Journal
of General Virology (1998), 79, 2349–2358. Printed in Great Britain
...................................................................................................................................................................................................................................................................................
Nucleotide sequence, genetic organization and expression
strategy of the double-stranded RNA associated with the
‘ 447 ’ cytoplasmic male sterility trait in Vicia faba
Pierre Pfeiffer
Institut de Biologie Mole! culaire des Plantes du CNRS, 12 Rue du Ge! ne! ral Zimmer, 67000 Strasbourg, France
The entire nucleotide sequence of the doublestranded (ds) RNA associated with the unconventional ‘ 447 ’ cytoplasmic male sterility (CMS) trait in
Vicia faba was determined from overlapping cDNA
clones and by RT–PCR. Confirming previous observations, it was found that the negative-strand was
continuous and 17 635 nt long, while the positivestrand featured an interruption, probably a nick,
that could potentially define two subgenomic RNAs
of 2735 nt and 14 900 nt, with the smaller RNA
being located on the 5« side. The entire positivestrand could encode a single in-frame ORF starting
at the first AUG at position 42–44 and ending with a
TGA at 17 517–17 519. This long potential polypeptide with a predicted molecular mass of 654 109
is the largest described to date in the plant kingdom
and contains conserved amino acid sequence motifs
typical of viral helicases and RNA-dependent RNA
Introduction
The cytoplasmic male sterility (CMS) trait of the ‘ 447 ’ line
of Vicia faba is unusual in that it does not result from
rearrangements of the mitochondrial DNA that lead to the
production of aberrant polypeptides as has been described in
several plant species (Dewey et al., 1986 ; Young & Hanson,
1987 ; Lewings, 1993), but rather correlates with the presence
of a double-stranded (ds) RNA of high molecular mass and
unknown origin (Grill & Garger, 1981). This dsRNA is
transmitted exclusively in a vertical mode, is absent from
maintainer lines and is permanently lost after restoration of
male fertility by crossing with a restorer line, or as a
consequence of spontaneous reversion to fertility (Scalla et al.,
1981). Early electron microscopic observations (Edwardson et
Author for correspondence : Pierre Pfeiffer.
Fax ­33 3 88 61 44 42. e-mail Pierre.Pfeiffer!ibmp-ulp.u-strasbg.fr
The EMBL accession number of the sequence reported in this paper is
AJ000929.
0001-5518 # 1998 SGM
polymerases (RDRP). Only limited sequence homology was detected with the ORF B encoded by the
hypovirulence-associated dsRNA of chestnut blight
fungus, a dsRNA replicon similarly contained in
host-derived membranous vesicles and considered
to share a common ancestry with potyviruses. By
contrast, the helicase and RDRP domains were in
the same respective arrangement and shared extensive sequence homologies with those identified
in the polyprotein encoded by the dsRNA isolated
from Japonica rice, another dsRNA replicon featuring a specific nick in the positive-strand. Although
no proteolytic self-cleavage activity has yet been
demonstrated, it appears likely that this long ORF is
a polyprotein that undergoes proteolytic maturation, with one of the polypeptides derived by selfcleavage being the determinant of the CMS trait.
al., 1976), performed at a time when CMS was considered to be
of virus aetiology, revealed that all tissues of the male-sterile
plants contain cytoplasmic membranous vesicles that were
dubbed ‘ cytoplasmic spherical bodies ’ or ‘ virus-like particles ’,
and which disappeared after restoration of male fertility. We
have previously established (Lefebvre et al., 1990) that these
membranous vesicles do indeed contain the dsRNA described
by Scalla et al. (1981) and Grill & Garger (1981), and that this
dsRNA is associated with a specific RDRP. These structures
probably correspond to a virus lacking a capsid protein which
is maintained in the form of a replicative complex contained in
host-derived membranous vesicles rather than as virions, very
reminiscent of the hypovirulence-associated viruses of the
chestnut blight fungus (Hansen et al., 1985). The structure of
this dsRNA was unusual, with the negative-strand being
continuous while the positive-strand was interrupted and
could therefore be potentially divided in two subgenomic
RNAs (Pfeiffer et al., 1993). Run-on RNA synthesis performed
in vitro with transcription}replication complexes isolated from
the cytoplasmic vesicles confirmed that a smaller subgenomic
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CDEJ
P. Pfeiffer
RNA was preferentially synthesized, but no information was
available on its coding capacity or on the various proteins
specified by the dsRNA in general. In addition, any evidence to
support the involvement of the dsRNA in the male sterility
trait remained circumstantial, since all attempts to convert
fertile plants to male sterility by inoculation or by transfer of
the dsRNA failed (Duc et al., 1984 ; Turpen et al., 1988). This
precluded demonstration of a cause-and-effect relationship and
prevented potential extension of this CMS to other plants of
agronomical interest.
In the case of the chestnut blight fungus, it has been
demonstrated that the virus-mediated changes do not result
from a general debilitation of the host, but rather from the
differential regulation of specific genes. Thus, ORF A of the
virus was identified as the determinant of traits associated with
alterations of fungal phenotype, such as reduced pigmentation,
and its proteolytic processing was, for instance, demonstrated
to be required for reduction of fungal conidiation (Craven et al.,
1993). ORF B controls virulence attenuation, and recent
progress in the elucidation of the signal transduction pathways
that govern fungal pathogenesis has identified the role of G
protein-mediated cAMP accumulation on the reduction of
fungal virulence (for a review, see Nuss, 1996).
The only data available so far for the ‘ 447 ’ dsRNA were
limited to the description of the degeneration of the microspores and tapetal cells in the male-sterile line, but provided no
clues as to the mechanism involved. The elucidation of the
complete sequence of the ‘ 447 ’ dsRNA should now allow us to
explore these avenues and to assess how the individual viral
gene products that are most probably generated by autoproteolysis of the polyprotein interfere with pollen viability.
Methods
+ Isolation and purification of dsRNA. Briefly, dsRNA was
extracted from flower buds or young leaves homogenized in grinding
buffer as described (Lefebvre et al., 1990). After phenol extraction and
ethanol precipitation, the nucleic acids were solubilized in TE, adjusted to
2 M LiCl, 5 mM EDTA, and incubated overnight at 0 °C. After
centrifugation, soluble nucleic acids (soluble RNA, dsRNA and DNA)
were ethanol-precipitated from the supernatant, pelleted and resuspended
in water. Contaminating plant DNA was digested with DNase, and
dsRNA was either purified by two cycles of CF-11 chromatography
(Morris & Dodds, 1979) or by electrophoresis through a 1 % agarose gel
run in TAE buffer and electroelution into dialysis tubing.
