IL‐1α cleavage by inflammatory caspases of the noncanonical inflammasome controls the senescence‐associated secretory phenotype

Abstract Interleukin‐1 alpha (IL‐1α) is a powerful cytokine that modulates immunity, and requires canonical cleavage by calpain for full activity. Mature IL‐1α is produced after inflammasome activation and during cell senescence, but the protease cleaving IL‐1α in these contexts is unknown. We show IL‐1α is activated by caspase‐5 or caspase‐11 cleavage at a conserved site. Caspase‐5 drives cleaved IL‐1α release after human macrophage inflammasome activation, while IL‐1α secretion from murine macrophages only requires caspase‐11, with IL‐1β release needing caspase‐11 and caspase‐1. Importantly, senescent human cells require caspase‐5 for the IL‐1α‐dependent senescence‐associated secretory phenotype (SASP) in vitro, while senescent mouse hepatocytes need caspase‐11 for the SASP‐driven immune surveillance of senescent cells in vivo. Together, we identify IL‐1α as a novel substrate of noncanonical inflammatory caspases and finally provide a mechanism for how IL‐1α is activated during senescence. Thus, targeting caspase‐5 may reduce inflammation and limit the deleterious effects of accumulated senescent cells during disease and Aging.

Senescence is a protective mechanism that induces permanent cell cycle arrest to prevent transmission of defects to the next generation, particularly to stop malignant transformation. Replicative senescence occurs after repeated cell division critically shortens telomeres, while induced senescence occurs after oncogene activation, mitochondrial deterioration, oxidative stress or DNA damage (Munoz-Espin & Serrano, 2014). As ageing drives tumorigenesis and telomere shortening, it induces senescence via both pathways.
We show that IL-1α is specifically cleaved and activated at a conserved site by caspase-5 and caspase-11, but not caspase-4.
Knockdown of caspase-5 or expression of a caspase site mutant reduces release of IL-1α after icLPS stimulation. IL-1α cleavage and release from murine macrophages after icLPS require only caspase-11, while IL-1β needs both caspase-11 and caspase-1. Importantly, we show caspase-5 and caspase-11 are required for senescent cells to establish the IL-1α-dependent SASP. Thus, IL-1α is a direct substrate for inflammatory caspases during noncanonical inflammasome activation and senescence.

| Caspase-5 cleavage of human IL-1α at a conserved site increases activity
Although noncanonical inflammasomes that utilize caspase-4 and caspase-5 result in IL-1α and/or IL-1β release (Kayagaki et al., 2011;Shi et al., 2014), their role in cleavage or secretion of IL-1α is unknown. Thus, we investigated whether IL-1α could be proteolytically activated by these caspases. Incubation of recombinant pro-IL-1α with active inflammatory caspases revealed that caspase-5 cleaved IL-1α to a ~19 kDa fragment, while caspase-1 or caspase-4 did not  (Afonina et al., 2011), with the Asp present in 81% of all sequences available (Ensembl). Cleavage at D 103 would produce a ~19 kDa C-terminal fragment, which is congruent with the size seen by Western blot (Figure 1a). Mutation of Asp 103 to Ala abolished caspase-5 cleavage of pro-IL-1α ( Figure 1f) and thus prevented activation ( Figure 1g). This shows that caspase-5 increases IL-1α activity by direct processing at a conserved site located adjacent to the calpain site ( Figure 1h).

| Release of cleaved IL-1α from human cells requires caspase-5
Due to the similar size of calpain-and caspase-5-cleaved IL-1α, we produced a peptide antibody that was reactive to the region between the calpain and caspase-5 sites (Supporting information Figure S1a) and developed an ELISA that recognized caspase-5-cleaved IL-1α, but not pro-IL-1α or calpain-cleaved IL-1α (Figure 2a and Supporting information Figure S2b). We also established an IL-1α ELISA that recognized total cleaved IL-1α, but not pro-IL-1α (Supporting information Figure S1c). To test IL-1α processing and release in cells, we

| Caspase-5 is required for the IL-1α-dependent senescent-associated secretory phenotype
The SASP directs clearance of senescent cells, but also drives chronic inflammation that can promote disease and unhealthy ageing. As IL-1α drives the SASP, we investigated whether caspase-5 is required for IL-1α release during senescence. We utilized the well- Yang, Wang, Ren, Chen, & Chen, 2017). As CASP5 is interferon responsive, we tested whether its expression was cGAS-dependent.
cGAS knockdown resulted in reduced release of IL-1α and IL-6, as expected, but also decreased CASP5 expression (Supporting information Figure S9a-d). Together, this suggests that release of cleaved fully active IL-1α and expression of the subsequent SASP is dependent on caspase-5, with upstream cGAS potentially controlling CASP5 expression.

| Caspase-11 is required for immune surveillance of senescent cells in vivo
To investigate whether caspase-11 is required for the SASP in vivo, we used a mouse liver model of NRAS-induced senescence.

