Advances in therapies for neurological lysosomal storage disorders

Lysosomal Storage Disorders (LSDs) are a diverse group of inherited, monogenic diseases caused by functional defects in specific lysosomal proteins. The lysosome is a cellular organelle that plays a critical role in catabolism of waste products and recycling of macromolecules in the body. Disruption to the normal function of the lysosome can result in the toxic accumulation of storage products, often leading to irreparable cellular damage and organ dysfunction followed by premature death. The majority of LSDs have no curative treatment, with many clinical subtypes presenting in early infancy and childhood. Over two‐thirds of LSDs present with progressive neurodegeneration, often in combination with other debilitating peripheral symptoms. Consequently, there is a pressing unmet clinical need to develop new therapeutic interventions to treat these conditions. The blood–brain barrier is a crucial hurdle that needs to be overcome in order to effectively treat the central nervous system (CNS), adding considerable complexity to therapeutic design and delivery. Enzyme replacement therapy (ERT) treatments aimed at either direct injection into the brain, or using blood–brain barrier constructs are discussed, alongside more conventional substrate reduction and other drug‐related therapies. Other promising strategies developed in recent years, include gene therapy technologies specifically tailored for more effectively targeting treatment to the CNS. Here, we discuss the most recent advances in CNS‐targeted treatments for neurological LSDs with a particular emphasis on gene therapy‐based modalities, such as Adeno‐Associated Virus and haematopoietic stem cell gene therapy approaches that encouragingly, at the time of writing are being evaluated in LSD clinical trials in increasing numbers. If safety, efficacy and improved quality of life can be demonstrated, these therapies have the potential to be the new standard of care treatments for LSD patients.


| INTRODUCTION
The Lysosome plays an essential role within the cell acting as the major recycling centre to dismantle surplus and functionally redundant material including carbohydrates, lipids, nucleic acids and proteins, originating from both inter-and intracellular environments. 1 Mutations in genes that encode lysosomal proteins give rise to Lysosomal Storage Disorders (LSDs), whereby the normal function of the lysosome is impaired, typically leading to the accumulation or "storage" of substrates. 2 Accumulation of storage material first occurs in the lysosomes before gradually overburdening other intracellular compartments, resulting in cellular dysfunction and ultimately cell death and organ dysfunction. Of the approximately 70 monogenic LSDs described to date, the majority follow an autosomal recessive inheritance pattern, with the exception of three X-linked disorders: Mucopolysaccharidosis (MPS) type II, Fabry and Danon disease. In isolation, individual LSDs are rare, however, combined they account for an overall prevalence reported to be as common as 1 per 4800 live births, depending on geographical location. [3][4][5][6] In addition to this, a greater number of adults are diagnosed compared to children, suggesting that LSDs are more common in adulthood. 5 Clinical manifestations of LSDs can be extremely variable depending on the type of the storage material and physical distribution throughout the body. Furthermore, disease traits can vary significantly even within the same LSD, which is particularly apparent in diseases such as Fabry, with a huge diversity of symptoms observed. [7][8][9] Many LSDs manifest with somatic and or neurological symptoms relatively rapidly after birth, or in rarer attenuated cases into late adulthood, although those with neurological manifestations are typically diagnosed in childhood. Build-up of storage material can afflict a number of different organs, bones and connective tissue and greater than 70% of all LSDs have neurological involvement to varying degrees, with congenital or infantile presenting LSDs often having the most severe clinical symptoms on the disease continuum. 10 Progressive neurodegeneration occurs and frequently leads to death before adulthood, with patients commonly presenting with some or all of the following clinical symptoms: cognitive impairment, brain atrophy, motor decline, behavioural problems, sleep disturbances, blindness, deafness, hyperactivity and seizures. LSDs displaying a neurological phenotype are often the most arduous to treat since any therapeutic or drug delivery intervention needs to overcome the protective blood-brain barrier (BBB) in order to deliver significant impact on treating disease burden in the brain. In addition to this, it is important to also highlight the difficulties in treating skeletal abnormalities and joint disease prevalent in LSDs, with spine and hip deformities in particular causing significant complications. [11][12][13] There is currently a pressing unmet clinical need to develop more efficacious therapeutic interventions to treat these severe disorders.
In this review, we focus attention on neurological LSDs and their treatments and explore the molecular and biological features of the BBB that can be exploited for transport of therapeutic agents into the brain under normal physiologic and pathological environments. We will discuss advantages and current limitations of central nervous system (CNS)-targeted therapeutics including chaperone and substrate reduction therapies, enzyme replacement therapies and gene therapies and describe new and emerging approaches in clinical development that take advantage of receptor-mediated transcytosis and other transport mechanisms to deliver therapeutics into the brain. We will focus particular attention on the current progress of clinical translation of these treatments.

| LSD CLINICAL MANIFESTATIONS
The majority of LSDs are caused by enzyme deficiencies and are routinely classified by the type of storage material bioburden. LSD clinical manifestations are broad ranging and depend on a number of factors including the type of gene affected, the specific mutation, toxicity of the storage material, the organs affected and the age of onset. Table 1 provides an extensive overview of the main classes of LSDs; sphingolipidoses, mucopolysaccharidoses, glycoproteinoses, glycogen storage disorders and neuronal ceroid lipofuscinoses, detailing the gene affected and enzyme deficiency, the type of storage material, epidemiology, clinical manifestations and the current treatment options. For a large majority of LSDs, effective treatments do not currently exist, with symptom management and supportive treatment the only option available.

