Targeted genome modifications in cereal crops

Abstract

The recent advent of customizable endonucleases has led to remarkable advances in genetic engineering, as these molecular scissors allow for the targeted introduction of mutations or even precisely predefined genetic modifications into virtually any genomic target site of choice. Thanks to its unprecedented precision, efficiency, and functional versatility, this technology, commonly referred to as genome editing, has become an effective force not only in basic research devoted to the elucidation of gene function, but also for knowledge-based improvement of crop traits. Among the different platforms currently available for site-directed genome modifications, RNA-guided clustered regularly interspaced short palindromic repeats (CRISPR)-associated (Cas) endonucleases have proven to be the most powerful. This review provides an application-oriented overview of the development of customizable endonucleases, current approaches to cereal crop breeding, and future opportunities in this field.

Introduction

According to the United Nations, the world population is expected to rise by 2 billion and reach 9.8 billion by 2050, which will increase the demand for food by 60% and require gains in production yield (FAO et al. 2018). Due to climate change and soil degradation, arable land that would otherwise support the growth of crops is being lost, thereby greatly limiting the food production chain. Crop breeding is a critical avenue for coping with these global challenges by harnessing genetic resources, such as naturally occurring and artificially generated variants, to increase yield and plant resilience against abiotic stress, pathogens, and pests.

Throughout the history of human farming, crop domestication and the development of new crop varieties have relied on spontaneous mutations in existing cultivated germplasm that led to more productive farming and/or were associated with improved product utility. In addition to passive selection of emerging traits in one crop species, targeted crossbreeding also opened the door to combining useful traits from different germplasm. However, spontaneous mutation rates are low across the genome and rarely affect a gene with potential use for crop improvement. To accelerate the discovery of agronomically relevant genetic diversity, methods aimed at inducing random mutations have been developed. Crop plants have been subjected to ionizing radiation such as X-rays, gamma-rays, or heavy ion beams, as well as chemical mutagens like ethyl methanesulfonate or sodium azide (Ahloowalia and Maluszynski 2001). To date, many induced mutant lines have been generated and incorporated into several crop breeding programs. For example, the short-straw barley (Hordeum vulgare) cultivar ‘Diamant’ was produced via gamma-ray irradiation; the causal denso mutation has been since introgressed into over 100 cultivars grown worldwide (Ahloowalia et al. 2004). However, the desired mutations are accompanied by many additional mutations randomly distributed across the genome, some of which remain due to linkage. In addition, identification of the causal gene is typically challenging, even when using segregating populations and selecting for the phenotype of interest, as the candidate interval is likely to contain more than one mutation, only one of which is causal. These problems were partially alleviated by the development of the Targeted Induced Local Lesions in Genomes (TILLING) method, which allows for the identification of mutations in a given candidate gene by reverse genetics on pooled DNA from mutant populations (McCallum et al. 2000). Efficient selection methods based upon DNA marker-trait associations were also developed and broadly embraced in modern plant breeding (Rasheed et al. 2017). In addition, advanced sequencing technologies have become indispensable for providing comprehensive genome data in all crop species (Bayer et al. 2020). Taking advantage of these resources, several new methods have emerged that facilitate the rapid cloning of genes of interest, even in the context of cereals with comparatively large genomes such as mapping-by-sequencing using MutMap, mutant chromosome sequencing (MutChromSeq), mutagenesis with resistance gene sequence capture (MutRenSeq), targeted chromosome-based cloning via long-range assembly (TACCA), and association genetics with resistance gene enrichment sequencing (AgRenSeq) (Abe et al. 2012, Arora et al. 2019, Bettgenhaeuser and Krattinger 2019, Sánchez-Martín et al. 2016, Steuernagel et al. 2016, Thind et al. 2017).

