Download Phylogeny of Geminiviruses - Journal of General Virology

J. gen. Virol. (1989), 70, 2717-2727.
Printed in Great Britain
2717
Key words: evolution/geminivirus/coat protein/taxonorny
Phylogeny of Geminiviruses
By A L A N J. H O W A R T H * AND G E O R G E J. V A N D E M A R K
Department of Plant Pathology, University of Arizona, Tucson, Arizona 85721, U.S.A.
(Accepted 14 June 1989)
SUMMARY
Amino acid sequences of 16 geminivirus replication-associated proteins and 15 coat
proteins were aligned and a new computer program was used to calculate the minimum
mutation distances for all possible pairwise comparisons. These data were used to
construct phylogenetic trees. Trees based on coat proteins had two main branches
which were positively correlated with vector specificities of the viruses. Trees based on
replication-associated proteins also had two main branches which were positively
correlated with viral host specificities for either monocotyledonous or dicotyledonous
plants. Therefore, evolutionary pressures on coat proteins and replication-associated
proteins are probably highly influenced by vectors and hosts, respectively. Geminiviruses that infect dicotyledonous plants may be divided further by geographical
origins into Old World and New World viruses. These results suggest the possible
geographical origins of some geminiviruses, that new taxa should be erected, and have
implications for distinguishing viruses and strains.
INTRODUCTION
The taxonomy of viruses has improved greatly over the years as more information has become
available about the physical and chemical nature of viruses. Early attempts at classifying viruses
were based on characters such as symptomatology and host range which depended on the
genotype of both viruses and host organisms (Matthews, 1983). Current classification schemes
for distinguishing viruses into families and genera, or groups, rely on inherent traits of the
viruses, including type of nucleic acid, particle morphology and mode of replication (Matthews,
1982). Within viral taxa distinctions between viruses are often based on characteristics such as
host range and serological relationships (Francki, 1983).
Geminiviruses are of world-wide importance as pathogens of crops including cereals such as
maize and wheat (Goodman, 1981a). They are among the smallest known autonomously
replicating viruses (Howarth & Goodman, 1982). As a result, geminiviruses are among the most
intensively studied viruses for their potential usefulness in genetic engineering of plants (Buck &
Coutts, 1983). The genomes of at least 16 geminivirus isolates have been sequenced (Table 1).
Nucleic acid type (circular ssDNA) and a unique geminate particle morphology distinguish
geminiviruses from all other known viruses (Matthews, 1982). However, a dichotomy among
geminiviruses has been evident from the time that they were recognized as worthy of taxonomic
distinction (Harrison et al., 1977). This dichotomy has been viewed in terms of vector specificity
(leafhoppers or whiteflies), host range [monocotyledonous (monocots) or dicotyledonous (dicots)
plants], and, more recently, genomic complexity (segmented or unsegmented; Fig. 1 ; Goodman,
1981 b; Stanley, 1983, 1985; Grimsley et al., 1987). Based on these criteria three subgroups have
been suggested: (i) whitefly-transmitted, segmented genome, infects dicots; (ii) leafhoppertransmitted, unsegmented genome, infects dicots; (iii) leafhopper-transmitted, unsegmented
genome, infects monocots. Although it has been useful to categorize geminiviruses in these
ways, new sequence information provided an opportunity for further examination of
relationships among geminiviruses.
We have chosen to use molecular distance data to evaluate the relationships among
geminiviruses. Only two proteins, coat protein (gene I) and the replication-associated protein
0000-8991 © 1989 SGM
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
Mung bean yellow mosaic
Tomato golden mosaic
Tomato yellow leaf curl
Wheat dwarf
Cassava latentt
Chloris striate mosaic
Digitaria streak
Maize streak
Abutilon mosaic
Beet curly top
Bean golden mosaic
Virus
Brazil
California, U.S.A.
Puerto Rico
Brazil
Kenya
Australia
Vanuatu
Kenya
Nigeria
South Africa
Thailand
Brazil
Israel
Czechoslovakia
Czechoslovakia
Sweden
AbMV
BCTV
BGMV-PR
BGMV-BZ
CLV
CSMV
DSV
MSV-K
MSV-N
MSV-S
MYMV
TGMV
TYLCV
WDV-C
WDV-CJI
WDV-S
B. tabaci
B. tabaci
Nesoclutha pallida
Nesoclutha declivata
Cicadulina spp.
Cicadulina spp.
Cicadulina spp.
