Molecular Cell, Vol.

8, 613621, September, 2001, Copyright

2001 by Cell Press

A Serine Protease, HtrA2,

Is Released from the Mitochondria
and Interacts with XIAP, Inducing Cell Death
Yasuyuki Suzuki,1,4 Yuzuru Imai,1,4
Hiroshi Nakayama,2,4 Kazuko Takahashi,1
Koji Takio,2 and Ryosuke Takahashi1,3
1
Laboratory
for
Motor
System
Neurodegeneration Brain Science Institute (BSI)
2
Division
of
Biomolecular
Characterization RIKEN (The Institute of
Physical
and Chemical Research)
Wako City, Saitama 351-0198
Japan

Summary
X chromosome-linked inhibitor of apoptosis (XIAP) is an
endogenous inhibitor of caspase-3, -7, and -9.
Smac/DIABLO, an inhibitor of XIAP, is released from
mitochondria upon receiving apoptotic stimuli and
binds to the BIR2 and BIR3 domains of XIAP, thereby
inhibiting its caspase-inhibitory activity. Here we re-port
that a serine protease called HtrA2/Omi is re-leased
from mitochondria and inhibits the function of XIAP by
direct binding in a similar way to Smac. More-over,
when overexpressed
extramitochondrially, HtrA2
induces atypical cell death, which is neither accompanied by a significant increase in caspase activity nor
inhibited by caspase inhibitors, including XIAP. A catalytically inactive mutant of HtrA2, however, does not
induce cell death. In short, HtrA2 is a Smac-like inhibitor of IAP activity with a serine protease-dependent cell
death-inducing activity.

Introduction
Apoptosis is a physiological cell suicide program critical
to the development and homeostasis of all animals (Jacobson et al., 1997). Abnormal inhibition of apoptosis is a
hallmark of cancer and autoimmune disease, whereas
excessive cell death has been implicated in neurodegenerative disorders (Thompson, 1995). The caspases,
a family of intracellular cysteine proteases, are the
central executioners of apoptosis (Thornberry and
Lazebnik, 1998). Effector caspases, such as caspase-3
and -7, are activated by initiator caspases, such as
caspase-9, through proteolytic cleavage. Once activated,
the effector caspases are responsible for the proteolytic
cleavage of a diverse array of structural and regulatory
proteins, re-sulting in an apoptotic phenotype
(Thornberry and Lazeb-nik, 1998).
Inhibitor of apoptosis proteins (IAPs), originally found
in baculoviruses, have been conserved in a number of
species, ranging from insects to humans, throughout
evolution and play a role in regulating apoptosis (Deveraux and Reed, 1999; Miller, 1999). Several members of
the human IAP family proteins, including XIAP, c-IAP1,

3
4

Correspondence: ryosuke@brain.riken.go.jp
These authors contributed equally to this work.

and c-IAP2, have been shown to be potent direct inhibitors of caspase-3, -7, and -9 (Deveraux et al., 1997,
1998; Roy et al., 1997). Among the above IAPs, XIAP is
the most potent inhibitor of caspases and apoptosis
(Deveraux and Reed, 1999). The structure of XIAP is
characterized by three tandem repeats of the baculovirus IAP repeat (BIR) domain at its N terminus and a
RING finger domain near its C terminus. Deletional
analysis indicates that the second BIR domain (BIR2) of
XIAP is sufficient to inhibit caspase-3 and -7 (Takahashi
et al., 1998), whereas a XIAP fragment encompassing
the third BIR domain (BIR3) specifically inhibits caspase9 (Dev-eraux et al., 1999; Sun et al., 2000). Collectively,
these results suggest that the BIR domains of IAP are
essential to inhibiting caspase activity.
Recently, a novel protein, Smac/DIABLO, has been
identified. It was found to promote caspase activation by
eliminating the caspase-inhibitory activity of IAPs (Du et al.,
2000; Verhagen et al., 2000). Smac is synthesized as a 239amino acid precursor protein, and the 55 amino acids at its
N terminus function as a mitochondrial tar-geting signal
peptide which is later cleaved following mitochondrial
transport. In response to various apo-ptotic stimuli, mature
Smac are released into the cytosol, where they are thought
to inhibit IAPs. Subsequent in vitro experiments have
revealed that Smac interacts with the BIR2 and BIR3
domains of XIAP via its N-terminal amino acids, starting with
AVPI. Moreover, mutations at the N terminus of Smac
abrogate its ability to inhibit XIAP. The N-terminal amino
acids of several Drosophila proapoptotic proteins, Hid,
Reaper, and Grim, are similar to Smac and are known to
inhibit IAP activity through their N terminus (Abrams, 1999;
Srinivasula et al., 2001). Although Smac is considered to be
a functional homolog of these fly proteins, it has not yet
been shown whether Smac is the only mammalian protein
that binds to and inhibits IAPs (Du et al., 2000; Holcik and
Korneluk, 2001).

In this study, we purified and identified a serine protease named HtrA2 (also referred to as Omi) as an XIAP
binding protein by amino acid sequence analysis using
Edman degradation and electrospray ionization tandem
mass spectrometry of the coimmunoprecipitate of overexpressed XIAP in cultured cells (Faccio et al., 2000;
Gray et al., 2000). The precursor form of HtrA2 is a 50
kDa protein whose N-terminal peptides are cleaved off
after mitochondrial transport, generating a 36 kDa mature protein with N-terminal amino acids resembling
those of Smac. The mature form of HtrA2 but not its
precursor binds to IAPs in a similar manner to Smac.
Although mature HtrA2 proteins are usually confined to
mitochondria, they are released into the cytosol as a
result of apoptotic stimulus. Moreover, extramitochondrially targeted overexpression of HtrA2 induces an unusual form of cell death in a caspase-independent, serine protease-dependent manner. Taken together, these
data demonstrate that HtrA2 is a Smac-like inhibitor of
IAP with a serine protease activity-dependent cell deathinducing effect.

