Lysine-11 ubiquitination drives type-I/III interferon induction by cGAS–STING and Toll-like receptors 3 and 4

TL;DR

The study reveals how immune receptors activate interferon production through a new ubiquitination mechanism involving ANKIB1 and Optineurin.

Contribution

Identifies ANKIB1-mediated lysine-11 ubiquitination as a novel mechanism for TBK1 activation in immune signaling.

Findings

ANKIB1 attaches K11-linked ubiquitin chains to TLR3 and cGAS–STING signalosomes.

Optineurin recruitment via K11 ubiquitination activates TBK1 and IRF3.

ANKIB1 deficiency reduces IFN induction and protection against HSV-1 in mice.

Abstract

Pattern recognition receptor (PRR)-induced interferon (IFN) is critical for effective immunity. The PRRs Toll-like receptor (TLR) 3, TLR4 and cyclic GMP–AMP synthase (cGAS), together with the stimulator of IFN genes (STING), signal through TANK-binding kinase 1 (TBK1), which activates the type-I/III IFN-inducing transcription factor interferon-response factor 3 (IRF3). The mechanism by which these PRRs activate TBK1 remains unresolved. Here we show that lysine-11 (K11)-linked ubiquitination drives TBK1 activation by these PRRs. The E3 ligase ANKIB1 attaches K11-linked ubiquitin chains to components of the TLR3- and cGAS–STING-induced signalosomes. This facilitates Optineurin recruitment to these complexes, in turn enabling recruitment and activation of TBK1 and IRF3, defining an uncharacterized signalling axis. In mice, ANKIB1 deficiency dampens IFN induction via TLR3 and cGAS–STING, reducing interferonopathy and compromising protection against HSV-1, respectively. Together, our results demonstrate an unanticipated and critical role for ANKIB1-generated K11-linked ubiquitination in the immune response activated by cGAS–STING, TLR3 and TLR4.

Betrancourt, Cinko et al. identify ANKIB1-mediated K11 ubiquitination and OPTN recruitment as a shared mechanism for driving immune signalling by cGAS, TLR3 and TLR4, with functional relevance in interferonopathy and herpes simplex virus-1 protection.

Figures

TLR3-mediated TBK1 and IRF3 activation requires ANKIB1, a member of the TLR3-SC, to promote type-I and type-III IFN induction.a,b, HeLa WT, HOIP-deficient (HOIP^*−*/*−*^) and HOIP^*−*/*−*^ cells reconstituted with enzymatically inactive HOIP^C885S^ (a) or ANKIB1-deficient (ANK^*−*/*−*^) (b) cells stimulated with either 10 µg ml^−1^ poly(I:C) or 100 ng ml^−1^ TNF for the indicated times and the lysates were analysed by western blotting. c, HT-29 ANK^*−*/*−*^ and WT cells were stimulated with 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were analysed by immunoblotting. d,e, IFN

The RBR and UIM domains of ANKIB1 are crucial for TBK1 phosphorylation.a, moTAP-ANKIB1 purified from HeLa cells was incubated with ubiquitin, E1, E2 UBE2D3 or UBE2L3 in ubiquitin buffer for 1 h at 37 °C and samples were analysed by western blotting. b, HeLa ANK^*−*/*−*^ cells reconstituted (rec.) with either moTAP-ANKIB1 WT (ANK WT) or ΔRBR (ANK ΔRBR) mutant were stimulated with 10 µg ml^−1^ poly(I:C) as indicated. c, HeLa cells stably expressing either ANK WT or ANK ΔRBR were stimulated with 15 µg ml^−1^ poly(I:C) as indicated and ANKIB1 was pulled down via its Flag-tag. d, HeLa ANK^*−*/*−*^

Optineurin is the crucial adaptor protein for TBK1 activation and requires K11-Ub generated by ANKIB1 for TLR3-SC recruitment.a, HeLa WT cells and cells deficient for TANK/NAP1/SINTBAD (TKO) were stimulated with either 100 ng ml^−1^ TNF or 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were subjected to western blot analysis. b, HeLa WT cells and cells deficient for TANK/NAP1/SINTBAD/OPTN (QKO) were stimulated with 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were subjected to western blot analysis c, HeLa WT, Optineurin-deficient (OPTN^*−*/*−*^) and ANK^*−*/*

TLR3-mediated TBK1 and IRF3 activation requires ANKIB1, a member of the TLR3-SC, to promote type-I and type-III IFN induction.a,b, HeLa WT, HOIP-deficient (HOIP^*−*/*−*^) and HOIP^*−*/*−*^ cells reconstituted with enzymatically inactive HOIP^C885S^ (a) or ANKIB1-deficient (ANK^*−*/*−*^) (b) cells stimulated with either 10 µg ml^−1^ poly(I:C) or 100 ng ml^−1^ TNF for the indicated times and the lysates were analysed by western blotting. c, HT-29 ANK^*−*/*−*^ and WT cells were stimulated with 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were analysed by immunoblotting. d,e, IFN

The RBR and UIM domains of ANKIB1 are crucial for TBK1 phosphorylation.a, moTAP-ANKIB1 purified from HeLa cells was incubated with ubiquitin, E1, E2 UBE2D3 or UBE2L3 in ubiquitin buffer for 1 h at 37 °C and samples were analysed by western blotting. b, HeLa ANK^*−*/*−*^ cells reconstituted (rec.) with either moTAP-ANKIB1 WT (ANK WT) or ΔRBR (ANK ΔRBR) mutant were stimulated with 10 µg ml^−1^ poly(I:C) as indicated. c, HeLa cells stably expressing either ANK WT or ANK ΔRBR were stimulated with 15 µg ml^−1^ poly(I:C) as indicated and ANKIB1 was pulled down via its Flag-tag. d, HeLa ANK^*−*/*−*^

Optineurin is the crucial adaptor protein for TBK1 activation and requires K11-Ub generated by ANKIB1 for TLR3-SC recruitment.a, HeLa WT cells and cells deficient for TANK/NAP1/SINTBAD (TKO) were stimulated with either 100 ng ml^−1^ TNF or 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were subjected to western blot analysis. b, HeLa WT cells and cells deficient for TANK/NAP1/SINTBAD/OPTN (QKO) were stimulated with 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were subjected to western blot analysis c, HeLa WT, Optineurin-deficient (OPTN^*−*/*−*^) and ANK^*−*/*

Type-I/III IFN induction by TLR4 and cGAS–STING, but not RIG-I, depends on ANKIB1.a, HT-29 WT and ANK^*−*/*−*^ cells were treated with 10 µg ml^−1^ LPS as indicated and subjected to western blot analysis. b,c, IFNβ (b) and IFNλ (c) mRNAs were measured by RT–qPCR 6 h after stimulation of HT-29 with 10 µg ml^−1^ LPS. *P* = 0.0043 and *P* = 0.0051, respectively. d, HeLa WT and ANK^*−*/*−*^ cells were treated with 10 µg ml^−1^ 2′3′cGAMP as indicated and subjected to western blot analysis. e,f, IFNβ (e) and IFNλ (f) mRNAs were respectively measured by RT–qPCR 6 h after stimulation of HT-29 with 10

ANKIB1 promotes ubiquitination of TRIF, STING, NEMO, OPTN and itself.a, HeLa WT, ANK^*−*/*−*^ and ANK RING1^mut^ cells were stimulated with 15 µg ml^−1^ poly(I:C) for the indicated times. The ubiquitinated proteins were pulled down using GST-TUBE pre-coupled with glutathione beads. Co-precipitated proteins were analysed by western blot. b, HeLa WT, ANK^*−*/*−*^ and ANK RING1^mut^ cells were stimulated with 10 µg ml^−1^ 2′3′cGAMP for the indicated times. The ubiquitinated proteins were pulled down using GST-TUBE pre-coupled with glutathione beads. Co-precipitated proteins were analysed by weste

ANKIB1 is crucial for TLR3- and cGAS–STING-mediated IFN induction in vivo*.*a–d, Control (ctrl) and *Ankib1*^*−/−*^ BMDMs were stimulated with 5 µg ml^−1^ poly(I:C) for 24 h (a), 200 ng ml^−1^ LPS for 4 h (b), 5 µg ml^−1^ ADU-S100 for 6 h (c) or 200 ng ml^−1^ 3p-hpRNA for 6 h (d) and the supernatants were collected to determine IFNβ concentration by ELISA. The graphs represent mean ± s.e.m. of three or four independent biological replicates. *P* values were determined by two-way ANOVA with uncorrected Fisher’s LSD multiple-comparisons tests with 95% confidence intervals. *P* < 0.0001, *P* < 0.

Full text

Main

Interferon (IFN) production is a key defence mechanism of innate immunity. Different pattern recognition receptors (PRRs) detect distinct pathogen- and damage-associated molecular patterns during infection and during tissue damage, rapidly mobilizing innate defences^1–3^. Toll-like receptors (TLRs) are archetypal and well-characterized PRRs^4^. The triggering of two of them, TLR3 and TLR4, results in the activation of TANK-binding kinase 1 (TBK1) in a biochemical context capable of directly activating IRF3, the transcription factor responsible for the induction of type-I and type-III IFNs^5^. This unique biochemical feature of the double-stranded (ds)RNA-sensing TLR3 and the lipopolysaccharide (LPS)-sensing TLR4 is shared with a small subset of other PRRs, most notably the cytosolic DNA sensor cyclic GMP–AMP synthase (cGAS) and the two dsRNA-sensing RNA helicases retinoic acid-inducible gene I (RIG-I) and melanoma differentiation-associated protein 5 (MDA-5)^6–9^. However, despite its critical role in innate immunity, the biochemical mechanism by which activation of these PRRs confers this capacity on TBK1 remains unknown.

TBK1 is also involved in other cellular processes, including inhibition of tumour necrosis factor (TNF)-induced cell death via RIPK1 phosphorylation^10^. We previously showed that, by enabling the recruitment of various kinases, linear ubiquitin (linUb) chains generated by HOIP, a component of the linUb chain assembly complex (LUBAC), are responsible for full gene activation and cell-death resistance in TNFR1 and TLR3 signalling^11–13^. While TBK1 is one such kinase in TNF signalling^10^, in TLR3 signalling LUBAC-generated linUb is dispensable for TBK1 activation^13^.

