Cell- and gene-based therapeutic approaches for neurological deficits in Mucopolysaccharidoses

Abstract

Mucopolysaccharidoses (MPS) are a group of lysosomal storage diseases that are resulted from abnormal accumulation of glycosaminoglycans. Among the progressive multi-organ abnormalities often associated with MPS diseases, the deterioration of central nervous system (CNS) is the most challenging manifestations to be tackled, due to the impermeability of the blood-brain-barrier (BBB). Evolved with recent development in stem cell biotechnology and gene therapy, several novel experimental approaches have been investigated in animal models. In this review, we will address different approaches attempting to bypass the BBB for neuropathic MPS treatment using cell- and gene-based therapies. Several neurological findings in CNS pathophysiology emerged with therapeutic investigation will also be discussed.

INTRODUCTION

Lysosomal storage diseases (LSDs) are a group of inherited metabolic disorders with defects in lysosomal function. It includes approximately 46 different genetic disorders that as a group have an incidence of 1 in 5000 - 7000 live births with 65% affecting the central nervous system (CNS)[1]. The Mucopolysaccharidoses (MPS) represent a subgroup of LSDs that consist of nearly 30% of all LSD patients, and are categorized by the abnormal degradation of mucopolysaccharides or more accurately glycosaminoglycans (GAGs). Eleven distinct enzyme deficiencies have been discovered that are associated with seven clinical MPS types (type I to IV, VI to VII and IX) (Figure1A). The clinical features in patients with MPS disorders are often associated with progressive multi-organ abnormalities, including skeletal (dysostosis multiplex), visceral and/or CNS manifestations. Among them, MPS I, II, III (all four subtypes) and VII have significant CNS involvements. The life spans for severe forms of MPS patients are largely shortened with premature death mostly occurred in childhood or early adulthood. The history and clinical manifestations of these disorders are well reviewed by others [1, 2].

Fig. (1).

A) Phenotypic manifestations in MPS types. The involvement of CNS and somatic symptoms are categorized based on the severe form of each MPS type. 2, significant; 1, mild/moderate; 0, absent. B) Classification of MPS types in correspondence to the types of GAGs which catabolism are abnormally blocked. The MPS types are listed by their unique genetic basis, and grouped by either single (right column) or multiple (left column) GAG accumulation. Oval, non-neuropathic MPS; rectangle, MPS types with CNS involvement.

Early studies of Hurler syndrome (severe form of MPS I) have provided key insights into metabolic correction and cross-correction from intercellular lysosome enzyme transfer in LSD [2]. Normal lysosomal enzyme trafficking to lysosome compartments is mediated by an intracellular mannose-6-phosphate (M6P) receptor system. However, a small proportion of lysosomal enzyme, which varies depending on the individual enzyme, cell type and gene expression status, can “leak” out to the extracellular environment and be available for reuptake and reuse by other cells through cell surface M6P receptor mediated endocytosis and/or direct cell-to-cell contact [3, 4]. This phenomenon of lysosomal enzyme release and reuptake among cells has provided the foundation for the development of current two main treatment options in MPS patients, namely allogeneic hematopoietic stem cell (HSC) transplantation and enzyme replacement therapy (ERT) [5]. Transplantation using healthy donors provides therapeutic benefit, including prolonging life and improving some of the visceral manifestations. Furthermore, transplantation early in life (<2 yrs) leads to significant improvement in CNS outcomes in MPS I, even through minimal or no response has been obtained for MPS II or for the reversal of pre-existing CNS abnormalities [6-9]. ERT is now available for several MPS disorders (MPS I, II and VI) and has been shown to ameliorate visceral manifestations in some patients. However, it is limited by likely poor penetration to the CNS and the need of frequent intravenous infusion for a lifetime. A therapeutic approach with the capacity to correct CNS deterioration is needed for MPS disorders with major CNS complications.

The extra challenges for neuropathic MPS treatment are presented by the fact that the progression of CNS manifestations may result in “irreversible” deficits; and that the blood-brain-barrier (BBB) is generally considered impermeable to lysosomal enzymes. To overcome these hurdles, researchers around the globe have investigated several therapeutic strategies with promising development, such as intrathecal ERT, direct in vivo gene therapy (intracranial or systemic delivery), small molecule therapies (substrate reduction and chaperone therapy) and cell-based therapies. While recent development in ERT will be discussed in other reviews of this special edition, this review will focus on recent progress in cell-based approaches and in vivo gene delivery, as well as neurological findings in CNS pathophysiology of MPS disorders. Regardless of variations in therapeutic strategies, the main questions remain the same, i.e., if and how therapeutic enzyme can reach the brain, and if the quantity and distribution within brain parenchyma are sufficient to make a significant impact on the clinical progression of the diseases.

