Why DNA-cutting counter-selection stalls recovery
Conventional CRISPR counter-selection works by cutting DNA. Cas9 or Cas12a is guided to a sequence that unedited cells still carry, and the resulting double-strand break removes them. In most prokaryotes, which lack robust non-homologous end joining, that break is effectively irreversible and fatal. The paper frames this as a double-edged sword: selection is strict, but survival pressure is excessive.
Two consequences follow, and the paper names both. The number of recoverable transformants drops sharply, a transformation bottleneck. And cells that inactivate the CRISPR machinery can survive as escaper mutants. Prior fixes attenuated nuclease activity, for example by splitting the guide RNA or mutating the crRNA repeat, but the authors describe these as a delicate trade-off: too much attenuation weakens selection, while any reliance on genomic cleavage still imposes physiological stress.
Cas12a2 turns a target transcript into a growth arrest
SURE, short for SuCas12a2-Universal Recombination Editing, uses a different trigger. Cas12a2 recognizes target RNA transcripts rather than DNA. Once activated, it unleashes collateral cleavage that degrades RNA, single-stranded DNA, and double-stranded DNA. The design intent is that unedited cells, which still express the target mRNA, enter an abortive infection-like state of dormancy or growth arrest rather than immediate death.
Edited cells that have removed the target sequence no longer produce the activating transcript, so they avoid the burden and gain a growth advantage. The authors describe this as a soft selection pressure meant to keep screening stringent without the transformation bottleneck caused by genomic destruction. The key distinction for readers: the selective agent is a transcript threshold, not a DNA break.
Up to 1,000-fold more transformants than DNA cutters
The headline comparison is transformation efficiency. The paper reports an increase of up to 1,000-fold compared with Cas9- or Cas12-based systems in E. coli MG1655. In vitro interference assays compared SuCas12a2 against FrCas9 and LbCas12a targeting the same sites, counting transformants per millilitre across inducer concentrations and normalizing to non-target controls.
This is a comparison of recoverable colony counts, not of editing purity. The two measures answer different questions: transformation efficiency reflects how many cells survive and can be recovered, while editing efficiency reflects what fraction of recovered clones carry the intended change. The paper's claim rests on improving the first without collapsing the second, and the editing numbers are reported separately.
Over 90% editing after tuning spacer and homology arms
The initial lacZ deletion protocol reached only about 40% efficiency, so the authors tuned parameters. A 30-nucleotide spacer outperformed 24, 35, and 40-nucleotide variants. Extending homology arms from 200 bp to 1,000 bp improved recombination, but pushing to 1,500 bp sharply reduced knockout efficiency for the 500 bp target fragment; the authors attribute this to limited recombinase protection of long single-stranded intermediates and to replication stalling, and set roughly 1,000 bp as the practical optimum. The λ-Red system outperformed RecET.
With these settings, SURE reached 90% efficiency for lacZ deletion, 85% for mCherry insertion, and 90% for replacing lacZ with mCherry. At other chromosomal loci, ahpC, katG, and ackA, deletions exceeded 30%. The gap between the lacZ locus and these other sites matters: the authors link it to transcription level, since transcriptome sequencing showed lacZ was transcribed far more highly than the other tested loci.
Weakly transcribed targets weaken the selection
Because activation depends on target mRNA, editing efficiency is tethered to transcriptional activity. The paper states directly that applying SURE to poorly transcribed genes, strictly non-coding RNAs, or structurally silent genomic regions presents a significant challenge, since basal transcript abundance may be insufficient to trigger the abortive infection state. This is the mechanism's own boundary, not an incidental caveat.
The authors propose future fixes rather than demonstrating them: engineered sensitized Cas12a2 variants, or coupling SURE with a dCas12f-σE machinery to drive promoter-independent activator RNA production. Readers should treat these as proposed directions, not validated results.
Double-stranded DNA degradation is the decisive activity
To find which activity actually drives selection, the authors built four Cas protein variants with mutations in the REC and RuvC domains and compared them with wild type. A fluorescence assay used a FAM-BHQ-labeled dsDNA reporter: cleavage separates fluorophore from quencher and raises signal, and relative activities were normalized to maximum initial velocities from the linear slope of the fluorescence increase.
The paper identifies collateral double-stranded DNA degradation as the decisive factor for efficient selection. This is a mechanistic claim about which cleavage output correlates with counter-selection strength, and it is what separates SURE from a purely RNA-degrading design.
Seed mismatches abolish selection but 10 kb deletions falter
Specificity was tested with 15 crRNA variants carrying consecutive double-base mismatches across the spacer. Tolerance was strongly position-dependent. Double mismatches in the core seed region, positions 3 to 10, almost completely abolished counter-selection, which supports target specificity. Mismatches at positions 1 to 2 next to the protospacer flanking sequence, or at positions 11 to 12, retained strong selection pressure, and mismatches beyond position 12 had progressively less impact. The practical reading is that seed-region design carries most of the off-target protection.
Scale is the other boundary. Using droplet digital PCR for absolute quantification, 500 bp, 1 kb, and 5 kb deletions were excised with over 90% efficiency, but a 10 kb fragment dropped to 4.7%. The authors attribute this decay to length-dependent instability of single-stranded DNA intermediates in the λ-Red machinery rather than to Cas12a2, and suggest sequential editing as a workaround. Whole-genome sequencing of a clone with a 1,000 bp lacZ deletion showed a read-depth drop confined to positions 365,250 to 366,249, with no unintended large deletions, rearrangements, or off-target editing detected elsewhere.
Three species edited and a plastic monomer consumed
The platform was extended beyond E. coli MG1655 to Bacillus subtilis BSZR and Shewanella oneidensis MR-1. For MR-1, the all-in-one knockout plasmid was introduced by electroporation; for B. subtilis, a xylose-inducible comK strain was used to induce natural competence. A single-plasmid design placed λ recombinase expression under arabinose, rhamnose, or IPTG control alongside the CRISPR-Cas components.
The applied demonstration engineered E. coli MG1655 to assimilate ethylene glycol, a plastic monomer, by integrating the NAD+-dependent etgB gene. The strain grew with ethylene glycol as the sole carbon source in M9 medium, and degradation was quantified by HPLC, including ethylene glycol derived from alkaline-hydrolysed PET after terephthalic acid removal. This shows the editing output feeding a metabolic phenotype, not just a genotype.
Sources used in this answer
Repurposing Cas12a2-mediated abortive infection for efficient and low-toxicity microbial genome editing
SURE uses Cas12a2 RNA-triggered abortive infection as a recoverable counter-selection, raising transformation efficiency up to 1,000-fold versus Cas9- or Cas12-based systems while exceeding 90% editing efficiency for knockout, insertion, and replacement in E. coli MG1655.