An empowered, clinically viable hematopoietic stem cell gene therapy for the treatment of multisystemic mucopolysaccharidosis type II

Abstract

Mucopolysaccharidosis type II (MPS II), or Hunter syndrome, is a rare X-linked recessive lysosomal storage disorder due to a mutation in the lysosomal enzyme iduronate-2-sulfatase (IDS) gene. IDS deficiency leads to a progressive, multisystem accumulation of glycosaminoglycans (GAGs) and results in central nervous system (CNS) manifestations in the severe form. We developed up to clinical readiness a new hematopoietic stem cell (HSC) gene therapy approach for MPS II that benefits from a novel highly effective transduction protocol. We first provided proof of concept of efficacy of our approach aimed at enhanced IDS enzyme delivery to the CNS in a murine study of immediate translational value, employing a lentiviral vector (LV) encoding a codon-optimized human IDS cDNA. Then the therapeutic LV was tested for its ability to efficiently and safely transduce bona fide human HSCs in clinically relevant conditions according to a standard vs. a novel protocol that demonstrated superior ability to transduce bona fide long-term repopulating HSCs. Overall, these results provide strong proof of concept for the clinical translation of this approach for the treatment of Hunter syndrome.

Graphical abstract

Biffi and colleagues provide proof of concept of efficacy and safety of a novel, lentiviral-based gene therapy approach for Hunter syndrome aimed at enhanced tissue IDS enzyme delivery. They implemented a novel protocol with superior ability to transduce bona fide long-term repopulating human HSCs, strongly supporting the clinical translation of this approach.

Introduction

Mucopolysaccharidosis type II (MPS II), also known as Hunter syndrome, is an inherited X-linked recessive pediatric lysosomal storage disorder (LSD) that affects 1.3 per 100,000 male live births.¹ It is due to mutations in the IDS gene, located on chromosome Xq28, which cause deficiency of the enzyme iduronate-2-sulfatase (IDS). The IDS enzyme is required for the stepwise degradation and recycling of complex glycosaminoglycans (GAGs), including dermatan sulfate and heparan sulfate. Loss of IDS activity leads to massive multisystemic accumulation of GAGs, resulting in progressive cellular and multi-organ dysfunction characterized by a wide variety of symptoms, including distinctive coarse facial features, skeletal abnormalities with short stature, and cardio-respiratory and neurologic involvement. The prognosis is highly variable. The clinical spectrum can be seen as a continuum, varying from a severe form with neurodegeneration to an attenuated form without neuronal involvement. Severe early-onset disease (60%–80% of cases) becomes clinically apparent within 2–4 years of age with patients exhibiting behavioral disturbances and progressive intellectual disability, concomitant with neurodegeneration, followed by death in adolescence due to obstructive airway disease and cardiac failure.² In contrast, patients with the mild, late-onset form preserve normal or slightly impaired intelligence and survive into late adulthood.

Targeted replacement of the gene encoding the defective enzyme represents a potentially effective therapeutic strategy for MPS II: the functional protein can be provided to the affected tissues and taken up by the deficient cells via the mannose-6-phospate receptor and other similar receptors, a process known as cross-correction. This can be achieved either by direct delivery of the recombinant protein, i.e., enzyme replacement therapy (ERT), or by generating a tissue source of functional enzyme, either via hematopoietic cell transplant (HCT) from healthy allo-compatible donors or by hematopoietic stem cell (HSC) gene therapy (GT).^3,4,5 To date, since its approval by the EMA in 2007, treatment with Elaprase is the only authorized medicinal product for MPS II in Europe. It is an ERT based on idursulfase, a recombinant human I2S enzyme (Takeda Pharmaceuticals International Ireland Branch) indicated for the long-term treatment of patients with Hunter syndrome.^1,6 In a phase II/III clinical trial enrolling children over the age of 5 years, idursulfase was shown to improve some of the somatic signs and symptoms of the disease, including walking ability, although not preventing or reversing the characteristic cardiac disease and cognitive decline observed in patients with severe disease.^7,8 This partial efficacy is likely due to the inability of the recombinant enzyme to efficiently cross the blood-brain barrier, and it renders ERT unable to prevent or mitigate disease manifestations in patients with neurological impairment. Despite allogeneic HCT being employed with beneficial effects on neurologic disease manifestations in other LSDs, experience in severe phenotypical Hunter syndrome is limited and, currently, HCT is not considered as a standard of care (SoC) for MPS II patients.⁹ However, recent clinical results support the use of HCT as a treatment strategy for early-stage MPS II patients, such as children diagnosed on pre-symptomatic testing or through newborn screening programs, as it might positively affect the neurological disease progression.^10,11

GT has been extensively explored as a therapeutic option to treat various LSDs, including MPS disorders.^12,13,14,15 In particular, ex vivo GT based on the use of autologous HSCs and integrating viral vectors is a valuable strategy to effectively and safely correct the genetic defect and provide clinical benefit in patients affected by other LSDs.^16,17,18,19 In this setting, autologous HSCs are genetically modified ex vivo to express supraphysiological levels of the functional lysosomal enzyme, hence establishing, upon engraftment and generation of a mature tissue progeny, a sustained and potentially lifelong source of therapeutic enzyme in multiple tissues, including the CNS. Importantly, the genetically corrected myeloid cell progeny of the transplanted HSCs in the CNS can provide functional lysosomal enzyme for metabolic rescue and also restore healthy microglia functions, including scavenging of the accumulated substrate and cell debris, and reverse the detrimental effects of microglial activation responses, which are involved in the LSD neurodegenerative process. Based on this rationale, we developed up to clinical readiness an HSC GT approach for MPS II that is characterized by a solid potential to exert therapeutic effects.

Results

HSC GT restores IDS activity at supranormal levels in peripheral blood of Ids^−/ mice

A third-generation lentiviral vector (LV) encoding a codon-optimized version of the human Ids cDNA under the control of the ubiquitous promoter for human phosphoglycerate kinase (PGK) (hIDS_LV) was developed and tested for the GT of MPS II. It is based on a self-inactivating (SIN) LV transfer plasmid and backbone, along with a promoter already validated in multiple LSD HSC GT clinical trials^17,18,19,20 able to drive a lysosomal enzyme overexpression in target tissues inducing a therapeutic benefit, coupled to a favorable safety profile. The hIDS_LV was used to generate the test item (TI) for the described proof-of-concept (PoC) study that consists of the murine equivalent of autologous human CD34⁺ hematopoietic stem and progenitor cells (HSPCs), obtained by lineage depletion of bone marrow (BM) harvested from young MPS II donor mice (6–8 weeks old), transduced with a laboratory-grade preparation of the hIDS_LV, as described.²¹ The control items (CIs) were lineage-depleted (Lin⁻) BM cells harvested from young (6–8 weeks old) Ids^−/ or wild-type (WT) CD45.2 C57BL/6 mice (CI^WT, transplanted into WT CD45.1 recipients) undergoing a mock transduction. A total of 15 batches of cell products were manufactured to complete the study. The homogeneity and consistency of the input TI and CIs were determined in each batch by several parameters, including the purity of Lin⁻ cells (90% ± 4.9%) and the mean proportion of Kit⁺Lin⁻Sca1⁺ cells within HSPCs (7.5% ± 1.7%). Clonogenic potential of the TI and CI, evaluated by colony-forming cell (CFC) assay, was similar for all test and control products, and falling in the expected range observed in CI^WT cells, with very low variability (mean ± SD; TI, 55.4 ± 9.2; CI, 57.3 ± 9.2; WT, 54.3 ± 21.1 [SD]). Vector copy number (VCN) was assessed on genomic DNA extracted from a pool of harvested hematopoietic progenitor colonies and from bulk populations of TI and CI after expansion in liquid culture (LC) for 14 days (Figure 1A). The efficient transduction of the TI led to robust IDS expression in its in vitro cell progeny: mean IDS activity was approximately 2-folds than that of the unmanipulated CI^WT cells (1,667 ± 239 nmol/4 h/mg vs. 851 ± 41 nmol/4 h/mg) and largely greater than the activity measured in CI progeny cells (0.6 ± 3.5 [SD] nmol/4 h/mg) (Figure 1B).

Figure 1.

