Guardian of the Genome

Research into a different kind of viral defense system may yield powerful treatments and solve some mysteries about HIV

By Philip Cohen, PhD

Four years ago, when Ann Sheehy took a trip to her local cinema to see the movie Gladiator, she knew she was in for one hell of a story.

Actor Russell Crowe had recently won an Academy award for his portrayal of Maximus, a once-great Roman general reduced to slavery, seemingly destined to die off the pages of history as a nameless gladiator. But after many gory encounters, Maximus rises from obscurity to battle and defeat his nemesis, the vile Emperor of Rome.

"I'm a history buff so the story of ancient Rome appealed to me, even if it was a bit bloody," remembers Sheehy, a postdoctoral researcher in Mike Malim's laboratory in King's College London. "I know for certain I wasn't thinking about HIV. When I see a movie it's one of the few times I try to forget about what I work on." Ironically, the same transformation of invisible serf into celebrated soldier would soon take place in her own laboratory—albeit on a molecular level.

In this version, the role of the unsung hero was to be played by a human protein known as a cytosine (or cytidine) deaminase. These enzymes can directly mutate genetic material, morphing the cytosine present in RNA or DNA into uracil by twisting around a few atoms. Back when Sheehy watched Crowe strut into the Coliseum, these enzymes seemed of little direct relevance to HIV's lifecycle. "I had never even heard of these enzymes," she says.

But pioneering work by Sheehy, Malim and their colleagues soon revealed that one of these mutator proteins, dubbed APOBEC3G, is actually on the front line in the battle against HIV, engaged in molecule-to-molecule combat nearly from the moment the virus enters a human cell. After a few years, a flurry of publications from top labs, and many recent twists in the tale, the story of APOBEC3G and similar enzymes has become one of the hottest areas of HIV biology—and researchers speak of these proteins in heroic, one might even say cinematic, terms. "They are intracellular guardians and defenders of genomes," Malim recently told scientists gathered at the Keystone Symposium on HIV Pathogenesis held in Banff, Canada.

One reason APOBEC3G is generating such excitement is that since its discovery researchers have been finding more and more human proteins with latent abilities to oppose HIV (see Keystone Report, this issue). "We are just starting to scrape the surface of the cellular defenses that are out there," says Mario Stevenson of the University of Massachusetts Medical School.

The science of these antiviral factors is at such an early stage that researchers haven't even agreed on what to call them. Some see these anti-HIV proteins as an extension of the traditional innate immune system. Others see them as something altogether different and prefer to call them host restriction factors or factors of intrinsic immunity. But whatever moniker they use, many researchers think that therapies based on this different kind of defense may help yield powerful treatments and studying them may solve some longstanding mysteries about the biology of HIV.

One of these mysteries, the function of the HIV protein Vif, is what originally drew Sheehy to Malim's laboratory. Vif is encoded by one of only nine genes HIV possesses, but the precise role of this protein had remained elusive for more than a decade. It was known that the vif gene is required to cause disease in animal models but viruses with the vif gene deleted can replicate normally in certain "permissive" cell lines, while infection of "non-permissive" cells produces only defective virions. Genetic evidence suggested that an unidentified factor in non-permissive cells inhibited propagation of the virus, but this factor was normally inactivated by Vif.

In an effort to unmask that antiviral factor and how it worked, Sheehy took advantage of two human T cell lines known to differ at much less than 1% of their genomes. Importantly, one of these cell lines was permissive to replication of vif-deleted virus while the other was not, suggesting that a defect in the unknown anti-HIV factor was one of the few genetic differences between them.

In 1999, she pulled out several promising candidate genes, the fifteenth of which was the one that would later be known as APOBEC3G. Sheehy initially called it CEM15 after the cell line from which it was cloned. After running CEM15's sequence through a human database, she threw it in the freezer. "I hate to be reminded of that," jokes Sheehy. "Nothing informative came out in that search, and there were some other genes that seemed promising. So I put it in a separate pile of things to look at later." Luckily, the impending completion of the human genome sequence was causing a gene cloning frenzy and new human gene sequences were being deposited in databases on a daily basis. She only had to wait for the right group to sequence the right genes to reveal the connection between CEM15 and other cytosine deaminases. Even so, it took three more years.

