A bacterial gene-drive system efficiently edits and inactivates a high copy number antibiotic resistance locus

Gene-drive systems in diploid organisms bias the inheritance of one allele over another. CRISPR-based gene-drive expresses a guide RNA (gRNA) into the genome at the site where the gRNA directs Cas9-mediated cleavage. In the presence of Cas9, the gRNA cassette and any linked cargo sequences are copied via homology-directed repair (HDR) onto the homologous chromosome. Here, we develop an analogous CRISPR-based gene-drive system for the bacterium Escherichia coli that efficiently copies a gRNA cassette and adjacent cargo flanked with sequences homologous to the targeted gRNA/Cas9 cleavage site. This “pro-active” genetic system (Pro-AG) functionally inactivates an antibiotic resistance marker on a high copy number plasmid with ~ 100-fold greater efficiency than control CRISPR-based methods, suggesting an amplifying positive feedback loop due to increasing gRNA dosage. Pro-AG can likewise effectively edit large plasmids or single-copy genomic targets or introduce functional genes, foreshadowing potential applications to biotechnology or biomedicine.


Supplementary Text
Since including GFP cargo sequences in the Pro-AG system for insertion into a high copy number plasmid (Fig. 3b) had no obvious impact on the copying efficiency relative to that observed with the minimal gRNA-only cassette (Figs. 2b and 3b), we speculated that this system also might be adapted for inserting cargo into single-copy sequences in the bacterial chromosome. Such enhancement of currently available precise genome editing tools18 could be useful for a wide variety of bacterial engineering applications.
We constructed an analogous set of plasmids to those described above for evaluating CRISPR-control versus Pro-AG protocols in the context of editing the lacZ chromosomal target gene ( Supplementary Fig. 10a). Comparing efficiencies of copying a minimal "gRNA-lacZ-only" versus a larger "gRNA-lacZ:GFP" cassette into the lacZ target site ( Supplementary Fig. 10a, Supplementary Table 1), we followed a similar selection and induction procedure described above for the three plasmid strategy ( Supplementary Fig.   1). In the context of the single-copy lacZ chromosomal target, induction of Cas9 following initial bacterial transformation resulted in a marked decrease in viable E. coli CFU following either the CRISPR-control or Pro-AG protocols ( Supplementary Fig.   10b). These profound reductions in recovered CFU are consistent with previously reported findings on Cas9-induced cleavage of single copy genomic targets that presumably reflect the lethal effects of unrepaired chromosomal double-stranded breaks19. Similarly, overnight cultures seeded from single colonies achieved significantly lower cell densities following induction of Cas9 alone or together with λRed than observed without Cas9 induction ( Supplementary Fig. 10c). Consistent with the lethal DNA-break hypothesis, induction of the λRed DNA repair cassette together with Cas9 significantly increased CFU recovery ( Supplementary Fig. 10d). Following induction of Cas9 alone, blue colonies comprised the overwhelmingly phenotype recovered on IPTG/X-Gal plates (intact lacZ gene function) ( Supplementary Fig. 10e), suggesting that Cas9 had not acted on the target sequences in these escaper colonies. Indeed, all tested CRISPR and Pro-AG induced blue colony escapers recovered following induction of Cas9 alone carried fully intact lacZ target sequences ( Supplementary Fig. 10a, bottom scheme). Induction of Cas9 + λRed applied to the CRISPR-control regimen only rarely produced white colonies (~2% of total colonies recovered, Supplementary Fig.   11a). In such white colonies, we were unable to amplify lacZ target sequences by PCR, perhaps reflecting large Cas9-induced deletion events ( Supplementary Fig. 6b, top scheme), as observed in prior studies20. In contrast, induction of Cas9 + λRed in the Pro-AG and Pro-AGFP regimens led to recovery of ~90% white colonies, suggesting lacZ inactivation by precise insertion of the homology flanked DNA cassettes ( Supplementary Fig. 6a). Indeed, all analyzed white colonies contained perfect genedisrupting insertions of the gRNA-lacZ-only or gRNA-lacZ:GFP cargo cassettes into the gRNA-lacZ directed Cas9 cleavage site in the genomic lacZ locus ( Supplementary Fig.   11b, middle and bottom schemes, respectively), mirroring prior results with the high copy plasmid. Because the GFP coding sequences were designed to be in-frame with lacZ, these Pro-AGFP edited bacteria also expressed the fluorescent GFP marker upon simultaneous induction of the lac operon with IPTG and activation of Cas9 + λRed (Supplementary Figs. 11c and d). We conclude that the Pro-AG system is well suited for high efficiency precision editing and delivery of DNA cargo into a chromosomal locus.