This unexpected observation, together with the fact that all proposed TBK1-recruiting adaptors in TLR3 signalling bind to polyubiquitin, including linUb^10,14^, led us to hypothesize that another, possibly linUb-generating and therefore probably HOIP-related E3 ligase may be required for TLR3-induced TBK1 activation. On the basis of this, we next screened the HOIP-comprising RING-between-RING (RBR)-domain-containing E3 subfamily for the elusive E3 ligase.

This resulted in the discovery of ANKIB1 as the E3 ligase responsible for TBK1 activation downstream of TLR3. Yet ANKIB1 does not generate linUb chains but instead attaches K11-linked ubiquitin (K11-Ub) chains to various components of the TLR3 signalosome, thereby enabling Optineurin (OPTN) recruitment and initiating an ANKIB1–K11-Ub–OPTN–TBK1–IRF3 signalling axis that bridges TLR3 stimulation to IRF3 activation and IFN induction. Remarkably, this signalling axis also drives activation of IRF3 following stimulation of TLR4 and cGAS, together with the stimulator of IFN genes (STING). In vivo, ANKIB1 deficiency limits TLR3-induced IFN induction and, consequently, reduces lethality in a TLR3-driven interferonopathy model, whereas it increases vulnerability of mice to herpes simplex virus-1 (HSV-1) infection, known to depend on cGAS–STING-mediated IFN induction for survival, providing compelling evidence that ANKIB1-mediated K11 ubiquitination is crucial for innate immune signalling in vivo.

Results

TLR3-mediated TBK1 and IRF3 activation requires ANKIB1, an unrecognized component of the TLR3-SC, to promote type-I and type-III IFN induction

After confirming that HOIP and its linUb-generating activity were dispensable for TBK1 activation downstream of TLR3, but not TNFR1 (Fig. 1a), we screened the RBR-E3 subfamily for the E3 ligase responsible for TLR3-induced TBK1 activation. Candidate RBR-E3s we could exclude a priori were HOIP, HOIL-1^15^ and PARKIN because it is not expressed by HeLa cells^16^. Individual RNA-interference (RNAi)-mediated silencing of the remaining 11 RBR-E3 members revealed that only suppression of ANKIB1 substantially reduced TLR3-induced TBK1 activation, but not p65 phosphorylation (Extended Data Fig. 1a–c), implying a specific role for ANKIB1 in activation of TBK1 and not NF-κB.Fig. 1TLR3-mediated TBK1 and IRF3 activation requires ANKIB1, a member of the TLR3-SC, to promote type-I and type-III IFN induction.a,b, HeLa WT, HOIP-deficient (HOIP^−/−^) and HOIP^−/−^ cells reconstituted with enzymatically inactive HOIP^C885S^ (a) or ANKIB1-deficient (ANK^−/−^) (b) cells stimulated with either 10 µg ml^−1^ poly(I:C) or 100 ng ml^−1^ TNF for the indicated times and the lysates were analysed by western blotting. c, HT-29 ANK^−/−^ and WT cells were stimulated with 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were analysed by immunoblotting. d,e, IFNβ (d) and IFNλ (e) mRNAs levels were assessed by RT–qPCR 6 h after stimulation of HT-29 with 10 µg ml^−1^ poly(I:C). P = 0.0199 and P < 0.0001, respectively. f, HeLa ANK^−/−^ cells were reconstituted (rec.) with either moTAP-ANKIB1 WT (ANK WT) or empty vector (empty vec.) and treated with 10 µg ml^−1^ poly(I:C) for the indicated times. Lysates were analysed by western blotting. g, Total RNA was isolated for IFNβ transcripts analysis by RT–qPCR after the stimulation of cells from f for the respective time points. P = 0.4470, P = 0.3506, P < 0.0001, P = 0.0018 and P = 0.0099, respectively. h, RNA-sequencing analysis of HeLa cells reconstituted with either ANK WT or empty vector was performed after 10 µg ml^−1^ poly(I:C) treatment for 2 h (n = 2). i, HeLa cells stably expressing ANK WT were stimulated with 15 µg ml^−1^ poly(I:C) and ANKIB1 was pulled down via its Flag-tag. j, HeLa TLR3^−/−^ cells reconstituted with moTAP-TLR3 were treated with 15 µg ml^−1^ poly(I:C) as indicated. moTAP-TLR3 was pulled down through its Flag-tag. FL, full length. All graphs represent mean ± s.e.m. of three independent experiments. P values were calculated by two-way ANOVA with uncorrected Fisher’s LSD multiple-comparisons tests with 95% confidence intervals. ^∗^P ≤ 0.05, ^∗∗^P ≤ 0.01 and ^∗∗∗^P ≤ 0.001. For western blots, one representative of at least three independent experiments is displayed. Non-specific bands are marked with an asterisk.Source data

To test whether ANKIB1 could indeed be the hitherto elusive E3 ligase required for TLR3-induced TBK1 activation, we next generated ANKIB1-knockout (KO) cell lines and analysed their response to TNF and polyinosinic:polycytidylic acid (poly(I:C)) stimulation in comparison to HOIP-KO and wild-type (WT) cells. This demonstrated that TLR3-induced TBK1 activation was virtually abolished in ANKIB1-KO cells, whereas TNF-induced TBK1 activation remained intact (Fig. 1b). Conversely, HOIP-KO impaired TNF- but not poly(I:C)-induced TBK1 activation (Fig. 1b). Extended kinetic analyses across different KO and WT cell lines confirmed that ANKIB1-KO did not merely delay but profoundly impaired TLR3- but not TNF-induced TBK1 phosphorylation (Extended Data Fig. 2a–c). Thus, TBK1 phosphorylation is enabled by two distinct RBR-E3 ligases in different signalling pathways; whereas HOIP facilitates TNF-induced TBK1 activation, ANKIB1 is required for TLR3-induced TBK1 phosphorylation.

Assessing the role of ANKIB1 in TLR3-mediated IRF3 activation and IFN induction^17,18^ revealed that TLR3-induced IRF3 phosphorylation and the induction of both β (type-I) and λ (type-III) IFN gene expression almost completely depended on ANKIB1 (Fig. 1c–e). Stably re-expressing modified tandem affinity purification-tagged (moTAP)-ANKIB1 at near endogenous levels in ANKIB1-KO cells fully restored TLR3-induced TBK1 and IRF3 phosphorylation (Fig. 1f and Extended Data Fig. 2d) and type-I IFN production (Fig. 1g), underscoring the crucial role of ANKIB1 in this pathway. RNA-sequencing analysis demonstrated that ANKIB1 is required for TLR3-mediated induction of type-I/III IFN genes and downstream IFN-stimulated genes. (Fig. 1h).

To epistatically place ANKIB1 in TLR3 signalling, we performed immunoprecipitation (IP) of moTAP-ANKIB1, which corroborated the TLR3-induced interaction of ANKIB1 with phosphorylated TBK1 (p-TBK1; Fig. 1i). To determine whether this interaction occurs at the TLR3 signalling complex (TLR3-SC) level, we generated TLR3-deficient HeLa cells expressing ectopic moTAP-TLR3. Apart from confirming recruitment of the known TLR3-SC components RIPK1 and HOIP^13,19^, immunoprecipitation of moTAP-TLR3 also revealed ANKIB1 as forming part of the TLR3-SC but not the TNFR1-SC (Fig. 1j and Extended Data Fig. 2e). Thus, ANKIB1 constitutes a previously unrecognized component of the TLR3-SC that interacts with TBK1 upon receptor activation and is required for TLR3-induced activation of TBK1 and IRF3 and induction of type-I/III IFNs.

The RBR and UIM domains of ANKIB1 are crucial for TBK1 phosphorylation

Although ANKIB1 is a member of the RBR-E3 subfamily^20^, its catalytic activity had not previously been demonstrated. In an in vitro ubiquitination assay, full-length recombinant ANKIB1 was capable of generating ubiquitin chains in the presence of the E2s UBE2L3 and UBE2D3, with higher activity observed with UBE2D3 (Fig. 2a), yet not with UBE2L6 (Extended Data Fig. 3a). The multimeric proteins formed by ANKIB1/UBE2D3 were degraded by the general deubiquitinase (DUB) USP2, confirming they were composed of ubiquitin polymers (Extended Data Fig. 3b).Fig. 2. The RBR and UIM domains of ANKIB1 are crucial for TBK1 phosphorylation.a, moTAP-ANKIB1 purified from HeLa cells was incubated with ubiquitin, E1, E2 UBE2D3 or UBE2L3 in ubiquitin buffer for 1 h at 37 °C and samples were analysed by western blotting. b, HeLa ANK^−/−^ cells reconstituted (rec.) with either moTAP-ANKIB1 WT (ANK WT) or ΔRBR (ANK ΔRBR) mutant were stimulated with 10 µg ml^−1^ poly(I:C) as indicated. c, HeLa cells stably expressing either ANK WT or ANK ΔRBR were stimulated with 15 µg ml^−1^ poly(I:C) as indicated and ANKIB1 was pulled down via its Flag-tag. d, HeLa ANK^−/−^ cells reconstituted with either ANK WT or RING1-mutant C333A/ I335A/ C336A/ C351A/ H353A (ANK RING1^mut^) were stimulated with 10 µg ml^−1^ poly(I:C) as indicated. The downstream pathways were analysed by immunoblotting. e, After performing an in vitro ubiquitination assay of ANKIB1 in combination with either UBE2D3 or UBE2L3, the samples were digested with trypsin overnight and subjected to LC–MS/MS analysis. Left: the intensity induced upon in vitro ubiquitin assay. Representative figure from two independent experiment. Right: pie charts representing the intensity average of two independent experiments of the different ubiquitin linkages generated by ANKIB1 in combination with either UBE2D3 or UBE2L3. f, moTAP-ANKIB1 purified from HeLa cells was incubated with ubiquitin, E1, E2 UBE2D3 in ATP regeneration solution for 1 h at 37 °C followed by digestion of the sample with deubiquitinase Cezanne for 1 h at 30 °C. g, HeLa cells stably expressing either ANK WT or moTAP-ANKIB1 ΔUIM (ANK ΔUIM) were stimulated with 15 µg ml^−1^ poly(I:C) and ANKIB1 was pulled down via its Flag-tag. h, HT-29 cells reconstituted (rec.) with either moTAP-ANKIB1 WT or empty vector (empty vec.) were stimulated with 15 µg ml^−1^ poly(I:C) and ANKIB1 was pulled down via its Flag-tag. The samples were subjected to analysis by western blotting. For all western blots, one representative of at least three independent experiments is displayed.Source data

