CNS-restricted Transduction and CRISPR/Cas9-mediated Gene Deletion with an Engineered AAV Vector.

Gene therapy using recombinant adeno-associated viral (AAV) vectors is emerging as a promising approach to treat central nervous system disorders such as Spinal muscular atrophy, Batten, Parkinson and Alzheimer disease amongst others. A critical remaining challenge for central nervous system-targeted gene therapy, silencing or gene editing is to limit potential vector dose-related toxicity in off-target cells and organs. Here, we characterize a lab-derived AAV chimeric (AAV2g9), which displays favorable central nervous system attributes derived from both parental counterparts, AAV2 and AAV9. This synthetic AAV strain displays preferential, robust, and widespread neuronal transduction within the brain and decreased glial tropism. Importantly, we observed minimal systemic leakage, decreased sequestration and gene transfer in off-target organs with AAV2g9, when administered into the cerebrospinal fluid. A single intracranial injection of AAV2g9 vectors encoding guide RNAs targeting the schizophrenia risk gene MIR137 (encoding MIR137) in CRISPR/Cas9 knockin mice resulted in brain-specific gene deletion with no detectable events in the liver. This engineered AAV vector is a promising platform for treating neurological disorders through gene therapy, silencing or editing modalities.

Introduction

Recombinant adeno-associated viral (AAV) vectors have met safety endpoints in several phase 1 gene therapy clinical trials for treating Hemophilia, Alpha-1 Antitrypsin deficiency, and Alzheimer disease amongst other indications.[1,2,3] Although vector redosing might be essential in some indications due to loss of gene expression observed in long term follow up studies,[4] preclinical studies continue to show promise and advance with cautious optimism. One concern noted in hemophilia gene therapy clinical trials is the potential for vector dose-related hepatotoxicity in patients as evidenced by a rise in transaminases.[2,5] Although resolvable by administration of anti-inflammatory steroids such as methyl prednisolone, permanent loss in gene expression has been observed.[2] The dose and composition of clinical AAV vectors has been shown to influence these outcomes in preclinical toxicity studies.[6]

Concurrent with the advent of AAV vector technology, recent studies in animal models have demonstrated the application of clustered, regularly interspaced, short palindromic repeats (CRISPR)/Cas9 technology for targeted disruption of genomic loci in vivo,[7,8,9,10] as well as disease correction.[8,11,12] While the recombinant AAV vector platform offers a versatile platform for therapeutic delivery of CRISPR/Cas9 in vivo, integration of the two technologies can pose unique challenges to clinical translation.

So far, over a hundred naturally occurring strains of AAV derived from human and animal tissues have been isolated.[13,14] Multiple strains ranging from AAV serotypes 1–9 and Rh.10 are under active development as gene therapy vectors for different clinical indications.[15,16] Recent data from clinical gene therapy trials to treat Parkinson and Alzheimer disease have provided information that is critical toward continued AAV vector development.[3,17] In addition, preclinical studies involving intravascular/intrathecal (IT) administration in mouse, canine, feline, and nonhuman primate models have supported the advancement of serotypes such as AAV9 as candidates for clinical trials.[18,19,20,21,22] Notably, AAV9 is currently being evaluated in phase 1 clinical trials for the treatment of Spinal muscular atrophy by intravascular injection and Giant axonal neuropathy through IT injection (Clinical trial identifiers NCT02122952 and NCT02362438).

