HTLV-I Tax self-association in optimal
trans
-activation function
HTLV-I Tax self-association in optimal trans -activation function
Dong-Yan
Jin
and
Kuan-Teh
Jeang*
Molecular Virology Section, Laboratory of Molecular Microbiology, National
Institute of Allergy and Infectious Diseases, National Institutes of Health,
Bethesda
, MD 20892-0460,
USA
Received September 4, 1996;
Revised and Accepted November 12, 1996
ABSTRACT
HTLV-I Tax protein is a potent transcriptional activator of viral and cellular
genes. Tax does not bind DNA directly but interacts through protein-protein contact with host cell factors that recognize the viral long
terminal repeat (LTR). Domains within Tax needed for protein-protein interaction have not been fully characterized. In studying transcriptional function in yeast cells, we
unexpectedly found that Tax functions optimally not as a monomer, but as a
homodimer. Here we have used the one hybrid and two hybrid genetic approaches
in yeast to investigate the region(s) within Tax necessary for self-association. Dimer formation was also confirmed biochemically by using
electrophoretic mobility shift (EMSA) and supershift assays. Twenty two Tax point mutants were utilized to map relevant residues. Genetic results from this series of mutants revealed that a necessary region for dimerization is contained within a previously characterized zinc finger domain. Two loss-of-function Tax mutants, each poorly active when assayed individually, were found to have complementing activity when co-expressed together. This genetic complementation suggests a mechanism
for
trans
-activation resulting from simultaneous but non-identical contact with a responsive target by each of two Tax
monomers in a dimer.
INTRODUCTION
Human T cell leukemia virus type I (HTLV-I) is etiologically associated with adult T cell leukemia (ATL) and
neurological disorders termed HAM/TSP (reviewed in
1
). In addition to Gag, Pol and Env proteins common to other retroviruses, the
HTLV-I genome encodes regulatory proteins such as Tax and Rex, which are
crucial for virus replication and pathogenesis (reviewed in
2
).
The 40 kDa Tax protein activates transcription of the viral genome through three
imperfectly conserved 21 nt direct repeats in the viral long terminal repeat
(LTR) (
3
-
6
). Tax also modulates the expression of other viruses, including the human
immunodeficiency virus (
7
) and cellular genes including interleukin-2 (
8
,
9
), interleukin-2 receptor [alpha]-chain (
8
,
10
), granulocyte macrophage colony stimulating factor (
11
), c-
fos
(
12
) and [beta]-polymerase (
13
,
14
). Tax does not bind DNA directly, but can interact with cellular DNA binding proteins, including cAMP responsive element binding protein (CREB) (
15
-
18
), nuclear factor [kappa]B (NF-[kappa]B) (
19
,
20
, reviewed in
21
), serum response factor (SRF) (
22
) and TATA binding protein (TBP) (
23
). In one perspective,
trans
- activation by Tax is proposed to be mediated through direct protein contact with these sequence-specific DNA binding factors. Another perspective suggests that Tax
facilitates/stabilizes the dimerization of transcription factors at the promoter (
24
-
26
, reviewed in
27
). Additional molecular mechanisms for Tax function have also been proposed (
20
,
21
,
28
-
30
).
Previously we have described aspects of Tax
trans
-activation in mammalian cells (
15
,
17
,
31
-
35
). In particular, we have focused on steps through which Tax communicates with the basal transcription machinery.
Although conflicting findings exist, there is good evidence that Tax can be
tethered to the promoter via contact with upstream transcription activators
such as CREB (
16
,
36
,
37
). Other evidence suggests that Tax can also directly bind basal factors such as
TBP (
23
). Taken together these findings suggest that the capacity for protein-protein contact might be an important functional component of the action
of Tax. Such supposition is not surprising, since many important regulatory
proteins, such as CREB, NF-[kappa]B and STAT, commonly employ homo- and hetero-dimerization as a means of regulating specificity and function (
38
-
41
).
Genetic studies in higher eukaryotic cells are complex and time consuming.
