Comprehensive Research Report

Claes–Jensen Type X‑Linked Intellectual Disability (MRXSCJ)

Prepared: 2026‑07‑31 · Target for: dismech KB entry kb/disorders/Claes-Jensen_Type_X-Linked_Intellectual_Disability.yaml

⚠️ Curation Preamble — Read Before Using This Report

1. Named Entity Confusion (NEC) preflight — PASSED. This disease sits in two high‑NEC‑risk classes (eponymic XLID syndrome; member of a numbered MRXS/MRXSCJ series). Identity was cross‑anchored on four independent sources before any content was gathered:

All four agree on gene = KDM5C. No mismatch. Other xrefs: DOID:0060809, UMLS:C1845243, MedGen:335139, GARD:0016744.

2. Snippet-verification caveat — MANDATORY. Abstracts in this report were retrieved via the NCBI E‑utilities efetch endpoint but were passed through a summarizing model, which imposes a ~125‑character quote ceiling. The quoted strings below are candidate snippets, not verified snippets. Before any of them is committed as a dismech evidence.snippet, run:

just fetch-reference PMID:XXXXXXXX
just validate-references kb/disorders/Claes-Jensen_Type_X-Linked_Intellectual_Disability.yaml

Likewise, every non‑HPO ontology ID suggested here (GO, CL, UBERON, CHEBI, NCIT) is a candidate and must pass just validate-terms. HPO terms in §3 are exceptions — they were pulled directly from the authoritative HPO/JAX annotation API for OMIM:300534 and carry real annotation frequencies.

3. Environment note. The MCP pubmed and ols-mcp servers and local runoak were not permission‑granted in this non‑interactive run; all data came from direct API fetches (NCBI E‑utilities, EBI OLS4 REST, HPO/JAX API, ClinGen, UniProt REST, MGI, PMC). Section 9 prevalence is consequently thinner than ideal — Orphanet's epidemiology table (API 401 / site bot‑walled) could not be read and should be filled in from the local ORPHA_85279 cache via just structured-rebuild-orphanet --id 85279.

1. Disease Information

Overview

Claes–Jensen type syndromic X‑linked intellectual disability (MRXSCJ) is a rare X‑linked chromatinopathy caused by loss‑of‑function variants in KDM5C (Xp11.22), encoding the histone H3 lysine‑4 di‑/tri‑methyl (H3K4me2/me3) demethylase JARID1C/SMCX. It is one of the more frequently mutated single genes in X‑linked intellectual disability (XLID).

The core clinical picture in hemizygous males is intellectual disability (usually moderate–severe) with short stature, microcephaly, hyperreflexia/spasticity, seizures, maxillary hypoplasia, and aggressive or disinhibited behaviour. Heterozygous females — historically dismissed as "carriers" — are now recognised to be affected far more often than the "X‑linked recessive" label implies, though more mildly.

Orphanet's definition (ORDO:85279):

"A rare multiple congenital anomalies/dysmorphic syndrome characterized by mild to severe intellectual deficit associated with variable clinical manifestations including spasticity, cryptorchidism, maxillary hypoplasia, alopecia areata, epilepsy, short stature, impaired speech, and behavioral problems."

Key Identifiers

Synonyms and Alternative Names

From OLS4 (MONDO) and MedGen: - Intellectual developmental disorder, X‑linked syndromic, Claes‑Jensen type - MRXSCJ, MRXSJ - Intellectual developmental disorder, X‑linked, syndromic 16 (MRXS16) - KDM5C‑related syndromic X‑linked intellectual disability (Orphanet preferred) - Claes‑Jensen syndrome (CJS) — used widely in recent literature - KDM5C‑NDD (KDM5C neurodevelopmental disorder) — the emerging, sex‑neutral, non‑eponymic term preferred by the 2026 RARE‑X cohort (PMID:41537560) - Historic/discouraged: "Mental retardation, X‑linked, syndromic, Claes‑Jensen type"; "…JARID1C‑related"; XLMR with short stature and hyperreflexia

Nomenclature recommendation for the KB entry: keep MONDO:0010355 as disease_term, but consider KDM5C-Related Neurodevelopmental Disorder in the description as the modern label, since the eponym encodes an outdated male‑only, recessive framing that the female data (§3, §9) refute.

Data Provenance Character

Information is aggregated disease‑level (OMIM, Orphanet, MONDO, HPO annotations, ClinGen) layered over individual‑patient case series. There is no EHR‑derived cohort. The two largest patient‑level aggregations are: - PMID:41537560 (Terry et al., Hum Mol Genet 2026) — the RARE‑X KDM5C Data Collection Program, a patient‑contributed registry (caregiver survey), 31 new individuals + literature meta‑analysis to 269 individuals / 130 families / 122 unique variants. This is the closest thing to a natural‑history dataset and is patient‑reported, not clinician‑abstracted. - PMID:39835750 (Ghasemi et al., Mol Genet Genomic Med 2025) — systematic literature review of 175 previously reported cases + 1 novel variant.

2. Etiology

Disease Causal Factors

Monogenic, genetic, fully penetrant in males. MRXSCJ is caused by germline loss‑of‑function variants in KDM5C. There is no infectious, environmental, or multifactorial component to disease causation. ClinGen's Intellectual Disability and Autism GCEP classified the KDM5C ↔ X‑linked syndromic intellectual disability relationship as Definitive (2018‑09‑19), inheritance X‑linked.

The causal chain is: KDM5C LoF → reduced/absent H3K4me2/me3 demethylase activity (± loss of non‑enzymatic scaffolding) → failure to restrain H3K4 trimethylation at CpG‑island promoters and enhancers → derepression of non‑neuronal, germline and cryptic transcriptional programs plus mistimed WNT signalling during corticogenesis → abnormal neuronal differentiation, dendritic arborisation and spine maturation → intellectual disability, seizures, behavioural phenotype.

Risk Factors

Genetic (causal, not "susceptibility"): - Hemizygosity for a pathogenic KDM5C variant in a 46,XY individual → essentially complete penetrance for ID (98% of males in meta‑analysis, PMID:41537560). - Heterozygosity in a 46,XX individual → incomplete, variable penetrance (56% with ID in meta‑analysis; 4/19 completely asymptomatic in Carmignac et al., PMID:32279304). - Being male is the single largest risk factor for severe expression. This is not merely dosage: KDM5C escapes X‑inactivation, so females normally express KDM5C from both X chromosomes and carry a higher baseline dose (Agulnik et al., PMID:7951230; Bonefas & Iwase, PMID:36831303 — "KDM5C escapes X-inactivation, thereby presenting at a higher level in females"). - Maternal carrier status (~10% of RARE‑X probands maternally inherited; historically the dominant mode in multiplex families). - De novo occurrence — >50% of the RARE‑X cohort; paternal inheritance rare (~1%). - Domain location of the variant is a severity/phenotype modifier: "Patients with mutated variants in the catalytic domain were more likely to experience seizures" (PMID:39835750).

Environmental: No established environmental risk factors. Advanced paternal age is a generic risk for de novo point mutation but has not been specifically demonstrated for KDM5C.

Protective Factors

• Favourably skewed X‑chromosome inactivation in heterozygous females is the classic candidate protective mechanism — but the evidence is inconsistent. Shen et al. (PMID:36536324) documented a symptomatic female with a de novo nonsense variant and explicitly "no significant skewed X-inactivation," and Carmignac et al. found asymptomatic and symptomatic females without a clean XCI correlate. Because KDM5C escapes XCI, XCI skewing is an incomplete explanatory model — this is a genuine open question worth curating as a KNOWLEDGE_GAP discussion.
• Residual enzymatic activity. Ghasemi et al. propose that "Missense mutations in catalytic domains may retain partial enzymatic activity, potentially producing milder phenotypes than nonsense mutations" (PMID:39835750). The R1115H variant is the extreme case — normal catalytic activity and stability, yet still pathogenic via a non‑enzymatic route (PMID:29670509).
• KMT2A dosage reduction — experimental, model‑organism only. Vallianatos et al. showed genetic epistasis: "Double mutation of Kmt2a and Kdm5c clearly reversed dendritic morphology, key behavioral traits" (PMID:32483278). This is a mouse result; it is a therapeutic hypothesis, not a human protective factor.
• No dietary, lifestyle, or nutritional protective factor is described.

Gene–Environment Interactions

Not established for MRXSCJ. Two leads worth flagging as hypotheses: 1. Illness as a decompensation trigger. Shaheen et al. (PMID:40346491) report developmental regression with loss of ambulation "following acute viral illness at 22 months." A single case — insufficient to assert a G×E interaction, but a candidate for a HUMAN_MODEL_MISMATCH/KNOWLEDGE_GAP note. 2. Immune/interferon axis. Liao et al. (PMID:41743791) found KDM5C‑mutant zebrafish have "disrupted antiviral and interferon-related signaling," raising the possibility that infectious exposure interacts with the KDM5C lesion. Zebrafish only; unreplicated in humans. 3. Upstream regulatory interaction with ARX is gene–gene, not gene–environment: "KDM5C, a gene known to be mutated in XLID-affected children and involved in chromatin remodeling, is directly regulated by ARX" (Poeta et al., PMID:23246292). ARX polyalanine expansions are hypomorphic for KDM5C transactivation — a shared regulatory path to ID+epilepsy.

3. Phenotypes

3.1 Authoritative HPO Annotations (HPO/JAX API, OMIM:300534)

These are the canonical, real HPO annotations with real observed fractions. n/m = affected/assessed. Use these fractions to justify any frequency: enum you assign (per docs/frequency-evidence-guidelines.md) — but note that most fractions come from small early cohorts and are superseded for common features by the 2025–2026 cohorts in §3.2.

Neurological / cognitive

Behavioural

Growth

Craniofacial

Ophthalmological

Genitourinary

Skeletal / limb

Other systems

Onset / inheritance annotations: HP:0011463 Childhood onset (3/3); HP:0001419 X‑linked recessive inheritance.

