Stalking HIV's Sleeper Cells

This year's HIV Pathogeneis meeting focused on latent HIV reservoirs—and how they might be eradicated

By Andreas von Bubnoff

Many scientists worry that as HIV research has become increasingly specialized, it appears more and more to be enclosing itself in an intellectual silo. This trend, they fear, hampers the exchange of valuable information with related areas of biological research and, if left unaddressed, could ultimately stunt the creativity and productivity of the field as a whole. To address such concerns, the organizers of the Keystone meeting on HIV Pathogenesis have in recent years invited colleagues immersed in other areas of inquiry to their annual conference.

This year, the gathering—convened in Whistler, British Columbia, from March 26 to March 31—was held jointly with a meeting on Virus Entry, Replication and Pathogenesis that covered general virology. “My hope is that everybody exchanges ideas [and] we get out of our silos,” said Michael Farzan, a virologist at Harvard Medical School and one of the organizers of the track that covered general virology.

This is not to say that HIV got short shrift at the meeting. Sessions dedicated to HIV covered everything from the mechanisms of elite HIV control to new strategies for blocking HIV’s entry into its target cells. But, more than anything else, researchers at Keystone were treated to the latest findings in cure research, most notably those related to the exploration and eradication of HIV reservoirs.

The T cell as Trojan horse

The depletion of HIV reservoirs is all but the ultimate prize of cure research. In other words, it’s a pretty tall order. Even those who have been on highly active antiretroviral therapy (HAART) for years and have suppressed HIV to undetectable levels in their blood, retain these reservoirs of chronic infection. The reservoirs consist mostly of CD4+ T cells that carry one or more copies of HIV in their chromosomes. These latently infected cells don’t ordinarily produce virus particles—until, that is, HAART is interrupted.

Another possible repository of latent HIV is in macrophages, although the in vivo relevance of this reservoir remains unclear. HIV researchers long suspected that it isn’t established very soon after HIV transmission, said Quentin Sattentau, a professor of immunology at the University of Oxford. This is because macrophages aren’t easily infected by free virus—not even by transmitted founder virus, which is the first and in many cases the only transmitted virus that generates productive infection. Sattentau, however, presented research at Keystone that calls that belief into question. His studies suggest that macrophages can, in fact, be efficiently infected through direct contact with infected CD4+ T cells, establishing a latent reservoir much earlier and more efficiently than previously suspected.

Sattentau and his colleagues isolated primary CD4+ T cells from volunteers, infected them with HIV, and mixed the cells with macrophages from the same volunteers. Typically, when HIV-infected cells contact uninfected ones, they form virological synapses—structures through which the virus buds and migrates directly into the uninfected cell. But Sattentau detected an entirely different path to infection in his experiments: the macrophages in his mixtures ingested the infected CD4+ T cells by phagocytosis. This ordinarily results in degradation or digestion of the ingested cell. But electron micrographs taken by Sattentau and colleagues revealed that virus-infected cells taken in by macrophages through phagocytosis were not rapidly degraded but, instead, released their virus into a subcellular compartment.

The researchers also found that three days after a one to two hour contact with infected T cells, 1%-10% of macrophages had HIV in their cytoplasm. This was not the case when macrophages were incubated with free HIV particles from the same number of CD4+ T cells. Sattentau concluded that the phagocytosis of infected T cells represents a new mechanism by which macrophages can get infected. This, he pointed out, doesn't mean that macrophages are never infected by free virus, but that such mechanisms of infection are probably less efficient.

Sattentau further reported that transmitted founder viruses can also infect macrophages via phagocytosis of infected CD4+ T cells. This suggests that the macrophage reservoir may be established far earlier than previously thought. Because it is the job of macrophages to ingest and digest dead and dying cells, Sattentau said, they are attracted to T cells that are dying from HIV infection. That likely happens on a large scale just days after infection, as soon as viral proliferation takes off in gut-associated lymphoid tissue (GALT).

The benefit of a nip in the bud

Still, most latent HIV is believed to reside in resting CD4+ T cells, and depleting such reservoirs might improve the prognoses of HIV infection. Mathias Lichterfeld of Massachusetts General Hospital in Boston reported data that support this hypothesis. He and his team found that people who start HAART during acute infection have HIV DNA levels in their latent CD4+ T-cell reservoir similar to that of elite controllers. Those levels are, moreover, about 100 times lower than those observed in people who start treatment only in the chronic phase of HIV infection.

