Design of ligand-controlled sgRNAs by inserting small molecule aptamers into the sgRNA. a Illustration of the design goal, where functional Cas9/sgRNA/target DNA complexes are stabilized either in the presence (top) or absence (bottom) of a small molecule ligand. b Aptamer insertion sites; sgRNA domains defined as in ref. . c Strategies for linking the aptamer to the sgRNA (Supplementary Table ). d Efficiency of in vitro Cas9 cleavage of DNA in the presence and absence of 10 mM theophylline for controls and selected designs. Design numbers refer to Supplementary Table  and are color-coded by aptamer insertion sites defined in b. Percent cut values (bottom) are the average of at least two experiments. All data shown are from a single gel (some lanes are excluded for clarity). e Dose-dependence of cleavage efficiency in response to increasing concentrations of theophylline for a representative design (#24). Source data are available in the source data file

Identification of robust ligRNAs using CRISPRi-based gene repression in E. coli. a Components used in the CRISPRi assay. dCas9 and any ligRNAs were expressed from plasmids, while the fluorescent reporters (GFP and RFP) were chromosomally integrated. The DNA regions targeted by different spacers (sgG1, sgR1, sgR2) used to repress the fluorescent reporters are indicated. b Regions randomized in each ligRNA library. c Schematic of the screen used to isolate ligRNA⁺. sgG1, sgR1, and sgR2 refer to spacers targeting GFP and RFP, respectively (Supplementary Table ). d, e Single-cell RFP fluorescence distributions for ligRNA⁺ (teal, d) and ligRNA⁻ (navy, e) targeting RFP using the sgR2 spacer with (solid lines) and without (dashed lines) theophylline. Control distributions are in grey (positive control: optimized sgRNA scaffold; negative control: G43C G44C). The mode of each distribution is indicated with a plus sign. RFP fluorescence values for each cell are normalized by both GFP fluorescence for that cell and the modes of the un-repressed control populations (i.e. apo and holo) measured for that replicate. f Efficiency of CRISPRi repression with increasing theophylline concentrations for ligRNA⁺ (teal) and ligRNA⁻ (navy). Controls are in grey. The fluorescence axis is the same as in d and e. The fits are to a two-state equilibrium model. g GFP mRNA levels after the addition or removal of theophylline. GFP mRNA levels were measured by qPCR and are normalized to 16S rRNA levels. Solid, colored lines: theophylline added at t = 0. Black, dashed lines: theophylline removed at t = 0. Grey lines: theophylline present (solid) or absent (dashed) for the whole experiment. Error bars reflect standard deviations from 3 technical replicates. h Change in the percentage of DNA cleaved in vitro in the presence and absence of theophylline for ligRNA⁺ and ligRNA⁻ for 24 representative spacers. Bar heights represent the mean of three or four measurements with a single spacer (except for spacer #24, where n = 2, Supplementary Table ). Data for each replicate are shown as faded plus marks. Pos and Neg denote the positive and negative control sgRNA scaffolds (Supplementary Table ). The control bars combine data for all 24 spacers

ligRNAs function in the context of the endogenous loci in different bacteria. a Schematic showing sites in the E. coli lac operon (the genes lacI and lacZ, the transcriptional operator sites A and O, and the promoter P driving LacZ expression) targeted by ligRNAs (black arrows). Note that LacI inhibits expression of LacZ (red arrow), so the ligRNAs targeting lacI are expected to increase LacZ expression when active (Supplementary Fig. ). Expression of LacZ is detected by conversion of o-nitrophenol-β-D-galactopyranoside (ONPG) into o-nitrophenyl (ONP, yellow) and galactose (Gal). b Fold change in LacZ activity for E. coli grown with or without theophylline, expressing either a control sgRNA (pos or neg, grey), ligRNA⁺ (teal), or ligRNA⁻ (navy), with spacers targeting the indicated part of the lac operon. Spacer sequences were taken from Qi et al. and are listed in Supplementary Table . Bar heights represent the mean of three biological replicates. Data for each replicate are shown as faded plus marks. The direction of the fold change (i.e. holo/apo or apo/holo) was chosen such that the resulting value is greater than one. ligRNA⁺ responds to theophylline for all 4 spacers, while ligRNA⁻ does so only for the two genes lacZ and lacI. c Schematic illustrating a growth assay in P. aeruginosa with ligRNA⁺ targeting the endogenous dihydrofolate reductase (DHFR) gene (folA). DHFR is neccesary for growth (green arrows) in the absence of thymidine, and is inhibited by the antibiotic trimethoprim (purple arrow). d Growth (OD600) of P. aeruginosa at 12 h in the presence (teal, solid) and absence (black, dashed) of theophylline at the indicated concentrations of trimethoprim (TMP). Inhibition of FolA by ligRNA⁺ in the presence of theophylline lowers the minimum concentration of trimethoprim needed to slow growth. Controls are in grey (high MIC: wildtype; low MIC: optimized sgRNA scaffold) and are also in the presence (solid) and absence (dashed) of theophylline. The spacer sequence is listed in Supplementary Table . Error bars are standard deviations of at least three biological replicates

Multiplexed temporal control of two genes with two ligands. a Schematic illustrating the constructs and the expected consequences of adding theophylline (theo) and 3-methylxanthine (3mx) for each fluorescent reporter. b GFP and RFP fluorescence measured at indicated time points by flow cytometry. Presence of theophylline leads to GFP expression; addition of 3-methylxanthine separately at a different time point also triggers RFP expression. Both effects are reversible (GFP and RFP are repressed when both ligands are absent) and expression can be triggered a second time with ligand addition. Bar heights represent the mean of three biological replicates. Data for each replicate are shown as faded plus marks. Unlike in Fig. , fluorescence is normalized by side-scatter because both fluorescent channels are being manipulated