ABSTRACT
The specific recognition of genomic positive strand RNAs as templates for the
synthesis of intermediate negative strands by the picornavirus replication
machinery is presumably mediated by
cis
-acting sequences within the genomic RNA 3
'
non-coding region (NCR). A structure-infectivity analysis was conducted on the 44 nt human rhinovirus 14
(HRV14) 3
'
NCR to identify the primary sequence and/or secondary structure determinants
required for viral replication. Using biochemical RNA secondary structure
probing techniques, we have demonstrated the existence of a single stem-loop structure contained entirely within the 3
'
NCR, which appears to be phylogenetically conserved within the rhinovirus
genus. We also report the
in vivo
analysis of a number of 3
'
NCR deletion mutations engineered into infectious cDNA clones which were
designed to disrupt the stem-loop secondary structure to varying degrees. Large deletions (up to 37
nt) resulted in defective growth phenotypes, although they were not lethal. We
propose that the absolute requirements for initiation of negative strand
synthesis are less stringent than previously postulated, even though defined
RNA secondary structure determinants may have evolved to facilitate and/or
regulate the process of viral RNA replication.
RNA-protein interactions are ubiquitous in nature and direct numerous
cellular processes intimately involved in the regulation of gene expression.
The ultimate expression of a eukaryotic gene product requires not only ribosome
recognition and utilization of a suitable mRNA template, but also pre-mRNA splicing as well as mRNA capping, transport, stabilization and
eventual degradation in a cell. Positive strand (mRNA sense) RNA viruses
utilize much of the host cell machinery to express viral-encoded gene products required to complete the virus life cycle. In
addition to some of the RNA-protein interactions characteristic of cellular messages, most RNA
viruses must also maintain a mechanism for the specific replication of the
virus genome in the presence of an abundance of cellular cytoplasmic RNAs. In
the case of picornaviruses, an important
cis
-acting molecular genetic determinant for this recognition process is
believed to reside in the 3' non-coding region (3' NCR) of positive strand genomic RNA. Primary sequence
determinants and/or secondary structure motifs in this region, in the context
of the polyadenylated RNA molecule, are presumably recognized by the viral RNA
replication complex to initiate the synthesis of negative strand RNA
intermediates (
1
,
2
).
The prototypic member of the
Picornaviridae
, poliovirus type 1 (PV1), has a 3' NCR of 72 nt immediately downstream of the two stop codons at the end of
the polyprotein coding region (
3
,
4
). Computer-generated RNA secondary structure predictions suggest the existence of a
pseudoknot structure in the PV1 3' NCR (
5
), which has been partially biochemically confirmed (
6
). The genomic RNA of human rhinovirus type 14 (HRV14), a closely related
picornavirus, contains a 3' NCR that is 44 nt in length and follows a single stop codon at the end
of the polyprotein coding sequence (
7
). The small size of the HRV14 3' NCR makes it a particularly attractive target for genetic manipulation
in order to identify the molecular features of this region of RNA required for
replication complex recognition to initiate the synthesis of viral negative
strand RNAs. In addition, an understanding of these features may ultimately aid
in the design of antiviral strategies against HRV, a major causitive agent of
the common cold.
In the following study, we have investigated the RNA secondary structure of the
HRV14 3' NCR and attempted to identify the primary sequence and/or secondary
structure determinants in the region required for viral RNA replication. Using
biochemical RNA secondary structure probing techniques, we have demonstrated
the existence of a single stem-loop structure, similar to that predicted by Pilipenko and colleagues (
5
). This single stem-loop structural motif appears to be conserved among different members of
the rhinovirus genus based on phylogenetic comparison (A.C.Palmenberg, personal
communication) and computer-predicted RNA secondary structure determination (
8
). We also report the
in vivo
analysis of a number of 3' NCR deletion mutations engineered into infectious cDNA clones.
Surprisingly, large deletions are tolerated within this region of RNA, although
they result in defective growth phenotypes. The largest engineered deletion
eliminates 37 nt of the HRV14 44 nt 3' NCR, eliminating the possibility of formation of any higher order RNA
structure resembling the wild-type stem-loop. These findings suggest that while the rhinoviruses have
evolved a highly conserved, predicted stem-loop in the 3' NCR of their genomic RNAs which may affect the efficiency of
template utilization by the viral replication complex, the absolute
requirements for viral RNA replication are much less stringent than previously
proposed. Our data support a mechanism for picornavirus negative strand RNA
replication initiation which utilizes functions that do not absolutely require
a highly specific
cis
-acting RNA recognition determinant at the 3'-end of virus genomic positive strand RNAs.
