TRAIL-death receptor endocytosis and apoptosis are selectively regulated by dynamin-1 activation

Significance

Clathrin-mediated endocytosis (CME) regulates receptor trafficking, thereby affecting several cellular signaling pathways. We discovered that dynamin-1 is selectively activated downstream of TNF-related apoptosis-inducing ligand–death receptors (TRAIL–DRs) to self-regulate their endocytosis, attenuate apoptotic signaling, and increase cell survival. Activation of initiator caspase-8 by TRAIL–DRs triggers spikes of Ca²⁺ through ryanodine receptor calcium channels, activating calcineurin, and in turn dephosphorylating dynamin-1 to promote cargo-selective endocytosis of TRAIL–DR. This study delineates specific mechanisms linking signaling downstream of cell-surface receptors to the regulation of cargo-selective CME, and thus their signaling properties. Cancer cell-specific adaptation of this bidirectional crosstalk between signaling and CME has implications for tumor progression and metastasis.

Abstract

Clathrin-mediated endocytosis (CME) constitutes the major pathway for uptake of signaling receptors into eukaryotic cells. As such, CME regulates signaling from cell-surface receptors, but whether and how specific signaling receptors reciprocally regulate the CME machinery remains an open question. Although best studied for its role in membrane fission, the GTPase dynamin also regulates early stages of CME. We recently reported that dynamin-1 (Dyn1), previously assumed to be neuron-specific, can be selectively activated in cancer cells to alter endocytic trafficking. Here we report that dynamin isoforms differentially regulate the endocytosis and apoptotic signaling downstream of TNF-related apoptosis-inducing ligand–death receptor (TRAIL–DR) complexes in several cancer cells. Whereas the CME of constitutively internalized transferrin receptors is mainly dependent on the ubiquitously expressed Dyn2, TRAIL-induced DR endocytosis is selectively regulated by activation of Dyn1. We show that TRAIL stimulation activates ryanodine receptor-mediated calcium release from endoplasmic reticulum stores, leading to calcineurin-mediated dephosphorylation and activation of Dyn1, TRAIL–DR endocytosis, and increased resistance to TRAIL-induced apoptosis. TRAIL–DR-mediated ryanodine receptor activation and endocytosis is dependent on early caspase-8 activation. These findings delineate specific mechanisms for the reciprocal crosstalk between signaling and the regulation of CME, leading to autoregulation of endocytosis and signaling downstream of surface receptors.

Receptor-mediated endocytosis plays a critical role in regulating signaling, by either promoting rapid endocytosis of ligand–receptor complexes and attenuating cell-surface signaling, or by promoting the formation of endosomes that can serve as signaling platforms for these complexes (1, 2). Clathrin-mediated endocytosis (CME) is one of the most important and well-characterized endocytic pathways in eukaryotes (3, 4). The CME core components—clathrin, dynamin, and adaptor protein 2 (AP2)—interact with several endocytic accessory proteins to initiate, stabilize, and promote the maturation of clathrin-coated pits (CCPs). Following maturation, CCP scission is catalyzed by the large GTPase dynamin, leading to the formation of cargo-containing vesicles (5, 6). Once thought to be a constitutive process, it is now recognized that CME can be highly regulated (7), but many questions remain as to the molecular mechanisms underlying the regulation of CME. Moreover, recent data have suggested that signaling G protein-coupled receptors (GPCRs) can directly regulate CCP dynamics through selective recruitment of dynamin and endocytic accessory proteins (8, 9). The extent of possible crosstalk between signaling receptors and CME has not been explored.

Dynamins are master regulators of CME. In addition to their role in promoting the fission of invaginated CCPs, dynamins control earlier rate-limiting steps of clathrin-coated vesicle formation (10–12). There are three dynamin isoforms in vertebrates: dynamin-1 (Dyn1) is predominantly expressed in neurons, dynamin-2 (Dyn2) is ubiquitously expressed, and dynamin-3 (Dyn3) is expressed in neurons, lung, and testis. Dyn1 and Dyn2 are distinct in their curvature sensing/generating properties: Dyn1 exhibits high curvature-generating properties when compared with Dyn2, making it more suited for rapid compensatory endocytosis (13). Hence, at the synapse, Dyn1 is well-suited to mediate rapid compensatory endocytosis and synaptic vesicle recycling following neurotransmission (14). Unexpectedly, we have recently shown that Dyn1 can also be activated in nonneuronal cells, downstream of an Akt/GSK3β signaling cascade to alter the rate and regulation of CME (15), linking endocytosis to signaling.

Activation of the apoptosis-signaling machinery by TRAIL (TNF-related apoptosis-inducing ligand) through the engagement of death receptors (DRs) has gained considerable interest as a potential anticancer strategy (16–19); however, its efficacy is limited by a variety of cancer cell-resistance mechanisms. TRAIL binding to its cognate death receptors (DR4 and DR5) triggers the formation of the death-inducing signaling complex by the recruitment of the adaptor FADD (Fas-associated death domain) and the initiator caspase-8 (20). Activated caspase-8 can then activate effector caspase-3 and -7, leading to cell-extrinsic apoptosis. Binding of human recombinant TRAIL to its cognate apoptosis-inducing DR (DR4 or DR5) stimulates their internalization via CME (21, 22). However, there are conflicting reports as to the effect of TRAIL-induced endocytosis of DRs on apoptotic signaling (21–23). Initial experimental data suggested that CME of TRAIL–DR complexes has an inhibitory effect on TRAIL-induced apoptosis (21, 22). However, using overexpressed dominant-negative dynamins to block CME, others have shown that endocytosis of TRAIL–DR is required for apoptosis, and proposed possible cell-type–specific differences in TRAIL signaling (23).

