henhaobio
Sign in

bioRxivGenomicsPreprint

Human-specific CCG-repeat insertion in the RBFOX1 promoter implicated in autism and speech enhances EGR1-responsive activity

Chul LeeiD, Matthew H. DavenportiD, Erich David JarvisiD

PDFDOIData
In-depth full-text digest available · Jump to digest ↓

Abstract

Human-specific promoter innovations have been predicted to shape human-specific traits. Here, we investigated the RBFOX1 splice regulator gene, which is specifically upregulated in human and songbird speech/song brain areas. We found that the human RBFOX1 promoter associated with autism and language impairment harbors a human-specific CCG trinucleotide insertion creating a functional CCG-repeat motif. Comparative single-nucleus transcriptome analyses of primate brains indicate that the human promoter drives differential expression in cortical excitatory neurons. Reporter assays show that the human CCG-repeat motif, relative to the chimpanzee allele, increases promoter activity with the EGR1 transcription factor. Genome-wide, the CCG-repeat motif marks core promoters of 104 genes, related to postsynaptic function and autism. In PTCHD1, a clinical duplication in autism patients creates another CCG-repeat motif that similarly increases EGR1-responsive promoter activity. Our findings suggest that small CCG-repeat variants in regulatory regions can exert large effects on gene regulation and apparent associated traits.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (12 Oct 2026) · CC BY 4.0

A human-specific CCG insertion in the RBFOX1 promoter boosts EGR1-driven reporter activity in HEK293T cells, as does an autism-linked CCG-gaining duplication in PTCHD1.

The paper says comparative screens for human regulatory innovation favour alignable sequences with accelerated substitution, so they underrepresent short lineage-specific insertions; neither HAR nor HAQER catalogs annotate this autism-linked RBFOX1 promoter. It uses long-read primate and mammal assemblies, a promoter-anchored aligner (ReAlignPro), AlphaGenome variant-effect estimates and reporter assays to test a three-base repeat gain.

Key results

  1. ShownThe RBFOX1 promoter carries a three-base CCG insertion found only in humans, which completes a five-copy CCG motif. It is shared with archaic hominins and invariant across 90 sequenced modern human haplotypes.

    Absent (three-base gap) in all 13 non-human sequences (non-human primates plus a flying lemur outgroup) of a 14-way alignment. ReAlignPro recovered RBFOX1 promoters from 143 of 153 VGP non-human mammal assemblies, none with the insertion. The motif is invariant across 90 HPRC haplotypes. Archaic hominin reads carry the same sequence. · A disease-susceptible promoter of RBFOX with a…

  2. ShownIn HEK293T luciferase reporters with EGR1, the human RBFOX1 promoter allele is more active than the chimpanzee-like allele, which is more active than a motif-null allele. The human allele also responds more strongly to EGR1 itself.

    Human vs chimpanzee 1.82-fold (95% CI 1.26-2.62; Holm-adjusted P = 0.011); human vs motif-null 3.22-fold; EGR1-over-GFP fold activation 1.79 times higher for human (paired P = 0.0017). Five independent transfections. EGR1 lacking its DNA-binding domain abolished the hierarchy. · The human insertion creates a functional motif…; Figure legends

  3. ShownThe same CCG motif is found in the core promoters of 104 genes, enriched for neurodevelopmental and postsynaptic functions. 16 of them are SFARI autism-risk genes, all differentially expressed in autism cortex in at least one cell cluster.

    BLAST found 266 hg38 loci, 104 of them in core promoters (TSS ± 150 bp). Of the 266 loci, 62 contain a human-specific base, 20 of those in core promoters. In PsychENCODE snRNA-seq (35 clusters), RBFOX1 and AUTS2 were both downregulated in layer 6 excitatory cluster EXT_9_L6. · The RBFOX1 CCG-motif recurs at neural and…; Human-derived motif promoters and a convergent…

  4. SuggestedAlphaGenome variant-effect predictions point the same way as the RBFOX1 reporter result, but the predicted effect varies by promoter, which the authors take to mean the repeat is not a sufficient portable activating module. Only AUTS2, besides RBFOX1, met all three prioritisation criteria.