CNS PATHOGENESIS

Heparan Sulfate and Neuropathic MPS

The deficiencies of lysosomal hydrolases in MPS disorders lead to a metabolic blockage in the stepwise degradation of one or more of the five GAGs, including heparan sulfate, dermatan sulfate, chondroitin sulfate, keratan sulfate and hyaluronan (Figure1B). They are unbranched polysaccharides consisting of a repeating disaccharide unit, and often form complexes with proteins as proteoglycans. These structural carbohydrates are a major component of extracellular matrix (ECS) in nearly all cell types, especially connective tissues. They may also play a role in ligand-receptor interactions and regulation of signal molecules within the matrix. Among the five GAGs, chondroitin sulfate proteoglycans (CSPG) is one of the main components in CNS ECM, followed by heparan sulfate proteoglycans (HSPG) that also exist in some ECM areas or cell surfaces [10]. Interestingly, patients with MPS IVA, that is caused by the abnormal catabolism of CS and keratan sulfate, exhibit normal intelligent with distinctive skeletal abnormalities. On the contrary significant CNS involvement appears to be always associated with HS accumulation in MPS disorders (Figure1B).

The HSGP plays a significant role during the formation of the CNS, and may have multiple functions relevant to cell signaling, cell motility, neuronal plasticity, neuroprotection and the extracellular hydric and ion homeostasis [11-13], as well as the control of the CNS injury response [14]. Syndecans and glypicans are the main cell surface HSPG, and have been detected in developing and adult CNS neurons and glia [15, 16]. Whereas agrin and perlecan are found in the ECS, with agrin mainly localized in the synapsis, and perlecan in the perivascular astrocyte feet [17]. There have been increasing evidence that cell surface HSPG syndecan and glypican play a role in promoting neuronal migration, axon guidance, and synapse formation during CNS development [18, 19]. In addition, CNS pathogenesis may also affect systemic symptom representations. For example, the role of CNS syndecans has been recognized recently in the regulation of body weight by mediating the activity of adiposity signals during neuronal development and synaptic organization [20]. This may, in part, contribute to the adipose storage deficiency observed in murine models of MPS I, MPS IIIB and MPS VII [21]. Nonetheless the relationships between HS recycling and HSGP synthesis and functions in CNS are largely unknown, even though the knowledge of the role of the endosome/lysosome (E/L) system in overall cell function has been expanded significantly in recent years [22]. It is clear that the complexity in brain pathogenic cascades of LSD involves not only the disruption of lysosomal function in neuronal cells, but also the interference with signal transduction, cellular homeostatic control and salvage processing normally controlled by the E/L system [23].

Other pathogenic signatures

Even through the enzymes defective in MPS diseases are known to affect GAG catabolism but not glycosphingolipids (GSL), the accumulation of gangliosides (GM2 and GM3), together with unesterified cholesterol, has been found in the brains of patients with MPS I, MPS II (severe), MPS IIIA, IIIB, and IIID, who have shown mental retardation, but not in those with normal intelligence [24-27]. Inhibition of GSL degradative enzymes, such as ganglioside sialidase, may have occurred secondary to GAG accumulation [1, 28], although the sequestered gangliosides do not typically co-localized with each other or with GAG storage [26].

Activation of microglia with chronic inflammation component has been observed in MPS I, MPS IIIB and IIID [29, 30]. The role of toll-like receptors (TLR) signaling has been highlighted as one of the major contributors to brain inflammatory responses and microglial cell activation with age and neurodegenerative diseases [31]. Recent findings by Ausseil et al. have demonstrated that microglial cell activation could be primed by heparan sulfate oligosaccharides via the TLR4 pathway in the brains of MPS IIIB mice at very early stage of the disease [32]. In addition, it has been reported that changes in ganglioside expression or availability at the plasmalemma may also alter TLR4 signaling pathway and trigger microglial activation [33]. Using TLR4-specific siRNA and a dominant-negative TLR4 gene, the functional importance of TLR4 was demonstrated in ganglioside-triggered activation of glia. Thus, the accumulation of both heparan sulfate and gangliosides in the brains of neuropathic MPS may have contributed to TLR4-mediated activation of glial cells.

Alteration of the blood-brain barrier in MPS diseases

A number of studies have demonstrated the possibility of BBB functional alterations during progressive CNS manifestation in MPS diseases [34] although these changes, if any, may be very slow with high heterogeneity depending upon the cells affected and their metabolic signatures. Recent studies using in vitro BBB models have shown that lipopolysaccharides-activated microglia could induce BBB dysfunction by releasing tumor necrosis factor-alpha (TNFα) [35] or by producing reactive oxygen species through NADPH oxidase [36]. In addition, activated microglial cells secrete interleukin 1β and TNFα that can both act on tight junctions to increase their permeability [37, 38]. As discussed above, microglia activation has been noted in several neuropathic MPS. In addition, brain pathology in MPS II patients has indicated a swelling and enlargement of the perivascular space around the brain micro-vasculatures [39]. Pro-inflammatory cytokines can also act to recruit and/or facilitate monocytes migration across the BBB, which in turn may further reinforce cytokine release and increase BBB permeability.

