Phylogenetic Analysis of Rubella Viruses Involved in Congenital Rubella Infections in France between 1995 and 2009

Abstract

Rubella is an acute infectious disease that normally has a mild clinical course. However, infections during pregnancy, especially before week 12 of gestation (WG), can cause severe birth defects known as congenital rubella syndrome (CRS). The aim of this study was to perform genotyping and molecular characterization of rubella viruses involved in congenital infections in France over the past 15 years (1995 to 2009). Amniotic fluid (AF) specimens (n = 80) from pregnant women with congenital rubella infections (CRI) before week 20 of gestation, and a few other samples available from children/newborns with CRS (n = 26), were analyzed. The coding region of the rubella virus E1 gene was amplified directly from clinical specimens by reverse transcriptase PCR, and the resulting DNA fragments were sequenced. Sequences were assigned to genotypes by phylogenetic analysis with rubella virus reference sequences. Sufficient E1 gene sequences were obtained from 56 cases. Phylogenetic analysis of the sequences showed that at least five different genotypes (1E, 1G, 1B, 2B, and 1h) were present in France and were involved in congenital infections, with a strong predominance of genotype 1E (87%). This is one of the very few comprehensive studies of rubella viruses involved in CRI. The results indicated that over the past 15 years, multiple introductions of the dominant genotype E caused most of the CRI cases in France. A few sporadic cases were due to other genotypes (1B, 1G, 1h, 2B).

Rubella is an acute infectious disease that normally has a mild clinical course. However, infections during pregnancy, especially before week 12 of gestation (WG), can cause severe birth defects known as congenital rubella syndrome (CRS) (2). Clinical signs of CRS include cataract(s), glaucoma, heart disease, loss of hearing, and pigmentary retinopathy. Congenital rubella infection (CRI) is usually confirmed by rubella virus-specific IgM (http://www.who.int/immunization_monitoring/diseases/rubella_surveillance). Infected fetuses from women who have had early abortions are also considered CRI cases.

Prenatal diagnosis is based mainly on the detection of rubella virus in amniotic fluid (AF) by reverse transcription-PCR (RT-PCR) (5, 10, 19, 25) or, less frequently, on the detection of rubella virus-specific IgM antibody in fetal blood (8, 21).

Live-attenuated vaccines against rubella virus have been available since the late 1960s (23). They are in use in more than 60% of countries worldwide, but vaccination coverage differs widely (29, 30). In France, a selective rubella vaccination program was introduced in 1970 for women of childbearing age, and measles-mumps-rubella childhood immunization has been recommended since 1986. France has adopted the WHO target to prevent CRI in the WHO European Region by 2010 (20). In 2004 to 2005, first-dose vaccination coverage was high (95.7%) among 11-year-old children, but it was only 80.6% among 2-year-old children. The second-dose coverage for 11-year olds was only 74.2%, with significant disparities between the north (82.8%) and the south (62.0%) of France (11).

Although no CRI cases were reported to the National Institute for Public Health Surveillance in 2006 or 2008, 2 cases were reported in 2007 and 1 case in 2009 (up to September) (22). CRI cases are not unexpected, considering the insufficient vaccination coverage, especially in some regions, and surveillance must be maintained, since occasional peaks of cases have been observed in the past (e.g., in 1993, 1997, and 2000). In France, this surveillance, implemented since 1976, is based only on a comprehensive voluntary laboratory-based reporting system for rubella among pregnant women, fetuses, and newborns. Although this system does not allow the detection of small epidemics in the community, it aims to assess the current immunization program. An increase in the incidence of infections among pregnant women would be considered an indicator of the active circulation of rubella virus among young adults and potentially in the whole community.

The use of molecular epidemiology has contributed to the understanding of the worldwide genetic diversity and transmission routes of rubella viruses and is considered important for supporting control and elimination activities (3, 31, 32). Based on partial E1 gene sequences, nine recognized and four provisional genotypes of rubella virus have been defined (32). Although knowledge about the genotypes of circulating viruses is available for some countries, especially those pursuing elimination (4, 7, 9, 13-17, 24, 26, 35), genotype data are lacking for many others, including France.

The aim of this study was to perform genotyping and molecular characterization of rubella viruses involved in congenital infections in France over the past 15 years.

MATERIALS AND METHODS

Clinical samples.

Samples from 106 CRI and CRS cases, 104 from all regions of France (including the West Indies) and 2 from abroad (1 from Portugal and 1 from Tunisia), collected between 1995 and 2009, were included in this study. Clinical samples consisted of 80 AF specimens, 16 urine samples, 9 cerebrospinal fluid specimens, and 1 lens aspirate. All samples were kept frozen at −70°C until further investigation. Case numbers are given in Fig. 2 and 3.

