This map indicates the location of the two districts where the poliomyelitis outbreaks occurred. Names of the PV strains isolated from the patients with AFP are indicated.

A given dose of virus was used to inoculate intracerebrally (IC) or intraperitoneally (IP) groups of PVR-Tg mice expressing the human poliovirus receptor (8–10 animals per virus). Animals were checked daily for 14 d following inoculation for paralysis and death. Percentages of affected mice following inoculation of viruses MAD 04 to MAD 07, MAD 29, and of a positive neurovirulent control virus S2/4568 are given. Strain S2/4568 was used only for IP inoculation. IP or IC inoculation of the vaccine strain Sabin 2 did not induce disease in animals.

(A) Plots of similarity between a set of aligned genomes (Sabin 2, MAD 04, MAD 29, and Sabin 1), using a 50-nt sliding window. Approximate nucleotide positions in the poliovirus genome are indicated. Average nucleotide identities between the 5′ vaccine part and the 3′ non-vaccine part of the recombinant genomes are shown.
(B) Sites of the recombination junctions in the recombinant VDPV genomes. The genetic organization of the PV genome is shown, including the 5′- and 3′-UTR. Genomic regions P1 to P3 encoding viral proteins (VP4 to 3Dpol) are indicated. Open and closed triangles indicate the approximate sites of the recombination junctions of the MAD 29 and MAD 04 recombinant genomes, respectively. Both junctions are located at the 3′ end of the protease 2A genomic region. MAD 04 and MAD 29 sequences (nt 3651–3890) on both sides of the recombination junctions are given and compared with Sabin 2 sequences (S2). Nucleotide numbering according to Sabin 2 sequences is given in italics.

These neighbor-joining trees were based on nucleotide sequence alignments of part of the 5′ UTR, the C-terminal third of the VP1-coding regions, part of the 2C-coding regions, and of the 3D^pol regions. Branch lengths were calculated using PUZZLE and the Hasegawa, Kishino, and Yano (HKY) model of substitution [63]. The genetic distance is indicated (bar). Numbers at nodes correspond to the percentage of 25,000 puzzle steps supporting the distal cluster; numbers in italics representing the percentage of 1,000 bootstrap pseudoreplicates obtained with the DNADist/Neighbor and SEQBOOT programs of PHYLIP [65] are shown in several cases. Sequences of some of the last wild PV strains (PV3.Mad95, PV1.Mad96a, PV1.Mad96b, and PV3.Mad97) isolated in Madagascar (1995–1997) were used in addition to HEV-C prototype strains (CA1 to CA24). Selected PV strains isolated during the last poliomyelitis outbreak in Finland (PV3.Fin84) and during the recent VDPV outbreaks in Egypt (PV2.Egy93; PV2.Egy88), Haiti (PV1.Hai00), Dominican Republic (PV1.Dor00), and the Philippines (PV1.Phi01) are included. Sequences of the recombinant type 3 PV isolated from a poliomyelitis case in Cambodia in 2002 (PV3.Cam02) were also used. Nucleotide sequences of echovirus 70 (EV70) were used as outgroup. Identifiers of the HEV-C coxsackie A viruses isolated from healthy children in 2002 in Madagascar consist of the given serotype as assessed from partial sequencing [33] and the laboratory number. New VDPV isolates from healthy children are indicated with their laboratory numbers. Spots of different colors differentiate HEV-C serotypes and polioviruses (red). VDPVs from Madagascar are indicated by oblong spots. According to the Eight Report of the ICTV [56], former CA15 and CA18 are considered as antigenic variants of CA11 and CA13, and are now named CA11v and CA13v, respectively. All field isolates from Madagascar, except isolates CA24.65902 and CA24.68187, were sequenced in the four regions. Vertical dotted lines indicate groups of isolates of particular interest that are discussed in the text.