A natural tandem array alleviates epigenetic repression of IPA1 and leads to superior yielding rice

Abstract

Super hybrid rice varieties with ideal plant architecture (IPA) have been critical in enhancing food security worldwide. However, the molecular mechanisms underlying their improved yield remain unclear. Here, we report the identification of a QTL, qWS8/ipa1-2D, in the super rice Yongyou12 (YY12) and related varieties. In-depth genetic molecular characterization of qWS8/ipa1-2D reveals that this newly identified QTL results from three distal naturally occurring tandem repeats upstream of IPA1, a key gene/locus previously shown to shape rice ideal plant architecture and greatly enhance grain yield. The qWS8/ipa1-2D locus is associated with reduced DNA methylation and a more open chromatin state at the IPA1 promoter, thus alleviating the epigenetic repression of IPA1 mediated by nearby heterochromatin. Our findings reveal that IPA traits can be fine-tuned by manipulating IPA1 expression and that an optimal IPA1 expression/dose may lead to an ideal yield, demonstrating a practical approach to efficiently design elite super rice varieties.

Figure 1. Map-based cloning of qWS8/ipa1-2D.

(a) Morphology of YYP1 and NIP. Scale bars, 20 cm. (b) Stems of YYP1 and NIP. Scale bar, 10 cm. (c) Panicles of YYP1 and NIP. Scale bar, 10 cm. (d) Coarse linkage mapping of qWS8/ipa1-2D. (e) High resolution mapping of qWS8/ipa1-2D to a ∼3 kb region of the NIP genome, which is denoted by double headed arrow. Number of recombinants between molecular markers and qWS8/ipa1-2D is indicated. The position of IPA1 is denoted by the single headed arrow. Schematic map of four key recombinants delimiting the mapping region for detailed progeny traits analysis is presented by different bars. Grey bars refer to the heterozygous allele, black and white bars to YYP1 and NIP homozygous alleles, respectively. (f–h) Trait comparison of stem diameter (f), panicle primary branch number (g) and tiller number (h) using F₄ sibling lines derived from corresponding recombinants in e. Blue and red columns indicate alleles of YYP1 and NIP, respectively. Values are means±s.d. (n=24). **P<0.01 or ***P<0.001, by Student's t-test.

Figure 2. Identification of large tandem repeats underlying qWS8/ipa1-2D for the IPA traits.

(a) Sequence polymorphisms of the region covering qWS8/ipa1-2D and IPA1 between YYP1 and NIP. Positions of all the SNPs identified were labelled as minus number from ATG (0 position) of IPA1. YYP1 bears three copies of the exact NIP sequence (3,137 bp in length) except for three SNPs. The primers only amplifying the repeats were labelled with blue arrows, and primers fail to detect the repeat structure were labelled with purple arrows. (b) Genotyping of the three varieties with identical YYP1 SNPs by Southern blotting with XbaI (top) and repeat-specific PCR (bottom) using primers indicated in a. XG-B, Xiangai B; G77-4, GENG77-4; JXB, Jinxibai. Note that JXB does not contain the triple repeats. (c–e) Phenotypic evaluation of stem diameter (c), panicle primary branch number (d) and tiller number (e) of five varieties including YYP1 and NIP. Values are means±s.d. (n=20). Different letters at top of each column indicate a significant difference at P<0.05 determined by Tukey's HSD test.

Figure 3. Association of IPA1 and microRNA expression with changes in IM development.

(a) Relative expression of IPA1 in NIL^ipa1-2D and NIL^IPA1, normalized to the rice Actin gene, in various organs including seedlings (S), root (R), mature leaf (L), different stages of IM (IM1–IM6), young panicle (YP) and flowering panicle (P). Values are means±s.d. (n=3). (b,c) Relative expression of miRNA156 (b) and miRNA529 (c) in various organs of NILs, normalized to rice 5 S rRNA. Values are means±s.d. (n=3). (d,e) Comparison of IM circumference (d) and area (e) between NILs at the IM1 stage in a. Values are means±s.d. (n=8). Triple asterisks represent significant difference between NILs determined by the Student's t-test at P<0.001. (f,g) Scanning electron microscope images of IMs in NIL^ipa1-2D (f) and NIL^IPA1 (g) at the IM1 stage in a. Scale bars, 50 μm. Dotted lines indicate the regions measured in (d,e). (h,i) Scanning electron microscope images of IMs in NIL^ipa1-2D (h) and NIL^IPA1 (i) at the IM3 stage in a. Scale bars, 100 μm.

Figure 4. Alleviation of epigenetic repression at the IPA1 promoter by qWS8/ipa1-2D.