    The chimpanzee-like allele was predicted to lower RBFOX1 RNA in six adult brain categories. Among 17 human-derived loci, some genes were predicted up and others down. RBFOX1 and AUTS2 met the one-sided but not the two-sided FDR threshold. The authors state that co-regulation is not demonstrated. · The human insertion creates a functional motif…; Human-derived motif promoters and a convergent…; Discussion

  5. ShownAn autism-associated 27-bp duplication in the PTCHD1 core promoter adds an extra uninterrupted 5xCCG run about 30 bp upstream of the shared motif and raises EGR1-responsive reporter activity. No difference appeared under control conditions.

    ASD vs REF 1.97-fold (95% CI 1.49-2.61; Holm-adjusted P = 0.0050); EGR1-over-GFP fold activation 1.95 times higher for the ASD allele (paired P = 0.0042). ASD/REF ratios stayed 0.99-1.04 under EGR1ΔBD, WT1, WT1ΔBD, GFP or no added factor. Five independent transfections. · An ASD-associated PTCHD1 promoter duplication…; Discussion

Methods, data and reproducibility

Long-read primate and mammal assemblies were aligned (14-way alignment plus the ReAlignPro promoter-anchored aligner) to find human-specific repeat changes. JASPAR motif scans and calibrated AlphaGenome effect predictions then guided luciferase reporter assays in HEK293T cells with EGR1 or GFP co-expression.

  • 14-way primatomorpha alignment: 13 primates plus 1 flying lemur outgroup, from VGP and T2T assemblies · multiple species (primates, Dermoptera outgroup) · n = 14 genomes
  • VGP non-human mammalian assemblies queried with the human RBFOX1 promoter · multiple mammals · n = 153 assemblies; promoters recovered from 143
  • Archaic hominin genomes (Neanderthals, Denisova 3) and AADR/Arcseqhub datasets · human (archaic hominins) · n = 3 Neanderthals, 1 Denisovan (BAM reads)
  • HPRC pangenome assemblies · human · n = 90 haplotypes
  • DECIPHER clinical structural variants overlapping the RBFOX1 promoter interval · human · n = 17 variants in 36 individuals
  • PsychENCODE frontal-cortex snRNA-seq (ASD scGENE Portal) reanalysis, ASD vs control · human · n = 35 cell-type clusters
  • Luciferase reporter assays in HEK293T cells (RBFOX1 and PTCHD1 promoter alleles) · human (HEK293T cell line) · n = five independent transfections per assay
Methods & tools
ReAlignPro, AlphaGenome, Alpha-GENIE, JASPAR CORE 2022, UCSC Genome Browser, BLAST, DECIPHER, FANTOM5, ENCODE4 cCREs, SFARI Gene, Luciferase reporter assay (HEK293T)
Reproducibility gaps
Alpha-GENIE web tool (https://alpha-genie.net); no source-code repository given. No repository URL for ReAlignPro appears in the supplied text. STAR Methods and Tables S1-S4 were not included, so calibration details and locus lists could not be checked.
Limitations (5)
  • AlphaGenome does not model activity-dependent neuronal states or individual genetic backgrounds. HEK293T reporters lack neuronal chromatin, long-range regulation, endogenous factor levels and speech/song brain-area regulation. · stated by the authors
  • The clinical RBFOX1 and PTCHD1 variants do not isolate CCG-repeat effects from surrounding sequence. Length- and GC-matched constructs are needed to isolate repeat-specific activity. · stated by the authors
  • The variant-effect screen predicts effects of the whole promoter sequence, not of the repeat alone, and does not demonstrate co-regulation. Only a subset of motif-bearing promoters is predicted to behave like RBFOX1. · stated by the authors
  • Reporter results rest on five independent transfections per assay in one cell line, and only RBFOX1 and PTCHD1 of the 104 motif-bearing promoters were tested experimentally. · noted in this summary
  • The speech-area, activity-dependent RBFOX1 expression model is a hypothesis built on general knowledge. No experiment in neurons or speech-related brain tissue tests it. · noted in this summary

Read this if you study how short repeat changes in promoters shape human-specific or disease-linked gene regulation, or you need long-read-based comparative promoter alignment.

Citations
3

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISSuresh '26Barker '25Darbandi '22Forti '25Lai '25Patel '21
52% semantic matchNeuroscience
Varun Suresh, Emilie M. Wigdor, Yuhan Hao+16 more·bioRxiv·Apr 3, 2026· 2 citations