JOURNAL OF VIROLOGY, Aug. 2002, p. 7987–7995
0022-538X/02/$04.00⫹0 DOI: 10.1128/JVI.76.16.7987–7995.2002
Copyright © 2002, American Society for Microbiology. All Rights Reserved.
Vol. 76, No. 16
Given the Opportunity, the Sendai Virus RNA-Dependent RNA
Polymerase Could as Well Enter Its Template Internally
Diane Vulliémoz and Laurent Roux*
Department of Genetics and Microbiology, University of Geneva Medical School, 1211 Geneva 4, Switzerland
Received 18 January 2002/Accepted 15 May 2002
Viral RNA synthesis on negative-stranded RNA virus infection consists of two distinct processes: genome replication and
transcription of the viral mRNAs. Replication initiates at the
very 3⬘-OH end of the viral RNA and proceeds via the synthesis of its full complement, the antigenome, which in turn serves
as the template to amplify the viral genome. The genomic
promoter (GP) and antigenomic promoter (AGP) for replication of the genome and the antigenome, respectively, are positioned at the RNA 3⬘ end of the genome and the antigenome,
respectively (for general information, see reference 23). For
Sendai virus, a member of the Paramyxoviridae family with a
nonsegmented genome of 15,384 nucleotides, the replication
promoters are within the first 96 nucleotides (34).
While the GP and AGP have the same gross primary
structure organization, only the first 12 nucleotides of the
promoters are identical. With the possible exception of members of the Pneumovirus genus (e.g., respiratory syncytial virus), the replication promoters of all members of the
Paramyxoviridae family are composed of two discontinuous
elements (graphically portrayed in reference 36). The first 20
to 30 nucleotides most proximal to the 3⬘ end are fairly conserved among members of the same genus and are likely to
contain the signal for RNA synthesis initiation. The second
element stretches into the internal extremity of the promoter.
A three-times-repeated motif common to the respiro- and
morbilliviruses [(CNNNNN)3, nucleotides 79 to 96] and the
rubulaviruses [(NNNNGC)3, nucleotides 77 to 94] is absolutely
required in number and location for replication to occur (26,
27, 34).
By use of a reverse genetic system that recently became
available for negative-stranded RNA viruses (1, 4, 8, 11, 13,
16–18, 20, 24, 29, 30, 37), the Sendai virus GP and AGP were
found to perform differentially, in that a minireplicon under
the control of the AGP replicated ⬃20-fold more efficiently
than that controlled by the GP (7, 35). This difference in
promoter strength depended on the promoter’s primary sequence, since reciprocal exchanges of the first 30 nucleotides
between the two promoters was sufficient to adjust their replication efficiencies (7). Finally, the amplitude of the difference
in promoter strength was found to depend on the expression of
the accessory C proteins (10), which appear to specifically
restrict replication at the GP (5, 33).
For Sendai virus, transcription initiates 56 nucleotides away
from the 3⬘ end of the viral RNA in a region of about 10
nucleotides found repeated at each mRNA start site. Once
initiated, transcription proceeds through the sequential synthesis of capped and polyadenylated messengers via a stop-restart
signal recognition mechanism (for general information, see
reference 23). It is generally accepted that the transcription
promoter overlaps the genomic replication promoter, although
it is clear that certain portions of the primary sequence must be
devoted to only one process, e.g., the transcription start signal,
devoted to transcription, or the first 12 to 30 nucleotides, which
are interchangeable between the GP and AGP and required
for replication. How the viral RNA-dependent RNA polymer-
* Corresponding author. Mailing address: Department of Genetics
and Microbiology, University of Geneva Medical School, CMU, 1 rue
Michel-Servet, 1211 Geneva 4, Switzerland. Phone: (44 22) 702 56 83.
Fax: (44 22) 702 57 02. E-mail: Laurent.Roux@medecine.unige.ch.
7987
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
The negative-stranded RNA viral genome is an RNA-protein complex of helicoidal symmetry, resistant to
nonionic detergent and high salt, in which the RNA is protected from RNase digestion. The 15,384 nucleotides
of the Sendai virus genome are bound to 2,564 subunits of the N protein, each interacting with six nucleotides
so tightly that the bases are poorly accessible to soluble reagents. With such a uniform structure, the question
of template recognition by the viral RNA polymerase has been raised. In a previous study, the N-phase context
has been proposed to be crucial for this recognition, a notion referring to the importance of the position in
which the nucleotides interact with the N protein. The N-phase context ruled out the role of the template 3ⴕ-OH
congruence, a feature resulting from the obedience to the rule of six that implies the precise interaction of the
last six 3ⴕ-OH nucleotides with the last N protein. The N-phase context then allows prediction of the recognition by the RNA polymerase of a replication promoter sequence even if internally positioned, a promoter
which normally lies at the template extremity. In this study, with template minireplicons bearing tandem
replication promoters separated by intervening sequences, we present data that indeed show that initiation of
RNA synthesis takes place at the internal promoter. This internal initiation can best be interpreted as the
result of the polymerase entering the template at the internal promoter. In this way, the data are consistent
with the importance of the N-phase context in template recognition. Moreover, by introducing between the two
promoters a stretch of 10 A residues which represent a barrier for RNA synthesis, we found that the ability of
the RNA polymerase to cross this barrier depends on the type of replication promoter, strong or weak, that the
RNA polymerase starts on, a sign that the RNA polymerase may be somehow imprinted in its activity by the
nature of the promoter on which it starts synthesis.
7988
VULLIÉMOZ AND ROUX
J. VIROL.
and Neubert and colleagues (28). Whenever explicit, the viral sequence is written
as DNA according to standard convention, 5⬘ to 3⬘, as the positive strand. The
plasmids expressing the Sendai virus N protein (pGem-N), P and C proteins
(pGem-P/Cwt), P without the C proteins (pGem-P/Cstop), and the L protein
(pGem-L) under the control of the T7 RNA polymerase promoter have been
described previously (6, 33). The structure of the minireplicon Sendai virus
DI-H4-(⫹)⌬DRA/AGP96 (referred to as pSV DIH4-⌬96, construct 1) and its
cloning into pSP65 have been described (34). As before (35), the 3⬘ ends of the
viral genome and antigenome contain the genomic (GP) and antigenomic (AGP)
promoters, respectively. In the case of the H4 RNA, which harbors the AGP on
both the minus- and plus-strand RNA, AGPL or AGPR, respectively, denote
these two promoters (see Fig. 1A).