The likelihood of BBB remodeling has been proposed for neuropathic MPS based on several reasons. First, both heparan sulfate and dermatan sulfate are important components of the basal lamina surrounding the BBB-forming microvascular endothelial cells, pericytes and glia. Secondly, matrix metalloproteinases (MMPs) are a group of neutral proteases that play a key role in remodeling the brain vascular bed (basal lamina) [40]. Among them, MMP9, whose synthesis and release can be induced by TNFα, has been shown to remodel the basal lamina by degrading laminin and other cell adhesion molecules [41]. The significance of BBB remodeling under MPS conditions has been highlighted recently toward the development of innovative CNS treatment [42]. To identify BBB-targeted peptides that bound to brain microvascular endothelial cells, Davidson’s group screened a phage library in vivo in MPS VII and normal mice. Remarkably, the peptide motifs enriched in MPS VII brain were distinctive from those in normal mice or those in another LSD mouse model (with lipid storage defects), suggesting a unique vascular remodeling process in MPS VII mice. The specificity of epitope-binding to MPS VII brain was resulted, in part, from accumulated chondroitin sulfate under disease state. It will be interesting to test if the epitopes identified from MPS VII brain would be applicable for BBB-targeting in other types of neuropathic MPS disorders.

DIRECT INTRACRANIAL DELIVERY APPROACHES

Direct intracranial delivery approaches, i.e., intracranial injection and intracerebroventricular administration, have been explored to bypass the BBB for treatment in neuropathic MPS animal models using either normal or genetically modified cells, or viral vectors as therapeutic agents. These strategies take advantages of refined, on-target administration, and subsequent controlled distribution of cellular vesicles or viral vectors for CNS benefits. Recent advances in stem cell technology have provided the prospects for the development of potentially powerful new therapeutic approaches toward neuropathic MPS treatment. If successful, this approach may be additive as a CNS treatment option to other therapies or may serve as a primary option for patients with MPS III who have minimal somatic symptoms. Main concerns are practical considerations with volume restriction, long-term survival of transplanted or transduced cells, and the limited diffusion of cell/vector from the injection sites. Procedure-induced injury may also trigger innate CNS immune response that may be followed by adaptive immune response prompted by self- or foreign-antigen recognition.

Cell-based brain injection

Stem cells are defined as primitive cells capable of self-renewal, multi-lineage differentiation and high rate of proliferation. Two broad subgroups of mammalian stem cells are categorized: pluripotent embryonic stem cells (ES) that are isolated from the inner cell mass of blastocysts, and multipotent somatic adult stem cells that are found in adult tissues. Human ES cells can grow indefinitely in culture while maintaining normal karyotype and pluripotency [43, 44], thus providing a potentially unlimited source of cellular vesicles for CNS transplantation [45]. A variety of applications have been studied using ES-derived neural stem cells [46] for direct CNS treatment in neuropathic MPS, although undifferentiated hES cells are highly associated with the formation of teratomas upon transplantation.

Neural stem/progenitor cells can be derived from various stem cell sources, and can give rise to the three major neuroepithelial-derived brain cell types: neurons, astrocytes and oligodendrocytes. These cells have been viewed for several reasons as the favorable target or cellular vesicle for gene therapy in treating acquired or inherited neurological diseases, especially neuropathic MPS, via brain injection. First, they can be readily derived with unlimited potential from more primitive stem cells such as embryonic stem cells or induced-pluripotent stem cells (iPS), and be capable of genetic modification to over-express the defective enzyme. Secondly, they can serve as a depot for continuous production and release of therapeutic enzyme that can be uptake by neighboring cells via MPR-mediated endocytosis with long-term benefit potential. Finally, neural stem cells can serve as cell replacement source to generate neuronal or glial cells lost in disease-affected brains.

The potential application of genetically engineered NSCs for treating neuropathic MPS was first demonstrated using a murine MPS VII model [47]. By direct injection of retrovirally transduced murine NSC into the cerebral ventricles of newborn mice, donor cells engraftment was observed throughout the neuraxis, resulting in widespread of therapeutic enzyme overexpressed by engrafted NSC and subsequent clearance of lysosomal storage in neurons and glia in affected brain. These observations were further expanded in treating symptomatic brain of adult MPS VII mice [48]. Lysosomal distention was cleared from neurons and glial cells in the vicinity of the grafts, showing that the secreted enzyme could reach the diseased cells and correct some of the established lesions in the severely diseased brain. With the development of human stem cell technology, similar therapeutic effects were also observed 25 days after transplanting genetically engineered human fetal neural stem cells into neonatal MPS VII mice, although significant loss of engrafted cells was detected due to apoptotic cell death [49]. Recently, the potential therapeutic application of murine ES-derived, genetically modified glial precursor cells has been evaluated as a cellular vehicle for sulfamidase delivery in the CNS of MPS-IIIA mice [50]. Relatively long-term engraftment (at least 12 weeks) was observed following neonatal ventricular brain injection with ~0.01% exogenous cells detected in the injection hemisphere of MPS IIIA brain 8 weeks post transplantation. More recently, neurological functional improvement has been evidenced in Sandhoff mice (a neuropathic LSD with abnormal lipid storage) when injecting NSC into neonatal or adult symptomatic brain [51, 52]. These reports indicate that neuronal stem/progenitor cells can serve as a valuable cellular vesicle for gene transfer, and an efficient depot with continuous enzyme distribution for the treatment of CNS pathology in MPS disorders.

