Progressive myoclonus epilepsies (PMEs) are diseases that afflict previously normal children with ever-worsening and soon-intractable myoclonus and epilepsy, usually associated with neurodegeneration, and eventual dementia and early death. PMEs are disorders of gray matter, and by and large neuroimaging reveals cerebral and cerebellar atrophy and no white matter disease. Autosomal recessively inherited PMEs may be divided pathogenetically into two main categories: non–lysosome-related, Lafora disease; and lysosome-related, Unverricht-Lundborg disease, the action myoclonus-renal failure syndrome, and forms of neuronal ceroid lipofuscinosis, sialidosis, and Gaucher disease (Shahwan et al., 2005; Lohi et al., 2006). Autosomal dominant PME is seen as part of dentatorubropallidoluysan atrophy (DRPLA) in patients with the largest expansions of the trinucleotide repeat underlying this disease (Ikeuchi et al., 1995). Finally, PME also occurs in various forms of mitochondrial cytopathies, especially in myoclonic epilepsy with ragged-red fibers (MERRF) (Fukuhara et al., 1980). In this review we summarize current molecular knowledge in PME, restricting to the autosomal recessive forms. The reader may refer to other work for updates on DRPLA and MERRF (DiMauro, 2004; Yamada et al., 2006).
Autosomal recessively inherited progressive myoclonus epilepsies (PMEs) include Lafora disease, Unverricht-Lundborg disease, the neuronal ceroid lipofuscinoses, type I sialidosis (cherry-red spot myoclonus), action myoclonus–renal failure syndrome, and type III Gaucher disease. Almost all the autosomal recessively inherited PMEs are lysosomal diseases, with the exception of Lafora disease in which neither the accumulating material nor the gene products are in lysosomes. Progress in identifying the causative defects of PME is near-complete. Much work lies ahead to resolve the pathobiology and neurophysiology of this group of devastating disorders.
Lafora Disease (LD)
First symptoms of Lafora disease (LD) appear between ages 8 and 18 years. There is an insidious near-simultaneous or closely consecutive appearance of headaches, difficulties in school work, myoclonic jerks, generalized seizures, and, in many cases, visual hallucinations of both epileptic and psychotic origin. The myoclonus, seizures, and hallucinations gradually worsen and become intractable. For many years, the patient struggles to maintain normal contact and communication, interrupted by extremely frequent myoclonic absence seizures. Gradually, dementia sets in, and by 10 years after onset the patient is in near-continuous myoclonus with absences, frequent generalized seizures, and profound dementia or vegetative state (Minassian, 2001; Striano et al., 2008). A distinctive pathology characterizes LD. Cells of various types exhibit dense accumulations of malformed and insoluble glycogen molecules, termed polyglucosans, differing from normal glycogen in lacking the symmetric branching that allows glycogen to be soluble. These polyglucosan accumulations are called Lafora bodies and in the central nervous system are present profusely in all brain regions and in the majority of neurons, specifically in their cell bodies and dendrites (Minassian, 2001; Striano et al., 2008).
LD is caused by mutations in either the EPM2A or EPM2B gene, encoding the interacting proteins laforin and malin (Minassian et al., 1998; Chan et al., 2003; Gentry et al., 2005; Ianzano et al., 2005). Laforin possesses a carbohydrate-binding domain with which it binds glycogen or polyglucosans and a dual-specificity phoshatase domain (Ganesh et al., 2000; Minassian et al., 2000, 2001; Wang et al., 2002; Chan et al., 2004; Wang & Roach, 2004). Malin is an E3 ubiquitin ligase, the specific ubiquitination target(s) of which remain unclear, although several have been suggested including glycogen synthase, glycogen debranching enzyme, PTG (a protein that targets the glycogen synthase activator phosphatase, PP1, to the glycogen molecule), and laforin (Gentry et al., 2005; Lohi et al., 2005; Cheng et al., 2007; Vilchez et al., 2007; Solaz-Fuster et al., 2008; Worby et al., 2008).
