Global reinforcement of DNA methylation through enhancement of RNA-directed DNA methylation ensures sexual reproduction in rice

DNA methylation is an important epigenetic mark that regulates the expression of genes and transposons. RNA-directed DNA methylation (RdDM) is the main molecular pathway responsible for de novo DNA methylation in plants. In Arabidopsis, however, mutations in RdDM genes cause no visible developmental defects, which raising the question of the biological significance of RdDM in plant development. Here, we isolated and cloned Five Elements Mountain 1 (FEM1), which encodes an RNA-dependent RNA polymerase. Mutation in FEM1 substantially decreased genome-wide CHH methylation levels and abolished the accumulation of 24-nt small interfering RNAs. Moreover, male and female reproductive development was disturbed, which led to the sterility of fem1 mutants. In wild-type (WT) plants but not in fem1 mutants, genome-wide CHH DNA methylation levels were greater in panicles, stamens, and pistils than in seedlings. The global increase of methylation in reproductive organs of the WT was attributed to enhancement of RdDM activity including FEM1 activity. More than half of all encoding genes in the rice genome overlapped with hypermethylated regions in the sexual organs of the WT, and many of them appear to be directly regulated by an increase in DNA methylation. Our results demonstrate that a global increase of DNA methylation through enhancement of RdDM activity in reproductive organs ensures sexual reproduction of rice.

overlapped with hypermethylated regions in the sexual organs of the WT, and many of them 21 appear to be directly regulated by an increase in DNA methylation.

22
Our results demonstrate that a global increase of DNA methylation through enhancement of 23 RdDM activity in reproductive organs ensures sexual reproduction of rice.  (Kankel et al., 2003;Saze et al., 2003;Woo et al., 2007;Stroud et al., 2013).

63
The mechanism of RdDM is well-understood in Arabidopsis, but because most Arabidopsis 8 increased expression levels (> 20-fold) compared to GAS were further selected and named five 116 elements mountain (fem) plants (Figure 1A;Supplemental Figures 1D and 1E). We isolated about 117 200 fem mutants, two of which (#45 and #30) were referred to as fem1-1 and fem1-2 because 118 subsequent, separate map-based cloning demonstrated that they were allelic to each other (see 119 below).

139
To overcome these problems, we knocked out FEM1 using CRISPR/Cas9 technology at two 144 alleles (fem1-3, fem1-4, and fem1-5) generated by sgRNA1 caused a premature stop codon and 145 led to a truncated protein without an intact RdRP domain, indicating that the three fem1 alleles 146 might be functional null mutants (Supplemental Figure 2B). The fem1-6 mutant generated by 147 sgRNA2 contained 49 amino acid (aa) mutations with a 12-aa deletion on its C-terminal 148 (Supplemental Figure 2B). Like the OsDCL3a and OsNRPD1 knock-down rice plants (Wei et al., 149 2014;Xu et al., 2020), fem1 mutants were dwarfs, and their flag leave angle was much larger than 150 that of WT plants with an empty vector (Supplemental Figure 2C). Interestingly, the anthers of 151 the fem1 mutants were smaller and paler than those of control plants ( Figure 1B). Furthermore, 152 iodine potassium iodide (I 2 -KI) staining showed that the pollen viability was dramatically 153 decreased in fem1 mutants (Figures 1C and 1D). Electron microscopy showed that the pollen understand the defect in anther development, we analyzed transverse sections of WT and fem1 156 anthers. At stage 6, when anther morphogenesis is complete, the WT anthers formed four somatic 157 layers (the epidermis, endothecium, middle layer, and tapetum), and the microsporocytes were 158 located at the center of each anther locule (Supplemental Figure 2D). In fem1 mutants, however, a 159 portion of the anther lobes was arrested at stage 6 and failed to form the organized sporophytic 160 cell layers, with the exception of the epidermis, indicating that FEM1 is essential for cell 161 differentiation during early anther development in rice (Supplemental Figure 2D). The anther 162 defects were much weaker in fem1-6 than in the other three fem1 mutants ( Figures 1B to 1E),

