Gene therapy for cross-correction of somatic organs and the CNS in mucopolysaccharidosis II in rodents and non-human primates

Abstract

Mucopolysaccharidosis II (MPS II) is a rare lysosomal storage disease characterized by deficient activity of iduronate-2-sulfatase (I2S), leading to pathological accumulation of glycosaminoglycans (GAGs) in tissues. We used iduronate-2-sulfatase knockout (Ids KO) mice to investigate if liver-directed recombinant adeno-associated virus vectors (rAAV8-LSP-hIDS_co) encoding human I2S (hI2S) could cross-correct I2S deficiency in Ids KO mouse tissues, and we then assessed the translation of mouse data to non-human primates (NHPs). Treated mice showed sustained hepatic hI2S production, accompanied by normalized GAG levels in somatic tissues (including critical tissues such as heart and lung), indicating systemic cross-correction from liver-secreted hI2S. Brain GAG levels in Ids KO mice were lowered but not normalized; higher doses were required to see improvements in brain histology and neurobehavioral testing. rAAV8-LSP-hIDS_co administration in NHPs resulted in sustained hepatic hI2S production and therapeutic hI2S levels in cross-corrected somatic tissues but no hI2S exposure in the central nervous system, perhaps owing to lower levels of liver transduction in NHPs than in mice. Overall, we demonstrate the ability of rAAV8-LSP-hIDS_co to cross-correct I2S deficiency in mouse somatic tissues and highlight the importance of showing translatability of gene therapy data from rodents to NHPs, which is critical for supporting translation to clinical development.

Graphical abstract

Chen and colleagues show that gene therapy for mucopolysaccharidosis II through administration of a liver-directed recombinant adeno-associated virus vector encoding human iduronate-2-sulfatase (rAAV8-LSP-hIDS_co) is able to cross-correct iduronate-2-sulfatase deficiency in somatic tissues of Ids knockout mice and demonstrate promising translational potential from the mouse model to non-human primates.

Introduction

Mucopolysaccharidosis II (MPS II; or Hunter syndrome [OMIM: 309900]) is a rare, X-linked, progressive, lysosomal storage disease.¹ The disease is caused by pathogenic variants of the iduronate-2-sulfatase gene (IDS), which encodes iduronate-2-sulfatase (I2S), a lysosomal enzyme responsible for catalyzing the hydrolysis of C2-sulfate groups of the glycosaminoglycans (GAGs) dermatan sulfate (DS) and heparan sulfate (HS). Deficient activity of I2S leads to pathological accumulation of GAGs throughout the body.²

Although a continuum of different phenotypes can be observed, MPS II is generally classified into two clinical forms: neuronopathic and non-neuronopathic.¹ Somatic clinical disease manifestations are present in all patients and can include organ failure, coarsening of facial features, bone and joint abnormalities, and short stature.³ Neuronopathic disease affects approximately two-thirds of patients with MPS II; in addition to somatic disease manifestations, these patients also experience profound central nervous system (CNS) involvement and cognitive impairment.^1,3

The current standard of care for MPS II is enzyme replacement therapy (ERT) with intravenous recombinant human I2S (hI2S; idursulfase, Takeda Pharmaceuticals USA, Lexington, MA, USA).¹ ERT with intravenous idursulfase is effective at improving or stabilizing many somatic disease manifestations⁴; however, it has some limitations, including fluctuating enzyme levels, the patient burden of weekly infusions, and limited penetration into some deep somatic tissues such as heart, lung, and bone.^5,6,7 Moreover, the high molecular weight (76 kDa) of hI2S and the absence of an active mechanism to cross the blood-brain barrier means that the clinical dose given systemically does not provide exposure to the brain to address neurological symptoms.^5,6

MPS II is a monogenic disease with relatively well-known pathological mechanisms, making it a potential candidate for gene replacement therapy.⁸ Recombinant adeno-associated virus (rAAV) is considered the delivery tool of choice for in vivo gene therapy because it is generally well tolerated, can provide long-term expression, and shows low immunogenicity.⁹ Gene therapy with rAAV-based vectors offers the possibility of long-term sustained expression of hI2S and the potential to normalize GAG levels in certain deep somatic tissues in patients with MPS II.

The successful development of intravenous ERT for MPS II included use of an integrated approach involving pharmacokinetic, biodistribution, and safety studies in rodents and non-human primates (NHPs).¹⁰ We developed a similarly integrated approach to investigate gene therapy for MPS II, using appropriate mouse models together with NHP studies to support the identification of a robust preclinical pharmacology framework for gene therapy optimization and translation into the clinic.

Here, we describe rAAV8-LSP-hIDS_co, a gene therapy vector comprising an adeno-associated virus serotype 8 capsid and a genome containing a liver-specific expression cassette encoding hI2S, using the liver as the in vivo hI2S production center (Figure 1A). We evaluated the ability of rAAV8-LSP-hIDS_co gene therapy to cross-correct I2S deficiency in somatic and CNS tissues of Ids knockout (Ids KO) mice and the translation of these mice data to NHPs.

Figure 1.

Vector schematic and Ids KO mice pharmacology 3 months after treatment with rAAV8-LSP-hIDS_co

(A) rAAV8-LSP-hIDS_co vector schematic. (B–G) Ids KO mice were treated with a single intravenous injection of rAAV8-LSP-hIDS_co at four dose levels ranging from 2.5 × 10⁸ vg/kg to 2.5 × 10¹¹ vg/kg and monitored for 3 months. An additional cohort was dosed with 2.5 × 10¹¹ vg/kg of a null vector, rAAV8-MY011, as a negative control. (B) Serum and (C) somatic tissues and the CNS were assayed for I2S activity. (D) Somatic tissues and the CNS were assayed for HS content and (E) DS content. (F) Somatic tissues and (G) the CNS were assayed by IHC for LAMP1 positivity. On all plots, data are presented as mean ± SD, n = 8 per group, except for the naive WT group (n = 5). p values for rAAV8-LSP-hIDS_co-treated animals compared with naive Ids KO were generated by one-way analysis of variance (ANOVA) with multiple comparisons using Dunnett’s correction. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ns, not significant versus naive Ids KO. ANOVA, analysis of covariance; CNS, central nervous system; DS, dermatan sulfate; I2S, iduronate-2-sulfatase; hI2S, human iduronate-2-sulfatase; hIDS_co, codon-optimized human iduronate-2-sulfatase gene; HS, heparan sulfate; Ids KO, iduronate-2-sulfatase gene knockout; IHC, immunohistochemistry; ITR, inverted terminal repeat; LAMP1, lysosome-associated membrane protein 1; LSP, liver-specific promoter; Poly A, polyadenylation; rAAV8, recombinant adeno-associated virus 8; vg, vector genome; WPRE, woodchuck hepatitis virus post-transcriptional regulatory element; WT, wild-type.

Results

Effect of rAAV8-LSP-hIDS_co on hI2S expression and disease-associated parameters at 3 months after dosing in Ids KO mice

To assess the ability of rAAV8-LSP-hIDS_co to cross-correct I2S deficiency, Ids KO mice (5–7 weeks of age) were treated with four rAAV8-LSP-hIDS_co doses ranging from 2.5 × 10⁸ vector genome (vg)/kg to 2.5 × 10¹¹ vg/kg via tail vein injection. An additional Ids KO mice cohort was dosed with a null vector, rAAV8-MY011 (2.5 × 10¹¹ vg/kg), as a negative control. Age-matched naive Ids KO and wild-type (WT) mice were also included as controls.