HSC GT restores IDS activity at supranormal levels and rescues neurocognitive, neuromuscular, and skeletal phenotype of treated Ids^−/ mice, prolonging their survival

(A) Vector copy number (VCN) in the cell culture progeny of the transduced HSPCs (TI HSPCs) compared to VCN retrieved from the pool of PB mononuclear cell CFCs of transplanted pre-symptomatic and symptomatic Ids^−/ mice. Individual animal data and mean values ±SD are shown. (B) IDS enzyme activity of the LC progeny of the transduced HSPCs (TI HSPCs; CI Ids^−/, WT CI, WT_CI) compared to IDS enzyme activity in murine PBMCs 6–8 weeks after transplantation in pre-symptomatic and symptomatic Ids^−/ mice and in control Ids^−/ and WT mice. Individual animal data and mean values ±SD are shown. (C and D) Behavioral assessment of experimental animals at mid-term (MT) and long term (LT): working memory, expressed as percentage of correct alternation at the Y-maze test (C), and motor learning skills at rotarod test (D), expressed as differential latency to fall at trial 5 vs. trial 1. (E–G) Representative measurements obtained from micro-CT scan images at MT and LT evaluation of experimental animals, including the head width (E), the zygomatic arch thickness (F), and the humerus thickness (G), with representative pictures shown in (H). Data are shown as mean (SEM). (I) Survival curves of experimental animals. The gray box indicates the timing of scheduled sacrifices. One-way ANOVA with Dunnett’s multiple comparisons test was applied to determine statistical differences (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05).

TIs and CIs were transplanted intravenously (i.v.) into Ids^−/ and WT recipient mice according to a study set up to mimic as closely as possible what would be conducted in a clinical setting (Figure S1A). Pre-symptomatic (6–8 weeks old) Ids^−/ mice (n = 20) and symptomatic (12–16 weeks old) Ids^−/ mice (n = 10) transplanted with the TIs were included and compared to Ids^−/ (n = 20) and WT (n = 10) mice transplanted with Ids^−/ and WT mock-transduced HSPCs, respectively (Tables S1 and S2). In WT transplanted mice, donor/recipient cell chimerism was determined by CD45 allele mismatch. Untreated syngeneic Ids^−/ controls (n = 10) were included in the analysis. Pre-symptomatic mice were conditioned using a busulfan-based myeloablative regime,²² while symptomatic recipients were myeloablated with total body irradiation (TBI) after receiving ERT for 2 weeks to mitigate any effect of accumulated GAGs on engraftment of the transduced cells (Figure S1A). As busulfan has no immunosuppressive effects, TBI was chosen over busulfan for the symptomatic cohort to minimize the presence of immune responses to Ids transgene product after exposure to ERT. After conditioning administration, mice received a cell dose of 0.9–1.2 × 10⁶ cells/animal, equivalent to approximately 40–60 × 10⁶ cells/kg, which exceeds the dose of human CD34+ cells normally used in clinical trials of HSPC-based therapy (10–20 × 10⁶/kg).^17,18,19,20 After transplant, mice were monitored 3 days/week for early identification of intercurrent deaths (ICDs). Among the transplanted mice, eight out of 61 transplanted mice died of BM aplasia, corresponding to the 13%, which is in line with other studies applying lethal myeloablative BM conditioning.²³ ICDs were randomly distributed throughout all the groups. At 6–8 weeks after transplantation, short-term evaluations were performed to verify engraftment of the transduced cells and hematopoietic reconstitution. At hemocytometric analysis, all animals showed hematologic recovery in all blood lineages (Figure S1B). Donor cell engraftment in peripheral blood (PB) of WT animals transplanted with CI^WT was 87.5% ± 5%, as determined by the proportion of CD45 cells that were CD45.1⁺ at flow cytometry. CFC assays were performed from PB mononuclear cells of transplanted and control mice (Figure S1C) and VCN analysis was performed on genomic DNA extracted from the pool of CFCs. A successful engraftment of the genetically modified cells was demonstrated by high VCN on colonies both in pre-symptomatic and symptomatic Ids^−/ mice (Figure 1A), consistent with the VCN detected in the TIs. Importantly, this was associated with rescue of IDS enzyme deficiency and robust IDS activity reconstitution in peripheral blood mononuclear cells (PBMCs) from Ids^−/ TI recipients that peaked at up to ≥40 times the WT enzyme levels (Figure 1B).

HSC GT rescues the phenotype of the treated Ids^−/ mice

All animals were observed for clinical signs and symptoms associated with the disease. At 4 or 6 months post transplant (in pre-symptomatic and symptomatic recipients, respectively), all treated and control animals underwent behavioral testing for neurocognitive and neuromuscular analysis and imaging by whole-body computed tomography (CT) to evaluate skeletal deformities. Then, 50% of the mice were euthanized for terminal evaluations. The remaining half of each cohort underwent further assessment for disease-related phenotype until scheduled long-term evaluations performed after 500 days of age when recipients were sacrificed and analyzed for the same terminal parameters.

Spatial working memory was evaluated using the Y-maze test, showing a decrease in spontaneous alternation between WT and untreated Ids^−/ mice (Figure 1C). At both mid- and long-term evaluations, a complete normalization of spontaneous alternation to WT levels was observed in pre-symptomatic and symptomatic mice treated with HSC GT. No differences in the number of total entries were detected between all tested groups, suggesting a true phenotypic rescue of cognitive symptoms in all treated animals, and no difference in performance with regard to absence vs. presence of symptoms at time of HSC GT administration was noticed. Neuromuscular coordination and learning memory were evaluated by the rotarod test (Figure 1D). The untreated/mock-treated Ids^−/ control animals, which were grouped in the same stage of the disease and age range, performed the test very poorly, with high internal variability. Treated mice showed a significant improvement of their performance as compared to untreated/mock-treated Ids^−/ controls, reaching a comparable performance to WT reference animals. Long term, an improved performance in neuromuscular coordination was recorded in HSC GT-treated mice, particularly of the animals treated at pre-symptomatic stage, as compared to the control groups, which were drastically impaired in motor coordination, balance, and learning. CT scans were performed to evaluate the skeletal abnormalities associated with MPS II and their modification upon treatment. Measurement of head width, zygomatic arch thickness, and humerus thickness 3D reconstruction of the scans revealed significant differences in untreated and mock-transplanted Ids^−/ mice compared to WT mice transplanted with CI^WT (Figures 1E–1H). The recorded disease-associated skeletal abnormalities (enlarge head, zygomatic arch of increased thickness, and humerus of increased thickness) were substantially prevented/corrected in pre-symptomatic and symptomatic Ids^−/ mice transplanted with transduced cells in the HSC GT groups. Within the symptomatic-treated cohort, depending on the parameter considered, normalization or amelioration of the abnormalities was observed. Ids^−/ mice have natural life span ranging between 10 and 14 months (300–430 days). Long-term follow-up survival evaluation was performed on the recipients not euthanized at the mid-term evaluation. Control Ids^−/ mice started showing very prominent disease-associated phenotypes at around 10 months and most of them deceased or were euthanized because of the overwhelming disease manifestations (humane endpoint) (defined altogether as ICDs) and were analyzed. In contrast, all the HSC GT-treated recipients were subjected to the terminal assessment upon their natural death (humane endpoint) at ∼530 days post transplantation. The resulting survival curves and proportion of animals that reached the end of study time points are reported in Figures 1I and S1D.

Consistent with these data, a robust and sustained engraftment of the transplanted cells was detected in the BM of the treated animals both at mid- and long-term time points, with average VCN values at study termination of 6.4 ± 2.3 in pre-symptomatic and 8.8 ± 0.04 (mean ± SD) in symptomatic cohorts (Table S3 and Figure S1E). IDS activity was restored in the BM of mice transplanted with the GT product up to above-normal IDS values (∼6-fold WT, on average) (Figure 2A). Notably, the transduced cells also engrafted in the brain of the treated animals, with higher VCN values detected in mice treated when symptomatic (Table S3 and Figure S1E). In the brain, IDS activity was also restored to approximately 20% to 40% of the WT reference values, on average (Figure 2B), with the highest activity levels in animals treated at a symptomatic stage of the disease. Interestingly, very robust IDS activity was also measured in the serum of treated animals (Figure 2C). GAG quantification on the liver of treated and control mice showed a significant reduction down to WT levels in the HSC GT-treated mice as compared to mock-treated Ids^−/ controls (Figure 2D). In the brain, GAG accumulation was observed in mock-treated Ids^−/ animals vs. WT controls, which was prevented/corrected to WT values by the treatment (Figure 2E). Tissue morphologic abnormalities associated with MPS II were studied on the tissues retrieved at necropsy. Morphologic findings considered consistent with the expected progression of the MPS II disease were evident on the tissues retrieved from the mock and untreated Ids^−/ mice (Figure 2F), including cell vacuolation and necrosis in various tissues, including brain and spinal cord. Comparing all animals, regardless of cohort (pre-symptomatic or symptomatic), sacrifice schedule (mid- or long-term or unscheduled), these disease-related findings were less frequent and/or less pronounced in MPS II mice treated with GT (Figure 2F). Notably, while some neuronal vacuolation was observed in GT-treated animals from all groups, little or no neuronal necrosis was observed, suggesting the treatments prevented or at least ameliorated the more severe sequelae. GFAP and Iba1 immunofluorescence staining and signal quantification revealed a trend toward normalization of the pathologic reactive gliosis in the treated mice vs. controls, with some variability (Figures 2G and 2H).

