Molecular Characterization of Tomato Spotted Wilt Virus Defective Interfering RNAs and Detection of Truncated L Proteins
ABSTRACT Junction sites of 25 different defective interfering (DI) RNAs of tomato spotted wilt virus (TSWV) were characterized. The DI RNAs varied in size from 2.0 to 5.2 kilobases (kb) and contained a single internal deletion. The absence of DI RNAs smaller than 2 kb suggested a size constraint for the survival of TSWV DI RNAs. This hypothesis was reinforced by the finding of a dimeric DI RNA formed by two 1.6-long monomers linked head to tail. Three types of junction sites were found, one type originating from a simple deletion; the second type contained a few extra nucleotides of unknown origin; and the third type contained a stretch of three to five nucleotides, originally occurring at both sides of the deletion and of which one was deleted. In 19 of the 25 DI RNAs studied, the original reading frame was maintained, suggesting a selective preference of DI RNAs with translational potency. Truncated proteins encoded by these DI RNAs were indeed detected in the nucleocapsid preparations. Folding studies of the complete L RNA revealed that the calculated minimal energy of folding was at 16°C lower than at 23°, indicating a higher stability of this molecule at low temperatures. The results suggest an involvement of locally folded secondary structures in the process of deletion, rather than the requirement of certain sequences around the deletion point. The DI RNA generation in TSWV is essentially, as discussed, similar to the process of RNA recombination described in many viruses.
- SourceAvailable from: virology.wisc.edu[show abstract] [hide abstract]
ABSTRACT: We have analyzed 11 picornaviral RNA genomic sequences by optimal and suboptimal minimum free energy folding algorithms. The systematic summation of all pairing partners for each base in the suboptimal structures (P-num value) shows a distinct pattern of alternating low and high values when plotted against the sequence length and indicate regions within each genome where secondary structure(s) are likely to play a significant role in virus biology. The individual folds augmented by data from phylogenetic folds, collectively suggest some revisions of existing models for 5′-untranslated regions of cardioviruses and enteroviruses that might better explain the functions of these regions.Seminars in Virology. 01/1997;
- [show abstract] [hide abstract]
ABSTRACT: The complete sequence of the tomato spotted wilt virus (TSWV) M RNA segment has been determined. The RNA is 4821 nucleotides long and has an ambisense coding strategy similar to that of the S RNA segment. The M RNA segment contains two open reading frames (ORFs), one in the viral sense which encodes a protein with a predicted size of 33-6K, and one in the viral complementary sense which encodes the pre- cursor to the G 1 and G2 glycoproteins, with a predicted size of 127.4K. Both ORFs are expressed via the synthesis of subgenomic mRNAs that possibly termin- ate at a stable hairpin structure, located in the intergenic region. The precursor for the glycoproteins contains a sequence motif (RGD) which is character- istic of cellular attachment domains. Significant sequence homology was found between the G1 glyco- proteins of members of the genus Bunyavirus and a corresponding region in the glycoprotein precursor of TSWV, indicating a close evolutionary relationship between these viruses. With the elucidation of the M RNA sequence, the complete nucleotide sequence of TSWV has been determined. TSWV represents the first member of the Bunyaviridae shown to contain two ambisense RNA segments.J. gen. Virol. 73 (1992) 2795-2804. 01/1992;
- [show abstract] [hide abstract]
ABSTRACT: Defective interfering (DI) RNA molecules derived from the genomic L RNA segment of tomato spotted wilt virus (TSWV) were generated during sequential passage of the virus at high multiplicity. Characterization of DI RNAs from four distinct isolates by Northern blot analysis and sequence determination revealed that both the 5' and 3' genomic termini were retained in these molecules. Each DI RNA contained a single internal deletion of approximately 60% to 80% of the L RNA segment. All DI RNAs studied maintain an open reading frame (ORF) which suggests that these defective molecules should be translatable by ribosomes. Detection of only defective molecules with ORFs indicates either that association with ribosomes or translation is a prerequisite for the selection and maintenance of replicating DI RNAs, or that the truncated proteins produced play a role in their selection or replication. Analysis of the junction sites in the DI RNAs showed that short nucleotide sequences are repeated, one at the release and another at the reinitiation point on the L RNA. One of these is lost during the generation of the DI molecules. The presence of repeated sequences at the junction sites seems to be unique for tospovirus DI L RNAs; they have not been described for other DI systems of either positive- or negative-strand RNA viruses. A model for TSWV DI RNA generation is proposed in which the viral polymerase can 'jump' across the internal sequences from one secondary structure to another containing the repeated sequences, during the replication of the viral complementary L RNA segment.Journal of General Virology 11/1992; 73 ( Pt 10):2509-16. · 3.13 Impact Factor
Molecular Characterization of Tomato Spotted Wilt Virus Defective
Interfering RNAs and Detection of Truncated L Proteins
Alice K. Inoue-Nagata,*,1Richard Kormelink,*Jean-Yves Sgro,† Tatsuya Nagata,*
Elliot W. Kitajima,‡ Rob Goldbach,*and Dick Peters*,2
*Department of Virology, Wageningen Agricultural University, Binnenhaven 11, 6709 PD, Wageningen, The Netherlands;
†Institute for Molecular Virology, University of Wisconsin, 1525 Linden Drive, Madison, Wisconsin 53706;
and ‡NAP/Department of Phytopathology, ESALQ, C.P. 9, 13418-900, Piracicaba, SP, Brazil
Received December 15, 1997; returned to author for revision January 12, 1998; accepted June 4, 1998
Junction sites of 25 different defective interfering (DI) RNAs of tomato spotted wilt virus (TSWV) were characterized. The
DIRNAs variedinsize from2.0 to 5.2 kilobases (kb) andcontaineda single internal deletion.The absence of DIRNAs smaller
than 2 kb suggested a size constraint for the survival of TSWV DI RNAs. This hypothesis was reinforced by the finding of a
dimeric DI RNA formed by two 1.6-long monomers linked head to tail. Three types of junction sites were found, one type
originating from a simple deletion; the second type contained a few extra nucleotides of unknown origin; and the third type
contained a stretch of three to five nucleotides, originally occurring at both sides of the deletion and of which one was
deleted. In 19 of the 25 DI RNAs studied, the original reading frame was maintained, suggesting a selective preference of DI
RNAs with translational potency. Truncated proteins encoded by these DI RNAs were indeed detected in the nucleocapsid
preparations.Folding studies ofthe complete L RNA revealedthatthe calculatedminimal energy offolding was at16°C lower
than at 23°, indicating a higher stability of this molecule at low temperatures. The results suggest an involvement of locally
folded secondary structures inthe process of deletion, rather thanthe requirement of certainsequences around the deletion
point. The DI RNA generation in TSWV is essentially, as discussed, similar to the process of RNA recombination described
in many viruses.
© 1998 Academic Press
Defective interfering (DI) RNAs are generated fromthe
original viral genome by deletions and/or rearrange-
ments and have a replicative advantage over wild type
genomes (Huang and Baltimore, 1970; Perrault, 1981).
Whereas their generation has for animal viruses been
studied extensively (Huang, 1973; Perrault, 1981; Barret
and Dimmock, 1986; Schlesinger, 1988), theirubiquity for
plant viruses was only confirmed during the last decade
(Hillmanetal., 1987;Roux etal., 1991;Graves etal, 1996).
Recently, the generation and accumulation ofDIRNAs
of tomato spotted wilt virus (TSWV), a negative stranded
plant virus belonging to the genus Topovirus was re-
ported (Resende et al., 1991, 1992; Inoue-Nagata et al.,
1997). The TSWV genome consists of two (S and M)
ambisense RNA segments and one (L) negative sense
RNA segment with respective sizes of 2916, 4821, and
8897 nucleotides (de Haan et al., 1990, 1991; Kormelink
et al., 1992). These RNAs have complementary termini
and form a complex with the nucleocapsid (N) protein,
referred to as ribonucleoprotein (RNP) particles, with a
pseudo-circular structure (de Haan et al., 1989; Peters et
al., 1991). Purified virus preparations contain four struc-
tural proteins, the N protein (29 kDa) encoded by the S
RNA, two glycoproteins (78 and 58 kDa) encoded by the
M RNA, and small amounts of the L protein (330 kDa)
encodedbythe L RNA andrepresenting the putative viral
polymerase (Mohamed et al., 1973; Tas et al., 1977; de
Haan et al., 1991; van Poelwijk et al., 1993).