At 3 months after dosing, I2S activity and hI2S concentration in serum and tissues had a dose-dependent increase in mice treated with rAAV8-LSP-hIDS_co (Figures 1B, 1C, S1, and S2). As expected, tissue I2S activity in rAAV8-MY011-treated KO mice (<1 nmol/h/mg) was similar to that in naive Ids KO mice. I2S activity in the group treated with the lowest dose (2.5 × 10⁸ vg/kg) was similar to that in naive Ids KO mice in all tissues except for liver. The rAAV8-LSP-hIDS_co dose groups had increasing I2S activity with increasing dose (2.5 × 10⁹ vg/kg to 2.5 × 10¹¹ vg/kg) in liver, spleen, kidney, heart, lung, bone marrow, quadriceps, skin, and brain tissues (Figure 1C). Differences in I2S activity between naive Ids KO mice and rAAV-LSP-hIDS_co-treated mice reached significance at different dose levels in tissues, namely at 2.5 × 10⁸ vg/kg or higher in the liver, at 2.5 × 10⁹ vg/kg or higher in the spleen, and at 2.5 × 10¹⁰ vg/kg or higher for all other somatic tissues except the lung, in which significance was achieved only with the 2.5 × 10¹¹ vg/kg dose. In Ids KO mice treated with the 2.5 × 10¹¹ vg/kg dose, I2S activity was greater in all somatic tissues than in WT mice, except for the brain, where activity was 10% of that in WT mice.

To quantify HS and DS, GAG levels were measured in tissues using a liquid chromatography with tandem mass spectrometry (LC-MS/MS) assay non-conforming with Good Laboratory Practice (GLP) guidelines. HS and DS concentrations saw similar patterns of decrease with increasing rAAV-LSP-hIDS_co dose across all tissues except for the kidney (Figures 1D and 1E). In the kidney, DS concentrations for the 2.5 × 10¹⁰ vg/kg and 2.5 × 10¹¹ vg/kg doses decreased to a greater extent than HS; however, this did not affect statistical significance, which was equivalent for HS and DS. Significant HS and DS reductions were observed in all somatic tissues at 2.5 × 10¹⁰ vg/kg and 2.5 × 10¹¹ vg/kg doses compared with naive Ids KO mice. The 2.5 × 10¹⁰ vg/kg dose achieved over 90% HS and DS reduction in somatic tissues and the 2.5 × 10¹¹ vg/kg dose normalized levels to WT. In mice treated with the 2.5 × 10¹¹ vg/kg dose, brain HS and DS levels were approximately 60% and 20%, respectively, of those recorded in naive Ids KO mice (Figure S3). Immunohistochemistry (IHC) analyses were performed to assess changes in lysosome-associated membrane protein 1 (LAMP1), a lysosomal membrane marker that is elevated in lysosomal disease model animals.¹¹ In rAAV8-LSP-hIDS_co-treated mice, there was a significant dose-dependent reduction in detectable LAMP1 compared with naive Ids KO mice in somatic tissues (Figure 1F) but not in the brain (Figure 1G).

Effect of rAAV8-LSP-hIDS_co on hI2S expression and disease-associated parameters at 12 months after dosing in Ids KO mice

To evaluate the long-term effects of rAAV8-LSP-hIDS_co over 12 months, Ids KO mice were treated with three dose levels: 2.5 × 10¹¹ vg/kg, 1.25 × 10¹² vg/kg, and 6.25 × 10¹² vg/kg. Vector administration partially rescued the diminished weight gain seen in Ids KO mice (Figure S4). At 12 months after dosing, a similar dose-dependent pattern to that seen in animals at 3 months after treatment was observed for I2S activity in serum and tissue (liver, spleen, kidney, heart, lung, and brain). After a peak at 14 days after the injection, serum hI2S activity gradually decreased over 12 months (Figure 2A). In treated Ids KO mice, increases in I2S activity levels in somatic tissues and the brain were directly proportional to the dose received (Figure 2B). I2S activity levels in somatic tissues of treated Ids KO mice were greater than in WT mice at all doses (Figure 2C), and somatic tissue GAG levels were normalized to WT levels (Figure 2D). Interestingly, brain I2S activity levels in the 1.25 × 10¹² vg/kg and 6.25 × 10¹² vg/kg dose groups were 20% and 30% of those in WT mice, which resulted in significant decreases in HS and DS GAG brain levels compared with Ids KO mice. Cerebrospinal fluid (CSF) HS GAG levels were measured, and reductions compared with Ids KO mice were observed for all rAAV8-LSP-hIDS_co groups (Figure 2E). Reduction in LAMP1 in heart sections correlated with the presence of hI2S (Figures S5 and S6). Significant LAMP1 reduction was detected in all brain regions of Ids KO mice treated with 1.25 × 10¹² vg/kg and 6.25 × 10¹² vg/kg doses (Figures 2F and S7 for immunostaining in the cortex) compared with naive Ids KO mice, and, histologically, I2S was visible in the choroid plexus and meninges (Figures 2G and S8).

Figure 2.

Ids KO mice pharmacology 12 months after treatment with rAAV8-LSP-hIDS_co

(A–G) Ids KO mice were treated with a single intravenous injection of rAAV8-LSP-hIDS_co at three dose levels ranging from 2.5 × 10¹¹ vg/kg to 6.25 × 10¹² vg/kg and monitored for 12 months. (A) Serum and (B) somatic tissues and the CNS were assayed for I2S activity. (C) I2S enzyme activity fold change above WT levels in serum, somatic tissues, and the CNS 12 months after treatment. (D) Somatic tissues and the CNS were assayed for HS and DS content. (E) CSF was assayed for HS content. Bars represent pooled CSF from all mice in each dose group and lack error bars due to pooling. Due to lack of sufficient sample volume, the WT group could not be tested. (F) CNS assayed by IHC for LAMP1 positivity. (G) Naive Ids KO and high rAAV8-LSP-hIDS_co dose assayed by IHC for hI2S in the choroid plexus and meninges. On all plots, data are presented as mean ± SD, n = 9–12 per group, except for the naive WT group (n = 6). p values for rAAV8-LSP-hIDS_co-treated animals compared with naive Ids KO were generated by one-way ANOVA with multiple comparisons using Dunnett’s correction.∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus naive Ids KO. CSF, cerebral spinal fluid; GAG, glycosaminoglycan.

Effect of rAAV8-LSP-hIDS_co on functional parameters in Ids KO mice

For imaging and behavioral experiments in Ids KO mice, rAAV8-LSP-hIDS_co was evaluated at doses of 2.5 × 10¹¹ vg/kg, 1.25 × 10¹² vg/kg, and 6.25 × 10¹² vg/kg. Three-dimensional surface visualizations of micro-computed tomography (CT) images of the zygomatic arch showed similarities in bone structure and width between Ids KO animals treated with rAAV8-LSP-hIDS_co and WT animals (Figure 3A). For dense bone volume measurements of the femur, tibia, humerus, radius, and zygomatic arch, vehicle-treated Ids KO mice displayed larger volumes than WT animals. At 13.5 months after dosing, there were no significant differences between animals treated with rAAV8-LSP-hIDS_co and WT animals, indicating a normalization of bone-related changes for dense bone volumes even at the lowest dose (2.5 × 10¹¹ vg/kg) (Figure 3B).

