The sequences used in this study consist of representative sequences from three earlier studies (Europe, North America, and Asia, 1961–1997 [8]; Italy, 1991–1997 [9]; and the United States, 1997–2001 [10]) and sequences of viruses collected in Russia, South Korea, New Zealand, Israel, China, Japan, and the United States that have not been reported (these new sequences are listed in Table 1; a complete sequence list is available online (http://www.cdc.gov/ncidod/EID/vol09no12/03-0242.htm#table1). The sequences from the three earlier studies have not previously been amalgamated into a single phylogenetic study. Phylogenetic trees produced with different programs yielded the same major branches, and only representative trees are shown (Figure 1). Branches were defined as reproducible clusterings of sequences by these different programs.

Most of the new sequences belonged to RGI, which emanate from a single node in the tree shown in Figure 1. For simplicity, the tree shown in Figure 1A contains only half of the RGI sequences; the RGI node with all of the RGI sequences is shown in Figure 1B. New sequences belonging to RGI included both from China, all three from New Zealand, all four from Japan, five of six from Israel, two of four from South Korea, and one of five from Russia. Additionally, all of the sequences from the U.S. study (10) and 19 of 21 from the Italy study (9) were RGI. Fourteen sequences were grouped on branches removed from the RGI node, including the three viruses originally used to define RGII, two RGII viruses isolated from Italy (1994) [9], and 9 of the new sequences: 4 from Russia (3 from 1967 to 1969 and 1 from 1997), 2 from South Korea (1996), 1 from Israel (1968), 1 from China (1979), and 1 isolated in the United States from a case contracted in India (2000).

In this study, we extended phylogenetic analysis of rubella viruses collected worldwide. The baseline for this study was an analysis (8) of viruses collected from Europe, North America, and Asia, 1961–1997; we found that an intercontinental genotypic group existed until ~1980 and was replaced by continent-restricted genotypic groups after ~1980. In this study, this analysis was augmented by inclusion of comprehensive collections from Italy (9) and the United States (10) as well as viruses collected from new locations. In the previous analysis (8), a second genotype was identified among a limited number of specimens collected from Asia and a specific goal of this study was to analyze viruses from new locations in hopes of learning more about this second genotype.

Use of the entire E1 gene resulted in excessive time being required to run phylogenetic programs with the large number of sequences included in this study. We therefore investigated using ~400-nt segments within the E1 gene. While the major genotypic and subgenotypic grouping were preserved in these segmental trees, the trees produced from the nt 8687–9088 window were most similar to those produced from the entire E1 gene. This window was named the “molecular epidemiology window.” As shown in Table 3, this window had a similar to somewhat higher intersequence variability than the E1 gene and preserved the GC content and intersequence transition/transversion ratio exhibited by the E1 gene. Although virus isolates were used in this study, amplification and sequencing of E1 gene segments directly from clinical specimens using reverse transcription-polymerase chain reaction (RT-PCR) are at hand (18–22). Our findings indicate that RT-PCR products from any region of the E1 gene will produce phylogenetic trees consistent with those from the entire E1 gene. However, thus far the regions amplified by in these studies have not been standardized.

Analysis of the segmental trees indicated that the FAR-GER89 isolate might have arisen by a recombination event within the E1 gene between a virus from the Europe 1986–1994 branch (TOM-UK86, FEL-UK88, D075-GER92) and a virus from the International 1964–1981 nodal cluster (THE-US64, RA27-USA64, JC1-NEZ81). Such an event is temporally consistent because viruses related to the International 1964–1981 nodal cluster have been isolated in Italy as late as 1991 (9) but were not included in this study. Although rubella virus recombination in cell culture has been documented (23), this report is the first of a natural recombinant. Although the FAR-GER89 isolate was related to the RA27/3 vaccine strain in the 3′ end of the E1 gene, the FAR-GER89 sequence contains none of the nucleotides characteristic of the RA27/3 sequence (7); thus, the recombination event did not involve a vaccine virus.

In our previous analysis (8), all of the viruses isolated in Europe, North America, and Japan were RGI. This finding was maintained with the expanded collections from all three regions used in this study with the exception of two RGII viruses isolated in Italy. Additionally, most of the viruses in the expanded RGI collection from these regions segued into the previously defined RGI branches (8), with the exception of a newly defined Europe 1972–1991 branch that contained viruses at the base of the Europe branch defined in the previous study and a novel branch consisting of two viruses from China (1999). In the case of viruses from Europe, with a larger number of viruses representing a longer time span, these viruses belonged to two branches, the second of which divided into two sub-branches. The RGI viruses from Israel were related to the European viruses and fit into the temporal pattern of isolation of the European viruses. The temporal pattern of isolation of viruses in the European branches and sub-branches (1972–1991, 1986–1994, and 1991–1998) indicated that temporal displacement of genotypic groups occurred, as had been noted in our previous study on viruses from Italy (9). These European branches and sub-branches have been more recently displaced by the International 1997–2000 sub-branch of the Japan-Hong Kong branch. The addition of viruses isolated recently in the United States (10) clarified that the US-Japan branch in the previous study (8) was a United States branch, viruses from which, when transported to Japan, caused an outbreak in 1991. Viruses in the USA 1990–2000 branch were related to the limited number of viruses available from Latin America.

