A Newly Built rAAV-SaCas9 Genome Editing System Enables Muscle- Directed Gene Editing to Improve Muscular Atrophy

Shaoting Weng (  89742715@qq.com ) Henan Agricultural University https://orcid.org/0000-0001-8953-4436 Xingyu Li Henan Agricultural University Yitian Zhao Henan Agricultural University Feng Gao Henan Agricultural University Mengmeng Shi Henan Agricultural University Haoxiang Yang Henan Agricultural University Liqiang Han Henan Agricultural University Jiang Wang Henan Agricultural University Guoyu Yang Henan Agricultural University


Background
The CRISPR/Cas system has become a commonly used technology in biomolecular research and biomolecular therapies [1,2]. This system can be used to create genetically modi ed animals [3,4], and it provides a way to resist viral invasion [5]. Moreover, genome-wide screening can be performed by using the programmable properties of CRISPR/Cas9 [6,7]. In addition, CRISPR/Cas9 has been used to genetically modify potential clinical treatments for diseases [8][9][10]. However, genome editing at speci c sites in vivo is affected by many factors, including the choice of vector, the e ciency of editing proteins and the in uence of the internal environment.
In vivo delivery of CRISPR/Cas9 has been achieved through hydrodynamic delivery or viral transduction, including with adenoviral or adeno-associated virus (AAV) vectors. In clinical settings, AAV as the vector for Cas9 provides a promising gene therapy method [11]. In particular, this technology can greatly bene t muscle atrophy treatment by enabling selective manipulation of gene expression in speci c muscle regions [12,13]. However, many of the exciting tools developed from CRISPR-SpCas9 technology are too bulky to meet the genome packaging limits of AAV (~ 4.85 kb including both ITRs). One way to overcome this technical hurdle is to take advantage of smaller orthologs of Cas9 derived from different prokaryotic species. Three Cas9 nucleases that have been recognized to be effective have similar properties to those of Streptococcus pyogenes Cas9 (SpCas9), but are markedly smaller. Notably, SpCas9 is ~ 4.3 kb, and it scarcely ts into the AAV genome when coupled with essential gene regulatory elements. However, Streptococcus thermophilus Cas9 (St1Cas9), Neisseria meningitidis Cas9 (NmCas9), and Staphylococcus aureus Cas9 (SaCas9) are all ~ 1 kb shorter than SpCas9 [14][15][16]. Thus, with the discovery of these smaller Cas9 nucleases, AAVs can now be engineered with either of these smaller Cas9 genes and gRNA expression cassettes to create AAV-based CRISPR/Cas9. This technical hurdle can also be overcome by optimizing the promoter. Mefferd and coworkers have recently reported that expression of sgRNAs can be driven by small tRNA promoters (~ 70 bp) with sizes approximately half that of an sgRNA-expressing cassette with a U6 promoter [17] for e cient delivery of CRISPR materials into cells. Tabebordbar and coworkers have reconstructed a new vector AAV-SaCas9 in which expression of SaCas9 is driven by EF1a-short promoters [13]. Other researchers have built the pX601-miniCMV-SaCas9-U6-sgRNA vector, in which expression of SaCas9 is driven by a miniCMV promoter, which is only 39 bp.
The e ciency of expression and the smaller sizes of two or more components of a single rAAV system would be necessary to perform experiments on the system's genome-editing capabilities in aspects, including rAAV tissue tropism, and e cient precise targeting and persistent performance of organizations [18]. For example, AAV9 has strong tropism in muscle tissue, and AAV9-Cas9 gene editing can reduce the harm caused by off-target editing and prevent other tissue mutations.
Myostatin (Mstn, also known as growth differentiation factor 8), is a member of the transforming growth factor β (TGF-b) signaling protein superfamily and a key regulator of muscle mass in vertebrates. Its expression inhibits the proliferation and differentiation of muscle cells. Myostatin signaling dysfunction promotes muscle growth, and myostatin-null animals have a characteristic supermuscle phenotype [19,20]. Unsurprisingly, in muscular dystrophy diseases, including sarcopenia, muscular dystrophy, and cancer-related cachexia, controlling myostatin signaling has become an attractive prospect for increasing functional muscle mass [21][22][23][24][25]. The generation of effective antibodies or inhibitors targeting myostatin signals to promote muscle growth remains clinically challenging. Many myostatin binding antibodies, designed to treat amyotrophic disease by inhibiting myostatin signaling, have failed in primary clinical phase II trials (bimagrumab by Novartis and PINTA 745 by Atara; Novartis, 2016; Atara Bio, 2015). Similarly, an ActRIIB receptor-Fc fusion (ACE-031 by Acceleron) was withdrawn from phase II trials because of safety concerns [22]. Currently, no myostatin inhibitors are approved for clinical use, primarily because the cross-reactivity of soluble antagonists with structure-related TGF-b superfamily growth factors inhibits the transmission of mature myostatin signals [26]. However, researchers are increasingly exploring treatments for muscular dystrophy by knocking out Mstn gene expression, which has been found to be a simple and rapid way to improve muscle traits [23,27,28].
