Two Diseases—One Preclinical Treatment Targeting Glycogen Synthesis

The more than 7,000 rare diseases impact > 350 million patients worldwide and their families, with only 5% having an FDA-approved treatment [1]. Approximately 80% of these diseases are due to inherited or de novo genetic mutations involving one or several genes. While many of these diseases are devastating, basic science can provide critical insights on how to treat several of them. Treatments for cystic fibrosis, Pompe disease, spinal muscular atrophy, and others have defined a path whereby dramatic progress can be made to define disease mechanisms and novel treatments can be implemented in the clinic. Researchers can identify the genetic locus/loci, determine if there is an altered protein, define the function of the expressed protein, and devise a therapy to fix the aberrant protein and/or develop a cellular work-around. Two drugs treating spinal muscular atrophy have recently been developed employing the paradigm of gene correction or cellular work-around. Onasemnogene abeparvovec (Zolgensma®) is an adeno-associated virus (AAV) gene therapy that introduces a functioning SMN1 gene to the affected motor neurons [2]. Nusinersen (Spinraza®) is an antisense oligonucleotide (ASO) that modulates alternative splicing, i.e., a work-around, to overcome loss of the SMN1 gene [3].

In this issue of Neurotherapeutics, Minassian and colleagues describe an innovative pre-clinical work-around to treat two independent neurogenetic diseases [4]. They develop an AAV vector to deliver an artificial microRNA (amiRNA) via a single dose to mediate RNA-interference (RNAi) of a single target that shows impressive benefits in multiple mouse models. Lafora disease (LD) and adult polyglucosan body disease (APBD) are autosomal recessive glycogen storage diseases (GSDs) that present with neurological symptoms in patients and neuro-centric phenotypes in the respective mouse models [5–7]. Both of these clinically distinct diseases exhibit aberrant glycogen-like aggregates in the cytoplasm of neurons and astrocytes called polyglucosan bodies (PGBs) that drive neuroinflammation, astrogliosis, and neurodegeneration. The PGBs in both diseases are synthesized by glycogen synthase (GYS1), and the PGBs are not broken down by normal cellular degradation pathways. Thus, PGB formation is analogous to a clogged sink with a running faucet (Fig. 1). Previous work from multiple laboratories independently demonstrated that genetic inhibition of PGB synthesis is a viable pre-clinical treatment path. These studies employed Gys1 knockout mice and other genetically modified mouse models that reduce glycogen synthesis crossed with the LD or APBD mice to establish that downregulating glycogen synthesis in LD and APBD mouse models rescued disease phenotypes [8–12]. Thus, inhibiting glycogen synthesis via the reduction of glycogen synthase expression is a means to decrease the faucet flow, inhibit PGB formation, and treat the diseases (Fig. 1).

Fig. 1.

Representation of polyglucosan body formation. Glycogen is synthesized by the combined activity of glycogen synthase (GYS1) and glycogen branching enzyme (GBE) and it is degraded by glycogen phosphorylase (PYGB, brain isoform) and glycogen debranching enzyme (GDE). Lafora disease (LD) and adult polyglucosan body disease (APBD) patients and mouse models exhibit aberrant, glycogen-like aggregates called polyglucosan bodies. When glycogen synthesis is reduced in LD or APBD, polyglucosan body levels are reduced

Herein, the authors generated an AAV-mediated RNAi strategy to downregulate brain-expressed Gys1 and achieved a 50% PGB reduction across multiple brain regions in both APBD and LD mouse models. These impressive results were accompanied with improvements in neuroinflammatory markers. This work provides an important proof-of-concept model whereby a single-dose therapy for APBD and LD could provide lifetime benefit.

APBD and LD are metabolic diseases that impact different steps of glycogen metabolism. Glycogen functions as an energy reserve largely found in the liver and skeletal muscle, but the brain is the organ most susceptible to decreases in glucose availability [13, 14]. Glycogen is comprised of glucose chains arranged in a linear fashion with branch points at regular intervals to allow for maximal glucose packing in a water-soluble molecule. Mutations in the genes encoding proteins that regulate glycogen metabolism result in GSDs [15]. The biochemical defect for many of the GSDs was defined before the molecular biology era, and research on glycogen metabolism resulted in six Nobel Prizes to Gerty and Carl Cori (1947) for their work on glycogen phosphorylase; Luis Leloir (1970) working on glycogen synthase; Earl Sutherland (1971) linking cAMP, epinephrine, and glycogen phosphorylase; and Edmond Fischer and Edwin Krebs (1992) for discovering glycogen phosphorylase and phosphorylase kinase stimulation by reversible phosphorylation. Due to their work and others, the GSDs have a rich biochemical history, and glycogen metabolism is a foundational bedrock in biochemistry textbooks (Fig. 2A).

