Abstract
Metachromatic leukodystrophy (MLD) is a rare, inherited, demyelinating lysosomal storage disorder caused by mutations in the arylsulfatase-A gene (ARSA). In patients, levels of functional ARSA enzyme are diminished and lead to deleterious accumulation of sulfatides. Herein, we demonstrate that intravenous administration of HSC15/ARSA restored the endogenous murine biodistribution of the corresponding enzyme, and overexpression of ARSA corrected disease biomarkers and ameliorated motor deficits in Arsa KO mice of either sex. In treated Arsa KO mice, when compared with intravenously administered AAV9/ARSA, significant increases in brain ARSA activity, transcript levels, and vector genomes were observed with HSC15/ARSA. Durability of transgene expression was established in neonate and adult mice out to 12 and 52 weeks, respectively. Levels and correlation between changes in biomarkers and ARSA activity required to achieve functional motor benefit was also defined. Finally, we demonstrated blood–nerve, blood–spinal and blood–brain barrier crossing as well as the presence of circulating ARSA enzyme activity in the serum of healthy nonhuman primates of either sex. Together, these findings support the use of intravenous delivery of HSC15/ARSA-mediated gene therapy for the treatment of MLD.
SIGNIFICANCE STATEMENT Herein, we describe the method of gene therapy adeno-associated virus (AAV) capsid and route of administration selection leading to an efficacious gene therapy in a mouse model of metachromatic leukodystrophy. We demonstrate the therapeutic outcome of a new naturally derived clade F AAV capsid (AAVHSC15) in a disease model and the importance of triangulating multiple end points to increase the translation into higher species via ARSA enzyme activity and biodistribution profile (with a focus on the CNS) with that of a key clinically relevant biomarker.
Introduction
Metachromatic leukodystrophy (MLD) belongs to a class of lysosomal storage disorders (LSDs) with an overall prevalence estimated to be between 1:40,000 and 1:160,000 (Gomez-Ospina, 1993). LSDs have in common a lysosomal dysfunction, membrane-enclosed organelles that contain an array of enzymes whose function is to break down proteins, nucleic acids, carbohydrates, and lipids. MLD is inherited as an autosomal recessive trait with the majority of cases because of mutations in the arylsulfatase-A (ARSA) gene, which encodes ARSA enzyme, whereas mutations in the prosaposin (PSAP) gene accounts for fewer cases (Siri et al., 2014). Disease-causing mutations in the ARSA gene lead to a reduction/absence of ARSA enzymatic activity, resulting in a toxic accumulation of sulfatides (Lukatela et al., 1998). Sulfatides are one of the most common sphingolipids in myelin and are present in cells of the PNS (Schwann cells) and CNS (oligodendrocytes). This disease initially has an impact on the nerve fiber myelin sheath, resulting in progressive motor and cognitive impairments (Siri et al., 2014; Rosenberg et al., 2016). Accumulation of sulfatides in Schwann cells, macrophages, oligodendrocytes, microglia, and neurons lead to axonal degeneration. Excess levels of sulfatides have also been associated with neuronal degeneration, astrocyte dysfunction, and may trigger an inflammatory response (i.e., astrogliosis and microgliosis; for review, see Shaimardanova et al., 2020). Indeed, microglial cells are assumed to play a role in the development of MLD, where it has been shown that its immune phenotype changes at the early stages of MLD development upstream of oligodendrocyte degeneration (Bergner et al., 2019). Sulfatides also accumulate in the cells of internal organs, such as the gall bladder (van Rappard et al., 2016; Kim et al., 2017; Almarzooqi et al., 2018). MLD patients commonly present with gait abnormalities or delays in early motor milestones, first presented by loss of ambulatory skills by 2 years of age. Subsequently, peripheral neuropathy, regression of speech, decrease in cognitive and motor abilities, deterioration of fine motor skills, spasticity, ataxia, convulsions, and visual and hearing impairments occur (Shaimardanova et al., 2020). Finally, the extent of sulfatide and lysosulfatide accumulation within peripheral sensory nerves and CSF is directly proportional to the severity of damage to the nervous tissue (Dali et al., 2015).
Libmeldy is a lentiviral-based ex vivo gene therapy for MLD, approved for use in Europe, for children with late infantile or early juvenile forms, without or with very early clinical manifestations of the disease. Other ongoing clinical trials include two intrathecal enzyme replacement therapy (ERT) approaches (IDs NCT01510028 and NCT03771898; Whiteman and Kimura, 2017). Given the aggressive and early onset of neurologic presentations in afflicted patients, rapidly delivering sufficient levels of ARSA to the PNS and CNS is thought to be critical. Ideally, the ARSA enzyme would be expressed in a fashion that mimics that of the endogenous ARSA biodistribution with respect to organs and cell types that normally express it. Although a direct CNS route of administration (ROA) can bypass the blood–brain barrier (BBB) limitations, there are drawbacks to this method, including the invasive nature of the required dosing procedures, higher risks for infections at the injection site, and lower biodistribution to a large number of peripheral organs (Puhl et al., 2019; Chen et al., 2020). Thus, to reach the full potential of the therapy, combining the optimal capsid and route of administration able to target the various desired organs and cell types is of great importance.
Herein, we evaluated ROA (intrathecal and intravenous) and adeno-associated virus (AAV) Clade F capsid serotypes AAVHSC15 (hereafter referred to as HSC15) and AAV9, for the potential to restore endogenous biodistribution and function of ARSA in the CNS in a well-established mouse model of MLD. We examined the modulation of clinically relevant CNS biomarkers and extended our biodistribution studies to healthy nonhuman primates (NHPs) to support translation to humans. Together, these findings support the use of an intravenous gene therapy approach to address the nervous system and peripheral manifestations of MLD.
Materials and Methods
Vector design and production of HSC15/ARSA
HSC15/ARSA is a recombinant version of a naturally occurring AAV (rAAV) serotype 15 (rAAVHSC15) expressing the human ARSA cDNA codon optimized by GenScript under the control of a ubiquitous CAG promoter (Niwa et al., 1991) and followed by the SV40 early polyadenylation signal. In our previous publications, HSC15/ARSA is referred to as HMI-202. The 3949 bp vector genome (vg) was packaged in the rAAVHSC15 capsid, a natural Clade F AAV variant that was isolated from CD34+ human peripheral blood stem cells from healthy adults (Smith et al., 2014). HSC15/ARSA was used to assess biochemical and behavioral efficacy in a MLD mouse model in vivo, as well as translatability in NHPs. In the latter, an additional vector was used, HCS15/ARSA-V5, which is an analog of HSC15/ARSA and is identical to HSC15/ARSA with the addition of a V5-tag (GKPIPNPLLGLDST) at the C terminal of the ARSA sequence to facilitate detection. This was designed to address the 96% sequence homology at the amino acid level between human and cynomolgus monkey arylsulfatase-a, making them indistinguishable when using availably commercial antibodies against ARSA. HSC15 vector stocks were produced by transient transfection using triple plasmid in HEK293 suspension cells and purified using affinity chromatography followed by anion exchange chromatography. The drug substance was formulated in an isotonic, neutral pH buffer suitable for in vivo administration. Vector preparation was tittered by droplet digital PCR (ddPCR) for vg concentration and sandwich enzyme-linked immunosorbent assay for capsid titer. Purity was assessed by capillary electrophoresis performed in the presence of sodium dodecyl sulfate and aggregation by size-exclusion chromatography (SEC). The percentage of empty capsids were determined using analytical ultracentrifugation. Vg titers were determined by ddPCR using the following ARSA primers and probe: forward ARSA primer, 5′-CCCTAGGCAGTCTCTGTTCT-3′; reverse ARSA-primer, 5′-TTGTACTTGCCGGTTCTCAC-3′; ARSA probe, 56-FAM/CTCCTATCC/ZEN/TGATGAGGTGCGGG/3IABkFQ.
Microglia culture
Primary iCell Microglia (FujiFilm) were cultured according to manufacturer specifications. Briefly, cells were cultured in iCell Glia Base Medium containing iCell Microglia Supplements A and B and iCell Neural Supplement C on poly-d-lysine-coated 96-well plates at a density of 20,000 cells/well. Every other day, half the medium was replaced with new culture medium containing the supplements. On day 2 after plating, iCell microglia were transduced with 100 K multiplicity of infection (MOI) using AAV9 eGFP or AAVHSC15 eGFP vectors. The cells were harvested and analyzed 72 h post-transduction for vgs and eGFP transcripts for expression.
