RHEB/mTOR hyperactivity causes cortical malformations and epileptic seizures through increased axonal connectivity

Proietti Onori M, Koene LMC, Schäfer CB, Nellist M, de Brito van Velze M, Gao Z, Elgersma Y, van Woerden GM. — PLOS Biology 19(5): e3001279 (May 26, 2021). DOI: 10.1371/journal.pbio.3001279

Citation

• Journal: PLOS Biology, Volume 19, Issue 5, e3001279
• Published: May 26, 2021
• DOI: 10.1371/journal.pbio.3001279
• License: Creative Commons CC-BY 4.0
• Corresponding author: Geeske M. van Woerden (g.vanwoerden@erasmusmc.nl), Department of Neuroscience, Erasmus Medical Center, Rotterdam, and ENCORE Expertise Center for Neurodevelopmental Disorders
• Uploaded PDF: storage/raw/uploads/slack/2026-04-18/1776499664496-pbio.3001279.pdf

Background (from the paper)

• RHEB (Ras Homolog Enriched in Brain) is a small GTPase and the direct upstream activator of mTORC1; it is normally kept inactive by the TSC1/TSC2 GAP complex.
• Germline and somatic mosaic variants in the PI3K-AKT-mTOR pathway (TSC1, TSC2, AKT3, PIK3CA, PIK3R2, MTOR, DEPDC5, NPRL2/3) cause a spectrum of "mTORopathies" — malformations of cortical development (MCD) including focal cortical dysplasia (FCD) type II, hemimegalencephaly, and megalencephaly — presenting with intellectual disability, autism, and early-onset refractory epilepsy.
• Reijnders et al. 2017 (Mol Psychiatry) first identified de novo missense variants in RHEB in patients with intellectual disability, megalencephaly, and epilepsy; p.P37L and p.S68P were the two recurrent variants. This paper mechanistically interrogates the p.P37L variant.
• A longstanding open question: where does the seizure actually originate in mTORopathy — from the visible malformation itself, or from cells distributed beyond it?

Study design

• Approach: In utero electroporation (IUE) at embryonic day E14.5 in mice to introduce constructs into layer 2/3 pyramidal neurons of the somatosensory cortex, expressing GFP alone (control), wild-type RHEB (RHEB-WT), or the patient RHEB p.P37L variant (RHEB-P37L).
• Phenotyping: post-hoc histology for heterotopic nodule formation, ectopic neuron placement, cell size, and S6 phosphorylation (mTORC1 readout); continuous 24/7 video-EEG at ~P45-P60 for spontaneous seizures; behavioral testing.
• Biochemistry: HEK293T transfection + co-immunoprecipitation to test whether RHEB-P37L can still bind TSC1/TSC2 and whether the TSC complex can still stimulate its GTPase activity (GAP assay).
• Pharmacology: rapamycin rescue experiments.
• Causality dissection:
  • Anatomic: compare mice with vs. without heterotopic nodules.
  • Temporal: tamoxifen-inducible Cre-ERT2 switched from RHEB-P37L back to WT at P14 — after malformation has formed — to ask whether ongoing mTOR hyperactivity is required for seizures.
  • Circuit: express tetanus toxin light chain (TeTxLC) to silence vesicle release specifically in RHEB-P37L neurons, or specifically in contralaterally projecting callosal axons, to localize where the excitability originates.

Key findings

1. RHEB p.P37L is a GAP-resistant gain-of-function variant. RHEB-P37L binds TSC1/TSC2 normally but is resistant to TSC-GAP-stimulated GTP hydrolysis, so it stays in the active GTP-bound state and constitutively drives mTORC1.

2. IUE of RHEB-P37L recapitulates human mTORopathy pathology.

   • Strongly elevated phospho-S6 (mTORC1 hyperactivity) in electroporated neurons.
   • Pyramidal neuron soma hypertrophy.
   • Heterotopic nodules of neurons failing to migrate to their proper cortical layer (subcortical / white-matter heterotopia resembling patient imaging).
   • Dysplastic cortex with disrupted lamination and ectopic neurons at positions incompatible with L2/3.

