Go Forth and Multiply

AIDS vaccine researchers are turning their focus from replication-deficient viral vectors to potentially more efficacious replication-competent approaches

By Andreas von Bubnoff

Just like politics can swing back and forth between conservative and progressive approaches, so can approaches to AIDS vaccine development. Early strategies focused on the traditional methods for developing a vaccine—using a weakened version of the pathogen that the vaccine is designed to protect against. For HIV, however, this strategy was shown to be unsafe.

As a result, many researchers focused on more conservative approaches, including using replication-deficient viral vectors to deliver HIV immunogens. But after Merck’s AIDS vaccine candidate that used a replication-incompetent adenovirus serotype 5 (Ad5) vector failed to provide any protection in the STEP trial, this approach came under mounting scrutiny. Several researchers have started to focus more recently on replication-competent viral vectors, hoping they will generate a robust and durable immune response against HIV and mimic the protection seen with a live-attenuated vaccine approach. “Today we know that replication-deficient vectors are not giving the kind of immune response that we are looking for,” says Eddy Sayeed of IAVI.

Exploring live-attenuated vaccines

In the early 1990s, researchers began tackling AIDS vaccine development with an historically effective, yet aggressive approach: evaluating the efficacy of live-attenuated simian immunodeficiency virus (SIV) vaccines in nonhuman primate (NHP) studies. In 1992, Ronald Desrosiers’ group at Harvard Medical School showed that vaccination of adult rhesus macaques with a live-attenuated SIV—replication competent but at a reduced level compared to the wild-type virus—could protect against SIV challenge (Science 258, 1938, 1992). The experiment was conducted with SIVmac239Δnef, which had a 182 base pair deletion in the nef gene, according to Matt Reynolds, who is currently studying this strain as a member of David Watkins’ laboratory at the University of Wisconsin-Madison.

“[Desrosiers] made the first observation that taking out the nef gene seemingly converts a pathogenic virus into a live-attenuated vaccine,” says Ruth Ruprecht of the Dana Farber Cancer Institute. The response in the field at the time was quite enthusiastic, with many thinking, “this is the breakthrough that we have needed,” recalls Ashley Haase of the University of Minnesota. The Washington Post wrote that scientists at the time called the study “one of the most impressive achievements to date in the search for an AIDS vaccine.”

And it wasn’t the only good news: A study of the so-called Sydney blood bank cohort, a group of eight people who had accidentally been infected, as it later turned out, with nef-deficient HIV through a blood transfusion, showed no apparent signs of disease progression (Lancet 340, 863, 1992; Science 270, 988, 1995). This corroborated Desrosiers’ finding in humans. “At that time there was a lot of optimism for the use of live-attenuated viruses,” says Paul Gorry of the Macfarlane Burnet Institute in Melbourne, who studies the Sydney blood bank cohort.

However, the optimism inspired by these observations was short lived. Longer-term studies showed that using live-attenuated HIV and SIV as vaccines was unsafe. The first evidence of this came from Ruprecht’s group. When they orally vaccinated newborn rhesus macaques with SIVmac239Δ3, which is more attenuated than Δnef, the viral load in the first vaccinated animal didn’t decline for months. “We said, gee, there is something funny going on,” Ruprecht recalls. The result was the same after vaccinating two additional newborn animals, Ruprecht says. The animals couldn’t clear the attenuated virus and although they grew normally and gained weight at first, their CD4+ T-cell levels declined and all animals vaccinated as infants eventually developed AIDS. “This was a big shocker,” Ruprecht says. When the results were published, she says it “hit like lightning” (Science 267, 1820, 1995).

Initially, some thought the infant macaques became ill because their immune systems were immature, but later Ruprecht showed that adult macaques also got sick when vaccinated with this live-attenuated SIV vaccine (Nature Medicine 5, 194, 1999; AIDS 17, 157, 2003). “Live-attenuated [SIV] is largely pathogenic, period,” Ruprecht concludes, adding that to this day, some researchers are still unaware of that. “Unfortunately even now some people still cite that the live-attenuated virus is pathogenic in newborns [only]. That’s really only part of the story.”

