Caspase-4/11 exacerbates disease severity in SARS-CoV-2 infection by promoting inflammation and thrombosis

SARS-CoV-2 is a worldwide health concern, and new treatment strategies are needed 1. Targeting inflammatory innate immunity pathways holds therapeutic promise, but effective molecular targets remain elusive. Here, we show that human caspase-4 (CASP4), and its mouse homologue, caspase-11 (CASP11), are upregulated in SARS-CoV-2 infections, and that CASP4 expression correlates with severity of SARS-CoV-2 infection in humans. SARS-CoV-2-infected Casp11-/- mice were protected from severe weight loss and lung pathology, including blood vessel damage, compared to wild-type (WT) and gasdermin-D knock out (Gsdmd-/-) mice. GSDMD is a downstream effector of CASP11 and CASP1. Notably, viral titers were similar in the three genotypes. Global transcriptomics of SARS-CoV-2-infected WT, Casp11-/- and Gsdmd-/- lungs identified restrained expression of inflammatory molecules and altered neutrophil gene signatures in Casp11-/- mice. We confirmed that protein levels of inflammatory mediators IL-1β, IL6, and CXCL1, and neutrophil functions, were reduced in Casp11-/- lungs. Additionally, Casp11-/- lungs accumulated less von Willebrand factor, a marker for endothelial damage, but expressed more Kruppel-Like Factor 2, a transcription factor that maintains vascular integrity. Overall, our results demonstrate that CASP4/11, promotes detrimental SARS-CoV-2-associated inflammation and coagulopathy, largely independently of GSDMD, identifying CASP4/11 as a promising drug target for treatment and prevention of severe COVID-19.

pathogen-sensing pathways and other inflammasome components in mediating host defense 70 versus immune-mediated pathology and thrombosis during SARS-CoV-2 infection in vivo remains 71 unclear 7 . While effector molecules downstream of infection-sensing pathways, such as specific 72 inflammatory cytokines, have been targeted in attempts to limit virus-induced tissue damage, 73 most of these strategies failed to exert major benefits in human clinical trials 8 . Therefore, 74 strategies targeting molecules upstream of multiple inflammatory cytokines or chemokines may 75 be more effective, though this remains to be experimentally tested. Here, we investigate the role 76 of a major member of the non-canonical inflammasome, caspase-11 (CASP11), and its 77 downstream effector Gasdermin D (GSDMD) in SARS-CoV-2 infection and disease severity using 78 knockout mouse models and mouse-adapted SARS-CoV-2.

79
Caspases are a family of cysteine proteases that specifically cleave their substrates at the 80 C-terminal side of aspartic acid residues. CASP11 is a murine protein that is critical for defense 81 against bacterial pathogens. Human caspase-4 (CASP4) displays high homology to murine 82 CASP11 9,10 and we have demonstrated that human CASP4 mediates many functions of mouse 83 CASP11 in macrophages during bacterial infections 9 . CASP4/11 is a component of the non-84 canonical inflammasome with multiple functions that remain to be fully characterized. One major 85 role for this protein is the cleavage of GSDMD 11 . Once cleaved, the GSDMD N-terminal fragment 86 inserts into the plasma membrane of eukaryotic cells to form pores that allow the release of IL-1β 87 and other molecules, sometimes leading to cell lysis and death known as pyroptosis 12 .

