Molecular Cell, Vol. 9, 401–410, February, 2002, Copyright 2002 by Cell Press
TNF-Induced Recruitment and Activation
of the IKK Complex Require Cdc37 and Hsp90
for dimerization of IKK? and IKK?, both the LZ and HLH
motifs are important for modulating the kinase activity
of IKK (Woronicz et al., 1997; Zandi et al., 1997, 1998;
Delhase et al., 1999). NEMO contains multiple putative
coiled-coil motifs and a potential zinc finger motif at the
C terminus. NEMO is an essential component of the IKK
complex because NEMO-deficient cells are unable to
plexalsodoes notformincells lackingNEMO(Yamaoka
et al., 1998).
Accumulating evidence suggests that the IKK com-
plex represents a converging point for transducing di-
verse upstream NF-?B-activating stimuli (Karin, 1999a,
1999b). However, the detailed mechanism by which IKK
is activated in response to various stimuli, in particular
TNF and IL-1, remains unclear. A simple model for IKK
activation may involve an upstream activating kinase,
or so-called IKK kinase (IKK-K) (Mercurio and Manning,
1999; Israel, 2000). However, the identity of such an
IKK-K, if it exists, remains elusive. Like the signal trans-
ducing molecules TRADD (Hsu et al., 1996b), RIP (Hsu
et al., 1996a), and TRAF2 (Shu et al., 1996), the IKK
complex is also recruited to TNF receptor 1 (TNF-R1)
upon TNF induction (Devin et al., 2000; Zhang et al.,
2000). Signal-dependent recruitment of IKK to TNF-R1
tion. Interestingly, enforced oligomerization of RIP,
inducing IKK and NF-?B activation (Poyet et al., 2000).
We have identified Cdc37 and Hsp90, which together
form a kinase-specific chaperone, as additional compo-
an Hsp90 binding agent, reduces the size of the IKK
complex by disrupting the association of IKK/NEMO
with Cdc37/Hsp90. GA treatment also abolishes TNF-
dependent recruitment of the IKK complex to TNF-R1
indicate that Cdc37 and Hsp90 play a physiological role
in TNF-dependent translocation and activation of the
Guoqing Chen, Ping Cao, and David V. Goeddel1
Two Corporate Drive
South San Francisco, California 94080
The IKK complex, containing two catalytic subunits
IKK? and IKK? and a regulatory subunit NEMO, plays
central roles in signal-dependent activation of NF-?B.
We identify Cdc37 and Hsp90 as two additional com-
ponents of the IKK complex. IKK?/IKK?/NEMO and
Cdc37/Hsp90 form an ?900 kDa heterocomplex, which
and with the kinase domain of IKK?/IKK?. Geldana-
vation of IKK and NF-?B. GA treatment reduces the
size of the IKK complex and abolishes TNF-dependent
recruitment of the IKK complex to TNF receptor 1
(TNF-R1). Therefore, heterocomplex formation with
tion and trafficking of IKK from the cytoplasm to the
NF-?B plays prominent roles in inducible expression of
genes involved in diverse biological processes, includ-
sis (for recent reviews, see Ghosh, 1999; Maniatis, 1999;
Karin and Ben-Neriah, 2000; Karin and Delhase, 2000).
In resting cells, the majority of NF-?B is sequestered in
the cytoplasm in complexes with a class of inhibitory
proteins termed I?B. Upon exposure of cells to a variety
of extracellular stimuli, including tumor necrosis factor
(TNF) and interleukin-1 (IL-1), I?B proteins become
tion of I?B. This allows NF-?B to translocate into the
nucleus, whereit activatestranscription oftarget genes.
The protein kinase responsible for stimulus-dependent
phosphorylation of I?B has been identified using different
approaches (DiDonato et al., 1997; Mercurio et al., 1997;
Regnier et al., 1997; Woronicz et al., 1997; Zandi et al.,
1997). Biochemical purification defines the I?B kinase
sisting of at least two catalytic subunits, IKK? and IKK?,
lator, also known as IKK?/IKKAP1/FIP3) (Chen et al.,
1996; Rothwarf et al., 1998; Yamaoka et al., 1998; Zandi
et al., 1998; Li et al., 1999; Mercurio et al., 1999). IKK?
and IKK? are two highly homologous kinases, both con-
taining a conserved N-terminal kinase domain and a
C-terminal region with a leucine zipper (LZ) and a helix-
loop-helix (HLH) motif. While the LZ motif is responsible
Characterization of HeLa Flag-IKK Stable Cells
To facilitate purification of the IKK complex, we gener-
tagged IKK? and IKK?. We monitored both TNF- and
IL-1-induced NF-?B activation in these stable cells by
transient transfection with a NF-?B-dependent lucifer-
ase reporter (Figure 1A). In comparison with parental
HeLa cells, HeLa cells stably expressing wild type IKK?
(?WT) demonstrated almost normal NF-?B activation,
while IKK?WT stable cells (?WT) displayed relatively
higher basal and induced NF-?B activities. In contrast,
stable expression of kinase-inactive IKK? (?KA) or IKK?
(?KA) significantly reduced basal and cytokine-induced
NF-?B activities. In spite of these differences, each of
Figure 1. Characterization of HeLa Cells Stably Expressing Flag-IKK
(A) NF-?B activation in Flag-IKK stable cells. Parental HeLa cells (HeLa) or cells that stably express Flag-tagged wild-type IKK? (?WT), IKK?
(?WT), or kinase-inactive IKK? (?KA), IKK? (?KA) were cotransfected with an NF-?B-dependent luciferase reporter and a control Renilla
luciferase reporter. Reporter activities were measured 4 hr after cells were stimulated with TNF or IL-1. Relative luciferase activities were
normalized to the control Renilla luciferase activities. Each bar is the average plus the standard deviation of three independent duplicate
(B) IKK activation in Flag-IKK stable cells. Extracts from above HeLa and stable cell lines that had been treated with (?) or without (?) TNF
were immunoprecipitated (IP) with anti-Flag antibodies. Each immunoprecipitate was analyzed by anti-Flag immunoblotting (IB) and in vitro
kinase assay (KA) using GST-I?B? (1–54) as the IKK substrate. Equal amounts of total cell extracts (TCE) from these cells were also immunoblot-
ted with anti-Flag, anti-IKK?/anti-IKK? (IKK?/?), and anti-NEMO (served as a loading control) antibodies (lower three panels) to determine
these HeLa cell lines exhibited roughly similar ratios
(induced to basal) of NF-?B induction by TNF or IL-1.
