Targeted, Genome-scale Overexpression in Proteobacteria

Abstract

Targeted, genome-scale gene perturbation screens using Clustered Regularly Interspaced Short Palindromic Repeats interference (CRISPRi) and activation (CRISPRa) have revolutionized eukaryotic genetics, advancing medical, industrial, and basic research. Although CRISPRi knockdowns have been broadly applied in bacteria, options for genome-scale gene overexpression face key limitations. Here, we develop a facile approach for genome-scale overexpression in bacteria we call, "CRISPRtOE" (CRISPR transposition and OverExpression). We first create a platform for comprehensive gene targeting using CRISPR-associated transposons (CAST) and show that transposition occurs at a higher frequency in non-transcribed DNA. We then demonstrate that CRISPRtOE can upregulate gene expression in Proteobacteria with medical and industrial relevance by integrating synthetic promoters of varying strength upstream of target genes. Finally, we employ CRISPRtOE screening at the genome-scale in the model bacterium Escherichia coli and the non-model biofuel producer Zymomonas mobilis, recovering known and novel antibiotic and engineering targets. We envision that CRISPRtOE will be a valuable overexpression tool for antibiotic mode of action, industrial strain optimization, and gene function discovery in bacteria.

Figure 1.

A CRISPR-associated transposition (CAST) system for targeted, genome-scale overexpression in Proteobacteria called “CRISPRtOE” (CRISPR transposition and OverExpression). CRISPRtOE precisely delivers a transposon with an outward facing promoter upstream of genes to facilitate overexpression. CRISPRtOE can upregulate gene expression in medically and industrially-relevant Proteobacterial species. Genome-scale CRISPRtOE screens in model (E. coli) and non-model (Zymomonas mobilis) Proteobacteria can be used to define genes required for resistance to antibiotics or biofuel production stressors.

Figure 2.

A mobilizable, selectable, dual-plasmid system for CRISPR-guided transposition. (A) Schematic of mechanism of Vibrio cholera Type I-F CRISPR-associated transposon (VcCAST) system(20, 21). (B) Schematic of strain construction using the CRISPRt targeted transposition system. Plasmids have pir-dependent origins of replication preventing replication in the bacterial recipient. One plasmid encodes the minimal Type I-F CRISPR-Cas (Cas678) and Tn7-like transposase (TnsABC and TniQ) machinery, and a second plasmid contains a transposon carrying the guide RNA and antibiotic resistance expression cassettes (see Figure S1 for details). These plasmids are transferred by co-conjugation to a recipient bacterium by E. coli donor strains with a chromosomal copy of the RP4 transfer machinery. Inside the recipient cell, the transposon (flanked in yellow) is inserted onto the chromosome at a site determined by the sequence of the guide RNA. Selection on antibiotic plates lacking diaminopimelic acid (DAP) selects for transconjugants and against the E. coli donor strains. (C) Specificity of CRISPRt disruption of gfp in E. coli measured by Tnseq. Mapping of location of transposon insertion sites to the E. coli genome after CRISPRt targeted transposition with the gfp1 guide (226,671 reads). (D) Transposition efficiency of 3 individual CRISPRt guides (T03, T09, T21) targeting mScarlet-I cassettes in the E. coli att_Tn7 site either with a promoter (+ transcription) or without a promoter (− transcription) measured by plating efficiency with and without selection (+/− kanamycin) in triplicate. Red arrows indicate fold change in efficiency without and with transcription. Error represents the average of n=3 assays, * indicates p<0.05, two-tailed t-test.

Figure 3.

Tunable overexpression of chromosomally located genes using CRISPRtOE. (A) Schematic of the CRISPRtOE targeted overexpression system. The test chromosomal target is an mScarlet-I gene preceded by a ‘Landing Pad’ (LP) with no promoter, a PAM, and the LZ1 protospacer (see Figure S6C). The CRISPRtOE construct has a transposon carrying guide RNA and antibiotic resistance expression cassettes and an outward facing promoter (see Figure S6B for details). Co-conjugation of strains carrying the CRISPRtOE construct and the CRISPRt-H plasmid (harboring the VcCAST machinery) with a promoterless ‘LP’ mScarlet-I reporter recipient strain results in a CRISPRtOE test strain. (B) mScarlet-I fluorescence analysis of E. coli CRISPRtOE isolates (no promoter or synthetic promoters A, V, and H) compared to the parent (promoterless ‘LP’ mScarlet-I) strain. CRISPRtOE transposon insertion position and fluorescence measurements were determined for twelve isolates. Fluorescence measurements were normalized to cell density (OD₆₀₀). On target, RL orientation isolates are shown here, see Figure S7B for all orientations and Table S6 for values. Error is expressed for the median value of 8–12 isolates in n=3 assays. (C) Fold effect of CRISPRtOE mScarlet-I overexpression using synthetic promoter H in eight Alpha- and Gammaprotebacteria (E. coli (Eco), Enterobacter cloacae (Ecl), Klebsiella pneumoniae (Kpn), Acinetobacter baumannii (Aba), Pseudomonas aeruginosa (Pae), Pseudomonas putida (Ppu), Zymomonas mobilis (Zmo), and Shewanella oneidensis (Son)). CRISPRtOE transposon insertion position and fluorescence measurements were determined for twelve isolates. Fluorescence measurements were normalized to cell density (OD₆₀₀). Values are shown for isolates with on-target, RL orientation CRISPRtOE insertions and fold changes for on-target and RL orientation CRISPRtOE isolates compared to a strain with promoterless mScarlet-I (see Table S6 for details). Error is expressed for the median value of 8–12 isolates in n=3 (Eco and Zmo) or n=2 (Ecl, Kpn, Aba, Pae, Ppu, and Son) assays.

