Derepression of specific miRNA-target genes in rice using CRISPR/Cas9

Abstract

MicroRNAs (miRNAs) target specific mRNA molecules based on sequence complementarity for their degradation or repression of translation, thereby regulating various developmental and physiological processes in eukaryotic organisms. Expressing the target mimicry (MIM) and short tandem target mimicry (STTM) can block endogenous activity of mature miRNAs and eliminate the inhibition of their target genes, resulting in phenotypic changes due to higher expression of the target genes. Here, we report a strategy to achieve derepression of interested miRNA-target genes through CRISPR/Cas9-based generation of in-frame mutants within the miRNA-complementary sequence of the target gene. We show that two rice genes, OsGRF4 (GROWTH REGULATING FACTOR 4) and OsGRF8 carrying in-frame mutants with disruption of the miR396 recognition sites, escape from miR396-mediated post-transcriptional silencing, resulting in enlarged grain size and increase in brown planthopper (BPH) resistance, in their respective transgenic rice lines. These results demonstrate that CRISPR/Cas9-mediated disruption of miRNA target sites can be effectively employed to precisely derepress particular target genes of functional importance for trait improvement in plants.

CRISPR/Cas9, de-repression, in-frame, miRNAs, rice

Introduction

MicroRNAs (miRNAs) are a class of single-strand non-coding RNAs, of typically 20-24 nt in length (Jones-Rhoades et al., 2006; Zhang et al., 2006; Li et al., 2017; Ghini et al., 2018). They bind to mRNAs by specific complementary base-pairing mechanisms, and mediate post-transcriptional gene silencing. In plants, miRNAs have been shown to be involved in numerous biological processes, including growth, development, stress responses and hormone signalling, making miRNAs strong targets for trait improvement (Khraiwesh et al., 2012; Tang and Chu, 2017). Thus, exploring the effective approaches to interrogate miRNA functions is necessary.

Target mimicry (MIM), consisting of a non-cleavable RNA that forms a non-productive interaction with a complementary miRNA, provides a powerful technology that can be used to inhibit the activity of miRNAs (Franco-Zorrilla et al., 2007). Based on the MIM technology, short tandem target mimicry (STTM), which is composed of two short sequences mimicking small RNA target sites, linked by a short spacer, induces the degradation of specific miRNAs (Yan et al., 2012; Zhao et al., 2019). Moreover, several studies have shown that high efficiency mutagenesis of miRNA genes can be obtained by CRISPR/Cas9 technology (Miao et al., 2020; Zhou et al., 2017). Such miRNA-targeted approaches provide effective tools to interfere with endogenous miRNA activity. In addition, the transgenic expression of miRNA-resistant targets, where multiple synonymous mutations are introduced in the miRNA binding site, is employed to escape the repression of endogenous miRNAs (Li and Millar, 2013; Song et al., 2013). Overall, multiple approaches have been developed to research on miRNA function and the miRNA-target regulatory module, providing guidelines for crop improvement.

miR396 is a miRNA that is conserved among the dicot and monocot plants. In Arabidopsis, miR396 targets a specific family of transcription factors, GROWTH-REGULATING FACTORS (GRFs), which are known to control cell proliferation during leaf development (Jones-Rhoades and Bartel, 2004; Liu et al., 2009). miR396 has also been found to down-regulate the abundance of GRFs mRNA, and is implicated in the control of cell differentiation and organogenesis in Poaceae. Several studies in rice have identified a natural occurring allele of OsGRF4, a 2 bp missense mutation in the miR396 recognition site, perturbing repression of miR396 and increasing grain size (Duan et al., 2015; Hu et al., 2015; Che et al., 2016; Chen et al., 2019). Therefore, it is possible that we can introduce miR396-resistant targets into plants to destroy the inhibition by miR396 and improve specific agronomic traits.

