Epilepsy in Rett syndrome, and CDKL5- and FOXG1-gene–related encephalopathies


  • Renzo Guerrini,

    1. Pediatric Neurology Unit and Laboratories, Children’s Hospital A. Meyer-University of Florence, Florence, Italy
    2. Research Institute IRCCS Stella Maris Foundation, Pisa, Italy
    Search for more papers by this author
  • Elena Parrini

    1. Pediatric Neurology Unit and Laboratories, Children’s Hospital A. Meyer-University of Florence, Florence, Italy
    Search for more papers by this author

Address correspondence to Renzo Guerrini, Pediatric Neurology Unit and Laboratories, Children’s Hospital A. Meyer-University of Florence, Viale Pieraccini 24, 50139 Florence, Italy. E-mail: r.guerrini@meyer.it


Rett syndrome is an X-linked neurodevelopmental disorder that manifests in early childhood with developmental stagnation, and loss of spoken language and hand use, with the development of distinctive hand stereotypies, severe cognitive impairment, and autistic features. About 60% of patients have epilepsy. Seizure onset before the age of 3 years is unlikely, and onset after age 20 is rare. Diagnosis of Rett syndrome is based on key clinical elements that identify “typical” Rett syndrome but also “variant” or “atypical” forms. Diagnostic criteria have been modified only slightly over time, even after discovering that MECP2 gene alterations are present in >90% of patients with typical Rett syndrome but only in 50–70% of atypical cases. Over the last several years, intragenic or genomic alterations of the CDKL5 and FOXG1 genes have been associated with severe cognitive impairment, early onset epilepsy and, often, dyskinetic movement disorders, which have variably been defined as Rett variants. It is now clearly emerging that epilepsy has distinctive characteristics in typical Rett syndrome and in the different syndromes caused by CDKL5 and FOXG1 gene alterations. The progressive parting of CDKL5- and FOXG1-gene–related encephalopathies from the core Rett syndrome is reflected by the effort to produce clearer diagnostic criteria for typical and atypical Rett syndrome. Efforts to characterize the molecular pathology underlying these developmental encephalopathies are pointing to abnormalities of telencephalic development, neuronal morphogenesis, maturation and maintenance, and dendritic arborization.

Rett syndrome is defined as an X-linked neurodevelopmental disorder characterized by loss of spoken language and hand use, with the development of distinctive hand stereotypies (Neul et al. 2010). Since the original description of Rett (1966) and the subsequent characterization of the disorder by Hagberg et al. (1983), diagnosis has been based only on consensus clinical criteria until methyl-CpG-binding protein 2 (MECP2) was identified as the causative gene (Amir et al., 1999) (Table 1). Such criteria, which have been modified only slightly over time (Neul et al. 2010), have retained key clinical elements that identify “classic” or “typical” Rett syndrome but also “variant” or “atypical” forms. These variants appear to cluster in somewhat distinct clinical subgroups, which have been described as “preserved speech variant,”“congenital variant,” and “early seizure variant” (Hagberg & Skjeldal, 1994). Mutation screening identifies MECP2 gene alterations in 95–97% of patients with typical Rett syndrome (Neul et al., 2008), but only in 50–70% of atypical cases (Percy et al., 2007), which indicates genetic, not just clinical, heterogeneity of so-called atypical cases.

Table 1.   Clinical features of MECP2-, CDKL5-, and FOXG1-gene–related encephalopathies
 Rett syndrome MECP2 duplication syndrome CDKL5 mutations/deletions FOXG1 mutations/duplications/deletions
  1. +, least severe; ++, severe; +++, most severe; GTC, generalized tonic–clonic.

DevelopmentRegression at 1–3 yearsSevere early delaySevere early delaySevere early delay
Social interactionPoorPoorPoorPoor
Muscle toneInitially normal, evolving to hypotoniaEarly hypotonia, evolving to spasticityHypotoniaHypotonia
Speech-languageAbsentMostly absentAbsentAbsent or minimal
WalkingImpaired (dyspraxic) or absentPossible (70%), impairedAbsent or impairedAbsent
Abnormal movementsHand stereotypies, myoclonus, mouthingChoreiform movements, hand stereotypiesHand stereotypies, bruxism (40%)Dyskinesia, stereotypies
Autonomic dysfunction+++++++
Breathing abnormalitiesHyperventilation during wakefulnessAbsent or minimalAbsent or minimalAbsent or minimal
Autistic behaviorOften presentOften presentPoor eye contact, reduced social interactionPoor eye contact, reduced social interaction
Sleep disturbancesPoor sleep patternPoor sleep patternPoor sleep patternPoor sleep pattern
MicrocephalyAcquired microcephalyAbsentMostly borderline (30%)Borderline head size at birth, severe postnatal microcephaly
Facial dysmorphismsMostly absentMild, always presentMostly absentOnly in patients with deletions
EpilepsyPartial complex and GTC seizuresMultiple seizure typesEarly onset epileptic encephalopathy with spasms, myoclonus and prolonged GTCInfantile spasms in association with duplications; complex partial, GTC, myoclonic

