Immune dysregulation in SHARPIN-deficient mice is dependent on CYLD-mediated cell death

Significance

The mechanisms underlying inflammatory disorders are poorly understood. In this study, we show that inappropriate cell death may cause uncontrolled inflammation. We found that CYLD, an enzyme that removes K63-linked polyubiquitination, is normally inhibited. But in the cpdm mouse strain that has a loss-of-function in the Sharpin gene, the brake on CYLD is no longer present. When CYLD is no longer inhibited, it turns on death signaling in cells exposed to the cytokine TNF, and the ensuing inappropriate cell death causes skin inflammation and other immune disorders in the cpdm mouse. Removing CYLD from the cpdm mouse prevents cell death and reverses the inflammation. We conclude that excessive CYLD activity leads to inappropriate cell death and inflammation.

Abstract

SHARPIN, together with RNF31/HOIP and RBCK1/HOIL1, form the linear ubiquitin chain assembly complex (LUBAC) E3 ligase that catalyzes M1-linked polyubiquitination. Mutations in RNF31/HOIP and RBCK/HOIL1 in humans and Sharpin in mice lead to autoinflammation and immunodeficiency, but the mechanism underlying the immune dysregulation remains unclear. We now show that the phenotype of the Sharpin^cpdm/cpdm mice is dependent on CYLD, a deubiquitinase previously shown to mediate removal of K63-linked polyubiquitin chains. Dermatitis, disrupted splenic architecture, and loss of Peyer's patches in the Sharpin^cpdm/cpdm mice were fully reversed in Sharpin^cpdm/cpdm Cyld^−/− mice. We observed enhanced association of RIPK1 with the death-signaling Complex II following TNF stimulation in Sharpin^cpdm/cpdm cells, a finding dependent on CYLD since we observed reversal in Sharpin^cpdm/cpdm Cyld^−/− cells. Enhanced RIPK1 recruitment to Complex II in Sharpin^cpdm/cpdm cells correlated with impaired phosphorylation of CYLD at serine 418, a modification reported to inhibit its enzymatic activity. The dermatitis in the Sharpin^cpdm/cpdm mice was also ameliorated by the conditional deletion of Cyld using LysM-cre or Cx3cr1-cre indicating that CYLD-dependent death of myeloid cells is inflammatory. Our studies reveal that under physiological conditions, TNF- and RIPK1-dependent cell death is suppressed by the linear ubiquitin-dependent inhibition of CYLD. The Sharpin^cpdm/cpdm phenotype illustrates the pathological consequences when CYLD inhibition fails.

Linear or M1-linked polyubiquitin modifications play a major role in regulating immune responses (1–3). This modification is catalyzed by the linear ubiquitin chain assembly complex (LUBAC) E3 ligase, which consists of the catalytic component RNF31/HOIP in complex with two other essential proteins with noncatalytic functions RBCK1/HOIL1 and SHARPIN (4–10). Linear ubiquitination crucially controls signaling downstream of several immune receptors including tumor necrosis factor receptor superfamily (TNFRSF) members, antigen receptors, and pattern recognition receptors (1, 11). In the case of TNF-α (TNF), LUBAC components are rapidly recruited to TNFR1 following receptor ligation to ubiquitinate target molecules, providing scaffolding for the formation of multimolecular complexes that propagate signaling. One critical signaling complex assembled by linear ubiquitin is the I-κB kinase (IKK) complex that phosphorylates I-κBα (5, 6, 12), a step required for subsequent degradation of I-κBα and induction of NF-κB–dependent gene transcription. NEMO (encoded by the Ikbkg gene), a structural component of the IKK complex, is modified by linear ubiquitination and also possesses a linear ubiquitin-binding domain that facilitates oligomerization of the IKK complex (5, 6, 12, 13). Cells deficient in linear ubiquitination have impaired IKK activity and NF-κB signaling in response to TNF stimulation (5, 7–9). Since IKK can phosphorylate substrates other than I-κB (14–16), whether linear ubiquitin regulates NF-κB–independent pathways and their downstream biological processes remains poorly understood.

Consistent with its role in the signaling of immune receptors, LUBAC components are critical for proper immune regulation. Loss-of-function mutations in RNF31/HOIP and in RBCK1/HOIL1 in humans cause immunodeficiency, autoinflammation, amylopectinosis, and lymphangiectasia (17, 18). In mice, a loss-of-function mutation in Sharpin has been found in the chronic proliferative dermatitis (cpdm) mice, which spontaneously develops dermatitis, multiorgan inflammation, and displays immunodeficiency (7–9, 19–21). Crossing of Sharpin^cpdm/cpdm mice with Tnf^−/− mice reversed the skin inflammation, indicating that some of the lesion in this mutant mouse are TNF driven (9). Sharpin^cpdm/cpdm cells exhibited enhanced cell death in response to TNF (8, 9), and Sharpin^cpdm/cpdm skin displays apoptotic signatures (22), suggesting that the phenotype of the Sharpin^cpdm/cpdm mice is due to inappropriate cell death. Subsequent genetic crosses with knockouts of death-signaling molecules provide further support for this hypothesis. Deletion of the necroptosis-signaling molecule Ripk3 delayed but did not prevent the dermatitis, but additional deletion of a single allele of the apoptosis-signaling molecule Casp8 or skin-specific deletion of Fadd reversed the dermatitis in the Sharpin^cpdm/cpdm mice (23, 24). Furthermore, crossing with a kinase-inactive Ripk1^tm1.1Gsk (Ripk1^K45A) allele or a deletion of a single copy of Ripk1 also reversed the phenotype of the Sharpin^cpdm/cpdm mice (25, 26). These genetic analyses suggest that the phenotype of the Sharpin^cpdm/cpdm mouse is due to excessive RIPK1-dependent cell death, and we suggested the term ripoptocide to denote this manner of cell death orchestrated by RIPK1 (27). One mechanism by which Sharpin deficiency could lead to aberrant RIPK1 death-signaling is impaired phosphorylation of RIPK1 catalyzed by members of the IKK family, which normally inhibits its death-promoting activity (28–32). Whether the Sharpin deficiency also affects other molecules involved in regulating cell death is unclear.

