Cromolyn as a novel pharmacophore of the Zα domain of the RNA-editing enzyme ADAR1p150

Abstract

ADAR1p150 is a critical RNA-editing enzyme for maintaining cellular homeostasis through dsRNA binding, protein–protein interactions, and adenosine-to-inosine (A-to-I) editing. Beyond its dsRNA binding domains, ADAR1p150 contains a Zα domain that can induce a conformational switch of dsDNA and dsRNA from stable B/A forms to a higher-energy left-handed Z form. By stabilizing Z-RNA, ADAR1p150 is thought to modulate immune activation by competing with dsRNA sensors like MDA5 and ZBP1. ADAR1p150’s editing activity minimizes dsRNA’s presence and prevents dsRNA-mediated inflammatory pathways’ activation. Our study employs NMR to introduce a novel pharmacophore model for Zα domain binders. We identify cromoglicic acid, also known as the FDA-approved drug cromolyn, as an ADAR1 Zα binder that competes with nucleic acid recognition. Cromolyn does not bind to ZBP1 Zα domains, which are the only other human Zα domains. Our work paves the way for effectively modulating ADAR1p150 function with small molecules, opening new avenues for enhancing anti-cancer immune responses.

1 |. INTRODUCTION

Adenosine Deaminase Acting on RNA 1 (ADAR1) is a crucial RNA-editing enzyme responsible for converting adenosine to inosine in double-stranded RNA (dsRNA), a process essential for distinguishing self from non-self RNA (Chung et al., 2018). Known for its role in the immune response, ADAR1 has also recently been identified as an anti-tumor target (Ishizuka et al., 2019). ADAR1 exists in two isoforms: the p150 isoform is primarily found in the cytoplasm and is interferon (IFN) inducible, while the shorter p110 localizes to the nucleus and is constitutively expressed. Both isoforms possess a deaminase catalytic domain, three dsRNA-binding domains, and a Zβ domain (Figure 1a). However, only p150 contains Zα within its N-terminal extension. Zα-ADAR1p150 binds to right-handed dsRNA and dsDNA and converts these duplexes to the higher energy left-handed Z-conformation (Nichols et al., 2024) (Figure 1b). Z-DNA-Binding Protein 1 (ZBP1) is the only other Zα domain-containing human protein, detecting Z-DNA/RNA to initiate cell death pathways (Cui et al., 2024; Karki et al., 2021) (Figure 1c). Another role of ZBP1 is to act as a cytosolic nucleic acid sensor, triggering inflammatory signaling pathways by interacting with RIPK1 and RIPK3 (Peng et al., 2022). Z-form stabilization by Zα domains occurs through structure-specific contacts, and mutation of key residues at the interface abolishes this ability (Karki et al., 2021; Nichols et al., 2023). ADAR1p150 is thought to prevent ZBP1-mediated activation of necroptosis and other inflammatory responses through competition exerted by its Zα domain (Zhang et al., 2022).

FIGURE 1.

Z-RNA binding domains (Zα) in ADAR1 and ZBP1. (a) Schematic of the domain architecture of ADAR1, highlighting the Zα domain in ADAR1p150 responsible for Z-RNA binding. (b) Crystal structure of two Zα domains of ADAR1p150 (red and green) bound to Z-RNA (blue and cyan) (PDB ID: 2GXB). (c) Domain architecture of ZBP1, showing two Zα domains (Zα1 and Zα2).

ADAR1 plays a critical role in both innate immunity and disease pathogenesis. When its activity is disrupted, problems arise, such as hypo-editing, leading to the accumulation of unedited dsRNA, triggering a detrimental interferon response as in diseases like Aicardi-Goutières Syndrome and type 1 diabetes. Conversely, hyper-editing is linked to autoimmune disorders such as rheumatoid arthritis (Zhang et al., 2025). In addition, some tumors become dependent on ADAR1’s immunosuppressive function to evade the immune system. This dual potential—to either suppress or activate the immune response depending on the therapeutic goal—makes ADAR1 a highly attractive and versatile drug target. It has been proposed that targeting ADAR1 would, for example, allow unmasking of tumor-associated antigens and make cancer cells vulnerable to immune checkpoint blockade therapies (Tamizkar & Jantsch, 2025).

Past research reported on computational approaches for screening large compound libraries for binding to Zα of ADAR1p150 (Choudhry, 2021; Hong et al., 2024). They were accompanied by various biochemical and sometimes cellular assays, leading support to the hypothesis that compounds can be identified that would act as ADAR1 inhibitors. One of the compounds identified as a potential binder through molecular docking was the FDA-approved drug cromolyn, a mast cell stabilizer used in the treatment of asthma, although its exact molecular mechanism of action remains unclear (Kuzubova et al., 2017). No experimental evidence was proposed to support such claims. Collectively, these studies did not systematically monitor direct binding to Zα, and whether such binding would translate into a disruption of Zα’s main function to recognize Z-D/RNA.