Expression of Tax in budding yeast (
42
-
44
) could provide a simple and useful model system for clarifying molecular
interactions within eukaryotes. Yeast and its associated genetics afford rapid
and powerful means to dissect molecular processes at the promoter. With this in
mind, we have employed yeast one hybrid and two hybrid assays (
45
) to explore intracellular Tax structure/function. In the course of these studies and in the process of
investigating how Tax might contact cellular effectors (Jin and Jeang, in
preparation), we found that this transcriptional activator functions optimally
as a homodimer. We show here that Tax, when targeted to the promoter via the
Gal4 DNA binding domain (Gal4bd), can potently activate transcription in yeast
and that such activation is mediated through the formation of Tax-Tax dimers. Structure-function studies using 22 Tax mutants implicated an N-terminal zinc finger region in Tax (
32
) as being important for dimerization. Interestingly, we found that two loss-of-function Tax mutants were poorly active when expressed individually, but were highly active when co-expressed together. This complementation is consistent with a mechanistic model in which each monomeric Tax molecule in a homodimer
exerts simultaneous but non-identical contact on a responsive target.
MATERIALS AND METHODS
Plasmids and yeasts
Yeast reporter strains
Saccharomyces cerevisiae
Y187 and HF7c have been described elsewhere (
46
,
47
). Strain Y187 expresses the
Escherichia coli
lacZ
gene driven by the Gal4-responsive GAL1 promoter. Y187 is deleted in its endogenous
gal4
gene. Strain HF7c expresses a
HIS3
reporter gene under the control of the GAL1 promoter.
Wild-type and mutant Tax cDNAs have been described previously (
33
,
48
). Mutations are designated by the amino acid to be changed, the position of the
residue and the replacement amino acid (e.g. Tax H52-Q). Amino acids that were removed in mutants are indicated in parentheses
[e.g. Tax [Delta](3-10)]. Tax cDNAs were inserted in-frame into vectors pGBDT9 or pGAD424 (
45
; Clontech Laboratories Inc.) that express the Gal4bd (amino acids 1-147) or Gal4 activation domain (Gal4ad, amino acids 768-881). Fusion proteins produced from these vectors were targeted to
the yeast nucleus using a nuclear localization sequence (NLS). Although Gal4bd
has an intrinsic NLS, an NLS from the SV40 large T antigen (LT) was also added
to the N-terminus of Gal4ad.
Gal4bd-p53 and Gal4ad-LT were from Clontech. A human cDNA encoding TBP was a gift from A.J.Berk (
49
). pGBDTax-T was constructed by replacing the P
ADH1
promoter in pGBDT9/Tax with a transcriptionally more active wild-type *P
ADH1
promoter from pGAD GH (Clontech). pGBDTax-L and pGBDTax-H were constructed by replacing the
TRP1
selective marker in pGBDTax-T with
LEU2
and
HIS1
respectively. Tax-expressing plasmids IEX, IEX Q9-G and IEX S132-A and reporter plasmid pU3RCAT, in which the CAT cDNA is
positioned downstream of the HTLV-1 LTR, have been described elsewhere (
33
).
Reporter assays
Yeasts were transformed by the LiAc method of Gietz
et al
. (
50
). [beta]-Galactosidase activity was measured in strain Y187. Expression of His3p was tested in strain HF7c. A qualitative colony-lift filter assay was performed using 5-bromo-4-chloro-3-indolyl-[beta]-D- galactopyranoside (X-gal) as substrate. Briefly, yeast transformants
were transferred to Whatman No. 1 filter paper, permeabilized by submerging in liquid nitrogen, allowed to thaw at room temperature and placed onto another filter presoaked in Z-buffer (60 mM Na
2
HPO
4
, 40 mM NaH
2
PO
4
, 100 mM KCl, 1 mM MgSO
4
and 50 mM [beta]-mercaptoethanol) containing 0.5 mg/ml X-gal. Strongly positive colonies (++++) appeared as dark blue within 20 min.
Positives (+++) were blue within 1 h and weak positives (++) were blue in 6 h.
Overnight incubation produced some very weak positives (+) with a light blue
color.
For quantitative [beta]-galactosidase assay, chlorophenol red-[beta]-D- galactopyranoside (CPRG) was used as
substrate. Overnight cultures grown in SD selective medium (0.67% bacto-yeast nitrogen base, 2% glucose, supplied with dropout nutrient solution
missing one or more amino acids) were diluted 5-fold in YEPD rich medium (1% bacto-yeast extract, 2% bacto-peptone, 2% glucose) and were grown until mid log phase (OD
600
of 0.5-0.8). Yeast cells were disrupted by freeze-thaw and by additional vortexing with acid-washed glass beads. [beta]-Galactosidase activity was assayed using the method of Miller
(
51
), except that CPRG was used for color development and the reaction buffer consisted of 100 mM HEPES, 150 mM NaCl, 2 mM MgCl
2
, 0.5 mM L-aspartate, 1% bovine serum albumin and 2 mM CPRG. [beta]-Galactosidase activity was expressed as relative CPRG units,
defined in the same manner as Miller units (
51
). Numbers presented in the tables throughout are representative of triplicate
determinations of duplicate transformants.