Given that the CRISPR/Cas9 system can produce in-frame and frame-shift mutants (Shan et al., 2013; Zhang et al., 2014; Zhu et al., 2019), we aimed to generate in-frame mutants in the miR396 recognition site of interested target genes, which would make the target genes escape from post-transcriptional inhibition, but also keep the function of target genes. To evaluate whether the strategy works, we selected OsGRF4 and OsGRF8, which are regulated by miR396 to control grain size and brown planthopper (BPH) resistance, respectively (Dai et al., 2019). Through the CRISPR/Cas9 system, we successfully destroyed the miR396-complementary sequence and produced in-frame osgrf4 and osgrf8 mutants in which the lengths of deletions or insertions were in multiples of three bases. Subsequently, the elevated expression of the in-frame osgrf4 and osgrf8 mutants led to enlarged grain size and increased BPH resistance respectively. Consequently, our results constitute a strategy for acquiring miRNA-resistant mutants using CRISPR/Cas9 that could be readily implemented in plants.

Materials and methods

Plant material and growth conditions

Oryza sativa L. indica ‘Shuhui143’ (‘S143’) was used as the wild type (WT) control and transformation host. The majority of the WT and transgenic plants were cultivated under a standard greenhouse at Fuzhou experimental station (26.08°N, 119.28°E), Fujian Province, China. The growing season began in May and extended to mid-October. Mature seeds were germinated in half-strength Murashige and Skoog (MS) medium in a 26 °C incubator for 14 d and then transferred to soil. The seedlings were grown under standard greenhouse conditions (14 h light at 30±1 °C/10 h dark at 25±1 °C).

Vector construction and rice transformation

The single guide RNA (sgRNA) sequences targeting the flanking sequences of the miR396 complementary sequence in the OsGRF4 and OsGRF8 alleles were designed as predicted by the CRISPR-P program (Lei et al., 2014). The oligonucleotide dimers of designed sgRNA sequence were cloned into a CRISPR/Cas9 plant expression vector VK005-01 (Viewsolid Biotech, Beijing, China) following the manufacturer’s specifications. The constructs carried a Cas9/sgRNA T-DNA and a designed sgRNA sequence, driven by the maize ubiquitin promoter and the rice U6 promoter, respectively.

The resulting constructs were confirmed by sequencing and then introduced into Agrobacterium tumefaciens strain EHA105. Next, the strains were transferred into rice calli of ‘S143’ by A. tumefaciens-mediated rice transformation, as described previously (Hiei et al., 1994).

Genotype analysis

Total genomic DNA from fresh leaves was extracted using the cetyltrimethylammonium bromide (CTAB) method (Stewart and Via, 1993). The Cas9 gene was amplified to evaluated the presence/absence of the Cas9/sgRNA T-DNA. Using specific primers, transgenic samples were performed to amplify the sgRNA target region of OsGRF genes and subsequently subjected to Sanger sequencing. Sequencing results were decoded with an established tool, as described previously (Liu et al., 2015). DNA sequences were aligned using Clustal Omega (Sievers et al., 2011).

Trait measurements

A total of 96 plants of T₂ mutagenesis of grf4-#2, grf4-#25, grf4-#1, grf4-#14 and T₁ mutants grf4-2-#7, grf4-2-#46 were grown for analysis. Grain length and grain width with 300 grains were measured when the grains were completely matured. A total of 1000 fully filled grains were chosen to investigate grain weight (1000-grain weight). Plant height, tiller number, primary branch number per panicle, secondary branch number per panicle, grain number per panicle and yield per plant were also measured and analyzed from the plant materials.

