Proc. Natl. Acad. Sci. USA
Vol. 94, pp. 3633–3638, April 1997
Biochemistry
AP-1 transcriptional activity is regulated by a direct association
between thioredoxin and Ref-1
(redox regulationydisulfide cross-linkingynuclear translocationymammalian two-hybrid assay)
KIICHI HIROTA*†, MINORU MATSUI*‡, SATOSHI IWATA*, AKIRA NISHIYAMA*, KENJIRO MORI†, AND JUNJI YODOI*§
†Department of Anesthesia, Kyoto University Hospital, Kyoto 606–01, Japan; and *Department of Biological Responses, Institute for Virus Research, Kyoto
University, Kyoto 606–01, Japan
Communicated by JoAnne Stubbe, Massachusetts Institute of Technology, Cambridge, MA, February 4, 1997 (received for review July 24, 1996)
ABSTRACT Thioredoxin (TRX) is a pleiotropic cellular
factor that has thiol-mediated redox activity and is important
in regulation of cellular processes, including proliferation,
apoptosis, and gene expression. The activity of several tran-
scription factors is posttranslationally altered by redox mod-
ification(s) of specific cysteine residue(s). One such factor is
nuclear factor (NF)-kB, whose DNA-binding activity is mark-
edly augmented by TRX treatment in vitro. Similarly, the
DNA-binding activity of activator protein 1 (AP-1) is modified
by a DNA repair enzyme, redox factor 1 (Ref-1), which is
identical to a DNA repair enzyme, AP endonuclease. Ref-1
activity is in turn modulated by various redox-active com-
pounds, including TRX. We here report the molecular cascade
of redox regulation of AP-1 mediated by TRX and Ref-1.
Phorbol 12-myristate 13 acetate efficiently translocated TRX
into the HeLa cell nucleus where Ref-1 preexists. This process
seems to be essential for AP-1 activation by redox modification
because co-overexpression of TRX and Ref-1 in COS-7 cells
potentiated AP-1 activity only after TRX was transported into
the nucleus by phorbol 12-myristate 13 acetate treatment. To
prove the direct active site-mediated association between TRX
and Ref-1, we generated a series of substitution-mutant cys-
teine residues of TRX. In both an in vitro diamide-induced
cross-linking study and an in vivo mammalian two-hybrid
assay we proved that TRX can associate directly with Ref-1 in
the nucleus; also, we demonstrated the requirement of cysteine
residues in the TRX catalytic center for the potentiation of
AP-1 activity. This report presents an example of a cascade in
cellular redox regulation.
Increasing evidence has indicated that cellular redox status
modulates various aspects of cellular events, including prolif-
eration and apoptosis (1). TRX is a small, ubiquitous protein
with two redox-active half-cystine residues, -Cys-Gly-Pro-Cys-,
in an active center (2–5). TRX also is known as adult T-cell
leukemia-derived factor that is involved in human T lympho-
tropic virus type I leukemogenesis (6, 7). TRX exists either in
a reduced or oxidized form and participates in redox reactions
through the reversible oxidation of this active center dithiol.
TRX serves functions inside (8–10) and outside (11–13) the
cell. One of its intracellular functions is the facilitation of
protein–nucleic acid interactions (3). In vitro and in vivo
experiments showed that TRX augmented the DNA-binding
and transcriptional activities of the p50 subunit of nuclear
factor (NF)-kB by reducing Cys-62 in its DNA-binding loop (8,
9). Recently, direct physical association of an active-site mu-
tant of TRX and an oligopeptide from NF-kB p50 was
demonstrated by a nuclear magnetic resonance (NMR) study
in vitro (14); in vivo direct association of TRX–NF-kB remains
to be demonstrated. Redox regulation of Jun and Fos mole-
cules has also been implicated. Various antioxidants strongly
activate the DNA-binding and transactivation abilities of ac-
tivator protein 1 (AP-1) complex (1, 15, 16). Interestingly,
TRX enhances the DNA-binding activity of Jun and Fos, and
this process requires other molecules such as a novel protein
called redox factor 1 (Ref-1) (17, 18). Ref-1 was identified in
a cell-free system as one of the factors restoring AP-1-DNA
binding and was found to be identical to an AP endonuclease
(17, 19). There has been no report describing whether TRX
and Ref-1 directly associate or require any intervening fac-
tor(s) for their interaction. In this paper we studied the
association of TRX and Ref-1 through the use of an in vitro
cross-linking and a mammalian two-hybrid system, with vari-
ous mutants of TRX, to investigate the interaction of TRX on
AP-1-mediated transcription.
