Resurrecting Ancient DNA Polymerases

Charles Hargrove¹ and Scott Nelson².

¹Undergraduate Research Associate, Roy J. Carver Department of Biochemistry, Biophysics, and Molecular Biology, Iowa State University.

²Professor, Roy J. Carver Department of Biochemistry, Biophysics, and Molecular Biology, Iowa State University.

Project at a glance: We are trying to bring ancient DNA-copying enzymes back to life in the lab. By using computational methods to reconstruct what ancestral DNA polymerases may have looked like millions or billions of years ago, we can synthesize these genes, purify the proteins, and test how they work. This project will help us understand how DNA replication evolved and may reveal unusually stable or useful enzymes with future applications in biotechnology. Crowdfunding will directly support gene synthesis and cloning, letting us move from computer predictions to real proteins that can be studied experimentally.

1. Abstract

Every living organism depends on DNA polymerases: molecular machines that copy genetic information. The polymerases found in living organisms today are only the surviving members of a much older enzyme family. Earlier versions of these enzymes helped copy DNA in organisms that lived tens to hundreds of millions of years ago, but those proteins are long gone. Enzymes do not leave behind fossils that can be dug up and tested. However, their sequences can be inferred from the DNA polymerases that survived into the modern world.

This proposal will fund the synthesis of four DNA polymerase genes: two reconstructed ancestral polymerases, which we call molecular fossil polymerases, and two matched modern polymerases. The two ancestral enzymes selected for this first round are:

Sulfur-lineage PolA: This ancestor comes from a bacterial lineage related to Sulfurovum and other Campylobacterota, organisms associated with sulfur-rich, low-oxygen environments.

Stalked-cell PolA: This ancestor comes from a Verrucomicrobiota-related lineage that includes unusual bacteria such as Roseimicrobium and Prosthecobacter relatives.

Under a simple molecular-clock model, these ancestral nodes may represent polymerase forms roughly 100 million years removed from many sampled modern relatives. At that time, Earth was in the Cretaceous Period: the planet was warmer, sea levels were much higher, flowering plants were spreading, and dinosaurs still dominated the land.

The requested funds are only for codon optimization and DNA synthesis. All downstream expression, purification, and biochemical testing will be carried out using existing laboratory resources. All resulting sequences, trees, structural models, expression results, gels, and biochemical data will be shared publicly through ResearchHub Notebook and an external open repository.

2. Introduction

DNA polymerases are the copying machines of life. Every genome, from a bacterium to a human, depends on enzymes that can read one strand of DNA and build another. These enzymes do more than copy letters. They balance speed, accuracy, proofreading, stability, and substrate recognition. Change that balance, and the biology changes with it.

Family A DNA polymerases, often called PolA enzymes, are one of the best-known DNA polymerase families. The textbook example is E. coli DNA polymerase I and its famous Klenow fragment, a shortened enzyme that retains the polymerase domain and the 3′ to 5′ proofreading exonuclease domain but lacks the N-terminal 5′ to 3′ exonuclease domain [2,3]. These polymerase cores are experimentally attractive because they contain the main DNA-copying and proofreading machinery in a compact, well-studied form.

Recent work has shown that the PolA family is far more diverse than the classic examples suggest [1]. A large-scale reclassification of family A DNA polymerases separated known PolA sequences into 19 major clusters, including several newly recognized subfamilies with distinctive structural and functional features [1]. This makes PolA an especially appealing system for ancestral sequence reconstruction, or ASR. Rather than trying to reconstruct one impossible “ancestor of all PolA,” we can reconstruct ancestors within coherent evolutionary clusters and then test them in the lab.

ASR uses modern sequences and phylogenetic trees to infer what ancient proteins may have looked like [4]. The inferred ancestral amino-acid sequence can then be converted into a synthetic gene, expressed in the lab, purified, and tested. In other words, ASR lets us do molecular paleontology. We are not extracting DNA from fossils and we are not reviving extinct organisms. Instead, we are using modern sequence data to build and test hypotheses about ancient enzymes.

