What LEDGF actually does for the fusion: it brings AF4, CDK9, BRD4, P-TEFb and Ser2P Pol II to the target gene, and H4K16ac with them
also called LEDGF recruitment-independent role · LEDGF licenses elongation
LEDGF is not the hook that holds the cancer protein onto DNA. It is what recruits the machinery that actually switches the gene on, so removing it silences the target genes without moving the cancer protein.
The same thing, for a scientist
Having shown that LEDGF is not required for fusion retention on chromatin, Zhu et al. asked what it is required for. In MLL-AF9-transformed HSPCs, LEDGF knockdown significantly reduced the chromatin occupancy of the AEP/SEC components **AF4 and CDK9** at Hoxa9 and Meis1, reduced **H4K16ac** at the same loci, and substantially reduced **BRD4, P-TEFb and Ser2-phosphorylated elongating Pol II**, while leaving H3K4me3 unchanged. The authors frame this explicitly as a *recruitment-independent role in modulating MLL-fusion protein functions*. This is the reading that reconciles the occupancy result with the transformation genetics: an IBM2 or F148A/F151A fusion that cannot engage LEDGF still reaches its genes and still cannot transform, because what it loses is the elongation licence rather than the address. It also relocates the therapeutic hypothesis: a LEDGF-directed agent would act like an upstream SEC/P-TEFb disruptor rather than like an eviction agent.
The chain, as measured
LEDGF knockdown -> H4K16ac down at Hoxa9/Meis1 -> BRD4 down -> P-TEFb down -> Ser2P Pol II down -> target gene expression down, with the fusion still sitting on the gene 1.
The H4K16ac step is the interesting one, because it is an acetylation mark and the paper's framing is that H4K16ac recruits the BRD4/P-TEFb module. That places a writer between LEDGF and the elongation machinery, which is a druggable link this graph has not modelled: gene-kat7 and the MOF/NSL axis sit exactly there, and the graph already records that five NUP98 fusions bind the NSL complex.
AF4 and CDK9 falling together is consistent with the SEC arriving as a unit rather than piecewise, which matches this graph's own SEC entries (complex-af4-mll-assembles-a-sixteen-component-elongation-machine, mechanism-cdk9-phosphorylates-aff4-sec-assembly).
Why this matters more than the occupancy result
The occupancy result is a negative: LEDGF is not the tether. This is the positive that replaces it, and it changes what a LEDGF drug would be expected to do.
- An eviction agent is judged by fusion ChIP signal. On this model that readout would be flat or would go the wrong way.
- A licence agent is judged by Ser2P Pol II, by nascent transcription at HOXA9/MEIS1, and by CDK9 occupancy. Those are the readouts a LEDGF pharmacodynamic assay should use, and they are not the readouts the field's rationale implies.
It also predicts a specific combination. If LEDGF loss removes BRD4 and P-TEFb from the target genes, then a LEDGF-directed agent and a BET or CDK9 inhibitor act on the same step, and the interaction should be epistatic rather than synergistic. That is a cheap, testable prediction and the opposite of what a naive two-target combination argument would assume.
The honest gap in the chain
None of the links has been shown to be direct. LEDGF knockdown reducing BRD4 occupancy is compatible with LEDGF recruiting an H4K16 acetyltransferase, with LEDGF binding BRD4 or the SEC, or with the whole cascade being a consequence of reduced transcription rather than a cause of it. The paper measures co-occurrence after perturbation, in one cell system, at two loci, by ChIP-qPCR.
There is one physical clue this graph should follow: LEDGF's IBD binds a conserved unstructured IBM consensus motif found in multiple cellular proteins, and the interactome defined that way is large 2. Whether any AEP/SEC subunit carries an IBM is a sequence question that could be answered in an afternoon and has not been asked in this graph.
| prediction | A LEDGF-directed agent should be epistatic with, not synergistic with, BET and CDK9 inhibition at MLL-fusion target genes. |
|---|
factors lost on ledgf knockdown
[ "AF4", "CDK9", "BRD4", "P-TEFb", "Ser2P Pol II", "H4K16ac" ]
factors unchanged
[ "H3K4me3" ]
factor increased
[ "MLL fusion protein occupancy" ]
loci
[ "Hoxa9", "Meis1" ]
- If LEDGF acts through the same elongation step as BET and CDK9 inhibitors, a LEDGF programme adds a third way to hit a node the field already has two drugs for, with none of their clinical data.