The construction of a nested series of 3' co-terminal, subgenomic, T7-based transcription vectors using the full-length HRV14 cDNA construct pT7RV(F.L.) has been
described, as well as the construction of an 8 nt deletion mutation in the cDNA
corresponding to the HRV14 3' NCR (
2
). Transcription vectors containing larger deletions in the cDNA sequence
corresponding to the HRV14 RNA 3' NCR (i.e. 18 and 21 nt) were constructed in a similar manner (
9
) using the previously described synthetic oligonucleotide, RV10[Delta](-) (5'-TGTTAACCTAAAAGAGGTCC-3') and an additional oligonucleotide,
RV14[Delta](+) (5'-GAGTAGAAGTAGGAGTTTAT-3'). A mutagenesis cassette was engineered into
the HRV14 3' NCR cDNA sequence by site-directed mutagenesis at nt 7196 using the heteroduplex method (
10
) with the mutagenesis oligonucleotide RVU7196A/G (5'-CACTTAATTTGAGRAGAAGTAGG-3', where R is A or G). The existing sequence, 5'-GAGTAG-3', in the wild-type cDNA was changed
to 5'-GAGGAG-3' to create the recognition sequence of the restriction
endonuclease
Bse
RI (5'-GAGGAG(N)
10/8
-3'). Digestion of the resulting plasmid with
Hpa
I (7169) and
Bse
RI (7193), followed by repair with the Klenow enzyme, religation and
transformation into
Escherichia coli
C600 cells resulted in the elimination of a 37 bp fragment from the 3' NCR cDNA sequence. Mutations in subgenomic cDNA constructs were
subsequently cloned into pT7-HRV14(ST), a reconstructed full-length HRV14 cDNA clone (S. Todd and B. L. Semler, unpublished
results). The resulting cDNA sequences of these mutations are described in
Results. All plasmids were sequenced using the modified T7 DNA polymerase.
Synthesis of non-radiolabeled RNAs was performed using the MEGAshortscript kit (Ambion)
with ~4 [mu]g of the appropriate
Pst
I- or
Cla
I-linearized plasmid DNA templates. The RNA was phenol/chloroform extracted
and ethanol precipitated. Aliquots (0.5 [mu]g/[mu]l) were then stored at -70oC.
Approximately 120 pmol of oligonucleotide RVoligoT+9(-) were incubated in the presence of 100 [mu]Ci 6000 Ci/mmol [[gamma]-
32
P]ATP and 60-100 U T4 polynucleotide kinase in 100 [mu]l for 45 min at 37oC. Unincorporated radiolabeled nucleotides were removed using a
Sephadex G-50 spin column. The resulting 5'-end-labeled oligonucleotides had a specific activity of ~3 * 10
5
c.p.m./pmol.
RNA secondary structure probing was performed using conditions modified from
published methods (
6
,
11
,
12
). Briefly, 0.5 [mu]g HRV14 3' NCR-specific RNA was incubated in the presence of 40 [mu]g
E.coli
tRNA in 0.7* TMK buffer (30 mM Tris-HCl, pH 7.4, 10 mM MgCl
2
, 270 mM KCl) or in TM(-K) buffer supplemented with different amounts of KCl or NaCl under
reducing conditions (18 mM 2-mercaptoethanol), in a total volume of 40 [mu]l. The RNA was then incubated successively at 68oC and 37oC and room temperature for 5 min each to allow RNA secondary
structure to form. Enzymatic treatment with RNase V
1
, T
1
, A, U
2
or an RNase isolated from
Baccillus cereus
was performed at room temperature for 5 min or on ice as detailed in Results.
The enzyme reactions were stopped by the addition of 155 [mu]l enzyme stop solution (0.3 M NaOAc, 10 mM EDTA, 0.3% SDS) followed by
phenol/chloroform extraction and precipitation with 2.5 vol. ethanol. Primer
extension was performed essentially as described in Eisenberg
et al.