To assess whether the E3 ligase activity of ANKIB1 is required for TBK1 activation, we employed a similar approach to the previously studied RBR-E3 ligases, all known to exert their catalytic activity via their respective RING2^21^. We therefore next generated ANKIB1 with mutations in the conserved cysteine residues (C519S/C522S/C532S/C537S/C540S/C545S/C548S) of its RING2 (ANKIB1-RING2^mut^) (Extended Data Fig. 3c). Unexpectedly, ANKIB1-RING2^mut^ expression in ANKIB1-deficient cells restored TLR3-induced TBK1 and IRF3 phosphorylation (Extended Data Fig. 3d). As we suspected that the point mutations we introduced in the RING2 of ANKIB1 might not have aborted its activity to the same extent as was seen in previous studies for other RBR-E3s^22–24^, we next expressed an ANKIB1 mutant lacking the entire RING2 domain (ANKIB1∆RING2) (Extended Data Fig. 3c), yet TBK1 and IRF3 activation were again restored to levels comparable to those observed in wt cells upon TLR3 stimulation (Extended Data Fig. 3e). To test whether ANKIB1∆RING2 bears residual catalytic activity, we performed in vitro ubiquitination assays. Despite lacking a RING2, this mutant RBR-E3 retained ubiquitin chain-forming activity with the more general E2 UBE2D3 but not with UBE2L3 (Extended Data Fig. 3f).

To fully abrogate any possible RBR-residing catalytic activity of ANKIB1, we next expressed mutant ANKIB1 lacking the entire RBR (ANKIB1∆RBR) in ANKIB1-KO cells (Extended Data Fig. 3c). This mutant failed to rescue TBK1 and IRF3 phosphorylation upon TLR3 activation, indicating that the catalytic activity of ANKIB1 indeed resides in its RBR and that the polyubiquitin chains it generates are essential for TBK1 activation (Fig. 2b). We next investigated whether the RBR domain is required for the recruitment of ANKIB1 to the TLR3-SC and/or its interaction with TBK1. Pull down (PD) revealed that the interaction of ANKIB1 with p-TBK1 was drastically reduced in absence of its RBR (Fig. 2c). By contrast, full-length ANKIB1 and ANKIB1∆RBR exhibited comparable interaction with TRAF3, a known component of the TLR3-SC^25–27^. Thus, the RBR of ANKIB1 is required for its TLR3-induced interaction with TBK1 and TBK1 activation but not for TLR3-SC recruitment.

Since the ANKIB1-RBR but not its RING2 was required, we next generated ANKIB1-deficient cells expressing a RING1-mutant ANKIB1 (ANKIB1-RING1^mut^) with mutations in several conserved residues (C333A/I335A/C336A/C351A/H353A) and, indeed, TLR3 stimulation failed to activate TBK1 in these cells (Fig. 2d). Hence, the RING1 of ANKIB1 is required and sufficient for its catalytic function and essential for promoting TBK1 activation.

Using mass spectrometry, we next determined the ubiquitin linkage type(s) ANKIB1 generates in the presence of UBE2D3 or UBE2L3. Surprisingly, K11-Ub was the most abundant linkage type formed by ANKIB1 in these in vitro assays with K63-Ub and K6-Ub linkages also detectable, albeit at substantially lower levels (Fig. 2e). No other linkage types, including linUb (M1-Ub) and K48-Ub, were detected. Performing a DUB assay using Cezanne (OTUD7B)^28^, a K11-Ub-specific DUB, resulted in the disappearance of the vast majority of ANKIB1-generated polyubiquitin chains (Fig. 2f). Thus, ANKIB1 preferentially generates K11-Ub chains and its catalytic activity facilitates TLR3-induced TBK1 activation, suggesting that ANKIB1-generated K11-Ub may be crucial for TBK1 recruitment and activation upon TLR3 stimulation.

Binding to ubiquitin or polyubiquitin chains is a well-established mechanism for recruiting proteins with the required specificity to signalling complexes^29,30^. As the ANKIB1-RBR was not required for TLR3-SC recruitment, we next examined the role of ANKIB1’s ubiquitin-interacting motif (UIM), a domain type previously reported to mediate polyubiquitin-chain binding (Extended Data Fig. 3c)^31^. Following poly(I:C) treatment, while wt-ANKIB1 interacted with known TLR3-SC components, including TRAF3 and HOIP, UIM-deficient ANKIB1 (ANKIB1∆UIM) failed to associate with these components, implying this domain is essential for TLR3-SC recruitment of ANKIB1 (Fig. 2g).

As TBK1 is recruited to different immune receptor signalling complexes via distinct ubiquitin-binding adaptor proteins, we next sought to identify which ANKIB1-interacting protein(s) might function as TBK1 recruiter(s) in the TLR3 pathway. OPTN was previously reported as adaptor for TBK1 in TLR3 and TLR4, but not in TNFR1 signalling^10,32^, whereas NAP1 was previously proposed to serve as adaptor for TBK1 in TLR3 and TNFR1 signalling^10,33^. To determine possible adaptor interactions, we performed ANKIB1 PD upon TLR3 stimulation followed by western blotting for these the different possible adaptors. This identified a selective interaction between ANKIB1 and OPTN but not NAP1 (Fig. 2g,h). Hence, ANKIB1 is recruited to the TLR3-SC via its UIM domain and, once recruited, utilizes its RBR domain to mediate TBK1 activation, probably via OPTN as adaptor.

Optineurin requires ANKIB1-generated K11-Ub for recruitment to the TLR3-SC and serves as essential adaptor for TBK1

Recruitment of OPTN, along with recruitment and activation of TBK1, correlates with the presence of ANKIB1 at the TLR3-SC (Fig. 2g,h). However, other proteins, including TANK, NAP1 and SINTBAD, have also been proposed as adaptors for TBK1 in TLR3 signalling^34–36^. To assess the contribution of these adaptors to TBK1 activation by TLR3, we next investigated poly(I:C)-induced TBK1 activation in triple knockout (TKO) cells lacking these three adaptors. Remarkably, TLR3-induced activation of TBK1 and IRF3 was unaffected in these cells whereas, consistent with our prior observations^10^, TNF-induced TBK1 activation was completely lost in the absence of these three proteins (Fig. 3a). Owing to the association between OPTN and ANKIB1 at the TLR3-SC and prior evidence implicating OPTN as TBK1 adaptor in TLR3 signalling^32,37,38^, we next generated quadruple knockout (QKO) cells lacking TANK, NAP1, SINTBAD and OPTN. In these QKO cells, TBK1 phosphorylation was completely abolished (Fig. 3b) upon poly(I:C) stimulation, indicating a critical role for OPTN as an adaptor in TBK1 recruitment. To test whether OPTN alone is sufficient and essential for TBK1 activation, we next generated OPTN single-KO cells. Strikingly, loss of OPTN alone recapitulated the phenotype of ANKIB1-deficient or QKO cells, with complete impairment of TBK1 and IRF3 activation in response to stimulation of TLR3 but not TNFR1 (Fig. 3c). Together, these results establish OPTN as the essential adaptor for TBK1 activation in TLR3 signalling.Fig. 3. Optineurin is the crucial adaptor protein for TBK1 activation and requires K11-Ub generated by ANKIB1 for TLR3-SC recruitment.a, HeLa WT cells and cells deficient for TANK/NAP1/SINTBAD (TKO) were stimulated with either 100 ng ml^−1^ TNF or 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were subjected to western blot analysis. b, HeLa WT cells and cells deficient for TANK/NAP1/SINTBAD/OPTN (QKO) were stimulated with 10 µg ml^−1^ poly(I:C) for the indicated times and the lysates were subjected to western blot analysis c, HeLa WT, Optineurin-deficient (OPTN^−/−^) and ANK^−/−^ cells were stimulated with either 100 ng ml^−1^ TNF or 10 µg ml^−1^ poly(I:C) for the indicated times. The lysates were analysed by immunoblotting. d, HeLa ANKIB1-proficient and deficient moTAP-TLR3 cells were stimulated with 15 µg ml^−1^ poly(I:C). moTAP-TLR3 was pulled down via its Flag-tag and the analysis of co-immunoprecipitated proteins was determined by immunoblotting. FL, full length. e, HeLa moTAP-TLR3 cells were stimulated with 15 µg ml^−1^ poly(I:C) for the indicated time points. moTAP-TLR3 was pulled down via its Flag-tag. After washing, the pulled-down samples were incubated with or without Cezanne (Cez) in DUB buffer for 1 h at 30 °C. Co-immunoprecipitated proteins and the resultant supernatant were subjected to western blotting analysis. One representative of at least three independent experiments is displayed. Non-specific bands are marked with an asterisk.Source data

Since the catalytic activity of ANKIB1 is required to promote TLR3-induced TBK1 activation (Fig. 2d) and because previous studies have implicated the UBAN domain of OPTN in TBK1 recruitment during TLR3 signalling^32,39^, we next tested whether ANKIB1 and its K11-Ub-generating activity are necessary for OPTN and TBK1 recruitment. A comparison of TLR3-SC formation in WT versus ANKIB1-KO cells revealed markedly reduced recruitment of p-TBK1 and OPTN in ANKIB1 deficiency (Fig. 3d).