In this study, we sought to determine the contribution of CME to the regulation of signaling via TRAIL–DR. Experimental down-regulation of all of the core components of the CME machinery revealed that dynamin isoforms differentially regulate CME of selected cargoes: whereas CME of constitutively internalized transferrin receptors (TfnR) requires Dyn2, CME of TRAIL–DR complexes is Dyn1-dependent. We discovered that TRAIL-induced early activation of caspase-8 results in ryanodine receptor (RyR)-mediated calcium release, causing the Ca²⁺ and calcineurin-dependent activation of Dyn1, Dyn1-dependent CME of TRAIL–DR, and suppression of TRAIL-induced apoptosis.

Results and Discussion

Dynamins Differentially Regulate Cargo-Selective Endocytosis and TRAIL-Induced Apoptosis.

Given the existence of conflicting reports (21–23), we directly tested the role of CME in regulating TRAIL-induced apoptosis by small-interfering RNA (siRNA)-mediated knockdown of core components of the CME machinery, namely the coat protein clathrin heavy chain (CHC), the AP2, and the ubiquitously expressed isoform of dynamin, Dyn2. TRAIL-sensitive MDA-MB-231 human breast adenocarcinoma cells and TRAIL-resistant A549 human lung adenocarcinoma cells depleted for CHC or AP2 showed a significant sensitization to TRAIL-induced cell death, compared with control siRNA-treated cells (Fig. 1A and Fig. S1 A and B). Unexpectedly, depletion of Dyn2 showed a relatively modest effect on TRAIL-induced cell death. Because Dyn1 has recently been shown to be activated in nonneuronal cancer cells (15), we also tested the effect of Dyn1 knockdown. Strikingly, Dyn1-depleted cells were as sensitized to TRAIL-induced apoptosis as those depleted of AP2 or CHC (Fig. 1A and Fig. S1 A and B). This finding was true for both DR4- and DR5-triggered apoptosis, as assessed using DR-selective variants of TRAIL (24, 25) (Fig. S1 C and D). Western blotting confirmed that the differential effects of Dyn1 and Dyn2 were not a result of differential efficiencies of knockdown or because of compensatory up-regulation of the nontargeted isoform (Fig. S2 A and B). Indeed, depletion of Dyn1 did not impact the protein levels of Dyn2 in A549 cells and MDA-MB-231 cells and Dyn2 knockdown only moderately (by ∼20%) increased the protein levels of Dyn1 in A549 cells (Fig. S2 A and B).

Fig. 1.

Differential regulation of cargo-receptor endocytosis and TRAIL-induced apoptosis by Dyn1. (A) TRAIL activity (100 ng/mL) in MDA-MB-231 (Upper) and A549 cells (Lower) treated with the indicated siRNAs. (B) FLAG-tag TRAIL uptake or (C) TfnR uptake in MDA-MB-231 (Upper) and A549 cells (Lower). (D) TRAIL activity in A549 WT, Dyn1^KO, and Dyn1^KO cells reconstituted with Dyn1-EGFP. (E) TRAIL-induced caspase-3/7 activation in parental A549 cells, Dyn1^KO and Dyn1^KO cells reconstituted with Dyn1-EGFP. (F) FLAG-tag TRAIL uptake in the indicated cell lines. Reduction in cell viability (%) was calculated relative to control wells containing no ligand. All internalization rates were calculated relative to surface-bound TfnR or FLAG-tag TRAIL. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. *P < 0.05, **P < 0.005, ***P < 0.0005; n.s., nonsignificant.

Fig. S1.

Dynamin isoforms differentially regulate TRAIL-mediated cell killing. TRAIL effects on cell viability in control, AP2, CHC, Dyn1, and Dyn2 siRNA-treated cells. (A) MDA-MB-231 and (B) A549 cells were incubated with the indicated concentrations of TRAIL and assayed for cell viability. TRAIL-resistant A549 were incubated with the indicated concentrations of a DR4-specific variant (TRAIL 4C7) (1–1,000 ng/mL) (C), or of a DR5-specific variant (TRAIL D269H/E195R) (D), and then assayed for cell viability. Reduction in cell viability (%) was calculated relative to control wells containing no ligand. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. *P < 0.05, ***P < 0.0005.

Fig. S2.

Down-regulation of CME components increased surface expression of DR. siRNA-mediated knockdown of Dyn1 and Dyn2 in A549 cells. (A) Immunoblots probed for Dyn1 and Dyn2. (B) Quantification of Dyn2 and Dyn1 expression differences in siRNA knockdown cells. Normalized levels of surface bound FLAG-tag TRAIL (relative to control) in MDA-MB-231 (C) and A549 cells (D) treated with Dyn1, Dyn2, AP2, and CHC siRNA. The results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. *P < 0.05, ***P < 0.0005.

To determine whether the isoform-specific functions of Dyn1 and Dyn2 were related to their roles in endocytosis, we assessed the effect of their down-regulation on TRAIL uptake. siRNA-mediated depletion of Dyn1 potently reduced TRAIL–DR endocytosis to the same extent as depletion of AP2 or CHC (Fig. 1B), resulting in increased levels of surface-bound TRAIL (Fig. S2 C and D). Depletion of Dyn2 was much less effective in reducing TRAIL–DR endocytosis. The opposite was true for uptake of the classic CME cargo, TfnR, which was mainly dependent on Dyn2 but not on Dyn1 (Fig. 1C). To confirm these effects, we generated CRISPR-Cas9n Dyn1 knockout (Dyn1^KO) A549 cells (Fig. S3A). As seen in siRNA-mediated knockdown cells, A549 Dyn1^KO cells exhibited increased sensitivity to TRAIL-induced apoptosis, as measured by cell viability (Fig. 1D) and caspase-3/7 activity (Fig. 1E), and showed a strong impairment in TRAIL–DR endocytosis (Fig. 1F). Importantly, both TRAIL resistance and TRAIL–DR endocytosis were restored by reconstitution of Dyn1^KO cells with Dyn1-EGFP (Fig. 1 D–F).