3. Reliable, reproducible spontaneous tonic-clonic seizures. ~100% of RHEB-P37L mice developed spontaneous tonic-clonic seizures with cortical ictal EEG correlates; WT and RHEB-WT-IUE controls did not. Seizure onset was typically in young-adult animals.

4. Rapamycin rescues biochemistry, histology, AND seizures. Systemic rapamycin normalized S6 phosphorylation, prevented heterotopic nodule formation when given perinatally, and abolished seizures in already-affected adults — confirming that mTORC1 hyperactivity is the driver and the phenotype is mTOR-inhibitor-reversible.

5. The heterotopic nodule is neither necessary nor sufficient for seizures. A subset of RHEB-P37L mice had no detectable heterotopic nodule on histology yet still developed spontaneous seizures; conversely, nodule presence did not cleanly predict seizure severity. This uncouples the visible MRI-scale malformation from the epileptogenic substrate.

6. Ongoing mTOR hyperactivity, not the frozen malformation, drives seizures. In Cre-ERT2 experiments where RHEB-P37L was switched back to WT at P14 (after cortical layering is complete), the malformation persisted but seizures did not develop — demonstrating that cortical-level malformation alone is insufficient; continuous mTORC1 hyperactivity in the affected cells is required.

7. Epileptogenesis is driven by enhanced long-range axonal connectivity, not local circuit changes.

   • RHEB-P37L neurons have enlarged axonal arbors and markedly increased contralateral callosal projections, with more synaptic puncta on normal-appearing neurons in the contralateral hemisphere.
   • Silencing vesicle release (TeTxLC) specifically in RHEB-P37L neurons abolished seizures.
   • Silencing release only at the contralateral projection targets also abolished seizures, despite the local malformation being untouched.
   • Conclusion: seizures originate because mTOR-hyperactive neurons send aberrantly many excitatory outputs onto structurally normal, genetically wild-type cortex in the opposite hemisphere, making that distal cortex hyperexcitable.

Paper's own takeaway

mTORopathies cannot be understood as "the visible malformation is the seizure focus." Even when a focal / mosaic mTOR-pathway variant produces a visible malformation on imaging, the epileptogenic substrate extends far beyond it through aberrantly enhanced long-range axonal connectivity onto normal-appearing cortex. Rapamycin rescue of both biochemistry and seizures supports mTOR-inhibitor therapy as a mechanism-directed treatment in mTORopathy epilepsies.

Relevance to Levi

This paper is directly relevant to Levi's current leading etiologic hypothesis (mosaic PI3K-AKT-mTOR pathway variant) and specifically changes how a structurally unremarkable brain MRI should be interpreted in his context.

