Procedures on animals were performed according to the European Community and Animal Research Ethical Committee guidelines and were approved by the IIS-FJD Animal Research Ethical Committee and the Consejería de Medio Ambiente y Ordenación del Territorio, Comunidad de Madrid (PROEX 038/19).

The study with human samples complied with ethical precepts formulated in Order SAS 3470/2009 and the Declaration of Helsinki of the World Medical Association on ethical principles for medical research and were approved by the institutional Research Ethical Committee (PIC026-19-FJD). Samples of patients were requested through written informed consent and collected under a biobank regimen.

Murine MCT cells are a well-characterized cell model of kidney tubular epithelium suitable to study molecular mechanisms of kidney injury (Haverty et al., 1988). MCT cells were grown in RPMI 1640 (GIBCO, Grand Island, NY) supplemented with 10% decomplemented fetal bovine serum (DFBS), 2 mM glutamine, 100 U/mL penicillin, and 10 mg/ml streptomycin, in 5% CO₂ at 37°C. For experiments, cells were stimulated with 100 ng/ml human TWEAK; 0.01 to 100 mUI/ml IFNα and IFNβ (R&D Systems Inc.); the cytokine mixture made of 100 ng/ml human TWEAK, 30 ng/ml TNFα, and 30 U/ml interferon-γ (IFNγ, PeproTech) or LPS (100 ng/ml, Merck). The following chemical inhibitors were used: 10 μM PF-06700,841 tosylate salt (Sigma-Aldrich, Merck); 50 μM amlexanox; 10 μM Parthenolide, and 2.5 μM IKK16 (MedChemExpress). All the inhibitors were added to cultured cells 1 h before the stimuli. Stock solutions of the stimuli and inhibitors were made according to the specified manufacturers’ instructions. Cells were also treated with the IFNAR neutralizing antibody for 3 h before the stimuli.

Gene transcription was analyzed through quantitative reverse transcription PCR (qRT-qPCR) by using predesigned gene expression assays (TaqMan^®, Applied Biosystems-Termofisher Scientific, Waltham, MA, United States). Proteins were assessed by western blot, ELISA, and immunocytochemical-immunofluorescence assays. Standard procedures were applied.

Samples were homogenized in lysis buffer (50 mmol/L Tris, 150 mmol/L NaCl, 2 mmol/L EDTA, 2 mmol/L EGTA, 0.2% Triton X-100, 0.3% NP-40, 0.1 mmol/L PMSF, 25 mmol/L NaF). Proteins were separated by 10% SDS-PAGE under reducing conditions, then blotted onto nitrocellulose membranes. Membrane blockade was accomplished with 5% defatted milk in TBS-T (0.05 mol/L Tris, 0.15 mol/L NaCl, 0.05% Tween 20, pH 7.8). Thereafter, membranes were overnight probed at 4°C with primary antibodies in the same blocking solution or 5% BSA in TBS-T and then incubated with secondary HRP-conjugated antibodies for 1 h at room temperature. The following primary antibodies were used to detect specific proteins of interest: rabbit polyclonal anti-p-STAT1 (Tyr701) (Invitrogen, 44-376G), p-TYK2 (pTyr1054) (Origene, TA333304), and p-IKKε (Ser172) (Sigma Aldrich, 06-1340); rabbit monoclonal anti-p-TBK1/NAK (S172) (D52C2) XP^® (Cell Signaling Technology, 1,675,483), TBK1/NAK (E8I3G) (Cell Signaling Technology, 38,066), IKKε (D61F9) XP^® (Cell Signalling Technology, 3416), pIRF3 (Ser396) (4D4G) (Cell Signalling Technology, 4,947) and IRF-3 (D83B9) (Cell Signalling Technology, 4,302); monoclonal mouse anti-pIKBα (Santa Cruz, sc-8404). Anti-α-Tubulin (Sigma-Aldrich, MAB374) and anti-GAPDH (Millipore, MAB374) were used to assess protein loading homogeneity.

Cells plated onto glass coverslips were fixed in 4% paraformaldehyde and permeabilized in 0.2% Triton X-100/PBS, washed in PBS, and overnight incubated with polyclonal rabbit anti-p-IRF3 (pSer396) antibody (1:50, Sigma-Aldrich, SAB4504031 or anti-p65 (1:100; Santa Cruz Biotechnology, sc-8008) followed by Alexa 488-conjugated secondary antibody (1:300; Invitrogen) (3). Nuclei were counterstained with DAPI.