During serial mechanical inoculation ofTSWV isolates
in susceptible hosts, DI RNAs of TSWV are readily gen-
erated (Resende et al., 1991; Inoue-Nagata et al., 1997).
GrowthofTSWV-infected plants atlow temperatures and
use of high inoculum concentrations positively affect
their generation and accumulation. All DI RNAs of TSWV
studied so far resulted from a single internal deletion
within the L RNA, and their occurrence is usually asso-
ciated with symptom attenuation in the inoculated plant
(Resende et al., 1991, 1992; Inoue-Nagata et al., 1997). In
other bunyaviruses, DI RNAs are poorly studied with the
exception of Bunyamwera virus (Kascsak and Lyons,
1978; Patel and Elliott, 1992; Scallan and Elliott, 1992).
Similarly to TSWV, they are all derived from the L RNA,
and some of them are shown to be associated with
persistent infections (Patel and Elliott, 1992; Scallan and
For three TSWV DI RNAs, the nucleotide sequence of
1Present address: CENARGEN/EMBRAPA, SAIN Parque Rural, C.P.
02372, CEP 70770–900. Brası ´ lia, DF, Brazil. Fax: 55–61-340–3624.
2To whom correspondence and reprint requests should be ad-
dressed. Fax: 31–317-484820. E-mail: Dick.Peters@medew.viro.wau.nl.
VIROLOGY 248, 342–356 (1998)
ARTICLE NO. VY989271
Copyright © 1998 by Academic Press
All rights of reproduction in any form reserved.
the region surrounding the junction sites was deter-
mined (Resende et al., 1992). In these three DI RNAs
studied, a small sequence repeat at both release and
reinitiation sides of the deletion site was observed. One
of these repeats was lost after the junction event. In
addition,twoofthese three DIRNAs showeda frameshift
in the ORF of the L RNA. Despite this frameshift, the
original ORF was potentially restored by the use of a
secondary start codon, suggesting an active ORF. How-
ever, the expression of these ORF’s was not further
investigated by the authors.
In this report, a large number of TSWV DI RNAs are
described, including the analysis of their junction sites
and their translatibility with the aim to recognise com-
mon molecular features. Based on the results obtained
and the predicted secondary structure of the complete L
RNA, a possible mechanism of DI RNA generation in
TSWV is discussed.
Nucleotide sequence of DI RNA molecules
Distinct DI RNAs of TSWV were generated by serial
mechanical inoculations under different inoculation con-
ditions in various hosts (Fig. 1). To gain more insightinto
the mechanism underlying the generation of these mol-
ecules,theywere clonedandthe nucleotide sequence of
the junction sites determined. The obtained sequences
were compared with the nucleotide sequence of the
wild-type TSWV BR-01 L RNA (de Haan et al., 1991) used
for generation of the DI RNAs. In total 25 DI RNA mole-
cules were isolated from 22 plant lines, which were
originally inoculated with wild-type (wt) TSWV (Inoue-
Nagata et al., 1997).
DIRNAs were amplified using primers specific to both
terminal ends of the L RNA, cloned into a sequencing
vector, and furtheranalyzed by restriction enzyme diges-
tion, and nucleotide sequence determination of the re-
gion surrounding the junction sites. The obtained data
revealed that all DI RNAs found contained one large
internal deletion (Table 1). However, the presence of
additional very small deletions orinsertions inthe L RNA
can not be completely ruled out. To exclude the possi-
bility that artificial DI cDNA fragments arose as a result
ofRT-PCR, a control reactionwas performedonwtTSWV
RNA using the same pair of primers. This reaction gave
rise to a few fragments in low amounts that did not
comigrate with any of the DI RNA fragments analyzed
(data not shown).
The size of DI RNAs varied between 2075 [DI RNA
16(15)-1] and 5191 (5–3) nts (Table 1). Most of the DI
RNAs analyzed (20 of 25) had shorter 5? than 3? terminal
ends of the viral complementary (vc) strand of L RNA
(Table 1, column 7). The shortest5? end, found in DIPe-2
(Table 1, Pe-2), consisted ofthe first93 nts, hence retain-
ing the start codon of the L ORF (located at position 34).
On the other hand, DI RNAs with short 3? terminal ends
could also be found, notably in DI Be-1. The deletion in
this DI RNA occurred between nt 3228–8732 (Table 1,
Be-1), which included the stop codon of the L ORF
Northern analyses of total RNA of a few plant lines
incubated at 16°C showed the presence of differently
sized DI RNAs after consecutive passages [Fig. 1, 16(5)
and 16(15)]. DI RNAs, differing in size and sequence,
were found between the 5th and 15th passage in lines
16–1 and 16–3 [Table 1, 16(5)-1 and 16(15)-1, 16(5)-3, and
16(15)-3], whereas the 3.4-kb DI RNA in line 16–2 re-
mained present and dominating during the examined
passages [Fig. 1; Table 1, 16(5)-2 and 16(15)-2]. These
results suggested that some DI RNAs, once formed,
could persist in a TSWV isolate throughout several pas-
sages, while others were possibly outcompeted by more
efficiently replicating DI RNA molecules. The occasional
presence of at least two distinct DI RNAs in the same
plant line, all in large, but different amounts, was occa-
sionallyobserved[Fig.1, 16(5)-3and5–2] andshows that
they can temporarily coexist.
FIG. 1. Northern blot hybridization of total RNA extracts fromTSWV DI RNAs infected plants using probes specific to TSWV L RNA. DI RNAs were
generated following serial inoculations ofTSWV inN. rustica plants incubated at16, 30, or23°C TSWV (labeled atthe top ofthe panel), and intomato
(To), sweet-pepper (Pe), N. benthamiana (Be), and Datura stramonium(Da) plants at 23°C (Inoue-Nagata et al., 1997). Plant lines incubated at 16°C
were tested after the 5th [16(5)] and 15th passage [16(15)]. For all other plant lines, only the 15th passage was used. Inoculation of plants incubated
at 23°C was performed with 5- and 50-fold diluted inoculum. Three individual plants were used for each condition (third row, 1, 2, and 3), and only
those plants showing DI RNA accumulation are shown. Position of wt L RNA and the size range of DI L RNA molecules are indicated at the left of
343 DI RNAs CHARACTERIZED IN TSWV
The occurrence of a DI RNA with a dimeric structure
Northernblotanalysis ofDIDa-2showeda single RNA
species ofca. 3.0 kb (Fig. 1). However, PCR amplification
of cDNA from this isolate resulted in two distinct DNA
fragments of 3.2 and 1.6 kb (data not shown). Complete
nucleotide sequence determination of both molecules
showed that the larger DI RNA (3184 nts) was formed by
two copies ofthe smallerDIRNA (1592 nts) linked head-
to-tail. Each monomeric sequence had a deletion be-
tween nt 893 and 8204 and an insertion of UUUUUA at
this junction site. The linkage point between the two
monomers had perfect 3? and 5? ends of the L RNA
sequence (Fig. 2A), by which a new XbaI site (TCTAGA)
was created. This linkage site was confirmed by PCR
amplification using the first strand DNA as template and
primers pDH003 (identical to nt 8339–8345 of vc L RNA)
and ZUP406 (complementary to nt 78–94 of vc L RNA).
The particular localization of these primers (Fig. 2A)
enabled amplification of a fragment of ca. 600 bp. Direct
sequencing of this fragment showed that the DI RNA
present in Da-2 was formed by a homogenous popula-
tion of the dimer.
To confirmwhether Da-2 was a true dimer, a Northern
blot analysis was made using four different probes (Fig.
