Because I work on HIV vaccine research, I often talk about vaccines and HIV vaccine design in particular. So far, there have been several phase I HIV vaccine trials, but very few have made it into phase III. One such case was the STEP trial, which was abruptly halted in 2007 after preliminary results showed that not only the vaccine was not protecting people from getting the virus, but the rate of HIV infections was actually higher in the vaccinated subjects compared to the subjects that received a placebo. Even more alarming was that this increase in acquisition risk lasted years after vaccination.
What went wrong in the STEP trial?
Vaccines are made of a "wimpy" version of the virus: you have to use enough genetic material from the virus in order to induce antibody production, but not enough to start an infection. We call the modified virus used in a vaccine "immunogen." The immunogen is only one part of the vaccine "recipe", the other part is what we call a vector, a structure that carries the immunogen and presents it to the immune system. Viruses make excellent vectors because they are like little "boxes" that are programmed to enter cells. And of all possible viral vectors, the most often used are adenoviruses because they are very common in the human population (they cause the common cold) and are therefore considered to be safe to "hijack" into carrying vaccine immunogens.
The STEP HIV vaccine was made of an adenovirus vector (recombinant adenovirus serotype 5 or rAd5) expressing the HIV proteins gag, net, and pol. When researchers looked back at what could've possibly gone wrong they found that the rates of infections were significantly higher in subjects that had been previously infected with Ad5 and had preexisting immunity against Ad5.
The HIV community feels so baffled by the failure of the STEP vaccine trial that at a recent conference I attended, the director of the Fred Hutchinson Cancer Research Center said quite vehemently that we should all move away from vector vaccines and do DNA vaccines instead. Since DNA is naturally absorbed by cells, DNA vaccines bypass the need of a vector.
In truth there's still strong hopes for vector vaccines. The natural question to ask in light of what happened with the STEP trial is: can we use a vector that instead of worsening the immune response actually makes it better?
It turns out that there is, and it's called Cytomegalovirus, or CMV. Like adenoviruses, CMV's are also very common in the human population and typically asymptomatic unless there are other underlying conditions.
If you remember roughly how the immune system works, we have two kinds of "sentinels" looking out for invaders: B-cells, which produce antibodies, and T-cells. While antibodies bind to viral particles, thus preventing the virus to enter and infect cells, T-cells recognize infected cells and destroy them. This recognition mechanism is based on the fact that infected cells express fragments of viral proteins (epitopes) on their surface. The T-cell recognizes those proteins as foreign and as a flag of infection and thus kill the cell before it starts replicating the virus.
Eliciting antibodies able to clear the HIV virus through a vaccine has proven very challenging (I discuss why in this post). But what about T-cell vaccines? In  Hensen et al. showed that a CMV vector SIV vaccine was able to elicit over three times greater breadth T-cell response in rhesus monkeys and about 50% of the vaccinated animals, once challenged with SIV (the simian version of HIV) were able to clear the infection without getting sick.
The vaccine was made of a recombinant rhesus monkey cytomegalovirus (strain 68-1 RhCMV) engineered to express simian immunodeficiency virus (SIV) genes.
"The key finding of Hansen et al. is that strain 68-1 RhCMV elicited CD8+ T cell responses that target SIV epitopes that were completely different from those generated by SIV infection itself, by other virus-based vectors, or by wild-type RhCMV expressing SIV genes ."Typically during an HIV infection, the immune system starts producing T-cells that attack a limited number of epitopes, in other words a limited number of viral protein fragments that infected cells express on their surface. So, the key finding in this study was that using a CMV vector increased the number and variety of epitopes that the T-cells were able to recognize.
"We conclude that RhCMV has an intrinsic ability to elicit CD8+ T cell responses to unconventional epitopes, distinct in quality and quantity from all infectious agents studied to date. ."As you know, HIV's winning strategy to evade the immune system is its ability to "hide" by constantly changing its genetic structure. This is favored by the fact that under normal circumstances T-cells recognize only a limited number of epitopes. In this light you can see why increasing the magnitude and breadth of the T-cell responses is effective in defeating the virus: once primed with the CMV vector, T-cells were not only able to recognize many more epitopes, but different "versions" of such epitopes, meaning that even when the virus came up with a mutation at a certain epitope, the T-cells were still able to recognize it and kill the infected cell.
These are remarkable results and I can't wait to follow this story as it moves to its next step -- human clinical trials.
 Nilu Goonetilleke, Andrew J. McMichael (2013). Antigen Processing Takes a New Direction Science DOI: 10.1126/science.1239649
 Scott G. Hansen, Jonah B. Sacha, Colette M. Hughes, Julia C. Ford, Benjamin J. Burwitz, Isabel Scholz, Roxanne M. Gilbride, Matthew S. Lewis, Awbrey N. Gilliam, Abigail B. Ventura, Daniel Malouli, Guangwu Xu, Rebecca Richards, Nathan Whizin, Jason S. Reed (2013). Cytomegalovirus Vectors Violate CD8+ T Cell Epitope Recognition Paradigms Science DOI: 10.1126/science.1237874