TSC is an autosomal dominant multi-system disorder caused by loss-of-function variants in TSC1 (hamartin, 9q34) or TSC2 (tuberin, 16p13.3). The reason these two very different genes produce one disease is that their protein products work as a single functional unit. Hamartin and tuberin assemble into the TSC complex, whose job is to keep the small GTPase Rheb in its inactive (GDP-bound) state. Active Rheb is the direct switch that turns on mTORC1, the master regulator of cell growth, protein synthesis, and metabolism. Lose either subunit and the brake fails: Rheb stays active, mTORC1 runs constitutively, and cells grow and proliferate inappropriately — producing hamartomas (benign but space-occupying overgrowths) in nearly every organ.

This is why a single pathway explains a bewildering multi-organ phenotype, and why the lesions are typically focal rather than diffuse. A person inherits one defective allele in every cell, but a lesion only forms where a second hit knocks out the remaining good allele (Knudson two-hit, as in tumor-suppressor biology). Each tuber, angiomyolipoma, or SEGA is essentially a clone that has lost both copies. ~2/3 of cases are de novo.

Neurological features arise because mTOR overactivity disrupts neuronal migration, dendritic arborization, and synaptic pruning during cortical development.

• Cortical tubers — focal regions where migration and lamination failed; disorganized, dysmorphic neurons make them highly epileptogenic
• Subependymal nodules (SENs) — calcified periventricular nodules
• SEGAs — SENs that cross into frank growth near the foramen of Monro; risk of obstructive hydrocephalus
• Epilepsy in ~85% (often infantile spasms in the first year)
• ASD in 40–50%; intellectual disability in ~50% — burden correlates with tuber load and early seizure control

Systemic features

• Cardiac rhabdomyomas — often the earliest sign (prenatal/neonatal); typically regress as the mitotically active fetal myocardium matures
• Renal angiomyolipomas — lifelong hemorrhage risk
• LAM (lymphangioleiomyomatosis) — cystic lung disease, predominantly adult females
• Facial angiofibromas — appear in childhood, pathognomonic

Diagnosis: 2012 criteria use major and minor features (2 major or 1 major + ≥2 minor = definite diagnosis). A pathogenic TSC1/TSC2 variant is independently sufficient — a notable feature, since a positive genetic test "counts" even when too few clinical features are present, reflecting confidence in the gene-disease link. The old Vogt triad is present in only a minority.

TSC is the paradigm for mechanism-based therapy in neurogenetics. Because the entire disease flows from one defect — too much mTORC1 signaling — a drug that inhibits mTOR addresses the root cause rather than a downstream symptom. This is rare in neurogenetics, where most disorders still have only supportive care.

Everolimus is a rapalog: it binds the intracellular protein FKBP12, and the complex allosterically inhibits mTORC1. By restoring the brake the TSC complex can no longer apply, it shrinks the overgrowths that mTOR hyperactivity drove. FDA-approved for:

• TSC-associated SEGA (reduces tumor volume, can avoid neurosurgery and shunting)
• Renal angiomyolipomas (reduces size, decreases hemorrhage risk)
• Adjunctive therapy for TSC-associated refractory focal seizures

A conceptual caveat: rapalogs are cytostatic, not curative — they suppress growth, so lesions often regrow when the drug is stopped, and treatment is typically long-term.

Vigabatrin — first-line for TSC-associated infantile spasms, with a striking ~95% response rate versus ~50% for ACTH. Vigabatrin is an irreversible GABA-transaminase inhibitor that raises CNS GABA; why TSC spasms respond so much better than spasms of other causes is not fully understood but likely reflects the specific mTOR-driven circuit pathology. The EPISTOP trial (Kotulska et al. 2021) tested a radical idea: treat before seizures appear. Infants were monitored with serial EEG, and preventive vigabatrin was started the moment the EEG became epileptiform but while the child was still clinically seizure-free. Preventive treatment reduced the incidence of clinical seizures and drug-resistant epilepsy and improved neurodevelopmental outcomes at 24 months — establishing a predict-and-prevent paradigm in which the EEG is a biomarker that lets you intervene ahead of the disease.

