Microarray-based techniques have been well established as powerful tools in various fields of molecular biology (6
). In virology, this method has been used for analyses of gene expression profiles of cells infected with various viruses. Recently, this assay has been applied for genotyping of selected viruses, including polioviruses, hepatitis B viruses, influenza viruses, hantaviruses, coxsackieviruses, papillomaviruses, measles virus, adenoviruses, flaviviruses, poxviruses, herpesviruses, and mumps viruses (3
). We reported previously the successful development and evaluation of an oligonucleotide microarray hybridization methodology for the identification of five clinically relevant human rotavirus G genotypes (i.e., G1 to G4 and G9) (8
). More recently, an increasing number of epidemiological studies have reported the detection in diarrheal patients of uncommon G types, including G5, G6, G8, G10, and G12, in both developing and developed countries. Thus, in this study, we modified the original microarray assay and expanded it to include the identification of such unusual G types as well as clinically relevant P genotypes (i.e., P, P, P, P, and P). Hence, this microarray assay can now identify almost all G and P genotypes of human rotaviruses that have been detected thus far. Epidemiological surveillance of rotavirus G and P types before and after an introduction of a candidate rotavirus vaccine in various parts of the world is important in order to evaluate (i) temporal and geographic changes/fluctuations of rotavirus genotype distribution, (ii) relationships between vaccine efficacy and circulating rotavirus strains, and (iii) horizontal transmission of vaccine strains.
The analytical sensitivity of the microarray method reported in this study is very high. The minimum concentration of cDNA that could be detected and typed was 8 ng/μl. In general, samples that produced a visible band in an ethidium bromide-stained 1% agarose gel after the first-round or nested PCR were all successfully P or G genotyped. By applying nested PCR to the first-round-PCR products of strains that gave no visible bands (and therefore could not be typed), we could successfully amplify and genotype both the VP4 and VP7 genes of such rotavirus strains. Thus, because of the high sensitivity of primer pairs used in the nested PCR in this study, the combination of first-round and nested PCR was shown to have a potential of amplifying both the VP4 and VP7 genes of strains bearing not only common but also uncommon G or P specificity.
The specificity of this microarray method is also very high. We could correctly and unambiguously identify both P and G genotypes of all reference rotavirus strains belonging to genotypes against which the genotype-specific microarray probes were designed. In addition, no cross-reactivity with rotavirus G or P genotypes against which specific probes were unavailable on the chip was detected. Thus, strains which do not generate fluorescent signals can safely be considered to bear a G genotype other than G1 to G6, G8 to G10, and G12 and/or a P genotype other than P, P, P, P, and P.
Only four G and two P genotype-specific probes were found to be cross-reactive with samples of heterologous genotypes. Of note is the finding, however, that each of the samples that exhibited cross-reactivity reacted with only one heterologous probe. Moreover, the intensity of all cross-reactive signals was low. It has been well recognized that it is necessary to distinguish mixed rotavirus infections from cross-reactivity in multiplex RT-PCRs (14
) by using a confirmatory assay for the individual genotypes. In contrast to PCR, the microarray method has the inherent ability to readily identify individual genotypes for the mixed infections. In such a case, unlike with cross-reactive signals, more than one probe of suspected G and/or P genotypes would produce strong positive fluorescence signals. Moreover, in this microarray assay, such cross-reactive signals can be utilized for accurate discrimination among various strains within the same genotype. For example, bovine G10 rotavirus KC-1 and B223 strains reacted with the heterologous G2-7 probe with low intensities, while human G10 rotavirus A64 strain exhibited no reactivity with the same probe.
While this study was in progress, Lovmar et al. reported (35
) a genotyping method of five G (G1 to G4 and G9) and four P (P, P, P, and P) types of rotavirus by using PE in a microarray format. Their method appears to differ from ours in various aspects. First, because discrimination of genotypes depends on the difference in the 3′ ends between genotype-specific probes and the target genome segments of samples, this method may not work well with strains that have a mutation in this region. For example, among the strains tested in the present study, (i) two G3 strains (HCR3 and Ro1843) which have 3′ ends different from that of each of three G3-specific probes (G3-1, G3-2, and G3-3) reported by Lovmar et al. could not be typed (false-negative result), and (ii) three or four sequence mismatches found between two G4 strains (Gottfried and SB1A) and the three G4-specific probes (G4-1, G4-2, and G4-3) described by Lovmar and coworkers may lead to a wrong conclusion (false-negative result). Second, in their assay, it may be difficult in some cases to distinguish mixed rotavirus infections from cross-reactions that may cause a false-positive result. For example, because there is only one base mismatch between the G10 strains (B223 and KC-1) tested in the present study and one of their G2-specific probes (G2-6) and because, in addition, 3′ ends are complementary to each other, such G10 samples may be misdiagnosed as a mixture of G10 and G2 viruses (false-positive result). Third, this method can provide only limited information about 3′-end differences among strains within the same genotype. Previously, we reported that our microarray method was capable of “subgenotyping” viruses in a given G genotype (G1 to G4 and G9) (8
). In the present study, not only the VP7 gene but also the VP4 gene of each rotavirus strain was found to display a unique hybridization pattern that correlated closely with its nucleotide sequences. Thus, one of the major advantages of our new microarray method is that it is ideal for analyzing such genetic polymorphisms of the VP7 and VP4 genes of various rotavirus strains within the same genotype. In addition, we found that the existence of even one nucleotide mismatch located near the 3′ or 5′ end of a probe changed its pattern of hybridization profile (data not shown). In general, rotavirus strains in a given genotype that are isolated (i) at different locations, (ii) in different years even at the same location, or (iii) from different animal species display different hybridization profiles. For example, two Japanese G2 strains, KUN and S2, displayed similar hybridization patterns which were distinct from that displayed by U.S. G2 strain DS-1 or Venezuelan G2 strain HN126 (data not shown). These results indicate that this assay can be applied to monitoring the evolutionary divergence and genetic drift of the selected rotavirus genes, including the VP7 and VP4 genes. Although sequencing of the target genome segment may be needed for detailed analysis, it is time-consuming, expensive, and not practical for analyses of a large number of field isolates. This microarray method is ideal for a rapid screening of a large number of rotavirus strains to detect and analyze genetic polymorphism before sequencing. Analyses by this assay of a possible genetic/antigenic drift of the VP7 and/or VP4 gene of G1P viruses collected longitudinally in three countries (Brazil, Spain, and the United States) are under way in this laboratory.