Genetic variation and microevolution of dengue 2 virus in Southeast Asia

V

aoLocv 172,

523-535 (1989)

Genetic Variation and Microevolution of Dengue 2 Virus in Southeast Asia' DENNIS W . TRENT,* , ' JOYCE A . GRANT,* THOMAS P . MONATH,* CHARLES L . MANSKE,t . FOXt MARC CORINA,t AND GEORGE F *Division of Vector Borne Viral Diseases, Center for Infectious Diseases, Centers for Disease Control, Public Health Service, U .S . Department of Health and Human Services, Fort Collins, Colorado 80522 ; and tDepartment of Biochemical and Biophysical science, University of Houston, Houston, Texas 77004 ReceivedAugust26, 1988 ; accepted June 13, 1989

Dengue 2 (DEN 2) virus strains collected from dengue hemorrhagic fever (DHF) patients and Aedes aegypti mosquitoes in Thailand, Burma, and Vietnam over a 25-year period have been analyzed by computer assisted TI-RNaseresistant oligonucleotide fingerprinting. Fifty-seven DEN 2 virus strains of the Thailand topotype were separated into four major clusters by phylogenetic analysis of 97 unique oligonucleotides identified in a common well-resolved region of the fingerprints . Similarities in the 57 fingerprints indicated that DEN 2 virus of a single, continually evolving genetic population has been involved in endemic transmission of the disease . Virus isolates from DHF cases and mosquitoes are genetically very similar, indicating that different genetic topotypes are not selectively the cause of severe DEN disease in Thailand . Microevolution of the DEN 2 virus genome from 1962-1986 was gradual with detectable changes in the pattern of oligonucleotides through time . Segregation of the DEN 2 virus fingerprints into the three decades (I 960s, 1970s, and 1980s) revealed the rate of genetic change to be one consensus oligonucleotide per year. Based on average association coefficient (Sae) values between the consensus fingerprints for each decade, the similarity between the consensus fingerprints decreased by 1 .4% per year . Genetic variation during each of the three decades was found to be essentially the same (0 .866 ± 0 .053) . Constancy in the microevolutionary rate and genetic variability suggests that a balance of genetic drift and natural selection acting on the viral population did not significantly change throughout the 25-year period . c 1969 Academic Press, Inc . INTRODUCTION

from epidemiologic and experimental studies indicate that immunopathologic mechanisms may be involved, with a higher risk of severe disease in infected persons with a history of prior infection with a heterologous DEN serotype (Halstead et a)., 1973 ; Sangkawibha et al., 1984 ; Burke et al., 1988 : Halstead, 1988) Other studies suggest that virulence differences between DEN virus strains may be responsible for the variability observed in disease expression (Barnes and Rosen, 1974 ; Rosen, 1977) . These two hypotheses, however, are not mutually exclusive (Burke at a/., 1988) . Movement of DEN viruses between different geographic areas is an important element in the epidemiol ogy of the disease- Outbreaks have frequently followed introduction of virus by viremic human beings ; these occurrences have been enhanced by the advent of air travel (Halstead, 1980, 1984) . Molecular characterization of DEN virus isolates can be used to define genetic variation between strains of the same serotype and follow geographic movement of DEN strains facilitating identification of the source of virus strains in new outbreaks (Vezza et al., 1980 ; 1 rent et al, 1983, 1989 ; Monath et al ., 1986 ; Repik et al ., 1983 : Walker et al ., 1988) . All four serotypes of DEN virus are actively transmitted in Thailand where major outbreaks have occurred

Dengue hemorrhagic fever (DHF), first recognized in 1954, is a severe disease caused by all four dengue (DEN) virus serotypes and characterized by bleeding diathesis, thrombocytopenia, hemoconcentration, and, in a small proportion of cases, circulatory shock [dengue shock syndrome (DSS)} (Quintos et al ., 1954) . DHF was first described in Thailand during 1958 when an epidemic occurred in Bangkok (Gunakasem et al ., 1981) . Between 1958 and 1971, DHF was endemoepidemic in Thailand, with a mean annual incidence of 4463 cases and 158 deaths (WHO, 1986a,b) . An increase in epidemic activity has occurred since 1972, with major outbreaks occurring at 3-4 year intervals (WHO, 1986a,b) . In the 1980s, the average annual incidence of DHF in Thailand exceeded 30,000 cases . DHF also occurs as an epidemic disease of major public health importance in Indonesia, Burma, and Vietnam (WHO, 1986a,b) . Host- and virus-specified factors which underlie the pathogenesis of DHF/DSS remain uncertain . Evidence 'The views, opinions, and/or findings contained in this report are those of the authors and should not be construed as an official Department of the Army position, policy, or decision unless designated so by other documertlulion . ~ To whom requests for reprints should be addressed . 523

0042-6822/89 $3 .00 CUpY~ lybl 6£ 1989 by Academic Press, Inc . Au dams of eI,rodecton in any Corm reserved .



524

TRENT ET AL .

annually since 1958 (Gunakasem et al., 1981 ; Sangkawibha eta/., 1984 ; Halstead, 1988) . DEN 2 virus has been consistently isolated during this period and until the early 1980s was the serotype most frequently associated with DHF/DSS (Sangkawibha eta!, 1984) . In 1984 the incidence of DHF/DSS cases reached the highest incidence during 27 years of DEN epidemics in Thailand (WHO, 1986a,b) . Oligonucleotide fingerprint and antigenic signature analysis of DEN 2 virus strains from Burma and Thailand suggested there are genetic and antigenic differences among DEN 2 virus strains circulating in Southeast Asia (Walker eta!, 1988 ; Trent et al., 1983 ; Monath et a!, 1986) . To understand the evolution and origin of DEN-2 virus strains involved in DHF/DSS disease in an endemic setting, we have examined the genetic variation of Thailand topotype DEN 2 viruses isolated from 57 DHF/DSS patients and mosquitoes in Southeast Asia over a 25-year period . MATERIALS AND METHODS Virus strains The DEN 2 virus strains analyzed in this study were obtained from the Armed Forces Research Institute of Medical Science, Bangkok, Thailand, and the Pacific Biomedical Research Center, Honolulu, Hawaii, The origin, strain designation, and year of virus isolation are presented in Table 1 . Clinical information relative to grade of disease severity and serologic response to flavivirus antigens are also given when data is available . Each isolate was passaged twice in C6/36 Aedes albopictus cell cultures prior to virus propagation for RNA oligonucleotide fingerprinting . All virus strains were reidentified by indirect immunofluorescence using DEN serotype-specific monoclonal antibodies (Henchal et al., 1983) . Virus propagation and purification C6/36 cells were grown at 28° in Dulbecco's modified minimal essential media (MEM) containing 10% heat-inactivated fetal calf serum . Ten 150-cc cell culture flasks were infected with each virus isolate at a multiplicity of 0 .01 to 0 .1 PFU per cell . Virus was allowed to adsorb for 1 .5 hr and MEM continuing 5% calf serum was added . Tissue culture medium containing the virus was removed after 5 days incubation at 28° and DEN virus purified by polyethylene glycol precipitation, and rate-zonal and isopycnic ultracentrifugation (Trent and Grant, 1980) . RNaseT1 oligonucleotide fingerprint preparation Extraction of the RNA from purified virions and endlabeling of the RNase Tt-resistant oligonucleotides

