Báo cáo khoa học: Caspase-2 is resistant to inhibition by inhibitor of apoptosis proteins (IAPs) and can activate caspase-7 pot

An artificially autoacti-vating version of caspase-2 rev-caspase-2, in which the small subunit precedes the prodomain and large subunit [29], killed yeast readily lane 6.. The catalyti-ca

apoptosis proteins (IAPs) and can activate caspase-7

Po-ki Ho1,2,3, Anissa M Jabbour1,2,3, Paul G Ekert1,4,5and Christine J Hawkins1,2,3

1 Murdoch Children’s Research Institute, Parkville, Australia

2 Children’s Cancer Centre, Royal Children’s Hospital, Parkville, Australia

3 Department of Paediatrics, University of Melbourne, Parkville, Australia

4 Department of Neonatology, Royal Children’s Hospital, Parkville, Australia

5 The Walter and Eliza Hall Institute, Royal Melbourne Hospital, Parkville, Australia

The caspases are a family of cysteine proteases that

typically cleave their substrates at aspartate residues

[1] Subclassification of family members has been based

on various criteria including substrate specificity or

structural features For example, caspases-1, -4 and -5

are involved in the proteolytic maturation of cytokines

including pro-interleukin-1b [2] and pro-interleukin-18

[3] Caspases-8 and -9 are components of cell death

signal transduction pathways and are classified as

api-cal caspases The primary role of these proteases, each

of which has a long prodomain containing a protein

interaction motif, is to proteolytically activate distal

caspases (such as caspase-3 and caspase-7), which then

catalyse the cleavage of numerous cellular substrates

[4] Despite being the second identified member of the

caspase family, the function of caspase-2 (Nedd-2⁄ Ich-1) remains somewhat elusive Its substrate prefer-ence more closely aligns with that of the pro-apoptotic caspases than their cytokine processing relatives [5] Of the mammalian caspases, caspase-2 is the most similar

to the nematode apoptotic caspase, CED-3 This would also tend to imply that caspase-2 plays a pro-apoptotic role, yet caspase-2 deficient mice have

an extremely subtle phenotype, arguing against a non-redundant role in programmed cell death [6,7]

Caspase-2 has recently received considerable attention, as several groups have sought to define its biological role in apoptosis signalling Overexpressing caspase-2 provoked the release of pro-apoptotic mole-cules (including cytochrome c) from mitochondria [8],

Keywords

caspase-2; protease; caspase-7;

S cerevisiae; enzyme activity

Correspondence

C Hawkins or P Ekert, Murdoch Children’s

Research Institute, Royal Children’s

Hospital, Flemington Road, Parkville, VIC

3052 Australia

Fax: +61 3 9345 4993 (CH); +61 3 9347

0852 (PE)

Tel: +61 3 9345 5823 (CH); +61 3 9345

2548 (PE)

E-mail: chris.hawkins@mcri.edu.au;

paul.ekert@mcri.edu.au

(Received 10 November 2004, revised 7

January 2005, accepted 18 January 2005)

doi:10.1111/j.1742-4658.2005.04573.x

Caspases are a family of cysteine proteases with roles in cytokine matur-ation or apoptosis Caspase-2 was the first pro-apoptotic caspase identified, but its functions in apoptotic signal transduction are still being elucidated This study examined the regulation of the activity of caspase-2 using recombinant proteins and a yeast-based system Our data suggest that for human caspase-2 to be active its large and small subunits must be separ-ated For maximal activity its prodomain must also be removed Consistent with its proposed identity as an upstream caspase, caspase-2 could provoke the activation of caspase-7 Caspase-2 was not subject to inhibition by members of the IAP family of apoptosis inhibitors

Abbreviations

AFC, 7-amino-4-trifluoromethyl coumarin; CARD, caspase activation and recruitment domain; GST, glutathione-S-transferease.

ferred by Apaf-1 downregulation, against apoptosis

induced by DNA damage [12] The involvement of

caspase-2 in TRAIL-induced apoptosis has also been

reported recently, placing this enzyme upstream of Bid

cleavage in the pathway [13]