Molecular Cell
614

Results
Identification and Cloning of HtrA2
In order to identify novel XIAP binding proteins, cell
lysates were prepared from 293T cells transiently transfected with N-terminal FLAG-tagged XIAP without a
RING finger motif (termed XIAPDRING). Following this,
they were incubated with anti-FLAG-mAb-coupled agarose beads; the XIAPDRING form was used because it
gives a better expression than full-length XIAP. After
extensive washing, bound protein was eluted under
acidic conditions and separated by SDS-PAGE. Two
protein bands with molecular weights of 23 kDa and 36
kDa were observed in a Coomassie-stained gel (Figure
1A). These bands were not observed in the immunoprecipitates of a control FLAG-tagged protein.
Since the band corresponding to 23 kDa appeared to
be Smac, the 36 kDa band was digested in gel with
Achromobacter Protease I (Lys-C). The resultant peptides were extracted from the gel, subsequently resolved by RP-HPLC, and sequenced by electrospray
ionization tandem mass spectrometry. Since database
searching of the uninterpreted MS/MS spectra to iden-tify
the peptide failed, we employed Edman degradation for
amino acid sequence analysis. The amino acid sequence of fraction 1 in Figure 1B perfectly matched the
N-terminal fragment of a mature form of human HtrA2
(Faccio et al., 2000; Gray et al., 2000). Moreover, subsequent analysis by MS/MS revealed that the seven Lys-Ctreated fractions covered the entire sequence of the
mature 13B form of HtrA2 without any gaps (Figures 1B
and 1C) (Gray et al., 2000). Although the 13B isoform of
human HtrA2 is thought to be a 458-amino acid protein,
133 of its N-terminal amino acids are cleaved off after
transport; thus, the resulting mature form of the 13B
HtrA2 protein is 36 kDa. Therefore, the band we
analyzed is identical in size to that of mature HtrA2.

Figure 1. Detection of XIAP Binding Protein and Identification of

HtrA2
(A) Overexpressed FLAG-XIAP lacking the RING finger motif
(XIAPDRING) in 293T cells was affinity purified using anti-FLAG (M2)
agarose (Sigma). Eluted fractions were subjected to SDS-PAGE and
then stained with Coomassie blue. The arrowhead indicates purified
FLAG-XIAPDRING. Coeluted proteins are represented by the letters
A and B (see text).
(B) RP-HPLC of Lys-C-digests of the 36 kDa XIAP binding protein
(protein band A in Figure 1A). Amino acid sequence analysis by
Edman degradation of fraction 1 revealed that this fraction contains
the N-terminal fragment of mature HtrA2 (the underlined sequence in
Figure 1C). Based on this information, fractions 2, 3, 4, 5, and 6 1 7
were shown to correspond to amino acids 325347, 238324, 349
395, 157237, and 396458 of full-length HtrA2 protein, respectively.

The Mature Form of HtrA2 but Not Its Precursor

Binds to IAPs
Since endogenous HtrA2 was coimmunoprecipitated with
overexpressed XIAPDRING, we checked whether
endogenous XIAP interacts with endogenous HtrA2 in
cells. As shown in Figure 2A, endogenous HtrA2 was
also coimmunoprecipitated with endogenous XIAP,
strongly suggesting that the interaction between XIAP
and HtrA2 is physiologically significant.
Next, to examine the interaction between IAP family
proteins and HtrA2, expression plasmids for various Nterminal FLAG- or Myc-tagged IAPs, including XIAP, cIAP1, c-IAP2, and survivin, were transiently trans-fected
with a plasmid encoding for human HtrA2 protein with a
C-terminal HA tag. The 36 kDa mature form of HtrA2 but
not the 50 kDa precursor was coimmunopreci-pitated
with XIAP, c-IAP1, and c-IAP2. Mature HtrA2 did not bind
to survivin, with which Smac reportedly associates (Du et
al., 2000) (Figure 2). These results indicate that mature
HtrA2 preferentially binds to those
(C) Amino acid sequence of the full-length 13B form of human HtrA2.
The sequence of fraction 1 in Figure 1B is underlined. The
arrowhead indicates the site at which cleavage occurs, thereby
generating mature HtrA2.

A Mitochondrial Cell Death-Inducing Factor

615

IAPs with potent caspase-inhibitory activity, such as

XIAP, c-IAP1, and c-IAP2 (Deveraux et al., 1997; Roy et
al., 1997; Tamm et al., 1998).
HtrA2 Inhibits the Caspase-Inhibitory Activity of
XIAP in a Similar Manner to Smac/DIABLO
The N-terminal amino acids preceding alanine 134 are
cleaved in order to produce mature HtrA2 (Savopoulos et
al., 2000) (Figure 3A). The first four amino acids of
mature HtrA2 protein are AVPS. This sequence is very
similar to that of the N-terminal sequence of Smac, which
is AVPI. Smac requires an intact N-terminal se-quence to
interact with the BIR2 and BIR3 domains of XIAP (Chai
et al., 2000). Moreover, other proteins with Smac-like Nterminal sequences, including the small sub-unit of
caspase-9, have been shown to bind to XIAP (Srinivasula et al., 2001). These results suggest that mature
HtrA2 binds to and inhibits XIAP in a similar manner to
Smac (Chai et al., 2000). To examine this possibility, we
assayed both wild-type and mutant recombinant mature
HtrA2, in which the first residue was changed from
alanine to methionine. These were designated
recHtrA2(AVPS) and recHtrA2(MVPS), respectively, and
were studied with regard to their ability to bind to a series
of XIAP fragments made into fusion proteins with
glutathione
S-transferase (GST). As expected,
recHtrA2(AVPS) bound to the BIR2 and BIR3 domains of
XIAP, whereas recHtrA2(MVPS) did not bind to XIAP at
all (Figure 3B). The nature of binding between HtrA2 and
XIAP was similar to that between Smac and XIAP. To
examine whether HtrA2, which is similar to Smac,
diminishes the caspase-inhibitory effect of XIAP, we
assayed recHtrA2(AVPS) and recHtrA2(MVPS) with
respect to their ability to promote caspase-3 activity and
caspase-3 activation. We employed a cell-free sys-tem in
which caspase-3 activation was reproduced in vitro by
the addition of cytochrome c to a HEK293 cyto-solic
extract (Deveraux et al., 1997). The BIR3 1 RING
domain has been found to block the in vitro processing of
procaspase-3 into large and small subunits by direct
inhibition of active caspase-9 (Figure 3C) (Deveraux et
al., 1999). The BIR2 domain, on the other hand, blocks
the generation of p19 and p17 from the large p20 subunit
by direct inhibition of caspase-3 (compare Figure 3C with
lanes 4, 7, and 10 of Figure 3D) (Deveraux et al., 1997).
RecHtrA2(AVPS) inhibited both BIR2-mediated and BIR3
1 RING-mediated inhibition of caspase-3 pro-cessing,
whereas recHtrA2(MVPS) did not inhibit the caspaseinhibitory activity of these XIAP fragments (Fig-ure 3D).
These results indicate that HtrA2 inhibits the caspaseinhibitory activity of XIAP in a similar manner to Smac.
XIAP Does Not Inhibit the Serine Protease
Activity of HtrA2
To examine whether XIAP inhibits the serine protease
activity of HtrA2, we performed an in vitro protease
assay using b-casein as a substrate. Recombinant mature HtrA2 completely cleaved b-casein at a concentration of 50 nM at 378C after 2 hr (Figure 4). XIAP, even at
a concentration of 2 mM, did not inhibit in vitro b-casein
cleavage by HtrA2. This result indicates that XIAP binds
to, but does not have an inhibitory effect on, the mature
HtrA2 serine protease.

