Using CRISPR-engineered biallelic 3XFLAG tagging of endogenous LEDGF in HEK293T cells, the authors obtained high-specificity ChIP-seq maps revealing that LEDGF is concentrated at transcription start sites (TSSs) β 95% of LEDGF-3XFLAG peaks overlapped H3K4me3 and 54% overlapped RNA Pol II (MACS2, FDR<0.05). Critically, 96% of LEDGF-3XFLAG peaks were significantly reduced upon shRNA knockdown of MLL1 (KMT2A), while deleting the IBD domain reduced 81% of LEDGF peaks. Conversely, 69% of MLL1 peaks overlapping LEDGF were reduced in LEDGF knockout cells, indicating reciprocal stabilization . These data are consistent with prior structural work showing the LEDGF IBD binds an MLL1 intrinsically disordered motif in a menin-dependent complex .
The most provocative negative result: SETD2 knockout (eliminating all H3K36me3) did not significantly alter integration frequencies per gene, recurrent integration gene (RIG) overlap coefficient was unchanged (no significant change), and SPIN-state preferences were unaltered. Only a modest upstream shift in the metagene integration profile was observed, consistent with H3K36me3 contributing to distribution within gene bodies but not to gene selection . This contradicts prior biochemical and structural evidence that the PWWP domain specifically recognizes H3K36me3-nucleosomes . The dPWWP construct retained strong TSS enrichment, while MLL1 knockdown did not recapitulate LEDGF knockout integration phenotypes (RIG overlap coefficient 0.95 vs. control), suggesting residual LEDGF at TSSs in MLL1-depleted cells remains sufficient for integration positioning.
The proposed model is elegant: at TSSs, MLL1 occupies the IBD, blocking IN access; during elongation, LEDGF transfers to elongation factors (IWS1, SPT6, TFIIS), freeing the IBD for IN binding and enabling integration across transcribed sequences . Evidence for this includes: (1) HOX genes show strong LEDGF/MLL1 co-enrichment across gene bodies but low integration; (2) IBD-deleted LEDGF loses TSS binding; (3) MLL1 KD does not alter integration patterns. However, no direct biochemical competition assay in cells demonstrates that MLL1 outcompetes IN for the IBD at TSSs. Prior structural work showed HIV-1 IN can outcompete MLL from the IBD in vitro (apparent ICβ β 39.7 nM) , which seems to contradict the authors' in-cell inference unless chromatin context, stoichiometry, or co-factor recruitment reverses this affinity relationship.
Additional context from related work: INLAIs targeting the LEDGF-binding pocket on IN confirm that disrupting this interface is antivirally relevant , and R-loop formation has been independently proposed as an additional integration determinant .
The study is confined to HEK293T cells β an embryonic kidney-derived line not representative of primary CD4+ T cells, the physiologically relevant HIV-1 target. ChIP-seq of endogenous 3XFLAG-LEDGF required ectopic rescue constructs expressed from Sleeping Beauty vectors (with 260,991 peaks called, suggesting substantial noise: only 1.7% overlapped endogenous LEDGF peaks). MLL1 was knocked down by shRNA (not genetically deleted), leaving open residual-protein explanations. The MLL1-blocking-IN mechanism remains untested in cells. Cross-linking-based ChIP may artifactually trap transient interactions, potentially explaining the TSS-enrichment pattern that differs from native-chromatin methods reporting more even LEDGF distribution .
Know what changed, what holds up, and what remains uncertain. Every Friday. No ads.