CRISPR technologies and the search for the PAM-free nuclease

Abstract

The ever-expanding set of CRISPR technologies and their programmable RNA-guided nucleases exhibit remarkable flexibility in DNA targeting. However, this flexibility comes with an ever-present constraint: the requirement for a protospacer adjacent motif (PAM) flanking each target. While PAMs play an essential role in self/nonself discrimination by CRISPR-Cas immune systems, this constraint has launched a far-reaching expedition for nucleases with relaxed PAM requirements. Here, we review ongoing efforts toward realizing PAM-free nucleases through natural ortholog mining and protein engineering. We also address potential consequences of fully eliminating PAM recognition and instead propose an alternative nuclease repertoire covering all possible PAM sequences.

Fig. 1. The PAM in target recognition and self/nonself-discrimination for CRISPR–Cas systems.

a Two checkpoints, the protospacer adjacent motif (PAM) and a flanking target matching the guide, for successful recognition by a Cas nuclease. Cas9 and a single-guide RNA (sgRNA) are used as a representative example. A matching PAM and target results in R-loop formation and target cleavage, whereas either a non-PAM or a nonmatching target block either recognition step by the Cas nuclease. b Role of the PAM in self- versus nonself-recognition in prokaryotic immune defense. Self refers to each spacer within the CRISPR array encoded on the host’s genome or endogenous plasmids, whereas nonself refers to invading nucleic acids such as phage or exogenous plasmids. The sgRNA has been engineered for ease-of-use as a fusion between the processed tracrRNA and CRISPR RNA molecules found in the native system.

Fig. 2. Phylogenetic relationship of PAM-characterized Cas9 and Cas12a nucleases found in nature.

a Cas9 nucleases with characterized PAMs. b Cas12a nucleases with characterized PAMs. Phylogenetic trees with corresponding consensus PAMs are shown to the right. PAM-engineered variants are in light-blue text. The colors refer to each nucleotide: A = green, C = blue, G = yellow, and T = red. Stacked colors represent the recognition of at least two or three different nucleotides at the same position. Phylogenetic trees were generated using Geneious (Geneious Prime, version 2019.2.3, Biomatters Ltd.) based on the complete protein amino-acid sequence. Some of the nucleases recognized consensus sequences that slightly varied based on the PAM determination method or contained multiple motifs that were preferentially recognized, including those for CjeCas9, ThermoCas9, PdCas12a, MbCas12a, and Pb2Cas12a. See Supplementary Data 1 for more details.

Fig. 3. Mutations in the PAM-engineered variants of SpyCas9.

a Domain architecture of SpyCas9. The location of the 34 mutated residues for all PAM-engineered variants (with the exception of chimeras) is indicated below the linear map. Each associated variant contains anywhere from 1 to 16 mutations. b Structure of SpyCas9 with mutated residues. Mutated residues are in salmon. The structural domains are colored according to the linear map in (a). The diguanine nucleotides in the PAM are shown as blue lines. Structural images were generated from PBD: 4UN3 using PyMOL (The PyMOL Molecular Graphics System, Version 2.4 Schrödinger, LLC).

Fig. 4. Implications of PAM engineering.

a Comparing target accessibility for Cas nucleases with relaxed or stringent PAM requirements. b The PAM-free nuclease versus a repertoire of nucleases that collectively recognize every possible sequence. Here, the repertoire consists of four nucleases recognizing one letter at the second position. c Qualitative comparison of a PAM-free nuclease, a PAM-relaxed nuclease, a PAM-stringent nuclease, and the nuclease repertoire. The nucleases are compared using different metrics of targeting performance. The more a bar is filled, the greater our prediction the associated nuclease can perform under that metric. We consider nuclease fidelity as the ability of the nuclease to ignore nontarget sequences at least partially matching the guide sequence. We consider multiplex ability as how readily the nuclease can be implemented to target any set of genomic sequences.