Ccl5 expression levels in the supernatants of cultured cells subjected to proinflammatory stimulation were assessed by ELISA (DuoSet ELISA Kit, R&D Systems, Minneapolis, MN) according to the manufacturers’ instructions.

MCT cells were grown in six-well plates and transfected with a mixture of a set of three specific siRNA for TBK1 (75 nM), IKKε (75 nM), or IRF3 (40 nM) (Stealth RNAi™, Invitrogen-TermoFisher Scientific, MA) and Lipofectamine RNAiMAX Transfection Reagent (Invitrogen) made in Opti-MEM I Reduced Serum Medium. After 18 h, cells were washed and cultured for another 48 h in a complete medium containing 10% BSA, and finally serum-deprived for 24 h before stimulation with TWEAK, TTI or z/TTI to evaluate gene mRNA expression and cell death. A siRNA negative control (Stealth RNAi™, Invitrogen) with the same GC content of specific siRNA oligonucleotides was used as a control. After transfection, cells were stimulated at time points when protein expression was reduced by approximately 90%.

For assessment of the overall death rate, cells were washed with PBS following stimulation and then incubated with 0.5 mg/ml MTT (Sigma, Merck) for 1 h at 37°C to detect changes in the metabolic activity. After this step, the MTT solution was withdrawn, and cells were allowed to air dry. Finally, deposits of reduced MTT were dissolved with DMSO, and their absorbance was read at 570 nm. In vivo cell death was assessed by a TUNEL assay performed in 3 µm-thick sections of paraffin-embedded renal tissue (In Situ Cell Death Detection Kit, Fluorescein, Roche Applied Science), according to the manufacturer’s protocol. TUNEL positive cells were counted in 10 randomly chosen fields with a fluorescent microscope.

Models were developed in wild-type0-12-week-old C57BL/6 mice (Charles River Chatillon-Sur-Charlaronne, France) or in TLR4 ^−/− mice from the same genetic background (Dr S Akira’s laboratory, Osaka University, Japan and generously provided by Dr C. Guerri, CIPF, Valencia, Spain). Four different models of kidney inflammation and injury were assessed. 1) Systemic administration of IFNβ was conducted as previously published (Van-Holten et al., 2004) and dose (0.5 μg/mouse, IP) was established in dose-response test models evaluating renal inflammation. At the times chosen for the experiments (4 and 24 h), recombinant mouse IFNβ (R&D Systems, McKinley Place, MN) caused no tubular cell death or loss of renal function. 2) The TWEAK murine model was previously standardized in our laboratory. It displays renal inflammation originating from classical and alternative activation of NF-κB (Sanz et al., 2010a). Wild-type or TLR4 ^−/− mice were challenged with recombinant human TWEAK (Merck, Darmstadt, Germany) and renal tissue was analyzed after 24 h. Additionally, some mice also received a second intervention: parthenolide (MedChemExpress, Monmouth Junction, NJ) (70 μg/mouse, IP) or amlexanox (Biorbyt, Cambridge, UK) (50 mg/kg) to inhibit NF-κB or TBK1/IKKε kinases respectively before TWEAK dosage. 3) Endotoxemia-induced kidney inflammation was assessed 24 h after the administration of 5 mg/kg IP bacterial lipopolysaccharide (LPS, Sigma) (LPS nephropathy, LPSN). The dose was chosen on the basis that it may induce a feasible inflammatory reaction and injury in the kidney without hemodynamic compromise and mortality (Panzer et al., 2009; Hato et al., 2019). TBK1/IKKε was inhibited with 50 mg/kg (PO) amlexanox whereas inhibition of the TI-IFN pathway was accomplished with 0.5 mg/kg/day (IP) neutralizing anti-IFNAR antibody (MAR1-5A3, Leinco Technologies). Inhibitors were administered 2 h before LPS. Rat IgG1 (MAB005, R&D Systems) was used as isotype control. 4) Murine folic acid-induced nephrotoxicity (FAN) is characterized by tubular cell death, leukocyte infiltration, and subsequent tubular regeneration that has been reported in humans (Metz-Kurschel et al., 1990; Martin-Sanchez et al., 2018; Yan, 2021). FAN was induced by a single IP injection of 250 mg/kg folic acid and analyzed at 24-96 h (Martin-Sanchez et al., 2018; Yan, 2021). Inhibition with amlexanox or anti-IFNAR antibody was scheduled 2 h before and 24 after FA injection.