This project focuses on two reconstructed PolA ancestors from a larger set of eleven ASR clusters. The first is Sulfur-lineage PolA, reconstructed from a cluster dominated by Sulfurovum and related Campylobacterota sequences. These bacteria are associated with sulfur-rich, low-oxygen environments, including hydrothermal and other chemically dynamic habitats. This cluster offers a chance to ask whether a polymerase lineage from a sulfur-linked ecological background carries unusual biochemical properties.

The second is Stalked-cell PolA, reconstructed from a cluster containing Roseimicrobium, Prosthecobacter, and other Verrucomicrobiota-related bacteria. This is an unusual bacterial neighborhood, distinct from the more familiar E. coli-like examples that dominate most DNA polymerase textbooks. It gives the project a second, non-redundant evolutionary branch to compare against the Sulfur-lineage enzyme.

These two clusters were chosen because they are not merely “more of the same.” They are biologically distinctive, phylogenetically separated, and appear underexplored at the biochemical level. Together they provide a compact first test of the project’s core idea: ancestral polymerases may reveal biochemical properties that are hidden, diluted, or lost in their modern relatives.

A rough molecular-clock analysis provides a sense of evolutionary scale. Using the tree files for these two clusters, and assuming a simple substitution rate of 10⁻⁸ substitutions per site per year, the estimated median distance from the ancestral node to modern tips corresponds to approximately:

~133 million years for Sulfur-lineage PolA
~111 million years for Stalked-cell PolA

These numbers should be treated as broad clock-model estimates, not exact dates. The trees are not perfect clocks, and protein branch lengths are not precise measures of time. Still, the estimates place these reconstructed enzymes near the Cretaceous Period, when Earth was warmer, sea levels were much higher than today, flowering plants were spreading, and dinosaurs still dominated the land.

3. Study Goals and Hypotheses

Primary goal

To synthesize and obtain initial biochemical data for two reconstructed ancestral Klenow-like DNA polymerases and two matched modern Klenow-like polymerases from the same evolutionary clusters.

Main hypothesis

Reconstructed ancestral Klenow-like PolA enzymes from well-supported bacterial clades can be synthesized, expressed, purified, and tested as folded, active enzymes.

Specific hypotheses

H1. Expression and activity
At least one ancestral polymerase will express as soluble protein and show detectable DNA primer-extension activity.

H2. Ancient-modern biochemical differences
At least one ancestral polymerase will differ measurably from its matched modern representative in a tractable biochemical property, such as thermal stability, primer-extension activity, temperature preference, or proofreading-associated behavior.

H3. Cluster-specific behavior
The Sulfur-lineage and Stalked-cell polymerases will show different biochemical profiles, suggesting that different PolA lineages explored different solutions to the same basic problem: copying DNA accurately enough, quickly enough, and under their own environmental constraints.

4. Experimental Constructs

This project will synthesize four genes total. Each gene will encode a Klenow-like PolA construct, meaning the polymerase and 3′ to 5′ exonuclease/proofreading region will be included, while the N-terminal 5′ to 3′ exonuclease domain will be excluded when present in the full-length modern enzyme.

Construct	Friendly name	Current name or accession	Source / cluster	Purpose
1	Sulfur-lineage PolA ancestor	A0A1W1C4H9_Node1	Reconstructed ancestor from Sulfurovum / Campylobacterota-like cluster	Molecular fossil polymerase
2	Modern Sulfur-lineage PolA	WP_011981001.1	Klenow-like region of DNA polymerase I from Sulfurovum sp. NBC37-1	Modern comparison enzyme
3	Stalked-cell PolA ancestor	A0A366HPX8_Node1	Reconstructed ancestor from Verrucomicrobiota / Roseimicrobium / Prosthecobacter-like cluster	Molecular fossil polymerase
4	Modern Stalked-cell PolA	WP_165205871.1	Klenow-like region of DNA polymerase I from Roseimicrobium sp. ORNL1	Modern comparison enzyme

The paired design is important. Synthesizing only the two ancestral enzymes would tell us whether the reconstructed proteins work, but not how they compare with living relatives. By synthesizing one modern representative from each cluster, we can ask a sharper question: what changed between the ancestral polymerase and a modern member of the same lineage?

