by Pepin
One could speculate that the extremely high male/female ratio in the Mayombe work camps (ten men for each woman) and the intense prostitution that ensued would have facilitated the transmission of HIV-1, possibly from a single worker who had been infected with SIVcpz. The time between the workers being sent to the CFCO camps and the development of their disease was not indicated in Auclert’s thesis, but was probably less than the ten years we usually see today between getting HIV-1 and the first symptoms of AIDS. This does not exclude anything: for complex virological reasons, it is possible that this incubation period was actually shorter soon after the virus was introduced into human populations. And even nowadays, some unfortunate patients develop AIDS within two years after their infection.9
Unless the original tissue blocks or some of the slides prepared from the biopsies performed during these autopsies could be miraculously located, this will remain a hypothesis. I contacted the Pales family, the Institut Pasteur, the Musée de l’Homme and Le Pharo, and no such material seems to have survived over the past seven decades. Unfortunately, there is no longer an Institut Pasteur in Brazzaville, and whatever archives may have existed seem to have been destroyed during the long periods of civil strife this country went through. So we will never know for sure. But the point that can be made from this story is that the supposed absence of a clinical condition recognised by early twentieth-century doctors cannot be used as a strong argument for dating the emergence of HIV.
Molecular clocks
We have just seen the limitations of what can be extrapolated from reports or publications by colonial-era clinicians. Fortunately, molecular biology offers some help with dating. For readers who are unfamiliar with molecular biology, this is the last section where we will need to talk about these particular concepts.
First we will review what scientists call ‘molecular clocks’. Their principle is simple, and is based on an assumption that the rate of genetic change is fairly constant over time. From a known mean rate of substitutions (which correspond to replication errors: mutations, deletions, etc.) within each gene, it is possible to estimate the chronology of the evolution of the organism. In its simplest form, if we assume a rate of evolution of, say, 0.002 substitutions per nucleotide per year at some given part of a gene and find, say, a 10% (0.1) difference in sequences between two isolates, we can back-calculate that they shared a common ancestor fifty years earlier, after which they diverged, each undergoing its own series of genetic changes. Of course, if the isolates were not obtained at roughly the same time, this will be taken into account.15
Viral recombination between two different subtypes of HIV-1 is the most difficult obstacle for molecular clocks. For instance, CRF02_AG isolates have part of their genome that originated from a subtype A, and part from a subtype G. It is not always clear which is which in the specific parts of the genome whose substitutions will be used to calculate molecular clocks. It is then impossible to evaluate how long the process of evolution has been going on, because the starting point is unclear. The counter-strategy is simple: all inter-subtype recombinant viruses must be excluded. This works well as long as the recombinants are identified, which is not always easy. Other challenges associated with molecular clocks are reviewed elsewhere.15
In a landmark paper that tried to address these limitations through complex analytical strategies, scientists at the Los Alamos National Laboratory estimated that the most likely date for the common ancestor of all HIV group M isolates (that is, all subtypes and recombinants within group M) was 1931. This common ancestor corresponds to the root in phylogenetic trees, the point after which all divergence occurred. As in opinion polls, they calculated a ‘confidence interval’, which in this case was pretty wide, from 1915 to 1941. In other words, there were nineteen chances out of twenty that the true date of the common ancestor was somewhere within this range.16
To verify the validity of this dating, they back-calculated the date of the oldest HIV-1 sequences (ZR59) obtained in Léopoldville in 1959, as well as the date of isolates from Thailand, a country where HIV-1 was introduced in 1986–7 according to extensive epidemiological studies. They dated the Léopoldville sequence between 1957 and 1960, while for Thailand the common ancestor was calculated as having existed in 1986. So in both cases their dating was similar to what could be inferred from historical information, which suggested that their 1931 dating for the common ancestor of HIV-1 group M was relatively accurate.
