Recent Advances in the Production of Genome-Edited Rats

The rat is an important animal model for understanding gene function and developing human disease models. Knocking out a gene function in rats was difficult until recently, when a series of genome editing (GE) technologies, including zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and the type II bacterial clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated Cas9 (CRISPR/Cas9) systems were successfully applied for gene modification (as exemplified by gene-specific knockout and knock-in) in the endogenous target genes of various organisms including rats. Owing to its simple application for gene modification and its ease of use, the CRISPR/Cas9 system is now commonly used worldwide. The most important aspect of this process is the selection of the method used to deliver GE components to rat embryos. In earlier stages, the microinjection (MI) of GE components into the cytoplasm and/or nuclei of a zygote was frequently employed. However, this method is associated with the use of an expensive manipulator system, the skills required to operate it, and the egg transfer (ET) of MI-treated embryos to recipient females for further development. In vitro electroporation (EP) of zygotes is next recognized as a simple and rapid method to introduce GE components to produce GE animals. Furthermore, in vitro transduction of rat embryos with adeno-associated viruses is potentially effective for obtaining GE rats. However, these two approaches also require ET. The use of gene-engineered embryonic stem cells or spermatogonial stem cells appears to be of interest to obtain GE rats; however, the procedure itself is difficult and laborious. Genome-editing via oviductal nucleic acids delivery (GONAD) (or improved GONAD (i-GONAD)) is a novel method allowing for the in situ production of GE zygotes existing within the oviductal lumen. This can be performed by the simple intraoviductal injection of GE components and subsequent in vivo EP toward the injected oviducts and does not require ET. In this review, we describe the development of various approaches for producing GE rats together with an assessment of their technical advantages and limitations, and present new GE-related technologies and current achievements using those rats in relation to human diseases.


Introduction
Rats (Rattus norvegicus) and mice (Mus musculus) belong to the same rodent family and have been the most widely used models in biomedical research for many years. However, there are several differences between these two animals. For example, the rat is larger (roughly about eight-to ten-fold) in size than the mouse, which provides a number of practical advantages, as exemplified by easier and more rapid microsurgery, multiple and melanocortin 4 receptor (Mc4r)) in a germ-line competent manner.

Tet1 Tet2 Tet3
Li W et al. (2013) [38] MI mRNAs TALENs (indels) WI KO rats generated with a markedly attenuated response to a lipopolysaccharide challenge; can be used as a model for studying ethanol action and general inflammatory conditions including septic shock.

Il2rg
Kaneko et al. KO rats generated to know that the bone morphogenetic protein receptor 2 (Bmpr2) mutations are linked to pulmonary arterial hypertension (PAH); displaying an intense pulmonary vascular remodeling at 3 months; suggesting that endothelial-to-mesenchymal transition (EndoMT) is linked to alterations in BPMR2 signaling.

Il2rg
Kaneko and Mashimo (2015) [54] MI mRNAs TALENs (indels) Unknown KO rats generated to examine the function of cold-inducible RNA-binding protein (Cirp) in heart; CIRP modulates cardiac repolarization by negatively adjusting the expression and function of Ito channels. A DNA cassette (composed of a green fluorescent protein (GFP) reporter sequence flanked by the two pairs of lox sites) was successfully inserted in the reverse orientation into the target (leucine-rich repeat-containing G-protein coupled receptor 5 (Lgr5)) locus, which was later inverted by Cre-mediated recombination.

SD-derived SSCs
The first creation of spermatogonial stem cell (SSC)-derived GE KO rats; when recipient male rats were transplanted with engineered SSCs into seminiferous tubules (STs), and then mated with wild-type (WT) females, the resulting progeny harbored indels with~10% efficiency.

Epsti1 Erbb3
Chapman et al.  [116] In vitro EP protein/gRNA CRISPR/Cas9 (indels) SD KO rats generated to examine the role of 5-hydroxytryptamine receptor 7 (Htr7) in blood pressure regulation; neither the male nor female KO rats were different in body size, fat weights, or mass of organs (kidney, heart, and brain) important to blood pressure; can be used to investigate the importance of Htr7 in blood pressure regulation.

Rat Strain Outcome Target Gene References
KO rats generated to examine the role of solute carrier organic anion transporter family member 1B2 (Slco1b2); in those rats, the serum levels of bilirubin and bile acids, the substrates of organic anion transporting polypeptide organic anion transport polypeptide 1b2 (OATP1B2), increased; can be a disease model to study hyperbilirubinemia-related diseases.

Slco1b2
Ma et al. (2020) [139] MI, in vitro EP mRNA/sgRNAs/dsDNA donor vectors (or mRNA/sgRNAs/lssDNA) A new powerful method, called Combi-CRISPR, was developed for achieving plasmid-based KI in rat embryos using the CRISPR/Cas9 system; in other words, this method is the combination of highly efficient editing via NHEJ and the low-efficiency but precise editing via HDR.
(2021) [140] i-GONAD protein/gRNA/ssODN CRISPR/Cas9 (KI) SD Effects of 3 gRNAs (which recognize different portions of the target locus, but also overlap each other in the target locus) and of commercially available KI-enhancing drugs on the KI efficiency; KI efficiency largely depends on the type of gRNA used; none of drugs are significantly effective for KI efficiency.

Delivery Method
For the production of GE rats, the choice of the delivery method for GE components in rat zygotes is important. The methods for the production of GE rats achieved by delivering GE reagents can be largely divided into six groups: the first is MI of GE reagents (in the form of DNA, mRNA, or protein) into the cytoplasm or nucleus of a zygote ( Figure 1A); the second is in vitro EP of zygotes in the presence of GE reagents ( Figure 1B); the third is in vivo EP of zygotes after instillation of GE reagents into the oviductal lumen of a pregnant female ( Figure 1C); the fourth is in vitro viral infection of zygotes with recombinant adenoassociated virus (rAAV) ( Figure 1D); the fifth is SSC-mediated production of GE rats ( Figure 1E); the sixth is ES cell-mediated production of GE rats ( Figure 1F).
In the following sections, each of these methods is described in detail.