+ Polyadenylation of dsRNA. Purified, undenatured dsRNA was
polyadenylated in vitro using E. coli poly(A) polymerase (BRL) as
described by Sippel (1973) ; more efficient polyadenylation was obtained
at later stages using yeast poly(A) polymerase (Amersham).
+ Reverse transcription of denatured dsRNA and molecular
cloning of cDNA. Several methods were used to generate cDNA from
dsRNA, and full denaturation of dsRNA was found to be the limiting
parameter for successful reverse transcription. Both oligo(dT) equipped
"(
with a cloning cassette (total length 36 nt) and random hexamers or
specific sequence-derived primers were used to generate cDNA clones.
Only short clones were obtained after denaturation by heat with or
CDFA
without DMSO (Cashdollar et al., 1982 ; Asamizu et al., 1985), possibly
due to incomplete denaturation or to rapid snapback of the separated
strands. On the contrary, denaturation with 10 mM methylmercuric
hydroxide (MMH) consistently yielded the best results, provided that the
denaturation mix was directly pipetted into the prewarmed reverse
transcription mix without prior neutralization of the MMH by mercaptoethanol (Antoniw et al., 1986 ; Jelkmann et al., 1989). First-strand
synthesis was directed by RNase H-minus moloney murine leukaemia
virus reverse transcriptase (Promega) at 42 °C for 1±5–2 h. Second-strand
synthesis was performed according to Gubler & Hoffman (1983). High
molecular mass cDNA was purified by centrifugation through GlassMax
columns (Gibco BRL) or Sephacryl S-400 spin columns (Pharmacia). Sizeselected cDNA was then phosphorylated with T4 polynucleotide kinase,
blunted with Klenow or T4 DNA polymerase, and either cloned in
dephosphorylated pKS (Stratagene) cleaved at the EcoRV or SmaI site, or
treated with restriction enzymes and cloned in an appropriately digested
vector.
+ Gap bridging by RT–PCR. To bridge the gaps between nonoverlapping clones, high molecular mass first-strand cDNA was selected
by precipitation with 8 % PEG (Paithankar & Prasad, 1991) after alkaline
hydrolysis of the RNA template. PCR (30 cycles) was performed after an
initial denaturation step of 5 min at 95 °C. Under these conditions, even
the RNA–DNA hybrid obtained in the first-strand synthesis reaction
could be directly used as a template. Primers were designed after initial
alignment of the various clones, and PCR was performed using a mixture
of enzymes with proof-reading capability (Expand Long Template PCR
system, Boehringer Mannheim) to prevent PCR-induced mutations. PCR
products were generally blunt-end cloned after phosphorylation and
polishing of the ends, but some recalcitrant PCR products had to be
cloned using the T-vector system of Marchuk et al. (1991).
+ Determination of the extremities of the dsRNA. Free 3« OH
termini were polyadenylated in vitro and oligo(dT) equipped with a
"(
cloning cassette was then used to prime cDNA synthesis. For sequence
determination of those extremities of the dsRNA that did not support
efficient addition of poly(A), cDNA primed with termini-proximal
oligonucleotides was submitted to PCR amplification using the 3« and 5«
RACE (rapid amplification of cDNA ends ; Frohman et al., 1988) or SLIC
(single-strand ligation to single-stranded cDNA ; Dumas et al., 1991)
techniques.
+ DNA sequencing and homology searches. cDNA clones were
sequenced by the dideoxynucleotide chain-termination method using T7
DNA polymerase (Promega). In regions difficult to sequence, compressions and strong stops were overcome using 7-deazaGTP. Incorporation close to the primer was improved by addition of Mn#+ ions.
One strand was completely sequenced from a series of overlapping Exo
III deletion series, while the other strand was sequenced from internal
deletions derived from the restriction map thus obtained. At later stages
of this study, dye terminator cycle sequencing was performed with an
Applied Biosystems 373 (Perkin-Elmer) automatic sequencer. Fig. 2
shows the position of the most significant clones on the dsRNA : at least
three independent cDNA clones were analysed for sequence determinations, and sequences determined by RT–PCR (see below) were
derived from at least five independent clones. The detailed list of the
primers used and of the cDNA clones generated can be obtained from the
author on request. Sequence analysis, nucleotide and amino acid
comparisons were performed using the GCG package, or the DNA
analysis software DNAid and DNA Strider. For protein homology
searches, the improved BLAST algorithm of Altschul et al. (1997) was
used to search the EMBL, NBRF and SWISS-PROT databases, and the
PROSITE database (Bairoch & Bucher, 1994) was used to identify
potential function signatures.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CMS-associated dsRNA in V. faba
Results
M
a
b
c
d
Cloning of the small subgenomic RNA
Early observations (Grill & Garger, 1981) had revealed that
the ‘ 447 ’ dsRNA probably has an unusual structure, with at
least one nicked strand. This was later confirmed by the studies
of Lefebvre et al. (1990) and Pfeiffer et al. (1993) who
demonstrated that the coding strand was interrupted, thus
defining two potential subgenomic RNAs. ‘ Run-on ’ experiments performed in vitro with the cytoplasmic vesicles that
contain the dsRNA associated with its specific RNA-dependent
RNA polymerase (RDRP) showed that only the smaller RNA
was elongated, reaching a limit size estimated at 4±5 kb by
electrophoresis on a formaldehyde gel. This indicated its
preferential, if not exclusive, transcription by the RDRP in
these complexes. The existence of this subgenomic RNA and
the preferential accessibility of its 3« OH end were further
confirmed by in vitro labelling of the purified dsRNA by T4
RNA ligase-mediated addition of [$#P]pCp, which then allowed
a tentative determination of this 3«-terminal sequence (Pfeiffer
et al., 1993). This analysis also revealed that none of the 3« ends
of the dsRNA were polyadenylated, but that the 3« OH
terminus of the small subgenomic RNA was an efficient
substrate for poly(A) polymerase. In vitro polyadenylated
dsRNA was therefore used as a template to prime reverse
transcription with oligo(dT), and specific primers were subsequently used to extend the clones obtained towards the 5«
end of the RNA.