In all derivatives used in this study, the 3⬘ end of the H4 minus-strand RNA
containing AGPL was never modified. Most of the plasmids used here were
derived from two different constructs containing two promoters in tandem at the
3⬘ end of the antigenome, Sendai virus DI-H4/AGP-GP and Sendai virus DIH4/AGP-AGP, described previously (36), and are referred to in this study as
AGPint-GPext (construct 3) and as AGPint-AGPext (construct 10), respectively.
The site-directed base substitution (G91C) was introduced in either promoter of
constructs 9 and 13 by fusion PCR with adequate overlapping oligonucleotides at
the inner modification sites and external primers containing the MunI site in the
minireplicon sequence and the BamHI site at the 3⬘ end of the ribozyme sequence (see Fig. 3A) (36).
The opening of an XhoI restriction site introduced between the adjacent
promoters by fusion PCR (constructs 4 and 11) (36) allowed the introduction of
gradually increasing DNA fragments originating from the Sendai virus M gene to
gradually extend the distance between the two promoters. Fragments of 42, 84,
120, 162, and 240 nucleotides were prepared by PCR amplification of portions of
the Sendai virus M gene (accession number X53056), comprising nucleotides 595
to 823, with primers flanked by XhoI sites (constructs 5, 6, 7, 8, 9, 12, and 13). The
introduction of the stretch of 10 A’s (10A stretch) in construct 9 to generate
construct 9A10 was done by fusion PCR. The 240(A10) cassette of 9A10 was then
removed by XhoI digestion and introduced in the XhoI site to generate constructs
9A10GP(G91C), 13A10, 13-A10Ex(G91C), and 13-A10In(G91C). The 10A stretch
preceding either AGP or GP in constructs 1A10 and 2A10 was introduced by
cloning a 10A cassette (5⬘-GTGAAAAAAAAAAGCTCACACTGG-3⬘) flanked
with two DraIII sites into the unique DraIII site. The portions of the constructs
that had undergone modifications were all extensively sequenced with the Licor
DNA 4000 automated sequencing apparatus.
Replication system. The intracellular replication system has been extensively
and repeatedly described before (6, 35, 36). Its main features are illustrated once
more in Fig. 1A and commented on in the first section of the Results. All the
replication assays were performed with the P/Cstop plasmid, i.e., in the absence of
the C proteins, except for the experiment presented in Fig. 1D, as indicated in
the figure legend.
RNA analysis. The encapsidated RNA (minireplicon) resulting from replication was analyzed by Northern blotting with a 32P-labeled riboprobe of positive
polarity described before (5⬘ex-probe) (25). The in vitro T7 RNA transcripts
produced from plasmids 1, 1A10, 2, and 2A10 (Fig. 3) were synthesized in
standard conditions. They were probed with a 32P-labeled Sp6 transcript of
negative polarity produced from plasmid 1 linearized at the DraIII site downstream of the 10A stretch so that the probe only scored the T7 RNA transcripts
that crossed the 10A stretch (see Fig. 3A). Primer extensions, used marginally in
this study to verify the precise initiation of the replicated RNA (see Fig. 4C),
were performed as previously described (36). Reverse transcription (RT)-PCR
amplification was performed after a further purification of the replicated minireplicon RNA through a 5.7 M CsCl cushion to get rid of possible plasmid
contamination. Moloney murine leukemia virus reverse transcriptase (Promega)
and Taq polymerase (Gibco-BRL Life Technologies) were used in standard
conditions with appropriate primers to obtain a 336-nucleotide-long amplified
product spanning the 10A stretch region (see Fig. 2D).
MATERIALS AND METHODS
RESULTS
Virus and cells. A549 cells and HeLa cells were grown in regular minimal
essential medium (MEM) and in Dulbecco’s modified medium (DMEM) supplemented with 5% fetal calf serum in a 5% CO2 atmosphere, respectively.
Recombinant vaccinia virus expressing the T7 RNA polymerase, vTF7-3, a gift
from Bernard Moss (National Institute of Health, Bethesda, Md.), has been
described by Fuerst and colleagues (14) and used accordingly. vTF7-3 stocks
were grown in HeLa cells, with titers around 5 ⫻ 108 PFU/ml.
Sequence and plasmids. The complete Sendai virus RNA primary sequence
(15,384 nucleotides) was taken from the work of Shioda and colleagues (31, 32)
Minireplicon replication system. The replication system
used in this study has been used successfully previously (6, 7,
33–36). It roughly corresponds to the method classically developed to replicate minireplicons of the other negative-stranded
RNA viruses (for reviews, see references 3, 9, and 15) and is
similar to that which allowed the recovery of infectious Sendai
virus from cDNA (16). This method is based on a multiplasmid
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
ase (RdRp) is committed to replication or transcription is still
a matter of debate and research.
The active genome of negative-stranded RNA viruses is not
the naked RNA but the nucleocapsid (NC), i.e., an RNAnucleocapsid protein (N) complex of helicoidal symmetry. Although the structure at atomic resolution of this nucleocapsid
has yet to be determined, the tightness of the RNA-N protein
complex varies among the virus families. The resistance to
high-salt conditions and/or to equilibrium density centrifugation (23), as well as the accessibility of the bases to watersoluble reagents and their ability to base pair (2, 19, 21a), are
criteria that allow this differentiation.
Compared to the orthomyxoviruses (e.g., influenza virus)
and the rhabdoviruses (e.g., vesicular stomatitis virus), the
paramyxoviruses (e.g., Sendai virus) appear to exhibit the highest binding of the RNA and N (19a). The interaction of the N
protein with six nucleotides possibly relates to the fact that the
rule of six applies to this family (12). This rule stipulates that
the total number of nucleotides in the genome must be a
multiple of six for efficient genome replication (6). Importantly, the rule implies that the nucleotides are seen by the viral
RdRp in the context of the nucleocapsid (N) protein (22).
