Evolution Consists Of Variation And Selection

In any form of evolution, mutagenesis creates variation within species for their genomes to explore the evolutionary landscape and select variants that confer fitness to the organism, driving adaptations to the forever changing conditions living organisms face.

Directed evolution aims to replicate natural evolution over timescales feasible in a research setting¹. By having a mechanism for rapidly generating variants of a given gene of interest (GOI) and selecting for efficiency, scientists have been able to produce more efficient proteins for a range of applications.

Many classical directed evolution procedures employ global mutagenesis techniques (e.g., irradiating samples with high UV, chemical mutagens, error-prone genome replication etc.). While this may increase the rate of landscape exploration in the entire genome, this can be deleterious. As mutations are occurring in many genes at once, acquisition of an advantageous variant of the gene of interest can be masked by the simultaneous generation of a disadvantageous variant of another gene – particularly if this off-target mutation is in an essential gene. Additionally, off-target mutations in an unrelated gene can confer changes in fitness in the absence of improving a GOI. This can constitute a breakdown in the mutagenesis-selection relationship, giving rise to so-called “evolutionary cheaters”. Suffice to say, directed evolution can be made a much more efficient process if mutagenesis is targeted.

In Vitro Directed Evolution Is Laborious

Other classical directed evolution techniques targeted their mutagenesis by library generation by sequential rounds of error-prone PCR of the GOI, transformation of thousands of variants, and appropriate selection of the best candidates. However, this process is limited by the demand for repetitive labour intensive steps – requiring DNA extraction of the desired variants, mutagenic PCR and subsequent transformation for the next round of selection. The efficiency of directed evolution could therefore be bolstered if the process were to occur completely in vivo in a labour-reduced, continuous manner².

The MutaT7 System Is A Step In The Right Direction

The MutaT7 system utilises a chimeric protein consisting of a T7 RNA polymerase fused to a base deaminase^3–5. Since the T7 polymerase demonstrates high specificity to its canonical promoter, the T7 promoter, MutaT7 presents an attractive system for targeted mutagenesis, whereby a GOI can be placed under the influence of a T7 promoter.

This is ideal for an in vivo directed evolution system since it allows mutagenesis to be coupled to host transcription of the GOI itself, removing much of the required direct input of traditional in vitro approaches. As the T7 polymerase transcribes the GOI , the fused deaminase cleaves amine groups from specific bases - which are subsequently repaired to different base identities, generating diversity in the GOI of all cells in a population.

For accessibility, we have designed a hypermutation plasmid - encoding two copies of the MutaT7 system. The first of the fusion encodes an adenine deaminase fusion; and the other a cytosine deaminase fusion - each under the influence of a pTet inducible promoter, allowing users to modulate the rate of mutagenesis. The ability of these two base editors to work together in one cell is something that has not shown to work in vivo before. The mechanistic detail of each deaminase is as follows:

• A cytosine deaminase will cleave an NH2 group from cytosine - which generates uracil. Since uracil pairs with A this process constitutes a C:G → A:T base pair substitution.
• An adenine deaminase will cleave an NH2 group from an adenine - which generates the non-canonical base inosine. Inosine can base pair with G, so eventually this will constitute a A:T → C:G base pair substitution.

The GOI will be placed on a ‘selection plasmid’, flanked by T7 promoters on opposing strands such that the strand bias of the base deaminases can be somewhat offset, allowing a larger set of possible mutations to occur. Additionally, use of flanking T7 terminator sequences can prevent the MutaT7 system from causing off-target mutations in adjacent loci. Using this system we are limited to the following base transitions of the GOI:

Tying GOI Fitness To Cell Fitness Is A Challenge

If directly evolving an essential gene – say, an antibiotic resistance gene – then selection is simple: as the gene improves, so does the cell fitness, allowing the improved gene to dominate in the population. Often, however, the evolving GOI is not essential to the cell - for example, a biotechnological enzyme - therefore there is a need for genetic circuitry that links the output of the GOI to overall cell fitness⁶. To fill this need, we have designed a second plasmid - the selection plasmid - that will contain one of two growth modulation based selection mechanisms: either positive or negative.

