Come Out, Come Out

A recent study suggests a novel treatment might flush out latent copies of HIV hiding in the body—and re-ignites discussion over the challenges of eradicating HIV infection

By Philip Cohen, PhD

There's a four-letter word that HIV researchers don't use very often: Cure.

But that word did grace a recent report authored by David Margolis of University of North Carolina, Chapel Hill and a collection of other notable HIV researchers (Lancet 366, 549, 2005). After describing an unusual treatment for HIV-infected people that might be targeting the burden of latent virus, the researchers wrote that the "finding, though not definitive, suggests that new approaches will allow the cure of HIV in the future."

The ensuing interest was not surprising. The story was widely reported in the mass media, thousands of websites picked up the story, and among the authors' colleagues grew a buzz of criticism over whether it was worth raising hopes of a cure in those already infected given what the test-of-principle study of four patients actually accomplished. "I used the word, so I'll take the heat," says Margolis. "But the position being staked out by some on the other side is that curing HIV isn't possible. And we don't know that. If you tell everyone this is impossible, it becomes self-fulfilling, because no one will work on this and no one will fund it."

Many researchers who study places that HIV might hide in the body, collectively known as the "reservoir," agree with Margolis that the goal of HIV eradication is worth pursuing, even if they are more reluctant to talk in terms of the c-word. "We just need to be a little bit humble about what we can achieve in the face of a tough problem like this," says Robert Siliciano of the Johns Hopkins School of Medicine, Baltimore.

But the sort of caveats researchers have raised about the Lancet paper reveal a lot about the scientific and technical difficulties in studying the reservoir and the large range of opinion on the feasibility of HIV eradication. Ask a dozen researchers about the prospects of eliminating HIV infection and you'll get a dozen shades of opinion ranging from optimism to pessimism surrounding a central fact: no one actually knows how or exactly where the virus is hiding, or the best way to lure it out and destroy it.

Today most researchers seem to lie on the glass-is-half-empty side on the question of HIV eradication. But that wasn't always the case. Talk about wiping out HIV infection was more popular in the late 1990s when powerful combinations of antiretroviral (ARV) drugs—highly active antiretroviral therapy (HAART)—first became available and were bringing some people with AIDS back from the brink. Where drugs were available, the death toll of the disease plummeted and the level of virus in the blood of infected people dropped below detection limits of conventional assays of the time, about 500 HIV copies/ml blood.

Even so, it was clear that drugs could only attack actively replicating virus. They could not touch another potential source of HIV—copies integrated into the chromosomes of resting CD4+ memory T cells. By their nature, these cells could be long-lived and the fear was that if the drugs were eventually withdrawn, some antigen would reactivate these cells and release the latent virus. But David Ho at the Aaron Diamond AIDS Research Center in New York and his colleagues calculated that if viral replication was completely stopped then existing latent cells harboring HIV could die off in less than three years (Nature 387, 188, 1997).

But researchers were to discover that while HIV can't run away from HAART, it can hide. This became clear when researchers experimented with removing drugs from HIV-infected people who had been receiving HAART for more than three years: the virus rebounded to pretreatment levels within weeks. "That means you are dealing with exponential growth from very small numbers of virus," says Siliciano. "It means partial reduction in the size of the reservoir is essentially useless. You are stuck with the virus unless you get every last latently-infected cell."

And Siliciano's measurements suggested that this reservoir of cells was extremely stable, displaying a half life of 44 months (Nat. Med. 5, 512, 1999). If that's true it means that even if HAART completely suppresses HIV replication and no new virions are produced, the ticking time bomb of latent cells would not be eliminated from the body for at least 60 years. When that finding was published, HIV eradication strategies switched their focus to directly targeting this source of HIV. The tactic adopted by several labs was to stimulate the CD4+ memory T cells to reactivate the virus, making it susceptible to drugs, an immune response, or some other therapy. These wake-up strategies involved treating patients with T-cell stimulating factors like the cytokines IL-2 and IL-7.

