Research ArticleCell Biology

The death-inducing activity of RIPK1 is regulated by the pH environment

See allHide authors and affiliations

Science Signaling  12 May 2020:
Vol. 13, Issue 631, eaay7066
DOI: 10.1126/scisignal.aay7066

Cell death on acid

The serine/threonine kinase RIPK1 is a critical mediator of cell death in response to the proinflammatory cytokine TNF. RIPK1 inhibitors are being investigated for the treatment of various infectious and inflammatory diseases. Moriwaki et al. found that the acidification that occurs in high-density cell cultures inhibited the kinase activity of RIPK1 and TNF-induced cell death. The effect of pH on RIPK1 activity was reversible and required histidine residues, which become protonated under low pH. These results suggest that cells in pathological situations in which intracellular pH becomes acidic are more resistant to RIPK1-dependent cell death.

Abstract

Receptor-interacting protein kinase 1 (RIPK1) is a serine/threonine kinase that dictates whether cells survive or die in response to the cytokine tumor necrosis factor (TNF) and other inflammatory stimuli. The activity of RIPK1 is tightly controlled by multiple posttranslational modification mechanisms, including ubiquitination and phosphorylation. Here, we report that sensitivity to TNF-induced, RIPK1-dependent cell death was tunable by the pH environment. We found that an acidic extracellular pH, which led to a concomitant decrease in intracellular pH, impaired the kinase activation of RIPK1 and autophosphorylation at Ser166. Consequently, formation of the cytosolic death-inducing complex II and subsequent RIPK1-dependent necroptosis and apoptosis were inhibited. By contrast, low pH did not affect the formation of membrane-anchored TNFR1-containing signaling complex (complex I), RIPK1 ubiquitination, and NF-κB activation. TNF-induced cell death in Ripk1−/− cells was not sensitive to pH changes. Furthermore, mutation of the conserved His151 abolished the pH dependence of RIPK1 activation, suggesting that this histidine residue functions as a proton acceptor to modulate RIPK1 activity in response to pH changes. These results revealed an unexpected environmental factor that controls the death-inducing activity of RIPK1.

INTRODUCTION

Receptor-interacting protein kinase 1 (RIPK1) is a serine/threonine kinase that determines cell survival and cell death downstream of various innate immune receptors, including tumor necrosis factor (TNF) receptor (TNFR) (1). Although TNF primarily triggers prosurvival and inflammatory signaling pathways, it induces two morphologically and mechanistically distinct modes of cell death in certain (often pathological) circumstances: caspase-dependent apoptosis and RIPK3 and mixed lineage kinase domain–like (MLKL)–dependent necroptosis (2). Upon ligation of TNF to the preassembled TNFR1 trimer, RIPK1 is rapidly recruited to TNFR1 through homophilic interaction of death domains (DDs), present at the C termini of both proteins. This membrane-associated TNFR1 complex is often referred to as “complex I,” in which RIPK1 is heavily conjugated by polyubiquitin chains by E3 ligases such as cellular inhibitor of apoptosis 1 (cIAP1), cIAP2, and linear ubiquitin chain assembly complex (LUBAC) (3, 4). These polyubiquitin chains serve as prosurvival platforms to recruit various signaling kinases, such as inhibitor of κB kinase α/β (IKKα/β) through nuclear factor κB (NF-κB) essential modulator (NEMO) adaptor, transforming growth factor–β–activated kinase 1 (TAK1), IKKε, and TRAF family member–associated NF-κB activator (TANK) binding kinase 1 (TBK1) (5). These kinases limit RIPK1 activation by directly or indirectly phosphorylating various serine/threonine sites on RIPK1 and thereby inhibit RIPK1 kinase activity–dependent cell death (612). Inhibition of these kinases or RIPK1 ubiquitination promotes RIPK1 kinase activity–dependent formation of a secondary complex in the cytosol termed as complex II, which is composed of RIPK1, FADD (Fas-associated DD protein), and caspase 8, and consequently, RIPK1-dependent apoptosis (3). When caspase 8 activity is inhibited, RIPK1 recruits an additional adaptor, the essential necroptosis kinase RIPK3, to complex II (13, 14). This alternative complex II is often referred to as the “necrosome.” In this complex, RIPK3 is activated and phosphorylates the downstream effector MLKL (15). Phosphorylated MLKL forms oligomer and executes necroptosis (1619).

RIPK1 has been implicated in various infectious and sterile inflammatory diseases such as bacterial and viral infections; ischemia reperfusion injury in heart, brain, and kidney; neurodegenerative diseases; and inflammatory bowel diseases (20). This has prompted clinical trials to test the therapeutic effect of RIPK1 kinase inhibitors on inflammatory bowel disease, psoriasis, rheumatoid arthritis, Alzheimer’s disease, and amyotrophic lateral sclerosis (21, 22). Under pathological conditions, the extracellular environment can undergo changes that affect cellular responses. For instance, a substantial drop of extracellular pH (<6.5) is caused by ischemia during myocardial infarction and stroke (2327). Changes in extracellular pH affect cellular behaviors (28). However, little is known about how extracellular pH might affect RIPK1 signaling. Here, we report that TNF-induced apoptosis and necroptosis were regulated by changes in extracellular pH. We found that reduced extracellular pH, which led to intracellular acidification, inhibited RIPK1 autophosphorylation and activation. Consequently, RIPK1-dependent formation of complex II and necrosome and subsequent activation of caspase 8 and RIPK3 were dampened. Hence, our results reveal an unexpected environmental control mechanism that can rapidly and reversibly modulate RIPK1 activity and cellular response to cell death signals.

