Optimal Tat-mediated activation of the HIV-1 LTR promoter requires a full-length TAR RNA hairpin
Optimal Tat-mediated activation of the HIV-1 LTR promoter requires a full-length TAR RNA hairpin
Koen
Verhoef
,
Marieke
Tijms
and
Ben
Berkhout*
Department of Human Retrovirology, Academic Medical Center, University of
Amsterdam, Meibergdreef 15, 1105 AZ
Amsterdam
,
The Netherlands
Received October 25, 1996
;
Revised and Accepted December 2, 1996
ABSTRACT
HIV-1 transcription from the LTR promoter is activated by the viral Tat
protein through interaction with the nascent TAR RNA hairpin structure. The
mechanism of Tat-mediated transcriptional activation has been extensively investigated with
LTR-CAT reporter genes in transient transfections and, more recently, in
infection experiments with mutant HIV-1 variants. Several discrepancies between these two assay systems have
been reported. For instance, whereas opening of the lower part of the TAR RNA
stem does not affect the promoter activity of an LTR-CAT plasmid in transient assays, the corresponding virus mutant is fully
replication-impaired. With the aim to resolve this controversy, we have examined the
activity of a set of TAR RNA mutants in transient transfection experiments with
a variety of cell types. We now demonstrate that truncated TAR motifs exhibit a
severe, but cell-type dependent transcription defect. Whereas full LTR activity is measured
in COS cells that have been used regularly in previous transfection assays, a
severe defect is apparent in a variety of human cell lines, including T cell
lines that are typically used in HIV-1 replication studies. These results suggest the presence of a human
protein that participates in Tat-mediated transcriptional activation through binding to the lower part of
the TAR stem. Several candidate co-factors have been reported in literature. This study resolves the
discrepancy between transfection and infection studies on the requirements of
the lower TAR stem structure. The evidence also implies that LTR transcription
studies should be performed preferentially in human cell types.
INTRODUCTION
The pathogenic human immunodeficiency virus type 1 (HIV-1) encodes not only the three structural genes (
gag
,
pol
and
env
) common to all retroviruses but also six unique gene products, including the
Tat protein (
1
). This protein potently activates transcription from the viral long terminal
repeat (LTR) promoter. Tat gains access to the DNA promoter region by binding
to the
trans
-acting-responsive RNA element (TAR) that folds into a stable stem-loop structure as part of the nascent transcript. The HIV-1 Tat-TAR axis has been one of the most intensively
investigated viral regulatory mechanisms (see reviews,
2
,
3
). Extensive mutational analyses have identified the active domains in both the
Tat protein and the TAR RNA element. The initial mutagenesis studies on the TAR
RNA motif were done primarily in transiently transfected cells containing a Tat
expression plasmid and a second plasmid encoding a reporter gene under
transcriptional control of the HIV-1 LTR promoter (
4
-
9
). Furthermore, appropriate cell-free assay systems have been developed to study the Tat-TAR interaction and the mechanism of transcriptional activation (
10
-
17
). More recently, replication studies with mutant HIV-1 variants demonstrated that both Tat protein (
18
-
22
) and its TAR RNA target sequence (
23
-
27
) are essential for viral replication, although some level of TAR-independent replication has been reported in activated T lymphocytes (
28
) and in astrocytic glial cells (
29
).
Several transient transfection studies suggested that the integrity of the TAR
stem, in particular of the upper domain including the single-stranded bulge and loop elements, is important for efficient
transcriptional activation by Tat (
8
,
9
,
30
). The identity of the base pairs surrounding the tri-nucleotide bulge was found to be critical both for efficient
trans
-activation (
31
) and high-affinity binding of the Tat protein (
32
). In comparison, no or relatively moderate defects were scored for mutations in
the lower stem domain of TAR (
5
,
31
). For instance, the detailed mutational analysis by Jakobovits
et al
. (
5
) in human epithelial 293 cells reported 34-39% LTR activity for TAR variants with a triple base substitution in the
+7/+18 region of the stem. This defect is relatively minor compared with triple
nucleotide substitutions in either the single-stranded bulge or loop element [6 and 2% LTR activity, respectively (
5
)].