3.2 Modern Cohort Frequencies (supersede HPO fractions for common features)

Ghasemi et al. 2025 (PMID:39835750) — 175 literature cases, sex‑stratified:

Terry et al. 2026 RARE‑X (PMID:41537560) — 31 new (19 M / 12 F) + meta‑analysis to 269 individuals (153 M / 112 F / 4 unspecified):

Meta‑analysis: - Intellectual disability: 82% overall — 98% of males, 56% of females - ID severity: males predominantly severe (70%); females predominantly mild (56%) - Seizures: 35% overall — 47% of males, 18% of females

"47% of males having (or had) seizures, and 18% females having (or had) seizures"

RARE‑X new cohort (caregiver‑reported, therefore higher ascertainment of "soft" features): - Brain/nervous system involvement 100%; communication challenges 93% - Seizures 48% (M 42%, F 56%); first seizure age 1–10 y, median 2 y - Growth issues 78%; short stature 75% of growth respondents - Behavioural concerns 78%; short attention span 88%, impulsivity 88%, anxiety 71%, ASD 65% - Vision/eye problems 74%; abnormal eye movement (strabismus/nystagmus) 91% - Digestive issues 74%; constipation 83%

⚠️ Discrepancy to record in the KB, not smooth over. Female seizure frequency is 18–19% in the two literature meta‑analyses but 56% in the RARE‑X new female cohort. Male seizure frequency is 47% (RARE‑X meta) vs 64% (Ghasemi). These are ascertainment artefacts running in opposite directions: registry self‑enrolment enriches for symptomatic females; historical XLMR‑family literature enriches for severely affected males. Curate the range with both citations, or omit frequency: per the frequency‑evidence SOP.

3.3 Phenotype Characteristics

Age of onset: Congenital/neonatal in the sense that the lesion is germline; clinically recognised in infancy to early childhood. HPO annotates HP:0011463 childhood onset. Developmental delay is typically noticed in the first 1–3 years (walking ~12 months but first words at ~3 years in the Liao proband, PMID:41743791). Seizure onset median 2 years (range 1–10 y). Prenatal‑onset short stature is documented (PMID:40125771).

Severity: Highly variable. Males skew severe (70% severe ID); females skew mild (56% mild). The full range spans severe ID with progressive spastic paraplegia (the original Claes family) to a documented case with no intellectual disability at all — Murati et al. describe "An 8-year-old boy with prenatal-onset short stature, ophthalmological abnormalities" carrying a "KDM5C variant typically linked to Claes-Jensen syndrome" but without ID (PMID:40125771).

Progression: The cognitive deficit is static/non‑degenerative — a neurodevelopmental, not neurodegenerative, disorder. However, several elements are progressive: - Spastic paraplegia: "severe mental retardation, slowly progressive spastic paraplegia, facial hypotonia, and maxillary hypoplasia" (Claes et al. 2000, PMID:10982473) - Seizures: episodic, may emerge in mid‑childhood after an initially seizure‑free period - Behavioural difficulties: often intensify around adolescence - One report of regression post‑viral illness with loss of ambulation (PMID:40346491) — isolated, needs replication

Quality‑of‑life impact (per phenotype): | Domain | Impact | |---|---| | Intellectual disability + communication (93%) | Dominant driver of dependency; lifelong support needs; most affected males non‑ or minimally verbal in the severe range | | Seizures (35–48%) | Injury risk, medication burden, driving/independence restriction, caregiver vigilance | | Behaviour: aggression, impulsivity (88%), anxiety (71%) | Frequently the leading caregiver stressor; drives placement decisions and psychotropic prescribing | | Spasticity / gait | Mobility loss, contractures, orthopaedic surgery, wheelchair dependency in the paraplegic subset | | Short stature (50–78%) | Cosmetic/psychosocial; endocrine work‑up burden | | Vision (74%; abnormal eye movement 91%) | Amblyopia risk if untreated; a highly actionable, under‑recognised domain (PMID:40125771) | | GI/constipation (83%) | Chronic discomfort, feeding difficulty, contributes to behavioural escalation |

No disease‑specific QoL instrument (EQ‑5D/SF‑36/PROMIS) has been applied to this cohort. This is a genuine gap.

4. Genetic / Molecular Information

Causal Gene

KDM5C — lysine demethylase 5C. Aliases: JARID1C, SMCX, XE169, DXS1272E. - Locus: Xp11.22 · hgnc:11114 · NCBIGene:8242 · OMIM 314690 · Ensembl:ENSG00000126012 · UniProt:P41229 - Reference transcript: NM_004187.5 - Protein: 1,560 aa; EC 1.14.11.67

Protein architecture (UniProt P41229):

Note the "PLU‑1"/Tower region is used in the clinical literature to describe the long linker between C5HC2 and PHD2 (PMID:41743791, PMID:39835750). Cofactor: Fe²⁺ (one ion per subunit, catalytic). Localisation: nucleus. Tissue specificity: ubiquitous, highest in brain and skeletal muscle — directly consistent with the neurological + growth/muscle phenotype.

Enzymatic specificity: demethylates H3K4me3 → me2 → me1, not to unmethylated product, and does not act on H3K9/K27/K36/K79 or H4K20.

"SMCX (JARID1C), which encodes a JmjC-domain protein, reversed H3K4me3 to di- and mono- but not unmethylated products" — Iwase et al., Cell 2007 (PMID:17320160)

Pathogenic Variants

Variant spectrum (RARE‑X meta‑analysis, 122 unique variants across 130 families; PMID:41537560):

Concordant with Ghasemi et al. (PMID:39835750; 80 unique variants): missense 41 (51%), nonsense 27 (34%), del/dup 6 (8%), splice 6 (8%).

Missense variant distribution by domain (PMID:41537560): JmjC 36%, interdomain 33%, ARID 11%, C5HC2 9%, PLU‑1/Tower 5%, PHD2 3%, JmjN 3%. The catalytic JmjC domain is the missense hotspot — and catalytic‑domain variants are enriched for seizures (PMID:39835750).

Landmark variants:

Variant classification and population frequency: - ClinVar: 528 KDM5C records classified Pathogenic or Likely Pathogenic (E‑utilities count, July 2026). - gnomAD constraint (via ClinGen): pLI = 1, LOEUF = 0.17 — extreme intolerance to loss of function, among the most constrained genes in the genome. Pathogenic variants are absent or vanishingly rare in gnomAD; there is no meaningful population allele frequency. - ClinGen Dosage Sensitivity (2023‑07‑27): Haploinsufficiency score 3 — Sufficient Evidence; Triplosensitivity 0 — No Evidence. Haploinsufficiency (not gain‑of‑function or triplosensitivity) is the established mechanism. - Origin: Germline only for MRXSCJ. (Somatic KDM5C mutation is a well‑known driver in clear cell renal cell carcinoma — PMID:39955388, PMID:37293154 — but that is a distinct, unrelated disease context and must not be conflated with MRXSCJ.)

Functional consequence class — LOSS OF FUNCTION, achieved by at least four distinct routes: 1. Transcript loss / NMD — "expression studies revealed the almost complete absence of the mutated JARID1C transcript" (PMID:15586325) 2. Protein destabilisation — missense variants that "compromise stability and enzymatic activity" (PMID:25666439) 3. Direct catalytic impairment — "Several XLMR-patient point mutations reduced SMCX demethylase activity" (PMID:17320160) 4. Non‑enzymatic / scaffolding loss — R1115H: "The KDM5C-R1115H substitution does not have an impact on enzymatic activity," yet fails to suppress targets → "KDM5C may have non-enzymatic roles in gene regulation" (PMID:29670509). Reinforced in Drosophila: "KDM5 operates in conjunction with local chromatin contexts to employ demethylase-dependent and independent mechanisms" (PMID:41340160).

Route 4 is mechanistically important for the KB: it means "loss of demethylase activity" alone is an incomplete pathophysiology node. Model the enzymatic and non‑enzymatic arms separately.

Modifier Genes

• KMT2A — the opposing H3K4 methyltransferase (Wiedemann‑Steiner syndrome gene). Mouse double mutants show mutual suppression: "shared reduced dendritic spines and increased aggression" in single mutants, reversed in doubles (PMID:32483278). The strongest candidate genetic modifier and the leading therapeutic hypothesis.
• KDM1A (LSD1) — cooperative, not opposing: double forebrain‑specific KO produces "stronger ectopic expression of non-neuronal genes in hippocampal neurons and thousands of de novo H3K4me3-enriched regions" and "more severe behavioral impairments than the single ifKOs" (PMID:40864554). A candidate severity modifier.
• ARX — upstream regulator, not a modifier per se: ARX polyalanine expansions reduce KDM5C transactivation, and "Kdm5c mRNA diminution led to a severe decrease in the KDM5C content during in vitro neuronal differentiation" (PMID:23246292).
• KDM5D (Y‑linked paralogue) — a theoretical male‑specific partial buffer, but KDM5D is expressed in a restricted manner and does not rescue; in fact its Y‑linkage is part of why males lack the female two‑copy advantage. KDM5C and KDM5D have demonstrably non‑equivalent consequences (PMID:39955388, ccRCC context).

Epigenetic Information

This disease is an epigenetic disorder, and it also produces a secondary, measurable epigenetic signature.

Primary epigenetic lesion: failure to remove H3K4me3/me2 at CpG‑island promoters. Iwase et al. found "94% of Kdm5c-bound promoters contain a CpG island, representing significant enrichment" (P < 1×10⁻²⁶), with the effect concentrated on lowly expressed genes: "Low-expressed Kdm5c-target genes showed most noticeable increase in expression (~7% increase, P = 1.4 × 10⁻⁸) and H3K4me3 (~12% increase, P < 2.2 × 10⁻¹⁶)". Crucially, "global levels of H3K4me1, me2 or me3 are comparable in WT and Kdm5c-KO neurons" — the defect is locus‑specific fine‑tuning, not a bulk chromatin collapse. This is the single most important mechanistic nuance to encode.

