Artificial miRNAs Reduce Human Mutant Huntingtin Throughout the Striatum in a Transgenic Sheep Model of Huntington's Disease

Abstract

Huntington's disease (HD) is a fatal neurodegenerative disease caused by a genetic expansion of the CAG repeat region in the huntingtin (HTT) gene. Studies in HD mouse models have shown that artificial miRNAs can reduce mutant HTT, but evidence for their effectiveness and safety in larger animals is lacking. HD transgenic sheep express the full-length human HTT with 73 CAG repeats. AAV9 was used to deliver unilaterally to HD sheep striatum an artificial miRNA targeting exon 48 of the human HTT mRNA under control of two alternative promoters: U6 or CβA. The treatment reduced human mutant (m) HTT mRNA and protein 50–80% in the striatum at 1 and 6 months post injection. Silencing was detectable in both the caudate and putamen. Levels of endogenous sheep HTT protein were not affected. There was no significant loss of neurons labeled by DARPP32 or NeuN at 6 months after treatment, and Iba1-positive microglia were detected at control levels. It is concluded that safe and effective silencing of human mHTT protein can be achieved and sustained in a large-animal brain by direct delivery of an AAV carrying an artificial miRNA.

Introduction

Huntington's disease (HD) is a fatal neurodegenerative disease caused by a genetic expansion of the CAG repeat region in the huntingtin (HTT) gene. The HTT gene, which was discovered in 1993,¹ encodes a ∼350 kDa protein comprising a 17 amino acid N-terminal region, followed by the CAG repeat, a polyproline repeat region, and a series of α-helical Huntingtin, elongation factor 3, protein phosphatase 2A, and TOR1 (HEAT) repeats.² Gene-silencing therapeutics, which include methods targeting RNA (antisense oligonucleotides, siRNAs, artificial miRNAs) and DNA (CRISPR/Cas9, Zinc finger nucleases), are promising, as they are directed to the most proximal cause of HD, the HTT mRNA or gene itself. Studies in HD mouse models support the feasibility of reducing the levels of the HD gene and its protein in the brain using antisense oligonucleotides, siRNAs, shRNAs, and miRNAs.^3–12 Intrathecal delivery of an antisense oligonucleotide against HTT is currently being tested for safety in humans¹³ but may have limited effect in the caudate–putamen.⁷ Long-lasting expression of shRNAs or artificial microRNAs can be achieved in the brain using a viral delivery approach and direct injection with stereotactic methods. Recombinant adeno-associated virus (rAAV) has been shown to deliver its cargo safely to neurons and other cell types in the striatum of mice.^14–16 Compared to an antisense approach, the advantage of AAV is that it is episomal and can be indefinitely active. Promoter choice determines expression levels of the AAV cargo, which may be an important determinant of the safety and silencing activity of potential RNAi therapies for HD.^17,18

A challenge in advancing human gene therapy for HD is scaling of therapeutics to a capacity that is comparable to treating humans. The sheep brain is significantly larger than that of mice and with a structure more similar to humans than that of rodents. Recently, it was shown that AAV-GFP could be accurately delivered to the neostriatum of normal Dorset sheep using an image-guided approach and convection-enhanced delivery.^19,20 HD transgenic sheep express the full-length human HTT cDNA with 73 CAG repeats from the human HTT promoter.²¹ They exhibit slow disease progression. Only a small number of animals have been analyzed to date, but reduced cannabinoid receptor type 1 (CB1) and DARPP-32 immunoreactivity have been reported in a single 7-month-old animal.²¹ At 6 months, there is a decrease in gamma-aminobutyric acid A receptor alpha 1 (GABA_A α1) immunostaining in the caudate nucleus and the putamen, and a decrease in leu-enkaphalin (Leu-ENK) in the globus pallidus.²² There are no detectable changes in immunostaining for glial fibrillary acidic protein (GFAP), a marker for astrogliosis, up to 36 months.²² Sparse cortical huntingtin aggregates appear by 18 months of age, and nuclear inclusions have been reported at 36 months.²² Young animals show no obvious behavioral phenotypes, but subtle disruptions in circadian rhythms appear by 18 months of age.²³ This study reports safe and successful targeting of full-length human HTT mRNA in a large-animal model of HD, a necessary first step in the treatment of patients with HD.

Materials and Methods

Animals and animal procedures

Merino sheep were bred at the South Australia Research and Development Institute (Rosedale, SA) and transferred to the South Australia Health and Medical Research Institute (Gilles Plains, SA) approximately 3 weeks prior to surgery. The sheep were provided with food and water as needed and were cared for according to the guidelines of the South Australia Health and Medical Research Institute (approval number 146/13) and the Primary Industries and Regions South Australia (10/13) animal ethics committees. Prior to the administration of anesthetic, the animals were fasted overnight for approximately 8 h. Animals were given a preoperative physical, including heart rate, respiratory rate, temperature, and weight. Baseline samples of serum (5 mL) and cerebrospinal fluid (CSF) were collected.