Like caspase-9, caspase-2 bears a caspase activation

and recruitment domain (CARD) in its amino-terminal

prodomain The role of the CARD (in caspase-9 at

least) is to permit binding to aggregated adaptor

pro-teins, leading to autoactivation through ‘induced

proxi-mity’ [14] Consistent with this, forced dimerization of

caspase-2 provoked its activation [15], and fusing the

caspase-2 prodomain to caspase-3 resulted in caspase-3

autoactivation [16] Recent findings by Baliga et al

indicated that dimerization is the key determinant for

initial activation of murine pro-caspase-2 [17] The

phy-siological mechanism through which the prodomain

might trigger activation of caspase-2 is still unclear A

molecular pathway has been proposed to link

caspase-2 to members of the tumour necrosis factor receptor

family via an adaptor molecule (RAIDD⁄ CRADD)

and intermediaries (RIP, TRADD, FADD and

TRAFs) [18,19] However, this has not been directly

demonstrated and death ligand-mediated apoptosis

proceeds normally in caspase-2-deficient cells [7] Other

putative caspase-2 adaptors have been proposed

[20,21], but verification of their relevance in

physiologi-cal settings has not yet been published Tinel and

Tschopp recently reported a complex they designated

the ‘PIDD-osome’ comprising caspase-2, RAIDD and

PIDD, the formation of which promoted apoptosis

following p53-dependent DNA damage [22] Further,

caspase-2 is recruited into a high molecular weight

complex independent of the apoptosome components

Apaf-1 and cytochrome c [23] It has also been recently

postulated that caspase-2 may influence apoptosis [24]

and⁄ or nuclear factor-jB activation [25] through

mech-anisms unrelated to its enzymatic activity

If caspase-2 functions as an apical caspase, it may

process and activate downstream caspases We sought

to characterize the molecular events downstream of

human caspase-2 activation In particular we focused

on the susceptibility of caspase-2 to suppression by

Results

High level expression of pro-caspase-2 is lethal

in yeast Properties of caspase-2 were assessed using a yeast-based system we have previously exploited to character-ize other caspases and apoptotic pathways [26–28] This system capitalizes on the observation that some caspases kill yeast upon enforced high-level expression

In order for caspases to kill yeast, they must both be able to autoactivate and their proteolytic specificity must permit cleavage of essential yeast proteins To assess the activity of caspase-2 in yeast, various con-structs encoding different forms of the protein (Fig 1) were transformed into yeast (Fig 2A) Expression of pro-caspase-2 using the Gal 1⁄ 10 promoter affected yeast growth only marginally (compare growth in lane

2 to that of an empty vector transformant in lane 1) Increasing the pro-caspase-2 expression level, by intro-ducing an additional expression construct under dif-ferent nutritional selection, elicited more substantial lethality (lane 3) A caspase-2 cleavage site mutant (D152A), from which the prodomain could not be removed, was also expressed at a high level using two plasmids with different nutritional selections Com-pared with equivalent expression of wild-type pro-caspase-2 (lane 3), this mutant exhibited only marginal toxicity (lane 4) suggesting that removal of the prodo-main contributes to full enzymatic activity Consistent with this observation, a truncation mutant lacking almost all of the caspase-2 prodomain (caspase-2D1)149) killed yeast more efficiently than full-length caspase-2 (compare lane 7 with lane 2) An artificially autoacti-vating version of caspase-2 (rev-caspase-2), in which the small subunit precedes the prodomain and large subunit [29], killed yeast readily (lane 6) The catalyti-cally inactive mutant pro-caspase-2C303Awas unable to kill yeast (lane 5) implying that the lethality of wild-type caspase-2 in yeast was due to its enzymatic activity The expression of the prodomain (caspase-2D150)435) had no effect on yeast viability (lane 8)

