Apoptosis of adherent cells by recruitment of caspase-8 to unligated integrins

Abstract

Integrin-mediated adhesion promotes cell survival in vitro, whereas integrin antagonists induce apoptosis of adherent cells in vivo. Here, we demonstrate that cells adherent within a three-dimensional extracellular matrix undergo apoptosis due to expression of unligated integrins, the beta subunit cytoplasmic domain, or its membrane proximal sequence KLLITIHDRKEF. Integrin-mediated death requires initiator, but not stress, caspase activity and is distinct from anoikis, which is caused by the loss of adhesion per se. Surprisingly, unligated integrin or beta integrin tails recruit caspase-8 to the membrane, where it becomes activated in a death receptor-independent manner. Integrin ligation disrupts this integrin-caspase containing complex and increases survival, revealing an unexpected role for integrins in the regulation of apoptosis and tissue remodeling.

Figure 1.

Expression of integrin αvβ3 increases death among cells attached to an ECM that does not ligate αvβ3. (A) T24E-L cells were derived from T24E cells by serial sorting to select a β3-lacking population. The T24E-Lβ3 cells, derived from T24E-L, were genetically reconstituted for αvβ3 expression. The relative αvβ3 integrin expression in these lines is shown by flow cytometry using monoclonal antibody LM609 (histograms), and by immunoblotting for the β3 integrin subunit with monoclonal antibody AP3 (inset). (B) Survival of these cell variants during culture in a collagen gel in the presence of serum (left) was determined. Death was scored by apoptotic morphology (condensation, satellite array; Cho and Klemke, 2000). Cell survival shown represents the mean ± SE of triplicate wells from a representative experiment. The viability of T24E, T24E-L, and T24E-Lβ3 cells were also assessed following serum deprivation (right). Viability was determined by immunofluorescent, microscopic assessment of PI exclusion at progressive time points. Data shown are from a representative experiment, and represent the mean ± SE viability of triplicate wells. (C) The stable expression of integrin αvβ3 in the CS1β3 cell line has been previously described (Filardo et al., 1995). CS1 parental cells (lacking αvβ3) or CS1β3 cells were grown as adherent to tissue culture (control CS1β3, black histogram), collagen cultured (gray histograms), or held in suspension (white histograms). After 20 h in complete media, cells were assessed for apoptosis by annexin-V staining and FACS^® analysis. A representative experiment, of two, is shown.

Figure 2.

Decreasing the expression of endogenous integrin αvβ3 reduces endothelial cell apoptosis during collagen culture. (A) Endothelial cells undergo apoptosis when cultured in a three-dimensional collagen matrix in the presence of complete media. HUVECs were labeled with Cell Tracker green (2 μM) and cultured in a three-dimensional collagen gel, fibrin gel, or in suspension (filled bars) or, alternatively, on the surface of collagen or fibrin gels (open bars) for 24 h. ECM-associated cells were scored for apoptosis by the morphological analysis of digital images. A representative experiment is shown, each bar is the mean survival, ± SE, of six low power fields. (B) Cell spreading was assessed in HUVECs attached to the surface of collagen (Col) or fibrin (Fb) gels (right) after capture of digital images of Cell Tracker–labeled cells (left). Cells were segmented and area calculated using IP Lab software. Data shown for each point is the mean ± SE cell area calculated at each time point from 10 low power fields. (C) The effect of treatment with a human-specific β3 integrin antisense gene, delivered by adenovirus (AdASβ3), or that of a control, nonsense adenovirus (AdNS), upon the expression of native αvβ3 in HUVECs was determined using LM609 for FACS^® analysis and AP3 for immunoblotting (inset) as described in Fig. 1 A. (D) AdASβ3-treated (or AdNS-treated) HUVECs were assessed for the percentage of surviving cells after a 24 h culture in a three-dimensional collagen gel (left) or fibrin gels (right) as described above. The results shown are the mean ± SE of two fibrin or four collagen experiments. (E) β3 antisense–treated HUVECs exhibited normal levels of ECM attachment. AdASβ3-treated HUVECs (filled bars) were assessed for adhesion to wells coated with ECM components vitronectin (VN), fibronectin (FN), laminin (LN), and collagen (COL). Adhesion was quantitated by staining of attached cells with crystal violet, followed by washing, reextraction, and quantitation of the bound dye (Filardo et al., 1995), and expressed relative to untreated HUVECs. AdNS-treated HUVECs (open bars) are shown as controls. The mean ± SE of two independent experiments is shown. (F) HUVEC survival was examined in collagen (left) or fibrin (right) gels, as described above, in the presence of 30 μg/ml monoclonal antibody P4C10 (anti-β1 integrin), monoclonal antibody LM609 (anti-αvβ3 integrin), both antibodies, or control monoclonal antibody 7G7B6 (anti-CD25). Survival shown is significantly decreased at 24 h (P < 0.05) in both ECMs in the presence of P4C10 or both antibodies. One of two similar experiments is shown.

