Caspase-9 Activation of Procaspase-3 but not Procaspase-6 is Based both on Local Context of Cleavage Site Motifs and on Sequence

Abstract

Studying the interactions between a protease and its protein substrates at a molecular level is crucial to identify the factors facilitating selection of particular proteolytic substrates and not others. These selection criteria include both the sequence and the local context of the substrate cleavage site where the active-site of the protease initially binds, and then performs proteolytic cleavage. Caspase-9, an initiator of the intrinsic apoptotic pathway, mediates activation of executioner procaspase-3 by the intersubunit linker (ISL) cleavage at site, ¹⁷²IETD↓S. Although procaspase-6, another executioner, possesses two ISL cleavage sites, site 1: ¹⁷⁶DVVD↓N and site 2: ¹⁹⁰TEVD↓A, neither of which is directly cut by caspase-9. Thus, caspase-9 directly activates procaspase-3 but not procaspase-6. To elucidate this selectivity of caspase-9, we engineered constructs of procaspase-3 (e.g., swapping the ISL site: ¹⁷²IETD↓S to DVVDN and TEVDA) and procaspase-6 (e.g., swapping the site 1: ¹⁷⁶DVVD↓N and site 2: ¹⁹⁰TEVD↓A to IETDS). Using the substrate-digestion data of these constructs, we show here that the P4-P1’ sequence of procaspase-6 ISL site 1 (DVVDN) is accessible yet uncleavable by caspase-9. We also found that caspase-9 can recognize the P4-P1’ sequence of procaspase-6 ISL site 2 (TEVDA); however, the local context of this cleavage site is the critical factor that prevents proteolytic cleavage. Overall, our data have demonstrated that both the sequence and the local context of the ISL cleavage sites play a vital role in preventing the activation of procaspase-6 directly by caspase-9.

Graphical Abstract

Introduction:

Proteolysis, a process of enzymes catalyzing the hydrolytic cleavage of their substrates, drives various biological pathways (e.g., cell cycle,¹ cell differentiation,² and cell death³) to maintain homeostasis in all living organisms. Examining the interactions between a particular protease and its protein substrate at a molecular level is beneficial in understanding their involvement in particular biological pathways. The Schechter-Berger convention has been widely adopted by the protease community to showcase the interactions between the peptide residues in the substrate cleavage-site (denoted as P) and sub-sites of protease active-site (denoted as S), where the cleavage (↓) occurs between P1 and P1’ (Figure 1A).⁴ The nature of the sub-sites on the enzyme together with the cleavage site motif within the substrate is considered to define protease preference. This notion is reinforced when the broad range of sequences detected by protein or peptide-based protease substrate profiling are reduced to a single “preferential” cleavage motif. In contrast to this narrow view of recognition specificity, detailed analyses of aggregate data on proteolysis suggest that reducing the concept of recognition motif to a single sequence is overly simplistic. For example, the preferred cleavage motif of caspase-7 protease using the peptide-based screening⁵ and the proteomics study⁶ was identified as DEVD↓ (P4-P1) and DEVD↓G (P4-P1’), respectively. However, from the same proteomics study’s data, we found that, out of 128 cleaved peptides by caspase-7, the cleavage sites of DEVD↓G, DEVD↓X, and DXXD↓X were only 2, 4, and 36, respectively.⁶ Another example is the family of ClpP proteases (from E. Coli, S. aureus, and human mitochondria) which favor peptide substrates composed of specific residues at the P3-P1 positions (e.g., natural amino acids preferred by human mitochondria ClpP at P3-P1 positions are F/W-M/T/L-M/L); nonetheless, proteomics studies on these proteases revealed very low cleavage preference.⁷ In fact, it was later discovered that proteolysis by ClpP protease (from Caulobacter crescentus) complexed with ClpX (an unfoldase which feeds ClpP) is mainly governed by local context (substrate cleavage occurred after every 10-13 residues) and not sequence dependent.⁸ Looking at the cleavage preferences for any protease and its suite of cleaved substrates, it becomes clear that i) no single peptide sequence can fully account for the cleavage properties observed for a natural protease, and ii) the sequence specificity alone cannot completely account for the selection of substrates that are cleaved. Thus, we hypothesized at the outset of this work that both the sequence and the local context of the substrate cleavage site may be important factors governing recognition by a protease.

Figure 1. Procaspase-3 and −6 share the same core structural fold, and intersubunit linker (ISL) cleavage is a pre-requisite for their activation.

(A) Schechter-Berger schematic diagram representing protease-specificity towards the cleavage sequence of the substrates.⁴ Peptide residues of the substrates and sub-sites of protease active-site are denoted as P and S, respectively. The cleavage of substrate occurs at the amide bond between the residues at P1 and P1’. Caspases prefer aspartate at the P1 position. (B) Procaspase-3 (light pink) and procaspase-6 (light cyan) structures with the missing loops/regions shown as dashes, which were modeled as described in the materials and methods. The PDB IDs used to model procaspase-3 and −6 were 4JQY and 3NR2, respectively. The active-site cysteines of both procaspases were substituted by alanine (C163A) to determine the structures of full-length uncleaved zymogen.^25,26 Active-site cysteines (substituted from alanine present in the original structures by using mutagenesis feature of PyMOL) of procaspase-3 and −6 are shown as spheres. Both procaspases possess a similar overall structural fold in the core. The greatest differences are present in the mobile loops. Loops containing the ISL are highlighted (dark pink for procaspase-3 and dark cyan for procaspase-6). Procaspase-3 has a known ISL cleavage after D175 (P1 position),^19,20 and procaspase-6 has two known ISL cleavages after D179 and D193.²¹ (C) Linear cartoons of procaspase-3 and −6 illustrating their prodomain (N), large subunit (Lg), ISL and small subunit (Sm). Simply removing the prodomain for both procaspases is not sufficient for their activation. ISL cleavage is required for both procaspases for activation.^41,42