The study was conducted in two parts with two different cohorts of sheep. For the first study, 41 transgenic animals (21 wethers, 20 ewes), aged approximately 8 months, were injected unilaterally with 300 μL of self-complementary AAV9 (scAAV9) vector at a titer of 1 × 10¹³ genome copies (gc)/mL for a total of 3 × 0¹² gc. Animals weighed between 39 and 55 kg at the time of injection. For the second study, 14 animals, aged 14 months, were injected with test vector and 14 with the control vector. At this time, animals weighed between 52 and 79 kg. Gadolinium was added to the vector formulation to allow postsurgical imaging of the injection spread. The animals were moved to the operating room and prepped for surgery. Buprenorpine and carprofen were administered, and animals were anesthetized using ketamine and diazepam. Animals were maintained under isoflurane during the surgery. They were rested in the sphinx position on a foam cushion on folded extremities or with extremities dangling. A stereotactic frame (large animal; Kopf Instruments) was used to hold the animal's head in place. CSF was collected via lumbar puncture using a 19-gauge spinal tap cannula. The rAAV was delivered directly to the striatum, targeting the internal capsule. The animal's head was shaved, prepped with betadine, and draped with clear plastic. A curvilinear incision was made using a #15 scalpel to expose the bregma. Once the bregma was identified, a 3–4 mm burr hole was placed 10 mm rostral to the bregma and 11 mm lateral of the midline using an electric drill. The convection enhanced delivery (CED) cannula (MRI Interventions) was secured in the manipulator and primed with agent to be injected to remove air from the line. The dura was opened with a 1.5 mm incision using a #11 scalpel, and the CED cannula was advanced 25 mm from dural surface to the target depth. The outer cannula (1.65 mm) sealed the dural incision to prevent CSF leakage during the infusion. The infusion began 5 min after cannula insertion to allow for tissue around the tip to stabilize. The infusion rate was set at 3.33 μL/min until a total volume of 300 μL was injected. Ten minutes after the infusion was completed, the cannula was slowly withdrawn, and a bone wax plug was used to repair skull and prevent CSF leakage. The wound was cleansed with saline and closed using a 3.0 vicryl suture. A standard anesthesia wake-up and recovery procedure was followed, according to guidelines at the South Australia Health and Medical Research Institute. Post-surgery magnetic resonance imaging was performed to determine the spread of gadolinium. One animal from the first study was excluded following surgery because no gadolinium was visible upon imaging, and a second animal from the second study was excluded because the gadolinium appeared to be primarily in the ventricle. After surgery, the animals were kept under observation for 3 days and housed indoors for 5 days. They were then transferred and housed outdoors in paddocks for the remainder of the study. They were monitored visually for signs of distress and changes in behavior throughout the study. Two animals suffered surgical complications, resulting in partial limb paralysis. This was thought to be due to the positioning of the animals under anesthesia. One was euthanized early, and one was moved from the 6-month to the 1-month cohort. Animals were weighed periodically throughout the post-injection period, and samples of CSF, blood, and serum were taken and saved for further analysis.

For cell counts and differentials, blood was collected via jugular venipuncture into a potassium EDTA blood collection tube (lavender top) and delivered to the Automated Directorate (SA Pathology, Royal Adelaide Hospital, Adelaide, Australia) for a complete blood examination with differential. For clinical chemistry, blood was collected via jugular venipuncture into a serum collection tube (red top). The samples were submitted to SA Pathology for multiple biochemical analysis.

At 1 and 6 months post injection, animals were harvested, with animals being used for either histology or biochemical analysis. Animals were transported to the operating table and placed in ventral recumbency while approximately 6 mL of CSF was collected. The animal was repositioned in dorsal recumbency. The carotid arteries were exposed and cannulated at a depth of 4 cm from the tip of the cannula. The jugular veins were exposed, and 200–500 IU of heparin/kg were injected into the jugular vein. Five minutes after administering the heparin, the sheep were euthanized by intravenous injection of Lethabarb (325 mg pentabarbitone sodium/mL) at 1 mL/2 kg of body weight. The infusion pump was primed with cold 9% NaCl and connected to the carotid cannulas. For immunohistochemical analysis, the animal was perfused with approximately 8 L of cold 9% NaCl at a pressure of 500 mmHg followed by infusion with 8 L of 4% paraformaldehyde at a pressure of 500 mmHg. The brain and liver were extracted. The tissues were post fixed in 4% paraformaldehyde for 24 h at 4°C and transferred to 30% sucrose in 1 × phosphate-buffered saline (PBS) for a minimum of 14 days at 4°C

For RNA, protein, and DNA assays, sheep were perfused with cold 9% NaCl, as described above. Collection of the peripheral tissue was performed in the following order: liver, adrenal gland, ovaries (if applicable), muscle, and heart. Cross-contamination was prevented by the use of different instruments and washing necropsy surfaces with 10% bleach and 70% ethanol. The organ was removed from the body and harvested using a 3 mm biopsy punch or cut to size using a scalpel. A total of 10 samples were collected from each organ: two were snap-frozen in liquid nitrogen, and eight were stored in RNAlater (Sigma–Aldrich) at 4°C for 24 h (300 μL of RNAlater for liver, muscle, and heart samples; 500 μl of RNAlater for adrenal gland and ovary samples).

