N4BP1 coordinates ubiquitin-dependent crosstalk within the IκB kinase family to limit toll-like receptor signaling and inflammation

Summary

The ubiquitin-binding endoribonuclease N4BP1 potently suppresses cytokine production by Toll-like Receptors (TLRs) that signal through the adaptor MyD88 but is inactivated via caspase-8-mediated cleavage downstream of death receptors, TLR3, or TLR4. Here we examined the mechanism whereby N4BP1 limits inflammatory responses. In macrophages, deletion of N4BP1 prolonged activation of inflammatory gene transcription at late time points after TRIF-independent TLR activation. Optimal suppression of inflammatory cytokines by N4BP1 depended on its ability to bind polyubiquitin chains, as macrophages and mice bearing inactivating mutations in a ubiquitin-binding motif in N4BP1 displayed increased TLR-induced cytokine production. Deletion of the non-canonical IκB kinases (ncIKKs), Tbk1 and Ikke, or their adaptor Tank phenocopied N4bp1 deficiency and enhanced macrophage responses to TLR1/2, TLR7, or TLR9 stimulation. Mechanistically, N4BP1 acted in concert with the ncIKKs to limit the duration of canonical IκB kinase (IKKα/β) signaling. Thus, N4BP1 and the ncIKKs serve as an important checkpoint against over-exuberant innate immune responses.

Graphical Abstract

eTOC blurb

The ubiquitin-binding endoribonuclease N4BP1 integrates signals from multiple innate immune receptors to regulate inflammatory cytokine responses. Gitlin et al. find that N4BP1 acts in concert with the non-canonical IκB kinases, TBK1 and IKKε, to suppress late-phase inflammatory gene transcription during Toll-like receptor responses, thereby limiting over-exuberant inflammation.

Introduction

Beyond its classical role in death receptor-induced apoptosis, there is burgeoning genetic evidence for the protease caspase-8 regulating both cell survival and inflammatory signaling¹. Human mutations impairing caspase-8 or its adaptor FADD cause severe immune-mediated diseases that encompass immunodeficiency and very-early onset inflammatory bowel disease^2-4. By contrast, mice lacking caspase-8 or FADD die in embryogenesis due to necroptosis driven by the kinase RIPK3 and pseudokinase MLKL^5-8. Although genetic deletion of Ripk3 or Mlkl enables Casp8-deficient mice to survive to adulthood, they eventually develop lymphadenopathy and splenomegaly owing to an accumulation of lymphocytes from impaired FAS signaling. In addition, young mice exhibit minimal sensitivity to an otherwise lethal dose of intraperitoneal lipopolysaccharide (LPS) and increased susceptibility to certain infections^9-12. Mouse macrophages deficient in Casp8 also produce abnormally low levels of inflammatory cytokines when exposed to the TLR4 agonist LPS in vitro. This defect stems, in large part, from a lack of N4BP1 cleavage by caspase-8. N4BP1 is a ubiquitin-binding endoribonuclease that suppresses, in an ill-defined manner, the production of certain cytokines and chemokines. Caspase-8 cleavage of N4BP1 downstream of TLR4, TLR3, TNFR1, or FAS inactivates its immunosuppressive function¹³.

TLR3 and TLR4 use the adaptor TRIF to signal to FADD and caspase-8, whereas TLRs that only engage the adaptor MyD88 do not stimulate caspase-8 activation. Consequently, the proinflammatory output of the TRIF-independent TLRs, when stimulated in isolation, is subject to regulation by N4BP1. If N4BP1 is lost, then TRIF-independent TLRs stimulate exorbitant proinflammatory cytokine production. However, when multiple innate immune receptors in a cell are triggered, inactivation of N4BP1 by caspase-8-activating receptors (e.g. TNFR1 or FAS) may license proinflammatory cytokine responses from non-caspase-8-activating receptors (e.g. the TRIF-independent TLRs). This model suggests that N4BP1 integrates innate immune signals, ensuring high levels of proinflammatory cytokines are only produced in response to the most threatening encounters with pathogens. However, the molecular mechanisms by which N4BP1 inhibits inflammatory cytokine production remain poorly understood.

Here, we examined the anti-inflammatory pathway by which N4BP1 controls cytokine responses in TLR-activated bone marrow-derived macrophages. In peritoneal macrophages and 293T cells, N4BP1 was reported to bind NEMO (also known as IKKγ), the regulatory subunit of the canonical IKK complex, and reduce the initial activation of IKKα/β¹⁴. However, IKKα/β activation peaks within minutes of TLR4 activation, whereas caspase-8 cleaves and inactivates N4BP1 later¹³. In mouse macrophages, genetic deletion of N4bp1 does not affect TLR4-induced cytokine production unless Casp8 is also deleted. This genetic evidence indicates that caspase-8 proteolytic activity prevents N4BP1 from inhibiting TLR4 signaling. Therefore, it is difficult to reconcile caspase-8 acting after the step that N4BP1 putatively regulates, namely IKKα/β activation. Another study proposed that N4BP1, like Regnase-1¹⁵, uses its endoribonuclease domain to cleave mRNA transcripts¹⁶. In this study, we found that N4BP1 did not require its endoribonuclease activity for cytokine suppression, nor did it substantially affect the initial activation of the IKKs. While we do not fully exclude a subtle effect of N4BP1 on IKK activation, we found it most prominently affected the later stages of the TRIF-independent TLR response, acting in concert with the noncanonical IKKs (ncIKKs) to curtail prolonged IKKα/β signaling.

The ncIKKs, TBK1 and IKKε, phosphorylate the transcription factors IRF3 and IRF7 to induce their nuclear translocation and type I interferon production^17,18. They also phosphorylate the kinase RIPK1, which suppresses RIPK1-driven cell death in response to TNF^19,20. Indeed, embryonic lethality of Tbk1^−/− mice is rescued by co-deletion of Tnf or inactivation of Ripk1^20-23. Moreover, a human autoinflammatory syndrome caused by biallelic mutations in TBK1 was ascribed to aberrant TNF and RIPK1-driven cell death²⁴. Our study identifies yet a third immunoregulatory pathway for the ncIKKs. As collaborators of N4BP1, ncIKKs suppress later waves of canonical IKK signaling downstream of TRIF-independent TLRs.

RESULTS

The RNAse activity of N4BP1 is dispensable for inflammatory cytokine suppression.

To investigate the role of N4BP1 endoribonuclease activity in the suppression of cytokine production by TRIF-independent TLRs, we generated N4bp1^D621A/D621A mice expressing N4BP1(D621A) from the endogenous locus. Mouse N4BP1(D621A), like its human counterpart N4BP1(D623A), lacked RNase activity and was unable to restrict HIV-1 in transfection studies (Fig. 1A)²⁵. Although N4BP1(D621A) was not expressed as well as wild-type (WT) N4BP1 in bone marrow-derived macrophages (BMDMs), N4bp1^D621A/D621A BMDMs produced only ~2.8-fold more IL-6 in response to TLR7 agonist R837 than WT N4bp1^+/+ BMDMs (Fig. 1B, C). By contrast, N4bp1^−/− BMDMs produced ~52-fold more IL-6 than WT BMDMs. CRISPR-mediated deletion of N4BP1(D621A) augmented IL-6 production to levels produced by N4bp1^−/− BMDMs (Fig. 1C). These data indicate that the RNase activity of N4BP1 is largely dispensable for limiting cytokine production.

Figure 1. RNAse activity and regulated mRNA decay appear to play little or no role in anti-inflammatory function of N4BP1.

(A) 293T cells were co-transfected with plasmids encoding HIV-1 (pNL4-3) and wild-type or mutant human (h) or mouse (m) N4BP1, as indicated. Plot depicts HIV-1 P24 protein levels in supernatants of transfected 293T cells, as measured by ELISA. Dots represent biological replicates; lines indicate mean ± SD. The ΔUb abbreviation refers to mouse N4BP1(I861A,F862A,P863A).

(B) Immunoblots of BMDMs of indicated genotypes showing protein levels of N4BP1. Lanes represent BMDMs from different mice.

(C) BMDMs of indicated genotypes were electroporated with Cas9 and control or N4bp1-targeting guide RNAs. At 7 days after electroporation, BMDMs were stimulated with R837 for 24 h and IL-6 secretion was measured in cell culture supernatants. Dots represent biological replicates; bars indicate the mean.

(D) Schematic for experimental workflow in panel E. Wild-type or N4bp1^−/− BMDMs were stimulated with R837 for 3 hours, at which point Actinomycin D was added to the cultures. RNA-Seq was performed at 3 hours post-R837 stimulation and after an additional 30 min, 1 h, 2, and 4 h of Actinomycin D treatment.

(E) Graphs indicate the transcript level in BMDMs of indicated genotypes after actinomycin D treatment relative to that at 3 h after R837 treatment. Dots represent biological replicates. Lines indicate the mean.

(F) Transcript levels in BMDMs of indicated genotypes after stimulation with R837. RPKM, reads per kilobase per million reads. Symbols represent biological replicates. Lines indicate the mean.

