During apoptosis, initiator caspases (8, 9 and 10) activate downstream executioner caspases (3, 6 and 7) by cleaving the IDC (interdomain connector) at two sites. Here, we demonstrate that both activation sites, site 1 and site 2, of caspase 7 are suboptimal for activation by initiator caspases 8 and 9 in cellulo, and in vitro using recombinant proteins and activation kinetics. Indeed, when both sites are replaced with the preferred motifs recognized by either caspase 8 or 9, we found an up to 36-fold improvement in activation. Moreover, cleavage at site 1 is preferred to site 2 because of its location within the IDC, since swapping sites does not lead to a more efficient activation. We also demonstrate the important role of Ile195 of site 1 involved in maintaining a network of contacts that preserves the proper conformation of the active enzyme. Finally, we show that the length of the IDC plays a crucial role in maintaining the necessity of proteolysis for activation. In fact, although we were unable to generate a caspase 7 that does not require proteolysis for activity, shortening the IDC of the initiator caspase 8 by four residues was sufficient to confer a requirement for proteolysis, a key feature of executioner caspases. Altogether, the results demonstrate the critical role of the primary structure of caspase 7's IDC for its activation and proteolytic activity.

INTRODUCTION

Proteases play crucial roles in regulating a wide range of events such as digestion, coagulation, development and elimination of superfluous and damaged cells. In animals, the latter two events essentially occur by apoptosis, an inflammation-free process in which cellular functions are halted and in which the cell is packaged into smaller bodies ready for efficient phagocytosis by professional macrophages or neighbouring cells. Apoptosis is driven by a family of ubiquitously expressed cytosolic cysteine peptidases called caspases. Although the primary role of caspases is to cause cellular demise [1], they are necessary also for some non-lethal processes [2,3].

Apoptotic caspases are divided into two groups: namely initiator (caspases 8, 9 and 10) and executioner (caspases 3, 6 and 7) caspases. They are all present in cells as latent zymogens but the mechanism by which they are activated differs. Initiator caspases are activated by dimerization on macromolecular platforms and they are further stabilized by cleavage of the IDC (interdomain connector) that separates the large and small subunits of the catalytic domain [47]. Therefore cleavage of the IDC is not essential for the proteolytic activity of initiator caspases. Conversely, the executioner caspases are always dimeric and activated solely by processing of the IDC by initiator caspases or other peptidases able to cleave the IDC [810]. Cleavage of the IDC, usually at two sites close in the primary structure, exposes the C-terminus of the large subunit and the N-terminus of the small subunit, which is sufficient for executioner caspase activation [11]. The processed neo-N-terminus of the small subunit is also key in the interaction of caspases 3, 7 and 9 with the endogenous inhibitor XIAP (X-linked inhibitor of apoptosis protein) [1215] and protein stability of caspases 3 and 7 [16]. Based on structural and biochemical analyses, the IDC is a highly flexible segment and is therefore very sensitive to proteolysis [8,17].

To study the relevance of the IDC primary sequence in the activation process of caspases, we chose caspase 7 as our model of executioner caspases. Our work is facilitated by the availability of several X-ray structures of the active and the zymogen forms of caspase 7 that has allowed us to make structure-based predictions. Like all apoptotic caspases, caspase 7 possesses two cleavage sites in the IDC. Site 1 is always cleaved before site 2 (and referred to as site 1 and site 2 throughout the text) in cellulo, in in vitro assays by initiator caspases and during bacterial protein expression in which caspases self-activate [18,19]. Notably, both cleavage sites are suboptimal for processing by initiator caspases and also by self-processing because the preferred recognition sequences of caspases 7, 8 and 9 do not match the sequence of the cleavage sites. Moreover, initiator caspase IDCs are longer than those of most executioners supporting the notion that initiator caspases must have enough leeway to become active without cleavage. This is necessary because there is no upstream protease in their respective pathways to activate them. Here, we have focused on the primary structure of the two activation sites of caspase 7 and how these sequences influence the activity and activation of the enzyme by the initiator caspases. We also show that the location of the cleavage site is paramount in dictating activation efficacy, which also suggests the contribution of other factors beyond the primary sequence. We also test the hypothesis that the IDC's length constrains the caspase by requiring cleavage to become active. Our results demonstrate the importance of the primary structure of the IDC in the activity and activation of caspase 7.

EXPERIMENTAL

Reagents

All caspase substrates were obtained from Biomol and were calibrated using the molar absorption coefficient (ϵ) of free pNA (p-nitroanilide)/afc [7-amino-4-(trifluoromethyl)coumarin] after full hydrolysis using excess caspase 3. The irreversible caspase inhibitor Z-Val-Ala-DL-Asp-CH2F [benzyloxycarbonyl-valyl-alanyl-DL-aspartyl-fluoromethane; also known as Z-VAD-FMK (‘benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone’) used to active site-titrate caspases was purchased from MP Biomedicals. Other chemicals were from Sigma–Aldrich or Laboratoire MAT.

Plasmids, recombinant protein expression and purification

Site-directed mutagenesis was performed using overlapping PCR, and the ensuing cDNAs were sequenced to ensure their integrity. Full-length caspase 7 cDNA subcloned into pET-23b(+) (Novagen) [10] and ΔDEDs [mutant lacking DEDs (death effector domains)] caspase 8 cDNA in pET-15b (Novagen) [20] were used for all constructs. Full-length caspase 9 and ΔCARD [mutant lacking CARD (caspase recruitment domain)] caspase 9 were described previously [6]. Caspases, CrmA (cytokine response modifier protein) and p35 were produced in the BL21(DE3)pLysS (Novagen) Escherichia coli strain as described previously [19,21,22]. Pro-caspase 7 mutants were produced following short expression times (25 min). In some instances, cleavage site mutants were activated by co-purification with untagged caspase 8 catalytic domain to cleave the caspase 7 zymogens that do not self-activate in E. coli. The remaining active caspase 8 was then neutralized with the poxvirus CrmA protein, which does not inhibit caspase 7 [22]. Protein concentrations of inactive caspases or of caspases that cannot be active site-titrated were determined using Coomassie protein assays and the Edelhoch relationship on purified material [23]. For mammalian expression, all cDNAs were subcloned by PCR into an engineered pcDNA3 plasmid that encodes a C-terminus FLAG epitope (LEDYKDDDDK).

Cell culture, transfection and apoptosis stimuli

The HEK (human embryonic kidney)-293Ad cells (MP Biomedicals) were propagated in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) fetal bovine serum (Wisent), 2 mM L-glutamine (Wisent) and antibiotics. For caspase 8-directed apoptosis, cells grown in 100 mm plates were transfected using FuGENE™ 6 (Roche Diagnostics) and 6 μg of the indicated plasmid DNA then split into different 6-well plates after 24 h. After another 24 h, cells were left untreated or treated with the indicated concentration of recombinant human soluble KillerTRAIL (Enzolife Sciences). Detergent extracts {mRIPA buffer [50 mM Tris/HCl (pH 7.4), 100 mM NaCl, 1% (v/v) Nonidet P40, 0.5% deoxycholate and 0.1% SDS] with protease inhibitors} were prepared and analysed by the method described in [18] after 12 h of treatment. For caspase 9-directed caspase activation, cells grown in 150 mm plates were transfected using FuGENE™ 6 (Roche) and 9 μg of the indicated plasmid DNA then split into three 150 mm plates after 24 h. After another 24 h, a hypo-osmotic extract was prepared as described in [24] in the presence of 30 μg/ml RNase A, and programmed to activate caspases using 1 μM horse cyt c (cytochrome c) and 1 mM dATP (both from Sigma–Aldrich) for the indicated period of time. Reactions were stopped with SDS gel loading buffer. In both cases, samples were analysed by immunoblotting as previously described [25] using M2 (0.5 μg/ml; F1804; Sigma–Aldrich), p23 (1:2000; MA3–414; Pierce, Rockland, IL, U.S.A.) or anti-hsp90 (heat-shock protein 90) (0.05 μg/ml; 610419; BD Biosciences) antibody.