The brain was removed from the skull using an oscillating saw and bone forceps. After extraction, the brain was weighed and placed ventrally in a custom made Plexiglass brain matrix (University of Massachusetts Machine Shop). Nine cuts were made to the brain to contain the striatum fully in four 6 mm blocks. The first cut was made posterior to the olfactory bulb attachment (approximately 18 mm from the beginning of the matrix), and the subsequent four cuts were made at 6 mm intervals (Supplementary Fig. S1A; Supplementary Data are available online at www.liebertpub.com/hum). The striatum was divided into four 6 mm blocks from posterior to anterior: 2p (posterior), 2m1 (medial 1), 2m2 (medial 2), and 2a (anterior). The striatal dissection was performed in the following order: 2p, 2m1, 2a. The striatum in the right (non-injected) hemisphere was dissected first in all blocks, and the scalpel blade was changed between hemispheres. The dissection was performed in a Petri dish on dry ice, and care was taken to remove as much white matter from the striatal tissue as possible. Once dissected out, the striatal pieces (caudate and putamen) were split in half, with the medial piece (closest to midline of block) stored in 1 mL of RNAlater at 4°C and the lateral piece snap-frozen in liquid nitrogen. The striatal dissection for the 6-month cohort in the CBA study was done in a manner to produce four striatal samples from both the caudate and putamen. The dorsal sections (both medial and lateral) were snap-frozen in liquid nitrogen, and the ventral sections (both medial and lateral) were stored in 1 mL of RNAlater at 4°C. RNAlater was removed after 24 h, and samples were stored at −80°C.

The 2m2 block was generously covered with OCT and frozen in a 2-methylbutane and dry ice bath. The remainder of the 2a, 2m1, and 2p block was frozen in the same manner. Ten cortex samples were taken from each block in a dorsal to ventral manner: two were snap-frozen in liquid nitrogen, and eight were stored in 1 mL of RNAlater at 4°C.

Sectioning of tissue for histological analysis

Prior to tissue sectioning for histological analysis, the striatum was isolated from the fixed and sucrose infiltrated brain, generously covered with OCT, and stored at −20°C for 24 h. Coronal sections, 40 μm thick, were cut with a sliding microtome (Reichert-Jung Tetrander sliding microtome) through the entire striatum. The sections were stored in 0.01% sodium azide in 1 × PBS at 4°C.

Vector cloning and rAAV9 production

For the first study, the test vector contained a U6 promoter driving an artificial miRNA based on the endogenous mir155 backbone (AAV9-U6-miR^HTT). The artificial miRNA targets human but not sheep huntingtin and has previously been described.¹⁷ A chimeric cytomegalovirus enhancer/chicken β-actin (CBA) promoter driving a chimeric intron was included to improve AAV packaging. The control vector (AAV9) contained only the empty CBA promoter and the intron. For the second study, the test vector contained the CMV enhancer and CBA promoter, the intron, and the miRNA-155 based artificial miRNA (AAV9-CBA-miR^HTT).

For packaging, the rAAV vector plasmid, a packaging plasmid, and an adenovirus helper plasmid were co-transfected into HEK 293 cells. The packaging plasmid expresses the regulatory and AAV9 capsid proteins leading to excision, replication, and packaging of the recombinant genome from the rAAV vector plasmid into AAV virions. The recombinant viruses were purified by standard CsCl gradient sedimentation and desalted by dialysis

Analysis of huntingtin mRNA levels

The RNA levels in the RNAlater preserved samples were analyzed using a branched DNA assay (bDNA).²⁴ Samples were processed according to the manufacturer's guidelines for preparation of tissue homogenates from tissues stored in RNAlater (Affymetrix eBioscience, Quantigene^® Sample Processing Kit). The homogenized samples were analyzed according to the manufacturer's guidelines for the bDNA assay (QuantiGene^® 2.0 Reagent System). The samples were analyzed with a probe to detect human huntingtin (human HD, SA-50339 from Quantigene^®), ovine huntingtin (sheep huntingtin, SF-10586 from QuantiGene^®), and ovine calnexin as a housekeeping gene (sheep calnexin, SF-10622 from QuantiGene^®). The assay results were measured with a Tecan Infinite M1000 PRO luminometer (integration time set at 200 ms).

Vector genome distribution

Genomic DNA was extracted from samples that had been snap-frozen in liquid nitrogen using the Gentra Puregene Tissue kit (Qiagen). The genomic DNA concentrations were measured using the NanoDrop ONE^c spectrophotometer. Droplet digital PCR (ddPCR; Bio-Rad Laboratories) was performed according to the manufacturer's recommendations using 50 ng of DNA as input and TaqMan assays detecting a region of the CB promoter common to all vectors and the HPRT reference gene. Results are expressed as vector genome per diploid genome (vg/dg).

