This page compiles research notes on enzymatic DNA synthesis, focusing on alternatives to traditional chemical oligonucleotide synthesis. The main approaches explored include direct control of DNA polymerase through various mechanisms (optical control, electronic control, template manipulation, forced misincorporation), and the use of template-independent enzymes such as Terminal Deoxynucleotidyl Transferase (TdT).

Optical control of DNA polymerase represents a proposed alternative approach to traditional phosphoramidite-based chemical oligonucleotide synthesis, aiming to harness the speed and fidelity of natural enzymatic DNA replication while achieving programmable sequence control. The core concept involves engineering DNA polymerases (or related enzymes like terminal deoxynucleotidyl transferase or CCA-adding enzymes) to respond to external light signals that would dictate which nucleotide to incorporate and when to proceed with synthesis. This approach leverages advances in optogenetics and protein engineering to create light-responsive conformational changes in the polymerase that could control nucleotide selection and incorporation timing, potentially enabling template-independent, de novo DNA synthesis without the wash steps, chemical modifications, or fragment assembly limitations of current methods.

See also DNA ticker tape in vivo memory systems.

  1. Basic ideas
  2. Enzymatic DNA synthesis group
  3. Terminal deoxynucleotidyltransferase (TdT)
  4. DARPA polymerase (GO) program
  5. Proposal: Light-dependent RNA polymerase (2011)
  6. More light considerations
  7. Telomerase approach
  8. Other telomerase approach: RATTLESNAE
  9. Exploring template-independent polymerases for automated DNA synthesis
  10. Using progressively larger backbones?
  11. DNA sequencing using electrical conductance measurements of a DNA polymerase
  12. Draft manuscript on enzymatic DNA synthesis
  13. Protein synthesis without DNA or RNA
  14. Ribosome approach
  15. More links
  16. Other notes
  17. more notes
  18. also check
  19. Large components
  20. Catalytic domain matching
  21. even more things

Basic ideas

  • electrical properties of DNA polymerase?
  • enzymatic synthesis of nucleic acids
  • chemoenzymatic synthesis
  • chemical oligonucleotide
  • in vivo DNA synthesis
  • in vivo gene synthesis

1) electronic control of polymerase 2) nucleotide gun made out of a nanotube pointed at the finger domain of some DNA polymerase 3) single-polymerase water droplet & add in a single dNTP at a time 4) physical display of dNTP as template for current base addition (i.e., on a stick) (not a protein template) 5) a protein that can undergo conformational changes that polymerase thinks represents the template strand 6) pull/push a template through DNA polymerase to control which dNTP it should be selecting for 7) protein-template DNA polymerase, where the polymerase itself has a giant protein that enzymatically encodes dNTP information (protein template, like in CCA-adding enzymes) 8) mechanical pressure on polymerase 9) some magical ultrasound method

Things I'd like to avoid in the best possible solution (I'd settle for sucky solutions though):

(1) Any step that involves dissociating from the growing molecule. Takes too long to re-associate.

(2) Mistakes.

(3) Wash steps. Takes too much time.

(4) Parallelism. Just one polymerase/enzyme, thank you very much, no "law of big numbers" stuff going on here.

(5) Pause/step mode. Including a step-inducing state change would be fantastic, but I think we can do continuous incorporation if we have to?

(6) if possible: off-enzyme nucleotide selection. The polymerase should be controlled, not the nucleotides - there's too many of those to reliably, quickly control.

(7) if possible: light. Complicates the setup, but isn't a huge deal in the end.

davidad replies: "I take issue with (7). I think that lasers are likely to be our only hope for controlling a single molecule using electronics. Using a nanowire to flow electric current or mechanical pressure directly into an enzyme is surely more expensive and technically challenging than optical techniques. It's possible that other electromagnetic fields than waves (electrostatics, magnetic moments) could be helpful, but I'd definitely be thinking Maxwell's Equations if I were you, not solid-state physics. I also take issue with (2). If you want to beat conventional synthesis, we're talking about 105 base pairs and up. Do you really think you can get error probability below 10^-5? I think it's probably better to accept that mistakes will be made, and incorporate a strategy for error correction."

nmz787 says:

you didn't mention electronic control of a DNA template, or some substrate (silicon, helical silicon (that might not exist, but I've heard of metallofullerenes being helical with a very similar twist/rotation as DNA http://www.jenlaurltd.com/Products.html)) If you could electronically control a "template", then you might not need anything special other than ddNTPs and polymerase

Now that I think of it though, polymerase probably needs/wants (in wild-type enzymes) to "grip" the template strand... which a solid plane of silicon/etc wouldn't allow.

I definitely think mistakes need to be suppressed to the point of natural in vivo levels... any error handling needs to be automatic and on-the-fly.

I think optical modification of the enzyme state is, like David said, the best way to interface. We can easily buy lots of different colored lasers and they work through vacuums, etc...

