Extensive location bias of the GPCR-dependent translatome via site-selective activation of mTOR

Significance

G protein–coupled receptors (GPCRs) play crucial roles in regulating essential physiological functions when bound by natural or synthetic ligands. Until recently, GPCR signaling was thought to occur solely from the cell surface, but it is now established that receptors can become active inside cells, including at the endosome. While intracellular GPCR signaling drives various physiological and pathological processes, the cellular and molecular mechanisms that underlie these site-selective effects remain poorly understood. Here, we employ a genome-wide approach to uncover a translational regulatory program controlled by the beta-adrenergic receptor. This previously poorly understood response to GPCR signaling is driven exclusively by intracellular receptors. Finally, we uncover that the translational regulation follows site-selective crosstalk between endosomal adrenergic receptors and another important signaling cascade, the mTOR pathway.

Abstract

G protein–coupled receptors (GPCRs) modulate various physiological functions by rewiring cellular gene expression in response to extracellular signals. Control of gene expression by GPCRs has been studied almost exclusively at the transcriptional level, neglecting an extensive amount of regulation that takes place translationally. Hence, little is known about the nature and mechanisms of gene-specific posttranscriptional regulation downstream of receptor activation. Here, we apply an unbiased multiomics approach to delineate an extensive translational regulatory program initiated by the prototypical beta2-adrenergic receptor (β2-AR) and provide mechanistic insights into how these processes are orchestrated. Using ribosome profiling (Ribo-seq), we identify nearly 120 gene targets of adrenergic receptor activity for which expression is exclusively regulated at the level of translation. We next show that all translational changes are induced selectively by endosomal β2-ARs and report that this proceeds through activation of the mammalian target of rapamycin (mTOR) pathway. Specifically, within the set of translational GPCR targets, we find significant enrichment of genes with 5’ terminal oligopyrimidine (TOP) motifs, a gene class classically known to be translationally regulated by mTOR. We then demonstrate that endosomal β2-ARs are required for mTOR activation and subsequent mTOR-dependent TOP mRNA translation. This site-selective crosstalk between the pathways is observed in multiple cell models with native β2-ARs, across a range of endogenous and synthetic adrenergic agonists, and for other GPCRs with intracellular activity. Together, this comprehensive analysis of drug-induced translational regulation establishes a critical role for location-biased GPCR signaling in fine-tuning the cellular protein landscape.

G protein–coupled receptors (GPCRs) are pivotal detectors and transducers of extracellular signals. At the cellular level, GPCRs respond to a range of sensory and chemical stimuli through induction of complex signaling networks that alter cell states. These begin with receptor-induced activation of heterotrimeric G proteins and proceed through the generation of second messengers, activation of effector kinases, and regulation of transcription and translation factors to ultimately rewire gene expression in a stimulus-dependent manner.

Much is known about how GPCR signaling tunes gene transcription via biochemical regulation of select transcription factors (1, 2). Remarkably, it has recently emerged that these cellular responses are also regulated by GPCR localization, whereby the same receptor/G protein complex can elicit distinct responses depending on its subcellular site of signaling. In fact, the activation of transcription factors and regulation of gene transcription were among the first GPCR processes recognized as “location-biased”. Specifically, several receptors have been shown to induce transcriptional signaling from endosomal compartments, as opposed to the plasma membrane (3–5). However, transcription is only the first of many steps that shape a cell’s protein repertoire in response to the environment. A significant amount of gene regulation takes place posttranscriptionally, and dedicated studies and global analyses have demonstrated that transcriptional and posttranscriptional regulatory programs often function as independent regulatory modules (6, 7). Among many notable examples, activity-dependent translation of preexisting mRNAs is a key mechanism underlying functional changes in the neuronal synapse linked to plasticity (8). This underscores the importance of investigating each stage of gene regulation separately. Yet, the nature and mechanisms of gene-specific posttranscriptional regulation downstream of GPCR activation, and how these may be shaped by site-selective receptor activity, are poorly understood.

Here, we leverage ribosome profiling (Ribo-seq) to define the GPCR-induced translatome and provide mechanistic insights into how these processes are orchestrated for a well-characterized receptor/second messenger pathway. We focus on the beta2-adrenergic receptor (β2-AR), a prototypical GPCR that mediates the effects of adrenaline and noradrenaline in the heart, lung, and central nervous system via Gαs-dependent production of cyclic AMP (cAMP) (9–11). We delineate an extensive β2-AR-dependent translational program and demonstrate that it is driven through spatially encoded crosstalk between the receptor and the mTOR pathway.

Results

Ribosome Profiling Accurately Captures Actively Translating RNA.

We turned to a deep sequencing-based experimental platform that uncouples the contributions of transcription and translation to gene expression regulation. Specifically, we utilized a multiomics approach that integrates conventional RNA sequencing to measure mRNA expression (RNA-seq, “transcriptome”), with ribosome profiling to quantify isolated ribosome-protected fragments of RNA (RPFs) and define ribosome positions across each transcript at nucleotide resolution (Ribo-seq, “translatome”) (Fig. 1A). While conventional proteomics methods measure steady-state protein levels, which reflect regulation at the stages of translation and protein stability, Ribo-seq is a powerful method to specifically interrogate genome-wide mRNA translational status that is not confounded by protein stability effects.

Fig. 1.

A snapshot of active global translation downstream of endogenous GPCR activity. (A) Schematic of the multiomics approach. HEK293 cells were treated with 1 µM Isoproterenol for 2 h. Lysates were prepared in parallel for sequencing of total RNA (“mRNA,” RNA-seq) and ribosome-protected fragments (“RPFs,” Ribo-seq). (B) Ribo-seq reads were mapped to genomic feature type using Ribotoolkit (12). (C) Ribo-seq and RNA-seq reads of RPF length (28 to 30 nt) evaluated for three nucleotide periodicity using Ribo-TISH (13). Data are mean of n = 6 biological replicates. Error bars = ± SEM. ****P < 0.0001, **P < 0.01 by one-way ANOVA test with Dunnett (B) or two-way ANOVA test with Tukey (C).