Advances in molecular genetics and study of copy number variants have over the last decade identified intragenic or genomic alterations causing different syndromes with severe cognitive impairment, epilepsy and, often, dyskinetic movement disorders, which have variably been defined as “Rett variants” (Mari et al., 2005; Scala et al., 2005; Papa et al., 2008; Jacob et al., 2009; Mencarelli et al., 2010; Philippe et al., 2010). If, on the one hand, it has become obvious that “atypical Rett” is indeed a heterogeneous collection of different etiologies, on the other hand, perplexity should be raised about how appropriate is an all-embracing Rett syndrome category for disorders whose common feature is profound cognitive disability and the unavoidably associated loss of purposeful hand use, with other clinical characteristics such as access to language, epilepsy type and severity, and associated movement disorders, being differently distributed. One might think that widespread, severe impairment of higher cortical functions leaves a residual, narrow behavioral repertoire, including lack of finalized hand use, hand stereotypies, and autistic-like features and that patients whose behaviour is so restricted have, unavoidably, overlapping features. There is an obvious ascertainment bias for those disorders that are initially characterized based on further molecular studies conducted on DNA that had been referred to a genetic laboratory for investigating a given syndrome, for instance Rett syndrome, when other genetic syndromes, with somewhat overlapping features were still unknown. These samples were stored as suspected Rett syndrome cases. If a new genetic defect is found, it inevitably becomes a new genetic cause for that syndrome. Only over time, after the new genetic defect has been screened in a larger population, with more careful phenotypic descriptions becoming available, it becomes apparent that we are dealing with different syndromes with their own specific clinical and pathophysiologic characteristics, not just “variants.” In this perspective it is clearly emerging that epilepsy has distinctive characteristics in typical Rett syndrome and in the different syndromes caused by cyclin-dependent kinase-like 5 (CDKL5)- and forkhead box G1 (FOXG1)-gene alterations. The progressive parting of CDKL5- and FOXG1-gene–related encephalopathies from the core Rett syndrome is reflected by the effort to produce clearer diagnostic criteria for typical and atypical Rett syndrome (Neul et al. 2010). We review herein epilepsy in Rett syndrome and in other MECP2-gene–related disorders, in CDKL5-gene and FOXG1-gene–related encephalopathies.

Rett Syndrome and Epilepsy

Mutations of the MECP2 gene in female patients cause a spectrum of phenotypes ranging from severe encephalopathy to asymptomatic carriers whose carrier status is detected after investigation for familial Rett syndrome (Wan et al., 1999). The asymptomatic carrier status is favored by extreme X-inactivation skewing, with the mutated allele being largely inactivated. The severe phenotypes include boys without the distinctive clinical features of Rett syndrome, severe early progressive encephalopathy (Kankirawatana et al., 2006), early onset myoclonic epilepsy (Leuzzi et al., 2004), and early death. The type and location of mutation dramatically affect the male phenotype, with severe inactivating mutations causing neonatal encephalopathy (Wan et al., 1999) and hypomorphic alleles yielding to X-linked mental retardation with or without psychosis (Cohen et al., 2002; Kleefstra et al., 2002). Rare girls with autism and no evidence of regression have also been described (Carney et al., 2003). Typical Rett syndrome remains by far the most frequent phenotype associated with MECP2 mutations. It occurs in 1.09/10,000 females. Epilepsy has been reported in 60% of patients in the largest available survey (Glaze et al., 2010), which, however, included both “classic” and “atypical” cases. In general, individuals with seizures were considered to have greater overall clinical severity and greater impairment of ambulation, hand use, and communication. This study, which included 315 classic and 74 atypical Rett patients, was based on parents’ report of seizures, as well as physician assessment of the clinical description and age at onset of seizures, without electrographic or video–electroencephalography (EEG) recordings and is therefore more informative from an epidemiologic, rather than clinical, perspective. Occurrence of seizures was comparable in the two groups (60% vs. 61%). MECP2 mutations were identified in 90% of participants, reaching 93% in the classic Rett subgroup. Seizures were reported in 59% of those with MECP2 mutations and in 73% of those without a mutation of this gene. Evaluation of the clinical description of reported seizures by Rett specialist physicians indicated that only 48% of the whole group was considered to have “true” epileptic seizures. A significant age-related occurrence of seizures was noticed, with none of the patients aged <2 years (nine) having seizures at initial visit, which had instead occurred in 33% of those aged between 3 and 5 years (99), in 62% of those aged 5–10 years (175), in 80% of those aged 10–15 years (94), in 86% of those aged 15–20 years (64), and in 84% of those aged 20–30 years (70). Sixty percent of patients carrying a mutation had one of the eight most common mutations (R106W, R133C, T158M, R168X, R255X, R270X, R294X, and R160C), 9% had C-terminal deletions, and 8.5% had large deletions. Age-adjusted analysis revealed that seizures were most frequent in patients with the T158M mutation. Other studies from Europe and Australia, reported higher rates of seizures, varying between 79% and 88% (Steffenburg et al., 2001; Jian et al., 2007; Pintaudi et al., 2010). Overall, these studies suggest that seizure onset before the age of 3 years is unlikely and onset after age 20 is rare.

The overall occurrence of seizures in Rett syndrome is probably overestimated. Indeed, video-EEG recordings suggest that paroxysmal nonepileptic manifestations such as motor activity, twitching, jerking, falling forward, as well as episodes of staring, laughing, breath holding, and hyperventilation are often misdiagnosed as seizures (Glaze et al. 1998). The consequence may be overtreatment and pseudo drug-resistance.