Figure 3.

Ids KO mice functional evaluations after treatment with rAAV8-LSP-hIDS_co

(A–E) rAAV8-LSP-hIDS_co was evaluated at single intravenous doses of 2.5 × 10¹¹ vg/kg, 1.25 × 10¹² vg/kg, and 6.25 × 10¹² vg/kg in Ids KO mice. An additional cohort was dosed with 6.25 × 10¹² vg/kg of a null vector (rAAV8-MY011) as a negative control, and age-matched vehicle-treated Ids KO and naive WT mice were included as controls. (A) Three-dimensional surface visualization of micro-CT images of the mouse skull, with zygomatic arch highlighted in yellow from each group 13.5 months after rAAV8-LSP-hIDS_co treatment. Scale bar, 2 mm. (B) Micro-CT measurements of mineralized dense bone volume 13.5 months after rAAV8-LSP-hIDS_co treatment. (C) Left ventricular mass measured by echocardiography 12.5 months after rAAV8-LSP-hIDS_co treatment. (D) Latency to fall assessed using the accelerating RotaRod motor coordination test in treated Ids KO mice up to 12 months. (E) Y-maze neurobehavioral assessment in treated Ids KO mice up to 12 months. All data are presented as mean ± SD with n = 7–10 animals per group for (B) and (C) and n = 10–12 animals per group for (D) and (E). p values for rAAV8-LSP-hIDS_co-treated animals compared with naive WT mice were generated by one-way ANOVA with multiple comparisons using Dunnett’s correction (B and C). Two-way ANOVAs with multiple comparisons using Dunnett’s correction were used to assess differences in behavioral performance between rAAV8-LSP-hIDS_co-treated animals and naive WT mice (D and E). ∗p < 0.05; ∗∗∗p < 0.001; ns, not significant versus naive WT.

Echocardiography was performed at 12.5 months after dosing. The cardiac phenotype of the Ids KO control group was mild, and no statistically significant differences were observed between Ids KO and WT mice in functional and morphometric analyses (data not shown). Treatment effects were also not statistically significant. For example, in the rAAV8-LSP-hIDS_co groups, there were reductions in left ventricular mass compared with Ids KO mice; however, reduction was also seen for the null vector-treated group (Figure 3C).

Surprisingly, motor performance testing found that naive Ids KO mice and null vector-treated mice stayed longer on the RotaRod than WT mice, which was the opposite of the expected phenotype.^12,13 Mice in our study were older than previous reports and we speculate that additional deficits may have contributed to inability of Ids KO mice to leave the rod voluntarily without penalty. The longer latency to fall was reversed at weeks 33, 45, and 57 in mice treated with the 1.25 × 10¹² vg/kg and 6.25 × 10¹² vg/kg doses (Figure 3D). The Y-maze test, used to assess exploratory behavior and working memory, reflected no meaningful difference in the percentage alternation among all groups at any time point, including WT and Ids KO controls (Figure 3E).

Lung function was measured with the flexiVent system in Ids KO mice at 4 months from treatment initiation with a single intravenous injection of rAAV8-LSP-hIDS_co at three dose levels ranging from 2.5 × 10¹¹ vg/kg to 6.25 × 10¹² vg/kg. An additional cohort was dosed with 6.25 × 10¹² vg/kg of the null vector, rAAV8-MY011, or with intravenous idursulfase 1 mg/kg once a week. The vehicle Ids KO group displayed statistically significant differences from WT mice in three of the five parameters measured (tissue elastance, quasi-static compliance, and inspiration capacity). Neither rAAV8-LSP-hIDS_co nor weekly ERT with intravenous idursulfase normalized any of these three parameters to WT levels with statistical significance (Figure S9). HS levels in the trachea were decreased but not normalized for the rAAV8-LSP-hIDS_co groups and the idursulfase-treated group, although the magnitude of decrease was greater for all rAAV-treated groups (Figure S10). Trachea DS concentrations decreased to WT levels in the rAAV-treated groups but not in the idursulfase-treated group (Figure S11).

Effect of rAAV8-LSP-hIDS_co on hI2S concentrations and activity in NHPs

Evidence from mice provided a framework for investigating the pharmacokinetic and biodistribution of rAAV8-LSP-hIDS_co in large animals. We administered rAAV8-LSP-hIDS_co to healthy male cynomolgus monkeys at low (1.25 × 10¹² vg/kg; n = 3) or high (6.25 × 10¹² vg/kg; n = 6) doses to investigate hI2S concentration and I2S activity in serum and multiple tissues over time (up to 3 or 9 months). These two doses were chosen for NHPs to match the same amount per kilogram of body weight given to Ids KO mice in the 12-month long-term study to understand species translation for the same vector genome based on body weight.

Comparing NHPs treated with the low dose with Ids KO mice treated with 2.5 × 10¹⁰ vg/kg, where more than 90% of HS reduction was observed in the somatic tissues, serum hI2S concentration was approximately five times higher and I2S activity nearly three times higher, with sustained levels up to 3 months (Figures 4A and 4B). Meanwhile, compared with Ids KO mice administered the same dose (1.25 × 10¹² vg/kg), NHPs had a serum hI2S concentration approximately 50 times lower and a serum I2S activity about 30 times lower. A peak in I2S activity was observed at 20 days, followed by an initial drop and stabilization. Anti-hI2S anti-drug antibodies (ADAs) were either transient or undetectable at 3 months in NHPs treated with the low dose (Figure S12). NHPs treated with the high dose showed a variable profile of hI2S concentrations (Figures 4C and 4D). In general, the decline in hI2S serum levels in these NHPs was associated with the presence of anti-hI2S ADAs (Figure S12). The rapid decline in I2S activity was only observed after 3 months, whereas the decline in hI2S concentration was observed as early as 1 month after dosing. For animal #3003, which lacked anti-hI2S ADAs, anti-AAV8 ADAs, and anti-AAV8 neutralizing antibodies (Nabs; Figure S12), the durability of hI2S expression and hI2S exposure in the serum was at least 9 months (planned sacrifice time point). As an additional investigation into the catalytic nature of the enzymatic activity, the levels of formylglycine on the animal-produced enzymes isolated from NHP serum were similar to those on the recombinant hI2S enzyme (Table S1).

Figure 4.

NHP serum transgene product concentration and enzyme activity profiles in individual animals at different time points after treatment with rAAV8-LSP-hIDS_co

(A) hI2S serum concentration and (B) I2S serum enzyme activity over time up to 3 months after administration with a low dose of rAAV8-LSP-hIDS_co (1.25 × 10¹² vg/kg). (C) hI2S serum concentration and (D) I2S serum enzyme activity over time up to 9 months after administration with a high dose of rAAV8-LSP-hIDS_co (6.25 × 10¹² vg/kg). Animals 2001, 2002, and 2003 were administered 1.25 × 10¹² vg/kg of rAAV8-LSP-hIDS_co; animals 3001, 3002, 3003, 4001, 4002, and 4003 were administered 6.25 × 10¹² vg/kg of rAAV8-LSP-hIDS_co. Animals 1001, 1002, 1003, and 1004 were administered formulation buffer. C_avg, average concentration; IV, intravenous.

