A novel quenched fluorescent activity-based probe reveals caspase-3 activity in the endoplasmic reticulum during apoptosis† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c5sc03207e

A selective quenched activity-based probe detects caspase-3 activity in the endoplasmic reticulum of cancerous cells during apoptosis.


Introduction
Apoptosis (programmed cell death) is the major process by which multi-cellular organisms eliminate excessive, damaged and potentially dangerous cells. 1 Both normal and pathological processes such as embryonic development, cancer, autoimmune disorders, ischemia and reperfusion or Parkinson's and Alzheimer's disease involve apoptotic cell death processes. Importantly, the induction of apoptosis in cancerous cells is the desired outcome of many cancer chemotherapeutic treatments and initiation or inhibition of apoptosis is a key element in numerous therapies. 2,3 Therefore, the availability of imaging tools for tracking cell death immediately aer treatment would not only be desirable for basic research but also of great benet for evaluating therapy success.
The caspases are a family of cysteine proteases that are critical for the execution of apoptosis. They are divided into two sub-families: the initiator caspases (caspases-2, -8, -9 and -10) which are initially activated by specic death stimuli from receptors or the mitochondria and the effector caspases (caspases-3, -6 and -7) that are triggered in response to initiator activation and overtake extensive substrate proteolysis leading nally to cellular destruction and death. 4 Caspase-3 is a key mediator of the apoptotic process and the most procient caspase, featuring an astonishing low K M and high k cat for several substrates. 5 The inactive zymogen form of caspase-3 is a dimer and activation is post-translationally regulated by distinct cleavages of a short interdomain linker between the large and small subunits. 6,7 These proteolytic events enable a rearrangement of the activation loops in the structure and thus the formation of the active sites for substrate cleavage. 3 Under normal physiologic conditions, caspase-3 is localized in the cytoplasm in its inactive zymogen form. Caspase-3 activation occurs in the cytosol, active caspase-3 is then translocated to the nucleus. In contrast, cellular mislocalization of active caspase-3 can lead to apoptosis resistance. 8 However, there is still not much known about the kinetics of caspase-3 activation, the exact sub-cellular localizations of caspase-3 and the precise molecular mechanism of active caspase-3 translocation to the nucleus. 9,10 Consequently, tools for monitoring caspase-3 at high resolution in real-time would be highly desirable to better understand the dynamics of caspase-3 activation, most importantly in the context of disease.
Different strategies for monitoring caspases or more specifically caspase-3 activity have already been developed using chemical biology methods. For example, chemical biology approaches for directed activation or for identifying caspase  Table 1. (c) Chemical structure, fluorescence intensity and quenching efficiency of probes 17, 18 and 19. (d) Direct labeling of recombinant caspase-3 (upper panel) and legumain and cathepsin B in RAW cell lysate (lower panel) by indicated qABPs. Recombinant caspase-3 was incubated with increasing probe concentrations for one hour, the reaction was stopped and separated on a SDS PAGE and scanned for Cy5 fluorescence. Samples marked with + were pretreated with a caspase inhibitor (AB46 peptide) 30 min prior to the probe treatment. Legumain and cathepsin B from RAW cell lysates were labeled by the indicated qABPs similarly to caspase labeling. Samples marked with a, b or c were pretreated for 30 min with the inhibitors AB46 peptide, GB111-NH 2 25 or 5 to selectively block caspase-3; cathepsin B or legumain, respectively. (e) Direct labeling of active caspase-3 in intact MM1s cells undergoing apoptosis. The indicated qABP showed covalent binding to active caspase-3, seen at 17 kDa. Samples marked with + represent the pretreatment with a caspase-3/legumain inhibitor (AB46 peptide) or cathepsin B inhibitor (GB111-NH 2 ) which was added 1 h prior to the probe.
substrates have been developed consisting for example of active caspase specic antibodies or small molecule probes and substrates also containing non-natural amino acids. [11][12][13][14][15] In particular uorescent activity based probes such as FLICA, a uorophore-labeled uoromethyl ketone probe 16,17 or AB50, an acyloxy methyl ketone based probe, 18 have been generated and applied to monitor caspase activity in cells and small animals. Additional uorescent substrate probes such as TCapQ, 19,20 NucView 21 and split bioluminescence caspase probes have also been developed. 22,23 Unfortunately, activity based probes as well as antibody based probes create the need for washes and manipulation of samples prior to imaging, preventing real-time imaging and manipulation-free examination. Furthermore, substrate probes oen also target other cysteine proteases such as cathepsin B or legumain, 24 and diffuse away from their target enzyme, thereby diminishing high resolution imaging capabilities. 5 To overcome these limitations and to generate a true highresolution real-time imaging tool, we therefore developed a selective quenched uorescent activity based probe (qABP) for caspase-3. We have based our design on our previously published qABPs strategy, 25,26 the non-quenched caspase probes AB50-Cy5 18 and the legumain probe LE28. 27 Quenched uorescent ABPs are small molecules that covalently bind their target in an activity dependent manner. A short sequence in the probe drives recognition by the target and an electrophile termed the "warhead" covalently binds the target enzyme. Unlike regular uorescent ABPs, quenched probes include an additional quencher unit that is removed when the target protease binds to the probe. Therefore, upon binding to the activated enzyme, the probe transitions from a non-uorescent free probe to an enzyme bound uorescent complex 25 (Fig. 1a). In the present study, we generated a selective caspase-3 quenched probe without cross-reactivity with cathepsin B and legumain by modifying both the prime and non-prime probe regions. The selective caspase-3 qABP was then applied to time lapse systems to study cellular localization at high resolution over time in both apoptosis sensitive and resistant cancer cells. Interestingly, we thereby discovered novel caspase-3 activity patterns in these cell populations that were subsequently validated biochemically.