Positive selection occurs based on the improving gene’s ability to induce a gene that encodes an alternative carbon source metabolism system, such that cells containing an improved GOI will outgrow cells that do not. Genetically, this consists of a promoter which senses the output of the GOI – and since this would be specific to a given researcher or project, this is designed to allow ease of cloning from a library of biosensors. Our designs incorporate a sorbitol 6-phosphate dehydrogenase enzyme from E. coli ⁷ such that selection occurs in defined media supplemented with sorbitol.

Negative selection occurs in a slightly more complex system. The plasmid contains a particular growth-slowing gene (of which there are many available options, most of which interfere with RNA polymerase or the ribosome⁸) which is regulated by the pT181 RNA regulatory system^9–11. Upon improvement of the GOI, the antisense RNA component of the pT181 system will be induced - and subsequently block expression of the growth slowing gene (i.e. a NOT gate), and therefore increasing growth rate of those with an improved GOI relative to those which do not.

Negative selection does not require a defined growth medium (e.g. M9 supplemented with sorbitol), and therefore would require less set-up for the user, compared to the positive selection strategy. However, the negative selection ‘not gate’ could be circumvented by suppressor mutations far more easily than positive selection (i.e. acquiring SDH activity spontaneously is far less likely than the inactivation of growth-slowing gene expression).

In order to link ‘enzyme improvement’ (i.e. increased production of an enzyme product) to cell fitness, a biosensor is required. Several impressive studies have generated libraries of small-molecule-inducible promoters^12,13 which we envision could be utilised for a modular directed evolution system such as the rEvolver. Our three-plasmid design is therefore designed with modularity at its core- the end-user would ideally only have to subclone the GOI and an appropriate biosensor element into the selection plasmid, with the hypermutation and knockdown plasmids remaining untouched.

CRISPRi Provides A Helping Hand

Despite the mutagenic efficiency of MutaT7, organisms have evolved to maintain genome integrity – therefore expressing a multitude of DNA repair pathways that prevent errors in replication or DNA damage from becoming fixated in the genome. Whilst important for cellular viability, some of these pathways have proven to limit the hyper-mutagenic potential of in vivo directed evolution systems^3,14. Previous studies have employed CRISPR interference (CRISPRi) to knockdown specific DNA repair pathways and have shown this increases the efficiency of MutaT7 mutagenesis up to 1000 fold.

To increase the efficiency of the rEvolver, we will encode a set of gRNAs which will target dCas9 to two key DNA repair enzymes involved in the base excision repair pathway. This will knock down their expression, increasing the success of mutagenesis by impeding their repair. These gRNAs will target:

• Uracil-DNA glycosylase (ung) is involved in recognising and repairing deoxyuracil residues found in DNA, which would prevent the mutagenic activity of the cytosine deaminase.
• Endonuclease V (nif) is involved in recognising and repairing deoxyinosine residues, which would prevent the mutagenic activity of the adenine deaminase.

Vibrio natriegens - Faster Than The Speed Of Life

Whilst E. coli has proved to be an invaluable tool in molecular biology, there is a desire in the field to accelerate workflow further. In light of this, the marine bacterium Vibrio natriegens has recently garnered much attention as a novel biotechnological chassis, particularly for its incredibly quick doubling time of as little as 7 minutes - making it the fastest-growing organism currently known^15,16. Since we expect the field of synthetic biology to increasingly incorporate V. natriegens into its work, we want to future-proof the rEvolver for V. natriegens compatibility in a couple of ways. Firstly, we will codon-optimise all of our genetic constructs for optimal expression in V. natriegens. Alongside this we will design a growth medium to facilitate fast growth of V. natriegens, using ‘Design of Experiments’ mathematical modelling to complement our work¹⁷. If directed evolution promises to compress evolutionary processes into research relevant timescales, then V. natriegens compatibility will prove invaluable for accelerating this further.

Testing MutaT7 System Via The ‘MutaT7 test cassette’

Before immediately attempting to assemble the ‘three plasmid system’ envisioned in our Project Description, we wanted to demonstrate targeted base deaminase activity at a specified gene of interest without compromising surrounding genome integrity in both E. coli and Vibrio natriegens. For this, we designed a codon reversion reporter assay, based on that of Moore et al. 2018³. This ‘MutaT7 test cassette’ (BBa_K4451021/BBa_K4451022) was assembled into a pBBRBB low-copy plasmid via NEB HiFi DNA Assembly.