But it was worth the wait. As Sheehy, Malim and their colleagues wrote in their now classic paper on APOBEC's discovery (Nature, 418, 646, 2002), "The sequence similarity between CEM15 and cytosine deaminases is potentially provocative." Among the known cytosine deaminases were APOBEC-1, a protein that changes a single cytosine (C) into uracil (U) in the messenger RNA for the intestinal protein apolipoprotein B, and Activation-Induced Deaminase (AID), which is involved in generating antibody gene diversity through the enzyme's ability to mutate DNA. The immediate implication was that human cells had somehow transformed the power of APOBEC enzymes to alter the cell's own genetic information into an antiviral weapon.

As the King's College London team and their colleagues, led by Michael Neuberger at the Medical Research Council Laboratory of Molecular Biology in Cambridge, rushed to test the model, many other top HIV labs joined the fray. Over the next year, papers or reviews on APOBEC enzymes and HIV were being published monthly. In 2004, that pace surged to once a week. From this flood of data, a compelling picture quickly emerged of the life and death struggle of APOBEC and HIV (Figure 1).


Figure 1. Original and New Models of APOBEC3G Antiviral Action. Panel A depicts the original model of APOBEC3G action. In a cell producing virus lacking functional Vif protein, APOBEC3G incorporates into nascent viruses budding from the producer cell (top of panel). When the virus infects a target cell (bottom of panel), APOBEC3G is released and mutates the nascent DNA, leading to its degradation or hypermutation. If Vif is present in the producer cell (top of panel), APOBEC3G is tagged with ubiquitin for transport to the proteasome and destruction. Panel B depicts a new mechanism by which a low molecular mass form of APOBEC3G (LMM APOBEC) inhibits virus. LMM APOBEC is found in resting CD4+ T cells and can block events in viral replication immediately following fusion through a mechanism yet to be determined. It can operate on wild-type virus in the presence of functional Vif protein. Panel C depicts another Vif-sensitive mechanism of APOBEC antiviral action uncovered by engineering a form of APOBEC that lacks DNA-mutating activity (APOBEC oval with X). Even in the absence of its mutational powers, APOBEC3G can demonstrate potent antiviral activity in the target cell. Where this block occurs in the viral replication cycle is still unknown.

[Large image

In this model, the APOBEC mutator enzyme effectively sterilizes viral progeny. It lies in wait in human cells for an invading retrovirus to reproduce, and then sneaks inside nascent virions. When one of those newly minted viruses infects another cell and switches on its reverse transcriptase enzyme to copy its RNA genome into a strand of DNA, the APOBEC stowaway strikes. It riddles the DNA strand with C to U mutations, potentially creating stop codons in the viral protein sequence and fouling splicing sites when the viral genome is later transcribed into RNA. If that isn't enough, since DNA doesn't usually contain U, this copy of the HIV genome can be recognized by cellular DNA repair enzymes which may chew it to bits.

Or at least that's what happens if HIV lacks the Vif protein. Before APOBEC can strike, Vif disarms it by branding its would-be attacker with ubiquitin molecules, chemical tags which mark it for destruction by the proteasome machinery that recycles the cell's own damaged proteins. With APOBEC removed, new virions are free to infect at will and create DNA copies that integrate into its host's chromosomes which spawn the RNA genomes of the next viral generation. So the invading viral army proceeds unhindered.

"It's a model that's makes a lot of sense, that's straightforward and simple," says Malim. This model of APOBEC's mechanism also explained a curiosity about HIV: it is one of the most adenosine(A)-rich retroviral genomes known. This makes sense considering that the C to U DNA changes that are the signature of APOBEC action would result in guanine (G) to A mutations in the RNA genome of new viruses. So a tendency to A-richness would seem a reasonable adaptation for a virus that has done battle with APOBEC-like enzymes for eons. But while the model of APOBEC as mutator turned viral birth control agent is appealing, it is also incomplete. APOBEC, it turns out, has other tricks up its sleeve.

One of the first hints that APOBEC had other talents came last year from the study of the effect of APOBEC on hepatitis B virus (HBV), which depends on the conversion of RNA to DNA in part of its lifecycle just as retroviruses do. It isn't clear if APOBEC plays a natural role in defending against HBV since the enzyme hasn't been found in liver cells. But Didier Trono's laboratory at the University of Geneva, Switzerland showed that, given the chance, APOBEC is fully capable of fighting HBV. His team genetically engineered tissue culture cells to express APOBEC3G and then infected them with HBV. As might have been expected, APOBEC was able to cripple the virus. But what caught the researchers off guard was that in at least in some of the cell lines they engineered, APOBEC3G managed to cripple the virus without causing an appreciable increase in C to U mutation (Science, 303, 1829, 2004).

