Real Time Analysis of Tumor Necrosis Factor-related Apoptosis-inducing Ligand/Cycloheximide-induced Caspase Activities during Apoptosis Initiation*

Employing fluorescence resonance energy transfer (FRET) imaging, we previously demonstrated that effector caspase activation is often an all-or-none response independent of drug choice or dose administered. We here investigated the signaling dynamics during apoptosis initiation via the tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) receptor pathway to investigate how variability in drug exposure can be translated into largely kinetically invariant cell death execution pathways. FRET-based microscopy demonstrated dose-dependent responses of caspase-8 activation and activity within individual living HeLa cells. Caspase-8 on average was activated 45-600 min after TRAIL/cycloheximide addition. Caspase-8-like activities persisted for 15-60 min before eventually inducing mitochondrial outer membrane permeabilization. Independent of the TRAIL concentrations used or the resulting caspase-8-like activities, mitochondrial outer membrane permeabilization was induced when 10% of the FRET substrate was cleaved. In contrast, in Bid-depleted cells, caspase-8-like activity persisted for hours without causing immediate cell death. Our findings provide detailed insight into the intracellular signaling kinetics during apoptosis initiation and describe a threshold mechanism controlling the induction of apoptosis execution.

Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) 2 is a potent cytotoxic ligand inducing apoptosis preferentially in tumor cells (1). New TRAIL-based treatment regimes for adjuvant chemotherapies therefore are currently being studied in phase I and II clinical trials (2). TRAIL binding to its cognate death receptors TRAIL-R1 and -R2 induces receptor trimerization. At their cytoplasmic domains, TRAIL-R1 and -R2 recruit the adaptor protein Fas-associated death domain into the so-called death inducing signaling complex (DISC). Via interaction of their death effector domains, Fas-associated death domain recruits procaspase-8 and -10 to the DISC, resulting in activation and processing of these initiator proteases (3). Although in some cell lines caspase-8/-10 can directly activate effector caspase-3 (type I signaling) (4), the majority of cells require caspase-8/-10 to initiate apoptosis by cleaving the BH-3-only protein Bid (type II signaling). Truncated Bid (tBid) then translocates to mitochondria and induces Bax/Bak-dependent mitochondrial outer membrane permeabilization (MOMP) (5)(6)(7)(8). MOMP results in the release of mitochondrial intermembrane space proteins, such as cytochrome c and Smac, from the mitochondria into the cytosol, mitochondrial depolarization, and subsequent apoptosis execution by effector caspases, such as caspase-3, -7, and -6 (9 -11).
Independent of the choice of stimulus or the dose applied, the induction of MOMP and the subsequent execution of apoptosis by effector caspases were shown to be kinetically invariant all-or-none signaling processes that guarantee cell death over a wide range of key protein concentrations (9,12,13). In the case of effector caspase activation, this switchlike response was shown to emanate as a systems property from feedback signaling loops in the caspase activation network (13,14). In contrast, little is known about the intracellular signaling dynamics during apoptosis initiation leading to the kinetically invariant activation of MOMP. Especially, since it is conceivable that at both physiological and therapeutic conditions, cells can be exposed to TRAIL receptor ligands over a wide range of different concentrations, it is unclear how this variability can be translated into a clear cell death decision.
Here, we address this question by using a fluorescence resonance energy transfer (FRET)-based approach to monitor the real time kinetics of caspase-8/-10 activation and activity via the TRAIL receptor pathway using single cell time lapse imaging. We found that TRAIL exposure induced dose-dependent kinetics of caspase-8/-10 activation and activity. We furthermore identified a threshold mechanism that permits us to integrate such differential caspase activities into an all-or-none decision of apoptosis execution. Importantly, we performed these analyses excluding the potential contribution of high abundant downstream effector caspases with similar substrate specificities.