164
In addition, stigmas of each gynoecium in fem1 lacked one or two hairbrushes ( Figures

171
To elucidate the role of FEM1 in the methylome, we conducted whole-genome bisulfite 172 sequencing (WGBS) for the panicles of fem1-3 and fem1-6 with the corresponding controls 173 created in the same regeneration process. The high quality of the methylome (Supplemental Data 11 CHH, i.e., the CHH methylation level dropped from 7.0% in WT replicate 1 to 2.9% in fem1-3, 176 and from 5.1% in WT replicate 2 to 3.9% in fem1-6 (Supplemental Data Set 1). For genes, CHH 177 methylation levels were higher for upstream and downstream regions in the WT than for the gene 178 body (Figure 2A). The high CHH methylation levels on the borders of genes were dramatically 179 reduced in fem1-6 and nearly eliminated in fem1-3 (Figure 2A), suggesting that CHH methylation 180 upstream and downstream of genes was dependent on FEM1. As was the case for genes, CHH 181 methylation but not CG or CHG methylation on TEs was also dependent on FEM1 ( Figure 2B).

182
CHH methylation depends on FEM1 not only in panicles but also in seedlings (Supplemental 183 Figure 3A and 3B). To determine which kind of TEs were dominant targets of FEM1, we divided 184 TEs into six subgroups based on length. On short TEs (< 500 bp), the CHH methylation level was 185 as high as 30% in the WT, but was 10% in the fem1-3 mutant ( Figure 2C). On long TEs (> 500 186 bp), however, the methylation level in the WT was very low, and the difference between the WT 187 and fem1 was small ( Figure 2C). That FEM1 mainly regulated DNA methylation on short TEs 188 was also apparent in seedlings (Supplemental Figure 3C).

203
Increased DNA methylation in panicles depends on FEM1

204
We also conducted WGBS for fem1-6 and WT seedlings (Supplemental Data Set 1). As is the 205 case in panicles, the CHH hypo-DMRs are the main type of DMRs in fem1-6 seedlings 206 (Supplemental Figures 3G and 3H;Supplemental Data Set 2), and the DMRs in seedlings are 207 mainly located on TEs (Supplemental Figure 3I). The number of 24-nt siRNAs in fem1-6 was 208 greatly decreased globally on the CHH hypo-DMRs in seedlings (Supplemental Figures 3J to 3L), 209 which was consistent with the features in panicles. However, the obvious difference between the 210 number and length of hypo-DMRs in fem1-6 panicles vs. fem1-6 seedlings indicated that the DNA 211 methylation level might differ among organs (Supplemental Figures 3D,3E,3G and 3H). To 212 understand this difference, we directly compared the GML in the WT panicles and seedlings, and 213 found that CHH methylation levels in panicles were 5.1% and 7.0%, which were substantially 214 higher than the 3.1% in seedlings (Supplemental Data Set 1). On both genes and TEs, the CHH 13 methylation level but not the CG or CHG methylation level was dramatically higher in WT 216 panicles than in WT seedlings ( Figures 3A and 3B). The increase in CHH methylation level on 217 transposons mainly occurred on short TEs ( Figure 3C). In contrast, the CG and CHG levels in 218 WT panicles and WT seedlings were similar on TEs of various lengths (Supplemental Figures 4A   219 and 4B). CHH hyper-DMRs were the main type of DMRs between panicles and seedlings 220 (Supplemental Figures 4C and 4D, the CHH hyper-DMRs were named "panicle hyper-DMRs"), 221 and these DMRs were also mainly located on TEs (Supplemental Figure 4E). On the panicle 222 hyper-DMRs, the CG methylation level was similar in WT panicles and WT seedlings 223 (Supplemental Figure 4F). The CHH level was substantially higher and the CHG level was 224 significantly higher in WT panicles than in WT seedlings (Supplemental Figure 4F). However, the 225 24-nt siRNA abundance on the panicle hyper-DMRs was not increased but was significantly 226 lower in WT panicles than in WT seedlings (Supplemental Figure 4F), suggesting that siRNA 227 levels were not responsible for the increase in DNA methylation.

