(Received for publication, October 31, 1996)
From the Unité de Biologie Moléculaire de l'Expression
Génique, The transcription factor NF- The transcription factor NF- In resting cells, NF- The molecular mechanism responsible for the cytosolic retention of
NF- Unlike I Under the effect of a stimulus, I Recently, the sites of phosphorylation of I In thymocytes, I A recent study concludes that TNF- The 70Z/3 murine pre-B cell line, EL-4 murine T cell
line (kindly provided by G. Milon, Pasteur Institute), E29.1, a
CD4-negative/CD8-negative mouse T cell hybridoma (34), and 293T
(embryonic kidney cells) were maintained in RPMI medium supplemented
with 10% fetal calf serum and 50 µM The antisera used were the following. Anti-RelA
antiserum 1226 (kindly provided by N. Rice), was raised against amino
acids 537-550 of human RelA. I Expression vectors for transfection into 70Z/3
cells were obtained by subcloning cDNAs encoding I 70Z/3 were pelleted and resuspended in
complete medium at 5 × 106 cells/0.5 ml, and the
cells were electroporated in 4-mm cuvettes with a Eurogentec Cellject
electroporator at 260 V, 1,500 microfarads, and infinite resistance.
Cells were diluted into 5 ml of complete medium and were left to
recover overnight before the addition of 1 mg/ml G418. Once
established, clones were maintained in complete medium supplemented
with 0.5 mg/ml G418.
Cells (5 × 106) were incubated for the indicated time at 37 °C in 1 ml of complete medium containing 100 units/ml TNF- After these treatments, cells were washed twice in phosphate-buffered
saline and lysed in 250 µl of 1 × TNE (50 mM Tris,
pH 8.0, 1% Nonidet P-40, and 2 mM EDTA), supplemented with
10 µg/ml each of the protease inhibitors leupeptin, aprotinin,
N-tosyl-L-phenylalanine chloromethyl ketone,
N-p-tosyl-L-lysine chloromethyl
ketone, and phenylmethylsulfonyl fluoride as well as the phosphatase
inhibitors sodium fluoride (50 mM) and sodium orthovanadate
(1 mM).
Cells were lysed as described above.
Specific polypeptides were then recovered by immunoprecipitation of
equivalent amounts of cellular protein, using either of the following
antibodies: anti-I Proteins were transferred to Immobilon
membranes (Millipore), and immunoblots were incubated with
anti-I For two-dimensional
analysis of cell extracts, 15 × 106 cells were lysed
with 30 µl of 1 × TNE supplemented with inhibitors as above.
Cytoplasmic extracts were reduced with 1% For
32Pi labeling, 293T cells were transiently
transfected with HA I In E29.1 T cell hybridoma, the two inhibitors I
To evaluate the discrepancy concerning the effect of
TNF-
To
determine whether one of the two variants of I We then asked whether the newly synthesized form of I Earlier studies demonstrated that treatment of cells with
the phosphatase inhibitors okadaic acid and calyculin A induced phosphorylation changes of I
It has been shown recently that two serine
residues (Ser-32 and Ser-36) spaced by a three-amino acid GLD motif and
two adjacent lysines (Lys-21 and Lys-22) play a critical role in
signal-induced proteolysis of I Using site-directed mutagenesis, Ser-19 and Ser-23 were replaced by
nonphosphorylatable alanine residues, and the resulting I We first examined the impact of Ser-19 and Ser-23 mutations on I
To test whether the mutation of Ser-19 and Ser-23 affected I Using the same type of analysis, we were able to demonstrate that the
presence of Lys-9 was not required for the signal-induced proteolysis
of I It has been reported that
signal-dependent degradation of I
To get a better
separation of I
To define the level of phosphorylation of wt
and A19 A23 I
Recent data have provided some insight into the mechanisms leading
to I The I Therefore, to get some insight into the mechanisms responsible for
these different behaviors, we carried out a systematic analysis of the
events leading to I Following treatment of E29.1 murine T cells with TNF- Treatment of form I with alkaline phosphatase results in a partial
conversion into a protein that comigrates with form II, suggesting that
form I is a hyperphosphorylated form of form II (data not shown). These
data suggest that this hyperphosphorylation might target form I for
degradation.
Based on the results obtained with I The I In contrast to I From these results we can conclude that serines 19 and 23 are critical
determinants of I Second, if I We thank Simon Whiteside, Gilles Courtois,
and Sylvie Mémet for helpful discussions; Christel Brou for
technical advice and for the plasmid pT7
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
B (nuclear
factor-
B) is neutralized in nonstimulated cells through cytoplasmic
retention by I
B inhibitors. In mammalian cells, two major forms of
I
B proteins, I
B
and I
B
, have been identified. Upon
treatment with a large variety of inducers, I
B
and I
B
are
proteolytically degraded, resulting in NF-
B translocation into the
nucleus. Recent observations suggest that phosphorylation of serines 32 and 36 and subsequent ubiquitination of lysines 21 and 22 of I
B
control its signal-induced degradation. In this study we provide
evidence that critical residues in the NH2-terminal
region of I
B
(serines 19 and 23) as well as its COOH-terminal
PEST region control I
B
proteolysis. However Lys-9, the unique
lysine residue in the NH2-terminal region of I
B
, is
not absolutely required for its degradation. We also demonstrate that
following stimulation, an underphosphorylated nondegradable form of
I
B
accumulates. Surprisingly, our data suggest that unlike
I
B
, I
B
is constitutively phosphorylated on one or two of
the critical NH2-terminal serine residues. Thus, phosphorylation of these sites is necessary for degradation but does
not necessarily constitute the signal-induced event that targets the
molecule for proteolysis.