Characterization of Junction Sites of TSWV DI RNAs
30-35009A in-?? 6244514 ??????????????????--
G 5065 ????????????????--
aSize in nucleotides of the complete DI RNA.
bType I junction site: deletion without any insertion or deletion at the junction site.
cType II junction site: unknown sequence added at the point of the junction. The extra sequence is shown, except for DI Da-3, which had an
insertion of 17 bases (Fig. 3B).
dType III junction site: repeated sequence is present at the junction site, one of them was lost. The number of repeated bases is shown.
ein, the deletion maintained the original reading frame downstreamofthe junction site; shift, the deletion caused a frame shift; *, purified particles
tested by Western blotting for the presence of truncated L proteins (Fig. 4).
fSchematic view ofDIRNAs inthe vc strand ofthe L RNA; ?, inframe sequence forthe L ORF; /, outofframe sequence forthe L ORF; -, noncoding
sequence. Numbers indicate position of junction sites.
g(5), 5th passage of N. rustica lines incubated at 16°C.
h(15), 15th passage of N. rustica lines incubated at 16°C.
iThis monomer occurred as a dimer (Fig. 2).
344INOUE-NAGATA ET AL.
2B). Two were homologous to both termini of the L RNA
corresponding to nucleotide regions 34–894 and 8339–
8665 (Probe 1). The two other probes were homologous
to the ends of the deleted stretch of the DI RNA, corre-
sponding to nucleotide regions 950–1499 and 6911–7974
(Probe 2). It was anticipated that Probe 2 would not
hybridize with this DI RNA when the 3.2 kb DI RNA was
formed by two identical monomers. Northern blot analy-
sis of total RNA extracted from wt TSWV and DI Da-2
infected plants confirmed that Da-2 lacked the internal
region fromnt 950–7974 (Fig. 2C). These results strongly
support that the 3.2-kb RNA molecule consists of a gen-
Distinct populations of DI RNAs in one line
Analysis ofmostlines withDIRNA molecules revealed
a single band (Fig. 1) corresponding to a single species
of DI RNA except for two plant lines which exhibited a
single band (Fig. 1, DI 50–3 and Da-1), but which con-
sisted of more than one species of DI RNA. DI 50–3
(Table 1, 50–3U, L-1, and L-2) contained three, and Da-1
(Table 1,Da-1–1and-2)twodistinctDIRNA species.PCR
amplification on DI50–3 cDNA resulted in two differently
sized DNA fragments (U, upper, and L, lower band).
Sequence analysis of obtained clones showed that the
larger DNA fragment had 4073 nts (50–3U), whereas the
smaller fragment consisted of a mixture of two different
DI RNA species (50–3L1 and 50–3L2) with almost iden-
tical sizes of 3767 and 3769 nts. Amplification of Da-1
resulted in one sized DNA fragment only. However, this
fragment contained two populations of similar DI RNAs
with sizes of 3961 and 3965 nts. The junction sites of
both DIRNAs were located in the same region and were
found to differ only in 10 nts at the 5? end and 19 nts at
the 3? end (Table 1, column 7).
In two lines, defective RNAs consisting of 2075 [Table
FIG. 2. (A) Schematic view of DI Da-2 dimer showing the nucleotide sequence of terminal ends, junction sites and linkage point of the two
monomers. Positions are shown below the sequences. Additional nucleotides notpresentin wtTSWV sequence are bolded. New XbaIsite is boxed.
Location and orientation of primers used for PCR amplification are indicated. (B) Location of probes (1 and 2) used for Northern blot hybridization.
Only the terminal ends of the DI Da-2 monomer are shown with the corresponding positions of junction sites in relation to the vc L RNA of TSWV.
(C) Northern blot hybridization of total RNA extracts from DI Da-2 (Da-2) and wt TSWV (wt) using Probes 1 and 2 (B). Positions of wt L RNA and DI
Da-2 are indicated.
345 DI RNAs CHARACTERIZED IN TSWV
1, 16(15)-1] and 2099 nts (Table 1, Pe-3) were found by
PCR, whereas Northern blot analysis on total RNA of
both plant lines (Fig. 1) only showed the presence of a
defective RNA species of ca. 4 kb, i.e., larger than those
obtained by the PCR studies. All attempts to amplify the
largerRNA molecules using several internal primercom-
binations failed (data not shown). These results indi-
cated that in the 15th passage of these lines, the 2075
and 2099 nts defective RNA species were present,
though below Northern blot detection levels. Most likely,
they were preferentially PCR amplified due to their
smaller size and might not represent true interfering
Inthe lines 16(5)-3 and 5–2, two majorDIRNAs (Fig. 1)
withapproximate sizes of2 and4 kbwere detected.PCR
amplification of 16(5)-3 and 5–2 cDNAs using primers
specific to both ends of L RNA resulted in single DNA
fragments of 2.2 and 2.4 kb, respectively, corresponding
insize to the smallerRNA species found [Table 1, 16(5)-3
and 5–2]. The larger RNA could not be PCR amplified,
even when primers directed to deleted regions of the
smaller DI RNA were used (data not shown), therefore
only the smaller DI RNA molecule was characterized in
Analysis of the junction sites of DI RNAs
Nucleotide sequence analysis of the junction sites of
the 25 DI RNAs revealed three types of junction sites.
One type (I) corresponded with the occurrence of a
deletion without any additional insertion or repetition of
specific nucleotide stretches (Table 1, column 3; Fig. 3A).
This type was found in 11 of the 25 (44%) cloned DI
molecules. A second type (II), found in 10 of the 25 DI
clones (40%), was characterized by the addition of a
number of nucleotides of unknown origin at the precise
junction site (Table 1, column 4; Fig. 3B). These additions
varied from 1 to 17 nts and were mostly ‘‘uracyl-rich.’’ In
the thirdtype (III), observedinfourclones (16%), a stretch
of3–5 nts was repeated atbothsides ofthe deletionsite
of which one was lost in the deletion event (Table 1,
column 5; Fig. 3C). Most of the junction sites were char-
acterizedbythe presence ofan‘‘AU-rich’’sequence inthe
vicinity of the deletion sites (Fig. 3).
Detection of four mutations in the L RNA
Nucleotide sequence analysis of the DI cDNA clones
revealed thatfourstable mutations had occurred inthe L
RNA molecule when compared with the previously pub-
lished L RNA sequence (de Haan et al., 1991). The C at
position 82 (C82), numbered on the vc strand, was sub-
stituted by U, U4691 was deleted, a U was inserted
between A4767 and U4768, and AUUU was inserted
betweenA8840andA8841.These changes resultedinan
L RNA with a size of 8901 nts instead of 8897 nts (de
Haanetal., 1991).The uniformityofthese mutations inall
the clones studied indicated that the present isolate
FIG. 3. Nucleotide sequence ofjunction sites ofRNAs fromDI16(5)-1 (A), Da-3 (B), and Be-1 (C) aligned with the corresponding region ofvc L RNA
(de Haan et al., 1991). Positions of junction sites are indicated by arrows. Nucleotide regions present in the DI RNAs are in bold. Insertions (Type II)
or repetitions (Type III) are doubly underlined. Adenine and uracyl residues are singly underlined.
346 INOUE-NAGATA ET AL.
BR-01 used as the inoculumsource differed slightly from
the original one. Other minor point mutations were oc-
casionally found throughout the sequenced molecules,
but were, unlike those described above, not invariably
Detection of truncated L proteins encoded
by DI RNA molecules
Analysis of DI RNAs revealed that 19 of 25 clones had
an in-frame deletion within the original ORF of the L
protein (‘‘in’’ in Table 1, column 6). These DI RNAs, there-
fore, could potentially encode a truncated L protein with
the original amino- (N) and carboxyl-terminal (C) amino
acids. Whenthe deletioncaused a frameshiftinthis ORF
(‘‘shift’’ in Table 1, column 6), a first stop codon was
present already a few codons downstream of the junc-
tion site in all remaining six DI RNAs (data not shown).
