Making a Monkey Out of HIV

More is becoming clear about a novel host factor that appears central to governing the species-specificity of retroviruses like HIV and could be a future antiviral target

By Philip Cohen, PhD

Popular accounts of scientific discoveries often involve a metaphorical light bulb popping up in some researcher’s head. In reality, those eureka moments aren’t always so illuminating. The cracking of some unsolved mysteries about HIV last year, for instance, came when cells located in Joseph Sodroski’s lab at Dana-Farber Cancer Institute in Boston failed to shine.

His graduate student, Matt Stremlau, was on the hunt for an elusive factor that protects some monkey cells from HIV infection. His approach seemed simple enough: add monkey genes to human cells, stir in HIV genetically engineered to glow in the dark, and pluck out the cells that remained dim—a sign that the dose of monkey DNA had snuffed out infection of the cells by the fluorescently-marked virus. “Finding the restriction factor was an enormous effort. Many labs had tried similar approaches and had failed to find anything,” says Sodroski. “Matt succeeded on his eighth try. It’s a testimony to his tremendous persistence.”

While the discovery was a technical coup, its true importance only became clear over the weeks and months that followed. Sodroski’s team and other labs showed that the protein produced by this gene, TRIM5α, was at the heart of a number of puzzles surrounding the species-specific resistance to various retroviruses. The study of TRIM5α also promises to shed light on an important but poorly understood part of HIV biology, the uncoating of the retroviral genome following infection.

More broadly, the identification of TRIM5α solidified an ongoing trend in HIV research—the discovery of host factors with potential antiviral activity. Only two years before, researchers identified another antiviral protein, APOBEC3G, that is now a focus of intense research (see Guardian of the genome, IAVI Report 9, 2, 2005). And it’s a good bet that there are more host proteins out there that can be turned against the virus. Some attendees of a Cold Spring Harbor meeting on retroviruses in May appeared a bit stunned by the number of putative virus-fighting factors presented at the meeting, says Stephen Goff of Columbia University, whose team studies an antiviral protein in rat cells. “Someone raised their hand and asked: what’s going on here, how many of these genes could there be?” says Goff. “I think the answer is that we’ll keep finding new factors until we’re close to scanning the real number of genes that are out there. And we don’t seem to be close to saturating our screens for those genes.”

Intrinsic immunity

Researchers are clamoring to study proteins like APOBEC3G and TRIM5α because they might offer a novel way to fight HIV—by tapping into what amounts to a previously unrecognized type of immunity, says Paul Bieniasz of the Aaron Diamond AIDS Research Center in New York City. A viral infection has long been known to trigger antibodies and immune cells specifically directed against the virus, as well as rally elements of innate immunity, including broadly-acting cellular responses like the release of interferon-α. But this new class of host restriction factors appears to be always present in the cell and they are highly specific for different viruses. “In a sense, they are even more innate than the classic innate response, because they don’t need to be triggered,” he says. He prefers the term “intrinsic immunity” to describe the protection these host factors provide.

While these restriction factors are now a hot topic in HIV research, the field can be traced back at least thirty years. In the 1970s an activity against so-called N strains of Friend murine leukemia virus (N-MLV) was discovered in some strains of mice and named the Friend virus susceptibility gene 1 (Fv1). But the field has picked up considerable steam in the last six years with the discovery of a similar anti-MLV activity, Ref1 (Resistance factor 1), in human cells and an anti-HIV-1 factor in the cells of New and Old World monkeys (Lv1, for lentivirus restriction factor 1).

In all these cases genetic experiments demonstrated that viral replication was blocked due to some inhibitor rather than the lack of some required factor. HIV, for example, while deadly to humans barely invades rhesus macaque cells before a molecular monkey wrench is thrown into the virus’s replication cycle, freezing it before the virus can reverse transcribe its RNA genome into DNA. While these antiviral activities were discovered separately, intriguing links were found between them. Genetic studies showed that both Ref1 and Fv1 targeted amino acid residue 110 of the capsid protein of N-MLV. And Lv1 turned out to also target the HIV capsid. Viral competition experiments revealed further links. In African green monkey cells, pretreatment of a cell with any restricted virus overwhelmed the cell’s ability to defend against any other, suggesting these activities shared at least one common element that could be saturated (Embo J. 22, 385, 2003).

Researchers got their first peek at the molecular identity of one of these factors when Fv1 was cloned by Jonathan Stoye’s group at the National Institute for Medical Research in London (Nature 382, 826, 1996). The Fv1 protein proved to be very similar to a retroviral coat protein, suggesting it insinuated itself as a decoy into some stage of the viral replication cycle. However, primates don’t have a similar gene, leaving what accounted for the host restriction activity in monkeys and humans unanswered.