Here, we used NMR and other biophysics approaches to demonstrate the binding of cromolyn to Zα-ADAR1p150. We also show evidence that cromolyn does not bind to ZBP1’s Zα domains, indicating that specificity to ADAR1 can be achieved through Zα binding. Importantly, we also reveal that cromolyn binds at the interface involved in Z-D/RNA recognition, leading to Z-D/RNA binding inhibition. Our work confirms Zα as a drug target and presents cromolyn or its derivatives as promising lead compounds for therapeutic development.

2 |. RESULTS AND DISCUSSION

Cromolyn binds Zα-ADAR1p150 and competes with nucleic acids. Because cromolyn was not in the top 10 of the best binders according to a previously published in silico study (Choudhry, 2021), we first tested the top three proposed candidates (alendronate, etidronate, and zoledronate; Figure S1a), using 2D ¹⁵N-Heteronuclear Single-Quantum Coherence (HSQC) NMR experiments. Such experiments show a single spectral peak for every amino acid except proline and are used as a fingerprint to test for binding. Upon binding of a small molecule, peaks of the amino acids affected by binding are expected to shift their position (chemical shift perturbation, CSP) or reduce their peak intensity (Henen et al., 2012). Zα peaks showed no shifts, or random non-significant shifts most probably due to local pH alterations upon addition of the ligands (Figure S1b). This indicated that the top candidates, according to the in silico study, do not bind to Zα-ADAR1p150.

Further down the list (Choudhry, 2021), cromolyn appeared as a more promising candidate, due to its distinctive chromone moiety (Figure 2a). HSQC data revealed that cromolyn binds to the Zα domain of ADAR1p150 (Figure 2b). Our mapping of the CSPs to the Zα structure showed that the peaks shifting the most are within α-helix 3 (Figure S2). This helix is a hotspot for binding to nucleic acids, including residue Tyr 177, a key amino acid for specific recognition of Z-D/RNA (Feng et al., 2011).

FIGURE 2.

Cromolyn binds Zα-ADAR1p150 at α-helix 3. (a) Structure of cromolyn, a clinically approved drug that binds to Zα domain of ADAR1p150; the chromone moiety is highlighted in red. (b) Overlay of NMR ¹⁵N-HSQC spectra of free Zα domain (orange) and Zα domain with cromolyn (green) indicates chemical shift perturbation specifically in α-helix 3 of Zα. A zoomed view of a selected region is provided on the right.

To check whether nucleic acid binding would be hampered by cromolyn binding to Zα-ADAR1, we used 2D ¹⁵N-HSQC NMR experiments to characterize the change in binding signature between dsDNA fragment d(CpG)₃ and Zα-ADAR1 in the absence or the presence of cromolyn. The data showed that some Zα peaks, which shifted upon DNA binding, moved back toward positions observed in the spectra of the free Zα domain or the cromolyn-bound Zα domain (Figure 3a). This further indicates that cromolyn is binding to α-helix 3 and competing with nucleic acids (Figure 3b).

FIGURE 3.

Cromolyn competes with DNA for binding to Zα-ADAR1p150 (a) Overlay of NMR ¹⁵N-HSQC spectra of free Zα domain (orange), and Zα domain with d(CpG)₃ alone (blue) or with d(CpG)₃ in the presence of cromolyn (cyan). The chemical shift perturbations caused by DNA binding to the Zα domain moved back toward positions observed in the spectra of the free Zα domain or the cromolyn-bound Zα domain, suggesting competitive binding between cromolyn and d(CpG)₃. A zoomed view of a selected region is provided on the right. (b) 3D representation of the Zα domain in complex with d(CpG)₃. Cromolyn binds to α-helix 3 (purple) in Zα-ADAR, which is the same binding site as for nucleic acids.

Cromolyn selectively binds to the Zα domain of ADAR1 but not to the Zα domains of ZBP1.

To test the in vitro selectivity of cromolyn against other Zα domains, we explored whether cromolyn was able to bind to other human Zα domains (Zα1 and Zα2 from ZBP1). We conducted 2D ¹⁵N-HSQC probed binding between cromolyn and Zα1 and Zα2, whose NMR resonances we have previously assigned (Beck et al., 2024). Our data showed that cromolyn does induce random insignificant CSP (Figure S3), indicating that it binds to neither of the Zα domains of ZBP1 (Figure 4a,b). Although preliminary, this finding indicates that selectivity is achievable for Zα binders.

FIGURE 4.

Cromolyn does not bind to Zα-ZBP1 domains. (a) Overlay of NMR ¹⁵N-HSQC spectra of free Zα1-ZBP1 domain (cyan) and Zα1 domain with cromolyn (red) shows no chemical shift perturbation. (b) Overlay of NMR ¹⁵N-HSQC spectra of free Zα2-ZBP1 domain (purple) and Zα2 domain with cromolyn (green) shows no chemical shift perturbation.

