The receptors at the heart of this work belong to the nuclear receptor superfamily, a large group of proteins that act as molecular switches controlling gene expression. Unlike their better-studied cousins such as the estrogen or glucocorticoid receptors, NR4A1 and NR4A2 are called orphan receptors because no natural ligand has been definitively identified for them. In most solid tumors, including colon cancer, both receptors behave as pro-oncogenic factors: they help cancer cells proliferate, resist apoptosis, invade surrounding tissue, and survive metabolic stress. Because they sit at the intersection of so many tumor-promoting pathways, they have long been considered attractive drug targets, but exploiting them therapeutically requires molecules that can bind and modulate their activity in a controlled way.

The Safe laboratory has spent years developing a class of synthetic ligands based on a chemical scaffold known as 1,2-bis(3′-indolyl)-1-(3,5-disubstitutedphenyl)methane, abbreviated DIM-3,5. These compounds are built around two indole rings, the same structural motif found in compounds derived from cruciferous vegetables, connected through a central methane carbon to a substituted phenyl ring. By varying the substituents on that phenyl ring, the chemists can tune how the molecule interacts with NR4A1 and NR4A2. In the new study, the researchers focused on dual NR4A1/2 ligands and asked a specific question: can these compounds push colon cancer cells into ferroptosis, an iron-dependent form of regulated cell death that has attracted enormous interest as a vulnerability in tumors resistant to conventional therapies?

Ferroptosis is mechanistically very different from apoptosis. Instead of the orderly cellular dismantling characteristic of programmed cell death, ferroptosis is driven by catastrophic oxidative damage to cell membranes. When the lipid bilayer is not adequately protected by antioxidant systems, polyunsaturated fatty acids in the membrane undergo peroxidation, generating reactive oxygen species and lipid peroxides that rupture the cell from the outside in. The central guardian against this process is the enzyme glutathione peroxidase 4, or GPX4, which detoxifies lipid hydroperoxides using glutathione as a cofactor. Closely linked to this defense is the cystine/glutamate antiporter SLC7A11, which imports cystine needed to replenish glutathione. Tumor cells, which already operate under high oxidative stress, are often exquisitely dependent on this GPX4-SLC7A11 axis, making it a promising therapeutic chokepoint.

The experimental results reported in the study show that the DIM-3,5 ligands do exactly what a ferroptosis-inducing agent should do. Treatment of colon cancer cells led to a marked enhancement of reactive oxygen species, detected using BODIPY fluorescence, a widely used probe whose signal shifts when lipid peroxidation accumulates in membranes. Consistent with this, the cells produced increased amounts of malondialdehyde, a well-established degradation product of lipid peroxides that serves as a chemical fingerprint of membrane damage. These biochemical changes were accompanied by a characteristic reprogramming of ferroptosis-related gene expression: the transferrin receptor, also known as TFR or CD71, was induced, while GPX4 and SLC7A11 were both decreased. Inducing the transferrin receptor increases iron uptake into the cell, feeding the Fenton chemistry that drives lipid peroxidation, while simultaneously suppressing the two main antioxidant safeguards. The net effect is a convergence of pressure on the membrane lipids from multiple directions at once.

One of the most technically significant contributions of the paper is the molecular dissection of how NR4A receptors control GPX4 expression in the first place. Using knockdown experiments, promoter analysis, and chromatin immunoprecipitation, the team demonstrated that GPX4 transcription is regulated by a partnership between NR4A2 and the transcription factor Sp4. Sp4 binds directly to specific DNA sequences in the GPX4 promoter, and NR4A2 functions as a ligand-dependent cofactor for the DNA-bound Sp4. In other words, NR4A2 does not simply float in the nucleus as an independent regulator; it is recruited to the promoter through its interaction with Sp4, and its activity there depends on the ligand occupying its binding pocket. When the DIM-3,5 ligands engage NR4A2, this cofactor arrangement is disrupted, and the GPX4 gene is no longer efficiently transcribed. This mechanism explains how a single compound can lower the cell’s antioxidant shield and provides a concrete structural and functional rationale for targeting NR4A2 in ferroptosis-based cancer therapy.

Importantly, the researchers did not stop at cell culture experiments. To test whether the compounds could work in a living organism with an intact immune system, they used a syngeneic mouse xenograft model in which CT26 mouse colon cancer cells are implanted into immunocompetent mice. Syngeneic models are particularly valuable because, unlike human tumor cells grown in immunodeficient mice, they allow the full interplay between the tumor and the host immune system to unfold. The team treated the animals with the 3,5-dichloro analog of the DIM-3,5 series, referred to as DIM-3,5-Cl2, at a dose of 2.5 milligrams per kilogram per day. The result was a significant inhibition of colon tumor growth, confirming that the ferroptotic and gene-regulatory effects observed in vitro translate into tangible antitumor activity in vivo.

