CDK12 - a different kinase with a different cyclin, and in KMT2A-rearranged lines it is an anti-dependency
also called CDK12 · CRKRS · CRK7 · CRKR · KIAA0904 · cyclin-dependent kinase 12
CDK12 is a cousin of CDK9 that also phosphorylates RNA polymerase, but it uses a different partner protein and controls a different set of genes, mostly DNA-repair genes. In KMT2A-rearranged cells, removing it is if anything slightly less harmful than in other leukaemias, so the CDK9 story does not transfer to it.
The same thing, for a scientist
CDK12 (UniProt Q9NYV4, 1,490 aa, 17q12; previous symbol CRKRS) pairs with cyclin K, not with cyclin T, and the cyclin K/CDK12 complex regulates Pol II CTD Ser2 phosphorylation over a small subset of genes - predominantly long, exon-rich genes, and most prominently the DNA damage response genes BRCA1, ATR, FANCI and FANCD2 [E1]. Cyclin K also binds CDK13, and the founding study showed the DNA-damage-response requirement is specific to CDK12 rather than CDK13 [E1]. Structurally the Cdk12/cyclin K kinase module is solved and its substrate specificity characterised [E2]. This entry exists to answer a question the brief posed: does this graph's CDK9 story survive being about CDK12 instead? **No.** CDK12 is the most anti-selective gene in this beat's set - +0.115 index-minus-comparator, rank 17,136 of 17,728 - and the direction holds in both lineage halves (-0.289 myeloid, -0.308 lymphoid against -0.390 in other leukaemia lines) [E3]. CDK12 is well expressed in all seven panel lines (median 21.1 nTPM) [E4], so this is not silence.
Why CDK12 is not interchangeable with CDK9 in this disease
Three separate reasons, and any one of them is sufficient:
- Different cyclin. Cyclin K, not cyclin T 1. The AFF-family cyclin T1-binding sequence that connects the SEC to P-TEFb has no counterpart here, so there is no documented route from a KMT2A fusion to CDK12.
- Different gene set. CDK12 loss predominantly reduces long, exon-rich genes and the DNA damage response programme 1, not the pause-release programme the fusion is proposed to hijack.
- Wrong direction in the data. KMT2A-rearranged lines are less dependent on CDK12 than other leukaemia lines, consistently across the lineage split 3.
There is a distinct reason to keep CDK12 in view, however, and it is not an elongation reason: a leukaemia whose survival is already precarious on DNA repair - and this graph carries a Fanconi-pathway beat - could be sensitised by CDK12 loss through BRCA1/FANCI/FANCD2 suppression. That is a different hypothesis from anything in this beat and is not claimed here.
| uniprot | Q9NYV4 |
|---|---|
| length AA | 1490 |
| gene symbol | CDK12 |
| cytoband | 17q12 |
| uniprot verified | Q9NYV4 = CDK12_HUMAN, primary gene name CDK12 (synonyms CRK7, CRKRS, KIAA0904), 1,490 aa, Homo sapiens; UniProt REST fetched and read back 2026-09-20 |
| cyclin partner | cyclin K (CCNK) - not cyclin T |
| programme | Pol II CTD Ser2; long exon-rich genes; DNA damage response (BRCA1, ATR, FANCI, FANCD2) |
| breakpoint side | not applicable - no documented KMT2A contact and no documented route from the fusion |
essentiality
{
"chronos_mean": -0.4097,
"n_lines": 1208,
"chronos_mean_kmt2ar_myeloid": -0.289,
"chronos_mean_kmt2ar_lymphoid": -0.3079,
"chronos_mean_other_leukemia": -0.3896,
"scan_rank": 17136,
"scan_n_genes": 17728,
"difference": 0.1148,
"interpretation": "anti-selective direction, consistent across the lineage split"
}expression hpa celline nTPM — the numbers behind the chart
{
"SEM": 32.1,
"RS4;11": 8.1,
"MV4-11": 17.3,
"THP-1": 31.4,
"NOMO-1": 21.4,
"MOLM-13": 19.1,
"OCI-AML-2": 21.1,
"median": 21.1
}related entries
[ "gene-cdk13", "gene-cdk9" ]
depmap kmt2a selectivity
{
"chronos_mean_kmt2a_rearranged": -0.281,
"chronos_mean_other_leukaemia": -0.3958,
"difference": 0.1148,
"pan_essentiality_all_lines": -0.4097,
"selectivity_rank": 17136,
"rank_denominator": 17728,
"provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
"how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}- CDK12 inhibition suppresses BRCA1 and the Fanconi genes, which in a paediatric patient means an induced DNA-repair defect in every dividing cell, not only the blast.