Figure 3.

The integrin cytoplasmic domain is sufficient to induce apoptosis. (A) Chimeric constructs composed of the cytoplasmic domains of integrins α5, β1, or β3 and the extracellular and transmembrane regions of CD25 (Tac) were expressed in COS7 cells, and viability of cells positive for Tac expression (FITC-7G7B6 positive) was determined via PI exclusion 36 h later. (B) Analysis of the effect of increasing expression of Tac-β3 or control Tac-α5 on cell viability was performed. As shown, ∼25–35% of all cells expressed elevated Tac (top, MED and HI populations). To quantitate death, transfected cells were separated based on mean fluorescence intensity (LO, MED, and HI; top). The viability of these populations was determined by PI exclusion (bottom). Approximately 30–50% of the total Tac-β3–expressing cells (MED and HI) die during the course of the assay, or ∼8–16% of the total COS7 cell population. (C) Tac-β3–expressing COS7 cells exhibited classic apoptotic markers. 18 h after transfection, nonadherent cells were removed and discarded; only those remaining attached were assessed for the onset of apoptosis by staining with annexin-V–FITC (top). Each bar represents the mean percentage of annexin-positive cells (± SE) from three independent experiments. The cleavage of the executioner caspase substrate PARP was assessed by Western blotting of total cell lysates 36 h after transfection. Caspase-cleaved PARP was detected as the 85-kD fragment that is characteristic of apoptosis (bottom). The relative intensity of the cleaved fragment was quantitated as a ratio relative to the uncleaved PARP, including PARP from nonexpressing cells, to give a relative indication of the activation of executioner caspases. The derived PARP ratios shown are: lipofectamine control, 0.004; Tac-β3, 0.17. (D) Caspase activity, as measured by the relative cleavage of PARP, was determined as a dose-dependent quantity of pCI-NeoTac transfection. COS7 cells were transfected with cDNA-encoding integrin tail constructs. Approximately 25–40% of cells expressed Tac, and comparable expression was observed between constructs at each dose transfected. The effect of expressing Tac-β1, Tac-β3, and the control, Tac-α5, as well as a chimera expressing the cytoplasmic domain of β5 (Tac-β5) was determined as the mean ± SE of caspase activity (measured as the intensity of p85/p115 from three independent experiments, as described above).

Figure 3.