In other sobering news, the HIV strain that had infected the people in the Sydney blood bank cohort also turned out to be pathogenic despite the nef deletion. “Some of them are progressing,” Gorry says. “Their CD4+ [T cells] are actually declining.” The donor, and two of the eight blood recipients eventually developed a reduced CD4+ T-cell count, although all of the progressors had low-level viremia (N. Engl. J. Med. 340, 1715, 1999; J. Infect. Dis.190, 2181, 2004; J. Acquir. Immune Defic. Syndr. 46, 390, 2007). “The take-home message is that even persistent very low-level replication places these individuals at risk of progression,” Gorry says. He classified low-level viremia as about 3,000-5,000 copies of HIV/ml of blood, compared with 60,000-100,000 HIV copies/ml of blood in a typical progressor not on antiretroviral (ARV) therapy. The Sydney blood bank cohort is “probably the best evidence that Δnef in human beings is certainly not safe enough,” concludes Ben Berkhout of the University of Amsterdam.

Other researchers experimented with a different attenuated strain called SIVmacC8, which has a 12 base pair deletion in the nef gene. While C8’s efficacy is “pretty impressive,” Ruprecht says, safety was also a problem (J. Virol. 73, 2790, 1999).

Deciphering the dangers

After these discoveries, researchers set out to unravel just how live-attenuated SIV can still cause disease. Additional research showed that these live-attenuated viruses eventually become pathogenic because of their ability to mutate. “The virus evolves,” Berkhout says. “There is some evidence that over time it will become pathogenic.” And there are several possible mechanisms that allow the virus to regain its pathogenicity. For example, Berkhout’s group showed that in cultured cells, SIVΔ3 virus, which has deletions in the long terminal repeat (LTR) promoter and in the nef and vpr genes, replicates more quickly after adding binding sites at its LTR promoter for a transcription factor called Sp1 (J. Virol. 73, 1138, 1999).

Duplicated NF-κB and Sp1 binding sites in combination with additional deletions in the nef gene might also account for the increased replication capacity of the HIV that eventually became pathogenic in some members of the Sydney blood bank cohort (Retrovirology 4, 66, 2007). “The virus is really deleting out all of the unnecessary material,” Gorry says. Still, he says, host differences probably also play a role in the Sydney cohort because some individuals with similar viral mutations eventually had progressive disease, while others didn’t.

Ruprecht also observed that the SIVΔ3 in macaques showing signs of disease had additional deletions. “The virus actually shrunk,” Ruprecht says. Her research group injected other macaques with DNA encoding these mutant virus strains and found that they actually caused much faster disease progression than the original strain.

Playing it safe

Together, these observations made it clear that live-attenuated HIV vaccines are not safe enough to test in humans. “I don’t think Δnef will ever be used in humans,” says Reynolds. But others aren’t as quick to dismiss the live-attenuated vaccine approach. Haase says a live-attenuated vaccine in humans may be justified in some cases. “[In] a population with a very high [HIV] prevalence and incidence, it’s a different calculus altogether,” he says. “You would have to look at the risk of the vaccine versus the very real risk of acquiring infection anyway.”

But for now, the field has largely given up on the idea of a live-attenuated HIV vaccine in humans. “What happened is everybody said it’s not safe and they backburnered it,” Haase says. Instead, the field focused more on using other viral vectors, many of which were replication deficient. But the recent failure of Merck’s Ad5 vaccine candidate in the STEP trial raised questions about whether replication-defective vectors are perhaps not immunogenic enough.

Even before the conclusion of the STEP trial, several replicating viral vectors with HIV gene inserts were in various stages of preclinical development. In addition, some researchers are studying how live-attenuated strains of SIV like SIVmac239Δnef protect to gain insights on how best to mimic this protection with a viral vector-based vaccine. Others are developing new, improved versions of live-attenuated SIV strains for evaluation in NHP studies (seeThe Mysteries of Protection). “Nothing is working, so people are going to try all sorts of things, including replicating viruses,” says Louis Picker of Oregon Health & Science University.

