The 6^th International Lafora Epilepsy Workshop: advances in the search for a cure.

Opening, remarks from NIH, and the LD registry

The meeting started with opening remarks and welcomes from Drs. Harris and Gentry. The first talk was given by Dr. Miriam Leenders, a NIH/NINDS program officer overseeing acquired and rare epilepsies and basic synaptic function projects. Dr. Leenders explained the strong commitment from NINDS to basic research and rare diseases, including the goal to fund discoveries of fundamental mechanisms to reduce the burden of neurological disorders. She highlighted that NINDS-funded research supports all neurological disorder and spans from basic, translational, to clinical research and, in fact, a large part of the funding goes to rare diseases and epilepsy ($416.9 million and $128.5 million USD in 2019, respectively), including LD funding to the LECI and many individual investigators (available at https://report.nih.gov/). In line with the strong commitment to reach their goals, she discussed the BRAIN Initiative, investments in tool development, and introduced a new program called Ultra-rare Gene Therapy (URGenT) Network.

Next, Dr. Juan Gonzalez (Fundación Jimenez Diaz Hospital) gave an update on the status of the LD registry, launched by Dr. Jose Serratosa, and he described how patient enrollment has progressed over the last year. As of October 2020, 44 patients are enrolled in the registry which has sites in Spain, Pakistan, Italy, and Germany and encompasses a dynamic prospective of patients worldwide. The registry is free of charge and includes teleconference visits every 6 months. The goal of the registry is to gather crucial data and provide an updated database of all LD patients. Gathering of these data will allow physicians and researchers to make critical decisions about treatment, clinical evaluations, perform genotype-phenotype correlations, and ultimately provide information for the natural history study. Dr. Gonzalez also explained how to enroll and stated that physicians and patients can email [email protected] for questions related to the registry.

Anti-sense oligonucleotides (ASO) as an LD therapeutic

The remainder of the first session focused on the development of anti-sense oligonucleotides (ASO) to suppress glycogen synthase 1 as an LD treatment. This therapy represents one of the exciting approaches being developed to treat LD patients (Fig. 1). ASO development, efficacy, and a roadmap for treatment of LD patients was presented by Dr. Paul Goldberg (Vice President of Clinical Development at Ionis Pharmaceuticals), Dr. Berge Minassian (University of Texas – Southwestern), and Dr. Silvia Nitschke (University of Texas – Southwestern).

Dr. Goldberg opened his talk emphasizing that Ionis has a team of scientist working to alleviate the burden of this disease and dedicated the talk to the LD children and their families. He then introduced the power of the ASO platform, which are nucleic acid-derived drugs that can very specifically control target protein expression. The ASO platform is highly precise and at least ten ASO drugs are currently FDA approved, with at least four from Ionis, and many more in human clinical trials (6). These approved drugs, exemplified by the drug Spinraza (Nusinersen), have already changed the future for many patients and provide similar hope for an LD ASO therapy (7–11). Dr. Goldberg then explained how in LD an imbalance between glycogen elongation and branching leads to the formation of glycogen aggregates known as Lafora bodies (LBs). The ASO works by suppressing glycogen synthase (GYS1) and LB formation. Preclinical data in the murine laforin deficient mouse model indicate that the suppression of glycogen synthesis dramatically decreases GYS1 mRNA and protein levels as well as glycogen and LBs. With the preclinical murine data in hand, Ionis has identified the most active and safe human drug candidate for patients. From here, the drug candidate must go through chronic toxicology studies as well as clinical studies.

Goldberg then transitioned into elucidating some of the challenges encountered in drug development for rare diseases, including a small patient population, under/misdiagnosis in patients, high phenotype variability, and lack of established efficacy endpoints. These parameters present challenges with recruitment, retention, trial size, power, trial design, and regulatory approval. Nevertheless, most of these issues can be addressed and overcome by the development of a natural history study and by utilizing specific regulatory pathways.

Natural history studies aim to track a disease’s course over time and understanding its normal progression. Such studies can be vital and help optimize the clinical study design by providing information such as spectrum of disease, rate, progression, clinical manifestation, optimal eligibility criteria for interventional study, and by identifying valid biomarkers and surrogate endpoints. An important LD prospective natural history studies was initiated January 2019 and is ongoing that includes 33 patients (ClinicalTrials.gov Identifier: NCT03876522). Additionally, retrospective data have been gathered and analyzed from patient charts in Spain, Serbia, and Italy. Finally, Goldberg emphasized that the natural history is vital to support the interventional study, allowing the drug to be administered to patients. Though most experiments were conducted in mice, the efficacy and safety in patients are expected to be similar to other ASOs which have been around for over 30 years. Goldberg also explained Ionis’ planned collaboration with the FDA to utilize expedited regulatory routes, which will result in a faster drug approval, and passionately closed highlighting Ionis’ commitment to the LD community.