(
13
) using 2-4 pmol (~10
6
c.p.m.) 5'-end-labeled oligonucleotide RVoligoT+9(-) and 10 U AMV reverse transcriptase (Life Sciences)
at 50oC for 20-30 min in primer extension buffer (20 mM Tris-HCl, pH 7.4, 10 mM MgCl
2
, 6 mM DTT, 300 [mu]M each dNTP, 80 [mu]g/ml actinomycin D). The extension reactions were phenol/chloroform
extracted, ethanol precipitated, resuspended in a formamide loading buffer and
resolved on 8% polyacrylamide-7 M urea gels. Sequencing ladders were generated from the appropriate
transcription vector using the modified T7 DNA polymerase with 5'-end-labeled oligonucleotide RVoligoT+9(-) as a primer. The gels were then dried and exposed to
Kodak AR or MR X-ray film.
Direct RNA secondary structure probing was performed using
in vitro
transcribed RNA which was dephosphorylated with intestinal alkaline phosphatase
and 5'-end-labeled with T4 polynucleotide kinase in the presence of 50-100 [mu]Ci 6000 Ci/mmol [[gamma]-
32
P]ATP. Unincorporated radiolabeled nucleotides were removed using a Sephadex G-50 spin column in TE8 buffer (10 mM Tris, 1 mM EDTA, pH 8.0) with 0.1% SDS
followed by a final ethanol precipitation in 2.5 M ammonium acetate. Enzymatic
treatment of ~0.5 [mu]g RNA was performed as described above, also in the presence of 40 [mu]g/reaction
E.coli
tRNA and 18 mM 2-mercaptoethanol. Reactions were stopped with enzyme stop buffer,
phenol/chloroform extracted, ethanol precipitated, resuspended in formamide or
8 M urea loading dye and analyzed as described above.
In vitro
transcription of full-length virus-specific RNAs from wild-type and mutated
Pst
I-linearized pT7-HRV14(ST)-based cDNA constructs was performed as described previously (
14
). DEAE-mediated RNA transfection of R19 HeLa cells and rhinovirus propagation in
tissue culture have also been described (
2
,
14
). Asymmetric RT-PCR sequencing of viral RNAs from total cytoplasmic RNA harvested from
infected monolayers (
15
) was performed as described in Todd
et al.
(
2
) using oligonucleotide primer RV7035(+) (5'-GCATGTTAGCATGGCACTCAGG-3'), which is identical to nt 7035-7056 within the polymerase coding sequence of
HRV14, along with either RVoligoT+9(-) (5'-TTTTTTTTATAAACTCC-3') or RVoligoT+2(-) (5'-TTTTTTTTTTTTTTTAT-3'), which are
complementary to the 3'-end of virus positive strand RNAs.
The computer-predicted RNA secondary structures of two HRV14 3'-end-specific RNA sequences using the Zuker FoldRNA algorithm (
16
) are shown in Figure
1
. In order to allow the direct comparison of secondary structures predicted
using the computer algorithms with subsequent biochemical probing data, the
sequence of the RNAs initially folded by computer were similar to those which
could be synthesized from available transcription vectors
in vitro
.
Pst
I-linearized, T7-based plasmids pT7RV7168A+, pT7RV7136A+, pT7RV7076A+ and pT7RV6338A+
served as transcription templates to generate RNAs which contain the HRV14 3' NCR with poly(A)
60-80
and 0, 32, 98 or 831 nt of 5'-proximal (3D
pol
) RNA sequence, respectively (
2
). RNAs with no sequences upstream of the 3' NCR or 36 nt of 3D
pol
coding sequence give rise to the type of secondary structure shown in Figure
1
A, which contains a single predominant stem-loop structure within the 3' NCR. The stop codon lies outside the stem structure (nt 7166-7168). Larger RNAs, with 98 or 831 nt of 5'-proximal sequence, give rise to a different
computer-predicted secondary structure, in which the stop codon lies outside a
different helical region, as shown in Figure
1
B. The predicted free energy change ([Delta]
G
) for the formation of the RNA secondary structure shown in Figure
1
A is -4.7 kcal/mol. The computer-predicted [Delta]
G
value using longer RNAs is more favorable (-10.4 kcal/mol for the 7136A+ RNA), although the 3' NCR stem-loop structure is unchanged. These modest -[Delta]
G
values made the computer predictions for RNA secondary structure unconvincing
by themselves, prompting the biochemical investigation of the RNA higher order
structure. The Zuker MFold algorithm (
8
), which is used for subsequent computer-generated RNA secondary predictions, predicts an RNA secondary structure
for the 7076A+ RNA which contains a helical region corresponding to the stem-loop shown in Figure
1
A.