Surveillance: regular brain MRI (SEGA monitoring to age 25), renal imaging (AML), echocardiography (infancy), CT chest (LAM screening in adult females), dermatology, ophthalmology, and serial EEG in infants — the last being precisely what makes the EPISTOP preventive strategy possible.

Fragile X is the most common inherited cause of intellectual disability and the most common single-gene cause of ASD. It results from a CGG repeat expansion in the 5'UTR of FMR1 (Xq27.3). The expansion sits in the untranslated region — it does not change the FMRP protein sequence at all. Instead, the disease is about whether the gene gets read.

Repeat ranges: normal <45; intermediate 45–54; premutation 55–200; full mutation >200. Once the repeat crosses ~200, it becomes a substrate for DNA methylation: the CpG island in the FMR1 promoter is heavily methylated, the chromatin condenses, and transcription shuts off. The result is transcriptional silencing — no FMR1 mRNA, no FMRP. So an expansion of non-coding DNA produces a functional null, an epigenetic loss of a perfectly normal protein.

Why losing FMRP matters — the mGluR theory. FMRP is an mRNA-binding protein that travels to dendrites and acts as a brake on local protein synthesis at the synapse. Normally, group-1 metabotropic glutamate receptor (mGluR5) signaling stimulates synaptic translation, and FMRP reins it in. Without FMRP, that translation runs unchecked: mGluR-dependent long-term depression is exaggerated, dendritic spines stay immature and overabundant, and synaptic plasticity is dysregulated. The phenotype is therefore not from a missing structural protein but from loss of translational control — which is why mGluR5 antagonists were a major (though clinically disappointing) therapeutic strategy.

Clinical features in affected males: moderate-to-severe ID, long face, prominent ears/jaw, macroorchidism (post-pubertal), anxiety, ADHD, hand flapping, gaze avoidance, joint hypermobility.

Females with full mutation: ~50% have some cognitive impairment. Because FMR1 is X-linked, a female's two X chromosomes are randomly inactivated; severity depends on what fraction of cells happen to silence the normal X versus the expanded one.

Anticipation: premutation alleles are unstable during maternal meiosis — the longer the repeat, the higher the chance it expands further in the next generation (>90 repeats → near-100% expansion risk to full mutation). Mechanistically, long repeats slip during DNA replication/repair. Paternal transmission of premutations is generally stable (and sperm rarely carry full mutations), so the full mutation is essentially always maternally transmitted.

Testing: FMR1 CGG repeat analysis (Southern blot/triplet-repeat PCR) — standard WES does NOT detect this. Must be specifically ordered.

The FMR1 premutation (55–200 repeats) is not clinically silent — and crucially, its diseases arise from the opposite molecular problem to Fragile X syndrome. In the full mutation the gene is silenced and there is too little product. In the premutation the repeat is not long enough to trigger methylation, so the gene stays on; in fact the cell senses the abnormal locus and transcribes it harder, producing elevated FMR1 mRNA (2–8× normal) studded with a long CGG hairpin. The damage is done by the mRNA itself, not by any protein deficiency.

This is a toxic RNA gain-of-function. The expanded-repeat transcript folds into stable hairpins that act as a sponge, sequestering RNA-binding proteins (e.g., Sam68, DGCR8, hnRNP proteins) away from their normal jobs and forming ubiquitin-positive intranuclear inclusions in neurons and astrocytes. A second contributor is RAN translation — repeat-associated non-AUG translation produces toxic homopolymeric peptides (e.g., FMRpolyG) even without a normal start codon. The mechanism is closely analogous to myotonic dystrophy type 1 (DM1), the other classic RNA-toxicity repeat disorder. The teaching point: same gene, same repeat, opposite mechanism — loss-of-protein at full mutation, gain-of-toxic-RNA at premutation — which is why the two ends of the spectrum cause entirely different diseases.

FXTAS (Fragile X-associated tremor/ataxia syndrome):

• Late-onset (>50 years), predominantly males
• Progressive intention tremor, cerebellar gait ataxia, executive dysfunction, neuropathy
• MRI hallmark: bilateral T2/FLAIR hyperintensity in the middle cerebellar peduncles (MCP sign)

FXPOI (primary ovarian insufficiency):

• Affects ~20–25% of female premutation carriers
• Premature menopause (<40 years), menstrual irregularity, infertility

Counseling: premutation females risk FXPOI + expansion to full mutation in offspring; premutation males risk FXTAS and transmit premutation (not full mutation) to all daughters. Cascade testing is critical.