5. Why These Two Polymerases?

5.1 Sulfur-lineage PolA: an enzyme from a chemically unusual world

The Sulfur-lineage PolA ancestor was selected because the BLAST neighborhood is dominated by Sulfurovum and related Campylobacterota sequences. Sulfurovum species are associated with sulfur-rich environments, and related vent-associated bacteria are known for chemolithoautotrophic lifestyles, using chemical energy rather than sunlight or organic food [5].

This makes the cluster exciting for two reasons: First, it is evolutionarily distinct from the more familiar polymerases used in standard laboratory biology. It is not just another close cousin of E. coli Pol I. Second, sulfur-rich and low-oxygen environments can impose unusual biochemical constraints. The polymerase itself may or may not reflect those constraints, but the lineage gives us a plausible reason to look.

The proposed modern comparison enzyme is the Klenow-like region of DNA polymerase I from Sulfurovum sp. NBC37-1, accession WP_011981001.1. This organism appears in the BLAST hit list for the ancestral sequence and provides a named, biologically interpretable modern comparator.

5.2 Stalked-cell PolA: an evolutionary oddball

The Stalked-cell PolA ancestor was selected because its BLAST neighborhood includes Verrucomicrobiota-related bacteria, including Roseimicrobium and Prosthecobacter relatives. This is an appealingly strange biological neighborhood. Prosthecobacter is especially famous among microbiologists because some members contain bacterial tubulin-like proteins, an unusual feature that gives the lineage a bit of evolutionary sparkle.

The selected modern comparison enzyme is the Klenow-like region of DNA polymerase I from Roseimicrobium sp. ORNL1, accession WP_165205871.1. This was the top named modern hit in the cluster and is therefore the cleanest direct comparison for the ancestral construct. Roseimicrobium sp. ORNL1 is a Verrucomicrobiota bacterium isolated from the rhizosphere of Populus deltoides, the Eastern cottonwood tree [6].

6. Methods

6.1 Computational selection

The starting point for this project was a family A DNA polymerase sequence resource and an iterative ancestral reconstruction workflow. The workflow included:

sequence clustering
multiple sequence alignment
domain-aware trimming
phylogenetic reconstruction
ancestral sequence reconstruction
BLAST analysis of reconstructed nodes
AlphaFold structural sanity checks
tree-based rough age estimation

Earlier pilot work showed that reconstruction quality depends strongly on how sequences are grouped and trimmed. Overly aggressive automated trimming can remove real polymerase or exonuclease sequence. A balanced trimming strategy, including testing with trimAl -gt 0.4 -cons 0.6, produced more plausible Klenow-like reconstructions in the pilot analyses.

From eleven reconstructed clusters, two were prioritized for this synthesis proposal because they combined:

distinctive evolutionary placement
clear BLAST-supported cluster identities
plausible Klenow-like sequence length and domain organization
preserved core catalytic motifs
interesting public-facing biological context
strong potential to generate preliminary data for a larger grant

6.2 Rough molecular-clock estimate

For each of the two selected tree files, the internal label Node1 was treated as the reconstructed ancestral node. The distance from Node1 to each modern tip was calculated using branch lengths in the tree. Because the trees are not ultrametric, the node-to-tip distances vary among descendants. Therefore, the median Node1-to-tip distance was used as a rough central estimate.

The median distances were:

Sulfur-lineage PolA: 1.332 substitutions per site
Stalked-cell PolA: 1.114 substitutions per site

Using a simple middle-rate clock scenario of 10⁻⁸ substitutions per site per year, this gives rough estimates of:

~133 million years for Sulfur-lineage PolA
~111 million years for Stalked-cell PolA

A faster clock of 10⁻⁷ substitutions per site per year would place the nodes around tens of millions of years ago, while a slower clock of 10⁻⁹ substitutions per site per year would push them toward billion-year scales. For this reason, the proposal will describe these as rough Cretaceous-scale molecular-clock estimates, not exact dates.

6.3 Gene design and synthesis

Each selected amino-acid sequence will be converted into a codon-optimized DNA sequence for expression in E. coli. Each construct is expected to be approximately 1,800 base pairs, corresponding to a Klenow-like fragment of roughly 600 amino acids.

