Systematically investigating the key features of the DNase deactivated Cpf1 for tunable transcription regulation in prokaryotic cells

Abstract

With a unique crRNA processing capability, the CRISPR associated Cpf1 protein holds great potential for multiplex gene regulation. Unlike the well-studied Cas9 protein, however, conversion of Cpf1 to a transcription regulator and its related properties have not been systematically explored yet. In this study, we investigated the mutation schemes and crRNA requirements for the DNase deactivated Cpf1 (dCpf1). By shortening the direct repeat sequence, we obtained genetically stable crRNA co-transcripts and improved gene repression with multiplex targeting. A screen of diversity-enriched PAM library was designed to investigate the PAM-dependency of gene regulation by dCpf1 from Francisella novicida and Lachnospiraceae bacterium. We found novel PAM patterns that elicited strong or medium gene repressions. Using a computational algorithm, we predicted regulatory outputs for all possible PAM sequences, which spanned a large dynamic range that could be leveraged for regulatory purposes. These newly identified features will facilitate the efficient design of CRISPR-dCpf1 based systems for tunable multiplex gene regulation.

Fig. 1

Mutation variants of dCpf1 induced differential gene repression. (A) Schematic representation of the cellular circuit for evaluating performance of the dCpf1-crRNA system. In the circuit, dCpf1 and crRNA were expressed from an inducible promoter (Ptac) and a constitutive promoter (J23119), respectively, and a reporter gene (super-folded gfp, sf-gfp) is repressed by the dCpf1-crRNA complex at promoter and transcribed regions. (B) Summary of repression abilities of different dFnCpf1 and dLbCpf1 variants. Repression fold is calculated as the ratio between fluorescence of the positive control and the test systems at 10³μM IPTG inducer concentration in (C) and (D). (C) Repression curves of three dLbCpf1 variants. The positive control (“P”) was of the strain with an empty crRNA plasmid, while the negative control (“N”) shows the background fluorescence of a strain with an empty gfp plasmid. (D) Repression curve of three dFnCpf1 variants. The positive and negative controls are the same as in (C). Error bars represent standard deviation of fluorescence for three independent experiments on different days. For crRNA sequences see Table S3.

Fig. 2

The effect of repeat and guide sequence lengths on gene repression by dFnCpf1-crRNA. (A) Aligned repeat sequences of different lengths used in crRNAs. The dashed line indicates the cleavage site on crRNA during crRNA processing by dFnCpf1. Red colored sequences remain in the mature crRNA, while the rest of the sequences are cleaved off. (B) Gene repression curves of dFnCpf1 with truncated repeat sequences. (C) Maximal repression folds of dFnCpf1 with the same set of truncated repeat sequence as in (B). (D) Aligned guide sequences of different lengths used in crRNAs. (E) The repression curves for different lengths of guide sequences in the dFnCpf1-crRNA system. (F) Maximal repression folds of dFnCpf1 with the same set of truncated guide sequences as in (E). Error bars represent standard deviation of fluorescence for three independent experiments on different days. Positive and negative controls are the same as in Fig. 1.

Fig. 3

Repression by dFnCpf1 with co-transcribed crRNAs. (A) Schematic representation of target sequences for each single crRNA, as well as the design of individual and combined multiple crRNAs. (B) Different lengths of repeat sequence in the triply-combined crRNA co-transcript. (C) Repression curves of dFnCpf1 with single or multiple crRNAs. (D) Repression curves of dFnCpf1 with varied repeat sequence lengths in the same triply-combined crRNA co-transcript. Error bars represent standard deviation of fluorescence for three independent experiments on different days. Positive and negative controls are the same as in Fig. 1. For crRNA sequences see Table S3.

Fig. 4

Gene repression on sliding targets with the canonical TTN PAM motif for dFnCpf1. (A) Top panel: targets T6-T8 were selected within the gfp coding sequence by 1-nt or 2-nt shifting. Red letters show the corresponding PAM sequences. Lower panel: gene repression by dFnCpf1 targeting the respective sequences. (B) Another two sets of 1-nt or 2-nt shifted target sequences and the respective repression curves by dFnCpf1. Error bars represent the standard deviation of fluorescence for three independent experiments on different days. Positive and negative controls are the same as in Fig. 1. For crRNA sequences see Table S3.

Fig. 5

Negative reporter screen for the PAM dependence of dFnCpf1's regulatory activity. (A) Design of the screening circuit. A randomized 6-nt PAM sequence (red) was placed upstream of a fixed target sequence, and a ribozyme insulator (RiboJ) was inserted between the target and the reporter gene to eliminate the effect of PAM sequences on yfp translation. (B) Fluorescence distributions measured by flow cytometry for clones carrying the specified PAM sequences or cell populations carrying the randomized PAM library, under the non-induced condition. (C) Fluorescence measured under the induced condition, for n = 200 clones carrying different PAMs randomly selected from flow-cytometer sorted library cells. Error bars represent the standard deviation of two to four independent experiments on different days. (D) Fluorescence for the 200 sample clones under two different IPTG inducer concentrations (20 μM and 100 μM).

Fig. 6

Predictions of PAM strengths for dFnCpf1. (A) Predictions of repression strengths for 4096 6-nt PAMs. PAMs are sorted by predicted values. Back predictions were made for measured PAMs in the sample pool (n = 200) from values predicted for unmeasured PAMs (n = 3896). For measured values (red dots), error bars show the standard error of mean for two to four independent measurements on different days. Sequence logos were obtained from measured PAMs with fluorescence strengths in ranges (0,200), (200, 600) and (600, 4000). (B) Site degeneracy for PAM positions 3–6. Box plots for measured PAM strengths (log fluorescence values, y-axis) of the sequence context specified on the x-axis. Percentages in parentheses indicate fractions of contexts that are degenerate by a 2-fold threshold. Numbers on top indicate the number of measured words within each sequence context. (C) Correlation between predictions and measured values in a sample run of cross-validation tests at 50% training set-testing set split. ρ: Pearson correlation between log values. (D) Summary of cross-validation tests. Distribution of correlation between predictions and measurements for the training (left) and testing (right) sets. In each run, data for training were randomly selected from and thus have the same distribution as the sample pool. 100 runs were conducted for each training set-testing set splitting ratio.