MATERIALS AND METHODS
Cells and Cell Culture.HeLa and COS-7 cells were cultured
in DMEM (GIBCOyBRL) supplemented with 10% fetal calf
serum and antibiotics at 95% humidityy5% CO2 in air at 378C.
Reagents. Phorbol 12-myristate 13-acetate (PMA) was pur-
chased from Sigma. Anti-human TRX mAb, 11-mAb, whose
epitope has not been determined (10), was produced and
provided by Fujirebio (Tokyo). Anti-human Ref-1 rabbit
polyclonal antibody (C-20), which recognizes the epitope
corresponding to amino acids 299–318, and anti-c-Fos mAb
(6–2H-2F), which recognizes the epitope corresponding to the
leucine zipper of c-Fos, were purchased from Santa Cruz
Biotechnology.
Mutagenesis of TRX and Transfection. Mutagenesis of
human TRX was performed by a PCR-based technique (20),
and authenticity of the sequences was verified by DNA se-
quencing (Fig. 1). The expression plasmids were introduced
into the cells by LipofectAMINE reagent (GIBCOyBRL)
according to themanufacturer’s instructions, using Opti-MEM
(GIBCOyBRL).
Luciferase Assay Using -73y163 Col (Human Collagenase
I)-LUC. The 273y163 Col-LUC was constructed with pGL2-
Basic (Promega) and a fragment from 273y163 Col in
pBLCAT3 (21). HeLa cells were plated in 6-well plates at a
density of 4 3 105 cells per well. The expression plasmids were
introduced into the cells by use of LipofectAMINE reagent. In
each transfection, 1mg of the pCDSRa-TRXwild or -TRXC32S/C35S
expression plasmid (6) andyor 1 mg of the pRcyCMV-Ref-1
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
Copyright q 1997 by THE NATIONAL ACADEMY OF SCIENCES OF THE USA
0027-8424y97y943633-c$2.00y0
PNAS is available online at http:yywww.pnas.org.
Abbreviations: TRX, thioredoxin; NF, nuclear factor; AP-1, activator
protein-1; Ref-1, redox factor-1; PMA, phorbol 12-myristate 13-
acetate; 6xHis, hexahistidine; DBD, DNA-binding domain.
‡Present address: Department of Biomedical Genetics, Faculty of
Pharmaceutical Sciences, University of Tokyo, Tokyo 113, Japan.
§To whom reprint requests should be addressed at: Institute for Virus
Research, Kyoto University, 53, Shogoin Kawahara-Cho, Sakyo-Ku,
Kyoto 606-01, Japan. e-mail: yodoi@virus.kyoto-u.ac.jp.
3633

expression vector with 1 mg of 273y163 Col-LUC were used.
The total amount of DNA was adjusted to 3 mg with pRcyCMV
(Invitrogen). After incubation of 16 hr, PMA (30 ngzml21) was
added and the cells were kept for 24 hr. The cells were harvested
after 24 hr and luciferase activity was measured by use of a
commercial assay system (Promega) with a luminometer. The
relative fold induction of luciferase activity was calculated.
Expression of Recombinant Proteins in Bacteria. Various
TRX recombinant proteins were expressed in Escherichia coli
as the N-terminal fused form with hexahistidine (6xHis) tag
using the pQE30 expression plasmids (Qiagen, Chatsworth,
CA). E. coli strain M15 [pRep4] was transformed with each
pQE–TRX expression vector. E. coli were grown at 378C with
vigorous shaking and were treated for 3 hr with 1mM isopropyl
b-D-thiogalactoside, and the pelleted cells were lysed in PBS
containing 1 mgzml21 lysozyme, 10 mM 2-mercaptoethanol, 1
mM phenylmethylsulfonyl f luoride, 50 mM 7-amino-1-chloro-
3-tosylamido-2-heptanone, and 0.8 mM imidazol and soni-
cated, and was clarified by centrifugation. The supernatant
containing 6xHis–TRX was loaded onto a Ni21-NTA (nitrilo-
tri-acetic acid) agarose column. The column was washed, and
the fusion protein was eluted by PBS containing 80 mM
imidazol. The eluted protein was dialyzed against PBS (pH 7.3)
containing 2 mM DTT. Human Ref-1 was expressed based on
the pDS56ref-1 (a gift form S. Xanthoudakis) (18) as a
6xHis-tagged protein by use of the same protocol as that used
for the expression of TRX, except that the Ref-1 eluate was
dialyzed against a storage buffer [50 mM sodium phosphate,
pH 7.3y50 mM NaCly5 mM MgCl2y1 mM EDTA, 5% (voly
vol) glycerol] containing 1 mM DTT. Truncated rat 6xHis-
tagged recombinant c-Fos proteins, wbFos and its cysteine-
to-serine mutant, wbFos (C1-S), were expressed and purified
as described, based on pDS56wbFos and pDS56wbFos (gifts
form T. Curran) (C1-S) (18).