These results suggest that in the cell lines examined
here, stable expression of various Flag-IKKs exerts only
minor effects on the endogenous NF-?B signaling path-
way. This behavior may be explained by the relatively
low levels of overexpression of various Flag-IKK pro-
teins, particularly in the ?KA and ?KA cell lines, in com-
parison with the endogenous levels of IKK? and IKK?
(Figure 1B, lower three panels). In addition, immunopre-
cipitation and in vitro kinase assay revealed that the
kinase activity of the Flag-IKK-containing complexes
was TNF-inducible in all four cell lines (Figure 1B, upper
2 panels). The observed IKK activation by TNF in ?KA
and ?KA cells is likely due to the presence of endoge-
nous IKK? and/or IKK? in the Flag-IKK-containing com-
To further purify the IKK complex, we fractionated the
Flag eluates by gel filtration with a Superose 6 column
(Figure 2B). Each fraction was then analyzed by immu-
noblotting with anti-IKK?/anti-IKK? (mix) and anti-NEMO
antibodies and in vitro kinase assays to monitor TNF-
dependent IKK activation. The peak levels of NEMO and
TNF-inducedIKK activitycoeluted infraction 10(marked
by arrowheads), corresponding to an exclusion size of
?900 kDa. This result is consistent with previous find-
ings that I?B kinase activity exists as a 700–900 kDa
Mercurio et al., 1997). To determine if known proteins
implicated in NF-?B signaling were present in our purified
a panel of antibodies against TRADD (Hsu et al., 1995),
RIP (Hsu et al., 1996a), TRAF2 (Hsu et al., 1996b), TRAF6
(Cao et al., 1996), I-TRAF (Rothe et al., 1996), MEKK1
(Mercurio et al., 1997), NIK (Malinin et al., 1997), AKT
(Ozes et al., 1999), ACT1/CIKS (Leonardi et al., 2000; Li
et al., 2000), MKP1 (Mercurio et al., 1997), and I?B? (Mer-
curio et al., 1997). We failed to detect any of these mole-
cules in the purified IKK complexes (data not shown).
Purification of the IKK Complex
from Flag-IKK Stable Cells
We next attempted to purify the IKK complex from the
Flag-IKK stable cell lines. A representative purification
of the IKK complex from the ?KA cell line is illustrated
in Figure 2. For each purification, we harvested 40 liters
of cells grown in suspension after cells were treated
with TNF or left untreated. Cells were then gently lysed
under isotonic conditions using 0.1% nonionic deter-
gent NP-40 to open the plasma membrane and release
the cytosol. Cleared lysates were loaded onto an anti-
Flag affinity column. After extensive washing, bound
proteins were eluted using Flag peptides. Aliquots of
the eluates derived from TNF stimulated (?) or unstimu-
lated (?) cells were separated by SDS-PAGE and visual-
ized by Coomassie blue staining (Figure 2A). Similar
polypeptide species were observed for all purifications.
Purified IKK Complexes Contain Cdc37 and Hsp90
We combined fractions containing peak IKK activities
(fractions 8 ? 9 and 10 ? 11) and resolved them by
preparative SDS-PAGE (Figure 2C). Coomassie blue
staining revealed almost identical polypeptides present
in the TNF-stimulated and -unstimulated IKK com-
plexes, except that a faint ?60 kDa polypeptide (labeled
p60) was detected only in the unstimulated complex.
We excised all visible bands and digested them with
trypsin. The tryptic peptide mixture from each in-gel
digest was analyzed by mass spectrometry. The re-
sulting peptide mass fingerprints were compared with
The IKK Complex Contains Cdc37 and Hsp90
Figure 2. Purification of the IKK Complex from Flag-IKK Stable Cells
(A) A Coomassie blue-stained SDS-PAGE gel containing samples affinity purified from TNF-unstimulated (?) or -stimulated (?) ?KA stable
cells. Samples represent Flag peptide eluates from the anti-Flag affinity columns and were used as the starting material (SM) for further
purification shown in (B) and (C).
(B) Gel filtration chromatography to isolate the IKK complex. The Flag peptide eluates from (A) were further fractionated on a Superose 6 gel
filtration column. Each fraction was analyzed by immunoblotting (IB) with anti-IKK?/anti-IKK? (IKK?/?) and anti-NEMO antibodies and in vitro
kinase assays (KA) to monitor the IKK enzymatic activities. An anti-Flag immunoblotting was also performed, revealing a pattern very similar
to that of anti-IKK (data not shown). Elution positions of protein standards run in a parallel experiment are indicated above corresponding
fractions. Arrowheads mark the peaks of purified IKK complexes.
(C) Identification of protein components of the purified IKK complexes. The peak fractions (8 ? 9 and 10 ? 11) from (B) were pooled,
concentrated, and resolved by preparative SDS-PAGE. Coomassie blue-stained bands were excised, and their identities (as indicated on the
right) were determined by peptide mass fingerprint mapping.
the theoretical tryptic peptide masses derived from the
NCBInr database. This allowed us to unambiguously
identify IKK? and NEMO in the purified complex (Figure
2C, right). IKK? was not found in the purified complex
by peptide mass fingerprint mapping, even though a
small amount of IKK? was indeed detected in this puri-
fied IKK complex by immunoblotting (data not shown).
The presence of IKK? explains why TNF-induced IKK
activation was observed for the IKK complex purified
from the ?KA cell line (Figure 2B). These results also
suggest that the majority of the IKK complex isolated
from the ?KA cell line is composed of overexpressed
Flag-IKK? and endogenous NEMO. In parallel to these
findings, complexes containing predominantly Flag-
IKK?and NEMOwerepurifiedfrom theFlag-IKK?stable
cell lines (data not shown).
In addition to IKK?/IKK? and NEMO, we consistently
observed the presence of Hsp90, Cdc37, and ?-tubulin
in the IKK complexes purified from different Flag-IKK
stable cells (Figure 2 and data not shown). Cdc37 and
Hsp90 have been previously shown to function as a
kinase-specific chaperone (Csermely et al., 1998; Pratt
et al., 1999). Coomassie blue staining revealed similar
amounts of Cdc37, Hsp90, IKK, and NEMO present in
the purified IKK complexes (Figure 2C and data not
shown), suggesting that Cdc37/Hsp90 and IKK?/IKK?/
NEMO form a heterocomplex at a stoichiometric ratio
of 1:1. Conversely, both ?-tubulin and p60 appear to
represent substoichiometic components of the IKK
complex. The in-gel digest of p60 generated a peptide
mass fingerprint that failed to match any known protein
in the database.
ble cells by immunoblotting (data not shown). To ex-
clude the possibility that this copurification is due to an
artifact of Flag-IKK overexpression in these cells, we
purified the endogenous IKK complex from nontrans-
fected HeLa cells by anti-IKK? affinity chromatography.