Perhaps the most striking findings came from the analysis of the tumor-infiltrating lymphocytes and other immune cells within the treated tumors. Flow cytometric profiling revealed that DIM-3,5-Cl2 treatment increased the population of CD8-positive T cells, the cytotoxic lymphocytes that directly kill tumor cells and are the primary mediators of successful cancer immunotherapy. At the same time, the treatment decreased the abundance of granulocytic myeloid-derived suppressor cells, or g-MDSC, a population of immature myeloid cells that potently suppresses antitumor immune responses and is associated with poor prognosis in many cancers. The ratio of M1 to M2 macrophages also shifted favorably, moving away from the immunosuppressive, tumor-promoting M2 phenotype and toward the inflammatory, tumoricidal M1 phenotype. In addition, the percentage of B cells within the tumor microenvironment increased, further supporting enhanced immune surveillance.

This dual mechanism, direct ferroptosis induction combined with immune reprogramming, is what makes the study noteworthy beyond the colon cancer field. Ferroptosis and immunity are increasingly recognized as deeply intertwined processes. When tumor cells die by ferroptosis, they release damage-associated molecular patterns and oxidized lipid mediators that can stimulate dendritic cells and promote cross-presentation of tumor antigens to T cells. Conversely, immunosuppressive myeloid cells within tumors secrete factors that help cancer cells resist lipid peroxidation. By simultaneously weakening the tumor’s antioxidant defenses and dismantling the suppressive immune circuitry that shields it, the DIM-3,5 ligands appear to attack the cancer from two complementary directions, a strategy that mirrors the logic of successful combination immunotherapies but achieves it with a single small molecule.

The work, which was supported by the Syd Kyle Chair Endowment, Texas A&M AgriLife, and the National Institutes of Health, is still at the preclinical stage, and the path from a mouse xenograft model to an approved human therapy is long and uncertain. Dose optimization, pharmacokinetics, toxicity profiling, and testing in additional tumor models will all be necessary before NR4A ligands could enter clinical trials. Nevertheless, the study establishes a clear mechanistic framework: NR4A1 and NR4A2, long appreciated as pro-oncogenic transcriptional regulators, can be pharmacologically exploited to collapse the GPX4-dependent antioxidant defense through the NR4A2-Sp4 cofactor mechanism, thereby triggering ferroptosis, while the same ligands reshape the tumor immune microenvironment in ways that favor immune-mediated tumor control. For a disease as stubborn as colon cancer, a molecule that turns the tumor’s own lipid chemistry against it and simultaneously invites the immune system to the scene represents a genuinely compelling new direction.

Subject of Research: NR4A nuclear receptor ligands as inducers of ferroptosis and immune modulation in colon cancer

Article Title: Orphan nuclear receptor 4A (NR4A) ligands induce ferroptosis and modify immune function in colon cancer

Article References: Oany, A. R., Upadhyay, S., Wright, G. A., Farkas, E., Sivaram, G., Tsui, W. N. T., Hailemariam, A., Chen, S.-H., & Safe, S. (2026). Orphan nuclear receptor 4A (NR4A) ligands induce ferroptosis and modify immune function in colon cancer. Cell Death Discovery. https://doi.org/10.1038/s41420-026-03369-4

Image Credits: AI Generated

DOI: 10.1038/s41420-026-03369-4

Keywords: NR4A1, NR4A2, ferroptosis, colon cancer, GPX4, DIM-3,5 ligands, reactive oxygen species, tumor microenvironment, CD8 T cells, myeloid-derived suppressor cells, immunotherapy, nuclear receptors

Enhancers act as molecular switches that integrate signals and guide gene expression programs. However, their activity depends on complex combinatorial arrangements of sequence features, such as transcription factor binding sites and other regulatory motifs. Until now, most insights into enhancer function have come from indirect biochemical assays or computational predictions, with little direct evidence connecting specific nucleotides within these regions to their regulatory activity in a living organism. The current study bridges this critical gap by utilizing a large-scale in vivo mutagenesis screen in transgenic mice to scrutinize the functional impact of targeted sequence alterations across multiple human enhancers active in developing key tissues.

