I’ve talked a fair bit (like here and here) about the evolution in situ that HIV undergoes in each of its hosts, evolving rapidly in response to local conditions such as immune responses. Much of the variation in the infected individual is selected, of course; selected in response to local conditions — those local conditions being the genetics of the host, and to a large extent the host’s immune system. The immune system puts tremendous pressure on the virus, and the only way it can escape from the prison is to cripple itself. HIV immune escape variants are usually relatively defective viruses, because the mutations that allow them to become invisible to the immune system, damage the virus’s ability to replicate and spread.
HIV infecting a macrophage2
Immune systems are idiosyncratic; yours is different than mine. When HIV is transmitted from one person to another the virus moves from one selection immune landscape to a very different one. The mutations that saved it in the first person are now probably no longer protective, yet the damage that those mutations were doing is still very much present - a double whammy. Fortunately for HIV (less so for humanity) it is still able to mutate its way back to a functional virus. If you examine HIV in one individual and then in the next step in the transmission chain, the new host’s virus will probably have started to revert back to the platonic essence of HIV; less, of course, the mutations that the new host’s immune system imposes on it.
A question is: What does the virus have to work with in this process? We speak of HIV as a quasi-species, traveling around as a cloud of related but different viruses. Within that cloud is variation that can be immediately selected. But is that true in transmission? How many viruses actually take that gigantic leap from one host to the next?