For detection of His3p expression, yeast transformants were tested for growth on SD medium without histidine. Positive colonies (+) that survive histidine dropout appeared within 48 h, while no growth was
observed after 60 h on negative plates (-). Indicated amounts of 3-aminotriazole (3AT) were added to the medium specifically to assess His3p expression. Growth in
escalating amounts of 3AT requires induced levels of His3p and the degree of 3AT
resistance reflects the degree of HIS3 transcription (
50
).
HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented with 2 mM L-glutamine, 10% fetal calf serum, 100 U/ml penicillin/streptomycin. For chloramphenicol acetyltransferase
(CAT) assay, cells were seeded at 5 * 10
5
cells/well into six-well culture plates. Calcium phosphate transfection was performed and CAT activity was assayed as previously described (
35
).
Extract preparation
Yeast whole cell extracts were prepared from yeast cells (strain Y187) grown in SD medium. Cells were resuspended in a 1/10 vol. of extraction buffer (20 mM HEPES, pH 7.5, 400 mM NaCl, 1 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 1 mg/ml aminoethyl benzenesulfonyl
fluoride, 1 mg/ml leupeptin, 1 mg/ml pepstatin and 1 mg/ml aprotinin). An equal
volume of acid-washed glass beads was added and the mixture was vortexed at 4oC for 5 min. The extract was separated from glass beads by
sedimentation and was clarified further by centrifugation. Protein concentrations were determined
by the Bradford dye binding procedure (
52
; BioRad) and were normalized by adding extraction buffer.
Western blotting
Proteins from yeast whole cell extract were trichloroacetic acid precipitated,
washed in 80% acetone and resuspended in SDS gel loading buffer (60 mM Tris
base, 2% SDS, 10% glycerol, 5% 2-mercaptoethanol). Samples containing equal amounts of protein (15 [mu]g) were separated by 12% SDS-PAGE and electroblotted onto Immobilon-P membranes (Millipore) using a Millipore semi-dry blotting apparatus as per the manufacturer's
protocol. The blot was probed with a previously described rabbit anti-Tax polyclonal antibody (
33
) at 1:1000 dilution. Bands were visualized by chemiluminescence (Western-Light; Tropix Inc.) using goat anti-rabbit alkaline phosphatase-conjugated secondary antibody.
Electrophoretic mobility shift assay (EMSA)
Oligonucleotides 5'-AGCTCGGAAGACTCTCCTCCGGATCC-3' and 5'-AGCTGGATCCGGAGGAGAGTCTTCCG-3' were annealed to produce a
consensus Gal4 binding site probe. The probe was labeled by filling-in using Klenow enzyme at 30oC for 30 min in the presence of [[gamma]-
32
P]dATP and unlabeled dGTP, dCTP and dTTP. Labeled probe was separated from unincorporated nucleotides through a Sephadex G25 spin column. Binding reactions were performed
in 20 [mu]l volumes containing 20 [mu]g protein, 0.5 [mu]g poly(dI[middot]dC), 0.5 ng labeled probe in EMSA buffer (10 mM HEPES, pH
7.5, 4 mM Tris-HCl, pH 7.5, 5% glycerol, 50 mM KCl, 0.5 mM dithiothreitol, 1 mM EDTA and
100 [mu]g/ml bovine serum albumin). After incubation at room temperature for 30 min,
the mixture was resolved in a 5% polyacrylamide gel with 0.25* TBE (Tris-borate EDTA; 1* TBE is 90 mM Tris-borate, 2 mM EDTA) as running buffer.
RESULTS
Definition of Tax activity in yeast
To verify yeast as an appropriate study model, we first defined Tax activity in
this system. We fused the full-length Tax protein (amino acids 1-353) to the C-terminus of the Gal4 DNA binding domain (Gal4bd; amino acids
1-147). In this manner, we tested the ability of the fusion protein to
stimulate transcription from reporter genes driven by the Gal4-responsive UAS. As a negative control, plasmids that express Gal4bd or a
Gal4bd-p53 fusion (
53
) were tested in parallel. Additionally, plasmids known to be biologically
active in this assay [e.g. Gal4bd fused with the Gal4ad (activation) domain and
a Gal4bd-TBP fusion;
54
] were used as positive controls.