Gel filtration analysis of the purified endogenous IKK
complex demonstrated that Cdc37/Hsp90 precisely co-
eluted with IKK?, IKK?, and NEMO at a peak corre-
sponding to an exclusion size of 900 kDa (Figure 3A,
arrowhead). Interestingly, Hsp90, Cdc37, and IKK?/?
were also present in some smaller molecular weight
fractions, perhaps reflecting the IKK complexes that
proteins, or that were disrupted during the fractionation
process. Aliquots of fractions 9, 10, and 11 derived from
the gel filtration column were resolved by SDS-PAGE
and visualized by silver staining (Figure 3B). Similar
amounts of Hsp90, IKK?, IKK?, Cdc37, and NEMO were
detectedin thepeakfractionof thepurifiedendogenous
IKK complex (arrowhead). Thus, all our purification re-
sultsprovide evidenceto stronglysupport theexistence
of a stoichiometric heterocomplex of IKK?/IKK?/NEMO
and Cdc37/Hsp90 in the cell.
Cdc37/Hsp90 Associates with the Kinase Domains
The presence of Cdc37/Hsp90 in the IKK complex was
further confirmed by an endogenous coimmunoprecipi-
tation experiment (Figure 4A). Both Cdc37 and Hsp90
were coprecipitated from HeLa cell extracts with anti-
NEMO and anti-IKK? antibodies, but not with a control
antibody. In addition, the specific interaction of Cdc37/
Hsp90 with IKK/NEMO is independent of TNF stimula-
tion. To further assess the stability of this interaction,
we subjected equal amounts of anti-NEMO immunopre-
cipitates to extensive washes with the lysis buffer alone
Purification of the Endogenous IKK Complex
We have verified the copurification of Cdc37 and Hsp90
with the IKK complexes isolated from the Flag-IKK sta-
Figure 3. Purification of the Endogenous IKK
(A) Gel filtration analysis of the endogenous
IKK complex purified from HeLa cells by anti-
IKK? affinity chromatography. Gel filtration
tion was analyzed by immunoblotting with
anti-Cdc37, and anti-NEMO antibodies. The
arrowhead denotes the peak of the purified
endogenous IKK complex that contains IKK,
NEMO, Cdc37, and Hsp90.
(B) A silver stained SDS-PAGE gel containing
aliquots of fractions 9, 10, and 11 from the
gel filtration fractionation in (A). The identity
of each polypeptide band in fraction 10 was
determined by mass spectrometry (as indi-
cated on the right). Asterisks denote protein
bands that either nonspecifically bind the
anti-IKK? affinity matrix or are absent in the
peak of the IKK complex.
or the buffer containing 1 M NaCl, 1% Triton X-100, or
2 M urea (Figure 4B). Immunoblotting analyses showed
that the association of IKK/NEMO with Cdc37/Hsp90
was vulnerable to high salt and denaturing agent
washes, while it tolerated most detergent washes. This
finding explains the earlier failure to identify Cdc37 and
Hsp90 as components of the IKK complex (DiDonato et
al., 1997; Mercurio et al., 1997), as these purifications
usually incorporated an initial step of stringent wash
with high salt or denaturing agent.
tagged IKK? and IKK? and immunoprecipitated cell ex-
tracts with anti-Flag antibodies. Endogenous Cdc37 and
Hsp90 were detected in the immunoprecipitates con-
taining the full-length Flag-IKK? and Flag-IKK? (Figure
Figure 4. Cdc37TargetsHsp90totheKinase
Domains of IKK
(A) Coimmunoprecipitation of endogenous
IKK/NEMO with Cdc37/Hsp90. Extracts from
anti-NEMO (NEMO), anti-IKK? (IKK?), or a
control IgG1 (CTRL) monoclonal antibody.
Each immunoprecipitate was immunoblotted
(IB) with rabbit polyclonal antibodies against
Hsp90, IKK?/IKK? (IKK?/?), Cdc37, and
(B) Stability of the interaction between IKK/
NEMO and Cdc37/Hsp90. Equal amounts of
those anti-NEMO precipitates in (A) were
washed extensively with the buffer alone
(lane 1) or the buffer containing either 1.0 M
NaCl (lane 2), 1% Triton X-100 (lane 3), or 2.0
M urea (lane 4) and subsequently analyzed
by immunoblotting as shown in (A). Asterisks
in both (A) and (B) mark those nonspecific
(C) Cdc37/Hsp90 interacts with the kinase
domains of IKK. HeLa cells were transfected
with vector alone or vectors expressing vari-
ous deletions of Flag-tagged IKK? or IKK?.
Cell extracts were immunoprecipitated with
anti-Flag antibodies, and coprecipitating
Cdc37 and Hsp90 proteins were determined
by immunoblotting with anti-Hsp90 and anti-
Cdc37 antibodies. Expression levels of IKK?
and IKK? variants were monitored by immu-
noblotting with anti-Flag (biotinylated) an-
(D) A direct interaction between Cdc37 and
the kinase domains of IKK. Equal amounts of purified HA-Cdc37 (INPUT) were incubated with the glutathione beads containing GST alone,
GST-IKK?/IKK?-kinase domain (KD), or GST-NEMO fusion proteins. After extensive washing, amounts of HA-Cdc37 retained on beads were
determined by anti-HA immunoblotting. The bottom is a Coomassie blue (CBB)-stained SDS-polyacrylamide gel containing purified HA-Cdc37
and various GST fusion proteins.
The IKK Complex Contains Cdc37 and Hsp90
Figure 5. TNF-Induced Activation of NF-?B
and IKK Requires Cdc37/Hsp90
plex formation between IKK/NEMO and
precipitated (IP) with a control IgG1 (CTRL)
or anti-NEMO (NEMO) monoclonal antibody.
The presence of Hsp90, Cdc37, IKK?/IKK?,
and NEMO in the immunocomplexes was de-
in Figure 4A, except that a goat anti-Cdc37
polyclonal antibody was used.
(B) Effects of GA treatment on NF-?B- and
AP1-dependent transcription. HeLa cells
were cotransfected with a NF-?B- (left) or
AP1-driven (right) luciferase reporter along
with a control Renilla luciferase reporter. 24
hr later, cells were treated with DMSO or GA
prior toTNF and IL-1 induction.Relative lucif-
erase activities were measured as described
in Figure 1A.
(C) Effects of GA treatment on IKK activation.
(?) TNF. The IKK complexes were isolated by
anti-NEMO immunoprecipitation and further
analyzed by in vitro kinase assays (KA) using
GST-I?B? (1–54) as the IKK substrate (bot-
tom). TheIKK protein level ineach precipitate
was also determined by immunoblotting.