Secondary DNA‑methylation episignature: Schenkel et al. defined a peripheral‑blood epi‑signature comprising 1,769 individual CpGs and 9 genomic regions in 7 male patients vs 56 controls, with 6 healthy female carriers showing intermediate changes (PMID:29456765). See §10 for diagnostic performance. Separately, Grafodatskaya/Chénier‑era work reported multilocus loss of DNA methylation in KDM5C‑mutant individuals (PMID:23356856), and a monozygotic‑twin methylation study exists (PMID:31419599).

Downstream chromatin consequences: patient fibroblasts show "local changes in chromatin conformation and gene expression" (PMID:25666439).

Chromosomal Abnormalities

MRXSCJ is predominantly a sequence‑level disorder. However: - Microdeletions involving KDM5C are within the ~3% "other" variant class (PMID:41537560) and one pathogenic hemizygous deletion was found by NGS gene‑dosage analysis in an XLID cohort (PMID:25649377). - ClinGen HI score 3 means an Xp11.22 CNV encompassing KDM5C is interpretable as causative in a male. - Larger contiguous Xp11.22 deletions may also involve neighbouring XLID genes (e.g. IQSEC2, SMC1A region) — expect a blended phenotype; check CMA breakpoints. - No recurrent translocation, inversion, or aneuploidy association.

5. Environmental Information

Not applicable as a causal category. MRXSCJ is a fully monogenic germline disorder.

• Environmental factors / toxins / radiation / occupational exposure: none implicated. No CTD‑curated chemical–disease association specific to MRXSCJ.
• Lifestyle factors: none causal. Relevant only as downstream management targets (nutrition for failure‑to‑thrive; activity for spasticity; sleep hygiene).
• Infectious agents: none causal. Two peripheral observations, both weak and non‑causal: (a) one case of regression after acute viral illness (PMID:40346491); (b) recurrent respiratory/URT infections are annotated phenotypes (HP:0002205, HP:0002788, both 3/3 in a small series) — likely secondary to hypotonia/aspiration rather than a primary immunodeficiency. The zebrafish interferon/TLR finding (PMID:41743791) is a transcriptomic dysregulation of antiviral pathways, not evidence of infectious causation.

6. Mechanism / Pathophysiology

6.1 The Causal Chain (upstream → downstream)

[MOLECULAR] KDM5C loss-of-function variant (LoF / destabilised / catalytically dead / scaffold-dead)
      │
      ├─► Reduced H3K4me3/me2 demethylase activity at CpG-island promoters + enhancers
      │        (locus-specific, NOT global — bulk H3K4me levels are normal)
      │
      └─► Loss of non-enzymatic scaffolding (REST/HDAC1-2/G9a complex; R1115H arm)
 │
[MOLECULAR/CELLULAR]  ▼
   Failure of transcriptional fine-tuning:
      • Derepression of REST target neuronal genes at NRSE elements (SCN2A, SYN1)
      • Spurious transcription: germline genes, non-neuronal genes, cryptic promoters
      • Failure to fine-tune activity-regulated enhancers
      • Mistimed canonical WNT signalling output
 │
[CELLULAR]       ▼
   • Premature/mistimed primary → intermediate progenitor transition; altered neurogenesis timing
   • Reduced ribosome biogenesis and translation (Drosophila arm)
   • Loss of neuronal identity maintenance (adult genome surveillance failure)
 │
[TISSUE]         ▼
   • Reduced dendritic arborisation (basolateral amygdala pyramidal neurons)
   • Reduced dendritic spine density (~45% of WT in BLA; 9% reduction in motor cortex)
   • Immature, thin (non-mushroom) spine morphology
   • Increased CA1 pyramidal neuron intrinsic excitability; altered ion channel expression
 │
[ORGANISM]       ▼
   Intellectual disability · seizures · aggression/anxiety/ASD · impaired social behaviour
   · memory deficits · short stature · spasticity

6.2 Molecular Pathways

(a) H3K4 methylation writer–eraser balance (the core axis). KDM5C is the eraser; KMT2A/MLL1 is the writer. The disorder is a stoichiometry disease of this pair. Suggested GO: GO:0032453 (histone H3K4 demethylase activity — verify label), GO:0034720/GO:0140939 (histone H3K4 demethylation — label changed in recent GO releases; must verify with OAK), GO:0006338 chromatin remodeling, GO:0005506 iron ion binding.

(b) REST/NRSF neuronal gene silencing.

"SMCX and REST co-occupy the neuron-restrictive silencing elements" · "loss of SMCX activity impairs REST-mediated neuronal gene regulation" (Tahiliani et al., Nature 2007, PMID:17468742)

SMCX assembles with HDAC1/HDAC2, the H3K9 methyltransferase G9a (EHMT2), and REST at NRSE elements in promoters of SCN2A and SYN1. RNAi depletion derepresses these targets with increased H3K4me3. Note the direct line from this to the seizure phenotype: SCN2A is itself a major epilepsy gene. Candidate GO: GO:0016575 histone deacetylation, GO:0045892 negative regulation of DNA-templated transcription.

(c) Canonical WNT signalling — the 2024 Nature mechanism, and the most therapeutically actionable. Karwacki‑Neisius et al. (PMID:38383780) established that "KDM5C is identified as a safeguard to ensure that neurodevelopment occurs at an appropriate timescale," acting by modulating WNT output during a defined developmental window to time the primary→intermediate progenitor transition. Critically, the deficit is pharmacologically reversible within that window: transient WNT inhibition "rescue[s] the transcriptomic and chromatin landscapes in patient-derived cells," and "WNT inhibition during this developmental period also rescues behavioural changes of Kdm5c knockout mice." The window matters — "only a transient alteration" is required, "WNT functioning in a transient nature to affect long-lasting cognitive function." Candidate GO: GO:0060070 canonical Wnt signaling pathway; GO:0021895 cerebral cortex neuron differentiation.

(d) Ribosome biogenesis / translation (invertebrate arm). Zamurrad et al. (PMID:29490272) found in kdm5^A512P flies "a striking downregulation of genes required for ribosomal assembly and function" with reduced translation, and "kdm5^A512P flies also showed impaired learning and/or memory." They argue "the primary defect of the KDM5A512P mutation is a loss of histone demethylase activity." Not yet demonstrated in mammals — flag as MODEL_ORGANISM and a candidate HUMAN_MODEL_MISMATCH.

(e) Interferon / Toll‑like receptor innate‑immune signalling (emerging, zebrafish only). Liao et al. (PMID:41743791): both novel variants produced overwhelmingly upregulated DEGs (355/363 and 320/326 up) enriched for "antiviral and interferon-related signaling," with TLR3, NFKB1, IFNB1, IRF7, SAT1a, SAT1b all up. The TLR inhibitor CU‑CPT 4a partially rescued morphology and restored spontaneous swimming. Candidate GO: GO:0034138 toll-like receptor 3 signaling pathway, GO:0060337 type I interferon-mediated signaling pathway. Zebrafish only; unreplicated; do not present as established human mechanism.

(f) ARX → KDM5C transcriptional axis (upstream). ARX directly binds a conserved noncoding element to activate KDM5C; polyalanine‑expanded ARX is hypomorphic. This links two XLID+epilepsy genes into one path (PMID:23246292).

6.3 Cellular Processes

• Neuronal differentiation and identity maintenance. Scandaglia et al.: "Kdm5c plays a critical role as a repressor responsible for the developmental silencing of germline genes" and, in the adult, "preventing the incorrect activation of non-neuronal and cryptic promoters in adult neurons" (PMID:28978483). This is a two‑phase role — developmental silencing plus lifelong genome surveillance — and both phases should be separate pathophysiology nodes.
• Enhancer fine‑tuning during activity‑dependent plasticity — "fine-tuning activity-regulated enhancers during neuronal maturation" (PMID:28978483).
• Dendritogenesis and spine maturation. Candidate GO: GO:0016358 dendrite development, GO:0060998 regulation of dendritic spine development.
• Neuronal excitability. Martín‑González et al. found "altered hippocampal expression of ion channels" and increased CA1 pyramidal excitability in KDM1A/KDM5C double KO (PMID:40864554) — a direct cellular substrate for seizures.
• Cell‑autonomous neuronal requirement. The RARE‑X Drosophila work is decisive: "Reducing the expression of its single Kdm5 gene in neurons, but not glia, led to spontaneous and stimulus-induced seizures" (PMID:41537560). Glia are dispensable for the seizure phenotype.
• Chromatin‑context dependence. "altered gene expression in both alleles correlates with preexisting chromatin signatures" (PMID:41340160) — KDM5C's effect is conditional on the local chromatin state, not uniform.

6.4 Protein Dysfunction

Four mechanistically distinct failure modes (detailed in §4): transcript loss/NMD, protein destabilisation, catalytic inactivation, and scaffold/recruitment failure with intact catalysis. A fifth, unusual mode: the R929X allele shows "The mRNA levels of the mutant gene were down-regulated, while the protein level" was up, with "Altering the subcellular localization of the protein" (PMID:39948613) — i.e. a truncated protein that escapes to the wrong compartment. No amyloid/aggregation mechanism. No dominant‑negative mechanism has been demonstrated; ClinGen's HI‑3 / TS‑0 assignment supports pure haploinsufficiency.

6.5 Metabolic Changes

No primary metabolic derangement; MRXSCJ is not an inborn error of metabolism. Biochemical work‑up in the original Belgian family was normal: "Biochemical investigations, neuroimaging and neuropathology were normal" (PMID:9377804). The only metabolic dimension is the enzyme's own cofactor dependence — Fe²⁺ and 2‑oxoglutarate (candidate CHEBI: CHEBI:29033 iron(2+), CHEBI:16810 2-oxoglutarate — verify). Endocrine abnormalities are more frequent in affected females ("Endocrine disorders were more frequent in females", PMID:32279304) but are unexplained and not mechanistically linked.

