Neurological glycogen storage diseases and emerging therapeutics

Abstract

Glycogen storage diseases (GSDs) comprise a group of inherited metabolic disorders characterized by defects in glycogen metabolism, leading to abnormal glycogen accumulation in multiple tissues, most notably affecting the liver, skeletal muscle, and heart. Recent findings have uncovered the importance of glycogen metabolism in the brain, sustaining a myriad of physiological functions and linking its perturbation to central nervous system (CNS) pathology. This link resulted in classification of neurological-GSDs (n-GSDs), a group of diseases with shared deficits in neurological glycogen metabolism. The n-GSD patients exhibit a spectrum of clinical presentations with common etiology while requiring tailored therapeutic approaches from the traditional GSDs. Recent research has elucidated the genetic and biochemical mechanisms and pathophysiological basis underlying different n-GSDs. Further, the last decade has witnessed some promising developments in novel therapeutic approaches, including enzyme replacement therapy (ERT), substrate reduction therapy (SRT), small molecule drugs, and gene therapy targeting key aspects of glycogen metabolism in specific n-GSDs. This preclinical progress has generated noticeable success in potentially modifying disease course and improving clinical outcomes in patients. Herein, we provide an overview of current perspectives on n-GSDs, emphasizing recent advances in understanding their molecular basis, therapeutic developments, underscore key challenges and the need to deepen our understanding of n-GSDs pathogenesis to develop better therapeutic strategies that could offer improved treatment and sustainable benefits to the patients.

Glycogen architecture and metabolism

Glycogen is the primary carbohydrate storage macromolecule in heterotrophic organisms ranging from bacteria to animals [1,2]. A single glycogen particle is composed of up to ∼55,000 glucose monomers that form α-1,4-linked linear chains with periodic α-1,6-linked branches (Fig. 1a). Branching maintains glycogen solubility and structures the particle in consecutive concentric tiers that enable rapid glycogen synthesis (glycogenesis) and degradation (glycogenolysis) as per metabolic needs (Fig. 1b) [1,3]. The dynamic processes of glycogen synthesis and utilization have been most extensively studied in liver and muscle, due to the central role of glycogen in metabolism and glucose homeostasis in these organs [1,[4], [5], [6]]. However, glycogen is synthesized in most organs, where its function is also critical for cellular and organ homeostasis [1,4,6]. Glycogen levels and architectures vary among tissue types and there is increased glycogen heterogeneity in glycogen-centric diseases. Both glycogen synthesis and degradation are multi-step enzymatic processes impacted directly or indirectly by numerous other enzymes and transporters. Mutations in the genes encoding these proteins result in Glycogen Storage Diseases (GSDs) that are characterized by impaired glycogen synthesis or degradation leading to deficient, excess, or aberrant glycogen [[7], [8], [9]].

Fig. 1.

Architecture and metabolism of glycogen. (a) Chemical structure of glycogen. Glycogen is comprised of linear glucose chains linked together via α-1,4 glycosidic bonds and branched via α-1,6 bonds. Due to this structure, glycogen contains multiple non-reducing ends (free C4-residues) and a single reducing end (free C1-residue). (b) Model of glycogen architecture in concentric tiers. Each chain contains an average of two branch points that form a structure of up to 12 concentric layers. Chains of different layers are color-coded for better visualization. (c) Schematic of brain glycogen synthesis and cytosolic degradation. Glycogen elongation is mediated by glycogen synthase (GYS1) that transfers the glucosyl moiety of UDP-glucose to non-reducing ends. Branching enzyme (GBE1) hydrolyzes one α-1,4 glycosidic bond and transfers the released chain to the same or an adjacent chain by forming α-1,6 bond. GBE1-transferred glucose units are highlighted in dark red. In the brain, the glycogen particle is degraded in the cytosol by the coordinated action of brain-type and muscle-type glycogen phosphorylases (PYGB an PYGM, respectively) and the glycogen debranching enzyme (GDE/AGL). PYGB and PYGM phosphorolyze terminal glucose residues to release glucose-1-phosphate (G1P). PYGB and PYGM only bind glucose chains longer than four glucose units, and GDE/AGL removes the shorter glucose chain branch points. GDE/AGL transfers the branch to the same or an adjacent linear chain, and one glucose is released during the process. PYGB-mediated phosphorolysis then resumes. (d) Schematic of lysosomal glycogen degradation. Glycogen is targeted to the lysosomal lumen via processes that are not fully resolved yet but likely involve both macroautophagy and glycogen-specific autophagy (called glycophagy; see Ref. [27] for review). The acid, α-glucosidase (GAA) is decorated by mannose-6-phosphate containing N-glycans that are recognized by the cation-independent mannose-6-phosphate receptor (CI-MPR). This interaction mediates GAA entry into lysosomal lumen. In the lysosomal acidic environment, GAA hydrolyzes glycogen into glucose that exits lysosomes via any of these potential transporters including GLUT1, GLUT2, GLUT3, GLUT8, SPNS1 and SLC37A2 [25,26]. Abbreviations: Glc: d-glucose; G1P: α-d-glucose-1-phosphate; Pi: orthophosphate; UDP: uridine-5′-diphosphate.

Glycogen synthesis is primarily driven by two enzymatic activities (Fig. 1c) [4]. First, glycogen synthase (GYS1 in muscle and throughout the body, GYS2 in the liver) transfers glucose from UDP-glucose to the non-reducing ends of elongating chains thereby forming the linear α-1,4 chains [10]. Mutations in either GYS2 or GYS1 cause GSD type 0a and 0b (Lewis disease), respectively, characterized by the inability to synthesize glycogen in the affected tissues [11,12]. A second enzyme, glycogen branching enzyme (GBE1) transfers a short linear chain from the growing glycogen particle to produce α-1,6-linked branches [4,6]. The ratio between these two activities is essential in defining glycogen architecture. Increased or decreased rates in either enzyme leads to perturbed glycogen architecture and disease. When glucose chains become too long, they form double helices and adopt a crystalline organization [13]. These aberrant glycogen particles are insoluble in nature, resistant to degradation, and are referred to as PolyGlucosan Bodies (PGBs) [13]. Mutations in GBE1 that cause severe loss of glycogen branching enzyme protein or enzymatic activity result in severe infantile-onset GSD IV (Andersen disease) [14], while mutations that only mildly affect enzyme activity result in adult-onset GSD IV known as Adult Polyglucosan Body Disease (APBD) [15,16]. Additionally, a third activity mediated by glycogenin is typically considered to prime glycogen synthesis and seed the growing particle [17]. Glycogenin catalyzes its self-glucosylation on a specific tyrosine residue to form a nascent glucose chain that is further elongated by glycogen synthase [18]. However, this role of glycogenin was recently challenged by a study showing that glycogenin knocked-out mice accumulate increased levels of glycogen in the heart and skeletal muscle [19]. Similarly, GSD XV patients who are devoid of glycogenin-1 (GYG1) expression or express a defective variant incapable of binding to glycogen synthase accumulate PGBs in skeletal muscles [20]. Thus, glycogenin seems dispensable for glycogenesis and its exact function requires further investigations.

Glycogen catabolism relies on two distinct and spatially separated pathways, one in the cytoplasm and the other in lysosomes (Fig. 1c & d). Cytosolic glycogen degradation involves glycogen phosphorylase (brain isoform: PYGB; muscle isoform: PYGM; liver isoform: PYGL) and glycogen debranching enzyme (GDE/AGL). Glycogen phosphorylase cleaves α-1,4 glycosidic bonds on non-reducing ends and releases glucose-1-phosphate (G1P). When four glucose units remain, GDE/AGL debranches the α-1,6-linkage (Fig. 1c). Mutations in PYGM, PYGL, and GDE/AGL are responsible for GSD V (McArdle disease), GSD VI (Hers disease), and GSD III (Cori disease), respectively [[21], [22], [23], [24]].

