ABSTRACT
We have recently elucidated the nature and function of transcription factors
present in Jurkat, glial and neuronal cells that interact with modulatory
region B, the nuclear receptor responsive element, in the long terminal repeat
of human immunodeficiency virus type 1 (HIV-1). Considering the key role that the combination of host cell proteins
plays in HIV-1 gene transcription, it appears essential to characterize proteins interacting with the adjacent region A.
In vitro
experiments revealed that the 5
'
-TGATTGGC-3
'
motif of region A is the target for at least three distinct proteins, one
belonging to the nuclear factor I family, while two others are related to the
cAMP response element binding (CREB) protein family. One of these proteins,
present in DNA-protein complex C2, is formed by distinct polypeptides of relative
molecular mass 43 000 and 50 000. We have purified the 43 kDa protein, which is distinct from CREB-43, and have shown that renatured p43 is able to specifically interact with site A. Transient expression
experiments with vectors containing wild-type or mutant motif A revealed that basal HIV-1 gene transcription in Jurkat cells is regulated by antagonistic
effects of the site A binding proteins.
Replication and gene expression of human immunodeficiency virus type 1 (HIV-1) depend upon multiple sequences present in the long terminal repeat
(LTR). Transcription is regulated by an interplay of viral and cellular host
proteins with regulatory elements of the LTR. Key elements such as the TATA,
Sp1 and NF-[kappa]B region, present in the proximal part of the LTR (+1 to -112) have been extensively characterized (for reviews see
1
,
2
). The role of the NF-[kappa]B element especially has been investigated in a variety of cell
types, including lymphocytes, monocytes and central nervous system-derived cells (
3
-
7
).
However, transcription control elements present in the -112 to -459 modulatory region of the LTR have been poorly characterized. This region includes regulatory elements that can mediate
transcription in many cell types, under a variety of growth and differentiation conditions (
8
). Recently several reports have described the nuclear receptor-responsive element (NRRE) located within the -356/-320 region as representing a site for complex regulatory
interactions among hormone receptors and orphan receptors, as well as the AP-1 transcription factor (
9
-
12
). We have recently reported that AP-1 is unable to directly bind to the NRRE sequence but interacts indirectly
via nuclear receptors belonging to the steroid-thyroid-retinoid nuclear receptor superfamily (
13
-
14
). Moreover, we have recently reported the role of retinoic acid receptors and
the orphan receptor COUP-TF on HIV-1 gene transcription in human central nervous system-derived cell lines (
14
). Since recent data have highlighted the importance of the U3 region of the LTR
in determining the pathogenicity of HIV-1 (
15
), studies concerning the nature and role of cellular transcription factors that
interact with various elements of the LTR appear essential.
In this study we have characterized the regulatory element spanning the -385 to -362 region of the LTR, which is directly adjacent to the NRRE. This region has previously been footprinted with nuclear extracts from Jurkat T cells and named site A (
16
-
17
). Our previous studies have revealed that this sequence is the binding site for
nuclear proteins present in a great variety of cell lines, such as Jurkat,
HeLa, astrocytoma, oligodendroglioma and neuronal cells (
13
). However, the nature and function of the nuclear proteins interacting with
site A remain elusive. Here we demonstrate that the TGATTGGC core sequence is
the binding site for at least three distinct proteins present in Jurkat and in
HeLa cells. By
in vitro
experiments we have analyzed the nature of the proteins forming the different
DNA-protein complexes. We have shown that one nuclear protein belongs to the
nuclear factor I (NFI) family (
18
), while other factors are related to the CREB protein family (for a review see
19
). We report here the purification of a nuclear factor of relative molecular
mass 43 000, distinct from CREB-43, which specifically binds to site A. Moreover, we provide evidence, by
transient expression experiments, that site A binding proteins play antagonistic roles in the control of
basal transcription of the HIV-1 genome in Jurkat cells.