RESULTS

TNF-induced necroptosis is inhibited at high cell density

We serendipitously observed that in TNFR2-expressing Jurkat 4E3 cells (29), the extent of necroptosis induced by TNF (T), the pan-caspase inhibitor z-VAD-fmk (Z), and second mitochondria-derived activator of caspase (SMAC) mimetic was significantly suppressed, with increasing cell density in culture (Fig. 1A). Cell density–dependent inhibition of necroptosis was also observed when we normalized the amount of TNF on a per-cell basis (fig. S1A), thus ruling out the possibility that ligand competition was responsible for the high cell density–dependent decrease in cell death.

Fig. 1 HCCM inhibits TNF-induced necroptosis.

(A) 4E3 cells were treated with z-VAD-fmk (Z), Smac mimetic LBW242 (S), and TNF (T) (100 ng/ml) at the indicated cell densities for 14 hours. Data are mean ± SEM of three independent experiments. (B to D) 4E3 (B) and HT29 (C and D) cells were treated with z-VAD-fmk (Z), Smac mimetic [S; either LBW242 in (B) or BV6 in (C) and (D)], and TNF for 14 (B) or 6 (C and D) hours in either fresh medium or HCCM. HCCM was prepared from 4E3 (B and D) and HT29 (C) cells. Data are mean ± SEM of three independent experiments. (E) Necroptosis was induced in I42 cells by stimulating with TNF alone for 14 hours. HCCM was prepared from 4E3 cells. Data are mean ± SEM of three independent experiments. Cell death was determined by flow cytometry using PI (A, B, and E) or CellTiter-Glo Luminescent Cell Viability Assay (ATP) (C and D). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.00001; unpaired t test with Welch’s correction.

We posited that a secreted factor might mediate the resistance to necroptosis at high cell density. To examine this possibility, we resuspended Jurkat 4E3 cells in medium from cells cultured at high cell density [high-density cell culture medium (HCCM)] or fresh culture medium at low cell density. When cells were cultured in HCCM, Z/S/T-induced necroptosis was inhibited (Fig. 1B). The inhibitory effect was also observed when HCCM was prepared from high–cell density culture of the human colon carcinoma HT29 (Fig. 1C). In addition, Z/S/T-induced necroptosis was suppressed when HT29 cells were cultured in HCCM collected from 4E3 Jurkat cells (Fig. 1D), indicating that the inhibitory effect was not cell type–specific. To exclude the possibility that the potency of z-VAD-fmk or Smac mimetic was compromised in HCCM, we induced necroptosis in FADD-deficient Jurkat I42 cells, which undergo necroptosis in response to TNF alone (30, 31). Again, TNF-induced necroptosis in FADD-deficient Jurkat I42 cells was also inhibited by HCCM (Fig. 1E). In addition, the inhibitory effect was observed regardless of the type of SMAC mimetic used (BV6 or LBW242) (Fig. 1, A to D). These results indicate that HCCM inhibits TNF-induced necroptosis in a cell type–independent and stimulus-independent manner.

The pH environment modulates TNF-induced necroptosis

We reasoned that soluble molecules secreted by tumor cells might be responsible for this inhibitory effect on necroptosis. At high cell density, these molecules could accumulate and inhibit necroptosis. One source of these putative inhibitory molecules could come from dying cells, which often release damage-associated molecular patterns that could alter cell physiology. We found that dead cell supernatant generated from repeated freeze-thaw cycles did not inhibit necroptosis (fig. S1B). Moreover, HCCM subjected to boiling and/or proteinase K treatment did not alter the necroptosis inhibitory effect of HCCM (fig. S1C). Moreover, the inhibitory effect was also retained after multiple freeze-thaw cycles (fig. S1D). Depletion of lipids by methanol and chloroform extraction or reactive gaseous species by evaporation also did not interfere with this inhibitory effect (fig. S1, E and F). These results suggest that proteins, lipids, and gaseous species are unlikely to contribute to the inhibition of necroptosis.

Proliferating cells consume sodium bicarbonate (NaHCO3) and release lactate, both of which could decrease the pH of culture medium. We found that the pH of HCCM was around 6.3. To test whether the low pH of HCCM was responsible for inhibition of TNF-induced necroptosis, we neutralized the pH of HCCM with NaHCO3 to 7.4. The inhibitory effect of HCCM was significantly reversed by pH neutralization (Fig. 2A). Necroptosis was also restored when Hepes and sodium hydroxide (NaOH) were used to neutralize the pH of HCCM to 7.4 (fig. S1G). To further determine whether reduced pH was responsible for the inhibitory effect of HCCM, we lowered the pH of fresh medium to 6.3 with hydrochloric acid (HCl). Reduced pH led to strong inhibition of Z/S/T-induced necroptosis in HT29, 4E3, HCT116, and SW480 and TNF-induced necroptosis in I42 cells (Fig. 2, B to D, and fig. S2, A to D). The inhibitory effect of acidic medium was still observed under serum-free conditions (Fig. 2E). In addition, neutralization of acidic medium by Hepes and NaOH negated the inhibitory effect (Fig. 2E). These results indicate that neither denaturation of proteins in the medium nor alteration of medium composition contributes to the inhibitory effect of acidic medium. Although we cannot rule out that cell confluency per se might partially affect the sensitivity to necroptosis, acidification of culture medium is a critical event for the inhibition of the cell death.

Fig. 2 Acidic extracellular pH condition inhibits TNF-induced necroptosis.

(A) HT29 cells were treated with z-VAD-fmk (Z), Smac mimetic BV6 (S), and TNF for 6 hours. HCCM was subjected to pH neutralization by NaHCO3. Data are mean ± SEM of three independent experiments. (B and C) HT29 (B) and 4E3 (C) cells were treated with z-VAD-fmk (Z), either Smac mimetic [S; either BV6 in (B) or LBW242 in (C)], and TNF in neutral or acidic medium for 6 (B) or 14 (C) hours. Data are mean ± SEM of three independent experiments. (D) I42 cells were stimulated by TNF alone in neutral or acidic medium for 14 hours. Data are mean ± SEM of three independent experiments. (E) HT29 cells were treated with z-VAD-fmk (Z), Smac mimetic BV6 (S), and TNF for 14 hours in a serum-free medium. Acidic medium was subjected to pH neutralization by Hepes and NaOH. Data are mean ± SEM of three independent experiments. Cell death was determined by CellTiter-Glo Luminescent Cell Viability Assay (ATP; A, B, and E) and flow cytometry using PI (C and D). *P < 0.05, **P < 0.01, and ***P < 0.001; unpaired t test with Welch’s correction.

