plant species. 602

603

ACKNOWLEDGEMENTS 604

This work was supported by the National Key Research and Development Program 605

of China (2016YFD0101004), the Major Program of the National Natural Science 606

Foundation of China (31290210), the National Natural Science of China (31471479), and 607

Chinese Universities Scientific Fund (2017TC035). 608

609

21

AUTHOR CONTRIBUTIONS 610

M.X., Z.N., and Q.S. conceived the project. G.Y., Y.Y., and H.P. collected the plant611

materials. G.Y. and M.F. performed the research. Z.L. and K.Y. analyzed the data. M.X., 612

Q.S., and Z.N. wrote the manuscript.613

22

FIGURE LEGENDS 614

Figure 1. Most Genes Exhibited the Expected Expression Ratios in Developing 615

Wheat Endosperm 616

Parental expression ratios plot for each reciprocal cross in diploid, tetraploid, and 617

hexaploid wheat species. The expression levels of paternal (y-axis) and maternal (x-axis) 618

alleles are represented by the log2-transformed read counts of the paternally- and 619

maternally-derived reads in the reciprocal crosses, respectively. The expression patterns 620

of 3,485, 4,612, and 5,063 genes with SNPs were analyzed in reciprocally crossed 621

endosperm for diploid, tetraploid, and hexaploid wheat, respectively. Of these, 434, 1,307, 622

and 1,630 genes were identified as parental biased expressed genes in diploid, tetraploid, 623

and hexaploid wheat endosperm, respectively, according to a Chi-Square Goodness-of-Fit 624

Test (FDR-adjusted p < 0.05). The dashed diagonal line represents the expected 2m:1p 625

ratio. DAP: days after pollination, CS: Chinese Spring. 626

Figure 2. Computational Identification of Imprinted Genes in Wheat Endosperm 627

(A) Ratio-based cutoff to identify MEGs and PEGs. Spots clustered in the upper-right 628

corners have more than 90% maternal reads (red, MEGs), whereas spots clustered in the 629

lower-left corners have more than 70% paternal reads (green, PEGs). Black dots 630

represent non-imprinted genes. The intersection of the dashed lines indicates a 2m:1p 631

ratio. Dots representing MEGs and PEGs are semitransparent. Y: Y177, R: RM220, J: 632

Jinying 8, S: SCAUP, D: Doumai, K: Keyi5214. 633

(B) Venn diagram analysis of imprinted genes. The number of imprinted genes identified 634

at 15, 20, and 25 DAP are shown in the red, green, and blue circles, respectively. 635

DAP: days after pollination, MEG: maternally expressed gene, PEG: paternally expressed 636

gene. 637

Figure 3. Expression of Imprinted Genes in Different Wheat Tissues 638

(A) The expression levels of MEGs and PEGs were examined in 13 wheat tissues. The 639

average expression level is higher in endosperm than in other tissues. MEGs appeared to 640

be more endosperm-specific than PEGs. Dashed line indicates the average expression 641

level of imprinted genes in different tissues. The color scale from blue (low) to red (high) 642

indicates relative gene expression level. 643

(B) The expression levels of MEGs and PEGs are higher than those of non-imprinted 644

23

genes in diploid, tetraploid, and hexaploid wheat species at all stages examined. The 645

number indicates the average expression level (log2-transformed FPKM). 646

Figure 4. Imprinted Genes that were Evolutionarily Conserved during 647

Hexaploidization 648

(A) Parent-of-origin expression patterns of imprinted genes are highly conserved among649

wheat species. The white bar indicates the number of imprinted genes in different 650

subgenomes of diploid, tetraploid, and hexaploid wheat; light gray bar indicates the 651

number of imprinted genes with SNPs in other wheat species; dark gray bar indicates the 652

number of conserved imprinted genes in different wheat species; black bar indicates the 653

number of conserved candidate imprinted genes with biased expression patterns in 654

different wheat species only considering the criterion of FDR-adjusted p value. 655