Lysosomal glycogen degradation is catalyzed by acid, α-glucosidase (GAA). In the acidic environment of the lysosome, GAA hydrolyzes both the α-1,4 and α-1,6 glycosidic bonds to release glucose that is then transported to the cytosol via multiple transporters [25,26] (Fig. 1d). Recently, a glycogen-selective autophagy process, termed glycophagy, has been proposed that may overlap with the lysosomal pathway or be distinct [27]. Mutations in GAA (GSD II) result in either a severe infantile onset Pompe disease (IOPD) or a milder late-onset Pompe disease (LOPD).

An interesting and under-appreciated aspect of glycogen resides in its phosphorylation. During the last two decades, the mechanistic understanding of glycogen phosphorylation and its impact on glycogen structure and metabolism started to unfold, particularly in the context of Lafora Disease (LD) [28]. LD is a GSD that manifests primarily with central nervous system (CNS) dysfunctions caused by mutation in either EPM2A or EPM2B/NHLRC1, that encode the glycogen phosphatase laforin and the E3 ubiquitin ligase malin, respectively [29]. LD is driven by the accumulation of hyperphosphorylated and aberrantly branched PGBs known as Lafora Bodies (LBs), resulting in epilepsy, childhood dementia, neurodegeneration, and is inevitably fatal [29,30].

Liver and muscle are primary glycogen storage organs and thus are commonly impacted by defects in glycogen metabolism. However, recent work has demonstrated that the CNS is highly sensitive to perturbations in glycogen metabolism due to its complex metabolic demands. These findings led to the recent designation of the neurological-GSDs (n-GSDs) [31]. The n-GSDs include GSDs with primarily CNS manifestation and significant peripheral impact as well as those with primarily peripheral manifestation and significant CNS impact. Recent observations in Pompe disease have been important in this classification. Pompe disease patients exhibit primary manifestations in peripheral organs, yet patients also exhibit significant CNS involvement which is clinically apparent in individuals undergoing peripheral enzyme replacement therapy (ERT) [[32], [33], [34]]. Together, the n-GSDs represent a significant category of neurological disorders with unmet clinical need [31].

This review focuses on the current understanding of CNS glycogen metabolism, how it is affected in n-GSDs, and summarizes the therapeutic strategies that have been or are currently being tested. Finally, still unanswered questions of CNS glycogen metabolism and the challenges to produce effective treatments are highlighted.

Unique aspects of brain glycogen

The brain is an energy-demanding organ with very high glucose consumption. In humans, the brain accounts for only 2% of the total body weight while consuming approximately 20% of body's total energy, largely to support neuronal functions [[35], [36], [37], [38], [39]]. Of note, brain energy demand in humans is dramatically higher during development and represents 60% and 50% of the total body basal metabolic rate in neonates and children, respectively [40]. Acute reduction in brain glucose results in rapid impairment of cognitive function, brain damage, and can be lethal [41,42]. Thus, maintenance of constant glucose supply to brain is critical to support its functions. While direct glucose transport from circulation into the brain has been long appreciated, a comprehensive understanding of the role of brain glycogen in this process is lacking.

Findings from hypoglycemia or hypoxia provided early appreciations that CNS glycogen can serve as a local emergency glucose reservoir to protect both astrocytes and neurons during stressful conditions [[43], [44], [45], [46]]. Indeed, in hypoglycemic conditions, glucose utilization via glycolysis or pentose phosphate pathway exceeds glucose uptake mediated by GLUT1 and GLUT3 transporters, inducing glycogenolysis to compensate for the diminished glucose availability [43,45,47]. Similarly, hypoxia leads to increased glycogen utilization to compensate the lack of oxidative phosphorylation and support brain energy demand [48].

Beyond its role as an emergency reserve, it has become increasingly clear that brain glycogen has an integral role in normal brain physiology. Glycogen-derived glucose-6-phosphate supports homeostasis of glutamate and GABA (γ-aminobutyric acid), two major neurotransmitters in the brain [49,50]. Also, astrocytic clearance of extracellular K⁺ by Na⁺/K⁺-ATPases, an essential process to avoid hyperexcitability and seizures, is energy-demanding and relies on glycogen-derived glycolysis and oxidative phosphorylation [51,52]. Additionally, glycogen is an essential participant in the metabolic interplay between astrocytes and neurons [53]. Metabolic shuffling between astrocytes and neurons is complex, with the highly controversial astrocyte-neuron lactate shuttle (ANLS) hypothesis proposing that astrocytic uptake of neuron-released glutamate triggers glycogenolysis and glycolysis in astrocytes. The glycogen-derived lactate is then proposed to shuttle from astrocytes to neurons to fuel neuronal energy demand [[54], [55], [56]]. However, the ANLS hypothesis fails to account for the experimental stoichiometry of brain energetics [57]. Especially, there are accumulating data that neurons increase their utilization of glucose upon stimulation, rather than metabolizing lactate [58]. Accordingly, the glucose sparing by glycogenolysis (GSG) model has recently proposed that astrocytes rely on glycogenolysis to spare free glucose during neuronal activity, increasing glucose availability for neuronal uptake to fulfill neuronal high energy demand [59].

Recent spatial studies have revealed key aspects of brain glycogen localization and function. Brain glycogen is distributed in a highly spatially heterogeneous manner throughout the brain. This spatial heterogeneity is likely attributed to distinct functional and energetic demands in specific brain regions and cell types. In humans, glycogen is enriched in the gray matter of the cortex, as well as granular layers of hippocampus and cerebellum [60,61]. While astrocytes are responsible for the majority of glycogen stores, neurons and microglia, once thought to be devoid of glycogen, take part in active glycogen metabolism and even exhibit higher flux [[62], [63], [64], [65]]. Beyond the brain, an unexpected new role of glycogen in pain signaling in the spinal cord was recently demonstrated [66,67]. Even at the sub-cellular level, glycogen is differentially localized and utilized. For example, astrocytic glycogen is not uniformly distributed within the cell, but it is highly accumulated in the perivascular endfeet and in the processes interacting with presynaptic terminals [68,69].

Key mechanistic insights building on these spatial studies demonstrated that brain glycogen is also chemically heterogeneous, and this complexity is critical for appropriate N-linked protein glycosylation and brain function. Strikingly, brain glycogen is comprised of glucose and approximately 25% glucosamine, while skeletal muscle and liver glycogen are comprised of only 1% and 0.1% glucosamine, respectively [70]. In the brain, this glucosamine pool is essential for proper N-linked protein glycosylation that modifies cell surface, secreted, and circulating proteins [71]. N-linked glycans participate in a diverse array of cellular processes critical for protein structure and function, including protein maturation, stability, enzymatic activity, subcellular localization, and protein-protein interactions [71]. Also, transporters on the plasma membranes are highly decorated with N-linked glycans so that perturbed N-linked glycosylation could directly impact neurotransmitter function. This work established a direct connection between brain glycogen metabolism and N-linked glycosylation, thus linking n-GSDs with congenital disorders of glycosylation (CDGs).

Finally, it is increasingly appreciated that the pleiotropic roles of brain glycogen are intimately related to higher order cognition. Early evidence regarding the involvement of brain glycogen in memory consolidation was obtained in chickens that received intracerebral administration of a glycogen phosphorylase inhibitor [72]. Disruption of glycogenolysis led to a dose-dependent inhibition of both learning and memory consolidation. Later, generation of mouse models lacking glycogen synthase 1 (Gys1) in the brain or specifically in Camk2a-expression forebrain pyramidal neurons led to similar deleterious outcomes in learning and memory [73,74]. Strikingly, a recent study analyzing the UK Biobank data discovered that glycogen metabolism is a critical pathway impacting cognitive performance in humans throughout the course of life [75].