Specifically, the research team focused on seven human enhancers that are known to be active in critical developmental structures including the brain, heart, limb, and face. These enhancers represent regulatory hubs that control genes essential for proper morphogenesis. Through the generation of more than 1,700 transgenic mice harboring over 260 distinct mutant enhancer alleles, the investigators systematically introduced mutations in 12-base-pair blocks tiled across the entire enhancer sequences. This design allowed them to probe every conceivable sequence feature within each enhancer, observing the downstream effects on reporter gene activity as a proxy for enhancer function.

The mutagenesis screen uncovered a striking revelation: nearly 69% of all mutated blocks were indispensable for the normal in vivo activity of their respective enhancers. This high proportion underscores a dense packing of functional elements scattered throughout the enhancer DNA, with mutations frequently leading to a loss of enhancer activity. Indeed, about 60% of the mutagenic events caused loss-of-function phenotypes, while a smaller subset, approximately 9%, resulted in gain-of-function effects. This asymmetry highlights the delicate sensitivity of enhancer sequences to disruption and suggests that most sequence features play a positive and essential role in maintaining regulatory output.

One of the study’s most compelling achievements was the application of machine learning models to predict which individual nucleotides within the enhancers are critical for function. By training algorithms on the mutagenesis data, the researchers generated high-resolution annotations pinpointing bases whose alteration would profoundly affect enhancer activity. Remarkably, 88% of sequence motifs identified by these predictive models corresponded with observed changes in functional activity in vivo. Furthermore, the models achieved a sensitivity rate of 59%, meaning they correctly flagged a majority of functionally relevant sequence blocks, representing a powerful computational tool for interpreting enhancer logic.

This integrated experimental and computational framework establishes a new paradigm for decoding enhancer architecture at base-pair resolution. The findings indicate that human developmental enhancers are not composed of sparse key motifs embedded within inert DNA, but rather possess a high density of functional features, each contributing to the nuanced regulation of gene expression. This complexity likely reflects the evolutionary pressures to finely tune developmental programs and may account for the sensitivity of these sequences to pathogenic mutations.

The implications of this research extend far beyond basic developmental biology. Since many disease-associated genetic variants reside in non-coding regions, including enhancers, the ability to map mutational sensitivities provides a critical foundation for interpreting how such variants perturb gene regulation. This in turn could facilitate more accurate diagnoses, enable the design of targeted therapies, and improve the prediction of phenotypic outcomes in genetic counseling.

Importantly, the study’s extensive dataset, generated through rigorous in vivo experimentation, serves as a rich resource for the broader scientific community. Researchers studying gene regulation, evolutionary biology, and human disease genetics can leverage these insights to explore enhancer function, identify disease-causing variants, and develop synthetic enhancers for therapeutic applications. The approach exemplifies how coupling systematic mutagenesis with sophisticated computational modeling can unravel the complexity of the non-coding genome.

Moreover, the study addresses long-standing questions about enhancer robustness and redundancy. Despite the high functional density, the observed gain-of-function mutations suggest that certain sequence alterations can create new regulatory activities, pointing to latent potential within enhancer sequences to evolve new functions. This plasticity within enhancer architecture could have significant evolutionary implications, contributing to phenotypic diversity and adaptation.

The research also demonstrates the power of using transgenic mice as an in vivo assay system to study human regulatory DNA. The conservation of regulatory mechanisms between humans and mice allows functional investigation while preserving the physiological context of developing tissues. This approach surpasses traditional in vitro assays by capturing the full complexity of chromatin landscape, three-dimensional genome architecture, and relevant cellular environments.

Looking forward, further refinement and expansion of mutagenesis-based enhancer mapping across additional tissues and developmental stages will deepen our understanding of the dynamic regulatory genome. Integration with single-cell transcriptomics, epigenomics, and 3D genomics could reveal how enhancers coordinate with one another and with other regulatory elements to choreograph developmental trajectories and maintain tissue homeostasis.

In conclusion, this pioneering work not only elucidates the dense and functionally rich nature of human developmental enhancers but also offers a robust methodological blueprint for disentangling the regulatory code embedded within non-coding DNA. As our ability to interpret non-coding variation improves, so too will our capacity to tackle genetic diseases and harness the power of genome editing for regenerative medicine and gene therapy.

Subject of Research: Functional architecture and mutagenesis sensitivity of human developmental enhancers in vivo

Article Title: In vivo mapping of mutagenesis sensitivity of human enhancers

Article References:
Kosicki, M., Zhang, B., Hecht, V. et al. In vivo mapping of mutagenesis sensitivity of human enhancers. Nature (2025). https://doi.org/10.1038/s41586-025-09182-w

Image Credits: AI Generated