| NEUROPATHOLOGY IN LSDS
Greater than two-thirds of LSDs exhibit neuropathology and the clinical indications in these patients typically arise at an early age, often leading to severe clinical manifestations and premature death. Developing therapeutic strategies to delay or prevent neurological decline in LSD patients is therefore of utmost importance. The sequence of events leading to neurodegeneration however is complex and it appears not solely dependent on accumulation of the primary storage material. 15 Typically, accumulation of primary storage material, the undegraded T A B L E 1 Lysosomal Storage Disorders (LSDs): Subtype, gene, enzyme, storage material, Incidence, disease severity, CNS affected, age of onset and life expectancy. Information sourced from LSDs: a practical guide, GeneReviews, rarediseases.org and omim.org. 14  lysosomal substrate, is followed by storage of other substrates that have presumably become backed up in the system. Often these are substrates stored in other lysosomal diseases. Autophagy is universally defective in these diseases, and there may be vesicular trafficking defects, resulting in for example synaptic dysfunction. 16,17 Another widespread finding is inflammation, typically in conjunction with reactive oxygen species build-up, suggestive of innate immune system activation, potentially caused by undegraded substrates and lysosomal disruption. 18,19 In MPSIII, for example, build-up of secondary storage materials such as monosialogangliosides (GM2 and GM3) and activation of inflammatory pathways make a significant contribution to neuropathology. [20][21][22] In addition, microglial cells, the resident immune cells of the CNS, become highly activated in the MPS brain and this is accompanied by the secretion of both anti and pro-inflammatory cytokines as well as activation of the TLR4 inflammatory response and ultimately the inflammasome. 23-2519 Astrogliosis is a further hallmark of these conditions and a prominent feature of neurodegenerative disease in general. 15,[26][27][28] Whether or not astrocytic responses represent a protective, restorative role or a detrimental, exacerbating affect, however, is still a matter of contention. 29 Another prominent characteristic of MPS and other LSDs is the development of swollen lysosomes throughout the CNS. 30,31 For example, lysosomal swelling can clearly be visualised in murine models of MPSIIIB by immunohistochemical analysis with lysosomal associated membrane protein-2 marker (LAMP-2) in neuronal cells and is accompanied with neuronal cell loss in cortical and other brain regions. 24,32 In metachromatic leukodystrophy (MLD), accumulation of sulfatide is observed in oligodendrocytes, microglia and neurons of the CNS and in Schwann cells and macrophages of the peripheral nervous system (PNS) leading to widespread demyelination, microglia damage and neurodegeneration. 33,34 Consequently, brain atrophy and abnormalities in the central white matter ensue that can be monitored throughout disease progression by magnetic resonance imaging (MRI). 35 Similar to MLD, the neurodegenerative phenotype of Krabbe disease is typically associated with disruption of the myelin sheath in the CNS, 31,36,37 and the characteristic neuropathology of the NCLs or Batten disease is the accumulation of autofluorescent storage material within the lysosome and widespread neuronal cell death, with astrocytosis and microgliosis also occurring early on in pathogenesis. 38,39 Although the specifics of brain degeneration vary from disease to disease, and appear to be somewhat dependent on primary storage substrates and their toxicity, all the neurological lysosomal diseases ultimately result in neuronal loss and severe disability.

| EXPLOITABLE TRANSPORT ROUTES ACROSS THE BLOOD-BRAIN BARRIER FOR LSD THERAPIES
The blood-brain barrier (BBB) is critical in maintaining homeostasis in the CNS by restricting the movement of harmful molecules, neurotoxins and cells from the blood to the brain, selectively regulating the uptake of nutrients and hormones and allowing for removal of metabolites. However, the semipermeable nature of the BBB also restricts the transport of larger molecules, that is, drugs or proteins greater than 500 kDa into the brain, leading to minimal drug bioavailability in the CNS. 40 This poses a significant hurdle in the development of therapeutics for neurodegenerative disease. 41 The BBB is comprised of an extensive network of brain microvascular endothelial cells (BMECs) supported by a capillary basement membrane consisting of astrocytes, pericytes and microglia and a further basement membrane consisting of extracellular matrix proteins including collagen, elastin, fibronectin, laminin and proteoglycans which provide membrane structural support and stability ( Figure 1). 40,42 The BBB is uniform in thickness with a negatively charged surface, the purpose of which is to regulate the passage of positively charged peptides such as heparan sulfate proteoglycans in a process called adsorptive-mediated transcytosis. 43 Bridging the BMECs is a complex array of tight junctions that restrict small ions diffusing through the paracellular pathway and contribute to the selective permeability of the BBB (Figure 1). An array of transporter molecules exist to facilitate movement of solutes and other molecules across the BBB and solute carriers that transport essential small molecular weight nutrients into the brain. In the reverse direction, energydependent ATP binding cassette (ABC) transporters including breast cancer-related peptide (BCRP), multidrug resistance proteins (MRPs) and P-glycoprotein (PgP/MDR) expedite solutes out of the brain across a concentration gradient (Figure 1).
Successful delivery of therapeutics to the CNS requires understanding of the complex physiology of the BBB, namely, its response to physical and chemical stimuli, its permeability under different pathological conditions and disease states and the molecular function of the various BBB transport receptors. Perivascular oedema and microvascular leakage has been observed in LSDs as disease progresses, notably in the brain of MPS III patients and associated animal models, compromising BBB integrity and function and contributing to brain atrophy. 44,45 In order to provide the best chance of treatment success, an LSD therapeutic would need to be delivered early before the onset of neurological decline and adverse BBB disruption. To facilitate delivery of therapeutic enzyme into the brain across an intact BBB, the most exploited transport routes to date for LSD therapies include Receptor-Mediated Transcytosis (RMT), which bestows a mechanism for selective uptake of larger macromolecules and Cell-Mediated Transcytosis (CMT), which relies on immune cells, including monocytes and macrophages, to cross the BBB to deliver therapeutic protein ( Figure 1). The alternative approach is to deliver directly into the intrathecal or intraparenchymal space in the brain to circumvent the barrier physically.

