ABSTRACT
Retroviral reverses transcriptases (RTs) are RNA- and DNA-dependent DNA polymerases that use a tRNA bound at the so-called primer binding site (PBS) located near the 5' end of the genomic RNA as primer. Thus, RTs must be able to accommodate both RNA and DNA in the primer strand. To test whether the natural primer confers some advantages to the priming process, we compared initiation of reverse transcription of avian and murine retroviral RNAs, using either their natural tRNA primer, tRNATrp and tRNAPro, respectively, or synthetic 18mer oligodeoxyribonucleotides (ODNs) and oligoribonucleotides (ORNs) complementary to their PBS. In both retroviral systems, the initial extension of ODNs was fast and processive. The initial extension of ORNs, tRNATrp and tRNAPro was much slower and distributive, giving rise to the transient accumulation of short pausing products. Synthesis of (-) strong-stop DNA was delayed when using ORNs and tRNAs, compared to ODNs. Even though ORNs and tRNAs were initially extended at the same rate, the short pausing products were more rapidly extended when using the tRNA primers. As a consequence, synthesis of (-) strong-stop DNA was much more efficient with tRNA primers, compared to ORNs. Taken together, these results suggest that the tRNA-primed synthesis of (-) strong-stop DNA is a two-step process, as already observed for HIV-1. The initiation mode corresponds to the initial non-processive nucleotide addition and extension of the short pausing products. It is more efficient with the natural primers than with ORNs. Initiation is followed by a more processive and unspecific elongation mode. Elongation is observed when the primer strand is DNA, i.e. when using the ODNs as primers or when the ORN and tRNA primers have been extended by a sufficient number (depending on the retroviral system) of deoxyribonucleotides.
Reverse transcription is a central event in the retroviral life cycle, that
converts two homologous genomic RNA molecules into double stranded DNA (
1
,
2
). Reverse transcription is carried out by the viral reverse transcriptase (RT),
which possesses RNA and DNA dependent DNA polymerase and RNase H activities
(for review see
3
,
4
). To initiate (-) strand DNA synthesis, RT uses a specific tRNA as primer (for review,
see
5
). tRNA
3
Lys
is the natural primer of Human Immunodeficiency Virus Type 1 (HIV-1) (
6
), while tRNA
Trp
(
7
) and tRNA
Pro
(
8
) are used by avian and most murine retroviruses, respectively. In retroviruses, the 3' 18 nucleotides (nt) of the primer tRNA are strictly complementary and bind to the 18
nt sequence of the viral Primer Binding Site (PBS) present in the 5' untranslated region of the genomic RNA (
9
-
11
).
In addition to this classical interaction, evidence for additional virus-specific primer-template interactions recently accumulated. Aiyar
et al.
(
12
,
13
) found evidence of an interaction between part of the T[psi]C arm of tRNA
Trp
and a 7 nt sequence upstream of the PBS of the genomic RNA of Rous sarcoma virus
that is involved in the efficiency of (-) DNA strong-stop synthesis (
12
,
13
). In HIV-1, the anticodon loop, the 3' part of the anticodon stem and parts of the variable loop of tRNA
3
Lys
interact with regions upstream of the PBS, leading to a binary tRNA/viral RNA
complex with a highly complex and compact structure (
14
-
16
). These HIV-1 RNA/tRNA
3
Lys
interactions require the presence of the post-transcriptional modifications of primer the tRNA
3
Lys
(
14
). A recent
in vivo
study clearly showed that the loop-loop interaction that takes place between the anticodon loop and the A-rich loop upstream of the PBS is important for maintaining a
specific primer tRNA during the HIV-1 replication cycle (
17
). Additional interactions also exist between the T- and D-arms of tRNA
i
Met
and three short regions of yeast retrotransposon Ty1 RNA downstream of the PBS
(
18
). These interactions play a role in reverse transcription of Ty1 retrotransposon (
19
) and are probably required for efficient annealing of tRNA
i
Met
to the genomic RNA (
18
).