Development of selective caspase-3 qABPs and their in vitro evaluation
We set out to generate selective qABPs for caspase-3. We based our initial design on probes from the Bogyo group: AB46-Cy5, a non-quenched probe for caspase-3, 18 AB50-Cy5 18 and LE28, 27 and cathepsin quenched probes. 25,26 AB46-Cy5 (Cy5-E8D-AOMK-DMBA, 8 stands for 2-amino butyric acid, see Table 1, bottom) was designed to be an ABP for caspase-3 but displayed crossreactivity with legumain and cathepsin B. LE28 is a qABP based on AB50-Cy5 that targets both legumain and caspase-3 and contains a Cy5 uorophore linked to a Glu-Pro-Asp (P3-P2-P1) peptide scaffold and an acyloxymethyl ketone dimethylterephthalate propane linker attached to a quencher moiety (structures in Table 1, bottom). It is obvious that the cross reactivity to the two lysosomal cysteine proteases cathepsin B and legumain signicantly lower the usage of caspase probes turning the development of more selective compounds highly attractable.
The selective caspase-3 qABP was generated in a stepwise manner. First, inhibitors lacking the uorophore were synthesized and biochemically evaluated for their binding to caspase-3 and legumain (Fig. 1b). The most potent and selective inhibitors were then converted into probes by attachment of a uorophore to the inhibitor. We started by generating a library of inhibitors with varied P2 position and the quencher and/or linker moieties in the prime site of LE28. Tryptophan and phenylalanine were selected as P2 amino acids based on their reported weak legumain interactions. 28 We therefore synthesized EWD, EPD, EFD (P3-P2-P1) sequences and incorporated diamino-linkers of different length and various quenchers in the prime site (see general structures of compounds 1-16 in Table 1).
The tripeptide was generated on solid support using solid phase peptide synthesis (SPPS) similar to previously described methods. 26 The protected tripeptide was cleaved from the resin and converted to a chloromethylketone (CMK) using diazomethane. It was then coupled to different resin bound 2,6dimethyl-terephthalic-amidesthat were generated from the various diamino linkers, thus forming the acyloxymethyl ketone (AOMK) warhead (ESI Scheme 1 †). The product was then cleaved using mild acidic conditions for maintaining the Boc and t-Bu side chain protections. The free amine was acetylated or coupled to QSY21 or BlackBerry 650 (BBQ), thereby incorporating the quencher units into the probe library. All compounds were puried by preparative HPLC and characterized using LCMS. The overall yields of the nal products varied between 4-30%.

Biochemical evaluation of compound library
Inhibitors 1-16 were tested for their inhibitory potency against recombinant caspase-3 and legumain using RAW cell lysates. First, each inhibitor was incubated with RAW cell lysate or recombinant caspase-3 enzyme and the residual enzyme activity was assayed with 4 (a caspase-3 activity based probe that was found nonspecic for legumain activity, data not shown). The bound and free probe were separated by SDS-PAGE and scanned for Cy5 uorescence (Fig. 1b). Compounds 7-11 with tryptophan at P2 inhibited legumain over 50% at low concentrations (ESI Fig. 1 †); they were not further evaluated. All compounds except compound 12 (EWD) showed over 90% inhibition of caspase-3. We found that most EPD based inhibitors displayed the highest legumain inhibition ($90%), followed by EFD ($60%) based inhibitors, whereas lower inhibition was observed for EWD-based compound 12. This nding demonstrates that changing P2 from proline to phenylalanine reduces the binding affinity for legumain. The reduced potency of the P2 tryptophan compound for caspase-3 might be caused by a steric clash and unfavorable hydrophobic interactions.
Additionally, the change of the quencher moiety from QSY21 in 3 (EPD) to Blackberry in 6 (EPD) signicantly reduced the Table 1 The compounds differ in their peptide sequence at the P2 position; R 1 represents the corresponding side chain at this P2 position in the probe sequence E-P2-D. *R 2 represents an acyl group or one of the two quenchers, i.e. QSY21 or BBQ (Blackberry quencher). **R 3 indicates if the compound was fluorescently labeled or not. ***n denotes the number of (CH 2 ) units and thus the length of diaminolinker, % ACN denotes the percentage of acetonitrile at which the compound eluted from the analytical HPLC. All synthesized compounds were purified via C-18 or C-4 preparative RP column after each synthetic step and characterized by LCMS. The final products were obtained in 3-32% yield after the final step of isolation and in over 95% purity (i.e. giving a single peak in the chromatogram at 215 as well as 254 nm) binding affinity towards legumain. This led us to conclude that legumain reactivity can also be affected by the interaction of the inhibitor with the prime site of the enzyme and not only with the non-prime site. We then attached Cy5 to selected EFD containing inhibitors and continued investigating direct labeling of these enzymes (see structures in Fig. 1c).