To examine these regulatory processes under native conditions, we selected HEK293 cells as a model, since these express endogenous β2-ARs (14). We stimulated adrenergic receptor signaling with the synthetic agonist isoproterenol for 2 h, and then split each lysate for parallel analysis by RNA-seq and Ribo-seq (Fig. 1A). We observed almost perfect correlation between biological replicates within each experimental setup, indicating high degree of reproducibility (Pearson coefficient > 0.95, P < 2.2 × 10⁻¹⁶, SI Appendix, Fig. S1 A and B). At the same time, replicates across the two platforms had Pearson coefficients of 0.40 to 0.50 (RNA-seq versus Ribo-seq, SI Appendix, Fig. S1C). This confirms that each method is reporting a distinct layer of gene expression regulation. Further, the Ribo-seq experiments successfully captured ribosome-occupied mRNAs based on several quality control metrics. First, we observed enrichment of fragments of characteristic RPF length within coding sequences (CDS) relative to untranslated regions (UTRs) (Fig. 1B and SI Appendix, Fig. S1D). Second, the aligned positions of RPFs revealed the expected stereotypical triplet periodicity within CDS, representing translocating ribosomes by three-nucleotide codons (Fig. 1C). Notably, these trends were neither expected nor observed in the RNA-seq data (Fig. 1C and SI Appendix, Fig. S1 E and F).

Given the high degree of reproducibility and the strong enrichment for known translational features observed in the Ribo-seq experiments, we next set out to identify target genes with significant change in translational status in response to β2-AR activation.

β2-AR Activation Mediates Distinct Transcriptional and Translational Programs.

To define the β2-AR-dependent translatome, we independently identified “transcriptional” and “translational” target genes based on differential expression analysis of the untreated and isoproterenol-stimulated conditions in the RNA-seq and Ribo-seq data, respectively.

Starting with the RNA-seq dataset, we detected 61 genes with isoproterenol-induced changes in mRNA abundance (Dataset S1). These genes subsequently also undergo changes in translation, and their agonist-dependent regulation is therefore classified as “forwarded.” Among these, a significant number of genes were already established as β2-AR transcriptional targets. Specifically, 17/61 genes were previously reported targets of β2-AR signaling by DNA microarray analysis of the same cell line under identical induction conditions (P < 2.2 × 10⁻¹⁶ by Fisher’s exact test). Next, we assessed agonist-stimulated changes in RPF abundance that did not exhibit concordant changes in RNA to pinpoint genes that are subject to translational regulation. For that, we used the change in the ratio of ribosome footprint to mRNA abundance within each gene’s CDS, a metric referred to as differential translation efficiency (ΔTE). We found 133 genes with ΔTE values of at least 1.5 SD above or below the mean that were classified as translationally regulated β2-AR targets (Fig. 2A and Dataset S2). From these, we further distinguished two subcategories of translational targets: “buffered” and “exclusive.” Specifically, we identified 15 translationally “buffered” genes that exhibited significant changes in transcription in response to agonist without corresponding significant changes in translation (Fig. 2A, light blue). On the other hand, the ΔTE for the remaining 118 genes was driven solely by change in RPF abundance with no significant transcription, and therefore these targets were classified as “exclusive” (Fig. 2A, red). Since changes in RNA and RPF abundance were measured at the same timepoint, we aimed to rule out the possibility that translational targets may have undergone transcriptional regulation at an earlier timepoint. For one of the robust β2-AR translational targets, FTH1 (Fig. 2B), we assessed transcript levels using RT-qPCR and protein expression via western blot at multiple time points ranging from 30 min to 6 h post isoproterenol stimulation. These parallel analyses confirmed protein induction without corresponding changes in mRNA levels (SI Appendix, Fig. S2 A and B).

Fig. 2.

Comparison of transcriptional and translational regulation of gene expression in response to GPCR stimulation. (A) Scatter plot depicting changes in mRNA and RPF abundance (log₂FC) following 2 h treatment with 1 µM Isoproterenol (“Iso”). “Exclusive” genes (red) are defined as genes with significant changes in ∆TE and RPF but not mRNA abundance. “Forwarded” genes (dark blue) are defined as genes with significant changes in both mRNA and RPF abundance. “Buffered” genes (light blue) exhibit significant changes in mRNA abundance only. (B) IGV browser tracks showing averaged normalized read coverage for select exons (boxes) and introns (lines) for the “exclusive” target, FTH1 (n = 3 biological replicates). Iso treatment leads to an increase in FTH1 translation with no change in mRNA abundance. The listed p_adj values were determined by the Wald test in DeSeq and log2-fold changes (L2FC) were calculated between Iso-treated (blue) and unstimulated (yellow) samples. The merge of the two tracks (Iso-treated vs Unstimulated) for each assay is shown in green. (C) Enriched Gene Ontology categories for "exclusive” translationally regulated genes (red, Top) and transcriptionally regulated genes (blue, Bottom). FDR < 0.05 was used as cut-off. All underlying data are summarized in Datasets S1–S3.

In all further analyses, we focused on the set of exclusive targets as these represent extensively translationally regulated events. Consistent with an autonomous nature of each gene regulatory mode, we noted several aspects that set apart the β2-AR translational from the transcriptional responses. First, the number of translational targets greatly exceeded the transcriptional repertoire, further underscoring translation as a key regulatory response downstream of GPCRs. Second, while the expression of virtually all transcriptional targets was increased by β2-AR signaling, the translational targets featured both induced and repressed genes (~70 and ~40 genes, respectively; Dataset S2). Presumably, the induction of GPCR-dependent transcription was mediated by CREB (cAMP response element-binding protein) (15). Supporting this established mechanism, our RNA-seq set contained a very significant number of CREB target genes (47/61 genes = 77%, P < 1.0 × 10⁻¹⁹ by Fisher’s exact test). In the case of translation, multiple regulatory pathways may exist allowing for both up- and downregulation of genes downstream of the receptor. Last, to provide additional insights into the functions of each gene set, we performed Gene Ontology (GO) analysis. Comparison of the enriched GO categories revealed distinct biological functions associated with each regulatory mode (Fig. 2C). Downstream of β2-AR stimulation, the induction of genes encoding factors with roles in cell differentiation, RNA polymerase II-dependent transcription, response to hormone stimulus, and MAPK regulation was mediated transcriptionally. On the other hand, genes encoding factors involved in translation, protein targeting, mRNA catabolism, and iron homeostasis were regulated at the level of translation (Fig. 2C and Dataset S3). Furthermore, network analysis of each GO category found that the encoded proteins participate in shared complexes, suggesting translational coregulation among associating proteins (SI Appendix, Fig. S2 C and D). Therefore, β2-AR signaling gives rise to discrete transcriptional and translational programs with unique cellular processes regulated by each mechanism.

Translationally Regulated Genes are Mediated by Endosomal β2-ARs.

The mechanisms that mediate translation downstream of GPCR activation are poorly understood. Following ligand binding, the β2-AR stimulates Gαs-dependent production of cAMP first from the plasma membrane and, upon internalization, subsequently also from early endosomes (3, 16). Therefore, as a first step in disentangling how β2-ARs regulate translation, we investigated whether and how the response may be shaped by receptor localization.