To study the potentially translational expression of
these truncated ORFs, nucleocapsids isolated fromviral
particles of nine DI RNA (Table 1, column 6, asterisk)
infected plants were analyzed by Western blotting. Intact
L protein, of predicted size of 330 kDa, was detected in
all preparations, and distinct smaller proteins in most of
them(Fig. 4). The latter reacted strongly and specifically
withthe antibodies to the N orC regions ofthe L protein,
indicating that these smaller, L-derived proteins con-
tainedbothoriginalterminalregions andmusthave been
expressed from DI RNAs. Such truncated proteins were
only detected for DI RNAs with an in frame deletion
(Table 1, column6), andnotinnucleocapsidpreparations
of DI 16–2 [16(15)-2], 5–3 and Da-1, in which the deletion
hadcauseda frameshiftinthe downstreamregionofthis
ORF after the junction site (Table 1, column 6).
The estimated sizes of the truncated proteins found in
DI 50–2, To-3, Pe-1, Be-3, and Da-3 (Fig. 4) were in good
agreement with the expected sizes derived from their
sequence (85, 93, 142, 124, and 98 kDa, respectively).
Only one truncated protein, i.e., DI Be-1 (123 kDa), re-
actedsolelywiththe antibodyagainstthe N-terminal part
ofthe L protein(Fig.4).The nucleotide sequence already
showedthatthe encoding DIRNA lostalmostits entire 3?
end of the L ORF (Table 1, Be-1), thus, explaining the
absence ofa positive reactionwithantibodies specific to
the C regionofthe L protein.The amountofthe truncated
protein relative to intactL protein varied with the DI RNA
isolate.Highamounts oftruncatedL proteins were found
in the nucleocapsid preparations of DI 50–2 and Be-3,
whereas loweramounts were foundinpreparations ofDI
To-3, Pe-1, Be-1 and Da-3. InTo-3, two truncated proteins
of slightly different sizes, but occurring in approximately
equal amounts, were found. The second protein was
most likely encoded by a distinct DI RNA present in the
plant line, though not clearly visualized on Northern blot
(Fig. 1, To-3) and not amplified by RT-PCR. Besides the
truncated L proteins, which could readily be detected in
most preparations, proteins giving a faint reaction were
occasionally observed, e.g., 5–3, To-3, Be-3 (Fig. 4). The
origin of these proteins, including the second protein
found in To-3, remains to be studied.
Secondary structure of the L RNA
DI RNAs of TSWV are more efficiently generated at
lower temperatures (Inoue-Nagata et al., 1997). To study
a possible correlation between the folding of the RNA
and the generation of DI RNAs, the L RNA, corrected for
the new mutations found (8901 nts), was folded at both
16 and 23°C using firstly its positive polarity (vc strand).
The optimal folding was different at these temperatures
(Figs. 5A and 5B). At 16°C, 5932 nts (66.6%) were in-
volved in base pairing (1714 A:U, 943 G:C, and 309 G:U
base pairs), while 5836 nts (65.6%) were at 23°C (1692
A:U, 934G:C, and292G:Ubase pairs).Mostofthe paired
nts assembled in 675 stems of length 2 or more for the
16°C foldand648at23°C.More thanhalfofthese stems
FIG. 4. Western blot analysis of purified nucleocapsid preparations fromN. rustica plants infected with TSWV containing DI RNAs. Healthy (H) and
wt TSWV (wt) infected plants were used as controls. Wt L protein is seen in all preparations (330), except for H. Identical reactions were obtained
when using antibodies against the N and C region of L protein (only one of themis shown), except for Be-1, in which anti-N (N) and anti-C reactions
are shown. Size marker is on the left.
347 DI RNAs CHARACTERIZED IN TSWV
FIG. 5. Counterclockwise two-dimensional representation of the optimal folding of TSWV positive (vc) L RNA strand at 16 (A) and 23°C (B) and of negative
(v) strand at 16°C (C). Deletion points are indicated for the 5? end position (gray) and 3? end position (black) for each clone (Table 1). The bases with lowest
P-num values are drawn in black, all other bases are gray.
348 INOUE-NAGATA ET AL.
349DI RNAs CHARACTERIZED IN TSWV
350INOUE-NAGATA ET AL.
(340 stems of length 2 or more) were common to both
folds. Some of these common stems were grouped and
could fold with the same pattern, see for example re-
gions of Fig. 5 around nt 2000, 5300, 6000, or 8900. The
latter showed the 5? and 3? ends in close contact, con-
necting the first 5? 14 nts to the last 3? nts with the
longest helix of the whole fold at either 16 or 23°C and
extended by four helices to base 41 (5?) and 8870 (3?).
This long stem was further forked and prolonged with
more helices bringing 5? to3? together.The 5? and3? end
long range stems harbored very low P-num values (the
number of all possible pairing partners for each base) in
both folds, and most clearly at 23°C, an indication that
the 5? and 3? ends are likely to be found paired in the
majority of the RNA molecules. The minimal energy cal-
culatedforfolding was lowerat16°C (?2859.1 kcal) than
at 23°C (?2350.2 kcal), suggesting a higher stability of
the RNA at lower temperatures because the RNA can
forma larger number of base pairs at the lower temper-
ature. Observed P-num values for each fold (see Mate-
rials and Methods) ranged from1 (ntC5, C9, A10, A8894,
and G8897) to 1103 (base U4924) for the 16°C fold and
from1 (nt C5, C9, A10, and G8897) to 1033 (base U8625)
for the 23°C fold. Overall, P-num values in both folds
were high (for 16 and 23°C folds average P-num values
were 205 ? 143 and 213 ? 146, means ? SD), suggest-
ing a large variabilityinthe pairing patternforsuboptimal
structures for each temperature fold (data not shown).
However, 3.3% of the nucleotides at 16°C but 18.2% at
23°C were at normalized P-num value lower than 3%of
the maximumP-numvalue at infinite energy, suggesting
that the corresponding regions could be found paired
more often at 23°C than at 16°C (see darker black
segments on Figs. 5A and 5B).
Considering that the negative (viral) strand fold would
not represent a symmetrical folding of the positive (vc)
one forthermodynamic reasons and also because ofthe
possibility of G–U pairs, the negative polarity was folded
at the temperature of 16°C serving as a model to com-
pare the secondary structure ofboth strands (Fig. 5C). In
the calculated optimal folding of the negative strand at
16°C, 5694 nts (64.0%) were base-paired (1714 A:U, 896
G:C, and 236 G:U base pairs) in 612 stems of length 2 or
more. The total energy of this optimal structure was
?2585.7 kcal. The lesser amount of bases engaged in
base-pairing observed in the negative polarity folding
may be a reflection of the skew in base composition.
Overall the base composition is 1/3 GC and 2/3 AU for
both polarities. For both 16°C folds, the number of A:U
base pairs was identical (1714) while the numberofbase
pairs involving a G was reduced in the negative polarity
which has only 1293 Gs (14.5%) vs 1692 (19%) for the
positive polarity. The observed P-num values ranged
from a value of 1 (nt C5, C9, A10, A8894, and G8897),
located in a 14 nucleotide terminal repeat common to
positive and negative strands at both ends, to a maxi-
mum value of 466 (base G758) and were on average
lower than that of positive folds (mean of 124 with 75 of
standard deviation). In the negative strand, 41% of the
bases had a normalized P-num values lower than 3%of
the maximum P-num value at infinite energy. Bases
within the negative strand at 16°C appeared therefore
more limited in the number of pairing partners they can
choose compared with the positive strand, suggesting a
larger number of stable conformations at this tempera-
ture (see darker black segments on Fig. 5C).
Within the structures obtained, the position of the
deletion (release and reinitiation) points from all se-
quencedDIRNAs were positionedinthe optimallyfolded
structure (Figs. 5A–5C). Analysis of the location of these
points in the structure revealed that they were concen-
trated in regions where the RNA was mostly folded
locally, i.e., not involved in long range base pairing.
These deletion points did not show a particular correla-
tion with low P-num regions unlike results from corona-
virus studies (Rowe et al., 1997). The release and reini-
tiationpoints were locatedinclusters whichwere distant
each other when analyzing in only two dimensions.
