ABSTRACT
Linear TAR RNA has previously been used as a decoy to inhibit HIV-1 transcription
in vitro
and HIV-1 replication
in vivo
. A 48 nucleotide circular RNA containing the stem, bulge and loop of the HIV-1 TAR element was synthesized using the self-splicing activity of a group I permuted intron-exon and was tested for its ability to function as a TAR
decoy
in vitro
. This small circular TAR molecule was exceptionally stable in HeLa nuclear
extracts, whereas a similar linear TAR molecule was rapidly degraded. The TAR
circle bound specifically to Tfr38, a peptide containing the TAR-binding region of Tat. The ability of Tat to
trans
-activate transcription from the HIV-1 promoter
in vitro
was efficiently inhibited by circular TAR RNA but not by TAR circles that contained either bulge or loop mutations. TAR circles did not inhibit
trans
- activation exclusively by binding to Tat since this inhibition was not
reversed by adding excess Tat to the transcription reaction. Together, these
data suggest that TAR circles act as decoys that inhibit
trans
- activation by binding to Tat and at least one cellular factor. These data
also demonstrate the utility of small circular RNA molecules as tools for biochemical studies.
RNA-protein interactions play important roles in a wide variety of biological
processes including transcription, splicing, nucleo-cytoplasmic transport, RNA degradation and translation. The use of RNA
molecules as biochemical tools for studying these processes may be limited by
the rapid degradation of RNA by nucleases present in biological samples.
Compared with linear RNA molecules, circular RNA molecules are often resistant
to nucleases (
1
-
3
) and could potentially function as useful biochemical tools for studying RNA-protein interactions. For example, if a circular RNA molecule has a structure that resembles an endogenous
linear RNA molecule, the circular RNA could be used as a decoy to probe RNA-protein interactions and to study the role of these interactions in
functional assays.
The human immunodeficiency virus type 1 (HIV-1)
trans
-activation response element (TAR) is a 59 nucleotide (nt) RNA stem-loop structure that forms the 5'-end of HIV-1 transcripts (
4
-
6
). TAR consists of a stem, a 4 nt bulge and a 6 nt loop (
7
-
11
). The HIV-1 viral protein Tat, which induces
trans
-activation of HIV-1 transcription by >100-fold
in vivo
(
12
-
14
), binds specifically to the TAR bulge
in vitro
(
15
-
17
). Several experiments suggest that at least one cellular factor binds to the
TAR loop (
18
-
25
). In fact, several cellular factors have been identified that bind to the TAR
loop
in vitro
(
23
,
24
,
26
-
28
), but none of these factors have been clearly shown to regulate HIV-1 transcription.
In vitro
transcription assays using HeLa nuclear extracts have been used to study HIV-1 transcription. In this system, as in
in vivo
transfection systems, Tat induces increased transcription from the HIV-1 promoter, and this increase in HIV-1 transcription depends on the presence of a functional TAR sequence
(
24
,
25
). Linear TAR RNA decoys functioned to inhibit Tat-mediated
trans
-activation in this system (
25
). In contrast, a linear TAR RNA that contained a loop mutation did not inhibit
HIV-1 transcription (
25
), suggesting that loop sequences are required for Tat-mediated
trans
-activation.
In this work, a 48 nt circular RNA molecule that contained stem, bulge and loop
sequences from the HIV-1 TAR element was synthesized using the auto-catalytic splicing of a group I permuted intron-exon. This circular TAR RNA, which was extremely stable in
HeLa nuclear extracts, functioned successfully as a TAR decoy in terms of its
ability to bind specifically to a Tat peptide and in terms of its ability to
specifically inhibit Tat-mediated
trans
-activation
in vitro
. This study demonstrates that stable circular RNA molecules designed to mimic known RNA structures can be useful tools for studying RNA-protein interactions.