But while some researchers were able to see impressive reductions in the size or composition of the pools of latent HIV in CD4+ memory T cells, the effect wasn't strong enough to purge the reservoir. One striking report came from Tae-Wook Chun and Anthony Fauci at the National Institute of Allergy and Infectious Diseases and colleagues in which they treated two HAART-recipients intermittently with IL-2. At first this treatment appeared successful. The number of latent HIV cells dropped until no such cells could be recovered from the patients. However, when the drugs were removed viral levels rebounded in just a few weeks (Nature 401, 874, 1999).

There was an obvious complication with these approaches, however, that could explain their failure. Attempts to stimulate resting CD4+ memory T cells to release virus also creates more active CD4+ T cells, and it has since become clear that these cells are HIV's preferential replication ground. That's why researchers began looking for a more subtle approach which could stimulate the slumbering HIV genome without prompting widespread activation of T cells.

Switching on the HIV genome requires the work of proteins called transcription factors that bind near the viral promoter and recruit the RNA polymerase machinery that transcribes the viral genome into RNA, which is then used to make protein. Margolis' team determined that in some latent cells the HIV genome is silent because of the action of an enzyme called histone deacetylase (HDAC) whose job in the cell is to shut off regions of the chromosomal DNA that contain unused genes. HDAC makes these regions inaccessible to transcription factors by chemically modifying histones, proteins that package chromosomal DNA, creating a large protein and DNA complex known as chromatin. Margolis found that sometimes HDAC ends up remodeling the region of chromosomal DNA around the HIV genome so that it is packed with histones and inhibitory proteins that cover the viral promoter, thereby blocking transcription factors from entering.

So researchers in Peter Dervan's laboratory at the California Institute of Technology developed hairpin-shaped molecules called polyamides that were designed to bind near the promoter of an integrated copy of HIV, freeing it from histones and allowing transcription to proceed. Margolis and his team used these polyamides to show in vitro that specifically opening chromatin around the integrated HIV genome induced HIV production from latent cells taken from patients. But these chemicals had never been used as approved drugs and so wouldn't be available to use in people for many years, if ever. But in 2001 Margolis' work unexpectedly took a giant step forward. "It was exactly what I had been looking for," he says. The answer, however, didn't come from his experiments. It arrived in his email inbox.

He was sitting in his office, eyes glazing over as he quickly skimmed through electronic tables of contents from the dozens of journals he regularly checked. One suddenly caught his eye. Sandwiched between reports of a novel protein from mouse testes and another on the biochemical minutiae of adrenaline pumping chromaffin cells was a seven page article titled "Histone Deacetylase Is A Direct Target of Valproic Acid, A Potent Anticonvulsant Mood Stabilizer, and Teratogen." (J. Biol. Chem. 276, 36734, 2001). For Margolis, it was an exciting moment.

Not only did the paper show that valproic acid inhibited the enzyme Margolis believed suppressed HIV transcription, but it is an FDA-approved drug used to treat many conditions including epileptic seizures and bipolar disorder. "Here was something I could give to patients right now," he says. His team first confirmed the drug could awaken HIV when applied in vitro to CD4+ memory T cells purified from infected volunteers. They also found valproic acid did so without activating the T cells or making them susceptible to new infection (AIDS 18, 1101, 2004). Armed with that data, Margolis and his colleagues felt they were ready for the clinic.

They recruited four HIV-infected volunteers who were receiving HAART and had viral loads below the threshold of current standard assays, about 50 HIV copies/ml. In a leukopheresis process that takes a couple of hours per patient, their blood was pumped through a centrifuge and white blood cells spun out. Cells expressing surface proteins characteristic of resting memory T cells were purified and then stimulated to reactivate any HIV, enabling the researchers to determine the size of the latent HIV reservoir at the start of the study.