The answer finally came from Stremlau, Sodroski, and their colleagues’ hunt for monkey genes that allow human cells to resist glow-in-the-dark HIV. They fingered two separate cell lines that were able to keep the viral lights low—and both proved to have the same rhesus macaque gene, TRIM5α. Further characterization confirmed that they had the right factor. The monkey TRIM5α strongly restricted HIV-1 but not SIV, a virus which successfully infects this species. Conversely, the human version of the protein was much less effective at halting HIV-1 replication than the monkey TRIM5α. Finally, when the rhesus TRIM5α protein was knocked down in monkey cells, these cells became more susceptible to HIV-1 replication (Nature 427, 848, 2004).

Jeremy Luban’s team at Columbia University discovered that TRIM5α also explains two bizarre and confusing observations about the cells of owl monkeys. These primates are unique among New World monkeys in that their cells block HIV-1 replication. And while immunosuppressive drugs called cyclosporins inhibit HIV-1 replication in human cells, cyclosporins have exactly the opposite effect in owl monkeys, strongly enhancing HIV-1 replication. Both observations turn out to stem from the unusual composition of owl monkey TRIM5α, which is naturally fused to cyclophilin A, a cellular protein that both binds the HIV-1 capsid (allowing the targeting and, therefore, restriction of this virus) and interacts with cyclosporins (which compete with virus for the cyclophilin A protein, releasing HIV restriction; Nature 430, 569, 2004).

All in the family

It turns out that TRIM5α is part of a large and poorly characterized gene family with at least 60 members. Before TRIM5α was linked to HIV restriction a number of aberrant TRIM proteins had been identified for their involvement in a wide array of hereditary diseases—for example, mutations in the TRIM family proteins PYRIN, MID1 and MUL respectively cause familial Mediterranean fever (characterized by episodes of fever and peritonitis), X-linked Opitz/GBBB syndrome (patients present with craniofacial, heart and genital abnormalities), and mulibrey nanism (involving growth delays and abnormal development of muscles, liver, brain, and eyes). And genetic breakage and fusion events that splice TRIM family genes PMLRFP and Tif1into other chromosomes were known to trigger the malignant transformation of cells. All these links between TRIM family genes and disease demonstrated that the proteins they encoded play important developmental and biological roles. But these data don’t reveal much about the biochemistry of TRIM5α or how it might fight viruses.

So researchers turned to dissecting the protein itself. The name of this protein family – TRIM, for TRIpartite Motif – refers to a trio of identifiable motifs its members typically contain: the zinc-binding motifs known as RING and B-box, plus a coiled-coil domain that is predicted to fold into a group of α-helices that wrap around each other. In addition, some family members, including TRIM5α, contain another distinctive amino acid sequence known as a SPRY domain, a motif found in many proteins including the antibody-like molecules of the immunoglobulin superfamily.

The SPRY domain came under early scrutiny since this part of TRIM5α can differ dramatically between primate species. By analyzing TRIM5α sequences from 20 primates, Michael Emerman, Harmit Malik, and their colleagues at the Fred Hutchinson Cancer Center in Seattle discovered that the SPRY sequence has undergone episodes of rapid change for at least 33 million years—long before lentiviruses such as HIV evolved about a million years ago. The nature, speed, and sporadic character of these changes are consistent with the theory that new versions of TRIM5α were fixed in each primate lineage following the emergence of new retroviruses (Proc. Natl. Acad. Sci. USA 102, 2832, 2005). “The effect of a new virus on the population might not have been immediately catastrophic,” says Emerman. “It’s possible that over long periods of time, animals with particular TRIM5α variants were protected from the virus, were healthier and produced more offspring, until the new variant took over.”

The SPRY sequence contains four variable regions that in various primate lineages are expanded, duplicated, or have changes in sequences (J. Virol. 79, 3139, 2005; J. Virol. 79, 6111, 2005). Researchers have been able to show directly that altering the SPRY domain of a TRIM5α protein alters the species-specificity of the viruses it restricts. Sodroski found that changing only three amino acids gives the human protein anti-HIV-1 potency rivaling that of the monkey protein. Both Sodroski’s and Stoye’s teams have shown that altering the human protein so that the amino acid arginine at position 332 is replaced by the proline present in rhesus TRIM5α (R332P human TRIM5α) significantly increases the ability of this mostly human protein to restrict HIV-1 (Curr. Biol. 15, 73, 2005; J. Virol. 79, 3139, 2005).

Intriguingly, the R332P human TRIM5α also has the power to restrict SIV more strongly than either of its parents, suggesting that some property of TRIM5α’s structure allows it to adapt to new viral threats with minor sequence variations. Sodroski speculates that this property might be the ability of TRIM5α to bind viral capsids based on recognition of a generalized pattern in these structures that viruses cannot easily alter. “This is a different model than suggesting there is a lock and key fit between TRIM5α and every viral target,” says Sodroski. He points out that this model is reminiscent of how Toll-like receptors recognize molecular patterns on invading microbes to trigger aspects of innate immunity. He thinks the specific sequences in the SPRY domain that target antiviral activity to specific viruses may fine tune this interaction or alter what TRIM5α does to the viral capsid after binding.

