Megalencephaly and hemimegalencephaly: Breakthroughs in molecular etiology

INTRODUCTION

The etiologies of focal malformations of cortical development have long been a puzzle. Unlike most forms of lissencephaly and microcephaly where brain morphology appears affected globally in most forms, suggesting a genetic disorder affecting the entire cortex, focal malformations such as focal cortical dysplasia (FCD), hemimegalencephaly (HME), and focal megalencephalies suggest another pattern in which only part of the developing brain appears to have experienced a genetic aberration. Brain overgrowth phenotypes range from very localized lesions to more diffuse multifocal forms. These are conditions were asymmetry is the rule, and where etiology had long been elusive. Non-genetic etiologies were previously sought to explain the presence of a severely overgrown and dysplastic portion of the brain with the neighboring cortex seemingly normal. However, one of the earliest and first recognized neurogenetic syndromes, tuberous sclerosis complex (TSC), featured just this type of patchy malformation, the cortical tuber. For over a decade, it has been known that germline loss of function mutations in TSC1 and TSC2, causing hyperactivation of the mammalian target of rapamycin (mTOR) signaling cascade, are associated with cytomegaly and disorganized lamination within the cerebral cortex, the hallmark features of TSC. Furthermore, loss of function mutations of PTEN, an upstream phosphatase that inhibits the PI3K-AKT-mTOR pathway, are well known causes of generalized megalencephaly (MEG) in syndromic (Cowden and Bannayan–Riley–Ruvalcaba syndromes) and non-syndromic phenotypes (MEG with autism). In this context, the identification of somatic mosaicism in HME and MEG phenotypes makes sense but has only recently come to attention. Mosaicism, as exemplified by these disorders, provides a new set of mechanisms previously under-appreciated as causes of neurological disease [Poduri et al., 2013]. Single cell sequencing of cortical tubers and the detection of somatic TSC1/2 mutations was a foreshadowing of our very rapidly improved understanding of the genetics of these disorders [Crino et al., 2010].

In this review, we will highlight genetic etiologies of brain overgrowth disorders broadly and then review the clinical spectrum and molecular pathogenesis of a recently emerging class of brain overgrowth phenotypes associated with mutations in the PI3K-AKT-mTOR pathway. Identification of new genes is ongoing. However, the biologic effects of these mutations on brain and body overgrowth and genotype–phenotype correlations are still under investigation, and we anticipate that these will be areas of continued discovery in the coming years.

SYNDROMES AND GENES ASSOCIATED WITH BRAIN OVERGROWTH: A GENERAL OVERVIEW

Megalencephaly (MEG) is classically defined as an oversized and overweight brain that exceeds the age-related mean by 2 or more standard deviations. Clinically, the distinction between megalencephaly (enlarged brain) and macrocephaly (enlarged head overall) relies on neuroimaging examination of the brain and recognition of enlarged cerebral structures. Whereas MEG is associated with specific syndromes, macrocephaly can be caused by a myriad of causes such as hydrocephalus or ventriculomegaly, enlarged extra-axial spaces, and thickened skull bones. Therefore, the distinction between MEG and macrocephaly is clinically helpful towards accurate diagnosis.

Megalencephaly (MEG) is classically defined as an oversized and overweight brain that exceeds the age-related mean by 2 or more standard deviations. Clinically, the distinction between megalencephaly (enlarged brain) and macrocephaly (enlarged head overall) relies on neuroimaging examination of the brain and recognition of enlarged cerebral structures.

MEG has long been classified based on pathogenesis into metabolic and non-metabolic (or anatomic) subtypes [DeMyer, 1972, 1986]. Cellular hypertrophy due to cellular edema or accumulation of metabolic substrates can cause MEG in a wide range of neurometabolic syndromes, such as Canavan disease, glutaric aciduria type I, lysosomal storage disorders, among others (Table I). A growing number of developmental (or non-metabolic) genetic syndromes are known to be associated with generalized or focal MEG, including HME. Table II represents a broad overview of genetic disorders where MEG is a defining/diagnostic or common feature. Brain overgrowth in these syndromes varies widely in severity, distribution and co-occurrence of other malformations of cortical development from a mild (and often relatively) enlarged brain with a normal cortex to bilateral MEG with diffuse cortical dysplasia, as discussed below. Finally, reciprocal copy number changes are known to be associated with brain growth dysregulation (i.e., MEG and microcephaly) (Table III).

