Molecular Characterization of Serotype G9 Rotavirus Strains from a Global Collection

Virology 278, 436–444 (2000) doi:10.1006/viro.2000.0682, available online at http://www.idealibrary.com on

Molecular Characterization of Serotype G9 Rotavirus Strains from a Global Collection M. Ramachandran,* ,1 C. D. Kirkwood,* L. Unicomb,† N. A. Cunliffe,‡ ,§ R. L. Ward,㛳 M. K. Bhan, ¶ H. F. Clark,** R. I. Glass,* and J. R. Gentsch* *Viral Gastroenteritis Section, Division of Viral and Rickettsial Disease, Centers for Disease Control and Prevention, Atlanta, Georgia 30333; †International Center for Diarrheal Disease Research, Dhaka, Bangladesh; ‡Department of Medical Microbiology and Genitourinary Medicine, University of Liverpool, United Kingdom; §Wellcome Trust Research Laboratories, Blantyre, Malawi; 㛳Division of Infectious Disease, Children’s Hospital Medical Center, Cincinnati, Ohio 45229; ¶Department of Pediatric Gastroenterology, All India Institute of Medical Sciences, New Delhi, India; **Section of Infectious Disease, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania 19104 Received May 4, 2000; returned to author for revision July 17, 2000; accepted September 22, 2000; published online November 22, 2000 Between 1992 and 1998, serotype G9 human rotavirus (RV) strains have been detected in 10 countries, including Thailand, India, Brazil, Bangladesh, Malawi, Italy, France, the United States, the United Kingdom, and Australia, suggesting the possible emergence of the fifth common serotype worldwide. Unlike the previously characterized reference G9 strains (i.e., WI61 and F45), the recent G9 isolates had a variety of gene combinations, raising questions concerning their origin and evolution. To identify the progenitor strain and examine the on-going evolution of the recent G9 strains, we characterized by genetic and antigenic analyses 16 isolates obtained from children with diarrhea in India, Bangladesh, the United States, and Malawi. Specifically, we sequenced their VP7 and NSP4 genes and compared the nucleotide (nt) and deduced amino acid sequences with the reference G9 strains. To identify reassortment, we examined the products of five gene segments; VP4, VP7, and NSP4 genotypes (genes 4, 9, and 10); subgroups (gene 6); electropherotypes (gene 11); and the genogroup profiles of all of the recent G9 isolates. Sequence analysis of the VP7 gene indicated that the recent U.S. P[6],G9 strains were closely related to the Malawian G9 strains (⬎99% nt identity) but distinct from G9 strains of India (⬃97% nt identity), Bangladesh (⬃98% nt identity), and the reference strains (⬃97% nt identity). Phylogenetic analysis identified a single cluster for the U.S. P[6],G9 strains that may have common progenitors with Malawian P[6],G9 strains whereas separate lineages were defined for the Indian, Bangladeshi, and reference G9 strains. Northern hybridization results indicated that all 11 gene segments of the Malawian P[6],G9 strains hybridized with a probe derived from a U.S. strain of the same genotype and may have the same progenitor, different from the Indian G9 strains, whereas the Bangladesh strains may have evolved from the U.S. G9 progenitors. Overall, our findings suggest that much greater diversity among the newly identified G9 strains has been generated by reassortment between gene segments than through the accumulation of mutations in a single gene. © 2000 Academic Press

selective pressures of wild strains or through genomic reassortment of wild strains with vaccine strains, raising concerns regarding our ability to detect novel RV variants. Monitoring prevalent serotypes using molecular methods is important to detect new strains before they become epidemic. The mature RV particle is composed of a triple-layered protein capsid enclosing 11 segments of doublestranded RNA genome, with each gene segment encoding at least one structural or nonstructural protein (Estes and Cohen, 1989). The outer capsid contains the two neutralizing proteins, VP4 and VP7, encoded by gene segments 4 and 7, 8, or 9, which are protective in nature and may be more sensitive to point mutations under selective immune pressures (Hoshino et al., 1985). Other proteins that may be important for virulence and protection include VP6, the group-specific protein, encoded by gene segment 6 and NSP4, a nonstructural protein encoded by gene segment 10 (Ball et al., 1996; Burns et al., 1996). During mixed infections, the 11 genome segments of the two parental RV strains could theoretically reassort into 2 11 different genome constellations, if reassortment

INTRODUCTION Rotaviruses (RVs) evolve by multiple mechanisms. The most common process is the accumulation of spontaneous mutations in the genome, which has been documented among all human and animal RVs (Pedley et al., 1984; Ramig, 1997). Larger genetic changes within a gene such as insertion, deletion, and concatemerization occur less frequently, and their role in generating viable human and animal RV populations is not fully assessed. Last, because the virus has a segmented genome, strains can also evolve through reassortment between gene segments (Ramig, 1997). Therefore, recent interest in the use of animal-human reassortant RV vaccines has raised important questions concerning the impact of the vaccine strains on the evolution of new variants. It is possible that new strains can emerge through immune

1 To whom correspondence and reprint requests should be addressed at Viral Gastroenteritis Section MS G04, Centers for Disease Control and Prevention, 1600 Clifton Rd., N.E., Atlanta, GA 30333. Fax: 404-639-3645. E-mail: [email protected].