To investigate whether its K11-Ub-generating capacity is also required, we purified the TLR3-SC, treated it with the K11-specific DUB Cezanne and analysed the resulting supernatant for proteins released from the complex. Strikingly, Cezanne-mediated hydrolysis of K11-Ub chains caused specific release of OPTN, but not NEMO, from the TLR3-SC (Fig. 3e). Hence, ANKIB1 and its K11-Ub-generating activity are essential for the recruitment of OPTN, which is, in turn, required for TBK1 activation upon TLR3 stimulation.

In summary, we define a signalling axis in which ANKIB1 is recruited to the TLR3-SC, where it catalyses the formation of K11-linked polyubiquitin chains, facilitating the recruitment of OPTN which, in turn, mediates the recruitment and activation of TBK1. Importantly, this previously unrecognized ANKIB1–K11-Ub–OPTN–TBK1 signalling axis is essential for TLR3-induced IRF3 activation and IFN production.

IFN induction by TLR4 and cGAS–STING is also driven by ANKIB1

We next investigated whether ANKIB1 plays a role in other PRRs known to induce type-I IFNs through TBK1 and IRF3. We first focused on the LPS-sensing TLR4, the only TLR capable of signalling through two pathways via distinct adaptors, a TRIF-dependent one that is similar to TLR3 signalling and an MyD88-dependent one. Importantly, only TRIF—but not MyD88— contains an IRF3-binding motif, which is essential for its recruitment, activation and subsequent induction of IFN genes^9,25,40^. In ANKIB1-KO cells, LPS-induced IRF3 phosphorylation and type-I/III IFN induction were substantially reduced in comparison with WT cells. TBK1 phosphorylation, however, was only partially decreased (Fig. 4a–c). Thus, similar to TLR3 signalling, the TRIF-dependent arm of TLR4 signalling requires ANKIB1 for effective TBK1 activation and downstream IRF3-dependent IFN induction. By contrast, MyD88-dependent signalling can also activate TBK1 but does not require ANKIB1^40^.Fig. 4. Type-I/III IFN induction by TLR4 and cGAS–STING, but not RIG-I, depends on ANKIB1.a, HT-29 WT and ANK^−/−^ cells were treated with 10 µg ml^−1^ LPS as indicated and subjected to western blot analysis. b,c, IFNβ (b) and IFNλ (c) mRNAs were measured by RT–qPCR 6 h after stimulation of HT-29 with 10 µg ml^−1^ LPS. P = 0.0043 and P = 0.0051, respectively. d, HeLa WT and ANK^−/−^ cells were treated with 10 µg ml^−1^ 2′3′cGAMP as indicated and subjected to western blot analysis. e,f, IFNβ (e) and IFNλ (f) mRNAs were respectively measured by RT–qPCR 6 h after stimulation of HT-29 with 10 µg ml^−1^ ADU-S100. P < 0.0001 and P = 0.0108, respectively. g, HeLa WT and ANK^−/−^ cells were infected with attenuated MVA virus for the indicated time points before IFNβ transcripts were measured by RT–qPCR. P = 0.0004 and P < 0.0001, respectively. hpi, hours post infection. h, HeLa cells stably expressing ANK WT or ANK RING1^mut^ were stimulated with 10 µg ml^−1^ ADU-S100 and the lysates were analysed by immunoblotting. i, HeLa cells stably expressing either ANK WT or ANK ΔUIM treated with 10 µg ml^−1^ 2′3′cGAMP. moTAP-ANKIB1 was pulled down via its Flag-tag. Immunoprecipitated proteins were analysed by immunoblotting. j, A549 WT and ANK^−/−^ cells were treated with 500 ng ml^−1^ 3p-hpRNA for the indicated times and subjected to western blot analysis. k,l, IFNβ (k) and IFNλ (l) mRNAs were measured by RT–qPCR 6 h after stimulation of HeLa WT and ANK^−/−^ cells with 500 ng ml^−1^ 3p-hpRNA. P = 0.8597 and P = 0.9458, respectively. m, A549 WT and ANK^−/−^ cells were infected with SeV for the indicated times before IFNβ transcripts were measured by RT–qPCR. P = 0.4326 and P = 0.7257, respectively. n, HeLa WT, OPTN^−/−^ and ANK^−/−^ cells were stimulated with 10 µg ml^−1^ ADU-S100. The lysates were subjected to immunoblotting. o, HeLa WT and TKO cells were treated with 500 ng ml^−1^ 3p-hpRNA for the indicated times and the lysates were subjected to western blot analysis. All graphs represent mean ± s.e.m. of three independent experiments. P values were determined by two-way ANOVA with uncorrected Fisher’s LSD multiple-comparisons tests with 95% confidence interval. ^∗^P ≤ 0.05, ^∗∗^P ≤ 0.01 and ^∗∗∗^P ≤ 0.001. For western blots, one representative of three independent experiments is shown. Non-specific bands are marked with an asterisk.Source data

We next investigated the role of ANKIB1 in cGAS–STING signalling, which also relies on TBK1 and IRF3 for IFN induction. In ANKIB1-KO cells, stimulation with the STING activator 2′3′cyclic GMP–AMP (2′3′cGAMP) or the STING agonist ADU-S100 failed to induce robust phosphorylation of TBK1, IRF3 and STING (Fig. 4d and Extended Data Fig. 4a,b). Following STING activation, ANKIB1 PD revealed an interaction with activated STING, together with p-TBK1, suggesting that ANKIB1 is recruited to the STING-associated signalling complex (STING-SC) (Extended Data Fig. 4c). ANKIB1-deficient cells exhibited significantly reduced type-I/III IFN induction, not only following STING agonist treatment but also after infection with the cGAS-sensed poxvirus modified vaccinia virus Ankara (MVA) (Fig. 4e–g). Hence, ANKIB1 is also crucial for effective type-I/III IFN induction in TLR4 and cGAS–STING signalling.

Using cells expressing ANKIB1-RING1^mut^ and ANKIB1ΔUIM, we next tested the role of ANKIB1 catalytic activity and ubiquitin-binding capacity in cGAS–STING signalling. Both mutants failed to rescue TBK1 and IRF3 phosphorylation following STING stimulation, indicating that the E3 ligase activity and UIM of ANKIB1 are also essential for its function in cGAS–STING signalling (Fig. 4h and Extended Data Fig. 4d). Furthermore, ANKIB1 required its UIM for recruitment to the STING-SC, where it interacted with phosphorylated STING, TBK1 and OPTN (Fig. 4i).

By treating cells with the RIG-I ligand 3p-hpRNA, we next assessed whether the RIG-I helicase, which recognizes cytosolic dsRNA and activates IFN transcription through mitochondrial antiviral signalling protein (MAVS)-driven activation of TBK1 and IRF3^41^, may also require ANKIB1 for TBK1 activation. However, ANKIB1-KO cells exhibited TBK1 and IRF3 activation and type-I/III IFN induction similar to WT controls (Fig. 4j–l), regardless of the 3p-hpRNA concentration used (Extended Data Fig. 4e,f). As further confirmation, we infected cells with Sendai virus (SeV) or influenza A virus (IAV), both of which are primarily sensed by RIG-I^42^. Upon infection, WT and ANKIB1-KO cells showed similar levels of TBK1 and IRF3 phosphorylation and MAVS protein levels remained unchanged (Extended Data Fig. 4g). Moreover, IFNβ induction was similar in WT and ANKIB1-KO cells (Fig. 4m). These findings suggest that ANKIB1 may not have a prominent role for RIG-I-mediated activation of TBK1 and IRF3 and subsequent IFN production.

To explain the divergence in ANKIB1 requirement across different PRRs, we hypothesized that distinct PRRs may engage different adaptor proteins for TBK1 recruitment and activation. Supporting this, we found that, as in TLR3 signalling, the absence of OPTN strongly impaired STING-induced activation of TBK1 and IRF3, whereas co-deletion of TANK, NAP1 and SINTBAD did not (Fig. 4n and Extended Data Fig. 4h). Conversely, TBK1 and IRF3 phosphorylation were strongly impaired in TANK/NAP1/SINTBAD TKO but remained unchanged in OPTN-KO cells upon RIG-I activation (Fig. 4o and Extended Data Fig. 4i). Thus, ANKIB1 promotes IFN production downstream of PRRs that depend on OPTN for TBK1 and IRF3 activation, namely TLR3, TRIF-mediated TLR4, and cGAS–STING. Instead, activation of TBK1, IRF3 and IFN production induced by stimulation of RIG-I is primarily mediated via TANK, NAP1 and/or SINTBAD.

ANKIB1 promotes ubiquitination of TRIF, STING, NEMO, OPTN and itself

Having established that the catalytic activity of ANKIB1 is essential for TBK1 activation downstream of TLR3 and cGAS–STING, we aimed to identify its potential substrates within the TLR3- and STING-SCs. To this end, we performed total ubiquitin PDs using tandem ubiquitin-binding entities (TUBE) following poly(I:C) or 2′3′cGAMP treatment. In WT cells, the adaptor proteins TRIF and STING were strongly ubiquitinated upon stimulation, whereas their ubiquitination was considerably reduced in ANKIB1-KO and ANKIB1-RING1^mut^-expressing cells (Fig. 5a,b). In addition, ubiquitination of NEMO and OPTN was also impaired in the absence of ANKIB1 or its catalytic activity (Fig. 5a,b), showing that the activity of ANKIB1 facilitates their ubiquitination. Intriguingly, ANKIB1 itself appeared to be constitutively ubiquitinated and underwent additional ubiquitination upon poly(I:C) or STING agonist treatment, while these ubiquitination events were absent in ANKIB1-RING1^mut^-expressing cells, (Fig. 5a,b), implying constitutive and PRR stimulation-dependent ANKIB1 auto-ubiquitination.Fig. 5ANKIB1 promotes ubiquitination of TRIF, STING, NEMO, OPTN and itself.a, HeLa WT, ANK^−/−^ and ANK RING1^mut^ cells were stimulated with 15 µg ml^−1^ poly(I:C) for the indicated times. The ubiquitinated proteins were pulled down using GST-TUBE pre-coupled with glutathione beads. Co-precipitated proteins were analysed by western blot. b, HeLa WT, ANK^−/−^ and ANK RING1^mut^ cells were stimulated with 10 µg ml^−1^ 2′3′cGAMP for the indicated times. The ubiquitinated proteins were pulled down using GST-TUBE pre-coupled with glutathione beads. Co-precipitated proteins were analysed by western blot. c, HeLa WT cells were stimulated with 15 µg ml^−1^ poly(I:C) for 60 min. The ubiquitinated proteins were pulled down using GST-TUBE pre-coupled with glutathione beads and the beads were incubated with K48 tetra-Ub chains before performing a deubiquitination assay using either 125 nM AMSH, Cezanne (Cez.) or 250 nM USP2. The samples were analysed by western blot. d, HeLa WT, and ANK^−/−^ cells were stimulated with 15 µg ml^−1^ poly(I:C) for the indicated times. K11-linked polyubiquitinated targets were co-precipitated using a K11 2A3/2E6 antibody pre-coupled with protein G. Co-immunoprecipitated proteins were analysed by western blotting. IP, immunoprecipitated. For western blots, one representative of three independent experiments is displayed. Non-specific bands are marked with an asterisk.Source data