Several adult stem cells, other than CNS-isolated NSC, have been shown to be capable of neural differentiation, and thus been proposed as alternative sources of cellular vesicles for CNS transplantation. They are mesenchymal stem cells derived from bone marrow or cord blood [53-58], adipose tissue stem cells [59], amniotic fluid [60]. These cells are attractive sources for generating recipient-derived autologous cell transplants, and currently under active investigation. However, adult stem cells, in general, are prone to senescence in culture with limited passages, thus have extra challenges in treatment development.

As a new class of pluripotent, potentially inexhaustible stem cells, iPS cells were recently generated from adult somatic cells by introduction of embryogenesis-related genes [61]. Four transcription factor genes were first identified as sufficient to switch murine embryonic and adult fibroblasts to iPS with properties similar to murine ES cells [62, 63]. These findings were further validated, and expanded from mouse to human with successful reprogramming of end-differentiated adult fibroblasts into ES-like pluripotent stem cells [64, 65]. Since then, there have been rapid progress with numerous reports to improve reprogramming efficiency, make iPS safer by reducing tumorigenecity and by eliminating permanent integration in iPS cells, as well as to explore the choices of the initial cell population. The number of factor genes required for successful iPS generation has reduced from 4 factors to 3 [66], and to 2 factors (in the presence of valproic acid) [67, 68]. More recently, the generation of one-factor human iPS cells from human fetal neural stem cells was reported by epigenetic expression of OCT4 [69]. The ability to obtain stem cells specific to an individual offers new prospects for regenerative medicine in general, and also open a door for autologous stem cell-based gene therapy in neuropathic MPS in particular. In fact, generation of murine iPS cells tailored to MPS VII mouse has been reported using primary MPS VII fibroblasts isolated from tail-clips [70]. While this new class of stem cells may provide opportunities to develop patient-specific pluripotent cell lines for cell-based CNS therapies, substantial research and developments are still required for their ultimate usage in clinics.

Toward future clinical application of cell-based direct brain injection approaches, several issues still need to be addressed, such as 1) identifying and/or optimizing ideal cell source(s) for cellular grafts that tailor to MPS conditions; 2) evaluating long-term safety and efficacy in large-animal models; and 3) maximizing the survival and engraftment of transplanted cells. In addition, it will be important to understand the dynamics of migration, proliferation and apoptosis, and the nature of the inflammatory response following cell transplantation at different brain regions. For example, a study evaluating intra-cerebral transplantation of mesenchymal stem cells into neonatal vs. adult mouse brain has indicated that cell proliferation in brain are regulated by the host microenvironment, while the unique molecular signatures of various MSC subpopulations convey an inherent capacity to engraft and migrate [71]. Such studies could provide valuable guideline to the optimization of differentiation and injection protocols, as well as the selection of cell sources for maximum migratory potential and subsequent widespread engraftment and enzyme distribution.

Viral vector-based strategies

A variety of non-neurotropic viral vectors have been investigated for in vivo CNS gene transfer in brain of MPS I, MPS IIIA, MPS IIIB and MPS VII mouse models using adeno-associated virus (AAV), adenovirus, and lentivirus (LV) [72, 73]. Therapeutic effects (storage reduction) have been demonstrated in all animal models injected at birth, and in some animal models injected as young adult using canine adenoviral vectors [74], lentiviral vectors [75-77], or adeno-associated viral vectors [78-82]. In general, neonatal injection resulted in more effective and long-lasting benefits, in comparison to treatment applied to adult brain (which is more relevant to clinically realistic situation). Moreover, these vectors show restricted spread from the injection site (mostly around the needle track), and thus result in relatively limited CNS benefit (especially when injected into adult brain).

Several strategies have been investigated to improve vector distribution and transduction efficiency. For example, targeting widely dispersed systems in the CNS as injection sites for potential gene dispersal has been explored by injecting AAV-9 into the ventral tegmental area, a region with numerous efferent and afferent projections, of MPS VII brain [83]. Widespread distribution of the vector genome was observed after a single injection, resulting in even further distribution of the enzyme product and reduction of the storage lesions throughout the entire brain. Alternative routes of vector administration have also been studied to improve vector distribution, including intravitreal injection [84] and combined intracisternal injection [85]. After injecting therapeutic rAAV directly into the vitreous humor of young adult MPS VII mice, elevated enzyme activity and reduced lysosomal distension were found in regions of the thalamus and tectum that received inputs from the injected eye, as well as adjacent nonvisual regions (the hippocampus and visual cortex) [84]. Another mode of administration was also studies for AAV-mediated gene therapy in MPS IIIB mice [85]. Following intravenous injection of mannitol, therapeutic AAV was injected intravenously and intracisternally into young adult MPS IIIB mice, resulting in significantly prolonged lifespan, improved behavioral performances, and variable correction of lysosomal storage pathology in the CNS.

Herpes simplex virus (HSV) has the ability to establish life-long latent infections in postmitotic neurons and to remain transcriptionally active, continuously expressing latency-associated transcripts while producing minimal disease. These properties have made HSV an attractive candidate vector for neuronal gene transfer with widespread in vivo transduction potential following a single injection. HSV-mediated CNS gene transfer was first demonstrated in MPS VII mice by corneal inoculation, resulting in expression of β-glucuronidase in a limited number of central nervous system cells up to four months post-inoculation [86]. Direct brain injection of neuroattenuated therapeutic HSV-1 vector at multiple sites of murine MPS VII brain has led to widespread correction of lysosomal storage throughout brain [87]. Axonal transport of vector and enzyme has been implicated, which was attributable to the correction of storage lesions in a large volume.

