Enzymatic DNA synthesis breakthrough

[Bryan Bishop]

Hi all,

I would like to share that our recent work on enzymatic DNA synthesis has been disclosed at the following link:

https://www.freepatentsonline.com/WO2024102228A1.html

We propose a novel method of DNA synthesis which avoids the problems of existing phosphoramidite and enzymatic syntheses: namely, long turnaround times, cost, fragment assembly, and low accuracy.

Our method has the potential to synthesize DNA in a single stretch that is long (>15 kb), accurate (<10^-5 errors per base), and rapid (seconds per synthesized base).

In brief: we immobilize a ribosome and a repetitive mRNA template in a reaction chamber. We can then sequentially flow in custom transfer RNAs, charged with nucleobase amino acids (NAAs). These NAAs are composed of an alpha-amino acid backbone, a linker, and one of the four canonical bases (A/C/G/T) in place of the side chain.

By generating a library of transfer RNAs (tRNAs) which have the *same* anticodon sequence but *differ* in the nucleobase they are charged with, we can effectively program the sequence of the synthesized polypeptide by injecting the desired (A/C/G/T) tRNA-nucleobase conjugate at each step. The ribosome, waiting at the corresponding codon in the mRNA, adds the new base at the next position in the polymer.

This allows for arbitrary conjugation of NAAs and natural amino acids in any desired order, connected via peptide bonds. Note that the sequence of the mRNA template does not predetermine the sequence of the resulting peptide because the mRNA template is a "universal" sequence of repeating codons. The mRNA functions only as a ratcheting mechanism to control the ribosome.

With this setup, we can synthesize a ‘nucleopeptide’ which consists of alternating nucleobase and interstitial (non-nucleobase) amino acids. A nucleopeptide with this structure is recognized as a template by a polymerase, which transcribes it into DNA, yielding the final product.

Animation:

https://www.youtube.com/watch?v=hZIugqTgTBw

Biotechnology has been severely hampered by a lack of fast and affordable synthetic DNA. Given the slow-to-nonexistent decline in the cost and speed of high-quality, profitable synthetic DNA over the past 20 years, and the lack of alternatives beyond the traditional basewise extension of chemical monomers, we believe that this method could potentially transform the field.

If anyone would like to know more, email us and ask for the Nucleostream whitepaper:

Bryan Bishop <br...@nucleostream.com>

Max Berry <m...@nucleostream.com>

Thank you,

Bryan & Max

- Bryan

https://twitter.com/kanzure

[Dan Bolser]

Congrats Bryan! 

Do you have to worry about secondary structure in the mRNA template? 

Cheers, 

--
You received this message because you are subscribed to the Google Groups "enzymaticsynthesis" group.
To unsubscribe from this group and stop receiving emails from it, send an email to enzymaticsynthe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/enzymaticsynthesis/CABaSBawwCAWPYsFSsGq3moFbT%2B3UxMG3gBG0fgammNU5%2B6gDcg%40mail.gmail.com.

[Max Berry]

Not really; avoiding excessive secondary structure is a consideration when designing the template, but since we can pick whatever codons we want (within reason), we should be able to avoid potential massive stem-loop structures with a decent design. And the ribosome is pretty effective at plowing through moderate amounts of secondary structure.

[Marinus]

Hi Bryan and Max,

Congratulations on developing this! I am curious about how this compares to EDS at the moment in terms of cost and accessibility of reagents (e.g. tRNA vs nucleotides, ribosome vs DNA polymerase). I am wanting to start up something in biotech and would love to know more about it if you guys would like to share the white paper!

Kind regards,

Matt

--

[Max Berry]

At the moment? EDS has benefited from about a decade of R&D, and hundreds of millions of dollars in funding to the big three EDS co.s (DNA Script, Ansa and Molecular Assemblies).

EDS isn't using off-the-shelf reagents, their nucleotides are modified - at least a cleavable protecting group (to prevent runaway reactions), and often directly conjugating them to the TdT. Nor is the TdT wild-type; they've had to spend most of their effort just mutating it to be heat-stable. Since they're working in aqueous solutions, the DNA being synthesized is prone to basepairing to itself, as it does in nature. This also blocks TdT from binding to the end of the strand, since TdT is specific for ssDNA.