0042-6822/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

436

MOLECULAR CHARACTERIZATION OF G9 ROTAVIRUSES

were random. However, surveillance studies conducted during the past two decades indicate that only four strains, P[8],G1, P[4],G2, P[8],G3, and P[8],G4, are common worldwide, suggesting that reassortment resulting in viable RV constellations may be restricted (Gentsch et al., 1996). Accordingly, genogrouping studies that examine the hybridization profile of all 11 gene segments have shown that most human RV strains are associated with only two major genogroups, Wa (P[8],G1) or DS-1 (P[4],G2), indicating the existence of strong determinants that allow genes within each “gene family” to be preserved (Flores et al., 1985; Nakagomi et al., 1989). Although speculative, recent reports of promiscuous reassortment associated with the newly reemergent serotype G9 RV strains was surprising, raising questions concerning the origin and evolution of the new G9 strains, whether they arose from previously characterized reference G9 strains (i.e., WI61 and F45), or whether they were able to reassort across the barriers of gene families to form new genomic constellations (Arista et al., 1997; Bon et al., 2000; Clark et al., 1987; Cubitt et al., 2000; Green et al., 1989; Griffin et al., 2000; IturrizaGomara et al., 2000; Linhares et al., 1996; Nakagomi et al., 1990; Palombo et al., 2000; Ramachandran et al., 1996, 1998; Unicomb et al., 1999; Urasawa et al., 1992). In this study, we characterized 16 G9 strains recently isolated from India, Bangladesh, Malawi, and the United States to understand their origins and genetic relationship to each other and to the earliest identified reference strains. We examined the nucleotide (nt) sequence of the VP7 genes to assess their genetic variation over the years. To identify genomic reassortment, we characterized five gene segments of these strains by antigenic and genetic methods, including RT–PCR for genotype VP4, VP7, and NSP4 genes; ELISA for subgroup VP6 antigen; and PAGE profile for gene 11. Genogrouping method was used to identify the origins of all the gene segments by hybridization with whole genome probes derived from standard strains Wa and DS-1. Comparison of results using these distinct approaches will allow us to assess whether evolution was more rapid by reassortment or through the accumulation of point mutations. This study raises serious concerns with the emergence of new, the reemergence of old, and the spread of RV strains across the geographic barriers. Although preliminary data support and provide evidence for the evolution and global spread of G9 strains, detailed characterization would allow us to assess their evolution at molecular level, within a single gene segment as well in the entire gene constellation. Further, there are important considerations for future RV vaccines because serotype G9 is not included in any of the previous RV vaccines and the inclusion of this serotype may be important for complete protection against RV disease, especially in areas where these new G9 strains are endemic.

437

RESULTS Serotype G9 RV strains We selected 16 G9 isolates from four countries that represented the four most common reassortants for molecular analysis (Table 1). These included two strains that had the same genotypic properties (P[8],G9 genotype, subgroup II, and long electropherotype [e-type]) as the reference G9 strains, WI61 and F45, and were grouped as prototypes for the purposes of distinction. Ten variant 1 (P[6],G9, SGI, S) strains, two each of variant 2 (P[6],G9, SGII,L) and variant 3 (P[6],G9, SGII,L) isolates, which had various genotype combinations that appeared to reflect reassortment of VP7 and VP4 genes between Wa and DS-1 genogroup strains were also selected for analysis. Overall, variant 1 represented the most prevalent isolate in this period and circulated in six U.S., three Bangladeshi, and one Malawian city (Cunliffe et al., 1999; Ramachandran et al., 1998; Unicomb et al., 1999). Variant 2 was detected less frequently, circulating in one U.S. and one Bangladeshi city. Variant 3 was found circulating in five Indian and one Bangladeshi city. The prototypes were found in two U.S. cities in the 1996–1997 RV season, whereas a single isolate for Bangladesh was found in 1996. P and G genotypes The VP7 (G) and (VP4) P genotypes of strains US430, US244, US321, and US1145 were determined by RT–PCR using primers specific for G9 strains described previously. All other strains included in this study have been genotyped elsewhere but were regenotyped in this study for confirmation and are listed in Table 1 (Cunliffe et al., 1999; Ramachandran et al., 1998; Unicomb et al., 1999). VP7 gene sequence comparisons The VP7 gene sequences of the 16 G9 strains obtained from the United States (n ⫽ 9), India (n ⫽ 2), Bangladesh (n ⫽ 3), and Malawi (n ⫽ 2), were analyzed and compared with each other and with the reference G9 strains (data not shown). All of the strains had ⱖ89% nucleotide (nt) sequence and ⬎96% deduced amino acid (aa) sequence identity to strain WI61, suggesting they belonged to serotype G9. The nt sequence data indicated the recently identified U.S. strains were more closely related to each other (⬍1.3% nt difference) than to the reference G9 strains WI61 and F45 (⬃89% nt identity), which were isolated 13–16 years ago. Within the United States, variants 1 and 2 were more homologous to each other (⬍0.4% nt difference) than to the prototype (⬍1.3% nt difference) isolated in the same year. Interestingly, the VP7 gene of the U.S. G9 strains were highly homologous to the Malawian strains at the level of nt and aa sequence identity (⬎99%) than to those from India and

438

RAMACHANDRAN ET AL. TABLE 1

P and G Types, Subgroup Specificity, and Electropherotypes of the G9 Strains Isolated from the United States, Malawi, Bangladesh, and India Gene segment/specificity Prototype/variant G9 strains

4

6

7–9

10/11 (e-type)

Location

Year isolated

Strain

Prototype

P[8]

SG II

G9

L

Variant I

P[6]

SG I

G9

S

Variant 2

P[8]

SG I

G9

S

Variant 3

P[6]

SG II

G9

L

Variant 4

P[11]

SG II

G9

L

Bangladesh a U.S. (Indianapolis) b U.S. (Indianapolis) b U.S. (Indianapolis) b U.S. (Little Rock) b U.S. (New Orleans) c U.S. (Philadelphia) d U.S. (Kansas City) b U.S. (Cincinnati) e U. S. (Cincinnati) e Malawi f Malawi f Bangladesh a Bangladesh a U.S. (Omaha) b India g India g India h