| Enzyme replacement therapy (ERT)
For a number of LSDs with little to no CNS involvement, ERT has proven to be an effective treatment strategy with a number of approved treatments commercially available ( Table 2). 46 First developed in the late-90s for Gaucher disease, ERT is based on the principle of intravenous delivery of functional recombinant enzyme to patients. Once administered, enzyme enters cells by receptor-mediated endocytosis via the mannose-6-phosphate receptor pathway and is then sorted to the lysosome, in a process commonly referred to as cross-correction. 47 In the case of the Mucopolysacchardisoses, intravenous delivery of ERT can provide significant therapeutic benefit for those nonneuropathic subtypes such as MPSI Hurler-Scheie/Scheie (Iduronidase), attenuated MPSII (Idursulfase/Idusulfase β), MPS VI (Galsulfase), MPS IV (Elosulfase α) and MPS VII (Vestronidase α) ( Table 2). [48][49][50][51][52][53][54][55][56][57][58][59][60] ERT is also a standard of care for several of the Sphingolipidoses such as Gaucher type I disease, the non-neuronopathic variant (Imiglucerase, Velaglucerase α, and Taliglucerase α). ERT is also approved for Fabry disease (Agalsidase β, Agalsidase α), Wolman disease (sebelipase α) and Pompe disease (Alglucosidase α), where ERT is effective at reversing cardiac abnormalities but not the skeletal issues, similar to observations seen in MPSII patients receiving ERT (Table 2). [61][62][63][64] Unfortunately, the CNS is poorly treated with ERT due to the impermeability of the BBB to larger proteins. 65 To improve targeted delivery of recombinant enzyme to the brain, two approaches have been evaluated in proof-of-concept animal studies and in the clinic: (1) direct administration of recombinant enzyme via intrathecal (IT) or intracerebroventricular (ICV) routes by-passing the BBB completely and (2) modification of F I G U R E 1 Transport routes across the blood-brain barrier. Schematic diagram highlighting the various transport mechanisms across the blood-brain barrier including receptor-mediated transcytosis, cell-mediated transcytosis, transcellular lipophilic pathway, paracellular aqueous pathway, efflux pumps, transport proteins and adsorptive transcytosis.  Twenty-four weeks post-treatment, spleen and liver volumes normalised in 7/9 and 9/9 patients, respectively, and developmental quotients (DQ) were stabilised in 5/7 patients. A phase 2 extension study, designed to evaluate the long-term safety, tolerability of AXE 250 and the long-term effect on cognitive function in MPSIIIB patients was initiated in 2018 and is currently ongoing at time of writing (NCT03784287) (now owned by Allievex). To date, this approach has demonstrated promising therapeutic outcomes when using a higher dosing regimen compared to earlier studies and the ICV route appears to be a more effective delivery method than IT. A similar therapeutic for MPSII delivering unmodified idursulfase β via the ICV route has resulted in regulatory approval in Japan. 70 Likewise, delivery of unmodified cerliponase α via the same ICV route has resulted in FDA approval for the treatment of CLN2. 70,71 Both are effective therapies at reducing neurological symptoms ( Table 2). As these successes are relatively recent and were dependent on delivery via the Ommaya reservoir, the alternative strategy of modifying recombinant enzymes to improve uptake and transport across the BBB when administered intravenously has been extensively explored. Researched methodologies that allow ERTs to more successfully penetrate the BBB include fusing the enzyme to molecules that recognise specific BBB carrier proteins including ApoB, ApoEII, Fc antibody fragments, IGFII and encapsulation into nanoparticles which predominantly facilitate transport across the BBB by RMT. [72][73][74][75][76][77][78][79][80][81][82] An IgG-IDUA fusion protein has been developed for the treatment of severe MPSI (valanafusp alpha), where the IgG domain targets the human insulin receptor on the BBB to facilitate transport of enzyme into the brain. In clinical trials treating paediatric MPSI patients with CNS involvement, valanafusp alpha treatment was demonstrated to stabilise DQ, cortical grey matter volume and somatic manifestations and improved urinary GAG levels, hepatic spleen volumes and shoulder range of motion in the majority of patients. 83 A similar approach was developed by Armagen (acquired by JCR pharmaceuticals) for MPSII, conjugated the anti-human insulin receptor antibody with iduronate-2-sulfatase with similar therapeutic benefits (AGT-182) (NCT02262338). Another approach is to couple the transferrin receptor Ab to recombinant enzyme to enhance BBB penetration. JCR Pharmaceuticals have combined anti-human transferrin receptor antibodies with α-L-iduronidase (JR-171) and iduronate-2-sulfatase (JR-141) to enable BBB penetration via IV administration. 84 A phase I/II study of JR-171 delivered by weekly IV administration in MPS I patients was concluded in 2020, however, results are yet to be published (NCT04227600). A follow on extension study (JR-171-101) to determine optimal dose began in 2021 (NCT04453085). A similar trial with weekly administration of JR-141 in MPSII patients reported HS and DS suppression in plasma and urine during a 4-week treatment period with a significant decrease of HS in the CSF at the 3-week time point (NCT03128593). 85 Following this, an open-label phase II/III study in MPS II patients was conducted in order to evaluate the safety and efficacy of JR-141 during 52 weeks of treatment (NCT03568175). Denali Therapeutics have developed a similar approach for MPSII by combined the antibody binding site of antihuman transferrin receptors with IDS to enhance uptake of IDS into the brain (DNL-310) and are currently performing a phase I/II trial in MPS II children with weekly DNL-310 IV administration over a 24 week period (NCT04251026).
For MPSIIIA, a chemically modified variant of recombinant human sulfamidase (SOBI003) has been shown to cross the BBB in mice and achieve pharmacologically relevant levels in CSF. 86 Following on from this, a first-in-man phase 1/2, open-labelled clinical study (NCT03423186) was conducted and its extension study (NCT03811028) to evaluate the long-term safety, tolerability, pharmacokinetics/ pharmacodynamics (PK/PD) and clinical efficacy of SOBI003 in MPSIIIA patients for up to 2 years. The results demonstrate SOBI003 was generally well tolerated, with serum concentrations of the drug increasing in proportion to the dose delivered and presence of SOBI003 in the CSF confirmed its ability to cross the BBB. 86 Anti-drug antibodies were present in the serum and CSF of all patients. An overall reduction in HS levels in CSF of 79% was recorded at the last assessment, together with reductions in HS levels in serum and urine. Neurocognitive development age-equivalent scores showed a stabilisation of cognition for all patients, whereas no clear overall clinical effect was observed on adaptive behaviour, sleep pattern or quality of life. 86 As promising as these enzyme-fusion modalities are for improving efficacy in the brain, the ERT approach has several fundamental disadvantages that limit its effectiveness as a long-term therapeutic strategy for LSDs. In addition to the challenge of neurological targeting, there is also typically poor penetrance of skeletal, cardiovascular and ocular symptoms and ERT requires repeated life-long weekly/bi-monthly administration due to the limited half-life of the enzymes, and this has a considerable impact on financial costs of the treatment for the lifetime of the patient, which are typically very high. Another major concern is that a large proportion of patients receiving ERT eventually develop neutralising allo-antibodies which can limit treatment effectiveness. 87 Cross-reactive immunological material (CRIM) statusnegative patients typically have a poor clinical response to ERT, secondary to high sustained antibody titres as seen in Pompe patient populations receiving ERT. 88 Promisingly for some LSDs where ERT is the standard of care, such as MPSI hurler-scheie/scheie and Pompe disease, effective strategies are in development to induce immune tolerance to ERT, effectively alleviating the neutralising effect of anti-drug antibodies. For example, in proof-of-concept studies using MPSI mice treated with laronidase, two courses of non-depleting anti-CD4 monoclonal antibodies were able to ablate immune responses to laronidase in comparison to methotrexate that was ineffective. 48,89,90