To confine β2-AR activation to the plasma membrane, we used an established pharmacological method to acutely inhibit dynamin-dependent endocytosis with the drug Dyngo-4a (“Dyngo”) (3, 16, 17) (SI Appendix, Fig. S3A). Then, we examined gene expression changes by parallel RNA-seq and Ribo-seq analyses (SI Appendix, Fig. S3B). Comparison between the unstimulated conditions revealed that Dyngo did not impact basal RNA or RPF abundance, supporting that inhibitor treatment alone does not alter steady-state transcription or translation (SI Appendix, Fig. S3 C and D). We previously established that β2-AR internalization is required for efficient activation of the downstream transcriptional responses (3). In agreement, the induction of transcriptional targets identified here through RNA-seq was significantly blunted by Dyngo (P < 1.1 × 10⁻¹⁰ by the Wilcoxon test, SI Appendix, Fig. S3E). This supports that the approach is effective at recapitulating known biology. Remarkably, we observed that blockade of endocytosis also led to pervasive inhibition of the β2-AR-dependent translational regulation (P < 1.7 × 10⁻¹³ by the Wilcoxon test, Fig. 3). Strikingly, we found an even greater dependence of gene translation on internalized receptors compared to gene transcription. Specifically, none of the 118 translational targets identified in cells with normal endocytosis undergo significant changes in translation following endocytic blockade (using P < 0.05 cut-off by the Wald test comparing untreated and isoproterenol-stimulated conditions in the presence of Dyngo). On the other hand, more than a fifth of the β2-AR transcriptional targets (14/61 genes = 23%) were mildly but significantly induced by isoproterenol also in the presence of Dyngo (Dataset S1). Hence, gene translation is regulated exclusively by internalized β2-ARs.

Fig. 3.

The GPCR-dependent translatome is spatially encoded. (A) Heatmap of the exclusive translational targets. Colors represent Log2 fold change (L2FC) between Isoproterenol (“Iso”)-treated vs unstimulated RPFs in the presence of vehicle (DMSO) or 30 µM Dyngo. K-means clustering was performed with the R-package pheatmap.

Endosomal β2-ARs Activate mTOR to Modulate TOP mRNA Translation.

To further pinpoint how endosomal β2-AR signaling mediates protein synthesis, we focused on categories of genes with robust isoproterenol-dependent changes in our Ribo-seq analysis. We recognized that the translation efficiency of mRNAs containing annotated terminal oligopyrimidine (TOP) motifs was disproportionally induced by receptor stimulation (P < 7.7 × 10⁻³ by Kolmogorov–Smirnov test, Fig. 4a), and a significant number of TOP mRNAs were among the set of ~120 robust translational targets (P < 1.0 × 10⁻¹⁰ by Fisher’s exact test). Importantly, this regulation required intact endocytosis (P < 1.6 × 10⁻³ by the Wilcoxon test). TOP mRNAs primarily encode components of the translation machinery and nearly all ribosomal proteins (18). Consistent with this, “Translation” was the most significantly impacted Ribo-seq GO category (P < 1.5 × 10⁻⁵, Fig. 2C). Mechanistically, the synthesis of ribosomal proteins, including TOP mRNAs, is regulated by the mTOR pathway. Therefore, we surmised that endosomal β2-ARs may selectively activate mTOR signaling to modulate the translation of this gene class.

Fig. 4.

Endosomal β2-ARs regulate TOP mRNA translation via mTOR activation. (A) Empirical cumulative distribution function (ECDF) of the differential translation efficiency for all genes in response to stimulation with 1 µM Isoproterenol (“Iso”). The distribution of genes with TOP motifs is shown in red. The resultant p-value was derived by the Kolmogorov–Smirnov test comparing the distributions of TOP and non-TOP motif-containing genes. (B) Schematic of mTOR-dependent regulation of translation via phosphorylation of RPS6 and 4E-BP1. (C and D) Endosomal β2-AR signaling is required for mTOR-dependent phosphorylation of RPS6 and 4E-BP1. HEK293 cells were grown in serum-free medium for 24 h, then pretreated with vehicle (DMSO), 100 nM Torin, or 30 µM Dyngo-4a for 20 min. Cells were then stimulated with 1 µM Iso for 10 min and lysates were subjected to western blot analysis with antibodies against total and phosphorylated RPS6 and 4E-BP1. Fold-change values represent the ratio of phosphorylated/total protein in Iso-treated vs unstimulated conditions. All data are mean of n = 6 to 20 biological replicates. (E) β2-AR signaling activates translation of an mTOR-dependent TOP reporter. HEK293T cells were transfected with TOP-Rluc reporter (schematic, Top) and a control plasmid encoding firefly luciferase (control-Fluc) for 24 h, then treated with 100 nM Torin, 100 nM Insulin, or 1 µM Iso for 6 h. Cells were measured by dual-luciferase assay and reporter translational induction was calculated as Rluc/Fluc and shown as L2FC (stimulated vs unstimulated cells) (Bottom). All data are mean of n = 18 biological replicates. (F) Endosomal β2-AR signaling activates TOP-mRNA translation via mTOR. HEK293T cells were transfected as in (E), pretreated with vehicle (DMSO), 100 nM Torin, or 30 µM Dyngo for 20 min, then stimulated with 1 µM Iso for 6 h. Data are shown as percent of the maximum response measured with Iso in the presence of vehicle. All data are mean of n = 18 biological replicates. Error bars ± SEM. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05 by unpaired Student's t test comparing vehicle to inhibitor-treated conditions in (D) and one-way ANOVA test with Dunnett in (E and F).

To test this, we began by interrogating whether β2-AR signaling impacts mTOR activity. Eukaryotic Initiation Factor 4E-Binding Protein 1 (4E-BP1) is phosphorylated directly by mTOR on residues Thr37/46, and ribosomal protein 6 (RPS6) is phosphorylated on sites Ser235/236 and Ser240/244 (19, 20) (Fig. 4B). Therefore, we monitored accumulation of phosphorylated 4E-BP1 and RPS6 by western blot analysis as markers of mTOR activation. Acute stimulation of β2-AR with isoproterenol led to significant induction of 4E-BP1 and RPS6 phosphorylation (Fig. 4 C and D, black bars). These events were dependent on mTOR, as pretreatment with the ATP-competitive inhibitor, Torin (21), abolished the regulation (Fig. 4 C and D, blue bars). The mTOR response also required PKA activity downstream of the β2-AR, consistent with a necessity for cAMP in this process (SI Appendix, Fig. S4A). More importantly, acute application of Dyngo completely blocked mTOR-induced phosphorylation of 4E-BP1 and RPS6 in response to GPCR agonist (Fig. 4 C and D, red bars). This supports that β2-AR endocytosis is essential to activate the mTOR pathway.