To investigate the auto-processing of pro-caspase-2

in yeast, we immunoblotted lysates obtained from

yeast expressing these different forms of caspase-2 with

an antibody recognizing an epitope in the large

subunit In lysates from yeast expressing wild-type

pro-caspase-2, a partial cleavage product was detected,

in addition to the fully processed large subunit

(Fig 2B) Like the wild-type enzyme, the cleavage site

mutant pro-caspase-2D152A was processed efficiently

between the large and small subunits, however, the

mutation at D152 prevented it from being further

processed to separate the prodomain from the large

subunit Caspase-2C303A remained intact as a result of

the abolished catalytic activity Rev-caspase-2, despite

its ability to efficiently kill yeast, was only incompletely

processed A proportion of caspase-2D1)149was cleaved

to remove the small subunit, thereby permitting

detec-tion of the dissociated large subunit

The activities of these different forms of caspase-2

were also analysed biochemically using a fluorogenic

caspase-2 substrate In this assay, the activity of an

enzyme is reflected by the efficiency with which it

cleaves the substrate to release free

7-amino-4-trimethyl coumarin (AFC) The caspase-2-specific

fluoro-genic synthetic peptide Z-VDVAD-AFC was used as a

substrate to assess caspase-2 activity [5] VDVADase activity was detected in lysates from yeast expressing all forms of caspase-2 that were capable of autoprocessing (Fig 2C) The most lethal forms of caspase-2 had the highest VDVADase activity (lanes 3, 6 and 7), while ly-sates from yeast that survived (lanes 1, 5 and 8) did not cleave the peptide substrate Yeast transformed with one wild-type caspase-2 plasmid or the D152A mutant were killed only inefficiently, however, their lysates exhibited significant VDVADase activity This may indi-cate that the biochemical assay is a more sensitive meas-ure of caspase-2 activity than the yeast death assay

Caspase-2 is not inhibited by mammalian IAP proteins

Members of the mammalian IAP family contribute to the regulation of apoptotic pathways in part by their inhibition of caspases-3, -7 and -9 [30] Other mamma-lian caspases (-1, -6, -8 and -10) are known to be resist-ant to inhibition by IAPs [30], but the susceptibility of caspase-2 to direct inhibition by IAPs has not been reported to date To explore the sensitivity of caspase-2

to IAP inhibition, we tested whether coexpression of IAPs would suppress caspase-2-dependent yeast death

Fig 1 Schematic illustration of the

caspase-2 proteins used in this study Mutated

resi-dues are listed above wild-type caspase-2

and are depicted with black circles.

We had previously established that the inhibitors p35

and p49 could rescue yeast from caspase-2 mediated

death [31], so these baculoviral proteins were used as

positive controls Caspase-3 effectively killed yeast and

this could be blocked by XIAP (also known as hILP),

MIHB (cIAP-1⁄ hIAP-2 ⁄ BIRC2) and MIHC

(cIAP-2⁄ hIAP-1), as well as p35 and p49 (Fig 3A) In contrast,

the mammalian IAPs could not inhibit yeast death

induced by expression either of full-length pro-caspase-2

(Fig 3B) or of truncated caspase-2 lacking the

prodo-main (Fig 3C) As expected, the baculoviral caspase

inhibitors p35 and p49 protected caspase-2-expressing

yeast (Fig 3B,C)

To confirm these observations using a biochemical

approach, purified caspase-2 was mixed with

recombin-ant XIAP or the inactive mutrecombin-ant XIAPD148A[32], then

assayed for its ability to cleave the fluorogenic

penta-peptide substrate Z-VDVAD-AFC Caspase-2 activity was not affected by the presence of XIAP (Fig 3D), whereas XIAP significantly reduced the activity of caspase-3, as demonstrated previously [33] The pres-ence of p35 led to a decrease in both caspase-2 and caspase-3 activities Inactive mutants of p35 (p35D87A) and XIAP (XIAPD148A) were unable to inhibit either caspase

Caspase-2 can promote caspase-7 catalytic activity

To explore the potential for caspase-2 to functionally interact with other caspases, we exploited the dose-dependent caspase-2-mediated yeast toxicity illustrated

in Fig 2 Caspase-2 was coexpressed in yeast from a single plasmid either alone (yielding weak lethality) or

B

C

Fig 2 Caspase-2 kills yeast (A) A semi-quantitative assay compares the effect of transgenes on yeast growth and viability Yeast cells were transformed with the indi-cated plasmids Suspensions of each trans-formant were prepared at standardized concentrations Serial dilutions were made and spotted onto solid inducing minimal media vertically down the plate Colony size indicates growth rate and colony number reflects cell viability (In every experiment, each dilution was also spotted onto a repressing plate to verify that equivalent numbers of each transformant were spot-ted; data not shown) (B) Anti-caspase-2 immunoblotting of lysates from the indica-ted transformants The presumed identities

of each band are shown to the left (pro, pro-domain; L, large subunit; S, small subunit) (C) The ability of caspase-2 to cleave the flu-orogenic peptide substrate Z-VDVAD-AFC Native lysates obtained from yeast were incubated with Z-VDVAD-AFC Fluorescence was monitored over time and the maximal rate of increase in free AFC was calculated and graphed Error bars indicate SD (n ¼ 4).

