OsDPE 2 can rescue panicle mutant phenotype of lax1-6 in Dular
We identified a mutant allele of LAX1, lax1-6, in rice (Oryza sativa L. subsp. japonica cv. ZH11) with defective lateral branch formation, by screening the lax panicle mutants from the T-DNA mutant library (Fig. S1). We found that the T-DNA insertion sites existed in 4905bp upstream of the LAX1 start codon of lax1-6 (Fig. S1a). LAX1 gene expression analysis in 2 mm young panicles of WT and lax1-6 showed that lax1-6 had a significantly reduced expression level of LAX1 than WT (Fig. S1c). To corroborate this observation, we derived the experimental rice lines from a cross-population between two parental rice lines (lax1-6 and Dular) to determine whether other loci, besides the LAX1 gene, are involved in the panicle morphogenesis of indica rice. Hybrid F2 populations were generated using lax1-6 and Dular as the parental lines. We screened out the mutant plants to observe and analyze their panicle phenotype after identifying their lax1-6 locus. After the analysis, 30 plants with the lax phenotype (osdpe2-d) and 83 plants with like wild-type phenotype (osdpe2-Dd or osdpe2-D) were selected (Fig. S2a). The ration of secondary branches/primary branches of lax1-6 genotypes was investigated using the separation ratio analysis, and the results showed that the OsDPE2 locus was consistent with the dominant inheritance of a single locus (Fig. S2b). After analyzing the panicle phenotype of lax1-6×Dular F2 plants, those with the lax1-6 mutant genotype were divided into osdpe2-d (lax1-6 type), osdpe2-Dd (like-wildtype), and osdpe2-D (wild-type), based on the ratio of secondary panicle branches/ primary panicle branches (Fig. 1a).
Two SSR markers (RM2201 and RM1301), two InDel markers (In1 and In2), and five SNP markers were developed for primary mapping and fine mapping of the lax1-6 locus (Table S2). Bulked separate analysis was used to map OsDPE2 between RM2201 and RM1301 on chromosome 7. Thereafter, 1600 F2 individuals screened for the lax1-6 mutant genotype were used to fine-map OsDPE2. We identified two recombinants between S5 and S8, and the mapping area was reduced to about 6 kb and contained only one candidate gene (LOC_07g46790) (Fig. 1b). Through genotypic identification and phenotypic observation of the recombinants identified between S5 and S8 using the progeny test, we identified the lax panicle phenotype in the progenies of OsDPE2-d1 and OsDPE2-d2. This showed that OsDPE2-d1 and OsDPE2-d2 recombinants were lax1-6 homozygotes of OsDPE2. Moreover, OsDPE2-D1, OsDPE2-D2, OsDPE2-D3, and OsDPE2-D4 recombinants of OsDPE2 were Dular homozygotes because their progenies displayed wild-type panicle phenotypes (Fig. S3).
Furthermore, we performed a parental comparative sequencing test for Dular, ZH11, and lax1-6 using S4 and S10 mapping regions to identify the OsDPE2 gene. The sequence analysis results showed nine SNPs in the S4 and S10 mapping regions of Dular, ZH11, and lax1-6 (Fig. 1c). Among these, two non-synonymous SNPs, which can alter the amino acid coding sequence, were detected in the exon region of the genome of rice at the 4th and 992nd positions of LOC_Os07g46790 ORF.
To verify the OsDPE2 candidate gene, we generated complementary transgenic plants in the lax1-6 background, denoted OsDPE2-Com. A Zhenshan 97 BAC was used to obtain 16kb fragments containing OsDPE2, exhibiting no genomic sequence differences with Dular. Fragments containing OsDPE2 were then fused with pCAMBIA2301 using one-step ligation. The progenies of the OsDPE2-Com T1 in the lax1-6 background were subjected to phenotypic analysis and positive transgenic detection (Fig. S4). Compared with lax1-6, OsDPE2-Com presented a significantly increased number of secondary panicle branches closer to those of the wild-type (Fig. 1d ~ g). Nevertheless, OsDPE2-Com exhibited a partially reductive phenotype, in which a portion of primary and secondary panicle branches retained a certain level of lax1-6. Thus, the candidate gene LOC_Os07g46790 was OsDPE2.