One prime example of the gulf between canonical recognition sequence and the actual cleavage propensity is exemplified in the caspases. Caspases are cysteine-aspartate proteases that play key roles in regulating multiple cellular pathways including apoptosis and inflammation. Dysfunction in caspase regulation is a hallmark of several diseases such as cancers,^9,10 autoimmune disorders,^11,12 and neurodegeneration.^13,14 Therefore, it is important to understand their cellular pathways at a molecular level. The defining feature of this family of proteases is their ability to cleave substrates at sites containing aspartate at the P1 position (Figure 1A), although glutamate and phosphoserine can also be recognized albeit at low frequency.^15,16 Apoptotic caspases are categorized into two groups: initiators (caspase-8, -9, and -10) and executioners (caspase-3, -6 and -7). Caspases are translated as full-length, inactive zymogens termed as procaspases. Initiator zymogens are recruited to activation platforms upon either intrinsic (e.g., procaspase-9 to form apoptosome)¹⁷ or extrinsic (e.g., procaspase-8 to form death-inducing signaling complex)¹⁸ cell-signaling to dimerize and ultimately achieve maturation. Excluding the mobile loops, executioner zymogens possess similar structural folds (Figure 1B) and are activated by a cleavage event at their intersubunit linker (ISL) generating large and small subunits (Lg and Sm) from each chain of the procaspase dimer (Figure 1C). Procaspase-3 has one ISL cleavage site at D175,^19,20 while procaspase-6 has two ISL cleavage sites, D179 and D193²¹ (Figure 1B and 1C). In an intrinsic apoptotic pathway, upon its activation, caspase-9 cleaves the ISL of procaspase-3 and −7, thereby activating them to the mature form. Activated caspase-3 mediates procaspase-6 ISL cleavage to activate caspase-6.^22-24 Activated executioners proteolyze their respective and common substrates evoking apoptosis. In this manuscript, we sought to address a key question in the field: Why can caspase-9 cleave the ISLs of procaspase-3 and −7 but not the procaspase-6 ISL cleavage sites? Addressing this question, while providing insights for caspase-9, is also applicable to proteases and their substrate selection generally.

Materials and Methods:

Generation of full-length procaspase-3 and −6 models by adding missing residues

We generated full-length models of procaspase-3 and −6 by employing structures from the Protein Data Bank, PDB: 4JQY and PDB: 3NR2, respectively.^25,26 The missing residues of procaspase-3 (1-31, 54-64, 165-185, 201-210, and 251-260) as well as procaspase-6 (1-30, 167-186, and 262-270) were modeled using UCSF Chimera/MODELLER integrated system.^27,28 The illustrations of procaspase-3 and −6 models were created using PyMOL (Schrödinger, LLC).

Sequence alignment of procaspase-3 and −6

We used Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) to align the sequences of procaspase-3 (Uniprot ID: P42574) and procaspase-6 (Uniprot ID: P55212).

DNA plasmids and generation of caspase constructs

Plasmids for expression of wild-type caspase-3 (pET23b-Casp3-His) and wild-type caspase-9 (pET23b-Casp9-His) were gifts from Guy Salvesen, and obtained from Addgene as plasmids, 11821 and 11829, respectively.^29,30 An E. coli codon-optimized wild-type caspase-6 gene (with a 6xHis at C-terminus) was synthetically produced (Celtek Bioscience) and ligated into the NdeI/BamHI sites of the pET11a vector. Wild-type caspase-3 and wild-type caspase-6 plasmids were used as templates to generate their respective constructs (substitutions and deletions) employing QuikChange mutagenesis (Agilent Technologies).

Expression and purification of recombinant caspase constructs

DNA plasmids for expression of caspase constructs were transformed into BL21 (DE3) T7 express strain of E. coli (New England Biolabs). A single colony was picked for each construct and overnight seed cultures were grown in 50 mL LB media (Research Products International) supplemented with 0.1 mg/mL ampicillin (Fisher BioReagents) by incubating at 37 °C. For large-scale growth, 3 mL of seed culture of each caspase construct was transferred into 1 L of LB medium containing 0.1 mg/mL ampicillin. Incubation at 37°C was carried until desired Abs₆₀₀ (0.6 for caspase-3 constructs, 0.8 for caspase-6 constructs, and 1.0 for caspase-9) was achieved. In each case, protein expression was induced by adding 1 mM of IPTG (GoldBio), and the temperatures were lowered (30 °C for caspase-3 constructs, 25 °C for caspase-6 constructs, and 25 °C for caspase-9). After 4 hours, cells were centrifuged at 5000 rcf for 7 minutes at 4 °C, and cell pellets were collected and stored at −80 °C until they were thawed and used for purification.

To purify all caspase constructs, we employed Ni²⁺-ion affinity chromatography followed by anion exchange. Caspase-3 constructs were purified as previously described wild-type caspase-3 purification protocol.³¹ Caspase-6 constructs were purified as previously described procaspase-6 (caspse-6 C163S) purification protocol.³² Caspase-9 was purified as previously described caspase-9 full-length purification protocol.³¹ Purity and concentrations of purified caspases were assessed by SDS-PAGE gel electrophoresis, and aliquots were stored at −80 °C until further usage for different assays.