Translation from mouse to NHPs requires comparison of tissue exposure and tissue activity of I2S between NHPs and Ids KO mice treated with rAAV8-LSP-hIDS_co (Table 1). In NHP deep tissues (lung, heart, and bone marrow), hI2S concentration in the low-dose cohort (in the absence of anti-hI2S ADAs) was sufficient to exceed the hI2S concentrations in mice that had normalized GAGs in these tissues at 2.5 × 10¹⁰ vg/kg. In the high-dose cohort, hI2S concentrations in NHP deep tissues were affected by the presence of anti-hI2S ADAs in the serum, meaning that hI2S concentrations were numerically lower in NHPs with anti-hI2S ADAs than in those without (Figure S13; Table S2).

Table 1.

Tissue exposure of transgene product in NHPs and Ids KO mice treated with rAAV8-LSP-hIDS_co

Tissue (dose)	hI2S concentration in NHPs without ADAs (above to vehicle control)	hI2S concentration (2.5 × 1010 vg/kg) in Ids KO micea	hI2S activity in NHPs without ADAs (% of WT mice)	hI2S activity (2.5 × 1010 vg/kg) in Ids KO mice (% of WT mice)	HS reduction (2.5 × 1010 vg/kg) in Ids KO mouse tissues (% of Ids KO vehicle)	hI2S activity (2.5 × 1011 vg/kg) in Ids KO mice (% of WT mice)	HS reduction (2.5 × 1011 vg/kg) in Ids KO mouse tissues (% of Ids KO vehicle)
Lung (low)	6.0 ng/mg	2.0 ng/mg	40	30	97	270	100
Lung (high)	35.0 ng/mgb		1.35a,c
Heart (low)	12.5 ng/mg	1.4 ng/mg	232	210	97	2,480	100
Heart (high)	23.2 ng/mg		442c
Kidney (low)	4.4 ng/mg	3.0 ng/mg	10	90	91	1,380	100
Kidney (high)	27.3 ng/mg		439c
Muscle (low)	all values < LLOQ	1.0 ng/mg	87	50	100	1,000	100
Muscle (high)	5.4 ng/mg		244c
Bone marrow (low)	29.5 ng/mg	15.1 ng/mg	210	310	100	46,000	100
Bone marrow (high)	198.0 ng/mg		1,574c
Spleen (low)	36.1 ng/mg	17.0 ng/mg	110	140	98	1,200	100
Spleen (high)	281.4 ng/mg		560c

ADA, anti-drug antibody; hI2S; human iduronate-2-sulfatase; HS, heparan sulfate; Ids KO, iduronate-2-sulfatase gene knockout; LLOQ, lower limit of quantification; NHP, non-human primate; rAAV8, recombinant adeno-associated virus 8; vg, vector genome; WT, wild-type.

^ahI2S concentration was evaluated at 3-month necropsy.

^bThe lack of correspondence between hI2S concentration and enzyme activity in lung tissues for the high-dose rAAV8-LSP-hIDS_co group was probably due to technical challenges involving the enzyme assay used to detect rAAV8-LSP-hIDS_co-mediated hI2S increase.

^cTissues from animals in the high-dose group at 9 months are normalized to the respective formulation-buffer-treated animals euthanized at 9 months; values obtained from formulation-buffer-treated animals at 9 months may vary from those obtained at 3 months.

There was no detectable hI2S above the NHP endogenous I2S in the CSF, in the spinal cord, and in different brain regions (data not shown) in either cohort throughout the study.

Liver expression profile of rAAV8-LSP-hIDS_co in NHPs and mice

To determine the liver expression profile of rAAV8-LSP-hIDS_co, liver biopsies were acquired from NHPs over time; vector copy number, hIDS messenger RNA (mRNA) transcripts, and liver hI2S concentrations were measured for the low- and high-dose groups (Figure 5).

Figure 5.

NHP liver vector genome, mRNA transcript, and transgene product concentration in individual animals at different time points after treatment with rAAV8-LSP-hIDS_co

(A) Analysis of the liver profile of vector genome, (B) mRNA transcript, and (C) transgene product concentration in NHPs treated with a high or low dose of rAAV8-LSP-hIDS_co for up to 3 or 9 months. Animals 2001, 2002, and 2003 were administered 1.25 × 10¹² vg/kg of rAAV8-LSP-hIDS_co; animals 3001, 3002, 3003, 4001, 4002, and 4003 were administered 6.25 × 10¹² vg/kg of rAAV8-LSP-hIDS_co.

After administration, a decreasing trend in vector copies was observed in all NHPs by day 50 (Figure 5A). In the high-dose group, the observed decline was slower after 3 months than in the first 90 days. Some animals showed a steady mRNA transcript profile, whereas others showed a decreasing trend in expression after vector administration, except for animal #4002 (administered with 6.25 × 10¹² vg/kg of rAAV8-LSP-hIDS_co), which experienced a rapid reduction trend in mRNA expression (Figure 5B). Hepatic hI2S concentrations mirrored the mRNA expression profile after administration in animals treated with either dose, including in animal #4002, in which a rapid reduction trend in protein concentration was also observed (Figure 5C).

Microscopic evaluation of hI2S by IHC in the liver of both Ids KO mice and NHPs at 1.25 × 10¹² and 6.25 × 10¹² vg/kg was conducted to characterize the distribution of hI2S protein expression (Figure 6). The expression of hI2S detected in the hepatocytes of Ids KO was widespread and dose dependent, with accumulations of hI2S in pericanalicular lysosomes and Kupffer cells (Figures 6A–6C). In NHPs, strong hI2S expression was detected in fewer, more scattered individual hepatocytes, with less dense accumulation of hI2S in the remaining hepatocytes and Kupffer cells than that seen in mice. In NHPs, there was a slight increase in the number of hI2S-expressing hepatocytes in the 6.25 × 10¹² vg/kg dose group compared with the 1.25 × 10¹² vg/kg group (Figures 6E and 6F). hI2S expression at the cellular level showed a marked difference in distribution between mouse and NHP livers. Scattered random individual cells were observed throughout the parenchyma in NHPs, in contrast with the relatively homogeneous dense distribution in all hepatocytes in mice.

Figure 6.

IHC of hI2S transgene product in Ids KO mice and NHP liver

Distribution of hI2S evaluated by IHC in the liver in both Ids KO mice and NHPs. Ids KO mice that received (A) vehicle control or (B) 1.25 × 10¹² vg/kg, or (C) 6.25× 10¹² vg/kg of rAAV8-LSP-hIDS_co 12 months after administration. NHPs that received (D) the vehicle control, or (E) 1.25 × 10¹² vg/kg, or (F) 6.25 × 10¹² vg/kg of rAAV8-LSP-hIDS_co 3 months after administration. Scale bar, 200 μm.

Vector genome levels in the liver and other tissues in NHPs were assessed together with those in treated WT mice (Figures 7A and 7B). In NHPs, non-liver tissues had a numerically lower number of viral vector copies than the liver at 3 months in animals in the low-dose group and at 3 and 9 months in animals in the high-dose group. A trend toward a dose-dependent increase in viral vector copies in all tissues at 3 months was observed. However, at 9 months, the viral vector copy number in the high-dose cohort showed a trend toward a decrease in all non-liver tissues (Figure 7A). Similarly, in mice, higher vector doses were associated with viral vector copy number, and non-liver tissues generally showed a lower vector copy number than the liver (Figure 7B). When each dose was evaluated across species, there were lower levels of vector genome, mRNA transcript, and hI2S in the livers of NHPs than in those of mice, in which the degree of difference was more evident in the mRNA transcript and hI2S levels (Figure 7C).

Figure 7.