Another form of polyubiquitin modification, one in which the covalent linkage occurs at K63 of the ubiquitin molecule, is also critical in TNF signaling (33, 34). This form of ubiquitin modification is mediated by TRAF2-dependent recruitment of the cIAP1/2 E3 ligases (encoded by Birc2 and Birc3) and one key molecule modified in this manner is RIPK1 (26, 35). Cells in which the ubiquitin acceptor site of RIPK1 at K377 was mutated were more sensitive to TNF-mediated apoptosis, similar to cells in which TRAF2 and cIAP1/2 were inhibited (36–38). Cells treated with second mitochondria–derived activator of caspase (SMAC) mimetics, which induce the degradation and loss of cIAP1/2, displayed reduced polyubiquitination of RIPK1 (37, 38). This cell-protective function of K63-linked polyubiquitination of RIPK1 is initially transcriptional independent (i.e., does not require NF-κB) but subsequently feeds into NF-κB–dependent induction of prosurvival genes. A unifying model proposes that there are two cell death checkpoints in the TNFR1 signaling pathway. The early checkpoint (Checkpoint 1) prevents RIPK1 from becoming a death-signaling molecule via K63-linked polyubiquitination, which is followed by the NF-κB–dependent induction of prosurvival genes during the late checkpoint (Checkpoint 2) (39–41). Disruption of Checkpoint 1 is often achieved in vitro by blocking RIPK1 ubiquitination using SMAC mimetics, which leads to CASP8-dependent apoptosis or RIPK3/MLKL-dependent necroptosis (38, 42–44). The relationship between linear and K63 ubiquitin linkages in regulating cell death and potentially in the immune dysfunction observed in LUBAC deficiencies is currently unclear.

We now report that the molecule CYLD is central to the immunopathology observed in the Sharpin^cpdm/cpdm mice. CYLD is a deubiquitinase that dismantles K63-linked polyubiquitin chains (45), and its activity was shown to be suppressed by IKK-β–mediated phosphorylation (46). CYLD has also been reported to associate with RNF31/HOIP via SPATA2 (47–51). We found that SHARPIN-deficient cells, which are deficient in IKK activity, display reduced CYLD phosphorylation following receptor ligation. Coincident with defective CYLD phosphorylation in the SHARPIN-deficient cells, there is enhanced association of its substrate RIPK1 with the death-signaling Complex II to induce apoptosis and necroptosis, which is reversed in SHARPIN and CYLD double-deficient cells. The phenotype of the Sharpin^cpdm/cpdm mice is reversed by a compound deletion of Cyld and importantly, conditional deletion of Cyld in myeloid cells significantly reversed the dermatitis. Our study provides evidence that linear ubiquitin-dependent suppression of CYLD is one mechanism that keeps ripoptocide at bay and to maintain immune homeostasis, disruption of which leads to immune dysfunction.

Results

Cyld Is Essential for the Development of Inflammation in Sharpin-Deficient Mice.

Sharpin^cpdm/cpdm cells exhibited heightened sensitivity to TNF-induced killing (9, 23, 24). Since CYLD has been reported to be required for TNF to induce RIPK1-dependent apoptosis and necroptosis (37, 52–56), we asked if CYLD has any role in the sensitivity of Sharpin^cpdm/cpdm cells to TNF-induced cell death. We first confirmed that Sharpin^cpdm/cpdm (abbreviated to Sharpin^c/c in the labeling of the figures) mouse embryonic fibroblasts (MEF) were more sensitive to TNF-induced necroptosis than their wild-type (WT) counterparts, an effect that was reversed when the Sharpin^cpdm/cpdm MEF were complemented with Sharpin but not with a control gene (SI Appendix, Fig. S1A). Comparison of the Sharpin^cpdm/cpdm MEF complemented with Sharpin versus the control gene showed SHARPIN-deficient cells to be more sensitive to TNF-induced necroptosis in a dose- and time-dependent manner (SI Appendix, Fig. S1 B–D). Furthermore, when both SHARPIN-sufficient and SHARPIN-deficient MEF were stably transfected with the nondegradable IκB superrepressor (I-κBSR) to block NF-κB induction, SHARPIN-deficient MEF remained more sensitive to death compared to SHARPIN-sufficient MEF (SI Appendix, Fig. S1E). Blotting of nuclear extracts from the Sharpin-complemented MEFs transfected with the I-κBSR confirmed the effectiveness of the I-κBSR in blocking translocation of the NF-κB p65 subunit to the nucleus (SI Appendix, Fig. S1F). This indicated that the loss of SHARPIN could sensitize cells to death through a NF-κB–independent mechanism, reminiscent of the effect of mutating the acceptor site on RIPK1 for K63-linked polyubiquitin (36). As an initial test of a potential role for CYLD, we examined the effect of knocking down CYLD on necroptosis in SHARPIN-deficient MEF. Knocking down CYLD significantly reduced the level of necroptosis in response to TNF (SI Appendix, Fig. S1 G and H). This observation suggested that the loss of SHARPIN led to CYLD-dependent death.

These results prompted us to test whether CYLD regulates the phenotype of the Sharpin^cpdm/cpdm mice. To test this hypothesis, we crossed Sharpin^cpdm/cpdm strain with the Cyld^tm1Scs strain (hereafter called Cyld^−/−) to generate Sharpin^cpdm/cpdm Cyld^−/− double-deficient animals. Strikingly, the loss of Cyld significantly ameliorated the spontaneous development of dermatitis in Sharpin^cpdm/cpdm mice (Fig. 1 A and B). Histological analysis showed that ulceration, thickening of the epidermis, and leukocytic infiltration observed in Sharpin^cpdm/cpdm skin were absent in Sharpin^cpdm/cpdm Cyld^−/− skin (Fig. 1C). Whereas Sharpin^cpdm/cpdm mice were smaller and weighed less than age-matched WT mice, sizes and weights of the Sharpin^cpdm/cpdm Cyld^−/− mice and the WT controls did not differ (Fig. 1A and SI Appendix, Fig. S2A). Skin of Sharpin^cpdm/cpdm mice was previously reported to be infiltrated by immune cells including granulocytes and macrophages (19). Our immunohistochemical analysis confirmed the presence of CD45⁺ hematopoietic cells in the skin from Sharpin^cpdm/cpdm mice but not from Sharpin^cpdm/cpdm Cyld^−/− mice (Fig. 1D). We also observed that thickening of the stratum corneum in the inflamed skin of the Sharpin^cpdm/cpdm mice was reversed in the Sharpin^cpdm/cpdm Cyld^−/− mice (SI Appendix, Fig. S2B). Thus, the dermatitis in the Sharpin^cpdm/cpdm mice was ameliorated by the compound loss of Cyld and Sharpin.

Fig. 1.

Cyld is essential for the development of inflammation in Sharpin^cpdm/cpdm mice. (A) Representative 16-wk-old Sharpin^+/+ (two males), Cyld^−/− (two females), Sharpin^cpdm/cpdm (one male and one female), and Sharpin^cpdm/cpdm Cyld^−/− (one male and one female) mice. Sharpin^cpdm/cpdm is labeled Sharpin^c/c throughout the figures. (B) A cohort of 16-wk-old female mice was examined by a pathologist in a double-blind fashion. The clinical score of mice from the indicated genotypes are shown (n = 4 per genotype). One-way ANOVA analysis was performed. *P < 0.05, **P < 0.01. (C) Histology of skin taken from female mice in (B) by hematoxylin and eosin (H&E) staining (100×). Representative photomicrographs from each genotype are shown (Scale bar, 200 μm). E: epidermis; D: dermis; HF: hypodermal fat; PCM: panniculus carnosus muscle; U: ulcerated area; DI: dermal inflammation. (D) Skin sections taken from female mice in (B) were analyzed by immunohistochemistry with anti-CD45 to label hematopoietic cells (100×). CD45+ cells are stained with HRP/AEC (red). Representative photomicrographs from each genotype are shown (Scale bar, 200 μm). Rectangular insert of the Sharpin^cpdm/cpdm skin section is enlarged for visualization of the CD45+ cells (Scale bar, 100 μm).