A key observation from preclinical studies is that central nervous system (CNS) administration of several naturally occurring AAV strains often results in vector sequestration within peripheral organs such as the liver and spleen via systemic leakage.[19,22,23,24] In this study, we address this potential concern through vector design. We have previously demonstrated grafting of the galactose binding motif from AAV9 capsid, onto other AAV surfaces as a viable strategy to engineer AAV vectors with novel gene transfer properties. Specifically, the viral protein 3 amino acid residues on AAV9 capsid required for galactose binding were used to replace corresponding residues on the AAV2 capsid (viral protein1 numbering: Q464V, A467P, D469N, I470M, R471A, D472V, S474G, Y500F, S501A) using site-directed mutagenesis. This resulted in rational engineering of a chimeric AAV strain (AAV2g9) which harbors the ability to engage either heparan sulfate (HS) or galactose as receptors for cell entry.[25,26] Earlier studies employing a similar approach were used to replace the HS binding motif of AAV2 (585-RGNRQA-590) with the corresponding residues of AAV8 (585-QQNTAP-590). This resulted in the engineering of HS binding deficient version of AAV2 capsid i.e., AAV2i8. The latter heparin-binding deficient capsid was then used as a template toward the generation of chimeric AAV2i8g9 (ref. 37). In this study, we demonstrate exceptional CNS transduction efficiency following single bolus injections of AAV2g9 as validated by two routes of administration in neonatal and adult rodents. More importantly, we establish potentially useful attributes of AAV2g9 for CNS-specific gene therapy applications, namely, preferential neuronal tropism, restricted CNS biodistribution and minimal transgene expression in off-target peripheral organs. Furthermore, we extend these findings to demonstrate the utility of AAV2g9 for targeted disruption of the MIR137 locus in CRISPR/Cas9 transgenic mouse.[9]

Results

AAV2g9 and AAV9 display similar spread and transduction efficiencies in the CNS

We first carried out neonatal (P0) mouse intracerebroventricular injections (ICV) comparing AAV2g9 and AAV9 vectors packaging a self-complementary green fluorescent protein (ScGFP) cassette driven by the ubiquitous chicken β actin hybrid (CBh) promoter. At 2 weeks post administration (3.5 × 109 vector genomes (vg)/mouse), diaminobenzidine immunohistochemistry and confocal microscopy analysis of postfixed mouse brain sections revealed extensive GFP expression across multiple sections in the rostrocaudal axis of the mouse brain for both AAV strains (). Specifically, we observed robust transduction in the olfactory bulb, striatum, hippocampus, and cortical regions of the mouse brain. High magnification insets further demonstrate that AAV9 injections result in GFP expression in cells with neuronal or glial morphology. On the other hand, mice injected with AAV2g9 show selective GFP expression in neurons. Furthermore, it is important to note that AAV2g9 injections result in robust GFP expression in the CNS tissue immediately surrounding lateral ventricular cerebrospinal fluid (CSF) space. Specifically, we observed strong transduction along the hippocampal boundary within the caudal sections of AAV2g9 injected mouse brains. On a cellular level, these are mixture of ependymal cells and neuronal projections within corpus callosum and medial forebrain bundle system. This robust transduction was not observed due to AAV9 administrations.

To further compare cellular transduction profiles of the two vectors, we generated higher magnification confocal micrographs of functionally relevant regions of the brain including the cortex, amygdala, hypothalamus, and hippocampus. Mice injected with AAV9 or AAV2g9 showed comparably robust GFP expression in these regions (). Additionally, we observed that while AAV9 treatment resulted in GFP+ cells with neuronal or glial morphology, AAV2g9 mediated GFP expression was mostly restricted to neurons. We further validated this observation using immune colocalization with the neuronal antigen marker, NeuN or glial fibrillary acid protein marker, glial fibrillary acid protein (GFAP). These results were further confirmed upon quantitation, which revealed a ~4 fold reduction in GFP+ cells with glial morphology for AAV2g9, but no significant differences between the two vectors in neuronal populations (Figure 3a,b). Overall, these results suggest that AAV2g9 spreads efficiently across the neonatal mouse brain, while demonstrating a robust and neurotropic transduction profile from a single unilateral ICV injection.

AAV2g9 and AAV9 display different cell-type specificities in the CNS

We then generated AAV9 and AAV2g9 vectors packaging single-stranded (ss) GFP cassettes driven by the neuron-specific human synapsin I (hSyn) or the glia-specific GFAP promoter. Vectors (3.5 × 109 vg/mouse) were administered in postnatal day 0 (P0) mice via the ICV route and postfixed brain tissues analyzed at 2 weeks postadministration by diaminobenzidine immunohistochemistry, immunostaining, and confocal microscopy analysis. Both AAV9 and AAV2g9 vectors packaging transgene driven by hSyn promoter displayed robust transduction across multiple sections of the mouse brain (). Specifically, we observed GFP+ cells colocalizing with NeuN+ cells (yellow) in various regions including the striatum, cortex, hippocampus, amygdala, and hypothalamus (). As expected, no glial staining was observed for either vector supporting the potential application of such neuron-specific promoters for CNS gene therapy studies.

