Mechanistic Target of Rapamycin and Megalencephaly: Novel Research Strategies for Therapeutic Discovery

Abstract

Megalencephaly (ME) is a malformation of cortical development defined by an enlarged brain. Individuals with ME often suffer from drug resistant epilepsy, intellectual disability, and autism spectrum disorder. Several clinical ME subtypes result from pathogenic variants in mTOR pathway genes (MPG) which cause diffuse brain overgrowth likely as a consequence of hyperactive mechanistic target of rapamycin (mTOR) signaling during brain development. Unfortunately, resected surgical or post-mortem ME brain tissue specimens are not widely available, and thus, there is only limited understanding of the histopathology of MPG associated ME. Thus, research strategies including new mouse models and human cerebral organoids have been developed to study the developmental pathogenesis of ME linked to MPG variants. These model systems provide a platform to study the mechanisms leading to brain overgrowth in ME as well as the establishment of the epileptic network. Perhaps most compelling, pre-clinical research approaches in ME models may pave the way for therapeutic development that could be deployed in utero to prevent ME formation.

Megalencephaly (ME) is a common malformation of cortical development (MCD) defined as an abnormally enlarged brain (occipito-frontal head circumference size >2 standard deviations over the mean for age). ¹ ME is associated with a heterogeneous, but durable clinical phenotype including drug resistant epilepsy (DRE), intellectual disability (ID), and autism spectrum disorder (ASD) as well as motor, coordination, and vision disorders that persists into adulthood. While epilepsy surgery is used to treat DRE in other MCD subtypes, because of the challenges in localizing the seizure onset zone in ME and the multifocal nature of epilepsy in ME, individuals with ME are typically not candidates for resective surgery. Thus, from a research perspective, few studies have analyzed human ME brain tissue specimens to define the characteristic histopathological features or investigate the mechanisms causing brain overgrowth. Clearly, there is a public health mandate to understand the pathogenesis of brain overgrowth in ME and to develop novel therapeutic strategies for DRE, ID, and ASD associated with ME.

Enlarged brain size in ME is due to increased growth of cerebral structures related to abnormal brain development that results clinically in macrocephaly. ² There is a particularly relevant association between ME and gene variants encoding components of the mechanistic target of rapamycin (mTOR) signaling pathway (mTOR pathway genes, MPG; Figure 1 ). mTOR is a serine/threonine kinase that exists in two functional complexes: mTOR complex-1 (mTORC1) defined by protein binding to Raptor, and mTORC2, defined by binding to Rictor. All known MPG variants causing MCD appear to affect mTORC1 signaling more substantially than mTORC2 (however, see 3). mTOR inhibitors (mTORi) such as rapamycin (sirolimus) and everolimus act primarily on mTORC1 with nominal effect on mTORC2. The mTOR signaling cascade is an integral regulator of brain growth, metabolism, and neural progenitor cell proliferation. ¹ Indeed, many of the components of the mTOR pathway are most highly expressed in the brain and certain brain cell types are under tight mTOR-regulatory control.^2,3 The primary function of mTOR signaling is regulation of protein translation in response to a converging set of 3 signaling vectors, for example, growth factors, cellular energy cues, and cellular amino acid levels ( Figure 1 and see ⁴ ) that govern mTOR pathway activation or inhibition. Interestingly, MPG variants causing ME are identified in each signaling vector suggesting a pivotal role of mTOR signaling in modulating brain size. Downstream, the primary substrates of mTORC1 are p70S6 kinase (Thr389) which in turn phosphorylates ribosomal S6 protein (pS6; Ser 235/236, Ser 240/244), and the eukaryotic translation initiation factor 4E (eIF4E)-binding protein 1(4E-BP1; Thr37/46). These target proteins are part of the 5’-translation initiation complex that governs cap-dependent translation of mRNA in all human cells. Thus, a likely consequence of MPG mutations is altered protein translation which may be a critical mechanism responsible for brain overgrowth in ME.

Figure 1.