together with the nonlethal caspases-3, -4, -6, -7 and -9

(Fig 4A) Yeast death was used as an indicator of

caspase activity Co-expression of caspase-2 with

caspase-7 led to a pronounced increase in yeast death,

compared to that triggered by either caspase alone

(compare lane 12 with lanes 2 and 11) Much weaker

synergy was also reproducibly observed between

caspase-2 and -3 (compare lane 6 with lanes 2 and 5)

We then tested the ability of lysates from these yeast

to cleave a fluorogenic caspase-3 substrate

(Ac-DEVD-AFC) or a caspase-2 substrate (Z-VDVAD-(Ac-DEVD-AFC)

Caspase-2 activity was not enhanced by coexpression

of caspases-3 or -7 However, significantly more

clea-vage of Ac-DEVD-AFC was observed when caspase-2

was coexpressed with caspase-7 (or, to a lesser extent

with caspase-3) (Fig 4B)

To further investigate the apparent synergy between

caspase-2 and caspase-7, plasmids encoding different

forms of these enzymes were transformed into yeast in

various combinations and their effects on enzyme

clea-vage, enzyme activity and yeast growth determined

(Fig 5) As before, high level expression of caspase-2

resulted in an active enzyme, able to efficiently kill

yeast, whereas lower expression levels of caspase-2 had

in vitro activity but weak killing activity (compare

lanes 2 and 4 in Figs 5A–C) Full length caspase-7 was unprocessed and did not kill yeast (lane 9), whereas caspase-7 coexpressed with caspase-2 was activated and toxic to yeast (lane 5) The activation of caspase-7

by caspase-2 depended on caspase-2 catalytic activity since coexpression of catalytically inactive caspase-2 with caspase-7 did not yield enzymatic activity (neither VDVADase nor DEVDase) and did not kill yeast (Figs 5A–C, lane 6) However, caspase-2 activation was independent of caspase-7 as caspase-2 proteolytic activity was the same in the presence of active or enzy-matically inactive caspase-7 (compare Fig 5B and C lanes 5 and 7) Two positive controls were used for caspase-7 activation First, caspase-7D1)53, which lacks the prodomain region and is constitutively active in mammalian cells [34] and in yeast [35] (lane 10) Second, as previously reported for caspase-3 [27], caspase-7 was activated by a constitutively active caspase-9 (rev-caspase-9) (lane 11) This autoacti-vating 9 protein, which could activate

caspase-3 [27] or caspase-7 (Fig 5A–C, lane 11), was not able

to co-operate with caspase-2 to kill yeast (Fig 4A, lane 14) Together, these data suggest that caspase-2 may lie upstream of caspase-7, and not downstream of caspase-9, in apoptotic pathways

A

B

C

D

Fig 3 IAPs do not inhibit caspase-2 The

caspase expression plasmids used to kill

yeast were (A) Caspase-3-lacZ (B)

pGALL-(LEU2)-caspase-2 with

pGALL-(URA)-cas-pase-2 or (C) pGALL-(URA)-caspGALL-(URA)-cas-pase-2D1)149.

Yeast transformed with the indicated

plas-mids were spotted as described in the

legend to Fig 1 (D) The indicated

combina-tions of caspase, fluorogenic substrate and

inhibitor were mixed together and the

fluor-escence resulting from the

caspase-medi-ated substrate cleavage was monitored and

calculated as described in the legend to

Fig 1 Error bars indicate SD (n ¼ 3).

The relationship between caspase-2 processing

and enzymatic activity

Previous work had illustrated that human

pro-caspase-2 can be processed at residue D152 to

remove its prodomain, and at residues D316 and

D330 to dissociate the large and small subunits and

release a small linker peptide [36] To investigate the

impact of these cleavage events on the enzymatic

activity of caspase-2, recombinant caspase-2 proteins

harbouring one or a combination of mutated D152,

D316 or D330 residues were generated and expressed

in bacteria and the protein purified (Figs 1 and 6A)