Biochemical characterization

All caspases were active site-titrated as previously described using Z-Val-Ala-DL-Asp-CH2F or p35 (a pan-caspase inhibitor from baculovirus) as the titrant and Ac-DEVD-afc (N-acetyl-Asp-Glu-Val-Asp-afc) (for caspase 7), Ac-IETD-afc (N-acetyl-Ile-Glu-Thr-Asp-afc) (caspase 8) or Ac-LEHD-afc (N-acetyl-Leu-Glu-His-Asp-afc) (caspase 9) as the substrate [19]. Kinetic parameters were determined by non-linear regression at 37 °C using the preferred substrate of each caspase [26] in standard caspase buffer [10 mM Pipes, pH 7.2, 100 mM NaCl, 10% (w/v) sucrose, 0.1% (w/v) Chaps, 1 mM EDTA and 10 mM DTT (dithiothreitol)] for executioner caspases or high salt buffer (50 mM Hepes, pH 7.4, 1 M sodium citrate, 50 mM NaCl, 0.01% (w/v) Chaps and 10 mM DTT) for initiator caspases [27]. In some instances, parameters were measured in pseudo-first-order kinetic conditions ([S]≪Km) when Km values were outside the workable substrate range (Km>200 μM). In all cases, initiator caspases at concentrations above 100 nM were pre-incubated for 1 h in high salt buffer at 37 °C to promote dimerization and maximal activity of the peptidases. Fluorescence was acquired using an M1000 Infinite platereader (Tecan) or an FLX800 (Bio-Tek Instruments). Results were analysed with GraphPad Prism 5 software.

Caspase cleavage assay by initiator caspases

Cleavage assays of pro-caspase 7 variants by initiator caspases were performed as follows. Initiator caspases (>100 nM) were preactivated at 37 °C for 1 h in high salt buffer prior to performing the assays for maximizing caspase activity [27]. For gel analyses, a range of concentrations of activated initiator caspase were incubated at 37 °C with a fixed amount of pro-caspase 7 (see the Figure legends for other assay conditions). After a fixed period of time, reactions were stopped by sedimenting proteins with trichloroacetic acid/acetone, resuspended in SDS/PAGE loading buffer and resolved on a gradient gel. Estimation of kcat/Km was obtained using the relationship p = (1-e-kcat·[E]/Km·t) [19] in which p is the fraction of pro-caspase 7 that is cleaved at time t (in units of seconds) and [E] is the initiator caspase concentration. The point at which 50% of the zymogen is processed (p = 0.5) was used to estimate kcat/Km. For pre-steady state activation assays, a fixed concentration of activated initiator caspase was incubated with various concentrations of pro-caspase 7 and the amidolytic activity on Ac-DEVD-pNA (N-acetyl-Asp-Glu-Val-Asp-p-nitroanilide) chromogenic substrate was measured continuously using a NanoQuant Infinite M200 or Infinite M1000 platereader (Tecan). Results were analysed by non-linear regression using GraphPad Prism 5 software as previously described [8,10,27] to determine the rates of activation. Importantly, site 2 mutations did not alter kinetic parameters of the active enzyme significantly (see Supplementary Table S1 at http://www.bioscirep.org/bsr/031/bsr0310283add.htm), which allowed us to easily compare these mutants. However, mutation of site 1 to LETD/LEHD caused the active enzyme to behave erratically precluding their thorough analysis by pre-steady state analyses.

SDS/PAGE and immunoblotting

PAGE was performed using gradient (8–18%, w/v) acrylamide gels in the 2-amino-2-methylpropan-1,3-diol/glycine buffer system [28]. Gels were either stained using GelCode Blue (Pierce) or proteins were transferred on to an Immobilon-P membrane (Millipore) using 10 mM Caps (pH 11) and 10% (v/v) methanol for 45 min at 0.4 A and at 4 °C. Immunoblotting was carried out using standard protocols with anti-His antibody (0.1 μg/ml; 34660; Qiagen) and were revealed with WestPico SuperSignal (Pierce).

RESULTS

Processing of caspase 7 activation site mutants

The cleavage site and its location within a substrate are two essential factors directing proteolysis efficacy [29]. At a given location in a substrate, the cleavage site sequence becomes critical. Based on results from the cleavage of short peptidic substrates [26], it appeared that caspase 7's primary activation site sequence (site 1, IQAD198↓S, caspase 7 residue numbering; Figure 1A) and the secondary site (site 2, NDTD206↓A) within the IDC are suboptimal recognition motifs for initiator caspases. To better understand the basis of this apparent suboptimal sequence usage, we generated a series of caspase 7 activation site variants by mutating individual motifs to the preferred sequence recognized by initiator caspase 9 (LEHD motif) or caspase 8 (LETD motif) as determined using short peptidic substrates [26]. We first analysed the ability of these mutants to auto-activate and characterized their proteolytic activity. Recombinant caspases were produced in bacteria as His-tagged proteins and purified by immobilized metal-affinity chromatography. To prevent cleavage of the alternate site while studying a particular site (i.e. site 1 versus site 2), the critical aspartate residues were mutated to an alanine residue, i.e. IQAD→IQAA (site 1) or NDTD→NDTA (site 2). Throughout the study, cleavage site mutants are named with the sequence of both sites as site 1/site 2 pairs; thus the WT (wild-type) enzyme would be termed caspase 7 IQAD/NDTD (P1 cleavage residues are underlined). Figure 1(B) shows that E. coli expression of site 1 mutants (caspase 7 LEHD/NDTA or caspase 7 LETD/NDTA) resulted in auto-processed enzymes displaying the typical large and small subunits. We noticed that a standard 8 h expression was insufficient to produce fully processed caspase 7 LEHD/NDTA (Figure 1B) when compared with WT caspase 7, suggesting that either the site is inadequate for caspase 7 self-activation or that the resulting enzyme is less active. To test the latter we measured the kinetic parameters of the processed enzyme. Interestingly, both caspase 7 site 1 mutants did not exhibit typical Michaelis–Menten behaviour (results not shown). Therefore we resorted to pseudo-first-order assays ([S]≪Km) to estimate activity. In such conditions, kcat/Kmv/[S][E] and the specific activity is directly related to the substrate hydrolysis v and inversely related to the substrate and enzyme concentrations. Using this method, we obtained specific activities that are 17–64 times less active than caspase 7 IQAD/NDTA (Table 1). These weak activities likely explain why we were unable to active site-titrate these enzymes.

Expression of caspase 7 cleavage site mutants

Figure 1
Expression of caspase 7 cleavage site mutants

(A) Schematic representation of caspase 7 and its IDC. The N-terminal peptide (N), the large and small subunits, and the IDC (sequence shown) are displayed. The star represents the catalytic cysteine. Cleavages of the IDC occur C-terminal to the motifs shown in boldface. (B) SDS/PAGE analysis of caspase 7 cleavage site mutants. It is noteworthy that all proteins lost the N-terminal peptide due to the activity of the enzyme although some are purified as the zymogen form. The underlined residues identify the processed cleavage sites.

Figure 1
Expression of caspase 7 cleavage site mutants

(A) Schematic representation of caspase 7 and its IDC. The N-terminal peptide (N), the large and small subunits, and the IDC (sequence shown) are displayed. The star represents the catalytic cysteine. Cleavages of the IDC occur C-terminal to the motifs shown in boldface. (B) SDS/PAGE analysis of caspase 7 cleavage site mutants. It is noteworthy that all proteins lost the N-terminal peptide due to the activity of the enzyme although some are purified as the zymogen form. The underlined residues identify the processed cleavage sites.

Table 1
Kinetic parameters of caspases 7 and 8 cleavage site mutants

Results are means±S.D. from two to five independent determinations. n.d., not determined because the enzyme did not display Michaelis–Menten behaviour at the substrate concentration range we used (see text); for these, kcat/Km were estimated using pseudo-first-order kinetics ([S]≪Km).