B. tabaci
B. tabaci
B. tabaci
Psammotettix alienus
P. alienus
P. alienus
Bemisia tabaci
Neoaliturus tenellus
B. tabaci
Vector
Malvaceae
44 families*
Leguminosae,
Malvaceae
Leguminosae
Solanaceae
Gramineae
Gramineae
Gramineae
Gramineae
Gramineae
Leguminosae
Solanaceae
Solanaceae
Gramineae
Gramineae
Gramineae
Host
Reference
Gilbertson et al. (1988)
Stanley & Gay (1983)
Andersen et al. (1988)
Donson et al. (1987)
Howell (1984)
Mullineaux et al. (1984)
Lazarowitz (1988)
Morinaga et al. (1987)
Hamilton et al. (1984)
N. Navot & H. Czosnek (unpublished)
Commandeur et al. (1987)
Woolston et al. (1988)
MacDowell et al. (1985)
Frischmuth et al. (1987)
Stanley et al. (1986)
Howarth et al. (1985)
* All 44 families which contain hosts of BCTV are dicotyledonous (Bennett, 1971).
t Cassava latent virus is the same as African cassava mosaic virus (Bock & Woods, 1983).
Origin
Abbreviation
T a b l e 1. O r i g i n s , v e c t o r s a n d h o s t s o f 16 g e m i n i v i r u s isolates
Z
t7
rn
Z
t7
>
o
~o
t,~
Geminivirus phylogeny
2719
(a)
DNA 1
(X
(b)
(c)
MSV-N
[!iiiiiiiiiili!ii!iiiii!!iii!!iiiiii!iiiiiii!ii!ii!i
BGMV
Fig. 1. Organization of geminivirus genomes and genes. The genomes of BGMV (a) and MSV (b) are
drawn schematically as represenative of dicot- and monocot-infecting geminiviruses, respectively.
Arrows depict the polarity and relative lengths of the conserved genes. Boxes are the common regions.
The small arc shows the binding site of the 80 nucleotide (approx.) 'primer' which is found with DNA in
virions of the monocot viruses. (c) Amino acid sequence similarity of MSV ORFs IIa and lib to BGMV
ORF II is indicated.
(gene II), are encoded in the genomes of all g e m i n i v i r u s e s characterized to date (Fig. 1 ; see
references in T a b l e 1). Phylogenetic trees were constructed based o n sequence a l i g n m e n t s of
each of these two proteins. I n a previous study i n v o l v i n g only six viruses, we used replicationassociated protein genes as the basis for a n a l y s i n g relationships b e t w e e n viruses ( H o w a r t h &
G o o d m a n , 1986).
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
2720
A. J. HOWARTH AND G. J. VANDEMARK
METHODS
The viruses analysed in this study and their abbreviations are listed in Table 1. Amino acid sequence alignments
of replication-associated proteins and coat proteins, respectively, were developed from the alignments of
MacDowell et al. (1985) and Stanley et al. (1986). Gaps were added as necessary to maximize similarities of
additional viral sequences. Then, each sequence was compared pairwise to every other sequence to determine
minimum mutation distances (Fitch & Margoliash, 1967), the minimum number of mutations required to convert
the sequence of a protein into the sequence of the protein with which it is compared. These numbers were
normalized for sequence length (Fitch & Margolish, 1967). Comparisons involving gaps were assigned a value of
zero and, thus, were omitted from the calculations. These computations were aided by a new computer program
which requires as input the aligned sequences and which gives as output a matrix of normalized mutation
distances. The program was designed to read an input file with a default length of l0 characters for the virus name
and 600 characters of single-letter code for the amino acid sequence of the peptide. The program can compare up
to 50 different viruses in one program execution. The name length, amino acid sequence length and maximum
number of viruses compared in one execution can be increased beyond the default limits by simple modifications
of the source code. The source code is a Pascal program which can be executed on any IBM-compatible personal
computer with a Turbo Pascal compiler or on any mainframe computer by invoking the Pascal compiler. The
program is available to interested users.
Rooted and unrooted trees were constructed from the minimum mutation distances using the Fitch and Kitsch
programs contained in the Phylip package of programs of Felsenstein 0985) as supplied for the IBM PC by
G. D. F. Wilson (Scripps Institution of Oceanography, La Jolla, Ca. 92093, U.S.A.). Phylogenetic trees
with the smallest sums of squares were obtained by at least three executions of each program using the jumble
and global rearrangement options.
RESULTS
Comparison of replication-associated proteins
Alignments of 16 amino acid sequences which correspond to the replication-associated
proteins of sequenced geminiviruses were compared in every pairwise combination. Minimum
mutation distances were calculated for each comparison (Table 2). After three executions of the
tree-construction program, we concluded that the best tree had been established because each
execution resulted in the same best tree and sum of squares (S.S. = 0.25). We used an F test
(Felsenstein, 1984) to compare the rooted tree produced by the Kitsch program, which assumes a
molecular clock, to the unrooted tree produced by the Fitch program, which assumes no clock.