Next, we explored whether N4BP1 might affect mRNA stability independent of its RNase activity. Using actinomycin D to block further transcription in BMDMs at 3 hours after R837 stimulation, we found that Il6, Tnf, and Ccl4 mRNAs decreased similarly in WT and N4bp1^−/− BMDMs (Fig. 1D, E). JunB and Cxcl1 transcripts, previously suggested to be direct targets of N4BP1-mediated degradation based on overexpression studies¹⁶, also exhibited comparable declines in WT and N4bp1^−/− BMDMs (Fig. 1E). We conclude that loss of N4BP1 does not discernably increase the persistence of the above-mentioned cytokine or chemokine transcripts following actinomycin D-mediated transcriptional blockade. A caveat of these experiments is that transcript levels in WT and N4bp1^−/− BMDMs were not all identical at 3 hours of R837 treatment when actinomycin D was first added to the cultures (Fig. S1A).

Transcriptional profiling of WT BMDMs stimulated with R837 or the TLR1/2 agonist Pam3CSK4 revealed that many proinflammatory transcripts, including Tnf, Cxcl1, Csf3, Il6, and Tnfaip3, reached near-maximal levels within 2 to 4 hours and then plateaued or declined (Fig. 1F). Levels of the same transcripts in N4bp1^−/− BMDMs were comparable at 2 hours, but by 4 hours, had accumulated to a higher level or, in the case of Tnfaip3, failed to decline. In addition, the increase in IL-6 and CXCL1 cytokine secretion by N4bp1^−/− versus WT BMDMs mirrored the kinetics and magnitude of increase in Il6 and Cxcl1 transcripts at these time points (Fig. 1F, S1B). Therefore, rather than controlling the initial burst of transcription, N4BP1 controls the late-phase of gene expression.

N4BP1 controls late-phase, TLR-induced inflammatory gene transcription in a cell intrinsic manner.

Cytokines and chemokines produced by N4bp1x^−/− BMDMs could potentially enhance late-phase gene expression in a non-cell-autonomous manner. To assess this in N4bp1^−/− BMDMs, we cultured eGFP⁺ BMDMs in a 1:1 ratio with either WT or N4bp1^−/− BMDMs and stimulated them with R837 for 6 hours (Fig. 2A). Single-cell (sc)RNA sequencing revealed that WT and eGFP⁺ BMDMs were largely indistinguishable from one another, whereas N4bp1^−/− BMDMs were shifted with respect to the eGFP⁺ BMDMs within the same co-culture (Fig. 2B and S1C, D). Consistent with our earlier transriptomics, N4bp1^−/− BMDMs expressed more Tnf, Il6, Cxcl1 and Csf3 than their co-cultured eGFP⁺ counterparts (Fig. 2C). Thus, autocrine and/or paracrine signaling by cytokines produced by N4bp1^−/− BMDMs cannot fully account for their enhanced late-phase inflammatory gene transcription, indicating that N4BP1 regulates TLR7-induced gene expression in a cell autonomous manner.

Figure 2. N4BP1 controls late-phase inflammatory gene transcription.

(A) Scheme for co-culturing BMDMs, created with BioRender.com. Wild-type eGFP+ BMDMs were co-cultured with either N4bp1^+/+ or N4bp1^−/− BMDMs and stimulated with R837 followed by single-cell (sc)RNA sequencing. Data for this experiment is displayed in panels B and C.

(B) UMAP plots of single-cell (sc)RNA sequencing data from co-cultured BMDMs, as depicted in panel A.

(C) Dot plots of Tnf, Il6, Cxcl1, and Csf3 expression from indicated cell populations of co-cultured BMDMs. Dot color represents average expression; size of dot represents percent expression.

(D) RNA Polymerase II ChIP-Seq profiles for Cxcl1 and Cxcl2 in N4bp1^+/+ and N4bp1^−/− BMDMs at baseline and stimulated with R837 for 2, 4, or 6 h.

(E) Plot shows the number of significantly differential ChIP-Seq peaks genome-wide among N4bp1^−/− versus N4bp1^+/+ BMDMs stimulated with R837. These data are part of the same experiment in panel D. Cutoffs for determining differential peaks were log₂FC > 0 or log₂FC < 0 and FDR < 0.05, using Wald test.

(F) Gene sets enriched in upregulated RNA Polymerase II ChIP-Seq peaks among N4bp1^−/− BMDMs stimulated with R837 for 4 h in panel E.

See also Figure S1.

We further defined a set of N4BP1-regulated TLR7-induced transcripts based on our bulk RNA-seq experiments (Fig. 1F). Overlaying this gene signature with our co-culture experiment revealed an enhanced N4BP1-regulated signature among N4bp1^−/− BMDMs (Fig. S1E, F), corroborating a cell-intrinsic effect of N4BP1 on gene expression. In addition to this cell-intrinsic effect, the enhanced cytokine production by N4bp1^−/− BMDMs may have caused paracrine effects on co-cultured eGFP⁺ BMDMs, as eGFP⁺ BMDMs that were co-cultured with N4bp1^−/− BMDMs appeared slightly shifted in their transcriptional profile compared to eGFP⁺ BMDMs that had been co-cultured with N4bp1^+/+ BMDMs (Fig. 2B, C and S1D).

To further characterize the TLR7-induced transcriptional response of N4bp1^−/− BMDMs, we evaluated the association of RNA Polymerase II (RNAPII) with the genome by chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-Seq). At 2 hours after R837 stimulation, RNAPII was associated with proinflammatory genes, including Cxcl1 and Cxcl2, to a similar extent in both WT and N4bp1^−/− BMDMs (Fig. 2D, E). From 4 to 6 hours, however, chromatin-bound RNAPII subsided in WT BMDMs, but persisted in N4bp1^−/− BMDMs. The RNAPII ChIP-seq peaks that differed globally at 4 hours after stimulation with R837 were enriched in inflammation-associated gene sets (Fig. 2F). Based on these findings, N4BP1 appears to curtail the duration of gene transcription following TLR stimulation, thereby attenuating the late-phase inflammatory response.

TANK and the non-canonical IKKs suppress cytokine and chemokine production by TRIF-independent TLRs.

TANK, an adaptor of the ncIKKs, has previously been shown to suppress cytokine production stimulated by TLR1/2, TLR7, and TLR9²⁶. To determine if TANK and N4BP1 might act in the same pathway, we compared BMDMs lacking TANK (ΔTank) or N4BP1 (ΔN4bp1) after CRISPR-mediated gene deletion (Fig. S2A). Control, ΔTank, and ΔN4bp1 BMDMs stimulated with LPS produced comparable amounts of IL-6, but ΔTank and ΔN4bp1 BMDMs both made significantly more IL-6 than control BMDMs in response to R837 (TLR7 agonist), Pam3csk4 (TLR1/2 agonist), or CpG-B DNA (TLR9 agonist) (Fig. 3A). R837 also induced more G-CSF, CXCL1, CXCL5, CCL3, CCL5, and TNF when BMDMs lacked TANK or N4BP1 (Fig. 3B). Thus, Tank deficiency grossly phenocopies N4bp1 deficiency in BMDMs.

Figure 3. Deletion of TANK or the ncIKKs phenocopies the effect of deleting N4BP1.

(A, B) Plots depict cytokine and chemokine secretion in cell culture supernatants of BMDMs electroporated with Cas9 plus control or N4bp1- or Tank-targeted guide RNAs and stimulated with LPS, R837, Pam3csk4 (Pam3), or CpG for 24 h.

(C) Plots depict IL-6 and G-CSF secretion in cell culture supernatants of Ripk3^−/− or Ripk3^−/− Casp8^−/− BMDMs that were electroporated with Cas9 plus plus control or N4bp1- or Tank-targeted guide RNAs and stimulated with LPS for 24 h.

(D) Plot depicts IL-6 secretion from N4bp1^+/+ and N4bp1^−/− BMDMs electroporated with Cas9 plus control or Tank-targeted guide RNAs and stimulated with R837 at 2, 0.5, or 0.125 ug/ml for 24 h.

(E) Plot depicts cytokine and chemokine secretion in cell culture supernatants of BMDMs electroporated with Cas9 plus control, Tbk1, Ikke or both Tbk1 and Ikke-targeted guide RNAs and stimulated with R837 for 24 h.

(F) Plot depicts cytokine and chemokine secretion in cell culture supernatants of unstimulated Ripk1^+/+ and Ripk1^D138N/D138N BMDMs electroporated with Cas9 plus control, Tbk1, Ikke or both Tbk1 and Ikke-targeted guide RNAs.

(G) Plots depict IL-6 secretion in cell culture supernatants of Ripk3^−/− Casp8^−/−, Ripk1^D138N/D138N, and Tnf^−/− BMDMs electroporated with Cas9 plus control or Tbk1 and Ikke-targeted guide RNAs and stimulated for 24 h with R837.

In all panels, symbols represent biological replicates; bars indicate the mean. Where indicated, p-values determined by Dunnett’s multiple comparisons test.

See also Figure S2.