Low extracellular pH affects necroptosis without interfering with TNF-induced de novo gene synthesis

RIPK1 and RIPK3 form the necrosome, an essential signal complex for necroptosis (13, 14). We found that RIPK1-RIPK3 interaction was inhibited in acidic medium (Fig. 3A). RIPK3 phosphorylation, which can be detected as a mobility shift on SDS–polyacrylamide gel electrophoresis, is critical for necrosome formation and subsequent amyloid conversion of the complex (32). We found that RIPK3 phosphorylation was inhibited in acidic medium and HCCM [Fig. 3, B (lanes 3 to 8) and C]. These results therefore indicate that low extracellular pH inhibits TNF-induced necroptosis at a step before RIPK3 activation.

Fig. 3 Acidic extracellular pH inhibits necroptosis independent of de novo gene expression.

(A to C, E, and F) HT29 cells were treated with either zVAD-fmk (Z), Smac mimetic LBW242, and TNF (100 ng/ml) (A to C and E) or TNF alone (100 ng/ml) (F) for the indicated times. WCEs were subjected to RIPK3 IP and then Western blotting (A) or directly Western blotted (B, C, E, and F). In (B), the medium was changed from neutral-pH medium (N) to acidic medium (A) containing Z/S 1 hour after TNF treatment (MC) (B and D). In (E), actinomycin D (ActD; 2.5 μg/ml) was used where indicated. C, control. RIPK3 phosphorylation as determined by an upward mobility shift was examined by Western blotting (B, C, and E). Blots are representatives of three (A to C and E) or two (F) independent experiments. (D) HT29 cells were treated as in (B) except for using BV6 instead of LBW242. After changing the medium to acidic medium, cells were cultured for another 13 hours in the presence of Z/S. Data are mean ± SEM of three independent experiments. (G) Rela−/− MEFs were treated with z-VAD-fmk (zVAD) and TNF for 14 hours. Where indicated, Nec1 was included. Data are mean ± SEM of three independent experiments. Cell death was determined by CellTiter-Glo Luminescent Cell Viability Assay (ATP; D and G). N, neutral medium. A, acidic medium. *P < 0.05, **P < 0.01, and ***P < 0.001; unpaired t test with Welch’s correction.

Acidic medium could interfere with TNF-TNFR1 interaction, which would inhibit necroptosis. To test this possibility, we stimulated cells with Z/S/T in fresh, neutral-pH medium for 1 hour to allow normal binding of TNF to the receptor. We then switched the medium to TNF-free, acidic medium. Under this condition, TNF-induced RIPK3 phosphorylation and necroptosis were still inhibited [Fig. 3, B (lanes 9 to 14) and D]. Hence, low–extracellular pH–mediated inhibition of necroptosis is unlikely due to impaired TNF-TNFR1 interaction.

TNF stimulates prosurvival gene expression through the NF-κB pathway. De novo synthesis of these survival factors can antagonize RIPK3 phosphorylation and necroptosis. However, the transcription inhibitor actinomycin D did not affect low–extracellular pH–mediated inhibition of Z/S/T-induced RIPK3 phosphorylation and necroptosis (Fig. 3E and fig. S3A). Furthermore, the phosphorylation and degradation of IκBα (inhibitor of κB alpha) were normal in acidic medium (Fig. 3F). RelA and TRAF2 (TNF receptor–associated factor 2) are two critical signal adaptors for NF-κB activation that protect cells from necroptosis. Mouse embryonic fibroblasts (MEFs) that are deficient for either one of these molecules are highly sensitive to Z/T-induced necroptosis (33, 34). In these cells, low extracellular pH also inhibited Z/T-induced necroptosis (Fig. 3G and fig. S3B). These results indicate that reduced extracellular pH inhibits necroptosis independent of NF-κB activation.

Changes in extracellular pH can modulate various intracellular signaling pathways, many of which are also activated in response to TNF (28). However, inhibitors against mitogen-activated protein kinases (MAPKs), phosphatidylinositol 3-kinase (PI3K), and cyclic AMP-dependent protein kinase (PKA) did not affect the inhibitory effect of acidic medium on necroptosis (fig. S3C). Collectively, these results indicate that low extracellular pH inhibits TNF-induced necroptosis independently of altering these intracellular signaling responses.

TNF-induced apoptosis is inhibited in the acidic environment

Besides driving necroptosis, TNF also stimulates apoptosis in the presence of intact caspase 8 function. To interrogate whether apoptosis is also regulated by the pH environment, we used Rela−/− MEFs, which underwent robust apoptosis marked by caspase 8 cleavage and activation in response to TNF alone (Fig. 4, A and B). Caspase 8 activation and apoptosis were also inhibited in acidic medium (Fig. 4, A and B). TNF stimulates formation of a membrane-associated complex (complex I) that contains RIPK1, TRAF2, and other signal adaptors. RIPK1 is ubiquitinated in this complex to facilitate NF-κB activation. RIPK1 and TRAF2 recruitment to TNFR1 was not impaired in acidic medium (Fig. 4C). In addition, RIPK1 polyubiquitination was not affected in acidic medium (Fig. 4C). Consistent with normal complex I formation, low-pH medium did not affect the phosphorylation and degradation of IκBα and phosphorylation of extracellular signal–regulated kinase (Erk) (Fig. 4D and fig. S3D). RIPK1 in complex I rapidly transitions to the cytosol to form a death-inducing complex termed as complex II that comprises RIPK1, FADD, and caspase 8. Formation of complex II was inhibited by acidic medium (Fig. 4E). These results indicate that low-pH medium halts the transition from complex I to complex II during TNF-induced apoptosis and necroptosis.