These deletion points were located either in single
stranded loop regions or in double stranded stems.
Recently, a large series of DI RNAs of TSWV was
generated during serial mechanical inoculations in
susceptible plants (Inoue-Nagata et al., 1997). To elu-
cidate the process of DI RNA formation in TSWV, the
nucleotide sequence of the junction site of 25 of these
DI RNAs was determined. The data obtained showed
that these DI RNAs were formed by a single internal
deletion ranging from42 to 77%of the complete L RNA
segment. Thus, only DI RNAs with intact 5? and 3?
ends were found, similarly to those previously found in
tospoviruses (Resende et al., 1991, 1992), Bunyamwera
virus (Patel and Elliott, 1992), as well as in influenza
virus (Nayak et al., 1982; Jennings et al., 1983). This
resultindicates thatthe cis-acting sequences required
for replication of the TSWV L RNA molecule are
present in both the 5? and 3? ends of the molecule. A
closer analysis of the smallest terminal ends of the DI
RNAs revealed that these essential sequences were
most likely within the last 165 nts of the 3? end (Table
1, Be-1) and, in analogy, within the first 71 nt? of the 5?
end (DI NL-11 3.3 kb, Resende et al., 1992).
The smallest DI RNA detected, 16(5)-3, had a size of
2216 nts. It exhibited a clear replicative advantage over
the wtTSWV L RNA as couldbe concludedfromthe large
amount of defective RNA detected by Northern blot hy-
bridizationstudies [Fig. 1, 16(5)-3]. This contrasts to other
L specific defective RNAs smaller than 2100 nts, which
were isolated during PCR cloning but that could not be
detected on Northern blots [Table 1, 16(15)-1 and Pe-3].
351DI RNAs CHARACTERIZED IN TSWV
These RNAs most likely represent defective noninterfer-
ing RNAs newly generated during replication of the iso-
lates, and which are preferentially amplified by PCR.
Because no DI RNAs smaller than 2.0 kb have been
reported for tospoviruses (Resende et al., 1992), a size
constraint may exist for their survival. This might be
related to RNA replication or stability, as suggested for
turnip crinkle carmovirus DI RNAs (Zhang and Simon,
1994). The strongest evidence for such size constraints
was the finding of a dimeric DI RNA. Its low fitness to
survive as a monomeric DI RNA may be reflected by our
inability to detectthe monomeronNorthernblots while it
was the dominant formamplified by RT-PCR. Dimer RNA
molecules functioning as DI or satellite RNA have also
previously been reported in tombusviruses (Dalmay et
al., 1995; Havelda et al., 1995; Finnen and Rochon, 1995)
and in carmoviruses (Simon et al., 1988; Cascone et al.,
1990).The mechanismleading tothe formationofdimers
is still not known (Carpenter et al., 1991; Finnen and
Rochon, 1995). However, the most accepted hypothesis
assumes that the polymerase, after finishing the replica-
tion, reinitiates another round of polymerization without
release of the nascent strand by switching to the 3? end
of the same or another template molecule (Carpenter et
al., 1991; Dalmay et al., 1995). In our studies, the difficul-
ties encountered to clone the RNAs observed in DI
16(15)-1, Pe-3, and the larger RNAs of 16(5)-3 and 5–2
may be related to dimer formation in these isolates, but
this needs further investigation.
The comparison of the nucleotide sequences of at
least two clones of each DI RNA indicated that possibly
mostofthemgrew as a single dominantDIRNA species.
However, the coexistence of at least two distinct DI
RNAs in the same line suggests that the DI RNAs occur
as a population of heterogenous DI RNAs even though
one species is often largely dominant.
Based on the obtained results, no answers can be
given on the question whether small TSWV DI RNAs do
evolve from larger ones or whether they are being inde-
pendently generated. The latter possibility seems more
likely, as supported by the appearance of larger rather
than smaller DI RNA molecules compared with the orig-
inally present ones. A thorough study on DI RNA gener-
ation within one plant line, though, is required to answer
the aforementioned question, and give more information
on the heterogeneity of DI RNA populations.
Maintenanceofareading framewas demonstratedtobe
essential for the survival of DI RNAs in several viruses
(Kuge et al., 1986; White et al., 1992; Pogany et al., 1995,
1997; van der Most et al., 1995). In our studies, the L ORF
was maintained in 19 of 25 TSWV DI RNAs, which sug-
gested a preferential selectionofthose DIRNAs withanin
frame deletion.Preliminaryattempts todetectthe truncated
L protein in crude sap preparation from TSWV infected
tissue failed(data notshown).However, truncatedproteins
with expected sizes could be detected successfully in pu-
rified virus particle preparations. This indicated that the
nucleocapsid (ribonucleoprotein complex, RNP) binding
domain was still retained in these truncated proteins and
that they were not selected out from the particles, in con-
trast with the situation in Bunyamwera (Patel and Elliott,
1992) and influenza virus (Akkina et al., 1984a,b) for which
truncated proteins were not found in the particles. The
occurrence, within the virus particle, of a truncated protein
encoded by DI RNA Be-1 in the virus particle, in which the
entire C-terminal region of the original L protein was ab-
sent, represents evidence that the nucleocapsid binding
domain is in the N-terminal extremity of the L protein be-
tween amino acids positions 1 and 1065. The presence of
a truncated L protein in Da-3 particle preparations, harbor-
ing a small N-terminal region of about 133 amino acids,
would even suggest that the binding domain is present
betweenaminoacids 1and133.This resultargues against
the role of the acidic domain of the C-terminal region in L
in the binding to nucleocapsids suggested previously
(Kormelink et al., 1992, 1994).
A major effect of the temperature of incubation on the
generation/accumulationofDIRNA molecules has recently
been demonstrated (Inoue-Nagata et al., 1997). The possi-
bility to analyze a large number of junction sites from DI
RNAs generated at distinct temperatures prompted us to
predict the secondary structure of the complete L RNA at
the temperature of 16 and 23°C for the positive (vc) strand
and at 16°C for the negative (viral) strand (Fig 5). As ex-
pected, all three structures differed in calculated minimal
energy, though the predicted conformation retained com-
mon local foldings (e.g., Figs. 5A and 5B: around nt 2000,
5300, 6000, and 8900; or Fig. 5C: around nt 6900, 3600,
2900, and 1). This result suggests that these foldings are
ofthe temperature.The deletionsites onthe foldedRNA at
both temperatures seem to be concentrated in regions
predicted to be in local folding configurations, either in
single or double stranded regions. These results are in
agreement with the increased occurrence of recombina-
tions in highly base-paired regions of a number of viruses
al., 1995; Dzianott et al., 1995; White and Morris, 1995;
Havelda et al., 1997). The global minimum energy of each
structure is an average of the overall favorable and unfa-
vorable thermodynamic and energy considerations for the
complete molecule.The lowerfree minimumenergy calcu-
lated for 16°C could be an indication that at this tempera-
ture the molecule can form more energetically favorable
structures that might facilitate the release of the polymer-
ase complex. This hypothesis could explain why DI RNAs
were readily generated inplants incubated ata lowertem-
perature (Inoue-Nagata et al., 1997). Alternative explana-
tions canreside insignificantchanges inthe physiology of
the plantand thus in the replicating machinery ofthe virus
orinthe greaterstabilityofDIRNAs atlowertemperatures.
352 INOUE-NAGATA ET AL.
However, the latterhypothesis is less plausible since incu-
bation of TSWV DI RNAs infected plants at higher temper-
atures does notfilterouttheDIRNAs fromtheTSWVisolate
(data not shown). Most of the 5? junction points (Table 1,
column 7) were located in the first (around nt 1) orsecond
branch (around nt 600) of positive polarity or in the first
(around nt 1) or last folded branch (around nt 8200) of
negative polarity structure. This resultsupports the hypoth-
esis that these regions are hot spots of occurrence of
polymerase stopping and release hence originating dele-
tions or inversely as structures favoring the landing of the
polymerase-nascent strand complex. Although no conclu-
sioncanbe drawnfromourdata as towhichstrandpolarity
is the siege of deletion events it is noteworthy that, in the
positive polarity, 21 of the 25 junctions (84%) are within the
first1200 bases while the deletion points atthe 3? end are
more evenly distributed along the 3? half of the molecule.