Vgs/cell
iCell microglia were washed in the plate using cold PBS twice and then lysed using a cell lysis buffer containing 10% deoxycholate, 0.45% Tween 20/HEPES solution and proteinase K (0.3 mg/ml). Samples were thoroughly mixed in the plate and then placed in a thermal cycler and incubated at 37°C for 1 h, 55°C for 2 h, and 95°C for 30 min. Samples were further diluted 1:10, and the diluted samples were then analyzed for vector genome copies via quantitative PCR (qPCR) using eGFP as well as APOB using the following primers and probes: forward eGFP primer, 5′-CTG CTG CCC GAC AAC CA-3′; reverse eGFP primer, 5′-GAC CAT GTG ATC GCG CTT CT-3′; eGFP probe, FAM/TAC CTG AGC ACC CAG TCC GCC CT/3IABKFQ/; forward APOB primer, 5′-TGA AGG TGG AGG ACA TTC CTC TA-3′; reverse APOB primer, 5′ CTG GAA TTG CGA TTT CTG GTA A-3′; and APOB probe, HEX/CGA GAA TCA CCC TGC CAG ACT TCC GT/3IABKFQ/.
eGFP transcript measurement via qRT-PCR
RNA was extracted from the transduced cells using the Qiagen RNeasy Mini Kit (catalog #74104). The RNA quality was assessed using nanodrop. Complementary DNA (cDNA) was synthesized using the High-Capacity cDNA Reverse Transcription Kit (catalog #4368814, Applied Biosystems). Transcript levels of eGFP were assessed using qPCR assay and primer/probe sets specific to eGFP transcript.
Arsa KO mice: 129S6/SvEvTac.Arsatm1Gies
Housing and breeding were in accordance with the Institutional Animal Care and Use Committees (IACUCs) at Homology Medicines and Taconic Biosciences. All procedures were performed in accordance with the IACUCs at Homology Medicines. For these studies, we have used a well-established MLD mouse model generated by the Gieselman lab and used in a number of investigational therapies for MLD (Hess et al., 1996). This model was rederived in house on a 129S6/SvEvTac background. Genotyping was performed by Transnetyx using real-time PCR. Homozygous Arsa mice are 129S6/SvEvTac.Arsa−/− (Arsa KO) and WT Arsa mice are 129S6/SvEvTac.Arsa+/+. Males and females of various ages were used in the studies, as described in the figure legends.
NHPs
Cynomolgus macaques (Macaca fascicularis) NHPs were used in the studies. Four NHPs ranging from 8 to 11.5 months of age and screened negative for anti-AAV9 neutralizing antibodies were used to determine ARSA biodistribution. Housing and all procedures were conducted by the Mannheimer Foundation veterinary staff in accordance with their IACUC, a laboratory accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). NHPs were group housed and fed Purina LabDiet 5049, with fruit and/or produce enrichment. HSC15/ARSA-V5 (n = 2) or vehicle (n = 2) was administered as 5 ml/kg as a slow 60 s bolus intravenous injection, and animals were killed at 29 d postadministration using a xylazine/ketamine combination. To determine ARSA protein concentrations in liver, brain, and plasma and ARSA enzyme activity in serum, eight additional NHPs (two males and two females per group) ranging from ∼11.5 to 12.5 months of age screened negative for anti-HSC15 neutralizing antibodies were used. Housing and all procedures were in accordance with the Standard Operating Procedures of Charles River Laboratories, an AAALAC-accredited laboratory. HSC15/ARSA or vehicle was administered as a 30 min continuous intravenous infusion on day 1 via the saphenous or other suitable vein via a percutaneously placed catheter. On day 29, animals were killed by intravenous solution administered under sedation according to the test facility standard operating procedure (SOP).
Vg copy number in mouse tissue
Vg copy number was determined by ddPCR assay using the Automated Droplet Generator from Bio-Rad. Genomic DNA was extracted from brain tissue samples using the QIAamp Fast DNA Tissue Kit according to the protocol from the manufacturer (catalog #51404, Qiagen). The ddPCRs were performed in 96-well plates and each plate was run with a no template control and study samples in duplicate. Vector-specific sequences were amplified using primers and probes for the codon-optimized ARSA, which are described as follows: forward ARSA primer, CTCTCTGATGGAGCTGGATGC; reverse ARSA primer, CGCTGCATCCGCCTCTAG; ARSA probe, CGCCGATTGCTGTCATCAGGGTGCCCACG. The ARSA probe consists of the following: HEX fluorescence reporter dye at the 5′ end of the probe, internal ZEN quencher (between the 9th and 10th nucleotides), and Iowa Black fluorescent quencher at the 3′ end of the probe. Samples were also evaluated for mouse genomic phenylalanine hydroxylase (Pah), which served to quantitate cell equivalents and as a template integrity control. Primer and probe sequences are as follows: forward Pah primer, CTGAGCAATGCATTCAGCAATAA; reverse Pah primer, GCAAGCTCCAGATCACCAATA; Pah probe, CCCTGAAACACCCTTGACAGAGCA. The Pah probe consists of the following: FAM fluorescence reporter dye at the 5′ end of the probe, internal ZEN quencher (between the 9th and 10th nucleotides), and Iowa Black® fluorescent quencher at the 3′ end of the probe.
Vg copy number in NHP tissue
Vg copy number was determined by a TaqMan-based qPCR assay using the QuantStudio 7 Flex Real-Time PCR Systems validated at Charles River Laboratories. Calibration standards were prepared using plasmid containing the ARSA sequence (pHM-05,000 plasmid) linearized by XhoI. Total DNA was extracted from NHP tissue homogenates and blood using QIAsymphony DSP DNA mini kit on QIAsymphony per test facility SOPs. Vector-specific sequences were amplified by qPCR using primers and probes for the codon-optimized ARSA, which are described as follows: ARSA-F2 forward primer, 5′-CTCTCTGATGGAGCTGGATGC-3′; ARSA R2 reverse primer, 5′-CGCTGCATCCGCCTCTAG-3′; ARSA P2 probe, 5′-CGCCGATTG/ZEN/CTGTCATCAGGGTG-3′. The ARSA probe consists of the following: 6-FAM, fluorescence reporter dye at the 5′ end of the probe, internal ZEN quencher (between the 9th and 10th nucleotides), and Iowa Black fluorescent quencher at the 3′ end of the probe.
ARSA mRNA levels in mouse tissue
Human ARSA mRNA levels were determined by ddPCR after RNA isolation from tissues using Trizol followed by purification using the RNEasy Mini Kit from Qiagen. cDNA synthesis was performed using the High-Capacity cDNA Reverse Transcriptase Kit (Thermo Fisher Scientific) according to instructions from the manufacturer. ARSA transcripts were detected using the primers and probes for the codon-optimized ARSA listed above. Samples were also evaluated for mouse glucuronidase beta or mouse hypoxanthine guanine phosphoribosyl transferase transcript using commercial primer and probe sets from Thermo Fisher Scientific, as a means of normalizing for differences in cDNA synthesis efficiency and assay input.
ARSA enzymatic activity in mouse and NHP tissues
A colorimetric enzyme assay catalog #N7251, Millipore Sigma that measures the cleavage of sulfate from the soluble substrate p-nitrocatechol-sulfate (pNCS) was used to determine arylsulfatase A-specific enzyme activity in tissue homogenates or serum samples from either mouse Arsa KO or NHP tissues. Nonspecific cleavage of sulfate from competing enzymes was eliminated from tissue lysates using an anti-ARSA-specific immunoprecipitation step (goat-anti-human ARSA antibody; catalog #PA5-47330, Thermo Fisher Scientific). Following immunoprecipitation, ARSA enzyme activity was determined as described for Recombinant Human Arylsulfatase A/ARSA (catalog #2485-SU, R&D Systems). ARSA enzyme activity was expressed as nanograms of desulfated pNCS, denoted pNC, per milligram of protein per hour and/or percentage of normal adult human brain ARSA enzyme activity. Protein concentration of tissue homogenates was determined using the Pierce BCA Protein Assay test kit (catalog #23225, Thermo Fisher Scientific). Normal adult human ARSA enzyme activity in brain tissues was determined as described above using two postmortem specimens from two adult males and frontal cortex samples from two adult females obtained from BioIVT.