MI-Based Production of GE Rats
Since the first development of pronuclear MI in mice by Gordon et al. [154], numerous GM mice and other animals such as rats, pigs, cows, and nonhuman primates have been created (reviewed by Clark et al. [155]). These animals are mainly produced by the introduction of exogenous DNA called "transgene", which comprises a tissue-specific or ubiquitous promoter, intron, cDNA as a gene of interest (GOI), and polyadenine tails (poly(A) site), into a male pronucleus of a zygote. The mode of transgene integration into the host chromosomes is generally random, leading to frequent transgene silencing. In contrast with this transgenic (Tg) system, GE-based modulation of endogenous target genes does not always require the chromosomal integration of the GE components.
In 2009, ZFN-based GE in rats was first reported by Guerts et al. [28], who performed a single pronuclear injection (PI) of plasmid DNA (ranging from 0.4 to 10 ng/µL) or mRNA (ranging from 0.4 to 10 ng/µL) or intracytoplasmic injection (IC) of mRNA (from 2 to 10 ng/µL) encoding ZFNs in rat zygotes (from Dahl salt-sensitive (SS), Sprague-Dawley (SD), or Fawn-hooded hypertensive (FHH) strains) to induce indels in different target genes (coding for integrated reporter green fluorescent protein (GFP), immunoglobulin M (IgM), or Ras-related protein Rab-38 (Rab38)). The PN of mRNA, PN of plasmid DNA, or IC of mRNA resulted in production of pups with the GE efficiencies of 5-29%, 6-11%, or 6-75%, respectively. These GE efficiencies shown in this first study accelerated the researchers to generate more numbers of GM rats with these technologies. For example, rat models for X-linked severe combined immunodeficiency (X-SCID) were successfully created by PI of mRNAs encoding custom-designed ZFNs (targeted to interleukin 2 receptor gamma (Il2rg) locus) with GE efficiencies of over 20% [29].
In 2011, TALEN-based GE in rats was first reported by Tesson et al. [32], who intended to disrupt the rat IgM locus to create heritable mutations that could eliminate IgM function.
In 2011, the 2nd wave of production of ZFN-based GE was brought by Cui et al. [33], who developed a conditional KI system in rats. They used ZFN technology in embryos to introduce sequence-specific modifications (KI) through homologous recombination (HR) in SD and Long-Evans (LE) rats. Briefly, Cui et al. [33] first constructed donors with an eight-base pair (bp) Not I restriction site inserted between the ZFN-binding sites, flanked by~800 bp of immediate homology on each side. Donor plasmid DNA (1 ng/µL) and respective ZFN mRNA (2.5 ng/µL) were co-injected into the pronucleus of rat zygotes. The loci targeted were rat multidrug resistance protein 1a (Mdr1a) and pregnane × receptor (Pxr) loci. When ZFN-treated fetuses were examined, 1 of 15 (7%) and 1 of 8 (13%) were GM fetuses with KI at the Mdr1a or Pxr loci, respectively. Notably, successful germline transmission of the indel mutations that occurred in the born founder rats was also confirmed. Next, they constructed GFP donors, replacing the Not I site with a 1.5-kilobase (kb) human phosphoglycerate kinase (PGK) promoter-driven GFP cassette. GFP is in the opposite orientation of transcription for the rat Mdr1a and Pxr loci. In the case of MI with GFP cassette into the Mdr1a locus of SD rats, of the 439 embryos transferred to recipients, 83 pups were born, and 2 (2%) were pups with successful KI and 21 (25%) were those having NHEJ-based mutations, suggesting that NHEJ occurs at a higher rate than targeted integration. In the eyes of founder Mdr1a pup No. 3, PGK-GFP was expressed and visually detectable.
In contrast, Brown et al. [36] first provided a system enabling the conditional ablation of a target gene by the ZFN/Cre system through MI of ZFNs into rat zygotes. Briefly, they first generated GM rats with loxP-flanked (floxed) alleles at the glutamate ionotropic receptor N-methyl-D-aspartate (NMDA) type subunit 1 (Grin1) or corticotropin releasing hormone receptor 1 (Crhr1) locus by co-injecting mRNA encoding two pairs of ZFNs for each gene and two donors (plasmids carrying the fragment in which loxP site is surrounded by left homology arm (800 bp) and right homology arm (800 bp) in both sides) for each target into the pronucleus of zygotes. Each ZFN pair cleaved in an intron flanking the exons of interest and delivered one loxP site. In case of Grin1 targeting, among 80 live births, 6 (8%) were identified as those containing loxP in both insertion sites. Additionally, 48 animals had single loxP insertions, and another 17 had only NHEJ events. Brown et al. [36] also generated a GM rat expressing Cre under the control of an endogenous tyrosine hydroxylase (Th) promoter by inserting an internal ribosomal entry site (IRES)-Cre cassette immediately after the translational stop site of the Th open reading frame by pronuclear MI of a pair of ZFNs along with a donor plasmid. Of the 41 live-born pups, 9 (22%) were founders. Through crossing between these two lines (the floxed line and the Cre-expressing line), Cre-dependent and tissue-specific gene disruption successfully occurred in vivo in the bigenic offspring.
In 2013, the production of GE rats using the CRISPR/Cas9 system was reported [37,38]. This CRISPR/Cas9 system enables multigene KO in one shot of gene delivery to mice [156,157]. This property is especially beneficial for the purpose of creating disease model animals, as certain types of diseases are known to be caused by multigene defects. Li et al. [37] first attempted to generate KO rats with multiple mutations (at least in two different genes) using a CRISPR/Cas9 system. They injected two sgRNAs (12.5 ng/µL for each) (targeting rat melanocortin 3 receptor (Mc3r) and melanocortin 4 receptor (Mc4r)) and Cas9 mRNA (25 ng/µL) into rat zygotes. The efficiency of indel formation caused by the Cas9 nuclease was considerably different because 13 of 15 (87%) F0 pups exhibited the KO phenotype for Mc4r, while 1 of 15 (7%) exhibited the KO phenotype for Mc3r. The authors concluded that a single injection can induce disruption of at least two different genes in the rat and also suggested that targeting efficiency depends on the choice of suitable gRNA. Furthermore, Li et al. [37] first demonstrated successful germ-line transmission of the KO phenotype obtained through the CRISPR/Cas9 system, although germ-line transmission has already been known using ZFNs and TALENs by others [28,32,33]. Similar to Li et al. [37], Li et al. [38] attempted to generate multiple gene mutations in rats in a germ-line competent manner using a CRISPR/Cas9 system. They designed six sgRNAs targeting six different genomic sites encoding rat Tet methylcytosine dioxygenase 1 (Tet1) (sgTet1-1 and sgTet1-2), Tet methylcytosine dioxygenase 2 (Tet2) (sgTet2-1 and sgTet2-2), and Tet methylcytosine dioxygenase 3 (Tet3) (sgTet3-1 and sgTet3-2) to increase the likelihood of successful targeting. Co-injection of sgRNAs for the three Tet genes (sgTet1-1, sgTet2-1, and sgTet3-1) together with Cas9 mRNA into rat zygotes resulted in the generation of a total of 22 newborn pups. Of these pups, 13 (59%) contained mutations of all three Tet genes. Notably,~60% of them had a biallelic or monoallelic mutation for each targeted locus. Li et al. [38] also demonstrated that mutations in the founder rats were transmitted to the next generation.
Ménoret et al. [35] demonstrated, for the first time, that meganucleases-naturally occurring restriction enzymes derived from lower-order animals and plants-can be used as a fourth editing platform. They microinjected a plasmid encoding an engineered meganuclease for recombination, activating gene 1 (Rag1) into the pronuclei of rat zygotes. Of the microinjected zygotes, 0.6% were found to be mutated rats. However, this technique exhibits design complexity and associated costs, making it accessible to only a few laboratories.
Although MI of GE reagents into zygotes is one of the most useful tools for generating GM animals, it has advantages as well as disadvantages. The advantages of this technology include the delivery of known quantities of nucleic acids into a zygote irrespective of the type of zygote and the introduction of a large-size cargo (carrying a GOI), which is a significant limiting factor when using viral vectors for gene delivery, as suggested by Sato et al. [158]. The disadvantages of this technology include the requirement of a micromanipulator system and egg transfer (ET) to pseudo-pregnant females for allowing the further development of MI-treated embryos.

In Vitro EP-Based Production of GE Rats
EP is known to be a useful and powerful gene delivery tool which can enable the transfer of exogenous substances (i.e., DNA, mRNA, and protein) into a cell by forming transient pores into the cell membranes under electrical stimulation in vitro and in vivo (reviewed by Young and Dean [162]).   In 2014, Kaneko et al. [43] first demonstrated that rat zygotes can be genome-edited by in vitro EP, which is shown in Figure 1B, schematically. Since then, many researchers have successfully induced gene edits using this technology in mice [163][164][165][166][167][168], rats [95,111], bovines [169], and pigs [170][171][172]. The merit of this technology is that it is simple, rapid, and convenient for GE in zygotes, compared to the previous MI-based technique. Notably, approximately 30 to 50 zygotes can be edited with one shot of in vitro EP. Furthermore, EP only requires a square pulse generator, called an electroporator, and not a more expensive micromanipulator system. According to Kaneko et al. [43], intact F344/Stm rat zygotes were subjected to in vitro EP using an NEPA21 electroporator (NEPA GENE Co. Ltd. Chiba, Japan) in the presence of a phosphate-buffered saline (PBS)-based solution containing ZFN (40 µg/mL) mRNA (targeted to the rat Il2rg gene) under the EP condition of a poring pulse (Pp) (voltage: 225 V; pulse interval: 50 ms; pulse width: 1.5 and 2.5 ms; number of pulses: 4). The NEPA21 electroporator employs a 3-step pulse system. In the first step, the Pp makes micro-holes in the ZP and oolemma. In the second step, several first transfer pulses (Tp) transfer mRNA into the cytoplasm. In the third step, the polaritychanged second Tp increase the chance of transferring mRNA into embryos. When rat embryos were electroporated with a pulse width adjusted to 1.5 or 2.5 ms, 73 and 75% of the resulting offspring, respectively, had an edited Il2rg locus. Notably, they reported no appreciable reduction in the developmental ability of the EP-treated embryos. Furthermore, the germ-line transmission of the ZFN EP-induced editing was also confirmed in the next generation. They named this technology "Technique for Animal Knockout system by Electroporation (TAKE)". Since then, many GE rats have been produced using EPbased GE technology [54,90,91,95,111,117,121,125,131,140,141,144]. Notably, in vitro EP can also be applied to frozen mouse and rat zygotes for the efficient introduction of CRISPR components, without affecting the embryo viability or development [173,174]. This will facilitate GE rat production in a more convenient manner.
At the initial stages of GE rat production, mRNAs encoding ZFNs, TALENs, or Cas9 have been frequently used [28][29][30][31][32][33][34][36][37][38][39][40][41][42]. However, mRNA transfection still relies on the translation of the mRNA into protein, resulting in variability of GE efficiency with timing. In the case of CRISPR/Cas9-based GE, instead of using Cas9 mRNA, the use of the Cas9 protein allows transient and faster editing [175]. For example, the recombinant Cas9 protein and the chemically synthesized guide RNA (gRNA) can be combined on the bench before being delivered into embryos; this Cas9-gRNA complex, which is called a ribonucleoprotein (RNP), is active immediately. If the RNP enters the nuclei of cells, GE should commence quickly, which will, in turn, lead to increased GE efficiency. Using this RNP-based GE system, a number of GE rats have been produced to date [57,58,68,90,91,95,111,113,115,117,121,126,131,142,150].
Although in vitro EP enables the effective delivery of a short (<1 kb) single-stranded DNA (ssDNA) or synthetic single-stranded oligodeoxynucleotide (ssODN) into mouse zygotes, it has been difficult to introduce longer (>1 kb) double-stranded DNA (dsDNA) such as plasmid DNA. Only a few laboratories have successfully achieved this, although extensive exploration of optimal EP conditions, which is very laborious and time-consuming, is strictly required [176]. Notably, Bagheri et al. [177] overcame this limitation by first injecting all GE components, including long plasmid-sized DNA templates, into the subzona pellucida (ZP) space, called the "peri-vitelline space". Therefore, they were retained, supporting subsequent in vitro EP. Bagheri et al. [177] suggest that this simple and welltolerated method achieves intracellular reagent concentrations sufficient to provide precise gene edits.
As for the HDR-based KI by in vitro EP, Miyasaka et al. [95] first designed long singlestranded DNA (lssDNA) (constructed using a simple method including custom-made plasmids using nicking endonucleases as previously reported by Yoshimi et al. [64]) as the HDR template. The lssDNA comprises a targeted exon (exon 2 of the rat vesicle-associated membrane protein-associated protein B/C (Vapb) gene with a P56S mutation, which is associated with amyotrophic lateral sclerosis (ALS) in humans) flanked by two loxP sites.
In vitro EP of rat zygotes was performed in the presence of Cas9 mRNA (400 ng/µL), lssDNA (40 ng/µL), and two gRNAs (200 ng/µL) in a chamber with 40 µL of serum-free media (Opti-MEM) with a 5-mm gap electrode in a NEPA21 electroporator under the following conditions: Pp of voltage 225 V, pulse width-2.0 ms, pulse interval-50 ms, and number of pulses-4. The first and second Tp were of voltage 20 V, pulse width-50 ms, pulse interval 50-ms, and number of pulses-5. Consequently, of six pups born, three showed floxed alleles with the P56S mutation. They called this technology "CRISPR with lssDNA inducing conditional knockout alleles (CLICK)". The CLICK technology will be useful as a tool for producing GM zygotes carrying transgenes driving tissue-specific Cre recombinase, which, in turn, will be applicable to the one-step generation of conditional KO animals.
When compared to the MI-based technique, in vitro EP allows the simultaneous processing of many zygotes in a short time (e.g., a batch of 30 to 50 zygotes in few seconds) without requiring expensive equipment and operators with extensive training and expertise. The disadvantages of this technology include the requirement of ET of the EP-treated embryos to pseudo-pregnant females to allow further development of the treated embryos.
Notably, the detailed protocols for in vitro EP-based GE in rats have been reported by Kaneko [178] and Remy et al. [179].