Polyglucosans, as poorly branched forms of glycogen, could arise from imbalance between the enzymatic activities of glycogen synthase, the enzyme that elongates glycogen strands, and branching enzyme, the enzyme that cleaves off the elongated parts and places them in internal branch points. Normally, the balanced activities between these two enzymes allow the glycogen particle to grow spherically instead of linearly, and thus allow it to remain soluble. Either overactive synthase or underactive branching could result in improperly branched glycogen, that is, polyglucosans. However, underactive branching cannot be implicated in LD, because branching enzyme deficiency causes a separate non-PME disease, type IV glycogenosis, in which polyglucosans form in all organs, including in brain, but in the latter in axons rather than in the somatodendritic compartment (Robitaille et al., 1980; Lossos et al., 1998). Overactive glycogen synthase appears to at least partly underlie the generation of polyglucosans in LD. It has been reported that the laforin–malin complex normally acts on PTG, the glycogen synthase activator, and on glycogen synthase itself, to downregulate glycogen synthase, via PTG and directly. In the absence of laforin or malin, glycogen synthase would be overactive and the balance between it and branching enzyme would be disturbed in favor of glycogen synthase and thus toward forming excessively long and insufficiently branched glycogen strands, that is, polyglucosans (Lohi et al., 2005; Vilchez et al., 2007; Solaz-Fuster et al., 2008; Worby et al., 2008). In a separate line of investigations it was shown that laforin dephosphorylates glycogen, and that glycogen in LD mouse models is hyperphosphorylated. What the role of phosphate is on normal glycogen is unknown, and how glycogen hyperphosphorylation in Lafora disease relates to the generation of polyglucosans and Lafora bodies is under investigation (Wang & Roach, 2004; Wang et al., 2006; Worby et al., 2006; Gentry et al., 2007; Tagliabracci et al., 2007; Wang et al., 2007).
Unverricht-Lundborg Disease (ULD)
Onset of Unverricht-Lundborg disease (ULD) is between 6 and 13 years of age. ULD diverges from other PMEs in that it appears to be progressive only in adolescence, with dramatic worsening of myoclonus and ataxia in the first 6 years after onset. Seizures usually are responsive to medication during this period, but myoclonus is not. By early adulthood, the disease stabilizes and myoclonus and ataxia may even improve. Seizures remain well-controlled and there is minimal to no cognitive decline. Adulthood is characterized by ongoing, sometimes severe, but no longer progressive myoclonus and ataxia, near-normal mental abilities, medically controlled epilepsy, and a possible normal lifespan. Pathology in ULD is unremarkable (Magaudda et al., 2006; Chew et al., 2008; Kalviainen et al., 2008; Santoshkumar et al., 2008).
The major gene involved in ULD (EPM1) has been identified as CSTB on chromosome 21, which encodes cystatin B, an inhibitor of cysteine proteases (Pennacchio et al., 1996). Almost all patients with ULD have mutations in this gene, although a second locus was recently mapped to a 15-Mb region on chromosome 12 (EPM1B) (Berkovic et al., 2005). Among CSTB patients, there are 10 identified mutations, with an unstable expanded polymorphic dodecamer repeat in the promoter region of CSTB accounting for greater than 90% of mutated alleles (Joensuu et al., 2008). Practically every ULD case has the dodecamer repeat expansion mutation in at least one of their alleles. This mutation leads to dramatic reduction in CSTB mRNA. However, a small amount of mRNA (∼10%) still remains, indicating that ULD is the phenotypic outcome not of complete absence but instead of vastly reduced CSTB (Lafreniere et al., 1997; Virtaneva et al., 1997; Lalioti et al., 1999; Larson et al., 1999).