Lymphoid organs in Sharpin^cpdm/cpdm mice are known to be abnormal (20, 21). Gross examination of immune organs showed that the enlarged spleen, shrunken thymus and mesenteric lymph nodes, and the enlarged liver typical of Sharpin^cpdm/cpdm mice were absent in the Sharpin^cpdm/cpdm Cyld^−/− mice (Fig. 2A and SI Appendix, Fig. S3 A and B). We observed no alterations in the weight of the hearts or the length of the colon in any of the genotypes (SI Appendix, Fig. S3 C and D). Whereas the splenic architecture of the Sharpin^cpdm/cpdm mice showed absence of lymphoid follicles, the Sharpin^cpdm/cpdm Cyld^−/− spleen contained follicles without overt defects, similar to spleens of control Sharpin^+/+ animals (Fig. 2B). While adult Sharpin^cpdm/cpdm mice lack Peyer's patches in their small intestine (20, 57), we observed normal Peyer’s patches in the Sharpin^cpdm/cpdm Cyld^−/− mice (Fig. 2 C and D). The lack of lymphoid tissue abnormalities in the Sharpin^cpdm/cpdm Cyld^−/− mice resembles that previously reported in Sharpin^cpdm/cpdm Casp8^+/− Ripk3^−/− mice, including the presence of Peyer’s patches (24). Since CASPASE 8 and RIPK3 propagate death signals, the similarity in the phenotype of the two strains is consistent with CYLD functioning as a death-signaling molecule. While Sharpin^cpdm/cpdm mice showed lower steady-state serum IgG and IgA in confirmation of previous findings by others (20), serum from the Sharpin^cpdm/cpdm Cyld^−/− mice contained normal levels of IgG and IgA (Fig. 2E). Similarly, we observed higher serum IgE level in the Sharpin^cpdm/cpdm mice, which was reversed in the Sharpin^cpdm/cpdm Cyld^−/− mice (Fig. 2E). Consistent with the lack of Peyer's patches, adult Sharpin^cpdm/cpdm mice have no fecal IgA (20), but this was also reversed in the Sharpin^cpdm/cpdm Cyld^−/− mice (Fig. 2F). Whereas several cytokines including IFNG, IL5, and IL12/23p40 were elevated in the serum of Sharpin^cpdm/cpdm mice compared to WT controls, these elevations were abrogated in the Sharpin^cpdm/cpdm Cyld^−/− mice (Fig. 2G). More modest changes in serum IL2 and IL6 but not eotaxin and CCL3 (MIP-1α) were observed in the Sharpin^cpdm/cpdm mice (SI Appendix, Fig. S3 E–H). Together, these data demonstrate that CYLD is critical for eliciting the multiorgan inflammation and abnormal lymphoid tissues observed in the Sharpin^cpdm/cpdm mice, suggesting dysregulated CYLD activity in the absence of SHARPIN.

Fig. 2.

Cyld deletion reverses lymphoid tissue defects in Sharpin^cpdm/cpdm mice. (A) Gross analysis of spleens, mesenteric lymph nodes (MLN), and thymi obtained from 16-wk-old mice in Fig. 1A. Spleens from two individual mice of each genotype are shown. (B) Histology of spleens taken from the mice shown in Fig. 1A with the indicated genotypes by H&E staining (100×). Representative photomicrographs from each genotype are shown (Scale bar, 200 μm). (C) H&E staining of Swiss rolls of small intestines obtained from mice shown in Fig. 1A with the indicated genotypes. White arrows indicate Peyer's patches. Images of individual Peyer's patches are shown Below. Note Sharpin^cpdm/cpdm mice lack Peyer’s patches. (D) The number of Peyer's patches was enumerated in each mouse of the different genotypes. One-way ANOVA analysis was performed. *P < 0.05, **P < 0.01. (E) Levels of serum IgG, IgA, IgE, and IgM from 10- to 16-wk-old mice of both sexes were measured by Luminex Multiplex assays. Each data point is from an individual mouse (n = 6 to 8 mice per genotype). Error bars represent mean ± SD, and Student's t test was performed. *P < 0.05, **P < 0.01, ***P < 0.001. (F) Stool IgA level in mice of both sexes from the indicated genotypes were analyzed by enzyme-linked immunosorbent assay (ELISA). Each data point is from an individual mouse (n = 6 to 8 mice per genotype). Error bars represent mean ± SD, and Student's t test was performed. *P < 0.05, **P < 0.01, ***P < 0.001. (G) Levels of serum IFNG, IL5, and IL12/23p40 from 10- to 16-wk-old mice of both sexes were measured by Luminex Multiplex assays. Each data point is from an individual mouse (n = 9 to 12 mice per genotype). Error bars represent mean ± SD, and Student's t test was performed. *P < 0.05, **P < 0.01.

CYLD Phosphorylation Is Disrupted by SHARPIN Deficiency.