CaspaseKm (μM)kcat (s−1)kcat/Km (M−1·s−1)
Caspase 7 (Ac-DEVD-Afc)    
 WT 12.8±0.9 2.7±0.2 2.1±0.3 × 105 
 IQAD/NDTA 22.6±3.6 2.3±0.1 1.0±0.2 × 105 
 DEVD/NDTA n.d. n.d. 1.6 × 104 
 LETD/NDTA n.d. n.d. 1566 
 LEHD/NDTA n.d. n.d. 5773 
 DQAD/NDTA n.d. n.d. 1690 
 DEAD/NDTA n.d. n.d. 288 
 IEAD/NDTA 24.5±0.6 3.4±0.1 1.4±0.1 × 105 
 IQAA/DEVD 58.8±5.1 1.3±0.1 2.1±0.3 × 104 
Caspase 8-activated caspase 7 (Ac-DEVD-Afc)    
 IQAA/NDTD 13.8±0.7 4.4±0.1 3.1±0.2 × 105 
 IQAA/LETD 13.3±1.1 3.6±0.1 2.7±0.3 × 105 
 IQAA/LEHD 14.6±0.8 3.2±0.1 2.2±0.2 × 105 
Caspase 7 (Ac-DEVD-Afc)    
 K212M* n.d. n.d. 2185 
 K212E n.d. n.d. <100 
 K212R 59.2±4.2 3.6±0.1 6.2±0.7 × 104 
 I213A n.d. n.d. 3183 
 I213M 56.5±5.4 2.6±0.1 4.7±0.6 × 104 
 I213E n.d. n.d. <100 
 I213L 78.2±6.5 3.8±0.1 4.7±0.5 × 104 
Caspase 7 insertion mutants (Ac-DEVD-Afc)    
 IQAD/NDTA+6 36.6±2.5 2.9±0.1 7.4±0.6 × 104 
 IQAA/NDTA+6 Inactive   
 IQAD/NDTA+14 27.7±2.1 6.0±0.1 2.2±0.2 × 105 
 IQAA/NDTA+14 Inactive   
Caspase 8 mutants (Ac-LETD-Afc)    
 WT 1.3±0.1 2.0±0.2 1.0±0.0 × 106 
 VETD/LEMA 0.3±0.0 0.5±0.0 1.9±0.2 × 106 
 VETA/LEMA 21.3±1.1 0.3±0.0 1.5±0.1 × 104 
 VETD/LEMA-2 8.6±1.4 5.8±0.21 6.7±1.3 × 106 
 VETA/LEMA-2 43.7±4.1 0.09±0.01 2022±388 
 VETD/LEMA-4 5.9±0.5 3.4±0.1 5.8±0.6 × 105 
 VETA/LEMA-4 Inactive   
CaspaseKm (μM)kcat (s−1)kcat/Km (M−1·s−1)
Caspase 7 (Ac-DEVD-Afc)    
 WT 12.8±0.9 2.7±0.2 2.1±0.3 × 105 
 IQAD/NDTA 22.6±3.6 2.3±0.1 1.0±0.2 × 105 
 DEVD/NDTA n.d. n.d. 1.6 × 104 
 LETD/NDTA n.d. n.d. 1566 
 LEHD/NDTA n.d. n.d. 5773 
 DQAD/NDTA n.d. n.d. 1690 
 DEAD/NDTA n.d. n.d. 288 
 IEAD/NDTA 24.5±0.6 3.4±0.1 1.4±0.1 × 105 
 IQAA/DEVD 58.8±5.1 1.3±0.1 2.1±0.3 × 104 
Caspase 8-activated caspase 7 (Ac-DEVD-Afc)    
 IQAA/NDTD 13.8±0.7 4.4±0.1 3.1±0.2 × 105 
 IQAA/LETD 13.3±1.1 3.6±0.1 2.7±0.3 × 105 
 IQAA/LEHD 14.6±0.8 3.2±0.1 2.2±0.2 × 105 
Caspase 7 (Ac-DEVD-Afc)    
 K212M* n.d. n.d. 2185 
 K212E n.d. n.d. <100 
 K212R 59.2±4.2 3.6±0.1 6.2±0.7 × 104 
 I213A n.d. n.d. 3183 
 I213M 56.5±5.4 2.6±0.1 4.7±0.6 × 104 
 I213E n.d. n.d. <100 
 I213L 78.2±6.5 3.8±0.1 4.7±0.5 × 104 
Caspase 7 insertion mutants (Ac-DEVD-Afc)    
 IQAD/NDTA+6 36.6±2.5 2.9±0.1 7.4±0.6 × 104 
 IQAA/NDTA+6 Inactive   
 IQAD/NDTA+14 27.7±2.1 6.0±0.1 2.2±0.2 × 105 
 IQAA/NDTA+14 Inactive   
Caspase 8 mutants (Ac-LETD-Afc)    
 WT 1.3±0.1 2.0±0.2 1.0±0.0 × 106 
 VETD/LEMA 0.3±0.0 0.5±0.0 1.9±0.2 × 106 
 VETA/LEMA 21.3±1.1 0.3±0.0 1.5±0.1 × 104 
 VETD/LEMA-2 8.6±1.4 5.8±0.21 6.7±1.3 × 106 
 VETA/LEMA-2 43.7±4.1 0.09±0.01 2022±388 
 VETD/LEMA-4 5.9±0.5 3.4±0.1 5.8±0.6 × 105 
 VETA/LEMA-4 Inactive   
*

This caspase 7 mutant was activated using excess caspase 8 during purification.

In a second series of mutants, we modified activation site 2 to the motifs preferred by caspase 8 or caspase 9 respectively. Unlike site 1 mutants, site 2 mutants (caspase 7 IQAA/LEHD and caspase 7 IQAA/LETD) did not auto-activate in bacteria (Figure 1B). To verify whether modifications to site 2 affected proteolytic activity, we determined the kinetic parameters of both mutants following their processing with caspase 8 and found the kcat and Km values to be similar to WT caspase 7 (Table 1). Both mutants retained typical Michaelis–Menten behaviour after processing with caspase 8 as compared with WT site 2 caspase 7.

Caspase 7 primary activation site is close to optimal for activation in trans by initiator caspases

Having characterized caspase 7 cleavage site mutants, weaddressed the ability of full-length and truncated initiator caspases to activate them. An important prerequisite for the use of a truncated caspase 9 is to demonstrate that initiator CARD do not influence substrate preference. Therefore we tested full-length and ΔCARD caspase 9 against a combinatorial peptidic substrate library covering the P4–P2 cleavage site residues (see Supplementary Experimental section and Figure S1 at http://www.bioscirep.org/bsr/031/bsr0310283add.htm) and found that both enzymes have the same substrate preference. Furthermore, similar experiments performed using full-length and truncated caspase 8 gave similar results [29a]. We first tested the capacity of full-length caspase 9 and a truncated version (lacking the CARD) to activate pro-caspase 7 (Figure 2A). In our assays, serial dilutions of activators were incubated with a fixed concentration of pro-caspase 7 IQAD/NDTA or LEHD/NDTA. Cleaved fragments are then resolved on SDS/PAGE. Figure 2(A) (top left) shows that processing of site 1 is not improved when the optimized LEHD motif is used compared with the IQAD sequence, which corresponds to an activation at WT site 1 [11]. This is also true for the activation by ΔCARD caspase 9, albeit the activation of all pro-caspases is better than with the full-length activator. Indeed, the lowest concentration used (35 nM) resulted in full cleavage of the pro-caspase 7. The discrepancy between the two caspase 9 activators remains a mystery at present. However, by varying the incubation time of the assays, we estimated that pro-caspase 7 mutant LEHD/NDTA was activated by ΔCARD caspase 9 with a rate that is twice as high as that of the WT site 1 (results not shown). Similarly, pro-caspase 7 LETD/NDTA is activated more efficiently (~2-fold) by caspase 8 as demonstrated by the fact that less activator was required for cleaving 50% of the zymogen (119 nM compared with 53 nM; Supplementary Figure S2 at http://www.bioscirep.org/bsr/031/bsr0310283add.htm). These results demonstrate that activation site 1 is close to optimal for activation by the initiators.

Activation of pro-caspase 7 by initiator caspase 9 can be improved at site 2

Figure 2
Activation of pro-caspase 7 by initiator caspase 9 can be improved at site 2

(A) SDS/PAGE analysis of activation kinetics of various caspase 7 mutants with full-length caspase 9 (left panels) or truncated (ΔCARD) caspase 9 (right panels). Assay conditions were as follows: left, 400 nM of the indicated pro-caspase 7, full-length caspase 9 (1/2 serial dilution from 500 nM down to 8 nM), 2 h at 37°C; right, 200 nM of the indicated pro-caspase 7, ΔCARD caspase 9 (2/3 serial dilution from 400 nM down to 35 nM), 12 min at 37°C. Trichloroacetic acid/acetone-precipitated proteins were resolved on SDS/PAGE and stained. A sample without initiator caspase (pro-casp-7) or with maximum concentration of the initiator (casp-9 or ΔCARD-casp-9) used was included as control. The full-length zymogen (empty arrowhead), the form lacking the N-terminal peptide (−N) and the cleaved large subunit (ls) are indicated. Note that most pro-caspase 7 preparations lack the N-peptide because it is removed extremely rapidly. Arrows indicate the midpoint at which 50% of the pro-caspase 7 is cleaved. Apparent kcat/Km was obtained as described in the Experimental section. Note that most values are underestimated because pro-caspase 7 concentrations are above the estimated Km (see Table 2). Molecular mass markers are 97.4 (full-length caspase 9 panels only), 66.2, 45 (*), 31, 21.5 and 14.4 kDa. (B) Activation kinetics by full-length caspase 9 (left), truncated caspase 9 (middle) and truncated caspase 8 (right) of pro-caspase 7 site 2 mutants. The concentration range (2/3 serial dilution from 150 nM down to 3.9 nM) of the indicated site 2 mutants was incubated with the indicated initiator caspases in the presence of 400 μM of the chromogenic substrate Ac-DEVD-pNA. The top graphs depict the amidolytic activity monitored over time at 405 nm for the 67 nM pro-caspase 7 samples. Values presented are background-corrected for the activity of the initiator caspase. Rates of pro-caspase 7 activation were estimated as described in the Experimental section and plotted against the pro-caspase 7 concentrations in the bottom graphs. Solid lines were obtained by non-linear regression of data using the Michaelis–Menten equation v = Vmax,app.[S]/(Km,app.+[S]), whereas broken lines are linear regressions and represent the particular case in which [S]≪Km and apparent kcat/Km is obtained from the slope over activator concentration. Kinetic parameters are reported in Table 2. Note the scale differences of y-axes. In bottom graphs, dotted lines indicate the 67 nM pro-caspase 7 concentration values extracted from the top panels data. Activator concentrations are: full-length caspase 9, 100 nM; truncated caspase 9, 25 nM; truncated caspase 8, 25 nM.