Both immunopreciptation and in vitro kinase
assay were performed essentially as de-
scribed in (C), except that an anti-JNK1
monoclonal antibody and the GST-Jun (1–79)
substrate were used for each assay.
4C). Furthermore, IKK deletions containing only the ki-
nase domain were also capable of binding Cdc37 and
Hsp90, while those without the kinase domain failed to
do so. A direct interaction between the IKKs and Cdc37
was further ascertained by a GST pull-down assay (Fig-
ure 4D). Purified HA-tagged Cdc37 was retained on
beads containing GST-IKK? or GST-IKK?-kinase do-
main, but not on beads with GST alone (the faint band
likely represents the shadow of overloaded input) or
GST-NEMO. Given the high input, the binding between
interacts more strongly with IKK in the presence of
Hsp90 and that purified Hsp90 by itself associates
poorly with GST-IKK fusions (data not shown). This sug-
gests a cooperative binding of Cdc37/Hsp90 to IKK.
Thus, Hsp90 is targeted to the IKK complex via a direct
interaction between Cdc37 and the catalytic region of
noblotting analysis showed that the treatment with GA
almost completely disrupted the interaction between
IKK/NEMO and Hsp90/Cdc37, but DMSO exhibited no
detectable effect. Thus, GA provides a powerful tool to
We next studied the impact of GA treatment on NF-?B
both TNF- and IL-1-induced NF-?B activation, but also
moderately inhibited (2- to 3-fold) the unstimulated NF-
?B activity (Figure 5B, left). To determine the specificity
of GA inhibition on NF-?B activation, we performed a
similar experiment with an AP1-dependent luciferase
partially inhibited TNF- and IL-1-induced AP1-mediated
reporter activities. These effects were much less severe
than those on NF-?B activation. In addition, GA treat-
ment had minimal effects on transcription of the internal
control Renilla luciferase reporter (?2-fold). These re-
sults indicate that GA selectively impedes the NF-?B
We also performed in vitro immunocomplex kinase
assays to measure the effects of GA treatment on TNF-
induced IKK activation (Figure 5C). HeLa cells were
treated with GA or DMSO prior to TNF induction, and
in vitro kinase assays were performed on the immuno-
precipitated IKK complexesusing exogenous GST-I?B?
fusion proteins as substrates. TNF stimulation resulted
TNF-Induced Activation of IKK and NF-?B
ing of the association of IKK/NEMO with Cdc37/Hsp90,
we employed an antitumor agent, geldanamycin (GA).
GA was previously shown to interact and interfere with
the normal functions of Hsp90 (Whitesell et al., 1994).
Using anti-NEMO immunoprecipitation, we isolated the
with either DMSO or GA (Figure 5A). Subsequent immu-
Figure 6. Roles for Hsp90/Cdc37 in Complex Assembly and Receptor Recruitment of IKK
(A and B) Gel filtration analysis of the IKK complexes. Equal amounts of total extracts from HeLa cells that had been treated with DMSO (A)
or GA (B) were fractionated on a Superose 6 column as described in Figure 2B. The IKK complex sizes were monitored by immunoblotting
(IB) analysis of each fraction with antibodies against IKK? and IKK? (IKK?/?) and NEMO.
(C) TNF-induced recruitment of IKK to TNF-R1 requires Hsp90/Cdc37. HeLa cells were treated with (?) or without (?) TNF after cells had
been preincubated with DMSO or GA. Equal amounts of cell extracts were immunoprecipitated (IP) with anti-TNF-R1 (985) monoclonal
antibodies. TNF-R1 and associated IKK?, Hsp90, RIP, TRAF2, and TRADD were detected by immunoblotting with the indicated polyclonal
antibodies (right panels). Total cell extracts (TCE) were also prepared and immunoblotted in the same way to monitor the effects of GA
treatment on cellular levels of these proteins (left panels). Asterisks indicate the protein bands that are nonspecifically recognized by respective
antibodies. Brackets mark the higher molecular weight species of RIP and TRAF2 that become visible only after TNF stimulation.
in a dramatic increase of IKK enzymatic activities. How-
ever, the pretreatment with GA almost completely abol-
ished this robust activation of IKK in response to TNF.
As a control, we also investigated if GA treatment might
affect c-Jun N-terminal kinase 1 (JNK1) activation by
TNF (Figure 5D). Similar to the DMSO control, GA exhib-
ited almost no effect on TNF-induced JNK1 activation,
even though the unstimulated JNK1 activity was slightly
reduced by GA treatment. In addition, GA had been
previously shown to block TNF-induced degradation of
I?B, indicating that GA acts on this arm of NF-?B re-
sponse (Lewis et al., 2000). These results together dem-
onstrate that the specific formation of a heterocomplex
between IKK/NEMO and Cdc37/Hsp90 is essential for
TNF-induced activation of IKK and NF-?B.
significantly increased amounts of NEMO in the lower
molecular weight fractions in comparison with those
from DMSO-treated cells (compare fractions 15–17 in
Figures 6A and 6B). The dramatic effects of GA on the
size of the endogenous IKK complex and the redistribu-
tion of IKK?/? and NEMO demonstrate a critical role for
Cdc37/Hsp90 in the assembly of functional IKK com-
plexes under physiological conditions.
TNF-Induced Recruitment of IKK to TNF-R1
We also examined if Cdc37/Hsp90 plays a role in shut-
tling the IKK complex from the cytoplasm to the plasma
membrane. HeLa cells were treated with DMSO or GA
and then briefly stimulated with or without TNF. TNF-
R1-associated signaling complexes were isolated from
these cells by immunoprecipitation using a monoclonal
antibody against the extracellular domain of TNF-R1.
Immunoblotting demonstrated TNF-dependent recruit-
to TNF-R1 in DMSO-treated cells (Figure 6C, lanes 5
to TNF-R1 after TNF stimulation, confirming that the
IKK/NEMO-Cdc37/Hsp90 heterocomplex is recruited to
TNF-R1 in a ligand-dependent manner. In GA-treated
cells, recruitment of TRAF2 and TRADD to TNF-R1 ap-
peared unchanged (Figure 6C, lanes 7 and 8). However,
GA treatment almost fully abrogated TNF-dependent re-
cruitment of IKK?, Hsp90, and RIP molecules to TNF-R1.
To determine the effects of GA treatment on expres-
sion levels of these proteins, we prepared total cell ex-
tracts and immunoblotted them with corresponding an-
tibodies (Figure 6C, left panels). GA treatment had little
or no effect on the cellular levels of IKK?, TRAF2,
To investigate if GA-mediated disruption of the interac-
tion between IKK/NEMO and Cdc37/Hsp90 might affect
the formation and composition of the IKK complex, we
fractionated equal amounts of extracts from HeLa cells
that had been treated with DMSO or GA by gel filtration.