Protein substrate digestion and determination of cleavage rate (k)

Different concentrations (0, 0.125, 0.25, 0.5, and 1 μM) of caspase-9 were individually incubated with 3 μM of an individual caspase-3 or −6 construct in an activity assay buffer (100 mM MES, pH = 6.5, 20% PEG 400, 5 mM DTT) for 1 hour at 37 °C. Each reaction was stopped by adding 1x SDS loading dye (New England Biolabs). These samples were denatured at 90 °C for 10 minutes. Each sample was analyzed using a 16% SDS-PAGE. SDS-PAGE gels were imaged using a ChemiDocTM MP imaging system (Bio-Rad Laboratories). Band intensities were quantified using Image Lab software (Bio-Rad Laboratories). The cleavage rates of caspase-9 to proteolyze each of the caspase-3 and −6 constructs were determined by following the previously described method.^33,34

Caspase-9 LEHD-ase activity and determination of kinetic parameters

We followed previously described protocol to derive the LEHD-ase activity of caspase-9.³¹ For a substrate titration, 10 μL of fluorogenic substrate, Ac-LEHD-afc (N-acetyl-Leu-Glu-His-Asp-7-amido-4-trifluoromethylcoumarin; Enzo Life Sciences, Inc.) with concentrations ranging from 0 to 3000 μM were placed into a 96-well black plate. Recombinant caspase-9 (90 μL of 800 nM) in an activity assay buffer (100 mM MES, pH = 6.5, 20% PEG 400, 5 mM DTT) was added into each well to make final volume of 100 μL. Immediately, fluorescence kinetics were measured (λ_ex of 400 nm and λ_em of 505 nm) at 37 °C for 7 minutes using a microplate reader (SpectraMax M5; Molecular Devices). Initial velocities versus substrate concentrations were plotted to a Michaelis-Menten curve using GraphPad Prism software, and K_M was determined. To derive the exact concentration of caspase-9, an active-site titration was performed using a covalent inhibitor, z-VAD-fmk (carbobenzoxy-Val-Ala-Asp-fluoromethylketone; Enzo Life Sciences, Inc.). For that, 2 μL of z-VAD-fmk (diluted in DMSO) with the concentrations ranging from 0 to 2 mM was added into 96 black-well plate containing 90 μL of 800 nM caspase-9 in an activity assay buffer. The plate was sealed using aluminum foil and incubated at 25 °C for 1.5 hours. Each aliquot (92 μL) was transferred in a duplicate 96 black-well plate containing 1 mM Ac-LEHD-afc (to make 100 μL as a total volume), and fluorescence kinetics were measured using a microplate reader as done for substrate titration. The exact concentration of caspase-9 (total enzyme concentration, E_T) was determined as the lowest concentration at which full inhibition was observed, and using this value, k_cat was calculated. Finally, caspase-9 LEHD-ase activity (k_cat/K_M) was calculated. Both the substrate titration and active-site titration assays were performed twice using two different aliquots on two separate days.

Generation of structural models of caspase-9 active-site accommodating peptide substrates

To generate models of caspase-9 active-site bound to peptides, IETD, DVVD and TEVD as P4-P1 positions, we employed a crystal structure of caspase-9 bound to a peptide substrate, z-EVD-Dcbmk (benzoxycarbonyl-Glu-Val-Asp-dichlorobenzylmethylketone) (PDB: 1JXQ). We used the mutagenesis feature of PyMOL (Schrödinger, LLC) to substitute the relevant residues. Rotamers of each substituted residue with the minimal clashes were selected to generate models.

Results:

Caspase-9 cleaves the ISL of procaspase-3 but not procaspase-6.

To examine the behavior of procaspase-3 and −6 as substrates of wild-type caspase-9 in vitro, we performed an SDS-PAGE-based substrate-digestion assay. Caspase-9 cleaved caspase-3 C163A (Figure 2A); however, it was unable to proteolyze caspase-3 C163A D175A (Figure 2B). Thus, caspase-9 cleaved procaspase-3 ISL at D175 resulting in two cleavage products: prodomain + large subunit (N+Lg) and small subunit (Sm) (Figure 2A). Caspase-9 cleaved the prodomain of caspase-6 C163S resulting in caspase-6 ΔN C163S; nevertheless, no significant ISL cleavage was observed (Figure 2C). A previous study showed³⁵ that removal of the prodomain while the ISL remained intact was not sufficient to achieve procaspase-6 activation. Thus, caspase-9 cannot activate procaspase-6 directly (Figure 2C). From a peptide-based profiling study,⁵ caspase-9 canonical recognition sequence is LEHD (P4-P1). Therefore, we carried out an LEHD-ase activity assay to determine the kinetic parameters (k_cat and K_M) of caspase-9 using a fluorogenic peptide substrate, Ac-LEHD-afc (Figure 2D). The catalytic efficiency, k (k_cat/K_M), of caspase-9 to hydrolyze Ac-LEHD-afc is (12.8 ± 1.1) x 10⁴ μM⁻¹s⁻¹ (Figure 2E). For the experiments (Figure 2A, 2B, and 2C), the substrate (caspase-3 or −6) concentration (3 μM) was in excess over the enzyme (caspase-9) concentration (0 – 1 μM). Limitations of the slow kinetics of caspase-9 coupled with the sensitivity of detecting cleaved products makes it impossible to monitor the reaction under pseudo-first order conditions as has been described previously.^33,34 Nevertheless, we used a similar approach³⁴ to estimate a relative rate (k) for caspase-9 cleavage of the ISLs of procaspase-3, caspase-3 D175A C163A, and procaspase-6 (Figure 2E). These apparent relative cleavage rates (k) for proteolysis of the ISL of procaspase-3, caspase-3 D175A C163A, and procaspase-6 by caspase-9 were 390 ± 30 M⁻¹s⁻¹, < 25 M⁻¹s⁻¹, and < 25 M⁻¹s⁻¹, respectively (Figure 2E). The sample calculation of estimating the rate (k) value of caspase-9 to cleave procaspase-3 (in M⁻¹s⁻¹) is shown (supporting information xlsx file). We used same approach throughout this manuscript to approximate rate (k) values. Our results complemented previous findings by various research groups demonstrating caspase-9 does not directly activate procaspase-6.^22-24,36 Excluding its prodomain and the ISL, procaspase-6 is about 43% identical in protein sequence to procaspase-3 (Figure S1 of supporting document docx). Due to the fact that they are the most divergent segments of caspase-6, we hypothesized that the prodomain and/or the ISL of procaspase-6 might protect procaspase-6 direct activation by caspase-9.