The resulting highly significant F statistic of 35-0 led us to reject the hypothesis that there was no
difference between the rooted and unrooted trees. Therefore, we rejected the validity of the
molecular clock for these data. However, the same clusters of viruses were obtained with both
programs.
Fig. 2 depicts the best tree of an average of 777 trees that were examined for each execution of
the program. The average percent standard deviation of this tree is only 3.3. The principal
feature of the tree is that there are two main branches which correspond to the host specificities
of the viruses for either monocots or dicots. The horizontal lines in the monocot branch are
longer than in the dicot branch, i.e. there is less divergence among dicot-infecting viruses than
among monocot-infecting viruses. All of the viruses in the monocot branch of the tree have a
gene organization as shown in Fig. 1 (b). The branch containing the viruses that infect dicots
divides into two branches. The viruses in one branch (AbMV, BCTV, BGMV-PR, BGMV-BZ
and TGMV) were collected in the New World whereas those in the other branch [CLV
(ACMV), MYMV and TYCV] were collected in the Old World. BCTV is the only virus in this
cluster which is not transmitted by the whitefly, Bemisia tabaci (Bennett, 1971). Further, BCTV
has a genome composed of a 2993 nucleotide DNA, which is very similar in its gene organization
D N A 1 of BGMV and other whitefly-transmitted viruses (Fig. 1a; Stanley et al., 1986). The
other dicot-infecting viruses in Fig. 2 have two genomic DNAs and the gene organization
depicted in Fig. 1 (a) (Lazarowitz, 1987).
The normalized mutation distances (Table 2) showed that the viruses which are considered to
be strains of WDV were very similar and had mutation distances of 10 or less, as did strains of
MSV. Even the most similar of other viruses had scores at least 10-fold larger. BGMV-PR and
BGMV-BZ were not each other's closest relative; BGMV-PR was nearest to AbMV and
BGMV-BZ was nearest to TGMV.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
0
237
199
153
240
506
496
511
511
513
259
190
241
469
469
470
232
0
220
216
231
497
487
502
502
504
250
210
232
459
460
460
203
226
0
179
223
490
480
494
494
496
242
131
224
452
452
453
153
208
173
0
220
486
476
491
491
493
239
t69
221
448
449
449
242
239
239
225
0
476
466
481
481
483
194
214
176
438
439
439
498
474
482
483
477
0
358
372
372
374
495
480
477
357
358
358
508
480
471
473
460
355
0
205
205
207
485
470
467
347
348
348
527
481
490
502
456
374
208
0
5
7
500
485
482
362
362
363
527
481
490
502
459
374
205
5
0
2
500
485
482
362
362
363
529
483
491
501
461
374
205
8
2
0
502
487
484
364
364
365
245
262
245
244
205
498
483
492
495
494
0
233
158
457
458
458
196
218
131
167
225
486
480
497
497
497
216
0
215
442
442
443
226
248
221
227
169
497
504
501
504
503
158
195
0
439
440
440
485
448
445
453
415
360
345
360
364
367
467
456
460
0
8
3
481
446
442
450
413
360
344
362
366
368
466
453
459
8
0
9
481
446
442
449
413
360
343
360
363
365
466
453
461
3
9
0
AbMV BCTV BGMV-BZ BGMV-PR CLV CSMV DSV MSV-K MSV-N MSV-S MYMV T G M V TYLCV WDV-C WDV-CJI WDV-S
Matrix of mutation distances* of geminivirus replication-associated proteins
* Numbers abovethe diagonalare normalized m i n i m u m m u t a t i o n distances. Below the diagonalare distancesreconstructed from thetree with the smallestsum
of squares.
AbMV
BCTV
BGMV-BZ
BGMV-PR
CLV
CSMV
DSV
MSV-K
MSV-N
MSV-S
MYMV
TGMV
TYLCV
WDV-C
WDV-CJI
WDV-S
T a b l e 2.
t~
to
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
~
0
4-2
,,60
.......
78.6
I7
173.6
17131
l ~
1~0~.
1
~2~
~
8
70"4
880
70.0
60-6
113.6
8%6
[
66.2
Ab~V
~v~v
BCTV
- TGMV
BGMV-BZ
BGMV-PR
CLV
86.8
146.7
lO8.2,
145.9
l
29.5
34-2
179.2
BGMV-PR
BGMV-BZ
I I 28.5
TYLCV
20.4 35.5 CLV
37 ] ' ~ 7 TGMV
~ 2
31.~'81 ~
20-7 AbMV
Fig. 3, Relationship of 15 geminiviruses based on coat protein alignments. The statistically best tree is shown. Lengths of horizontal arms are proportional to the
mutation distances indicated by the numbers, Lengths of vertical lines are arbitrary.
