infants. This may be related to the lower incidence of rotavirus infection in this environment relative to E. coli infection. The incidence of rotavirus diarrhea in Bangladesh is estimated to be about 0.5 episodes per child per year (Black et al., 1982a). Neonatal infection is most commonly asymptomatic; the vast majority of neonates with evidence of rotavirus in the stool can be classified as carriers based on both lack of symptoms and the absence of an antibody response (Champsaur et al., 1984).

Rotavirus is a double-stranded RNA virus in the Reoviridae family, with a distinctive genome of 11 segments. Serological classification has been somewhat confusing; however, recent work permits separation of distinct serotypes based on outer capsid antigens detected by neutralization with hyperimmune sera (Wyatt et al., 1982). Serotype specificity may, in fact, be determined by two distinct genes, as is the case with influenza virus (i.e., the genes for neuraminidase and hemagglutinin). Four human serogroups have been defined, two of which contain cross-reactive animal rotaviruses, and at least three other serogroups exist containing animal rotaviruses (Hoshino et al., 1984). Epidemiological studies are in progress to determine the prevalence of these serotypes in different parts of the world. The present data indicate that serotypes 1 and 2 are present worldwide. Serotype 3 appears to be less prevalent, and serotype 4 has been found only in Europe (Kapikian, personal communication, 1984). Some heterologous cross-reactivity has been reported between animal and human serotypes 3 and 4 (Hoshino et al., 1984). The number and cross-reactivity of serotypes is obviously important for vaccine development.

Rotavirus serotypes may be divided into subgroups based on inner capsid antigens detected by complement fixation, ELISA (enzyme-linked immunosorbent assay), or immune adherence assays (Kapikian et al., 1981). The two well-defined subgroups, 1 and 2, also can be identified by differences in RNA patterns detected by electrophoresis in polyacrylamide-agarose gels (Kalica et al., 1981). Subgroup and serotype antigens are controlled by different segments of the virus genome.

In vitro cultivation of human rotaviruses has been difficult in the past. Strain Wa, the prototype serotype 1 rotavirus, was originally propagated in African green monkey kidney cells following 11 passages in newborn, germ-free piglets (Wyatt et al., 1980). Other strains, including DS-1, the serotype 2 prototype strain, were grown following rescue by genetic reassortment with readily cultivated bovine rotaviruses (Greenberg et al., 1981). Recently, many human rotaviruses (up to 75 percent of stool isolates) have been grown successfully in MA-104 cells, a primary embryonic cynomolgus monkey kidney line, following pretreatment of virus by trypsinization and low speed centrifugation (Sato et al., 1981; Urasawa et al., 1981).

Protective antigens have not been well defined. There is evidence of cross-protection between animal and human viruses, but the responsible determinants have not been identified (Wyatt et al., 1979).