A high-throughput CRISPR interference screen for dissecting functional regulators of GPCR/cAMP signaling

Abstract

G protein-coupled receptors (GPCRs) allow cells to respond to chemical and sensory stimuli through generation of second messengers, such as cyclic AMP (cAMP), which in turn mediate a myriad of processes, including cell survival, proliferation, and differentiation. In order to gain deeper insights into the complex biology and physiology of these key cellular pathways, it is critical to be able to globally map the molecular factors that shape cascade function. Yet, to this date, efforts to systematically identify regulators of GPCR/cAMP signaling have been lacking. Here, we combined genome-wide screening based on CRISPR interference with a novel sortable transcriptional reporter that provides robust readout for cAMP signaling, and carried out a functional screen for regulators of the pathway. Due to the sortable nature of the platform, we were able to assay regulators with strong and moderate phenotypes by analyzing sgRNA distribution among three fractions with distinct reporter expression. We identified 45 regulators with strong and 50 regulators with moderate phenotypes not previously known to be involved in cAMP signaling. In follow-up experiments, we validated the functional effects of seven newly discovered mediators (NUP93, PRIM1, RUVBL1, PKMYT1, TP53, SF3A2, and HRAS), and showed that they control distinct steps of the pathway. Thus, our study provides proof of principle that the screening platform can be applied successfully to identify bona fide regulators of GPCR/second messenger cascades in an unbiased and high-throughput manner, and illuminates the remarkable functional diversity among GPCR regulators.

Fig 1. A robust CREB transcriptional reporter for GPCR/cAMP activity.

(A) Schematic representation of the CREB reporter used in the CRISPR-based screen. Two cAMP response elements were fused to a ProteoTuner destabilizing domain (degron) followed by a green fluorescent protein, and inserted into a lentiviral vector. A clonal cell line was generated by transducing HEK293 cells with the reporter and dCas9-BFP-KRAB, sorting individual cells, and growing and verifying clonal lines for high reporter expression and efficient dCas9-dependent gene silencing. (B) The CREB reporter responds robustly to direct stimulation of adenylyl cyclase/cAMP signaling with forskolin. Reporter cells were treated with 10 μM forskolin (FSK) or DMSO (vehicle), and 1 μM Shield-1 was added simultaneously to stabilize the degron domain. After 4 h, reporter expression was analyzed by flow cytometry. Data from n = 4 per condition. (C) The CREB and PKA inhibitors, 666–15 and H89, respectively, significantly diminish FSK-induced accumulation of the reporter. Cells were pre-incubated with 100 nM 666–15 or 10 μM H89 versus DMSO (vehicle) for 30 min, then treated with 10 μM FSK in the presence of 1 μM Shield-1. After 4 h, reporter accumulation was analyzed by flow cytometry. Data plotted are from n = 5 for 666–15 and n = 7 for H89. (D) Timecourse for β2-AR-dependent reporter induction. Reporter cells were treated with 1 μM isoproterenol / 1 μM Shield-1 for indicated times or treated with 1 μM Shield-1 alone (no isoproterenol) for 5 h (“untreated”), and reporter expression was analyzed by flow cytometry. Data from n = 5. (E) Dose-response curve for isoproterenol-dependent reporter expression. Reporter cells were treated with indicated doses of isoproterenol and 1 μM Shield-1 for 4 h, and reporter expression was analyzed by flow cytometry. Data plotted are means of GFP expression, n = 3 per condition. EC₅₀ curve-fitting was performed using Prism6 GraphPad software. In each flow cytometry experiment, 10,000 cells total were analyzed and gated for singlets. Error bars = ± s.e.m. **** = p ≤ 0.0001; ** = p ≤ 0.01 by unpaired two-tailed Student’s t-test.

Fig 2. The reporter cell line is compatible with pooled functional genomic screening.

