Huntingtin silencing delays onset and slows progression of Huntington’s disease: a biomarker study

Abstract

Huntington’s disease is a dominantly inherited, fatal neurodegenerative disorder caused by a CAG expansion in the huntingtin (HTT) gene, coding for pathological mutant HTT protein (mHTT). Because of its gain-of-function mechanism and monogenic aetiology, strategies to lower HTT are being actively investigated as disease-modifying therapies. Most approaches are currently targeted at the manifest stage, where clinical outcomes are used to evaluate the effectiveness of therapy. However, as almost 50% of striatal volume has been lost at the time of onset of clinical manifest, it would be preferable to begin therapy in the premanifest period.

An unmet challenge is how to evaluate therapeutic efficacy before the presence of clinical symptoms as outcome measures. To address this, we aim to develop non-invasive sensitive biomarkers that provide insight into therapeutic efficacy in the premanifest stage of Huntington’s disease. In this study, we mapped the temporal trajectories of arteriolar cerebral blood volumes (CBVa) using inflow-based vascular-space-occupancy (iVASO) MRI in the heterozygous zQ175 mice, a full-length mHTT expressing and slowly progressing model with a premanifest period as in human Huntington’s disease.

Significantly elevated CBVa was evident in premanifest zQ175 mice prior to motor deficits and striatal atrophy, recapitulating altered CBVa in human premanifest Huntington’s disease. CRISPR/Cas9-mediated non-allele-specific HTT silencing in striatal neurons restored altered CBVa in premanifest zQ175 mice, delayed onset of striatal atrophy, and slowed the progression of motor phenotype and brain pathology.

This study—for the first time—shows that a non-invasive functional MRI measure detects therapeutic efficacy in the premanifest stage and demonstrates long-term benefits of a non-allele-selective HTT silencing treatment introduced in the premanifest Huntington’s disease.

neurovascular biomarker, Huntington’s disease, huntingtin silencing, functional MRI

Introduction

Huntington’s disease preferentially involves the basal ganglia—especially the striatum—but also affects other brain regions and has no cure or disease-modifying treatment.^1,2 While cognitive and emotional changes are increasingly recognized as causes of disability, motor changes remain key and easily quantifiable aspects of clinical onset and progression.³ Huntington’s disease is caused by a CAG repeat expansion in the huntingtin gene (HTT)² and, as a monogenic disorder, can serve as a model for studying other neurodegenerative diseases. Because of the gain-of-function mechanism of mutant HTT (mHTT), lowering HTT levels is rapidly emerging as a potential first disease-modifying therapy.^4,5

Most therapeutic approaches target the manifest stage when clinical outcomes can be used to evaluate the effectiveness of therapy. However, structural MRI studies have shown that almost 50% of striatal volume has already been lost by the onset of manifest Huntington’s disease.^6-10 Thus, it would be preferable to begin treatment in the premanifest period, and equally important to have biomarkers indicative of treatment response prior to the development of clinical symptoms. Striatal volume is a potentially valuable biomarker, but it changes slowly, and the extent to which it will demonstrate a response to treatment is unknown. CSF and blood biomarkers may also be useful^11–15; however, their response to treatment remains less well understood. Therefore, we have been developing sensitive and non-invasive biomarkers, such as functional neuroimaging measures.¹⁶

The clinical diagnosis of Huntington’s disease is based on the presence of movement disorders. However, functional changes in the brain can precede motor onset by many years.^17–22 CAG repeat expansion length can be used after predictive genetic testing during the premanifest period to approximate the age of motor onset (clinical manifest). Given the availability of predictive genetic testing, mutant HTT carriers can be identified decades before clinical manifestation, providing a unique opportunity to identify premanifest biomarkers, which can facilitate the development of disease-modifying therapies. Moreover, sensitive biomarkers are crucial for determining the optimal time to begin treatment.^1,18,23–25

The brain is a high energy-demanding organ despite its limited intrinsic energy storage.²⁶ The balance between substrate delivery from blood flow and brain energy demand is precisely regulated in a healthy condition, and cerebral blood flow/volume are strongly coupled with brain metabolism.²⁷ A recent study indicated that impaired mitochondrial oxidative phosphorylation in the most vulnerable striatal neurons occurs long before manifest stage.²⁸ As arterioles are actively regulated blood vessels, arteriolar cerebral blood volume (CBVa) may be sensitive to premanifest alterations in the Huntington’s disease brain.^29–31 We and others have reported elevated CBVa in human premanifest Huntington’s disease.^16,27

HTT-lowering approaches are being actively pursued in clinical trials. These include antisense oligonucleotides and AAV-mediated delivery of RNA interference (RNAi) reagents.⁴ A more recent gene-editing approach using RNA-guided clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas) systems, which induce DNA double-strand breaks and permanently disable mutant gene function in target cells, has been applied in preclinical studies of Huntington’s disease.^32,33 It remains an unmet challenge to reliably evaluate the effectiveness of HTT-lowering treatments in clinical trials, particularly in premanifest Huntington’s disease.

Mouse models have helped elucidate disease pathogenesis, experimental therapeutics, as well as biomarker development in Huntington’s disease.^34,35 While prior focus has been on neurons in Huntington’s disease, it is becoming increasingly apparent that other components such as blood vessels may contribute to or at least be reflective of pathogenesis.^36–38 The heterozygous full-length HTT knock-in zQ175 mouse model^39,40 has the advantage of a relatively slow development of pathology with a period resembling the human premanifest stage.

In this study, we used a CRISPR/Cas9 system to lower mutant (and wild-type) HTT expression in striatal neurons of heterozygous zQ175 mice. We then used CBVa measured by inflow-based vascular-space-occupancy (iVASO) MRI in conjunction with behavioural assessments and structural MRI to evaluate the effectiveness of HTT silencing introduced at the premanifest stage longitudinally. We demonstrate that lowering HTT in striatal neurons restores CBVa in premanifest zQ175 mice, delays onset, and slows the progression of motor phenotype and striatal atrophy. Our data suggest that CBVa measure may be a promising non-invasive biomarker for premanifest clinical trials, and that introducing treatment at the premanifest stage has long-lasting benefits in Huntington’s disease.