Figure 2. Interaction between HtrA2 and IAP Family Proteins

(A) Endogenous XIAP associates with endogenous HtrA2. Whole-cell
lysates (WCL) from HEK293 cells (1 mg protein) were incubated with
1 mg of control mAb and anti-XIAP mAb conjugated with 5 ml of
protein G-Sepharose at 48C for 4 hr. Immunoprecipitates (IP) were
separated by SDS-PAGE, transferred to PVDF membranes, and analyzed by immunoblotting (IB) using anti-HtrA2 and anti-XIAP rabbit
polyclonal Ab. The bands for the heavy chain and the light chain of
IgG are indicated as IgG HC and IgG LC, respectively.
(B) Mature HtrA2 but not its precursor associates with IAP family
proteins. N-terminal FLAG-tagged XIAP, c-IAP1, c-IAP2, Myc-tagged
XIAP, and survivin cDNA were transfected with C-terminal HA-tagged
full-length HtrA2 cDNA into HEK293 cells. Whole-cell lysates (WCL)
and immunoprecipitates with anti-FLAG mAb (IP: FLAG) or anti-Myc
mAb (IP: Myc) were analyzed by immunoblotting (IB) using antiFLAG, anti-Myc, or anti-HA mAb.

HtrA2 Is Localized within the Mitochondria

and Released into the Cytosol as a Result
of Apoptotic Stimuli
PSORT (see http://psort.ims.u-tokyo.ac.jp) analysis revealed that HtrA2 might be a mitochondrial protein, although previous reports have documented its presence
in either the endoplasmic reticulum or the nucleus (Fac-

Molecular Cell
616

Figure 4. XIAP Does Not Inhibit the Serine Protease Activity of HtrA2
Coomassie blue-stained polyacrylamide gel of His 6-tagged recombinant HtrA2 after 1 hr and 2 hr of incubation. b-casein (6 mM) as a
generic substrate was incubated with 50 nM recombinant HtrA2
serine protease in the presence of the indicated proteins. Even when
incubated with 2 mM GST-XIAP, the amount of cleaved b-casein was
almost identical to that obtained from incubation without XIAP.

Figure 3. Mature HtrA2 but Not Its N-Terminal Missense Mutant

Binds to and Inhibits XIAP in a Similar Manner to Smac
(A) Schematic representation of full-length HtrA2 (HtrA2-FL) and mature
HtrA2. The transmembrane (TM) segment, the trypsin-like catalytic
domain, and the PDZ domain are located among amino acids 105121,
182330, and 390445 of HtrA2-FL, respectively. The serine residue of
the active site is labeled S306. RecHtrA2(AVPS) and recHtrA2(MVPS)
designate the bacterially produced recombinant proteins of wild-type and
N-terminal mutant HtrA2, respectively. Although the N-terminal sequence
of RecHtrA2(AVPS) was originally constructed as MAVPS, the initial
methionine is cleaved off in bacte-ria, generating wild-type mature HtrA2
protein.
(B) RecHtrA2(AVPS) but not recHtrA2(MVPS) binds to the BIR2 and

BIR3 regions of XIAP. Recombinant wild-type and mutant HtrA2

protein was pulled down with a series of GST-fused XIAP deletion
mutants, after which immunoblot analysis was performed using antipenta His or anti-GST Abs. The arrowhead indicates C-terminal His 6tagged recombinant mature HtrA2.
(C) Schematic representation of the processing of procaspase-3 and
the points at which XIAP fragments inhibit this processing. The BIR3
1 RING (BIR31R) fragment inhibits the processing of procaspase-3
(p32) by caspase-9 to generate p20 and p12 frag-ments, while the
BIR2 fragment inhibits the processing of p20 by caspase-3 to yield
p19 or p17 fragments.
(D) RecHtrA2(AVPS) but not recHtrA2(MVPS) inhibits both caspase-3

cio et al., 2000; Gray et al., 2000). In addition, the Smaclike activity of HtrA2 prompted us to reexamine the
subcellular localization of this protease. Subcellular
fractionation of HeLa cells revealed the presence of
HtrA2 in the heavy membrane fraction where cytochrome
c was cofractionated (Figure 5A). Immunostaining of
HeLa cells with polyclonal Ab generated against recombinant HtrA2 protein showed remarkable colocalization of
HtrA2 with cytochrome c in the mitochondria. How-ever,
when apoptosis was induced by UV irradiation, both
HtrA2 and cytochrome c staining changed from a
punctate mitochondrial pattern to a diffuse cytosolic
pattern. Both proteins started to exhibit a diffuse pattern
of staining 4 hr after UV irradiation (Figure 5B). To confirm these immunostaining results through biochemical
analysis, cytosolic fractions were isolated from the cells
with or without UV irradiation. As shown in Figure 5C, in
nonirradiated cells HtrA2 and cytochrome c were not
detected in the cytosolic fractions. However, 4 hr after
UV irradiation, HtrA2 and cytochrome c were concurrently observed in the cytosol. These results indicate that
HtrA2, like Smac, is a mitochondrial protein released into
the cytosol as a result of apoptotic stimuli.
HtrA2 Increases the Sensitivity of Cells to UV
To study the role of HtrA2 in cells, we transiently transfected HeLa cells with cDNA encoding either full-length
HtrA2 or full-length Smac with a C-terminal Myc tag. We
then induced apoptosis by UV irradiation. As shown in
Figure 6, both HtrA2 and Smac transfectants showed
significantly higher caspase-3-like DEVDase activity than

