Interfering with HIV

A fundamental biological process that was first discovered only eight years ago could revolutionize research and medicine, and may hold promise for HIV infection

By Andreas von Bubnoff, PhD

This year's Nobel Prize for the discovery of RNA interference (RNAi) to Andrew Fire and Craig Mello is the preliminary end point of a rise to prominence that can only be described as meteoric. The award came just eight years after Fire and Mello found in the nematode worm Caenorhabditis elegans that pieces of double-stranded (ds) RNA are much more powerful than single-stranded RNA in specifically inhibiting expression of genes with the corresponding sequence (Nature 391, 806, 1998).

The dsRNA pieces Fire and Mello used were several hundred bases long, too long to specifically inhibit gene expression in mammalian cells. That's because they would induce the interferon response, a non-specific general shutdown of gene expression. But only three years later, it became clear that very short dsRNA pieces—around 21 nucleotides, so-called short interfering RNAs (siRNAs)—can specifically inhibit genes in mammalian cells as well (Nature 411, 494, 2001).

It is now clear that these siRNAs inhibit gene expression through the same natural phenomenon that cells normally use to regulate their own genes. The cells do this by transcribing genes that encode micro RNAs (miRNAs), which are the functional equivalents of siRNAs. A ribonuclease protein called Dicer helps process these miRNAs into short dsRNAs that look just like siRNAs, and from there the cell treats both (siRNAs or miRNAs) the same, in that one strand is incorporated into an enzyme complex called the RNA-induced silencing complex (RISC; Figure 1). Once that strand binds a complementary target mRNA, the target MRNA is degraded or is not translated into protein. The complex acts catalytically, meaning it is recycled and can act again and again. That, combined with the exquisite specificity afforded by the nucleotide sequence matching, explains why RNAi has such potential, in research and possibly medicine too.

Gene knockout

For molecular biologists, RNAi has become a powerful and specific tool to study gene expression and function by making it much easier to knock out genes than ever before, says Kuan-Teh Jeang, who studies miRNAs at the US National Institutes of Health (NIH). "RNAi has now become the poor man's fast knockout tool," he says. This has made it possible to screen siRNA libraries to find host cell genes that are required for HIV replication or infection, says John Rossi of the City of Hope Comprehensive Cancer Center in Duarte, California.

Almost immediately after the discovery that siRNAs can work in mammalian cells, researchers started thinking about possible applications, including prevention and treatment of HIV. "HIV was an obvious [target]," says Bryan Cullen of Duke University Medical Center in Durham, North Carolina. Cullen's group was among the first to show that siRNA can inhibit HIV replication in cultured human T cells, one of the main cell types that HIV infects (J. Virol.76, 9225, 2002).

Initially, researchers transfected siRNAs transiently into cells, but soon they found a way to coax cells into expressing them constitutively. They infected cells with viruses engineered to insert genes for short hairpin RNAs (shRNA) into the host cell genome. The host cell processes them in a similar way to endogenous miRNAs to inhibit the expression of target genes.

These advances opened the door to a gene therapy approach by introducing cells that stably express shRNAs into HIV-infected patients. Several groups are planning to start Phase I clinical trials in the next two years.

Delivery obstacle

At the same time, Judy Lieberman's lab at Harvard Medical School is working on ways to deliver siRNAs directly to cells to treat or prevent HIV infection. One major obstacle, she says, as with all gene therapy approaches, is delivery.

"[siRNAs] don't naturally get into cells," Lieberman says. In a study three years ago she literally forced the siRNAs into the livers of mice. She used high volume injection into the blood, temporarily damaging the cell membranes of liver cells so the siRNAs got in and protected the cells from hepatitis infection. The problem was that the volume was so large that the treatment also resulted in heart failure.

In more recent experiments, Lieberman has encapsulated siRNAs into liposomes to get mucosal surfaces to take them up, for example in the genital tract of mice. She has found that this approach can silence genes for more than a week, and has shown that siRNAs targeting herpes simplex virus type 2 (HSV-2) genes can protect mice from HSV-2 infection by silencing any viral genes that enter cells in a potential transmission event (Nature 439, 89, 2006). HSV-2 is the leading cofactor for HIV transmission in the world, increasing people's susceptibility to HIV infection (see HIV prevention in a pill?, IAVI Report 9, 4, 2005).

