Unexpected Detection of Animal VP7 Genes among Common Rotavirus Strains Isolated from Children in Mexico

Abstract

In the course of characterizing 103 rotaviruses from children in Mexico, we found that the majority of strains were globally common types (55.4% of total), while uncommon types represented 5.7%, mixed infections with common types represented 14.8%, and partially or fully nontypeable isolates represented about 24%. Serotype G9 was detected for the first time in Mexico. We sequenced a subset of strains that were G nontypeable by reverse transcriptase PCR and found surprisingly that two strains having common human rotavirus P genotypes (8 and 6) had serotype G3 and G4 VP7 gene sequences that shared closer homology with canine and porcine strains, respectively, than with human strains, suggesting that these isolates represented reassortants between human and animal rotaviruses.

Group A human rotaviruses are the most common etiologic agent of severe diarrhea in infants and young children throughout the world and are an important cause of gastroenteritis in a wide range of animals (13). The rotavirus genome consists of 11 double-stranded RNA segments, and the mature virus particle contains two neutralizing capsid proteins, VP4 and VP7, which determine P and G types, respectively (4). Although rotaviruses have a wide host range, strains appear to be host restricted, and direct transmission from animals to humans has never been documented (13). However, the segmented nature of the genome makes reassortment between strains during mixed infections possible, and the isolation in humans of unusual strains containing gene segments of both human and animal origin, or exclusively of animal origin, suggests that interspecies transmission and reassortment of rotaviruses do occur in nature (20).

Fourteen G serotypes and 16 P serotypes (including three subtypes) have thus far been identified, and of these, 10 G types and 10 P types have been detected in humans. A 15th G serotype and an additional unique P genotype have been proposed based on sequence information of a bovine rotavirus (24). While some G serotypes appear to be host restricted, 9 of the 10 G types known to infect humans have also been isolated from at least one animal species (13). Only serotypes G1 to G4 traditionally have been considered as important causes of diarrhea in humans worldwide (6), although in the last decade, the intensification of rotavirus surveillance, together with the development of better characterization methods for typing previously nontypeable strains, has resulted in the detection of more G types in humans (e.g., G5, G6, G8, and G10) that were previously thought to be restricted to animals (for a listing of all of these serotypes, see recent reviews [4, 13]). Serotype G5 has most commonly been isolated from pigs, while serotypes G6, G8, and G10 are commonly recognized as pathogens in cattle (13). It was originally believed that animal rotaviruses did not infect humans under natural conditions (12), but these findings suggest that interspecies transmission of rotaviruses from one animal species to another, either as whole virions or by gene segment reassortment, may be taking place in nature at a relatively high frequency, especially in developing countries, where mixed infections are more common and humans and animals often live in close physical contact (11, 14, 25, 26).

The increasing detection of strains containing G serotypes that were previously believed to be restricted to animal hosts also raises questions as to the origins of common G1 to G4 serotype strains isolated from humans (13). Serotypes G1 to G4 infect animals as well as humans (13), but since such strains are the expected outcomes of rotavirus surveillance in human populations, they are crudely lumped together as belonging to the common group of humans strains and are thoroughly characterized and studied less often than the unusual G5, G6, G8, and G10 strains, which are only occasionally isolated from humans. It is possible that many G1 to G4 isolates, rather than being closely related to other human strains of the same serotype, are animal strains that have recently crossed the species barrier to infect humans or are the products of reassortment events between animal and human strains. Such interspecies transmission events have been documented previously, as illustrated by the isolation from humans of serotype G3 strains that have genomes of partial or complete canine and feline origin (21, 22).

In the present study, during the course of characterizing the P and G types of rotavirus strains isolated from children in Mexico, we unexpectedly encountered two such rotavirus strains having the common human serotypes G3 and G4 but whose VP7 genes belonged to canine and porcine lineages, respectively. Our findings contribute to a growing body of evidence suggesting that interspecies transmission and reassortment events may be occurring at a much higher rate than previously thought and that they may not be limited to the G serotypes rarely seen in humans, such as G5, G6, G8, and G10.