The development of genetic transformation techniques based upon Agrobacterium (Agrobacterium tumefaciens)-mediated or direct delivery of recombinant DNA to totipotent plant cells has opened another chapter in crop breeding history (Kumlehn and Hensel 2009). This technology makes it possible to ectopically express, overexpress, or downregulate the expression of genes of interest to achieve desirable modifications of plant traits. In addition, the ability to easily introduce transgenes into most plant backgrounds has greatly facilitated basic plant research aimed at deciphering gene function (Kumlehn et al. 2019). The Flavr Savr tomato (Solanum lycopersicum), the first genetically engineered crop, was released for commercial use in 1994. Several other genetically engineered crops have since been developed. However, their commercial use is still largely confined to herbicide- and pest-resilient varieties due to public concern about the potential but unproven human risks associated with this technology, even though each new genetically engineered crop goes through a comprehensive and costly review process prior to approval.

More recently, emerging customizable endonucleases have begun to pave the way for the introduction of mutations or predefined sequence modifications at any genomic position, followed by the removal of transgenes by simple segregation in the progeny (Koeppel et al. 2019). This new technology, which is commonly referred to as genome editing, has already been implemented and demonstrated to work in most crop species. Customizable endonucleases hold great promise to greatly accelerate mutation breeding in crops, either via the precise introduction of previously known allelic states conferring desired traits into advanced breeding backgrounds or by the generation of novel genetic diversity at target genes. In this review, we introduce this novel technology to readers and provide examples of its applications in the major cereal crops rice (Oryza sativa), maize (Zea mays), wheat (Triticum aestivum), and barley. Finally, we discuss the potential of employing customizable endonucleases for future plant breeding programs.

Principles of targeted genome modification

To ensure the survival of organisms in the face of DNA damage, their constituent cells must repair DNA breaks efficiently, which mostly relies on any one of three major repair mechanisms that are equally relevant for site-directed genome modification methods (Kumlehn et al. 2018, Fig. 1). Non-homologous end-joining (NHEJ) repair is the predominant cellular repair mechanism for DNA double-strand breaks in plants. NHEJ involves the recognition and re-ligation of free DNA ends, which is error-prone and therefore leads to insertions and deletions of usually one or a few nucleotides. NHEJ repairs the majority of events induced during site-directed mutagenesis efforts, after the customized endonuclease has targeted and cleaved the intended site of modification, although the resulting modification itself is random. Other repair mechanisms rely on sequence homology and result in more predictable repair outcomes. Even small sequence repeats located on either side of the DNA break induced by the nuclease allow for their overlapping annealing, since opposite single strands of both DNA ends are complementary along these repeats; any remaining single-stranded DNA ends are degraded before the newly assembled double strand is cured by ligation. This mechanism of microhomology-mediated end-joining (MMEJ) repair is very common in plants and produces precise, predictable deletions of one of the repeat sequences and the sequence between them. Another virtually error-free basic mechanism of DNA repair is based on homologous recombination (HR) between one damaged DNA region and the corresponding and intact sequence of the sister chromatid or a homologous chromosome, resulting in longer stretches of identical sequences than in MMEJ. HR may be harnessed to introduce a foreign DNA fragment as long as it is flanked by regions with homology to the region surrounding the site of DNA cleavage. While HR is essential for meiosis, the underlying enzymes are comparatively rarely active in somatic plant cells. In addition, the sister chromatids that are predominantly recruited as native repair templates are exclusively present in the G2 phase of the cell cycle. These limitations contribute to the current difficulties in implementing HR in plant genetic engineering and for precise genome editing in crop plants.

Fig. 1.

Diagram summarizing the various ways in which customizable endonucleases can be used to introduce modifications into target genes. Customizable endonucleases trigger double-strand breaks in selected target sites of the plant genome. The DNA is then repaired via one of three major mechanisms, resulting in modifications at the target site: I, non-homologous end-joining (NHEJ) repair; II, microhomology-mediated end-joining (MMEJ) repair; and III, homologous recombination (HR). These mechanisms result in possible insertions, deletions, and/or substitutions of individual or multiple nucleotides. The size of the insertion or deletion is random following NHEJ-mediated DNA repair, whereas MMEJ and HR lead to predictable modifications and can thus be utilized for precise genome editing.

