Diverse Dengue Type 2 Virus Populations Contain Recombinant and Both Parental Viruses in a Single Mosquito Host

Abstract

Envelope (E) protein genes sampled from populations of dengue 2 (DEN-2) virus in individual Aedes aegypti mosquitoes and in serum from dengue patients were copied to cDNA, cloned, and sequenced. The nucleotide sequences of the E genes in more than 70% of the clones differed from the consensus sequence for the corresponding virus population at up to 11 sites, and 24 of the 94 clones contained at least one stop codon. Virus populations recovered up to 2 years apart yielded clones with similar polymorphisms in the E gene. For one mosquito, the clones obtained fell into two genotypes. One group of sequences was closely related to those of viruses recovered from dengue patients in the same locality (Yangon, Myanmar) since 1995 and were classified as Asian 1 genotype. The second group were Cosmopolitan genotype viruses which were also circulating in Yangon in 2000 and which were related to DEN-2 viruses sampled from southern China in 1999. Finally, one clone was identified as a recombinant genome composed of portions of these two “parental” genotypes. This is the first report of recombinant and parental dengue viruses in a single host.

Infections with dengue viruses result in a spectrum of disease ranging from inapparent to fever (dengue fever), hemorrhage (dengue hemorrhagic fever [DHF]), and hypovolemic shock (dengue shock syndrome), which, if untreated, may be fatal in up to 30% of patients (1). With perhaps one-third of the world's population, between the Tropics of Capricorn and Cancer, at risk of infection with these viruses, dengue is a significant public health burden (13). Considerable effort is being expended in developing vaccines to prevent dengue (5, 25), but the success of these vaccines will be influenced by the diversity and the genetic stability, over time, of the virus populations against which the vaccines are directed. Similarly, there is evidence that the ability of mosquitoes to transmit dengue viruses and the severity of disease in humans are influenced by the genotypes of both hosts and the virus (2, 6, 14, 18, 19, 20). Changes in virus genotype, therefore, have the potential to greatly influence patterns of disease and of disease transmission.

Genomes of RNA viruses, dengue viruses included, are expected to rapidly accumulate mutations due to the error-prone nature of RNA polymerases (9, 10). This diversification may lead to fitness losses due to the buildup of deleterious mutations or to rapid and unpredictable fitness gains (7, 11). Although this genetic variation is ultimately produced by mutation, it also could be shaped by natural selection, most likely from two sources. The first is the requirement for dengue viruses to replicate in multiple tissues in genetically diverse populations of mosquitoes in order to move from the gut to the saliva to infect a new host (15). It is less clear whether dengue virus replication in multiple tissues in humans is required to maintain the transmission cycle (4). A second potential source of selective pressure is cross-reactive antibodies in hosts infected previously with dengue virus serotypes which differ from that causing a current infection or in hosts infected previously with other flaviviruses. However, there are only two reports of efforts to quantify the diversity of dengue virus populations sampled from individual hosts, both dengue 3 (DEN-3) viruses and both from humans (29, 30). Furthermore, phylogenetic analyses suggest that dengue viruses in nature are subject to very weak positive selection pressure (26, 27). However, more-rapid genetic changes have been reported on occasion. In particular, there have been sudden changes in the genotypes of DEN-2 and DEN-3 virus populations in Thailand, which could be attributed to lineage extinction (21, 30), and there are reports of intraserotypic recombination within all four dengue virus serotypes which have produced viruses significantly different from the putative parental viruses (17, 24, 28, 31). Because of gaps in the virological record, it has not been possible, in most cases, to determine when or where these intraserotypic-recombination events might have occurred or which viruses were the actual parents. Furthermore, all dengue virus recombination events described to date have been detected in viruses passaged in vitro prior to study.