• RHEB is an explicit candidate gene for Levi. Reijnders et al. 2017 described exactly the phenotype Levi has — intellectual disability, megalencephaly, and epilepsy — from de novo RHEB missense variants. Levi has symmetric proportional overgrowth at the 99th percentile from ~12 months (including head circumference), regression, autism, and DEE-SWAS. RHEB deserves explicit coverage confirmation in his trio exome / WGS reports and should be on any mosaic-sensitive PROS / mTOR-pathway panel ordered on buccal swab or skin punch.
• Structurally unremarkable MRI does NOT exclude mTOR-driven epilepsy. Levi's April 7, 2026 MRI showed only nonspecific R>L periventricular deep white-matter FLAIR signal with no cortical dysplasia, heterotopia, TSC stigmata, or hemimegalencephaly. This paper shows (finding #5) that seizures from a mTOR-pathway gain-of-function can occur without a visible heterotopic nodule, and (finding #7) that the true seizure substrate is long-range aberrant connectivity onto normal-appearing cortex. In a low-VAF mosaic scenario where too few cells are affected to produce an MRI-scale malformation, the same mechanism could still produce widespread SWAS-like cortical hyperexcitability. This meaningfully softens the argument "normal MRI ⇒ down-weight mTORopathy" and supports keeping the mosaic PI3K-AKT-mTOR branch as Theory 1 even after the MRI.
• Mechanism reconciles DEE-SWAS physiology. DEE-SWAS is defined by near-continuous bilateral multifocal spike-wave activation in sleep. A mechanism in which mTOR-hyperactive cells drive hyperexcitability across structurally normal, widely distributed cortex — rather than a single focal onset zone — is compatible with the multifocal, bilateral, sleep-activated pattern Levi showed on the Stanford EMU (wake SWI 78%, sleep SWI 95-100%, multifocal discharges at O1/O2/P4/T3/T4-T6).
• Rapamycin / mTOR-inhibitor rescue of both biochemistry and seizures is direct pre-clinical support for everolimus/sirolimus as a mechanism-directed treatment if a PI3K-AKT-mTOR pathway variant (germline or mosaic) is ever identified in Levi. This should be captured in the treatments workspace as a contingent, diagnosis-triggered lever with non-trivial prior probability given the current differential.
• Directly raises the diagnostic value of tissue-based mosaic-sensitive sequencing. If the epileptogenic substrate can be driven by a minority of mTOR-hyperactive cells whose malformation is invisible on MRI, then the clinical bar for chasing a low-VAF somatic variant should be lower, not higher. Reinforces the current top-ranked diagnostic: buccal swab / skin punch to a PROS or PI3K-AKT-mTOR panel with explicit deep coverage of RHEB, MTOR, AKT3, PIK3CA, PIK3R2, TSC1, TSC2, DEPDC5, NPRL2, NPRL3, PTEN.
• Confirms that age-at-switch-off matters. The Cre-ERT2 rescue result (finding #6) shows that stopping ongoing mTORC1 hyperactivity after malformation has frozen prevents seizures — suggesting that mTOR inhibition could be therapeutic in Levi even if any subtle structural substrate is already fixed.
• Outreach value. The van Woerden / Elgersma groups at ENCORE / Erasmus MC in Rotterdam are an active mTORopathy research program with both mouse and patient-facing sides; they are a reasonable long-tail outreach target for the somatic-mosaicism / mTORopathy line of investigation, complementary to the Mirzaa / Poduri / Aldinger groups already prioritized.

Limitations / interpretive caveats

• This is a mouse model with forced overexpression via IUE, not a knock-in, so absolute expression levels may exceed what a human mosaic variant produces. The qualitative mechanistic claims (GAP resistance → mTORC1 hyperactivity → enlarged axonal arbors → distal hyperexcitability → seizures) are nonetheless well-supported.
• The paper studies one specific variant (RHEB p.P37L); generalization to other PI3K-AKT-mTOR-axis variants (PIK3CA, MTOR, AKT3, etc.) is biologically plausible but not directly demonstrated.
• The "seizures without nodule" result (finding #5) is an observation in a subset of IUE mice; the paper interprets it as evidence that nodule presence is not required, but the underlying VAF distribution in those animals is not directly comparable to a low-VAF human somatic mosaic. Still, the direction of the evidence is unambiguous.
• Tetanus-toxin silencing argues for a necessary role of synaptic output from RHEB-P37L cells; it does not rule out additional contributions from local circuit changes or network-level plasticity downstream.

Provenance

• Ingested 2026-04-18 from a PDF uploaded by the user via Slack (original filename pbio.3001279.pdf).
• Stored at storage/raw/uploads/slack/2026-04-18/1776499664496-pbio.3001279.pdf.
• Canonical external source: PLOS Biology open access article, DOI 10.1371/journal.pbio.3001279.
• Companion research note: content/research/notes/2026-04-18-rheb-mtor-axonal-connectivity-mechanism.md.