Anthocyanins are widely distributed in one or more parts of rice (Oryza sativa L.) plants, including seed coat, stigma, apiculus, leaf sheath and leaf blade, and are the main pigments used in rice to achieve different colors (Hou et al, 2009; Aizza and Dornelas, 2011). In rice, tissue-specific color traits (especially the color of apiculus, namely the lemma and palea of the spikelet) are not only important for rice variety identification but also important for linkage analysis and rice domestication research (Saitoh et al, 2004; Fan et al, 2007; Lin et al, 2019). The apiculus color is controlled by the complementary functions of three pairs of dominant genes, C, A and P. Gene C (chromogen) is a pigment gene, which is the basic gene for producing pigments. Gene A (activator) activates gene C, converting the chromogen into anthocyanins, and gene P (purple) controls the distribution of anthocyanins in various organs (Reddy, 1996; Sakamoto et al, 2001). OsC1, which determines apiculus color, locates on chromosome 6 (Zhao et al, 2016). S₅ⁿ, which controls wide compatibility in crosses between rice subspecies, locates 427 kb ± 94 kb away from OsC1 in physical distance on chromosome 6, and is genetically tightly linked with OsC1 (Sheng et al, 2008). In general, there is genetic linkage between desirable and undesirable genes, in which case, if the desirable gene is introduced into recurrent parents by backcrossing, the undesirable gene would be also introduced. This genetic phenomenon is regarded as linkage drag. In breeding, it is difficult to break linkage drag by selecting for rare recombinants, especially for tightly linked genes. Therefore, a new method is needed to address this problem. CRISPR/Cas9 system provides a convenient and powerful tool for targeted genome editing in different species, and it has been widely used in rice for improvement of a range of traits (Zhou et al, 2016; Shao et al, 2017; Sun et al, 2017; Tang et al, 2017; Zhang et al, 2018). In Arabidopsis, a double mutant of closely linked genes M320 and M330 has been created using the CRISPR/Cas9 system (Pang et al, 2018). In the current study, our goal was to overcome the tight linkage between OsC1 and S₅ⁿ using the CRISPR/Cas9 system to create new sterile line resources with wide-compatibility and colorless apiculus.

In this study, we confirmed that there was linkage disequilibrium between OsC1 and S₅ⁿ within 132 japonica accessions (Supplemental Fig. 1). To overcome the tight functional linkage between the two genes, we constructed a pC1300-Ubi::Cas9-sgRNA^OsC1 expression vector, with the target site being specifically chosen on exon 1 of OsC1in 4008S (a japonica-type wide-compatibility photosensitive genic male-sterile line with purple apiculi) (Fig. 1-A). The target contained restriction sites that were used for screening mutations using a PCR-based restriction enzyme (PCR/RE) digestion assay. T₀ plants were screened for the presence of frame-shift mutations in OsC1, and five homozygous mutant plants (osc1-2, -3, -10, -12 and -13) were identified from 15 transgenic plants using the PCR/RE assay (Fig. 1-B). A total of four different mutation types were identified from the five homozygous mutant plants by the Sanger sequencing (Fig. 1-C). These four different mutation types all create a stop codon in the coding area, thus leading to the production of an immature protein of OsC1. In this study, we randomly chose the osc1-2 mutant to produce the T₁ generation. Previous studies indicate that there must be a difference of at least two bases between the sgRNA and the target sequence if off-targeting events are to be avoided (Li et al, 2016). To determine whether off-targeting had taken place in the osc1-2 plant line, four putative off-target sites which showed high homology with the sgRNA of OsC1 were searched, and we found that none of the four putative off-target sites produced off-target events (Supplemental Table 1). To determine whether Cas9 and the hygromycin resistance gene (hptII) sequences were present in these mutant lines, PCR amplification was performed using the primers designed to specifically amplify Cas9 and hptII sequences, respectively (Supplemental Fig. 2-A). Seven possible transgene-free homozygous osc1-2mutant plants were isolated from 18 T₁ plants (Supplemental Fig. 2-B). These possible transgene-free mutant T₁ plants represented a new germplasm resource for rice breeding and were chosen for further studies.