Finding the right balance

Several researchers predict that replicating viral vectors will induce more durable and potent immune responses, but it appears that the more replication capacity the viral vector has, the less safe it becomes. “We have learned [that] with replicating vectors, there is an inverse correlation between the level of virus attenuation and vaccine efficacy,” Berkhout says. “The more attenuated, the weaker the protection.”

To account for these observations, Ruth Ruprecht has developed a threshold hypothesis suggesting that much of the efficacy and safety characteristics of replicating vaccines can be explained by how much they replicate (Curr. Opin. Infect. Dis. 17: 17, 2004). According to this hypothesis, there must be some minimal “vaccine threshold” of replication of a live-attenuated virus to achieve protection. However, above a certain higher “disease threshold,” that same live-attenuated virus can cause disease. The goal is to find a vaccine virus that replicates within a “window of opportunity,” Ruprecht says, exceeding the vaccine threshold but staying below the threshold at which the virus can cause disease. James Hoxie of the University of Pennsylvania agrees. “If it replicates too much, you are going to get disease,” he says, “if too little, you will have a wimpy virus that can’t generate any protection.”

This appears to be the case with live-attenuated versions of SIV. In general, the more genes that are removed from SIV, the more replication is compromised, and the safer it is. But it also becomes less efficacious, Picker says. He is currently conducting the first systematic comparison of the immune responses to different versions of live-attenuated SIVs in a study involving 120 rhesus macaques. Until now NHP studies have been much more limited in size. Picker’s study will compare T-cell, innate, and antibody responses induced by different live-attenuated SIV strains.

Chris Parks of IAVI’s AIDS Vaccine Development Laboratory says the inverse correlation between safety and efficacy generally holds true for viral vaccine approaches (see Figure 1). Some of the most effective vaccines to date are typically based on live-attenuated viruses, such as those against measles, mumps, and rubella, Parks says, while DNA-based or protein subunit vaccines are typically safe, but not as efficacious. There are notable exceptions to this, such as the Hepatitis B vaccine, which is a protein subunit and is highly efficacious. But the gp120 Env protein subunit AIDS vaccine that was developed by VaxGen and tested in a Phase III efficacy trial was not effective.

Figure 1: Safety and efficacy of replicating and non-replicating vaccine approaches."

View Larger Image

Replicating vector systems

Several different replicating viral vectors are currently under investigation that can carry HIV or SIV genes, but do not integrate into the host genome. Researchers hope that replicating vectors will be more efficient since replication is thought to stimulate the immune system over a longer period of time. It is also possible that replicating vectors will induce mucosal immune responses, which are considered important for protection from HIV infection (see The great barrier, IAVI Report, March-April, 2008). In the case of polio and measles, studies showed that the live-attenuated version was better at protecting from infection than an inactivated version, says Frédéric Tangy of the Institut Pasteur in Paris.

Replicating vectors have other advantages as well. They induce better innate immune responses, according to Tangy. Also, less virus is required for vaccination because more virus gets made through replication in the host cell. For example, Tangy is using measles vaccine as a replicating vector, and only 1,000-10,000 copies of the virus are necessary to achieve an immune response, he says, much less than the one billion copies of the replication-deficient Ad5 vector used in the MRKAd5 vaccine candidate. The high dose of replication-deficient Ad5 used in the STEP trial might have generated additional target cells for HIV by recalling a memory response to preexisting Ad5 immunity. “This is one possibility of increasing infection due to vaccination,” Tangy says.

Still, there is little evidence that replicating viral vectors will be superior for AIDS vaccine candidates and experts say that the assumption that they will work better is still hypothetical at this point. Berkhout says it’s not really known how the immune responses to replication-competent vectors are different from those induced by replication-deficient vectors. And not everyone believes that replicating vectors will be able to protect as well as live-attenuated HIV or SIV. “I would have my doubts that these other replicating systems will mimic an HIV-1 infection, because it’s happening in different cell types,” Berkhout says. “But the tests should be done. I would be happy if I am wrong.”