The test cassette contains two silenced antibiotic resistance genes with ATA/ACG in place of an ATG start codon. Upon transcription of these genes by T7 RNA polymerase, the tethered base deaminase will cause DNA damage to bases within the boundaries of the T7 terminators. Streptomycin resistance phenotype is indicative of MutaT7 reconstituting the START codon via an A→G transition (for the ATA test cassette, induced by the adenine deaminase TadA18), or a C→T transition (for the ACG test cassette, induced by the cytosine deaminase evoAPOBEC1-BE4max¹⁹).

To show whether diversification is confined to the intended region, a silent chloramphenicol acetyltransferase gene is located immediately adjacent to one of the synthetic T7 terminators. Translation of chloramphenicol acetyltransferase and the resultant chloramphenicol resistance phenotype is therefore indicative of mutagenesis occurring outside of the flanking T7 promoters, and reconstitution of the ATG codon. Therefore, confirmation of targeted diversification would be shown by a streptomycin resistance phenotype, with retained sensitivity to chloramphenicol.

Preliminary results in the E. coli expression strain BL21 suggested that the silent start codons were indeed not translated, and a low (undetectable) rate of codon reversion due to background mutation, rather than targeted deaminase activity (although this rate would be expected to be far higher in cells lacking uracil and inosine repair pathways as intended in our three-plasmid system).

The hypermutation plasmid described in our project description was first assembled as two intermediate plasmids containing a single base deaminase-T7RNAP fusion gene (pJKR-TetR-CD-T7RNAP and pJKR-TetR-AD-T7RNAP), each under the control of an aTc-inducible promoter, which would allow us to tune the rate of hypermutation. The final construct (BBa_K4451019) could not be assembled before the project deadline, so we instead co transformed the test cassette with the intermediate plasmids BBa_K4451003 and BBa_K4451004, since both of these contained a full transcription unit for one of the chimeric proteins.

Although we encountered few problems cotransforming the pBBRBB-MutaT7 test cassette plasmid with the pJKR hypermutation plasmids into Keio collection E. coli K12 Δung cells, we found it exceptionally difficult to transform plasmids into the commercial V. natriegens strain Vmax Express (Synthetic Genomics Inc.) using either heat-shock or electroporation when following the manufacturer’s protocols. After talking to Daniel Stuckenberg (Human Practices), formerly of the Marburg 2018 team ‘VibriGens’, he agreed to send us a Δdns mutant of the V. natriegens type strain ATCC14048²⁰ for us to use for the remainder of the project, as this is reportedly far easier to transform. Sadly, these cells did not arrive in Sheffield before the project deadline, meaning that we were unable to demonstrate targeted diversification in V. natriegens.

PROTOCOLS

Plasmid Assembly

All gene fragments and oligonucleotides used in the project were synthesised by Integrated DNA Technologies and were resuspended following the manufacturer’s instructions (see below for a full list of primers). Plasmids were donated by the Hitchcock lab (School of Biosciences, University of Sheffield, UK) and were amplified using PCR into linear fragments to allow insertion of our constructs via HiFi assembly.

Overlaps suitable for HiFi assembly (New England Biolabs) were added through PCR in 50µl reactions as follows: 1µl of 10 ng/µl template DNA was combined with 2x 1µl 10uM forward and reverse primers, Q5 High-Fidelity 2X Master Mix (NEB), and 22µl nuclease-free water (33 cycles of 95℃ for 15s, X℃ for 15s, and 72℃ for Y s, followed by a final extension of 2 mins at 72℃ , where X and Y are determined by primer annealing temperature, and template length, respectively).

All PCR products were visualised via gel electrophoresis. Desired bands were gel-purified using the FastGene Gel/PCR Extraction Kit (Nippon Genetics). Amplified backbones were digested with 1µl 10U/µl Dpn I (Agilent) for 1 hour prior to HiFi assembly.