Immunoblotting revealed that the peak of the IKK com-
plex in GA-treated cells (Figure 6B, fraction 12) eluted
with an exclusion size smaller than in DMSO-treated
cells (Figure 6A, fraction 11). Similarly, GA treatment
causes a reduction in size of the purified IKK complex
that lacks Cdc37/Hsp90 (data not shown). However, no
apparent effect was observed on the IKK activity when
GA was directly added to the in vitro kinase assays
with the purified IKK complexes (data not shown). This
suggests that GA blocks the process of IKK activation
by TNF rather than inhibits the IKK enzymatic activity
per se. GA treatment also resulted in the presence of
The IKK Complex Contains Cdc37 and Hsp90
plex in response to TNF. These results reveal a physio-
ing NF-?B activating signals.
Functions of Cdc37/Hsp90 in IKK Activation
Hsp90, one of most abundant cytosolic proteins in eu-
karyotic cells, plays an important role in the maturation,
activation, and trafficking of proteins involved in signal
transduction, cell cycle control, development, and tran-
scriptional regulation (Csermely et al., 1998; Pratt et
al., 1999). Unlike the more general Hsp70 and Hsp60
chaperones, Hsp90 appears to be a dedicated chaper-
steroid hormone receptors and protein kinases. Hsp90
functions in concert with a set of cochaperones that link
the mammalian homolog of the yeast cell cycle control
protein Cdc37 acts as a kinase-specific targeting sub-
unit for Hsp90 (Dai et al., 1996; Stepanova et al., 1996).
A growing list of kinases, including Src family tyrosine
kinases and the serine/threonine kinases Raf and MEK,
are known to exist as heterocomplexes with Cdc37/
Hsp90. Cdc37/Hsp90 contributes to the stabilization,
activation, and/or translocation of these kinases.
IKK? andIKK? cannow beadded tothe list ofkinases
Hsp90. We find that theHsp90 chaperone has no appar-
ent effect on the stability of IKK. In contrast, Hsp90
affects the integrity of RIP, another important kinase in
the TNF/NF-?B signaling pathway. This is in an agree-
ment with a previous study showing that GA selectively
induces degradation of RIP and blocks TNF-induced
NF-?B activation (Lewis et al., 2000). IKK?, IKK?, and
(Devin et al., 2000; Zhang et al., 2000). Our current study
extends these findings by revealing that Cdc37/Hsp90
is part of the IKK complex and is also recruited to TNF-
R1 in a stimulus-dependent manner. Given that RIP is
dispensable for IKK recruitment to TNF-R1, our results
reveal a critical role for Cdc37/Hsp90 in shuttling IKK
to the membrane.
GA treatment causes the dissociation of Cdc37/
Hsp90 from the IKK complex, a reduction in size of the
IKK complex, and the inhibition of TNF-induced activa-
logical rolefor Cdc37/Hsp90in theTNF/NF-?B signaling
pathway. The potent blockade of NF-?B signaling by
GA is likely due to multiple effects on the pathway. First,
GA disrupts the interaction of IKK with Cdc37/Hsp90
and prevents the recruitment of IKK to the membrane.
Second, GA treatment causes degradation of RIP, re-
sulting in the assembly of a TNF-R1 signaling complex
lacking RIP and defective in IKK activation. It is possible
that Cdc37/Hsp90 may also provide a link to compo-
nents of the upstream TNF-R1 complex, allowing the
recruitment of IKK to the receptor. For example, both
RIP and IKK associate with Cdc37/Hsp90, which may
serve as a bridging factor to bring them into proximity.
Interestingly, a protein highly homologous to Hsp90 has
been proposed to directly bind the intracellular region
of TNF-R1 (Song et al., 1995).
Figure 7. Kinetics of GA-Mediated Inhibition of IKK Recruitment to
TNF-R1 and Complex Formation with Cdc37/Hsp90
Equal numbers of HeLa cells were treated with GA for the indicated
times (0–10 hr) and then stimulated with TNF for 10 min. Aliquots
of extracts prepared from these cells were immunoprecipitated (IP)
witheither anti-TNF-R1(A)or anti-NEMO(B) monoclonalantibodies.
Anti-TNF-R1 immunoprecipitates were further analyzed by immu-
noblotting (IB) with anti-IKK? and anti-TRADD polyclonal antibodies
(A), while anti-NEMO immunoprecipitates were immunoblotted with
anti-IKK? and anti-Cdc37 polyclonal antibodies (B).
TRADD, and TNF-R1 proteins, while it modestly upregu-
lated Hsp90 expression. Incontrast, RIP expression lev-
els were greatly diminished in GA-treated cells, sug-
gesting a role for Cdc37/Hsp90 in stabilizing RIP. This
is in an agreement with a previous report that RIP is
also an Hsp90-associated kinase and GA treatment se-
lectively causes RIP degradation (Lewis et al., 2000).
Thus, the defective recruitment of RIP to TNF-R1 in GA-
treated cells could be due simply to the rapid turnover
of RIP when it is dissociated from Hsp90. However, RIP
is not required for IKK recruitment to TNF-R1 (Devin et
al., 2000), and GA treatment has no obvious effect on
the levels of IKK and NEMO proteins. In addition, time
course experiments indicate that the kinetics of GA-
induced inhibition of IKK recruitment to TNF-R1 corre-
late with those of GA-mediated disruption of the IKK-
Cdc37/Hsp90 complex (Figure 7). GA treatment for 5 hr
significantly reduced the amount of IKK? recruited to
TNF-R1 and the amount of Cdc37 associated with IKK,
whereas 10 hr treatment almost completely abolished
both interactions. Taken together, these results support
an important role for Cdc37/Hsp90 in trafficking the IKK
complex from the cytoplasm to the membrane. It is also
possible that Cdc37/Hsp90 may be required for the in-
teraction of the IKK complex with components of the
upstream TNF-R1 signaling complex.
We report the purification and identification of Cdc37
and Hsp90 as additional components of the IKK com-
plex. Several lines of evidence argue that the formation
of a heterocomplex with Cdc37/Hsp90 is essential for
assembly, translocation, and activation of the IKK com-
Stoichiometry of the IKK-Cdc37/Hsp90 Complex
Our gel filtration analysis has revealed that the endoge-
reported previously for the native IKK complex isolated
from HeLa cells (DiDonato et al., 1997; Mercurio et al.,
1997). While the native IKK complex contains a hetero-
dimer of IKK? and IKK? (Mercurio et al., 1997; Rothwarf
et al., 1998; Woronicz et al., 1997), purified recombinant
ular weight (MW) of 230–250 kDa (Zandi et al., 1998).
kDa (Pratt et al., 1999). Cdc37 has a size of 130–150 kDa
under native conditions, suggesting that it is a dimer or
trimer (Kimura et al., 1997). NEMO is also known to form
suggest that there are roughly equal amounts of IKK,
NEMO, Cdc37, and Hsp90 in the purified IKK complex.