In Vitro Disulfide Cross-linking Interaction Assay. In the
cross-linking assay, 6xHis–TRXs (200 ng), 6xHis–Ref-1 (200
ng) and 6xHis–Fos (200 ng) were incubated for 30 min at room
temperature with 1 mM DTT. Then they were mixed and
incubated in the presence of 10 mM diamide for 30 min at
room temperature in 20 ml of PBS (pH 7.3). The reaction
mixture was denatured for 5 min at 908C in dissociation buffer
with or without 2-mercaptoethanol (1%). Each mixture was
applied to a 15% SDSyPAGE and electrophoresed (18). After
electroblotting, polyvinylidene difluoride membranes (Milli-
pore) were blocked, then incubated with primary antibodies
followed by incubation with horseradish peroxidase-
conjugated anti-IgG antibody (Amersham). The antigens on
the membrane were visualized with the ECL Western blot
detection kit (Amersham).
Indirect Immunofluorescence Cell Staining. Cells were
fixed with 3.7% paraformaldehyde in PBS containing 10%
fetal calf serum for 20 min at room temperature followed by
permeabilization for 10 min using 0.2% (wtyvol) Triton X-100
in PBS. After incubation (1 hr) with primary antibodies (for
TRX, monoclonal 11-mAb; for Ref-1, polyclonal antibody,
C-20) slides were incubated (1 hr) with either dye-labeled
anti-mouse or anti-rabbit IgG. Slides with stained cells were
mounted in 90% glycerol with 1 mgzml21 p-phenylenediamine.
Cells were examined using a confocal microscopy, MRC 600
(Bio-Rad).
Mammalian Two-Hybrid Assay (22). A complementary
DNA of TRX, its mutants, or Ref-1 were fused in frame to
pCMX-GAL4 (23) or pCMX-VP16 (24) (gifts from K. Ume-
sono, Nara Institute of Science and Technology). The COS-7
cells were plated in 6-well plates at a density of 2 3 105 cells
per well. The expression plasmids were introduced into the
cells by LipofectAMINE reagent. In each transfection, 1 mg of
GAL4-fused plasmid and 1mg of viral protein 16 (VP16)-fused
plasmid were used together with 1 mg of reporter construct,
ptk-GALpx3-LUC (a gift from K. Umesono, Nara Institute of
Science and Technology) made from pUC8 by inserting three
copies of Gal4-binding site and a minimal thymidine-kinase
promoter-luciferase expression cassette (this reporter plasmid
has no AP-1-binding site which may influence the expression
of luciferase). As an internal control, 1 mg of b-galactosidase
expression plasmid was used. The cells were harvested after 30
hr, and luciferase activity was determined. The relative fold
induction of luciferase activity was calculated by normalizing
to the b-galactosidase activity.
RESULTS
Transient Expression of TRX andyor Ref-1 Activate AP-1 in
PMA-Stimulated HeLa Cells. We examined whether overex-
pressed TRX andyor Ref-1 resulted in transcriptional activa-
tion of AP-1 in vivo. Thus we assayed the activity of the human
collagenase I promoter, which contains an AP-1-binding site,
by determining the luciferase activity in HeLa cells (Fig. 2).
Transient overexpression of TRX andyor Ref-1 had only a
FIG. 1. Schematic representation of mutant constructs of TRX
used in this experiment.
FIG. 2. Effect of transient expression of TRX andyor Ref-1 on
AP-1 transactivation in HeLa cells with or without PMA treatment. In
each transfection, 1 mg of the pCDSRa-TRXwild (lanes 2 and 5) or
-TRXC32S/C35S (lanes 3 and 6) expression plasmid andyor 1 mg of the
pRcyCMV–Ref-1 (lanes 1–3) expression vector with 1 mg of273y163
Col-LUC were used. The total amount of DNA was adjusted to 3 mg
with pRcyCMV (Invitrogen). After incubation for 16 hr, PMA (30
ngzml21) was added and the cells were kept for 24 hr. The cells were
harvested after 40 hr, and luciferase activity was determined. The
results are the means 6 SE of three experiments (each in duplicate)
and presented as fold increases in luciferase activity over the baseline
seen in lane 4 without PMA treatment.