Mice were euthanized under anaesthesia with 35 mg/kg ketamine (Ketolar/Pfizer) and 5 mg/kg xylazine (Rompun/Bayer). Blood for serum analytical assessment was drawn from the saphena vein and collected on tubes coated with EDTA. Plasma was obtained by centrifugation (1500 rpm, 5 min). Kidneys were perfused in situ with cold saline before removal. One kidney was snap-frozen in liquid nitrogen for RNA and protein studies and the other fixed and paraffin-embedded for immunohistochemistry.

Urea plasma levels were assessed by biochemical methods intended for automatic measurements based on the enzymatic decomposition with urease, then followed by colorimetric detection of the reaction product (Roche/Hitachi cobas^® c701/702).

Transcriptomics arrays of kidney tissues from mice 24 h after folic acid or vehicle (n = 3/group) were previously published (González-Guerrero et al., 2018; Fontecha-Barriuso et al., 2019). Transcriptomics were performed at Unidad Genómica Moncloa, Fundación Parque Científico de Madrid, Madrid, Spain. Affymetrix microarray analysis followed the manufacturer’s protocol. Image files were initially obtained through Affymetrix GeneChip^® Command Console^® Software (AGCC) (Affymetrix, Thermo Fisher Scientific, Santa Clara, CA). Subsequently, robust multichip analysis (RMA) was performed using Affymetrix Expression Console^® Software Affymetrix, Thermo Fisher Scientific. Starting from the normalized RNA, a significance analysis of microarrays was performed using the limma package (Babelomics, www.babelomics.org), using a false discovery rate (FDR) of 5% to identify genes that were significantly differentially regulated between the analyzed groups. Canonical pathway enrichment analyses were performed using the public database Reactome (www.reactome.org) (supported by United States National Institutes of Health; Toronto University; European Union and the European Molecular Biology Laboratory) 34,788,843 (Gillespie et al., 2022) and Interferome (www.interferome.org) (Monash Institute of Medical Research, University of Cambridge) (Samarajiwa et al., 2009).

Paraffin-embedded sections were stained using standard histology procedures. Immunostaining was performed in 3 μm thick tissue sections that were deparaffinized and antigen retrieved using the PT Link system (Dako Diagnostics, Barcelona, Spain) with Sodium Citrate Buffer (10 mM) adjusted to pH 6-9, depending on the marker. For colorimetric immunohistochemistry, endogenous peroxidase was blocked and then sections were incubated overnight at 4°C with the following primary antibodies: polyclonal rabbit anti-human CD3 (ready to use; DAKO A0452), anti-human MPO (ready to use; DAKO IS511); monoclonal mouse anti-mouse T-bet/Tbx21 [4B10] (1:200; Abcam ab91109); monoclonal rat anti-mouse F4/80 (1:50, MCA497, Bio-Rad); polyclonal rabbit anti-human IFNAR2 (1:100, LS-B13369, LSBio); polyclonal rabbit anti-human IFN beta (1:400, PA5-20390, TermoFisher). Finally, sections were washed, stained with 3,3′-diaminobenzidine (DAB) as chromogen (Dako, Denmark), counterstained with Carazzi`s hematoxylin, dehydrated, and mounted in DPX medium (Merck). Fluorescent immunohistochemistry was developed in tissue sections that were first permeabilized with 0.2% Triton X-100 for 5 min, then blocked with 4% BSA/10% species-specific serum (the same animal species that animal source of primary antibodies), followed by sequential primary and secondary antibodies incubation for 90 m or 60 m, respectively, and finally, the nuclei stained with DAPI for 5 m. Primary antibodies used were: polyclonal rat anti-mouse F4/80 (1:50, MCA497, Bio-Rad); FLEX monoclonal mouse anti-human CD31 (ready to use, GA610, clone JC70A DAKO); polyclonal rabbit anti-human IFN beta (1:100, PA5-20390, TermoFisher). Primary antibody binding to specific antigens was revealed by using the following secondary antibodies (1:200, Invitrogen): Alexa fluor donkey anti-rat 488 (A21208), goat anti-mouse 633 (A21050), goat anti-rabbit 633 (A21070), and donkey anti-rabbit 488 (A21206).

