Rotavirus Serotype G9 Strains Belonging to VP7 Gene Phylogenetic Sequence Lineage 1 May Be More Suitable for Serotype G9 Vaccine Candidates than Those Belonging to Lineage 2 or 3

Abstract

A safe and effective group A rotavirus vaccine that could prevent severe diarrhea or ameliorate its symptoms in infants and young children is urgently needed in both developing and developed countries. Rotavirus VP7 serotypes G1, G2, G3, and G4 have been well established to be of epidemiologic importance worldwide. Recently, serotype G9 has emerged as the fifth globally common type of rotavirus of clinical importance. Sequence analysis of the VP7 gene of various G9 isolates has demonstrated the existence of at least three phylogenetic lineages. The goal of our study was to determine the relationship of the phylogenetic lineages to the neutralization specificity of various G9 strains. We generated eight single VP7 gene substitution reassortants, each of which bore a single VP7 gene encoding G9 specificity of one of the eight G9 strains (two lineage 1, one lineage 2 and five lineage 3 strains) and the remaining 10 genes of bovine rotavirus strain UK, and two hyperimmune guinea pig antisera to each reassortant, and we then analyzed VP7 neutralization characteristics of the eight G9 strains as well as an additional G9 strain belonging to lineage 1; the nine strains were isolated in five countries. Antisera to lineage 1 viruses neutralized lineage 2 and 3 strains to at least within eightfold of the homotypic lineage viruses. Antisera to lineage 2 virus neutralized lineage 3 viruses to at least twofold of the homotypic lineage 2 virus; however, neutralization of lineage 1 viruses was fourfold (F45 and AU32) to 16- to 64-fold (WI61) less efficient. Antisera to lineage 3 viruses neutralized the lineage 2 strain 16- to 64-fold less efficiently, the lineage 1 strains F45 and AU32 8- to 128-fold less efficiently, and WI61 (prototype G9 strain) 128- to 1,024-fold less efficiently than the homotypic lineage 3 viruses. These findings may have important implications for the development of G9 rotavirus vaccine candidates, as the strain with the broadest reactivity (i.e., a prime strain) would certainly be the ideal strain for inclusion in a vaccine.

Group A rotavirus is the leading cause of severe diarrhea in infants and young children worldwide and has been estimated to be responsible for a median of 440,000 deaths each year among children under 5 years of age, predominantly in the developing countries (37). In the United States alone, approximately 500,000 physician visits, 50,000 hospitalizations, and about 20 deaths are estimated to result among approximately 2.7 million children under 5 years of age who get affected by rotavirus diarrhea yearly (3, 11, 23, 37). Although 10 VP7 G serotypes and 13 VP4 P serotypes have been detected in humans thus far (23), only 4 of 10 G serotypes (G1, G2, G3, and G4) and only 1 of 13 P serotypes (P1A[8] and P1B[4]) have consistently been shown to be epidemiologically important worldwide (10, 28; N. Santos and Y. Hoshino, unpublished data). For these reasons, we have developed rhesus rotavirus (RRV)- and bovine rotavirus (UK)-based reassortant vaccines which are designed to cover G and P serotypes of epidemiologic significance (15, 17, 31, 32).

The licensure in 1998 by the U.S. Food and Drug Administration of an RRV-based quadrivalent vaccine (RotaShield; Wyeth-Lederle Vaccines and Pediatrics, Philadelphia, Pa.) (1) was a strong impetus to establish rotavirus strain surveillance programs throughout the world to monitor rotavirus strain diversity before and after vaccine introduction. Thus, today there are various national (e.g., the U.S. National Rotavirus Strain Surveillance Program and the Australian Rotavirus Surveillance Program) as well as regional (e.g., the African Rotavirus Network and the Asian Rotavirus Surveillance Network) groups that have been established to conduct studies on rotavirus epidemiology and disease burden (reviewed in reference 27). Such studies have repeatedly shown the continuing worldwide distribution and clinical importance of serotypes G1, G2, G3, and G4 and the existence of serotypes other than G1 to G4 in various parts of the world, including G5, G6, G8, G9, G10, and G12 (7, 10, 28; Santos and Hoshino, unpublished). Although the occurrence of such uncommon serotypes as G5, G6, G8, G10, and G12 is still focal, the G9 strain appears to be distributed worldwide and to be clinically important (reviewed in references 7, 28, and 40). With regard to its overall distribution in comparison to the well-established serotypes G1 to G4, an analysis of a total of 42,757 rotavirus strains collected globally from 108 studies from 50 countries on five continents that were published between 1989 and 2003 (Santos and Hoshino, unpublished) indicated the relative distribution of human rotavirus G types as follows: G1 (66.0%), G2 (12.1%), G4 (8.6%), G3 (3.5%), and G9 (2.7%). Although G9 viruses have been detected in association with a variety of P types including P[4], P[6], P[8], P[9], P[11], and P[19], the P[8],G9 has been reported to be most widely distributed worldwide (Santos and Hoshino, unpublished). Thus, in order to “be ready and prepared,” an RRV- or UK-based serotype G9 as well as G5, G8, and G10 single VP7 gene substitution reassortant vaccine candidates for potential addition to a multivalent rotavirus vaccine were recently constructed and characterized (17).