Rett syndrome is X-linked dominant, caused by de novo loss-of-function variants in MECP2 (Xq28) — the gene-disease link established by the landmark report of Amir et al. 1999. It affects almost exclusively females, and the reason is instructive about X-linked biology. A female is a mosaic: random X-inactivation means roughly half her neurons express the mutant MECP2 and half the normal one, so she has a survivable mixture of healthy and affected cells. A hemizygous male has no such backup — every neuron is affected — so males with null variants typically have severe neonatal encephalopathy and early death; surviving males usually have somatic mosaicism, Klinefelter syndrome (47,XXY) (an extra X providing a normal copy), or hypomorphic variants. >95% of variants are de novo, the vast majority arising on the paternal X (explaining the female preponderance, since fathers pass their X only to daughters).

Why MECP2 loss is so damaging. MeCP2 is a reader of the epigenome: it binds methylated CpG sites across the genome and recruits co-repressor and chromatin-remodeling complexes to fine-tune transcription. It is expressed at extremely high levels in mature, post-mitotic neurons, where it acts as a genome-wide modulator ("rheostat") of neuronal gene expression rather than an on/off switch for a few genes. Its loss therefore does not kill neurons or block their birth — development is initially normal — but it derails the maintenance and refinement of synaptic gene programs, which is why the disorder is one of regression appearing only after a period of normal early development. See the Epigenetics module for more on methylation mechanisms.

Classic presentation: apparently normal development to 6 months, then stagnation and regression (6–18 months) with loss of hand skills and speech, emergence of hand stereotypies (wringing, washing), and gait abnormalities.

Four stages: I — early stagnation (6–18 mo, subtle slowing, head growth deceleration); II — rapid regression (1–4 yr, hand/speech loss, stereotypies, breathing irregularities); III — plateau (2–10 yr, some social improvement, seizures peak, scoliosis); IV — late motor deterioration (>10 yr, rigidity, loss of ambulation).

Additional features: seizures (60–80%), acquired microcephaly, breathing irregularities (hyperventilation/apnea), prolonged QTc (cardiac monitoring needed), severe scoliosis.

Atypical variants: CDKL5 disorder (early seizures before regression — now a distinct entity); FOXG1 (congenital variant with severe impairment from birth).

Therapy and the dosage problem: MECP2 is exquisitely dosage-sensitive, and this single fact dominates therapeutic strategy. Too little MeCP2 causes Rett; too much causes MECP2 duplication syndrome — a distinct disorder (severe ID, spasticity, recurrent infections, mostly in males). The protein must sit within a narrow band. That makes naive gene replacement dangerous: an AAV vector cannot easily titrate expression neuron-by-neuron, so cells that already had a working copy (the mosaic female's normal half) would be pushed into the duplication range. The clever workarounds under investigation reflect this constraint — self-regulating vectors that cap their own expression, and X-reactivation approaches that switch the silenced wild-type allele back on (since females already carry a normal copy on the inactive X). In the meantime, trofinetide (an IGF-1 analog, FDA-approved 2023) is the first approved Rett treatment. It sidesteps the dosage trap entirely by not touching MECP2: it acts downstream to address neuroinflammation, synaptic maturation, and dendritic deficits.

Angelman syndrome results from loss of function of the maternal UBE3A allele at 15q11-q13. UBE3A encodes an E3 ubiquitin ligase that tags target proteins for degradation; in neurons this housekeeping function is essential for normal synaptic protein turnover and plasticity, so its loss derails synaptic development.

Why parent-of-origin matters — imprinting. Most genes are expressed from both the maternal and paternal copy, giving a built-in spare. UBE3A is an exception: in neurons it is imprinted, with the paternal copy silenced. The silencing is not a mutation but an epigenetic act — a long non-coding antisense transcript (UBE3A-ATS), expressed only from the paternal chromosome, runs across the paternal UBE3A gene and shuts it off. So a neuron has one working UBE3A allele by design: the maternal one. Knock out the maternal copy and there is no spare to fall back on — the paternal copy is still epigenetically switched off — and neuronal UBE3A drops to near zero. This is also exactly why the SAME 15q11-q13 region produces a mirror-image disorder, Prader-Willi syndrome, when the paternally-expressed genes are lost instead. See Methylation & Imprinting for the full imprinting mechanism.