WDV-CJI 1-0
BCTV
WDV-C 0-0
WDV-S 0)1
csMv
C--]
osv~2
67~
I 38.0
0.5
MSV-K ]~ .3
MSV-S 2 g[-- 69.8
MSV_N J
~
Fig, 2, Relationship of 16 geminiviruses based on replication-associated protein alignments. The tree with the smallest sum of squares (Felsenstein, 1984) as
calculated by the Fitch program of the Phylip package is shown. Lengths of horizontal arms are proportional to the mutation distances indicated by the numbers.
Lengths of vertical lines are arbitrary.
WDV-CJ1
0I~~
wov_~ ~.....
WDV-C 1.0
95.2
183,7
MSV-K
2.6
107.4
DSV
CSMV
6
t
MVS-S 2.
MSV-N
Z
t7
Z
,.q
o
tO
tO
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
0
387
32
47
118
413
410
414
415
416
38
111
423
423
423
377
0
378
374
381
318
315
319
320
321
372
374
328
328
328
32
377
0
38
108
404
401
405
405
407
29
101
414
414
414
49
375
38
0
105
400
397
401
402
403
33
98
410
410
410
AbMV BCTV BGMV-BZ BGMV-PR
110
381
116
105
0
407
404
408
409
410
103
64
417
417
417
392
318
394
398
402
0
214
218
218
220
398
400
285
285
285
CLV CSMV
406
308
401
400
400
215
0
139
139
141
395
397
282
282
282
413
318
413
412
410
218
139
0
4
3
400
401
287
287
287
409
317
409
409
408
218
138
4
0
6
400
402
287
287
287
DSV MSV-K MSV-N
412
318
412
411
411
218
142
3
6
0
402
403
289
289
289
40
373
28
32
105
393
401
410
407
409
0
96
408
408
408
MSV-S T G M V
104
383
108
96
64
391
397
410
408
410
97
0
410
410
410
TYLCV
Matrix ofmutation distances*forgeminiv~us coatproteins
413
333
411
411
411
295
283
283
287
283
418
411
0
2
0
412
333
410
410
412
295
284
283
287
283
417
412
2
0
2
WDV-C WDV-CJI
413
333
411
411
411
295
283
283
287
283
418
411
0
2
0
WDV-S
* Numbers above the diagonal are normalized minimum mutation distances. Below the diagonal are distances reconstructed from the tree with the smallest sum
of squares.
AbMV
BCTV
BGMV-BZ
BGMV-PR
CLV
CSMV
DSV
MSV-K
MSV-N
MSV-S
TGMV
TYLCV
WDV-C
WDV-CJI
WDV-S
T a b l e 3.
l,O
2724
A. J. HOWARTH AND G. J. VANDEMARK
Comparison o f coat proteins
Alignments of 15 coat protein amino acid sequences (the coat protein sequence of MYMV
was not available) were compared in every pairwise combination, as described above for the
replication-associated proteins. Minimum mutation distances were computed (Table 3). Three
executions of the Fitch program of the Phylip package each yielded the same best tree and sum
of squares (S.S. = 0.088). An average of 593 trees were examined per program execution. The
best tree had an average percent standard deviation of only 2-0. An F test (F = 65.6) led us to
reject the validity of the molecular clock, therefore Fig. 3 depicts an unrooted tree. The main
feature of this tree is dichotomous branching into clusters of geminiviruses that correlate with
their vectors, either whiteflies or leafhoppers. The horizontal lines of the leafhopper-transmitted
viruses were longer than those of the whitefly-transmitted viruses, showing that divergence was
greater among the leafhopper- than among the whitefly-transmitted viruses, a result also
concluded from the tree based on replication-associated proteins (Fig. 2). According to this
scheme, BCTV is nearest to viruses which are also transmitted by leafhopper vectors, but differs
from them in having a host range restricted to dicots and a gene organization like DNA 1 of Fig.
1 (a). In other regards, the tree in Fig. 3 is similar to the tree in Fig. 2. Again, BGMV-PR and
BGMV-BZ were each more closely related to another virus, TGMV, than to each other.