B1 plays a
central role in the regulation of genes implicated in the immune
response and in inflammatory processes. NF-
B is composed of homo-
and heterodimeric complexes of members of the Rel/NF-
B family of
polypeptides. In vertebrates, this family comprises p50, p65 (RelA),
c-Rel, p52, and RelB.
B is cytosolic, but the nuclear translocation
of this factor can be induced by multiple stimuli that act at different
levels in the cell. Some, like TNF-
, IL-1, LPS, or antibodies
against the T cell receptor-CD3 complex, act on an extracellular
receptor, whereas others, like PMA and double-stranded RNA, activate
intracellular second messengers (for review, see Refs. 1 and 2).
B involves its association with the inhibitory ankyrin repeat-containing members of the I
B family of proteins. This family
of inhibitors is mainly represented by I
B
and I
B
(3, 4) but
also includes I
B
, Bcl-3, p105, and p100 (for review, see Ref. 5).
p105 and p100, which are also the precursors of the p50 and p52
subunits of NF-
B, function as I
B proteins through association
with p50, c-Rel or p65 (6-8). Among the different I
Bs, I
B
and
I
B
play a major role in the regulation of NF-
B. These two
proteins have been originally identified by partial purification (3,
4). A cDNA clone has been isolated which encodes a 36-kDa protein
called MAD-3, which appears to be identical to I
B
(9). The
cloning of I
B
cDNA is more recent, and this molecule has thus
been characterized less thoroughly (10).
B
, the I
B
gene is positively regulated by NF-
B
and glucocorticoids (10, 11-14). I
B
and I
B
associate with p50-p65 heterodimers and prevent the nuclear translocation of these
complexes by masking their nuclear localization sequence. These two
molecules are structurally similar as they contain multiple ankyrin
repeats and a COOH-terminal PEST domain, a sequence known to be highly
correlated to rapid protein turnover. The PEST domain of I
B
is
involved in its degradation (15-19) and in its inhibition of the DNA
binding activity of NF-
B (20).
B
becomes phosphorylated and is
subsequently degraded, allowing NF-
B to translocate into the
nucleus. The use of protease inhibitors has shed some light on the
proteases responsible for I
B
degradation (21-27). In the presence of an NF-
B inducer, proteasome inhibitors stabilize a
phosphorylated form of I
B
characterized by a slow electrophoretic mobility. The observation that this retarded form is still associated with NF-
B invalidates the former hypothesis that I
B
phosphorylation induces its dissociation from NF-
B.
B
have been identified
as two closely spaced serines at positions 32 and 36 in the
NH2-terminal part of the protein (15-19, 28). Mutation of these two serines to nonphosphorylatable residues prevents I
B
degradation, suggesting that their phosphorylation is a prerequisite for degradation. In addition, an in vitro study has also
shown that only the hyperphosphorylated form of I
B
is degraded
(29). It has been reported that degradation of I
B
is triggered by ubiquitination (30). Ubiquitination occurs primarily on two adjacent
lysines (Lys-21 and Lys-22) (31, 32).
B
is the main inhibitor of NF-
B, and the
disruption of I
B
gene by homologous recombination results in a
constitutively elevated level of nuclear NF-
B (33). In contrast to
hematopoietic cells, I
B
/
embryonic fibroblasts behave as wild
type cells because of the major role of the I
B
molecule.
or PMA induces degradation of
I
B
but not of I
B
, suggesting that these two proteins are
regulated differentially (10). The mechanisms responsible for the
differential behavior of the two I
B molecules remain unresolved. We
provide here some clues as to why these two molecules are regulated
differentially. We first show that I
B
degradation is blocked by
proteasome inhibitors, suggesting an involvement of the
ubiquitin-proteasome pathway, similar to what has been described for
I
B
. These similarities are confirmed by the identification of two
critical sites of phosphorylation (Ser-19 and Ser-23) whose mutations
decrease the rate of signal-induced degradation of I
B
. However,
I
B
contains only one lysine NH2-terminal to ankyrin repeats, and its mutation does not prevent the signal-induced degradation of the mutant protein. We also demonstrate that I
B
preexists as two electrophoretically different variants: the major, slow migrating form, is degraded following stimulation; the minor, faster migrating form accumulates. Our study also suggests that, unlike
I
B
, I
B
is phosphorylated on Ser-19 and/or Ser-23 in noninduced cells. Therefore, these results suggest that I
B
might differs from I
B
in that the critical event that targets I
B
for degradation is not the induced phosphorylation of the two conserved
serine residues located in the NH2-terminal region of the
molecule.
Cells
-mercaptoethanol.
70Z/3 derivatives expressing human I
B
(hI
B
) have been
described previously (10). 70Z/3 HA I
B
wt represents a stable
transformant expressing hemagglutinin-tagged, wild type I
B
. 70Z/3
expressing I
B
HA-tagged variants: HA A19, HA A23, HA A19 A23, HA
R9, and HA R9 A19 A23 expressed HA I
B
containing the following
point mutations: (A19, Ser-19
Ala; A23, Ser-23
Ala; R9, Lys-9
Arg). 70Z/3 HA wt
PEST and 70Z/3 HA A19 A23
PEST expressed
amino acids 1-306 of HA I
B
wt and HA I
B
A19 A23,
respectively.
B
immunoblots were probed with
anti-I
B
52008, generated against recombinant human I
B
or
with an antibody kindly provided by M. Karin. Anti-I
B
immunoblots
were probed with rabbit polyclonal anti-I
B
sera 37015 and 41276 raised against a glutathione S-transferase fusion protein
containing amino acids 258-360 of mouse I
B
or an antibody raised
against recombinant I
B
(10).
B
or its
derivatives into the plasmids pRc-CMV or pcDNA-3 (Invitrogen). Some
of the mutants were first cloned into the plasmid
pT7
globin-HA,2 and then the fragment
containing the
-globin 5
-untranslated region, the HA epitope, and
I
B
coding sequences was subcloned into pRc-CMV or pcDNA-3.