3634 Biochemistry: Hirota et al. Proc. Natl. Acad. Sci. USA 94 (1997)

marginal effect for induction of luciferase activity. However
when treated by PMA, the luciferase activity increased about
4-fold (lane 2) over the mock transfection (lane 4). Expression
of either wild-type TRX (lane 5) or Ref-1 (lane 1) alone
resulted in less than 1.5-fold or 2-fold activation, respectively.
Expression of TRXC32S/C35S, which lacked reducing activity,
caused only marginal induction (lane 6). Thus, overexpressed
TRX and Ref-1 seemed to potentiate the AP-1 transcriptional
activity.
PMA Treatment Translocated TRX from the Cytoplasm
into the Nucleus in HeLa Cells. We next examined the
intracellular locations of TRX and Ref-1. We analyzed HeLa
cells without or with PMA treatment (to induce transcriptional
activation of AP-1) (15, 21), by an indirect immunofluores-
cence method, using antibodies raised against TRX or Ref-1.
As reported elsewhere, Ref-1 was localized in the nucleus and
remained there after PMA stimulation (Fig. 3D) (17, 25). In
contrast, TRX was located predominantly in the cytoplasm
before PMA stimulation (Fig. 3A) but was translocated into
the nucleus after 1 hr of PMA treatment (Fig. 3B); this process
was almost complete in 24 hr (Fig. 3C).
TRX Associates with Ref-1 through Cysteines of Catalytic
Center in Vitro. As suggested by Holmgren (3), an oxido-
reductase like TRX should form an intermediate through
disulfide linkage. Therefore, we reasoned that it would be
possible to trap the transient physical association using cross-
linking reagent(s) such as diamide, which converts free sulf-
hydryls to disulfides by cysteine oxidation (26, 27). To deter-
mine which of the five cysteines in TRX is responsible for
intermolecular disulfide bond formation, three TRX mutants,
which contained cysteines-to-serines or cysteine-to-alanine
substitution(s), were purified as the 6xHis-tagged form. A
series of experiments was performed with the 6xHis-tagged
Ref-1 and the various mutants of TRX (Fig. 1). It is known that
TRXC62S/C69S/C73S retains its reducing activity (28, 29), while
substitution mutants (TRXC35A, TRXC32S/C35S) at either
Cys-32 or Cys-35 lose their in vitro reducing activity (3). In spite
of the loss of reducing activity, TRXC35A is able to involve the
covalent mixed disulfide intermediate formation between its
Cys-32 and protein substrate (3, 30, 31) .
Under reducing conditions, the wild-type TRX and all the
TRX mutants (TRXC32S/C35S, TRXC35A, and TRXC62S/C69S/C73S)
migrated as a single 13-kDa band in SDSyPAGE. Treatment of
TRXwild or TRXC32S/C35S with diamide generatedmultiple bands,
suggesting oligomer formation. In the case of TRXC62S/C69S/C73S,
however, we detected only a monomeric form after diamide
treatment (data not shown). Ref-1 has seven cysteines, and it
moved as a monomer under reducing conditions (Fig. 4A, lanes
1, 3, 5, and 7). In oxidizing conditions, an additional band,
migrating faster than the reduced form, was observed; this
probably was a different monomeric form of Ref-1, with intramo-
lecular disulfide bond (Fig. 4A, lanes 2, 4, 6, and 8) (18).We found
no additional bands of higher molecular size, suggesting that
Ref-1 did not form cysteine-mediated multimers under the
thiol-targeted oxidizing conditions. Next, wild-type or TRX mu-
tants were incubated with Ref-1, and then their migration pat-
terns were analyzed. Both antibodies against Ref-1 (Fig. 4A) and
TRX(Fig. 4B) recognized the same extra bandsmigrating around
50 kDa in lanes of TRXwild, TRXC35A, andTRXC62S/C69S/C73S (Fig.
4A, lanes 2, 4, and 8; Fig. 4B, lane 4) under oxidizing conditions,
but not of TRXC32S/C35S species (Fig. 4A, lane 6). Only TRXC62S/
C69S/C73S was presented in Fig. 4 B andC because the other TRXs
FIG. 3. Effect of PMA treatment on subcellular distribution of TRX and Ref-1 in HeLa cells. Indirect immunofluorescence labelling, using
anti-TRX antibody (A–C) or anti-Ref-1 antibody (D). (A) Before PMA treatment. (B) After 1-hr treatment. (C and D) After 24-hr treatment.