DISCUSSION
Serology or amino acid composition of capsid proteins are commonly used to suggest
phylogenies of plant viruses (Shukla & Ward, 1988; Fauquet et al., 1986). Serological studies of
geminiviruses have shown that whitefly-transmitted geminiviruses are related sufficiently to
cross-react with polyclonal and monoclonal antisera; however, there is a general lack of
serological cross-reactivity among leafhopper-transmitted geminiviruses (Roberts et al., 1984),
with the exceptions of MSV and DSV (Dollet et al., 1986) and BCTV and tobacco yellow dwarf
virus (Thomas & Bowyer, 1980). No antigenic cross-reactivity has been substantiated between
whitefly- and leafhopper-transmitted geminiviruses (Thomas et al., 1986). These observations,
together with the fact that coat protein is the only protein that has been purified from most
geminiviruses, limits the usefulness of a serological approach in a comparative study. We have
overcome these problems in determining phylogenetic relationships of geminiviruses by
analysis of alignments of amino acid sequences of both gene products which all geminiviruses
contain, the coat protein and the replication-associated protein (MacDowell et al., 1985 ; Stanley
et al., 1986). Thus, even the most disparate geminiviruses were compared.
The geminiviruses that infect either monocots or dicots havc different genome organizations.
First, the monocot-type DNAs have two non-coding (intcrgenic) regions, whereas dicot-type
DNAs have only one intergenic region in DNA 1. Second, the monocot-type viruses have small
(approximately 80 nucleotides) complementary-strand DNAs which contain a few covalently
bound, 5'-terminal ribonucleotides and are associated with and probably base-paired to viral
DNA inside nucleocapsids (Howell, 1984; Donson et al., 1984, 1987; Andersen et al., 1988;
Hayes et al., 1988; Lazarowitz, 1988). These small DNAs are thought to be primers of
complementary strand DNA synthesis. No such 'primer' molecules have been reported for
viruses with dicot-type DNAs. Third, monocot viruses encoded an 11K to 12K protein in a gene
(alpha in Fig. 1 b) with the same polarity as, and just upstream from, the coat protein gene
(Mullineaux et al., 1988). No comparable gene is conserved in the genomes of the dicot viruses.
Similarly, monocot viruses lack genes III and IV that are conserved in dicot viruses. Fourth,
monocot viruses have two open reading frames (ORFs) in the complementary strand (labelled
IIa a n d l I b in Fig. 1b), the products of which have significant amino acid sequence similarities
to the product of gene II of the dicot viruses (Mullineaux et al., 1985). Schalk et al. (1989) have
shown that splicing of WDV RNA results in removal of an intron and, presumably, expression
of only one protein from these two ORFs. These points suggest that although monocot and dicot
geminiviruses are grossly similar in genetic organization, close examination reveals that there
may be significant differences in their mechanisms of replication and gene expression.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
Geminivirus phylogeny
2725
The replication-associated protein phylogenetic tree (Fig. 2) has two main branches one of
which contains the viruses that infect dicots and the other those which infect monocots. These
results suggest that this tree is consistent with classification of geminiviruses based on two
constantly associated characteristics, host range limited to either monocots or dicots and type of
genome organization. Further, the results also suggest that evolution of replication-associated
proteins is influenced by viral hosts. Thus, the original divergence of monocot and dicot
geminiviruses may have been, in part, due to evolution of the replication-associated protein
genes to function eMciently in the different plant hosts. One could speculate, then, that gene
organization has become an effective determinant, but not necessarily the only determinant, of
host range. For example, the two ORFs that encode the replication-associated protein of the
monocot viruses probably require RNA splicing as a necessary step in expression of that gene
(Schalk et al., 1989). Perhaps dicots do not process these genes properly and are, thus, eliminated
as potential hosts of these viruses.
The coat protein tree (Fig. 3) is similar to the replication-associated protein tree (Fig. 2) in that
it also divides into two main branches. Unlike the replication-associated protein tree, however,
the phylogeny derived from analysis of coat proteins correlates with vector specificities of the
viruses for either leafhoppers or whiteflies. The relationships among the monocot viruses are
unchanged from the tree in Fig. 2. The whitefly-transmitted viruses have some small differences
in that BGMV-PR and BGMV-BZ switch places in their relative associations with T G M V and
AbMV. CLV and TYLCV, the only whitefly-transmitted viruses of the Old World for which
coat protein sequences were available, form their own branch of the tree apart from the whiteflytransmitted New World viruses to suggest that, as with the replication-associated protein tree,
geographical isolation has played a role in their phylogenetic history. The main difference
between the coat protein and replication-associated protein trees is that the position of BCTV
has changed from near the other viruses of dicots, all of which are whitefly-transmitted, to near
the monocot viruses, all of which are leafhopper-transmitted. This result supports the hypothesis
that coat protein is intimately associated with vector transmission (Roberts et al., 1984;
Gardiner et al., 1988), the corollary of which is that evolution of the coat protein gene is
constrained by the need to accommodate vector transmissibility. Thus, it seems logical that
phylogeny based on coat proteins would correspond to vector specificity. We favour the
replication-associated protein tree as the best representation of the phylogeny of geminiviruses
because the genome organization, structure of its replication-associated protein gene and host
range of BCTV are like those of other geminiviruses that infect dicots.