The concentration of deletion points at the 5? end of the
positive polaritystrandcouldreflectthe earlyrelease ofthe
polymerase as it tries to duplicate the positive strand from
the 3? endofthe negative strandtemplate. The occurrence
of only type I junction site was observed in all DI RNAs
the RNA folding remains unknown. The lowest P-numval-
ues foranenergybracketof12kcalwere locatedwithinthe
firstandthe lastdozenbases atthe extremities suggesting
that the 5?–3?repeats have evolved to pair together. The
circularization of the RNA may thus be the result of select-
ing against alternative pairing partners elsewhere in the
sequence (Palmenberg and Sgro, 1997). This result con-
seeninthe electronmicroscope (Peters etal.,1991).One of
the limitations of using this secondary structure prediction
is thatthe folding is based ona naked RNA sequence and
notonthe folding ofthe RNP complex.However, the use of
a varietyofenzymatic andchemicalprobes forstudying the
RNA conformation strongly indicated that the RNA is ex-
posed outside of the RNP structure in influenza virus (Jen-
nings etal., 1983; Baudinetal., 1994). Assuming thatRNPs
of TSWV are structured identically, RNA–RNA interactions
forming a structure similar to that predicted of naked RNA
may represent a real situation. The confirmation of the
reliability of the secondary structure awaits further studies
using biochemical and biophysical techniques (e.g., Raikar
et al., 1988). However, the strong clustering of the deletion
sites atthe 5? endofvc L RNA (orinversely,atthe 3? endof
structures suggested the potential usefulness of the used
technique topredictthe conformationofthe RNA innature.
Ourresults suggestthatsecondary structure canaffectthe
efficiency of replication particularly at lower temperatures
atwhichthe RNA molecules couldfoldenergeticallystable
motifs putatively disrupting the polymerase complex.
Nucleotide sequence analysis of the junction sites
of TSWV DI RNAs revealed three types of deletions.
Most junction sites did not show any repeated se-
quences or acquisition of additional nucleotides. This
indicates that TSWV DI RNAs are mostly generated by
nonhomologous recombination events, as only 16%
homologous sequences were found surrounding the
deletion site. However, analysis of each of the junction
sites found revealed strong similarities with aberrant
homologous recombinationseeninbrome mosaic bro-
movirus (Nagy and Bujarski, 1996), like the presence
ofAU-rich sequences close to the deletion site and the
addition of extra nts at the junction site, most of them
rich in uracyl (Cascone et al., 1990, 1993; White and
Morris, 1994). Many models, developed to explain the
occurrence of recombinations and deletions during DI
RNA formation, are based on the ‘‘copy-choice’’ model.
In this model the polymerase dissociates carrying the
nascent chain following to precise realignment and
continuation of replicase activity on a new template or
on a new site in the same template. It was originally
proposed by Cooper et al. (1974) for recombination in
polioviruses and later corroborated by Lazzarini et al.
(1981), Perrault (1981), and Lai (1992). We believe that
in TSWV a strongly folded RNA promotes occasional
transcriptional pausing (Mills et al., 1978) and tem-
plate switching (Cooper et al., 1974), thus generating
deletions in the RNA genome. The finding of two DI
RNAs from one plant line with closely located dele-
tions sites (Table 1, Da-1–1 and -2) suggested that a
large variety of DI RNAs are indeed generated but not
isolated due to a strong selective pressure. The high
frequency of generation of TSWV DI RNAs supports
the great importance of recombinatorial events in RNA
evolution and virus adaptation.
MATERIALS AND METHODS
Virus isolates containing DI RNAs
DI RNAs of TSWV were generated in Capsicum an-
nuum, Datura stramonium, Lycopersicon esculentum,
Nicotiana benthamiana, and N. rustica plants by serial
mechanical inoculations (Inoue-Nagata et al., 1997) of
the wt TSWV isolate BR-01 (de A´vila et al., 1993). Leaf
material collected from plants infected with TSWV DI
RNA was stored at ?70°C until analysis.
Cloning and sequencing of DI RNAs
Total RNA was extracted from leaf material infected
with TSWV containing DI RNAs according to the method
ofde Vries etal.(1982),exceptforN.benthamiana plants,
in which the method of Logermann et al. (1987) was
used. First strand cDNA was synthesized using primer
pDH001 (Table 2), and prior to PCR amplification and
subsequent cloning, samples were treated with 1 U of
RNase H (GIBCO-BRL) at 37°C for 20 min. PCR amplifi-
cation was performed according to manufacturer’s rec-
ommendations (Stratagene) using Taq ExtenderPCR Ad-
353 DI RNAs CHARACTERIZED IN TSWV
ditive (Stratagene). Primers used for amplification are
described in Table 2. Each cycle consisted of 30 s of
denaturation at 94°C, 1 min of annealing at 52°C and
extension at 72°C for 3–6 min, depending on the size of
the DIRNA. Afteramplification, the DNA was gel purified
using Glassmax (GIBCO BRL) and ligated into pGEM-T
(Promega), pUC 19 orpSK, as described by Sambrook et
al. (1989). The location of the junction sites were deter-
mined by restriction enzyme analysis and PCR amplifi-
cation using internal primers. The sequence of the re-
gion around the junction site was determined using an
internal primer(Table 2) on an ABI automatic sequencer.
Nucleotide sequences were compiled and analyzed us-
ing programs ofthe GCG package (Devereux etal., 1984,
GCG, Madison, WI) and DNAsis (Hitachi Software Engi-
neering, Japan). The sequence of each fragment ana-
lyzed was based on at least two DI cDNA clones.
Secondary structure prediction of the TSWV L RNA
RNA secondary structure folding of the 8901 nucleo-
tides (nts) vc strand of TSWV L RNA (accession number
D01230) was predicted according to Zuker (1989) with
version 2.2 of the stand-alone MFOLD program (ftp://
snark.wustl.edu). The programwas compiled and run on
a Silicon Graphics (Mountain View, CA, USA) Crimson
workstation with 256 MB of RAM. Each run required
about 1 week of uninterrupted computation.
MFOLD calculated the global minimum free energy
following thermodynamic rules and parameters (Jaeger
et al., 1989). By pairing each nucleotide with all possible
partners in the sequence, MFOLD also simultaneously
determined all suboptimal RNA secondary structures
with less optimal free energy.
MFOLD assumed a temperature of 37°C, therefore
appropriate thermodynamics tables for 23 and 16°C
were calculated with the programNEWTEMP distributed
P-num(Jaeger et al., 1989) was extracted for all struc-
tures within ?12 kcal of the global minimumfree energy
and normalized with respect to base composition, se-
quence length and the number of pairing partners for
each base atinfinite energy (Palmenberg and Sgro, 1997
and unpublished data).
Graphic representations were generatedforthe global
minimum free energy calculated by MFOLD using the
program NAVIEW (Bruccoleri and Heinrich, 1988) as
adapted by Zuker (1989; ftp://snark.wustl.edu).
Northern blot analysis using DIG-labeled probes
The presence of wt L and DI RNA molecules was
followed by Northern blot hybridizations using total RNA
extracts from systemically infected leaves. Total RNA
extraction and subsequent agarose gel electrophoresis
were done essentially as described by Inoue-Nagata et
DIG-labeled probes, generated by standard PCR includ-
ing 1 mM DIG-11-dUTP (Boehringer Mannheim) using
primers of17–19nts locatedatbothsides ofthe fragment
of interest. After hybridization, the filters were incubated
with alkaline-phosphatase anti-DIG polyclonal antibod-
ies (BoehringerMannheim) andsubsequently developed
using NBT/BCIP (GIBCO-BRL) as substrate.
