In-Frame and Frame-Shift Editing of the Ehd1 Gene to Develop Japonica Rice With Prolonged Basic Vegetative Growth Periods

Japonica rice has become increasingly popular in China owing to its superior grain quality. Over the past decades, “indica to japonica” projects have been proposed to promote cultivation of japonica rice in low latitudes in China. Traditionally, japonica varieties were planted mainly in mid latitudes in the northeast plain and Yangtze River region. The key obstacle for introducing elite mid-latitude japonica varieties to low latitudes is the severe shortening of growth period of the japonica varieties due to their sensitivity to low-latitude short photoperiod and high temperature. Here we report development of new japonica rice with prolonged basic vegetative growth (BVG) periods for low latitudes by targeted editing the Early heading date 1 (Ehd1) gene. Using CRISPR/Cas9 system, we generated both frame-shift and/or in-frame deletion mutants in four japonica varieties, Nipponbare, Longdao16, Longdao24, and Xiushui134. When planting at low-latitude stations, the frame-shift homozygous lines exhibited significantly longer BVG periods compared with wild-types. Interestingly, we observed that minor deletion of the first few residues within the receiver domain could quantitatively impair the function of Ehd1 on activation of Hd3a and RFT1, resulting in an intermediate-long BVG period phenotype in the homozygous in-frame deletion ehd1 lines. Field investigation further showed that, both the in-frame and frame-shift lines exhibited significantly improved yield potential compared with wild-types. Our study demonstrates an effective approach to rapid breeding of elite japonica varieties with intermediate-long and long BVG periods for flexible cropping systems in diverse areas or under different seasons in southern China, and other low-latitude regions.


INTRODUCTION
Rice (Oryza sativa L.) is one of the most important crops in China. Cultivated rice is comprised of japonica and indica subspecies. Compared to indica varieties, japonica varieties in general have more superior grain quality -higher head rice rates and gel consistency, lower chalky rice rates, chalkiness degree, and amylose contents (Feng et al., 2017). Over the past decades, japonica rice has become increasingly popular in China. To meet the demands of consumers, "indica to japonica" projects have been proposed to promote cultivation of japonica varieties in southern China Leng et al., 2019).
Traditionally, japonica varieties were planted mainly in mid latitudes in the northeast plain and Yangtze River region, and indica varieties were planted in low latitudes in southern China (Min et al., 2011;Wang et al., 2015). Over the past decades, many elite japonica varieties have been bred for mid latitudes in China. However, most of the midlatitude japonica varieties were sensitive to photoperiod or temperature, and they were not suitable for cultivating in lowlatitude regions. Under short photoperiod and high temperature environmental conditions in low latitudes, the mid-latitude japonica varieties commonly display early flowering, resulting in low grain yield Wei et al., 2009;Yuan et al., 2009).
CRISPR/Cas9 technology has emerged as a powerful approach for rice research and breeding (Shan et al., 2014;Xie et al., 2015;Lu et al., 2017;Xu et al., 2017;Li et al., 2018). CRISPR/Cas9-mediated gene modification has been applied in rice for improvement of important agronomic traits, such as photo-sensitive and thermo-sensitive genic male sterility Zhou et al., 2016), herbicide-resistance (Sun et al., 2016), resistance to blast and tungro spherical virus (Wang et al., 2016;Macovei et al., 2018), low cadmium accumulation , grain yield (Miao et al., 2018), glutinous trait , chilling tolerance (Mao et al., 2019), red pericarp , and fixation of heterosis (Khanday et al., 2019;Wang et al., 2019). More recently, Cui et al. (2019) assessed the effect of ten heading time genes on reproductive transition and yield components in rice using a CRISPR/Cas9 system, revealing that heading time was often associated with yieldrelated traits. In the present study, we report development of new japonica rice exhibiting prolonged BVG periods for low latitudes by targeted editing the receiver domain region of Ehd1. Using CRISPR/Cas9 system, we generated both frame-shift and/or in-frame deletion mutants in four japonica varieties, Nipponbare, Longdao16, Longdao24, and Xiushui134. When planting under low-latitude conditions, the frame-shift mutants of Nipponbare, Longdao16, Longdao24, and Xiushui134 exhibited significantly longer BVG periods compared with their wild-type parents. Interestingly, the inframe mutants of Longdao16, Longdao24, and Xiushui134 possessed intermediate-long BVG periods. Field investigation further showed that, both the in-frame and frame-shift mutant lines showed significantly improved yield potential compared with wild-type plants. Our study demonstrates an effective approach for breeding new japonica varieties with intermediatelong and more extremely long BVG periods for flexible cropping systems in southern China, as well as in other lowlatitude regions.