1995–96 1996–97 1996–97 1996–97 1996–97 1996 1995 1996–97 1997–98 1997–98 1997 1997 1995–96 1995–96 1996–97 1994 1994 1985

BD524 a US1212 b US1205 b US1206 b US1111 b US430 g US1145 c US1071 b US244 d US321 d MW 47 e MW 69 e BD426 a BD431 a US1343 b INL1 f ING16 f 116E

a

Unicomb et al., J. Clin. Microbiol. 37, 1999. Ramachandran et al., J. Clin. Microbiol. 36, 1998. c J. Maffei, unpublished data. d H. F. Clark and V. Jain, unpublished data. e R. L. Ward, unpublished data. f Cunliffe et al., J. Med. Virol. 57, 1999. g Ramachandran et al., J. Clin. Microbiol. 34, 1996. h Das et al., Virology 197, 1994. b

Bangladesh (⬃97% aa identity). The Indian and Bangladesh strains were less closely related (⬃96% nt identity) to each other, suggesting possibly independent origins for these strains. The VP7 gene of strain 116E, the neonatal strain from India, was distinct from the recent Indian, Bangladeshi, Malawian, and U.S. G9 isolates (⬃88% nt identity). Comparison of the variable antigenic regions of the VP7 protein All G9 isolates from within the same country had identical deduced aa sequences in the antigenic regions A, B, C, D, and F of the VP7 gene, irrespective of their e-types, P-genotype, or subgroup specificity. Therefore we selected a representative G9 strain for each country, including strain US1205 for the United States, MW47 for Malawi, BD431 for Bangladesh, and INL1 for India (Table 2). Among the strains isolated from the United States, six aa substitutions were detected between strain US1205 and the reference strains WI61 and F45 in regions A, C, D, and F. Specifically, single aa substitutions in regions A and F at positions 87 and 242, respectively; two substitutions in region C at positions 208 and 220; and two substitutions in region D at positions 73 and 74 were identified among all of the U.S. G9 strains, including

strain US1205. The two G9 strains from Malawi were identical to each other and to the U.S. G9 strains in all antigenic regions (A–F) of the VP7 gene. The Indian strains, INL1 and ING16, varied from the U.S. strains in regions B and F, whereas single aa substitutions were detected at positions 142 and 242, respectively. The Bangladeshi strains varied from the U.S. strains in regions A and D, where a single aa substitution was detected at positions 87 and 66, respectively. Overall, the VP7 antigenic regions were closely related among the recent G9 strains, differing from each other by only a few aa in the five antigen regions examined. Furthermore, the aa substitutions were specific and well conserved among the strains within the same country, suggesting possible common progenitors for these strains and regional evolution within each country. Phylogenetic analysis The phylogenetic analysis of the VP7 nt and aa sequences to understand the genetic relationship between the G9 strains was performed with 100 bootstrap repetitions using the Paupsearch program (Wisconsin Genetic Computer Groups), and the trees were generated using the Treetool program (Devereux et al., 1984). A non-G9 VP7-encoding gene obtained from a rhesus RV

MOLECULAR CHARACTERIZATION OF G9 ROTAVIRUSES

439

TABLE 2 Comparison of the Antigenic Regions A–D and F for Human RV Strains G9 strain A (87–101) WI61/F45 116E US1205 a MW47 b BD431 c INL1 d B (142–152) WI61/F45 116E 1205 a MW47 BD431 c INL1 d C (208–221) WI61/F45 116E 1205 a MW47 b BD431 c INL1 d D (65–76) WI61/F45 116E 1205 a MW47 b BD431 c INL1 d F (235–245) WI61/F45 116E 1205 a MW47 b BD431 c INL1 d a

Strain Strain c Strain d Strain b

Antigenic region

A I T T I T

E -

A -

S -

T -

Q -

I -

G -

D -

T -

E -

W -

K -

D G -

M V V

K -

Y -

D -

S -

T -

L -

E -

L -

D -

T T I I I I

T -

T -

N -

T -

A -

T -

F -

E -

E -

V -

A -

A T T T T T

S N -

T -

A E -

Y -

A T -

N -

S -

S -

Q -

L Q Q Q Q Q

D E E E E E

T -

F -

H -

K -

L -

D -

V -

T -

T N N N N K

T -

T -

C -

T -

I -

T -

US1205 (USA) was identical to eight other U.S. G9 strains in variable regions A–D and F. MW47 (Malawi) was identical to MW69, also from Malawi, in regions A–D and F. BD431 (Bangladesh) was identical to two other Bangladesh strains, BD426 and BD524, in regions A–D and F. INL1 (India) was identical to strain ING16, also from India, in regions A–D and F.

strain, RRV, belonging to serotype G3 was included as an outgroup to understand the relationship between the G9 RV strains in comparison with non-G9 strains. The relationship between the aa sequences of the G9 strains is depicted in Fig. 1. Phylogenetic analysis identified a single cluster for the U.S. variant I G9 strains, with closely linked origins to the Malawian variant 1 G9 strains, whereas separate genetic lineages were defined for the Indian and Bangladesh G9 strains. Within the United States, prototypes were more closely related to variant 1 despite possessing different subgroup, P-type, and etype specificities than to the reference strains, F45 and WI61. Similarly, the prototype isolated in Bangladesh was closer to variants I and II isolated in the same country than to the reference strains, F45 or WI61, or to the prototypes isolated in the United States. The Indian G9

strains clustered in one distinct group and were different from all other G9s, including the human-bovine reassortant G9 strain, 116E, previously isolated from Indian neonates. Genogroup analysis Strain US1205, a representative of the predominant serotype 9 strain circulating in the United States in 1996– 1997 (variant I), was recently shown to belong to the DS-1 genogroup by Northern hybridization using whole genome probes (Kirkwood et al., 1999b). To determine whether the US1205-like strains were prevalent in other countries as well and to investigate the origins of all the gene segments of variant 1, 2, and 3 G9 strains isolated from India, Bangladesh, and Malawi, Northern hybridiza-

440

RAMACHANDRAN ET AL.