were as described previously (Trent et al., 1983) . Twodimensional polyacrylamide gel electrophoresis of the 32 P-labeled RNase T1-resistant oligonucleotides was performed according to DeWachter and Fiers (1972) . Computer-assisted fingerprint image analysis Autoradiograms of DEN viral fingerprints were electronically captured and digitized using an Eyecom III (Spatial Data Systems, Goleta, CA) image analysis and acquisition system with a PDP 11/73 (Digital Equipment Corporation, Maynard, MA customized by Plessey Peripheral Systems, Irvine, CA) serving as the host computer . Spot boundaries were identified by a second derivative algorithm (Lemkin and Lipkin, 1983) and images were edited to remove artifacts and spots outside the analysis region using Fortran IV programs written for the PDP 11/73 . These files were then transferredto a Microvax II minicomputer (Digital Equipment Corporation, Maynard, MA) for further processing . Electrophoretic mobilities of the oligonucleotides varied during individual runs, making it difficult to judge whether spots in different images are actually the same . Variations were compensated by transforming the X, Ycoordinates for the spots in each image to one common coordinate system, that of strain D79-069 . This strain is similar to the original representative selected for the Burma topotype (Trent et al., 1983) but is more representative of the virus collection used for this study . First, 10 to 15 likely matching spots ("register points") between each image and D79-069 were visually identified . At least 10 register points were identified in the digitized image of each fingerprint by computer analysis, and these points were used to transform each image to the D79-069 coordinate system by 3rd order polynomial spatial warping (Hall, 1979) . The best transformation was visually selected and used to identify matching spots between each image and D79-069 . Since all spots were brought into a single coordinate system, it was possible to identify matches among spots not contained in D79-069 . From this information, a database of 97 oligonucleotides containing the positions of all well-resolved oligonucleotides in all of the fingerprint images was constructed . Since the computer programs were developed concurrently with the fingerprint analysis, all database assignments were confirmed by visual comparisons . Construction of phylogenetic trees Patterns of genetic relatedness among the viruses was analyzed by constructing phylogenetic trees based on similarity measured for all possible pair-wise comparisons . The similarity between any two strains was measured according to an association coefficient



525

GENETIC VARIATION AND MICROEVOLUTION OF DEN 2

(Fox and Stackebrandt, 1987) used in the analysis of 16 S ribosomal RNA fingerprints . The association coefficient S ab is defined as Sab = 2Nab/(Na + Nb),

(1)

where N a is the total number of oligonucleotides in the analysis area of image a, Nb is the total number of oligonucleotides in the analysis area of image b, and Nab is the number of matching oligonucleotides shared by both images . We suggest using "percentage similarity" ratherthan "homology" for the percentage version of this similarity measure (Reeck et al ., 1987) . A matrix of Sab values for all possible comparisons among the images was assembled and used to construct phylogenetic trees according to the unweighted pair-group method using arithmetic averages (UPGMA ; Sheath and Sokal, 1973) and the reference ratio method (RRM ; Manske and Chapman, 1987) . According to UPGMA, the distance to a branch point is simply the average S ab of all viruses joined at that branch point . This method is most accurate when the rate of genetic change is constant (Wilson et al., 1977) ; however, this condition is seldom met with molecular data (Manske and Chapman, 1987) . Nonuniformities in evolutionary rates in different branches of a tree may be more accurately estimated by using the reference ratio method in which the matrix of S ab similarities is convened into a matrix of genetic distance values, D ab = 1 - S ab . In this way the difference values are directly related to the length of the branches in the dendrogram . Nonuniforrities in evolutionary rates within different branches of a tree can be detected from the ratio of the distance of each member, i, of a set of viruses, S, to an outside "reference" group of viruses, R, to produce a reference ratio, D;R/DsR . Values less than one indicate evolutionary rates less than the average rate while ratios greater than one identify viruses evolving faster than the average rate . The nonuniformities in a raw Dab distance matrix were normalized by dividing each distance value, D,,, by the reference ratios for i and j : D'„=Dr,(Dsp)'1D,RD;R .

(2)

Each FIRM tree was produced by applying the UPGMA clustering algorithm to a distance matrix modified according to equation 2 . In the resulting tree (e .g ., Fig . 5), the distance from each bifurcation in the tree to the tips of the descendants from that bifurcation is half the average distance (D ab ) between the two descending strains or group of strains . For consistency with the style of dendrograms reported using the S ab method (Fox and Stackebrandt, 1987), the horizontal scale in Fig . 6 is labeled so that each bifurcation is the Sar , of the descendants .

Both UPGMA and RRM assign equal amounts of "genetic distance" when calculating the branch lengths descending from each bifurcation in a tree ; in other words, all terminal branches end at the same level or "extinction time" as shown in Fig . 5 . This assumption is not true, however, if the amount of evolution expected to occur between viruses with the earliest and most recent isolation dates is significant compared to the total amount of evolution measured for the deepest branches of a tree . Therefore, a method was developed to build dendrograms that better illustrate the evolutionary history of DEN 2 viruses . For each strain i, an amount of genetic change, C,, expected to occur if the strain survived to the time of the most recent strain (1986) was estimated by multiplying the estimated average rate of genetic change (0 .018 S ab per year, explained under Results) by the quantity (1986 year of isolation of strain r) . These values were then added to the measured amounts of genetic change in the original distance matrix : =D,;+C;+C .

(3)

This modified difference matrix, which contains genetic distances as if all the strains had survived to 1986, was then processed exactly as described above to produce UPGMA and RRM trees . In the final tree (Fig . 6), however, each C, value was removed from the line segments leading to their respective locations in the tree, so the branch end points are staggered to show the different time of isolation . To summarize, the addition of expected amounts of evolution allows the assignment of equal distances to both descendants from each bifurcation in the tree, and removal of these distances from the final tree restores the ''actual" (measured) distances and places terminal branch points according to their relative times of "extinction ." This method is easily demonstrated to produce the correct tree topology and branch distances if the measured genetic distances between each pair of viruses are proportional to the actual genetic distances, and the rate of genetic change is constant through time . Because of the reality of nonuniform rates of evolution in real genetic systems such as the DEN 2 viruses, RRM was included in the isolation date algorithm . RESULTS

The RNase T1 oligonucleotide fingerprint images of all DEN 2 virus isolates were initially analyzed by visual comparison of the autoradiograms, resulting in the selection of a representative genotype strain designated D79-069 . Direct comparison of this fingerprint with the original 1976 Burma/Thailand topotype (Trent et al., 1983) revealed that 83%b of the 38 large oligonucleo-

5 26

TRENT ET AL . TABLE I CHARACTERIZATION OF DENGUE

2

VIRUS STRAINS ISOLATED IN SOUTHEAST ASIA

Strain designation

Geographic location

Source

BKM 59-62 BKM418-62 SKM309-62 5991-62 1249-62 5029-62 2358-62 BKM68-63 BKM68/63 3227/63 1635-63 14256-64 TC 16677-64 TC 15460 TC 14914 TC 14849-64 TC 14543-64 TC 16681-64 23753-66 BKM551/66 53428-73 53544-73 53700-73 1143-73 1388/74 S-16803 D79-062 S-40916 S-40921 1413-76 D76-152 D76-008 D77-191 2780-77 D78-025 D79-069 D79-175 D80-100 D80-317 D80-084 D80-401 DB1-081 D82-010 D82-033 D83-066 D83-076 D84-015 D84-181 D84-074 D84-087 D84-237 D85-044 D85-055 D85-76 D85-249 D86-004

Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Vietnam Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Burma Burma Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand Thailand

Aedesaegypti Aedesaegypri Asides aegypri Human Human Human Human Aedesaegypri Aedesaegypti Human Human Human Human Human Human Human Human Human Aedesaegypri Aedesaegypti Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human Human

illness/grade

Seroresponse

Year 1962 1962 1962

DHF°/NA° DHF/NA DHF/NA DHF/NA

DHF/NA DHF/NA DHF/NA DHF/NA DHF/NA DHF/NA DHF/NA DHF/NA DHF/NA DHF/NA DHF/grill DH F/grIV DHF/gr III DHF/grIII DHF/gr1 DHF/NA DHF/grII DF/NA DF/NA DHF/grIV DHF/grIII DHF/gr IV DHF/grIII DF/NA DHF/grIII DHF/gr11 DHF/grII DHF/grII DHF/gr III DHF/gr11 DHF/grll DHF/grIII DHF/grll DHF/gr1 DHF/gr111 DHF/grIII DHF/gr II DHF/grIII DHF/gr III DHF/gr IV DHF/grI DHF/gr III DHF/gr IV DHF/grII DHF/grlI DHF/gr 1111

° The seventy of DHF is classified into four grades (I-IV) according to clinical and laboratory findings (WHO, 1 986c) . ° Clinical data and seroresponse not available . ` Determined by hemagglutination inhibition testing ; 1', primary ; 2°, secondary (WHO, 1986c) .

2 2° 2° 2° 2° 2°

2° 2° 2° 2° 2° 2° 2° 2° 1° 2° 1° 2° 2° 2° 1° 2° 2° 2° 2° 2° 2° 1° 2° 2° 2° 2° 2°

1962 1962 1962 1962 1963 1963 1963 1963 1964 1964 1964 1964 1964 1964 1964 1966 1966 1973 1973 1973 1973 1974 1974 1975 1975 1976 1976 1976 1976 1977 1977 1978 1979 1979 1980 1980 1980 1980 1981 1982 1982 1983 1983 1984 1984 1984 1984 1984 1985 1985 1985 1985 1986



GENETIC VARIATION AND MICROEVOLUTION OF DEN 2

527

121

Qct o C) 00 ~o o

C77

6

®O

~ ~ O Q f 76 O

Q ®

6 7s'92,'90

77,

ii 10® 89 7P 25 ® 93 6' III® jP®2431 29 04 2(D E)9 ®5 •96 30®43 &4 94567 66 649 •41 9 ® 4'®® 3 997 65® 3 3 21' _- 13 , qg 60%39 6, E)62 _ .Wa 53 I 1 2• 38 ® ®5: ~4 16 7® 19 ®7 ® ®59 ® 58 I8 9 • 52 955 37 ED FIG . 1 . A Composite RNase T1 oligonucleotide fingerprint map of Thailand DEN 2 viruses isolated from 1962-1986 . (0) oligonucleotides in fingerprint of strain D79-069 and (®) oligonucleotides in fingerprints of the other 56 strains .

tides in D79-069 are present in the Thailand/Burma isolate . Fifty-seven fingerprints from the strains listed in Table 1 were digitized and subjected to a detailed computer-assisted analysis . All the digitized images were transformed to a single frame of reference (i .e ., that of D79-069) which allowed the intercomparison of spots between any binary pair of images . The analysis area selected encompasses a region containing 38 spots in D79-069, as illustrated by the solid-filled spots in Fig . 1 . The analysis was restricted to this region because it was fully resolved on all 57 fingerprints and contained a minimal number of large unresolved spots . The major advantage of the computer-assisted approach is that it facilitates the tracking of any spot that occurred in the analysis area in any image throughout the entire data set . A total of 97 such unique large oligonucleotides were identified in the analysis area for the entire data set ; their locations are identified in Fig . 1 . The number of large oligo-

nucleotides in individual isolates was remarkably constant ; 39 ± 2 .7 oligonucleotides per image . This is highly suggestive of both constancy of size and base composition of the RNA throughout the 25-year period . Relatively conserved oligonucleotides were identified and used to construct "consensus fingerprints" which is a hypothetical construct consisting of all oligonucleotides ('consensus oligonucleotides") that occur in the majority (>50%) of the set of images under consideration . Figure 2 shows the consensus fingerprints for the 57 strains analyzed and the three subsets divided according to the consensus fingerprint pattern through time . The complete database of 97 oligonucleotides shown in Fig . 3 is arranged according to the phylogenetic analyses discussed below and presents the consensus oligonucleotides for the four major phylogenetic clusters . Including all oligonucleotides in the analysis made possible calculation of the association coefficient, 6 a1 .



528

TRENT ET AL .

5 Gained - 5 Lost

0

70s

0 x

•

• • •

lam x • •%B 2 •

73 0

27 ,4z

•

0 68

• • : •• • •• 4

0 95 96

•

5 Gained 6 Lost

7 3 76 • •

6•

77 • '

027

7• 6•

60s

°• 3•2•

1 22 • w •4 a i Sz 44 •49 '6 2i •19 • 40 •4 9 36

•

•69

42

8

7

90

93• •9 5

s2

•5 2

3 . 687

B

All Strains

7•

26•

60 12

22

3

sfi 69 0

•

2

15

70 * 9 3 9p

1

• •,

•

21 s . 36

•49 •52

6 2

FiG . 2 . Consensus fingerprints (containing oligonucleotide spots in at least 50% of strains) of DEN 2 strains . Spot numbers in the 1 970s and 1 980s that did not change relative to the 1 960s pattern are omitted for clarity.

(equation 1), for each pair of images and application of cluster analysis procedures to the data . Phylogenetic trees obtained by using the UPGMA (Sneath and Sokal, 1973) revealed clearly isolated subgroups . The effect of nonuniformities of evolutionary rates within subgroups was reduced using the RRM to yield better estimates of the true branching order (topology) and branch distances as compared with UPGMA alone (Manske and Chapman, 1987) . In the UPGMA tree of all 57 viruses, strains 1635-63 and 3227-63 clearly diverged at the deepest bifurcation (Fig . 5) . These two strains were therefore ideal ''reference strains" for construction of the RRM dendrogram containing the remaining 55 virus strains . The fine structure in dendrograms presented in Figs . 5 and 6 can be affected by changing the interpretation regarding matches of as few as two or three spots . The most trustworthy inferences can therefore be made where there is greater difference between the S ass of adjacent branches or major clusters . The DEN 2 virus fingerprints were divided into four genetically related clusters, identified in Figs . 3 and 5, in order to estimate genetic variation among the isolates . Cluster I contains 14 isolates obtained between 1962 and 1966 and one isolate from 1978 (Fig . 5) . A representative isolate can be selected for a cluster by