Plant Materials
The japonica variety Nipponbare and three elite japonica varieties Longdao16, Longdao24 and Xiushui134 were used in this study.

Construction of sgRNA-Cas9 Plant Expression Vector and Rice Transformation
A single guide RNA (sgRNA) sequence targeted the receiverdomain region of the Ehd1 gene were designed according to Shan's program (Shan et al., 2014). A sgRNA-Cas9 vector pVK005-01 (Viewsolid Biotech, Beijing, China) in which a codon-optimized Cas9 gene was under control of the maize ubiquitin promoter and the customized sgRNA was driven by the rice U6 promoter, was used for generation of the sgRNA-Cas9 construct for Ehd1. DNA oligos (Supplementary Table S1) corresponding to the designed sgRNA sequence were synthesized and the dimer was cloned into pVK005-01 according to the manufacturer's instruction.
The Ehd1 sgRNA-Cas9 expression vector was introduced into Agrobacterium tumefaciens strain EHA105 by electroporation. Agrobacterium-mediated transformation of rice calli induced from seeds of Nipponbare, Longdao16, Longdao24, and Xiushui134 was carried out as described previously (Hiei et al., 1997).

Genotyping and Off-Target Analysis
Genomic DNA was extracted from rice leaves using the CTAB method. The T 0 transgenic plants were determined by PCR amplification of the hygromycin phosphotransferase gene. To genotype mutations in the Ehd1 gene, PCR amplification was performed using primer pairs flanking the Ehd1 target site (Supplementary Table S1). The PCR products were sequenced by Sanger method. DNA sequences were aligned using Clustal Omega (Sievers et al., 2011).
Transgene-free homozygous mutant plants/lines were screened in T 1 generation derived from self-pollination of T 0 mutant plants, and were double confirmed in T 2 generation. To identify transgene-free plants, PCR amplification was performed with DNA extracted from individual plants using specific primers (Supplementary Table S1) for the hygromycin phosphotransferase gene and the Cas9 gene.
Forty transgene-free T 2 mutant plants in Nipponbare, Longdao16, Longdao24, and Xiushui134 backgrounds were selected for evaluation of the potential off-target effects. PCR amplification was performed using specific primers (Supplementary Table S1) for the potential off-target sites. The resulting PCR products were sequenced for mutations by Sanger method.

Plant Growth Conditions and Phenotype Analysis
T 0 generation plants were grown in a standard greenhouse. Wild-type parents, T 1 and T 2 transgene-free homozygous mutant plants were investigated for agronomic traits. The plants were cultivated at Sanya experimental station, Hainan Province, under nature SD condition (December to next April, c. 11-12.5 h, 18 • 14 N, 109 • 31 E), and at Fuzhou experimental station, Fujian Province, under nature LD condition (mid-May to mid-September, c. 13.5 h, 26 • 08 N, 119 • 28 E). Longdao16 and Longdao24 were also cultivated at Haerbing experimental station, Heilongjiang Province, under nature LD condition (April to September,45 • 51 N,126 • 50 E). About 100 seedlings of each line were transplanted in the field. Column space and row were both 20 cm. The field management was followed the normal agricultural practice. Head rice rate was determined by the percentage of the total head rice weight produced after milling. Grain chalkiness traits were determined using image analysis method (Fang et al., 2015), and grain amylose content was determined by iodine reaction (Juliano, 1971). Gel consistency and alkali spreading value were determined as described previously (Leng et al., 2013).
One T 2 Nipponbare mutant line N-ehd1-#1, and two T 2 Longdao16 mutant lines L16-end1-#2, L16-end1-#3 were used for analyzing gene expression of Hd3a and RFT1 at different developmental stages. The frame-shift mutant lines N-ehd1-#2, L16-end1-#1, and in-frame mutant line L16-end1-#6 were also analyzed for the mRNA expression levels of Ehd1, Hd3a, and RFT1 at 50d after sowing. All the lines were grown under LD condition (mid-May to mid-August, c. 13.5 h, before heading) in nature filed and SD condition (September to next January, c. 12.7-11.5 h) in a greenhouse at Fuzhou experimental station. Full expanded leaves were sampled 2 h after dawn. Three independent plants were used as biological replicates.