Mass spectrometry for ARSA protein quantitation in NHP tissues
Samples were collected from the liver (left lateral lobe and median lobe) and the brain (right hemisphere, hindbrain, sections 4 and 5b; Bolon et al., 2018) from all NHPs at 28 d postdose and placed into cryotubes, flash frozen on dry ice, and stored frozen at −60° to −90°C. For evaluation of ARSA protein in plasma, blood samples were collected via the femoral artery/vein into microtainer tubes containing K2EDTA, centrifuged at 2°C to 8°C, aliquoted into cryovials, and stored at −60° to −90°C until analysis. The quantitation of ARSA protein in liver and brain was performed at KCAS Bioanalytical and Biomarker Services using a non-GLP (good laboratory practice) liquid chromatography (LC)-MS/MS direct digest peptide method involving a direct digestion of tissue homogenates with trypsin, followed by sample clean-up via SPE. Next, samples were diluted to reduce matrix effect, and a human ARSA-specific GGL peptide (GGLPLEEVTVAEVLAAR) was used to quantitate ARSA in samples on a Sciex API6500 LC-MS/MS system. The internal standard (IS), a flanking heavy labeled (R-13C, 15N) peptide which yields the GGL peptide on digestion, was added to samples after homogenization to ensure accuracy, precision, and robustness. The assay was linear within the range of 3.00–1280 ng/ml and has been qualified for use on cynomolgus monkey tissue samples obtained during nonclinical studies. In addition, BCA analysis was performed using the Pierce BCA Protein Assay test kit within the range of 25–2000 µg/ml. Plasma ARSA protein was also determined by using a non-GLP hybrid LC-MS/MS method qualified at KCAS Bioanalytical and Biomarker Services. This method was similar to that used to quantitate ARSA in tissue samples; however, sample cleanup and enrichment was achieved by an immunoprecipitation step using streptavidin-coated magnetic beads (catalog #KS0-9533, Phenomenex) and biotinylated anti-human ARSA antibodies (catalog #LS-C688795, LSBio; catalog #BAF2485, R&D Systems). The internal standard, with flanking sequences, was added to samples after the immunoprecipitation step and before trypsin digestion. After digestion, samples were diluted 10-fold to minimize matrix effects before injection on a Sciex API6500 LC-MS/MS system, where the ARSA-specific GGL peptide was quantitated to determine ARSA protein concentration. The assay was linear within the range of 3.00–1280 ng/ml.
Sulfatide levels in mouse tissue
Analysis of sulfatide isoforms was conducted at PureHoney Technologies. Briefly, brain tissue from each level was weighed and homogenized in water in a Precellys bead homogenizer, and aliquots of the homogenate were removed for Pierce BCA protein assay quantification. Acetonitrile was added to each homogenate and the mixture was homogenized a second time. The homogenate was centrifuged at 14 000 × g for 15 min, and the centrifuge-clarified supernatant was removed and diluted 5× in 75% acetonitrile for RapidFire MS analysis. C19:0 sulfatide (catalog #1888; Matreya) was used as the IS and monitored together with C18:0, C18:1, C24:0, and C24:1 sulfatides in multiple reaction monitoring mode on a Sciex API4000 triple quadrupole mass spectrometer. Each sample was injected eight times with eight different concentrations of C19:0 sulfatide IS to generate a unique standard curve for each sample, which was used to calculate the concentration of each analyte.
Neurophysiological rotarod assay
A standard ramping 4–40 rpm rotarod assay was conducted at Charles River Laboratories. Briefly, mice on the study were all weighed before each study run and benchmarked on the rotarod performance test before dosing. The rotarod performance was composed of 1 d of testing at a constant speed (7 rpm, two trials of 120 s each), followed by a single day of testing in accelerating mode (4–40 rpm, three trials of 300 s each).
Immunohistochemistry in mouse and NHP tissues
All immunohistochemistry (IHC) was performed at Premier Laboratories. All mouse and NHP tissue samples for IHC were postfixed in 4% paraformaldehyde (PFA) in 0.1 m phosphate buffer (PB; catalog #S2558, Poly Scientific R&D) for 24 h at 4°C and then transferred to PBS containing 0.05% sodium azide (catalog #P0202, Teknova) for long-term 4°C storage. Stored samples were shipped at ambient temperature to Premier Laboratory for embedding in Tissue-Tek Paraform (catalog #7052, Sakura Finetek) paraffin blocks. The blocks were sectioned at a thickness of 4–6 µm, and IHC was performed using an Autostainer Plus Link (Agilent). Briefly, after overnight drying, deparaffinization, and epitope retrieval, sections were incubated with a blocking solution (catalog #X0909, Agilent) followed by primary antibody, either a polyclonal goat anti-ARSA (0.5 µg/ml; catalog #PA5-47330, Thermo Fisher Scientific), a polyclonal rabbit anti-LAMP1 primary antibody (0.25 µg/ml; catalog #ab24170, Abcam) or a polyclonal anti-GFAP antibody (1.93 µg/ml; catalog #Z0334, Agilent) for 30 min at room temperature. All conditions were followed by a 15-min-duration incubation in either EnVision+ System-horseradish peroxidase (HRP) Labeled Polymer Anti-Rabbit (catalog #K4001, Agilent) or Goat-on-Rodent-HRP Polymer (catalog #GHP516, Biocare Medical) depending on the host the antibodies were raised in. Diaminobenzidine-positive (DAB+) Chromogen Solution (Agilent, K4071) was applied for 5 min, then slides were manually rinsed in tap water and counterstained with a modified Harris Hematoxylin (Dako Agilent Pathology Solutions, S330130-2) for 5 min to highlight cellular nuclei. The sections were then dehydrated, cleared with xylene, and coverslipped. Slides were scanned on a Aperio AT2 ScanScope (Leica Biosystems) digital scanner, and digital images were visualized and captured using the Aperio eSlide Manager Software (Leica Biosystems).
Immunofluorescence in NHP tissue
NHP tissue samples for immunofluorescence (IF) were postfixed in 4% PFA in 0.1 m PB for 24 h at 4°C and then transferred to PBS containing 0.05% sodium azide for long-term 4°C storage. Stored NHP samples were shipped at ambient temperature to NeuroScience Associates for embedding and sectioning using Large Format technology. These blocks were sectioned at a thickness of 40 µm, and IF was performed on free-floating sections. Sections were incubated in a blocking solution composed of 5% normal donkey serum (catalog #ab7475, Abcam) and 0.25% Triton X-100 Surfact-Amps Detergent Solution (catalog #85111, Thermo Fisher Scientific) in 1× Tris buffered saline (TBS; catalog #28358, Thermo Fisher Scientific). Sections were incubated overnight at 4°C with a set of primary antibodies, including polyclonal goat anti-ARSA (10 µg/ml; catalog #PA5-47330, Thermo Fisher Scientific) and/monoclonal mouse anti-V5 (11 µg/ml; catalog #R960-25, Thermo Fisher Scientific). Sections used as negative controls were incubated with IgG isotype control antibodies (goat IgG, catalog #ab37373; mouse IgG, catalog #ab37355, Abcam) matching the host species and concentration of the primary antibodies used. Sections were incubated for 2 h at room temperature in secondary antibodies conjugated to different Alexa Fluor dyes (catalog #A-11055 and #A-31570, Thermo Fisher Scientific). Sections were mounted onto large glass microscope slides (catalog #2947, Corning) and allowed to dry partially. Then 1× TrueBlack Lipofuscin Autofluorescence Quencher (catalog #23007, Biotium) was added to the slide for 30 s and washed several times with 1× TBS. All slides were coverslipped after adding ProLong Gold Antifade Mountant with DAPI (catalog #P36931, Invitrogen) and imaged using an LSM 800 with Airyscan confocal microscope (Zeiss) and Zeiss ZEN (blue edition) imaging software.
Image analysis in mouse tissue
LAMP-1 image analysis was performed using ZEN software version 3.2 (blue edition, Zeiss). The total intensity of anti-LAMP1-positive pixels was analyzed from 20,000 µm2 [ventral white matter (vWM)] or 40,000 µm2 [dorsal white matter (dWM) and ventral gray matter (vGM)] square-millimeter regions of interest (ROIs) taken from spinal cord sections (n = 2–3 per animal). GFAP image analysis was performed using ImageJ software version 1.53c (National Institutes of Health). Briefly, the IHC Toolbox plug-in was used to remove the hematoxylin-stained nuclei so that only the DAB stain of interest (anti-GFAP) remained in the image. The mean gray value of DAB-positive pixels was analyzed from 40,000 µm2 ROIs taken from brain or spinal cord sections (n = 2 per animal).
Statistical analysis
Indicated statistical tests were performed as appropriate using GraphPad Prism software, version 9.0.0. Unpaired two-tailed Student's t tests were used to compare differences between two unpaired experiments. For studies comparing multiple groups, an ordinary one-way ANOVA analysis with Tukey's multiple-comparison test was performed. For behavioral study, a two-way ANOVA mixed-effect analysis with multiple-comparison Dunnett's test was used. In all graphs, the mean ± SEM is illustrated, and symbols depict individual samples per group. Details of the corresponding statistical analysis are included in the corresponding figure legends.
Study approval
Housing and breeding were in accordance with the IACUC at Homology Medicines and Taconic Biosciences. All procedures were performed in accordance with the IACUC at Homology Medicines. For Housing and all procedures on NHPs were in accordance with the Institutional Animal Care and Use Committee at the Mannheimer Foundation or SOPs of Charles River Laboratories.
Results
MLD is a rare, debilitating lysosomal disease with deleterious impact to the nervous system. We set out to design a gene therapy by identifying the best ROA and capsid combination to address the complex presentation of this disease. We focus primarily on the nervous system, the first organ severely impacted by this disease (Shaimardanova et al., 2020), to achieve sufficient ARSA activity to correct sulfatide metabolism while achieving disease-modifying benefits.