inhibition by GST-BIR2 (BIR2) and caspase-9 inhibition by GST-BIR3

1 RING (BIR31R). Cytosolic extracts prepared from HEK293 cells
were incubated with cytochrome c/dATP to induce caspase-3
activation in vitro in the presence or absence of the indicated
proteins. Following this, the extracts were subjected to SDS-PAGE
followed by immu-noblot with anti-caspase-3 polyclonal Ab, as
described elsewhere (Deveraux et al., 1999; Takahashi et al., 1998).
The open arrowhead indicates contamination with a small amount of
protein from E. coli which crossreacts with anti-caspase-3 Ab.

A Mitochondrial Cell Death-Inducing Factor

617

Figure 5. HtrA2 Is Released from Mitochondria to the Cytosol in Response to UV Irradiation

(A) Subcellular fractionation of HeLa cells.
HtrA2 and cytochrome c were observed to
colocalize within the fraction containing mitochondria. Protein disulfide isomerase (PDI)
and XIAP represented the endoplasmic reticulum resident protein and the cytosolic protein, respectively (Duckett et al., 1996; Ferrari
and Soling, 1999). Although PDI was detected
in both the heavy membrane fraction and cytosolic extract, presumably due to leakage, its
presence in the light membrane fraction
containing the endoplasmic reticulum was
evident. HM, heavy membrane fraction; LM,
light membrane fraction; CE, cytosolic extract.
(B) Immunostaining of HtrA2 and cytochrome
c. Before and 4 hr after UV (100 mJ/cm 2) irradiation, the cells were fixed and immunostained with anti-HtrA2 and anti-cytochrome c
mAb.
(C) Cytosolic translocation of HtrA2 and cytochrome c after UV irradiation. Cytosolic extracts were prepared from UV (100 mJ/cm 2)
irradiated cells at the time indicated and immunoblotted with anti-HtrA2 polyclonal Ab,
anti-cytochrome c mAb, or anti-actin mAb.

mock transfectants 4 hr after 100 mJ/cm2 UV irradiation

(Du et al., 2000) (Figures 5B and 5C). These results also
support the notion that HtrA2 has proapoptotic activity,
as does Smac, and that both HtrA2 and Smac achieve
this via a similar mechanism of action.
Extramitochondrially Expressed HtrA2 Induces
Atypical Cell Death
Recently, it has been reported that a variant of Smac that
lacks the N-terminal IAP binding domain and displays an
extramitochondrial cortical distribution retains the proapoptotic activity of Smac with the same potency (Roberts et al.,
2001). This observation prompted us to examine the IAPindependent function of HtrA2. We made the same
constructs as were used for producing the bacte-rial proteins
recHtrA2(AVPS) and recHtrA2(MVPS), ex-cept for the
epitope tags (Figure 3A). Unexpectedly, the binding assay
revealed that recHtrA2(MVPS) or recHtrA2 (AVPS) would
not bind to or only scarcely bind to XIAP, respectively,
presumably due to posttranslational modi-fication. We
therefore used the HtrA2(AVPS) construct

for further experimentation. Mature HtrA2 protein derived from this construct was immunolocalized to the
cytosol (data not shown).
Surprisingly, overexpression of HtrA2(AVPS) induced
dramatic apoptosis-like morphological changes, including cell rounding and shrinkage (Figures 7A and 7B).
However, this remarkable change in cell shape was not
accompanied by membrane blebbing, apoptotic body
formation, or nuclear morphological changes, as evaluated by DAPI staining (data not shown). Moreover, a
trypan blue exclusion assay revealed that very few
HtrA2-overexpressing cells were dead; the ratio of trypan blue-positive cells among mock and HtrA2 transfectants was less than 5%, even 60 hr after transfection.
While the HtrA2-overexpressing cells exclude the dye,
they did not seem to recover or proliferate and finally
detached from the plate up to 100 hr after transfection.
We therefore deemed these atypical morphological
changes atypical cell death. There was only a background level of caspase activity in the lysate of HtrA2
(AVPS)-overexpressing cells, whereas caspase activity

Molecular Cell
618

Figure 6. HtrA2 Enhances DEVDase Activity Induced by UV Irradiation

HeLa cells were cotransfected with 0.8 mg of pcDNA3 (Mock),
pcDNA3-Smac-Myc (Smac), or pcDNA3-HtrA2-Myc (HtrA2) together
with 0.2 mg of pEGFP plasmid, incubated for 24 hr, then stimulated
by UV (100mJ/cm2) irradiation. After 2 or 4 hr, cell lysates were
prepared, and DEVDase activity within 30 mg of lysate from each
preparation was measured. The error bars represent the SE calculated from triplicate samples.

