The capacity to diversify genetic codes advances our ability to understand and engineer biological systems1,2. A method for continuously diversifying user-defined regions of a genome would enable forward genetic approaches in systems that are not amenable to efficient homology-directed oligonucleotide integration. It would also facilitate the rapid evolution of biotechnologically useful phenotypes through accelerated and parallelized rounds of mutagenesis and selection, as well as cell-lineage tracking through barcode mutagenesis. Here we present EvolvR, a system that can continuously diversify all nucleotides within a tunable window length at user-defined loci. This is achieved by directly generating mutations using engineered DNA polymerases targeted to loci via CRISPR-guided nickases. We identified nickase and polymerase variants that offer a range of targeted mutation rates that are up to 7,770,000-fold greater than rates seen in wild-type cells, and editing windows with lengths of up to 350 nucleotides. We used EvolvR to identify novel ribosomal mutations that confer resistance to the antibiotic spectinomycin. Our results demonstrate that CRISPR-guided DNA polymerases enable multiplexed and continuous diversification of user-defined genomic loci, which will be useful for a broad range of basic and biotechnological applications.
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Ravikumar, A., Arzumanyan, G. A., Obadi, M. K. A., Javanpour, A. A. & Liu, C. C. Scalable continuous evolution of genes at mutation rates above genomic error thresholds. Preprint at https://www.biorxiv.org/content/early/2018/05/03/313338 (2018).
Esvelt, K. M., Carlson, J. C. & Liu, D. R. A system for the continuous directed evolution of biomolecules. Nature 472, 499–503 (2011).
Camps, M., Naukkarinen, J., Johnson, B. P. & Loeb, L. A. Targeted gene evolution in Escherichia coli using a highly error-prone DNA polymerase I. Proc. Natl Acad. Sci. USA 100, 9727–9732 (2003).
Ma, Y. et al. Targeted AID-mediated mutagenesis (TAM) enables efficient genomic diversification in mammalian cells. Nat. Methods 13, 1029–1035 (2016).
Hess, G. T. et al. Directed evolution using dCas9-targeted somatic hypermutation in mammalian cells. Nat. Methods 13, 1036–1042 (2016).
Wang, H. H. et al. Programming cells by multiplex genome engineering and accelerated evolution. Nature 460, 894–898 (2009).
Costantino, N. & Court, D. L. Enhanced levels of λ Red-mediated recombinants in mismatch repair mutants. Proc. Natl Acad. Sci. USA 100, 15748–15753 (2003).
Troll, C., Alexander, D., Allen, J., Marquette, J. & Camps, M. Mutagenesis and functional selection protocols for directed evolution of proteins in E. coli. J. Vis. Exp. 49, e2505 (2011).
de Boer, J. G. & Ripley, L. S. An in vitro assay for frameshift mutations: hotspots for deletions of 1 bp by Klenow-fragment polymerase share a consensus DNA sequence. Genetics 118, 181–191 (1988).
Jinek, M. et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337, 816–821 (2012).
Bambara, R. A., Uyemura, D. & Choi, T. On the processive mechanism of Escherichia coli DNA polymerase I. Quantitative assessment of processivity. J. Biol. Chem. 253, 413–423 (1978).
Sarkar, S., Ma, W. T. & Sandri, G. H. On fluctuation analysis: a new, simple and efficient method for computing the expected number of mutants. Genetica 85, 173–179 (1992).
Drake, J. W. A constant rate of spontaneous mutation in DNA-based microbes. Proc. Natl Acad. Sci. USA 88, 7160–7164 (1991).
Jagessar, K. L. & Jain, C. Functional and molecular analysis of Escherichia coli strains lacking multiple DEAD-box helicases. RNA 16, 1386–1392 (2010).
Slaymaker, I. M. et al. Rationally engineered Cas9 nucleases with improved specificity. Science 351, 84–88 (2016).
Minnick, D. T. et al. Side chains that influence fidelity at the polymerase active site of Escherichia coli DNA polymerase I (Klenow fragment). J. Biol. Chem. 274, 3067–3075 (1999).
Loh, E., Salk, J. J. & Loeb, L. A. Optimization of DNA polymerase mutation rates during bacterial evolution. Proc. Natl Acad. Sci. USA 107, 1154–1159 (2010).
Wang, Y. et al. A novel strategy to engineer DNA polymerases for enhanced processivity and improved performance in vitro. Nucleic Acids Res. 32, 1197–1207 (2004).