The synthesis order will include:

A0A1W1C4H9_Node1, Sulfur-lineage ancestral PolA
Klenow-like region of WP_011981001.1, modern Sulfurovum sp. NBC37-1 PolA
A0A366HPX8_Node1, Stalked-cell ancestral PolA
Klenow-like region of WP_165205871.1, modern Roseimicrobium sp. ORNL1 PolA

The funding request covers only codon optimization and DNA synthesis. Expression testing, purification, and biochemical assays will be performed using existing lab infrastructure and reagents [9,10].

6.4 Expression and purification

Synthesized genes will be expressed in a standard bacterial expression system. Constructs will be designed for purification using an existing lab-compatible affinity tag, likely a His-tag or comparable fusion strategy. Initial purification will use immobilized metal affinity chromatography followed by buffer exchange.

The first-pass expression screen will evaluate:

whether the protein is expressed
whether it is soluble
whether it can be purified in useful amounts
whether the apparent size matches expectation by gel electrophoresis

If a construct expresses poorly, expression conditions such as temperature, induction time, induction strength, and host strain may be adjusted using standard lab workflows.

6.5 Initial biochemical characterization

The first biochemical tests will be deliberately simple. The goal is not to fully characterize all enzyme kinetics in this small pilot. The goal is to find out whether these molecular fossils can be made and whether they do anything interesting.

The initial assays will include:

Primer-extension activity
Can the enzyme copy DNA from a defined primer-template substrate?
Temperature response
Does the enzyme work better or worse at different assay temperatures?
Thermal stability
Does the ancestral enzyme retain activity or structure after heat challenge better than its modern comparison enzyme?
Proofreading-associated behavior
Does the enzyme show detectable 3′ to 5′ exonuclease or mismatch-excision activity?
Fidelity
Does the enzyme make mistakes more or less often?

7. Planned Analyses

7.1 Feasibility outcomes

The primary feasibility outcomes will be:

successful synthesis
soluble expression
purification yield
detectable primer-extension activity

7.2 Comparative biochemical outcomes

For constructs yielding adequate protein, the following comparisons will be made between each ancestral enzyme and its modern representative:

relative activity under matched primer-extension conditions
apparent thermal stability
activity after heat challenge
temperature dependence of activity
proofreading-associated activity, if feasible

A construct will be considered especially promising if it shows any of the following:

at least a 2-fold difference in primer-extension activity under matched conditions
at least a 5 °C difference in apparent thermal stability or heat-survival behavior
a clear difference in proofreading-associated behavior
unusually strong expression or solubility compared with the modern representative
activity under conditions where the modern representative performs poorly

7.3 Interpretation

Because this is a four-construct pilot, results will be interpreted by effect size, reproducibility, and biochemical plausibility rather than by formal population-level statistics. This is a feasibility and discovery experiment, not a powered clinical or ecological study.

If the ancestral polymerases express and show activity, the project will have demonstrated that the ASR workflow can generate testable, functional PolA enzymes.

If one ancestral enzyme behaves differently from its modern representative, that enzyme becomes a lead candidate for deeper study.

If both ancestral enzymes fail, that still provides useful information about construct boundaries, alignment quality, reconstruction confidence, or expression strategy.

If the ancestors behave very similarly to the modern representatives, that is also informative: it would suggest that these core polymerase properties may be conserved across deep evolutionary time in these clades.

8. Expected Outcomes and Broader Impact

This project has two major possible impacts.

8.1 Understanding enzyme evolution

The first impact is scientific. DNA polymerases sit at the center of genetic life. By comparing reconstructed ancestral polymerases with modern representatives, we can begin to ask how the copying machinery of life changed over evolutionary time.

Did ancient polymerases tend to be more stable?

Did proofreading behavior change?

Did polymerase activity shift with environmental context?

Did different bacterial lineages evolve different tradeoffs between stability, speed, and accuracy?

These are large questions, but a small four-gene pilot can provide the first experimental foothold.

8.2 Discovering useful polymerases

The second impact is practical. DNA polymerases are among the most useful enzymes in biotechnology. PCR, sequencing, cloning, mutagenesis, diagnostics, and synthetic biology all depend on polymerases. The history of Taq polymerase shows how one enzyme with useful properties can transform biology.

Ancestral polymerases may contain useful combinations of properties that are uncommon in modern enzymes. They may be more stable, more tolerant of unusual conditions, or different in how they balance synthesis and proofreading. This project will not claim to discover the next Taq polymerase in one step. But it will open the door to testing whether ancient PolA sequence space contains useful biochemical surprises.

