Activity of Protein Kinase RIPK3 Determines Whether Cells Die by Necroptosis or Apoptosis

See allHide authors and affiliations

Science  21 Mar 2014:
Vol. 343, Issue 6177, pp. 1357-1360
DOI: 10.1126/science.1249361

Life and Cell Death

Trying to protect animals from one form of cell death may lead to death by another. Two protein kinases, known as RIPK1 and RIPK3 promote signaling that leads to cell death by necroptosis. However, Newton et al. (p. 1357, published online 20 February; see the Perspective by Zhang and Chan) found that inhibition of RIPK3 was not always beneficial. Instead, mice expressing a form of RIPK3 with no catalytic activity died from increased apoptotic cell death, but animals lacking the RIPK3 protein entirely, did not die perhaps because RIPK3 restrains apoptosis mediated by caspase-8 by an independent mechanism.


Receptor-interacting protein kinase 1 (RIPK1) and RIPK3 trigger pro-inflammatory cell death termed “necroptosis.” Studies with RIPK3-deficient mice or the RIPK1 inhibitor necrostatin-1 suggest that necroptosis exacerbates pathology in many disease models. We engineered mice expressing catalytically inactive RIPK3 D161N or RIPK1 D138N to determine the need for the active kinase in the whole animal. Unexpectedly, RIPK3 D161N promoted lethal RIPK1- and caspase-8–dependent apoptosis. In contrast, mice expressing RIPK1 D138N were viable and, like RIPK3-deficient mice, resistant to tumor necrosis factor (TNF)–induced hypothermia. Cells expressing RIPK1 D138N were resistant to TNF-induced necroptosis, whereas TNF-induced signaling pathways promoting gene transcription were unperturbed. Our data indicate that the kinase activity of RIPK3 is essential for necroptosis but also governs whether a cell activates caspase-8 and dies by apoptosis.

Tumor necrosis factor receptor 1 (TNFR1), Toll-like receptors (TLRs), and antigen receptors trigger necroptotic cell death when caspase-8 is inhibited and cells express receptor-interacting protein kinase 3 (RIPK3) and its substrate MLKL (mixed lineage kinase domain-like) (1). Wild-type RIPK3, but not catalytically inactive RIPK3 mutants, can reconstitute necroptosis in RIPK3-deficient cells (24), indicating that the kinase activity of RIPK3 is critical. TNFR1 engages RIPK3 through RIPK1, whereas TRIF (Toll/interleukin-1 receptor domain–containing adaptor inducing interferon-β) links RIPK3 to TLRs 3 and 4 (1). In the absence of antibodies that label cells dying by necroptosis, the contribution of necroptosis to pathology has been inferred from the resistance of Ripk3−/− mice to pancreatitis, retinal degeneration, atherosclerosis, liver injury, and systemic inflammation (1). The caveat here is that loss of RIPK3 could indicate a scaffold function for RIPK3 rather than a requirement for its kinase activity. Necrostatin-1 (nec-1) inhibits RIPK1 and ameliorates disease in some animal models too, although nec-1 might also inhibit other cellular processes via inhibition of indoleamine 2,3 dioxygenase (5).

We sought genetic proof that the kinase activity of RIPK3 drives necroptosis and disease by generating Ripk3 knock-in mice with Asp161 of the catalytically important DFG motif mutated to Asn (fig. S1, A and B). Whereas Ripk3−/− mice are viable (6), Ripk3D161N/D161N mice died around embryonic day 11.5 (E11.5) with abnormal yolk sac vasculature (Fig. 1, A and B). (Single-letter abbreviations for the amino acid residues are as follows: D, Asp; N, Asn. In the mutants, other amino acids were substituted at certain locations; for example, D161N indicates that aspartic acid at position 161 was replaced by asparigine.) Staining for the endothelial cell marker PECAM-1 (platelet/endothelial cell adhesion molecule 1) and cleaved caspase-3, a marker of apoptosis, revealed significantly increased apoptosis in the Ripk3D161N/D161N yolk sac vasculature (Fig. 1C). RIPK3 is expressed in endothelial cells and gut epithelial cells during development (fig. S1C). Ripk3D161N/– mice, which are expected to make half the amount of RIPK3 D161N, were viable. Therefore, there may be a threshold above which the RIPK3 D161N protein becomes lethal. Ripk3D161N/+ mice had a normal life span, arguing against RIPK3 D161N acting as a dominant negative mutant.