The 3' NCR of HRV14 was subjected to biochemical analysis by enzymatic
treatment followed by primer extension analysis of the resulting RNA cleavage
products. The enzymes (and their specificities) used in this study included
RNase V
1
(dsRNA), T
1
(Gp <=> N), A (Cp <=> N, Up <=> N), U
2
(Ap <=> N) and an RNase isolated from
B.cereus
(herein refered to as
Bc
) (Cp <=> N, Up <=> N). Figure
2
A shows the results of secondary structure probing with all the above listed
enzymes on 7076A+ RNA generated from the pT7RV7076A+ transcription vector. The
reactions were performed either at room temperature for 5 min (odd numbered
lanes) or on ice for 20 min (even numbered lanes). The 5'-end-labeled 3' NCR-specific deoxyoligonucleotide primer RVoligoT+9(-) was used for both primer extension
analysis of the resulting RNA fragments as well as for generating the
accompanying sequencing ladder by dideoxy sequencing of the pT7RV7076A+ plasmid
DNA (lanes 1-4). As a result of the different mechanisms of cDNA termination (the
incorporation of a dideoxynucleotide nucleotide analog compared with template
scission), the reverse transcribed primer extension products are 1 nt shorter
than the corresponding cDNA product of the sequencing ladder [e.g. the primer
extension product resulting from cleavage after G(7193) by RNase T
1
co-migrates with position 7194 of the sequencing ladder].
Direct RNA secondary structure probing using [gamma]-
32
P-5'-end-labeled HRV14 3' NCR-specific RNA was employed to confirm the
results obtained using the primer extension method as well as to assess the
role of the 3'-terminal 9 nt of the 3' NCR and the poly(A)
n
tract in the formation of RNA secondary structure. Since the relative distance
of the RNA sequence of interest from the radiolabeled 5'-terminal nucleotide of the template RNA determines the number of
nucleotides to be resolved by polyacrylamide gel electrophoresis, shorter RNAs
were used for direct structure probing than those employed in the primer
extension experiments shown in Figure
2
. Figure
3
A shows the results of such probing using 5'-end-labeled 7136A+ and 7136A- RNAs. The 7136A- RNA lacks a poly(A) tract but contains the
intact HRV14 3' NCR and terminates with two non-viral nucleotides (-CG-3') acquired from the
Cla
I restriction site used to linearize the pT7RV7136A- transcription vector.
As described in Materials and Methods, most partial RNase digests were carried
out in 200 mM KCl (final concentration) supplied by TMK buffer. To determine
the effect of salt concentration on RNA secondary structure, enzymatic probing
was carried out in TM(-K) buffer, which lacked KCl, either with no additional salt or
supplemented with 75, 150 or 300 mM NaCl or KCl (final concentration).
Increasing NaCl concentration resulted in increased helical secondary
structure, as demonstrated by RNase V
1
sensitivity between nucleotides 7184 and 7192 and 7208 and 7210, inclusive
(data not shown). Numerous nucleotides between positions 7166 and 7192, most
notably U(7177), U(7179) and U(7187), became less susceptible to single strand-specific nuclease attack with increased salt concentration, while U(7196),
in the proposed distal loop region, remained RNase A-sensitive regardless of the salt concentration. The bulged G(7181) was
consistently RNase T
1
sensitive even at higher NaCl concentrations. Similar results were obtained
when the reactions were supplemented with KCl instead of NaCl (data not shown).
RNA secondary structure probing of 5'-end-labeled 7168A+ and 7168A- (which contain no 5'-proximal 3D
pol
coding sequence) under the identical range of NaCl and KCl concentrations
showed the same RNase sensitivity pattern (data not shown), indicating that 5'-proximal nucleotides were not interacting with the distal stem-loop under high salt conditions to alter the observed RNA
secondary structure. The predominant stem-loop structure in the 3' NCR was thus detected reproducibly under moderate to high salt
conditions (bracketing physiological intracellular conditions) and was
disturbed by extremely low salt conditions.