Deletion of the PEST region of I
B
was obtained by digestion with
HindIII. Mutations of I
B
were constructed by
site-directed mutagenesis using polymerase chain reaction and verified
by sequencing. All constructions were linearized with ScaI
before electroporation into 70Z/3 cells.
(Pharmingen), 15 µg/ml LPS (Sigma), 50 ng/ml PMA (Sigma), 10 µg/ml cycloheximide (Sigma), 150 nM calyculin A (Sigma), 100 µM
N-acetyl-Leu-Leu-norleucinal (calpain inhibitor I, Sigma) as
indicated in the figure legends.
B
serum 52008, affinity-purified anti-I
B
serum 41276, anti-RelA serum 1226, or mouse anti-HA 12CA5 monoclonal
antibody. Immune complexes were collected with Staphylococcus
aureus protein A (Pansorbin, Calbiochem) or protein G (Sigma),
washed three times in lysis buffer, and resolved by sodium dodecyl
sulfate (SDS)-polyacrylamide gel electrophoresis. For certain
experiments, RelA and associated proteins were eluted from specific
antibodies with an excess of the 1226 peptide, and the eluate was mixed
with 2 × sample buffer. Subsequent immunoblots were performed
using the protocol outlined below.
B
, anti-I
B
, or anti-RelA serum diluted 1/1,000 (for
ECL) or 1/200 (for 125I-protein A), as indicated in the
figure legends, and revealed with the Amersham ECL system (for direct
immunoblotting of total cell extracts) or by incubation with
125I-protein A (Amersham) (for immunoblotting of
immunoprecipitates). Immunoreactive products were detected by
autoradiography, and 125I-labeled proteins were quantitated
with a PhosphorImager (Molecular Dynamics).
-mercaptoethanol and
denatured with 0.3% SDS. For immunoprecipitations, 5 × 107 cells were lysed with 300 µl of 1 × TNE
supplemented with inhibitors and immunoprecipitated as above, except
that the beads were resuspended in 30 µl of SDS buffer (0.3% SDS,
1%
-mercaptoethanol, and 50 mM Tris, pH 8.0). Samples
were boiled for 3 min and flash frozen in liquid nitrogen. After
lyophilization, samples were resuspended in 30 µl of sample buffer
(9.95 M urea, 4% Nonidet P-40, 2% ampholytes, pH 5-7,
and 100 mM dithiothreitol) and centrifuged for 2 min. Samples were then loaded onto the isoelectrofocusing gel (pH 4-8; Millipore) and run for 20,000 Vh. The second dimension was performed as
described previously on a 10% acrylamide gel (35). Relative isoelectric points were determined by parallel migration of a carbamylated muscle creatine phosphokinase standard (BDH), and the
relative molecular weights of the proteins were determined according to
molecular weight markers applied to an adjacent slot on the same gel.
After SDS-polyacrylamide gel electrophoresis, proteins were blotted as
for the one-dimensional polyacrylamide gels.
B
wt or HA A19 A23 I
B
plasmids and
were incubated for 3 h in phosphate-free RPMI medium supplemented
with 0.8 mCi of 32Pi (carrier free; ICN)/ml and
2% dialyzed fetal calf serum. They were then washed with
phosphate-free medium and lysed in TNE buffer containing protease and
phosphatase inhibitors, as described above. I
B
polypeptides were
recovered by immunoprecipitation and resolved in 8% SDS-polyacrylamide
gels. Trypsin cleavage was performed, and phospholabeled I
B
polypeptides were resolved in Tricine-SDS-polyacrylamide gels
electrophoresis as described elsewhere (36) and detected by
autoradiography.
TNF-induced Degradation of IB
in E29.1 T Cell
Hybridoma
B
and I
B
behaved differently in response to TNF-
. TNF-
treatment resulted in rapid (15 min) and transient degradation of
I
B
(data not shown, and see Fig. 2), whereas I
B
was
degraded slowly (Fig. 1) and incompletely resynthesized
3 h after induction (lanes 7 and 9).
Increasing the time of stimulation to 6, 9, and 24 h led to the
reappearance of I
B
(lanes 10-12). This longer time
course allowed the identification of two different molecular species of
I
B
: the upper form migrates at 50 kDa and is totally degraded upon stimulation, whereas persistent activation causes the slow accumulation of a lower molecular mass product of approximately 48 kDa.
Fig. 2.
Effect of TNF-, LPS, and PMA stimulation
on I
B
degradation in various cell lines. Different cell
lines were treated with NF-
B inducers for the indicated periods of
time. Panel A, TNF-
causes the degradation of I
B
in
E29.1 T cell hybridoma. Cell lysates (20 µg of proteins) from TNF-
treated cells were prepared and subjected to parallel I
B
(right panel) or I
B
(left panel)
immunoblotting. Panel B, I
B
is not degraded upon TNF-
stimulation of EL-4 T cells. Cell lysates (20 µg of proteins) from TNF-
-treated cells were prepared as described in panel
A and subjected to parallel I
B
(right panel) or
I
B
(left panel) immunoblotting. Panel C,
LPS and PMA cause the degradation of I
B
in 70Z/3 cells. After
lysis, 20 µg of proteins for each lane was probed with anti-I
B
(left panel) or anti-I
B
(right panel) antibodies. All of the anti-I
B
blots were probed with 37015 antiserum except for the PMA experiment (panel C), which was
blotted with an anti-I
B
antibody provided by S. Ghosh. The
positions of I
B
, I
B
, and nonspecific products are shown by
an arrow on the right of each
panel.
[View Larger Version of this Image (33K GIF file)]
Fig. 1.