Fig. 1 Ripk3D161N/D161N mice die during embryogenesis.

(A) Numbers of surviving offspring from Ripk3D161N/+ parents. Mouse strains were derived from independent Ripk3D161N/+ embryonic stem cell clones. (B) E11.5 embryos and their yolk sacs. (C) E11.5 yolk sacs stained for PECAM-1 (red) and cleaved caspase-3 (CC3) (green). Scale bar, 200 μm. Each square in the scatter plot represents the yolk sac of one embryo. CC3+ cells were counted in 8 to 17 fields per embryo, and the mean number per field is plotted. P value determined with Student’s t test.

The genetic evidence for caspase-8 and its adaptor FADD (Fas-associated via death domain) inhibiting necroptosis caused by RIPK1 and RIPK3 is compelling (1). We tested whether RIPK3 D161N promoted caspase-8–dependent apoptosis (Fig. 2A) by breeding the Ripk3D161N allele onto a caspase-8–null background (fig. S2A). Casp8−/− embryos, like Ripk3D161N/D161N embryos, die mid-gestation (7), but double mutant Casp8−/− Ripk3D161N/D161N mice were viable (Fig. 2B). Halving the Casp8 gene dosage delayed lethality until ~E15.5 (fig. S2B). Casp8−/− Ripk3D161N/D161N mice resembled Casp8−/− Ripk3−/− mice (8, 9) because they accumulated T cells expressing the surface markers CD3 and B220 (fig. S2C) and developed the lymphadenopathy and splenomegaly (Fig. 2, C and D) that is characteristic of impaired apoptosis signaling by the death receptor Fas. These data confirm that (i) RIPK3 D161N promotes caspase-8–dependent apoptosis during embryogenesis and (ii) the kinase activity of RIPK3 is required for the death of Casp8−/− embryos. The kinase activity of RIPK3 also appears to mediate the proliferation defect of Casp8−/− T cells (10) because Casp8−/− Ripk3D161N/D161N T cells proliferated normally in response to antibodies to CD3 and CD28 (fig. S2D).

Fig. 2 RIPK1- and caspase-8–dependent embryonic lethality caused by RIPK3 D161N.

(A) Proposed mechanism for control of necroptosis and apoptosis by RIPK3 and RIPK3 D161N. (B) Numbers of surviving offspring from Casp8+/− Ripk3D161N/+ parents. (C) Spleen and lymph nodes from mice aged 16 to 20 weeks. Scale bar, 1 cm. (D) Leukocytes in the spleen or brachial, inguinal, mesenteric, and axillary lymph nodes were counted. Bars represent the mean ± SD. (E) Numbers of P3 to P6 offspring from Ripk1+/− Ripk3D161N/+ parents. (F) E18.5 littermates.

Although a direct interaction between RIPK3 D161N and caspase-8 during embryogenesis is a possibility, we think that RIPK3 may influence caspase-8 through RIPK1 and FADD. RIPK3 and RIPK1 interact via their RIP homotypic interaction motifs (RHIMs) (11); RIPK1 and FADD can interact through their death domains (12); and FADD and caspase-8 interact through their death effector domains (13). Consistent with this model, RIPK1 deficiency (fig. S3A) rescued embryonic lethality caused by RIPK3 D161N (Fig. 2, E and F). Ripk1−/− mice died soon after birth with increased cell death in the colon, brown adipose tissue, and thymus (fig. S3B). RIPK1 is abundant in the wild-type intestine at E18.5 and can also be detected in brown fat and liver (fig. S3C). Ripk1−/− Ripk3D161N/D161N mice died soon after birth with the same intestinal lesions as Ripk1−/− mice (fig. S3B), suggesting that the defects caused by RIPK1 loss do not require RIPK3-dependent necroptosis. Halving the Ripk1 gene dosage delayed death due to RIPK3 D161N (Fig. 2F). At E18.5, pale Ripk1+/− Ripk3D161N/D161N embryos exhibited none of the lesions associated with RIPK1 loss but contained fewer hematopoietic cells and less hepatic glycogen (fig. S3D). Ripk3D161N/D161N embryos lacking MLKL, TRIF, the RHIM protein DAI (DNA-dependent activator of interferon-regulatory factors), the RIPK1 de-ubiquitinase CYLD (cylindromatosis), TNFR1, DR3 (death receptor 3), or FLIP (FLICE-inhibitory protein) did not survive beyond ~E11.5 (fig. S4), indicating that these proteins are dispensable for activation of caspase-8 by RIPK3 D161N.