Here, we built a smaller rAAV-SaCas9 gene editing system, which generated EF1α-driven SaCas9 and tRNA-driven sgRNAs. We then inserted the eGFP sequence into the rAAV-SaCas9 and demonstrated the expression effects of eGFP and SaCas9 gene editing in vivo and in vitro. Furthermore, we demonstrated the e cacy of the system in mice with muscle wasting. This study provides a reference for clinical treatment applications.

Materials And Methods
Construction of SaCas9 and sgRNA plasmids.
Procedures were carried out as previously described [29]. For the production of rAAV-DJ/8, rAAV9 viruses, HEK293T cells were seeded into 20 dishes (100 ⋅ 10 mm; 5 × 10 6 cells per dish). After 24 h, the cells were triple-transfected with pX601/psgMstn1, pAAV-DJ/8-RC pAAV9-RC and pHelper according to the instructions provided with the 1 × PEI transfection reagent. After 72 h, the cells were harvested by scraping into medium, centrifuged at 1000 ⋅ g for 10 min and resuspended in 1 mL of 1 × phosphate-buffered saline. The cell suspension was subjected to three freeze-thaw cycles at 80 °C and at 37 °C. After fast centrifugation and ltration, the cell debris was cleared. The viral solution was concentrated with PEG 8000 and puri ed on a cesium chloride density gradient column. After two rounds of ultracentrifugation, the high-density viruses were separated and extracted, and run through dialysis bags for desalting [30]. The titers of the puri ed rAAV-DJ/8, rAAV9 viruses were determined with a RT-PCR-based method described previously [31]. pX601 was diluted from 10 9 copies/µL to 10 3 copies/µL as the standard solution. The primers were ITR-QPCR-F: 5 - Adeno-associated virus (AAV) serotypes differ broadly in transduction e cacy and tissue tropism. Recombinant AAV-DJ vectors provide superior in vitro transduction e cacy to that of any other wild type serotype. E cient transduction is bene cial for gene editing in the SaCas9 system. After digestion and resuspension of C2C12 cells in logarithmic growth phase, the cells were cultured in DMEM supplemented with 10% FBS in 12-well plates at a seeding density of 1 × 10 5 and grown overnight. The cells were cultured to 50-60% con uence and infected with ~ 2 × 10 9 vg and 1 × 10 10 vg rAAV-DJ/8-sgMstn1, and control cells were infected with ~ 2 × 10 9 vg and 1 × 10 10 vg rAAV-DJ/8-CN. Then, the medium was replaced with DMEM (3% FBS) after 24 h. Fluorescence was determined, and DNA and protein were extracted at 96 and 120 h.
Mice. 120 C57BL/10 male mice in SPF grade (6 weeks of age) were purchased from the Center of Experimental Animal of Guangdong province (Guangzhou, China) and maintained in a speci c-pathogen-free animal facility according to the Guide for the Care and use of Laboratory Animals and the related ethical regulations at Henan Agricultural University. Then, these mice were fed to the age of 8 weeks and were healthy without any abnormalities. rAAV9 transduction in vivo.
Compared with AAV-DJ, AAV9 is more addicted in muscle tissue; it is conducive to gene editing of muscle cells and is often applied in muscle tissue. In addition, the packaging of different types of AAV allows for broad applicability of the psgMstn1 plasmid. 60 C57BL/10 male mice were divided into four groups, and rAAV9-sgMstn1 and rAAV9-CN were injected into male C57BL/10 mice in two doses. Three points on the thigh muscle in the left thigh in the treatment group were injected with ~ 1 × 10 10 vg/1 × 10 11 vg of rAAV9-sgMstn1 in 100 µL of phosphate-buffered saline (30 µL/point,N = 15), and the control group were treated similarly with ~ 1 × 10 10 vg/1 × 10 11 vg of rAAV9-CN in 100 µL of phosphate-buffered saline (30 µL/point, N = 15). After 6, 8 and 10 weeks, the expression of the tissue uorescence was determined, and the thigh muscles were collected for genomic DNA extraction, western blotting.

Construction of mouse model of muscular atrophy and rAAV9 transduction.