Gene Expression Analysis
Total RNAs were isolated from rice leaves using RNA isolater Total RNA Extraction Reagent (Vazyme Biotech Co., Ltd., Nanjing, China) according to the manufacturer's instructions. The RNA samples were treated with Recombinant DNase I (TaKaRa, Shiga, Japan). First cDNA was synthesized with 2 µg of total RNA using a RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, UC, Lithuania). The synthesized cDNAs were used as templates for quantitative real-time RT-PCR with ChamQ TM SYBR qPCR Master Mix (Vazyme Biotech, Nanjing, China) on a QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific, Singapore). All experiments were conducted three times. The primers used for gene expression analysis are listed in Supplementary Table S1.

Mediated by CRISPR/Cas9 System
To generate transgenic plants with targeted mutation in Ehd1, a sgRNA-Cas9 construct targeting the first exon of the Ehd1 gene ( Figure 1A) was designed and used for rice transformation. A total of 30, 28, 25, and 20 T 0 independent transgenic plants were generated from calli of japonica rice varieties Nipponbare, Longdao16, Longdao24, and Xiushui134, respectively (Table 1 and Supplementary  Table S2). Genotyping of the transgenic plants identified 53 (14, 15, 13, and 11 in Nipponbare, Longdao16, Longdao24, and Xiushui134 backgrounds, respectively) with mutations in the predicted target site of Ehd1 (Figure 1, Table 1, and  Supplementary Table S2). The mutations in T 0 transgenic plants included homozygous, bi-allelic and heterozygous genotypes ( Table 1 and Supplementary Table S2). Most of the mutations were frame-shift types in mutants in Nipponbare, Longdao16, Longdao24, and Xiushui134 backgrounds. In addition, there were two, three, and one mutants with inframe deletion in the target site in Longdao16, Longdao24, and Xiushui134 backgrounds, respectively ( Figure 1B and Supplementary Table S2). The mutants in Nipponbare, Longdao16, Longdao24, and Xiushui134 backgrounds were thus referred to as N-ehd1, L16-ehd1, L24-ehd1, and X-ehd1, respectively.
Transgene-free T 1 and T 2 mutant lines were screened (Supplementary Figures S1, S2) and T 2 mutant lines were examined for potential off-target mutations. Six potential off-target sites containing four or five mismatched bases were retrieved using the CRISPR-P 2.0 tool . PCR products amplified from 40 transgene-free T 2 mutant plants (9, 11, 10, and 10 in Nipponbare, Longdao16, Longdao24, and Xiushui134 backgrounds, respectively) were sequenced. No mutations were found in all six potential sites ( Supplementary Table S3), indicating the high specificity   Figure 2 and Supplementary Figure S3, the flowering times of the frame-shift mutants were significantly delayed compared with that of wild-type in Nipponbare, Longdao16, Longdao24, and Xiushui134 backgrounds. For example, the heading dates of N-ehd1-#1, an 1-bp deletion homozygous mutant line in Nipponbare background (Figure 2A), were 98.1 ± 3.1 days under SD and 92.3 ± 2.5 days under LD, respectively. In contrast, the heading dates of wild-type Nipponbare were about 57.7 ± 2.3 days and 61.1 ± 2.3 days under SD and LD, respectively (Figure 2A). The heading dates of L16-ehd1-#1, an 1bp deletion homozygous mutant line in Longdao16 background, were 108.5 ± 2.4 days under SD and 106.4 ± 2.5 days under LD, which were about 40 days more than that of wildtype Longdao16 plants under SD (71.2 ± 2.2 days) and LD (64.9 ± 2.3 days), respectively ( Figure 2B). Similar results were observed in other tested frame-shift mutant lines in all four backgrounds (Figure 2), indicating that frame-shift mutations in Ehd1 could result in long BVG periods in japonica varieties under low-latitude conditions.