Serotype G9 rotaviruses were not recognized prior to 1983, when the first G9 strain, WI61, was isolated from an 18-month-old child with diarrhea at Children's Hospital of Philadelphia, Philadelphia, Pa. (5). In that study, 5 of 59 rotavirus-positive diarrhea stools collected during the 1983 to 1984 rotavirus season were reported to have a WI61 electropherotype. The second G9 strain, F45, was isolated in Osaka, Japan, in 1985 (N. Ikegami, K. Akatani, T. Hosaka, and H. Ushijima, Abstr. 7th Int. Congr. Virol., p. 113, 1987). An epidemiologic survey of rotavirus strains reported that a considerable number of strains similar to F45 were prevalent nationwide in Japan during the 1985 to 1986 rotavirus season (Ikegami et al., Abstr. 7th Int. Congr. Virol. 1987). However, in both the United States and Japan, soon after their detection, the G9 viruses became undetectable for about a decade. Both WI61 and F45 strains carry P1A[8] specificity. Strain 116E, which was shown to carry G9 and P8[11] specificities, was isolated from an asymptomatic neonate in India in 1986 (8, 9). In 1989, two G9 strains bearing P[19] specificity were isolated from diarrheic children in Thailand (48). The mid-1990s heralded the detection of serotype G9 rotavirus strains around the world: in India in 1993; in the United States and the United Kingdom in 1995; in Bangladesh, Japan, and Thailand in 1996; and in Malawi, Brazil, and China in 1997. Since then, the detection of the G9 viruses has been reported from a number of countries on each of the five continents (reviewed in references 7 and 29; Santos and Hoshino, unpublished).

This emergence or reemergence of the G9 strains has aroused considerable scientific interest on various aspects of this G serotype, including the epidemiology, evolutionary origin, and genetic composition of the virus as well as vaccine development against this serotype. For example, the VP7 gene of many G9 isolates has been sequenced and analyzed. Such studies have demonstrated the existence of at least three phylogenetic sequence lineages among the G9 rotavirus VP7 genes (2, 19, 25, 29, 30, 35, 40, 42, 43, 52, 53). In addition, differences in the reactivity pattern of certain G9-specific monoclonal antibodies (MAbs) against selected G9 isolates in lineages 1 to 3 were reported (6, 25, 39, 52, 53). However, a comparative systematic characterization of the VP7 neutralization specificities of G9 rotavirus strains belonging to lineage 1, 2, or 3 has not been made, which forms the aim of this study. This type of study is essential for choosing the best G9 strain (i.e., the strain with the broadest reactivity) as a vaccine candidate.

MATERIALS AND METHODS

Rotavirus strains, cell cultures, culture medium, virus titration, neutralization assay, and hyperimmune antiserum.

