The Monkey Diaries

One of the clearest lessons to emerge from the 2007 STEP trial came, oddly enough, not from human volunteers but from monkeys.


Editor’s Note: This article includes work Harvard University scientist Norman Letvin presented several weeks before his death on May 28, following a long battle with pancreatic cancer.

One of the clearest lessons to emerge from the 2007 STEP trial came, oddly enough, not from human volunteers but from monkeys. The MRKAd5 candidate tested in that study, which failed to demonstrate efficacy and derailed the large Phase IIb test-of-concept trial, had only advanced into human trials because preclinical data in nonhuman primates suggested the outcome would be very different. Most importantly, monkeys given a candidate vaccine closely resembling MRKAd5 and challenged with the hybrid simian-human immunodeficiency virus (SHIV) 89.6P strain had a 1-3 log decrease in viral load compared to unvaccinated monkeys challenged with the same SHIV strain. But in humans, the MRKAd5 vaccine candidate showed no impact on the viral load of vaccinated volunteers who became infected during the trial. Worse, it also appeared to increase susceptibility to HIV among certain subgroups of volunteers who had been exposed to the adenovirus serotype 5 from which the vector was derived (see Getting It Right Early, IAVI Report, Sep.-Dec. 2007).

Even before the STEP trial results, some scientists suggested that using SHIV 89.6P could be leading researchers down the wrong path (1). At any rate, researchers have strained in recent years to engineer SHIVs that are more predictive of what happens following HIV transmission and that infect monkeys in a manner that better reflects HIV’s infection of humans.

At the May 15-16 meeting of the AIDS Vaccine Research Subcommittee (AVRS), which was devoted entirely to NHP studies, attendees discussed advances in such efforts. Jeff Lifson, the pony-tailed director of the AIDS and Cancer Virus Program SAIC-Frederick, Maryland, at the US National Cancer Institute, said the ideal SHIV should be mucosally transmissible, replicate to moderate or high peak viral loads and be as pathogenic as HIV and SIV strains. Ideally, SHIVs should also contain env genes from early transmitted/founder virus sequences that are neutralization sensitive and reflect the predominant clades—B, C, and A/E—found in populations with high HIV incidence.

“You have to be as clever as you can in designing your SHIV,” said Lifson. “You have to be clever about which env you pick. When the in vivo results do not match the in vitro results, you rinse and repeat.”

Scientists have generally preferred the SIV challenge model. Though it is a virus unique to NHPs, in certain species of monkeys—namely rhesus macaques—it follows a similar disease course as HIV (see VAX October 2008 Primer on Understanding Animal Models of HIV Infection). But the SIV model has its own limitations. To be appropriately evaluated in macaques, AIDS vaccine candidates must include non-infectious fragments of SIV, rather than HIV.  Researchers have thus turned to the hybrid SHIVs, because they are in theory closer mimics of HIV. Efficacy trials like STEP, however, suggested that may not be the case.

Though developing newer generation SHIVs has proved challenging, researchers are making some progress. In what was probably his last talk before his death on May 28, Norman Letvin, who pioneered the use of NHPs in AIDS vaccine design, told AVRS members that his laboratory is getting close to rolling out a SHIV strain called KB9-05 derived from clade B viruses. Letvin reported that KB9-05 reached peak viremia levels of 7-8 logs and persisted for 250 days. Letvin said the SHIV also had tier 2-like antibody sensitivity, though he would not classify KB9-05 as a tier 2 virus. “This virus has more resistance to neutralization by these usual standard antibodies than previous SHIVs,” said Letvin. “So we are taking KB9-05 forward and preparing it as a challenge stock, As soon as it is titered and prepared we will distribute it to everyone in the field who wants it for any purpose.”

Letvin also described three other SHIVs in the pipeline—one derived from a clade B virus that reached a peak viral load of 10(7), persisted 800 days but which only had tier 1 antibody sensitivity; a clade C SHIV that reached a peak viral load of 10(6) and persisted for 170 days; and a circulating recombinant form 01 (CRF01) clade A/E SHIV being developed as a low-dose mucosal challenge stock that reached a peak viral load of 10(6) and persisted for 100 days. Antibody sensitivity studies continue with the clade C and CRFOI A/E SHIVs, said Letvin.

AVRS Subcommittee member Jerry Sadoff, the chief medical officer and head of development at the Dutch-based pharmaceutical company, Crucell, said progress in the SHIV pipeline was welcome news. But he wondered whether the field was too preoccupied with engineering SHIVs using different HIV clades. Perhaps, he said, it might be more instructive to instead construct SHIVs that encode a variety of epitopes that a vaccine candidate would need to target in order to provide protection.