GONAD-Based Production of GE Rats
In 2015, a novel method (termed "GONAD") enabling in situ CRISPR/Cas9-based GE toward early embryos was first reported by Takahashi et al. [180] using mice. In this system, both Cas9 mRNA (up to 1.1 µg/µL) and sgRNA (0.6 µg/µL) were co-injected into the oviductal lumen of pregnant females at Day 1.4 of pregnancy (corresponding to the 2cell stage). Technically, this was simply achieved by inserting a glass micropipette through the oviductal wall under observation using a dissecting microscope, as shown in Figure 1C schematically. Next, a small amount of a solution (1-1.5 µL) containing GE components was injected into the oviductal lumen with the aid of a mouthpiece-controlled micropipette. Subsequently, in vivo EP was performed using tweezer-type electrodes and a square-wave pulse generator (T820; BTX Genetronics Inc. San Diego, CA, USA) under the EP condition of eight square-wave pulses with a pulse duration of 5 ms and an electric field intensity of 50 V. GE occurrence in the 2-cell embryos can be later checked by molecular analysis of the mid-gestational fetuses dissected after the GONAD. The KO efficiency of these fetuses was approximately 29%. Later, this technology was elaborated by the same group, who injected a solution containing RNP (comprising Cas9 protein (1 µg/µL Integrated DNA Technologies, Inc. Coralville, IA, USA) and sgRNA (30 µM)) (for KO experiment), RNP + ssODN (1-2 µg/µL) (for KI experiment), or RNP + ssDNA (0.85-1.4 µg/µL) (for KI experiment) at Day 0.7 of pregnancy (corresponding to the late 1-cell stage) [181]. In vivo EP using the NEPA21 electroporator was performed under the following conditions: Pp: 50 V, 5-ms pulse, 50-ms pulse interval, three pulses, 10% decay (±pulse orientation); Tp: 10 V, 50 ms pulse, 50 ms pulse interval, three pulses, and 40% decay (±pulse orientation). This modification resulted in 97% of the embryos exhibiting indels in the target locus and approximately 50% of embryos containing KI alleles. Thus, this improved technology was re-named the "improved GONAD (i-GONAD)". Since the successful development of GONAD/i-GONAD in mice, the technology has proven useful for other species, such as rats [90,91,121] and hamsters [182].
Successful i-GONAD in rats was first reported in 2018 [90,91]. For examples, Takabayashi et al. [90] demonstrated that the appearance of non-pigmented eyes in fetuses (showing the KO phenotype due to indels at the tyrosinase gene (Tyr), a gene coding for protein essential for eye pigmentation) was induced with an efficiency of 56%, when F1 embryos from pigmented Brown Norway (BN) × albino SD rat crosses were used for i-GONAD targeting of Tyr. In this case, the in vivo EP condition used was almost the same as that used for i-GONAD in mice. They also tested the possibility of KI (targeted Tyr locus) using albino Lewis (LEW) rats. The i-GONAD-mediated KI efficiency, which was evaluated by the presence of fetuses with pigmented eyes, was as low as~5%. Kobayashi et al. [91] provided data similar to that of Takabayashi et al. [90] using the NEPA21 electroporator under the following EP conditions: Pp: 30, 40, or 50 V, 5-ms pulse, 50-ms pulse interval, number of pulses 3, and 10% decay (±pulse orientation); Tp: 10 V, 50-ms pulse, 50-ms pulse, number of pulses 6, and 40% decay (±pulse orientation). According to Kobayashi et al. [91], intraoviductal injection of RNP (1 µg/µL of Cas9 protein +30 µM of sgRNA) targeted to Tyr into oviducts of pregnant albino WKY females (which had been successfully mated with dark agouti (DA) males) or pigmented DA females (which had been successfully mated to Wistar-Kyoto (WKY) males) resulted in the production of GE pups with efficiencies of 59% or 42%, respectively, when 50 V of the voltage (Pp) was employed. These results suggest no significant difference in the gene-editing efficiency between these two strains. They also attempted to recover the coat-color mutation in WKY females using an ssODN-based KI approach. Consequently, the KI efficiency was 27% in the pups born. They named this rat-based i-GONAD as "rGONAD".
For performing i-GONAD (or rGONAD) in rats, the exploration of optimal EP conditions is strictly required. For example, when a current of >500 mA was employed, albino SD and albino LEW rats were successfully genome-edited; however, the acquisition of offspring derived from pigmented BN rats (fetuses/newborns) failed. In contrast, i-GONAD under a current of 100-300 mA using the NEPA21 electroporator led to the production of GE BN rats with efficiencies of 75% to 100% [90]. Similarly, i-GONAD under a current of 150-200 mA using another electroporator CUY21EDIT II (BEX Co., Tokyo, Japan) led to the production of genome-edited BN rats with efficiencies of 24% to 55% [121].
The advantages of this technology are that GONAD/i-GONAD requires an electroporator but no other special equipment. Furthermore, no recipients are required unlike the case of MI-or in vitro EP-based GE rat production. Disadvantages of this technology are that technical skill may be required for intraoviductal injection of a solution into the lumen of oviducts under a dissecting microscope using a mouthpiece-controlled glass micropipette.
The detailed protocols for GONAD/i-GONAD have been reported by Gurumurthy et al. [183] and Ohtsuka and Sato [184] in mice. The GONAD/i-GONAD-based production of GE rats is also possible using the same approach shown in mice. Notably, Namba et al. [185] demonstrated the protocols for GONAD/i-GONAD in rats.

rAAV-Mediated Production of GE Rats
rAAVs are non-pathogenic and non-enveloped ssDNA viruses enabling highly efficient transduction of both dividing and non-dividing cells (reviewed by Daya and Berns [186]). Unlike other viral vectors such as retroviruses and adenoviruses, rAAVs have been used for manipulating the target genome sequences through simple co-incubation with ZPenclosed preimplantation mammalian embryos for the production of GM animals such as mice [168,[187][188][189][190][191]. For example, Mizuno et al. [187] observed that, among the rAAVs tested, the rAAV with serotype 6 (which is hereinafter referred to as rAAV-6) showed the highest transduction efficiency when ZP-intact murine embryos were co-cultured with rAAVs. More importantly, Mizuno et al. [187] demonstrated successful CRISPR-based KI in mice. As shown in Figure 1D, zygotes were first subjected to in vitro EP in the presence of RNP and then infected with rAAV-6 carrying a 1.8-kb GFP expression cassette flanked by two 100-bp Rosa26 homology arms. The KI efficiencies in the Rosa26 locus were 16% and 6%, when assessed at the blastocyst and newborn stages, respectively. Notably, Mizuno et al. [187] demonstrated that both ZP-intact rat and bovine embryos were effectively infected by rAAV-6, suggesting that rat genome can be effectively edited through the simple incubation of ZP-intact rat zygotes with a solution containing rAAVs.
This rAAV-based gene delivery system to early embryos was also observed in vivo. Yoon et al. [188] injected a solution containing two vectors, rAAV6-Cas9 (carrying spCas9 gene derived from Streptococcus pyogenes) and rAAV6-gTyr (carrying gRNA expression unit targeted Tyr), into the oviducts of pregnant female mice at Day 0.5 of pregnancy, similar to GONAD/i-GONAD. Out of the 29 pups that were obtained, three (10%) were found to have indels. All mutated founder mice generated albino offspring, indicating germ-line transmission. These results suggest that AAV is a powerful tool for inducing GE in the ZP-enclosed early embryos in vivo. According to Sato et al. [158], this in vivo approach is referred to as "AAV-based GONAD".
The advantages of this technology are that no special equipment is required in the case of using the AAV-based GONAD. Notably, the use of highly concentrated rAAV would be one of the key factors for achieving a high rate of transduction of preimplantation embryos. According to Yoon et al. [188], zygotes treated with 6 × 10 8 genome copies (GCs) for one day in vitro resulted in live births with 100% indel frequency after ET. However, the editing frequency dropped to 25% for fetuses and 20% for newborns at 6 × 10 7 GCs. No edited animals were detected from the 6 × 10 6 GC treatment group. These results clearly suggest that gene delivery and GE efficiency are rAAV dose-dependent.