Figure 2
Activation of pro-caspase 7 by initiator caspase 9 can be improved at site 2

(A) SDS/PAGE analysis of activation kinetics of various caspase 7 mutants with full-length caspase 9 (left panels) or truncated (ΔCARD) caspase 9 (right panels). Assay conditions were as follows: left, 400 nM of the indicated pro-caspase 7, full-length caspase 9 (1/2 serial dilution from 500 nM down to 8 nM), 2 h at 37°C; right, 200 nM of the indicated pro-caspase 7, ΔCARD caspase 9 (2/3 serial dilution from 400 nM down to 35 nM), 12 min at 37°C. Trichloroacetic acid/acetone-precipitated proteins were resolved on SDS/PAGE and stained. A sample without initiator caspase (pro-casp-7) or with maximum concentration of the initiator (casp-9 or ΔCARD-casp-9) used was included as control. The full-length zymogen (empty arrowhead), the form lacking the N-terminal peptide (−N) and the cleaved large subunit (ls) are indicated. Note that most pro-caspase 7 preparations lack the N-peptide because it is removed extremely rapidly. Arrows indicate the midpoint at which 50% of the pro-caspase 7 is cleaved. Apparent kcat/Km was obtained as described in the Experimental section. Note that most values are underestimated because pro-caspase 7 concentrations are above the estimated Km (see Table 2). Molecular mass markers are 97.4 (full-length caspase 9 panels only), 66.2, 45 (*), 31, 21.5 and 14.4 kDa. (B) Activation kinetics by full-length caspase 9 (left), truncated caspase 9 (middle) and truncated caspase 8 (right) of pro-caspase 7 site 2 mutants. The concentration range (2/3 serial dilution from 150 nM down to 3.9 nM) of the indicated site 2 mutants was incubated with the indicated initiator caspases in the presence of 400 μM of the chromogenic substrate Ac-DEVD-pNA. The top graphs depict the amidolytic activity monitored over time at 405 nm for the 67 nM pro-caspase 7 samples. Values presented are background-corrected for the activity of the initiator caspase. Rates of pro-caspase 7 activation were estimated as described in the Experimental section and plotted against the pro-caspase 7 concentrations in the bottom graphs. Solid lines were obtained by non-linear regression of data using the Michaelis–Menten equation v = Vmax,app.[S]/(Km,app.+[S]), whereas broken lines are linear regressions and represent the particular case in which [S]≪Km and apparent kcat/Km is obtained from the slope over activator concentration. Kinetic parameters are reported in Table 2. Note the scale differences of y-axes. In bottom graphs, dotted lines indicate the 67 nM pro-caspase 7 concentration values extracted from the top panels data. Activator concentrations are: full-length caspase 9, 100 nM; truncated caspase 9, 25 nM; truncated caspase 8, 25 nM.

Site 2 can be improved for activation by initiator caspases

We also performed on-gel activation assays with pro-caspase 7 site 2 mutants. Compared with pro-caspase 7 IQAA/NDTD (i.e. WT site 2), pro-caspase 7 IQAA/LEHD has a ~5-fold better rate of caspase 9-mediated activation and ~4-fold increase using ΔCARD caspase 9 as an activator (Figure 2A). Because our gel-based assays are performed with a pro-caspase 7 concentration that may exceed the Km value, it is possible that activation assays were not performed in pseudo-first-order conditions, which would greatly underestimate the calculated kcat/Km. Therefore we performed pre-steady state assays to better characterize the activation at site 2 because we observed an important improvement in activation when this site is mutated. In these assays, serial dilutions of pro-caspase 7 are incubated with a fixed amount of activator. The ensuing activation is then monitored using a caspase 7 substrate (Figure 2B, top panels). Using an equation describing the rate of activation of a protein in a coupled reaction, we determined activation rates that we plotted against pro-caspase 7 (Figure 2B, bottom panels) and calculated Km, kcat and kcat/Km using non-linear regression (Table 2). For all three activator caspases tested, we observed a 10–36-fold improvement in specific activity (kcat/Km) when site 2 is modified to the preferred recognition motif of either caspase 9 or caspase 8. Interestingly and irrespective of the initiator caspase studied, we obtained activation rates at the mutated site 2 that are similar to the rates obtained for WT site 1. Another finding is that for both forms of caspase 9 we tested, Km values for WT site 1 (IQAD) are assumed to be high because we were unable to ascribe a Michaelis–Menten behaviour to the activation of pro-caspase 7 IQAD/NDTA. Finally, a striking observation was that very low kcat values were obtained. The reason for that is unknown, but our kcat values are consistent with previously published kcat/Km [4] (~104 M−1·s−1) and Km values [30] (~100 nM). Taken together, we conclude that site 2 can be improved for activation.

Table 2
Activation kinetics of pro-caspase 7 cleavage site 2 mutants

Means±S.D. for two or three independent datasets of 4–10 samples are presented. n.d., not determined because [S]≪Km. The parameters kcat and Km could not be determined independently because Km≫[S].

ActivatorPro-caspase 7Km, app. (nM)kcat, app. (s−1)kcat, app./Km, app. (×104 M−1·s−1)
Caspase 9 IQAD/NDTA n.d. 0.34  
 IQAA/NDTD n.d. <0.01  
 IQAA/LEHD 180±10 0.0007±0.0001 0.36 
ΔCARD caspase 9 IQAD/NDTA n.d. 29.5  
 IQAA/NDTD 76.0±8.2 0.0020±0.0001 2.6 
 IQAA/LEHD 20.0±14.5 0.005±0.002 25 
ΔDEDs caspase 8 IQAD/NDTA 38.9±6.7 0.0019±0.0002 4.9 
 IQAA/NDTD n.d. <0.3  
 IQAA/LETD 17.5±7.3 0.0012±0.0004 7.3 
ActivatorPro-caspase 7Km, app. (nM)kcat, app. (s−1)kcat, app./Km, app. (×104 M−1·s−1)
Caspase 9 IQAD/NDTA n.d. 0.34  
 IQAA/NDTD n.d. <0.01  
 IQAA/LEHD 180±10 0.0007±0.0001 0.36 
ΔCARD caspase 9 IQAD/NDTA n.d. 29.5  
 IQAA/NDTD 76.0±8.2 0.0020±0.0001 2.6 
 IQAA/LEHD 20.0±14.5 0.005±0.002 25 
ΔDEDs caspase 8 IQAD/NDTA 38.9±6.7 0.0019±0.0002 4.9 
 IQAA/NDTD n.d. <0.3  
 IQAA/LETD 17.5±7.3 0.0012±0.0004 7.3 

Both activation sites can be improved in complete apoptotic paradigms

To further demonstrate that activation sites are not optimal, we used apoptotic models. First, we employed a cell-free system that recapitulates the intrinsic apoptotic pathway led by caspase 9 that we trigger with cyt c and dATP [24]. Extracts were prepared from cells expressing the FLAG-tagged catalytic mutant of the cleavage site mutants (named ‘reporter’). By doing so, each reporter is incorporated into the apoptotic framework as a bystander substrate of the initiator, thus reporting activation. Importantly, the reporter has no observable effect as indicated by the rate of Ac-DEVD-pNA hydrolysis (caspase 3/7 activity) in the cell extract compared to that observed with the extract from empty plasmid-transfected cells (Figure 3A, graph inset). The generation of cleaved FLAG-tagged reporter-carrying WT sites (Figure 3A, middle panel) appears within 10 min of cyt c/dATP addition followed at 30 min by the cleavage of the death substrate hsp90 co-chaperone p23 [31]. Similar profiles were obtained with the reporter with IQAD/NDTA or LEHD/NDTA sites (Figure 3, left panels) demonstrating that in agreement with our activation kinetics (Table 2), site 1 is relatively optimal for caspase 9-mediated activation. In contrast, no activation was observed at WT site 2 (IQAA/NDTD), which is what we observed in assays with recombinant proteins (Figure 2). Interestingly, its replacement with the LEHD motif allows some limited processing (Figure 3A, right panels). In all cases, p23 processing was not altered, suggesting again that ectopic expression of caspase 7 activation reporter does not interfere with the apoptotic machinery.