FIG. 2.

Phylogenetic tree based on the WHO-recommended minimal window region of rubella virus and the neighbor-joining method. Bootstrap values above 50 are shown, and the samples obtained during this study are marked with black dots.

FIG. 3.

Clusters showing genotype 1E sequences (a) and the position of strain 2 (b) from France in a phylogenetic tree based on the minimum acceptable window and the neighbor-joining method and including all sequences available in GenBank (September 2009). Bootstrap values above 50 are shown. The samples obtained during this study are marked with black dots, and the Japanese vaccine strain (GenBank accession number D50676) is marked with a gray triangle.

RNA extraction and PCR.

Phenol-chloroform RNA extraction and reverse transcription were performed as previously described (10, 19). Briefly, after RNA extraction, reverse transcription was done using primer Rurt E1 (5′-TTT TTT TTT CTA TGC AGC AAC-3′) by incubation of the reaction mixture for 1 h at 37°C, followed by 5 min at 95°C. Eight primers (Table 1) were used to amplify the entire E1 gene region of 1,446 nucleotides (nt) (nt 8258 to 9703), which includes the 739 nt (8731 to 9469) corresponding to the minimum acceptable window defined by WHO for routine molecular epidemiology (31). Forward and reverse primers were paired (EA5 with RU4, 8669F with SPR8, SPF9 with 86, and 615 with 9′) in order to obtain 4 overlapping fragments that cover the E1 gene. The parameters for all amplifications were exactly the same: 95°C for 10 min; 35 cycles of 95°C for 1 min, 55°C for 2 min, and 72°C for 2 min; and finally 72°C for 10 min, followed by 4°C for 10 min. PCR amplification was verified by gel electrophoresis, and PCR products were purified on Microcon columns (Amicon; Millipore, Bedford, MA).

TABLE 1.

PCR primers used for the amplification and sequencing of the E1 gene region

Primer	Genomic position (nt)	Sequence (5′-3′)	Source or reference
Forward
EA5	8244-8264	CTATGGCGAGGAGGCTTTCAC	Designed by E. S. Abernathy and J. Icenogle
8669F	8669-8689	GTGATGAGCGTGTTCGCCCTT	1
SPF9	8968-8988	CCTGACTGCTCGCGGCTTGTG	12
615	9122-9140	CTCCACATACGCGCTGGAC	26
Reverse
RU4	8824-8847	GTCGGGCGGGACCTGGACCTCGAG	10
SPR8	9105-9125	GCGCGCCTGAGAGCCTATGAC	26
86	9524-9540	TGGTGTGTGTGCCATACA	26
9′	9737-9756	TATACAGCAACAGGT	Modified from reference 12

Sequencing.

Purified PCR products were sequenced in both directions with the BigDye Terminator cycle sequencing kit, version 3.1 (Applied Biosystems, Nieuwerkerk, Netherlands), on a capillary sequencer (model 3100; Avant; Applied Biosystems) using the PCR primers.

Data analysis.

Sequence analysis was done with the SeqScape program, version 2.5 (Applied Biosystems). Phylogenetic analysis using MEGA, version 3.1 (17a), was based both on the entire E1 gene region comprising 1,446 nt and on the 739 nt corresponding to the minimum acceptable window defined by WHO (31). For genotype determination, reference sequences (32) were included in each analysis. In addition, all sequences spanning the window region available on GenBank (September 2009) were downloaded for comparison. The neighbor-joining method with the Kimura 2-parameter model and the maximum-parsimony method were employed, by choosing 1,000 bootstrap replications. Distances were also calculated with MEGA by using the within-group means and pairwise options. Samples were named according to the convention provided by WHO (31). All dates (week and year) refer to the time of sample collection. For easier reference, the samples are referred to by number in the text, while in the phylogenetic trees both the sample numbers and the official WHO names are displayed.

Nucleotide sequence accession numbers.

The sequences obtained during this study are available under GenBank accession numbers FN546966 to FN547021.

RESULTS

Between January 1995 and September 2009, 91 congenital infections were reported to the French National Institute for Public Health Surveillance after prenatal diagnosis. For three cases, prenatal diagnosis relied only on the detection of rubella virus-specific IgM in fetal blood. For 88 cases (97% of the reported cases), the rubella virus genome was detected in AF (Fig. 1). Two main peaks of cases were observed—one in 1997 (22 cases) and the other in 2000 (15 cases)—while no cases were reported in 2006 and 2008 (Fig. 1). Out of 88 CRI cases detected by prenatal diagnosis on AF in France during the study period, 78 (88.6%) were in fact diagnosed in our laboratory and were therefore available for the study. Two other AF samples were addressed to our laboratory from abroad (Portugal and Tunisia) during the study period, and we also included these in the study. The 26 other clinical samples (urine samples, cerebrospinal fluid samples, and a lens aspirate) were available from French children/newborns with CRS.

