Neurovirulent vaccine-derived polioviruses in sewage from highly immune populations

Abstract

Background: Vaccine-derived polioviruses (VDPVs) have caused poliomyelitis outbreaks in communities with sub-optimal vaccination. Israeli environmental surveillance of sewage from populations with high (>95%) documented vaccine coverage of confirmed efficacy identified two separate evolutionary clusters of VDPVs: Group 1 (1998-2005, one system, population 1.6x10(6)) and Group 2 (2006, 2 systems, populations 0.7x10(6) and 5x10(4)).

Principal findings: Molecular analyses support evolution of nine Group 1 VDPVs along five different lineages, starting from a common ancestral type 2 vaccine-derived Sabin-2/Sabin-1 recombinant strain, and independent evolution of three Group 2 VDPVs along one lineage starting from a different recombinant strain. The primary evidence for two independent origins was based on comparison of unique recombination fingerprints, the number and distribution of identical substitutions, and evolutionary rates. Geometric mean titers of neutralizing antibodies against Group 1 VDPVs were significantly lower than against vaccine strains in all age-group cohorts tested. All individuals had neutralizing titers >1:8 against these VDPVs except 7% of the 20-50 year cohort. Group 1 VDPVs were highly neurovirulent in a transgenic mouse model. Intermediate levels of protective immunity against Group 2 VDPVs correlated with fewer (5.0+1.0) amino acid substitutions in neutralizing antigenic sites than in Group 1 VDPV's (12.1+/-1.5).

Significance: VDPVs that revert from live oral attenuated vaccines and reacquire characteristics of wild-type polioviruses not only threaten populations with poor immune coverage, but are also a potential source for re-introduction of poliomyelitis into highly immune populations through older individuals with waning immunity. The presence of two independently evolved groups of VDPVs in Israel and the growing number of reports of environmental VDPV elsewhere make it imperative to determine the global frequency of environmental VDPV. Our study underscores the importance of the environmental surveillance and the need to reconsider the global strategies for polio eradication and the proposed cessation of vaccination.

Figure 1. Genomic fragments of type 2 VDPVs amplified by RT-PCR for sequencing and phylogenetic analysis.

The polyprotein encoded by the Sabin 2 genome is represented by a rectangle on the genomic RNA. The 5′ UTR (partial), P1, VP1, and 3D polymerase regions that were amplified by RT-PCR and sequenced are represented by 4 rectangles with dark cross hatching below the genomic RNA. They correspond to nt 183 to 747, 748 to 3384, 2482 to 3384, and 5986 to 7368, respectively, of Sabin 2 (Acc: X00595). The downward pointing arrow indicates the start of the hypervariable region of 5′UTR. The upward-pointing arrow indicates the location of the genetic recombination in the 3D polymerase gene. The letters R1 through R6 correspond to the separate regions used for sequence comparisons and phylogenetic analysis as shown in Figs. 2, 3 and 4.

Figure 2. NICER Identity Analysis of the aVDPVs: genomic recombination fingerprint in the 3D polymerase gene, and identity plots for the P1 and VP1 regions.

Upper panel: An Identity Plot created by the NICER program of the nucleotide sequences of Group 1 (G1) and Group 2 (G2) aVDPVs compared with Sabin 1 (S1) and Sabin 2 (S2) at each nucleotide position of 3D where Sabin 1 and Sabin 2 differ. The nucleotides of the query sequences are scored as identical to Sabin 1 (light orange/light grey(B&W)), identical to Sabin 2 (dark brown/dark grey(B&W)), or different from both (grey/white(B&W)). The upward pointing arrow (G1-X), indicating the most likely position where identity of the Group 1 sequences shifts from Sabin 2 to Sabin 1, has been placed between positions 19 and 20 which corresponds to nucleotides 6693 and 6705 of Sabin 2 (Acc: X00595). The downward pointing arrow (G2-X); indicating the most likely position where identity of the Group 2 sequences shifts from Sabin 2 to Sabin 1, has been placed between positions 52 and 53 which corresponds to nucleotides 6771 and 6774 of Sabin 2 (Acc: X00595). Middle panel: An Identity Frequency plot to indicate the frequency of occurrence of nucleotide positions in P1 (region R3) of the reference sequence (Sabin 2) where the equivalent nucleotide on at least one of the query sequences differs was created by the NICER program. The number and frequency of identical positions (all of query sequences differ and all have the identical substitution; N/N) and near identical positions (all but one of the query sequences have identical substitutions; N-1/N) is shown for analysis of 6 Group 1 aVDPVs isolated between 1998 and 2004 (A); the 6 Group 1 isolates in A plus the 3 Group 1 aVDPVs isolated in 2005 (B); the 6 Group 1 isolates in A plus the 3 Group 2 aVDPVs isolated in 2006 (C); and all 9 Group 1 and 3 Group 2 isolates (D) is shown. For example; in plot A there were 70 positions where 6 of 6 query sequences had identical substitutions and this identity occurred in 12.5% of all positions where at least one of the query sequences differed from the reference sequence. Bottom panel: An Identity Frequency Distribution plot to indicate the nucleotide position where the equivalent nucleotide on at least one of the query sequences differs from the nucleotide in VP1 (region R4) of the reference sequence (Sabin 2) and % of query sequences that had the maximum number of identical substitutions at each of those positions. Analysis of the 6 Group 1 aVDPV isolates plus either the 3 Group 1 aVDPVs isolated in 2005 (B.1) or the 3 Group 2 aVDPVs isolated in 2006 (C.1).