FIG. 1. Phylogenetic analysis of the VP7 nucleotide sequences of the G9 strains was determined using a computer software program, Paupsearch, using the seqlab component of the GCG programs of the Wisconsin Genetic Computer Group. All of the G9 strains were rooted to strain RRV and drawn using Treetool Program. *Prototype G9 strains (P[8];G9, SGII, long e-type): WI61, US1212 (United States); F45 (Japan); BD524 (Bangladesh). **Variant I G9 strains (P[6];G9, SGI, short e-type): US1205, US 1206, US1111, US 1071, US1145, US 430, US 1343, US 321, US244 (United States); MW47, MW 69 (Malawi); strain BD 426 (Bangladesh). *** Variant II G9 strains (P[8];G9, SGI, short e-type): 1343 (United States), BD431 (Bangladesh). @Variant III G9 strains (P[6];G9, SGII, long e-type): ING16, INL1. @@Variant IV G9 strain (P[11];G9, SGII, long e-type): 116E. The VP7 gene of RRV, a rhesus RV strain belonging to serotype 3, was included as an outgroup.

tion was done using whole genome probes derived from double-strained RNA of strains US1205 and Wa. All of the segments of the representative Malawian G9 strain MW47 hybridized with the US1205 whole genome probe, indicating that the Malawi and the U.S. strains may have a closely linked origin (Fig. 2B, lane 4). The Bangladeshi strain BD426, which had a similar e-type and the same P-type specificity, produced only 10 bands when hybridized with US1205 probe (Fig. 2B, lane 3), suggesting that the gene segment 7 of strain BD426, which did not bind to 1205 probe, may have a separate origin. Strain US1343, a P[8],G9 strain from the United States possessing SGI and a short e-type pattern, hybridized with all of the segments of US1205 probe except segment 4, which designates the P[8] VP4 specificity. The Indian strain INL1 produced only two bands corresponding to gene segments 4 and 9 when hybridized with US1205 probe (Fig. 2B, lane 2). Strain BD524, from Bangladesh, having P[8],G9 genotypes, SG II specificity, and a long e-type, produced more than nine hybrids with Wa probe (data not shown), whereas only one band was detected, corresponding to segment 9, when the same strain was hybridized with 1205 probe (Fig. 2B, Lane 6). Thus strain BD524 was more closely related to the Wa genogroup than to US1205, which belonged to the DS-1 genogroup. Similarly, the U.S. strain US1212, having P[8],G9 genotypes, SGII specificity, and long e-type, was shown to belong to the Wa genogroup along with the Indian strain INL1 with P[6],G9 genotypes, SGII specificity, and a long

e-type, producing at least five hybrid bands. We were unable show Wa probe results in the form of a figure because the amount of RNA from strains BD524 and INL1 was limiting. NSP4 genotypes The origin, diversity, and clustering of the NSP4 gene among the G9 reassortant RV strains were examined by sequence analysis and phylogenetic clustering methods. The NSP4 sequences of the U.S. G9 strains US1205, US1071, US1145, and US1212 have been described previously and used in this study for comparisons and phylogenetic analysis (Kirkwood et al., 1999a). In this study, we sequenced NSP4 gene from nine G9 strains using previously described primers and compared their sequences with those of other NSP4 genes in the database. Altogether, these results strengthened our genogrouping data and provided independent confirmation for the reassortment of the VP7 gene of G9 strains between Wa and DS-1 gene families. Variants I and II, with SGI specificity and short e-type, belonged to NSP4 genotype I, whereas the prototypes and variant III, with SGII specificity and long e-type, belonged to NSP4 genotype II. Phylogenetic analysis identified two main lineages for the two NSP4 genotypes, similar to previously published sources (data not shown) (Cunliffe et al., 1997; Horie et al., 1997; Kirkwood and Palombo, 1997). In many respects, clustering of NSP4 genes was similar to VP7

MOLECULAR CHARACTERIZATION OF G9 ROTAVIRUSES

FIG. 2. Genogroup analysis of serotype G9 strains. RV dsRNA of G9 strains US1206, INL1, BD426, MW47, US1343, and BD524 (lanes 1–6) was separated in polyacrylamide gels, visualized by ethidium bromide staining (A), and electrically transferred to nylon membrane, UV crosslinked, and probed with whole genome probe derived from strain US1205 (B).

gene clusters. For example, variants I and II had closely related NSP4 sequences (⬎99.4% nt identity), similar to their VP7 genes (⬎99.3% nt identity), and clustered together. The NSP4 genes of variant III and the prototypes had an nt difference of ⬃1.5%, similar to their VP7 genes (⬃1.5% nt difference), and formed clusters in the same main lineage. DISCUSSION This study characterizing 16 G9 strains from four geographic locations has provided new insights into the origin and evolution of human RVs. First, a number of G9 strains with closely related VP7 sequences were associated with a variety of other gene segments. For example, in the United States, little difference in the VP7 gene sequences of many G9 strains was noted; nonetheless