We reported previously a discrepancy between transient transfection assays and
virus infectivity studies with TAR-mutated HIV-1 LTR promoter motifs (
23
). Specifically, we found that a mutant HIV-1 virus with base substitutions in the lower TAR stem is replication-incompetent, although the same TAR mutant is fully transcriptionally
active in transient transfections with an LTR-CAT (chloramphenicol acetyltransferase) reporter construct (
31
). These combined results may suggest an additional role of the TAR hairpin
structure in the virus replication cycle, but the observed difference in TAR
requirement may also reflect a variation in transcription in these two experimental settings. For instance, the
difference may reflect transcription from an integrated versus an unintegrated
LTR promoter in the infected and transfected cells, respectively (
33
). Furthermore, whereas infections are routinely performed in CD4
+
T cells, transfections are frequently performed in the non-T cell line HeLa or the non-human cell line COS. Here, we present evidence that a full-length TAR element is required for efficient LTR-transcription in human cell types, but this effect is
less obvious in other cells.
MATERIALS AND METHODS
DNA constructs
Details on the LTR-CAT plasmids used in this study can be found in previous publications (
31
,
34
); the wild-type LTR-CAT contains the complete HIV-1 LTR U3 region upstream, and the
cat
reporter gene downstream of TAR (cloned in the
Hin
dIII site at position +78). Nucleotide numbers refer to the position on the HIV-1 RNA genome, with +1 being the transcriptional start site. The revertant
TAR motifs that restore replication of the Xho+10 mutant virus were described
previously (
23
). The proviral sequences were subcloned into LTR-CAT by exchange of the
Pvu
II-
Hin
dIII fragment (position -22/+76). The Tat-expression plasmid pcDNA3-Tat contains the Tat-Rev genomic segment of the HIV-1 isolate pLAI (
35
) inserted downstream of the cytomegalovirus immediate early promoter in pcDNA3
(Invitrogen). Details of this vector will be described elsewhere [Verhoef and
Berkhout, in preparation, see also (
36
)].
Cell culture, DNA transfection and CAT enzyme analysis
The lymphocytic T cell lines (SupT1, A3.01, MT2, C8166) were maintained at 37oC and 5% CO
2
in RPMI 1640 medium containing 10% fetal calf serum (FCS). Adherent cell lines
(COS, HeLa) were grown in Dulbecco's modified Eagle's medium with 10% FCS.
Transient transfections were performed with DEAE-dextran [COS and HeLa cells,
see (
7
)] or by electroporation [all T cell lines, see (
37
)]. Co-transfection of COS cells (~70% confluency on 60 mm dishes) was performed with 1 [mu]g LTR-CAT and 0.1 [mu]g pcDNA3-Tat. HeLa cells (~70% confluency on 60 mm dishes) were transfected
with 1 [mu]g of each plasmid, and all T cell lines (5 * 10
6
) were electroporated with 2 [mu]g LTR-CAT and 5 [mu]g pcDNA3-Tat plasmid. We did verify that these Tat levels are within the
linear range of LTR activation. To measure basal promoter activity, we used 30 [mu]g LTR-CAT to transfect SupT1 cells. Cell extracts were prepared at day 3 post-transfection and assayed for CAT activity with butyryl-CoA in combination with the phase-extraction method as described (
38
). CAT activities were quantitated in the linear range of the reaction.
RESULTS
Truncated TAR hairpins are transcriptionally defective in human cells
The lower TAR stem, but not the loop or bulge elements, function in a cell type-specific manner
Although many HIV-1 functions have been reported to contribute to viral replication in a
cell type-specific manner, such an effect has not been described for Tat/TAR-mediated LTR transcription. Three explanations can be proposed for
this cell-type dependent phenotype. First, it is possible that a cellular factor,
involved in Tat-mediated
trans
-activation through binding to the lower TAR stem, is lacking or inactive
in COS cells. Second, LTR transcription in COS cells may not require co-factors that bind the lower TAR stem. Third, the mechanism of LTR-transcription is a multistep process and it is possible that the
rate of this process is controlled in COS cells at a slow, rate-limiting step that is not apparent in human cells. In all three scenarios,
opening of the TAR stem will not influence the transcription rate in COS cells.