The results in Fig. 1 and Fig. 2 showed a clear genetic interaction between SHARPIN and CYLD, but the mechanistic relationship between the two molecules was unknown. Following TNFR1 ligation, an array of signaling scaffold and effector molecules including LUBAC, cIAP1/2, TRAF2, and IKK are recruited to TNFR1. This receptor complex, also known as Complex I (58), propagates a survival signal. CYLD can also be recruited to the receptor complex via a SPATA2-dependent association with RNF31/HOIP (47–51), but how defects in linear ubiquitin could lead to CYLD activation has not previously been determined. CYLD is regulated by posttranslational mechanisms, and cleavage by CASP8 and MALT1 have been reported to regulate its function (56, 59). Another potential regulatory mechanism may be phosphorylation. CYLD phosphorylation at a cluster of serine residues around Ser-418 carried out by IKK-β in a stimulus-dependent fashion is known to suppress its enzymatic activity (46). Phosphorylation of CYLD at Ser-418 by IKK-ε was also reported to have the same effect (60). Since SHARPIN-deficient cells were reported to have diminished IKK activity (8), we postulated that SHARPIN deficiency would also lead to diminished CYLD phosphorylation, which would result in a more active CYLD to initiate cell death. To test this, we examined the phosphorylation kinetics of CYLD in primary adult dermal fibroblast (ADF). Using an antibody that detects phosphorylated Ser-418 of CYLD, we observed an induction in CYLD phosphorylation in Sharpin^+/+ cells and observed that this phosphorylation was diminished in Sharpin^cpdm/cpdm cells (Fig. 3A). To confirm that the defect in CYLD regulation is caused specifically by SHARPIN deficiency, we examined the SHARPIN-complemented MEF described in SI Appendix, Fig. S1. Complementation of Sharpin^cpdm/cpdm cells with Sharpin but not with a control gene restored CYLD phosphorylation (Fig. 3B). Consistent with the previous study that IKK-β phosphorylates CYLD (46), blockade of IKK-β with the highly selective chemical inhibitor [5-(p-Fluorophenyl)-2-ureido]thiophene-3-carboxamide (TPCA-1) resulted in reduced CYLD phosphorylation in Sharpin^+/cpdm MEF, similar to that observed in Sharpin^cpdm/cpdm MEF (Fig. 3C). To extend our observations beyond fibroblasts, we examined splenic B cells. Sharpin^cpdm/cpdm B cells also showed diminished CYLD phosphorylation when compared to their WT counterparts (Fig. 3D). However, there is a significant reduction in full-length CYLD, accompanied by the presence of smaller fragments of CYLD in the Sharpin^cpdm/cpdm B cells. Likewise in total splenocytes, there is a loss in full-length CYLD and appearance of smaller fragments in the Sharpin^cpdm/cpdm splenocytes. These observations suggest that in some cell types, SHARPIN deficiency may also lead to altered proteolytic processing of CYLD. In addition to attenuated IKK activity in SHARPIN-deficient cells, another explanation for reduced CYLD phosphorylation could be defective CYLD recruitment to the TNFR1 complex due to the instability of HOIP in the absence of SHARPIN. We immunoprecipitated the receptor after 5 and 15 min of stimulation with FLAG-TNF and observed a modest diminution in receptor-associated CYLD in SHARPIN-deficient MEF (SI Appendix, Fig. S4).

Fig. 3.

CYLD phosphorylation is impaired by SHARPIN deficiency. (A) ADFs from Sharpin^+/+ and Sharpin^cpdm/cpdm mice were treated with mTNF (100 ng/mL) for the indicated minutes. CYLD phosphorylation was detected by blotting with anti–phospho-CYLD (Ser-418), followed by blotting for total CYLD, SHARPIN, and β-ACTIN. (B) Sharpin^+/cpdm, Sharpin^cpdm/cpdm, and Sharpin^cpdm/cpdm MEF complemented with Sharpin or control gene were stimulated with mTNF (100 ng/mL) for the indicated times. CYLD phosphorylation was analyzed as in (A). (C) Sharpin^+/cpdm and Sharpin^cpdm/cpdm MEF were stimulated with 100 ng/mL mTNF for 10 min in the presence of 0 to 50 μM TPCA-1, a highly selective inhibitor against IKK-β. CYLD phosphorylation was analyzed as in (A). (D) Purified B cells from Sharpin^+/+ and Sharpin^cpdm/cpdm mice were treated with mTNF (100 ng/mL) for the indicated minutes. CYLD phosphorylation was analyzed as in (A). N-terminal proteolytic fragments of CYLD are indicated. (E) Total splenocytes from Sharpin^+/+ and Sharpin^cpdm/cpdm mice were treated with mTNF (100 ng/mL) for the indicated minutes. CYLD phosphorylation was analyzed as in (A). N-terminal proteolytic fragments of CYLD are indicated. Data in panels A–E are representative of at least three independent experiments conducted for each panel.

SHARPIN-Deficient Cells Are More Susceptible to CYLD-Mediated Cell Death.

Since our earlier result in MEF showed a role for SHARPIN and CYLD in regulating cell death (SI Appendix, Fig. S1), we examined the sensitivity of ADF obtained from Sharpin^+/+, Sharpin^cpdm/cpdm, Cyld^−/−, and Sharpin^cpdm/cpdm Cyld^−/− mice to TNF-induced cell death (Fig. 4A). These assays showed higher susceptibility of Sharpin^cpdm/cpdm ADF to TNF-induced apoptosis and necroptosis when compared to their Sharpin^+/+ counterpart. In contrast, we observed abolishment of this increased susceptibility to both forms of death in Sharpin^cpdm/cpdm Cyld^−/− cells (Fig. 4A), consistent with observations obtained earlier by knocking down CYLD (SI Appendix, Fig. S1). Similarly, Sharpin^cpdm/cpdm ADF exhibited enhanced apoptotic markers (i.e., cleaved CASP8, CASP3, and PARP) and a necroptotic marker (i.e., phospho-MLKL) when compared to Sharpin^+/+ ADF and reduction of these death signatures in the Sharpin^cpdm/cpdm Cyld^−/− ADF (Fig. 4 B and C and SI Appendix, Fig. S5 A and B). Experiments performed with MEF showed comparable results (SI Appendix, Fig. S6 A–E). Together, the findings support the conclusion that enhanced sensitivity to cell death in SHARPIN-deficient cells is dependent on CYLD. CYLD initiates death signaling by removing K63-linked ubiquitin chains from RIPK1, thereby converting RIPK1 from a survival-signaling to a death-signaling effector (33, 56, 61). This conversion is biochemically detected by the translocation of RIPK1 to the FADD/CASP8-containing Complex II. We therefore tested whether cells that lack functional SHARPIN show 1) more conversion of RIPK1 to its death-signaling form and 2) the effect is dependent on CYLD. In ADF treated with TNF, in the presence of a combination of cycloheximide and zVAD-fmk to stabilize Complex II, we detected RIPK1 in the FADD pulldowns of Sharpin^cpdm/cpdm ADF more rapidly than in Sharpin^+/+ cells (Fig. 4D). The assays showed abrogation of RIPK1 translocation to the Complex II in Sharpin^cpdm/cpdm Cyld^−/− ADF (Fig. 4D), demonstrating that CYLD is necessary for the conversion of RIPK1 to a death-signaling molecule when SHARPIN is functionally absent. Studies using MEF also revealed that RIPK1 translocation to Complex II is CYLD dependent (SI Appendix, Fig. S6F). We confirmed expression, or lack therefore, of SHARPIN and CYLD in the MEF of the indicated genotype (SI Appendix, Fig. S6G). As an additional confirmation, we reconstituted Sharpin^cpdm/cpdm Cyld^−/− MEF with retroviral vectors encoding SHARPIN only, CYLD only, or both molecules together (SI Appendix, Fig. S6H). These reconstituted MEF showed similar data (SI Appendix, Fig. S6I) to that observed previously in ADF (Fig. 4D) and MEF (SI Appendix, Fig. S6F).

Fig. 4.