We sought to examine the diversity and evolutionary dynamics of DEN-2 populations within individual mosquito and human hosts. Dengue viruses were detected in serum from patients in the Yangon Children's Hospital, Yangon, Myanmar, and in female Aedes aegypti mosquitoes collected inside homes in metropolitan Yangon by adding 140 μl of serum or mosquito homogenate to 25-cm² monolayers of C6-36 Aedes albopictus mosquito cells and incubating the cultures for 7 days at 30°C. Individual mosquitoes were homogenized in 500 μl of sterile phosphate-buffered saline, pH 7.3, in sterile polypropylene microcentrifuge tubes with a sterile plastic pestle. Cells from the monolayers of C6-36 cells were stained by indirect immunofluorescence with monoclonal antibodies specific for flaviviruses (4G2), dengue viruses (2H2), or individual dengue virus serotypes (3, 12). The nomenclature employed to identify viruses and virus sources was as follows: country of origin.sample number (clone number if a sequence from a clone)/year of sample, for example, Myanmar.35004/99 and Myanmar.376-7/00.

RNA was extracted from serum, mosquito homogenate, or virus isolates (supernatant from cultures of dengue virus-infected C6-36 cells) with QIAamp viral RNA minicolumns (Qiagen, Hilden, Germany), according to the manufacturer's instructions. The RNA was reverse transcribed with Expand reverse transcriptase (Roche, Mannheim, Germany) and random hexanucleotide primers (Boehringer, Mannheim, Germany) (3). The envelope (E) protein gene region of the resultant cDNA was amplified by 36 cycles of PCR with a mixture of Taq and Pwo polymerases (Expand Long Template DNA polymerase; Roche) and oligonucleotide primers P789 and cP2504 (Table 1). The PCR product was analyzed in 1% (wt/vol) agarose-Tris-acetate-EDTA gels, and bands of cDNA of interest were recovered and purified with QIAquick gel extraction kits (Qiagen) according to the manufacturer's instructions. 5′ ATP extensions were added to the cDNA by incubating it with 10 μM ATP (Roche) and 5 U of Taq polymerase (Roche) for 1 h at 72°C. This cDNA was purified with a QIAquick PCR purification kit (Qiagen), ligated into pGEM-T Easy plasmids, and used to transform Escherichia coli JM109 according to the manufacturer's instructions (Promega, Madison, Wis.). Plasmids were purified from individual bacterial colonies which grew on Luria-Bertani agar supplemented with 100 μg of ampicillin/ml, 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and 80 μg of 5-bromo-4-chloro-3-indoyl-β-d-galactosidase/ml by using QIAprep Miniprep kits (Qiagen), and the DEN-2 virus E protein gene cDNA inserts were sequenced in both directions by using the oligonucleotide primers shown in Table 1, BigDye terminator cycle sequencing kits, and an automated sequencer (Applied Biosystems, Foster City, Calif.).

TABLE 1.

Oligonucleotide primers used for PCR or sequencing of DEN-2 E protein genes

Primera	Sequence (5′-3′)	Locationb
T7	TAATACGACTCACTATAGGG	Plasmid
P789	GAAACATGCCCAGAGAATTGAAACT	789-813
P900	TCATCCTACTGACAGCTGTCGCTCC	900-926
P1546	AAGCTTGGCTGGTGCACAGGCAATGGTT	1547-1574
cP1567	TGGCAACGGCAGGTCTAGGAACCATTG	1593-1567
cP2200	TGTAAACACTCCTCCCAGGGATCCAAA	2226-2200
cP2504	GGGGATTCTGGTTGGAACTTATATTGTTCTG	2537-2504
SP6	GATTTAGGTGACACTATAG	Plasmid

^aP, sense primer; cP, antisense primer.

^bNucleotides are numbered according to reference 8.

Twenty clones were sequenced from the DEN-2 virus population in each of three female Aedes aegypti mosquitoes: Myanmar.25/00, collected in June 2000 at the beginning of the dengue “season,” Myanmar.196/00, collected in July 2000 at the peak of the dengue season, and Myanmar.376/00, collected in September 2000 at the end of the dengue season. Fourteen clones from DEN-2 viremic patient serum Myanmar.32309/98 (DHF, 1998), 12 clones from serum sample Myanmar.35004/99 (DHF, 1999), and 11 clones from serum sample Myanmar.36273/00 (DHF, 2000) were sequenced. The E protein gene cDNA (PCR product) used to prepare the clones also was sequenced.