HiFi assembly itself was carried out according to the manufacturer’s instructions (we found that extended incubation times of 1-2 hours were required for our larger assemblies e.g. pJKR-evoAPOBEC1-T7RNAP). 2-5µl of assembled DNA was heat-shocked into NEB 5-alpha Competent E. coli, then spread onto LB agar with the appropriate antibiotic for selection and incubated overnight at 37℃.

Colonies were screened through colony PCR using MyTaq DNA polymerase (Meridian Bioscience). 10µl MyTaq Red Mix was added to 2x 2.5µl 10uM forward and reverse primers, and 9µl nuclease-free water. Colonies were transferred into the reaction tubes using a sterile pipette tip, then vortexed thoroughly. PCR conditions were run as described above, with the addition of a preceding 5 minute lysis stage at 95℃.

Positive transformants were grown overnight in 5ml LB (+ antibiotic) at 37℃, then treated with the Nippon Genetics FastGene Plasmid Mini Kit, following manufacturer’s protocols, to obtain purified plasmid DNA, which could be verified by Sanger sequencing (Eurofins Genomics).

For pBBRBB-MutaT7_Test(ATA) (BBa_K4451022), BBa_K4451021 was first constructed; the SmR false start codon was then converted from an ACG → ATA via QuickChange II Site-Directed Mutagenesis Kit (Agilent), following the manufacturer’s protocols.

Transformation of E. coli knockouts, and V. natriegens

Competent E. coli cells were made following the method described by Sambrook and Russell (2001). E. coli Keio collection knockout strains for ung, nfi, and the Keio collection parent strain, BW25113, were donated by the School of Biosciences, University of Sheffield. Individual colonies were grown overnight at 37℃. Cotransformation of pJKR-evoAPOBEC1-T7RNAP (BBa_K4451003) and pBBRBB-MT7_Test_ACG (BBa_K4451021) into chemically-competent Keio wild-type and Δung strains was performed by a 50-second heat-shock using 2µl of each plasmid, and selecting with kanamycin/ampicillin plates.

Electrocompetent V. natriegens was prepared as follows: 10ml Vmax Express cells (Synthetic Genomics Inc.) were grown overnight at 30℃ in e2YT media. 1ml of this culture was used to inoculate 100ml e2YT, which was then incubated at 37℃ with shaking at 200rpm until reaching an absorbance at 600nm of 0.5. The culture was split into two 50ml falcon tubes, incubated on ice for 15 mins, then centrifuged at 6500 rpm for 20 mins at 4℃. Cells were then washed to remove growth media which may interfere with electroporation; cell pellets were gently resuspended in 5ml electroporation wash buffer (680mM sucrose, 7mM K₂HPO₄), then made up to 35ml with the addition of further wash buffer, and finally spun at 6500 rpm for 15 mins. Pelleted cells were washed a further two times, then resuspended to OD600 = 16 (compared to electroporation buffer blank). Cells were frozen at -80℃ in 100μl aliquots.

Electroporation of Vmax Express cells was most successful when a pulse of 900V was used, followed by a 2hr recovery time in pre-warmed e2YT media.

IPTG Induction

IPTG induction was required for inducing GFP expression at different IPTG concentrations Overnight cultures were diluted in 50ml rich media (LB for E. coli or e2YT for V. natriegens) to OD600 0.05 and grown at 37℃ in a shaking incubator. Once cultures reached OD600 0.4-0.6, uninduced control samples were taken, and IPTG was added to the remaining culture (for pET21a-BBa_K4451026, this was 1mM final concentration). Cultures were then grown for a further 3hrs at 20℃. For pET21a-BBa_K4451026, 1ml of induced V. natriegens culture was concentrated to 100μl, then spread onto M9 minimal agar plates containing 0.2% glucose, 0.2% sorbitol, or neither (all plates had been spread prior with IPTG). Plates were incubated at 37℃ for 48hrs.

Growth Media

All E. coli strains were grown in LB-lennox media (10 g/L tryptone, 5 g/L yeast extract, 5 g/L NaCl) with appropriate antibiotics. Solid media was made with the addition of 1% w/v agar (bacteriological). V. natriegens was grown in liquid enhanced 2YT media (32 g/L tryptone, 20 g/L yeast extract, 17 g/L NaCl, 0.2% w/v glucose and 17.6 mM Na₂HPO₄), or on solid LB-Miller (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl).