Based on these data, we predict that the complete IKK
a homodimer of Hsp90, and two to three molecules
of NEMO and Cdc37. Such a complex would have an
estimated MW of ?800 kDa, matching the size of the
native IKK complex and providing further evidence for
that Cdc37 and Hsp90 are bona fide subunits of the IKK
ity in the TNF-R1 signaling complex is directly responsi-
ble for activating IKK. The diversity of NF-?B stimuli
forecasts the existence of different upstream activators
involved in signal-dependent IKK activation. It is thus
ing pathways may utilize a pathway-specific factor to
directly trigger IKK activation. To date, however, most
of these putative direct activators of IKK have not been
tional components, in particular substoichiometric and/
or transiently associated subunits, will improve our un-
derstanding of the mechanism of IKK activation as well
as the specificity and diversity of NF-?B signaling.
Recombinant human TNF and IL-1 were obtained from Genentech
(South San Francisco, CA) and BioSource International (Camarillo,
CA), respectively. Geldanamycin (GA), anti-Flag M2 affinity gel, Flag
peptide, anti-Flag M2, and biotinylated anti-Flag M2 monoclonal
anti-NEMO, anti-Hsp90, anti-TRAF2, anti-HA, anti-TRADD, anti-
TNF-R1, and anti-RIP polyclonal antibodies as well as goat anti-
Cdc37 polyclonal antibody were products from Santa Cruz Biotech-
nology (SantaCruz, CA).Rabbit polyclonalanti-Cdc37 antibodywas
purchased from StressGen (Victoria, BC, Canada). Anti-NEMO (3F6)
monoclonal antibody was provided by Z. Cao (Tularik), and anti-
JNK1 monoclonal antibody was purchased from Pharmingen (San
Diego, CA). Anti-TNF-R1 (985) monoclonal antibody has been de-
scribed previously (Hsu et al., 1996b). Anti-HA matrix (3F10) was pur-
chased from Roche Molecular Biochemicals (Indianapolis, IN), and
HA peptides were synthesized by L. Huang (Tularik). Purified GST-
and GST-I?B? (1–54) proteins were provided by W. Liu (Tularik).
Mechanisms of IKK Activation
Althoughthis studyandothershave shownacorrelation
between the recruitment of IKK to TNF-R1 and the acti-
vation of IKK, the exact mechanism of IKK activation is
still an enigma. It is plausible that the recruitment of the
IKK complex to TNF-R1 may represent an initial step.
Once recruited and activated, IKK may be released from
the receptor, reentering the cytoplasm and in trans acti-
like many signaling molecules, IKK may traverse the
cytoplasm simply by diffusion and become trapped at
the sites of action by protein-protein interactions (e.g.,
those involved in the formation of TNF-R1 signaling
complexes at the internal surface of the membrane). It
is also possible that the IKK complex may move through
the cytoplasm via a transport system involving Cdc37/
Hsp90, even though nothing is known about how this
movement is initiated or directed to the cell membrane.
RIP plays an essential role in TNF-mediated IKK acti-
vation (Devin et al., 2000; Hsu et al., 1996a; Kelliher et
al., 1998; Ting et al., 1996; Zhang et al., 2000). Following
the recruitment of IKK to TNF-R1, therefore, a RIP-
dependent process or factor may directly trigger IKK
activation. One possibility is that an upstream kinase
(e.g., IKK-K) is brought to the receptor complex and
that enforced oligomerization of RIP, NEMO, IKK?, or
2000). These results suggest that such an intermediate
factor or IKK-K may not be necessary for IKK activation.
RIP itself may cause some conformational changes of
the IKK complex, resulting in IKK activation. However,
To fully understand the mechanism of TNF-mediated
Expression vectors for Flag-tagged wild-type IKK? and IKK?, and
their kinase inactive (KA) mutants, have been described previously
(Ling et al., 1998). pRK7-N-Flag-IKK? and pRK5-C-Flag-IKK? con-
structs for expressing various Flag-tagged IKK deletions, as well
as pGEX4T1-NEMO, pGEX4T1-IKK? (1–304), and pGEX4T1-IKK?
(1–316) for producing GST-NEMO and -IKK fusion proteins, were
generated by PCR. The full-length Cdc37 cDNA was isolated from
human HeLa Marathon-Ready cDNA library by PCR (Clontech, Palo
Alto, CA) and further cloned into pVL1392 vector (Pharmingen) for
generating recombinant baculovirus expressing HA-tagged Cdc37.
NF-?B-dependent (pNF-?B5x-Luc) and AP1-dependent (pAP17x-Luc)
firefly luciferase reporters were products from Stratagene. The Re-
nilla luciferase reporter utilized to normalize transfection efficiency
in reporter assays has been described (Chen et al., 1999).
Transfection and Reporter Assays
HeLa cell lines stably expressing Flag-IKKs were generated using
standard methods. Transient transfection was performed as de-
scribed previously (Regnier et al., 1997), except using the Effectene
transfection reagent (Qiagen, Valencia, CA). Dual-luciferase assays
were conducted with the kit from Promega. Luciferase activities
were normalized to the control Renilla luciferase activities as de-
scribed (Chen et al., 1999). For reporter assays in the GA-treated
cells, HeLa cells were transfected with indicated reporters. 24 hr
posttransfection, cells were incubated with 0.5 ?M of GA in serum-
free medium for 15 hr before TNF or IL-1 stimulation. GA concentra-
tion was maintained at 0.5 ?M throughout stimulation.
