Protein kinase R is an innate immune sensor of proteotoxic stress via accumulation of cytoplasmic IL-24

Abstract

Proteasome dysfunction can lead to autoinflammatory disease associated with elevated type I interferon (IFN-αβ) and NF-κB signaling; however, the innate immune pathway driving this is currently unknown. Here, we identified protein kinase R (PKR) as an innate immune sensor for proteotoxic stress. PKR activation was observed in cellular models of decreased proteasome function and in multiple cell types from patients with proteasome-associated autoinflammatory disease (PRAAS). Furthermore, genetic deletion or small-molecule inhibition of PKR in vitro ameliorated inflammation driven by proteasome deficiency. In vivo, proteasome inhibitor-induced inflammatory gene transcription was blunted in PKR-deficient mice compared with littermate controls. PKR also acted as a rheostat for proteotoxic stress by triggering phosphorylation of eIF2α, which can prevent the translation of new proteins to restore homeostasis. Although traditionally known as a sensor of RNA, under conditions of proteasome dysfunction, PKR sensed the cytoplasmic accumulation of a known interactor, interleukin-24 (IL-24). When misfolded IL-24 egress into the cytosol was blocked by inhibition of the endoplasmic reticulum-associated degradation pathway, PKR activation and subsequent inflammatory signaling were blunted. Cytokines such as IL-24 are normally secreted from cells; therefore, cytoplasmic accumulation of IL-24 represents an internal danger-associated molecular pattern. Thus, we have identified a mechanism by which proteotoxic stress is detected, causing inflammation observed in the disease PRAAS.

Fig. 1.. CRISPR-Cas9–mediated deletion of iβ5 or iβ1 generates cell lines that recapitulate the molecular hallmarks of PRAAS pathology.

CRISPR-Cas9 gene editing was used to delete the inducible proteasome subunit genes PSMB8 and PSMB9 from the monocytic cell line, THP-1. (A) Accumulation of ubiquitinated proteins and successful deletion of PSMB8 and PSMB9 was assessed by immunoblotting for total ubiquitin and the proteasomal subunits iβ5 and iβ1 (encoded for by PSMB8 and PSMB9, respectively). Phosphorylation of p65 in THP-1, iβ5^−/−, and iβ1^−/− cell lines was assessed by immunoblotting with actin used as a loading control. (B) Secretion of IFN-β and the IFN inducible chemokine CXCL10 by THP-1, iβ5^−/−, and iβ1^−/− lines were assessed using ELISA. (C) Levels of IFN-α and phosphorylated (p)STAT1 in parental THP-1 (black line), iβ5^−/− (red), and iβ1^−/− (maroon) cell lines were assessed by flow cytometry, gray lines represent unstained cells. Histograms are representative of four independent experiments summarized in column graphs. (D) iβ5^−/−, iβ1^−/−, and THP-1 cells were treated for 14 hours with the nonspecific DUB inhibitor: PR619 (5 μM) and expression of stated inflammatory genes were assessed by qPCR. Data are pooled from at least three experiments, and statistical significance was assessed by a ratio paired t test, where * indicates P < 0.05 and ** indicates P < 0.01.

Fig. 2.. Proteasome dysfunction–induced inflammation is PKR dependent in vitro and in vivo.

(A) CRISPR-Cas9 gene editing was performed on THP-1 cells to generate UNC93B1 deficient THP-1 cells (UNC93B1^−/−). Cells were treated with oprozomib (500 nM) for 16 hours, and qPCR was used to assess IFNB1 gene induction. (B) MEF cell lines deficient in MAVS (MAVS^−/−), TMEM173 (STING^−/−), and EIF2AK2 (PKR^−/−) were treated with delanzomib (200 nM) or oprozomib (500 nM), and IFNB1 gene induction was assessed by qPCR. (C) EIF2AK2 was CRISPR-Cas9 deleted from THP-1 cells (PKR^−/−), and cells were treated with delanzomib (200 nM) or the STING agonist: cGAMP (10 μg/ml), as indicated, for 16 hours, and induction of IFNB1 mRNA was assessed by qPCR. (D) EIF2AK2 was deleted from iβ5^−/− and iβ1^−/− THP-1 cells to generate PKR^−/−iβ5^−/− and PKR^−/−iβ1^−/− cell lines. Immunoblotting was performed on THP-1, iβ5^−/−, iβ1^−/−, PKR^−/−, PKR^−/−iβ5^−/−, and PKR^−/−iβ1^−/− cells to assess phosphorylation (p) of STAT1, PKR, and eIF2α. (E) Induction of IFNB1 mRNA upon PR619 (5 μM) treatment of THP-1, iβ5^−/−, and PKR^−/−iβ5^−/− cells was assessed by qPCR. (F) iβ5^−/− or THP-1 cells were cultured with the PKR inhibitor: C16 (0.5 μM) and/or PR619 (5 μM) for 14 hours, induction of IFNB1 was assessed by qPCR. (G) PKR^−/− and littermate (PKR^+/+) mice were treated with bortezomib (1 mg/kg) or vehicle control (Ctrl) for 18 hours. Cardiac bleeds were processed for RNA, and expression of Ifnb1, Ifna4, Il6, and Usp18 was assessed by qPCR. Data are pooled from at least three experiments, and statistical significance was assessed by ratio paired t test for in vitro experiments and by Student’s t test for in vivo results. * indicates P < 0.05 and *** represents P < 0.001.