| Chaperones and substrate reduction therapy
Pharmacological chaperone therapy (PCT) is an approach that utilises small chaperone molecules that are designed to bind mutant proteins and correct their conformation, thereby improving structural stability, enzymatic function and lysosomal trafficking. 91 In LSDs with missense mutations, enzymes are synthesised but typically prevented from trafficking to the lysosome due to incorrect folding and retained in the ER, making these diseases potential candidates for PCT. One of the main advantages that chaperones have compared to ERT is superior biodistribution throughout the body, with greater penetrance of cells and tissues and the ability to cross the BBB to allow treatment of the CNS. Another benefit is that PCT drugs can be ingested orally, unlike ERT which requires weekly infusions, thereby reducing the need for frequent hospital visits. [92][93][94] Several studies suggest a low threshold enzyme activity of 10%-15% maybe sufficient to prevent storage in some LSDs. 24,25,95,96 Therefore, if there is potential to recuperate enzyme functionality with PCT, even by a small degree, it will likely improve disease pathology and have a positive impact on patients' quality of life. Following several successful PCT proof-ofconcept studies in animal models, chaperones have been evaluated in the clinic setting for several LSDs including Fabry, Gaucher, Pompe and Niemann-pick C disease (NCT03135197, NCT03950050 and NCT03911505). [97][98][99] Migalastat, a molecular inhibitor of alpha-galactosidase (alpha-GalA), demonstrated significant therapeutic benefit in patients in clinical trials and as such became the first PCT to gain FDA approval for the treatment of adult Fabry patients with amenable mutations. 100,101 Similarly, the pharmacological GCase chaperone ambroxol, has been evaluated in the clinical for Gaucher patients demonstrating improvements in neurocognitive function when administered at higher doses. 102,103 In Tay-Sachs disease, specific mutations lead to misfolded HexA enzymes that degrade too rapidly before trafficking to the lysosome can occur. In proof-of-concept studies, the chaperone compound pyrimethamine was found to stabilise HexA in cells with the late-onset mutation alphaG269S, however, when evaluated in the clinic, the treatment did not lead to significant improvements in patients with several adverse side effects reported. 104,105 Substrate reduction therapy (SRT) was developed with the aim of reducing the quantity of excess substrates, produced in LSDs, by inhibiting its synthesis. Substrate synthesis inhibitors, like PCTs, have the advantage of being small enough to cross the BBB and therefore have the potential to treat the CNS in neurodegenerative LSDs. Miglustat, a glucosylceramide synthase inhibitor, is approved for the management of mild to moderate type I Gaucher disease for patients not eligible for ERT. 106 In addition, miglustat was the first approved therapy for Niemann-Pick type C patients, gaining approval in the EU, Brazil and South Korea for the treatment of the progressive neurological symptoms observed in both paediatric and adult patients. 107 Eliglustat, another glucosylceramide synthase inhibitor, is also approved for the treatment of Gaucher and approved by the FDA in 2014. 108 Important to note, however, is that both Miglustat and Eliglustat are not approved or recommended for use in children or teens with Gaucher disease, and these SRT drugs can have unpleasant side effects including diarrhoea, stomach problems and neuropathy. 109 Conversely, Miglustat appears to be relatively well tolerated in children with GM1 type 2 gangliosidosis, with early clinical data suggesting the drug is more effectively utilised to slow down, rather than treat, both visceral and neurological manifestation, as in Niemann-Pick type C paediatric patients. 110,111 The rarity, rapid clinical progression, and truncated lifespan of LSD patients such as GM1 gangliosidosis type 2, make clinical surveillance and evaluation of therapeutic efficacy tough, as effective interventions are only possible before irreversible damage occurs. Further studies with larger patient cohorts and extended follow-up are required to better understand the long-term therapeutic effects of Miglustat in GM1 type II gangliosidosis. 112 Genistein (4 0 ,5,7-trihydroxyisoflavone), a soy-derived isoflavone, is another SRT that has been evaluated as a potential treatment for MPS. 113 Genistein inhibits tyrosine kinase activation of the EGF receptor, thereby disrupting transcription of enzymes involved in GAG synthesis observed in MPS disorders. Whilst demonstrating neurological improvements and correction of the behavioural phenotype in a proof-of-concept MPSIIIB murine study, Genistein proved ineffective when assessed in MPSIII patients during a single centre, double-blinded, randomised, placebo-controlled clinical trial with no clinically meaningful reductions in biomarkers or improvement in neuropsychological outcomes. 113,114 6 | GENE THERAPY TREATMENTS FOR NEUROPATHIC LSDS Similar in principle to ERT, gene therapy approaches can be utilised to deliver enzyme for LSD treatment. A major advantage of gene therapy-based approaches is the potential to provide a long-term, sustained therapy without the need for continual re-administration. To treat neurological manifestations in LSDs, gene therapies can be delivered directly to the CNS, systemically or by ex vivo administration in combination with haematopoietic stem cell transplant. At the time of writing Adeno-Associated Viruses (AAVs) (Figure 2) and lentiviral-mediated Haematopoietic stem cell gene therapy (HSC-GT) (Figure 3), approaches offer the most potential for long-term clinical benefit with numerous applications proving effective in in vivo proof-of-concept studies that have been taken forward for evaluation in the clinic. [115][116][117] 6.1 | AAV gene therapy AAV vectors are highly effective gene delivery vehicles that can be engineered to deliver therapeutic DNA to range of specified target cells, making them particularly attractive for the treatment of monogenic diseases. AAV vectors comprise of a protein shell encapsulating a singlestranded DNA genome of up to 4.8 kilobases (kb). 118 The inverted terminal repeat (ITR)-flanked transgenes form circular concatemers that persist as extrachromosomal episomes in the nucleus of transduced cells, with limited integration into the host genome. 118 AAV gene therapies can deliver high levels of gene expression in the target cell type and crucially this can occur very rapidly, providing the potential for long-lasting therapeutic benefit from a single treatment that could be transformative to patients. Indeed, AAV vectors are fast emerging to be the in vivo gene therapy platform of choice for treatment of neurological diseases, with numerous AAV serotypes described in recent years that facilitate specific targeting to the CNS. 119,120 Intravenous (IV) administration of CNS-targeted AAV gene therapy vectors would be the preferred, noninvasive delivery route for the treatment of neuropathic LSDs. Indeed, AAV serotypes 9 and rh.10 have proved effective at crossing the BBB and transducing CNS cells when delivered IV in a number of in vivo proof-ofconcept studies and several phase I/II clinical trials are currently ongoing for MPSIIIA, Pompe, GM1 gangliosidosis and Gaucher type 2, evaluating safety and efficacy with escalating vector doses delivered IV (Table 3). An issue with IV administration is the substantial volume of vector typically required compared to direct CNS delivery methods which could cause issues relating to cost and scale-up for clinical translation. 121,122 More importantly, systemic delivery of neurotrophic vectors, particularly at higher doses, can cause toxicity and typically results in non-CNS organs being targeted, predominantly the liver, which can trigger adverse immune responses, thereby limiting both efficacy and safety of treatment. 123,124 Nevertheless, strides are being made to improve AAV CNS targeting whilst reducing off-target transduction. In a recent study, IV-administered AAV.CAP-B10, selected from an engineered AAV capsid library, was found to have high specificity for neurons in the CNS of treated mice with significantly reduced transgene expression in peripheral organs including the liver. 125 Furthermore, this finding was reproduced in a larger animal model where robust transgene expression was observed in the CNS of adult marmoset, with minimal expression in the liver, following IV administration of AAV.CAP-B10 and demonstrated a significant improvement compared to AAV9 and AAV-PHP.eB vectors. 125 To address peripheral disease such as skeletal and cardiac manifestations common to a number of LSDs, IV administration of AAV vectors with a broad serotype tropism that can target the CNS and desired somatic organs will likely be the best F I G U R E 2 AAV-mediated gene therapy for LSDs. AAV gene therapy vectors packaging a single-stranded (ss) DNA transgene sequence expressing the correct form of the defective gene can be delivered intravenously or by direct administration routes to the central nervous system (CNS). Direct CNS administration routes include intraparemchymal, IntraCisternalVentricular (ICV), IntraCisternaMagna (ICM), IntraThecal (IT) delivery, intranasal and via the neuromuscular junction. Neurotrophic serotypes such as 1, 5, 9, and rh.10 more effectively target the AAV vector to the CNS. Once inside the body the AAV vector fuses with the target cell membrane, before being quickly internalised via receptor-mediated endocytosis through clathrin-coated pits and trafficked to the nucleus. Following shedding of the viral coat the viral DNA is expressed and transgene mRNA released from the nucleus before being translated into protein.
F I G U R E 3 HSC-GT for LSDs. A lentiviral vector encoding the therapeutic transgene is used to transduce isolated haematopoietic stem cells from the patient. Following LV fusion and release of viral contents, the ssRNA is reverse transcribed and trafficked to the nucleus where the therapeutic transgene is then integrated into the host cell genome. The genetically modified stem cells are transplanted back to the patient after they receive a conditioning regime to promote engraftment. The modified cells are able to differentiate and traffic to various organs including the brain where they can then manufacture and secrete enzyme that can be taken up by neighbouring cells via the mannose 6 phosphate receptor (M6PR).
