Non-canonical regulation of SPL transcription factors by a human OTUB1-like deubiquitinase defines a new plant type rice associated with higher grain yield

Abstract

Achieving increased grain productivity has long been the overriding focus of cereal breeding programs. The ideotype approach has been used to improve rice yield potential at the International Rice Research Institute and in China. However, the genetic basis of yield-related traits in rice remains unclear. Here, we show that a major quantitative trait locus, qNPT1, acts through the determination of a 'new plant type' (NPT) architecture characterized by fewer tillers, sturdier culms and larger panicles, and it encodes a deubiquitinating enzyme with homology to human OTUB1. Downregulation of OsOTUB1 enhances meristematic activity, resulting in reduced tiller number, increased grain number, enhanced grain weight and a consequent increase in grain yield in rice. Unlike human OTUB1, OsOTUB1 can cleave both K48- and K63-linked polyubiquitin. OsOTUB1 interacts with the E2 ubiquitin-conjugating protein OsUBC13 and the squamosa promoter-binding protein-like transcription factor OsSPL14. OsOTUB1 and OsSPL14 share common target genes, and their physical interaction limits K63-linked ubiquitination (K63Ub) of OsSPL14, which in turn promotes K48Ub-dependent proteasomal degradation of OsSPL14. Conversely, loss-of-function of OsOTUB1 is correlated with the accumulation of high levels of OsSPL14, resulting in the NPT architecture. We also demonstrated that pyramiding of high-yielding npt1 and dep1-1 alleles provides a new strategy for increasing rice yield potential above what is currently achievable.

Figure 1

Positional cloning of qNPT1. (A) Appearance of a mature plant. RIL52 was selected from a Chunjiang06 (CJ06) × IR66167-27-5-1-6 cross. Scale bar, 20 cm. (B) Panicle morphology. Scale bar, 5 cm. (C-E) Comparison of RIL52 with its parents with respect to (C) grain number, (D) tiller number and (E) culm diameter. Data are shown as the mean ± SEM (n = 30). The presence of the same lowercase letter denotes a non-significant difference between means (P < 0.05). (F) QTL mapping for grain number, tiller number and culm diameter. (G) Positional cloning of qNPT1. The locus was mapped to a ∼ 4.1 Kbp genomic region flanked by P139 and P143. The numbers below the lines indicate the number of recombinants between qNPT1 and an adjacent marker. The candidate gene was predicted to generate two alternative transcripts. The arrowhead indicates the target site designed for CRISPR/Cas9-based genome editing. (H) Sequence variants at the NPT1 locus in both the promoter and coding regions shown in G. The specific nucleotide variants of the npt1 allele are indicated by the pink boxes.

Figure 2

Effect of functional OsOTUB1 on plant architecture and grain yield. (A) Mature plant morphology. Scale bar, 20 cm. (B) Loss-of-function mutations of OsOTUB1 generated via CRISPR/Cas9-based genome editing. The target sequence indicated by the green boxes is located on the reverse strand of OsOTUB1.1 shown in Figure 1G. The osotub1-C1 mutant harbors a 7-bp deletion (in red) that results in a premature stop codon. (C) Heading date. (D) Plant height. (E) Diameter of the uppermost internode. (F) Number of tillers per plant. (G) Panicle length. (H) Number of primary branches per panicle. (I) Number of secondary branches per panicle. (J) Number of grains per panicle. (K) 1 000-grain weight. (L) Overall grain yield per plant. Data are shown as the mean ± SEM (n = 288). All phenotypic data were measured in paddy-grown plants under normal cultivation conditions. The presence of the same lowercase letter denotes a non-significant difference between means (P < 0.05).

Figure 3

Pyramiding of the npt1 and dep1-1 alleles enhances panicle branching and grain yield in rice. (A) Levels of OsOTUB1 transcript present in the organs of NIL plants. R, root; C, culm; LB, leaf blade; LS, leaf sheath; SAM, shoot apical meristem; YP0.2, YP6, YP12: young panicles, with a mean length of 0.2 cm, 6 cm or 12 cm, respectively. Relative expression levels are presented as the relative number of copies per 1 000 copies of rice Actin1. Data are shown as the mean ± SEM (n = 3). (B) Plant morphology. Scale bar, 20 cm. (C) Panicle morphology. Scale bar, 5 cm. (D) Grain morphology. Scale bar, 2 mm. (E-J) A quantitative comparison of the two NILs. (E) Heading date. (F) Plant height. (G) Number of tillers per plant. (H) Number of secondary branches per panicle. (I) Number of grains per panicle. (J) 1 000 grain weight. (K) An image of the shoot apical meristem. Scale bar, 50 μm. (L) Scanning electron microscopy image of a culm. Scale bar, 25 μm. (M) Culm vascular system. Scale bar, 500 μm. (N), Total number of large and small vascular bundles shown in M. (O) Overall grain yield per plant. Data are shown as the mean ± SEM (n = 288). All phenotypic data were measured from paddy-grown plants under normal cultivation conditions. Student's t-test was used to generate the P values.

Figure 4

Phenotypes of transgenic ZH11 plants over-expressing OsOTUB1.1. (A) Two independent transgenic lines showed reduced tiller number and dwarfism. Scale bar, 10 cm. (B) Panicle size was reduced. Scale bar, 5 cm. (C) Leaves suffered from necrosis. Scale bar, 3 mm. (D) Apoptosis was induced in the flag leaves, assayed via Evans Blue staining. Scale bar, 3 mm. (E) Abundance of OsOTUB1.1 transcript in the young panicle. Transcription relative to the level in ZH11 plants (set to one). Data are shown as the mean ± SEM (n = 3). (F) Plant height. (G) Number of tillers per plant. (H) Number of grains per panicle. Data are shown as the mean ± SEM (n = 60). All phenotypic data were measured in paddy-grown rice plants under normal cultivation conditions. The presence of the same lowercase letter denotes a non-significant difference between means (P < 0.05, panels F to H).