Materials and methods

Animals

Heterozygous zQ175 mice and wild-type littermates (both genders) were used. The zQ175 breeder mice were obtained from the Jackson Lab. Genotyping and CAG repeat size were determined by PCR of tail snips at Laragen Inc. The CAG repeat length was 220 ± 3 in the zQ175 mice used in the study. All mice were housed under specific pathogen-free conditions with a reversed 12-h light/dark cycle maintained at 23°C and provided with food and water ad libitum. All behavioural tests and longitudinal MRI measures were done in the mouse dark phase (active). The study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and approved by Animal Care and Use Committee at the Johns Hopkins University. A rigorous study design was implemented. Mice were randomized into groups to avoid bias. Balanced gender ratio was considered in all experiments. Data were collected using animal ID and analysed by investigators who were blinded to genotype and treatment. Data from different genders were analysed grouped when there was no gender-dependent difference in the outcome measures (Supplementary Table 1).

MRI acquisition

MRI scans were performed on a horizontal bore 11.7 T Bruker Biospec system (Bruker) equipped with a physiological monitoring system. The MRI scanner was tuned and calibrated before each experiment. A 72-mm quadrature volume resonator was used as a transmitter, and the receiver was a four-element (2 × 2) phased-array coil. Anatomical images covering the entire brain were acquired using a multi-slice Rapid Acquisition using Relaxation Enhancement (RARE) sequence: resolution = 0.10 × 0.10 × 0.50 mm³, echo time/repetition time =30/3000 ms, and RARE factor = 8. A multi-slice iVASO sequence optimized for typical blood flow and vascular transit times in mice and relaxation times at 11.7 T was performed for CBVa mapping: repetition time/inversion time = 1924/800, 1636/700, 1365/600, 1109/500, 867/400, 750/350, 636/300, 525/250, 415/200 ms, echo time = 4.6 ms, resolution = 0.15 × 0.15 × 0.80 mm³, and RARE factor = 16. All mice were anaesthetized with an induction of 2% isoflurane in medical air, and then 1.0–1.5% isoflurane during the MRI scan. The mouse head position was fixed with a bite bar and two ear pins. During MRI scans, the animal was placed on a water-heated animal bed equipped with respiratory controls. The animal’s respiration rate was constantly monitored using an animal monitoring system (SAII). Mice were ventilated to maintain stable physiological conditions (respiratory rate at 60–80 breaths/min). The anaesthesia level was carefully controlled and the physiological conditions including heart rate and respiration were monitored during MRI scans for each animal. Animals were ventilated to maintain stable physiological conditions (respiratory rate 70 ± 10 breaths/min) during MRI scans.

iVASO MRI image analysis

Statistical parametric mapping (SPM) (Version 8, Wellcome Trust Centre for Neuroimaging, London, UK; http://www.fil.ion.ucl.ac.uk/spm/) and in-house programs coded in MATLAB (MathWorks, Natick, MA, USA) were used. Motion correction was performed for all iVASO images, and the difference in iVASO signal was calculated using the surround subtraction method.⁴¹ Note that since CBVa in iVASO was calculated by comparing a blood nulled scan with a subsequent (a few seconds apart) control scan where blood signal is not affected in each mouse, the difference signal should be minimally affected by potential variations from the MRI scanner. Whole-brain CBVa maps were calculated using the iVASO theory.⁴² Cortical thicknesses obtained from structural MRI scan were used to correct partial volume effects on the iVASO signal, as described in our previous work.^16,43 Therefore, the final CBVa values used in subsequent analyses have been adjusted according to regional atrophy measured in each mouse. The iVASO images were co-registered with structural images and were normalized to the Franklin and Paxinos’s mouse brain atlas⁴⁴ using SPM routines from which specific regions in the brain are identified. Two regions of interest were delineated in each scan: (i) motor cortex; and (ii) striatum. Average CBVa values were obtained in each region of interest. Two investigators (H.L. and C.Z.) manually delineated regions of interest of the striatum and motor cortex. Intra-rater and inter-rater variabilities were measured by repeating manual delineation multiple times (three to four times) until the reproducibility reached a plateau. The final intra- and inter-rater variabilities were measured using another dataset three times. Two double-blind investigators performed an image analysis for the same samples: the inter-rater variability was 2.1%, and the intra-rater variability was 1.5%, indicating our analysis method is reproducible and reliable. The two operators (H.L. and C.Z.) processed data independently and we adopted the average of the two measurements.

Structural MRI image analysis

Images were analysed as established in our laboratory and described previously.^45–47 Skull-stripped, rigidly aligned images were analysed using Landmarker software (www.mristudio.org). The intensity-normalized images were submitted by Landmarker software to a Linux cluster, which runs large deformation diffeomorphic metric mapping (LDDMM). The transformations were then used for quantitative measurement of changes in local tissue volume among different mouse brains, by computing the Jacobian values of the transformations generated by LDDMM. More details of LDDMM can be found in the Supplementary material.

Behavioural tests

Balance beam

Testing was conducted on an 80-cm long and 5-mm wide square-shaped balance beam that was mounted on supports of 50 cm in height. A bright light illuminated the start platform, and a darkened enclosed 1728 cm³ escape box (12 × 12 × 12 cm³) was situated at the end of the beam. Mice were trained to walk across the beam twice at least 1 h prior to testing. The time for each mouse to traverse the balance beam was recorded with a 125-s maximum cut-off, and falls were scored as 125 s.

Tapered beam

Testing was conducted on a beam that is 1 m in length tapering from 3.5 cm to 0.5 cm with underhanging ledges 1.0 cm in width on either side. The beam was placed at a 30° angle of incline, with the narrowest end at the highest point. The time for each mouse to traverse the tapered beam was recorded with a 125-s maximum cut-off, and falls were scored as 125 s.

Locomotor activity

The open field locomotor activity test was performed during the dark phase of the diurnal cycle under red light conditions. Locomotor activity was measured by an automated Open Field Activity System and the data were analysed by Activity Monitor software (Columbus Instrument Inc., OH).