These results suggest the existence of a specific ternary complex of initiation of reverse transcription. This hypothesis is supported by the
fact that all lentiviral RTs that use tRNA
3
Lys
as natural primer are not able to extend the tRNA3
Lys
/HIV-1 RNA primer/template (
20
). Indeed, in the case of HIV-1, we showed the existence of an initiation phase that can be
distinguished from the elongation phase of reverse transcription on the basis of biochemical and kinetic experiments (
21
,
22
). Whether, the specific initiation of reverse transcription followed by
unspecific elongation that we demonstrated in the case of HIV-1 exists in other retroviruses (and retrotransposons) is presently
unknown.
Here, we compared the ability of murine retroviral primers, tRNA
Pro
and
synthetic 18mer oligodeoxyribo- (ODN
MLV
) and oligoribonucleotides (ORN
MLV
) complementary to MLV PBS, or of avian retroviral primers, tRNA
Trp
, ODN
AMV
and ORN
AMV
, to support (-) strand strong-stop DNA synthesis in the presence of their cognate RT and
template. Our results are consistent with the existence of distinct initiation
and elongation phases of reverse transcription.
An RNA spanning nucleotides 1-725 of MLV genomic RNA was prepared by
in vitro
transcription using plasmid pIS1 linearised with
BstE
II as previously described (
23
). The RNA corresponding to nts -52 to 622 of the RSV genome was synthesised with RNA polymerase from
phage SP6, using the pLAD4 plasmid (
24
) cut with
Xho
I as template. Transcription was in 40 mM Tris-HCl, pH 7.5, 13 mM MgCl
2
, 10 mM NaCl, 4 mM spermidine, 10 mM DTT, 4 mM of each deoxyribonucleotide triphosphate (dNTP), 6 mg/ml of BSA, 10 U/ml of RNasin, 1400 U/ml SP6 RNA polymerase and 100 [mu]g/ml of plasmid for 3 h at 37oC. RNAs were purified by selective lithium chloride precipitation and
their integrity was checked as described (
14
).
tRNA
Pro
and tRNA
Trp
were purified from total beef liver tRNA prepared according to Fournier
et al.
(
25
) using published procedures: BD-cellulose column chromatography (
26
) followed by two dimensional PAGE (
27
). The fractions containing tRNA
Pro
or tRNA
Trp
were located by dot-blot hybridisation (
28
) with probes complementary to their 3' ends (nucleotides 59-73). 3' end labelling of tRNA
Trp
and tRNA
Pro
with [[alpha]-
32
P]ATP (Amersham) was as described (
14
).
ORN and ODN primers were chemically synthesised and 5' end labelled with [[gamma]-
32
P]ATP (Amersham) and polynucleotide kinase from phage T4 (USB).
RTs from MLV and AMV were from Gibco and Life Sciences, respectively.
In reverse transcription experiments, the RNA template was hybridized with a
32
P-labelled primer at a 2:1 molar ratio as previously described (
14
). The primer/template complex was incubated at 37oC for 4 min with 9 or 27 nM of RT in 50 mM Tris-HCl pH 8.0, 50 mM KCl, 6 mM MgCl
2
and 1 mM dithioerythritol. Reverse transcription was initiated by addition of
the four dNTPs (50 [mu]M final concentration each) and stopped with formamide containing 50 mM EDTA
at times ranging from 15 s to 25 min. The reaction products were analyzed on 8
or 12% polyacrylamide (1/20 bis-acrylamide) denaturing gels and quantified with a BioImager BAS 2000 (Fuji). The single turnover experiments were performed using poly(rA)
.
(dT)
15
(10:1 w:w) at a final concentration of 1.66 [mu]M of (dT)
15
in presence of the four dNTPs, as described in (
21
).