SSC-Mediated Production of GE Rats
Cultured SSCs can display long-term sperm-forming potential when they are transplanted into the seminiferous tubules (STs) of sterile rat testes (shown in Figure 1E). It is now possible to culture SSCs for a long time and perform gene modification toward SSCs derived from mice and rats (reviewed by Sato et al. [158]).
Chapman et al. [60] first demonstrated that CRISPR/Cas9-mediated GE can be successfully applied in SSCs to create KO rats. They used SSCs isolated from tgGCS-EGFP rats (with the genetic background of SD) showing germ cell-specific expression of enhanced green fluorescent protein (EGFP) [192]. The SSCs were then transfected with the pX330 vector (conferring expression of both Cas9 and gRNA targeted epithelial stromal interaction 1 (Epsti1) gene) together with a neomycin resistance gene expression plasmid. The surviving clones after G418 selection were then transplanted into the STs of recipient sterile rats to promote spermatogenesis. When the recipient rats transplanted with G418-selected cells were mated with WT females at~65 days post-transplantation, a total of 87 pups were obtained. Of these,~10% (9) were found to harbor Epsti1 mutants. A similar attempt was also made by Noto et al. [102], who successfully obtained immunodeficient rats by grafting recombination activating gene 2 (Rag2) KO SSCs into rat testes.

ES Cell-Mediated Production of GE Rats
As mentioned previously, acquisition of a rat ES cell line is now possible. Yamamoto et al. [14] first succeeded in producing KI rats using CRISPR/Cas9-engineered ES cells. Initially, they obtained GM ES cell clones using a conventional gene-targeting technology but failed because of no recombinant ES cells were obtained. Next, in vitro EP of a Cas9expressing plasmid, gRNA-expressing plasmid, and KI vector (carrying two homology arms) in rat ES cells was performed. Consequently, the number of drug-resistant colonies increased and the HR efficiency was considerably enhanced from 0 to 36%. These recombinant clones resulted in the successful production of chimeras, which were later proven to have the ability to transmit the KI allele to their offspring after mating with WT rats (as shown in Figure 1F). According to Yamamoto et al. [14], the methodology presented can overcome the problem encountered in conventional gene targeting that prevented the production of humanized rats. Later, successful GM rat production through ES cells carrying CRISPR/Cas9-based KI of the Discosoma coral red fluorescent protein (dsRed) gene at the Sry-box transcription factor 10 (Sox10) locus was reported by Chen et al. [15].

Other Techniques and Factors to Modify the Rat Genome
As shown above, GE tools such as ZFNs, TALENs, and CRISPR/Cas9 are considered useful for the production of GE rats. However, several techniques and factors still influence their performance, which must be addressed. These include the allele-specific GE using ssODNs, large genomic fragment deletion, ssODN-mediated KI without the need to construct homology arms, the use of inhibitors for efficient KI, the importance of choosing the appropriate gRNA for efficient KI, and base editor systems enabling precise genome modifications.
In this section, these techniques or factors are described in detail.

Allele-Specific GE for the Correction of Mutated Phenotypes
Numerous reports have demonstrated that CRISPR/Cas9 editors combined with 100 bp ssODNs, instead of using a relatively long template (~1 kb), allow for precise HDR-mediated GE in mice [156,157,193]. These results suggest that ssODNs co-injected as donor templates preferentially support the activation of HDR relative to the NHEJ pathway.
To explore the applications of ssODN-mediated GE in rats, Yoshimi et al. [46] successfully used ssODNs to manipulate the coat-color genetics of F344 rats. They used a WT ssODN to exchange a single base pair in Tyr, thus reverting the albino allele back to the WT sequence and restoring a non-albino pigmentation pattern. A similar strategy using ssODN for the correction of mutated phenotypes in rats was also employed by others [64,90,91,106,114,[125][126][127]129,141,142].

Large Genomic Fragment Deletion
The GE system has enabled the creation of targeted deletion spanning the large genomic region over 10 kb in size via the direct MI of two sgRNAs (~10-30 ng/µL) and TALEN mRNA (45 ng/µL) or Cas9 mRNA (~67-100 ng/µL) into the pronuclei or cytoplasm of mouse zygotes [194,195]. Wang et al. [58] achieved 53-kb deletion of a fragment spanning long non-coding RNA (lncRNA) GM14005 genomic DNA by injecting the Cas9 protein (30 nM) into the cytoplasm of mouse zygotes together with two sgRNAs (12.5 ng/µL). Among five F0 pups, one founder was confirmed as a mouse with a large deletion. Wang et al. [58] also showed that the 53-kb deletion was efficiently transmitted to the F1 generation.
Yoshimi et al. [46] succeeded in the deletion of a 7-kb fragment from the rat genome. They used two flanking gRNAs to delete a 7098-bp endogenous retroviral element (ERV) within the first intron of the Kit gene, using ssODN containing homology arms on either side of the junction to bridge across the deletion (Figure 2). They injected 100 ng/µL Cas9 mRNA, 50 ng/µL gRNAs (targeted Kit), and 120 bp ssODNs (50 ng/µL) into the pronuclei of Long-Evans hooded (LEH) rat zygotes. Thus, GE rats exhibited restoration of the function of the Kit gene with a non-hooded pigmentation pattern. The frequency of the appropriately edited pups was 4%, showing the feasibility of performing template-directed correction of mutated portions using ssODNs. Yoshimi et al. [46] succeeded in the deletion of a 7-kb fragment from the rat genome. They used two flanking gRNAs to delete a 7098-bp endogenous retroviral element (ERV) within the first intron of the Kit gene, using ssODN containing homology arms on either side of the junction to bridge across the deletion (Figure 2). They injected 100 ng/μL Cas9 mRNA, 50 ng/μL gRNAs (targeted Kit), and 120 bp ssODNs (50 ng/μL) into the pronuclei of Long-Evans hooded (LEH) rat zygotes. Thus, GE rats exhibited restoration of the function of the Kit gene with a non-hooded pigmentation pattern. The frequency of the appropriately edited pups was 4%, showing the feasibility of performing template-directed correction of mutated portions using ssODNs. If both DSBs occur at the same time, the result will be a large deletion of the region of interest. To the best of our knowledge, the biggest deletion achieved to date in rats is 24.4 mega base pairs (Mb) [196].

ssODN-Mediated KI with CRISPR/Cas9 for Large Genomic Regions in Rat Zygotes
Although the CRISPR/Cas9 system is recognized as a powerful tool for generating GM animals, the creation of targeted KI animals via HR is still challenging. Yoshimi et al. [64] developed a novel KI method without the need to construct homology arms, thus simplifying the process of genome engineering in living organisms. In other words, they If both DSBs occur at the same time, the result will be a large deletion of the region of interest. To the best of our knowledge, the biggest deletion achieved to date in rats is 24.4 mega base pairs (Mb) [196].

ssODN-Mediated KI with CRISPR/Cas9 for Large Genomic Regions in Rat Zygotes
Although the CRISPR/Cas9 system is recognized as a powerful tool for generating GM animals, the creation of targeted KI animals via HR is still challenging. Yoshimi et al. [64] developed a novel KI method without the need to construct homology arms, thus simplifying the process of genome engineering in living organisms. In other words, they tried to knock-in longer plasmids (including the 1.2-kb chicken β-actin-based promoter (called "CAG") and a 0.8-kb GFP cassette sequence) at the targeted site (rat Rosa26 locus) (Figure 3). In more detail, they constructed two gRNAs: one targeting the rat Rosa26 locus to cleave genomic DNA, and the other targeting the 5 end of the CAG promoter sequence for concurrent cleavage of the plasmid DNA (Figure 3). Two 80-bp ssODNs (ssODN-1 and -2) were designed to ligate the two cut ends, as shown in (Figure 3). A mixture of 100 ng/µL of the Cas9-poly(A) mRNA (which appended an 81-bp poly(A) to the transcription terminator to increase the efficacy of GE with the CRISPR/Cas9 system), 50 ng/µL of each of the two gRNAs (gRNA:rRosa26 and gRNA:CAGGS), 50 ng/µL of each of the two ssODNs (ssODN-1 and -2), and 5 ng/µL of the CAG-GFP plasmid was co-microinjected into the pronuclei of rat zygotes. Of the 17 pups obtained, four (#6, #7, #8, and #11) were selected as candidates with successful KI. Sequence analysis revealed that #11 had accurate conjunction at both of the two cut ends, #6 had a 6-bp deletion on one side, and #7 had several base-pair deletions and insertions on both sides. They confirmed germ-line transmission of the GFP allele. No off-target effect was observed at any candidate sites in these rats. Furthermore, uniform expression of the GFP gene in most cells and tissues examined was confirmed. Based on these observations, this strategy was called "two-hit by gRNA and two-oligo with plasmid (2H2OP)".  . Schematic representation of the two-hit by gRNA and two-oligo with plasmid (2H2OP) method for the production of CAG-GFP knock-in (KI) rats generated using the CRISPR/Cas9 system. In the first step, Cas9, together with two gRNAs targeting the rat Rosa26 locus and the CAG promoter in the GFP plasmids, cuts the target sites. In the second step, two ssODNs ligate each cut end to join the genomic DNA and the plasmid DNA via HDR. This figure was drawn in-house, based on the data shown in the paper of Yoshimi et al. [64]. Abbreviations: CAG, chicken β-actinbased promoter; GFP, green fluorescent protein; KI, knock-in; ssODNs, single-stranded oligodeoxynucleotides; pA, poly(A) sites.