Figure 2. Caspase-9 cleaves the ISL of procaspase-3 but not procaspase-6.

(A, B, and C) Different concentrations of recombinant wild-type caspase-9 ranging from 0 to 1 μM were incubated with 3 μM of (A) recombinant procaspase-3 (caspase-3 C163A), (B) caspase-3 C163A D175A and (C) procaspase-6 (caspase-6 C163S) for 1 hour at 37 °C. In the last lane of each gel, 1 μM of caspase-9 was added as a control. Caspase-9 cleaved (A) procaspase-3 at the ISL resulting in two cleavage products: the prodomain + large subunit (N+Lg) and the small subunit (Sm), while no significant cleavage of (B) caspase-3 D175A C163A was observed, verifying that caspase-9 cleaves the ISL of procaspase-3 at D175. Caspase-9 removed the prodomain of (C) procaspase-6 (denoted as ΔN); however, no significant ISL cleavage was observed. (D) The peptide substrate, Ac-LEHD-afc, was employed to measure the LEHD-ase activity of caspase-9. Each experiment (A-D) was performed twice using different samples on two different days, and mean ± S.D. values were used to derive the rate constant (k) values. (E) We derived the cleavage rate (k) of caspase-9 to proteolyze the protein substrates, (A) procaspase-3, (B) caspase-3 D175A C163A, and (C) procaspase-6 by plotting [ln (1-ρ)/t] versus E, where ρ is the % total band intensity of the ISL cleavage products, t is time in sec, and E is caspase-9 concentration in μM. The k values of caspase-9 to cleave the ISL of procaspase-3 and −6 were derived as the negative value of the slope, and for a peptide substrate, Ac-LEHD-afc, we reported cleavage rate (k) as k_cat/K_M from (D).

Prodomain of procaspase-6 does not play a significant role in protecting direct activation by caspase-9.

To test the hypothesis that the procaspase-6 prodomain might protect the procaspase-6 ISL from cleavage, we used caspase-6 ΔN C163S, an uncleaved form of caspase-6 lacking the prodomain. If the prodomain prevents procapase-6 ISL cleavage by caspase-9, then we should observe more ISL cleavage for caspase-6 ΔN C163S than what we observed for caspase-6 C163S (Figure 2C). The ISL cleavage of caspase-6 ΔN C163S by caspase-9 was almost undetectable (Figure 3A). From the comparison, we observed no significant difference in the ISL cleavage of caspase-6 C163S as compared to caspase-6 ΔN C163S cleavage by caspase-9 (Figure 3B). The estimated rate (k) values of caspase-9 to cleave the ISL of both constructs were < 25 M⁻¹s⁻¹ (Figure 3C). These data imply that the prodomain does not play a significant role in preventing cleavage of the procaspase-6 ISL by caspase-9. This suggests that the procaspase-6 ISL cleavage sites, site 1: ¹⁷⁶DVVD↓N and site 2: ¹⁹⁰TEVD↓A may contribute to substrate selection and should be tested individually.

Figure 3. Prodomain of procaspase-6 does not play significant role in protecting the ISL from direct activation by caspase-9.

(A) 3 μM of caspase-6 ΔN C163S was incubated with concentrations of caspase-9 ranging from 0 to 1 μM for 1 hour at 37 °C. No significant ISL cleavage was observed. (B) Comparison of ISL cleavage between procaspase-6 (Figure 2C) and caspase-6 ΔN C163S by caspase-9. Each experiment was performed twice on two separate days using different samples. The mean ± S.D. of these two replicates were used for statistical analysis using unpaired two-tailed t-test. The p value parameters are ns (not significant) for p ≥ 0.05 and * for p < 0.05. No significant difference (shown as ns) in the ISL cleavage was observed between the two constructs for all caspase-9 concentrations. (C) The kinetic rates (k) of caspase-9 to cleave the ISL of procaspase-6 and caspase-6 ΔN C163S were derived as described in Figure 2.

Caspase-9 does not recognize the procaspase-6 ISL site 1, ¹⁷⁶DVVD↓N.

