NUDT2 loss defines quantitative limits for dinucleoside polyphosphate action on the cGAS–STING–TBK1 axis

  1. Paul Weiland
  2. Renuka Dharani Shivakumar
  3. Ekaterina Jalomo-Khayrova
  4. Julia Schmidt
  5. Victor Zegarra
  6. Ying Wang
  7. Nicole Paczia
  8. Stephan Kiontke
  9. Andreas Burchert
  10. Gert Bange

  • Curated by eLife

    eLife logo

    eLife Assessment

    This valuable study investigates the role of dinucleoside polyphosphates in regulating the cGAS-STING-TBK1 axis of innate immunity. The data collected using multiple biochemical and cellular assays, plus X-ray crystallography, are solid, but testing additional polyphosphates and using NUDT2KO cells will further strengthen the main conclusion. This paper will be broadly relevant to those studying various innate immune pathways.

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

NUDT2 is an emerging candidate for therapeutic intervention in cancer, and its inhibition or loss is known to elevate adenosine-containing dinucleoside polyphosphates (ApnNs), including diadenosine tetraphosphate (Ap4A). Ap4A is a stress- and immune-associated nucleotide metabolite proposed to act as a second messenger, raising the possibility that NUDT2 targeting may unintentionally affect important nucleotide-sensitive signaling pathways. One such pathway is cGAS-STING signaling, a central innate immune axis that detects cytosolic DNA, produces the nucleotide second messenger 2′3′-cGAMP, and drives type I interferon responses. Because cGAS-STING also contributes to antitumor immunity and is being pharmacologically targeted in cancer, we asked whether sustained Ap4A accumulation perturbs this pathway. We systematically evaluated Ap4A and related dinucleoside polyphosphates across the cGAS-STING-TBK1 axis using biophysical, enzymatic, structural, and cellular approaches. Contrary to a previous model, STING did not bind Ap4A, Ap3A, or Ap4G, despite robust binding of canonical cyclic dinucleotides. Although cGAS bound these nucleotides with micromolar affinities, DNA-activated cGAMP synthesis was inhibited only at high, supra-substrate ratios. Similarly, TBK1 inhibition required extreme Ap4A ratios beyond physiologically relevant levels. In a THP-1 cell model, NUDT2 knockout caused strong Ap4A accumulation, but the resulting intracellular dinucleoside polyphosphate levels remained below the ratios required to inhibit cGAS or TBK1 in vitro. This study thus distinguishes biochemical possibility from physiological relevance and argues that NUDT2-linked Ap4A accumulation is unlikely to directly compromise cGAS-STING pathway activity.

Article activity feed

  1. eLife

    eLife Assessment

    This valuable study investigates the role of dinucleoside polyphosphates in regulating the cGAS-STING-TBK1 axis of innate immunity. The data collected using multiple biochemical and cellular assays, plus X-ray crystallography, are solid, but testing additional polyphosphates and using NUDT2KO cells will further strengthen the main conclusion. This paper will be broadly relevant to those studying various innate immune pathways.

  2. eLife

    Reviewer #1 (Public review):

    Summary:

    The manuscript entitled "NUDT2 loss defines quantitative limits for dinucleoside polyphosphate action on the cGAS-STING-TBK1 axis" investigates whether intracellular levels of dinucleoside polyphosphates can influence signaling through the cGAS-STING pathway. The study was motivated in part by a previous report proposing that Ap4A can bind to STING and inhibit its activity.

    In the present work, the authors prepared recombinant human and mouse cGAS, STING, and TBK1 and systematically investigated their interactions with Ap3A, Ap4A, and Ap4G, which represent biologically relevant dinucleoside polyphosphates. To characterize these interactions, they employed a broad range of biochemical and structural approaches, including isothermal titration calorimetry (ITC), HPLC analysis of enzymatic reactions, X-ray crystallography, and a time-resolved fluorescence kinase assay.

    First, the authors prepared human and mouse STING in both wild-type form and as specific mutants previously associated with altered binding of the canonical STING ligand cGAMP. ITC measurements showed that, whereas cGAMP binds STING with nanomolar affinity, none of the tested dinucleoside polyphosphates displayed detectable binding to any of the STING variants. The authors then investigated binding of ATP, GTP, and the dinucleoside polyphosphates to a mutant form of cGAS by ITC. Interestingly, all three dinucleoside polyphosphates showed apparent binding in a range similar to that of the tested NTPs. However, subsequent structural analysis did not reveal electron density corresponding to the dinucleoside polyphosphates in the active site.

    The authors further examined the effect of Ap4A on TBK1 and found that Ap4A can inhibit TBK1 activity, but only at very high concentrations. Finally, they quantified intracellular ATP, GTP, Ap3A, Ap4A, and Ap4G levels in NUDT2-knockout cells. Although Ap4A and Ap4G concentrations increased substantially upon NUDT2 deletion, the measured intracellular concentrations generally remained below those required to significantly affect components of the cGAS-STING-TBK1 pathway in vitro.