Use of Inhibitor for Efficient KI
KI is an event of precise modifications at a target locus and is often mediated by HR in the presence of donor DNA, as exemplified by ssODN. Generally, HDR-mediated KI is more difficult than NHEJ-based indels. For example, in proliferating human cells, NHEJ has been reported to repair 75% DSBs, whereas HDR repaired the remaining 25% [197]. To enhance the HDR efficiency, several approaches are now being attempted. In the case of a CRISPR/Cas9-mediated KI event, Scr7, which was first identified as an anti-cancer compound and was considered a potential NHEJ inhibitor, was used to increase the HR- Figure 3. Schematic representation of the two-hit by gRNA and two-oligo with plasmid (2H2OP) method for the production of CAG-GFP knock-in (KI) rats generated using the CRISPR/Cas9 system. In the first step, Cas9, together with two gRNAs targeting the rat Rosa26 locus and the CAG promoter in the GFP plasmids, cuts the target sites. In the second step, two ssODNs ligate each cut end to join the genomic DNA and the plasmid DNA via HDR. This figure was drawn in-house, based on the data shown in the paper of Yoshimi et al. [64]. Abbreviations: CAG, chicken β-actin-based promoter; GFP, green fluorescent protein; KI, knock-in; ssODNs, single-stranded oligodeoxynucleotides; pA, poly(A) sites.

Use of Inhibitor for Efficient KI
KI is an event of precise modifications at a target locus and is often mediated by HR in the presence of donor DNA, as exemplified by ssODN. Generally, HDR-mediated KI is more difficult than NHEJ-based indels. For example, in proliferating human cells, NHEJ has been reported to repair 75% DSBs, whereas HDR repaired the remaining 25% [197]. To enhance the HDR efficiency, several approaches are now being attempted. In the case of a CRISPR/Cas9-mediated KI event, Scr7, which was first identified as an anticancer compound and was considered a potential NHEJ inhibitor, was used to increase the HR-mediated precise genetic modification [198,199]. For example, co-injection of murine zygotes with a mixture containing Cas9 mRNA, sgRNA, template ssODNs, and Scr7 significantly improved the efficiency of HDR-mediated insertional mutagenesis [198]. Chu et al. [200] also demonstrated the usefulness of Scr7 for abolishing NHEJ activity and increasing HDR in both human and mouse cell lines. However, the function of Scr7 in promoting HDR remains controversial. Some researchers have demonstrated that Scr7 failed to increase HDR rates in rabbit embryos [201] and porcine fetal fibroblasts [202].
However, the opposite results were recently achieved by Takabayashi et al. [90] and Aoshima et al. [141]. Takabayashi et al. [90] performed i-GONAD using a solution containing 1 µM Scr7, RNP (1 µg/µL of Cas9 protein +30 µM of sgRNA), and 2 µg/µL of ssODN resulted in the production of 27 fetuses, of which only two (7%) had pigmented eyes. Notably, in the absence of Scr7, i-GONAD-mediated KI efficiency was as low as~5%, suggesting that addition of Scr7 appeared not to affect i-GONAD-mediated KI efficiency in rats. Aoshima et al. [141] examined whether commercially available KI-enhancing drugs (including Scr7, L755,507, RS-1, and HDR enhancer), all of which are known to enhance KI efficiency, can increase the KI efficiency using i-GONAD in rats. Aoshima et al. [141] demonstrated that some drugs (e.g., Scr7, L755,507, and HDR enhancer) were found to be slightly effective in vivo, but their effects were not statistically significant when compared to those in the control group (without any KI-enhancing drugs).

Choice of gRNA Is Very Important for Achieving High KI Efficiency in Rats
According to Raveux et al. [203], ssODN-mediated KI efficiency varied from 0% to 40%, depending on the type of Cas9, the injection site, the length of the homology arm, and the gRNA-binding site. The nucleotide composition of a target sequence is one of the most important factors determining KI and KO efficiency. For example, Graf et al. [204] screened the published gRNA activity datasets and demonstrated that TT-and GCC-motifs located in the four PAM-proximal bases of the targeting sequence (the efficiency-modulating sequence) were sufficient to block CRISPR/Cas9-mediated KI. Moreover, the presence of these motifs in gRNAs resulted in a 10-fold reduction in gene KO frequencies.
Aoshima et al. [141] constructed three gRNAs showing high efficiencies, ranging from 51% to 57%, based on the CHOPCHOP predictions [205]. These three gRNAs (termed crRNA1, crRNA2, and crRNA3) used here recognize the Tyr sequence containing a point mutation (G to A) and overlap with each other (Figure 4). When KI-based i-GONAD was performed using ssODN (2 µg/µL), Cas9 protein (1 µg/µL), and gRNA (crRNA1, 2 or 3) (30 µM), the KI efficiency in fetuses generated after i-GONAD with crRNA2 was significantly higher (24%) than that obtained after i-GONAD with crRNA1 (5%) or crRNA3 (0%). The discrepancy between predicted and actual efficiencies was unexpected. Notably, no inhibitory sequences exist in the gRNAs used, unlike the suggestion of Graf et al. [204]. Furthermore, in the case of in vitro GE, the GE efficiency is known to vary, with predicted efficiencies typically higher than observed levels [156,206,207]. The data of Aoshima et al. [141] suggest the importance of selecting the appropriate gRNA for effective KI in rats.
GONAD was performed using ssODN (2 μg/μL), Cas9 protein (1 μg/μL), and gRNA (crRNA1, 2 or 3) (30 μM), the KI efficiency in fetuses generated after i-GONAD with crRNA2 was significantly higher (24%) than that obtained after i-GONAD with crRNA1 (5%) or crRNA3 (0%). The discrepancy between predicted and actual efficiencies was unexpected. Notably, no inhibitory sequences exist in the gRNAs used, unlike the suggestion of Graf et al. [204]. Furthermore, in the case of in vitro GE, the GE efficiency is known to vary, with predicted efficiencies typically higher than observed levels [156,206,207]. The data of Aoshima et al. [141] suggest the importance of selecting the appropriate gRNA for effective KI in rats.  [141]. The target sequence (exon 2 of Tyr) recognized by crRNA1, 2, and 3 is shown in green. The PAM sequences are underlined. Single-stranded oligodeoxynucleotide (ssODN) (containing wild-type nucleotide "G" that corresponds to mutated nucleotide "A") is shown in orange below the target sequence. The nucleotide "A/T" marked in red is the mutation causative of the albino phenotype. This figure was drawn in-house, based on the data shown in the paper of Aoshima et al. [141].

Base Editor Systems as An Efficient Tool Enabling Precise Genome Modifications
Single-base editing systems without the induction of DSB or template DNA, mediated by cytidine deaminase combined with Cas9 nickase (nCas9) or dead Cas9 (dCas9),  [141]. The target sequence (exon 2 of Tyr) recognized by crRNA1, 2, and 3 is shown in green. The PAM sequences are underlined. Single-stranded oligodeoxynucleotide (ssODN) (containing wild-type nucleotide "G" that corresponds to mutated nucleotide "A") is shown in orange below the target sequence. The nucleotide "A/T" marked in red is the mutation causative of the albino phenotype. This figure was drawn in-house, based on the data shown in the paper of Aoshima et al. [141].