Intravenous delivery of HSC15/ARSA achieves an endogenous arsa-like CNS biodistribution in the arsa KO mice
Natural serotypes belonging to Clade F, such as AAV9 and AAVHSCs, have been demonstrated to have the most adequate profile when it comes to BBB penetration in larger species (Mendell et al., 2017; Ellsworth et al., 2019). Our approach consisted of identifying a capsid capable of delivering ARSA systemically with the lowest efficacious dose across the diseased cell types, including those of the CNS using a well-characterized mouse model for MLD (Rosenberg et al., 2016; Hess et al., 1996). We compared two natural Clade F serotypes known to cross the BBB that are either approved (Zolgensma, using AAV9) or currently being evaluated in the clinic (NCT05222178 and NCT05238324 for HSC15). Using a single intravenous delivery of AAV9/ARSA or HCS15/ARSA in adult Arsa KO mice, we confirmed that both cross the BBB and blood–nerve barrier (BNB; Fig. 1A,B). Yet, at the same dose, we observed that the levels and biodistribution of anti-ARSA detected across the entire nervous axis were significantly higher in HSC15/ARSA-treated mice when compared with AAV9/ARSA-treated mice. When comparing vgs (Fig. 1C), ARSA transcript (mRNA; Fig. 1D) and the resulting levels of ARSA activity (Fig. 1E) achieved in brain tissue, we observed that all three biochemical end points were significantly higher in HSC15/ARSA-treated mice brain samples, suggesting that HSC15/ARSA transduces the CNS more efficiently than AAV9/ARSA at the dose tested.
HSC15/ARSA biodistribution following intravenous administration in Arsa KO mice. A–B, Comparison of anti-ARSA biodistribution in brain, spinal cord, and DRG following a 4 week, single intravenous administration of HSC15/ARSA or AAV9/ARSA at 4E + 13 vg/kg compared with endogenous murine Arsa in adult WT mice. Individual points represent individual male mice (n = 4/group). C–E, Resulting respective vgs (C; mean ± SEM; unpaired t test with Kolmogorov–Smirnov correction; *p < 0.05), copies of ARSA transcript (D; mean ± SEM; unpaired t test; *p < 0.05), and brain ARSA activity (E; mean ± SEM; unpaired t test; ***p ≤ 0.0001) at study termination. Scale bars: 1000 µm (brain), 250 µm (spinal cord), and 100 µm (DRG; n = 2 repeats; *p < 0.05; ***p ≤ 0.0001; mean ± SEM). F, Anti-ARSA biodistribution in neuronal and glial-like profiles in the motor cortex (neurons and astrocytes), cerebellum (Purkinje cells), axonal tract (fornix; oligodendrocytes), choroid plexus (ependymal cells), and spinal cord (motoneurons and astrocytes). Representative images for each dosed group are shown. Scale bars: 125 µm (motor cortex), 50 µm (all other regions).
At the subcellular level, although both capsids successfully targeted neurons, glia (astrocytes), and ependymal cells of the choroid plexus (also of glial origin) as observed by their characteristic cellular profiles and cytoarchitecture (Fig. 1F), we observed that the biodistribution of the anti-ARSA immunoreactivity in the HSC15/ARSA-treated Arsa KO mice mimicked more closely that of the WT murine Arsa mice. Note that for AAVHSC15, the nature of the cell types transduced in these mice correlated with data published in NHPs dosed with the same capsid including peripheral and central neurons, astrocytes, and oligodendrocytes centrally, ependymal cells, and peripheral glial cells (satellite cells; Ellsworth et al., 2019). Furthermore, when examining motor cortex sections at a higher magnification, ARSA-transduced microglial-like profiles were more robust and numerous in HSC15/ARSA-treated Arsa KO mice (Fig. 2A, white arrowhead). Transduction of microglia is highly desirable in diseases of the nervous system, including MLD, where it is assumed to play an important role in the development of the disease (Bergner et al., 2019). Thus, to further confirm our in vivo murine observations, we investigated microglial transduction capability of HSC15 and AAV9/eGFP using iCell human microglia in culture. These commercially available cells exhibit functional characteristics like those described in human microglia including phagocytosis and cytokine-mediated inflammation and express relevant microglial markers (Abud et al., 2017). Vgs were detectable in cells treated with both capsids albeit at lower levels in AAV9-treated cells (Fig. 2B), yet only HSC15/eGFP-treated cells had significant eGFP mRNA expression (Fig. 2C), suggesting that HSC15 was superior at transducing and expressing in human microglial cells, further supporting the transduction capability and cellular biodistribution observed in these mice. Together, these data suggest that although either construct can successfully access the nervous system, HSC15/ARSA is superior at transducing more broadly brain tissue and associated cell types, including microglia, raising the possibility of achieving endogenous-like distribution, at lower doses and potentially resulting in efficacy.
HSC15/ARSA transduces murine microglia in vivo and human microglia cells in vitro. A, Insets and white arrowheads, Anti-ARSA biodistribution detected in microglial-like profiles in the motor cortex of HSC15/ARSA and AAV9/ARSA intravenous-treated Arsa KO mice at 4E + 13 vg/kg, as well as in neuronal-like profiles in HSC15/ARSA. Representative images for each dosed group are shown. Scale bar, 25 µm. B, HSC15/eGFP vgs in human microglia cell culture 72 h post-transduction (MOI 100 K) when compared with control untransduced cells and AAV9/eGFP-treated cells (mean ± SEM, 1-way ANOVA analysis, ****p ≤ 0.0001; ns = not significant; n = 3 repeats). C, Corresponding HSC15/eGFP transcript levels in human microglia cell culture when compared with control untransduced cells and AAV9/eGFP-treated cells (mean ± SEM, 1-way ANOVA analysis, ****p ≤ 0.0001; ns; n = 3 repeats).
Next, we assessed two different ROAs to restore a murine Arsa-like biodistribution in the brain following administration with HSC15/ARSA. We compared ARSA biodistribution following either intravenous or intrathecal delivery at doses that would lead to similar overall levels of ARSA activity in brain tissue in adult Arsa KO mice (average ARSA activity intravenous, 181.0 ± 45.1 ng pNC/mg protein/h and average ARSA activity intrathecal, 201.0 ± 61.5 ng pNC/mg protein/h (p value = 0.7951; n = 3 mice/group, respectively). For this, ARSA activity was evaluated in the midbrain and hindbrain of these mice to account for differential brain tropism and corresponding biodistribution. We confirmed that both ROAs led to ARSA protein in the CNS (Fig. 3A), albeit with a different biodistribution, and liver tissue (Fig. 3B). When compared with WT mice, the ARSA biodistribution in the intravenously dosed Arsa KO cohort resembled more closely that of the endogenous Arsa than that obtained in intrathecally treated Arsa KO mice (Fig. 3A). The intravenous HSC15/ARSA-treated Arsa KO mice adopted a more uniform brain ARSA biodistribution across the rostrocaudal and dorsoventral axes when compared with the outside-in pattern obtained in intrathecally dosed mice accompanied by a cerebellum bolus (Fig. 3A). At the subcellular level, the biodistribution in neuronal- and glia-like cell types was broader in the intravenous-treated Arsa KO mice (motor and sensory cortex, the hippocampus and putamen), with the exception of the cerebellar tissue, which showed the highest (supraphysiological in both anti-ARSA intensity and resulting ARSA activity) transduction in intrathecally treated mice (Fig. 3C). Together, these data suggested that the intravenous delivery of HSC15/ARSA best addresses the complex cell type and CNS biodistribution of this broadly expressed enzyme and is further characterized here on.
Biodistribution of human ARSA protein following either intravenous or intrathecal administration of HSC15/ARSA. A, Anti-ARSA biodistribution in brain tissue following a 4 week, single intravenous (n = 4/group) or intrathecal (n = 4/group) administration of HSC15/ARSA at 4E + 13 vg/kg or 4E + 12 total vgs to adult male Arsa KO mice, respectively. Adult male WT and vehicle-treated Arsa KO mice were processed identically and shown as controls. Scale bars: 1000 µm. B, Anti-ARSA biodistribution in liver tissue following a 4 week, single intravenous or intrathecal administration of HSC15/ARSA in Arsa KO mice. Scale bars, 50 µm. C, Anti-ARSA biodistribution in neuronal and glial-like profiles in the motor and sensory cortex (neurons and astrocytes), hippocampus, putamen, and cerebellum (Purkinje cells). Representative images for each dosed group are shown. Scale bars: 125 µm (cortex), 50 µm (all other regions).