Cystatin B localizes in the cytosol, the nucleus, and the outer surface of lysosomes (Alakurtti et al., 2005). It interacts with and inhibits papain-family lysosomal cysteine proteases cathepsins B, L, S, and H (Turk & Bode, 1991; Estrada et al., 1998; Rinne et al., 2002). These findings suggest a neuroprotective role for cystatin B, quenching excess lysosomal hydrolases that leach into the cytoplasm during times of cellular stress. However, Cstb knockout mice, which nicely replicate the ULD PME phenotype (Pennacchio et al., 1998; Shannon et al., 2002; Vaarmann et al., 2006; Franceschetti et al., 2007), when crossed with mice deficient of cathepsins B, L, or S, showed no improvement in the ataxia or the PME, although there was significant reduction in cerebellar granule cell apoptosis in one of the crosses (Cstb knockout mice lacking cathepsin B) (Houseweart et al., 2003). These results show that the theory that the clinically relevant function of cystatin B in the regulation of cathepsins is inadequate to explain the disease. Thirteen years since the identification of the causative defect in ULD, pathogenesis of this common PME remains mostly not understood.
The Neuronal Ceroid Lipofuscinoses (NCLs)
There are at present eight genetic forms of neuronal ceroid lipofuscinoses (NCLs) (CLN1, CLN2, CLN3, CLN5, CLN6, CLN7, CLN8, and CLN10). CLN4 is reserved for the elusive and possibly nonexistent adult-onset Kuf’s disease (see below). CLN9 is from a patient cell line, which appears not to have mutation in the known genetic forms (Schulz et al., 2004).
The NCLs are PMEs, because they affect previously normal children, are progressive, and include worsening myoclonus during these children’s short lives. They are grouped together on pathologic grounds due to the common presence of neuronal and extraneural autofluorescent pigment accumulations. Under the electron microscope, the accumulated material takes three different forms: granular osmiophilic deposits (GRODs), curvilinear profiles, and fingerprint bodies. The form that predominates in a particular patient correlates with the particular mutated gene. For example, in CLN1 only GRODs are seen and in CLN2 curvilinear profiles predominate, whereas CLN3 is distinguished by a preponderance of fingerprint bodies.
Clinical progression varies among the different NCLs. For example in CLN2 an aggressive intractable myoclonic epilepsy occurs early in the disease course, even at onset, whereas in CLN3 the myoclonus is mild and appears late in the disease. Conversely, visual loss and blindness are early features of CLN3 and mild and late features of CLN2. Both diseases have a comparable progression of dementia.
First symptoms appear between 8 and 18 months of age with irritability, and the baby becomes difficult to comfort. This is followed by rapid psychomotor deterioration, central hypotonia, and deceleration of head growth. Myoclonic jerks and other seizures are present, and blindness occurs, with optic atrophy. Hand-wringing often develops during the disease course, which, along with the slowing of head growth and developmental regression, raises the relatively optimistic differential of Rett syndrome, but unlike the latter, CLN1 does not stabilize, continuing instead to deteriorate until death in early childhood (Mole et al., 2005).
Of all the NCLs, CLN1 has the widest range of ages of onset, depending on the particular mutation type in the CLN1 gene. Although the majority of patients have infantile onset as described previously, some have late-infantile, juvenile, and even adult onset as late as 40 years of age (Wisniewski et al., 2001). It is possible that most if not all Kuf’s disease cases are actually adult onset CLN1. Whichever the age of onset, all CLN1 patients will have GRODs as their electron microscopic manifestation.
CLN1 encodes a palmitoyl protein thioesterase (PPT1), a lysosomal enzyme that removes palmitate residues from proteins (Vesa et al., 1995).The latter accumulate in CLN1 to result in the GRODs. Palmitoylation plays a critical role in neuronal vesicular transport and intracellular signaling (Greaves & Chamberlain, 2007), and in neurons PPT1 is found in nonlysosomal compartments in presynaptic terminals (Ahtiainen et al., 2006). These findings suggest that CLN1 is not exclusively due to the abnormal storage.