However, the researchers could not directly address the question of when the cross-species transmission, from chimpanzee to man, had occurred. If the 1931 common ancestor was a human virus, the cross-species event would have occurred in the preceding decade (otherwise the patient would not have survived until 1931). But could the 1931 common ancestor still have been a simian virus at the time? This latter scenario was thought to be highly unlikely since it would have required multiple and close to simultaneous cross-species transmissions, all after 1931, that were all epidemiologically successful. On the contrary, analyses suggested that each group of HIV-1 (M, N, O and now P) represented a distinct cross-species transmission event, and that for pandemic HIV-1 group M the most likely number of such events was one. In other words, the 1931 ancestor indeed lay within a human host – and from this single individual, the true ‘patient zero’, more than sixty million people across the world were subsequently infected!17–20
Then, from a database of HIV-1 isolates from the DRC, and using a sophisticated mathematical approach, the population dynamics of HIV-1 in that country were reconstructed by estimating, year after year, the number of infected individuals. This showed a very slow growth between 1930 and 1940, with an exponential growth later on. Prior to 1930, the number of HIV-1-infected individuals in the Belgian Congo was estimated at somewhere between 0 and 100.17
For a long time, ZR59 was the only ancient specimen of HIV-1. The recent discovery of DRC60, obtained from a lymph node biopsy performed in Léopoldville in 1960, provided additional information. ZR59 and DRC60 diverged by 12%, and were clearly phylogenetically distinct: ZR59 is an ancestor of subtype D while DRC60 is related to subtype A. This implies that HIV-1 group M began to diversify into human populations and entered Léopoldville a few decades before 1960. The inclusion of DRC60 changed the measure of the most recent common ancestor of HIV-1 group M, which was now dated at 1921 (confidence interval: between 1908 and 1933). Other models yielded slightly different dates. Reconstruction of the HIV-1 dynamics in the Belgian Congo was again compatible with a total of fewer than 100 HIV-1-infected individuals for a long time, then a very slow increase in the number of infected individuals until the mid-1950s, when exponential growth supervened.21
This gives us an appreciation of the lack of certainty in such measures. The addition of a single isolate, DRC60, changed the estimates by ten years, and the confidence interval remained wide. In practice, the common ancestor of all the subtypes of HIV-1 group M probably existed sometime in the first three decades of the twentieth century. To keep things simple, I will use ‘1921’ from now on, but this should be viewed as indicating a period rather than a specific year.
For how long has SIVcpz been present in chimpanzee populations? The same methods were used, based on sequences in the Los Alamos database. The results varied according to which gene was used for the molecular clocks and lacked precision. By and large, they suggested that the emergence of SIVcpz in chimpanzees preceded its cross-species transmission to humans by something in the order of a few hundred years, not thousands or tens of thousands. This is probably why SIVcpz is not found in P.t. verus and P.t. ellioti populations: the virus appeared long after these subspecies diverged from P.t. troglodytes.22
To summarise, there is compelling evidence that the common ancestor of HIV-1 existed in a human being sometime in the first three decades of the twentieth century, and that the whole group M pandemic was started by a single cross-species transmission. Now we will try to understand how the virus crossed species, what happened after 1921,
and how HIV-1 eventually expanded into a global pandemic.
4 The cut hunter
The next question to be addressed is: how did the virus cross species to infect humans? How did the simian immunodeficiency virus of P.t. troglodytes chimps become the human immunodeficiency virus type 1? Again, science started out with an intuition: this must have occurred through the handling of chimpanzee meat by hunters, or their wives who would cut up the animals before cooking them. We will now examine whether this theory remains plausible after reviewing the various pieces of evidence accumulated over the past decade.