The integrin cytoplasmic domain is sufficient to induce apoptosis. (A) Chimeric constructs composed of the cytoplasmic domains of integrins α5, β1, or β3 and the extracellular and transmembrane regions of CD25 (Tac) were expressed in COS7 cells, and viability of cells positive for Tac expression (FITC-7G7B6 positive) was determined via PI exclusion 36 h later. (B) Analysis of the effect of increasing expression of Tac-β3 or control Tac-α5 on cell viability was performed. As shown, ∼25–35% of all cells expressed elevated Tac (top, MED and HI populations). To quantitate death, transfected cells were separated based on mean fluorescence intensity (LO, MED, and HI; top). The viability of these populations was determined by PI exclusion (bottom). Approximately 30–50% of the total Tac-β3–expressing cells (MED and HI) die during the course of the assay, or ∼8–16% of the total COS7 cell population. (C) Tac-β3–expressing COS7 cells exhibited classic apoptotic markers. 18 h after transfection, nonadherent cells were removed and discarded; only those remaining attached were assessed for the onset of apoptosis by staining with annexin-V–FITC (top). Each bar represents the mean percentage of annexin-positive cells (± SE) from three independent experiments. The cleavage of the executioner caspase substrate PARP was assessed by Western blotting of total cell lysates 36 h after transfection. Caspase-cleaved PARP was detected as the 85-kD fragment that is characteristic of apoptosis (bottom). The relative intensity of the cleaved fragment was quantitated as a ratio relative to the uncleaved PARP, including PARP from nonexpressing cells, to give a relative indication of the activation of executioner caspases. The derived PARP ratios shown are: lipofectamine control, 0.004; Tac-β3, 0.17. (D) Caspase activity, as measured by the relative cleavage of PARP, was determined as a dose-dependent quantity of pCI-NeoTac transfection. COS7 cells were transfected with cDNA-encoding integrin tail constructs. Approximately 25–40% of cells expressed Tac, and comparable expression was observed between constructs at each dose transfected. The effect of expressing Tac-β1, Tac-β3, and the control, Tac-α5, as well as a chimera expressing the cytoplasmic domain of β5 (Tac-β5) was determined as the mean ± SE of caspase activity (measured as the intensity of p85/p115 from three independent experiments, as described above).

Figure 4.

IMD requires a membrane localized region of the integrin cytoplasmic domain. (A) A series of mutant integrin chimeras with truncations of the native integrin β3 cytoplasmic domain (residues 715–762) and the extracellular domain of Tac were constructed in pCI-Neo and expressed in COS7 cells. (B) Expression of integrin tail mutant constructs leads to cell death. COS7 cells expressing the integrin constructs were assessed for viability based on PI exclusion by flow cytometry. Within each experiment, the relative capacity of each construct to induce death was normalized to the death induced by the full length tail (which ranged from 20 to 40% of expressing cells in these studies). (C) Executioner caspase activity is induced by integrin truncation mutants. For each mutant, the relative caspase activation (assessed by the p85/p115 ratio) was assessed and normalized to that induced by intact Tac-β3. The results shown are the mean ± SD of the normalized PARP ratios for three experiments. (D) Mutation of the NPXY motif within the tail of integrin β3 blocks αvβ3-directed cell migration, but does not inhibit IMD. T24-EL cells reconstituted to express wild type β3 (T24-ELβ3) or a point mutant in the NPXY motif (T24-ELN744A; shown in bold, Fig. 4 A) at similar levels were assessed for their migration on vitronectin (left) or for survival within a collagen ECM (right). Migration was measured microscopically as the average wound closure plotted, as the mean ± SE of six sites on a vitronectin-coated plate after 24 h. Collagen survival was assessed microscopically by direct cell counting to assess exclusion of PI (five fields per gel, three gels). The mean ± SE of three experiments are shown. (E) Membrane-anchored forms of the integrin cytosolic domain induce apoptosis. COS7 cells were transfected with three different β3 cytoplasmic domain constructs; Tac-β3, as well as a second membrane-anchored, GFP-β3, and an unanchored, His-β3, variant of the β3 cytoplasmic domain. Detergent lysates of these cells were subjected to immunoblotting to assess PARP cleavage (top) and to confirm expression of the individual constructs (GFP, top; Tac/His, bottom). One of two similar experiments is shown. (F) Cell spreading was assessed via analysis of digital images of cells expressing GFP-β3 or GFP-TM (control, GFP with a signal sequence and the transmembrane domain from CD25) as described above (Fig. 2 B) 12, 24, and 48 h after transfection. (G) Ligation of β3 constructs suppresses IMD. COS7 cells expressing submaximal levels of integrin tails were allowed to attach to substrates coated with anti-Tac monoclonal antibody (Ligation +) or maintained as a normal, adherent tissue culture (Ligation −). Adherent cells were lysed and assessed for the relative expression of death-inducing construct, i.e., integrin, present, as well as to determine executioner caspase activation (through quantitation of p85/p115 ratios) via immunoblot analysis. Bars represent the mean ± SE of three experiments.