Next, we aimed to establish a connection between mTOR induction by intracellular receptors and the regulation of TOP gene translation. EEF2 was among the TOP mRNAs significantly upregulated by β2-AR signaling based on the Ribo-seq analysis without concomitant changes in mRNA levels (Dataset S2 and SI Appendix, Fig. S4B). Therefore, as an orthogonal approach to monitor TOP mRNA translation, we obtained a validated reporter which utilizes Renilla luciferase expression downstream of the EEF2 5’UTR harboring its TOP sequence (“TOP-Rluc”) (22). As expected, treatment with Torin alone led to TOP-Rluc repression, while insulin yielded a significant increase (Fig. 4E). These represent responses to inhibition and activation of mTOR that are independent of GPCR signaling. Next, we assessed reporter levels under β2-AR induction conditions. TOP-Rluc expression was induced by isoproterenol in a PKA-dependent manner, and the magnitude of this change was greater than that following mTOR activation with insulin (Fig. 4E and SI Appendix, Fig. S4C). These reflect reporter translation, because neither isoproterenol nor insulin impacted TOP-Rluc mRNA levels (SI Appendix, Fig. S4D). We additionally verified that this is the case using a pharmacological inhibitor of transcription, which had no effect on reporter induction by β2-AR activation (SI Appendix, Fig.S4E). We then asked whether mTOR signaling is required for this regulation. Coapplication of Torin drastically repressed the isoproterenol-dependent accumulation of TOP-Rluc (Fig. 4F, blue bar). In support of the regulatory significance of the TOP motif, an analogous luciferase reporter which harbors a mutated TOP motif within its UTR (22) displayed minimal β2-AR-dependent accumulation that was not affected by Torin (“TOPmut-Rluc,” SI Appendix, Fig. S4F). Last, endocytic blockade completely precluded the translational induction of TOP-Rluc in isoproterenol-stimulated cells (Fig. 4f, red bar). On the other hand, Dyngo had no impact on the insulin-dependent regulation of TOP-Rluc (SI Appendix, Fig. S4G), corroborating a critical role for internalized β2-AR activity in this process.

A Conserved Requirement for Intracellular GPCR Signaling in mTOR Regulation.

We next interrogated whether the coupling between intracellular GPCR signaling and mTOR activation represents a broadly conserved mechanism.

First, we asked whether this regulation is ligand-specific. To this end, we stimulated the β2-AR with additional agonists in the presence or absence of receptor endocytosis, including endogenous (epinephrine, norepinephrine) and synthetic (formoterol) ligands. Across all stimulation conditions, we observed robust agonist-driven TOP-Rluc induction exclusively in cells with intact trafficking, which closely paralleled isoproterenol treatment (Fig. 5A).

Fig. 5.

Intracellular GPCR signaling regulates mTOR and TOP gene translation. (A) Endosomal β2-AR drive TOP mRNA translation in response to different adrenergic agonists. HEK293T cells were transfected with TOP-Rluc and control-Fluc for 24 h, then pretreated with 30 µM Dyngo or Vehicle (DMSO) for 20 min. Cells were treated with the β2-AR agonists 1 µM epinephrine (Epi), 1 µM norepinephrine (Norepi), or 100 nM formoterol (Form) for 6 h. Cells were measured by dual-luciferase assay, reporter translational induction was calculated as Rluc/Fluc and shown as L2FC (stimulated vs unstimulated cells). All data are mean of n = 6 biological replicates. (B) β2-AR activates mTOR in endocytosis-dependent manner in NRVMs. NRVMs were serum-starved for 48 h, and then pretreated with vehicle (DMSO) or 30 µM Dyngo for 20 min. Cells were then stimulated with the selective β2-AR agonist, 100 nM formoterol (Form), for 30 min and lysates were subjected to western blot analysis with antibodies against phosphorylated and total 4E-BP. Fold-change values are ratios of phosphorylated/total protein in Form-treated vs unstimulated conditions. All data are mean of n = 6 biological replicates. (C) Endosomal VIPR1 activates TOP mRNA translation. HEK293T cells were transfected and pretreated as in (A), then stimulated with 500 nM vasoactive intestinal peptide (VIP) for 6 h. Data are mean of n = 12 biological replicates. (D) Intracellular β1-ARs activate TOP mRNA translation. HEK293T cells were transfected with TOP-Rluc, control-Fluc, and β1-AR for 24 h, and then pretreated with vehicle (DMSO), 100 µM metoprolol (Met) for 20 µM sotalol (Sot) for 20 min. Cells were then stimulated with 100 nM dobutamine (Dob) for 6 h. Data are mean of n = 6 biological replicates. (E) Model: Intracellular GPCR/cAMP signaling drives TOP mRNA translation via site-selective activation of mTOR. Error bars ± SEM. ****P < 0.0001, ***P < 0.001, *P < 0.05, by unpaired Student’s t test comparing vehicle to inhibitor-treated conditions.

To ensure that these results apply to a more physiologically relevant context for the β2-AR, we next turned to neonatal rat ventricular myocytes (NRVMs). A recent study demonstrated that β2-ARs can signal from endosomes of NRVMs (23) and, therefore, we asked whether this intracellular fraction is required to activate mTOR. Stimulation of the β2-AR with the subtype-selective agonist, formoterol, induced mTOR signaling. Notably, this activation was blocked by Dyngo (Fig. 5B). These results support that the pathway crosstalk occurs in response to intracellular receptor activity in multiple cell models expressing native β2-ARs.

Last, we examined two additional Gαs-coupled GPCRs known to signal from intracellular organelles, VIPR1 and β1-AR. Similar to the β2-AR, the vasoactive intestinal peptide receptor 1 (VIPR1) signals via cAMP production at the plasma membrane and from endosomes following ligand-induced internalization (24). It is natively expressed in the kidney and, therefore, endogenously present and functional in HEK293 cells (24). Activation of VIPR1 with vasoactive intestinal peptide (VIP) led to accumulation of the TOP reporter, which was blocked by preincubation with a dynamin inhibitor (Fig. 5C), supporting that endosomal VIPR1s are required for mTOR activation. The beta1-adrenergic receptor (β1-AR) can be activated by ligands at the cell surface or at the Golgi (25). Unlike the β2-AR or VIPR1, the Golgi fraction of β1-ARs is delivered from the biosynthetic pathway. Therefore, to functionally distinguish plasma membrane activity from internal activity for this receptor, we employed a pharmacological strategy. We stimulated β1-ARs throughout the cell using the permeable adrenergic agonist dobutamine and compared its effects on mTOR to those upon cotreatment with either an impermeable antagonist (sotalol, blocks only plasma membrane β1-AR signaling) or permeable antagonist (metoprolol, blocks all β1-AR signals). Cells coexpressing β1-AR and TOP-Rluc exhibited efficient reporter induction when treated with dobutamine, which was fully occluded by coapplication of metoprolol (Fig. 5D, purple). In contrast, sotalol did not have a significant inhibitory effect on TOP-Rluc induction (Fig. 5D, orange). Thus, intracellular β1-AR activation is similarly required for mTOR regulation.

