Researchers are starting to apply the tools of systems biology to better understand human immune responses to vaccination, including HIV vaccine candidates
By Andreas von Bubnoff
Increasingly, researchers are using a more integrative approach to understand complex biological systems, such as the human immune system. Systems biology, as this approach is called, has only recently become possible because of the availability of technologies such as microarrays that can simultaneously measure the expression of many of an organism’s genes. This results in the so-called transcriptome, the level of RNA transcripts of some or all of the genes expressed in a cell. It is now also possible to measure the levels of all of the proteins or metabolites in a cell, resulting in the proteome or the metabolome, respectively.
One advantage of a system-wide approach is that it collects a massive amount of data without requiring researchers to operate on hypotheses. “You accumulate a lot of data [in an] unbiased way, which I think is a wake-up call compared to what everybody has been doing,” says Rafick Sékaly, scientific director of the Vaccine & Gene Therapy Institute in Florida. Michael Katze, a professor of microbiology at the University of Washington, has a similar view. “The traditional reductionists will say that [we will] get the answers by doing empirical reductionist biology. The people who are more converted to systems biology like me will say we will only get the answers when we have enough data. In my mind it’s an unacceptable failure that over 25 years after the discovery of the AIDS virus and all the billions of dollars that have been spent on it we don’t understand what a good vaccine is. My argument, my hypothesis, is that we don’t know anything, so let’s discover what’s going on. Let’s take more of a holistic rather than a reductionist approach.”
But just getting people to agree on the definition of systems biology can be a challenge. “It means different things to different people,” says Stephen Popper, a research associate at Stanford University. “To some people, doing microarrays or some other sort of expression profiling is systems biology,” he says. To others, he adds, it is integrating different types of information such as gene expression and proteomics, or building dynamic networks.
Bali Pulendran, a professor of immunology at Emory University, says systems biology is much more than microarrays and genomics. “The real goal is to use high throughput data such as microarrays to understand the behavior of complex biological systems and ultimately predict the behavior of such systems, including vaccine-induced host defenses against HIV.”
Other areas of biomedical research, such as cancer research, have been using the tools and concepts of systems biology for some time. To encourage the HIV vaccine field to explore possible uses of this new technology, the Global HIV Vaccine Enterprise in 2008 helped organize one of the first meetings that brought together systems biologists and HIV vaccine researchers. “I think there has been a silo between people who do systems biology and people who do HIV vaccinology, and I think one of our jobs is to help bring people together and encourage collaboration,” says Alan Bernstein, executive director of the Global HIV Vaccine Enterprise. Katze agrees. Systems biology projects, he says, involve “a lot of people who don’t normally talk to each other—clinicians, biostatisticians, virologists. It defies the way classical experimentalists have grown up.”
Researchers are now using the tools of systems biology to characterize the immune responses induced by existing vaccines and identify the gene expression changes that occur following vaccination. They are also beginning to apply systems biology approaches to the development of HIV vaccine candidates and to better understand HIV pathogenesis.
Responses to existing vaccines
Recently, Pulendran’s and Sékaly’s labs used a systems biology approach for the first time to study the immune response induced by a vaccine (1,2; see Research Briefs, IAVI Report, Jan.-Feb. 2009). Using microarray analysis, they measured gene expression changes in the innate immune response to the yellow fever vaccine, starting days after vaccination. The vaccine against yellow fever is one of the most effective vaccines ever developed. “The reason why we decided to do yellow fever was because it works, so we wanted to understand what a good vaccine looks like,” says Alan Aderem, director and cofounder of the Institute for Systems Biology in Seattle, who collaborated on Pulendran’s study.
The studies found a group of genes, or a signature, which was consistently expressed more strongly after yellow fever vaccination. Pulendran’s group also showed that this signature could be used to predict the level of the later adaptive B- and T-cell immune response to the vaccine.
Researchers in Sékaly’s group have also identified a similar gene expression signature in yellow fever vaccinated rhesus macaques and plan to test whether this signature can predict protection. They are also trying to see if the gene expression signatures induced by the yellow fever vaccine are absent in people who got sick despite yellow fever vaccination. This is not easy, however, because almost all yellow fever vaccinees are protected.