Primers Used to Amplify and Sequence TSWV DI RNAs
aPosition of the primer in the L RNA molecule (de Haan et al., 1991).
bAll DI RNAs were amplified with pDH001 as the first primer and a second primer as mentioned in the table.
cJunction sites of these DI RNAs were determined using the primer in the first column.
dClones Pe-1 and Be-1 could be amplified using only the primer pDH001 due to the high complementarity of the L RNA terminal ends.
354 INOUE-NAGATA ET AL.
Detection of TSWV L RNA translation products
To detect wt and truncated L protein species, Western
immunoblot studies were performed on nucleocapsids
released from particles of TSWV DI RNAs. Viral nucleo-
capsids were extracted by homogenizing systemically
infected N. rustica leaves in 0.1 M potassiumphosphate
buffer, pH 7.0, containing 0.01M sodiumsulfite.Thenthis
extract was centrifuged at 13,000 g for 15 min. The
resulting pellet, which contained the particle fraction,
was resuspended in resuspension buffer (0.01 M potas-
siumphosphate buffer, pH 7.0, containing 0.01 M sodium
sulfite) plus 1% Nonidet-P40 to release nucleocapsids
from the virus particles. Plant debris and starch were
eliminated by centrifugation at 13,000 g for 15 min. Nu-
cleocapsids were then precipitated by centrifugation at
185,000 g for 1 h through a 30% sucrose cushion. The
pellet was resuspended in resuspension buffer (10 ?l
buffer/1 g starting leaf material) and stored at ?70°C.
Nucleocapsid preparations containing approximately
30 ?g (5 ?l) of proteins were subjected to denaturing
electrophoresis in a 5–15%gradientSDS-polyacrylamide
gel according to Laemmli (1970) buffered with Tris pH
8.8. Proteins were transferred to a PVDF membrane
(Immobilon) by a semi-dry transfer blot (Transblot SD,
Bio-Rad) according to manufacturer’s recommendations.
Filters were blocked with 5% nonfat milk in 0.01 M po-
tassium phosphate buffer, pH 7.2, containing 0.14 M
NaCl, and 0.05% Tween 20 (PBS-T). Filters were then
incubated with antibodies directed to the amino- (amino
acids positions 1–287) orcarboxyl-terminal (amino acids
positions 2293–2875) ends of the L protein (van Poelwijk
et al., 1993) in PBS-T. The antigen-IgG complex formed
was detected using horseradish conjugated anti-rabbit
antibodies, and visualized using the ECL detection kit of
Alice K.Inoue-Nagata was a doctoral studentofUniversity ofBrası ´ lia
(Brazil) and was supported by a fellowship fromthe National Research
Council of Brazil (CNPq). Jean-Yves Sgro is supported by a grant from
the Lucille P. Markey Charitable Trust. We thank Marcel Prins and
Erwin Cardol for helpful discussions.
Akkina, R. K., Chambers, T. M., and Nayak, D. P. (1984a). Expression of
defective-interfering influenza virus specific transcripts and polypep-
tides in infected cells. J. Virol. 51, 395–403.
Akkina, R. K., Chambers, T. M., and Nayak, D. P. (1984b). Mechanismof
interference by defective-interfering particles of influenza virus: Dif-
ferential reduction of intracellular synthesis of specific polymerase
proteins. Virus Res. 1, 687–702.
Barret, A. D. T., and Dimmock, N. J. (1986). Defective interfering viruses
and infections ofanimals. Curr. Top. Microbiol. Immunol. 128, 55–84.
Baudin, F., Bach, C., Cusack, S., and Ruigrok, R. W. H. (1994). Structure
ofinfluenza virus RNP. I. Influenza virus nucleoprotein melts second-
ary structure in panhandle RNA and exposes the bases to the
solvent. EMBO J. 13, 3158–3165.
Bruccoleri, R. E., and Heinrich, G. (1988). An improved algorithm for
nucleic acid secondary structure display. Comput. Appl. Biosci. 4,
Carpenter, C. D., Cascone, P. J., and Simon, A. E. (1991). Formation of
multimers of linear satellite RNAs. Virology 183, 586–594.
Carpenter, C. D., Oh, J. W., Zhang, C., and Simon, A. E. (1995). Involve-
ment of a stem-loop structure in the location of junction sites in viral
RNA recombination. J. Mol. Biol. 245, 608–622.
Cascone, P. J., Carpenter, C. D., Li, X. H., and Simon, A. E. (1990).
Recombination between satellite RNAs of turnip crinkle virus. EMBO
J. 9, 1709–1715.
Cascone, P. J., Haydar, T. F., and Simon, A. E. (1993). Sequences and
structures required for recombination between virus-associate
RNAs. Science 260, 801–805.
Cooper, P. D., Steiner-Pryor, A., Scotti, P. D., and Delong, D. (1974). On
the nature of poliovirus genetic recombinants. J. Gen. Virol. 23,
Dalmay, T., Zittya, G., and Burgya ´n, J. (1995). Generation of defective
interfering RNA dimers of cymbidiumringspot tombusvirus. Virology
de A´vila, A. C., de Haan, P., Smeets, M. L. L., Resende, R. de O.,
Kormelink, R., Kitajima, E. W., Goldbach, R. W., and Peters, D. (1993).
Distinct levels of relationships between tospovirus isolates. Arch.
Virol. 128, 211–227.
de Haan, P., Kormelink, R., Resende, R. de O., van Poelwijk, F., Peters,
D.,andGoldbach,R.(1991).Tomatospottedwiltvirus L RNA encodes
a putative RNA polymerase. J. Gen. Virol. 71, 2207–2216.
de Haan, P., Wagenmakers, L., Peters, D., and Goldbach, R. W. (1989).
Molecularcloning andterminal sequence determinationofthe S and
M RNA of tomato spotted wilt virus. J. Gen. Virol. 70, 3469–3473.
de Haan, P., Wagenmakers, L., Peters, D., and Goldbach, R. W. (1990).
The S RNA segment of tomato spotted wilt virus has an ambisense
character. J. Gen. Virol. 71, 1001–1007.
Devereux, J., Haeberli, P., and Smithies, O. (1984). A comprehensive set
of sequence analysis programs for the VAX. Nucleic Acids Res. 12,
de Vries, S. C., Springer, J., and Wessala, J. G. H. (1982). Diversity of
abundant mRNA sequences and patterns of protein synthesis in
etiolated and greened pea seedlings. Planta 156, 129–135.
Dzianott, A., Flasinski, S., and Bujarski, J. J. (1995). Foreign complemen-
tary sequences facilitate genetic RNA recombination in brome mo-
saic virus. Virology 208, 370–375.
Finnen, R. L., and Rochon, D. M. (1995). Characterization and biological
activity of DI RNA dimers formed during cucumber necrosis coinfec-
tions. Virology 207, 282–286.
Graves, M. V., Pogany, J., and Romero, J. (1996). Defective interfering
RNAs anddefective viruses associatedwithmultipartite RNA viruses
of plants. Sem. Virol. 7, 399–408.
Havelda, Z., Dalmay, T., andBurgya ´n, J.(1995).Localizationofcis-acting
sequences essential for cymbidium ringspot tombusvirus defective
interfering RNA replication. J. Gen. Virol. 76, 2311–2316.
Havelda, Z., Dalmay, T., and Burgya ´n, J. (1997). Secondary structure-
dependent evolution of cymbidium ringspot virus defective interfer-
ing RNA. J. Gen. Virol. 78, 1227–1234.
Hillman, B. I., Carrington, J. C., and Morris, T. J. (1987). A defective
interfering RNA that contains a mosaic of a plant virus genome. Cell
Huang, A. S. (1973). Defective interfering viruses. Annu. Rev. Microbiol.
Huang, A. S., and Baltimore, D. (1970). Defective viral particles and viral
disease processes. Nature 226, 325–327.
Inoue-Nagata, A.K., Kormelink, R., Nagata, T., Kitajima, E.W., Goldbach,
R., and Peters, D. (1997). Temperature and host effects on the gen-
eration of tomato spotted wilt virus defective interfering RNAs. Phy-
topathology 87, 1168–1173.