(B–D) Thirteen pairs of homeologs show similar imprinted expression patterns in 656

tetraploid and hexaploid wheat. Vertical lines indicate the 13 groups of homeologous 657

wheat genes. Blue (low), white (medium), and red (high) represent the relative expression 658

levels of maternal or paternal alleles. DAP: days after pollination, S: SCAUP, J: Jinying 8, 659

D: Doumai, K: Keyi5214, MEG: maternally expressed gene, PEG: paternally expressed 660

gene.661

662

24

REFERENCES 663Akhunov, E.D., Akhunova, A.R., Anderson, O.D., Anderson, J.A., Blake, N., Clegg, M.T., 664

Coleman-Derr, D., Conley, E.J., Crossman, C.C., Deal, K.R., Dubcovsky, J., Gill, B.S., Gu, 665Y.Q., Hadam, J., Heo, H., Huo, N.X., Lazo, G.R., Luo, M.C., Ma, Y.Q., Matthews, D.E., 666McGuire, P.E., Morrell, P.L., Qualset, C.O., Renfro, J., Tabanao, D., Talbert, L.E., Tian, C., 667Toleno, D.M., Warburton, M.L., You, F.M., Zhang, W.J., and Dvorak, J. (2010). Nucleotide 668diversity maps reveal variation in diversity among wheat genomes and chromosomes. BMC 669Genomics 11, 702. 670

Akhunova, A.R., Matniyazov, R.T., Liang, H., and Akhunov, E.D. (2010). Homoeolog-specific 671transcriptional bias in allopolyploid wheat. BMC Genomics 11, 505. 672

Andrews, S. (2010). FastQC: a quality control tool for high throughput sequence data. Reference 673Source. 674

Autran, D., Baroux, C., Raissig, M.T., Lenormand, T., Wittig, M., Grob, S., Steimer, A., Barann, M., 675Klostermeier, U.C., Leblanc, O., Vielle-Calzada, J.P., Rosenstiel, P., Grimanelli, D., and 676Grossniklaus, U. (2011). Maternal epigenetic pathways control parental contributions to 677Arabidopsis early embryogenesis. Cell 145, 707-719. 678

Bolger, A.M., Lohse, M., and Usadel, B. (2014). Trimmomatic: a flexible trimmer for Illumina sequenc679e data. Bioinformatics 30, 2114-2120. 680

Chague, V., Just, J., Mestiri, I., Balzergue, S., Tanguy, A.M., Huneau, C., Huteau, V., Belcram, H., 681Coriton, O., Jahier, J., and Chalhoub, B. (2010). Genome-wide gene expression changes in 682genetically stable synthetic and natural wheat allohexaploids. New Phytol. 187, 1181-1194. 683

Chaudhury, A.M., Ming, L., Miller, C., Craig, S., Dennis, E.S., and Peacock, W.J. (1997). 684Fertilization-independent seed development in Arabidopsis thaliana. Proc. Natl. Acad. Sci. 685USA 94, 4223-4228. 686

Chen, C., Li, T., Zhu, S., Liu, Z., Shi, Z., Zheng, X., Chen, R., Huang, J., Shen, Y., Luo, S., Wang, L., 687Liu, Q.Q., and E, Z. (2017). Characterization of imprinted genes in rice reveals 688post-fertilization regulation and conservation at some loci of imprinting in plant species. 689bioRxiv, doi.org/10.1101/143214. 690

Costa, L.M., Marshall, E., Tesfaye, M., Silverstein, K.A., Mori, M., Umetsu, Y., Otterbach, S.L., 691Papareddy, R., Dickinson, H.G., Boutiller, K., VandenBosch, K.A., Ohki, S., and 692Gutierrez-Marcos, J.F. (2014). Central cell-derived peptides regulate early embryo patterning 693in flowering plants. Science 344, 168-172. 694

Curley, J.P., Barton, S., Surani, A., and Keverne, E.B. (2004). Coadaptation in mother and infant 695regulated by a paternally expressed imprinted gene. Proc. Biol. Sci. 271, 1303-1309. 696