Compound
The quenched probes 17, 18 and 19 were found to have dramatically reduced uorescence when compared to the nonquenched probe AB46-Cy5, with AB46-Cy5 showing 230-fold, 150-fold and 190-fold higher uorescence, respectively (Fig. 1c). Direct labeling of recombinant caspase-3 (Fig. 1d, upper panel) and legumain in RAW cell extracts (Fig. 1d, lower panel) was evaluated for these probes. The enzyme or lysates were incubated with increasing probe concentrations, with or without pretreatment of specic inhibitors for caspase, legumain or cathepsins. The free probes were separated from the enzyme- bound probe using gel electrophoresis aer which the gel was scanned for uorescence (Fig. 1d). We found legumain (36 kDa) in the upper band and cathepsin B (28 kDa) in the lower band of labeled RAW cell lysate. While AB46-Cy5 maximal labeling of caspase-3 was achieved at a concentration of 0.1 mM, probes 17-19 were less potent and reached their maximum labeling at 0.5 mM. Notably, they were considerably less potent towards legumain and cathepsin B in the tested concentration range.
We next tested the probe's ability to selectively label active caspase-3 in intact MM1s cells that were induced to undergo apoptosis by treatment with etoposide, a cytotoxic anticancer topoisomerase inhibitor. 29 Aer apoptosis induction, the cells were pre-treated with vehicle, caspase-3 or cathepsin B inhibitors, followed by AB46-Cy5 or the qABPs 17, 18 or 19 (Fig. 1e).
The uorescent probes were found to covalently bind active caspase-3, detected by a 17 kDa band on the gel. While AB46-Cy5 also targeted legumain, probe 19 labeled legumain only very weakly. Labeling of legumain by AB46-Cy5 is not surprising since P1 aspartic acid probes have been reported to bind legumain. 28 Unfortunately, cathepsin B was also labeled by probes 17-19, although in signicantly less intensity than caspase-3, as seen by the cathepsin inhibitor GB111-NH 2 25 sensitive 28 kDa protein band.
We have previously reported that cathepsin B has reduced affinity to compounds with steric hindrance in the prime site. 26 Therefore, we modied the primed binding site of probes 18 and 19 in an attempt to lower cathepsin B binding affinity. We generated two glycine-AOMK based qABPs: 21 with QSY21 and 22 with BBQ, in which the bulky quencher was incorporated close to the AOMK to induce a steric clash to the prime site of cathepsin B (Scheme 1a). To this end, the tripeptide chloromethyl ketone was condensed with commercial trityl protected glycine. The quencher and uorophore were then attached sequentially (Scheme 1b and ESI † Methods).
To further evaluate potential cross reactivity towards legumain in intact cells, human Colo-205 colorectal cancer cells that naturally express high levels of legumain and cathepsin B were treated with the various probes. Aer probe incubation, cell lysates were collected and separated by SDS PAGE and were then scanned for uorescence. We detected no non-specic labeling with 22 and very weak cathepsin B labeling with 21 (ESI Fig. 2 †). As previous studies indicated that it is difficult to achieve selective caspase-3 inhibition due to cross-reactivity with caspase-7, 30 we evaluated the potency of 22 towards recombinant caspase-7. Although we could detect weak binding of caspase-7 at higher concentrations, this binding does not interfere with caspase-3 labeling (ESI Fig. 3 †). Since compound 22 is potent and selective towards caspase-3 and also highly quenched, we selected it for subsequent studies in different biological systems.

Detecting caspase-3 activity in apoptotic cells using a next generation qABP
Our qABP is designed as a tool for real time visualization of caspase-3 activity during apoptosis. We rst determined the optimal time window for detecting caspase-3 activity in apoptotic cells using biochemical methods. We monitored caspase-3 activity aer chemotherapy-induced apoptosis in a time course experiment in a time interval of 2 to 24 hours in chemotherapy-resistant KMS11 and sensitive MM1s cells. Apoptosis was initiated with etoposide and probe 22 was added three hours before lysing the cells. Crude lysates were run on an SDS PAGE and labeled enzymes were detected by a uorescent gel scan (Fig. 2d). The caspase-3 signal was detected as a $17 kDa uorescent band that reached its maximum intensity between 12-16 hours aer etoposide treatment in the sensitive MM1s cells. Not surprisingly, we detected very little caspase-3 activity in the resistant KMS11 cells. Probe specicity was veri-ed by addition of the caspase-3 inhibitor ZVAD-FMK and by blotting with a caspase-3 antibody (Fig. 2e). These data show that our probe specically labels caspase-3 activity in intact cells and can be used to assess the dynamics of apoptosis in biological systems. Cisplatin was added to the growth media for indicated times to intact cells one day after seeding, samples marked with (+) were treated with ZVAD-FMK one hour before probe treatment. All samples were labeled by AB46-Cy5 for 3.5 hours, lysed and equal protein was loaded on a 14% SDS-PAGE gel. Fluorescent bands were detected as described above. (b) Fluorescent microscopy of caspase-3 activity in A2780 and A2780cis cells. In order to visualize the localization of active caspase-3 during apoptosis, A2780 and A2780cis cells were treated with 5 mM cisplatin for the indicated times, qABP 22 was added 24 hours after cisplatin addition and fluorescence was visualized using confocal microscope between 27-72 hours after apoptosis induction, scale bar ¼ 30 mm.
As probe 22 only becomes uorescent aer caspase-3 specic cleavage, we next examined the dynamics of caspase-3 activity in apoptotic cells using microscopy. MM1s cells were treated with etoposide and probe 22 was added for the last three hours of treatment before uorescent images were collected with a confocal microscope. Legumain and caspase inhibitors were included as controls (Fig. 3a-f). Images were then quantied for percent of Cy5 uorescent cells per eld (Fig. 3g). We detected no decrease in the uorescent signal with the legumain inhibitor while ZVAD-FMK (caspase inhibitor) signicantly reduced the signal. In order to conrm that the observed uorescent signal corresponded to labeling of active caspase-3, we analyzed the same samples by SDS-PAGE (Fig. 3h). The microscopy images closely matched the specic signal form the gel analysis, thereby showing specic labeling of cleaved caspase-3 by 22.
Fluorescence-Activated Cell Sorting (FACS) in combination with probe 22 was also used to analyze MM1s and KMS11 cells aer etoposide treatment. In the MM1s cells Cy5 uorescence increased from 23% (basal uorescence) to 88% within 18 hours aer apoptosis induction with etoposide. Addition of the inhibitor ZVAD-FMK reduced the uorescence back to the basal level of 23.2%, thereby verifying the specicity of the signal for cleaved caspase-3 (Fig. 3i). In contrast, we did not detect any signicant increase in uorescence in the resistant KMS11 cells: the basal apoptosis levels were 15.2% and were induced to 20.4% by etoposide treatment. Moreover, ZVAD-FMK had no signicant impact on the uorescent signal of KMS11 cells (Fig. 3i).