Normal levels of ARSA activity in the CNS and potential for systemic cross-correction are achieved following an intravenous administration of HSC15/ARSA
To better understand the translation between findings in a mouse model and the desired clinical outcome in humans, we determined the levels of arylsulfatase-A enzymatic activity in normal adult WT mice (Arsa) and human (ARSA) brain samples to better define the required levels of reintroduced ARSA required to achieve efficacy. The lowest desirable levels were set based on ARSA pseudodeficiency (PD) in individuals, clinically characterized by a significant decrease in ARSA activity (as low as 5–20% of normal levels) that remains sufficient for sulfatide degradation and maintenance of normal life (Patil and Maegawa, 2013; Doherty et al., 2019). This observation in asymptomatic ARSA-PD individuals suggests that reintroducing activity levels as low as 5–20% could be sufficient to rescue the phenotypes in Arsa KO mice. First, we observed that arylsulfatase-A activity levels in normal adult brain tissue were similar across these two species, 121.4 ± 6.3 ng pNC/mg protein/h (n = 14) and 123.9 ± 23.4 ng pNC/mg protein/h (n = 4) for mice and human, respectively). Next, following a single intravenous administration of HSC15/ARSA, we investigated the doses that would give rise to levels of ARSA activity ranging from PD to normal, in relationship to Arsa levels, while maintaining a biodistribution profile in brain tissue as observed in WT mice. A dose–response relationship for ARSA protein (Fig. 4), HSC15/ARSA vgs, transcript (mRNA), and activity (Extended Data Fig. 4-1) was observed in CNS tissue following a single intravenous administration of HSC15/ARSA ranging from embryonic day 0.5E + 13 to 6E + 13 vg/kg; where “E” stands for exponent. At the histologic level, although we observed transduction and resulting ARSA protein in both brain (Fig. 4A,B) and spinal cord (Fig. 4C) tissue at all doses tested, it appeared that levels and biodistribution of ARSA in the spinal cord preceded that in the brain, suggesting a caudorostral gradient in ARSA expression following an intravenous administration and/or a difference in the permeability of the blood–spinal barrier (BSB). This gradient was more noticeable in brain tissues from mice having received a lower dose of HSC15/ARSA (Fig. 4A,C). Similarly, at the lowest doses tested, microglia-like profiles were apparent (Fig. 4B, blue arrowheads), consistent with earlier findings in mice and cultured human microglia cells. Although this cell type continues to be transduced at higher doses of HSC15/ARSA, the dense neuropil anti-ARSA IHC makes it harder to resolve. Finally, we confirmed that HSC15/ARSA successfully crossed the BSB and BBB, at all the doses tested. These broad ARSA biodistribution patterns resulted in increasing levels of ARSA activity achieving supraphysiological levels at the highest dose evaluated (normal ARSA human brain activity levels, 159.28 ng pNC/mg protein/h). Together, these data demonstrate the potential to achieve sufficient levels of ARSA activity to address the CNS manifestations of MLD over a broad range of doses.
Dose–response in ARSA protein detection across the CNS 4 weeks after single intravenous administration of HSC15/ARSA in adult Arsa KO mice. A–C, Anti-ARSA biodistribution in brain (A–B), spinal cord, and (C) tissue across doses ranging from 0.5E + 13 to 6E + 13 vg/kg. Individual points represent individual mice (n = 5/group; males). Higher magnification of motor cortex tissue (B) highlights transduction in the brain at the lowest doses evaluated. Blue arrowheads point to microglial-like profiles. Representative images for each dosed group are shown. Scale bars: 2500 µm (low-magnification brain), 50 µm (high-magnification brain) and 250 µm (spinal cord). D, Anti-ARSA biodistribution in peripheral organs of the gastrointestinal tract following intravenous dosing of HSC15/ARSA at 4E + 13 vg/kg. Representative images for each dosed group are shown. Scale bar, 100 µm. E, Resulting ARSA activity levels in duodenum, colon, and liver (mean ± SEM). F, Resulting ARSA activity levels in serum of HSC15/ARSA-treated Arsa KO mice in comparison with age-matched vehicle-treated WT and Arsa-KO mice (mean ± SEM). Extended Data Figure 4-1 contains more information.
Figure 4-1
ARSA vgs, mRNA and enzyme activity in adult Arsa KO mice CNS following a single intravenous administration of HSC15/ARSA. HSC15/ARSA doses ranging from 0.5E + 13 vg/kg to 6E + 13 vg/kg were administered intravenously. A, ARSA vgs copies (mean ± SEM; Kruskal–Wallis with Dunn's multiple comparison test; *p ≤ 0.05, ***p ≤ 0.0005; unlabeled comparisons were not ns). ns, Not significant. B, ARSA transcript levels (mean ± SEM; Kruskal–Wallis with Dunn's multiple comparison test; *p ≤ 0.05, ***p ≤ 0.0001, ****p ≤ 0.0001; unlabeled comparisons were not ns). C, ARSA activity (mean ± SEM; Kruskal–Wallis with Dunn's multiple comparison test; *p ≤ 0.05, **p ≤ 0.0002, ***p ≤ 0.0001; unlabeled comparisons were not ns) were observed. Individual points represent individual mice (n = 5 males/group). Download Figure 4-1, EPS file.
Next, we evaluated the levels of ARSA activity in key peripheral organs and serum following a single intravenous administration of HSC15/ARSA. A dose of 4E + 13 vg/kg was chosen based on its ability to reach near normal cerebral levels of ARSA activity. Figure 4, D–F, demonstrates transduction in matrices beyond liver, in particular peripheral organs involved in the gastrointestinal tract, which are severely impacted in MLD patients (Kim et al., 2017; Almarzooqi et al., 2018). At the enzymatic levels, ARSA activity was confirmed in all serum and peripheral organs evaluated, with the highest levels detected in liver tissue (Fig. 4E,F) supporting the prospect of addressing broad and peripheral manifestation of the disease as well as the potential for cross-correction in tissues with limited biodistribution of HSC15/ARSA.
HSC15/ARSA improves behavioral motor deficit in arsa KO mice
Although lacking the demyelinating features seen in MLD patients (Saravanan et al., 2004), Arsa KO mice develop a mild motor deficit that has been described by multiple independent groups (Hess et al., 1996; Consiglio et al., 2001; Wittke et al., 2004; Biffi et al., 2006; Eckhardt et al., 2007; Sevin et al., 2007). We extended our evaluation of the effect of HSC15/ARSA on behavioral motor function in Arsa KO mice with the goal of identifying the level of ARSA activity required to address this motor deficit by performing a longitudinal dose-range-finding study in presymptomatic HSC15/ARSA-treated Arsa KO mice. Dose ranges led to ARSA activity levels from low to supraphysiological. First, in a ramping rotarod assay, we confirmed the establishment of a mild motor deficit at ∼9 months of age (Fig. 5A–C, open circles), demonstrated by a lower latency to fall on a classic 4–40 rpm ramping rotarod protocol. Interestingly, at that age both Lamp1 and sulfatides levels are elevated in the CNS, whereas neuroinflammation (astrogliosis and microgliosis) and cerebellar Purkinje neuronal loss have not yet occurred (Hess et al., 1996; Sevin et al., 2006, 2007; Zerah et al., 2015; Audouard et al., 2021), indicating that the latter phenotypes are not the driving factors leading to the mild motor deficit observed. The mild motor deficit, illustrated by a shortened latency to fall observed in vehicle-treated Arsa KO, mice was ameliorated, and within range of WT performance, in HSC15/ARSA-treated mice at all doses tested (Fig. 5B,C), reaching statistical significance at the two highest doses of 4E and 6E + 13 vg/kg. Although the mid-dose of 2E + 13 vg/kg of HSC15/ARSA evaluated trended with the higher-dosed groups, it did not reach statistical significance over the course of this study. The study was terminated at 11 months of age, when Lamp-1 levels and neuronal sulfatide levels are known to be elevated in brain, neuroinflammation has not yet emerged, and Purkinje cells are intact in Arsa KO mice tissue (Sevin et al., 2006, 2007; Piguet et al., 2012). Moreover, any potential confounds associated with aging in mice (≥12 months of age), independent of the group genotypes, were also eliminated. At study termination, ARSA activity was determined in all groups of mice. As shown (Fig. 5D), the prevention of the motor deficit was observed in groups with ARSA activity levels nearing ∼50% of WT mouse levels (and higher), supporting the potential of HSC15/ARSA to prevent the development of motor deficit over a broad range of doses, albeit higher than that predicted based on ARSA activity levels in asymptomatic ARSA-PD individuals.
Motor improvement in Arsa KO mice following a single intravenous administration of HSC15/ARSA. A, Significant motor deficit was detected in vehicle-treated adult Arsa KO mice at 9 months of age when compared with age-matched WT littermates (mean ± SEM; unpaired 2-tailed student t test; *p < 0.01). B, C, Rescue of motor deficit in HSC15/ARSA-treated Arsa KO mice at all doses assessed, where B corresponds to the last time point tested (10 months) in C. Statistical significance was achieved at doses of 4 and 6E + 13 vg/kg (mean ± SEM; 2-way ANOVA mixed-effect analysis with multiple-comparison Dunnett's test; *p < 0.05). Lower dose evaluated was excluded from graph for clarity purposes. Gray dashed line corresponds to the average latency to fall of the WT group across the last three reads. Individual points represent individual mice (n = 8–12/group). D, Corresponding brain Arsa/ARSA activity levels in vehicle- and HSC15/ARSA-treated Arsa KO mice at study termination. 100% on the y-axis (right) represents normal levels of ARSA activity in brain samples of normal adult human samples.