calculating Sab values for all members as compared to the consensus oligonucleotides for a cluster . Using this approach, the most representative isolate of cluster I was strain 5991-62, which has an Sab of 0 .94 with the consensus oligonucleotides of this cluster . Cluster II contains 14 isolates obtained between 1962 and 1977 . It is best represented by isolate TC16677-64, which has an Sab of 0 .94 to the consensus oligonucleotides of this cluster . Cluster II has an average of 40 .2 ± 2 .5 consensus oligonucleotides in the fingerprint (Fig . 5) . Cluster III contains five viruses ; two Burma isolates and two distantly related Thai isolates, 1388-74 and Thai isolate D84-181 . Thai strain S-16803 exactly matches the consensus fingerprint for cluster III, which has an average of 39 .9 ± 5 .2 oligonucleotides per image . Cluster IV contains 19 isolates including most of those collected in the 1980s . Strain D69-175 is the best representative of this cluster, with an S ab of 0 .93, to the consensus oligonucleotides . The transformation reference strain, D79-069, is related to strain D81-081 as the second closest isolate (S ab to the consensus is 0 .935) . The average number of oligonucleotides per image for cluster IV is 37 .8 ± 1 .9 . In addition to strains 1635-63 and 3227/63, the strains used for the reference ratio method, six other isolates did not cluster in an obvious way with the four major clusters (Fig . 5) . These strains were not included in the within-cluster analyses presented above but were included in determining the consensus fingerprint of all 57 isolates and in the decade-specific analyses . Examination of the distribution of isolation dates in the dendrogram establishes a time-dependent variation in the consensus fingerprint of the Thailand topotype over the 25-year period . To explore the evolution of the viral genome through time, the spot database was divided into three groups according to the decades in which the viral strains were isolated . The 1960s group contained 21 strains with an average date of isolation, weighted according to the number of strains per year, of 1963 .3 . There are 17 strains in the 1970s group with an average isolation date of 1975 .5 . The 1980s group included 19 strains with an average isolation date of 1983 . A consensus fingerprint was constructed to represent viruses from each decade (Fig . 2) . Examination of the changes in the consensus fingerprints revealed that the 1 960s consensus fingerprint changed at a remarkably uniform rate (Fig . 2) . If an evolutionary ''event" is counted as the loss or gain of one consensus oligonucleotide from one decade to the next, linear regression analysis of the average isolation dates compared to the number of events yields a constant evolutionary rate of 1 .0 event per year with a correlation coefficient of 0 .997 . Correlation of the S an





529

GENETIC VARIATION AND MICROEVOLUTION OF DEN 2 Spot Number :

5 6 7 8 9 2 3 4 1 123456789D1234567890123456789012345678901234567890123456789012345678901234567890123 56789012 4

Strain - D79-069 D80-084 D82-010 D80-100 D86-004 D85-076 D79-175 23753-66 D81-081 14 D76-008 DOD-317 D84-237 D85-044 D85-055 D83-076 D84-087 D84-015 D77-191 D82-033 Consensus

345

67

Year (79) (80) (82) (80) (86) (85) (79) (66) (81) (76) (80) (84) (85) (85) (83) (84) (84) (77) (82)

- S-16803 (74) S-4D92) (76) III S-40916 (76) 1388/74 (74) D84-181 (84) -- Gonsensus

.'. o •

•

u

.

•

D84-074 D83-066

(84) (83)

• . .' '. . .'. .`. .' ' . .'

D85-249

(85)

.S

080-401

(80)

, ..

L '.

// . • . '

.I . . .' 1 I . . .' .i . ., .' ' . .,A ' ~._ .~ .~.~ ' . . • 0 .1 .1 . . .1 .-I a . .i .i . . .n 0 .1 ~. .~ . .~ .~.~.~ .N . . .U .U. .t ./ .

- D76-152 (76) 1413-76 (76) 53700-73 (73) D75-062 (75) 17 (74) 7016677-64(64) TG14543-64(64) BRM328-63 (63) BKH309 .62 (62) 53544-73 (73) - Consensus

. . . . . . . . .. . .. . . . . . I I

.

1143-74

l'i ∎ ∎ . .∎. . .`/ ∎ .∎ 1.0

53428-73 (73)

2780-77

∎ . .'

.. . .

, . .~~ '.1 . .

∎. ., .∎ -. V './ ∎ . .,.∎ ' ' .∎.' . .,.' . .

(77)

- 3016681-64(64) TC14914 (64) TG15460 (64) 5991-62 (62) TC14849-64(64) 14256-64 (64) 8815159-62 (62) S M68-63 (63) 814 168/63 (63) BKM551/66 (66) 88151418-52 (62) 2358-62 (62) 1249-62 (62) 5029-62 (62) D78-025 (78) - Don .vensss

. .. . . . . . . . . .. . . . . . . . . . .. . . . . . . . .. . . . . . . . .

.

1635-63 3227/63

(63) (63)

. .ti

. . .i. .r . . . .a

.. i .11 . . . . .rtr0"J .w 1 ."Va.I'll

. .1. i.,

Consensus for e71 stains 1 2 4 3 5 6 8 9 1234567890123456789012345fi78901234567890123456789012345678901234567890123456789012345678901234567 11

FIG . 3 . Oligonucleotide occurrence database for 57 DEN 2 viruses arranged according to genetic relationships . Solid boxes indicate presence of oligonucleotide spots in the analysis area of a fingerprint . M indicates uncertainty in recognizing a spot-each was counted as one half for the average spot number and spot frequency calculations, and they were considered neither present nor absent, but not included in the total numbers (N. or N4 in equation 1) during S et calculations . Consensus oligonuoleotides are identified for clusters I N, identified by the phylogenetic analysis presented in Fig . 5 . and for all strains .



630

TRENT ET AL .

•

230

•

[V

®

5

•0

06, 69

129

• • •

749 730 750

0

•

••0®0% 32 35

••

9 96

063

948 O • 51

i 721 1

• •

M

258028

73

689

93

: 120 : ®44 3•••• 9%5%

•

9 96

7 962

73 4®

•

•

• • •6

ss •

68 • ®

93® 095

044

•

•

5

• •• e % •4 •3 5 • • •

1 1s

662

were isolated at different times and not at the same contemporary time as is usual for molecular phylogenetic analyses . Using the 1 .4% rate estimated as an average rate of genetic change, another evolutionary tree (Fig . 6) was constructed using a simple algorithm that compensates for different isolation dates . The major limitation of this method is the lack of resolution in the dates of isolation, which is accurate only to the nearest year . The clusters identified in Fig . 5 are mostly retained in the isolation date tree in Fig . 6, although the divisions between clusters II, III, and IV are much less distinct in Fig . 6 . Besides some shuffling in branching patterns within each cluster, strains 23753-66, D77191, 1388/74, and D85-249 showed the greatest change in location as compared with Fig . 5 . The gradual evolution of the DEN 2 Thailand topotype through time is clearly evident in the dendrogram shown in Fig . 6 . DISCUSSION

