Engineering Immunity

Researchers are developing new approaches to introduce genes for antibodies as a novel way to protect against HIV infection

By Andreas von Bubnoff

“We really ought to try heroic measures because we have nothing to lose,” says David Baltimore, a professor at the California Institute of Technology (Caltech), referring to AIDS vaccine research. Baltimore, who was a co-recipient of the Nobel Prize in 1975 for his work in the discovery of reverse transcriptase, is pushing the envelope in HIV research by pursuing gene therapy in an effort he refers to as “engineering immunity” against the virus.

So far researchers have had little success in inducing protective immunity against HIV. Early approaches to induce neutralizing antibody responses against the virus were unsuccessful, causing many researchers to shift strategies and focus instead on candidates that induced primarily cellular immune responses against HIV. But after the failure of Merck’s T-cell based vaccine MRKAd5 last year—regarded as one of the most promising cellular immunity vaccine candidates—this approach was also called into question. “I suspected that T cells were not going to be the whole answer,” says Baltimore. “I didn’t think they would turn out as badly as they did,” he adds, referring to the failure of MRKAd5. “Antibody [approaches weren’t] going anywhere [either], so it looked to me like we were in a position where it was possible we were going to end up with no AIDS vaccine,” Baltimore explains.

While many researchers are now focusing on innovative ways to develop improved AIDS vaccine candidates that can induce both neutralizing antibodies and cellular immune responses against HIV, Baltimore and others, including Philip Johnson, chief scientific officer at the Children’s Hospital of Philadelphia, are trying a novel approach. They are developing ways to introduce genes encoding antibodies into people that are capable of neutralizing a variety of HIV isolates in vitro. “The more standard measures [to develop an AIDS vaccine] were getting a lot of attention from a lot of people,” Baltimore says. “If there was a way to make them work they were going to get them to work, so they didn’t need me for that.”

Both Baltimore and Johnson are using viral vectors to introduce antibody genes, so far with varying degrees of success. Baltimore is leading a gene therapy project that uses an HIV-derived lentivirus as a vector to carry antibody genes into bone marrow-derived hematopoietic stem cells (HSCs). These, he hopes, will then become antibody-expressing B lymphocytes. Meanwhile, Johnson is working on a gene transfer approach, using an adeno-associated virus (AAV) as a vector to carry antibody genes. Johnson is collaborating with Ron Desrosiers, a professor at Harvard Medical School, who is using a monkey herpes virus vector to introduce antibody genes into cells.

So far Baltimore and his collaborators have been able to successfully incorporate genes into HSCs of mice, and are developing ways to evaluate the expression of antibody genes in human B cells after introducing them into HSCs. Meanwhile Johnson has shown expression of antibody genes in muscle cells of nonhuman primates that offer some protection against simian immunodeficiency virus (SIV).

Identifying antibodies

Although researchers are actively searching for broadly neutralizing antibodies against HIV, only about five have been identified so far from HIV-infected people. There is some disagreement among researchers about how many different isolates an antibody has to neutralize to earn the classification of being broadly neutralizing. “It depends where you draw the line,” says Dennis Burton, professor of immunology and molecular biology at the Scripps Research Institute. Even though the handful of already identified antibodies has been well studied, it’s still unknown how to induce them through vaccination. “Nobody knows how to design an immunogen that you can inject into people to give rise to these responses,” Baltimore says.

But there is evidence that if you could induce them in humans, they might do the trick. In passive immunization experiments, administration of the already identified broadly neutralizing antibodies has been shown to protect humanized mice—mice engineered to have human immune cells—as well as rhesus macaques from challenge with HIV or a hybrid simian/human immunodeficiency virus (SHIV), respectively (Nat. Med. 3, 1389, 1997; J. Virol. 73, 4009, 1999; Nat. Med. 6, 207, 2000; J. Virol. 75, 8340, 2001; Nat. Med. 9, 343, 2003).

Generally, relatively high serum antibody concentrations are required to provide complete protection in monkeys, according to Burton. In one study, intravenous transfer of 25 mg/kg body weight of the human broadly neutralizing monoclonal antibody b12 six hours prior to challenge protected four out of four monkeys from vaginal SHIV challenge, but a smaller dose of five mg/kg body weight only protected two out of four monkeys (J. Virol. 75, 8340, 2001).

Similar challenge studies are obviously impossible in humans, but infusing antibodies into people already infected with HIV has been shown to delay rebound of viral load after interruption of antiretroviral (ARV) therapy (Nat. Med. 11, 615, 2005; J. Virol. 81, 11016, 2007). However, to ensure a constant supply of antibodies in humans, they would have to be administered continuously, and this is impractical over the long term. In the 2005 study in HIV-infected individuals, participants had to come in once a week for two hours to receive antibody infusions, according to Alexandra Trkola, a professor at the University Hospital of Zurich. “It’s impractical to inject someone for the rest of their lives with an antibody,” adds Caltech professor Pamela Bjorkman, who collaborates with Baltimore.