A mouse model of muscular atrophy was constructed as previously described [32]. 40 C57BL/10 male mice were divided into two groups: a DEX treatment group, which received DEX treatment once every other day for 14 days (DEX, Decadron, 0.5 mg/kg of body weight, i.p., N = 30), and a control group, which received saline during the same period (N = 10). DEX and saline injections were simultaneously given each day (9:00-10:00 a.m.). Then, 20 C57BL/10 male mice with signi cant weight changes were selected from the DEX treatment group, and ~ 1 × 10 11 vg of rAAV9-sgMstn1 or rAAV9-CN was injected into the thigh muscle in muscular atrophy model mice (N = 10). Control group mice were injected with saline. After 8 weeks, the thigh muscles were collected for genomic DNA extraction, western blotting and tissue slice experiments. Body weight was measured daily in each week of the experiment. ACACACCTACCTTTGGAGTAAG-3 . The fragment sizes ampli ed by these primer sets were 521 bp, 548 bp and 537 bp, respectively. The PCR products were digested with T7 endonuclease 1 (NEB, Boston, USA) and resolved with 1.5% agarose gel electrophoresis. The primers used for tracking of insertions and deletions (indels) by decomposition (TIDE) were the same as the MSTN-Test-F/R primers. Genomic DNA (100 ng) was used for PCR ampli cation with a High Fidelity 2⋅ PCR Master Mix (NEB). For TIDE analysis, 300 ng of PCR product was puri ed with a QIAquick PCR Puri cation Kit (Qiagen, Hilden, Mannheim, Germany) and sent for Sanger sequencing with the forward primer MSTN-Test-F. Indel values were obtained with the TIDE web tool (https://tide.deskgen.com/) as described previously. Targeted deep-sequencing analysis was performed for C2C12 cells and gDNA from mouse muscle with a PCR ampli cation approach. Brie y, the off-target locus was identi ed through the website http://www.rgenome.net/caso nder/. On-target or off-target locus-speci c primers (Table 1)  Protein extracts were prepared on ice by homogenization of pieces of frozen tissues or cell pellets in 500 µL of RIPA lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS and 2 mM MgCl 2 ) supplemented with 1:100 protease inhibitor solution (Roche, Basel, Switzerland) by passage through a syringe. With each loose-and tight-tting piston, the samples received 30 strokes and were then centrifuged at 1000 ⋅ g for 5 min at 4 °C to remove debris. The supernatant (whole cell lysate) was collected. For the isolation of membrane fractions, the supernatant was further centrifuged at 13200 ⋅ g for 20 min at 4 °C, and the pellet was resuspended (2 µL/mg tissue) in sample buffer (2.7 M urea, 3.3% SDS and 0.167 M Tris, pH 6.7). The protein concentration was estimated with a BCA assay. For protein detection, 30 µg of sample was separated with 10% SDS-PAGE and then transferred to a polyvinylidene uoride membrane. After incubation in 5% nonfat milk for 1 h, the membrane was incubated with rabbit polyclonal anti-MSTN/GAPDH antibody (1:1000, Bioss Antibodies, China, catalogue number: bs-23012R/bs-2188R) overnight at 4 °C. Then membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit (1:2000, Bioss Antibodies, catalogue number: bs-0295G) antibodies for 1 h at room temperature. The target proteins were detected with Luminata ™ Crescendo immunoblotting HRP Substrate (Millipore, USA).
Histology and immuno uorescence analyses.
The quadriceps muscles were xed in para-formaldehyde and embedded in para n for further histopathological investigations. Formalin-xed quadriceps muscles sections were stained to investigate the density and diameter of muscle bers by hematoxylin and eosin according to a standard protocol [33]. After the staining procedure, the slides were scanned with a microscope. At least ve elds at 200⋅ magni cation were randomly selected from each section in each group for imaging. The number of bers was counted in each eld, then converted to the number of bers per mm 2 . Each section was analyzed to calculate the area of individual muscle bers (mm 2 ), by selecting the area of every muscle ber (mm 2 ). All data were obtained and analyzed with Image Pro Plus 6.0 software. For immunostaining, the para n sections of muscle tissue were dewaxed as previously described [34], placed in a solution of sodium citrate at 100 °C for 10 minutes and then soaked in hydrogen peroxide. Cross-section samples were immunostained with mouse monoclonal anti-SaCas9 primary antibody (1:200, Epigentek, catalogue number: A-9001) and goat anti-mouse IgG-FITC (1:500, Abcam, catalogue number: ab6785) secondary antibody, and DAPI for nuclei. Muscle sections were imaged with standard uorescence microscopy.
Muscle weights of mice.
The experimental C57BL/10 male mice were weighed on electronic scales. The body weights of the mice were recorded from weeks 1 to 8. For the measurement of muscle weight, quadriceps and adductor muscles from the left side in the experimental mice were dissected 2 months later, and the average weight was used for each muscle. Ten mice were weighed per group.