| D ISCUSS I ON
Due to the powerful, pleiotropic effects of IL-1, unprecedented levels of control and feedback pathways are required to mount an immune response that resolves an insult without causing overt tissue damage. For example, multiple inflammasomes have evolved to distinguish tissue damage, environmental pollutants, bacteria, viruses, fungi and protozoa. IL-1 also drives homeostatic processes such as thermoregulation (Dinarello, 2009), regulatory B-cell differentiation (Rosser et al., 2014), intestinal integrity (Jung et al., 2015), and the immune surveillance and clearance of senescence cells (Kang et al., 2011). However, how IL-1α is cleaved, activated and released was unknown.
We find that IL-1α is directly cleaved by caspase-5 and caspase-11 at a conserved site, leading to full cytokine activity. Expression of a caspase site mutant IL-1α or knockdown of CASP5 prevents release of cleaved IL-1α from cells, while IL-1α cleavage and release from murine macrophages require only caspase-11, with IL-1β needing both caspase-11 and caspase-1. Importantly, the IL-1α-dependent SASP requires caspase-5 and caspase-11 in vitro and in vivo. Thus, IL-1α is a direct substrate of inflammatory caspases during noncanonical inflammasome activation and senescence.
Although independent publications show cleavage of IL-1α greatly increases activity (Afonina et al., 2011;Zheng et al., 2013), others contest this (Kim et al., 2013).  by HPLC, which typically denatures proteins due to the organic solvent mobile phase, and we find denatured and refolded pro-IL-1α has more activity than native pro-IL-1α (Zheng et al., 2013).
In murine systems, caspase-11 binds LPS, increases caspase-1 cleavage of IL-1β (Wang et al., 1998), activates gasdermin D (Kayagaki et al., 2015;Shi et al., 2015) and now also directly cleaves and activates IL-1α. In humans, an ancestral gene duplication resulted in CASP4 and CASP5, with caspase-4 suggested as the closer homologue of caspase-11 (Shi et al., 2014). Both caspase-4 and caspase-5 bind LPS and target gasdermin D, but only caspase-5 can complement Casp11 −/− cells (Shi et al., 2014), only CASP5 and Casp11 are upregulated in response to LPS (Figure 3a,b), and only caspase-5/11 can cleave IL-1α, suggesting a closer functional connection. We suggest caspase-4 and caspase-5 have most likely undergone subfunctionalization, with functionality of the ancestral Casp11-like gene now distributed between both. However, if each performs nonredundant roles, and what these are, is still unknown.
Although IL-1α induces sterile inflammation upon release from damaged cells, it also drives the SASP, but in this context IL-1α release or surface presentation occurs without cell death (Figure 4g and Gardner et al., 2015). This excludes the usual notion that IL-1α is passively released (Kayagaki et al., 2011). IL-1α expression during the SASP is controlled in part by ATM/ATR liberation of GATA4 from p62-directed autophagy (Kang et al., 2015) and/or an mTORC1-dependent pathway (Laberge et al., 2015). Also, redox balance and Ca 2+ levels may enhance calpain processing of IL-1α in senescent cells Together, this suggests that cleavage of IL-1α by caspase-5/11 during senescence leads to release of active IL-1α only, which drives the SASP. Interestingly, CASP5 mutations are associated with cancers (Offman et al., 2005), while CASP1 and CASP4 mutations are not (Soung et al., 2008), suggesting a caspase-5-specific protective role against tumorigenesis that is in keeping with SASP-driven senescence surveillance.
In conclusion, we show that IL-1α is a direct substrate of the noncanonical inflammatory caspase-5 and caspase-11, with cleavage increasing cytokine activity and release. Conservation of the cleavage site between species implies that activation of IL-1α by caspase-5 or caspase-11 is important. Indeed, cleavage of IL-1α by caspase-5 or caspase-11 is essential for the SASP. Thus, directly targeting caspase-5 may reduce inflammation and limit the deleterious effects of senescent cells that accumulate during disease and ageing.

| MATERIAL S AND ME THODS
All material from Sigma-Aldrich unless otherwise stated. (Clontech), as previously described (Young et al., 2009) crystal violet staining as previously described (Gardner et al., 2015).
Specific IL-1 activity was inferred by the effect of neutralizing antibodies (2 µg/ml; same as for Westerns), on IL-6 production. Every experiment contained a media only negative control and a recombi-

| Induction of mouse hepatocyte senescence in vivo
Mice were handled and kept under pathogen-free conditions in accordance with UK law and institutional guidelines at the University of Cambridge and CRUK Cambridge Institute. Transposon-mediated gene transfer by hydrodynamic tail vein injection was as previously described (Dauch et al., 2016;Kang et al., 2011). Briefly, short-hairpin oligomers targeting Casp11 (5'-TACCATCTATCAGATATTCAA-3') were cloned into MSCV-miR30-puro before subcloning into pCaNIG-miR30 (from Lars Zender (Dauch et al., 2016)

| Western blotting and cytokine detection
Western blotting was performed as previously described. Blocked cleavage site in IL-1α was generated by immunizing rabbits with a KLH-conjugated peptide, followed by affinity purification and biotinylation (Innovagen). This was then used as the detection reagent in a standard sandwich ELISA using a goat anti-human IL-1α antibody (R&D) as capture.

| Gene expression analysis
RNA was isolated (RNeasy, Qiagen) and converted to cDNA using AMV reverse transcriptase (Promega). qPCR used TaqMan probes with AmpliTaq Gold (Life Technologies) in a Rotor-Gene thermocycler (Corbett). Gene expression was evaluated using the 2 −ΔΔCT method using GUSB as a reference gene. qPCR data are displayed as 2 −ΔΔCT , but statistical analysis was performed on the untransformed ΔΔCT values. RNA-Seq data were from GEO GSE72404.

| Immunofluorescence and Immunohistochemistry
After processing, paraffin sections were cleared before antigen

| Statistical Analysis
All statistical analyses were carried out using Prism 7 (GraphPad).
Data were analysed by unpaired t test (two-tailed), one-way ANOVA with Dunnett's post hoc or one-way ANOVA with Tukey's post hoc. n = individual biological replicate performed in duplicate. The in vitro experiments typically required sample sizes of n = 3-4 to provide adequate power, and produced data that were normally distributed.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y
The data that support the findings of this study are available from the corresponding author upon reasonable request.