Figure 3. Phylogenetic relationships among Type 2 aVDPVs at 6 locations along the genome.

Phylogenetic trees 3A through 3F are maximum likelihood phylogenetic trees for 6 regions distributed along the polioviral genome, two in the 5′ non-coding region, 2 in the region encoding structural genes and 2 in the region encoding non-structural genes (R1–R6 in Fig. 1). The non-coding regions analyzed are R1, the 5′UTR encoding domains III to VI (Fig. 3A ; nt 196 to nt 645), and R2, the 5′UTR hypervariable region (Fig. 3B : nt 646 to nt 712). The regions encoding structural genes analyzed are R3, the P1 region encoding all capsid proteins (Fig. 3C ; nt 748 to nt 3384), and R4, VP1 (Fig. 3D ; nt 2482 to nt 3384). The regions encoding nonstructural genes are R5, the RNA polymerase gene before the recombination point (Fig. 3E ; nt 5986 to nt 6690), and R6, the RNA polymerase gene after the recombination point (Fig. 3F ; nt 6700 to nt 7368). Nucleotide positions are given relative to Sabin 2 (Acc: X00595). Trees were prepared using the dnaml maximum likelihood application of the PHYLIP program. The transition:transversion ratio was set at 10, data was bootstrapped 100 times and the order of sequences for each bootstrap randomized 3 times. Trees A through E are rooted to Sabin 2, while tree F is rooted to Sabin 1(Acc: V01150). Branch lengths correspond to the number of times a given branch appears among all bootstrapped trees. All branches with bootstrap values <90% have been collapsed. The trees were visualized using the njplot program.

Figure 4. Estimation of the date of initial exposure to Sabin-2.

Fig. 4.A. is a method for estimating of the date of initial exposure based on regression analysis of the accumulation of the % of 3^rd position synonymous substitutions from all codons with single nucleotide substitutions in VP1 (region R4 in Fig. 3D). The date of initial exposure to Sabin 2 was determined by extrapolation backwards to 0% substitutions. (r² = 0.886). Group 1 isolates are represented by open circles while Group 2 Isolates are represented by open triangles. Fig. 4.B is a rooted, neighbor-joining, phylogenetic tree of the same region, region, R4, analyzed in Figs. 3D and 4.A. Unlike Fig. 4.A, substitutions in all three codon positions were taken into account. The tree was prepared with ClustalX. Data was bootstrapped 1000 times. The numbers on the branches are the bootstrap values. The tree was visualized using the njplot program. Branch lengths correspond to genetic distances. Since dates of isolation are know, branch lengths were converted to time. The time for the distance between Sabin 2 and the first branch was calculated proportionally to the minimum and maximum branch “times” between different pairs of aVDPVs.

Figure 5. Geometric mean titers (GMT) of neutralizing antibodies against Type 2 strains by age group.

The GMT for each strain was calculated for each age group after data was normalized by comparison to titers from a standard pool of 20 positive sera included in each analysis as described in Materials and Methods. The cohort (see Table 2) for which each GMT point was determined is shown at the top of the figure. The symbols for each isolate are in Fig. 5B. Fig. 5.A represents GMT of neutralizing antibodies to poliovirus in sera from cohorts aged 15 months to 15 years in which the complete immunization history of all individuals was documented. Fig. 5.B represents the GMT in sera from cohorts aged 18 years to 50 years from convenient samples obtained from the serum banks of the Israel Center for Disease Control (ICDC).

Table 1. Amino acid substitutions in neutralizing antigenic sites of Israeli aVDPVs.