Figure 2.

HSC GT effectively and safely restores IDS activity in vivo, resulting in reduced neurotoxicity, storage accumulation, and gliosis of treated Ids^−/ mice

(A‒C) IDS activity quantified in BM (A), brain cell lysate (B), and blood serum (C) of mice sacrificed at MT and LT evaluation. Individual animal data and mean values (SD) are shown. (D and E) GAG concentration in liver (D) and brain cell lysates (E) from mice sacrificed at MT and LT evaluation. Individual animal data and mean values (SD) are shown. (F) MT and LT pathology data: quantification of the frequency of vacuolation/necrosis of increasing severity (as indicated) on brain, spinal cord, liver, and kidney samples retrieved from treated and control animals (data are expressed as percentage of samples showing no, minimal, mild, or moderate findings/total examined animals, as per pathologist scoring). (G and H) Staining for Iba-1 (G) and GFAP (H) antigens in the cortex of the experimental mice and signal quantification. Individual animal data and mean values (SD) are shown. One-way ANOVA with Dunnett’s multiple comparisons test was applied to determine statistical differences (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05).

HSC GT was well tolerated by the treated animals

Safety data, such as body weight, food consumption, and clinical signs, were also collected along the entire course of the study. As expected, transient body weight loss was observed in all the animals that received conditioning (Figure S1F). In total, 24 mice spread across all groups were euthanized due to poor body conditions before the scheduled termination. At sacrifice, these animals, as well as terminally euthanized mice, underwent hemocytometric analysis on PB, clinical chemistry and evaluation of cell composition by flow cytometry, full necropsy, and tissue collection. At necropsy, macroscopic observations revealed that disease-associated organomegaly was prevented by HSC GT in animals treated at both pre-symptomatic and symptomatic stages (Figure S1G). In the animals euthanized due to poor body conditions before the scheduled termination, abdominal ascites was observed in the abdominal cavity, frequently accompanied by the presence of a solid tumor mass. These findings were observed only in the groups that received busulfan intraperitoneally (i.p.), as previously reported.²⁴ At hemocytometric and blood cell composition analysis, no significant differences were observed between all the groups, suggesting that long-term hematopoiesis was preserved in all treated groups (Figure S1H). An extensive histopathological examination was also conducted by a board-certified pathologist at an external CRO. No major findings were noticed in GT-treated mice. Mice transplanted with mock-transduce Ids^−/ cells and untreated Ids^−/ mice showed typical disease-associated signs such as hepatosplenomegaly, and skeletal abnormalities, including thickened digits, swollen hocks, and distorted craniofacial bones (see below). No unexpected findings or pathologies suggestive of toxicity related to the administration of the GT product were reported in the treated mice. Minor findings were considered spontaneous and/or incidental, because they occurred at similar incidence and severity to normal WT control groups or were observed as rare individual findings. Such findings were as expected for mice and were not considered GT product related.²⁵

A novel protocol allows efficient transduction of human HSPCs by the therapeutic LV at large scale

To implement a transferability plan for this therapeutic approach, in vitro and in vivo studies were developed with the vector employed in the PoC of efficacy study. A pre-GMP viral batch of the hIDS_LV was produced in the process development lab, but representative of the GMP process tested, characterized, and released by AGC Biologics. This vector was employed to produce drug product (DP) equivalents for testing, made of human CD34⁺ HSPCs derived from mobilized PB of healthy volunteers of commercial origin transduced with the therapeutic LV (pre-GMP batch) or with another LV, used as control, produced in large scale for an unrelated indication (encoding the human programmed death ligand 1 [PD-L1_LV]), used as control, by AGC Biologics in medium scale (in a range of 20–80 million cells per condition). Two transduction protocols were tested with the hIDS_LV: a protocol already employed for a commercial DP, characterized by two overnight rounds of transduction at MOI 100 over a total of 60 h of in vitro culture (A), and a new protocol optimized by AGC Biologics characterized by a single round of transduction at MOI 80 over a total of 24 h of in vitro culture (B). Only the new protocol was tested in the case of the PD-L1_LV. Control mock-transduced cell CIs were also generated as per the same two protocols. After transduction, the DPs and CIs were cryopreserved and made available with release specifications to the laboratory for intended use (Table S4). By immunophenotype analysis, the DPs produced with the two processes differed mainly in the relative number of primitive HSCs and committed progenitors, with a significantly higher content of primitive HSCs (CD34⁺/CD90⁺) and uncommitted (CD34⁺/CD15⁻/CD19⁻, CD34⁺/CD15⁻) progenitors in the DP produced with the AGC process (Figures 3A and S2). At thawing, part of the DPs and CIs were maintained in LC as a bulk cell population or plated as individual progenitors for a CFC-assay, and part was saved for transplantation into immunodeficient mice for an in vivo safety/biodistribution study described below. CFC assay resulted in a comparable number of colonies independently from the transduction process (not shown). Pools of CFCs and LC cells were collected to evaluate the VCN by digital-droplet PCR (ddPCR). Interestingly, an efficient and comparable transduction of human HSPCs by the two protocols was observed, with the new protocol equally performing with the two different LVs (Figure 3B). A trend toward higher VCN detected on CFCs vs. LC cell progeny was noticed. IDS activity determined in the LC cells resulted ∼3,000 and ∼4,000 nmol/mg/h in DP-A and DP-B respectively, representing a 13- to 22-fold increase with respect to the corresponding CIs (Figure 3C).

Figure 3.

A novel transduction protocol allows efficient transduction of human HSPCs and efficient repopulation by gene-modified human HSCs of an immunodeficient mouse model

(A) Immunophenotypic analysis of human CD34⁺ HSPCs transduced with the therapeutic hIDS_LV employing the process A (DP-A) or the new process (DP-B). Frequency of primitive HSCs (CD34⁺/CD90⁺), uncommitted progenitors (CD34⁺/CD15⁻/CD19⁻), and myeloid committed progenitors (CD34⁺/CD15⁺) evaluated in the LC progeny of transduced cells after 14 days post transduction is shown. As control, immunophenotypic analysis of human CD34⁺ HSPCs transduced with the unrelated LV employing the new process (UDP-B) is reported. (B) VCN detected on the LC and on pooled CFC cell progeny of DP-A, DP-B, and UDP-B. (C) IDS activity assessed in the liquid-cultured progeny of DP-A, DP-B, UDP-B, and the corresponding control mock-transduced cell items (CI-A, CI-B). Mean values (SD) are shown. (D and E) Frequency of human CD45⁺ cells in hematopoietic organs (D) (BM, spleen [Spl], thymus [Thy]) and in brain (BR) and spinal cord (SC) (E) of immunodeficient NSG recipients transplanted with transduced (DP-A, DP-B, UDP-B) or mock-transduced (CI-A, CI-B) cells and sacrificed at 14 weeks post transplant. Individual animal data and mean values (SEM) are shown. (F) Human-cell VCN retrieved in tissues collected from mice transplanted with DP-A and DP-B; human-cell VCN retrieved in the BM of mice transplanted with UDP-B is also reported. Individual animal data and mean values are shown. (G) Fold increase of the VCN measured in CFC progeny cells (in vitro) and in the BM of transplanted mice (in vivo) over the VCN values measured in the corresponding LC progeny cells (in vitro). The data are calculated according to the following formulas: VCN in CFCs/VCN in LC and VCN in BM/VCN in LC. Individual data and mean values ±SD are shown. (H) IDS activity detected in BM, spleen, and liver of mice transplanted with DP-A, DP-B, and CI_A&B. Individual data and mean values are shown.