Molecular Cloning of the IETD FRET Probe-A DEVDaseresponsive enhanced CFP-DEVD-Venus FRET cassette was obtained from pSCAT3 (15) (a generous gift of Masayuki Miura, RIKEN Brain Research Institute, Wako, Saitama, Japan) by digestion with BamHI and HindIII. This was subcloned into pTK-RL (Promega, WI), digested at the same sites, and treated with shrimp alkaline phosphatase (U. S. Biologicals/Amersham Biosciences). The resulting plasmid was digested with SacI and KpnI to remove the DEVDase substrate cassette and ligated with excess annealed oligonucleotides, forming an IETDase substrate cassette. This IETD cassette was generated using the following complementary oligonucleotides 5Ј-G AGC GGA ATC GAG ACC GAT GGTAC-3Ј and 5Ј-C ATC GGT CTC GAT TCC GCT CAGCT-3Ј. This generated a cassette consisting of a coding sequence for IETD flanked NH 2 -terminally by a flexible glycine-serine dipeptide and COOH-terminally by a KpnI site (encoding glycine-threonine) and a flexible serineglycine-serine tripeptide. The oligonucleotide sequences immediately adjacent to the overhangs were designed to destroy the SacI site upon ligation with corresponding vector overhangs and to generate a diagnostic PvuII site to exclude multiple concatenated oligonucleotides. The resulting pTK-reverse SCAT8 vector was digested with BamHI and HindIII to obtain the enhanced CFP-IETD-Venus FRET cassette. This was ligated into pcDNA3.1 vector to generate pSCAT8.
Cell Culture and Transfection-All cells were cultured in RPMI 1640 medium supplemented with penicillin (100 g/ml), streptomycin (100 g/ml), and 10% fetal calf serum (Sigma). Cells were transfected with 500 ng of pSCAT8 plasmid DNA, and 4 l of Metafecten (Biontex Laboratories, Munich, Germany) per milliliter of serum-free medium at 37°C for 4 h. For the generation of stable cell lines, cells were selected in the presence of 50 g/ml G418 (Invitrogen) for 3-4 weeks, and fluorescent clones were enriched.
Generation of Stable Bid-depleted Cells-Three different short hairpin RNA sequences specific for human Bid mRNA were designed using the Dharmacon siRNA design tool (available on the World Wide Web). The following 19-nucleotide sequences were chosen and subjected to BLAST analysis to avoid significant homology to other human genes: Bid-1 sense (5Ј-AAGCTGTTCTGACAACAGC-3Ј), corresponding to nucleotides 78 -97 downstream of the Bid mRNA start codon; Bid-2 sense (5Ј-AAGGAGAAGACCATGCTGG-3Ј), homologous to nucleotides 430 -449; and Bid-3 sense (5Ј-AAGAATA-GAGGCAGATTCT-3Ј, spanning nucleotides 210 -229. The Bid-specific short hairpin RNA duplexes along with a scram-bled control sequence were ligated into the pSilencer 2.1-U6 hygro vector (Ambion, Cambridgeshire, UK) via their BamHI and HindIII sites. To generate stable knockdown cell lines, HeLa cells were transfected with the different short hairpin RNA constructs using Metafectene (Biontex, Munich, Germany) according to the manufacturer's instructions. 24 h posttransfection, the cells were serially diluted and transferred to 96-well plates, and stable clones were selected using hygromycin B (160 g/ml). Bid expression in HeLa cells stably expressing Bid siRNA was compared with the Bid expression in parental HeLa cells by Western blotting, and the clone with the strongest Bid depletion was selected for this study. After background subtraction, chemiluminescence intensities from n ϭ 3 independent whole cell protein extracts were densitometrically measured using AlphaEase FC software (Alpha Innotech, San Leandro, CA).
Adenoviral Infection-The generation of adenoviruses for XIAP (X-linked inhibitor of apoptosis protein) expression and control viruses and the infection procedure were described previously (13,16). In brief, parental and Bid-depleted HeLa cells expressing the SCAT8 FRET probe were grown on WillCo dishes (Willco BV, Amsterdam, The Netherlands). Cells were washed twice with phosphate-buffered saline and infected at a multiplicity of infection of 1000 in serum-free medium for 2 h and subsequently cultured in full growth medium. Experiments were carried out Ͼ24 h postinfection.