The volunteers then intensified their ARV therapy and added valproic acid to their drug regimen and three months later the level of latently infected cells was measured again. Based on reports on the stability of the latent HIV reservoir reported in the literature, the researchers set a level of 50% decline as significant. In three of the patients, they measured declines of 68, 72, and 84%. While this level of reduction is unlikely to have any clinical benefit, the downward trend was encouraging.

This pilot study has its weaknesses, which Margolis readily acknowledges. The study involved only four volunteers and had no control group. And even an 84% reduction is on the border of confidence of the assay, say researchers who regularly conduct measurements of latent HIV in memory T cells. Also, since the patients received two additional drugs as well as HAART—they were also given the fusion inhibitor T20 (or enfuvirtide) in order to make sure any reactivated released virus did not infect neighboring cells—in theory either drug or their combination may have been responsible for any depletion of latently-infected cells, making the mechanism of depletion unclear. Finally, one patient receiving the ARV zidovudine as part of HAART developed anemia during the study, possibly because valproic acid increases the bioavailability of this drug. This side effect sets limits on how widely the treatment could be used. But as the first test of an HDAC inhibitor in the clinic, many experts agree the results are intriguing and deserve to be followed up with a larger study.

Blanketed, bulldozed or barely replicating?

But some of the questions raised about the paper reflect wider disagreement in the field. Some experts are wondering, for instance, whether the mechanism of latency that Margolis' team is targeting is the most important one. Siliciano believes that latently-infected cells are a major contributor to the HIV reservoir responsible for rebound, but he thinks valproic acid is targeting the wrong process to awaken the virus in most of these cells.

Consider that by inhibiting HDAC valproic acid helps remove a blanket of inhibitory proteins that obscure the promoter of the integrated HIV genome, preventing it from expressing its genes and spooling off RNA copies of itself to form new viruses. This would suggest that latent HIV would be located in transcriptionally inactive regions of chromosome. But Siliciano says that recent studies have revealed that HIV actually prefers to integrate into active genes, which suggests a completely different mechanism by which its genome could be shut down: transcriptional interference (Trends Mol. Med. 10, 525, 2004). In this scenario the HIV genome is silent because the cellular gene in which it has integrated is so active that transcription factors can't gain access to the viral promoter—they are bulldozed out of the way by active cellular transcription. "RNA polymerase complexes would be going right through these HIV genomes all the time," explains Siliciano.

And there could be many other reasons why the virus may lay fallow. For example, some factors that HIV needs to transcribe its genes may not be fully active in some cells. Indeed, Dean Hamer at the US National Cancer Institute and his colleagues have shown in vitro that some latent HIV genomes can be revived by synthetic molecules called diacylglycerol lactones. These molecules stimulate protein kinase C, an enzyme known to boost HIV transcription by driving the activation of the cellular transcription factors NF-κB, c-Jun, and TAR-binding factors, as well as the virally encoded Tat transcription factor (J. Virol. 77, 10227, 2003).

Some researchers also question what role latent cells play in sustaining the reservoir given the numerous other possible places that HIV might conceal itself. "Reservoir is a garbage bag term, I don't think it is one location or type of cell," says Roger Pomerantz of the pharmaceutical research company Tibotec. To him the rapid recovery of virus after HAART cessation suggests the primary source for this rebound is not latent virus at all. "I think you have to have ongoing replication for it to happen so fast," he says.

In 1999 Pomerantz's team found evidence for this "cryptic" replication when they analyzed blood from 22 HIV-infected people on HAART with ultra-sensitive detection methods. They found that virus was present in the blood, but at levels below the 50 copies/ml level of detection of conventional tests (JAMA 282, 1627, 1999).