The mTOR signaling cascade. Growth factors, such as insulin-like growth factor (IGF1), signal to mTORC1 via a canonical pathway including PI3K, PTEN, and AKT to the TSC1/TSC2/TBC1D7 complex (gray) and then to Rheb. Glucose and energy levels (ATP) signal to mTORC1 through LKB1/STRADA/MO25 via AMPK (orange). Ambient levels of cellular amino acids signal to mTOR via a series of regulatory protein complexes (GATOR1, GATOR2, KICSTOR; green). The lysosome is where mTORC1 will reside in its active conformation. Downstream mTOR effectors modulate cellular metabolism and protein translation. All mTOR pathway regulators associated with ME gene variants are denoted in red.

MPG variants are detected in autosomal recessive ME or as de novo germline variants in sporadic ME as a subtype of “mTORopathies”.^5,6 MPG variants, for example, PI3KCA, ⁷ AKT3, ⁸ MTOR, ⁹ PTEN, ¹⁰ DEPDC5, ¹¹ RHEB, ¹² STRADa, ¹³ KTPN, ¹⁴ and TBC1D7, function as either gain-of-function or loss-of-function variants and cause constitutive hyperactivation of the mTOR pathway in fetal brain progenitor cells impairing assembly of the cortex and presumably leading to brain overgrowth. There are very limited published post-mortem analyses of ME associated with MPG variants, that is, AKT3,^7,8 PI3KCA,^15,16 and STRADA. ¹⁷ Reported histopathological findings in ME include an enlarged brain with increased brain weight and disorganized structure which may be asymmetric and may differentially affect cortical versus subcortical structures. Cerebellar tonsillar ectopia and callosal dysgenesis may be seen. Cortical gyral abnormalities such as polymicrogyria or focal cortical dysplasia, and laminar abnormalities including enlarged (cytomegalic) neurons, disorganized cortical lamination, and subcortical heterotopic neurons are observed.

Defining ME pathogenesis has clear therapeutic development implications for DRE, ID, and ASD or even prevention of ME as technologies advance with prenatal genetic screening and fetal therapies. Indeed, enlarged fetal head circumferences can be detected as early as the second trimester via fetal ultrasound or magnetic resonance imaging (MRI) and thus, there may be an early window for genetic ascertainment and therapeutic intervention during fetal development. Since most current evidence suggests that MPG variants drive brain overgrowth in ME via enhanced mTOR pathway signaling, targeting the mTOR pathway seems a logical first step. However, the downstream molecular events linking augmented mTOR signaling to brain overgrowth are largely unknown and thus, further pre-clinical investigation using rodent and human cellular models is necessary. This review summarizes how experimental models including transgenic mouse and human cerebral organoids (hCOs) of two ME syndromes linked to MPG variants can help define ME pathogenesis. The first, Polyhydramnios, MEgalencephaly, and Symptomatic Epilepsy (PMSE) syndrome is caused by biallelic loss-of-function mutations in STRADA, and the second, KPTN (kaptin)-related disorder (KRD) is caused by biallelic loss-of-function mutations in KPTN.

Mouse Models of ME

All MPG variants result in mTOR pathway hyperactivation that produces the phenotypes typically observed in mTOR-associated ME including seizures, cortical dyslamination, and pS6 labeled cytomegalic neurons. ⁵ Recently, two rodent models of ME have emerged as promising platforms to investigate ME pathogenesis and test novel therapeutics.^18,19

To model PMSE, a rodent model was created using a loxP neocassette to permanently knockout (KO) exons 9–13 of mouse Strada on a C57BL/6N background at the earliest stages of embryonic development creating a global KO. ¹⁸ Homozygous (Strada-/-) animals have a markedly reduced survival rate compared to both wildtype (Strada +/+) and heterozygous (Strada+/-) littermates with the majority of Strada-/- animals dying shortly after birth. Homozygous animals that do survive are smaller than littermates and rarely live past postnatal day 5 with a single animal living ten months. Histological examination of Strada-/- animals revealed ventriculomegaly but no changes in the overall size of the cerebral cortex. However, increased mTOR pathway signaling was observed, as measured by pS6, in the cortex, hippocampus, and thalamus (Figure 2). Examination of the superficial cortical layer marker Cux1 showed an increase in the number of neurons in deeper cortical layers and subcortical white matter compared to Strada+/- controls. ¹⁸

Figure 2.

