HIV Integrase (IN) works like other integrases: it jams viral DNA into human DNA so that the human cell machinery will "read" the viral DNA and make viral products. It is a promising drug target because it has this essential function in the HIV life cycle, and because it is unique to viruses. That is, its a disease-only enzyme with no mammalian equivalent. A drug with good affinity to IN wouldn't touch any human proteins, and that's a good thing.
IN is obviously more intensively studied than other integrases because HIV is the retrovirus that is kicking our collective ass the most.
There are two main challenges to fully understanding and hammering IN:
HIV has a quick turnaround (lifecycle as short as 1.5 days) and a high error rate during the reverse transcription process. Raltegravir was the first FDA approved integrase inhibitor, and has only been used in HAART since 2007. There are already several resistant HIV strains. Modelling this resistance (creating and trialling mutants in a lab) indicates that this is most likely due to two simultaneous point mutations - two separate amino acids in the IN protein have changed to nullify the drug.
- IN is hard to look at. It doesn't suit the normal approach, when modelling a protein and its interactions. Ideally, you start by crystallising it. You actually try for several crystals, to see it in several states - alone, bound to DNA or other protein, bound to the drug of interest. You then perform X-ray crystallography on these, and get a highly detailed image of its structure - especially of its catalytic site. You can then use this image to work out how it interacts, with old school stick-and-ball modelling or computer / molecular dynamics modelling.
To explain point 2, the issue with IN is that “Direct physical and structural studies of full-length integrase have been impeded by its propensity to form large aggregates under reaction conditions.” (IN fragment structure, Nature 2001). This means that it doesn't form crystals, just gooey clots.
Only parts (some domains) of IN have been crystallised successfully. IN is multi-domain, with the links between the domains being both flexible and variable, so the shape of the entire complex (especially when bound to DNA) was conjectural. In 2010, a complete and bound integrase from a different virus, HFV, was crystallised, however. As HFV has an identical catalytic site to HIV, this analogous integrase structure should be useful in drug design.
This was / is being written to help writer's block with an assignment / literature review on the topic.