Statistical analyses.
Data are expressed as the mean ± standard error of the mean. Unpaired Student's t-test was used to analyze differences between two groups, and oneway ANOVA analysis of variance with Bonferroni's post-test was used for multiple group comparisons with Prism 6 (GraphPad). * P values less than 0.05 and greater than 0.01 were considered signi cant. ** P value less than 0.01 was considered to be of greater signi cance.

Results
SaCas9/CRISPR system exhibits e cient genome editing in vitro.
In our experiment, we focused on the design of a compact CRISPR/Cas9 system suitable for AAV delivery. We constructed a viral plasmid with gene editing e ciency. The CRISPR/Cas9 system plasmid contains an EF1α promoter controlling expression of the SaCas9 coding region fused to a selfcleaving P2A sequence attached to an eGFP sequence, and it contains a constitutively expressing sgRNA expression cassette controlled by a tRNA promoter. The P2A sequence enables the SaCas9 and the eGFP domain to yield two independent proteins from the same cistron, and the translated proteins from the two genes retain their functions. This approach can be used to indirectly understand the transfection ability of recombinant AAV and the gene editing effect mediated by SaCas9 protein through the expression level of GFP protein (Fig. 1A).
To examine the genome editing capabilities of our system, we transfected NIH/3T3 cells with the pX601-:EF1α:SaCas9-eGFP-tRNA:sgRNA plasmids in the treatment groups. The sgRNAs were designed to target the mouse Mstn locus (sgMstn1/ sgMstn2/ sgMstn3); the control groups were transfected by pX601 plasmids with an empty sgRNA (Fig. 1B). We chose to target the Mstn locus because it is a well-studied gene known for its important role in muscle of overgrowth and is important in research on some amyotrophic lateral diseases. The three sgRNAs are located at the N-terminal of the rst and second exon of Mstn gene, and these sites play an important role in the synthesis of MSTN precursor proteins. At 72 h post transfection, the cells were observed and collected by centrifugation. The expression of eGFP from the transfected plasmid psgMstn1/psgMstn2/psgMstn3 was con rmed by cellular immuno uorescence in NIH/3T3 cells (Fig. 1C). A ow cytometry experiment was performed on NIH/3T3 cells transfected with the psgMstn1/psgMstn2/psgMstn3 plasmids. The results con rmed expression of the recombinant vectors, at an expression e ciency of 46.40%, 43.81% and 42.42%, respectively (Fig. 1D). Genomic DNA was isolated, and the 521 bp, 548 bp and 537 bp regions from the sgMstn1/ sgMstn2/ sgMstn3 loci were PCR ampli ed and examined for SaCas9-mediated editing with a T7 endonuclease digestion identi cation method. With this assay, if no editing occurred, a single band was visible on the gel. The results were as shown in the control groups. When editing did occur, the DNA was cut at the location of the editing, thus resulting in two smaller bands of 181 and 340 bp in length for sgMstn1; 185 and 352 bp in length for sgMstn2; and 122 and 415 bp in length for sgMstn3 (Fig. 1E). For TIDE analysis, 300 ng of PCR product was sent for Sanger sequencing with the forward primer MSTN-Test1/ 2/ 3-F.
Compared with the results in the control group transfected with plasmid pX601, the sequencing results of psgMstn1/psgMstn2/psgMstn3 plasmids showed that the crest of the edited cells at the knockout site was not a single, more disorderly (Fig. 1F). Indel values were obtained with the TIDE web tool (https://tide.deskgen.com/). Cells that received the pX601 vectors did not exhibit editing of the MSTN locus. However, cells transfected with psgMstn1/psgMstn2/psgMstn3 plasmids displayed an average of 31.43%, 18.14% and 27.94% editing, respectively (Fig. 1G). Genome editing percentages were averaged from three independent samples per group. From the above data, we determined that all three plasmids had gene editing ability. Compared with that of sgMstn2 and sgMstn3, the gene editing e ciency of sgMstn1 was higher, and the knock-off e ciency was lower. Therefore, sgMstn1 was selected as the rst choice for in vivo and in vitro targeting of packaged rAAV vectors.
The rAAV-DJ/8-SaCas9 system e ciently targets the Mstn site in C2C12 cells. First, we built rAAV-DJ/8-sgMstn1, pseudotyped as DJ/8 serotype ( Fig. 2A). Next, we tested the rAAV-DJ/8-sgMstn1 infectivity with two doses in C2C12 cells, as well as the knockout effect on MSTN protein in the cells.