88
Interestingly, accumulating evidence has posited potential roles for GSDMD downstream of

104
revealed that CASP4 is highly expressed in the airway of SARS-CoV-2-infected patients, and that 105 expression levels increase with disease severity (Fig. 1a). Additionally, we found that human lung 106 sections from COVID-19 patients show higher levels of CASP4 staining compared with healthy 107 lung controls (Fig. 1b), owing to greater numbers of CASP4 positive cells in the infected lung 108 tissue (Fig. 1c). We then performed intranasal infection of C57BL/6 wild-type (WT) mice with 109 pathogenic mouse-adapted SARS-CoV-2 (strain MA10) 15 , and found that infection strongly 110 induces Casp11 expression throughout murine lung tissue within 4 days of infection as detected 111 by RNAscope in situ hybridization (ISH) (Fig. 1d) and confirmed by qRT-PCR (Fig. 1e). The level 112 of CASP11 protein, likewise, went from below detection to highly-expressed in response to SARS-

123
MA10 for comparison of weight loss, a commonly used indicator of overall infection severity in 124 mice 16 . We found that WT mice lost a significant percentage of their body weight between days 125 1 and 4 post-infection, followed by partial recovery of weight up to day 7, at which point we ended 126 our experiments (Fig. 2a). Casp11 -/mice, on the other hand, lost weight only up to day 3, and 127 then rapidly recovered fully to their original weight by day 5 (Fig. 2a). In comparison, weight loss 128 of Gsdmd -/mice was not significantly different from that of WT mice (Fig. 2a). These data indicate 129 that CASP11 promotes disease severity during SARS-CoV-2 infection, and that this function is 130 not mediated by GSDMD.

131
To determine whether differences in disease severity could be explained by differences in 132 viral replication, we quantified live virus titers in WT, Casp11 -/and Gsdmd -/mouse lungs at 2 and 133 4 days post-infection. We found that viral loads were similar with no statistical difference between 134 the mice genotypes at either time point (Fig. 2b). We also observed that, in agreement with 135 previous reports 17 , viral titers were decreased at day 4 compared with day 2 in all groups, 136 demonstrating that neither CASP11 nor GSDMD are required for viral clearance mechanisms ( Fig. 2b). To corroborate these findings, lung sections from WT and Casp11 -/mice were stained 138 for SARS-CoV-2 nucleocapsid protein and similar staining patterns were observed with prominent 139 infection of cells lining the airways and neighboring alveoli (Fig. 2c). Overall, these results 140 demonstrate that loss of CASP11, but not GSDMD, prevents severe disease in SARS-CoV-2 141 infection without affecting virus replication or clearance. Gsdmd -/infected lungs were seen using dimensionality reduction approaches, while the Casp11 -

149
/infected lung profiles showed greater divergence in gene expression patterns (Fig. 2d). We   Fig. 1b,c). 170 signaling as highlighted in (Fig. 2e). Knockout of Gsdmd, had less impact on the magnitude of cytokine and chemokine expression compared to Casp11 knockout (Fig. 2f)   infected control tissue (Fig 3a). However, WT and Gsdmd -/lung sections showed more severe 182 tissue consolidation and cell infiltration throughout a greater portion of the lung than that seen in 183 Casp11 -/mice. We thus quantified cell area versus airway space to determine cellularity scores 184 indicative of pathology for tissue sections from individual mice 21 . We observed significantly 185 decreased SARS-CoV-2-induced lung pathology in Casp11 -/mice compared to WT and Gsdmd -186 /mice ( Fig. 3a,b), correlating with the preservation of Casp11 -/mice body weight and their faster 187 recovery that (Fig. 2a).

188
Guided by our transcriptomic results indicating that a critical subset of inflammatory 189 mediators are controlled by CASP11 (Fig. 2f), we measured levels of CXCL1, IL-1β, and IL-6 by 190 ELISA in lung homogenates from infected animals at 2 and 4 days post-infection (Fig. 3c,d).  1β was lower in the lungs of both Casp11 -/and Gsdmd -/mice at 2 days post-infection when 192 compared to WT (Fig. 3c). Moreover, IL-1β staining in lung tissue sections revealed more IL-1β 193 in WT SARS-CoV-2 infected mice than Casp11 -/infected ones (Supplementary Fig. 2). On the 194 other hand, the production of CXCL1 was dependent on Casp11 at both time-points, and 195 independent of Gsdmd (Fig. 3c,d). Average levels of IL-6 were partially decreased in Casp11 -/-196 lungs with a statistically significant difference between WT and Casp11 -/lungs at day 4 ( Fig.   197 3c,d). These results corroborate and expand our day 2 transcriptomic analysis in which 198 expression of Il1b and Cxcl1 was decreased (Fig. 2f), and demonstrate that production of a critical 199 subset of inflammatory mediators in the lung is dependent on CASP11 during SARS-CoV-2 200 infection.