If a particular rate-limiting step determines the level of LTR transcription in COS cells,
other HIV-1 mutants defective in LTR-transcription should also be less overt in this cell type. We
therefore tested the cell-type specificity for other TAR mutants and the partially defective Tat
protein mutant Y26H. Two additional TAR mutants were tested, either with
substitutions in the single-stranded bulge or loop domain (mutants B123 and L135, respectively, see
legend to Figure
3
for further details on the substitutions in these mutants). Their
transcriptional activity was scored upon transfection of COS and SupT1 cells
(Fig.
3
). Unlike the Xho+10 stem-mutant that was defective exclusively in SupT1 cells, these TAR mutants
were equally defective in the two cell types. Furthermore, the Y26H Tat mutant
demonstrated a similarly reduced activity in both cell types (Fig.
3
). These results suggest that the cell-type dependent phenotype is rather specific for lower TAR stem mutants,
consistent with the first two models.
Repair of the truncated TAR stem restores transcriptional activity
The analysis so far indicates that an extended RNA hairpin structure is
important for optimal TAR function in human cells. However, we cannot exclude
alternative mechanistic explanations. For instance, the inactivity of the
Xho+10/5's and 5'd mutants in human cells may reflect a sequence-specific binding of a human co-factor to the +3/+16 TAR sequences, either in the RNA
or DNA form. When the Xho+10 mutant virus was used to select for faster
replicating revertants, we observed restoration of base pairing of the lower
TAR stem by acquisition of additional mutations (
23
). Several of these TAR revertant structures, obtained in two independent
reversion experiments, were now tested in the transient LTR-CAT assay. Three consecutive TAR variants of reversion route I (Fig.
4
, upper pathway) and the final structure that evolved in route II (lower
pathway) were subcloned into the LTR-CAT plasmid to measure their transcription properties.
In route I, base pairing is gradually restored in the consecutive TAR samples by
acquisition of mutations (Fig.
4
; Xho+10 mutant, 11 bp; revertants Ib12, 16 bp; Ib1, 17 bp; Id15, 21 bp).
Whereas the Xho+10 mutation dramatically reduced the level of Tat-activated LTR transcription in SupT1 cells, the TAR revertants improved
the expression level step-by-step and the final ID15 variant produced wild-type levels of CAT enzyme (Fig.
5
A). The latter TAR mutant corresponds with a fast replicating revertant virus (
23
). Thus, the increased LTR expression levels correlated perfectly with the
stability and length of the TAR RNA stem. Combined with the results obtained
with the initial set of TAR mutants (Fig.
1
), these results suggest a correlation between the length of the TAR duplex and
its transcriptional activity in human cells.
To further corroborate these results we tested this set of TAR mutant-revertants in two other T cell lines, A3.01 and MT2, and we used COS cells
to provide a cellular milieu in which the Xho+10 defect is not manifest. The
results of several transfections are summarized in Figure
6
and clearly demonstrate the cell-type specific defect of mutant Xho+10 and the subsequent recovery of
transcriptional activity in route I revertants. To test whether the
transcriptional defect/repair was specific for Tat-mediated transcription from the LTR promoter, we tested this same set of
mutant-revertant TAR motifs in the absence of the Tat
trans-
activator protein. We found that the basal LTR promoter activity was not
sensitive either to opening of the TAR stem in Xho+10 or to the subsequent
repair of this motif in the three TAR revertants of pathway I (Fig.
5
B).
DISCUSSION
We suggested earlier that TAR RNA may have a role in the HIV-1 replication cycle separate from its contribution to transcription (
23
). This idea originated from an apparent discrepancy in the activity of a
particular TAR mutant in two experimental systems. Specifically, a truncated
TAR RNA hairpin was transcriptionally competent in transient transcription
assays, but the corresponding virus mutant was replication-incompetent. The evidence presented in this study resolves this issue.