Site-directed genome modification is initiated by the cleavage of a given target sequence in genomic DNA by customizable endonucleases. Several effective endonucleases are available: meganucleases, zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and clustered regularly interspaced short palindromic repeats (CRISPR)-associated (Cas) endonucleases. Meganucleases, such as I-SceI, I-CreI, and I-DmoI, include a transit peptide for nuclear localization, a site-specific double-stranded DNA-binding domain of 12 to 40 nucleotides, and an endonuclease domain to cleave DNA. Although meganucleases have high target sequence specificity, the small number of DNA-binding sequences identified makes it difficult to design a derivative that targets other sequences of interest, unlike ZFNs, TALENs, or Cas endonucleases (see below). These nucleases are typically deployed in plant cells by introducing a transgene including a promoter that drives their expression and is recognized by RNA polymerase II, such as the cauliflower mosaic virus (CaMV) 35S or UBIQUITIN promoters. ZFNs and TALENs are chimeric proteins composed of a DNA-binding domain fused to a nuclease domain derived from the restriction enzyme FokI. The DNA-binding domain of ZFNs is composed of three to six zinc-finger modules derived from DNA-binding transcription factors. Each zinc-finger module specifically recognizes a unique nucleotide triplet. By contrast, the DNA-binding domain of TALENs consists of approximately 20 modules of four basic types, each with specificity to one of the four nucleobases A, C, G, or T (Gurushidze et al. 2014). This DNA-binding principle was adapted for biotechnological use from transcription activator-like effectors produced and delivered by plant pathogenic bacteria to bind to specific host genomic sequences to manipulate gene expression in infected cells. Due to differences in the architecture of the modules, the design and construction of TALENs is more straightforward compared to ZFNs (Budhagatapalli et al. 2016, Hensel and Kumlehn 2019). Since FokI only cleaves DNA as a dimer, two versions of ZFNs or TALENs are typically expressed, with each monomer binding to a DNA sequence on either side of the intended cleavage site and separated by an appropriate distance to direct endonucleolytic cleavage.

Cas endonucleases were originally discovered as components of microbial CRISPR-Cas immune systems. SpCas9 isolated from the bacterium Streptococcus pyogenes is currently the most widely used Cas endonuclease for biotechnological purposes due to its high cleavage efficiency within a physiologically relevant range of temperatures in a variety of hosts. The deployment of Cas endonucleases in microbial cells relies on a single-stranded guide RNA (gRNA) derived from the same native immune system that targets a DNA sequence that is recognized as foreign and must therefore be eliminated. The gRNA can be customized to target any sequence of interest for Cas-dependent cleavage. In most SpCas9 applications, the gRNA is a single-stranded chimeric RNA harboring two parts whose native predecessors were independent molecules: the CRISPR RNA (crRNA), which includes ~20 bp of customizable sequence that binds to the DNA target motif via complementary base-pairing, and trans-activating CRISPR RNA (tracrRNA) scaffold, which binds to the Cas endonuclease. In general, polymerase II-type promoters are used to drive the expression of the genes encoding Cas endonucleases, while polymerase III-type promoters from small non-coding RNAs are mainly used to express gRNAs. A Cas endonuclease target motif is not confined to the sequence bound by the gRNA 5ʹ end via complementary base-pairing, as it must also include a short binding site for the Cas endonuclease. In the case of SpCas9, this so-called protospacer-adjacent motif (PAM) consists of two guanine nucleobases preceded by any nucleobase (i.e., NGG). This general pattern must therefore be given some thought when searching for appropriate Cas endonuclease target motifs within the host genome. Several powerful online platforms are now available that greatly facilitate the identification of target sequences within genes of interest based on experimental considerations such as the Cas endonuclease variant and the host species used (e.g., WU-CRISPR, http://crisprdb.org/wu-crispr/; CRISPR-P 2.0, http://crispr.hzau.edu.cn/CRISPR2; CRISPOR, http://crispor.tefor.net; and CRISPR Guide RNA Design, https://www.benchling.com/crispr).