Recently, the importance of the N protein phase context
received experimental support with the demonstration that
recognition by the viral RdRp of the replication promoters
depended on the position of the nucleotides vis-à-vis the six N
protein interaction sites (36). This result was obtained with
minigenomes carrying two promoters in tandem separated by
six nucleotides of intervening sequence that allowed changing
the N-phase context of the inner promoter while keeping the
original N-phase context of the external promoter. Despite
efficient replication from the external promoter, the internal
promoter was not used under these conditions. This contrasted
with the use of both promoters when the inner promoter’s
original N-phase context was conserved as well. More than
supporting the importance of the N-phase context for promoter recognition, these experiments raised questions about
the mechanism of template recognition by the viral RdRp. For
instance, when replication initiation took place at the internal
promoter, was the viral RdRp entering the template at the
internal promoter location or was it initiating internal RNA
synthesis by scanning from the template extremity?
In this study, we address this question and we provide data
that support entry at the internal promoter as well. Moreover,
by confronting the viral RdRp with what appears to be an
elongation barrier, we provide evidence for an imprinting of
the viral RdRp by the type of promoter (GP or AGP) on which
it first starts. The importance of these results to our understanding of the mechanisms of viral RNA synthesis is discussed.
VOL. 76, 2002
TEMPLATE ENTRY OF PARAMYXOVIRUS RNA POLYMERASE
7989
transfection of cells (Fig. 1A) previously infected with a vaccinia virus recombinant expressing the T7 RNA polymerase
(vTF7-3).
The minireplicons used here were derived from a naturally
amplified copyback-defective interfering RNA which originally
harbored the antigenomic replication promoter (AGP) at the
3⬘ end of both the positive and the negative strands. The
template plasmids are designed so that the T7 RNA transcript
produced is of positive polarity and contains the exact viral 5⬘
and 3⬘ natural extremities due to the adequate positioning of
the T7 RNA promoter and the presence of the hepatitis virus
delta ribozyme (Fig. 1A, diagram b) (6). The T7 transcript is
then encapsidated by the N protein (Fig. 1A, diagram c), and
as such becomes a template for the Sendai virus RdRp. The
viral RdRp now copies this template and generates an encapsidated RNA of negative polarity that becomes the hallmark of
replication (Fig. 1A, diagram d). At the end of the reaction
(generally 42 h posttransfection or postinfection), the encapsidated RNA is purified and analyzed by Northern blotting or
primer extension, with scoring only for the RNA of negative
polarity (Fig. 1A, diagram e) (for details, see Materials and
Methods) (36).
In the present study, the promoter manipulations were performed exclusively at the 3⬘ end of the positive-strand RNA
(AGPR, Fig. 1A, diagram c), leaving the 3⬘ end of the negativestrand RNA (AGPL) unaltered. In consequence, the replication potential of the minigenomes was only differentially affected by the promoter (or the double promoters) of the
positive-strand RNAs. Multiple rounds of replication with the
minus-strand minireplicon (and AGPL) as the template were
in principle nondiscriminative, since all the minus-strand minireplicons harbored identical AGPs.
Internal initiation is independent of the length of intervening sequence between the two promoters. In our previous
study (36), the two promoters in tandem, in the configuration
AGPint-GPex (Fig. 1B, construct 3), were adjacent or separated
by at most six nucleotides (construct 4). To verify whether the
use of the internal AGP (AGPint) was dependent on the proximity of the external GP (GPext), the intervening distance between the two promoters was sequentially increased from 6 to
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
FIG. 1. Replication of the AGPint-GPext constructs. (A) Schematic representation of the replication system. For a detailed description, see
Results. NoP and PE refer, respectively, to the Northern blot riboprobe and to the primer extension oligonucleotide, both of positive polarity,
complementary to the replicated RNA, depicted in diagram e. (B) Schematic representation of the AGPint-GPext series constructs. 0, Partial
scheme of the pSP65 plasmid harboring the basic copyback minireplicon pSV-DI-H4⌬96 flanked by the T7 RNA polymerase promoter (T7p) and
the hepatitis delta virus ribozyme (Rbz). 1, T7 RNA transcript produced from 0, carrying at its 5⬘ and 3⬘ ends the complementary sequence of the
antigenomic promoter (AGPLC) and the antigenomic promoter sequence (AGPR), respectively. 2, Minireplicon T7 transcript harboring at its 3⬘
end the genomic promoter (GP) instead of the AGP. 3, Minireplicon T7 transcript harboring at its 3⬘ end an internal AGP sequence adjacent to
an external GP. 4 to 9, Series of minireplicon T7 transcripts harboring at their 3⬘ ends a double promoter, as in construct 3, but separated by
increasing intervening sequences as indicated (nt, nucleotides). Note that all the constructs harbor the same 5⬘ end sequence (AGPLC).
(C) Northern blot analysis (see Materials and Methods) of the encapsidated minireplicon RNAs of negative polarity produced upon transfection
of A549 cells with the various plasmids described for panel B in replication assays made in the absence of the C proteins (P/Cstop conditions; see
Materials and Methods). Open and solid arrowheads, external and internal initiation, respectively. (D) As for panel C, but the replication assays
were performed in HeLa cells in the presence of the C proteins (Cwt conditions; see Materials and Methods).
7990
VULLIÉMOZ AND ROUX
FIG. 2. Effect of the 10A stretch on minireplicon replication.
(A) Schematic representation of the single-promoter minireplicons
used. 0, as described in the legend to Fig. 1B, with further features
related to the present experiment: the restriction sites [DraIII, BamHI
(BH1), and MunI], the sequencing primer (SE), and the Sp6 promoter.
1 and 2, as for Fig. 1B. 1A10 and 2A10, Derivatives of 1 and 2, respectively, containing a 10A stretch inserted at the DraIII site. (B) Northern blot analysis of the in vivo-replicated minireplicons. Lanes ⫺L,
replication assays performed in the absence of the L-expressing plasmid. (C) Northern blot analysis of in vitro T7 RNA polymerase transcriptions of pSp65 plasmids carrying the indicated minireplicons. The
plasmids were linearized at the BamHI (BH1) site (see panel A). The
probe of negative polarity resulting from an Sp6 RNA polymerase
transcription of plasmid 1 or 2 linearized at the DraIII site (indicated
as probe ⫺ in panel A) was used. Open arrowhead, plasmids. Solid
arrowhead, T7 transcripts. ⫺T7p, in vitro transcription without T7
RNA polymerase. (D) The plasmid-carrying construct 1A10 and the
RT-PCR product amplified (see text) from the in vivo-replicated minireplicon 1A10 were sequenced with a primer situated as indicated in
panel A (SE). Only the relevant portions of the sequence are shown.