The quality of the cDNA synthesis primed by oligo(dT)
was routinely monitored by RT–PCR using as primers
oligo(dT) and the antisense oligonucleotide 141CT (nt
1985–2004) mapping at an estimated 750 bp from the 3« end
of the smaller subgenomic RNA : the intensity of the 786 nt
long PCR product obtained (Fig. 1, lane b) provided a semiquantitative estimate of the overall efficiency of the polyadenylation and of the RT reaction.
Surprisingly, a negative control performed with oligo(dT)
alone (Fig. 1, lane a) gave a self-PCR product of about 2±8 kb.
This suggested the presence of a run of U residues at that
distance from the 3« end on the small subgenomic RNA, with
the corresponding run of A residues on the cDNA acting as a
target for hybridization of oligo(dT). The cloned product was
2735 nt long (excluding the added homopolymeric tails) and
contained a single continuous ORF (ORF A) starting at the first
AUG, encountered at position 42–44. No significant ORFs
were found in any of the other phases or on the complementary
strand, and the presence of two stop codons in the same frame
upstream of this AUG suggested that this sequence constituted
the 5« untranslated region of the smaller subgenomic RNA.
However, AUG (42–44) does not lie in a favourable
nucleotide context (GCAAAUGAA) compared with the
consensus sequence for initiation of translation in plants
(AACAAUGGC, Lu$ tcke et al., 1987). Possibly, the reduced
length of the 5« untranslated sequence and the apparent
kbp
3·0
2·0
1·6
1·0
0·5
0·4
0·2
(+) strand
2·7 kb
141 CT (A)n
(A)n 141 NT Xho
(+) strand
14·9 kb
L15CT (A)n
(–) strand
Fig. 1. RT–PCR products derived from in vitro polyadenylated dsRNA.
Oligo(dT) (broken line) was used as a primer for reverse transcription, and
the primers used for subsequent PCR were oligo(dT) alone (lane a) or
oligo(dT) with 141CT as antisense primer (lane b). The intensity of the ca.
0±8 kb band of 141CT–oligo(dT) PCR product gives a semi-quantitative
estimate of the overall efficiency of the polyadenylation and RT reactions.
The 2±8 kb self-PCR product obtained with oligo(dT) alone (arrowhead in
lane a) resulted from a ‘ megaprimer ’ extension of the partially overlapping
cDNAs primed at the 3« OH of the small subgenomic RNA and at the 3«
OH of the negative-strand. The yield of this 2±8 kb band could be
increased by adjusting the PCR conditions (not shown). The band
intensities of the PCR products obtained with oligo(dT) and either 141NTXho (nt 364–339, lane c) or L15CT (nt 16946–16971, lane d) similarly
reflected the variable efficiency of in vitro polyadenylation of the 3« OH
extremities of the dsRNA, and their sizes matched exactly those in the
predicted structure.
preferential synthesis of this subgenomic RNA (Lefebvre et al.,
1990) may compensate for this poor context ; alternatively,
translation may initiate at the second AUG in this ORF, at
position 726–728 (ACCAAUGGG), which lies in a better
context, but which is located downstream of three other outof-frame AUGs. The use of such a strategy of ‘ leaky scanning ’
has recently been demonstrated in plum pox virus, a member
of the potyvirus family (Simo! n-Buela et al., 1997), and in vitro
transcription}translation experiments are underway to clarify
this point.
Attempts to extend cDNA clones further towards the 5«
end using oligonucleotides 141NT-Xho (nt 364–339) and
141NT-Ext (nt 105–83) as primers led to cDNA clones
extending up to nt 24. Similarly, PCR products obtained by 5«
RACE (Frohman et al., 1988) or SLIC (Dumas et al., 1991)
extended to nt 17. No clones longer than those corresponding
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CDFB
P. Pfeiffer
to the 2±8 kb oligo(dT)-directed self-PCR product were
obtained, and all of these started at nt position 1 or 2. Finally,
RT–PCR performed on oligo(dT)-primed cDNA with 141NTXho as antisense primer produced a PCR product corresponding to the sequence between the polyadenylated 3« OH
of the negative-strand and this primer. The length of the band
was 400 nt (Fig. 1, lane c), as predicted for termination near nt
1, and sequencing confirmed that the 5« end of the small
subgenomic RNA had been reached. The intensity of this band
was lower than that of the 141CT-oligo(dT) product, reflecting
the lower efficiency of in vitro polyadenylation of the 17 kb
negative-strand vs that of the small subgenomic RNA in the
dsRNA (Fig. 1, compare lanes b and c). This indicated in turn
that the ca. 2±8 kb PCR product obtained with oligo(dT) alone
(Fig. 1, lane a) resulted from a ‘ megaprimer ’ extension of
oligo(dT)-primed cDNA copies of the smaller subgenomic
RNA which hybridized to antisense cDNA copies of the
polyadenylated 17 kb negative-strand, and sequencing confirmed that the 5« end of the small subgenomic RNA mapped
opposite the 3« end of the negative-strand.
Reappraisal of the arrangement of the small and large
subgenomic RNAs and generation of cDNA clones
representative of the complete sequence of the ‘ 447 ’
dsRNA
In keeping with the general 3«-coterminal localization of
subgenomic RNAs in RNA plant viruses, the preferential
accessibility of the smaller subgenomic RNA to enzymes that
require a free 3« OH to add nucleotide(s) had favoured
mapping the smaller subgenomic RNA at the 3« end of the
positive-strand of the dsRNA molecule (Pfeiffer et al., 1993).
This implied in turn that the 3« end of the longer subgenomic
RNA would map at the discontinuity in the positive-strand,
where its lack of reactivity [e.g. to poly(A) polymerase]
suggested that it was either blocked as a 3« or a 2«–3« cyclic
phosphate (presumably produced by a nucleolytic cleavage
event), or inaccessible to these enzymes due to steric hindrance.
The reverse arrangement of the two subgenomic RNAs
was confirmed by taking advantage of a NcoI site (nt
2500–2505) to generate a clone (NcoB23) that extended
186 nt beyond the 3«-terminal nucleotide of the small subgenomic RNA. Assuming that the single-stranded interruption
is limited to a nick in the phosphodiester bond, the first AUG
codon encountered was at position 124–126 downstream of
that nick. Sequence determination with clones extending
further into the larger subgenomic RNA confirmed that it
encoded a continuous ORF (tentatively designated as ORF B)
that ended with a UGA stop codon at position 17 517–17 519
(i.e. 14 782–14 784 nt from the nick). All other reading frames
present on both strands contained numerous stop codons and
are unlikely to be translated into proteins.