Hunters and their prey
Hunters and/or cooks can acquire infectious agents from their prey, including primates. For instance, Herpes B virus is a rare but highly lethal infection of individuals who handle monkeys, and especially laboratory technicians working with rhesus and cynomolgus macaques. Monkeypox is a smallpox-like but benign viral infection associated with exposure to monkeys. Highly lethal Ebola and Marburg haemorrhagic fevers have been reported in veterinarians and villagers who handled the carcasses of apes that had died in the wild from these infections. Recently, a retrovirus called simian foamy virus (SFV) has been associated with human exposure to monkeys and apes, and its sequencing allows the identification of the exact simian source. American veterinarians and animal caretakers working in primate centres or zoos were found to be infected with SFV acquired from chimpanzees. Fortunately, this virus does not seem to be pathogenic for humans, and no person-to-person transmission has ever been documented.1–3
Because of their intelligence, agility and aggressiveness, chimpanzees have no predators in the forest apart from leopards and, of course, humans. However, they are by no means easy prey for hunters, and rarely a specific target unless the apes had the bad idea of destroying the crops in plantations near the villages. Although they might occasionally fall into wire snares, lianas knots, nets or pits of all kinds set up for other game animals, chimpanzees can escape from many such traps. Pygmies hunted year round with bows, crossbows and assegais. Targeting chimps would have been a dangerous undertaking, but the fact that pygmies could hunt elephants without firearms (albeit at considerable risk) was testament to their skills. For reasons still unclear, HIV-1 has remained remarkably rare among pygmies, and when present seems to have been acquired through their contacts with Bantus rather than from apes. Bantus hunted mostly during the dry season when the forest was easier to penetrate and when they had less farm work to do and needed additional sources of food until the next crop. If the hunt was successful, carcasses of great apes were generally first cut up in the forest to make them easier to carry, and then cut into smaller pieces in the village before being sold and eaten.4–8
It is difficult to hunt chimpanzees without firearms, and the firearms with small pellets generally available in the bush are not powerful enough to kill an ape. French colonisers had to deal with a number of armed rebellions from populations who opposed their rule, especially in the south-west of Oubangui-Chari and adjacent areas of the Moyen-Congo. Therefore, regulations made it difficult for the natives legally to procure powerful weapons. In France‘s annual reports to the League of Nations concerning Cameroun Français, the exact number of firearms and bullets imported into the country was spelled out (for instance, in 1922, 789 shotguns, 41 revolvers and 6,740 kg of ammunition). Africans could, however, own locally made piston firearms for small game hunting.6,9,10
In the Belgian Congo, regulations were looser. In 1927 an astounding 122,804 firearms permits were issued. This number gradually increased to 245,644 (a quarter of a million!) by 1945. The situation was made worse by a regulation forcing employers to provide their workers with meat at least once a week: hunting was much cheaper than farming. By 1925, Professor Leplae of Université Catholique de Louvain, an expert who also worked for the Ministry of Colonies, was complaining that many species of game animals were being butchered at such a rate that extinction would soon ensue. He estimated that 25,000 elephants were killed each year in the Belgian Congo, often with automatic rifles. Thirty years later, game animals of all kinds were indeed nearly extinct, and this was attributed to a combination of factors: the regulations concerning meat for workers, the development of large cities which created lucrative markets for hunting products, the widespread availability of firearms and all kinds of materials that could be used for trapping, the bad example set by some European hunters and the disappearance of customary hunting regulations through which an equilibrium was maintained between human populations and game animals in the pre-colonial era.11–13
In edicts in April 1901 and December 1912, the Belgian Congo‘s governor prohibited the hunting of a number of animals, including chimpanzees, by both natives and Europeans. The ban was upheld in subsequent amendments of the law in 1934 and 1937, which divided game animals into four categories: gorillas belonged to category I and could not legally be hunted apart from what was required by scientific institutions, while chimpanzees, listed in category II, could be hunted by those who could afford to buy the appropriate, more expensive, permit. A tax also had to be paid for each animal killed: 1,500 francs ($30) for a P.t. schweinfurthii, and 3,000 francs for a Pan paniscus. That was far more expensive than the 50 francs charged for a monkey but a lot less than the 25,000 francs to be paid for a white rhinoceros. In practice, only expatriates could afford to pay such a high tax for killing a chimp. However, the extent to which the regulation could be enforced in such a huge territory was an entirely different matter.14
In French colonies of Africa, decrees issued in April 1930 and November 1947 prohibited the hunting of some species including chimpanzees. Sanctions for offenders included fines from 50 to 2,000 francs, confiscation of firearms or jail terms from six days to six months. The French colonial administrations did not have either the human resources or the will to enforce such regulations in remote, self-subsistent, communities but these decrees made it more difficult for hunters to sell chimpanzee meat openly in the markets of small provincial towns or in the workers‘ camps of private companies established in rural areas for logging or agriculture.15
Furthermore, in several ethnic groups of central Africa there were traditional cultural taboos against the consumption of chimpanzee meat because of their similarity to humans. For instance, among the Bayombe of the DRC and the Bakota of Gabon, eating apes is culturally prohibited for fear that women will give birth to apes, or to children with a simian face. In the Equateur region of the Belgian Congo, the bonobo was considered a human which had been transformed in the distant past.6,16–18
Thus, apart from pygmies, who lived in the forests, hunted daily and were not too concerned about government regulations (but were limited by their lack of firearms), the above factors tended to limit the number of natives who might potentially handle chimpanzee carcasses and acquire SIVcpz from scratches or cuts on their hands. This may have changed in recent decades, as human populations increased and moved deeper into forested areas, using the dirt roads built by logging companies.
