Scientists have long known that the virus behind AIDS can hide in plain sight. Antiretroviral therapy suppresses active replication of HIV-1, but the proviruses embedded in resting memory CD4+ T cells persist for decades, transcriptionally silent yet fully capable of rebounding once treatment stops. A study published in EMBO Molecular Medicine on 6 October 2026 now reports that a small molecule targeting the cellular transcriptional kinases CDK12 and CDK13, together with their cyclin partner Cyclin K (CCNK), can reverse this latency in laboratory models and in cells from people living with HIV-1, without any additional stimulus.
The compound at the center of the work, SR-4835, is not a conventional kinase inhibitor. Molecular modeling based on the X-ray structure of CDK12 predicts that SR-4835 binds the hinge kinase region in an ATP-competitive fashion. Reported IC50 values fall in the range of 60 nanomolar for CDK12/CCNK and 366 nanomolar for CDK13/CCNK, with dissociation constants of roughly 94 and 168 nanomolar respectively. Importantly, the drug shows no inhibitory activity toward CDK7, CDK8 or CDK9, and only moderate, tenfold weaker inhibition of CDK10, making it among the most selective CDK12/CDK13-targeting agents described to date.
To test whether this target matters for HIV-1, the team treated a panel of Jurkat-derived CD4 T cell latency models with SR-4835 for just two hours. These included 2D10 cells, which carry an integrated HIV-based cassette driving a destabilized GFP reporter, and the J-Lat 6.3 and J-Lat 9.2 clones, which harbor a replication-defective full-length provirus in which GFP replaces the nef gene. Across all three models, increasing concentrations of the drug produced a clear, dose-dependent rise in GFP expression, indicating reactivation of proviral transcription. In J-Lat 9.2 cells, where reactivation is notoriously difficult, the increase was statistically significant, and RT-qPCR confirmed elevated HIV-1 RNA levels. Cell viability assays showed relatively low toxicity, around 20 to 25 percent, under the short treatment conditions used.
A central concern in latency-reversal research is that apparent reactivation may be an artifact of cellular stress or apoptosis rather than genuine proviral awakening. The authors addressed this by flow cytometry, gating only on live Annexin V-negative, propidium iodide-negative cells. Reactivation was observed specifically within the viable population, and higher drug concentrations produced only minimal apoptosis. This control matters because stress-induced, non-specific activation of HIV-1 has confounded earlier candidate therapies, and the finding supports the interpretation that SR-4835 acts directly on the transcriptional machinery governing proviral expression.
The study also explored whether the compound could amplify the effect of existing latency-reversing agents, or LRAs. Cells were pulsed with SR-4835 and then exposed to Prostratin, SAHA, JQ1 or I-BET, each of which reactivates HIV-1 through a different pathway. In 2D10 cells, combination with I-BET, SAHA or JQ1 produced roughly fourfold increases in reactivation, while Prostratin showed no synergy in that model. In J-Lat 9.2 cells, combinations with JQ1 or I-BET yielded the strongest latency reversal, about fivefold. A Bliss independence analysis confirmed that the observed combination effects exceeded additivity.
Mechanistically, the work converges on P-TEFb, the CDK9/Cyclin T complex that releases paused RNA polymerase II into productive elongation. In cells, a large fraction of P-TEFb is sequestered in an inactive reservoir within the 7SK snRNP, where HEXIM1 binds Cyclin T and blocks CDK9 kinase activity. Immunoprecipitation experiments showed that after SR-4835 treatment, markedly less CDK9 co-precipitated with HEXIM1, while total CDK9 and HEXIM1 levels in input lysates were unchanged. RNA immunoprecipitation revealed a substantial decrease in 7SK snRNA associated with CDK9, confirming release of the complex from its inhibitory reservoir. The dependence on CDK9 was further validated pharmacologically and genetically: the CDK9 inhibitors Flavopiridol and AZD-4573, as well as CDK9-specific siRNAs, abolished the SR-4835-induced reactivation.
Chromatin-level analyses reinforced this picture. ChIP-qPCR in HIV-1-infected Jurkat cells showed approximately sixfold higher H3K4me3 at the viral promoter after treatment, alongside increases in CDK9 occupancy (2.2-fold), total RNA polymerase II (4.7-fold), and Ser2- and Ser5-phosphorylated polymerase CTD (twofold each), consistent with promoter activation and pause release. Genome-wide CUT&RUN profiling documented elevated CDK9 occupancy at transcription start sites and gene bodies, together with increased H3K4me3, H3K27ac and polymerase phosphorylation. Genetic depletion of CDK12, CDK13 or CCNK by siRNA or shRNA mimicked the drug’s effect, though more modestly, indicating that loss of the complex, rather than an off-target action, drives reactivation.
The most clinically suggestive experiments used primary human cells. In CD4+ T cells transduced with a dual-reporter HIV-1 and treated with antiretroviral drugs mimicking combination therapy, SR-4835 produced roughly 2.5-fold reactivation with low toxicity. In PBMCs from nine people living with HIV-1 whose viremia was suppressed on treatment, a two-hour exposure sufficed to increase full-length viral RNA in eight of nine donors, although the magnitude varied substantially between individuals, likely reflecting differences in proviral load, integration sites and epigenetic constraints. Reactivation was also observed, though subtly, in monocyte-derived macrophages and in U1 monocytic cells carrying a transcriptionally repressed provirus, hinting at cell-type-specific restrictions that P-TEFb release alone cannot overcome.
The drug’s effects were not confined to the viral genome. RNA sequencing identified 30,226 differentially expressed genes, with 1,448 upregulated and 2,059 downregulated among significant hits. Proto-oncogenes such as MYC, FOS and JUN rose, while DNA damage response genes including BRCA1, FANCD2 and FANCI fell. Notably, short transcripts under 2.5 kilobases were disproportionately enriched among upregulated genes, consistent with a model in which CDK12/CDK13 loss shifts P-TEFb toward short-gene transcription while long genes suffer elongation defects. The authors frame this as a previously unrecognized compensatory interplay between transcriptional kinases, with implications for both HIV cure strategies and cancer biology. They also caution that SR-4835 targets CCNK, a shared subunit, making it impossible to disentangle CDK12 from CDK13 contributions, and that therapeutic application will require substantial further investigation, including studies in larger patient cohorts to identify the parameters governing responsiveness.
Subject of Research: Pharmacological inhibition of CDK12/CDK13-Cyclin K reverses HIV-1 latency by releasing P-TEFb from the 7SK snRNP complex.
Article Title: Inhibition of CDK12/CDK13-CYCLIN K rewires P-TEFb signaling to reverse HIV-1 latency and modulate cellular gene expression
Article References: Murugavelu, P., Rani, J., Glebko, H., Ambiel, I., Tibroni, N., Tickotsky, N., Heyn, F., Schommers, P., Levin, L., Kuzmina, A., Fackler, O. T., & Taube, R. (2026). Inhibition of CDK12/CDK13-CYCLIN K rewires P-TEFb signaling to reverse HIV-1 latency and modulate cellular gene expression. EMBO Molecular Medicine. https://doi.org/10.1038/s44321-026-00536-z
Image Credits: AI Generated
DOI: 10.1038/s44321-026-00536-z
Keywords: HIV-1 latency, CDK12, CDK13, Cyclin K, SR-4835, P-TEFb, 7SK snRNP, CDK9, latency reversal, transcriptional elongation, molecular glue degrader, HIV cure
News Source: Ophelia Keating. (October 30, 2026). Drug That Degrades Cyclin K Flushes Latent HIV-1 From Infected Cells. Scienmag.