In contrast, to hSyn promoter driven expression, we observed striking differences in the patterns of GFP expression mediated by AAV9 and AAV2g9 vectors packaging the GFAP promoter driven transgene cassette. Notably, AAV9 injections resulted in widespread GFP expression in cells with glial morphology across the entire brain parenchyma in multiple brain sections (). In contrast, mice injected with AAV2g9 exhibit spatially restricted GFP expression within close proximity of the site of CSF injection (lateral ventricles) or along the immediate point of contact between the CSF and brain parenchyma, the inner meninges. To further confirm these observations, we coimmunostained different sections of the mouse brain for GFP expression (green) with neuronal marker-NeuN or glial marker-GFAP (red) as before. High magnification confocal micrographs generated at different brain regions show robust GFP+ expression from AAV9 injections in GFAP+ cells, but not NeuN+ cells (). These observed patterns of transgene expression were further confirmed by quantitative analysis, which revealed ~2.5-fold reduction in glial transduced area throughout the brain with AAV2g9 vectors under GFAP promoter activity (,). These results demonstrate that AAV9 vectors packaging GFAP promoter driven expression cassettes are excellent candidates for glial gene transfer applications, while AAV2g9 appears to more neurotropic and hence suitable for CNS gene transfer applications targeting neuronal populations. From the AAV biology perspective, AAV2g9 appears to preferentially transduce neurons over glia, while this is patently not the case with AAV9.

HS is critical for preferential neuronal transduction by AAV2g9 vectors

The interaction of AAV2 vectors with HS proteoglycans is associated with preferential neuronal transduction in the immediate vicinity to site of CNS administration.[27,28] Coadministration of soluble heparin has been shown to block the ability of AAV2 to bind hep and alter transduction profiles in vitro and in vivo.[29,30,31] Therefore, we compared the CNS transduction profiles of AAV2 and AAV2g9 to coinjections of AAV2g9 with 1 µg of soluble heparin in the neonatal mouse brain. Further, we generated a hep binding mutant, AAV2i8g9, as described in earlier studies by our lab.[26] Specifically, we injected equal viral titers (3.5 × 109 vg) of AAV2, AAV2g9, and AAV2g9 + soluble heparin or AAV2i8g9 packaging CBh-ScGFP transgene via the ICV route in P0 mice. As shown in , we observed a low level of GFP+ expression as evidenced by diaminobenzidine immunostaining within the hippocampal sections of AAV2 injected mouse brains. Consistent with earlier results, AAV2g9 displayed robust and widespread CNS transgene expression supporting the notion that the gal footprint enhances transduction efficiency and spread. Further, we observed a decrease in GFP+ expression across the brain parenchyma in AAV2g9 + soluble heparin treated animals as well as AAV2i8g9 treated mouse cohorts in comparison with AAV2g9 (–).

High magnification confocal micrographs of multiple regions of the mouse brain, namely; the somatosensory cortex, piriform cortex, motor cortex, dentate gyrus, amygdala, and hippocampal CA1, CA2, and CA3 regions were then generated. As seen in the top panel of , we observed sporadic expression in neurons and astrocytes in case of AAV2-injected mice. On the other hand, extensive and preferential neuronal transduction was observed across these regions in case of AAV2g9 as observed earlier (, second row panel). In contrast, loss of hep binding, either by competitive inhibition with soluble heparin or with the AAV2i8g9 mutant resulted in a decreased neuronal transduction in multiple brain regions (third row and bottom row panels, ). In addition, loss of neuronal transduction was accompanied by an increase in GFP+ cells with glial (astrocytic) morphology in these cohorts. Taken together, our observations suggest that hep binding is critical in determining the preferential neuronal tropism displayed by AAV2g9.