Fig. S3.

Influence of GSK3β inhibition on TRAIL activity in A549 (WT and Dyn1^KO) cells. (A) Immunoblots of two representative single clones. Clone 11 was chosen for additional assays (50, 25, and 12.5 µg of loaded protein per each condition). (B) Levels of surface bound FLAG-tag TRAIL (100 ng/mL) in A549 WT and Dyn1^KO cells after different incubation times at 4 °C. (C) TRAIL effects on cell viability in A549 WT and Dyn1^KO cells, in the presence or absence of the GSK3β inhibitor Chir99021 (10 µM). (D) Effectiveness of Chir99021 after 15-min incubation, as measured by phosphorylated S6K. Cells were preincubated with Chir99021 for 30 min, followed by treatment with several concentrations of TRAIL (1–1,000 ng/mL) and assayed for cell viability. Reduction in cell viability (%) was calculated relative to control wells containing no ligand. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. ***P < 0.0005.

Knockdown of Dyn1, AP2, and CHC resulted in an increase in cell surface binding of TRAIL, presumably because of inhibition of constitutive DR endocytosis. Thus, prolonged inhibition of CME could have increased TRAIL-mediated apoptosis simply by increasing the basal levels of surface DR before TRAIL exposure. To test this theory, we titrated cell surface binding of TRAIL in A549 WT and Dyn1^KO cells. As seen in Fig. 2A, at subsaturating concentrations of TRAIL, and within the range that showed enhanced apoptotic signaling in Dyn1 knockdown cells (i.e., 1–100 ng/mL), there were no differences in TRAIL binding when comparing WT to Dyn1^KO cells. The same was true even after 4 h of incubation at 4 °C with 100 ng/mL TRAIL (Fig. S3B). The increased surface levels of DR in Dyn1^KO cells were only apparent at high, saturating concentrations of TRAIL (>500 ng/mL). Although DR–ligand interactions will be different at 4 °C vs. 37 °C, these results nonetheless suggest that apoptosis occurs primarily through the inhibition of TRAIL-induced DR endocytosis, and not because of differences in the surface expression of DRs, in agreement with previous reports wherein DR expression was not predictive of clinical responses (26, 27).

Fig. 2.

TRAIL induces calcineurin-mediated dephosphorylation of Dyn1 and TRAIL–DR endocytosis to reduce apoptosis. (A) TRAIL binding to WT and Dyn1^KO A549 cells measured at 4 °C. (B) Effects of calcineurin inhibition by CsA (20 µM) on TRAIL activity in A549 WT and Dyn1^KO cells. (C) TRAIL activity (100 ng/mL) in A549 WT, Dyn1^KO, and Dyn1^KO reconstituted with Dyn1 WT (KO + WT) or the nonphosphorylatable mutant, Dyn1 S774A/S778A (KO + AA), in the presence or absence of CsA. (D) FLAG-tag TRAIL uptake in control and CsA pretreated cells. (E) TfnR uptake in the presence or absence of CsA. (F) Caspase-3/7 activation in TRAIL-treated MDA-MB-231 cells in the presence or absence of CsA. (G) Dyn1 phosphorylation (pSer774) profiles upon TRAIL stimulation (100 ng/mL) in MDA-MB-231 cells in the presence or absence of CsA, and quantification of pS774-Dyn1/total-Dyn1. (H) TRAIL-induced apoptosis in MDA-MB-231 WT and Dyn1 overexpressing cells in the presence or absence of CsA. All internalization rates were calculated relative to surface bound TfnR or FLAG-tag TRAIL. Reduction in cell viability by TRAIL (%) was calculated relative to control wells incubated with solvent alone or 20 µM of CsA. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. *P < 0.05, **P < 0.005, ***P < 0.0005; ns, nonsignificant.

Taken together, our findings confirm previous reports (21, 22) of an important role for TRAIL–DR endocytosis in the negative regulation of TRAIL-induced apoptosis, and furthermore, establish the existence of cargo-selective, dynamin isoform-specific mechanisms for endocytosis.

TRAIL Induces Calcineurin-Mediated Dephosphorylation of Dyn1 and DR Endocytosis.

We next investigated the mechanism by which Dyn1 is activated downstream of DRs. At the synapse, Dyn1 activity is regulated by cycles of phosphorylation/dephosphorylation, including its phosphorylation at Ser778 and Ser774, the latter by GSK3β (28). Thus, we examined whether inhibition of GSK3β, which activates Dyn1 (15), might alter the cell sensitivity to TRAIL-induced apoptosis in the context of Dyn1 expression, using A549 WT and Dyn1^KO cells. Pretreatment with Chir99021, which potently inhibits GSK3β, did not alter the sensitivity to TRAIL-induced apoptosis in either cell line (Fig. S3 C and D).