In one study in which a large number of patients had video-EEG recordings, complex partial seizures and generalized tonic–clonic seizures were considered by far the most frequent seizure types both at onset and during follow-up (Pintaudi et al., 2010), although the number of patients with recorded seizures was not specified.

Cortical reflex myoclonus, manifested as multifocal, arrhythmic, and asynchronous jerks involving distal limbs, is frequently observed in Rett syndrome and exhibits particular neurophysiologic features, including a prolonged intracortical delay of the long-loop reflex (Guerrini et al., 1998).

Seizure severity and drug resistance are difficult to assess. This is due in part to difficulties in differential diagnosis between epileptic and nonepileptic seizures and in part to the variation in severity observed at different ages. In one study, 165 patients were divided in three subgroups, including 21% who had never had epileptic seizures, 49% who were defined as drug responsive and 30% who were drug resistant (Pintaudi et al., 2010). In another study, 36% of the patients with Rett syndrome and epilepsy experienced no seizures during a 6-month follow-up, 27% had had monthly seizures, 20% weekly seizures, and 11 had daily seizures (Glaze et al., 2010); 20% of patients were on no drugs, 37% received one antiepileptic drug (AED), and 8% received three and or more AEDs.

Neuropathologic and neuroimaging studies have demonstrated an early and marked decrease in brain size in Rett syndrome and have been considered to provide converging evidence for a role of MeCP2 in neuronal maturation and maintenance (Chahrour & Zoghbi, 2007). The morphologic correlates include a generalized reduction in neuronal soma size and dendritic arborizations, primarily affecting the cerebral cortex (Jellinger, 2003; Armstrong, 2005). There is also a decrease in the size of cortical minicolumns (Casanova et al., 2003). Immunohistochemical studies revealed a reduced expression of microtubule-associated protein (MAP)2 in the neocortex, which indicates a marked disruption of a major cytoskeletal component (Jellinger, 2003). Using a complementary semiautomated Talairach- and voxel-based magnetic resonance imaging (MRI) approach, Carter et al. (2008) studied 23 girls with Rett syndrome and MECP2 mutations and demonstrated absolute volumetric reductions distributed throughout the brain. Selective volume reductions occurred in the dorsal parietal gray matter and in girls with a more severe phenotype; anterior frontal lobe volumes were relatively more reduced. Mild and diffuse reduction in cortical white matter (the white matter underlying the cortex) also occurs, which is likely related to axonal pathology, as no changes in T2 signal intensity that might indicate myelination abnormalities have been reported (Carter et al., 2008). Brain proton magnetic resonance spectroscopy (MRS) shows decreased N-acetyl-aspartate (NAA) in both the gray and white matter, particularly involving the frontal and parietal lobes and the insular cortex, reflecting reduced neuronal and dendritic tree size and also decreased neuronal function (Jellinger, 2003). Densities of receptors for glutamate and for γ-aminobutyric acid (GABA), which are localized to dendritic spines, have been shown to be abnormal in postmortem brain tissue from young female individuals with Rett syndrome compared with densities from age-matched control samples (Blue et al., 1999a,b).

MECP2 Duplication Syndrome

The MECP2 Duplication Syndrome, now described in >100 individuals, is characterized by infantile hypotonia, severe to profound mental retardation, autistic features, poor or absent speech, recurrent infections, epilepsy, progressive spasticity, and dyskinetic movements (Ramocki et al., 2010) (Table 1). Developmental regression is sometimes observed and is usually associated with the worst prognostic outlook. Overall it has been estimated that 72% of reported patients achieved ambulation and almost 40% died before their 25th birthday, usually from respiratory infections (Ramocki et al., 2010). The syndrome is fully penetrant in male individuals, who most often inherit the genomic abnormality from their mother; however, a few de novo cases have been reported, especially in relation of functional disomy resulting from X-Y and X-autosome translocations (Clayton-Smith et al., 2009; Ramocki et al., 2010). Carrier females are unaffected but some may exhibit neuropsychiatric disorders. Most reported duplications are small, spanning from 0.3 to 4 Mb, and are only visible by real time polymerase chain reaction (PCR), multiplex ligation-dependent probe amplification (MLPA) or array-comparative genomic hybridization (array-CGH). Larger, cytogenetically visible duplications have been described (Carvalho et al., 2009; Ramocki et al., 2010). Triplication of the MECP2 locus may also occur and is associated with a similar, although more severe, phenotype (Del Gaudio et al., 2006). Genotype–phenotype correlation studies indicate that the minimal duplicated region that is necessary to cause the core phenotype involves MECP2 and the Interleukin-1 receptor-associated kinase 1 (IRAK1) genes (Del Gaudio et al., 2006; Lugtenberg et al., 2009).

Transgenic mice engineered to overexpress the wild-type human MECP2 protein, at approximately twofold wild-type levels, developed a progressive disorder with seizures, reduced activity, repetitive movements, spasticity, and premature death (Collins et al., 2004). Seizures are described as clonic and “akinetic” episodes during spike and wave EEG discharges. These types of seizures are different clinically and electrophysiologically from those observed in the loss-of-function mice that recapitulates the Rett phenotype (Shahbazian et al., 2002a). It should be noted that MECP2 overexpression mice exhibited enhanced synaptic plasticity, as measured by increased paired-pulse facilitation (PPF) and long-term potentiation (LTP), whereas the loss-of-function mutants show decreased PPF and LTP (Collins et al., 2004).