(a) Schematic map showing the sites of two methylation sensitive enzymes (HpaII and MspI) and probe for methylation detection by Southern blotting. The blue and orange double arrows indicate the mapping region and single repeat region respectively. Mn1–Mn4, positions of PCR detection after MNase digestion. (b) Southern blot analysis of DNA methylation differences between NIL^ipa1-2D and NIL^IPA1. The bands reflecting different methylation pattern are denoted by arrows. Note that the smallest band is a direct reflection of different methylation at sites −2,545 and −2,974. M, DNA markers. (c) Distribution of three cytosine contexts and methylation pattern in ∼800-bp promoter region of IPA1. Filled circles, methylated cytosine; empty circles, unmethylated cytosine. Two distinct regions are labelled by blue lines and the junction region is labelled by black frame. Note that the junction region presents obvious methylation difference especially for CHH and overlaps with the DH site labelled by black double arrows. (d–f) Methylation levels of CG (d), CHG (e) and CHH (f) in eight 100 bp-windows of the promoter region shown in c. Blue, NIL^ipa1-2D; Red, NIL^IPA1. Dark yellow shades highlight the junction region. (g) Schematic map of the approach to detect open chromatin. Chromatin with loose nucleosome occupation is more susceptible to mononuclease (MNase) digestion and inefficiently amplified by PCR. (h) Sensitivity of four regions (Mn1–Mn4) to increasing dosage of MNase digestion between NIL^IPA1 and NIL^ipa1-2D. The position of each region is labelled in a. Note that the Mn3 region was more sensitive to the digestion in NIL^ipa1-2D. (i) Ratio of the NIL^ipa1-2D allele in the Mn3 region after MNase digestion of nucleus from heterozygous NIL. Two alleles are distinguished by the SNP at position −419 shown in a.

Figure 5. IPA1 dosage affects IPA characteristics.

(a) Morphology of plants carrying different IPA1 gDNA copy number. 1C, 2C, 3C, and 4C represent plants with 1, 2, 3 and 4 insertions of the transgenic construct, compared with the wild-type (WT) NIP. Scale bar, 15 cm. (b) Copy number detection of four lines by Southern blotting with WT as control. A probe from hygromycin phosphotransferase gene was used to detect the transgene. M, DNA markers. (c–e) IPA traits of plants with increasing IPA1 copy number, including tiller number (c), stem diameter (d) and panicle primary branch diameter (e). Values are means±s.d. (n=12). Different letters at top of each column indicate a significant difference at P<0.05 determined by Tukey's HSD test. (f–h) Plots of relative IPA1 expression normalized to rice Actin with trait performance in different lines, including tiller number (f), stem diameter (g) and panicle primary branch number (h). Curves fitting the trait change are calculated by quadratic equation with R² values.

Figure 6. Effects of qWS8/ipa1-2D allele in reaching optimal yield potential.

(a) Overall morphology of three genotypes in the NIL population. Scale bar, 20 cm. (b) Panicle morphology of three genotypes. Scale bar, 6 cm. (c) Relative expression of IPA1 in three genotypes, normalized to the rice Actin gene. Values are means±s.d. (n=3). (d–i) Yield related trait performance of three alleles in the Shanghai and Hainan experimental stations, including panicle primary branch number (d), panicle secondary branch number (e), spikelet number per panicle (f), tiller number (g), grain weight (h) and yield per plant (i). The homozygous ipa1-2D and IPA1 alleles are denoted by black and white columns and heterozygous allele is denoted by grey columns respectively. Values are means±s.d. (n=67, 124, 62 in Shanghai and 39, 78, 42 in Hainan for three genotypes). Different letters at top of each column indicate a significant difference among genotypes in the respective locations at P<0.05 determined by Tukey's HSD test.

Figure 7. De novo design of super hybrid rice using ipa1-2D and ipa1-1D.

(a) Overall morphology of newly developed hybrid rice JYZK-4 (ipa1-2D) and JYZK-6 (ipa1-1D). Scale bar, 20 cm. Note that JYZK-4 has more productive tillers than JYZK-6. (b) Panicle morphology of JYZK-4 and JYZK-6. Scale bar, 10 cm. (c,d) Contribution of ipa1-2D (c) and ipa1-1D (d) alleles to tiller number in the F₂ population of the two hybrids as showed by box plot. The centre line, box limits and whiskers represent mean, 25% and 75% confidence limits, and min and max values, respectively. Different letters at top of each box indicate a significant difference among genotypes at P<0.05 determined by Tukey's HSD test. R² calculated by one-way ANOVA showed the percentage of tiller variation explained by three genotypes in each population, indicating that ipa1-2D has a less effect on tillering than ipa1-1D. (e,f) Chromatogram of IPA1 transcript abundance of ipa1-2D/ipa1-1D plants recovered from RT-PCR with primers flanking the miRNA cleavage site in YP (e) and seedling (f). Arrows indicate the SNP discriminating the two alleles. (g) Relative expression of IPA1 in three genotypes derived from the ipa1-2D/ipa1-1D plant in YP, normalized to rice Actin. Values are means±s.d. (n=3). (h) Model of yield potential shaped by the combination of IPA1, ipa1-2D and ipa1-1D alleles. The model predicts that the higher yield can be obtained in plants with modest IPA1 expression, which generates big panicles and moderate tiller numbers.