Concise verdict

Salinas‑Luypaert et al. (Nature Genetics, 2025) provide strong, multi-modal experimental evidence that centromeric CpG methylation acts as a functional boundary that constrains CENP‑A positioning and limits CENP‑B binding; acute targeted demethylation causes CENP‑A expansion, altered centromere architecture, replication stress and centromere-proximal aneuploidy, while slow demethylation enables cellular adaptation (Salinas‑Luypaert et al., Nature Genetics 2025)

Visual paper analysis — DNA methylation influences human centromere positioning and function

Visual summary (figures first, text second)

Evidence, methods and key strengths

• Targeted perturbations: Centromere‑targeted DBD‑TET1‑AID (inducible/degradable) and DBD‑MQ1 Q147L constructs produced rapid demethylation or hypermethylation specifically at centromeric α‑sat arrays (validated by COBRA, pyrosequencing and nanopore methylation calls) (DBD‑TET1‑AID targeted demethylation).
• Single‑molecule mapping: DiMeLo‑seq (CENP‑A tethered adenine methyltransferase read by nanopore) and Fiber‑seq measured CENP‑A localization and chromatin accessibility on long single molecules, showing CENP‑A accumulation within and beyond canonical centromere dip regions (CDRs) after demethylation (DiMeLo‑seq/Fiber‑seq mapping).
• Biochemical validation: In vitro microscale thermophoresis showed CENP‑B has ≈8× weaker affinity for methylated CENP‑B boxes vs unmethylated, providing a mechanistic explanation for increased CENP‑B at hypomethylated centromeres (CENP‑B methylation-sensitive binding).
• Functional consequences: Rapid demethylation increased mitotic errors, micronuclei, copy‑number alterations concentrated near centromeres, and reduced colony formation and viability; CENP‑B knockout partially rescued defects, implicating excess CENP‑B binding in instability (Phenotypes and CENP‑B KO rescue).
• Adaptation vs acute toxicity: Slow demethylation (TET1 low expression or HELLS degradation modeling ICF) produced elevated CENP‑A/B but cells adapted and survived, recapitulating clinical observations that ICF patients can reach adulthood despite hypomethylation (Gradual demethylation adaptation).

Critical appraisal — strengths and limitations

Strengths

• Orthogonal, state‑of‑the‑art methods (targeted epigenetic editing, nanopore methylome, DiMeLo‑seq, Fiber‑seq, expansion microscopy, scGS) produce a convergent, mechanistic story with single‑molecule resolution (Multimodal evidence integration).
• Careful controls: catalytically dead constructs, single‑copy vs multicopy integrations, auxin‑degron reversibility, and rescue/KO experiments (CENP‑B KO) reduce risk of off‑target or construct‑binding artefacts (Robust experimental controls).

Limitations, blindspots and alternative explanations

• Cell models vs in vivo: All data are from cultured human cell lines (DLD‑1, RPE‑1, CHM13 clones, HCT116, ICF patient fibroblasts). Extrapolation to tissues/organisms is untested; tumor cell lines (DLD‑1) have preexisting genomic instability that could modulate sensitivity to centromere perturbation (Model systems used).
• Demethylation kinetics and off‑targets: Rapid, forced demethylation by tethered TET1 may produce transient DNA breaks or TDG/BER‑mediated repair activity that contributes to instability; authors acknowledge TET1‑mediated DNA damage could contribute (TET1 demethylation side-effects).
• Long‑term dynamics unclear: The study documents early CENP‑A expansion and some longer (weeks) adaptation, but whether centromere positioning will drift further over months or across development/gametogenesis is unknown; population genomics (pangenome centromere studies) show centromere variability, but long‑term in vivo dynamics need study (Long‑term centromere drift uncertainty).
• Heterogeneity across chromosomes/HORs: Authors report even demethylation kinetics across centromeres in their system, but biological centromeres vary in HOR sequence, CENP‑B box density and existing methylation patterns (T2T/pangenome studies). The generality across all human centromere haplotypes therefore requires population‑level testing (Centromere haplotype variation caveat).

What would falsify the paper's central claims?

1. Show that targeted centromeric demethylation (using an orthogonal demethylation method without TET1 oxidation intermediates) does not change CENP‑A/CENP‑B localization or centromere function in multiple cell types.
2. Demonstrate that CENP‑A expansion occurs without any change in DNA methylation (i.e., methylation is correlated but not causal).
3. Prove that increased CENP‑B binding after demethylation is an epiphenomenon unrelated to chromosome instability (for example, by showing equal CNA rates after demethylation even when CENP‑B binding is blocked genetically or chemically).

Practical takeaways & follow‑up experiments (concise)

1. Test demethylation using an enzymatic approach that avoids BER intermediates (e.g., targeted passive demethylation by inhibiting maintenance methyltransferase during S phase) to separate TET1‑damage effects from methylation loss effects.
2. Perform population‑scale analyses (T2T pangenome centromere haplotypes) to test whether natural hypomethylated CDRs and CENP‑B box density predict CENP‑A boundaries and aneuploidy propensity across individuals.
3. Use orthogonal in vivo models (e.g., organoids or mouse humanized centromere constructs) to test organismal phenotypes and trans‑generational effects.

Run deeper / Reproduce / Data

All sequencing data are available via SRA BioProject PRJNA1270043 and Zenodo deposits; analysis code is provided on Zenodo per paper (see Methods and Data Availability) — good for reproducibility and reanalysis (Data & code availability).

Primary citation used in this review:

Salinas‑Luypaert et al., Nat Genet 2025