Drs. Berge Minassian and Silvia Nitschke from UT-Southwestern Medical Center detailed the key pre-clinical ASO efficacy data for the treatment of LD. In collaboration with Ionis Pharmaceuticals, they investigated the pharmacological properties of the ASO targeting the GYS1 mRNA. Nitschke presented data from several studies where they treated LD mice at an early (1-3 months), intermediate (3-6 months) and late stage (8-14 months) of the disease in two LD mouse models. The data suggest that an early intervention reduces both glycogen and LB levels, successfully halting disease progression in mice. Their studies indicate that the formation of large LBs can be prevented with the early treatment yielding better results. ASO treatment also reduced neuroinflammation, evident by defining the levels of the inflammatory markers Cxcl10, Ccl5, and GFAP. These data cumulatively show a promising effect of the ASO treatment capable of halting LD progress and perhaps preventing the disease with early intervention (12).

Repurposing drugs to ameliorate LD

In the search for immediate ways to ameliorate the progression of LD, the community is also investigating the use of repurposing drugs (Fig. 1). A LD hallmark that could be addressed by repurposed drugs is neuroinflammation. Three talks presented by Drs. Subramaniam Ganesh (Indian Institute of Technology Kanpur), Pascual Sanz (CSIC-Valencia), and Joan Guinovart (IRB-Barcelona) focused on the role of neuroinflammation and autophagy in LD and the use of repurposed drugs to address these hallmarks.

Dr. Sanz briefly highlighted past work that resulted in the orphan-drug designation of metformin to treat LD (European Medicines Agency no. EU/3/16/1803, Food and Drug Administration, #DRU-2017-6161) (13, 14). He then focused on recent work demonstrating that in LD mice, genes encoding proinflammatory and phagocytosis proteins are dramatically upregulated (15). This discovery inspired multiple studies where they utilized repurposed drugs to treat neuroinflammation as an alternative way of using disease-modifying drugs. Sanz elegantly explained the vicious cycle of LD pathophysiology where LBs trigger other pathophysiological processes, e.g., neuroinflammation, which then cause further damage. He also explained how the use of repurposed drugs is a promising alternative since they have the potential to lower overall drug development costs and shorten development and approval timelines. He then presented data from studies where they investigated the pharmacological properties and treatment possibilities of three drugs affecting the inflammatory response in mice: cannabidiol, epigallocatechin gallate (EGCG), and propranolol (16, 17). While the treatment with cannabidiol proved disappointing, propranolol, and EGCG improved memory and attention deficits in the LD mice. Furthermore, propranolol reduced astrogliosis and microglia activation while not affecting LB numbers. These studies demonstrate a promising therapeutic potential of treating neuroinflammation in LD with repurposed drugs.

Dr. Ganesh presented on the correlation between neuroinflammation, progressive myoclonus epilepsy genes, and autophagy. He emphasized that neurodegeneration, inflammation, and epilepsy correlate positively with age while autophagy and heat shock response presents a negative correlation (18–23). Interestingly, heat shock proteins are also reduced in LD mouse models and the treatment with the repurposed drug ibuprofen led to an increase in heat shock proteins and a reduction in LBs. These data suggest a link between inflammation and LBs and provide an exciting new avenue to explore. However, it was noted that these data are based on murine studies and utilized a high dose of ibuprofen whereas, in humans, chronic high dose treatment is associated with health risks.

Dr. Guinovart concluded the session with an intriguing talk focusing on a novel role for p62 in LD. p62 plays a fundamental role in the autophagy machinery and degradation of proteins (24). Furthermore, deficits in autophagy are associated with the lack of laforin and malin (25, 26). However, the specific role of p62 in LB formation and if it could be a putative drug target in LD was unknown. Guinovart presented data showing increased p62 in the malin KO mouse model of LD and hypothesized that p62 plays a role in LB formation. They generated a mouse lacking both malin and p62 and observed an ablation of muscle LBs and a change in morphology of brain LBs. These data suggest that p62 plays a key role in LB formation or clearance and that it is required for muscle LB accumulation.

Cumulatively, neuroinflammation, autophagy, and associated pathways are promising targets for LD disease modifying drugs. The true value of repurposed drugs affecting these pathways will be realized when future human trials are employed.