Our results may have a bearing on the issue of origins of geminiviruses. For example, Oman
(1969) suggested that BCTV originated in the New World and perhaps was transmitted by a less
aggressive insect which has since been displaced by the introduced vector, Neoaliturus tenellus
(Circulifer tenellus). Bennett (1971) argued that BCTV was introduced to North America along
with the vector. On the face of it, our results with a California isolate of BCTV support Oman's
conclusion because this isolate of BCTV lies within the cluster of New World viruses (Fig. 2).
However, it is also possible that BCTV originated in the Old World and has mutated since its
introduction to the New World, in response to new host plants, to become similar to other New
World geminiviruses. Similar to the issue of the origin of BCTV is that of CLV. Our results
suggest that CLV originated in Africa and was present there before the introduction of cassava
from South America. The lack of a South American geminivirus of cassava also supports this
idea. However, we cannot eliminate the possibility that CLV originated in South America and
was introduced to Africa with cassava.
Our results may also have a bearing on the definition of viruses and strains. We included
examples of viruses that have been regarded as three strains each of MSV and WDV. Our results
corroborate the use of the term 'strain' for these viruses. For example, MSV strains not only have
the same vector and hosts, but they have only very small mutation distances (Tables 1 and 2).
Their mutation distances were 8 or less in the replication-associated protein tree and 6 or less in
the coat protein tree, whereas the smallest mutation distance between any other pair of viruses
was more than 13-fold higher (131 for T G M V and BGMV-BZ) in the replication-associated
protein tree and four-fold higher (28 for T G M V and BGMV-BZ) in the coat protein tree. It is
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
2726
A.J.
H O W A R T H AND G . J .
VANDEMARK
interesting, but not too surprising, that the two viruses named BGMV-PR and BGMV-BZ are
not each other's closest relative. Reasons for this may include that they were collected at
distantly separated sites, Puerto Rico and Brazil; BGMV-PR was selected as a mechanically
transmissible isolate and BGMV-BZ was not; BGMV-PR was originally collected from
Phaseolus lunatus and BGMV-BZ was collected from Phaseolus vulgaris; ultimately, they were
named BGMV because they cause golden mosaic symptoms on bean and are transmitted by
whiteflies. We conclude that BGMV-PR and BGMV-BZ may not be strains of the same virus
because of their closer relationships to other viruses than to each other.
It seems clear that additional taxa should be erected to classify the two different types of
geminiviruses that are included in the group. If current nomenclature is retained then subgroups
for monocot- and dicot-infecting geminiviruses, respectively, should be defined. It would be
simple to convert this hierarchical structure to a family with two genera (Matthews, 1985), if
such a system is approved in the future. This classification would remove the problem of having
MSV as the type geminivirus (Matthews, 1982; Stanley, 1985), a status which is not consistent
with what we know about the structure and other features of these viruses. As more
geminiviruses are characterized, standardization in naming the two D N A segments is
desirable; the precedent of Stanley & Gay (1983) should be observed.
We thank D. P. Maxwell, H. Jeske and T. Frischmuth, and H. Czosnek and N. Navot for sharing sequences
prior to publication, G. D. F. Wilson for providing the MS-DOS version of the Phylip package, and J. G.
Utermohlen, K. D. Hirschi, and M. C. Hawes for comments on the manuscript. This work was supported by
USDA Hatch and C R G O (85-CRCR-l-1793) grants and this paper is no. 7043 in the journal series of the Arizona
Agricultural Experiment Station.
REFERENCES
ANDERSEN,M. T., RICHARDSON,K. A., HARBISON,S. & MORRIS,B. A. M. (1988). Nucleotide sequence of the geminivirus
Chloris striate mosaic virus. Virology 164, 443--449.
BENNETt, C. W. (1971). Curly Top Disease of Sugarbeet and Other Plants, pp. 1-81. St. Paul: American
Phytopathological Society.
BOCK, K. R. & WOODS, R. D. (1983). Etiology of African cassava mosaic disease. Plant Disease 67, 994-995.
BUCK, K. W. & cotrrrs, R. H. A. (1983). The potential of geminiviruses in the construction of plant host vector
systems. Plant Molecular Biology 2, 351-354.
COMMANDEUR,U., MATZErr, V. & GRONENBORN, B. (1987). The D N A sequence of WDV-C, a wheat dwarf virus
isolate from Czechoslovakia. In Abstracts, VII International Congress of Virology, p. 235. Ottawa: National
Research Council Canada.