Jaeger, J.A., Turner, D.H., andZuker, M.(1989).Improvedpredictions of
355DI RNAs CHARACTERIZED IN TSWV
secondary structures for RNA. Proc. Natl. Acad. Sci. USA 86, 7706–
Jennings, P. A., Finch, J. T., Winter, G., and Robertson, J. S. (1983). Does
the higher order structure of the Influenza virus ribonucleoprotein
guide sequence rearrangements in influenza viral RNA? Cell 34,
Kascsak, R. J., and Lyons, M. J. (1978). Bunyamwera virus. II. The
generation and nature of defective interfering particles. Virology 89,
Kormelink, R., de Haan, P., Meurs, C., Peters, D., and Goldbach, R.
(1992).The nucleotide sequence ofthe M segmentoftomatospotted
wilt virus, a bunyavirus with two ambisense RNA segments. J. Gen.
Virol. 73, 2795–2804.
Kormelink, R., Storms, M., van Lent, J., Peters, D., and Goldbach, R.
(1994). Expression and subcellular location of the NSm protein of
tomato spotted wilt virus (TSWV), a putative viral movement protein.
Virology 200, 56–65.
Kuge, S., Saito, I., and Nomoto, A. (1986). Primary structure ofpoliovirus
defective-interfering particle genomes and possible generation
mechanisms of the particles. J. Mol. Biol. 192, 473–487.
Laemmli, U. K. (1970). Cleavage of structural proteins during the as-
sembly of the head of bacteriophage T4. Nature 227, 680–685.
Lai, M. M. C. (1992). RNA recombination in animal and plant viruses.
Microbiol. Rev. 56, 61–79.
Lazzarini, R. A., Keene, J. D., and Schubert, M. (1981). The origin of
defective interfering particles of the negative-strand RNA viruses.
Cell 26, 145–154.
Logermann, J., Schell, J., and Willmitzer, L. (1987). Improved method for
the isolation of RNA from plant tissues. Anal. Biochem. 163, 16–20.
Mills, D. R., Dobkin, C., and Kramer, F. R. (1978). Template-determined,
variable rate of RNA chain elongation. Cell 15, 541–550.
Mohamed, N. A., Randles, J. W., and Francki, R. I. B. (1973). Protein
composition of tomato spotted wilt virus. Virology 56, 12–21.
Nagy, P. D., and Bujarski, J. J. (1993). Targeting the site of RNA-RNA
recombination in brome mosaic viruses with antisense sequences.
Proc. Natl. Acad. Sci. USA 90, 6390–6394.
Nagy, P. D., and Bujarski, J. J. (1996). Homologous RNA recombination
in brome mosaic virus: AU-rich sequences decrease the accuracy of
crossovers. J. Virol. 70, 415–426.
Nayak, D. P., Chambers, T. M., and Akkina, R. K. (1982). Structure of
defective-interfering RNAs of Influenza viruses and their role in
interference. In‘‘The Influenza Viruses’’ (R. M. Krug, Ed.), pp. 269–317.
Plenum Press, New York.
Palmenberg, A. C., and Sgro, J.-Y. (1997). Topological organization of
picornaviral genomes: Statistical prediction of RNA structural sig-
nals. Sem. Virol. 8, 231–241.
Patel, A. H., and Elliott, R. M. (1992). Characterization of Bunyamwera
virus defective interfering particles. J. Gen. Virol. 73, 389–396.
Perrault, J. (1981). Origin and replication of defective interfering parti-
cles. Curr. Top. Microbiol. Immunol. 93, 151–207.
Peters, D., de A´vila, A. C., Kitajima, E. W., Resende, R. de O., de Haan,
P., and Goldbach, R. W. (1991). An overview of tomato spotted wilt
‘‘Proceedings ofUSDA Workshop, Beltsville, MD’’ (H.-T. Hsuand R. H.
Lawson, Eds.), pp. 1–14. National Technology Information Service,
Pogany, J., Romero, J., Huang, Q., Sgro, J.-Y., Shang, H., andBujarski, J.J.
(1995).De novogenerationofdefective interfering-like RNAs inbroad
bean mottle bromovirus. Virology 212, 574–586.
Pogany, J., Romero, J., and Bujarski, J. J. (1997). Effect of 5? and 3?
terminal sequences, overall length, and coding capacity on the
accumulation of defective RNAs associated with broadbean mottle
virus in planta. Virology 228,: 236–243.
Raikar, A., Rubino, H.M., andLockard, R.E.(1988).Chemical probing of
adenine residues within the secondary structure of rabbit ribosomal
RNA. Biochemistry 27, 582–592.
Resende,R.de O.,de Haan,P.,de A´vila,A.C.,Kitajima,E.W.,Kormelink,
R., Goldbach, R., and Peters, D. (1991). Generation of envelope and
defective interfering RNA mutants of tomato spotted wilt virus by
mechanical passage. J. Gen. Virol. 72, 2375–2383.
Resende, R. de O., de Haan, P., van de Vossen, E., de A´vila, A. C.,
Goldbach, R., and Peters, D. (1992). Defective interfering L RNA
segments of tomato spotted wilt virus retain both genome termini
and have extensive internal deletions. J. Gen. Virol. 73, 2509–2516.
Roux,L.,Simon,A.E.,Holland,J.J.(1991).Effects ofdefective interfering
viruses onvirus replicationandpathogenesis invitroandinvivo.Adv.
Virus Res. 40, 181–211.
Rowe, C. L., Fleming, J. O., Nathan, M. J., Sgro, J.-Y., Palmenberg, A. C.,
and Baker, S. C. (1997). Generation of coronavirus spike deletion
variants by high-frequency recombination at regions of predicted
RNA secondary structure. J. Virol. 71, 6183–6190.
Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). ‘‘Molecular Cloning.
A Laboratory Manual,’’ 2nd ed. Cold Spring HarborLaboratory Press,
Cold Spring Harbor, NY.
Scallan, M.F., andElliott, R.M.(1992).Defective RNAs inmosquitocells
persistently infected with Bunyamwera virus. J. Gen. Virol. 73, 53–60.
Schlesinger, S. (1988). The generation and amplification of defective
interfering RNAs. In ‘‘RNA Genetics’’ (E. Domingo, J. J. Holland, and P.
Ahlquist, Eds.), Vol. 2, pp. 167–185. CRC Press, Boca Raton, FL.
Simon, A. E., Engel, H., Johnson, R. P., and Howel, S. H. (1988).
Identification of regions affecting virulence, RNA processing and
infectivity in the virulent satellite of turnip crinkle virus. EMBO J. 7,
Tas, P. W. L., Boerjan, M. L., and Peters, D. (1977). The structural
proteins of tomato spotted wilt virus. J. Gen. Virol. 36, 267–279.
van der Most, R., Luytjes, W., Rutjes, S., and Spaan, W. J. M. (1995).
Translation but not the encoded sequence is essential for the effi-
cientpropagationofthe defective interfering RNAs ofthe coronavirus
mouse hepatitis virus. J. Virol. 69, 3744–3751.
van Poelwijk, F., Boye, K., Oosterling, R., Peters, D., and Goldbach, R.
(1993).Detectionofthe L proteinoftomatospottedwiltvirus.Virology
White, K. A., Bancroft, J. B., and Mackie, G. A. (1992). Coding capacity
determines in vivo accumulation of a defective RNA of clover yellow
mosaic virus. J. Virol. 66, 3069–3076.
White, K. A., and Morris, T. J. (1994). Nonhomologous RNA recombina-
tion in tombusviruses: Generation and evolution of defective inter-
fering RNAs by stepwise deletions. J. Virol. 68, 14–24.
White, K. A., and Morris, T. J. (1995). RNA determinants of junction site
selection in RNA virus recombinants and defective interfering RNAs.
RNA 1, 1029–1040.
Zhang, C., and Simon, A. E. (1994). Effect of template size on accumu-
lation of defective interfering RNAs in protoplasts. J. Virol. 68, 8466–
Zuker,M.(1989).Onfinding all suboptimal foldings ofanRNA molecule.
Science 244, 48–52.
356 INOUE-NAGATA ET AL.