The yellow fever findings show that, in principle, it is possible to use gene expression signatures as an early biomarker to indicate whether a vaccine will work. This could save time in vaccine development. “The idea is a biomarker that [lets] you have an earlier endpoint for measuring how well your vaccine is working,” Popper says. “[It] gives you a readout that you can use as you go [to] some next stage of development. If you collect all the right samples and identify a relevant signature, then the next round of development will profit from increased information and hopefully shorten the time cycle.” Aderem agrees. “[A signature] will be critical because you’d be able to assess early on whether or not a vaccine is a good one or not,” he says.
|Building Models of Complex Biological Systems
One goal of systems biology is to develop models of complex biological systems. “It may be quite challenging to develop all-encompassing models for complex physiological systems such as the immune system,” says Leor Weinberger, a biochemist and virologist at the University of California in San Diego. “But a number of researchers have already developed simple models that are effective and predictive.”
Weinberger has used computer models to predict how HIV Tat protein expression influences the expression of HIV genes and then verified the predictions in experiments. Also, Alan Aderem, director and co-founder of the Institute for Systems Biology in Seattle, and colleagues have analyzed and simulated the signaling cascade activated by Toll-like receptors, which is part of the innate immune response to bacteria and viruses.
Others are using computer models that simulate components of the immune system itself. Vladimir Brusic, director of the cancer vaccine center bioinformatics core at the Dana-Farber Cancer Institute, has been involved in a project that used a model of the immune system that contains about 50-100 different elements such as the major cell types, cytokines, and other molecules of the immune system to predict the immune responses to breast cancer vaccine regimens in mice. The model predicted that the number of vaccinations could be reduced by about 40% without a loss in vaccine efficacy. Subsequent experiments showed that the prediction was largely correct. In principle, Brusic says, this approach could also be used to assist in the development of HIV candidate vaccine regimens.
A recent study led by Arup Chakraborty, a professor of chemistry, chemical engineering, and biological engineering at the Massachusetts Institute of Technology, and Bruce Walker, a professor of medicine at Harvard Medical School, used computer models of relevant parts of the immune system to predict that CD8+ T-cell repertoires that are more cross reactive to different HIV peptides lead to better control of viral load (3; see Research Briefs, IAVI Report, May-June, 2010). The study then validated the prediction with data from elite controllers and HIV-infected people with progressive infection. —AvB
Researchers have also begun to study other licensed vaccines to determine whether they induce similar gene expression signatures to those induced by the yellow fever vaccine. Influenza vaccination is the focus of the first major project at the Center for Human Immunology, Autoimmunity and Inflammation (CHI), that was recently established by the US National Institutes of Health (NIH) to use systems biology tools to analyze, in unprecedented detail, the normal and perturbed state of the human immune system. CHI-affiliated researchers are collecting data from more than 150 people of different ages and ethnic backgrounds who have been vaccinated with the killed flu virus. Researchers will collect data prior to vaccination to establish baseline measurements and help define the normal so-called “immunome.” They will also collect post-vaccination data on day one for the innate response, on day seven for the adaptive response, and on day 70 for the memory response, according to Ron Germain, deputy chief of the Laboratory of Immunology at the National Institute of Allergy and Infectious Diseases (NIAID), and an associate director of CHI.
In each individual, researchers will analyze, among other things, genome-wide gene expression, more than 60 cytokines in blood, and 700,000 gene variants, known as single nucleotide polymorphisms (SNPs), to help track the effects of host genes on the immune response. In a related study, researchers are also looking at the early innate immune response in blood of flu vaccinees as early as four hours after vaccination. CHI is also considering assessing the immune status of people who are given the live-attenuated flu vaccine, as well as other vaccines with or without adjuvants, Germain adds. “Over the next year or two we will be getting much more information at a much larger scale that will let us set the baseline standards to try to understand what it [looks] like when you have a normal human immune system [or] when you begin to perturb it with various vaccine challenges,” he says.
Pulendran and colleagues are also studying gene expression changes after vaccination with the inactivated or the live-attenuated seasonal influenza vaccines. They have found that gene expression signatures at days three and seven after vaccination with the inactivated flu shot predict the magnitude of the later antibody responses to the flu protein hemagglutinin. They also found that the expression level of a gene called tumor necrosis factor (TNF) receptor superfamily 17, which promotes B-cell maturation, predicts the later antibody response after both yellow fever and flu vaccination. “This is quite exciting to me because it hints at the likelihood that there might be some common predictors in different vaccines,” says Pulendran. “It’s possible that [such genes] may also be predictive of antibody responses against HIV.”
Recently, Pulendran was awarded a grant by the NIH to establish a Center for Systems Vaccinology at Emory University, which will study whether this approach can be used to predict the effectiveness of vaccines, including those against influenza, pneumococcal disease, and shingles. The US$15.5 million center will be established as part of a $100 million nationwide research initiative by the NIH to define changes in the human immune system in response to infection or vaccination.