Enzymatic DNA synthesis group

https://groups.google.com/group/enzymaticsynthesis

The enzymatic synthesis group was set up in 2012 to discuss and brainstorm different methods and ideas for achieving enzymatic DNA synthesis and the control of enzymes or polymerase for the purposes of the synthesis of new DNA molecules.

https://github.com/kanzure/enzymaticsynthesis

Mechanisms of attack:

(1) template DNA/RNA
(2) selected dNTP
    - electronic control to force a specific dNTP to be selected

Direct control of polymerase

1) template shams
2) forced misincorporation

crazy ideas:

  1. electronic control of polymerase
  2. nucleotide gun made out of a nanotube pointed at the finger domain of dna polymerase
  3. single-polymerase water droplet & add in a single dNTP at a time
  4. physical display of dNTP as template for current base addition (i.e., on a stick)
  5. a protein that can undergo conformational changes that polymerase thinks represents the template strand
  6. pull/push a template through DNA polymerase to control which dNTP it should be selecting for
  7. protein-template DNA polymerase, where the polymerase itself has a giant protein that enzymatically encodes dNTP information
  8. graph grammar rules of valid thumb/fingers/pad/palm domains of polymerases to generate a library of possible polymerase variants
  9. mechanical pressure on polymerase
  10. ultrasound

mRNA polyadenylation (addition of adenosine monophosphates to an mRNA fragment)

"Primase is of key importance in DNA replication because no known DNA polymerases can initiate the synthesis of a DNA strand without an initial RNA or DNA primer (for temporary DNA elongation)."

What is the actual change in DNA polymerase that decides whether a base should be accepted or not? Is primase template-dependent?

replicase - RNA-dependent RNA polymerase

In the transfer RNA (tRNA) CCA-adding enzyme, both the tRNA backbone and the protein contribute to the specificity of the incoming nucleotide (24). Eukaryotic translesion synthesis polymerases thus use a variety of means of DNA polymerization, which include Watson-Crick base-pairing by Pols n (25) and kappa (26), Hoogsteen base-pairing by Pol?? (18, 22), and protein template-directed synthesis by Rev1.

primer-template slippage

Terminal deoxynucleotidyltransferase (TdT)

EC2.7.7.31 -- terminal deoxynucleotidyltransferase

TDT

http://en.wikipedia.org/wiki/Terminal_deoxynucleotidyl_transferase

http://www.rcsb.org/pdb/explore/explore.do?pdbId=1jms

http://www.rcsb.org/pdb/explore/explore.do?pdbId=1coe

author: Basu (again)

(02:43:38 PM) Cathal Garvey: Here, this enzyme: Terminal Transferase and easily removed blocking groups on NTP washes = Easy homebrew DNA synthesis reaction? http://en.wikipedia.org/wiki/Terminal_deoxynucleotidyl_transferase
(02:44:13 PM) Cathal Garvey: In other words, not so dissimilar to the existing method, but without requiring the chemistry of actually joining nucleotides with such care. In fact, you could probably do something like this:
(02:44:21 PM) Cathal Garvey: Oligo annealed to solid substrate
(02:44:29 PM) Cathal Garvey: Ends not phosphorylated at beginning of each step.
(02:45:09 PM) Cathal Garvey: Add a kinase to prime DNA, add NTP + TDT
(02:45:32 PM) Cathal Garvey: Interesting stuff!
(02:45:44 PM) Cathal Garvey: I love all the little methods devised by nature for this function
(02:45:49 PM) Cathal Garvey: By the way, my error there:
(02:46:06 PM) Cathal Garvey: the method I suggest is based on the assumption that the enzyme might bind the oligo end and await an NTP,
(02:46:16 PM) Cathal Garvey: and that you can use NDPs rather than NTPs
(02:46:35 PM) Cathal Garvey: so that it won't polymerise further than a single nucleotide without a step involving kinase
(02:47:06 PM) Cathal Garvey: So you can use two enzymes and NDPs in washes, if that prior assumption regarding the enzyme holds.
(02:47:15 PM) Cathal Garvey: Which is, of course, a huge achilles heel.

DARPA polymerase (GO) program

DARPA's Generative Optogenetics (GO) program aims to engineer a novel protein complex termed the nucleic acid compiler (NAC) that enables template-independent, light-programmable de novo synthesis of DNA or RNA sequences within living cells using endogenous nucleotide monomers, thereby achieving massless, photonic transfer of genetic information that integrates seamlessly into the central dogma. RO1 mandates development of the NAC to address three core challenges—optogenetic signal transduction into molecular motion, nucleotide-specific polymerization with processive stability, and multi-domain integration for synchronized population-level control—leveraging advances in optogenetics, polymerase structure-function, and generative protein design models, while eschewing bioprospecting or novel optics. Optional RO2 focuses on error mitigation strategies, such as mismatch detection in palindromes or duplexes, toggling modes, or modular add-ons, without exploratory foundational work. The 42-month program proceeds in two phases with progressive milestones targeting 3-6 kb sequences at high fidelity and rates, sequential programming (<1 hour turnaround), supported by commercialization/policy working groups, community workshops, and CUI protections for NAC sequences, promising spatial/temporal precision and matter-free dissemination for biomanufacturing, agriculture, medicine, and space applications, contrasting inefficient extrinsic oligonucleotide assembly.