441

three variants emerged, with distinct phylogenetic sublineages and a diverse assortment of genetic constellations. Second, we found virtually identical G9 strains circulating in two different countries: the United States and Malawi. The variant I G9 strains from these two countries were closely related at the level of nt and aa sequences of the VP7 gene and shared identical antigenic regions (VP7) and overall genetic constellations (DS-1 genogroup), indicating that these strains may have very recent common origins. On the other hand, strains from the two neighboring countries, India and Bangladesh, were less homologous at the level of nt and aa sequences (VP7 gene) and had independent lineages by phylogenetic analysis. Overall, our results provide evidence that much greater diversity among the RV populations is generated by reassortment between gene segments than by accumulation of spontaneous mutations within a single gene (VP7, NSP4) over a period of years. New variant P[6],G9 strains were commonly identified in all four countries, and in the United States and Bangladesh, prototype G9s were also detected cocirculating with the variants, raising questions of whether the variant G9s evolved from the reference G9 strains, WI61 and F45, that were first isolated 16 and 13 years ago in the United States and Japan, respectively (Clark et al., 1987; Green et al., 1989) or represented an independent introduction of novel G9 strains from humans or animals. Comparisons of VP7 gene sequences indicate that the newly identified G9 strains, both variants and prototypes, were less homologous to the reference G9 strains, WI61 and F45, at the level of nt and sequences, having several distinct aa substitutions in the key antigenic regions indicating evolution of this gene over a period of years. Although our data do not allow us to determine the origin of the new variants, we speculate that a common progenitor of the new G9 strains may have been introduced to various countries where the strains evolved locally through gene reassortment during mixed infections and with accumulation of genetic mutations over a period of years. Our present findings that serotype G9 strains were commonly associated with both Wa and DS-1 genogroups were unusual and contradicted the earlier notion that genes within each gene family were genetically linked. Several studies in the past have demonstrated that reassortment across Wa and DS-1 genogroups occurred at very low frequencies and attributed this phenomenon to the existence of strong but undefined determinants that allow genes within each gene family to be preserved (Brown et al., 1988; Flores et al., 1988; Garbarg-Chenon et al., 1984; Nakagomi et al., 1987; Sethi et al., 1988; Steele and Alexander, 1988). Even during experimental mixed infections in cell culture, crosses between strains with serotype 2, SGI, and short e-type (DS-1 gene family) and serotype 1, SGII strains with long e-type (Wa gene family), occurred at very low frequencies

442

RAMACHANDRAN ET AL.

compared with reassortment within each genogroup (Garbarg-Chenon et al., 1984; Ward and Knowlton, 1989; Ward et al., 1988). Therefore, ours and other recent reports of reassortment of G9 serotype across gene families resulting in a variety of stable G9 constellations are intriguing (Cunliffe et al., 1999; Griffin et al., 2000; Ramachandran et al., 1998; Unicomb et al., 1999; Urasawa et al., 1992). It is not clear whether the physical properties of the serotype 9 VP7 protein or its interaction with other structural proteins is important in allowing such stable G9 reassortants to thrive in the community and spread extensively across various continents. This study has several limitations. First, only 16 G9 strains from four countries were characterized and these strains may not fully represent the serotype. Inclusion of G9 strains obtained before year 1994, especially in the United States, will be helpful and may possibly identify evolutionary intermediates or progenitor strains for the recent G9 isolates. So far, two studies conducted in the United States and Bangladesh have reported that new variant G9 strains were not identified before 1995 based on their analysis of earlier collections spanning 10 years (Jain et al., 1999; Unicomb et al., 1999). However, collections in other countries need to be examined more carefully. Second, we were restricted to characterization of only 5 of 11 RV segments and may well have overlooked additional differences by not detailing other gene segments. Nonetheless, results of whole genome hybridization suggest that these differences may be marginal. Overall, our findings have important implications for RV evolution, and together with other reports, challenge the previous belief that only four serotypes were globally common. For the first time, a new serotype, G9, has been documented in 10 countries, across all five continents and is capable of causing substantial numbers of episodes of diarrhea in children, suggesting that G9 may be the fifth globally important serotype along with types G1–G4. Identification of diverse G9 reassortants from children with natural RV infections has raised important questions concerning the number of serotypes that need to be targeted by polyvalent RV vaccines. Concerns are being raised about the possibility of emergence of new strains by reassortment between vaccine viruses and human serotype G9 RV strains. Finally, our study has important implications for the emergence of new variants in G9 endemic regions where the vaccinees may be naturally infected with G9 strains. MATERIALS AND METHODS RV isolates Sixteen human G9 RV strains isolated from stools of patients hospitalized with diarrhea were used for molecular studies. Of these, nine G9 strains were obtained from the United States, including five from the collaborating laboratories of the National Rotavirus Strain Sur-

