Lafora Disease: current biology and therapeutic approaches

Neurodegenerative diseases, ubiquitin biology and Lafora disease:

Abnormal inclusions in the central nervous system are hallmarks of various neurodegenerative disorders. These accumulations often perturb cellular physiology and homeostasis. Globally, the burden of neurological disorders continues to increase as populations are growing and ageing, demanding more research in the basic molecular mechanisms of neurodegenerative diseases associated with toxic inclusions [1].

Protein ubiquitination and related cellular pathways are frequently altered in neurodegenerative diseases [2–4]. Ubiquitin is a small protein (8 kD), which is covalently attached to other proteins via a process that requires three consecutive reactions. The first is activation of the ubiquitin molecule by a ubiquitin activating enzyme E1 through an ATP-dependent mechanism. This is followed by the transfer of the activated ubiquitin from E1, by a ubiquitin conjugating enzyme E2 to itself. Finally, a ubiquitin ligase enzyme, E3, covalently ligates the ubiquitin molecule to its protein substrate. Ubiquitination either targets a substrate to degradation by the ubiquitin proteasome system (UPS) or by autophagy, or serves to regulate the protein substrate in its cellular pathways, including immune signaling, endocytosis, DNA damage repair, cell cycle progression, mitophagy, autophagy and others [5, 6]. Because of this myriad functions of ubiquitin biology, it is not surprising that aberrations of ubiquitination have been implicated in multiple neurological disorders including but not limited to Alzheimer’s disease, Parkinson’s disease, Huntingtin’s disease and Angelman Syndrome [3, 7]. In a recent study, a manual database search found that among ~700 known and predicted human E3 ligases, 83 are mutated in different neurological disorders, highlighting the importance of this class of proteins in neurological disease [8].

In certain neurological disorders, the inclusions are not proteinaceous but glucan in nature, namely abnormal structured, insoluble forms of glycogen [9, 10]. In some, there is disturbance in ubiquitin biology [11, 12]. One such disease is Lafora Disease (LD), a teenage-onset intractable, fatal progressive myoclonus epilepsy [13]. In this article we review LD with a focus on ubiquitin biology. First, we briefly review clinical aspects of the disease. We next describe the unknowns about the E3 ubiquitin ligase gene mutated in the disease including its hypothetical role in glycogen metabolism and neurodegeneration. Finally, we discuss currently available and future treatment options for LD.

The clinical syndrome of Lafora disease

LD is inherited in autosomal recessive fashion. The majority of patients have a relatively uniform phenotype, with exceptional cases with missense mutations or certain splice site mutations having a later onset and protracted course. Typically, a period of decline in school performance in pre-teen years predates appearance between ages 10 and 15 years of overt symptoms, which are any or all of myoclonus, visual hallucinations or convulsive seizures [14]. Given the onset of myoclonus and seizures in teenage years, a tentative diagnosis of Juvenile Myoclonic Epilepsy (JME) is frequently made, until the EEG is obtained. The latter shows background slowing and generalized and occipital discharges of irregular conformations, indicating something more severe than JME, and leading to genetic testing [15]. All symptoms inexorably worsen in the subsequent years. Seizures become intractable to any and all combinations of anti-seizure drugs, and myoclonus becomes near constant. The myoclonus is associated with atypical absenses leading, in its constancy, the patient to become unable to complete thoughts or sentences. Dementia sets in, ambulation is lost, and eventually the patient becomes vegetative. Usually, the young patient succumbs to airway control problems and/or status epilepticus within ten years of onset [13, 14]. Thankfully, LD is rare, with a frequency not higher than 1/100,000 [16]. But as mentioned, it is a horrendous disease definitely needing better understanding and a treatment.

Malin is an E3 ubiquitin ligase deficient in approximately 50% of patients with LD:

In LD, long chained, lesser branched glycogen is formed and precipitated into insoluble polyglucosan bodies (PBs) (Lafora bodies; LBs, named after Gonzalo Lafora who first described them [17]), which likely drive the neurodegeneration and epilepsy (see below) [14, 18]. In the vast majority of patients the brain is the only clinically affected organ, even though LBs also form in other organs including skeletal muscle, heart, liver and skin [19, 20]. There have been exceedingly rare instances of cardiac arrhythmia and heart failure [21]. Typically, LD starts in otherwise healthy teenagers with myoclonus, visual and convulsive seizures, which over time that become frequent and intractable. This is associated with behavioral and cognitive decline; the patient ultimately reaches a vegetative state and dies within 10 years of onset [22]. The molecular mechanisms that underlie LD are not fully understood. However, recent research revealed unsuspected players in glycogen metabolism, namely the glycogen phosphatase laforin and its interacting protein partner, the ubiquitin E3 ligase malin.