Comparison of vector genome, mRNA, and transgene product in WT mice and NHPs

Vector genome in tissues of (A) NHPs treated with rAAV8-LSP-hIDS_co for up to 3 or 9 months and (B) WT mice treated with rAAV8-LSP-hIDS_co for up to 3 months. (C) Comparison of vector genome, transgene mRNA, and transgene product of liver in WT mice and NHPs treated with rAAV8-LSP-hIDS_co. Data are presented as mean ± SD of n = 6 animals.

GLP toxicology study in mice

The aim of this study was to determine any potential toxicity of rAAV8-LSP-hIDS_co and to assess the pharmacokinetics, tissue exposure of hI2S, and tissue vector genome concentrations during a 13-week observation period. rAAV8-LSP-hIDS_co administered to WT mice at 1.25 × 10¹² vg/kg, 6.25 × 10¹² vg/kg, and 3.0 × 10¹³ vg/kg was well tolerated at all dose levels. All in-life, clinical, and anatomical pathology parameters were within normal limits. The no-observed-adverse-effect level (NOAEL) was 3.0 × 10¹³ vg/kg, the highest dose level in the study. Pharmacokinetic parameters for hI2S in serum are summarized in Table S3. Serum hI2S levels were similar among WT and Ids KO male mice (data not shown). Sex-dependent differences in serum hI2S concentrations were observed; namely, male mice tended to have hI2S concentrations more than twice as high as those of female mice (Table S3).

Discussion

In contrast to current treatments for MPS II, gene therapy with IDS could potentially lead to long-term sustained and constant vector-mediated I2S production, allowing penetration into deep tissues not adequately addressed by intravenous ERT. Using the liver as a depot organ for a systemically delivered rAAV-based vector can lead to cross-correction of other organs via the systemic circulation while also potentially providing immunotolerance to the transgene product.¹⁴

In the MPS II mouse model, we demonstrated that, when rAAV8-LSP-hIDS_co was administered to young mice, hepatic hI2S production was sufficient to normalize GAG burden in somatic tissues at doses below 1 × 10¹² vg/kg. Doses higher than 2.5 × 10¹⁰ vg/kg partially reduced brain GAG levels and improved histopathology. We observed normalization of bone density, suggesting that deep tissue exposure of liver-produced hI2S could have translational benefits to skeletal and joint clinical pathology. In RotaRod motor behavioral testing, the reduced latency to fall phenotype expected for Ids KO mice was not observed, possibly owing to age-related mice vision deficits. Nevertheless, the longer-latency knockout phenotype was corrected. Moreover, cardiorespiratory and neurobehavioral functional measurements were uninformative owing to the weak phenotype of the mouse model.

Although several rAAV-based gene therapies have shown promising results in knockout mice models,^15,16,17 side-by-side evaluations in larger species remain unfortunately rare in preclinical publications. Knowing that rAAV-driven protein delivery is characterized by lower expression efficiency in larger species,^18,19,20,21 we believe that cross-species studies are critical to evaluate whether rAAV-based treatments reach desirable therapeutic levels. To increase the translational understanding of rAAV8-LSP-hIDS_co from Ids KO male mouse studies, the transduction efficiency of rAAV8-LSP-hIDS_co gene therapy in the liver of NHPs was assessed together with its ability to cross-correct deep tissues (lung, heart, kidney, brain, and bone marrow), where the therapeutic hI2S levels in somatic tissues, but not in the CNS, were observed. At similar doses to those evaluated in a mouse toxicology/biodistribution study (1.25 × 10¹² to 6.25 × 10¹² vg/kg), NHPs treated with rAAV8-LSP-hIDS_co showed sustained hI2S production in the liver. The comparison of the measurements of vector genome, mRNA, and hI2S levels in a mouse toxicology/biodistribution study with those observed in NHPs revealed a lower efficiency in one or multiple steps, from receptor binding and transgene production and secretion to tissue cross-correction in NHPs, than in mice: this may have contributed to the lower transgene product exposure levels observed in NHPs. The observation of the species differences is not unique to rAAV8-LSP-hIDS_co but has also been reported by other investigators.^22,23 As a result, we speculate that the partial GAG correction observed may be explained by the greater supraphysiological amount in mice than in NHPs of circulating transgene product, which enters the brain by diffusion crossing the blood-CSF barrier through the choroid plexus as suggested by IHC detection of hI2S in mouse choroid plexus and meninges. Although rAAV8-LSP-hIDS_co gene therapy has only demonstrated partial CNS GAG normalization in Ids KO mice models and a lack of CNS uptake in NHPs, it has shown promising, translatable therapeutic effects in the cardiorespiratory, skeletal, and bone systems, addressing unmet medical needs.

Safety evaluation of rAAV8-LSP-hIDS_co in NHPs was integrated into the study design. With this objective, the animals were not treated with immunosuppressants to avoid potential confounding factors. Without immunosuppressants, anti-hI2S ADAs were only observed in the high-dose cohort, suggesting that there is a circulating level of transgene product above which ADA generation is triggered due to breakthrough of the tolerization threshold that can be species dependent. Given that rAAV8-LSP-hIDS_co was designed to have the liver secrete the therapeutic protein into the systemic circulation and cross-correct other tissues, the level of hI2S in the circulation determines its amount in cross-corrected tissues. In animals with anti-hI2S ADAs, hI2S exposure in cross-corrected tissues (e.g., lung, heart, and bone marrow) correlated with the low circulating hI2S level. In contrast, animal #3003 had sustained systemic hI2S levels throughout the study in the absence of ADAs and had consistent hI2S exposure and enzyme activity in somatic tissues. In addition, the impact of anti-capsid NAbs in the absence of anti-hI2S ADAs was observed in animal #4002, which belonged to the high-dose group. This animal was pre-screened for anti-AAV8 NAbs and was confirmed to be seronegative before dosing. Owing to the lack of liver samples before 3 months, it is difficult to know if the vector had a rapid decline of transduction (i.e., vector genome levels) immediately after dosing or if it started with low baseline transduction in this animal. However, based on data obtained at 3 months, we speculate that T cell-mediated cytotoxicity may have inhibited or reduced the hepatocyte transduction in this animal, leading to low mRNA and protein levels. These observations further confirm the need to include immunosuppressants as part of rAAV-based gene delivery studies, should transgene product durability be measured.

In summary, liver-directed rAAV8-LSP-hIDS_co gene therapy demonstrated significant reduction of GAG levels in difficult-to-treat systemic tissues and partially reduced CNS GAG levels in Ids KO mice. Although meaningful hI2S levels were achieved in almost all measured somatic organs in NHPs, the level of circulating hI2S was not sufficient to enable adequate amounts of enzyme to cross passively into the CNS. The current study established a foundation for a correlation between dose, circulating hI2S level, and tissue exposure levels (somatic and CNS tissues) in rodents and large animal models, enabling translational understanding of gene therapy development for MPS II. The difference in CNS exposure between rodents and NHPs, probably due to circulating hI2S levels being lower in NHPs than in rodents at the same dose, further emphasizes the importance of evaluating gene therapy products in NHPs.

Overall, these results demonstrate that appropriately designed preclinical rodent and NHP studies are critical to support the translation, including dose selection, of gene therapies into the clinic. Further studies with higher doses, improvement of vector potency, or proteins using an active mechanism to access the CNS would be required to investigate the feasibility of systemic gene therapy for both somatic and CNS benefits for patients with MPS II.