Activation of pro-caspase 7 can be improved in a complete apoptotic system

Figure 3
Activation of pro-caspase 7 can be improved in a complete apoptotic system

(A) Site 1 is optimal for caspase 9 activation but site 2 can be ameliorated. Immunoblot analyses of cyt c/dATP-activated extract from HEK-293Ad cells expressing the various pro-caspase 7 cleavage site mutants. After the indicated period of time, samples were boiled in gel loading buffer and analysed using anti-FLAG, p23 (caspase substrate) and hsp90 (loading control). Because pro-caspase 7 has a C-terminal FLAG tag, the anti-FLAG antibody detects the full-length zymogen (empty arrowhead), the form lacking the N-terminal peptide (-N) and the cleaved small subunit (ss). The inset shows similar assays analysed for caspase activation using the chromogenic substrate Ac-DEVD-pNA. Note that no difference is observed between cyt c/dATP-activated samples in which pro-caspase 7 cleavage site mutants are individually expressed. The asterisk denotes the group of traces obtained in the absence of cyt c/dATP for each corresponding extract. The underlined residues identify the processed cleavage sites. (B) Both sites can be improved for activation by caspase 8. HEK-293Ad cells transfected to express the indicated pro-caspase 7 activation site mutant were challenged without (first lane) or with increasing doses of recombinant TRAIL ligand (0.5, 1, 2, 5, 10, 20, 50 or 100 ng/ml). Processing of FLAG-tagged pro-caspase 7 was monitored by immunoblotting with an anti-FLAG antibody at 12 h after TRAIL addition. The underlined residues identify the processed cleavage sites.

Figure 3
Activation of pro-caspase 7 can be improved in a complete apoptotic system

(A) Site 1 is optimal for caspase 9 activation but site 2 can be ameliorated. Immunoblot analyses of cyt c/dATP-activated extract from HEK-293Ad cells expressing the various pro-caspase 7 cleavage site mutants. After the indicated period of time, samples were boiled in gel loading buffer and analysed using anti-FLAG, p23 (caspase substrate) and hsp90 (loading control). Because pro-caspase 7 has a C-terminal FLAG tag, the anti-FLAG antibody detects the full-length zymogen (empty arrowhead), the form lacking the N-terminal peptide (-N) and the cleaved small subunit (ss). The inset shows similar assays analysed for caspase activation using the chromogenic substrate Ac-DEVD-pNA. Note that no difference is observed between cyt c/dATP-activated samples in which pro-caspase 7 cleavage site mutants are individually expressed. The asterisk denotes the group of traces obtained in the absence of cyt c/dATP for each corresponding extract. The underlined residues identify the processed cleavage sites. (B) Both sites can be improved for activation by caspase 8. HEK-293Ad cells transfected to express the indicated pro-caspase 7 activation site mutant were challenged without (first lane) or with increasing doses of recombinant TRAIL ligand (0.5, 1, 2, 5, 10, 20, 50 or 100 ng/ml). Processing of FLAG-tagged pro-caspase 7 was monitored by immunoblotting with an anti-FLAG antibody at 12 h after TRAIL addition. The underlined residues identify the processed cleavage sites.

Employing a different apoptotic model that uses the death ligand TRAIL (tumour-necrosis-factor-related apoptosis-inducing ligand) to activate the extrinsic pathway (caspase 8/10), we showed that both cleavage sites can be improved by using the LETD motif (Figure 3B). Indeed, HEK-293Ad cells transfected with a plasmid encoding the reporter with LETD/NDTA or IQAA/LETD sites showed better processing than with the respective WT site. Similarly to caspase 9, these results also mirrored our in vitro activation assays. Taken together, these results demonstrate that caspase 7 activation sites can be improved by substituting them with the preferred motif of the respective initiator caspase.

Self-activation of caspase 7 and cleavage site location

Caspase 7 does not self-activate at the endogenous concentration, whereas the high concentration of enzyme achieved during E. coli expression allows self-activation. To better understand the mechanism by which caspase 7 auto-activates, we first inserted an optimal caspase 7 cleavage sequence (DEVD) at site 1. Like WT caspase 7, the introduction of a DEVD motif (caspase 7 DEVD/NDTD or caspase 7 DEVD/NDTA) resulted in a fully processed enzyme (Figure 4A). Unexpectedly, tracking of caspase 7 DEVD/NDTA processing during its expression revealed that the zymogen form could easily be produced and that most of the processing occurs after 2 h of expression, like WT caspase 7 (Figure 4B). This is likely due to the enzyme's weak intrinsic activity (Table 1) similar to that observed for the LETD/LEHD site 1 mutants. Indeed, caspase 7 DEVD/NDTA did not display typical Michaelis–Menten behaviour, suggesting that the IQAD→DEVD substitution altered some fundamental properties of the enzyme similarly to site 1 mutation to LETD/LEHD. Further, by deconstructing the DEVD motif (IQAD→DQAD/DEAD), we realized that the P4 residue Ile195 is critical in supporting robust enzyme activity, whereas mutation of P3 Gln196 alone (IQAD→IEAD) had a minimal impact on the activity (Table 1). We next asked why Ile195 is so important as suggested by our results and recent work by Witkowski and Hardy [32]. One reason for the loss of activity of site 1 mutants with altered Ile196 may reside in the intricate contacts between this residue of one catalytic unit of the dimer with the other seen in the crystal structure of inhibitor-bound caspase 7 [33]. In the crystal structure of inhibitor-bound caspase 7, Ile195 is in close proximity to Lys212′ and Ile213′ of the other catalytic unit of the dimer (often called the L2–L2′ loop bundle). To further test this observation, these residues were mutated, and enzyme activity was assessed. As expected, disruptive mutation of Lys212 to a methionine/glutamine residue or Ile213 to an alanine/glutamine residue significantly affected the activity of the caspase (Table 1). In support of the perturbation of caspase activity by non-conservative mutations, three of these mutants required longer expression times or transactivation by caspase 8 to obtain processed enzymes. Conversely, conservative amino acid substitutions (Lys212→arginine residue or Ile213→methionine/leucine residue) preserved caspase 7 activity and proper activation time (Table 1).

Auto-activation of caspase 7 mutants

Figure 4
Auto-activation of caspase 7 mutants

(A) SDS/PAGE analysis of purified caspase 7 cleavage site mutants. zym., zymogen. (B) Self-activation at site 1. The indicated caspase 7 were expressed in 50 ml cultures for the indicated period of time, purified and analysed by SDS/PAGE. Arrowheads indicate the transition point from the zymogen to the active form; this transition is absent from caspase 7 IQAA/NDTD that cannot self-activate [18]. (C) Swapping activation site sequences. The indicated caspase 7s were expressed for 8 h, purified and analysed by SDS/PAGE. (D) Self-activation at site 2. Lysates of E. coli expressing the indicated caspase 7s for 30 min or 8 h were analysed by immunoblotting using an anti-His tag antibody. Note that only processed caspase is detected for caspase 7 IQAA/DEVD. WT and caspase 7 IQAA/NDTD are presented as controls for the fully processed and zymogen forms respectively. The underlined residues identify the processed cleavage sites.

Figure 4
Auto-activation of caspase 7 mutants

(A) SDS/PAGE analysis of purified caspase 7 cleavage site mutants. zym., zymogen. (B) Self-activation at site 1. The indicated caspase 7 were expressed in 50 ml cultures for the indicated period of time, purified and analysed by SDS/PAGE. Arrowheads indicate the transition point from the zymogen to the active form; this transition is absent from caspase 7 IQAA/NDTD that cannot self-activate [18]. (C) Swapping activation site sequences. The indicated caspase 7s were expressed for 8 h, purified and analysed by SDS/PAGE. (D) Self-activation at site 2. Lysates of E. coli expressing the indicated caspase 7s for 30 min or 8 h were analysed by immunoblotting using an anti-His tag antibody. Note that only processed caspase is detected for caspase 7 IQAA/DEVD. WT and caspase 7 IQAA/NDTD are presented as controls for the fully processed and zymogen forms respectively. The underlined residues identify the processed cleavage sites.

We showed that despite the improvement of site 2 by introducing preferred initiator caspase recognition motifs, caspase 7's site 1 was superior for activation, suggesting, as proposed by Timmer et al. [29], that location of the site within the IDC is also important. To address this, we first transferred the site 1 motif (in addition to the P1′ serine) to site 2 generating caspase 7 IQAA/IQAD-S. Interestingly, this mutant was expressed only as a zymogen (Figure 4C). Moreover, transplanting site 2 motif (NDTD) to site 1 failed to produce a caspase able to auto-activate, even after 24 h (results not shown). This demonstrates that location of the activation site within the IDC is important. We then replaced site 2 by the preferred caspase 7 motif DEVD, thus generating caspase 7 IQAA/DEVD. Surprisingly, we obtained this protein in a fully processed form (Figure 1A), demonstrating that it is possible to cleave site 2 first. Nevertheless, this mutant displayed a 10-fold decrease in specific activity compared to WT caspase 7 (Table 1). Despite the reduced activity of caspase 7 IQAA/DEVD, this caspase was activated so effectively that even a 25 min expression immediately followed by immunoblotting failed to show the zymogen form of the enzyme (Figure 4D). Altogether, these results show that site 1 is better positioned than site 2 for activation in trans, but that the propensity for processing of site 2 can be improved significantly.