Discovered in 2003 [23], the gene for malin, EPM2B (also known as NHLRC1), maps in chromosomal band 6p22.3 and consists of a single exon expressing an N-terminal RING (Really Interesting New Gene) domain and a C-terminal six repeats of NHL (protein domain present in Ncl-1, HT2A and Lin-41 protein) domains. The NHL domains presumably confer protein-protein interaction, and the RING domain the E3 ubiquitin ligase activity. There are over 80 LD causing mutations that have been identified in the EPM2B gene, affecting both RING and NHL domains (http://projects.tcag.ca/lafora/). The evolutionary lineage of the EPM2B gene is restricted to only vertebrate species and a branchiopod. This is in contrast to its interacting protein partner laforin, encoded by EPM2A gene, which has much broader appearance in both vertebrates, invertebrates, protozoa as well as in certain algae indicating possible malin-independent functions of laforin especially in lower class of organisms [24]. Approximately 50% of LD patients have mutations in the EPM2B gene, and the remainder in the EPM2A gene.

The malin-laforin complex may act in quality control of glycogen synthesis:

The role of malin in glycogen biology remains a knotty question, which we review starting with a focused summary of glycogen metabolism. Glycogen is the largest molecule in the cytoplasm and a critical energy store. Glucose chains 10 units or longer tend to precipitate, yet 55,000-unit glycogen is soluble, a requisite to its metabolism [25]. How this solubility is achieved was part of the foundational classical biochemical discoveries of the 20^th century and until recently considered settled [26]. Glycogen branching enzyme (GBE) activity proportionate to that of glycogen synthase (GS) was thought enough to generate molecules with adequate branching (and therefore adequately short branches) to allow molecule-through hydration and thus solubility. However, in both the laforin and malin knockout mouse models, the amount of soluble GS, its activation state and activity levels remain unaltered, indicating that there are additional mechanisms in the structuring of normal glycogen beyond just balanced activities between GS and GBE at the whole-cell level [27]. One of the fascinating findings in this aspect was the discovery that laforin is a glycogen phosphatase, consistent with the observation that LD glycogen is hyperphosphorylated [28–31]. However, subsequent studies showed that glycogen hyperphosphorylation, and therefore deficiency of laforin’s glycogen phosphatase function, are not at the root of abnormal glycogen (polyglucosan) formation in LD. Expression of a phosphatase-defective laforin in laforin knockout mice rescued the mice, including correcting the chain lengths of glycogen and preventing LB formation. This indicated that a laforin function other than its phosphatase is necessary to prevent LD. Expression of the same phosphatase-defective laforin in malin knockout mice did not rescue any of the above. Together these results indicated that laforin most likely acts via malin, and emphasized the critical role of malin’s ubiquitin E3 ligase activity in the regulation of glycogen structure [32, 33] [34–36]. [33].

A recent study shed on the native properties of soluble glycogen added further detail. Analyses of glycogen fractions from both laforin knockout and malin knockout mouse models showed that i) soluble glycogen is heterogeneous in branch chain length distribution, ii) molecules with longer chains have a propensity to precipitate, and iii) the insoluble glycogen fractions from both mouse models are enriched with molecules with the longest chains [27]. These observations highlighted the fact that glycogen branch lengths are key determinant to solubility, and that certain soluble glycogen molecules are at the risk of precipitation due to possessing an excess of longer chains. Laforin contains a carbohydrate binding motif (CBM) [31, 37] that preferentially binds longer, versus shorter, glucan chains [38]. Considering all above, one can speculate that the malin-laforin complex tethers to these longer, ‘at-risk’, soluble glycogen molecules through laforin’s CBM, and may act to prevent ‘at-risk’ molecules from proceeding to precipitate.