FIG. 1.

Numbers of CRI cases in pregnant women in France between 1995 and 2009 that were reported (shaded bars), investigated by Antoine Béclère's laboratory (filled bars), and genotyped (open bars). (Only CRI cases detected by prenatal diagnosis on AF are reported in this figure.)

All these rubella cases were detected using a diagnostic PCR different from the PCR generating the fragments for E1 gene sequencing and genotyping. During storage, partial RNA degradation may have taken place, no longer allowing the amplification of the larger fragments needed for genotyping. Therefore, complete or nearly complete E1 gene sequences (1,437 to 1,446 nt) were obtained from 49 cases. From seven additional samples (samples 9, 30, 53, 78, 84, 104, and 121), regions between nt 848 (sample 53) and nt 1280 (sample 121) of the E1 gene, including the complete window region, were amplified. Of these 56 sequences, 1 was derived from the lens sample (sample 2), 2 from cerebrospinal fluid (samples 84 and 92), 3 from urine (samples 66, 73, and 97), and 50 from amniotic fluid.

Phylogenetic analysis of these sequences showed that at least five genotypes (1E, 1B, 1G, 2B, and 1h) were present in France during the years 1995 to 2009 (Fig. 2). Of the 56 sequences obtained, 49 (87.50%) clustered with genotype 1E reference sequences. The sequences within genotype 1E were quite similar (within-group distance, 0.70%) and clustered interspersed with sequences from America and Europe (Fig. 3a). Several pairs of 1E sequences were identical over the total length of sequence information obtained (samples 115 and 120, 3 and 42, 48 and 53, 58 and 61, 78 and 81, and 85 and 97). The two most distant viruses within genotype 1E (samples 104 and 26) differed by 1.93% over the E1 gene region. Besides genotype 1E sequences, two representatives each of genotypes 1G (samples 9 and 51; Kimura distance, 1.43%) and 2B (samples 114 and 121; Kimura distance, 1.58%) were found (Fig. 2). One sequence each grouped with 1B (sample 92) and 1h (sample 5), while another single strain (sample 2) clustered separately from all reference sequences as an outlier to the 1G/1h/1i group of sequences, close to the 1B reference strain RVi/Jerusalem.ISR/75 (Fig. 2). Analysis based on the complete E1 gene region showed a similar picture, with strain 2 again an outlier to the 1G/1h/1i group of viruses, again with no bootstrap support (data not shown). When all available window sequences from GenBank were included in the analysis, strain 2 still clustered separately from all other viruses (Fig. 3b).

The sequence most similar to strain 2 identified by BLAST was a Japanese vaccine strain first isolated in 1967 (GenBank accession number D50676; Kimura distance, 2.48%), which clusters with the 1B viruses (Fig. 3b). Different sequences from Italy showed the highest similarity to the genotype 1G viruses (GenBank accession number AY161361, collected in 1993, to strain 9 [Kimura distance, 0.35%]; GenBank accession numbers AY161371, AY161372, and AY161373, all collected in 1995, to strain 51 [Kimura distance, 0.71%]), while a sequence determined in 2000 in the United States from a virus considered to have been imported from India was the closest match of both 2B strains (GenBank accession number AY968220; Kimura distances to strains 114 and 121, 1.77% and 2.15%, respectively). The closest fit to the 1B sequence (sample 92) was a virus from Israel collected in 1979 (GenBank accession number AY968208; Kimura distance, 3.22%). The 1h sequence was most similar to a sequence from Western Siberia from 2004 (GenBank accession number EF421977; Kimura distance, 1.62%). The 1E viruses obtained during this study showed the highest similarity with viruses from Italy collected in 1997 (GenBank accession numbers DQ085343 [Kimura distance, 0.28 to 1.36%] and AY161378 [Kimura distance, 0.12 to 0.85%]).

Several unique amino acid changes relative to 169 complete or nearly complete E1 gene sequences from GenBank were identified in single sequences: 28L (sample 26), 50V (sample 100), 103T (sample 68), 138A (sample 99), 142L (sample 34), 152R (sample 35), 181K (sample 68), 196A (sample 71), 215R (sample 84), 285L (sample 84), 316R (sample 104), 340L (sample 99), 347A (sample 62), and 405T (sample 26). At least one of these mutations (215R) occurs in the epitope of a neutralizing monoclonal antibody (33).