Other important overgrowth conditions include Sotos syndrome, Weaver syndrome, Simpson Golabi Behmel syndrome, and nevoid basal cell carcinoma syndrome. While they are beyond the scope of this review, many of the same principles apply to these disorders as to the conditions that affect the brain predominantly.

MOLECULAR PATHWAYS OF MEGALENCEPHALY AND HEMIMEGALENCEPHALY

Cellular growth of neuronal elements is an intricately orchestrated process, as discussed by Drs. Alcantara and O'Driscoll in this series. Dysregulation of a number of critical pathways is known to be associated with human brain overgrowth phenotypes, as highlighted in Table II. Dysregulation of two particular critical cellular pathways, the Ras/mitogen-activated protein kinase (MAPK) pathway and the PI3K-AKT-mTOR pathway, appear to account for the largest number of known MEG/HME syndromes. Both pathways are associated with multiple diverse cellular functions including cellular proliferation, differentiation, cell cycle regulation, survival, and metabolism. Not surprisingly, these two pathways are functionally related. Given their critical developmental roles, germline mutations of genes in both pathways are believed to be embryonic lethal. When dysregulated, regardless of the specific gene or protein alteration, the ensuing syndromes exhibit numerous overlapping phenotypic features spanning many organ systems. Furthermore, both pathways have been extensively studied in the cancer field and constitute very attractive targets for pathway (small molecule) inhibitor therapy to treat various malignancies.

Whereas most mutations in the RASopathies are germline, the emerging spectrum of PI3K-AKT-mTOR pathway phenotypes (particularly in upstream components such as PIK3CA) includes predominantly post-zygotic mutations present in a mosaic pattern. The related MEG syndromes involve almost every organ system in the body including the brain (intellectual disability, autism, epilepsy, hydrocephalus, Chiari malformation), heart and vascular system (conduction defects, heart-great vessel anomalies), skin (capillary malformations, epidermal nevi), connective tissue (skin laxity, joint hypermobility), skeleton (polydactyly, syndactyly), and others. Mutations within the PI3K-AKT-mTOR pathway are associated with the most severe brain overgrowth phenotypes including marked brain overgrowth (occipito-frontal circumference, OFC, more than 4 standard deviations above the mean) and HME, a serious medical condition typically associated with severe early onset intractable epilepsy and poor developmental outcome. While some syndromes in both pathways are considered cancer syndromes, as identified mutations are “activating” causing enhanced pathway activation, some of the novel germline and mosaic mutations are not as robustly activating as those associated with oncogenesis and require further study. We will specifically focus on an area of rapidly increasing knowledge related to PI3K-AKT-mTOR-related brain overgrowth phenotypes, their clinical and neuroimaging features, and molecular pathogenesis.

PI3K-AKT RELEATED MEG AND HME: THE CLINICAL AND NEUROIMAGING SPECTRUM

In addition to PTEN-related disorders, upstream mutations in core components of the PI3K-AKT-mTOR pathway have now been identified in classic HME and a variety of MEG syndromes including the MEG-capillary malformation syndrome (MCAP) and the MEG-polymicrogyria-polydactyly-hydrocephalus syndrome (MPPH). Collectively, these disorders not only overlap molecularly but also share clinical, neuroimaging, and neuropathologic features.

