NAME: Alan Gewirtz
POSITION: Professor, medicine/pathology, University of Pennsylvania Medical School
— MD, State University of New York, Buffalo, 1976
— MA, microbiology, State University of New York, Buffalo, 1976
— AB, marine biology, Colgate University, 1971
This month, a team of researchers from the University of Pennsylvania showed that “higher-order structures formed by RNA and bound proteins,” in addition to messenger RNA structure, play a key role in determining the efficacy of antisense-directed post-transcriptional gene silencing.
“We hypothesize that strategies aimed at removing RNA-binding proteins might significantly improve AON-mediated PTGS in vivo,” the team noted in their study, which appears in the current Proceedings of the National Academy of Sciences.
Last week, RNAi News spoke with Alan Gewirtz, the senior author on the PNAS paper, about the findings.
You began the research detailed in PNAS looking for the effects of messenger RNA structure on antisense and siRNA silencing, but you got some curious results. Could you touch on what you found?
We were interested in how siRNAs and oligonucleotides resolve their structural requirements. I have always thought that the reasons that siRNAs perform better [than antisense oligos] is because they deal with structure better somehow. In a cell-free system, the RISC may or may not help resolve structure; I think there are data to suggest that it does not.
The bottom line is that if you put one, two, three, four siRNAs [into a cell], you’re almost always bound to find one that will work, and that is not the case with an antisense DNA. It just seems to take a lot more effort to find an antisense DNA that will target the RNA you want as well as an siRNA might. I just wondered why that is.
[Some] obvious things to think about [are] the structural requirements. I thought that the siRNAs in the RISC were able to deal with the RNA structure better than DNA, but I wasn’t sure I understood why that was. So we just started to look at the structural requirements for an antisense DNA versus an siRNA targeted to something that is known. Flanking sequence being involved with intramolecular duplex formation — maybe that matters. Or maybe [it is] how many base pairs are free in the actual target region.
I don’t think anybody had ever looked at a comparison like that with a DNA molecule and an siRNA molecule, and that’s how we started this off.
I was presuming that the antisense DNA would be a lot more particular about the structural requirements, meaning a lot more finicky, and that would explain, then, the biological observation that for antisense DNAs, [it’s] just harder to find one that will work compared to an siRNA.
When we began to do the experiments, we found exactly what I was not expecting. The first thing was that the antisense DNAs seemed to be a lot less finicky than the siRNAs with respect to structure. The second thing we found was that the actual cleavage rates were much faster with the antisense DNAs than the siRNAs. You would expect, based on previous literature, that something that is less finicky with respect to structure and cleaved the structure a lot faster would just be a lot better.
But that was not the case?
That [is] not the case when you dump it in a cell. So what’s the problem here? We did a lot of biochemistry to look at the structural requirements: how many bases free does an antisense DNA need versus an siRNA; what happens if you impose structure from the 5’ end as opposed to the 3’ end; what happens when you just march in; what happens if you have a loop or a bulge structure, which is likely the way these things are going to exist in vivo?
[When we found that] the siRNAs were much more particular, and the kinetics of the cleavage were slower, I said, “why don’t [antisense oligos] work [as well as siRNAs] when you put them in a cell?” and [hypothesized that] maybe it’s not just the structure but the proteins that are associated with the structure because RNAs are almost never naked within a live cell. They’re always associated with proteins.
So we set up this known model where we had an RNA with a protein that bound to it [to] see whether that mattered in terms of being able to cleave a target or not. That [protein] was alphaCP, [which bound] to this R7alpha structure.
When RNA was covered, the oligo wouldn’t cut [its target] at all. When we competed it off, it cut just fine. That seemed to support this notion that the RNA is basically being hidden by the proteins, and if there is protein around the area you’re trying to target, it’s going to be difficult to hit that target, at least with the antisense DNA.
We could never really do the comparison with the siRNA because we could never get one to cleave that RNA target. We really tried hard … but after awhile we just gave up.
The take-home message for us was that if you’re targeting a region that is accessible, [then] antisense DNA is as good or better than an siRNA. I don’t think that’s novel; there have been other people saying the same thing. What we are proposing is that if you could develop, let’s say, a chemistry that might have a greater ability to invade a structure, to displace a protein, that might work very well. Or maybe what you need to do is develop some way of actually getting the protein attracted off … [to make a target region] accessible to an antisense DNA in vivo.
Is there anything going on in your lab to follow up on that?
Yeah because … speaking in a therapeutic context, I think that it’s still cheaper, faster, and easier to make a single-stranded molecule than a double-stranded RNA.
In the end, they both have the same problem of delivery. Everybody thinks that siRNAs are going to be so much better in the clinic, and I’m not sure that that’s correct because, first of all, somebody has to figure out a way to get [them] into something other than a liver cell. … Then, once they get in, then it’s not clear whether a good antisense sequence … would be just as good as an siRNA [so that] you wouldn’t even need the siRNA. Maybe what you need is a way to get any kind of nucleic acid molecule that is of interest to you into a cell.
I could imagine putting multiple DNAs within this packaging system and they’ll work just fine. Maybe at the end of the day, it would be a lot cheaper of a way to make something you’d use as a medicine.
When you think about ways to go after these RNA-binding proteins to boost the efficacy of antisense oligos, at least in the work you’re doing, is this something that varies from target tissue to target tissue, or cell? Or do you think there might be a general strategy?
The general strategy, I think, would come with some sort of chemical modification that gives greater strand-invasion possibilities.
I may be using the term strand invasion incorrectly. What I want to be able to do is get in under the proteins. One of the things I’ve learned is that these associations aren’t static. They’re dynamic in a live cell. Whatever proteins are involved with a messenger RNA will depend on where in the stage of transcription one actually is, and then where the RNA is physically located.