RNA isolation and qRT-PCR analysis

Total RNA was extracted from rice young panicles and leaves using TRIZOL (Life technologies, Carlsbad, CA, USA). The cDNAs were reverse transcribed using the First Strand cDNA Synthesis Kit (Thermo, Waltham, MA, USA). Quantitative RT-PCR was performed with the SYBR Premix Ex Taq (Takara, Otsu, Shiga, Japan) as per the manufacturer’s instructions, and the rice UBIQUITIN and OSACTIN1 genes were used as endogenous controls. For the detection of miR396, reverse transcription was conducted using stem-loop RT primers (Supplementary Table S1), and the U6 gene was regarded as an internal reference in quantitative RT-PCR. Relative expression was analysed using the 2^-∆∆Ct method. Data from three biological replicates and three technical repetitions were collected, the mean value with SD was plotted for each individual genome editing event and the controls based on three biological replicates.

Detection of brown planthopper resistance and determination of total flavonoids

Approximate 3-week-old seedlings were used for brown planthopper (BPH) resistance test. Each line of WT ‘S143’, in-frame and frame-shift osgrf8 mutants was infested with 12 second-instar BPH nymphs with at least three replicates (Zhao et al., 2016). A week later, the resistance level (susceptible or resistance) of plants were evaluated (Dai et al., 2019).

Approximately 1 g milled powder from fresh leaves of 30-day-old seedling were extracted in 10 ml of 50% ethanol for 1 h with shaking and repeated three times. The extract was centrifuged at 12 000 × g for 10 min, and 0.5 ml of supernatant was diluted to 5 ml with 30% ethanol, After 5 min, 0.3 ml of 0.5 M NaNO₂, 0.3 ml of 0.3 M AlCl₃, and 2 ml of 1 M NaOH were added and the reaction was made up to 10 ml using distilled water. After 15 min of incubation, the absorbance at 510 nm was recorded against a blank sample. The total flavonoid content was calculated as the equivalents of Rutin used as the standard (Loyola et al., 2016). Each measurement was repeated three times.

Results

miR396-resistant variants with in-frame osgrf4 mediated by the CRISPR/Cas9 system

We picked an elite rice variety, indica ‘Shuhui143’ (‘S143’) to obtain miRNA-resistant mutants. Sequence analysis revealed that OsGRF4 of ‘S143’ contained the miR396-complementary sequence of 21 nucleotides in the third exon (Fig. 1A; Supplementary Fig. S1A), To interfere with miR396-mediated repression, a Cas9/sgRNA construct targeting the flanking sequences of miR396 recognition site was designed (Fig. 1A). Through Agrobacterium-mediated genetic transformation, a total of 43 T₀ independent transgenic plants were generated from calli of ‘S143’. For T₀ lines, 26 out of 43 were identified as mutants, of which more than half (53.85%) were putatively homozygous, while heterozygous and bi-allelic lines accounted for 11.54% and 34.62% of the total mutants, respectively (Table 1; Supplementary Table S2). All of the T₀ mutants destroyed the miR396-complementary sequence of OsGRF4, which would get rid of miR396-mediated post-transcriptional regulation.

Table 1.

Summary of mutation frequencies at targeted miR396-complementary sequence of OsGRF4 allele in T₀ lines

No. of transgenic plants in ‘S143’	No. of plants with mutation: number, ratioa	T0 zygosityb			No. of plants with in-frame osgrf4 alleles: number, ratiob
		Homozygous: number, ratio	Heterozygous: number, ratio	Bi-allelic: number, ratio
43	26, 60.47%	14, 53.85%	3, 11.54%	9, 34.62%	3, 11.54%

^a Percentages were calculated over the total number of transgenic plants.

^b Percentages were calculated over the total number of plants with mutations.

Fig. 1.

Development of large grains through CRISPR/Cas9-mediated miR396-resistant mutants of the OsGRF4 allele. (A) Gene structure of OsGRF4 in ‘S143’ (WT) and the miR396 target site. The sgRNA targeting the flanking sequences of the recognition site is indicated by red letters. The miR396 recognition sequence is marked in grey background. (B) Mutations in T₀ plants harbouring in-frame osgrf4 and frame-shift osgrf4 variants. The introduced deletions and insertions are indicated by black dashes and red letters, respectively. Numbers on the right side indicate the lengths of indels compared with WT. –: deletion; +: insertion; combined mutations are distinguished by ‘/’. (C) Grain morphology of T₀ mutant plants and the corresponding WT. Scale bar: 0.5 cm. (D) Expression of OsGRF4 in young panicles of T₀ mutant plants. Data are means ± SD. **P < 0.01 compared with WT using Student’s t-test. NS: no significant difference.