Dyn1 can also be activated through dephosphorylation by the calcium (Ca²⁺) and calmodulin-dependent phosphatase, calcineurin (29, 30). Inhibiting calcineurin by either cyclosporin A (CsA) (Fig. 2B) or FK506 (Fig. S4), resulted in a significant increase in sensitivity to TRAIL-induced apoptosis in A549 WT cells, comparable to the levels seen in Dyn1^KO cells. Previous studies have shown that high concentrations of CsA can cause endoplasmic reticulum (ER) stress (31), which in turn can lead to apoptosis via DR5 independently of TRAIL (32). However, in the cell lines tested and at the incubation times and concentrations of CsA used (below 40 µM), cytotoxicity remained absolutely dependent on TRAIL (Fig. S5). Moreover, CsA did not further sensitize Dyn1^KO cells to TRAIL-induced cell death (Fig. 2 B and C); however, sensitivity to CsA was fully restored in A549 Dyn1^KO cells by reintroducing Dyn1 WT. Importantly, Dyn1^KO cells reconstituted with a nonphosphorylatable, and hence constitutively active mutant Dyn1^S774A/S778A (KO+AA), were both resistant to TRAIL-induced apoptosis and insensitive to CsA (Fig. 2C). Similar results were obtained in H1299 Dyn1^KO cells reconstituted with the above-mentioned proteins (Fig. S6A). These results establish that the effects of calcineurin inhibition on TRAIL-induced apoptosis are mediated primarily through Dyn1 activation. The effects of Dyn1 depletion and calcineurin inhibition on sensitivity to TRAIL-induced apoptosis were also observed in several other cancer cell lines, including HCT116 colorectal carcinoma, SCC61 squamous carcinoma, HT1080 fibrosarcoma, and MV3 melanoma cells, but not, as expected, in the caspase-3–deficient TRAIL-resistant MCF-7 mammary carcinoma cells (Fig. S6 B and C). These results establish that calcineurin-mediated dephosphorylation of Dyn1 on Ser774/778 plays a critical role in TRAIL sensitization in diverse cancer cell lines.

Fig. S4.

Calcineurin inhibition by FK506 enhances TRAIL-mediated cell killing. Effects of calcineurin inhibition on TRAIL activity in MDA-MB-231 cells (A) or A549 cells (B). Control cells or cells pretreated with FK506 (20 µM) were incubated with the indicated concentrations of TRAIL and assessed for cell viability. Reduction in cell viability by TRAIL (%) was calculated relative to control wells incubated with 0 or 20 µM of FK506. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. ***P < 0.0005.

Fig. S5.

CsA treatment does not affect cell viability in the absence of TRAIL. (A) Different concentrations of CsA (up to 40 µM) do not affect cell viability in A549 cells. (B) CsA enhances TRAIL-mediated cell death in A549 cells. Cells were preincubated with CsA (0-40 µM) for 30 min, followed by treatment with or without 100 ng/mL of TRAIL and assayed for cell viability. Reduction in cell viability (%) was calculated relative to control wells containing no ligand. All of the results are mean values ± SD (n = 3).

Fig. S6.

Dyn1 activation by calcineurin blocks TRAIL-induced apoptosis in H1299 cells and influences TRAIL activity in several cancer cells. (A) TRAIL activity (100 ng/mL) in H1299 WT, Dyn1^KO, and H1299 Dyn1^KO cells reconstituted with Dyn1 WT (KO + WT) and Dyn1 S774A/S778A (KO + AA), in the presence or absence of calcineurin inhibitors. Cells were preincubated with CsA (20 µM) for 30 min, followed by treatment with 100 ng/mL of TRAIL and assayed for cell viability. Reduction in cell viability (%) was calculated relative to control wells containing no ligand. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. **P < 0.005; ns, nonsignificant. (B) TRAIL activity (100 ng/mL; overnight incubation) in the indicated cell lines, in both control (si-Ctrl) and Dyn1-depleted (si-Dyn1) cells, calculated relative to control wells containing no TRAIL. (C) Effects of calcineurin inhibition by CsA on TRAIL activity in several human cancer cell lines, including HCT116 colorectal carcinoma, SCC61 squamous carcinoma, HT1080 fibrosarcoma and MV3 melanoma cells, and caspase-3–deficient MCF-7 mammary carcinoma cells. Reduction in cell viability by TRAIL (%) was calculated relative to control wells incubated with 0 or 20 µM of CsA. All of the results are mean values ± SD (n = 3). (D) TRAIL-induced caspase-8 cleavage and activation is enhanced in CsA pretreated cells. MDA-MB-231 cells were treated with TRAIL (1 μg/mL) in the presence or absence of CsA for the time points indicated and analyzed by immunoblot for processing of caspase-8.

We next tested whether inhibition of calcineurin activity effected TRAIL–DR endocytosis. Treatment of A549 WT cells with the calcineurin inhibitor CsA significantly inhibited TRAIL–DR uptake, comparable to the reduced kinetics of TRAIL–DR uptake seen in Dyn1^KO cells. CsA had no effect on the residual TRAIL uptake in Dyn1^KO cells (Fig. 2D). In marked contrast, TfnR uptake remained unaltered under all these conditions (Fig. 2E). These results establish that calcineurin modulates TRAIL–DR endocytosis, thus regulating DR signaling, via the activation of Dyn1.