in the lysates from Bax-overexpressing cells showed

more than 3-fold increase compared with that in control
cell lysates (Figure 7C). Moreover, HtrA2-induced cell
death was not inhibited by caspase inhibitors, including
XIAP and z-VAD-fmk, whereas Fas-induced apoptosis
was markedly suppressed by both caspase inhibitors
(Figure 7D). Notably, the serine protease-inactive catalytic mutant of HtrA2(AVPS) almost lost its cell deathinducing activity (Figure 7B). These data indicate that
extramitochodrially overexpressed mature HtrA2 induces caspase-independent cell death and that this peculiar activity requires the serine protease activity of
HtrA2.
Discussion
Mitochondria play a key role in physiological cell death
(Green and Reed, 1998). Various proapoptotic stimuli
induce an increased permeability of the outer mitochondrial membrane, which allows the release of multiple mitochondrial proapoptotic proteins in the intermembrane space. These proteins include cytochrome c,
caspase-2, -3, -7, -9, and apoptosis-inducing factor (AIF)
(Kroemer and Reed, 2000). Smac/DIABLO is a re-cently
identified mitochondrial proapoptotic protein which is
released from mitochondria and directly binds to and
inhibits the caspase-inhibitory activity of XIAP (Du et al.,
2000; Verhagen et al., 2000). Here we identified a serine
protease, HtrA2, as a XIAP binding protein and
demonstrated that HtrA2 inhibits XIAP in a similar manner to Smac.
HtrA2 belongs to the HtrA family of serine proteases
that is well conserved from bacteria to humans (Faccio et
al., 2000; Gray et al., 2000; Hu et al., 1998; Savopoulos
et al., 2000). The bacterial HtrA gene product is one of
the most well-characterized proteins of the HtrA family of
serine proteases; Htr stands for high temperature

requirement. HtrA endoprotease is localized within the

periplasmic space of bacteria, and its presence is necessary for bacterial thermotolerance (Lipinska et al.,
1988, 1990; Seol et al., 1991). Moreover, it has recently
been shown that bacterial HtrA has a dual role, acting as
a chaperone at normal temperatures and as an active
protease at high temperatures (Spiess et al., 1999). Human HtrA2 serine protease has extensive homology with
bacterial HtrA and L56, another human HtrA, at the C
terminus of its polypeptide (Hu et al., 1998). HtrA2 protein has a putative transmembrane domain, a typsin-like
catalytic domain, and a PDZ domain, located among
amino acids 105121, 182330, and 390445, respectively (Figure 3A) (Faccio et al., 2000). Although it has
been reported that HtrA2 is localized within either the
endoplasmic reticulum or the nucleus (Faccio et al.,
2000; Gray et al., 2000), we have shown here that HtrA2
is a mitochondrial protein. Because mature HtrA2 protein is released from the mitochondria in response to
apoptotic stimuli, under normal conditions mature HtrA2
protein is found within the intermembrane space, as are
cytochrome c and Smac. The fact that the first four Nterminal amino acids of mature HtrA2 protein, AVPS, are
very similar to those of the mature Smac protein, AVPI,
provides evidence that the two proteins function in a
similar way to inhibit IAPs (Chai et al., 2000). Northern
blot analysis has revealed that Smac mRNA is only
scarcely expressed in several tissues, including brain,
placenta, lung, and leukocytes, whereas HtrA2 is expressed rather ubiquitously (Gray et al., 2000). These
two proteins might have complementary roles in various
tissues.
The unique feature that differentiates HtrA2 from Smac
is its ability to induce atypical cell death when its mature
form is overexpressed in the cytosol. Although cells
overexpressing HtrA2 in the cytosol underwent marked
rounding and shrinkage, they did not show membrane
blebbing or apoptotic body formation, and the integrity of
their membrane was maintained. Al-though this HtrA2induced cell death could not be inhib-ited by caspase
inhibitors, a serine protease-inactive mutant of HtrA2 lost
the ability to induce cell death (Figures 7B and 7D).
Since we could not express the mature HtrA2 protein
with an intact N-terminal amino acid sequence in cells, it
is not yet certain whether the cell death-inductive activity
of HtrA2 is inhibited by asso-ciation with XIAP. However,
recombinant XIAP did not inhibit the serine protease
activity of mature HtrA2, sug-gesting that XIAP is likely
ineffective in suppressing HtrA2-induced cell death,
regardless of its ability to bind with HtrA2.
HtrA2 seems to promote or induce cell death through
two different mechanisms. One is the direct binding to
and inhibiton of IAPs and is accompanied by a significant increase in caspase activity (Figure 6). The other is
through the IAP inhibition-independent, caspase-independent and serine protease activity-dependent mechanism shown in Figure 7. Although the latter form of
HtrA2-induced cell death seems to be a nonapoptotic cell
death, it has been observed only when overex-pressed.
The precise characterization of mature HtrA2-induced
cell death requires further examination in more
physiological settings.

A Mitochondrial Cell Death-Inducing Factor

619

Figure 7. Extramitochondrially Overexpressed HtrA2 Induces Caspase-Independent Atypical Cell Death

(A) The expression of mature HtrA2 induces atypical cell death. HEK293 cells were cotransfected with 1 mg of pcDNA3 (Mock) or pcDNA3mature HtrA2-Myc (mature HtrA2) together with 0.2 mg of pEGFP plasmid. After 30 hr, GFP-positive cells were visualized by confocal laser
microscopy.
(B) Protease activity of mature HtrA2 is required for atypical cell death. HEK293 cells were cotransfected with pcDNA3 (Mock), pcDNA3 encoding
mature HtrA2-Myc (mature HtrA2), active site Ser to Ala mutant of mature HtrA2-Myc [mature HtrA2(S/A)], full-length HtrA2-Myc (FL-HtrA2), or
full-length Smac-Myc (FL-Smac) along with pEGFP plasmid as in (A). After 24 hr, GFP-positive cells with round or shrunken shapes were counted.
The error bars represent the SE calculated from triplicate samples.
(C) DEVDase activity is not significantly induced by mature HtrA2. HEK293 cells were cotransfected with the indicated constructs along with
pEGFP plasmid as in (A). After 24 hr, cell lysates were prepared and DEVDase activity in 100 mg of each lysate sample was measured. The error
bars represent the SE calculated from triplicate samples.
(D) Caspase inhibitors cannot block the effect of mature HtrA2. HEK293 cells were cotransfected with the indicated constructs along with pEGFP
plasmid as in (A). During transfection and following incubation, the broad-spectrum caspase inhibitor zVAD-fmk (100 mM) was added as
indicated. The error bars represent the SE calculated from triplicate samples.

