Gut microbiome may have interfered with experimental HIV vaccines

Dr. James Kublin, executive director of the HVTN, which is headquartered at Fred Hutchinson Cancer Research Center, said the study “provides profound insights into the nature and evolution of human immune responses.”

“As soon as we enter the world … we’re constantly exposed to microbes that we have to identify as non-self but also not overreact to, otherwise our bodies would be in a constant state of immune activation and battling the microbial communities that are all around and within us,” said Kublin, who was not an author of the microbiome study but helped lead the original HVTN 505 trial. “There is this enormous impact that the microbiome has on the evolution of our immune system.”

A viral evasion

Haynes, Williams and their colleagues, including Duke and HVTN researcher Dr. Georgia Tomaras and Fred Hutch and HVTN statistician Dr. Peter Gilbert, set out to study the immune responses of HIV-negative volunteers from HVTN 505 and smaller trials that had tested the same vaccine. The Duke researchers looked at the immune molecules known as antibodies in trial participants’ blood.

They were surprised to find that the antibodies elicited by the vaccine were the kind of molecule typically made early in life — not following a vaccine received as an adult.

“We thought that was strange,” Haynes said.

His team had seen similarly atypical antibodies in another setting: natural HIV infection. The researchers previously found that people infected with the virus make antibodies that recognize both HIV and gut bacteria. They hypothesized that by mimicking natural gut microbes, HIV evades our immune system because the immune cells already know not to reject helpful bacteria.

At the time, the scientists thought that making these antibodies may be a process unique to HIV-positive people, Haynes said, because HIV infection damages the gut and prompts a large immune response. So they were surprised by what they found in HIV vaccine study participants who’d remained uninfected.

When they tested the antibodies from several people who’d received the experimental vaccine, they saw a similar phenomenon. The vaccine boosted production of primarily a class of antibodies that bound both an HIV protein and several different intestinal bacteria.

And they also found microbe- and viral protein-binding antibodies in a blood sample taken from one study participant before that person received the vaccine, showing that the primary vaccine-driven immune response comes from a pool of B cells (the immune cells that make antibodies) that already exists in the body.

Even though the vaccine tested in HVTN 505 spurred people to make antibodies against an HIV protein, those antibodies clearly didn’t prevent infection, and that could be because people’s bodies were trained in early childhood not to eliminate healthy bacteria — or, by extension, anything that looks to the immune system like those healthy bacteria.

“As yet it remains a hypothesis that it’s the microbiome driving this,” Haynes said. “We do know that the vaccine … is able to recruit this preexisting pool of B cells.”

What vaccine researchers can learn from ‘failure’

Although the vaccines tested in HVTN 505 and every other — but one — HIV vaccine efficacy trial to date didn’t protect people against infection, vaccine researchers don’t call these trials a failure. The studies are purposely designed so researchers can study how people respond to each vaccine, information that will ultimately help design a working vaccine, Gilbert said.

“The vaccine didn’t work and so it’s important to characterize what immune response the vaccine induces so we can learn what a vaccine looks like that doesn’t work,” he said. “This gives us rational criteria for screening out vaccine candidates … There are so many aspects of the immune system to measure. This study is useful because it suggests markers we should consider using more routinely.”

Going forward, Haynes proposes three possible solutions to get around the B cell diversion. The vaccine tested in HVTN 505 contains several different pieces of the virus, but the antibodies that cross-reacted to the vaccine and to bacteria bound to just one of those pieces, a section of the HIV protein known as gp41.

So it’s possible that simply leaving that piece out of future vaccines would circumvent the problem. The so-called Thai trial — the only vaccine trial to show any hint of protection against HIV/AIDS — didn’t have gp41 in its vaccine. Alternatively, there may be ways to mask that part of the vaccine so that the immune system doesn’t recognize it, Haynes said.

Finally, the researchers propose inoculating infants rather than adults, as young children may not yet have established the ineffectual immune cells that seem to have diverted the HVTN 505 vaccine from working. Haynes and his team are currently testing that idea in animal studies, he said. IMPAACT, a pediatric HIV trials group, plans to start an HIV vaccine study in the next year including children potentially young enough to test the microbial influence hypothesis.

“However, whether or not getting around this problem would result in an efficacious vaccine is yet to be tested,” Kublin said.

Other groups are also testing whether giving people or animals probiotics or antibiotics influences how candidate HIV vaccines work, Kublin said. Based on the findings from Haynes’ group, the HVTN has now made a specific tweak to all its ongoing trials: “The bottom line is we’re now collecting poop from all of our study subjects,” Kublin said, with the aim of adding gut microbiome characterization to immune response studies for HVTN trial volunteers.