7• t

43

26&27

699

21 ••

j

• •

2

22, 3•9•4 2 =46 32 9 15®20"O 19 •3 6 •52

FIG . 4. Consensus fingerprints of DEN 2 viruses divided into clusters I to IV according to genetic relationships . (•) Consensus oligonucleotides present in all four clusters, ((D) present in three out of four, (6) present in two out of four, and (0) present in just one of the four clusters . The latter may be considered "characteristic" spots within each cluster . Spots missing in one cluster, but present in the other three also characterize a cluster . For example, spots 12, 35 . 68, 73, 74, and 90 are missing just from cluster I . Spot numbers in clusters II, III, and IV that are also in cluster I are omitted for clarity .

values between the 1 960s consensus fingerprint and the other two decades yields an evolutionary rate of -0 .014 Sob units per year with a correlation coefficient of 0 .994 . Thus, the similarity between consensus fingerprints decreases 1 .4% per year. This change is accompanied by an inherent variability in the virus population which is essentially constant over the 25-year period . This was measured by calculating the Sab for each strain to the consensus oligonucleotides for a particular decade ; the average Sab s are 0 .861/0 .053 (1 960s), 0 .861/0 .058 (1970s), and 0 .875/0 .048 (1980s) . Separate UPGMA and RRM dendrograms were constructed for each of the three decades and also for all possible combinations of two decades . With very few exceptions, individual strains separated into the same four clusters as those shown in Fig . 5 . Construction of a dendrogram for rapidly evolving RNA viruses which reflects the evolutionary history of the strains is complicated because individual strains

Analysis of the genetic variation among DEN 2 viruses isolated in Thailand, Burma, and Vietnam by oligonucleotide fingerprinting extended previously reported results demonstrating a close relationship between the Burma and Thailand DEN 2 viruses isolated between 1962 and 1985 (Trent et al ., 1983) . Repik at al., (1983) previously demonstrated an apparent evolution of the DEN 1 virus genome over time ; however, the viruses studied came from different geographic regions . The fingerprints of DEN 2 viruses from Southeast Asia demonstrated slow, continual, genetic changes in the RNA genorne over a period of 25 years . Similar genetic evolution of the flavivirus, St . Louis encephalitis, has been previously demonstrated (Trent et al., 1981) . The pattern of genetic change illustrated in Fig . 6 suggests that the genotypes present in the 1960s, forming primarily clusters I and II, underwent genetic change leading to the dominance of genotypes typical in clusters III and IV by the 1980s . These changes may have been caused by directional selection, random genetic drift, or a combination of both . Fingerprints prepared from viruses before and after 22 serial passages in mosquito cell cultures were identical indicating that mosquito cell culture passage did not select forvariants in the population (Trent, unpublished results) . The collection of viruses used in this study are unique because of the 25-year range in isolation dates . Comparison of the dendograms in Figs . 5 and 6 reveals the effect of the reference ratio method (RRM) on constructing dendrograms . By this technique, the earlier isolates were treated as if they had slow evolutionary rates because the genetic distances to the reference



GENETIC VARIATION AND MICROEVOLJTION OF DEN 2

0 .63

0 .70

D .75

0.00

0 .85

0 .95

0 .90

0* 06101

strains, 1635-63 and 3227-63, were less than the average distance in all strains to the reference strains . Ge-

Year

079-069

(79) -

D00-084

(80)

082-5D

(52)

netic distances among the earlier isolates within cluster I, are exaggerated as shown in Fig . 5 . This, how-

960-100

(ew

ever, does not invalidate the usefulness of RRM since

086 004

(86)

- D83-076

(83)

the topology of the RRM tree is more accurate than that produced by the unmodified UPGMA tree . The dendro-

079-175

(79)

gram in Fig . 6 more accurately shows the branch dis-

23753-65 (667 000 001

(81)

076-0 D8

0767

900117

0097

tances among earlier isolates and unequivocally would be considered the "best'' tree if not for the limitation in resolution of the isolation dates . This tree better repre-

08423

I

sents the pattern of DEN 2 virus evolution because ter-

005 044

(05)

D83 055

0050

905976

(85)

084-007

(84)

D&-015

(84)

077 191

(77)

D82.033

(821 -

5- 603

(74) -

1D9 -4

(700

5-40076

(767

1388774

(740

Du-191

(947

D84-074

184)

D83 066

(837

minal branches end at different locations on the horizontal scale, and reflect the time of isolation or "extinction'' relative to the strains that were isolated most

Day

I I

J

a

08040'

160)

076-152

(76) -

1413-76

(76)

53700-73 (733

I

075 .62

(75)

1143-74

(74)

rc+mn-6f<6)

I

1014543-64(6.7 a

e

10+309 6E (62) 53544-73 (733 53428-77 (73)

278D 77

(017

156681-64(64) 04 C1491

(64)

1015469

(64)

599162

002)

101484964(64) 14256-64 (64)

F-

I

0059-62 (62)

recently . The comparison of 57 strains in our study illustrates the ''quasispecies" nature of viruses (Eigen and Schuster, 1977 ; Reanney, 1982) . At any given time

111



531

(e .g ., a vertical section through the dendrogram in Fig . 6), if all members of a group of isolates are fingerprinted, one finds that each yields a somewhat different fingerprint . Thus, at any evolutionary time the genome that defines the virus "species'' is in reality a population of more or less similar genomes . For various reasons, descendants of some portions of the population will be more greatly represented in subsequent populations and as a consequence the "typical'' genome will slowly change through time . In addition, the population at any time contains minor components that are descendants from other components of the earlier population that have survived but not flourished . To characterize this process, two types of information are specially relevant . These are the population diversity at any one time and the rate of population change overtime . If these two diversity measures remain constant, as they appear to be for the DEN 2 strains in our study, major changes in selective forces have not occurred . Therefore, a group of strains is best represented by a consensus fingerprint rather than by any individual fingerprint . The number of consensus oligonucleotides is 39 for the 1960s, 38 for the 1 970s, and 38 for the 1 980s groups (Fig . 2), and is 39 for each of clusters I-IV identified by phylogenetic analysis (Figs . 4 and 5 .) This is es-

68068-63 (637 0070/63 {63) 001551765 166)

I

004416-62 002)

I

i 0 .65

0 .70

0 .8.

1 9 .35

` .90

I 095

Sa7

25)0 62

(62)

1249-62

(62)

5029-62

(627

970-025

(78)

163563

(67)

3227763

4637

Fic . 5 . Dendrogram of genetic relationships among Thailand DEN 2 viruses . The San of the deepest branch, from strains 1635-63 and 3227/63 to the others, is the average S,, of these strains to the others, the deepest branch as determined by the UPGMA tree-building method . Branching order and distances to the upper 55 strains were calculated using the reference ratio algorithm /RRM) ; strains 163563 and 3227/63 were the reference strains . The horizontal scale is labeled so that each bifurcation !vertical line in the tree) is located at the average S,,, of the descendant branches-





9 •S

56 0

06'0

59'0

5®'0

02 0

02 0

59 0

09'0

55'0

as-0

157) 59-622£

0)_a

I

(58) £9-5£94

•

0

(59) 59/09900 (59) £9-99940 (29) (9. 651109 (99) 99-9291

(99) 99.6999u+ (29) 29-Lf6S (9) 0995071 (99) 9069171 (99) 99- (OWL)) (29) 29-6205 (29) 29692(

I

129) 29-MR (29) 29-91997.9 (02) 52(920 -'

• •

.