8.3 Public engagement

This project is also easy to explain to non-scientists: “Around 100 million years ago, Earth was a greenhouse planet [8]. Sea levels were high, shallow oceans spread across continents, flowering plants were expanding, and dinosaurs still walked the land. These polymerases are molecular fossils from that ancient biological world, inferred from modern DNA sequences and brought back into the lab for testing.”.

9. Pilot Data

Several pieces of pilot work have already been completed.

9.1 ASR workflow development

A computational pipeline has been established for PolA ancestral reconstruction. The workflow includes sequence clustering, alignment, trimming, tree inference, ancestral reconstruction, BLAST analysis, and structural prediction (Fig. 1 below). Pilot work showed that clade-specific reconstruction is more credible than attempting a single family-wide reconstruction across all PolA diversity.

Figure 1. Pipeline for reconstructing and testing ancestral PolA polymerases. Modern family A DNA polymerase sequences were filtered to retain Klenow-like polymerase cores, clustered into evolutionary lineages, aligned, and used to infer phylogenetic trees. Ancestral Node1 sequences were reconstructed for each cluster, then prioritized using BLAST results, reconstruction confidence, and structure predictions.

9.2 Eleven reconstructed clusters

The initial ASR effort produced eleven candidate ancestral PolA clusters (Fig. 2 below). These clusters span multiple bacterial lineages, including Gemmatimonadota-like, Campylobacterota-like, Chloroflexota-like, Verrucomicrobiota-like, and several distinct Firmicute/Bacillota-like groups.

This proposal narrows the first synthesis round to the two clusters with the strongest combination of scientific interest, evolutionary distinctiveness, and public-facing story:

Sulfur-lineage PolA, A0A1W1C4H9_Node1
Stalked-cell PolA, A0A366HPX8_Node1

Figure 2. Evolutionary context for the selected fossil polymerases. Dense phylogenetic overview of Klenow-like PolA sequences from the eleven sequence clusters used for ancestral reconstruction. Most lineages are shown in gray. The two lineages selected for synthesis and biochemical comparison are highlighted: Stalked-cell PolA in red and Sulfur-lineage PolA in blue. These two clades represent the ancestral sequences A0A366HPX8_Node1 and A0A1W1C4H9_Node1, respectively.

9.3 Rough age estimates from tree files

Tree files for the two selected clusters were parsed to estimate how far the reconstructed Node1 ancestors sit from sampled modern relatives. Under a middle molecular-clock assumption of 10⁻⁸ substitutions per site per year, the two reconstructed ancestors fall around:

~133 million years for Sulfur-lineage PolA
~111 million years for Stalked-cell PolA

These estimates are intentionally rough. They are not being used as strict evolutionary dates. Instead, they provide a useful scale for public communication: these are plausibly Cretaceous-scale molecular fossils under a simple clock model.

9.4 Structural plausibility

The two reconstructed polymerases selected for synthesis were evaluated by AlphaFold structure prediction [7] using their Klenow-like core sequences (Fig. 3). Both the Sulfur-lineage PolA ancestor and the Stalked-cell PolA ancestor produced compact, multi-domain models consistent with family A DNA polymerase/Klenow-like architecture [9,10].

Figure 3. Predicted structures of selected fossil polymerases.
AlphaFold predictions of the reconstructed Klenow-like PolA cores selected for synthesis and biochemical comparison. (A) Sulfur-lineage PolA, corresponding to the reconstructed ancestral node A0A1W1C4H9_Node1. (B) Stalked-cell PolA, corresponding to the reconstructed ancestral node A0A366HPX8_Node1. Both models adopt compact, multi-domain architectures consistent with family A DNA polymerase/Klenow-like folds, supporting the use of these reconstructed sequences for synthesis, expression, and preliminary biochemical testing.

10. Ethics and Biosafety

This project does not involve human participants, human tissue, protected health information, vertebrate animals, clinical interventions, environmental release, or infectious organisms. Therefore, IRB and IACUC approval are not required.

