📄 synthetic biology

Continuous target-specific mutagenesis and rapid gene evolution by diversity-generating retroelements in Escherichia coli

This paper demonstrates that diversity-generating retroelements (DGRs) can be engineered in *Escherichia coli* and coupled with horizontal gene transfer to create a simple, programmable system for the continuous, iterative, and position-specific mutagenesis of target genes, enabling the rapid evolution of proteins such as pyrrolysyl-tRNA synthetase for biotechnological applications.

Original authors: Liu, Y., Gu, Y., Agam, G., Rehm, F., Spinck, M., Chin, J., Holliger, P.

Published 2026-07-22✓ Author reviewed ⓘ

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Liu, Y., Gu, Y., Agam, G., Rehm, F., Spinck, M., Chin, J., Holliger, P.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine you are a master chef trying to invent a new recipe. You could try to change the whole cookbook at once, but that's chaotic and likely to ruin the dish. Or, you could try to change just one specific spice in a single sentence, but doing that by hand is slow and tedious. This is the challenge of "directed evolution," a field of science where researchers try to speed up nature's trial-and-error process to create better proteins, medicines, and materials. For a long time, scientists had to mix chemicals in a lab to create random mutations, pick the winners, and then start over again. It was like trying to find a specific needle in a haystack by building a new haystack every hour. Recently, scientists discovered a way to do this inside living cells, letting them evolve continuously. However, these living "factories" often got tired, confused, or broke down the very instructions they were supposed to follow. The big question was: Could we build a system that keeps evolving a specific part of a protein, over and over again, without the factory collapsing or the instructions getting garbled?

This paper introduces a clever solution called HGT-DGR, which acts like a high-speed, self-renewing assembly line for evolving proteins. The researchers took a natural "hyper-mutation" system found in viruses, called Diversity-Generating Retroelements (DGRs), and installed it inside E. coli bacteria. Think of the DGR as a magical photocopier that doesn't just copy a page; it intentionally scribbles random new words onto specific letters (only the letter 'A') of a target sentence. The problem was that this photocopier was messy: it kept scribbling on its own instruction manual (the template), eventually erasing the rules it needed to keep working, and the bacteria hosting it got exhausted.

To fix this, the team invented a "relay race" strategy. Instead of letting one batch of bacteria run the race until they collapse, they pass the baton—the gene they want to evolve—to a fresh, healthy group of bacteria every few days. They did this using a natural bacterial process called conjugation, which is like two bacteria shaking hands and swapping a small, circular piece of DNA (a plasmid). By constantly moving the evolving gene into new hosts that have a fresh, uncorrupted instruction manual, the system can keep mutating the target gene for weeks without breaking down.

The results are impressive. After running this relay race for seven days, the team found that about 40% of the bacteria in the culture carried a mutated version of the target gene. The mutations weren't random chaos; they were precise, hitting specific spots as designed. The average gene in the library had about 6% of its "A" letters changed, creating a massive diversity of new protein versions. To prove this worked, they used the system to evolve an enzyme called MmPylRS, which is used to add special, non-standard building blocks to proteins. They successfully guided the enzyme to accept two different new building blocks, showing that the system can be reprogrammed to target different parts of a gene and evolve new functions. The paper suggests this method is a simple, low-cost way to generate huge libraries of protein variants, which could be incredibly useful for designing new drugs or for training artificial intelligence models to understand how proteins work.

Technical Summary: Continuous Target-Specific Mutagenesis and Rapid Gene Evolution by Diversity-Generating Retroelements in Escherichia coli

Problem Statement
Directed evolution is a cornerstone of bioengineering, yet traditional in vitro methods (e.g., error-prone PCR, site-saturation mutagenesis) are often slow, costly, and limited in their exploration of sequence space. While in vivo continuous evolution systems (such as PACE) offer speed and autonomy, they frequently lack precise control over mutation parameters and target sequences. Existing targeted in vivo mutagenesis strategies, including those utilizing orthogonal episomes, base deaminases, or retron editors, are often constrained by limited mutational target windows, restricted mutational spectra, or a lack of orthogonality with genome replication. Crucially, there is a need for a sustainable, in vivo system capable of programmable, site-saturation mutagenesis at single-nucleotide resolution across user-defined positions to enable structure-guided targeting of evolved molecules.

