Looks like a really bad article on a possibly promising treatment...
It's like trying to describe why a particular compiler optimization doesn't work in an edge case to your mother[1].
---
Based upon the actual article @ PLoS Biology: http://www.plosbiology.org/article/info:doi/10.1371/journal....
The take home message is that we at one point had a gene that can block HIV infection. Somewhere along the way it got a mutation that produced a truncated protein. These guys found it, fixed the glitch, and showed that the resulting protein is effective at blocking HIV infection. They then took the next step of figuring out a way to "fix the glitch" in humans without using some complicated gene therapy approach. All in all, this is very impressive.
---
[1] I'm assuming your mother isn't a committer to GCC :)
If you take the article at face value, we just cured HIV. And IMHO articles like that are VERY bad because over the years they have built a consensus of science is wrong and confusing and disappointing into the general public.
Anything that's repeatedly over hyped will end up like that.