MA tried to fix this by chemically blocking the nucleobases until the final step so that they couldn't pair, but they seem to have abandoned that. The only alternative is to heat up the reaction enough that the DNA can't basepair. Since TdT is a human enzyme, it denatures much above 40C. Anyway, once they have a mutant TdT that's capable of surviving high temps, they can make it like any other protein. Their nucleotides are still custom synthesis, but any DNA synthesis reaction uses vanishingly small quantities of each molecule at scale.

Now on to our stuff: we can pull ribosomes directly out of cells via ultracentrifugation. The plan is to use close to wild-type ribosomes, since the cell needs them to make its own proteins before we steal them. Luckily even a regular ribosome is perfectly capable of handling these kinds of non-canonical amino acids (NCAAs). We can make some modifications without stressing the producer cells too much either.

Synthesis of the nucleobase amino acids has been around for >20 years. To stick them to the tRNAs, Flexizyme is the natural choice, but Church's lab claims to have an even better method for conjugating NCAAs (albeit pre-publication) so we'll see.

It's hard to compare cost since we're still developing our method, but the vast majority of the cost (money and time) for a company like Twist isn't in per-base synthesis; it's the labor and machinery required to sequence-verify every oligo before it's assembled into a gene. Since we can one-shot an entire plasmid-length construct, we avoid that hassle.

As I mentioned, any DNA synthesis reaction uses tiny quantities of reagents, so the differentiator is in stepwise efficiency and speed. We use the ribosome's natural ratcheting mechanism, extremely high accuracy, and speed. That way we can avoid the multiple addition/wash/deprotection/wash steps of regular synthesis, which each need a high efficiency. In regular synthesis (phosphoramidite or TdT), you want to wait several minutes at each of those steps to give the chemical reaction time to approach completion. We will be more limited by how fast we can push tRNAs to the ribosome.

To view this discussion on the web visit https://groups.google.com/d/msgid/enzymaticsynthesis/CAMmtKC0PS9oyTu03nrSXgGbtZd6M3o2Lu_Z-gn8e-THKnd-T1w%40mail.gmail.com.

[Marinus]

Right - so the enzyme is wild-type. But if you have to express and purify it anyway is there an advantage over a recombinant one?

With continuous flow of the tRNAs, I'm also wondering how you would avoid homopolymers in the product. Don't you need wash steps if the codons are the same?

The codon sequence space is 64 so I'm also wondering if you guys have considered mini-batching synthesis with more diverse mRNA templates.

To view this discussion on the web visit https://groups.google.com/d/msgid/enzymaticsynthesis/CAO_nO-uBS59j%3DLrtWQbQRLuz7%3Dr4ROS1NzzccC05imtWn5KC9w%40mail.gmail.com.

[Max Berry]

By 'enzyme' you mean the ribosome? Assembling a functioning ribosome recombinantly is just barely possible now, and a huge pain in the ass. It is an option if we need to engineer a ribosome that's incapable of translating native proteins for some reason.

Homopolymers are avoided by using different codons, in a repetitive order. For example, a template like [...ABCDABCDABCD...] where each letter is a different codon. If we inject an A-matching tRNA, by the time the ribosome gets to another A codon, any residual A tRNAs have been flushed out while injecting the B, C and D tRNAs. We can increase the number of codons in the template if needed ([...ABCDEFGHABCDEFGH...] etc), to increase the time between A tRNAs being injected, or just add more washing in between. With the simplest [...ABABABABAB...] template we'd probably spend too much time washing to avoid runaways, which would be counterproductive vs just using a more complex template.

There are other considerations in using more codons, since it makes the device more complex, the ribosome more prone to frameshift slipping, and synthesis of the tRNA library more annoying. But we'll find the best compromise.

To view this discussion on the web visit https://groups.google.com/d/msgid/enzymaticsynthesis/CAMmtKC2KjosA2jqpF5DMkmzphPdJgO2spP7gDWHbb5nAjeD24g%40mail.gmail.com.