The novel protocol with the therapeutic LV transduces at high efficiency NSG-repopulating human HSCs

Thawed DPs and CIs were xenotransplanted in myeloablated immunodeficient mice to evaluate their long-term engraftment capacity, repopulation ability, biodistribution, and absence of long-term toxicity associated with the cell product. Female Nod-Scid-Gamma (NSG, NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ) mice 6–8 weeks old were used as test system in this study. Female mice were preferentially used because human HSPC engraftment is significantly more efficient than in male mice^26,27; nevertheless, a small cohort of males was also included, being MPS II an X-linked condition. Mice received a myeloablative conditioning regimen with i.p. busulfan administration and were then transplanted with 0.75–1 × 10⁶ CD34⁺ cells (equivalent to 18.75 × 10⁷ cells/kg) by tail vein injection, as per Table S5. Four days after human cells transplant, recipient animals received 4 × 10⁶ unmanipulated total BM cells from NSG donors to rescue them from myeloablation. The unrelated DP (UDP) was infused according to the same experimental design and protocol in the setting of a separate experiment (Table S6). After cell transplantation, mice were monitored daily for general health status and survival, and weekly for body weight; animals experienced transient weight loss as an effect of conditioning. Overall, 11 animals (seven males and four females, belonging to different groups) were reported as ICDs, as they died or reached humane endpoint for sacrifice (body weight loss >20%) before study termination. No ICDs were registered in the UDP animal study. By gross body examination, the cause of dead was likely related to the conditioning regimen (i.e., BM aplasia or residual busulfan solution in the abdomen). The study lasted 14 weeks post transplant, as human-cell engraftment declines afterward, as reported.^27,28 At the end of the study, a blood sample was collected for hemocytometric analysis, and the mice were euthanized. An extensive organ perfusion was performed with saline solution, and a full necropsy was conducted before collecting the organs for further evaluations. White blood cell (WBC), red blood cell, hemoglobin, and platelet counts were comparable in all the animals, irrespective of the group (Figures S3A and S3B). Hematopoietic organs (blood, BM, spleen, thymus), liver, gonads, brain, and spinal cord were collected and processed for evaluating human-cell engraftment and biodistribution by multicolor flow cytometry and ddPCR, respectively, and to determine human-cell VCN. Organ-to-body-weight ratio assessed in spleen, liver, and kidneys showed lack of abnormal expansions and absence of any significant differences between mice receiving DPs or CIs and untreated animals (Figure S3C).

A robust human-cell engraftment in hematopoietic organs (Figure 3D) was documented in all the tested conditions and experimental groups, with no significant differences among the groups. The engrafted human cells showed a multilineage differentiation (Figure S3D). Engraftment of the human myeloid cells in the CNS was also clearly documented, at levels consistent with previous reports^23,29 and with no significant differences between the groups (Figure 3E). Human-cell engraftment data were confirmed by ddPCR, highlighting a minimal human-cell biodistribution in non-hematopoietic organs (Figure S3E). Human-cell VCN in the different collected organs, including non-hematopoietic ones, was also quantified by ddPCR; the measured VCN was very consistent across the different tested organs in each experimental group, resulting in ∼5 and ∼10 in groups A and B, respectively (Figure 3F). The VCN measured in the BM of animals transplanted with unrelated LV employing the new process (UDP_B) was ∼5. Notably, these values were substantially higher than those measured in the in vitro LC cell progeny of the infused products, with the in vivo VCN values representing on average a 2.5-fold (group receiving DP_A) and >5-fold (groups receiving either DP_B or UDP_B) increase over the corresponding infused cell products (Figure 3G). The VCN values measured in the CFC cell pools were intermediate between the in vitro LC and the in vivo data (Figure 3G). The higher VCN values measured in the tissues of the animals receiving DP-B resulted in the correspondingly higher IDS enzyme activity values measured in the BM and spleen of group B mice with respect to group A and control mice (Figure 3H). The IDS activity values in the organs where human-cell engraftment was lower than in the BM and spleen were not detectable over the background, the basal IDS activity of the murine WT tissue (not shown).

The therapeutic hIDS_LV has a safe vector integration profile

An extensive lentiviral analysis of hIDS_LV integration sites was performed in vitro in the DP-A and DP-B cells and in vivo in a pool of BM cells collected from each individual mouse at the end of the study. LV integration sites (ISs) were recovered by ligation-mediated PCR (LM-PCR) followed by high-throughput Illumina sequencing as previously described,³⁰ followed by an established bioinformatic read processing³⁰ to retrieve the hIDS_LV ISs in the human genome (assembly Hg38). We obtained ∼30,000 univocal ISs in DP-A and DP-B, and 5,590 and 8,282 ISs in vivo, in the pool of BM cells from group A (BM-A) and B (BM-B) mice, respectively. As expected, hIDS_LV showed the canonical integration profile in the human genome, with a marked preference for intronic and intergenic regions, independently from the transduction protocol (Figure 4A). Intragenic integration events targeted 6,620 and 6,698 genes in DP-A and DP-B, respectively, 70% of which was in common (Figure 4B). In the set of the preferential LV target genes in DP-A and DP-B (Figure 4B), we retrieved most of the genes frequently targeted in pre- and post-transplantation samples in multiple LV-based human GT trials studies (e.g., DACH1, FANCA, FCHSD2, KDM2A, NPLOC4, PACS1).^{20,31,32,33,34,35,36} More than 80% of the LV target genes retrieved in vitro were maintained in vivo in the corresponding BMs, suggesting the absence of a substantial in vivo selection for cell clones carrying LV integrations in particular genes (Figure 4C). The number of original reads accounting for a particular IS can be used to estimate the relative abundance of a cell clone harboring a particular LV integration. Our analysis showed a very polyclonal cell population in the BMs, with the absence of relatively abundant clones in vivo, independently from the LV transduction protocol (Figure 4D).

Figure 4.

The therapeutic hIDS_LV has a safe vector integration profile

(A) Genomic distribution of LV ISs in the infused DPs and the BM of recipient mice. (B) On the top is shown a Venn diagram reporting the total number of LV-targeted genes identified in each individual DP in parentheses, the number of LV-targeted genes shared between the DPs, or univocally identified in each DP. On the bottom, Word Cloud charts showing the top 100 most targeted genes by hIDS_LV in the two DPs; the size of each word is proportional to the relative vector integration frequency. (C) Pie charts reporting the frequency of LV-targeted genes identified in the infused DPs and in the recipient BMs ex vivo. (D) Most abundant cell clones identified by IS read count in the pool of BM cells collected from mice receiving each individual DP. (E) IVIM RF of the control samples mock transduced (MOCK) or the mutagenic γ-retroviral vector RSF91 and the test vector hIDS_LV, compared to data of a meta-analysis for control samples (Mock-MA, RSF91-MA, and LV-SF-MA, an SIN LV containing the mutagenic γ-retroviral SF promoter in the internal position). Differences in the incidence of positive and negative assays relative to Mock-MA or RSF91-MA were analyzed by Fisher’s exact test with Benjamini-Hochberg correction (∗∗∗p < 0.001; ∗∗p < 0.01; ∗p < 0.05; NS, not significant). Bars indicate mean RF. (F) Principal-component (PC) analysis of gene expression data from samples transduced with mutagenic vectors (light red), neutral vectors (light blue), and MOCK controls from previous assays (light gray). Current samples are depicted in red (RSF91), black (MOCK), and blue (hIDS_LV).

The excellent safety profile of the hIDS_LV was confirmed in an in vitro immortalization (IVIM) assay and a surrogate assay for genotoxicity assessment (SAGA). As we proved the substantial equivalence of the hIDS_LV integration characteristics with the A and B transduction protocols, we performed IVIM and SAGA assays only with DP-B, which resulted in the highest VCN in vivo, therefore retaining a higher genotoxic potential. In three independent transductions of murine lineage⁻ HSCs, hIDS_LV did not elicit the growth of insertional mutants in the IVIM assay (Figure 4E). Furthermore, in eight out of nine transductions, the vector did not perturb the normal gene expression profile in the SAGA (Figure 4F). Overall, the hIDS_LV resulted in the range of the reference low genotoxic LVs in both the in vitro assays, indicating a safe and low genotoxic potential.

Discussion

Despite the advances in designing enzyme replacement therapies, MPS II still represents a condition with unmet medical need, especially when it severely affects the CNS. Lentiviral-based ex vivo GT is a one-time therapy that aims at targeting the underlying genetic cause of the disease, offering the potential for long-term correction or amelioration of cognitive development and maintenance of motor function, based on stable production and release of the therapeutic enzyme in the CNS and systemically. This potential for a durable therapeutic effect is testified by the first HSC GT registered for an LSD (atidarsagene autotemcel [arsa-cel]) that has recently received full market authorization in Europe for patients with early-onset metachromatic leukodystrophy.¹⁸ The potential of HSC GT has been also already tested in the context of MPS II in preclinical animal disease models.^37,38 However, the clinical relevance of these early studies is limited as they have employed experimental conditions and reagents that are not applicable or informative for clinical translation. Moreover, these studies did not challenge the most frequent clinical scenario in Hunter syndrome, which is represented by patients who show neurological manifestations of the disease and are already undergoing ERT at the time they would be considered for GT administration. Thus, to obtain clinical-trial-enabling data on the efficacy of HSC GT for MPS II and overcome the limitations of previous approaches, we designed a study that challenged the therapeutic potential of HSC GT in both young pre-symptomatic and symptomatic Ids^−/ mice, exploiting an LV encoding the human Ids cDNA and based on a backbone already extensively tested in the clinics, as well as clinically relevant treatment and conditioning protocols. The overall purpose of the study was to assess the efficacy, feasibility, and safety of an ex vivo GT approach for MPS II, employing first murine HSPCs transduced with the clinical candidate LV (hIDS_LV) produced in laboratory scale and the animal model of the disease in a PoC design challenging symptomatic treatment in animals undergoing ERT besides pre-symptomatic mice. In the PoC study, we did not include an allogeneic transplantation group (Ids^−/ mice receiving WT cells) as clinical practice extensively demonstrated the limited efficacy of allogeneic HCT in MPS II patients, in whom the transplant is not effective on skeletal, corneal, and cardiac abnormalities, nor can it ameliorate cognitive and intellectual disease manifestations,²⁵ likely because supraphysiological levels of enzyme expression in donor cells are needed to favor the cross-correction of deficient recipient cells.