Supplemental Fig. 1.

Figure Option ViewDownloadNew Window

Supplemental Fig. 1. Pairwise linkage disequilibrium heatmap (LD heatmap) of the OsC1 and S5n gene. Heatmap displaying pairwise linkage disequilibrium between polymorphic single nucleotide polymorphisms (SNPs) of OsC1 and S5n gene. The heat color scale represents squared correlation (R2) between pairs of SNPs. Color intensity corresponds to higher R2 values according to the legend. The higher the R2 value between the SNPs, the tighter the linkage.

Supplemental Table 1.

Supplemental Table 1.

Supplemental Table 1. Mutations detected in putative CRISPR/Cas9 off-target sites. Plant line was sequenced Target Gene Off-target site No. of mismatching bases No. of plants sequenced No. of plants with mutations Mutation rate (%) osc1-2 OsC1 Os06g0205100 GAGCCAATGGCCTTCGCCATGGG 3 36 0 0 CTCCCTCTTGCCTTCGCCATGGG 4 36 0 0 GCGCCGCTGGCGTTCGCCATCGG 3 36 0 0 CCGCCCATGGCCTTCGCCATAGG 4 36 0 0 The PAM motif (NGG) is shown in green; mismatch bases are shown in red.

Supplemental Table 1. Mutations detected in putative CRISPR/Cas9 off-target sites.

Supplemental Fig. 2.

Figure Option ViewDownloadNew Window

Supplemental Fig. 2. Isolation of possible transgene-free T1 plants from osc1-2. A, A schematic diagram of the position of specific primers used to amplify the hptII and Cas9 sequences in T1 plants. B, hptII and Cas9 DNA fragments could not be detected in some T1 plants of the osc1-2 mutant (numbers 2, 6, 8, 9, 10, 15, and 18). The endogenous OsACTIN1 was used as a control. M, DNA molecular weight marker.

To determine the phenotypic pattern in the mutants, we examined the color trait of the wild type and osc1-2 in the apiculi. The wild type 4008S exhibited purple apiculi at the heading stage. However, osc1-2displayed colorless apiculi (Fig. 1-D to -F). To examine whether the tight functional linkage between OsC1 and S₅ⁿ present in the wild type was overcome in the osc1-2 mutant, primers (S₅ⁿ-JD-F/R and OsC1-JD-F/R) were designed to specifically amplify S₅ⁿ and OsC1 sequences in the wild type and the osc1-2 mutant, respectively (Supplemental Table 2). The results indicated that the sequence ofS₅ⁿ can be amplified in both the wild type and the osc1-2 mutant, whereas the sequence ofOsC1 can be amplified in only the wild type but not in osc1-2 (Fig. 1-G). In addition, we also measured the expression of OsC1 in different organs of wild type and osc1-2 by quantitative real-time PCR (qRT-PCR). The data indicated that OsC1was constitutively expressed in the root and particularly in the leaf of the wild type, whereas expression was significantly down-regulated in both organs of the mutant line (Fig. 1-H), which is consistent with the expression profile reported by Zhao et al (2016). These results indicated that, using the CRISPR/Cas9 system, we were able to overcome the co-inheritance of the wide-compatibility allele S₅ⁿ and the purple apiculus allele OsC1in 4008S.

Supplemental Table 2.

Supplemental Table 2.