Quantifying the exposure
Now we will try to estimate the number of individuals who could have been occupationally infected with SIVcpz in the 1920s. We will need to review findings from several studies which we will then assemble.
A few years ago, researchers visited remote villages in the Cameroonian rain forest and identified people who reported having had contacts (bites, scratches, wounds or other injuries) with animals at any time in their lives. This mostly involved a contact with small monkeys or non-primate animals, from rats to leopards and elephants. Twenty-nine individuals reported contact with gorillas or chimpanzees, up to fifty-three years earlier, and some had the scars to prove it. Antibodies against SFV, an innocuous retrovirus highly prevalent among apes and monkeys, were more common among villagers exposed to apes than those exposed to monkeys, presumably because the wounds, bites or scratches inflicted by the former were more severe.1
The exposure to apes was better quantified in seventeen other Cam
eroonian villages, where almost 4,000 adults were interviewed. A large majority of the exposures to primates involved monkeys rather than apes. Hunting was limited to men, 10% and 12% of which reported having hunted gorillas and chimpanzees respectively. Similar proportions of men and women reported having butchered these apes at least once in their lives. However, when asked about direct contact with primate blood or saliva (scratches, bites or other injuries) during hunting and butchering, only four men reported such exposures to chimpanzees, and seven to gorillas. No woman did so. In other words, only 0.1% of adults reported having had at least one direct contact with chimpanzee blood or body fluids, and 0.2% with gorillas.19
This enables us to estimate roughly the number of individuals living in the relevant parts of central Africa around 1921 who might have had at least one contact with SIVcpz-containing chimpanzee blood at some point in their lives. Let us assume that no exposure occurred among children under sixteen, that only exposure to chimpanzees was relevant (the prevalence of SIVgor infection in gorillas is lower and its relevance to human transmission unclear) and that the frequency of lifetime exposure to chimpanzee blood among inhabitants of areas populated by P.t. troglodytes was the same in 1921 as in the recent past.
Around 1930, when reliable censuses became available, about 2.3 million persons lived in the areas inhabited by P.t. troglodytes: 900,000 in Cameroon south of the Sanaga River, 387,000 in Gabon, 664,000 throughout Moyen-Congo, 120,000 in continental Equatorial Guinea, 130,000 in the south-west of the Central African Republic, and at most 100,000 in the Cabinda enclave and the adjacent Mayombe area of the Belgian Congo. At the time, the natural growth of central African populations each year (the difference between the birth and death rates) was 0.6%, so we can estimate that the relevant populations in 1921 were around 2,177,000. Of all inhabitants, 62% were aged sixteen years or over (the population was older than today because of the high child mortality). If we multiply 2,177,000 by 62% by 0.1%, it can be estimated that 1,350 adults living in 1921 had been exposed to chimpanzee blood at least once in their lives.