DISCUSSION

Sequences of rubella viruses from a number of countries throughout the world have been determined (4, 7, 9, 13, 15-17, 24, 26, 35), but often only a few sequences have been characterized. This is one of the few studies with a large number of rubella virus strains with broad regional and temporal distribution in a single country. It is also one of the very rare studies investigating a significant number of rubella viruses involved in congenital infections. The strains were collected from all regions of France over a 15-year period (1995 to 2009), providing the baseline information about the circulating genotypes as required for the documentation of rubella/CRS elimination (32).

Phylogenetic analysis of the 56 sequences obtained during this study showed that at least five different genotypes were involved in congenital infections, with a strong predominance of genotype 1E. It is possible that genotype 1E has a higher propensity to cause congenital infections than other genotypes, but it is more likely that 1E viruses were simply the most prevalent. Although there is no genotype information from acute rubella cases in France other than that from this study of pregnant women, most WHO European countries for which genotyping data are available for the same period also reported 1E (9/12 [75%]) (28). Recent studies suggest a very wide circulation of this genotype (7, 9, 13, 34, 35), which may well be the most prevalent contemporary genotype worldwide. Phylogeny shows that the French 1E sequences are interspersed with viruses from other European countries and America, possibly indicating multiple exchanges between these countries. This is also supported by the four genotype 1E viruses found in Portugal, Tunisia, and Martinique/West Indies (samples 28, 50, 85, and 104). The other genotypes were relatively rare in France and possibly correspond to sporadic, imported cases with limited spread. Interestingly, most of the non-1E viruses were found during the 1990s, suggesting that genotype 1E may have later displaced other genotypes in France and perhaps beyond, although genotypes 1G, 1h, and 2B are still found in other countries in Europe (32, 35). While the geographic source of the rare genotypes detected during this study cannot be determined, it is noteworthy that both the genotype 1E and 1G viruses showed the highest similarity to sequences from Italy. Whether this corresponds to a true epidemiological link or is biased by the limited number of rubella virus sequences available from the past 15 years is unclear. The low genetic diversity (0.70%) among the genotype 1E sequences, despite the inclusion of four viruses acquired outside metropolitan France (samples 28, 50, 85, and 104 [see above]), points at least to a comparatively short duration of circulation in France.

The oldest virus sequenced during our study (sample 2) was isolated from a lens sample collected in 1995 from a young child presenting with a congenital cataract due to CRI acquired in 1992. This virus clustered separately from all rubella virus reference sequences and was most similar by a BLAST search to a Japanese vaccine strain first isolated in 1967 (GenBank accession number D50676). This vaccine strain clustered with the 1B viruses (Fig. 3b). However, D50676 was on a short branch in the phylogenetic analysis, indicating that these BLAST results should be interpreted cautiously. Furthermore, when all available sequences (in GenBank as of September 2009) spanning the window region were included, strain 2 clustered separately from all other sequences and as an outlier to the 1G/1h/1i group of viruses (Fig. 3b). BootScan analysis of strain 2 using SimPlot, version 3.5.1 (18), showed no evidence of recombination. Thus, strain 2 may belong to a genotype 1B that is more diverse than previously thought. Alternatively, it may be the so far unique representative of a new cluster or genotype of rubella viruses. The status of strain 2 should be readdressed by the WHO Rubella Nomenclature Committee when more similar viruses become available.

The 1B virus found in this study (sample 92) was isolated from a baby, born in Paris in 2001, who presented with a congenital cataract and intracerebral calcifications. This virus was most similar to an isolate obtained in Israel more than 20 years earlier (in 1979) (GenBank accession number AY968208), but with quite a high Kimura distance of 3.22%.

Based on data reported to the National Institute for Public Health Surveillance (Institut de Veille Sanitaire), 2 years with a high incidence of rubella in pregnant women (1997 and 2000) were observed during the study period (Fig. 1). All the cases investigated from these 2 years (13 in 1997, 8 in 2000) were due to genotype 1E, and the patients came from different regions of France. Although no data on rubella cases are available for the general population, there must have been a high number of acute cases to produce the numerous fetal infections recorded and reported here, and perhaps even nationwide outbreaks. The genetic diversity of the 1E viruses analyzed in relation to time and location did not yield evidence of outbreaks in certain regions of France or during certain time periods.

In conclusion, the present study suggests that during the past 15 years, multiple introductions of the dominant genotype 1E caused most of the CRI/CRS cases in France. A few sporadic cases were caused by other genotypes, including a strain that could not be clearly attributed to any of the current genotypes. Even though only 3 CRI cases were reported during the past 3 years, decreasing vaccination coverage rates, particularly in some parts of the country, represent a considerable threat to the long-term success of the WHO elimination goal. Ongoing surveillance and renewed efforts to substantially improve vaccination coverage are therefore warranted.