Fig. 3.. Cytosolic IL-24 activates PKR to drive inflammatory signaling under conditions of proteasome inhibition.

(A) CRISPR-Cas9 gene editing was used to delete IL24, PRKRA, and TARBP2 genes in THP-1 cell lines to generate cells deficient in IL-24, PACT, and TRBP (respectively). Cells were then treated for 16 hours with delanzomib (200 nM), poly(dA:dT) (1 μg/ml), or vehicle control (Ctrl) as indicated, and induction of IFNB1 mRNA was assessed by qPCR. (B and C) IL24 was CRISPR- Cas9– deleted from iβ5^−/− or iβ1^−/− cells, and (B) induction of IFNB1 mRNA was assessed by qPCR in iβ5^−/−, IL-24^−/−, iβ5^−/−IL-24^−/−, and parental THP-1 cell lines. (C) Immunoblotting was performed on parental, iβ5^−/−, and iβ1^−/− THP-1 cells sufficient or deficient in IL-24 to assess phosphorylation (p) of STAT1, PKR, and eIF2α. (D) IL-24 tagged with HA or empty vector control was stably expressed in IL-24^−/− or IL-24^−/−PKR^−/− THP-1 cells by lentiviral transfection. Cells were treated with delanzomib (200 nM) for 16 hours, and induction of IFNB1 was assessed by qPCR. (E) The IL-24 receptor subunit, IL20RB was CRISPR-Cas9–deleted from THP-1 cells, cells were then treated with delanzomib (200 nM) for 16 hours and induction of IFNB1 was assessed by qPCR. (F) PKR-IP and WCLs from THP-1, iβ5^−/−, iβ5^−/−PKR^−/−, and iβ5^−/− IL-24^−/− cell lines were immunoblotted for PKR, IL-24, and actin (loading control). (G) IL-24–Halo was stably expressed in U2OS cells. Super-resolution microscopy was then performed to monitor IL-24–Halo cellular localization during proteasome inhibitor treatment (delanzomib, 200 nM). Egress of IL-24–Halo from the ER to the cytosol could be inhibited by kifunesine. Representative images of specified timepoints are displayed, IL-24–Halo in red, ER-Tracker in green and DAPI in blue. White scale bars, 10 μM. (H) Three to seven representative cells from each time point from (G) were analyzed using Imaris software to determine percentage IL-24–Halo outside the ER (cytoplasmic IL-24). (I) THP-1 cells were treated with delanzomib (200 nM) and/or kifunesine (50 nM) for 14 hours and induction of IFNB1 and IL6 was assessed by qPCR. (J) Expression of IL-24 splice isoforms, and β-tubulin mRNA expression was analyzed by PCR in PRAAS patient and HC fibroblast lines, under normal conditions (lanes 1 to 5) and serum starvation (lanes 6 to 10). (K) THP-1 cells were stably transfected with HA-tagged full length IL-24 or IL-24 ΔEx5 (as indicated). Differentiation into macrophages was performed using PMA (72 hours), and IF for IL-24 (HA tag) in red and DAPI in blue was assessed at baseline and 14 hours after treatment with delanzomib (200 nM). Data are pooled from three to four experiments, and statistical significance was assessed by ratio paired t test, where * indicates P < 0.05, ** indicates P < 0.01, and ns indicates not significant, with the exception of (H) where statistical significance was assessed by two-way ANOVA with Bonferrioni post tests for individual time points, where ** denotes P < 0.01, and * indicates P < 0.05.

Fig. 4.. PKR mediates inflammatory signaling in PRAAS patient samples.

(A) Lysates from patient fibroblasts were immunoblotted for (p)PKR and (p)eIF2α, as well as IL-24 and actin. (B) Patient PBMCs were treated with C16 (0.5 μM) for 8 hours, and protein lysates were then immunoblotted for (p)PKR, (p)STAT1, and (p)eIF2α, as well as IL-24 and actin. (C) PBMC samples from PRAAS and SAVI patients as well as HC samples were cultured with C16 for 8 to 14 hours. Expression of IFNB1 and IFNA1 was assessed by qPCR. IFN-αβ induction of ISGs was assessed by NanoString Technologies to generate an ISG score. Significance was assessed by Wilcoxon matched-pair signed-rank test, where * indicates P < 0.05 and ** denotes P < 0.01. (D) iPSC-derived macrophages from PRAAS patient #3 and a HC were assessed for expression of IFNB1, IFNA1, and IFNA2 at baseline and after 8 hours of treatment with C16 (1 μM). Significance was assessed by Student’s t test, * indicates P < 0.05.