AAV2g9 and AAV9 display similar transduction profiles in adult mice after IT administration

To compare the two strains in adult mice, we carried out IT infusions of AAV9 or AAV2g9 packaging CBh-ScGFP reporter transgene into the lumbar CSF space of 8 week old mice. At 3 weeks postadministration of 1 × 1011 vg/mouse, we subjected postfixed tissues to immunohistochemical analyses and confocal imaging as outlined in online methods. As shown in , IT infusions of both AAV9 and AAV2g9 resulted in strong transgene expression in the lumbar, thoracic, and cervical spinal cord regions. Of note, we observed GFP expression in multiple gray and white matter regions of the spinal cord. Further, we compared cellular transduction profiles resulting from IT infusions of AAV9/2g9 vectors. Specifically, we focused on different regions of the adult CNS, namely: ventral horn, dorsal horn, intermediate gray matter region, ventral funiculus, lateral funiculus, ventral commissure, and rubrospinal tract. As shown in , both AAV2g9 and AAV9 infused mice exhibit robust GFP expression in these regions of the spinal cord. In both AAV9 and AAV2g9 injected mice, we observed that the GFP+ cells (green) in the gray matter regions of the spinal cord (ventral horn, dorsal horn, and intermediate) extensively colocalized with NeuN+ as well as GFAP+ cells. Within the white matter regions of spinal cord (ventral funiculus, lateral funiculus, ventral commissure, and rubrospinal tract), the GFP+ (green) cells from either injection extensively colocalized with GFAP+ cells (). In addition, we observe a consistent drop in transduction in progressive rostral sections of the brain from IT administrations of AAV9 or AAV2g9 (data not shown). It is important to note that parameters such as vector dose, volume, and infusion rates will likely need to be optimized further to achieve widespread gene transfer in the brain following IT injections. Overall, these observations also underscore the influence of route of injection on the cellular tropism of different AAV vectors.

AAV2g9 and AAV9 display distinct systemic biodistribution and off-target transduction profiles following CNS administration

Having established the robustness of the rationally engineered AAV2g9 strain, we compared the systemic leakage, vg biodistribution and gene expression profiles in peripheral organs resulting from ICV or IT injections of AAV9 and AAV2g9 in both neonatal and adult mice. At 3 days postadministration, we extracted whole genomic DNA from different organs and determined vg copies using quantitative polymerase chain reaction (PCR) in different peripheral organs including the liver, heart, and spleen. We observed a striking difference between the off-target biodistribution profiles of AAV9 and AAV2g9 vectors following ICV or IT injections. As seen in –, in the AAV9 injected neonatal mouse cohort, ~40, 30, and 3% of the viral genomes were recovered from the liver, heart, and spleen, respectively, in comparison with the brain (normalized to 100%). In contrast, AAV2g9 vg copy numbers ranged from ~1.4 0.7, and 0.003% in these peripheral organs. In the adult mouse cohort, AAV9 vg copies recovered in the liver, heart, and spleen were 188, 81, and 59%, respectively in comparison with vg copies recovered from the spinal cord (normalized to 100%). In contrast, only ~3, 4.5, and 10% of vg copies were recovered from these systemic organs in the AAV2g9-treated cohort (–). These results suggest that unlike AAV9, AAV2g9 is sparingly sequestered in off-target, systemic organs following CNS administration.

To further corroborate the biodistribution results, we harvested different peripheral organs from the ICV and IT mouse cohorts treated with either AAV2g9 or AAV9 and subjected postfixed tissue to immunostaining and confocal imaging analysis as outlined earlier. We observed robust GFP expression in the heart, liver, and spleen of both neonatal (ICV) () and adult (IT) () mice treated with AAV9. Consistent with biodistribution data, AAV2g9 treated mice exhibit near background levels of GFP expression in both ICV and IT cohorts. Together, these results are also consistent with the reduced leakage of AAV2g9 vectors (>100-fold) into the blood circulation that was observed in comparison to AAV9 vectors (). Briefly, AAV9 vector genomes were observed in blood as early as 15 minutes postadministration and continued to decline over 2 days. These results are consistent with earlier studies reported by our lab and others.[32,33] In contrast, AAV2g9 vector genomes remained near background levels with a slight increase in copy number observed at the 24- and 48-hour time intervals. Together, our results confirm that unlike AAV9, AAV2g9 undergoes significantly reduced systemic leakage following CNS administration and consequently displays low levels of sequestration and transgene expression in peripheral organs.