A series of deletion mutations was generated in the HRV14 3' NCR. An 8 nt deletion ([Delta]8) in the 3' NCR of HRV14 has been described previously (
2
). This sequence was selected for site-directed mutagenesis based on its conservation in the 3' NCRs of other rhinovirus genomes (A.C.Palmenberg, personal
communication) and its disruptive effect on the computer-predicted secondary structure of the 3' NCR. The [Delta]8 virus showed delayed onset of RNA replication and reduced
accumulation of viral RNA in infected cells, based on RNA slot blot analysis
using virus derived from the pT7RV(F.L.)-based cDNA transcription vector (
2
). In order to further investigate the requirement for an intact stem-loop structure in the HRV14 3' NCR to support viral replication, larger deletion mutations were
engineered which were designed to abolish formation of the distal stem-loop structure. An 18 nt deletion ([Delta]18; nt 7175-7192) results in the deletion of half the 3' NCR stem sequence, while a 21 nt deletion ([Delta]21) results in the additional deletion of most of
the loop region (nt 7193-7196). A 37 nt deletion ([Delta]37; nt 7172-7209) results in the removal of all but 7 nt of the
HRV14 3' NCR (5'-GTTTTAT-3'), excluding the possibility that any elaborate
RNA secondary structure can be assumed by the remaining RNA sequence.
The [Delta]18, [Delta]21 and [Delta]37 mutations were constructed into the pT7-HRV14(ST) plasmid background to generate RNAs for
transfection into tissue culture cells to examine their abilities to produce
infectious rhinovirus. The [Delta]8 mutation, originally studied in a different infectious cDNA plasmid
background (
2
), was also cloned into a pT7-HRV14(ST)-based transcription vector. Surprisingly, genome-length RNAs bearing these larger deletions gave rise to viral
progeny with impaired growth characteristics similar to those of the original [Delta]8 virus. While DEAE-mediated transfection of wild-type HRV14 RNA resulted in complete destruction of a HeLa R19
monolayer in <48 h, mutant HRV14 viruses required 10-11 days to effect complete cell lysis of a monolayer following separate
transfections with [Delta]8, [Delta]18 or [Delta]37 RNAs. Wild-type, [Delta]8, [Delta]18 and [Delta]37 virus isolates obtained from
liquid overlays were used to infect additional HeLa R19 cell monolayers and
total cytoplasmic RNA was harvested from monolayers beginning to show
cytopathic effects (CPE) following infection (8, 16, 15 and 13.5 h,
respectively). The resulting RNA was then subjected to asymmetric RT-PCR sequencing as described previously (
2
) using the RV7035(+)/RVoligoT+9(-) or the RV7035(+)/RVoligoT+2(-) primer set. The 3'-end of RVoligoT+9(-) contains 9 nt which are complementary to the
wild-type HRV14 3' NCR and should therefore anneal to wild-type, [Delta]8 and [Delta]18 (or [Delta]21) RNAs but not to [Delta]37 RNAs, which harbor a deletion
extending into this region of complementarity. RVoligoT+2(-) will amplify any RNA sequence terminating with 5'-AT(A)
n
-3', including the [Delta]37 RNAs. The identities of the wild-type, [Delta]8 and [Delta]18 viruses were confirmed using the
RV7035(+)/RVoligoT+9(-) primer set, while no sequence was obtained from RNA isolated from [Delta]37-infected monolayers (arguing against the possibility of
virus stock contamination; data not shown). Wild-type and [Delta]37 RNAs were then sequenced following amplification using the
RV7035(+)/RVoligoT+2(-) primer set, which confirmed the existence of the [Delta]37 lesion in the transfection-derived virus. The demonstration of 3' NCR deletions in virus harvested from cells showing
CPE, which correspond to the deletions engineered into the pT7-HTV14(ST)-based transcription vectors, is compelling evidence that the mutated
RNAs are capable of being autonomously replicated by the viral RNA replication
machinery.
We have used RNA secondary structure probing techniques to examine the structure
of the wild-type and mutated HRV14 RNA 3' NCRs in solution. A diagrammatic representation of the wild-type HRV14 3' NCR secondary structure, based on the data presented
here, is shown in Figure
5
. The 3' NCR appears to fold into a single stem-loop structure which does not involve 5'-proximal 3D
pol
coding sequence or the 3' poly(A) tract. There are several bulged (i.e. non-paired) nucleotides along the length of the stem between nt ~7176 and 7192, notably G(7181) and A(7189). G(7201) was
slightly more sensitive than neighboring G(7198) and G(7204/5), supporting the
model in which it is bulged opposite A(7189). The inability to detect the
computer-predicted bulge A(7184) or A(7206) by biochemical probing is noteworthy.