Qualitative (filter assay, Table
1
) and quantitative (CPRG units, Table
1
) assays for [beta]-galactosidase activity in yeast Y187 and growth of transformed yeast
HF7c in selective histidine dropout medium are summarized. In all experiments,
Gal4bd-Tax activated the promoter-downstream reporter. In contrast, neither Gal4bd (Table
1
, line 1) nor Gal4bd-p53 (Table
1
, line 3) showed detectable activity. The level of activation by Galbd-Tax, although lower than that induced by native Gal4 (compare lines 2 and
5, Table
1
), was comparable with that stimulated by a known activator, Gal4bd-TBP (
54
; compare lines 2 and 4, Table
1
). Furthermore, when we increased the amount of Gal4bd-Tax produced by either using a stronger promoter (Fig.
1
, lane 2) or by expressing Gal4bd-Tax simultaneously from two or three stably transforming plasmids each
with a different selectable marker (Fig.
1
, lanes 3 and 4), the level of [beta]-galactosidase activity was concordantly elevated (Fig.
1
). Thus, under these conditions, there is a dose-dependent activation of transcription by Gal4bd-Tax.
Dimerization of Tax contributes to activity
In the process of characterizing transcriptional function, we often find,
depending upon expression level, non-linear changes in Tax activity in a manner suggestive of protein
oligomerization. Previously, biochemical cross-linking results compatible with Tax homodimers have been reported (
56
). Tax activity in yeast affords a powerful genetic system to define
dimerization in detail. To delineate this interaction better, assays were
performed in which separate plasmids expressing either Gal4bd-Tax or Gal4ad-Tax were used to co-transform yeasts engineered to permit measurement of either
a
lacZ
or a
HIS3
response. Because Gal4ad-Tax contains the strong ad activation domain and Gal4bd-Tax contains the specific bd binding domain for the promoter-upstream UAS, the rationale is that if Gal4bd-Tax and Gal4ad-Tax could dimerize via Tax-Tax protein interaction, then the
strong ad activation domain would be brought proximal to the promoter. As a
positive control, the well-characterized interaction of p53 with SV40 LT (
53
) was assayed in parallel (Table
3
, line 20). As negative controls, plasmids Gal4bd and Gal4ad-Tax or Gal4bd-Tax and Gal4ad (Table
3
, lines 21 and 22) were paired in separate co-transformations. CPRG units from transformants were compared with baseline
activity from Gal4bd-Tax transformants (Table
3
, line 1). Because both high and low level expression of His3p can result in
growth on selective medium (Table
3
, column 6), 30 mM 3AT was added to differentiate high from low activation
(Table
3
, column 7).
In the above setting, the reporter activities induced by Gal4bd-Tax and Gal4ad-Tax (Table
3
, line 2) were found to be comparable with those from Gal4bd-p53 and Gal4ad-SV40 LT (Table
3
, line 20). Readouts from both the qualitative His3p test and the CPRG
measurement (2726 U for Gal4bd-Tax/Gal4ad-Tax and 3280 U for Gal4bd-p53/Gal4ad-SV40 LT; Table
3
) were similar for the two pairs. Thus, at this level of assessment, the
affinity of Tax-Tax interaction is very similar to that dictating the binding of p53 and
SV40 LT (
52
). The specificity of Tax-Tax dimerization is supported by findings that Tax interacted with
neither Gal4bd alone nor Gal4ad alone (Table
3
, lines 21 and 22).
The genetics of Tax dimerization were further corroborated biochemically by
EMSA. For this series of assays, whole cell extracts were prepared separately
from yeast Y187 transformed with plasmids expressing Gal4bd-Tax, Gal4bd-Tax + Gal4ad, Gal4bd-Tax + Gal4ad-Tax or Gal4ad-Tax. The extracts were then challenged with
an excess of probe. Tax-Tax dimerization was monitored by the ability of one molecule of Tax to
supershift a second probe-bound Gal4bd-Tax molecule. We titrated the amount of Gal4bd-Tax such as to favor monomeric Gal4bd-Tax-probe interaction. Indeed, when Gal4ad-Tax was added subsequently to a Gal4bd-Tax protein assembled onto a
Gal4 binding site probe (Fig.
4
, lane 4), a supershifted band (open arrow, Fig.