Clones spanning the 3« region of ORF B were generated
with several specific oligonucleotides and by 3« RACE (Fig. 1,
lane d), and the one that extended the farthest after the stop
CDFC
codon of ORF B was assumed to contain the 3« terminus of the
larger subgenomic RNA, which was 14900 nt long. The 3«
untranslated region was 116 nt long and devoid of a poly(A)
tail in the native dsRNA. The nucleotide sequence is deposited
in the EMBL database (AJ000929) and is not duplicated here.
One or two ORFs?
The ‘ 447 ’ dsRNA shares remarkable similarities with the
dsRNA associated with hypovirulence in the chestnut blight
fungus [Cryphonectria (Endothia) parasitica]. This dsRNA replicon is also contained in membranous vesicles (Hansen et al.,
1985), in which it is associated with a specific RDRP (Fahima et
al., 1993). The virus-like genetic organization of the chestnut
blight fungus dsRNA and its expression strategy (Shapira et al.,
1991) suggest that it shares a common ancestry with
potyviruses (Koonin et al., 1991). It was first referred to as
HAV, for chestnut blight fungus hypovirulence-associated
virus, and later renamed CHV1, for Cryphonectria hypovirulence-associated virus 1. This dsRNA features two ORFs
that are translated in a ‘ stop}restart ’ mode and undergo
cotranslational autoproteolysis (Choi et al., 1991 b). ORF A by
itself is able to suppress fungal sporulation and reduce
pigmentation and laccase accumulation (Choi & Nuss, 1992),
and this suppressive activity was mapped by Craven et al.
(1993) to the autocatalytic papain-like protease p29, resembling the potyvirus-encoded protease HC-Pro (Choi et al.,
1991 a). ORF B encodes the ‘ house-keeping genes ’ responsible
for the helicase, polymerase and a second protease function.
Depending on the virus strain, ORF A may or may not encode
a protease.
Sequencing multiple clones of cDNA and of the 141CToligo(dT) PCR product confirmed, however, that there was no
stop codon delimiting ‘ ORF A ’ and that the ORF continued
through the discontinuity into the other subgenomic RNA.
Such in-frame continuation of ‘ ORF A ’ into ‘ ORF B ’ suggests,
therefore, that translation of a single, continuous, 5825 aa long
polypeptide can occur, possibly from those rare molecules of
dsRNA that feature an uninterrupted positive-strand (Pfeiffer et
al., 1993). Given their size and the relationship of the ‘ 447 ’
dsRNA with potyviruses (see below), it is at any rate expected
that such polyprotein(s) undergo(es) proteolytic self-maturation. By analogy with CHV1 (Shapira et al., 1991) and for
comparison purposes, however, the two blocks of genetic
information located upstream and downstream of the nick will
be referred to as ‘ ORF A ’ and ‘ ORF B ’, respectively. Fig. 2
highlights the important features of the ‘ 447 ’ CMS-associated
dsRNA of V. faba and summarizes its genetic organization as
deduced from its complete nucleotide sequence.
Comparison of the polypeptide(s) encoded by the
V. faba dsRNA and by other RNA replicons : ‘ ORF A ’ is
specific to the ‘ 447 ’ dsRNA
All cases of CMS elucidated to date result from the
expression of abnormal or chimaeric mitochondrial protein
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CMS-associated dsRNA in V. faba
Fig. 2. Map of the major cDNA and RT–PCR clones (A), coding capacity (B), main sequence features (C) and structure (D) of
the dsRNA associated with the ‘ 447 ’ CMS in V. faba. (A) Clones pZO140–143 (Turpen et al., 1988) are indicated by hatched
shading, the major cDNA clones by open boxes, 3«RACE clones of the extremities are black, and RT–PCR clones bridging the
gaps are shaded grey. The positive-strand features a 41 nt long 5« UTR with two in-frame stop codons upstream of the AUG
42–44, and a 116 nt long 3« UTR with no ORFs of significant length after UGA (17517–17519). The single long ORF encoded
by the dsRNA is interrupted by a discontinuity in the coding strand at nt 2735 (arrow). The region ‘ B ’ of the 5825 aa long
potential protein encoded by this ORF contains the conserved helicase and RDRP motifs typical of RNA viruses which are
detailed in Fig. 4.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CDFD
P. Pfeiffer
variants that usurp the role of their functional counterparts in
the anthers, thus disrupting mitochondrial function in tissues
where energy demand peaks during pollen development
(Lewings, 1993). To find clues for the mechanism of ‘ 447 ’
CMS, ‘ ORF A ’ (suspected to be the CMS determinant) was
analysed for sequence motifs that could be indicative of a
specific function. The most prominent feature of this tentative
polypeptide was the presence of a fumarate lyase signature
spanning residues 492–501 (GSTNMFSKNN), which matches
the short conserved sequence (GSx2Mx2KxN) centred around
an essential methionine residue probably involved in catalytic
activity (Bairoch & Bucher, 1994). In addition, a putative
arabinogalactan protein motif SPPAP (C. Reuzeau, personal
communication) was found at the N terminus (residues 13–17)
of ‘ ORF A ’ (see Fig. 2 B), suggesting that ‘ ORF A ’, or a
derivative thereof, can be glycosylated. In this respect, it is
noteworthy that most of the antigenicity of the dsRNAcontaining cytoplasmic vesicles (Dulieu et al., 1988) corresponds to epitopes that contain sugars (Desvoyes & Dulieu,
1996). No further significant or extensive sequence homology
could be detected between ‘ ORF A ’ and any other proteins of
virus or other origin, strengthening the contention that the
‘ ORF A ’ region is a specific hallmark of this dsRNA.
‘ ORF B ’ encodes the replication functions
Assuming that the positive-strand is interrupted by a
simple nick and that the larger subgenomic RNA actually
represents a functional messenger, the 5«-proximal initiation
codon for ORF B would be the AUG at nt position 124–126
from that nick. No significant ORFs were found in any of the
other phases, nor was there any other AUG codon found outof-frame upstream of the potential initiator. The nucleotide
context of this first AUG (AAACAUGUGC) is not optimal,
and here too, in vitro translation experiments are required to
decide whether this AUG can be effectively used to initiate
translation, or whether initiation may occur at other downstream AUG codons that lie in a more favourable context.
