Non-Canonical Base Pairs and SHAPE: To Boost or Not to Boost?

  • 4
  • Article
  • Updated 7 years ago
Why do "unboosted" designs seem to fare better than "boosted" designs in lab? I don't know, but I'm concerned that it is an illusion. The reason depends on the mechanism behind "boosting" and how we evaluate our designs.

- Why does placing G residues at "boost points" lower energy?
The pairs permitted in EteRNA aren't the only pairs that can form in nature. The energy-lowering effect of strategically-placed G residues is an attempt to implicitly account for the formation of G-A pairs at the end of a stack. This effect is particularly striking in certain tetraloops, but it appears in internal loops as well. Here's an example from a 3-3 loop:



-How do we judge how "good" our results are?
SHAPE. It's a technique based on the idea that a paired base is held in a position that makes it less reactive toward certain chemicals.

-So what's the problem?
Loop energy can be lowered by pairing between supposedly unpaired bases. The EteRNA GUI doesn't display these pairing, but designs are judged based on a technique that can potentially detect them.

I'm concerned that we are avoiding especially stable loop sequences because they contain non-canonical base pairs that are detected by SHAPE. If non-canonical pairings reduce accessibility to the SHAPE reagent, it could give the appearance that the loop has misfolded.

Non-WC base pairing appears to be reflected in SHAPE data for GNRA tetraloops in the lab "bulged star".


Stardust by deep_thought (above)



It may also explain the reason that the most stable 2-2 loops sequences appear to fail in lab. Starryjess recently posted a thread about this.
Photo of Quasispecies

Quasispecies

  • 100 Posts
  • 9 Reply Likes

Posted 8 years ago

  • 4
Photo of Freywa

Freywa

  • 41 Posts
  • 2 Reply Likes
Heh. It turns out that G and A can pair if sufficient space is given to allow such pairing (such as in internal loops). This would not happen inside a stack, as this paper by Crick shows. The reason such Gs at boost points lower the free energy (at least on the eteRNA GUI) may be due to the fact that this AG pair would be adjacent to another pair, meaning that it would be part of a stack, hence lowering its stability and pushing the loop apart.

Freywa
Photo of Quasispecies

Quasispecies

  • 100 Posts
  • 9 Reply Likes
The unusual base pairs at boost points aren't forcing loops apart - they're making the loops more stack-like, lowering their energy in the process.

You were checking out PDB the other day, right? Check out the structures 1MIS and 1GUC (2-2 loops with GA and GU mismatches, respectively). The mismatched bases are rather stack-ish. The same thing happens in the sarcin loop. There's a very obvious bulge, but several unusual base pairs adjacent to it are part of a helix.
Photo of Quasispecies

Quasispecies

  • 100 Posts
  • 9 Reply Likes
GNRA Tetraloops

Stacking and H-bonding interactions may constrain certain the bases in the loop to specific positions. This would reduce the mobility of a substituent at the 2' position and possibly its accessibility to the SHAPE reagent. The bases that are most paired in the GNRA tetraloop tend to be the least paired in our results. Likewise, the second base in the tetraloop (the "N") is the least paired of the loop bases in both our results and in nature.





GNRNA Pentaloops

This 5-nt loop structure appears to adopt a structure similar to that of the GNRA tetraloop by "flipping" the fourth base out of the loop. The structures available show this loop associated with a protein, however, so it's not entirely clear that the same pairings occur in other situations.

Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 123 Reply Likes
@quasispecies, thanks for a sophisticated and very accurate post, and a really nice hypothesis & backup. [I certainly hadn't seen the cross-comparison of the pentaloops -- very cool]

There are three things that we can do to better handle noncanonical pairs. First is to allow them to form in the game -- but then what scores should we give them? There aren't much experimental data on the energetics of these base pairs (whatever there are are folded into the present rules, actually, e.g., a GNRA pair is more stable than other loops). One option might be to let you use sliders to set the energetics.

Second, it would be great to get more data on the RNAs in these regions, besides the SHAPE chemistry. This is in the pipeline actually (using chemistry based on modification by dimethyl sulfate, etc.). We can in fact do these experiments now with no problem, but they cost time & money; as you may have heard elsewhere we are trying to substantially parallelize the experiments to test far more designs, and it is our dream to then also carry out 3 to 4 other chemical mapping experiments on every player design. These additional data should help clarify whether a low SHAPE score is due to Watson/Crick or these noncanonical base pairs. My lab is actually writing 1-2 papers on this now -- its very much a cutting-edge question.

Third, the ideal scenario might be model the RNAs in full 3D, where some/many of the energetic effects of noncanonical structure can be computed. Creating such predictive 3D modeling algorithms is in fact the core goal of my lab, and the original version of EteRNA was a hack of Foldit that enabled RNA 3D modeling. It was unfortunately hard to make base pairs in it -- but I think we now have quite sophisticated code that will enable interactive modeling of 3D structure feasible and fun. It may take several months to work out the kinks (mainly b/c we're a small team and are also preoccupied with writing papers and setting up the new expt pipeline!), but the developers are very excited about this.

In the meanwhile, I (and, I'm guessing, the other players) find your analyses really compelling and insightful... keep them coming, and I am highly optimistic that we will be able to test your hypotheses comprehensively in the near future.
Photo of hoglahoo

hoglahoo

  • 141 Posts
  • 39 Reply Likes
Are your lab's papers published to the public and if so, where? thanks
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 123 Reply Likes
Yes definitely, all the papers are freely available as PDFs at my lab's website.
Photo of jandersonlee

jandersonlee

  • 551 Posts
  • 125 Reply Likes
A new discussion related to this thread over here.