Western blot analysis of TNF--stimulated
E29.1 cells reveals the presence of two electrophoretically different
forms of I
B
. 20 µg of total cell extracts from E29.1 cells
exposed to TNF-
for the indicated times were separated by
SDS-polyacrylamide gel electrophoresis, transferred to Immobilon, and
probed with anti-I
B
(37015) antiserum.
[View Larger Version of this Image (10K GIF file)]
B
Is Not Degraded in All T Cell Lines in Response to TNF-
Stimulation
on I
B
degradation (10, 33), we examined the consequence
of TNF-
treatment on a second murine T cell line, EL-4 (Fig.
2B). Our results suggest that there may be some
cell specificity in I
B
degradation in response to TNF-
since
this treatment affected only I
B
in EL-4 T cells, whereas it
resulted in the loss of I
B
and I
B
in E29.1 (compare Fig. 2,
A and B). However, treatment of EL-4 T cells with
IL-1 or PMA induced a complete degradation of I
B
(data not
shown). We also confirmed the detection of a doublet of bands in the
murine preB 70Z/3 cell line in response to LPS and PMA (Fig.
2C). This doublet, noted previously in the I
B
original
cloning paper (10), represents specific forms of I
B
since it is
detected with our I
B
antibody 37015 (Fig. 2, A and
C), with an antibody kindly provided by S. Ghosh (Fig. 2C) and also by immunoblotting p65 immunoprecipitates
with anti-I
B
antibodies (Fig. 3B).
Fig. 3.
Characterization of the two
electrophoretically different forms of IB
. Panel A,
I
B
is de novo synthesized as a low molecular mass
product following stimulation. E29.1 T cell were incubated with TNF-
for the indicated times with (lanes 9-12) or without
(lanes 5-8) cycloheximide. Cycloheximide was added 30 min
prior to induction for lanes 9-12. To normalize cell treatments, cells were also treated with cycloheximide alone for the
indicated time (lanes 1-4). Total lysates were prepared as described under "Materials and Methods" and assayed by
immunoblotting with anti-I
B
(37015) (top panel) or
anti-I
B
(with an antibody provided by M. Karin) (bottom
panel). I
B
degradation cannot be totally observed in the
absence of cycloheximide as it is already resynthesized at 30 min (see
Fig. 2A). Panel B, the two forms of I
B
are
associated with NF-
B. E29.1 cells were stimulated for the indicated
time with TNF-
alone (lanes 3-6) or in the presence of
ALLN (lane 7). After immunoprecipitation with normal rabbit
serum (NRS) or 1226 anti-p65 antiserum, coprecipitated proteins were eluted from bound antibodies with an excess of the 1226 peptide. The eluates were subjected to 8% SDS-polyacrylamide gel
electrophoresis and probed with anti-p65 (1226) (top panel) or anti-I
B
(37015) (bottom panel) antisera. These two
immunoblots were derived from the same gel. The migration of p65 and
I
B
is indicated on the left.
[View Larger Version of this Image (36K GIF file)]
B
B
described above
could be derived from the other, we tested the effect of the protein
synthesis inhibitor cycloheximide on the TNF-
-induced degradation of
I
B
(Fig. 3A, top panel). To perform this
experiment, we pretreated E29.1 T cells with cycloheximide for 30 min.
Alone this treatment had no effect on the I
B
and I
B
levels
(lanes 1-4); but, followed by stimulation with TNF-
for
the indicated period (lanes 9-12), we observed a complete
loss of the low molecular mass product of I
B
, whereas the upper
form disappeared with kinetics similar to non-cycloheximide-treated
cells (compare lanes 5-8 with lanes 9-12). We
thus consider the low molecular mass variant of I
B
to be a newly
synthesized product. As expected, I
B
had a rapid turnover and was
entirely resynthesized after 60 min (Fig. 3A, bottom
panel, lanes 5-8) but did not reappear when protein
synthesis was blocked by cycloheximide (lanes 10-12).
B
was
specifically bound to NF-
B. To this end, p65-containing complexes were precipitated with anti-p65 antiserum 1226, eluted with the cognate
peptide, and analyzed by immunoblotting with anti-p65 (Fig.
3B, top panel) or anti-I
B
antibody
(bottom panel). The two I
B
variants were not present
in the immune complexes obtained using preimmune serum (lane
1, bottom panel) but were both present in p65
immunoprecipitates (lanes 2-6), indicating that both
variants were associated with p65 with a similar affinity. In contrast to the rapid disappearance of the slower migrating form of I
B
, the newly synthesized form accumulated following TNF-
stimulation. We also demonstrated that a proteasome inhibitor, ALLN, efficiently blocked degradation of I
B
and did not lead to its dissociation from NF-
B (lane 7).
B
for Proteolysis and Releases I
B
from
NF-
B
B
which contribute to its degradation (18, 24, 27, 28). To investigate the impact of phosphatase inhibitor
treatment on I
B
stability and association with NF-
B, coimmunoprecipitations were performed. E29.1 cells were treated for
different periods of time with calyculin A. NF-
B·I
B complexes were recovered by immunoprecipitations using anti-I
B
, anti-p65, or affinity-purified anti-I
B
antibodies. The abundance of
I
B
, I
B
, and p65 in these immunoprecipitates was established
by parallel immunodetection of these proteins. As predicted (Fig.
4), calyculin A caused a rapid depletion of I
B
concomitant with the disappearance of associated p65 (lanes
2-4). However, calyculin A treatment induced a progressive shift
of I
B
mobility, with almost no associated degradation. The
association of I
B
with NF-
B, monitored by its association with
p65, was abolished by calyculin A treatment (lanes 6-8),
suggesting that this modified form of I
B
is stable even when not
associated with NF-
B. Calyculin A treatment also induced the
appearance of slow migrating forms of p65 which likely correspond to
hyperphosphorylated molecules (lanes 9-12).