Even if translation of the larger subgenomic RNA were to
initiate at the first initiation codon found after the nick, the
resulting translation product corresponding to ORF B would
still contain 4886 aa and would be longer than the 4572 residue
long protein encoded by the dsRNA present in cultivated rice
(Moriyama et al., 1995), and reported to be the longest ORF in
the plant kingdom to date. (It is, however, likely that the
unique ORF is translated into a single 5825 aa polyprotein
covering ‘ ORF A ’ and ‘ ORF B ’, and amino acid residue
numbering will therefore start from the first methionine residue
encoded by the AUG at nt position 42–44.)
ORF B featured two short CCGVN perfect repeats
separated by 50 aa (aa 1306–1310 and 1360–1364), and two
longer repeats RLE(S}F)NGFTNFNT(N}E)LVN (aa 2650–
2665 and 2672–2687) separated by six amino acids and
followed by a short incomplete RLEF sequence (aa 2752–2755),
CDFE
Fig. 3. Alignment of the helicase and polymerase regions of the
polyprotein encoded by the V. faba ‘ 447 ’ dsRNA (VFdsRNA) and by the
rice dsRNA (RDR). The highest degree of sequence conservation is found
around the invariant amino acid residues of helicase motifs I (GXGKS) and
II (DE), and around the RDRP motifs V (TGQXXTXXXN) and VI (GDD)
indicated in bold.
which may indicate gene duplication events or be associated
with transmembrane regions.
The sequencing of the dsRNA associated with hypovirulence in Cryphonectria (Endothia) parasitica (Shapira et al.,
1991) and of the symptomless rice dsRNA (Moriyama et al.,
1995) has established without ambiguity that such high
molecular mass dsRNAs, which are associated with a specific
polymerase and are contained within membranous vesicles
(Lefebvre et al., 1990 ; Fahima et al., 1993), have a virus-like
genetic organization and synthesize polyprotein precursors
that are subsequently processed into functional polypeptides
by the dsRNA-encoded protease function(s). In the case of
Cryphonectria hypovirus 1-713 (CHV1), co-translational processing has been demonstrated to occur for both ORF A and
ORF B (Shapira et al., 1991) and mutational analysis has further
confirmed the existence of two virus-encoded papain-like
proteases (Choi et al., 1991 b ; Shapira & Nuss, 1991).
Like ORF B of CHV1 and the single ORF of the rice dsRNA
(Fukuhara et al., 1993), ORF B of the ‘ 447 ’ dsRNA contained
the conserved domains typical of RNA-helicase and RDRP
functions (Koonin et al., 1991) that are required for the
replication and transcription of these virus-like genetic elements. Computer-assisted sequence comparisons revealed that
the ‘ ORF B ’ region of the ‘ 447 ’ dsRNA was indeed distantly
related to ORF B of CHV1, but a much stronger relationship
was found with the single ORF encoded by the dsRNA from
Japonica rice (Moriyama et al., 1995 ; Fig. 3).
The matrix comparison of these two proteins (not shown)
revealed three regions of high homology. The highest
incidence and length of clusters of identical amino acids was
found around the conserved motifs of the polymerase region
located towards the C terminus of ORF B (aa 5250–5650, with
41 % amino acid identity and 58 % similarities with only three
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CMS-associated dsRNA in V. faba
gaps introduced), followed by the motifs in the helicase region
(aa 1950–1996 with 47 % amino acid identity and 56 %
similarities, and aa 2023–2043 with 35 % amino acid identity
and 56 % similarities), with a BLASTP score of 392. A more
detailed analysis of the phylogeny of the dsRNA replicons and
of their classification based on these sequence similarities will
be published elsewhere (M. J. Gibbs, R. Koga, H. Moriyama, P.
Pfeiffer & T. Fukuhara, unpublished data), but it is interesting to
note that the next closest protein in the databases was the
replicase of hepatitis E virus, with a BLASTP score of 118. In
this virus, however, the helicase region precedes immediately
the RDRP region, while these are separated by some 3000 aa
residues in the ‘ 447 ’ dsRNA.
In addition to the helicase and RDRP motifs which we
anticipated to find in the polyprotein, searches using the
PROSITE base identified a potential β1–3 glucanase signature
(VLMGLENEVL) where the underlined conserved glutamic
acid is an active site residue in glycosyl hydrolases (Py et al.,
1991). It remains to be established, however, whether any
specific enzymatic activities correspond to the sequence
homologies detected.
Discussion
High molecular mass dsRNAs have been found in several
non-virus-inoculated plant species where they have long
remained undetected because of the lack of obvious symptoms.
Their presence is generally restricted to a single line or cultivar
in which they are transmitted efficiently in a vertical manner
through both egg and pollen. Zabalgogeazcoa et al. (1993), for
example, traced the presence of a 13±2 kb dsRNA in the
‘ Barsoy ’ cultivar of barley to an ancestral progenitor imported
from Japan at the beginning of this century, and retrieved this
dsRNA from the various intermediate crosses performed. The
presence of the dsRNA seemed therefore to confer some
desirable agronomic trait(s) during the selection scheme rather
than cause adverse effects.
Such unique, vertically transmitted 12–14 kb linear
dsRNAs are also present in specific cultivars of pepper
(Valverde et al., 1990), in the Black Turtle Soup bean
(Wakarchuk & Hamilton, 1985) and in cultivated rice Oriza
sativa ssp. japonica (Wang et al., 1990). In the latter two species,
their presence was thought to be associated with CMS, but this
hypothesis was later disproved (Mackenzie et al., 1988 ;
Fukuhara et al., 1993). In addition, elucidation of the molecular
mechanisms of CMS (for a review, see Lewings, 1993)
subsequently confirmed that this phenotype does not to result
from a virus infection, but rather is associated with mitochondrial DNA rearrangements that lead to the production of
chimaeric, non-functional polypeptides.
The dsRNA found in the ‘ 447 ’ male-sterile line of V. faba
is the only apparent exception to this mechanism, since malesterile and restored F1 plants have indistinguishable mitochondrial DNA patterns (Turpen et al., 1988). By contrast,
plants that have spontaneously reverted to fertility no longer
contain the dsRNA, and both the dsRNA and the male-sterility
trait are lost irreversibly in the progeny obtained after
fertilization of the male-sterile line with restorer pollen.