Base Editor Systems as an Efficient Tool Enabling Precise Genome Modifications
Single-base editing systems without the induction of DSB or template DNA, mediated by cytidine deaminase combined with Cas9 nickase (nCas9) or dead Cas9 (dCas9), were recently developed to modify mammalian genomic DNA. This system is called the "cytosine base editor (CBE)" which achieves C·G to T·A conversion with a window of approximately five nucleotides [208,209]. Furthermore, Gaudelli et al. [210] developed a new base-editing (BE)-based system, called the "adenine base editor (ABE)", which efficiently enables the induction of A·T to G·C conversion through fusion of an engineered tRNA adenosine deaminase (ecTadA) with nCas9. Thus, together with CBE, ABE enables the introduction of all four nucleotide transitions (C to T, A to G, T to C, and G to A) in the target genomic sequence [210]. Indeed, using these systems, the correction of genetic disease has been successfully performed in mice [211,212].
With respect to the production of GM rats using the BE systems, Yang et al. [97] reported the successful production of KO rats through MI of ABE mRNA together with sgRNA (targeted glycogen storage disease type 2 (Gsd2) gene) to generate an animal model for glycogen storage disease type II. Consequently, 85% (28/33) of rats had single or multiple A > G substitutions between position A3 and A7 in the target protein (GSD2), leading to I646V or D645G mutations. Similarly, Ma et al. [104] injected ABE mRNA together with the sgRNA-targeted hemogen (Hemgn), N-deacetylase (Ndst1), or N-sulfotransferase 4 (Ndst4) gene into rat zygotes. For Hemgn targeting, the zygotes were microinjected with ABE mRNA and sgRNA (targeted Hemgn); a total of 99 injected zygotes were transferred to three pseudo-pregnant female rats, and 15 pups were born. Consequently, fourteen rats (14/15) contained an A to G conversion at the 14th base distal from the PAM, indicating the high base-editing efficiency. According to Ma et al. [104], these BE tools can provide a much safer approach compared with the WT CRISPR/Cas9 system for the gene correction of human disease.
Recently, Qi et al. [153] demonstrated that a bacterial toxin deaminase (DddA) from Burkholderia cenocepacia, which can convert cytosine to uracil specifically within dsDNA, can be used for mediating mitochondrial DNA (mtDNA) editing in rats. It has been demonstrated that the toxin domain of DddA (DddAtox, 1264-1427 amino acids) is engineered and incorporated with the mitoTALE system to efficiently achieve C · G-to-T · A conversion in mtDNA of human cell lines [213]. This system is therefore termed "DddA-derived cytosine base editor (DdCBE)" [213]. Qi et al. [153] injected mRNAs of a DdCBE pair with the best performance in rat zygotes, but the DdCBE-mediated BE failed. They reasoned that this failure was due to the inactive mtDNA replication in zygotes because this system works during the period when mtDNA replication is active. To overcome this problem, the DdCBE pair was cloned into the piggyBack (PB) transposon vector, and then, the resulting PB vector was co-injected into rat zygotes with PB transposase mRNA. Thus, one out of four pups obtained was successfully edited with 36% efficiency for one target locus. According to Qi et al. [153], almost all DdCBE pairs failed to integrate into the rat genome by the transposon system; however, the expression of the DdCBE pair from the PB vector might have lasted longer than that from mRNA injection, facilitating C · G-to-T · A conversion in mtDNA. Using the DdCBE system, animals with precise mtDNA mutations can be produced on demand.

Disease Models in Rats
For a decade, the advent of genetic engineering tools such as ZFNs, TALENs, and CRISPR/Cas9 has led to a revolution in obtaining specific and targeted genetic mutations in rats for the study of human genetic diseases. Several important genetic diseases have been modeled in rats. A brief description of the most useful models is provided below.

Models for Cardiovascular Diseases
Monoallelic mutations in the gene encoding bone morphogenetic protein receptor 2 (Bmpr2) are the main genetic risk factors for heritable pulmonary arterial hypertension (PAH) with incomplete penetrance. Several Bmpr2 Tg mice have been reported to develop mild spontaneous PAH. Ranchoux et al. [53] generated Bmpr2 KO rats using the ZFN technology. The resulting KO rats with a heterozygous 140 bp deletion in the first exon of Bmpr2 displayed intense pulmonary vascular remodeling. The same group [116] also generated rat lines with mutations (deletion of 71 bp in exon 1 of Bmpr2) and showed that the heterozygous rats developed age-dependent spontaneous PAH with a low penetrance (16-27%), similar to that in humans. They concluded that this new genetic rat model represents a promising tool to study PAH pathogenesis. Concerning the pathogenesis of PAH, mutations in the potassium channel subfamily K member 3 (KCNK3) gene, which encodes an outward rectifier K+ channel, have been identified in PAH patients. Lambert et al. [109] generated Kcnk3 KO rats through the CRISPR/Cas9 technology. The resulting rats developed age-dependent PAH associated with low serum-albumin concentrations. Lambert et al. [109] concluded that KCNK3 loss of function is a key event in PAH pathogenesis.

Models for Neurological Diseases
Animal models are required to understand the pathogenesis of autism spectrum disorder (ASD). Despite the apparent advantages of mice for neural studies, rats have not been widely used for ASD studies, probably owing to the lack of convenient genome manipulation tools. Hamilton et al. [42] generated two rat models for ASD, one syndromic and one non-syndromic model, through the ZFN system, by destroying a gene (Fmr1) coding the Fragile X mental retardation protein (FMRP), the protein responsible for the pathogenesis of Fragile X syndrome (FXS), or a gene (Ngln3) coding for Neuroligin3 (NLGN3), a member of the neuroligin synaptic cell-adhesion protein family, respectively. Both FMRP and NLGN3 have been implicated in the pathogenesis of human ASD. Both KO rat lines exhibited abnormalities in ASD-relevant phenotypes including juvenile play, perseverative behaviors, and sensorimotor gating, suggesting the utility of these rats as genetic models for investigating ASD-relevant genes. Later, Tian et al. [77] produced Fmr1 KO rats through the CRISPR/Cas9 technology targeted to exon 4 of Fmr1. Consistent with the previous reports, deletion of the Fmr1 gene in rats specifically impairs long-term synaptic plasticity and hippocampus-dependent learning in a manner resembling the key symptoms of FXS.
DEP-domain containing 5 gene (Depdc5), encoding a repressor of the mechanistic target of rapamycin complex 1 (mTORC1) signaling pathway, has recently emerged as a major gene mutated in familial focal epilepsies and focal cortical dysplasia. Marsan et al. [67] produced a Depdc5 KO rat using the TALEN technology. Homozygous KO embryos died from Day 14.5 of pregnancy. Heterozygous KO rats developed normally and exhibited no spontaneous electroclinical seizures; however, they showed altered cortical neuron excitability and firing patterns. This rat model is considered a relevant model to study pathogenic mechanisms underlying those disorders.
One subtype of ASD is associated with mutations in the methyl-CpG-binding protein 2 (Mecp2) gene, causing an X-linked neurodevelopmental disorder called Rett syndrome (RS). Patients with RS have cognitive defects and circadian clock dysfunction, as exemplified by abnormal sleep patterns. According to Zhai et al. [61], there is an urgent need for new animal models for RS because the existing Mecp2 KO mouse models fail to fully mimic the pathogenesis and symptoms of patients with RS. Mecp2 KO rats were successfully produced by the two groups using ZFN [62] and CRISPR/Cas9 [61] technologies. The resulting KO rats exhibited significant abnormalities in growth (body weight loss) as well as behavioral function (anxiety tendency and cognitive deficits) [61,62]. Because these phenotypes well recapitulate the major symptoms of RS patients, these Mecp2 KO rats will provide an alternative tool for future studies of MeCP2 functions.
Mutations in fused in sarcoma (Fus), a nuclear DNA/RNA-binding protein, cause familial ALS and occasionally frontotemporal dementia. Zhang et al. [87] produced KI rats expressing a Fus point mutation (R521C) as a model for ALS using the CRISPR/Cas9 technology. The mutant animals developed adult-onset learning and memory behavioral deficits, with reduced spine density in the hippocampal neurons. Remarkably, the sleep-wake cycle and circadian abnormalities preceded the onset of cognitive deficits. These results suggest a new role of Fus in sleep and circadian regulation and demonstrate that functional changes in FUS could cause sleep-wake and circadian disturbances as early symptoms.
Emmert et al. [113] produced a novel rat model of X-linked hydrocephalus (XLH) by CRISPR-mediated mutation in the L1 cell-adhesion molecule (L1cam) gene on the X chromosome. Hemizygous male mutants developed hydrocephalus and delayed development. The mutant rats did not show reactive gliosis, but exhibited hypomyelination and increased extracellular fluid in the corpus callosum.
GABAergic dysfunctions have been implicated in the pathogenesis of schizophrenia, especially the associated cognitive impairments. The level of the GABA synthetic enzyme glutamate decarboxylase 67 kDa isoform (GAD67) encoded by the GAD1 gene is downregulated in the brains of schizophrenia patients. Furthermore, a schizophrenia patient harboring a homozygous mutation of GAD1 has recently been discovered. Gad1 KO mice exhibited perinatal lethality [214], which precluded characterization at adult stages. Fujihara et al. [125] generated Gad1 KO rats using the CRISPR/Cas9 technology. Surprisingly, 33% Gad1 KO rats survived to adulthood, which made further characterization possible. The Gad1 KO rats exhibited impairments in both spatial reference and working memory without affecting adult neurogenesis in the hippocampus. In addition, Gad1 KO rats showed a wide range of behavioral alterations, such as enhanced sensitivity to an NMDA receptor antagonist, hypoactivity in a novel environment, and decreased preference for social novelty. These results suggest that Gad1 KO rats could be a novel model for studying cognitive deficits. Furthermore, Fujihara et al. [125] claimed the necessity to check species differences in the mode of phenotype manifestation when animal models of human diseases are considered.
Angelman syndrome (AS) is a rare genetic disorder characterized by severe intellectual disability, seizures, lack of speech, and ataxia. The gene responsible for AS is the ubiquitin protein ligase E3A (Ube3a) gene, which encodes for ubiquitin ligase E6-associated protein (E6AP). Dodge et al. [122] generated Ube3a KO rats using the CRISPR/Cas9 system. The resulting KO rats phenotypically mirrored human AS with deficits in motor coordination as well as learning and memory. This model can, thus, offer a new avenue for the study of AS.
Koster et al. [142] produced a pigmented KO rat model for lecithin retinol acyltransferase (LRAT) using the CRISPR/Cas9 system. The introduced mutation (c.12delA) is based on a patient group harboring a homozygous frameshift mutation in the Lrat gene (c.12delC), causing a dysfunctional visual (retinoid) cycle. The resulting KO rats exhibited progressively reduced electroretinography potentials from two weeks of age onwards and overall retinal thinning. Vision-based behavioral assays confirmed the reduced vision. These KO rats are a novel animal model for retinal dystrophy, especially for early-onset retinal dystrophies.