Fig. 4.
Calyculin A dissociates IB
from p65 but
does not induce its degradation. E29.1 T cells were stimulated for
the indicated period with calyculin A. Immunoprecipitates with
anti-I
B
, anti-I
B
, or anti-p65 were analyzed by
immunoblotting in the same conditions used in Fig. 1A. The
migration of prestained molecular size markers is shown on the
right, with the different antibodies used to blot the
corresponding piece of the membrane. The positions of p65, heavy chain
of IgG (Ig), I
B
, and I
B
are indicated on the
left.
[View Larger Version of this Image (31K GIF file)]
B
Proteolysis
B
(8, 15, 16, 19, 30-32).
Analysis of the I
B
protein sequence reveals two serines (Ser-19
and Ser-23) in a similar environment and only one lysine (Lys-9) in the
entire NH2-terminal domain of I
B
. It was thus
tempting to examine whether Ser-19 and Ser-23 as well as Lys-9 play a
similar role in signal-induced degradation of I
B
.
B
mutants were stably introduced into murine 70Z/3 pre-B cells. To
distinguish between endogenous I
B
and exogenous, mutated versions
of I
B
, the latter were tagged at the NH2 terminus
with an HA epitope that is specifically recognized by the 12CA5
monoclonal antibody (anti-HA).
B
proteolysis (Fig. 5A). After treatment with LPS,
cells were lysed, and the "tagged" mutants were specifically
immunoprecipitated with anti-HA antibody and then detected by
anti-I
B
immunoblotting. We also verified that transfected
proteins remained associated with NF-
B by immunoblotting the immune
complexes with an antibody to the p65 subunit of NF-
B. As shown
above for endogenous I
B
, we also observed two forms of the
exogenously expressed I
B
derivatives. The slower migrating
species of exogenously expressed wild type I
B
was the only
species to be degraded efficiently following a 1-h treatment of cells
with LPS (Fig. 5A, lanes 2 and 3). In contrast, the amount of the faster migrating form was strongly increased after 2 h of stimulation (lane 4). On the
other hand, mutation of either Ser-19 or Ser-23 or both completely
abolished the degradation of I
B
in response to LPS (lanes
5-16). As indicated by parallel p65 immunoblotting, these mutants
were still capable of interacting with NF-
B. We verified that the
observed lack of degradation of these mutants was not a result of
differences in the level of proteins, nor in LPS induction variability,
by directly probing the total cell extracts with anti-p65 or
anti-I
B
antiserum (data not shown).
Fig. 5.
Ser-19 and Ser-23 but not Lys-9 are necessary
for IB
proteolysis in response to LPS. 70Z/3 cells were
stably transfected with HA-tagged wt or mutated murine I
B
expression vectors. Cells were induced with LPS for the indicated
periods. Panel A, extracts from 70Z/3 cells which stably
express tagged wt and serine 19 and 23 to alanine double (A19 A23) or
single point mutants (A19 and A23) were immunoprecipitated with anti-HA
monoclonal antibody and transferred to an Immobilon membrane that was
then separated in two pieces to allow anti-p65 (1226) or anti-I
B
(37015) immunoblotting. Panel B, 70Z/3 clones expressing
tagged wt, R9 A19 A23 I
B
or R9 I
B
mutants were pretreated
for 30 min with cycloheximide before LPS stimulation. Extracts were
then immunoprecipitated with anti-HA monoclonal antibody, transferred
to an Immobilon membrane, and probed with anti-I
B
(37015)
antiserum.
[View Larger Version of this Image (27K GIF file)]
B
degradation in response to other inducers, we analyzed the effect of
IL-1 and PMA. We found that, like LPS stimulation, IL-1 and PMA caused
a complete degradation of endogenous and HA I
B
wt but not of the
Ser-19 and/or Ser-23 mutants (data not shown).
B
. To measure the effect of Lys-9 mutation on I
B
half-life upon LPS treatment, experiments using cycloheximide were
carried out (Fig. 5B). Under these conditions, we found that HA I
B
wt was completely degraded in 60 min, whereas the triple mutation of Lys-9, Ser-19, and Ser-23 prevented its degradation. Surprisingly, following quantification of the blots by PhosphorImager, we did not observe detectable changes when I
B
was mutated on Lys-9 alone (the basal levels of expression of the two molecules are
different, but compare lanes 1 and 2 with
lanes 11 and 12). These results suggest that, in
contrast to Lys-21 and Lys-22 of I
B
, Lys-9 does not seem to be an
exclusive site of ubiquitination of I
B
, but, as for I
B
, two
phosphorylatable serines are required to target I
B
for
degradation (see "Discussion").
B
PEST Region
B
requires the
carboxyl-terminal PEST sequence (16, 17, 19). To determine whether
I
B
PEST sequences are also necessary for signal-dependent proteolysis, we constructed truncated
proteins lacking amino acids 307-360 (
PEST) in the context of both
I
B
wt and I
B
A19 A23. Cell lysates were prepared and
subjected to immunoprecipitation using anti-HA antibody followed by
anti-I
B
immunoblotting (Fig. 6, A and
B, top panels). We observed that both tagged
constructs, HA wt
PEST and HA A19 A23
PEST, were stable upon LPS
stimulation (lanes 2, 3, 7, and
8). However, during the 1-h treatment with LPS, the amount
of HA wt
PEST did not increase, in contrast to HA A19 A23
PEST,
suggesting that the former construct was still partially susceptible to
signal-induced degradation. This was confirmed by pretreatment of the
cells with ALLN (Fig. 6A, top panel, compare
lane 3 with 5, and lane 8 with 10). Immunoblotting of total cell extracts with anti-p65
antibody indicated that protein loading was identical in each
lane (middle and bottom panels). To
analyze this point further, cells were pretreated with cycloheximide
for 1 h and then stimulated with LPS (Fig. 6B). Under
these conditions I
B
was not resynthesized (bottom
panel, lanes 3, 4, 7, and
8), and endogenous I
B
was degraded completely
(middle panel, lanes 3, 4,
7, and 8). This experiment revealed a small
decrease in the amount of HA wt
PEST after 1 h (upper
panel, lane 3) but no change for HA A19 A23
PEST (lanes 6-8). However, compared with the effect of
cycloheximide on HA wt (Fig. 5B), this degradation was
minor. We thus consider that the deletion of I
B
PEST sequence
largely protects I
B
from proteolysis and that this effect is
reinforced by Ser-19 and Ser-23 mutations.