Table 1 summarizes the rotavirus strains employed in this study. Figure 1 depicts the phylogenetic VP7 sequence lineage of each of the nine G9 strains analyzed. Three lineage 1 viruses, each of which bore P[8] specificity, were strain WI61 from the United States (5), and strains F45 (Ikegami et al., Abstr. 7th Int. Congr. Virol. 1987) and AU32 (33) from Japan. Strain 116E, which was the only lineage 2 virus available, was isolated in India and carried P[11] specificity (8, 9). Each of the lineage 1 and 2 strains in this study displayed a “long” electropherotype (41). Five lineage 3 viruses were strains R44 and R143 from Brazil, which bore P[9] and P[6] specificity, respectively (43); strain US1205 from the United States, carrying P[6] specificity (39); strain INL1 from India bearing P[6] specificity (39); and strain BD524 from Bangladesh which bore P[8] specificity (47). Strains R44, INL1, and BD524 had a “long” electropherotype, whereas strains R143 and US1205 had a “short” electropherotype (an electropherotype displaying slow-moving gene segments 10 and 11) (22). The earliest detected G9 strain (WI61) was isolated in 1983, and the most recently detected G9 strain (R143) was isolated in 1999. Single VP7 gene substitution rotavirus reassortant D × UK (G1,P7[5]), used in the present study in the construction of the G9 × UK virus reassortants as described below, was an experimental vaccine suspension (lot HDBRV-1 [i.e., D × UK]) prepared in fetal rhesus monkey lung diploid cell strain (FRhL-2) cultures (31). Primary cultures of African green monkey kidney (AGMK) cells (Diagnostic Hybrids, Athens, Ohio) or an established monkey kidney MA104 cell line was used for virus amplification, genetic reassortment, and plaque purification. The MA104 cell line was used for virus titration and plaque reduction neutralization (PRN) assay. Eagle's minimum essential medium supplemented with 0.5 μg of trypsin (γ-irradiated trypsin; Sigma Chemical, St. Louis, Mo.) per ml and antibiotics was used as maintenance medium, and Leibovitz L-15 medium (Quality Biological, Gaithersburg, Md.) supplemented with antibiotics was employed when making virus or serum dilutions. The PRN assay was performed by using 50 to 60 PFU per 250 μl of the virus as described previously (16). Agarose (SeaKem ME; BMA, Rockland, Maine) was used as a solidifying reagent in the plaque assay. Hyperimmune antiserum to each of the reassortants was raised in specific-pathogen-free guinea pigs (Charles River, Wilmington, Mass.) which were free of rotavirus-neutralizing antibodies (titer of <1:20 versus AU32) as determined by PRN assay. Rotavirus immunogens were prepared as previously described (50). The intramuscular immunization was performed first with Freund complete adjuvant followed (2 weeks later) by two immunizations (2 weeks apart) with Freund incomplete adjuvant and finally, 2 weeks after the third immunization, with immunogen without adjuvant. Seven days after the fourth immunization, guinea pigs were euthanized, and blood was collected. Sera were inactivated before use by heating at 56°C for 30 min.

TABLE 1.

Antigenic characterization of outer capsid glycoprotein VP7 of selected G9 rotavirus strains belonging to VP7 gene phylogenetic sequence lineage 1, 2, or 3

Strain	Rotavirus characteristics						Reciprocal of 60% PRN antibody titer of guinea pig hyperimmune antiserum to indicated rotavirus reassortant (VP7 gene lineage)b
	Country of origin	Year collected	Type		E typea	VP7 gene lineage	UK × WI61 (1)	UK × WI61 (1)	UK × AU32 (1)	UK × AU32 (1)	UK × 116E (2)	UK × 116E (2)	UK × R44 (3)	UK × R44 (3)	UK × R143 (3)	UK × R143 (3)	UK × US1205 (3)	UK × US1205 (3)	UK × INL1 (3)	UK × INL1 (3)	UK × BD524 (3)	UK × BD524 (3)
			P	G
WI61	U.S.	1983	[8]	9	L	1	20,480	20,480	20,480	40,960	1,280	160	80	<80	640	640	<80	80	<80	80	<80	160
DS-1 × WI61 reassortant			[4]	9	S	1	10,240	20,480	20,480	40,960	640	160	160	<80	640	640	<80	<80	80	80	80	160
F45	Japan	1985	[8]	9	L	1	10,240	40,960	40,960	40,960	5,120	2,560	640	2,560	2,560	5,120	640	320	2,560	2,560	320	5,120
AU32	Japan	1986	[8]	9	L	1	40,960	40,960	81,920	81,920	5,120	2,560	1,280	2,560	2,560	2,560	320	320	2,560	2,560	640	5,120
116E	India	1986	[11]	9	L	2	10,240	10,240	20,480	20,480	20,480	10,240	640	640	2,560	2,560	640	640	1,280	2,560	640	2,560
R44	Brazil	1997	[9]	9	L	3	40,960	20,480	40,960	40,960	10,240	10,240	10,240	20,480	40,960	81,920	10,240	10,240	20,480	81,920	40,960	40,960
R143	Brazil	1999	[6]	9	S	3	40,960	20,480	81,920	81,920	20,480	10,240	20,480	20,480	81,920	81,920	10,240	20,480	20,480	40,960	40,960	81,920
US1205	U.S.	1996-1997	[6]	9	S	3	40,960	20,480	40,960	40,960	10,240	10,240	10,240	10,240	40,960	81,920	20,480	10,240	20,480	40,960	81,920	81,920
INL1	India	1994	[6]	9	L	3	40,960	10,240	81,920	81,920	10,240	20,480	10,240	20,480	81,920	81,920	20,480	20,480	40,960	81,920	40,960	81,920
BD524	Bangladesh	1995-1996	[8]	9	L	3	10,240	10,240	10,240	10,240	10,240	10,240	10,240	10,240	40,960	81,920	10,240	10,240	40,960	81,920	40,960	>81,920
UK	UKd	1975	[5]	6	L	NAc	2,560	640	320	640	320	640	640	640	160	640	640	320	640	640	2,560	640