During the vegetative growth phrase, Hd3a and RFT1 were expressed at relatively high levels in wild-type varieties Nipponbare and Longdao16. In contrast, Hd3a and RFT1 were expressed at low levels in the frame-shift mutant lines N-ehd1-#1, N-ehd1-#2, L16-end1-#1, and L16-ehd1-#2 ( Figure 4A and Supplementary Figure S6). The expression levels of Hd3a and RFT1 in the in-frame deletion lines L16-ehd1-#3 and L16-end1-#6 were lower than that in wildtype Longdao16, but were higher than that in L16-end1-#1 and L16-ehd1-#2 ( Figure 4A and Supplementary Figure  S6), although expression patterns were not exactly the same under SD and LD, respectively. We further performed the liner fitting analysis of the relationship between heading date and expression levels of Hd3a and RFT1 in L16-ehd1-#2, L16-ehd1-#3, and Longdao16. The analysis revealed that there was a significantly negative correlation between heading date and expression levels of Hd3a and RFT1 (Supplementary Figure S7). These results thus supported the conclusion that transcriptional activation of Hd3a and RFT1 mainly relied on the presence of functional Ehd1 (Doi et al., 2004;Zhao et al., 2015;Cho et al., 2016;Nemoto et al., 2016). Our results also indicated that minor deletions in the receiver domain could quantitatively impair the function of Ehd1 on activation of Hd3a and RFT1, resulting in a moderately-delayed flowering phenotype ( Figure 4B).

In-Frame and Fame-Shift Mutant Lines Exhibited High Grain Yield and Good Grain Quality in Low-Latitude Regions
Homozygous in-frame and frame-shift mutant lines were investigated under both SD and LD in low-latitude regions for yield-associated traits, including plant height, panicle number per plant, panicle length, 1000-grains weight, spikelets per panicle, and grain yield per plant. Due to the prolongation of growth period, both the in-frame and frame-shift mutant lines exhibited significantly higher plant height, panicle length, especially spikelets per panicle, which in turn resulted in higher grain yield per plant compared with wild-types ( Figure 5 and Table 2). For example, the grain yields per plant of the in-frame mutant lines L16-ehd1-#3 and L16-ehd1-#6 were about 1.37-to 1.43-fold, and the grain yields per plant of the frame-shift lines L16-ehd1-#1, L16-ehd1-#2, and L16-ehd1-#4 were about 1.90-2.28-fold compared with that in wild-type Longdao16 plants under SD or LD ( Figure 5F). Similar results were observed in other homozygous frame-shift lines in Nipponbare, Longdao24, or Xiushui134 backgrounds (Figures 5E,G,H).
Rice grains harvested from wild-types and the ehd1 mutant lines cultivated in Fuzhou, and from the two Heilongjiang varieties Longdao16 and Longdao24 cultivated in Haerbing, Heilongjiang, were evaluated for quality characteristics, including head rice rate, chalkiness degree, chalky rice rate, amylose content, gel consistency, and alkali spreading value. The results showed that, Longdao16 and Longdao24 cultivated in Heilongjiang had higher head rice rate, lower chalky rice rate and chalkiness degree compared with Longdao16, Longdao24 and the ehd1 mutant lines cultivated in Fujian. There were  Figure S8). When compared with wild-types cultivated in Fuzhou, the in-frame and frame-shift ehd1 mutant lines showed higher head rice rate, lower chalky rice rate and chalkiness degree (Supplementary Figure S8), which may be due to that the ehd1 mutant lines were exposed to relatively lower temperature during grain-filling stage. There were no significant differences in amylose content, gel consistency and alkali spreading value between the ehd1 mutant lines and their corresponding wild-types (Supplementary Figure S8). Overall, the ehd1 lines exhibited high grain yield and good grain quality when cultivated in low-latitude regions.