The plasmids pTC, pTC-31/34 and pTC-BL were used as templates for the synthesis of TAR circle RNA, 31/34
circle RNA and bulgeless circle RNA, respectively. These plasmids were
constructed by inserting synthesized oligonucleotides containing the sequence 5'-CTAGCCAGATCTGAGCCTGGGAGCTCTCTGG-3' (pTC), 5'-CTAGCCAGATCTGAGCCCAAAAGCTCTCTGG-3' (pTC-31/34) or 5'-CTAGCCAGAGAGCCTGGG-
AGCTCTCTGG-3' (pTC-BL) into the
Nhe
I site of plasmid pPR120 (
3
). The correct sequence and orientation of inserts for each plasmid was
confirmed by DNA sequencing using the dideoxy method (Sequenase, US
Biochemicals). The plasmid pT7-157 was used as a template for synthesis of linear TAR RNA. This plasmid
was derived from pT7-TAR (
27
) but was modified such that a
Hin
dIII site was inserted at position +63 immediately downstream of the TAR stem
sequence. The plasmids pFLBH (
29
) and pBC12/HIV/SEAP (
30
) were used as templates for transcription in HeLa nuclear extracts to produce
the AdML and HIV-1 transcripts respectively. All plasmid DNA was isolated from overnight
cultures by the alkaline lysis method and purified by equilibrium density centrifugation in cesium chloride containing ethidium bromide (
31
).
RNA synthesis was carried out as described by Milligan and Uhlenbeck (
32
) with minor modifications. The plasmids pTC, pTC-31/34, pTC-BL and pT7-157 were digested with
Hin
dIII (New England Biolabs) to create linear templates for
in vitro
transcription using T7 RNA polymerase. T7 RNA polymerase was prepared as described by Grodberg and Dunn (
33
). Transcription reactions containing 40 mM Tris-HCl (pH 8.1), 26 mM magnesium chloride, 1 mM spermidine, 5 mM dithiothreitol, 1 mM ribonucleotide triphosphates, 0.1 mg/ml linear plasmid DNA, 10 [mu]g/ml T7 RNA polymerase, 0.01% Triton X and 0.04% polyethylene glycol
(average molecular weight 8000) were incubated for 4 h at 37oC. In order to increase the efficiency of auto-catalytic splicing of precursor RNA, the reactions were incubated at
42oC for an additional 2 h. Radiolabelled RNA was prepared by including 0.2 nM
[[alpha]-
32
P]UTP (3000 Ci/mmol; New England Nuclear) in the transcription reaction. The
reaction mixtures were then separated by electrophoresis on 12% urea-polyacrylamide gels. Gel slices containing the RNA of interest were
identified by UV shadowing and were cut out of the gel. This RNA was eluted
from the gel slices overnight in a buffer containing 0.1% SDS, 0.5 M ammonium
acetate and 10 mM magnesium acetate, recovered by ethanol precipitation, and
quantitated by measuring the optical density at 260 nm and/or by measuring
radioactivity with a scintillation counter.
Partial hydrolysis of RNA was accomplished by incubating 0.1 pmol of circular or linear TAR RNA in 50 mM sodium bicarbonate (pH 9.0) and 1 mM
EDTA for 3 min at 90oC. The partially hydrolyzed RNA was separated by electrophoresis on a 24%
urea-polyacrylamide gel and was visualized by autoradiography on Hyperfilm
(Amersham).
Radiolabelled circular or linear TAR RNA (0.1 pmol) was incubated with 10 [mu]l HeLa nuclear extract (150 [mu]g protein) in 100 [mu]l reactions containing 50 mM Tris-HCl (pH 7.5), 100 mM sodium chloride, 10 mM magnesium
chloride and 0.1 mg/ml yeast tRNA for 0-12 h. The reactions were stopped by adding an equal volume of 25 mM EDTA
and 80% formamide. Aliquots of the reaction mixtures were then separated by
electrophoresis on a 6% urea-polyacrylamide gel. Bands on the gel were visualized by autoradiography and were quantified on a Molecular Dynamics
PhosphorImager.
Radiolabelled circular TAR RNA (0.5 pmol) was incubated at room temperature for
10 min with 1 pmol of Tfr38 peptide (
17
) in the presence or absence of varying amounts of cold competitor RNA in a
buffer containing 20 mM HEPES (pH 7.9), 10 mM magnesium chloride, 100 mM
potassium chloride, 5% glycerol and 0.5 mg/ml yeast tRNA in a volume of 20 [mu]l. A 4 [mu]l volume of loading buffer containing 50% glycerol, 0.25% bromophenol
blue and 0.25% xylene cyanol was added to each reaction, and then the reactions
were separated by electrophoresis in a non- denaturing 8% polyacrylamide gel (acrylamide to bis-acrylamide ratio of 62:1) using a running buffer containing 45 mM Tris-borate, 45 mM boric acid and 2 mM EDTA. The gel was dried and bands were visualized by autoradiography and
quantified on a Molecular Dynamics PhosphorImager.
HeLa nuclear extracts were prepared as previously described (
34
).