Danilevskaya, O.N., Hermon, P., Hantke, S., Muszynski, M.G., Kollipara, K., and Ananiev, E.V. (2003). 697Duplicated fie genes in maize: expression pattern and imprinting suggest distinct functions. 698Plant Cell 15, 425-438. 699

Dickinson, H., Costa, L., and Gutierrez-Marcos, J. (2012). Epigenetic neofunctionalisation and 700regulatory gene evolution in grasses. Trends Plant Sci. 17, 389-394. 701

Dvořák, J. (1976). The relationship between the genome of Triticum urartu and the A and B genomes 702of Triticum aestivum. Can. J. Genet. Cytol. 18, 371-377. 703

Dvorak, J., and Akhunov, E.D. (2005). Tempos of gene locus deletions and duplications and their 704relationship to recombination rate during diploid and polyploid evolution in the 705Aegilops-Triticum alliance. Genetics 171, 323-332. 706

25

Dvorak, J., Luo, M.C., Yang, Z.L., and Zhang, H.B. (1998). The structure of the Aegilops tauschii 707genepool and the evolution of hexaploid wheat. Theor. Appl. Genet. 97, 657-670. 708

Feldman, H.A. (1995). On the allometric mass exponent, when it exists. J. Theor. Biol. 172, 187-197. 709Feldman, M., and Levy, A.A. (2005). Allopolyploidy - a shaping force in the evolution of wheat 710

genomes. Cytogenet Genome Res. 109, 250-258. 711Fouquet, R., Martin, F., Fajardo, D.S., Gault, C.M., Gomez, E., Tseung, C.W., Policht, T., Hueros, G., 712

and Settles, A.M. (2011). Maize rough endosperm3 encodes an RNA splicing factor required 713for endosperm cell differentiation and has a nonautonomous effect on embryo development. 714Plant Cell 23, 4280-4297. 715

Gehring, M., Missirian, V., and Henikoff, S. (2011). Genomic analysis of parent-of-origin allelic 716expression in Arabidopsis thaliana seeds. PLoS One 6, e23687. 717

Gill, B.S., Appels, R., Botha-Oberholster, A.-M., Buell, C.R., Bennetzen, J.L., Chalhoub, B., Chumley, 718F., Dvořák, J., Iwanaga, M., and Keller, B. (2004). A workshop report on wheat genome 719sequencing. Genetics 168, 1087-1096. 720

Grimanelli, D., Perotti, E., Ramirez, J., and Leblanc, O. (2005). Timing of the maternal-to-zygotic 721transition during early seed development in maize. Plant Cell 17, 1061-1072. 722

Grossniklaus, U., Vielle-Calzada, J.P., Hoeppner, M.A., and Gagliano, W.B. (1998). Maternal control of 723embryogenesis by MEDEA, a polycomb group gene in Arabidopsis. Science 280, 446-450. 724

Guitton, A.E., and Berger, F. (2005). Loss of function of MULTICOPY SUPPRESSOR OF IRA 1 725produces nonviable parthenogenetic embryos in Arabidopsis. Curr. Biol. 15, 750-754. 726

Haig, D., and Westoby, M. (1989). Parent-specific gene-expression and the triploid endosperm. Amer. 727Nat. 134, 147-155. 728

Han, Y., Xin, M.M., Huang, K., Xu, Y.Y., Liu, Z.S., Hu, Z.R., Yao, Y.Y., Peng, H.R., Ni, Z.F., and Sun, 729Q.X. (2016). Altered expression of TaRSL4 gene by genome interplay shapes root hair length 730in allopolyploid wheat. New Phytol. 209, 721-732. 731

Hatorangan, M.R., Laenen, B., Steige, K., Slotte, T., and Kohler, C. (2016). Rapid evolution of 732genomic imprinting in two species of the Brassicaceae. Plant Cell 28, 1815-1827. 733