The numbers of nucleotide changes in clones from human serum (43 × 10⁻⁶ changes/nucleotide [nt]/cycle of PCR) and from mosquitoes (266 × 10⁻⁶/nt/cycle) were greater than (P < 0.05; χ² test) that predicted on the basis of reverse transcriptase (10⁻⁴) and Taq polymerase (4 × 10⁻⁶ to 9 × 10⁻⁶/nt/cycle) errors combined (22) and that measured following reverse transcription-PCR with a DEN-3 E gene template (5.3 × 10⁻⁶/nt/cycle) (29). The nucleotide sequences of less than 30% of the clones from any virus population matched the consensus sequence (Table 2). However, with one exception, the consensus nucleotide sequences derived by aligning the sequences of the cloned E genes from each host were the same as the consensus sequence derived by sequencing the viral cDNA (PCR product) from that host, which was used for cloning, suggesting that no marked bias had been introduced during the cloning process. The one exception was nt 1629 of virus from mosquito Myanmar.196/00, which was an unambiguous G when the PCR product was sequenced but which was A in 11 of the 20 clones (the remaining 9 clones had G). The sequences of three clones derived from mosquito Myanmar.376/00 were so divergent that they were excluded from the following analyses and are discussed in more detail below.

TABLE 2.

Nucleotide polymorphisms in DEN-2 virus E protein gene clones

Source	No. of clones	No. with consensus sequence	No. of clones with indicated no. of nucleotide changes											No. of stop codons	No. of sites at which nucleotides varied	dN/dSa	Mean diversity (%)b
			1	2	3	4	5	6	7	8	9	10	11
Human Myanmar. 32309/98	14	4	2	4	3	0	1	0	0	0	0	0	0	1	21	0.85	0.12
Human Myanmar. 35004/99	12	1	4	3	1	2	0	1	0	0	0	0	0	0	23	2.0	0.15
Human Myanmar. 36273/00	11	2	1	1	0	4	3	0	0	0	0	0	0	3	30	1.0	0.21
Mosquito Myanmar. 196/00	20	0	2	3	1	3	3	5	2	1	0	0	0	17	48	0.58	0.31
Mosquito Myanmar. 025/00	20	5	1	0	2	1	1	3	1	3	2	0	1	4	46	0.74	0.32
Mosquito Myanmar. 376/00	17	1	2	0	3	1	5	3	0	0	1	0	1	4	47	0.69	0.29
Isolate Myanmar. 376/00	20	5	4	3	0	5	3	1	0	0	1	0	0	6	33	0.82	0.20

^aRatio of nonsynonymous to synonymous nucleotide changes.

^bNumber of substitutions divided by the number of nucleotides sequenced times 100.

The mean diversity ([number of substitutions/total length of sequence {1,485 nt}] × 100) in the DEN-2 virus populations from mosquitoes and from humans ranged from 0.12 to 0.32%. These values are within the range calculated by others (0.1 to 0.84%) for a 430-nt portion of the E protein genes of DEN-3 viruses from serum (29). The proportions of variable sites in the clones of E genes from humans and from mosquitoes were not significantly different (P > 0.05; χ² test). Strikingly, the same synonymous polymorphisms occurred at two sites in three or more clones from virus populations in individual sera (nt 1908, A-C, Myanmar.35004/99; nt 1941, C-T, Myanmar.32309/98). Similarly, in the virus populations from individual mosquitoes, there were 17 sites at which the nucleotide sequences of three or more clones differed from the consensus sequence for that population, and seven of these “hypervariable” sites were present in two or more of the mosquito-derived virus populations (Table 3). There were seven sites at which the nucleotide sequences of clones from both mosquitoes and humans varied from their consensus sequences (Table 4), although only one of the nucleotides (nt 2093) was a site at which there were multiple changes within an individual virus population. Viral clones from mosquitoes and serum sampled in 2000 had the same polymorphisms at nt 2093 and 1879, whereas the polymorphisms from patients in 1998 and 1999 differed from those at the corresponding sites in clones from the mosquitoes (all collected in 2000). Finally, three of the polymorphic changes (at nt 1301 [two mosquitoes], 1366 [one patient], and 1937 [one mosquito]) altered the deduced amino acid sequence of the E protein at sites that had previously been shown to be under positive selection in DEN-2 viruses (27). However, the precise nature of these selection pressures is uncertain.

TABLE 3.