    Overall, the manuscript is clearly written and represents a careful and rigorous piece of biochemical work. I particularly appreciate the authors' effort to critically examine and clarify conclusions from previous studies. The experimental work is comprehensive.

    Strengths:

    A major strength of the study is the breadth and rigor of the experimental strategy. The authors combine binding measurements, enzymatic assays, structural analysis, kinase assays, and direct measurements of intracellular metabolite concentrations. Their data provide convincing evidence that the tested dinucleoside polyphosphates do not bind STING with physiologically relevant affinity and that inhibition of TBK1 by Ap4A occurs only at concentrations substantially higher than those normally reached in cells. Importantly, the authors also quantify the intracellular concentrations of these metabolites in cells lacking NUDT2 and show that, although Ap4A and Ap4G levels increase significantly, they generally remain below the concentrations required to affect the tested signaling proteins in vitro. This quantitative comparison between biochemical effects and intracellular metabolite levels is particularly valuable and strengthens the physiological interpretation of the study.

    Weaknesses:

    Interpretation of the pppGpA density in the cGAS structure.

    Importantly, when Ap4G was used in the crystallographic experiments with cGAS, the authors observed density corresponding only to pppGpA. They interpret this observation as potentially resulting from processing of Ap4G by DNA-bound cGAS to generate a canonical reaction intermediate.

    Based on my direct experience with structural studies of an unrelated enzyme using commercially supplied Ap4G from Jena Bioscience, I believe that an alternative explanation should be carefully considered. In our experience, Ap4G preparations can contain small amounts of pppGpA, likely at the per mille level, as a side product of the chemical synthesis. If the studied protein has substantially higher affinity for pppGpA than for Ap4G, even such a minor contaminant can become strongly enriched in the active site during crystallization and consequently dominate the observed electron density.

    Indeed, traces of a corresponding contaminant may already be visible in the HPLC profile presented in Figure S2.3C. I therefore strongly suggest repurifying Ap4G by HPLC and repeating, at minimum, the HPLC-based experiments shown in Figure S2.3C using the purified material. Ideally, the purity of the Ap4G preparation used for crystallography should also be carefully documented.

    Until this possibility has been excluded experimentally, I would recommend that the authors substantially temper the conclusion that pppGpA observed in the crystal structure originates from cGAS-mediated processing of Ap4G.

  3. eLife

    Reviewer #2 (Public review):

    Summary:

    Dinucleoside polyphosphates have been found to increase in responses to cellular stress and are involved in the regulation of transcription, antibiotic resistance, DNA replication and innate immunity in metazoa. Increased Ap4A levels have been previously found to regulate the cGAS/STING pathway, leading to the proposal that modulation of Ap4A levels may be a suitable pathway to pharmacologically modulate cGAS/STING signalling (Guerra et al., 2020). Here, Weiland et al. present compelling biochemical, biophysical and structural evidence that dinucleoside polyphosphates (Ap3/4A and Ap4G) do not interact with STING nor are likely to affect cGAS signalling in vivo.

    The evidence that they present evaluates the interactions of Ap3/4A and Ap4G s with cGAS, STING and TBK1 kinase as key signalling molecules in the cGAS/STING innate immune response. Previous work has identified Ap4A as a regulator of cGAS/STING signalling, and through precise characterisation of the biophysical interaction, crystallographic as well as modelling data provide evidence that these effects are too weak to affect these processes in vivo. Their work identifies that there are no direct interactions between STING and dinucleoside polyphosphates at physiological concentrations and at hyperactivated levels in NUDT2 mutants. There are detectable interactions of both Ap4A/and Ap4G with cGAS. In the presence of DNA, cGAS shows either ATP when bound with Ap4A or an intermediate of the cGAMP synthesis when crystallised with Ap4G. This is an interesting observation, showing Ap4G appears to serve as a weak substrate for canonical cGAMP production in the presence of DNA. Finally, the effect of Ap4A on the catalytic activity of TBK1 was evaluated and showed very weak effects. Competition assays in the presence of ATP and Ap4A/Ap4G showed a higher inhibitory effect for Ap4G, but only at concentrations 5x or greater than intracellular ATP levels and much weaker effects for Ap4A (~32x ATP levels).

    Strengths:

    The approaches used provide robust data for a subset of physiological Ap4Ns. The strength of their evidence comes from the complementary structural and biophysical approaches investigating the interactions of cGAS, STING and TBK1 independently. These data, including the dissociation constants for Ap4A and Ap4G with cGAS, allow for comparison with physiological levels of the dinucleoside polyphosphate in vivo. The use of quantitative mass spectrometry provides some increased rigor to the conclusions by integrating the in vitro and in vivo studies.

    Weaknesses:

    The data presented for Ap4G and Ap4A are convincing, and conclusions are consistent with the evidence presented. However, there are other Ap4Ns (Ap4C/U) that may be active against cGAS/STING/TBK1 that are not included in this study, which weakens the conclusions. This could be rectified relatively simply. The data would be more robust if the effects of Ap4C and Ap4U were included for cGAS using ITC, and also the additive effect of competition assays with physiological ATP and the levels of all Ap4Ns in the same reaction as in Figure 3A. At present, each Ap4N is investigated in isolation. To determine the effect of physiological Ap4N levels, reactions with all 4 Ap4Ns at maximum physiological levels as determined from NUDT2 KO cells would provide a robust test of their activities.

