*2.1. Generation of Rice FWL Gene Mutants Using CRISPR*/*Cas9*

Two target sites were designed in the coding region of each of the eight rice *FWL* genes for CRISPR/Cas9 gene editing (Table 1). The GC content in these target sites was in the range 45–75%. The synthesized oligos were inserted into the CRISPR/Cas9 binary vector (Figure S1). Subsequently, the 16 constructed vectors were transformed into the *Japonica* rice variety Zhonghua 11 using the *Agrobacterium*-mediated method.

Of the 16 vectors, we successfully generated transgenic T0 lines for 15 vectors (Table 2). We detected targeted mutations in all those T0 lines. The mutation rates varied from 26.7% to 100%, and the average mutation rate was 81.6% (Table 2), suggesting that the CRISPR/Cas9 system constructed in this study is efficient in rice gene editing. Bi-allelic mutants were detected in T0 plants from each vector, with detection percentages varying from 20.0% to 87.5% (Table 2). Homozygous mutants were detected in T0 plants from 13 vectors, with the highest detection percentage being 64.3%. By contrast, heterozygotes and chimeras were detected only in T0 plants from three vectors (Table 2). The percentage of heterozygotes and chimeras in all T0 plants was only 1.3% and 4.0%, respectively. Detailed sequencing results of all T0 mutants are shown in Table S1.


**Table 1.** Target sites of rice *FWL* genes for clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9)-mediated gene editing.

<sup>1</sup> The protospacer adjacent motif (PAM) sequences are shown in green.

**Table 2.** Identification of targeted mutations in T0 plants.


Sequencing analyses revealed that most mutations were short indels; 62.3% of indels were 1 bp changes (Figure 1A,B). A majority of the 1 bp insertions (83.2%) were either A or T, which is consistent with previous reports [27,29].

Of the 223 T0 plants, 41 plants did not contain mutations. To test whether failed editing of these plants was caused by a lack of the CRISPR/Cas9 construct, the presence of *hygromycin phosphotransferase* (*HPT*), sgRNA, and *Cas9* transgenes in these 41 plants was examined. Two plants did not contain *HPT*, sgRNA, and *Cas9* sequences (Table S2), which suggests that these plants escaped hygromycin selection. Twenty-five plants did not contain sgRNA and/or *Cas9* sequences (Table S2) which suggests that incompleteness of the sgRNA/*Cas9* expression cassette led to failed mutagenesis in these plants. Interestingly, when unmutated T0 plants without the complete sgRNA/*Cas9* construct were excluded, all targets except Osfwl1a, Osfwl4a, and Osfwl6a had a mutation rate of 100% (Table 2 and Table S2). The score of sgRNA activity in all targets predicted using the sgRNA Scorer 2.0 varied from −0.64 to 1.09 (Table S3), indicating moderate efficiency of the sgRNAs [30].

**Figure 1.** Characterization of on-target and off-target mutations. (**A**) Frequencies of different types of on-target mutations. (**B**) Frequencies of different lengths of on-target mutations. (**C**) Zygosity of off-target mutations. (**D**) Types of off-target mutations. Legend: i, insertion; d, deletion; s, substitution; c, combined mutation; Ho, homozygote; Bi, bi-allele; He, heterozygote.

The inheritance patterns of targeted mutations in later generations were also examined. Mutations of most homozygous T0 plants were stably transmitted to the T1 generation (Table S4). However, unexpected genotypes were detected in the T1 generation of four of the five bi-allelic T0 plants. Additionally, a large proportion of the progeny of a chimeric T0 plant were chimeras (Tables S4 and S5). The transmission of mutations of several randomly selected T1 lines that did not contain transgenes ('transgene-free'; see Section 2.2) in the T2 generation was also examined. The genotypes of all these lines were faithfully transmitted to T2 plants (Table S6).