Recently, nonapoptotic caspase-independent cell

death has attracted enormous attention because of its
implica-tion in cancer and neurodegenerative diseases
(Kita-naka and Kuchino, 1999; Sperandio et al., 2000). In
this regard, a very interesting question is whether HtrA2
is involved in caspase-independent cell death in either
pathophysiological or physiological conditions. Understanding the biochemical pathway by which HtrA2 induces peculiar cell death may reveal the pathogenesis

of such diseases or lead to the development of new

therapeutics for their treatment.
Experimental Procedures
Plasmids and Antibodies
The plasmids encoding FLAG epitope-tagged wild-type XIAP (pcDNA3FLAG-XIAP) and the RING finger-deleted form of XIAP (pcDNA3-FLAGXIAPDRING) have been described elsewhere (Suzuki et al., 2001b). To
generate pcDNA3-FLAG-c-IAP1 and -c-IAP2, the EcoRI-

Molecular Cell
620

XhoI fragment of the full-length c-IAP1 of pcDNA3-Myc-c-IAP1 and

the reverse transcription-PCR (RT-PCR) product of human c-IAP2
were cloned into pcDNA3-FLAG-N, respectively. The plasmids encoding the Myc-tagged IAPs, including pcDNA3-Myc-survivin,
pcDNA3-Bax, and pCMV-Fas, were generously provided by Dr. John
Reed (Burnham Institute). The cDNA encoding human HtrA2 was
generated by RT-PCR using the following primers: forward, ccgcaat
tggccATGGCTGCGCCGAGGGCG; reverse, ctctcgagTTCTGTGACCTCAGGGGTC. The MfeI-XhoI fragment of HtrA2 was subcloned
into the EcoRI-XhoI sites of pcDNA3-Myc-C or -HA-C plasmids in
order to generate C-terminal epitope-tagged constructs (pcDNA3HtrA2-Myc or -HtrA2-HA, respectively). The various mutants of the
HtrA2 constructs were made in the same way. The cDNA encoding
human Smac/DIABLO was generated by RT-PCR, and the BamHIXhoI fragment of Smac was subcloned into the BamHI-XhoI site of
pcDNA3-Myc-C. Proper construction of all the plasmids was confirmed by DNA sequencing. Antisera specific for HtrA2 and XIAP
were prepared in rabbits using the E. coli expressed C-terminal His6tagged mature form of human HtrA2 protein and N-terminal GSTtagged XIAP-BIR1 protein (Suzuki et al., 2001a) as antigen, respectively.
Purification of XIAP Binding Proteins
FLAG-XIAPDRING-transfected 293T cells (1.25 3 108) were homogenized in lysis buffer (10 mM Tris [pH 8.0], containing 120 mM NaCl, 5 mM
EDTA) with Complete Protease Inhibitors (Roche Diagnostics). The
soluble fraction of the suspension was immunoprecipitated with anti-FLAG
M2 agarose (Sigma) and then washed five times in lysis buffer without
protease inhibitors. The fractions eluted with 0.1 M glycine-HCl (pH 2.8)
were resolved by SDS-PAGE. The bands de-tected by CBB staining were
excised for in-gel digestion.

HPLC-MS/MS and Amino Acid Sequencing

In-gel digestion was performed as described elsewhere (Kawasaki et
al., 1990). Briefly, the CBB-stained band of 36 kDa was excised and
treated with 0.1 mg of Achromobacter Protease I (Lys-C) at 378C for
12 hr in 0.05 M Tris-HCl (pH 9) containing 0.1% SDS (Masaki et al.,
1981). The digest extracted from the gel was subjected to online
HPLC-MS/MS and Edman degradation. An aliquot of the peptide
mixture was separated on columns of DEAE-5PW (1 3 10 mm;
Tosoh) and Mightysil C18 (1 3 50 mm; Kanto Chemical) con-nected
in series using a model 1100 liquid chromatograph (Agilent
Technologies) with a diode-array detector. The peptides were eluted
at a flow rate of 30 ml/min using a 30 min linear gradient of 2%60%
solvent B, where solvents A and B were 0.09% (v/v) aqueous trifluoroacetic acid and 0.075% (v/v) trifluoroacetic acid in 80% (v/v) acetonitrile, respectively. The eluate was analyzed by a Finnigan LCQ ion
trap mass spectrometer with an electrospray ionization probe. The
residual peptide mixture was separated as described above and
fractionated. The selected peptides were subjected to Edman degradation using a model 477A automated protein sequencer connected online to a model 120A PTH analyzer (Applied Biosystems).
Transfections, Coimmunoprecipitation,
and Immunoblot Analysis
HEK293 and HeLa cells were transiently transfected using LipofectAMINE 2000 Reagent (GIBCO-BRL) according to the manufacturers
instructions. For coimmunoprecipitation, a total of 2 3 106 HEK293
cells were plated on a 10 cm dish, washed the following day, and
then transfected using a total of 6 mg of plasmid DNA containing 5
mg of FLAG- or Myc-tagged IAP and 1 mg of HA-tagged HtrA2expressing plasmid. After 6 hr of incubation, the medium was replaced with fresh complete medium. At 24 hr following transfection,
the medium was replaced with fresh complete medium containing 5
mM of the proteasome inhibitor MG132 (Peptide Institute), cells were
cultured for another 16 hr, then washed with PBS and recov-ered by
scraping. Precipitated cells were lysed in lysis buffer (20 mM HEPES
[pH 7.4], 120 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10% glycerol)
containing 20 mM of MG132 and Complete Protease Inhibitors
(Roche Diagnostics). Cellular debris was removed by cen-trifugation
at 15,000 3 g for 30 min, and the supernatant was incu-bated with 5
ml of anti-FLAG M2 affinity gel (Sigma) or agarose-conjugated 9E10
anti-Myc mAb (Santa Cruz Biotechnology) at 48C