Real-time localization and kinetics studies of cleaved caspase-3 with probe 22
To determine the suitability of the qABP 22 for real-time imaging of active caspase-3 in apoptosis, we performed experiments with the adherent ovarian carcinoma cells A2780 and A2780cis which are sensitive and resistant to cisplatin treatment, respectively, a widely used anticancer chemotherapy. 31 First, we performed a time course study of caspase-3 activation over 3 days of cisplatin-induced apoptosis. At various time points throughout the experiment, AB46-Cy5 was added to label active caspase-3. Cells were then collected, washed, lysed and an equal protein amount per lane was loaded and resolved by SDS-PAGE, followed by scanning for Cy5 uorescence. We detected prolonged caspase-3 activation between 24 and 72 hours, with a maximum at 72 hours aer treatment in A2780 cells (Fig. 4a). The specicity of the labeled bands was veried by competition with the inhibitor ZVAD-FMK. In A2780cis cells, however, the signal of cleaved caspase-3 was observed at a later time interval aer cisplatin treatment (54 to 72 hours) and was signicantly less intense.
To visualize the localization of active caspase-3 during cellular apoptosis, we used probe 22 and captured images of cisplatin-treated A2780 and A2780cis cells for 72 h with a confocal microscope (Fig. 4b). In A2780 cells, a uorescent signal could be detected aer 27 hours and 51 hours. The signal corresponded to active caspase-3 since ZVAD-FMK treatment extinguished the uorescent signal. As expected, uorescence intensity was weaker in the cisplatin resistant A2780cis cells, although it was also specic to cleaved caspase-3 as demonstrated by competition with ZVAD-FMK.
Since we detected dened punctate patterns rather than diffuse signals of active caspase-3, we set out to investigate the localization of active caspase-3 during the apoptosis process in higher resolution. We used 22 as a real-time imaging probe and visualized live intact cells undergoing apoptosis in the presence of three markers for cellular compartments using confocal microscopy. A2780 cells were placed in a "live imaging" incubation chamber and were treated with cisplatin for 72 hours. The qABP 22 was added aer 24 h together with three different organelle-specic trackers, i.e. Lysotracker (an acidotropic marker) for localizing lysosomes, Mitotracker for mitochondria, and ER-Tracker to visualize the endoplasmic reticulum (ER) (Fig. 5a-c). We detected active caspase-3 in the cytoplasm and co-localized with mitochondria and the endoplasmic reticulum.
To visualize the kinetics of caspase-3 activation during the apoptotic process, a real time movie of A2780 cells treated with cisplatin was captured in a time lapse movie over 78 hours (ESI Movie 1 †). Snapshots from representative periods are presented in Fig. 5d. The movie reveals dened uorescent punctate representing caspase activity throughout the cellular compartments. Of note, we observed some variability of caspase activation during the time course in some cells, most likely due to their non-synchronicity.
To validate that active caspase-3 is indeed also localized to the ER, we performed a similar experiment in which A2780 cells were induced to undergo apoptosis by cisplatin treatment and analyzed them by uorescence microscopy aer addition of a cleaved caspase-3-specic antibody and an ER-marking calnexin antibody. Images show a clear co-localization between the Fig. 6 Caspase 3 activity is localized in the ER. (a) A2780 cells were treated with cisplatin for 50 hours; cells were fixed, permeabilized, blocked and treated with calnexin and caspase-3 antibodies. Cells were imaged with a Zeiss fluorescent microscope. Red, cleaved caspase-3; green, calnexin; blue DAPI, scale-bar is 5 mm. (b) A2780 and A2780cis cells were treated with cisplatin for indicated times. Probe 22 was added to cell lysate and cell fractionation was performed. Equal protein amounts from all fractions were spread by a 14% SDS-PAGE, in-gel fluorescence shows caspase bands in the ER fraction. (c) Identification of caspase-3 in ER fractions. The ER fractions from cisplatin treated and untreated A2780 cells were labeled with 22 and subjected to immuno-precipitation with an antibody against Cy5. Labeled bands and corresponding areas from untreated samples were cut out and subjected to mass spectrometry. Several caspase-3 peptides were identified from the P17 subunit, two in high confidence shown in green. (d) Western blot of ER and cytosolic fractions from A2780 cells that were cisplatin treated or untreated for 3 days with a cleaved caspase-3 antibody. cleaved caspase-3 antibody and the calnexin marked ER (Fig. 6a). In addition, we detected cleaved caspase-3 in the nucleus of many cells undergoing cisplatin-mediated apoptosis (ESI Fig. 4 †).
To conrm and further study the kinetics of caspase-3 activity appearance in cellular compartments, we performed cell fractionation experiments with cells undergoing apoptosis over the course of 72 hours. A2780 or A2780cis cells were treated with cisplatin and probe 22 was added aer cell lysis. The mitochondria, ER, nuclear and cytosolic fractions were then collected. We could for the rst time detect caspase-3 activity in the ER fraction aer two days of treatment in A2780 cells. From this time-point on, the intensity continuously increased over time. Interestingly, we found in A2780cis cells lower and delayed caspase-3 activity levels but detected active caspase-3 in the ER already one day aer treatment started. In both cell lines, caspase-3 activity was also detected in the mitochondria, cytosol and nucleus fractions, as expected from previous reports 32-34 (Fig. 6b).
In order to conrm the quality of our fractionation method, proteins extracted from each fraction (e.g. mitochondria, endoplasmic reticulum, nucleus and cytosol) were digested and analyzed by mass spectrometry-based label-free quantication and sub-sequent analysis for pathway enrichment (ESI Fig. 5a †). As anticipated, pathways corresponding to metabolic activities such as glycolysis and oxidative phosphorylation were highly enriched in the cytoplasm and mitochondria fractions, while RNA processing and protein folding were enriched in the nucleus and endoplasmic reticulum, respectively. Interestingly, we detected a few inhibitors of apoptosis in the ER fraction, such as cell division cycle and apoptosis regulator 1 (CCAR1), and apoptosis inhibitor 5 (API5). A full list of proteins detected in each cell fraction is found in ESI Table 1 †.
To corroborate that the labeled bands were indeed caspase-3, the labeled ER fractions were subjected to immuno-precipitation with an anti-Cy5 antibody followed by mass spectrometry analysis of the labeled bands. Six caspase-3 peptides were detected in the cisplatin treated samples while they were absent in the control samples. Two of the found sequences were high condence tryptic peptides that originated from the caspase-3 p17 subunit, which enabled detection of caspase-3. In addition, two short tryptic peptides and two trypsin miscleavage peptides were detected with lower condence, strengthening caspase-3 is the specic and only target of 22 in apoptotic cells (Fig. 6c). Finally, the ER fractions were subjected to western blot with an antibody against cleaved caspase-3, here again cleaved caspase-3 was detected in the ER fraction only aer cisplatin treatment (Fig. 6d). In addition, to eliminate the possibility that the signal in the ER was generated by binding to caspase-7, a western blot with a caspase-7 antibody was performed with cellular fractions aer cisplatin treatment. The cleaved caspase-7 was not detected in the ER fraction (ESI Fig. 6 †).