Polymorphic sites in the DEN-2 E gene at which three or more clones in each mosquito varied from the consensus sequence for the population

Nucleotidea	Change	Amino acid change
1139	C-T	E68, T-I
1163	C-T	E76, T-I
1300	A deletion
1386*	A-G	—b
1399	A-G	E155, T-A
1549	G-A	E205, A-T
1629*	G-A	Stop
1722	A-G	—
1800*	A-G	—
1891	A-G	E319, T-A
1897	G-A	E321, V-I
1937	A-T	E324, K-I
1946	T-C	—
1983*	T-A	—
2093*	A-C	E386, Q-P
2144*	A-G	E403, E-G
2373*	A-G	—

^a*, clones from at least two of the three mosquitoes contained the variations at this site.

^b—, no change.

TABLE 4.

Polymorphic sites in the DEN-2 E protein gene at which the nucleotide sequences of clones from both humans and mosquitoes varied from their respective consensus sequences

Nucleotide	Nucleotide and amino acid changes		Yr
	Mosquitoes	Humans
972	G-A, silent	A-T, silent	1998
1340	T-C, E135 L-P	T-C, E135 L-S	1999
1556	T-C, E207 L-P	T-A, E207 L-Q	1998
1877	A-G, E314 E-G	A-T, E314 E-V	1998
1879	A-C, E315 T-P	A-G, E315 T-A	1998
2093	A-C, E386 Q-P	A-C, E386 Q-P	2000
2304	A-G, silent	A-G, silent	2000

Similar polymorphisms were found in virus populations recovered from dengue patients over the 2-year sampling period. If Aedes aegypti mosquitoes survive for approximately 2 weeks after taking an infected-blood meal (16) and the average duration of viremia in patients is 7 days (1), the polymorphisms shared between populations of virus from 1998 and 2000 (Table 4) would have survived at least 28 cycles of human-mosquito transmission.

Of particular note was the number of polymorphic changes in the E gene clones which coded for stop codons (4 in 37 clones from serum and 25 in 57 clones from mosquitoes; Table 2). In many alphaviruses (for example, Sindbis virus; Aura virus; eastern, western, and Venezuelan equine encephalitis virus; and Ross River virus), production of nonstructural protein nsP4, the viral RNA polymerase, is regulated by the necessity to read through an OPAL (UGA) codon near the 3′ end of the preceding nsP3 gene (23). Given the frequency of nucleotide changes which generated OPAL codons in the dengue viruses in this study, it is possible that similar read-through of OPAL codons also may occur with dengue viruses. This is a question which could be addressed by making corresponding changes in dengue virus infectious clones.

Virus isolate Myanmar.376/00, recovered following culture of homogenate of mosquito Myanmar.376/00 with C6-36 cells for 7 days at 30°C, was analyzed in the same manner as virus populations in other mosquitoes or in serum. The consensus nucleotide sequence of the E protein gene of virus in the mosquito was the same as the consensus nucleotide sequence of the isolate. In addition to the polymorphisms which were unique to the viruses in the mosquito or to the isolate, there were nine sites where the parental virus and the isolate shared polymorphisms (three synonymous [nt 1386, 2349, and 2373], two resulting in stop codons [nt 1307 and 1629], and four nonsynonymous changes [nt 1495, C→T, E amino acid 187 [E187] P→S; nt 1769, T→C, E278 L→P; nt 2093, A→C or G, E386 Q→P or R, respectively; nt 2144, A→G, E403 E→G).