The IDC length is a crucial determinant

The longer length of the IDC is postulated to be a crucial determinant in the ability of initiator caspases to display activity without cleavage [17]. Executioner caspases 3 and 7 have a shorter IDC and, unlike the initiators, which do not require cleavage for activity [5,6,20], proteolysis is the driving force of executioners' activation. Therefore, not only is the primary structure of caspase 7's IDC key to the activation process but also its length may be important. To address this issue, we extended the IDC from 25 residues to 31 (the length of caspase 8's IDC) with the hope of creating an active executioner without cleavage of the IDC. To do so, and guided by the studies described above, we inserted three glycine–serine pairs N-terminal of site 2 of the double cleavage site mutant to generate caspase 7 IQAA/NDTA+6 (Figure 5A). Figure 5(B) shows that this mutant did not self-process during bacterial expression unlike the cleavable site 1 version of this construct (IQAD/NDTA+6). We also extended the IDC by adding seven glycine-serine pairs but were unsuccessful in obtaining an active enzyme (Table 1, and results not shown). In a second strategy, we constructed two mutants by replacing the IDC of caspase 7 with that of caspase 8: (a) DGIQADSGPINDTDANPRYKIPV→KGIPVETDSEEQPYLEMDLSSPQTRYIPD (linker casp-8a) to preserve Asp192 [10] or (b) QADSGPINDTDANPRYKIPV→PVETDSEEQPYLEMDLSSPQTRYIPD (linker casp-8b) to also preserve the critical Ile195 of caspase 7 (broken lines in Figure 5A). However, these mutants did not undergo auto-activation nor did they exhibit any proteolytic activity (Figure 5B and results not shown), but processing by caspase 8 did activate caspase 7/linker casp-8b (results not shown). A corollary to our hypothesis is that the length of the initiator caspase 8 IDC is optimal to support caspase 8 activity without cleavage of the linker. Therefore, as a validation of our strategy, we performed the reverse approach with caspase 8, and attempted to create a caspase 8 requiring proteolysis for activity. To do so, two (Glu377-Gln, caspase 8 numbering system) or four residues (Glu377-Gln-Pro-Tyr) were deleted from caspase 8's IDC, and we compared the activity of these deletion mutants possessing a cleavable (caspase 8 VETD/LEMA) or uncleavable (caspase 8 VETA/LEMA) IDC (Figure 5A). Removal of 4, but not of 2, residues resulted in a caspase 8 requiring cleavage for activity (Figure 5B; Table 1). Indeed, caspase 8 VETA/LEMA-4 was inactive in high salt buffer, whereas caspase 8 VETD/VETA-4 was cleaved (Figure 5B) and fully active compared with WT caspase 8 in the same conditions (Table 1). These results demonstrate that the length of the IDC is critical in the activation process of caspases.

Changing the length of the IDC

Figure 5
Changing the length of the IDC

(A) Schematic representation of caspase 8 and its IDC. DEDs, the large and small subunits, and the IDC (sequence shown) are displayed. The star identifies the catalytic cysteine. Cleavage of the IDC occurs C-terminal to the motifs shown in boldface at two sites (Asp374 or Asp384, caspase 8 numbering system). The six-residue insertion (in caspase 7) and 2/4 amino acid deletions in caspase 8 are indicated under the sequence. The broken lines show which parts of the IDC were swapped between caspase 8 and caspase 7. (B) SDS/PAGE analysis of caspase 7 insertion and caspase 8 deletion mutants after 8 or 6 h of expression respectively. Caspase 7/linker casp-8b is shown. The underlined residues identify the processed cleavage sites.

Figure 5
Changing the length of the IDC

(A) Schematic representation of caspase 8 and its IDC. DEDs, the large and small subunits, and the IDC (sequence shown) are displayed. The star identifies the catalytic cysteine. Cleavage of the IDC occurs C-terminal to the motifs shown in boldface at two sites (Asp374 or Asp384, caspase 8 numbering system). The six-residue insertion (in caspase 7) and 2/4 amino acid deletions in caspase 8 are indicated under the sequence. The broken lines show which parts of the IDC were swapped between caspase 8 and caspase 7. (B) SDS/PAGE analysis of caspase 7 insertion and caspase 8 deletion mutants after 8 or 6 h of expression respectively. Caspase 7/linker casp-8b is shown. The underlined residues identify the processed cleavage sites.

DISCUSSION

All caspases have an IDC that is poorly conserved among family members, yet it is always cleaved during activation. Cleavage of executioner caspases is the driving force for their activation [810], whereas cleavage of initiator caspase IDC is insufficient for activation [47,34]. The IDC of human apoptotic caspases has two cleavage sites that are processed during apoptosis or E. coli expression. In the present study, we addressed the structure relationship of the two activation sites of executioner caspase 7.

Here, we show that the sequences of both activation sites of caspase 7 are suboptimal for activation by initiators in vitro and in cellulo. Changing site 1 to the preferred motif recognized by caspase 8 resulted in a higher activation rate, whereas activation by caspase 9 seemed optimal. Interestingly, most site 1 mutants we examined have poor kinetic efficacies demonstrating that it also supports enzyme activity once cleaved. Indeed, crystal structures of many cleaved caspases show that residues upstream of site 1 are implicated in contacts with the neighbouring catalytic unit within a dimer to form the so-called L2–L2′ loop bundle (Supplementary Figure S3A at http://www.bioscirep.org/bsr/031/bsr0310283add.htm). Many studies [10,32] have confirmed that Asp192 and Ile195 are critical for caspase 7 activity, explaining why many site 1 mutants have reduced kinetic performance. In the crystal structures, these residues contact amino acids C-terminal to the IDC. We have tested two of these residues, Lys212 and Ile213, and found that only conservative mutations at these positions supported robust activity. Both residues are conserved in the closely related homologue caspase 3, whereas only the hydrophobic character of position 213 is preserved in other caspases. We speculate that the lack of conservation may explain why swapping the IDC between caspases failed to produce activity because the IDC must also complement features of the active site to stabilize it. Suboptimal cleavage sites in protease zymogen activation have been reported before, such as for plasminogen [35]. In this particular context, structural requirements over-ride sequence preference, and so caspase 7 activation demonstrates an interesting analogy to the requirements of plasmin activation. Less than optimal activation processes pave the way for rigorous regulation. Indeed, coupled with the difference in reactivity towards its endogenous inhibitor XIAP [12], a controlled cleavage at two different sites can enhance the regulation of the whole system.

Our results obtained with site 1 and site 2 mutants suggest that the location has considerable impact on the efficacy of activation. We had previously demonstrated that caspase 7 cannot directly undergo auto-activation at site 2 [18]. Unexpectedly, inserting the best motif recognized by caspase 7 (DEVD) at site 2 allowed caspase auto-activation directly at this site. However, we could not obtain a caspase able to self-activate using the site 1 motif (IQAD) despite the fact that we demonstrated that it was close to optimal for self-activation. Therefore it seems that only an optimal motif could be used at site 2, but these results demonstrate that it is the positioning within the IDC that is critical. Neither the available crystal structures of the active form nor the ones of the zymogen form of caspase 7 reveal well-defined electron density for residues surrounding site 2 (listed in [17,32,36,37]). However, these structures show that residues close to or part of site 1 (Asp192 to Gln196) stabilize the active form by immobilizing the catalytic cysteine.

Recently, the NMR structure of monomeric pro-caspase 8 was published [38]. It shows the IDC adopting several different conformations, and potentially making several contacts with the core of the protein. Because of its short length, caspase 7's IDC cannot adopt these configurations. Instead, the available structures of pro-caspase 7 demonstrate that the IDC must travel from one end to the other end of the molecule as an extended loop with contacts made only by residues C-terminal to site 2. Therefore we can propose that site 2 is not as accessible as site 1 because of these contacts, which restrain IDC movement. However, mutation of site 2 to DEVD would disturb these interactions enough to render that mutant highly susceptible to self-activation. Indeed, pro-caspase 7 structures show that residue P2-Thr205 of site 2 contacts Lys212′ of the other molecule of the dimer, which is likely disrupted in the DEVD mutant (Supplementary Figure S3B) [39,40]. Furthermore, Ile213′ and the side chain of Lys212′ occupy a hydrophobic pocket, with the ϵ-amino group of the lysine residue participating in hydrogen bonds with a nearby water molecule. Because cleavage of the IDC results in an important repositioning of Lys212 and Ile213, we can assume that insertion of the IDC into the central cavity of the dimer is reversible allowing rapid exchange with the other IDC, thus rendering site 2 of both catalytic units less available for cleavage.