Given that caspase-8 inactivates N4BP1 and prevents it from suppressing LPS-induced cytokine responses¹³, we tested whether caspase-8 also negates a suppressive effect of TANK on TLR4 signaling. Specifically, we deleted Tank or N4bp1 in Ripk3^−/− or Ripk3^−/− Casp8^−/− BMDMs (Fig. S2B; Ripk3 deficiency is used to prevent necroptosis in Casp8-deficient cells). Tank deficiency phenocopied N4bp1 loss and increased LPS-induced IL-6 and G-CSF production in Ripk3^−/− Casp8^−/− BMDMs, but not Ripk3^−/− BMDMs (Fig. 3C). Therefore, caspase-8 restricts the function of both N4BP1 and TANK.

To further assess whether N4BP1 and TANK might act in the same pathway, we CRISPR-deleted Tank in WT or N4bp1^−/− BMDMs (Fig. S2C). Tank deficiency caused a modest increase in R837-induced IL-6 production by N4bp1^−/− BMDMs, whereas ΔTank and ΔTank N4bp1^−/− BMDMs made similar amounts of IL-6 (Fig. 3D). These data suggest functional overlap between N4BP1 and TANK, with N4BP1 contributing to some, but not all TANK-dependent functions.

Excessive cytokine production by Tank-deficient BMDMs led us to consider whether the ncIKKs might suppress cytokine production downstream of the TRIF-independent TLRs. Deletion of Tbk1 or Ikke (encoding IKKε) individually has been reported to have modest effects on TLR signaling^22,27. BMDMs lacking both TBK1 and IKKs (Fig. S2D) produced more G-CSF, IL-6, CXCL1, CXCL2, CXCL5, CCL3, CCL4, CCL5, and TNF in response to R837 than BMDMs lacking only TBK1 or IKKε (Fig. 3E). Thus, TBK1 or IKKε is sufficient to suppress TLR7-induced cytokine and chemokine secretion. In addition, human THP-1 monocytic cells CRISPR deleted for both TBK1 and IKKE and stimulated with Pam3csk4 produced higher levels of G-CSF, CXCL1, CCL3, CCL4, CCL5, and TNF than did singly-deficient or control cells (Fig. S2E, F). It should be noted, however, that ΔTbk1, Ikke BMDMs differed from ΔTank or ΔN4bp1 BMDMs because they made low levels of detectable CXCL1, CXCL2, and CCL5 even in the absence of stimulation. This spontaneous CXCL1, CXCL2, or CCL5 secretion was not seen when Tbk1 and Ikke were deleted in BMDMs expressing catalytically inactive RIPK1(D138N) (Fig. 3F, S2G). By contrast, co-deletion of both Tbk1 and Ikke still led to enhanced R837-induced IL-6 production in BMDMs expressing RIPK1(D138N), deficient in Tnf, or lacking both Ripk3 and Casp8 (Fig. 3G, S2G-I). Therefore, excessive TLR7-induced cytokine and chemokine production in ΔTbk1, Ikke BMDMs can occur independent of aberrant TNF-induced cell death. Collectively, our data support a model wherein N4BP1, TANK and the ncIKKs act together to limit cytokine production by the MyD88-dependent TLRs but point to additional TANK- or N4BP1-independent roles of the ncIKKs.

Role of ncIKK kinase activity in inflammatory cytokine suppression and transcriptional overlap with TLR-activated N4bp1-deficient macrophages

To determine whether the kinase activity of the ncIKKs is required to suppress TLR7-induced cytokines, we generated Tbk1^cKD/cKD Rosa26^CreERT2/+ mice that express catalytically inactive TBK1(D135N) after tamoxifen-induced Cre recombination (Fig. S3A). Tbk1^cKD/cKD Rosa26^CreERT2/+ and Tbk1^+/+ Rosa26^CreERT2/+ BMDMs expressed comparable amounts of TBK1 after treatment with 4-hydroxytamoxifen (Fig. S3B), but as expected^22,23, Tbk1^KD/KD BMDMs made less IFNβ than Tbk1^+/+ BMDMs when stimulated with the STING agonist 2’3’-cGAMP (Fig. 4A). TLR-induced IL-6 and G-CSF production was not greatly enhanced in Tbk1^KD/KD BMDMs compared with Tbk1^+/+ BMDMs (Fig. 4B). However, Tbk1^KD/KD BMDMs lacking Ikke (Fig. S3B), similar to ΔTbk1, Ikke BMDMs (Fig. 3E), made significantly more IL-6 and G-CSF than control BMDMs after stimulation with R837 (TLR7 agonist), Pam3csk4 (TLR1/2 agonist), or CpG-B DNA (TLR9 agonist) (Fig. 4B). Tbk1^KD/KD ΔIkke BMDMs also exhibited the spontaneous secretion of CXCL1, CXCL2, and CCL5 that was characteristic of ΔTbk1, Ikke BMDMs (Fig. S3C). These findings suggest that, in the absence of IKKε, the kinase activity of TBK1 suppresses cytokine and chemokine production.

Figure 4. Role of TBK1 kinase activity and overlap among ncIKK- and N4BP1-regulated transcriptional profiles.

(A) Plot shows IFNβ secreted from Tbk1^+/+ R26^CreERT2/+ and Tbk1^cKD/cKD R26^CreERT2/+ BMDMs treated with 4OHT and stimulated with 2’3’-cGAMP. Biological replicates shown as dots; lines indicate the mean.

(B) Plots show secretion of G-CSF (left) or IL-6 (right) from BMDMs of indicated genotypes that had been electroporated with Cas9 plus control or Ikke-targeted guide RNAs and stimulated for 24 h with poly(I:C) (p(I:C)), LPS, R837, Pam3csk4 (Pam3), or CpG. Biological replicates shown as dots; bars represent mean values.

(C) Four-way plot comparing differential gene expression among N4bp1^+/+ and N4bp1^−/− BMDMs stimulated with R837 (x-axis) versus differential gene expression among 4OHT-treated Tbk1^cKD/cKD R26^CreERT2/+ ΔIkke and Tbk1^WT/WT R26^CreERT2/+ ΔControl BMDMs (y axis) stimulated with R837.

(D) Pie charts show the proportion of N4BP1-regulated R837-induced genes whose expression is also upregulated in R837-stimulated Tbk1^cKD/cKD R26^CreERT2/+ ΔIkke BMDMs (upper), and the proportion of genes upregulated in R837-stimulated Tbk1^cKD/cKD R26^CreERT2/+ ΔIkke BMDMs that are also upregulated in R837-stimulated N4bp1^−/− BMDMs (lower).

See also Figure S3.

A comparison of the R837-induced transcriptional changes in Tbk1^KD/KD ΔIkke BMDMs with those in N4bp1^−/− BMDMs revealed that 62% of genes upregulated by N4bp1 deficiency were also upregulated by ncIKK deficiency (Fig. 4C, D). By contrast, only 20% of genes upregulated by ncIKK deficiency were also upregulated by N4bp1 deficiency. These correlative findings, together with our earlier cytokine secretion data, reveal a high degree of overlap in the set of TLR7-induced genes upregulated by deficiency in N4bp1 and the ncIKKs.

Prolonged IKKα/β activation in cells lacking N4BP1 or the non-canonical IKKs.

The ncIKKs have been suggested to dampen both canonical and non-canonical NF-κB signaling^28-30. Therefore, we tested if NF-κB transcription factors were required for enhanced R837-induced IL-6 production in N4bp1^−/− BMDMs. Deletion of RelA and Nfkb1 suppressed TLR7-induced IL-6 production in both WT and N4bp1^−/− BMDMs, whereas deletion of RelB and Nfkb2 did not (Fig. 5A, B). Therefore, enhanced late phase gene expression in N4bp1^−/− BMDMs is not driven solely by non-canonical NF-κB signaling through RelB and NF-κB2. Whether canonical NF-κB signaling through RelA and NF-κB1 drives late phase gene expression as well as early phase gene expression cannot be discerned from this experiment.

Figure 5. ncIKK-mediated suppression of TLR7-dependent cytokine production.

(A) Immunoblots of BMDMs electroporated with Cas9 plus the indicated guide RNAs.

(B) IL-6 secretion from N4bp1^+/+ and N4bp1^−/− BMDMs electroporated with control or Rela/Nfkb1-, or Nfkb2/Relb-targeted guide RNAs and stimulated with R837 for 24 h. Symbols represent biological replicates. Bars indicate the mean.

(C) Immunoblots of BMDMs electroporated with control or Tank-targeted guide RNAs and stimulated with R837 or LPS for indicated lengths of time.

(D) Immunoblots of N4bp1^+/+ and N4bp1^−/− BMDMs untreated or stimulated with R837 for 3 h.

(E) Immunoblots of N4bp1^+/+ and N4bp1^−/− BMDMs either untreated or stimulated with R837 after pre-treatment with TPCA-1 or DMSO as control.

Immunoblots representative of 2 or 3 independent experiments.

