Nanopore sequenceing and DNA synthesis
If you have ever tried to create a gene construct or make specific mutations to genes via cloning and PCR methods you would appreciate the potential time and sanity saving benefits of DNA synthesis.
As the ambitions of synthetic biology increase the limitations of the aforementioned methods become glaringly apparent. This is true even with standardization attempts, such as biobrick alpha (www.partsregistry.org) which simplifies the process greatly. The problem with BBa (biobrick alpha) is that you have to standardize the part beforehand. For some genes doing this via PCR is rather difficult and time consuming. Then you still have to clone afterward.
I have been strugling to standardize a 1.8 kb gene for a while now. The PCR primers are designed to amplify the gene, add the BBa prefix and suffix, add a ribosome binding site, and make a synonymous mutation to get rid of an EcoRI site. Problem is… it still hasn’t worked. It’s been close, but close is still wrong. I got a quote to synthesize the gene at $1500 it was above our budget. The cost of synthesizing goes up dramatically above lengths of 1.0 kb.
Today’s DNA synthesis companies still work with phosphoramidite chemistry. Double stranded DNA is synthsized in the same way as oligonucleotides, but with a few costly steps afterward. To drive the cost down and make longer sequences cheaply obtainable a whole new approach is needed. I recently had a discussion with some colleagues (Andre, and Ricardo) about in vivo DNA synthesis, that would be a simple as shining different wavelengths of light at cells and having them assemble a molecule for you. This could involve an enzyme similar to a telomerase that would have an RNA function as a template. The input would align the nucleotide of choice with the active site of the enzyme, and it would function as a sort of DNA typewriter. I suppose that would be the holy grail, but there are going to have to be some intermediate steps toward that goal. I say an enzyme based in vitro approach could bridge that gap.
Oxford Nanopore Technologies alpha-hemolysin nanopore (ribbon diagram) with covalently attached cyclodextrin (teal) transiently binds a base (red) traversing the pore.
I was thinking on the drive home yesterday and I remembered having read about nanopore sequencing. Nanopore sequencing is quite amazing, and works by using a hemolysin nanopore (essentialy a protein with a nice beta-barrel) as a scaffold to support a exonuclease and a cyclodextrin molecule. The exonuclease nicely deposits nucleotides into the pore where they transiently bind the cyclodextrin molecule. When the nucleotides bind the cyclodextrin they then block an electrical current carried by the cylodextrin. The degree to which they do this depends on which nucleotide is bound. That allows the sequence to be read.
Here is when I begin to speculate. What if the same process was doable in reverse. If the cyclodextrin would bind and release a specific deoxynucleotidetriphosphate (dNTP) depending on the amplitude of an electrical current it could work. The exonculease could be replaced with a DNA polymerase something like the DNA typewriter enzyme in the last paragraph. The electrical/magnetic field produced by the electrical current could be used to move the RNA template around, and the pore assembly would feed the polymerase the right dNTP. The whole thing could be designed to exclude random dNTPs from getting into the polymerase active site. With two dNTP specific steps involved in the incorporation of each nucleotide it could be a robust system. The reaction could also be carried out in a pyrosequencing mixture such that when a nucleotide was incorporated there would be a flash of light. This would give the computer controlling the process a way of monitoring the quality of the reaction. This would work in theory if the reaction was working with a single copy of the assembly, but with multiple copies it would be impossible to tell if one of the copies has missed the incorporation of a nucleotide. You would want multiple copies to get high yields of synthetic DNA. Well I can’t think of a way to remedy that problem right now. This is still some crazy speculation anyway.
All thoughts on this wild speculation are welcome.
References:
Cheap Third Generation Sequencing, Nature Methods 6, 244 (2009)
doi:10.1038/nmeth0409-244a
Dos Margaritas Por Favor: Science, Music, Photography, and Other Rants 
The pyrosequencing-as-you-go idea is genius! It solves a number of problems. When you detect that an synthesiser has created an in correct sequence, you can simply stop synthesising.
Creating large amounts of synthetic DNA is not necessarily that important. At worst case, you can always do it by high-fidelity PCR. It might also be possible to do this inside a phage, in which case, 1 copy is probably enough.
There is one addition I would make to your pipeline: the addition of another DNA polymerase. The primary DNA polymerase (the anti-exonuclease), will probably have to create single-stranded DNA. If it can be made to construct it in a 3′->5′ direction, then normal DNA polymerase can construct the complementary strand in it’s natural 5′->3′ direction. Moreover, it means that addition of each nucleotide will cause a pyro-flash of the desired nucleotide, and, after a short delay, the complement. This could be used for a more thorough verification.
I was thinking the same thing about the “anti-exonuclease” and the 3`-5` synthesis, at a stoplight. If I just drove all the time I might have better ideas. Anyway, then I thought that even though that idea is cleaner it might just be easier to have the required parts for lagging strand synthesis, but that is adding lots of parts. A lot more parts that can cause derailment.
Maybe their is an example of a DNA polymerase domain that does the 3`-5` synthesis in the literature somewhere? That would be cool. I’ll keep my eyes open. Maybe I should look up the details on telomerase again.
The thing with the pyrosequencing as you go is that it will only tell that a nucleotide has been incorporated. If nucleotide presence was controlled (when adding T only T was present etc. etc.) Then we could be sure. If nucleotides were just floating around and the nanopore thing was picking them out of the environment the pyrosequencing only tells us that yes a nucleotide was incorporated and assuming the nanopore performed its function it was a T.
I’m really not sure this nanopore thing would be able to pick out nucleotides, haha but who is? There has to be some way of designing a molecule that will selectively bind different nucleotides depending on an applied current/field.
Now we are talking metasynthetic biology here ;).
very interesting indeed….