Based on genotype–phenotype correlations and experience on animal models MECP2 is therefore considered to be the primary dosage-sensitive gene contributing to the neurologic phenotype in boys with the duplicated Xq28 region. Xq28 duplication including MECP2 was the most common submicroscopic telomeric rearrangement in 5,380 consecutive patients referred for microarray testing (Shao et al., 2008), and it is estimated that the MECP2 duplication syndrome may represent about 1% of cases of X-linked mental retardation (Ramocki et al., 2010). Mild dysmorphic features are observed in some individuals, including brachycephaly, large ears, and midface hypoplasia. Slowing of background EEG, with multifocal and generalized discharges, are observed in most cases (Ramocki et al., 2010). Epilepsy has been described as focal or generalized but has not been well characterized. Echenne et al. (2009) studied five patients and reported multiple seizure types, usually resistant, including atypical absence, generalized convulsive, myoclonic, and focal seizures, with frequent status epilepticus. Vignoli et al. (2012) studied eight patients and found atonic seizures with falling forward of the head and trunk to be the most frequent seizure type. Generalized tonic–clonic seizures were rarely observed. The most frequent EEG pattern was with generalized slow spike and wave asynchronous discharge. Drug resistance was a significant feature. The relative rarity of the syndrome and lack of a systematic approach to the study of epilepsy makes it impossible to know whether a sufficiently characteristic epilepsy phenotype exists in boys with MECP2 duplication. Neither does published work allow to know what is the percentage of patients with epilepsy and the rate of intractability, although it appears that most have intractable seizures. Information on the epilepsy characteristics and other relevant clinical features will undoubtedly arise from global initiatives such as the “MECP2 Duplication Syndrome Seizure Status,” launched by a U.S.-based association of families affected by MECP2 duplication (http://www.mecp2duplication.com). This association is now reporting the results of a poll for parents or primary caregivers of individuals with MECP2 duplication syndrome. Responses were provided for 102 individuals with the syndrome. Overall, 46% of individuals were seizure free, 17% had experienced seizures that were considered to be well controlled, 37% had seizures that were considered to be poorly controlled. Seizures appeared to be more prevalent and more difficult to control with age and among those who lived into adulthood; the true prevalence of seizures approached 100%. This kind of initiative brings precious information that, if adequately supported by standardized seizure questionnaires and analysis of data by expert physicians, may obviate to the unavoidable slowness that is necessary to characterize epilepsy in complex and rare disorders.

The MECP2 Gene and Protein

The MECP2 gene maps on Xq28 and undergoes X chromosome inactivation. The MECP2 protein is expressed in all tissues (Fig. 1A). In the brain, MECP2 is expressed in neurons and astrocytes, both in the nucleus and cytosol (Nagai et al., 2005). MECP2 is a transcriptional regulator involved in chromatin remodeling and the modulation of RNA splicing. In resting neurons, the MECP2 protein regulates gene expression (Chen et al., 2003) and acts as a transcriptional activator for some genes (Chahrour et al., 2008). Neuronal activity induces MECP2 phosphorylation, dissociation of MECP2, and the corepressor complex from promoters and target gene expression (Brooks-Kayal, 2011). MECP2 mutations cause loss of activity-dependent changes in gene expression that may disrupt synaptic plasticity (Chahrour & Zoghbi, 2007). MECP2 loss induces changes in the expression of a number of genes involved in synaptogenesis (Smrt et al., 2007), although the precise mechanism by which loss of MECP2 results in epilepsy, severe disability, and autistic features remains uncertain (Brooks-Kayal, 2011). Studies in a mouse model of Rett syndrome suggest that abnormal cortical glutamatergic synaptic responses and excitatory connectivity result in a relative excess of inhibition compared to excitation, which precede deficits in plasticity and might play an important role in causing the clinical features (Dani et al., 2005; Dani & Nelson, 2009). The finding that MECP2 is expressed in mature neurons and that its amount increases postnatally in these cells throughout childhood in humans argues that its function is critical in mature neurons, likely after they have engaged in synaptogenesis and are undergoing adolescent synaptic pruning (Shahbazian et al., 2002b).

Figure 1.