Fig. 4 TNF-induced apoptosis is inhibited when external pH is low.

(A) Rela−/− MEFs were treated with TNF for 14 hours, and cell death was determined by CellTiter-Glo Luminescent Cell Viability Assay (ATP). Data are mean ± SEM of three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001; unpaired t test with Welch’s correction. (B to E) Rela−/− MEFs were treated with TNF (20 ng/ml) in neutral or acidic medium. WCEs were subjected to Western blotting (B and D) or IP and then Western blotting (C and E). Pro, IM, and active represent proform, intermediate form, and active form of caspase 8, respectively. Arrowhead in (E) indicates IgG heavy chain. Blots are representatives of three (B, D, and E) or two (C) independent experiments.

Low extracellular pH inhibits necroptosis downstream of RIPK1 deubiquitination

We reasoned that low pH might regulate a common checkpoint in apoptosis and necroptosis. One such checkpoint is the deubiquitination of RIPK1, which facilitates the transition of complex I to complex II during both apoptosis and necroptosis. The deubiquitinating enzyme CYLD (cylindromatosis) targets RIPK1 for deubiquitination to promote complex II assembly (35). Because complex II, but not complex I, was affected by acidic medium (Fig. 4, C and E), we asked whether CYLD might be the cellular target of pH regulation. Consistent with previous reports and the lack of effect of acidic medium on RIPK1 deubiquitination (Fig. 4C) (3537), Z/S/T-induced necroptosis, which was reduced in Cyld−/− MEFs (fig. S4A), was still sensitive to acidic medium inhibition in Cyld−/− MEFs (fig. S4A). Thus, CYLD is dispensable for the inhibitory effect of low extracellular pH. In addition to CYLD, the caspase 8 homolog cFLIP (cellular FADD-like interleukin-1β converting enzyme-like inhibitory protein, also known as CFLAR) also regulates TNF-induced apoptosis and necroptosis (38). However, Cflar−/− MEFs were also sensitive to low–extracellular pH–mediated inhibition of necroptosis (fig. S4B). These results indicate that CYLD and cFLIP are not the molecular targets of low–extracellular pH regulation of TNF-induced necroptosis.

The kinase activity of RIPK1 promotes TNF-induced apoptosis and necroptosis in Rela−/− MEFs

In addition to TNF-induced necroptosis (Fig. 5A, left), RIPK1 is also a key adaptor for TNF-induced apoptosis in certain cell types and conditions. A previous study reported that Rela−/− MEFs undergo apoptosis in response to TNF in a RIPK1 kinase activity–independent manner (7). However, the embryonic lethality of Rela−/− mice is rescued by combined deletion of FADD and RIPK3 or by introducing a kinase-dead RIPK1 mutant (Ripk1K45A) (39). In agreement with this latter report, we found that TNF-induced apoptosis in Rela−/− MEFs was partially, but substantially, inhibited by four different RIPK1 kinase inhibitors (Fig. 5A, right), indicating that RIPK1 promotes apoptosis in Rela−/− cells through its kinase activity. The RIPK1 kinase inhibitor GSK’963 reduced TNF-induced caspase 8 activation and complex II formation (Fig. 5, B and C), but not the phosphorylation and degradation of IκBα in Rela−/− MEFs (Fig. 5D). Silencing of RIPK1 by short hairpin RNA (shRNA) inhibited TNF-induced apoptosis and caspase 8 activation (fig. S5, A and B), but not the phosphorylation and degradation of IκBα (fig. S5C). As expected, RIPK1 knockdown inhibited Z/T-induced necroptosis in Rela−/− MEFs (fig. S5D). Low extracellular pH was as effective as GSK’963 in inhibiting TNF-induced apoptosis in Rela−/− MEFs (Fig. 5E). Moreover, low extracellular pH did not further reduce apoptosis in the presence of GSK’963 (Fig. 5E). In Rela-replete cells, TNF induces apoptosis in the presence of SMAC mimetic or TAK1 inhibitor in a RIPK1 kinase activity–dependent manner. In both cases, TNF-induced, RIPK1-dependent apoptosis was inhibited by acidic medium (Fig. 5, F and G). RIPK1 kinase activity was also involved in cytokine production in the presence of caspase inhibitor (40). Z/S/T stimulation induced robust transcription of endogenous Tnf in a RIPK1 kinase activity–dependent manner, which was inhibited by acidic medium (Fig. 5H). These results suggest that low extracellular pH suppresses RIPK1 kinase activity.

Fig. 5 RIPK1 mediates pH-dependent effects on TNF-induced apoptosis.

(A) Rela−/− MEFs were treated with TNF and the indicated RIPK1 inhibitors in the presence (left) or absence (right) of zVAD. Cell death was measured by LDH release assay after 4.5 hours (for TNF and zVAD) or 14 hours (for TNF alone). Data obtained from three independent experiments are shown. (B to D) Rela−/− MEFs were treated with TNF (20 ng/ml) and the RIPK1 inhibitor GSK’963 as indicated. In (C), whole-cell lysates were subjected to FADD IP. Western blotting was performed, as indicated in (B) and (D). Arrowhead in (C) indicates IgG heavy chain. Blots are representative of three independent experiments. (E) Rela−/− MEFs were treated with TNF for 14 hours under the indicated conditions. Data obtained from three independent experiments are shown. (F and G) HeLa cells were treated with TNF for 14 hours. Where indicated, SMAC mimetic BV6 or TAK1 inhibitor 5Z-7 was added 1 hour before TNF stimulation. Data are mean ± SEM of three independent experiments. (H) HT29 cells were treated with z-VAD-fmk (Z) and Smac mimetic BV6 (S) and TNF (10 ng/ml) for 6 hours. RNA was subjected to real-time RT-PCR for Tnf. Data are mean ± SEM of three independent experiments. Cell death was determined by CellTiter-Glo Luminescent Cell Viability Assay (ATP) (E and G) or flow cytometry using annexin V/PI (F). **P < 0.01 and ****P < 0.00001; unpaired t test with Welch’s correction.