Prevention or treatment

Next Lieberman wants to use the approach to develop a microbicide women could use to prevent HIV transmission. "Because the silencing lasts for a while, you don't have to remember to use it immediately before you have sex," she says. It could also be cheap because it uses very small amounts of siRNA; according to Lieberman, one dose in humans could cost as little as US$8.

She is also developing a method that can direct siRNAs to HIV-infected cells inside the body. To this end she has made fusion proteins of protamine (a protein that binds and condenses the DNA in sperm) to bind the siRNA, and an antibody that recognizes proteins on target cells like the HIV Env protein. This approach can suppress HIV replication in cultured T cells, she says. In whole animals, an Env-specific antibody directed the fusion protein to tumor cells expressing Env in the flank of mice (Nat. Biotech. 23, 709. 2005). These experiments show that cell-specific delivery of siRNAs is possible, but it will be a long way until clinical trials, Lieberman says. Depending on the antibody targets, the method could be used to prevent HIV infection in the first place or to treat infected patients.

Still, some experts say that to treat a chronic disease like HIV, gene therapy is a better approach than delivering siRNAs directly, which only has a temporary effect. "A constant supply of siRNAs is required to control a chronic virus infection," says Ben Berkhout of the University of Amsterdam. "The best way to achieve [that] is with a gene-therapy approach in which the siRNAs are stably expressed."

Berkhout and others are planning Phase I gene therapy trials. They will use an HIV-derived lentivirus to introduce HIV-suppressive shRNA genes into the genome of CD34+ hematopoietic stem cells taken from HIV-infected individuals. CD34+ cells give rise to T cells and macrophages, two of the major cell types infected with HIV. These CD34+ cells will then be reintroduced into the patient's blood. Berkhout says that the hope is that the protected stem cells will preferentially survive and reconstitute the patient's immune system, as only the untreated, non-protected cells will be killed by HIV.

So far, animal studies suggest that the approach could be safe and efficient. In transgenic mice, shRNAs can be expressed without deleterious consequences for the host. What's more, Rossi's lab has shown that CD34+ stem cells transduced with shRNA could still block HIV replication even after they differentiate into T cells and macrophages.

Safety issues

However, the safety of gene therapy in general is still a major concern. The latest major set back came when three children treated for severe combined immunodeficiency disease (SCID) in a clinical trial in France got leukemia, most recently last year, because the virus used to treat them-murine leukemia virus-had integrated into sites upstream of an oncogene, activating its expression.

But the results of a recent Phase I clinical trial suggest that an HIV-derived lentivirus did not show such dangerous insertion events after 21 months (Proc. Nat. Acad. Sci. 103, 17372, 2006). In the trial, five advanced-stage AIDS patients who did not respond to at least two current antiviral regimens were treated once with autologous lentivirus-transduced CD4+ T cells expressing an antisense RNA to the HIV env gene. "It is the first report of patients being treated with any kind of lentiviral vector," says lead researcher Carl June at the University of Pennsylvania. "We have now followed up the first two patients for three years, [and] there are no adverse events. It is safe from what we have observed." June adds that the researchers also looked at almost 200 integration sites and did not observe that the virus integrated into any regions on chromosomes known to be problematic.

What's more, the lentivirus is derived from HIV itself, which does not seem to cause cancer in HIV-infected individuals. "There has never been a known case of a viral insertion causing a cancer in any [HIV-infected] patient," Rossi says. He has also looked at about 130 lentivirus integration sites and found that it almost always integrated into introns, genomic regions that are inactive. "It's relatively benign," Rossi says of the lentivirus vector, adding that he is now ready to go into patients since experiments in mice and monkeys have indicated that it is safe.

Interference escape

Still, even if gene therapy turns out to be safe, there are additional challenges that any RNAi approach needs to overcome. One major obstacle is the high mutation rate of HIV that allows the virus to escape the RNAi inhibition. "If there is one point mutation, it doesn't work anymore," says Daniel Boden of the Aaron Diamond AIDS Research Center in New York. He found that in cultured T cells HIV can escape from shRNA targeting the HIV tat gene by mutating after just 25 days (J. Virol. 77, 11531, 2003). Boden, for his part, is skeptical as to whether the escape problem can be solved, in part because HIV is not a clone but a quasispecies that varies greatly within an infected individual. "I don't see this as something that can be done," Boden says.