In retrospect, it isn't surprising to find some viral replication. No drug combination is perfect. Cocktails that are 99.9999% effective would, by definition, allow for replication at one millionth of pretreatment level of about 10 billion virions per day. And that would be the case even if the drugs could reach every nook and cranny in the body. In fact, the availability of drugs in different tissue compartments can vary considerably. And other factors can limit how well the drugs operate. Many ARVs depend on cellular enzymes to convert them from precursors to an active form—and not all cells carry the same levels of these enzymes. It also turns out that some cells are equipped with membrane-spanning molecular pumps called P-glycoproteins (PGPs) whose job it is to spit toxins out of cells. But unfortunately PGPs do just as good a job removing some drugs, particularly protease inhibitors.

The relative importance of low levels of viral replication versus latent virus for sustaining the reservoir, and the relationship between these two viral sources, is unclear. Siliciano favors the idea that leaks of latent virus from sporadically-activated resting cells are the major source of rebound virus, suggesting that activation of the latent virus and its destruction is crucial to eradication. This viewpoint is partly based on the evidence of the longevity of this source of virus.

But some evidence suggests that latently-infected memory T cells may be less stable than Siliciano's data suggests. A recent paper from Chun and Fauci argues that the latent reservoir is frequently refilled by virus replicating in sporadically-activated CD4+ T cells, extending the apparent half life of latently-infected memory T cells (J. Clin. Invest. 115, 3250, 2005). This leads the authors to propose a very different recipe for eradication: intensification of therapy to further cripple replication, plus drugs to dampen activation of latent cells.

Intriguingly, the Margolis paper could be taken as evidence for either model. Remember that besides valproic acid, the researchers intensified therapy aimed at stopping replication of the virus with a fusion inhibitor. So any actual reduction in the level of the latent reservoir could have been caused either by direct depletion of the reservoir, or slowing its refilling by active replication or a combination of both effects. Margolis suspects that viral latency and low-level replication may contribute to viral persistence to a greater or lesser extent in different patients on different regimens, and that both obstacles will need to be breached before eradication can be achieved.

A set of new studies may resolve some of these issues. One trial headed by Joseph Eron at the University of North Carolina at Chapel Hill is studying whether just the addition of T20 to HAART accelerates the decay of the latent HIV in CD4+ memory T cells ("Clinical Trial: Effect of an Enfuvirtide-Based Anti-HIV Drug Regimen on Latent HIV Reservoirs in Treatment Naive Adults"). Margolis and his team have begun a new small study in which his patients receive only HAART plus valproic acid. And Siliciano is looking for signs of rapid reservoir depletion in the "natural" experiment of HIV-infected people who were prescribed valproic acid because they are bipolar, or have seizures or headaches. "We've found there are lots of people who have been on valproic acid plus HAART for a long time," he says.

Ultimately, developing the perfect cocktail of drugs to purge HIV may require understanding all the ways the virus manages to evade destruction by HAART for so long. Researchers say it will be important to develop better animal models of viral latency, such as SIV in the macaque, to take the hunt for hidden virus to a whole new level. "We can give the animals drugs and if the virus disappears from one compartment we can see if it's gone somewhere else or just disappears," says Margolis. "We can look through all the blood and every tissue, if that's necessary. We can't do that in people."

Even when all of HIV's hiding places are mapped, the prospects or timeline for turning that knowledge into a cure is uncertain. Hamer, for example, has argued that multiple steps may be required to allow HIV-infected people to live drug-free: developing HAART regimens that completely suppress viral replication throughout the body, developing drugs that activate latent HIV copies, targeting those activated cells for destruction, and some ongoing form of non-drug therapy (he suggests genetically-engineering cytotoxic T cells as one possibility) to stop any residual HIV copies from rekindling the infection (Curr. HIV Res. 2, 99, 2004).

But Hamer also retains a guarded optimism shared by many reservoir researchers. "There's no evidence that HIV is genetically programmed to persist in the body," he says. Instead, he argues, the virus simply benefits from the limits of current therapy. "So eradication might actually turn out to be quite simple. Nobody knows. Adding one right drug to HAART may push the virus down to a level where it doesn't rebound. There's no evidence that's not the case."