OsDPE2 encodes rice Disproportionating Enzyme 2 located in the cytoplasm
OsDPE2 encodes a rice DPE2, whose protein sequence was obtained from the Rice Genome Annotation Project (RGAP). We analyzed the protein sequence with SMART software and found that OsDPE2 has two carbohydrate-binding modules (CBM-2) and a glycoside hydrolase family 77 (Glyco-hydro-77). To examine the evolutionary characteristics of the DPE2 protein family, we conducted phylogenetic analyses of DPE2 protein sequences from rice, Arabidopsis, soybean, tomato, sorghum, black cottonwood, grapes, purple false brome, moss, and Chlamydomonas (Fig. 2a). The results showed that OsDPE2 has no homologous proteins in rice.
According to a prediction by TAIR, OsDPE2 protein may chiefly be located in the cytoplasm. To verify this, we transfected theOsDPE2 coding sequence fused with a Green fluorescent protein (GFP) into rice protoplasts gathered from etiolated seedlings. We found that OsDPE2::GFP proteins pointed were located in the cytoplasm (Fig. 2b), suggesting that OsDPE2 protein chiefly localizes in the cytoplasm. To understand the spatiotemporal expression of OsDPE2 among Dular, lax1-6 and ZH11, we analyzed its expression pattern in the mature leaf (ML), young leaf (YL), stem (ST), root (RO), and in several developmental stages of young panicles (1 mm (P1), 2 mm (P2), 3–5 mm (P3), and > 10 mm (P4) young panicles) via RT-qPCR (Fig. 2c). OsDPE2 was constitutively expressed in Dular, and its highest expression was detected in the leaves and 2 mm panicles of ZH11 and lax1-6 background.
Furthermore, OsDPE2-specific antisense chain probes were designed and used for in situ hybridization of the paraffin sections of leaves and young panicles at various stages of ZH11 background samples. We observed hybridization signals in mesophyll cells and the outer cells at the initial stage of inflorescence meristem (P1). Moreover, the hybridization signals were also found in the outer cells at the initial stage of the primary branch meristem (P2). The initial stage of the secondary branch meristem (P2) had hybridization signals in the outer cells of the secondary branch meristem and spikelet meristem. Additionally, the initial stage of spikelet meristem (P3) exhibited hybridization signals in the outer cells of anther meristem, while the spikelet maturation stage (P4) had the hybridization signals in the outer cells of the anthers and glume (Fig. 2d).
To understand the role of OsDPE2 in rice, we generated OsDPE2 gene-editing transgenic plants in Dular by CRISPR; however, upon screening, the T0 plants tested negative for the osdpe2Dular(m). To determine the characteristics of these plants, we screened osdpe2Dular(H) plants for plant and panicle yield characteristics. The results revealed that osdpe2Dular(H) plants exhibited short panicles, infertility, and dwarfism (Fig. S5). Moreover, we screened 12 allelic mutants from transgenic ZH11 plants containing the Cas9 target site sequence. The edited amino acid sequences of the 12 allelic mutants (Fig. 2e) were identified via Sanger sequence analysis, and osdpe2#01 was used for subsequent functional analysis in ZH11.
OsDPE2 affects vegetative plant development of rice by DPE2 activity
We speculated that OsDPE2, just like DPE2, may affect the vegetative plant development. To clarify this, we grew WT and osdpe2#01 plants under continuous light (CL) and continuous dark (CD). After analyzing the growth rate of WT and osdpe2#01 under CL and CD conditions, we found that osdpe2#01 was inhibited significantly under CD but had a similar growth rate with WT under CL (Fig. S6a and b). We further analyzed DPE2 activity in the WT plants under CL and CD using an in-gel DPE2 activity assay and OD values of the DPE2 activity. The results revealed that the DPE2 activity level had a transitory increase associated with the continuous decline in CL condition and a continuous increase associated with the transitory decline in CD (Fig. S6c and d).