for 4 hr. Immunoprecipitate was washed five times with 500 ml of

lysis buffer without protease inhibitor at 48C. Following this, the
samples were separated by SDS-PAGE using polyacrylamide gel,
transferred to PVDF membranes (Immobilon, Millipore), and analyzed by immunoblot analysis with the appropriate Abs. Antibody
detection was performed using an enhanced chemiluminescence
detection kit (Amersham Pharmacia).
Expression and Purification of Recombinant Protein
The N-terminal GST-tagged recombinant XIAP proteins were prepared as described elsewhere (Suzuki et al., 2001a). The C-terminal
His6-tagged mature HtrA2 and its mutant were subcloned into
pET28a (Novagen), and the recombinant proteins were prepared as
described elsewhere (Savopoulos et al., 2000). Protein concentrations were determined using a Coomassie Plus Protein Assay Reagent Kit or a Micro BCA Protein Assay Reagent Kit (Pierce).
Subcellular Fractionation
HeLa cells were harvested by centrifugation at 600 3 g for 10 min at
48C. The cell pellets were washed once with ice-cold PBS and
resuspended with 3 volumes of buffer A (20 mM HEPES-KOH [pH
7.5], 10 mM KCl, 1.5 mM MgCl 2, 1 mM EDTA, 1 mM EGTA, and 1
mM dithiothreitol) containing 250 mM sucrose and protease inhibitor. The cells were passed through a 29G needle ten times, and the
homogenates were centrifuged twice at 750 3 g for 10 min at 48C.
The supernatant was centrifuged at 10,000 3 g for 15 min at 48C,
and the resulting pellets were resuspended in lysis buffer as
described above. This was followed by centrifugation at 15,000 3 g
for 30 min at 48C prior to recovery of the resulting supernatant (the
heavy membrane fraction). Supernatant recovered from the 10,000 3
g spin was further centrifuged at 100,000 3 g for 1 hr at 48C, and the
resulting pellets (the light membrane fraction) and the resulting
supernatant (the cytosolic extract) were recovered. For immnoblot
analysis of subcellular fractions, anti-cytochrome c mAb (1:500;
PharMingen) and anti-PDI (2 mg/ml; Stressgen) were used.
Immunostaining
For immunocytochemical analysis, HeLa cells were plated in 8-well
chamber slides and cultivated for 2448 hr. The cells were washed
with PBS, fixed with 4% paraformaldehyde, permeabilized with 0.2%
Triton X-100, and then incubated with anti-human HtrA2 antisera
(1:200 dilution) and anti-cytochrome c mAb (PharMingen; 1:500).
The primary Abs were detected by secondary Abs conjugated to
Alexa 488 or 546 (Molecular Probes). Stained cells were mounted in
SlowFade (Molecular Probes), then analyzed with a laser scanning
confocal microscope system (Fluoview, Olympus).
DEVDase Activity and Cell Death Assay
A total of 45 3 105 HeLa or HEK293 cells were plated on a 6-well
plate, washed the following day, and then transfected with a total of
13 mg of plasmid DNA containing 0.2 mg of pEGFP N3 (Clontech).
After 24 hr of incubation, the HeLa cells were stimulated with UV as
indicated, then the cell lysates were recovered. Caspase activity was
assayed as described elsewhere (Suzuki et al., 2001b). To per-form
the cell death assay, transfected HEK293 cells were cultivated for
2436 hr, after which GFP-positive cells exhibiting round or shrunken
morphology were counted.
Acknowledgments
We thank Dr. Douglas Green for invaluable advice and encouragement, Drs. John Reed and Masayuki Miura for critical reading of the
manuscript, and Yui Nakabayashi for technical assistance. This work
was funded by research grants from RIKEN BSI; grants from Grantsin-Aid for Scientific Research on Priority Areas from the Ministry of
Education, Science, Sports and Culture of Japan; grants from the
Ministry of Health and Welfare of Japan; and a Grant-in-Aid for
Encouragement of Young Scientists from the Japan Society for the
Promotion of Science.
Received August 14, 2001; revised August 29, 2001.

A Mitochondrial Cell Death-Inducing Factor

621

References

M4971. I. Purification and some enzymatic properties. Biochim.

Biophys. Acta 660, 4450.

Abrams, J.M. (1999). An emerging blueprint for apoptosis in Drosophila. Trends. Cell Biol. 9, 435440.

Miller, L.K. (1999). An exegesis of IAPs: salvation and surprises from

BIR motifs. Trends Cell Biol. 9, 323328.

Chai, J., Du, C., Wu, J.W., Kyin, S., Wang, X., and Shi, Y. (2000).
Structural and biochemical basis of apoptotic activation by Smac/
DIABLO. Nature 406, 855862.

Roberts, D.L., Merrison, W., MacFarlane, M., and Cohen, G.M.

(2001). The inhibitor of apoptosis protein-binding domain of Smac is
not essential for its proapoptotic activity. J. Cell Biol. 153, 221228.

Deveraux, Q.L., and Reed, J.C. (1999). IAP family proteins

suppres-sors of apoptosis. Genes Dev. 13, 239252.

Roy, N., Deveraux, Q.L., Takahashi, R., Salvesen, G.S., and Reed,
J.C. (1997). The c-IAP-1 and c-IAP-2 proteins are direct inhibitors of
specific caspases. EMBO J. 16, 69146925.

Deveraux, Q.L., Takahashi, R., Salvesen, G.S., and Reed, J.C.

(1997). X-linked IAP is a direct inhibitor of cell-death proteases.
Nature 388, 300304.
Deveraux, Q.L., Roy, N., Stennicke, H.R., Van Arsdale, T., Zhou, Q.,
Srinivasula, S.M., Alnemri, E.S., Salvesen, G.S., and Reed, J.C.
(1998). IAPs block apoptotic events induced by caspase-8 and cytochrome c by direct inhibition of distinct caspases. EMBO J. 17, 2215
2223.
Deveraux, Q.L., Leo, E., Stennicke, H.R., Welsh, K., Salvesen, G.S.,
and Reed, J.C. (1999). Cleavage of human inhibitor of apoptosis
protein XIAP results in fragments with distinct specificities for caspases. EMBO J. 18, 52425251.
Du, C., Fang, M., Li, Y., Li, L., and Wang, X. (2000). Smac, a
mitochon-drial protein that promotes cytochrome c-dependent
caspase acti-vation by eliminating IAP inhibition. Cell 102, 3342.
Duckett, C.S., Nava, V.E., Gedrich, R.W., Clem, R.J., Van Dongen,
J.L., Gilfillan, M.C., Shiels, H., Hardwick, J.M., and Thompson, C.B.
(1996). A conserved family of cellular genes related to the baculovirus iap gene and encoding apoptosis inhibitors. EMBO J. 15, 2685
2694.
Faccio, L., Fusco, C., Chen, A., Martinotti, S., Bonventre, J.V., and
Zervos, A.S. (2000). Characterization of a novel human serine protease that has extensive homology to bacterial heat shock endoprotease HtrA and is regulated by kidney ischemia. J. Biol. Chem. 275,
25812588.
Ferrari, D.M., and Soling, H.D. (1999). The protein disulphide-isomerase
family: unravelling a string of folds. Biochem. J. 339, 110.