Sequencing of the gRNA constructs from Pro-AG escaper (blue) colonies following Cas9 + λRed induction revealed deletions in the tet operator region and sequence insertion downstream HA1 in 100% of clones analyzed ( Supplementary Fig. 7b), suggesting escape dependent on inactivation of the gRNA donor plasmid.
Dosage amplification of gRNA2 accompanied its copying from its low-copy number launch plasmid to its high-copy plasmid targets, and experiments in which the gRNA component was either included in the amplifying cassette (gRNA-In) or left outside of the cassette (gRNA-Out) revealed that the enhanced efficiency of the Pro-AG system depended on the gRNA being placed within the homology arms cassette (Fig. 3a). In contrast, when targeting a single-copy E. coli chromosomal locus, we hypothesized that the absence of gRNA amplification in this context would translate to equal efficiencies of the gRNA-in and gRNA-out configurations. We compared Pro-AGFP (gRNA-In) versus gRNA-out configurations for insertion of the GFP cargo gene into E. coli chromosomal lacZ locus ( Supplementary Fig. 12a) and observed that both configurations performed similarly, with no significant differences in CFUs recovered following parallel editing protocols ( Supplementary Fig. 12b). Induction of Cas9 + λRed applied to gRNA-Out configuration yielded more >90% edited colonies ( Supplementary Fig. 12c), with perfect insertion of GFP cargo into Cas9 cleavage site in the genomic lacZ locus at similar efficiency to the Pro-AGFP (gRNA-In) strategy ( Supplementary Fig. 12d).

Supplementary Figures
Supplementary Fig. 1 Pro-AG-mediated gene-editing and CRISPR-Control. Scheme of the method used for editing Amp resistance cassette from multi-copy plasmid through CRISPR-control a or Pro-AG b editing configurations Escherichia coli MG1655 cells were transformed with pET, pCas9 and pCRISPR or pPro-AG derivative plasmids in a or b, respectively. Cells were selected on LB plates containing triple antibiotic selection: ampicillin, spectinomycin and chloramphenicol (+Amp+Spm+Cm) in the presence or in the absence of anhydrotetracycline (aTc, Cas9 induction). Following transformation, single colonies were selected for an editing round, where bacteria was grown over night in LB broth in the presence of Amp and under induction of Cas9 with aTc. Additionally, in Pro-AG configuration b, recombinogenic Red enzymes were induced with the addition of L-arabinose. Finally, aliquots were diluted and plated for selection and quantification of Amp or Gm resistant colonies in each editing configuration.
Supplementary Fig. 2 Escherichia coli MG1655 colony counts and cell density are not affected after transformation and plating with Cas9 switched off or on. a Quantification of Ampicillin resistant E. coli expressing CRISPR-control configuration that mediates Cas9 targeting of bla gene. Colony forming units (CFU) were quantified on ampicillin agar plates in the presence (+aTc, blue dots) or in the absence (-aTc, black dots) of the Cas9 inducer anhydrotetracycline. b Optical densities measured at 600 nanometers (O.D 600nm) from E. coli over-night cultures grown in the presence (+aTc, blue bars) or in the absence (-aTc, black bars) of anhydrotetracycline after CRISPR-control-mediated editing round AmpgRNA1 or AmpgRNA2 were expressed. Supplementary Fig. 3 Number of Escherichia coli MG1655 colonies and cell density is not affected after transformation and overnight culture steps, respectively under CRISPR-control or Pro-AG editing configurations. a Quantification of Ampicillin resistant E. coli after transformation with the three plasmids system for CRISPR-control or Pro-AG editing configurations, that mediate Cas9 targeting of bla gene. Colony forming units (CFU) were quantified on LB ampicillin, spectinomycin and chloramphenicol agar plates and in the presence (+ aTc, blue dots) or in the absence (-aTc, black dots) of the Cas9 inducer anhydrotetracycline. b Optical densities (O.D.) measured at 600 nanometers from single E. coli colonies over-night cultures grown in the presence (+aTc, blue bars) or in the absence (-aTc, black bars) of anhydrotetracycline, and in combination of Red enzymes induction with arabinose and Cas9 with aTc (+aTc +arab, red bars) after CRISPR-control or Pro-AG mediated editing rounds.