Neurological glycogen storage diseases

Key insights into the function of CNS glycogen has emerged from the disruptions observed in n-GSD patients, particularly those with LD, APBD and Pompe disease [6,29,31,76,77]. In addition, recent studies suggest a role of CNS glycogen in glycogen storage disease type 0b (GSD 0b), Cori disease (GSD III), polyglucosan body myopathy type 1 (PGBM1) and Glut1 deficiency syndrome (Glut1-DS) that lead us to define these diseases as emerging n-GSDs [12,[78], [79], [80]]. Similarities and unique aspects of the pathophysiology of different n-GSDs are providing key insights to decipher the function of glycogen in the CNS while suggesting potential therapeutic strategies.

Lafora disease

LD (OMIM #254780) is one of the first n-GSDs described. LD is a fatal, ultra-rare childhood dementia and progressive myoclonic epilepsy with an onset during adolescence in apparently healthy teens. LD is an autosomal recessive disorder caused by mutations in either the EPM2A or EPM2B/NHLRC1 genes leading to impaired function of the glycogen-phosphatase laforin or the E3 ubiquitin ligase malin, respectively (Fig. 2) [29,[81], [82], [83]]. The majority of patients exhibit classical LD, and there are less deleterious mutations with patients presenting a slower progressing clinical course [84,85].

Fig. 2.

Molecular aspects of three neurological glycogen storage diseases (n-GSDs). In unaffected individuals, brain glycogen is synthesized by GYS1 and GBE1 and catabolized by either GAA in the lysosome or the combination of PYGB, PYGM and GDE/AGL in the cytosol. Deficient GBE1 activity causes Adult Polyglucosan Body Disease (APBD) that is characterized by accumulation of polyglucosan bodies (PGBs). PGBs differ from glycogen by their larger size and possess longer glucose chains that form helical structures rendering PGBs insoluble. Deficiency of the glycogen phosphatase laforin or the E3 ubiquitin ligase malin are responsible for Lafora Disease (LD). Laforin dephosphorylates glycogen particles, which contains low levels of covalently attached phosphate in unaffected individuals (P in full red circle represents phosphorylation). Laforin interacts with malin, which ubiquitinates several glycogen-related enzymes, including laforin. LD is characterized by the accumulation of highly phosphorylated PGBs. Disruption of lysosomal glycogen degradation by GAA deficiency in Pompe Disease leads to a large accumulation of intra-vesicular glycogen particles that largely exhibit normal architecture. Recent report also suggests that higher levels of cytoplasmic glycogen may also be present in Pompe Disease [120].

LD is characterized by the accumulation of cytosolic, hyperphosphorylated PGBs, referred to as LBs, that are observed in many cell types from multiple organs including brain, skeletal muscles, cardiac muscle, and skin [13]. In the brain, PGBs accumulate in neurons, astrocytes, and microglia [[86], [87], [88]]. Interestingly, PGB structure varies between neurons and glial cells with neurons exhibiting large PGBs and glial cells containing numerous small PGBs [89]. Multiple laboratories generated Epm2a^−/− and Epm2b^−/− LD mouse models that recapitulate PGB accumulation in brain and other organs [[90], [91], [92], [93]]. PGBs are accompanied by neurophysiological alterations in hippocampal synaptic activity, increased susceptibility to kainate-induced epilepsy, and progressive neuronal cell death [[94], [95], [96], [97], [98], [99], [100]]. Data from LD mouse models suggest that neuronal PGBs underlie the LD epileptic phenotype while astrocytic PGBs cause neurodegeneration [97,98]. By crossing these LD mouse models with genetic models that decrease glycogen synthesis, several groups independently and conclusively demonstrated that PGBs drive LD aberrant electrophysiological properties, kainate-induced epilepsy, neuronal cell death, and inflammation [94,95,99].

LD patients develop tonic-clonic seizures, ataxia, dysarthria, visual hallucinations, and exhibit a decline in their cognitive functions that is associated by rapid dementia and eventually a vegetative state prior to death [101]. As frequency of epileptic episodes increases, patients become resistant to antiepileptic drugs, and death typically occurs within 10 years from onset [29,[101], [102], [103]]. There is currently no cure or disease modifying therapy for LD.

Adult polyglucosan body disease

APBD (OMIM #263570) is another rare, autosomal recessive n-GSD. It is one of the several pathological subcategories of GSD IV, caused by deficiency in the glycogen branching enzyme GBE1, and is also known as adult-onset GSD IV (Fig. 2). APBD patients exhibit PGB accumulation in multiple tissues including the CNS, peripheral nervous system, heart, muscles and liver [76,104,105]. APBD is characterized by progressive spastic paraparesis, sensory deficits in the distal limbs, neurogenic bladder, gait impairment and subsequently loss of ambulation, and premature death due to complications of myelopathy and peripheral neuropathy, with an age of onset ≥40 years [15,16,106]. Leukodystrophy is also a common symptom of APBD, associated with medullary and spinal atrophy that can be easily identified by magnetic resonance imagery [[106], [107], [108]]. Approximately 50% of APBD patients suffer from cognitive decline, notably memory deficit, and more rarely present atypical clinical manifestations including Alzheimer-like dementia and stroke-like episodes [76,109].

Similar to LD, PGBs drive APBD disease physiopathology and accumulate both in astrocytes and neurons [[110], [111], [112]]. Recent pre-clinical therapeutic studies in APBD mouse models, discussed in section 4.6, have demonstrated that suppression of PGBs accumulation is a viable therapeutic strategy [[113], [114], [115]].

Pompe disease

Pompe disease (OMIM #232300), caused by deficiency in the acid α-glucosidase GAA, is the most common GSD, with a recent newborn screening study that determined a prevalence of 1:18,711 [8,116] (Fig. 2). Pompe disease is both a GSD and the first identified lysosomal storage diseases (LSDs), which now encompasses more than 50 disorders [[117], [118], [119]]. Accumulation of glycogen in Pompe disease mainly occurs in lysosomes with some cytoplasmic glycogen accumulation [120]. Presence of insoluble cytoplasmic PGBs, characteristic to LD and APBD, is rare in Pompe disease [121]. IOPD is the more severe form and manifests in utero, at birth, or during the first months of life, with a median onset at 2 months of age [122,123]. It is characterized by hypertrophic cardiomyopathy, muscle weakness, failure to thrive and respiratory distress that leads to death from heart or respiratory failure within one year from onset [124,125]. LOPD is generally a milder form that manifests during childhood, adolescence or adulthood [126,127]. Symptoms associated with LOPD usually involve progressive muscle weakness, notably degeneration of the diaphragm, which induces several respiratory disorders including sleep apnea, dyspnea, and ultimately respiratory failure [[128], [129], [130], [131], [132], [133], [134], [135]]. CNS involvement in the disease was recognized since the 1960s, though this contribution has only recently been appreciated [[136], [137], [138], [139]]. IOPD patients exhibit extensive neuronal pathophysiology, including massive glycogen accumulation in the CNS, neuronal ballooning, electroencephalographic abnormalities, impaired breathing, white matter abnormalities and decline in cognitive functions [136,[138], [139], [140], [141], [142]]. Studies in animal models indicate the respiratory motor neurons may be particularly susceptible to glycogen accumulation and histopathology in Pompe disease [143]. LOPD patients accumulate large amounts of glycogen in the CNS and present neurological symptoms that include hearing loss, leukoencephalopathy, cerebral aneurysms, and executive functions impairment [77,[144], [145], [146], [147]]. Assessment of the diaphragm indicates prominent neurologic impairment in a subset of LOPD patients [129,148].