Biochemistry: Hirota et al. Proc. Natl. Acad. Sci. USA 94 (1997) 3635

(wild, C35A, C32SyC35S) form multimers by themselves, which
are difficult to discriminate from heterocomplexes.
To exclude the possibility that diamide could cross-link any
cysteine-containing proteins nonspecifically, c-Fos–TRX inter-
action was assayed in the same condition. c-Fos, which also has
a well-conserved cysteine in its DNA-binding domain (DBD),
is one of the redox-sensitive transcription factors and is known
to be controlled by Ref-1. Fos polypeptides, wbFos and wbFos
(C1-S), were incubated with or without TRXC62S/C69S/C73S. As
shown in Fig. 4C, we could not detect covalent dimers either
between wbFos themselves or with TRXC62S/C69S/C73S.
These results showed that TRX and Ref-1 preferentially
formed a covalent heterodimer in vitro and further suggested
that this association involved Cys-32 in the active center of
TRX.
Demonstration of a Physical Association Between TRX and
Ref-1 by the Mammalian Two-Hybrid Assay in COS-7 Cells.
We investigated if TRX could associate with Ref-1 in the
nucleus using the two-hybrid system with cultured mammalian
cells (22). A complementary DNA of TRX or Ref-1 was
subcloned downstream of the DBD of GAL4 (23) or the VP16
transactivation domain (24). These fusion plasmids were then
coexpressed in COS-7 cells with a plasmid containing the
luciferase gene as a reporter.
As shown in Fig. 5A, when a DBD–TRX or DBD–Ref-1 was
expressed in COS-7 cells with VP16, there was no luciferase
activity (lanes 1 and 3). When VP16–TRX or VP16–Ref-1 was
expressed with DBD, only marginal inductions were detected
(lanes 2 and 4). In contrast, coexpression of the wild-type TRX
and Ref-1 fusion constructs induced the luciferase activity
more than 10-fold (lanes 5 and 6) over the basal level (lane 7).
Next, we used TRX mutants to identify the cysteine residue(s)
responsible for this interaction (Fig. 5B). A triple mutation of
the noncatalytic cysteines (Cys-62, Cys-69, and Cys-73) did not
diminish the luciferase activity (lane 5), suggesting that these
cysteines are dispensable for the activity of TRX to interact
with Ref-1. Compatible with our in vitro experiments, signif-
icant luciferase induction was observed by coexpression of
TRXC35A and Ref-1 fusion constructs (lane 4). On the other
hand, coexpression of TRXC32S/C35S and Ref-1 fusion con-
structs failed to induce luciferase activity (lane 3). These in
vitro and in vivo data showed that the direct physical associa-
tion between TRX and Ref-1 was mediated through the active
center of TRX. In addition to substitution mutants, we con-
structed a truncated TRX, TRXD87 (Fig. 1) by deleting the
C-terminal domain from amino acid 88. TRXD87 has all its five
cysteines preserved but has lost its dithiol reducing activity.
Coexpression of TRXD87 and Ref-1 fusion constructs showed
no significant induction of luciferase activity (lane 6).
DISCUSSION
Redox regulation of transcription factors is an interesting and
important issue. The result of in vitro gel shift assays has
FIG. 4. Evidence that TRX associated with Ref-1 through the redox-active cysteines in vitro. Various TRXmutants containing cysteine-to-serine
or cysteine-to-alanine substitutions were cross-linked with Ref-1 or wbFos by diamide. After incubation, the complexes were resolved by
electrophoresis under reducing or oxidizing conditions and detected by antibodies. (A) Western blot analysis of various TRX and Ref-1 complex
by use of anti-Ref-1 antibody. (B) Western blot analysis of TRXC32S/C69S/C73S and Ref-1 complex by use of anti-TRX antibody. (C) Western blot
analysis of TRXC32S/C69S/C73 and wbFos by use of anti-c-Fos antibody or anti-TRX antibody.
3636 Biochemistry: Hirota et al. Proc. Natl. Acad. Sci. USA 94 (1997)

established that theDNA-binding activity of a growing number
of transcription factors, including AP-1, NF-kB, Myb, and Ets,
is regulated by a thiol-redox control mechanism (8, 32–35).
Although some chemical agents, such as 2-mercaptoethanol
and DTT, can activate DNA binding of transcription factors,
the issue of which endogenous molecules act as physiological
redox regulators is still to be clarified. In the case of NF-kB,
TRX is one of the candidate molecules: Qin et al. (30)
demonstrated direct physical association between TRX and an
oligopeptide from the DNA-binding loop of NF-kB p50. That
and previous reports strongly suggest that NF-kB is a target
molecule of redox regulation by TRX (8, 9, 36).