AAV reagents and stereotaxic injection

CRISPR/Cas9-related viral vectors (PX551, PX552) were kindly provided by Dr Xiaojiang Li’s lab (Emory University, then). The original viral vector was obtained from Addgene (plasmids #60957 and 60958). AAV₉-HTT-gRNAs were generated by inserting guide RNAs (gRNAs) into PX552 via SapI restriction sites. Sequences for the gRNAs are as follows: T1: GGCCTTCATCAGCTTTTCCAggg, T3: GGCTGAGGAAGCTGAGGAGGcgg, and control gRNA: ACCGGAAGAGCGACCTCTTCT (PAM sequence is shown in lower case). AAV₉-Mecp2-Cas9 vector was generated in PX551 with CMV promoter (658 bp) using XbaI and AgeI restriction sites. AAVs were packaged by the Viral Vector Core at Emory University (NINDS-supported service centre).

The mice were anaesthetized with 1.5% isoflurane inhalation and stabilized in a stereotaxic instrument (David Kopf Instruments). Mice were injected into the striatum using the stereotaxic coordinates: 0.62 mm rostral to bregma, ±1.75 mm lateral to midline and 3.5 mm ventral to the skull surface. AAV₉-gRNA and AAV₉-spCas9 were mixed at a ratio of 1:3, and 2 μl of the mixed viruses (1.0 × 10¹³ particles/μl) were injected into the striatum using a Hamilton syringe infusion pump (World Precision Instruments) at a perfusion speed of 0.2 μl/min.

Immunohistochemistry and quantification

Mice were anaesthetized and perfused transcardially with PBS followed by 4% paraformaldehyde. Brains were post-fixed overnight followed by immersion in 30% sucrose for 24 h. Coronal brain sections (40 μm) were immunostained with following antibodies, EM48 (MAB5347, 1:200, Merck Millipore), NeuN (26975-I-AP, 1:200, Proteintech), Collagen IV (2150–1470, 1:100, Bio-Rad), Acta-2 (C6198, 1:200, Millipore Sigma), and GLUT1 (SPM498, 1:200, Thermo Fisher). Briefly, the sections were washed three times with PBS, then permeabilized by incubating with 0.3% Triton^TM X-100 for 5 min, followed by incubation with blocking solution containing 5% donkey serum or 3% goat serum and 0.3% Triton^TM X-100 for 1 h. The sections were then incubated with primary antibody at 4°C overnight. After three washings with PBS, the sections were incubated with fluorescently-labelled secondary antibody for 2 h at room temperature. Sections were mounted onto Superfrost® slides (Fisher Scientific) dried and then covered with anti-fade mounting solution. Fluorescence images were acquired with Keyence BZ-X700 All-in-One florescence microscope.

For image analysis, the samples were coded with an ID and images were analysed by investigators who were blinded to the genotypes and treatment. Analysis results were then calculated statistically and decoded by different investigators at the end. The results from three microscopy fields per slide and three sections per mouse were calculated for each mouse brain. GLUT1 green fluorescence intensity was quantified using Fiji-ImageJ (NIH) in each microcopy field.

Quantification of mHTT aggregates

High resolution fluorescent images were obtained using a Zeiss LSM 900 confocal microscope. EM48 positive puncta quantities were determined using the Fiji-ImageJ’s analyse particles plugin function. The numbers of EM48 positive puncta per square millimetre were determined at a constant threshold for each stain using ×20 confocal images for quantifications.

Blood vessel morphological analysis

MATLAB Image Processing ToolboxTM was used to provide a comprehensive image processing and morphological analysis. To reduce processing time and enhance signal-to-noise ratio, we converted the RGB colour image to greyscale and used the imaging adjusting tool to increase contrast in low-contrast areas and sharpen differences between black and white, which can help identify blood vessels. To reduce noise in the background and normalize vessel intensities, a 2D Gaussian filter was applied.⁴⁸ To quantify vessel structure, Bradley’s method was used to create a binarized image.⁴⁹ A skeletonized image was generated by using a morphological method described previously,^50,51 and the vascular diameters were calculated by executing the Euclidean distance transform.⁵¹ Vessel segment was defined as the fragment between two branch points.⁵² Blood vessel density was quantified as the percentage of collagen IV and Acta2 positive immunofluorescence within microscopic field. For the entire process, bootstrapping sampling method was applied.⁵³ This program randomly samples a 0.1 mm² area 1500 times on the images and provides a mean value for the designated area. All analysis was done in a blinded manner, the slides were coded with ID, and the investigator doing imaging analysis did not know the genotype information. Imaging results are calculated statistically and decoded at the end. Three sections were quantified and averaged to get the value for each individual mouse. Four mice per group and all data points are shown in the figures.

Western blotting

Brain samples were homogenized in a buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.1% (w/v) SDS, 1.0% NP-40, 0.5% sodium deoxycholate, and 1% (v/v) protease inhibitors. Proteins (30 μg) were separated in a 4–20% gradient gel and transferred to a nitrocellulose membrane. The membrane was blotted with the following primary antibodies: MAB 2166 (anti-HTT, 1:1000, Millipore Sigma), MCA2050 (mouse anti HTT, 1:1000, Bio-Rad), MCA2051 (mouse anti HTT, 1:1000, Bio-Rad), MW1 (anti-poly-Q, 1:1000, Millipore Sigma), and β-actin (Sigma, mouse monoclonal antibody, 1:5000). After incubation with HRP-conjugated secondary antibodies, bound antibodies were visualized by chemiluminescence.

Statistics

Data are expressed as the mean ± standard error of the mean (SEM) unless otherwise noted. Statistical analysis was performed with SPSS using two-tailed Student’s t-test, one-way ANOVA, two-way ANOVA, with Bonferroni post hoc tests. The P-values < 0.05 were considered as statistically significant. The n numbers are reported in the figure legends.

Data availability

All the data supporting the findings of this study are available upon reasonable request.