Immunology of intracranial delivery

The human CNS has been traditionally considered “immunologically privileged” because graft survival is prolonged significantly comparing to non-CNS sites. Several factors may have contributed to the prevention of efficient adaptive immune response activation, including BBB protection, antigen-presenting cell paucity and limited lymphatic drainage [88]. However, post-morterm examinations of fetal NSC graft recipients have revealed activated microglia and host immune cell infiltration at the host-graft interface and throughout the graft, suggesting that inflammation may have compromised graft survival and functional outcome [89, 90]. Moreover, neuroinflammation has been considered a disease component in MPS I and MPS III (as discussed previously). The survival and the ultimate fate of transplanted human stem cells and their differentiated progeny will more likely be compromised by donor/host human leukocyte antigens (HLA) mismatch in MPS patients. The potential application of iPS cells derived from patients’ somatic cells for ex vivo gene transfer will open a door for new strategies toward autologous brain transplantation in neuropathic MPS disorders.

The surgical procedure during intracranial injection may activate local innate CNS immune response by inducing immune signaling following injury [91]. The activated cells produce an array of pro-inflammatory cytokines and chemokines that, in turn, recruit circulating leukocytes to the injury site. Invading macrophages and leukocytes then mediate the adaptive response in which foreign- or self-antigen recognition triggers T and B cell activation and amplification that may eventually induce cellular or humoral cytolysis [92]. This may be, in part, attributable to the loss of transplanted or in situ transduced cells after direct cell or vector injection.

As an alternative approach to overcome the difficulties relating to immune response, cell microencapsulation has been explored to enclose cell lines, such as normal or genetically modified NSCs, within a generally polymeric, semi-permeable membrane [93, 94]. The underlining potential advantages include 1) that transplanted cells do not have physical contact with host cells, thus protecting cells from immune mediators; 2) and that the risk of uncontrolled proliferation of transplanted cells may be put under check. Significant β-glucuronidase production and release has been observed in MPS VII brain after transplantation of encapsulated enzyme-secreting fibroblasts [95], or immortalized human amniotic epithelial cells [96] into adult mice. Direct in vivo producer cell-mediated gene transfer was also investigated by transplanting encapsulated retroviral packaging cells into murine MPS VII brain for continuous vector production and gene transfer for up to 6 weeks [97]. Microencapsulation may have a potentially significant future for CNS disease treatment, if further improvement can be achieved to tackle issues on performance, biosafety, biocompatibility, retrievability, purity and characterization.

PERIPHERAL DELIVERY OF GENE-MODIFIED HSC

The surface area of the human brain microvasculature available for protein/vector transport is ~20 m², which has the potential to reach the entire brain volume. In fact, the microvasculature is so dense that all neuron or glial cells are within 20 um from the nearest capillary [98]. Therefore, if a cellular vesicle or vector is capable to cross the BBB, systemic delivery via circulation can provide the ideal noninvasive mean for rapid and wide distribution of therapeutics throughout the brain. Peripheral administration of hematopoietic stem cells (with or without genetic modification) or viral vectors for neuropathic MPS will be discussed.

Allogeneic hematopoietic stem cell transplantation

Allogeneic transplantation of hematopoietic stem cells (HSCT) (including bone marrow, HSC-enriched CD34⁺ cells, and, more recently, umbilical cord blood as sources) has been one of the main treatment options currently available for MPS disorders that have significant CNS involvement. More than two hundreds of patients with MPS disorders (mostly MPS I) have been treated with allogeneic HSCT since the first bone marrow transplantation was performed about 30 years ago [99, 100]. In the last 5 years, many MPS patients have been treated with umbilical cord blood from unrelated donor, allowing rapid and increased access to transplantation with favorable outcomes [7, 100]. Clinical experiences and outcome have been well documented for MPS I, II, III and VII [101-103]. Expression of enzyme from engrafted normal donor blood cells results in 1-10% of normal serum enzyme levels in MPS patients with successful BMT [99]. These levels have been associated with clinical benefits in MPS I and MPS II patients, including prolonging life and ameliorating some of the clinical manifestations.

The neuropsychological outcomes have varied widely after HSCT. For example, when treated early in the course of disease (under 2-year old), children with MPS I experience preservation of normal intellectual function [104]. However, minimal or no response has been observed in reversing pre-existing CNS abnormalities in patients with severe MPS II or MPS III [6-9, 105]. Factors that may affect the CNS outcome include the type and severity of the MPS disorder, the genotype of the donor (normal donor is better than carrier), the degree of clinical involvement, and the age at the time of transplantation.