Salis, H. M. The ribosome binding site calculator. Methods Enzymol. 498, 19–42 (2011).
Funatsu, G., Schiltz, E. & Wittmann, H. G. Ribosomal proteins. XXVII. Localization of the amino acid exchanges in protein S5 from two Escherichia coli mutants resistant to spectinomycin. Mol. Gen. Genet. 114, 106–111 (1972).
Zheng, X., Xing, X.-H. & Zhang, C. Targeted mutagenesis: a sniper-like diversity generator in microbial engineering. Synth. Syst. Biotechnol. 2, 75–86 (2017).
Timms, A. R., Steingrimsdottir, H., Lehmann, A. R. & Bridges, B. A. Mutant sequences in the rpsL gene of Escherichia coli B/r: mechanistic implications for spontaneous and ultraviolet light mutagenesis. Mol. Gen. Genet. 232, 89–96 (1992).
Brocklehurst, P. Antibiotics for gonorrhoea in pregnancy. Cochrane Database of Systematic Reviews 2, CD000098 https://doi.org/10.1002/14651858.CD000098 (2002).
McKenna, A. et al. Whole-organism lineage tracing by combinatorial and cumulative genome editing. Science 353, aaf7907 (2016).
Koboldt, D. C. et al. VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 22, 568–576 (2012).
Hall, B. M., Ma, C.-X., Liang, P. & Singh, K. K. Fluctuation analysis CalculatOR: a web tool for the determination of mutation rate using Luria-Delbruck fluctuation analysis. Bioinformatics 25, 1564–1565 (2009).
Rosche, W. A. & Foster, P. L. Determining mutation rates in bacterial populations. Methods 20, 4–17 (2000).
Truniger, V., Lázaro, J. M., de Vega, M., Blanco, L. & Salas, M. phi 29 DNA polymerase residue Leu384, highly conserved in motif B of eukaryotic type DNA replicases, is involved in nucleotide insertion fidelity. J. Biol. Chem. 278, 33482–33491 (2003).
de Vega, M., Lazaro, J. M., Salas, M. & Blanco, L. Primer-terminus stabilization at the 3′-5′ exonuclease active site of phi29 DNA polymerase. Involvement of two amino acid residues highly conserved in proofreading DNA polymerases. EMBO J. 15, 1182–1192 (1996).
Ducani, C., Bernardinelli, G. & Högberg, B. Rolling circle replication requires single-stranded DNA binding protein to avoid termination and production of double-stranded DNA. Nucleic Acids Res. 42, 10596–10604 (2014).
Povilaitis, T., Alzbutas, G., Sukackaite, R., Siurkus, J. & Skirgaila, R. In vitro evolution of phi29 DNA polymerase using isothermal compartmentalized self replication technique. Protein Eng. Des. Sel. 29, 617–628 (2016).
Badran, A. H. & Liu, D. R. Development of potent in vivo mutagenesis plasmids with broad mutational spectra. Nat. Commun. 6, 8425 (2015).
Greener, A., Callahan, M. & Jerpseth, B. An efficient random mutagenesis technique using an E. coli mutator strain. Mol. Biotechnol. 7, 189–195 (1997).
We thank S. McDevitt at the University of California, Berkeley Vincent J. Coates Genomics Sequencing Laboratory for assistance with high-throughput sequencing, the Arkin laboratory for supplying E. coli strain RE1000, W. DeLoache for helping edit our manuscript, and the Innovative Genomics Institute for funding.
The Regents of the University of California have filed a provisional patent application (62/662,043 and 62/556,127) related to the technology described in this work to the United States Patent and Trademark Office; S.O.H. is listed as the inventor.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Previous tools enabling diversification of user-defined loci by substituting cytosines and guanines limit the protein coding space that can be explored4,5. This chart shows which amino acids can (green) and cannot (red) be reached by mutating cytosines and guanines to any other base for each of the 64 codons, highlighting that only 32% of missense mutations are achievable with targeted cytidine deaminases. The white area depicts the original amino acid identity.
Extended Data Fig. 2 The direction of EvolvR-mediated mutagenesis relative to the gRNA is dependent on which strand is nicked.