DISCUSSION
Rapid Breeding of New Elite Japonica Varieties With Prolonged BVG Periods and High Yield for Low-Latitude Regions Through Genome Editing of the Ehd1 Gene Previous studies revealed that grain yield of rice is positively correlated with growth period. A longer BVG period could promote the accumulation and allocation of more resources to reproductive organs, resulting in higher yield (Xue et al., 2008;Wei et al., 2010;Hu et al., 2013;Yan et al., 2013;Zhang et al., 2015;Leng et al., 2019). Severe shortening of growth period is considered to be the main obstacle for introducing northern japonica varieties to southern areas in China. Therefore, breeding of japonica varieties with long BVG periods would be of great significance to promote "indica to japonica" projects in southern China (Leng et al., 2019). A growing number of studies showed that enhanced expression of or mutations in flowering time genes could lead to heading date changes in rice (Xue et al., 2008;Wei et al., 2010;Hu et al., 2013;Koo et al., 2013;Yan et al., 2013;Zhang et al., 2015;Wu et al., 2017;Leng et al., 2019). For example, enhanced expression of Ghd7, an upstream negative regulator of Ehd1, could lead to delayed heading in rice under LD. In contrast, natural mutants with reduced function of Ghd7 exhibited early flowering (Xue et al., 2008). DTH7, DTH8 are two suppressors in the signal network of photoperiodic flowering. Expression of DTH7 or DTH8 led to delayed heading by down-regulating expression of Ehd1 and Hd3a in rice under LD (Wei et al., 2010;Yan et al., 2013;Gao et al., 2014). Natural mutation of DTH7 or DTH8 caused weak photoperiod sensitivity. More recently, a preponderant allele of Heading date 1 (Hd1) has been identified from indica cultivars, and backcrossing the preponderant Hd1 allele led to prolongation of growth period of a japonica variety Chunjiang06 (Leng et al., 2019). These studies also demonstrated that, due to the prolongation of growth period, rice lines with enhanced expression of Ghd7, DTH7, DTH8 or introgression of the preponderant Hd1 allele possessed significantly increased grain yield (Xue et al., 2008;Wei et al., 2010;Yan et al., 2013;Gao et al., 2014;Zhang et al., 2015;Leng et al., 2019).
In this study, we aimed to breed japonica rice with prolonged growth periods under both SD and LD. While many rice flowering time genes, such as Ghd7, DTH7 or DTH8, function only under LD, Ehd1 functions to promote rice flowering under both SD and LD (Doi et al., 2004;Komiya et al., 2008). More recently, Cui et al. (2019) demonstrated that a single−nucleotide insertion in the third exon of the Ehd1 gene resulted in delayed heading date of the japonica rice variety Sasanishiki. In the present study, we chose the first exon of Ehd1 as a target for CRISPR/Cas9-mediated genome editing for regulating heading date in rice. We generated both frame-shift and/or inframe deletion mutants in four japonica varieties, Nipponbare, Longdao16, Longdao24, and Xiushui134. Similar to previous studies Zhou et al., 2016;Macovei et al., 2018;, we were able to obtain transgene-free mutant lines by segregating away the integrated T-DNA in T 1 generations (Supplementary Figure S2). Field investigation further showed that, in all four japonica variety backgrounds, the flowering time of in-frame and frame-shift ehd1 mutant lines was significantly delayed compared with that of wild-type plants under both SD and LD. In the present study, we also observed that the delay of flowering times of the Xiushui134 mutant lines was not as great as that of the ehd1 mutant lines in Nipponbare, Longdao16, or Longdao24 backgrounds, suggesting that the effects of Ehd1 on heading date varied under different genetic backgrounds.
Accordingly, the ehd1 mutant lines displayed significantly improved yield potential (Figure 5). When planting as middleseason rice in mid-May at Fuzhou station, wild-type Nipponbare, Longdao16, Longdao24, and Xiushui134 flowered in late-July to early-August, and were exposed to high temperatures during grain-filling stage. In contrast, the in-frame and frame-shift ehd1 mutant lines started to flower in late-August to early-September. While high temperature during grain-filling stage has been shown to lead to higher chalk values (Yamakawa et al., 2007), the delay of flowering time allowed the ehd1 mutant lines to be exposed to relatively lower temperatures during grainfilling stage, resulting in higher head rice rate, lower chalky rice rate and chalkiness degree (Supplementary Figure S8). In addition, we did not observe obvious differences in stress tolerance, disease resistance or lodging resistance between the ehd1 mutant lines and their wild-type parents while they were cultivated in the field at Sanya station and Fuzhou station over two generations (data not shown). Taken together, our study demonstrates an effective approach to rapid breeding of elite japonica varieties with prolonged BVG periods and high yield for low-latitude regions through genome editing of Ehd1.