IAVI’s Vector Design Consortium has several investigators working on novel replication-competent viral vectors. IAVI also has a vaccine development program in partnership with DNAVEC, a biotechnology company in Tsukuba, Japan, to develop a replication-competent Sendai virus vector. This vector is expected to be in Phase I clinical trials by 2010. It is considered safe for humans because mice are the natural host and its ability to replicate in humans is greatly restricted, preventing it from causing illness, Parks says. In addition, Sendai infects the upper respiratory tract, raising hopes that it might induce mucosal immunity. IAVI is also planning to do intranasal administration of the Sendai vaccine with a device similar to the one used to administer FluMist to generate mucosal immune responses. The safety of intranasal administration will be studied in mice, rabbits, and monkeys.

Tetsuro Matano’s group at the University of Tokyo showed a few years ago that a DNA prime followed by a boost with a Sendai vector expressing HIV gag could protect five of eight rhesus macaques from intravenous SIVmac239 challenge, in that they had undetectable viral loads five weeks after challenge (J. Exp. Med. 199, 1709, 2004). In the study, both replication-competent and replication-deficient Sendai vector vaccines showed some protection: The replication-competent version protected two out of four monkeys, and the deficient version protected three out of four. There was no observed difference in the induced immune response between the replication-competent and incompetent Sendai vector in this study, Matano says. “We have not shown evidence that replication can result in longer duration of the immune response,” Matano adds.

IAVI is also collaborating with Picker’s lab to develop Cytomegalovirus (CMV) for use as an AIDS vaccine vector. Picker is currently studying the efficacy of simian CMV vectors carrying SIV genes in rhesus macaques. One advantage of using CMV is that it results in persistent infection and induces a high level immune response for life. There are concerns that preexisting immunity could be an issue because many people have been exposed to the human CMV, but so far, preexisting immunity doesn’t seem to make a difference in immune response in studies with NHPs. “Preexisting immunity doesn’t affect CMV, unlike virtually all other vectors today,” says Picker. His group is currently developing attenuated strains of CMVs by deleting immune evasion genes to handicap the virus. These strains remain immunogenic and efficacious, but are less likely to cause disease in immunocompromised people, Picker says.

Vesicular stomatitis virus (VSV), a virus that naturally infects cattle, is also being explored as a replicating vector. Data from John Rose’s lab at Yale University indicated that VSV expressing HIV genes could protect rhesus macaques from SHIV challenge (Cell 106, 539, 2001). “The data from Rose’s lab was the first to demonstrate that VSV was a promising HIV vaccine vector candidate,” Parks says. Earlier studies showed that VSV can spread to the central nervous system of young mice exposed to the virus, although monkeys vaccinated intranasally or through intramuscular injection did not experience any CNS complications, Parks says. Wyeth later developed highly-attenuated VSV vectors that retained high levels of immunogenicity in preparation for clinical trials (seeRenewed promise, IAVI Report, Sept.-Oct., 2005).

Measles is another possible replicating vector for AIDS vaccines. It is considered safe since measles vaccine has been used in millions of people, and inexpensive because manufacturers are already making it, Tangy says. Measles, similarly to HIV, targets T cells, macrophages, and dendritic cells. Tangy is currently collaborating with GSK to develop a measles vector-based AIDS vaccine candidate expressing HIV gagpol, and nef for a Phase I clinical trial.

Preexisting immunity to measles is an obvious concern but animal experiments suggest it might not be an issue (J. Virol. 78, 146, 2004). In the experiments, researchers first primed animals with standard measles vaccine and then, one year later, gave them two injections with the measles vaccine encoding HIV proteins. This regimen was able to induce HIV-specific immune responses. Tangy’s group is also working on a chimeric virus where the surface glycoproteins of measles are replaced with a modified HIV surface gp160 Env protein. The hope is that this chimeric virus will circumvent preexisting immunity from measles vaccination and will enter HIV-specific target cells with CD4 and CCR5 or CXCR4 receptors.

Marjorie Robert-Guroff of the US National Cancer Institute is also planning a Phase I safety trial of an orally-administered replicating adenovirus serotype-4 vaccine. She says this was shown to be safe and effective in protecting American soldiers against acute respiratory disease. A group at St. Jude Children’s Research Hospital has initiated a Phase I safety trial with a replicating vaccinia virus vector based on the smallpox vaccine (Eur. J. Clin. Microbiol. Infec. Dis. 23, 106, 2004). The vector is given in combination with DNA and a protein in an approach called D-V-P (DNA-Vaccinia-Protein).