Lichterfeld and his colleagues drew these conclusions from studies measuring the levels of HIV DNA in CD4+ T cells from nine HIV-infected people. All of them had been on HAART for about 10 years, but three were put on the regimen before they had HIV antibodies in their blood (which typically occurs two to four weeks after infection), while six started the treatment later than that. The HIV DNA detected probably included defective virus, but Lichterfeld said the amount of replication-competent viral DNA was also a lot lower in those who started treatment early. He and his team further found that the HIV-infected individuals who started HAART early had a microRNA expression profile—used as a marker of disease activity—similar to that observed in HIV negative controls, suggesting that they were healthier than those who started HAART late.

“The data suggest that you benefit from early treatment initiation,” Lichterfeld argued, “because it does reduce the amount of reservoir that you have in your system. If treatment options become available that target the reservoir, it would probably be of benefit if you have a low amount of reservoir to start with.”

Replicating with HAART?

Draining HIV reservoirs in people on HAART requires not only the eradication of latently infected cells but an end to their replenishment as well. Researchers at Keystone presented conflicting data on whether residual HIV replication—the kind that would replenish latent cell reservoirs—occurs in people on HAART who have no detectable HIV in their blood.

Sarah Palmer, of the Karolinska Institute in Sweden, and her colleagues found no evidence of viral evolution in such people. That, she argued, suggests an absence of HIV replication—at least in the tissues examined.

Palmer and her colleagues determined the RNA sequence of single HIV particles in plasma taken from eight individuals just days before they started HAART. (Five of them started HAART during acute infection, and three during chronic infection.) They then compared these with sequences of single integrated proviral HIV DNAs in CD4+ T cells from samples of the same patients’ peripheral blood and GALT taken between four and 12 years later.

While individuals who started treatment later had a more diverse set of unique HIV sequences, none of the volunteers showed any difference in the kinds of sequences found in the samples taken shortly before they started HAART and years later. This indicates that there is no sequence evolution in peripheral blood and GALT in people on HAART and, to Palmer, suggests that there is little or no residual viral replication. Palmer also said that most CD4+ T cells in these tissues only harbored one copy of integrated HIV DNA. This, she noted, is good news for HIV eradication efforts. “If you have to purge your reservoirs, you don’t have multiple HIV molecules that have to be purged,” she said. “There is not a large amount of replication that continues to reseed the reservoir. [So] hopefully, once we purge this reservoir, it’s not being reseeded as long as you stay on HAART.”

Not everyone agrees. Tae-Wook Chun, an associate scientist at the National Institute of Allergy and Infectious Diseases (NIAID), said that HIV might replicate at such a low level that many more sequences in tissue sites might have to be determined to be sure it isn’t evolving. “It might take several years to see meaningful viral evolution,” Chun said.

Mario Stevenson of the University of Miami added that the HIV sequences Palmer looked at in tissues came from integrated provirus and could therefore have been generated from HIV that replicated a long time ago, before it integrated into the cell’s genomes. A better way to detect recent HIV replication, he suggested, is to study nonintegrated HIV DNA, which only stays in cells a few days after HIV replication and then gets degraded. In previous studies, Stevenson and his colleagues found nonintegrated DNA in the GALT and lymph nodes of about a third of a group of people who had been on HAART for three and a half years on average (1).

Stevenson said this probably indicates that HIV is replicating in these supposedly latently infected tissues. He cautioned, however, that the presence of nonintegrated DNA could be the result of only one cycle of infection, where a cell got infected but didn’t pass the virus on to other cells. To distinguish between the two possibilities, Stevenson and his colleagues are currently sequencing single, unintegrated HIV DNAs in the GALT and lymph nodes collected from HAART patients to see if the virus is evolving. Such sequence evolution would suggest multiple rounds of HIV replication.

Stevenson also reported results from a study in which biopsies of GALT and lymph node tissues were taken from patients at zero, one, three and six months after they started HAART. The tissues were then assayed for viral RNA, nonintegrated viral DNA and intracellular drug levels in CD4+ T cells. Virus was undetectable in the plasma of the patients one month after HAART was started, but still detectable in GALT and lymph nodes; six months into therapy, most of the 15 individuals studied so far actually had more viral RNA and nonintegrated viral DNA in GALT and lymph node tissue than they had before they started treatment. The lymph nodes, which had the most viral transcripts, actually had the lowest intracellular drug levels in their CD4+ T cells, suggesting that drug concentrations there were insufficient to suppress viral replication. “Some drugs in lymph node are undetectable,” Stevenson observed.