Enhanced mTOR signaling in Strada KO mouse model. Increased phosphorylation of ribosomal S6 protein (pS6, red) in Strada KO mouse compared to wildtype mouse (DAPI blue counterstain; Adapted from Dang et al, 2020). Increased pS6 is observed in cerebral cortex (larger arrows), hippocampus (small arrow), and thalamus (asterisk). Scale bar, 200 microns.

Together these data demonstrate that several hallmarks of PMSE are recapitulated with KO of exons 9–13 in rodents including increased mTOR pathway signaling in neurons, cortical dyslamination, and ventriculomegaly. However, there are several limitations to this model. First, Strada-/- animals do not have spontaneous behavioral or electrographic seizures, which are universal in PMSE patients. Second, the severe perinatal lethality of these animals makes it difficult to obtain sufficient specimens to study and it is therefore unknown whether mTOR inhibition can rescue any of the rodent phenotypes. Future experiments attempting to rescue the perinatal lethality are clearly warranted.

KICSTORopathies (ME caused by variants in the KICSTOR complex, ITFG2, KICS2, SZT2, KPTN) have recently emerged as a new subclassification of MPG variant-associated MCD. A model of one KICSTORopathy, KRD, has produced a mouse strain that aligns with patient phenotypes. ¹⁹ The mouse model was created using a splice-trap neocassette inserted between exons 1 and 2 of the mouse Kptn gene, resulting in termination of gene transcription and producing a KO (Kptn-/-). MRI skull reconstruction and histopathological examination of Kptn-/- mouse brain sections demonstrated a post-natal increase in skull and brain size compared to Kptn+/+ control mice. In addition, mild cortical dyslamination was also observed. mTOR pathway activation (neuronal pS6 labeling) was also observed in the cortex, similar to observations in Strada-/-. However, unlike Strada-/- animals where pS6 positive neurons were observed in the thalamus and hippocampal subfields, increased pS6 labeling was only observed in the dentate gyrus in Kptn-/- animals. Behavioral assays revealed that Kptn-/- mice had increased locomotion, supporting a hyperactivity phenotype and anxiety-like behaviors were observed using the light/dark transition test where Kptn-/- animals spent more time in the dark and traveled to the light areas of the cage less frequently than Kptn+/+ mice. Lastly, assays to test hippocampal and non-hippocampal memory yielded non-significant results with the exception of an olfactory habituation/dishabituation assay where Kptn-/- animals were observed to spend more time interacting with non-novel stimuli compared to controls. ¹⁹ While these data demonstrate concordance between several KRD phenotypes and Kptn-/- mice, unlike KRD patients, Kptn-/- mice do not have seizures.

hCOs as Models of ME

hCOs and assembloids provide human-specific insight into ME. hCOs derived from induced pluripotent stem cells (iPSCs) are a central platform for modeling early human corticogenesis in vitro.^20,21 In three-dimensional culture, they self-organize into ventricular-like zones and cortical plate–like layers, recapitulating key cytoarchitectural and transcriptional features of the developing cortex.^22,23 These systems are informative for disorders of brain size, because outer radial glia (oRG)-rich outer subventricular zones are expanded in human neocortex and sensitive to mTOR signaling, linking progenitor behavior to cortical overgrowth.^24,25

Directed patterning generates dorsal hCOs enriched in glutamatergic neurons and ventral hCOs resembling ganglionic eminences. When dorsal and ventral hCOs are fused into assembloids, GABAergic interneurons born in ventral organoids migrate tangentially into dorsal hCOs and integrate into excitatory circuits, enabling direct visualization of early interneuron development and probing how connectivity and excitatory–inhibitory (E/I) balance are perturbed.^26,27 Compared with mouse models, hCOs and assembloids offer several advantages but also important limitations. They capture human-specific progenitors such as oRGs ²⁴ and preserve human temporal dynamics of neurogenesis; when generated from patient-derived iPSCs, they allow direct modeling of individual genetic phenotypes.^28,29 However, most hCOs remain developmentally immature, vary between lines and differentiations, lack clear lamination and subcortical structures, fall short of the cellular complexity of cortex, and provide only a partially defined epileptic phenotype.^20,21,30 In practice, mouse models define circuit- and organism-level consequences of mTOR dysregulation, whereas hCOs and assembloids pinpoint early human-specific cellular events that drive brain overgrowth and epileptogenesis.