Determination of the entire sequence of this dsRNA has now
demonstrated that it belongs to a family of RNA replicons
encoding long polyproteins that are subsequently processed
into functional peptides by an endogenous protease, a strategy
used by many RNA viruses (Spall et al., 1997).
The size, genome organization and sequence similarities
found in the helicase, polymerase and protease domains of
CHV1-713, the type member of Hypoviridae (Hillman et al.,
1995), led Koonin et al. (1991) to suggest that this dsRNA
replicon shares a common ancestry with potyviruses. The rice
dsRNA and the ‘ 447 ’ dsRNA differ, however, from CHV1 in
the order of the helicase and polymerase domain, and in the
absence of a long non-coding sequence and poly(A) tail ; in
addition, they differ in possessing a single ORF spanning the
entire length of the dsRNA instead of two ORFs being read in
a stop}restart mode in CHV1 and related hypoviruses. Such
differences, together with the presence of a nick in the coding
strand, suggest that these two dsRNAs may be classified into
a family other than Hypoviridae. They are at any rate more
closely related to positive-stranded than to dsRNA viruses,
and computer-assisted analysis showed a clear-cut affinity of
the helicase and RDRP regions with hepatitis E virus, a virus
that also encodes a papain-like cysteine protease. Due to their
low degree of conservation, protease functions are difficult to
identify in viruses (Koonin et al., 1992 ; Rozanov et al., 1995)
and are generally inferred from functional studies and sitedirected mutagenesis that allow assignment of catalytic
residues and target sequences. Preliminary analysis of the long
ORF encoded by the ‘ 447 ’ dsRNA has identified an ILHAY
sequence (residues 572–576) similar to the IVHAY sequence of
the X domain (Koonin et al., 1992) found upstream of the
papain-like protease domain of Ross River virus, a member of
the ‘ alpha-like ’ supergroup ; the invariant proline located two
residues downstream in the X domain was, however, missing in
the polypeptide encoded by the ‘ 447 ’ dsRNA and in vitro
proteolysis studies will therefore be required to detect and
identify eventual protease activity associated with the expression of the ‘ 447 ’ dsRNA genome.
Unencapsidated dsRNA replicons have a plasmid-like lifecycle ; their copy number is controlled by the host but can vary
with the differentiation and developmental stage. The rice
dsRNA, for instance, is kept at a constant concentration
estimated at 100 copies per cell, but this concentration increases
10-fold in suspension cultures and 100-fold in pollen without
affecting its viability (H. Moriyama, personal communication).
Likewise, the dsRNA content of the ‘ 447 ’ line of V. faba was
found to depend on the nuclear background (G. Duc, personal
communication).
Sequence comparisons have revealed that the ‘ 447 ’ dsRNA
is most closely related to the rice dsRNA, with stretches of
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CDFF
P. Pfeiffer
Fig. 4. Similarities in the genetic and
functional organization of potyviruses,
bymoviruses and dsRNA replicons
possibly derived from potyviruses or
other ssRNA viruses. Pro, protease
domain ; Hel, helicase domain ; Pol, RDRP
domain. The arrow in the rice dsRNA and
in the V. faba dsRNA points to the
discontinuity in the coding strand. The
slanted line in CHV1-713, the
hypovirulence-associated dsRNA in
chestnut blight, indicates the stop/restart
that separates ORF A from ORF B. It is
suggested that the discontinuity in the
coding strand, and the stop/restart
arrangement represent steps towards a
bipartite genetic organization as found in
the bymoviruses.
protein sequence being highly conserved around the helicase
and polymerase domains. An additional, intriguing feature
specific to these two RNAs is the presence of a nick in the
positive-strand, interrupting the long ORF they encode
(Fukuhara et al., 1995 ; and this work). Although it is not known
at present how this interruption is generated and whether it is
relevant to the translation of these dsRNAs, it is tempting to
speculate that this could constitute an alternative means of
expressing the genetic information in two separate blocks
instead of using the stop}restart arrangement of the hypovirulence-associated virus CHV1, or a bipartite genome as in
the bymoviruses. In this respect, it is striking that the most
conserved part of these RNA replicons or viruses corresponds
to the domain of the polyprotein endowed with the functions
related to RNA replication and proteolytic self-maturation,
while ORF A or its equivalent is much more variable and
probably reflects the acquisition of specific properties, such as
transmission by specialized vectors or phenotypic effects on
their host. Fig. 4 summarizes the similitudes in the genetic
organization of these dsRNA replicons compared to that of
potyviruses, bymoviruses and alpha-like viruses.
It is not known at present whether the mRNA(s) derived
from the ‘ 447 ’ dsRNA feature a 5«-terminal structure such as a
cap or a VPg, and thus, whether a ‘ leaky scanning ’ translation
strategy may be used as in potyviruses (Simo! n-Buela et al.,
1997). Similarly, a polyprotein self-processing strategy can
only be inferred from comparisons with CHV1, potyviruses
and alpha-like viruses where this mechanism has been investigated in detail. Finally, it is hoped that transformation
with ‘ ORF A ’ or with specific domains of ORF A­B will
result in male-sterile plants and will allow a better understanding of the mechanism by which the ‘ 447 ’ dsRNA induces
CDFG
male sterility. Indeed, several signatures – fumarate lyase,
thioredoxin and β1–3 glucanase – have been identified in the
polyprotein(s) and suggest various potential mechanisms can
cause male sterility. In petunia for instance, male sterility has
been attributed to mis-timing of callose wall degradation by
callase, a β1–3 glucanase normally expressed at a specific stage
in the anther locule (Izhar & Frankel, 1971). This situation could
be reproduced experimentally in transgenic tobacco (Worall et
al., 1992) expressing a modified β1–3 glucanase, and in which
premature breakdown of the callose deposits resulted in male
sterility. It is tempting to speculate that a similar mechanism
may operate in the ‘ 447 ’ CMS, and expression studies should
help resolve this point.
I wish to thank Philippe Hammann for excellent assistance with
automated sequencing, Dr Rodrigo Valverde for help with some of the
experiments and useful discussions, Drs Hiromitsu Moriyama, Christophe
Reuzeau and Ge! rard Duc for communication of unpublished results, and
Drs Vivek Prasad and Ken Richards for help with the manuscript.
