CRISPR-Cas enzymes are directed to specific genomic DNA targets by guide RNA (gRNA). At the target site, the Cas enzyme must first bind the DNA before cleaving it. These 2 distinct steps, binding and cleavage, have separate and measurable kinetics. CRISPR-Cas enzymes can be highly specific, but off-target effects (OTEs) are also often detected, in regions not intended to be targeted. Many new Cas enzymes are currently being discovered and are not yet well characterized, so Zhang et al. reasoned that rapid and convenient new methods are needed to understand the specificity of Cas enzymes .
Zhang et al. developed 2 assays to run in parallel. The “specificity measured by sequencing” (Spec-seq) assay measures how DNA binding affinity is affected by mismatches and mutations. Meanwhile, the “sequence-specific endonuclease activity measurement by sequencing” (SEAM-seq) assay measures DNA cleavage activity. When used simultaneously, these assays constitute the Spec/SEAM-seq assay. The Spec-seq assay uses a library of dsDNA strands produced by PCR amplification of ssDNA. Some of the DNA sequences include a perfect target site, but most instead include a variety of target site mutants. The pooled dsDNA strands are combined with RNP targeting the predetermined site. However, this RNP is made of the catalytically inactive “dead Cas” (dCas) enzyme, rather than active Cas, along with gRNA. (Zhang et al. cloned various necessary mutant dCas enzymes so they could study many different Cas enzymes’ binding activity.) The dCas enzyme will bind but not cleave the target, and will also sometimes bind—but only with varying degrees of affinity—to the mutant targets. The mixture of RNP with the library is then run on an electrophoretic mobility shift assay (EMSA)-like gel. If DNA and dCas are tightly bound, the complex is heavy and does not migrate far through the gel; otherwise, the complex will dissociate, and the unbound DNA will migrate rapidly through the gel. Bands are excised, and DNA is purified and sequenced to determine the relative binding affinity of dCas to each sequence.
SEAM-seq uses catalytically active Cas enzyme, with the same gRNA and the same library used in Spec-seq. Therefore, there is cleavage and destruction of many of the DNA sequences in the library. After this occurs, the remaining DNA is sequenced, telling the researchers which sequences are cut and which are not.
When Spec-seq and SEAM-seq were used simultaneously, the sequences that were determined to be bound did not always correlate with the sequences that were found to be cleaved. Using a series of calculations and graphing techniques, Zhang et al. performed not only quantification, but also modeling and statistical analysis for SpCas9, Hifi‑SpCas9, ScCas9, AsCas12a, and DpbCasX. These 5 enzymes did not behave identically to each other with regard to binding and cleavage specificities.
SpCas9 was studied first, because its binding and cleavage properties have been extensively characterized in the literature. In agreement with published findings, Spec/SEAM-seq demonstrated that SpCas9 binding specificity is most heavily dependent on base-pairing of gRNA with target DNA within the PAM and in the base positions nearest the PAM, not in the target positions further from the PAM. Interestingly, for some targets, mismatches near the PAM were characterized by decreased binding affinity but increased cleavage when excess RNP is used. This situation was named “mismatch activation.” The authors carried out a thorough kinetic analysis of this observation and compared their model to the previously described “Doench rules,” which use a Cutting Frequency Determination (CFD) scoring matrix. They demonstrated the consistency and even the potential for superiority of Spec/SEAM-seq compared to this other approach. They also showed that their findings with regard to SpCas9 binding and cleavage specificity were in agreement with previous literature.
After demonstrating the validity of the Spec/SEAM-seq approach with SpCas9, the researchers went on to investigate binding and cleavage specificities of HiFi‑SpCas9, ScCas9, AsCas12a, and DpbCasX with the same approach. HiFi‑SpCas9 is a bioengineered mutant of wild-type SpCas9 developed by IDT to have increased target specificity with high activity. Spec/SEAM-seq showed that the increased target specificity of HiFi-SpCas9 is due to specificity in the cleavage step, not in the binding step. ScCas9 shares ~90% sequence identity to SpCas9, but has been reported to recognize a minimal “NNG” PAM . Spec/SEAM-seq redefined the optimal PAM site of ScCas9 as “NAG(B: A/G/C)(K: G/T)”, enabling consistent performance of editing in human cells similar to SpCas9 at NGG PAM sites.
AsCas12a has a reputation for low off-target activity, but the reason for this had not been thoroughly characterized before this study. The researchers determined using Spec/SEAM-seq that binding specificity was the primary contributor to this enzyme’s overall on-target specificity.
A recently discovered Cas enzyme from Deltaproteobacteria (Dpb) is known as DpbCasX. This enzyme uses a PAM site with the sequence TTCN, and displays high specificity. Similar to SpCas9, the researchers showed that DpbCasX can bind mismatched DNA but not cleave it, so additional specificity lies in the cleavage step.
The research team concluded that Spec/SEAM-seq is an excellent approach for studying Cas enzyme specificity in terms of its low cost, its requirement for only commonly accessible equipment, its speed, and its reliability. They emphasized that Spec/SEAM-seq gives results consistent with previously published approaches and the Doench rules. They also observed that the very high specificity of DpbCasX makes it an excellent candidate for future genome editing research.