Rational engineering of a modular bacterial CRISPR-Cas activation platform with expanded target range

Abstract

CRISPR-Cas activator (CRISPRa) systems that selectively turn on transcription of a target gene are a potentially transformative technology for programming cellular function. While in eukaryotes versatile CRISPRa systems exist, in bacteria these systems suffer from a limited ability to activate different genes due to strict distance-dependent requirements of functional target binding sites, and require greater customization to optimize performance in different genetic and cellular contexts. To address this, we apply a rational protein engineering approach to create a new CRISPRa platform that is highly modular to allow for easy customization and has increased targeting flexibility through harnessing engineered Cas proteins. We first demonstrate that transcription activation domains can be recruited by CRISPR-Cas through noncovalent protein-protein interactions, which allows each component to be encoded on separate and easily interchangeable plasmid elements. We then exploit this modularity to rapidly screen a library of different activation domains, creating new systems with distinct regulatory properties. Furthermore, we demonstrate that by harnessing a library of circularly permuted Cas proteins, we can create CRISPRa systems that have different target binding site requirements, which together, allow for expanded target range.

Figure 1.

Synthetic transcription activation with dCas9 fusions to ADs. (A) Schematic of a CRISPRa system composed of dCas9 fused to the ω subunit of RNAP (dCas9-ω). For characterization, dCas9-ω in complex with the sgRNA was targeted upstream of a promoter driving RFP expression. Activation is achieved by local recruitment of RNAP to promoter elements. (B) Fluorescence characterization of a dCas9-ω system in wild-type strain E. coli K-12 BW25113 (WT) and a modified strain that lacks the gene encoding the ω subunit (ΔrpoZ). (C) Fluorescence characterization of dCas9 with N- or C-terminal fusions to different ADs through a two-alanine linker in E. coli MG1655 strain. ADs are derived from RNAP subunits: ω, α and αNTD; or transcription factors: SoxS and LuxR. Fluorescence measurements (measured in units of fluorescence [FL]/optical density [OD] at 600 nm) were performed with E. coli cells transformed with an RFP reporter plasmid, a plasmid encoding the dCas9 fusions, and a sgRNA-encoding plasmid or a no-sgRNA control plasmid. sgRNA variants used targeted PAMs located at 80 bp upstream of the promoter TSS on the template strand (+ sgRNA T) or 81 bp upstream on the non-template stand (+ sgRNA NT). Data represent mean values and error bars represent s.d. of n = 4 biological replicates. A two tailed Student's t test was used to calculate P value comparing against the no-sgRNA control. * P < 0.0001; P > 0.0001 has no asterisk.

Figure 2.

Evaluation of different ADs and fusion strategies. (A) Characterization of different protein fusion strategies to recruit the αNTD AD through dCas9 by a covalent two-alanine (AA) linker, covalent 16 amino acid XTEN linker, or a noncovalent SYNZIP interaction domain. Schematics of DNA constructs shown under each data set. (B) Schematic of the modular CRISPRa design. The dCas9 and AD are independently fused to SYNZIP domains that form heterodimers. (C) A schematic illustrating creation and screening of an AD library using our modular CRISPRa platform. Different ADs were identified from αNTDs derived from 8 different bacterial species and translationally fused to a SYNZIP domain. These AD-encoding plasmids were then co-transformed with dCas9-SYNZIP and sgRNA-encoding plasmids to create new CRISPRa systems. (D) Characterization of the AD library. Fluorescence characterization (measured in units of fluorescence [FL]/optical density [OD] at 600 nm) was performed with MG1655 E. coli cells transformed with an RFP reporter plasmid, a dCas9-AD plasmid or separate dCas9- and AD-encoding plasmids, and a sgRNA-encoding plasmid or a no-sgRNA control plasmid. sgRNA variants used targeted PAMs located at 80 bp upstream of the promoter TSS on the template strand (+ sgRNA T) and 81 bp upstream on the non-template stand (+ sgRNA NT). Data represent mean values and error bars represent s.d. of n = 4 biological replicates. A two tailed Student's t test was used to calculate p value comparing against the no-sgRNA control. * P < 0.0001; P > 0.0001 has no asterisk.

Figure 3.

Characterization of distance-dependent activation patterns. Measuring the distance-dependent activation patterns of CRISPRa systems using XTEN and SYNZIP linkers. These systems were targeted to PAMs located between 71 and 101 bp upstream of the promoter TSS on both the template and non-template strands. Fluorescence characterization (measured in units of fluorescence [FL]/optical density [OD] at 600 nm) was performed in MG1655 E. coli cells transformed with an RFP reporter plasmid, a dCas9 plasmid, an AD plasmid and a sgRNA-encoding plasmid or a no-sgRNA control plasmid. Fold activation was calculated by dividing the [FL]/[OD] obtained in the presence of a targeting sgRNA against the no-sgRNA control within each reporter plasmid. Data represent mean values and shading represent s.d. of n = 4 biological replicates.

Figure 4.

Circularly permuted dCas9 (cpdCas9) variants allow for an expanded target range. (A) cpdCas9 variants were fused to a SYNZIP through the new N- or C-termini to generate a cpdCas9 SYNZIP library that was evaluated as part of the modular CRISPRa system. cpdCas9 variants used in this study are shown on the crystal structure of Cas9 (PDB: 5F9R), DNA is colored purple, sgRNA is colored blue and dCas9 is colored grey. (B) Characterization of the distance-dependent activation patterns of CRISPRa systems using three dCas9 variants: wild-type dCas9 (SYNZIP-dCas9), cpdCas9¹⁰²⁹ (SYNZIP-cpdCas9¹⁰²⁹) and cpdCas9¹⁹⁹ (SYNZIP-cpdCas9¹⁹⁹). Distance-dependent activation patterns were measured by targeting PAMs located between 61 and 71 bp upstream of the promoter TSS on the non-template strand (61NT to 71NT). Fluorescence characterization (measured in units of fluorescence [FL]/optical density [OD] at 600 nm) was performed in MG1655 E. coli cells transformed with an RFP reporter plasmid, a plasmid encoding the cpdCas9 or dCas9 variant, a plasmid encoding αNTD-SYNZIP, and a sgRNA-encoding plasmid or a no-sgRNA control plasmid. Fold activation was calculated by dividing the [FL]/[OD] obtained in the presence of a targeting sgRNA against the no-sgRNA control within each reporter plasmid. Data represent mean values and shading represent s.d. of n = 4 biological replicates. (C) Structural analysis of the angles corresponding to the different positions of the N-terminal SYNZIP fusions in the dCas9, cpdCas9^1029,and cpdCas9¹⁹⁹ relative to the axis of the DNA double helix (purple). (D) Fold activation values obtained from the different dCas9 variants targeting positions 61NT to 71NT shown as a function of the targeted position in a DNA helical wheel that was set to 10.5 bp per turn.