Figure 5.

IMD depends on initiator caspase activity. (A) Executioner caspase activity is blocked by crmA but not by inhibitors of stress or Fas-mediated apoptosis. Coexpression of crmA, DN-FADD, Bcl_xl, or catalytically inactive caspase-9, DN Casp 9 (4 μg) with the proapoptotic Tac-β3 construct (2 μg) was performed to determine whether these checkpoint-specific apoptosis inhibitors could suppress IMD. Cells were assessed after 36 h using the activation of executioner caspase activity (PARP cleavage immunoblot assay) as an indicator. Data are expressed as a ratio of PARP cleavage found in (Tac-β3 + inhibitor) to that observed in (Tac-β3 + control vector) transfected cells, and represents the normalized results of three to four independent experiments. To confirm that constructs with no activity in IMD were functional, apoptosis was also induced via Fas overexpression (2 μg; hatched bars), or Bad expression (1 μg; gray bars). (B) Cell viability is maintained by crmA. Cells were allowed to express constructs (as per A, above), but in this case apoptosis was quantified by FACS^®. Cells harvested 36 h after transfection were labeled with Alexa dye–conjugated anti-CD25 (7G7B6-Alexa488) to label Tac-expressing cells. Dead cells were identified by uptake of PI (500 ng/ml). Results are expressed as the percentage of dead, Tac-expressing cells in each group, and represent the mean ± SE of two experiments. (C) Peptides that inhibit stress-induced apoptosis are not active in IMD. COS7 cells, induced to die via expression of Tac-β3, were treated with 40 μM of caspase-inhibitory peptides (zIETDfmk, zDEVDfmk, or zLEHDfmk) as a single dose 12 h after transfection, and then monitored for executioner caspase activation, i.e., PARP cleavage, 24 h later. Data are expressed relative to cells treated with diluent (DMSO) alone and represent the mean ± SD of two experiments. (D) IMD induced by endogenous integrins is blocked by crmA. COS7 cells coexpressing Ds-RED fluorescent protein, a transfection marker, and checkpoint- specific inhibitors of apoptosis, as listed above, were assessed for survival after a 48 h culture on collagen gels in complete growth medium via morphologic scoring, as described (Fig. 2). Data shown is the mean survival ± SE. A representative experiment of two is shown.

Figure 6.