Sharpin^cpdm/cpdm cells are more susceptible to CYLD-mediated cell death. (A) Sharpin^+/+, Sharpin^cpdm/cpdm, Cyld^−/−, and Sharpin^cpdm/cpdm Cyld^−/− ADF were treated with mTNF (10 ng/mL) for 24 h in the presence or absence of zVAD (20 μM) to induce necroptosis or cycloheximide (CHX, 1 μg/mL) to induce apoptosis. Cell death was analyzed by propidium iodide staining and flow cytometry. Data from three independent experiments are shown. Error bars represent mean ± SD, and one-way ANOVA analysis was performed. ***P < 0.001 Sharpin^cpdm/cpdm versus Sharpin^+/+ ADF. P value of Sharpin^cpdm/cpdm Cyld^−/− versus Sharpin^+/+ ADF is not significant. (B) Sharpin^+/+, Sharpin^cpdm/cpdm, Cyld^−/−, and Sharpin^cpdm/cpdm Cyld^−/− ADF were stimulated with mTNF (100 ng/mL) in the presence of CHX (1 μg/mL) for 3 and 6 h. Apoptosis was examined by blotting for cleaved CASP8 and cleaved CASP3 (C). For each panel, data shown are representative of at least three independent experiments. (C) Necroptosis was analyzed by blotting for phospho-MLKL (Ser-345) in Sharpin^+/+, Sharpin^cpdm/cpdm, Cyld^−/−, and Sharpin^cpdm/cpdm Cyld^−/− ADF stimulated with mTNF (25 ng/mL) in the presence of zVAD (20 μM) for 1.5 and 3 h. Result shown is representative of at least three independent experiments. (D) Sharpin^+/+, Sharpin^cpdm/cpdm, Cyld^−/−, and Sharpin^cpdm/cpdm Cyld^−/− ADF were stimulated with mTNF (100 ng/mL) in the presence of CHX (1 μg/mL) and zVAD (20 μM) for 2 and 4 h. The death-signaling Complex II was isolated by FADD immunoprecipitation and sequentially blotted for RIPK1 and FADD. Experiment was repeated twice with similar results. (E) Cyld^+/+ and Cyld^−/− MEFs transfected with small guide RNA (sgRNA) that is nontargeting (WT) or Sharpin-targeting (KO) were stimulated with mTNF (100 ng/mL) for 0, 5, and 15 min. Lysates were subjected to Flag-K63-TUBE pulldown to isolate K63-ubiquitinated proteins followed by blotting for RIPK1. Whole-cell lysates were also blotted for RIPK1 and ACTIN. Result shown is representative of three independent experiments. (F) Residues 411 to 445 of WT mouse CYLD encompassing the cluster of seven serines equivalent to those reported to be phosphorylated on human CYLD, and the corresponding ser-to-ala mutations in the nonphosphorylatable CYLD-S7A mutant are depicted. Ser-417 of mouse CYLD is equivalent to Ser-418 of human CYLD. Cyld^−/− MEF were transduced with retroviruses encoding a control (triangles), WT CYLD (circles), or CYLD-S7A (squares). Transduced cells plated in triplicates were treated with media or 0.1 ng/mL mTNF plus 20 μM zVAD-fmk and imaged for 24 h in the presence of YOYO-3. Aliquots of the complemented MEF were also analyzed by Western blotting to demonstrate defective phosphorylation in the mutant CYLD but equivalent expression of WT and mutant CYLD. Error bars represent mean ± SD from three biological replicates, and paired Student’s t test was performed. *P < 0.05.

As a complementary approach, we used CRISPR-Cas9 to knock out Sharpin in previously generated Cyld^+/+ and Cyld^−/− MEF (62). We then analyzed the sensitivity of these cell lines to TNF-induced apoptosis using the IncuCyte S3 instrument with a fluorogenic CASP3/7 substrate. Knockout of Sharpin sensitized Cyld^+/+ MEF to TNF-induced apoptosis as evidenced by the time-dependent increase in caspase activity, whereas knockout of Sharpin did not lead to apoptosis in the Cyld^−/− MEF (SI Appendix, Fig. S7 A and B). The assays showed that enhanced sensitivity of the SHARPIN-deficient cells to apoptosis was accompanied by enhanced recruitment of RIPK1 to Complex II and that this required CYLD (SI Appendix, Fig. S7C). Therefore, multiple lines of evidence point toward CYLD playing a critical role in the susceptibility of cells to cell death in the absence of functional SHARPIN.

While RIPK1 recruitment to the death-signaling Complex II is clearly regulated by SHARPIN and CYLD, it is not clear whether ubiquitination level of RIPK1 could account for the differing sensitivity of the different cell lines to death. We first analyzed this by immunoprecipitating the TNFR1 complex with anti-FLAG beads following 5 and 15 min of stimulation with FLAG-TNF and blotting with anti-RIPK1. We observed minimal alterations in ubiquitinated RIPK1 in the receptor complex in cells of the different genotypes without a clear correlation to their cell death sensitivity phenotype (SI Appendix, Fig. S8A). Examinations of linear ubiquitin level in the receptor complex were also inconclusive, as they did not reveal any correlation with sensitivity/resistance to cell death (SI Appendix, Fig. S8B). Since the SHARPIN-deficient MEF generated by CRISPR exhibited a mild defect in the level of linearly ubiquitinated proteins recruited to Complex I (SI Appendix, Fig. S8B), we compared this response to SHARPIN-deficient MEF generated from Sharpin^cpdm/cpdm mice. Under our experimental conditions, SHARPIN-deficient MEF generated by CRISPR or from a Sharpin^cpdm/cpdm mouse were comparable, and both displayed a mild defect (SI Appendix, Fig. S8C). Moquin et al. had previously concluded that CYLD does not function at the TNFR1 receptor complex but deubiquitinates RIPK1 in the cytosolic necrosome during necroptosis (63), suggesting that examining ubiquitination of RIPK1 in the nonreceptor pool may be more informative. Thus, we directly analyzed K63-linked ubiquitination of RIPK1 in the total detergent-soluble pool using K63-TUBE affinity isolation followed by blotting for RIPK1 (Fig. 4E). Interestingly, a pronounced cluster of ubiquitinated RIPK1 migrating above the 130-kDa marker appears in the SHARPIN-deficient cells after 15 min of TNF stimulation when compared against WT cells, suggesting that editing of ubiquitin linkages on RIPK1 may be occurring in the absence of SHARPIN. This cluster of ubiquitinated RIPK1 was not observed in the SHARPIN and CYLD double-deficient cells, indicating that the editing is CYLD dependent. In addition, we examined the possibility that highly ubiquitinated RIPK1 species may not be migrating into the resolving gel by including the stacking gel in our transfer to the nitrocellulose membrane. Significant amounts of RIPK1 from TNF-stimulated CYLD-deficient cell lines remain in the stacking gel, indicating that in the absence of CYLD, a significant proportion of RIPK1 remains highly ubiquitinated and cannot migrate into the resolving gel. Thus, the increased sensitivity of SHARPIN-deficient cells to death correlated with enhanced editing of ubiquitin chains on RIPK1 that is dependent on CYLD. Conversely, the failure of the CYLD-deficient cells to undergo cell death could be explained by the preservation of large ubiquitin chains on RIPK1.