As a representative model of redox regulation of transcrip-
tion factors, the molecular mechanism of AP-1 redox regula-
tion has been intensively studied in vivo. Xanthoudakis et al.
(19) showed that AP-1 was not a direct substrate of TRX.
Because the amino acid sequence surrounding the cysteine is
entirely different from that of NF-kB (37, 38), not surprisingly
TRX alone is unable to activate AP-1 DNA binding by cysteine
reduction in vitro. Moreover, in the case of AP-1, interestingly
redox regulation of AP-1 involves a conserved cysteine residue
in the DBD (1, 32). Substitution of the cysteine with a serine
results in a gain-of-function phenotype (37); this substitution
occurs spontaneously during the generation of the v-jun on-
cogene andmay contribute to v-jun’s oncogenic properties (38,
39). Analysis of mutated fos genes in cell transformation assays
suggests that redox regulation also affects Fos activity in vivo
via these conserved cysteines. These reports show that mod-
ification of redox-sensitive cysteines of the AP-1 molecules
may control its transcriptional activity and regulate subsequent
gene inductions. These types of substitutions in NF-kB bring
complete loss of function (8). Thus, redox regulation of AP-1
seems necessary for appropriate gene expression.
Ref-1 was cloned as a molecule that stimulated DNA
binding activity of AP-1 (19). However, no report has de-
scribed the electron donors for Ref-1 molecules. In this paper,
we have shown that TRX is one of the endogenous redox
molecules and regulates the AP-1 transcriptional activity
through its direct association with Ref-1. In other words, TRX
is one of the hydrogen donors of Ref-1. Indeed, Qin et al. (40)
solved the solution structure of an oligopeptide comprising
amino acid residues 59–71 of Ref-1 and TRX complex, thus
demonstrating the direct association of TRX and the partial
peptide of Ref-1. Our in vitro data, using whole recombinant
molecules, and in vivo experimental results show that Cys-32
and Cys-35 of TRX, which constitute the catalytic center, are
involved in the association. Site-directed mutagenesis studies
showed that two cysteines of the redox domain of Ref-1, Cys-63
and Cys-95, are redox sensitive (41, 42) and can be targets of
TRX.
As a way of transcriptional control, the subcellular locations
of some transcription factors are strictly regulated. One of the
best examples is a nuclear translocation of NF-kB, which has
a nuclear localization signal sequence (NLS) under activation
stimuli (43, 44). In contrast, TRX has no authentic NLS.
Although the minute translocational mechanism is to be
investigated in a further study, we have observed that UV
irradiation, which also potentiates the AP-1 activity, translo-
cates TRX into the nucleus in the HSC-1 keratinocyte cell line
and HeLa cells (45). As shown in Fig. 2, overexpression of
TRX and Ref-1 alone was not sufficient to induce significant
luciferase expression, and this result suggests that nuclear
translocation of TRX is indispensable for potentiating the
AP-1 transcriptional activity. These data collectively suggest
that PMA translocates TRX into the nucleus, allowing its
interaction with Ref-1, and that this association promotes the
direct activation of AP-1 by Ref-1. Recently, nuclear translo-
cation of mitogen-activating protein kinases on stimulation has
been implicated (46, 47). Interestingly, redox-acting molecules
as well as kinases regulate transcription activity in the nucleus.
We failed to immunoprecipitate Ref-1 with the anti-TRX
antibody from HeLa cells. If TRX is a hydrogen donor for
Ref-1, their association would be transient and too weak to be
detectable by standard immunoprecipitation methods. To get
higher sensitivity for detection of an association between TRX
and Ref-1, we used a diamide-induce cross-linking assay and
the in vivo two-hybrid assay. Considering that diamide is a
potent sulfhydryl-oxidizing agent, it should be able to cross-
link various proteins nonspecifically, depending on the assay
conditions. However, this is not the case in our experiments
because c-Fos and TRXC62S/C69S/C73S did not form such a
complex in the same condition. The mammalian two-hybrid
assay has been used to study various protein–protein interac-
tions, and it is especially useful in confirming interactions
FIG. 5. Demonstration of a physical association between TRX and
Ref-1 by mammalian two-hybrid assay in COS-7 cells. (A) An asso-
ciation between wild-type TRX and Ref-1 was monitored by a
luciferase assay. (B) An association between TRX mutants and Ref-1
was monitored by a luciferase assay. The relative fold induction of
luciferase activity was calculated by normalizing to the b-galactosidase
activity. The results are the means 6 SE of three experiments (each
in duplicate) and presented as fold increases in luciferase activity over
the baseline seen in lane 1.
