Humanity's war against the HIV has been a constantly shifting battle, thanks to the wily nature of the virus. But humans, being quite wily themselves, have persevered in the search for a cure despite setback after setback. Recently, scientists found a way to snip HIV's genetic code out from the genomes of infected cells. The technique's only been tried in cultured cells so far, but it could represent a powerful new weapon in the arsenal against AIDS.

But just what makes HIV such a formidable foe?

In most viruses, the genetic transfer of information mimics a host's: DNA is transcribed to RNA, which is translated into a protein. But HIV is a retrovirus (in the family Retroviridae), meaning that it carries its genes around in the form of a single strand of RNA, rather than a DNA strand or double-helix. In infected persons, HIV hacks the cells of the body with the help of an enzyme called reverse transcriptase, which allows it to create a DNA version of the viral genome after it invades the cell. The viral DNA strand becomes integrated into a host's genome, and the cell copies out the instructions in those genes, creating new copies of the virus that can burst forth and infect other cells.

(More from World Science Festival: Cinema Peer Review: 'I Origins')

In normal DNA replication, a cell employs a variety of molecular proofreaders to help avoid changes to the genetic code. But in reverse transcription, there aren't those checks, meaning that a retrovirus' genome can mutate very quickly. This is one reason HIV is particularly good at ducking treatments — it mutates so quickly, it's hard to figure out what anchor a drug or vaccine can latch onto.

HIV is also a member of the Lentivirus genus within the retroviruses, giving it another nasty set of armaments. Other retroviruses can only infect cells in the midst of division; lentiviruses aren't bound by that restriction.

The AIDS virus is also hard to destroy because it tends to infect the very cells designed to destroy it: a kind of white blood cell called a CD4 lymphocyte. These types of cells can be activated to fight pathogens, but can also lie inactive. If infected, these dormant CD4 cells can maintain a hidden reservoir of HIV that can bring the infection roaring back to life even after a successful course of treatment; most drugs can't touch these reservoirs.

(More from World Science Festival: The long shadow of the Manhattan Project)

As of now, there are two main strategies for a potential AIDS cure. One is a bone marrow transplant, first used by German doctors in 2008 to treat an HIV-infected man who was also suffering from leukemia. The doctors gave the patient marrow from a donor who possessed a special mutation called Delta 32, which helps resist HIV infection. Nearly two years after the treatment, the doctors could not find any trace of HIV in the patient's blood, bone marrow, or other organs. But bone marrow transplants are too risky a procedure to become a widespread cure.

One option for a "functional cure" involves patients taking antiretroviral medicines, which interfere at various stages of HIV infection and replication, very soon after infection. This approach has been shown to leave patients with very low levels of HIV in their blood, even after they stop taking the drugs. But there are limitations: a patient has to start treatment right after infection, and it's still unclear just how long the effects can last. The "Mississippi baby," seemingly cured of HIV after an intense antiretroviral therapy regimen lasting for 18 months shortly after her birth, has shown signs of HIV infection again.

(More from World Science Festival: Five alternate uses for the Moon)

And the genome-editing techniques unveiled this week, though promising, still have to be proven in human subjects. There could be other complications as well. To edit HIV out of the genome, Temple University researchers Kamel Khalili and Wenhui Hu use a specially designed small strand of RNA that sniffs out the virus' code. But, as mentioned before, HIV has a predilection for mutation, which could complicate creating effective code sniffers. Hopefully, further testing of Khalili and Hu's technique won't send them back to the drawing board once more.