Durability of intravenous administered HSC15/ARSA and impact of age at dosing in brain and liver tissue
In mice it has been demonstrated that androgen differences between males and females significantly influence hepatocyte tropism across a broad range of serotypes (Davidoff et al., 2003; Paneda et al., 2009). Indeed, higher transduction was reported in adult male hepatocytes compared with adult females while sparing other nonhepatic tissues (Davidoff et al., 2003). Thus, the impact of gender on the durability of HSC15/ARSA administered intravenously in adult Arsa KO mouse brain and liver tissues was assessed. In the brain, we observed sustained levels of ARSA activity, as early as 1 week postdose (earliest time point evaluated) and out to 52 weeks postdose (end of study), with no statistical difference between males and females (Fig. 6A; Extended Data Fig. 6-1). In liver tissue, supraphysiological ARSA activity levels were detected as early as 4 weeks postdose (earliest time point evaluated) and out to 12 weeks postdose, with a decrease observed in samples from the females from the 4 and 8 week postdosing cohorts when compared with the age-matched and dosing-matched male cohorts (Fig. 6B), consistent with the literature (Davidoff et al., 2003; Paneda et al., 2009).
Durability of ARSA activity in brain and liver tissue in adult and neonate Arsa KO mice. A, B, Brain and liver tissue of adult Arsa KO mice following intravenous administration of HSC15/ARSA at 4E + 13 vg/kg, respectively. Males are depicted as black circles and females as gray circles. Individual points represent individual mice (n = 6/group, equal number of males and females). Gender differences in liver ARSA activity were observed in adult liver tissue at 4 and 8 weeks postdosing (*p ≤ 0.02). C, D, Brain and liver tissue of neonate Arsa KO mice following intravenous administration of HSC15/ARSA at 4E + 13 vg/kg. Males are depicted as black circles and females as gray circles. Individual points represent individual mice (n = 6/group, equal number of males and females). Extended Data Figure 6-1 contains more information.
Figure 6-1
ARSA protein in the CNS of HSC15/ARSA-treated adult and neonate Arsa KO mice. Anti-ARSA biodistribution in brain and spinal cord tissue was determined at different time points in adult (n = 5–6/group, equal numbers of males and females) and neonate (n = 6/group, equal numbers of males and females) Arsa KO mice following a single intravenous dosing of HSC15/ARSA of 4E + 13 vg/kg. Boxed regions highlight matching time points postdosing of HSC15/ARSA. Representative images for each group are shown. Scale bars: 5,000 µm (brain) and 1,000 µm (spinal cord). Download Figure 6-1, EPS file.
The difference in expression observed in hepatic tissue between gender raises a limitation in gene therapy with regard to our goal of addressing, in addition to the nervous tissue, the somatic tissues also affected. Indeed, gene therapy relies primarily on episomal expression of the reintroduced transgene. In mice, there is an active growth phase that takes place over the course of the first 4 postnatal weeks followed by a steady-state plateau (Nomura, 1976). To assess episomal loss during somatic tissue growth and the impact of gender, we evaluated ARSA enzymatic activity levels in HSC15/ARSA-treated neonate male and female Arsa KO mice. In postmitotic brain tissue, we observed sustained near-normal levels of ARSA activity (adult normal human brain, 129.1 ng pNC/mg protein/h), as early as 1 week postdose (earliest time point evaluated) and out to 12 weeks postdose (end of study) with no detectable loss in expression or impact on gender (Fig. 6C; Extended Data Fig. 6-1). At the histologic level, we also reproduced the broad and murine-like Arsa CNS biodistribution observed in adult treated Arsa KOs (Extended Data Fig. 6-1).
Unlike brain tissue, the liver is expected to grow substantially during the first postnatal month (Nomura, 1976). In liver tissue we observed supraphysiological ARSA activity levels at 1 week postdose followed by a ninefold decline, leading to a steady-state plateau by 3 weeks postdose and to the end of the study (Fig. 6D). To better define the biological impact of residual ARSA activity levels in neonate hepatic tissue, we compared the resulting hepatic ARSA activity levels in HSC15/ARSA-treated neonate mice to those of healthy normal human liver tissue (normal adult liver, 483.96 ng pNC/mg protein/h) and determined that it reached near-normal human range and remained above normal murine Arsa levels (adult mouse liver, 123.00 ng pNC/mg protein/h; Fig. 6D). Interestingly, no gender differences were observed in ARSA activity in neonate liver tissue samples (Fig. 6D), further supporting that hormonal levels in sexually mature mice are most likely responsible for the observed differences in transduction reported in the literature for this species and observed in our studies (Fig. 6B).
Together, these data suggest that regardless of the age at dosing the CNS biodistribution of ARSA protein achieves a murine-like biodistribution and remains constant throughout the duration of the study. Moreover, contrary to postsynaptic tissue such as the brain, the dosing age in somatic tissue, such as the liver, has an impact on the resulting levels of ARSA activity, albeit a supraphysiological plateau of expression is reached following the active tissue growth phase.
Rescue of cerebral biomarkers across a caudorostral gradient following intravenous administration of HSC15/ARSA
Biomarkers play a critical role in guiding clinical diagnosis, estimating disease risk, evaluating disease stage, and monitoring progression or response to therapy. We evaluated the effect of HSC15/ARSA on three key CNS biomarkers, including levels of sulfatides (Hess et al., 1996; Sevin et al., 2006; 2007); lysosomal associated membrane protein 1 (Lamp1), indicative of lysosomal burden (Meikle et al., 1997; Wright et al., 2018); and Gfap, reflective of the overall astrogliosis (Frati et al., 2018), with the goal of determining the biomarker(s) most predictive of disease progression and therapeutic impact.
Sulfatide isoform levels are well-established clinical biomarkers for MLD. Here, we investigated sulfatide modulation in two treated Arsa KO mouse cohorts receiving HSC15/ARSA at different ages over the progression of the disease, including both neuronal (C18:0 and C18:1) and glial (C24:0 and C24:1) sulfatide isoforms; where C stands for shortened fatty acid chain length. The first cohort received treatment at what is believed to be a presymptomatic age/stage in Arsa KO mice [early treatment (ET), 1.5–2 months of age at time of dosing; 52-week-duration study; Fig. 7A,B] at a dose of 4E + 13 vg/kg, whereas the second received treatment at a later time point [late treatment (LT), ∼4.5 months of age at time of dosing; 24-week-duration study; Fig. 7C,D] at 0.5E + 13 and 4E + 13 vg/kg. Neuronal sulfatide isoforms (C18:0 and C18:1) remained elevated across the CNS axis in vehicle-treated Arsa KO mice in all CNS regions evaluated across the groups (Fig. 7, open circles). A significant reduction in CNS neuronal sulfatides reaching levels from those measured in age-matched WT tissue was observed in both HSC15/ARSA-treated Arsa KO age groups (Fig. 7). Of interest in the LT cohort, although the lowest dose of HSC15/ARSA was insufficient to modulate neural sulfatide accumulation in the brain, it significantly reduced it to WT levels in the spinal cord, suggesting that spinal sulfatide modulation precedes that of brain tissue and/or requires much lower levels of active ARSA, regardless of the disease progression state. Levels of glial sulfatide (C24:0 and C24:1) were not significantly elevated in brain tissue of ET or LT vehicle-treated Arsa KO mice when compared with those observed in WT littermates (data not shown), possibly in part because of technical limitations (sensitivity of our assay) and/or the lack of demyelination reported in these mice (Saravanan et al., 2004).