228
The increased methylation on panicle hyper-DMRs and on the FEM1-dependent DMRs in 229 both panicles and seedlings occurred mainly on short TEs and in CHH context, suggesting that

244
To determine whether the increased methylation in panicles involves an increase in 245 methylation of the reproductive organs, we performed WGBS of stamens and pistils of the WT, 246 fem1-3, and fem1-6 (Supplemental Data Set 1). On a genome-wide scale, the CHH methylation 247 level was 5.5% in WT stamens and 4.7% in WT pistils, but was only 3.1% in WT seedlings 248 (Supplemental Data Set 1). The CHH methylation level was higher on both genes and TEs in 249 stamens than in seedlings ( Figures 4A and 4B). On transposons, CHH methylation level in 250 stamens mainly occurred on short TEs (Supplemental Figure 5A). There were 98,434 CHH 251 hyper-DMRs between stamens and seedlings (Supplemental Data Set 3), which covered 17.8 Mb

252
(Supplemental Figures 5B and 5C), and among the stamen hyper-DMRs, 59.4% were located on 253 TEs (Supplemental Figure 5D). Mutation in FEM1 produced 162,809 CHH hypo-DMRs in 15 one-tenth of the rice genome (Supplemental Figure 5F), and 58.6% of the CHH DMRs were on 256 TEs (Supplemental Figure 5G). About 93.1% of the stamen hyper-DMRs overlapped with 257 FEM1-dependent CHH methylation regions in stamens ( Figure 4C). In the overlapped regions, 258 the CHH methylation level was higher in WT stamens than in WT seedlings, and mutation in 259 FEM1 almost eliminated CHH methylation in both seedlings and stamens ( Figure 4C). Like the 260 overlapped DMRs, the CHH levels on the specific FEM1-dependent DMRs were higher in WT 261 stamens than in WT seedlings ( Figure 4C), suggesting that the increased methylation in stamens 262 was fully FEM1-dependent. However, siRNA abundance was not higher in WT stamens than in 263 WT seedlings on those regions ( Figure 4C), indicating an inconsistency between siRNA 264 abundance and methylation levels.

265
As was the case for panicles and stamens, CHH levels on genes and TEs in WT pistils were 266 higher than the levels on corresponding genes and TEs in WT seedlings ( Figures 4D and 4E), and 267 the increase in methylation occurred on short TEs (Supplemental Figure 5H). We identified a total 58.6% of them were located on TEs (Supplemental Figure 5N). Among the pistil hyper-DMRs, 274 98.0% overlapped with hypo-DMRs in fem1-3 pistils ( Figure 4F), suggesting that the increase in CHH methylation was largely FEM1-dependent. On the specific FEM1-dependent DMRs, the 276 CHH methylation level was much higher in WT pistils than in WT seedlings ( Figure 4F),

277
suggesting that the increase in methylation might occur on all FEM1 targets in pistils. The 24-nt 278 siRNA levels, however, were not increased on all FEM1-dependent DMRs ( Figure 4F).

281
However, the three types of DMRs largely overlapped with each other (Supplemental Figure 5O).

282
On seven groups of DMRs, the CHH methylation levels were higher in panicles, stamens, and 283 pistils than in seedlings, and the increase in CHH methylation was reversed in fem1 mutants of 284 various organs (Supplemental Figure 5P), suggesting that methylation reinforcement largely 285 occurred on common regions in panicles, stamens, and pistils, and that this reinforcement was 286 FEM1-dependent.