Seeking a connection between N4BP1, TANK and the ncIKKs, we analyzed R837- or LPS-induced activation loop phosphorylation of the ncIKKs in ΔTank (Fig. 5C) and N4bp1^−/− BMDMs (Fig. 5D). Consistent with the role of TANK as an adaptor for the ncIKKs²⁹, ΔTank BMDMs exhibited markedly less IKKε phosphorylation and slightly less TBK1 phosphorylation than control BMDMs after treatment with R837 or LPS (Fig. 5C). Activation of the ncIKKs was detected within 15 minutes of R837 treatment but became more prominent after 1 and 3 hours, consistent with prior work and the late-phase gene expression effects observed in N4bp1^−/− BMDMs^28,29. Curiously, TBK1 phosphorylation appeared normal in N4bp1^−/− BMDMs, whereas IKKε phosphorylation was higher than in WT BMDMs (Fig. 5D). TPCA-1, a canonical IKK inhibitor³¹, blocked R837-induced IκBα phosphorylation as well as enhanced IKKε phosphorylation in N4bp1^−/− BMDMs (Fig. 5E). This data suggested that N4BP1 might modulate crosstalk between the canonical and noncanonical IKKs.

Negative regulation of IKKα/β by N4BP1 and the ncIKKs

The ncIKKs are thought to downregulate canonical IKK activity through inhibitory phosphorylation of the IKKβ C-terminus^28,29,32-35. Therefore, we examined activation loop phosphorylation on IKKα and IKKβ in WT and N4bp1^−/− BMDMs stimulated with R837 or LPS. Given the limited sensitivity of phospho-specific IKKα/β antibodies, we first enriched for the canonical IKK complex by immunoprecipitating IKKγ/NEMO and IKKβ (Fig. 6A). N4bp1^−/− BMDMs had slightly more IKKα and IKKβ phosphorylation than WT BMDMs at 1 and 3 hours after R837 treatment, whereas no difference in IKKα and IKKβ phosphorylation was observed when WT and N4bp1^−/− BMDMs were stimulated with LPS. The increased IKKα and IKKβ activation loop phosphorylation in R837-stimulated N4bp1^−/− BMDMs coincided with reduced inhibitory phosphorylation of IKKβ Ser740 (Fig. 6B). Enhanced IKK activation loop phosphorylation and reduced inhibitory phosphorylation was also observed in R837-stimulated TBK1^KD/KD ΔIkke BMDMs (Fig. 6C). These results suggest that N4BP1 facilitates crosstalk between the canonical and noncanonical IKK complexes, promoting negative regulation of the canonical IKK complex by the ncIKKs to limit late phase proinflammatory gene expression.

Figure 6. Negative regulation of persistent canonical IKK activation by N4BP1 and the ncIKKs.

(A) Immunoblots of NEMO/IKKβ immunoprecipitates and input lysates from N4bp1^+/+ and N4bp1^−/− BMDMs stimulated with R837 or LPS as indicated.

(B) Immunoblots of NEMO/IKKβ immunoprecipitates from N4bp1^+/+ and N4bp1^−/− BMDMs stimulated with R837 as indicated.

(C) Immunoblots of NEMO/IKKβ immunoprecipitates from 4OHT-treated Tbk1^+/+ R26^CreERT2/+ and Tbk1^cKD/cKD R26^CreERT2/+ BMDMs electroporated with Cas9 plus control or Ikke-targeted guide RNAs and treated with R837 as indicated.

Each blot is representative of 2-3 independent experiments.

Role of K63/linear polyubiquitin chain binding by N4BP1 in suppressing inflammatory cytokine responses in vivo and in mediating interaction with ncIKKs

TANK is thought to recruit ncIKKs to innate immune signaling complexes that are modified with K63- and linear polyubiquitin chains^19,28,29,32. N4BP1 has also been shown to bind to K63- and/or linear polyubiquitin^14,36. To explore whether ubiquitin binding is required for the anti-inflammatory function of N4BP1, we first confirmed the region of N4BP1 required for binding to linear polyubiquitin. Consistent with previous reports^14,36, C-terminal residues 844-893 containing the CUE-like domain were required for binding to linear polyubiquitin (Fig. S4). X-ray crystallography of the CUE-like domain of N4BP1 bound to linear di-ubiquitin revealed that N4BP1 binds to the linear linkage by simultaneously interacting with the distal and proximal ubiquitin moieties (Fig. 7A and Table S1). Ile861/Phe862/Pro863 in N4BP1 appeared to interact with the Ile44 patch of one ubiquitin molecule. Accordingly, WT N4BP1 coimmunoprecipitated linear polyubiquitin generated by HOIP and HOIL-1 in 293T cells, but N4BP1(I861A,F862A,P863A) did not (Fig. 7B; for simplicity, this triple-mutation in N4BP1 is abbreviated as ΔUb). Intriguingly, N4BP1(ΔUb) still repressed production of HIV-1 p24 in transfection studies (Fig. 1A).

Figure 7. K63/Linear polyubiquitin-binding by N4BP1 facilitates cytokine suppression and enables co-association with noncanonical and canonical IKK complexes.

(A) Co-crystal structure of linear di-ubiquitin (grey with blue and yellow) bound by the CUE-like domain of N4BP1 (green).

(B) Immunoblots of input lysates and Flag immunoprecipitates of 293T cells transfected with the plasmids indicated.

(C) Immunoblots of N4bp1^+/+ and N4bp1^ΔUb/ΔUb BMDMs treated with LPS or TNF for indicated lengths of time.

(D) IL-6 and G-CSF secretion from N4bp1^+/+, N4bp1^ΔUb/ΔUb, and N4bp1^−/− BMDMs stimulated with R837 for 24 h. Symbols represent biological replicates. Bars indicate the mean.

(E) Plots depict serum cytokine levels among N4bp1^+/+ and N4bp1^ΔUb/ΔUb mice at baseline or 6 h after intraperitoneal injection of Pam3csk4. Dots represent individual mice. Lines indicate the mean.

(F) Immunoblots of input lysates and Flag immunoprecipitates from 293T cells transfected with indicated plasmids.

Results representative of 2-3 independent experiments.

See also Figures S4 and S5 and Table S1.

Next, we created mice expressing N4BP1(ΔUb) from the endogenous locus. Homozygous N4bp1^ΔUb/ΔUb mice were viable and overtly normal (Fig. S5A, B). They did not develop the spontaneous lymphoid aggregates seen in the lungs of N4bp1^−/− mice¹³, nor significant signs of inflammation by 18-20 weeks of age (Fig. S5C-F). N4bp1^ΔUb/ΔUb spleens displayed marginally enlarged germinal centers (Fig. S5D, F), but this finding was variable and not as pronounced as in N4bp1^−/− mice¹³. WT and N4bp1^ΔUb/ΔUb BMDMs expressed comparable amounts of N4BP1, but TNF- or LPS-induced cleavage of N4BP1 was attenuated, but not ablated, in N4bp1^ΔUb/ΔUb BMDMs (Fig. 7C). This result suggests that N4BP1 binding to polyubiquitin promotes, but is not required for, caspase-8 cleavage of N4BP1. As a caveat, although N4BP1(ΔUb) appears to retain the ability to diminish HIV-1 p24 production in 293T transfection studies (Fig. 1A), we cannot exclude a defect in N4BP1(ΔUb) protein folding as a cause of its impaired cleavage by caspase-8.

N4bp1^ΔUb/ΔUb BMDMs stimulated with R837 made significantly more IL-6 and G-CSF than WT BMDMs, but slightly less than N4bp1^−/− BMDMs (Fig. 7D). Indeed, analysis of a larger set of R837- and CpG-B-induced cytokines revealed a greater increase in their production by N4bp1^−/− compared to N4bp1^ΔUb/ΔUb BMDMs (Fig. S6A). Moreover, CRISPR-mediated deletion of N4BP1(ΔUb) in N4bp1^ΔUb/ΔUb BMDMs further augmented cytokine production (Fig. S6B), indicating that N4BP1(ΔUb) possesses greatly imapired, but residual, cytokine-suppressive activity. These observations may account for the less prominent immune phenotype of N4bp1^ΔUb/ΔUb relative to N4bp1^−/− mice (Fig. S5D-F). Nevertheless, when challenged intraperitoneally with the TLR1/2 agonist Pam3csk4, N4bp1^ΔUb/ΔUb mice produced more serum TNF, G-CSF, IL-6 and CXCL1 than WT littermates (Fig. 7E). Thus, polyubiquitin binding by N4BP1 is important for optimal suppression of proinflammatory cytokine and chemokine production downstream of TRIF-independent TLRs. Although we could not capture endogenous N4BP1 complexes in BMDMs successfully, WT N4BP1 expressed in 293T cells coimmunoprecipitated with TANK, IKKε, and the IKK complex only in the presence of HOIP, HOIL-1, and linear polyubiquitin (Fig. 7F). N4BP1(ΔUb) binding to TANK, IKKε and the canonical IKK complex was greatly attenuated. In addition, we tested two poorly understood domains within N4BP1 — the di-KH and UBM-like domains^37,38 — for binding to the canonical and non-canonical IKKs and linear polyubiquitin. Deletion of the N4BP1 di-KH domain did not affect N4BP1 binding to linear polyubiquitin, TANK, IKKε, or the canonical IKK complex. In contrast, deletion of the UBM-like domain in N4BP1 diminished its interaction with TANK, IKKε, and the canonical IKK complex, while binding to linear polyubiquitin was retained (Fig S6C). These data are consistent with a model whereby N4BP1 uses its CUE-like domain to bind linear polyubiquitin and mediates a UBM-like domain-dependent association with canonical and noncanonical IKKs, facilitating crosstalk between these kinases to limit late phase inflammatory gene expression downstream of TRIF-independent TLRs.