The imprinting logic also explains the leading experimental therapy: antisense oligonucleotides (ASOs) that knock down the paternal UBE3A-ATS to un-silence the dormant paternal UBE3A — a built-in, sequence-correct backup copy waiting to be reactivated.

Four molecular mechanisms (they differ in severity and recurrence risk):

• Maternal 15q11-q13 deletion (~70%) — most common and generally most severe.
• Paternal uniparental disomy (UPD) (~2-5%) — generally milder.
• Imprinting defect (~2-5%).
• UBE3A pathogenic variant (~10%).

Clinical features

• Severe ID with absent or minimal speech.
• A happy demeanor — frequent laughter and smiling, excitability, hand-flapping.
• Ataxic, jerky gait and tremulousness.
• Acquired microcephaly, tongue protrusion, and evolving facial features.
• Seizures (onset typically 1-3 years, multiple types) with characteristic high-amplitude EEG patterns — the EEG is often suggestive even before clinical seizures.
• Sleep disturbance (reduced sleep need) and a notable fascination with water.

Diagnosis — DNA methylation analysis of the 15q11-q13 region detects ~80% of cases (deletion, UPD, or imprinting defect); if methylation is normal but suspicion remains high, UBE3A sequencing identifies the ~10% caused by UBE3A variants.

Management is supportive — seizure control, sleep (melatonin), and PT/OT/speech-language therapy with AAC for communication. UBE3A-reactivation and antisense (ASO) approaches are in clinical trials.

Each disorder requires a distinct testing approach, and the unifying lesson is that the testing method must match the mutation type. The key pitfall is assuming WES/WGS detects everything — it does not, because the dominant lesions in these disorders are precisely the things short-read exome sequencing is blind to: large repeat expansions, methylation/imprinting defects, and copy-number changes.

Fragile X — the canonical exome blind spot: FMR1 CGG repeat analysis (Southern blot/triplet-repeat PCR) is the gold standard. Standard WES does NOT detect this. The reason is technical: short reads (~100–150 bp) cannot span or accurately size a repeat tract that may be hundreds to thousands of bases of near-identical CGG, and the GC-rich expanded region also amplifies poorly. So the disease-causing element is literally unsequenceable by routine methods and must be specifically ordered. ACMG recommends FMR1 testing as first-tier in any male with unexplained ID. (Modern WGS pipelines are beginning to flag some short-tandem-repeat disorders, but dedicated FMR1 analysis remains definitive.)

Angelman — methylation, not sequence: the most common mechanisms (deletion, UPD, imprinting defect) are invisible to sequencing because they are about which parent's copy is active, not about a letter change. This is why DNA methylation analysis comes first — it reads the epigenetic state directly and captures ~80% of cases regardless of which of those three mechanisms is responsible; only the ~10% UBE3A point variants then need sequencing.

TSC: TSC1/TSC2 sequencing detects variants in ~85% of clinical TSC; ~15% are mutation-negative by standard sequencing. These "no mutation identified" cases often hide a deep intronic variant, a structural/copy-number change (needing MLPA), or — importantly — low-level mosaicism, where the variant is present in only a fraction of cells and falls below standard detection thresholds (calling for deep or long-read sequencing). Genetic confirmation alone is sufficient for diagnosis.

Rett: MECP2 sequencing plus del/dup analysis detects >95% of classic Rett — the del/dup arm matters because a meaningful minority are large deletions a sequencing-only panel would miss. If negative with a Rett-like phenotype, test CDKL5 and FOXG1.

Counseling highlights

• TSC: AD, 2/3 de novo; recurrence ~1–2% for apparently de novo (germline mosaicism)
• Fragile X: X-linked with maternal anticipation; counsel premutation carriers about FXTAS/FXPOI
• Rett: >95% de novo; low but non-zero recurrence (~1% germline mosaicism)