Detection and Quantification of GPCR mRNA: An Assessment and Implications of Data from High-Content Methods

Abstract

G protein-coupled receptors (GPCRs) are the largest family of membrane receptors and targets for approved drugs. The analysis of GPCR expression is, thus, important for drug discovery and typically involves messenger RNA (mRNA)-based methods. We compared transcriptomic complementary DNA (cDNA) (Affymetrix) microarrays, RNA sequencing (RNA-seq), and quantitative polymerase chain reaction (qPCR)-based TaqMan arrays for their ability to detect and quantify expression of endoGPCRs (nonchemosensory GPCRs with endogenous agonists). In human pancreatic cancer-associated fibroblasts, RNA-seq and TaqMan arrays yielded closely correlated values for GPCR number (∼100) and expression levels, as validated by independent qPCR. By contrast, the microarrays failed to identify ∼30 such GPCRs and generated data poorly correlated with results from those methods. RNA-seq and TaqMan arrays also yielded comparable results for GPCRs in human cardiac fibroblasts, pancreatic stellate cells, cancer cell lines, and pulmonary arterial smooth muscle cells. The magnitude of mRNA expression for several Gq/11-coupled GPCRs predicted cytosolic calcium increase and cell migration by cognate agonists. RNA-seq also revealed splice variants for endoGPCRs. Thus, RNA-seq and qPCR-based arrays are much better suited than transcriptomic cDNA microarrays for assessing GPCR expression and can yield results predictive of functional responses, findings that have implications for GPCR biology and drug discovery.

Figure 1

Detection of GPCRs by RNA-seq, TaqMan GPCR arrays, and Affymetrix (Affy) arrays. (a–c) GPCRs for which the relevant primer probes are present by each method. Representative data are shown for an individual CAF replicate as an example; similar numbers were detected by all three methods in a second replicate. (d) Differences in GPCR expression between RNA-seq and TaqMan arrays. (e) False-positive detection of GPCRs from Affymetrix HG U133plus2.0 arrays; these GPCRs were detected by neither RNA-seq (plotted above) nor TaqMan arrays (not shown). (f) False negatives from Affymetrix arrays that we detected by the other methods; expression is plotted for such GPCRs identified by RNA-seq.

Figure 2

Comparison of GPCR expression levels by RNA-seq, TaqMan GPCR arrays, and Affymetrix (Affy) arrays with independent qPCR and comparisons of expression changes. (a) Data from RNA-seq compared to that of TaqMan arrays; (b) data from Affymetrix HG U133plus2.0 arrays compared to that of TaqMan arrays. Representative data are shown for an individual CAF sample. (c) Validation of TaqMan GPCR array data by qPCR, for N = 5 CAF samples; the data shown are mean and standard error of the mean (SEM) of ΔCt vs 18 S rRNA. (d, e) Correlation between expression ratios of GPCRs in two CAF samples (CAF2 and CAF3) evaluated by (d) RNA-seq and TaqMan arrays and (e) Affymetrix HG U133plus2.0 and TaqMan arrays. (f, g) Number of GPCRs in two CAF samples as detected by (f) TaqMan arrays and (g) Affymetrix HG U133plus2.0 arrays.

Figure 3

GPCR expression in other cell types. (a–c) Number of GPCRs detectable by RNA-seq and TaqMan arrays in individual human lines of (a) primary fetal cardiac fibroblasts, (b) PSCs, and (c) PASMCs. (d−f) The number of GPCRs expressed by the AsPC-1 pancreatic ductal adenocarcinoma cell line (determined by Affymetrix HG U133plus2.0 arrays; CCLE), TaqMan GPCR arrays (Insel Lab), and RNA-seq (CCLE and EBI). (g, h) GPCR expression of MDA-MB-231 breast cancer cells (determined by Affymetrix HG U133plus2.0 arrays and RNA-seq; CCLE). (g) Correlation in the GPCR detection by the two methods for the 62 commonly detected GPCRs. (h) The number of commonly or uniquely identified GPCRs using RNA-seq or Affymetrix HG U133plus2.0 arrays. (i) For CCLE data, expression in AsPC-1 cells of five Gq-coupled GPCRs tested for functional effects in Figure 7, linearized (for Mas5 data) and normalized to the expression of NTSR1, the highest expressed of these receptors as per RNA-seq data.

Figure 4

Comparison of GPCR expression data in TCGA samples generated by Affymetrix arrays and RNA-seq. (a, b) The correlation of expression of commonly detected GPCRs and (c, d) Venn diagrams showing the overlap in GPCR expression of randomly selected tumor samples assessed by Affymetrix HG U133a arrays or RNA-seq of ovarian cancer (OV; TCGA24-1418-01) and LUSC (TCGA-37-4141-01) tumor samples in TCGA. (e) The correlation of expression ratios and (f) Venn diagrams of GPCRs detected by RNA-seq or Affymetrix HG U133a array for a randomly selected pair of TCGA OV samples.

Figure 5

Expression of splice variants of GPCRs in CAFs: CAF3 as an example. (a) The number of GPCRs detected with multiple transcripts expressed at >0.2 TPM expression threshold. (b) As an example, the expression of different transcripts for ADGRE5 (aka CD97). (c) GPCRs with multiple transcripts detected and the number of transcripts expressed at >0.2 TPM for each GPCR.

Figure 6

Gene expression and the dynamic range of detection by different methods. (a) Venn diagram of the detection of all protein-coding genes by Affymetrix HG U133plus2.0 arrays and RNA-seq in pancreatic CAFs. (b) Correlation of expression values for commonly detected genes by both the methods. (c, d) Cumulative distribution functions (CFDs) showing (c) the dynamic range of RNA-seq and Affymetrix arrays (HG U133plus2.0) for all genes; (d) the same as (c), but for GPCRs, detected by RNA-seq, Affymetrix arrays, and TaqMan arrays.

Figure 7

Signaling and functional response to agonists for Gq-coupled GPCRs in AsPC-1 cells. (a) Maximal GPCR agonist-promoted increase in intracellular calcium [“calcium response”, relative to 5 μM ionomycin-induced response (blue line)] for agonists of the indicated GPCRs that are expressed at different TPM in AsPC-1 cells (as determined by RNA-seq in CCLE). Data shown are the mean and SEM from three independent experiments. (b) Concentration–response curves for peak calcium response by the indicated GPCR agonists compared to GPCR expression as in panel (a). Data shown are mean and SEM, from three independent experiments. (c) Kinetics of calcium response by agonist concentrations that yield half-maximal response and kinetics of the ionomycin positive control; data shown are representative from individual wells in a 96-well plate; other replicates showed similar behavior. (d) Impact of treatment with GPCR agonists on the migration of AsPC-1 cells over 24 h; N ≥ 6 for each treatment. Agonist concentrations were: oxytocin (5 μM); histamine (10 μM); 2-Thio-UTP (0.5 μM), neurotensin (0.1 μM); *: p < 0.05; **: p < 0.001; ***: p < 0.0001; significance was evaluated via one-way ANOVA with Tukey multiple comparison testing. (e) The relationship between the increased rate of migration and GPCR expression [as in panel (a)]. (f) The relationship between maximal calcium response promoted by the GPCR agonist concentrations indicated in (d) and the increase in the rate of migration of AsPC-1 cells.