Affinity measurement reveals weak affinity binding of cromolyn.

We carried out microscale thermophoresis (MST) to specifically derive the dissociation constant (K_D) of cromolyn binding to Zα. As MST requires minimal protein amounts, it overcomes the high ligand concentration and solubility limitations often encountered in NMR-based or biochemical approaches. Our MST titration of cromolyn into a solution containing Zα-ADAR1 indicated the presence of binding; however, saturation was not achieved due to weak affinity. We can therefore estimate K_D to be >774 μM ± 301 μM (Figure 5). This is consistent with the micromolar range expected at this early stage of a drug discovery project (Deloche et al., 2023; Silvestre et al., 2013). It is not uncommon for FDA-approved drugs (e.g., vemurafenib (Flaherty et al., 2011) or venetoclax (Fairbrother et al., 2019)) to originate from compounds with K_D values of several hundreds of μM. Furthermore, our measurements indicate that MST provides a reliable starting point for our upcoming medicinal chemistry efforts to study cromolyn derivatives with improved characteristics.

FIGURE 5.

Cromolyn binding affinity of Zα-ADAR1 measured by MST. Ten serial dilutions of cromolyn were mixed with fluorescently labeled Zα-ADAR1 in premium capillary tubes. The K_D value is 774 ± 301 μM.

A promising scaffold for developing Zα-ADAR1p150 inhibitors.

We explored the potential of chromone, the primary structural ring of cromolyn, as a scaffold for future medicinal chemistry approaches aiming at improving binding and specificity. Chromone derivatives are much sought after scaffolds in medicinal chemistry, due to their structural versatility and various pharmacological properties leading to diverse bioactivities (Benny et al., 2021). We utilized 6-fluorochromone-2-carboxylic acid as a model compound, due to its single ¹⁹F atom incorporated into the chromone moiety. This allowed us to take advantage of the precision and sensitivity of ¹⁹F-NMR spectroscopy, a powerful tool for studying ligand interactions with remarkable efficiency (Vulpetti & Dalvit, 2013). Our data showed two fluorine peaks corresponding to a major and a minor conformation/state of the compound (Figure 6). Upon adding Zα at different concentrations (10–40 mM) to 1 mM of the small molecule, we observed over 50% reduction in the major peak’s intensity and CSP. At higher Zα concentrations, peak broadening further masked J-coupling splitting, suggesting more broadening upon addition of more protein. Interestingly, the minor peak showed CSPs and an increase in intensity, suggesting a shift in the population toward this second state upon interaction. We speculate this second state to be arising from a second rotamer (syn vs. anti) of the carboxylic group relative to the chromone ring due to rotation about the C—C bond that links the carboxyl to the chromone system. These findings demonstrate the feasibility and promise of using chromone derivatives to identify novel binders with better affinity and subsequently use these ligands for cell-based assays.

FIGURE 6.

¹⁹F NMR confirms the binding between fluorochromone and Zα domain. Overlay of NMR ¹⁹F spectra of free fluorochromone (green) with fluorochromone in the presence of 10, 20, and 40 μM of Zα domain (red, blue, and purple, respectively). The major peak of the fluorine atom loses more than 50% of its intensity upon the addition of Zα, and shows chemical shift perturbation and peak broadening. The minor peak shows an increase in intensity and chemical shift perturbation.

3 |. CONCLUSIONS AND PERSPECTIVES

Our study provides biochemical and biophysical insights into the interaction between Zα-ADAR1p150 and cromolyn, an FDA-approved drug for treating asthma and allergic rhinitis (Kuzubova et al., 2017). We showed experimental evidence that cromolyn binds to Zα-ADAR1p150 with low affinity, yet it does not bind to Zα domains of ZBP1. This emerging selectivity paves the way for developing chromone-based ADAR1 modulators. Interestingly, our preliminary findings about a biological effect potentially correlated to hampered Z-form recognition indicate that dsRNA levels in CD8 T cells expressing ADAR1p150 increase upon addition of cromolyn (Figure S4). This is intriguing, as a rise in dsRNA levels could be due to a decreased editing activity of ADAR1p150, which is compatible with cromolyn interfering with normal ADAR1p150 functions. However, this observation deserves further scrutiny, as Zα does not seem to generally be involved in editing (Nichols et al., 2025), and it is hard to reconcile the effect observed in cells at a 25 μM cromolyn concentration when the measured K_D is much higher. Nonetheless, our work proposes that cromolyn, through its chromone moiety, may be used as a pharmacophore to develop novel potential therapy to enhance immune response in cancer treatment.

Supplementary Material

Additional supporting information can be found online in the Supporting Information section at the end of this article.

DATA AVAILABILITY STATEMENT

Research data are not shared.