ABSTRACT
During the initial infection of B lymphocytes by Epstein-Barr virus (EBV) only a few viral genes are expressed, six of which
encode the EBV nuclear antigens, EBNAs 1-6. The majority of EBNA mRNAs share common 5
'
-ends containing a variable number of two alternating and repeated exons
transcribed from the
Bam
HI W major internal repeats of the viral DNA. These sequences can also exist as
independent small RNA species in some EBV-infected cell types. We present evidence that transcripts from these W
repeat regions can exert a
trans
-acting effect on protein synthesis, through their ability to activate the
dsRNA-dependent protein kinase PKR. UV cross-linking and filter binding assays have demonstrated that the W
transcripts bind specifically to PKR and can compete with another EBV-encoded small RNA, EBER-1, which was shown previously to bind this kinase. In the
reticulocyte lysate system the W RNAs shut off protein synthesis through an
ability to activate PKR. In contrast to EBER-1, the W RNAs are unable to block the dsRNA-dependent activation of PKR. Using a purified preparation of the
protein kinase we have shown that the W transcripts directly activate PKR
in vitro
. The results suggest that EBV has the ability both to activate and to inhibit
PKR through the actions of different products of viral transcription.
Epstein-Barr virus (EBV) is linked with a number of human malignancies, including
Burkitt's lymphoma, nasopharyngeal carcinoma and lymphoproliferative diseases
in immunocompromised patients.
In vitro
, EBV is able to efficiently immortalise normal human B lymphocytes and in the
majority of the infected cells the virus enters into a latent state in which
only a few viral genes are expressed (
1
). These genes can encode up to six nuclear antigens (EBNAs 1-6), three membrane proteins (LMP 1, 2A and 2B) and two small RNAs (EBER-1 and -2) (
2
). Since the coordinated action of these gene products appears to be sufficient
to promote cell proliferation and immortalisation, it is important to
understand their biological effects and the mechanisms by which their
expression is regulated.
The EBNAs are translated from mRNAs which arise from complex primary transcripts
that span a large portion of the viral genome. Two distinct promoters that can
direct synthesis of such RNAs have been identified, located in the
Bam
HI C unique region and the
Bam
HI W major internal repeat sequence of the viral genome respectively (
3
-
6
). The messages for each individual EBNA are generated by differential splicing
and these RNAs can have a number of variations of exon structure at their 5'-ends (
7
-
9
). Additional promoters for the synthesis of transcripts encoding EBNA-1 have also been identified in the
Bam
HI F and Q regions of the genome (
10
-
14
). With the exception of this last case, all the EBNA messages contain near
their 5'-ends multiple copies of a sequence encoded by a pair of exons, W1
(66 nt) and W2 (132 nt), within each
Bam
HI W repeat. These W1W2 exon pairs usually occur within the 5' untranslated regions (5' UTRs) of the EBNA mRNAs, although in the case of EBNA-5 (EBNA-LP) a different splicing pattern creates an upstream
AUG which permits translation of the W repeats (
15
). The significance of the W repeat sequences in the 5' UTRs of mRNAs encoding the EBNAs is unclear, but these regions are GC-rich and may adopt stable secondary structures (
16
). Such features in eukaryotic cell mRNAs are known to reduce the efficiency of
translation of downstream open reading frames (
17
,
18
).
The significance of transcription of the W repeat region of the EBV genome is
further complicated by the evidence for the presence of small RNA species
containing W exons in cell lines latently infected with the virus (
16
,
19
). These small RNAs are of unknown function, but may act
in trans
to regulate gene expression at the post-transcriptional level.
The interferon-inducible protein kinase PKR (also known as DAI and p68) is intimately
involved in the regulation of protein synthesis in response to viral infections
(
20
,
21
). This enzyme is dependent for its activation on double-stranded RNA (dsRNA), which can be produced by symmetrical transcription
of viral genomes. Extensive stem-loop structures in mRNAs may also activate PKR (
22
,
23
). The computer-predicted secondary structure of the W1W2 exon pair transcribed from the
Bam
HI W repeat sequence of EBV indicates such a possible hairpin structure (Fig.
1
). Indeed, limited RNase cleavage of end-labelled W1W2 transcripts confirms that this RNA adopts a complex
secondary structure (data not shown). These data suggested to us that this RNA
sequence may have the ability to bind to PKR and activate the protein kinase.
Moreover, the presence of the W-encoded RNA sequence at the 5'-ends of the EBNA mRNAs is reminiscent of the situation with
HIV-1 RNAs, where the TAR sequence is a common 5'-element (
24
). TAR has been shown to bind to PKR and regulate the activity of the enzyme,
although there has been some controversy over whether PKR is activated or
inhibited by TAR (
25
-
29
).