To verify these results, we transplanted four-leaves stage seedlings of WT and osdpe2#01 under long day (LD, 16 h light/ 8 h dark) and short day (SD, 8 light/ 16 dark) conditions. We then analyzed the growth rate of the plants after 30 days and found that osdpe2#01 was significantly inhibited under SD but exhibited a similar growth rate with WT under LD (Fig. 3a ~ c).
We assumed that OsDPE2 protein forms a heteroglycan-enzyme complex and transfers the heteroglycan to the non-reducing end of maltose. To verify that, we evaluated DPE2 activity using an in-gel DPE2 activity assay and OD values of the DPE2 activity. The extracts for the assays were from mature leaf (ML), young leaf (YL), stem (ST), root (RO), and various panicle sections (1 mm (P1), 2 mm (P2), 3–5 mm (P3). and > 10 mm (P4) young panicles) of WT and osdpe2#01 plants. The results revealed higher DPE2 activity levels in ML, YL, P1, and P2 of WT; however, no activity was observed in osdpe2#01 (Fig. 3d and e). Additionally, 48-hour rhythmic expression analysis of the WT plants under LD and SD conditions was conducted to determine whether OsDPE2 expression was associated with plant growth under light or dark conditions. The results showed the expression level of OsDPE2 had rhythmic oscillation with day and night alternation in LD and SD conditions (Fig. 3f). Further analyses revealed the expression of OsDPE2 exhibited a transitory increase associated with the continuous decline in the light and a continuous increase associated with the transitory decline in the dark. Meanwhile, a 48-hour rhythmic DPE2 activity analysis of WT and osdpe2#01 under LD and SD conditions showed that, similar to OsDPE2 expression, the OsDPE2 activity had a rhythmic oscillation with day and night alternation in LD and SD conditions. It also had a transitory increase associated with the continuous decline in the light and a continuous increase associated with the transitory decline in the dark (Fig. 3g and h). To sum up, OsDPE2 affect the vegetative plant development of rice by DPE2 activity.
OsDPE2 regulates panicle morphogenesis of rice
After analyzing the panicle phenotypic characteristics of the 12 allelic OsDPE2 mutants, we found that the degradation branches significantly increased in OsDPE2 mutants than in the WT plants, and protein variation caused by editing correlated with the number of degenerated branches (Fig. 4a and b, Fig. S7). To determine whether OsDPE2 affects the reproductive growth in rice, we grew 30-day seedlings of WT and osdpe2#01 plants under LD and SD conditions until flowering. After comparing panicle phenotypic characteristics of WT and osdpe2#01 in LD and SD, we found that number of the secondary branches of the panicle and the number of spikelets per panicle reduced in osdpe2#01 plants as the sunlight duration changed from long to short (Fig. 4c). Moreover, plant and panicle yield traits analyses showed that osdpe2#01 had significantly decreased primary branches of the panicle, secondary branches of the panicle, spikelets per panicle, panicle length, and tiller number, and a significant increase in plant height than WT under LD and SD conditions (Fig. 4d ~ i). Further analysis revealed that, except for the plant height, other plant and panicle yield traits of the WT and osdpe2#01 significantly decreased under SD than LD conditions. For osdpe2#01, the number of secondary branches of the panicle, spikelets per panicle, and tiller significantly decreased under SD compared to LD. This indicated that OsDPE2 affected panicle morphogenesis by regulating plant growth and development in the dark. The above results showed that OsDPE2 might play a role of regulating panicle morphogenesis of rice.
OsDPE2 regulates panicle morphogenesis by affecting starch content.
Through iodine staining analysis of starch in various vegetative organs of wild-type rice, we found that ML and YL of osdpe2#01 had stronger staining signals than the wild-type (Fig. S8). Additionally, the starch staining test of ML under CL and CD conditions using iodine revealed that the starch content of osdpe2#01 was significantly higher than WT under CD conditions (Fig. 5a). We conducted a 48-hour rhythmic starch content analysis of the WT and osdpe2#01 plants under LD and SD to determine whether starch content was associated with OsDPE2 activity. The results showed that compared with WT in the plants dark, osdpe2#01 plants had significantly increased starch content (Fig. 5c). Therefore, we postulated that OsDPE2 might engage in the biological pathways of starch breakdown during the vegetative growth of rice.