Researchers were still scratching their heads over this result last January when Sheehy, Malim and their colleagues dropped another paradigm-challenging paper in their laps. Their team was trying to understand exactly how APOBEC3G unleashed its mutational weapon. In similar cytosine deaminase proteins, a quartet of amino acids—histidine, glutamic acid and two cysteines – bound to a zinc atom are responsible for carrying out the alchemy of converting Cs to Us. APOBEC3G has two of these domains. As might be expected then, the researchers found that crippling both domains stripped the protein of its anti-HIV powers, while keeping either intact largely preserved them. All that made good sense, until the team directly assessed the ability of each domain to mutate DNA, and found that only the C-terminal one had this ability (Current Biology, 15, 166, 2005). That means that when the C-terminal domain was inactivated by mutation, APOBEC was completely stripped of its mutational power, and yet remained a formidable force against HIV. "This surprised us, to say the least," says Malim. APOBEC appeared to have the power to mutate HIV to death. Yet that might not be its only weapon, or even its most powerful.

And APOBEC had at least one more surprise in store. Another hallmark of APOBEC's particular kind of combat was supposed to be that it strikes not directly at incoming virus, but its progeny. But in April, Warner Greene's group at the University of California, San Francisco reported an exception to this rule. His team was studying another classic HIV puzzle. Why does the virus replicate efficiently in activated T cells but in resting T cells replicates very poorly? Many explanations have been offered over the years, but Greene's team decided to investigate whether APOBEC enzymes play a role.

They were prompted to do so when they discovered that in most cells APOBEC3G doesn't stand on sentry duty alone but is part of a large or high molecular mass complex that includes uncharacterized RNA molecules. However they found that in resting T cells APOBEC3G is primarily found in a much more streamlined, low molecular mass form. Greene's team recently reported that if this APOBEC3G-lite is eliminated in resting T cells, a known block to HIV reverse transcription largely disappears. And although slimmed-down APOBEC3G appears to have cytosine deaminase activity, Greene's team found that only about 8% of viral sequences isolated from these cells showed evidence of C to U hypermutation (Nature, 435, 108, 2005).

The finding is remarkable for three reasons. It shows that, in some circumstances at least, the APOBEC enzyme can act directly and immediately on the incoming virus, that it can stop the replication of wildtype HIV even if functional Vif is present, and that this activity doesn't appear to require mutation.

But the work also raises an interesting question. If APOBEC has all these powers to stop HIV in its tracks, why didn't our cells evolve to use it? Harmit Malik at the Fred Hutchinson Cancer Research Center in Seattle thinks one possibility is that cells didn't optimize APOBEC to fight HIV because the enzyme evolved for another purpose. His team, together with that of his colleague Michael Emerman, has found evidence that APOBEC proteins have been active guardians of primate genomes for 35 million years, well before lentiviruses evolved about one million years ago. To understand what APOBEC was doing all that time their two labs studied the selective forces on different members of this enzyme family in primates.

The prediction of this type of analysis is that proteins which perform a physiological role for the cell, such as AID, would be driven by the selective forces of evolution to perform that function as efficiently as possible. Once the sequence of the protein has been pushed to near perfection, any change in its sequence caused by random mutation tends to lower its efficiency and it is said to be under negative selection. In contrast, some proteins provide functions that are under constant pressure to change, a feature called positive selection. A good example is a protein like APOBEC3G, which is suspected to interact with proteins from a pathogenic virus. Any change in this type of protein will eventually be countered by a deflecting change in its target. This arms race leads to a genetic conflict where both combatants constantly explore new sequences to cope with each others shifting tactics.

In addition, if it was true that some members of the APOBEC3 gene family (primates have a total of eight similar enzymes designated APOBEC3A, APOBEC3B, etc.) evolved to deal with HIV or similar viruses, then this leads to three predictions. First, some of these proteins should experience positive selection. Secondly, the positive selection should be sporadic, present in lineages whose natural history included a devastating encounter with such viruses while absent in others. Third, the positive selection should be restricted to just the region of each APOBEC gene which interacted with viral proteins like Vif that directly engage it.

Unexpectedly, only one of these predictions turned out to be correct. Six of the APOBEC3 proteins were under positive selection, but that was true in all ten primate lineages the researchers examined. This suggested these proteins were not fighting sporadic viral infections but some ever-present danger. And when they examined the APOBEC3G sequence in fine detail they found that areas of positive selection were not limited to those mapped biochemically as important for the protein's interaction with Vif, but extended across its whole length making it one of the fastest evolving of all known human proteins (PLoS Biology, 2, E275, 2004). This suggests that in its long history every piece of APOBEC3G has been used to engage some attacker. "The analogy of a Swiss army knife is pretty good," says Malik. "Think of it as having different blades that have been honed to different purposes."