Preparation of Whole Cell Extracts and Western Blotting-Cells were collected at 1000 rpm for 3 min and washed with phosphate-buffered saline. The cell pellet was resuspended in lysis buffer (62.5 mM Tris-HCl, pH 6.8, 10% (v/v) glycerin, 2% (w/v) SDS, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 1 g/ml leupeptin, and 5 g/ml aprotinin) and heated at 95°C for 20 min. Protein content was determined with the Pierce Micro-BCA protein assay (Pierce). An equal amount of protein (20 g) was loaded onto SDS-polyacrylamide gels. Proteins were separated at 100 V for 2.5 h and then blotted to nitrocellulose membranes (Protean BA 83; 2 m; Schleicher & Schuell) in transfer buffer (25 mM Tris, 192 mM glycine, 20% methanol (v/v), and 0.01% SDS) at 18 V for 60 min. The blots were blocked with 5% nonfat dry milk in TBST (15 mM Tris-HCl, pH 7.5, 200 mM NaCl, and 0.1% Tween 20) at room temperature for 1 h. Membranes were incubated with the following antibodies: a rabbit polyclonal caspase-3 antibody (Cell Signaling Technology, Danvers, MA), a mouse monoclonal caspase-8 antibody (Alexis Biochemicals, San Diego, CA), a mouse monoclonal green fluorescent protein antibody (Clontech), a goat polyclonal Bid antibody (R&D Systems, Abingdon, UK), a mouse monoclonal ␣-tubulin antibody (Sigma), and a mouse monoclonal ␤-actin antibody (Sigma). Membranes were washed with TBST three times for 5 min and incubated with anti-mouse or anti-rabbit peroxidase-conjugated secondary antibodies (Jackson Laboratories) for 1 h. Blots were washed and developed using the enhanced chemiluminescence detection reagent (Amersham Biosciences). Chemiluminescence was detected at 12-bit dynamic range using a Fuji LAS 3000 CCD system (Fujifilm UK Ltd., Bedfordshire, UK).
Determination of Caspase-3-like Protease Activity-After exposure to staurosporine, TRAIL/CHX, or vehicle, culture medium was aspirated, and cells were lysed in 200 l of lysis buffer (10 mM HEPES, pH 7.4, 42 mM KCl, 5 mM MgCl 2 , 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5% CHAPS, and 1% protease inhibitor mixture (Sigma)). Fifty microliters of this extract were added to 150 l of reaction buffer (25 mM HEPES, pH 7.5, 1 mM EDTA, 0.1% CHAPS, 10% sucrose, 3 mM dithiothreitol). The reaction buffer was supplemented with 10 M Ac-DEVD-AMC, the preferred fluorigenic substrate for effector caspase-3 and -7. Production of fluorescent AMC was monitored over 120 min using a fluorescent plate reader (excitation 360 nm, emission 465 nm). Autofluorescence of blanks containing lysis buffer only were subtracted. Protein content was determined using the Pierce Coomassie (Bradford) protein assay kit (Pierce). Caspase activity was expressed as change in fluorescent units/h/g of protein.
Confocal Microscopy and Acceptor Bleaching-To confirm resonance energy transfer within the IETD FRET probe, the acceptor Venus was bleached, and the donor dequenching was analyzed. HeLa cells expressing the FRET probe were placed on the heated stage of an LSM 510 Meta confocal microscope equipped with a Plan-Apochromat ϫ63 numerical aperture 1.4 oil immersion differential interference contrast objective (Carl Zeiss, Jena, Germany). The emission spectra were recorded at 10.3-nm step size using the 405-nm laser line (attenuated to 1.0%), and the 405/514 multichroic beamsplitter and the Meta detector in the range of 442-614 nm. Following each cycle of bleach scans with the nonattenuated 514 argon laser line (laser was run at 50% of its maximal power), spectral images were recorded.
Mitochondrial Membrane Potential (⌬⌿ M ) and FRET Disruption-After background subtraction, the cellular TMRM fluorescence intensity was calculated for each cell. Caspase cleavage kinetics were detected at the single-cell level by FRET analysis, as described previously (12). Images were processed using MetaMorph 7.1r1 software (Molecular Devices Ltd., Wokingham, UK). CFP/YFP and FRET/YFP emission ratio traces were obtained by dividing the fluorescence intensity values in the CFP and FRET channels by the YFP emission of single cells after background subtraction. The ratiometric readout automatically corrects for nonspecific changes in fluorescence intensities that affect all channels in parallel, such as slight drifts in optical focus or cellular movements.