M9 minimal agar for testing V. natriegens’ ability to utilise sorbitol as a carbon source was prepared as detailed in Lee et al. (2016) using 0.4% sorbitol (final w/v).

Primers Used In Our Study

Here is a list of ALL the primers used in our project!

1. Molina, R. S. et al. In vivo hypermutation and continuous evolution. Nat. Rev. Methods Primer 2, 1–22 (2022).

2. Morrison, M. S., Podracky, C. J. & Liu, D. R. The developing toolkit of continuous directed evolution. Nat. Chem. Biol. 16, 610–619 (2020).

3. Moore, C. L., Papa, L. J. & Shoulders, M. D. A Processive Protein Chimera Introduces Mutations across Defined DNA Regions In Vivo. J. Am. Chem. Soc. 140, 11560–11564 (2018).

4. Álvarez, B., Mencía, M., de Lorenzo, V. & Fernández, L. Á. In vivo diversification of target genomic sites using processive base deaminase fusions blocked by dCas9. Nat. Commun. 11, 6436 (2020).

5. Park, H. & Kim, S. Gene-specific mutagenesis enables rapid continuous evolution of enzymes in vivo. Nucleic Acids Res. 49, e32–e32 (2021).

6. Tizei, P. A. G., Csibra, E., Torres, L. & Pinheiro, V. B. Selection platforms for directed evolution in synthetic biology. Biochem. Soc. Trans. 44, 1165–1175 (2016).

7. Pérez-Ramos, A. et al. Characterization of the Sorbitol Utilization Cluster of the Probiotic Pediococcus parvulus 2.6: Genetic, Functional and Complementation Studies in Heterologous Hosts. Front. Microbiol. 8, 2393 (2017).

8. Roucourt, B. & Lavigne, R. The role of interactions between phage and bacterial proteins within the infected cell: a diverse and puzzling interactome. Environ. Microbiol. 11, 2789–2805 (2009).

9. Kumar, C. C. & Novick, R. P. Plasmid pT181 replication is regulated by two countertranscripts. Proc. Natl. Acad. Sci. 82, 638–642 (1985).

10. Brantl, S. & Wagner, E. G. H. Antisense RNA-mediated transcriptional attenuation: an in vitro study of plasmid pT181. Mol. Microbiol. 35, 1469–1482 (2000).

11. Takahashi, M. K. & Lucks, J. B. A modular strategy for engineering orthogonal chimeric RNA transcription regulators. Nucleic Acids Res. 41, 7577–7588 (2013).

12. Townshend, B., Xiang, J. S., Manzanarez, G., Hayden, E. J. & Smolke, C. D. A multiplexed, automated evolution pipeline enables scalable discovery and characterization of biosensors. Nat. Commun. 12, 1437 (2021).

13. Rogers, J. K. et al. Synthetic biosensors for precise gene control and real-time monitoring of metabolites. Nucleic Acids Res. 43, 7648–7660 (2015).

14. Farzadfard, F., Gharaei, N., Citorik, R. J. & Lu, T. K. Efficient retroelement-mediated DNA writing in bacteria. Cell Syst. 12, (2021).

15. Lee, H. H. et al. Vibrio natriegens, a new genomic powerhouse. 058487 Preprint at https://doi.org/10.1101/058487 (2016).

16. Weinstock, M. T., Hesek, E. D., Wilson, C. M. & Gibson, D. G. Vibrio natriegens as a fast-growing host for molecular biology. Nat. Methods 13, 849–851 (2016).

17. Gilman, J., Walls, L., Bandiera, L. & Menolascina, F. Statistical Design of Experiments for Synthetic Biology. ACS Synth. Biol. 10, 1–18 (2021).

18. Gaudelli, N. M. et al. Directed evolution of adenine base editors with increased activity and therapeutic application. Nat. Biotechnol. 38, 892–900 (2020).

19. Thuronyi, B. W. et al. Continuous evolution of base editors with expanded target compatibility and improved activity. Nat. Biotechnol. 37, 1070–1079 (2019).

20. Payne, W. J., Eagon, R. G. & Williams, A. K. Some observations on the physiology of Pseudomonas natriegens nov. spec. Antonie Van Leeuwenhoek 27, 121–128 (1961).