We performed P and G genotyping of rotavirus strains gathered from children less than 5 years of age with diarrhea in Mexico City during the periods from 1988 to 1991 and 1998 to 2002. Stool specimens were received at the National Institute of Nutrition, Salvador Zubiran, in Mexico City, Mexico, and a total of 103 specimens tested positive for rotavirus, using a commercial rotavirus detection kit (Rotaclone; Meridian Diagnostics, Inc., Cincinnati, Ohio). Rotavirus RNA was manually extracted from 200 μl of clarified 20% (wt/vol or vol/vol) fecal suspensions by use of a NucliSens Extractor kit (Organon Teknika Corporation, Durham, N.C.) as described previously (1). All samples were analyzed to determine G and P genotypes by use of a multiplex, seminested reverse transcription (RT)-PCR protocol similar to those described previously and primers designed to detect G genotypes G1 to G4 and G9 and P genotypes P[4], P[6], P[8], P[9], and P[10] (3, 5, 7). For a subset of the strains whose G type could not initially be determined by these methods (39 of 61 nontypeable isolates), the VP7 gene was amplified by RT-PCR using previously described degenerate versions (9) of Beg9 and End9 primers (7) and sequenced on an ABI377 automated sequencer (Applied Biosystems Inc., Foster City, Calif..). The G type was then determined by comparison of the sequence to other rotavirus VP7 gene sequences available in GenBank (8). The vast majority of the 39 G-nontypeable strains sequenced turned out to be serotype G1 (n = 30), although G2 (n = 4), G3 (n = 3), and G4 (n = 2) were also found. Although the reason why these strains were nontypeable by RT-PCR genotyping has not been determined, all of the G1 strains had four sequence mismatches with the G1 typing primer used, which may explain in part or whole why many G1 strains could not be typed by RT-PCR. The G2 strains, however, had no mismatches with the typing primer, so other factors, such as residual stool inhibitors in the RNA extracts and/or low RNA concentrations in some samples, may also be responsible for some strains being nontypeable. We speculate that the remaining 22 G-nontypeable strains that were not sequenced may have failed to type for similar reasons. These issues will need to be addressed in a future study by redesigning the appropriate typing primers and retesting the 61 strains that were originally nontypeable.

Our results showed that globally common rotaviruses—P[8],G1, P[4],G2, P[8],G3, and P[6],G9—were in the majority, representing 55.4% of the 103 rotavirus isolates, while strains with common G types but uncommon P-G combinations—P[6],G3, P[6],G4, and P[8],G2—represented almost 6% of the strains characterized (Table 1). An additional 14.8% of the strains were from mixed infections with common G types and, for the most part, common P types. For the first time in Mexico, serotype G9 was detected, appearing in one isolate in combination with genotype P [6] and in a mixed infection having a G-P combination of P[6+8],G1+9, providing further support for the identification of serotype G9 as a globally common serotype.

TABLE 1.

Distribution of rotavirus types isolated from children in Mexico between 1988 and 2001^a

Strains	G type	P type	No. (%) of strains
			In 1988-1991	In 1998-2001	Total
Common	G1	P[8]	9	32	41 (39.8)
	G2	P[4]	0	12	12 (11.7)
	G3	P[8]	2	1	3 (2.9)
	G9	P[6]	0	1	1 (1.0)
Uncommon	G2	P[8]	1	1	2 (1.9)
	G3	P[6]	2	0	2 (1.9)
	G4	P[6]	1	1	2 (1.9)
From mixed infections	G1	P[6 + 8]	0	1	1 (1.0)
	G1	P[4 + 8]	0	1	1 (1.0)
	G1 + 2	P[4 + 8]	0	1	1 (1.0)
	G1 + 2 + 3	P[8]	1	0	1 (1.0)
	G1 + 4	P[8]	0	1	1 (1.0)
	G1 + 3 + 4	P[8]	0	2	2 (1.9)
	G1 + 9	P[6 + 8]	0	1	1 (1.0)
	G2	P[4 + 8]	0	5	5 (4.9)
	G2	P[4 + 6]	0	1	1 (1.0)
	G2 + 4	P[4]	0	1	1 (1.0)
Partially or fully nontypeable	G1	P[NT]	1	1	2 (1.9)
	G4	P[NT]	1	0	1 (1.0)
	GNT	P[8]	2	11	13 (12.6)
	GNT	P[4 + 8]	0	3	3 (2.9)
	GNT	P[6 + 8]	0	2	2 (1.9)
	GNT	P[NT]	0	4	4 (3.9)
Total			20	83	103

^aNT, not typeable.