The kinase activity of RIPK1 is inhibited by low pH

To address whether RIPK1 kinase activity is controlled by pH, we examined RIPK1 autophosphorylation at Ser166, an event that is widely used as a proxy to measure RIPK1 activation (41). We found that the phosphorylation of Ser166 was inhibited in acidic medium (Fig. 6A). Next, we performed in vitro kinase assay to directly test whether RIPK1 kinase activity is tunable by pH. First, we determined the intracellular pH within the first 30 min after cells were placed in acidic medium (Fig. 6B). This early time point was chosen because RIPK1 kinase activity is activated within 30 min upon TNF stimulation (42). The kinase activity of RIPK1 was diminished at pH 6.42, the intracellular pH of cells cultured in low-pH medium (Fig. 6C). To test whether low pH–mediated inhibition of RIPK1 kinase activity is reversible, we incubated immunoprecipitated RIPK1 under acidic conditions and subsequently performed in vitro kinase assay under neutral conditions. We found that pH-dependent regulation of RIPK1 kinase activity was reversible (Fig. 6D). The phosphorylation of IκBα by IKK (Figs. 3F and 4D and fig. S3D) and the phosphorylation of Erk and JNK (c-Jun N-terminal kinase) (Fig. 4D and fig. S6A) were normal in acidic medium, suggesting that reduced pH did not globally impair kinase activities. To further investigate the role of RIPK1 in pH-dependent regulation of cell death, we used Ripk1−/− 3T3 cells, which also undergo apoptosis in response to TNF. Unlike in RIPK1-replete cells, apoptosis and caspase 8 activation induced by TNF alone or TNF/cycloheximide (CHX) were not inhibited by low extracellular pH in Ripk1−/− cells (Fig. 6, E to H, and fig. S6B). These results implicate RIPK1 kinase activity as a key molecular target of pH regulation of TNF-induced apoptosis and necroptosis.

Fig. 6 Acidic pH inhibits RIPK1 kinase activity.

(A) HT29 cells were treated with z-VAD-fmk (Z) and Smac mimetic BV6 (S) and TNF (100 ng/ml). WCEs were subjected to Western blotting for RIPK1 pSer166. Blots are representative of three independent experiments. (B) Intracellular pH of cells cultured in neutral or acidic medium was determined using pHrodo Red AM intracellular pH indicator. Data are mean ± SEM of three independent experiments. (C and D) RIPK1 immunoprecipitates from HT29 cells were subjected to in vitro kinase assays to measure the autophosphorylation of RIPK1 Ser166 in neutral and acidic pH environment. Where indicated, Nec1 was included. Blots are representative of three (C) or two (D) independent experiments. (E to H) Ripk1−/− 3T3 fibroblasts were treated with either TNF (E and G) or TNF and CHX (1 μg/ml) (E and H). Data are mean ± SEM of three independent experiments (E and F). Western blotting for caspase 8 was performed in (G) and (H). Blots are representative of three independent experiments. (I) HT29 cells were treated with z-VAD-fmk (Z), Smac mimetic LBW242 (S), and CH11 for 14 hours. Nec1 was added where indicated. Data are mean ± SEM of three independent experiments. (J) HT29 cells were treated with z-VAD-fmk (Z), Smac mimetic BV6 (S), and TRAIL for 6 hours. Data are mean ± SEM of three independent experiments. (K) 3T3-SA cells were primed with IFN-β (10 ng/ml) for 24 hours in neutral medium. After washing IFN-β out, the primed cells were treated with z-VAD-fmk (Z), Smac mimetic BV6 (S), and poly(I:C) for 15 hours in neutral or acidic medium. Data are mean ± SEM of three independent experiments. Cell death was determined by CellTiter-Glo Luminescent Cell Viability Assay (ATP) (E, F, and I to K). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.00001; unpaired t test with Welch’s correction.

RIPK1 is also critical for necroptosis induced by the death receptors Fas and TRAIL (TNF-related apoptosis-inducing ligand) receptors (30), which was confirmed by treatment with necrostatin 1 (Nec1) (Fig. 6I and fig. S6C). Reduced extracellular pH also significantly inhibited Fas receptor- and TRAIL receptor–mediated necroptosis (Fig. 6, I and J). Although necroptosis induced by Toll-like receptor 3 (TLR3) or interferon (IFN) receptor (IFNR) is intact in RIPK1-deficient cells (43, 44), RIPK1 kinase inhibitors nonetheless effectively block necroptosis induced by these receptors in wild-type (WT) cells (4346). Necroptosis induced by the TLR3 ligand polyinosinic:polycytidylic acid [poly(I:C)] was also significantly reduced in acidic medium (Fig. 6K). These results suggest that low pH locks RIPK1 in an inactive state and thereby inhibits RIPK1 kinase activity–dependent cell death.