Using sequence alignment, we compared the procaspase-6 ISL site 1 (¹⁷⁶DVVD↓N) with the procaspase-3 ISL site (¹⁷²IETD↓S) (Figure 4A). To understand the inability of caspase-9 to hydrolyze the procaspase-6 ISL site 1, we engineered a construct in which we introduced the site 1 sequence from procaspase-6 into procaspase-3 to generate caspase-3 C163A ¹⁷²IETD↓S to DVVDN. Using this construct, we asked whether the DVVDN sequence could be cleaved by caspase-9 if it was presented in the background of procaspase-3. We also engineered another construct by introducing the procaspase-3 ISL site into procaspase-6, caspase-6 C163S ¹⁷⁶DVVD↓N to IETDS. Using this construct, we can determine whether caspase-9 can recognize an appropriate procaspase-3 cleavage site sequence in the context of procaspase-6. Caspase-9 did not proteolyze caspase-3 C163A ¹⁷²IETD↓S to DVVDN (Figure 4B) suggesting that the sequence DVVDN is not optimized for cleavage by caspase-9. In contrast, caspase-9 can cleave the ISL of caspase-6 C163S ¹⁷⁶DVVD↓N to IETDS resulting in generation of cleavage products: Lg and small subunit + intersubunit linker (Sm+ISL) (Figure 4C). This indicated that the context of the 176 to 180 region is available to caspase-9 for recognition. A previous peptide-based profiling study⁵ demonstrated that the most preferred peptide-substrate sequence of caspase-9 is LEHD↓ (P4-P1). Therefore, we constructed a third construct, caspase-6 C163S ¹⁷⁶DVVD↓N to LEHDS, to analyze how efficiently caspase-9 can proteolyze the canonical consensus sequence. This construct was cleaved at the ISL by caspase-9 resulting in the formation of Lg and Sm+ISL (Figure 4D). No significant difference in the ISL cleavage was detected between caspase-6 C163S and caspase-3 C163A ¹⁷²IETD↓S to DVVDN (Figure 4E). The cleavage rate (k) of caspase-9 to proteolyze the ISL of caspase-3 C163A ¹⁷²IETD↓S to DVVDN was < 25 M⁻¹s⁻¹ (Figure 4F). Thus, it appears that the caspase-9 active-site is unable to recognize the sequence of procaspase-6 ISL site 1, ¹⁷⁶DVVD↓N. There was a significant difference in the ISL cleavage when procaspase-6 site 1 was replaced from ¹⁷⁶DVVD↓N to IETD/LEHD (Figure 4E). We did not observe significant changes in the ISL cleavage among the constructs, caspase-3 C163A, caspase-6 C163S ¹⁷⁶DVVD↓N to IETDS, and caspase-6 C163S ¹⁷⁶DVVD↓N to LEHDS (Figure 4E). Moreover, the ISLs of these three constructs were cleaved by caspase-9 with a similar estimated cleavage rate, k (Figure 2E and 4F). These results demonstrated that caspase-9 active-site can access the procaspase-6 ISL site 1; however, the sequence, DVVDN (P4-P1’) is unfavorable for proteolytic cleavage.

Figure 4. Caspase-9 does not recognize the sequence of procaspase-6 ISL site 1, ¹⁷⁶DVVD↓N.

(A) Sequence alignment of the ISL of procaspase-3 and −6. The ISL site of procaspase-3 (¹⁷²IETD↓S) and ISL site 1 of procaspase-6 (¹⁷⁶DVVD↓N) are shown in boxes. (B, C, and D) Different concentrations of recombinant caspase-9 ranging from 0 to 1 μM were incubated with 3 μM of (B) caspase-3 C163A ¹⁷²IETD↓S to DVVDN, (C) caspase-6 C163S ¹⁷⁶DVVD↓N to IETDS, and (D) caspase-6 C163S ¹⁷⁶DVVD↓N to LEHDS for 1 hour at 37 °C. In the last lane of each gel, 1 μM of caspase-9 alone was loaded as a control. Caspase-9 could cleave the ISL cleavage of (B) caspase-3 C163A IETDS to DVVDN was observed. In contrast, caspase-9 cleaved the ISL for (C) caspase-6 C163S ¹⁷⁶DVVD↓N to IETDS and (D) caspase-6 C163S ¹⁷⁶DVVD↓N to LEHDS resulting into Lg and small subunit + intersubunit linker (Sm+ISL). (E) Comparison of ISL cleavage among different constructs of caspase-3 and −6 by caspase-9. No significant difference was observed between ISL cleavage of procaspase-6 (Figure 2C) and caspase-3 C163A IETDS to DVVDN. No significant difference was observed among ISL cleavage of procaspase-3 (Figure 2A), caspase-6 C163S DVVDN to IETDS/LEHDS. Each experiment (B-D) was performed twice on two different days using different samples. The mean ± S.D. of the two replicates were used to perform statistical analysis using the unpaired two-tailed t-test as parameters defined in Figure 3B where ns is for p ≥ 0.05 and * is for p < 0.05. For t-test analysis, we compared each construct with the other four constructs individually. (F) Cleavage rate (k, as defined in Figure 2) of caspase-9 to cleave ISL sequence constructs of caspase-3 and −6.

Caspase-9 does not cleave the procaspase-6 ISL site 2, ¹⁹⁰TEVD↓A, mainly due to the local context.

We compared the procaspase-3 ISL site (¹⁷²IETD↓S) and procaspase-6 ISL site 2 (¹⁹⁰TEVD↓A) by aligning the sequences (Figure 5A). To investigate why caspase-9 does not cleave the procaspase-6 ISL site 2, which is the sequence first recognized by caspase-6,²⁵ we engineered constructs, caspase-3 C163A ¹⁷²IETD↓S to TEVDA, caspase-6 C163S ¹⁹⁰TEVD↓A to IETDS, and caspase-6 C163S ¹⁹⁰TEVD↓A to LEHDS (the canonical caspase-9 recognition motif⁵). Caspase-9 cleaved the ISL of caspase-3 C163A ¹⁷²IETD↓S to TEVDA resulting in the formation of cleavage products: N+Lg and Sm (Figure 5B). Though we see a band indicating removal of the prodomain (ΔN), the ISL cleavages for caspase-6 C163S ¹⁹⁰TEVD↓A to IETDS and caspase-6 C163S ¹⁹⁰TEVD↓A to LEHDS were not observed (Figure 5C and 5D). We observed a significant difference in the ISL cleavage between the constructs, procaspase-6 and caspase-3 C163A ¹⁷²IETD↓S to TEVDA (Figure 5E). Thus, caspase-9 can recognize the procaspase-6 ISL site 2 in the context of the procaspase-3 background. We observed a significant difference in the ISL cleavage between the constructs, caspase-3 C163A and caspase-3 C163A ¹⁷²IETD↓S to TEVDA (Figure 5E). Caspase-9 cleaved the ISL of procaspase-3 with much higher efficiency than caspase-3 C163A ¹⁷²IETD↓S to TEVDA (Figure 2E and 5E). Thus, caspase-9 cannot efficiently recognize the sequence of procaspase-6 ISL site 2, ¹⁹⁰TEVD↓A, even when it was presented in an optimal context in procaspase-3. No significant change in the ISL cleavage was observed among the constructs: caspase-6 C163S, caspase-6 C163S ¹⁹⁰TEVD↓A to IETDS, and caspase-6 C163S ¹⁹⁰TEVD↓A to LEHDS (Figure 5E). Caspase-9 cleavage rates, k, to cleave the ISL of these three constructs were < 25 M⁻¹s⁻¹ (Figure 2E and 5F). Thus, even replacing procaspase-6 ISL site 2 (¹⁹⁰TEVD↓A) to IETDS or LEHDS (the most preferred sequence of caspase-9 active-site for P4-P1 positions⁵), resulted in almost no ISL cleavage. These results strongly suggest that in addition to the sequence, the local context of procaspase-6 ISL site 2 (¹⁹⁰TEVD↓A) plays a vital role in blocking the access of caspase-9.