Results

Capturing abnormal arteriolar cerebral blood volume trajectories in zQ175 mice

We previously observed elevated CBVa in human premanifest Huntington’s disease brains.¹⁶ In the present study, we mapped temporal trajectories of CBVa and conducted motor phenotype and striatal volume assessments in the heterozygous zQ175 mice. Longitudinal characterization was conducted at 3, 6 and 9 months of age, representing prior to onset (premanifest, 3 months), onset (6 months), and post-onset (manifest, 9 months) of striatal atrophy and motor phenotypes (Fig. 1A). Mouse brains were imaged using a 3D acquisition protocol, which allowed continuous scanning of 0.1-mm thick slices of the whole brain to minimize partial volume sampling effects in the dataset. A region of interest-based analysis was performed in the basal ganglion system with focuses on the striatum and motor cortex. We found that CBVa was significantly elevated in the striatum (Fig. 1B and C, P = 0.034) and motor cortex (Fig. 1D and E, P = 0.041) of zQ175 mice at 3 months compared to wild-type littermate controls. No significant differences were observed in the striatal volume, motor function, locomotor activity, and body weight at this age between zQ175 mice and controls (Supplementary Fig. 1A–E). These data indicate that altered CBVa occurs prior to motor symptoms and striatal atrophy in heterozygous zQ175 mice, and that the CBVa changes in this mouse model are reminiscent of those observed in the premanifest human Huntington’s disease brain.

Figure 1

Abnormal trajectories of arteriolar cerebral blood volume in zQ175 mice. (A) Timeline for in vivo experiments. Onset indicates the age when striatal atrophy is detectable in zQ175 mice. (B) Representative CBVa maps calculated from inflow-based vascular-space-occupancy (iVASO) images in mouse brains from indicated genotypes and ages. Top: The raw images, and the red regions of interest indicate the quantified brain region-striatum. Bottom: The CBVa maps for zQ175 mice and wild-type (WT) littermates. The scale bars are shown on the right and warmer colours represent higher CBVa values. Scale bar = 3 mm. (C) Longitudinal CBVa changes in the striatum of zQ175 mice and wild-type littermates. (D) Representative CBVa maps calculated from iVASO images in mouse brains from indicated genotypes and ages. Top: Raw images, and the red regions of interest indicate the quantified brain region-motor cortex. Bottom: CBVa maps for zQ175 mice and wild-type (WT) littermates at the indicated ages. The scale bars are shown on the right and warmer colours represent higher CBVa values. Scale bar = 3 mm. (E) Longitudinal CBVa changes in the motor cortex of zQ175 mice and wild-type littermates. Longitudinal data are mean ± SEM, n = 8 (four males and four females/genotype). *P < 0.05 versus the values of the wild-type group at matched ages by Student’s t-test.

We then examined the trajectories of CBVa changes longitudinally in the zQ175 mice. CBVa values from multiple regions of interest were included as the dependent variables, genotype was included as the independent variable, and Genotype × Time was included as the interaction variable. Significant Genotype × Time interaction was observed between zQ175 mice and controls. We observed that control mice could maintain a stable level of CBVa in basal ganglion regions over the course of 6 months (Fig. 1C and E). In contrast, zQ175 mice exhibited progressively declining CBVa in the striatum [n = 10, Age × Genotype F(2,28) = 6.736, P < 0.01] and motor cortex [Age × Genotype F(2,42) = 7.046, P < 0.01] over the course of disease progression. Although striatal volumes in this model have been reported before,^39,40 our measurements were made in the same cohort in which CBVa was measured. This provides more detailed information on possible age-dependent structural changes and more importantly, helps identify the potential influence of structural changes on CBVa alterations. Our data indicate that zQ175 mice experienced striatal atrophy onset at 6 months (Supplementary Fig. 1A, P = 0.012), which progressively worsened at 9 months (P = 0.001). The zQ175 mice also displayed motor deficits on the tapered beam (Supplementary Fig. 1B, P = 0.046) and 5 mm balance beam (Supplementary Fig. 1C, P = 0.018), and decreased locomotor activity in the open field apparatus (Supplementary Fig. 1D, P = 0.043) at 9 months of age. zQ175 mice also had slower body weight gain than control mice (extended data Fig. 1E) [Age × Genotype F(6,102) = 8.311, P < 0.01]. Taken together, the longitudinal data support that altered CBVa occurs at the premanifest stage, which progressively declines with the onset and progression of disease in the zQ175 model.

Morphological analysis of the cerebral vasculature in zQ175 mice

Given that mHTT may affect the physiology of cells comprising the cerebral vasculature, we analysed the morphology and structures of arterioles and cerebral vasculature in zQ175 mice at the premanifest age (3 months) and manifest age (9 months) when significant changes in CBVa were detected between zQ175 mice and controls. Acta2 labels arterioles whereas collagen IV labels all blood vessels including arterioles, capillaries and venules.⁵⁴ At the premanifest stage, there were no significant differences in the vessel density and mean diameter of Acta2+ arterioles and collagen IV+ blood vessels in the striatum (Fig. 2A–E) and motor cortex (Supplementary Fig. 2A–E). To compare variability in the diameter of collagen IV+ vessels, we used MATLAB software tools to generate skeletonized images^50,51 and calculated vascular diameters by executing the Euclidean distance transform.⁵¹ We located vessel branch points using the skeletonized images and defined a vessel segment as the fragment between two branch points.⁵² Further analysis of blood vessel morphology excluded capillaries (<2 µm) and focused on smaller vessel diameter ranges at 2–10 µm and 10–20 µm. We did not detect differences in the numbers of vessel segments in the striatum (Fig. 2F and G) and motor cortex (Supplementary Fig. 1F and G) between zQ175 and age-matched controls. These data indicate that elevated CBVa in the premanifest zQ175 mice is not due to permanent morphological changes in cerebral blood vessel densities or diameters, and that it may reflect a transient neurovascular functional response to altered brain energy demand, such as enhanced blood vessel dilation.