6.6 Immune System Involvement

No autoimmunity, no immunodeficiency established. Two threads: recurrent respiratory infections as an annotated phenotype (likely secondary), and the zebrafish interferon/TLR overactivation (§6.2e) — "suggesting aberrant immune activation" (PMID:41743791). Treat as an emerging hypothesis, MODEL_ORGANISM evidence only.

6.7 Tissue Damage Mechanisms

There is no tissue destruction. No oxidative stress, ischaemia, fibrosis, necrosis, or inflammation‑driven damage. Adult Kdm5c‑KO mice show "no gross abnormalities in the cytoarchitecture of the adult Kdm5c-KO cerebral cortex, hippocampus, or amygdala" (PMID:26804915). The pathology is structural‑microscopic and functional — dendritic/spine hypoplasia and transcriptional miswiring, not lesional. The one progressive element (spastic paraplegia) implies a corticospinal‑tract dysfunction of unclear substrate; human neuropathology was normal (PMID:9377804).

Curation implication: do not conform this entry to a degeneration/fibrosis/injury module. If a module fits at all, it is developmental/chromatin, and the relevant near‑neighbour is the chromatinopathy class rather than any existing dismech module. Consider proposing a chromatin_h3k4_writer_eraser_imbalance module — KDM5C/KMT2A/KDM1A/KMT2D form a genuinely conserved, recurrent writer–eraser axis across Claes‑Jensen, Wiedemann‑Steiner, Kabuki and KDM1A‑related NDD.

6.8 Epigenetic Changes

Covered in §4 (primary H3K4me3 dysregulation; secondary DNA‑methylation episignature; multilocus methylation loss).

6.9 Molecular Profiling

Transcriptomics: - Mouse (Iwase 2016, PMID:26804915): "larger number of up-regulated genes than down-regulated genes is consistent with the enzymatic activity of Kdm5c, which removes the active chromatin mark H3K4me2/3." Brain‑region‑specific effects — "Synaptic pathways such as 'Glutamate Neurotransmitter Release Cycle' and 'Nicotinic acetylcholine receptors' are down-regulated in the KO amygdala but not in the frontal cortex." A class of genes was "down-regulated in KO amygdala but unchanged in KO frontal cortex… highly relevant for neuronal differentiation, neuron-projection development, and synapses." - Mouse, sex‑stratified (Bonefas & Iwase 2023, PMID:36831303): "gene expression and behavioral abnormalities are readily detectable in Kdm5c-heterozygous female mice." - Zebrafish (PMID:41743791): 363 DEGs (355↑/8↓) for c.3019del; 326 DEGs (320↑/6↓) for c.782‑2A>T — strongly derepression‑biased, matching the eraser‑loss prediction. - Human patient‑derived cells (PMID:38383780): WNT‑inhibitor treatment rescued transcriptomic and chromatin landscapes. - Drosophila (PMID:29490272): ribosomal assembly gene downregulation.

Epigenomics: ChIP‑seq for KDM5C occupancy and H3K4me3 (PMID:26804915, PMID:28978483, PMID:40864554, PMID:41340160); genome‑wide DNA methylation array (PMID:29456765, PMID:23356856).

Proteomics / metabolomics / lipidomics: No dedicated studies. Genuine gap.

Single‑cell / spatial transcriptomics: No published single‑cell or spatial dataset specific to MRXSCJ. Given the strong brain‑region‑ and cell‑type‑specificity of the mouse phenotype (amygdala ≫ cortex), single‑cell profiling is the highest‑value missing experiment. Record as a KNOWLEDGE_GAP with a proposed_experiments entry.

Functional genomics screens: KDM5C appears in a genome‑wide screen for 2‑cell‑like state regulators (PMID:37488355) — developmental biology context, not MRXSCJ.

7. Anatomical Structures Affected

Organ Level

Primary: Brain (UBERON:0000955) — the overwhelmingly dominant target. Within it: - Cerebral cortex (UBERON:0000956) — frontal/motor cortex spine density reduction - Hippocampal formation (UBERON:0002421) — ectopic non‑neuronal gene expression, CA1 hyperexcitability - Amygdala, specifically the basolateral amygdala — the most severely affected structure in the mouse (dendritic length ↓, spine density ~45% of WT). Candidate UBERON for BLA: UBERON:0002873 — verify with OAK - Forebrain (UBERON:0001890) — the conditional‑KO target region - Corticospinal tract — implied by progressive spastic paraplegia; no direct imaging/pathology evidence

Secondary / other systems: | System | Involvement | UBERON candidate | |---|---|---| | Musculoskeletal / growth axis | Short stature, brachydactyly, cubitus valgus, pectus excavatum, contractures | UBERON:0002204 musculoskeletal system | | Craniofacial skeleton | Maxillary hypoplasia, micrognathia/prognathia, high palate | UBERON:0002397 maxilla | | Visual system | Strabismus, refractive error, abnormal eye movement (91%) | UBERON:0000970 eye | | Reproductive | Cryptorchidism, small testes, micropenis | UBERON:0000473 testis | | GI | Constipation (83%), cholelithiasis | UBERON:0001155 colon; UBERON:0002110 gallbladder | | Respiratory | Recurrent infections (likely secondary) | UBERON:0001004 respiratory system | | Integumentary | Patchy alopecia / alopecia areata | UBERON:0002097 skin of body | | Skeletal muscle | High KDM5C expression (UniProt); no described myopathy | UBERON:0001134 skeletal muscle tissue |

Tissue and Cell Level

Nervous tissue is the affected tissue type. Cell populations, with candidate CL terms (all require OAK verification):

Subcellular Level

• Nucleus (GO:0005634) — KDM5C's exclusive localisation (UniProt). Mislocalisation is itself pathogenic for the R929X allele (PMID:39948613).
• Chromatin / nucleosome (GO:0000785) — the substrate. Specifically CpG‑island promoters and activity‑regulated enhancers.
• Dendritic spine (GO:0043197) — the principal affected structure.
• Synapse (GO:0045202) — downstream. No mitochondrial, ER, lysosomal, or peroxisomal involvement.

Localization and Lateralization

Brain involvement is bilateral and symmetric. Microcephaly is generalised. Spastic paraplegia is bilateral and lower‑limb predominant (HP:0006895 lower limb hypertonia, HP:0002395 lower limb hyperreflexia) — a length‑dependent corticospinal pattern. Strabismus may be unilateral or alternating. Cryptorchidism may be unilateral or bilateral. Alopecia is patchy/focal (HP:0002232). One neuroimaging finding: "faint hyperintensities in posterior periventricular white matter suggesting dysmyelination" (PMID:40346491) — single case, bilateral posterior periventricular.

8. Temporal Development

Onset

• Biological onset: prenatal — WNT‑dependent progenitor mistiming occurs during corticogenesis, in utero (PMID:38383780). Prenatal‑onset short stature is documented (PMID:40125771).
• Clinical onset: infancy to early childhood. HPO annotates HP:0011463 childhood onset (3/3).
• Onset pattern: insidious / chronic. Not acute. Presents as failure to attain milestones rather than loss of them.
• Typical presentation sequence: hypotonia and feeding/growth concerns in infancy → gross‑motor delay (walking ~12 mo or later) → marked expressive language delay (first words often ~3 y) → behavioural difficulties in preschool/school years → seizures (median 2 y, range 1–10 y) → spasticity/gait deterioration in some, from mid‑childhood.

Progression

Stages — no formal staging system exists. A pragmatic natural‑history framing:

Progression rate: The cognitive deficit is static — it does not degenerate. The motor phenotype is slowly progressive in the spastic‑paraplegia subset: "slowly progressive spastic paraplegia" (PMID:10982473). Overall course pattern: chronic, lifelong, static‑with‑a‑slowly‑progressive‑motor‑component, punctuated by episodic seizures.

Duration: Lifelong. Not self‑limited.

Patterns

Remission: No spontaneous remission of ID. Seizures may be well controlled or remit with antiseizure medication in a subset — no quantitative data available. Behavioural difficulties may improve with intervention and maturation.

Critical periods — the most important temporal fact in this disease. The 2024 Nature work established a discrete, closable developmental window during which WNT modulation is corrective: use of "WNT signalling modulators at specific times reveal that only a transient alteration" is needed, with WNT inhibition in that window rescuing both molecular and behavioural phenotypes in Kdm5c‑KO mice. This defines a time‑limited therapeutic opportunity in embryonic/early‑postnatal corticogenesis — and, by implication, means that intervention after that window may not be corrective for the cognitive phenotype. Additional practical windows: early amblyopia detection (visual critical period), and early language/behavioural intervention.

9. Inheritance and Population

Epidemiology

Point prevalence: not established. No population‑based prevalence or incidence estimate exists for MRXSCJ. Orphanet's epidemiology class for ORPHA:85279 could not be retrieved in this run (API 401 / site bot protection) — retrieve it from the local structured cache (just structured-rebuild-orphanet --id 85279) before populating a Prevalence record. Given ORPHA:85279's designation as a rare multiple‑congenital‑anomaly syndrome, expect BELOW_1_IN_1000000 or NOT_YET_DOCUMENTED.

What is quantified is the gene's share of XLID — a case‑fraction, not a prevalence. Curate this with Genetic.case_fractions, not as a Prevalence record:

The widely repeated "2.8–3.3%" figure is best read as the upper bound from ascertained multiplex XLMR families, and 0.7% as the yield in unselected/regional cohorts. Report the range with both citations; do not pick one.

For context: ID affects "up to 2% of the population world-wide" (PMID:29670509), and XLID is genetically heterogeneous — "a genetically heterogeneous condition involving more than 100 genes" (PMID:32279304).

Genetic Etiology Parameters

Inheritance pattern: X‑linked. HPO annotates HP:0001419 X‑linked recessive inheritance and OMIM titles it X‑linked recessive — but this label is empirically wrong for females and should be curated with a caveat. With 56% of heterozygous females affected (PMID:41537560), the disorder behaves as X‑linked with markedly sex‑biased severity / incomplete female penetrance, not clean recessive.