Figure 5. Caspase-9 does not cleave procaspase-6 ISL site 2, ¹⁹⁰TEVD↓A, mainly due to the local context.

(A) Sequence alignment of the ISL of procaspase-3 and −6. The ISL site of procaspase-3 (¹⁷²IETD↓S) and ISL site 2 of procaspase-6 (¹⁹⁰TEVD↓A) are shown in boxes. (B, C, and D) Recombinant caspase-9 ranging from 0 to 1 μM was incubated with 3 μM of (B) caspase-3 C163A ¹⁷²IETD↓S to TEVDA, (C) caspase-6 C163S ¹⁹⁰TEVD↓A to IETDS, and (D) caspase-6 C163S ¹⁹⁰TEVD↓A to LEHDS for 1 hour at 37 °C. In the last lane of each gel, 1 μM of caspase-9 was loaded as a control. We observe the ISL cleavage of (B) caspase-3 C163A ¹⁷²IETD↓S to TEVDA indicating that TEVDA can be recognized by caspase-9. The ISL cleavage of (C) caspase-6 ¹⁹⁰TEVD↓A to IETDS and (D) caspase-6 ¹⁹⁰TEVD↓A to LEHDS was very low to be detected. (E) Comparison of the ISL cleavage among different constructs of caspase-3 and −6 by caspase-9. No significant difference was observed between ISL cleavage of procaspase-6 and caspase-6 C163A ¹⁹⁰TEVD↓A to IETDS/LEHDS. Replacing ISL site of procaspase-3 from ¹⁷²IETD↓S to TEVDA resulted into significantly less ISL cleavage compared to procaspase-3 (Figure 2A). Each experiment (B-D) was performed twice on two different days using different samples. The mean ± S.D. of the two replicates were used to perform statistical analysis unpaired two-tailed t-test as defined in Figure 3C and 4E where ns is for p ≥ 0.05 and * is for p < 0.05. For the t-test analysis, each construct was compared with other four constructs individually. (F) Cleavage rate (k, as defined in Figure 2) for caspase-9 to cleave different constructs of caspase-3 and −6.

Caspase-9 active-site preference for P4-P1’ positions is IETDS > TEVDA > DVVDN

We used a previously solved crystal structure (PDB: 1JXQ) to model the caspase-9 active-site bound to different peptides.³⁷ This reported crystal structure has a peptide-based inhibitor, z-EVD-Dcbmk (benzoxycarbonyl-Glu-Val-Asp-dichlorobenzylmethylketone) bound at the active-site pocket occupying the P4-P1 positions (Figure 6A). Using the mutagenesis feature of PyMOL, we generated structural models of caspase-9 active-site binding to the peptides, IETD, TEVD, and DVVD (Figure 6B, 6C, and 6D). The S4 sub-site of the caspase-9 active-site contains residues, I396, Y397, W354, and W362, rendering it hydrophobic. Therefore, isoleucine, a hydrophobic residue, can be accommodated at the P4 position (Figure 6B). Aspartic acid present in DVVDN is a very hydrophilic residue that may resist residing inside the S4 sub-site (Figure 6C). Moreover, the presence of D356 nearby S4 sub-site may further decrease affinity for a negatively charged aspartic acid at the P4 position (Figure 6C). Threonine is uncharged and less hydrophilic than aspartic acid. Thus, the S4 sub-site may more readily accept threonine (Figure 6D). Peptides IETD and TEVD have an advantage of possessing glutamic acid as P3 position, which can interact with R355 via hydrogen bonding (Figure 6B and 6D). In contrast, valine, a hydrophobic residue, is not suitable for S3 sub-site binding (Figure 6C). The threonine of IETD peptide can interact with K292 by making a hydrogen bond (Figure 6B). Looking at the cavity of the S1’ sub-site, small sized residues such as glycine, serine and alanine are more preferred than asparagine as P1’ position (not shown in the figure because P1’ was not present in the original structure, Figure 6A). Therefore, the caspase-9 active-site preference for P4-P1’ positions is IETDS > TEVDA > DVVDN. Calculation of the solvent accessible surface area (SASA) also provides insights into recognition preferences (Table S1). While the cleavage site in procaspase-3 and site 1 in procaspase-6 are both highly accessible, site 2 in procaspase-6 is much less accessible. The structural models (Figure 6) and SASA analysis support our cleavage assays data that caspase-9 i) cleaves the procaspase-3 ISL site (¹⁷²IETD↓S) with a higher cleavage rate (Figure 2), ii) does not recognize the procaspase-6 ISL site 1 (¹⁷⁶DVVD↓N) (Figure 4), iii) can cleave the sequence of procaspase-6 ISL site 2, ¹⁹⁰TEVD↓A, with a lower cleavage rate (Figure 5), and iv) procaspase-6 site 2 is buried by the local context which play important role in preventing the cleavage by caspase-9.

Figure 6. Caspase-9 active-site preference for P4-P1’ positions in substrate is IETDS > TEVDA > DVVDN.

