Why do restriction enzymes act on palindromic sequences?

[joigus]

Here's an idea that, if too out there, I'd wholeheartedly thank you to dismiss as nicely and informatively as possible.

Something that very much drew my attention years ago about restriction enzymes is the fact that they always seem to act on palindromic sequences of DNA, not in the usual language-related sense, but in the sense that, for a bunch of code-bases, the sequences usually (maybe always?) are palindromes of their inverses in the "daughter" thread, like,

AGGCCT

TCCGGA

 

(Sorry if there's a specially reserved codon there, I wasn't particularly careful in the example.)

Now, I don't think that's a coincidence and I'm pretty sure there must be a physical reason for it. My best guess is that there must be a physical action, like a cutting torque at the molecular level (my background is theoretical physics) leading to the controlled chopping of the polymer precisely at that spot. Opposite pulling or pushing forces would lead to that in a way that's very intuitively easy to picture. Does that make any sense at all?

If so, could a similar mechanism work for endonucleases on tRNA in opposite sequences attached for splicing?

I'm not even sure that tRNA splice at palindromes or even if that's been understood in any detail.

Thank you very much in advance.

Edited May 1, 2020 by joigus
mistyped

[Endy0816]

Wiki mentions some types do act on non-palindromic sequences.

https://en.wikipedia.org/wiki/Restriction_enzyme#Types

[joigus]

13 hours ago, Endy0816 said:

Wiki mentions some types do act on non-palindromic sequences.

https://en.wikipedia.org/wiki/Restriction_enzyme#Types

Thank you very much.

[Endy0816]

17 minutes ago, joigus said:

Thank you very much.

Welcome, sorry I couldn't be of more help though. Is an interesting question.

Don't know if the types that do select palindromes, whether they are simply nonspecific in terms of direction or if there a physical reason like you suggest.

[Dagl1]

I may have to brush my knowledge about tRNA splicing, but could you elaborate what you mean by :

20 hours ago, joigus said:

If so, could a similar mechanism work for endonucleases on tRNA in opposite sequences attached for splicing?

I'm not even sure that tRNA splice at palindromes or even if that's been understood in any detail.

I have been trying to find if SEN (TSEN in humans) tRNA splicing endonucleases actually cuts the opposite strand, but so far I have not found that they do: 
It definitely doesn't seem palindromic though. From quickly scanning these few articles it seems that for tRNA splicing there are two catalytic (TSEN2, TSEN34) subunits and two structural (TSEN15, TSEN 34). The last article (sci-hub link) shows this through mutating SEN (non-human version) 2 and 34 (see last picture).
Please note that I have included the first article because it shows the structure of the complex quite nicely (I think), but that article is about specific mutations that cause pontocerebellar hypoplasia.

Hope this is of interest/relevant to you!

Articles (hopefully in order of the pictures):
https://www.nature.com/articles/ng.204? 
http://www.genesilico.pl/rnapathwaysdb/Pathway/step/72/ 
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4090602/
sci-hub.tw/10.1007/s00018-008-7393-y 

[CharonY]

There is also the quite simple aspect that many restriction enzymes bind as dimers, and each half recognizes the sequence on the opposite strand resulting in a palindromic recognition site. There are thermodynamic aspects to it, too, in terms how the enzyme bends the strand but that has a bit more to do with how the enzymes bend the DNA while nicking (I am hazy on details though).

[joigus]

6 hours ago, Dagl1 said:

I may have to brush my knowledge about tRNA splicing, but could you elaborate what you mean by :

On 5/1/2020 at 4:52 PM, joigus said:

If so, could a similar mechanism work for endonucleases on tRNA in opposite sequences attached for splicing?

I'm not even sure that tRNA splice at palindromes or even if that's been understood in any detail.

Sure. Thanks a lot for your interest. I was referring to the possibility of selectively cleaving sequences in a similar way to how restriction enzymes are used to cleave DNA to mass-produce genomic libraries. In this case, though, the target would be RNA. The only example I know of RNA that has complementary (let's say "locally" double-stranded) sub-sequences is tRNA, and from what I remember cleavage of nucleic acids in Nature only happens on double-stranded sequences, e.g. tRNA cleaved by eukaryotes in the splicing process --and as I've just learnt from the references you provided some archaeas[!?]. In other words: would it be possible to mimic endonucleases' job with tRNA, synthesize them, maybe modify them for human purposes?

Sorry I said "opposite" instead of "complementary" and the like.

And thanks a lot for the references.

 

[Dagl1]

9 hours ago, joigus said:

Sure. Thanks a lot for your interest. I was referring to the possibility of selectively cleaving sequences in a similar way to how restriction enzymes are used to cleave DNA to mass-produce genomic libraries. In this case, though, the target would be RNA. The only example I know of RNA that has complementary (let's say "locally" double-stranded) sub-sequences is tRNA, and from what I remember cleavage of nucleic acids in Nature only happens on double-stranded sequences, e.g. tRNA cleaved by eukaryotes in the splicing process --and as I've just learnt from the references you provided some archaeas[!?]. 