4
, lane 2) with a slower migration in gel was observed. A similar supershifted
signal was obtained when the extract was made from yeast that simultaneously co-expressed Gal4bd-Tax and Gal4ad-Tax (Fig.
4
, lane 1). Parallel assays show that proteins which lack the bd domain (i.e.
Gal4ad-Tax and Gal4ad alone; Fig.
4
, lanes 5 and 6) do not bind the probe. A further control showed that addition
of Gal4ad alone to Gal4bd-Tax produced a single monomeric probe complex (filled arrow, Fig.
4
, lane 3) without a second supershifted band. This is consistent with the
interpretation that Tax needs to be present on both DNA-bound and DNA-free moieties for formation of a supershifted protein-protein-DNA complex.
Biochemical and genetic evidence for dimerization prompted us to define a
protein domain int Tax specifying homologous association. We genetically
analyzed 15 Tax mutants for their ability to bind a Gal4ad-tagged wild-type Tax protein (Gal4ad-Tax, Table
3
). The approach challenges different versions of UAS-bound Gal4bd-Tax protein for recruitment of the strong ad-containing Gal4ad-Tax wild-type protein to the promoter. Successful
formation of a Gal4ad-Tax/Gal4bd-Tax complex at the promoter is reflected in activated expression
of the downstream
lacZ
/
His3
cDNA. Thus, we observed that the 15 Tax mutants fell into three groups. Nine
mutants, Tax [Delta](3-6), Tax C23-S, Tax S32-A, Tax S113-A, Tax S150-A, Tax S258-A, Tax L296-G, Tax L320-G and Tax [Delta](337-353),
bound Gal4ad-Tax well (Table
3
, lines 3, 5, 7, 10 and 12-16). These mutants can self-associate (see for example Table
3
, line 18) and can also bind with another from within the same group (data not
shown). Two mutants, Tax Q9-G and Tax S132-A, were partially active and bound Gal4bd-Tax modestly (Table
3
, lines 4 and 11). In contrast, three other mutants, Tax C29-S, Tax C36-S and Tax H52-Q, failed to exhibit any protein binding potential (Table
3
, lines 6, 8 and 9).
Strikingly, the grouping of the mutants into three protein-protein association phenotypes correlated perfectly with their respective
phenotypic groupings derived from the one hybrid assay used to define
activation profiles [i.e. proteins (not/weakly) competent for homologous
association were also (not/weakly) competent for activation; compare Table
3
with Table
2
and Fig.
2
]. Thus, one interpretation of these results is that self-association is a pre-requisite for transcriptional activity (or vice versa).
One criticism of the above assay is the use of an artificially fused ad domain
in measuring activity. Indeed if dimerization is a physiologically important
step in the process of Tax
trans
-activation, then it should be possible to demonstrate, under conditions
that favor dimer formation, an increased activity from the activation domain(s)
inherent to Tax (
33
,
57
). Hence, we repeated the previous assay, however, in this instance we assayed
for the activity profile of promoter-bound Gal4bd-Tax when coupled via protein-protein interaction with a non-fused Tax (Table
4
). The prediction is that without the involvement of the strong ad domain, a
Gal4bd-Tax-Tax complex should, nonetheless, have significantly greater
activity than Gal4bd-Tax-alone. The experimental findings were indeed consistent with this
prediction (Table
4
, compare lines 1 and 2). [beta]-Galactosidase activity from Gal4bd-Tax + Tax co-transformed yeast was 152 relative CPRG units, which
was 10 times higher than that from yeast expressing Gal4bd-Tax alone (16 U; Table
4
, line 1) and six to >100 times more than two other sets of controls [Gal4bd
alone and Tax (26 U; Table
4
, line 19) and Gal4bd-p53 and Tax (<1 U; Table
4
, line 17)].
DISCUSSION
Yeast as a simple model to study Tax function
Several models have been suggested to explain Tax function. These include: (i)
indirect tethering of Tax to the promoter via upstream transcription factors
such as CREB (
22
, reviewed in
21
); (ii) direct contact of Tax with basal transcription factors such as TBP (
23
); (iii) induced degradation of inhibitor proteins such as I[kappa]B which results in translocation of NF-[kappa]B into the nucleus (
20
,
58
); (iv) stimulation of DNA binding by enhancing dimerization of the bZIP family
of transcription factors (
24
-
26
); amongst others. The settings in which each of these mechanisms is operative
remain to be clarified fully.