Figure 2

Morphological analysis of cerebral vasculature in the striatum of zQ175 mice. (A) Representative images of immunostaining signals of Acta2 (red, arterioles) and collagen IV (green, all blood vessels) in the mouse striatum from indicated genotypes and ages. Three months (3 M) represents the premanifest stage in zQ175 mice; 9 months (9 M) represents the manifest age of zQ175 mice. Scale bar = 40 µm. (B and C) Quantitative analysis of Acta 2+ arteriolar density (B) and mean diameter (C) in the striatum of zQ175 mice and wild-type (WT) controls at 3 and 9 months of age. (D and E) Quantitative analysis of collagen IV+ blood vessel density (D) and mean diameter (E) at indicated ages. (F) The representative skeletonized image of collagen IV+ blood vessels at the indicated diameters in the zQ175 and wild-type mice at 3 and 9 months of age. (G) Quantitative analysis of the number of segments per 0.1 mm² in the collagen IV+ blood vessels with indicated diameters. (H and I) Representative images of Glut1 immunostaining in the mouse striatum from indicated genotypes and ages (H) and quantitative data of Glut1 immunofluorescent intensity (I). Scale bar = 40 µm. All data are mean ± SEM, n = 4 (two males and two females/genotype). *P < 0.05 versus the values of wild-type group at the corresponding ages by Student’s t-tests.

We next analysed the morphology of arterioles and cerebral vasculature in manifest zQ175 mice (9 months old) when CBVa significantly declined. At this age, zQ175 mice exhibited increased density (Fig. 2B, P = 0.046) and decreased diameters (Fig. 2C, P = 0.043) in Acta2+ arterioles in the striatum. Although there was still an increasing trend, no significant differences in arteriolar density in the motor cortex of zQ175 mice were observed (Supplementary Fig. 2B, P = 0.27). Notably, significantly reduced arteriolar diameter was observed in the motor cortex of zQ175 mice (Supplementary Fig. 2C, P = 0.004). Moreover, we observed significantly increased density (Fig. 2D, P = 0.009) and reduced vessel diameter (Fig. 2E, P = 0.005) in all collagen IV+ striatum vasculatures of 9-month-old zQ175 mice. The effects of increased density and reduced vessel diameter were smaller in collagen IV+ vessels compared to Acta2+ arterioles in the striatum of 9-month-old zQ175 mice. Further analysis by separating collagen IV+ blood vessels into different groups based on diameter as above for 3-month-old mice showed that the number of segments in smaller blood vessels (diameter at 2–10 µm) was significantly higher (Fig. 2F and G, P = 0.004) in the striatum of zQ175 mice, while no significant changes in the numbers of larger blood vessel segments (diameter at 10–20 µm) (Fig. 2F and G) were observed. Similar changes in the morphology of collagen IV+ blood vessels were also detected in the motor cortex of zQ175 mice (Supplementary Fig. 2F and G). These results suggest that lower CBVa in manifest zQ175 mice may be at least partially due to these permanent morphology changes of blood vessels, particularly reduced arteriolar diameter. These structural changes in the collagen IV+ cerebral blood vessels of manifest zQ175 mice were consistent with reported vascular alterations in other mouse models of Huntington’s disease.³⁷ Because the cerebral vascular system provides essential support for effective brain functioning,⁵⁵ morphological abnormalities in the cerebral vasculature likely further compromise neuronal integrity and function, thereby contributing to pathology and disease progression.

CBVa is a haemodynamic measure that may be an indirect indicator of neuronal metabolism and function,⁵⁶ as altered neuronal activity and/or glucose utilization can result in differences in CBVa readouts. It is worth noting that blood vessel diameters measured from histology may not reflect the blood vessel diameters measured during in vivo MRI scans when the animals are alive. Because we did not find histological changes in the density and diameter of arterioles and cerebral blood vessels in premanifest zQ175 mice (3 months), we hypothesize that elevated CBVa measured with iVASO MRI at the premanifest stage may represent a compensatory vascular dilation in response to impaired neuronal metabolism. If this is the case, glucose transporters may also be upregulated to meet the neuronal energy demands in the premanifest stage. We thus conducted immunofluorescent staining to examine the major glucose transporter GLUT1 in endothelial cells. Significantly increased GLUT1 immunofluorescent intensity was observed in the striatum (Fig. 2H and I, P = 0.033) and motor cortex (Supplementary Fig. 2H and I, P = 0.020) of premanifest zQ175 mice (3 months), suggesting upregulated glucose transport in the zQ175 mouse brain. We next asked whether such upregulation of the glucose transporter will persist to the manifest stage. In contrast, manifest zQ175 mice (9 months) exhibited significantly decreased Glut1 immunofluorescent intensity in the striatum (Fig. 2H and I, P = 0.049) and motor cortex (Supplementary Fig. 2H and I, P = 0.036). The altered GLUT1 levels in endothelial cells may represent a cerebral vascular response to metabolic changes in Huntington’s disease brain, as the total GLUT1 protein levels, which are also expressed in astrocytes and other cells, did not show significant changes in zQ175 Huntington’s disease mice compared to age-matched controls (Supplementary Fig. 3). Taken altogether, our data suggest that the zQ175 brain initiates a compensatory cerebral vascular response to altered neuronal activity and/or energy metabolism in the premanifest stage, while impaired vasculature structure leads to lowered CBVa and possibly compromised compensatory regulation—such as reduced glucose transporter levels—thus exacerbating pathology and disease manifestation.

HTT silencing in the striatal neurons restores altered arteriolar cerebral blood volume in premanifest zQ175 mice

We next tested whether suppression of mHTT in neurons could normalize altered CBVa in the premanifest zQ175 mouse brain. Lowering mHTT by introducing CRISPR/Cas9-mediated HTT silencing to the striatal neurons has been reported to improve motor function in another mouse model with rare off-target effects.³² As demonstrated in our longitudinal characterization data, heterozygous zQ175 mice do not show motor phenotype and striatal atrophy at 3 months of age (Supplementary Fig. 1). Therefore, HTT silencing agents were administered into the striatum of 2-month-old mice.

