53 Therapeutic Window for the Treatment of Lafora Disease

Glycogen is a polymer of glucose, and it is the only energy reservoir in the brain. Within the central nervous system (CNS) and other tissues, such as skeletal muscle and heart, glycogen is synthesized by the muscle isoform of glycogen synthase (MGS, codified by the GYS1 gene), whereas in the liver it is synthesized by the liver isoform (LGS, codified by the GYS2 gene). MGS activity is tightly regulated by several mechanisms, including allosteric activation by glucose-6-phosphate and inactivation by phosphorylation at multiple sites by several kinases (Goldberg and O’Toole, 1969; Skurat et al., 1994). The generation of mice lacking MGS specifically in the CNS has greatly contributed to unraveling the physiological role of glycogen in the brain (Duran et al., 2013; López-Ramos et al., 2015). These animals are viable and have a normal life span; however, interestingly, they present deficiencies in learning and memory and have an increased susceptibility to epilepsy.

In physiological conditions, most brain glycogen is located in the astrocytic population (Cataldo and Broadwell, 1986), where its metabolism is regulated in response to a variety of external factors, mainly neurotransmitters and extracellular glucose levels (Hertz et al., 2010; Öz et al., 2016; Briski et al., 2021). The physiological role of glycogen in astrocytes is associated with the preservation of neuronal function and the regulation of synaptic activity. In this regard, astrocytic glycogen acts as a glucose reservoir for neurons and it also supports K⁺ and neurotransmitter uptake (Swanson and Choi, 1993; Wender et al., 2000; Xu et al., 2013). Until recently, neurons were considered not to have an active glycogen metabolism in physiological conditions. However, the observation that several glycogen storage diseases are characterized by neuronal accumulation of intracellular aggregates of glycogen, as occurs in Lafora disease, already indicated that neurons have the machinery necessary for the synthesis of this polysaccharide. The presence of an active glycogen metabolism in this cell population was first demonstrated in vitro (Vilchez et al., 2007; Saez et al., 2014) and has been confirmed more recently in vivo (Duran et al., 2019) in experiments with mice in which the expression of MGS is suppressed specifically in CamK2a-positive neurons. These animals showed that glycogen metabolism in this neuronal population contributes to learning and memory processes.

Lafora disease (LD; OMIM#254780 ORPHA501) is a fatal neurodegenerative childhood dementia that was first described by the Spanish neurologist Gonzalo Rodríguez Lafora in the early 1900s (Lafora and Glueck, 1911) and eventually identified as a glycogen storage disease (Sullivan et al., 2017; Gentry et al., 2018). It typically manifests during childhood as myoclonus epilepsy with symptoms that worsen progressively, including myoclonus, occipital seizures, atypical absence seizures, and generalized tonic-clonic seizures. The disease progresses swiftly with neurodegeneration and rapidly leads to dementia. LD is caused by the recessive inheritance of mutations in one of two genes, either laforin (EPM2A) or malin (NHLRC1 or EPM2B) (Turnbull et al., 2016). Laforin is a dual-specificity phosphatase with a carbohydrate-binding domain through which it binds to glycogen (Wang et al., 2002; Ganesh et al., 2004). Through this direct interaction, laforin can dephosphorylate glycogen. Malin is an E3-ubiquitin ligase that binds laforin and promotes the degradation of enzymes of glycogen metabolism, namely glycogen synthase (MGS), protein targeting to glycogen (PTG), and laforin itself (Gentry et al., 2005; Vilchez et al., 2007; Nitschke et al., 2018; Verhalen et al., 2018). Mutations in malin and laforin generate an indistinguishable disease, indicating that both genes participate in the same physiological process. The generation of mouse models lacking laforin (Machado-Salas et al., 2012; Pederson et al., 2013; López-González et al., 2017) or malin (DePaoli-Roach et al., 2010; Valles-Ortega et al., 2011; García-Cabrero et al., 2014) has provided valuable tools for the study of the disease. These animals develop progressive neurodegeneration and neuroinflammation and have an increased susceptibility to epilepsy, mimicking the main symptoms of the disease. Many other factors have contributed to the rapid increase in our understanding of LD. These factors include the cross-breeding of LD models with other mice with an altered glycogen metabolism, rapid tissue fixation methods that allow the preservation of glycogen, the application of immunofluorescence and imaging techniques, and the development of specific molecular tools, among others. These methods revealed how the aberrant glycogen is distributed in the tissue. Indeed, these tools have brought about great advances in our understanding of the role of glycogen in the brain and the contribution of each of the cell populations involved in the progression of LD.