One possible explanation for this, said Stevenson, is that the cells in lymph nodes are more activated than cells in blood, and more activated cells are more likely to actively export drugs. “If that’s the case, then we could come up with strategies to inhibit the drug exporters,” he said. In any case, he added, his results suggest that the latent reservoir in people on HAART is constantly replenished by a low level of ongoing replication in some tissues. “I don’t think we can be arrogant enough to say that the viral reservoir is simply viral latency, and nothing else is going on,” Stevenson said. “If we are to eradicate the reservoir, the first thing we are going to have to do is take care of replenishment.”

HAART in simian models

Studies of the latent HIV reservoir in HAART patients are difficult to conduct because they involve repeated biopsies. There is thus an urgent need for animal models that mimic HAART. Researchers at Keystone reported that they may have come up with such models.

Binhua Ling from the Tulane National Primate Research Center accomplished the feat in Chinese rhesus macaques infected with the mac239 strain of simian immunodeficiency virus (SIV). All four of the chronically infected animals she studied had undetectable levels of SIV—defined as less than 30 copies per ml of plasma—for two months in response to treatment with a combination of three antiretroviral medicines (ARVs). (One, however, did have detectable viral load at one time point.)

Jeffrey Lifson of the AIDS and Cancer Virus Program at SAIC Frederick, Inc., Frederick National Laboratory, who collaborated with Ling on her study, reported that in a separate study, he and his colleagues achieved suppression of SIVmac239 viral load to undetectable levels in six Indian rhesus macaques with a combination of several ARVs. This was probably tougher to pull off, said Lifson, because SIVmac239 tends to replicate to higher levels in Indian rhesus macaques than in the Chinese variety.

Reactivating the reservoir

One currently favored strategy for eradicating viral reservoirs involves inducing HIV replication in latent cells, so that they die as a result of renewed virus replication or can be targeted by drug treatment or immune responses. One drug that researchers hope might roust HIV from its hiding places is the histone deacetylase inhibitor SAHA. Lifson, in fact, is currently testing that possibility in his Indian rhesus macaques.

SAHA, as it happens, is also being tested in humans. Recently, David M. Margolis of the University of North Carolina at Chapel Hill reported promising results from one of the first Phase I human trials examining this drug’s potential as a rouser of latent HIV. He recently reported that, in a handful of HIV-infected individuals on HAART, a single dose of the drug leads to an increase in cell-associated HIV RNA expression (see In Pursuit of a CureIAVI Report, Jan.-Feb. 2012). At Keystone, Margolis shared further results that now included seven patients from the study. On average, SAHA treatment led to a roughly 5-fold increase of cell-associated HIV RNA detected in CD4+ T cells in the blood.

However, Chun reported that a 48-hour ex vivo SAHA treatment of resting, latently infected CD4+ T cells isolated from the blood of HIV-infected individuals on ART did not increase the number of free viruses (measured by counts of cell-free HIV RNA) produced by the cells, as compared to untreated cells.

Margolis said one possible reason for the difference between his observations and Chun’s could be that 48 hours might be too long as a treatment. “Prolonged exposure to these drugs at this concentration [could] probably have nonspecific effects on the cell,” he said, adding that 48 hours is much longer than it takes for one dose of the drug to be cleared out of the body. Chun, however, countered that the cells in his experiments were very viable, suggesting that his treatment wasn’t too toxic, and that he got similar results after treating cells for less than 24 hours. Still, Margolis isn’t convinced: Even if normal cells are not killed by SAHA in vitro, he said, such exposure may prove too toxic for use in the clinic. Chun, for his part, offered a different explanation for the discrepancy in their findings, noting that he assayed RNA levels of free HIV particles, whereas Margolis measured the levels of HIV RNA inside cells. It is thus possible, Chun argued, that SAHA induces an increase in HIV RNA transcription, but that this does not necessarily translate into an increase in HIV virions actually released from the cell.

Insights into the Elite Control of Infection   

One reason some people can suppress HIV even without HAART is that their CD8+ T cells are better able to kill HIV-infected cells. This is especially true of those who have certain MHC receptor variants that infected cells use to present HIV antigens to CD8+ T cells. Variants called HLA B27 and B57 have, for example, been associated to some degree with the ability to suppress viral load to very low levels.

Yet, oddly enough, most people who have those two variants don’t display lower viral loads, suggesting that other factors contribute to the suppression of HIV in elite controllers who carry these genes. To identify those factors, a team led by Bruce Walker of the Ragon Institute studied 10 HIV-infected people, all of whom had the HLA B27 variant, but only half of whom were able to control their viral load.