DISCUSSION

Our findings deconvolute the roles of N4BP1, TANK and the ncIKKs in controlling inflammatory gene expression instigated by MyD88-dependent TLRs. In humans, homozygous loss-of-function mutations in TBK1 were recently reported to cause an autoinflammatory syndrome without an apparent increase in viral disease susceptibility²⁴. TBK1 and IKKs may act redundantly in anti-viral IFN-I production, whereas TBK1 may be more crucial than IKKε for suppressing proinflammatory RIPK1-dependent cell death. Our findings highlight yet a third immunoregulatory function of the ncIKKs, namely N4BP1-dependent dampening of canonical IKK activity. Precisely how N4BP1 allows ncIKKs to target inhibitory phosphorylation sites on the canonical IKK complex warrants further study.

N4BP1 was previously reported to bind to NEMO¹⁴, and this was proposed to impede NEMO oligomerization and thereby inhibit canonical IKK activation. This study employed thioglycolate-elicited peritoneal macrophages from N4bp1-deficient mice, which differ from the BMDMs used in our studies¹³ in that they express Il6 transcripts and exhibit NF-κB signaling without additional stimulation. It is possible that the proinflammatory effects of thioglycolate are exacerbated in N4bp1-deficient mice, and therefore the differences observed are context-dependent. Although we cannot exclude a subtle effect of N4BP1 on initial IKK activation, we find that the more prominent effect of N4BP1 in TRIF-independent TLR signaling lies in controlling crosstalk between canonical and non-canonical IKK complexes during late phases of the response.

The conventional view of the ncIKKs is that they primarily induce IFN-I downstream of immune receptors that utilize the adaptors TRIF (TLR3 and TLR4), MAVS (RIG-I and MDA5) or STING (cGAS). Using these adaptor molecules as platforms during innate immune responses, the ncIKKs phosphorylate and thereby activate the IFN-I-inducing transcription factors, IRF3 and IRF7^18,39-45. However, in BMDMs, the larger class of TRIF-independent TLRs also induces ncIKK activation via the adaptor MyD88, yet downstream phosphorylation of IRF3 does not generally occur during these responses²⁹. Moreover, MyD88-mediated TLR responses are not thought to directly trigger RIPK1 and RIPK3/caspase-8-dependent cell death⁴⁶. These observations raise the possibility that MyD88-mediated ncIKK activation might serve a distinct function independent of IRF3/7-dependent IFN-I production and RIPK1-driven cell death. However, elucidating such a function has been difficult due to the embryonic lethality of Tbk1^−/− mice and the genetic redundancy among ncIKKs uncovered herein. Our findings shed light on this issue, highlighting a potent immunosuppressive function of the ncIKKs, coordinated by N4BP1, that constrains the proinflammatory potential of MyD88-dependent TLR responses.

An important question raised by our studies is why the ncIKKs evolved the dual ability to induce IFN-I downstream of certain receptors (e.g. the TRIF-TLRs, cGAS/STING, RIG-I, and MDA5), while suppressing proinflammatory cytokine responses and cell death downstream of other receptors (e.g. the TRIF-independent TLRs and TNFR1, respectively). We hypothesize that these intertwined functions of the ncIKKs ‘guard’ against pathogens seeking to evade IFN-I-mediated immunity. In plant immunity, the ‘guard hypothesis’ holds that perturbation of immune pathways is monitored and defended against by reserve pathways in case initial host defenses are breached^47-49. Strategically positioned downstream of IFN-I-inducing receptors, death receptors, and the MyD88-dependent TLRs, the ncIKKs may act as a central ‘guarding’ hub. When the ncIKKs are inhibited by pathogens, both cell death and proinflammatory cytokine responses can be unleashed to extinguish the infection. It should be noted that this hypothesis does not currently have support in our data and may not be mutually exclusive with other non-guarding roles for the cytokine suppressive activities of N4BP1 and the ncIKKs.

Limitations of the study

In this study, we provide genetic evidence for a pathway by which N4BP1, TANK, and the ncIKKs curtail persistent canonical IKK activity to limit inflammatory cytokine responses. However, the detailed mechanisms by which these proteins accomplish this function remain unknown. For example, how N4BP1 facilitates the ability of the ncIKKs to negatively regulate the canonical IKKs will require further study. Moreover, our experiments focused on murine macrophages. Whether these proteins act to similarly control inflammatory responses by this mechanism in other cell types and in a manner conserved across different species is unknown. Finally, direct evidence is required to substantiate the idea that perturbation of ncIKK-mediated interferon responses by pathogens is guarded by the proinflammatory cytokine responses held in check by the ncIKKs.

STAR METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Alexander D. Gitlin (gitlina@mskcc.org).

Materials availability

All unique reagents generated in this study are available upon request of the lead author and completion of a materials transfer agreement with Genentech.

Data and code availability

ChIP-seq, RNA-seq and scRNA-seq data have been deposited to GEO (accession # GSE218958).

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

The Genentech or Memorial Sloan Kettering institutional animal care and use committees responsible for ethical compliance approved all animal protocols. N4bp1^−/−13, Tnf^−/− (Jackson laboratory strain # 003008)⁵⁰, Ripk1^{D138N/D138N 51}, Ripk3^{−/− 52}, R26^{CreERT2/+ 53}, eGFP⁺ (Jackson laboratory strain # 006567)⁵⁴ and Casp8^{−/− 51} mice have been previously described. The N4bp1^D621A and N4bp1^ΔUb alleles were generated using established CRISPR methodology and electroporation of Cas9 HiFi (IDT) in complex with sgRNA, along with a single-stranded donor template into C57BL/6N zygotes.

To generate the N4bp1^D621A allele, the following sgRNA was used: 5'-CUGGAAUUAAAAAACGAACC-3'. For the N4bp1^ΔUb allele, the following sgRNA was used: 5'-CAGCGCCUCCCUCAGCUCGC-3'. For both alleles, a long single-stranded oligo (IDT Ultramer) with a number of silent mutations to prevent re-cutting in addition to the Alanine-encoding mutation(s) was used as template. For N4bp1^D621A, the resulting modified genomic sequence, corresponding to mm10 chr8:86,860,443-86,860,493 (reverse strand) is: 5'- TTA AAg AAt GAg CCc GGg CGA GCc GAT cTG AAG CAc ATT GTg ATA Gcc GGA-3'. For the N4bp1^ΔUb allele, the resulting modified genomic sequence, corresponding to mm10 chr8:86,844,777-86,844,818 (reverse strand) is: 5'- ACC tcC GAa CTc Aga GAG GCG CTG CTg AAG gcC GcC gCc GAC. Wildtype nucleotides in uppercase, introduced mutations in lowercase, Alanine-encoding sequence(s) are underlined. Resulting mosaic founders were screened for predicted off-targets, backcrossed to C57BL/6N, and off-target-free N1 heterozygous mice with the desired mutation(s) were bred with C57BL/6N to establish a breeding colony.

To generate Tbk1^cKD/+ mice, a targeting vector was designed to replace 230 bp of genomic Tbk1 sequence corresponding to the 3’ end of intron 4 and the beginning of exon 5, mm10 Chr. 10:121,571,978-121,572,208 (reverse stand). This sequence was replaced with a loxP-flanked cassette containing the 3’ end of a CMV intron, a cDNA consisting of Tbk1 exons 5-21, a 4x polyadenylation signal, and an FRT-flanked neomycin selection marker followed by a modified exon 5 containing a sequence encoding the D135N mutation that replaced 5’ end of the endogenous exon 5. This vector was confirmed by DNA sequencing, linearized with Notl and used to target C57BL/6 C2 embryonic stem (ES) cells. Correctly targeted clones were transfected with a Flpe-encoding plasmid to remove the neomycin selection cassette and the resulting ES cells were injected into blastocysts. Germline transmission was obtained after crossing resulting chimaeras with C57BL/6N females and the mice were subsequently maintained on a C57BL/6N background. In the Pam3csk4 sepsis model, mice were injected intraperitoneally with 30 mg/kg of Vaccigrade Pam3csk4 (Invivogen) in PBS as described previously¹³. Hematology assays were run on a Sysmex XT-2000iV automated hematology analyzer.

Cell culture

HEK293T cells (ATCC CRL-3216; tested for mycoplasma contamination but not authenticated) were cultured in the high glucose version of DMEM supplemented with 10% heat-inactivated fetal bovine serum, 2 mM glutamine, 1× non-essential amino acids solution, 100 U/ml penicillin and 100 μg/ml streptomycin. Cells were transfected for 24-72 hours with plasmids encoding wild-type or indicated mutant N4BP1 constructs containing N-terminal 3×Flag tags, HIV-1 (pNL4-3), Myc-IKKα, Myc-IKKβ, Myc-NEMO, Myc-TANK and Myc-IKKε. A plasmid co-expressing HOIP and HOIL-1 was previously described⁵⁵. Plasmids were transfected using Lipofectamine 2000 or 3000 (Thermo Fisher Scientific) or Fugene (ProMega) according to manufacturer’s protocol.