Caspase-8 is recruited to the membrane and becomes activated during IMD. (A) Integrin tails colocalize with caspase activity during IMD. The cellular location of death-inducing integrin tails and active caspases was assessed by immunofluorescence using a 1024 Bio-Rad confocal microscope. Cells were induced to undergo apoptosis by integrin cytoplasmic domain expression, or were rescued by ligation (as described in Fig. 4 G). Before fixation in PBS/4% paraformaldehyde, cells were labeled with FAM-VAD in serum free media for 60 min to label active caspases (fluorescein, green channel). Cells were then probed for Tac expression (monoclonal antibody 7G7B6, red channel). Signals for each image were gathered serially as independent scans using the 488 nm (middle row) and 566 nm (top row) laser lines. Colocalization within the z-sections is shown as a yellow/orange signal (bottom row). (B) Confocal images of COS7 cells expressing the Tac–integrin chimeras described in A were acquired by blinded observers. The coefficient of colocalization between the 488 and 566 nm signal within the integrin-containing z-sections was determined (Bio-Rad Lasersharp Software). The mean ± SE of 10 random fields for each group are plotted. (C) Caspase-8 and -3 are labeled during IMD. COS7 cells induced to die by the expression of integrin tail were incubated with biotin-VAD to label active caspase. Active caspases were then isolated with avidin-Sepharose and resolved via SDS-PAGE on 8–16% nonreducing gradient gels. Immunoblotting was performed with polyclonal rabbit antisera to caspases-3 (lane 2) or -6 (lane 4) (control and not detected, respectively) or monoclonal antisera to caspase-8 (lane 3). (D) Caspase-8–like activity is elevated during IMD. Lysates from COS7 cells expressing Tac constructs with either the integrin α5 cytoplasmic domain (gray circles), the integrin β3 cytoplasmic domain (open circles), or mock transfectants (open squares) were assessed for their relative ability to cleave the caspase-8 substrate, IETD-pNA. The induced IETDase activity in the Tac-β3 expressing cells was suppressed by substrate ligation of the chimera (as described in Fig. 2 G), as shown (black circle). (E) The subcellular distribution and predominant forms of caspase-8 change in cells undergoing IMD. Cell fractionation of COS7 cells expressing GFP-β3 (induces IMD) or GFP-β3^Δ717 (inactive truncation) was performed 36 h after transfection with the individual constructs (4 μg). The relative distribution and forms of caspase-8 were assessed by immunoblot analysis of the cytosolic (Csol), membrane (Mem), and cytoskeletal/nuclear (Cskn) fractions. Relative molecular mass (kD) of the detected species is indicated. Bar, 10 μm.

Figure 6.

Caspase-8 is recruited to the membrane and becomes activated during IMD. (A) Integrin tails colocalize with caspase activity during IMD. The cellular location of death-inducing integrin tails and active caspases was assessed by immunofluorescence using a 1024 Bio-Rad confocal microscope. Cells were induced to undergo apoptosis by integrin cytoplasmic domain expression, or were rescued by ligation (as described in Fig. 4 G). Before fixation in PBS/4% paraformaldehyde, cells were labeled with FAM-VAD in serum free media for 60 min to label active caspases (fluorescein, green channel). Cells were then probed for Tac expression (monoclonal antibody 7G7B6, red channel). Signals for each image were gathered serially as independent scans using the 488 nm (middle row) and 566 nm (top row) laser lines. Colocalization within the z-sections is shown as a yellow/orange signal (bottom row). (B) Confocal images of COS7 cells expressing the Tac–integrin chimeras described in A were acquired by blinded observers. The coefficient of colocalization between the 488 and 566 nm signal within the integrin-containing z-sections was determined (Bio-Rad Lasersharp Software). The mean ± SE of 10 random fields for each group are plotted. (C) Caspase-8 and -3 are labeled during IMD. COS7 cells induced to die by the expression of integrin tail were incubated with biotin-VAD to label active caspase. Active caspases were then isolated with avidin-Sepharose and resolved via SDS-PAGE on 8–16% nonreducing gradient gels. Immunoblotting was performed with polyclonal rabbit antisera to caspases-3 (lane 2) or -6 (lane 4) (control and not detected, respectively) or monoclonal antisera to caspase-8 (lane 3). (D) Caspase-8–like activity is elevated during IMD. Lysates from COS7 cells expressing Tac constructs with either the integrin α5 cytoplasmic domain (gray circles), the integrin β3 cytoplasmic domain (open circles), or mock transfectants (open squares) were assessed for their relative ability to cleave the caspase-8 substrate, IETD-pNA. The induced IETDase activity in the Tac-β3 expressing cells was suppressed by substrate ligation of the chimera (as described in Fig. 2 G), as shown (black circle). (E) The subcellular distribution and predominant forms of caspase-8 change in cells undergoing IMD. Cell fractionation of COS7 cells expressing GFP-β3 (induces IMD) or GFP-β3^Δ717 (inactive truncation) was performed 36 h after transfection with the individual constructs (4 μg). The relative distribution and forms of caspase-8 were assessed by immunoblot analysis of the cytosolic (Csol), membrane (Mem), and cytoskeletal/nuclear (Cskn) fractions. Relative molecular mass (kD) of the detected species is indicated. Bar, 10 μm.