Pulendran has also shown that gene expression changes can be used to predict the kind of immune cells that are induced by the flu vaccine. Using publicly available microarray data, Pulendran has compiled libraries of genes that are typically expressed in different types of immune cells such as macrophages, B cells, or CD4+ and CD8+ T cells. He has validated these cell-type specific signatures in gene expression data from whole blood in people vaccinated against flu. For example, flu vaccinees express many genes that, according to the library, belong to plasma B cells, and as predicted, Pulendran found more B cells in these vaccinees.
Pulendran also plans to study people who received pneumococcal vaccine, which is given to the elderly to prevent pneumonia, or herpes zoster vaccine, which is given to the elderly to protect against shingles. Because these vaccines don’t protect all vaccinees, it is possible to compare gene expression signatures in protected and unprotected vaccinees. “In the elderly population, a substantial number of people are not protected,” Pulendran says. “We want to use the biomarker approach to identify those people.”
According to Germain, there are also discussions about comparing hepatitis B vaccinees, 80% to 90% of whom are protected, to recipients who receive the standard hepatitis B vaccine with an adjuvant called CpG (an artificial DNA oligonucleotide containing the CpG motif), all of whom are protected. If CHI can participate in such a study, it would include gene expression analysis and genomic analysis of SNPs to see if there are any genetic variants that are associated with a lack of response to the vaccine.
Systems biology and HIV vaccines
While researchers are collecting clues from studies of existing vaccines, they are also beginning to use systems biology approaches to study experimental vaccines, such as HIV vaccine candidates tested in recent clinical trials. Bob Palermo, a principal research scientist in Katze’s lab, is studying genome-wide gene expression in 200 volunteers from the STEP trial, a phase IIb trial of Merck’s adenovirus serotype 5 (Ad5)-based vaccine candidate known as MRKAd5. Palermo is looking for hints about why MRKAd5 failed to protect against HIV infection, and initially even seemed to increase the risk of HIV acquisition in vaccinees that were uncircumcised, had a high titer of preexisting Ad5 antibodies, or both.
Samples from RV144—an HIV vaccine efficacy trial conducted in more than 16,000 volunteers in Thailand that showed a modest 31% protection against HIV infection—will also be analyzed for genome-wide RNA transcriptional analysis by Sékaly’s lab, according to Nelson Michael, director of the US Military HIV Research Program, one of the agencies that conducted the RV144 trial. The initial analysis will include samples from the time of the first vaccination, six months later (two weeks after the last vaccination), which is close to the peak of immune responses, and one year after the initial vaccination. An initial pilot study will only use samples from HIV-uninfected RV144 trial volunteers, and will include samples from 80 vaccine recipients (40 male and 40 female), and 20 placebo recipients (10 male and 10 female). Afterwards, a case-control study will look at possible gene expression differences between approximately 40 vaccinees who later became infected and 200 who did not.
In most vaccinees, the RV144 candidate vaccine regimen, a canarypox vector-based candidate, ALVAC-HIV, administered in a prime-boost combination with an engineered HIV gp120 protein, AIDSVAX B/E, resulted in gp120-binding antibody responses. Michael says that the titers of antibodies that bind gp120 will be determined in 100 RV144 vaccinees that didn’t get infected with HIV. To see if there are any genetic differences between the vaccinees with high antibody responses to the vaccine and those with lower antibody responses, David Goldstein, director of the Center for Human Genome Variation at Duke University Medical Center, will then determine the genome sequence of the 20 RV144 vaccinees with the lowest and the highest antibody titers, respectively. If genetic differences can be identified, they would make it possible to better interpret the efficacy of vaccines in the future, Michael says. “There could be people that just inherently are more able to respond to a vaccine,” he says. “It’s important to know if the deck is already stacked one way or the other against people.”
Researchers are also starting experiments in rhesus macaques to study early gene expression changes in response to experimental vaccine regimens against simian immunodeficiency virus (SIV). One advantage of the animal model is that researchers can later infect, or challenge, the vaccinated animals and compare those that are protected to those that become infected. In principle, this should enable them to find early gene expression changes in the innate immune response to a vaccine regimen that can predict whether an animal will be protected against challenge.
Sékaly’s group is collaborating with Adrian McDermott, director of immunology and vaccine design at IAVI, to analyze gene expression changes in macaques four days, two weeks, and three weeks after vaccination with SIVmac239Δnef, a live-attenuated version of SIVmac239. The researchers found that the macaques that were protected from SIVmac251 challenge a year later had a different early gene expression profile from the ones that became infected. “Both the innate and the adaptive immune responses are quite distinct,” Sékaly says.