To do so, we infected C2C12 cells with ~ 2 × 10 9 vg/1 × 10 10 vg of rAAV-DJ/8-sgMstn1 in the treatment groups. The control groups comprised C2C12 cells infected with ~ 2 × 10 9 vg/1 × 10 10 vg of rAAV-DJ/8-CN. All cells were observed and harvested 96 and 120 h later and processed as described above. The expression of eGFP in the treatment groups was con rmed by cellular immuno uorescence in C2C12 cells. No eGFP expression was observed in the control groups ( Fig. 2A). A ow cytometry experiment was performed on C2C12 cells infected with the rAAV-DJ/8-sgMstn1 to con rm expression of the transgene. Their editing e ciency for low and high doses, respectively, was 4.36% and 6.21% at 96 h, 17.31% and 24.84% at 120 h (Fig. 2B). The eGFP expression e ciency of the high dose group was higher than that of the low dose group in vitro. Genomic DNA was isolated from a 521 bp region from the sgMstn1 locus, PCR ampli ed and examined for SaCas9-mediated editing with a T7 endonuclease digestion identi cation method. The results indicated two smaller bands 181 and 340 bp in length for sgMstn1 (Fig. 2C). In TIDE analysis, low and high doses rAAV-DJ/8-sgMstn1 resulted in an average of 8.42% and 11.94% editing at 96 h, and 16.75% and 21.88% editing at 120 h (Fig. 2D). Cells that received rAAV-DJ/8-CN did not exhibit editing of the MSTN locus. Genome editing percentages were averaged from three independent samples per group. Compared with the control groups, the treatment groups infected with low or high doses rAAV-DJ/8-sgMstn1 from 96 to 120 h showed a gradual decrease in MSTN protein levels (Fig. 2E). A stacked histogram showing the percentage distribution of MSTN expression at different time points in C2C12 cells indicated 80.6% and 71.5% at 96 h, and 69.4% and 62% at 120 h. Data represent the means ± SD from three technical replicates (Fig. 2F). E cient editing and MSTN protein reduction indicated the success of the delivery and activity of rAAV-DJ/8-sgMstn1 at the Mstn locus, and demonstrated that gene editing of the cells with high doses of the virus was more effective.
The rAAV9-SaCas9 system e ciently targets the Mstn site in mouse thigh muscle. We sought to assess whether our rAAV9-SaCas9 system could be used for steady expression in vivo. In these experiments, we produced rAAV9-sgMstn1, pseudotyped as 9 serotype. In the treatment groups, rAAV9-sgMstn1 were infected at low and high doses into the left thigh muscle in C57BL/10 male mice, each at a titer of ~ 1 × 10 10 vg/1 × 10 11 vg. At 6, 8 and 10 weeks later, the mice were sacri ced, and eGFP uorescence in their bodies was detected with an in vivo imaging system. The control groups were infected with rAAV9-CN. We found that the uorescence intensity of eGFP protein increased with time, and the uorescence intensity of the high-dose rAAV9-sgMstn1 group was signi cantly higher than that of the low-dose group. (Fig. 3A). Genomic DNA from mouse muscle was isolated, and a 521 bp region from the sgMstn1 locus was PCR ampli ed and examined for SaCas9-mediated editing with a T7 endonuclease digestion identi cation method. The results revealed two smaller bands 181 and 340 bp in length for sgMstn1 in the treatment groups, which were absent in the control groups. (Fig. 3B). For TIDE analysis, low dose rAAV9-sgMstn1 displayed an average of 6.31%, 13.11% and 18.8% editing, and high doses of rAAV9-sgMstn1 displayed an average of 9.29%, 17.59% and 21.68% editing at 6, 8 and 10 weeks, respectively (Fig. 3C). Muscle from mice that received the rAAV9-CN did not exhibit editing of the Mstn locus. The genome editing percentages were averaged from three independent samples per group. Compared with the control group muscles, in the muscle infected with low or high doses of rAAV9-sgMstn1, MSTN protein levels were gradually reduced at 6, 8 and 10 weeks (Fig. 3D) weeks, respectively. The data represent means ± SD from three technical replicates (Fig. 3E). E cient editing and MSTN protein reduction indicated the success of the delivery and activity of rAAV9-sgMstn1 at the Mstn locus in mice. In addition, the results con rmed that gene editing was more e cient with high viral doses in vivo.
The rAAV9-SaCas9 system effectively edited genes and improved muscle mass in muscle-atrophic mice.