Regulatory concerns

A variety of other replicating vectors are also currently in development and many groups are now considering the possibility of additional regulatory hurdles to advancing them into clinical trials. There are some safety concerns for replicating vectors and regulatory agencies, like the US Food and Drug Administration (FDA), might set more stringent requirements for preclinical studies of candidate vaccines that use replicating vectors.

One concern is that replication could go uncontrolled in immunocompromised people. Shiu-lok Hu of the University of Washington in Seattle recalls a case in the early 1990s when individuals in a therapeutic AIDS vaccine trial died after receiving a vaccine preparation containing replication-competent Vaccinia virus that was insufficiently inactivated. To address this issue, Heinz Feldmann of the National Microbiology Laboratory of the Public Health Agency in Canada is currently checking whether immunocompromised macaques infected with SIV get sick from vaccines that use a VSV vector to carry the Marburg or Ebola virus glycoprotein gene. So far, he says, it seems that there are no problems.

Another concern might be that using animal viruses as vaccine vectors exposes humans to viruses they wouldn’t otherwise come in contact with.

Possible shedding of a replicating virus in urine or stool is also being explored. Tangy says the clinical trial of a measles vector-based vaccine will need to measure where in the body the vector replicates, whether it is shed by the body of vaccinated people, and if the released virus is infectious. “We have to document the shedding and biodistribution,” Tangy says.

Despite these concerns, some say it is a necessary avenue to pursue. “My view is let’s find something that works and then we can go and make it safe,” Picker says, referring to preclinical studies. “There is no point in making a vector safe if it doesn’t work.” But at the same time, researchers insist other strategies should not be neglected. “I don’t know what it takes to have a safe and effective vaccine,” Ruprecht says. “As long as we don’t know what it takes, we need [to] test various approaches.”

The Mysteries of Protection 

While live-attenuated AIDS vaccines may be off limits in humans, understanding how they protect may provide clues about how to develop other, safer vaccines with similar levels of protection. Even though SIVΔnef was described over 15 years ago, it’s still unclear just how it protects in rhesus macaques. “It’s just been curious [that] nobody has been able to figure out exactly why,” says Louis Picker of Oregon Health & Science University. The protection could be due to cellular, humoral, innate immunity, or some combination of these. “None of this has yet been ruled out,” says Ruth Ruprecht of Dana Farber Cancer Institute. Viral interference might also be at play, she adds, meaning that once one type of a virus is in the host, it could interfere with a second virus being able to “set up shop.”

Researchers are now conducting studies to try to elucidate the mechanism of protection. “If we had some insight into how [it protects], we would have an idea what an efficacious vaccine might actually look like,” says Ashley Haase of the University of Minnesota. The goal is to understand and then mimic the mechanism with another approach that’s safer, for example with another persistent virus such as CMV, Picker says.

In one project, in conjunction with IAVI’s Live Attenuated Consortium (LAC), Haase will combine tetramer staining with in situ hybridization to observe where the immune responses to Δnef occur. The technique will make it possible to directly count the ratio between effector and HIV target cells. “We can actually see the battle going on,” Haase says. “We can directly see where the tetramer positive cells are spatially in relationship to the infected cells.”

He already knows from natural history studies that the higher that ratio, the better the control of infection. Haase has injected Δnef into 18 monkeys and plans to challenge them soon. The hypothesis is that in vaccinated animals, many effector cytotoxic T lymphocytes (CTLs) eliminate HIV-infected cells at the portal of entry before they even have a chance to expand and disseminate the infection systemically. One theory is that Δnef continues to replicate at a low level, thereby continuing to stimulate the immune response. “That’s why we think these attenuated viruses may actually prove to be superior [to non-replicating vectors],” Haase says. “That’s what we are trying to prove.”

But maintaining a constant stimulation of the immune system is not the only reason why live-attenuated viruses protect. They also express most viral genes, whereas many non-replicating vectors only express a few. “There might be a more narrow immune response,” says Matt Reynolds, a member of David Watkins’ laboratory at the University of Wisconsin-Madison. Live-attenuated viruses also direct the immune system toward the same cells that will be infected by the pathogenic strains, Reynolds adds.