That’s why researchers, including Baltimore and Johnson, are developing ways to produce a continuous supply of these antibodies by introducing the genes that encode them into human cells. At the same time, researchers are also trying to engineer improved antibodies or antibody-like molecules that will perhaps be even better than the naturally occurring ones at inhibiting HIV from infecting its target cells.

Success in mice

A few years ago, Baltimore showed in experiments with mice that it is possible to use a retrovirus to deliver T-cell receptor genes specific for certain kinds of tumors into HSCs isolated from their bone marrow. When the HSCs were transferred back into the bone marrow of mice, they developed into T cells that expressed the tumor specific T-cell receptor (Proc. Natl. Acad. Sci. 102, 4518, 2005). “I said well, if you can do it with T-cell receptors, you should be able to do it with antibody genes,” says Baltimore.

He is now testing this approach with an HIV-derived lentivirus to carry antibody genes into CD34+ HSCs. He hopes these cells will multiply and develop into antibody producing B cells, which could then provide a lifelong supply of antibodies. Ultimately, Baltimore’s goal is to utilize this gene therapy approach as a strategy to prevent HIV infection, but for now he is focusing on testing this approach in HIV-infected people because he says it is easier to conduct gene therapy trials when it is a therapeutic approach.

Still, even as an experimental therapy, this is an elaborate strategy. For example, Baltimore’s approach would require a bone marrow transplant—researchers would remove bone marrow, insert the antibody genes into HSCs isolated from the marrow ex vivo, and then transfer them back into the volunteer. Each procedure could cost as much as a couple of thousand dollars for a single volunteer and could be especially difficult to do in developing countries, says Pin Wang, an assistant professor at the University of Southern California, who is collaborating with Baltimore’s group.

Wang is currently trying to develop a way to introduce the antibody genes with a simple injection, rather than a transplant. To accomplish this he has engineered an HIV-derived lentivirus vector that he hopes can target CD34+ HSC target cells following injection directly into the bone marrow. The vector carries two proteins on its surface, an antibody that recognizes the CD34+ receptor on the HSCs, and a fusion protein that enables it to fuse with the HSCs so it can introduce its genetic payload. This approach would lower the cost substantially to about US$100 per procedure, Wang estimates.

To provide proof of concept Wang is conducting experiments using injections of the engineered lentivirus vector to try to introduce a luciferase gene into the bone marrow of humanized mice that have CD34+ HSCs derived from human cord blood. Initial results look promising, Wang says. Within a few weeks, he found expression of theluciferase gene in the bone marrow in the legs of mice, suggesting that in principle the luciferase gene does successfully incorporate into the genome following injection and gets expressed in the human HSCs (see Figure 1). Wang confirmed that indeed the CD34+ cells express the luciferase gene by isolating these cells from the mice in his experiment.

Figure 1. Testing for gene expression in humanized mice.

View Larger Image

But after the antibody genes are incorporated into the genome of the HSCs, it’s still unclear if the B cells derived from these human HSCs can express the antibody proteins the same way they do naturally. Baltimore says experiments are underway to test this in vitro, as well as in humanized mice. To test it in vitro, Baltimore matures human HSCs into B cells by treating them with a certain combination of growth factors. In humanized mice, researchers can infect human CD34+ cells with the lentivirus carrying antibody genes in vitro, and then inject the HSCs into the mice, where they mature into human B and T cells, Baltimore says. If the antibody proteins are expressed by the human B cells, then researchers can test if they provide protection by infecting the human T cells in these mice with HIV. “We are in the process of doing that,” Baltimore says. “If we can make that work then the next step might be monkeys or maybe even humans depending on how it looks.”

But even if successful in nonhuman primate studies, gene therapy has a checkered past and conducting this type of study in humans might raise safety concerns. Several years ago, a gene therapy trial to treat children with x-linked severe combined immunodeficiency (X-SCID) using a retroviral vector caused leukemia because the vector integrated into the region of an oncogene. Baltimore says there is less of a concern of cancer with using an HIV-derived lentivirus as a vector because HIV integration happens millions of times every day in HIV-infected people without causing cancer. “That is one reason I consider our method safer than the trials for X-SCID,” he says.

Better than nature

Baltimore’s group is currently working with the broadly neutralizing monoclonal antibody b12, among others, and if his gene therapy approach proves successful, eventually researchers could also utilize this method to deliver modified antibodies that are even better than the naturally occurring ones that have already been characterized. “If you could deliver antibody genes to HSCs and have them function in B cells,” Baltimore says, “then you would liberate the whole scientific community to use their design methods to make better antibodies or antibody-like proteins.” He is currently collaborating with Bjorkman’s group, which is trying to engineer improved antibodies that Baltimore’s lab will then test in animal models.

“We can introduce whatever we want into our synthetic gene,” Bjorkman says. She expresses her engineered proteins in cultured cells and then, in a neutralization assay, infects cultured cells with HIV to test if HIV can still infect these cells in the presence of the engineered proteins. “We dream up strange things that we then test for neutralization,” Bjorkman says.