Hsieh, T.F., Shin, J., Uzawa, R., Silva, P., Cohen, S., Bauer, M.J., Hashimoto, M., Kirkbride, R.C., 734Harada, J.J., Zilberman, D., and Fischer, R.L. (2011). Regulation of imprinted gene expression 735in Arabidopsis endosperm. Proc. Natl. Acad. Sci. USA 108, 1755-1762. 736

Huang, S., Sirikhachornkit, A., Su, X., Faris, J., Gill, B., Haselkorn, R., and Gornicki, P. (2002). Genes 737encoding plastid acetyl-CoA carboxylase and 3-phosphoglycerate kinase of the 738Triticum/Aegilops complex and the evolutionary history of polyploid wheat. Proc. Natl. Acad. 739Sci. USA 99, 8133-8138. 740

Keverne, E.B., and Curley, J.P. (2008). Epigenetics, brain evolution and behaviour. Front. 741Neuroendocrinol. 29, 398-412. 742

Kihara, H. (1944). Discovery of the DD-analyser, one of the ancestors of Triticum vulgare (abstr). 743Agric. Hortic. 19, 889-890. 744

Kiyosue, T., Ohad, N., Yadegari, R., Hannon, M., Dinneny, J., Wells, D., Katz, A., Margossian, L., 745Harada, J.J., Goldberg, R.B., and Fischer, R.L. (1999). Control of fertilization-independent 746endosperm development by the MEDEA polycomb gene in Arabidopsis. Proc. Natl. Acad. Sci. 747USA 96, 4186-4191. 748

Klosinska, M., Picard, C.L., and Gehring, M. (2016). Conserved imprinting associated with unique 749epigenetic signatures in the Arabidopsis genus. Nat. Plants 2, 16145. 750

26

Kohler, C., Hennig, L., Bouveret, R., Gheyselinck, J., Grossniklaus, U., and Gruissem, W. (2003). 751Arabidopsis MSI1 is a component of the MEA/FIE Polycomb group complex and required for 752seed development. EMBO J. 22, 4804-4814. 753

Konieczny, A., and Ausubel, F.M. (1993). A procedure for mapping Arabidopsis mutations using 754co-dominant ecotype-specific PCR-based markers. Plant J. 4, 403-410. 755

Kradolfer, D., Wolff, P., Jiang, H., Siretskiy, A., and Kohler, C. (2013). An imprinted gene underlies 756postzygotic reproductive isolation in Arabidopsis thaliana. Dev. Cell 26, 525-535. 757

Langmead, B., and Salzberg, S.L. (2012). Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 758357. 759

Leach, L.J., Belfield, E.J., Jiang, C.F., Brown, C., Mithani, A., and Harberd, N.P. (2014). Patterns of 760homoeologous gene expression shown by RNA sequencing in hexaploid bread wheat. BMC 761Genomics 15. 762

Leroy, O., Hennig, L., Breuninger, H., Laux, T., and Kohler, C. (2007). Polycomb group proteins 763function in the female gametophyte to determine seed development in plants. Development 764134, 3639-3648. 765

Levy, A.A., and Feldman, M. (2004). Genetic and epigenetic reprogramming of the wheat genome 766upon allopolyploidization. Bio. J. Linn. Soc. 82, 607-613. 767

Li, H. (2011). A statistical framework for SNP calling, mutation discovery, association mapping and 768population genetical parameter estimation from sequencing data. Bioinformatics 27, 7692987-2993. 770

Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., Durbin, 771R., and Proc, G.P.D. (2009). The sequence alignment/map format and SAMtools. 772Bioinformatics 25, 2078-2079. 773

Li, Y., Zheng, L., Corke, F., Smith, C., and Bevan, M.W. (2008). Control of final seed and organ size by 774the DA1 gene family in Arabidopsis thaliana. Genes Dev. 22, 1331-1336. 775

Lin, B.Y. (1984). Ploidy barrier to endosperm development in maize. Genetics 107, 103-115. 776Liu, Z.S., Xin, M.M., Qin, J.X., Peng, H.R., Ni, Z.F., Yao, Y.Y., and Sun, Q.X. (2015). Temporal 777

transcriptome profiling reveals expression partitioning of homeologous genes contributing to 778heat and drought acclimation in wheat (Triticum aestivum L.). BMC Plant Biol. 15. 779