Discussion
While numerous tools to study caspase activity have been produced, here we show for the rst time a specic quenched activity-based probe (qABP) to enable high-resolution imaging. We have developed a selective caspase-3 qABP 22 that covalently modies active caspase-3 and lacks background uorescence when free in solution while generating a uorescent signal when it is bound to its enzyme target. We found 22 to have versatile applications including high-resolution real-time imaging, real-time monitoring of caspase-3 activation during apoptosis and various biochemical applications.
In contrast to activity-based probes for proteases such as the cathepsins or the proteasome for which highly selective probes are easily created, 25,35 generating selective probes towards caspases had so far proven to be a signicant challenge. We found that the reactivity of chemical probes to legumain and cathepsin B is directly affected by the interaction of the inhibitor with the enzyme's prime binding sites. Therefore, we undertook a rational approach to lower their affinity to these proteases. To this end, we modied our qABP at both the P2 position (non-prime site) and by appropriate positioning of a blackberry quencher to impair prime sites interactions. Importantly, these modications did not reduce the potency towards caspase-3, since we included a glycine residue at the prime site, that is well accepted by caspase-3, 36 leading to a selective caspase qABP.
The use of ABPs to detect caspase activity instead of antibodies for cleaved caspase-3 is of great importance since the cleaved "active" caspase-3 may be inhibited by an active site steric blockage that is induced by different endogenous inhibitors. In such cases the cleaved caspase-3 antibody might recognize a cleaved-inhibited form of caspase-3, while the ABP would be able to detect the "true" activity. Furthermore, although the lack of enzymatic amplication causes covalent modiers to have weaker signals than substrate based reporters, they are still highly valuable since they can report on the exact location of the target and be applied for SDS-PAGE analysis.
In this report we applied our newly developed probe to investigate caspase-3 localization during apoptosis in the ovarian carcinoma cell lines A2780 and A2780cis. In addition to the expected cytosolic activity several punctate stains appeared ( Fig. 4b and 5). Using cellular tracker stains together with cellular fractionation we discovered that the activity in the organelles is located in the mitochondria and the ER. There have been numerous reports describing the localization of caspase-3 mainly as cytosolic and nuclear, 10,37 a few reports on caspase-3 localized to the mitochondria 32,33 and one report indicating active caspase-3 found in the ER of rabbit hippocampus upon aluminium-induced neurotoxicity. 38 Furthermore, caspase-7 was reported to be localized in the ER following anti FAS treatment in mouse liver. 39 Therefore, ER localization of active caspase-3 is supported by literature precedence but is now detected for the rst time in cancer cells.
The ovarian carcinoma cell lines A2780/A2780cis cells exhibited differential activation of caspase-3 during apoptosis (Fig. 4). Caspase-3 activity was greatly reduced in A2780cis cells which might explain their resistance to cisplatin treatment. We anticipate that this reduction in caspase-3 activity originates from endogenous inhibitor binding. In fact, high basal levels of the caspase-3 inhibitor, XIAP, and other apoptosis inhibitors such as survivin, and C-IAP-1 in addition to the antiapoptotic protein BCL-2 were found independent of cisplatin treatment in these cells 40 while the sensitive counterpart A2780 cells showed reduced levels of these proteins upon cisplatin treatment.
What is the physiological relevance of directing caspase-3 to the ER? We suggest two hypotheses: the cell tries to limit cellular destruction by positioning caspase-3 in the vicinity of ER-based apoptosis inhibitors, thereby preventing caspasemediated cell death. Alternatively, caspase-3 is localized to the ER in order to dismantle this compartment as a part of the apoptosis process. As we show in Fig. 6b, we detected different caspase-3 kinetics in ER localization between the resistant A2780cis and sensitive A2780 cells in which the resistant cells show more rapid ER localization. Therefore, we favor the rst hypothesis as shuttling or activating caspase-3 in the ER reduces its destructive capabilities in an attempt to diminish apoptosis. Nevertheless, both hypotheses require further investigations to elucidate the underlying molecular mechanisms.
There are a few published reagents for in vivo imaging of apoptosis which contain radiolabeled reporters like 18 F tracers. These include the Aposense compounds [41][42][43] or Isatin based compounds. 44 Several chemical modications of probe 22 would be necessary to enable detection of apoptosis in vivo since 22 includes an ester moiety that would be hydrolyzed by esterases found in the blood stream leading to probe destruction. Nevertheless, 22 is a unique reagent that could be widely used in cell cultures to investigate caspase activity and dynamics in the live cell. Taken together, we present probe 22 as a tool to investigate the changes in activity levels and localization of caspase-3. In our work we apply the probe to further investigate impaired caspase-3 activation in cells that are resistant to apoptosis by chemotherapy induction. We anticipate that a broader use of this novel chemical tool will allow gaining insight into caspase-3 regulation in numerous biological and disease based systems.