Statistics-Data are given as means Ϯ S.D. or S.E. For statistical comparison, Student's t test or analysis of variance and subsequent Tukey's test were used for normal distributed data. Otherwise, Mann-Whitney U test or Kruskal-Wallis H test were used. p values smaller than 0.05 were considered to be statistically significant. Kruskal-Wallis H tests returning statistical significances were subsequently followed up with Bonferroni adjusted Mann-Whitney U tests.

Acceptor Bleaching Confirms Resonance Energy Transfer in a
New IETDase FRET Probe-Previously, recombinant FRET probes based on green fluorescent protein variants were successfully employed to analyze effector caspase activity in single living cells (12,13,15,17). Here, we developed a new FRET probe based on CFP and Venus, an enhanced yellow fluorescent protein, which are interconnected by a short linker containing a preferred caspase-8/-10 recognition site IETD (18,19). Upon CFP excitation, energy is transferred to the acceptor fluorophore Venus (Fig. 1A). Cleavage of the linker disrupts the resonance energy transfer as the distance between donor and acceptor increases and results in enhanced CFP emission (Fig.  1B). The probe was generated from a previously described FRET probe template for effector caspases, which provided a very high signal/noise ratio upon proteolytic cleavage (15). This high signal/noise ratio facilitates the detection of low caspase activities as expected in the case of caspase-8/-10 during apoptosis initiation. Efficient resonance energy transfer was observed when expressing the IETD FRET probe in HeLa cervical cancer cells; upon photobleaching the acceptor fluorophore Venus in individual cells, CFP emission significantly increased, as observed in cellular fluorescence emission profiles ( Fig. 1 Bid Depletion Can Prevent Apoptosis Execution but Not IETD Probe Cleavage following TRAIL Exposure-Biochemically, it has been shown that besides caspase-8 and -10, downstream effector caspase-3 and -6 can cleave IETD recognition sites as well (19). Similar overlapping specificities were reported for synthetic caspase inhibitors, such as IETD-and DEVD-fmk, effectively preventing the selective inhibition of individual caspases (18,19). Using the above described new IETD FRET probe, we therefore investigated whether we could establish a model system that enabled us to temporally separate the downstream execution phase from caspase-8/-10-dependent signaling during the initiation phase of TRAIL-induced apoptosis.
HeLa cells are type II signaling cells and consequently depend on the mitochondrial pathway for the activation of effector caspases during death receptor-induced apoptosis (20). We generated HeLa cells stably depleted of Bid expression by siRNA transfection to impair MOMP and subsequent effector caspase activation ( Fig. 2A). Bid expression was reduced to ϳ4.5 Ϯ 3.0% of the wild type expression level as densitometrically quantified from Western blots.
To investigate whether Bid depletion was sufficient to impair effector caspase activation during death receptor-induced apoptosis, we next biochemically characterized the response of parental and Bid-depleted HeLa cells to TRAIL/CHX (hereafter referred to as TRAIL) and intrinsic apoptosis stimulus STS. TRAIL was administered in combination with 1 g/ml CHX to suppress the activation of translation-dependent survival signaling in response to TRAIL exposure (3). In both parental and Bid-depleted cells, procaspase-8 was processed into active subunits with apparently identical kinetics in response to TRAIL (Fig. 2B). Although parental cells can activate caspase-3, Biddepleted HeLa cells did not show any reduction in procaspase-3, and active subunits were only detected in minute amounts (Fig. 2, B and C). To analyze whether the observed caspase-3 processing results in caspase-3-like activity, we employed fluorigenic DEVD-AMC substrate assays. In these assays, Bid depleted HeLa cells exhibited massively reduced cleavage rates following death receptor stimulation with TRAIL, suggesting that downstream apoptosis execution is significantly inhibited in these cells (Fig. 2D). In contrast, Bid depletion did not influence the activation of caspase-3 or caspase-8 (Fig. 2E), which is largely activated downstream of caspase-3 during STS-induced apoptosis (20); nor was caspase-3 like activity significantly affected in fluorigenic assays (Fig. 2F).