Analysis of huntingtin protein levels: sample preparation

Small pieces of tissue were removed from frozen blocks and homogenized on ice in 200 μL of 10 mM HEPES pH 7.2, 250 mM of sucrose, and 1 mM of EDTA + protease inhibitor tablet (mini, complete, EDTA-free Roche #11836170001). Samples were sonicated for 10 s, and protein concentration was determined using the Bradford method (Bio-Rad #500-0006).

Mesoscale detection assay for mHTT protein

A 96-well QuicPlex standard plate (MSD) was coated with rabbit monoclonal anti-HTT proline 1220 region antibody (D7F7; Cell Signaling Technology; 1:250) in PBS overnight at 4°C. The plate was washed three times for 10 min with PBST (PBS +0.05% Tween20) and blocked with 3% bovine serum albumin (BSA) in PBS for 2 h at room temperature. After washing three times for 10 min with PBST, technical duplicates of samples with 20 μg of protein in 25 μL of homogenization buffer or blanks (homogenization buffer) were distributed into the plate and incubated overnight at 4°C on an orbital shaker. The plate was washed three times for 10 min in PBST and incubated in secondary/detection antibody mix as follows. For detection of mHTT, mouse monoclonal anti-polyQ antibody MW1 (DSHB) was mixed with anti-mouse SulfoTag detection antibody (MSD) at 1 μg/mL of each antibody in 1% BSA in PBS. Detection antibody mix (30 μL) was applied per well and incubated for 3 h at room temperature on an orbital shaker. The plate was washed three times for 10 min in PBST, and 150 μL of 2 × Read Buffer (MSD) was applied per well right before readout on a QuickPlex SQ120 (MSD).

Western blotting

Equal concentrations of protein (25 μg) were separated by SDS-PAGE on 3–8% Tris-Acetate gels (Life Technologies; #EA03785BOX) and transferred to nitrocellulose using TransBlot Turbo (Bio-Rad). Blots were blocked in 5% nonfat dry milk in Tris-buffered saline +0.1% Tween-20 (TBST) for 1 h and incubated overnight in primary antibody at 4°C diluted in blocking solution. Primary antibodies used were: anti-poly-Q (MW1, Coriell, 1:500 or 3B5H10, Sigma–Aldrich, 1:1,000), anti-huntingtin (MAB2166, EMD Millipore, 1:1,000 or Ab1, 1:1,000),²⁵ anti-DARPP32 (#ab40801, Abcam, 1:10,000), anti-actin (A4700, Sigma–Aldrich, 1:1,000), and anti-spectrin (MAB1622, EMD Milliopore, 1:4,000). Blots were washed in TBST, incubated in peroxidase labeled secondary antibodies (Jackson Immunoresearch), diluted 1:5,000 in blocking solution for 1 h at room temperature, washed in TBST, and incubated in SuperSignal West Pico Chemiluminescent Substrate (Pierce #34080). Images were obtained with a CCD imaging system (Alpha Innotech) and Hyperfilm ECL (GE Healthcare). Densitometry was performed on the digital images using ImageJ software (NIH). Statistical analysis was performed using unpaired t-tests, and results were expressed as mean value for the injected side.

Immunohistochemistry for DARPP32, NeuN, and Iba1

To quantify the DARPP32-positive cells, every 20th section was incubated for 3 min in 3% hydrogen peroxide in 1 × PBS, for 20 min in 0.5% Triton-X-100, and for 4 h in 1.5% normal goat serum (Vector Labs, S-1000) in 1 × PBS. Sections were incubated in anti-DARPP32 (AbCam, ab40801, 1:10,000 dilution) in 1.5% normal goat serum overnight at 4°C. Sections were then incubated in biotinylated goat, anti-rabbit IgG antibody (Vector Labs, AP-1000, 1:200 dilution) in 1 × PBS for 10 min. The sections were incubated with 2% Elite A and 2% Elite B reagent from the Vectastain Elite ABC Kit (Vector Labs, PK-6100) in 1 × PBS for 5 min. The Metal Enhanced DAB kit (Thermo Fisher Scientific, 34065) was used to visualize the DARPP32-positive cells. The sections were incubated in 1 × 3,3′-diaminobenzidine in stable peroxide buffer.

To quantify the NeuN-positive cells, every 20th section was incubated for 3 min in 3% hydrogen peroxide in 1 × PBS, 20 min in 0.5% Triton-X-100, and then overnight in 1.5% normal goat serum (Vector Labs, S-1000) in 1 × PBS at 4°C overnight. The sections were incubated in anti-NeuN (Chemicon, MAB377, 1:1,000 dilution) in 1.5% normal goat serum for 1 h at 4°C. The sections were then incubated for 40 min in a fluorescent AF594 goat, anti-mouse IgG (Thermo Fisher Scientific, A-11005, 1:2,000 dilution) to visualize the NeuN-positive cells.