I

(99) 99/05900

• - • • . . .

(R) R-995££ (99)99-5)59471 (99) 992299171 (58) 29105900 (£9) 59-925100 (90 £1-D2nS

(50) 612100 (2L) 22-0012 (92) 91/8951 (92) 92-£941 (5L) 29-510 (5L) u-052£5 (9L) 921191 -' ' ' ' ' ' (9L) 250920

(00) 409-090 42) L64-LLO (99) 99-55LR (A7 (e1-905 (91) 94601-5 ( 9(1 2609-.

L.

( 91) 50991-s (00) 990-580 (98) 9(0909

(29) 550-200 (91) 000-900

I

(00) L45-W0 ((0) 580400 (6L) 514-620 (90) 540-989 v0- 009400 (50) 9L0-580 (50) 550-580 (58) M(-WO (90) 252990 (58) 920-580 (00) WL-WO (26) 510-290 (09) 958-000 (90) 900-00 -' ( 6U 699-610



9R

56'0

06'0

50'0

58'0

SL'0

OL'0

lV 13 1N3i1

I

59-0

I

09 . 0

SS'a

OS'O

59'0

09'0

Z£9

GENETIC VARIATION AND MICROEVOLUTION OF DEN 2

sentiafly the same value, 39/2 .6, as the average number within the analysis area of all 57 fingerprints analyzed . We do not ascribe any particular taxonomic importance to the four clusters (I-IV) or to specific relationships among closely associated viruses because of the limitations and uncertainties of tree-building techniques . The overall pattern of evolution, especially the gradual shift in viral genotypes through time, is the essential result . The demarcation of a tree into subgroups or clusters is useful primarily for analyses, which do not depend on detailed dendrogram structure . Another consequence of studying a large number of strains is a problem common to many taxonomic investigations-how to classify individuals into useful taxonomic units . The term "topotype" was proposed by Trent etal. (1981) to describe a quasispecies collection of genotypes from a particular geographic area, as revealed by the classification of oligonucleotide fingerprints of St . Louis encephalitis viruses from three geographic regions in the United States . For DEN 2 viruses, the topotype concept has been a practical taxonomic unit because, with few exceptions, collections of viruses isolated from different geographic regions show obvious oligonucleotide fingerprint similarities within a region and quite distinct fingerprint patterns when comparing different regions (Trent et al., 1989) . The fingerprinting approach lends itself to a pragmatic, albeit subjective, definition of a topotype : If two fingerprints are not recognizably similar, then isolates being compared are of a different topotype . In the case of DEN 2 isolates from different regions of the world, e .g ., Jamaica vs Thailand (Trent et al., 1983), there is clearly significant divergence and thus strains have been assigned to different topotypes . Taxonomic issues arise when examining strains isolated within one region, such as in the present analysis and that of Walker et al . (1988) . At what S ah value should two isolates be considered different topotypes? We propose that the degree of evolutionary diversity in the Thailand strains constitute a single topotype because there seems to be no practical taxonomic value to further subdividing the essentially continuous evolutionary pattern illustrated in Figs . 5 and 6 . Walker et al. (1988) compared fingerprints of 22 DEN 2 strains from Thailand and concluded that one strain, D80-141 (not in our analysis), represented a new topotype present in Thailand during the 1 980s because it was only about

533

56% similar to three other 1980 strains with which it was compared . To relate their results to ours, we compared the three possible similarity measurements among three strains that were analyzed in both studies (D80-100, D80-084, and TC16681-64) . Even though Walker et al . (1988) included about 1 1 more large oligonucleotides in the analysis region, their similarities agree well with ours . Although the similarity value of D80-141 to the other three strains (0 .56 Sob) would seem to be more divergent than any of the strains in our analysis, the significant genetic diversity revealed by our larger database (Fig . 3) does not support assigning new topotypes based on a single strain . The average similarity of the most divergent strain in our study, 1635-63, from the rest of the strains is 0 .65, but the individual Sab s range from 0 .49 to 0 .78 . We favor inclusion of strain D80-141 in the Thailand topotype and agree with Walker et al. (1988) that it is notably different than most of the strains of this topotype . If the apparently constant rate of microevolution for the Thailand strains, -0 .014 S,,/year, is assumed outside of the 1962--1986 time period, then an upper limit of the ''life expectancy" of the Thailand lopotype may be defined as that time when a consensus fingerprint pattern shares no oligonucleotides with the current consensus fingerprint, 71 .4 years . Assuming rates of evolution similar to the Thailand topotype, the geographic separation of a DEN 2 into two populations about 36 years ago would yield different topotypes . In practice, however, less stringent criteria have been used, hence viruses logically separated into different topotypes still have measurable similarities (e .g ., Repik et al., 1983 ; Trent et al., 1983) . Based on examination of hundreds of DEN 2 fingerprints obtained from strains isolated worldwide, we have identified 10 topotypes of DEN 2 viruses (Trent et al., 1989) ; each topotype consists of strains with at least 50% similarity among their fingerprints . Oligonucleotide fingerprint analysis of genome variability is an extremely sensitive technique so that strains which are 50% similar in fingerprints (Sob - 0 .5) are about 95% similar in their sequence (Aaronson etal., 1982) . High rates of mutation in RNA viruses relative to DNA viruses is expected because of the lack of error correction by RNA replicases (Reanney, 1982) However, the observed evolution of DEN 2 viruses within Thailand occurs at a rate much slower than rates measured for other RNA viruses (Holland et al_ 1982 : Palese and

FIG . 6 . Dendrogram of genetic relationships among Thailand DEN 2 viruses adjusted to reflect the different isolation dates of the viruses . The year of isolation of each virus is given in parentheses . The location of each bifurcation does not reflect the average S b of descendants as in Fig . 5 ; the S t between any two strains or groups of strains is the sum of all horizontal branch segments along the path connecting the strains or average of these segments where more than one strain descends from either side the bifurcation connecting them .