^aAntigenic characterization for strains was done by analysis of their guinea pig hyperimmune antiserum neutralization profile. E type; electropherotype; L, long; S, short.

^bThe homologous 60% PRN antibody titers ranged from 20,480 to >81,920 in previous tests in this laboratory. Values belonging to the same VP7 gene sequence lineage are in boldface type. VP7 homologous values are underlined.

^cNA, not applicable.

^dUnited Kingdom.

FIG. 1.

Phylogenetic comparison of the VP7 nucleotide sequences of the G9 strains employed in this study determined by the unweighted pair group method with arithmetic mean program. Numbers above branches are bootstrap nucleotide percentage values. The GenBank accession numbers for the VP7 sequence are as follows: 116E, L140720; WI61, see reference 12; F45, see reference 12; AU32, AB045372; INL1, AJ250277; BD524, AJ250543; R44, AF438227; US1205, AF060487; and R143, AF274969.

Construction, identification, and characterization of single G9 VP7 gene substitution rotavirus reassortants.

Roller tube cultures of primary AGMK or MA104 cells were coinfected at a multiplicity of infection of approximately 1 with human rotavirus strain WI61, 116E, R44, R143, US1205, INL1, or BD524 and the reassortant rotavirus D × UK. When approximately 75% of the infected cells exhibited cytopathic effects, the cultures were frozen and thawed once, and the lysate was plated onto primary AGMK or MA104 cells in a 6-well plate (Costar; Corning Inc., Corning, N.Y.) in the presence of G1-specific VP7 neutralizing MAb 2C9 (44) for selection of the desired UK × G9 (P7[5],G9) reassortants. Coinfection of the DS-1 (P1B[4],G2) strain and WI61 strain was also performed at a multiplicity of infection of approximately 1. The infected cell culture lysate was plated onto MA104 cells in a 6-well plate in the presence of G2-specific neutralizing MAb S2-2G10 (46) for selection of the desired DS-1 × WI61 (P1B[4],G9) reassortant. Each of the desired single VP7 gene substitution reassortants was selected and identified and then plaque purified three times. The origin of genes of each reassortant was identified by polyacrylamide gel electrophoresis of its genomic RNAs. The origins of certain genes which were not able to be determined with certainty by polyacrylamide gel electrophoresis were studied further by constant denaturant gel electrophoresis as previously described (21). Generation and characterization of UK × AU32 reassortant vaccine candidate was reported previously (17). Hyperimmune guinea pig antiserum to each reassortant was analyzed for VP7-specific antibodies to selected human and animal rotavirus strains by 60% PRN assay (16, 50).

RESULTS

Generation of eight single VP7 gene substitution rotavirus reassortants and guinea pig hyperimmune antiserum to each of the eight reassortants.

Since the interaction of VP4-VP7 outer capsid proteins of rotavirus has been reported to affect the expression of selected phenotypes of one or both proteins (4, 26, 38), we constructed eight single VP7 gene substitution rotavirus reassortants, each of which had 10 genes of bovine rotavirus strain UK and only the VP7 gene of human G9 rotavirus strain WI61, AU32, 116E, R44, R143, US1205, INL1, or BD524 (Fig. 2). Thus, each UK × G9 rotavirus reassortant carried an identical genetic and antigenic configuration, except for outer capsid glycoprotein VP7. We also generated guinea pig hyperimmune antiserum to each reassortant for use in neutralization characterization of the VP7 protein of each of nine G9 rotavirus strains (three lineage 1, one lineage 2 and five lineage 3 strains) (Table 1). At least two guinea pigs were used for antiserum production against each reassortant in an attempt to obtain a more accurate neutralization profile.