Supplemental Table 2. Primers used in this research. Primer name Sequence (5′ -3′ ) Purpose OsC1-g++ OsC1-g-- T3 OsC1-JC-F OsC1-JC-R hptII F GGCAGCGCCACTTGCCTTCGCCAT AAACATGGCGAAGGCAAGTGGCGC ATTAACCCTCACTAAAGGGA CCAGATCGCTCAGTCTCACA TAGGCCGGAGATAGTTGAGC GCTGTTATGCGGCCATTGTC Vector construction Vector construction Vector construction Detection of target mutations Detection of target mutations Genotyping hptII R GACGTCTGTCGAGAAGTTTC Genotyping Cas9 F ACCAGACACGAGACGACTAA Genotyping Cas9 R ATCGGTGCGGGCCTCTTC Genotyping Actin-F TGCTGACAGGATGAGCAAGG Genotyping Actin-R CCCAACCATGCAAAGCTCAC Genotyping OsC1-qPCR-F ATGGGGAGGAGAGCTTGCTG qPCR OsC1-qPCR-R CGGAGATAGTTGAGCCACC qPCR Ubi-qPCR-F GCTCCGTGGCGGTATCAT qPCR Ubi-qPCR-R CGGCAGTTGACAGCCCTAG qPCR S5n-JD-F CTTCATTCCCAGCGAGCGG Genotyping S5n-JD-R ATGGGCGGAGGCATTGGT Genotyping OsC1-JD-F GCAAAGGAAGGGATGAAGAG Genotyping OsC1-JD-R CGTCATCGCCGTCTCCTAATT Genotyping

Supplemental Table 2. Primers used in this research.

Fig. 1.

Figure Option ViewDownloadNew Window

Fig. 1. CRISPR/Cas9 induced mutation in OsC1 gene break the tight functional linkage between OsC1andS5n. A, Schematic representation of OsC1gene structure and the genomic RNA (gRNA) target site on its genomic sequence. Introns are depicted as lines, and exons are depicted as blue boxes. The targeted site is labeled with a red arrowhead, and its sequence is depicted by blue uppercase letters; the protospacer adjacent motif (PAM) (CCC) is marked red. The recognition sequence of NcoI is underlined. B, Identification of mutations in osc1 via PCR-based restriction enzyme assay in the T0 generation. The PCR products of osc1-2, -3, -10, -12 and -13 are resistant to NcoI digestion. Nipponbare (Nip) was used as the negative control. M, Marker. C, Sequencing results at the target sites in T0 plants. The targeted sequence is highlighted in blue, and the PAM sequence is in red. Green uppercase letter indicates substituted nucleotide. Red lowercase letter indicates inserted nucleotide. Black dashes indicate deleted nucleotides. ‘ +’ and ‘ -’ on the right indicate insertion and deletion, respectively, and numbers indicate number of nucleotides. WT, Wild type. D, Phenotypes of WT and osc1-2 plants. E, Grain shape of WT. F, Grain shape of osc1-2. G, Analysis of PCR production of OsC1 and S5n in WT and osc1-2 on 1% agarose gel. M, Marker. H, Quantitative real-time PCR analysis of the transcription of OsC1 in roots and leaves of WT and osc1-2 plants. * , P < 0.05; * * , P < 0.01 (Student’ s t-test).

In addition to the apiculus color trait, we also investigated agronomic traits to determine whether there were any pleiotropic effects of the frame-shift mutation. The data indicated that the plants of the osc1-2 mutant were significantly higher than the wild type, but there was no significant difference in either the tiller number or the number of grains per panicle between the wild type and osc1-2 (Supplemental Fig. 3-A to -C). In addition, the seed-setting rate, the 1000-grain weight, and the average yield per plant of osc1-2 also exhibited no significant difference from those of the wild type (Supplemental Fig. 3-D to -F). The grain length of the mutant was significantly greater than that of the wild type, but there were no significant differences in the grain width and the grain thickness of the wild type and the mutant, meaning that the length-to-width ratio of the mutant was significantly higher than that of the wild type (Supplemental Fig. 3-G to -J). These results indicate that, except for apiculus color and plant height, the osc1-2 line exhibited normal agronomic traits relatively unchanged from the wild type. Overall, a frame-shift mutation of OsC1 in the genetic background of 4008S overcame the detrimental effect of the tight linkage between OsC1 and S₅ⁿ, providing a new germplasm resource for rice breeders.

Supplemental Fig. 3.