(A) MECP2 is a member of the methyl-CpG binding protein family and is composed of three domains: the methyl-CpG binding domain (MBD, light blue box), the transcriptional repression domain (TRD, light yellow box), and a WW domain binding region with two highly conserved tryptophans (WWD, red box), in addition to two nuclear localization signals (NLS, purple box). The MBD specifically binds to methylated CpG dinucleotides, with preference for CpG sequences with adjacent A/T-rich motifs. MBD also binds to unmethylated four-way DNA junctions with a similar affinity, implicating a role for the MECP2 MBD in higher-order chromatin interactions. The more downstream TRD is able to recruit corepressor complexes that mediate repression through deacetylation of core histones, with consequent compaction of DNA into heterochromatin. WWD facilitates MECP2 binding to naked DNA and to the nucleosomal core. It also contains evolutionarily conserved poly-proline runs that can bind to group II WW domain splicing factors. MECP2-A (486 amino acids) and MECP2-B (498 amino acids) protein isoforms are produced by alternative splicing of the transcript and differ in the N-terminal regions (pink boxes) according to the exon involved (exon 2 in MECP2-A and exon 1 in MECP2-B). (B) CDKL5 is a serine/threonine kinase protein which is characterized by a catalytic domain (CD, light gray box) and a long C-terminal extension involved in the regulation of the catalytic activity of CDKL5 and in the subnuclear localization of the protein. Two kinase signatures are contained in the CD domain: an ATP-binding region (ABD, yellow box) and a serine-threonine protein (STK, orange box). In addition, CDKL5 has a Thr-Xaa-Tyr sequence that corresponds to the TXY activation motifs of classic mitogen-activated protein (MAP) kinases, and a putative signal peptidase I serine active site (SPISA, green box). (C) FOXG1 is a DNA-binding transcription factor with a forkhead domain (FHD, blue box) that represses target genes during brain development with a key role in early telencephalon patterning. FOXG1 recruits transcriptional corepressor proteins, a histone demethylase and Groucho (Gro), via two further protein-binding domains, the Lysine-specific demethylase 5B binding domain (KBD, grey box) and the about 20-residue Gro-binding domain (GBD, light green box), respectively. Structurally, the FHD consists of three alpha helices and one beta hairpin (two beta strands and one loop), whereas the GBD and KBD are random coiled.

Almost 70% of all MECP2 mutations are caused by C>T transitions at 8CpG dinucleotides, in the third and fourth exons. The most common mutation is R168X. Mutations are distributed throughout the gene, but missense mutations tend to cluster in the methyl-CpG binding domain (MBD). Null mutations are distal to the MBD, and large multinucleotide deletions occur in the C-terminal domain (Matijevic et al., 2009). Despite the high number of mutations reported, there are no clear genotype–phenotype correlations. Patients carrying missense mutations or deletions within the hotspot region for deletions appear to be less severely affected, whereas either complete or partial truncations of the region coding for the nuclear localization signal would cause more severe consequences than other truncating mutations (Huppke et al., 2003). Reports of discordant phenotypes in sisters carrying the same intragenic deletion raise the question of the modulatory role of epigenetic factors or of additional genetic influences (Scala et al., 2007). For example, skewing of X chromosome inactivation appears to remarkably influence phenotype severity. Among patients with the same MECP2 mutation, those with a significant increase of the active mutated allele in blood leukocytes have a more severe phenotype, likely resulting from the unfavorable X chromosome inactivation skewing (Archer et al., 2007). The large majority of MECP2 mutations occur de novo. Familial cases have originated either from mutations inherited from healthy or mildly affected mothers, with favorable X inactivation skewing, or from gonadal mosaicism (Matijevic et al., 2009).

Mecp2-null mice and conditional mouse mutants exhibit some features that recapitulate the clinical manifestations of Rett syndrome. Mecp2-null mice have reduced brain weight and neuronal cell size (Chen et al., 2001). Shahbazian et al. (2002a) generated mice with a truncating mutation similar to those found in Rett syndrome. These mice appeared normal and exhibited normal motor function for about 6 weeks, but then developed progressive neurologic manifestations including tremors, motor impairment, hypoactivity, increased anxiety-related behavior, seizures, kyphosis, and stereotypic forelimb motions.

CDKL5-Gene–Related Early Onset Epileptic Encephalopathy

In 2003, Kalscheuer et al. identified two unrelated girls with seemingly balanced translocations (46, X,t(X;7)(p22.3;p15) and 46,X,t(X;6)(p22.3;q14)), exhibiting infantile spasms and profound developmental delay. In both patients the CDKL5 gene was disrupted by a breakpoint on the X chromosome. Due to some overlapping clinical similarities between the two reported patients and the early seizure variant of Rett syndrome, in the following years, other researchers studied the CDKL5 gene in patients who had been diagnosed with classical or variant Rett and were mutation negative to MECP2 testing, and identified intragenic CDKL5 mutations/deletions in girls with early onset severe seizures (Tao et al., 2004; Weaving et al., 2004; Mari et al., 2005; Scala et al., 2005; Bahi-Buisson et al., 2008a). In the last several years, the large number of patients reported, with detailed description of epilepsy and EEG features, has permitted the delineation of a phenotypic spectrum spanning from milder forms—which include the possibility of autonomous walking and less severe epilepsy that is amenable to control—to severe forms featuring intractable seizures, more severe microcephaly and absence of motor milestones (Table 1). In this spectrum several of the distinctive clinical features of Rett syndrome are lacking. For instance, girls with CDKL5 mutations do not exhibit a clear period of regression, neither do they present the intense eye gaze and impaired neurovegetative function typically seen in girls with Rett. The central feature of CDKL5-related phenotype is an early onset epileptic encephalopathy with seizures starting within the fifth month of life, severe developmental delay, deceleration of head growth, impaired communication and, often, hand stereotypies. Several dozen girls and a few boys with CDKL5 gene mutations, or genomic deletions, have been reported, all having early onset intractable seizures. CDKL5 mutations/deletions were found in 8–16% of girls with early onset seizures (Bahi-Buisson et al., 2008b; Nemos et al., 2009; Mei et al., 2010) and, more specifically in 28% of girls with early onset seizures combined with infantile spasms (Bahi-Buisson et al., 2008a).