The events that follow that binding could be the key to TRIM5α’s viral restriction and researchers have been busy genetically analyzing the RING, B-box and coiled-coiled domains of the protein for clues. RING is known to give some proteins the ability to attach a chemical tag called ubiquitin to other proteins, which are then quickly shuttled into the cell’s proteasomal disposal pathway. In their very first paper on TRIM5α, Sodroski’s team proposed that the rhesus TRIM5α may destroy HIV by ubiquinating its capsid, but their later work argues against that simple model. When they engineered mutations into this domain they found that the antiviral activity of the protein was reduced but not completely eliminated (J. Biol. Chem. 280, 26933, 2005).

In contrast, mutations in the B-Box domain completely remove the antiviral powers of TRIM5α. What the B-box does isn’t known, although in other proteins B-boxes promote protein-protein interaction. So it’s possible the one in rhesus TRIM5α may recruit other proteins to help it fight HIV-1. Coiled-coils are also known to foster protein-protein interactions. When TRIM5α mutants lacking restriction activity, but containing intact coiled-coils, are expressed in the same cell as an active TRIM5α, they suppress its activity. This argues that the coiled-coil domain helps TRIM5α molecules associate with each other and that inactive molecules can inhibit the antiviral activity of this protein complex.

The ability of TRIM5α to form large complexes has made moving beyond these genetic experiments to direct biochemical characterization difficult. But an important step forward was recently reported when Luban’s team showed for the first time in a test tube that direct, specific binding occurs between human TRIM5α and the capsid of N-MLV (Retrovirology 2, 40, 2005).

Structure is function?

The self-association of TRIM5α molecules makes a biochemist’s life difficult, but it may also be important to this protein’s antiviral activity. In fact, one common property of TRIM family proteins is their formation of various structures in cells named for their appearance under the microscope: filaments, speckles, aggregates, and—in the case of TRIM5α—cytoplasmic bodies. “An intriguing idea is that TRIM5α forms one of these bodies around a viral core and separates it from the rest of the cell,” says Tom Hope of the University of Illinois at Chicago. His team is also exploring the possibility that these bodies serve as storage centers that regulate the concentration of free TRIM5α. At this year’s Keystone Symposia on HIV Pathogenesis and HIV Vaccines in Banff, Canada, Hope showed TRIM5α movies his team had made using high tech “deconvolution” microscopy, a computer reconstruction of the cell’s image that eliminates light distortion to create bright, detailed images. These films show cytoplasmic bodies zooming around human cells and reveal these are dynamic structures in which TRIM5α protein is actively exchanged and not insoluble blobs as some researchers had thought. But Hope hasn’t yet shown that the formation of these structures correlates with antiviral activity.

Another theory for how TRIM5α restricts retroviruses is that it corrupts the uncoating process of the viral capsid by either accelerating or delaying disassembly of the viral capsid. Experiments suggest that for viruses like HIV to efficiently replicate, uncoating must happen at precisely the right time, although the reason for that crucial timing isn’t clear. “We generally call this part of the viral life cycle a black box,” says Luban.

Understanding how TRIM5α manages to halt virus replication could have a number of practical benefits. Bieniasz says his team and others are busy mapping the molecular interactions between rhesus TRIM5α and HIV-1 capsid. This would be an important first step to engineering an HIV-1 strain that would evade the monkey protein and productively infect the animals. “For vaccine trials, that would allow you to test the same immunogens in monkeys that you want to test in humans,” he says.

The ability of primate TRIM5α to restrict HIV-1 also suggests a number of theoretical strategies for antiviral therapy. The most direct would be to engineer a human protein that potently restricts HIV-1 and use gene therapy to introduce this gene into the immune cells of HIV-infected individuals, making them HIV-1 resistant. Presumably the molecules that enhance the ability of the human protein to target the HIV-1 capsid would also give it the power to stop the virus. But designing such drugs will be difficult until the biochemical basis of this recognition is better understood.

Indeed as quickly as research on TRIM5α is moving, it is still at a very preliminary stage. And the potential for TRIM5α as an HIV therapy will largely depend on how it is acting, what other proteins are involved in its activity, and how easy it is to target drugs to this pathway, says Sodroski. “Part of what makes this exciting is that TRIM5α acts at an early stage in the viral life cycle where there are many unknown elements, and potentially many opportunities to understand new steps and interrupt them.”

Luban agrees that the biggest payoff from TRIM5α may be what it reveals about HIV. The better researchers know their adversary, the better they are equipped to fight it, he says. “If you want to bring down a tiger, the more you find out about its behavior, what it needs, what it wants, the more chance you have to catch it and not get hurt in the process.”

Figure 1. Shedding light on TRIM5α’s role.

Four cells expressing rhesus TRIM5α protein (visualized in red) were infected with HIV-1 particles (in green). TRIM5α is dispersed throughout the cytoplasm and also clustered in roughly spherical cytoplasmic bodies. Small white squares in the left panel mark cytoplasmic bodies that appear to overlap with HIV-1 particles. The three right panels show close ups of these regions. These studies are consistent with the theory that TRIM5α cytoplasmic bodies may restrict HIV-1 replication. Alternatively, cytoplasmic bodies may help regulate levels of active TRIM5α or be an inactive form of the protein.