- Any pan-CDK or dual CDK9/12/13 inhibitor carries this liability whether or not CDK12 is the intended target.
- Does CDK12 loss sensitise KMT2A-rearranged cells to DNA-damaging chemotherapy through BRCA1/FANCI/FANCD2 suppression? A fitness screen in untreated culture cannot see this.
- Is cyclin K limiting in these cells, and does CDK12 compete with CDK13 for it?
- Do the dual CDK9/CDK12/CDK13 inhibitors in development owe their leukaemia activity to CDK9, to CDK13, or to the combination? No published study separates them in a KMT2A-rearranged cell.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
A 70-kDa cyclin K binds CDK12 and CDK13 to form two distinct complexes; cyclin K/CDK12 regulates Pol II CTD Ser2 phosphorylation and the expression of a small subset of genes, predominantly long genes with many exons and most prominently the DNA damage response genes BRCA1, ATR, FANCI and FANCD2, and it is cyclin K/CDK12 rather than cyclin K/CDK13 that is necessary for their expression; genetic inactivation of cyclin K in mice causes early embryonic lethality.
The Cyclin K/Cdk12 complex maintains genomic stability via regulation of expression of DNA damage response genes. · Blazek D, Kohoutek J, Bartholomeeusen K, Johansen E, Hulinkova P, Luo Z, Cimermancic P, Ule J, Peterlin BM · Genes Dev 2011What this evidence cannot show
Human cell lines with expression microarrays, nuclear run-on and ChIP; the CDK12-versus-CDK13 distinction rests on depletion experiments in one system. No leukaemia cell.
Evidence that argues against this (1)
- Later work in other systems reports CDK13 cooperating with CDK12 over global Pol II processivity, so the clean division of labour in the founding paper is not the whole picture. Bosken CA et al. 2014
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The structure of human Cdk12/cyclin K has been solved and its substrate specificity characterised.
The structure and substrate specificity of human Cdk12/Cyclin K. · Bosken CA, Farnung L, Hintermair C, Merzel Schachter M, Vogel-Bachmayr K, Blazek D, Anand K, Fisher RP, Eick D, Geyer M · Nat Commun 2014What this evidence cannot show
A structure of the kinase module, not of a chromatin-engaged complex. This is a structural-biology result with no disease context.
Evidence that argues against this (1)
- This project's own rule applies: a PDB entry existing is not a structure suitable for design, and nothing here bears on whether CDK12 is a target in leukaemia. Blazek D et al. 2011
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
CDK12 has mean Chronos -0.410 across 1,208 DepMap lines and ranks 17,136th of 17,728 on index-minus-comparator difference (+0.115), with the KMT2A-rearranged myeloid half at -0.289 and the lymphoid half at -0.308 against -0.390 in other leukaemia lines - an anti-selective direction consistent across the lineage split.
DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API, re-queried for this beat on 2026-09-20 and split by lineage inside the KMT2A-rearranged arm; genome-scale ranks read from this repository's analyses/depmap-kmt2a-selective-scan-full/results.json (17,728 genes, 1,208 lines) · Broad Institute DepMap; Open Targets Platform; Oncofusion KG internal scan · Databases 2026What this evidence cannot show
Six myeloid and four lymphoid lines. +0.115 sits inside the +/-0.15 noise band this project uses at that n; the claim supported is 'no KMT2A-rearranged selectivity', not 'a protective effect'.
Evidence that argues against this (1)
- A CRISPR fitness screen in cell culture cannot detect a dependency that only appears under DNA-damaging therapy, which is precisely where CDK12 loss is expected to matter. Blazek D et al. 2011
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
CDK12 is detected in all seven KMT2A-rearranged panel lines the Human Protein Atlas carries, median 21.1 nTPM, so the anti-selective dependency is not an expression artefact.
Human Protein Atlas cell-line RNA nTPM, rna_celline.tsv (24,315,372 rows, 1,206 cell lines), downloaded and filtered for this beat on 2026-09-20 · Human Protein Atlas · Database 2026What this evidence cannot show
Bulk poly-A cell-line RNA, not protein.
Evidence that argues against this (1)
- Expression says nothing about cyclin K availability, which is the limiting partner for both CDK12 and CDK13. Blazek D et al. 2011
- related to →CDK13 - rank 171 of 17,728, equally deep in both lineage halves, and essentially unstudied in leukaemia
both use cyclin K; the DNA-damage-response requirement is CDK12-specific
a transcriptional CDK the CDK9 story does not transfer to: different cyclin, different gene set, anti-selective in KMT2A-rearranged lines
- regulates →Plain-English: what transcription elongation is, and why a traffic light explains this leukemia
Pol II CTD Ser2 on long, exon-rich genes
the paralogue that shares cyclin K and moves in the opposite direction