The first targeted treatment for Pompe disease reached the market in 2006 with the approval of human recombinant GAA (alglucosidase alfa) as an ERT that increased the life expectancy of IOPD patients [149]. As patients aged, and since peripherally delivered alglucosidase alfa cannot degrade CNS glycogen, these patients started to present neurologic symptoms including sensorineural hearing loss, white matter alterations, and cognitive impairment that were previously not observed because of the early fatality [[32], [33], [34],[150], [151], [152], [153], [154]]. The Gaa-deficient rodent model of the disease demonstrates substantial glycogen accumulation and histopathology in respiratory motoneurons including diaphragm (phrenic) and tongue (hypoglossal) motoneurons [77,140,143,155,156]. We have advanced the hypothesis that such respiratory motor neurons ultimately contribute to reduced neuronal output and impaired breathing [141]. This emphasizes the role of the nervous system in respiratory manifestations of Pompe disease [148]. Numerous studies also reported that seizures are part of the clinical manifestations of both IOPD and LOPD, which suggests seizures may be a common symptom shared by diseases involving brain glycogen perturbations [33,[157], [158], [159], [160]].

Emerging n-GSDs

Glycogen storage disease type 0b (GSD 0b, OMIM #611556) is an ultra-rare disease caused by deficiency in GYS1, the glycogen synthase isoform mainly expressed in skeletal muscles, heart, and brain [12,161]. GSD 0b is characterized by the absence of glycogen in muscle fibers. Early studies reported a childhood onset associated with exercise intolerance, tonic-clonic seizures and sudden death by cardiac arrest [12,162,163]. One patient also presented myalgia and loss of consciousness [163]. A recent study reported two cases with adult onset (>40 years) presenting milder phenotypes that were limited to skeletal muscle manifestations [161]. Despite the lack of direct evidence, the neurological manifestations are presumably impacted by low brain glycogen levels. Due to its extremely low prevalence, there is currently no known therapy in development for GSD 0b.

Glycogen storage disease type III (GSD III, Cori disease, OMIM #232400) is an autosomal recessive disorder caused by mutations in AGL/GDE gene leading to deficiency of glycogen debranching enzyme involved in glycogen breakdown [164]. GSD III patients exhibit increased glycogen accumulation primarily in the liver, skeletal, and cardiac muscles that clinically manifest as hepatomegaly (enlarged liver), hypoglycemia (low blood sugar), and hyperlipidemia. Additionally, patients may experience growth retardation, delayed puberty, and in severe cases, cardiomyopathy, or liver cirrhosis [164]. To date, therapy for GSD III primarily focuses on dietary management, aiming to maintain stable blood glucose levels and prevent hypoglycemia [165]. In addition to reports of CNS involvement in GSD III patients [78], recent studies using GSD III mouse models demonstrate increased CNS glycogen accumulation [70]. Thus, further studies are required to comprehensively grasp the mechanisms of CNS involvement in this disease.

RBCK1-associated polyglucosan body myopathy type 1 (PGBM1, OMIM #615895) is an ultra-rare fatal GSD caused by mutations in the RBCK1 gene encoding the E3 ubiquitin ligase RBCK1 (also known as HOIL1). RBCK1 is a component of the linear ubiquitin chain assembly complex (LUBAC) reportedly involved in ubiquitination of multiple targets, including carbohydrates [166]. Monoubiquitylation of unbranched glucosaccharides may act as a quality control mechanism to remove poorly branched glycogen from cells, thus preventing PGBs accumulation [166]. Deficiency of RBCK1 leads to progressive accumulation of PGBs in several tissues including skeletal muscle, cardiac muscle, and brain both in PGBM1 patients and transgenic mice models [[167], [168], [169]]. The patients primarily present myopathy, cardiomyopathy, autoinflammation, and immunodeficiency as predominant clinical phenotypes. While neurological involvement, including cognitive impairment, dementia, or psychiatric disturbances, has also been reported [79], further research is needed to understand its prevalence and establish mechanisms linking neurological aberration with PGBs accumulation in this disease.

Glucose transporter type 1 deficiency syndrome (Glut1-DS, OMIM #606777) is a rare autosomal dominant neurometabolic disorder caused by haploinsufficiency of SLC2A1 encoding GLUT1 [80]. Mutations in the SLC2A1 result in reduced glucose flux in the brain affecting brain functions resulting in clinical phenotypes including but not limited to developmental encephalopathy with medication-refractory infantile-onset seizures affecting the cognitive, behavioral, and motor functions [80]. To our knowledge, only one study investigated brain glycogen accumulation in a Glut1-DS mouse model and reported a >50% decrease in the Glut1-DS mice compared to wild-type controls [170]. Currently, an effective treatment for Glut1-DS remains elusive.

Targeting CNS glycogen in n-GSDs: Current challenges and future advances

Current challenges in the treatment of n-GSDs

Current strategies for developing n-GSD therapeutics focus on either clearing existing glycogen or PGBs using enzyme-based therapies, such as ERT, or preventing glycogen or PGB formation via Substrate Reduction Therapy (SRT) (Fig. 3). Due to their relative rarity, there is a strong desire to develop generalizable strategies that collectively treat n-GSDs. Developing effective therapies for n-GSDs is challenging due to the unique properties of PGBs. Additionally, depending on the n-GSD, glycogen particles and PGBs are observed in different organs, cell types and subcellular compartments and locations that still need to be precisely determined. Moreover, the blood-brain barrier (BBB) restricts the entry of many substances into the CNS, including most therapeutic agents [171,172]. Current approaches to CNS therapy overcoming these barriers are rapidly developing, both at foundational and technical levels, and so this is an exciting time for the development of n-GSD therapeutics [172]. Foundational advances include a growing understanding of disease mechanisms, the intricacies of the CNS functions, BBB architecture and dynamics, identifying specific molecular targets involved in CNS diseases, and development of small molecules that can modulate disease targets. Technical advances include next-generation carrier platforms that either cross or bypass the BBB and other advanced drug delivery systems that deliver drugs directly into the CNS.

Fig. 3.

Representation of the therapeutic approaches to treat n-GSDs. In unaffected individuals, there is a balance of glycogen synthesis and degradation that maintains low levels of brain glycogen. In n-GSDs that accumulate high levels of glycogen or PGBs, glycogen degradation is decreased and glycogen levels and/or aberrant PGBs accumulate. Several therapeutic approaches aim at reducing aberrant glycogen. Substrate reduction therapy (SRT) reduces glycogen synthesis via glycogen synthase (GYS1) inhibition. Gene therapy and enzyme replacement therapy (ERT) both aim at replacing the deficient enzyme with an unaffected enzyme to normalize glycogen levels. Additionally, antibody-enzyme fusions (AEF) directly degrade glycogen using a different enzymatic activity than the one impacted by the disease.

Enzyme replacement therapy (ERT)

The most established strategy for GSD therapeutics is clearing glycogen or PGBs via ERT (Fig. 3). In ERT, the missing or defective enzyme in a disease is administered to the patient, typically through intravenous injections. For Pompe disease, recombinant human GAA (rhGAA, Alglucosidase alfa/Myozyme®/Lumizyme®), developed by Sanofi Genzyme, was approved as an ERT therapy in 2006 [149,173]. Long-term ERT with rhGAA improves the natural course of Pompe disease leading to improved cardiac, respiratory, and motor functions, but with variable efficacy in correcting skeletal and smooth muscle dysfunction. Over the last two decades, extensive studies on the efficacy of rhGAA in preclinical models and human patients yielded an understanding of the challenges that hinder its effectiveness targeting elevated glycogen levels in skeletal, cardiac, and smooth muscle. These challenges include the instability of rhGAA at physiological pH, its rapid clearance by non-muscle tissues (liver, spleen), development of neutralizing antibodies, and poor bioavailability. Additionally, it is important to realize that <1% of total intravenously injected rhGAA reaches muscles, due to low abundance of cation-independent mannose 6-phosphate receptor (CI-MPR) on target cells and/or low bis-phosphorylated oligosaccharides on the rhGAA [174,175]. Strikingly, some Pompe disease patients treated long term with the ERT exhibit neurologic phenotypes despite an early treatment regime [33,176]. These CNS clinical signs are thought to be revealed upon peripheral treatment since rhGAA does not cross the BBB and therefore cannot decrease CNS glycogen levels.