The branch termed a Japan-Hong Kong branch in our earlier study (8) contained viruses from Japan and Hong Kong isolated between 1976 and 1991. Ten representative viruses were selected from that branch for this study, including 2 viruses from Japan (1976 and 1977) and 1 virus from Hong Kong (1985) that grouped at the base of the branch as well as several viruses from Japan (1987–1991) that formed a sub-branch. This branch was expanded by the addition of four sub-branches containing viruses from New Zealand, 1991; Japan and Korea, 1994–1996; Japan and the Philippines, 1997; and an international sub-branch that contained viruses associated with epidemics and outbreaks in Europe and the United States during 1997 to 2000 that were unrelated to viruses previously isolated from those regions. Recent evidence indicates that these viruses are closely related to viruses from China (Z. Zhen et al., unpub. data). Thus, the appearance of this sub-branch in Europe and the United States was likely due to intercontinental transport. Considering that viruses from this sub-branch have been predominant in Europe since 1997 and have been one of the two genotypic groups isolated recently in the United States (USA 1990–2000 is the other), an international genotypic branch of rubella virus appears to have emerged after a ~20-year hiatus since the previous demonstrable international branch. Along with this branch, the two other currently active branches are the USA 1990–2000 branch and the China 1999 branch, only represented by two viruses from the collection used in this study.

Figure 3

Figure 3. Histogram of genetic distances between rubella virus sequences. The histogram, showing the distribution of all of the pairwise distances between the rubella virus sequences in the study, was constructed from the...

With its expanded virus collection, this study demonstrates that RGI rubella virus isolates segregate into discrete subgenotypic groups (i.e., branches) that exhibit geographic and temporal consistency. These groups could provide the basis for a standard classification scheme, as has been developed for other viruses. In this study, the number of available RGII viruses also increased to 14 in contrast to the 3 available in the previous analysis; the RGII viruses segregated into three discrete clusters (RGIIA, RGIIB, and RGIIC). RGI viruses formed a discrete cluster clearly distinguishable from the most closely related RGII cluster (RGIIB) with no indication of intermediate viruses. The maximum diversity within RGI was 5.8%, and RGI viruses were an average of >7% different from the RGII viruses (Table 2); when one considers the extensive nature of the RGI virus collection, this is a working definition of a rubella virus genotype. RGII was originally defined to contain a limited number of sequences (three) that distinctly differed from the RGI viruses that made up most of the collection (8). The sequences in each of the RGII clusters vary by up to 5.4%, and the average distance between viruses in each of the clusters is >7% (Table 2). Thus, each of these clusters could represent a genotype. This hypothesis is strengthened by a histogram of the intersequence distances that show a bimodal distribution with peaks from 0.5%-6.0% to 6.5%-11.0%, corresponding to intra- and intergenotypic distances (Figure 3). The bimodal distribution also indicated equivalency of genotypic groups (24). However, in light of the limited number of RGII specimens, more specimens are needed to fully characterize the extent of the diversity within non-RGI viruses, and criteria need to be established for definition of additional genotypes. In this regard, as represented by the viruses in this collection, only one of the RGII clusters (RGIIA) has been repeatedly active in the past decade. However, viruses belonging to the RGII C cluster (China) have been recently isolated (Z. Zhen et al., unpub. data).

Figure 4

Figure 4. Distribution of rubella genotypes.

Figure 4 shows the countries from which RGI and RGII viruses have been isolated. RGII has not been isolated from an indigenous case outside of the Eastern Hemisphere. Only RGI viruses have been isolated from the United Kingdom, Belgium, Germany, Japan, and New Zealand; other studies have shown RGI viruses in Brazil (10) and Greece (23). Only RGII viruses have been isolated from India, and both genotypes have been isolated from Russia, Italy, Israel, China, and Korea. In Italy and Korea, the two genotypes were isolated in the same year. In both Italy and Israel, isolation of RGII viruses was only during a single year. By contrast, RGI viruses were isolated in both previous and subsequent years. This finding indicates that these RGII viruses were imports, although the relative distant relatedness of the two Italy isolates suggests at least two importation events, albeit in a single year (9). In China, RGII viruses were isolated in 1979 and 1980; the most recent isolates (1999) were RGI, although RGIIC viruses were recently isolated (Z. Zhen et al., unpub. data). These data indicate that RGI has a wider worldwide distribution than does RGII and that in much of the world RGI is the sole genotype. Recent RGII activity is confined to Asia and overlaps with RGI; however, the dataset from areas in which RGII viruses appear to circulate is limited. Additionally, as shown in Figure 4, large regions of the world remain to be sampled to complete the rubella virus genotypic picture.