We aimed to screen in-frame mutations in which the number of insertion or deletion nucleotides were in multiples of three. Among the T₀ mutants, three plants that contained an in-frame osgrf4 allele were identified (Fig. 1B; Table 1). Compared with the WT OsGRF4 protein, the in-frame mutants (grf4-#2, grf4-#8, grf4-#25) had minor deletions ranging from 1–3 amino acids (Supplementary Fig. S2). These results indicated the possibility that miR396-resistant mutants with in-frame osgrf4 might have been successfully obtained by the CRISPR/Cas9 system.

Putative off-target analysis and T₀ rice plants of in-frame osgrf4 mutants exhibit increased grain size phenotype

The miR396-resistant plants were evaluated for potential off-target mutations. We examined five potential off-target sites carrying two to four mismatched bases retrieved by the online tool CRISPR-P (http://skl.scau.edu.cn/offtarget/). To further confirm whether the target sgRNA affected OsMIR396 and other family genes additionally, both the OsMIR396 members and other OsGRF genes were also detected (Supplementary Table S3). Then the PCR products that were amplified from the osgrf4 mutants were sequenced (all primers are listed in Supplementary Table S1), and no mutation events were found across all potential off-target sites, OsMIR396 and other OsGRF genes (Supplementary Table S3), showing that the OsGRF4-targeting sgRNA had high specificity in inducing miR396-resistant variants.

In T₀ mutants harbouring in-frame osgrf4 variants (grf4-#2, grf4-#25), the grain shape was larger than that of ‘S143’ WT (Fig. 1C). The corresponding OsGRF4 transcripts were obviously elevated (Fig. 1D), indicating that OsGRF4 was protected from the cleavage of miR396, and the function was preserved. Similarly, the expression of OsGRF4 in the frame-shift mutant grf4-#14 was increased. There was no significant difference (P>0.05) in the OsGRF4 transcripts between WT and grf4-#1, which may be because the 1 bp insertion did not affect the regulation of miR396. However, the frame-shift osgrf4 mutants produced inactivated OsGRF4 and thus exhibited smaller grain traits (Fig. 1C, D; Supplementary Fig. S3A). To evaluate the effects of miR396 on other target genes after the de-repression of OsGRF4 and the expression of OsMIR396 members, we tested the transcripts of OsGRFs and OsMIR396 genes (Supplementary Fig. S4A, B). The results showed that there were no significant differences (P>0.05) in the expression of other target genes and OsMIR396 members between WT and in-frame mutants (Supplementary Fig. S4C, D; Fig. S3B). These findings corroborated that our method could make the specific target gene escape from miRNA suppression, thereby effectively improving particular traits.

Inheritance of miR396-resistant variants analysis in T₁ generation

To investigate the pattern of transmission of miR396-resistant mutants and segregation of the Cas9/sgRNA T-DNA, several T₁ progeny were obtained by strict self-separation and used for testing, including in-frame variants (grf4-#2, grf4-#25) and frame-shift mutants (grf4-#1, grf4-#14). In all T₁ progeny of frame-shift variants, both the homozygous mutant grf4-#1 and the heterozygous mutant grf4-#14 showed the same phenotypes as that of respective T₀ generations (Table 2). However, seeds of the T₁ bi-allele in-frame mutant plant grf4-#25 showed the same large-grain phenotype as the T₀ plant; besides, some T₁ progeny displayed small-grain traits due to the isolation of the +1 bp genotype (Fig. 2). These data indicated that the mutations in the T₀ lines were stably transferred to the T₁ generation and the phenotypes of grain size were consistent with the corresponding mutations. In addition, the identification of the Cas9/sgRNA T-DNA by PCR amplification determined the presence or absence of the transgene. The results showed that the T-DNA region could be segregated out in most T₁ lines. Moreover, the transgene-free plants with in-frame osgrf4 mutants were observed in both T₁ progenies of grf4-#2 and grf4-#25 (Table 2).