Next, we tested whether the IETD FRET probe can still be cleaved in Bid-depleted cells during TRAIL-induced apoptosis. Following TRAIL exposure, the full-length probe was cleaved into fragments corresponding to the molecular sizes of CFP linker and Venus, as shown by Western blotting (Fig. 2G). The addition of caspase inhibitor benzyloxycarbonyl-VAD-fmk blocked FRET substrate cleavage, showing this to be a caspasedependent process (Fig. 2G).
Taken together, these results show that Bid depletion efficiently inhibited apoptosis execution in response to TRAIL addition. Bid depletion, however, did not impair the processing of procaspase-8; nor did it prevent IETD FRET probe cleavage.
TRAIL-induced Caspase-8-like Activities Can Persist for Hours without Causing Immediate Apoptotic Cell Death in Biddepleted HeLa Cells-To analyze profiles of caspase-8-like activities in single living cells by time lapse imaging, we then expressed the IETD FRET probe in the Bid-depleted HeLa cells. Following administration of TRAIL, we observed that probe cleavage began individually in single cells rather than it being a synchronous response throughout the whole population in the field of view (Fig. 3A). Probe cleavage was detected as an intensity increase in the CFP/YFP emission ratio images (Fig. 3A) and a decrease in FRET/YFP emission ratio images (not shown).
Previous studies in HeLa cells have shown a loss in ⌬⌿ M in response to MOMP-induced cytochrome c release (9,11,21,22). Using TMRM as a ⌬⌿ M -sensitive dye in parallel with the FRET probe, we found that in Bid-depleted cells, mitochondria either did not depolarize (Fig. 3A) or depolarized only at very late times during or after FRET probe cleavage (not shown). Importantly, classical apoptotic morphology, such as cellular shrinkage or membrane blebbing could not be observed as long as the cells did not depolarize. Corresponding to our biochemical analyses (Fig. 2), this indicates that MOMP-dependent mitochondrial depolarization and apoptosis execution are efficiently inhibited by Bid depletion.
We next plotted the CFP/YFP and FRET/YFP emission ratios of individual Bid-depleted cells to graphically follow substrate cleavage over time. Probe cleavage began following an initial lag time, as observed by an increase in the CFP/YFP and a concomitant decrease in the FRET/YFP ratios (Fig. 3B). In response to 100 ng/ml TRAIL, probe cleavage on average began 56 Ϯ 25 min after drug addition and lasted for 5.5 Ϯ 2.8 h (data are mean Ϯ S.D. from n ϭ 20 cells analyzed). To exclude the possibility that type I-like direct activation of caspase-3 by caspase-8/-10 contributed to the activity measured (4), in additional control experiments we employed adenoviral overexpression of XIAP, the most efficient intracellular inhibitor of downstream caspase-3, -7, and -9. XIAP overexpression did not affect IET-Dase activity in Bid-depleted cells in response to TRAIL (supplemental Fig. 1).
These findings demonstrate that HeLa cells individually activate IETDases in response to TRAIL exposure and that this caspase-8-like activity can be a surprisingly stable response that can persist for hours without causing immediate apoptotic cell death.   AUGUST

TRAIL-induced Profiles of Caspase-8-like Activities Upstream of MOMP in Parental HeLa
Cells-We next analyzed caspase-8-like activities in parental HeLa cells following TRAIL exposure. In contrast to Bid-depleted cells, we observed that parental HeLa cells cleaving the FRET probe subsequently underwent cellular condensation and membrane blebbing characteristic of apoptotic cell death (Fig. 4A). Furthermore, in parental cells substrate cleavage was accompanied by MOMP, as shown by pronounced mitochondrial depolarization (Fig. 4A).
Plotting the temporal profiles for CFP/YFP and FRET/YFP ratios for individual cells indicated that after an initial lag time, the IETD FRET probe was cleaved at a low rate until eventually the IETDase activity sharply increased. Comparing the highest slopes during times of low and high rates of substrate cleavage suggested that IETDase activity increased ϳ12-fold (n ϭ 16 cells exposed to 100 ng/ml TRAIL analyzed) (Fig. 4B).
