As some of you may know a means of synthesising DNA that is accessible to the ambitious and well informed amateur has been a long term intellectual pursuit of mine. A kind of theoretical moby dick. Here I'd like to present something like a paper (for a certain value of paper) containing my latest thoughts. If these ideas pan out it would permit synthesis of DNA fragments potentially 100s of nucleotides long with out any pre made oligos (or at least only non gene specific ones), with out invoking potentially toxic and hard to obtain phosphoramidite based chemistry or any thing similar. A method based almost entirely on enzymatic processes. Of course there are a lot of ifs. However I'm not in a position at the moment to take the theory any further into actual experimentation. So I'm opening the idea up to the community in the hope that you can collectively make something of it. It's not really a better alternative for those of you who can already easily order DNA, it's not nearly as efficient as the phosphoramidite method not even in theory. It's a significantly more complex process. However for those of you who can't get synthesised DNA or who want to make your own as a project this may be of interest.
(I've just saw the pictures =D) Particularly on the page 12, scheme 17. Scheme 20 seens good too, because it has regioselectivity. Forgive me if I am missing something...
Well I've thought about this at length actually. Two posable strategies for alternative blocking groups have occurred to me. The first it may, just may, be posable to use a disaccharide synthase to add a sugar based compound to the 3 prime position of a dNTP. You can see scheme 31 in that paper shows a way of using a disaccharide synthase to add at the 5 prime position but there are a lot of disaccharides in nature and lots of different enzymes to make them. You'd have to screen a lot of disaccharide synthases till you found one with week activity for 3 prime addition then do directed evolution to get something that could give you dNTPs blocked by joining another monosaccharide to the ribose. To the best of my knowledge all disaccharides break down into monosaccharides in sufficiently hot water. So you could make your own thermal liable blocked dNTP. It won't be cheeper to do this than buy cleanamp.
The other option might be to use dirty ends. When dna breaks it doesn't always leave clean 3 prime ends. Often you get a lot of 3 prime ends with PO4H2 groups on them. You can't polymerise off this. But there is an enzyme PNKP that nips off the PO4H2 group so polymerisation can continue. If you used dirty dNTPs and used PNKP instead of heat to deblock that might be an option. However the only simple way I can think of to get dirty dNTPs is to get a DNA sample, subject it to lots of oxygen, possibly ozone. Use exonucleases to break it down into clean and dirty dNMPs painstakingly purify out the dirty dNMPs turn them into dirty dNTPs using NDP / NMP kinases then separate them by base. This will also not be cheeper or easier than buying cleanamp.
So the question in my mind is how hard would it be to make a dirty dNTP chemically with out an enzyme. Is there a reagent I could add to dNTPs and phosphoric acid that would attach PO4H2 to the 3 prime location with out the base or 5 prime end needing protecting groups. There are 4 OH groups hanging off the phosphors at the 5 prime end so can we expect some PO4H2 groups added here? What if I turned the phosphoric acid into a free radical before adding it using electrical discharge? All things I've considered. Basically I'm a sucky chemist and it takes a very talented chemist to do serious organic synthesis only using reagents you can purify out of things you can buy at your local supermarket or buy from amazon.
And again none of this is going to be cheeper or easier than cleanamp. But as I said a cheep solution is unrealistic this method is aimed at those who a) want a learning experience or b) can't get industry made DNA sequences.
So your consern is that if you needed a custom enzyme people wouldn't be bothered to do it? Perhaps. But you'd only need to do that if you couldn't buy cleanamp in which case you might not be able to buy lots of enzymes like tdt etc.
I wanted to put some thought into how people could make a dna synthesis technique work with minamal cooperation from the existing biotech industry. I don't know how the Americans feel but here in the uk the goverment and the established biotech industry doesn't seem to see much potential in the diybio movement but does see the potential for harm. The diybio movement has been compared to the coding movement that's gave birth to the raspberry pi. If so DNA would be the code and dna synthasis the compiler. Imagin if every raspberry pi came with no compiler. Instead you have to send your source code off to a company that might compile it, if they approve of it, for a charge. That situation would be unthinkable but it's more or less where we are with diybio.
So as I said I wanted to look into how you might make dna assuming virtually no cooperation from industry. But if you can order off the shelf enzymes I don't see why you wouldn't. Unless you were making enough for economies of scale to apply., I've been thinking about it and making it using bogstandard DNAse II and endonucliases might even work better.