Researchers are also developing better live-attenuated vaccines, in part because the ones that are available so far don’t protect well enough. For example, a more recent study showed that SIVΔ3 protection from homologous challenge disappeared several years after vaccination, Ruprecht says (J. Virol. 79, 8131, 2005). And neither Δnef nor Δ3 completely protect against heterologous challenge with SIV E660. E660 is derived from SIV-infected rhesus macaques and is believed to more closely resemble naturally occurring SIV. It is not a clone of a single SIV strain, but made up of several SIV subspecies. In one study, two out of four Δ3 vaccinated animals got sick by 96 weeks after the E660 challenge (J. Virol. 73, 8356, 1999).

Recently, David Watkins’ group vaccinated 10 macaques with SIVmac239Δnef and challenged them with an intravenous injection of E660. Eleven months after the challenge, only four of the 10 could suppress viral load.

Others are working on improved live-attenuated SIVs beyond Δnef and Δ3. James Hoxie of the University of Pennsylvania has been developing a live-attenuated SIV that, so far, appears to be safe and protect against heterologous SIV challenge in pigtail macaques. The strain, called SIVmac239ΔGY (J. Virol. 75, 278, 2001), has a deletion of two amino acids in the cytoplasmic tail of the SIV Env protein. This motif is highly conserved in all SIVs and HIVs and mediates endocytosis of Env by binding to clathrin adaptor complexes, Hoxie says. Without GY, Env is incorporated less into virions and perhaps more Env that’s not incorporated into virions remains on infected cells, making them more susceptible to immune responses by the host.

After vaccination, ΔGY initially replicates at levels similar to that of wild-type SIV, but then an undetectable level of virus is established in the context of an emerging immune response. “This is a virus that’s not a wimp,” Hoxie says, “it replicates quite well.” In fact, he adds, it replicates to higher levels than what has been reported for Δnef.

The immune responses elicited by ΔGY seem to be, at least in part, mediated by CD8+ T cells, since depleting these cells 600 days after ΔGY vaccination leads to a brief burst of ΔGY replication. And so far, ΔGY appears to protect from heterologous challenge: Two pigtail macaques vaccinated with ΔGY seven years ago remain healthy three years after E660 challenge, Hoxie says. In another experiment, two out of three animals vaccinated four years ago remain healthy two years after the E660 challenge. The third animal died after its CD8+ T cells were experimentally depleted because it couldn’t control its E660 levels anymore.

Still, Hoxie cautions that it remains to be seen if ΔGY is safer than Δnef, because it took years for some of the Δnef vaccinated animals to get sick. “I wouldn’t try to argue that this is a safe virus,” Hoxie says. One animal did develop a high viral load and AIDS within the first month of ΔGY infection after the virus developed an apparent compensatory mutation. However, such mutations only seem to occur soon after vaccination, Hoxie says.

In a different approach, Ben Berkhout’s group at the University of Amsterdam has engineered a live-attenuated SIV that depends on the antibiotic doxycycline to replicate (J. Virol. 81, 11159, 2007). After two to three weeks, when administration of the antibiotic is stopped, replication also ceases. That, he hopes, will limit replication and thereby the ability of the virus to mutate and become more virulent. He has tested the system in cell culture and in SCID mice that have a human immune system. He will soon test it in macaques in collaboration with Neil Almond at the UK-based National Institute for Biological Standards and Control, as part of IAVI’s LAC.

“That’s where this [live-attenuated approach] is taken to the next level,” says Paul Gorry of the Macfarlane Burnet Institute. However, Gorry also says he doubts a regulatory agency will allow testing an approach like Berkhout’s in humans. “Even though you can turn it on and off,” adds Chris Parks of IAVI’s AIDS Vaccine Development Laboratory, “you are going to end up with nucleic acid integrating into the host. I think that’s the drawback of the system.”

Berkhout does not necessarily believe his approach will make it into human clinical trials anytime soon, but he says it will be useful for elucidating the mechanism of protection by live-attenuated vaccines. -AVB