DOLLET, M., ACCOTTO,G. P., LISA, V., MENLSSIER,J. & BOCCARDO,G. (1986). A geminivirus, serologically related to
maize streak virus, from Digitaria sanguinalis from Vanuatu. Journal of General Virology 67, 933-937.
DONSON,J., MORRIS-KRSINICH,B. A. M., MULL1NEAUX,P. M., BOULTON,M. I. & DAVIES,J. W. (1984). A putative primer
for second-strand D N A synthesis of maize streak virus is virion-associated. EMBO Journal 3, 3069-3073.
DONSON,J., ACCOTTO,G. P., BOULTON,M. I., MULLINEAUX,P. M. & DAVIES,J. W. (1987). The nucleotide sequence of a
geminivirus from Digitaria sanguinalis. Virology 161, 160-169.
FAUQUET, C., DEJARDIN, J. & THOUVENEL,J.-C. (1986). Evidence that the amino acid composition of the particle
proteins of plant viruses is characteristic of the virus group, lntervirology 25, 1-13.
FELSENSTEIN, J. (1984). Distance methods for inferring phylogenies: a justification. Evolution 38, 16-24.
FELSENSTEIN,J. (1985). Confidence limits on phylogenies: an approach using the bootstrap. Evolution39, 783-791.
FITCH, W. M. & MARGOLIASH,E. (1967). Construction of phylogenetic trees. Science 155, 279-284.
FRANCKI,R. I. B. (1983). Current problems in plant virus taxonomy. In A CriticalAppraisalof Viral Taxonomy, pp.
63-104. Edited by R. E. F. Matthews. Boca Raton: CRC Press.
FRISCHMUTH,T., ABOUZID,A. & JESKE,H. (1987). Comparison between nucleotide sequence of abutilon mosaic virus
and other geminiviruses by computer analysis. In Abstracts, VII International Congress of Virology, p. 233.
Ottawa: National Research Council Canada.
GARDINER, W. E., SUNTER, G., BRAND, L., ELMER, J. S., ROGERS, S. G. & BISARO, D. M. (1988). Genetic analysis of
tomato golden mosaic virus: the coat protein is not required for systemic spread or symptom development.
EMBO Journal 7, 899-904.
GILBERTSON, R. L., FARIA, J. C., MORALES,F., LEONG, S. A., MAXWELL,D. P. & AHLQUIST, P. G. (1988). Molecular
characterization of geminiviruses causing bean golden mosaic. Annual Meeting of American Phytopathological Society, San Diego, California, U.S.A.
GOODMAN,R. M. ( 1981 a). Geminiviruses. In Handbook of Plant VirusInfectionsand ComparativeDiagnosis, pp. 879910. Edited by E. Kurstak. Amsterdam: Elsevier/North-Holland.
GOODMAN, R. U. (1981 b). Geminiviruses. Journal of General Virology 54, 9-21.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01
Geminivirus phylogeny
2727
6RIMSLEY, N., HOHN, T., DAVIES, J. W. & rtorrN, B. (1987). Agrobacterium-mediated delivery of infectious maize
streak virus into maize plants. Nature, London 325, 177-179.
HAMILTON, W. D. O., STEIN, V. E., COUTTS, R. H. A. & BUCK, K. W. (1984). Complete nucleotide sequence of the
infectious cloned D N A components of tomato golden mosaic virus: potential coding regions and regulatory
sequences. EMBO Journal 3, 2197-2205.
HARRISON, B. D., BARKER,H., nOOK, K. R., GUTHRIE, E. J., MEREDITH, ~. & ATKINSON,M. (1977). Plant viruses with
circular single-stranded D N A . Nature, London 270, 760-762.
HAYES, R. J., MACDONALD,H., COUTTS,R. H. A. & BUCK, K. W. (1988). Priming of complementary D N A synthesis in
vitro by small D N A molecules tightly bound to virion D N A of wheat dwarf virus. Journal of General Virology
69, 1345-1350.
HOWARTH, X. J. & GOODMAN, R. M. (1982). Plant viruses with genomes of single-stranded D N A . Trends in
Biochemical Sciences 7, 180-182.
rlowARTH, x. J. & C-,OODMAN,R. U. (1986). Divergence and evolution of geminivirus genomes. JournalofMolecular
Evolution 23, 313-319.
HOWARTH,A. 1., CATON,J., BOSSERT,M. & GOODMAN,R. M. (1985). Nucleotide sequence of bean golden mosaic virus
and a model for gene regulation in geminiviruses. Proceedingsof the NationalAcademy of Sciences, U.S.A. 82,
3572-3576.
rIOWELL, S. H. (1984). Physical structure and genetic organisation of the genome of maize streak virus ( K e n y a n
isolate). Nucleic Acids Research 12, 7359-7375.