Table 2.

Segregation patterns of the introduced mutations during the T₀ to T₁ generation

Line	T0		T1
	Zygositya	Genotypeb	No. of plants tested	No. of plants with mutation phenotypec	Cas9 positive: Cas9 negative	No. of transgene-free plants with in-frame osgrf4 allele
grf4-#2	Bi-allele	+1,-6	24	5(+1); 12(+1,-6); 7(-6)	13:11	9
grf4-#25	Bi-allele	+1,-3	24	7(+1); 11(+1,-3); 6(-3)	15:9	6
grf4-#1	Homozygote	+1	24	24(+1)	18:6	0
grf4-#14	Heterozygote	-7	24	6(WT); 12(-7,WT); 6(-7)	19:5	0

^a The zygosity of homozygote, heterozygote and bi-allele in T₀ plant lines were putative.

^b The numbers indicate the lengths of indels compared with OsGRF4 gene of wild type. +: insertion; -: deletion;,: Bi-allelic mutations^.

^c The numbers in brackets indicate the lengths of insertions or deletions compared with OsGRF4 gene of wild type.

Fig. 2.

Segregation analysis of T₁ progeny of in-fame grf4-#25 mutant. (A) Alignment of sequences of the indels site in T₀ and T₁ progeny of grf4-#25. †Bi-allelic sequences were decoded from sequencing chromatograms using an online program DSDecode (http://dsdecode.scgene.com/). (B) Examples of sequencing chromatograms from different mutants presented in (A). (C) Grain phenotypes of segregations in T₁ progeny of grf4-#25. Scale bar: 1 cm.

The analyses of yield traits for T₂ progeny of miR396-resistant mutant plants

To further examine the characteristics of miR396-resistant mutations in later generations, the phenotypes of T₂ plants were analysed in detail. The transgene-free homozygous mutants for the OsGRF4 were selected to investigate the yield-associated traits (Fig. 3). Consistent with the roles of the OsGRF4 allele in regulating many important agronomic traits as reported previously (Duan et al., 2015; Hu et al., 2015; Che et al., 2016; Chen et al., 2019), grf4-#2 and grf4-#25 carrying in-frame osgrf4 had a significantly enlarged (P<0.01) grain volume that was 20.21% longer and 10.56% wider than WT (Fig. 3A, C, D), resulting in a remarkable increase (23.08%) of the 1000-grain weight (Fig. 3E). In contrast, the grain size of the frame-shift osgrf4 mutants, grf4-#1 and grf4-#14, was obviously smaller and lighter than that of WT plants (Fig. 3A, C-E). As demonstrated by scanning electron microscopy of the grain husk, the volume and number of outer epidermal cells in the in-frame mutants were enlarged and increased compared with the WT (Supplementary Fig. S5). In contrast, the cell size and cell number of the frame-shift variants were significantly reduced (P<0.01) in the outer surface of glumes (Supplementary Fig. S5). Results showed that in-frame osgrf4 could remarkably increase grain size by promoting cell elongation and cell proliferation(Supplementary Fig. S5), which confirmed that OsGRF4 positively regulated cell size and cell number to enlarge grain size (Hu et al., 2015).

Fig. 3.

Phenotypes and grain yield analysis of T₂ generation mutants. (A) Grain morphology of osgrf4 mutants and the wild type ‘S143’ (WT). Scale bar: 1 cm. (B) Morphological phenotypes of osgrf4 mutants and the corresponding WT. Scale bar: 20 cm. (C-G) Statistical data of the grain length (C), grain width (D), 1000-grain weight (E), plant height (F), till number (G), and grain yield per plant (H) of the WT and osgrf4 variants. Values are means ± SD. **P < 0.01 compared with WT using Student’s t-test. NS: no significant difference.