The delivery of enzyme across the BBB in HSCT treatment may be dependent upon diapedesis of HSC-derived macrophage-monocytes into the brain, a presumed mechanism of CNS improvement after HSCT [106]. That is donor-derived mononuclear cells are continually crossing the BBB and remain in the CNS for prolonged periods of time. Some of them are microglial precursors and gradually become repopulated as CNS resident microglia. These donor-derived cells can serve as a cell source for replacement of enzyme-deficient, defective microglia, and also as an enzyme source for continuous release of functional enzyme that can be uptake by neighboring brain cells, thus contributing to CNS benefits. However, the microglia turnover process is very slow and the progression of CNS manifestation can result in “irreversible” damage to the brain. In addition, several factors that are specific to individual disease type may also affect the CNS outcome, including the degree of microglia involvement, the levels (proportion) of enzyme release by donor-derived cells, and the amount of enzyme required for cross-correction in main pathogenic cell types. Thus, the overall outcome of HSCT might be sufficient to meet the catalytic demand in patients of one MPS condition but not in another.

Although allogeneic HSC transplantation has significantly modified the natural history of the diseases and improved survival in some MPS patients, this treatment is not considered curative for neuropathic MPS. Despite significant progress in medical procedures and the availability of banked umbilical cord blood, allogeneic HSCT is greatly limited by the risks inherent to transplantation with significant mortality, and by complications related to graft-versus-host disease as long-term sequelae of chemotherapy may place the patients at the risk of secondary malignancy.

Ex vivo HSC gene transfer with autologous transplantation

Ex vivo HSC gene transfer followed by autologous transplantation is an attractive alternative for treating MPS disorders that could provide life-long therapeutic effects without the morbidity and mortality of allogeneic transplantation. Moreover, over-expression of the missing protein in HSC-derived, transduced microglial cells may provide additional benefit for CNS treatment. The growth of gene therapy for neuropathic MPS has been through ups and downs with major advances and issues of the gene therapy field as a whole, and still in a relatively early stage of development. Recent developments on the application of lentiviral vectors and related safety improvement have reinforced the promise that HSC gene therapy has hold in treating neuropathic MPS. A phase I clinical trial of LV-mediated ex vivo HSC gene therapy for children with MPS VII has been approved in 2006 and are open for patient enrollment (US-758, http://www.wiley.co.uk/genetherapy/clinical/).

Based on preclinical studies of retrovirus-mediated gene transfer for MPS II [107, 108], the first gene therapy clinical trial for MPS conditions was conducted with autologous lymphocytes for patients with MPS II (Hunter Syndrome) using a conventional murine leukemia virus-based retroviral vector [109] (NCT00004454). Although proof-of-concept safety concerns were addressed in this early gene therapy trial, very limited benefits were observed, which was mainly resulted from insufficient gene transfer efficiency, lose of transduced cells and the short-life of lymphocytes (our unpublished data).

With advances in human HSC technology and vectorology during late 90th and the start of 21^st century, a peak of retroviral vector-mediated ex vivo HSC gene therapy clinical trials emerged for treatment of a variety of inherited and acquired diseases. The feasibility of transducing human primitive hematopoietic progenitors from MPS I patients was demonstrated using a hollow-fiber bioreactor system [110]. However, the frequencies of transduced and successfully engrafted HSC have been low in general in most gene therapy clinical trials (<1%). Exceptions have emerged recently when long-term clinical efficacy was achieved using RV-mediated ex vivo HSC gene transfer in several clinical trials for children with inherited immunodeficiencies [111-113] and chronic granulomatous disease (CGD) [114]. The genetically corrected cells in these patients likely had a selective advantage over the uncorrected cells, which had compensated for the relatively low frequency of transduced and successfully engrafted HSC normally observed in all other gene therapy clinical trials. Successful functional correction of T and B lymphocyte deficiency was evidenced in 14 of 15 treated children four years after infusion of retrovirally transduced autologous HSCs. However, this selective advantage is not available in most of other diseases. Moreover, inadvertent activation of cellular proto-oncogenes by provirus insertion resulted in secondary leukemogenesis in these otherwise successful clinical trials [115-117].

As a strategy to increase the frequency of transduced and already engrafted HSC, which is vital for the success of HSC gene therapy, the potential application of “adapted” in vivo selection was investigated using methylguanine-DNA-methyltransferase (MGMT). MGMT is an alkyltransferase that functions to repair cellular DNA damage at the O⁶ position of guanine. In vitro studies have demonstrated the utility of a MGMT variant (MGMT^P140K), which is resistant to inhibition by 6-benzylguanine (BG), for the enrichment of transduced HSCs or their progeny [118, 119]. This was then further advanced and confirmed by successful in vivo selection in mice [119-121], large animal models [122, 123] and non-human primates [124] following time- and dose-intensive treatment using different O (6)-alkylating drugs such as 1,3-bis (2-chloroethyl)-1-nitrosourea (BCNU) or temozolomide (TMZ). Both BCNU and TMZ can cross blood brain barrier, and may have variable degrees of effects in non-hematologic organs such as liver, heart, lung and CNS. The feasibility of co-selection of hepatocytes with a dosage effective for in vivo HSC selection has been demonstrated using enzyme-deficient, primary hepatocytes isolated from MPS I mice [125]. The escalation of intracellular and extracellular therapeutic enzyme levels in transduced enzyme-deficient hepatocytes was found to directly correlate with increasing selection doses, resulting in phenotypic correction in transduced and neighboring non-transduced hepatocytes. The preclinical application of this strategy for HSC gene therapy in MPS diseases is yet to be determined.