Figure 4.

Genome-scale CRISPRtOE. (A) Schematic of genome-scale CRISPRtOE experiment including design and construction of libraries, experimental screen, bioinformatic data processing, and validation. Bioinformatic steps are in purple and experimental steps are in blue. Ten gRNAs were designed to guide insertion of the Tn6677 transposon (with or without a promoter) upstream of all genes. Pooled CRISPRtOE libraries grown in the presence of sub-lethal concentrations of ihibitors (e.g. E. coli: trimethoprim and fosfomycin antibiotics; Z. mobilis: switchgrass hydrolysate) and were screened for fitness by Tn-seq before and after treatment. Data were analyzed by counting reads that were in the R-L orientation upstream of genes to discover fitness phenotypes dependent on gene overexpression. Individual strains overexpressing genes with CRISPRtOE or from a plasmid were constructed to validate phenotypes outside of the pooled context. (B) Positions of CRISPRtOE insertions upstream of all genes in the genome. Insertion site distance d was calculated based on the distance in bp from the 3′ end of the Tn to the 5′ end of the gene. CRISPRtOE Tn insertions shown in red. For comparison, random/untargeted insertions from a Tn5 library are shown in green. (C) CRISPRtOE insertions into the E. coli lac locus showing precise placement upstream of genes compared to a library constructed with Tn5 random transposition. The scale is capped at 20 reads for visual clarity.

Figure 5.

E. coli genome scale CRISPRtOE screen with trimethoprim. (A) Volcano plot of gRNA spacer counts in trimethoprim (TMP) versus dimethylsulfoxide (DMSO) control. Each dot represents a strain with a spacer targeting upstream of a gene. Those that are not statistically significant (false discovery rate (FDR) > 0.05) are shown in light grey and significant hits (FDR ≤ 0.05) are shown in dark gray. Several outliers are colored: e.g. folA (red), metF (blue), rffM/wecG (green). (B) Top 10 gene hits with increased fitness in TMP treatment. Genes must have at least a 4-fold change in the CRISPRtOE-Promoter H data and show a 4-fold difference in a comparison between Promoter H (red bars) and no promoter (blue bars) to be considered. Dots represent unique insertions. Error bars represent standard deviation (SD). (C) Liquid MIC assay showing resistance of strains overexpressing metF from a plasmid to TMP. Error bars represent SD of triplicate assays. Significance was determined by a t test (≤0.05(*), ≤0.01(**), ≤0.001(***)). (D) E. coli Enterobacterial Common Antigen (ECA) biosynthesis (rff/wec) operon. (E) Liquid MIC assay showing resistance of strains overexpressing rffM from a plasmid to TMP. Error bars represent SD of triplicate assays. Significance was determined by a t test (≤0.05(*), ≤0.01(**), ≤0.001(***)). (F) Growth assay of strains overexpressing rffM from a plasmid in additional chemical conditions (Biolog phenotypic array plate PM16 comprised of 24 compounds). Two concentrations of CRISPRtOE screen chemicals FOS (no response expected) and TMP (response expected), and of two additional chemical that had a significant response (diamide and rifamycin) are shown. Duplicate assays are shown (Rep1 and Rep2). Significance was determined by a t test (≤0.05(*), ≤0.01(**). See figure S14 for additional details.

Figure 6.

E. coli genome scale CRISPRtOE screen with fosfomycin. (A) Top gene hits with increased fitness in fosfomycin (FOS) treatment. Genes must have at least a 4-fold change in the CRISPRtOE-Promoter H data and show a 4-fold difference in a comparison between Promoter H (red bars) and no promoter (blue bars) to be considered. Dots represent unique insertions. sError bars represent standard deviation (SD). (B) CRISPRtOE-Promoter H insertions with increased fitness in FOS in the E. coli phosphonate utilization (phn) operon represented by arrows. (C) Quantification of MIC test strip (Liofilchem) assay showing resistance to FOS of individual CRISPRtOE-Promoter H strains overexpressing phnC or phnH. Error bars represent SD of triplicate assays (also see figure S15B). (D) Schematic of phnMNOP chromosomal deletion and phnM complementation experiment. (E) Quantification of FOS MIC test strip (Liofilchem) assays in WT, ΔphnMNOP deletion, or CRISPRtOE-Promoter H-phnC plus ΔphnMNOP deletion strains that are overexpressing phnM from a plasmid (or empty vector control). Error bars represent SD of triplicate assays (also see figure S15D).

Figure 7.

Z. mobilis genome scale CRISPRtOE screen with plant hydrolysate. (A) Volcano plot of gRNA spacer counts in 65% hydrolysate versus Zymomonas rich defined medium (ZRDM) control. Each dot represents a strain with a spacer targeting upstream of a gene. Those that are not statistically significant (false discovery rate (FDR) > 0.05) are shown in light grey and significant hits (FDR ≤ 0.05) are shown in dark gray. Several outliers are colored: fumA (green), rpoE (orange). (B) Top gene hits with increased fitness in 65% hydrolysate. Genes must have at least a 4-fold change in the CRISPRtOE-Promoter H data and show a 4-fold difference in a comparison between Promoter H (red bars) and no promoter (blue bars) to be considered. Dots represent unique insertions. Error bars represent standard deviation (SD). (C) Growth assay of individually constructed CRISPRtOE-Promoter H rpoE strains versus the parent strain in 65% hydrolysate or ZRDM control. Error bars represent standard deviation (SD) of three independant cultures. (D) Relative fitness of strains shown in panel (C). Growth was quantified by calculating the area under the curve (auc_e) using GrowthCurver. Fitness was normalized to control condition (parent strain in ZRDM). Significance was determined by a t test (≤0.005(***). See also Figure S16C and S16D.