Brain-restricted deletion of MIR137 gene using AAV2g9 in CRISPR-Cas9 knock-in mice

We have established that CNS administration of AAV2g9 results in limited systemic exposure and efficient gene transfer within the neural tissue. To further validate the potential advantages of the latter attributes, we generated AAV2g9 vectors packaging two guide RNAs (gRNAs) for targeted disruption of the MIR137 gene in the brain. Specifically, the gRNAs were designed to recognize both ends of the 85bp pre-MIR137 region. Equal doses of AAV2g9 packaging MIR137gRNA or controlgRNA were administered into Cas9 transgenic mice[9] via unilateral ICV injections. At 2 weeks post vector administration the mice were sacrificed and organs were harvested. Brain and liver tissues were then subject to genomic DNA extraction. To evaluate gene disruption events, we utilized the droplet digital PCR (ddPCR) technique. Briefly, primers were designed to amplify 206 bp mouse genomic regions flanking the MIR137 target locus. Fluorescent probes were designed to bind MIR137gRNA target region (FAM, red) and an unspecific downstream locus (HEX, green). Successful disruption of MIR137 locus results in exclusive excitation of HEX probe alone (green), whereas both FAM and HEX probes are excited (orange) in case of no gene disruption events (). Next, ddPCR analysis revealed that the frequency of MIR137 eliminated alleles was significantly higher (green dots) in mice that received AAV2g9-MIR137gRNA, as compared with AAV2g9-controlgRNA cohort (, red arrow). Correspondingly, quantitative analysis of this phenomenon demonstrated a significant increase in mutant allele frequency within the MIR137gRNA injected mouse brains ().

Next, we examined whether AAV2g9 vector-mediated gene disruption events were specific to the brain. Identical ddPCR analysis conducted on liver DNA extracted from the same mouse cohorts displayed near background levels of MIR137 elimination in the systemic organ (). Further quantitation demonstrated negligible mutant allele formation due to both AAV2g9-controlgRNA and AAV2g9-MIR137gRNA injections in the liver (). In addition, consistent with earlier results, AAV2g9 vector genome biodistribution analysis demonstrated significant enrichment of AAV genomes within the brain in comparison with systemic liver tissue (). Overall, our data demonstrates CNS-restricted gene disruption within the Cas9 transgenic mouse using ICV administration of engineered AAV2g9 vector.

Discussion

Recent large scale GWAS studies conducted across ~40,000 patients led to the discovery that disruption in MIR137 gene function is strongly correlated with Schizophrenia disease occurrence.[34,35] Here, we utilize this well-known gene target to illustrate the utility of a novel, lab-derived AAV vector for CNS-specific gene transfer. Since MIR137 and CAG promoter driven Cas9 are both expressed constitutively in brain and liver tissue,[9,36] it is safe to assume that brain specific MIR137 disruption relies on AAV2g9 vector design. The efficiency of AAV-mediated gene editing achieved in the current study is similar to that demonstrated earlier with the CRISPR/Cas9 knock-in mouse model.[9] However, improvements to vector design could further improve gene editing efficiency. More importantly, from the vector development perspective, these studies have broad implications for gene therapy, silencing and editing.

First, it should be noted that the CNS transduction efficiency and systemic gene expression from a single ICV/IT dose of AAV9 is striking. Although this might not be ideal for certain CNS-specific indications due to the systemic leakage profile, AAV9 is certainly an optimal vector candidate for treatment of lysosomal storage disorders and other such diseases, which are characterized by multiorgan involvement in addition to the CNS. On the other hand, AAV2g9 displays favorable attributes for therapeutic applications targeting neuronal populations within the CNS. Further, the features displayed by the engineered AAV2g9 strain exemplify the feasibility of making rational improvements to AAV vector design with the goal of imparting favorable biodistribution profiles. When combined with cell-specific elements such as the hSyn promoter, as established in the current study or miRNA targeting elements,[22,23] such approaches could help restrict transgene expression to neuronal populations within the CNS and reduce systemic exposure. In addition, therapeutic delivery of gene editing nucleases using AAV2g9 could help mitigate the risk for off-target effects at the organ level.