CLN2 mutations present a late infantile disease onset between 2 and 4 years of age marked by particularly severe early myoclonus and seizures. Gross and fine motor abilities, cognitive functions, and, later, vision are rapidly lost within 3 years of onset. Spasticity, truncal hypotonia, loss of head control, near-continuous myoclonus, frequent seizures, and an extended vegetative state characterize the rest of the child’s life until death in early adolescence. Death is often due to aspiration pneumonia. Electroencephalography (EEG) includes characteristic occipital spike responses to slow flash (1–2 Hz) stimulation before the onset of seizures, which exaggerates as the disease progresses. Electroretinography is diminished even before noticeable visual loss (Wisniewski et al., 1998; Goebel & Wisniewski, 2004).
Presence of pure curvilinear membrane-bound lysosomal aggregates without clear lipid droplets is the hallmark for CLN2 mutation disease. A few cases have presented late (∼8 years), exhibiting slow regression with death as late as 40 years of age (Sleat et al., 1999).
CLN2 encodes the lysosomal enzyme tripeptidyl peptidase (TPP1), a member of the serine carboxyl proteinase family (Rawlings & Barrett, 1999).
This group of enzymes removes tripeptides from N-termini of small proteins such as the subunit c of mitochondrial ATP synthase. In addition to the lysosome, TPP1 localizes to the Golgi apparatus, lipid rafts, and endosomes, and interacts with CLN5, CLN3, and CLN8 (Mole et al., 2005; Persaud-Sawin et al., 2007). Again, it appears there is more that meets the eye than substrate accumulation in the pathogenesis of this NCL.
CLN3 mutations cause juvenile NCL (onset between 5 and 10 years of age). Many cases, mostly northern European, are due to a common ancestral 1-kb deletion mutation (Munroe et al., 1997). Loss of vision and intellectual deterioration occur rapidly, followed by seizures and psychosis later. Ocular pathology is initially a pigmentary retinopathy often misdiagnosed as retinitis pigmentosa or cone dystrophy. In adolescence, speech, mobility, and cognitive skills deteriorate and seizures set in. Children have behavioral problems such as aggressiveness, psychosis, mood disturbance, and anxiety. Speech becomes dysarthric and dysfluent with echolalia. As the disease progresses, myoclonic jerks and parkinsonian features and gait develop.
Ultrastructurally, CLN3 cases exhibit fingerprint profiles. These could appear pure within the lysosomal residual body, or in conjunction with curvilinear or rectilinear profiles, or as a small component within large membrane-bound lysosomal vacuoles. This particular NCL is diagnosable through the identification of vacuolated blood lymphocytes.
CLN3 encodes a transmembrane protein that has been reported to localize to membrane lipid rafts in lysosomes, endosomes, synaptosomes, and the cell membrane, as well as in mitochondria (Phillips et al., 2005). Numerous roles have been attributed to CLN3, and much work is needed to reconcile these disparate functions. In mitochondria, CLN3 was suggested to aid the processing of mitochondrial membrane proteins such as the ATPase subunit c, which accumulates in this as synaptic vesicle transport (Margraf et al., 1999). CLN3 has been implicated in the regulation of lysosomal pH, transport of basic amino acids into the lysosome, and lysosomal size (Golabek et al., 2000; Holopainen et al., 2001; Ramirez-Montealegre & Pearce, 2005). An antiapoptotic role has been ascribed to CLN3; its C-terminus appears to participate in cell cycle regulation, and mutations of this terminus result in slow growth and increased apoptosis (Puranam et al., 1999). Cln3 knockout mice show neutralizing antibodies against glutamic acid decarboxylase (GAD65), suggesting that an autoimmune response against GAD65 might contribute to preferential loss of GABAergic neurons in this disease. However, it is not understood if these autoantibodies contribute to the pathogenesis or if they are secondary entities during neurodegeneration, although their presence may influence excitotoxic mechanisms (Chattopadhyay et al., 2002). A second mouse model suggested that CLN3 plays an active role during early embryogenesis (Eliason et al., 2007). Recently, an enzymatic function has been found in CLN3, namely palmitoyl-protein Δ-9 desaturase activity (Narayan et al., 2006). Another group showed a correlation between CLN3 expression and the synthesis of bis monoacylglycerol phosphate (BMP) and suggested that CLN3 may play a role in the biosynthesis of BMP (Hobert & Dawson, 2007).