Fig. 2.

Glycogen metabolism. A Summary of major steps of glycogen metabolism. Glucose enters cells via a glucose transporter (Glut1) and is phosphorylated by hexokinase (HK) to generate glucose-6-phosphate (G6P). G6P is converted into glucose-1-phosphate by phosphoglucomutase (PGM) and then to UDP-glucose (UDP-Glc) by UDP-glucose pyrophosphorylase (UGP). Glycogen synthase (GYS1) and glycogen branching enzyme (GBE) catalyze glucose chains and branching, respectively, to synthesize glycogen. Glycogen is catabolized by the sequential removal of G1P monomers via glycogen phosphorylase (PYGB) with glycogen debranching enzyme (GDE) removing branch points. Adult polyglucosan body disease (APBD) is caused by mutations in GBE. B Lafora disease (LD) is caused by autosomal recessive mutations in the gene encoding laforin or malin. Laforin is a glycogen phosphatase that removes phosphate from glycogen (component 1). Malin is an E3 ubiquitin ligase that forms a complex with laforin and ubiquitinates a number of enzymes that synthesize or catabolize glycogen (component 2). Perturbed laforin or malin activity results in aberrant glycogen metabolism and PGB formation, and PGBs drive LD progression

APBD is an adult-onset neurodegenerative disease characterized by spastic paraplegia, neurogenic bladder dysfunction, axonal neuropathy, and cognitive impairment [7, 16]. APBD is caused by autosomal recessive mutations in GBE1 that encodes glycogen branching enzyme, and there is currently no cure for the disease. APBD patients retain some residual GBE activity, while patients with little to no GBE activity suffer from a fatal perinatal or childhood-onset disease, which is classified as Andersen disease (GSD IV). GBE catalyzes regular branch points in the glucose chains that form glycogen and acts in concert with glycogen synthase to catalyze glycogen as a water-soluble macromolecule (Fig. 2A). In APBD, glycogen synthase continues to elongate the glucose chains while GBE fails to appropriately form branches. Thus, APBD results from an imbalance between glucose chain branching and chain elongation resulting in PGB formation. APBD PGBs are observed in astrocytic processes but are most prevalent in neuronal axons [17]. Similar to LD, multiple laboratories have demonstrated that APDB PGBs are the driver of APBD pathophysiology and, thus, constitute a therapeutic target [8, 18–20].

LD is a fatal form of progressive myoclonus epilepsy characterized by severe childhood epilepsy and dementia [5, 6]. Onset occurs in healthy adolescent children with myoclonic jerks, seizures, and cognitive decline. Initial response to antiseizure medications is lost within a few years, and patients develop constant myoclonus, frequent generalized tonic–clonic seizures, and severe dementia in their teen years. The patients become bedridden despite any combination of anti-seizure medications, and death comes after a protracted decade of unceasing myoclonus from respiratory issues, sudden unexpected death, or status epilepticus. LD is caused by autosomal recessive mutations in EPM2A, which encodes the glycogen phosphatase laforin, or EPM2B, which encodes the E3 ubiquitin ligase malin that ubiquitinates a number of glycogen metabolism enzymes (Fig. 2B) [21–26]. Perturbations in the activity of laforin or malin result in aberrant glycogen metabolism and PGB formation. Using LD mouse models, multiple laboratories have demonstrated that LD PGBs impact nearly every region of the brain and are the driver of LD pathophysiology and disease progression, and therefore PGBs are the therapeutic target [8–12, 20, 27–33].

For both APBD and LD, multiple groups utilized disease mouse models and biochemical insights to identify the disease target, and for both diseases, the goal is to decrease PGBs. Excitingly, there has been recent progress in developing pre-clinical therapeutic platforms to treat both diseases via antibody-enzyme therapy to degrade PGBs, and employing antisense oligonucleotide (ASOs) or small molecules to decrease glycogen synthesis [18, 19, 27, 31, 32, 34]. The antibody-enzyme therapy and ASO each displayed impressive target engagement in LD mouse models and rescued aspects of the phenotype. However, the antibody-enzyme therapy and ASOs both require continuous or multiple intrathecal injections to deliver the drug since they do not readily cross the blood–brain barrier. While repeated intrathecal administration is utilized in the clinic, an AAV approach offers a single lifetime dose.