Supplementary Fig. 4 Cas9 and Red induction with anhydrotetracycline and arabinose, respectively, do not affect Escherichia coli viability. a Quantification of E. coli colony forming units (CFU) on LB agar plates with (+Amp) or without (-Amp) after CRISPRcontrol or Pro-AG editing rounds when cells were grown in the presence (blue dots) or in the absence (black dots) of anhydrotetracycline (aTc) or in combination of aTc and arabinose (red dots). b Quantification of E. coli ampicillin resistant colony forming units (CFU) on LB ampicillin agar plates after CRISPR-control or Pro-AG editing rounds when cells were grown in the presence (+, red dots) or in the absence (-, black dots) of arabinose (arab), and under triple antibiotic selection, corresponding to the three plasmids used in our editing protocols.
Supplementary Fig. 5 RecA is partially involved in the mechanism for E. coli escape from Pro-AG-mediated editing. a Comparison of CFU selection from wild type E. coli MG1655 (WT) vs. MG1655 isogenic recA mutant (ΔRecA) strains on ampicillin (Amp, filled dots) or gentamicin (Gm, open dots) plates following Pro-AG-mediated targeting of the dual antibiotic resistant target plasmid pETg, from single colonies grown in the absence (-aTc, black dots), in the presence (+aTc, blue dots) of anhydrotetracycline for induction of Cas9 only, or in combination with arabinose for induction of both Cas9 and λRed (+aTc+arab, red dots). b DNA sequence analysis of plasmids isolated from single ΔRecA colonies in panel a recovered from the Pro-AG regimen (GmR plates; blue shaded box). All 12 Pro-AG-derivative clones analyzed carried a perfect insertion of the homology flanked gRNA2 expression cassette into the bla gene (zoom-in bottom scheme). The gRNA expression cassette is composed of the gRNA scaffold (purple), 20 bp gRNA targeting sequences (pink and black), and the constitutive tet promoter (grey). Also indicated are the Cas9 cleavage site, the protospacer adjacent motif (PAM), and homology arms (HA1, dark yellow box and HA2, light yellow box) that flank the gRNA2 cleavage site in the bla target gene carried on the pETg plasmid. Data in panel a are plotted as the mean ± s.e.m and analyzed by Student's t-test. **(P<0.01); **** (P<0.0001).
Supplementary Fig. 6 Sequence analysis of gRNA donor plasmids from Pro-AG escapers. Schematic of pPro-AG (Amp) a and pPro-AG (lacZ) b plasmids. Homology arms 1 (HA1, orange) and 2 (HA2, yellow) flanking bla a and lacZ b gRNAs, replication origin (pSC101ori, red), tet operator (cyan), gRNA (pink arrow), gRNA scaffold (turquoise arrow) and spectinomycin resistant gene (SpmR, grey arrow) are indicated. a Sequencing analysis of pPro-AG (Amp) plasmids from 20 escaper colonies of Pro-AG editing bla plasmid-encoded shows deletions (red lines) across the plasmid sequence in more than 50% of the plasmids. The remaining plasmids analyzed show intact wild type sequences (continuous black lines). b Sequencing analysis of pPro-AG (lacZ) plasmids from 10 escaper colonies of Pro-AG editing chromosomal lacZ gene shows similar pattern of tet operator deletion (red line) and sequence insertion (*). Length of insertion (discontinuous black line) and deletions (discontinuous red line) is indicated (bp). The 20 bp upstream and downstream specific insertions or deletions is also shown from the coding DNA strand (5'-3').