See darpa-generative-optogenetics-dna-synthesis and https://www.darpa.mil/research/programs/go

https://news.ycombinator.com/item?id=46268153

Proposal: Light-dependent RNA polymerase (2011)

manuscript: http://sphere.chronosempire.org.uk/~HEx/tmp/prop.pdf

discussion: https://groups.google.com/g/enzymaticsynthesis/c/6GZT8zFNOfo

""" I want to construct a light-sensitive RNA polymerase that functions in vivo, where the template for RNA synthesis is conveyed via light pulses from outside the cell. I believe such a construct would have wide-ranging applications if it could be achieved.

It seems clear to me that existing chemical methods for de novo nucleic acid synthesis are incredibly suboptimal when compared with the methods used by living cells. But harnessing the optimized nucleic acid manipulation tools used by living systems is a problem, because the main way that life encodes information in nucleic acids is by copying existing nucleic acids. (Other methods, such as relying on the combination of error-prone copying and natural selection, are useful to life, but less useful as a method for creating speci␜c nucleic acid sequences. See ␜gure 1.)

Thus the chicken-and-egg problem of creating nucleic acids to order without a template to copy has traditionally been solved by chemical methods, as life does not appear to o␛er a useful solution we can borrow. Such methods are slowly improving, but are nonetheless inherently messy and expensive.

RNA polymerases come in two types: template-dependent, where the template is provided by an existing nucleic acid strand, typically dsDNA; and templateindependent, where the template is encoded implicitly in the structure of the enzyme. Examples of the latter include poly(A) polymerases (PAPs) that add a poly(A) tail onto mRNA, and CCA-adding enzymes (CCAEs), that add the "CCA" sequence onto the end of tRNAs. Both sets of enzymes have been extensively studied and crystal structures are known[2]. I am primarily going to consider CCA-adding enzymes as they already have the capability for adding more than one type of base, and are extremely speci␜c in their action. CCAEs are found universally (even in organisms where the terminal CCA motif is explicitly encoded in tRNA genes, such as E. coli), although some organisms the CCA-adding function is split between a CC-adding enzyme and an A-adding enyzme. CCAEs can be divided into two classes, based upon structural motifs, class I enzymes are found in archaea and class II enzymes are found in prokaryotes and eukaryotes[10]. The most-studied class I enzyme is that from Archaeoglobus fulgidus, a thermophile[7]. Class II enzymes have been studied in Thermotoga maritima, Aquifex aeolicus and in Bacillus stearothermophilus, in addition to E. coli and Homo sapiens. Class II enzymes have several well-de␜ned structural motifs, some of which are conserved across other polymerases including PAPs[6]. In particular, there is just a single NTP binding pocket. Previous experiments have succeeded in modifying a class II CCAE to add other bases instead of C and A [4], and in creating a chimeric poly(CCA) polymerase from a class II CCAE and a PAP[1]. Such results indicate that class II CCAEs seem a good starting point for modi␜cation.

Light-dependence. TODO: mostly questions here. Do there exist natural photoenzymes that are not optimized for energy extraction? Do we need cofactors? [5] claims that only two enzymes (protochlorophyllide oxidoreductase and DNA photolyase) are known to require light for catalysis. Photoreceptors seem common however. Opsins[8]? Can we get them out of the membrane? (Yes: [3]) Do we need to? How small can they be made? Can they be stuck to a CCAE?

Light-dependent polymerase. The light-sensitive elements could be part of the enyzme or part of the nucleotides. I assumed the former, but the existence of the latter possibility should not be forgotten. Getting an enzyme to reliably add just a single nucleotide of our choice will be hard. My initial idea involved four wavelengths of light, one per nucleotide, but conceivably a ␜fth wavelength could be employed as a ratchet to drive the reaction forward and prevent multiple additions. (E.g. enzyme has two states, A and B. State A: accept a pulse of light, bind the appropriate nucleotide, go to state B. State B: accept a non-information-carrying "driver" pulse, add nucleotide to RNA strand, go to state A.) For this to be useful as a practical means of RNA synthesis, yield will be critical. Even with 99% e␞ciency per step, fewer than half of the resulting RNA molecules will be correct and complete after 70 steps, and a typical small mRNA of 500nt would have a yield of only 0.6%, while the remaining incorrect 99.4% would remain in the cell to cause problems. Initiation? Maybe a small RNA could be expressed in a conventional fashion, and the polymerase would bind speci␜cally to it. (CCAEs are extremely picky in what RNA they will bind to, for example ␕ can we change the binding to some other speci␜c RNA sequence?) Termination? Maybe existing RNA hairpin loop mechanisms could be employed to dissociate the growing RNA strand from the polymerase. But: hairpins disrupt DNA:RNA binding? Might not work.