Fig. 6.
The PEST sequence is required for
signal-dependent degradation of IB
. Panel
A, 70Z/3 cells stably transformed with HA-tagged wt
PEST and
A19 A23
PEST were induced with LPS for the indicated periods
following (lanes 4-5, 9-10) a 1-h pretreatment with ALLN. Top panel, HA wt
PEST and HA A19 A23
PEST
were immunoprecipitated with anti-HA monoclonal antibody, transferred
to an Immobilon membrane, and then immunoblotted with an anti-I
B
anti-serum provided by S. Ghosh. Middle panel, direct
anti-p65 (1226) immunoblot of total cell extracts. Bottom
panel, direct anti-I
B
(37015) immunoblot of total cell
extracts. The positions of tagged I
B
mutants (HA I
B
),
endogenous I
B
, and p65 are shown on the left. The
altered mobility of the HA A19 A23
PEST construct versus wt has been reproducibly observed. Panel B, 70Z/3 cells
stably transformed with HA-tagged wt
PEST and A19 A23
PEST were
pretreated with cycloheximide and stimulated with LPS for the indicated
periods. Cell extracts were analyzed by immunoprecipitation with
anti-HA monoclonal antibody, transferred to an Immobilon membrane, and then immunoblotted with an anti-I
B
antiserum provided by S. Ghosh
(top panel). Western blotting was performed on total cell extracts with anti-I
B
antibody (middle panel), the
part corresponding to the transfected HA-tagged molecule has been cut
out) or anti-I
B
antibody (with an antibody provided by M. Karin)
(bottom panel). The positions of a nonspecific band
(NS), of tagged I
B
mutants (HA I
B
), and of
endogenous I
B
and I
B
are shown on the
left.
[View Larger Version of this Image (28K GIF file)]
B
B
isoforms, total lysates from unstimulated E29.1
T cells were subjected to two-dimensional gel analysis and
immunoblotted with anti-I
B
antiserum (Fig. 7A). As for I
B
(16, 19) multiple isoforms
were detected which likely represent differentially phosphorylated
isoforms of I
B
. Upon stimulation with TNF-
, the spots
disappeared progressively (panels B and C).
Strikingly, the simultaneous presence of ALLN did not lead to the
appearance of additional spots (panel D). This result was
confirmed by two-dimensional analysis of mixed lysates from
unstimulated and TNF-
-induced cells in the presence of ALLN
(panel E). This is in contrast to the situation observed with I
B
, which, upon signaling, becomes phosphorylated on two critical serines, resulting in the appearance of additional isoforms of
higher apparent molecular mass and more acidic
migration.3 (16, 19). This intriguing
observation led us to evaluate the possibility that I
B
was
constitutively phosphorylated on Ser-19 and/or Ser-23. To this end, we
decided to use 70Z/3 cells expressing
PEST constructs in the hope
that, as for I
B
, it would simplify the overall pattern of
isoforms (19). Interestingly, both HA wt
PEST and HA A19 A23
PEST constructs still gave rise to several isoforms of different
apparent molecular mass and isoelectric points (panels F and
H), suggesting that unlike I
B
, additional phosphorylations take place in the ankyrin repeat region (see "Discussion"). In agreement with what we observed for wt I
B
(panels D and E), pretreatment of HA wt
PEST
with ALLN did not result in the appearance of additional spots in
response to LPS (panels G and I). But the most
striking observation was that the HA A19 A23
PEST isoforms present a
distribution identical to that of the HA wt
PEST isoforms, except
that they exhibit a much more basic isoelectric point (panel
J). The simplest explanation for this observation is that all of
the isoforms are constitutively phosphorylated on Ser-19 and/or Ser-23
in HA wt
PEST constructs.
Fig. 7.
Analysis of constitutive and inducible
phosphorylation of IB
by two-dimensional gel
electrophoresis. E29.1 T cells were analyzed before stimulation
(panel A) and after 30 min (panel B) and 90 min
of induction by TNF-
in the absence (panel C) or presence
(panel D) of ALLN. Cell extracts from treated and untreated cells were subjected to two-dimensional gel electrophoresis as described under "Materials and Methods." Endogenous I
B
proteins were analyzed by immunoblotting with anti-I
B
antiserum
37015. A mixing experiment (panel E) of lysates obtained
from uninduced and TNF-
-stimulated cells in the presence of ALLN
demonstrated the identical migration of I
B
isoforms observed in
these two conditions. 70Z/3 cells expressing HA wt
PEST
(panels F and G) or HA A19 A23
PEST
(panel H) were treated as indicated. Total cell extracts
were precipitated with anti-HA antibody and subjected to
two-dimensional gel electrophoresis as described under "Materials and
Methods." Immunoblots were probed with an anti-I
B
antiserum provided by S. Ghosh. Extracts analyzed in panels F and
G were mixed in panel I; extracts analyzed in
panels F and H were mixed in panel J.