Finally, we also evaluated the impact of IDC length on the activity of caspase 7. Several reports have demonstrated that the zymogen of initiator caspases displays activity without cleavage [5,6,20], which is not the case for executioners. The length of the IDC is one of the critical features that are hypothesized to allow activity without cleavage [17]. Consequently, our rationale was to elongate caspase 7's IDC hoping to produce an active caspase 7 without proteolysis. However, neither the insertion of a highly flexible sequence of up to 14 residues nor the replacement with the IDC of an initiator caspase succeeded in removing this requirement. A corollary to the IDC length hypothesis is that the initiator IDCs must permit activity without proteolysis. In fact, shortening the IDC by four residues was sufficient to produce a caspase 8 that also requires cleavage for its activation. This result reinforces the idea that IDC length is essential for the higher zymogenicity of initiator caspases compared with executioners, and that for executioners, it serves also to stabilize the active form. In their study, Keller et al. [38] also concluded that caspase 8 IDC length, and the various determinants it contains, were critical in providing activity to the zymogen. However, they could not establish how the length controls the auto-cleavage process of caspase 8 because IDC truncation mutants were expressed as separate subunits and then refolded to yield the catalytically active protease. Recent studies on an uncleaved but active caspase 3 mutant [41] (shorter IDC than caspase 7) and the crystal structure of pro-caspase 6 [42] (longer IDC than caspase 7 and caspase 8) draw a more complex picture of how the IDC controls the zymogen state and the activity of caspases. It is likely that a combination of several molecular determinants rules this process on an individual basis and that general rules may not apply even among executioner caspases.

Caspase activation has been extensively studied, but many of the underpinning molecular determinants of activation remain unknown. Our results highlight the importance of the primary structure of caspase 7 IDC for its activation and activity. They also demonstrates, using a real-life proteolytic event, the need to also study molecular determinants other than the immediate cleavage site and the substrate binding site to fully understand how proteases reach high levels of efficacy and specificity.

Abbreviations

     
  • Ac-DEVD-pNA

    N-acetyl-Asp-Glu-Val-Asp-p-nitroanilide

  •  
  • afc

    7-amino-4-(trifluoromethyl)coumarin

  •  
  • CARD

    caspase recruitment domain

  •  
  • CrmA

    cytokine response modifier protein

  •  
  • cyt

    c, cytochrome c

  •  
  • DED

    death effector domain

  •  
  • DTT

    dithiothreitol

  •  
  • HEK

    human embryonic kidney

  •  
  • hsp90

    heat-shock protein 90

  •  
  • IDC

    interdomain connector

  •  
  • pNA

    p-nitroanilide

  •  
  • TRAIL

    tumour-necrosis-factor-related apoptosis-inducing ligand

  •  
  • WT

    wild-type

  •  
  • XIAP

    X-linked inhibitor of apoptosis protein

  •  
  • Z-Val-Ala-DL-Asp-CH2F

    benzyloxycarbonyl-valyl-alanyl-DL-aspartyl-fluoromethane

AUTHOR CONTRIBUTION

Dave Boucher designed, performed and interpreted the experiments, with the following exceptions: Véronique Blais did all of the in cellulo experiments and Marcin Drag made the substrate library and performed assay screens with the library. Jean-Bernard Denault contributed to the conceptual and experimental design, and in writing the manuscript.

We thank Professor Richard Leduc, Professor Guy Salvesen and Professor Pierre Lavigne for helpful discussions.

FUNDING

This work was supported by the Canadian Institutes of Health Research (CIHR) [grant number MOP-86563], Cancer Research Society (CRS), Canadian Foundation for Innovation (CFI) [grant number 18637] and the Banting Research Foundation. J.-B.D. is a Chercheur-boursier Junior 1 of the Fonds de la Recherche en Santé du Québec. M.D. is supported by the Foundation for Polish Science.

References

References
1
Fischer
 
U.
Janicke
 
R. U.
Schulze-Osthoff
 
K.
 
Many cuts to ruin: a comprehensive update of caspase substrates
Cell Death Differ.
2003
, vol. 
10
 (pg. 
76
-
100
)
2
Lamkanfi
 
M.
Festjens
 
N.
Declercq
 
W.
Vanden Berghe
 
T.
Vandenabeele
 
P.
 
Caspases in cell survival, proliferation and differentiation
Cell Death Differ.
2007
, vol. 
14
 (pg. 
44
-
55
)
3
Yi
 
C. H.
Yuan
 
J.
 
The Jekyll and Hyde functions of caspases
Dev. Cell
2009
, vol. 
16
 (pg. 
21
-
34
)
4
Pop
 
C.
Timmer
 
J.
Sperandio
 
S.
Salvesen
 
G. S.
 
The apoptosome activates caspase-9 by dimerization
Mol. Cell
2006
, vol. 
22
 (pg. 
269
-
275
)
5
Pop
 
C.
Fitzgerald
 
P.
Green
 
D. R.
Salvesen
 
G. S.
 
Role of proteolysis in caspase-8 activation and stabilization
Biochemistry
2007
, vol. 
46
 (pg. 
4398
-
4407
)
6
Stennicke
 
H. R.
Deveraux
 
Q. L.
Humke
 
E. W.
Reed
 
J. C.
Dixit
 
V. M.
Salvesen
 
G. S.
 
Caspase-9 can be activated without proteolytic processing
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
8359
-
8362
)
7
Donepudi
 
M.
Mac Sweeney
 
A.
Briand
 
C.
Gruetter
 
M. G.
 
Insights into the regulatory mechanism for caspase-8 activation
Mol. Cell
2003
, vol. 
11
 (pg. 
543
-
549
)
8
Zhou
 
Q.
Salvesen
 
G. S.
 
Activation of pro-caspase-7 by serine proteases includes a non-canonical specificity
Biochem. J.
1997
, vol. 
324
 (pg. 
361
-
364
)
9
Boatright
 
K. M.
Salvesen
 
G. S.
 
Mechanisms of caspase activation
Curr. Opin. Cell Biol.
2003
, vol. 
15
 (pg. 
725
-
731
)
10
Denault
 
J. B.
Békés
 
M.
Scott
 
F. L.
Sexton
 
K. M. B.
Bogyo
 
M.
Salvesen
 
G. S.
 
Engineered hybrid dimers: tracking the activation pathway of caspase-7
Mol. Cell
2006
, vol. 
23
 (pg. 
523
-
533
)
11
Srinivasula
 
S. M.
Ahmad
 
M.
MacFarlane
 
M.
Luo
 
Z.
Huang
 
Z.
Fernandes-Alnemri
 
T.
Alnemri
 
E. S.
 
Generation of constitutively active recombinant caspases-3 and -6 by rearrangement of their subunits
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
10107
-
10111
)
12
Scott
 
F. L.
Denault
 
J. B.
Riedl
 
S. J.
Shin
 
H.
Renatus
 
M.
Salvesen
 
G. S.
 
XIAP inhibits caspase-3 and -7 using two binding sites: evolutionarily conserved mechanism of IAPs
EMBO J.
2005
, vol. 
24
 (pg. 
645
-
655
)
13
Shiozaki
 
E. N.
Chai
 
J.
Rigotti
 
D. J.
Riedl
 
S. J.
Li
 
P.
Srinivasula
 
S. M.
Alnemri
 
E. S.
Fairman
 
R.
Shi
 
Y.
 