T A B L E 3 Current active/recruiting AAV Gene Therapy Clinical trials for LSDs reported on ClinicalTrials.gov as of Mar2023. approach going forward, in combination with a transgene driven by ubiquitous promoters such as the human cytomegalovirus (CMV) or the hybrid CMV enhancer/ chicken β-actin (CBA) for the capacity of high overall gene expression in a variety of different tissue types. 126,127 AAV serotypes 1, 5, 9, and rh.10 are particularly effective at transducing neuronal cells when directly administered to the CNS, with various delivery routes possible dependent on the target region and distribution required, these include intraparemchymal, IntraCisternalVentricular (ICV), IntraCisternaMagna (ICM), IntraThecal (IT) and intranasal delivery ( Figure 2). 119,[128][129][130][131][132][133][134] Numerous CNS targeting AAV treatments for LSDs are being evaluated to determine the most suitable serotype and delivery method for optimal therapeutic benefit. In Krabbe disease, for example, ICV delivery of AAVhu68 expressing codon-optimised human GALC (AAVhu68. GALCco) was compared to IV administration in neonatal Twitcher mice in a dose escalating study. 135 High-dose IV administration (1 Â 10 11 GC, equivalent to 1 Â 10 14 GC/kg) increased median survival from 40.5 days in vehicle-treated mice to 49 days in AAV-treated mice, whilst a five-fold lower dose, 2 Â 10 10 GC (1.3 Â 10 11 GC/g brain), administered ICV produced a greater survival benefit of 62 days median survival. The high dose of 1 Â 10 11 GC (6.7 Â 10 11 GC/g brain) administered ICV achieved the longest median survival of 130 days. The intended clinical route, a single ICM administration dose of AAVhu68.GALCco at 3.0 Â 10 13 GC (6.0 Â 10 11 GC/g brain, equivalent to highest mouse dose), was then evaluated in a canine model of Krabbe disease demonstrating efficacy of treatment, including normalisation of peripheral nerve function. The same group then performed a toxicology study in rhesus macaques confirming the safety of the approach in preparation for a first-in-human trial. 135 In vivo proof-of-concept studies in murine and rodent models of neuronal ceroid lipofuscinosis type 7 (CLN7) disease have demonstrated efficacy and safety of IT-delivered AAV9 expressing MFSD8, the membranebound lysosomal protein effected in the condition. 136 Mfsd8 À/À (KO) mice, were injected IT at P7-P10 (presymptomatic) or P120 (early symptomatic) with a single high (5 Â 10 11 vg/mouse) or low (1.25 Â 10 11 vg/mouse) dose of the AAV9/MFSD8 vector. IT-delivery of the AAV9/ MFSD8 vector resulted in a dose-dependent increase of MFSD8 vector DNA across the CNS and peripheral organs and a significant effect on survival in both an agedependent and dose-dependent manner, with the early treatment and high-dose group, showing a larger increase in survival with the greatest improvements in behavioural, biochemical and immunohistochemical outcomes. 136 In MPSIIIA preclinical studies, a chimeric sulfamidase containing an alternative signal peptide to improve enzyme secretion and sulfatase-modifying factor 1 (SUMF1) to increase sulfamidase post-translational activation rate was shown to enhanced enzyme biodistribution in wild-type pigs upon IT-delivered AAV9-mediated gene delivery. 128 Unique hybrid AAV serotypes, such as AAV-TT (truetype), have recently been developed that significantly improve AAV neurotropism and distribution throughout the mouse brain compared to benchmark AAV9 and AAVrh10 vectors, with improved ability to treat the neurological phenotype in MPSIIIC where the defective lysosomal enzyme, HGSNAT, resides transmembrane. 137 In addition, various injection techniques are currently being researched to improve vector delivery to the CNS and to minimise localised tissue damage. For example, improved convection enhanced delivery techniques can allow a greater volume of vector to be administered to the CNS whilst keeping the flow rate low to limit localised damage. 138,139 6.2 | AAV gene therapy clinical trials for LSDs CNS-targeted AAV vectors have demonstrated encouraging therapeutic profiles when evaluated in various LSD animal models including GM1 gangliosidosis, Sandhoff disease, MLD, Krabbe disease, Tay-Sachs disease, Pompe disease and several of the MPS subtypes including MPSI, II, III and VI. 117,121,129,[140][141][142][143][144][145][146][147][148] In the majority of in vivo preclinical studies, AAV vectors offer an effective means to correct or substantially improve the neurodegenerative disease phenotype, as such the most promising approaches have been taken forward for evaluation in LSD patients in numerous AAV clinical trials. 116,149 Table 3 summarises current active and recruiting AAV clinical trials for LSDs at time of writing. Details of the outcomes of these trials, many of which are still ongoing and remain largely unreported are discussed in the following review. 150 In a phase I/II clinical trial for MPSIIIA (Lysogene), four children with cognitive impairment received intracerebral injections of an adeno-associated viral vector serotype rh.10-SGSH-IRES-SUMF1 vector (NCT01474343 and NCT02053064). Vector was delivered at a dose of 7.2 Â 10 11 viral genomes/patient simultaneously via 12 needles as deposits of 60 μL over a period of 2 h bilaterally to the white matter anterior, medial, and posterior to the basal ganglia. Brain atrophy evaluated by magnetic resonance imaging was stable in two out of four patients but increased in the other two patients. Neuropsychological evaluations suggested a possible although moderate improvement in behaviour, attention, and sleep in three patients. It was concluded that neurocognitive benefit was more likely in patients treated at a younger age. 151 Tardieu et al., also evaluated AAV gene therapy in MPSIIIB patients in a similarly designed phase I/II trial. 152 Four patients were enrolled with treatment consisting of 16 intraparenchymal deposits (four in the cerebellum) of a rAAV2/5 encoding human α-N-acetylglucosaminidase (NAGLU) plus immunosuppressive therapy. Compared to natural history, neurocognitive progression was improved in all four patients receiving treatment, with the youngest patient having function close to that of a healthy child. In the oldest patient, neurocognitive skills declined over follow-up but slower than natural disease course. Longer follow-up is required to determine long-term safety outcomes and to assess whether improved cognitive development persists or whether the disease reverts to its natural course over time. 152 Lysogene reworked its AAV vector (SAF302CAG) to include a CAG promoter driving human SGSH expression and delivered under an AAV2/rh10 vector intracerebrally administered into both sides of the brain through image-guided tracks (NCT03612869). 153 The study resulted in localised findings on MRI at the intracerebral injection sites and a trial halt for MPSIIIA. 154,155 It is hard to know at this point if the MRI findings are linked to the injection, but AAV delivery by this route does result in very high proximal expression at the delivery site and thus it is possible that AAV or gene related toxicity is a factor. In trials conducted by Abeona for GM2 and MPSIIIA delivering AAV therapies by IV, the youngest patients (those under 30 months of age) have responded best whilst older treated individuals demonstrated little improvement. 156,157 This suggests that neurological lysosomal diseases may have a shelf life for treatment, which in this disease is very young indeed.
The first-in-human evaluation of AAV gene therapy in Tay-Sachs disease was recently conducted. 148 Two patients were treated, the first at 30 months by IT administration because of severe thalamic degeneration. The second was treated at 7 months, by bilateral thalamic injections followed by IT delivery. Generally, the procedure was well tolerated in both patients with no neurologic deficits or other notable acute procedural complications, however, patient outcomes were not as positive as expected with only slight deviation from natural history of the disease in patients treated at a younger age. 148 Recent reports have suggested that IV and ICV injections may result in a preferential transduction of glial cells in the brain, whilst intraparenchymal injection often appears to result in better transduction of neuronal cells. 158 Obviously serotype also plays a role, but overall glial transduction is likely to result in more rapid turnover and loss of AAV as most of the vector remains episomal, and thus dilutes with cell division, whilst the slow turnover of neurons lends itself to long-term sustained expression. Our own work on the development of convection enhanced delivery of AAV to the brain, by reducing proximal expression to intraparenchymal injection sites whilst improving distal expression through for example detargeting of AAV could be a positive step forward.
Whilst the efficacy and safety of the AAV approach has been well demonstrated in preclinical animal studies, there are several drawbacks that can impede successful translation to the clinic. Insufficient distribution within the brain remains an issue even when administered ventricularly as in comparison to in vivo disease models the human brain is considerably larger in volume with greater structural complexity. Scale-up manufacturing for both IV (high dose required) and brain delivery (low volume) methods can be considerable especially if high titres are required. 159 Another challenge facing the successful translation of effective AAV gene therapy treatments to the clinic is host humoral immunity. Typically, 20%-40% of potential patients have a high prevalence of pre-existing anti-AAV antibodies (αAAV-Abs) rendering them ineligible for treatment. 160 In addition, immune responses to AAV are common following rAAV gene therapy treatment and routinely patients require immune suppression to be prescribed during and after treatment, which may introduce potential adverse side-effects ( Table 3). The presence of αAAV-Abs can inhibit the effectiveness of AAV gene therapy treatment and limit the possibility of re-administration. Various strategies have been investigated to circumvent pre-existing αAAV-Abs including AAV capsid modification, plasmapheresis and pharmacological immunosuppression with varying levels of success. 161 In a recent study, recombinant Ide-S op , an extracellular cysteine protease designed to cleave IgG molecules, was effectively utilised to rapidly deplete human and rabbit IgGs in vitro and in vivo in rabbitized mouse models with high levels of αAAV9 IgGs and αAVV9-Nabs. 162 An IV injection of high dose AAV9-hSGSH op vector (5 Â 10 13 vg/kg) at 24 h post Ide-S op treatment led to transduction as effective in αAAV9-Abs+ MPS IIIA mice, as in αAAV9-Abs-negative controls. These tools could potentially address the current issues with pre-existing AAV-Abs in LSD patients suitable for rAAV gene therapy.