Targeting LBs and a LD-patient mutation mouse model

Data from multiple laboratories have demonstrated that decreasing glycogen synthesis is a promising approach to inhibit LB formation and LD progression (27–29). Dr. Olga Varea, a postdoctoral fellow in the Guinovart laboratory, presented their recent data about glycogen synthesis suppression as a treatment for LD and the importance of intervention timing. Previous findings showed that deletion of GYS1 prevented LB formation and reduced inflammatory markers in malin knockout (MKO) mice (29). Varea and colleagues sought to determine whether it was possible to modify LB accumulation via reducing GYS1 protein levels once the disease initiated. To answer this question, they generated an LD MKO mouse with a tamoxifen-inducible deletion of GYS1 (30). They induced GYS1 deletion at 4-, and 6-months in MKO and control animals. Early suppression of GYS1 by 4-months arrested LB accumulation in the brain and muscle. In the muscle, this intervention not only blocked new LBs, but also reduced the number of LBs. In the brain, it was only possible to arrest LB numbers with an early intervention and GYS1 suppression at 4-month was insufficient to prevent the neuroinflammatory response. Using a different model, the Minassian laboratory also recently published similar results (31). These data highlight that an early LD genetic diagnosis is crucial for the efficiency of glycogen synthase-based therapeutics.

Dr. Tom Hurley (Indiana University) provided an update on the Roach, DePaoli-Roach, and Hurley laboratories’ efforts to identify small molecule inhibitors of glycogen synthase (GS) (Fig. 1) and to perform structural and mechanistic studies. This team identified initial compounds that can bind to the GS active site and then they screened analogs and improved potency by 300-fold (32). Currently, their lead compound, a competitive inhibitor at the UDP-glucose active site, has an inhibition constant of 1.3 μM (32). Excitingly, these compounds have also shown cellular efficacy by reducing glycogen accumulation in cell lines, making them very promising to advance to animal models. Hurley then reported on their recent LD mouse results. They generated an LD mouse model expressing a catalytically dead version of laforin (laforin-C265S) at the endogenous locus. They crossed this mouse with the MKO model and found that the absence of laforin catalytic activity does not affect the MKO phenotype. These results infer that since different mutations yield similar phenotypes, treatments that are successful for laforin-driven LD or malin-driven LD should successfully impact all LD mutations. Finally, Hurley presented work using CRISPR technology to manipulate the endogenous malin gene by introducing a myc-tag at the C-terminus of malin. This work was initiated due to the lack of quality malin antibodies. Initial metabolic analyses yielded similar profiles for WT and Malin-myc mice, suggesting that the malin-myc protein is functioning and providing an excellent new tool in LD research.

Dr. Serratosa (Fundación Jimenez Diaz Hospital) highlighted his team’s work on generating and characterizing the first LD mouse model with a patient mutation, the laforin-R240X knock-in mouse model. They assessed WT, laforin KO and laforin-R240X behavioral patterns, memory, learning, and motor skills as well as neurologic functions and the epileptic phenotype at 3-and 6-months. The laforin-R240X mice exhibited similar LB numbers and levels of reactive astrocytes as laforin KO mice at 3-month. Strikingly, the laforin-R240X mice presented with laforin aggregates that are absent in the laforin KO mice. Additionally, they displayed worse learning and memory deficits and a trend towards increased myoclonic jerks while motor coordination was preserved compared to laforin KO mice. At 6 months, they present with worse memory and cognition, a trend towards decreased motor coordination compared to the KO model, increased seizure length and higher lethality. The laforin-R240X mice exhibited increased LBs, more pronounced memory deficits, increased PTZ-induced myoclonic jerks, and longer PTZ-induced generalized tonic-clonic seizures, compared to laforin KO. Thus, this mouse model provides LD patient mutation mouse model to test LD therapies.