Figure 5

OsOTUB1 displayed cleavage activity for K48- and K63-linked ubiquitin tetramers. Cleavage activity towards K48- and K63-linked ubiquitin tetramers (Tetra-ub) was analysed using OTUB1, His-OsOTUB1.1 or OsOTUB1.2. The inputs (Ub₄) and their cleavage products (trimers (Ub₃), dimers (Ub₂) and monomers (Ub₁)) are labelled on the left. The products were visualised via western blotting using an anti-ubiquitin antibody.

Figure 6

The interaction between OsOTUB1 and OsUBC13 proteins regulates plant architecture. (A) Yeast two-hybrid assays. (B) Pull-down assays using recombinant GST-OsOTUB1 and His-OsUBC13. (C) BiFC assays in rice protoplasts. Scale bar, 10 μm. (D) Morphology of transgenic ZH11 plants. Scale bar, 20 cm. (E) Cross-section of the uppermost internodes. Scale bar, 500 μm. (F) Effect of OsUBC13 on panicle branching. Scale bar, 5 cm. (G) Grain size and shape. Scale bar, 2 mm. (H) Abundance of the OsOTUB1 transcript in young panicles relative to the level in ZH11. Data are shown as the mean ± SEM (n = 3). (I) Number of tillers per plant. (J) Number of grains per panicle. (K) 1 000-grain weight. (L) Diameter of the uppermost internode. Data are shown as the mean ± SEM (n = 30). All phenotypic data were measured in paddy-grown rice plants under normal cultivation conditions. The presence of the same lowercase letter denotes a non-significant difference between means (P < 0.05, panels I to L).

Figure 7

The OsOTUB1-OsSPL14 interaction controls plant architecture. (A) BiFC assays. The N-terminus of YFP-tagged OsSPL14, the SBP domain or a deleted version of OsSPL14 was co-transformed into rice protoplasts along with the C-terminus of YFP-tagged OsOTUB1.1. Panels (from left to right), DAPI staining, YFP signal, differential interference contrast image, merged channels. Scale bar, 10 μm. (B) Co-immunoprecipitation of OsOTUB1.1-GFP and OsSPL14. IB, Immunoblot; IP, immunoprecipitation. (C) Plant morphology. Scale bar, 20 cm. (D) Panicle morphology. Scale bar, 5 cm. (E) OsSPL14 transcript abundance. Transcription relative to the level in ZH11 plants was set to one. Data are shown as the mean ± SEM (n = 3). (F) Number of tillers per plant. (G) Number of grains per panicle. (H) Culm diameter. All phenotypic data were measured in field-grown plants under normal cultivation conditions. Data in F-H are shown as the mean ± SEM (n = 120). The presence of the same lowercase letter denotes a non-significant difference between means (P < 0.05).

Figure 8

OsOTUB1 and OsSPL14 antagonistically regulate common target genes. (A) Number and overlap of OsSPL14-activated and OsOTUB1-repressed target genes. RNA-seq was performed using young panicles (< 0.2 cm in length) of the NIL plants. (B) The abundances of OsSPL14-regulated genes examined in young panicles relative to the levels in ZH11. Data are shown as the mean ± SEM (n = 3).

Figure 9

OsOTUB1 promotes the degradation of OsSPL14. (A) Accumulation of OsSPL14 in ZH11 and ZH11-npt1 plants. The abundance of the HSP90 protein was used as a loading control. (B) Treatment with the proteasome inhibitor MG132 stabilizes OsSPL14. Total protein was extracted from young panicles (< 0.2 cm in length) of ZH11 plants exposed to either 0 or 50 μM MG132. The immunoblot was probed with either anti-OsSPL14 or anti-HSP90 antibodies. (C) OsOTUB1 destabilizes OsSPL14. The lysates from young panicles of ZH11 and ZH11-npt1 plants were co-incubated with GST-OsSPL14 in the presence or absence of His-OsOTUB1. The lysates were harvested at various times and immunoblotted to assess the accumulation of OsSPL14 and HSP90. (D) Ubiquitination of OsSPL14. The protein extracts from young panicles were immunoprecipitated using an anti-Myc antibody, then analysed using anti-ubiquitin, anti-K48-linked ubiquitin or anti-K63-linked ubiquitin chain conjugates. (E) Flag-OsSPL14 can be modified via K48-ubiquitin linkage. Rice protoplasts were co-transfected with Flag-OsSPL14 and HA-ubiquitin (either HA-tagged WT or K48R ubiquitins), and the ubiquitinated forms of Flag-OsSPL14 were immunoprecipitated using an anti-Flag antibody and then analysed using an anti-HA antibody. (F) The K63-linked ubiquitination of OsSPL14 is regulated by OsOTUB1. Rice protoplasts were co-transfected with Flag-OsSPL14 and HA-ubiquitin (either HA-tagged WT, K48R, K63R, K48O or K63O ubiquitins) in the presence or absence of Myc-OsOTUB1; lysates were then harvested and immunoblotted to assess the accumulation of OsSPL14 and analysed for ubiquitinated forms of Flag-OsSPL14, as described in E.