KB recommendation: carry HP:0001419 (matching the authoritative annotation) but add a second Inheritance block or a notes field stating the empirical female penetrance. Do not silently assert HP:0001417 (X-linked dominant) — that overcorrects.

Modes of transmission (RARE‑X cohort, PMID:41537560): - De novo: >50% - Maternally inherited: ~10% - Paternally inherited: ~1% (rare; an affected/mosaic father transmitting to daughters) - Remainder untested/unknown

This is a major shift from the historical picture. Early gene discovery was done in multiplex XLMR families, which by construction were maternally transmitted; contemporary trio exome ascertainment reveals that de novo occurrence is now the majority mode.

Penetrance: - Males: essentially complete for ID — 96% (PMID:39835750) to 98% (PMID:41537560). The 8‑year‑old boy without ID (PMID:40125771) is the documented exception and shows penetrance is not literally 100%. - Females: incomplete — 56–79% with ID/learning disability; Carmignac et al. found 4/19 heterozygous females completely asymptomatic (PMID:32279304).

Expressivity: highly variable, in both sexes. Range spans no‑ID to severe ID with progressive spastic paraplegia, within and across families. Carmignac: "All affected individuals presented with learning disabilities or ID (mostly moderate)."

Genetic anticipation: Not applicable. No repeat expansion mechanism.

Germline mosaicism: Not specifically documented for KDM5C, but must be assumed possible for recurrence‑risk counselling given the high de novo rate (standard practice: quote ~1% empiric recurrence risk after an apparently de novo variant, higher if maternal mosaicism is detected).

Founder effects: None reported. Variants are private/family‑specific; the spectrum is dominated by unique variants (122 unique variants across 130 families — near‑complete allelic heterogeneity).

Consanguinity: Not a factor — X‑linked, not autosomal recessive.

Carrier frequency: Not estimable at population level. Given pLI=1 / LOEUF=0.17 and the near‑absence of LoF variants in gnomAD, carrier frequency is very low and dominated by de novo events.

Population Demographics

Affected populations: No ethnic predilection. Cases reported from Belgium (index family), Australia, USA, Germany, Netherlands, Italy, France, Brazil, China, Thailand, Estonia, Palestine, and elsewhere. First Latin American screen: PMID:24583395. First Palestinian case: PMID:40346491.

Geographic distribution: Worldwide, no clustering. Reporting bias favours countries with established exome‑sequencing diagnostics — apparent geography reflects diagnostic access, not biology.

Geographic distribution of specific variants: None; no recurrent founder allele.

Sex ratio: Historical literature: strongly male‑predominant (the RARE‑X literature meta‑analysis is 153 M : 112 F ≈ 1.4:1, already far less skewed than the classic "X‑linked recessive" expectation). The RARE‑X new cohort is 61% M : 39% F ≈ 1.6:1. Females are systematically under‑ascertained; the true molecular sex ratio in an unbiased sequencing cohort is likely closer to 1:1, with severity — not occurrence — being the sex‑biased variable.

Age distribution: RARE‑X participants ranged 2–20 years (mean 10.6 y males, 12.4 y females). Adults are markedly under‑represented in the literature — most published individuals are children, which biases natural‑history and prognosis data toward the paediatric course. Oldest well‑described individuals include a 27‑year‑old woman (PMID:36553533) and a 48‑year‑old (PMID:40544030).

10. Diagnostics

Clinical Tests

Laboratory tests: No diagnostic biochemical abnormality. Routine metabolic work‑up is normal — "Biochemical investigations, neuroimaging and neuropathology were normal" (PMID:9377804). Standard ID work‑up (CK, thyroid function, plasma amino acids, urine organic acids, acylcarnitines) serves to exclude alternatives. Endocrine evaluation (IGF‑1, growth‑hormone axis, thyroid) is warranted for short stature and for the female endocrine excess (PMID:32279304). No LOINC‑coded disease‑specific analyte.

Biomarkers: The DNA‑methylation episignature is the only validated biomarker (below). No protein or metabolite biomarker.

Imaging: Brain MRI is usually normal or non‑specific — its role is exclusionary. Reported findings are limited to "faint hyperintensities in posterior periventricular white matter suggesting dysmyelination" in a single case (PMID:40346491). Microcephaly is a clinical/OFC measurement, not an imaging diagnosis. Bone age and skeletal survey may be indicated for short stature/brachydactyly.

Functional tests: Formal neuropsychological/developmental assessment (Bayley, WISC, Vineland) is essential for ID diagnosis and severity grading. Ophthalmological assessment — explicitly under‑used and high‑yield given 91% abnormal eye movement and the amblyopia risk; PMID:40125771 argues "importance of ophthalmological assessments in X-linked syndromes."

Electrophysiology: EEG is indicated in all patients — 35–48% have seizures, tonic seizures documented (HP:0032792). No pathognomonic EEG signature described. EMG/NCS is not routinely indicated (the motor phenotype is upper‑motor‑neuron), though distal lower‑limb amyotrophy (HP:0008944) may prompt it. ECG not indicated.

Biopsy / pathology: Not diagnostically indicated. Human neuropathology was unremarkable (PMID:9377804). Skin biopsy for fibroblast culture is a research tool for functional variant assays (PMID:25666439).

Genetic Testing

Recommended approach. MRXSCJ is clinically non‑specific enough that gene‑first testing is standard:

1. Trio exome sequencing (WES) or genome sequencing (WGS) — first‑line for unexplained global developmental delay/ID. Trio design is essential given >50% de novo.
2. Chromosomal microarray (CMA) — either first‑line alongside, or reflex; detects the Xp11.22 microdeletion subset (~3% of the variant spectrum) that sequencing may miss.
3. XLID gene panel — a valid alternative in a family with an X‑linked pedigree. Tzschach et al.'s 107‑gene XLID panel achieved ">10 reads for approximately 96% of coding bases at mean coverage of 124 reads" and yielded 26% in familial vs 5% in sporadic cases (PMID:25649377). KDM5C is on all commercial XLID and ID/epilepsy panels; GTR lists dedicated single‑gene tests (e.g. GTR 581685).
4. Single‑gene KDM5C sequencing — only for targeted familial‑variant testing / cascade screening once a variant is known.
5. DNA methylation episignature (EpiSign) — see below; use as a reflex for VUS resolution and carrier confirmation.

Modalities not indicated: karyotype (too low resolution; normal in these patients), FISH (no recurrent rearrangement), mtDNA testing (not mitochondrial), repeat‑expansion testing (no repeat mechanism) — except as differential‑diagnosis exclusions (e.g. FMR1 CGG for fragile X).

A specific pitfall: synonymous and deep‑intronic variants can be pathogenic via splicing — c.633G>C (p.Arg211Arg) was predicted to create "an Exonic Splicing Enhancer sequence" and co‑segregated (PMID:24583395); c.782‑2A>T is a canonical splice‑acceptor change producing a PTC (PMID:41743791). RNA studies should be considered before dismissing a segregating synonymous or splice‑region variant.

Omics‑Based Diagnostics

Epigenomics — the standout. Schenkel et al.'s Claes‑Jensen episignature (PMID:29456765) is clinically deployed within the EpiSign framework: - Derived from 7 male patients vs 56 matched controls; 1,769 CpGs, 9 genomic regions - 6 healthy female carriers showed intermediate, distinguishable changes - "Highly specific computational model using the most significant methylation changes demonstrated 100% accuracy" in the training cohort - "The 100% specificity of this unique epi-signature was further confirmed on additional 500 unaffected controls" plus 600 ID/DD patients including other episignature cohorts - Clinical use: "can be used for molecular diagnosis and carrier identification and assist with interpretation of genetic variants" of unknown significance

Real‑world confirmation: Koparir et al. (PMID:41957673) applied EpiSign to 400 NDD individuals — "Seventeen percent of individuals (67/400) harbored variants in chromatinopathy-associated genes," "26 individuals (43%) exhibited disorder-specific episignatures," with KDM5C among the confirmed genes; "Integration of EpiSign analysis facilitated variant reclassification." Methylation profiling has been used specifically "to the reclassification of a variant" (PMID:35781022).

⚠️ Important caveat — do not overstate episignature performance. Husson et al.'s independent multicentre evaluation (PMID:37872275) found published episignatures perform very unequally: "While ATRX, DNMT3A, KMT2D, and NSD1 signatures displayed a 100% sensitivity, CREBBP-RSTS reached <40%," concluding "Episignatures do not perform equally well. Some signatures are ready for confident use" and "It is imperative to characterise the actual validity perimeter and interpretation of each episignature." The KDM5C signature's original 100% figures come from a 7‑patient training cohort and reflect specificity against large control sets more than sensitivity across the full allelic spectrum. Curate the 100% claim as training‑cohort performance with this limitation stated.

RNA sequencing: Research tool; potential clinical value for splice‑variant resolution. Proteomics, metabolomics, liquid biopsy: no diagnostic role.

Clinical Criteria

No formal, standardised diagnostic criteria exist (no DSM/ICD/society criteria for MRXSCJ). Diagnosis is molecular: a pathogenic/likely pathogenic KDM5C variant in a compatible phenotype. Historical clinical suspicion criteria remain useful for gene prioritisation — Abidi et al. concluded "male patients with mental retardation, short stature and hyperreflexia should be considered candidates for mutations in the JARID1C gene" (PMID:18697827), and reported in their nine males "mental retardation (100%), short stature (55%), hyperreflexia (78%), seizures (33%) and aggressive behaviour (44%)."

Differential diagnosis:

Because the phenotype is non‑specific, the differential is effectively "all XLID/ID" and is resolved by sequencing, not by clinical discrimination.

