Using speed of polymeraze to sequence DNA?

[Duda Jarek]

There are considered some approaches to sequence DNA base by base - for example by making it go through nanoscale hole and measure its electric properties using some nanoelectrodes.

Unfortunately even theoretical simulations says that identifying bases this way is already extremely difficult ...

http://pubs.acs.org/doi/abs/10.1021/nl0601076

 

Maybe we could use nature's ways to read/work with DNA?

For example somehow mount polymerase or ribosome and somehow monitor its state...

 

I thought about using speed of process to get information about currently processed base.

For example DNA polymerase to process succeeding base has to get from environment corresponding nucleoside triphosphat - there are only four of them - we can manipulate their concentrations.

If we would choose different concentrations for them, there would be correlations between type of the base and time of its processing - by watching such many processes we could determine the sequence.

 

Is it doable?

What do you think about such 'base by base' sequencing methods?

How to use proteins developed by nature for this process?

[CharonY]

I am not sure what precisely you propose. Both Sanger and pyrosequencing utilize polymerases and identify bases by subsequently adding or leaving the respective base out of the reaction. The detection of these reactions is far more accurate and direct.

[Duda Jarek]

Sanger is completely different - it cuts into short pieces and use electrophoresis.

Pyrosequencing I've just read about, is a bit closer what I'm thinking - it sequentially adds nucleotides and watch if they were used by polymerase.

Steps of such sequence are quite long and so expensive.

 

The idea is not to use such macroscopic time sequences, but rather a natural process which goes many orders of magnitude faster.

For example - somehow mount polymerase on the cantilever of atomic force microscope, so that it can 'watch' its speed of DNA processing.

Now use different concentrations of the 'carriers' of nucleotides - so that the speed of the process depends on the current base.

So there should be correlations between base sequence and forces observed by the microscope - processing given sequence a few times this way, we should be able to fully determine base sequence ... many orders of magnitude faster than using pyrosequencing.

Eventually we could mount ssDNA and optically watch the speed of polymerase (for example by attaching to it something producing light, like luciferase).

Edited December 21, 2009 by Duda Jarek

[CharonY]

First, Sanger does not cut anything. You measure when the polymerase stalls. There are other approaches out there, including using the AFM, but most without good results. Mind you, they use intact DNA. Measuring an active process with AFM is hardly practical. Also which forces would you think would change by the insertion of different bases from the viewpoint of the polymerase? The actual interactions happens with the second strand, does it not? The polymerase mostly senses the back of the nucleotides. Incidentally this is also one of the problems with sequencing with nanopores.

 

In other words, the elements that you propose (speed of polymerase) are hard to measure and almost impossible with the required accuracy.

Monitoring the insertion of nucleotides whether directly or indirectly is much more accurate. You may be interested in a paper in Science a year or so back where they sequenced a viral genome with fluorescence microscopy.

Finally the speed is increased dramatically in the next gen systems due to massive parallelization so getting the required coverage is somewhat less of an issue (though assembly still is).

[Duda Jarek]

Yes - monitoring the insertion of nucleotides is more accurate, but it's difficult to make it fast - usually each base requires some separate cycle of macroscopic time. So for such methods the only way to make it practical is to use massive parallelism as you written ... but it's still slow and expensive ...

 

We should also look for smarter methods analyzing base by base for example while it comes through nanopore/protein/ribosome ...

Such polymerase naturally process large portions of chromosome in minutes/hours - if we could monitor this process, we would get faster and cheaper methods ...

 

Measuring position of polymerase is in fact really difficult ... optical methods are rather not precise enough ... making many snapshots using electron microscope could damage it ...

Alternative approach is attaching polymerase to cantilever of atomic force microscope - it should 'feel' single steps it make ... so when we would use large differences in concentrations (like 1:10:100:1000), times between steps would very probably determine base. Of course we would have to process given strand a few time to get required accuracy.

[CharonY]

Again, what force difference do you expect to measure with the AFM? The strongest force is the attachment of the polymerase to the ssDNA. Then what? Where do you expect to see steps and how do they translate to amplitude changes? More importantly, how do you expect to differentiate between bases? Dilutions do not play a factor as at any given time you only measure the interaction of one molecule.

Also, why do you think that monitoring the direct polymerization would be any slower? It is the rate limiting step after all. You do not need different cycles but different dyes.