Specific histidine residues in RIPK1 mediate pH sensing

Changing pH can alter the protonation status of certain amino acids. Specifically, the pKa (where Ka is the acid dissociation constant) of histidine is close to physiological pH, rendering it sensitive to charge changes through direct protonation as the cellular pH changes. These charge changes can affect intermolecular and intramolecular interactions (47). Histidine is a key functional residue in many pH-sensing receptors (48). RIPK1 has several highly conserved histidine residues throughout its kinase domain, the intermediate domain, and the DD. Asparagine is similar to histidine in size and positions of hydrogen bond donor and acceptor but has a side chain that is not titratable with pH changes. To disrupt the putative pH-sensing function of RIPK1, we introduced asparagine substitutions at the conserved histidines. Among the nine histidine-to-asparagine substitutions tested, H136N and H151N abolished the necroptosis inhibitory effect by acidic pH (Fig. 7, A and B), suggesting that these residues contribute to pH sensing. Because H198N abrogated the necroptosis in even neutral medium, we could not evaluate contribution of this residue in the inhibitory effect (Fig. 7, A and B). In WT RIPK1-expressing cells, activation of RIPK1 and RIPK3 was impaired by acidic medium (Fig. 7C). However, their kinase activation was below detection limit in cells expressing the histidine mutants (Fig. 7C). This was because basal kinase activity was also reduced by H136N mutation and, to a lesser extent, H151N mutation (Fig. 7D), suggesting that these histidine residues are crucial for RIPK1 activation. His136 and His151 are located near the adenosine 5′-triphosphate (ATP)–binding pocket in RIPK1 (Fig. 7E) (49). In particular, His136 is part of the HKD motif, which is the critical motif in the catalytic loop. Therefore, we stimulated the mutant cells with a higher dose of TNF. Under this condition, we found that activation of RIPK1 and RIPK3 was no longer inhibited by acidic medium in H151N mutant cells (Fig. 7F). These results suggest that protonation of highly conserved His151 on RIPK1 compromises RIPK1 kinase activity by charge-dependent conformational change (Fig. 7G). Similarly, RIPK3 activation, which requires RIPK1 activation, was not inhibited by acidic medium in H136N mutant cells (Fig. 7F). However, RIPK1 activation was still below detection limit for this mutant. Therefore, we could not definitively conclude the role of His136 in the pH-dependent regulation of RIPK1 activity.

Fig. 7 Specific histidine residues in RIPK1 sense pH changes.

(A and B) WT or mutant RIPK1 tagged by HA at the N termini was expressed in Ripk1−/− 3T3 cells under the control of a doxycycline-inducible promoter. After induction of RIPK1 expression by treating with doxycycline (0.2 μg/ml) for 24 hours, the cells were treated with z-VAD-fmk and TNF for 14 hours. Cell death was determined by CellTiter-Glo Luminescent Cell Viability Assay (ATP). Data obtained from three independent experiments are shown. RIPK1 expression was confirmed by Western blotting (B). Blots are representative of three independent experiments. (C and F) After induction of RIPK1 expression by doxycycline (0.2 μg/ml) (C) or doxycycline (1 μg/ml) (F) for 24 hours, cells were treated with z-VAD-fmk, SMAC mimetic BV6, and either TNF (1 ng/ml) (C) or TNF (100 ng/ml) (F). The lanes were run on the same gel but were noncontiguous (C). Blots are representative of two independent experiments. (D) In vitro kinase assay was performed for WT, H136N, and H151N RIPK1. Blots are representative of two independent experiments. (E) His136 and His151 are shown on the crystal structure of human RIPK1 kinase domain (Protein Data Bank: 4ITH). The catalytic triad residues Lys45, Glu63, and Asp156 and the autophosphorylation site Ser166 are shown. (G) Alignment of amino acid sequences of RIPK1 from indicated species. His136 and His151 in mouse RIPK1 are highlighted in yellow. *P < 0.05, **P < 0.01, and ***P < 0.001; unpaired t test with Welch’s correction. H. sapiens, Homo sapiens; P. troglodytes; Pan troglodytes; G. gorilla, Gorilla gorilla; C. hircus, Capra aegagrus; M. musculus, Mus musculus; R. norvegicus, Rattus norvegicus; G. gallus, Gallus gallus; C. mydas, Chelonia mydas; X. tropicalis, Xenopus tropicalis; D. rerio, Danio rerio.

DISCUSSION

RIPK1 is a master regulator of cell fate under various stress and inflammatory conditions. The activity of RIPK1 is tightly regulated by posttranslational modifications such as phosphorylation and ubiquitination. In this study, we found another layer of posttranslational regulation of RIPK1 activity. Changes in extracellular pH is sensed by a series of proton-sensing receptors expressed on the plasma membrane, which can activate various intracellular signaling pathways such as MAPK, PI3K, and PKA and de novo gene expression (50). However, we found that neither de novo gene expression nor the kinase activity of PI3K, PKA, or MAPK contributed to the pH regulation of TNF-induced cell death. Instead, reduced pH in the extracellular environment caused a concomitant decrease in intracellular pH, which inhibited RIPK1 autophosphorylation at Ser166 and its kinase activity. Although we cannot rule out the possibility that there are other pH-sensitive molecular targets, our results indicate that RIPK1 is a major sensor that tunes TNF-induced apoptosis and necroptosis in response to changes in pH in the environment.

How does pH regulate the kinase activity of RIPK1? Our results showed that substitutions of His151 with non-titrating amino acid asparagine rendered RIPK1 insensitive to low–extracellular pH–mediated inhibition of necroptosis, suggesting that protonation on the residue was crucial for the inhibitory effect. His151 is located in close proximity to the DLG motif. Therefore, it is tempting to speculate that direct protonation on this residue could certainly affect critical activating rotation of the DLG motif (51) and locks RIPK1 in its inactive state. The negative charge from the phosphorylation of the N-terminal kinase domain of RIPK1 causes a charge repulsion–induced conformational change that could either facilitate amyloid assembly followed by RIPK1 activation or inhibit the kinase activity, depending on the site phosphorylated (612, 32). Our results provide additional support for the model that charge-dependent structural changes are a crucial regulatory lever that controls RIPK1 activation. Structural analysis of RIPK1 in posttranslationally modified states will further give an insight into its activation mechanism.