In our nal experiment, we sought to test our single rAAV9-SaCas9 system in mice with muscle atrophy. We tested rAAV9-sgMstn1 in C57BL/10 male mice with muscle atrophy that showed listlessness and weight loss, and muscle cells shrink and die at the cellular level. E cient editing and MSTN protein reduction were considered to indicate successful delivery and activity of rAAV9-sgMstn1 at the Mstn locus. Mstn de ciency led to skeletal muscle proliferation in mice. Compared with the mice injected with rAAV9-CN, in mice injected with rAAV9-sgMstn1, muscle was substantially added in the left thigh. Wild type mice were used as a negative control (Fig. 4A). There was no substantial difference in body weight between the groups injected with rAAV9-sgMstn1 and rAAV9-CN at the beginning of the experiment, whereas at the end of the experiment, especially in the last 3 weeks after injection, an increase was observed, owing to increased weights following injection with rAAV9-sgMstn1 (P < 0.01) (Fig. 4B). In addition, we clearly found the weight change of quadriceps and adductor at the injection site. Treatment with rAAV9-sgMstn1 compared with the rAAV9-CN signi cantly increased the quadriceps weight (P < 0.01) and adductor weight (P < 0.01) (Fig. 4C and D). These results revealed muscle weight increases in the Mstn knockout muscle. Our negative control was wild type mice, which had a steady level of muscle growth and received saline. To elucidate the successful gene editing in mice muscles, we examined the expression of SaCas9 protein in rAAV9-sgMstn1 in muscle cells. Immunostaining revealed green uorescence in the muscle sections of mice infected with rAAV9-sgMstn1 (Fig. 4E). Hematoxylin and eosin staining of the rAAV9-sgMstn1-treated, rAAV9-CN-treated and saline-treated groups were performed, and the transverse sections of the muscle bers showed different density distributions.
The rAAV9-CN-treated group had signi cantly more lytic and atrophic cells than the rAAV9-sgMstn1-treated and the saline-treated groups. And the ber areas and numbers clearly differed in size (20 µm) when the transverse sections of the muscle bers were enlarged (Fig. 4F). According to analysis of the quadriceps muscle sections, we observed that the average area of muscle ber cells was signi cantly greater in the rAAV9-sgMstn1-treated group than in those from the rAAV9-CN treated (Fig. 4G) (0.01 < P < 0.05). By analyzing the number of muscle bers per unit area, we found that the number of muscle bers treated with rAAV9-sgMstn1 increased (Fig. 4H) (0.01 < P < 0.05). The results revealed that the number of muscle bers increased, and muscles enlarged, with decreased MSTN expression in muscle tissues.
SaCas9 is highly speci c in vivo.
The possibility of off-target edits is a major concern in therapeutic CRISPR/Cas9 genome editing. Some researchers have found that SaCas9 is a naturally highly occurring genome editing platform in mice [35] [36]. Using targeted deep-sequencing of mouse cell DNA products obtained by GENEWIZ Inc., we screened for off-target sites in the mouse genome to determine whether SaCas9 maintained its minimal off-targeting pro le in mice cells.
NIH/3T3 cells were transfected with psgMstn1, psgMstn2 and psgMstn3 plasmids. The resulting genomic DNA was subjected to sequencing analysis.
Sequencing revealed few off-target sites in the mouse genome, a result consistent with our previous observations in mouse cells. Six potential offtarget sites were identi ed for sgMstn1, ve for sgMstn2 and another nine for sgMstn3, with a low probability of off-target sites in the mouse genome (Fig. 5A). We concluded that SaCas9 editing is intrinsically speci c. We also performed targeted deep-sequencing with genomic DNA from the muscles of mice to validate these off-target sites. According to this more sensitive readout, indels detected more accurately above background at all of the offtarget locus 1 sites with Mstn (Fig. 5B). These results indicated that rAAV9-sgMstn1 is a promising and safe candidate for in vivo applications.

Discussion
Several CRISPR/Cas9-based-inducible systems have been developed, but few have been developed for the SaCas9 variant. Since the characterization of SaCas9 [16], several groups have extended its application in targeted mutagenesis in a variety of models such as plants, mice and zebra sh [37][38][39].
Because of its smaller size, SaCas9 provides substantial advantages in the delivery and expression of Cas9, especially when AAVs are used. Recent work has shown that SaCas9 has higher activity than the other Cas9 variants such as SpCas9 and FnCpf1 [40].
To our knowledge, there are currently few available single AAV viruses containing both the Cas9 and sgRNA. Among them, the SaCas9 system has been shown to be effective for gene editing in AAV in vivo [16]. These systems not only provide comparable editing levels to those of other SpCas9 systems, but also have the advantage of being within the AAV packaging limit. Moreover, researchers can manipulate gene expression at precise sites and to precise degrees of expression. Therefore, these SaCas9 systems can probably be used to edit genes at speci c sites to improve tissue traits and treat chronic diseases [41,42]. First, we report the development of an SaCas9 system with a different sgRNA promoter (tRNA pro ~ 72 bp) and SaCas9 promoter (EF1α pro ~ 212 bp) that can be delivered to cells via rAAV. This single vector system contains a smaller Cas9, a uorescent protein and a gRNA expression cassette. Only one virus is needed to mediate genome editing in animals or cell lines. We believe that this system should be extremely useful and may be used separately or in combination with other expression cassettes in experiments.