(A, B, C, D) The active-site of caspase-9 accommodating peptides, (A) z-EVD-Dcbmk (benzoxycarbonyl-Glu-Val-Asp-dichlorobenzylmethylketone), PDB: 1JXQ,³⁷ (B) IETD, (C) DVVD, (D) and TEVD were modeled as residues in the P4-P1 positions. IETD, DVVD, and TEVD peptides were generated using the mutagenesis feature of PyMOL. The hydrogen bonds between each peptide residue and caspase-9 active-site residues are shown as black dashes. The threonine in (B) IETD peptide is predicted to make a hydrogen bond with K292. Thus, the P2 position of IETD is more preferred by caspase-9 compared to other peptides in (A, C, and D). Due to the presence of valine as P2 position, fewer hydrogen bond interactions with R355 is predicted to form for (C) DVVD peptide compared to other peptides in (A, B, and D). The S4 sub-site of caspase-9 is hydrophobic consisting of residues I396, Y397, W354, and W362. Hydrophobic residues at P4 position in (A) z-EVD-Dcbmk and (B) IETD peptides are preferred; however, aspartate (a very hydrophilic residue) at the P4 position in (C) DVVD peptide is expected to lack enthalpically-driven interactions, as illustrated by red clashes.

Discussion:

The order of caspases activation during the intrinsic apoptotic pathway has been studied in-depth previously.^22,38 Based on these studies, we created a schematic diagram of intrinsic apoptosis focusing on caspase-9 activation via apoptosome complex and the order of activation of the executioners (caspase-3, −6, and −7) (Figure 7A). Once triggered via formation of the apoptosome, caspase-9 mediates the activation of procaspase-3 and −7. In the case of procaspase-3, caspase-9 cuts the ISL site (¹⁷²IETD↓S) (Figure 2A and 2B). This proteolytic cleavage event at the ISL is required to activate procaspase-3 ^39,40 (Figure 7B). After this ISL cleavage, to achieve complete maturation, the prodomain of caspase-3 is removed (first cleavage at D9 and then at D28) likely via self-proteolysis or caspase-3-like activity.^41,42 Activated caspase-3 further processes caspase-9 via a feedback mechanism – a cleavage event at D330²² (Figure 7A). Procaspase-7 contains three cleavage sites: D23, D198, and D206.⁴³ In order to activate procaspase-7, prodomain (residues: 1-23)⁴⁴ and then N-terminal peptide (residues: 23-28)⁴⁵ are removed perhaps by active caspase-3. These cleavage events facilitate procaspase-7 activation via ISL cleavage by initiator caspases (e.g., caspase-9 and −8). Caspase-9 activates procaspase-7 (Figure 7A) via cleavage at ISL site 1 (¹⁹⁵IQAD↓S) and ISL site 2 (²⁰³NDTD↓A).⁴⁵ The cleavage rates (k) for caspase-9 to cut procaspase-7 ISL site 1 (¹⁹⁵IQAD↓S) and ISL site 2 (²⁰³NDTD↓A) were derived to be ~0.4 x 10⁴ M⁻¹s⁻¹ and < 200 M⁻¹s⁻¹, respectively.³³ Thus, caspase-9 proteolyzes procaspase-7 ISL site 1 with a much higher cleavage rate than ISL site 2. Swapping the procaspase-7 ISL site 2 from ²⁰³NDTD↓A to LEHDA robustly enhanced the cleavage rate of caspase-9 (from < 200 M⁻¹s⁻¹ to ~0.1 x 10⁴ M⁻¹s⁻¹)³³ exemplifying the importance of the cleavage site motif sequence in proteolysis. Nevertheless, this improved cleavage rate of caspase-9 for this swapped construct of procaspase-7 was about four times less than the cleavage rate to hydrolyze at ISL site 2, ¹⁹⁵IQAD↓S.³³ Thus, the ISL site 1 of procaspase-7 appears to be more accessible to caspase-9 active-site than the ISL site 2, illustrating the importance of local context of a cleavage site in the process of proteolysis.

Figure 7. In the intrinsic apoptotic pathway, caspase-9 cannot directly activate caspase-6 due to the sequence of ISL cleavage site 1 and the local context of ISL cleavage site 2.

(A) Schematic diagram adapted from the literature illustrating the order of caspase activation during the intrinsic path of apoptotsis.^22,38 Mitochondrial stress releases cytochrome c, which interacts with Apaf-1 (apoptotic protease activating factor 1, colored orange) to form the apoptosome complex which recruits procaspase-9 to mediate activation of caspase-9. Caspase-9, then activates caspase-3 and −7; however, it is unable to activate caspase-6 directly. Activated caspase-3 processes procaspase-9 via feedback mechanism. Caspase-3 also cleaves and activates procaspase-6. In the absence of active caspase-3, activated caspase-7 activates procaspase-6. The apoptotic executioners (caspase-3, −6, and −7) cleave their respective downstream substrates to evoke apoptosis. (B) Schematic diagram showing caspase-9 activating procaspase-3 by the ISL cleavage at D175. (C) Schematic diagram showing caspase-9 can access D179 of procaspase-6 but the sequence, DVVDN, is uncleavable (P4-P1’ cleavage sites for each monomer are shown as transparent blue spheres). Caspase-9 is unable to hydrolyze D193 due to the local context (inaccessibility is shown as clashes). Though caspase-9 can remove the prodomain of procaspase-6, the resultant product (ΔN caspase-6) is inactive which cannot hydrolyze downstream substrates.³⁵