The neuroinflammation found in LD is characterized by the presence of profound astrogliosis and microgliosis, especially in the hippocampus (Duran et al., 2014; Lahuerta et al., 2020; Varea et al., 2021). It has recently been shown that microgliosis is a consequence of astrocytic activation (Duran et al., 2021). Modulation of the inflammatory response has been proposed as a therapeutic strategy since inflammation might contribute to the progression of epileptic seizures once patients show clinical manifestations (Lahuerta et al., 2020; Mollá et al., 2021). The analysis of inflammatory markers corresponding to these two cell populations, together with the measurement of cytokine and pro-inflammatory gene expression, provides a good readout of the general status of the brain, which is particularly important when evaluating a possible therapy for LD (Nitschke et al., 2020; Varea et al., 2021; Ahonen et al., 2021).

The hallmark of LD is the accumulation of intracellular insoluble aggregates of abnormal glycogen known as Lafora bodies (LBs) in several tissues, including the brain, heart, skeletal muscle, and skin. This pathological accumulation of large carbohydrate particles was first described by Lafora in his initial characterization of the disease. While the mechanisms regulating the formation of LBs and the specific way through which laforin and malin control glycogen metabolism and the formation of aberrant glycogen are not yet clear, it is known that the phosphatase activity of laforin is not relevant in the regulation of LB formation (Gayarre et al., 2014). Studies in laforin-deficient and in malin-deficient mice have shown that LB accumulation in the brain underlies the appearance of neuroinflammation, the increased susceptibility to epilepsy, and the neurodegeneration of LD. Indeed, full or partial suppression of glycogen accumulation in mouse models of LD from conception prevents the progression of the disease (Pederson et al., 2013; Duran et al., 2014; Turnbull et al., 2014). On the basis of these results, it was determined that the accumulation of glycogen in the CNS represents a new pathological mechanism underlying neurodegeneration, dementia, and epilepsy.

In patients, the time course of LB accumulation or that of the appearance of neuroinflammation is still unknown. In LD mouse models, LBs are already present in significant numbers in the brains of 1-month-old mice (unpublished data from our laboratory). These numbers slowly increase until 6 months and, strikingly, after this period LB accumulation accelerates until the age of 15 months, as seen in malin-deficient mice (malin^KO) (Varea et al., 2021). Neuroinflammation in early stages is detectable only by gene expression analysis but becomes clearly visible from 4 months in the form of profound astrogliosis and from 11 months in the form of microgliosis (López-González et al., 2017; Lahuerta et al., 2020; Varea et al., 2021; Duran et al., 2021).

In addition to glycogen, LBs contain several proteins, including enzymes related to glycogen metabolism, such as MGS and laforin (in malin^KO models), and also the autophagy adaptor p62 (Duran et al., 2014; Auge et al., 2018). The accumulation of p62 at the LBs led some researchers to hypothesize that malin and laforin participate in the regulation of autophagy and that the accumulation of LBs was the result of impaired autophagy (Criado et al., 2012). However, it was later demonstrated that p62 deposition is actually a consequence of the buildup of LBs (Duran et al., 2014). It has been recently shown that LBs are a product of the accumulation of aberrant glycogen that is packed with the aid of p62. Indeed, p62 exerts a beneficial effect, lowering the toxicity of unpacked aberrant aggregated glycogen (Pellegrini et al., 2021). This mechanism driven by p62 to protect cells from toxic aggregates has previously been proposed in other neurodegenerative conditions, such as Huntington disease (Bjørkøy et al., 2005; Saitoh et al., 2015). Accordingly, the neuroinflammation and increased susceptibility to epilepsy characteristic of malin^KO mice are worsened in the absence of p62 (Pellegrini et al., 2021). In summary, LB formation can be defined as follows: abnormal glycogen is generated as a by-product of defective glycogen metabolism due to the lack of functional malin or laforin. The increase of cytoplasmic aberrant glycogen is toxic for cells. The autophagy adaptor p62 acts as an aggregating agent to contain the accumulation of glycogen at specific sites in the cytoplasm. The mechanism driven by p62 by which packed LBs are formed is beneficial compared to the presence of diffuse aberrant glycogen throughout the cytoplasm. However, the intracellular accumulation of LBs also alters cell homeostasis and induces neuronal death, neuroinflammation, and the appearance of susceptibility to epilepsy.