To identify the nature of the ubiquitin linkages involved, we performed TUBE PDs from poly(I:C)-treated cells followed by DUB assay using K63-specific AMSH, K11-specific Cezanne and non-linkage-specific USP2^28,43,44^. TRIF ubiquitination was sensitive to both AMSH and Cezanne, indicating the presence of both K63- and K11-Ub-linked chains (Fig. 5c). OPTN ubiquitination was predominantly reduced by AMSH and USP2, with a modest Cezanne effect, suggesting predominant K63-Ub of OPTN yet also the presence of K11-Ub. For NEMO, only the high-molecular-weight ubiquitin chains were cleaved by either DUB (Fig. 5c), suggesting the presence of complex, probably K63–K11-branched or mixed–linkage ubiquitin chains that resist complete processing by individual DUBs. By performing K11-Ub-specific PDs following TLR3 activation, TRIF was found to be K11-ubiquitinated in WT but not in ANKIB1-KO cells, providing further proof of TRIF as substrate of ANKIB1 (Fig. 5d). Together, these results identify TRIF, STING, NEMO, OPTN and ANKIB1 itself as targets of the K11-Ub chain-generating activity of ANKIB1. K11-Ub, placed on components of the TLR3- and STING-SCs enables OPTN recruitment, thereby promoting activation of TBK1, IRF3 and subsequent IFN induction.

ANKIB1 is crucial for IRF3-activating, IFN-inducing TBK1 activation in primary murine immune cells upon activation of TLR3/4 and cGAS–STING, but not RIG-I

To understand the relevance of ANKIB1 in primary cells and in vivo, we generated Ankib1^−/−^ mice. ANKIB1 deficiency was not embryonically lethal and did not result in any overt phenotype (Extended Data Fig. 5a). However, ANKIB1 is differentially expressed across tissues and particularly enriched in the spleen, lungs and, to a lesser extent, the colon and Peyer’s patches (Extended Data Fig. 5b). To assess whether ANKIB1 deletion altered basal immune homeostasis or inflammation, we evaluated spleen morphology and immune cell composition. Both spleen/body weight ratio and immune cell numbers were comparable between control and Ankib1^−/−^ mice (Extended Data Fig. 5c,d). Histological characterization of liver, spleen, lungs and intestine did not show any prominent alterations between Ankib1^−/−^ and control mice (Extended Data Fig. 5e).

We next assessed the role of ANKIB1 in innate immune signalling by examining primary murine immune cells ex vivo. Consistent with our findings in human cell lines, stimulation of TLR3 led to substantially decreased TBK1 phosphorylation in bone marrow-derived macrophages (BMDMs) from Ankib1^−/−^ compared with Ankib1^+/−^ controls (Extended Data Fig. 5f). In TLR4-stimulated splenocytes, TBK1 phosphorylation was preserved—probably via MyD88-dependent signalling—yet IRF3 phosphorylation was abolished and STAT1 phosphorylation, a downstream readout of type-I IFN signalling, was largely diminished in Ankib1^−/−^ cells (Extended Data Fig. 5g). Similarly, STING stimulation induced robust TBK1 activation and IFN production in Ankib1^+/−^ primary murine immune cells but was substantially attenuated in Ankib1^−/−^ cells (Extended Data Fig. 5h). By contrast, RIG-I signalling remained intact in Ankib1^−/−^ cells, with comparable phosphorylation of TBK1, IRF3 and STAT1 to controls (Extended Data Fig. 5i). These observations were further confirmed by cytokine quantification: control BMDMs secreted high levels of IFNβ upon TLR3, TLR4 or STING stimulation, whereas Ankib1^−/−^ BMDMs showed blunted IFNβ production upon TLR3 and TLR4 stimulation and a substantial decrease therein upon STING activation (Fig. 6a–c). By contrast, IFNβ production in response to RIG-I stimulation was unaffected by ANKIB1 deficiency (Fig. 6d). These results demonstrate that ANKIB1 is a crucial driver of IRF3 activation and consequent IFN induction downstream of TLR3, TLR4 and cGAS–STING.Fig. 6ANKIB1 is crucial for TLR3- and cGAS–STING-mediated IFN induction in vivo*.a–d, Control (ctrl) and Ankib1^−/−^ BMDMs were stimulated with 5 µg ml^−1^ poly(I:C) for 24 h (a), 200 ng ml^−1^ LPS for 4 h (b), 5 µg ml^−1^ ADU-S100 for 6 h (c) or 200 ng ml^−1^ 3p-hpRNA for 6 h (d) and the supernatants were collected to determine IFNβ concentration by ELISA. The graphs represent mean ± s.e.m. of three or four independent biological replicates. P values were determined by two-way ANOVA with uncorrected Fisher’s LSD multiple-comparisons tests with 95% confidence intervals. P < 0.0001, P < 0.0001, P < 0.0001 and P = 0.6573, respectively. e, A workflow of the sHLH model. f, Poly(I:C) (10 mg kg^−1^) was injected intraperitoneally to 8–12-week-old WT littermate control and Ankib1^−/−^ mice, followed 24 h later by intraperitoneal administration of 5 mg kg^−1^ LPS. The survival rates of the mice were monitored for 5 days and a log-rank Mantel–Cox test was used to determine the statistical significance. P = 0.0018. hpi, hours post-injection. g, Spleens of 9–10-week-old WT littermate control and Ankib1^−/−^ mice were collected for RNA extraction after 10 mg kg^−1^ intraperitoneal poly(I:C) injection for 3 and 6 h. The RNA-sequencing analysis is represented as a heat map, using two independent biological replicates per group for each genotype. h, IFNα response gene set enrichment analysis was performed using spleen samples for both the 3-h and the 6-h time points. Statistical significance of the data was determined using the Kolmogorov–Smirnov test. i, WT littermate control and Ankib1^−/−^ mice were challenged with 2 × 10⁸ colony-forming units Escherichia coli-induced sepsis. The survival rates of the mice were monitored for 5 days and a log-rank Mantel–Cox test was used to determine the statistical significance. P = 0.9001. j, Eight to ten-week-old WT littermate control and Ankib1^−/−^ mice were challenged with 1 × 10^6^ p.f.u. IAV. The weight loss of the mice was monitored for 15 days. Data show the mean ± s.d. of independent biological replicates. P values were determined by two-way ANOVA. P = 0.2721. dpi, days post-infection. k, Twelve to sixteen-week-old male Ankib1^−/−^ (n = 5) and WT littermate control mice (n = 8) were challenged with 1 × 10^6^ PFU HSV-1 SC16. The survival rates of the mice were monitored for 14 days and a log-rank Mantel–Cox test was used to determine the statistical significance. P = 0.0395. l, GO analysis was performed using RNA samples from the brainstem of WT littermate control and Ankib1^−/−^ mice at 3 days post infection (dpi) with HSV-1. m,n, The respective cytokines in lung homogenates of 12-week-old WT littermate control (n = 4) and Ankib1^−/−^ (n = 5) mice were measured at day 5 post infection with HSV-1 SC16 by multiplex ELISA. P = 0.007 and P = 0.0209, respectively. ELISA graphs represent mean ± s.d. of independent biological replicates. P values were determined by an unpaired two-tailed Student’s t-test. ^∗^P ≤ 0.05, ^∗∗^P ≤ 0.01, ^∗∗∗^P ≤ 0.001 and ^∗∗∗^P ≤ 0.0001 o, GO analysis was performed using RNA samples from brainstems of WT littermate control and Ankib1^−/−*^ mice at 5 dpi with HSV-1. Panel e created with BioRender.com.

ANKIB1 is crucial for TLR3- and cGAS–STING-mediated IFN induction in vivo

During infection, type-I IFNs are critical for orchestrating an effective immune response^45,46^. However, excessive or uncontrolled type-I IFN production can be detrimental, as seen in various so-called interferonopathies^47^, such as Aicardi–Goutières syndrome and STING-associated vasculopathy with onset in infancy, as well as secondary haemophagocytic lymphohistiocytosis (sHLH)^47–52^. The latter is characterized by excessive macrophage activation, leading to hyperinflammation and multi-organ failure^53^. As no natural pathogen is known to elicit IFN exclusively through TLR3, we employed an in vivo model of sHLH (Fig. 6e) in which TLR3-driven type-I IFN converts an otherwise sublethal TLR4 stimulation (Extended Data Fig. 6a) into a lethal challenge^47^. Strikingly, ANKIB1 deficiency significantly protected from TLR3-driven sHLH, as 78% of Ankib1^−/−^ mice survived whereas all control mice succumbed to the challenge (Fig. 6f). RNA-sequencing of spleens collected after poly(I:C) treatment revealed that, while poly(I:C) induced robust type-I/III IFN expression in control mice, this response was largely absent in Ankib1^−/−^ mice (Fig. 6g). Importantly, while NF-κB target genes such as IL-6 and IL-1β were equally induced in both genotypes, IFN-stimulated genes were selectively upregulated in controls. Gene set enrichment analysis further showed a marked suppression of IFNα and IFNγ response signatures in Ankib1^−/−^ mice, whereas TNF, IL-6 and MAPK signalling pathways were not strongly affected (Fig. 6h and Extended Data Fig. 6b–e). Together, these results establish ANKIB1 as a key driver of TLR3-induced type-I/III IFN production and responses in vivo, and identify it as a critical mediator of TLR3-induced lethality in interferonopathy.