Mutations in CLN5 cause the Finnish variant of late-infantile NCL (onset between four and 7 years, i.e., slightly older than the range in CLN2). Characteristic clinical symptoms include developmental regression, visual impairment, ataxia, and myoclonus epilepsy, similar to the other NCLs. Neurophysiologic examination shows giant visual evoked potentials, exaggerated somatosensory potentials, and occipital spikes in response to photic stimulation, similar to CLN2. Lipopigments are distributed in the central nervous system and extracerebrally, and include fingerprint bodies, curvilinear profiles, lamellar inclusions, and occasionally condensed fingerprint figures associated with lipid droplets.
CLN5 encodes a transmembrane protein and is synthesized as four precursors, all directed to the lysosome. The longest form is associated with the lysosomal membrane and interacts with CLN2 and CLN3 (Vesa et al., 1995). Similar to CLN3-defective fibroblasts, CLN5-deficient fibroblasts also exhibit elevated intralysosomal pH (Kyttala et al., 2006).
Whereas CLN5 is predominately a Finnish disease, CLN6 has been found in multiple different ethnicities, similar to CLN1, CLN2, and CLN3. Age of onset of CLN6 straddles the ages of onset of CLN1, CLN2, and CLN3 ranging from 18 months to 8 years, with the majority between 3 and 5 years. Early features include gait and speech disturbance, seizures, and developmental delay. Lipopigments include fingerprint, curvilinear, as well as rectilinear patterns. CLN6 encodes a protein of unknown function with seven transmembrane domains localizing to the endoplasmic reticulum (ER) (Heine et al., 2004; Mole et al., 2004).
In this NCL, onset is between 2 and 7 years. Psychomotor regression or seizures can be the initial presenting signs. This form, like CLN3, can be diagnosed by pathologic analysis of peripheral lymphocytes. Unlike CLN3 though, where vacuolation is seen, in CLN7 dense fingerprint profiles are present in the lymphocytes. The CLN7 gene product is a lysosomal integral membrane protein with features suggestive of transport function (Siintola et al., 2007).
Depending on the mutation, CLN8 has presented with childhood-onset (5–10 years) intractable epilepsy followed by progressive cognitive decline (Ranta et al., 1999) or mild developmental delay in infancy followed by a florid PME with progressive myoclonus, seizures, retinopathy, and psychomotor regression starting between 3 and 6 years (Striano et al., 2007). GROD, curvilinear, and fingerprint profiles have been reported on electron microscopy, in various tissues, including lymphocytes (Striano et al., 2007). The CLN8 protein appears to localize to the ER and the ER–Golgi intermediate compartment (Lonka et al., 2000), and its likely function is as an enzyme in the pathway of ceramide synthesis (Winter & Ponting, 2002).
Here, complete loss of function mutation leads to a congenital form of NCL with encephalopathy, status epilepticus, and death due to respiratory insufficiency in the first days and weeks of life with massive GRODs in the central nervous system (Siintola et al., 2006). Missense mutations, so far in one patient, have caused a childhood-onset neurodegenerative disease with ataxia, retinopathy, severe cognitive decline, and apparently no seizures at age 17. The pathologic correlate was GRODs (Steinfeld et al., 2006). The mutated gene encodes cathepsin D, a major lysosomal protease (Siintola et al., 2006; Steinfeld et al., 2006).