To outline the difference with respect to Rett syndrome, the definition of CDKL5 gene–related epileptic encephalopathy has been proposed (Melani et al., 2011). Males are at the more severe end of the phenotypic spectrum with virtually no motor acquisitions (Van Esch et al., 2007; Sartori et al., 2009; Melani et al., 2011). Patients with CDKL5-gene abnormalities are reported to be normal in the first days of life to subsequently exhibit early signs of poor developmental skills, including poor sucking, and poor eye contact, even before seizure onset. Subsequently, absent purposeful hand use, severe developmental delay, and absent language skills become apparent (Archer et al., 2006; Bahi-Buisson et al., 2008b; Elia et al., 2008; Nemos et al., 2009; Mei et al., 2010; Neul et al. 2010; Melani et al., 2011). About one third of patients will eventually be able to walk (Bahi-Buisson et al., 2008b). Epilepsy is typically manifested as an epileptic encephalopathy, with infantile spasms starting between the first days and fourth month of life (Archer et al., 2006; Bahi-Buisson et al., 2008b; Mei et al., 2010). Patients show a peculiar seizure pattern with “prolonged” generalized tonic–clonic events, lasting 2–4 min, consisting of a tonic-vibratory contraction, followed by a clonic phase with series of spasms, gradually translating into repetitive distal myoclonic jerks (Melani et al., 2011). The EEG correlate of this ictal event is a bilateral and synchronous initial flattening, followed by repetitive sharp waves and spikes. A generalized ictal pattern in neonates or very young infants has been described, but is considered to be exceptional, owing to the lack of both the functional cortical organization that is necessary to propagate and sustain the electrical discharge and the failure of interhemispheric transmission resulting from commissural immaturity (Hirsch et al., 1993). Whether this peculiar ictal pattern represents a unique seizure type or just an unusual expression of infantile spasms in the immature, and possibly malformed, brain of children with CDKL5-gene abnormalities is open to speculation.

Follow-up studies of children older than 3 years indicate that about half of them can experience seizure remission, whereas the remaining continue to have intractable spasms, often associated with multifocal and myoclonic seizures (Bahi-Buisson et al., 2008b; Mei et al., 2010). Early EEG findings vary from normal background to moderate slowing, with superimposed focal or multifocal interictal discharges and, in some cases a suppression burst pattern (Melani et al., 2011). Atypical hypsarrhythmia is often seen in infancy, and multifocal abnormalities predominate in older patients (Bahi-Buisson et al., 2008b).

There are no neuropathologic studies that have examined the brain of patients with CDKL5 mutations. Imaging data are also scanty and nonquantitative. Bahi-Buisson et al. (2008b) reported “cortical atrophy” in 13 of 20 girls, associated with areas of increased T2 signal in the white matter, especially in the temporal lobes in some.

The CDKL5 Gene and Protein

CDKL5 was identified through an exon trapping method designed to screen candidate genes in Xp22, a region where several genetic disorders have been mapped (Montini et al., 1998). Sequence analysis suggests that CDKL5 is a member of a proline-directed kinase subfamily that has homology to both MAP and cell-cycle–dependent kinases known as the cyclin-dependent kinase-like (CDKL) kinases (Lin et al., 2005).

CDKL5 is a ubiquitous protein (Fig. 1B) but is expressed mainly in the brain (cerebral cortex, hippocampus, cerebellum, striatum, and brainstem), thymus, and testes (Lin et al., 2005). In the developing mouse brain, CDKL5 expression is strongly induced in early postnatal stages, and in the adult brain CDKL5 is present in mature neurons, but not in astroglia. CDKL5 levels are low in the embryonic cortex and are strongly induced at perinatal and postnatal stages in maturing neurons in the cerebral cortex and hippocampus. This expression profile suggests a role of the CDKL5 protein in neuronal maturation (Rusconi et al., 2008). CDKL5 shuttles between the cytoplasm and the nucleus, which suggests a function in both cellular compartments. Its presence in the cell nucleus varies at the regional level of the adult brain and is developmentally regulated (Rusconi et al., 2008). CDKL5 is considered a kinase on the basis of sequence homologies. However, its phosphorylation targets are poorly known. In vitro, CDKL5 phosphorylates the product of the MECP2 gene, suggesting a common signaling pathway between these two proteins (Mari et al., 2005). A series of experiments by Chen et al. (2010) provides compelling evidence that CDKL5 plays a critical role in neuronal morphogenesis, and that migratory defects may cause early seizures in patients with CDKL5 mutations. These authors observed that downregulating CDKL5 by RNA interference (RNAi) in cultured cortical neurons inhibits neurite growth and dendritic arborisation, whereas overexpressing CDKL5 has opposite effects (Chen et al., 2010). Knocking down CDKL5 in the rat brain by in utero electroporation resulted in delayed neuronal migration and severely impaired dendritic arborization (Chen et al., 2010). No mouse model for CDKL5-gene–related epileptic encephalopathy is currently available.