From the AAV biology perspective, these studies provide new insights into the role of capsid-glycan interactions in determining CNS spread and systemic leakage. Earlier studies in our lab and others have demonstrated that the diverse cellular tropisms and biodistribution profiles of different AAV strains is intricately linked to their glycan receptor binding profiles.[37,38] For instance, intracranial injection of the HS binding strain, AAV2 shows restricted, neuronal gene expression.[27] This preferential neuronal tropism of AAV2 appears to correlate with the higher concentration of heparan sulfate proteoglycans on the surface of neurons compared with glia.[39] Accordingly, AAV2g9 retains its preferential neuronal tropism, likely due to its ability to bind HS within the CNS. Correspondingly, both competitive inhibition with soluble heparin coadministration and disrupting HS binding with loss of function capsid mutation (AAV2i8g9) resulted in ablated neuronal transduction in vivo. However, as a caveat to this neurotropic bias, the ability of AAV2 to bind HS with high affinity also restricts vector spread and the effective volume of CNS tissue transduced.[28] In contrast, the galactose binding strain, AAV9, shows both neuronal and glial transduction as well as efficient CNS spread.[18,40] One possible explanation for the increased spread of AAV2g9 as compared with AAV2 can be attributed to capsid interactions with proteins in the CSF that might facilitate spread and transport within the brain. Alternatively, while axonal transport of AAV2 is exclusively restricted to the anterograde direction, AAV9 is known to travel in both anterograde and retrograde directions to cover long distances within the neural tissue.[16,21,37,40] We acknowledge that such hypotheses warrant further evaluation and such detailed studies will likely shed light on the mechanism(s) driving enhanced CNS spread of the dual glycan binding AAV2g9. Consistent with the latter attributes, AAV9 capsids also appear to be prone to systemic leakage and transduction in the liver and other organs following CNS administration.[19,22,23] In this study, the dual glycan binding strain has inherited traits from both parental AAV serotypes, i.e., neurotropism from AAV2 and robust CNS spread and transduction from AAV9.

Materials and Methods

Recombinant AAV vector production. An updated triple plasmid transfection protocol was used to generate recombinant AAV vectors. The pXR2g9 and pXR2i8g9 plasmids have been described earlier.[26] All other plasmids used for AAV production in this study were obtained from the UNC vector core. Briefly, the transfection mixture contained (i) the pXR9 helper plasmid or pXR2g9/2i8g9 helper plasmids; (ii) the adenoviral helper plasmid pXX6-80; and (iii) pTR-CBh-ScGFP, pTR-hSyn-EGFP, pTR-GFAP-EGFP or pTR-CBA-Luc plasmids encoding the GFP or luciferase (Luc) reporter genes driven by the CBh, hSyn, GFAP or the chicken β actin (CBA) promoters, flanked by inverted terminal repeats derived from the AAV2 genome. Vector purification was carried out using iodaxinol gradient ultracentrifugation protocol, buffer exchange and concentration using vivaspin2 100 kDa molecular weight cut-off (MWCO) centrifugation columns (F-2731-100 Bioexpress, Kaysville, UT). Vg titers were obtained by quantitative PCR (Lightcycler 480, Roche Applied Sciences, Pleasanton, CA) using primers designed to selectively bind AAV2 inverted terminal repeats (forward, 5′- AAC ATG CTA CGC AGA GAG GGA GTG G-3′; reverse, 5′- CAT GAG ACA AGG AAC CCC TAG TGA TGG AG-3′) (IDT Technologies, Ames, IA).