If it is postulated that the malin-laforin complex acts on these ‘at risk’ soluble glycogen molecule, how does the complex regulate solubility of glycogen? The answer may resides in finding the glycogen metabolism related protein substrate(s) for the malin E3 ubiquitin ligase. Numerous malin ubiquitination substrates have been proposed at least based on in-vitro overexpression experiments (reviewed in [12]). Among them GS and Protein Targeting to Glycogen (PTG) are notable because of their direct importance in glycogen biology [39–42]. Indeed, genetic downregulation of GS (described later in ‘Therapeutic approaches to LD’ of this review) activity corrected glycogen chain lengths and prevented polyglucosan formation in LD mouse models, indicating, unsurprisingly, that the latter is dependent on GS. Additionally, knocking out subunits of protein phosphatase 1α (which is a GS activator), namely the protein phosphatase 1 regulatory subunit 3c (PP1R3c, also known as PTG) or the protein phosphatase 1 regulatory subunit 3d (PP1R3d, also known as R6), reduced LB formation, emphasizing that the regulation of glycogen synthase activity and chain elongation are key to LD pathogenesis [43–46].

These and other results lead to the proposal of a mechanistic model for the malin-laforin complex as a quality control system for glycogen metabolism (Figure 1, adapted from [47]): The hypothesis is that the GBE to GS activities ratio is occasionally out of balance at localized sites of particular glycogen molecules, engendering overlong chains and resultant risk of precipitation for those specific molecules. An unknown kinase phosphorylates these long chains, preventing their intertwining (in amylopectin, phosphorylation separates double helices to allow hydration and access for chain-digesting enzymes [48]). The phosphorylated long chains attract laforin, and with it malin, which ubiquitinates and inhibits GS and/or GS activating enzymes locally. The pathogenic over-elongation stalled, the chain-digesting enzyme glycogen phosphorylase re-shortens the long chains but cannot proceed beyond the phosphate (a known feature of this enzyme [49]). Laforin removes the phosphate, allowing phosphorylase to complete the shortening/remodeling (in plants the phosphatase SEX4, orthologous to and complementable by laforin, removes added phosphates to permit continued chain shortening by plant phosphorylases/exoamylases [50, 51]). Finally, malin ubiquitinates and removes laforin. In the absence of laforin or malin, chain over-elongations are not checked, and the at-risk molecules precipitate and accumulate slowly over time. Overall tissue or cell level GS and GBE activities are normal (i.e. not measurably altered), because the causative GBE/GS imbalance is local at a very small fraction of total glycogen molecules at any one time.

LD at a crossroad of neuroinflammation, neurodegeneration and epilepsy:

Another major unknown in LD research is how the abnormal glycogen accumulation drives the neurodegenerative and epileptic phenotypes? Laforin and malin knockout mouse models show neurodegenerative phenotypes of neuronal loss and astrogliosis [45, 52–57]. The PBs are present in both astrocytes and neurons with distinctive characteristics [58, 59]. Neuronal PBs are spherical and large in size and present in the somatic region, whereas astrocytic PBs are smaller, found in clusters mainly in the astrocytic processes, and have resemblance to Corpora Amylacea (CA) of normal aged brains [59]. How the inclusion of two different types of PBs lead to neurodegeneration is not understood. However, one can assume that accumulation of these ‘unwanted’ toxic products in neurons and astrocytes may perturb the normal physiology of brain. The accumulating LBs may also be drivers of an inflammatory or autoimmune process. In fact, malin knockout brains show reactive astrocytes, microglia and high levels of proinflammatory markers compared to wild type [60]. Indeed, both laforin and malin knockout mouse brains overexpress neuroinflammatory genes compared to WT in progressive fashion with increasing age [60, 61]. In addition, PBs may affect particular cell types, which may explain the particularly severe epilepsy. In a recent study, when glycogen synthase was ablated specifically from astrocytes in malin knockout mice, the mice no longer showed PB accumulation, astrogliosis/microgliosis and autophagy impairment. However in these mice, the epileptic phenotype was not corrected. On the other hand, over-accumulation of glycogen in astrocytes led to a neurodegnerative phenotype [62]. These results reinforce the idea that astrocytic PB accumulation could be the driving force behind neurodegeneration, but perhaps neuronal pathology underlies the epilepsy component of the disease. PBs were noted in GABAergic interneurons of laforin knockout mice, associated with a particular diminution of the numbers of these inhibitory neurons. Also, these same studies revealed decreased levels of neuronal growth factors, which in turn may contribute to the neurodegenerative process [63]. Lastly, astrocytic glutamate transporter, GLT-1 which is responsible for removing excess of glutamate from synaptic cleft, was shown to have altered distribution with less glutamate uptake activity in cultured primary astrocytes from both laforin and malin knockout mouse models [64, 65]. Recently, the Sanz group reported a possible role of malin-laforin complex in ubiquitination-mediated delay for the endocytosis of GLT-1 receptor in a cell culture-based system suggesting the importance of the complex in GLT-turnover [66]. Indeed, as shown previously, altering the activity of either GLT-1 or blocking GABA mediated neurotransmission by pharmacological interventions in malin knockout mice resulted in significantly increased glutamate levels in the brain suggesting a plausible role of glutamate homeostasis in the epileptic phenotype of LD [65].