Given the established role of ANKIB1 in TRIF- but not MyD88-dependent TLR4 signalling, we next determined whether ANKIB1 deficiency influences TLR4-driven, MyD88-dependent pathology in vivo. To accomplish this, we employed an E. coli-induced sepsis model in which the lethality is driven by MyD88-dependent TLR4 signalling^54^. Notably, ANKIB1-deficient and control mice exhibited comparable susceptibility to E. coli-induced lethality (Fig. 6i), indicating selective requirement for ANKIB1 in TRIF-dependent, but not MyD88-dependent, TLR4 signalling in vivo.

We next assessed the role of ANKIB1 in the context of viral infection. First, we used IAV as an RNA virus model, in which viral RNA is known to be primarily sensed by RIG-I, leading to a potent IFN response essential for host survival^45,55,56^. In this model, weight loss and recovery were indistinguishable between control and Ankib1^−/−^ mice (Fig. 6j), providing in vivo confirmation that ANKIB1 is not essential for RIG-I signalling.

Finally, we intranasally infected mice with HSV-1 SC16, a DNA virus primarily sensed by the cGAS–STING pathway whose induction of IFN is indispensable for survival in this model^57–59^. As expected, 75% of control mice survived infection (Fig. 6k). Remarkably, Ankib1^−/−^ mice exhibited severe susceptibility, with only 20% of the mice surviving the challenge (Fig. 6k), implicating ANKIB1 as a key regulator of antiviral defence against HSV-1. Interestingly, the susceptibility of Ankib1^−/−^ mice to HSV-1 infection is similar to that previously reported for Optn^−/−^ mice in the same model^60^, which, in both cases, is lower than expected for complete loss of cGAS–STING signalling^59^. This result is probably reflective of a substantial, yet incomplete dependence of cGAS–STING-induced IFN induction on ANKIB1–OPTN signalling also in vivo, in line with our observations in human cell lines (Fig. 4n) and primary murine immune cells ex vivo (Fig. 6c and Extended Data Fig. 5h).

To determine the mechanism underlying this increased lethality, we collected brainstems—the primary site of HSV-1 replication in this model^59^ at day 3 post infection and analysed cytokine expression. Notably, IFNβ response signalling was significantly downregulated in HSV-1-infected Ankib1^−/−^ compared with control mice (Fig. 6l). However, NF-κB target genes showed heterogeneous expression, indicating that ANKIB1 primarily affected IFN signalling. We also observed a 50% reduction in IFNβ levels in pulmonary tissues from infected mice, while NF-κB-driven cytokines such as IL-6 remained largely unchanged (Fig. 6m and Extended Data Fig. 6f). Consequently, immune activation and cytotoxic effector responses were also impaired, as evidenced by reduced granzyme B expression in the lungs and the analysis of the downregulated genes in the brainstems of Ankib1^−/−^ mice (Fig. 6n,o). Thus, ANKIB1 is an important driver of cGAS–STING-dependent IFN induction and antiviral immunity against HSV-1 infection in vivo.

In summary, the results demonstrate that ANKIB1 is crucial for IFN-dependent biological effects triggered by TLR3 and cGAS–STING in vivo, whereas this is not the case for RIG-I and the MyD88-dependent arm of TLR4.

Discussion

We here describe ANKIB1 and the product of this catalytic activity, K11-Ub, as previously unknown crucial drivers of TBK1 and IRF3 activation and type-I/III IFN induction downstream of TLR3, TLR4 and cGAS–STING. This is achieved via a hitherto unknown signalling axis comprised of ANKIB1–K11-Ub–OPTN–TBK1, which acts as a molecular bridge linking activation of TLR3, TLR4, and cGAS–STING to TBK1 activation, thereby enabling IRF3 activation and IFN induction. We provide evidence for the physiological relevance of ANKIB1 as the apical component of a previously unrecognized signalling axis in in vivo disease models of TLR3-dependent interferonopathy and cGAS–STING-mediated, IFN induction-dependent antiviral immunity.

Mechanistically, ANKIB1 is recruited to the TLR3-SC via its UIM domain, where it K11-ubiquitinates TRIF, NEMO, OPTN and itself. K11-Ub was previously shown to regulate cell cycle progression by enabling proteasomal degradation^61^, with more recent studies suggesting that, rather than acting alone, K11-Ub triggers proteasomal degradation in concert with other chain types, including K48-Ub or K63-Ub^62^ with non-degradative signalling roles^63^. We show that ANKIB1-generated K11-Ub is essential for the recruitment of OPTN to the TLR3-SC, which in turn serves as critical adaptor for TBK1 recruitment and activation, driving IRF3 phosphorylation and IFN production. Interestingly, Dynek et al. reported that K11-Ub confers a scaffold property to RIPK1 in TNFR1 signalling and that NEMO, which contains a UBAN domain similar to that of OPTN, is able to bind to such K11-Ub^64^.

We also identified a role for ANKIB1 in TRIF-dependent TLR4 signalling. TLR4 signals via both MyD88 and TRIF^65,66^, with TRIF containing the IRF3-binding motif for IFN production^9,40^. While TANK mediates TBK1 and IKKε activation in MyD88 signalling^34^, it is dispensable for TBK1 and IRF3 activation downstream of TRIF. Our findings align with prior studies showing that OPTN—but not TANK—is essential for IFN production in TRIF-dependent TLR4 signalling^32,37^. On the basis of our findings, we propose that the ANKIB1–K11-Ub–OPTN–TBK1 signalling axis we identified as responsible for IFN induction downstream of TRIF in TLR3 signalling operates likewise in TLR4 signalling.

Finally, we demonstrated a key role for ANKIB1 in cGAS–STING signalling. Prior studies reported a direct TBK1–STING interaction via a C-terminal TBK1-binding motif^7,67^, whereas others highlighted a requirement for ubiquitination^68–70^. Our results reconcile these models by showing that the ANKIB1–OPTN dependence of cGAS–STING-induced TBK1 activation and IFN induction is substantial yet incomplete. It is possible that ANKIB1–OPTN-independent, cGAS–STING-induced TBK1 activation is due to direct binding of TBK1 to STING, as previously shown^7,67^; it remains to be shown, however, whether this is truly the case.

Consistent with previous work that OPTN deficiency impairs STING-induced IFN induction, thereby increasing susceptibility to HSV-1 infection^60,71^, we show that OPTN is required for effective TBK1 activation downstream of STING. We identify STING, NEMO, OPTN and ANKIB1 itself as substrates of ANKIB1 and show, by using ANKIB1-RING1^mut^ cells, that the catalytic activity of ANKIB1 enables OPTN and TBK1 recruitment, contributing substantially to IRF3 activation and type-I and type-III IFN production. By contrast, ANKIB1 and OPTN are dispensable for robust RIG-I–MAVS signalling, whereas TANK, NAP1 and/or SINTBAD are required for it, as previously reported^36^.

Interestingly, the biochemical mechanism of the catalytic activity of ANKIB1 is distinct from that of other RBR-E3s such as PARKIN or HOIP^72,73^ as it depends on the RBR-RING1 rather than the RBR-RING2. The identification of this currently unique mechanism for ANKIB1 among the RBR-E3s proposes the existence of two distinct types of RBR-E3s, one relying on RING2 and the other one on RING1. Structural studies will be required to determine how ANKIB1 accomplishes this.

We identify ANKIB1 as crucial driver of activation of TBK1, IRF3 and type-I/III IFN induction in signalling triggered by TLR3, TRIF-dependent TLR4 and cGAS–STING, but not RIG-I. Contrasting with the latter, it was recently reported that ANKIB1 would interfere with RIG-I-induced IFN induction, that is, exert the opposite function to the one we identified for ANKIB1 in TLR3, TLR4 and cGAS–STING but not RIG-I–MAVS signalling, by mediating the proteasomal degradation of MAVS through K48-linked ubiquitination^74^. However, we neither observed such interference with RIG-I-induced IFN induction in ANKIB1-proficient cells nor did we detect proteasomal degradation of MAVS upon treatment of such cells with the RIG-I ligand 3p-hpRNA or by infection with the RIG-I-stimulating viruses IAV and SeV. We attribute this differential requirement to distinct biochemical mechanisms of TBK1 activation downstream of various PRRs. Whereas cGAS–STING and TRIF-dependent TLR3 and TLR4 signalling rely on OPTN-mediated TBK1 phosphorylation, which requires recruitment via ANKIB1-generated K11-linked ubiquitin chains, TBK1 activation downstream of RIG-I–MAVS primarily depends on the adaptor proteins TANK, NAP1 and SINTBAD, whose recruitment, as we show, is ANKIB1-independent.

In summary, we here identify a previously unrecognized ANKIB1–K11-Ub–OPTN–TBK1 signalling axis as the crucial driver of IRF3-mediated IFN induction downstream of TLR3, TLR4 and cGAS–STING (Extended Data Fig. 5j). Notably, ANKIB1 deletion confers significant protection in an otherwise lethal TLR3-dependent interferonopathy mouse model. Future strategies aimed at inhibiting the catalytic function of ANKIB1 or degrading it altogether may open additional avenues for treating IFN-driven pathologies.

Methods

All animal procedures were conducted in accordance with European, national and institutional guidelines and approved by local government authorities (LAVE NRW; permission Az. 81-02.04.2020.A022). Study protocols and animal experiments performed in Zaragoza were approved by the Animal Experimentation Ethics Committee of the University of Zaragoza (numbers PI 18/2, PI 18/23, PI 62/22 and PI 24/25) and in accordance with institutional, national and European ethical animal regulations (Protection of Animals Act).

Generation of ANKIB1 expression constructs

The coding sequence of ANKIB1 was amplified using Phusion High-Fidelity PCR Master Mix (New England Biolabs) and cloned to the pBABE-Puro plasmid using XhoI and AgeI (New England Biolabs). Deletions and point mutations were introduced to into the plasmid using the primers presented in Supplementary Table 1 with the Q5 Site-Directed Mutagenesis kit (New England Biolabs) according to the manufacturer’s instructions.