Dr. Gentry provided the final talk of this session and highlighted the central idea and ultimate goal of inhibiting or ablating LBs and/or targeting other LD hallmarks. These efforts have established five therapeutic options that are currently being pursued (Fig 1). Gentry started with the recent work on an antibody-enzyme fusion (AEF) with cell penetrating properties that degrades LBs and normalizes brain metabolism after treatment (33, 34). Gentry and colleagues then analyzed glycogen from liver, skeletal muscle, and brain, and they found that while liver and muscle glycogen is 95-99% glucose that brain glycogen is comprised of ~75% glucose and 25% glucosamine. Glucosamine is a precursor of UDP-N-acetylglucosamine (GlcNAc), which is subsequently used in N-glycosylation and O-GlcNAcylation. Collaboration with Hurley, Roach, and Depaoli-Roach yielded a crystal structure of glycogen synthase bound to UDP-GlcNAc as well as enzymatic data demonstrating that UDP-GlcNAc is a substrate of glycogen synthase. In collaboration with Dr. Ramon Sun (University of Kentucky), they demonstrated that glucosamine is released from glycogen by the concerted actions of glycogen debranching enzyme (GDE) and glycogen phosphorylase (PYGB). Thus, these data suggest that glucosamine is regularly incorporated into and liberated from brain glycogen. To test this hypothesis, they generated primary astrocytes with reduced levels of Pygb and observed glycosylation defects. Additional data from laforin KO mice corroborated the in vitro data and a similar glycosylation defect. Thus, normal brain glycogen is a reservoir for GlcNAc that can be converted to UDP-GlcNAc and utilized for protein glycosylation. In LD, GlcNAc is sequestered in LBs and cannot be used for protein glycosylation. To further test this hypothesis, they collaborated with Drs. Peter Roach, Anna DePaoli-Roach and Richard Taylor using a novel glycogen storage disease II mouse model lacking GDE. Strikingly, the GDE KO mice exhibit increased glycogen-derived glucosamine and decreased glycan-derived GlcNAc, mirroring the laforin KO phenotype. Finally, Gentry presented data that the AEF therapy rescued the glycosylation phenotype in the brain of LKO mice. These results demonstrate that the disruption of brain glycogen metabolism in LD is also associated with abnormal glycosylation.

Clinical Insights

Three talks from LD neurologists, Drs. María Machío, (Fundación Jimenez Diaz Hospital), and Alison Dolce, (University of Texas-Southwestern), and Delgado-Escueta (University of California-Los Angeles), provided clinical insights into EEG abnormalities and LD stages. Two clinical talk by Drs. Machio and Dolce focused on the power of EEG to define early-stage LD. Dr. Dolce described intriguing differences between EEG recordings from two siblings where one was pre-symptomatic and the other symptomatic. Her data demonstrated that even the pre-symptomatic patient exhibited EEG abnormalities. Although the background EEG had normal organization with preserved sleep architecture and identifiable sleep stages, there was intermittent slowing (notched FIRDA, frontal intermittent rhythmic delta activity) with intermixed multifocal spikes along with high amplitude generalized irregular spike or polyspike and slow wave discharges. The symptomatic sister displayed both abnormal background with loss of normal sleep staging and a diffuse spike/polyspike pattern in the EEG as previously described for LD patients as an electrical firestorm (27).

Machío described the EEG findings in four patients with LD (two pre-symptomatic and two in early stage) illustrating that the EEG is grossly abnormal very early in the evolution of the disease. Abnormalities were mainly reflected in slowing of the background activity and in the appearance of interictal epileptiform discharges. These findings highlight the importance of the possible use of EEG as an early indication of LD even before clinical signs appear.

Delgado-Escueta emphasized the importance of knowing the pathology and progression of LD to conduct valuable clinical trials, including issues related to late diagnosis, age of onset, and semiology of seizures. He emphasized that many of these issues can be addressed by enrollment of patients in the natural history study. He then addressed some of the parameters and patient groups that will be valuable to include in phase I and phase II clinical trials such as EEG recordings, asymptomatic patients as well as patients with myoclonus in later clinical studies. Delgado-Escueta then elegantly defined five LD stages from analyzing a wealth of patient data that he and his colleagues have accumulated. Defining the criteria for clinical trials and elucidating an understanding of LD progression is critical for successful LD treatments.

Conclusions

Work presented at this meeting and published by the highly collaborative LECI groups have uncovered a series of LD hallmarks and mechanisms that are key for understanding LD pathology and identifying treatments. These groups have demonstrated that mutations in EPM2A or EPM2B result in increased levels of glycogen accumulation, formation of LB aggregates, mitochondrial dysfunction, autophagy dysfunction, neuroinflammation, metabolic dysregulation at the cellular level, glycosylation defects and glutamate transporter dysfunction (Fig. 2). Excitingly, work presented at this meeting described five parallel and potentially synergistic therapeutics to clear existing LBs and prevent de novo synthesis of new LBs. The groups are also elucidating the needed foundational knowledge of timing, target, delivery, and intervention strategy to translate pre-clinical discoveries into effective therapeutic patient treatment. The collaborative effort between the groups increases the likelihood that discoveries can be transferred from their laboratories to patients living with the fatal epilepsy LD and the discoveries have implications for other glycogen storage diseases. The progress in LD research also suggests a model for other rare diseases: identify the genetic locus, generate appropriate mouse models, determine the biochemical function of the proteins involved, elucidate a drug target, initiate pre-clinical drug development, and partner with industry all while working with patient advocacy groups.