Materials and methods

Animals

Animal welfare complied with all applicable regulations, and all in vivo procedures described were approved by the Takeda Pharmaceuticals Institutional Animal Care and Use Committee.

Hemizygous Ids KO male (B6N.Cg-Ids^tm1Muen) mice containing an X-linked mutation and recapitulating several biological and pathological aspects of MPS II disease and C57BL/6 WT mice were bred at Taconic Laboratories.^24,25 Upon arrival at Takeda, animals were housed with their littermates. C57BL/6 WT mice for the GLP toxicology study were received from Charles River Laboratories, Raleigh, NC, USA and housed at the testing facility (Charles River Laboratories Ashland, Ashland, OH, USA). All mice were maintained on a 12-h light/dark cycle, and water and food were provided ad libitum.

Cynomolgus monkeys (Macaca fascicularis) were supplied by Worldwide Primates, Miami, FL, USA. Blood samples (500 μL of serum) from 100 animals were screened and evaluated for antibodies against the adeno-associated virus serotype 8 (AAV8) vector. In total, 25 animals that tested negative for AAV8 Nabs were shipped to Charles River Laboratories, Reno, NV, USA for study activities and were acclimated to laboratory housing for at least 7 days before dosing initiation. Animals were housed singly upon arrival until completion of dose administration, then in groups of up to three animals of the same dosing group for the remainder of the study.

Vector structure and production

The vector genome for rAAV8-LSP-hIDS_co contains a 5′ inverted terminal repeat (ITR), three hepatocyte-specific cis-regulatory modules, a liver-specific transthyretin promoter, a minute virus of mice intron, a codon-optimized human IDS (hIDS_co) coding sequence, a woodchuck hepatitis virus post-transcriptional regulatory element (WPRE), a bovine growth hormone polyadenylation sequence, and a 3′ ITR (Figure 1A). The IDS coding sequence was optimized based on codon-usage frequency in the liver, a reduced number of CpG sites, and additional modifications to improve expression (e.g., removing repeat and inverted repeat sequences and potential splicing sites).

The rAAV8-LSP-hIDS_co vector was produced using standard triple-transfection methods. In brief, HEK293 cells were co-transfected with three plasmids: the ITR plasmid containing the IDS transgene, a RepCap expression cassette, and an adenoviral helper gene expression cassette. After 72–96 h, the cells were chemically lysed and the cell pellet and medium were collected. The cell lysate was clarified and treated with benzonase. The clarified lysate was loaded to a POROS CaptureSelect AAVX column connected to an AKTA purification system, and AAV vector-containing fractions of the elute were collected according to the manufacturer’s recommended conditions. Fractions containing AAV vectors were pooled and subjected to a cesium chloride continuous gradient to enrich for full capsids. Fractions containing mostly full capsids were pooled and subjected to buffer exchange. The AAV vectors were formulated in phosphate buffered saline with 0.001% Pluronic F-68 and were subjected to standard characterization, including a quantitative polymerase chain reaction (qPCR) for titrating, silver staining for purity, an amebocyte lysate assay (Endosafe) for endotoxin measurement, and an in vitro transduction assay to determine biological activity. Characterized vectors were aliquoted and frozen at −80°C.

Study design and vector administration

All mice for pharmacology experiments were dosed at Takeda in Cambridge, MA, USA; samples were tested at the same facilities or transferred to external contract research organizations (CROs), depending on the experiment.

A single intravenous rAAV8-LSP-hIDS_co injection via tail vein was administered to Ids KO mice at different doses ranging from 2.5 × 10⁸ to 6.25 × 10¹² vg/kg, depending on the experiment. Ids KO mice were dosed at 5–7 weeks of age (16 weeks of age for the respiratory function studies). In some experiments, an additional Ids KO mice cohort was dosed with a null vector (rAAV8-MY011, vector genome with no promoter and transgene sequences) as a negative control; age-matched naive Ids KO and WT mice were also included as controls. For the respiratory function study, an Ids KO mice cohort was also dosed with intravenous idursulfase 1 mg/kg once a week. The Ids KO mice cohort used for the imaging measurements was assessed at Invicro (San Diego, CA, USA) at different time points up to 13.5 months after dosing. For the functional studies, mice were tested at PsychoGenics (Paramus, NJ, USA) for behavioral assessments beginning at 21 weeks after dosing.

A GLP toxicology study was also performed in WT mice to assess the tolerability of rAAV8-LSP-hIDS_co and to identify the NOAEL. For this study, rAAV8-LSP-hIDS_co was evaluated at doses of 1.25 × 10¹² vg/kg, 6.25 × 10¹² vg/kg, and 3.0 × 10¹³ vg/kg administered via a single intravenous injection to WT mice at Charles River Laboratories Ashland. An additional WT mouse cohort was dosed with formulation buffer as a control.

Data from the mice studies were used to identify the most appropriate vector dose to treat NHPs and, therefore, to assess serum and tissue transgene product concentrations, enzyme activity profiles, and vector biodistribution in large animals. Healthy male cynomolgus monkeys aged 2.1–2.9 years were administered with rAAV8-LSP-hIDS_co at two different doses, 1.25 × 10¹² vg/kg and 6.25 × 10¹² vg/kg, through a single 30-min intravenous infusion via a suitable peripheral vein using a calibrated infusion pump. Formulation buffer was administered to the control animals.

Necropsy and sample collection

Mice were euthanized using carbon dioxide inhalation at different time points after vector administration depending on the experiment (see “results” section). Blood was collected via cardiac puncture and mice were perfused with 10 mL of saline. Serum was obtained by centrifuging blood at 13,200 revolutions per minute (rpm) for 3 min at room temperature. Collected brain and somatic tissues (liver, kidneys, spleen, heart, lung, bone marrow, quadriceps, skin) were divided into multiple sections, placed in 10% formalin for histology analysis, or frozen at −80°C for further analyses.

At 3 or 9 months after AAV injection, NHPs were humanely euthanized by induction into deep, unrecoverable anesthesia with ketamine-pentobarbital administration. This was followed by blood collection for analysis of parameters, and saline perfusion to clear assay-interfering circulating transgene proteins and endogenous serum proteins from the tissues to be evaluated. Serial CSF and blood samples were collected throughout the duration of the study. Following collection of CSF samples, total-body perfusion was performed with ice-cold 0.9% saline for approximately 5–10 min at a rate of 250 mL/min until no blood was visible in the perfusate. Selected peripheral tissues (liver, kidney, spleen, heart, lung, skin, bone marrow, spinal cord, and quadricep) and prespecified brain regions were collected for IHC analysis or frozen at −80°C for further analyses.

hI2S serum and tissue ELISA

hI2S concentrations from serum and tissue were analyzed by ELISA using a purified goat anti-I2S capture antibody (B852) and a horseradish peroxidase-conjugated goat anti-I2S detection antibody (B852). Samples exceeding the high end of the calibration curve were further diluted and retested. Concentrations from serum were expressed as ng/mL, and results from tissues were normalized to total protein in extracts as determined by bicinchoninic acid (BCA) assay (Pierce Biotechnology, Rockford, IL, USA) and were expressed as ng/mg protein.