Screening

• Newborn screening: Not included in any programme; no treatable metabolic marker. Not appropriate under current criteria.
• Carrier screening (population): Not offered; not on expanded carrier screening panels.
• Cascade / family screening: yes, and important. Once a familial variant is known, test at‑risk female relatives — both for reproductive counselling and because carrier females are frequently affected and may benefit from their own diagnosis. The episignature independently identifies healthy carriers (PMID:29456765).
• Prenatal / preimplantation: Available for known familial variants (see §13).
• Risk stratification: No validated model. Male sex and a catalytic‑domain variant are the two crude severity predictors currently available.

11. Outcome / Prognosis

Evidence quality warning: there is no published survival study, no mortality analysis, and no longitudinal natural‑history cohort with adult outcomes for MRXSCJ. The RARE‑X cohort — the largest prospective dataset — has a maximum age of 20 years. Everything in this section about survival is inference, and should be curated as such (or omitted) rather than asserted.

Survival and Mortality

• Survival rate (5‑/10‑year/overall): no data.
• Life expectancy: Not established. There is no evidence of shortened lifespan attributable to the disorder itself — it is not a degenerative or organ‑failure condition, and individuals into their 40s are described (PMID:40544030). Any excess mortality would be expected to arise from the generic ID‑population risks: seizure‑related events (including SUDEP, given 35–48% epilepsy), aspiration in the hypotonic/dysphagic subset, and immobility complications in the spastic‑paraplegia subset. None of these has been quantified for MRXSCJ.
• Mortality rate / disease‑specific mortality: no data.

Morbidity and Function

Morbidity is substantial and lifelong, dominated by cognitive and behavioural burden: - 82% ID overall (98% males, 56% females); males predominantly severe (70%) - Communication challenges 93%; most severely affected males have minimal expressive language - Behavioural concerns 78% — impulsivity 88%, short attention span 88%, anxiety 71%, ASD 65% - Mobility: progressive spastic paraplegia in a subset → contractures, gait loss, wheelchair dependency - Most affected males require lifelong supported living; independent living is realistic only for mildly affected individuals, disproportionately female - ICF domains affected: learning and applying knowledge, communication, mobility (subset), self‑care, interpersonal interactions

Quality‑of‑life measures: No EQ‑5D, SF‑36, PROMIS, or disease‑specific QoL instrument has been administered. The RARE‑X caregiver survey is the closest proxy and captures symptom burden rather than validated QoL. This is a clear, actionable research gap.

Disease Course and Complications

Recovery potential: None for the established cognitive deficit under current therapy — this is a static, structural neurodevelopmental condition. The WNT data (PMID:38383780) are the first credible evidence that the phenotype is biologically reversible, but only within an early developmental window and only in mice/patient cells.

Prediction

Prognostic factors (all weak, none validated): 1. Sex — the strongest predictor. Male → severe (70% severe ID); female → mild (56% mild). 2. Variant domain — catalytic (JmjC/C5HC2) variants associate with seizures (PMID:39835750). 3. Variant class — the hypothesis that "Missense mutations in catalytic domains may retain partial enzymatic activity, potentially producing milder phenotypes than nonsense mutations" (PMID:39835750). Unproven, and complicated by R1115H (normal activity, still pathogenic) and by D87G (normal activity, disease‑associated). Do not curate as established. 4. Early seizure onset — plausibly predicts worse cognitive outcome, as in most DEEs; not demonstrated in MRXSCJ. 5. Residual expressive language at age 5 — a general ID prognostic anchor; not MRXSCJ‑specific.

Prognostic biomarkers: None. The episignature is diagnostic, not prognostic — no correlation between episignature strength and severity has been established.

12. Treatment

There is no disease‑modifying therapy, no approved drug, no gene therapy, and no interventional clinical trial for MRXSCJ. Management is entirely symptomatic, supportive, and multidisciplinary. Everything below labelled "experimental" is preclinical.

Pharmacotherapy (all symptomatic)

Pharmacogenomics: ClinGen reports 0 CPIC and 0 PharmGKB records for KDM5C. No gene‑specific PGx guidance. Standard CYP2D6/CYP2C19 considerations apply to the psychotropics used, unrelated to KDM5C.

Advanced Therapeutics

• Gene therapy / gene replacement: none. Conceptually challenging — a 1,560‑aa protein whose dose must be balanced (both loss and excess are deleterious, cf. the KMT2A epistasis) and whose critical window may be prenatal.
• Gene editing: none.
• RNA‑based therapies (ASO/siRNA/mRNA): none. Note that a KDM5C upregulation strategy (e.g. targeting a repressive element or NMD‑escape approach) is theoretically attractive for haploinsufficiency but entirely unexplored.
• Cell therapy, immunotherapy, targeted therapy: none.

Experimental / Preclinical Strategies

Three distinct, non‑overlapping preclinical leads — worth curating as mechanistic_hypotheses with status: EMERGING and explicit MODEL_ORGANISM evidence tagging:

1. Transient WNT inhibition during a developmental window (strongest lead). "WNT inhibition during this developmental period also rescues behavioural changes of Kdm5c knockout mice" and rescues "the transcriptomic and chromatin landscapes in patient-derived cells" (PMID:38383780, Nature 2024). Limitation: the window may close before postnatal diagnosis is possible — the central translational obstacle.
2. Rebalancing the H3K4 writer–eraser pair (KMT2A inhibition). "Double mutation of Kmt2a and Kdm5c clearly reversed dendritic morphology, key behavioral traits," supporting "balancing a single writer-eraser pair to ameliorate their associated disorders" (PMID:32483278). A genetic, not pharmacological, proof of concept; MLL1/menin inhibitors exist in oncology and are a conceivable repurposing route.
3. Toll‑like receptor pathway inhibition (CU‑CPT 4a). In zebrafish, treatment at ½ LC50 "partially restored morphological defects, including head area, body length and eye size" and "spontaneous swimming activity was restored"; the authors propose "Targeting the regulation of TRL related receptors (such as TLR3) may become a potential strategy" (PMID:41743791). Weakest of the three — zebrafish only, single study, unreplicated, and the interferon signature has no human correlate yet.

ClinicalTrials.gov: no interventional trial registered for MRXSCJ/KDM5C‑NDD. The RARE‑X KDM5C Data Collection Program (PMID:41537560) is an observational patient‑registry, not a trial, and is the appropriate referral for families seeking research participation.

Surgical and Interventional

• Orchidopexy for cryptorchidism (NCIT:C15329 / NCIT:C16186 candidates)
• Strabismus surgery and refractive correction (NCIT:C15329)
• Orthopaedic surgery for contractures, foot deformity (talipes equinovarus/calcaneovarus), scoliosis (NCIT:C16186 Orthopedic Surgical Procedure)
• Cholecystectomy if symptomatic cholelithiasis
• Gastrostomy in the failure‑to‑thrive/dysphagia subset
• Epilepsy surgery: no reported role

Supportive and Rehabilitative — the mainstay

Treatment Outcomes

Response rates: No quantitative data for any intervention. Adverse events: none disease‑specific; standard profiles for the symptomatic agents used. Aggression and hyperactivity in this population frequently drive polypharmacy — a recognised iatrogenic risk in ID generally.

Treatment Strategy

No published clinical practice guideline or care pathway exists for MRXSCJ. Practical algorithm, synthesised from the phenotype frequencies:

1. At diagnosis: baseline developmental/neuropsychological assessment; EEG; formal ophthalmological exam (91% abnormal eye movement — highest‑yield under‑performed test); growth chart with OFC; feeding/GI assessment; genital exam (males); genetic counselling; offer episignature if variant is a VUS.
2. Ongoing surveillance: annual growth/OFC; annual vision; developmental re‑assessment; seizure review; behavioural review; spasticity/gait exam; bowel review.
3. Escalate: ASM for seizures; behavioural intervention before psychotropics; PT/orthopaedics for progressive spasticity.
4. Family: cascade testing of at‑risk females; recurrence‑risk counselling; connect to RARE‑X registry and patient advocacy.

Combination therapies / personalised medicine: No genotype‑guided treatment exists. The nearest thing to precision stratification is domain‑based seizure risk (catalytic‑domain variants → heightened seizure surveillance) — reasonable clinical prudence, but not a validated rule.

13. Prevention

Primary prevention of the disease itself is not possible — it is a germline monogenic condition, and >50% of cases arise de novo, meaning most cases are unpredictable and unpreventable. "Prevention" here means recurrence prevention within families plus prevention of secondary complications.

Prevention Levels

Primary (preventing occurrence): Limited to reproductive options in families with a known variant (below). No vaccination, no risk‑factor modification, no environmental avoidance is relevant.

Secondary (early detection and intervention): This is where real benefit lies. - Early molecular diagnosis via trio WES/WGS in unexplained global developmental delay — ends the diagnostic odyssey, enables targeted surveillance, and (given the WNT critical‑window data) may eventually enable window‑timed intervention. - Cascade testing of at‑risk female relatives — identifies affected/at‑risk females who are currently under‑diagnosed. - Episignature testing to resolve VUS and confirm carriers (PMID:29456765). - Early ophthalmology to prevent amblyopia — the clearest preventable morbidity. - Early EEG and seizure recognition. - Early speech/AAC intervention.

Tertiary (preventing complications in affected individuals): - Seizure control to reduce injury and SUDEP risk - Spasticity management and stretching/orthotics to prevent contractures - Bowel regimen to prevent chronic constipation and impaction - Nutritional support to prevent FTT sequelae - Orchidopexy to reduce cryptorchidism‑related fertility/malignancy risk - Behavioural support to prevent crisis and placement breakdown - Fall/fracture prevention in the immobile subset

Immunization

No disease‑specific vaccine strategy. Routine childhood immunisation is indicated and should not be deferred — recurrent respiratory infections are an annotated feature (HP:0002205, HP:0002788), making influenza and pneumococcal vaccination particularly worthwhile. Note that the zebrafish interferon findings do not constitute any contraindication.