General methods
The reagents were purchased from commercial suppliers and used without further purications. All solvents were HPLC grade. All water-sensitive reactions were performed in anhydrous solvents under a positive pressure of argon. Cy5 succinimidyl ester (SE) was purchased from GE Healthcare. QSY 21 SE was bought from Molecular probes, Invitrogen and BBQ SE was acquired from Berry & Associates. Reactions were characterized by liquid chromatography-mass spectrometry (LC-MS) (Thermo Scientic MSQ-Plus attached to an Accela UPLC system) using reverse-phase chromatography with a water/ acetonitrile gradient. For preparative purication of the compounds reverse-phase HPLC was conducted with a Chromeleon (Dionex) system using C18 or C4 columns. Fluorescent gels were scanned with a Typhoon FLA 9500 atbed laser scanner (GE Healthcare Bio-Sciences AB, Uppsala, Sweden).

Probe synthesis
The synthesis of compounds 1-19 is described in ESI Scheme 1 and Table 1 and in detail in the ESI. † The synthesis of qABP 21 and 22 is shown in Scheme 1.

Competition assay in RAW cell lysate
Legumain inhibitory potency of probes was determined in RAW cell lysate using phosphate citrate buffer, pH 5.3, 5 mM DTT and 25 mg total protein per sample. The lysate was incubated with increasing concentrations of the corresponding probe for 1 h at 37 C (with or without pretreatment of an inhibitor, AB46 peptide) followed by labeling with 4 for an additional 30 min. The reaction was stopped with 4Â sample buffer (40% glycerol, 0.2% Tris/HCl pH 6.8, 20% b-mercaptoethanol, 12% SDS and 0.4 mg mL À1 bromophenol blue) and boiled. Lysates were separated by SDS-PAGE 12.5% acrylamide and visualized by uorescence scanning of the gel with an Odyssey atbed laser scanner (excitation/emission 680/700 nm).

Competition assay of caspase-3
Recombinant human caspase-3 (19 mM, a gi from Guy Salvesen, Burnham Institute, La Jolla, CA) was diluted in reaction buffer (100 mM Tris pH 7.4, 0.1% CHAPS, 10% sucrose and 4 mM DTT) to 100 nM concentration. Increasing probe concentrations were added to caspase-3 at 37 C for 1 h (with or without pretreatment with AB46 peptide), 18 4 was added and the resulting solution was mixed for additional 30 min. The reaction was stopped as described above. Samples were separated on a 14% SDS PAGE and gels were scanned for uorescence by an Odyssey atbed laser scanner, excitation/emission 680/700 nm.

Direct labeling of recombinant enzyme
Enzymes were incubated with increasing probe concentrations, with or without pretreatment with an inhibitor (AB46 peptide, GB111-NH 2 or compound 5, which bind to caspase 3, cathepsin B and legumain, respectively). The free probe was separated from the enzyme-bound probe using gel electrophoresis and the gel was scanned for uorescence to detect the amount of enzyme bound probe in a similar fashion to the indirect labeling experiments.
A2780 and A2780cis ovarian carcinoma cells were cultured in RPMI medium (GIBCO) adjusted to pH 6 supplemented with 10% FBS, 50 mg mL À1 of gentamicin at 37 C under 5% of CO 2 .

Probes penetration and labeling in apoptotic cells
MM1s and KMS11 cells were seeded in a 24 wells plate in varying density 2-4 Â 10 6 cells per well. Cells were induced to undergo apoptosis by addition of 75 mM of etoposide for 4-24 h. Cathepsin B inhibitor (5 mM of GB111-NH 2 ) or caspase inhibitor (10 mM of AB46 peptide) or vehicle (DMSO) were added. Probes 17-22 were then added to the growth media for an additional 3 h unless otherwise noted. The nal DMSO concentration was kept at 0.1% in all wells. Cells were washed once with PBS, lysed by incubation in hypotonic lysis buffer (10 mM PIPES, 10 mM KCl, 5 mM MgCl 2 , 2 mM EDTA, 1% NP40 and 4 mM DTT) for 10 min on ice followed by transferring through a 30G syringe. The lysates were frozen and thawed three times and nally supernatants were collected aer a 20 min spin-down at 4 C. The protein concentration was quantied by a Bradford assay. 30-100 mg total protein was separated by a 14% SDS PAGE and the gel was scanned for uorescence by an Odyssey laser scanner (680 nm excitation/700 nm emission).
In experiments with A2780 or A2780cis cell cultures, cells were seeded 1.5 Â 10 6 cells per well 24 h prior to apoptosis induction with 5 mM of cisplatin for 24-72 h and lysates were prepared as described above.

FACS studies
MM1s and KMS11 cells were seeded in a 24 wells plate in density 1 Â 10 6 per well and were treated with DMSO vehicle or induced to undergo apoptosis by addition of 75 mM etoposide for 18 h. The caspase inhibitor (ZVAD-FMK, 50 mM) was added to selected wells. Aer 15 h, probe 22 (1 mM) was added to the growth media for an additional 3 h, the nal DMSO concentration was kept at 0.1% in all wells. Cells were then washed once with PBS and re-suspended in RPMI without phenol red. Flow cytometry was performed to determine the labeling of cleaved caspase-3 using a LSRII ow cytometer (BD Biosciences). The resulting data were analyzed with FACS express soware (BD, BD Biosciences).