PTEN-Related Disorders

Loss of function mutations of PTEN have been identified in a wide range of MEG phenotypes including two well-known cancer predisposition syndromes: Cowden and Bannayan–Riley–Ruvalcaba syndrome (BRRS) syndromes [Liaw et al., 1997; Marsh et al., 1997]. These disorders constitute a clinical spectrum associated with prenatal onset MEG, hamartomas, lipomas, intestinal polyps, and various types of cutaneous vascular malformations [Gorlin et al., 1992; Tan et al., 2011]. Neurologically, affected individuals have hypotonia, delayed gross motor skills and, in some, proximal myopathy [Marsh et al., 1999]. Prenatal onset progressive MEG is typical and OFCs are usually 4–5 (and up to 8) standard deviations above the mean, whereas body overgrowth is typically mild (+1–3 SD). PTEN mutation carriers are at increased risk for various tumors, most notably of the breast, thyroid, and endometrium. Finally, PTEN mutations constitute the largest single gene defects in autistic children with MEG, with estimates of mutations between 1% and 17% in this cohort [Butler et al., 2005; Buxbaum et al., 2007]. While OFCs in this cohort vary, one of the earliest reports showed OFCs of +7–8 SD above the mean [Butler et al., 2005]. An average OFC of +4.35 SD was recently reported in 6 PTEN mutation-positive individuals with autism [Hobert et al., 2014]. Pten loss of function mutant mice develop macrocephaly and behavioral abnormalities such as reduced social activity, increased anxiety and sporadic seizures, closely resembling the human phenotype of PTEN-related disorders [Kwon et al., 2001; Ogawa et al., 2007].

Pten loss of function mutant mice develop macrocephaly and behavioral abnormalities such as reduced social activity, increased anxiety and sporadic seizures, closely resembling the human phenotype of PTEN-related disorders.

While most children with PTEN mutations have uniform bilateral MEG with grossly normal cortical cytoarchitecture (Fig. 1A–C), there are several published cases of asymmetric or focal brain phenotypes in individuals with germline PTEN mutations. These include HME and linear epidermal nevi in a child whose family has features of BRRS, and FCD with focal intractable epilepsy in a child with features of Cowden syndrome [Merks et al., 2003; Elia et al., 2012] (Fig. 1D–F).

The Megalencephaly-Polymicrogyria-Polydactyly-Hydrocephalus (MPPH) Syndrome

This rare developmental syndrome is characterized predominantly by prenatal onset MEG and bilateral perisylvian PMG. Hydrocephalus and postaxial polydactyly are more variable features seen in nearly half of reported individuals [Kariminejad et al., 2012; Mirzaa et al., 2013]. A subset of children also has a distinctly thick corpus callosum (mega-corpus callosum). De novo germline mutations in three core PI3K-AKT-mTOR pathway genes are now known to be associated with MPPH including, in order of frequency, PIK3R2, CCND2, and AKT3 [Rivière et al., 2012; Nakamura et al., 2014; Mirzaa et al., 2014] (Fig. 2A–C, G–L). These mutations are mostly germline mutations with a narrow mutational spectrum. For example, a single recurrent PIK3R2 (p.Gly373Arg) mutation has been reported in most MPPH children. Postaxial polydactyly is a more frequent feature in PIK3R2 versus CCND2-positive children.

PI3K-AKT-mTOR RELEATED MEG AND HME: MOLECULAR SPECTRUM AND INSIGHTS INTO MOLECULAR PATHOGENESIS

Mutations of the above mentioned PI3K-AKT-mTOR pathway genes, whether loss of function mutations of PTEN or gain of function mutations of PIK3CA, AKT3, and PIK3R2, all share a common functional endpoint, namely activation of the pathway (Table IV). Mutations of PTEN, PIK3R2, and AKT3 have been predominantly germline. Whereas mutations of PIK3CA and the Proteus syndrome gene, AKT1, have been mosaic, providing a molecular explanation for the wide phenotypic variability in their attendant overgrowth phenotypes [Lindhurst et al., 2011]. The phenotypic spectrum of PIK3CA-related disorders is particularly wide and, while the exact mechanisms by which mutations result in these manifestations are currently under study, some preliminary genotype–phenotype correlations can be suggested. For example, the same PIK3CA mutation (p.Glu545Lys) has been identified in the four children with HME so far [Lee et al., 2012]. The mutational spectrum of MCAP syndrome, on the other hand, is wide and has not included any of the so-called mutation “hotspots” seen in cancer (p.Glu542Lys, p.Glu545Lys, p.His1047Arg, p.His1047Leu) [Samuels and Ericson, 2006; Samuels and Waldman, 2010; Rivière et al., 2012; Mirzaa et al., 2013]. While these mutations are all activating, they may not as robustly activating as those seen in cancer, given that the cancer risk in these phenotypes does not appear to be much increased; although natural history, and particularly cancer risk, data are lacking.