For each panel, the horizontal axis represents
the isoelectric focusing dimension; the vertical axis
represents the molecular mass dimension. The acidic side of the gel is
on the right.
[View Larger Version of this Image (31K GIF file)]
B
molecules, 293T cells were transiently transfected
with HA wt or HA A19 A23 I
B
and were metabolically labeled with
32Pi. Then, phospholabeled I
B
molecules
were recovered by immunoprecipitation as described under "Materials
and Methods" (Fig. 8A). We found that wt
I
B
polypeptides were prominently phosphorylated in 293T cells, to
an extent apparently similar to that of A19 A23 I
B
. However, to
define precisely the sites of phosphorylation on these molecules,
phospholabeled I
B
molecules were subjected to peptide mapping
studies using trypsin (Fig. 8B). Trypsin digestion removes from I
B
a unique 103-amino acid fragment of approximately 13 kDa
which contains serines 19 and 23. Strikingly, we found that a 13-kDa
peptide was phosphorylated in wt I
B
molecules but was devoid of
phosphorylation in A19 A23 I
B
molecules. This is the largest
peptide generated by trypsin digestion (the next largest is 37 amino
acids).
Fig. 8.
Phosphorylation state of wt IB
and A19
A23 I
B
in 293T cells. Panel A, in vivo
32Pi labeling of nontransfected
(NT) or transiently transfected 293T cells (HA wt and HA A19
A23). Cells were labeled with 32Pi, and
I
B
molecules were recovered by immunoprecipitation with affinity-purified anti-I
B
serum 41276. Exposure, 1 h.
Panel B, trypsin peptide mapping studies. In vivo
phosphorylation of serines 19 and 23 was determined by trypsin cleavage
of 32Pi-labeled wt I
B
and A19 A23
I
B
. The migration of molecular mass markers is indicated on the
right. Exposure, 6 days.
[View Larger Version of this Image (30K GIF file)]
B
degradation (15-19, 21, 30-32, 37). The signal-induced degradation of this molecule is dependent upon the presence of an
intact COOH-terminal PEST region as well as the induced phosphorylation of serine residues 32 and 36. This phosphorylation probably targets the
molecule for ubiquitination on multiple residues, lysines 21 and 22 playing a central role (31, 32). This in turn targets the molecule for
degradation by the 26 S proteasome (30), which takes place in the
absence of dissociation between the inhibitor and NF-
B.
B
molecule was originally purified and shown to be
inactivated in vitro by dephosphorylation, in contrast to
the situation observed with I
B
(38). The recent cloning of
I
B
(10) demonstrated that the protein contained, like I
B
, a
COOH-terminal PEST region, several ankyrin repeats, and an
NH2-terminal region with two serines (at positions 19 and
23) in an environment similar to I
B
. A unique lysine residue was
present in this region, at position 9. However, some differences could
be observed between the two molecules following stimulation. First,
although PMA, TNF, LPS, and IL-1 all induce degradation of I
B
,
only LPS and IL-1 induced degradation of I
B
in 70Z/3 cells (10).
Second, the I
B
molecule is resynthesized rapidly following
degradation, partly because the promoter of the I
B
gene is
positively regulated by NF-
B. On the contrary, the I
B
molecule
is not resynthesized before the stimulus has ceased. This suggests that
the promoter of the I
B
gene does not respond to NF-
B. However,
a more recent study seems to indicate that in different cell types the
I
B
molecule is also degraded in response to TNF-
(33).
B
degradation. In contrast to previous work
(10), we found that I
B
could be degraded in response to TNF-
and PMA in various murine cell lines (Fig. 2). However, the EL-4 T cell
line shows a different susceptibility to TNF-
since only I
B
is
degraded. This observation raised some important questions concerning
the signaling pathway leading to the loss of I
B
and I
B
. If
the two inhibitors were targeted for proteolysis by the same pathway,
they would be affected equally in response to a given stimulus. It is
however possible that the weakness of the signal via the TNF-
receptor in this cell line could account for the phenomenon observed,
since I
B
can indeed be degraded by other stimuli (IL-1 and PMA)
(data not shown).
, two forms of
I
B
become detectable: a major upper form (form I), which is
degraded progressively; and a minor lower form (form II), which
accumulates progressively and appears to be resistant to degradation in
response to TNF-
stimulation. The same kinetics of degradation is
observed during treatment of 70Z/3 cells by LPS or PMA. The two forms
of I
B
are both associated with p65. Using cycloheximide, we
demonstrated that form II is not derived from form I but requires
ongoing protein synthesis to accumulate (Fig. 3A).
B
, we asked whether serines
19 and 23 were the sites of phosphorylation responsible for I
B
proteolysis. In accordance with this model, we observed that mutation
of either of these two serines into an alanine abolished I
B
degradation (Fig. 5). These results are consistent with a recent report
(39). However, the A19 A23 I
B
mutant still exists as form I and
form II, suggesting that form II is not derived from form I by
dephosphorylation on Ser-19 and/or Ser-23 (Fig. 5, A and
B).
B
PEST sequence is constitutively phosphorylated by a highly
ubiquitous conserved kinase, casein kinase II. I
B
possesses several consensus sites for casein kinase II, and mutation of one of
these sites increased the I
B
half-life (40). It has thus been
hypothesized that the role of basal phosphorylation is to allow
degradation of excess free I
B
by reducing its half-life. This
PEST sequence also plays an important role in
signal-dependent degradation of I
B
(16, 17, 19). As
reported for I
B
, we also find that I
B
-inducible proteolysis
requires the PEST sequence present in its COOH-terminal region.
However, deletion of the PEST sequence results in the disappearance of
form II (Fig. 6). This suggests that form II differs from form I in its
level of phosphorylation in the PEST region.