Images were obtained by optical (BX53F2 model, Olympus Spain) or confocal (SP5, Leica Microsystems, Spain) microscopy and quantified with ImageProPlus software (Media Cybernetics, Bethesda, MD). Results are shown as the number of positive cells from 10 randomly chosen fields per kidney (20–4×0 objective) or as the percentage of stained area, considering the total area as 100%. Negative controls include non-specific immunoglobulin and no primary antibody.

Biopsies from AKI patients (n = 6) and healthy kidney tissue from nephrectomy specimens (n = 4) were obtained from the IIS-Fundación Jimenez Diaz Biobank (IIS-Fundación Jimenez Diaz, Madrid, Spain). Clinical characteristics of human samples intended for this study are provided in Supplementary Table S1.

Statistical analysis was performed using GraphPad Prism (Dotmatics, San Diego, CA), expressing results as sem ± sd. Significance (p < 0.05) was assessed by a non-parametric Mann-Whitney test for two independent samples.

Recent emphasis on the association of enhanced IFN-I signaling and kidney injury has focused on glomerular injury (Lodi et al., 2022), without considering associated tubulointerstitial injury. Pathway enrichment analysis in a previously reported transcriptomics dataset of murine folic acid nephropathy (FAN) (González-Guerrero et al., 2018; Fontecha-Barriuso et al., 2019) identified multiple activated upstream regulators of the TI-IFN pathway having an absolute z-score > 2.0 and p-value < 0.05 (Supplementary Table S2). Moreover, 168 of the 865 (19.4%) most upregulated genes (fold change ≥1.5 times; p and FDR <0.05) and 37 of the 333 (11.1%) most downregulated genes (fold change ≥0.5 times; p and FDR <0.05) were classified as interferon regulated genes (IRG) in the Interferome database, based on several datasets of mouse cells challenged with TI-IFNs. Subsequent bibliography-based expression profiling analysis also showed overexpression (p and FDR <0.05) of an ISG mRNA signature (Supplementary Table S3). Hence, kidney transcriptomics identified a direct activity of the TI-IFN pathway on ISG transcription or the propagation of an intricate signaling network resulting from the engagement of this pathway in murine tubulointerstitial injury.

The increased gene expression of some of these ISGs, IFNβ, and IFNAR1/2, taken as reporters of the TI-IFN pathway activation, were validated by PCR in mice with both FAN and LPS-induced nephropathy (LPSN) (Figures 1A–C). Moreover, in FAN and LPSN mice, increased renal expression of IFNAR2 was detected by immunohistochemistry in kidney tubules (Figure 1D). By contrast, expression of IFNβ, which remained undetectable in control mice, was markedly increased in LPSN and FAN mice in the tubulointerstitial space (Figure 1E) where its fluorescence signal heavily overlaps with the fluorescence signal of endothelial cells (CD31) but hardly with the fluorescence signal of mononuclear phagocytic cells (F4/80) (Figure 1F). In addition, immunohistochemical IFNAR2 staining increased in tubular cells in patients with tubular injury, supporting the clinical relevance of this finding (Figure 1G).