- P-TEFb and BRD4 are required at most active genes, so an agent working through this step inherits the toxicity profile of elongation inhibition rather than the clean profile the DepMap selectivity number implies.
- The causal chain is inferred from knockdown co-occurrence in one murine system; if it is wrong, the pharmacodynamic readouts recommended here are wrong too.
- Does any AEP/SEC subunit carry an IBD-binding motif? The IBM consensus is published and the SEC subunit sequences are in UniProt; nobody in this graph has run the match.
- Is the LEDGF-BRD4 link mediated by an H4K16 acetyltransferase, and if so which one? KAT7/MOF is the obvious candidate and this graph already holds gene-kat7.
- Is a LEDGF-directed agent epistatic with a BET or CDK9 inhibitor, as this model predicts? A three-drug matrix in SEM would answer it and would be the cheapest discriminating experiment on this axis.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
LEDGF knockdown in MLL-AF9-transformed HSPCs significantly reduced chromatin occupancy of the AEP/SEC components AF4 and CDK9 at Hoxa9 and Meis1, reduced H4K16ac at the same loci and substantially reduced BRD4, P-TEFb and Ser2-phosphorylated Pol II, while H3K4me3 was unchanged and fusion occupancy rose - which the authors interpret as a recruitment-independent role for LEDGF in MLL-fusion function.
ASH1L Links Histone H3 Lysine 36 Dimethylation to MLL Leukemia. · Zhu L, Li Q, et al. · Cancer Discov 2016What this evidence cannot show
Single system (murine MLL-AF9-transformed HSPCs), ChIP-qPCR at two loci, knockdown rather than knockout. No direct physical interaction between LEDGF and any SEC subunit is demonstrated, and the causal ordering within the H4K16ac-BRD4-P-TEFb chain is inferred from the published literature rather than measured here. Read via efetch db=pmc after fullTextXML returned HTTP 500.
Evidence that argues against this (1)
- Every factor in the chain could fall simply because transcription fell, making the measurements a consequence of the expression phenotype rather than its cause. The paper does not include a transcription-inhibition control that would separate these. Zhu L et al. 2016
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The LEDGF/p75 IBD binds a conserved unstructured consensus motif (the IBM) present in multiple unrelated cellular proteins, which defines the IBD interactome as a motif-based rather than fold-based recognition system.
Multiple cellular proteins interact with LEDGF/p75 through a conserved unstructured consensus motif. · Tesina P, Čermáková K, et al. · Nat Commun 2015What this evidence cannot show
The paper defines the motif and its structural basis; it does not enumerate whether AEP/SEC subunits carry it, which is the specific question raised here.
Evidence that argues against this (1)
- A short degenerate consensus motif over-predicts: a sequence match to an IBM is not evidence of binding, and the same paper shows the interaction is affinity-switched by phosphorylation, so motif presence alone would not establish an AEP/SEC contact. Sharma S et al. 2018
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
IBD-binding motifs are phospho-switches: phosphorylation of IBM serines raises IBD affinity, and eliminating the IBM phosphorylation sites on MLL1 disrupts the oncogenic potential of primary MLL1-rearranged leukaemic cells.
Affinity switching of the LEDGF/p75 IBD interactome is governed by kinase-dependent phosphorylation. · Sharma S, Čermáková K, et al. · Proc Natl Acad Sci U S A 2018What this evidence cannot show
Cited here to constrain motif-based inference, not for the KMT2A residues themselves, which the graph already holds in interface-psip1-ibd-kmt2a-phosphoswitch.
Evidence that argues against this (1)
- The transformation defect of the phospho-null mutant was not shown to act through LEDGF binding specifically, so the switch could operate through another reader of the same serines. Sharma S et al. 2018
supplies the positive role that replaces the tether model
P-TEFb and CDK9 occupancy fall on LEDGF knockdown
BRD4 occupancy falls on LEDGF knockdown
AF4 occupancy at Hoxa9/Meis1 falls on LEDGF knockdown
the locus at which the chain was measured
the recruitment-independent function of this protein
the measured chain the hypothesis generalises