534

TRENT ET

Young, 1982) . For example, poliovirus isolated from different epidemics within the same country show no obvious genetic relationship to each other by fingerprint analysis (Nottay et al., 1981) . Young et al. (1979) provided evidence that genetic changes in human influenza A occur sequentially and many changes are retained in the new variants that emerge . The rate of evolution calculated for DEN 2 viruses in this study and the fact that 13 of the 36 consensus oligonucleotides (Figs . 2 and 3) are present in 90% or more of all viruses support a common ancestor for the viruses and the existence of conserved and variable regions in the RNA genome . Nucleotide sequence data for the entire genome of DEN 4 (Mackow et al., 1987) and DEN 2 viruses (Hahn et at, 1988 ; Deubel et at, 1988) and genome structural regions of DEN 1 virus (Mason et at, 1987 ; M . Chu, personal communication) have confirmed that regions of the genome are highly conserved . Factors that may contribute to the conservative rate of evolution of DEN 2 relative to other RNA viruses include fewer errors by the DEN 2 replicase during genome replication, reduced quantities of nucleotide fixations in transcribed regions imposed by codon usage (Grantham et al., 1981), and constraints imposed by secondary and tertiary interactions within the RNA genome essential to its replication and packaging (Trent etat, unpublished data) . The antigenic structure of the DEN 2 envelope protein including the critical neutralization site is stable (Hahn et al., 1988 ; Monath et al., 1986 ; Roehrig et al., 1983) . In the nonstructural region, NS3 and NS5 are the most highly conserved and the proteins NS1 and NS4 more variable (Deubel et at, 1988) . Thus, the surface glycoprotein E and putative polymerase genes NV3 and NV5 are highly conserved to maintain the integrity of virus functions necessary for growth in both vertebrate and mosquito host . Since DEN viruses are maintained and transmitted by mosquito vectors to humans without substantial genetic change, the virus must replicate conservatively in both hosts . This would imply that major genetic changes in the flavivirus genome are not tolerated in the biological cycle of mosquito and host, and that molecular constraints placed upon virus replication are imposed by both vector and host . Because of the redundancy of the genetic code, however, selection at the amino acid level cannot totally explain the conservative evolution observed at the RNA level . Studies of poliovirus epidemiology using genome sequence relatedness have suggested that viruses, characteristic of a particular endemic focus, persist and are infrequently displaced (Rico-Hesse etal ., 1 987) . In contrast, the introduction of DEN virus strains by infected persons or mosquitoes from areas of the world where virus is endemic, to new areas where conditions are

AL . suitable for transmission, continues to pose an important epidemiologic problem (Halstead, 1988) . Spread of virus strains from one geographic area to another together with immunological potentiation of severe DHF are problems not yet thoroughly understood (Rosen, 1977 ; Halstead, 1988) . Among the 57 viruses in this study, only one strain, 2780-77, was isolated from a patient with less severe symptoms, so that we cannot address the hypothesis that more severe disease (DHF/DSS) is caused by genetically different strains (Rosen, 1977) . Eight of the virus strains studied were isolated from Aedes aegypti mosquitoes between 1962 and 1966 (Table 1) . Fingerprints of these isolates were similar to those of the viruses isolated during the same time period from sera of patients with DHF/DSS . Assuming that the viruses isolated from mosquitoes represent a random sample from the virus population, our data do not support significant genetic differences between strains causing severe and mild disease . This was also concluded by Walker et al. (1988), whose study included four strains isolated from patients with mild symptoms . However, we cannot dismiss the possibility of increased pathogenicity caused by a few nucleotide changes in the viral genome, or changes in regions of the genome not represented in the large oligonucleotides examined in fingerprints . ACKNOWLEDGMENTS The authors thank Drs . D . Burke, S . B . Halstead, B . Innis, and L . Rosen for DEN virus strains used in this study . Dr . A . V. Vorndam, R . Tsuchiya, and S . Sviat provided technical support and interest . This work was supported by U .S . Army Medical Research and Development Command Contract Nos . PP-3809 and DAMD17-83C-3167 .

REFERENCES AARONSON, R . P ., YOUNG, J . F ., and PALESE, P . (1982). Oligonucleotide mapping : Evaluation of its sensitivity by computer simulation . Nucleic Acids Res . 10, 237-246 . BARNES, W . J . 5 ., and ROSEN, L . (1974) . Fatal hemorrhagic disease and shock associated with primary dengue infection on a Pacific Island . Amer. .J Trop. Med. Hyg . 23, 495-506 . BURKE, D. S ., NISALAK, A ., JOHNSON, D . E ., and SCOTT, R . MEN, (1988)A prospective study of dengue infections in Bangkok . Amer. J. Trop . Med. Hyg . 38, 172-180 . DEUHEL, V ., KINNEY, R . M ., and TRENT, D . W. (1988) . Nucleotide sequence and deduced amino acid sequence of the nonstructural proteins of dengue type 2 virus, Jamaica genotype : Comparative analysis of the full-length genome . Virology 165, 234-244 . DEWACHTER, R ., and FIERS, W . (1972). Preparative two-dimensional polyacrylamide gel electrophoresis of 32P-labeled RNA. Analy. aiochem. 49, 184-197 . EiGEN, M ., and SCHUSTER, P, (1977) . The hypercycle, a principle of natural self-organization . Part A, Emergence of the hypercycle . Naturwissenschaften 64, 541-565 . Fox, G . E ., and STACKEBRANDT, E . (1987) . The application of 16S rRNA cataloging and 5S rRNA sequencing in bacterial systematics . Methods Microbiol. 19, 405-458 . GRANTHAM, R . C ., GAUTIER, C ., Gouv, M ., JACOSEONE M ., and MERCIER

GENETIC VARIATION AND MICROEVOLUTION OF DEN 2

(1981) . Coach catalog usage is a genome strategy modulated for gene expressivity. Nucleic Acids Res. 9, 143-174 . GUNAKASEM, P ., CHANTRASRI, C ., CHAIYANLIN, S ., SIMASATHIEN, P ., and SANGPETCHSONG, V . (1981) . Surveillance of dengue hemorrhagic fever cases in Thailand . Southeast Asian J. Trop . Med . Public Health 12, 338-343 . GUZMAN, M . G . . KOURI, G . P ., BRAVO, J ., CALUNGA, M ., SOLER, M ., VASOUPZ, S ., SANTOS, M ., VILLAESCUSO, R, BASANTA, P ., INDAN, G ., and BALLESTER, J . M . (1984a) . Dengue haemorrhagic fever in Cuba . II

Clinical Investigations . Trans. R . Sac, Trop . Med. Hyg . 78, 239241 . GUZMAN, M . G ., Kouel, G . P ., BRAVO, J ., CALUNGA, M ., SOLER, M ., VAS

QUEZ S ., and VENFRO, C . (1984b) . Dengue hemorrhagic fever in Cuba . I Serological confirmation of clinical diagnosis . Trans . R . Sec . Trop . Men. Hyg .78,235-238 . HAHN, S ., GALLES, R ., HUNKAPILLER, T ., DALRYMPLE, J . M ., STRAUSS, J . H ., and STRAUSS, E . G . (1988) Nucleotide sequence of dengue 2 RNA and comparison of the encoded proteins with those of other flaviviruses . Virology 162, 167-180 . HAIL, E . L . (1979) "Computer Image Processing and Recognition .'' Academic Press, New York .

535

OuINros, F . N ., Lim, L. E ., JULIANO, L . A ., REYES, A ., and LACSON, P . (1954) . Hemorrhagic fever observed among children in the Philippines . Philipp. J, Pediatr. 3,1-19 . REANNEY, D . C . (1982) . The evolution of RNA viruses . Annu . Rev. Microbiol . 36, 47-73 . REECK, G . R ., DE HAEN, C„ TELLER, D . C ., DOOLITTLE, R . F ., FITCH, W . M ., DICKERSON, E ., CHAMBON, P ., MCLACHLAN, A . D ., MARGOLI ASH, E ., JUKES, T . H ., and ZUCKERKANDI, E . (1987) . "Homology" in proteins and nucleic acids : A terminology muddle and a way out of it. Cell 50, 667 . REPIK, P . M_ DALRYMPLE, 1 . M ., BRANDT, W . E ., McCowN, J . M ., and RUSSELL, P. K . (1983) . RNA fingerprinting as a method for distinguishingdengue 1 virus strains . Amer . I Trop. Med. Hyg . 32, 32,577 589 . RICO-HESE, R ., PALLANSCH, M . A ., NoTTAY, B . K., and KEW, 0 . M . (1987) . Geographic distribution of wild-poliovirus type 1 geno-

types . Virology 160, 311-322 . ROEHRIG, J . T ., MATHEWS, J . H ., and TRENT, D . W . (1983) . Identification

of epitopes on the E glycoprotein of Saint Louis encephalitis virus

HALSI EAO, S . B . (1980) . Dengue haemorrhagicfever-a public health problem and a field for research . Bull . WHO 58,1-21 .

using monoclonal antibodies . Virology 128,118-126 . ROSEN, L . (1977) . ''The emperor's new clothes" revisited, or reflections on the pathogenesis of dengue hemorrhagic fever . Amer. J. Trop . Med. Hyg.26,337-343 .