Numerous studies have been conducted to develop a vector system that is more suitable for efficient and stable HSC gene transfer [126]. HIV-based lentiviral vectors have become the most attractive system because of their ability to transduce quiescent cells and mediate stable integration, resulting in sustained expression. Since the early in vivo studies initiated in 1996, a series of step-width improvements have emerged to make LV both safer and more efficacious. In the 1^st-generation HIV-based LV, the expression of viral trans-acting products was separated from the cis acting components and placed into a packaging plasmid with the expression cassette using the cytomegaloviral (CMV) immediate-early enhancer/promoter and a cellular polyA addition signal [127]. The 2^nd generation vectors focused on identifying the minimal trans-acting requirements. Additional accessory genes (including vif, vpr, vpu, and nef) were deleted from the packaging plasmid and yet the transfer vector was still capable of transducing non-dividing cells and prolonged gene expression [128, 129]. The 3^rd generation LV introduced two additional safety features. First, the removal of the transactivating transcriptional element (tat), which acts as a crucial transactivator for HIV replication, abolished a critical feature of lentivirus infectivity. Second, Rev, a HIV regulatory protein responsible for nuclear export and inhibition of viral mRNA splicing, was removed from the packaging plasmid and placed on a separate plasmid [130]. Thus, the possibility of generating replication competent lentiviruses (RCL) by DNA recombination was further reduced with the 4-plasmid packaging system. Moreover, self-inactivating LV (SIN-LV) was developed on the transfer vector backbone by introducing deletions of the TATA box, SP-1 and NF-κB sites into the U3 promoter region of the 3′LTR [46, 131]. Upon reverse transcription and integration of SIN-LV into target cells, the deletion in 3′LTR would be transferred to the 5′LTR, resulting in inactivation of both LTR. SIN-LV also has more flexibility for targeted gene expression, as an internal promoter is the sole transcriptional control for transgene. In fact, these vectors are virtually unable to mobilize with replication-competent HIV in the “worse case” scenario [132].

In addition to the capability of transducing quiescent cells (such as HSC) and introducing stable integration for sustained expression, several other features make LV more suitable vehicles for clinical application of HSC-mediated gene therapy than other viral vectors. For example, substantial silencing of transgene expression was observed in embryonic stem cells and pre-implantation embryos transduced by retroviruses, apparently by methylamine-dependent and –independent epigenetic mechanisms [133]. However, in similar settings, lentivirus vectors were able to demonstrate successful transgene expression in ES and different tissues of chimeric animals that were generated from transduced ES using either ubiquitously active promoter [134] or tissue-specific promoters [135]. In addition, integration site mapping studies have revealed that MLV preferentially integrates in and around gene promoters (±1 kb from CpG islands); whereas, HIV-1 integrates mostly within transcriptional units [136, 137]. In vivo tumorigenesis studies of LV- and RV-transduced human CD34⁺ in tumor-prone mouse model revealed that the risk factors related to insertional mutagenesis for HIV-based LV could be less than that of MLV-based RV [138], although additional studies are required to fully determine the relevance of these differences. Moreover, the feature of “self-inactivating” LTR in LV has been found to have significant safety advantages over the intact 3′-LTR used in most RV, which can function as a strong promoter to activate adjacent coding sequences in both directions [139, 140].

With the development of the “advanced” 3^rd-generation SIN-LV featuring significant improvements of safety and performance, the application of HIV-based LV for autologous HSC-mediated gene therapy has become reality in recent years with a total of 24 clinical trials ongoing worldwide (http://www.wiley.co.uk/genetherapy/clinical). Among them, a phase I LV-mediated ex vivo HSC gene therapy clinical trial has been approved for children with MPS VII (US-758). The advantage of using HIV-1 based lentiviral vector for CNS treatment has recently been demonstrated in autologous HSC gene therapy clinical trial for children with adrenoleukodsytrophy (ALD), a fatal monogenic demyelinating disease of the CNS [141]. Remarkably, approximately 15% HSC gene transfer efficiency has been achieved, superior to any RV-mediated human clinical trials without endogenous in vivo selective advantage for gene-modified cells. Long-lasting over-expression of the missing gene in transduced, patients’ own hematopoietic stem cells and their progeny was sufficient to ensure neurological benefits in ALD patients, a clinical outcome comparable to those obtained by successful allogeneic HCT with more than 80% donor-derived engraftment [142]. This clinical trial has clearly demonstrated the feasibility and potential efficacy of HSC-based LV-mediated gene therapy in treating neurological disorders, including the neuropathic MPS disorders.