Primary BMDMs were differentiated from bone marrow for 7 days in Roswell Park Memorial Institute (RPMI) medium supplemented with 10% heat-inactivated fetal bovine serum, 2 mM glutamine, 50 μM 2-mercaptoethanol (Gibco), 100 U/ml penicillin, 100 μg/ml streptomycin (Gibco) and 50 ng/ml of recombinant M-CSF (R&D Systems).

CRISPR-mediated deletion of N4BP1, TANK, TBK1, IKKε, p100, RelB, p105, and p65 in mouse BMDMs of indicated genotypes was accomplished by Cas9 ribonucleoprotein electroporation as previously described^13,56,57. In brief, BMDMs were differentiated for 5 days in the above-mentioned RPMI medium on non-tissue culture treated plates. After collecting and washing BMDMs with PBS, 5 × 10⁶ BMDMs were electroporated with recombinant Cas9 protein v.3 (IDT) complexed with two target-specific gRNAs per intended gene deletion. The seed sequence/alt-R crRNA sequences used were as follows:

Negative control (gRNA 1): 5’-GGTGGTGCAGATGAACTTCA-3’

Negative control (gRNA 2): 5’-GGCATCGACTTCAAGGAGGA-3’

ΔN4bp1 (gRNA 1): 5’-GCTGCAACTCCGCGGCGCGC-3’

ΔN4bp1 (gRNA 2): 5’-AGCAGTAACGGTTCTGTAGA-3’

ΔTank (gRNA 1): 5’-CTGGAAAAGAATCCGCCAA-3’

ΔTank (gRNA 2): 5’-CAAAGAATACGCGAGCAAC-3’

ΔTbk1 (gRNA 1): 5’-TTGAACATCCACTGGGCGA-3’

ΔTbk1 (gRNA 2): 5’-AGGAGCCGTCCAATGCGTA-3’

ΔIkke (gRNA 1): 5’-GACAATGCCATTCTCCCGC-3’

ΔIkke (gRNA 2): 5’-AGCTATCGGCGACCTCCTG-3’

ΔNfkb2 (gRNA 1): 5’-GGATCCTTAGGCTCCACGA-3’

ΔNfkb2 (gRNA 2): 5’-CAGCGGGTTCCGTGCGCGG-3’

ΔRelb (gRNA 1): 5’-GCTTCCGCTACGAGTGCGA-3’

ΔRelb (gRNA 2): 5’-CTGAGTGAGATGCCGCGCC-3’

ΔNfkb1 (gRNA 1): 5’-TCTCGGACAGCTTCGGTGG-3’

ΔNfkb1 (gRNA 2): 5’-GAGTCACGAAATCCAACGC-3’

ΔRela (gRNA 1): 5’-GATTCCGCTATAAATGCGAG-3’

ΔRela (gRNA 2): 5’-ATCGAACAGCCGAAGCAACG-3’

The above-listed cRNAs were annealed to Alt-R tracrRNA (IDT). Annealed gRNAs were complexed with Cas9 and electroporated into BMDMs using nucleofector solution P3 (Lonza). BMDMs were then maintained in above-described RPMI medium containing M-CSF for the next 7 days.

BMDMs were treated as indicated throughout the text and figures with poly(I:C) (2 μg/ml, Invivogen), LPS (from Escherichia coli K12, Invivogen; 100 ng/ml), R837 (2 μg/ml or as indicated, Invivogen), Pam3csk4 (1 μg/ml, Invivogen), CpG-B (ODN 1826, 5 μM, Invivogen), 2'3'-cGAMP (100 μg/ml, Invivogen), TNF (100 ng/ml, Genentech), Actinomycin D (5 μm/ml, Sigma), or TPCA-1 (Selleckchem, 5 μM). For cytokine secretion assays, stimulation of BMDMs with TLR agonists was performed in above-described DMEM with indicated stimuli in 96-well tissue culture plates with 40,000-100,000 cells per well. Cytokines were measured from sera or cell culture supernatants using Milliplex MAP Mouse Cytokine/Chemokine Magnetic Bead Panel – Premixed 32-plex – Immunology Multiplex Assay (Millipore) on a Luminex Flex-Map 3D. Interferon-beta was measured from cell culture supernatants using the LumiKine Xpress mIFN-β 2.0 according to manufacturer’s instructions. HIV-1 P24 was measured using a SimpleStep ELISA kit (Abcam, ab218268) according to manufacturer’s instructions.

For the Human THP-1 monocyte cell line experiment, THP-1 cells were cultured in Roswell Park Memorial Institute (RPMI) medium supplemented with 10% heat-inactivated fetal bovine serum, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (Gibco). CRISPR-mediated deletion of TBK1 and IKKE was accomplished by Cas9 Neon^™ electroporation. In brief, after washing THP-1 cells with PBS, 100,000 cells were resuspended in 10 uL R buffer and mixed with Alt-R CRISPR-Cas9 (IDT) protein complexed with the following multi-guide sgRNAs (Synthego):

Negative Control, scrambled sgRNA (mod) #1 (Synthego)

Negative Control, scrambled sgRNA (mod) #2 (Synthego)

ΔTBK1 (guide 1): 5’-UGCCAUUUAGACCCUUUGA-3’

ΔTBK1 (guide 2): 5’-CAGUUGAUCUUUGGAGCAUU-3’

ΔTBK1 (guide 3): 5’-ACUGCUCUCUCAUACAUAUC-3’

ΔIKKE (guide 1): 5’-CUCGCGGGGCCGCAGGUAGC-3’

ΔIKKE (guide 2): 5’-AGAGCUUGACAAUGUUCUGG-3’

ΔIKKE (guide 3): 5’-AUGGCAGAAAUCCGGAGAGC-3’

Cells were eletroporated using the Neon^™ Transfection System under the following conditions: Voltage 1400, Width 10, Pulse 3. After electroporation, cells were immediately plated in 0.5 mL pre-warmed RPMI media without antibiotics. Cells were cultured for 7 days and then treated with Pam3csk4 (1 μg/ml, Invivogen) for 24 hours. Stimulation was performed in 96-well tissue culture plates containing 20,000 cells per well. Cytokine concentrations were measured from cell culture supernatants using Milliplex MAP reagents (MilliporeSigma, St. Louis, MO) according to manufacturer’s recommended protocol.

METHOD DETAILS

Flow cytometry

Flow cytometry was performed as previously described¹³. In brief, after preparing single-cell suspensions from organs, cells were incubated with anti-CD16/CD32 antibody (clone 93, eBioscience), and subsequently stained in PBS containing 2% heat-inactivated fetal bovine serum and 1 mM EDTA at 4 °C. The following fluorophore-conjugated antibodies were used: CD4 (RM4-5, eBioscience), CD8α (53-6.7, eBioscience), CD3ε (145-2C11, BD Biosciences), TCRβ (H57-597, BD Biosciences), B220 (RA3-6B2, BD Biosciences), CD44 (IM7, Biolegend), CD62L (MEL-14, Biolegend), CD19 (1D3, BD Biosciences), CD38 (90, eBioscience) and FAS (Jo2, BD Biosciences). Antibodies were used for flow cytometry at 1:400 dilution. CD4⁺ T cells were gated as: Live/Singlets/B220⁻/TCRβ⁺/CD4⁺/CD8⁻. CD8⁺ T cells were gated as: Live/Singlets/B220⁻/TCRβ⁺/CD4⁻/CD8⁺. For CD4⁺ and CD8⁺ T cells, the fraction of CD44⁺CD62L⁻ and CD44⁺, respectively, were further calculated. Germinal center B cells were gated as: Live/Singlets/B220⁺/CD3ε⁻/CD38⁻/FAS⁺. Samples were acquired with BD FACSDiva software v.8.0 on FACSCantoll (BD Biosciences) and analysed with FlowJo software version 9.9.6.

Western blotting and immunoprecipitation

Cells were lysed in 20 mM Tris HCl pH 7.4, 135 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 1% (v/v) Triton X-100, 10% (v/v) glycerol containing 100x Halt protease/phosphatase inhibitor cocktail (Thermo Fisher). Insoluble material was removed by centrifugation at 20,000 g before addition of LDS sample buffer followed by incubation at 70 °C for at least 10 min. For immunoprecipitation experiments, samples were immunoprecipitated in the same lysis buffer as above using ~1 mg of antibody per 1 mg of lysate using Pierce Protein A/G Agarose (Thermo Fisher) or Pierce Protein A/G Magnetic Beads (Thermo Fisher) by rotation at 4 °C for 4 hours or overnight. Alternatively, where indicated, magnetic anti-Flag beads (M2, Sigma Aldrich) were used for immunoprecipitation. Samples were washed in lysis buffer three times before elution in LDS sample buffer and analysis by immunoblotting. Immunoprecipitation of the canonical IKK complex was performed using antibodies targeting NEMO (AF4365, R&D Systems) and IKKβ (AF4535, R&D Systems). For the experiment in Figure S4, transfected cells were lysed as described above with the addition of 10 mM N-ethylmaleimide (Sigma Aldrich) in lysis buffer. Immunoprecipitation of flag-tagged N4BP1 constructs was performed using anti-Flag beads (M2, Sigma Aldrich) by rotation at 4 °C for 2 hours, after which samples were washed four times with lysis buffer. Subsequently, 500 μL of lysis buffer containing 0.5 μg of linear tetra-ubiquitin (UC-710b-025, Boston Biochem) was added to washed beads and further incubated by rotation at 4 °C for 2 hours. Samples were then washed in lysis buffer four further times and then eluted in LDS sample buffer for immunoblot analysis.

