Transcription-inhibitor payloads

In one sentence

Transcription-inhibitor payloads are ADC cargoes that block RNA polymerase II, the enzyme every cell uses to read its genes, so they can kill a cell whether or not it is dividing.

The intuition

A cell's DNA is a library. Copying DNA is making a second library, which happens only when a cell divides. Transcription is reading a book aloud so the cell can build the protein described in it, which happens every minute of a cell's life. Most antibody-drug conjugate (ADC) payloads sabotage the copying: microtubule inhibitors stop the cell pulling its chromosomes apart, and topoisomerase I (TOP1) inhibitors turn an unwinding enzyme into a DNA-breaking one when the copying machinery runs into it. Both work best on cells that are busy dividing — the reason chemotherapy hits dividing cells hardest.

A transcription poison shuts the reading room. A cell that is resting still needs to read its genes to replace proteins, so it is not spared.

Where the analogy breaks: healthy resting cells also read their genes, so there is no built-in sparing of normal tissue. Everything depends on the antibody delivering the payload only where it should.

How it works

The enzyme. RNA polymerase II transcribes all protein-coding genes. Its largest, catalytic subunit is encoded by the gene POLR2A. Alpha-amanitin — the principal toxin of the death cap mushroom — binds this complex and inhibits it specifically; a co-crystal structure of amanitin bound to RNA polymerase II has been solved [1]. Without new messenger RNA, short-lived proteins are not replaced and the cell dies, by a route that does not require the cell to be in the cell cycle.

Killing cells that are not dividing. The RNA polymerase II complex was pursued as a target precisely because it is needed in both proliferating and dormant cancer cells. An amanitin conjugate directed at a myeloma surface antigen killed both proliferating and resting myeloma cells in the laboratory while sparing antigen-negative cells, and single low doses produced durable complete remissions in mouse models [2]. This matters wherever slow-cycling or dormant cells are suspected of surviving treatment — stated as biology, not as a demonstrated clinical benefit.

The collateral-vulnerability bet. POLR2A sits on chromosome 17p, next to TP53, the most frequently inactivated tumor suppressor in human cancer. When a tumor deletes one copy of the TP53 region, it usually takes the neighbours with it, so cells with hemizygous TP53 deletion carry only one copy of POLR2A — see copy number and loss of heterozygosity. In colorectal cancer, POLR2A expression tracked its copy number, and partially suppressing the enzyme — with amanitin or with small interfering RNA — selectively inhibited cells with hemizygous TP53 loss, independently of p53 itself; low doses of an amanitin conjugate caused complete tumor regression in mouse models with that deletion [3]. The gene is not a cancer driver; it is a bystander whose single remaining copy leaves less margin. This is the logic of synthetic lethality applied to a deletion rather than to a mutation. In triple-negative breast cancer, where TP53 is the most frequently mutated or deleted gene, POLR2A was identified by the same reasoning as a collateral vulnerability, and nanoparticle-delivered small interfering RNA against it reduced growth of tumors with hemizygous POLR2A loss [4].

Two independent reasons for this payload: a dosage window created by co-deletion (top) and a toxin that only antibody delivery makes usable (bottom).

Why it cannot be given as a drug. Amanitin is a mushroom toxin that is harmful to the liver and kidneys [5]. Cells in the liver take it up through a specific transporter, OATP1B3, as shown in engineered cells and primary human liver cells [6]. It therefore needs a vehicle that directs it to chosen tumor cells, which is why its clinical development has been as an ADC payload [7].

Why it matters in cancer: which failure it addresses

In the three-level frame on cross-resistance between ADCs, this is a payload-level switch, and a deliberately distant one. A cell that has mutated topoisomerase I, or that over-expresses an efflux pump suited to a TOP1 payload, has changed nothing about RNA polymerase II. The class also claims reach into cells that are not dividing, which neither tubulin nor TOP1 payloads kill efficiently. It does nothing about delivery failure: if the antibody never reaches or enters the cell, the payload is irrelevant.

What could break the bet

• Liver and kidney toxicity. The organ damage of amatoxin poisoning sets the floor on tolerability. Any payload release in the bloodstream, or uptake by cells using OATP1B3, points the toxin at the liver. The nonclinical work reports a promising therapeutic index [2], which is a laboratory claim, not a clinical one; see therapeutic index and ADC payload toxicity.
• Potency forces a narrow design. Because so little toxin suffices, these conjugates carry few payload molecules per antibody and depend on a linker that does not release the drug in circulation; the published conjugates report long half-lives without premature payload release in serum [8]. A linker that leaks turns a delivery vehicle into a poison infusion. See ADC conjugation chemistry.
• Little help for antigen-negative neighbours. Amanitin is taken into liver cells by a transporter rather than by passive diffusion [6], so it is not an obvious candidate for the bystander killing that makes deruxtecan payloads effective in tumors with uneven target expression. Antigen-negative cells inside the tumor may simply be missed — the problem described on target expression versus ADC response.
• The collateral-vulnerability window may be too narrow. The bet is that one gene copy versus two changes sensitivity enough to matter in a patient, with the tumor's single copy also sitting in cells that may up-regulate it. The published work measured copy number and expression in cell lines and models, not in treated patients [3][4].
• One supporting report has been retracted. A 2021 study of an amanitin-conjugated HER2 antibody in HER2-low breast cancer with 17p loss was retracted in 2026; it is not evidence and is not relied on here.