Despite the inability to detect RNase V
1
cleavage between nucleotides ~7176 and 7182, we believe this region is base paired because of the absence
of cleavage by single strand-specific RNases and the series of RNase V
1
-sensitive U nucleotides at the 3'-end of the 3' NCR (nt 7208-7210) (
17
). Secondary structure probing using shorter RNAs, including the 3' NCR with only two non-viral nucleotides at the 3'-end (7168A-) indicates that any secondary structure model
must account for the base pairing within the stem structure without involving
nucleotides outside the 3' NCR (data not shown). The residues at positions A(7184) and/or A(7206)
are not sensitive to single strand-specific RNases, although there is no obvious means of base pairing these
nucleotides without generating extremely unlikely RNA secondary structure
conformations (based upon MFold predictions). It is more plausible that the
opposing bulged nucleotides are not susceptible to RNase U
2
as a result of their orientation within the flanking helical structure or that
the nucleotides are non-canonically paired within the helix (
18
). DMS methylation followed by primer extension did show A(7184) to be
methylation sensitive (data not shown), however, no DMS data are available for
A(7206) due to the limitations of the primer extension method. The ability to
detect G(7181) with RNase T
1
treatment is probably due to: (i) breathing of the weak G(7181)-U(7209) base pair flanked by A-U base pairs; (ii) the existence of a repeating dinucleotide
sequence between nucleotides 7176 and 7179 (5'-AUAU-3') which could allow an alternative stem structure to
form displacing G(7181); (iii) the overall weak duplex structure predicted to
exist at the base of the 3' NCR stem-loop (for a review see
18
).
The RNA secondary structure outside the 3' NCR was less striking than in the 3' NCR itself, although there were clearly distinct helical and
single-stranded segments of RNA. We propose the existence of a short stem-loop structure immediately upstream of the stop codon with a
stretch of pyrimidines (5'-UCUUUU-3') in the loop region, however, this structure is
probably a fortuitous result of the 3D
pol
coding sequence. Our preliminary data suggest that the formation of this
upstream stem-loop structure is not required for viral infectivity (unpublished
observations). The genomic RNA sequences of HRVs 1A, 1B, 2, 9, 16, 85 and 89
are distinctly different from HRV14 in this region of the genome in that they
lack the 8 nt pyrimidine tract (HRV14 nt 7159-7166) and instead have a string of four adenosines preceding a UUU codon
(UUC in the case of HRV16) for phenyalanine, which is highly conserved as the C-terminal amino acid of picornavirus 3D
pol
polypeptides (A.C.Palmenberg, personal communication). Although an attractive
possibility and consistent with secondary structure models for the 3' NCR of EMCV (
19
), we have found no experimental evidence that the four uridylate residues (nt
7163-7166) interact with the polyadenosine tract of the HRV14 genomic RNA. In
addition, we have found no evidence for the existence of any long range RNA-RNA interactions or pseudoknot structures involving the HRV14 3' NCR, as have been described for the prototypic picornavirus,
poliovirus (
5
,
6
).
The data obtained from mutagenesis of the HRV14 3' NCR suggest that while a stem-loop structure is phylogenetically conserved among the
rhinoviruses, it is not absolutely required for initiation of RNA replication.
The deletion of 8 nt at the base of the stem-loop structure results in a severely debilitated RNA replication
phenotype
in vivo
which we previously postulated was the result of the abrogation of an RNA-protein interaction between the HRV14 3' NCR and a 34-36 kDa host cell protein (
2
). Investigation of this RNA-protein interaction using the [Delta]18, [Delta]21 and [Delta]37 RNAs suggests that the extreme 3'-end of the 3' NCR may be a major molecular
determinant in this interaction, although an intact stem-loop may be the preferred binding site for the host proteins (unpublished
data). Further deletion mutagenesis clearly demonstrates that maintenance of
the 3' NCR stem-loop structure is not absolutely essential for virus infectivity
and, hence, replication complex recognition and utilization of the mutated RNA
template. The infectivity of RNAs harboring the [Delta]37 mutation was most remarkable. Approximately 84% of the 44 nt 3' NCR was deleted in the [Delta]37 3' NCR, leaving only the primary sequence 5'-GTTTTAT-3' between the stop codon
and the poly(A)
n
tract. Nonetheless, virus recovered from a [Delta]37 RNA transfection displayed a growth phenotype similar to that of the [Delta]8, [Delta]18 and [Delta]21 viruses.