⬎⬎ indicates further extension of A’s past the 10A stretch, ⴱ indicates
longer replication products, and ⬍⬍⬍ indicates the large accumulation
of heterogeneous shorter products.
can it have the same effect on the T7 RNA polymerase, since
the level of the T7 RNA transcript obtained in vitro from the
corresponding template plasmid was not affected (Fig. 2C,
compare lanes 1 and 2 with 1A10 and 2A10).
In Fig. 3B (left panel), comparison of lanes 2 and 2A10 again
shows the blocking effect of the 10A stretch on the single-GP
construct. When positioned between GP and AGP, the 10A
stretch had a similar inhibitory effect on GPext (lane 9A10).
AGPint, however, remained perfectly functional (compare with
lane 9), showing that replication from the internal promoter
did not depend on replication from the external promoter.
The function of the external promoter was also obliterated
by a G91C point mutation shown to prevent replication (34,
36). In this case, internal initiation was observed in the total
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
240 nucleotides (Fig. 1B, constructs 4 to 9). Figure 1C presents
a Northern blot of the replication products where the use of
AGPint corresponds to the signal of construct 1 (single-AGP
construct) (Fig. 1B) or 2 (single-GP construct), and the use of
GPext is reflected by the upper signals whose sizes increase
sequentially with the increasing intervening sequence (lanes 5
to 9). The constant presence of the lower band was evidence
that AGPint was used regardless of its distance from GPext (the
initiation at both promoters for constructs 3 and 4 was verified
by primer extension; data not shown) (36).
In construct 9, the 240 nucleotides of intervening sequence
represented more than 3 helix turns of distance between the
two promoters. As shown previously (7), the replication of the
minigenome carrying a single GP (construct 2) was less efficient than that harboring an AGP (construct 1) (difference of
about 10-fold in the experiments presented). For the promoters in tandem, this was still the case, but the difference was
statistically reduced to 1.5- to 2.0-fold (not shown).
The replication assay in Fig. 1C was performed in the absence of the C proteins (P/Cstop conditions; see Materials and
Methods). Since the presence of the C proteins was shown to
represent more stringent conditions for the viral RdRp function (35), it was of interest that initiation at both AGPint and
GPext was taking place as well in the presence of the C proteins
(P/Cwt conditions) (Fig. 1D). It is finally noteworthy that the
two replicated products accumulated with similar kinetics (not
shown), which rules out a possible precursor-product relationship between the two bands and minimizes a possible competition between the two minireplicons for amplification.
Does internal initiation depend on the activity of the external promoter? The results presented above were consistent
with the independent activity of each promoter and could
therefore speak for entry of the viral RdRp at each promoter.
To be consistent with viral RdRp internal entry, replication
from the internal promoter should ideally take place under
conditions in which entry at GPext is prevented. Since no direct
method to prevent or even to assess entry at the promoter is
available, methods that attempted to prevent the viral RdRp
activity from the external promoter were developed. A stretch
of 10 adenosines (template strand) was inserted just after the
inner border of GPext after it was found by serendipity that
such a configuration could significantly lower (or even abolish)
replication initiated at a perfectly competent promoter (compare replication of constructs 1 and 2 with that of 1A10 and
2A10 in Fig. 2B). This decrease or block in replication likely
results from a stuttering of the viral RdRp at the 10A sequence.
Indeed, when the RNA from the 1A10 construct replicated
(Fig. 2B), the sequencing of its RT-PCR product clearly
showed sequence degeneration with further extension of A’s
past the 10A stretch (Fig. 2D, RT-PCR sequence). This should
lead to longer replication products, as observed, while the large
accumulation of heterogeneous shorter products could rather
speak for a backward stuttering (Fig. 2D). Following stuttering,
the 10A stretch may eventually cause the dissociation of the
viral RdRp from its template. Concomitantly, stuttering on the
10A stretch could also result from the production of replicating
molecules out of the rule of six, which would subsequently be
amplified much less efficiently. Whatever the mechanism, the
10A stretch remains an obstacle for the viral RdRp. In no way
J. VIROL.
VOL. 76, 2002
TEMPLATE ENTRY OF PARAMYXOVIRUS RNA POLYMERASE
absence of initiation at GPext [Fig. 3B, lanes 9GP(G91C) and
9A10GP(G91C)]. If anything, in the presence of 10A, internal
initiation could be slightly more efficient [compare lanes 9 and
9GP(G91C) with lanes 9A10 and 9A10GP(G91C), respectively]. Inactivating the internal promoter with the same mutation,
however, led to exclusive replication from the external promoter [Fig. 3B, right panel, lane 9AGP(G91C)]. Note that, in
this configuration, the presence of the 10A stretch prevented
any replication by inhibiting replication from the external promoter [lane 9A10AGP(G91C)]. In general, in the whole series,
the activity of AGPint corresponded to that of the single-GP
construct (Fig. 3, construct 2), which itself is about 5- to 10-fold
lower than that of the single-AGP construct (Fig. 1C, constructs 1 and 2). This was verified even when GPext was mutated [Fig. 3, construct 9 GP(G91C)]. This may indicate that
internal RNA synthesis is somehow less efficient than that
starting at the 3⬘ end of the template.
Despite this observation, this first series of experiments were
supportive of independence of function of the internal and
external promoters and as such were compatible with independent entry of the viral RdRp at each promoter.
Does internal initiation depend on the nature of the external
promoter? A second series of minigenomes carrying double
promoters was prepared, this time with two AGPs, two strong
FIG. 4. Replication of the AGPint-AGPext constructs. (A) Schematic representation of AGPint-AGPext series constructs. Construct 1,
as described in the legend to Fig. 1. Constructs 10 to 13, minireplicons
harboring two AGP promoters, adjacent (construct 10) or separated by
increasing intervening sequence (nucleotides [nt]) as indicated.
(B) Northern blot analysis (see the encapsidated minireplicon RNAs
produced from the plasmids described for panel A). The open rectangle is meant to help show the different migration of the bands. Note the
presence of a single band regardless. (C) Primer extension (PE) analysis (see Materials and Methods) on construct 10 to better demonstrate the preponderant presence of the replicated product initiating at
AGPext. The end position of the product elongated on the RNA is
given by comparison with the sequence performed on the respective
plasmid with the same primer. 3⬘-end sequence (Seq) of AGP,
5⬘. . .TTGTCTGGT-3⬘. Adjacent ribozyme sequence, 5⬘. . .CCGGC
CGA. . .3⬘. Only portions of the sequencing gel relevant for the 3⬘ ends
of the replicated RNAs are shown.
promoters, separated by increasing intervening sequences (Fig.