ICV administration. Animal experiments reported in this study were conducted with Balb/C or C57/Bl6 mice bred and maintained in accordance to NIH guideline as approved by the UNC Institutional Animal Care and Use Committee (IACUC). Postnatal day 0 (P0) Balb/c pups which were rapidly anesthetized on ice for 2 minutes followed by stereotaxic ICV injections. Specifically, AAV9 or AAV2g9 vectors packaging different transgenes were injected into the left lateral ventricle (total volume < 3 µl) using a Hamilton 700 series syringe with a 26s gauge needle (Sigma-Aldrich, St. Louis, MO), attached to a KOPF-900 small animal stereotaxic instrument (KOPF instruments, Tujunga, CA). All neonatal injections were performed 0.5mm relative to the sagittal sinus, 2 mm rostral to transverse sinus and 1.5 mm deep. Following vector administration, mice were revived under a heat lamp and rubbed in the bedding before being placed back with the dam. At 2 weeks post vector administrations (P14) the mouse brains were harvested, postfixed and immunostained as described below in detail. For coadministration of soluble heparin (generously provided by Dr Jian Liu, School of Pharmacy, UNC), AAV vectors (3.5 × 109 vg) were mixed with 1 µg of heparin and a total volume of 2–3 µl administered per mouse for 30 minutes on ice prior to ICV injections in neonatal mice.

IT administration. AAV vectors were infused into the mouse IT cerebrospinal fluid space using Alzet mouse IT catheter and pump (Alzet, 0007743, Durect, Cupertino, CA). Briefly, the pumps were primed with 0.9% NaCl for ~12 hours followed by AAV vectors. Later 8 weeks old C57/Bl6 males were anesthetized with intraperitoneal injection of Avertin (1.25%, 2,2,2-tribromoethanol in phosphate-buffered saline) at 0.23 ml/10 g body weight prior to infusions. A 23 gauge needle was used to expose the L5–L6 intervertebral space and the osmotic pump was implanted and sutured under the skin. The vector was then infused at the rate of ~8 µl/hr for ~24 hours to infuse a total 1 × 1011 vg of AAV9 and AAV2g9 vectors packaging the CBh-ScGFP transgene. At 3 weeks post administration of the vectors, the mice were sacrificed, postfixed in paraformaldehyde, sectioned and immunostained as described below. Another cohort of 8 week old Balb/C mice was injected via IT bolus injections with AAV9 and AAV2g9 vectors as described elsewhere.[41] Briefly, equal titers of vectors packaging the CBA-Luc transgene cassette (total volume 4–5 µl) was free-hand injected into the L5 and L6 intervertebral space. The injections were carried out an angle of 20–30 degrees from spinal column using the 30G disposable needle attached to 50 µl Luer-hub Hamilton syringe (Sigma-Aldrich).

Vector genome biodistribution and pharmacokinetic analyses. Cohorts of neonatal and adult mice used for biodistribution studies were sacrificed 3 days post vector administration via the ICV or IT bolus routes. The genomic DNA was extracted from the tissue lysates and blood using the DNeay kit (Qiagen, Valencia, CA). To calculate viral genome copy numbers quantitative polymerase chain reaction was performed with primers specific to luciferase transgene 5′-AAAAGCACTCTGATTGACAAATAC-3′ and 5′-CCTTCGCTTCAAAAAATGGAAC-3′. The vector genome copy numbers were normalized to mouse lamin B2 locus as the housekeeping gene using the primers 5′-GGACCCAAGGACTACCTCAAGGG-3′ and 5′- AGGGCACCTCCATCTCGGAAAC-3′. The vector biodistribution was represented as the ratio of vector genomes per cell recovered in the peripheral organs to the CNS site of injection (brain or the spinal cord). For pharmacokinetic studies, the vector genome copy numbers were calculated from total DNA isolated from 10 µl blood.