Several ME subtypes have been modeled using hCOs and organoid models of germline mTORopathies reveal shared and gene-specific mechanisms of brain overgrowth. For example, PTEN loss produces enlarged and folded hCOs with expanded progenitor zones and delayed neuronal differentiation, phenomena not reproduced when PTEN is deleted in mouse organoids, underscoring species-specific differences in progenitor behavior and growth responses.^22,28 hCO models of other mTOR regulators, including TSC1/2 and additional PI3K–AKT–mTOR components, similarly support the idea that mTOR hyperactivation promotes progenitor expansion and cytomegaly as a cellular substrate for ME.^25,31,32 PMSE syndrome is an example in which hCOs reveal an early pathogenic window. Multimodal work in mouse and human neurons shows that STRADA loss increases mTOR activity, cell size, neuronal excitability, and disrupts cortical lamination.^17,33–35 Patterned dorsal and ventral hCOs show that STRADA loss perturbs excitatory and inhibitory neurogenesis, with mTORC1 hyperactivation and partial rescue by rapamycin linking germline STRADA mutations, ME, and early network hyperexcitability.^36,37

Assembloids bring multiple lineages together in the context of a developing forebrain network, mature more rapidly than single-region hCOs, and are compatible with functional readouts such as multielectrode array recordings ³⁸ and calcium imaging. ³⁹ Although their application to mTORopathies with ME is still in its early stages, these systems are being applied to other genetic neurodevelopmental disorders to probe similar mechanisms, underscoring common pathogenic mechanisms of interneuron development deficits and E/I imbalance.^26,27,40 hCO models of mTORopathies have important translational implications for both ME and epilepsy and indeed, organoid studies inform timing and targets for emerging therapies in ME and epilepsy. They help define the developmental windows during which mTOR dysregulation first derails human cortical development: PTEN deletion alters growth trajectories over several weeks of early corticogenesis,^22,28 whereas STRADA-mutant hCOs show overgrowth and perturbed progenitor dynamics within the first two weeks of development. ³⁶ hCOs also provide a mechanistic framework for mTOR-targeted therapies. Evidence from hCO systems demonstrates that pharmacologic inhibition of the PI3K–AKT–mTOR axis can normalize progenitor proliferation, reduce cytomegaly, and at least partly rescue overgrowth phenotypes in several mTORopathies.^22,37,41 By incorporating functional readouts, hCOs and assembloids can link structural abnormalities to network hyperexcitability and serve as a preclinical platform to evaluate how more targeted therapies influence both growth and E/I balance.^26,42 Together, these findings point to a therapeutic window in fetal or very early postnatal life, highlighting the need for time-sensitive disease-modifying strategies in treating mTORopathies (see below). ³⁴

Fetal Therapeutics for Human ME

The neurobehavioral phenotypes of ME including DRE, ID, and ASD provide significant challenges for individuals with ME. Based on fetal imaging studies, ME pathological progression begins in utero. Altered brain size and disruption of brain structure seen in mouse models of ME serve as a key feature of ME.^34,35,43 Thus, a compelling therapeutic possibility is fetal intervention during the critical period of neurogenesis which offers a window of opportunity to potentially prevent brain overgrowth and the establishment of epileptic networks.

Several key factors must be considered when developing safe and effective fetal therapy for ME:

i. Necessity and Justification. The first consideration is always whether the therapy needs to be fetal, based upon whether the developmental process targeted occurs in utero. For all MPG variants causing ME, fetal imaging clearly shows onset early in brain development (second trimester). In PMSE, the loss of STRADA protein and consequent mTOR hyperactivity is associated with evidence of a neural migratory defect radiographically and histologically, suggesting a prenatal therapeutic target.^17,33