One effort to improve upon nature involves combining structures from several naturally occurring antibodies into one protein (see Figure 2). Developing a combination protein from several antibodies might help prevent HIV escape mutants, Bjorkman says, likening it to the way ARV therapy is most effective with a combination of three or more drugs. In addition, a combination antibody protein will bind to HIV proteins like gp120 more tightly than just one antibody, because it would be more difficult to dissociate all of them. “If you link five things together, it’s almost impossible to remove it,” Bjorkman says. “So it’s kind of glued there.”

Figure 2. Engineering novel antibody-like proteins.

View Larger Image

Bjorkman is now trying to combine the genes for binding sites from several of the known broadly neutralizing antibodies to see if they are better at blocking HIV from infecting its target cells or are effective on a wider range of strains. She is also trying to engineer smaller versions of the known antibodies because she says they are often too big to fit in the narrow space between the target cell and HIV. “We can link [several of] them together in a smaller format,” Bjorkman says.

Some antibodies will only bind to gp120 that is bound to the CD4+ receptor, so another option is to link the antibody proteins to a CD4+ receptor, which would bind to the HIV Env protein and induce a conformational change. Additionally, her lab screens for mutants in antibody genes that allow them to bind more tightly to gp120, and uses 3D protein imaging to identify structures that would fit better. This might result in engineered antibodies that would be effective at lower concentrations, Bjorkman adds. “We are still far from a complete understanding,” she cautions. “Using 3D protein imaging to enable a rational design of new architectures is still quite challenging.”

So far, improving on nature has not been so easy. Some of the engineered proteins, for example, are impossible to express in cultured cells. In addition, Bjorkman says, “We keep managing to make things that decrease the ability to neutralize HIV [in vitro].” However, she says there are a few promising leads. “We have some multimeric versions of existing antibodies that work better [than the parental antibodies].”

Muscling their way in

Meanwhile Johnson is working on a different approach that involves using AAV to carry DNA encoding antibody into muscle tissue, to make the cells express the HIV antibodies. In contrast to Baltimore’s gene therapy approach, the DNA carried by AAV does not integrate into the genome of the muscle cells. Instead, it stays in the nucleus as a so-called episome, and the cell then expresses the antibody genes, Johnson says.

He says the primary goal of this work is to develop a way to prevent HIV infection. In some ways this may be easier than treating HIV-infected individuals, according to Johnson. Once HIV has mutated extensively it is more likely to develop escape mutations to the antibody. In contrast, he says, there are very few different viruses that are responsible for actually establishing an HIV infection in a single person. A recent analysis of env sequences in HIV-infected people shows that often a single virus is responsible for establishing infection (Proc. Natl. Acad. Sci.105, 7552, 2008). “There are very few viruses that make it through the bottleneck,” Johnson says. “You have an Achilles’ heel for the virus if you have the antibodies there at the right time, and that is at the time of initial infection.”

Johnson’s approach has already been tested in mice and rhesus macaques with encouraging results. His group showed a few years ago that an injection of AAV carrying DNA encoding the broadly neutralizing monoclonal antibody b12 into mice leads to expression of antibodies in their blood (J. Virol. 76, 8769, 2002). And in preliminary, still unpublished experiments, his group injected AAV with DNA encoding three different antibodies that are all known to neutralize SIVmac316 in vitro, into rhesus macaques. One year after vaccination the nonhuman primates are still expressing high levels of antibody in their serum. “We are very encouraged that this will go on for a very long time,” Johnson says.

He vaccinated three groups, each of three monkeys, with slightly different antibody genes and one month later challenged them with intravenous injection of SIVmac316, a derivative of SIVmac239. All six control monkeys became infected, and four of them have since developed AIDS. But of the vaccinated monkeys, all three in one group were protected, two of three were protected in a second group, and one of three in the third group remained uninfected. The reason for the variation in the level of protection is still being investigated.

In his next experiments, Johnson wants to determine the dose of antibody that will be necessary to achieve protection. He also plans to challenge the vaccinated macaques either rectally or vaginally to more closely mimic the primary mode of HIV transmission. To be protected from such a mucosal challenge, the antibodies would have to be present in mucosal tissues. Johnson says a recently approved protein-based vaccine against human papilloma virus provides evidence that a vaccine injected intramuscularly can protect against sexually transmitted viruses, suggesting antibodies are available at the site of infection.

His current goal is to conduct a clinical trial to see if injecting humans with an AAV vector carrying genes for broadly neutralizing antibodies like b12, results in production of the antibody. The proof that this can work is already there from monkey experiments, he says. “We have clearly shown that we can do this in monkeys and that it’s effective,” Johnson says, adding that the challenge now is to show the approach is safe in humans. Due to issues surrounding regulatory approval, it will be at least three years until a clinical trial of this approach will be underway, says Johnson. “The challenge is how you answer the safety questions the FDA [US Food and Drug Administration] and others are going to have.”