Possible role of the malin-laforin complex in ubiquitination mediated glycophagy:

Several observations suggest a role for defective autophagy in LD, including: i) presence of ubiquitin positive proteinaceous materials in LBs [67]; ii) in-vitro interaction of malin-laforin complex with autophagy related PI3KC3 complex [68] and adaptor protein P62 [69]; iii) most often, in in-vitro conditions, malin-laforin complex ubiquitinates protein substrates with K-63 linked ubiquitin linkage – a type of ubiquitination that strictly associates with autophagy (reviewed in [12]) iv) a decrease in proteosomal activity and activation of autophagy related mTOR signaling pathway in laforin knockout mouse model [70], and lastly v) impairment of autophagosome formation in malin knockout mice [67]. Furthermore, it has been observed that autophagy related pathways such as mitophagy [71], lysosomal pathway [72], endoplasmic reticulum stress [73–75] and reactive oxygen species generation are all affected in LD [76]. However, the precise connection between autophagy and LD pathology remains much debated with contrasting results from LD mouse model tissues where several of the tested autophagy markers remain unaltered [33, 77]. Nevertheless, it could be a possibility that autophagy is a secondary effect of LB formation where malin-laforin complex tethers K-63 linked ubiquitin chains onto their protein substrates attached to LBs and targets it to the autophagosome for degradation. Some interesting findings substantiate this hypothesis, such as: i) striking similarities between CA from aged brain and LBs in terms of their shape, size and low phosphate content [20] ii) CA and LBs both being decorated with ubiquitin and P62 [58, 78] and most intriguingly iii) presence of a neo-epitopes on CA that could be recognized and cleared by the body’s natural immune system [58, 79]. Overall, the precise connection between LD and autophagy is yet to be clarified.

Therapeutic approaches to LD:

The current basis of therapy for patients with LD is to control the severity and frequency of seizures and myoclonus. Anti-epileptic drugs are the only available treatments [13, 80]. Despite some advancement in understanding the disease mechanism, patients are still lacking targeted or curative therapy for the disease [13, 81].

Gene therapy offers much hope for debilitating genetic disorders and has considerably advanced the development of treatments for hereditary diseases [82–84]. The limited number of genes (EPM2A or EPM2B) involved in LD makes the disease a good candidate for gene replacement therapies. The functional copy of the mutated gene can be delivered to compensate for the deficiency. In addition to the gene replacement therapies, downregulation of glycogen synthesis by targeting GS at the DNA, RNA, or protein level and degradation of LBs are also being investigated as possible therapeutic options for LD (Figure 2). Moreover, other currently available dietary modifications could be used as an addition to these more targeted approaches. In a recent paper, Israelian et al. showed that the ketogenic diet can reduce abnormal glycogen accumulation in a mouse model of LD [85]. They suggested initiation of the diet at the diagnosis of LD, preferably through an internationally organized clinical trial to clarify the particular role of the diet in patients.

Conclusion:

LD is an orphan disease with much unknown basic mechanism. Therefore, the disease presents unique opportunities for basic science researchers to study diverse mechanistic pathways including glycogen biology, protein ubiquitination, neuroinflammation, neurodegeneration and epilepsy. These studies will uncover important unknown mechanisms in the brain, and also beyond. For example, LD related work recently disclosed a role for malin in small cell lung cancer [130, 131]. Finally, LD is ripe for therapy and rife with possible therapeutic approaches, including disease gene replacement, but also intervention in the already uncovered biochemical disease pathway. Advances in this or other rare diseases also lays the groundwork to therapies of more common and complex diseases of the brain.