| HSC-GT for treatment of LSDs
Haematopoietic stem cell transplantation (HSCT), also referred to as an allogeneic stem cell transplant, has been the standard of care for several non-neurological LSD subtypes since the 1980s. 163 HSCT is a technique that involves the transplantation of stem cells from a healthy donor (related or non-related) to a patient with the aim of replacing their own HSCs with donor HSCs that can express a functional copy of the defective gene, thereby allowing the patient to produce functional enzyme. HSCT requires high-intensity conditioning, typically by a chemotherapy regiment, in order to abolish the patient's own stem cells. This weakens the immune system to help keep the body from rejecting the donated cells posttransplant and also to provide a niche in the bone marrow to promote successful engraftment to occur. Transplanted healthy donor cells secrete functional enzyme which is then taken up by the affected cells in the recipient through cross-correction. HSCT is routinely prescribed as the standard of care treatment for MPSI Hurler following an early disease diagnosis, often greatly increasing life expectancy and improving several clinical features including correction of hepatosplenomegaly, clearing of corneal clouding, reduced and less severe cognitive decline and improvements to reaching development milestones. [164][165][166] Despite successful transplantation, many skeletal, cardiac and CNS manifestations remain partially refractory to correction and many continue to require surgical interventions. 163 A combination therapy where ERT treatment is provided post-HSCT may potentially alleviate the poorly corrected somatic manifestations by HSCT alone. 167 HSCT also significantly improves the high incidence of neutralising allo-antibodies observed in Hurler patients after receiving ERT. 168 Allogeneic transplantation has several drawbacks to consider. Despite prescribed immunosuppression, the patient's body may still reject the donated stem cells before they are able to engraft in the bone marrow, recognising them as foreign and destroy them. Moreover, immune cells originating from the donor may attack healthy patient cells commonly referred to as "graft-versus-host-disease" (GVHD). In a comprehensive study of 25 Hurler patients receiving HSCT, 9 patients were reported to have GVHD to varying degrees of severity, with the skin, intestines, liver, muscles, joints and eyes predominantly affected. 163 Treatments are available for GVHD, but in severe cases, GVHD can be fatal.
Autologous haematopoietic stem cell gene therapy (HSC-GT) is one of the more appealing therapeutic strategies for the treatment of neurodegenerative LSDs, since patients' own haematopoietic stem cells can be genetically engineered to express supraphysiological levels of functional lysosomal enzyme. Critical to the success of the strategy is that engrafted cells, predominantly of the macrophage lineage, migrate to affected tissues and can circumvent the BBB by cell-mediated transcytosis and engraft within the CNS (Figure 3). 35 Since multiple copies of the functional gene can be integrated into the HSC genome, supra-normal expression of therapeutic enzyme can be achieved and taken up by affected neurons, enabling the breakdown of storage material and cross-correction of afflicted CNS cells as well as in the periphery, more effectively so compared to a standard allogeneic transplant. 22,24,95 Unlike with an allogenic transplant there is no risk of graft versus host disease since the patients' own HSCs are used. Furthermore, the intrinsic self-renewal ability of HSCs means that HSC-GT is likely to be a one-off treatment without the need for repeated dosing, as is the case with current enzymereplacement therapies and possibly with AAV-based therapies due to potential episomal loss over time.
The HSC-GT approach was first pioneered as a treatment for primary immunodeficiency disorders such as adenosine deaminase severe combined immunodeficiency (ADA-SCID) and Wiskott-Aldrich syndrome (WAS), and more recently for X-linked chronic granulomatous disease, demonstrating life-changing clinical benefits. [169][170][171][172][173][174] Biffi and co-workers were the first group to adapt the HSC-GT approach for the treatment of an LSD, initially demonstrating proof-of-concept in a murine model of MLD. A lentiviral vector (LV) expressing the ARSA gene, driven by the ubiquitous PGK promoter, was utilised to genetically modify donor MLD HSCs that were then transplanted into MLD recipient mice resulting in prevention of neurological damage. 175 The approach has since been translated to the clinic and evaluated in MLD patients in phase I/II clinical trials with significant clinical improvements in neurocognitive function and overall therapeutic efficacy observed. 35 Follow-up studies are ongoing in these treated patient populations to evaluate if therapeutic benefit is sustained over an even longer extended period, nonetheless, this product was approved in Europe in 2021 as Libmeldy (Table 2).
Following the success of HSC-GT for MLD, the approach has been tailored to treat several of the mucopolysaccharidoses where substantial CNS involvement is prevalent. MPSIIIA is a prime example where previous therapeutic strategies, based on ERT and allogenic HSCT, have had poor therapeutic outcomes likely due to the limited bioavailability of functional SGSH enzyme in the CNS with these delivery methods. 176,177 Sergijenko et al. refined the LV genome used in previous MLD studies by switching to a myeloid specific promoter, CD11b, in order to improve transgene expression in the myeloid cells trafficking to the brain. MPSIIIA donor HSCs transduced with CD11b.coSGSH LV were able to effectively correct behaviour and neuropathology of MPSIIIA recipient mice 6 months post treatment, and more effectively so than the ubiquitous PGK promoter. 25 Subsequent preclinical evaluations have confirmed safety and efficacy of MPSIIIA HSC-GT in a humanised mouse model and in human CD34+ HSCs prior to first-in man studies. 178 A phase I/II clinical trial in MPSIIIA patients has since opened, primarily to assess safety with efficacy as a secondary outcome measure (ClinicalTrials.gov Identifier NCT04201405, Table 3). 179 The clinical and research teams conducting the trial have published clinical trial design guidelines applicable for the treatment of MPS disease by HSC-GT. 180 In addition to MPSIIIA, efficacy of the HSC-GT approach has been successfully demonstrated for the similar LSD MPSIIIB and MPSI Hurler in preclinical proofof-concept studies. 24,181 At time of writing, a HSC-GT approach for MPSIIIB is currently being developed for clinical evaluation, whilst a clinical trial of HSC-GT for MPSI Hurler is ongoing with encouraging interim findings (NCT03488394) ( Table 3). 24,182 After a median interim follow-up period of 2.1 years, Hurler patients receiving lentiviral vector-based HSC-GT confirmed overall safety and feasibility of the treatment with promising metabolic and early clinical outcomes. 182 Median vector copy number was 0.98 per genome in PBMCs at 9 to 12 months post treatment and at 12 months all patients demonstrated supraphysiological blood IDUA activity with a steep decline in pathological GAG excretion in the urine. Encouragingly, early neurological outcomes indicate detectable IDUA levels in the CSF and all patients progressively acquired motor skills. Brain MRI performed at 12 and 24 months post treatment showed reductions in white-matter and perivascular space abnormalities compared to baseline in five of eight patients evaluated to date.