veillance System (NRSSS) isolated in the 1996–1997 RV season, one from an outbreak in a neonatal intensive care unit in New Orleans in 1996, one from Philadelphia isolated in 1995, and two from Cincinnati isolated in the 1997–1998 season (Table 1). Seven other G9 strains were also included: two from India isolated in 1994, three from Bangladesh isolated between 1995 and 1996, and two from Malawi isolated in the 1997–1998 season (Table 1). G serotype and subgroup analysis using monoclonal antibodies The VP7-specific monoclonal antibodies WI61:1, F45:1, and F45:8 and subgroup-specific monoclonal antibodies 255-60 and 631-9 used in this study have been previously described (Greenberg et al., 1983; Kirkwood et al., 1993). Strains US430, US1343, US244, US321, US1145, INL1, ING16, MW47, and MW69 were G-serotypes and subgrouped by ELISA, using methods described previously (Kirkwood et al., 1993). The G-serotypes and subgroup analysis of all other strains were determined previously and described elsewhere (Cunliffe et al., 1999; Ramachandran et al., 1998; Unicomb et al., 1999). RNA extraction and electropherotype analysis RV RNA was extracted from 10% stool extracts using the glass powder method, as previously described (Gentsch et al., 1992), and resuspended in RNase-free diethyl pyrocarbonate-treated water. The e-types of RV isolates US430, US1343, INL1, ING16, MW47, and MW69 was determined by PAGE and visualized by silver staining (Pereira et al., 1983; Raj et al., 1989). All other strains have been e-typed earlier (Cunliffe et al., 1999; Ramachandran et al., 1998; Unicomb et al., 1999). RT–PCR and automated nucleotide sequencing P and G genotypes of the G9 isolates were determined using heminested PCR, agarose gel electrophoresis, and ethidium bromide staining (Gentsch et al., 1992; Gouvea et al., 1990). The full-length PCR products of the gene coding for the VP7 protein of all 16 G9 strains were amplified using a mixture of degenerate Beg9 and End9 primers. Degenerate primers were based on serotypes G1, G2, G3, G4, G5, G8, and G9 VP7 nt 1–28 for Beg9 primers and nt 1036–1062 for End9 primers and have been described previously (Gouvea et al., 1990; Ramachandran et al., 1998). The PCR products were gel purified using a Qiagen kit (Qiagen Inc., Valencia, CA), and the nt sequence of the entire open reading frame of the VP7 gene was determined using dideoxynucleotide chain terminator method with a BigDye sequencing kit in an automated sequencer (ABI 377; Applied Biosystems Inc., Foster City, CA). The primers used for sequencing have been described previously (Ramachandran et al., 1998). Overlapping sequences were analyzed by Sequencher program, Version 3.0 (Gene Codes Corp. Inc.,

MOLECULAR CHARACTERIZATION OF G9 ROTAVIRUSES

Ann Arbor, MI) and compared with sequences in the GenBank using the Wisconsin Genetics Computer Group programs, Version 8.0. Phylogenetic analysis We examined the phylogenetic relatedness of the VP7 gene of the G9 strains by comparing nt and deduced aa sequence of 17 strains using a computer software program, Paupsearch with 100 bootstrap repetitions (Wisconsin Genetics Computer Group programs, Version 8). Phylogenetic trees summarizing the VP7 nucleotide sequence relationship between the G9 strains were constructed using the program Treetool. Genogroup analysis To determine the origin of all the gene segments of the G9 strains, the strains were hybridized with whole genome probes derived from double-stranded RNA (dsRNA) of standard strain Wa, belonging to the Wa genogroup and strain US 1205, a representative of the recent G9 isolates from the United States, which belongs to the DS-1 genogroup. The probes were generated by labeling cDNA, derived by reverse transcription of purified dsRNA with random hexanucleotide primers and digoxigenin (DIG)-11-dUTP (Kirkwood et al., 1999b; Palombo and Bishop, 1995). Viral dsRNA was electrophoresed in a 10% gel and transferred to a nylon membrane. Prehybridization and hybridization were performed using conditions described previously (Kirkwood et al., 1999b), and the hybridized bands were detected using anti-DIGlabeled antibody conjugated to alkaline phosphatase (Boehringer-Mannheim, Indianapolis, IN) and the chemiluminescent substrate CSPD (Boehringer-Mannheim). NSP4 Genotyping The full-length PCR product of the gene coding for NSP4 protein was amplified by RT–PCR and sequenced using specific primers described previously (Cunliffe et al., 1997). Nucleotide sequence accession numbers The VP7 gene sequences submitted in this study have been deposited in the EMBL nucleotide sequence database and assigned the following accession numbers: strain US1071, AJ250268; US1111, AJ250269; US1145, AJ250270; US1206, AJ250271; US1212, AJ250272; US1343, AJ250273; US244, AJ250274; US321, AJ250275; US430, AJ250540; G16, AJ250276; INL1, AJ250277; BD426, Aj250541; BD431, Aj250542; BD524, AJ250543; MW47, AJ250544; and MW69, AJ250545. The NSP4 gene sequences submitted in this study have been assigned the following accession numbers: strain BD524, AJ400634; BD431, AJ400635; BD426, AJ400636; US1111, AJ400637; US1206, AJ400638; US1343, AJ400639; US244, AJ400640;

443

US321, AJ400641; MW47, AJ400642; MW69, AJ400643; and US430, AJ400644. ACKNOWLEDGMENTS M.R. and C.D.K. were supported by a fellowship from Atlanta Research and Education Foundation, VA Hospital (Atlanta, GA). N.P.C. is a Welcome Research Training Fellow in Clinical Tropical Medicine (Grant 049485/z/96). We thank James Gathany for assistance in photography and Ann Risinger for help with the manuscript. We thank Barbara Coulson, Ruth Beshoff, and Harry Greenberg for providing monoclonal antibodies used in this study.

