Review papers with raw data transparency

Core mechanistic model (visual + key evidence)

1. PHF2, PHF8 and KIAA1718 form the human KDM7 family — each contains an N-terminal PHD reader and a JmjC catalytic demethylase; this modular architecture links recognition and catalysis (KDM7 family architecture (review)).
2. PHDs bind H3K4me3 (anchoring at active promoters); JmjC removes nearby repressive marks (H3K9me2/1, H3K27me2/1, H4K20me1), converting chromatin to a more permissive state for RNAPII-driven transcription (PHD anchoring & JmjC demethylation (review)).
3. Functional coupling: PHD binding to H3K4me3 can stimulate demethylation of adjacent tails (cis) for some family members (PHF8) but steric constraints or rigid linkers mean different family members show distinct behaviour (KIAA1718 inhibited when bound in cis) (Cis vs trans demethylation (review)).

Why this matters biologically

The review links KDM7 activity to rRNA and mRNA transcriptional co-activation, neuronal differentiation and disease: PHF8 mutations associate with X-linked mental retardation and craniofacial defects; knockdowns in zebrafish and mouse ES cells produce neurodevelopmental phenotypes (Developmental & disease links (review)).

Major strengths

• Integrative: synthesizes structural, biochemical, genomic and genetic data into a cohesive mechanistic model (Integrative synthesis (review)).
• Mechanistic specificity: shows how domain geometry (linker flexibility/rigidity) predicts biochemical outcome — a falsifiable structural-to-function claim (Structure-function predictions (review)).

Key limitations, blind spots and potential biases

• Overreliance on domain structures and in vitro assays — full-length protein structures and native nucleosome-context assays are limited (Domain vs full-length limitation (review)).
• Subtle in vivo methylation changes and variable target overlap across studies — raises possibility of context-dependent roles or compensatory mechanisms (Subtle in vivo effects (review)).
• Species/model differences: evidence comes from human cell lines, mouse ES cells, zebrafish and C. elegans; cross-species generalization is not guaranteed (Cross-species caveat (review)).
• Potential for publication bias / positive-result bias in the literature synthesized; the review does not present raw datasets for reanalysis (no supplied underlying data) (Data availability limitation (review)).

What would disprove the central model?

1. Show that PHD binding to H3K4me3 is dispensable for promoter localization and transcriptional co-activation of PHF8/PHF2/KIAA1718 in native cells (genetic separation-of-function mutants).
2. Demonstrate that demethylase catalytic activity is unnecessary for co-activation in physiological contexts (e.g., catalytically dead full-length mutants rescuing loss-of-function phenotypes).
3. Prove consistent in vivo demethylation of the proposed substrates at endogenous loci across relevant neuronal cell types in patient-derived material.

Concise recommended next experiments

• Endogenous CRISPR knock-in of point mutants separating PHD binding from JmjC catalysis, with ChIP-seq and quantitative mass-spec histone PTM profiling in neuronal progenitors.
• Cryo-EM of full-length KDM7 proteins bound to nucleosomes to resolve linker geometry and validate cis vs trans demethylation models.