In-Frame Editing of Ehd1 for Developing Rice With Intermediate-Long BVG Periods Japonica varieties with extremely long BVG periods are expected to be suitable for cultivating in low-latitude regions. In production practice, however, rice varieties with various growth periods are required for cultivation in diverse areas or under different cropping seasons. For example, low-altitude and singleseason cropping areas are more likely to grow varieties with long growth periods. In contrast, varieties with intermediatelong growth periods would be more applicable to planting in relatively high altitude areas or under double-season cropping system. In the present study, we found it interesting that the in-frame deletion ehd1 lines in Longdao16, Longdao24, and Xiushui134 backgrounds, displayed a moderately-delayed flowering phenotype. The flowering time of the five in-frame deletion ehd1 lines in Longdao16 and Longdao24 backgrounds was delayed about 3 weeks compared with that of wild-type plants, but was about 2 weeks earlier than that of the frameshift lines (Figure 5). Similar moderately-delayed flowering phenotypes were observed in the in-frame deletion Xiushui134 mutant line X-ehd1-#5, although the delay of flowering time was less extreme. Ehd1 has a receiver domain and a GARP DNA-binding domain (Doi et al., 2004;Cho et al., 2016). In the receiver domain, the conserved Asp-Asp-Lys (D17th -D63th -K117th) motif, is known to be important for Ehd1 function (Takahashi et al., 2009;Cho et al., 2016). For example, phosphorylation of the middle Asp is required for homodimerization of Ehd1, which is crucial for Ehd1 activity (Cho et al., 2016). In the in-frame deletion ehd1 lines, the editing caused minor deletion mutations of the first few residues of the Ehd1 receiver domain. These residues might not be critical for basic function of Ehd1. However, the moderatelydelayed flowering phenotype of the six lines suggested that the minor in-frame mutations could quantitatively impair the biological function of Ehd1. Expression of Hd3a and RFT1 in L16-ehd1-#3 was lower than that in wild-type Longdao16, but higher than that in the frame-shift line L16-ehd1-#2 (Figure 4), further supporting that the in-frame deletion ehd1 allele had a moderate effect on activation of Hd3a and RFT1. Overall, our results showed that the in-frame deletion ehd1 alleles have common effect among different backgrounds. Therefore, the in-frame deletion ehd1 alleles could be valuable gene resources in developing rice varieties with intermediate-long BVG periods.
The majority of important agronomic traits in crops are quantitative traits. Thus, manipulation of quantitative traits has been extremely desired by breeders. Recently, CRISPR/Cas9-based cis-regulatory and upstream open reading frames (uORFs) mutagenesis approaches have been developed to create quantitative variations for crop breeding (Rodríguez-Leal et al., 2017;. Editing of the cisregulatory elements or uORFs could cause transcriptional change of the targeted genes or the downstream primary ORFs (pORFs), leading to quantitative variation of the traits . Using the CRISPR/Cas9 system, Rodríguez-Leal et al. (2017) generated a continuum of variants of three major productivity traits (fruit size, inflorescence branching, and plant architecture) in tomato;  created mutant plant lines with varying amounts of mRNA translation in four pORFs. These results demonstrated the great feasibility of the two approaches for engineering quantitative traits for crop improvement. In the present study, we observed that minor in-frame deletion of non-critical residues of the Ehd1 receiver domain could result in quantitatively delayed flowering in rice. Our finding suggested that in-frame editing of noncritical residues could potentially be an alternative approach to creating quantitative variants of important traits. Unlike targeted editing in cis-regulatory regions that may change the expression patterns of the genes and cause unpredictable phenotypic variation (Rodríguez-Leal et al., 2017), in-frame editing does not affect expression pattern of the targeted genes, and may be applicable to traits controlled by specific expression patterns.
In summary, we report the development of japonica rice varieties with prolonged BVG periods and high yield potential by CRISPR/Cas9-mediated editing of Ehd1. Our results showed that, while frame-shift mutations in Ehd1 resulted in more extremely long BVG periods, minor in-frame deletion mutations in the receiver domain region of Ehd1 led to moderate delays of heading date in japonica rice. Field investigation further showed that, both the in-frame and frame-shift lines exhibited significantly improved yield potential compared with wild-types. Therefore, our study demonstrated an effective approach to rapid breeding of elite japonica varieties with intermediatelong and more extremely long BVG periods. The ehd1 mutant lines with the frame-shift or the novel in-frame deletion ehd1 alleles represent valuable resources to develop elite japonica rice varieties for flexible cropping systems in diverse areas or under different seasons in southern China, and other low-latitude regions. In addition, our study suggests that in-frame editing of non-critical residues could potentially be an effective approach for engineering quantitative traits for crop improvement.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ACKNOWLEDGMENTS
We thank Dr. Yin Meng at Tillage and Cultivation Researching Institute, Heilongjiang Academe of Agriculture Sciences for providing japonica varieties Longdao16 and Longdao24, and Prof. Haigen Yao at Jiaxing Academy of Agricultural Sciences for providing japonica variety Xiushui134.