Directed evolution. Although rational design of proteins has produced some successes, it is not clear to me that it is at the level of sophistication required for such a major challenge in protein engineering. Therefore I propose using directed evolution to circumvent the time-consuming endeavour of actual understanding in lieu of a more pragmatic ␐whatever works␑ approach. TODO: what can we usefully select for? Need a plan of action. """

More light considerations

https://groups.google.com/g/enzymaticsynthesis/c/eVnfgbNCnto

I think we can divide the problems that need solving into two main problems that may or may not need solving independently:

  1. Exactly controlling when the polymerase adds precisely one nucleotide.

  2. Controlling which nucleotide is added at each step.

If we can solve problem one, we can probably anchor the polymerases and simply wash in the next nucleotide before allowing the polymerase to continue. This might be a slow process, and a crude solution to problem two, but it should work.

It doesn't seem entirely unreasonable that we should be able to create a polymerase that can be activated and deactivated by external factors, yet even if we can accomplish that, it will probably not be possible to reliably active the polymerase for exactly long enough to add precisely one nucleotide, given the stochasticity of polymerase activity. We need a polymerase that gets stuck in a wait-state after adding a single nucleotide, where it can only proceed when given a signal. Even then, we may get into a situation where the minimum reliable time for the activation signal to be 100% effective will sometimes allow a polymerase to activate twice and add two nucleotides. We may need two different signals to control the polymerase, or figure out some other clever way of avoiding this scenario. If we're lucky maybe it won't be a problem at all, but we should do some research on the variability of time taken for photo- isomerization of enzyme to predict is this will be an issue we need to plan for.

The stochasticity of polymerases has another nasty side-effect: We will have to wait long enough between adding nucleotides that the slowest polymerase has finished adding the current nucleotide before proceeding to the next step. This could mean that controlled enzymatic synthesis will always be significantly slower than synthesis by natural polymerases.

In order to get something like fast whole-genome synthesis, we will also need to solve problem two in some way that is fast and reliable. Modifying the nucleotides themselves to be deactivated or activated by pulses of light would be an interesting way of accomplishing this, if such a thing is possible in a reliable way.

In single-molecule real-time sequencing, Pacific Biosciences uses nucleotides with a fluorescent molecules attached to the phosphate chain:

https://en.wikipedia.org/wiki/Single_molecule_real_time_sequencing

Their technique is sequencing by synthesis, and the attached fluorescent molecules are automatically decoupled from the nucleotide as the polymerase attaches it to the DNA, apparently without affecting synthesis.

So we know that it's possible to have something linked to the nucleotide, and the question becomes if it's possible to attach a photo-isomerizable protein that will block the nucleotide from binding to the polymerase in one isomer and allow it in another.

Maybe your protein+nucleotide complex could be reversed. Instead of using light to dissociate the complex, you could have light cause a change in the enzyme that allows it to hook up to a hacked polymerase, then passing on the nucleotide cargo to the polymerase, then dissociating. I can't think of an enzyme that does this sort of physical handoff of cargo, but it's probably no more crazy than all this other crap we're smoking.

One class of enzymes that does this hand-off operation is the elongation factor class.

"During protein synthesis, tRNAs are delivered to the ribosome by proteins called elongation factors (EF-Tu in bacteria, eEF-1 in eukaryotes), which aid in decoding the mRNA codon sequence. Once delivered, a tRNA already bound to the ribosome transfers the growing polypeptide chain from its 3’ end to ...."

"This complex transiently enters the ribosome, with the tRNA anticodon domain associating with the mRNA codon in the ribosomal A site. If the codon-anticodon pairing is correct, EF-Tu hydrolyzes guanosine triphosphate (GTP) into guanosine diphosphate (GDP) and inorganic phosphate, and changes in conformation to dissociate from the tRNA molecule."

https://en.wikipedia.org/wiki/EF-Tu

https://en.wikipedia.org/wiki/EEF-1

So the requirements:

(1) four engineered enzymes that bind to a specific NTP whenever they don't have cargo

(2) in their default state, these carrier enzymes do not bind to polymerase

(3) when stimulated with light, a specific carrier enzyme (w/ cargo) changes shape to bind to polymerase. Non-cargo carrying enzymes need to be excluded with some additional conformational change? i.e. they should only bind to polymerase when they have an NTP and a signal has been given to change it into the "bind to polymerase" state.

(4) polymerase does not bind to free NTPs (no "nucleotide pocket" at least as they currently exist), but rather only binds to this carrier enzyme

(5) the carrier enzyme binds to and plugs polymerase, dissociates from the NTP releasing the NTP inside a pocket on polymerase

(6) polymerase incorporates the NTP, waits for a step signal; if we do some fluorophore stuff we could theoretically detect incorporation

(7) at this point, we switch all the carrier enzymes to "do not bind to polymerase". Maybe this step dissociates the carrier enzyme from the polymerase, or maybe it happens in the next step.

(8) when there's a step signal, polymerase moves (and maybe dissociates the carrier enzyme during this step?)