Gross morphologies of the osgrf4 mutants were similar to WT plants (Fig. 3B). No significant differences (P>0.05) in plant height and tiller number were observed among the WT, in-frame and frame-shift mutants (Fig. 3F, G). Furthermore, to investigate the actual grain yield, we used 24 plants in the corresponding lines. As shown in Supplementary Fig. S6, the in-frame osgrf4 panicle length were significantly longer (P<0.05) than WT. By contrast, branch number and grain number per panicle of in-frame osgrf4 mutants were similar to those of WT plants. Thus, the in-frame osgrf4 variants were notably increased by 24.15% in the grain yield per plant (Fig. 3H). Conversely, the yield per plant of frame-shift mutants was significantly reduced (P<0.01) because obvious changes had taken place in the yield-associated traits (Fig. 3H; Supplementary Fig. S6). Overall, we could effectively change the regulation of OsGRF4 on grain size and rice yield by editing the miR396 recognition site, so that it would be a feasible method for rice genetic breeding.

Application of the de-regulation of OsGRF4 in rice breeding

To evaluate the applicability of this strategy in genetic improvement of grain shape, 30 rice accessions were collected and sequenced for the miR396 recognition site of OsGRF4 allele. However, none of these rice cultivars contained the base substitutions (Supplementary Fig. S7), implying that our strategy had a great potential to take advantage of OsGRF4 allele in developing the high-yield elite varieties. Notably, the sgRNA of OsGRF4 gene designed in this study was applicable to most rice varieties (Supplementary Fig. S7), which would meet the requirements of employing the CRISPR/Cas9 system to specifically de-repress the OsGRF4 gene and improve grain size in the interested cultivars.

BPH resistance increased in the miR396-resistant line with in-frame osgrf8 variants

To verify the feasibility of this strategy, we selected another functional gene inhibited by miRNA for further investigations. Previous studies reported that miR396 regulated OsGRF8 to affect brown planthopper (BPH) resistance (Dai et al., 2019). Consequently, a sgRNA targeting the flanking sequences of the recognition site in the OsGRF8 allele was designed to circumvent miR396 suppression (Fig. 4A; Supplementary Fig. S1B). Both the in-frame and frame-shift mutants disrupted the miR396 recognition sequence and escaped the inhibition by miR396 (Fig. 4B). Hence, OsGRF8 transcripts were significantly elevated (P<0.01) in the mutant plants (Fig. 4C; Supplementary Fig. S3C). However, OsGRF8 in the frame-shift mutants were inactivated and lost their normal functions. In addition, to investigate the specificity of the OsGRF8 target site, the OsMIR396 genes and other OsGRF family alleles were selected (Supplementary Table S4). We found that all osgrf8 variants had no mutations in the OsMIR396 and other OsGRF genes, probably because the fourth base of the sgRNA was unique, resulting in a high specificity of the OsGRF8 target site.

Fig. 4.

Development of BPH resistance rice through CRISPR/Cas9-mediated miR396-resistant mutants of the OsGRF8 allele. (A) Gene structure of OsGRF8 in ‘S143’ (WT) and the miR396 target site. The sgRNA targeting the flanking sequences of the recognition site is indicated by red letters. (B) Mutations in T₀ plants harbouring in-frame Osgrf8 and frame-shift Osgrf8 variants. The miR396 recognition sequence was marked in grey background. The introduced deletions and insertions are indicated by black dashes and red letters, respectively. Numbers on the right side indicate the lengths of indels compared with WT. –: deletion; +: insertion; combined mutations are distinguished by ‘/’. (C) Expression of OsGRF8 in leaves of T₀ mutant plants compared with WT. (D) Individual tests to determine the BPH resistance of the WT, in-frame variants and frame-shift mutants. Scale bar: 5 cm. (E) Statistical analysis of the survival rates of WT plants and Osgrf8 mutants after BPH feeding (n=3). (F) Expression of OsF3H in leaves of WT and in-frame Osgrf8 plants. (G) Quantitative determination of flavonoid content in the WT and in-frame Osgrf8 mutants (n=3). Data are mean ± SD. **P<0.01 compared with WT using Student’s t-test.