HeLa cells were grown in spinner culture in RPMI 1640 (Gibco) supplemented with 10% heat-inactivated fetal calf serum (Imperial Laboratories) and diluted every two or three days to between 1.5 and 4 * 10
5
cells/ml. Wild-type PKR was purified from a ribosomal salt wash obtained from 40 l of
cells that had been treated with human interferon [alpha](1000 U/ml for 24 h) (
30
). The protein was subjected to chromatography on DEAE-cellulose (where it elutes in the flow-through fraction), followed by chromatography on phosphocellulose.
Proteins were eluted with a 50-500 mM KCl gradient and the PKR-containing fractions (eluting at ~250 mM KCl) were identified by Western blotting. All buffers used during purification contained aprotinin (1 [mu]g/ml), leupeptin (1 [mu]g/ml), pepstatin (1 [mu]g/ml) and PMSF (1 [mu]M). The K296R mutant of PKR was purified from
insect cells infected with a recombinant baculovirus expressing this protein as
described previously (
31
,
32
).
Plasmids containing various numbers of copies of the W1W2 exon pair, cloned in
the pSP64 transcription vector, were kindly provided by Prof. Paul Farrell
(Ludwig Institute for Cancer Research, London). These were modified to remove a
region of the multiple cloning site lying between the SP6 RNA polymerase
promoter and the W1W2 sequences. Plasmids containing one, two or seven W1W2
repeats were generated and shown to be suitable for transcription of the
corresponding RNAs using SP6 polymerase. Plasmids encoding the EBV small RNA
EBER-1, derived from that described by Clarke
et al.
(
33
), and a
Xenopus laevis
tRNA
Phe
precursor (a gift from Dr G.Pruijn, University of Nijmegen, The Netherlands)
were transcribed with T7 RNA polymerase as previously described (
34
).
SP6 and T7 RNA polymerases were purchased from Cambio and poly(I)[middot]poly(C) and 2-aminopurine from Sigma Chemical Co. Radiochemicals were from ICN
Flow or Amersham International. Reticulocyte lysates were a kind gift from Drs
Simon Morley and Jenny Pain (University of Sussex, Falmer, UK).
Small RNAs were synthesised by
in vitro
transcription from recombinant plasmids using either SP6 or T7 RNA polymerases
(
35
). Plasmids were linearised with the appropriate restriction enzymes such that
there would be no extraneous sequences at the 3'-ends of the transcripts and the templates were gel purified before
use. Transcription was performed in a final volume of 100 [mu]l for 3 h at 37oC. The DNA template was removed by adding 15 U DNase I at 37oC for 15 min and the RNA was phenol extracted and precipitated
with isopropanol.
Radiolabelled transcripts were produced using a similar protocol except that
unlabelled UTP was omitted from the reaction and replaced with 50 [mu]Ci [[alpha]-
32
P]UTP. Labelled RNA was quantified by precipitating 1-2 [mu]l of a dilution of the transcript with cold 10% trichloroacetic acid,
0.5% sodium pyrophosphate, followed by scintillation counting.
RNA preparations synthesised
in vitro
often include a minor population of dsRNA contaminants (
33
,
36
). In experiments involving assays of PKR activation it is therefore often
necessary to purify transcripts to remove such contaminants and this was
carried out as described by Mellits
et al.
(
36
). Approximately 150-300 [mu]g RNA were loaded onto a 5% denaturing polyacrylamide gel using
formamide loading buffer as previously described. After electrophoresis the RNA
was detected by UV shadowing, eluted for 3 h at 37oC into buffer (0.5 M NaCl, 5 mM EDTA, 80 mM HEPES, pH 7.6) containing 1200 U/ml RNAguard (Pharmacia) and ethanol
precipitated with 20 [mu]g glycogen as carrier. The eluted RNA was then run on a 5% non-denaturing polyacrylamide gel and recovered using the same procedure
as before.
UV cross-linking of RNA samples to purified PKR was carried out as described
previously (
37
). For convenience the K296R mutant form of PKR was used in some experiments, as
this protein can be easily produced from insect cells infected with a
recombinant baculovirus (
31
,
32
). Although this mutant has lost all protein kinase activity it retains the RNA binding characteristics of the wild-type enzyme (reviewed in
21
).