201
To determine the role of CASP11 in the response of lung macrophages to SARS-CoV-2, 202 we purified mature primary macrophages from lungs of WT and Casp11 -/mice, and infected them 203 with SARS-CoV-2 MA10. Culture supernatants and cellular RNA were collected and measured 204 for IL-1β, IL-6 and CXCL1 protein and transcript levels, respectively. Compared with non-infected cells, CXCL1 protein and RNA transcripts were detected at high levels upon infection of WT 206 macrophages, but were poorly produced by Casp11 -/cells (Fig. 3e,f). Interestingly, IL-1β 207 transcripts were also induced in a CASP11-dependent manner, but secreted protein was not 208 detected in either group (Fig. 3e,f). Distinctly, protein and transcript levels of IL-6 did not 209 significantly differ between WT and Casp11 -/cells (Fig. 3e,f). These results confirm our in vivo 210 measurements and further demonstrate that CASP11 is an important cellular regulator of specific 211 cytokines and chemokines, including CXCL1 and IL-1b, in response to SARS-CoV-2.  (Fig. 3g, Supplementary Fig. 1d). Additionally, genes 220 associated with the response to tissue damage from neutrophils (Slpi and Lair1) were also 221 decreased in the absence of Casp11 relative to WT lungs (Fig. 3g). These results are consistent 222 with decreased gene expression for the neutrophil chemoattractant CXCL1 (Fig. 3c,d), as well 223 as with previous reports of neutrophil regulation by CASP11 through effects on actin 9,19 .

230
Overall, these analyses most prominently demonstrate that CASP11 is required for robust 231 production of specific inflammatory mediators as well as neutrophil recruitment and functions in 232 the lung during SARS-CoV-2 infection.

247
Notably, Casp11 -/neutrophils failed to form NETs in response to all conditions. In contrast,

272
KLF2 is an endothelial protective transcription factor that exerts anti-inflammatory and anti-thrombotic functions in the vascular endothelial cell and maintains the integrity of the endothelial 274 vasculature. We found that KLF2 expression is significantly reduced after SARS-CoV-2-infection 275 in WT lungs, but largely preserved in Casp11 -/infected lungs (Fig. 4e). Moreover, we examined 276 the vascular architecture in the cleared lungs of SARS-CoV-2 infected mice by using fluorophore 277 conjugated albumin and tissue clearing (Supplementary Fig 6). The vascular tracing revealed 278 distinctive vascular features in WT SARS-CoV-2-infected lungs with pronounced vascular 279 thickening and angiogenesis/neovascularization (Fig. 4g). In stark contrast, Casp11 -/infected 280 lung vasculature did not show these abnormalities, confirming less endothelial 281 damage/dysfunction (Fig. 4g). Taken together, we conclude that CASP11 contributes to         Together, our findings suggest that targeting the CASP11 homologue, human CASP4, 372 during COVID-19 will prevent severe pneumonia, inflammation, tissue damage as well as 373 thrombosis and accompanying repercussions such as low oxygen, lung failure, need for 374 ventilators and perhaps long-term sequela. These advantageous effects will be achieved without compromising viral clearance. It is also plausible that the level of expression of CASP4 could 376 serve as a biomarker to identify patients who will succumb to severe Covid. Targeting CASP4 377 alone can achieve benefits that will exceed and replace the administration of a large number of 378 individual anti-inflammatory agents and anti-thrombotics given to SARS-CoV-2 patients. Further 379 research is needed to develop therapeutics in this regard.