A next generation rhGAA; Avalglucosidase alfa (NEXVIAZYME™; avalglucosidase alfa-ngpt) received US Food and Drug Administration (FDA) approval in 2021 [[177], [178], [179]]. Avalglucosidase alfa is rhGAA conjugated with multiple synthetic bis-mannose-6-phosphate (bis-M6P)-tetra-mannose glycans to enhance its cellular uptake by CI-MPR in skeletal muscle. In a preclinical study, this modified enzyme displayed improved glycogen clearance (approximately fivefold over alglucosidase alfa) in cardiac and skeletal muscles of Pompe disease mouse model [180]. In a randomized double-blind phase 3 trial, avalglucosidase treated Pompe disease patients also exhibited statistically significant improvement in some measures including respiratory function and muscle strength compared to existing standard of care, alglucosidase alfa [178]. However, Pompe disease mice treated with avalglucosidase alfa did not provide any additional benefit at the CNS level as the drug failed to cross BBB and reduce CNS glycogen [180].

In 2023, Amicus Therapeutics received approval in the European Union for use of a combination therapy Cipaglucosidase alfa (Pombiliti™). This ERT utilizes cipaglucosidase alfa that has higher M6P content than standard alglucosidase alfa and is provided in combination with a small-molecule pharmacological chaperone miglustat (N-butyl-deoxynojirimycin [NB-DNJ], Opfolda™) for the treatment of LOPD. Cipaglucosidase alfa displays improved enzyme stability, provided by miglustat, and the pharmacological properties are improved compared to alglucosidase alfa, leading to better uptake and improved muscle pathology in Pompe disease mice [181,182]. To date, Cipaglucosidase alfa has not been reported to cross the BBB or the decrease glycogen in the CNS.

ERTs have been effectively utilized to treat multiple LSDs due to the intrinsic ability of many cell types to take up lysosomal enzymes from circulation into lysosomes via the CI-MPR [183,184]. However, ERTs require regular, lifelong infusions to maintain therapeutic enzyme levels. Additionally, most n-GSDs do not exhibit lysosomal glycogen accumulation. Indeed, Pompe disease can be targeted with ERT, due to it being an LSD, allowing M6P-mediated uptake into lysosomes, but this is not the case for other n-GSDs that exhibit aberrant cytoplasmic glycogen accumulation [185].

Antibody-enzyme fusions (AEFs)

Since most n-GSDs exhibit cytoplasmic PGBs, there is strong motivation to develop methods to deliver PGB-active enzymes to the cytoplasm of affected cells. Antibodies offer an intriguing delivery platform that can be engineered as fusions with enzymes to deliver them to specific tissues, organs, cell types, or sub-cellular locations [186]. Antibodies have emerged at the forefront of pharmaceutical research as targeting molecules and/or cell penetrating modalities. Antibody-directed therapies as antibody-drug conjugates, immune modulators, and antibody-directed enzyme prodrugs are being successfully employed as therapies in hematology, rheumatology, and oncology. The approval of etanercept in 1998 for the treatment of rheumatoid arthritis led to the development of antibody-enzyme fusions (AEFs) that received approval for autoimmune and inflammatory diseases, cancers, neurological, and lysosomal diseases [186,187]. Recently, there has been significant pre-clinical progress to develop AEFs for n-GSDs [187,188].

One strategy to promote cytoplasmic delivery of PGB-active enzymes, involves fusion with the humanized antibody fragment of 3E10 [189]. 3E10 is derived from a murine lupus anti-DNA antibody that is non-pathogenic, penetrates cell membranes via the equilibrative nucleoside transporter 2 (ENT2, SLC29A2), and localizes to the nucleus and cytoplasm [[190], [191], [192], [193], [194]]. The 3E10 Fab can deliver payloads up to 155 ​kDa into the cytoplasm of ENT2-expressing cells. Parasail Therapeutics (formerly Valerion Therapeutics) developed VAL-1221 that is comprised of 3E10 Fab fused to rhGAA. Fusing 3E10 with rhGAA imparts the ability for VAL-1221 to enter cells via two mechanisms: the intrinsic mannose-6-phosphate receptor mediated rhGAA uptake and ENT2-mediated 3E10 uptake. Thus, VAL-1221 can potentially target glycogen pools in the cytoplasm and degrade aberrant lysosomal glycogen. Importantly, rhGAA was able to maintain some level of activity even at neutral pH [189]. Indeed, VAL-1221 was stable in Pompe disease mice upon intravenous (i.v.) delivery, trafficking to both cytoplasmic and lysosomal compartments, and lowering both cytoplasmic and lysosomal glycogen levels [189]. VAL-1221 successfully completed a phase 1/2 clinical trial (NCT02898753) in advanced Pompe disease patients where the primary endpoint was to test safety [195]. Current efforts are focused on efficacy studies.

Thus, AEFs combine engineered antibody-dependent functionality with highly specific enzymatic activity for cellular and sub-cellular delivery. This is an important step, but biologics for n-GSDs require CNS delivery, and no current ERT or AEF modalities cross the BBB.

Intracerebroventricular administration

The most immediate solution to CNS delivery is direct delivery. Intracerebroventricular (i.c.v.) administration of therapeutic agents bypasses the BBB, introducing the agent directly into the cerebrospinal fluid (CSF), and allowing broad biodistribution throughout the brain [[196], [197], [198]]. Distribution of drugs in the CNS is highly dependent on CSF movement and this dependency is even more critical for biologics. Defining CNS fluid movement is an active field of research with ongoing analyses of biological drug biodistribution via i.c.v. and intrathecal (i.t.) delivery [[199], [200], [201]]. Recently, FDA approved i.c.v.-based ERT therapy for a fatal CNS-centric LSD, ceroid lipofuscinosis type 2 (CLN2) [202]. This approval was based on promising canine data of repeated i.c.v. administration of cerliponase alfa (recombinant human tripeptidyl peptidase-1 (rhTPP1) leading to clearance of lysosomal storage material and delayed onset and slower progression of neurodegeneration [[203], [204], [205]].

Direct CNS delivery of a PGB-targeting AEF was recently reported in a pre-clinical setting. 3E10 was fused with pancreatic α-amylase for cytoplasmic PGB-targeting to generate the AEF designated VAL-0417. VAL-0417 robustly degraded PGBs in vitro [206]. When administered to LD mice by intramuscular (i.m.) or i.v. injections, VAL-0417 degraded peripheral PGBs in skeletal muscle or skeletal muscle and cardiac muscle, respectively, but failed to efficiently cross the BBB [206]. Continuous i.c.v. administration of VAL-0417 for 28 days into an LD mouse model resulted in its broad brain biodistribution deep into the brain parenchyma with robust intra-cellular localization [207]. Conversely, pancreatic α-amylase lacking the 3E10 Fab exhibited little to no intra-cellular staining in any brain regions, demonstrating the need for AEF utilization in the CNS [207]. Based on these promising results, an increased dose of VAL-0417 was continuously i.c.v. administrated for 7-days into LD mice. VAL-0417 treatment ablated PGBs in every region of the brain, including the cerebral cortex, thalamus, cerebellum, and brainstem, and these data were corroborated by biochemical analysis [206]. Strikingly, degradation of the PGBs rescued the N-linked hypo-glycosylation phenotype of LD mice demonstrating physiological normalization was possible even after accumulation of a significant PGB load making this a very promising strategy as a therapeutic [70].