To lower mHTT efficiently, we used combinations of two guiding RNAs targeting human HTT exon-1 regions (Fig. 3A). Two gRNAs were expressed under the U6 promoter in an AAV vector that expresses red fluorescent protein (RFP) (AAV-HTT gRNAs) for the purpose of tracking. The knock-down efficiency of these HTT gRNAs has been demonstrated previously.³² The spCas9 was expressed in a separate AAV vector under the Mecp2 promoter (AAV-Mecp2-spCas9) to drive neuronal expression.

Figure 3

CRISPR/Cas9-mediated HTT silencing in the striatal neurons restores altered CBVa in the striatum of zQ175 mice. (A) Schematics of the designed HTT-gRNA (T1 and T3) and AAV vectors. HA = human influenza haemagglutinin; ITR = inverted terminal repeat; KASH = Klarsicht, ANC-1, Syne Homology; NLS = nuclear localization sequence; RFP = red fluorescent protein; WPRE = woodchuck hepatitis virus post-transcriptional regulatory element. (B) Timeline for AAV injections, HTT level measures and longitudinal CBVa assessment. (C) RFP (red) fluorescence showing the transduction of AAV-HTT-gRNAs in the striatum. Green colour representing autofluorescence and showing the outline of striatum. Scale bar = 900 µm (left two images) and 40 µm (right two images). (D) The images of western blotting with indicated antibodies to HTT, NeuN, and β-actin. (E) Quantification results of mHTT (MW1), total HTT (combining upper and lower bands in the MCA2050 + MCA2051 blot), wtHTT (lower and major band in the blot with MAB 2166 antibody), and NeuN in the striatum injected with AAV-HTT gRNA + Cas9 (HTT gRNA, right striatum-R) or AAV-control gRNA + Cas9 (Con gRNA, left striatum-L). n = 3, *P < 0.05 versus the values of control gRNA group by Student’s t-tests. (F) Representative CBVa maps in mouse brains from indicated genotypes and treatment groups. Top row: The raw images, and the red regions of interest indicate the quantified brain region. Bottom row: The representative CBVa maps in the mice at the indicated genotypes and treatment at 3 months of age. The scale bars are shown on the right and warmer colour represents higher CBVa values. (G) Quantification of longitudinal CBVa changes in the striatum of zQ175 mice and wild-type littermates with indicated treatment at indicated ages. Mean ± SEM, n = 7 (four male and three female) in WT+Con, zQ175+Con, and WT+ HTT gRNA groups; n = 6 (three male and three female) in zQ175+HTT gRNA group. *P < 0.05, comparison between zQ175 mice injected with HTT gRNAs (+ spCas9) versus zQ175 mice injected with control RNA (+ spCas9) by two-way ANOVA with Bonferroni post hoc analysis.

To confirm the efficacy of CRISPR/cas9-mediated HTT lowering in zQ175 mice, we designed experiments as indicated (Fig. 3B). HTT gRNAs and spCas9 were injected into one side of the striatum (right side) and the contralateral striatum (left side) was injected with control gRNA and spCas9. RFP fluorescence in the striatum indicated virus transduction (Fig. 3C). Western blotting with different HTT antibodies indicated that levels of mutant HTT (MW1 blot) and total HTT were significantly reduced by HTT gRNA + spCas9 injection (Fig. 3D and E) 4 weeks after injection. NeuN (a pan neuronal marker) levels (Fig. 3D and E) were examined to verify that reduced HTT levels were not due to neuronal damage.

Confirming that the HTT gRNAs worked efficiently in zQ175 mice, we injected HTT gRNAs (or control) + spCas9 to striatum bilaterally of 2-month-old mice and evaluated outcomes at 3 months of age (when no striatal atrophy or motor deficits are present in zQ175 mice). Remarkably, zQ175 mice injected with HTT gRNA had rescued altered CBVa in the striatum (Fig. 3F, P < 0.05). Mice injected with HTT gRNA + spCas9 in the striatum exhibited a CBVa curve close to normal mice [Fig. 3G, wild-type (WT) + control (Con) versus zQ175 + Con, Age × Genotype F(2,24) = 19.51, P < 0.01; zQ175 + Con versus zQ175 + HTT gRNA, Age × Treatment F(2,24)= 4.7, *P < 0.05]. Lowering HTT in striatal neurons did not significantly affect the altered CBVa in the motor cortex (Supplementary Fig. 4A and B). These results suggest that the altered CBVa in premanifest zQ175 mice is most likely due to neuronal changes in either activity or metabolism. Our data confirmed that this non-allele-specific HTT lowering did not affect body weight or locomotor activity in control mice as well as zQ175 mice (Supplementary Fig. 4C and D). Because the iVASO MRI technique has been developed and successfully applied to the human brain,⁴² the present study provides a proof-of-principle to consider CBVa measured by iVASO as a biomarker to evaluate therapeutic efficacy in premanifest Huntington’s disease clinical trials.

HTT silencing in striatal neurons delays onset and slows progression of disease in zQ175 mice

Given that the onset and severity of clinical manifestations of Huntington’s disease does not only depend on neuronal loss but also on neuronal dysfunction, which may occur at an early stage of the disease,⁵⁷ we asked whether lowering HTT at the premanifest stage could delay or even prevent the onset of phenotype and pathology in zQ175 mice. HTT gRNAs (or control) + spCas 9 were administered to mice at 2 months of age. Longitudinal measures were assessed from 3 months to 9 months (Fig. 4A). Similar to untreated zQ175 mice, control gRNAs injected zQ175 mice displayed significant striatal atrophy (Fig. 4B, P < 0.01) and motor impairment (Fig. 4C and D, P < 0.05) at 6 months. In contrast, zQ175 mice treated with HTT gRNAs + spCas9 had no significant striatal atrophy (Fig. 4B) or motor deficits (Fig. 4C and D) at 6 months, indicating that HTT lowering at the premanifest stage delays the onset of motor phenotype and striatal atrophy in zQ175 mice.