Mechanism of XIAP-mediated inhibition of caspase-9
Mol. Cell
2003
, vol. 
11
 (pg. 
519
-
527
)
14
Srinivasula
 
S. M.
Hegde
 
R.
Saleh
 
A.
Datta
 
P.
Shiozaki
 
E.
Chai
 
J.
Lee
 
R. A.
Robbins
 
P. D.
Fernandes-Alnemri
 
T.
Shi
 
Y.
Alnemri
 
E. S.
 
A conserved XIAP-interaction motif in caspase-9 and Smac/DIABLO regulates caspase activity and apoptosis
Nature
2001
, vol. 
410
 (pg. 
112
-
116
)
15
Takahashi
 
R.
Deveraux
 
Q.
Tamm
 
I.
Welsh
 
K.
Assa-Munt
 
N.
Salvesen
 
G. S.
Reed
 
J. C.
 
A single BIR domain of XIAP is sufficient for inhibiting caspases
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
7787
-
7790
)
16
Choi
 
Y. E.
Butterworth
 
M.
Malladi
 
S.
Duckett
 
C. S.
Cohen
 
G. M.
Bratton
 
S. B.
 
The E3 ubiquitin ligase cIAP1 binds and ubiquitinates caspase-3 and -7 via unique mechanisms at distinct steps in their processing
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
12772
-
12782
)
17
Fuentes-Prior
 
P.
Salvesen
 
G. S.
 
The protein structures that shape caspase activity, specificity, activation and inhibition
Biochem. J.
2004
, vol. 
384
 (pg. 
201
-
232
)
18
Denault
 
J. B.
Salvesen
 
G. S.
 
Human caspase-7 activity and regulation by its N-terminal peptide
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
34042
-
34050
)
19
Stennicke
 
H. R.
Salvesen
 
G. S.
 
Caspases: preparation and characterization
Methods
1999
, vol. 
17
 (pg. 
313
-
319
)
20
Boatright
 
K. M.
Renatus
 
M.
Scott
 
F. L.
Sperandio
 
S.
Shin
 
H.
Pedersen
 
I.
Ricci
 
J.-E.
Edris
 
W. A.
Sutherlin
 
D. P.
Green
 
D. R.
Salvesen
 
G. S.
 
A unified model for apical caspase activation
Mol. Cell
2003
, vol. 
11
 (pg. 
529
-
541
)
21
Zhou
 
Q.
Snipas
 
S.
Orth
 
K.
Dixit
 
V. M.
Salvesen
 
G. S.
 
Target protease specificity of the viral serpin CrmA: analysis of five caspases
J. Biol. Chem.
1997
, vol. 
273
 (pg. 
7797
-
7800
)
22
Zhou
 
Q.
Salvesen
 
G. S.
 
Viral caspase inhibitors CrmA and p35
Methods Enzymol.
2000
, vol. 
322
 (pg. 
143
-
154
)
23
Edelhoch
 
H.
 
Spectroscopic determination of tryptophan and tyrosine in proteins
Biochemistry
1967
, vol. 
6
 (pg. 
1948
-
1954
)
24
Li
 
P.
Nijhawan
 
D.
Budihardjo
 
I.
Srinivasula
 
S. M.
Ahmad
 
M.
Alnemri
 
E. S.
Wang
 
X.
 
Cytochrome c and dATP-dependent formation of Apaf-1/caspase-9 complex initiates an apoptotic protease cascade
Cell
1997
, vol. 
91
 (pg. 
479
-
489
)
25
Denault
 
J. B.
Eckelman
 
B. P.
Shin
 
H.
Pop
 
C.
Salvesen
 
G. S.
 
Caspase 3 attenuates XIAP (X-linked inhibitor of apoptosis protein)-mediated inhibition of caspase 9
Biochem. J.
2007
, vol. 
405
 (pg. 
11
-
19
)
26
Thornberry
 
N. A.
Rano
 
T. A.
Peterson
 
E. P.
Rasper
 
D. M.
Timkey
 
T.
Garcia-Calvo
 
M.
Houtzager
 
V. M.
Nordstrom
 
P. A.
Roy
 
S.
Vaillancourt
 
J. P.
, et al 
A combinatorial approach defines specificities of members of the caspase family and granzyme B. Functional relationships established for key mediators of apoptosis
J. Biol. Chem.
1997
, vol. 
272
 (pg. 
17907
-
17911
)
27
Denault
 
J. B.
Salvesen
 
G. S.
 
Mor
 
G.
Rutherford
 
T. J.
 
Apoptotic caspase activation and activity
Apoptosis and Cancer
2007
Totowa, NJ
Humana Press
pg. 
272
 
28
Bury
 
A.
 
Analysis of protein and peptide mixtures: evaluation of three sodium dodecyl sulphate-polyacrylamide gel electrophoresis buffer systems
J. Chromatogr.
1981
, vol. 
213
 (pg. 
491
-
500
)
29
Timmer
 
J. C.
Zhu
 
W.
Pop
 
C.
Regan
 
T.
Snipas
 
S. J.
Eroshkin
 
A. M.
Riedl
 
S. J.
Salvesen
 
G. S.
 
Structural and kinetic determinants of protease substrates
Nat. Struct. Mol. Biol.
2009
, vol. 
16
 (pg. 
1101
-
1108
)
29a
Pop
 
C.
Oberst
 
A.
Drag
 
M.
Van Raam
 
B. J.
Riedl
 
S. J.
Green
 
D. R.
Salvesen
 
G. S.
 
FLIPL induces caspase 8 activity in the absence of interdomain caspase 8 cleavage and alters substrate specificity
Biochem. J.
2011
, vol. 
433
 (pg. 
447
-
457
)
30
Malladi
 
S.
Challa-Malladi
 
M.
Fearnhead
 
H. O.
Bratton
 
S. B.
 
The Apaf-1*procaspase-9 apoptosome complex functions as a proteolytic-based molecular timer
EMBO J.
2009
, vol. 
28
 (pg. 
1916
-
1925
)
31
Walsh
 
J. G.
Cullen
 
S. P.
Sheridan
 
C.
Luthi
 
A. U.
Gerner
 
C.
Martin
 
S. J.
 
Executioner caspase-3 and caspase-7 are functionally distinct proteases
Proc. Natl. Acad. Sci. U.S.A.
2008
, vol. 
105
 (pg. 
12815
-
12819
)
32
Witkowski
 
W. A.
Hardy
 
J. A.
 
L2′ loop is critical for caspase-7 active site formation
Protein Sci.
2009
, vol. 
18
 (pg. 
1459
-
1468
)
33
Wei
 
Y.
Fox
 
T.
Chambers
 
S. P.
Sintchak
 
J.
Coll
 
J. T.
Golec
 
J. M.
Swenson
 
L.
Wilson
 
K. P.
Charifson
 
P. S.
 
The structures of caspases-1, -3, -7 and -8 reveal the basis for substrate and inhibitor selectivity
Chem. Biol.
2000
, vol. 
7
 (pg. 
423
-
432
)
34
Oberst
 
A.
Pop
 
C.
Tremblay
 
A. G.
Blais
 
V.
Denault
 
J. B.
Salvesen
 
G. S.
Green
 
D. R.
 
Inducible dimerization and inducible cleavage reveal a requirement for both processes in caspase-8 activation
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
16632
-
16642
)
35
Ding
 
L.
Coombs
 
G. S.
Strandberg
 
L.
Navre
 
M.
Corey
 
D. R.
Madison
 
E. L.
 
Origins of the specificity of tissue-type plasminogen activator
Proc. Natl. Acad. Sci. U.S.A.
1995
, vol. 
92
 (pg. 
7627
-
7631
)
36
Agniswamy
 
J.
Fang
 
B.
Weber
 
I. T.
 
Plasticity of S2-S4 specificity pockets of executioner caspase-7 revealed by structural and kinetic analysis
FEBS J.
2007
, vol. 
274
 (pg. 
4752
-
4765
)
37
Fu
 
G.
Chumanevich
 
A. A.
Agniswamy
 
J.
Fang
 
B.
Harrison
 
R. W.
Weber
 
I. T.
 
Structural basis for executioner caspase recognition of P5 position in substrates
Apoptosis
2008
, vol. 
13
 (pg. 
1291
-
1302
)
38
Keller
 
N.
Mares
 
J.
Zerbe
 
O.
Grutter
 
M. G.
 
Structural and biochemical studies on procaspase-8: new insights on initiator caspase activation
Structure
2009
, vol. 
17
 (pg. 
438
-
448
)
39
Chai
 
J.
Wu
 
Q.
Shiozaki
 
E.
Srinivasula
 
S. M.
Alnemri
 
E. S.
Shi
 
Y.
 
Crystal structure of a procaspase-7 zymogenMechanisms of activation and substrate binding
Cell
2001
, vol. 
107
 (pg. 
399
-
407
)
40
Riedl
 
S. J.
Fuentes-Prior
 
P.
Renatus
 
M.
Kairies
 
N.
Krapp
 
R.
Huber
 
R.
Salvesen
 
G. S.
Bode
 
W.
 
Structural basis for the activation of human procaspase-7
Proc. Natl. Acad. Sci. U.S.A.
2001
, vol. 
98
 (pg. 
14790
-
14795
)
41
Walters
 
J.
Pop
 
C.
Scott
 
F. L.
Drag
 
M.
Swartz
 
P.
Mattos
 
C.
Salvesen
 
G. S.
Clark
 
A. C.
 
A constitutively active and uninhibitable caspase-3 zymogen efficiently induces apoptosis
Biochem. J.
2009
, vol. 
424
 (pg. 
335
-
345
)
42
Wang
 
X. J.
Cao
 
Q.
Liu
 
X.
Wang
 
K. T.
Mi
 
W.
Zhang
 
Y.
Li
 
L. F.
Leblanc
 
A. C.
Su
 
X. D.
 
Crystal structures of human caspase 6 reveal a new mechanism for intramolecular cleavage self-activation
EMBO Rep.
2010
, vol. 
11
 (pg. 
153
-
160
)

Supplementary data