4A). As shown best by the product obtained with construct 13
(AGP-240-AGP), only one replication signal was observed that
corresponded to the use of the external promoter only (Fig.
4B, lane 13). For the three remaining constructs (construct 10,
11, and 12), the difference in size between the putative products made from the internal or external promoter was too small
to be informative. Therefore, primer extension analysis of the
replicated products was performed (see legend to Fig. 4C and
Materials and Methods).
For construct 10, the almost exclusive product observed extended to the end of AGPext (Fig. 4C, PE 10), a result verified
for the other two constructs (not shown). In conclusion, in
contrast to what had been observed in the context of GPext
(Fig. 3), AGPint was not used in the context of AGPext. A
possible trivial explanation for this discrepancy could be the
higher strength of the external promoter in the series AGPintAGPext. If any components were limiting, then the strong external promoter could outcompete the activity of the internal
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
FIG. 3. Replication of the AGPint-GPext constructs harboring a
10A stretch in the intervening sequence. (A) Schematic representation
of the minireplicons used. 2 and 9 are as described in the legend to Fig.
1. 2A10, 9A10, 9A10GP(G91C), and 9AGPA10(G91C) harbor a 10A
stretch (positive polarity sequence) adjacent to the internal border of
GP (see Materials and Methods). 9GP(G91C), 9A10GP(G91C),
9AGP(G91C), and 9A10AGP(G91C) harbor a G-to-C substitution at
position 91 of GP or AGP. (B) Northern blot analysis of the replicated
minireplicons. When two bands are visible, the upper and lower one
correspond to initiation at GP and AGP, respectively. Open and solid
arrowheads, as for Fig. 1.
7991
7992
VULLIÉMOZ AND ROUX
one. A series of replication assays were therefore performed in
which the template level was progressively reduced, keeping
the other parameters unchanged. No matter how reduced the
replication from the external promoter was, following the reduced delivery of the template plasmid, a higher replication
from the internal promoter was never observed (construct 13)
(Fig. 5B). This argues against a limiting factor that would
prevent activity of the internal promoter.
In Fig. 6, the effect of introducing the 10A stretch between
the two AGPs was analyzed. As found previously, the inhibitory effect of the 10A was less pronounced when it was downstream of a single AGP (Fig. 6B, compare constructs 1 and
1A10). Consequently, replication from the external AGP in the
double promoter was still observed (construct 13A10), although it was partially decreased compared to that of construct
13. Interestingly, for construct 13A10, evidence of internal initiation was present (Fig. 6B, see arrow). The replicated product from the internal promoter was very reproducibly observed
(for another example, see Fig. 5, 13A10) and could represent 15
to 20% of the total (statistical analysis not shown). Decreasing
the template in otherwise unaltered replication conditions did
not change this fraction (Fig. 5, 13A10). Internal initiation was,
in contrast, never observed when the activity of the internal
promoter was inactivated by the G91C mutation [Fig. 6B, 13In(G91C) and 13-A10In(G91C)].
Note that the internal promoter was active in the doubleAGP construct, since upon inactivation of AGPext by the G91C
mutation, replication took place exclusively internally [Fig. 6B,
13-Ex(G91C) and 13-A10Ex(G91C)]. In general, the replication product issued from the double-AGP constructs, either
from AGPext (Fig. 6, construct 13) or from AGPint [Fig. 6,
13-Ex(g91C)], was consistently lower than that produced by the
FIG. 6. Replication of the AGPint-AGPext constructs harboring a
10A stretch in the intervening sequence. (A) Schematic representation
of the constructs. Constructs 1, 13, and 13A10, as described in the
legend to Fig. 5. 13-Ex(G91C) and 13-In(G91C), as for construct 13
but harboring the G91C mutation in the external or the internal AGP,
respectively. 13-A10Ex(G91C) and 13-A10In(G91C), as for 13Ex(G91C) and 13-In(G91C) but with the 10A stretch in the intervening sequence. (B) Northern blot analysis of the replicated products.
Construct 9 (Fig. 3A) was used a size marker for the replication
products initiating at both promoters. Open and solid arrowheads, as
for Fig. 1.
single-AGP construct (Fig. 6, construct 1), alluding to the fact
that, as for the series AGPint-GPext, there is some price to pay
for the presence of the double promoter on the original template. In summary, this AGPint-AGPext series of constructs
appeared to demonstrate that the activity of the external promoter was somehow limiting the activity of the internal one, a
notion that contrasted with the conclusion drawn from the
results obtained with the AGPint-GPext series.
Does internal initiation depend on GP’s being only the external promoter? The difference between the first and second
series of constructs lies in the nature of the external promoter.
In the end, only GPext was associated with internal initiation.
Two new constructs harboring two GP promoters separated by
an intervening sequence of 240 nucleotides were therefore
analyzed, so that, as in the first series, GP was again the
external promoter (Fig. 7). This GPext (construct 14), (Fig. 7B)
was used with the same efficiency as a single-GP construct
(compare with construct 2) and with an efficiency that could
even exceed that exhibited by GPext of the AGPint-GPext construct (construct 9), where the internal promoter is AGP. Despite this high activity, replication from GPint was not apparent
(construct 14, solid arrowhead position). The insertion of the
10A stretch between the two GPs (14A10) reduced the replication from GPext, as expected (compare with construct 2A10).
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
FIG. 5. Gradually diminishing the template availability. (A) Schematic representation of the minireplicons used. Constructs 1 and 13, as
described in the legend to Fig. 4. 13A10 harbors a 10A stretch (positive
polarity sequence) adjacent to the internal border of external AGP.
(B) Northern blot analysis of the replicated products. g, micrograms
of the template plasmid added. Note the presence of a faint signal
below the main band for lanes 13A10, 5 and 0.5 g (see text). Open and
solid arrowheads, as for Fig. 1.