In sum, these results support that the crosstalk between intracellular GPCR/cAMP signaling and the mTOR pathway spans various receptors and cell types, highlighting a mechanism of widespread relevance (Fig. 5E).

Discussion

Numerous extracellular signals affect physiological responses by binding to GPCRs, leading to changes in gene expression. However, understanding of how GPCRs influence the capacity of cells to dynamically regulate protein content in response to stimuli has remained limited, as studies on receptor-mediated gene expression have focused almost exclusively on transcription. The present work takes advantage of the sensitivity and high-throughput nature of Ribo-seq to delineate an extensive regulatory program orchestrated by a GPCR that consists of nearly 120 genes whose translation is regulated independently of mRNA levels. These genes have not been previously associated with β2-AR and cAMP signaling. Notably, at a matched time point, the repertoire of translationally regulated genes exceeds the transcriptional one, and each set orchestrates distinct biological functions and processes (Fig. 2 and Dataset S3). Therefore, by focusing on gene translation and its regulation by receptor activity, these findings provide insights into the comprehensive mechanisms by which GPCR signaling shapes cell states.

Over the past decade, it has been recognized that GPCR signaling is subject to intricate spatial regulation, and that the selective modulation of localized receptors can have distinct effects on physiology (26). Despite this growing appreciation for the importance of compartmentalized GPCR signaling, how different subcellular fractions of receptors shape unique cellular behaviors has remained poorly understood. Our findings indicate that a salient molecular mechanism involves the exclusive coupling between gene translation and endosomal receptors (Fig. 3). We further define how intracellular GPCRs mediate these processes by demonstrating that the translational induction of TOP mRNAs by agonist proceeds through receptor endocytosis-dependent activation of mTOR (Figs. 4 and 5). Of note, an interplay between the GPCR and mTOR signaling pathways has been documented previously. In some contexts, GPCR activation appears to regulate mTOR via β-arrestin-dependent mechanisms (27, 28), and in others- via cAMP/PKA activity (29–33). Our data are consistent with the majority of studies showing that the pathway crosstalk proceeding via PKA (SI Appendix, Fig. S4). Interestingly, the reported impact of receptor signaling on mTOR varies depending on the study, with some demonstrating that cAMP stimulates mTOR (29, 33), while others support inhibition (31, 34). Typically, these discrepancies have been attributed to differences in experimental models and/or experimental design (35). However, the significance of localized receptor activity in the regulation of mTOR was not considered. Our findings ascertain signal compartmentalization as a pertinent axis of regulation underlying the crosstalk between these two pathways. Hence, we speculate that this unappreciated location-biased nature of the β2-AR/mTOR interplay could be a critical parameter in further resolving the seemingly contradictory findings regarding whether cAMP accumulation activates or inhibits mTOR. There are known precedents in the context of GPCR crosstalk with other cascades that support such model. For example, distinct subcellular pools of the same second messenger were shown to exert opposing effects on phospholipase C-ε activity (36). It could similarly be the case that cAMP levels are not monotonically related to mTOR status, and the regulation is dictated by the spatiotemporal dynamics of the signal instead. In this study, we tested three Gαs-coupled GPCRs with intracellular activity and found that, in all cases, internal receptor signaling triggers TOP gene translation (Figs. 4 and 5). In contrast, we note that multiple studies that reported cAMP-dependent inhibition of mTOR have relied on experimental approaches that induce superphysiological cAMP levels, such as adenylyl cyclase stimulation with high doses of forskolin combined with phosphodiesterase inhibitors (31, 37, 38). It is well established that these are saturating activation regimes that abolish GPCR and cAMP compartmentalization (39). Therefore, it would be imperative to systematically dissect how factors and/or genetic contexts that lead to loss of GPCR/cAMP compartmentalization impact the interplay with mTOR activation and, subsequently- translational regulation of gene expression.

Another unresolved puzzle arising from our findings is how GPCR activity and its spatial bias modulate the mTOR pathway. Despite the documented examples of interplay between the cascades, a commonly agreed-upon detailed mechanistic model linking GPCRs and mTOR is mostly lacking (29, 33, 34). Previous reports have suggested that the GPCR effects are independent of several known mTOR regulatory inputs, including Rheb, the Rheb GAPs TSC1/2, AMPK, or Rag GTPases (31, 34). Instead, the regulation may proceed through direct PKA-dependent phosphorylation of mTOR complex 1 (mTORC1) (29, 33, 40), although it is not clear whether and which of the documented events take place across multiple contexts, and which are context-specific. For example, it was shown that PKA phosphorylates the mTOR kinase itself in one model (31), but not in another (34). Similarly, while PKA can phosphorylate the mTORC1 negative regulator PRAS40 in vitro (33), it remains to be determined whether this takes place in cells. One mechanism that has garnered independent support from multiple groups involves the phosphorylation of RAPTOR (Regulatory-Associated Protein of mTOR) by PKA. This process was found to be required in the context of both activation and inhibition of mTOR by cAMP (31, 34), although it is unknown how this particular phosphorylation of RAPTOR alters mTORC1 activity. Nevertheless, a recent report uncovered that this is dependent on the scaffolding protein, AKAP13 (A-kinase anchor protein 13), which brings PKA in physical proximity to mTORC1 (40). AKAPs are of recognized significance in defining cAMP/PKA compartmentalization through the scaffolding of multimolecular “signalosomes” (41). Interestingly, in our hands, inhibition of AKAP/PKA anchoring with a cell-permeable peptide, Ht31, significantly diminished β2-AR-dependent translational regulation of TOP reporter and its dependence on endosomal receptors (SI Appendix, Fig. S4H). Therefore, it is tempting to speculate that the AKAP13/PKA/RAPTOR complex may render that “nanodomain” of PKA selectively responsive to increase in local cAMP concentrations and, thus, play an important role in establishing the location bias of GPCR-dependent mTOR regulation. It is pertinent to examine this directly and to further dissect the precise mechanisms through additional candidate-based and unbiased approaches.