Since the first point mutations in the CDKL5 gene were reported, several dozen different sequence variations have been described resulting in missense, nonsense, splice, and frameshift mutations. Null mutations causing the premature termination of the protein are distributed throughout the open reading frame, whereas missense mutations are located almost exclusively within the kinase domain. Missense mutations in the catalytic domain would cause impaired autophosphorylation and phosphorylation of target proteins such as MeCP2 (Bertani et al., 2006). Mutations affecting the catalytic activity of CDKL5 have been associated with a more severe phenotype (Bahi-Buisson et al., 2008b; White et al., 2010). Only a few mutations have been reported in more than one patient (p.A40V, p.R178W, c.2635_2636del2, c.145+2 C>T, and (p.R59X). Affected boys harbor various germline mutation types but may also carry somatic mosaic mutations Guerrini R and Mei D, unpublished data and Masliah-Plachon et al., 2010). Most girls with CDKL5 mutations exhibit a random pattern of X chromosome inactivation in blood leukocytes (Bahi-Buisson et al., 2008b; Nemos et al., 2009).

Microdeletions involving several exons or the whole CDKL5 gene have been detected using array-CGH or MLPA. Larger rearrangements have also been reported, removing only a few exons of the CDKL5 gene or a segment, or the entire genomic region of CDKL5 and several contiguous genes (Bahi-Buisson et al., 2010; Mei et al., 2010; Castrén et al., 2011). The epilepsy phenotype of patients with large deletions does not differ from that seen with intragenic sequence variations, suggesting that CDKL5 is the major determinant of the phenotype, whereas contribution of contiguous deleted genes is minor, if any. Large deletions, which escape gene sequencing in girls, should be systematically searched using MLPA, real-time quantitative PCR, or array-CGH, when investigating a severe early onset epileptic encephalopathy, as their occurrence is probably underestimated (Mei et al., 2010). With the exception of one instance of familial occurrence, likely due to gonadal mosaicism (Weaving et al., 2004), all reported cases are sporadic. Although the risk of gonadal mosaicism, and consequent familial recurrence is low, genetic counseling should be offered to couples with a child with CDKL5-related epileptic encephalopathy.

The FOXG1 Syndrome and Epilepsy

Intragenic mutations or duplications/deletions of the FOXG1 gene have been reported initially in patients with a developmental disorder classified as the congenital variant of Rett syndrome (Neul et al., 2010). Although some clinical similarities exist, there are also a number of clinical features that are consistently different, and sufficiently distinct, to allow clinical recognition of the FOXG1 syndrome (Kortüm et al., 2011) (Table 1). Patients with FOXG1 alterations have a complex association of clinical features that includes postnatal growth deficiency, severe postnatal microcephaly, severe developmental delay with absent speech, defective social reciprocity resembling autism, poor sleep, stereotypies in combination with dyskinesia, and epilepsy. Almost none of the patients achieve autonomous walking. There appears to be no history of regression. Brain MRI reveals a simplified gyral pattern, with hypogenesis of the corpus callosum and cortical thickening in the frontal lobes (Kortüm et al., 2011).

The initial reports that associated FOXG1 with a syndrome of severe developmental delay resembling Rett were prompted by the finding of interstitial de novo deletions in 14q12 in patients who had resulted mutation negative to MECP2 screening (Jacob et al., 2009; Mencarelli et al., 2009). From this initial reports, FOXG1 was soon identified as a strong candidate gene for the syndrome, due to its high expression in the developing brain and the reported developmental abnormalities in the telencephalon of both heterozygous and homozygous mouse mutants (Eagleson et al., 2007; Siegenthaler et al., 2008). The ensuing screening of FOXG1 in small series of patients with similar phenotypes and no mutation of MECP2, identified nonsense, frameshift, and missense mutations. Although some overlapping with Rett syndrome did exist (nonetheless due to selection bias), it was acknowledged that no early period of normal development could be identified. Severe early onset epilepsy emerged as a main feature of the FOXG1-related phenotype in subsequent reports (Yeung et al., 2009; Brunetti-Pierri et al., 2011). A recent report, describing 11 new patients carrying different sequence changes involving FOXG1, pointed out the relevant differences from Rett and other so-called “Rett variants,” and suggests that FOXG1 syndrome is a distinct form of developmental encephalopathy (Kortüm et al., 2011). Seizures were documented in eight patients, with onset between 3 months and 6 years, and were classified as tonic and generalized tonic–clonic, and complex partial. EEG data were incomplete and showed focal or multifocal abnormalities. None of the patients in this series had infantile spasms. Of the 15 additional patients with FOXG1 mutations or deletions reported by other authors and reviewed by Kortüm et al. (2011), 10 had epilepsy; although information was largely incomplete, seizures were reported to appear between age 6 months and 14 years to be complex partial, generalized tonic–clonic, or myoclonic. Data on seizure semiology, however, should be considered with due caution as in genetic papers seizure description and terminology are usually reported loosely. On the other hand, the association between 14q12 duplication including FOXG1 and infantile spasms appears to be clearly documented, based on descriptions of 6 patients with spasms of 10 reported patients with duplications (Yeung et al., 2009; Brunetti-Pierri et al., 2011; Striano et al., 2011). Differences in the epilepsy phenotype between patients with null mutations and those with duplications suggest that overexpression of FOXG1 has a specific role in the pathogenesis of infantile spasms. Infantile spasms appear before severe developmental delay and behavioral phenotype typical of the FOXG1 syndrome become apparent, and almost all patients with FOXG1 duplications were reported to have normal head circumference before age 3 years. The clinical presentation is therefore that of cryptogenic infantile spasms. Although the prevalence of 14q12 duplications in cryptogenic infantile spasms is to be determined in large series, they might not be rare (2 of 38 in the series of Striano et al., 2011), indicating that array-CGH testing is indicated in this patient population (Paciorkowski et al., 2011).