Live cells imaging
A2780 and A2780cis cells were seeded at a density of 100 000 cells per well in an 8-well coverslip chamber (Nunc Lab-Tek). One day later, cells were treated with cisplatin (5 mM) for 24 h. The media was replaced with RPMI without phenol red but still containing cisplatin (5 mM). DMSO or 1 mM of probe 22 was added to the wells for additional 3 h of incubation (controls in which ZVAD-FMK 100 mM was added before treatment were applied). The uorescent signal of cleaved caspase-3 (excitation/emission Cy5 lter) was captured using a confocal microscope (Olympus FV10i inverted microscope, Tokyo Japan, equipped with a Â60 Oil DIC lens) over 72 h and analyzed using FLUOVIEW soware.

Co-localization studies
A2780 and A2780cis cells were seeded in an 8-well coverslip chamber (Lab-Tek) at a density of 100 000 cells per well and grown at 37 C under 5% of CO 2 , 24 hours later, cells were treated with cisplatin (5 mM) for 24 h. The media was replaced with RPMI without phenol red, supplemented with 3% FCS and either DMSO or 1 mM of probe 22 was added to the wells. Aer 48 h, the different cellular trackers were added in order to perform co-localization studies of both the tracker and the cleaved caspase-3 labeled by 22. To reveal the location of mitochondria, probe-containing medium was withdrawn and medium with Mitotracker-orange (Life technologies, M7510, conc. 500 nM) was added 30 min prior to imaging, then washed for 5 min. To localize the endoplasmic reticulum, medium containing probe was withdrawn and ER-tracker solution (Life technologies, 500 nM in HBSS buffer) was added for the last 30 min prior to imaging. To reveal whether the signal is lysosomal, LysoTracker (Life technologies, 50 nM) was added minutes before imaging. Cells were live imaged using a Zeiss LSM 710 Axio Observer.Z1 with a 63Â/1.4 Oil DIC M27 lens, in Cy5, FITC and M-cherry channels.
Immuno-uorescence co-localization studies A2780 cells were seeded at a density of 15-30 000 cells per well in an 8-well coverslip chamber (Nunc Lab-Tek) and incubated at 37 C under 5% of CO 2 . On the next day, cells were treated with cisplatin (5 mM) for 50 h. The media was then withdrawn and the cells were xed using 4% paraformaldehyde. Cells were permeabilized using 0.1% Triton-X for 15 min, washed and incubated with 2% BSA for 20 min at RT. Cleaved caspase-3 Asp175 (Cell signaling #9664) staining was done using a 1 : 100 dilution of Cas-Block solution (Life Technologies) overnight at 4 C. Simultaneously, calnexin (goat monoclonal antibody, cell signaling) staining was done using a 1 : 200 dilution of Cas-Block solution overnight at 4 C. Cells were washed and incubated with an anti-rabbit secondary antibody conjugated to Cy5 and an antigoat secondary antibody conjugated to Cy3. Cells were imaged with a confocal microscope (Zeiss LSM 710 Axio Observer.Z1 with a 63Â/1.4 Oil DIC M27 lens, in Cy5, Dapi and Dsred channels).

Cell fractionation
Proteins from the nuclear, mitochondrial, cytosol and ER fractions were extracted as described previously. 38 In brief, A2780 cells were trypsinized, washed with PBS and centrifuged for 5 min at 1200 rpm. Cells were gently homogenized through syringe needles in 2 volumes of cold suspension buffer (20 mM HEPES-KOH (pH 7.5), 250 mM sucrose, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA). The homogenates were labeled with 1 mM probe 22 for 20 min at 37 C.
Aer labeling, homogenates were rst centrifuged at 750 Â g at 4 C for 10 min to isolate the nuclear fraction, and then at 8000 Â g for 20 min at 4 C to separate the mitochondrial from the cytosolic fraction. The 8000 Â g pellets were re-suspended in cold buffer without sucrose and used as the mitochondrial fraction. The supernatant was further centrifuged at 100 000 Â g for 60 min at 4 C to separate the cytosolic from the ER fraction. Protein concentrations were determined with Bradford protein assay reagent (Sigma). Equal amount of proteins from the nuclear, mitochondrial, cytosolic and microsomal fractions were separated by SDS-PAGE (14% gel), and scanned for uorescence by a Typhoon scanner FLA 9500 (excitation/emission 630/670 nm) (GE Healthcare Bio-Sciences AB, Uppsala, Sweden).

Western blotting
Protein lysates or cellular fraction were separated on SDS-PAGE gels were transferred to PVDF membranes for immunoblotting.
Membranes were probed with cleaved caspase-3 Asp175 (Cell signaling #9664) and calnexin (BD Biosciences #610523) antibodies and detected using the appropriate HRP-conjugated secondary antibody, followed by an ECL assay (Biological Industries, Kibbutz Beit Haemek, Israel). Visualization of the chemiluminescent protein bands was performed using the Bio-Rad ChemiDoc XRS chemiluminescence detection system.

Immuno-precipitation
For each sample, 1 mg total protein was incubated overnight in rotation at 4 C. 100 mL protein A/G (Santa Cruz, CA, USA #2003) was added to each sample for two hours in rotation at 4 C. Aer washing, 25 mL 2Â sample buffer was added to the beads and boiled for 10 min. Proteins were separated on a 14% SDS-PAGE gel.