Establishment of targeted genome modification in cereals

Among cereal crops, three types of nucleases mentioned above (meganucleases, ZFNs, and TALENs) were used to generate targeted genome modifications prior to the advent of Cas endonucleases. The first example of targeted genome modification using customizable endonucleases was by D’Halluin et al. (2008), who succeeded in inserting the 35S promoter upstream of a promoterless herbicide resistance transgene using a meganuclease in maize. The following year, Shukla et al. (2009) reported the first use of ZFNs for site-directed mutagenesis at the maize INOSITOL-1,3,4,5,6-PENTAKISPHOSPHATE 2-KINASE 1 (IPK1) gene as well as site-directed DNA insertion of a PHOSPHINOTHRICIN ACETYLTRANSFERASE (PAT) gene at the same locus. For TALENs, Li et al. (2012) were the first to successfully modify the promoter sequence of the bacterial leaf blight susceptibility gene SUGARS WILL EVENTUALLY BE EXPORTED TRANSPORTER 14 (OsSWEET14) in rice.

Cas9 endonuclease-based targeted genome modification was first achieved in rice and wheat by Shan et al. (2013), Miao et al. (2013), and Feng et al. (2013), who used SpCas9 to target genes associated with chlorophyll biosynthesis and thus with easily scorable phenotypes: PHYTOENE DESATURASE (PDS), CHLOROPHYLL A OXYGENASE 1 (CAO1), and YOUNG SEEDLING ALBINO (YSA).

Besides SpCas9, endonucleases derived from other CRISPR systems have also been co-opted for targeted genome modification. For example, Endo et al. (2016), Tang et al. (2017b), and Yin et al. (2017) induced mutations by expressing the Cas12a (CRISPR from Prevotella and Francisella 1 or Cpf1) endonuclease in rice. In contrast to Cas9, Cas12a functions with a native single crRNA. Like several other types of Cas endonucleases, Cas12a broadened the choice of possible target sequences because it requires a T-rich PAM (5ʹ-(T)TTV-3ʹ) that resides upstream of the nucleotides bound by the gRNA. Unlike Cas9, Cas12a results in staggered cleavage of the target sequence.

A single nucleotide substitution method called base editing has also been developed. Base editing relies on Cas derivatives (so-called nickases) that cleave only one of the two DNA strands, after which a fused cytidine deaminase or adenosine deaminase domain converts a single nucleobase. For example, Li et al. (2018) expressed a Cas9-adenosine deaminase fusion in rice and wheat and introduced T/A to C/G conversions in target genes.

More recently, another approach for precise genome modifications called prime editing was developed, based upon chimeric fusion proteins between a Cas9 nickase and a reverse transcriptase. Such proteins are coupled with gRNAs carrying a 3ʹ extension that acts as a primer to specify the replacement of particular nucleobases within the target sequence (Anzalone et al. 2019). This method was successfully implemented in rice and wheat among other plants (Hua et al. 2020, Lin et al. 2020).

Highly versatile, precise genome editing can be achieved using homology-based approaches in which customizable nucleases are used along with a repair template that carries DNA fragments with homology to the target region while also harboring the desired modified sequence to be incorporated at the genomic target site (Fig. 1). Proof-of-concept studies demonstrated precise genome editing using meganucleases and TALENs in barley, although these initial examples were confined to the modification of previously inserted transgenes (Budhagatapalli et al. 2015, Watanabe et al. 2015). Begemann et al. (2017) subsequently used the Cas12a endonuclease to perform site-directed DNA insertions in rice, while Gil-Humanes et al. (2017) developed a site-directed DNA insertion method using the replication system of wheat dwarf virus combined with the Cas9 endonuclease, which resulted in increased frequency of knock-in of foreign DNA at a genomic target region in wheat.