REFERENCES Arista, S., Vizzi, E., Ferraro, D., Cascio, A., and Di Stefano, R. (1997). Distribution of VP7 serotypes and VP4 genotypes among rotavirus strains recovered from Italian children with diarrhea. Arch. Virol. 142, 2065–2071. Ball, J. M., Tian, P., Zeng, C. Q.-Y., Morris, A. P., and Estes, M. K. (1996). Age-dependent diarrhea induced by a rotaviral nonstructural glycoprotein. Science 272, 101–104. Bon, F., Fromantin, C., Aho, S., Pothier, P., and Kohli, E. (2000). G and P genotyping of rotavirus strains circulating in France over a three year period: Detection of G9 and P[6] strains at low frequencies. J. Clin. Microbiol. 38, 1681–1683. Brown, D. W. G., Mathan, M. M., Mathew, M., Martin, R., Beards, G. M., and Mathan, V. I. (1988). Rotavirus epidemiology in Vellore, South India: Group, subgroup, serotype, and electrophoretype. J. Clin. Microbiol. 26, 2410–2414. Burns, J. W., Majid, S.-P., Krishnaney, A. A., and Greenberg, H. B. (1996). Protective effect of rotavirus VP6-specific IgA monoclonal antibodies that lack neutralizing activity. Science 272, 104–107. Clark, H. F., Hoshino, Y., Bell, L. M., Groff, J., Hess, G., Bachman, P., and Offit, P. A. (1987). Rotavirus isolate W161 representing a presumptive new human serotype. J. Clin. Microbiol. 25, 1757–1762. Cubitt, W. D., Steele, A. D., and Iturrize, M. (2000). Characterisation of rotaviruses from children treated at a London hospital during 1996: Emergence of strains G9P2A[6] and G3P2A[6]. J. Med. Virol. 61, 150–154. Cunliffe, N. A., Woods, P. A., Leite, J. P. G., Das, B. K., Ramachandran, M., Bhan, M. K., Hart, C. A., Glass, R. I., and Gentsch, J. R. (1997). Sequence analysis of NSP4 gene of human rotavirus allows classification into two main genetic groups. J. Med. Virol. 53, 41–50. Cunliffe, N. A., Gondwe, J. S., Broadhead, R. L., Molyneux, M. E., Woods, P. A., Bresee, J. S., Glass, R. I., Gentsch, J. R., and Hart, C. A. (1999). Rotavirus G and P types in children with acute diarrhea in Blantyre, Malawi, from 1997 to 1998: Predominance of novel P[6]G8 strains. J. Med. Virol. 57, 308–312. Devereux, J., Haeberli, P., and Smithies, O. (1984). A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Res. 12, 387–395. Estes, M. K., and Cohen, J. (1989). Rotavirus gene structure and function. Microbiol. Rev. 53, 410–449. Flores, J., Perez-Schael, I., Boeggeman, E., White, L., Perez, M., Purcell, R., Hoshino, Y., Midthun, K., Chanock, R. M., and Kapikian, A. Z. (1985). Genetic relatedness among human rotaviruses. J. Med. Virol. 17, 135–143. Flores, J., Taniguchi, K., Green, K., Perez-Schael, I., Garcia, D., Sears, J., Urasawa, S., and Kapikian, A. Z. (1988). Relative frequencies of rotavirus serotypes 1, 2, 3, and 4 in Venezuelan infants with gastroenteritis. J. Clin. Microbiol. 26, 2092–2095. Garbarg-Chenon, A., Bricout, F., and Nicolas, J.-C. (1984). Study of genetic reassortment between two human rotaviruses. Virology 139, 358–365. Gentsch, J. R., Glass, R. I., Woods, P., Gouvea, V., Gorziglia, M., Flores,

444

RAMACHANDRAN ET AL.

J., Das, B. K., and Bhan, M. K. (1992). Identification of group A rotavirus gene 4 types by polymerase chain reaction. J. Clin. Microbiol. 30, 1365–1373. Gentsch, J. R., Woods, P. A., Ramachandran, M., Das, B. K., Leite, J. P., Alfieri, A., Kumar, R., Bhan, M. K., and Glass, R. I. (1996). Review of G and P typing results from a global collection of strains: implications for vaccine development. J. Infect. Dis. 174(Suppl. 1), 30–36. Gouvea, V., Glass, R. I., Woods, P., Taniguichi, K., Clark, H. F., Forrester, B., and Fang, Z. Y. (1990). Polymerase chain reaction amplification and typing of rotavirus nucleic acid from stool specimens. J. Clin. Microbiol. 28, 276–282. Green, K. Y., Hoshino, Y., and Ikegami, N. (1989). Sequence analysis of the gene encoding the serotype-specific glycoprotein (VP7) of two new human rotavirus serotypes. Virology 168, 429–433. Greenberg, H., McAuliffe, V., Valdesuso, J., Wyatt, R., Flores, J., Kalica, A., Hoshino, Y., and Singh, N. (1983). Serological analysis of the subgroup protein of rotavirus using monoclonal antibodies. Infect. Immun. 39, 91–99. Griffin, D. D., Kirkwood, C., Parashar, U. D., Woods, P. A., Bresee, J. S., Glass, R. I., Gentsch, J. R., et al. (2000). Surveillance of rotavirus strains in the United States: Identification of unusual strains. J. Clin. Microbiol. 38, 2784–2787. Horie, Y., Masamune, O., and Nakagomi, O. (1997). Three major alleles of rotavirus NSP4 proteins identified by sequence analysis. J. Gen. Virol. 78, 2341–2346. Hoshino, Y., Sereno, M. M., Midthun, K., Flores, J., Kapikian, A. Z., and Chanock, R. M. (1985). Independent segregation of two antigenic specificities (VP3 and VP7) involved in neutralization of rotavirus infectivity. Proc. Natl. Acad. Sci. USA 82, 8701–8704. Iturriza-Gomara, M., Cubitt, D., Steele, D., Green, J., Brown, D., Kang, G., Desselberger, U., and Gray, J. (2000). Characterisation of rotavirus G9 strains isolated in the UK between 1995 and 1998. J. Med. Virol. 61, 510–517. Jain, V., Clark, H. F., Dennehy, P., Zangwill, K., Kirkwood, C. D., Glass, R. I., and Gentsch, J. R. (1999). Identification of serotype G9 in archival collections: Implications for United States vaccine implementation. Abstract presented at The American Society for Virology Meeting, Amherst, MA, July 1999, W43–4. Kirkwood, C. D., and Palombo, E. A. (1997). Genetic characterization of the rotavirus nonstructural protein, NSP4. Virology 236, 258–265. Kirkwood, C. D., Masendycz, P. J., and Coulson, B. S. (1993). Characteristics and location of cross-reactive and serotype-specific neutralization sites on VP7 of human G type 9 rotaviruses. Virology 196, 79–88. Kirkwood, C. D., Gentsch, J. R., and Glass, R. I. (1999a). Sequence analysis of the NSP4 gene from human rotavirus strains isolated in the United States. Virus Genes 19, 113–122. Kirkwood, C. D., Gentsch, J. R., Hoshino, Y., Clark, H. F., and Glass, R. I. (1999b). Genetic and antigenic characterization of a serotype P[6]G9 human rotavirus strain isolated in the U.S. Virology 256, 45–53. Linhares, A. C., Gabbay, Y. B., Mascarenhas, J. D. P., de Freitas, R. B., Olivera, C. S., Bellesi, N., Monteiro, T. A. F., Lins-Lainson, Z., Ramos, F. P., and Valente, S. A. (1996). Immunogenicity, safety and efficacy of tetravalent rhesus-human, reassortant rotavirus vaccine Belem, Brazil. Bull. WHO 74, 491–500. Nakagomi, O., Nakagomi, T., Hoshino, Y., Flores, J., and Kapikian, A. Z. (1987). Genetic analysis of a human rotavirus that belongs to subgroup 1 but has an RNA pattern typical of subgroup II human rotaviruses. J. Clin. Microbiol. 25, 1159–1164.