| HSC-GT to deliver BBB penetrating enzymes
With the aim to further improve therapeutic levels of functional enzyme in the brain, the HSC-GT approach has been enhanced in order to deliver therapeutic enzyme fused to BBB-penetrating peptides, thus allowing enzyme to circumvent the BBB by both receptormediated (peripheral expressed enzyme) and cellmediated transcytosis (myeloid cells expressing enzyme). Gleitz and colleagues evaluated a brain-targeted HSC-GT for MPSII using lentiviral IDS fused to ApoEII (IDS. ApoEII) and compared to a lentivirus expressing normal IDS and a standard HSCT. All treatments corrected peripheral disease, but only IDS.ApoEII-mediated complete normalisation of brain pathology and behaviour, providing significantly enhanced correction compared to non-modified IDS whilst a normal bone marrow transplant achieved no brain correction. IDS.ApoEII was taken up and transcytosed across brain endothelia significantly better than IDS via both heparan sulfate/ ApoE-dependent receptors and mannose-6-phosphate receptors. 95,183 A similar BBB-targeting peptide system, iduronidase fused to apolipoprotein E, was successfully utilised for application of HSC-GT to treat MPSI Hurler, effectively normalising and improving the CNS deficits in Hurler mice providing long-term brain metabolic correction and normalisation of exploratory behaviour deficits. 184 The major challenges faced when translating HSC-GT to the clinic will lie in attaining of good manufacturing practice (GMP) grade lentiviral vector which is currently difficult to produce at a sufficient, cost effective scale. With HSC-GT therapy the treatment window is narrow and patients will likely need to be treated before the onset of clinical symptoms to see the best clinical outcome. Several months can pass before sufficient monocytes have trafficked to and distributed throughout the brain following bone marrow engraftment with further time required for therapeutic levels of enzyme to be produced in the CNS. Patients typically require a full conditioning regiment in order to provide the niche for genetically modified HSCs to engraft. Busulfan conditioning is not without risk and typically there is a small mortality risk associated with the conditioning treatment, however, refinements in transplantation procedures over the last decade have significantly improved the safety of bone marrow transplantation. 185