The study highlighted the benefits associated with HSC GT with the candidate hIDS_LV in the murine model of the disease. The high transduction efficiency and consequent above-normal IDS activity in vitro resulted in significant increase of IDS activity that reached highly supraphysiological levels, up to >40-fold the WT enzyme one, in the PB of treated animals at dose proof. At mid-term evaluation and up to the end of the study, 17–18 months post transplantation, a robust and sustained engraftment of the infused cells was reported, indicating an efficient LV transduction and engraftment of long-term HSCs, leading to a long-term supraphysiological IDS enzymatic activity in BM mononuclear cells (>10-fold above WT enzyme levels) and serum. Similarly, in the brain, engraftment of the transplanted cells/their progeny was testified by VCN values in a range consistent with previous data,²³ with a trend toward greater VCN values in symptomatic treated animals. This would suggest that disease-associated abnormalities, i.e., neuroinflammation, present at time of transplant, might have favored the homing, migration, and engraftment of the transplanted cells and/or their progeny. Engrafted myeloid cells in the brain restored IDS brain tissue activity to approximately 15% of WT levels, with a trend toward higher values in the animals treated in symptomatic stage, consistent with VCN values. Notably, the VCN and IDS activity values measured on the mid- vs. long-term cohorts were superimposable, indicating that, already at mid-term evaluation, namely between 4 and 6 months post transplants (in pre- and symptomatic-treated mice, respectively), a plateau of transduced cell engraftment in the brain and enzyme delivery was reached. The supplemental IDS activity measured in the CNS was sufficient to determine a reduction of GAG accumulation in the brain down to normal levels. The correction of GAG accumulation was also detected systemically, as measured in the liver. This is in line with clinical data where 5%–15% of the normal enzyme level is required to maintain a healthy condition and have a positive impact on clinical manifestations.^39,40 Indeed, HSC GT with hIDS_LV resulted in normalization of the behavioral and neurocognitive disease manifestations of the MPS II mice treated as pre-symptomatic and symptomatic, as determined by the Y-maze and rotarod tests. Similarly, benefit was also reported on the skeletal phenotype, which was normalized by HSC GT administration. Overall, these effects ultimately resulted in a significantly prolonged survival of the Ids^−/ mice treated either at pre- or symptomatic stage. This effect was also documented by the lower incidence of ICDs in the GT-treated cohorts enrolled in the long-term evaluation as compared to control animals. Pathological analysis confirmed a strong reduction or complete absence of vacuolated neuronal degeneration and necrosis in the CNS and other organs of the treated mice compared to Ids^−/ controls. Overall, these results indicate that the proposed HSC GT approach based on clinically applicable vector and clinically relevant settings has potential for preventing and correcting the MPS II biological and clinical manifestations. In particular, the hIDS_LV was able to efficiently and safely transfer multiple copies of a therapeutic IDS cDNA in long-term repopulating HSCs that could deliver, through their mature progeny, therapeutic levels of IDS enzyme to the brain and peripheral tissues, including the skeleton, of the treated animals whose phenotype was prevented and/or corrected. The long-term time point of observation (up to 18 months of age, >12 months post treatment) suggests that these ameliorations are stable over time. Of note, the recently proposed approach of HSC GT exploiting the use of a brain-targeted enzyme still needs to be observed beyond the 8-month time point,⁴¹ as well as similar approaches employing LVs engineered with blood-brain barrier-crossing peptides that still failed to achieve high levels of IDS enzyme activity in the brain, consequently with only partial amelioration of MPS II phenotype in mice.⁴² To translate these results toward clinical application, we tested the therapeutic hIDS_LV for its ability to efficiently and safely transduce human HSCs in clinically relevant conditions according to a standard vs. a novel protocol that demonstrated superior ability to transduce bona fide long-term repopulating HSCs. Pre-GMP viral lots of the hIDS_LV were tested in a standard protocol already employed for a commercial DP and compared to a newly optimized protocol characterized by a shorter time of in vitro culture and lower MOI, features that are relevant for clinical feasibility of HSC GT in terms of drug manufacturing. Both protocols resulted in efficient transduction and induction of robust IDS expression and activity in human HSPCs when tested in large-scale, clinically informative conditions. The two processes differed mainly in the resulting relative amount of HSPC subpopulations at end of transduction, with a significantly higher content of primitive HSCs and uncommitted progenitors in the DP produced with the new process. This finding was confirmed upon transduction with an unrelated LV, indicating that the observed result is independent from the transgene encoded by the integrated cassette. The human HSPCs transduced with both protocols similarly showed a preserved functionality in repopulating immunodeficient mice in a biodistribution and tolerability study, with a sustained and comparable to mock-transduced cells long-term engraftment and multilineage differentiation in all tested organs, including brain. Interestingly, however, a homogenously higher LV content (VCN) was measured in the hematopoietic and in the other tested organs of animals receiving human HSCs transduced with the newly developed process with respect to mice transplanted with human HSCs transduced with the standard protocol. Overall, the VCN measured in vivo was >5× the in vitro LC VCN with the new protocol, vs. a previously observed slight increase in the range of 2.5×, with CFC VCN values being between the in vitro and the in vivo values for both the conditions. This increase in vector context in group B vs. A animals was paralleled by a higher IDS enzyme activity in the ex vivo-tested samples. These findings could also be interpreted considering the enrichment of long-term HSCs vs. progenitors observed with the new vs. old protocol and could depend on a preferential transduction of the more primitive HSCs that would be ultimately responsible for long-term in vivo engraftment. Recently, the preferential expansion of HSCs in vitro has been shown to increase the frequency of in vivo repopulating gene-modified cells using a combination of small molecules in the ex vivo culture,⁴³ corroborating our findings. Indeed, the newly developed process, mostly thanks to strict cell-to-cell contact, could similarly determine a preferential proliferation of HSCs vs. more committed progenitors in vitro, resulting in enhanced susceptibility of HSCs for gene modification and consequently in the high frequency of efficiently gene-modified cells found in vivo.

Initially, we reasoned that upregulation of IDS enzyme activity could be contributing to the enhanced engraftment of highly transduced HSCs because of the enzyme overexpression perturbing the sphingolipid pathway. Indeed, sphingolipid modulation has been shown to be associated with autophagy activation in the HSCs, but not in the progenitors, restricting the proliferation of progenitors in the ex vivo culture.⁴⁴ However, increased VCN values were also retrieved in vivo in mice transplanted with HSCs transduced with an unrelated LV (PD-L1_LV) with the new protocol B, suggesting that culturing conditions are the main factor responsible of the findings observed. Notably, a similar phenotype of the transduced cells was also observed at end of transduction employing the new protocol with the PD-L1_LV. Overall, these findings suggest that the progenitor cells and HSCs respond differentially to culture conditions and that the newly developed protocol, favoring the transduction and survival of long-term HSCs, could induce the enrichment of highly transduced immature cells capable of long-term engraftment and differentiation in vivo. This may be of great relevance as, over the years, many efforts and extensive research have been conducted to expand and/or maintain primitive HSCs during transduction.⁴⁵ However, a clinically relevant protocol that could guarantee maintenance of these features along a short in vitro culture, achieving high levels of transduction at a low vector input (one round of transduction), would be highly desirable, particularly for conditions such as MPS II where high levels of therapeutic gene expression are needed to obtain significant therapeutic benefit. Notably, this was obtained without the use of transduction enhancers (with the exception of RetroNectin bag coating) and without addition to a standard cytokine cocktail, making the novel protocol highly amenable to clinical translation.

Importantly, the use of the hIDS_LV in both the mouse-into-mouse and human-into-mouse transplant settings was well tolerated. Indeed, HSC GT was well tolerated by the Ids^−/ animals, as the infusion of the genetically corrected HSCs resulted in a complete hematologic recovery after myeloablation and up to long-term assessment, and clinical chemistry and pathological analysis revealed no unexpected findings or alternate pathologies suggestive of toxicity related to the administration of the GT product in the transplanted animals. Abdominal ascites, frequently accompanied by the presence of solid tumor mass, were observed only in the groups that received busulfan i.p., independently from the GT treatment, thus likely ascribing these findings to busulfan conditioning, as previously reported.²⁴ The human-into-mouse study, showing no abnormal findings in animals receiving treated vs. control cells, also demonstrated a safe vector integration profile of hIDS_LV resulting in a very polyclonal cell population in the BM, with absence of relatively abundant clones in vivo, independently from the LV transduction protocol used. These findings were confirmed by the IVIM and SAGA assays. Overall, based on these data, we propose an immediate clinical translation of this treatment approach in MPS II patients.