Cullen agrees that an escape of the virus is not a question of if, but when. "Every possible mutant is there 1000 times every day," he says, and the virus only has to change one nucleotide to become resistant, which is easier than changing an amino acid to become resistant to drugs. "This is very easy for the virus to escape from," Cullen says.

But Berkhout says the escape problem can be solved, for example by simultaneously targeting multiple conserved parts of the HIV genome with different shRNAs. "It just becomes a numbers game," Berkhout says. "At some point, the virus just won't escape anymore." To escape from several shRNAs, the virus would have to mutate all of the target sequences at the same time, and Berkhout thinks that's extremely unlikely. Even if it eventually escapes there could well be a fitness cost that renders the virus less virulent, he adds. So far Berkhout hasn't seen escape after two months with a combination of four different shRNAs transduced into cell lines using a lentivirus (Mol. Ther. 14, 883, 2006).

Other groups are targeting host cell mRNAs to get around the escape problem, since these are less likely to mutate. One such target is CCR5, a host cell co-receptor that HIV needs to enter the cell. There is some evidence that there would be little or no side effects since some people who have a deletion in their CCR5 gene are resistant to HIV infection but seem otherwise fine.

Rossi plans to overcome escape in his gene therapy trial by using three different RNA-based mechanisms. One of them uses a ribozyme that specifically cuts the host cell's CCR5 mRNA in an enzymatic manner. The second mechanism involves shRNAs against HIV targets, and the third is a so-called decoy RNA that binds the HIV transcription factor Tat to keep it from activating the transcription of viral genes. "We think we can avoid resistance with this approach because we are going after a cellular as well as a viral target," Rossi says. In a recent study in cultured CD34+ cells, he didn't see any viral replication for 72 days when using this combined approach (Mol. Ther. 12, 900, 2005).

But some caution that CCR5 inhibition with RNAi may not be the best idea. Berkhout points to the recent observation that CCR5 may make symptomatic West Nile virus infection less likely. Boden says targeting CCR5 in late stage HIV-infected patients-which is what most of the trials plan to do-may not be sufficient because the virus tends to switch to a different host cell co-receptor, CXCR4, later in the course of infection. And pharmaceutical companies like Pfizer have CCR5-inhibiting drugs in clinical trials. Given the delivery problems of RNAi, these are more likely to emerge as anti-CCR5 agents than RNAi, Cullen says.

Toxic effects

Escape is not the only thing to worry about. Toxicity also needs to be addressed. Rossi says gene therapy could be less toxic then the current drug treatments for HIV patients such as RT (reverse transcriptase) and protease inhibitors. But Mark Kay's lab at Stanford University showed that mice expressing too much shRNA in all liver cells died of liver failure (Nature 441, 537, 2006). "It overloaded the system," Kay says. "It gives some indication that it is very important to be in the right dosing range." The introduced shRNAs probably interfered with something in the normal processing of endogenous micro RNAs, because both use the same pathway, Kay says. However, he does not expect this to necessarily be a problem for the gene therapies planned by the labs of Rossi and Berkhout. "They are using a vector that generally gives a lower amount [of shRNA] than the vector we use," Kay says, adding that in the worst case, the treated CD34+ stem cells would simply die.

Cell death is exactly what happened when Irvin Chen's lab at University of California, Los Angeles expressed an shRNA under two different promoters in T cells-the promoter that expressed more shRNA was more potent, but also killed the T cells, Chen says (Mol. Ther. 14, 494, 2006). "There is a balance between potency and safety," he says. "We need the most potent shRNA combined with a promoter that expresses the minimum amount of siRNA to minimize the cytotoxic effect," says Dong Sung An, lead author of the study.

Given all these challenges, will RNAi ever be available as a treatment for HIV infection? "The reason that so many people are working on this approach is because we are optimistic that it can work," says the NIH's Jeang. "[But] we are far short of RNAi as a therapeutic agent against HIV."

It's unlikely to expect that RNAi approaches will completely replace other HIV treatments, says Stanford University's Fire. "Throwing away all the drugs and just using RNAi, I don't think that's going to happen," he says.

"I don't think RNAi has any potential against HIV in the next 5-10 years," says Cullen when asked how soon RNAi therapies will be available for patients. Jeang says he doesn't know. "If I could tell you that, I would go to Wall Street and buy a lot of stock."