To determine the evolutionary history of these viruses and particularly the diversity within mosquito Myanmar.376/00, a phylogenetic analysis was performed. The sequences of 10 viruses sampled from Myanmar (including mosquito and human consensus sequences as well as representative clones from mosquito Myanmar.376/00) were compared with 20 E gene sequences collected from GenBank and representing the global genetic diversity of DEN-2, including that in regions close to Myanmar. The sequences of the Myanmar human viruses fell into Asian genotype 1 (26), as did the sequences of two clones from the Myanmar.376/00 mosquito virus population. However, three mosquito virus consensus sequences (Myanmar.025/00, Myanmar.196/00, and Myanmar.376-15/00) and the sequences of 17 of the 20 clones from mosquito Myanmar.376/00 (represented by clone Myanmar.376-15/00) formed part of the genetically distinct Cosmopolitan genotype (Fig. 1). Hence, an individual mosquito (Myanmar.376/00) contained viruses representing different DEN-2 genotypes, although none of the clones from the isolate contained cDNA with nucleotide sequences identical to either the Asian 1 or the recombinant genotype defined below. More strikingly, clone Myanmar.376-3/00, also from the mosquito with the mixed-genotype virus infection, changed phylogenetic position depending on which region of the E gene sequence was analyzed. As such a change in tree topology is indicative of recombination, this possibility was explored more fully by using a maximum-likelihood method (program LARD [17]). Under this analysis, clone Myanmar.376-3/00 was confirmed as recombinant with significant support (P < 0.005) for breakpoints at nt 1186 and 1506 in the E gene. This was confirmed in phylogenetic trees constructed for three sequence regions, nt 937 to 1186, 1187 to 1506, and 1507 to 2422 (Fig. 1). For the regions of nt 937 to 1186 and 1507 to 2422, Myanmar.376-3/00 fell into the Cosmopolitan genotype, showing a close evolutionary relationship with clone Myanmar.376-15/00, which is therefore its most likely parental sequence. However, in the region of nt 1187 to 1506 Myanmar.376-3/00 clustered in Asian genotype 1, with clone Myanmar.376-7/00 its most likely parental form. In all cases the conflicting phylogenetic positions of clone Myanmar.376-3/00 received strong bootstrap support. All trees were constructed by using a maximum-likelihood method, with the rate of substitution between each nucleotide type, the base composition, the extent of “among-site” rate variation, and the proportion of invariant sites estimated from the data (parameter values available from the authors on request). Bootstrap trees were reconstructed by the neighbor-joining method using the substitution parameters defined above. All these analyses were performed by using the PAUP* package (D. L. Swofford, PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4, Sinauer Associates, Sunderland, Mass., 2002).

FIG. 1.

Maximum-likelihood phylogenetic trees for different regions of the DEN-2 E gene depicting the recombinant structure of clone Myanmar.376-3/00 (boxed). All mosquito-derived sequences are shown in boldface. Genotypes are named according to reference 25. Bootstrap support values are shown only for the nodes which depict the different phylogenetic positions of strain Myanmar.376-3/00. All horizontal branch lengths are drawn to scale. Other DEN-2 viruses used in this analysis were as follows (GenBank accession numbers, where available, in parentheses): Brazil/90 (L10041), China.H1/99, China.H2/99, China.M1/99, CookIslands/97 (AF004020), India/57 (L10043), India/94 (strain CAMR10; AF410374), NewGuineaC/44 (AF038403), Philippines/96 (strain SLMC 179; AF297008), PuertoRico/69 (L10046), Trinidad/53 (L10053), Thailand/84 (strain D84-016; AF195033), Thailand/87 (strain D87-1040; U34934), Thailand/89 (strain D89-828; AF195041), Thailand/94 (strain K0008; AF100459), Thailand/95 (strain C0390; AF100462), Thailand/96 (strain C0167; AF100464), Thailand/98 (strain CAMR14; AF410377), Vietnam/97 (strain CTD113; AF410358), and Vietnam/98 (strain CTD226; AF410367).

The validity of the detection of a recombinant genome and both parental genomes in a single host is strengthened by the following associated observations. Viruses with a genotype similar to that of one parent (Asian 1 genotype) had been circulating for several years in the locality in which the mosquito was caught (Yangon; Fig. 1), and viruses with the genotype of the other parent (Cosmopolitan) were recovered from other mosquitoes in Yangon in 2000. The oligonucleotide primers used to prepare the cDNA for cloning were at least 300 nt 3′ and 5′ of the sites of recombination, and the sequence of clone Myanmar.376-3/00 differs from those of all other Myanmar viruses in the recombinant region (Fig. 1). However, it is not possible to determine whether the recombination event occurred in the mosquito in which the recombinant virus was detected or in a human host from whom a blood meal was taken, although mosquito Myanmar.376/00 was not blood engorged when collected, suggesting that it had not fed for at least 48 h and that some virus replication had occurred in it (15). Recombination may be much more frequent in dengue virus populations than present data suggest, but it may not be being detected because almost all genetic analyses have employed only consensus sequences.