Researchers recently published evidence of a vaccine sieve effect in the STEP trial, meaning that the vaccine may have kept virus strains carrying antigens most similar to those presented by the vaccine from establishing infection (2). A vaccine-induced sieve effect was harder to detect in the RV144 trial in Thailand, which demonstrated modest 31.2% efficacy, because the serum samples were obtained less frequently during acute infection. But researchers were able to find evidence of a sieve effect in Env-C1, which corresponds to a common antibody-dependent cellular cytotoxicity (ADCC) epitope, and in the V2 loop among vaccinated volunteers who became infected during the trial through natural exposure (see A Slew of Science in Seattle, IAVI Report, Mar-Apr. 2012).  All but 10 of the 50 vaccinees and 71 placebo recipients included in the sieve analysis were infected with a clade E virus, one of the subtypes represented in the re-engineered gp120 AIDSVAX B/E protein boost administered at 12 weeks and 24 weeks after the ALVAC-HIV prime.

While the overall efficacy in the RV144 trial was modest, the protective effect of the ALVAC/AIDSVAX regimen reached nearly 60% during the first year. Sadoff said this raises questions about why the other 40% were not protected. One reason was that the immune responses elicited by the RV144 vaccine regimen, which peaked two weeks after the last boost, waned over time. But another reason could be that vaccinated volunteers who became infected may have been infected with viruses whose epitopes did not match epitopes in the V2 region found in the vaccine candidate. Sadoff said unpublished data suggests the latter occurred. He believes the field should strive to construct SHIV models that have variations in the V2 region. This, he said, could provide valuable clues to the design of more effective vaccine candidates.

“I’m not against making SHIVs with different clades,” he said. “But we also need to make SHIVs that represent a variety of epitopes that a vaccine [candidate] might have to protect against.”

Duke University scientist David Montefiori, who studies antibody responses in HIV-infected humans and in NHP models, also favors refining the SHIV models. “We have been talking about this for 20 years and we are still challenging with the same strains of SHIV and SIV. There are very few SHIVs that have actually made it out there in the field for people to use. We need greater diversity in the challenge stock.”

SIV transmission models

SIV has some similarities to HIV. Some species of SIV-infected rhesus macaques display high viral loads and a marked decline in the number of CD4+ T cells, primary targets of both SIV and HIV. Such challenge models are useful for studying viral transmission and pathogenesis.
In recent years,  the use of single-genome amplification in humans has provided strong evidence that 90% of heterosexual transmissions are caused by a single transmitted virus, and approximately 60% of infections in men who have sex with men are caused by a single transmitted/founder virus. This phenomenon has been recapitulated in the SIV NHP model. Today, many researchers use the SIVmac251 or SIVsmE660, biological isolates, or swarms. Scientists favor swarms as challenge stocks because they contain many different virus variants or quasispecies, similar to HIV, making them useful for enumerating the number of founder viruses and studying various vaccine modalities as well as HIV pathogenesis (see Looking for the Perfect Challenge, IAVI Report, July-Aug. 2009.)

Yet tracking early HIV evolution and pathogenesis using these SIV models can be a tricky process. The individual SIVmac251 challenge stocks used by different laboratories can vary by 2% to 3% within each stock, while the diversity between SIVmac251 challenge stocks and SIVsmE660 can be as high as 20%. This means that challenge studies using different SIVmac251 stocks could result in different outcomes and may not represent a truly homologous challenge model. “Every animal is infected with a slightly different virus, but all from the same pool,” explained Brandon Keele, principal scientist in the NCI’s AIDS and Cancer Virus Program at the Frederick National Laboratory, during his presentation at AVRS. “Each one of those differences adds up to variation. And variation is the enemy here.”

Researchers have been searching for ways to standardize these swarms so that even though they may contain different virus variants or quasispecies, the stocks essentially all behave the same way. Keele, who has already authored several seminal papers—including one that established that productive HIV infection is usually caused by a single founder virus (see Luck Favors the Prepared, IAVI Report, Sep.-Oct. 2010)—has devised a tool that could help do that. 

Keele highlighted ongoing work by his laboratory to design unique replication-competent SIV stocks. His team introduced synonymous mutations (also referred to as silent mutations) of 2-3 nucleotides in the integrase gene of SIVmac239 to generate a phenotypically identical synthetic swarm, but marked or tagged to identify unique or individual virions. Keele said they introduced these synonymous mutations in Int because it has conserved sites where researchers could simply and accurately amplify all variants in the pool. The specific nucleotides were chosen to represent natural polymorphisms. 

Using this bar-code approach, Keele was able to produce 10 DNA plasmids, each of which was identical to the parental SIVmac239 except for the 2-3 nucleotides that had been introduced. These plasmids were then used to generate a large transfection stock of virus. Single-genome amplification and 454 sequencing methods—tools that identify minor changes in viral populations—validated the proportion of each variant with “remarkable consistency,” meaning that a stock now exists with equal proportion of each of 10 variants acting as a synthetic swarm, with all the benefits of both an isolate and a clone.

Keele said they are now in the process of assessing the stocks in animals challenged intravaginally, intrarectally and intravenously, as well as through penile challenge. Once testing is completed, challenge virus will be made available to the research community. Though not as far along, similar experiments are also underway using newly-generated transmitted/founder SIVmac251 and SIVsmE660 stocks. —Regina McEnery

1. J. Virol. 79, 15547, 2005
2. Nat. Med. 17, 366, 2011