(A) Schematic representation of the CRISPR interference-based genomic screen. Reporter cells were transduced with sgRNA libraries for 1 week, β2-AR signaling was activated with 1 μM isoproterenol/1 μM Shield-1 for 4 h, and edited cells were sorted based on differential expression of the CREB reporter into “low”, “medium” or “high” bins. Genomic DNA was extracted from each fraction, amplified, barcoded and subjected to deep sequencing and statistical analysis of sgRNA enrichment. (B) Volcano plot of gene enrichment in the “high” versus “low” reporter-expressing sorted fractions after infection with the H1 CRISPRi library (S2 Table). Fold enrichment is calculated as the log₂ of the mean counts for 3/5 sgRNAs targeting a given gene, while p values are calculated based on the distribution of all 5/5 sgRNA targeting a gene relative to the NTCs in the library. PKA subunit-encoding genes are shown in blue, GRKs are shown in red. (C) Enrichment of ADRB2-targeting sgRNAs in the each sorted population are plotted and color-coded based on the log₂ enrichment of sgRNA counts as indicated in the legend. ADRB2-specific sgRNAs are colored in red, negative control sgRNAs are colored grey. sgRNAs used for validation experiments in S2 Fig are indicated. p-values were calculated by Mann-Whitney test using sgRNA activity relative to the negative control distribution.

Fig 3. A high-throughput CRISPR-based genomic screen identifies novel modulators of GPCR signaling.

(A) Volcano plot of gene enrichment in the “high” versus “low” fractions from all seven CRISPRi libraries. Significant hits with “hit strength” FDR ≤ 10% are color-coded based on their phenotype, and genes selected for follow-up validation are indicated with arrows. (B) Validation of the functional effects on β2-AR-dependent CRE-GFP reporter upregulation for a subset of novel regulators identified by the CRISPRi screen. Two sgRNAs per gene were selected based on screen phenotype, individually cloned and tested in batches. Cells were treated with 1 μM Shield-1 and 1 μM isoproterenol for 4 h and analyzed by flow cytometry. Mean GFP signal of the two control sgRNAs (5443 and 5444) for each batch was set to 100% and used to normalize the mean GFP values for all sgRNAs from that batch, as described in the “Materials and Methods” section. Data shown are mean from n = 3–12 per gene-specific sgRNA, and n = 41–42 for NTCs. (C) Enriched Gene Ontology categories (S5 Table) among High-confidence hits identified through “high”/”low” comparison. Error bars = ± s.e.m. **** = p ≤ 0.0001; *** = p ≤ 0.001; ** = p ≤ 0.01; * = p ≤ 0.05 by one-way ANOVA test.

Fig 4. Novel hits identified by the screen regulate GPCR/cAMP signaling through distinct mechanisms.

(A) PKMYT1 depletion blunts isoproterenol-dependent accumulation of an endogenous CREB target gene, PCK1. Cells were left untreated (basal, ND = no drug), treated with 1 μM isoproterenol, or with 10 μM forskolin for 1 h, RNA was isolated, reverse transcribed and analyzed by qPCR. Data plotted are GAPDH-normalized mean levels from n = 3. (B) NUP93 knockdown diminished cAMP production following GPCR-dependent and direct induction of adenylyl cyclase enzymes. Basal cAMP levels (left) were measured by cAMP ELISA assay, and induced levels (right) were measured using a luciferase-based biosensor in real time upon addition of 1 μM isoproterenol, 10 μM NECA or 10 μM forskolin (FSK). ELISA values were normalized to total protein per sample, and maximum biosensor values were normalized to account for plasmid transfection efficiency (see “Materials and methods”). All values shown are percent of averaged controls. Data are mean from n = 5–7 (ELISA) and n = 5–6 (biosensor). (C) Effects of NUP93 depletion on trafficking of β2-AR analyzed by flow cytometry. Cells transiently transfected with flag-tagged β2-AR were either left untreated or were induced with 1 μM isoproterenol for 20 min, and cell-surface levels of flag-tagged receptor were measured by incubation with Alexa-647-M1 on ice and flow cytometry analysis. Average control values were set to 100% and used to normalize the data as described in the “Materials and Methods” section. Data are mean from n = 5–6. Error bars = ± s.e.m. **** = p ≤ 0.0001, ** = p ≤ 0.01, * = p ≤ 0.05 by one-way ANOVA test.

Fig 5. Model depicting proposed regulatory roles for validated GPCR/cAMP mediators.

Proposed roles are assigned based on functional and localization annotations of the factors (Table 1) and experimental evidence provided in this study.