The key difference between controllers and non-controllers, it turned out, wasn't in the types of epitopes presented to CD8+ T cells, Walker reported. In people with HLA B27, just one HIV Gag epitope called KK10 is presented to CD8+ T cells. A closer look at the KK10-specific CD8+ T cells did reveal, however, that in the elite controllers these cells were better able to kill HIV-infected cells and inhibit virus replication in vitro than in the non-controllers.

Because there was no difference in the antigen presenting cells between the two groups, the researchers reasoned that the secret of better control might reside on the other side of the equation: the T-cell receptors (TCRs) on CD8+ T cells that recognize the KK10 epitopes. And indeed, TCR gene sequences in KK10-specific CD8+ T cells revealed that controllers had TCR types that were also found in the CD8+ T cells that could better kill virus in vitro. By contrast, none of the TCR types found in the non-controllers were found in the more efficient CD8+ T cells.

Walker noted that knowledge of the TCR types induced by a vaccine might allow researchers to assess if a vaccine response is likely to be effective. He and his colleagues are already looking at which TCR types are induced in volunteers who received a vaccine candidate containing HIV Gag. —AvB

A direct route to the cure?

The only person whose HIV infection is believed to have been cured is Timothy Brown, who underwent a number of anti-cancer treatments for acute myeloid leukemia before receiving a risky bone marrow transplant. The marrow donor was homozygous for the CCR5Δ32 allele, which abrogates expression of the HIV co-receptor CCR5 on cells, leaving carriers resistant to HIV infection (see In Pursuit of a CureIAVI Report, Jan.-Feb. 2012). It remains unclear exactly which part of Brown’s complex cancer therapy cured him, and whether this feat can be replicated in others. Researchers are therefore interested in reproducing the Timothy Brown result.

John Mellors from the University of Pittsburgh noted that Brown had received an allogeneic bone marrow transplant, in which the replenished marrow comes from a different person. The donor cells (which have to be matched in at least seven of the eight MHC alleles) used in such transplants recognize the remaining host cells as foreign and kill them. This makes it more likely, said Mellors, that the donor cells will fully replace the host’s immune cells if introduced after whole body irradiation. But there is also a chance that the host cells will reject and kill the donor cells. In that case, the individual would die from leukemia because the remaining leukemic cells would outgrow the normal cells from the donor. This, Mellors explained, is partly why allogeneic stem cell transplants have a 25% mortality rate.

Mellors and his colleagues thus studied 10 HIV-infected individuals from three medical centers in the US who had received a less risky variation of the treatment: an autologous bone marrow transplant, in which the patient’s immune cells are replaced with his or her own bone marrow cells. This approach has a mortality rate of less than 5%. Its disadvantage, though, is that some of the transplanted cells might carry integrated HIV DNA. “You can’t sort cells for latent HIV,” Mellors said, adding that this is probably one of the reasons why this approach has not cured any of the 10 HIV-infected individuals he studied. Another possible reason is that the approach doesn’t result in the replacement of all host cells because the donor cells don’t kill existing cells that they recognize as self.

Mellors and colleagues are therefore enrolling 15 HIV-infected individuals with leukemia or end stage refractory lymphoma, who need an allogeneic bone marrow transplant, into a clinical trial. They will also try to find donors that are homozygous for the CCR5Δ32 allele, as was Brown’s.

If this results in a cure in at least some cases, it will prove that Timothy Brown’s case was not just a fluke, and could help elucidate why he was cured. Whatever happens, though, the approach isn’t likely to ever become routine, given its high risk. “It’s just proof of principle, that’s all,” Mellors said. “It’s got no practical significance whatsoever, none. This is heroic, and totally impractical.”

Perhaps, but the Timothy Brown story has also inspired the formulation of more practical cure strategies. Paula Cannon of the University of Southern California reported that she and her colleagues are preparing a human trial that involves autologous bone marrow transplants. Before the cells are transplanted back to the patients, however, they will be treated with modified adenoviral vectors that carry the gene for an enzyme that knocks out the CCR5 gene. This typically results in destruction of the CCR5 gene in about 5% of the cells, she noted. But studies in humanized mice suggest that because the descendants of cells successfully modified this way are resistant to HIV, they have a competitive advantage once infused into an HIV-infected animal, and eventually become the dominant population of immune cells (2). Cannon said her team is preparing to apply for FDA approval to test the strategy in humans.