(9) switch the next carrier enzyme into the "bind to polymerase" state.

(10) our previously dissociated carrier enzymes go and pick up some free nucleotides from solution somewhere

Conformational states of this carrier enzyme:

(1) default - unable to bind to polymerase, can bind to one type of nucleotide (may or may not have nucleotide)

(2) bindable-inactive - unable to bind to polymerase, has nucleotide

Note: "change to a state where you can bind to polymerase" signal + no nucleotide should /not/ allow it to bind to polymerase.

(3) bindable-active - able to bind to polymerase, has nucleotide ("able to bind to polymerase" caused by light/laser)

(4) polymerase-bound -- bound to polymerase, released/releasing its nucleotide

After #4, it should stay bound to polymerase until a signal switches it back to the "default" state. So there needs to be a way to switch it "on" (per type) and a way to switch all "off" regardless of whether or not the enzyme is already "off" (unless we want 8 wavelengths to deal with?).

So the "bind to polymerase" state should only occur when (A) it already has bound to a nucleotide and (B) the bind-to-polymerase signal has been given. When the "bind to polymerase" signal is given and it has not found a free nucleotide, the carrier enzyme should be unable to bind to polymerase. If it's easy to do, then in this scenario (bind signal + no nucleotide) making it unable to bind to polymerase until it finds itself a nucleotide is okay. I don't see any reason to force the sequence of steps to be locked in order. But the real determinant of this is which way the protein design is easiest, of course.

It is not immediately obvious to me how EF-Tu avoids binding to a ribosome when it has no tRNA. But there seems to be lots of good studies on this elongation factor :-).

Telomerase approach

Juul says: "Another interesting enzyme is Telomerase, as it uses a built-in RNA template to add repeats of six nucleotides to the 3' end of DNA. Not template-independent, but the template is part of the enzyme complex."

Cathal adds:

Importantly, the template is hackable. I have a bunch of papers zipped up somewhere detailing research into this, where they messed around with the templating region in telomerase template RNA to see how it worked. Turns out that, once you account for secondary structures at the boundary of the template (which define the "start" and "finish" of the template region, in effect), everything else is exactly as you'd expect. If the template can bind loose 3' DNA, the rest of the template is written. Assuming there's no minimum length to the template, you could have a set of telomerases with different minimal templates, maybe 3n long; 2n for reading the prior two nucleotides, and 1n for templating the next addition. For synthesis of DNA with this system, you'd therefore need 43 enzymes; 64. Think we could work with a 64-ink printer? :)

Hehe. The binding may be too weak with so few nucleotides. One way to fix that, at least to some degree, could be to use Locked Nucleic Acid (unfortunately patented) instead of the normal RNA template.

Other telomerase approach: RATTLESNAE

RATTLESNAE: Read Artificial Telomerase Templates for Low Entropy Synthesis of Nucleic Acids by Enzyme

https://groups.google.com/g/enzymaticsynthesis/c/haZ31200JNM

RATTLESNAE exploits the demonstrated tolerance of human telomerase for TERC template mutations (validated by tumor-derived variants encoding alternative repeats like TTGGGG and TTAGTG) to implement optogenetically-controlled enzymatic DNA synthesis encoding 2 bits per extension cycle. The system employs four engineered TERC variants differing at position 6 of the 11-mer template (5'-CCAACXCCAAC-3', where X ∈ {C,A,G,U}), selected via combinatorial red/green photoswitching, to direct TERT-mediated incorporation of corresponding hexameric repeats (TTGGGG, TTGGTG, TTGGCG, TTGGAG) with the variable base encoding the binary state. Template boundary constraints are maintained by preserving helix P1b structure rather than sequence, while positions 1-5 retain wobble-compatible bases to ensure proper template realignment and processivity. A photocleavable protecting group ratchet prevents runaway processive synthesis: after each extension cycle, a complementary oligonucleotide bearing a protecting group hybridizes to the nascent repeat, blocking further elongation until blue-light-activated deprotectase removes the block, permitting controlled single-repeat advancement and deterministic sequence programming.

Exploring template-independent polymerases for automated DNA synthesis

https://groups.google.com/g/enzymaticsynthesis/c/LAg2CUQAC4M

Enzyme-based DNA synthesis could lower the cost and complexity of assembling DNA code by several orders of magnitude. Enzymes that assemble DNA are called polymerases. They are present in all living cells and work to copy genetic material at a rate of about 1000 bases per second. Typically, one strand of DNA is used as a “template” for the polymerase to produce the complimentary matching strand, together forming the familiar DNA double helix.

In 1997, it was reported that an archaebacterial polymerase could generate long DNA repeats (50,000 bp or more) spontaneously, even in the absence of template DNA[3]. The repeat sequences are in effect "programmed" directly into the polymerase protein structure. This was an unexpected and counter-intuitive discovery, found when control experiments lacking template DNA still produced bands on gels. Even now many biologists remain unaware that polymerases have this capability.