Screening and Early Detection

• Population screening / newborn screening: not indicated, not available (no treatable metabolic marker; does not meet Wilson‑Jungner criteria).
• Carrier screening: not on expanded carrier panels.
• Genetic screening in families with a known variant:
• Prenatal diagnosis — CVS or amniocentesis for the known familial variant
• Preimplantation genetic testing for monogenic disease (PGT‑M) — technically straightforward for a known KDM5C variant
• Both require prior identification of the familial variant; counsel explicitly that a female fetus carrying the variant has a substantial (~56%) chance of being affected, which materially changes historical "carrier daughters are unaffected" counselling
• Risk stratification: no validated model.

Behavioural Interventions

None reduce disease risk. Behavioural intervention is treatment (§12), not prevention.

Counselling

Genetic counselling is the central preventive intervention (NCIT:C15240). Key content, updated for current evidence:

1. De novo is now the majority mode (>50%) — a substantially lower recurrence risk than the classic X‑linked‑recessive family framing implies. Quote empiric ~1% for gonadal mosaicism after an apparently de novo variant.
2. Carrier mother: 50% transmission to each child; sons who inherit will be affected (~98%); daughters who inherit have ~56% chance of ID/learning disability and ~18–19% chance of seizures — they are not reliably unaffected.
3. Affected male: all daughters obligate carriers, no sons affected (rare paternal transmission documented, ~1%).
4. Offer maternal testing, cascade testing of maternal relatives, and prenatal/PGT‑M options.
5. Counsel on the wide variable expressivity including the documented no‑ID case — prognosis cannot be predicted precisely from genotype.
6. Connect to the RARE‑X KDM5C Data Collection Program and patient advocacy.

Carmignac et al. make the counselling point directly: consideration of "XLID genes in females, even in sporadic affected individuals" is required (PMID:32279304).

Public Health / Environmental Interventions / Prophylaxis

Not applicable — no environmental determinant, no infectious transmission, no prophylactic medication or procedure.

14. Other Species / Natural Disease

Taxonomy

KDM5C orthologues are present across vertebrates and, as a single ancestral KDM5 gene, across bilaterians:

The Drosophila situation is analytically valuable: a single autosomal Kdm5 gene collapses the KDM5A/B/C/D paralogue redundancy and removes sex‑chromosome dosage effects. The RARE‑X team exploited exactly this — fly seizures show "no sex differences in flies (unlike humans), reflecting autosomal Kdm5 location" (PMID:41537560), cleanly attributing the human sex bias to X‑linkage/XCI‑escape rather than to KDM5 biology itself.

Breed

Not applicable. No breed‑associated KDM5C disorder in any domestic species. No VBO identifier applies.

Natural Disease in Other Species

None documented. An OMIA search for KDM5C returned "No phene records found" — there is no naturally occurring KDM5C disease in companion animals, livestock, or wildlife. All non‑human disease models are engineered, not natural.

Veterinary relevance: none.

Comparative Biology

Evolutionary conservation is high and functionally validated. Three independent lines: - Sequence: the S451R variant residue "is conserved" across JARID1 family members "and in mouse and fruit fly" (PMID:16538222); pathogenic missense variants "alter evolutionarily conserved amino acids" (PMID:15586325); ARID‑domain disease variants "are located in a highly-conserved part of the ARID structure" (PMID:26580603). - Enzymatic: H3K4me3 demethylase activity is shared across the whole KDM5 family — "Other family members including SMCY, RBP2, and PLU-1 also demethylated H3K4me3" (PMID:17320160). - Phenotypic: the cognitive/behavioural consequence of KDM5 loss is conserved from fly (impaired learning/memory, seizures) through zebrafish (behavioural and morphological defects) to mouse (memory deficits, aggression, social deficits) to human (ID, seizures, aggression). This cross‑phylum concordance is unusually strong for an ID gene and materially raises confidence in the model systems.

Comparative pathology — differences worth noting: - Mouse Kdm5c‑KO shows no gross brain cytoarchitectural abnormality, mirroring the largely normal human MRI — good fidelity. - The mouse phenotype is regionally selective (basolateral amygdala ≫ motor cortex); no human equivalent has been sought. - Fly Kdm5 loss produces a ribosome/translation deficit not yet demonstrated in mammals — a candidate HUMAN_MODEL_MISMATCH. - Zebrafish mutants show cartilage/craniofacial defects and an interferon signature; the craniofacial arm loosely echoes human maxillary hypoplasia, but the interferon arm has no human correlate.

Transmission

Not applicable — non‑infectious, no zoonotic potential, no cross‑species susceptibility.

15. Model Organisms

15.1 Mouse — the principal mammalian model

Resource: MGI:99781. 38 total mutations and alleles (7 endonuclease‑mediated, 24 gene‑trapped, 7 targeted); 22 strains/lines available via IMSR. "22 phenotypes from 4 alleles in 4 genetic backgrounds." MGI explicitly curates a mouse model of "Syndromic X-linked intellectual disability Claes-Jensen type (OMIM:300534)."

Model types available: constitutive knockout, gene‑trap, targeted/conditional (floxed), endonuclease‑mediated (CRISPR) alleles, and forebrain‑specific inducible KO (ifKO) used in the Barco lab studies.

Phenotype recapitulation — Iwase et al. 2016 (PMID:26804915), the flagship:

Model limitations (Iwase 2016): - Seizures were not a reported phenotype, despite 35–48% seizure frequency in humans. The seizure phenotype had to be modelled in Drosophila instead (PMID:41537560) — a real gap in the mouse. - Anxiety runs the wrong direction (mouse anxiolytic‑like, human anxious). - Heterozygous females were not examined: "The KDM5C gene is X-linked in humans and mice, and affected human individuals are predominantly male, so we focused our analyses on male hemizygous animals." This omission was corrected seven years later by Bonefas & Iwase (PMID:36831303), who found "gene expression and behavioral abnormalities are readily detectable in Kdm5c-heterozygous female mice" and identified "sex-specific consequences of a reduced KDM5C dose in social behavior, gene expression." This is a case study in how a male‑only model design propagated the false 'unaffected carrier' assumption. - No craniofacial or skeletal phenotype characterised; no epilepsy, GI, or ophthalmological modelling.

Other key mouse studies: - Scandaglia et al. 2017 (PMID:28978483) — Kdm5c‑null + forebrain‑specific inducible KO. Established the developmental‑repressor and adult‑surveillance dual role. - Vallianatos et al. 2020 (PMID:32483278) — Kmt2a;Kdm5c double mutant; mutual suppression. The therapeutic proof of concept. - Martín‑González et al. 2025 (PMID:40864554) — Kdm1a;Kdm5c double inducible forebrain KO; synergistic loss of neuronal identity and increased CA1 excitability. - Karwacki‑Neisius et al. 2024 (PMID:38383780) — Kdm5c‑KO + human patient‑derived cells; WNT‑window rescue. The most translationally significant mouse result to date.

15.2 Zebrafish (Danio rerio)

Two generations of work: - Iwase et al. 2007 (PMID:17320160) — original: zebrafish and mammalian neuron studies revealed "roles in neuronal survival and dendritic development linked to demethylase activity." - Liao et al. 2026 (PMID:41743791) — a full patient‑variant model. Two clinical variants (c.3019del, c.782‑2A>T) expressed in zebrafish: - Morphology: "significantly reduced head area, body length, and eye size compared with control and WT groups" — directly models microcephaly and short stature; Alcian blue showed "impaired cartilage development" - Specificity control performed: "Co-injection with WT KDM5C mRNA rescued these phenotypic defects" — an important rigour marker - Behaviour: "All behavioral parameters were significantly altered in the c.3019del and c.782-2A>T groups" (reduced distance travelled and swimming speed), partially rescued by WT mRNA - Transcriptomics: 363 and 326 DEGs, overwhelmingly upregulated, enriched for antiviral/interferon responses; six validated genes (TLR3, NFKB1, IFNB1, IRF7, SAT1a, SAT1b) - Pharmacological rescue: CU‑CPT 4a at ½ LC50 — "Treatment partially restored morphological defects" and "spontaneous swimming activity was restored"

Strengths: rapid, scalable variant‑function assay with built‑in WT rescue control; models the growth/microcephaly axis the mouse handles less directly. Limitations: morpholino/mRNA‑injection transient models rather than stable germline mutants; the interferon signature is unreplicated and may be an injection artefact; no cognitive readout; single study.

15.3 Drosophila melanogaster

Single autosomal Kdm5/lid gene — the analytical advantage described in §14.

• Zamurrad et al. 2018 (PMID:29490272) — kdm5^A512P, a knock‑in of the fly residue equivalent to a human KDM5C disease missense variant. Found "a striking downregulation of genes required for ribosomal assembly and function" and reduced translation; "kdm5^A512P flies also showed impaired learning and/or memory." Concluded "the primary defect of the KDM5A512P mutation is a loss of histone demethylase activity."
• Terry et al. 2026 (PMID:41537560) — the seizure model, and the most clinically decisive fly result:
• Neuronal knockdown (elav>shKdm5): mechanical stress "Significantly more knockdown flies exhibited seizures (53%) than controls (19%)"; heat stress 58% vs 9%; spontaneous 6.74% vs 0%
• Cell‑type specificity: "Reducing the expression of its single Kdm5 gene in neurons, but not glia, led to spontaneous and stimulus-induced seizures"
• Dissociation from gross morphology: mushroom body reduction caused morphological defects but not seizures — separating the structural from the excitability phenotype
• Yheskel et al. 2025 (PMID:41340160) — compared demethylase‑dead (Kdm5^JmjC*) vs pathogenic ID variant (Kdm5^L854F). Found the two "produced divergent effects on H3K4me3 distribution" yet "similar transcriptional dysregulation" not correlated with recruitment, H3K4me3, or accessibility; instead "altered gene expression in both alleles correlates with preexisting chromatin signatures." Conclusion: "KDM5 operates in conjunction with local chromatin contexts to employ demethylase-dependent and independent mechanisms."
• Related: Hatch et al. on the KDM5–Prospero axis in mushroom body development; PMID:39677601 (bioRxiv preprint) on KDM5 insulator activity in the brain — preprint, not peer‑reviewed; do not cite as evidence.