Using the procedure previously described for HIV-1, we studied the initiation of avian and murine reverse transcription by
directly evaluating utilisation of natural and synthetic primers in the
initiation phase and by analysing the synthesis of (-) strong-stop DNA. The RNAs used as template cover the 5' region of the retroviral genome containing the PBS. The
primers were labelled at their 5' (ODNs and ORNs) or 3' (tRNAs) end, allowing direct quantification of the extended and
unextended species, and hence of the priming efficiency, which is proportional
to the sum of all extended products normalised to the initial amount of primer.
Avian and murine primers were hybridized to the corresponding template and
synthesis of (-) strong-stop DNA was performed by addition of the four dNTPs in presence of
homologous AMV or MLV RT. (-) strong-stop DNA results from run off reverse transcription of the
template.
Using either MLV or AMV RT, more than 90% of the initial amount of all primers,
including ODNs, was eventually extended upon prolonged incubation (Fig.
1
B and D). This result indicates that, when using the ODN primers, cleavage of
the PBS by RNase H activity of the wild type MLV and AMV enzymes during the 4
min preincubation was very limited and could be neglected. However, RNase H
activity became noticeable after prolonged incubation (see below).
In addition to the priming efficiency, we also quantified the synthesis of (-) strong-stop DNA. For this purpose, we worked at a three-fold excess of RT (27 nM) in order to level off the priming
efficiency of the natural and synthetic primers. By comparing Figure
2
A and B, it appears that a major difference between avian and murine systems was
the time needed for appearance of (-) strong-stop DNA. When comparing the corresponding primers (ODN
AMV
and ODN
MLV
, ORN
AMV
and ORN
MLV
, tRNA
Trp
and tRNA
Pro
), (-) strong-stop DNA was always detected at significantly shorter incubation
times in the avian system than in the murine one (Fig.
2
). This result probably reflects an intrinsic lower processivity of MLV RT
compared to AMV RT. In each system, significant differences were also observed
between the primers. When using ODN
AMV
, 20% of the primer was extended into (-) strong-stop DNA in about 15 s, while the same elongation level was obtained only after 1 and 2 min with tRNA
Trp
and ORN
AMV
, respectively (Fig.
2
A). However, after prolonged incubation (25 min), tRNA
Trp
and ORN
AMV
yielded a higher amount of (-) strong-stop DNA than ODN
AMV
(Fig.
2
A). One surprising result highlighted by these experiments is that synthesis of
(-) strong-stop DNA was delayed when ORN
AMV
was used as primer instead of tRNA
Trp
(Fig.
2
A). This result has to be compared with the priming efficiency data discussed
above, where we found that ORN
AMV
and tRNA
Trp
- primed initiation were identical. Thus, the more efficient (-) strong-stop DNA synthesis with tRNA
Trp
compared to ORN
AMV
resulted from a step occurring after the initial extension of the primers.
Synthesis of (-) strong-stop DNA was not a completely processive event. Rather, many short
intermediate products, resulting from pauses of the RT, appeared during DNA
synthesis (Figs
3
-
6
). For a given viral system, the profile of these pausing sites was different
with each primer (Figs
3
-
6
). ODNs are the primers that gave the lowest amount of short transient products
during incorporation of the first nucleotides (+1 to + 11 products for ODN
MLV
and +1 to + 6 products for ODN
AMV
) (Fig.
3
A and
5
A). The profile of pauses resulting from tRNA-primed reaction (+1 to + 8 products for tRNA
Pro
and +1 to + 28 products for tRNA
Trp
) was also very characteristic, showing a strong decrease of the pausing
products with incubation time (Fig.
3
B and
5
B). On the opposite, the ORNs yielded high amounts of pausing products (+1 to +
11 products for ORN
MLV
and +1 to + 6 products for ORN
AMV
) which only hardly decreased with time (Fig.
3
C and
5
C). Pauses with all the primers except tRNA
Trp
are located within the first 11 nt. In the case of tRNA
Trp
, the pausing sites from positions +20 to +28 could be correlated to the
additional base pairing proposed by Aiyar
et al.