HALSTEAD, S . B . (1984) . Selective primary health care : Strategies for control of disease in the developing world . XI . Dengue . Rev. Infect. Dis . 6, 251-264 .

SANGKAWIBHA, N ., ROIANASUPHOT, S ., AHJANDRIK, S ., VIRIYAPONGSE, S ., JATANASEN, S ., SALITUL, V., PHANTHUMACHINDA, B ., and HALSTEAD, S . B . (1984) . Risk factors in dengue shock syndrome : A prosper

HALS LAD, S . B . (1988) Pathogenesis of Dengue : Challenges to Mo-

five epidemiologlc study in Razong, Thailand . 1 . The 1980 outbreak . Amer. J. Epidemiol 120, 653 -669 . SNEATH, P . H . A ., and SOKAL, R . R . (1973) ''Numerical Taxonomy .''

lecular Biology. Science 239, 476-481 . HA[ STEAD, S . 8 ., CHOW, J . S ., and MARCHETTE, N . J . (1973) . Immunological enhancement of dengue virus replication, Nature (London) 243,24-26 . HENCHAL, E . A ., MCCOWN, J . M ., SEOUIN, M . C ., GENTRY, M . C ., and BRANDT, W . E . (1983) . Rapid identification of dengue virus isolates by using monoclonal antibodies in an indirect immunofluorescence assay . Amer. .Trop . Med. Hyg. 32, 164-169 . HOLLAND, J ., SPINDLER, K ., HAROOYSKI, F ., GRABAU, E ., NICHOL, S„ and VANDEPAL, S . (1982) . Rapid evolution of RNA genomes . Science

215,1577-1585 . IGARASHI, A . (1978) . Isolation of Singh'sAeries albopictus cell clone sensitive to dengue and chikungunya viruses . J. Gen . Virol . 40, 531-544, LEMKIN, P . F ., and LPKIN, L . E . (1983) In ''Computing in Biological

Science' (Geisow and Barrett, Eds .) . pp . 181-231 . Elsevier Biomedical Press, Amsterdam/New York . MACKOW, E ., MAKING, Y ., ZHAO, B ., ZHANG, Y.-M ., MARKOEE, L., BucKi FIR WHITE, A ., GUILER, M ., CHANOCK, R ., and LAI, C .-1 . (1987) . The nucleotide sequence of dengue 4 virus : Analysis of genes coding for nonstructural proteins . Virology 159, 217-228 . MANSKE, C . L., and CHAPMAN, D . J . (1987) Nonuniformity of nucleo-

Freeman, San Francisco . TRENT, D . W ., CHU, M ., KLIKS, S . C ., MANSKE, C . L ., Fox, G . E ., and MONATH, T . P . (1989) . The molecular epidemiology of Dengue vi-

ruses : Genetic variation and microevolution . In ''Applied Virology Research : Vol . 2 . Virus variation and epidemiolgy" (F . Kurstak, R, G . Marusyk, F . A . Murphy, and M . H . V . Regenmortel, Eds .), (in press) Plenum, New York . TRENT D . W ., and GRANT, J . A . (1980) . A comparison of New World alphaviruses in the western equine encephalitis complex by immunochemical and oligonucleotide fingerprint techniques, J, Gen . Virol . 47, 261-282 . TRENT, D . W ., GRANT, J . A ., ROSEN, L ., and MONATH, I . P . (1983) . Ge-

netic variation among dengue 2 viruses of different geographic origin . Virology 128, 271-284 . TREN-, D . W ., GRANT,1 . A, VORNDAM, A . V ., and MONATH, 7 . P .0981) . Genetic heterogenicity among Saint Louis encephalitis virus isolates of different geographic origin . Virology 114, 319--332 . L ., REPIK, P ., DALRYMPLE, J . M ., and BISHOP, D . H . L . (1980) . Characterization of the viral RNA species of prototype dengue viruses . Amer . J, Trop . Med. Hyg . 29, 643-652 .

VFZZA, A ., ROSEN,

tide substitution rates in molecular evolution : Computer simulation and analysis of 5S ribosomal RNA sequences . l. Mot. EvoL 26, 226-251 . MASON, P . W ., McADA, P. C ., MASON, T . L ., and FOURNIER, M . J .

YOUNG, 1 . F ., DESSELeERGER, J ., and PALESE, P . (1979) . Evolution of human influenza A viruses in nature : Sequential mutations in the

(1987) . Sequence of the dengue-1 virus genome in the region en-

BURKE, D . S ., ROBBINS, S . L, and GORMAN, B . M . (1988) . Variation in Dengue type 2 viruses isolated in Bangkok during 1980 . J Gen . Virol. 69, 591-602 .

coding the three structural proteins and the major nonstructural protein NS1 . Virology 161, 262-267 . MONAIH, T . P ., WANDS, J . A., HILL, L. 1 ., BROWN, N . V ., MARCINIAK, R . A ., WING, M . A ., GENTRY, M . K ., BURKE, D . S ., GRANT, J . A ., and TPE+r, D . W . (1986) . Geographic classification of Dengue-2 virus strains by antigen signature analysis . Virology 154, 313-324 . NOTTAY, B . K ., KEW, 0 . M ., HATCH, M . H ., HEYWARD J . T ., and OsuESKI, J . F . (1981) . Molecular variation of type 1 vaccine related and wild

polioviruses during replication in humans . Virology 108, 405-423 . PAL ESE, P ., and YOUNG, J . F . (1982) . Variation of influenza A, B, and C viruses . Science 215, 1468-1474 .

genomes of new H 1 NIT isolates . Cell 18, 73-83 . WALKER, P . J ., HENCHAL, E . A ., BLOK, J ., PERK, P . M ., HENCHAL, L . S . .

WILSON, A . C ., CARLSON, S . S ., and WHITE, T. J . (1977) . Biochemical

evolution . Annu . Rev. Biochem. 46, 5 73-639 . World Health Organization (1986a) . Dengue and dengue hemorrhagic fever South-East Asia region . Weekly Epidemiol. Rec. 61, 20-21 . World Health Organization (1986b) . Dengue and dengue hemorrhagic fever . Weekly Epidemiol. Rec . 61, 205-208 . World Health Organization (1986c) . Dengue hemorrhagic fever . Diagnosis, treatment and control . WHO, Geneva .