I2S activity assay

The fluorometric assay for I2S enzymatic activity is a two-step reaction under acidic conditions.²⁶ During the first step, tissue or serum samples and controls were diluted, mixed with 4-methylumbelliferyl α-L-iduronide-2-sulfate (4-MUS) (Santa Cruz Biotechnology, Dallas, TX, USA; or Biosynth, San Diego, CA, USA) and added on to black microtiter 96-well plates (Sigma-Aldrich, St Louis, MO, USA). Following 1 h of incubation at 37°C, I2S hydrolyzed 4-MUS to 4-methylumbelliferyl α-L-iduronide (MUBI). In the second step, the enzyme alpha-L-iduronidase was added to the plate and incubated at 37°C to hydrolyze MUBI to the final product, 4-methylumbelliferone (4-MU). After 4 h of incubation, the reaction was stopped and a standard curve of 4-MU sodium salt (Sigma-Aldrich, St Louis, MO, USA) ranging from 3.91 to 4,000 nM for mice and from 1.47 to 6,000 nM for NHPs was added to the plate. Fluorescence was read at 365-nm excitation, 450-nm emission, and 435-nm cutoff. Serum data were expressed as nmol/h/mL; results from tissues were normalized to total protein in tissue extracts as determined by BCA assay (Pierce Biotechnology, Rockford, IL, USA) and expressed as nmol/h/mg protein.

GAG quantification

For HS and DS quantification, mouse liver, kidney, spleen, heart, lung, bone marrow, quadriceps, skin, and brain tissues were homogenized in 25 mM ammonium formate and 1 mM calcium acetate buffer followed by digestion with 12 mU/μL heparinase I, 4 mU/μL heparinase II, and 1.4 mU/μL heparinase III (all from New England Biolabs, Ipswich, MA, USA) at 30°C for 16 h for HS or 0.1 μg/μL chondroitinase B (R&D Systems, Minneapolis, MN, USA) at 35°C for 16 h for DS to degrade the polysaccharide forms to HS or DS disaccharides. Digestion reactions were precipitated with chilled methanol and the supernatant dried and resuspended in 20 μL of methanol and 150 μL of 5 mM tributylamine, 5 mM ammonium formate, and 0.025% formic acid in 95:5 water:acetonitrile. Ten microliters was injected on a Sciex API5000 mass spectrometer (AB Sciex, Framingham, MA, USA) containing a Waters ACQUITY UPLC BEH C18 column (130Å, 1.7 μm, 2.1 × 50 mm) at 50°C. Following gradient liquid chromatography separation,²⁷ turbo ion spray ionization in negative-ion mode was coupled with multiple reaction monitoring for disaccharides. Intact polysaccharides of HS from porcine intestinal mucosa (Calbiochem, San Diego, CA, USA) and DS from porcine mucosa (Iduron, Alderley Park, Alderley Edge, Cheshire, UK) were used as reference standards. Four major disaccharides degraded from HS (ΔUA, GlcNAc,6S [II-A]; ΔUA, GlcNS [IV-S]; ΔUA, GlcNS,6S [II-S]; and ΔUA, GlcNAc [IV-A]) were selected as signature disaccharides monitored in HS quantitation. Two major disaccharide isomers, DS4S (DS disaccharide UA, GalNAc,4S) and DS6S (DS disaccharide UA, GalNAc,6S), resulting from specific digestion of DS were used in DS quantitation. Calibration standards were used for quality control (QC) and acceptance criteria set at back-calculated concentration not deviating from more than ±25.0%. Lower limit of quantification for HS and DS was 50 ng/mL. Tissue homogenates were measured for protein concentration using BCA assay (Pierce), GAG quantifications were normalized to the tissue protein levels, and concentrations expressed as μg GAG/mg protein.

Functional studies in mice

Imaging measurements

Micro-CT imaging of whole-body skeletal anatomy was performed using an Inveon CT scanner. For CT imaging, mice were anesthetized with 2%–3% inhaled isoflurane and maintained with 1%–2% inhaled isoflurane in oxygen throughout image acquisition. Each mouse was positioned in the scanner such that the left hindlimb was tucked under the animal and the right hindlimb was extended behind, making sure all limbs and the head were in the field of view. Phantom scans were performed with a mineral density phantom (three densities) immediately before in vivo image acquisition. Scans were set to medium magnification (79.66-μm pixels) using standard X-ray tube settings and exposure (80 kV, 500 μA, 275 ms) and acquiring two bed positions in step + shoot mode with 360° projections. The femur, tibia, humerus, radius, cranium, and zygomatic arch were segmented from the whole-body CT scans. The dense bone volume was calculated for each bone using a threshold-based approach. A voxel Hounsfield unit (HU) value was chosen as a low threshold corresponding to dense bone; then, all voxels with HU values above this threshold were segmented, counted, and converted to a volume measurement of dense bone. Test article identities were blinded to investigators for CT and echocardiography measurements.

For echocardiography, mice were anesthetized with 2%–3% inhaled isoflurane and maintained with 1%–2% inhaled isoflurane in oxygen throughout imaging. Isoflurane was adjusted as necessary to maintain a target heart rate of 400–600 beats per minute in each animal. Following induction of anesthesia, animals were placed supine on a heated stage (41°C–42°C) and images were obtained using a Visual Sonics Vevo 2100 ultrasound system with MS550S 32- to 56-MHz linear echocardiography transducer. Measurements acquired were ejection fraction, fractional shortening, stroke volume, cardiac output, heart rate, left ventricular wall thickness in systole and diastole, myocardial area change, left ventricle mass, and left ventricular mass corrected.

RotaRod behavioral test

All mice were tested on a RotaRod (Ugo Basile Instrument, Varese, Italy) over 3 days. Test article identities were blinded to investigators. An accelerating RotaRod testing program was selected based on previous Ids KO mouse phenotyping.^13,28 Mice were trained on the RotaRod on day 1 with two training trials. Trial 1 was a habituation session at 5 rpm for 60 s. If mice fell during trial 1, they were replaced on the rod until 60 s was reached. After a 30-min inter-trial interval, mice began trial 2, a 5-min session with an accelerating speed of 5–40 rpm. Both training trials were unrecorded. All mice were then tested in six testing trials, with two trials per day for 3 days. All testing trials were recorded and conducted over 5 min with a speed of 5–40 rpm and an inter-trial interval of 5 min. Latency to fall was recorded for all testing trials.

Y-maze behavioral test

Spontaneous alternation was evaluated in a single 8-min trial in a Y-maze consisting of three identical arms (40 × 10 × 20 cm) and each mouse was allowed to see distal spatial landmarks. Test article identities were blinded to investigators. The test mouse was placed in the middle of the three arms and allowed to explore freely. Arm entry was considered successful when hind paws were placed in the arm in full. Spontaneous alternation was described as successive entries into three arms, in overlapping triplet sets. The effect was calculated as percentage alternation = [number of alternations/(total number of arm entries −2)] × 100. Total entries were recorded as an indication of ambulatory activity, and mice that performed fewer than 12 entries in 10 min were excluded from the analysis.