J. VIROL.
VOL. 76, 2002
TEMPLATE ENTRY OF PARAMYXOVIRUS RNA POLYMERASE
7993
Interestingly, replication from GPint was now visible. The replicated product from GPint was, as expected, lower than that
obtained when AGP was the internal promoter (construct 9,
solid arrowhead). However, it reproducibly exceeded by twoor threefold the product issued from GPext. In summary, the
presence of GPext was not sufficient to permit visible internal
initiation.
DISCUSSION
FIG. 8. Schematic summary of the data. The explanations of the symbols are presented at the right of the figure. The two promoters are shown,
in their wild-type and mutated configurations, as well as the intervening sequence and the 10A stretch. Below, the curved arrows portray external
or internal initiations of RNA synthesis, as well as interdicted initiations. The sizes of the arrows reflect the inherent activity of the promoter,
AGP⬎⬎GP. This activity is further modulated by the relative use of the promoter, as indicated by the width of the straight arrows (in the left part
of the figure) depicting the importance of the replicated products. In the left part of the figure, only the two promoters at the 3⬘ end of the
encapsidated T7 RNA transcripts are shown. (A) AGPint-GPext series. (B) AGPint-GPext series. (C) GPint-GPext series.
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
FIG. 7. Replication of the GPint-GPext constructs. (A) Schematic
representation of GPint-GPext series constructs. Construct 2, as described in the legend to Fig. 1. Construct 2A10, as for Fig. 2. Constructs
9 and 9A10, as for Fig. 3. Constructs 14 and 14A10, minireplicons
harboring two GP promoters, adjacent or separated by 240 nucleotides, respectively. (B) Northern blot analysis of the replicated constructs. 2, 2A10, 9, and 9A10 are shown for comparison and as size
markers. 14 and 14A10 constitute the newly analyzed constructs GPintGPext. Open and solid arrowheads, as for Fig. 1. ⴱ, residual non-fully
denatured RNA, occasionally seen upon poor solubilization. Note its
position close to the main band.
The ability of the viral RdRp to initiate replication at a
promoter not positioned at the 3⬘ end of the template is evident in the three series of experiments presented above. This
took place efficiently when the internal promoter was inherently stronger than the external one, as in the series AGPintGPext. In the AGPint-AGPext and GPint-GPext series, where the
two promoters were of equal strength, internal initiation was
seen only when the activity of the external promoter was impaired (10A stretch) or obliterated (G91C mutation) (see schematic summary of the results in Fig. 8).
The mechanism by which this internal initiation took place is
open to question. The following observations have to be taken
into account in order to try to derive such a mechanism. First,
and this is not new, replication from AGP was more efficient
than that from GP. Second, the strength of the promoter appeared to be adjusted by its relative position. Replication from
the external position appeared to be favored relative to that
from the internal position. In consequence, when the two promoters were identical, replication from the external position
alone was observed. Replication from the internal promoter
occurred only when the inherent strength of the promoter
compensated for the internal positioning, as in the first series
7994
VULLIÉMOZ AND ROUX
way “imprint” the viral RdRp with different properties vis-à-vis
the internal promoter recognition. This interpretation, however, as appealing as it appears, does not agree with the results
of the GPint-GPext series, for which it predicts an internal
initiation at GPint, a result that was not observed (Fig. 7,
construct 14).
More generally, these models, although impossible to absolutely disprove, are difficult to reconcile with the results obtained when RNA synthesis is impaired by the G91C mutations
or by the 10A stretch insertion. For example, scanning would
have to take place without apparent RNA synthesis initiation
at the external promoter. Finally, the minireplicons carrying
only the internal promoter could have been generated during
the replication of the opposite strand by the polymerase terminating at the internal promoter. Again, this possibility cannot be formerly excluded. It is, however, not consistent with the
lack of replication resulting from internal initiation in the
AGPint-AGPext or GPint-GPext. In conclusion, although the
data presented do not provide direct and unquestionable evidence for internal entry of the viral RdRp, they can still be best
interpreted in that way.
Furthermore, the possibility of internal entry agrees with
and confirms our previous proposal that the proper N-phase
context is sufficient for the viral RdRp to recognize its template, this at the expense of the template 3⬘-end congruence
that was found not to be critical (36). This implies that the viral
RdRp must detect or interact with the nucleotides of the promoter region despite the tight interactions between the RNA
and the N protein subunits. The three cytidines of the threetimes-repeated motif interacting in position 1 with the N subunits might represent the detected signal. Present in the second helix turn, directly superimposed over the three most
proximal hexamers of the promoter, they exist in a configuration that is not found in any of the 13 other hexamers constituting the promoter (34). Alternatively, some nucleotides
(again the three-times-repeated motif?) of the promoter region might induce the nucleocapsid to adopt a local structure
sufficiently different from the rest of the nucleocapsid to represent this signal. The two hypotheses may not be exclusive.
Until the high-resolution structure of the nucleocapsid is available, it will be difficult to tackle this type of question.
If imprinting vis-à vis the internal promoter recognition, as
discussed above, is not supported by our data, the difference
with which the 10A stretch impairs replication starting at the
AGP or GP could well be interpreted as the difference with
which the viral RdRp negotiates the barrier to RNA synthesis
elongation. Starting at AGP appears to make the viral RdRp
less prone to terminate synthesis when crossing the 10A
stretch, as if the AGP RdRp would have inherently more
processivity than the GP RdRp.
Considering that RNA synthesis initiation at AGP leads
naturally to a replication-only process, while initiation at GP
can proceed to replication or transcription, it would not be
surprising if the two promoters were to imprint the viral RdRp
with a different sensitivity for recognition of downstream signals. In fact, as artificial as the double-promoter constructs
presented here can be—in nature the replication promoters
are normally found at the extremity of the templates—an old
report describing a vesicular stomatitis virus defective interfering (DI) RNA relates to the double promoter and the imprint-
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
(AGPint-GPext). Finally, whenever replication from the external promoter was impaired, replication from the internal promoter took place regardless of the promoter configuration.