Recombinant proteins, reagents and cell lines

E1, E2s, USP2 and ubiquitin were purchased from R&D systems. moTAP-TNF was produced and purified as described in ref. ^75^. Poly(I:C) HMW was purchased from Invivogen. LPS was purchased from Enzo Life Sciences. ADU-S100 was purchased from MedchemExpress and 2′3′cGAMP was purchased from Selleck Chemicals. All primers listed in Supplementary Table 1 were produced by Sigma-Aldrich.

WT HeLa (ATCC CCL-2), HT-29 (ATCC HTB-38) and A549 (ATCC CCL-185) cell lines used herein were purchased from the ATCC. ANKIB1 and TLR3 were depleted using the plasmid px458 with the respective gRNA sequences presented in Supplementary Table 1 and performing single-cell sorting upon transfection. TANK/NAP1/SINTBAD/OPTN were generated using the gRNA published in ref. ^10^. The generation of viral particles and reconstitution of cancer cell lines with different ANKIB1 mutants or TLR3-moTAP were performed as described previously^10^. Reconstituted cells were either selected via GFP-positive cell sorting by fluorescence-activated cell sorting (FACS) or via puromycin treatment.

SDS–PAGE and western blotting

Proteins were separated using 4–15% Mini- or Midi-PROTEAN-TGXTM-gels (Bio-Rad) with Tris/glycine/SDS running buffer. Proteins were transferred on Mini- or Midi-0.2-μm nitrocellulose membranes (Bio-Rad transfer packs) using the Trans-Blot Turbo Transfer System from Bio-Rad. Proteins were analysed by western blotting using the respective antibodies presented in Supplementary Table 2 and either LLC Western Lightning Plus ECL (Revvity Health Sciences) for chemiluminescent signal or a Li-Cor Odyssey CLx for fluorescent signal.

Mice

C57BL/6N Ankib1^−/−^ tm1a mice were developed by the European Conditional Mouse Mutagenesis Program using Knockout-First-Reporter Tagged Insertion allele technology. The mice used in this study were maintained and bred in the animal facility of the CECAD Research Centre, University of Cologne. Their microbiological status was examined as recommended by the Federation of European Laboratory Animal Science Associations and the mice were free of all listed pathogens.

Isolation of BMDMs and splenocytes

Bone marrow isolation was performed as described previously^10^. Cells were plated in non-coated 10-cm dishes at a concentration of 2 × 10^6^ cells ml^−1^ and treated with 20 ng ml^−1^ M-CSF for 7 days. The cells were then detached with 10 mM EDTA for 10 min and plated in a 12-well plate for the experiments. Splenocytes were isolated from 8–12-week-old mice. Spleens were mashed on a 75-μm filter and red blood cells were removed as for BMDMs. Cells were counted and plated at 2.5 × 10^6^ cells ml^−1^ in DMEM containing 1% L-glutamine and 1% penicillin/streptomycin with 10% decomplemented FBS.

RNA isolation and RT–qPCR

Total RNAs from HeLa and HT-29 cells with or without the respective treatment were extracted using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions From the cDNA generated by the Iscript Reverse Transcription Supermix (Bio-Rad), RT–qPCR was performed using Itaq Universal SYBR (Bio-Rad), with the primers listed in Supplementary Table 1 and the following conditions: initial polymerase activation step of 30 s at 95 °C, followed by 35 amplification cycles of 5 s at 95 °C and 30 s at 64 °C. Three technical replicates were generated for each sample and run on a Bio-Rad CFX Opus 96. GAPDH was used as reference gene and the relative expression of the gene transcripts was analysed using the 2^−ΔCt^ method.

For RT–qPCR after in vitro infection with either MVA or SeV, samples were collected with 250 µl lysis buffer (4 M guanidine thiocyanate, 25 mM Tris–HCl pH 7 and 143 mM 2-mercaptoethanol) followed by the addition of 250 µl 70% ethanol. Total RNA was extracted using a silica column (Epoch Life Science) and washed once with wash buffer 1 (1 M guanidine thiocyanate, 25 mM Tris–HCl pH 7 and 10% ethanol) and twice with wash buffer 2 (25 mM Tris–HCl pH 7 and 70% ethanol). Total RNA was eluted using nuclease-free water. Then 500 ng of RNA was used to produce cDNA using SuperScript III reverse transcriptase (Thermo Scientific) following the manufacturer’s protocol. Samples were diluted and used as input in a 20 µl reaction using the KAPA SYBR Fast System (Roche) following the manufacturer’s protocol.

RNA sequencing

Total RNA from HeLa cells treated with poly(I:C) were extracted using the RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. RNA sequencing was performed by CeGaT GmbH. In brief, 10 ng RNA of each sample were used for library preparation with the SMART-Seq stranded total RNA kit (Takara). Libraries were sequenced on a NovaSeq 6000 machine (Illumina) with 2× 100 bp. The sequencing reads were demultiplexed with Illumina bcl2fastq (2.20) and adaptors were trimmed with Skewer (version 0.2.2). Quality trimming of the reads was not performed. The raw counts derived from the mapping contain the number of reads that map to each gene ID. Based on these numbers, the normalized counts were calculated. Normalized counts were calculated with DESeq2 in R.

Total RNA from the spleen of poly(I:C)-injected mice were extracted using the RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. mRNA sequencing for mouse spleen samples was performed by Cologne Center for Genomics. Briefly, 3′ mRNA libraries were generated from 200 ng (spleen samples) total RNA using the QuantSeq 3′ mRNA-Seq FWD V2 Library Prep kit (Lexogen) according to the manufacturer’s protocol and using 17 cycles for library amplification. After validation (using a TapeStation, Agilent Technologies) and quantification (Qubit, Thermo Fisher Scientific) individual libraries were pooled. The library pools were quantified using the Collibri Library Quantification kit (Thermo Fisher Scientific) and the QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). The libraries were subsequently sequenced on an Illumina NovaSeq 6000 instrument using a 1× 100 bp sequencing protocol and aiming for 10–12 million reads per sample.

ELISA

Supernatants were collected upon respective treatment. Before freezing at −80 °C, samples were spun down for 10 min at 450g and the supernatants were transferred. The IFNβ ELISA was conducted following the manufacturer’s protocol (R&D).

Cell activation and immunoprecipitation

For cell activation followed by immunoprecipitation, cells were treated for the indicated times and then lysed in 1 ml lysis buffer (30 mM Tris–HCl pH 7.5, 120 mM NaCl, 1,7 mM DDM, 2 mM KCl, 2 mM EDTA, 10% glycerol, phosSTOP (Roche) and EDTA-free protease inhibitor cocktail (Roche)) at 4 °C for 20 min before centrifugation at 17,500g at 4 °C for 30 min. Subsequently, the cell lysates were incubated with M2 anti-Flag magnetic beads (Sigma-Aldrich) overnight at 4 °C. The following day, beads were washed five times with lysis buffer and proteins were eluted by boiling in reducing sample buffer. Samples were analysed by western blotting as previously described.

Production of TUBE and deubiquitinases

The coding sequence of AMSH was cloned into the GEX6-P2 vector. pOPINK-Cezanne (OTU, aa 53–446) was a gift from David Komander (Addgene plasmid #61581; RRID: Addgene_61581). The glutathione S-transferase (GST)-tagged TUBE (GST-TUBE) was previously described^76^. Expression vectors were transformed into BL21 (DE3) bacteria and the respective proteins were purified as described previsouly^75^. After purification, the GST tag was removed from AMSH and Cezanne using human rhinovirus 3C protease (Cytiva 27084301) according to the manufacturer’s instructions. The protein concentration was measured with a Nanophotometer (Implen).

In vitro ubiquitin assay

MoTAP tagged-ANKIB1 WT and ΔRING2 were purified by Flag-tag. The amount and quality of purification was assessed by Coomassie brilliant blue G-250. Different versions of ANKIB1 were incubated for 1 h at 37 °C in a 30 μl reaction mixture containing either 200 nM UBA1; 1 μM UBE2L3, 1 μM UBE2D3 or 1 μM UBE2L6; 100 μg ml^−1^ ubiquitin (R&D systems); 2 mM DTT; 30 mM Tris–HCl, pH 7.5; 5 mM MgCl_2_ and 1× Energy Regenerating Solution (Enzo Life Sciences). For negative controls, ubiquitin buffer was added instead of the E1 and E2. Afterward, the samples were analysed by western blot. For subsequent deubiquitinase assay, ANKIB1 was removed with M2-Flag beads and either USP2 or Cezanne were added at a concentration of 1 μM and 0.5 µM, respectively, and samples were incubated at 30 °C for 1 h. The reaction was terminated by adding and boiling in reducing sample buffer.

K11-Ub antibody production

The amino acid sequence of the K11-Ub linkage-specific 2A3/2E6 antibody was as previously described^60^, and the coding sequence was synthesized by Thermo Scientific and cloned into the pVITRO plasmid using AgeI, BsrGI, BspeI and BamHI restriction enzymes (New England Biolabs). The antibody was transiently expressed in Expi293F cells (Gibco, A14527) and purified using a Protein G column (Sigma-Aldrich, GE17-0618-01).