Strengths: the only system that has reproduced the seizure phenotype; enables clean neuron‑vs‑glia and enzymatic‑vs‑non‑enzymatic dissection; no paralogue redundancy or sex‑chromosome confound. Limitations: no mammalian cortex; "learning/memory" assays are only loosely homologous to human cognition; the ribosome finding remains fly‑specific.

15.4 Cellular and In Vitro Models

• Patient‑derived primary fibroblasts — the workhorse for variant functional assay: protein stability, demethylase activity, and "local changes in chromatin conformation and gene expression" (PMID:25666439).
• Patient‑derived cells for WNT rescue — used in PMID:38383780; the substrate for demonstrating pharmacological reversibility.
• Primary cortical/hippocampal neuron culture with KDM5C overexpression — used to show R1115H's non‑enzymatic defect in post‑mitotic neurons (PMID:29670509).
• Biochemical/structural: recombinant ARID domain with urea‑induced unfolding and binding free‑energy calculations (PMID:26580603) — a COMPUTATIONAL + IN_VITRO hybrid.
• iPSC / cerebral organoids: no published MRXSCJ iPSC or organoid model. Given the WNT‑timed progenitor mechanism, human cortical organoids are the obvious missing system and the single highest‑value model gap. Curate as KNOWLEDGE_GAP with proposed_experiments.

15.5 Model Databases

MGI (MGI:99781), IMSR (22 strains), IMPC, KOMP/EuMMCR, ZFIN, FlyBase, WormBase, Alliance of Genome Resources, Cellosaurus (for patient fibroblast lines, where deposited).

Appendix A — Consolidated Reference List

Landmark and current sources, with PMIDs for just fetch-reference.

Disease definition and clinical delineation | PMID | Citation | Evidence source | |---|---|---| | 9377804 | Claes S et al. Clin Genet 1997 — original Belgian family (linkage then placed at Xq27‑28) | HUMAN_CLINICAL | | 10982473 | Claes S et al. Am J Med Genet 2000 — "Novel syndromic form of X-linked complicated spastic paraplegia" | HUMAN_CLINICAL | | 15586325 | Jensen LR et al. Am J Hum Genet 2005 — gene discovery, 7 mutations in 210 XLMR families | HUMAN_CLINICAL | | 16541399 | Tzschach A et al. Hum Mutat 2006 — 5 novel mutations | HUMAN_CLINICAL | | 16538222 | Santos C et al. Eur J Hum Genet 2006 — S451R | HUMAN_CLINICAL | | 18697827 | Abidi FE et al. J Med Genet 2008 — ID + short stature + hyperreflexia triad; frequencies | HUMAN_CLINICAL | | 24583395 | Gonçalves TF et al. Eur J Med Genet 2014 — Brazilian screen, 0.7% frequency | HUMAN_CLINICAL | | 25649377 | Tzschach A et al. Eur J Hum Genet 2015 — 107-gene XLID NGS panel | HUMAN_CLINICAL | | 32279304 | Carmignac V et al. Clin Genet 2020 — female phenotype, 19 new individuals | HUMAN_CLINICAL | | 39835750 | Ghasemi et al. Mol Genet Genomic Med 2025 — 175-case review, sex-stratified frequencies | HUMAN_CLINICAL | | 41537560 | Terry et al. Hum Mol Genet 2026 — RARE-X, 269 individuals + Drosophila seizures | HUMAN_CLINICAL + MODEL_ORGANISM (split the evidence items) |

Case reports expanding the spectrum | PMID | Citation | |---|---| | 36536324 | Shen R et al. BMC Neurol 2022 — female, de novo p.S1178X, no skewed XCI | | 36553533 | Lintas C et al. Genes 2022 — 27-y-old female, de novo p.Glu1283 | | 39948613 | Meng Y et al. Ital J Pediatr 2025 — p.R929X, NMD + mislocalisation | | 40346491 | Shaheen MM et al. BMC Pediatr 2025 — first Palestinian case, post-viral regression | | 40125771 | Murati FA et al. J Pediatr Ophthalmol Strabismus 2025 — Claes-Jensen without ID* |

Mechanism | PMID | Citation | Evidence source | |---|---|---| | 7951230 | Agulnik AI et al. Hum Mol Genet 1994 — SMCX escapes X-inactivation | IN_VITRO | | 17320160 | Iwase S et al. Cell 2007 — KDM5 family are H3K4 demethylases | IN_VITRO | | 17468742 | Tahiliani M et al. Nature 2007 — SMCX–REST/HDAC/G9a, SCN2A/SYN1 | IN_VITRO | | 23246292 | Poeta L et al. Am J Hum Genet 2013 — ARX → KDM5C regulatory axis | MODEL_ORGANISM + IN_VITRO | | 23356856 | BMC Med Genomics 2013 — multilocus loss of DNA methylation | HUMAN_CLINICAL | | 25666439 | Brookes E et al. Hum Mol Genet 2015 — protein stability + activity | IN_VITRO | | 26580603 | Peng Y et al. Int J Mol Sci 2015 — ARID domain variants | COMPUTATIONAL + IN_VITRO | | 26804915 | Iwase S et al. Cell Rep 2016 — Kdm5c-KO mouse | MODEL_ORGANISM | | 28978483 | Scandaglia M et al. Cell Rep 2017 — spurious transcription, enhancer fine-tuning | MODEL_ORGANISM | | 29670509 | Vallianatos CN et al. Front Mol Neurosci 2018 — R1115H, non-enzymatic role | IN_VITRO | | 29490272 | Zamurrad S et al. Cell Rep 2018 — Drosophila kdm5^A512P, ribosome/translation | MODEL_ORGANISM | | 32483278 | Vallianatos CN et al. Commun Biol 2020 — KMT2A/KDM5C mutual suppression | MODEL_ORGANISM | | 34536985 | Hatch HAM & Secombe J FEBS J 2022 — review | OTHER | | 36831303 | Bonefas & Iwase Cells 2023 — sexually dimorphic; heterozygous females affected | MODEL_ORGANISM | | 38383780 | Karwacki-Neisius V et al. Nature 2024 — WNT window, rescue | MODEL_ORGANISM + IN_VITRO | | 40864554 | Martín-González AM et al. Cell Rep 2025 — KDM1A/KDM5C cooperation | MODEL_ORGANISM | | 41340160 | Yheskel M et al. Epigenetics Chromatin 2025 — chromatin-context dependence | MODEL_ORGANISM | | 41743791 | Liao et al. Front Mol Neurosci 2026 — zebrafish, TLR/interferon, CU-CPT 4a | MODEL_ORGANISM |

Diagnostics / epigenetics | PMID | Citation | |---|---| | 29456765 | Schenkel LC et al. Clin Epigenetics 2018 — Claes-Jensen episignature | | 31419599 | Eur J Med Genet 2020 — monozygotic twin methylation fingerprint | | 35781022 | Eur J Med Genet 2022 — methylation profiling for variant reclassification | | 37872275 | Husson T et al. Eur J Hum Genet 2024 — independent episignature evaluation (caveat source) | | 41957673 | Koparir A et al. Clin Epigenetics 2026 — chromatinopathies, 400 individuals, EpiSign |

Non-literature resources consulted: EBI OLS4 (MONDO, ORDO); HPO/JAX annotation API (OMIM:300534); NCBI MedGen 335139; NCBI ClinVar (528 P/LP records); ClinGen (Gene-Disease Validity: Definitive, ID & Autism GCEP 2018-09-19; Dosage HI=3/TS=0, 2023-07-27; pLI=1, LOEUF=0.17); UniProt P41229; MGI:99781; OMIA (no entries).

Appendix B — Curation Notes for the dismech Entry

Suggested pathophysiology node chain (each biological_scale tagged; keep nodes atomic — see the single-value discipline in CLAUDE.md):

Module conformance: No existing dismech module is a good fit. epilepsy_excitation_inhibition_imbalance is a partial fit at the seizure node only (epilepsy_excitation_inhibition_imbalance#Excitation-Inhibition Imbalance) — the KDM1A/KDM5C ion-channel and CA1-excitability data (PMID:40864554) and the neuron-specific fly seizure result (PMID:41537560) support conformance there. Do not force-fit any degeneration, fibrosis, or storage module. Consider proposing a new chromatin_h3k4_writer_eraser_imbalance module (see §6.7).

Recommended discussions entries: - KNOWLEDGE_GAP — no single-cell or spatial transcriptomic data despite strong region-specificity (amygdala ≫ cortex) in mouse; proposed_experiments: snRNA-seq of human post-mortem or iPSC-derived cortical/amygdalar tissue. - KNOWLEDGE_GAP — no iPSC/cerebral organoid model; the WNT critical-window mechanism cannot be tested in human tissue without one. - KNOWLEDGE_GAP — no validated QoL instrument, no survival/mortality data, no adult natural-history cohort (RARE-X max age 20 y). - KNOWLEDGE_GAP — XCI skewing does not reliably explain the female phenotype spectrum, and KDM5C escapes XCI; the determinant of female severity is unknown. - HUMAN_MODEL_MISMATCH — mouse Kdm5c-KO shows reduced anxiety-like behaviour whereas 71% of humans have anxiety; and the mouse does not seize despite 35–48% human epilepsy (seizures required a Drosophila model). Both are direction/presence mismatches, not absent evidence. - HUMAN_MODEL_MISMATCH — the Drosophila ribosome/translation deficit (PMID:29490272) and the zebrafish interferon/TLR signature (PMID:41743791) have no demonstrated human correlate.

Do not curate as established: the TLR/interferon therapeutic axis (zebrafish, n=1 study); the "missense = milder than nonsense" genotype–phenotype rule (contradicted by R1115H and D87G); episignature 100% sensitivity (training-cohort figure, n=7, with PMID:37872275 as the counterweight); post-viral regression as a disease feature (single case).