Whether TI-IFNs modulate inflammatory responses in tubular cells has not been firmly established. Thus, we next explored the impact of IFNβ on cultured murine proximal tubular MCT cells. IFNβ dose-dependently increased the mRNA expression of the canonical ISGs Oasl2 and Usp18, peaking at 6 h and decreasing following 24 h, suggesting TI-IFN pathway activation (Figure 2A). Accordingly, IFNβ activated signaling downstream TI-IFN formation through JAK1, TYK2 and STAT1 phosphorylation (Figure 2B) and the transcription of NF-κB target cytokines like Cxcl10 (also considered an ISG), Ccl2, and Il-6, with similar kinetics to canonical TI-IFN-responsive genes (Figure 2C). Supporting crosstalks between the TI-IFN and NF-κB pathways, IFNβ induced IκBα phosphorylation and the nuclear translocation of NF-κB/RELA (p65) (Figure 2D). IFNAR blockade with a specific anti-IFNAR1/2 neutralizing antibody (Figure 2E) or inhibiting TYK2/JAK1 with PF-06700841 (Figure 2F), dampened canonical ISG and cytokine transcription programs. IFNβ also induced phosphorylation of TBK1, IRF3 nuclear location (Figure 1G), and Ifna/Ifnb1 mRNA expression (Supplementary Figure S1A), reflecting a positive activation loop in which IFNβ promotes the expression of TI-IFNs. Inhibition of both TBK1 and IKKε by amlexanox (Reilly et al., 2013) did not modify Oasl2 mRNA levels, which is consistent with IFNAR signaling directly impacting the ISG response. However, amlexanox downregulated Cxcl10 and Ccl2 mRNA expression, without modifying Il6 mRNA levels (Figure 2H), suggesting that IFNAR signaling engages TBK1/IKKε to promote the expression of NF-κB-dependent genes (Figure 2I). Indeed, IFNβ interacts with other key proinflammatory pathways because pre-exposure to IFNβ potentiated TWEAK-induced cytokine transcription to levels above an additive response (Supplementary Figure S1B). IFNα also elicited the ISG and cytokine responses in cultured tubular cells (Supplementary Figure S1C). Consistently, the IFNAR agonist RO8191 (Konishi et al., 2012) mimicked the strong IFNα/β-induced ISG response and the positive activation loop over the NF-κB (Supplementary Figure S1D). Thus, both TI-IFNs share a proinflammatory impact on tubular cells.

In vivo, a single dose of IFNβ (Figure 2J) transiently (peak 4 h) increased the kidney expression of ISGs (still mildly increased at 24 h) and proinflammatory cytokines (back to baseline at 24 h) (Figure 2K) resulting in tubulointerstitial inflammation characterized by infiltration by F4/80 + mononuclear phagocytes and CD3⁺ T lymphocytes that persisted for at least 24 h (Figure 2L). Collectively, these data establish that TI-IFNs promote proinflammatory and antiviral activities in cultured tubular cells and in the kidney in vivo.

We next studied whether proinflammatory stimuli known to cause kidney injury activate the TI-IFN pathway in MCT cells. TWEAK and LPS phosphorylate/activate TBK1, IKKε (Figure 3A), and IRF3 (which translocates into nuclei) (Figure 3B). These results are consistent with previously uncharacterized TWEAK-dependent noncanonical IKKs recruitment, and corroborated TI-IFN production following canonical LPS ligation to TLR4 (Uematsu & Akira, 2007).

Furthermore, increases in the IFNβ secretion in TWEAK treatments and in transcription levels of IFNα/β and IFNAR1/IFNAR2 in TWEAK and LPS treatments were found (Figure 3C). TWEAK and LPS also stimulated STAT1 and TYK2 phosphorylation (Figure 3D), i.e. signaling downstream of IFNAR, and correspondingly increased ISG mRNA expression at 6 and 24 h (Figure 3E). The autocrine/paracrine recruitment of the TI-IFN pathway was confirmed by IFNAR blockade (Figure 3F) or TYK2/JAK1 inhibition with PF-06700841 (Supplementary Figure S2A), as both prevented TWEAK- or LPS-induced ISGs transcription (Oasl2) without modifying the cytokine response.

Pretreatment with the IKKα/β inhibitor parthenolide limited long-lasting (120 m) TWEAK-induced TBK1 and IKKe phosphorylation (Supplementary Figure S2B) and both parthenolide and a second IKK inhibitor, IKK16, decreased the expression of TBK1/IKKε-dependent ISG genes in cultured tubular cells stimulated with TWEAK and LPS (Figure 3G). These results are consistent with reports of IKKα/β-mediated activation of TBK1/IKKε by NF-κB agonists such as TNFα, IL1β, or LPS, in BMDM and MEFs (Clark et al., 2011).

Overall, inflammatory stimuli that cause kidney injury, such as TWEAK and LPS, recruit the canonical NF-κB pathway (IKKα/β) to activate TBK1/IKKε and IRF3-mediated autocrine/paracrine loop of TI-IFN on IFNAR, leading to ISG transcription.