References
Altschul, S. F., Madden, T. L., Scha$ ffer, A. A., Zhang, J., Zhang, Z.,
Miller, W. & Lipman, D. J. (1997). Gapped BLAST and PSI-BLAST : a
new generation of protein database search programs. Nucleic Acids
Research 25, 3389–3402.
Antoniw, J. F., Linthorst, H. J. M., White, R. F. & Bol, J. F. (1986).
Molecular cloning of the double-stranded RNA of beet cryptic viruses.
Journal of General Virology 67, 2047–2051.
Asamizu, T., Summers, D., Motika, M. B., Anzola, J. V. & Nuss, D. L.
(1985). Molecular cloning and characterization of the genome of wound
tumor virus : a tumor-inducing plant virus. Virology 144, 398–409.
Bairoch, A. & Bucher, P. (1994). PROSITE : recent developments.
Nucleic Acids Research 22, 3583–3589.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CMS-associated dsRNA in V. faba
Cashdollar, L. M., Espaza, J., Hudson, G. R., Chmelo, R., Lee, P. W. K.
& Joklik, W. K. (1982). Cloning of the double-stranded RNA genes of
reovirus : sequence of the cloned S2 gene. Proceedings of the National
Academy of Sciences, USA 79, 7644–7648.
Choi, G. H. & Nuss, D. L. (1992). Hypovirulence of chestnut blight
fungus conferred by an infectious viral cDNA. Science 257, 800–803.
Choi, G. H., Pawlyk, D. M. & Nuss, D. L. (1991 a). The autocatalytic
protease p29 encoded by a hypovirulence-associated virus of the chestnut
blight fungus resembles the potyvirus-encoded HC-Pro. Virology 183,
747–752.
Choi, G. H., Shapira, R. & Nuss, D. L. (1991 b). Cotranslational
autoproteolysis involved in gene expression from a double-stranded
RNA genetic element associated with hypovirulence of the chestnut
blight fungus. Proceedings of the National Academy of Sciences, USA 88,
1167–1171.
Craven, M. G., Pawlyk, D. M., Choi, G. H. & Nuss, D. L. (1993). Papainlike protease p29 as a symptom determinant encoded by a hypovirulenceassociated virus of the chestnut blight fungus. Journal of Virology 67,
6513–6521.
Desvoyes, B. & Dulieu, P. (1996). Purification by monoclonal antibody
affinity chromatography of virus-like particles associated with the ‘ 447 ’
cytoplasmic male sterility of Vicia faba and investigation of their
antigenic composition. Plant Science 116, 239–246.
Dewey, R. E., Levings, C. S., III & Timothy, D. H. (1986). Novel
recombinations in the maize mitochondrial genome produce a unique
transcriptional unit in the Texas male-sterile cytoplasm. Cell 44, 439–449.
Duc, G., Scalla, R. & Lefebvre, A. (1984). New developments in
cytoplasmic male sterility in Vicia faba L. In Vicia faba : Agronomy,
Physiology and Breeding, pp. 254–260. Edited by P. D. Hebblethwaite,
T. C. K. Dawkins, M. C. Heath & G. Lockwood. The Hague}Boston}
Lancaster : Martinus Nijhoff}Dr W. Junk Publishers.
Dulieu, P., Penin, P., Dulieu, H. & Gautheron, D. C. (1988). Purification
of virus-like particles from Vicia faba and detection by ELISA in crude leaf
extracts. Plant Science 56, 9–14.
Dumas, J. B., Edwards, M., Delort, J. & Mallet, J. (1991). Oligodeoxyribonucleotide ligation to single-stranded cDNAs : a new tool for
cloning 5« ends of mRNAs and for constructing cDNA libraries by in vitro
amplification. Nucleic Acids Research 19, 5227–5232.
Edwardson, J. R., Bond, D. A. & Christie, R. G. (1976). Cytoplasmic
sterility factors in Vicia faba L. Genetics 82, 443–449.
Fahima, T., Kazmierczak, P., Hansen, D. R., Pfeiffer, P. & Van Alfen,
N. K. (1993). Membrane-associated replication of an unencapsidated
double-strand RNA of the fungus, Cryphonectria parasitica. Virology 195,
81–89.
Frohman, M. A., Dush, M. K. & Martin, G. R. (1988). Rapid production
of full-length cDNAs from rare transcripts : amplification using a single
gene-specific oligonucleotide primer. Proceedings of the National Academy
of Sciences, USA 85, 8998–9002.
Fukuhara, T., Moriyama, H., Pak, J. Y., Hyakutake, H. & Nitta, T.
(1993). Enigmatic double-stranded RNA in Japonica rice. Plant Molecular
Biology 21, 1121–1130.
Fukuhara, T., Moriyama, H. & Nitta, T. (1995). The unusual structure of
a novel RNA replicon in rice. Journal of Biological Chemistry 270,
18147–18149.
Grill, L. K. & Garger, S. J. (1981). Identification and characterization of
double stranded RNA associated with cytoplasmic male sterility in Vicia
faba. Proceedings of the National Academy of Sciences, USA 78, 7043–7046.
Gubler, U. & Hoffman, B. J. (1983). A simple and efficient method for
generating cDNA libraries. Gene 25, 263–269.
Hansen, D. R., Van Alfen, N. K., Gillies, K. & Powell, W. A. (1985).
Naked dsRNA associated with hypovirulence of Endothia parasitica is
packaged in fungal vesicles. Journal of General Virology 66, 2605–2614.
Hillman, B. I., Fulbright, D. W., Nuss, D. L. & Van Alfen, N. K. (1995).
Hypoviridae. In Virus Taxonomy. Sixth Report of the International Committee
on Taxonomy of Viruses, pp. 261–264. Edited by F. A. Murphy, C. M.
Fauquet, D. H. L. Bishop, S. A. Ghabrial, A. W. Jarvis, G. P. Martelli,
M. A. Mayo & M. D. Summers. Vienna & New York : Springer-Verlag.
Izhar, S. & Frankel, R. (1971). Mechanism of male sterility in Petunia :
the relationship between pH, callase activity in the anthers, and the
breakdown of microsporogenesis. Theoretical and Applied Genetics 41,
104–108.