Although our typing results showed a distribution of G and P types in Mexico similar to that in other countries (6), the VP7 gene of 2 of the 39 strains analyzed by sequencing produced unexpected results (Table 2). The fully sequenced VP7 gene of the P[8],G3 strain, 7177-1042, demonstrated the closest sequence homology (91.9% on the nucleotide level and 99.1% on the amino acid level) to Italian canine isolate RV198/95 (17) and only a slightly lower homology to the unusual human isolate HCR3. Strain HCR3 was isolated from a healthy infant in the United States in 1984 (15) and was shown by RNA-RNA hybridization to form 11 hybrid bands with canine and feline rotavirus strains but not with common human rotavirus strains, suggesting that interspecies transmission of whole virions to humans had occurred (22). One other G3 strain, Ro1845, that was isolated from a child with diarrhea in Israel in 1985 (19), also shared high homology by RNA-RNA hybridization with feline and canine strains (21). Unlike strains HCR3 and Ro1845, which have the rare P serotype P5A[3] (13, 16, 18), isolate 7177-1042 had the globally most common human P genotype, P[8]. Thus, the finding that 7177-1042 VP7 had higher homology to animal strains than to human rotavirus P[8],G3 strains was a surprise. It will be interesting to determine whether 7177-1042, like HCR3 and Ro1845, has other genes of animal origin in addition to its VP7 gene. Since strain 7177-1042 has a typical human rotavirus VP4 gene (P[8]), it may not represent an example of transmission of whole virions from animals to humans, as is believed to be the case for HCR3 and Ro1845. Sequencing of the rest of the genome of 7177-1042 or analysis by RNA-RNA hybridization will be needed in order to test this hypothesis that strain 7177-1042 may have other genes related to canine and/or feline rotaviruses.

TABLE 2.

VP7 nucleotide and amino acid similarity between Mexican rotavirus isolates and other human strains of animal and/or human origin

Strain	% Similarity witha:
	D151 (G4)		7177-1042 (G3)
	nt	aa	nt	aa
Wa (G1)	74.7	85.3	75.9	86.8
S2 (G2)	71.3	76.4	72.9	81
Yo (G3)	73.5	82.5	80.3	93.3
HCR3 (G3)	73.9	83.7	91.1	99.1
RV198/95 (canine G3)	73.9	83.7	91.9	99.1
ST3 (G4)	84.4	95.1	73.8	83.4
cr117 (G4)	85.5	95.4	73.9	83.4
ICB2185 (porcine G4)	88.2	96.9	75.3	85

^ant, nucleotide; aa, amino acid.

Another unexpected finding was that sequence analysis of the VP7 gene of the P[6],G4 strain D151, isolated in 1988, had the closest nucleotide sequence homology (88.2% on the nucleotide level and 96.9% on the amino acid level) with strain ICB2185, an atypical porcine rotavirus isolated in Brazil in 1995 to 1997 (23), rather than to the typical P2A[6],G4 human rotavirus reference strain ST3 (84.4% on the nucleotide level and 95.1% on the amino acid level). The strain with the second highest homology (85.5% on the nucleotide level and 95.4% on the amino acid level) to D151 was a human strain, cr117, which was recently isolated in Beijing, China (10). This prior example of a human G4 strain with a VP7 gene sequence closely related to the cognate gene of a porcine G4 strain further supports the hypothesis that transmission of rotaviruses with prevalent human serotypes between humans and animals may be a more common occurrence than previously thought. It is possible that many G1 to G4 strains isolated from humans, rather than being closely related to other common human strains, are actually more closely related to animal strains of the same serotype. Since these strains are assumed to be of human origin and are rarely studied intensively, the frequency at which G1 to G4 strains of animal origin occur in humans remains unknown.

The factors that promote interspecies transmission of animal rotavirus genes or virions to humans are poorly understood. It can be speculated that close contact of children with animals may sometimes result in interspecies infections with animal rotaviruses and/or coinfections with human and animal rotaviruses, resulting in reassortant formation. Subsequent spread to other children could then result in establishment of these strains in humans. In the present study, the two children from whom the canine-like and porcine-like rotaviruses were isolated had dogs and pigs living in their respective houses. Thus, close contact with fecal material from these animals could have resulted in interspecies infections.

At present, it is not possible by standard G and P typing methods alone to establish whether a strain shares genetic relationships with animal rotaviruses. To determine whether isolates bearing the common G and P types are of human or animal origin, routine gene sequencing or RNA-RNA hybridization methods would need to be adopted. As a more practical, feasible alternative, close analysis of the nucleotide sequences of human and animal strains having the same serotype could allow us to identify conserved regions within the gene that discriminate between the animal and human variants. With this, new primers could be designed for use in multiplex RT-PCR systems or in methods based on microarray technology (2) that differentiate between human and animal strains having the same serotype. Unless these extra investigative efforts are made, the novelty of these animal, or animal reassortant, strains having serotypes common to human strains will remain cryptic.

Nucleotide sequence accession numbers.

The VP7 gene sequences of the strains characterized here have been deposited in the GenBank database and assigned accession numbers AJ488586 (strain D151) and AJ488587 (strain 7177-1042).