Main

A short description of five samples of oral polio vaccine (OPV) held by the Wistar Institute is given in Table 1. CHAT pool 13 was widely used to vaccinate thousands of people (mainly children) in Léopoldville3 in the former Belgian Congo (now Kinshasa, Democratic Republic of Congo) and some in Poland. Three further samples of CHAT vaccine held at the Centers for Disease Control (CDC) for 20–40 years were included in the study. The combinations of oligonucleotide primers and probes used can detect a wide variety of HIVs, including HIV-1 M, O and N isolates, as well as chimpanzee isolates Gab1, US and ANT70. The PCR assays have a sensitivity of about 50–80 RNA copies per ml. No sample proved positive for HIV-1-related nucleic acids by any of four primer pairs we used. By contrast, all samples were positive for poliovirus using PCR with reverse transcription, demonstrating that there was not extensive degradation of viral RNA (Table 1).

Table 1 Oral poliovirus vaccine samples tested
Full size table

To identify the source of the kidney epithelial monolayers used to culture OPV, we analysed a small region (140 base pairs) within the mitochondrial 12S ribosomal RNA gene4. The primers we used amplify DNA from animals, including the African green monkey, chimpanzee, rhesus monkey and sooty mangabey (see supplementary information). All but one of the OPV samples were positive for mitochondrial (mt) DNA. With the exception of CHAT 1FL (see below), most sequences (86%) were highly related to those of the rhesus (Macaca mulatta) or cynomolgus monkey (Macaca fascicularis), and differed by only 0–2 bases from a known haplotype (Fig. 1).

Figure 1: Neighbour-joining tree of 12S mtDNA sequences (see ftp://ftp.pasteur.fr/pub/retromol/wistar01).
figure 1

The three groups correspond to the Macaca, Gorilla/Homo/Pan and Cercopithicus/Erythrocebus/Cercocebus lineages. Of the latter, C. atys and C. torquatus are Cercocebus lineage. Wistar and CDC OPV samples and mtDNA controls are in yellow, green and red, respectively; the Sabin-1 positive control is in blue. Other sequences are from databases. The phylogenetic tree was derived by the neighbour-joining method applied to pairwise sequence distances calculated using the Kimura two-parameter method transition/transversion ratio set to 10. Branch lengths are drawn to scale: bar, 0.1 nucleotide replacements per site. Numbers indicate the percentage of bootstrap samples in which the cluster is supported (only bootstrap values over 60% are given). Rhesus macaques are split into two groups, corresponding to the Chinese and Indian subspecies. Remaining sequences are nuclear insertions of mtDNA sequences (see supplementary information).

Full size image

A few sequences were nuclear paralogues of mtDNA. The amplification of bona fide mtDNA sequences also picks up these nuclear insertions of mitochondrial sequences (NUMTs). They behave as pseudogenes and fix mutations more slowly than mtDNA itself5. A few of these NUMTs, three out of more than 250 sequences, were related to mtDNA sequences of a variety of Cercopithecus monkeys. These Cercopithecus-like sequences never exactly corresponded with a known sequence in the databases, but differed by about 4 base pairs. Most rhesus monkey mtDNA sequences corresponded to known mtDNA haplotypes. Some of the Cercopithecus-like sequences were also derived from single rhesus monkey samples (not shown). We conclude that these really are NUMTs and not evidence that some African monkey kidneys, along with those of macaques, had been used to grow OPV.

Given that OPV was prepared from pools of culture supernatants derived from individual pairs of kidneys, the presence of multiple mitochondrial haplotypes in the same sample is not unexpected. This would also explain the 3:2 mix of M. mulatta and M. fascicularis haplotypes in CHAT type 1 Wy4B. The detection of Homo sequences in CHAT 1FL stems from the fact that the virus was grown on the FL human diploid cell line.

To confirm these findings, we studied a mitochondrial 12S-rRNA locus (101 base pairs) just 5′ to the previous segment. This too was highly informative, with ten point substitutions and one insertion/deletion distinguishing macaque and chimpanzee sequences. All of our findings were confirmed. As the Wistar CHAT pool-13 lot was widely used in Léopoldville, we increased the resolution of the analysis. No chimpanzee sequences were found at a resolution of 1.7%. As this frequency is below the reciprocal of the number of animals generally used to make a lot of OPV, we conclude that none of these OPVs were prepared using chimpanzee kidneys.

In addition, none of the samples was positive when amplified with primers specific for chimpanzee nuclear DNA (data not shown). The extensive use of rhesus and cynomolgus monkeys in the preparation of all the Wistar samples described here not only confirms earlier claims6,7,8, as well as those denying the use of chimpanzee kidneys in the preparation of OPV9, but also reflects the procedures used in the late 1950s by Salk and Sabin.

CHAT pool 13 was used in Léopoldville between August 1958 and April 1960. The earliest bona fide HIV-1 sequence was derived from a 1959 serum sample taken from an adult living in the suburbs of Léopoldville10. According to the OPV/AIDS hypothesis1,2, the CHAT pool-13 sample should contain chimpanzee DNA, but as only macaque sequences were found, this tight geographical and temporal association would seem to have been a coincidence.

There is a corollary to our conclusion that the Wistar OPV samples were prepared using macaque kidneys. Given that simian immunodeficiency viruses (SIVs) are confined to African primates, the kidneys could not have been positive for SIV. In other words, the absence of SIV nucleic acids in all samples do not represent false negatives. We find no evidence to support the OPV/AIDS hypothesis.