Models for Muscular Diseases
Duchenne muscular dystrophy (DMD) is an X-linked lethal muscle disorder caused by mutations in the Duchenne muscular dystrophy (Dmd) gene encoding dystrophin. DMD model animals, such as Mdx (X-linked muscular dystrophy) mice and canine X-linked muscular dystrophy dogs, have been widely utilized in the development of a treatment for DMD. However, according to Larcher et al. [51], large animal models such as dogs are expensive and difficult to handle. In contrast, Mdx mice only partially mimic the human disease, with limited chronic muscular lesions and muscle weakness. Their small size also imposes limitations on analyses. In this context, a rat model could represent a useful alternative because rats are bigger than mice and could better reflect the lesions and functional abnormalities observed in DMD patients.
Dmd KO rats were successfully produced by two groups using CRISPR/Cas9 technology targeting two exons of the rat Dmd [50] or using TALENs targeting exon 23 [51]. These resulting KO rats exhibited a decline in muscle strength and the emergence of degenerative/regenerative phenotypes in the skeletal muscle, heart, and diaphragm. Furthermore, these phenotypes were transmitted to the next generation [50]. Notably, Dmd KO rats, but not mice, present cardiovascular alterations close to those observed in humans, which are the main cause of death of patients [51].
Desminopathy is a clinically heterogeneous muscle disease caused by over 60 different mutations in the desmin (DES) gene. The most common mutation with a clinical phenotype in humans is an exchange of arginine to proline at position 350 of desmin leading to p.R350P. Langer et al. [126] first produced a KI rat model for a muscle disease through the CRISPR/Cas9 technology using ssODN carrying the missense mutation Des c.1045-1046 (AGG > CCG) in exon 6 of Des. While muscle weights did not differ between the mutant rats and WT rats, the levels of many muscle-related proteins such as dystrophin, syntrophin, dysferlin, and annexin A2 increased in the mutant rats, showing the phenotype of desminopathy. This rat model will be a useful tool for furthering our understanding of the disease and testing therapeutic approaches to delay disease progression.

Models for Pulmonary Diseases
Cystic fibrosis (CF) is characterized by airway and digestive pathology with a reduced life expectancy and is one of the most common genetic diseases in western populations. The most common mutation is the missense mutation p.Phe508del (or F508del) in the cystic fibrosis transmembrane conductance regulator (CFTR) gene, which leads to abnormal CFTR function and mucus accumulation. Tuggle et al. [52] first generated Cftr KO rats using the ZFN technology. The resulting KO rats lacked CFTR activity and exhibited abnormalities in the ileum and increased intracellular mucus in the proximal nasal septa. Airway surface liquid and periciliary liquid depths were reduced, and the submucosal gland size was abnormal. Similar results have also been observed in another Cftr KO rat developed by Dreano et al. [106] using the CRISPR/Cas9 technology. Furthermore, Dreano et al. [106] produced another KI rats (carrying F508del mutation). These rats showed residual CFTR activity and milder phenotype than the Cftr KO rats. These findings suggest that these rat models can be a CF animal model that recapitulates various aspects of the human disease. Notably, a humanized CF rat strain expressing the G551D variant was produced through KI of the human CFTR cDNA carrying the G551D mutation downstream of the endogenous Cftr promoter using the ZFN technology [130].

Models for Metabolic Diseases
The low-density lipoprotein receptor (LDLR) and apolipoprotein E (APOE) genes control normal levels of cholesterol and other forms of fat in the blood. A deficiency in ApoE is involved in several age-related fatty acid diseases. Wei et al. [56] established ApoE KO rats through TALEN-mediated gene targeting. After being fed with a high-cholesterol diet (HCD) for 12 weeks, the ApoE KO rats displayed typical dyslipidemia, although there was no obvious atherosclerotic lesion in the en face aortas. Notably, partial ligation caused the formation of plaques consisting of lipids and macrophages in carotid arteries from ApoE KO rats. Wei et al. [56] concluded that the ApoE KO rats can be a novel model for dyslipidemia. Lee et al. [110] produced ApoE KO rats using Cas12a (previously named Cpf1), an RNA-guided endonuclease, as a part of the CRISPR system. The resulting KO rats displayed hyperlipidemia and aortic lesions.
A deficiency in LDLR is a cause of familial hypercholesterolemia (FH). Zhao et al. [92] and Lee et al. [110] created Ldlr KO rats. These ApoE and Ldlr KO rats mimic pathological changes observed in hyperlipidemia and atherosclerosis in humans with genetic deficiencies and in normal individuals, suggesting usefulness in the research of atherosclerosis.
Melanocortin-3 and -4 receptors (MC3R and MC4R) regulate energy homeostasis. You et al. [70] generated Mc3r and Mc4r single-and double-KO (DKO) rats using the CRISPR/Cas9 system. Mc3r KO rats displayed hypophagia and decreased body weight, whereas Mc4r KO and DKO rats exhibited hyperphagia and increased body weight. All three mutants showed increased white adipose tissue mass and adipocyte size. These mutant rats will be important in defining the complicated signaling pathways of MC3R and MC4R. According to You et al. [70], both Mc4r KO and DKO rats are good models for obesity and diabetes research.
Hereditary tyrosinemia type I (HT1) is caused by a deficiency in fumarylacetoacetate hydrolase (FAH) enzyme. Fah-deficient mice and pigs are phenotypically analogous to human HT1, but do not recapitulate all chronic features of the human disorder, especially liver fibrosis and cirrhosis. Zhang et al. [63] produced Fah KO rats through MI of CRISPR/Cas9 components to obtain HT1 models. The Fah KO rats faithfully represented hypertyrosinemia, liver failure, and renal tubular damage. More importantly, they developed remarkable liver fibrosis and cirrhosis, which have not been observed in Fah mutant mice or pigs. These data suggest that Fah KO rats may be used as an animal model of HT1 with liver cirrhosis.
Pseudoxanthoma elasticum (PXE)-a heritable ectopic mineralization disorder-is caused by mutations in the ATP-binding cassette subtype C number 6 (ABCC6) gene primarily expressed in the liver and kidneys. These mutations result in generalized arterial calcification throughout the body in infancy. Li et al. [75] generated Abcc6 KO rats as models of PXE using the ZFN technology. The plasma inorganic pyrophosphate (PPi) level was reduced (<30%), leading to a lowered PPi/inorganic phosphate plasma ratio. When in situ liver and kidney perfusions were performed, the PPi levels in the perfusates were significantly reduced, but those in the liver of WT rats remained high. Li et al. [75] speculate that hepatic ABCC6 may play a critical role in contributing to plasma PPi levels, identifying the liver as a target of molecular correction to counteract ectopic mineralization in PXE.
Wolfram syndrome (WS) is a rare autosomal-recessive disorder caused by mutations in the Wolfram syndrome 1 (WFS1) gene and characterized by juvenile-onset diabetes, optic atrophy, hearing loss, and a number of other complications. According to Plaas et al. [76], no mutant Wfs1 mice displayed fasting hyperglycemia. In other words, previous mouse models of WS involved only partial diabetes and other disease symptoms. Plaas et al. [76] generated Wfs1 KO rats, in which exon 5 of the Wfs1 gene was deleted, resulting in a loss of 27 amino acids from the WFS1 protein. The resulting KO rats showed progressive glucose intolerance, glycosuria, hyperglycemia, and severe body weight loss by 12 months of age. They also exhibited neuronal abnormality such as axonal degeneration and disorganization of the myelin. The phenotype of these KO rats indicates that they have the core symptoms of WS.
Leptin is a cytokine-like hormone principally produced by white adipose tissues. Defects in leptin production cause severe obesity. Leptin receptor (Lepr) encoded by the diabetes (db) gene is highly expressed in the choroid plexus. Bao et al. [59] produced Lepr KO rats through the CRISPR/Cas9 technology. The resulting KO rats exhibited obesity, hyperphagia, hyperglycemia, glucose intolerance, hyperinsulinemia, and dyslipidemia. In contrast, Chen et al. [79] generated Lepr KO rats through the TALEN technology. These rat models could complement the existing models (db/db mice and Zucker rats) [215,216] and be useful for biomedical and pharmacological research in obesity and diabetes.
Angiopoietin-like protein 8 (ANGPTL8) is a liver-and adipocyte-derived protein that controls plasma triglyceride levels. Izumi et al. [89] generated Angptl8 KO rats through the CRISPR/Cas9 technology to clarify the roles of Angptl8 in glucose and lipid metabolism. The resulting KO rats exhibited decreased body weight and fat content, associated with impaired lipogenesis in adipocytes. Izumi et al. [89] suggest that ANGPTL8 might be an important therapeutic target for obesity and dyslipidemia.
Hereditary aceruloplasminemia (HA) is a genetic disease characterized by iron accumulation in the liver and brain. The mutation of the ceruloplasmin (Cp gene is thought to be related to HA pathogenesis. Kenawi et al. [115] generated Cp KO rats through the CRISPR/Cas9 system. The Cp KO rats exhibited decreased iron concentration, transferrin saturation, plasma ceruloplasmin, and ferroxidase activity, which is considered essential for macrophage iron release. Thus, Cp KO rats can mimic the iron hepatosplenic phenotype in HA, which will form the basis to understand and treat the disease.