FIG. 2.

Electrophoretic migration patterns of genomic RNAs of human rotavirus strain WI61 (lane 1), reassortant UK × WI61 (lane 2), reassortant D × UK (lane 3), reassortant UK × AU32 (lane 4), human rotavirus strain AU32 (lane 5), human rotavirus strain 116E (lane 6), reassortant UK × 116E (lane 7), reassortant D × UK (lane 8), reassortant UK × R44 (lane 9), human rotavirus strain R44 (lane 10), human rotavirus strain R143 (lane 11), reassortant UK × R143 (lane 12), reassortant D × UK (lane 13), reassortant UK × US1205 (lane 14), human rotavirus strain US1205 (lane 15), human rotavirus strain INL1 (lane 16), reassortant UK × INL1 (lane 17), reassortant D × UK (lane 18), reassortant UK × BD524 (lane 19), and human rotavirus strain BD524 (lane 20) in 10% polyacrylamide gel. Genomic RNAs were electrophoresed at 13 mA for 15 h, and the resulting migration patterns were visualized by staining the gel with silver nitrate.

Neutralization capabilities of guinea pig hyperimmune antisera raised against each of the eight UK × G9 single gene substitution reassortants.

Table 1 summarizes the antigenic characterization of outer capsid glycoprotein VP7 of each of the eight G9 rotavirus strains belonging to VP7 gene phylogenetic sequence lineage 1, 2, or 3 by analysis of their guinea pig hyperimmune antiserum neutralization profile. Guinea pig hyperimmune antisera raised against the VP7 protein of strain WI61 or AU32 which belonged to lineage 1 neutralized lineage 2 and 3 strains to at least within eightfold of the homotypic lineage 1 viruses, indicating that the VP7 proteins of all nine G9 strains analyzed were closely related antigenically at least in this one-way direction. The extent of the relatedness among the three lineages was analyzed further by examining the reciprocal neutralization specificities; antisera to the lineage 2 strain 116E VP7 neutralized lineage 1 virus AU32 to within fourfold of the homotypic value, indicating that these lineage 1 and 2 viruses were similar, if not identical, in both directions, as determined by the 20-antibody rule for relatedness (14). However, antisera to lineage 2 strain 116E VP7 neutralized lineage 1 prototype strain WI61 16- to 64-fold less efficiently than the homotypic strain, indicating that the VP7 of the lineage 2 virus was distantly related to or antigenically distinct from the WI61 strain, suggesting that WI61 could be considered the prime strain (at least with the results from one of the two guinea pig sera). Antisera to each of the lineage 3 virus VP7s neutralized the WI61 lineage 1 strain 128- to 1,024-fold less efficiently than the homotypic strains, indicating that the WI61 strain was the prime strain for each of the five lineage 3 strains. In addition, antisera to four of five lineage 3 strain VP7s neutralized the AU32 and F45 lineage 1 strains 16- to 128-fold less efficiently, indicating that the AU32 and F45 strains were the prime strains for certain lineage 3 strains. The lineage 3 R44 VP7 antisera neutralized the AU32 strain within eightfold of the homotypic strain, indicating a two-way antigenic relationship.

Reassortant DS-1 × WI61 (P1B[4],G9) was generated for use in analyzing whether the VP4 protein (P1A[8]) of the WI61 strain affected the observed reactivity of the WI61 virus against each of the 16 guinea pig hyperimmune antisera tested in neutralization. The reassortant DS-1 × WI61 was shown to behave in neutralization similar to the parental WI61 virus (Table 1), indicating that the observed neutralization characteristics of the WI61 were unique to the VP7 of WI61.

It was noteworthy that antisera to lineage 2 strain 116E VP7 neutralized lineage 3 viruses to at least twofold of the homotypic virus, suggesting a close antigenic relationship in this direction. However, in the reciprocal analysis, antisera to the lineage 3 virus VP7s neutralized the lineage 2 virus 116E 16- to 64-fold less efficiently, indicating that the lineage 2 virus was the prime strain for certain lineage 3 viruses.

Finally, antisera to each of the lineage 3 virus VP7s neutralized lineage 3 viruses to a high titer regardless of the country of origin (Bangladesh, Brazil, or the United States) or year of isolation (1994 to 1999), indicating a high degree of antigenic relatedness in both directions and serotypic identity by neutralization.