These observations are consistent with the ability of the viral
RdRp to initiate replication at either promoter, provided that
the promoter is accessible. While accessibility to the external
promoter in its wild-type configuration is permanent, that of
the internal promoter appears to be regulated by the use of the
external promoter. Taking these observations into consideration, we propose the following mechanism to best account for
the results obtained. Internal initiation reflects internal entry,
an entry regulated by the availability of the internal promoter,
a promoter that can be occluded by the use of the external
promoter. In the light of this model, the higher strength of
AGP over GP allows replication from the internal promoter in
the AGPint-GPext series (Fig. 8A) by minimizing promoter
occlusion resulting from GPext activity. Promoter occlusion is
furthermore decreased by addition of the 10A stretch, seen
here mainly by the decrease in replication from GPext, and not
so much by an increased replication from AGPint, which must
reach the upper limit of the system. Replication from AGPint
takes place a fortiori when the activity of GPext is abolished by
the G91C mutation (see Fig. 3).
For the AGPint-AGPext series (Fig. 8B), both promoters are
of equal strength, so that replication from AGPint cannot overcome the occlusion resulting from AGPext. The 10A stretch, by
decreasing the ability of the viral RdRp to elongate, results in
a slight decrease in promoter occlusion, allowing minimal replication from AGPint. Note that the release of promoter occlusion by the 10A stretch, evidenced by the importance of replication from AGPint, is less important than that seen in the
AGPint-GPext series, a feature that correlates with a lower
effect of the 10A stretch in blocking replication from AGP than
from GP. In contrast, the G91C mutation, by inhibiting replication from AGPext, fully releases promoter occlusion, so that
replication from AGPint becomes maximal (see Fig. 6).
Finally, in the GPint-GPext series, the equal strength of the
internal and external promoters prevents, as in the AGPintAGPext series, the release of promoter occlusion, and only
replication from GPext is observed. As in the AGPint-GPext
series, the 10A stretch almost completely abolishes replication
from GPext and in this way releases promoter occlusion, with
concurrent appearance of replication from GPint. Although
weak, replication from GPint may in fact correspond to the
highest possible replication from a weak promoter positioned
internally. It is indeed higher than that initiating at GPext (see
Fig. 7).
Although this model appears to account most fully for the
results obtained, other possibilities can be envisaged. These
include external entry followed by template scanning to reach
the internal promoter or by reinitiation at the internal promoter after release of the nascent chain initiated at the external promoter. These models could integrate the effect that
initiation at the external promoter could have on the ability of
the viral RdRp to recognize the internal promoter. Considering the AGPint-GPext and AGPint-AGPext series, one could
conclude that a viral RdRp starting replication at GPext would
still be in a position to recognize AGPint, while an initiation at
AGPext would render the viral RdRp insensitive to an internal
promoter. Entry and initiation at GPext or AGPext would in a
J. VIROL.
VOL. 76, 2002
TEMPLATE ENTRY OF PARAMYXOVIRUS RNA POLYMERASE
ing concepts (21). An internal DI deletion (DI-LT2) carrying,
before the genomic 3⬘ end, 70 nucleotides complementary to
the 5⬘ end of the vesicular stomatitis virus genome (in other
words, carrying an AGP sequence in front of the normal GP
sequence) was described. When this template was assayed in
vitro for RNA synthesis, neither leader RNA nor capped vesicular stomatitis virus mRNA was detected, suggesting that
the viral RdRp starting on AGP was not sensitive to downstream signals. The concept of promoter imprinting is thus
appealing, but further studies, under way, will be required to
estimate its reality.
ACKNOWLEDGMENTS
REFERENCES
1. Baron, M. D., and T. Barrett. 1997. Rescue of rinderpest virus from cloned
cDNA. J. Virol. 71:1265–1271.
2. Baudin, F., C. Bach, S. Cusack, and R. W. H. Ruigrok. 1994. Structure of
influenza virus RNP. I. Influenza virus nucleoprotein melts secondary structure in panhandle RNA and exposes the bases to the solvent. EMBO J.
13:3158–3165.
3. Boyer, J.-C., and A.-L. Haenni. 1994. Infectious transcripts and cDNA clones
of RNA viruses. Virology 198:415–426.
4. Bridgen, A., and R. M. Elliott. 1996. Rescue of a segmented negative-strand
RNA virus entirely from cloned complementary DNAs. Proc. Natl. Acad.
Sci. USA 93:15400–15404.
5. Cadd, T., D. Garcin, C. Tapparel, M. Itoh, M. Homma, L. Roux, J. Curran,
and D. Kolakofsky. 1996. The Sendai paramyxovirus accessory C proteins
inhibit viral genome amplification in a promoter-specific fashion. J. Virol.
70:5067–5074.
6. Calain, P., and L. Roux. 1993. The rule of six, a basic feature for efficient
replication of Sendai virus defective interfering RNA. J. Virol. 67:4822–4830.
7. Calain, P., and L. Roux. 1995. Functional characterisation of the genomic
and antigenomic promoter of the Sendai virus. Virology 211:163–173.
8. Collins, P. L., M. G. Hill, E. Camargo, H. Grosfeld, R. M. Chanock, and
B. R. Murphy. 1995. Production of infectious human respiratory syncytial
virus from cloned cDNA confirms an essential role for the transcription
elongation factor from the 5⬘ proximal open reading frame of the M2 mRNA
in gene expression and provides a capability for vaccine development. Proc.
Natl. Acad. Sci. USA 92:11563–11567.
9. Conzelmann, K.-K. 1996. Genetic manipulation of nonsegmented negativestrand RNA viruses. J. Gen. Virol. 77:381–389.
10. Curran, J., and D. Kolakofsky. 1990. Sendai virus P gene produces multiple
proteins from overlapping open reading frames. Enzyme 44:244–249.
11. Durbin, A. P., S. L. Hall, J. W. Siew, S. S. Whitehead, P. L. Collins, and B. R.
Murphy. 1997. Recovery of infectious human parainfluenza virus type 3 from
cDNA. Virology 235:323–332.
12. Egelman, E. H., S.-S. Wu, M. Amrein, A. Portner, and G. Murti. 1989. The
Sendai virus nucleocapsid exists in at least four different helical states.
J. Virol. 63:2233–2243.
13. Erlenhoefer, C., W. J. Wurzer, S. Loffler, S. Schneider-Schaulies, V. ter
Meulen, and J. Schneider-Schaulies. 2001. CD150 (SLAM) is a receptor for
measles virus but is not involved in viral contact-mediated proliferation
inhibition. J. Virol. 75:4499–4505.
14. Fuerst, T. R., E. G. Niles, F. W. Studier, and B. Moss. 1986. Eukaryotic
transient-expression system based on recombinant vaccinia virus that synthesizes bacteriophage T7 RNA polymerase. Proc. Natl. Acad. Sci. USA
83:8122–8126.