Until recently, it was widely believed that LBs accumulate exclusively in neurons. However, it has been demonstrated that astrocytes also accumulate these glycogen aggregates (Valles-Ortega et al., 2011). Strikingly, most LBs found in the brains of LD mouse models are actually present in astrocytes (Rubio-Villena et al., 2018; Duran et al., 2021; Auge et al., 2018). Neuronal LBs (nLBs) are normally found in the form of one large round LB per neuron, located close to the nucleus (Auge et al., 2018). In contrast, astrocytic LBs are known as corpora amylacea-like (CAL) because of their resemblance to corpora amylacea— abnormal glycogen aggregates found in aged brains. Astrocytes accumulate numerous CAL that are distributed throughout their cytoplasm, including cell projections, and present a variety of sizes and morphologies (Auge et al., 2018b). The astrocytes containing CAL are highly abundant in the hippocampus of young malin^KO mice, and their number increases in this region, as well as in others such as the prefrontal cortex, as the disease progresses. Furthermore, CAL are responsible for the inflammatory response since blocking glycogen synthesis specifically in astrocytes prevents the increase in neurodegeneration markers, autophagy impairment, and metabolic changes characteristic of the malin^KO model. Interestingly, astrocytic glycogen does not drive susceptibility to epilepsy, since malin^KO mice lacking astrocytic LBs maintain an increased susceptibility to epilepsy. This pathological trait would therefore be dependent on the accumulation of aberrant glycogen in neurons (Duran et al., 2021). In this regard, it is known that parvalbumin-positive interneurons of the hippocampus accumulate LBs (Valles-Ortega et al., 2011). These GABAergic inhibitory neurons are widely known for their contribution to synaptic plasticity and neural network homeostasis (Caillard et al., 2000; Chevaleyre and Piskorowski, 2014). Deficits in the inhibitory system lead to cognitive impairment (Bast et al., 2017; Xu et al., 2020) and epileptic seizures (Khazipov, 2016). Therefore, the functional impairment of the interneuronal population driven by the accumulation of nLBs could induce an imbalance of the excitatory and inhibitory neuronal network, thus leading to the epileptic phenotype of the disease. Additionally, other regions play a role in epileptogenesis (Aroniadou-Anderjaska et al., 2008; Li et al., 2014; Martín-López et al., 2017), but their contribution to the clinical manifestations of LD remains to be elucidated.

The identification of LBs as the underlying cause of LD paved the way for the design of therapeutic strategies based on the prevention of glycogen accumulation. The first strategy tested for the degradation of LBs in vivo was the introduction, via intracerebroventricular injection (ICV), of fusion antibodies containing pancreatic human alpha-amylase into young laforin-deficient mice. Remarkably, alpha-amylase is capable of degrading the glycogen aggregates and restoring the metabolomics profile of the brain (Brewer et al., 2019).

Targeting MGS to prevent glycogen accumulation emerged from the demonstration that the suppression of MGS from conception impedes the appearance of LD (Pederson et al., 2013; Duran et al., 2014). However, LD is usually diagnosed with the onset of the first clinical symptoms, a stage of the disease at which LBs have already accumulated in the brain and other tissues. It was not clear whether the suppression of MGS expression would be effective for the treatment of the disease in these conditions. To assess the effectiveness of MGS suppression in a condition with preexisting LBs, conditional models for the genetic suppression of MGS were combined with either malin- or laforin-deficient mice (Nitschke et al., 2020; Varea et al., 2021). In malin^KO mice, MGS suppression was induced at two stages of LD in which LBs are already present: an early stage and a more advanced stage of the disease. An advanced stage of LD in a mouse model would correspond to a patient who has been diagnosed with LD and already shows clinical manifestations. Both groups of mice were analyzed for the presence of LBs, as well as for signs of neuroinflammation. These experiments revealed that early suppression of MGS arrests the accumulation of LBs in the brain, both in neurons and astrocytes. However, the neuroinflammatory response persisted. Furthermore, late MGS suppression was far less effective in halting the accumulation of LBs. Equivalent results were obtained when suppressing MGS in a laforin-deficient mice (Nitschke et al., 2020).