In summary, our study provides a comprehensive analysis of how GPCR signaling impacts gene translation and presents mechanistic insights into these processes. Ultimately, this work extends the concept of compartmentalized receptor activity to a previously underexplored layer of gene expression regulation to fundamentally enrich the current understanding of the remarkable diversity of endogenous cellular responses and their location bias. While we demonstrate mTOR as one of the downstream pathways through which intracellular receptors exert translational control, much remains to be learned about the precise molecular underpinnings of this interplay. The mTOR pathway is a master regulator of cell growth and proliferation that is impaired in many pathological conditions, including cancer, metabolic disorders, and neurodegeneration (42). Given the equally critical functions of GPCRs as coordinators of human physiology and pathophysiology and their immense pharmacological potential (43), further mechanistic dissection of the GPCR/mTOR crosstalk may help identify candidates for therapeutic targeting in the context of human diseases arising from dysfunctions in these pathways. Along these lines, endosomal β2-ARs likely also mediate gene translation via mTOR-independent mechanisms. The elucidation of additional mechanisms linking GPCR signaling to gene translation is therefore another key avenue for future exploration. Last, we envision that these analyses can be extended to investigate how GPCR activation impacts global as well as local mRNA translation in other cell systems, including neurons, where protein synthesis from preexisting mRNAs is of recognized physiological importance.

Materials and Methods

Chemicals and Antibodies.

(−)- Isoproterenol hydrochloride (Sigma-Aldrich, Cat #I6504) was dissolved in water/100 mM ascorbic acid to 10 mM stock and used at 1 μM final concentration. Dyngo-4a (Abcam, Cat #120689) was resuspended in DMSO to 30 mM stock and used at 30 μM final concentration. Torin-1 (MedChemExpress, Cat #HY-13003) was resuspended to 100 μM in DMSO and used at a final concentration of 100 nM. Insulin (Sigma -Aldrich, Cat #91077C) was resuspended to 1 mM in water/HCl (pH 2.0) and used at a final concentration of 100 nM. Coelenterazine (GoldBio, Cat #CZ) was resuspended to 10 mM in ethanol and stored protected from light. D-luciferin (Goldbio, Cat# LUCNA-100) was resuspended in 10 mM HEPES buffer and used at 150 μg/mL final concentration. H89 (Cayman Chemical, Cat #10010556) was dissolved in DMSO to 10 mM and used at 10 μM final concentration. sc-Ht31 ammonium (MedChemExpress, Cat #HY-P2624A) was resuspended in DMSO and used at 20 μM final concentration. Dobutamine hydrochloride (Sigma-Aldrich, Cat #D0676) was resuspended in DMSO and used at 100 nM final concentration. Metoprolol tartrate (Sigma-Aldrich, Cat #M5391) was resuspended in water and used at 100 μM final concentration. Sotalol hydrochloride (Sigma-Aldrich, Cat #S0278) was resuspended in water and used at 20 μM final concentration. Norepinephrine (Sigma-Aldrich, Cat #A7257) was resuspended in water/100 mM ascorbic acid and used at 1 μM final concentration. Epinephrine (Sigma-Aldrich, Cat #E4250) was resuspended in water/100 mM ascorbic acid and used at 1 μM final concentration. Formoterol fumarate dihydrate (Sigma-Aldrich, Cat # F9552) was resuspended in DMSO and used at 100 nM final concentration. α-Amanitin (MedChemExpress, Cat # HY-19610) was resuspended in water and used at 50 μg/mL final concentration. VIP (Sigma-Aldrich, Cat #V3628) was resuspended in 1% acetic acid and used at 500 nM final concentration.

RPS6 mouse monoclonal antibody (Cell Signaling Technologies, Cat #2317), phospho-RPS6 (S235/S236) rabbit monoclonal antibody (Cell Signaling Technologies, Cat #4858), phospho-RPS6 (S240/S244) rabbit monoclonal antibody (Cell Signaling Technologies, Cat #5364), 4E-BP1 rabbit monoclonal antibody (Cell Signaling Technologies, Cat #9644), phospho-4E-BP1 rabbit monoclonal antibody (T37/T46) (Cell Signaling Technologies, Cat #2855), FTH1 rabbit monoclonal antibody (Cell Signaling Technologies, Cat #4393), and α-Tubulin mouse monoclonal antibody (Cell Signaling Technologies, Cat #3873) were all used at 1:1,000 for western blots. The following secondary antibodies were used at 1:10,000 dilutions: donkey anti-mouse-680 (LICOR Biosciences, Cat #926-68072) and donkey anti-rabbit-800 (LICOR Biosciences, Cat #926-32213).

Cell Culture.

HEK293 cells were obtained from ATCC and HEK293T cells were obtained from Clontech. Cell lines were grown at 37 °C/5% CO2 in Dulbecco’s Modified Eagle Medium (ThermoFisher Scientific, Cat #11965118) supplemented with 10% fetal bovine serum (Sigma-Aldrich, Cat #F2442). HEK293 cells with stable Flag-β2-AR for internalization assays were previously described (44).

Neonatal rat ventricular myocytes (NRVMs) were isolated as previously described (45). Hearts were excised from 2-day-old Sprague-Dawley rats, removed, and washed in D-PBS. Debris and atria were removed from the hearts and the remaining ventricular tissue was minced by razor. The minced heart tissue was digested to form a single-cell suspension with an ADS buffer (116 mM NaCl, 20 mM HEPES, 80 μM Na₂HPO₄, 56 mM glucose, 5.4 mM KCl, 80 mM MgSO₄; pH 7.4) and an enzyme solution containing Pancreatin (Thermo, Cat #J19880.28), Collagenase II (Thermo, Cat #17101015), and 50 nM CaCl₂. A series of three enzymatic digestions were done before cells were strained and resuspended in Ham’s F-10 Complete Medium (Gibco, Cat #11550043) with 10% Horse serum, 5% FBS, and 1% Pen-Strep. Cells were preplated for 2 h to remove endothelial cells and fibroblasts which adhere faster than myocytes. The nonadhered myocyte population was removed and counted before plating on matrigel-coated (VWR, Cat #47743) 6-well plates at 1 million cells/well in F-10 Complete Medium with horse serum. The next day the cells were changed into a Reduced Serum F-10 Complete Medium (5% FBS, 1% Pen-Strep).

Ribosome Profiling.

Samples were harvested and ribosome footprints generated using the standard strategy of McGlincy and Ingolia (46) with slight modifications to facilitate the use of commercial kits for rRNA depletion and sequencing library preparation. The monosome isolation step was eliminated, as previous work has shown that size selection for RNA fragments with lengths characteristic of RPFs (~25 to 35 nt) is sufficient for capturing actively translating mRNA (47).