To test whether RIPK3 D161N was deleterious in adult mice, we engineered Ripk3LSL/LSL mice with loxp-Stop-loxp (LSL) transcription termination cassettes preventing expression of RIPK3 D161N during development (fig. S1B). This strategy was possible because Ripk3−/− mice develop normally (6). We deleted the LSL cassettes and switched on expression of RIPK3 D161N in adults using ubiquitously expressed tamoxifen-inducible cre recombinase (CreERT2) (14). All Ripk3LSL/LSL CreERT2 mice given tamoxifen developed diarrhea and lost between 13 and 28% of their body weight after 6 to 7 days, requiring that they be euthanized. No morbidity was observed in Ripk3+/+ or Ripk3LSL/+ mice expressing CreERT2 (Fig. 3A and fig. S5A). Histological analyses of Ripk3LSL/LSL CreERT2 mice revealed degeneration of intestinal crypts, villous blunting and fusion, and mild enteritis (Fig. 3B). Lesions were mainly in the small intestine. Many mice had segmental, submucosal edema in the large intestine. Pyknotic and karyorrhectic lymphocytes in secondary lymphoid organs and thymic atrophy were consistent with lymphocyte death (fig. S5B). Ripk3LSL/LSL CreERT2 mice showed no signs of morbidity after 5 days, but staining for cleaved caspase-3 revealed increased apoptosis in the small intestine (Fig. 3C). These results suggest that RIPK3 D161N affects caspase-8 in the small intestine like it does in the embryo vasculature.

Fig. 3 Lethal effect of RIPK3 D161N expressed in adult mice.

(A) Effect of tamoxifen-induced RIPK3 D161N expression on adult mouse survival. Mice (n = 8 per genotype) received daily tamoxifen injections on days 1 through 5. (B) Hematoxylin and eosin stained small intestine on day 7. Arrows indicate blunted and fused villi. Asterisks indicate crypt degeneration. Magnification is 20×. (C) Small intestine harvested on day 5. Cells expressing cleaved caspase-3 are stained brown. Magnification is 20×. (D) Survival of MEFs cultured in 4-hydroxytamoxifen (4-OHT). Percent survival = 100 × [percent viable cells (unstained by propidium iodide) after 4-OHT treatment]/[percent viable cells in medium alone]. Data represent the mean ± SD of three lines per genotype. P value on day 3 determined with Student’s t test. (E) Western blots of FADD-containing protein complexes immunoprecipitated from CreERT2 MEFs cultured for 2 days in medium, 4-OHT, or 4-OHT and zVAD.fmk.

We sought biochemical evidence of RIPK3 D161N affecting caspase-8 in Ripk3LSL/LSL CreERT2 mouse embryo fibroblasts (MEFs) immortalized with adenovirus E1A. Culture with 4-hydroxytamoxifen, to promote expression of RIPK3 D161N, caused increased cell death that was not seen in Ripk3+/+ CreERT2 cultures (Fig. 3D). This death appeared to be apoptosis because it was blocked by the pan-caspase inhibitor zVAD.fmk (carbobenzoxy-valyl-alanyl-aspartyl-[O-methyl]-fluoromethylketone), but not by nec-1 (fig. S5C). As a control, nec-1 blocked the death of wild-type E1A-immortalized MEFs treated with TNF/zVAD.fmk (fig. S5D). Inhibition of apoptosis caused by RIPK3 D161N allowed us to isolate a complex containing FADD, caspase-8, RIPK1, and RIPK3 D161N (Fig. 3E). An equivalent complex was not recovered from similarly treated Ripk3+/+ CreERT2+ cells. Further emphasizing the ability of RIPK3 D161N to promote apoptosis, the mutant RIPK3 protein was less abundant than was wild-type RIPK3. Poor expression of RIPK3 D161N did not result from decreased transcription of Ripk3D161N mRNA (fig. S5E). In the absence of structural data for RIPK3 D161N, it is unclear why this catalytically inactive mutant engages RIPK1, FADD, and caspase-8 but wild-type RIPK3 does not. Regardless, our results indicate that the kinase activity of RIPK3 dictates whether a cell dies from necroptosis or apoptosis. As such, small-molecule inhibitors of RIPK3 may have limited therapeutic benefit because of their potential to promote apoptotic cell death.