32
P-Labelled RNA (10
5
c.p.m.) was incubated with PKR preparations for 15 min at 30oC in the presence of 80-100 mM KCl, 10 mM Tris-HCl, pH 7.5, in a final volume of 25 [mu]l. RNA was cross-linked to protein by irradiation at 254 nm for 5
min at 4oC. The samples were then made 0.5% in
N
-laurylsarcosine. Each sample was treated with 20 U RNase T
1
and 20 [mu]g RNase A for 1 h at 37oC and analysed by SDS gel electrophoresis. Cross-linked complexes were identified by autoradiography.
The formation of RNA-protein complexes was also assayed by retention of radioactivity on
cellulose nitrate filters. Varying concentrations of RNA were incubated with
mutant PKR in the presence of 10 mM Tris, pH 7.6, 100-150 mM KCl and 0.8 mM Mg acetate for 15 min as described previously (
34
). Labelled RNAs were present at 10
4
-10
5
c.p.m. When competition assays were performed a second non-radioactive RNA was also included at increasing molar excess over the
labelled species.
Endogenous protein synthesis was measured in rabbit reticulocyte lysates by the incorporation of L-[
14
C]leucine into acid-insoluble material, in the presence or absence of the synthetic dsRNA
poly(I)[middot]poly(C) (0.1 [mu]g/ml) (
33
,
38
). Reactions were incubated at 30oC for 1 h and triplicate aliquots removed at the end of the incubation. Hot
trichloroacetic acid-insoluble radioactivity was measured by liquid scintillation counting.
Where indicated the PKR inhibitor 2-aminopurine was included at a concentration of 10 mM.
Activation of PKR was assessed by autophosphorylation of the protein kinase in
the presence of [[gamma]-
32
P]ATP. Incubations (20 [mu]l) contained: 10 mM Tris-HCl, pH 7.5, 100 mM KCl, 2 mM MnCl
2
, 0.1 mM EDTA, 20%(v/v) glycerol and 4.5 [mu]M [[gamma]-
32
P]ATP (5-10 [mu]Ci). Poly(I)[middot]poly(C) or other RNA species were added as indicated in the
figure legends. After 20 min at 30oC proteins were denatured with SDS gel sample buffer and separated by
electrophoresis on 15% polyacrylamide-SDS gels. The dried gels were analysed by autoradiography.
Since W exon transcripts are widely expressed in cells after infection with EBV
and are present at the 5'-ends of several EBNA-encoding mRNAs, as well as existing as small cytoplasmic RNA
species (
16
,
19
), we have investigated the possibility that these W sequences may have a
trans
-acting function in the cytoplasm of EBV-infected cells. Structural analysis of the W1W2 RNA suggests that
this GC-rich sequence has extensive secondary structure (Fig.
1
and data not shown). We therefore examined whether the small W RNAs can
interact with the dsRNA-dependent protein kinase PKR. This enzyme shows specific and tight binding
of a number of small viral RNAs, including VA
I
RNA of adenovirus (
27
,
39
,
40
), the TAR RNA of HIV-1 (
26
,
28
) and EBER-1 and EBER-2 of EBV (
34
,
37
).
To assess their ability to bind to PKR, RNA species containing one, two and
seven copies of the 198 nt W1W2 repeat sequence were synthesised by
in vitro
transcription using SP6 RNA polymerase. For direct binding studies each small
RNA was labelled by transcription in the presence of [[alpha]-
32
P]UTP, incubated with purified preparations of wild-type or K296R mutant PKR and subjected to UV cross-linking as described in Materials and Methods. Figure
2
shows that, on analysis by SDS gel electrophoresis, all three W1W2 RNAs were
found to be cross-linked to both forms of the 68 kDa PKR protein. The (W1W2)
7
RNA gave a somewhat stronger signal than the W1W2 and (W1W2)
2
species. As reported previously (
34
,
37
), the EBV-encoded small RNA EBER-1 could also be cross-linked to wild-type PKR and the K296R mutant (although cross-linking to the wild-type is weak in Fig.
2
A), but a tRNA transcript could not be cross-linked (Fig.
2
C). These data indicate that, like the other small viral RNAs described above,
the W1W2 RNAs can bind to PKR
in vitro
. As has been clearly established in the case of other ligands, the protein
kinase activity of PKR is not required for RNA binding.