Figure 7.

Caspase-8 associates with integrins during IMD. (A) Caspase-8 associates with native integrin αvβ3. Integrin αvβ3 and LRP complexes are expressed at similar levels on COS7 cells (unpublished data) and were subjected to clustering and immunoprecipitation followed by immunoblot analysis for caspase-8 to determine if caspase-8 was recruited to an integrin DISC-like complex (Scaffidi et al., 1999). (B) Integrin cytosolic domains are sufficient to recruit caspase-8. Expressed GFP-β3 (proapoptotic) or GFP-β3^Δ717 (inactive) constructs were subjected to immunoprecipitation with GFP-Sepharose followed by immunoblot analysis for the presence of associated caspase-8 (top) or the GFP chimera (bottom). (C) Ligation of integrin disrupts caspase association. Integrin-expressing COS7 cells (Tac-β1, unligated Tac-β3, Tac-α5) and Tac-β3–expressing cells replated onto anti-Tac substrate (ligated Tac-β3) (as described in Fig. 4 F) were lysed and subjected to immunoprecipitation with anti-Tac monoclonal antibody. The precipitates were resolved via SDS-PAGE and analyzed for the presence of associated caspase-8 (top) or Tac (bottom) via immunoblotting. (D) An αvβ3 integrin ligand rescues apoptosis and interrupts caspase-8 recruitment to integrins in a three-dimensional ECM. HUVECs were cultured in three-dimensional fibrin or collagen gels, or maintained in suspension (apoptosis positive control) for 16 h. Endothelial cells were then recovered from the gels, lysed, and subjected to immunoprecipitation using LM609 (anti-αvβ3). The precipitates were resolved and analyzed for the presence of caspase-8 by immunoblotting, as described above.

Figure 7.

Caspase-8 associates with integrins during IMD. (A) Caspase-8 associates with native integrin αvβ3. Integrin αvβ3 and LRP complexes are expressed at similar levels on COS7 cells (unpublished data) and were subjected to clustering and immunoprecipitation followed by immunoblot analysis for caspase-8 to determine if caspase-8 was recruited to an integrin DISC-like complex (Scaffidi et al., 1999). (B) Integrin cytosolic domains are sufficient to recruit caspase-8. Expressed GFP-β3 (proapoptotic) or GFP-β3^Δ717 (inactive) constructs were subjected to immunoprecipitation with GFP-Sepharose followed by immunoblot analysis for the presence of associated caspase-8 (top) or the GFP chimera (bottom). (C) Ligation of integrin disrupts caspase association. Integrin-expressing COS7 cells (Tac-β1, unligated Tac-β3, Tac-α5) and Tac-β3–expressing cells replated onto anti-Tac substrate (ligated Tac-β3) (as described in Fig. 4 F) were lysed and subjected to immunoprecipitation with anti-Tac monoclonal antibody. The precipitates were resolved via SDS-PAGE and analyzed for the presence of associated caspase-8 (top) or Tac (bottom) via immunoblotting. (D) An αvβ3 integrin ligand rescues apoptosis and interrupts caspase-8 recruitment to integrins in a three-dimensional ECM. HUVECs were cultured in three-dimensional fibrin or collagen gels, or maintained in suspension (apoptosis positive control) for 16 h. Endothelial cells were then recovered from the gels, lysed, and subjected to immunoprecipitation using LM609 (anti-αvβ3). The precipitates were resolved and analyzed for the presence of caspase-8 by immunoblotting, as described above.