Another novel approach that also largely alleviates the need for stable transformation of the host species was reported by Budhagatapalli et al. (2020), who pollinated wheat plants with pollen from Cas9-gRNA-transgenic maize plants, resulting in haploid wheat plants that could be subjected to whole-genome duplication to produce a homozygous mutant and transgene-free wheat. This method was demonstrated using a panel of five diverse genotypes including durum wheat (Triticum durum) and common wheat.

Targeted genome modifications may not require the introduction of a transgene at all, relying instead on preassembled Cas9 or Cas12a nuclease protein-gRNA ribonucleoprotein complexes (Kim et al. 2017, Toda et al. 2019, Woo et al. 2015). Toda et al. (2019) introduced Cas9-gRNA ribonucleoproteins into totipotent rice zygotes to generate plants carrying mutant alleles at GRAIN WIDTH 7 (GW7), DROOPING LEAF (DL), and PSEUDO-RESPONSE REGULATOR 37 (PRR37). Likewise, Zhang et al. (2016) introduced in vitro-transcribed transcripts for Cas9 and the gRNA into common wheat by accelerated microparticles and obtained primary mutants. Subsequently, Liang et al. (2017) successfully performed targeted genome modification of GW2 in common wheat using Cas9-gRNA ribonucleoproteins.

Targeted genome modification for agronomic trait improvement in rice

Rice is one of the most important cereal crops worldwide, particularly in Asian countries, where it is consumed as a staple food. The rice genome is smaller than those of other cereal crops and was the first cereal genome to be sequenced. Rice plants are also easier to transform than other members of the Poaceae family, explaining why it has long been used as a model system for monocot plants. In recent years, rice has also been used as an experimental platform to test and implement novel targeted genome modification tools. Such tools have been used to modify genes related to high yield and production stability in ways that would not have been possible via conventional breeding. In this section, we describe how genome modification tools have been used to modify genes of high agronomic importance in rice.

Table 1 lists several examples of targeted genome modification for agronomically important traits in cereal crops. In two pioneering studies, Li et al. (2012) and Zhou et al. (2015) used TALENs and Cas9 endonuclease to modify the promoter regions of the bacterial leaf blight susceptibility genes SWEET14 and SWEET13, respectively. The resulting mutant plants showed increased resistance to Xanthomonas oryzae pv. oryzae, the pathogen that causes bacterial leaf blight in rice. Following these reports, Oliva et al. (2019), Xu et al. (2019), and Blanvillain-Baufumé et al. (2017) also created mutants that are highly resistant to the same pathogen by targeting the promoter regions of SWEET11, in addition to SWEET13 and SWEET14. In another study, Wang et al. (2016) used Cas9 nuclease to modify ETHYLENE RESPONSIVE TRANSCRIPTION FACTOR 922 (OsERF922), thereby increasing resistance to rice blast caused by Magnaporthe oryzae, one of the most destructive plant diseases worldwide. The authors further showed that the induced mutations were inherited by the progeny. Compared to wild-type plants, mutant lines exhibited much smaller blast lesions at both the seedling and tillering stages. Nawaz et al. (2020a) also successfully improved resistance to rice blast by modifying another gene, PYRICULARIA ORYZAE RESISTANCE 21 (PI21). Other examples of mutants generated via site-directed genome modification for higher tolerance to biotic stress were reported by Kim et al. (2019) and Li et al. (2020a) for bacterial blight resistance and by Macovei et al. (2018) for virus resistance.