We next investigated whether the binding of the W repeat RNAs to PKR has any
functional significance for the activity of the protein kinase. When activated
by dsRNA PKR undergoes autophosphorylation and then phosphorylates the [alpha] subunit of protein synthesis initiation factor eIF2 at position Ser51 (reviewed in
21
). This inhibits polypeptide chain initiation by blocking the activity of the
guanine nucleotide exchange factor eIF2B (
41
). The reticulocyte lysate cell-free translation system contains endogenous PKR and is a suitable assay
system for examining the effects on protein synthesis of RNA molecules that
regulate this protein kinase (
28
,
33
). The effects of gel-purified W1W2 RNA in this system were assayed in the presence and absence
of the synthetic PKR activator poly(I)[middot]poly(C) (Fig.
5
A). In the absence of poly(I)[middot]poly(C) the W RNA caused a marked shut-off of protein synthesis. The maximum level of inhibition was
identical to that caused by poly(I)[middot]poly(C) itself, with 50% of maximum being achieved at a concentration of
between 0.1 and 1 [mu]g/ml (although this was somewhat variable between experiments; compare Fig.
5
A and B). In the presence of poly(I)[middot]poly(C) at a concentration sufficient for maximal activation of PKR the
W RNA had no further effect on overall protein synthesis, suggesting that even
at concentrations as high as 100 [mu]g/ml the W RNA is unable to block activation of PKR by dsRNA. Inhibition of
protein synthesis in the reticulocyte lysate by the W1W2 RNA, like that caused
by poly(I)[middot]poly(C), was reversed by the PKR inhibitor 2-aminopurine (Fig.
5
B). This suggests that the effect of the W RNA is mediated by PKR activation
rather than being a consequence of a non-specific inhibition of translation.
In this paper we have shown that transcripts encoded by the
Bam
HI W region of the major internal repeat (IR1) of the EBV genome bind to the
interferon-inducible protein kinase PKR and induce autophosphorylation and activation
of the enzyme. Binding to both wild-type PKR and a recombinant mutant form of the protein which is
catalytically inactive has been demonstrated. Competition experiments using
other PKR ligands and RNA species that do not bind to the protein kinase have
demonstrated that the W repeat RNAs bind in a specific manner. The data suggest
that this binding may occur at the same site(s) that bind(s) dsRNA activators
and other RNA ligands, i.e. the dsRNA binding domains near the N-terminus of the protein (
40
,
45
), although this has not been tested directly. Two types of assay, namely
autophosphorylation of purified wild-type PKR and inhibition of protein synthesis in the reticulocyte lysate,
have shown that gel-purified W transcripts can function as activators of the RNA-dependent protein kinase.
Several studies have demonstrated the complexity of the patterns of
transcription and splicing of the
Bam
HI W internal repeats during EBV infection (
7
-
9
). Although the W exons are present close to the 5'-ends of mRNAs for all the EBNA proteins, except in some situations
EBNA-1 (
10
,
11
), in the majority of cases they are not translated into protein but constitute
a major part of the 5' UTRs of these mRNAs. However, in the case of EBNA-5 (EBNA-LP) the W1W2 exon repeats code for a repeated amino acid
sequence, as a consequence of the creation of an upstream AUG by a different
splicing event (
15
).
Previous evidence from our laboratory suggested that, in addition to the
presence of the W exons in the EBNA mRNAs, these sequences could be found in a
family of small RNAs expressed in relatively larger amounts in the cytoplasm of
EBV-infected cells (
16
). The exact nature of these small W RNAs remains to be established, but the
presence of several species with size differences which are multiples of ~200 nt (
16
) suggested that they may contain multiple numbers of W1 and W2 exon pairs
(consisting of 198 nt/repeat;
8
). Alfieri
et al.
(
19
) also detected a 200 nt RNA species which was present in both acutely EBV-infected lymphocytes and latently infected IB-4 cells and which hybridized to an EBNA-5 probe containing W repeat sequences. The molecular basis for
the appearance of these small RNAs is not clear, but the phenomenon may reflect
the tendency for RNA polymerase II to terminate transcription prematurely when
RNA sequences with extensive stable secondary structure are formed (
46
,
47
). The existence of the small W RNAs is reminiscent of the situation with the
TAR RNA sequence of HIV-1. This highly structured RNA is present at the 5'-ends of all HIV mRNAs, but also exists independently in
infected cells as a result of premature termination of viral transcription (
46
,
47
). Interestingly, because of its stable hairpin loop structure, TAR RNA also
binds to PKR and has been reported to be able to activate the kinase (
25
,
29
).