Our previous fluctuation analysis in Fig. 1e demonstrated that nCas9(D10A)–PolI3M mutates a window 3′ of the nick site. Here we directly tested whether mutations are generated 5′ of the nick site using a different gRNA. Because DNA polymerases synthesize in the 5′-to-3′ direction, we anticipated that nCas9(D10A)–PolI3M would not provide an elevated mutation rate 5′ of the nick site. We indeed found that expressing a guide RNA which targeted nCas9(D10A)–PolI3M to nick 16 nucleotides 3′ from the nonsense mutation (indicated by a red cross) did not show targeted mutagenesis. We hypothesized that we could induce targeted mutagenesis using the same gRNA by using a Cas9 variant harbouring the H840A mutation, which nicks the DNA strand non-complementary to the gRNA, rather than the D10A mutation, which nicks the strand complementary to the gRNA. nCas9(H840A)–PolI3M increased the mutation rate 16 nucleotides 3′ from the nick by 52-fold compared to the global mutation rate of cells expressing an off-target gRNA. We used the D10A nCas9 variant for all subsequent experiments. Data are mean ± 95% confidence intervals from ten biologically independent samples. *P < 0.0001; two-sided Student’s t-test.
Extended Data Fig. 3 PolI5M elevates mutation rates 1 nucleotide, but not 11 nucleotides, from the nick compared to PolI3M.
PolI3M with additional F742Y and P796H mutations (PolI5M) elevates the mutation rate 33-fold 1 nucleotide from the nick compared to PolI3M. PolI5M did not have a higher mutation rate than PolI3M 11 nucleotides from the nick. Data are mean ± 95% confidence intervals from ten biologically independent samples. *P < 0.0001; two-sided Student’s t-test.
Extended Data Fig. 4 Fusing a highly processive DNA polymerase to enCas9 increases the target window length.
PolI was exchanged for a more processive and higher-fidelity bacteriophage Phi29 DNA polymerase (Phi29). Owing to Phi29 not having a flap endonuclease, residues 1–325 of PolI were inserted between enCas9 and Phi29. Using gRNAs targeting different distances from the nonsense mutation, we found that Phi29 with two previously reported fidelity-reducing mutations (N62D and L384R) elevated the mutation rate 56 nucleotides from the nick compared to the global mutation rate28,29. When we expressed Phi29’s single-stranded binding protein (ssb), which is known to improve the activity of Phi29, we observed an elevation in the targeted mutation rate30. Finally, because the activity of Phi29 is known to decrease at temperatures above 30 °C and the fluctuation analysis was performed at 37 °C, we added mutations previously reported to improve the thermostability of Phi29 (iPhi29) and observed a targeted mutation rate 347 nucleotides from the nick site that was significantly greater than the global mutation rate31. Unfortunately, mutations decreasing Phi29’s fidelity are known to decrease its processivity explaining our inability to identify Phi29 variants that retain high processivity while offering as high of a mutation rate as PolI3M28. Data are mean ± 95% confidence intervals from ten biologically independent samples. *P < 0.0001; two-sided Student’s t-test.
Extended Data Fig. 5 Removing internal ribosome binding sequences decreases EvolvR-mediated off-target mutagenesis.
enCas9–PolI3M–TBD was codon optimized to remove strong ribosome binding sites in the EvolvR coding sequence that were predicted to produce an untethered DNA polymerase. The off-target mutation rate decreased 4.14-fold when expressing enCas9–PolI3M–TBD-CO compared to enCas9–PolI3M–TBD (P = 0.000482) whereas the on-target mutation rate only decreased 1.23-fold. Data are mean ± 95% confidence intervals from ten biologically independent samples. *P < 0.0001; two-sided student’s t-test.
Extended Data Fig. 6 EvolvR-mediated mutagenesis can be coupled with a non-selectable genetic screen.
a, To test the capability for coupling EvolvR-mediated mutagenesis with a non-selectable genetic screen, we designed a target plasmid containing a GFP cassette with an early termination codon in the GFP coding sequence (pTarget-GFP*). After co-transforming pEvolvR with pTarget-GFP* and growing for 24 h, we analysed and sorted the GFP-positive fraction. In the two replicates expressing an off-target gRNA, we did not detect or sort any GFP cells. By contrast, for the two replicates expressing a gRNA nicking four nucleotides away from the chain-terminating mutation in the coding sequence of GFP, we found that 0.06% and 0.07% of the total cells were GFP positive. These results agree with sequencing outcomes from Fig. 1b, which showed that expressing nCas9–PolI3M for 24 h produces substitutions in the target region at frequencies between 0.5% to 1%. b, After culturing the sorted populations, both replicates expressing an off-target gRNA did not show growth, whereas both replicates expressing the on-target gRNA grew bright green.