We then more closely examined whether this increase in activity temporally was related to MOMP and mitochondrial depolarization. Indeed, when analyzing the TMRM fluorescence and CFP/YFP ratios for individual cells, we observed that probe cleavage increased very soon after the onset of mitochondrial depolarization (Fig. 4C). On average, the increased rate in substrate cleavage was observed within 4.1 Ϯ 2.2 min after MOMP (mean Ϯ S.D. from n ϭ 23 cells), thus closely resembling times between MOMP and effector caspase activation reported for HeLa cells before (13). All cells that activated IET-Dases proceeded through to MOMP and increased substrate cleavage, followed by cellular shrinkage and apoptotic membrane blebbing.
These results, and the results shown in Fig. 3 and supplemental Fig. 1 thus suggest that during TRAIL-induced apoptosis, low IETDase activities attributable to initiator caspase-8/-10 can be observed upstream of MOMP and that IETDase activities downstream of MOMP dramatically increase within living cells. This increase in activity may be largely subject to activation of effector caspases during apoptosis execution.
No Detectable IETDase Activities Upstream of MOMP during Activation of the Intrinsic Apoptosis Pathway by STS-To further analyze whether the post-MOMP increase in IETDase activity indeed was attributable to engagement of the apoptosis execution phase, we performed additional analyses in parental and Bid-depleted HeLa cells treated with intrinsic death stimulus STS (Fig. 5). It was previously reported that the activation of effector caspases following MOMP is largely independent of the concentration and type of stimulus used to induce apoptosis (11,12), so that we expected largely identical IETDase activities downstream of MOMP during TRAIL-and STS-induced apoptosis.
In response to STS, mitochondrial depolarization preceded the onset of IETDase activity in HeLa cells, and the slopes of FRET substrate cleavage downstream of MOMP closely resembled those observed downstream of MOMP during TRAIL-induced apoptosis (Figs. 4C and 5, A and B). Similar results were observed in Bid-depleted HeLa cells upon STS exposure (Fig. 5,  C and D). No cells were observed showing IETDase activity upstream of MOMP. The presence or absence of Bid apparently did not affect the caspase activity during apoptosis execution, since the time required for complete substrate cleavage in response to STS was largely identical in both groups (Fig. 5E). Similar results were obtained with a DEVD FRET probe optimized for effector caspase-3 and -7 activities (supplemental Fig. 2).

IETDase Activities Upstream of MOMP Show Dose-dependent Differences in Onset of Substrate Cleavage and Caspase
Activity-We next analyzed in more detail how extracellular variability in TRAIL stimulation manifests within individual cells. To this end, we first investigated the apoptosis initiation phase in response to TRAIL concentrations ranging from 10 to 1000 ng/ml. For all cells measured, we determined the following parameters from the temporal IETDase profiles (Fig. 6A): (i) the time required from TRAIL addition to caspase-8 activation, which biologically comprises the processes from TRAIL binding to its respective receptors, receptor trimerization, and DISC formation; (ii) the time from caspase-8 activation until mitochondrial depolarization, indicating how long caspase-8-like activity needs to persist to induce MOMP.
We found that the lag time between stimulus addition and caspase-8/-10 activation was indeed a dose-dependent response that was kinetically saturated at high TRAIL concen- trations (Fig. 7A). At high doses, the caspase-8/-10 response was observed on average 45 min following drug addition. In contrast, at submaximal TRAIL concentrations, the caspase-8/-10 response was delayed by ϳ10 h and became increasingly asynchronous between individual cells (Fig. 7B). Quantifying the duration of caspase-8/-10 activity until MOMP indicated that higher stimulus concentrations resulted in an accelerated MOMP response, indicating that higher doses of TRAIL result in higher caspase-8/-10 activities (Fig. 6C).
Although at submaximal doses, caspase-8/-10 activity persisted for ϳ50 min, in response to high TRAIL doses ϳ15 min of caspase-8/-10 activity was sufficient to induce MOMP.