The project will synthesize individual enzyme-coding genes and express them in standard laboratory expression hosts. No living Sulfurovum, Roseimicrobium, or related environmental organism will be cultured as part of this proposal.

Standard institutional recombinant DNA and biosafety practices for non-pathogenic cloning and protein expression will be followed.

11. Data Management and Open Science

All major project outputs will be shared publicly, including:

ancestral amino-acid sequences
codon-optimized DNA sequences
modern comparison sequences
multiple sequence alignments
phylogenetic trees
ancestral reconstruction summaries
AlphaFold or other structure-prediction files
plasmid maps
expression gels
purification data
primer-extension assay data
thermal stability data
analysis scripts
negative results and troubleshooting notes

Data will be posted to ResearchHub Notebook and mirrored to Open Science Framework.

12. Timeline

The anticipated timeline is approximately 8 to 12 weeks from final construct approval to synthesized DNA in hand and initial expression testing.

Stage	Estimated time
Final construct review and order placement	1 week
Gene synthesis and delivery	2 weeks
Expression and purification	2 weeks
First activity assays	1 week
Stability assays	1 week
Exonuclease assays	1 week
Fidelity assays	1 week

13. Budget

This proposal requests funding for four codon-optimized DNA constructs to be synthesized by Gene Universal.

Gene Universal charges $0.19 per base, and each construct is expected to be approximately 1,800 base pairs.

Codon-optimized synthetic gene, ~1,800 bp

4

$342

$1,368

All downstream cloning, expression, purification, and biochemical testing will be carried out using existing laboratory resources. The crowdfunding request is therefore focused on the one thing currently missing: the physical DNA needed to bring these molecular fossils into the lab.

14. References

1. Czernecki, D., Nourisson, A., Legrand, P. & Delarue, M. Reclassification of family A DNA polymerases reveals novel functional subfamilies and distinctive structural features. Nucleic Acids Research 51, 4488–4507 (2023). doi:10.1093/nar/gkad242.

2. Brutlag, D., Atkinson, M. R., Setlow, P. & Kornberg, A. An active fragment of DNA polymerase produced by proteolytic cleavage. Biochemical and Biophysical Research Communications 37, 982–989 (1969).

3. Beese, L. S., Derbyshire, V. & Steitz, T. A. Structure of DNA polymerase I Klenow fragment bound to duplex DNA. Science 260, 352–355 (1993). doi:10.1126/science.8469987.

4. Spence, M. A., Kaczmarski, J. A., Saunders, J. W. & Jackson, C. J. Ancestral sequence reconstruction for protein engineers. Current Opinion in Structural Biology 69, 131–141 (2021). doi:10.1016/j.sbi.2021.04.001.

5. Nakagawa, S. et al. Deep-sea vent ε-proteobacterial genomes provide insights into emergence of pathogens. Proceedings of the National Academy of Sciences USA 104, 12146–12150 (2007). doi:10.1073/pnas.0700687104.

6. Podar, M., Turner, J., Burdick, L. H. & Pelletier, D. A. Complete genome sequence of the novel Roseimicrobium sp. strain ORNL1, a Verrucomicrobium isolated from the Populus deltoides rhizosphere. Microbiology Resource Announcements 9, e00617-20 (2020). doi:10.1128/MRA.00617-20.

7. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021). doi:10.1038/s41586-021-03819-2.

8. National Park Service. Cretaceous Period, 145.0 to 66.0 MYA. U.S. National Park Service educational resource. Published April 27, 2023.

9. Milton, M. E., Choe, J. Y., Honzatko, R. B. & Nelson, S. W. Crystal structure of the apicoplast DNA polymerase from Plasmodium falciparum: the first look at a plastidic A-family DNA polymerase. Journal of Molecular Biology 428, 3920–3934 (2016). doi:10.1016/j.jmb.2016.07.016.

10. Lo, C. Y., Ung, A. R., Koley, T., Nelson, S. W. & Gao, Y. Cryo-EM structures of the Plasmodium falciparum apicoplast DNA polymerase. Journal of Molecular Biology 436, 168842 (2024). doi:10.1016/j.jmb.2024.168842.

Resurrecting Ancient DNA Polymerases

Resurrecting Ancient DNA Polymerases

Funding Progress

Funders

Nonprofit Recipient

Topics