We observed that full length recombinant caspase-2 has about a 10-fold lower activity than a commonly used amino-terminal truncation lacking most of the prodomain (D1–149) [37] (Fig 6B) We therefore used this truncated caspase-2 to test the effects on activity

of mutating the D152, D316 and D330 residues We immunoblotted the purified caspase-2 enzymes with

an antibody recognizing an epitope within the large subunit of caspase-2, to determine whether enzyme autoprocessing had occurred (Fig 6A) and then tes-ted cleavage of a caspase-2 specific fluorogenic sub-strate (Fig 6B) to determine enzyme activity Retention of at least one cleavage site between the

B

Fig 4 Co-expression of caspases-2 and -7 enhances yeast lethality (A) The indicated combinations of caspases were coexpre-ssed in yeast and their ability to promote yeast death was compared to the lethality arising from expression of single caspases (B) The activities of caspases were assayed

in lysates from yeast expressing individual caspases, or the indicated combinations of caspases Substrate cleavage was calcula-ted from the maximal rate of free AFC released through cleavage by 30 lg of yeast lysate Error bars indicate SD (n ¼ 3).

large and small subunits (D316 and⁄ or D330)

permit-ted autocatalytic separation of the subunits (Fig 6A)

and yielded active enzymes (Fig 6B) Fusion of the

linker to the small subunit (D330A) had a slightly

greater deleterious effect on enzyme activity than

fusion to the large subunit (D316A) (Fig 6B) In

contrast, mutation of both D316 and D330 sites abolished auto-processing (Fig 6A) and dramatically reduced enzymatic activity (Fig 6B) Using higher amounts of enzyme (100 nm), it was evident that mutation of both of these cleavage sites decreased activity by 840-fold (data not shown) The C303A

A

B

C

Fig 5 Caspase-2 activates caspase-7 in

yeast (A) The indicated plasmids were

transformed into yeast and transformants

spotted onto inducing medium to visualize

their impact on yeast growth (B)

Immuno-blotting was used to detect caspase

pro-cessing The presumed identities of each

band are shown to the left (pro, prodomain;

L, large subunit; S, small subunit) (C) The

abilities of the yeast lysates to cleave the

caspase-2 (Z-VDVAD-AFC) and caspase-7

(Ac-DEVD-AFC) substrates were assayed.

Substrate cleavage was calculated from the

maximal rate of free AFC released through

cleavage by 30 lg of yeast lysate Error bars

indicate standard deviations (n ¼ 3).

active site mutant was completely inactive (Fig 6B),

even at 100 nm (data not shown) Full length

clea-vage site mutants were expressed from two plasmids

in yeast and their impact on yeast viability assayed

The D316A and D330A single mutants were toxic

to yeast, however, yeast expressing the double D316,

330A mutant survived (Fig 6C) Together, these data

suggest that the proteolytic activity of human

caspase-2 correlates with the degree to which the large subunit is separated from the small subunit

We also tested the abilities of the caspase-2 mutant enzymes to cleave protein substrates Cellular sub-strates (Bid, PARP, catalytically inactive pro-caspase-2 and pro-caspase-7) were expressed as glutathione-S-transferease (GST)-fusion proteins, incubated with the various caspase-2 enzymes and subjected to

B

C

Fig 6 Caspase-2 processing is necessary for activation (A) Caspase-2 enzymes with the indicated mutations were generated in bacteria and immunoblotted to determine the extent of auto-processing The pre-sumed identities of each band are shown to the right (pro, prodomain; L, large subunit;

S, small subunit) (B) The abilities of wild-type recombinant caspase-2 or the indicated mutants to cleave the fluorogenic substrate Z-VDVAD-AFC were monitored as described

in previous legends Two independent prep-arations of each enzyme were used (C) The indicated plasmids encoding wild-type or cleavage site mutants of casapse-2 (or empty vectors) were transformed into yeast and transformants spotted onto inducing medium to visualize their impact on yeast growth.