MOMP Is Induced at 10% IETD FRET Substrate Cleavage in Response to High and Low TRAIL Concentrations, Regardless of the Large Variability in Onset of Substrate Cleavage and
Caspase Activity-To provide an explanation how these differences in IETDase activities upstream of MOMP are translated into an all-ornone-response at the level of MOMP, we quantified additional parameters for all cells analyzed; we determined the amount of cleaved IETD FRET substrate at the onset of mitochondrial depolarization as a measure for the integrated caspase-8/-10 activity over time (Fig. 6A, iii). We also determined the time from mitochondrial depolarization to the subsequent increase in IETDase activity, which reflects the duration from MOMP until activation of effector caspases (Fig. 6A, iv).
Evaluating the amount of FRET probe cleaved at the time of MOMP, we found that independent of the TRAIL dose applied, mitochondrial depolarization occurred when ϳ10% of the FRET substrate was cleaved (Fig. 8A). The subsequent time between MOMP and effector caspase activation was dose-independent as well, confirming that, following MOMP apoptosis, signaling proceeded in an all-or-none fashion (Fig. 8B).

DISCUSSION
TRAIL was shown to induce apoptotic cell death in transformed cells with high selectivity, rendering this cytotoxic ligand a prime candidate as a drug for novel anti-cancer therapies (1,23). To provide insight into how the consequences of TRAIL exposure manifest within the complex biological environment inside living cells, we analyzed the intracellular dynamics of caspase-dependent signaling during apoptosis initiation and execution by time lapse imaging of IETDase activity profiles. Our study showed that although the upstream caspase-8-like activity during TRAIL-induced apoptosis is clearly a dose-dependent response, this dose dependence can be translated into a dose-independent all-or-none response of cell death execution. Our data demonstrate that this is functionally achieved at a conserved threshold of substrate cleavage, reflecting the integration of caspase-8/-10 activity over time.
Upon TRAIL stimulation, procaspase-8 and -10 are recruited to and activated at the DISC. Both caspase-8 and -10 were described as closely related IETDases with largely overlapping  substrate specificities. However, it was shown that TRAIL-induced apoptosis initiation seems to be largely dominated by caspase-8, since caspase-10, although recruited to the DISC, cannot compensate for a loss in caspase-8 expression (24 -28). The majority of IETDase activity we observed during the apoptosis initiation phase therefore can probably be attributed to caspase-8.
Measuring the IETDase activities of caspase-8/-10 in the presence of the complexity of the full caspase signaling network is challenging, since other caspases were shown to possess significant affinities toward IETD recognition sites as well. In vitro assays identified that effector caspase-3 and -6 exhibited activities toward the supposedly specific caspase-8/-10 recognition site IETD, which were twice as high as the activities of caspase-8/-10 themselves (19,29). Similarly, commercially available synthetic caspase inhibitors were shown being too nonspecific to selectively inhibit effector caspases, since they affect caspase-8/-10 as well (18,19). We observed an ϳ12-fold increase in IETDase activity following MOMP, which seems to support the hypothesis that the majority of intracellular IETDase activity reflects effector caspase activity. We cannot exclude the possibility that activation of additional caspase-8/-10 via caspase-3/-6-dependent feedback loops downstream of MOMP might contribute to this increased IETDase activity as well (30). However, this contribution may be relatively small, since our biochemical characterizations suggested that the efficiency of procaspase-8 processing was independent of whether or not apoptotic signaling could proceed beyond MOMP (Fig. 2). Our experiments demonstrate that the contribution of downstream effector caspases needs to be excluded when determining the activities of initiator caspase-8/-10 based on substrate cleavage. We accomplished this in living cells by separating initiation and execution phases by parallel detection of MOMP.