DISCUSSION

Among 10 rotavirus G serotypes found in humans (23), serotype G9 displays various unique characteristics, examples of which are as follows. (i) Soon after its detection in the United States (in 1983) and in Japan (in 1985), the G9 viruses were not reported to occur in either country for about a decade and reemerged in the mid-1990s, and today they represent the fifth most common G serotype of clinical importance throughout the world. (ii) The VP7 gene of many G9 rotavirus isolates has been sequenced and reported to belong to one of at least three VP7 phylogenetic sequence lineages. (iii) Unlike other globally common G1, G2, G3, or G4 strains, which are detected almost exclusively in conjunction with P[8] or P[4], G9 strains have been detected in association with a variety of P types, including P[4], P[6], P[8], P[9], P[11], and P[19]. The G9 strains employed in this study, for example, bore either P[6] (three isolates), P[8] (four isolates), P[9] (one isolate), or P[11] (one isolate) specificity. In addition, the G9 strains have been isolated in association with a variety of electropherotypes (short or long pattern) and subgroup specificities (I or II). Rapid global spread and successful establishment in various communities of the G9 serotype may be related to this unique antigenic “promiscuity.”

Previously, using enzyme immunoassay (EIA), Coulson et al. (6) examined three lineage 1 (WI61, F45, and AU32) and one lineage 2 (116E) G9 strains against a panel of nine MAbs raised to lineage 1 G9 virus (WI61 or F45) VP7 and reported that only one of nine MAbs reacted to a high titer with the 116E virus. Thus, the 116E strain was suggested to be a monotype or a subtype of lineage 1 G9 strains. Later, two additional MAbs to the WI61 or F45 VP7 which reacted in EIA with both lineage 1 and lineage 2 viruses were reported to also react with selected lineage 3 G9 strains isolated in the United States (39). More recently, using the same set of G9- and G4-specific MAbs, Kirkwood et al. (25) reported that Australian G9 lineage 3 viruses could be grouped into three monotypes. Using EIA, Zhou et al. (53) examined two lineage 1 Japanese and six lineage 3 Thai G9 strains against three G9-specific MAbs (one raised to lineage 1 F45 and the other two raised to lineage-unknown Thai G9 strain CM17-1) and reported that two lineage 1 strains (F45 and J830 [TK830]) reacted with all three MAbs, whereas six lineage 3 strains reacted with two MAbs raised to lineage-unknown G9 strain but not with MAb raised to the lineage 1 F45 strain. Thus, these studies have shown that although all the G9 strains tested share common antigens, subtle differences in antigenic composition of the VP7 protein may exist among lineage 1, 2, and 3 strains. Indeed, although the VP7 proteins of G9 strains employed in this study, for example, were highly related to each other (>90.8% overall amino acid identity), a comparison of the amino acid sequences in the VP7 variable regions (VRs) (some of which have been reported to be antigenic sites through epitope mapping studies [reviewed in reference 23]) demonstrates the existence of certain amino acid substitution(s) in such region(s) among lineage 1, 2, and 3 viruses or even in strains within the same lineage. For example, the U.S. lineage 1 strain WI61 differs from the Japanese lineage 1 strains F45 and AU32 in VR4 (D) at amino acids (aa) 70, 73, and 74 and strain AU32 in VR7 (B) at aa 149 and from lineage 2 strain 116E in VR4 (D) at aa 68, 73, and 74; in VR5 (A) at aa 100; in VR7 (B) at aa 142 and 145; in VR8 (C) at aa 220 and 221; and in VR9 at aa 242 (Fig. 3). However, the effect of such amino acid substitution(s) on the qualitative and quantitative nature of anti-VP7 antibodies as a whole has not been explored.

FIG. 3.

Comparison of the deduced amino acid sequence of the VP7s of nine serotype G9 strains employed in the present study. Three VRs (VR1 to VR3) which are not antigenic sites are marked in black boxes. Six VRs (VR4 to VR9) and seven amino acid residues which have been demonstrated to be involved in the formation of antigenic sites (shown as letters in parentheses) through epitope mapping studies (reviewed in reference 23) are marked in red boxes.