Insights into HIV and SIV infection
Researchers also hope that systems biology tools will give them insight into the effects of HIV infection in unprecedented detail, and at earlier time points than ever before. “We really do not know how any virus triggers the events that ultimately lead to virulence and pathogenesis,” Katze says. “Now we have the tools.”
Already, Katze and colleagues have infected CD4+ T-cell lines in vitro with HIV and have looked at changes in the abundance of more than 1,000 proteins as early as a few hours after infection. “Cell lines are an easy start,” says Palermo, adding that ultimately, the goal is to do this kind of analysis with primary human cells. “We would like to do it with truly bona fide target cells [of HIV],” he says. The analyses will then also look at gene expression as early as one hour after infection, Palermo adds.
Damien Chaussabel, an associate investigator at the Baylor Institute for Immunology Research, hopes to correlate gene expression signatures soon after HIV infection with the clinical outcome later on. He does expression profiling of blood samples taken from HIV-infected people at five time points from as soon as one to two weeks after infection, and up to 24 weeks later. The samples, provided by the Center for HIV/AIDS Vaccine Immunology (CHAVI), come from HIV-infected people in Malawi, South Africa, and North Carolina.
Systems biology analyses are also underway to better understand why certain people, dubbed elite controllers, can control viral load below detectable levels for decades without treatment. It is already known that their CD8+ T cells function better than those of progressors, which are thought to be deficient in their ability to multiply or to secrete cytokines (see Research Briefs, IAVI Report, Jan.-Feb. 2009). Consistent with this, Sékaly and Elias Haddad, an associate scientist at the Vaccine & Gene Therapy Institute in Florida, have done gene expression analyses that suggest that HIV-specific T cells from elite controllers survive longer than the same cells from chronic progressors. Next, they want to see if this difference is also present at the protein level, and if the T cells show corresponding functional differences. If gene expression in elite controllers is found to be similar to gene expression in people vaccinated with yellow fever vaccine, it would suggest that it might be possible to eventually use such gene expression signatures of elite controllers as a guide to assess HIV vaccine candidates.
W. Nicholas Haining, an assistant professor of pediatrics at the Dana-Farber Cancer Institute and Harvard Medical School, says that gene expression analysis of antigen-specific CD8+ T cells from elite controllers and progressors has allowed him to identify new classes of genes that have not previously been associated with T-cell dysfunction in HIV.
Researchers are also using transcriptional profiling to understand why some nonhuman primates (NHPs) such as African green monkeys (AGMs) don’t get sick from SIVagm infection, while other NHPs such as pigtail macaques do. Looking at gene expression changes in cells taken from the colon, lymph nodes, and blood, Katze found that while SIV induces immune activation and interferon-stimulated genes in both AGMs and pigtail macaques, these responses only resolve in the AGMs. In addition, AGMs have more active cell survival pathways (4). Other groups made similar observations when they compared gene expression in SIV-infected AGMs or sooty mangabeys—neither of which get sick when infected with SIV—with rhesus macaques, which do get sick (5,6). They also observed that activation of interferon activated genes was transient in the species that don’t get sick, and chronic in rhesus macaques.
New tools, new questions
As researchers are implementing systems biology approaches, the tools available to them keep evolving. To analyze the expression of thousands of genes, for example, many researchers currently use microarrays, which only probe for the expression of the genes that are known to be encoded by RNAs. But such analyses are already being replaced by another method called RNAseq, which uses ultrafast high-throughput next-generation sequencing to sequence and also count all RNAs in a sample, including ones whose functions are completely unknown.
Recently, Katze used RNAseq to sequence all RNAs expressed in the lungs of severe acute respiratory syndrome (SARS)-infected mice. The results were rather “scary,” he told the audience at this year’s Conference on Retroviruses and Opportunistic Infections, which took place in February. “You see all these novel RNAs that are not annotated to anything in the genome,” Katze said. “Nobody knows what the hell these RNAs do.”
Systems biology will probably enable researchers to make a lot of new discoveries, says Louis Picker, a professor at Oregon Health & Science University, who works with Sékaly and Aderem on systems biology analysis of candidate SIV vaccine regimens in rhesus macaques. “It’s like Christopher Columbus—you can see that there is a whole new world out there but you don’t know quite what’s out there yet.”
1.J. Exp. Med. 205, 3119, 2008
2. Nat. Immunol. 10, 116, 2009
3. Nature 465, 350, 2010
4. PLoS Pathog. 5, e1000296, 2009
5. J. Clin. Invest. 119, 3544, 2009
6. J. Clin. Invest. 119, 3556, 2009