To quantify the Iba1-positive cells, every 20th section was incubated for 1 h in a solution of 5% normal goat serum (Vector Labs, S-1000), 1% BSA (Sigma–Aldrich, A-3059), 0.2% Triton-X-100, and 0.03% hydrogen peroxide in 1 × PBS. The sections were incubated in anti-Iba1 (Wako Chemicals, 019-19741, 1:1,000 dilution) in 5% normal goat serum (Vector Labs, S-1000) and 1% BSA (Sigma–Aldrich, A-3059) at 4°C overnight. Sections were incubated in biotinylated goat, anti-rabbit IgG antibody (Vector Labs, AP-1000, 1:200 dilution) in 1 × PBS for 10 min. The sections were incubated with 2% Elite A and 2% Elite B reagent from the Vectastain Elite ABC Kit (Vector Labs, PK-6100) in 1 × PBS for 5 min. The Metal Enhanced DAB kit (Thermo Fisher Scientific, 34605) was used to visualize the reaction by incubating section in 1 × 3′,3-diaminobenzidine in stable peroxide buffer.

The quantification of DARPP32 and Iba1-positive cells in the left and right hemisphere of the brain was done by taking images (20 × for DARPP32 and 40 × for Iba1) with a Nikon Eclipse E600 microscope of each section. In order to capture images consistently between different sections, the first image was captured in the medial, dorsal edge of the striatum, and the stage was moved 0.5 cm toward the ventral edge. Once the ventral edge was reached, the stage was moved 0.5 cm laterally and 0.5 cm dorsally until 10 images were captured. Random numbers were assigned to each image to eliminate bias when quantifying cells. The cells were counted using ImageJ software (NIH).

The quantification of NeuN-positive cells was performed using the Nikon Eclipse E600 with a Chiu Technical Corporation Mercury 100-W lamp at 60 × . The stereological method used for capturing DARPP32 and Iba1 images was also used to quantify the NeuN-positive cells.

The area of the striatum, caudate, and putamen for each section was measured by manually circling the DARPP32 stained regions using ImageJ software (NIH) on the injected and non-injected sides of every 20th section through the striatum (29–35 sections per side per animal). The observer was blinded to the conditions. Total volume for each region was determined by multiplying the area by the section thickness (40 microns) by the number of sections between slides (20) and adding together for each animal. Statistical analysis was performed using Microsoft^® Excel, paired and unpaired t-tests, with n = 3 or 4 animals per group.

Results

Vector genome were widely distributed in the neostriatum

Silencing of an expanded mouse huntingtin in a knock-in model of HD¹⁴ and of the human mHTT transgene mRNA in a transgenic mouse model of HD have been previously reported.¹⁷ The striatum was unilaterally injected in two cohorts of HD sheep (study 1 and study 2). In study 1, the sheep were injected at 8–9 months of age with scAAV9-U6-miR^HTT (AAV9miR^HTT) or scAAV9-CBA-empty (AAV9), where a non-coding sequence is inserted between the promoter and the poly-A signal. In study 2, the sheep were injected at 14 months of age with scAAV9-CBA-miR^HTT or scAAV9-CBA-empty (AAV9). The brains were harvested 1 and 6 months after AAV9 administration. Vector genome copies were determined in a subset of regions (Supplementary Fig. S1) by ddPCR (Fig. 1). The genome copies were highest in the caudate and putamen on the injected compared to the non-injected side, and were at the highest levels in the scAAV9-U6-miR^HTT-treated group at 1 and 6 months post injection. Small amounts of vector genome were present in the cortex and liver but were undetectable in the adrenals (Fig. 1).

Figure 1.

Adeno-associated virus (AAV) vector genomes in control (AAV9) and treated (AAV9miR^HTT) animals. In brain samples (A), open circles represents the injected side, and filled squares represents the non-injected. Results from the liver and adrenals are reported in (B). Vector genomes were measured by digital droplet polymerase chain reaction (ddPCR) using genomic HPRT as the reference gene. The values are plotted on a log scale.

A single administration of scAAV9-miR^HTT reduced the human mutant huntingtin mRNA in caudate and putamen at 1 and 6 months post injection

HTT mRNA was measured in the anterior and medial striatum using a branched DNA assay that specifically recognizes human and not sheep HTT mRNA.²⁴ At 1 month post injection in the medial 1 block (Supplementary Fig. S1), scAAV9-U6-miR^HTT (study 1) reduced human HTT mRNA by >50% in both the caudate and putamen (Fig. 2). No significant silencing was detected in the anterior striatum (Fig. 2), which was farthest from the injection site (Supplementary Fig. S1). At 6 months post injection, mRNA silencing was pronounced in the caudate (Fig. 2).