Figure 4

HTT silencing at the premanifest stage delays the onset and slows the progression of phenotype and pathology in zQ175 mice. (A) Timeline for AAV injections and outcome measures. (B) Longitudinal striatal volume data were quantified from structural MRI scans from indicated groups at indicated ages. (C) Mice were tested on a tapered beam and time crossing the beam (traverse time) was recorded from different groups at indicated ages. (D) Mice were tested on a 5 mm beam and time crossing the beam (traverse time) was recorded from different groups at indicated ages. All data in B–D are mean ± SEM, n = 7 (four male and three female) in WT+Con, zQ175+Con, and WT+ HTT gRNA groups; n = 6 (three male and three female) in zQ175+HTT gRNA group. *P < 0.05, comparison between zQ175 mice injected with HTTg RNAs (+ spCas9) versus zQ175 mice injected with control RNA (+ spCas9) by two-way ANOVA with Bonferroni post hoc analysis. (E) Mutant HTT (mHTT) aggregates were labelled by immunostaining with EM48 antibody in the striatum of zQ175 mice at 9 months (9 M) of age. EM48-positive mHTT aggregates (green), pan neuronal marker NeuN (red), and nucleus (blue, Hoechst) were indicated. Scale bar = 10 μm. (F) Numbers of mHTT aggregates/mm² were quantified in the striatum. Mean ± SEM, n = 4 (two male and two female) per group, **P < 0.01 versus the values of con gRNA group by standard Student’s t-tests.

HTT silencing in the striatal neurons also significantly slowed the progression of striatal atrophy [Fig. 4B, WT + Con versus zQ175 + Con, Age × Genotype F(2,24)= 6.926, P < 0.01; zQ175 + Con versus zQ175 + HTT gRNA, Age × Treatment F(2,22)= 3.584, ^#P < 0.05] and ameliorated motor deficits on both tapered beam [Fig. 4C, WT + Con versus zQ175 + Con, Age × Genotype F(2,28) = 5.562, P < 0.01; zQ175 + Con versus zQ175 + HTT gRNA, Age × Treatment F(2,40)= 2.1, *P < 0.05] and 5 mm beam tests [Fig. 4D, WT + Con versus zQ175 + Con, Age × Genotype F(2,27)= 4.635, P  < 0.05; zQ175 + Con versus zQ175 + HTT gRNA, age × Treatment F(2,35) = 3.382, *P < 0.05] in zQ175 mice. CBVa levels in the motor cortex, body weight, and locomotor activity were not significantly modulated by lowering HTT in the striatum (Supplementary Fig. 3A–D).

A histopathological hallmark of Huntington’s disease brains is mHTT aggregates, we determined whether HTT silencing by CRISPR/Cas9 affected mHTT aggregation, as mHTT aggregates may form due to mutant HTT that increases the cellular concentrations or renders them aggregation-prone. Thus, HTT silencing may reduce the mHTT aggregates if mHTT is efficiently lowered. After the last longitudinal assessment at 9 months, zQ175 mice were sacrificed and mHTT aggregates were labelled by immunostaining with the EM48 antibody (Fig. 4E). We found that HTT gRNA + Cas9-injected zQ175 mice had significantly less mHTT aggregates in the striatum compared with mice injected with control gRNAs (Fig. 4F, **P < 0.01), suggesting that this HTT silencing strategy efficiently lowered mHTT levels in the neurons.

Discussion

A challenge in developing treatments for Huntington’s disease is the identification of demonstrable outcome measures for evaluating intervention efficacy in delaying onset or improving underlying pathological processes prior to measurable clinical symptoms. Our study has shown for the first time that a sensitive physiological MRI measure responds to an HTT-lowering treatment in the premanifest period, and demonstrates the feasibility of detecting arteriolar perfusion changes in the premanifest Huntington’s disease brain with a robust MRI technique that is suitable for longitudinal evaluations of therapeutic efficacy. Moreover, we provide the first evidence that introducing HTT-lowering treatment before the occurrence of motor symptoms and striatal atrophy delays onset and slows progression of pathology and phenotype in a full-length mHTT expressing mouse model. The present results also suggest that altered CBVa in premanifest zQ175 mice is secondary to the effects of mHTT on neural activity/metabolism, and that a reduced rate of oxygen/nutrient delivery due to reduced CBVa and decreased glucose transporter GLUT1 across a compromised neurovascular network in the manifest stage might eventually trigger neuronal dysfunction and degeneration.

Rodent models play an important role in elucidating phenotypic progression and in investigating potential underlying mechanisms of clinical symptoms of Huntington’s disease.⁵⁸ The validation of potential biomarkers in animal models has aided in the preclinical development of new therapeutic modalities. However, neuroimaging, especially functional and physiological brain imaging methods, has only recently begun to be applied to mouse models. In the present study, we utilized iVASO MRI, developed initially for the human brain, and observed similar alterations in premanifest mouse and human brains of Huntington’s disease. Our findings demonstrate that significant changes in CBVa occur before striatal atrophy and motor symptoms, further supporting that altered cerebrovascular function is an early event in Huntington’s disease.

The fact that CBVa changes in the striatum emerge over time supports that dysfunction in the basal ganglion circuitry is associated with disease phenotype, though underlying mechanisms that link CBVa and neuronal dysfunction (or compensation for dysfunctional neuronal metabolism) in premanifest Huntington’s disease remain to be explored. These changes indicate that there is a premanifest therapeutic window to test interventions in this mouse model, which is not present in faster progressing mouse models. While no animal model replicates all of the features of Huntington’s disease, the heterozygous zQ175 model offers an alternative system to study functional changes in the premanifest period.

The use of iVASO MRI for CBVa assessment may provide an improved means for the bidirectional translation of findings between animal models and human Huntington’s disease. Reverse translation of biomarkers to rodents also provides an important tool for evaluating the therapeutic effects of candidate treatment,⁵⁹ which is particularly relevant as several clinical trials on lowering HTT using different strategies, including antisense oligonucleotides and RNAi, are ongoing.⁴ It is notable that most mouse models of Huntington’s disease, including the zQ175 model, are not characterized by overt neuronal death, which argues against a direct involvement of neuronal loss. Such interventions may target neuronal dysfunction and consequently modify disease progression. In fact, the severity of clinical manifestations in Huntington’s disease likely does not depend solely on neuronal loss but also on neuronal dysfunction and circuitry reorganization.⁵⁷ Our data support the existence of cerebrovascular abnormality in Huntington’s disease that may be independent from the neuronal loss characteristic of the disease. Overall, this study not only reveals a promising imaging biomarker for premanifest Huntington’s disease, but also uncovers effective therapeutic windows. Future studies will need to address the predictive validity of using CBVa in Huntington’s disease clinical trials and giving HTT-lowering treatment in premanifest subjects.