Similar to the cell-to-cell asynchrony in apoptosis execution that can be observed for individual cells within a population (12,31), we found that also the upstream activation of initiator caspases is a response that occurs individually at the single cell level. These asynchronies were most prominent at submaximal TRAIL concentrations. Such asynchronies may arise from cellto-cell differences in cell cycle phases or gene and protein expression patterns (32) and highlight the requirement to analyze signaling at the single cell level. A limited number of "activated" receptors at low TRAIL doses could reduce the probability of receptor trimerization. This stochastic factor could contribute to the observed cell-to-cell variability in caspase-8/-10 activation time. Several mechanisms may further reduce the level of activated receptors and increase the variability of the onset time of the response. TRAIL decoy receptors (TRAIL-R3 and TRAIL-R4) might negatively regulate the cellular responsiveness to TRAIL by scavenging TRAIL before it can bind to TRAIL-R1 or -R2. In addition, TRAIL-R4 was shown to oligomerize with TRAIL-R2, resulting in a nonfunctional DISC as a potential second level of regulation mediated by decoy receptors (33,34). Probably more important, the binding of c-FLIP, an inactive caspase-8 homologue, might significantly delay or even fully inhibit formation of mature DISC complexes if it is expressed sufficiently high in relation to procaspase-8 (35,36). Indeed, it was recently shown in type I cells that c-FLIP controls the apoptotic response in a threshold-dependent manner, which is most prominent following low dose stimulation of CD95 death receptors (37,38).
Although type I cells can activate apoptosis execution by direct positive feedback between caspase-8 and -3, type II cells, such as HeLa cells, require activation of MOMP for efficient apoptosis execution (20). The question therefore remains whether in type II cells the activation of caspase-8 at the DISC already represents a cell death decision. The caspase-8/-10 activities we observed were dependent on the TRAIL dose and therefore differed from the dose-independent all-or-none signaling during effector caspase activation downstream of MOMP. However, caspase-8/-10 activation was not a transient response, but instead activities were detected to persist for hours in Bid-depleted cells. This suggests that a cell death decision can be made at the level of DISC formation during TRAILinduced apoptosis, as long as the subsequent signaling steps leading to MOMP and apoptosis execution are not blocked. In our experiments, we deliberately uncoupled caspase signaling from parallel transcription/translation-dependent survival signaling pathways that can be activated upon TRAIL receptor stimulation by the addition of CHX. We therefore may have suppressed NF-B and/or survival kinase signaling cascades that otherwise might have allowed for additional modification of the caspase-8/-10 response (3).
Regardless of the TRAIL concentration used, MOMP coincided with a FRET probe cleavage of ϳ10% under our experimental conditions. We found that at this threshold, dosedependent signaling kinetics were translated into a dose-independent apoptosis execution. This finding suggests that the MOMP decision threshold could correspond to the accumulation of a conserved critical amount of tBid. The requirement of the signaling network for such a threshold arises from the biological necessity to be insensitive to low amounts of tBid. Such low levels of intrinsic proapoptotic noise could arise e.g. from zymogenic activities of procaspases or accidental caspase autoactivation (40) and could sufficiently be filtered out by a threshold mechanism. It is noteworthy that the observed threshold for the MOMP decision was found at the lower end of the substrate cleavage range. This effectively means that HeLa cells have the possibility to cleave much more Bid than actually required for MOMP induction. This additional capacity can be crucial to override potential changes in the balance of pro-and antiapoptotic Bcl-2 family members toward compositions that otherwise would not allow MOMP to proceed. Corresponding to our findings, previous studies have shown that indeed submaximal amounts of tBid seem sufficient to effectively induce mitochondrial permeabilization and that a large proportion of Bid cleavage during extrinsically induced apoptosis seems to occur downstream of MOMP (6,41).
However, post-translational modifications of Bid may additionally delay or inhibit MOMP in some scenarios. It has been reported that Bid phosphorylation as well as impaired degradation of the NH 2 -terminal fragment following Bid cleavage can further modify cellular susceptibility to active caspase-8/-10 (42)(43)(44)(45). Indeed, nonlethal activation of caspase-8/-10 was shown to be a prerequisite for NF-B activation during lymphocyte activation (46) as well as for monocyte and trophoblast differentiation (47)(48)(49). Correspondingly, our analyses in Biddepleted cells have shown that caspase-8/-10 activation in the absence of MOMP in itself does not cause an immediate cell death response. In these scenarios, caspase-8/-10 activation is clearly not a cell death decision.
Based on analyses within living cells, our study thus provides mechanistic insight into how the initiation network of extrinsic apoptotic signaling is balanced to avoid unwanted induction of apoptosis execution and at the same time is sensitive enough to guarantee robust cell death responses following TRAIL administration.