HEK293 cells were plated in 15 cm dishes for 70% confluency on the day of the experiment. For endosome inhibition experiments, cells were preincubated for 20 min in serum-free DMEM with either 30 µM Dyngo-4a or DMSO. Cells were stimulated with 1 µM isoproterenol in serum-free DMEM for 2 h, then rinsed twice with ice-cold PBS (GenClone, Cat #25-507) and lysed in-dish with 400 µL freshly prepared lysis buffer [20 mM Tris (Sigma, Cat #T2194), 150 mM NaCl (Thermo, Cat #AM9759), 5 mM MgCl₂ (Thermo, Cat #AM9530G), 1 mM DTT (Sigma-Aldrich, Cat #43816), 1% Triton X-100 (Acros Organics, Cat #AC32737100), 100 µg/mL cycloheximide (Sigma-Aldrich, Cat #C4859), and 25 U/mL TURBO DNase (Thermo, Cat #AM2238)]. Following incubation in lysis buffer for at least 10 min on ice, lysates were triturated 10 times through a 25G needle and then clarified by centrifugation (10 min/20,000xg/4 °C) prior to flash-freezing in liquid nitrogen and storage at −80 °C. An aliquot of lysate for each sample was purified with the Zymo RNA Clean & Concentrator-25 kit (Genesee, Cat #11-353) following the manufacturer’s protocol for later use in preparing poly-A selected mRNA libraries.

An aliquot of lysate containing 30 µg total RNA was taken for input to the ribosome profiling workflow. After bringing the volume to 200 µL with freshly prepared polysome buffer (lysis buffer without cycloheximide or DNase), lysates were incubated with 15 U (1.5 µL) RNase I (VWR, Cat #76081-704) for 45 min at room temperature with gentle agitation. The digestion reaction was terminated by adding 200U (10 µL) SUPERase RNase inhibitor (Thermo, Cat #AM2696) and 20 µL 10% SDS (Promega, Cat #V6551) and then purified using the Zymo RNA Clean & Concentrator kit (Genesee, Cat #11-353/11-325), following manufacturer’s modified protocol to isolate small RNAs (<200 nt).

The entirety of the eluate was input to an end repair reaction for 1 h at 37 °C to generate 3’-OH/5’-PO4 RNA fragments compatible with downstream adapter ligation. Final reaction volume of 40 µL contained ~24 µL of sample, 40U (4 µL) T4 polynucleotide kinase enzyme (T4 PNK; NEB, Cat #M0201S), 4 µL 10X T4 Ligase Buffer (NEB, Cat #B0202S; final reaction concentrations of 50 mM Tris-HCl, 10 mM MgCl₂, 1 mM ATP, 10 mM DTT, pH 7.5), 4 µL 10 mM ATP (NEB, Cat #P0756S), 40U (2 µL) SUPERase inhibitor, and nuclease-free water. Reactions were terminated by transferring the tube to ice, and samples were purified with the same modified protocol for the Zymo RNA Clean & Concentrator kit. A supersaturating final reaction concentration of 2 mM ATP was chosen because T4 PNK enzymatic activity is irreversible and therefore, flooding the reaction serves to minimize the impact of between-sample differences in digestion or total RNA quantity on availability of fragments compatible with downstream adapter ligation.

Ribosome-protected fragments (RPFs) between ~25 to 35 nt were size-selected by gel electrophoresis on 15% TBE-Urea gels (Bio-Rad, Cat #4566055), then eluted via crush-and-soak extraction in buffer containing 300 mM NaOAc (Thermo, Cat #AM9740), 1 mM EDTA (Thermo, Cat #AM9260G), 10% SDS (Promega, Cat #V6551). Crushed RPF-containing bands were eluted overnight at 4 °C, then slurries were centrifuged through Spin-X cellulose filters (Sigma-Aldrich, Cat #CLS8163) at 14,000 RPM for 2 min. To precipitate RNA from clarified eluates, 20 µL of 3 M NaOAc and 2 µL Glycoblue (Thermo, Cat #AM9515) were added and mixed well, followed by addition of 500 µL isopropanol and further vigorous mixing before overnight precipitation at −80 °C. Then, eluates were centrifuged for 1 h at 21,100×g at 4 °C and pellets containing RPFs were rinsed once with ice-cold freshly diluted 80% ethanol, air dried by 5 min Speedvac at room temperature, and resuspended in 14 µL nuclease-free water.

Samples were depleted of rRNA using the Human riboPOOL for Ribosome Profiling RNA kit (Galen Molecular, Cat #dp-K024-000042) with a modified protocol omitting the 50 °C/5 min incubation following magnetic bead addition to the hybridized RNA-rRNA oligo reaction. After aspirating supernatant away from rRNA-capturing magnetic beads, RPFs were further purified using the small RNA purification protocol for the Zymo RNA Clean & Concentrator kit and Zymo-5 columns and eluted in 6 µL nuclease-free water. The entire yield was used to generate sequencing libraries following the manufacturer’s protocol for NEBNext Multiplex Small RNA Library Prep Set for Illumina (New England Biolabs, Cat #E7300S), using 12 PCR cycles. The final product was purified using the Zymo DNA Clean & Concentrator-5 kit (Genesee, Cat #11-302) following a modified protocol for small DNA fragments (<200 nt), and ~140 to 150 nt DNA fragments were then size-selected by gel electrophoresis with 5% TBE gels (Bio-Rad, Cat #4565015) following NEB library prep kit protocol. Libraries were sequenced with Illumina HiSeq 4000.

Polyadenylated mRNA Library Preparation.

Total RNA was isolated from an aliquot of each lysate above using the manufacturer’s standard protocol for the Zymo RNA Clean & Concentrator-25 kit (Genesee, Cat #11-353) and 1 µg total RNA was used as input to NEBNext Poly(A) mRNA Magnetic Isolation Module for NEBNext Ultra II RNA Library Prep Kit for Illumina (New England Biolabs, Cat #E7775), following the manufacturer’s protocol and using 7 PCR cycles for library amplification. Additional unpaired mRNA libraries were prepared using NEBNext Ultra II Directional RNA Library Prep Kit for Illumina, following the manufacturer’s protocol (New England Biolabs, Cat #E7776S). Libraries were sequenced on Illumina HiSeq 4000.

RNA-seq and Ribo-seq Data Processing.

Sequencing data were trimmed BBDuk (BBMap version #38.63) to remove adaptors. Sequences aligning to ncRNA, tRNA, and rRNA were removed using Bowtie (version 2.3.5) and cleaned files were aligned to the Genome Reference Consortium Human Build 38 (GRCh38) using STARaligner (version 2.7.5). FeatureCounts (Subread version 1.6.3) was used to align and quantify coding sequences, with RPF reads being narrowed to reads of length 26 to 34 nucleotides.