SDS⁄ PAGE Cleavage of the substrates was assessed

by staining with Coomassie blue (Fig 7A) and

immu-noblotting (Fig 7B) All caspase-2 proteins that were

active in the fluorogenic assay (Fig 6B) were also able

to cleave Bid at aspartate 60 [10] and a catalytically

inactive GST-tagged pro-caspase-2 (Fig 7A) The size

of the cleavage product implied that processing

occurred between the large and small subunits PARP

was cleaved by caspases-3 and -7, but not by

caspase-2 GST-tagged pro-caspase-2C303A was not processed

by caspases-3, -7 or -8; cleavage products were not

detected by Coomassie blue staining (Fig 7A) or by

immunoblotting (Fig 7B) (Processing of PARP or Bid

by these enzymes confirmed that they were active)

Having observed the activation of caspase-7 by

caspase-2 in yeast, we examined the processing of

GST-tagged pro-caspase-7C186A by caspase-2 in this

system The cleavage of GST-pro-caspase-7C186A by

active caspase-2 (as well as by caspases-3 and -8) was

detected by immunoblotting with an antibody that

recognizes cleaved caspase-7 (Fig 7B)

Discussion

A unique merit of the yeast system used here is that it

is free from the potential interference of other mamma-lian apoptotic signal transduction pathway compo-nents, allowing the expression of the gene of interest in

a naive yet eukaryotic cell-based environment We have previously used this system to reconstitute caspase-9 activation by Apaf-1 [27] and the core nematode pro-grammed cell death pathway [28] Here, we harnessed this system to analyse the regulation of caspase-2 activ-ity, exploiting the observation that overexpressed caspase-2 kills yeast in a concentration dependent man-ner, requiring a catalytically active enzyme Purified, recombinant proteins were also used to verify much of the data generated from the yeast system

We have shown that prodomain removal increases caspase-2 activity, when expressed in yeast or in bac-teria For generation of active human caspase-2, pro-cessing is also required between the small and large subunits (at D316 and⁄ or D330) Mutation of either site had little effect on enzyme activity or toxicity to yeast but mutation of both sites abolished both enzyme activity and yeast killing These observations differ somewhat from previously reported analyses of murine caspase-2 Firstly, the human and mouse enzymes vary in their propensity for autoprocessing between the large subunit and the linker We have shown that human caspase-2 almost completely auto-processes at this point (D316), as indicated by the effi-cient separation of large and small subunits of the D330A mutant In contrast, mutation of the murine equivalent of the human residue D316 (D333) alone prevented autoprocessing of caspase-2 [17,38] This species difference persisted when the mouse and human mutants were generated using the identical bac-terial expression systems (B Baliga and S Kumar, personal communication), ruling out any technical explanations for the variation Secondly, human caspase-2 which was prevented from autoprocessing between the large and small subunits was almost totally inactive, however, the D333G mutant of murine caspase-2 that could not autoprocess retained about one-fifth of wild-type enzyme activity [17,38] Further investigations will hopefully clarify the mechanisms underlying these curious species differences

Coexpression of caspase-2 with caspase-7 in yeast was significantly more toxic than expression of either protein alone Although this result could reflect an additive effect of two mildly lethal stimuli, two pieces of evidence suggest that caspase-2 activation of caspase-7 accounts for the combined lethality Firstly, caspase-2 cleaved caspase-7 in vitro Secondly, lysates from yeast

A

B

Fig 7 Substrate cleavage by wild-type caspase-2, its cleavage site

mutants and other caspases (A) GST-tagged, enzymatically inactive

pro-caspase-2 or caspase substrates were incubated with the

indi-cated purified recombinant caspases (as detailed in the

experimen-tal section) The reactions were then subjected to SDS⁄ PAGE and

the gels stained with Coomassie blue to visualize cleavage (B) The

more sensitive technique of immunoblotting was used to detect

cleavage of catalytically inactive pro-caspase-2 or pro-caspase-7.