The inability of caspase-9 to directly activate procaspase-6 (Figure 7A) has been reported in previous studies.^22-24,36 Proteolytic cleavage at any of the ISL cleavage sites (either at D179 or D193) of procaspase-6 is a prerequisite to activate procaspase-6.^21,35 Moreover, ISL cleavage of procaspase-6 results in structural-dynamics changes rendering the active-site more accessible to solvent.⁴⁶ Based on our results from this manuscript (Figure 2, 4, and 5), we prepared a schematic diagram showcasing the interactions between caspase-9 and procaspase-6 (Figure 7C). We demonstrate that the procaspase-6 ISL site 1 (¹⁷⁶DVVD↓N) is accessible (in other words, the local context of this cleavage site is not creating hindrance) to the active-site of caspase-9. However, the sequence of this cleavage site (DVVDN as P4-P1’), illustrated as transparent blue spheres for each monomer (Figure 7C), is unrecognizable (Figure 4). We also found that the active-site of caspase-9 can recognize the sequence of the procaspase-6 site 2 (¹⁹⁰TEVD↓A). Nonetheless, the local context of this ISL cleavage site, shown as clashes (Figure 7C), is blocking the access of caspase-9 active-site (Figure 5). Thus, both the sequence as well as the local context of procaspase-6 ISL cleavage sites play vital roles in preventing direct activation by caspase-9. We observed that caspase-9 only partially cleaved the prodomain of procaspase-6, but not any other sites, although caspase-9 was present at concentrations above those expected physiologically (Figure 2C). The removal of only the prodomain is not sufficient for procaspase-6 activation.³⁵ Thus, procaspase-6 proteolysis by caspase-9 simply results in an inactive ΔN version of caspase-6 (Figure 7C).

Upon its activation by caspase-9, caspase-3 hydrolyzes procaspase-6 into mature caspase-6^22,23 (Figure 7A). Active caspase-3 proteolyzes procaspase-6 at all three cleavage sites: D23, D179, and D193.^23,25 Among these three cleavage sites of procaspase-6, caspase-3 directly cuts at D23 and D179, and only then proteolysis at D193 can occur.^23,25 These findings strongly suggested that the procaspase-6 ISL site 1 (¹⁷⁶DVVD↓N) is readily accessible and recognizable to the active-site of caspase-3. Moreover, the local context of procaspase-6 ISL site 2 (¹⁹⁰TEVD↓A) initially blocks the caspase-3 active-site, and only allows the access after prior cleavage event at D179,²³ although is it difficult to disentangle the contribution of a more favorable recognition sequence. Activated caspase-6 can also proteolyze procaspase-6 intermolecularly at all three cleavages sites: D23, D179, and D193.²⁵ Among these three cleavage sites, active caspase-6 robustly cleaves at D23 but inefficiently proteolyzes at D179 and D193.^23,25 Active caspase-6 was able to hydrolyze the procaspase-6 ISL site 1 with much higher efficiency when the sequence ¹⁷⁶DVVD was substituted with TETD (P4-P1 sequence of the procaspase-6 prodomain).²⁵ Thus, as we determined for caspase-9 (Figure 4), the sequence of procaspase-6 site 1 is not well recognized by caspase-6 via intermolecular recognition. Nevertheless, when it engages in an intermolecular interaction, in contrast to caspase-3, active caspase-6 can cut procaspase-6 at D193 without any need for prior cleavage at D179.^23,25 Thus, unlike active caspase-9 (Figure 5) and caspase-3,^23,25 active caspase-6 can access and directly cleave the procaspase-6 ISL site 2 (¹⁹⁰TEVD↓A). Despite the fact that caspase-3 hydrolyzes its preferred fluorogenic peptide substrate (DEVD-ase activity)⁴⁷ with ~5-fold higher k_cat/K_M value than caspase-6 (VEID-ase activity),^48,49 the inability of caspase-3 and the ability of caspase-6 to directly proteolyze the procaspase-6 ISL site 2 via intermolecular interactions signifies the importance of the local context of cleavage sites in the process of proteolysis. Early work in the field showed that caspase-3 but not caspase-7 activated caspase-6.⁵⁰ More recently, in at least one study of the intrinsic apoptotic pathway,³⁸ in the absence of caspase-3, activated caspase-7 was able to activate procaspase-6 to caspase-6 (Figure 7A); however, the molecular factors facilitating this activation remain to be identified.

Peptide-based screening and proteomics-wide studies of an individual protease are insightful in designing active-site inhibitors/peptide substrates^5,26,51 and obtaining the list of protein substrates,^52-54 respectively. However, this information is not enough to completely understand the substrate preference of an individual protease. Such knowledge at a molecular level can be determined by performing biochemical and/or biophysical studies on a protease and its substrates. Interactions between many proteases and their protein substrates (e.g., caspase-9 and the executioners, procaspase-3, −6, and −7) are transient; therefore, deriving the structural information of their complexes using biophysical techniques such as X-ray crystallography, NMR, and cryo-EM is challenging (e.g., currently, there is no structure deposit of the complex of caspase-9 and any of its protein substrates in the Protein Data Bank, https://www.rcsb.org/). In such cases, using site-directed mutagenesis coupled with protein substrate digestion assays, as we have undertaken herein, can help in elucidating the mechanisms of biological pathways at a molecular level. This study demonstrates two key points: i) That the molecular architecture of procaspase-6 renders it insensitive to caspase-9 activation, although its ISL cleavage sites are similarly positioned (within the ISL) to those in procaspase-3/-7. ii) That sequence as well as the local context of the cleavage sites enables productive proteolytic activation.

Abbreviations.

P

peptide residues of the substrates

S

sub-sites of protease active-site

ISL

intersubunit linker

N

N-terminal prodomain

ΔN

lacking prodomain

Lg

large subunit

Sm

small subunit

Ac-LEHD-afc

N-acetyl-Leu-Glu-His-Asp-7-amido-4-trifluoromethylcoumarin

z-EVD-Dcbmk

benzoxycarbonyl-Glu-Val-Asp-dichlorobenzylmethylketone

z-VAD-fmk

carbobenzoxy-Val-Ala-Asp-fluoromethylketone

SASA

solvent accessible surface area