To purify the IKK complex, 40 liters of Flag-IKK HeLa cells were
grown in suspension and harvested after cells were treated with or
in 10 volumes of the isotonic protein lysis buffer IP150containing 50
mM HEPES (pH 7.6), 150 mM NaCl, 1 mM EDTA, 0.1% NP-40, 10%
The IKK Complex Contains Cdc37 and Hsp90
glycerol, 1 mM DTT, 0.5 mM PMSF, 20 mM ?-glycerophosphate,
20 mM p-nitrophenyl phosphate, 1 mM Na3VO4, and 1? complete
were precipitated with anti-Flag M2 affinity gel at a volume ratio of
200:1. After extensive washing, beads were loaded onto a column,
and bound proteins were eluted with 4-bead volumes of the IP150
buffer containing 0.3 mg/ml of Flag peptide. Flag eluates were frac-
tionated with a Superose 6 HR 30/10 column connected to a FPLC
system (Amersham Pharmacia, Piscataway, NJ). The column was
precalibrated withstandards blue dextran (2,000kDa), thyroglobulin
(670 kDa), and ?-amylase (200 kDa). Similarly, the endogenous IKK
complex was isolated from untransfected HeLa cells using an anti-
IKK? affinity matrix (Santa Cruz Biotechnology). GST pull-down
assays were performed as described previously (Chen et al., 1999).
Dai, K., Kobayashi, R., and Beach, D. (1996). Physical interaction of
mammalian CDC37 with CDK4. J. Biol. Chem. 271, 22030–22034.
Delhase, M., Hayakawa, M., Chen, Y., and Karin, M. (1999). Positive
and negative regulation of I?B kinase activity through IKK? subunit
phosphorylation. Science 284, 309–313.
Devin, A., Cook, A., Lin, Y., Rodriguez, Y., Kelliher, M., and Liu, Z.
(2000). The distinct roles of TRAF2 and RIP in IKK activation by
TNF-R1: TRAF2 recruits IKK to TNF-R1 while RIP mediates IKK
activation. Immunity 12, 419–429.
DiDonato, J.A., Hayakawa, M., Rothwarf, D.M., Zandi, E., and Karin,
M. (1997). A cytokine-responsive I?B kinase that activates the tran-
scription factor NF-?B. Nature 388, 548–554.
Ghosh, S. (1999). Regulation of inducible gene expression by the
transcription factor NF-?B. Immunol. Res. 19, 183–189.
Hsu, H., Xiong, J., and Goeddel, D.V. (1995). The TNF receptor
1-associated protein TRADD signals cell death and NF-?B activa-
tion. Cell 81, 495–504.
Hsu, H., Huang, J., Shu, H.B., Baichwal, V., and Goeddel, D.V.
(1996a). TNF-dependent recruitment of the protein kinase RIP to
the TNF receptor 1 signaling complex. Immunity 4, 387–396.
Hsu, H., Shu, H.B., Pan, M.G., and Goeddel, D.V. (1996b). TRADD-
TRAF2 and TRADD-FADD interactions define two distinct TNF re-
ceptor 1 signal transduction pathways. Cell 84, 299–308.
Israel, A. (2000). The IKK complex: an integrator of all signals that
activate NF-?B? Trends Cell Biol. 10, 129–133.
Karin, M. (1999a). How NF-?B is activated: the role of the I?B kinase
(IKK) complex. Oncogene 18, 6867–6874.
Karin, M. (1999b). The beginning of the end: I?B kinase (IKK) and
NF-?B activation. J. Biol. Chem. 274, 27339–27342.
Karin, M., and Ben-Neriah, Y. (2000). Phosphorylation meets ubiqui-
tination: the control of NF-?B activity. Annu. Rev. Immunol. 18,
Karin, M., and Delhase, M. (2000). The I?B kinase (IKK) and NF-?B:
key elements of proinflammatory signalling. Semin. Immunol. 12,
Kelliher, M.A., Grimm, S., Ishida, Y., Kuo, F., Stanger, B.Z., and
Leder, P. (1998). The death domain kinase RIP mediates the TNF-
induced NF-?B signal. Immunity 8, 297–303.
Kimura, Y., Rutherford, S.L., Miyata, Y., Yahara, I., Freeman, B.C.,
Leonardi, A., Chariot, A., Claudio, E., Cunningham, K., and Sieben-
list, U. (2000). CIKS, a connection to I?B kinase and stress-activated
protein kinase. Proc. Natl. Acad. Sci. USA 97, 10494–10499.
Mass Spectrometry Analysis
the gel, washed, and dehydrated. Dried gel slices were reduced
to in-gel digest with trypsin. Aliquots of tryptic peptides derived
from each in-gel digest were spotted on the matrix-assisted laser
desorption/ionization (MALDI) sample plate. Mass spectra were ac-
quired using a Voyager-DE STR Biospectrometry Workstation (Per-
Septive Biosystems, Framingham, MA), operated in reflectron mode
with delayed extraction. Proteins were identified by comparing ob-
served peptide mass fingerprints with those theoretically derived
from NCBInr database using the Protein Prospector AutoMS-Fit
Immunoprecipitation and Kinase Assays
For immunoprecipitation of overexpressed proteins, cells were
transfected with 0.5 ?g of each indicated expression vector. 24 hr
later, cells were lysed in the buffer IP150. Cleared cell extracts were
incubated with 2 ?g of anti-Flag antibody and 20 ?l of protein G
agarose beads (Upstate Biotechnology, Lake Placid, NY). For pre-
cipitating endogenous IKK complexes, 2 ?g of anti-NEMO (3F6) or
isotype-matched monoclonal antibodies were used. For TNF-R1
immunoprecipitation, HeLa cells were induced with TNF (100 ng/
ml) for 5 min and lysed in the IP150buffer containing 1% Triton X-100.
buffer. For kinase assays, each immunopreciptate was incubated
with or without 1 ?g of exogenous substrate GST-I?B? (1–54) or
GST-c-Jun (1–79) in 20 ?l of kinase buffer containing 20 mM HEPES
(pH 7.6), 20 mM MgCl2, 20 mM ?-glycerophosphate, 1 mM DTT, 10
?M ATP, and 5 ?Ci [?-32P]ATP. Samples were separated by SDS-
PAGE, and32P-labeled proteins were visualizedby autoradiography.
We thank Zhaodan Cao, Xiong Gao, Lei Ling, Wei Liu, and Linda
Huang for providing various reagents and Miki Rich for DNA se-
ating the HeLa Flag-IKK stable cell lines and to Xiaohong Liu for
assistance with gel filtration chromatography. We thank Zhulun
Wang and Nigel Walker for their help in the preparation of figures.
Lewis, J., Devin, A., Miller, A., Lin, Y., Rodriguez, Y., Neckers, L.,
tion of the death domain kinase, receptor-interacting protein (RIP),
and blockage of tumor necrosis factor-induced nuclear factor-?B
activation. J. Biol. Chem. 275, 10519–10526.
Received June 1, 2001; revised November 12, 2001.
Li, Y., Kang, J., Friedman, J., Tarassishin, L., Ye, J., Kovalenko, A.,
Wallach, D., and Horwitz, M.S. (1999). Identification of a cell protein
(FIP-3) as a modulator of NF-?B activity and as a target of an adeno-
virus inhibitor of tumor necrosis factor ?-induced apoptosis. Proc.