Nec-1 blocked necroptosis induced by TNF/zVAD.fmk (fig. S5C), was protective in an inflammatory disease model (15), and does not induce apoptotic cell death, suggesting that inhibition of RIPK1 rather than RIPK3 may have therapeutic benefit. To mimic RIPK1 inhibition in the whole animal, we generated Ripk1 knock-in mice with Asp138 of the HKD motif necessary for kinase activity mutated to Asn (fig. S6, A and B). Whereas Ripk1−/− mice died soon after birth, Ripk1D138N/D138N mice were viable. Consistent with a critical role for the kinase activity of RIPK1 in TNF-induced necroptosis, Ripk1D138N/D138N bone marrow–derived macrophages (BMDMs) and E1A-immortalized MEFs were as resistant as Ripk3−/− cells to killing by TNF/zVAD.fmk (Fig. 4A and fig. S6C). Ripk1D138N/D138N cells expressed normal amounts of RIPK3 and MLKL (Fig. S6D). Ripk1D138N/D138N mice also resembled Ripk3−/− mice in their systemic response to TNF, exhibiting less hypothermia than did their wild-type counterparts (Fig. 4B). Unlike RIPK1 loss, RIPK1 D138N did not rescue embryonic lethality caused by RIPK3 D161N (fig. S4). This result is consistent with nec-1 not protecting MEFs expressing RIPK3 D161N (fig. S5B) and indicates that the kinase activity of RIPK1 is not required for activation of caspase-8 by RIPK3 D161N.

Fig. 4 Requirement of RIPK1 catalytic activity for TNF-induced necroptosis and hypothermia.

(A) Survival of primary BMDMs stimulated for 7 hours. Bars represent the mean ± SD, n = 3 mice per genotype. (B) Body temperatures of mice dosed with TNF. Data represent the mean ± SD, n = 10 mice per genotype. The letter “x” denotes when a wild-type mouse was found dead or had to be euthanized because its body temperature fell below 23.6°C. (C) Western blots showing the response of primary BMDMs to TNF.

RIPK1 is required for TNF-induced nuclear factor κB (NF-κB) and mitogen-activated protein kinase signaling (16, 17). Wild-type RIPK1 and RIPK1 D138N restored these signals in Ripk1−/− cells (18), suggesting that RIPK1 has an essential scaffold function in this setting, whereas its kinase activity is dispensable. Indeed, Ripk1D138N/D138N and wild-type BMDMs were indistinguishable in their phosphorylation of inhibitor of NF-κB α (IκBα), c-Jun N-terminal kinase (JNK), and extracellular signal-regulated kinase (ERK) in response to TNF (Fig. 4C). These results, together with the viability of Ripk1D138N/D138N mice, are encouraging because they suggest that inhibiting the kinase activity of RIPK1 has no deleterious effects, at least in the short term. These Ripk1D138N/D138N mice can be used to explore the contribution of RIPK1 and necroptosis to various mouse models of human disease.

Supplementary Materials

Materials and Methods

Figs. S1 to S6

References (1925)

References and Notes

  1. Acknowledgments: We thank J. Starks, K. O’Rourke, A. Maltzman, M. Schlatter, T. Soukup, L. Tam, R. Pattni, R. Newman, H. Ngu, and D. Siler for technical assistance. S. Sun, W. Alexander, S. Akira, and Y. He kindly provided knockout mice. All authors were employees of Genentech.
View Abstract

Stay Connected to Science

Navigate This Article