A titer obtained by a rotavirus neutralization assay is a sum of neutralization titer due to anti-VP4 neutralizing antibodies and neutralization titer due to anti-VP7 neutralizing antibodies, because VP4 and VP7 are independent neutralization antigens (13, 18, 34). For example, the neutralization titer of anti-WI61 antiserum is a sum of neutralization titers due to anti-WI61 VP4 (P1A[8]) and anti-WI61 VP7 (G9) antibodies, and the neutralization titer of anti-116E antiserum is a sum of neutralization titers due to anti-116E VP4 (P8[11]) and anti-116E VP7 (G9) antibodies. Therefore, in order to characterize the qualitative and quantitative nature of anti-VP7 antibodies of each of the eight G9 strains employed in this study, the neutralization effect of anti-VP4 antibodies needs to be eliminated. Thus, we first constructed eight single VP7 gene substitution reassortants, each of which bore a single VP7 gene encoding G9 specificity of one of the eight G9 strains (two lineage 1, one lineage 2, and five lineage 3 strains) and the remaining 10 genes of the bovine rotavirus UK strain and then generated two guinea pig hyperimmune antisera to each reassortant. The UK strain (49) was chosen because neutralization specificities on the UK VP4 were quite distinct from those on human rotavirus VP4s of epidemiologic importance (15). Each hyperimmune antiserum thus generated contained antibodies to the identical UK VP4 (P7[5]) and antibodies to the VP7 of one of the eight G9 strains. With such antisera, it became possible to characterize the true nature of neutralization specificities on the VP7 protein of each of the eight G9 strains.

As stated earlier, since overall amino acid identity of the VP7 of the eight G9 strains analyzed in this study was >90.8%, it was surprising to find that the nature and spectrum of anti-VP7 neutralizing antibodies generated by each of the eight G9 strains were characteristically different (i.e., genotype-specific neutralization profiles were observed). Unexpectedly, antibodies to lineage 1 VP7 exhibited the broadest cross-reactivity by neutralizing all nine G9 rotavirus strains to high titers, regardless of their VP7 phylogenetic lineages. Thus far, the G9 strains belonging to lineage 1 have been isolated only in the United States and Japan. Interestingly, however, in both countries, strains bearing the lineage 1 VP7 have not been detected during the last 7 to 10 years. Instead, all the G9 strains isolated during that time period in both countries have been reported to bear lineage 3 VP7. More recently, however, the reemergence of the lineage 1 G9 viruses has been reported in Japan (52). Regarding the lineage 2 G9 VP7, it has never been detected outside of India, which may suggest its distinct evolutionary lineage, different from the lineage 1 or 3 VR7. Indeed, its distinctness is also reflected in the unique amino acid substitutions of the VP7 (e.g., at aa 142 and 145 in VR7 [B]). The origin of the lineage 3 G9 VP7, which is carried by rotavirus strains detected throughout the world since the mid-1990s and which is the predominant G9 strain causing diarrhea today, is unknown. Recently, by analyzing a total of 40 G9 strains isolated in the United States (1996 to 2001) and in India (1993 and 1996 to 1998) using various assays, Laird et al. (29) reported that the lineage 3 G9 strains were detected as early as 1993 in Indian neonates who shed viruses asymptomatically.

Two (F45 and AU32) of three lineage 1 strains tested were isolated in Japan during the same rotavirus season; the F45 was isolated in 1985 in Osaka and the AU32 was isolated in 1986 in Yamagata. The two cities are approximately 1,200 km (750 miles) apart. It is of interest that the F45 and the AU32 strains share an identical VP7 amino acid sequence between residues 1 and 75, including mutations at aa 22, 37, 70, 74, and 75 relative to the WI61 VP7 (lineage 1 U.S. isolate), whereas the WI61 shares an identical amino acid sequence with the F45 (but not with the AU32) between residues 76 and 326. Yet the F45 exhibited a neutralization profile almost identical to that of the AU32 against antisera to lineages 2 and 3 viruses, which was different from the neutralization profile demonstrated by the WI61.