Biochemistry: Hirota et al. Proc. Natl. Acad. Sci. USA 94 (1997) 3637

identified by two-hybrid library screens in yeast. This assay can
also be used in combination with deletional or site-directed
mutagenesis to rapidly map the amino acids responsible for the
interaction between two proteins. In the two-hybrid assay,
reporter gene expressions imply the association of two fusion
proteins in living cell nuclei. Accordingly, in this report, the in
vitro cross-linking assay and the mammalian two-hybrid assay
were complementary in proving the physiological association
of TRX and Ref-1. Our results showed that TRX associates
with Ref-1 via the catalytic center of TRX in living cells.
Coexpression of TRX lacking a C-terminal domain and Ref-1
fusion constructs showed no significant induction of luciferase
activity; this result suggests that the C-terminal region of TRX
is indispensable for its interaction with Ref-1 as well as for its
reducing activity.
In the present study, we demonstrated the PMA-dependent
translocation of cytoplasmic TRX to the nuclear compartment
and the direct physical association between TRX and Ref-1 in
vitro and in vivo. The precise mechanism and the role of
TRX–Ref-1 interaction in the regulation of gene expression is
to be clarified.
We thank S. Xanthoudakis and K. Umesono for plasmid vectors, T.
Curran for useful consultation, S. Toyama and H. Kagoshima for
technical advice, W. Brown and R. Yamaguchi for review of the
manuscript, G. M. Clore and A. Gronenborn for communicating their
results before publications, and Y. Kanekiyo and C. Ogawa for
secretarial help. This work was supported in part by a Grant-in-Aid for
Scientific Research from the Ministry of Education, Science and
Culture of Japan.
1. Pahl, H. L. & Baeuerle, P. A. (1994) BioEssays 16, 497–502.
2. Laurent, T. C., Moore, E. C. & Reichard, P. (1964) J. Biol. Chem
239, 3436–3444.
3. Holmgren, A. (1985) Annu. Rev. Biochem. 54, 237–271.
4. Holmgren, A. (1989) J. Biol. Chem. 264, 13963–13966.
5. Buchanan, B. B., Schurmann, P., Decottignies, P. & Lozano,
R. M. (1994) Arch. Biochem. Biophys. 314, 257–260.
6. Tagaya, Y., Maeda, Y., Mitsui, A., Kondo, N., Masutani, H.,
Hamuro, J., Brown, N., Arai, K., Yokota, T., Wakasugi, H. &
Yodoi, J. (1989) EMBO J. 8, 757–764.
7. Yodoi, J. & Uchiyama, T. (1992) Immunol. Today 13, 405–411.
8. Matthews, J. R., Wakasugi, N., Virelizier, J. L., Yodoi, J. & Hay,
R. T. (1992) Nucleic Acids Res. 20, 3821–3830.
9. Okamoto, T., Ogiwara, H., Hayashi, T., Mitsui, A., Kawabe, T. &
Yodoi, J. (1992) Int. Immunol 4, 811–819.
10. Sasada, T., Iwata, S., Sato S., Kitaoka, T., Hirota, K., Nakamura,
K., Nishiyama, A., Y. Taniguchi, Y., Takabayashi, A. & Yodoi,
J. (1996) J. Clin. Invest. 97, 2268–2276.
11. Silberstein, D. S., McDonough, S., Minkoff, M. S. & Balcewicz
Sablinska, M. K. (1993) J. Biol. Chem. 268, 9138–9142.
12. Biguet, C., Wakasugi, N., Mishal, Z., Holmgren, A., Chouaib, S.,
Tursz, T. &Wakasugi, H. (1994) J. Biol. Chem. 269, 28865–28870.
13. Wakasugi, N., Tagaya, Y., Wakasugi, H., Mitsui, A., Maeda, M.,
Yodoi, J. & Tursz, T. (1990) Proc. Natl. Acad. Sci. USA 87,
8282–8286.
14. Qin, J., Clore, G. M. & Gronenborn, A. M. (1994) Structure
(London) 2, 503–522.
15. Meyer, M., Schreck, R. & Baeuerle, P. A. (1993) EMBO J. 12,
2005–2015.
16. Schenk, H., Klein, M., Erdbru¨gger, W., Dro¨ge, W. & Schulze-
Osthoff, K. (1994) Proc. Natl. Acad. Sci. USA 91, 1672–1676.