Type II sialidosis is a severe infantile-onset disease associated with bony deformities, dysmorphism, myoclonus, cherry-red spot in the fundus, and early lethality. Type I is a typical PME with a wide range of age of onset anywhere from childhood to adulthood. Ataxia is prominent, and presence of cherry-red spot on funduscopy is highly indicative of the disease in the context of a clinical presentation with PME. The type and severity of disease depends on the particular mutation in the NEU1 gene encoding the lysosomal neuraminidase, an enzyme that removes sialic acid from various macromolecules in the lysosome. The disease can be diagnosed by the presence of sialo-oligosaccharides in urine (Bonten et al., 2000).
Action Myoclonus–Renal Failure Syndrome (AMRF)
This disorder may present with proteinuria and glomerulosclerosis as early as 9 years of age. The neurologic syndrome does not manifest until early adulthood, after age 17 but before age 25. In a number of cases the proteinuria is not noted until the neurologic disease appears. The latter begins with bilateral hand tremor worsened with activity. Soon this progresses to include myoclonus that is highly inducible by a wide range of stimuli, highly emotionally enhanced, and rapidly progressive and debilitating. Seizures also make their entry into the course but are relatively controllable. Dementia is minimal to not present in most patients, but ataxia is pronounced. The renal syndrome (collapsing glomerulopathy) is also progressive and, if not treated with transplantation, fatal. The neurologic phenotype is not connected to the renal failure, as it progresses unaltered after renal function is corrected by transplantation (Badhwar et al., 2004). Pathology reveals some accumulation of autofluorescent material seemingly not in neurons, perhaps in astrocytes, in contradistinction to the NCL where the material is very much neuronal (Badhwar et al., 2004; Berkovic et al., 2008).
The AMRF gene is SCARB2/LIMP2, which encodes a lysosomal membrane protein (Balreira et al., 2008; Berkovic et al., 2008). One important function of this protein appears to be to escort glucocerebrosidase, the enzyme defective in Gaucher disease, from the ER to the lysosome (Reczek et al., 2007) (glucocerebrosidase is one lysosomal enzyme that is not transported to the lysosome via the canonical mannose 6-phosphate pathway). However, the glucocerebrosidase transport function of SCARB2/LIMP2 is not sufficient to explain AMRF. Additional functions have been proposed, such as a role in the biogenesis of lysosomes and endosomes (Eskelinen et al., 2003), which await further elucidation.
Gaucher disease is characterized by hepatosplenomegaly, anemia, thrombocytopenia, bone pain, and other systemic features. When the central nervous system is involved, the disease is classified as type II (early onset and severe) or type III (late onset and slowly progressive). The latter has been further subdivided. Patients with types IIIB and IIIC have aggressive systemic Gaucher and relatively mild neurologic features that do not include myoclonus. Patients with type IIIA have relatively mild systemic Gaucher and a PME (Patterson et al., 1993; Bohlega et al., 2000). Occasionally, the systemic disease in type IIIA is so mild that it is unrecognized for some time into the course of the PME, and the disease is confused for a pure PME such as LD (Filocamo et al., 2004).
At the cellular level, the divulgation of the recessively inherited PME genes has revealed that most are lysosomal disorders. Even the few NCLs where the protein product does not localize to the lysosome have abnormal lysosomal storage. The major exception is LD, where the accumulating polyglucosans in the brain are not membrane-bound and the two disease gene products, laforin and malin, are not lysosomal.
This survey highlights how far we have come in the characterization of these diseases and the understanding of their genetic cause. However, our near-complete knowledge of causation contrasts starkly with our very incomplete understanding of pathogenesis. We know our diseases well, and we know their causes. Now, we need to know how cause connects to disease, so that we may hope to intervene and help our patients.
This work was supported by the Canadian Institutes of Health Research. B. A. Minassian holds the Canada Research Chair in Paedriatic Neurogenetics.
We have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
Disclosure: The authors have no conflicts of interest to declare.