Evidence so far, graded honestly

No human breast cancer data of any kind exist for this payload class. The deletion-based selection hypothesis has not been tested prospectively in people; one preclinical preprint describes enhanced activity in myeloma models engineered to mimic 17p deletion by knocking down POLR2A [10].

What evidence would settle it

A phase 1 study that reports liver enzymes and dose-limiting toxicities against the delivered payload dose, and that pre-specifies POLR2A copy number or 17p status as a stratifying factor. Without that stratification, a response in an unselected group cannot confirm or refute the collateral-vulnerability idea; and a signal in myeloma says little about a solid tumor, where the antibody has to penetrate tissue first.

Common confusions

• Transcription is not DNA copying. Blocking transcription stops the cell making messenger RNA; blocking replication stops it making a second genome. Only the second is confined to dividing cells.
• "Kills resting cells" is not "kills only cancer cells". Resting normal cells transcribe too. The selectivity comes from the antibody and, hypothetically, from gene dosage — never from the payload alone.
• POLR2A loss is not a driver. It is a passenger deletion. Nothing about it helps the tumor; it is a side effect of deleting TP53.
• Hemizygous is not homozygous. The bet needs one working copy. A cell with no copies of POLR2A is not viable, and a cell with two copies has normal margin.
• Amanitin is not a conventional cytotoxic drug that happens to be strong. It has no stand-alone clinical use at all, for the transporter-driven liver reason above.

In Diana's case

How this applies to her is owned by the drug-panel preferences page.

Related concepts

• ADC payload classes
• Cross-resistance between ADCs
• DNA-damage-response inhibitor payloads
• Next-generation TOP1 payloads
• Central dogma
• Copy number and loss of heterozygosity
• Synthetic lethality
• Oncogene addiction versus synthetic lethality
• Metastasis and dormancy
• ADC conjugation chemistry

Sources and scope

Source check: November 7, 2026. Mechanism and preclinical claims from primary papers; clinical status from trial registry records read on that date. One frequently cited breast cancer report in this area has been retracted and is excluded. A different transcription-directed design, a TROP2 dual-payload conjugate pairing exatecan with the transcription inhibitor triptolide, has laboratory work [11] and an unquantified early human report; see the plain-language payload summary.

References

1. Bushnell et al., 2002 — structural basis of transcription: the alpha-amanitin-RNA polymerase II co-crystal.
2. Figueroa-Vazquez et al., 2021 — an anti-BCMA amanitin conjugate kills proliferating and resting myeloma cells, spares antigen-negative cells, and induces tumor regression in mouse models; RNA polymerase II named as a target in proliferating and dormant cancer cells.
3. Liu et al., 2015 — POLR2A is co-deleted with TP53; hemizygous TP53 loss creates sensitivity to POLR2A suppression by alpha-amanitin, and an amanitin conjugate caused complete regressions in colorectal cancer models. An erratum was published in 2021.
4. Xu et al., 2019 — POLR2A as a collateral vulnerability in triple-negative breast cancer, targeted with small interfering RNA in nanoparticles. Two errata were published.
5. Liu et al., 2020 — systematic review of amatoxin poisoning treatment: gastrointestinal illness with liver and renal failure, and mortality figures across 506 reported patients.
6. Letschert et al., 2006 — OATP1B3 identified as the human hepatic uptake transporter for amatoxins; inhibitors of the transporter prevented the cellular lesion.
7. Pahl, Lutz and Hechler, 2018 — review of amanitins as an ADC payload and of the development of the technology around the toxin.
8. Papacharisi et al., 2025 — TROP2-directed amanitin conjugates: picomolar potency, half-life of about 10 days without premature payload release in serum, and tumor eradication in pancreatic and triple-negative breast cancer xenograft models at tolerated doses.
9. ClinicalTrials.gov records read through the registry application programming interface: the BCMA-directed amanitin conjugate dose-finding study and a second early study. Response counts: sponsor nine-month 2026 report (not peer-reviewed; 13 of 31 evaluable pooled across dose groups 5–9).
10. Singh et al., 2024 — preprint, not peer-reviewed: enhanced activity of the same conjugate in models engineered to mimic 17p deletion.
11. Ren et al., 2026 — a dual-payload ADC platform integrating exatecan and triptolide; preclinical.

Used in

• Read a new payload strategy
• Aiming a drug at a surface target