It may be possible to harness this de novo synthetic ability and communicate to the polymerase the DNA sequences to generate. This approach would closely mimic reaction conditions in a living cell and use natural, unmodified reagents. For example, it may be possible to dictate the polymerase's synthetic activity using precise, shaped electrical perturbations. Here the goal would be to transmit information about which DNA base(s) to add as well as start-stop instructions. If successful, the enzyme would effectively become a molecular DNA "typewriter".

https://patents.google.com/patent/US12168210B2/en

Analysis and control of untemplated DNA polymerase activity for guided synthesis of kilobase-scale DNA sequences (2024) -- "DNA polymerases are complex molecular machines able to replicate genetic material using a template-driven process. While the copying function of these enzymes is well established, their ability to perform untemplated DNA synthesis is less well characterized. Here, we explore the ability of DNA polymerases to synthesize DNA fragments in the absence of template. We use long-read nanopore sequencing and real-time PCR to observe the synthesis of pools of DNA products derived from a diverse set of natural and engineered DNA polymerases across varying temperatures and buffer compositions. We detail the features of the DNA fragments generated, enrichment of select sequence motifs, and demonstrate that the sequence composition of the synthesized DNA may be altered by modifying environmental conditions. This work provides an extensive data set to better discern the process of untemplated DNA polymerase activity and may support its potential repurposing as a technology for the guided synthesis of DNA sequences on the kilobase-scale and beyond."

Using progressively larger backbones?

https://groups.google.com/g/enzymaticsynthesis/c/E94B-CKLa78

Some of the DNA alternatives like XNA and GNA offer some interesting options. Perhaps it would be possible to find some DNA-equivalent system with significantly larger components. Unfortunately there are probably some scaling limits to the size of a polymerase enzyme. Larger "nucleotides" would be useful because they would be easier to physically manipulate and synthesis could be much easier by physically moving larger building blocks around. We'd need something on the order of 0.1 microns or something else we can physically grab.

http://gnusha.org/logs/2015-07-07.log

DNA sequencing using electrical conductance measurements of a DNA polymerase

RETRACTED

http://www.nature.com/nnano/journal/v8/n6/full/nnano.2013.71.html

https://ir.nctu.edu.tw/bitstream/11536/22351/1/000319979400018.pdf

""" The development of personalized medicine—in which medical treatment is customized to an individual on the basis of genetic information—requires techniques that can sequence DNA quickly and cheaply. Single-molecule sequencing technologies, such as nanopores, can potentially be used to sequence long strands of DNA without labels or amplification, but a viable technique has yet to be established. Here, we show that single DNA molecules can be sequenced by monitoring the electrical conductance of a phi29 DNA polymerase as it incorporates unlabelled nucleotides into a template strand of DNA. The conductance of the polymerase is measured by attaching it to a protein transistor that consists of an antibody molecule (immunoglobulin G) bound to two gold nanoparticles, which are in turn connected to source and drain electrodes. The electrical conductance of the DNA polymerase exhibits well-separated plateaux that are ~3 pA in height. Each plateau corresponds to an individual base and is formed at a rate of ~22 nucleotides per second. Additional spikes appear on top of the plateaux and can be used to discriminate between the four different nucleotides. We also show that the sequencing platform works with a variety of DNA polymerases and can sequence difficult templates such as homopolymers. """

Draft manuscript on enzymatic DNA synthesis

2017 draft: https://diyhpl.us/~bryan/papers2/bio/Electronic%20manipulation%20of%20polymerase%20for%20DNA%20synthesis.2017-12-04.pdf

Electronic manipulation of polymerase for DNA synthesis: "In this review, we explore the concept of modifying existing DNA polymerase enzymes such that their DNA synthesis activity could be programmed and manipulated in real-time by external control signals including optical stimulation and physical electronic interface. We review a number of concepts and techniques which could be useful for carrying out such a project."

In vitro: The assumption for the most of this paper is that the artificial DNA polymerase will be for in vitro usage. This is expected to be an environment such as a test tube, or some MEMS/CMOS scaffold that interfaces with the polymerase. It is assumed that the polymerase will be meant to function in an aqueous environment. In vitro, polymerase could be attached to a flat surface (such as a CMOS semiconductor microchip) or to the surface of beads, or have no tethering at all and be free floating in droplets, emulsions, or other liquid environments.

In vivo: One alternative is that a polymerase could be made that operates in vivo. There can also be a mix of methods where the polymerase is evolved partially outside of cells in vitro, and then for a second round of selection it is evolved inside of ecoli (or something similar) in cell culture. One proposal is to use a toxin that can be defeated by giving a trivial sequence to the ecoli. Then, use optical stimulation of the ecoli to send signals to the polymerase to construct a certain antitoxin molecule or protein.