We confirmed that calcineurin inhibition acts at early stages in DR signaling by analyzing the effects of CsA treatment on caspase-8 processing. Both the rate of onset and the extent of TRAIL-induced caspase-8 cleavage were markedly increased in CsA-treated cells, compared with control cells (Fig. S6D). Moreover, the increased sensitivity of MDA-MB-231 cells to TRAIL-induced apoptosis upon inhibition of calcineurin was especially apparent when measuring caspase-3/7 activation (Fig. 2F). We also sought direct evidence that TRAIL and calcineurin inhibitors were indeed acting through Dyn1 by modulating its phosphorylation. To this end, MDA-MB-231 cells were incubated with TRAIL for different time periods, and changes in Dyn1 phosphorylation levels were measured using a phospho-S774–specific antibody. Our results show a time-dependent decrease in Dyn1 phosphorylation upon TRAIL incubation, relative to total Dyn1 (Fig. 2G). Importantly, TRAIL-mediated dephosphorylation of Dyn1 was completely dependent on calcineurin, as CsA abolished Dyn1 dephosphorylation mediated by TRAIL (Fig. 2G). From these results, we conclude that TRAIL-induced activation of Dyn1 is regulated by calcineurin activation.

Dyn1 is overexpressed in several cancers, including leukemia, lung, and colon adenocarcinomas (33, 34). Therefore, we tested the impact of Dyn1 overexpression on TRAIL-induced apoptosis in TRAIL-sensitive MDA-MB-231 cells. Transient overexpression of Dyn1-EGFP in MDA-MB-231 cells increased their resistance to TRAIL-induced cell death. The effect of Dyn1 overexpression was abrogated by treatment with CsA, which dramatically enhanced sensitivity to TRAIL-induced cell death in both Dyn1 overesspression and control cells (Fig. 2H).

Calcium Release by RyRs Regulates TRAIL–DR Endocytosis.

Given that calcineurin is a Ca²⁺-dependent serine-threonine phosphatase, we explored the source of Ca²⁺ needed for its activation. Pretreatment of MDA-MB-231 cells with the intracellular Ca²⁺ chelator BAPTA-AM increased their sensitivity to TRAIL-induced apoptosis, whereas chelating extracellular Ca²⁺ by EGTA had no effect, suggesting that intracellular Ca²⁺ is involved in enhanced sensitization to TRAIL-mediated apoptosis (Fig. 3A). Intracellular Ca²⁺ ions are generally stored in the ER, and released into the cytosol upon extracellular signaling by either of two ER calcium channels: inositol triphosphate receptors (IP3R) or RyRs (35). We tested the involvement of each of these by incubating cells with Xestospongin C (XesC) or Ryanodine (Ry), specific inhibitors of IP3R or RyR, respectively. Preincubation of MDA-MB-231 cells with up to 20 µM XesC, which potently inhibits IP3R (Fig. S7A), did not affect TRAIL-induced apoptosis (Fig. 3B). In contrast, treatment with Ry, known to inhibit RyR at higher concentrations (>10 µM), resulted in a dose-dependent increase in sensitivity to TRAIL-induced cell death (Fig. 3C), and a corresponding TRAIL-dependent increase in caspase-3/7 activation (Fig. 3D). As predicted, TRAIL–DR endocytosis was also inhibited upon blocking RyR (Fig. 3E), whereas TfnR uptake remained unaltered (Fig. 3F). Moreover, Dyn1^KO A549 cells reconstituted with the constitutively active Dyn1^S774A/S778A mutant, but not WT, were insensitive to RyR inhibition (Fig. S7B), confirming that RyR activity is upstream of Dyn1 activation.

Fig. 3.

TRAIL-mediated Ca²⁺ release by RyR regulates TRAIL–DR endocytosis and apoptosis in MDA-MB-231 cells. (A) The effect of TRAIL on cell viability in control, BAPTA-AM (10 µM) and EGTA (1 mM) -treated cells. Cell viability after cells are pretreated or not with (B) XesC, an IP3R inhibitor (20 µM), or (C) with different concentrations of Ry, a RyR inhibitor, followed by TRAIL incubation. (D) TRAIL induced caspase-3/7 activation in the presence or absence of 40 µM Ry. (E) FLAG-tag TRAIL uptake in control and Ry pretreated cells. (F) TfnR uptake in control and Ry pretreated cells. Internalization rates were calculated relative to surface bound TfnR or FLAG-tag TRAIL. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. *P < 0.05, ***P < 0.0005.

Fig. S7.

Validation of IP3R inhibition by XesC and evidence that the effects of RyR inhibition on apoptosis depend on Dyn1 phosphorylation. (A) XesC reduces IP3R-mediated release of calcium induced by bradykinin (BK, 300 nm; gray traces), compared with A549 control (black traces) cells. Results shown mean ± SD from 20 cells per condition. (B) Dyn1 KO A549 cells reconstituted with nonphosphorylatable and constitutively active Dyn1^AA (KO + AA) are insensitive to Ry. A549 cells, as indicated, were preincubated with Ry for 30 min, followed by treatment with 100 ng/mL of TRAIL and assayed for cell viability. Reduction in cell viability (%) was calculated relative to control wells containing no ligand. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. ***P < 0.0005.

TRAIL-Induced Caspase-8 Activation Promotes RyR-Dependent Calcium Spikes.

We next sought direct evidence for TRAIL-stimulated ER Ca²⁺ release using MDA-MB-231 cells transfected with a genetically encoded calcium sensor, GCaMP6f (36). Cells imaged after incubation with TRAIL exhibited periodic and transient spikes of elevated Ca²⁺ (Fig. 4 A and B and Fig. S8), which were completely blocked when RyR was inhibited by high concentrations of Ry (Fig. 4B). Calcium signaling downstream of TRAIL–DR activation has not previously been reported, raising the question of the mechanism for coupling DRs to RyR. Interestingly, another TNF superfamily member, TNF-α, potentiates terminal afferent responses through a RyR-mediated calcium-release mechanism (37). TNF-α–induced caspase-8 activation induces RyR S-nitrosylation, resulting in Ca²⁺ release (38). Thus, we tested for the specific involvement of caspase-8 on TRAIL-induced Ca²⁺-signaling. We observed a significant decrease in TRAIL-stimulated calcium spike formation in caspase-8–inhibited cells, with most cells showing much fewer and less-intense calcium elevations, compared with TRAIL-treated cells (Fig. 4 B and C and Fig. S9A). Consistent with the block in Ca²⁺ release, inhibition of caspase-8—or its siRNA-mediated knockdown (Fig. S9B)—also inhibited TRAIL–DR endocytosis, without significantly affecting the uptake of TfnR (Fig. 4D). Thus, activation of caspase-8 can either trigger or negatively regulate apoptotic signaling, presumably reflecting a threshold effect that controls apoptosis.