On Jul 16, 2016 12:39 PM, "CodeWarrior" <code.w...@gmail.com> wrote:
>
> Ah that's workable for supper short >sequences but the yield will be low and >the needed input masive. Imagine you >tune your TdT reaction perfectly and get >a 30% yield of single nucliotide aditions
30% sounds far from perfect... wouldn't perfect be 100%? Also why do you assume 30% when phosphoramidite chemistry is close to 99% already, and we know enzymes work and don't have similar error issues, less stringent chemistry (enzymes are all water based, phosphoramidite synthesis hates water).
> (which I expect is optimistic). Then let's >argue your seperation procedure is >100% lossless (highly unlikely). After 8 >rounds you'll have less than 0.007% of >the oligos you started with. You've also >had to do 8 seperation a probably 8 >consecutive PAGE procedures.
Oh, i am talking about nanofluidic channels, where you don't need gel or anything more than maybe counter-ion buffers.
>It's just not efficient to add 8 nucliotides >to a sequence. After that you need to >jam a primer on the end and do some >PCR or you'll be working with a sample >diluted out of existence.
Meh, just shove the at-least single molecule into an e.coli with electroporation... 'low' yield 'problem' solved. Seems a boon to me to be able to require at minimum a single molecule of output. Also, I am talking lab-on-a-chip tech, not garage-scale reaction apparatus.
sadly I’m a bit busy but I’ll try to work through the maths quickly.
so we have 4 populations start oligos f0, oligos with one nucleotide additions f1 and oligos larger than that fm we also have a finite supply of nucleotides n. our system of ODEs is
dn/dt=-a n f0-a n f1-a n fm
df0/dt=-a n f0
df1/dt=a n f0-a n f1
dfm/dt=a n f1
the terms in turn are, a n f0 reactions where a nucleotide is consumed changing a start oligo to an oligo with one nucleotide, a n f1 reactions where the desired one nucleotide addition oligos are turned into longer many nucleotide added oligos. a n fm reactions where many nucleotide oligos grow still larger. Notice all reactions are equally likely as TdT doesn’t care about oligo length once you get past oligos a few bases long. The non linear system can be simplified by noticing that df0/dt+df1/dt+dfm/dt=0 implies f0+f1+fm=I the initial number of start oligos. this means we can solve for n and substitute getting a linear system. if you solve this for initial condition then find the time t-max where f1 reaches a maximum and then find f1(t-max) you get I/e. all the other variables cancel out.
adding an exonuclease would be an interesting alteration. in vivo TdT works against exonuclease activity to produce short oligo additions maybe you can get past the 36% yield limit that way.
This is an awesome project. Kind of puzzled why I don't see much publications on the topic.
It's possible it has been already shifted to patent applications? (Found one using acid decapping meanwhile)
I saw pretty much a lot of ideas thrown out there, is there anyone that started testing? Ideas are running wild here, but about the basic experiments to validate the procedure?
Microfluidics seems the most straightforward approach, especially if you look for solid state synthesis. I'm thinking to give a try to the theory with the digital microfluidic device we've built at digi.bio. Using magnetic beads fuctionalized to oligonucleotides would allow programmable exposure of elongating chain to different reagents, concentrations and incubation times.
Something like elongate>wash>exonuclease>wash>decap>wash>elongate
If I make enough devices I can test probably 100x different reaction protocols and map the search space for optimal conditions.
@codewarrior I was discussing your proposal at the diybio meetup in amsterdam, I was pretty surprised to see so much involvement. We thought starting a community project there.
Best,
Fede
The TdT thus bound is still only non covalently bound to the DNA though. Even if an extreem PH solvent capable of removing it can't be found a sufficiently hot solvent will denature the protien alowing it to unbind but leaving the DNA intact.
One could invision a sequence in which one fist adds a modified TdT with a dNTP mix causing the majority of DNA strands to elongate. The solid suport would then be washed removing free TdT and dNTP. Then exposed to a solvent to denature / unbind TdT blocking strand elangation in preperation for further additions.
Does that sound viable? The question in my mind is how dificult might it be to protien engenear TdT incapable of progressing along or unbinding from a DNA strand end? I've no facilities to produce or test TdT mutants in a lab at the moment but given time I may be able to investigate the matter in silico.
The TdT thus bound is still only non covalently bound to the DNA though. Even if an extreem PH solvent capable of removing it can't be found a sufficiently hot solvent will denature the protien alowing it to unbind but leaving the DNA intact.