Figure 2.

scAAV9-anti-HTT-6433 reduces human mutant huntingtin mRNA in the striatum. Data shown are the signal for HTT mRNA normalized to sheep calnexin (CANX). Asterisks indicate significant differences in means between treatment groups (AAV9 or AAV9miR^HTT) at p ≤ 0.03 with unpaired t-tests. The U6-promoter driven artificial miRNA significantly lowers human mutant HTT mRNA in the caudate and putamen at 1 month post injection and in the putamen at 6 months post injection. The CBA-promoter driven artificial miRNA lowers the HTT mRNA in the caudate, putamen, and anterior striatum at 1 month post injection and in the caudate and putamen at 6 months post injection. The medial region of the caudate, lateral putamen, and anterior striatum were examined in the analysis.

At 1 month post injection, AAV9-CBA-miR^HTT (study 2) silenced HTT mRNA by >50% in the caudate, 35% in the putamen, and 20% in the anterior striatum (Fig. 2). Six months post injection, 40–60% silencing was observed in the caudate, with up to 40% silencing in the putamen (Fig. 2). Interestingly, no silencing was observed in the lateral anterior striatum (area 1), but there was a non-significant trend toward lowering in the medial anterior striatum (area 2), indicating that there could be differences in silencing within a single block. This may explain some of the observed variability in the data. There was no significant silencing of the endogenous sheep HTT mRNA (Supplementary Fig. S2A).

Western blot assay and mesoscale discovery immunoassay show that AAV9-miR^HTT reduced human mutant huntingtin protein in the caudate and putamen

HTT protein was detected by electrochemiluminescence sandwich assay (MSD; Fig. 3)²⁶ and Western blot (Fig. 4) in the same sample preparations. Since antibodies that detect mHTT may have different sensitivities and selectivities, multiple antibodies were used: 3B5H10, MAB2166, and MW1 to detect human mHTT protein (Table 1). In the HD transgenic sheep, MAB2166 recognizes only human huntingtin and not sheep HTT.²² 3B5H10 preferentially detects mHTT (mutant HTT) compared to normal HTT, and MW1 preferentially recognizes the expanded polyglutamine region in HTT²⁷ and was used for Western blotting and for detection of mHTT in the mesoscale discovery (MSD) assay.

Figure 3.

Human mutant HTT levels detected by MSD assay at 1 and 6 months post injection in study 1 (U6 promoter) and study 2 (CBA promoter). Graph shows distribution of individual values and means (horizontal bars) for sheep treated with either AAV9 (control) or AAV9-miR^HTT. Results are shown for different striatal regions (caudate, putamen, and anterior striatum). Asterisks indicate significant difference on the injected side between AAV9 and AAV9-miR^HTT at p < 0.05 based on unpaired t-tests.

Figure 4.

Western blot assays show that AAV9-miR^HTT reduces the human mutant huntingtin protein in the striatum. (A) Sample Western blots of putamen from studies 1 and 2 show mutant HTT detected with antibody 3B5H10 and actin as loading control. (B) Graph shows distribution of individual values and mean (horizontal bar) for sheep treated with either AAV9 (control) or AAV9-miR^HTT. Shown are results for different striatal regions (caudate, putamen, and anterior striatum) in studies 1 and 2 and at 1 and 6 months post injection. In study 2, 6 months post injection, two areas (areas 1 and 2) were examined in each region. Asterisks indicate significant difference on the injected side between AAV9 and AAV9-miR^HTT at p < 0.05 based on unpaired t-tests.

Table 1.

Mean percent of mutant huntingtin protein lowering by Western blot and MSD assays in studies 1 and 2

	Study 1 (U6)				Study 2 (CBA)
Study # (promoter) Assay mutant htt antibody	Western blot (3B5H10)		MSD MW1		Western blot (3B5H10)		Western blot (MAB2166)		Western blot (MW1)		MSD MW1
Post-injection interval (months)	1	6	1	6	1	6	1	6	1	6	1	6
Caudate	78**	30	71**	74**	16	30, 43	61	46, 58*	50*	65,* 70	43*	40,* 47*
Putamen	61*	47*	73**	50*	40*	55,* 44	68*	65,* 67*	54	51,* 56*	42	63,** 56**
Anterior striatum	63**	22	49*	33	46	60,** 48	46	81,* 62*	–8	74,* 53*	22	62,** 67**

The human mutant huntingtin protein was measured by Western blot with anti-htt polyQ antibody 3B5H10 in study 1, and antibodies 3B5H10, MAB2166 (anti-HTT443-456), which does not recognize sheep HTT,²² and anti-polyQ monoclonal antibody MW1 in study 2. In the MSD assays, MW1 was used as the detection antibody. This table reports the mean percent mHTT lowering for the caudate, putamen, and anterior striatum. Percent lowering was calculated by dividing the average signal for the injected side in the AAV9miR^HTT-treated sheep by the average signal for the injected side in the AAV9 alone treated animals. Study 2, 6 months post injection, two areas per region were analyzed.