If Sharpin^cpdm/cpdm cells are more sensitive to death due to defective CYLD phosphorylation, then a nonphosphorylatable CYLD mutant would be predicted to mimic the SHARPIN deficiency (i.e., more sensitive to cell death). Mutations in the cluster of seven serines around residue 418 of human CYLD was shown to result in a gain-of-function in deubiquitinase activity (46). We generated an analogous mutant of mouse CYLD with seven serine-to-alanine substitutions (CYLD-S7A) and complemented Cyld^−/− MEF with either a control gene, WT, or mutant CYLD. The complemented cells were stimulated in the presence of zVAD-fmk in order to prevent CYLD cleavage by CASP8 (56) and thus negate the potential confounding issue of differential cleavage of CYLD-WT versus CYLD-S7A by CASP8. Under this stimulation condition, CYLD-complemented cells undergo necroptosis (56). We induced necroptosis with 0.1 ng/mL TNF plus 20 μM zVAD-fmk and analyzed death using the cell-impermeant nucleic acid stain YOYO-3 iodide in the IncuCyte instrument (64). Consistent with our hypothesis, these assays showed that CYLD-S7A–expressing cells are more sensitive to necroptosis than their CYLD-WT counterparts (Fig. 4F). The enhanced sensitivity of CYLD-S7A–complemented cells to cell death is consistent with the notion that phosphorylation and therefore inhibition of CYLD is a key regulatory event that suppresses cell death in response to TNF. Our findings imply dysregulated CYLD activity in SHARPIN-deficient cells.

CYLD-Mediated Death of Myeloid Cells Causes Skin Inflammation.

The leukocytic infiltration in the skin of the Sharpin^cpdm/cpdm mice consists of granulocytes and macrophages (19), prompting us to examine the role of macrophages in the pathogenesis of skin lesions in these mice. We recently reported that macrophages are highly susceptible to CYLD-dependent autonecroptosis mediated by the TNF that is produced in response to TLR4 ligation (61). Bone marrow–derived macrophages (BMDM) deficient in SHARPIN showed a defect in CYLD phosphorylation when stimulated with TNF (Fig. 5A), albeit the defect was modest. We also observed a similar modest defect when these cells were stimulated with lipopolysaccharides (LPS) to activate TLR4 or polyIC (pIC) to activate TLR3 (Fig. 5 B and C). Therefore, we tested to see if these macrophages also exhibit altered necroptosis. In response to low-dose LPS that elicits TNF-dependent necroptosis (61), we observed that Sharpin^cpdm/cpdm BMDM were more sensitive to necroptosis compared to their WT counterparts, and this was abrogated in the Sharpin^cpdm/cpdm Cyld^−/− BMDM (Fig. 5D). We observed a similar effect using pIC to trigger necroptosis (Fig. 5E). Analysis of Complex II formation showed that RIPK1 was detected in FADD pulldowns of Sharpin^cpdm/cpdm BMDM more rapidly than in Sharpin^+/+ cells treated with TNF (Fig. 5F). Similar to that observed previously in ADF (Fig. 4D), the pulldown assays showed abrogation of RIPK1 association with Complex II in Sharpin^cpdm/cpdm Cyld^−/− BMDM.

Fig. 5.

Sharpin^cpdm/cpdm myeloid cells are more susceptible to CYLD-mediated cell death. (A–C) Sharpin^+/+ and Sharpin^cpdm/cpdm BMDM were stimulated with (A) 100 ng/mL mTNF, (B)10 ng/mL LPS, and (C) 2.5 μg/mL polyIC for the indicated times. CYLD phosphorylation was analyzed as in Fig. 3A. Semiquantitative analysis of the Western blots was conducted using Image J. Values shown under the blots are the ratio of the signal of the pS418-CYLD band relative to the signal of the total CYLD band. Blots shown are representative of three independent experiments performed with each stimulus. (D and E) Sharpin^+/+, Sharpin^cpdm/cpdm, Cyld^−/−, and Sharpin^cpdm/cpdm Cyld^−/− BMDM were stimulated with the indicated doses of (D) LPS and (E) polyIC in the presence of dimethyl sulfoxide (DMSO) or zVAD (20 μM) for 24 h. Cell death was analyzed as in Fig. 4A. Error bars represent mean ± SD, and one-way ANOVA analysis was performed. ***P < 0.001 Sharpin^cpdm/cpdm versus Sharpin^+/+ BMDM. P value of Sharpin^cpdm/cpdm Cyld^−/− versus Sharpin^+/+ BMDM is not significant. (F) Sharpin^+/+, Sharpin^cpdm/cpdm, Cyld^−/−, and Sharpin^cpdm/cpdm Cyld^−/− BMDM were stimulated with mTNF (100 ng/mL) in the presence of CHX (0.25 μg/mL) and zVAD (20 μM) for 4 and 6 h. Translocation of RIPK1 to Complex II was analyzed as in Fig. 4D. Result shown is representative of three independent experiments performed.

The genetic crosses conducted using germline knockouts showed CYLD to be the cause of the phenotype observed in Sharpin^cpdm/cpdm mice. However, the contribution of CYLD-dependent death in specific cellular compartments to the Sharpin^cpdm/cpdm phenotype is not clear. The data in Fig. 5 A–F prompted us to ask whether CYLD-dependent macrophage death has any role in the pathogenesis of the Sharpin^cpdm/cpdm mice. We crossed the Sharpin^cpdm/cpdm strain to the LysM-cre–driven conditional knockout of Cyld (Cyld^M-KO) we previously generated (61). The Sharpin^cpdm/cpdm Cyld^M-KO mice showed significant improvement in their dermatitis compared to the Sharpin^cpdm/cpdm mice, which develop discernable dermatitis by week 8 in our colony (Fig. 6A). In our cohort, two-thirds of Sharpin^cpdm/cpdm Cyld^M-KO mice did not show skin inflammation by week 16 while the remaining one-third had delayed progression (SI Appendix, Fig. S9A). Histological analysis showed that skin of the Sharpin^cpdm/cpdm Cyld^M-KO mice lacked the typical ulceration, thickening of the epidermis, and leukocytic infiltration observed in Sharpin^cpdm/cpdm skin (Fig. 6B). However, when we examined the spleen and thymus, the defects in these organs of Sharpin^cpdm/cpdm mice were not reversed in the Sharpin^cpdm/cpdm Cyld^M-KO mice (Fig. 6 C and D). We confirmed that BMDM generated from Sharpin^cpdm/cpdm Cyld^M-KO mice were more resistant to cell death than those from Sharpin^cpdm/cpdm mice and that the LysM-cre was effective in knocking out CYLD in our BMDM cultures (SI Appendix, Fig. S9 B–D). We also crossed the Sharpin^cpdm/cpdm strain to a Cx3cr1-cre–driven conditional knockout of Cyld (Cyld^MP-KO), which targets mononuclear phagocytes including macrophages (65). The Cx3cr1-cre–mediated deletion of Cyld recapitulated the effect observed earlier with LysM-cre (SI Appendix, Fig. S10 A–F). Results from the use of two different cre strains that can delete Cyld in macrophages led us to conclude that CYLD-dependent macrophage cell death may be a major driver of the inflammatory response in the skin of the Sharpin^cpdm/cpdm mice.