Rescue of neuronal sulfatide isoforms (C18:0 and C18:1) across the CNS axis in HSC15/ARSA-treated adult Arsa KO mice. B–E, Combined schematic illustrating the two age groups evaluated in separate studies. ET (B, C) was dosed at 6–8 weeks of age (at the onset of Lamp1 accumulation in the brain; Hess et al., 1996), and the study was conducted 52 weeks out, when the mice reached ∼14 months of age. The LT (D, E) group were dosed at ∼4.5 months of age (after the reported accumulation of sulfatides in the brain; Hess et al., 1996), and the study was conducted 24 weeks out, when the mice reached 11 months of age. B, C, Fifty-two weeks postintravenous administration of HSC15/ARSA at 4E + 13 vg/kg, neuronal sulfatide levels (C18:0 and C18:1) were significantly reduced, down to WT levels in (A) forebrain (**p ≤ 0.01) and (B) hindbrain tissue (**p ≤ 0.01, ***p ≤ 0.001; ns, Not significant; mean ± SEM; ordinary 1-way ANOVA with Tukey's multiple-comparison test). Individual points represent individual male and female mice (n = 5–6/group), where forebrain and hindbrain values were obtained from each mouse in the study. D, E, Twenty-four weeks postdosing of HSC15/ARSA (at 0.5 E + 13 or 4E + 13 vg/kg) in adult male Arsa KO mice, neuronal sulfatide levels (C18:0 and C18:1) were reduced to WT levels in (D) forebrain tissue at a dose of 4E + 13 vg/kg (C18:0, **p ≤ 0.01; ****p ≤ 0.0001 C18:1, *p ≤ 0.05; **p ≤ 0.01; unlabeled comparisons, ns) but ns at 0.5E + 13 vg/kg. In spinal cord tissue (E), neuronal sulfatide levels were substantially ameliorated at both doses, with a greater impact (full rescue) of C18:1 over C18:0 (C18:0, *p < 0.05, **p ≤ 0.01, ****p ≤ 0.0001; C18:1, ****p ≤ 0.0001; unlabeled comparisons were not ns; mean ± SEM; ordinary 1-way ANOVA with Tukey's multiple-comparison test). Individual points represent individual mice (n = 8–10/group), where forebrain and spinal cord values were obtained from each mouse on the study. Extended Data Figure 7-1 contains more information.
Figure 7-1
LAMP1 and GFAP modulation in the CNS of treated adult Arsa KO mice 12 or 52 weeks postdose. A, B, Vehicle or HSC15/ARSA was administered intravenous at 4E + 13 vg/kg. LAMP-1 protein levels, detected by IHC in a cross section of the ventral spinal cord at 12 weeks (A) and 52 weeks (B) were significantly reduced in treated mice when compared with vehicle-treated control mice. A, dWM, p = 0.0011; vWM, p = 0.0059; vGM, p = 0.0254); B, **p = 0.0030, **p = 0.0003, ****p < 0.0001; mean ± SEM; 1-way ANOVA with Tukey's multiple comparisons test). C, D, GFAP protein levels, detected by IHC in a cross section of the ventral spinal cord, were reduced to WT levels in spinal cord tissue (C) in HSC15/ARSA treated mice (*p = 0.0107, ***p = 0.0003); ns, Not significant (D) of HSC15/ARSA treated mice (*p = 0.0166, **p = 0.0047, ****p < 0.0001; mean ± SEM; 1-way ANOVA with Tukey's multiple comparisons test). Individual points represent individual sections analyzed (n = 2 sections/mouse and 3 mice/group). Download Figure 7-1, EPS file.
Lysosomal burden and astrogliosis were also evaluated in these mice. The analysis was aimed at the ET-treated group based on the lack of neural sulfatide correction in the brain of LT-treated mice and on the study duration of the latter, ending before the onset of astrogliosis. Indeed, astrogliosis has been reported to occur in the second year of life in this animal model (Hess et al., 1996). In brain tissue of HSC15/ARSA-treated Arsa KO mice (ET), we observed an amelioration in Lamp1 protein accumulation as early as 12 weeks postdosing (earliest time point evaluated) followed by a full rescue (indistinguishable from those in age-matched WT mice) by study termination (Extended Data Fig. 7-1). Next, we evaluated astrogliosis and determined that it was fully prevented in spinal tissue (Extended Data Fig. 7-1) and ameliorated in brain tissue (fimbria; Extended Data Fig. 7-1) of HSC15/ARSA-treated Arsa KO mice at study termination, highlighting a caudorostral gradient in biomarker rescue (spinal cord before brain) consistent with that seen herein for biodistribution, ARSA expression, and sulfatide modulation.
Transduction in NHPs treated with HSC15/ARSA intravenously
Translation of gene therapies across higher-order species is of critical importance to inform clinical trial designs. We evaluated the systemic and nervous system (PNS and CNS) biodistribution of HSC15/ARSA-V5 (encoding ARSA fused to a C-terminal V5-tag) following a single intravenous administration to 8- to 13.5-month-old NHPs and determined whether the broad distribution across cell types observed in mice was maintained. Disease-relevant tissues were selected, including the dorsal root ganglia (DRGs), spinal cord, brain, and a range of peripheral organs.
ARSA-V5 was detected in ganglia [trigeminal ganglion (TG) and DRG] of the PNS (Fig. 8A; Extended Data Fig. 8-1), brain (highlighted by choroid plexus, motor cortex, cerebellum, and putamen) and spinal cord (Fig. 8B–F), as well as in multiple peripheral organs (liver, heart, lung, kidney (Extended Data Fig. 8-1), indicating that HSC15/ARSA-V5, similarly to HSC15/ARSA distribution in Arsa KO mice, distributes to disease-relevant cell types and tissues of the nervous system in NHPs.
ARSA protein immunofluorescence dose-dependent increases in ARSA protein in liver and brain tissue following a single intravenous injection in NHPs. A–F, ARSA was detected (via V5 tag) in neuronal and glial-like cell profiles in DRG (A), spinal cord (B), choroid plexus (C), motor cortex (D), cerebellum (E), and putamen (F). Anti-ARSA is in light blue, and anti-V5 is in magenta (n = 2 monkeys per group; HSC15/ARSA-V5 and vehicle). Representative images for each group are shown. Scale bars: A, 200 µm; B, 25 µm; C, F, 50 µm; D, E, G, 100 µm. H, In brain tissues collected from level 5b, ARSA protein levels in NHP-administered vehicle or HSC15/ARSA at 6E + 13 vg/kg and 1E + 14 vg/kg. Individual points represent individual monkeys (n = 4/group); where no symbol is shown, the value for the individual was below the limit of detection. I, Plasma levels of ARSA protein in NHPs following a single intravenous administration of HSC15/ARSA. Individual points represent individual monkeys (n = 4/group); where no symbol is shown, the value for the individual was below the limit of detection. NHPs in the vehicle group did not have ARSA protein levels above the limit of quantitation (30.0 ng/ml). J, Serum levels of ARSA enzyme activity in NHPs following a single intravenous administration of HSC15/ARSA. Bars indicate group mean, and individual points represent individual monkeys (n = 4/group); where no symbol is shown, the value for the individual was below the limit of detection. NHPs in the vehicle group did not have ARSA enzyme levels above the limit of quantitation (1.76 µg/ml). Extended Data Figures 8-1 and 8-2 contain more information.
Figure 8-1
ARSA immunostaining in HSC15/ARSA and vehicle treated-NHPs across a broad range of tissues. Human ARSA was detectable in PNS tissue (trigeminal ganglion), as well as multiple peripheral organs including kidney, heart, liver, and to a lower degree, lung of intravenously treated NHP at 4E = 13 vg/kg. Note that the anti-ARSA antibody used detects both NHP and human ARSA protein, thus explaining the detectable signal in healthy normal vehicle-treated NHPs. Scale bars: 1,000 µm (low magnification TG), 5,000 µm (low magnification all other organs), 50 µm (high magnification TG); 100 µm (high magnification liver), 500 µm (high magnification testis), and 250 µm (high magnification all other organs). Representative images for each group are shown. Download Figure 8-1, EPS file.
Figure 8-2
Levels of vgs and ARSA protein in liver and brain are significantly correlated at 28 days following a single intravenous administration of 6E + 13 and 1E + 14 vg/kg HSC15/ARSA. A, B, A two-tailed Pearson correlation demonstrated significant positive relationships between vg levels in the liver [r(6) = 0.87, p < 0.01] and (B) brain level 5b [r(6) = 0.81, p < 0.05]. Data points represent values for individual NHPs (n = 4/group) and line represents simple linear regression. Symbol x represents the single NHP that was determined to be positive for anti-HSC15 neutralizing antibodies prior to dosing. Download Figure 8-2, EPS file.
In a separate NHP study, levels of human ARSA protein (no tag) in liver, brain, and plasma were determined using specific quantitative hybrid LC-MS/MS 28 d following dosing with HSC15/ARSA. ARSA protein levels increased in a dose-dependent manner in the liver and brain (Fig. 8G,H). Interestingly, ARSA protein levels were higher in the most caudal section of the brain (level 5b, 29.93 ± 14.19 ng/mg vs level 4, 9.04 ± 3.32 ng/mg; Bolon et al., 2018), consistent with the caudorostral gradient observed in mice. ARSA protein in plasma (Fig. 8I) and ARSA activity in serum (Fig. 8J) increased in a dose-dependent manner following a single intravenous administration of HSC15/ARSA at 6E + 13 and 1E + 14 vg/kg on day 8 postdose. ARSA activity was detectable on day 15 but were generally undetectable by day 28 postdose, which coincided with the presence of anti-ARSA antibodies. Vg levels and ARSA protein levels were well correlated in both liver and brain sections (Extended Data Fig. 8-2).