Both TWEAK and LPS activated the TBK1/IKKε-dependent TI-IFN pathway in tubular cells. Therefore, we explored whether TWEAK recruited TLR4, the LPS receptor, to signal through TBK1/IKKε in TWEAK-treated WT and TLR4^−/− mice (Figure 4A). In WT mice, TWEAK administration upregulated the kidney mRNA expression of proinflammatory chemokines that are dependent on TBK1/IKKε signaling in cultured tubular cells (i.e., Cxcl10 and Ccl5) (see below, Figure 5B) and this response was milder in TLR4 ^−/− mice (Figure 4B). In cultured tubular cells, the TLR4 blocker CLI095 prevented the LPS-induced upregulation of cytokine and ISG gene expression, as expected for LPS binding and activation of TLR4 (Figure 4C). To further characterize the in vivo observations in TLR4 ^−/− mice, we next tested the direct impact of CLI095 on TWEAK-stimulated cultured tubular cells. TLR4 blockade with CLI095 prevented the TWEAK-induced nuclear accumulation of RELA/P65 (Figure 4D) and phosphorylated IRF3 (Figure 4E) and the corresponding increase in Ccl5 and Ifit1 mRNA (6 and 24 h) as well as the persistence of Cxcl10 mRNA upregulation (24 h), without modifying Ccl2 mRNA levels (Figure 4F). Collectively, these data show that TWEAK can transactivate TLR4 to initiate antiviral and NF-κB-mediated transcriptional programs modulated by TBK1/IKKε/IRF3 signaling.

In exploring the participation of TBK1 and IKKε in antiviral and inflammatory responses elicited by proinflammatory stimuli other than IFNβ in tubular cells, amlexanox decreased TWEAK-induced ISG (Figure 5A) and cytokine (Figure 5B) mRNA expression. Consistent with these findings, amlexanox reduced nuclear p-IRF3 levels and NF-κB/p65 location (Figures 5C,D) and decreased p-IκBα levels (Figure 5E). Amlexanox also diminished the LPS-induced gene expression of ISGs and Ccl5 with a time course like TWEAK (Figures 5A,B), whereas Cxcl10 and Ccl2 downregulation was observed by 24 h (Figure 5F). In TWEAK-injected mice (Figure 5G), amlexanox also reduced kidney ISG and cytokine mRNA, consistent with cell culture observations (Figure 5H).

In summary, inflammatory mediators such as TWEAK and LPS recruit TBK1/IKKε in tubular cells through pathways that include canonical IKKs and TLR4. Thereby, TBK1/IKKε represents a signaling node located at the crossroad between inflammation and the antiviral response (Figure 5I).

Finally, we dissected the functional relevance of each component of the TBK1/IKKε/IRF3 pathway for TWEAK-induced NF-κB and antiviral response activation. Gene silencing reduced TBK1, IKKε, and IRF3 protein levels in tubular cells (≈90% at 48 h) (Supplementary Figure S3A). After further culture for 6 h, TBK1 silencing decreased the basal and TWEAK-induced mRNA expression of ISGs (Figure 6A), and Ccl5 mRNA expression (Figure 6B) and secretion (Supplementary Figure S3B), but it did not modify Cxcl10, Ccl2, and Il-6 mRNA induced responses (Figure 6B). By contrast, together with Ccl5, they were suppressed at 24 h (Figure 6C), suggesting that recruitment of TBK1 promotes the persistence of TWEAK-induced inflammatory responses over time. Combining TBK1 silencing with amlexanox (Figure 6B) or the simultaneous silencing of TBK1 and IKKε thoroughly inhibited the TWEAK-induced mRNA expression of cytokines at 6 h (Figure 6D) suggesting that IKKε is required for early TWEAK-induced NF-κB activation after TBK1 engagement. However, IKKε silencing alone did not modify TWEAK-induced ISG or cytokine responses at 6 h (Figure 6E) or 24 h (Supplementary Figure S3C). Hence, coordinated TBK1 and IKKε activation is required for efficient activation of early TWEAK-induced inflammatory responses whereas only TBK1 is required to activate the ISG program.