To establish whether OsDPE2 played a similar role in biological pathways of starch metabolism under the reproductive growth of rice, we performed the starch iodine staining analysis of the various young panicle stages of WT, osdpe2#01, and lax1-6. The results revealed that young panicles of osdpe2#01 and lax1-6 had a stronger staining signal than WT, suggesting that OsDPE2 may be involved in the biological pathways of starch breakdown during the development stages of young rice panicles (Fig. 5c). We conducted a DPE2 activity analysis of various young panicle stages of the WT, osdpe2#01, and lax1-6 plants. The results revealed that, like osdpe2#01, there was no DPE2 activity in the various young panicle stages of lax1-6 (Fig. 5d and e). The above results imply that OsDPE2 regulates panicle morphogenesis of rice by affecting starch content.
Haplotype OsDPE2(AQ) with higher DPE2 activity increase the panicle yield of rice
Rice genome data from RiceHapmap and Ricevarmap databases were used to analyze the haplotypes and their evolution at the OsDPE2 locus in wild rice (430 samples) and cultivated rice (504 samples). The results revealed that the 4th and 992nd SNPs of the OsDPE2 CDS were divided into three haplotypes: OsDPE2 (TA), OsDPE2 (AQ), and OsDPE2 (AA). Among these, OsDPE2 (TA) was mainly distributed in Temperate japonica (55.91%), while OsDPE2 (AQ) was mainly distributed in indica (53.85%), aus (15.22%) and Intermedia rice (50%). OsDPE2 (TA) was mainly found in Temperate japonica (40.36%), Tropical japonica (90.70%), indica (42.86%), aus (84.78%), and intermedia (46.43%) (Fig. 6a, Figure S9a). Furthermore, we used Phyre2 to predict the protein spatial structure of OsDPE2 haplotypes and found that their protein spatial structures differed from one another, possibility because of protein variation (Fig. 6b) (Kelley et al., 2015). We conducted phylogenetic analyses of OsDPE2 in the OsDPE2 haplotypes, wild rice, and other rice germplasm resources using EggNOG v5.0 software to uncover the evolution of OsDPE2 haplotypes. The result revealed that O. rufipogon was closely related to OsDPE2 (AA), OsDPE2 (AQ), and OsDPE2 (TA) (Figure S6b).
To determine whether OsDPE2 haplotypes have functional variations, we performed a starch assay and bound the recombinant proteins to OsDPE2 haplotypes and the CBM2 domain. We found no obvious differences among the OsDPE2-GST haplotypes, and CBM2-1(A) had a higher binding affinity than CBM2-1(T) but lower than that of CBM2-2(Q) (Fig. 6c and Fig. S10). The binding of recombinant proteins to the maltose assay of OsDPE2 haplotypes was also conducted using differential scanning fluorimetry (DSF) to distinguish the functional strength of the OsDPE2 haplotypes. The Melting temperature (℃) of each OsDPE2-GST haplotype increased logarithmically with the maltose concentration from 100 to 1000 mM (Fig. S11). The results revealed that OsDPE2 (AQ) had a significantly higher binding ability than the other two haplotypes (Fig. 6d). Similarly, the DPE2 activity assays revealed that OsDPE2 (AQ) had significantly higher DPE2 activity than the other two haplotypes (Fig. 6e and f). Thus, these results suggested that OsDPE2 (AQ) was a higher functioning haplotype than the others.
Therefore, these results suggested that OsDPE2 haplotypes may have evolved with rice domestication and could affect rice panicle yield. To evaluate this assumption, we generated a transgenic plant line, OsDPE2(AQ)ZH11 using the OsDPE2(AQ) sequence of Dular ligated to pC2301 vector skeleton, which was transformed into ZH11(haplotype of OsDPE2 is OsDPE2(TA)). Analysis of the panicle yield characteristic of OsDPE2(AQ)ZH11 and ZH11 was then conducted for three consecutive years. The results showed that OsDPE2(AQ)ZH11 had bigger panicles and more panicle branches and spikelets per panicle than ZH11 (Fig. 6g ~ m). Thus, in addition to higher DPE2 activity, the haplotype OsDPE2 (AQ) could increase the panicle yield of rice.