    Overall, the in vivo data suggest that Ap4A and Ap4G can reduce cGAS activity, but only weakly and only at levels that are unachievable even where NUDT2 is inactivated. The data suggest that Ap4A/Ap4G do not affect cGAS/STING/TBK1 signalling at physiological levels and that NUDT2 targeting would be ineffective for the pharmacological modulation of cGAS/STING/TBK1 signalling. This conclusion is accurate only if Ap4C/U do not contribute to this inhibition. At present in the absence of these data, this conclusion cannot be made, as their activity could in principle affect cGAS/STING signaling.

  4. eLife

    Reviewer #3 (Public review):

    Recent work reported that STING binds Ap4A and proposed that elevated Ap4A levels could modulate the cGAS-STING pathway. Intracellular Ap4A levels are controlled by the hydrolase NUDT2, itself a cancer drug target, and loss of NUDT2 raises cellular Ap4A pools. Building on this, the authors set out to investigate whether and how Ap4A modulates the cGAS-STING pathway. This question was motivated by the fact that the mechanism by which STING might bind a linear dinucleotide such as Ap4A remained unclear, and by the existence of alternative routes of crosstalk between Ap4A and the pathway. These include the chemical similarity between Ap4Ns and pppGpA, the linear intermediate generated by cGAS during cGAMP synthesis, raising the possibility that Ap4A could inhibit cGAS directly, as well as a potential downstream effect on the kinase TBK1.

    Using in vitro biochemistry, the authors show that STING does not, in fact, bind Ap4A, contrary to what was previously claimed, and they further examine the impact of NUDT2 loss on the cGAS-STING pathway. This clarifies for the field whether Ap4A could influence the pathway in any way, whether through binding to cGAS, STING, or TBK1.

    The authors find that cGAS can bind Ap4A but cannot process it. cGAS also binds Ap4G, but processes it only inefficiently and only when supplied at very high concentrations in vitro. This argues against physiological relevance in cells, since inhibition of canonical cGAMP synthesis would require Ap4Ns to be present in excess over ATP and GTP, a condition unlikely to be met at physiological nucleotide concentrations even when NUDT2 is absent. The authors also test for effects on TBK1, showing that while it can bind Ap4A, the concentrations required for inhibition are similarly not physiological.

    The authors conduct a thorough series of biochemical and biophysical experiments to clarify the underlying mechanism, systematically ruling out a plausible effect of Ap4A in cells at each step. Even where an effect on cGAS or TBK1 can be detected at very high, non-physiological concentrations, this remains confined to in vitro conditions. The experiments are sound and conducted with scientific rigor, and the results support the authors' conclusions.

    This work provides clear evidence that STING does not bind Ap4Ns, and that even where binding does occur elsewhere in the pathway, there is no plausible mechanism by which Ap4Ns could modulate the cGAS-STING pathway under physiological conditions. This resolves an open question for the field: if Ap4Ns do have biological effects related to this pathway, these are unlikely to arise from a direct interaction with cGAS, STING, or TBK1, and must instead be mediated through other processes. The methods and data will be a useful reference for the community when evaluating future claims of nucleotide second-messenger crosstalk.

    Prior studies have shown that NUDT2 disruption, which elevates cellular Ap4A levels, is associated with down-regulation of interferon-response pathways. This is notable in light of the current findings, which show no direct biochemical interaction between Ap4A and cGAS, STING, or TBK1. Together, this suggests that if elevated Ap4A does suppress IFN production, it likely does so indirectly, through mechanisms other than direct engagement of the cGAS-STING-TBK1 axis.

  5. Version published to 10.7554/elife.112828.1 on eLife
  6. Version published to 10.7554/elife.112828 on eLife
  7. Mentioned by @rridrobot.bsky.social on Bluesky

    RRIDs were included in this in bioRxiv paper. We value the author's support of reproducibility. #reproducibility #STMpublishing #methodsmatter

  8. Mentioned by @biorxivpreprint.bsky.social on Bluesky

    NUDT2 loss defines quantitative limits for dinucleoside polyphosphate action on the cGAS-STING-TBK1 axis https://www.biorxiv.org/content/10.64898/2026.08.18.745545v1

  9. Mentioned by @biorxiv-biochem.bsky.social on Bluesky

    NUDT2 loss defines quantitative limits for dinucleoside polyphosphate action on the cGAS-STING-TBK1 axis https://www.biorxiv.org/content/10.64898/2026.08.18.745545v1

  10. Mentioned by @biorxivpreprint.bsky.social on Bluesky

    NUDT2 loss defines quantitative limits for dinucleoside polyphosphate action on the cGAS-STING-TBK1 axis https://www.biorxiv.org/content/10.64898/2026.08.18.745545v1

  11. Version published to 10.64898/2026.08.18.745545v1 on bioRxiv