In the search for a generalized n-GSD therapeutic, the efficacy of i.c.v. rhGAA (Alglucosidase alfa/Myozyme®) in clearing cytoplasmic PGBs in the brain of LD mouse models was recently reported. A single i.c.v. administration of increased doses of rhGAA failed to reduce PGB numbers in the hippocampus of 12-month-old LD mice. Similarly, a continuous i.c.v. administration of rhGAA for 2-weeks in 6-month-old LD mice and 4-weeks in 9-month-old LD mice also resulted in similar outcomes. Additionally, this continuous i.c.v. rhGAA administration also failed to improve behavior abnormalities including memory, anxiety, or sensitivity to PTZ in LD mice. These data establish that rhGAA does not degrade cytoplasmic PGBs in LD mouse models [208], emphasizing the significance of targeting compartment specific glycogen store in different n-GSDs.

Receptor-mediated transport

While i.c.v. delivery provides a direct administration route, it comes with obvious clinical concerns. There have been significant efforts in developing methods to deliver biologics to the CNS via BBB transcytosis, which would then allow i.v. administration. Receptor-mediated transport (RMT) across the BBB via receptors localized on the endothelial cells of brain capillaries provides access to the brain for large molecules [209]. These receptors include insulin-like growth factor (IGF) receptors (IGFRs, including CI-MPR, also known as IGF2R) and transferrin receptor (TfR) [[210], [211], [212], [213]].

Initial efforts focused on use of the IGF pathway for rhGAA delivery. Infusion via i.v. of the IGF2-rhGAA fusion reveglucosidase alfa (BMN 701) exhibited superior efficacy over rhGAA in clearing muscle glycogen in a murine Pompe disease model (GAAtm1Rabn/J) [214]. Strikingly, both the murine model and LOPD patients treated with reveglucosidase alfa exhibited enhanced pulmonary and respiratory functions [214,215], an indication of improved CNS related pathology. A phase III trial of reveglucosidase alfa was terminated due to patient safety against the side effects of hypoglycemia, potentially induced by the IGF2 moiety of reveglucosidase alfa (NCT01924845, ClinicalTrials.gov). Nevertheless, this work provides optimism that similar brain penetrating biologics may be helpful in clearing CNS PGBs.

The transferrin receptor is localized on brain capillary endothelial cells and has been targeted to transfer cargos across the BBB for multiple diseases without the metabolic side effects of the IGF pathway [212,213]. Denali Therapeutics successfully developed multiple pre-clinical therapies that achieved CNS target engagement with some assets moving into the clinic [216]. Additionally, Sanofi recently presented a novel fusion of an anti-human transferrin receptor (hTfR) antibody with hGAA (Fig. 4 and Table 1) [217]. They highlighted that a single i.v. dose (180 ​nmol/kg, equivalent to 20 ​mg/kg Alglucosidase alfa) of anti-hTfR-GAA in Pompe mice expressing hTfR reduced glycogen in the skeletal muscle and heart to an extent comparable to treatment with either Alglucosidase alfa or Avalglucosidase alfa. Strikingly, this single i.v. dose of anti-hTfR-GAA significantly reduced brain and spinal cord glycogen in these mice by nearly 70%. More impressively, i.v. infusions for 12-weeks dramatically reduced skeletal muscle glycogen, CNS glycogen, and neuroinflammation. The 12-week regimen also reduced urine glucose tetrasaccharide levels, a biomarker for somatic tissue glycogen, and CSF maltotetraose, which correlates with reduced brain and spinal cord glycogen. These proof-of-concept data suggest an exciting clinical course for anti-hTfR-GAA and may provide a framework for more generalizable n-GSD therapeutics.

Fig. 4.

Summary of the different approaches and therapeutics in development for n-GSDs. Schematic summarizing therapeutics listed in Table 1. Abbreviations: UDP-glucose: uridine-5′-diphosphate-glucose; Glc: glucose; RNAi: RNA interference; ASO: antisense oligonucleotide; shRNA: small hairpin RNA; amiRNA: artificial microRNA; AAV: Adeno-associated virus; GAA: α-glucosidase, acid; Gys1: Glycogen synthase 1; GBE1: glycogen branching enzyme; BBB: blood-brain barrier.

Table 1.

Pre-clinical therapeutics to treat CNS aspect of n-GSDs.

Drug class	Disease	Therapeutic molecules	Molecular Target	Mode of Administration	References
ERT	Pompe disease	anti-hTfR-hGAA	GAA	i.v.	[217]
AEF	LD	Val-0417	Cytoplasmic glycogen	i.c.v.	[206]
SRT	LD/APBD	Gys1-ASO	Gys1	i.c.v.	[222,223]
	LD/APBD	AAV9-SaCas9-Gys1	Gys1	i.c.v.	[224]
	LD/APBD	AAV-amiRNA	Gys1	i.c.v.	[114]
Small molecules	APBD	Guaiacol	Gys1	Drinking water	[115]
	APBD	GHF201	Lamp1	i.v. and subcutaneous	[248]
	LD	Metformin	Not defined	Drinking water	[250,251]
Gene therapy	Pompe disease	rAAV9-hGAA	GAA	Intrapleural injection	[156]
	Pompe disease	AAVrh10-hGAA or AAV9-hGAA	GAA	Intrathecal injection	[235]
	Pompe disease	AAV9/3-GAA	GAA	i.c.v.	[236]
	Pompe disease	AAV-PHP.B-hGAA	GAA	i.v.	[237]
	Pompe disease	AAV8-hGAA (sec)	GAA	i.v.	[[241], [242], [243]]
	Pompe disease	AAV9.CAG.BiP.vIGF2.hGAA	GAA	i.v.	[244]
	LD	rAAV2/9-hEPM2A	Laforin	i.c.v.	[245]
	APBD	AAV9-hGBE	Glycogen branching enzyme (GBE1)	i.v.	[247]

LD: Lafora disease; APBD: Adult Polyglucosan Body Disease; ERT: Enzyme replacement therapy; SRT: Substrate reduction therapy; AEF: Antibody-enzyme fusion; ASO: Antisense oligonucleotide; AAV: Adeno-associated virus; GAA: α-glucosidase, acid; Gys1: Glycogen synthase 1; LAMP1: Lysosomal-associated membrane protein 1; i.c.v.: Intracerebroventricular; i.v.: Intravenous.

Substrate reduction therapy

SRT is an alternative strategy where the biosynthesis rate of the disease-driving biomolecule is reduced to offset decreased catabolism, improving the synthesis and catabolism balance [218,219]. For n-GSDs, the SRT approach is to target GYS1, the enzyme responsible for glycogen synthesis in skeletal muscle, cardiac muscle, smooth muscle, and the CNS (Fig. 3). This strategy would spare hepatic glycogen synthesis that relies on GYS2 and therefore averts glucose related homeostatic disturbances that may occur upon targeting the liver specific enzyme.

In a proof-of-concept study using genetically modified mice, genetic depletion of Gys1 in Pompe disease mice (Gaa/Gys1-KO mice) resulted in reduced lysosomal glycogen and functional improvement in the heart and skeletal muscles of the mice [220]. Similarly, genetic reduction of Gys1 by ∼50% in LD and APBD mice resulted in reduced neurological PGBs and marked improvement in brain physiology [96,113,221]. Corroborating these data, genetic depletion of the glycogen synthase activator PTG (Protein Targeted to Glycogen) successfully eliminated brain PGBs in LD and APBD mice, leading to significant rescue and improvement of disease specific phenotypes in the murine models of both diseases [94,95,113]. Based on these important data from genetic models, multiple modalities for SRT are being developed including antisense oligonucleotide (ASOs), small interfering RNAs (siRNAs), short hairpin RNAs (shRNAs), and small molecules (Fig. 4 and Table 1).