 

I think I may be misunderstanding, so apologies if I am missing the mark, but: in humans, each TSEN enzyme/subunit cuts one strand, and they don't seem to be directly complementary of each other. Also, a lot of mRNA is 'double-stranded' or forms hairpins/secondary structures, non coding RNA even more (I think). Important to note that mRNA normally binds to many mRNA-binding proteins, and therefore predicted structures (minimum free energy calculations) will often not actually form inside the cell (or be different from the predictions). The 5'UTR (Untranslated region) and 3'UTR of mRNA contains many structures as well. About 190 human genes (I think) contain an internal ribosomal entry site (IRES) within their 5'UTR, which can modulate ribosomal activity, these generally have the shape of hairpins as well.
A viral enzyme of herpes virus, SOX, cuts at a single point, but recognises specific sequences in hairpins, so both structure and sequence is important.

9 hours ago, joigus said:

In other words: would it be possible to mimic endonucleases' job with tRNA, synthesize them, maybe modify them for human purposes?

I don't really understand what you mean?

 

https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1000307

 https://journals.plos.org/ploscompbiol/article/figure?id=10.1371/journal.pcbi.1003517.g004 

https://en.wikipedia.org/wiki/Insulin-like_growth_factor_II_IRES (couldn't find an article with eukaryotic IRES that shows it nicely)

 


 

 

 

Edited May 3, 2020 by Dagl1

[joigus]

4 hours ago, Dagl1 said:

I think I may be misunderstanding, so apologies if I am missing the mark, but:

 

4 hours ago, Dagl1 said:

I don't really understand what you mean?

Please, don't apologize. I wasn't making much sense biologically, and now I realize that I'm confusing really basic stuff in RNA maturation process. I really must go over my notes and books before I make more of a mess and take much more of your time.

What does make sense in what I'm saying, I think, is that different reflection symmetries in the chain to be cut must play an important role in the cutting process, because the enzyme's job is to cut a line of covalent bonds, so a very high energy barrier must be overcome. As CharonY says:

20 hours ago, CharonY said:

in terms how the enzyme bends the strand but that has a bit more to do with how the enzymes bend the DNA while nicking

(my emphasis)

So the general idea that I get from these comments is something like: Oh, so the enzymes that do the job of cutting must either twist the strand, bend it, then act like scissors... Now, these physical actions all require some handling of the object with different configurations of strong opposite pairs of force. This is very different from what is required in, e.g., helicases, which only need to overcome hydrogen bonds to untangle the double strand for reading and don't require any kind of symmetric grasping of the molecule that I can think of.

In one of the images that you so kindly (but so much overestimating my understanding of biology) have attached, I've found something very interesting. I can see a single-stranded sequence under the tag hsa-miR-25-5p that reads,

TCCGCCT

Now, I don't know what the significance of that sequence is, but it is a palindrome in a different sense than palindromes in double-stranded DNA are. This is a palindrome of itself in the sense of ordinary-language palindromes. I.e., if you read it in a 5'-to-3' direction, instead of 3'-to-5', it doesn't change. The palindromes selected for cutting in double-stranded DNA, for example, are different. They are palindromes only if you apply a sequence of two "inversions". Take, for example, my blah-blah example,

AGGCCT

First invert (read 5' to 3' instead of 3' to 5'): AGGCCT --> TCCGGA

Then complementary invert (A-->T, C-->G, G-->C, T-->A): TCCGGA --> AGGCCT

And you're back where you started. The fact that different kinds of palindromes pop up when cutting, twisting, etc. are involved; I don't think is coincidental.

Free-energy considerations don't interest me so much at this point, important though they are.

Please don't trust me when I say anything strictly biological, as it's well over my head there. And do feel free to drop the conversation at any point if you don't find it useful or revealing or anything.

 

20 hours ago, CharonY said:

restriction enzymes bind as dimers,

Do 'dimers' here --or elsewhere in biology-- refer to primary structure only? Two symmetrically-placed terciary-structure blobs of protein weakly attached to each other wouldn't be a dimer, would they? My ignorance shows, I know.

Thank you very much.

Edited May 3, 2020 by joigus
mistyped

[Dagl1]

I am not sure about the relevance of the the HSA-miR palindrome, but my biochemistry knowledge is also not very up to date regarding how endonucleases cut exactly, maybe Charon can elaborate

[CharonY]

A dimer are complex consisting of two full (sub)-units. In this case we are talking about a homodimer, i.e. two copies of the the protein interacting with each other and the substrate. It is related to the concept of quarternary structure.