Luo, M., Bilodeau, P., Koltunow, A., Dennis, E.S., Peacock, W.J., and Chaudhury, A.M. (1999). Genes 780controlling fertilization-independent seed development in Arabidopsis thaliana. Proc. Natl. 781Acad. Sci. USA 96, 296-301. 782

Luo, M., Taylor, J.M., Spriggs, A., Zhang, H., Wu, X., Russell, S., Singh, M., and Koltunow, A. (2011). 783A genome-wide survey of imprinted genes in rice seeds reveals imprinting primarily occurs in 784the endosperm. PLoS Genet. 7, e1002125. 785

McFadden, E., and Sears, E. (1946). The origin of Triticum spelta and its free-threshing hexaploid 786relatives. J. Hered. 37, 107-116. 787

Meng, D., Zhao, J., Zhao, C., Luo, H., Xie, M., Liu, R., Lai, J., Zhang, X., and Jin, W. (2017) 788Sequential gene activation and gene imprinting during early embryo development in maize. Pl789ant J., accepted, doi:10.1111/tpj.13786. 790

Pfeifer, M., Kugler, K.G., Sandve, S.R., Zhan, B., Rudi, H., Hvidsten, T.R., Mayer, K.F., Olsen, O.A., 791and Consortium, I.W.G.S. (2014). Genome interplay in the grain transcriptome of hexaploid 792bread wheat. Science 345, 1250091. 793

Pignatta, D., Erdmann, R.M., Scheer, E., Picard, C.L., Bell, G.W., and Gehring, M. (2014). Natural 794

27

epigenetic polymorphisms lead to intraspecific variation in Arabidopsis gene imprinting. Elife 7953,e03198. 796

Pires, N.D., and Grossniklaus, U. (2014). Different yet similar: evolution of imprinting in flowering 797plants and mammals. F1000prime Rep. 6, 63. 798

Pont, C., and Salse, J. (2017). Wheat paleohistory created asymmetrical genomic evolution. Curr. Opin. 799Plant Biol. 36, 29-37. 800

Qiu, Y., Liu, S.L., and Adams, K.L. (2014). Frequent changes in expression profile and accelerated 801sequence evolution of duplicated imprinted genes in Arabidopsis. Genome Biol. Evol. 6, 8021830-1842. 803

Raissig, M.T., Bemer, M., Baroux, C., and Grossniklaus, U. (2013). Genomic imprinting in the Arabido804psis embryo is partly regulated by PRC2. PLoS Genet. 9, e1003862. 805

Rodrigues, J.A., and Zilberman, D. (2015). Evolution and function of genomic imprinting in plants. 806Genes Dev. 29, 2517-2531. 807

Schreiber, A.W., Sutton, T., Caldo, R.A., Kalashyan, E., Lovell, B., Mayo, G., Muehlbauer, G.J., Druka, 808A., Waugh, R., Wise, R.P., Langridge, P., and Baumann, U. (2009). Comparative 809transcriptomics in the Triticeae. BMC Genomics 10, 285. 810

Scott, R.J., Spielman, M., Bailey, J., and Dickinson, H.G. (1998). Parent-of-origin effects on seed 811development in Arabidopsis thaliana. Development 125, 3329-3341. 812

Shirzadegan, M., Christie, P., and Seemann, J.R. (1991). An efficient method for isolation of RNA from 813tissue cultured plant cells. Nucleic Acids Res. 19, 6055. 814

Swaney, W.T., Curley, J.P., Champagne, F.A., and Keverne, E.B. (2007). Genomic imprinting mediates 815sexual experience-dependent olfactory learning in male mice. Proc. Natl. Acad. Sci. USA 104, 8166084-6089. 817

Vielle-Calzada, J.P., Baskar, R., and Grossniklaus, U. (2000). Delayed activation of the paternal 818genome during seed development. Nature 404, 91-94. 819