To construct the 1N/tk-CAT, 1L/tk-CAT, 1Lm/tk-CAT, 2*1L/tk-CAT and 2*1Lm/tk-CAT vectors, one copy of the 1N, 1L
and 1Lm oligonucleotides and two copies of the 1L and 1Lm oligonucleotides
respectively were subcloned in sense orientation in the blunt-ended
Sal
I site of pBLCAT2, containing the herpes simplex virus thymidine kinase promoter
in front of the
CAT
gene (
20
; a gift of C.Kedinger, Unit INSERM 184, Strasbourg, France). The 1L-LTR-CAT and 1Lm-LTR-CAT vectors were constructed by site-directed mutagenesis with oligonucleotides 1L and
1Lm (Transformer TM Site-directed Mutagenesis Kit; Clontech). Mutations were controlled by DNA
sequencing.
Jurkat and HeLa cells were respectively grown in RPMI 1640 and DMEM medium supplemented with 10% fetal calf serum and 10 mM HEPES in the presence of penicillin and streptomycin (100 U/ml). Nuclear extracts were prepared as previously described (
21
). A phosphatase inhibitor, 0.5 mM [beta]-glycerophosphate, was added during the preparation of nuclear
extracts from Jurkat cells.
Jurkat cells (5 * 10
6
-10
7
cells/transfection) were transfected by the DEAE-dextran technique with 3 pmol plasmid DNA. After 48 h, cells were
collected and chloramphenicol acetyltransferase (CAT) assays were performed as
described previously (
21
) with 20 [mu]g protein extract and a 2 h incubation at 37oC.
Protein-DNA binding reactions were performed with 5 or 10 [mu]g crude nuclear extract as previously described (
13
). For competition assays, unlabeled oligonucleotides were added at a 50- or 200-fold molar excess at the same time as the probe. When purification
columns were analyzed, 10 [mu]l of the fractions were used. The sequences of the oligonucleotides used as
probes or competitors are shown in Figure
1
. In supershift assays performed as previously described (
13
), we used antibodies directed against CREB-1 and ATF-1 and reactive with ATF-1 p35, CREB-1 p43 and CREM-1 (Santa Cruz Biotechnology). Binding reactions
were carried out with the protein CREB bZIP (Santa Cruz Biotechnology).
Methylation interference assays were performed with oligonucleotide 1L as described previously (
13
).
Crude nuclear extract (10 [mu]g) from Jurkat cells and partially purified proteins (10 [mu]g) of a 0.4 M KCl fraction from the heparin-Sepharose column were incubated for 15 min on ice with the 5'-end-labeled 1L probe substituted with
bromodeoxyuridine under standard mobility shift assay conditions. Following
electrophoresis, the gel was irradiated for 12 min using a UV transilluminator
(254 nm; Bioblock Scientific). Radioactive bands corresponding to complexed and
free DNA identified by autoradiography were excised and incubated in 40 [mu]l denaturing buffer (62 mM Tris-HCl, pH 6.7, 2% SDS, 10% glycerol, 50 mM dithiothreitol) and heated
at 100oC for 5 min. The gel slices and the denaturing buffer were loaded on a 10% SDS-polyacrylamide gel. The gel was fixed, dried under vacuum and the proteins
directly involved in the DNA interaction were identified by autoradiography. We
used prestained molecular weight markers (BioRad).
Nuclear proteins partially purified by heparin-Sepharose (20 [mu]g) or by DNA affinity chromatography (5 [mu]g) were resolved on a 10% SDS-PAGE gel and electrotransferred onto nitrocellulose
membranes. The membranes were washed three times for 1 h with renaturation buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 10 mM dithiothreitol, 2.5% Nonidet P-40, 10% glycerol, 3% BSA) (
22
). The membranes were then placed in a heat-sealable pouch in binding buffer (10 mM Tris-HCl, pH 7.5, 40 mM NaCl, 1 mM DTT, 1 mM EDTA, 8% v/v glycerol, 3%
BSA, 60 [mu]g poly(dI[middot]dC), 5 mM MgCl
2
, 5 [mu]g/ml
32
P-labeled oligonucleotide 1L or 1Lm). After 16 h constant agitation,
membranes were washed for 2 h at room temperature with 10 mM Tris-HCl, 50 mM NaCl with several changes and autoradiographed on Kodak X-OMAT film.