Western blotting antibodies included: actin (C4, MP Biomedicals), N4BP1 (EPNCIR118, Abcam and 717, Genentech), p-IKKα/β (C84E11, Cell Signaling Technologies), IKKβ (D30C6, Cell Signaling Technologies), IKKα (Y463 ab32041, Abcam), p-IκBα (14D4, Cell Signaling Technologies), NEMO (EPR16629 ab178872, Abcam), p-IKKβ S740 (PAB15911, Abnova), TBK1 (D1B4, Cell Signaling Technologies), p-TBK1 (D52C2, Cell Signaling Technologies), IKKε (D61F9 or D20G4, Cell Signaling Technologies), p-IKKε (D1B7, Cell Signaling Technologies), Vinculin (E1E9V, Cell Signaling Technologies), TANK (R&D Systems, AF4755), HOIP (11D6H2G5, Genentech), HOIL-1 (2E2, Millipore), linear ubiquitin (1F11/3F5/Y102L, Genentech), Caspase-8 (1G12, cat. no. ALX-804-447-C100, Enzo Life Sciences), RelB (Abcam, ab309084), p100/52 (4882, Cell Signaling Technologies), p105/50 (D7H5M, Cell Signaling Technologies), p65 (D14E12, Cell Signaling Technologies), and Jackson Immunoresearch (HRP-anti-mouse, cat. no. 115-035-174; HRP-anti-rabbit, cat. no. 211-032-171; HRP-anti-goat, cat. no. 205-032-176). For western blots, antibody for actin was used at a 1:10,000 dilution; antibody for N4BP1, was used at 1:4,000 dilution; secondary anti-mouse and ant-rabbit antibodies were used at 1:10,000 dilution; all other antibodies were used at 1:1,000 dilution.

RNA sequencing data analysis

GSNAP (version 2013-11-10)⁵⁸ was used to align raw FASTQ reads to the mouse reference genome (GRCm38/mm10) with following parameters “-M 2 -n 10 -B 2 -i 1 -N 1 -w 200000 -E 1 --pairmax-rna=200000 --clip-overlap”. Reads were filtered to include only the uniquely mapped reads. Limma⁵⁹ R package was used to perform differential expression analysis. For the four-way plot, genes were considered differentially expressed if the log2 fold change was either > 1 or < −1 and the adjusted p-value < 0.05.

Chromatin Immunoprecipitation (ChIP)-Seq

Samples were sent to Active Motif (Carlsbad, CA) for ChIP-Seq. Active Motif prepared chromatin, performed ChIP reactions, generated libraries, sequenced the libraries and performed basic data analysis. In brief, cells were fixed with 1% formaldehyde for 15 min and quenched with 0.125 M glycine. Chromatin was isolated by adding lysis buffer, followed by disruption with a Dounce homogenizer. Lysates were sonicated and the DNA sheared to an average length of 300-500 bp with Active Motif’s EpiShear probe sonicator (cat# 53051). Genomic DNA (Input) was prepared by treating aliquots of chromatin with RNase, proteinase K and heat for de-crosslinking, followed by SPRI beads clean up (Beckman Coulter) and quantitation by Clariostar (BMG Labtech). Extrapolation to the original chromatin volume allowed determination of the total chromatin yield.

An aliquot of chromatin (30 ug) was precleared with protein G agarose beads (Invitrogen). Genomic DNA regions of interest were isolated using 4 ug of antibody against Total RNA Pol II antibody AbFlex^® (Active Motif 91151). Complexes were washed, eluted from the beads with SDS buffer, and subjected to RNase and proteinase K treatment. Crosslinks were reversed by incubation overnight at 65 °C, and ChIP DNA was purified by phenolchloroform extraction and ethanol precipitation.

Quantitative PCR (QPCR) reactions were carried out in triplicate on specific genomic regions using SYBR Green Supermix (Bio-Rad). The resulting signals were normalized for primer efficiency by carrying out QPCR for each primer pair using Input DNA.

Illumina sequencing libraries (a custom type, using the same paired read adapter oligonucleotides described previously⁶⁰) were prepared from the ChIP and Input DNAs on an automated system (Apollo 342, Wafergen Biosystems/Takara). After a final PCR amplification step, the resulting DNA libraries were quantified and sequenced on Illumina’s NextSeq 500 (75 nucleotide reads, single end).

Raw FASTQ files were aligned to the mouse reference genome (GRCm38) using BWA⁶¹. Reads that mapped uniquely to the genome and aligned with no more than 2 mismatches were used for downstream analysis. MACS2 (v2.1.0)⁶² was used to call the peaks relative to the pooled input. Differential peak analysis was done using DiffBind v3.16 R package. GREAT v4.0.4⁶³ web application was used to perform gene ontology analysis on the peaks that were enriched in the N4bp1^−/− BMDMs relative to the N4bp1^+/+ BMDMs after 4 hours of R837 treatment. WashU epigenome browser⁶⁴ was used to look at locus specific ChIP-Seq read enrichment.

scRNA-Seq data analysis

10X 3’ single cell RNA-sequencing data was processed with cellranger (v3.1.0)⁶⁵. A gene by barcode matrix was generated by mapping reads to custom mouse reference genome (GRCm38) that includes the eGFP sequence.

Seurat (v4.0.4)⁶⁶ was used for downstream analysis. Cells with number of genes detected between 500 and 8000 for eGFP+ and N4bp1^+/+ samples, between 500 and 7500 for eGFP+ and N4bp1^−/− co-culture samples, between 500 and 6000 for eGFP⁺ and N4bp1^+/+ co-culture samples treated with R837, and between 500 and 5000 for eGFP⁺ and N4bp1^−/− co-culture samples treated with R837 were used for subsequent analysis. After filtering, we detected 36,174 cells across four samples. After log normalization, we performed dimensionality reduction using principal component analysis (PCA). We ranked the principal components (PCs) by the amount of variance explained and found that after 20 PCs the explained variance plateaus. Clusters were then identified using the first 20 PCs with resolution of 0.15. The same PCs were also used to generated UMAP projections. Violin plots for Il6, Csf3, Tnf, and Cxcl1 were generated using ggplot2 using the log transformed counts.

The N4BP1-regulated gene signature was derived from bulk RNA-seq data by comparing N4bp1^−/− samples with N4bp1^+/+ samples after 6 hour treatment with R837. The differentially expressed genes were further filtered to have log-fold change greater than 2.5 and adjusted p-value less than 0.05. Unpaired wilcoxon test was used to test for significance between eGFP⁺ and eGFP⁻ groups for each co-culture.

Protein production and crystal structure

The N4BP1 CUE-like domain (Q845-D893) was expressed as a His-tag fusion construct with a cleavable TEV linker in Escherichia coli overnight at 20 °C. Linear di-ubiquitin was expressed as a His-tag fusion construct with a Thrombin-cleavable linker in E. coli for 3 hours at 37 °C. The constructs were purified via gravity flow using Cobalt Resin. The His and the His-MBP tags were cleaved overnight during dialysis and removed by reverse gravity flow, followed by size exclusion chromatography equilibrated. The CUE-like domain and linear di-ubiquitin complex was formed by incubating both proteins at a ratio of 5:1 (CUE-like:linear di-ubibquitin) on ice for 30 minutes. The resulting complex was separated from unbound N4BP1 CUE-like domain by size exclusion chromatography pre-equilibrated in 25 mM Tris 7.5, 100 mM NaCl.