Fig. 6.

CYLD-mediated death of myeloid cells is necessary to cause skin inflammation in Sharpin^cpdm/cpdm mice. (A) Photograph of 10- to 16-wk-old Sharpin^cpdm/cpdm Cyld^M-KO mice in comparison to a 9-wk-old Sharpin^cpdm/cpdm mouse. The 13- and 14-wk-old mice were males, the rest were females. (B) H&E staining of skin sections from 16-wk-old Sharpin^cpdm/cpdm and Sharpin^cpdm/cpdm Cyld^M-KO mice (100×) (Scale bar, 200 μm). Top arrow: orthokeratotic hyperkeratosis and serocellular crust; Middle arrow: thickened epidermis; Bottom arrow: thickened dermis. (C) Gross analysis of spleens, mesenteric lymph nodes, and thymi from 16-wk-old Sharpin^+/+, Sharpin^cpdm/cpdm, and Sharpin^cpdm/cpdm Cyld^M-KO mice. (D) H&E staining of spleen sections from 16-wk-old Sharpin^+/+ and Sharpin^cpdm/cpdm Cyld^M-KO mice (Scale bar, 200 μm). (E) Model of CYLD suppression by linear ubiquitin. In WT cells, the LUBAC-dependent IKK phosphorylates CYLD to inactivate it. LUBAC could also suppress CYLD via a phosphorylation-independent mechanism. These mechanisms suppress CYLD, thereby keeping RIPK1 in a survival mode. In addition, IKK can directly phosphorylate RIPK1 to inhibit its death-signaling function. Linear ubiquitin cooperates positively with K63-linked ubiquitin to maintain cell survival by preventing the dismantling of K63-linked ubiquitin chains. In SHARPIN-deficient cells, there is diminished suppression of CYLD. Therefore, CYLD is active and removes K63-linked ubiquitin chains from RIPK1. This enables RIPK1 to associate with Complex II to induce cell death leading to multiorgan inflammation, abnormal lymphoid tissues, and immunodeficiency.

Discussion

This study now provides genetic data that adds insight into the mechanisms underlying the Sharpin^cpdm/cpdm phenotype. We propose that in WT cells, SHARPIN-dependent regulation of CYLD suppresses its function, allowing K63-linked ubiquitin chains on RIPK1 to be sustained, thus preventing RIPK1 from becoming a death-signaling molecule (Fig. 6E, Left). In the absence of SHARPIN, there is a failure to suppress CYLD, which dismantles K63-linked ubiquitin chains from RIPK1 to initiate the RIPK1-dependent death cascade (Fig. 6E, Right). We note that there is reduced phosphorylation of serine 418 of CYLD in SHARPIN-deficient cells, which is known to be phosphorylated by IKK-β and IKK-ε (46, 60). Reduced CYLD phosphorylation could be due to reduced IKK activity in SHARPIN-deficient cells (7–9). In addition, CYLD is recruited to the receptor complex in a SPATA2 and RNF31/HOIP-dependent manner (47–51), and since Sharpin^cpdm/cpdm cells have reduced RNF31/HOIP levels (7–9), this could reduce CYLD recruitment to the receptor complex and subsequent phosphorylation by IKK. Since phosphorylation at the cluster of serines around residue 418 inhibits its enzymatic activity (46, 60), defective phosphorylation may enhance CYLD’s editing of ubiquitin chains on RIPK1. This model is supported by the CYLD-dependent alteration in RIPK1 ubiquitination pattern observed in the SHARPIN-deficient cells. A role for CYLD in trimming polyubiquitin chains is further supported by the presence of high–molecular-weight RIPK1 that fail to migrate into the resolving gel in CYLD-deficient cells.

In addition to phosphorylation, CYLD can also be negatively regulated by proteolysis mediated by CASP8 and MALT1 (56, 59). In some cell types (e.g., B cells), there is reduced full-length CYLD accompanied by the presence of smaller CYLD fragments in SHARPIN-deficient cells. The role of the increased proteolytic processing of CYLD in regulating the sensitivity of SHARPIN-deficient cells to death is unclear at present. In mix population (e.g., splenocytes), it is possible that changes in CYLD expression in SHARPIN-deficient cells could be due to changes in the cellular composition of the spleens of Sharpin^cpdm/cpdm mice (20). Furthermore, CYLD can be phosphorylated by multiple members of the IKK family including IKK-β, IKK-ε, and TBK1, suggesting that there may be redundancy in regulating CYLD phosphorylation in different cell types. This redundancy could explain the modest defect in CYLD phosphorylation in the Sharpin^cpdm/cpdm BMDM as macrophages express IKK-ε and TBK1 (66). The interrelationship between these multiple posttranslational mechanisms in regulating CYLD remains to be fully characterized. In addition, a role for SHARPIN in regulating CYLD activity via an alternative mechanism independent of phosphorylation remains a possibility. The IKK kinases also directly phosphorylate RIPK1 to inhibit the death-signaling function of RIPK1 (28–32). Thus, the emerging picture is that a fully functional Checkpoint 1 needed to suppress ripoptocide is dependent on multiple posttranslational modifications of several molecules.

These observations point toward CYLD suppression as central to immune homeostasis as its perturbation, observed in the Sharpin^cpdm/cpdm mice, leads to multiorgan inflammation and disruption of lymphoid tissues. It also demonstrates that CYLD functions as a proinflammatory molecule since its deletion resolves the inflammation seen in the Sharpin^cpdm/cpdm mice. These aspects of CYLD in the proper functioning of the immune system have not been appreciated to date. Previous studies with CYLD have described it as having an anti-inflammatory function (67–69), primarily because of its inhibitory effect on the NF-κB pathway. We now show that CYLD can be highly proinflammatory via its death-inducing function, which is revealed upon the loss of an upstream inhibitory mechanism (e.g., SHARPIN/LUBAC). The notion that CYLD mediates RIPK1-dependent death-signaling is supported further by the observation that deleting Cyld in Sharpin^cpdm/cpdm mice has the same effect on dermatitis as inactivating the kinase activity of Ripk1 (Ripk1^K45A) (25), deletion of a single copy of Ripk1 (26), the combined deletion of Ripk3 and a single copy of Casp8 (24), or combined deletion of Ripk3 and epidermal deletion of Fadd or Tradd (23). The reversal of extracutaneous systemic inflammation in the Sharpin^cpdm/cpdm mice by Cyld deletion, inhibition of the kinase activity of Ripk1 (25), and a compound deletion of Ripk3 and Casp8 (24), is further indication that these molecules function in the same biochemical pathway. Genetic crosses of Sharpin^cpdm/cpdm mice with other null strains have also been conducted. For instance, Sharpin^cpdm/cpdm Casp1^−/− Casp11^−/− mice do not developed dermatitis, but inflammation remains in the extracutaneous organs (70, 71), suggesting that the inflammasome either plays a direct role downstream of SHARPIN in a cell-autonomous manner only in skin tissue, or it has an accessory role.