Table 1. Site-directed modification of agronomically relevant genes in cereal crops

Plant	Modified gene	Purpose	Nuclease	DNA repair by	Reference
rice	ACC	Herbicide resistance	Cas9 nuclease	NHEJ	Liu et al. (2020)
	ALS	Herbicide resistance	TALENs, Cas9 nuclease	HDR/TSI	Li et al. (2016c), Sun et al. (2016)
	BADH2	Fragrant rice	TALENs, Cas9 nuclease	NHEJ	Shan et al. (2015), Ashokkumar et al. (2020)
	CRTI and PSY	Carotenoid accumulation	Cas9 nuclease	HDR/TSI	Dong et al. (2020)
	CSA	Male sterility	Cas9 nuclease	NHEJ	Li et al. (2016b)
	DEP1	Grain yield	Cas9 nuclease	NHEJ	Huang et al. (2018)
	ELF4G	Virus resistance	Cas9 nuclease	NHEJ	Macovei et al. (2018)
	EPSPS	Herbicide resistance	Cas9 nuclease	NHEJ/TSI	Li et al. (2016a)
	ERF922	Blast disease resistance	Cas9 nuclease	NHEJ	Wang et al. (2016)
	GA20OX2	Plant height	Cas9 nuclease	NHEJ	Nawaz et al. (2020b)
	GN1a	Grain yield	Cas9 nuclease	NHEJ	Huang et al. (2018)
	GS3	Grain yield	Cas9 nuclease	NHEJ	Zeng et al. (2020b)
	GW2	Grain yield	Cas9 nuclease	NHEJ	Xu et al. (2016)
	GW5	Grain yield	Cas9 nuclease	NHEJ	Xu et al. (2016)
	LCT1	Cadmium accumulation	Cas9 nuclease	NHEJ	Liu et al. (2019)
	LOX3	Storage tolerance	TALENs	NHEJ	Ma et al. (2015)
	MTL (PLA1)	Haploid production	Cas9 nuclease	NHEJ	Yao et al. (2018)
	MYB30	Cold tolerance	Cas9 nuclease	NHEJ	Zeng et al. (2020b)
	NRAMP5	Cadmium accumulation	Cas9 nuclease	NHEJ	Tang et al. (2017a), Liu et al. (2019), Yang et al. (2019)
	PI21	Blast disease resistance	Cas9 nuclease	NHEJ	Nawaz et al. (2020a)
	PIN5b	Grain yield	Cas9 nuclease	NHEJ	Zeng et al. (2020b)
	RC	Proanthocyanidin and anthocyanin accumulation	Cas9 nuclease	NHEJ	Zhu et al. (2019)
	RR22	Salinity tolerance	Cas9 nuclease	NHEJ	Zhang et al. (2019)
	SBEIIb	Starch composition	Cas9 nuclease	NHEJ	Sun et al. (2017)
	SD1	Plant height	Cas9 nuclease	NHEJ	Hu et al. (2019), Biswas et al. (2020)
	SE5	Plant height	Cas9 nuclease	NHEJ	Hu et al. (2019)
	SWEET11	Bacterial leaf blight resistance	Cas9 nuclease	NHEJ	Oliva et al. (2019), Xu et al. (2019)
	SWEET13	Bacterial leaf blight resistance	Cas9 nuclease	NHEJ	Zhou et al. (2015), Oliva et al. (2019), Xu et al. (2019)
	SWEET14	Bacterial leaf blight resistance	TALENs, Cas9 nuclease	NHEJ	Li et al. (2012), Blanvillain-Baufumé et al. (2017), Oliva et al. (2019), Xu et al. (2019)
	TGW6	Grain size	Cas9 nuclease	NHEJ	Xu et al. (2016), Han et al. (2018)
	TMS5	Male sterility	Cas9 nuclease	NHEJ	Zhou et al. (2016)
	VP1	Germination speed	Cas9 nuclease	NHEJ	Jung et al. (2019)
	WAXY	Starch composition	Cas9 nuclease	NHEJ	Han et al. (2018), Zhang et al. (2018a), Huang et al. (2020), Zeng et al. (2020a)
	XA13	Bacterial blight resistance	Cas9 nuclease	NHEJ	Kim et al. (2019), Li et al. (2020a)