Evidence from both
in vitro
translation experiments and direct assays of PKR autophosphorylation indicates
that the W RNAs have the potential to activate the protein kinase. Such
activation appears to be a true property of these RNA species and cannot be
attributed to the presence of artefactual dsRNA contaminants in the
in vitro
transcripts. In this respect the W RNAs thus appear to be similar to TAR RNA (
25
,
29
), regions of the S1 mRNA of reovirus (
22
,
23
) and the collapsed circular RNA of the hepatitis delta virus (
48
) in functioning as PKR activators. It is not yet known what structural features
of the W RNAs are necessary or important for PKR activation, but preliminary
nuclease digestion studies indicate the presence of considerable secondary
structure. Nevertheless, it is rather surprising that the one repeat W RNA is
able to activate PKR because it does not possess any long tracts of
uninterrupted base pairing (data not shown). A similar situation appears to
exist in the case of nt 416-575 of reovirus S1 mRNA, which is a potent PKR activator (
22
). It seems likely that additional aspects of the structure of such small RNAs,
such as tertiary interactions, are important in the regulation of PKR activity.
A single repeat W RNA is also probably too small to bind more than one molecule
of PKR at a time, in which case it might be expected that dimerization-dependent activation (
21
) would not occur. The fact that this RNA can indeed activate the enzyme
suggests that RNA-mediated dimerization is not a prerequisite or that only one monomer in a
PKR dimer actually needs to be bound to the RNA for activation to take place.
However, it is notable that a seven-repeat W transcript is significantly better than one- or two-repeat RNAs, on a molar basis, in promoting PKR
autophosphorylation (Fig.
6
). Thus larger numbers of the repeat sequence linked in tandem, as occurs
in vivo
in EBNA mRNAs, may be more effective because they can bind greater numbers of
PKR monomers and thus facilitate trans-phosphorylation reactions.
The competition binding studies using W RNAs and EBER-1 indicate that these species may bind to the same (or at least
overlapping) sites on PKR. By means of filter binding we have previously
measured the
K
d
for the association of EBER-1 with PKR as being ~0.3 nM (
34
), similar to the value estimated by others for dsRNA binding to PKR (
40
,
49
) and other proteins (
50
) using different methods. It would seem likely that the W RNAs bind to PKR with
a comparable affinity to that of EBER-1 and other structured RNAs (note that 0.3 nM for (W1W2)
7
RNA is equivalent to a concentration of 0.14 [mu]g/ml). Similar concentrations of the W RNAs and EBER-1 (0.1-1 [mu]g/ml and above) are required to activate and inhibit PKR
respectively (this paper and
34
). Unlike EBER-1, the W RNAs are not inhibitory towards the dsRNA-dependent activation of the protein kinase, even at very high
concentrations.
The importance of the activation of PKR by transcripts from the major internal
repeat region of the viral genome is not yet known, but it may be advantageous
to the virus to temporarily shut down host protein synthesis during the early
stages of infection. Such an effect has been seen in cells infected with a
number of other (predominantly lytic) viruses (
51
-
53
) and is probably necessary to allow establishment of infections and the
accumulation of early viral mRNAs and proteins. EBNA mRNAs containing W repeat
sequences appear at relatively early times after EBV infection of B lymphocytes
(
19
,
54
).
Later during infection, in spite of the potential activation of PKR by the W
RNAs, protein synthesis must be stimulated in order to support B cell
proliferation. In this connection it may be significant that the PKR inhibitor
EBER-1 appears relatively late in the time course of infection (
19
), perhaps at a time when the activity of PKR needs to be down-regulated. Since PKR is an interferon-induced protein it may also be pertinent that B cell infection by
EBV is sensitive to inhibition by interferons at early but not at late times
after exposure of the cells to the virus (
55
). The mechanisms responsible for this time-dependent phenomenon have not been elucidated, however, and there is no
evidence for or against a role for PKR in the inhibition of EBV infection by
interferons. Future studies of the extent of activation of PKR at different
times after infection, as well as in established EBV-transformed cell lines, should help to determine whether the ability of
EBV to encode RNAs capable of both activating and inhibiting PKR constitutes a
means of fine tuning the activity of this protein kinase at different stages in
the viral life cycle.
We are grateful to Prof. P.J.Farrell and Drs D.R.Gewert, M.G.Katze, D.H.Levin
and V.M.Pain for gifts of materials. This work was funded by grants from the
Leukaemia Research Fund, the Cancer Research Campaign, the Wellcome Trust and
the Sylvia Reed Cancer Fund. A.E. was supported by a studentship from the
Medical Research Council.
*To whom correspondence should be addressed at present address: School of
Biological Sciences, University of Sussex, Brighton BN1 9QG, UK.
Tel: +44 1273 678544; Fax: +44 1273 678499; Email: m.clemens@sussex.ac.uk
REFERENCES
Return