Microvascular abnormalities in different segments of the microvasculature (arteriolar and venous blood vessels) have been identified in neurodegenerative diseases.^8,21 The supply of adequate oxygen and energy substrates for local metabolic demands is controlled by blood vessels and the status of the microvasculature is closely associated with brain functions and energy metabolism. Different types of blood vessels have distinct functions and physiology and may be differentially affected by disease pathology. Arterioles are the most actively regulated blood vessels, and thus may be more sensitive to metabolic disturbances in the brain.^29–31 The iVASO MRI approach can measure CBV in small pial arteries and arterioles (CBVa),^29,42 the primary regulator of local tissue perfusion. The iVASO method has been used to measure neurovascular abnormalities in human brains of several disease conditions, such as brain tumours,⁶⁰ Huntington’s disease,¹⁶ Parkinson’s disease,⁶¹ Alzheimer’s disease dementia,^61,62 and schizophrenia.⁴³ In the current study, we demonstrated that CBVa was significantly increased at premanifest stage in heterozygous zQ175 mice, while structural MRI measures showed significant striatal atrophy later, indicating that iVASO CBVa measures may be a sensitive biomarker for assessing therapeutic response in the premanifest clinical trials.

Current ongoing clinical trials of HTT-lowering approaches target all cells in the brain. Our study, for the first time, demonstrated that HTT silencing in striatal neurons can restore CBVa in the basal ganglion circuitry of premanifest Huntington’s disease brains, consequently delaying the onset of striatal atrophy and motor deficits, and slowing pathological progression in zQ175 mice. Our results implicate that elevated CBVa in premanifest Huntington’s disease brain is likely a compensatory response to altered energy demands in striatal neurons, as reducing mHTT levels in those neurons rescued the elevated CBVa in the striatum of premanifest zQ175 mice. This early compensatory response could be impaired over the course of disease progression due to permanent morphological changes in the cerebral vasculature during the manifest period. Thus, an inability to increase blood supply in response to increased energy demands may accelerate neuronal dysfunction and degeneration in the long term.

An underlying assumption in our study was that CBVa precisely reflected specific pathophysiological processes. However, we are aware that iVASO MRI techniques still offer a limited characterization of vascular properties. Consequently, it is important to exert caution about the observed CBVa changes with disease progression. Although abnormal cerebrovascular trajectories may reflect a tentative ordering in which pathophysiological events occur, our results should be interpreted within the scope of biomarker sensitivity to disease progression instead of causal pathological interactions. Another potential limitation of our study is that all evaluations were performed within a linear regression framework, which could mean that obtained results primarily reflect linear tendencies in the analysed outcome measures. These traditional limitations suggest the need to study disease progression not only in terms of alterations of specific biomarkers, but also through analysis of the multifactorial causal pathological interactions that take place on different spatiotemporal scales.

In the current study, region of interest-based analysis focusing on the motor cortex and striatum was adopted for the comparison of CBVa. The two brain regions are known to be the most affected in Huntington’s disease in general and in this particular mouse model as evidenced from our previous work on structural brain changes,⁴⁵ and are also appropriate for our treatment scheme. Voxel-based analysis on CBVa in the entire brain was also attempted. For instance, representative t-maps (maps of t-scores) from the voxel-based comparison between the zQ175 and wild-type mice at 9 months old are shown in Supplementary Fig. 6. While decreasing trends in the motor and striatum regions can be observed, few voxels can survive the multiple comparisons correction mainly due to the relatively small sample size in the current study. However, the results from the current study will allow us to adjust the power analysis in subsequence studies for whole brain voxel-based analysis.

There is increasing evidence that a number of neurodegenerative disorders are associated with alterations in the cerebral vasculature.⁵⁵ A multifactorial data-driven analysis for all biomarkers and a tentative temporal ordering of disease progression indicate that cerebrovascular dysregulation is an early pathological event during Alzheimer’s disease development.⁶³ Our present findings implicate a similar pathology in Huntington’s disease. The highly regulated neurovascular coupling system ensures that regional blood supply is increased to meet energy needs and remove metabolic waste. This system is often impaired by disease pathology. Our data suggest that elevated CBVa may be one of the initial steps contributing to observed vascular modifications in premanifest Huntington’s disease and restoration of CBVa during this particular period by the HTT-lowering approach or other disease-modifying treatment could be an indication of improved neuronal function/metabolism. Additionally, we demonstrate that early longitudinal measures of CBVa changes may be predictive of later therapeutic efficacy, and that lowering HTT before clinical symptoms manifest has long-lasting beneficial effects in Huntington’s disease. Further validation of these findings in human clinical trials will facilitate the development of efficient therapeutic interventions for premanifest subjects, with a goal to delay or even conceivably prevent onset of manifest Huntington’s disease.

Acknowledgements

We thank Dr Xiaojiang Li (at Emory University then) for providing CRISPR/Cas9-related viral vectors (PX551, PX552) and HTT gRNAs constructs.

Funding

This work is supported by the National Institutes of Health under awards R21NS104480 (NINDS) to J.H. and W.D. and R01NS082338 (NINDS) to W.D. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Competing interests

C.A.R. is a consultant for: Huntingtin Study Group (HSG), Annexon, Mitoconix, NeuBase, NeuExcell, Roche/Genentech, SAGE, Spark, TEVA, uniQure, Wave. P.C.M.vZ. has a patent on VASO technology (the parent technology for iVASO) that has been licensed to Philips Healthcare. All other authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.

 
• CBVa

  arteriolar cerebral blood volume

 
• iVASO

  inflow-based vascular-space-occupancy