Sunday, June 04, 2006

Dodging bullets with animal research   posted by Coffee Mug @ 6/04/2006 01:51:00 AM
Share/Bookmark

My favorite bookstore in Denver is the Tattered Cover. This can be wholly attributed to the fact that they have Nature on the shelf in the science periodicals section. This enabled me to sit down and peruse the actual journal today while killing time before watching the Mavs take the western conference championship (I'm a Nowitzness). Something different happens when I'm looking at the actual magazine for some reason. I hadn't noticed this article (Grimm et al. 2006, pdf) before, but I think it's a nice example given recent negativity about the usefulness of animal research. Plus we get to talk about RNA interference (RNAi), and RNAi is cool.

I've argued that a large part of the promise of whole genome research won't be fulfilled until we can use that information medically. We need to be able to pick which genes and variants we want more or less of in which cell-types and turn them up or down in adult humans (i.e. gene therapy). Gene therapy hasn't really got its act together just yet, and we've had some really tragic consequences. The problem thus far in most cases has been that, while the gene therapy did what it was supposed to do, it also did some other things... like causing leukemia or overblown immune response and death.

We've made quite a bit of progress in determining how to get gene therapy vectors into the right cells without disrupting other important segments of DNA if we wanted to turn up a gene. In parallel, the discovery and characterization of RNA interference has provided a good way to knockdown a gene that is causing a ruckus. I'm planning on explaining the details of the RNA interference pathway in a later post, but essentially RNAi allows scientists to co-opt endogenous regulatory pathways and virus fighting machinery to specifically silence any gene you'd like. If the cell comes across double-stranded RNA (dsRNA, many viruses look like this) it chops it up and uses one half of the RNA as a template, hunting around for any other matching RNA to silence. So if we want to knockdown a gene, you just shoot a vector in there that codes for a dsRNA with complementarity to the gene of interest. The system works best if there is a little loop connecting the two strands which makes the whole construct look like a little hairpin, so they're called short hairpin RNAs (shRNAs).

The authors discovered a huge negative side-effect of a therapy in development in mice that we might not have expected and pinned down its cause, so they can now work on figuring out a ways to circumvent it. I'm not sure what started this line of research. Probably they were trying to test out the efficiency of their adeno-associated viral vector (AAV8) and noticed that their mice kept dropping dead, and it turned into a Nature paper. The first couple of experiments they report are something to this effect. They had mice expressing either luciferase in their liver (a fluorescing molecule) or a human gene called hAAT (it's not really important what it is). They then targeted these genes with specific liver-directed shRNAs. Injecting high doses of almost all of their different AAV8-shRNA constructs knocked down the gene for a little bit, but within about a month the mice keeled over from liver toxicity. In fact, the shRNA didn't even have to have a target to cause toxicity. shRNA vectors aimed at nothing in particular caused the same sorts of problems.

Others have shown that trying to induce RNA interference with certain constructs can lead to negative side-effects due to inflammation. You get a reaction from sort-of the wrong virus-detecting widget in the cell, and all these inflammatory proteins get made and the whole system goes haywire and cells die. Grimm et al. couldn't find any evidence that this was happening in their mice. Nor was there any evidence that the delivered constructs were disrupting the cell-cycle. The first clue to the actual cause of toxicity was the fact that the constructs that were causing major toxicity tended to make a major amount of shRNA. Following this up, the authors found that toxicity was also associated with a decrease in overall microRNA levels. MicroRNAs are short regulatory RNAs that happen to use a lot of the same machinery as the shRNA system does.

Here's the punchline: shRNAs that are too highly expressed clog up the works, so essential miRNAs don't get made. Grimm et al. performed a couple of different experiments showing competition for resources between their shRNAs and endogenous miRNAs. Both shRNAs and miRNAs are transcribed in the nucleus just like your average everyday messenger RNAs, but they have their own special exit from the nucleus called exportin-5. This is one of many nodes where the competition could be occuring, but it turns out to be an important one. When Grimm et al. overexpressed exportin-5 so there was enough to go around, the competition effects dropped off.

Since this step in the pathway pretty much runs exactly the same in mice and humans it serves as a good warning for folks who are developing shRNA-based therapies. More expression in this case is not better. The aim will be to produce only as much shRNA as can get the job done. Grimm seems particularly interested in using this technique against hepatitis. While they haven't figured out exactly what elements differentiate a non-toxic, effective construct from a toxic one, they did report one that works to stably knockdown one of the hepatitis B virus particles specifically in the liver for over 5 months after transfection. The main feature of the effective construct as far as I can tell is that it is shorter than all the rest. I guess maybe its shortness leads to weaker expression for some reason so it doesn't crowd the door out of the nucleus.

It's great they were able to catch this effect so early. Since the recent "elephant man trial" (discussed in the Slate article linked above), animal rights fanatics are trying to argue that discoveries in animal research don't translate to humans, so we should stop being so mean to the poor mousies. Here's an example of a promising new approach to treating a debilitating human disease that is being developed with the help of mice. This potentially lethal side-effect was caught before the final therapeutic construct has even been designed. Not to mention that screening for side-effects probably accounts for the smallest portion of animal experiments. You have to get the basic biology down and find treatments that work against animal disease models before you start worrying about this off-target side-effect business. How hard is it to decide it was worth a few mouse lives to avoid lethal liver toxicity in hepatitis patients?