Since current RNA-based therapeutics do not readily cross the BBB, the most common strategy to deliver them into the CNS is via i.t., i.c.v. administration, or viral based vectors. The i.c.v. administration of a Gys1 ASO in LD and APBD mouse models, developed in collaboration with Ionis Pharmaceuticals, displayed promising results in both models. In one-month old LD mice, a single i.c.v. administration of Gys1 ASO resulted in >50% reduction in Gys1 protein expression in the cortex, hippocampus, and spinal cord two-weeks after the injection [222]. In young LD mice, i.c.v. administration of the Gys1 ASO prevented PGB formation. Administration in older mice with pre-existing PGBs halted further PGB accumulation, reduced neuroinflammation and slowed disease progression in two LD mouse models [222,223]. This work demonstrated the in vivo efficacy and disease modifying ability of Gys1 ASO in reducing CNS glycogen and/or abnormal glycogen deposits and alleviating pathological outcomes in multiple preclinical n-GSDs models. In methods conceptually similar to AEFs, protein targeting modalities are also being developed for ASO delivery by Aro Biotherapeutics, among others (NCT06109948, ClinicalTrials.gov).

A recent study utilized AAV9 (adeno-associated virus 9) virus packaged with Staphylococcus aureus Cas9 and a guide RNA targeting Gys1 that was delivered into the brain of neonatal APBD and LD mouse models by i.c.v. injection [224]. The AAV9-Gys1 construct yielded decreases in both Gys1 mRNA and Gys1 protein in multiple brain regions, diminished glycogen and PGB accumulation, and amelioration of neuroinflammation [224]. Another AAV vector was reported to deliver an artificial microRNA (amiRNA) targeting Gys1 mRNA for RNA-interference. When delivered by a single i.c.v. injection to APBD and LD mice, the Gys1 AAV-amiRNA achieved 15% reduction of Gys1 mRNA and 50% reduction of PGBs across multiple brain regions in both APBD and LD mouse models [114]. This study provided proof-of-principal that a single dose AAV-based SRT therapy yields therapeutic benefits to multiple n-GSDs.

Lentiviral (LV) mediated short hairpin ribonucleic acids (shRNAs) knockdown of Gys1 in primary myoblasts derived from Pompe disease mice significantly reduced Gys1 levels and resulted in decreased cytoplasmic and lysosomal glycogen [225]. Further, a single intramuscular injection of recombinant AAV1 expressing shGys1 into the gastrocnemius muscle of newborn Pompe disease mice led to a significant reduction in glycogen accumulation, demonstrating further pre-clinical evidence of therapeutic efficacy of Gys1 targeting [225].

Until recently, there were no clinically relevant small molecule inhibitors of Gys1. Guaiacol was identified in an in vitro high-throughput screen of FDA approved compounds as reducing PGB and glycogen levels in APBD mouse embryonic fibroblasts and verified in APBD patient-derived skin fibroblasts [115]. Guaiacol administered in drinking water to APBD mice reduced PGB levels in the liver, heart, and peripheral nerves, but did not reduce PGB levels in the skeletal muscle or brain (Fig. 4 and Table 1) [115]. APBD mice administered guaiacol in drinking water every other day for 12-months exhibited improved grip strength and dramatically improved lifespan. However, the pharmacokinetics data showed negligible levels of guaiacol in the murine brain after oral gavage. Nevertheless, guaiacol is a potential therapeutic avenue for GSDs associated with peripheral neuropathy, like APBD. Conversely, treatment of n-GSDs with guaiacol require further studies via alternative administration routes (i.e. continuous i.c.v.) that could bypass the BBB.

More recently, Maze Therapeutics developed MZ-101 as a potent and highly selective in vitro and in vivo GYS11 inhibitor that, importantly, does not inhibit GYS2 [226]. MZ-101 treatment markedly reduced glycogen concentrations in fibroblasts from Pompe disease patients and unaffected controls as well as in mice. Administration of MZ-101 was well tolerated in a Pompe disease mouse model and 12-weeks of treatment reduced skeletal muscle glycogen to levels similar to ERT (Alglucosidase alfa/Lumizyme®). Strikingly, the combination of MZ-101 and ERT exhibited additive effects and normalized glycogen levels in the Pompe mice gastrocnemius, peripheral blood mononuclear cells, diaphragm, and heart as well as normalizing Glc4 in urine. Impressively, MZ-101 treatment substantially corrected abnormal transcriptional and metabolic profiles of the Pompe mice. MZ-101 was granted Orphan Drug Designation by the FDA and completed a Phase I clinical trial (NCT05249621). To date, no data have been presented regarding MZ-101 ability to cross the BBB.

Gene therapy

The goal for n-GSDs is not just effective therapeutics but to develop n-GSD cures. Gene therapy is actively under development as it holds the compelling potential of restoring long-lasting endogenous production of defective enzymes and a long-term therapeutic benefit. AAV and LV vectors have exhibited impressive promise in delivering stable and long-term protein expression in a wide range of human tissues including the CNS. Employing advanced design and quality control tools, recombinant AAVs (rAAVs) can be tailored for specific CNS diseases to achieve better CNS tropism, cell type specific targeting, and longer-term therapeutic effect. Two rAAV-based therapies recently received FDA approval, Luxturna for inherited retinal dystrophy and Zolegensma for spinal muscular atrophy [227,228]. Among the AAV1-13 serotype, AAV9, AAV rh.8, and AAV rh.10, display high efficiency to cross the BBB after i.v. and or other administration routes [[229], [230], [231], [232], [233]].

In a foundational gene therapy attempt targeting the CNS in Pompe disease, AAV5-GAA was injected into the spinal cords of the Pompe disease mice [234]. Four-weeks post-treatment, the mice exhibited improved spinal cord GAA activity and reduced glycogen level. Additionally, the treated mice displayed increased phrenic motor output and improved ventilation, suggesting that these phenotypes are related to spinal cord glycogen. This study provided strong evidence that correcting CNS glycogen improves peripheral dysfunctions in Pompe disease [234]. Another study demonstrated that a single intrapleural injection of rAAV9-hGAA resulted in improved cardiac and respiratory function via a direct effect and improved respiratory muscle function via retrograde transport of rAAV9 to the motor neuron/spinal cord, further implicating CNS in Pompe disease pathology (Fig. 4 and Table 1) [156]. More recently, a single intrathecal delivery of AAVrh10-hGAA or AAV9-hGAA (human GAA) in one month old Pompe disease mice sustainably improved neurologic, neuromuscular, and cardiac functions [235]. Surprisingly, the muscle glycogen levels were unaffected by this treatment suggesting that phenotypic improvements were directly related to restoration of CNS functions. Similarly, neonatal i.c.v. injection of AAV9/3-GAA, driving neuron-specific expression of GAA, in Pompe disease mice improved GAA activity and glycogen levels in the CNS and substantially improved motor coordination without improving muscle glycogen (Fig. 4 and Table 1) [236]. Moreover, a single i.v. injection of a recently developed AAV9 variant (AAV-PHP.B), exhibiting 40–60 times higher transduction efficiency in mouse brains compared to AAV9 [237], demonstrated nearly complete glycogen clearance in the brain, heart, and muscles of a Pompe disease mouse model (Fig. 4 and Table 1).

Hepatic expression of secreted GAA promotes immunologic tolerance, increases enzyme activity in peripheral organs, and is an attractive therapeutic strategy for multisystem disorders [[238], [239], [240], [241]]. In a Pompe disease mouse model, systemic delivery of AAV expressing secreted GAA (sec-hGAA) under tandem liver-neuron specific promoters alleviated glycogen accumulation in both skeletal muscle and spinal cord, leading to improved muscle strength and respiratory functions [242]. Likewise, expression of secreted GAA using AAV8 vectors (AAV8-GAA) in Pompe mice, both at early and advanced stages of disease, resulted in reduced glycogen levels in muscle, heart, and CNS, and improved cardiac, muscle and respiratory functions (Fig. 4 and Table 1) [241,243]. Further, a recent study illustrated that single i.v. administration of AAV9 expressing secreted IGF2-tagged hGAA in both young and old Pompe mice resulted in effective clearance of CNS glycogen (Fig. 4 and Table 1) [244]. Additional strong evidence supporting the superior enzyme bioavailability and therapeutic effectiveness of secreted AAV vectors compared to ERT [245] positions them as the preferred vector for CNS diseases including n-GSDs. Notably, a phase I clinical trial investigating secreted AAV2/8-based gene therapy for Pompe disease is currently in progress (NCT03533673).