Extended Data Fig. 7 EvolvR enables targeted genome diversification without affecting viability or growth rate.
a, The viability of TG1 E. coli expressing EvolvR targeted to the essential rpsE gene was significantly higher than TG1 E. coli transformed with the MP6 plasmid and induced with 25 mM arabinose and 25 mM glucose (a previously developed plasmid for continuous non-targeted mutagenesis32, P = 0.0108) as well as XL1-Red E. coli (a previously developed strain for continuous non-targeted mutagenesis33, P = 0.0105). Viability was measured relative to TG1 E. coli transformed with an empty control plasmid. Data are mean ± s.d. from three biologically independent samples. *P < 0.05; two-tailed t-test. b, TG1 E. coli transformed with an empty control plasmid and TG1 E. coli transformed with pEvolvR targeting the rpsE gene resulted in similar growth curves whereas XL1-Red E. coli and TG1 E. coli transformed with MP6 plasmid and induced with 25 mM arabinose and 25 mM glucose grew much slower and saturated at lower final optical densities. Shaded area represents mean ± s.d. from three biologically independent samples. c, The spectinomycin-resistant CFUs per ml saturated culture of TG1 E. coli targeting EvolvR to the rpsE gene was significantly higher than XL1-Red E. coli (P = 0.022) and TG1 E. coli transformed with MP6 plasmid and induced with 25 mM arabinose and 25 mM glucose (P = 0.0049). Data are mean ± s.d. from three biologically independent samples. *P < 0.05; two-tailed t-test.
Extended Data Fig. 8 EvolvR-mediated mutagenesis performs better than a previous non-targeted diversification technique.
To compare the performance of EvolvR and the previously developed non-targeted mutagenesis plasmid MP6 in screen-based directed evolution applications, we co-transformed pEvolvR (enCas9–PolI3M–TBD) or MP6 with a target plasmid containing a GFP cassette with an early termination codon in the GFP coding sequence (pTarget-GFP*). The cultures expressing EvolvR were grown for 24 h and the MP6 cultures followed a two day growth–induction protocol as previously described. Flow cytometry revealed that cultures expressing EvolvR and an on-target gRNA resulted in 28-fold more GFP-positive cells than MP6 cultures.
Extended Data Fig. 9 Locations of gRNA targets relative to the rpsE gene and mutations in ribosomal protein S5 that confer spectinomycin resistance.
a, enCas9–PolI3M–TBD was targeted to five dispersed loci in the endogenous rpsE gene using gRNAs that nick after the 119th, 187th, 320th, 403rd or 492nd base pair of the 504-bp rpsE coding sequence. The locations of the previously identified rpsE mutations that provide spectinomycin resistance are coloured orange, and the region where we identified new spectinomycin-resistance mutations is highlighted in red. b, The mutations that we discovered confer spectinomycin resistance would be expected to move Lys26 (which is predicted to hydrogen bond with spectinomycin) relative to the spectinomycin-binding pocket. We hypothesized that mutations that move Lys26 relative to the spectinomycin-binding pocket remove that hydrogen bond and destabilize the interaction of spectinomycin with the ribosome, thereby conferring spectinomycin resistance. c, Therefore, we tested whether deleting any single amino acid between residues 16 and 35 confers spectinomycin resistance. We found that deleting residues 23, 24, 25, 26, 27 or 28 provides spectinomycin resistance whereas deleting any of the residues between 16 and 22 or 29 and 35 does not. These results support the hypothesis that one mechanism of resistance to spectinomycin is disruption of the interaction between Lys26 and spectinomycin. Data are mean ± s.d. from three biologically independent samples.
About this article
Cite this article
Halperin, S.O., Tou, C.J., Wong, E.B. et al. CRISPR-guided DNA polymerases enable diversification of all nucleotides in a tunable window. Nature 560, 248–252 (2018). https://doi.org/10.1038/s41586-018-0384-8
Journal of Microbiology (2020)
Various strategies of effector accumulation to improve the efficiency of genome editing and derivative methodologies
In Vitro Cellular & Developmental Biology - Animal (2020)
Conservation Biology (2020)
BMC Plant Biology (2020)