Previous work from the Minassian laboratory identified and employed an ASO targeting brain-expressed Gys1 in laforin knockout (KO) and malin KO LD mouse models [32]. The Gys1-ASO was administered via multiple intracerebroventricular injections and prevented PGB formation in young mice that had not yet formed PGBs and inhibited further PGB accumulation in older mice. In each case, PGB inhibition was associated with decreased astrogliosis and correction of neuroinflammatory markers.

In the current study, the authors developed an artificial microRNA (amiRNA), utilizing a similar siRNA Gys1-targeting sequence from exon 8 of Gys1 cloned into the natural human/mouse miR-30 primary microRNA scaffold, followed by the GFP coding sequence and a 3′ bGH poly(A) signal to stabilize the transcript that is driven by the CBh promoter. The cassette is flanked by AAV2 inverted terminal repeats and cloned in a self-complementary fashion to generate the Gys1-targeting amiRNA packaged in AAV9. The amiRNA technology is superior to earlier short hairpin loop RNAs (shRNAs) since the amiRNAs are expressed at lower levels and processed more efficiently. Thus, amiRNAs do not exhibit higher cellular toxicity profiles because they do not saturate the endogenous RNAi machinery, which is observed with shRNAs.

After construct optimization in N2A cells, the authors tested the construct in both LD mouse models, laforin KO and malin KO, and the APBD mouse model, Gbe1^Y239S. The Gys1-targeting AAV-amiRNA or vehicle was delivered by a single intracerebroventricular injection at postnatal day 2, and mice were sacrificed and analyzed at 3 months of age. Impressively, GFP expression was observed across a broad distribution of the brain in all three mouse models after a single dose. The Gys1-amiRNA provided a 13–18% reduction in Gys1 mRNA and an impressive 40–50% reduction in Gys1 protein levels. Previous work crossing the LD or APBD mice with Gys1 heterozygous mice demonstrated that a 50% reduction in Gys1 protein levels dramatically decreased PGB formation and rescued disease pathology [9, 11, 12]. In agreement with these previous results, the authors observed a 40–60% decrease in PGBs in the hippocampal area in all three mouse models. Each of these mouse models exhibit astrogliosis and microgliosis by 3 months of age that is associated with increased immune or inflammatory pathway gene transcripts [35, 36]. In each of the three mouse models, the authors observed a reduction in immune or inflammatory transcript levels that is in agreement with previous work and demonstrates a cellular response to inhibition of PGB formation.

This work provides an exciting step toward the clinic in developing a treatment for multiple diseases with potentially a single dose providing a lifetime benefit. Future work will focus on defining how late in disease progression the treatment can be administered to provide beneficial results, as well as how long the Gys1-amiRNA AAV decreases Gys1 protein expression. The more robust impact on Gys1 protein compared to its transcript is a reoccurring theme, as two other recent studies focused on Gys1 observed similar effects [20, 33]. If a similar trend is confirmed in human cells with the humanized construct, then this mechanism would represent an important advantage in progressing to the clinic since lower viral genome counts could be sufficient and/or utilization of promoters with a lower strength could be employed. In addition to this AAV amiRNA strategy to treat LD and APBD patients, there are a number of GSDs where patients present with a neurological component. The most common of the GSDs is Pompe disease where enzyme replacement therapy (ERT) has been the standard of care since 2006. However, the ERT does not cross the blood–brain barrier, and recent work has highlighted the unmet need for addressing the neurological manifestations of Pompe patients [37]. Thus, as this AAV amiRNA Gys1 therapy moves toward the clinic, it could be utilized for multiple diseases.

Funding

This work was supported by National Institute of Health grants R35 NS116824 and P01 NS097197 to M.S.G.

Declarations

Disclosures

M.S.G. has research support and research compounds from Maze Therapeutics, Valerion Therapeutics, and Ionis Pharmaceuticals. M.S.G. has provided consulting services to Maze Therapeutics, PTC Therapeutics, Aro Biotherapeutics, Valerion Therapeutics, Enable Therapeutics, Chelsea’s Hope, and the Glut1-Deficiency Syndrome Foundation. M.S.G. is a co-founder of Attrogen LLC.