The FOXG1 Gene and Protein

Foxg1 is a conserved transcriptional repressor that regulates telencephalon development from the early embryonic to adult stages through multiple mechanisms (Roth et al., 2010). The Foxg1 protein (Fig. 1C) is strongly expressed in progenitor cells of the ventricular zone, in early postmitotic neurons in the developing telencephalon, and in visual structures (Bourguignon et al., 1988; Tao & Lai, 1992). This protein involved in multiple key mechanisms of telencephalon development, including the timing of neuronal differentiation and in the specification of subdomain identity and neural progenitor versus Cajal-Retzius cell fate and progenitor cell cycle length (Xuan et al., 1995; Dou et al., 1999; Hanashima et al., 2004; Muzio & Mallamaci, 2005). Heterozygous Foxg1+/− mice mutants exhibit several developmental abnormalities, including reduced volume of the cerebral cortex, hippocampus, and striatum, and reduced cortical thickness (Eagleson et al., 2007; Siegenthaler et al., 2008). Homozygous Foxg1+/− mutants die at birth and exhibit a severe reduction in brain size and depletion of neural progenitor pools. These findings in animal models predict that FOXG1 haploinsufficiency in humans would result in microcephaly, thin cerebral cortex with abnormal cytoarchitecture, and developmental delay (Kortüm et al., 2011).

Overall 18 intragenic sequence changes of FOXG1 have been reported (reviewed by Kortüm et al., 2011), including 14 null (6 nonsense and 8 frameshift) and 4 missense mutations. All missense mutations affect evolutionarily highly conserved amino acid residues in the forkhead domain, which is involved in DNA binding. Both missense and loss of function mutations can have a severe impact on FOXG1 function (Kortüm et al., 2011). The number of reported mutations is too limited to disclose genotype–phenotype correlations, if any, or to reveal mutation hot spots, although the c.460dupG mutation has already been reported in three patients and might represent one. Patients with microdeletions close to, but not covering FOXG1, whose phenotype overlaps with that seen in FOXG1-mutation positive individuals, have been reported (Shoichet et al., 2005; Mencarelli et al., 2009; Kortüm et al., 2011), prompting the hypothesis that regulatory mutations and mutations directly affecting FOXG1 cause similar phenotypes. Duplications of 14q12 including FOXG1 have been found in 10 patients (Yeung et al., 2009; Brunetti-Pierri et al., 2011; Striano et al., 2011) whose phenotype is similar to that observed in mutation-positive patients, although epilepsy appeared to be more severe, with 6 of the 10 patients having infantile spasms.

Concluding Remarks

Efforts have been made to unravel the molecular bases that might underlie phenotypic similarities between MECP2-CDKL5-FOXG1 encephalopathies. In our opinion, these similarities have been overemphasized and might at least in part be related to the severe neurodevelopmental impairment related to mutations or duplications/deletions of these genes and to the consequently reduced behavioral repertoire of affected patients. The characteristics of associated epilepsy are obviously different.

CDKL5 and MECP2 are widely coexpressed in the brain and are similarly activated during neuronal maturation and synaptogenesis (Mari et al., 2005; Rusconi et al., 2008). A direct interaction between MECP2 and CDKL5 has been demonstrated. CDKL5 can bind and phosphorylate MECP2 in vitro, and MECP2 can in turn regulate CDKL5 gene expression, at least in certain brain regions (Mari et al., 2005; Bertani et al., 2006; Carouge et al., 2010). Moreover, both proteins bind to different regions of DNA methyltransferase 1 (DNMT1) further suggesting their possible participation to common pathways (Kameshita et al., 2008).

The expression profile of FOXG1 overlaps with the described MECP2 expression domain in cortical tissues, in differentiating and mature neurons (Ariani et al., 2008). At the single-cell level, FOXG1 localizes in the nuclear compartment but is excluded from the MECP2-positive heterochromatic foci both in nonneural and primary neurons. These findings suggest that, differently from MECP2, FOXG1 is not a transcriptional repressor stably associated with heterochromatin. However, both proteins exhibit a large colocalization domain in other nuclear compartments. Overall, these data suggest that FOXG1 may exert some additional functions in differentiating and mature neurons, thus sharing similarities with those described for MECP2 (Ariani et al., 2008).

With the expansion of knowledge related to Rett syndrome and the progressive parting of CDKL5- and FOXG1-gene–related encephalopathies from the core Rett syndrome, it is becoming obvious that other genes may be involved in the heterogeneous spectrum of phenotypes defined as “atypical” Rett. Indeed, several patients with phenotypes suggestive of CDKL5- and FOXG1-gene–related encephalopathies remain without a molecular diagnosis. As with any monogenic disorder, mutations located in deep introns or in the promoter and regulatory regions of these genes, as well as mutations in novel genes, possibly related to MECP2, CDKL5, and FOXG1 in their function, might explain the phenotype of these patients. It is to be expected that in the near future, exome sequencing of homogeneous phenotypic subgroups of “atypical” Rett patients will lead to the identification of new entities thinning down the syndrome spectrum to typical cases.


The authors declare no relevant conflicts of interest. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.