| HSC-GT clinical trials for LSDs
In early clinical trials conducted pre-2000, several failures were reported that demonstrated variable levels of longterm transgene expression and limited clinical effect, due to sub-optimal transduction conditions, conditioning regimes and or vector design. [186][187][188][189] Since then, there have been a number of successful HSC-GT clinical trials, particularly for primary immunodeficiencies. 190 Strimvelis has since been EMA approved for ADA-SCID in 2016, however, its use is currently suspended pending an investigation into a potential link between the drug and a case of Lymphoid T-cell leukaemia. 191 Table 4 summarises the current active HSC-GT clinical trials for LSDs at time of writing.

| GENE EDITING TECHNOLOGIES FOR LSDs
The recent advancements in genome editing technologies such as clustered regularly interspaced short palindromic repeat-CRISPR-associated protein 9 (CRISPR-Cas9), zinc finger nucleases (ZFN) and transcriptional activator-like effector nucleases (TALEN), brings the possibility of new treatments for a range of genetic disorders including LSDs. 192,193 These gene editing systems allow for correction of mutations at specific sites via double-strand break and homologous recombination. Ex vivo and in vivo genome editing ZFN platforms have been under evaluation primarily for the treatment of MPSI and MPSII in murine models. 194,195 These promising animal studies led the way to evaluation in clinical trials for MPSI and MPSII (ClinicalTrials.gov, NCT02702115 and NCT03041324). Unfortunately, outcomes of these trials have been disappointing and datasets are not published. These treatments, however, are only intended for patients with no CNS involvement due to manufacture of enzyme in the liver which cannot readily cross the BBB. Preclinical studies using CRISPR-Cas9 to treat MPSI were also unsuccessful at treating neurological symptoms of disease, limiting the scope for the treatment of severe neurological LSDs with current gene editing methodologies. 196 An ex vivo gene editing strategy to modify haematopoietic stem cells may offer a way of reaching the CNS, but currently, this approach has several drawbacks over HSC-GT. There are multiple gene mutations that can give rise to the same disease phenotype meaning multiple versions of the gene editing tools will need to be manufactured to address all patient genotypes. Furthermore, the gene editing approach is not 100% efficient and, therefore, will have limited efficacy and will likely only restore gene function to normal or subnormal levels, rather than over expression as seen in HSC-GT. This would mean treatment would only be as efficacious as an allogenic bone marrow transplant, which we know to be ineffective for treatment of neurological MPS. An alternative is the delivery of a replacement gene into a safe harbour locus, where disruption does not have adverse effects on the cell and robust transcription can be achieved but this is fraught with the reduced efficiency of homologous recombination versus non-homologous end joining. The relatively low repair percentage and the fact that AAV6 mediated repair of HSCs appears to reduce the repopulation potential of these cells in immunodeficient mice, so there are currently significant technical issues to solve. 197 Finally, there is a significant challenge in determining safety of therapeutic gene editing approaches before translating to the clinic, with particular emphasis on evaluating potential off-target affects, tumourigenicity and immunogenicity.

| CONCLUDING REMARKS
The development of efficacious, safe and affordable therapies is of paramount importance for neurological LSDs where virtually no effective treatment strategies currently exist. New emerging therapy methodologies offer an exciting prospect to be new standard of care treatments for neurodegenerative LSDs with the potential to deliver significant improvements to disease pathology, quality of life and life expectancy. Advancements in ERT design, fusing recombinant enzymes to peptides that facilitate uptake across the BBB, hold great promise for the future LSD treatments. Both ex vivo and in vivo gene therapy delivery systems have their unique benefits and challenges for treatment of CNS disease. HSC-GT offers a permanent treatment modality, with fewer scale-up considerations, provides tolerance, is well distributed to the CNS and periphery and more cost effective to produce, however, myeloablation still carries a small but significant mortality risk, there is a 3-4 week delay before any measurable CNS effect and there is the potential for insertional mutagenesis, although no incidences have thus far been reported in any of the completed HSC-GT clinical trials. Treatment with AAV vectors is likely to provide long-term benefit, with the potential for multiple routes of delivery, AAV vectors can be produced at high titre and their effect is likely to be immediate given the direct route of delivery. However, the presence of neutralising antibodies in up to 40% of candidate patients may limit eligibility for treatment or impair the effectiveness of the treatment. The high doses required and scale-up manufacture of AAV vector is another challenge that will need to be addressed. Inefficient distribution to different brain regions can hopefully be overcome with novel serotypes in the future, several of which appear to be significantly more efficacious than AAV9, the current gold standard for the brain. Another important point to consider is that treatment of patients post-symptomatic will mean that damage already present cannot be reversed, consequently treatment may stabilise disease and halt further progression but will not undo damage that has already occurred. Therefore, new-born screening and early intervention will be of paramount importance to facilitate successful treatment of LSD patients going forward.
There are many active gene therapy trials but unfortunately several good programmes that are being terminated by their funders. Often this is due to mixed data outcomes that better trial design, more advanced outcome measures or patient stratification could have addressed. It is important to remember that gene therapy improvements tend to be iterative with improving vectors, improving immune modulation programmes and improving methods of targeting specific organs including the brain. We should not forget that improved trial design and a capacity to play the long game are both important factors for success in developing treatments for neurological LSDs.

AUTHOR CONTRIBUTIONS Stuart Ellison and Brian Bigger wrote the review. Helen
Parker made the figures.

CONFLICT OF INTEREST STATEMENT
As this is a review there is no associated data in a data repository.

DATA AVAILABILITY STATEMENT
My manuscript has associated data in a data repository, including all data for which data deposition is mandatory.

INFORMED CONSENT AND ANIMAL RIGHTS
This article does not contain any studies with human or animal subjects performed by the any of the authors.