Figure Option ViewDownloadNew Window

Supplemental Fig. 3. Characterization of agronomic traits of wild-type (WT) and osc1-2 plants. A, Plant height. B, Tiller numbers per plant. C, Grain number per panicle. D, Seed-setting. E, Weight of 1, 000 grains. F, Average yield per plant. G, Grain length. H, Grain width. I, Grain thickness. J, Length-to-width ratio. All data are from mature plants grown in the field under normal agricultural conditions. Data are presented as Mean ± SD (n = 10). Differences between the two genotypes were compared using Student’ s t-test (* * , P < 0.01).

The purple apiculus trait controlled by OsC1 is an easy visual marker, and often used for varietal selection and evaluation of the purity of a seed population, which is tightly linked in coupling to the wide-compatibility allele S₅ⁿ. Such tight gene linkage is common in rice and can be used for varietal improvement in rice breeding, as tight linkage of desirable alleles can be beneficial in the selection of improved varieties. For example, the two QTLs (qSPP5 for spikelet number per panicle and qTGW5 for grain weight) are tightly linked on chromosome 5, which can be valuable for improving rice yields (Luo et al, 2013). However, the tight functional linkage between desirable and undesirable genes is complicated and challenging in terms of its application in rice breeding. For example, the tight linkage between the positive QTLs GS5 and Chalk5 (Supplemental Fig. 4), which regulate grain size and white belly grain rate, respectively, is not conducive to the development of high-yielding and good-quality rice varieties (Li et al, 2011). Another example is the tight linkage between the desirable rice blast resistance gene and the undesirable high chalky gene, which causes a barrier in breeding of good-quality and high disease-resistant rice varieties (Fukuoka et al, 2009).

Aggregation of large numbers of desirable genes by hybridization is the basis for developing improved varieties by traditional breeding. However, this method has certain disadvantages. Firstly, depending on the crop and its breeding system, it takes more than a decade to hybridize and accumulate multiple desirable genes, and the workload is large. Secondly, when desirable and undesirable genes are tightly linked, it can be difficult to eliminate the undesirable gene by selection, which hinders the use of the desirable genes. Therefore, it is necessary to develop effective methods to overcome the tight linkage between desirable and undesirable genes.

Supplemental Fig. 4.

	Figure Option ViewDownloadNew Window
	Supplemental Fig. 4. Pairwise linkage disequilibrium heatmap (LD heatmap) in the Chalk5and GS5 genes. Heatmap displaying pairwise linkage disequilibrium between polymorphic SNPs of Chalk5 and GS5 genes. The heat color scale represents R2 between pairs of SNPs. Color intensity corresponds to higher R2 values according to the legend. The higher the R2 value between SNPs, the tighter the linkage.

The CRISPR/Cas9 system can precisely edit the sequence of any gene, with the advantages of simple operation, a low technical threshold and a short experimental period (Baltes and Voytas, 2015). In our study, the problem of the tight functional linkage between OsC1 and S₅ⁿ was successfully overcome by mutating the purple apiculi allele into the colorless apiculus allele (Supplemental Fig. 1), resulting in the production of a valuable new genetic resource for breeding, the osc1mutant combined with the wide-compatibility gene S₅ⁿ but with colorless apiculi. Furthermore, through self-pollination, the T-DNA sequences can be removed and possible transgene-free plants were generated from the T₁ generation in only one year (Supplemental Fig. 2-B). Therefore, the CRISPR/Cas9 system is an effective method for rapidly breaking the linkage drags. Although genetic modification (GM) crops still have great controversy in China, genome editing is, in many ways, even more precise and predictable than transgenesis. Crops bred by genome editing are defined as genome-edited crops (GECs) and the US Department of Agriculture does not consider GECs to be GM organisms as long as GECs do not contain DNA from plant pests. Moreover, in many cases, GECs are more promising than those from conventional breeding, which would be offered more opportunities for ensuring global food and nutrition security (Huang et al, 2016). In a word, this strategy would have wider applications in significantly accelerating the use in breeding application of beneficial genes that had previously been limited by the presence of undesirable linkage genes.