In vivo methodology has some other interesting benefits. Usually the purpose of DNA synthesis is to construct DNA and then express the DNA inside of another organism. If we are already going to be handwaving about the construction of a magical polymerase that conforms to our high expectations, then why not also handwave about some magic to make it also work in a cell by purely optical phenomena? This would allow for the construction of DNA at the destination where the DNA is going to be used. This has many advantages such as a reduced reagent cost, a self-renewing system that only requires agarose or other cell culture medium, and can completely skip sequencing by checking for cell survivability when exposed to antimicrobial selection. An interesting example of electronic control over cellular activities is the field of optogenetics, where real-time optical stimulation is used to control the action potential firing of neurons in vivo [5].

Long-term performance of an electronic polymerase .... ultimately our goal should be to construct entire synthetic genomes in a matter of minutes. Using arrays of thousands or millions of polymerases, we can construct many billions of nucleotides per second of novel synthetic DNA. The cost of DNA synthesis should be reduced to be pennies per genome. Making a mistake, like not correctly writing an amino acid sequence for a custom protein, should not be a multi-million dollar mistake for synthetic genome engineering. The design-build iterative cycle needs to be shortened as much as possible

Protein synthesis without DNA or RNA

https://groups.google.com/g/enzymaticsynthesis/c/3YEEv0OULo0

Here's an idea. Since no particular progress has been made on the problem of enzymatic DNA synthesis, what about focusing on enzymatic synthesis of proteins instead?

Currently, in vitro protein synthesis protocols require DNA or mRNA.... I don't think this is impossible to override, though. This would require either hacking a ribosome or hacking tRNA synthetase. Ultimately the protocol would probably end up with steps like "select one of 20 vials of tRNAs already attached to tRNA and amino acids, insert into ribosome reaction".

Antibiotics already cause mistranslation in ribosomes, so those mechanisms could be studied to determine how easy or difficult it may be to cause only mistranslation in a ribosome.

Additionally, some method of capping or blocking a ribosome would be necessary, so that after delivering an amino acid to the ribosome that no other tRNA synthetase could make a delivery. Next, the protocol would have to call for a wash step to remove the excess tRNA synthetase and amino acids, next decap, then proceed on to the next step. These wash/wait steps would have to be compatible with the ribosome.

The advantage of this method is that unlike oligonucleotide synthesis, large proteins can be constructed. Personally I would find this most useful for the construction of DNA polymerases, which so far cannot be directly synthesized in one molecule constructed by the phosphoramidite method. by experimenting with the amino acid design of polymerases, perhaps more efficient synthesis procedures can be made. But in the mean time, an effective cell-free, in vitro protein synthesis would be available without DNA and without RNA (or rather, without special/synthesized DNA).

Cathal's solution: How about this: degenerate mRNA nucleotides and orthologous tRNAs, so an arbitrary "next triplet" can be loaded with a chosen amino? To avoid repeats have two incompatible triplets staggered and two sets of 21 aminos per triplet. Feed charged tRNA stepwise, wash between. No modified ribosome required, minimal modified tRNAs.

Edward Jung's solution: https://patents.google.com/patent/US7923533/en

Ribosome approach

https://groups.google.com/g/enzymaticsynthesis/c/Fx0WSeqFrjs/m/1ljdUcodCAAJ

https://patents.google.com/patent/US20240166682A1/en

More links

adam marblestone's molecular nanotechnology concept page: https://longitudinal.blog/bottleneck-analysis-positional-chemistry/

julian englert's broken "rotary ribosome" concept: https://www.youtube.com/watch?v=5R1EQAe9vio

Other notes

See refs for some more notes.

more notes

DNA-directed vs DNA-dependent ??

also check

  • recombinase
  • CRISPR stuff
  • helicase
  • replicase
  • DNA ligase
  • RNA ligase
  • RNA ligase ribozyme
  • RNA helicase
  • RNA replicase

terminal deoxynucleotide transferase

terminal deoxyribonucleotidyltransferase terminal transferase TdT addase activity

https://en.wikipedia.org/wiki/Terminal_deoxynucleotidyl_transferase

Large components

Taq polymerase http://www.uniprot.org/uniprot/P19821

http://www.ebi.ac.uk/interpro/entry/IPR019760

""" DNA polymerases (EC:2.7.7.7) can be classified, on the basis of sequence similarity [PMID: 3479792, PMID: 2196557], into at least four different groups: A, B, C and X. Members of family X are small (about 40 kDa) compared with other polymerases and encompass two distinct polymerase enzymes that have similar functionality: vertebrate polymerase beta (same as yeast pol 4), and terminal deoxynucleotidyl-transferase (TdT) (EC:2.7.7.31). The former functions in DNA repair, while the latter terminally adds single nucleotides to polydeoxynucleotide chains. Both enzymes catalyse addition of nucleotides in a distributive manner, i.e. they dissociate from the template-primer after addition of each nucleotide. DNA-polymerases show a degree of structural similarity with RNA-polymerases.