LAZAROWITZ,S. G. (1987). The molecular characterization of geminiviruses. Plant Molecular Biology Reporter 4,
177-192.
LAZAROWITZ,S. a. (1988). Infectivity and complete nucleotide sequence of the genome of a South African isolate of
maize streak virus. Nucleic Acids Research 16, 229-250.
MACDOWELL,S. W., MACDONALD,H., HAMILTON, W. D. O., COUTTS, R. H. A. & BUCK, K. W. (1985). The nucleotide
sequence of cloned wheat dwarf virus D N A . EMBO Journal 4, 2173-2180.
MATTHEWS, R. E. F. (1982). Classification and nomenclature of viruses. Intervirology 17, 1-199.
MATTHEWS, R. E. V. (1983). The history of viral taxonomy. In A Critical Appraisal of Viral Taxonomy, pp. 1-35.
Edited by R. E. F. Matthews. Boca Raton: C R C Press.
MATTHEWS, R. E. F. (1985). Viral taxonomy for the nonvirologist. Annual Review of Microbiology 39, 451-474.
MORINAGA,T., IKEGAMI,M. & MIURA,K. (1987). Complete gone structure of m u n g b e a n yellow mosaic virus D N A . In
Abstracts, VII International Congress of Virology, p. 258. Ottawa: National Research Council Canada.
MULLINEAUX,P. M., DONSON, J., MORRIS-KRSINICH,B. A. M., BOULTON,M. I. & DAVIES, J. W. (1984). T h e nucleotide
sequence of maize streak virus D N A . EMBO Journal 3, 3063-3068.
MULLINEAUX,P. M., DONSON, L, STANLEY,J., BOULTON,M. I., MORRIS-KRSINICH,B. A. M., MARKHAM,P. G. & DAVIES,
J. W. (1985). Computer analysis identifies sequence homologies between potential gone products of maize
streak virus a n d those of cassava latent virus and tomato golden mosaic virus. Plant Molecular Biology 5,
125-131.
MULLINEAUX,P. M., BOULTON,M. I., BOWYER,P., VAN DER VLUGT, R., MARKS,M., DONSON, J. & DAVIES, J. W. (1988).
Detection of a non-structural protein of Mr 11000 encoded by the virion D N A of maize streak virus. Plant
Molecular Biology 11, 57-66.
OMAN, P. (1969). Criteria of specificity in virus-vector relationships. In Viruses, Vectorsand Vegetation, pp. 1-22.
Edited by K. Maramorosch. N e w York: John Wiley & Sons.
ROBERTS, L M., ROBINSON,D. J. & HARRISON, B. D. (1984). Serological relationships and genome homologies among
geminiviruses. Journal of General Virology 65, 1723-1730.
SCHALK,H.-J., MATZEIT,V., SCHILLER, B., SCHELL, J. & GRONENBORN,B. (1989). W h e a t dwarf virus, a geminivirus of
graminaceous plants needs splicing for replication. EMBO Journal 8, 359-364.
SHUKLA, D. D. & WARD, C. W. (1988). A m i n o acid sequence homology of coat proteins as a basis for identification
and classification of the potyvirus group. Journal of General Virology 69, 2703-2710.
STANLEY, J. (1983). Infectivity of the cloned geminivirus genome requires sequences from both D N A s . Nature,
London 305, 643-645.
STANLEY, J. (1985). The molecular biology of geminiviruses. Advances in Virus Research 30, 139-177.
STANLEY, J. & GAY, M. R. (1983). Nucleotide sequence of cassava latent virus D N A . Nature, London 301, 260-262.
STANLEY,J., MARKHAM,P. G., CALLIS,R. J. & PINNER, M. S. (1986). The nucleotide sequence of an infectious clone of
the geminivirus beet curly top virus. EMBO Journal 5, 1761-1767.
THOMAS, J. E. & BOWYER, J. W. (1980). Properties of tobacco yellow dwarf and bean s u m m e r death viruses.
Phytopathology 70, 214-217.
THOMAS, J. E., MAX~ALSKI,P. R. & HARRISON, B. D. (1986). Production of monoclonal antibodies to African cassava
mosaic virus and differences in their reactivities with other whitefly-transmitted geminiviruses. Journal of
General Virology 67, 2739-2748.
WOOLSTON,C. J., BARKER, R., GUNN, H., EOULTON, M. I. & MULLINEAUX,P. M. (1988). Agroinfection and nucleotide
sequence of cloned wheat dwarf virus D N A . Plant Molecular Biology 11, 35-43.
(Received 31 March 1989)
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Fri, 16 Jun 2017 22:41:01