Materials and methods

LV

Human IDS cDNA was codon optimized and synthesized by GenScript (NJ, USA) and was cloned into a third-generation lentiviral pCCL backbone under the control of a human PGK promoter to produce pCCLsin.cPPT.hPGK.IDSco.WPRE LVs (hIDS-LVs). The human PD-L1 cDNA was also cloned in the same backbone under the control of the PGK promoter to produce pCCLsin.cPPT.hPGK.PD-L1.WPRE LVs (hIDS-LVs). LVs were produced and tittered according to previous published protocols.²¹

Animal genotyping and conditioning

MPS II (Ids^−/) mice were obtained from The Jackson Laboratories (B6N.Cg-Ids^tm1Muen/J, stock no. 024744, Bar Harbor, ME). Ids^+/− females were bred with Ids⁺ (WT) males from the same strain to generate Ids^−/ mice (MPS II) males and WT males used in the study. CD45.1 mice were obtained from The Jackson Laboratories (B6.SJL-Ptprca Pepcb/BoyJ). All the mice were maintained at the Boston Children’s Hospital animal research core facility. Procedures involving animals and their care were conducted in conformity with the institutional guidelines according to the international laws and policies (EEC Council Directive 86/609, OJ L 358, 1 December 12, 1987; NIH Guide for the Care and Use of Laboratory Animals, US National Research Council, 1996). The specific protocols Institutional Animal Care and Use Committee (IACUC) #16-07-3170 and #19-06-3942R, covering the studies described in this paper, were approved by the Boston Children’s Hospital IACUC. Mice were identified by genotyping the toe tissue sample collected by toe clipping for the purpose of identification at around 8–10 days of age. Stock no. 024744, Protocol 27335: Standard PCR Assay - Ids<tm1Muen>-Alternate 2 from Jackson Laboratory was used. Three different primers were used: (1) 5′ GGAAACTGAACCCCAAAGA 3′ common forward primer, (2) 5′ GAGGATGGATGATAAGGT GGA 3′ WT reverse primer, and (3) 5′AAAAGAGGACTGCGTGTGGG3′ mutant reverse primer, which, upon PCR amplification, results in different band sizes as mutant, 190 bp; Het, 185 and 190 bp; and WT, 185 bp. Before cell transplantation, (1) pre-symptomatic recipient mice were conditioned using a previously optimized busulfan-based myeloablation regimen (27 mg/kg i.p. for four consecutive days) ²²; (2) symptomatic recipient mice were treated with ERT, idursulfase (Elaprase, Shire Human Genetic Therapies, Cambridge, MA) (10 mg/kg i.p., four doses) 2 weeks before transplantation and were myeloablated with TBI at 1,000 cGy 2 h before transplant.

Isolation, transduction, and transplantation of murine HSPCs

Ids^−/ mice or WT CD 45.1 mice were euthanized with CO₂ at 6–8 weeks of age. BM was harvested by crushing the femurs, tibias, humerus, and iliac crest. HSPCs were enriched by lineage depletion (Lin−) of BM using the Miltenyi Biotec Lineage Cell Depletion Kit (Bergisch Gladbach, Germany) with magnetic separation using the autoMACS separator following manufacturer’s instruction. Isolated Lin− cells were transduced with the LVs for 14–16 h at MOI 100 in StemSpan Culture Medium (StemCell Technologies, Cambridge, UK) with Pen/Strep and cytokines (interleukin [IL]-3, IL-6, FLT-3, and SCF) at 37°C according to our standard protocol.²¹ The transduced cells were harvested, washed, and formulated in PBS at a cell concentration of 0.8–1.2 × 10⁶ cells in 110 μL prior to i.v. injection into conditioned recipient mice.

The percentage of c-Kit⁺ Sca⁺ Lin⁻ cells was determined by cytofluorimetric analysis using the appropriate antibodies as described in Visigalli et al.²¹ Then 4 × 10³ hematopoietic progenitors transduced or mock treated were plated in a methylcellulose-based medium (MethoCult M3434, StemCell Technologies, Cambridge, UK) for colony forming unit (CFU) assay to confirm their clonogenic potential. Fourteen days later, colonies were counted and scored for lineage commitment and subsequently harvested. The harvested colonies were washed, pooled, and lysed with proteinase K digest for DNA extraction and ddPCR analysis for the detection of LV sequences. A small fraction of transduced cells (0.2–0.5 × 10⁶ cells) was cultured for 14–16 days in LC (RPMI 1640 medium supplemented with 10% fetal bovine serum [FBS], 2 mM L-glutamine, 1 mM Pen-Strep, mIL3, mFLT3-L, mIL-6, and mSCF) to assess VCN by ddPCR and IDS enzymatic activity assay.

Lin− cells were injected i.v. (0.9–1.2 × 10⁶ cells/mouse) 24 h after the last busulfan dose administration in the pre-symptomatic 6- to 8-week-old MPS II Ids^−/ mice and WT mice and 2 h after TBI into symptomatic 14- to 16-week-old MPS II Ids^−/ mice. The i.v. administration was performed by tail vein injection.

Human HSPC culture and transduction

Human CD34⁺ cells from healthy donors were purified by immuno-magnetic selection from mobilized PB (Charles River Laboratories) using a Miltenyi CliniMACS Prodigy instrument and cryopreserved. After thawing, cells were resuspended in their specific growth medium (CellGenix SCGM for protocol A, serum- and xeno-component-free medium optimized for the expansion of isolated HSCs for protocol B) and then were seeded in RetroNectin-coated bags for pre-stimulation. Twenty million cells (protocol A with hIDS-LV), 35 million cells (protocol B hIDS-LV), and 80 million cells (protocol B with PD-L1_LV) were transduced (two hits with washout in between for protocol A and one hit for protocol B) or mock-transduced (CI) in the same medium of pre-activation at an MOI of 100 (protocol A) and 80 (protocol B). After transduction, cells were washed and resuspended at target cell dose of 5 × 10⁶ cells/mL in saline solution with a final concentration of 7% w/v HSA and 5% v/v DMSO (cryoformulation buffer) for cryopreservation.

Transplantation of human HSPCs

The 6- to 8-week-old female NSG mice (JAX stock # 005557) were purchased from Jackson Laboratory (Bar Harbor, ME, USA) and maintained at the Istituto di Ricerca Pediatrica Città della Speranza animal research facility (Padua, Italy). Procedures involving animals and their care were conducted in conformity with the institutional guidelines according to the international laws and policies and authorized by the Italian Ministry of Health (Authorization no. 447/2021-PR). The specific protocols covering the studies described in this paper were approved by the Istituto di Ricerca Pediatrica Città della Speranza Animal Care and Use Committee.

On the day of the transplant, DPs and CIs were thawed, washed, and resuspended in PBS (Lonza Bioscience, Walkersville, MD, USA), and injected in mouse tail vein. Before transplantation, NSG mice were conditioned using a myeloablative busulfan-based regimen (16.25 mg/kg i.p. for four consecutive days). Five days post transplant, 4 × 10⁶ total BM nucleated cells from syngeneic donors were provided to transplanted mice for support. Mice were maintained for 16 weeks post transplantation, monitoring them as described above. At sacrifice, mice were euthanized under deep anesthesia. BM, spleen, thymus, brain, spinal cord, kidneys, liver, and gonads were collected and differentially processed: BM cells were collected by flushing the femurs with PBS 2% FBS (Lonza Bioscience, Walkersville, MD, USA); cells from other organs were mechanically disaggregated to obtain a single-cell suspension in PBS 2% FBS.

Animal monitoring

After transplant, mice were monitored for early identification of ICDs 3 days/week. In the case of ICDs, proper assessments were conducted as described below. ICDs up to day 31 are expected as consequence of the conditioning regimen; therefore, no specific analyses were performed. ICDs or mice euthanized due to poor clinical conditions (>15% weight loss and hunched posture) occurring from day 31 up to the end of study were processed, whenever feasible, to analyze the following parameters: (1) transduced and mock-transduced cell engraftment on PBMCs; (2) reconstitution of IDS activity in PBMCs, serum, BM cells, and brain tissue; (3) VCN analysis on BM and brain samples; and (4) GAG quantification in liver and brain lysates. Necropsy, histopathological, and immunohistochemical examinations were performed on a selected set of tissues.