Five regions of similarity are found in all the polymerases of this entry. The signature of this entry is to the conserved region, known as 'motif B' [PMID: 2196557]; motif B is located in a domain which, in E. coli polA, has been shown to bind deoxynucleotide triphosphate substrates; it contains a conserved tyrosine which has been shown, by photo-affinity labelling, to be in the active site; a conserved lysine, also part of this motif, can be chemically labelled, using pyridoxal phosphate. """

Catalytic domain matching

even more things

"Flaviviruses produce a polyprotein from the ssRNA genome. The polyprotein is cleaved to a number of products, one of which is NS5. Recombinant dengue type 1 virus NS5 protein expressed in Escherichia coli exhibits RNA-dependent RNA polymerase activity. This RNA-directed RNA polymerase possesses a number of short regions and motifs homologous to other RNA-directed RNA polymerases.[11]"

a template-independent polymerase like Terminal deoxynucleotidyl Transferase (TdT) or poly(A) polymerase

""" According to certain aspects, oligonucleotide sequences may be prepared using ink jet techniques known to those of skill in the art, electrochemical techniques known to those of skill in the art, microfluidic techniques known to those of skill in the art, photogenerated acids known to those of skill in the art, or photodeprotected monomers known to those of skill in the art. Such techniques have the advantage of making oligonucleotides at high speed, low cost, fewer toxic chemicals, enhanced portability and ability to interleave DNA biochemistry (e.g. modifications, polymerases, hybridization etc.) with de novo (digital or analog) synthesis. For example, spatially patterned light, either directly from camera optics or from Digital Micromirror Display devices (DMD), can be used with aqueous chemistry. See US2003/0228611. For example, a template-independent polymerase like Terminal deoxynucleotidyl Transferase (TdT) or poly(A) polymerase—alternatively, a template-dependent polymerase like Taq or Phi29 derivatives, can have their basic polymerase function, base-specificity or fidelity programmable by light by incorporating an azobenzene amino acid (see Hoppmann C, Schmieder P, Heinrich N, Beyermann M. (2011) Chembiochem. 12(17):2555-9. doi: 10.1002/cbic. 201100578. Epub 2011 October 13, Photoswitchable click amino acids: light control of conformation and bioactivity) into the active site of the polymerase or 5′43′ exonuclease domains (if present). """

... can have their basic polymerase function, base-specificity or fidelity programmable by light by incorporating an azobenzene amino acid (see Hoppmann C, Schmieder P, Heinrich N, Beyermann M. (2011) Chembiochem. 12(17):2555-9. doi: 10.1002/cbic. 201100578. Epub 2011 October 13, Photoswitchable click amino acids: light control of conformation and bioactivity) into the active site of the polymerase or 5′43′ exonuclease domains

see Hoppmann C, Schmieder P, Heinrich N, Beyermann M. (2011) Chembiochem. 12(17):2555-9. doi: 10.1002/cbic. 201100578. Epub 2011 October 13, Photoswitchable click amino acids: light control of conformation and bioactivity) into the active site of the polymerase or 5′43′ exonuclease domains

""" Light sensitive neurons (optogenetics) can trigger ion-sensitive polymerases (see Zamft B, Marblestone A, Kording K, Schmidt D, Martin-Alarcon D, Tyo K, Boyden E, Church GM (2012) Measuring Cation Dependent DNA Polymerase Fidelity Landscapes by Deep Sequencing. PLoS One, in press) or, for some applications, the ion flux patterns themselves can constitute the stored datasets. """

""" The nucleotide type incorporated can be determined by: a) the intersection of a light pulse coincident with a particular dNTP (or rNTP or other monomer class) present at that time point in a cyclic pattern of dNTP solutions. b) ‘caged’ (i.e. photo-activatable or photo-inactivatable) dNTPs, rNTPs or cations. c) base-specific, light-modulated steric or conformational selectivity (see Hoppmann C, Schmieder P, Heinrich N, Beyermann M. (2011) Chembiochem. 12(17):2555-9. doi: 10.1002/cbic.201100578. Epub 2011 October 13. Photoswitchable click amino acids: light control of conformation and bioactivity). Poly(A) polymerase is particularly useful since its specificity for ATp relative to other rNTPs is due to a conformational change which can be mimicked by a photo-sensitive amino acid linkage (like azobenzene, with or without crosslinking). """

Photoswitchable Click Amino Acids: Light Control of Conformation and Bioactivity http://onlinelibrary.wiley.com/doi/10.1002/cbic.201100578/abstract

"Click the switch: By using a photoswitchable click amino acid (PSCaa) a light-induced intramolecular thiol-ene click reaction with a neighboring cysteine under very mild conditions results in an azobenzene bridge (see figure). By expanding the genetic code for PSCaa the specific incorporation of photoswitch units into proteins in living cells can result in an exciting approach for studying light-controllable activity, in vivo."

Reversible photocontrol of biological systems by the incorporation of molecular photoswitches

Genetically encoding photoswitchable click amino acids in Escherichia coli and mammalian cells

SELEX

phage display

ribosome display

mRNA display

phage-assisted continuous evolution

TRAP display

liposome display

megavalent bead surface display

mRNA display selection

  • azobenzene-matching aptamer, introduce azobenzene into the aqueous environment ?