Integrated Multi-Omics and Interactome Analysis of CDK8 Inhibition Reveals Erythroid Differentiation Programs and BET Synergy in AML Stem-like Cells
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background. Acute myeloid leukemia (AML) remains among the most therapeutically challenging hematologic malignancies, driven largely by the self-renewal capacity, quiescence, and therapy resistance of leukemic stem cells (LSCs). CDK8, a kinase component of the Mediator complex, has emerged as a regulator of oncogenic transcription. We demonstrated that the selective CDK8/CDK19 inhibitor RVU120 (Romaciclib; SEL120-34A) potently targets AML cells with CD34⁺/pSTAT5-high LSC-like characteristics. However, the full spectrum of epigenetic and transcriptional events triggered by CDK8 blockade, and combinatorial strategies that may exploit them, remains incompletely defined. Methods. Using the TEX LSC-like cell line treated with CDK8 inhibitors, RVU120 and CCT251921, we performed time-resolved (3h, 24h, 72h) RNA-Seq, whole-proteome and phosphoproteomics mass spectrometry (MS). Chromatin occupancy of CDK8, Polymerase II RNA (RNAP2), BRD3, MLL-4, NFRKB and histone marks (H3K27ac, H3K4me1, H3K4me3) was profiled with CUT&Tag. CDK8 protein–protein interactions (PPIs) were surveyed by co-immunoprecipitation MS (co-IP-MS) and validated across five AML models. Synergy between CDK8 inhibitors and Pelabresib (BET inhibitor) or CB6644 (RUVBL1/2 inhibitor) was assessed by high-content screening microscopy in three cell lines and three patient-derived xenograft (PDX) models. Results. Both inhibitors suppressed STAT5 Ser726/731 phosphorylation; over subsequent days, TEX cells lost their CD34⁺/CD38⁻ phenotype and acquired surface markers of erythromegakaryocytic differentiation. Transcriptomics revealed a concordant response to both CDK8 inhibitors (r = 0.82–0.95), including early cholesterol biosynthesis, followed by inflammatory programs, and erythroid commitment with GATA1 upregulation. Whole-proteome changes significantly correlated with the transcriptomic response (r = 0.65–0.78). Phosphoproteomics identified 33 significantly different phosphosites with overrepresentation of nucleic acid-binding proteins. In the TEX cell line, co-IP-MS identified 125 CDK8-interacting proteins and a conserved PPI network across five AML models, including Mediator subunits, NFRKB, and BRD3. CUT&Tag genomic profiling demonstrated that CDK8 inhibition triggers widespread enhancer activation with coordinated redistribution of RNAP2, BRD3, and NFRKB, while uncoupling proliferative from immune-regulatory chromatin programs. CDK8 combined with Pelabresib acted synergistic in MOLM-16 cells and two of three PDX models. Conclusions. These findings position CDK8 as a central node in LSC transcriptional control and establish a mechanistic rationale for combined CDK8-BET inhibition as a precision therapeutic strategy in molecularly defined AML subsets.