Nakagomi, O., Nakagomi, T., Akatani, K., and Ikegami, N. (1989). Identification of rotavirus genogroups by RNA-RNA hybridization. Mol. Cell. Probes 3, 251–261. Nakagomi, T., Ohshima, A., Akatani, K., Ikegami, N., Katsushima, N., and Nakagomi, O. (1990). Isolation and molecular characterization of a serotype 9 human rotavirus strain. Microbiol. Immunol. 34, 77–82. Palombo, E. A., and Bishop, R. F. (1995). Genetic and antigenic characterization of a serotype G6 human rotavirus isolated in Melbourne, Australia. J. Med. Virol. 47, 348–354. Palombo, E. A., Masendycz, P. J., Bugg, H. C., Bogdanovic-Sakran, N., Barnes, G. L., and Bishop, R. F. (2000). Emergence of serotype G9 human rotaviruses in Australia. J. Clin. Microbiol. 38, 1305–1306. Pedley, S., Hundley, Chrystie, I., McCrae, M. A., and Desselberger, U. (1984). The genomes of rotaviruses isolated from chronically infected immunodeficient children. J. Gen. Virol. 65, 1141–1150. Pereira, H. G., Azeredo, R. S., Leite, J. P. G., Candeias, J. A. N., Racz, M. L., Linhares, A. C., Gabbay, Y. B., and Trabulsi, J. R. (1983). Electrophoretic study of the genome of human rotaviruses from Rio de Janeiro, Sao Paulo and Belem, Brazil. J. Hyg. (Cambridge) 90, 117–125. Raj, P., Bhan, M. K., Prasad, A. D. K., Kumar, R., Bhandari, N., and Jayashree, S. (1989). Electrophoretic study of the genome of human rotavirus in rural Indian community. Indian J. Med. Res. 89, 65–68. Ramachandran, M., Das, B. K., Vij, A., Kumar, R., Bhambal, S. S., Kesari, N., Rawat, H., Bahl, L., Thakur, S., Woods, P. A., Glass, R. I., Bhan, M. K., and Gentsch, J. R. (1996). Unusual diversity of human rotavirus G and P genotypes in India. J. Clin. Microbiol. 34, 436–439. Ramachandran, M., Gentsch, J. R., Parashar, U. D., Jin, S., Woods, P. A., Holmes, J. L., Kirkwood, C. D., Bishop, R. F., Greenberg, H. B., Urasawa, S., Gerna, G., Coulson, B. S., Taniguchi, K., Bresee, J. S., Glass, R. I., and The National Rotavirus Strain Surveillance System Collaborating Laboratories. (1998). Detection and characterization of novel rotavirus strains in the United States. J. Clin. Microbiol. 36, 3223–3229. Ramig, R. F. (1997). Genetics of the rotaviruses. Annu. Rev. Microbiol. 51, 225–255. Sethi, S., Olive, D., Strannegard, O., and Al-Nakib, W. (1988). Molecular epidemiology of human rotavirus infections based on genome segment variations in viral strains. J. Med. Virol. 26, 249–259. Steele, A. D., and Alexander, J. J. (1988). The relative frequency of subgroup I and II rotaviruses in black infants in S Africa. J. Med. Virol. 24, 321–327. Unicomb, L. E., Podder, G., Gentsch, J. R., Woods, P. A., Hasan, K. Z., Faruque, A. S. G., Albert, M. J., and Glass, R. I. (1999). Evidence of high-frequency genomic reassortment of group A rotavirus strains in Bangladesh: Emergence of type G9 in 1995. J. Clin. Microbiol. 37, 1885–1891. Urasawa, S., Hasegawa, A., Urasawa, T., Taniguchi, K., Wakasugi, F., Suzuki, H., Inouye, S., Pongprot, B., Supawadee, J., Suprasert, S., Rangsiyanond, P., Tonusin, S., and Yamazi, Y. (1992). Antigenic and genetic analyses of human rotaviruses in Chiang Mai, Thailand: Evidence for a close relationship between human and animal rotaviruses. J. Infect. Dis. 166, 227–234. Ward, R. L., and Knowlton, D. R. (1989). Geneotypic selection following coinfection of cultured cells with subgroup 1 and subgroup 2 human rotaviruses. J. Gen. Virol. 70, 1691–1699. Ward, R. L., Knowlton, D. R., and Hurst, P.-F. L. (1988). Reassortant formation and selection following coinfection of cultured cells with subgroup 2 human rotaviruses. J. Gen. Virol. 69, 149–162.