ii. Targetability. For ME disease modification, therapeutic targets can include genes, enzymes, or metabolites. Treatments can include gene therapy, antisense oligonucleotide delivery, or pharmacological therapy with inhibitors, activators, or molecular chaperones. As MPG encode proteins involved in mTOR pathway regulation, mTOR inhibition presents an attractive option for treating these disorders. In PMSE, rapamycin (sirolimus) treatment in progenitor cell and fetal mouse models has been shown to prevent key disease features. ³⁴

iii. Timing. Initiation of prenatal therapy is limited by the timing of disease diagnosis. Known genetic predispositions and early radiographic findings can enable prenatal discovery of ME. Once the correct mechanistic target is established, it is critical to optimize therapeutic timing to maximize benefit relative to risk. Risks can be minimized by selecting the latest developmental timepoint when the therapy will still be effective. For PMSE, postnatal sirolimus therapy demonstrated transient benefits in seizure reduction, which diminished over a five-year course, likely reflecting the necessity of altering epileptic network formation during the earlier fetal periods of corticogenesis in order to achieve a durable effect. ³⁴ Based on data supporting that neurogenesis in humans continues from gestational weeks (GW) 8 through 30 and that human inhibitory interneuron progenitor pools can persist through GW 39, we postulate that therapy to interrupt aberrant corticogenesis in mTOR-driven ME should be initiated sometime between GW 8–39, with greater efficacy from earlier initiation.^44–46 However, it is critical to balance this with increased risks at earlier timepoints, and therapy should not be initiated before the completion of organogenesis at GW 8.

iv. Therapeutic Delivery Method. Whether or not a fetal therapy has good transplacental penetrance can determine whether it can be given to the mother or will instead need to be delivered directly to the fetus. In the latter case, it needs to be determined whether the therapy crosses the blood-brain barrier (BBB), which is often less of a factor with prenatal therapies as the BBB is incompletely formed through the second trimester, or whether it will need to be injected directly into the fetal central nervous system. ⁴⁷

v. Dosing. Initial dosing for fetal therapies can be informed by precedent from clinical case reports as well as interspecies conversions from animal studies. Drug bioavailability can be estimated through circulating levels in maternal serum concentrations. However, case reports of sirolimus use suggest that concentrations in the neonate can be higher than those in the mother, due to possible active transport mechanisms across the placenta. ⁴⁸ Additionally, the time course of treatment needs to be determined—whether the goal is to interrupt a time-limited developmental process or whether the therapy needs to be continued postnatally. In the first case report of sirolimus delivered prenatally for obstructive cardiac rhabdomyomas in TSC, Barnes and colleagues found a persistent inverse correlation between tumor size and sirolimus treatment, such that the medication required postnatal continuation to prevent tumor regrowth. ⁴⁸ This parallels the effect of everolimus treatment for TSC-associated subependymal giant cell astrocytomas (SEGAs) in the EXIST trials, where tumors can regrow during pauses in mTOR inhibition.^49–51

vi. Risks and Off-Target Effects. Both long- and short-term effects on both the fetus/child and the mother will need to be carefully monitored. Prenatal sirolimus in TSC has been associated with intrauterine growth restriction. ⁵² Additionally, Tsai and colleagues found that wildtype mice treated with fetal rapamycin demonstrated postnatal behavioral deficits, suggesting that supratherapeutic mTOR inhibition needs to be monitored to avoid unfavorable developmental outcomes. ⁵³ In contrast to postnatal therapy, fetal therapy creates a second patient in the mother, who may also be affected by the treatment. Thus, it is crucial to closely monitor maternal effects, such as immunosuppression with mTORi, throughout the course of treatment.

vii. Outcomes. The success of fetal therapy for ME can be measured through anatomical and functional outcomes. Anatomical measures can include head circumference and other radiographic pathologies associated with ME including increased brain size, gyral abnormalities, and corpus callosal dysgenesis. Functional measures could include survival, gestational age at delivery, seizure onset, frequency and type, need for anti-seizure medications, and longer-term tracking of neurodevelopmental milestones.

Fetal therapeutics for ME will require a coupling of targeted therapies, determined through translational research, with delivery techniques designed to optimize safety and efficacy, capitalizing on critical developmental windows to prevent brain overgrowth and interrupt the establishment of epileptic networks. Preventative fetal interventions are poised to redefine the landscape of ME.