^*p < 0.05; ^**p < 0.001, unpaired t test; N = study 1, 1 month, 6 AAV9, 8 AAV9-miR^HTT; study 1, 6 months, 6 AAV9, 7 AAV9-miR^HTT; study 2, 1 month, 4 AAV9, 4 AAV9-miR^HTT; study 2, 6 months, 6 AAV9, 6 AAV9-miR^HTT.

Results with the MSD assay using MW1 for detection showed that scAAV9-U6-miR^HTT treatment (study 1) significantly lowered mHTT protein levels in the caudate, putamen, and anterior striatum at 1 and 6 months post treatment (Fig. 3). Western blotting using 3B5H10 indicated that 1 month after treatment, miR^HTT significantly reduced mHTT protein in the caudate, putamen, and anterior striatum when compared to injection with AAV9 lacking the miRNA (Fig. 4). At 6 months post treatment, there was a significant reduction of mHTT protein in the putamen. The data are summarized in Table 1.

In study 2, the MSD assay showed that AAV9-CBA- miR^HTT reduced mHTT protein by >40% in the caudate at 1 month post injection, and in the caudate, putamen, and anterior striatum 6 months after treatment (Fig. 3). Western blotting using 3B5H10 indicated significant lowering of mHTT in the putamen at 1 month and in the putamen area 1 and anterior striatum area 1 at 6 months (Fig. 4). Table 1 compares the mean percent lowering of mHTT detected by Western blot with three anti-mHTT antibodies (3B5H10, 2166, and MW1) and by MSD assay. All three antibodies in Western blot analysis detected significant mHTT lowering in multiple neostriatal regions in study 2 (49–81%). Results of mHTT lowering by MSD assay were consistent when two samples from the same striatal region were analyzed in study 2. A comparison of the results by Western blots and by MSD assays with MW1 in study 2 are also noteworthy. There was good agreement between these two different methods of mHTT detection in the magnitude of mHTT lowering.

By Western blot analysis, the cortex overlying the AAV9-miR^HTT-injected striatum did not show a decline in mHTT protein levels compared to the AAV9-injected cortex (Supplementary Fig. S3A), nor was there a reduction of mHTT protein in caudate or putamen on the contralateral, non-injected side (Supplementary Fig. S3B).

An important question was whether treatment with AAV9-miR^HTT against the human HD gene affected the levels of endogenous sheep HTT. The levels of the human transgene mHTT were compared directly to levels of endogenous sheep HTT by Western blot analysis by taking advantage of differences in migration of the two proteins on SDS-PAGE (Supplementary Fig. S2B). Western blot analysis with Ab1 antibody, which recognizes HTT1-17, showed that unlike human mHTT, the endogenous sheep HTT was not lowered by treatment with miR^HTT (Supplementary Fig. S2B).

DARPP32-labeled neurons and striatal volume were unaffected by miRNA treatment

DARPP32 immunoreactivity can be used as a marker for the presence of medium spiny neurons. To examine the safety of injection of the AAV vectors, immunohistochemistry for DARPP32 was performed (representative images Supplementary Figs. S4 and S5), and the number of DARPP32-positive cells were counted within every 20th section in the striatum. There was no significant difference between the number of cells in the AAV9-miR^HTT-treated and AAV9 control-treated groups (Supplementary Table S1). Although there was a trend toward loss of DARPP32-positive cells between 1 and 6 months in the first study, this difference did not reach statistical significance, nor were differences detected between wild-type and HD animals in study 2. The number of cells staining for NeuN, a marker of neuronal cells, also showed no significant difference between groups. Striatal volumes were determined using cross-sectional area measurements of striatum in DARPP32-labeled sections and were found to be unchanged compared to controls after miRNA treatment (Supplementary Table S2).

A transient increase in activated microglia occurred after direct injection with AAV9-miR^HTT and AAV9 control

Immunohistochemical localization of Iba1 a protein that is localized to microglia and upregulated upon their activation, was examined (examples are shown in Supplementary Figs. S4 and S5).²⁸ Labeled cells were identified based on morphology as resting or activated microglia (Supplementary Table S3). Injection of scAAV9-U6-miR^HTT or the corresponding control vector increased the number of activated microglia on the injected side at 1 month post treatment. On the non-injected side, activated cells comprise <1% of the total number of Iba1-positive cells, whereas on the injected side, they are 49–58% of the total. Six months after treatment, the injected and non-injected sides were indistinguishable. In the second study, the microglial response was only examined at the study end point (6 months). At this time, there was a trend toward higher Iba1 positive activated microglia in the AAV9 (control) injected HD sheep when compared to the AAV9 (control) injected wild-type animals, but this did not reach statistical significance. The possibility cannot be excluded that HD and control sheep have a small difference in microglia response at this age, which the current study was not powered to detect. Additional experiments are required to rule out this hypothesis. The number of Iba1-positive activated microglia is small—between 2% and 12% of the total number of Iba1-positive cells—and there was no significant difference between any of the groups. The findings suggest that the transitory increase in activated microglia is independent of AAV cargo and can occur with any vector or with surgery alone.