Fig. 4.

TRAIL induces RyR-dependent calcium spike formation. (A) Total internal reflection fluorescence microscopy (TIRFM) time-lapse images of MDA-MB-231 expressing GCaMP6f after TRAIL incubation (100 ng/mL) in the presence or absence of Ry. (Scale bars: 5 μm.) (B) Quantification of Ca²⁺ oscillations in control, TRAIL, and TRAIL+Ry-treated cells. Boxes illustrate the time course where the Ca²⁺ oscillations induced by TRAIL shown in A occurred. (C) Quantification of Ca²⁺ oscillations in TRAIL + caspase-8 inhibitor (Z-IETD-FMK, 20 µM) -treated cells. (D) TRAIL and TfnR uptake in control cells (Ctrl) and cells pretreated with caspase-8 inhibitor 30 min before the uptake assay. Internalization rates were calculated relative to surface bound TRAIL or TfnR, respectively. All of the results are mean values ± SD (n = 3). Two-tailed Student’s t tests were used to assess statistical significance. *P < 0.05, **P < 0.005.

Fig. S8.

TRAIL induces transient Ca²⁺ spikes in MDA-MB-231 cells expressing GCaMP6. TIRFM time-lapse images of representative MDA-MB-231 expressing GCaMP6f after TRAIL incubation. Boxes illustrate the time course where the Ca²⁺ oscillations occurred, at the locations numbered in the Upper panel.

Fig. S9.

Caspase-8 inhibition reduces TRAIL induced transient Ca²⁺ spikes and TRAIL endocytosis in MDA-MB-231 cells. (A) Representative TIRFM time-lapse images of MDA-MB-231–expressing GCaMP6f after TRAIL incubation in the presence of caspase-8 inhibitors (10 movies). Boxes illustrate the time course where the Ca²⁺ oscillations occurred, at the locations numbered in the Upper panel. (B) TRAIL or TfnR uptake in control (si-Ctrl) and caspase-8–depleted cells (si-Casp8) and Western blot showing efficiency of caspase-8 depletion. Internalization rates were calculated relative to surface bound TRAIL or TfnR, respectively. All of the results are mean values ± SD (n = 3).

Conclusions and Perspectives

Endocytosis is a major regulator of cellular signaling (2, 39). Herein we present evidence for the direct, reciprocal regulation of CME by signaling receptors. Our studies uncover an early feedback loop relying on TRAIL–DR-mediated activation of initiator caspases to promote TRAIL–DR endocytosis. We show that TRAIL-activated DRs trigger RyR-dependent Ca²⁺ release from ER stores, induce calcineurin-dependent dephosphorylation, and thereby activation of Dyn1, leading to cargo-selective uptake of DRs and the attenuation of their apoptotic signaling (Fig. 5). Although DRs are the first surface-signaling receptor shown to regulate their own CME and signaling, through a dynamin-isoform–specific mechanism, we predict that other surface signaling receptors participate in this bidirectional crosstalk between signaling and CME. Indeed, others have shown that GPCRs can regulate CCP dynamics via recruitment of dynamin (8), although isoform-specificity and effects on signaling have not been studied.

Fig. 5.

Proposed model for Dyn1-dependent endocytosis of TRAIL–DR complexes to suppress apoptotic signaling. TRAIL-induced DR activation leads to caspase-8 cleavage/activation, which in turn activates RyR-mediated Ca²⁺ release from ER stores and calcineurin activation to dephosphorylate and activate Dyn1. The selective regulation of TRAIL–DR endocytosis suppresses TRAIL-mediated apoptosis in cancer cells.

Interestingly, the α-adaptin subunit of AP2 is a substrate for caspase-8–dependent cleavage, resulting in removal of the C-terminal appendage domain (21). We previously showed that Dyn1 is activated in cells expressing a similar truncation mutation of α-adaptin, altering the regulation of CME (15). Thus, this cleavage, which occurs at later times after exposure to TRAIL (21), may be a mechanism for sustained regulation of CME that is likely less cargo-selective.

The overexpression and activation of Dyn1 in cancer cells might be an adaptation that enhances their survival, migration, and other properties that contribute to cancer cell aggressiveness. In support of this possibility, Dyn1—but not Dyn2—is overexpressed in many lung tumor-derived lung cancer cells relative to normal bronchial epithelial cells (Fig. S10A). Moreover, lower survival rates in lung cancer patients are linked to high levels of Dyn1, but not Dyn2 expression, especially among smokers (Fig. S10B). Thus, Dyn1-mediated modulation of endocytic trafficking emerges as an important physiological regulator of DR-mediated signals, further implicating the regulation of CME by dynamins as a mechanism exploited by cancer cells to escape apoptotic cell death. More generally, given that Dyn1 can also be activated downstream of oncogenic Akt signaling (15), Dyn1 becomes a potential nexus for reciprocal regulation of cell-surface receptor signaling and CME in cancer cells. Further exploration of the differential endocytic adaptations exploited by cancer cells may provide new targeting strategies to combat this disease.