Waters, A.J., Bilinski, P., Eichten, S.R., Vaughn, M.W., Ross-Ibarra, J., Gehring, M., and Springer, N.M. 820(2013). Comprehensive analysis of imprinted genes in maize reveals allelic variation for 821imprinting and limited conservation with other species. Proc. Natl. Acad. Sci. USA 110, 82219639-19644. 823

Waters, A.J., Makarevitch, I., Eichten, S.R., Swanson-Wagner, R.A., Yeh, C.T., Xu, W., Schnable, P.S., 824Vaughn, M.W., Gehring, M., and Springer, N.M. (2011). Parent-of-origin effects on gene 825expression and DNA methylation in the maize endosperm. Plant Cell 23, 4221-4233. 826

Wilkins, J.F., and Haig, D. (2003). What good is genomic imprinting: the function of parent-specific 827gene expression. Nat. Rev. Genet. 4, 359-368. 828

Wolf, J.B., and Hager, R. (2006). A maternal-offspring coadaptation theory for the evolution of 829genomic imprinting. PLoS Biol. 4, e380. 830

Wolff, P., Jiang, H., Wang, G., Santos-Gonzalez, J., and Kohler, C. (2015). Paternally expressed 831imprinted genes establish postzygotic hybridization barriers in Arabidopsis thaliana. Elife 4. 832

Wolff, P., Weinhofer, I., Seguin, J., Roszak, P., Beisel, C., Donoghue, M.T., Spillane, C., Nordborg, M., 833Rehmsmeier, M., and Kohler, C. (2011). High-resolution analysis of parent-of-origin allelic 834expression in the Arabidopsis endosperm. PLoS Genet. 7, e1002126. 835

Xin, M., Yang, R., Li, G., Chen, H., Laurie, J., Ma, C., Wang, D., Yao, Y., Larkins, B.A., Sun, Q., 836Yadegari, R., Wang, X., and Ni, Z. (2013). Dynamic expression of imprinted genes associates 837with maternally controlled nutrient allocation during maize endosperm development. Plant 838

28

Cell 25, 3212-3227. 839Xu, W., Dai, M., Li, F., and Liu, A. (2014). Genomic imprinting, methylation and parent-of-origin 840

effects in reciprocal hybrid endosperm of castor bean. Nucleic Acids Res. 42, 6987-6998. 841Yang, S., Johnston, N., Talideh, E., Mitchell, S., Jeffree, C., Goodrich, J., and Ingram, G. (2008). The 842

endosperm-specific ZHOUPI gene of Arabidopsis thaliana regulates endosperm breakdown 843and embryonic epidermal development. Development 135, 3501-3509. 844

Zhang, M., Li, N., He, W., Zhang, H., Yang, W., and Liu, B. (2016). Genome-wide screen of genes 845imprinted in sorghum endosperm, and the roles of allelic differential cytosine methylation. 846Plant J. 85, 424-436. 847

Zhang, M., Zhao, H., Xie, S., Chen, J., Xu, Y., Wang, K., Zhao, H., Guan, H., Hu, X., Jiao, Y., Song, W., 848and Lai, J. (2011). Extensive, clustered parental imprinting of protein-coding and noncoding 849RNAs in developing maize endosperm. Proc. Natl. Acad. Sci. USA 108, 20042-20047. 850

851

Diploid Wheat Mapped to CS_D

Tetraploid Wheat Mapped to CS_A&B

Hexaploid Wheat Mapped to CS_A&B&D

2m:1p

2m:1p 2m

:1p

0

5

10

15

20

P

ater

nal E

xpre

ssio

n (L

og -

trans

form

ed R

ead

Cou

nts)

0 5 10 15 200 5 10 15 20 0 5 10 15 20

No. of genes with SNP: 3,485No. of genes: 434 (p < 0.05)

No. of genes with SNP: 4,612No. of genes: 1,307 (p < 0.05)

No. of genes with SNP: 5,063No. of genes: 1,630 (p < 0.05)

