HIV’s high level of genetic diversity, the large difference in nucleotide sequence from strain to strain of the virus, presents a particular challenge to the HIV vaccine field. Even if a vaccine existed that perfectly protects against infection from one HIV strain, there are no guarantees it would also protect against other strains. Further complicating matters, the virus can mutate incredibly quickly within its human host, so “escape mutations” that allow a given strain to evade the immune system’s grasp are common.
To understand which of the many genetic variations in HIV are important and are likely to affect vaccine efficacy, researchers need sophisticated computational tools to compare and contrast multiple different viral sequences. Dr. Paul Edlefsen, new assistant member in VIDD, is creating tools to aid in this comparison, called multiple sequence alignment.
“For highly evolving viruses, the immune system might be great at protecting against what you told it to protect against with the original vaccine, but the virus will just adapt away from that,” Edlefsen said. “Ultimately understanding how the virus adapts could help us come up with strategies to develop a vaccine that would thwart that escape.”
One possible way that comparing HIV sequences could help in vaccine design is by identifying which regions are conserved among multiple strains of HIV, meaning those regions are too important to change drastically. A vaccine constructed against those important regions might be harder for HIV to escape by mutating. “Although this sounds easy, it’s really not,” Edlefsen said.
Sequence alignment is an important tool to many fields of biology, not just HIV research, but it’s not always a straightforward procedure. Especially among related sequences with high levels of variation, such as in HIV, it can be difficult to see which regions of one sequence are equivalent to those in another. Understanding how these sequences evolved from a common ancestor can go a long way in helping this alignment, Edlefsen said. But in many cases researchers don’t know which nucleotides existed in the evolutionary ancestor and which regions mutated more recently. So most sequence alignment tools just take their best guess.
“It’s a chicken and egg problem,” Edlefsen said, because evolutionary trees are used to construct alignments, and those alignments are then used in turn to make judgments about the different sequences’ evolution.
During his doctoral research at Harvard University, Edlefsen approached this problem looking not at HIV sequences, but at transposons, self-replicating pieces of DNA scattered throughout the genome that are thought to be ancestral remnants of old viruses. Like HIV, transposon sequences are highly variable, and picking out a transposon from background DNA sequences can be tricky. To better identify new transposons, Edlefsen adapted a standard statistical tool, called profile hidden Markov models, to create several different sequence alignments instead of just one. His computational approach then looks at that distribution of alignments and creates a distribution of evolutionary relationships between the sequences for each possible alignment.
Eventually, Edlefsen hopes to design tools that will simultaneously create distributions of both sequence alignments and evolutionary relationships, so that the alignments don’t have to depend on faulty assumptions of ancestry and evolution. He then plans to apply his approaches to HIV sequences. Currently, Edlefsen is collaborating with VIDD member Dr. Peter Gilbert on ongoing “sieve analysis” of HIV vaccine trials. The researchers compare the HIV strains of placebo and vaccine recipients that became infected during the trial, to see whether the tested vaccine influenced the types of HIV strains that were able to infect. In this way, the scientists can learn whether even those candidate vaccines that did not reduce overall rates of HIV infection had some influence on the virus, and can use that information to inform better design of future vaccines.
In the future, Edlefsen hopes to use these computational approaches to study how HIV evolves within a single human host to continually evade the immune system, using sequence alignments of either viral or human sequences to better understand both how the virus adapts and the differences in different people’s immune systems, information that would also help inform vaccine design. “I’m hopeful that the specific kinds of genomic research that I’ve been doing will be helpful here,” Edlefsen said.