Respiratory function measurements

Respiratory mechanics were assessed by the flexiVent (SCIREQ, Montreal, QC Canada) system in mice. An endotracheal tube was inserted and connected to a flexiVent ventilator. Before transfer and connection, the flexiVent circuit was primed with isoflurane to minimize potential for animal recovery. Maintenance anesthesia was achieved with 1.0%–3.5% isoflurane. Respiratory parameters, including elastance, resistance, and compliance, were collected via forced oscillation techniques. The CRO conducting the flexiVent measurements (QTest Labs, Columbus, OH, USA) was not blinded to the experimental groups. If measurements could not be accurately recorded, individual mice were excluded from the analysis (n = 2 omitted from the 6.25 × 10¹² vg/kg rAAV8-LSP-hIDS_co group; n = 1 omitted from the naive Ids KO group; n = 4 omitted from the idursulfase-treated group). Throughout each terminal experiment, analog signals were digitally sampled (500–2,000 Hz) and recorded continuously with a data acquisition system (IOX, emka TECHNOLOGIES, Paris, France; and flexiWare, SCIREQ Scientific Respiratory Equipment, Montreal, QC, Canada). Ids KO animals treated with idursulfase were terminated approximately 7 days after the last idursulfase infusion.

Vector biodistribution and hIDS mRNA

Vector biodistribution and hIDS mRNA quantification were analyzed using qPCR and reverse transcription qPCR (RT-qPCR). Briefly, genomic DNA and total RNA were extracted from RNAprotect Tissue Reagent (Qiagen, Valencia, CA, USA) from preserved mouse and NHP tissues using a QIAsymphony DSP DNA Mini Kit (Qiagen, Valencia, CA, USA) or a QIAsymphony RNA Kit (Qiagen, Valencia, CA, USA).

For vector biodistribution analysis, purified genomic DNA was analyzed using a qPCR assay, a set of transgene-specific primer/probe for vector genome copy number quantification. An endogenous single-copy gene (transketolase gene) on mouse and NHP chromosomes was also measured by qPCR to quantify the copy number of haploid genome for normalizing the genomic DNA concentration. The measured vector genome copy number and the normalized genomic DNA concentration were then used to calculate the copy/μg genomic DNA or copy/diploid genome units.

For hIDS mRNA quantification, total RNA was analyzed by RT-qPCR for the mRNA of hIDS transgene and three housekeeping genes, namely the actin γ1 gene (ACTG1), the ribosomal protein S18 gene (RPS18), and the ribosomal protein S27 gene (RPS27), separately with the corresponding primers and TaqMan probes. Relative hIDS mRNA expression was calculated using the average of the three housekeeping gene measurements and the $2^{-\Delta \Delta {\mathrm{C}}_{\mathrm{T}}}$ method.²⁹

Histology

For mouse studies, somatic tissues and left hemispheres of brains were fixed in 10% neutral buffered formalin for 24 h. Brain was trimmed into four slices from anterior to posterior, namely striatum, thalamus, hippocampus, and cerebellum. All tissues were processed for paraffin block and 5-μm sections were prepared for I2S and LAMP1 IHC staining. IHC staining for hI2S and LAMP1 detection was performed with BondRX Stainer. Briefly, after antigen retrieval with BOND epitope retrieval solution 1 (pH 6.0), the following primary antibodies were applied: rabbit anti-hI2S polyclonal antibody (1:500; LSBio, Seattle, WA, USA) for detection of hI2S expression and distribution, and rabbit anti-LAMP1 polyclonal antibody (Abcam, Cambridge, UK; ab24170; 1:2,000 for brain and 1:3,000 for other tissues) to detect LAMP1, a marker of lysosomes. A BOND Polymer Refine kit (Leica Biosystems, Buffalo Grove, IL, USA) containing a peroxide block, post primary, polymer reagent, DAB chromogen, and hematoxylin counterstain was supplied as ready-to-use detection system for the automated BOND stainer. All stained slides were dehydrated with gradient alcohol, cleared with xylene, and mounted with mounting medium, then scanned with an Aperio ScanScope AT2 scanner (Leica Biosystems, Lincolnshire, IL, USA). The whole digital slide for each tissue or the main brain regions, including cortex, striatum, thalamus, hippocampus, and cerebellum, were viewed and analyzed, and representative images were taken by ImageScope (Leica Biosystems, Lincolnshire, IL, USA). For LAMP1 quantitative analysis, the Aperio Positive Pixel Count algorithm (Leica Biosystems, Lincolnshire, IL, USA) was selected and adjusted to cover LAMP1-positive staining accurately. The data were presented as LAMP1 positivity percentage (positivity percentage = positive area [pixels]/total analyzed area [pixels] × 100%).

For the NHP IHC study, formalin-fixed, paraffin-embedded male cynomolgus monkey liver tissue was cut at 3 μm and kept at room temperature before staining. Following deparaffinization and rehydration, the slides were incubated in 1× Target Retrieval Solution pH 6 (Agilent, Dako, Santa Clara, CA, USA) for 20 min at 97°C and then placed in a Thermo Autostainer (Agilent, Santa Clara, CA, USA). Endogenous peroxidase activity was blocked by using 0.3% hydrogen peroxide for 20 min. Slides were then incubated in either 1.0 μg/mL of rabbit anti-I2S antibody (LSBio, Seattle, WA, USA; catalog number LS-C341717) or an isotype control rabbit immunoglobulin G (Vector Laboratories, Burlingame, CA, USA; catalog number I-1000) for 60 min followed by the secondary antibody, Dako EnVision+ System labeled polymer (goat anti-rabbit; Agilent, Santa Clara, CA, USA; catalog number K400311-2) for 30 min. Antibody binding was detected using 3,3′-diaminobenzidine tetrahydrochloride as the chromogen. Slides were then counterstained in hematoxylin and dehydrated through a progressive series of alcohols and xylene before cover slipping.

I2S ADA assay

The I2S ADA assay comprised a homogeneous “bridging” electrochemiluminescence (ECL) format with biotin and ruthenium-labeled recombinant hI2S (idursulfase, Takeda Pharmaceuticals USA, Lexington, MA, USA) bound together (bridged) by ADAs in test samples for detection.

rAAV8 capsid ADA assay

For the rAAV8 capsid ADA assay (a sandwich ECL assay), serum samples were diluted to minimum required dilution and added to an AAV8-coated plate; after incubation for 1 h, the plate was washed and the rAAV8 ADA bound to the AAV8 capsid was detected with SULFO-TAG-labeled Protein A/G/L (Meso Scale Diagnostics, Rockville, MD, USA).

rAAV8 capsid NAb assay

The rAAV8 capsid NAb assay was a cell-based assay that used a rAAV8 vector coding for luciferase as a reporter. Briefly, serum samples were first incubated with a rAAV8-CMV-Luc vector for 1 h, then the mixture was added to wells containing pre-seeded HEK293T cells pretreated with etoposide. The cells were lysed after 2 days and the luciferase in the cell lysate was measured. The luciferase signal intensity inversely correlates with the concentration of NAb in the serum sample. The assay threshold was established at a 50% reduction in luciferase signal, and any sample with a signal reduction of more than 50% was determined as NAb positive.

GLP toxicology study in mice

Potential toxicity of rAAV8-LSP-hIDS_co, pharmacokinetics, tissue exposure of transgene product hI2S, and tissue vector genome concentrations were evaluated across a 13-week observation period. Standard histopathology and safety assessments were performed on 10 animals/sex/group at necropsy on days 30 and 91.

Statistical analysis

All data were analyzed with Prism 7 (GraphPad Software, San Diego, CA, USA) unless otherwise noted. Data are presented as either individual values or mean ± standard deviation (SD). Data were evaluated using one-way or two-way analysis of variance (ANOVA) with multiple comparisons using Dunnett’s correction. p < 0.05 was considered statistically significant.

Data availability

Restrictions apply to the availability of information and materials related to the viral vectors (cis-regulatory modules), which were used under license for this study. Please contact the corresponding author with any inquiries.