PB analysis

At 6–8 weeks after transplantation, the transduced cell engraftment was determined by (1) IDS activity assay on PBMCs; (2) VCN in DNA obtained by CFU assay from PBMCs (same procedure as described for Lin− cells, plating 200 μL of peripheral blood after red blood cell lysis); (3) hemocytometric analysis was performed with HemaVet 950FS (DREW Scientific) hemocytometer on tail vein PB with anti-coagulant (EDTA). The parameters measured were hemoglobin, hematocrit, total red blood cell count, mean cell volume, mean cell hemoglobin, mean cell hemoglobin concentration, platelet count, and total leukocyte count. Engraftment in CT^WT-transplanted mice was monitored by flow cytometry exploiting CD45.1 CD45.2 allele mismatch.

Pre-symptomatic recipients were evaluated 6 months post transplant and symptomatic recipients were evaluated 4 months post transplant. All recipients underwent (in sequential order) behavioral testing and whole-body CT imaging to evaluate skeletal deformities. Subsequently, blood was collected from half of each cohort for IDS activity measurement, hemocytometric analysis, and cytofluorimetry and clinical chemistry, after which the mice were euthanized. Specifically, the aorta, BM, brain, eyes, gall bladder, heart, kidney, liver, lung, lymph nodes, spinal cord, spleen, bone sternum, testes, and thymus were collected, weighed, and processed for biochemical analysis or fixed in neutral buffered formalin for histopathological evaluations. Tissues were processed for analysis of tissue GAG concentration, IDS activity, and VCN.

The remaining half of each cohort underwent further observation for disease-related phenotype until scheduled long-term evaluations performed from 500 days of age. For all the mid-term evaluations, recipients underwent behavioral testing and whole-body CT imaging to evaluate skeletal deformities. Blood samples were collected, and the animals were euthanized for tissue analysis as described above.

Behavioral tests

The Y-maze test evaluates spatial working memory by exploiting the innate preference of mice to explore novel arms over recently explored arms. Spatial working memory is quantitated as spontaneous alternation percentage over a 10-min period. In a correct alteration sequence, the mouse had to move into a new arm each time after leaving a previous arm; e.g., A to B to C or C to A to B. If the animal moved from A to B to A, that was considered an incorrect alternation. The test room was lit using Anglepoise lamps angled away from the Y maze to create diffuse lighting to remove any shadows in the maze. The light level at the end of each arm was ∼30 lux. Each trial lasted a total of 8 min. The trial was started by placing a mouse at the end of an arm (called the start arm for that trial) and the mouse was recorded for a period of 8 min using the video tracking software Ethovision XT v11.5 (Noldus, the Netherlands). The total number of entries into the different arms of the Y maze was used as a proxy measure of overall activity to normalize the impact that skeletal abnormalities may have in the MPS II mouse model.⁴⁶

Neuromuscular coordination and learning memory were evaluated by the rotarod test. Rotarod is a 2-day test: on the first day, mice were trained to walk on the rotarod for 5 min at four rotations per minute (RPM). On the second day, each mouse was placed on the platform at 4 RPM with an acceleration of 0.1, and the time was recorded until the mouse fell off the platform. Each mouse was tested for five times with a break of 3–5 min between each trial, and the latency was calculated as time it stayed on the platform on trial 5 minus trial 1 (T5 − T1). Touchscreen Rota Rod (Panlab) Model 76–0770 (Harvard Apparatus, Holliston, MA) was used for the study.

Computational tomography scansion

CT scans were performed using a Bruker Albira instrument (Bruker Corporation, Billerica, MA), which utilizes an X-ray source with a 35-μm focal spot that can operate at 10–50 kVp in conjunction with a 12 × 12-cm (2,400 × 2,400 pixels) detector to provide spatial resolution between 40 and 120 μm to evaluate the skeletal abnormalities associated with MPS II and their modification upon treatment. Upon 3D reconstruction of the scans, the following parameters were measured: head width, zygomatic arch thickness, femur length, and humerus thickness.

Hematological parameter evaluation

Cytofluorimetric analysis of PB was performed with a BD LSR Fortessa cytofluorimeter at termination. Approximately 50 μL of PB was collected from the tail vein, in EDTA-coated vials. Non-specific binding sites were blocked with CD16/CD32 mouse Fc Block (BD #553142) 1:100 for 10 min at room temperature prior to staining with antibodies against the cell surface markers CD3 (BD #561799 PE), B220 (BD #552772 PECY7), and CD11b (Invitrogen #17-0112-83, APC) used at 1:100 in PBS 2% FBS. After staining, red blood cells were lysed with ammonium-chloride-potassium (ACK) lysing buffer and the percentage of CD3, B220, CD11b-positive cells was analyzed by cytofluorimetry. Flow cytometry data were analyzed with FlowJo 10.8.

Murine tissue collection and processing

Mice were euthanized under deep anesthesia (200 mg/kg ketamine and 20 mg/kg xylazine) by extensive intra-cardiac perfusion with PBS for 15 min after clumping the femur. Organs were then collected and differentially processed. BM cells were collected from the clumped femur by flushing. The brain was removed and divided into two longitudinal halves. For immunohistochemistry analysis, half brain was fixed for 24 h in 4% PFA and equilibrated in sucrose gradient (from 10% to 30%) before embedding in OCT compound and stored at −80°C. A 1-mm-thick section of the other half brain was fixed in formalin for histopathology and the rest was flash frozen for biochemical analysis.

VCN and human-cell chimerism determination by ddPCR

gDNA extraction was performed with DNeasy Blood & Tissue Kit (Qiagen, USA) and 20 ng of gDNA were used for VCN and human-cell chimerism determination by duplex ddPCR (Bio-Rad, USA) as per manufacturer’s instructions, with specific primers annealing on lentiviral psi region and murine Rpp30 gene for murine-cell VCN, lentiviral psi region and human ALB for human-cell VCN, human ALB and murine Actb gene for human-cell chimerism. The specific primer/probe sequences are as follows: viral psi (FW, TGAAAGCGAAAGGGAAACCA; RW, CCGTGCGCGCTTCAG; probe, AGCTCTCTCGACGCAGGACTC); murine Rpp30 gene (FW, CCAGCTCCGTTTGTGATAGT; RW, CAAGGCAGAGATGCCCATAA; probe, CTGTGCACACATGCATTTGAGAGGT); human ALB gene (FW, GCTGTCATCTCTTGTGGGCTGT; RW, ACTCATGGGAGCTGCTGGTTC; probe, CCTGTCATGCCCACACAAATCTCTCC); murine Actb gene (FW, AGAGGGAAATCGTGCGTGAC; RW, CAATAGTGATGACCTGGCCGT; probe, CACTGCCGCATCCTCTTCCTCCC).

IDS enzymatic activity assay

Homogenates were prepared by sonication of tissue samples with homogenization buffer or repeated freeze-thaw of cells in saline. Reaction mixture consisted of 10 μL of homogenate to which was added 20μL of 1.25 mmol/L 4-methylumbelliferone (4MU)-αIdoA-2S in 0.1 mol/L sodium acetate buffer pH 5.0, containing 10 mmol/L lead acetate and 0.02% sodium azide. The reaction mixtures were incubated for 4 h at 37°C, thereafter 40 μL of concentrated McIlvain’s buffer pH 4.5 (0.4 mol/L Na-phosphate/0.2 mol/L citrate) and 10 μL (16 μg of protein) of partially purified α-iduronidase were added and a second incubation of 24 h at 37°C was carried out. Reactions were terminated by the addition of 200 μL of 0.5 mol/L Na₂CO₃/NaHCO₃, pH 10.7 and the fluorescence of 4MU was measured. The activity of IDS was determined using 4-methylumbellyferil iduronide (4MUI) as the substrate. IDS will catalyze the cleavage of the non-fluorescent 4MUI into a fluorescent 4MU. To calculate the exact IDS activity in each reaction, the standard curve made by 4MU was performed in each assay. The fluorescence of the reaction was measured using a plate reader with excitation at 355 nm and emission at 460 nm.

GAG analysis

GAG concentration was quantified form the liver and brain lysates using a sulfated GAGs Assay Kit (catalog # 6022, Chondrex, Woodinville, WA). It uses a cationic dye 1,9-dimethyl methylene blue (DMB), which binds to highly charged sulfated GAGs. The absorbance was measured at 525 nm, plotted against the standard concentration of chondroitin-6-sulfate, and was expressed as μg/mg protein.

Immunohistochemistry

Immunohistochemistry was performed on the Leica Bond III automated staining platform using the Leica Biosystems Refine Detection kit (Richmond, IL). Polyclonal anti-GFAP (catalog # ab7260, Abcam, Cambridge, MA) was run at 1:3,000 dilution using the Leica Biosystems Refine Detection kit with citrate antigen retrieval. Polyclonal antibody IBA1 (catalog # 019–19741, FujiFilm Wako Chemicals, Richmond, VA) polyclonal was run at 1:500 dilution using the Leica Biosystems Refine Detection kit with citrate antigen retrieval.