The complex was concentrated to ~76 mg/ml and used for crystallization screens. Crystals appeared after 2 days in 0.1M Tris pH 8.5, 30% (w/v) PEG 4K, 0.2 M LiSO₄. Crystals were harvested and diffraction data was collected at the Advanced Photon Source (APS) beamline 24-IDC. The crystal structure was determined by molecular replacement in Phaser using ubiquitin (pdb 1UBQ, residues 1-71) and KIAA0323 (pdb, 2N5M). The solution was built to 2.2 Å from the models, in multiple rounds of model building in Coot and refinement in PHENIX.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using Prism version 10 software (GraphPad). Data are represented by symbols representing biological replicates and bar graphs or lines representing mean values, unless otherwise indicated. Error bars represent standard error of the mean or standard deviation, as indicated. P-values < 0.05 were considered statistically significant, and no statistical tests were performed to predetermine sample sizes. Statistical analyses of cytokine secretion or immune cell subset analyses were performed by unpaired two-sided t-test, one-way ANOVA with Tukey’s multiple comparisons test, or Dunnett’s multiple comparisons test, as indicated in figure legends.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-Actin	MP Biomedicals	C4; RRID: AB_2335127
Anti-N4BP1	Abcam	EPNCIR118; ab133610
Anti-p-IKKα/β	Cell Signaling Technologies	C84E11; RRID: AB_2079379
Anti-IKKβ	Cell Signaling Technologies	D30C6; RRID: AB_11024092
Anti-IKKα	Abcam	Y463; ab32041; RRID: AB_733070
Anti-p-IκBα	Cell Signaling Technologies	14D4; RRID: AB_561111
Anti-NEMO	Abcam	EPR16629; ab178872; RRID: AB_2847887
Anti-p-IKKβ S740	Abnova	PAB15911; RRID: AB_10696238
Anti-TBK1	Cell Signaling Technologies	D1B4; RRID: AB_2255663
Anti-p-TBK1	Cell Signaling Technologies	D52C2; RRID: AB_10693472
Anti-IKKε	Cell Signaling Technologies	D61F9; RRID: AB_1264180
Anti-IKKε	Cell Signaling Technologies	D20G4; RRID: AB_1147662
Anti-p-IKKε	Cell Signaling Technologies	D1B7; RRID: AB_2737061
Anti-Vinculin	Cell Signaling Technologies	E1E9V; RRID: AB_2728768
Anti-TANK	R&D Systems	AF4755; RRID: AB_2199741
Anti-HOIP	Genentech	11D6H2G5
Anti-N4BP1	Genentech; this paper	Clone 717
Anti-HOIL-1	Millipore	2E2; RRID: AB_2737058
Anti-linear ubiquitin	Genentech	1F11/3F5/Y102L
Anti-Caspase-8	Enzo Life Sciences	1G12, Cat #ALX-804-447-C100; RRID: AB_2050952
Anti-RelB	Abcam	EPR26077-23; ab309084
Anti-p100/52	Cell Signaling Technologies	Cat #4882; RRID: AB_10695537
Anti-p105/50	Cell Signaling Technologies	D7H5M; RRID: AB_2687614
Anti-p65	Cell Signaling Technologies	D14E12; RRID: AB_10859369
HRP-anti-mouse	Jackson Immunoresearch	Cat #115-035-174; RRID: AB_2338512
HRP-anti-rabbit	Jackson Immunoresearch	Cat #211-032-171; RRID: AB_2339149
HRP-anti-goat	Jackson Immunoresearch	Cat #205-032-176; RRID: AB_2339056
Anti-NEMO	R&D Systems	AF4365; RRID: AB_2233641
Anti-IKKβ	R&D Systems	AF4535; RRID: AB_2122305
Anti-CD4-eFluor450	eBioscience	RM4-5; Cat #48-0042-82; RRID: AB_1272194
Anti-CD4-PE	BD Biosciences	H129.19; Cat #553653; RRID: AB_394973
Anti-CD8α-FITC	eBioscience	53-6.7; Cat #11-0081-82; RRID: AB_464915
Anti-CD3ε-APC-Cy7	BD Biosciences	145-2C11; Cat #557596; RRID: AB_396759
Anti-TCRβ-PE-Cy7	BD Biosciences	H57-597; Cat #560729; RRID: AB_1937310
Anti-TCRβ-APC	Biolegend	H57-597; Cat #109212; RRID: AB_313435
Anti-B220-PE	BD Biosciences	RA3-6B2; Cat #553090 and 561878; RRID: AB_394619
Anti-B220-V500	BD Biosciences	RA3-6B2; Cat #561226; RRID: AB_10563910
Anti-CD44-APC-Cy7	Biolegend	IM7; Cat #103028; RRID: AB_830785
Anti-CD44-FITC	BD Biosciences	IM7; Cat #553133; RRID: AB_2076224
Anti-CD62L-BV510	Biolegend	MEL-14; Cat #104441; RRID: AB_2561537
Anti-CD62L-PE-Cy7	BD Biosciences	MEL-14; Cat #560516; RRID: AB_1645257
Anti-CD19-PE	BD Biosciences	1D3; Cat #557399; RRID: AB_395050
Anti-CD38-PE-cy7	eBioscience	90; Cat #25-0381-82; RRID: AB_2573344
Anti-CD38-BV421	BD Biosciences	90; Cat #562768; RRID: AB_2737781
Anti- CD8α-BV421	BD Biosciences	53-6.7; Cat #563898; RRID: AB_2738474
Anti-FAS-FITC	BD Biosciences	Jo2; Cat #554257; RRID: AB_395329
Anti-CD16/CD32	eBioscience	clone 93; Cat #14-0161-82; RRID: AB_467133
Anti-RNA Pol II antibody AbFlex®	Active Motif	91151; RRID: AB_2793789
Anti-Flag M2-HRP	Sigma-Aldrich	Cat #A8592; RRID: AB_439702
Bacterial and virus strains
N/A	N/A	N/A
Biological samples
N/A	N/A	N/A
Chemicals, peptides, and recombinant proteins
R837	Invivogen	Cat #tlrl-imq-10
Actinomycin D	Sigma	Cat #SBR00013; CAS: 50-76-0
Pam3csk4	Invivogen	Cat #tlrl-pms
Vaccigrade Pam3csk4 (for Fig. 7E)	Invivogen	Cat #vac-pms
CpG-B ODN 1826	Invivogen	Cat #: tlrl-1826
Lipopolysaccharide, Escherichia coli K12	Invivogen	Cat #tlrl-peklps
Recombinant Cas9	IDT	Cat#10000735
Poly(I:C)	Invivogen	Cat #tlrl-pic
2’3’-cGAMP	Invivogen	Cat #tlrl-nacga23-02
Linear tetra-ubiquitin	Boston Biochem	Cat #UC-710b-025
TPCA-1	Selleck Chemicals	Cat #S2824; CAS: 507475-17-4
Critical commercial assays
HIV-1 P24, SimpleStep ELISA kit	Abcam	ab218268
MILLIPLEX MAP Mouse Cytokine/Chemokine Magnetic Bead Panel - Premixed 32 Plex	Millipore	MCYTMAG70PMX3 2BK
MILLIPLEX Human Cytokine/Chemokine/Growth Factor Panel A 38 Plex Premixed Magnetic Bead Panel - Immunology Multiplex Assay	Millipore	HCYTA-60K-PX38
Lipofectamine 2000	Thermo Fisher Scientific	Cat #11668027
Lipofectamine 3000	Thermo Fisher Scientific	Cat #L3000008
LumiKine Xpress mIFN-b 2.0	Invivogen	Cat #luex-mifnbv2
Deposited data
ChIP-Seq	This paper	GSE218958
Bulk RNA-Seq	This paper	GSE218958
scRNA-Seq	This paper	GSE218958
Structure of the N4BP1 CUE-like domain in complex with linear di-ubiquitin	This paper	PDB ID: 8T48
Experimental models: Cell lines
HEK293T	ATCC	CRL-3216
THP1 monocytes	ATCC	TIB-202 (Lot 58999776)
Experimental models: Organisms/strains
N4bp1−/−	Gitlin et al.13	MGI: 6490626
Tnf−/−: B6; 129S-Tnftm1Gkl/J	The Jackson Laboratories50	RRID: IMSR_JAX:003008
Ripk3−/−	Newton et al.52	MGI: 3028853
Casp8−/−	Newton et al.51	MGI: 5614580
R26CreERT2/+	Siebler et al.53	MGI: 5551325
eGFP+: C57BL/6-Tg(CAG-EGFP)131Osb/LeySopJ	The Jackson Laboratories54	RRID: IMSR_JAX:006567
N4bp1 ΔUb/ΔUb	This paper
Ripk1D138N/D138N	Newton et al.51	MGI: 5614572
N4bp1D621A/D621A	This paper
Oligonucleotides
Sequences for gRNAs, See Table S2.
Recombinant DNA
HIV-1	BEI Resources	pNL4-3
Myc-IKKα	This paper	N/A
Myc-IKKβ	This paper	N/A
Myc-NEMO	This paper	N/A
Myc-TANK	This paper	N/A
Myc-IKKε	This paper	N/A
HOIP and HOIL-1	Genentech55	N/A
3xFlag-hN4BP1 (WT)	This paper	N/A
3xFlag-hN4BP1 (D623A)	This paper	N/A
3xFlag-mN4BP1 (WT)	This paper	N/A
3xFlag-mN4BP1 (D621A)	This paper	N/A
3xFlag-mN4BP1 (ΔUb)	This paper	N/A
3xFlag-mN4BP1 Δdi-KH (Δ2-148)	This paper	N/A
3xFlag-mN4BP1 ΔUBM (Δ158-202)	This paper	N/A
3xFlag-mN4BP1 (Δ341-389)	This paper	N/A
3xFlag-mN4BP1 (Δ619-750)	This paper	N/A
3xFlag-mN4BP1 (Δ844-893)	This paper	N/A
N4BP1 CUE-like domain (Q845-D893; His-tag fusion with cleavable TEV linker)	This paper	N/A
Linear di-ubiquitin (His-tag fusion with thrombin cleavable linker)	This paper	N/A
Software and algorithms
Prism	GraphPad	Version 10.1.1
Flowjo	BD Biosciences	V10
Other
N/A	N/A	N/A

Highlights.

• N4BP1 curtails late-phase inflammatory gene transcription during TLR signaling

• Ubiquitin-binding by N4BP1 is critical to suppress inflammatory cytokine responses

• N4BP1 acts in concert with the non-canonical IKKs to limit inflammation