Gray, C.W., Ward, R.V., Karran, E., Turconi, S., Rowles, A.,
Viglienghi, D., Southan, C., Barton, A., Fantom, K.G., West, A., et al.
(2000). Characterization of human HtrA2, a novel serine protease
involved in the mammalian cellular stress response. Eur. J. Biochem.
267, 56995710.
Green, D.R., and Reed, J.C. (1998). Mitochondria and apoptosis.
Science 281, 13091312.
Holcik, M., and Korneluk, R.G. (2001). OPINIONXIAP, the guardian
angel. Nat. Rev. Mol. Cell. Biol. 2, 550556.
Hu, S.I., Carozza, M., Klein, M., Nantermet, P., Luk, D., and Crowl,
R.M. (1998). Human HtrA, an evolutionarily conserved serine protease identified as a differentially expressed gene product in osteoarthritic cartilage. J. Biol. Chem. 273, 3440634412.
Jacobson, M.D., Weil, M., and Raff, M.C. (1997). Programmed cell
death in animal development. Cell 88, 347354.
Kawasaki, H., Emori, Y., and Suzuki, K. (1990). Production and
sepa-ration of peptides from proteins stained with Coomassie brilliant
blue R-250 after separation by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Anal. Biochem. 191, 332336.
Kitanaka, C., and Kuchino, Y. (1999). Caspase-independent programmed cell death with necrotic morphology. Cell Death Differ. 6,
508515.
Kroemer, G., and Reed, J.C. (2000). Mitochondrial control of cell
death. Nat. Med. 6, 513519.
Lipinska, B., Sharma, S., and Georgopoulos, C. (1988). Sequence
analysis and regulation of the htrA gene of Escherichia coli: a sigma
32-independent mechanism of heat-inducible transcription. Nucleic
Acids Res. 16, 1005310067.
Lipinska, B., Zylicz, M., and Georgopoulos, C. (1990). The HtrA
(DegP) protein, essential for Escherichia coli survival at high temperatures, is an endopeptidase. J. Bacteriol. 172, 17911797.
Masaki, T., Tanabe, M., Nakamura, K., and Soejima, M. (1981). Studies on a new proteolytic enzyme from A chromobacter lyticus

Savopoulos, J.W., Carter, P.S., Turconi, S., Pettman, G.R., Karran,

E.H., Gray, C.W., Ward, R.V., Jenkins, O., and Creasy, C.L. (2000).
Expression, purification, and functional analysis of the human serine
protease HtrA2. Protein Expr. Purif. 19, 227234.
Seol, J.H., Woo, S.K., Jung, E.M., Yoo, S.J., Lee, C.S., Kim, K.J.,
Tanaka, K., Ichihara, A., Ha, D.B., and Chung, C.H. (1991). Protease
Do is essential for survival of Escherichia coli at high temperatures:
its identity with the htrA gene product. Biochem. Biophys. Res.
Commun. 176, 730736.
Sperandio, S., de Belle, I., and Bredesen, D.E. (2000). An alternative,
nonapoptotic form of programmed cell death. Proc. Natl. Acad. Sci.
USA 97, 1437614381.
Spiess, C., Beil, A., and Ehrmann, M. (1999). A temperature-dependent switch from chaperone to protease in a widely conserved heat
shock protein. Cell 97, 339347.
Srinivasula, S.M., Hegde, R., Saleh, A., Datta, P., Shiozaki, E., Chai,
J., Lee, R.A., Robbins, P.D., Fernandes-Alnemri, T., Shi, Y., and
Alnemri, E.S. (2001). A conserved XIAP-interaction motif in caspase9 and Smac/DIABLO regulates caspase activity and apoptosis.
Nature 410, 112116.
Sun, C., Cai, M., Meadows, R.P., Xu, N., Gunasekera, A.H., Herrmann, J., Wu, J.C., and Fesik, S.W. (2000). NMR structure and mutagenesis of the third Bir domain of the inhibitor of apoptosis protein
XIAP. J. Biol. Chem. 275, 3377733781.
Suzuki, Y., Nakabayashi, Y., Nakata, K., Reed, J.C., and Takahashi,
R. (2001a). X-linked inhibitor of apoptosis protein (XIAP) inhibits
caspase-3 and -7 in distinct modes. J. Biol. Chem. 276, 27058
27063.
Suzuki, Y., Nakabayashi, Y., and Takahashi, R. (2001b). Ubiquitinprotein ligase activity of X-linked inhibitor of apoptosis protein promotes proteasomal degradation of caspase-3 and enhances its antiapoptotic effect in Fas-induced cell death. Proc. Natl. Acad. Sci. USA
98, 86628667.
Takahashi, R., Deveraux, Q., Tamm, I., Welsh, K., Assa-Munt, N.,
Salvesen, G.S., and Reed, J.C. (1998). A single BIR domain of XIAP
sufficient for inhibiting caspases. J. Biol. Chem. 273, 77877790.
Tamm, I., Wang, Y., Sausville, E., Scudiero, D.A., Vigna, N., Oltersdorf, T., and Reed, J.C. (1998). IAP-family protein survivin inhibits
caspase activity and apoptosis induced by Fas (CD95), Bax, caspases, and anticancer drugs. Cancer Res. 58, 53155320.
Thompson, C.B. (1995). Apoptosis in the pathogenesis and treatment of disease. Science 267, 14561462.
Thornberry, N.A., and Lazebnik, Y. (1998). Caspases: enemies
within. Science 281, 13121316.
Verhagen, A.M., Ekert, P.G., Pakusch, M., Silke, J., Connolly, L.M., Reid,
G.E., Moritz, R.L., Simpson, R.J., and Vaux, D.L. (2000). Identifi-cation of
DIABLO, a mammalian protein that promotes apoptosis by binding to and
antagonizing IAP proteins. Cell 102, 4353.