Supplementary Fig. 7 Differences in antibiotic resistance gene inactivation between Pro-AG and CRISPR-control configurations are partially dependent on gRNA levels. a Schematic of CRISPR-control and Pro-AG-mediated editing configurations of the betalactamase gene (bla) encoded on high-copy plasmids (pETg-CRISPR and pETg-Pro-AG) conferring resistance to ampicillin (AmpR, tan arrow) in E. coli MG1655. CRISPRcontrol configuration expresses gRNA2 targeting bla gene from the same pETgCRISPR target plasmid. Pro-AG configuration expresses the gRNA2 cassette flanked by bla (AmpR) homology arms (HA1 and HA2) that directly abut the gRNA2 cleavage site from the same pETg-Pro-AG target plasmid. pETg-CRISPR and pETg-Pro-AG plasmids also express a gentamicin resistance marker (GmR, green arrow). Cells were also transformed with a low copy plasmid, pCas9, maintained under chloramphenicol (CmR) selection, encodes Cas9 under control of an anhydrotetracycline (aTc) inducible promoter, and with a second low copy plasmid, pΔgRNA, maintained under spectinomycin (Spm) selection, encodes λRed under control of an arabinose (arab) inducible promoter. b Top panel. Selection for E. coli CFU on ampicillin (Amp, filled dots) or gentamicin (Gm, open dots) plates following CRISPR-control-or Pro-AGmediated targeting of the dual antibiotic resistant target plasmids pETg-CRISPR and pETg-Pro-AG, respectively, from single colonies grown in the absence (-aTc, black dots), in the presence (+aTc, blue dots) of anhydrotetracycline for induction of Cas9 only, or in combination with arabinose for induction of both Cas9 and λRed (+aTc+arab, red dots). Bottom panel. DNA sequence analysis of plasmids isolated from single colonies in top panel recovered from the Pro-AG regimen (GmR plates; blue shaded box). All 10 Pro-AG-derivative clones analyzed carried plasmids target with a perfect insertion of the homology flanked gRNA2 expression cassette into the bla gene (zoomin bottom scheme). The gRNA expression cassette is composed of the gRNA scaffold (purple), 20 bp gRNA targeting sequences (pink and black), and the constitutive tet promoter (grey). Also indicated are the Cas9 cleavage site, the protospacer adjacent motif (PAM), and homology arms (HA1, dark yellow box and HA2, light yellow box) that flank the gRNA2 cleavage site in the bla target gene carried on the pETg-Pro-AG plasmid. Data in panel b are plotted as the mean ± s.e.m and represent three independent experiments. Data analyzed by Student's t-test. N.S, not significant (P>0.05); ***(P<0.001); **** (P<0.0001).
Supplementary Fig. 8 Pro-AG acts via a self-amplifying mechanism with a copied gRNA that remains functional on further editing events. a Recovery of spectinomycin resistant (SpmR) E. coli CFU from single CRISPR-control or Pro-AG edited colonies and following growth at 30°C (black dots) or 37°C (white dots). b Selection for E. coli CFU on ampicillin plates from single CRISPR-control or Pro-AG edited colonies and following growth at 30°C (filled dots) or 37°C (open dots) in the absence (-aTc, black dots) or in the presence of anhydrotetracycline in combination with arabinose (+aTc+arab, red dots). Bacteria growth at 30°C and 37°C conditions are permissive and non-permissive for gRNA-harboring plasmids replication, respectively. Data in panel b are plotted as the mean ± s.e.m and analyzed by Student's t-test. N.S, not significant (P>0.05); ***(P<0.001); **** (P<0.0001).