Aliquots of 25 mg nuclear protein extract, obtained from 5 * 10
9
Jurkat cells were applied to a 5 ml heparin-Sepharose column equilibrated with buffer Z (20 mM HEPES, pH 7.9, 0.2 mM EDTA, 20% glycerol, 1 mM dithiothreitol) containing 50 mM KCl. Proteins
were separated using a step gradient from 0.1 to 1 M KCl in buffer Z. Activity was detected in fractions eluted between 0.3 and 0.4
M KCl by gel mobility shift assay. These fractions were pooled and dialyzed
against buffer Z containing 0.1 M KCl. The pool was mixed with salmon sperm DNA
(25 [mu]g/ml), incubated on ice for 15 min and loaded onto a 2 ml oligonucleotide
1Lm affinity column equilibrated in buffer Z containing 0.1 M KCl. The column was washed with the same solution. The flow-through was loaded onto a 2 ml oligonucleotide 1L affinity column equilibrated in
buffer Z containing 100 mM KCl. The column was washed with the same solution
and eluted with a step gradient from 0.1 to 1 M KCl in buffer Z.
Oligonucleotide 1L binding activity was detected in fractions at 0.5 M KCl.
These fractions were pooled, dialyzed against buffer Z containing 0.1 M KCl and
recycled on the same column. An aliquot (10 [mu]l) of each fraction was assayed by gel retardation assay using the
oligonucleotide 1L as probe. The active fractions were also analyzed by 10% SDS-PAGE. The sequence-specific DNA affinity columns were prepared according to Kadonaga and Tjian (
23
) using CNBr- activated Sepharose 4B (Pharmacia).
Fractions (20 [mu]g) containing peak activity after heparin-Sepharose chromatography were subjected to a preparative SDS-10% PAGE gel. After electrophoresis, the gel was cut
horizontally into 1 mm slices between 40 and 70 kDa. The proteins were eluted
from these slices as described by Hager and Burgess (
24
). After elution, proteins were precipitated with acetone and resuspended in
buffer H (50 mM HEPES, pH 7.9, 0.5 mM EDTA, 0.05% Nonidet P-40, 20% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 M NaCl and 5 [mu]g/ml leupeptin and pepstatin) containing 6 M guanidine-HCl for 20 min. The denaturing agent was diluted 50-fold with buffer H and renaturation was allowed for 4 h at 4oC. Activity of renatured proteins was tested by gel
mobility shift assay in the absence of non-specific competitors.
The sequence spanning the -385/-362 distal region of the HIV-1 LTR has previously been footprinted with nuclear extracts
from different cell types and named site A (
13
,
16
,
17
). A number of studies have shown that the adjacent region B is the binding site
of proteins belonging to the nuclear receptor superfamily (
11
-
13
). In order to analyze the nuclear proteins interacting with site A, we
performed DNA mobility shift assays using oligonucleotides 1N, 1L and 1Lm (Fig.
1
A). The 1N sequence, which corresponds to site A, gave rise to four DNA-protein complexes, C1-C4, with extracts from HeLa cells (Fig.
1
B, lane 1) and to three complexes, C1, C2 and C4, with extracts from Jurkat
cells (lane 3). Specificity of binding was ascertained by competition experiments using a 50-fold molar excess of unlabeled homologous 1N oligonucleotide. The formation of complexes C1, C2 and C3 was
specifically competed by oligonucleotide 1N (lanes 2 and 6). Under similar
conditions the formation of complex C4 appeared less specific, since half of
the complex was still detected (lanes 2 and 6).