In vitro
transcription reactions were carried out for 30 min at 30oC in a 25 [mu]l volume containing 10 [mu]l nuclear extract (150 [mu]g protein), 14 mM HEPES (pH 7.9), 14% glycerol, 68 mM potassium chloride, 15 mM sodium chloride, 7 mM magnesium chloride, 4 mM sodium citrate, 250 ng poly I-poly C, 300 ng poly dI-poly dC, 1 mM DTT, 10 mM creatine phosphate, 0.1 [mu]M EDTA, 625 [mu]M each of ATP, CTP and GTP, 40 [mu]M UTP, 10 [mu]Ci [[alpha]-
32
P]UTP (3000 Ci/mmol; New England Nuclear), 100 ng HIV-1 template (pBC12/HIV/SEAP cut with
Bam
HI) and 250 ng AdML template (pFLBH cut with
Aat
II). Some reactions also contained 15-1500 ng Tat protein (
35
) and/or 1-10 pmol circular RNA. The circular RNA was the last component added to
the reactions. The reactions were stopped, and newly transcribed RNA was
isolated and separated by electrophoresis on 6% urea- polyacrylamide gels as previously described (
35
,
36
). Bands on the gel were visualized by autoradiography and quantified using a Molecular
Dynamics PhosphorImager.
Group I permuted intron-exons, which undergo auto-catalytic splicing to generate circular products (
3
,
37
), were used to synthesize circular TAR RNA. The sequence from +18 to +44 of HIV-1 TAR was shown by mutational analysis to represent the minimal TAR
sequence for
in vivo
function (
4
). The plasmid pPR120 (
3
), which contained a group I permuted intron-exon sequence, was modified by inserting this minimal TAR sequence into
the exon sequence. After linearization with a restriction endonuclease, this plasmid was used as a template for transcription by T7 RNA polymerase. The linear precursor RNA was expected to undergo self-splicing at 42oC in the presence of magnesium and GTP to yield a circular RNA
product and two linear products (Fig.
1
a). The products of an incomplete splicing reaction are shown in Figure
1
b (lane S). The bands were identified by comparison with auto-catalytic RNA splicing products produced from the parent plasmid (
37
). The identity of the circular RNA product was confirmed by partial alkaline
hydrolysis (Fig.
1
b). Random nicking of linear TAR RNA (TAR) produced multiple products that
appeared as a ladder when separated by electrophoresis. In contrast, random
nicking of circular TAR RNA (TAR Circle) produced a single band that displayed
increased electrophoretic mobility. The expected sequence of the circular TAR
RNA is shown in Figure
1
c. The sequence of the splice junction has been previously determined by primer
extension in circular RNAs produced from the parent plasmid (
37
).
Other circular RNA molecules have been shown to be relatively resistant to
degradation by nucleases (
1
-
3
). If circular TAR RNA is also resistant to nucleases, it could be a very useful
biochemical tool. Therefore, the stability of circular TAR RNA was compared
with linear TAR RNA in HeLa nuclear extracts. Linear TAR RNA was rapidly
degraded (Fig.
2
) with a half life of only 20 min. In contrast, circular TAR RNA was very stable
and was not appreciably degraded after 12 h. Circular TAR RNA is not completely
resistant to nucleases, however, since it is degraded in the presence of serum
(data not shown). The marked stability of circular RNA in nuclear extracts
suggests that the TAR circle would be a very useful tool if it possessed
functional properties of native TAR RNA.
Several studies have shown that the HIV-1 Tat protein binds through its basic domain to the TAR bulge
in vitro
(
15
,
17
,
38
-
40
). A Tat-derived peptide, Tfr38, consisting of the C-terminal 38 amino acids of Tat, was shown to bind to TAR with
similar specificity as the full-length Tat protein (
17
). Circular TAR RNA was examined for its ability to interact with the Tfr38
peptide. As shown in Figure
3
, the Tfr38 peptide bound specifically to radiolabelled circular TAR RNA (TAR
circle) as detected using an electrophoretic mobility shift assay. Using this
assay, the affinity of Tfr38 for circular TAR RNA was equivalent to the
affinity for linear TAR RNA (data not shown). In the experiment shown in Figure
3
, most of the binding by Tfr38 to radiolabelled TAR circle was competed with a
10-fold excess of unlabelled TAR circle (lanes 3-5). A 10-fold excess of a circular TAR RNA containing a mutation in
the loop (31/34 circle) also competed for Tfr38 binding but to a slightly
lesser extent (lanes 6-8). In contrast, a circular TAR RNA in which the bulge was deleted (BL
circle) competed poorly for Tfr38 binding even at a 100-fold molar excess (lanes 9-11). Compared with the TAR circle, the affinities of the 31/34
circle and the BL circle for the Tfr38 peptide were 4-fold and 50-fold lower, respectively. This pattern of binding is consistent with
previously reported binding of Tat to linear TAR in which bulge mutations
abolished Tat binding while loop mutations had little or no affect (
15
,
41
-
45
). These data suggest that the TAR circle may have bulge and loop structures
similar to native TAR RNA.