Natl. Acad. Sci. USA 96, 1042–1047.
Cao, Z., Xiong, J., Takeuchi, M., Kurama, T., and Goeddel, D.V.
(1996). TRAF6 is a signal transducer for interleukin-1. Nature 383,
Chen, Z.J., Parent, L., and Maniatis, T. (1996). Site-specific phos-
phorylation of I?B? by a novel ubiquitination-dependent protein ki-
nase activity. Cell 84, 853–862.
Chen, G., Fernandez, J., Mische, S., and Courey, A.J. (1999). A
functional interaction between the histone deacetylase Rpd3 and
the corepressor Groucho in Drosophila development. Genes Dev.
Csermely, P., Schnaider, T., Soti, C., Prohaszka, Z., and Nardai, G.
(1998). The 90-kDa molecular chaperone family: structure, function,
Li, X., Commane, M., Nie, H., Hua, X., Chatterjee-Kishore, M., Wald,
D., Haag, M., and Stark, G.R. (2000). Act1, an NF-?B-activating pro-
tein. Proc. Natl. Acad. Sci. USA 97, 10489–10493.
Ling, L., Cao, Z., and Goeddel, D.V. (1998). NF-?B-inducing kinase
activates IKK-? by phosphorylation of Ser-176. Proc. Natl. Acad.
Sci. USA 95, 3792–3797.
Malinin, N.L., Boldin, M.P., Kovalenko, A.V., and Wallach, D. (1997).
MAP3K-related kinase involved in NF-?B induction by TNF, CD95
and IL-1. Nature 385, 540–544.
Maniatis, T. (1999). A ubiquitin ligase complex essential for the
Molecular Cell Download full-text
NF-?B, Wnt/Wingless, and Hedgehog signaling pathways. Genes
Dev. 13, 505–510.
Mercurio, F., and Manning, A.M. (1999). Multiple signals converging
on NF-?B. Curr. Opin. Cell Biol. 11, 226–232.
Mercurio, F., Zhu, H., Murray, B.W., Shevchenko, A., Bennett, B.L.,
Li, J., Young, D.B., Barbosa, M., Mann, M., Manning, A., et al. (1997).
IKK-1 and IKK-2: cytokine-activated I?B kinases essential for NF-?B
activation. Science 278, 860–866.
Mercurio, F., Murray, B.W., Shevchenko, A., Bennett, B.L., Young,
D.B., Li, J.W., Pascual, G., Motiwala, A., Zhu, H., Mann, M., et al.
(1999). I?B kinase(IKK)-associated protein 1, acommon component
of the heterogeneous IKK complex. Mol. Cell. Biol. 19, 1526–1538.
Ozes, O.N., Mayo, L.D., Gustin, J.A., Pfeffer, S.R., Pfeffer, L.M., and
Donner, D.B. (1999). NF-?B activation by tumour necrosis factor
requires the Akt serine-threonine kinase. Nature 401, 82–85.
Poyet, J.L., Srinivasula, S.M., Lin, J., Fernandes-Alnemri, T., Ya-
maoka, S., Tsichlis, P.N., and Alnemri, E.S. (2000). Activation of the
I?B kinases by RIP via IKK?/NEMO-mediated oligomerization. J.
Biol. Chem. 275, 37966–37977.
Pratt, W.B., Silverstein, A.M., and Galigniana, M.D. (1999). A model
for the cytoplasmic trafficking of signalling proteins involving the
Hsp90-binding immunophilins and p50Cdc37. Cell. Signal. 11,
M. (1997). Identification and characterization of an I?B kinase. Cell
Rothe, M., Xiong, J., Shu, H.B., Williamson, K., Goddard, A., and
Goeddel, D.V. (1996). I-TRAF is a novel TRAF-interacting protein
thatregulates TRAF-mediatedsignal transduction.Proc. Natl.Acad.
Sci. USA 93, 8241–8246.
Rothwarf, D.M., Zandi, E., Natoli, G., and Karin, M. (1998). IKK? is
an essential regulatory subunit of the I?B kinase complex. Nature
Shu, H.B., Takeuchi, M., and Goeddel, D.V. (1996). The tumor necro-
sis factor receptor 2 signal transducers TRAF2 and c-IAP1 are com-
ponents of the tumor necrosis factor receptor 1 signaling complex.
Proc. Natl. Acad. Sci. USA 93, 13973–13978.
Song, H.Y., Dunbar, J.D., Zhang, Y.X., Guo, D., and Donner, D.B.
(1995). Identification of a protein with homology to Hsp90 that binds
the type 1 tumor necrosis factor receptor. J. Biol. Chem. 270, 3574–
Stepanova, L., Leng, X., Parker, S.B., and Harper, J.W. (1996). Mam-
malian p50Cdc37 is a proteinkinase-targeting subunit of Hsp90 that
binds and stabilizes Cdk4. Genes Dev. 10, 1491–1502.
Ting, A.T., Pimentel-Muinos, F.X., and Seed, B. (1996). RIP mediates
tumor necrosis factor receptor 1 activation of NF-?B but not Fas/
APO-1-initiated apoptosis. EMBO J. 15, 6189–6196.
Whitesell, L., Mimnaugh, E.G., De Costa, B., Myers, C.E., and Neck-
ers, L.M. (1994). Inhibition of heat shock protein HSP90-pp60v-src
heteroprotein complex formation by benzoquinone ansamycins: es-
sential role for stress proteins in oncogenic transformation. Proc.
Natl. Acad. Sci. USA 91, 8324–8328.
I?B kinase-?: NF-?B activation and complex formation with I?B
kinase-? and NIK. Science 278, 866–869.
Yamaoka, S., Courtois, G., Bessia, C., Whiteside, S.T., Weil, R.,
cloning of NEMO, a component of the I?B kinase complex essential
for NF-?B activation. Cell 93, 1231–1240.
Zandi, E., Rothwarf, D.M., Delhase, M., Hayakawa, M., and Karin,
M. (1997). The I?B kinase complex (IKK) contains two kinase sub-
units, IKK? and IKK?, necessary for I?B phosphorylation and NF-?B
activation. Cell 91, 243–252.
Zandi, E., Chen, Y., and Karin, M. (1998). Direct phosphorylation of
I?B by IKK? and IKK?: discrimination between free and NF-?B-
bound substrate. Science 281, 1360–1363.
Zhang, S.Q., Kovalenko, A., Cantarella, G., and Wallach, D. (2000).
Recruitment of the IKK signalosome to the p55 TNF receptor: RIP
and A20 bind to NEMO (IKK?) upon receptor stimulation. Immunity