Identification of the proteins binding to site A was first achieved using
competition gel shift experiments. A close inspection of the 1N oligonucleotide
revealed the sequence TGGCA, which corresponds to the consensus binding site of
proteins of the CTF/NFI family (
18
). We therefore used as a probe the mutant oligonucleotide 1L containing the
TGACA 1 bp mutation; interestingly this mutation abolished formation of complex
C1 (lane 4). When oligonucleotide 1L was used in a 50-fold molar excess as a competitor of the 1N probe, formation of complex C1
was unaffected, while formation of complexes C2-C4 was abolished (lane 7). To further assess that the protein forming
complex C1 belongs to the NFI family, we used as a competitor oligonucleotide
CR, previously shown to be the binding site of the NFI protein in the promoter
of the human transferrin gene (
25
). As expected, in the presence of a 50-fold molar excess of CR competitor, formation of complex C1 was
specifically prevented (lane 11). As a control, the mutant oligonucleotide 1Lm,
which contains 4 bp mutations, was unable to give any specific complex when
used as a probe (lane 5) or to compete for formation of any complex in
competition experiments (lane 8). A recent report described that transcription
factor YB-1, a member of the NFI family, is able to interact with site A of the HIV-1 LTR (
26
). However, supershift experiments with antibodies directed against YB-1 did not alter formation of complex C1 (results not shown). Taken
together, these results suggest that complex C1 is formed by a member of the
NFI protein family, distinct from YB-1.
Oligonucleotide 1L contains the TGAC half-site consensus sequence of the cAMP response element (CRE), in direct
repeat with the sequence TGAT, which varies by one base from the CRE consensus
(
19
). We therefore performed gel shift assays, using as a competitor in 200-fold excess a CRE oligonucleotide containing the consensus palindromic sequence TGACGTCA. Formation of complexes C2
and C4 was totally abolished (lane 9). As a control, a mutant CRI
oligonucleotide containing mutations in the CRE site did not affect formation
of complexes C1 and C2 (lane 10). Similarly, when oligonucleotide TRE
containing the TGACTCA binding site for proteins of the AP-1 family was used as a competitor, the pattern was unaffected (results not
shown). These results suggest that the protein forming complex C2 is related to
the ATF/CRE binding protein family. We therefore performed shift assays in the
presence of antibodies directed against CREB-1 (
27
) or ATF-1 (
28
). All the complexes were unaffected (results not shown), indicating that the
CRE-binding protein(s) in complex C2 is (are) different from CREB-1 and ATF-1.
We next tested whether a truncated CREB-1 protein, containing the binding and dimerization domain, was able to
bind to oligonucleotides 1L and 1N (Fig.
2
). Interestingly, with the 1N probe CREB-1 was able to form a weak retarded complex. With the 1L probe, containing
a TGAC site, and with the CRE probe, containing the palindromic TGACGTCA site,
CREB-1 gave a more intense retarded complex. This result demonstrates that CREB
proteins are able to interact with the 1L sequence and with a weaker affinity
with the 1N sequence. Taken together, these results indicate that complex C2 is
formed by a protein distinct from CREB-1 and ATF-1 and related to proteins of the CREB/ATF family. In HeLa cells, the
additional complex C3 had a behavior similar to that of complex C2 (results not
shown), suggesting that proteins forming complex C3 are also related to the
CREB/ATF family.
To evaluate the functional effect of the different proteins interacting with site A of the LTR, we constructed a series of pBLCAT2 reporter plasmids
containing one copy of either oligonucleotide 1N, 1L or 1Lm or two copies of 1L
or 1Lm in front of the herpes simplex virus thymidine kinase promoter upstream
of the
CAT
gene. These vectors were transfected into Jurkat cells and CAT activities were
determined (Fig.