Jelkmann, W., Martin, R. R. & Maiss, E. (1989). Cloning of four plant
viruses from small quantities of double-stranded RNA. Phytopathology
79, 1250–1253.
Koonin, E. V., Choi, G. H., Nuss, D. L., Shapira, R. & Carrington, J. C.
(1991). Evidence for common ancestry of a chestnut blight hypo-
virulence-associated double-stranded RNA and a group of positivestrand RNA plant viruses. Proceedings of the National Academy of Sciences,
USA 88, 10647–10651.
Koonin, E. V., Gorbalenya, A. E., Purdy, M. A., Rozanov, M. N., Reyes,
G. R. & Bradley, D. W. (1992). Computer-assisted assignment of
functional domains in the nonstructural polyprotein of hepatitis E virus :
delineation of an additional group of positive-strand RNA plant and
animal viruses. Proceedings of the National Academy of Sciences, USA 89,
8259–8263.
Lefebvre, A., Scalla, R. & Pfeiffer, P. (1990). The double-stranded RNA
associated with the ‘ 447 ’ cytoplasmic male sterility in Vicia faba is
packaged together with its replicase in cytoplasmic membranous vesicles.
Plant Molecular Biology 14, 477–490.
Lewings, C. S., III (1993). Thoughts on cytoplasmic male sterility. Plant
Cell 5, 1285–1290.
Lu$ tcke, H. A., Chow, K. C., Mickel, F. S., Moss, K. A., Kern, H. F. &
Scheele, G. A. (1987). Selection of AUG initiation codons differs in
plants and animals. EMBO Journal 6, 43–46.
Mackenzie, S. A., Pring, D. R. & Bassett, M. J. (1988). Large double-
stranded RNA molecules in Phaseolus vulgaris L. are not associated with
cytoplasmic male sterility. Theoretical and Applied Genetics 76, 59–63.
Marchuk, D., Drumm, M., Saulino, A. & Collins, F. (1991). Construction
of T-vectors, a rapid and general system for direct cloning of unmodified
PCR products. Nucleic Acids Research 19, 1154.
Moriyama, H., Fukuhara, T. & Nitta, T. (1995). Double-stranded RNA
in rice : a novel RNA replicon in plants. Molecular General Genetics 248,
364–369.
Morris, T. J. & Dodds, J. A. (1979). Isolation and analysis of doublestranded RNA from virus-infected plant fungal tissue. Phytopathology 68,
854–858.
Nuss, D. L. (1996). Using hypovirus to probe and perturb signal
transduction processes underlying fungal pathogenesis. Cell 8, 1845–
1853.
Paithankar, K. R. & Prasad, K. S. N. (1991). Precipitation of DNA by
polyethylene glycol and ethanol. Nucleic Acids Research 19, 1346.
Pfeiffer, P., Jung, J.-L., Heitzler, J. & Keith, G. (1993). Unusual structure
of the double-stranded RNA associated with the ‘ 447 ’ cytoplasmic male
sterility in Vicia faba. Journal of General Virology 74, 1167–1173.
Py, B., Bortoli-German, I., Haiech, J., Chippaux, M. & Barras, F. (1991).
Cellulase EGZ of Erwinia chrysanthemi : structural organization and
importance of His98 and Glu133 residues for catalysis. Protein Engineering
4, 325–333.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54
CDFH
P. Pfeiffer
Rozanov, M. N., Drugeon, G. & Haenni, A. L. (1995). Papain-like
Turpen, T., Garger, S. J. & Grill, L. K. (1988). On the mechanism of
proteinase of turnip yellow mosaic virus : a prototype of a new viral
proteinase group. Archives of Virology 140, 273–288.
cytoplasmic male sterility in the 447 line of Vicia faba. Plant Molecular
Biology 10, 489–497.
Scalla, R., Duc, G., Rigaud, J., Lefebvre, A. & Meignoz, R. (1981).
Valverde, R., Nameth, S., Abdallha, O., Al-Musa, O., Desjardins, P. &
Dodds, J. A. (1990). Indigenous double-stranded RNA from pepper
RNA-containing intracellular particles in cytoplasmic male sterile faba
bean (Vicia faba L.). Plant Science Letters 22, 269–277.
Shapira, R. & Nuss, D. L. (1991). Gene expression by a hypovirulenceassociated virus of the chestnut blight fungus involves two papain-like
protease activities. Journal of Biological Chemistry 266, 19419–19425.
Shapira, R., Choi, G. H. & Nuss, D. L. (1991). Virus-like genetic
organization and expression strategy for a double-stranded RNA genetic
element associated with biological control of chestnut blight. EMBO
Journal 10, 731–739.
Simo! n-Buela, L., Guo, H. S. & Garcı! a, J. A. (1997). Cap-independent
leaky scanning as the mechanism of translation initiation of a plant viral
genomic RNA. Journal of General Virology 78, 2691–2699.
Sippel, A. E. (1973). Purification and characterization of adenosine
triphosphate : ribonucleic acid adenyltransferase from Escherichia coli.
European Journal of Biochemistry 37, 31–40.
Spall, V. E., Shanks, M. & Lomonossof, G. P. (1997). Polyprotein
processing as a strategy for gene expression in RNA viruses. Seminars in
Virology 8, 15–23.
CDFI
(Capsicum annuum). Plant Science 67, 195–201.
Wakarchuk, D. A. & Hamilton, R. I. (1985). Cellular double-stranded
RNA in Phaseolus vulgaris. Plant Molecular Biology 5, 55–63.
Wang, B., Li, Y. N., Zhang, X. W., Hu, L. & Wang, J. Z. (1990). Doublestranded RNA and male sterility in rice. Theoretical and Applied Genetics
79, 556–560.
Worall, D., Hird, D. L., Hodge, R., Paul, W., Draper, J. & Scott, R.
(1992). Premature dissolution of the microsporocyte callose wall causes
male sterility in transgenic tobacco. The Plant Cell 4, 759–771.
Young, E. G. & Hanson, M. R. (1987). A fused mitochondrial gene
associated with cytoplasmic male sterility is developmentally regulated.
Cell 50, 41–47.
Zabalgogeazcoa, I., Cox-Foster, D. C. & Gildow, F. E. (1993). Pedigree
analysis of the transmission of a double-stranded RNA in barley cultivars.
Plant Science 91, 45–53.
Received 18 February 1998 ; Accepted 17 June 1998
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 17 Jun 2017 20:39:54