Several models have been proposed to explain the initiation of virus negative
strand synthesis from a genomic positive strand RNA template (for a review see
1
). One model suggests that a uridylylated VPg molecule (VPg-pU-pU), perhaps in the context of a larger protein precursor (i.e. 3AB)
serves as a primer for the viral RNA-dependent RNA polymerase (3D
pol
) (
20
,
21
). Uridylylation, polymerase priming and proteolytic maturation of viral
replication proteins could occur as concerted events within a membrane bound
replication complex. A second model proposes that negative strand synthesis is
initiated upon the formation of a snap-back hairpin loop structure involving the 3'-end of positive strand RNAs following the addition of 3' uridylate residues to the poly(A)
n
tract by a cellular enzyme such as terminal uridylyl transferase (
22
,
23
). The proposed role of the positive strand 3' NCR in these models is to direct the replication initiation complex to
the authentic template RNA through as yet unidentified RNA-protein contacts (possibly involving cellular factors). While our results
do not disprove either of these models, they argue that specific sequence
and/or secondary structure determinants within the 3' NCR are not required to facilitate the basic mechanism of polymerase
recognition of the positive strand template, even though these sequences may
have evolved to enhance or otherwise regulate the initiation of negative strand
synthesis. The correct subcellular localization of viral genomic RNAs to
membrane bound replication centers (
24
-
26
) and the suggested requirement for concurrent translation of an RNA destined
for use as a replication template may be additional considerations which lend
efficiency and fidelity to the initiation of negative strand RNA synthesis in
an infected cell (
27
,
28
).
Previous studies using poliovirus have described mutagenesis of the PV1 3' NCR which interfered with RNA replication, presumably as the result of
disruption of the proposed pseudoknot structure in this region (
6
,
29
). Characterization of large deletions in the poliovirus 3' NCR has recently been described (
30
), however, these studies may not have detected poorly replicating mutant
viruses due to the use of inefficient DNA transfection methodologies. Other
than the previous study (
2
) and the results reported here, mutagenesis of the HRV14 3' NCR has only been described in the context of a HRV14 3' NCR/poliovirus chimeric replicon in a CAT reporter assay. While
these results also suggested the tolerance of extensive primary sequence
variation in this region, the mutations were not studied
in vivo
in the context of infectious virus (
31
). The possible contribution of the RNA sequence or secondary structure at the 3'-terminus of the 3D
pol
coding region toward directing specific template utilization has not been
addressed. The potential conformational changes which are likely to occur at
the 3'-end of the viral RNA template upon assembly of the protein
components of the replication complex will clearly be difficult to investigate,
although these studies are ongoing.
We have biochemically confirmed the existence of a single stem-loop structure in the 3' NCR of the HRV14 genomic RNA which, based on phylogenetic primary
and secondary structure comparisons, is highly conserved among the
Rhinoviridae
. The stem-loop structure is independent of 5'-flanking coding sequence or the presence of a 3'-flanking poly(A)
n
tract. The deletion of most of the primary sequence within the HRV14 3' NCR, which abolishes formation of the stem-loop structure, results in a severely impaired growth phenotype
but does not result in lethality, indicating that initiation of negative strand
replication does not require the intact stem-loop structure. Taken together, these results offer the possibility of
defining an absolute minimal sequence requirement at the 3'-end of genomic RNA to support the initiation of negative strand
synthesis [herein reduced to 5'-GTTTTAT(A)
n
-3' following the stop codon] which will ultimately lead to the
identification of the underlying molecular mechanism responsible for the
process, as well as the opportunity to study the role of the 3' NCR stem-loop structure in greatly enhancing the efficiency of negative
strand synthesis through additional macromolecular contacts with the viral RNA
replication machinery.
We are grateful to Holger Roehl and Louis Leong for critical reading of the
manuscript. We are also indebted to Hung Nguyen and Tri Ho for assistance with
cell culture. ST was a pre-doctoral trainee of a Public Health Service training grant (GM07134). This
work was supported by Public Health Service grant AI22693 from the National
Institutes of Health.
REFERENCES
Return