In gel digestion
Eluted proteins were resolved on a 12.5% SDS-PAGE and visualized in situ by scanning uorescence with a Typhoon FLA 9500 uorescence scanner. Gel regions containing labelled proteins were then excised using the Blind-Cut method. 45 A printout of uorescently labeled proteins was placed below the gel and regions corresponding to labeled proteins were cut out using a razor blade as described in reference 46. Proteins were in gel reduced with 2.8 mM DTT (60 C for 30 min) and alkylated with 8.8 mM iodoacetamide in the dark at room temperature for 30 min. Protein digestion was carried out using modied trypsin (Promega) at a 1 : 10 (w/w) enzyme-to-substrate ratio in the presence of 10% acetonitrile in 10 mM ammonium bicarbonate. The reaction was carried out at 37 C overnight and tryptic digests were resolved by reverse-phase silica capillaries (J&W). Peptides were separated by an acetonitrile water gradient including 0.1% formic acid. Mass spectrometry analysis was performed on a Q-Exactive plus (QE, Thermo) in positive mode. A full MS scan was followed by High energy Collision Dissociation (HCD) of the 10 most intense ions selected from the rst (precursor) MS scan. The MS 2 spectra were uploaded by Proteome Discoverer 1.4 soware to Sequest (Thermo) and Mascot (Matrix Science) search engines and searched against the UniprotKB human database.

In solution digestion
Proteins from cellular compartments fractionation (10 mg) were reduced with 10 mM DTT in 8 M urea for 1 hour followed by alkylation with 50 mM iodoacetamide for another 30 min in the dark. Protein digestions were then carried out with LysC (Wakochemicals) at 1 : 100 (w/w) enzyme-substrate ratio for 3 hours at 37 C aer which the urea concentration was reduced to 1.6 M with 25 mM ammonium bicarbonate and trypsin (Promega) was added at 1 : 50 (w/w) enzyme-substrate ratio for another 16 hours at 37 C. 47 Tryptic peptides were de-salted on homemade C 18 -Empore (Sigma-Aldrich) StageTips. 48 Peptides were then taken up in 0.1% formic acid and analyzed by LC-MS/MS.

LC-MS/MS
Peptides from the in-solution digests were analyzed on an Orbitrap Elite instrument (Thermo) 49 that was coupled to an EASYnLC-1000 (Thermo) liquid chromatography (LC) system. The LC was operated in two-column mode. A 3.5 cm pre-column (100 mm Â 3.5 cm, fused silica capillary with home-made integrated frit) packed with Reprosil-Pur 120 C18-AQ 3 mm resin (Dr Maisch) was placed in front of the analytical column (75 mm Â 23 cm), fused silica capillary with integrated pico frit emitter; New Objectives PF360-75-15-N-5) which was packed in-house with Reprosil-Pur 120 C18-AQ 3 mm resin (Dr Maisch). The analytical column was hooked up to a nanospray ex ion source (Thermo). The LC was equipped with two mobile phases: solvent A (0.1% formic acid, FA, in UPLC grade water) and solvent B (0.1% FA in acetonitrile, ACN). Peptides were delivered to the pre-column via the integrated autosampler at a ow rate of 2-5 mL min À1 in 100% solvent A. Peptides were subsequently resolved on the analytical column by running a 70 min gradient of solvent A and solvent B (start with 2% B; gradient 2% to 35% B for 60 min; gradient 35% to 100% B for 5 min; 100% B for 5 min) at a ow rate of 300 nL min À1 .
The Orbitrap Elite mass spectrometer was operated using Xcalibur soware (version 2.2 SP1.48). The mass spectrometer was set in the positive ion mode. Precursor ion scanning was performed in the Orbitrap analyser (FTMS) in the scan range of m/z 300-1300 and at a resolution of 120 000 with the internal lock mass option turned on (lock mass was 445.120025 m/z), polysiloxane. 50 Product ion spectra were recorded in a data dependent fashion in the ion trap (ITMS) in a variable scan range and at a rapid scan rate. The ionization potential (spray voltage) was set to 1.6-2.0 kV. Peptides were analyzed using a repeating cycle consisting of a full precursor ion scan (1.0 Â 10 6 ions) followed by 15 product ion scans (1.0 Â 10 4 ions) where peptides are isolated based on their intensity in the full survey scan (threshold of 500 counts) for tandem mass spectrum (MS 2 ) generation that permits peptide sequencing and identication. CID collision energy was set to 35% for the generation of MS 2 spectra. During MS 2 data acquisition dynamic ion exclusion was set to 120 seconds with a maximum list of excluded ions consisting of 500 members and a repeat count of one. Ion injection time prediction, preview mode for the FTMS, monoisotopic precursor selection and charge state screening were enabled. Only charge states bigger than 1 were considered for fragmentation.

Peptide and protein identication
The recorded RAW les were searched against the UniprotKB human database using the Andromeda search engine as integrated in MaxQuant (version 1.5.2.8). Andromeda searches allowed oxidation of methionine residues (16 Da) and N-terminal acetylation (42 Da) as variable modication as well as carbamidomethylation on cysteine (57 Da) as a static modication. All other settings for MaxQuant were the default settings. Most importantly the FDR for protein identication was set to 0.01 and the label-free quantication (LFQ) and match between runs options were enabled.

Data processing and statistical analysis
Data analysis was done in Perseus (version 1.5.0.9) and performed on log 2 scaled normalised label free quantication (LFQ) values. Proteins were excluded from the analysis if they were either identied only by their variable modications, identied using the reverse database or classied as a potential contaminant. Three replicates of each cellular fraction were grouped and valid values required at least two numeric values in at least one group. Missing values were replaced from the normal distribution and Anova statistical analysis performed to identify differentially expressed proteins with FDR ¼ 0.05 and S0 ¼ 1. Fischer's exact test was used for analysis of annotation enrichments with Benjamini-Hochberg FDR of 0.02. For uorescent microscopy studies ANOVA test using Sidak's correction for multiple comparisons was done using Graphpad Prism 6 Soware San Diego.

Conclusions
We have developed a specic quenched uorescent activitybased probe that serves as a tool to investigate the changes in activity levels and localization of caspase-3. The application of our selective probe reveals that caspase-3 is active in the endoplasmic reticulum during apoptotic cell death of cancer cells.