(
12
), which takes place between the T[psi]C arm of tRNA
Trp
and a single stranded region 25-31 nt upstream of the PBS.
The results we described so far indicate a strong difference between tRNAs and
ODNs in their ability to prime reverse transcription. To further investigate
this difference, we performed extension of tRNA
Pro
, tRNA
Trp
, ODN
MLV
and ODN
AMV
with their homologous RT in a single turn-over experiment in which recycling of the enzyme was prevented by using
poly(rA)
.
(dT)
15
as a trap for the free enzyme. Extension of both ODN
MLV
and ODN
AMV
primers (compare Fig.
3
A to Fig.
7
A and Fig.
5
A to Fig.
7
B) was reduced in the presence of the trap. Some extension products were
observed with ODN
MLV
(Fig.
7
A), with stronger pausing sites within the first six nucleotides, as compared to
the standard extension experiment, and complete dissociation of MLV RT occurred
after incorporation of 51 nucleotides. This region also corresponded to strong
pausing sites when reaction was conducted without trap (Fig.
3
A). Thus, MLV RT dissociated before complete synthesis of the (-) strong-stop DNA. In the avian system, despite the lower level of strong-stop DNA as compared to the extension without trap after 25
min of incubation (Fig.
5
A and
7
B), AMV RT, thanks to its high processivity, was able to reach the 5' end of the template without dissociating. On the contrary, almost no
extension of tRNA
Pro
and tRNA
Trp
was observed in the presence of poly(rA)
.
(dT)
15
(Fig.
7
C and D). Only a very faint band of (-) strong-stop DNA was detected in the tRNA
Trp
-primed reaction. These results, together with the accumulation of +1 and
+1/+2 products when tRNA
Pro
and tRNA
Trp
, respectively, are used as primers, indicate that RT dissociation from the
binary tRNA/viral RNA complexes is very fast. On the opposite, RT dissociates
slowly from the ODN/viral RNA complexes.
The results that we obtained here on a murine and on an avian system clearly
show that, as in the case of HIV-1 system (
21
,
22
,
29
), ODN-primed reverse transcription differs from ORN- and tRNA-primed reactions. First, single turnover experiments performed
in the presence of a trap indicate that nucleotide incorporation is faster than
enzyme dissociation when ODNs are used as primer, while dissociation is faster
than nucleotide addition with the tRNA primers. Second, the ODN-primed synthesis of (-) strong-stop DNA is faster than the ORN- and tRNA-primed reactions. Third, large amounts of short
pausing products transiently accumulate with the ORN primers, while they are
not observed or are extremely rapidly extended when using ODN primers. The
dissociation of MLV RT at position +7 of the template with ODN
MLV
(Figs
3
and
7
) probably reflects a local low processivity of this enzyme. The difference with
the pauses observed with the RNA primers is clear when looking at the time
course of the reaction. The pausing products observed with the ORNs and tRNAs
accumulate over time, while the +7 product observed with the ODN
MLV
is readily extended.
The priming efficiency.
In addition to the differences observed between ODNs and the ribonucleotidic
primers, there are also noticeable, although less dramatic, differences between
ORN- and tRNA-primed (-) strong stop DNA synthesis. Even though, in the two
retroviral systems used here, the priming efficiency of the ORN and tRNA
primers is the same, differences appear after the initial extension of the
primer. First, the short pausing products, which are observed with ORN
MLV
and ORN
AMV
as well as with tRNA
Pro
and tRNA
Trp
, are more rapidly and efficiently extended when using the latter primers.
Second, the synthesis of the (-) strong stop DNA is also more efficient with the natural primers,
compared to the ORNs.