Linear TAR RNA decoys have been shown to inhibit Tat-mediated
trans
-activation in an
in vitro
system (
25
). In order to determine if circular TAR RNA could also inhibit Tat-mediated
trans
- activation, circular TAR RNA was added to
in vitro
transcription reactions containing HeLa nuclear extracts. As seen in Figure
4
a, transcription from the HIV-1 promoter increased when exogenous Tat was added to the reaction while transcription from the adenovirus
major late (AdML) promoter remained relatively unchanged (lanes 1 and 2). This
specific increase in HIV-1 transcription was inhibited by the addition of TAR circle RNA to the
reaction, and the amount of inhibition increased as the amount of TAR circle
was increased from 1 to 10 pmol (lanes 3-5). In contrast,
trans
-activation was not inhibited by 31/34 circle RNA (lanes 6-8) or bulgeless circle RNA (lanes 9-11). Quantification of three independent transcription
experiments revealed similar results (Fig.
4
b). The TAR circle inhibited the Tat-mediated increase in HIV-1 transcription by 77.4%, whereas the 31/34 circle and the bulgeless
circle had no significant effect. These data suggest that bulge and loop
structures are both required for TAR to function as a decoy in this assay.
Basal HIV-1 transcription in the absence of Tat was not affected by TAR circle,
31/34 circle, or bulgeless circle RNA (Fig.
5
a and data not shown).
A 48 nt circular RNA containing the stem, bulge and loop of HIV-1 TAR RNA was synthesized and examined to determine its utility as a
biochemical tool. This circular TAR RNA was found to be extremely stable in
HeLa nuclear extracts while a similar linear TAR RNA molecule was rapidly
degraded. Although the mechanism for this increased stability has not been
defined, the circular RNA does not have ends that are accessible to
exonucleases. It is also possible that the extensive secondary structure of the
circular TAR RNA could contribute to its stability.
Circular TAR RNA functioned as a TAR decoy by binding specifically to the Tfr38
peptide. The binding of Tfr38 to the circular TAR RNA appeared to involve the
bulge since deletion of the bulge reduced the relative binding affinity by 50-fold. The relative affinity of Tfr38 binding to a circular TAR RNA in
which the loop sequence from +31 to +34 was altered (Fig.
1
c) was also determined. This particular mutation in the TAR loop was chosen
because the equivalent mutation
in vivo
nearly abolished Tat-mediated
trans
-activation (
15
,
46
,
47
). The 31/34 circle had a similar but slightly lower affinity (~4-fold) for Tfr38 binding compared with the TAR circle. This result
differs slightly from the result of Roy
et al
. (
15
) who found that linear TAR RNA and linear 31/34 RNA have the same relative affinities for Tat. Roy
et al
., however, examined only a single point rather using titrated amounts of cold
competitor RNA, and they could have missed small differences in relative
affinities. It is unclear whether the slight decrease in the relative affinity
resulting from the 31/34 loop mutation seen in these
in vitro
binding experiments could explain the dramatic decrease in
trans
-activation caused by the 31/34 mutation
in vivo
. Overall, the results reported here are consistent with previous reports that
mutation of the TAR bulge nearly abolished Tat binding while mutation of the
TAR loop had little or no effect (
15
,
17
,
42
-
45
,
48
) and suggest that the structure of the bulge and loop in circular TAR RNA is
similar to linear TAR RNA.
We thank K. Weeks and D. Crothers for the kind gift of the Tfr38 peptide. We also thank C. Suñé, Z. Pasman and M. Velaz-Faircloth for their critical reading of this manuscript. The Keck
Foundation is acknowledged for their generous support of the Levine Science
Research Center at Duke University, where much of this work was performed.
P.R.B. was supported by a Howard Hughes Medical Institute Postdoctoral
Fellowship for Physicians. R.A.C. was a predoctoral fellow in the MSTP at Duke
University. This work was supported by grants from the VA to M.A.G.-B. and from the NIH to M.D.B.
REFERENCES
Return