[Duda Jarek]

In short, polymerase cycle is - catch and insert the proper nucleoside triphosphate, then GO TO THE NEXT BASE, don't it?

This movement is rather active (uses ATP), so when polymerase would be attached, it would have to pull DNA - AFM observes single atom interactions, so it should also observe forces created while this 'pulling' step.

 

To distinguish between bases we would have just to watch time between succeeding steps - because of large difference in concentrations of 'nucleotide carriers', time required to find complementary base for different nucleotides would be e.g. like 1:10:100:1000.

Of course its accuracy wouldn't be perfect, so we would have to read it a few times.

 

Probably this pulling would be to difficult for polymerase and we would have to help it in controlled way.

For example ssDNA it is working on could just go through a nanopore - cantilever of AFM would be just behind the nanopore and polymerase would work toward the nanopore. Thanks of it polymerase would work on unfolded ssDNA and we can control speed of releasing DNA through nanopore by changing electric field to ensure optimal working conditions.

Such nanopores are already working http://www.physorg.com/news180531065.html

 

 

I agree that dyes are more reliable, but it's difficult to use different ones for different nucleotides, so in pyrosequencing it has to be done in separate machine cycle.

The perfect would be attaching different inactive dyes to nucloside triphosphates so that when polymerase catches it, it by the way breaks this connecting, what would release the dye and somehow activate it ... sounds great, but how to do it??

The only fast and simple way of distinguishing bases without drastically modifying biochemical machinery (or for example attaching electrodes to it) I can think of is modifying its speeds by changing 'nucleotide carriers' concentrations ...

Edited December 23, 2009 by Duda Jarek

[CharonY]

I think you will have to think about how a AFM force profile looks like a bit more. Assumeing you have the polymerase attached to the cantilever. You come close to the DNA, the interaction force changes the amplitude. So far so great. Now at this point you make a leap in assumptions. How do you measure movement on the DNA? How it is tethered and which force change do you assume for each base (hint there are several, depending on how you fix the system)? Even if one is heavily diluted you cannot assume that the diluted one will take precisely 1000th the time of the undiluted one. It would be an estimate with an unacceptable failure rate. Bottom line is that you have problems untangling the involved forces AND you will not be able to distinguish between the bases.

 

Regarding dyes, just check Science for an article last year, they actually sequenced a viral genome just based on fluorescence measurement. Why should it be difficult to have each nucleotide differently labelled? They do it routinely for years even for Sanger sequencing (using one channel capillaries).

 

Also you may want to solve the Michaelis-Menten kinetics (probably assuming three dimensional freedom for this issue) to see how much time difference you would actually expect on average for the given reaction (and think whether there is a way to statistically derive the right base from it at all. Finally, take into account that if the wrong bases are vastly overrepresented, it increases the chance of polymerase errors and stalls. In large reaction it is less of an issue as you will have enough polymerases not doing it, but if you let a reaction continue to go one while monitoring (with whatever means) a single molecule you will run into deep trouble.

Edited December 23, 2009 by CharonY

[Duda Jarek]

Ok, I've finally found the article You are referring to ... yes - it's how it should be done! ... http://www.pacificbiosciences.com/index.php

I didn't realize that there are so large difference in time scale between searching and incorporating the new base ... in this case indeed the dye doesn't have to be activated - it can be always active, but we can use that "it takes several milliseconds to incorporate it" ...

 

It's simpler and better than my idea, but still I wanted to defend it...

As I imagine, simpler polymerases makes that succeeding 'nucleotide carriers' tries if it suits to actually considered base.

If polymerase doesn't analyze the current base, these 'nucleotide carriers' are taken from environment completely at random - if their concentration is 1:10 means that statistically per 11 'draws' there would be tried one of the first type ... I completely agree that it's simplified, but generally we can choose concentrations to select differences in time required for search as we want.

It's stochastic process - this time varies - it's why there would be required a few runs for given strand to increase preciseness.

I agree that these differences in concentrations would increase the number of errors made, but we don't have to use these duplicated strands (for example using nanopore - duplicates are on the other side)

The problem is that time required to incorporate it is much larger, but it should be rather practically constant.

"How do you measure movement on the DNA?"

The forces read by AFM should be rather small and smooth, until the active movement of polymerase to the next base - they should be seen as 'peaks' of force (along DNA strand - the tip should be not on the bottom of cantilever, but rather on it's front)