Supplementary Fig. 9 Pro-AG configuration is well suited for manipulation of large plasmids. a Schematic of CRISPR-control and Pro-AG-mediated configurations for inactivation of beta-lactamase gene (bla) encoded on long cosmid vector (Super-Cos SV3B05) conferring resistance to ampicillin (AmpR, tan arrow) in E. coli MG1655. Super-Cos SV3B05 cosmid also expresses a kanamycin resistance marker (KmR, green arrow). Cells were also transformed with a low copy plasmid, pCas9, maintained under chloramphenicol (CmR) selection, encodes Cas9 under control of an anhydrotetracycline (aTc) inducible promoter. CRISPR control configuration expresses gRNA2 targeting bla gene from pCRISPR Amp plasmid. Pro-AG configuration expresses the gRNA2 cassette flanked by bla (AmpR) homology arms (HA1 and HA2) that directly abut the gRNA2 cleavage site from pPro-AG Super-Cos plasmid. Also carried on plasmids pCRISPR Amp and pPro-AG Super-Cos are the recombinogenic Red enzymes (Red), which can be induced with L-arabinose (arab). b Top panel. Selection for E. coli CFU on ampicillin (Amp, filled dots) or kanamycin (km, open dots) plates following CRISPR-control-or Pro-AG-mediated targeting of the dual antibiotic resistant target cosmid Super-Cos SV3B05, from single colonies grown in the absence (-aTc, black dots), in the presence (+aTc, blue dots) of anhydrotetracycline for induction of Cas9 only, or in combination with arabinose for induction of Cas9 and λRed (+aTc+arab, red dots). Bottom panel. DNA sequence analysis of plasmids isolated from single colonies in top panel recovered from the Pro-AG regimen (KmR plates; blue shaded box). All 10 Pro-AG-derivative clones analyzed carried a perfect insertion of the homology flanked gRNA2 expression cassette into the bla gene (zoom-in bottom scheme). The gRNA expression cassette is composed of the gRNA scaffold (purple), 20 bp gRNA targeting sequences (pink and black), and the constitutive tet promoter (grey). Also indicated are the Cas9 cleavage site, the protospacer adjacent motif (PAM), and homology arms (HA1, dark yellow box and HA2, light yellow box) that flank the gRNA2 cleavage site in the bla target gene carried on the Super-Cos SV3B05 cosmid. Data in panel b are plotted as the mean ± s.e.m and analyzed by Student's t-test. N.S, not significant (P>0.05); **** (P<0.0001).
Supplementary Fig. 10 Comparison between CRISPR-control and Pro-AG editing configurations when targeting chromosomal Escherichia coli lacZ gene. a Schematic of the three plasmids used for editing the genomic lacZ locus: CRISPR-control (pCRISPRlacZ); gRNA-only or Pro-AG (pPro-AGlacZ); and gRNA + GFP or Pro-AGFP (pPro-AGFPlacZ). Precise insertion of homology flanked cassettes would inactivate the lacZ chromosomal E. coli locus (lacZ, blue arrow) and in the case of the pPro-AGFPlacZ construct, result in in-frame fusion with GFP. b Quantification of E. coli colony forming units (CFU) following initial transformation with the corresponding CRISPR-control or Pro-AG editing configuration plasmids. CFUs were quantified on LB spectinomycin and chloramphenicol agar plates and in the presence (+aTc, blue dots) or in the absence (-aTc, black dots) of the Cas9 inducer anhydrotetracycline. c Optical densities measured at 600 nanometers (O.D 600nm) from single E. coli colonies after transformation and over night cultures when cells were grown in the absence (-aTc, black bars), or presence of anhydrotetracycline (+aTc, blue bars) or in combination of aTc and arabinose (+aTc+arab, red bars) compared under the following configurations: CRISPR, Pro-AG, or Pro-AGFP. d Quantification of E. coli colony forming units (CFU) after CRISPR-control, Pro-AG and Pro-AGFP mediated editing of lacZ loci. Cells were grown with (+aTc, blue dots) or without (-aTc, black dots) Cas9 induction with anhydrotetracycline, or in combination of Cas9 and Red induction with aTc and arabinose, respectively (+aTc +arab, red dots). e Appearance of dilutions spots carrying isolated colonies obtained under CRISPR-control (top) versus Pro-AG (bottom) configurations and after induction of Cas9 only. Data in panels b, c and d are plotted as the mean ± s.e.m and represent three independent experiments. Data analyzed by Student's t-test. ** (P<0.01); **** (P<0.0001).