Fig. S10.

High Dyn1 expression in lung tumor cells and its effect on patient survival. (A) Dyn1 and Dyn2 mRNA expression in lung tumor-derived (n = 108) and normal (n = 50) bronchial epithelial cell lines. (B) Dyn1, but not Dyn2, expression is a prognostic indicator of survival rates in lung cancer patients, including patients with smoking history. Data were derived from kmplot.com/lung.

Materials and Methods

Detailed methods are provided in SI Materials and Methods. In brief, cells lines used in this study, including MDA-MB-231 breast adenocarcinoma, H1299 nonsmall-cell lung cancer cells, A549 lung carcinoma, HT1080 fibrosarcoma, MV3 melanoma cells, HCT116 colon adenocarcinoma cells, MCF-7 breast adenocarcinoma, and Scc61 squamous cell carcinoma cells were cultured as described in SI Materials and Methods.

siRNA transfections were performed with previously established siRNA sequences using RNAiMAX (Life Technologies), as described in SI Materials and Methods.

Dyn1 KO A549 and H1299 cells were generated using a CRISPR-Cas9n double-nicking strategy. Cell viability assays were performed by using the CCK-8 Counting Kit (Dojindo), according to the manufacturer’s instructions. Caspase-3/7 activation was assessed using the Apo-ONE Homogeneous Caspase-3/7 assay (Promega).

Endocytosis of Transferrin receptors or N-terminal FLAG-tag TRAIL was measured using the anti-TfnR (HTR-D65) or anti-FLAG mouse mAbs. TfnR and TRAIL receptor cell surface expression were measured in parallel by incubating MDA-MB-231 or A549 cells with the respective ligands at 4 °C for 30 min. Internalized ligand was expressed as the percentage of the total surface-bound ligand at 4 °C (i.e., without the acid wash step), measured in parallel.

Calcium imaging was performed using MDA-MB-231 transiently expressing GCaMP6f or A549 cells preincubated with Fura2.

SI Materials and Methods

Cell Culture and Reagents.

MDA-MB-231 breast adenocarcinoma (from R. Brekken, University of Texas Southwestern Medical Center, Dallas), HT1080 fibrosarcoma (from Sarah Courtneidge, Sanford Burnham, San Diego), MV3 melanoma cells (a gift from Sean Morrison, University of Texas Southwestern Medical Center, Dallas) were grown in DMEM containing high glucose medium (Life Technologies), supplemented with 20 mM Hepes and 10% (vol/vol) FCS (HyClone). A549 lung carcinoma, H1299 (nonsmall-cell lung carcinoma) and HCT116 colon adenocarcinoma cells (from John Minna, University of Texas Southwestern Medical Center, Dallas) were grown in RPMI supplemented with 20 mM Hepes and 5% (vol/vol) FCS. MCF-7 breast adenocarcinoma (from ATCC) were grown in RPMI supplemented with 20 mM Hepes and 10% (vol/vol) FCS. Scc61 squamous cell carcinoma cells (from ATCC) were grown in DMEM high-glucose medium supplemented with 20 mM Hepes, 0.4 μg/mL of hydrocortisone, and 20% (vol/vol) FCS (HyClone). The GSK3β inhibitor CHIR-99021, calcineurin inhibitors, cyclosporin A and FK506, and the calcium chelator BAPTA-AM were purchased from Sigma-Aldrich. EGTA was from Fisher-Scientific. Akt inhibitor X was purchased from CalBiochem. XesC and Ry were purchased from Santa Cruz Biotechnology. Human recombinant TRAIL WT (FLAG-tag TRAIL) was from Enzo Life Sciences. DR-selective mutants were expressed, purified, and characterized as described previously (24, 25).

siRNA Transfection.

MDA-MB-231 and A549 cells were treated with previously established siRNA sequences using RNAiMAX (Life Technologies) to silence endogenous proteins, following the manufacturer’s instructions. Briefly, 110 pmol of the indicated siRNA and 6.5 µL of RNAiMAX reagent were added in 2 mL of OptiMEM (Life Technologies) in each well of a six-well plate containing cells for 4 h. Transfection was performed at days 1 and 3 after plating, and experiments were performed at day 5. Oligos used to target the following proteins were endogenous α-adaptin (5′-GAGCAUGUGCACGCUGGCCA-3′), dynamin-1 (5-GGCUUACAUGAACACCAACCACGAA-3), dynamin-2 (Dyn2_1: 5-CCGAAUCAAUCGCAUCUUCUU-3 and Dyn2_2: 5 -GACAUGAUCCUGCAGUUCAUU-3), as well as CHC, as previously described (40). The AllStars Negative siRNA nontargeting sequence was purchased from Qiagen.

Generation of Dyn1 KO A549 Cells by CRISPR-Cas9n.