Unlike other mechanisms that regulate RIPK1 function, pH-mediated control of RIPK1 activity was rapid and reversible. Such a rapid response may be critical for cellular adaptation in response to different physiological and pathological cues. For instance, during ischemia, interruption of blood supply causes reduced extracellular and intracellular pH due to anaerobic metabolism and lactate accumulation. Although reperfusion restores the supply of oxygen and nutrient and reestablishes pH environment, it paradoxically induces and exacerbates cell death, followed by tissue injury (52). RIPK1 is involved in ischemia/reperfusion injury, and pharmacological or genetic inhibition of RIPK1 kinase activity ameliorates injury severity (5356). It could be that reperfusion lifts the low pH–mediated restriction of RIPK1 kinase activity and thereby induces RIPK1-dependent cell death. Given the wide spectrum of biological responses controlled by RIPK1, it is tempting to speculate that pH-dependent regulation of RIPK1 activity may play a key role in pathological conditions in which intracellular pH changes.

MATERIALS AND METHODS

Reagents

z-VAD-fmk, Nec1, and TRAIL were purchased from Enzo Life Sciences. CHX, actinomycin D, and LY294002 were purchased from Sigma-Aldrich. PD98059 was purchased from ChemScene. H89 was purchased from Cayman Chemical. (5Z)-7-oxozeaenol was purchased from Santa Cruz Biotechnology. Doxycycline was purchased from MP Biomedicals. Anti-human Fas agonistic antibody CH11 was purchased from Millipore. Poly(I:C) HMW (high molecular weight) was purchased from InvivoGen. IFN-β was purchased from BioLegend. Smac mimetics LBW242 and BV6 were provided by Novartis and Genentech, respectively. GSK’481, GSK’728, and GSK’963 were provided by GlaxoSmithKline. Cells were pretreated with 10 μM z-VAD-fmk, 5 μM LBW242, 0.25 μM BV6, 2 μM Nec1, 2 μM GSK’481, 2 μM GSK’728, or 2 μM GSK’963 for 1 hour before TNF stimulation, unless otherwise stated.

Cell culture

4E3 and I42 cells were cultured in RPMI 1640. Rela−/− MEFs, HT29, Ripk1−/− 3T3, and 3T3-Swiss albino (3T3-SA) cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM). Ten percent fetal calf serum (FCS), 2 mM glutamine, penicillin (100 U/ml), and streptomycin (100 μg/ml) were added to the media. HCT116 cells overexpressing RIPK3–green fluorescent protein and SW480 cells overexpressing RIPK3 were generated in our previous study (57). RIPK1-knockdown Rela−/− MEFs were generated, as described before (58). pGIPZ/puro vector carrying shRNA against mouse Ripk1 (Open Biosystems, V2LMM_9240) was used. Nonsilencing scrambled shRNA pGIPZ vector was used as control (Open Biosystems, RHS4346). N-terminally hemagglutinin (HA)–tagged RIPK1 was cloned into a modified lentiviral tet-on pTRIPZ/puro vector. Site-directed mutagenesis was performed to generate asparagine mutants. Ripk1−/− 3T3 cells were transduced by lentivirus encoding WT RIPK1 or these mutants and selected by puromycin as described before (58). Dead cell medium was generated by subjecting 3 or 6 × 106 cells to five times freeze-thaw cycles, resuspending the dead cells with 3 ml of fresh RPMI medium (5 and 10% of 20 × 106 cells/ml), and then removing cell debris by centrifugation and filtration. The medium was used to culture cells.

Preparation and treatment of HCCM

For preparation of HCCM, 4E3 cells were cultured at 20 × 106 cells/ml unless otherwise stated or at indicated concentrations in serum-free RPMI medium supplemented with penicillin and streptomycin for 24 hours. Five or 25 × 106 cells of HT29 cells were seeded in 10-cm tissue culture dish and then cultured in serum-free DMEM supplemented with penicillin and streptomycin for 24 hours. HCCM was collected, centrifuged, and then filtered through 0.22-μm filter twice to remove cell debris. For proteinase K treatment, HCCM was incubated with proteinase K (200 μg/ml; Invitrogen) at 55°C for 4.5 hours. As a control, same amount of 10 mM tris-HCl (pH 7.5) was added. After the incubation, the HCCM was boiled for 20 min, centrifuged at 14,000 rpm for 10 min, and used for stimulation. Another control sample was prepared by adding proteinase K just before boiling. To deplete lipids, HCCM was mixed with 0.1 M KCl, 2.5 times volume of methanol, and 1.25 times volume of chloroform. Ten minutes later, the HCCM was further mixed with 1.25 times volume of chloroform and water and then centrifuged. An aqueous layer was collected, mixed with 0.25 times volume of chloroform, and centrifuged. An aqueous layer was collected, subjected to evaporation, resuspended with water, and used for stimulation. pH of HCCM was adjusted to around 7.4 either by adding NaHCO3 (23.81 mM) or Hepes (50 mM) and NaOH. FCS was added at 10% in HCCM before use. All the media were equilibrated under 5% CO2 and 37°C before use.

Preparation of acidic fresh medium

HCl was added in RPMI medium so that the pH became 6.3. FCS was added at 10% in the acidic fresh medium before use. The media were equilibrated under 5% CO2 and 37°C before use.

After cells were incubated with neutral or acidic RPMI medium for 30 min, intracellular pH was determined by using pHrodo Red AM intracellular pH indicator according to the manufacturer’s protocol (Thermo Fisher Scientific).