4
). Interestingly, the activity of all the vectors was higher than that of the
control pBLCAT2 vector. Transfection of the 1Lm/tk-CAT and 2*1Lm/tk-CAT vectors resulted in 1.8- and 4.0-fold stimulation of CAT activity respectively,
reflecting the positive action of the protein(s) forming complex C4.
Transfection of the 1L/tk-CAT and 2*1L/tk-CAT constructs led to 3.2- and 6.2-fold transcriptional activation respectively, indicating that proteins
forming complexes C2 and C4 are able to optimally stimulate transcription. Interestingly, the CAT
activity of the 1N/tk-CAT construct was reduced, when compared with that of 1L/tk-CAT, which could suggest mutual inhibition between proteins forming
complexes C1 and C2. Another explanation could be that complex C1 contains a
protein(s) that acts as an inhibitor of site A promoter activity and that is
overwhelmed by the positive activity of proteins forming complexes C2 and C4.
To estimate the apparent molecular mass of the protein(s) present in the major
complex C2, we performed UV cross-linking experiments. Gel shift assays were carried out with
bromodeoxyuridine-substituted
32
P-labeled 1L probe in the presence of either crude nuclear extract from
Jurkat cells or with partially purified proteins eluted from a heparin-Sepharose column. After electrophoresis, the native gel was UV irradiated
and the band corresponding to complex C2 was excised and subjected to SDS-10% PAGE. As shown in Figure
6
, two oligonucleotide-protein complexes were detected with apparent molecular masses of 67 and
60 kDa. As a control, similar experiments performed with the C4 complex revealed a unique band migrating in the 80 kDa range (results not shown). These results suggest that the C2 complex is
composed of two proteins, with apparent molecular masses of ~43 (+- 5) and 50 (+- 5) kDa, while the C4 complex is formed of one protein of ~60-66 kDa.
Recent studies have revealed the increasing importance of the modulatory region
of the LTR in regulation of HIV-1 gene expression. The -356/-320 NRRE is the site of interaction of transcription factors
of the nuclear receptor superfamily which regulate transcription in a cell-type specific manner (
10
-
14
,
16
). However, the nature and role of nuclear host proteins which interact with the
previously described (
13
,
16
,
17
) -385/-362 adjacent region A remain to be investigated.
Here we describe a preliminary characterization of the nature and functional
role of nuclear proteins present in HeLa and Jurkat cells which interact with
region A. Our studies reveal the complexity of the interactions, since multiple
distinct protein species have the ability to interact with this distal region.
In Jurkat and HeLa cells respectively three and four distinct DNA-protein complexes were formed in gel mobility shift assays with
oligonucleotide 1N, corresponding to site A.
Our gel retardation competition experiments suggest that the protein forming
complex C1 belongs to the NFI family of proteins. Moreover, our data indicate
that the proteins forming the specific complex C2 in HeLa and Jurkat cells and
complex C3 in HeLa cells interact with the core sequence 5'-TGATTGGC-3'. The importance of this core sequence for binding of
the distinct proteins was revealed both by methylation interference and by gel
mobility shift data obtained with the 1 nt mutated oligonucleotide 1L,
containing the modified sequence TGATTGAC. While this mutation abolished
formation of complex C1, it increased the binding ability of proteins forming
complexes C2 and C3. Interestingly, the core sequence corresponds to a direct
repeat of a mutant TGAC half-site of the consensus CRE, TGACGTCA (
19
). Our competition data with the CRE and mutant oligonucleotides raise the possibility that proteins forming complexes C2 and C3 may be related
to the CREB/ATF family. However, our data show that CREB-1 and ATF-1 are not present in complexes C2 and C3. It is known that many
proteins are able to bind
in vitro
to CRE-related elements and that members of the ATF/CREB family form cross-family heterodimers with members of the AP-1 family (
29
). In that respect it is worth noting that proteins forming complexes C2 and C3
were not able to interact with the related TRE binding site of proteins of the
AP-1 family. Interestingly, our data show that although the 1L sequence
contains a mutated direct repeat of the CRE consensus sequence, it was able to
bind the CREB-1 protein. As expected, the binding affinity of CREB-1 for the 1N sequence was lower than that for the 1L and CRE
sequence. Interaction of ATF/CREB proteins with the LTR region of human T cell
lymphotropic virus type I (HTLV-I) has been reported (
30
). However, binding to the HIV-1 LTR of a protein related to the CREB family has not yet been described.