AAV9-miR^HTT treatment did not affect blood counts, electrolytes, or liver and kidney function

Blood samples were taken at four times: baseline (pretreatment), 28 (or 30) days, 90 days, and 180 days post treatment. Complete blood count, electrolytes, liver and kidney function tests were measured (Supplementary Table S4). No meaningful changes in any of these measurements were found between AAV9-miR^HTT-injected sheep and controls. In addition, there were no differences in weight or normal movement between groups.

Discussion

This study addressed important concerns for therapy of HD. First, a single treatment with the scAAV9-miR^HTT effectively lowered mutant human huntingtin in the neostriatum of an animal with a large brain. Second, lowering huntingtin mRNA and protein has been reported in rodent models,^3,4,6,11 but only a handful of studies focused on the full-length mutant huntingtin.^11,26 The transgenic sheep model used here is the only large animal with a full-length human mHTT. A critical advantage of this HD model is that additional downstream exons, compared to N-terminal models of disease, are available for miR targeting. Third, mutant human huntingtin reduction could be achieved and sustained with two different promoters, confirming the robustness of the miR^HTT activity.

Lowering mutant human huntingtin protein was the goal of the miRNA-mediated silencing. This goal was achieved. Mutant human huntingtin mRNA was reduced, but somewhat less consistently than huntingtin protein lowering, which was detected by Western blot with multiple antibodies and by MSD assay. Several speculations are offered that might account for the more consistent huntingtin protein reduction compared to mRNA. Huntingtin mRNA may have a compartmental distribution in neurons, exhibiting cytoplasmic and nuclear localization or differential distribution to cell bodies, axons or dendrites. These different pools of mRNA may respond to miRNA silencing differently. Furthermore, the cumulative effect of small differences in injection and sampling site may increase the variability of measurements. Samples for mRNA, protein, and DNA vector genome analysis were taken from adjacent sites, but the block size was reasonably large. It was shown that using this method, there can be differences in observed silencing within a single sample block. In future studies, freezing samples in smaller, single-use blocks and measuring protein and RNA from the same sample would be recommended. Finally, mRNA might be less stable and the methods used to measure it less reproducible than protein. These observations highlight the importance of measuring both mRNA and protein, if possible, by more than a single method. Measurements of protein and mRNA represent bulk amounts in brain regions, rather than in individual cells. Huntingtin mRNA estimations in neurons vary,¹⁴ and huntingtin protein variance has not been established. The current study presents a composite of mRNA silencing. Using these measures, the extent of mHtt reduction cannot be determined on a cellular or subcellular level.

Within the brain, the vector was mainly concentrated near the injection site between the caudate and putamen, indicating that the local spread by convection enhanced delivery surmounted anatomic restraints that might be imposed by the internal capsule. Much lower steady-state levels of AAV9 vector genome were detected in the ipsilateral cortex. It will be interesting in a follow-up study to detect vector genome in neurons collected by laser capture. It is speculated that a higher amount of AAV9 volume or titer might have huntingtin-lowering properties in cortex. The small amount of AAV9 vector genome in the contralateral neostriatum was not associated with mutant huntingtin protein change. Outside of the brain, only a little AAV9 was detected in the liver.

The transgenic HD sheep offer a value-added to the study of mutant huntingtin protein lowering. Further characterization of the model, including a detailed analysis of the number of DARPP32-positive neurons and microglia activation, may reveal additional phenotypes that will further enhance the value of the model. However, concerns about huntingtin protein lowering remain. In addition to the human huntingtin transgene, the model has two sheep huntingtin alleles. This approach successfully reduced mutant human huntingtin and did not affect the sheep huntingtin. Most current and broadly applicable Htt targeting therapies affect both huntingtin alleles. Allele-specific artificial miRNA would require a complex strategy involving generation of multiple AAV-miRs targeted to single nucleotide polymorphism heterozygosities, and would apply to only a subset of patients. A broad artificial miR would reduce the toxic moiety and could be beneficial for all patients, but it is not yet clear if long-term elimination of huntingtin is safe.^29,30 A future study in nonhuman primates with the same miR^HTT, which would silence both alleles, would help to clarify this issue. This study focused on young sheep that lack huntingtin inclusions or neuronal loss. For the most part, HD is a disease of adults; study of older sheep would have virtue, especially if the HD transgenic sheep were to develop signs of disease or neuropathology.

Acknowledgments

We wish to acknowledge the staff at the Preclinical, Imaging, and Research Laboratories, South Australian Health and Medical Research Institute, for their help with the study. The authors acknowledge the facilities and the scientific and technical assistance of the National Imaging Facility at the SAHMRI Large Animal Research and Imaging Facility. We also thank Eric Mick for help and consultation on the study analysis and members of the Khvorova lab for suggestions on sample collection, study design, and the branched DNA assay. This work was supported by CHDI foundation agreement A-5038.

Author Disclosure

C.M., E.P., N.A., and G.G. have filed patents pertaining to this work. All other authors have no competing financial interests.