Recently, other approaches aiming to obtain a high degree of MGS suppression have been developed. These include the ICV delivery of an adenovirus vector-based (AAV)-Crispr/Cas9 technology to neonatal LD mice (Gumusgoz et al., 2021), and the use of antisense oligonucleotides (ASOs) targeting MGS in laforin-deficient mice (Ahonen et al., 2021). Both approaches have offered promising results for their future application in LD patients, providing an amelioration of the inflammation when applied at early stages of the disease.

As a single-gene disorder, LD could potentially benefit from a gene replacement therapy–based approach in which a normal copy of the defective gene (laforin or malin) is introduced. LD patients present mutations in laforin or malin (Chan et al., 2003; Gentry et al., 2005). Among the approximately 60 malin mutations described in LD patients, the most abundant are malin missense mutations and malin loss-of function mutations that affect its interaction with laforin (Gentry et al., 2005). The use of malin restoration as a gene replacement therapeutic approach for LD has recently been validated by generating malin-deficient mice with inducible expression of malin (Varea et al., 2022). Malin restoration was induced in malin^KO mice at two distinct stages of LD, namely an early point corresponding to a stage before the detection of neuroinflammation, and a later point after the appearance of neuroinflammation. Interestingly, the introduction of malin did not trigger the degradation of preexisting glycogen aggregates in the cortex or the hippocampus but resulted in the degradation of some of the protein components of the preexisting LBs, namely MGS and laforin (Fig. 53–1) (Varea et al., 2022). Both proteins are known targets for malin ubiquitination (Gentry et al., 2005; Vilchez et al., 2007), but it was unknown whether they would be accessible to malin when they are located at the LBs. In contrast to MGS and laforin, p62 remained unaltered and attached to the aberrant glycogen at the LBs (Varea et al., 2022).

The effectiveness of malin restoration was also tested at an advanced stage of the disease where there was massive accumulation of LBs and profound neuroinflammation. These experiments evidenced that malin restoration arrests the accumulation of LBs even at advanced stages. This effect was the result of changes in both neurons and astrocytes. Most importantly, malin restoration promoted a reduction of the neuroinflammatory response, observed by both reduced microgliosis and a lower expression of inflammatory markers (Varea et al., 2022). This result implies that, for the first time, a therapeutic strategy for LD shows beneficial effects even when started at advanced stages of the disease (Fig. 53–2). Further studies are needed to elucidate whether malin introduction has a beneficial effect on susceptibility to epilepsy in these mice. In this regard, our previous results indicate that strategies to decrease neuronal LB accumulation during the progression of the disease could be sufficient to reduce susceptibility to epilepsy (Duran et al., 2014, 2021).

From the results obtained, the following mechanism can be proposed: when restored, malin drives the degradation of MGS and laforin (Varea et al., 2022) and probably other LB components, such as PTG, by promoting their ubiquitination. This effect per se could lower the cellular toxicity caused by these aggregated proteins (Lovestone and McLoughlin, 2002; Stefani and Dobson, 2003). Additionally, malin prevents the formation of new LBs both in neurons and astrocytes (Varea et al., 2022). In all, malin restoration ameliorates the cellular homeostasis and reduces the inflammatory response (Fig. 53–3). This is the first time that a therapeutic approach has proven effective at an advanced stage of LD, thus offering the opportunity of applying this strategy to patients that already have clinical manifestations.