The functional role of the site A binding proteins was first examined in Jurkat
cells by transfecting tk-CAT reporter vectors containing the different 1N, 1L and 1Lm
oligonucleotides. Transient expression data revealed that in the context of the
tk promoter, all the nuclear proteins binding to the various 1L, 1N and 1Lm
sequences, corresponding to the wild-type and mutated site A, were able, to a different extent, to activate
transcription. Transcriptional activity was highest with the 1L sequence,
forming the major complex C2 in gel shift assays. Activity was decreased with
the 1N sequence, forming the two major specific complexes C1 and C2, suggesting
either mutual inhibition between two proteins binding to the same site or an
ability of the protein present in complex C1 to act as a repressor. In
addition, the functional role of the various site A binding proteins was
established in the context of the entire HIV-1 LTR region. Transcriptional activity was highest with the 1Lm-LTR-CAT vector, was decreased with 1L-LTR-CAT and was lowest with the wild-type LTR-CAT vector. These data reveal that in
the context of the LTR, proteins forming complexes C1 and C2 are able to
inhibit the positive action of protein(s) forming complex C4.
Since the protein(s) forming complex C2 functions as a modulator(s) of HIV-1 gene transcription, we deemed it essential to analyze these factors more closely. According to UV cross-linking data, complex C2 is formed by a heterodimer consisting of 43 and 50
kDa proteins. As suggested by our gel competition data, at least one member of
this heterodimer has a DNA binding domain able to recognize the CRE sequence.
It will be interesting to determine in future which member of the dimer
possesses the CRE binding activity.
DNA affinity chromatography, using concatemers of oligonucleotide 1L and 1Lm, was a convenient method to purify a protein of relative molecular
mass 43 000 which was recognized by the 1L probe on a Southwestern blot. After
elution and renaturation, p43 was able to specifically bind to the 1L probe and
to form a DNA-protein complex C of faster mobility than complex C2 formed with crude
nuclear proteins, which is consistent with our UV cross-linking data.
Protein p43 appears to be a novel nuclear factor, since a search for homologous
N-terminal amino acid sequences in protein databanks did not reveal any
homology with a known protein (M.H.Metz-Boutigue; C.Schwartz, D.Aunis and E.Schaeffer, unpublished results). To
definitely resolve the question of whether p43 or p50 are related to the
CREB/ATF family, it will be necessary to carry out comparative studies using
cloned cDNAs. Therefore, cloning of the corresponding cDNA represents an essential step for complete identification of this protein and for
elucidation of its role in modulation of HIV-1 gene transcription. Interestingly, region A is a binding site for
proteins of lymphoid and non-lymphoid cells; future studies are necessary to investigate precisely the nature as well as the role of the cellular site A binding
proteins in different cell types infected by HIV-1.
We thank H.D.Royer for YB-1 antibodies, B.E.Sawaya for helpful discussions and D.Filliol for
excellent technical assistance. This work was supported by the Institut
National de la Santé et de la Recherche Médicale (INSERM), the Agence Nationale des Recherches sur le
SIDA, the Association `Ensemble contre le SIDA', the Fondation pour la
Recherche Medicale (FRM) and the Association pour la Recherche sur le Cancer.
C.S. received a fellowship from INSERM and F.C.H. from FRM.
*To whom correspondence should be addressed. Tel: +33 3 88 45 67 18; Fax: +33 3
88 60 08 06; Email: schaeffer@neurochem.u-strasbg.fr
REFERENCES
Return