A double-nicking strategy was used to knock out endogenous Dyn1 in A549 cells. Two pairs of 20-bp-long Dyn1 sgRNAs (single guide) were designed using the CRISPR design tool, available at www.genome-engineering.org/crispr (41). The guide RNA pairs (+ and − strands) were cloned into a bicistronic expression vector (pX335) containing a human codon-optimized Cas9n and necessary RNA components (Addgene). The top two double-nickase “hits” targeting exon 1 of human Dyn1 were chosen for cloning into pX335 vector. The sgRNA pair (A): 5-TTCCATGCCGCGGTTGCCCATGG-3 and 5-ATCTCATCCCGCTGGTCAACCGG-3 and the sgRNA pair (B): 5-CCGGCTGCCGCTAGCGCTCCCGG-3 and 5-AGCCATGGGCAACCGCGGCATGG-3 were used. The single-guide RNAs (sgRNAs) in the pX335 vector (1 µg each for a sgRNA pair) were mixed with pmaxGFP plasmid (0.2 µg; Lonza) and cotransfected into A549 cells using lipofectamine 2000 transfection reagent (Life Technologies). Forty-eight hours posttransfection, the cells were trypsinized, washed with PBS, and resuspended in PBS containing 2% serum. GFP⁺ cells were single-cell sorted by FACS (The Moody Foundation Flow Cytometry Facility, University of Texas Southwestern Medical Center, Dallas) into a 96-well plate format into RPMI containing 5% FBS. Single clones were expanded and screened for Dyn1 expression by Western blotting using the anti-Dyn1 rabbit monoclonal antibody (EP772Y, Abcam).

Cell Viability and Caspase-3/7 Activity Assays.

Measurement of cell viability/cytotoxicity was performed by using the CCK-8 Counting Kit (Dojindo). Briefly, 1 × 10⁵ cells/mL cells were treated with several concentrations of TRAIL in 96-well plates for 16 h at 37 °C, in the presence or absence of inhibitors as indicated (all inhibitors were preincubated for 30 min), further incubated with WST-8, and further analyzed according to the manufacturer’s instructions. Caspase-3/7 activation was assessed using the Apo-ONE Homogeneous Caspase-3/7 assay (Promega). Cells were treated with several concentrations of TRAIL in 96-well plates for 5 h at 37 °C, in the presence or absence of the indicated inhibitors, and then lysed in homogeneous caspase-3/7 reagent containing caspase substrate. Lysates were incubated at room temperature for 30 min before reading with a fluorometer (Biotek Synergy H1 Hybrid Reader) at 485/530 nm.

Transferrin Receptor and TRAIL–DR Uptake Assays.

TfnR internalization was performed using the anti-TfnR mAb (HTR-D65). TRAIL–DR internalization was assessed by using an N-terminal FLAG-tag TRAIL and a mouse anti-FLAG mAb (Sigma). Cells were grown for 4 h in gelatin-coated 96-well plates at a density of 2.0 × 10⁵ cells/mL, and then incubated with 4 µg/mL of HTR-D65 or with 1 µg/mL of FLAG-Tag TRAIL in PBS⁴⁺ buffer (PBS supplemented with 1 mM MgCl₂, 1 mM CaCl₂, 5 mM glucose, and 0.2% BSA) at 37 °C for the indicated time points. Cells were then immediately cooled to 4 °C to arrest internalization. Following a washing step to remove unbound ligand (3× PBS⁴⁺), the remaining surface-bound ligand was removed from the cells by an acid wash step (5 × 2 min, 0.2 M acetic acid, 0.2 M NaCl, pH 2.5). TfnR and TRAIL receptor cell surface expression were measured in parallel by incubating MDA-MB-231 or A549 cells with the respective ligands at 4 °C for 30 min. Cells were then washed with PBS and fixed in 4% paraformaldehyde (PFA; Electron Microscopy Sciences) in PBS for 20 min and further permeabilized with 0.1% Triton X-100/PBS for 5 min. Internalized HTR-D65 and TRAIL ligand was assessed using a goat anti-mouse HRP-conjugated antibody (Life Technologies) and anti–FLAG-tag HRP-conjugated antibody (Cell Signaling), further developed with OPD (P1536, Sigma-Aldrich), and the reaction was stopped by using 5 M H₂SO₄. The absorbance was read at 490 nm (Biotek Synergy H1 Hybrid Reader). Internalized ligand was expressed as the percentage of the total surface-bound ligand at 4 °C (i.e., without the acid-wash step), measured in parallel. For HTR-D65 and TRAIL internalization experiments using the GSK3β inhibitors CHIR-99021 (10 µM; Sigma), the Akt inhibitor X (10 µM, Calbiochem), calcineurin inhibitors CsA (20 µM; Sigma), and FK506 (20 µM; Sigma) and Ry (40 µM) cells were preincubated in the absence (i.e., control) or presence of the indicated inhibitors for 30 min at 37 °C, followed by incubation with HTR-D65 or FLAG-tag TRAIL in PBS⁴⁺ at 37 °C for the indicated time points, in the absence or in the presence of the respective inhibitors. Internalization assays were then performed as described above. Percentage of TfnR and TRAIL uptake was calculated relative to the initial total surface-bound ligand at 4 °C for all of the assays.

Calcium Imaging.

Calcium imaging was performed using MDA-MB-231 or A549 cells. MDA-MB-231 cells transiently expressing GCaMP6f were treated with 100 ng/mL of TRAIL for 10 min before imaging, or preincubated with Ry (30 min), followed by TRAIL incubation for 10 min and immediately imaged. Image sequences for the indicated conditions were acquired using a 100 × 1.49 NA Apo TIRF objective (Nikon) mounted on a Ti-Eclipse inverted microscope equipped with the Perfect Focus System (Nikon) at 2 s per frame and an exposure time of 100 ms using a pco-edge 5.5 sCMOS camera with 6.5-μm pixel size. Determination of intracellular calcium signals was presented as change in fluorescence normalized to resting fluorescence (ΔF/F₀), where F₀ represents basal fluorescence measured before the onset of calcium signaling and ΔF represents the change in fluorescence (F − F₀). A549 cells, preloaded with Fura-2, were treated with bradykinin (300 nM) to trigger IP3R activation in the presence or absence of XesC (20 µM). Intracellular Ca²⁺ release as measures as changes in fluorescence (excitation 380 nm, emission 510 nm) over time.