We observed favorable gene editing e ciency by AAV in vitro across our experiments. Through infection of C2C12 cells, we observed uorescence expression of the eGFP protein in rAAV-DJ/8-sgMstn1. One study has indicated that when the SaCas9 system targets the HT1080 cell line in vitro, the genome editing rate is ~ 40-50% [43]. In contrast, our system showed relatively low editing e ciency ~ 17.31% in vitro with the same amount of virus, but by increasing the amount of virus, we substantially improved the editing e ciency approximately ~ 24.84%. We suspect that the recombinant system maybe slightly less e cient at editing in vitro, owing to lower sgRNA expression from the U6 to tRNA promoters. This possibility seems reasonable, given that the success of system depends on maintaining tight levels of promoter control of sgRNA expression. We found that increasing the amount of rAAV-DJ/8-sgMstn1 improved the expression of eGFP protein and the e ciency of SaCas9 protein editing. We also found that rAAV-SaCas9 was expressed stably in cells for a long time in our experiments. Therefore, genome editing e ciency can be improved by increasing the expression time and titer of recombinant viruses.
Because of differences in AAV serotypes, transcription factors and injection sites, AAV-mediated delivery of transgenes in tissues often differs [29]. In addition, the titer of the AAV virus and the method of virus injection also greatly affect the e ciency of AAV-mediated transgene delivery. Intramuscular injection was used because it has a highly targeted effect on tissues. The effect is more easily analyzed by comparing the differences in muscle tissues, cells and corresponding proteins without considering the overall effect on local areas. In addition, systemic injection may cause adverse effects speci c to nerves, immune cells, hormones and other factors, and may cause a large area of off-target editing and even cancer mutations in offspring. Mstn knockout in mice also resulted in changes in body characteristics. Therefore, we focused on local targeted knockout experiments in gene editing. In our in vivo experiments, we infused our rAAV-SaCas9 system into the left thigh in C57BL/10 male mice. Therefore, genome editing was restricted to the SaCas9 system within this region. We found that the higher the rAAV9-sgMstn1 titer, the higher the uorescence intensity of eGFP protein; moreover, we were able to clearly observed the location of the rAAV9-sgMstn1 through eGFP protein uorescence. We observed the feasibility and effectiveness of genome editing in these systems, which achieved approximately ~ 13.11% genome editing at 8 weeks. These editing e ciencies are higher than those in previous studies from laboratory examining SaCas9 genome editing [23,44]. In agreement with the results of in vitro experiments, increasing the amount of recombinant virus in vivo also promoted the genome editing e ciency of the virus. The genome editing e ciency in the high virus titer group was 17.59% at 8 weeks, and the greatest amount of genomic editing was ~ 21.68%. Therefore, within a certain period of time, when the amount and expression time of the virus increased, the gene editing of the virus in vivo achieved the best effect by using rAAV9-Mstn1.
Muscle wasting occurs with aging and in a wide range of catabolic diseases, such as cancer, diabetes, chronic renal disease and heart failure, which can dramatically decrease quality of life and increase disease mortality [23]. Recently, in vivo genome editing targeting muscle tissues with CRISPR-Cas systems has shown e cacy in a mouse model of Duchenne muscular dystrophy [45]. With an aim to establish a therapeutic treatment for muscle-wasting syndrome, we used CRISPR/Cas9 to directly target Mstn in vivo in skeletal muscle cells to prevent loss of muscle mass. We believed that by blocking the MSTN protein expression and myostatin pathway of some skeletal muscle cells, these cells should be able to prevent the atrophy of target cells and adjacent cells, thus maintaining muscle function to a certain extent [46]. Our constructed rAAV system has been shown to be effective for gene editing of thigh muscles. Our AAV-SaCas9 viruses target muscle, owing to the natural tropism of AAV9 for muscle within the mouse leg. We immunostained the SaCas9 protein to indicate its expression in muscle and possible editing in vivo. Thomas and Joulia et al. have found that MSTN up-regulates the activity and level of cyclin-dependent kinase inhibitors, thus preventing G1 phase to S phase transition, inhibiting the proliferation of muscle cells, and decreasing the number of muscle bers [47,48]. McPherron and coworkers have found that mutant mice have signi cantly enlarged muscles; in mice with gene knockout of Mstn, the number of muscle bers was 86% greater than that in wild-type mice, thus indicating that Mstn knockout causes muscle ber growth and hypertrophy [19]. The results of our experiment are consistent with that conclusion. Treatment with rAAV9-Mstn1 resulted in signi cantly higher body weight, quadriceps weight and adductor weight than those in the muscular dystrophy group. Through analysis of muscle tissue, we con rmed that partial knockout the Mstn gene improved the volume and number of muscle bers, an effect particularly important for the treatment of muscular atrophy.