Figure 1. Most Genes Exhibited the Expected Expression Ratios in Developing Wheat Endosperm. Parental expression ratios plot for each reciprocal cross in diploid, tetraploid, and hexaploid wheat species. The expression levels of paternal (y-axis) and maternal (x-axis) alleles are represented by the log -transformed read counts of the paternally- and maternally-derived reads in the reciprocal crosses, respectively. The expression patterns of 3,485, 4,612, and 5,063 genes with SNPs were analyzed in reciprocally crossed endosperm for diploid, tetraploid, and hexaploid wheat, respectively. Of these, 434, 1,307, and 1,630 genes were identified as parental biased expressed genes in diploid, tetraploid, and hexaploid wheat endosperm, respectively, according to a Chi-Square Goodness-of-Fit Test (FDR-adjusted < 0.05). The dashed diagonal line represents the expected 2m:1p ratio. DAP: days after pollination, CS: Chinese Spring.

Maternal Expression (Log -transformed Read Counts)2

2

p

2

15 DAP 20 DAP 15 DAP 20 DAP

15 DAP 20 DAP 25 DAP

Mat

erna

l Rea

ds in

Y ×

R (%

)

0

20

40

60

80

100

Mat

erna

l Rea

ds in

J ×

S (%

)

0

20

40

60

80

100

Maternal Reads in R × Y(%) (Diploid) Maternal Reads in S × J (%) (Tetraploid)

Mat

erna

l Rea

ds in

K ×

D (%

)

0

20

40

60

80

100

Maternal Reads in D × K (%) (Hexaploid)

Maternally expressed gene

Paternally expressed gene

Non-imprinted gene

0 20 40 60 80 100 0 20 40 60 80 100

B

A

18 440 12 314 25 1550 14 1021

275

4

8

357

8

12

114

1

164

4

62 MEGs 29 PEGs 90 MEGs 45 PEGs 94 MEGs 52 PEGs Diploid Wheat Tetraploid Wheat Hexaploid Wheat

15 DAP 20 DAP 25 DAP

0 20 40 60 80 100 0 20 40 60 80 100

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

Figure 2. Computational Identification of Imprinted Genes in Wheat Endosperm.(A) Ratio-based cutoff to identify MEGs and PEGs. Spots clustered in the upper-right corners have more than 90% maternal reads (red, MEGs), whereas spots clustered in the lower-left corners have more than 70% paternal reads (green, PEGs). Black dots represent non-imprinted genes. The intersection of the dashed lines indicates a 2m:1p ratio. Dots representing MEGs and PEGs are semitransparent. Y: Y177, R: RM220, J: Jinying 8, S: SCAUP, D: Doumai, K: Keyi5214.(B) Venn diagram analysis of imprinted genes. The number of imprinted genes identified at 15, 20, and 25 DAP are shown in the red, green, and blue circles, respectively. DAP: days after pollination, MEG: maternally expressed gene, PEG: paternally expressed gene.

6.06.57.07.58.08.59.09.5 Average expression level

Coleoptile

RootEmbryo

Seedling root

Seedling crown

Seedling leaf

Immature in

floresce

nce

BractsPisti

ll

Anthesis

3-5 DAP caryo

psis

22 DAP embryo

22 DAP endosperm

Germinating seed

Before anthesis

051015

6.06.57.07.58.08.59.09.5

Log

-Tra

nsfo

rmed

Exp

ress

ion

Leve

l

A

B

(MEG

s)(P

EGs)

Figure 3. Expression of Imprinted Genes in Different Wheat Tissues.(A) The expression levels of MEGs and PEGs were examined in 13 wheat tissues. The average expression level is higher in endosperm than in other tissues. MEGs appeared to be more endosperm-specific than PEGs. Dashedline indicates the average expression level of imprinted genes in different tissues. The color scale from blue (low) to red (high) indicates relative gene expression level. (B) The expression levels of MEGs and PEGs are higher than those of non-imprinted genes in diploid, tetraploid, and hexaploid wheat species at all stages examined. The number indicates the average expression level (log -transformed FPKM).