15. Garcia-Sastre, A., and P. Palese. 1993. Genetic manipulation of negativestrand RNA virus genomes. Annu. Rev. Microbiol. 47:765–790.
16. Garcin, D., T. Pelet, P. Calain, L. Roux, J. Curran, and D. Kolakofsky. 1995.
A highly recombinogenic system for the recovery of infectious Sendai
paramyxovirus from cDNA: generation of a novel copy-back nondefective
interfering virus. EMBO J. 14:6087–6094.
17. He, B., R. G. Paterson, C. D. Ward, and R. A. Lamb. 1997. Recovery of
infectious SV5 from cloned DNA and expression of a foreign gene. Virology
237:249–260.
18. Hoffman, M. A., and A. K. Banerjee. 1997. An infectious clone of human
parainfluenza virus type 3. J. Virol. 71:4272–4277.
19. Iseni, F., F. Baudin, D. Blondel, and R. W. Ruigrok. 2000. Structure of the
RNA inside the vesicular stomatitis virus nucleocapsid. RNA 6:270–281.
19a.Iseni, F., F. Baudin, D. Garcin, J.-B. Marq, R. W. H. Ruigrok, and D.
Kolakofsky. Chemical modification of nucleotide bases and mRNA editing
depend on the hexamer or nucleoprotein phase in Sendai virus nucleocapsids. RNA, in press.
20. Kato, A., Y. Sakai, T. Shioda, T. Kondo, M. Nakanishi, and Y. Nagai. 1996.
Initiation of Sendai virus multiplication from transfected cDNA or RNA
with negative or positive sense. Genes Cells 1:569–579.
21. Keene, J. D., I. M. Chien, and R. A. Lazzarini. 1981. Vesicular stomatitis
virus defective interfering particle containing a muted internal leader RNA
gene. Proc. Natl. Acad. Sci. USA 78:2090–2094.
21a.Kolakofsky, D., C. Bellocq, and R. Raju. 1987. The translational requirement
for La Crosse virus S-mRNA synthesis. Cold Spring Harb. Symp. Quant.
Biol. 52:373–379.
22. Kolakofsky, D., T. Pelet, D. Garcin, S. Hausmann, J. Curran, and L. Roux.
1998. Paramyxovirus RNA synthesis and the requirement for hexamer genome length: the rule of six revisited. J. Virol. 72:891–899.
23. Lamb, R. A., and D. Kolakofsky. 2001. Paramyxoviridae: the viruses and their
replication, p. 1305–1340. In D. M. Knipe and P. M. Howley (ed.), Fields
virology, 4th ed. Lippincott Williams & Wilkins, Philadelphia, Pa.
24. Lawson, N. D., E. A. Stillman, M. A. Whitt, and J. K. Rose. 1995. Recombinant vesicular stomatitis viruses from DNA. Proc. Natl. Acad. Sci. USA
92:4477–4481.
25. Mottet, G., and L. Roux. 1989. Budding efficiency of Sendai virus nucleocapsids: influence of size and ends of the RNA. Virus Res. 14:175–188.
26. Murphy, S. K., Y. Ito, and G. D. Parks. 1998. A functional antigenomic
promoter for the paramyxovirus simian virus 5 requires proper spacing
between an essential internal segment and the 3⬘ terminus. J. Virol. 72:10–
19.
27. Murphy, S. K., and G. D. Parks. 1999. RNA replication for the paramyxovirus simian virus 5 requires an internal repeated (CGNNNN) sequence
motif. J. Virol. 73:805–809.
28. Neubert, W. J., C. Eckerskorn, and H. E. Homann. 1991. Sendai virus NP
gene codes for a 524 amino acid NP protein. Virus Genes 5:25–32.
29. Radecke, F., P. Spielhofer, H. Schneider, K. Kaelin, M. Huber, C. Dötsch, G.
Christiansen, and M. A. Billeter. 1995. Rescue of measles viruses from
cloned DNA. EMBO J. 14:5773–5784.
30. Schnell, M. J., T. Mebatsion, and K.-K. Conzelmann. 1994. Infectious rabies
viruses from cloned cDNA. EMBO J. 13:4195–4203.
31. Shioda, T., Y. Hidaka, T. Kanda, H. Shibuta, A. Nomoto, and K. Iwasaki.
1983. Sequence of 3⬘ 687 nucleotides from the 3⬘ end of Sendai virus genome
RNA and the predicted amino acid sequences of viral NP, P and C proteins.
Nucleic Acids Res. 11:7317–7330.
32. Shioda, T., K. Iwasaki, and H. Shibuta. 1986. Determination of the complete
nucleotide sequence of the Sendai virus genome RNA and the predicted
amino acid sequences of the F, HN, and L proteins. Nucleic Acids Res.
14:1545–1563.
33. Tapparel, C., S. Hausmann, T. Pelet, J. Curran, D. Kolakofsky, and L. Roux.
1997. Inhibition of Sendai virus genome replication due to promoter-increased selectivity: a possible role for the accessory C proteins. J. Virol.
71:9588–9599.
34. Tapparel, C., D. Maurice, and L. Roux. 1998. The activity of Sendai virus
genomic and antigenomic promoters requires a second element past the
leader template regions: a motif (GNNNNN)3 is essential for replication.
J. Virol. 72:3117–3128.
35. Tapparel, C., and L. Roux. 1996. The efficiency of Sendai virus genome
replication: the importance of the RNA primary sequence independent of
terminal complementarity. Virology 225:163–171.
36. Vulliemoz, D., and L. Roux. 2001. “Rule of six”: how does the Sendai virus
RNA polymerase keep count? J. Virol. 75:4506–4518.
37. Whelan, S. P. J., L. A. Ball, J. N. Barr, and G. W. Wertz. 1995. Efficient
recovery of infectious vesicular stomatitis virus entirely from cDNA clones.
Proc. Natl. Acad. Sci. USA 92:8388–8392.
Downloaded from http://jvi.asm.org/ on July 24, 2015 by guest
We, particularly D.V., thank Dominique Garcin for exciting discussions and Daniel Kolakofsky for critical reading of the manuscript.
This work was supported by a grant from the Swiss National Foundation for Scientific Research.
7995