Finally, like for TBK1 (Figure 6B), IRF3 silencing decreased basal and TWEAK-stimulated ISG mRNA expression (Figure 6F) but only the NF-κB-dependent Ccl5 mRNA and protein upregulation at 6 h (Figure 6G, Supplementary Figure S3D), and the expression of Cxcl10, Ccl2, and Il-6 mRNA at 24 h (Figure 6H), suggesting that the TBK1 axis contributes to the persistence of TWEAK-induced inflammatory responses over time.

In vivo, cells are simultaneously exposed to multiple proinflammatory cytokines that may amplify tubular cell death. A cell microenvironment combining TWEAK, TNFα, and IFNγ (TTI) triggers both an inflammatory response and tubular apoptosis (Justo et al., 2006). Like TWEAK alone, TTI increased cytokine and ISG mRNA levels in cultured tubular cells and this was prevented by inhibition of TBK1/IKKε by amlexanox (Figure 7A). Amlexanox also decreased tubular stress and injury markers Kim-1 and Ngal (Figure 7B) and cell death (Figure 7C) elicited by TTI, suggesting a role of TBK1/IKKε in driving inflammation-induced tubular cell death that is independent of IFNAR, as IFNAR blockade did not prevent TTI-induced cell death (Figure 7D). Altogether, these results support a role for TBK1/IKKε beyond mediating inflammation in also modulating cell death in tubular cells immersed in an inflammatory environment (Figure 7E).

Based on the differential expression of genes related to the TI-IFN pathway in experimental and human kidney injury and the impact of TI-IFNs pathways on tubular cell biology, we addressed the impact of inhibiting TBK1/IKKε signaling or blocking IFNAR in sterile kidney tubulointerstitial inflammation and injury.

LPS administration to mice causes endotoxemia and nephrotoxicity. IFNAR blockade by anti-IFNAR antibodies (Figure 8A) decreased the LPS-induced upregulation of mRNA encoding several ISG, including Cxcl10, which targets Th1 lymphocytes and is involved in sepsis-associated kidney injury and other kidney disease conditions (Herzig et al., 2014; Gao et al., 2020) (Figure 8B). IFNAR blockade also decreased the interstitial infiltration by F4/80 + mononuclear phagocytes (Figure 8C), MPO + immune cells and Th1 (T-BET+) lymphocytes (Supplementary Figure S4), and restricted tubular cell death as assessed by TUNEL (Figure 8D). These results suggest that the TI-IFN pathway contributes to LPS-induced kidney inflammation and tubular cell death. We next inhibited TBK1 and IKKε with amlexanox (Figure 8A). Amlexanox protected from kidney dysfunction (Figure 8E), kidney infiltration by F4/80 + cells (Figure 8F), and cell death (Figure 8G).

In murine folic acid-induced nephropathy (FAN) (Figure 9A), IFNAR neutralization did not preserve renal function assessed by plasma urea levels (FAN: 245.9 ± 43.3 mg/ml; FAN/IFNAR-Ab: 274.0 ± 25.6 mg/dl; p = ns) but did decrease kidney cytokine and ISG mRNA expression (Figure 9B) and kidney infiltration by F4/80 + phagocytes (Figure 9C), supporting the contribution of the TI-IFN pathway to kidney inflammation. IFNAR blockade also decreased tubular cell death assessed by TUNEL (Figure 9D). Likewise, treatment with amlexanox (Figure 9A) protected from kidney dysfunction (Figure 9E), kidney infiltration by F4/80+ and MPO + cells (Figures 9F,G), and cell death (Figure 9H). Finally, activating the TBK1/IKKε/IRF3 and TI-IFN pathways from the top with DMXAA (Vadimezan), that mimics signaling of TBK1/IKKε initiated by viral dsDNA or bacterial cyclic dinucleotides (Roberts et al., 2007) increased renal cytokine mRNA expression and aggravated the course of FAN following 72-96 h by reducing mice survival. These results suggest deleterious effects of virus or bacteria-induced TBK1/IKKε/IRF3 and TI-IFN pathways over the course of renal injury (Supplementary Figure S4B).

Overall, results from nephrotoxic mouse modeling suggest that upstream (TBK1 and IKKε) and downstream (IFNAR) signals within the TI-IFN pathway promote kidney inflammation and injury.