The similar priming efficiency of ORN
AMV
and ORN
MLV
,
compared to tRNA
Trp
and tRNA
Pro
, respectively, contrasts with previous observations on HIV-1. In the case of HIV-1, the post-transcriptional modifications of tRNA
3
Lys
are required for efficient primer extension; a synthetic tRNA
3
Lys
lacking the post- transcriptional modifications or an 18mer ORN complementary to the PBS
(ORN
HIV
) are very inefficient primers (
21
,
22
,
30
). In the case of MLV, the results described above may not be surprising.
Indeed, contrary to HIV-1 RT, MLV RT does not specifically interact with its primer tRNA (
31
,
32
), and encapsidation of tRNA
Pro
in murine retroviral particles is less specific than that of tRNA
3
Lys
in HIV-1 and does not require the RT precursor (
33
,
34
). On the opposite, tRNA
Trp
(
35
) and tRNA
3
Lys
(
36
) packaging processes require specific interactions between the primer tRNA and
the RT domain of the
gag-pol
precursor. Moreover, extensive studies on AMV RT/tRNA
Trp
interaction provide evidence for a specific recognition between RT and the
primer (
31
,
37
,
38
), as proposed in the case of the HIV-1 RT/tRNA
3
Lys
complex (
39
-
42
) (but see also
43
). Thus, the difference of specificity in the initiation of reverse
transcription in HIV-1 and avian retroviruses is rather unexpected.
The short pausing products.
When ORN or tRNAs were used to prime reverse transcription of avian and murine
genomic RNAs, large amounts of short pausing products transiently accumulate.
We noticed that in both systems, the short pausing products are hardly extended
when using ORNs as primers, while they are more efficiently extended into
larger products in the tRNA-primed reactions. Thus, RT binding to the primers extended by a few
nucleotides appears to be much more efficient when the primer is a tRNA than
when the primer is an ORN. This indicates that MLV and AMV RTs are able to
recognise specific features of the annealed tRNA primers. A similar situation
also exists in HIV-1 (
21
,
30
); short pausing products are observed when tRNA
3
Lys
or ORN
HIV
are used as primers. We showed that the extended interactions between tRNA
3
Lys
and HIV-1 RNA, and in particular base-pairing of the anticodon loop with a viral A-rich loop, are required for efficient extension of these short
pausing products (
21
). These interactions take place with the natural tRNA
3
Lys
, but not with a synthetic one lacking the post-transcriptional modifications or with ORN
HIV
(
14
,
15
), and hence efficient extension of the short pausing products is observed only
with the natural primer (
21
). The involvement of this loop-loop interaction in the replication of HIV-1 was recently proved
in vivo
by Wakefield
et al
. (
17
). Extended virus specific primer/template interactions also exist in avian
retroviruses. In particular, the T[psi]C loop of tRNA
Trp
interacts with a 7 nt sequence located upstream of the PBS (
12
,
13
). This interaction is required for efficient synthesis of (-) strong-stop DNA
in vitro
and for efficient viral replication
in vivo
(
12
,
13
,
44
). Based on sequence analysis, a similar interaction was postulated to take
place in MLV (
12
). This proposal is not experimentally supported and a detailed structural
probing analysis on the tRNA
Pro
/MLV viral RNA complex performed in our laboratory failed to detect any
primer/template interaction outside of the expected annealing of the 3' end of tRNA
Pro
to the PBS (P. Fossé
et al.
, to be published). Therefore, specific interactions between the annealed tRNA
Pro
and MLV RT, rather than extended primer-template interactions, probably facilitate extension of the short pausing products in MLV.
Synthesis of (-) strong stop DNA
. The low processivity during the initial extension of the ribonucleotidic
primers probably explains why the synthesis of (-) strong-stop DNA is delayed when using these primers instead of ODNs.
Furthermore, the more efficient extension of the short pausing products when
using tRNA primers, as compared to ORNs, is reflected at the level of (-) strong stop DNA synthesis. However, it is surprising that despite an
initial delay in the (-) strong stop DNA synthesis in the tRNA-primed, compared to the DNA-primed reaction, more (-) strong stop DNA is synthesized in the tRNA-primed reaction upon prolonged incubation. This
result suggests that following the initial tRNA extension taking place with low
processivity, the tRNA primer may increase processivity of the c-DNA extension, compared to the ODN primers.