Models for Kidney Diseases
The renin (REN)-angiotensin system plays an important role in the control of blood pressure and renal function. Moreno et al. [31] produced Ren KO rats through ZFNmediated GE system targeted to exon 5 of Ren. The resulting rats exhibited reduced body weight, lower blood pressure, and abnormal renal morphology (as exemplified by cortical interstitial fibrosis and abnormally shaped glomeruli). These results suggest the role of REN in the regulation of blood pressure and kidney function.
Primary hyperoxaluria type 1 (PH1) is an inherited disease caused by mutations in the mitochondrial localized alanine-glyoxylate aminotransferase (Agxt) gene, leading to abnormal metabolism of glyoxylic acid in the liver, subsequent endogenous oxalate overproduction, and deposition of oxalate in multiple organs, mainly the kidney. Patients with PH1 often suffer from recurrent urinary tract stones and finally renal failure. There is no effective treatment other than combined liver-kidney transplantation. Zheng et al. [94] produced Agxt KO rats as PH1 models through the CRISPR/Cas9 system. The resulting Agxt KO rats excreted more oxalate in the urine than WT animals and exhibited crystalluria with mild fibrosis in the kidney. These data suggest that Agxt-deficiency in mitochondria impairs glyoxylic acid metabolism and leads to PH1 in rats. This rat strain would be a valuable tool for developing innovative drugs and therapeutics.
Urate oxidase (uricase) encoded by UOX gene is a key enzyme whose disfunction causes hyperuricemia. Because of the low survival rate of Uox-deficient mice [217], Yu et al. [128] generated Uox KO rats through the CRISPR/Cas9 system targeting the exons 2 to 4 of Uox. The resulting Uox KO rats, called "Kunming-DY rats", were apparently healthy, with more than a 95% survival up to one year. The male rats' serum uric acid increased at levels significantly higher than those of WT rats. Kunming-DY rats exhibited histological renal changes including mild glomerular/tubular lesions, suggesting an alternative model animal to study hyperuricemia and associated diseases mimicking human conditions.

Models for Ophthalmology Diseases
The aryl hydrocarbon receptor (AHR) is a ligand-activated transcription factor which plays a role in the development of multiple tissues and is activated by a large number of ligands, including 2,3,7,8-tetrachlorodibenzo-p-dioxin. To examine the roles of the AHR in both normal biological development and response to environmental chemicals, Harrill et al. [40] produced Ahr KO rats through the ZFN technology targeting exon 2 of the Ahr and compared with an existing Ahr KO mouse model. Ahr KO rats, but not Ahr KO mice, displayed pathological alterations to the urinary tract, as exemplified by bilateral renal dilation (hydronephrosis). In contrast, abnormalities in vascular development were observed in Ahr KO mice, but not in rats. These findings suggest the differences in the role of AHR in tissue development, homeostasis, and toxicity between rats and mice.

Models for Hematological Systems
Hemophilia A is a genetic bleeding disorder resulting from factor VIII (FVIII or F8) deficiency. In preclinical hemophilia research, an animal model that reflects both the phenotype and pathology of the disease is required. Nielsen et al. [41] produced KO rats lacking detectable F8 activity through the ZFN technology targeting exon 16 of the F8 gene. Episodes of spontaneous bleeding requiring treatments were observed in 70% of F8 KO rats. Shi et al. [120] produced F8 KO rats in which nearly the entire rat F8 gene was inverted, causing translational stop six amino acids after the signal sequence, through the CRISPR/Cas9 technology to examine whether platelet F8 expression can prevent severe spontaneous bleeding in F8 KO rats. They showed that the severe spontaneous bleeding phenotype in F8 KO rats was successfully rescued by platelet-specific F8 expression through bone marrow cell transplantation.
Autoimmune regulator (AIRE) deficiency in humans induces a life-threatening generalized autoimmune disease called autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED), and no curative treatments are available. Several models of AIREdeficient mice have been generated; although they have been useful in understanding the role of AIRE in central tolerance, they do not reproduce the APECED symptoms accurately and, thus, there is still a need for an animal model displaying APECED-like disease. Ossart et al. [86] produced an Aire KO rat model using ZFN technology. The resulting KO rats exhibited several of the key symptoms of APECED disease, including alopecia, skin depigmentation, and nail dystrophy, which are much more pronounced than those in Aire KO mice and closer to manifestations in humans.

Others
Estrogens play pivotal roles in the development and function of many organ systems, including the reproductive system. Rumi et al. [48] generated estrogen receptor 1 (Esr1) KO rats using the ZFN system targeting exon 3 of Esr1. Both male and female Esr1 KO rats were infertile. Esr1 KO males had small testes with distended and dysplastic STs, whereas Esr1 KO females possessed large polycystic ovaries, thread-like uteri, and poorly developed mammary glands. In addition, the uteri of Esr1 KO rats failed to 17β-estradiol treatment. This rat model provides a new experimental tool for investigating the pathophysiology of estrogen action.
The forkhead box N1 (FOXN1) gene is known as a critical factor for the differentiation of thymic and skin epithelial cells. Goto et al. [69] generated Foxn1 KO rats through the CRISPR/Cas9 technology. The resulting Foxn1 KO rats exhibited thymus deficiency and incomplete hairless, which was characterized by splicing variants.
Multidrug resistance 1 (MDR1; also known as P-glycoprotein) is a key efflux transporter that plays an important role not only in the transport of endogenous and exogenous substances, but also in tumor MDR, one of the most important impediments to the effective chemotherapy of cancer. In rodents, two isoforms, Mdr1a and Mdr1b, encode MDR1. Liang et al. [112] produced DKO rats (in which both Mdr1a and Mdr1b had been disrupted) through the CRISPR/Cas9 system. Pharmacokinetic studies of digoxin, a typical substrate of MDR1, confirmed the deficiency of MDR1 in resulting KO rats. This rat model is a useful tool for studying the function of MDR1 in drug absorption, tumor MDR, and drug-target validation.

Perspective
Genome-engineering technology, especially the CRISPR/Cas9 system, has opened ways to create KO at the specific genes and targeted KI, perform gene replacement, and conditionally ablate gene expression with more ease. Furthermore, recently, CRISPR-based new tools such as use of other types of Cas9 endonucleases (i.e., Staphylococcus aureus Cas9 (SaCas9), Francisella novicida Cas9 (FnCas9), Neisseria meningitidis Cas9 (NmCas9), Streptococcus thermophilus Cas9 (St1Cas9), and Brevibacillus laterosporus Cas9 (BlatCas9)) and Cas12a (Cpf1) endonucleases (i.e., Acidaminococcus sp. Cpf1 (AsCpf1) and Lachnospiraceae bacterium Cpf1 (LbCpf1)) (reviewed by Nakade et al. [218]), prime editing (PE), and BE have emerged. For example, BE systems, as exemplified by ABE, have already been applied in rats [97,104]. The PE system is the latest addition to the CRISPR genome-engineering toolkit and represents a novel approach to expand the scope of donor-free precise DNA editing [219]. PE-based manipulation of mouse genome has already been successfully performed [220]. This approach will soon be applied to modify the rat genome. Manipulation of mtDNA by DdCBE coupled with the PB transposon system has become possible in rats [153]. These new technologies, together with the newly developed gene delivery tools such as in vitro or in vivo EP-based gene delivery (TAKE and i-GONAD/rGONAD), will accelerate the creation of more GM rats, which will shed light on the new role of the previously known genes, whose function had been investigated using mouse models but was judged as being far from the function inferred from human diseases.

Conclusions
To date, tools for modifying the mouse genome as exemplified by traditional genetargeting-based KO and KI technologies have been extensively employed in mice and have proven useful for understanding the molecular mechanism underlying human diseases. However, despite these efforts, the consequences obtained from the KO or KI mice were often different from those observed in human diseases. According to Barbaric et al. [221], discrepancies between the symptoms of human diseases and mouse phenotypes may be ascribed to redundant gene networks or alternative pathways or modifiers. This point was further strengthened when a number of GM rats were successfully produced through newly developed GE technologies (see Table 1). For example, Dmd KO rats (but not mice) presented cardiovascular alterations close to those observed in humans, which are the main cause of death in patients. Furthermore, Fah KO rats developed remarkable liver fibrosis and cirrhosis, which have not been observed in Fah mutant mice or pigs. These findings encourage the speculation that rats may better mimic the human situation than mice.