Human patients with mutations in the RNF31/HOIP and RBCK1/HOIL1 subunits of LUBAC have immune defects that overlap with the Sharpin^cpdm/cpdm mice (17, 18). Patients with loss-of-function mutations in OTULIN/FAM105b, the deubiquitinase specific for linear ubiquitin, also develop systemic autoinflammation (72, 73). Mice with germline defects in Otulin are embryonic lethal (74, 75). Mechanistic studies using an inducible expression of a catalytically inactive mutant Otulin^iC129A in adult mice indicate that the autoinflammation is in part due to excessive cell death caused by hyper-ubiquitination and disruption in the function of LUBAC (74). Whether CYLD plays a role in the lesions observed in these patients with defects in linear ubiquitin modification remains to be determined.

Excessive CYLD-dependent cell death in the Sharpin^cpdm/cpdm mice can cause skin inflammation due to 1) death of epithelial cells that could result in cellular erosion and ulcers in and of itself, 2) loss of barrier function in the epithelial layer and subsequent breach by commensal microbiota and environmental contaminants (bedding and food), and/or 3) excessive cellular debris from dying cells that are not properly cleared by phagocytes. The reversal of skin inflammation in Sharpin^cpdm/cpdm mice following deletion of Fadd or Tradd in keratinocytes indicates that cell death in this cellular compartment is critical for inflammation (23). Our observation that skin inflammation was reduced by deletion of Cyld in macrophages suggests that death of macrophages may amplify the inflammatory response initiated by keratinocyte cell death. In summary, the Sharpin^cpdm/cpdm mouse illustrates the pathophysiology when the TNF response switches from cell survival to cell death by virtue of a genetic defect in CYLD regulation. The conserved cell-killing function of TNF has been postulated to serve an evolutionary role against microbial infection (41, 76). The analysis of the Sharpin^cpdm/cpdm mice suggests that a biological consequence of TNF/CYLD-mediated ripoptocide is inflammation and could serve an antimicrobial function that is beneficial to the host (31, 77).

Materials and Methods

Mice.

Sharpin^cpdm/cpdm Cyld^−/− mice were generated by crossing Sharpin^cpdm/cpdm (C57BL/KaLawRij-Sharpin^cpdm/RijSunJ) provided by Dr. John Sundberg with Cyld^−/− mice (B6;129S-Cyld^tm1Scs) provided by Dr. Shao-Cong Sun (78). Experiments were conducted using Sharpin^+/+, Sharpin^cpdm/cpdm, and Cyld^−/− littermates as controls. Sharpin^cpdm/cpdm Cyld^M-KO mice were generated by crossing Sharpin^cpdm/cpdm strain with the Cyld^flox/flox (B6.129S-Cyld^tm1.1Attg) x LysM-cre (B6.129P2-Lyz2^tm1(cre)Ifo/J) strain previously described (61). Sharpin^cpdm/cpdm Cyld^MP-KO mice were generated by crossing Sharpin^cpdm/cpdm strain with the Cyld^flox/flox (B6.129S-Cyld^tm1.1Attg) x Cx3cr1-cre (B6J.B6N(Cg) Cx3cr1cre^tm1.(cre)Jun/J) strain. The Cx3cr1-cre was provided by Dr. Sergio Lira. All experiments involving the use of animals were performed in agreement with approved protocols by the Institutional Animal Care and Use Committee at the Icahn School of Medicine at Mount Sinai.

Clinical Scoring.

The gross scoring scheme was modified from a system developed for a different mouse disease, ulcerative dermatitis, that occurs in C57BL/6 substrains (79). The clinical score combines gross evaluation with histologic scoring from skin sections. In brief, four criteria were evaluated: character of lesion (0 = none, 1 = alopecia or excoriations only, or 1 small punctuate crust, 2 = multiple, small punctuate crusts or coalescing crusts (>2 mm), regions affected (head/cervical, thoracic, and/or abdominal/caudal; 0 = none, 1 = <25%, 2 = 25 to 50%, 3 = >50%). The “regions affected” score in Hampton et al. (79) was based on the progression of the disease. Ulcerative dermatitis in B6 mice often started in the intercapsular/dorsal back area (region 2) and progressed, thus encroachment on the head and face (region 1) was considered to be a severe score. In this study, the lesions often start in the ventral chin/neck region (region 1) and progress to the thorax, abdomen, and back. The scoring system was modified to increase in severity based on the number of regions affected rather than the specific region affected because of this difference in progression of skin disease. For the histologic lesions, scoring was a subjective: 0 = none, 1 = mild, 2 = moderate, and 3 = marked scale. The scoring of inflammation in the dermatitis was done using the following scale: 0 = none, 1 = mild (focal or few foci of inflammation in superficial dermis), 2 = moderate (multiple foci with inflammation in deep dermis, with or without ulcerations), and 3 = severe (regionally extensive with inflammation extending to the subcutis and ulceration).

Cell Death Assays.

Cellular death analyses were performed by Annexin V or propidium iodide staining and flow cytometry as previously described (56). Cell death was also quantified using the CellTiter-Glo Luminescent Cell Viability Assay (Promega). Biological triplicates were used in the experiments for ADF and BMDM cell death analysis. For real-time analysis of apoptosis, MEF were seeded at 7,500 cells/well in a clear flat-bottom 96-well plate in complete media. At 24 h later, the complete media was replaced with phenol red-free complete media containing CellEvent Caspase-3/7 Green Detection Reagent. After 30 min of incubation, triplicate cells were treated with TNF and analyzed in an IncuCyte S3 (Essen Biosciences) instrument for 24 h. For real-time analysis of necroptosis, cells were stimulated with TNF in the presence of 20 zVAD-fmk and cultures imaged in the presence of the cell-impermeant nucleic acid stain YOYO-3 iodide (64).

Statistical Analysis.

Statistics were performed using one-way ANOVA, Student's t test, or Mann–Whitney U test. (*P < 0.05, **P < 0.01, ***P < 0.001). Analysis was carried out using Prism 8 (GraphPad). Additional materials and methods are provided in SI Appendix.

Data Availability

All study data are included in the article and/or SI Appendix.