Mutations of the above mentioned PI3K-AKT-mTOR pathway genes, whether loss of function mutations of PTEN or gain of function mutations of PIK3CA, AKT3, and PIK3R2, all share a common functional endpoint, namely activation of the pathway.

For more than a decade now, hyperactivation of the mTOR signaling downstream of PI3K-AKT (due to loss of function mutations of TSC1 and TSC2, for example) have been considered to provide a pathological link between TSC, HME, and FCD by extensive studies [Crino, 2007; Lim and Crino, 2013]. Further, mTOR inhibition reversed neuronal hypertrophy in Pten-deficient mice and ameliorated a subset of Pten-associated abnormal behaviors, thereby substantiating evidence that the mTOR pathway downstream of PTEN is critical for its complex phenotype [Kwon et al., 2003; Zhou et al., 2009]. However, the recent identification of mutations in CCND2 in MPPH syndrome sheds a novel insight into the molecular pathology of these phenotypes [Mirzaa et al., 2014]. CCND2 is a member of the D-type cyclin family critically required for G1/S transition during the cell cycle [Matsushime et al., 1991; Inaba et al., 1992; Ross et al., 1996; Glickstein et al., 2006, 2009]. Identified mutations within CCND2 affect highly conserved terminal residues that include targets for glycogen synthase kinase 3β (GSK-3β)-phosphorylation and, ultimately, its' ubuiquitin mediated degradation [Kida et al., 2007]. Recent data demonstrate accumulation of degradation resistant CCND2 in individuals with MPPH and also, interestingly, in lymphoblastoid cell lines of individuals with upstream mutations in PIK3CA, PIK3R2, and AKT3 [Mirzaa et al., 2014]. These data implicate for the first time the involvement of another critical effector pathway downstream of PI3K-AKT: the cell cycle pathway downstream of GSK3b (Fig. 3). Clearly, further investigation is necessary to determine the detailed molecular mechanisms of brain overgrowth in this pathway and how these various critical downstream pathways interact in these phenotypes.

Recent data demonstrate accumulation of degradation resistant CCND2 in individuals with MPPH and also, interestingly, in lymphoblastoid cell lines of individuals with upstream mutations in PIK3CA, PIK3R2, and AKT3.

THERAPIES AND FUTURE DIRECTIONS

The past few years have witnessed exciting advances in our understanding of the molecular pathogenesis of brain overgrowth. It follows that individuals with disorders of the PI3K-AKT-mTOR and the interacting Ras/MAPK pathways may have the option in the future of pathway-based rational treatment. Some of the key functional deficits of these disorders, such as intellectual disability, autism, epilepsy, hydrocephalus, are naturally attractive targets for treatment that today are still treated empirically. For example, the epilepsy associated with some MEG syndromes is treated with anti-epileptic drugs that do not specifically address the molecular defects that we now know to be the basis of their disease. The use of numerous small molecule PI3K-AKT-mTOR pathway inhibitors to alleviate some of these developmental defects is currently under study. While specific therapy is already available for the downstream TSC-mTOR pathway by everolimus [Kingwell, 2013; Krueger et al., 2013], it is clear that the brain and somatic overgrowth phenotypes associated with upstream PI3K-AKT-mTOR pathway mutations result from increased activation of multiple pathways downstream of AKT, leading us to predict that successful treatment strategies will need to downregulate more than one of these pathways.

Some of the key functional deficits of these disorders, such as intellectual disability, autism, epilepsy, hydrocephalus, are naturally attractive targets for treatment that today are still treated empirically. For example, the epilepsy associated with some MEG syndromes is treated with anti-epileptic drugs that do not specifically address the molecular defects that we now know to be the basis of their disease.