288
To determine whether the global increase of DNA methylation regulates gene expression in 289 WT panicles, we checked genes near CHH hyper-DMRs (DMRs overlapped with genes from -2 290 kb in the promoter through 500 bp in the terminator). We found that panicle CHH hyper-DMRs 291 overlapped with 65.6% of all rice genes (36,708 genes, Figure 5A). To investigate the effects of 292 CHH methylation on gene expression, we performed mRNA-seq for WT and fem1-6 seedlings   Figure 6B). The genes were expressed at differential levels in panicles vs.

307
We also performed mRNA-seq for stamens (stage 12) of the WT, fem1-3, and fem1-6 308 (Supplemental Data Set 1, and Supplemental Data Set 5). Stamen CHH hyper-DMRs overlapped 309 with 62.4% of all rice genes (34,939 genes, Figure 5E). Among the 34,939 genes near DMRs, 310 11,744 exhibited differential expression between stamens and seedlings ( Figure 5F). Of the higher in WT stamens than in WT seedlings, and the methylation level was reduced in fem1 318 stamens (Supplemental Figure 6D). Consistent with the latter findings, the transcriptional levels 319 of DTC1 and OsGAMyb were greater in WT stamens than in WT seedlings, and the differences in 320 their transcriptional levels were even greater between fem1 stamens and WT seedlings 321 (Supplemental Figure 6D), indicating that their proper expression depends on their 322 hypermethylation.

323
Stage FG8 pistils of the WT, fem1-3, and fem1-6 were also subjected to mRNA-seq analysis   Table 1) in seedlings, panicles, pistils, and stamens of the WT. The genes 344 necessary for 24-nt siRNA biosynthesis, including FEM1, OsNRPD1, OsSHH1, CLSYs, and 345 OsDCL3, were upregulated in panicles relative to seedlings, and their transcript level was even 346 greater in pistils (Supplemental Figure 7A). Among those genes, however, only CLSY-like 1 347 expression was increased in stamens (Supplemental Figure 7A). Except the genes related to 348 siRNA biosynthesis mentioned above, many of the RdDM pathway genes involved in the 349 methylation of DNA methylation and heterochromatin formation, including OsNRPE1b, OsAGO4,

350
OsDMS3, OsDRM2, and MORC6-like 1, were upregulated in panicles and pistils (Supplemental 351 Figure 7A). To confirm that most of RdDM genes were upregulated in the sexual organs of rice,

352
we searched the transcriptomes previously reported for rice seedlings, leaves, shoots, panicles, 353 stamens, and pistils (http://rice.plantbiology.msu.edu/expression.shtml). The search indicated that 354 the expression levels of most RdDM genes were higher in panicles and pistils than in leaves, 20 seedlings, or shoots (Supplemental Figure 7B). Moreover, the expression levels of those genes 356 were much higher in young panicles than in mature panicles (Supplemental Figure

364
To determine the molecular mechanism underlying the upregulation of RdDM genes in sexual 365 organs of rice, we examined the DNA methylation levels on 46 DNA methylation-related genes in 366 different organs (Supplemental Table 1), and found that the majority of those genes overlapped 367 with three types of hyper-DMRs (40 genes with panicle hyper-DMRs, 41 genes with stamen 368 hyper-DMRs, and 35 genes with pistil hyper-DMRs, Figure 6B). For example, DNA methylation 369 levels on FEM1, OsDRM2, and OsNRPD1b were greater in panicles, stamens, and pistils than in 370 seedlings (Supplemental Figure 7C; Figure 6C). The high methylation levels on those regions 371 were reduced in panicles, stamen, and pistils of fem1 mutants (Supplemental Figure 7C; Figure   372 6C), suggesting that the increase in the methylation of those genes was FEM1-dependent. The

390
In the current investigation of rice, a monocotyledonous plant that diverged 150 millions years 391 ago and is more representative of higher plants than Arabidopsis is in terms of genome size and 392 TE content, we found that mutation in FEM1 greatly reduced de novo DNA methylation and 393 caused total sterility because of developmental defects in both male and female reproductive 394 organs. We demonstrated that the upregulation of RdDM activity including FEM1 activity is