Similar AAV-centric, proof-of-concept gene replacement therapy has been successful in LD and APBD mouse models. In LD mice lacking Epm2b, proof of concept experiments demonstrated that restoration of Epm2b expression using tamoxifen inducible Epm2b halted PGB accumulation in the brain and muscle and ameliorated neuroinflammation [246]. These data suggested that gene therapy could be a possible therapeutic approach to treat LD and stop disease progression. A subsequent study demonstrated that a single dose i.c.v. injection of rAAV2/9-human laforin in 3-month-old LD laforin-deficient mice reduced PGB loads, epileptic seizures, neuronal hyperexcitability, and ameliorated histopathological and molecular neurological alterations (Fig. 4 and Table 1) [245]. Likewise, a single i.v. injection of AAV9-human GBE in 14-day-old APBD mice led to a significant surge in GBE enzyme activity and concomitant glycogen reduction in skeletal muscle, heart, and brain at 3-months of age [247]. These experiments in preclinical animal models of n-GSDs provide important proof-of-concept evidence that gene therapy could successfully reverse aberrant glycogen storage in CNS and peripheral organs and ameliorate pathogenic features of the disease.

Alternative therapeutic targets and strategies

Recently, the FDA granted Orphan Drug Designation to the small molecule GHF201 following promising results in APBD mice. GHF201 (also known as 144DG11) demonstrated effectiveness in reducing PGB/glycogen levels in the brain and in peripheral tissues like the liver, heart, and peripheral nerves in an APBD mouse model (Table 1) [248]. Data suggested that the molecular target of GHF201 is the lysosomal membrane protein LAMP1 and that it increases autophagic flux, indicating potential increased lysosomal degradation of the PGBs. These findings suggested that GHF201 could be explored as a potential treatment for the n-GSDs. However, a more recent study from the same group, along with additional collaborators, demonstrated that GHF201 failed in reducing PGBs in an LD mouse model [249]. Effectiveness of GHF201 on one PGB-associated n-GSD mouse model and not another is perplexing and requires further investigation. Additionally, the efficacy of GHF201 in other n-GSDs warrants further exploration.

Metformin is a clinically safe pharmacological agent known for its pleiotropic biological effects, acting on multiple cellular targets through various mechanisms. In LD, metformin exhibits a moderate beneficial impact in LD mice by reducing LB accumulation and slowing neurological deterioration when administered in utero (Table 1) [250,251]. In 2016, metformin was designated by the European Medicines Agency as an orphan drug for the treatment of LD (EU/3/16/1803). While in utero data in mice were positive, this administration option in humans is not realistic.

In addition to identifying or designing new therapeutics, improved drug delivery strategies could also be tested to deliver therapeutics into the brain for treating n-GSD. Focused ultrasound (FUS) combined with microbubbles locally, transiently, and reversibly disrupts the BBB, facilitating the delivery of drugs into the brain [[252], [253], [254]]. This approach is being extensively studied in preclinical settings to enhance the bioavailability of multiple therapeutic agents in specific brain regions for treating neurological disorders [254,255]. Additionally, FUS has demonstrated a favorable safety and tolerance profile in several early-phase clinical trials, showing promise as a treatment [256] and drug delivery strategy [[257], [258], [259]]. To date, long-term, repeated FUS treatment has been limited. One recent study evaluated Alzheimer's disease patients that received three FUS-directed treatments performed three weeks apart [260]. The study reported no apparent cognitive worsening 1-year after FUS treatment. However, n-GSD patients would likely need repeated FUS treatment for decades. Thus, repeated FUS treatment must be carefully assessed, as neuroinflammation and adverse side effects like edema do occur [252,261]. Moreover, FUS was designed to precisely target a desired brain region, such as a glioma, and current technology may not be an optimal therapeutic approach for n-GSDs where multiple brain regions need to be targeted. Nonetheless, FUS offers a potential avenue to deliver small molecules and biological therapeutics to cross the BBB for more effective and targeted neurological treatments.

Key challenges

Over the last two decades, there have been significant advancements in therapeutic approaches for CNS diseases, moving beyond conventional symptom-targeting treatments toward therapies that can modify the course of the disease itself [262]. Despite these strides, the effectiveness of various treatments remains constrained by specific and shared challenges. For n-GSDs, specific challenges involve gaps in our understanding of disease-specific mechanisms as highlighted below.

• •Which brain/CNS regions are primarily affected in different n-GSDs?
• •Which cell types are involved in glycogen/PGB accumulation?
• •How do glycogen/PGBs compartmentalize at the subcellular level?
• •How does glycogen distribution change in brain regions in n-GSDs?
• •How does glycogen structure, architecture, and composition change in n-GSDs?
• •What is the region- and/or cell type-specific molecular/metabolic signatures linked to perturbed glycogen metabolism in n-GSDs?
• •What molecular mechanism determines whether glycogenolysis occurs through lysosomal hydrolysis or cytoplasmic phosphorolysis?
• •Developing molecules or therapeutic approaches capable of targeting both the CNS and peripheral components.

Concluding remarks and future perspectives

The challenges associated with the development of n-GSDs therapeutics emphasize the need for thorough investigations into disease mechanisms, target identification, selection of active agent and appropriate therapeutic approaches. It is noteworthy that great strides are being rapidly made in multiple classes of n-GSD therapeutics which provides hope to n-GSD patients and their families. As the safety and efficacy of these new treatment approaches are clinically evaluated, their integration into more broad treatment plans offer novel ways to alleviate n-GSD and more broadly CNS disease pathologies and improve patient quality of life.

While n-GSDs pose significant clinical challenges, continued collaborative efforts between clinicians, researchers, and patient advocacy groups are crucial for advancing the field and ultimately transforming the landscape of n-GSD management. Overall, the treatment landscape for rare diseases within the same class (like LSDs) is rapidly evolving, with ongoing research efforts focused on developing innovative therapies and improving the understanding of disease pathophysiology. We hope that addressing the knowledge gaps highlighted here about CNS glycogen metabolism will catalyze further progress toward understanding its importance in CNS functions, human health, and n-GSDs management.

Author contributions

M.Colpaert, P.K.S., M.S.G., and C.W.V.K. wrote the initial draft. M.Colpaert and P.K.S. generated figures. K.J.D., N.T.P., D.D.F., M.Corti, B.J.B. and R.C.S. each wrote sections specific to their areas of expertise. All authors participated in editing and generating the final draft.

Declaration of competing interest

R.C.S. has received research support and consultancy fees from Maze Therapeutics and is a member of the Medical Advisory Board for Little Warrior Foundation. M.S.G. has received research support, research compounds, or consultancy fees from Maze Therapeutics, Valerion Therapeutics, Ionis Pharmaceuticals, PTC Therapeutics, and the Glut1-Deficiency Syndrome Foundation. R.C.S. and M.S.G. are co-founders of Attrogen LLC. M.Corti has received research support from Sanofi, Friedreich Ataxia Research Alliance (FARA), Amicus, AavantiBio, Lacerta, Provention Bio, Sarepta, Duchenne Research Fund, Muscular Dystrophy Association (MDA), GoFAR, Cydan, Audentes. M.Corti has received consulting fees from AavantiBio, Reata, Lilly, Avexis and Gilbert foundation, SwanBio and PCT Therapeutics. B.J.B. has received research support from SolidBio, ProventionBio, Barth Syndrome Foundation. B.J.B. has received consulting fees from AavantiBio, Amicus Therapeutics, Rocket Pharma, Pfizer, Sanofi, and Sarepta Therapeutics. M.Corti and B.J.B. are co-founders of Ventura, LLC.