Western blot and immunoprecipitation

Whole-cell extracts (WCEs) were prepared by radioimmunoprecipitation assay (RIPA) lysis buffer (150 mM NaCl, 50 mM tris-HCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS). Western blotting was performed using anti-human RIPK3 (13), RIPK1 (BD Biosciences, 38/RIP), human RIPK1 (Abcam, ab178420), human RIPK1 pSer166 (Cell Signaling Technology, D1L3S), mouse RIPK1 pSer166 (Cell Signaling Technology, no. 31122), mouse RIPK3 pThr231/Ser232 (Abcam, ab222320), mouse caspase 8 (Enzo Life Sciences, 1G12), mouse FADD (provided by A. Winoto, University of California, Berkeley), IκBα (Cell Signaling Technology, L35A5), phospho-IκBα (Cell Signaling Technology, 5A5), phospho-Erk1/2 (Cell Signaling Technology, no. 4370), Erk1/2 (Cell Signaling Technology, no. 4695), phospho-JNK (Cell Signaling Technology, no. 4668), JNK1/2 (Cell Signaling Technology, no. 9252), phospho-Akt (Cell Signaling Technology, no. 4060), Akt (Cell Signaling Technology, no. 4691), phospho-CREB (cyclic adenosine monophosphate response element–binding protein)/ATF1 (activating transcription factor 1) (Cell Signaling Technology, no. 9196), ATF1 (Proteintech, 11946-1-AP), β-actin (ProSci, 3779), HSP90 (BD Biosciences, 610418), glyceraldehyde-3-phosphate dehydrogenase (EMD Millipore, 6C5), TRAF2 (Santa Cruz Biotechnology, N-19), TNFR1 (Abcam, ab19139), and HA (Roche, 3F10) antibodies. For RIPK3 and FADD immunoprecipitation (IP), WCE was prepared by RIPA lysis buffer and immunoprecipitated by anti-human RIPK3 (as described above) and mouse FADD (7A2, provided by A. Strasser, Walter & Eliza Hall Institute) antibodies. For TNFR1 IPs, WCE was prepared by R1 lysis buffer (150 mM NaCl, 20 mM tris-HCl, 1% NP-40, and 1 mM EDTA) and immunoprecipitated by anti-TNFR1 antibody (R&D Systems, MAB425). Protease inhibitor (Roche) and phosphatase inhibitor (Sigma) were added to the lysis buffer.

In vitro kinase assay

WCEs were subjected to IP using anti-RIPK1 antibody (BD Biosciences, 38/RIP) and Protein G Sepharose 4 Fast Flow (GE Healthcare). The immunoprecipitates were washed with cold lysis buffer thrice and subsequently twice with cold kinase reaction buffer (20 mM Pipes, 1 mM EDTA, 20 mM MgCl2, and 20 mM MnCl2; pH was adjusted to either 7.05 or 6.42 for neutral or acidic kinase reaction buffer by NaOH, respectively). Then, the immunoprecipitates were resuspended with either cold neutral or acidic kinase reaction buffer containing 2 mM dithiothreitol and 300 μM ATP. Where indicated, Nec1 was included. To test the reversibility of RIPK1 kinase activity, immunoprecipitated RIPK1 was incubated in cold acidic kinase reaction buffer for 1 hour, washed with cold neutral kinase reaction buffer, and subsequently subjected to in vitro kinase assay in neutral kinase reaction buffer. Kinase reaction was performed at 30°C for 30 min and then terminated by adding SDS loading dye and boiling for 5 min. The samples were subjected to Western blotting for RIPK1 pSer166.

Cell death assay

Cell death was determined by CellTiter-Glo Luminescent Cell Viability Assay (Promega), CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega), or flow cytometry using propidium iodide (PI) (Sigma) or PI/annexin V (BioLegend).

Real-time RT-PCR

RNA extraction was performed using the NucleoSpin RNA kit (Macherey-Nagel) according to the manufacturer’s protocol. The RNA was subjected to reverse transcription (RT) using PrimeScript Reverse Transcriptase (Takara Bio) followed by real-time polymerase chain reaction (PCR) using FastStart Universal Probe Master Mix (Roche Diagnostics) and a ViiA 7 Real-Time PCR System (Applied Biosystems). Glucose-6-phosphate dehydrogenase (G6PD) was used as an internal control. The following primers were used: 5′-CAGCCTCTTCTCCTTCCTGAT-3′ (forward) and 5′-GCCAGAGGGCTGATTAGAGA-3′(reverse) for Tnf and 5′-GCAAACAGAGTGAGCCCTTC-3′ (forward) and 5′-GGCCAGCCACATAGGAGTT-3′ (reverse) for G6pd. The data were analyzed by the comparative Ct (cycle threshold) method.

Statistical analysis

P values were calculated by unpaired t test with Welch’s correction. Normality was not formally assessed. Instead, individual data points were shown. P values less than 0.05 were considered statistically significant.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/13/631/eaay7066/DC1

Fig. S1. Low pH mediates the inhibitory effect of HCCM on TNF-induced necroptosis.

Fig. S2. Acidic extracellular pH inhibits TNF-induced necroptosis in multiple cell types.

Fig. S3. De novo gene expression does not contribute to the inhibitory effect of low extracellular pH on necroptosis.

Fig. S4. CYLD and cFLIP are not the targets of pH-dependent regulation of necroptosis.

Fig. S5. Knockdown of RIPK1 reduces TNF-induced apoptosis and necroptosis in Rela−/− MEFs.

Fig. S6. Low extracellular pH does not globally affect kinase function.

REFERENCES AND NOTES

Acknowledgments: We thank P. Gough and J. Bertin (GlaxoSmithKline) for GSK’481, GSK’728, and GSK’963, Z. G. Liu (NIH/NCI) for Rela−/− MEFs, A. Winoto (University of California, Berkeley), and A. Strasser (The Walter and Eliza Hall Institute) for anti-FADD antibody, Novartis for LBW242, and D. Vucic (Genentech) for BV6. Funding: This work was supported by NIH grant AI119030 (F.K.-M.C.) and the Takeda Science Foundation (K.M.). Author contributions: K.M. and F.K.-M.C. designed the project. K.M. performed experiments with assistance from S.B. and analyzed data. K.M. and F.K.-M.C. wrote the paper. F.K.-M.C. supervised the project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: GSK’728 requires a material transfer agreement from GlaxoSmithKline. All data and materials needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.

Stay Connected to Science Signaling

Navigate This Article