Tissue processing, immunofluorescence, and confocal microscopy. Neonatal and adult mouse cohorts were sacrificed 2 weeks and 3 weeks post vector administrations respectively. The mice were overdosed with tribromoethanol (avertin) (0.2 ml/10 g of 1.25% solution) via the intraperitoneal route. This was followed by transcardial perfusions of phosphate-buffered saline and 4% paraformaldehyde in phosphate-buffered saline. The organs were removed and postfixed for 24 hour prior to sectioning. Briefly, 50 μm thick sections were obtained using a Leica VT 1,200S vibrating blade microtome (Leica Biosystems, Buffalo Grove, IL). The spinal cords were cryo-sectioned by UNC animal histopathology core. The sections of mouse organs from various treatments were blocked in 10% goat serum (Sigma-Aldrich) and 1% Triton X (Sigma-Aldrich) in phosphate-buffered saline for 1 hour. This was followed by overnight incubation with primary monoclonal antibodies at 4̊C. The primary antibodies utilized as a part of this study are as follows: Rabbit anti-GFP (Life-Technologies-G10,362, 1:750) mouse anti-GFAP (Abcam-10,062, 1:1,000), rabbit anti-NeuN (Abcam-104,225, 1:750), chicken anti-GFP (Abcam-13,970, 1:750). Secondary antibodies were raised in goats and conjugated to Alexa 488 (anti-rabbit Abcam-96,883, anti-chicken Abcam-96,947), Alexa 594 (anti-Rabbit Abcam-96,885) or Alexa 647 (anti-mouse Abcam-150,115). The secondary antibodies were used at a standard dilution of 1:500. The immunohistochemical analyses of GFP expression was conducted using Vectastain ABC kit (Rabbit IgG PK-4001 kit, Vector biolabs, Burlingame, CA). We used Zeiss CLSM 700 confocal laser scanning microscope for imaging sections of different organs after immunostaining (Microscopy Services Laboratory, UNC). The images were stitched, pseudocolored, and analyzed on the Zen Black v 10.0 software.

CRISPR gRNA design. Two gRNAs were designed to detect 98bp Pre-MIR137 flanking region and analyzed by COSMID (crispr.bme.gatech.edu) to check potential off-target sites.[42] Target sequencing used to make the gRNAs are following: mir-137-g1; CGTCACCGAAGAGAGTCAG, mir-137-g3; GTAGTCGAGGAGAGTACCAG. For control gRNA, we used a 20 bp sequence, which recognizes an unspecific backbone region of plasmid DNA (Con-g1; GTCGACTCTAGAGGATCCAC). Tandem repeat of U6 promoter-target sequence-guideRNA is conjugated with EF1core promoter-tdTomato-P2A-PuroR and subcloned into the pTR vector for packaging into AAV.

Droplet digital PCR. PureLink Genomic DNA extraction kit (Thermo Fisher Scientific, K182,002) was used to obtain DNA from brain and liver tissues. Following sequences of primers and probes were used for the ddPCR assay. MMIR137L206ddPCR; GCAGCAGTGACAGCGGTAGC, MMIR137R206ddPCR; TGGCAACCGGGAGCTTTTAG, MMIR137MTFAMddPCR; /56-FAM/TCCACCCAA/ZEN/GAATACCCGTCACCG/3IABkFQ/, MMIR137WTHEXddPCR; /5HEX/CCCTCCCAG/ZEN/CCCACCAGCTG/3IABkFQ/.

For ddPCR, 2× ddPCR Supermix for Probes (No dUTP; Bio-rad, 1,863,023), the QX100 Droplet generator and reader (Bio-rad) were employed. 10 ng genomic DNAs were subjected to PCR amplification by C1,000 thermal cycler (Bio-rad) with the following condition: 95°C 10 minutes × 1, (4°C 30 seconds, 65°C 1 minutes) ×40, 98°C 10 minutes ×1; 2°C/sec. Obtained data was analyzed by the QuantaSoft software (Bio-rad).

Study approval. All neonatal and adult Balb/c or C57/Bl6 mice used in the study were bred, maintained and utilized in accordance to NIH guidelines as approved by the UNC Institutional Animal Care and Use Committee (IACUC) (Protocol # 15–109).

Author contributions

G.M. and A.A. conceived the study. G.M., L.R., and T.C. carried out vector production, administration, histology, and image analysis. K.S. and P.S. conducted the gRNA design and gene disruption analyses. D.W. and G.G. assisted with IT administration and provided input for the overall study. G.M. and A.A. analyzed the data and wrote the manuscript.