Methodology
The authors engineered a synthetic system in E. coli (which lacks natural DGRs) based on the Diversity-Generating Retroelements (DGRs) found in the Bordetella bacteriophage BPP1. The system comprises three core components:

Template Region (TR): A DNA sequence transcribed into RNA, serving as the template for reverse transcription.
Reverse Transcription Cassette: Includes the BPP1 reverse transcriptase (bRT) and the accessory variability determinant protein (Avd).
Synthetic Retrohoming Machinery: To facilitate integration in E. coli, the authors introduced the ssDNA annealing protein CspRecT (from Collinsella stercoris) and a dominant-negative MutL mutant (mMutL E32K) to suppress the host mismatch repair (MMR) system.

To overcome the instability of DGR systems in continuous culture—specifically the degradation of the TR sequence due to mutagenesis and the growth burden of the mutagenesis machinery—the authors coupled DGR mutagenesis with F-plasmid-mediated horizontal gene transfer (HGT). This "HGT-DGR" strategy involves iterative cycles where:

A donor strain carries the target gene (Variable Region, VR) and constant DGR components on an F-plasmid.
The F-plasmid is conjugated into a recipient strain harboring a fresh, unmutated TR.
DGR mutagenesis is induced in the recipient, accumulating mutations in the VR.
The mutated VR is then transferred to a new recipient with a fresh TR, allowing for the progressive accumulation of mutations without eroding the template.

The system was optimized by varying TR length, utilizing an exonuclease-deficient host strain ( $\Delta sbcB, \Delta recJ, \Delta exoX$ ) to prevent cDNA degradation, and selecting optimal antibiotic markers (nourseothricin) and induction conditions (low-level arabinose) to balance mutation frequency with cell viability.

Key Results

Reprogramming and Specificity: The BPP1 DGR system was successfully reprogrammed to target user-defined sequences. When applied to the sfGFP gene, mutagenesis occurred exclusively at adenine positions within the TR-specified region, leaving adjacent non-adenine residues unchanged. The mutational spectrum showed a bias toward Guanine and Thymine substitutions at adenine sites, with negligible indels.
Optimization: Extending the TR homology arm length (up to 93 bp) and using the exonuclease-deficient strain nearly doubled the mutation frequency.
Sustainable HGT-DGR: In a standard continuous culture without HGT, DGR activity ceased to increase after three days due to TR mutation and host selection pressure. The HGT-DGR approach enabled stable, iterative mutagenesis over seven days.
Diversity Generation: After seven cycles of HGT-DGR, the population exhibited an average mutation load of approximately 6% at adenine positions in the target segment. About 40% of the population consisted of mutant sequences, with a median pairwise Hamming distance of 4 among variants, indicating a diverse library with low duplication.
Directed Evolution Application: The system was used to evolve Methanococcus mazei pyrrolysyl-tRNA synthetase (MmPylRS) for the incorporation of non-canonical amino acids (ncAAs).
- CbzK: A single round of evolution successfully enriched variants capable of incorporating Nε-benzyloxycarbonyl-L-lysine (CbzK), including the known Y306G mutation.
- BenzK to CbzK: By iteratively targeting different active site regions (TR3 and TR4) in successive conjugation rounds, the authors evolved MmPylRS variants that could incorporate Nε-benzyl-L-lysine (BenzK) and subsequently shifted specificity to CbzK or broadened substrate preference to accept both. Mass spectrometry confirmed the dual activity of selected variants.

Significance and Claims
The paper claims to provide a simple, low-cost, and sequence-programmable system for iterative, position-specific, and tunable in vivo mutagenesis. By decoupling the mutagenesis machinery from the target gene via HGT, the authors address the critical limitations of TR instability and host fitness costs that have hindered the long-term application of DGRs in continuous evolution.

The authors assert that HGT-DGR enables the rapid diversification of user-specified gene segments, generating large libraries of variants suitable for biotechnology and medicine. They highlight the system's potential for evolving binding interfaces (e.g., antibody CDR loops) and catalytic sites. Furthermore, the paper suggests that the ability to generate large, structured sequence-function datasets with defined mutational spaces could be particularly valuable for training and benchmarking artificial intelligence models in biomolecular engineering. The system is presented as a complementary tool to existing directed evolution methods, offering a unique balance of programmability and continuous evolution capabilities.

More like this