Conclusion
Gene editing for all of the systems could probably be optimized further by using highly e cient sgRNAs and high titer AAVs, increasing the incubation time of these viruses in vivo, and using serotypes of AAV that transduce target cells with high e ciency. In this paper, we demonstrate that the constructed rAAV-SaCas9 genome editing system confer stable expression in vitro and in vivo. Moreover, The uorescence intensity of eGFP protein and the gene editing e ciency of SaCas9 protein increased with increasing rAAV-SaCas9 titer. The uorescence intensity of the eGFP protein was also used to determine the infection effects and the location of the rAAV-SaCas9. In addition, we edited muscle-atrophic mice with a high dose of rAAV9-SaCas9 virus, which effectively improved the muscle properties of mice. The results show that the rAAV-SaCas9 system has substantial therapeutic potential for muscular atrophy. Figure 1 The SaCas9/CRISPR system exhibits e cient genome editing in vitro. (A) AAV vector maps depicting EF1α:SaCas9-eGFP-tRNA:sgRNA. EF1α:SaCas9-eGFP-tRNA:sgRNA is an AAV vector that consists of an EF1α promoter controlling the expression of the coding regions of SaCas9, followed by a self- h post transfection, cells were harvested, and genomic DNA was isolated and puri ed. The 521 bp, 548 bp and 537 bp regions of the sgMstn1/sgMstn2/sgMstn3 locus, including the site targeted for editing, were PCR ampli ed from genomic DNA and analyzed for genome editing with a T7 endonuclease digestion identi cation method. The top band on the gel is uncut DNA at 521 bp, 548 bp and 537 bp in length, and the two smaller bands are edited DNA at 181 and 340 bp in length for sgMstn1, 185 and 352 bp in length for sgMstn2, and 122 and 415 bp in length for sgMstn3. The nucleic acid marker was a 50 bp DNA ladder. (F) The sgMstn1/sgMstn2/sgMstn3 knockout sequence peak was obtained by Sanger sequencing. For a control, genomic DNA of the pX601-treated group was sequenced. (G) Stacked histogram showing the percentage distribution of indels at sgMstn1, sgMstn2 and sgMstn3 in NIH/3T3 mutant cells, as measured by sequencing analyses. Data represent means ± SD from three technical replicates.  The rAAV9-SaCas9 system e ciently targets the Mstn site in C57BL/10 male mice thigh muscle. (A) For the rAAV9-SaCas9 system, the expression range was 2 × 1010-5 × 1010photons/second/cm 2. The images are shown at 6, 8 and 10 weeks post-injection (dpi). Owing to the weak uorescence expression of eGFP, uorescence imaging results were not visible through the mouse fur. Therefore, we killed the mice rst and then stripped off the fur for uorescence imaging. To exclude rAAV9-CN expression in the thigh, a control group is included in the rst image for the low/high dose rAAV9-sgMstn1 treatment groups (ventral position). (B) Mouse muscles were infected with low/high rAAV9-sgMstn1 in the treatment groups, and the control groups were infected with the same dose of rAAV9-CN. At 6, 8 and 10 weeks (with) post infection, mouse muscle was harvested, and genomic DNA was isolated and puri ed. A 521 bp region of the sgMstn1 locus including the site targeted for editing was PCR ampli ed from the genomic DNA and analyzed for genome editing with a T7 endonuclease digestion identi cation method. The top band on the gel is uncut DNA at 521 bp in length in all groups, but the two smaller bands are edited DNA at 181 and 340 bp in length for sgMstn1 at 6, 8 and 10 weeks only in the treatment groups. The  The rAAV9-SaCas9 system e ciently improves muscle mass in muscle-atrophic mice. (A) Changes in left thigh muscle mass in C57BL/10 male mice in the WT, rAAV9-CN treated and rAAV9-sgMstn1-treated groups (ventral position). Compared with that in the rAAV9-CN treated group, the muscle mass in the rAAV9-sgMstn1-treated group was signi cantly increased (red cycle). (B) Changes in body weight from 1 to 8 weeks in different experimental groups. (C) and (D) The weights of the quadriceps and adductor muscle in the different groups at 8 weeks. N=10/group, *: 0.01 < P < 0.05 **: P < 0.01 for the rAAV9-sgMstn1-treated group compared with the WT and rAAV9-CN treated groups. (E) Tissue immuno uorescence tests were performed on