Human DPs and CIs characterization

Before intended in vitro and in vivo use, DPs and CIs produced by AGC Biologics with the two processes (protocol A and B) were characterized to confirm the following critical attributes: strength, identity, purity, and potency. A summary of applied methods is hereafter reported.

Cell count and viability

The aim of the test was to determine the viability and the total number of viable cells. The test was done by manual count and using trypan blue exclusion dye. The test was performed on two independent samplings and was compliant with EP 2.7.29.

Clonogenic capacity (number of colonies per 10⁶ cells)

The aim of the test was to evaluate the clonogenic capacity of the hematopoietic progenitors. It was performed according to Pharmacopoeia Europaea (Ph. Eur.) 2.7.28 (colony-forming assay for human hematopoietic progenitor cells).

Different concentrations of cells were seeded (four replicates per concentration) in MethoCult H4435 enriched methylcellulose medium (STEMCELL Technologies, USA) and incubated at +37°C for 13–14 days. At the end of incubation, CFU-GM, BFU-E + CFU-E, and CFU-GEMM colonies were counted.

Transduction efficiency (%) (only for transduced DPs)

The transduction efficiency was evaluated in terms of percentage of LVV-positive cells. The assay was performed on colonies obtained at the end of the clonogenic assay. The colonies from the clonogenic test plates were randomly picked (40 BFU-E and/or CFU-E and 40 CFU-GM). The colonies were resuspended in PBS buffer and washed and lysed in lysis buffer. Each colony was analyzed in qPCR using two sets of primers, one designed to span over the HIV long terminal repeat (LTR) region and GAG sequence of the provirus and the other one on the human telomerase reverse transcriptase (hTERT) gene in order to evaluate the presence or absence of integrated provirus. A colony was considered evaluable if it was composed of more than 24 cells through the detection of the hTERT gene. It was then assessed to be negative or positive for integrated LVV. A colony was positive if copies of the integrated provirus were present. The transduction efficiency was calculated as the percentage of positive colonies on the total number of colonies analyzed.

VCN (LV copies/cell) (only for transduced DPs)

The assay quantifies the mean provirus copy number per cell (mean of LV copy number/cell) in the DP after an extended LC. A 14-day LC is essential to remove the effect of residual episomal forms of the LV transfer plasmid on total copy number, and therefore it only allows the detection of integrated copies. The test is based on qPCR system. The mean number of integrated proviruses (qPCR HIV) is normalized to the mean number of cells analyzed (qPCR h-tert gene).

Immunophenotype characterization

This method uses flow cytometry analysis to determine the phenotype of the cellular population and ultimately determine the percentage of CD34⁺ cells. The identification of each sample’s phenotype was performed utilizing the principles of European guidelines relevant for advanced therapy medicinal products EP 2.7.23 and EP 2.7.24.

Briefly, samples were stained with fluorochrome conjugated monoclonal antibodies (CD45, CD34, CD15, CD19, CD16, CD56, CD3, and 7AAD as mortality marker). The percentage of each population analyzed was calculated on total viable CD45⁺ cells, with the exception of CD34⁺/CD15⁻/CD19⁻ cells, which were evaluated on total CD34⁺ cells. In addition, a separate analysis was also performed as part of further stem cell characterization to monitor the expression of the markers CD90 and CD133.

LV integration retrieval and analysis

LV integration site retrieval and analysis was performed as previously published.³⁰ Ligation-mediated PCR libraries were barcoded and sequenced with an Illumina MiSeq machine at IGA Technology Services (Udine, Italy). Raw reads resulting from Illumina paired-end sequencing were bioinformatically trimmed to recover the human genome sequences adjacent to the proviral LTR as previously described,³⁰ and they were mapped on the reference genome (human NCBI Hg38) by the BWA-MEM software (>95% identity). The genomic coordinates of the first nucleotide in the host genome adjacent to the viral LTR were indicated as ISs. ISs originated by different reads mapping in the same genomic position were collapsed recovering the number of corresponding reads (read count). Unique ISs were annotated on the RefSeq gene database as intergenic or intragenic (exonic or intronic). Genes (defined by the most upstream transcription start site to the most downstream end) hosting at least one IS were identified as target genes. Significant over-targeted genes were defined as previously described.³⁰

IVIM assays

The IVIM assay was performed as previously published.⁴⁷ Briefly, lineage-negative cells were isolated from 10- to 12-week-old C57Bl/6J mice (Janvier Labs, France), transduced with two hits of vector transduction, and expanded up to 15 days. To compare test vectors to viral gene transfer vectors with a documented potential to induce in vitro immortalization by insertional mutagenesis, a positive control γ-retroviral vector pRSF91.mCherrygPRE (RSF91) and a SIN-LV RRL.PPT.SF.eGFP.pre containing the mutagenic spleen focus-forming virus promoter/enhancer (SF) in the internal position (LV-SF) were included in the assay.⁴⁸ At day 4, analysis of GFP-control vectors by flow cytometry followed by isolation of DNA from one-tenth of the cell material was performed. The mean VCN was determined by ddPCR with a TaqMan approach on a QX200 system. Samples were measured in triplicate using 100 ng of genomic DNA. The number of viral sequences was normalized to a genomic reference sequence. The WPRE element detected viral sequences, whereas primers targeting the Ptbp2 gene were used for normalization of genomic DNA.⁴⁹ After the limiting dilution step performed on day 15, cells were cultured for an additional 2 weeks. Afterward, each well of a 96-well plate was screened microscopically and scored for clonal outgrowth, determining the number of positive wells. The frequency of mutants with a replating phenotype was calculated according to Poisson statistics using the L-Calc software (Stem Cell Technologies, Vancouver, BC, Canada). For statistical comparison, the incidence of positive (replating frequency [RF] ≥3.17 ×10⁻⁴) and negative plates (RF < 3.17 ×10⁻⁴) were compared by a Fisher’s exact test with Benjamini-Hochberg correction. The incidence of the samples in the current assay was compared to meta-data of mock transduced (MOCK) and RSF91. The mean RF of the current RSF91 and the hIDS-LV were compared to the meta-data of RSF91 using a Mann-Whitney test. A p value <0.05 was considered statistically significant.

SAGA

The molecular SAGA was performed as previously described.⁵⁰ The cell culture component until day 15 was the same as that of the IVIM assay. While the IVIM assay uses a microscopic and enzymatic readout, SAGA is based on gene expression analysis. Briefly, instead of the limiting the dilution step of the IVIM assay, total mRNA was isolated from the bulk cultures for total RNA isolation using the Direct-Zol RNA MiniPrep Kit (Zymo Research) with on-column DNAse treatment. Microarrays were performed by the Research Core Unit Genomics (RCUG) of Hannover Medical School. Slides were scanned using the Agilent Micro Array Scanner G2565CA (pixel resolution 3 μm, bit depth 20). Data extraction was performed with the Feature Extraction Software V10.7.3.1. Extracted features were analyzed using R 3.3.2 and Bioconductor 3.4 as described.⁵¹ Raw data were log2 transformed, quantile normalized, and within-array replicates condensed using the R package “limma.”⁵² Probe annotations supplied by Agilent for the Whole Mouse Genome Oligo Microarray 4 × 44K v2 (Design ID 026655) were used. Batch correction between different SAGA assays was performed with the ComBat algorithm of R package “sva.”⁵³ Results of gene expression analysis were visualized as 2D plots. The “prcomp” function from the R package “stats” was used to perform principal-component analysis. Differentially expressed genes between the respective MOCK control of one assay and the control or test vector transduced samples were calculated with the moderated t test procedure of the “limma” package with Benjamini-Hochberg multiple testing correction. Differential gene expression sets were ranked in descending order resulting in a pre-ranked list of gene symbols consisting of 15,801 genes. The gene lists were subsequently analyzed with the Broad gene set enrichment analysis (GSEA) software using “GSEA-preranked” with the (1000) permutation type set to Gene_set.⁵⁴

Statistical analysis

All statistical tests were two sided. Normality of the samples was first verified by applying the Shapiro-Wilk test. Then, the Student’s t test or Mann-Whitney non-parametric test was used for two-group comparisons. For comparisons with more than two groups, one Way ANOVA with Dunnett’s multiple comparisons test was used, comparing every column with each of the others, if not otherwise indicated in the figure legend. In the case of longitudinal behavioral data, repeated-measures ANOVA was applied. Survival and endpoints analysis were performed by log rank test. Differences were considered statistically significant at a value of ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Data and code availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.