Based on biochemical (
21
) and kinetic (
22
) evidence, we proposed that the tRNA
3
Lys
-primed synthesis of (-) strong stop DNA is a two-step process. An initial weakly processive DNA synthesis
requiring specific interactions between primer, template and RT, is followed by
a non-specific and processive mode of DNA synthesis (
21
,
22
). We called these two steps the initiation and elongation phases of reverse
transcriptase, respectively. The experimental evidence presented here suggests
that a similar interpretation holds true for the avian and murine retroviruses.
The strong pausing products observed in these systems with the
oligoribonucleotidic primers reflect the low processivity of the initiation
phase. Extension of these products is more efficient when using the tRNA
primers, compared to ORNs, because of specific interactions existing in the
natural tRNA/viral RNA/RT initiation complex. Elongation is observed when the
primer strand is DNA, i.e. when using the ODNs as primers or when the ORN and
tRNA primers have been extended by a sufficient number (depending on the retroviral system) of deoxyribonucleotides. The fact that the DNA-primed reverse transcription is unspecific is routinely used in primer
extension experiments, in which almost any DNA or RNA sequence can be
transcribed by AMV or MLV RT using complementary oligodeoxyribonucleotides as
primer. The existence of an initiation phase when using tRNAs or ORNs, but not
ODNs, as primers is certainly due to minor structural differences existing
between RNA/RNA and RNA/DNA hybrids. Indeed, although both helical structures
belong to the A family, significant differences exist between the two
structures in solution (
45
). Thus, these two hybrids probably make different contacts with RT that may
explain why RNA/RNA hybrids are not cleaved by RNase H (
45
), but also the differences that we observed in the priming mechanism.
An alternative interpretation could be that our data reflect a stronger binding
of RT to DNA/RNA compared to RNA/RNA primer/template and that the differences
between tRNAs and ORNs are due to specific interactions of RTs with their
cognate primers. This interpretation would not require the existence of two
distinct processes. Even though one cannot discount this interpretation solely
on the basis of the present data, it is not supported by detailed analysis of the literature. First, the specificity of the initiation complex is not due to specific interactions of RTs with their
cognate tRNA primers, but to specific interactions of RTs with their cognate
tRNA/viral RNA primer/templates. This is shown by the importance of the
extended interactions between the primer tRNA and the genomic RNA of Rous
sarcoma virus (
12
,
13
), HIV-1 (
14
,
15
,
21
), and the yeast retrotransposon Ty1 (
18
) in the synthesis of (-) strong stop DNA. At the opposite, RTs from HIV-2, feline immunodeficiency virus and equine infectious anaemia
virus are unable to extend the tRNA
3
Lys
/ HIV-1 RNA primer/template although their natural primer is tRNA
3
Lys
(
20
). These data point towards the existence of a highly specific ternary
initiation complex. Second, in HIV-1, which is most studied system, strong differences exist between the
initiation and elongation modes, even though the binding affinity of RT for the
primer/template is very close in the two modes (
22
). Indeed, the initiation and elongation phases are catalytically distinct: the
elongation phase is inhibited by Mn
2+
while initiation is not (
21
), and the rate of nucleotide incorporation is 50-fold faster during elongation, as compared to initiation (
22
). Thus, we propose that, as in HIV-1, (-) strong stop DNA synthesis in avian and murine retroviruses is a
two step process, namely specific and weakly processive initiation followed by
processive and non-specific elongation.
Jean-Marc Lanchy is thanked for fruitful discussions. This work was supported
by the French `Agence Nationale de Recherches sur le SIDA' (ANRS).
*To whom correspondence should be addressed. Tel: +33 3 88 41 70 91; Fax: +33 3
88 60 22 18; Email: marquet@ibmc.u-strasbg.fr
REFERENCES
Return