B
, mutation of lysine at position 9 inhibited
weakly, if at all, I
B
degradation, indicating that if this is a
site of ubiquitination, its mutation does not prevent ubiquitination at
other sites (Fig. 5). Furthermore, no upshift of the molecule could be
observed following stimulation in the presence of ALLN, a proteasome
inhibitor that prevents degradation of I
B
and I
B
and allows
accumulation of a hyperphosphorylated form of I
B
. However this
lack of upshift does not necessary imply a lack of hyperphosphorylation
of the molecule, as hyperphosphorylated murine I
B
migrates like
its hypophosphorylated counterpart and is identified only by
two-dimensional gel electrophoresis3 (19). To reveal
hyperphosphorylated forms of I
B
, we performed a two-dimensional
gel analysis following various treatments of 70Z/3 cells. We observed
that I
B
preexists as multiple isoforms that most likely
correspond, as for I
B
, to differentially phosphorylated forms.
Interestingly, the isoforms observed in the presence of LPS and ALLN
coincide exactly with those present in untreated cells, suggesting that
their net charge does not change following stimulation.
B
degradation but that their phosphorylation status does not seem to change following activation. One intriguing possibility is that they could be constitutively phosphorylated in
untreated cells. To test this hypothesis, we compared HA wt
PEST and
HA A19 A23
PEST proteins, as the number of isoforms should be
reduced following deletion of the PEST region. As hypothesized, we
observe that mutation of serines 19 and 23 to alanines shifted all
isoforms toward more basic pIs. Since these serine to alanines substitutions per se do not result in a modification of
calculated pI, the observed difference in pI between the two proteins
(0.1 pH unit) suggests that Ser-19 and/or Ser-23 is constitutively phosphorylated. Consequently, we found that the peptide containing serines 19 and 23 was constitutively phosphorylated in wt I
B
molecules but not when serine 19 and 23 were mutated (Fig. 8). However
this result has been obtained in transiently transfected cells as we
never obtained enough radioactivity from 32P-labeled cells
to carry out the analysis of I
B
-derived phosphopeptides. The
intriguing observation that Ser-19 and/or Ser-23 is constitutively phosphorylated is supported by the the fact that following treatment with calyculin A, a potent inhibitor of serine/threonine phosphatases, I
B
is degraded quickly, whereas the I
B
molecule is not;
but, surprisingly, it dissociates from p65 (Fig. 4). Although this situation is not very physiological, since treatment with calyculin A
induces multiple nonspecific phosphorylations (see the multiple bands
observed for p65 in Fig. 4, lanes 10-12), it tells us that hyperphosphorylation is unable to induce I
B
degradation, thus supporting our hypothesis that Ser-19 and/or Ser-23 is constitutively phosphorylated and that this is not sufficient to trigger I
B
proteolysis. In this context, several speculations can be made concerning the signaling pathways leading to the inducible degradation of I
B
and I
B
. First, if the critical serine residues of
I
B
and I
B
are phosphorylated by the same kinase, then there
should exist a mechanism maintaining serines 32 and 36 of I
B
unphosphorylated in uninduced cells. Since degradation of I
B
is
induced by phosphatase inhibitors, it might be that a constitutive
phosphatase, which is inactivated upon induction, is involved but does
not recognize I
B
. Alternatively, serines 32 and 36 of I
B
might be masked in resting cells and thus inaccessible to the kinase
activity. Finally, the kinase might be constitutively associated with
I
B
but recruited to I
B
upon NF-
B activation.
Alternatively, different kinases may phosphorylate the two I
B
molecules.
B
is constitutively phosphorylated on Ser-19 and/or
Ser-23, the nature of the signal responsible for targeting I
B
for
degradation remains to be identified. Because we have been unable to
detect any inducible modification of I
B
mobility after
two-dimensional electrophoresis, we think it is unlikely that other
I
B
-specific phosphorylation events are involved. Alternatively,
it might be that degradation per se is induced. For example,
enzymes responsible for the ubiquitination of I
B
could be
activated following induction. A possible candidate would be the
specific E3 involved. Indeed it has been demonstrated that in the case of E6-AP, activity can be regulated (41, 42). A
change in activity of the 26 S proteasome following induction could
also constitute the inducible signal, as has been shown in the case of
stimulation by interferon-
(43).
*
This research was sponsored in part by grants from
l'Association pour la Recherche sur le Cancer, l'Agence Nationale de
Recherche contre le SIDA, INSERM, and the ligue Nationale
Française contre le Cancer.The costs of publication of this
article were defrayed in part by the
payment of page charges. The article
must therefore be hereby marked
"advertisement" in
accordance with 18 U.S.C. Section
1734 solely to indicate this fact.
Recipient of long term fellowship from ANRS.
¶
To whom correspondence should be addressed: Unité de
Biologie Moléculaire de l'Expression Génique, URA 1149 CNRS, Institut Pasteur, 25 Rue du Dr. Roux, 75724 Paris Cedex 15, France. Tel.: 33-1-4568-8553; Fax: 33-1-4061-3040; E-mail:
aisrael{at}pasteur.fr.
1
The abbreviations used are: NF-B, nuclear
factor-
B; TNF, tumor necrosis factor; IL, interleukin; LPS,
lipopolysaccharide; PMA, phorbol 12-myristate 13-acetate; HA,
hemagglutinin; wt, wild type; Tricine,
N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine; ALLN,
N-acetyl-Leu-Leu-norleucinal.
2
C. Brou, unpublished data.
3
S. T. Whiteside, C. Laurent-winter, and A. Israël, unpublished observations.
globin-HA; Michael Karin,
Nancy Rice, and Sankar Ghosh for the antisera; and Paolo Truffa-Bachi
and Geneviève Milon for the E29.1 and EL-4 T cells.
©1997 by The American Society for Biochemistry and Molecular Biology, Inc.