17. Xanthoudakis, S. & Curran, T. (1992) EMBO J. 11, 653–665.
18. Xanthoudakis, S., Miao, G. G. & Curran, T. (1994) Proc. Natl.
Acad. Sci. USA 91, 23–27.
19. Xanthoudakis, S., Miao, G., Wang, F., Pan, Y. C. & Curran, T.
(1992) EMBO J. 11, 3323–3335.
20. Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K. & Pease,
L. R. (1989) Gene 77, 51–59.
21. Angel, P., Baumann, I., Stein, B., Delius, H., Rahmsdorf, H. J. &
Herrlich, P. (1987) Mol. Cell. Biol. 7, 2256–2266.
22. Nerlov, C. & Ziff, E. B. (1995) EMBO J. 14, 4318–4328.
23. Sadowski, I. & Ptashne, M. (1989) Nucleic Acids Res. 17, 7539.
24. Perlmann, T., Rangarajan, P. N., Umesono, K. & Evans, R. M.
(1993) Genes Dev. 7, 1411–1422.
25. Barzilay, G. & Hickson, I. D. (1995) BioEssays 17, 713–719.
26. Isoda, K. & Nu¨sslein-Volhard, C. (1994) Proc. Natl. Acad. Sci.
USA 91, 5350–5354.
27. Kishigami, S., Kannaya, E., Kikuchi, M. & Ito, K. (1995) J. Biol.
Chem. 270, 17072–17074.
28. Oblong, J. E., Berggren, M., Gasdaska, P. Y. & Powis, G. (1994)
J. Biol. Chem. 269, 11714–11720.
29. Tonissen, K.,Wells, J., Cock, I., Perkins, A., Orozco, C. &Clarke,
F. (1993) J. Biol. Chem. 268, 22485–22489.
30. Qin, J., Clore, G. M., Kennedy, W. M. P., Huth, J. R. & Gronen-
born, A. M. (1995) Structure (London) 3, 289–297.
31. Forman-Kay, J. D., Clore, G. M. & Gronenborn, A. M. (1992)
Biochemistry 31, 3442–3452.
32. Abate, C., Patel, L., Rauscher, F. J. d. & Curran, T. (1990)
Science 249, 1157–1161.
33. Wasylyk, C. &Wasylyk, B. (1993) Nucleic Acids Res. 21, 523–529.
34. Myrset, A. H., Bostad, A., Nadege, J., Pierre-Noel, L., Toma, F.
& Gabrielsen, O. S. (1993) EMBO J. 12, 4625–4633.
35. Guehmann, S., Vorbruehggen, G., Kalkbrenner, F. & Moelling,
K. (1992) Nucleic Acids Res. 20, 2279–2286.
36. Hayashi, T., Ueno, Y. & Okamoto, T. (1993) J. Biol. Chem. 268,
11380–11388.
37. Okuno, H., Akahori, A., Sato, H., Xanthoudakis, S., Curran, T.
& Iba, H. (1993) Oncogene 8, 695–701.
38. Ng, L., Forrest, D. & Curran, T. (1993) Nucleic Acids Res. 21,
5831–5837.
39. Chida, K. & Vogt, P. K. (1992) Proc. Natl. Acad. Sci. USA 89,
4290–4294.
40. Qin, J., Clore, G. M., Kennedy, W. M. P., Kuszewski, J. &
Gronenborn, A. M. (1996) Structure (London) 4, 613–620.
41. Walker, L. J., Robson, C. N., Black, E., Gillespie, D. & Hickson,
I. D. (1993) Mol. Cell. Biol. 13, 5370–5376.
42. Walker, L. J., Craig, R. B., Harris, A. L. & Hickson, I. D. (1994)
Nucleic Acids Res. 22, 4884–4889.
43. Henkel, T., Zabel, U., Zee, K. V., Mu¨ller, J. M., Fanning, E. &
Baeuerle, P. A. (1992) Cell 68, 1121–1133.
44. Yoneda, Y., Imamoto-Sonobe, N., Matsuoka, Y., Iwamoto, R.,
Kiho, Y. & Uchida, T. (1988) Science 242, 275–278.
45. Masutani, H., Hirota., K., Sasada, T., Ueda-Taniguchi, Y.,
Taniguchi, Y., Matsui, M. & Yodoi, J., Immunol. Lett., in press.
46. Tyrrell, R. M. (1995) BioEssays 18, 139–148.
47. Karin, M. & Hunter, T. (1995) Curr. Biol. 5, 747–757.
3638 Biochemistry: Hirota et al. Proc. Natl. Acad. Sci. USA 94 (1997)