wheat	α-GLIADINS	Low gluten	Cas9 nuclease	NHEJ	Sánchez-León et al. (2018)
	CM3	Low α-amylase/trypsin inhibitor (allergenic proteins)	Cas9 nuclease	NHEJ	Camerlengo et al. (2020)
	CM16	Low α-amylase/trypsin inhibitor (allergenic proteins)	Cas9 nuclease	NHEJ	Camerlengo et al. (2020)
	EDR1	Powdery mildew disease resistance	Cas9 nuclease	NHEJ	Zhang et al. (2017)
	GW2	Grain size	Cas9 nuclease	NHEJ	Zhang et al. (2018b), Wang et al. (2018a, 2018b)
	HRC	Fusarium head blight disease resistance	Cas9 nuclease	NHEJ	Su et al. (2019)
	MLO	Powdery mildew disease resistance	TALENs	NHEJ	Wang et al. (2014)
	MS1	Male sterility	Cas9 nuclease	NHEJ	Okada et al. (2019)
	MS45	Male sterility	Cas9 nuclease	NHEJ	Singh et al. (2018)
	NP1	Male sterility	Cas9 nuclease	NHEJ	Li et al. (2020b)
	NFXL1	Fusarium head blight disease resistance	Cas9 nuclease	NHEJ	Brauer et al. (2020)
	QSD1	Grain dormancy	Cas9 nuclease	NHEJ	Abe et al. (2019), Liu et al. (2021)
	SBEIIb	Starch composition	Cas9 nuclease	NHEJ	Li et al. (2021)
	SD1	Plant height	Cas9 nuclease	NHEJ	Budhagatapalli et al. (2020)
barley	β-1,3-GLUCANASE	Increased callose formation	Cas9 nuclease	NHEJ	Kim et al. (2020)
	MORC	Fungal resistance	Cas9 nuclease	NHEJ	Kumar et al. (2018)
	NUD	Non-adherent hull	Cas9 nuclease	NHEJ	Gasparis et al. (2018), Gerasimova et al. (2020)
maize	AAD1	Herbicide resistance	ZFNs	HDR/TSI	Ainley et al. (2013)
	ARGOS8	Drought stress tolerance	Cas9 nuclease	HDR/TSI	Shi et al. (2017)
	GA20OX3	Plant height	Cas9 nuclease	NHEJ	Zhang et al. (2020)
	IPK1	Phytate accumulation in grain	ZFNs	NHEJ	Shukla et al. (2009)
	LOX3	Resistance to fungal infection	Cas9 nuclease	NHEJ	Pathi et al. (2020)
	MS45	Male fertility	Cas9 nuclease	NHEJ	Svitashev et al. (2016)
	MTL (PLA1)	Haploid production	Cas9 nuclease	NHEJ	Kelliher et al. (2017)
	PAT	Herbicide resistance	ZFNs	HDR/TSI	Shukla et al. (2009), Ainley et al. (2013)

NHEJ; non-homologous end-joining, HDR; homology-directed repair, TSI; targeted sequence insertion.

In concerted efforts to improve grain yield, quality, and composition in rice, Tang et al. (2017a), Liu et al. (2019), and Yang et al. (2019) modified NATURAL RESISTANCE-ASSOCIATED MACROPHAGE PROTEIN 5 (NRAMP5) and LOW-AFFINITY CATION TRANSPORTER 1 (LCT1) using Cas9 nuclease to reduce cadmium accumulation, which causes Itai-itai disease. The resulting mutant grains accumulated less cadmium compared to the wild-type, thereby contributing to a safer food supply. Sun et al. (2017) modified STARCH BRANCHING ENZYME IIb (SBEIIb) using Cas9, which produced genome-edited plants whose grains had high resistant starch content. Likewise,