There have been many difficulties with ex vivo HSC manipulation for maintaining stem cell properties, including loss of reconstitution potential during in vitro HSC culture and transduction, and cytokine stimulation that may activate unwanted signaling pathways that could potentially increase the risk of non-random mutagenic events during provirus insertion [140, 143]. In situ bone marrow stem cell gene transfer by intrafemoral injection of viral vectors may provide an attractive, alternative strategy for in vivo stem cells gene transfer and avoid some of the difficulties of ex vivo HSC gene transfer. This approach may also take the advantages of both the natural HSC cycling for efficient transduction, and the supportive microenvironment in bone cavity for maintaining stem cell viability and capacities. The potential for in situ RV-mediated HSC gene transfer as a clinical relevant approach has been demonstrated by intrafemoral injection of RV in Jak3 knock-out SCID mice that had been pretreated with sublethal dose of 5-fluorouracil [144]. We further demonstrated that efficient HSC and MSC could be genetically modified successfully in their natural “niche” by LV-mediated in vivo gene transfer in mice without any preconditioning [145, 146]. This approach may potentially provide an innovative mode of gene delivery into stem cells residing in bone cavity for disease treatment.

Proviral integration into hematopoietic stem cells may provide life-long therapeutic benefit, but may also trigger oncogenesis as a consequence of insertional mutagenesis [140, 143, 147]. Restricting transgene expression to late erythroblasts and reticulocytes may reduce the risk of activating oncogenes in HSC and its offspring in all lineages, and take the advantage of the formidable protein synthesis machinery in maturing erythroid cells may compensate for the generally low HSC gene transfer frequency. Recent studies by Chang et al demonstrated the feasibility of long-term secretion of therapeutic levels of human factor IX (hFIX) in plasma of hemophilia B mice from HSC-derived erythroid cells [148]. This erythroid cell-derived systemic protein delivery system could also be enhanced by MGMT-mediated in vivo selection from minimal initial HSC gene transfer, resulting in stable, curative hFIX levels in mice [149]. Moreover, we have demonstrated that maturing erythroid cells could be reprogrammed, when transduced with a tissue-specific LV, to produce and release a functional lysosomal enzyme successfully and continuously at supra-physiological levels in the circulation for a total of 9-months in MPS I mice [150]. Significant improvement of neurological function and brain pathology were also observed in MPS I mice by the erythroid-derived, higher-than-normal peripheral therapeutic protein. This approach would provides a paradigm for the utilization of red blood cell precursors as a depot for efficient, sustained, seemingly safer, systemic delivery of lysosomal proteins by ex vivo HSC gene therapy for MPS disorders involving visceral and CNS abnormalities.

PERIPHERAL DELIVERY OF VIRAL VECTORS BY INTRAVENOUS INJECTION

It would be ideal if a multi-organ disease can be corrected by a single injection of therapeutics. Systemic delivery of vectors may hold the promise of just that, with the potential of low morbidity, less invasive application, and long-term transgene expression. Preclinical evaluation of intravenous injection of therapeutic vectors has been explored in several animal models of neuropathic MPS disorders. A variety of vector systems have been investigated, including adenovirus, adeno-associated virus, retrovirus, lentivirus or plasmids, and resulted in transgene expression primarily in the liver and secretion of the relevant enzyme into the circulation. A comprehensive account of this approach for treatment in mucopolysaccharidoses has been discussed elsewhere recently [151], and is beyond the scope of this review. It is worth noticed, in general, that neonatal treatment often leads to correction and prevention of GAG accumulation in multiple organs of MPS animal models, with significant CNS improvement (or even normalization) in some studies. However, postnatal treatment would likely result in partial visceral correction and less impact on the brain manifestations. In addition to age factor (developmental stage), the unique BBB remodeling in neuropathic MPS brains may also play a key role on the development of novel CNS treatment approach [42, 152]. A recent in vivo gene therapy study on adult MPS VII mice has shown that modification of the capsid of rAAV with BBB-targeted, MPS VII-specific motifs could shift AAV tropism from the liver to the brain, and enabled therapeutic gene delivery into BBB-forming brain microvascular endothelial cells in a disease-specific manner. Moreover, widespread enzyme distribution was observed throughout the brain, leading to significant improvement of neuropathology. Finally, the risk of inadvertent germline gene transfer related to insertional mutagenesis has raised a broad array of ethical issues and safety concerns, especially with systemic gene transfer approach [153].

CONCLUSION

Significant progress has been made toward better understanding of disease pathogenesis in neuropathic MPS, and the development of novel strategies to bypass the BBB, although major hurdles still remain. Unlike any other GAGs, the accumulation of heparan sulfate is always associated with CNS involvement, although the relationship between HS recycling and HSGP biosynthesis or functions in CNS is largely unknown. Recent progress in identifying BBB-targeted epitope-modified AAV has revealed disease- and species-dependent, molecular differences in the vascular bed of the brain [42, 152], suggesting BBB remodeling in neuropathic MPS. Intracranial stem cell- or vector-injection may present a complementary approach to patients with visceral correction by HSCT. However, localized delivery may be impractical to meet the demand of global CNS benefits in treating neuropathic MPS in human. Autologous LV-mediated HSC gene therapy in the ALD clinical trial has shown promising prospects in treating CNS abnormalities [141] with improved HSC gene transfer efficiency. However, there are still many technical issues, long-term efficacies and safety concerns are to be addressed. With the development of other therapeutic strategies in ERT and small molecule therapies, a comprehensive approach would be needed in treating neuropathic LSDs that have high phenotype-genotype heterogeneity.