The present study used yeast to examine one mechanistic aspect, specifically the
mechanism of activation by Tax when it is tethered directly to promoter DNA (
33
). We used yeast because many facets of transcription are conserved between
yeast and mammals and because yeast cells are simple to manipulate genetically.
For example, there are many engineered yeast strains that `knockout' or
overexpress certain factors (
59
,
60
) and others that have activated or repressed pathways (
61
,
62
). Hence, future detailed characterizations of specific factor-factor interactions would be possible once the basic system for Tax is
defined.
We found that Tax, when directly targeted to DNA in yeast, capably activated
transcription from a minimal promoter which had no binding sites for known
accessory factors (i.e. CREB, AP-1, SRF, etc.) (Fig.
1
). This recapitulates previous findings in human cells of a direct promoter-proximal activity of Tax on a minimal TATA promoter (
33
) and establishes the relevance of a yeast model for this aspect of
transcription. We did note some slight differences in the activity of a few Tax
mutants in yeast and human cells. However, this could stem from the fact that
the activation assays in yeast measure only the direct promoter-proximal effect of Tax and do not reflect activities in mammalian cells
mediated through other paths (e.g. CREB).
Dimerization as a mechanism that drives Tax function
Perhaps one of the more informative findings from yeast is the suggestion that
Tax oligomerization contributes to optimal promoter-proximal activity. Although we do not exclude the participation of higher
order structures, Tax-Tax dimer formation offers one simple explanation for the experimental
results. Indeed, the idea of dimerization agrees with physical findings from others using Tax-Tax affinity chromatography (Chou-Zen Giam, personal communications). Thus the genetic and biochemical
findings here of Tax-Tax dimerization present a parallel with one of the oft proposed
functions for Tax, stimulation of bZIP protein dimerization (
24
-
26
).
Mechanistically, Tax self-association could guide the dimerization of bZIP and other Tax-interactive proteins (
28
,
28
,
37
,
63
,
64
). In this scenario, a Tax monomer would bind a partner monomer (e.g. CREB) and
then, through the (as yet not understood) process of self-association, Tax would bring partner molecules into close proximity, thus
increasing efficiency of partner dimerization. This would be attractive in
settings where Tax expression is high (e.g. HTLV-I-transformed cells) and the level of Tax-associated factor (e.g. CREB) is relatively limiting.
Functional complementation between two partially active Tax mutants
Independent of the role of Tax dimerization on Tax-associated proteins, evidence suggests that this mechanism contributes
directly to promoter-proximal Tax activity. This is most clearly illustrated by complementation
results in yeast from two weakly active mutants (Figs
5
and
6
). Although there are other possible explanations, one interpretation of
complementation is that Tax and Tax-responsive target interaction is dictated by multiple contacts in which the contact points are distributed asymmetrically on more than one Tax molecule (Fig.
6
). Therefore, coalescence of more than one Tax molecule presents an optimal
activation surface conformation. The proper conformation is dictated by Tax-Tax contact and cannot be substituted by Gal4bd-Gal4bd dimerization (
18
). Thus, although the Gal4 DNA binding domain can dimerize (
18
), the fact that Galbd-Tax Q9-G/Tax Q9-G (Fig.
5
A, lane 5) or Galbd-Tax S132-A/Tax S132-A (Fig.
5
A, lane 9) are functionally suboptimal supports the idea that it is not
di(multi)merization
per se
but the conformation of the protein complex after di(multi)merization which is important for function (Fig.
6
). Consistent with this line of thought, a Tax di(multi)mer would physically
lead to a larger possible array of distinct patterns on the contact surface
than that from a Tax monomer. This could, in part, explain the commonly
observed pleiotropic ability of Tax to activate many different promoters that
share no apparent cognate similarities (reviewed in
27
).
ACKNOWLEDGEMENTS
We thank O.J.Semmes, D.Trinh, L.Derr and A.J.Berk for gifts of plasmids and
R.F.Chun, M.Benkirane, C.Neuveut, H.Xiao and V.Giordano for critical reading of
the manuscript.
REFERENCES
1 Hollsberg,P. and Hafler,D.A. (1993) New Engl. J. Med., 328, 1173-1182.
46 Bartel,P.L., Chien,C.-T., Sternglanz,R. and Fields,S. (1993) In Hartley,D.A. (ed.), Cellular Interactions in Development: A Practical Approach. Oxford University Press, Oxford, UK, pp. 153-179.