“Antigenic drift” of a virus is thought to be driven by specific immunologic pressures exerted by the susceptible hosts. The antihemagglutinin antibodies to type A influenza viruses, for example, are the driving force to generate influenza viruses bearing the mutant hemagglutinin (antigenic drift in hemagglutinin) (45). It is well known that antibodies to parental influenza A viruses, from which variant influenza A viruses emerge through antigenic drift, are no longer capable of neutralizing the mutant viruses efficiently. Thus, each year, when specific mutant influenza A viruses are suspected to become prevalent in humans, a new vaccine that is targeted to such mutant influenza viruses needs to be produced. In general, a vaccine constructed against a mutant influenza virus can induce antibodies capable of efficiently neutralizing not only the mutant influenza viruses but also their parental influenza viruses. Of note is the finding in this study that antibodies to the VP7 of lineage 1 viruses (the oldest G9 strains to be isolated chronologically) neutralized the lineage 3 viruses (which emerged 10 or more years after the first detection of lineage 1 viruses) as efficiently as the homologous viruses. Surprisingly, however, antibodies to each of the five lineage 3 virus VP7s did not neutralize the lineage 1 viruses efficiently. Thus, it is unlikely that the lineage 3 VP7 evolved from lineage 1 VP7 through antigenic drift. Rather, the lineage 1 and 3 VP7s are more likely to have independent evolutionary origins. This assumption is consistent with the hypothesis that the contemporary G9 strains belonging to lineage 3 may have evolved from an unknown progenitor strain different from that of lineage 1 G9 stains (reviewed in reference 29). It is of interest that serotype G9 rotavirus strains have been detected in other animal species, including pigs and sheep (23). A possible involvement of such animal G9 strains in the evolution of human G9 strains of lineages 1, 2, and 3 is being investigated in our laboratory.

Comparative sequence analysis of the VP7 gene of selected field isolates of epidemiologic importance has demonstrated the existence of genetic variation and possible genetic drift of the VP7 genes relative to G1, G2, G3, and G4 strains employed to construct RRV-based quadrivalent vaccine (RotaShield, the first licensed rotavirus vaccine) (reviewed in reference 36). Jin et al. (20) analyzed neutralization capabilities of serum samples collected from infant vaccinees who received RRV-based quadrivalent or monovalent (G1) vaccine against prototype G1 strain Wa (which had a VP7 sequence almost identical to that of the strain D in the vaccine) and against a G1 strain (which by sequence analysis was shown to belong to a phylogenetic lineage different from that of strain D) isolated from vaccine “failures,” i.e., vaccinees who experienced diarrheal episodes. They reported that postvaccination serum neutralizing antibody titers to the Wa strain were significantly higher than those to the circulating G1 strain. Although no correlation was found between neutralizing antibody titers to the clinical G1 isolate and protection against diarrhea in infants who got infected after vaccination, that study suggested the presence of antigenic differences between the circulating G1 strain (isolated during the 1991 to 1992 season) and the Wa and the vaccine strain D (both of which were isolated in 1974). Systematic comparison between the VP7 antigenic characteristics of G1 to G4 strains present in various existing vaccine candidates and those of the more contemporary circulating G1 to G4 strains needs to be explored.

As stated earlier, serotype G9 has recently established itself as the fifth globally important rotavirus G type. For example, during the 1998 to 1999 season in Tokyo and Sapporo, Japan, the G9 strains were the most prevalent type, with an incidence of 52.9 and 71.4%, respectively (51); and during the 2001 to 2002 season in Australia, serotype G9 was the most prevalent type nationwide (40.4%), followed by G1 (38.9%) (24). Although the parameters of protection against rotavirus disease have not been firmly established, it appears that type specificity of antirotavirus antibodies plays an important role in protection against rotavirus disease; and thus, a G9 component needs to be included in the formulation of an effective rotavirus vaccine. This study has provided a basis regarding the construction of G9-specific rotavirus vaccine candidates; the VP7 of a G9 strain capable of inducing the most broadly reactive neutralizing antibodies should be utilized for the development of G9-specific vaccine candidates. In this study, we found that antibodies to lineage 1 VP7 neutralized not only the lineage 1 strains but also the single lineage 2 strain and each of the five lineage 3 strains to high titers, whereas antibodies to the lineage 2 and 3 VP7s were not as efficient in consistently neutralizing the lineage 1 strains. In several instances, lineage 1 strains were found to represent the clear prime strains (i.e., 20 U of antibody to the lineage 1 strain would recognize the lineage 2 or 3 strains but not vice versa). Thus, it is important to include a lineage 1 strain as the G9 representative in a multivalent rotavirus vaccine. The recently constructed G9 vaccine candidates bear the lineage 1 VP7 gene (i.e., UK × AU32 and RRV × AU32) (17). Safety and immunogenicity as well as efficacy of such candidate vaccines against lineage 2 and 3 G9 viruses remain to be evaluated. In addition, it would be interesting to analyze sera from infants or young children who underwent natural primary G9 rotavirus infections by neutralization against G9 viruses belonging to lineage 1, 2, or 3 to determine if such infections yield similar or differing responses to those described in this study with hyperimmune guinea pig sera.