Work by Olivier Schwartz at the Institut Pasteur in Paris suggests that viruses alone may not have driven APOBEC to develop these diverse set of skills. Instead the protein may have been shaped by fighting a danger from within, endogenous retroviruses. Endogenous retroviruses have a life cycle similar to HIV, including the reverse transcription of their genome and incorporation of a DNA copy into a new chromosome. The only difference is that endogenous retroviruses never leave a host's cells. They reside within a genome and their raison d'être is simply to spread more copies of themselves. But because these retroviruses can splice themselves into functioning genes, cells have mechanisms to suppress this activity. Now Schwartz's team has shown that human APOBEC3G can provide another layer of protection against some of these retroviruses (Nature, 433, 430, 2005).

Retroviruses would fit the bill of a constant and rapidly shape-shifting adversary for APOBEC3G to combat, and this enzyme is expressed in reproductive tissues where the threat of retroviruses to species survival is most serious. But even if fighting HIV isn't APOBEC's day job, some new data suggests that variations in how it works from individual to individual could partly explain why an HIV infection progresses rapidly to disease in some people but causes disease slowly in others.

One study of more than 3000 HIV-infected people found that a variant of APOBEC3G common in the African American population is associated with acceleration to the development of AIDS symptoms (J. Virol., 78, 11070, 2004). However, a French study of APOBEC3G gene variants in 245 HIV-infected Caucasians with slow disease progression and 82 with rapid progression found no significant association between gene variants and disease (J. Infect. Diseases, 191, 159, 2005), suggesting the influence of APOBEC genetic background could differ between populations.

Genetic differences in the infecting virus may also have an effect on disease progression. At the recent Conference on Retroviruses and Opportunistic Infection in Boston, Australian researchers reported that in a group of 227 HIV patients they found that virus from about 3% showed evidence of G to A hypermutation and, in all these cases, the integrated proviruses showed evidence of defective vif genes. Importantly, these patients also showed a lower viral load before receiving antiretroviral drugs.

While the clinical picture is still developing, any evidence of APOBEC's influence on the course of disease fuels the hope that boosting APOBEC activity could one day be a powerful therapeutic. "Ideally, you'd want to stabilize APOBEC enzymes, raise their level or make them resistant to degradation," says Stevenson. The most actively pursued strategy is to screen chemical libraries for small molecules that block Vif's ability to mark APOBEC for destruction. Dana Gabuzda at the Dana-Farber Cancer Institute in Boston has been pursuing this approach and now Stevenson and Greene say they are independently joining the hunt.

But some of the recent research into APOBECs suggests that other types of therapeutics could also prove effective. One human APOBEC family member, APOBEC3B, appears naturally impervious to Vif but is not normally expressed in cells which HIV infects (Current Biology, 14, 1392, 2004). Drugs that upregulate theAPOBEC3B gene in these cells could therefore erect a new barrier to viral replication. A completely different strategy would be to promote the formation of the low molecular mass version of the APOBEC3G protein in more cells that, as Greene's team has shown, could freeze the virus early after its invasion into the cell.

No one is under the illusion that developing a drug to target the APOBEC pathway will be easy. The road to developing a new drug against HIV is a hard, long one with no guarantee of success—witness the 20-year struggle to get an effective and safe HIV integrase inhibitor to market. And until further research nails down the normal physiological role of these enzymes, worries about possible toxic side effects of APOBEC drugs are likely to make companies cautious about trying to exploit their potential. "Pharmaceutical companies don't embrace novel approaches and they haven't embraced APOBEC," says Stevenson. But he adds that companies will be more likely to pay attention if academic labs come up with some promising compounds.

If such compounds can be found, Sheehy speculates that targeting the APOBEC pathway could give medications unusual bang. After all, current drugs simply block critical functions of the virus. She compares APOBEC-based pharmaceuticals to vaccination. "Vaccines tend to work so well because they harness the strength of your own body to fight the virus," she says. "Similarly a drug that can keep APOBEC super active would help one of our own proteins fight the virus for us." In the end, the APOBEC enzymes may not be the type of heroes that Hollywood rushes to immortalize in a big budget movie. But researchers seem convinced that a drug which boosts the ability of these enzymes to fight HIV may still be worthy of the term "blockbuster."