TUBE PD and in vitro deubiqutination assay

HeLa WT, ANKIB1-deficient and stably expressing RING1-mutant ANKIB1 cells were treated with either poly(I:C) or 2′3′cGAMP for the indicated times and then sonicated and lysed in denaturing TUBE PD lysis buffer (30 mM Tris–HCl pH 7.5, 120 mM NaCl, 2 mM EDTA, 2 mM KCl, 1 mM CaCl_2_, 1 mM MgCl_2_, 0.5% Triton-X 100 1% SDS, 10% glycerol, phosSTOP (Roche) and EDTA-free protease inhibitor cocktail (Roche)) and 45 µM PR619 DUB inhibitor (Selleckchem) at 4 °C for 30 min before centrifugation at 17,500g for 30 min. Samples were then diluted with SDS-free TUBE PD lysis buffer, thus decreasing the SDS concentration to 0.1%, and then incubated with glutathione sepharose beads (Cytiva) freshly pre-coupled with GST-TUBE (20 μg per sample) overnight at 4 °C. The following day, beads were washed with SDS-free TUBE PD lysis buffer and proteins were either eluted from beads by boiling with reducing sample buffer followed by immunoblotting or subjected to an ‘in vitro deubiquitination assay’ after washing beads twice with DUB reaction buffer (50 mM Tris–HCl pH 7.5 and 100 mM NaCl). In short, the beads were incubated with 600 µg ml^−1^ K48-tetraubiquitin chains (Adipogen) in DUB reaction buffer with 1 mM DTT at 37 °C for 30 min. Meanwhile, AMSH, Cezanne and USP2 (R&D Systems) were diluted to 2× the indicated concentrations in DUB dilution buffer (25 mM Tris–HCl pH 7.5, 150 mM NaCl and 1 mM DTT) and incubated at room temperature for 10 min. The respective DUBs were then mixed with the beads and incubated at 37 °C for 1 h. The reactions were terminated by boiling the samples in reducing sample buffer before analysis by immunoblotting.

HSV-1 infection and disease monitoring

Age-matched (12–16-week-old) male WT littermate control (n = 8) and Ankib1^−/−^(n = 5) mice were anaesthetised in a mouse anaesthesia induction chamber with 2–3% isoflurane. Mice were infected intranasally with 1 × 10^6^ plaque-forming units (p.f.u.) of HSV-1 SC16 strain, using a total volume of 40 µl. Mice were weighed daily and a clinical score was applied when mice showed signs of disease. The clinical score evaluated mouse appearance, level of consciousness, activity, response to stimuli, eye appearance and frequency and quality of respiration, standardized to a five-point scale ranging from 0 to 4. The humanitarian end point was reached with a 25% weight loss, a score equal to or higher than 15 or if any respiratory characteristics increased by more than 3 (ref. ^77^). Mice were euthanized following human end point standards.

Secondary HLH and E. coli sepsis model

The model of secondary HLH was performed using age-matched (8-week-old) female and male WT littermate control (n = 8) and Ankib1^−/−^ (n = 9) mice as described previously^78^. For the sepsis model, a strain of E. coli isolated from the blood of a WT mouse with CLP-induced sepsis was used^79^. Sepsis was induced by inoculating 2 × 10⁸ colony-forming units in 200 µl of PBS to 8-week-old female and male WT littermate control (n = 9) and Ankib1^−/−^ (n = 8) mice via intraperitoneal injection. The mice were weighed and observed daily, with sepsis scoring performed regularly, and survival monitored over 5 days. A humane end point was applied when the murine sepsis score exceeded 14 or when body weight loss was greater than 25%.

IAV infection

Age-matched (12–16-week-old) WT littermate control (n = 7) and Ankib1^−/−^ (n = 7) female and male mice were infected with a mouse-adapted pandemic influenza strain H1N1 A/PR/8/3424 virus. The mice were anesthetized in a chamber with 2–3% isoflurane and intranasally infected with 2 × 10^3^ p.f.u. of IAV in 40 µl. The mice were weighed and observed daily, with sepsis scoring performed and survival monitored as described previously^77^.

In vitro infection assays

Monolayers of HeLa or A549 cells in 6-well plates were washed once with PBS and infected with the indicated viruses at a multiplicity of infection of 5 (MVA, HSV-1 and IAV) or 1 (SeV) in diluted serum-free media for 2 h at 37 °C. Inoculum was then removed and replaced with complete DMEM. Cell lysates were collected on ice at the indicated time points with RIPA buffer (50 mM Tris–HCl pH 8, 150 mM NaCl, 1% NP40, 0.1%SDS and 0.5% Na Deoxycholate) supplemented with complete EDTA-free protease inhibitors and PhosSTOP phosphatase inhibitor cocktail for western blot analysis.

nCounter RNA Analysis

At day 3 post-infection with HSV-1, brainstems of mice were isolated and subjected to RNA extraction using the RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. Total RNA was quantified using a spectrophotometer. For each reaction, 300 ng of purified total RNA was hybridized with nCounter Reporter and Capture probe sets (nCounter XT PGX MmV1 Cancer Immune) according to the manufacturer’s protocol. Hybridization was performed in a total volume of 15 μl at 65 °C for 18 h. Following hybridization, samples were loaded on the nCounter cartridge using the nCounter Prep Station and data were collected with the nCounter digital analyser. Raw counts were normalized using the housekeeping genes included in the nCounter XT Codeset and the nSolver software’s Basic Analysis tool. Normalized counts were exported for downstream analysis. Heat maps were generated using the pheatmap function from the pheatmap R package (vR 4.5.1).

Pathway analysis was performed with an overrepresentation test (Fisher’s exact test) using PANTHER v19.0 and the Gene Ontology (GO) database (10.5281/zenodo.15066566, released on 16 March 2025). In brief, downregulated genes with a moderate to high effect (log_2_FC <−0.5) were used as a gene list and the entire gene panel measured by the nCounter assay was used as the background. Gene symbols were used as identifiers.

K11-Ub immunoprecipitation

HeLa WT and ANKIB1-deficient cells were treated with poly(I:C) for the indicated times and lysed in denaturing K11 immunoprecipitation-lysis buffer (30 mM Tris–HCl pH 7.5, 120 mM NaCl, 2 mM EDTA, 2 mM KCl, 1% Triton-X 100, 1% SDS, 10% glycerol, phosSTOP, EDTA-free protease inhibitor cocktail and 45 µM PR619 DUB inhibitor) and processed as described in TUBE PD but then incubated with protein G-beads freshly pre-coupled with K11 antibody overnight at 4 °C. The following day, beads were washed with SDS-free K11 immunoprecipitation-lysis buffer and proteins were eluted by boiling them in reducing sample buffer. Samples were analysed by immunoblotting.

Cytokine/chemokine quantification

Lung samples were obtained from a weighed portion of murine lung, homogenized in 500 µl DMEM with a GentleMACS dissociator (Miltenyi) and centrifuged (900g for 5 min). The addition of 0.5% Triton-X 100 and protease inhibitor (Complete, Roche) to the homogenates inactivated the virus and prevented degradation.

The multiplex ELISA was performed via a custom Luminex Discovery assay (R&D Bio-Techne) according to the manufacturer’s instructions. Samples were centrifuged at 10,000g for 5 min and diluted at a 1:2 ratio in Calibrator Diluent (R&D Bio-Techne) before analysis. Samples were measured with a Luminex 200 xMAP system (Luminex) and xPONENT software was used for data collection and analysis.

Ubiquitin linkage type identification

For the in vitro ubiquitin assay, samples were prepared for mass spectrometry analysis after in-gel digestion overnight with trypsin. All samples were analysed by the CECAD Proteomics Facility on aQ Exactive Plus Orbitrap mass spectrometer coupled to an EASY nLC (both Thermo Scientific). Peptides were loaded with solvent A (0.1% formic acid in water) onto an in-house packed analytical column (50 cm, 75 µm inner diameter, filled with 2.7 µm Poroshell EC120 C18, Agilent). Peptides were chromatographically separated at a constant flow rate of 250 nl min^−1^ using the following gradient: 3–5% solvent B (0.1% formic acid in 80% acetonitrile) within 1.0 min, 5–30% solvent B within 121.0 min, 30–40% solvent B within 19.0 min, 40–95% solvent B within 1.0 min, followed by washing and column equilibration. The mass spectrometer was operated in data-dependent acquisition mode. The MS1 survey scan was acquired from 300 to 1750 m/z at a resolution of 70,000. The top ten most abundant peptides were isolated within a 1.8-Th window and subjected to HCD fragmentation at a normalized collision energy of 27%. The AGC target was set to 5e5 charges, allowing a maximum injection time of 110 ms. Product ions were detected in the Orbitrap at a resolution of 35,000. Precursors were dynamically excluded for 10.0 s.

All mass spectrometric raw data were processed with Maxquant v2.0.3.0 (ref. ^80^) using default parameters against the Uniprot canonical Human database (UP5640, downloaded 26 August 2020) with the match-between-runs option enabled between replicates. Follow-up analysis was done in Perseus 1.6.15 (ref. ^81^). Protein groups were filtered for potential contaminants and insecure identifications. The log_2_ intensities values of the different linkage types found in the samples were used to identify their respective abundance.

Statistics and reproducibility

All data collection and analysis for in vivo experiments were performed blindly by three researchers. GraphPad Prism V9.4 software (Graphpad) was used for statistical analyses. The normality of all data was confirmed using Anderson–Darling or Shapiro–Wilk tests. Error bars denote either s.e.m. or s.d. The statistical significance of data was assessed by either two-way analysis of variance (ANOVA), uncorrected Fisher’s least significant difference (LSD) multiple-comparisons test, unpaired two-tailed Student’s t-test or Kolmogorov–Smirnov or Fisher’s exact test with 95% confidence intervals. Survival curves were compared using a log-rank Mantel–Cox test with 95% confidence intervals. No statistical methods were used to predetermine sample sizes for in vivo experiments but our sample sizes are similar to those reported in previous publications^59,78,79^. Mice of the indicated genotype were randomly assigned to groups. No data were excluded from any experimental group.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41556-026-01886-z.

Supplementary information

Reporting Summary Supplementary Table 1. List of primers used in this study. Supplementary Table 2. List of antibodies used in this study.

Source data

Source Data Fig. 1. Unprocessed western blots. Source Data Fig. 2. Unprocessed western blots. Source Data Fig. 3. Unprocessed western blots. Source Data Fig. 4. Unprocessed western blots. Source Data Fig. 5. Unprocessed western blots. Source Data Extended Data Fig./Table 1. Unprocessed western blots. Source Data Extended Data Fig./Table 2. Unprocessed western blots. Source Data Extended Data Fig./Table 3. Unprocessed western blots. Source Data Extended Data Fig./Table 4. Unprocessed western blots. Source Data Extended Data Fig./Table 5. Unprocessed western blots. Source Data Extended Data Fig./Table 6. Statistical source data.