Another approach to curing HIV infection relies on the excision of integrated HIV DNA from cells that carry it. Jan Chemnitz from the Heinrich Pette Institute in Hamburg reported that he and his colleagues have made humanized mice that express Tre recombinase, an enzyme that can excise HIV DNA from the genomes of infected cells. To do so, the team transduced human CD4+ T cells or CD34+ hematopoietic stem cells with a lentiviral vector expressing Tre recombinase, and injected the genetically modified cells into humanized mice so that the cells could populate their immune system.

When they infected the mice with HIV-1, they found that the mice expressing the enzyme had significantly lower viral load and more CD4+ T cells than those that did not express it. Chemnitz said this is the first therapeutic approach that can remove proviral DNA of circulating wildtype HIV strains from the genome of an infected cell, and the first proof in an in vivo model that such an approach can attenuate HIV infection. Some animals were more or less cured, he said.

In its current form, the strategy does not excise HIV DNA from latently infected cells because HIV Tat (which is only present in productively infected cells) must be present to induce expression of the recombinase. But infusing patients with CD4+ or CD34+ cells that can express the recombinase should help increase the proportion of healthy immune cells that can easily get rid of HIV once they get infected. If combined with a treatment that reactivates the latent reservoir, this healthier immune system might be better able to kill infected as well as reactivated cells, Chemnitz suggested.

Another limitation is that the Tre recombinase used in the experiments can only excise integrated HIV DNA from an HIV clade A strain from Tanzania. The researchers are now designing a recombinase that they hope will excise many other HIV strains as well. Chemnitz plans a human trial using an enzyme with a broader specificity in about two years. He and his colleagues also hope to modify their strategy to target the latent HIV reservoir, developing lentiviral vectors that can be used to excise HIV DNA in latently infected cells even in the absence of HIV Tat.

Molecular decoys for HIV

Because the gp120 part of HIV’s Envelope spike needs to bind both CD4 and CCR5 proteins to infect target cells, free-floating molecules that effectively mimic these receptors could serve as decoys that inhibit the process. Unfortunately, efforts to use CD4 itself as a competitive inhibitor of infection have so far failed, said Farzan. But new approaches appear to hold more promise.

Farzan shared with the audience at Keystone the results of studies using a chimeric molecule to effect such inhibition. The molecule bears not only the part of the CD4 receptor that binds gp120, but also the portion of the CCR5 co-receptor that binds most tightly to gp120. That part, identified by Farzan’s lab in 1999, consists of four sulfotyrosines (tyrosines that carry a sulfate group). Farzan and colleagues fused these sulfotyrosines to a modified version of CD4, called CD4 Ig, that consists of the two gp120-binding immunoglobulin domains of the CD4 receptor fused to the constant (Fc) domain of an antibody.

Farzan said that the resulting molecule has a 100% breadth of binding, and a potency similar to recently identified broadly neutralizing antibodies (bNAbs), such as VRC01. His chimeric protein neutralized all 22 HIV-1 strains tested, including the most evasive variants and strains that use the CXCR4 co-receptor to infect cells. It even neutralized three SIV strains and two HIV-2 strains, something no bNAb has yet accomplished. “That’s a degree of breadth that’s unprecedented,” Farzan said.

He noted that because the molecule binds to both the CD4 and the CCR5 binding sites on gp120, HIV variants that mutate to escape it would in all likelihood also be much less efficient at infecting cells and replicating. Because it is a protein, Farzan said, the molecule may not be suitable for use in a topical microbicide gel formulation. Instead, he and his team plan to express it in muscle cells, by injecting adeno-associated virus (AAV) carrying a gene encoding the molecule into muscle cells. This gene transfer approach, developed by Philip Johnson of The Children’s Hospital of Philadelphia, has already been shown to express truncated antibodies, including CD4 Ig, in rhesus macaques for almost two years, Farzan said, adding that such persistent expression protected the animals from challenge with an easily neutralized virus (3).

Farzan hopes to use this AAV gene transfer approach in rhesus macaques to test whether it can protect against challenge with more robust viruses and drive down viral load in infected animals. The approach should be quite safe in humans, Farzan suggested, because it uses a modified AAV that does not integrate into the genome but stays in muscle cells—which don’t divide—as episomal DNA that doesn’t multiply. It therefore does not carry the risk of causing cancer.

1. Nat. Med. 16, 460, 2010
2. Nat. Biotechnol. 28, 839, 2010
3. Nat. Med. 15, 901, 2009