Gene therapy has become a promising therapeutic tool for many neurodegenerative diseases (Zhao et al., 2016; Rosenberg et al., 2018; Ingusci et al., 2019; Rodrigues et al., 2019; Martier and Konstantinova, 2020; Amado and Davidson, 2021). In particular, gene replacement has been approved for other neurodegenerative conditions such as spinal muscular atrophy (SMA) due to its proven beneficial effect in clinical trials (Mendell et al., 2021). It is important to note that the strategies presented in this chapter were described as proof-of-concept experiments and were performed using genetically modified mouse models. In the mouse models presented in this chapter, both the suppression of MGS and the induction of malin expression were achieved by the activation of a tamoxifen-dependent Cre/loxP system. Following tamoxifen administration, the deletion of the required fragment of DNA located at the inserted cassette (specific exons of Gys1 to prevent MGS expression, or a stop codon located between the promoter and the malin gene to activate malin expression) was induced. The application of either of these strategies to LD patients will require the external introduction of a vector containing the molecules of interest for reducing MGS expression in the first case, or for restoring malin expression in the second. This implies that more work is needed before any of these approaches can be used on patients. In this context, it is therefore particularly important to channel efforts into the development of molecular tools to target the brain in a specific and efficient manner. As shown in this chapter, the dysfunction of both neurons and astrocytes contributes to the progression of LD. The targeting of both cell types is therefore necessary to obtain maximal efficiency of the strategy applied. Gene therapy has seen rapid progress, thanks to the improvement of viral vector-based approaches, as well as the introduction of naked material, such as ASOs. An advantage of a classical gene-therapy approach involving the replacement of malin or viral-vector-based MGS suppression over the use of ASOs is that, in the latter, periodic injections are required, in contrast to a single intervention for a viral-based approach. The current viral vectors, mainly adeno-associated viruses or AAVs, are designed to reduce the immune response to their introduction, to target specific cells, and to offer a long response by remaining in dividing and nondividing cells for long periods (Choi et al., 2014; Rabinowitz et al., 2019; Ronzitti et al., 2020; Shirley et al., 2020). For the introduction of genetic material into the brain, viral vectors with the capacity to cross the blood–brain barrier have been designed (Gray et al., 2010; Bourdenx et al., 2014; Bors and Erdő, 2019). For direct brain delivery, an intracerebroventricular injection would be recommendable to target inner structures such as the midbrain and limbic structures (Castle et al., 2018). The malin gene is single-exon with a small locus size, which offers an advantage compared to most genes studied for gene therapy, since this opens the door to the introduction of the whole malin gene locus—containing the native promoter and all the regulatory elements—in the vector. By introducing these elements, a physiological regulation of malin expression in the targeted cells would be achieved, and safety would therefore be enhanced. Future work will explore the possibility of using laforin for a gene replacement strategy for laforin-deficient patients and will also help to clarify the mechanisms by which laforin impedes the appearance of the disease.

The accumulation of abnormal glycogen in the form of LBs in neurons and astrocytes underlies the pathophysiology of LD. Therapeutic approaches targeting MGS or based on malin/laforin restoration to halt the accumulation of glycogen are promising tools to stop the progression of LD. Both strategies are effective even when LBs have already accumulated, but malin restoration has proven significantly more effective in halting the progression of the disease at advanced stages. Malin gene replacement restores the physiological functions of malin, acting on its targets even when they are attached to the aberrant glycogen at the LBs. Further work will clarify the mechanisms of the initiation and progression of the disease, and this knowledge will be highly relevant for obtaining the best outcome from the application of any of these possible interventions. The observation that preexisting glycogen is maintained even after the introduction of malin indicates the difficulty of degrading existing aberrant insoluble glycogen in tissues and possibly the need for a combined therapeutic approach to fully degrade all the components of LBs.

The results shown in this chapter highlight the importance of early diagnosis of the disease, since early treatments are far more effective than later ones. The LD inheritance pattern allows the establishment of genetic screening for asymptomatic siblings. This type of preclinical diagnostic approach will be crucial for the early application of these strategies, which will help to slow down the progression of the disease and, conceivably, prevent it.

IRB Barcelona is the recipient of a Severo Ochoa Award of Excellence from MINECO (Government of Spain). This study was supported by grants from MINECO (BFU2017-84345-P to JD and JJG, and PID2020-118699GB-I00 to JD) and a grant from the National Institutes of Health (NIH-NINDS) (P01 NS097197) to JJG.

The authors declare no relevant conflicts of interest.