To evaluate the response to BPH, the T₂ generation of osgrf8 mutants and the WT ‘S143’ were selected for testing. After BPH infestation for 7 d, the in-frame mutants grf8-#3 and grf8-#9 showed greater resistance to BPH than the WT, and accordingly, the survival rate increased (Fig. 4D, E). On the contrary, the frame-shift lines grf8-#1 and grf8-#5 were the most sensitive to brown planthopper (Fig. 4D, E). Subsequently, the expression of OsF3H activated by OsGRF8 was detected in the WT and in-frame plants, which positively correlated with flavonoid content and BPH resistance. As expected, the expression of OsF3H was elevated in the in-frame variants (Fig. 4F; Supplementary Fig. S3D). Correspondingly, the increased flavonoid content was also found in the in-frame plants (Fig. 4G). These findings confirmed that the strategy could introduce the miR396-resistant mutants in rice and improve BPH resistance in the in-frame mutants.

Discussion

Plant miRNAs have recently presented as promising targets for crop improvement, and the target mimicry and STTM have been developed to inhibit the activity of miRNAs. However, neither of the MIM or STTM approaches guarantees the strongest silencing efficacies of miRNA activity (Liang et al., 2014; Reichel et al., 2015). The 1 bp mutations mediated by CRISPR/Cas9 are a little difficult to block the suppression of miRNA on the target genes, due to the fact that the base-frame shift may not impact the process of miRNA maturation, and may generate the functionally redundant miRNAs (Zhang et al., 2021; Zhou et al., 2017). Meanwhile, these miRNA-target technologies probably produce the compounded phenotypes when a miRNA possesses multiple target genes (Zhang et al., 2017). To precisely promote the expression of specific target genes regulated by miRNA, we make in-frame mutations on the miRNA recognition site, and circumvent miRNA suppression successfully. Given a list of studies reported that altering the expression of target genes repressed by miRNAs could significantly affect grain yield (Zhao et al., 2017; Jiang et al., 2018; Zhao et al., 2019), and change stress responses (Dai et al., 2019), it is conceivable that our strategy may be promising to accurately improve elite traits in crop breeding.

The strategy appears to be simple and precise for improving specific traits. However, we have to take into consideration whether the deletions of amino acids in the miRNA binding region disrupt the protein function. If so, this strategy will not work. OsGRF4 belongs to the GRF transcription factor family, and possesses conserved QLQ, WRC and TQL domains (Li et al., 2016). The in-frame mutants, grf4-#2, grf4-#8 and grf4-#25, has amino acids missing ranging from 1–3 in the C-terminal of the WRC domain (Supplementary Fig. S2). However, the deletions avoided destroying the CCCH zinc-finger motif, which preserves the function of the WRC domain. Because the conserved QLQ, WRC and TQL domains were maintained (Supplementary Fig. S2), the in-frame variants still retained the regulation of OsGRF4 on the grain size. Therefore, the in-frame variants could precisely manipulate agronomic traits regulated by miRNAs, on the premise that the major domains were retained.