15 20 15 20 15 20 25 (DAP)Diploid Tetraploid Hexaploid

Non-imprinted genesPEGsMEGs

10

5

0

-5

3.7

3.8

2.7

2.4

3.9

3.9

2.5

2.2

2.0

1.8

1.9

2.2

2.0

2.3

1.4

1.0

2.1

2.3 0.8

1.7

2.3

2

2

91

59

76

55

69

198

1314

1123

63 411

411

33 613

817

40

20

40

60

80

100Imprinted genesCandidates with SNPs in other wheat speciesCandidates with imprinted expression in other speciesCandidates with biased expression

Diploid A A BB AA BB DD

Tetraploid Hexaploid DD

TRIAE_CS42_1AL_TGACv1_000492_AA0013430TRIAE_CS42_1BL_TGACv1_030377_AA0088570TRIAE_CS42_2AS_TGACv1_113503_AA0356980TRIAE_CS42_2BS_TGACv1_146985_AA0476180TRIAE_CS42_3AL_TGACv1_193561_AA0613050TRIAE_CS42_3B_TGACv1_220594_AA0710320TRIAE_CS42_6AL_TGACv1_471030_AA1501260TRIAE_CS42_6DL_TGACv1_526797_AA1692130TRIAE_CS42_7BS_TGACv1_592641_AA1942120TRIAE_CS42_7DS_TGACv1_623437_AA2053190

TRIAE_CS42_1AL_TGACv1_000533_AA0014130TRIAE_CS42_1BL_TGACv1_031004_AA0105520TRIAE_CS42_2AL_TGACv1_094451_AA0297830TRIAE_CS42_2BL_TGACv1_129489_AA0385850TRIAE_CS42_6AS_TGACv1_488409_AA1575430TRIAE_CS42_6BS_TGACv1_514220_AA1657120TRIAE_CS42_7AL_TGACv1_557085_AA1776270TRIAE_CS42_7BL_TGACv1_577884_AA1885250TRIAE_CS42_7AS_TGACv1_569561_AA1819220TRIAE_CS42_7DS_TGACv1_622780_AA2045220

TRIAE_CS42_1AL_TGACv1_000879_AA0020900TRIAE_CS42_1BL_TGACv1_031537_AA0115800TRIAE_CS42_2AS_TGACv1_114436_AA0367220TRIAE_CS42_2BS_TGACv1_147972_AA0489480TRIAE_CS42_4BL_TGACv1_321258_AA1058000TRIAE_CS42_5AL_TGACv1_375999_AA1230490

Tetraploid15 20 (DAP)

JXS

SXJJX

S SXJ

(PEG)

020406080100

Maternal Reads (%)A B

C D

Hexaploid15 20 25 (DAP)

KXDDXK

KXDDXK

KXDDXK

(MEG)

Hexaploid15 20 25 (DAP)

KXDDXK

KXDDXK

KXDDXK

(PEG)

020406080100

Maternal Reads (%)

020406080100

Maternal Reads (%)

Figure 4. Imprinted Genes that were Evolutionarily Conserved during Hexaploidization. (A) Parent-of-origin expression patterns of imprinted genes are highly conserved among wheat species. The white barindicates the number of imprinted genes in different subgenomes of diploid, tetraploid, and hexaploid wheat; light gray barindicates the number of imprinted genes with SNPs in other wheat species; dark gray bar indicates the number of conservedimprinted genes in different wheat species; black bar indicates the number of conserved candidate imprinted genes withbiased expression patterns in different wheat species only considering the criterion of FDR-adjusted value.(B–D) Thirteen pairs of homeologs show similar imprinted expression patterns in tetraploid and hexaploid wheat. Verticallines indicate the 13 groups of homeologous wheat genes. Blue (low), white (medium), and red (high) represent the relativeexpression levels of maternal or paternal alleles. DAP: days after pollination, S: SCAUP, J: Jinying 8, D: Doumai, K: Keyi5214,MEG: maternally expressed gene, PEG: paternally expressed gene.

p