Overall, these observations confirm that as seen in the MLD mouse model, HSC15/ARSA-V5 and HSC15/ARSA successfully crossed the BNB and BBB in a large animal model, was broadly distributed in the nervous system, and led to the generation of active ARSA enzyme in tissues as well as in circulation supporting the potential for cross-correction.
Discussion
Herein, we explore in vivo gene therapy approaches to address the manifestations of MLD in the Arsa KO mouse model. Our findings demonstrate the prevention of motor deficits, restoration of ARSA CNS biodistribution, accompanied by normalization of disease biomarkers following a single intravenous administration of an HSC15 serotype expressing ARSA.
LSDs often present with CNS manifestations, in addition to the peripheral organ phenotypes. Although ERT has been used effectively in addressing organ function stabilization and/or slowing down disease progression, applicability is limited in patients with CNS dysfunctions as the recombinant enzymes cannot cross the BBB (Barton et al., 1991). This, compounded with a need for systemic expression, could be addressed by a gene therapy able to cross the BBB. Gene therapy leverages viral vectors derived from five main classes, adenovirus, AAVs, herpes simplex-1 viruses, retroviruses, and lentivirus (Puhl et al., 2019). The latter has made great progress in the clinic, in particular for the treatment of MLD. Libmeldy (Orchard Therapeutics) is a lentiviral-based ex vivo gene therapy reintroducing modified hematopoietic stem progenitor cells (HSPCs) in MLD children with late infantile or early juvenile forms. This ex vivo approach leverages genetically modified HSPCs, followed by bone marrow irradiation and transplantation via intravenous delivery. Although promising, the field continues to seek further improvements to address some of the limitations. Ex vivo gene therapies rely on myeloablation, and the engraftment of the modified reintroduced cells takes up to 3 months post-transplantation to achieve peak expression (Zijlmans et al., 1998; Srikanthan et al., 2020). Some AAVs can cross the BBB following an intravenous administration and have the potential of extending the biodistribution of the therapeutic agent in the brain while leveraging a less invasive delivery route (Audouard et al., 2021), but the translatability to larger species of the capsids evaluated remains to be determined in the clinic. Furthermore, because of their rapid onset, AAV gene therapy allows the possibility to tackle early symptomatic late infantile patients, currently excluded from trials.
Targeted AAV delivery in the brain is most commonly achieved via direct intracerebral injections. This method, albeit highly invasive, is preferred when the targeted region is limited and defined to a specific cerebral localization or if targeting deeper brain structures such as the thalamus and putamen are necessary (Hocquemiller et al., 2016; Taghian et al., 2020). In the field of MLD, despite long-lasting restoration of ARSA activity in the CSF, this clinical trial approach (using AAVrh.10) failed to demonstrate any effect (NCT01801709; Dali et al., 2020) while alleviating most long-term disease manifestations in MLD mice (Sevin et al., 2006, 2007; Piguet et al., 2012). The authors of the trial suggest that the intracerebral delivery was probably not enough to achieve sufficient ARSA activity in the CNS (Audouard et al., 2021), further supporting our goal of achieving an endogenous-like biodistribution that mirrors the expression of the native enzyme across the rostrocaudal axis of the brain to increase the chances of success in the clinic. The systemic intravenous route remains, to date, the least invasive for viral gene therapy and has the advantage of addressing the largest number of organs and cell types (Yang et al., 2014; Ellsworth et al., 2019; Belur et al., 2020; Chen, 2020). We prioritized and explored two Clade F AAV capsids (AAV9 and HSC15) currently in the clinic for CNS diseases, known to cross the BBB and BSB in multiple species, and selected the capsid and ROA leading to the broadest and highest biodistribution of ARSA across the CNS. We determined that at equal dose, following an intravenous delivery, HSC15/ARSA expressed significantly higher levels of active ARSA across the CNS axis, suggesting that therapeutic benefits could be observed at a lower dose. Although it is possible that AAV9 could have reached equivalent levels of activity at a higher dose, the need for a rapid onset and a high level of active ARSA protein at the lowest possible dose led to the selection of HSC15/ARSA. Moreover, when compared head to head with a well-documented direct CNS delivery route (intrathecal) at comparable overall levels of active enzyme, the biodistribution of corresponding HSC15/ARSA protein across the CNS mirrored that of the endogenous mouse distribution in the intravenously treated Arsa KO mice. Moreover, the characterized robust and broad biodistribution of HSC15/ARSA across the rostrocaudal axis of the CNS administered intravenously in mouse and NHP demonstrate the potential of accessing deep brain structures originally limited to intracerebral dosing (Hocquemiller et al., 2016); Ellsworth et al., 2019; Taghian et al., 2020), while offering a broader tropism to access the phenotypic presentation of peripheral organs affected in MLD patients, such as the gastrointestinal tract. Finally, cross-correction through reuptake via the mannose-6-phosphate receptor pathway is a well-understood biological feature of LSDs (for review, see Rastall and Amalfitano, 2015; Rosenberg et al., 2016). We also demonstrated the presence of active ARSA enzyme in the serum of HSC15/ARSA-dosed Arsa KO mice and healthy NHPs, supporting the additional potential for cross-correction following a single intravenous administration.
Levels of ARSA enzyme activity necessary to modulate key CNS biomarkers and behavioral deficits in the clinic remain an unresolved question in the field. Many have based their targeted therapeutic levels on the reported levels of ARSA activity in ARSA-PD individuals. Given the reported correlation between residual ARSA enzyme activity and the severity of clinical presentation (Dali et al., 2015), a therapy designed to restore ARSA activity to levels near or above a threshold of ∼5–20% of normal adult human brain ARSA levels, as seen in ARSA-PD samples, would be anticipated to have the potential to benefit all clinical subtypes of MLD. To better understand the relationship between CNS neuronal sulfatides and motor deficit, we assessed the motor performance of presymptomatic Arsa KO mice following an intravenous administration of HSC15/ARSA. We determined that ARSA activity levels >50% of normal were required to prevent the mild motor deficit in this MLD mouse model. In agreement with this observation, transplantation of WT cells via HSCT in Arsa KO mice did not achieve significant benefit, further supporting a critical role of enzyme overexpression (Biffi et al., 2006). Furthermore, we observed that although near-normal levels of ARSA activity can successfully prevent cerebral neuronal sulfatide (C18:0 and C18:1) accumulation, lower levels (closer in range to ARSA-PD) can modulate neuronal sulfatides in the spinal cord. These data suggest that the enzyme activity levels required to modulate this key disease-associated biomarker across the entire CNS (brain and spinal cord) are much higher than the previously targeted ARSA-PD levels proposed in the literature and that modulation of neuronal sulfatides across the entire axis is required to predict motor benefits in this MLD mouse model.
A similar caudorostral gradient in biomarker modulation was observed for lysosomal burden and astrogliosis modulation. Of note, the reduction seen in the cerebral tissue evaluated in the deep regions of the brain (fimbria) may reflect the presence of an additional gradient in AAV diffusion that would require more evaluation to further elucidate. Nevertheless, although a rescue in lysosomal burden correlated with in vivo behavioral efficacy, an amelioration in astrogliosis was sufficient, suggesting that this neuroinflammation component is not directly linked with the observed motor deficits. Indeed, the later onset of this biomarker well passed the onset of the reported motor deficits is in agreement with this hypothesis. One major limitation in the preclinical data generated in the mouse model across the field is the incomplete parallel between the clinical presentation in patients and the phenotypes characterized in the MLD mouse model, of note a milder overall phenotype, absence of demyelination in the PNS and CNS, and the absence of disease impact on the Arsa KO mouse expectancy (Hess et al., 1996). On the one hand, the data gathered in this mouse model, including sulfatide accumulation, neuroinflammation (astrogliosis and microgliosis), and neuronal cell death (e.g., cerebellar Purkinje neurons) at different stages of development, may better reflect more accurately early stages of the human disease and/or pinpoint species differences in the metabolic pathways of the storage compound (Hess et al., 1996; Sango et al., 1995).
In summary, we have designed an intravenous HSC15/ARSA gene therapy vector capable of rapidly crossing the BNB, BSB, and BBB, reaching sustained, normal ARSA activity levels leading to a prevention of motor deficits in treated adult Arsa KO mice. The ARSA activity levels achieved, compounded with its broad endogenous-like cellular biodistribution along the brain rostrocaudal axis, correlated with a sustained normalization of key biomarkers to levels indistinguishable from those in age-matched WT controls. In addition, broad nervous system distribution of HSC15/ARSA in NHPs, along with the presence of enzymatically active ARSA across numerous tissues and serum, support the translation to larger species. Thus, a single intravenous delivery of HSC15/ARSA is a promising approach to treating all aspects of this genetic disease, including those affecting the nervous system of MLD patients.
Footnotes
This work was supported by Homology Medicines.
All authors are past or present remunerated employees of Homology Medicines.
- Correspondence should be addressed to Jacinthe Gingras at jgingras{at}homologymedicines.com