Previous studies reported that GS^AA, which was substituted at the miR396 recognition site of OsGRF4, increased the grain size (Duan et al., 2015; Hu et al., 2015; Che et al., 2016; Chen et al., 2019). But the in-frame osgrf4 mutants we obtained had amino acid deletions. In order to identify whether the emerged large-grain phenotype was caused by the newly generated protein function or by miR396 de-regulation, we designed a target site sgRNA-2 on the miR396-complementary sequence of OsGRF4 (Supplementary Fig. S8A). The in-frame mutants, grf4-2-#7 and grf4-2-#46, which had lost amino acids 12 and 7 respectively, produced different OsGRF4 proteins (Supplementary Fig. S8B, C). However, the T₁ progeny of grf4-2-#7 and grf4-2-#46 still displayed a large-grain phenotype, consistent with the traits of grf4-#2 and grf4-#25 (Supplementary Fig. S8D, E). These results indicated that the large-grain phenotype of in-frame osgrf4 mutants were caused by getting rid of miR396 inhibition rather than the new protein functions.

The miRNA-resistant mutants with in-frame targets relieve the influence of endogenous miRNA regulation, elevating the expression of corresponding target genes, and manipulating agronomic traits precisely. Most of the knockout mutants mediated by CRISPR/Cas9 system usually produce 1 bp indels between the third and fourth bases upstream of the Protospacer Adjacent Motif (PAM ) region (Zhang et al., 2014). Whereas, the acquisition rate of mutants with 3n bp indels is relatively low. To increase the number of in-frame mutants, we can introduce two adjacent sgRNAs simultaneously whose intervals between PAM sequences is 3n bp. Thereby, the likelihood of the missing bases harbouring a multiple of three would increase (Guo et al., 2018). Similar to obtaining miRNA-resistant targets, the improved base editing and prime editing technologies can be designed to replace the miRNA recognition sequence of the target gene, which seem to have more potential (Kim et al., 2017; Anzalone et al., 2019). However, the unsatisfactory editing efficiency (Butt et al., 2020; Xu et al., 2020) and the off-target editing effects of cytosine base editor have limited their applications in crop breeding presently (Jin et al., 2019). Thus, employing CRISPR/Cas9 system to accurately develop specific traits inhibited by miRNAs is promising.

In summary, CRISPR/Cas9-mediated destruction of miR396 inhibition elevated the OsGRF4 and OsGRF8 transcripts in the corresponding in-frame variants, leading to the enlarged grain size and increased BPH resistance. These results proved that our strategy could be effectively employed to precise de-regulation of target genes repressed by miRNAs and would be of great significance for production of elite traits in plants.

Supplementary data

The following supplementary data are available at JXB online.

Fig. S1. Sequences of the third exon of the OsGRF4 and OsGRF8 alleles in ‘S143’.

Fig. S2. Multiple alignment of the deduced amino acid sequences of in-frame osgrf4 mutants.

Fig. S3. qRT–PCR analysis of a second reference gene UBIQUITIN.

Fig. S4. Analysis of the expression of miR396 target genes and OsMIR396 members.

Fig. S5. Scanning electron microscopy observation of spikelet lemma.

Fig. S6. Analyses of panicle traits for wild type ‘S143’ and osgrf4 mutants.

Fig. S7. Alignment of the miR396 complementary site in OsGRF4 from rice conventional varieties.

Fig. S8. Development of large grain through the sgRNA-2 of the OsGRF4 allele.

Table S1. Primers used in this study.

Table S2. Genotypes of T₀ mutant plants in ‘S143’.

Table S3. Evaluation of sgRNA-OsGRF4 potential off-target sites.

Table S4. Evaluation of sgRNA-OsGRF8 potential off-target sites.

Author contributions

YL, YZ, FW, and LC conceived the project and designed the experiments; YL, YZ, YC, RC, ZC, GL, MF, YL, XG, and XZ performed the experiments; YL, YZ, and FW analysed the data; YL, YZ, and FW wrote the manuscript.

Funding

This work was supported by the Genetically Modified Organisms Breeding Major Projects, China (2016ZX08001001), National Key Research and Development Program of China (2016YFD0100903) and Natural Science Foundation of Fujian Province (2015J01055).

Data availability

All data supporting the findings of this study are available within the paper and within its supplementary materials published online.

Author notes

These authors contributed equally to this article.