Opening of the bottom part of the TAR stem inhibited the transcription function
of the HIV-1 LTR promoter in human T cells, the cell type ordinarily used in HIV-replication studies. Previous transfection studies overlooked this
transcription defect because it is cell-type dependent and not observed in COS cells. Thus, the integrity of the
TAR RNA stem is critical for efficient Tat-mediated LTR transcription and the first definition of the borders of the
minimal TAR domain [position +19 to +42, (
34
)] should be modified accordingly.
The cell-type specific behaviour of the lower TAR stem mutant is rather unique in
that other transcriptionally defective TAR mutants display no such cell-type differences. This argues against a general phenomenon, for instance
caused by a particular rate-limiting step during LTR-transcription in COS cells. We therefore favour a more direct
explanation in which the lower TAR stem contributes in a cell-type specific manner to the transcription process by binding of a co-factor. LTR transcription in COS cells does either not require such
a lower TAR stem binding factor, or this co-factor is absent from COS cells. Several TAR RNA-binding factors have been reported (
42
-
55
), but only those proteins that require an extended TAR stem for binding can
explain the defect of TAR mutants such as Xho+10. The revertant data suggest
that binding of this cofactor occurs without strict specificity for the
sequence of the lower stem region. We cannot exclude an alternative mechanism,
with a COS cell-specific factor that binds to the upper TAR region and that can overcome
the requirement for an extended TAR stem.
The present study demonstrates that full-length TAR is required for optimal LTR transcription, thereby explaining
the severe replication defect of HIV-1 variants with a truncated TAR stem. However, this does not rule out any
additional role for this sequence/structure motif in the viral replication
cycle, and several putative functions have been proposed [reviewed in (
56
)]. In this respect, we note that TAR is part of the repeat (R) region of the
LTR that encodes a double hairpin motif [TAR and the `polyA-hairpin', (
57
)] that is present at both the extreme 5' and 3' ends of all HIV-1 transcripts. These structures may actively participate in
one of the many functions of the HIV-1 leader RNA that involve multiple RNA-RNA and RNA-protein contacts, e.g. RNA packaging, dimerization or
reverse transcription. Interestingly, a recent study proposed a role for TAR
RNA in the process of initiation of reverse transcription (
58
). In this process, a cellular tRNA primer that is annealed to the downstream
primer-binding site (PBS) is extended by the viral reverse transcriptase enzyme
over the upstream leader sequences, including the TAR element. Alternatively,
relatively simple functions can be considered for the 5'- and 3'-terminal tandem hairpin motif. For instance, the
structures may protect the viral RNA from degradation in the infected cell or
virion particle.
Finally, this study underscores the concept that viral mechanism should be
studied in experimental systems that model as closely as possible natural HIV
infections. There is a growing list of controversies in HIV-1 research that are caused primarily by a difference in cell type used in
the experiments. For instance, several HIV-1 accessory gene products are required for efficient virus replication in
some cell types, but not in others (
59
) and the efficiency of reverse transcription was shown to differ significantly
in T cell lines versus primary cells (
60
,
61
). This study reveals that the mechanism of Tat/TAR-mediated transcription should be performed preferentially in human T cell
lines.
ACKNOWLEDGEMENTS
We thank Bep Klaver for providing the Xho+10 revertant plasmids and Atze Das for
critical reading of the manuscript. This work was supported by the Dutch AIDS
Foundation.
57 Berkhout,B., Klaver,B. and Das,A.T. (1995) Virology207, 276-281.MEDLINE Abstract
58 Harrich,D., Ulich,C. and Gaynor,R.B. (1996) J. Virol. 70, 4017-4027.MEDLINE Abstract
59 Subramanian,R.A. and Cohen,E.A. (1994) J. Virol. 68, 6831-6835.
60 O'Brien,W.A., Namazi,A., Kalhor,H., Mao,S.-H., Zack,J.A. and Chen,I.S.Y. (1994) J. Virol. 68, 1258-1263.
61 Back,N.K.T., Nijhuis,M., Keulen,W., Boucher,C.A.B., Oude Essink,B.B., van Kuilenburg,A.B.P., Van Gennip,A.H. and Berkhout,B. (1996) EMBO J.15, 4040-4049.