and -8 capable of efficiently processing known

physio-logical substrates (PARP or Bid) could not cleave an

inactive mutant of pro-caspase-2 This discrepancy

probably relates to differences in the relative

concen-trations and⁄ or purity of the enzymes and substrates

used The studies cited above used either unspecified

amounts of unpurified enzyme or enzyme

concentra-tions four times [40] or over 11 times [36] that used

here In the previous studies, reticulocyte lysates

con-taining 35S-labelled wild-type caspase-2 were used as

substrates These lysates would contain endogenous

reticulocyte proteins that may potentially influence

the processing of caspase-2 To avoid any such

indi-rect effects, we used purified, catalytically inactive

caspase-2 as a substrate

The IAP family of apoptosis inhibitors exert their

pro-survival effect, at least in part, through

suppres-sion of caspases-3, -7 and -9 [33,41,42] The IAPs

XIAP, MIHB and MIHC could not inhibit other

casp-ases including -1, -6, -8 and -10 [30], but their ability

to directly inhibit caspase-2 has not been previously

published Caspase-2-dependent yeast death was

unaf-fected by coexpression of XIAP, MIHB and MIHC

although, as we previously reported [31], p35 and p49

could inhibit caspase-2 in this system Furthermore,

XIAP, the most potent caspase inhibitor of the IAP

family, did not impede the ability of recombinant

caspase-2 to cleave a synthetic substrate It was

previ-ously observed that IAPs partially protected tissue

cul-ture cells from apoptosis induced by caspase-2

overexpression [43] In the light of our findings, this

inefficient protection was probably due to

IAP-medi-ated inhibition of caspase-7, which was likely activIAP-medi-ated

by the overexpressed caspase-2

In summary, this study illustrated that, at least in

the absence of an activating adaptor, generation of

active human caspase-2 requires separation of its large

and small subunits In the context of autoactivation,

removal of the prodomain also enhances proteolytic

activity Caspase-2 can act as an apical caspase,

pro-moting the activation of caspase-7 Unlike caspases-3,

-7 and -9, caspase-2 was resistant to inhibition by

members of the IAP family

caspase-7D1)53, p35 and pGALL-(HIS3)-p49 have been reported [27,31,35] Other plasmids were constructed as follows: Pro-caspase-2 PCR product, gener-ated with primers 1 and 2, was cut with BglII⁄ XbaI and ligated into BamHI⁄ XbaI cut vectors to produce pGALL-(HIS3)-caspase-2 and pGALL-(HIS3)-FLAG-caspase-2 To make pGALL-(LEU2)-rev-caspase-2, the carboxyl terminal fragment was amplified with primers 3 and 4, digested with BglII⁄ XbaI and ligated into pGALL-(LEU2) to give pGALL-(LEU2)-rev-caspase-2-C The amino-terminal frag-ment was generated with primers 5 and 6, cut with XhoI⁄ XbaI, and ligated into pGALL-(LEU2)-rev-caspase-2-C

to generate the final construct

pGALL-(LEU2)-caspase-2C303A, pGALL-(URA)-caspase-2C303A, pGALL-(HIS3)-caspase-2C303A and pGALL-(HIS3)-FLAG-caspase-2C303A

were produced by replacing a SpeI⁄ BamHI cut fragment with a PCR product generated with primers 1 and 7 pGALL-(HIS3)-FLAG-caspase-2D1)149was cloned by ligat-ing a NdeI-digested and blunt-ended then BamHI cut frag-ment from pET23a-caspase-2D1)149 into SpeI-digested and blunt-ended then BamHI cut pGALL-(HIS3)-FLAG-ca-spase-2 A PCR product generated with primers 1 and 8 was cut with BglII⁄ XbaI and ligated into BamHI ⁄ XbaI cut vector to produce pGALL-(LEU2)-caspase-2D153)435 pGALL-(HIS3)-FLAG-caspase-2D152Aand pGALL-(HIS3)-FLAG-caspase-2D316A were made by replacing a SalI⁄ BamHI cut fragment with PCR products generated with primer pairs 9, 10 and 11, 12, respectively pGALL-(HIS3)-FLAG-caspase-2D330A was produced by replacing a

Bam-HI⁄ XbaI cut fragment with a PCR product generated with primers 13 and 14 To make pGALL-(HIS3)-FLAG-caspase-2C303A; D152,316A, a SalI⁄ BamHI fragment in pGALL-(HIS3)-FLAG-caspase-2C303A was replaced with a PCR product generated using primers 9 and 12 It was sub-sequently used to produce pGALL-(HIS3)-FLAG-cas-pase-2C303A; D152, 316, 330Aby replacing a SalI⁄ BamHI fragment into pGALL-(HIS3)-FLAG-caspase-2D330A SpeI⁄ BamHI fragments isolated from

pGALL-(HIS3)-FLAG-Caspase-2D1)149or pGALL-(HIS3)-FLAG-Caspase-2D152Awere used

to replace part of the coding region in pGALL-(HIS3)-Caspase-2, pGALL-(URA)-Caspase-2 and pGALL-(LEU2)-Caspase-2 to make pGALL-(HIS3)-pGALL-(LEU2)-Caspase-2D1)149 and

pGALL-(HIS3)-Caspase-2D152A and pGALL-(LEU2)-Caspase-2D152A,