DESeq2 (version 3.16) was used for differential expression analysis of raw reads. Significant changes at either the RNAseq or Ribo-seq level were determined by the Wald test with an adjusted p-value of 0.05 (48). For ΔTE calculations, unstimulated RNA samples were used as an additional covariate in the DESeq input as described in (49). Genes with significant changes in mRNA or RPFs were further categorized based on their respective ΔTE values. “Exclusive” genes were defined as genes with ΔTE values at least 1.5 SD above or below the mean in the same direction as the corresponding significant RPF change (P < 0.05 by the Wald test). “Buffered” genes were defined as genes with ΔTE values less than 1.5 SD below the mean in the same direction as the corresponding significant mRNA abundance change (P < 0.05 by the Wald test). The remaining genes with significant mRNA changes were categorized as “Forwarded.”

Ribotish and Ribotoolkit were used for periodicity and feature-type analysis of the Ribo-seq data, respectively (12, 13). Feature count analysis of RNAseq data was performed using GenomicFeatures in R (50). Correlation analysis to compare Pearson correlation coefficients for log-transformed counts between conditions and across library preparations was performed using bigPint (51). For gene ontology analysis, categorized hits were entered into the GOrilla GO term analysis software and significantly regulated pathways were determined by FDR < 5%. Protein interaction matrices were generated by String-DB (52) and analyzed in Cytoscape (53). Genomic alignment traces were generated in Interactive Genomics View (IGV) (54) by combining coverage BAM files and overlaying in Adobe Illustrator. Nonuniform coverage of Ribo-seq results reflects the variation in ribosome distribution across translating transcripts. These can arise from various translational regulatory mechanisms including ribosome pausing, codon bias, structured RNA elements.

TOP-Rluc Reporter Assay.

HEK293T cells were seeded on 6-well dishes for 24 h, then transfected with TOP-Rluc or TOPmut-Rluc and control Firefly Luciferase plasmids using Lipofectamine-2000 (Invitrogen, Cat #11668027) following recommended protocols. For experiments with β1AR, HEK293T cells were additionally transfected with flag-β1AR plasmid. After 24 h, cells were lifted with Accutase (Gibco, Cat #A1110501), pelleted, and resuspended in serum-free DMEM. Cells were split across black 96-well assay plates (Corning, Cat #3603) and incubated with inhibitor or DMSO for 20 min as described. For experiments involving transcriptional inhibition, cells were preincubated with the inhibitor α-Amanitin for 6 h. For experiments with PKA/AKAP disruption, cells were preincubated with the cell-permeable peptide inhibitor Ht31 and/or Dyngo for 1 h. Following 6 h stimulation with agonist, cells were incubated with 150 μg/mL D-luciferin for 40 min and an initial Firefly luciferase reading was taken. Afterward, cells were lysed in Renilla luciferase lysis buffer (4 mM EDTA (Sigma, Cat #ED4SS), 1% glycerol (Sigma, Cat #G5516), 0.1% Triton X-100 (Sigma, Cat #X100), 50 mM HEPES, 1 mM DTT, and 8 µM coelenterazine). To measure luciferase signal, the plate was read on the Hamamatsu μCell plate reader at 37 °C. Samples were normalized by dividing Renilla luciferase by Firefly luciferase, and the normalized values of treated samples were compared to untreated controls.

Quantitative Real-Time PCR.

HEK293T cells were seeded on 6-well dishes for 24 h. For measurements of reporter mRNA expression cells were transfected with TOP-Rluc and Firefly Luciferase plasmids for 24 h. Cells were treated with isoproterenol for 1 to 6 h in serum-free media. Total RNA was extracted from samples using the Quick-RNA Mini-Prep Kit (Zymo, Cat #11-328). Reverse transcription was carried out on 1,000 ng purified total RNA with SuperScript II Reverse Transcriptase (Thermo, Cat #18064071). Power SYBR Green (Applied Biosystems, Cat #4368706) and primers described below were used to determine expression of target genes. All target gene expression levels were normalized to housekeeping gene, GAPDH.

• PCK1 Forward: CATTGCCTGGATGAAGTTTGACG

• PCK1 Reverse: GGGTTGGTCTTCACTGAAGTCC

• FTH1 Forward: TGAAGCTGCAGAACCAACGAGG

• FTH1 Reverse: GCACACTCCATTGCATTCAGCC

• EEF2 Forward: CCTCTACCTGAAGCCAATCCAG

• EEF2 Reverse: CCGTCTTCACCAGGAACTGGTC

• Rluc Forward: TCATGGCCTCGTGAAATCCCGT

• Rluc Reverse: GCATTGGAAAAGAATCCTGGGTCCG

• GAPDH Forward: CAATGACCCCTTCATTGACC

• GAPDH Reverse: GACAAGCTTCCCGTTCTCAG

β2-AR Internalization Assay.

HEK293 cells with stable β2-AR were seeded on 12-well plates and grown to 80% confluency. Cells were treated with 30 µM Dyngo-4a or vehicle (DMSO) for 20 min in serum-free DMEM, then treated with 1 µM isoproterenol for 20 min to induce receptor internalization. Cells were washed with PBS, then stained for 1 h at 4 °C in PBS with 1:1,000 Alexa-647 conjugated M1 mouse anti-FLAG antibody (Invitrogen, Cat #MA1-91878). Cells were moved to flow cytometry (BD FACS Canto2), and 10,000 cells were analyzed per sample. The mean Alexa 647 of the singlet population was used to quantify surface β2-AR. The fraction of internalized receptors was calculated as 100% − (Iso surface β2-AR)/(Unstimulated surface β2-AR) × 100%.

Protein Extraction and Western Blot.

Cells were serum-starved for 24 to 48 h, and then pretreated with vehicle (DMSO), 100 nM Torin, or 30 µM Dyngo-4a for 20 min. Cells were then stimulated with 1 µM Iso for 10 min (HEK293 cells) or 100 nM Form for 30 min (NRVMs) and lysed with RIPA buffer (Sigma, Cat #R0278) containing protease inhibitors cocktail (Sigma, Cat #P8340), phosphatase inhibitors cocktail (Sigma, Cat #P0044), and 0.1 μM PMSF (Sigma, Cat #P7626). Lysates were then transferred to microcentrifuge tubes and spun for 15 min at 14,000 RPM at 4 °C and the supernatant was used for western blot experiments. Protein concentration was determined by Pierce BSA Protein Assay Kit (Thermo Fisher, Cat #23225). Protein samples were prepared for western blot analysis by adding 4x sample buffer (Bio-Rad, Cat. #1610747) and 1 mM 2-mercaptoethanol (Sigma, Cat #M6250). Samples were run on a 4 to 20% MINI-PROTEAN TGX Stain-Free gels (Bio-Rad, Cat #4568093) and transferred to a nitrocellulose membrane for 2 h at 85 V in 4 °C. Membranes were blocked with Intercept Blocking Buffer (LICOR Biosciences, Cat #927-80001) for 1 h and incubated with primary antibodies/TBST against proteins of interest for 3 h at room temperature. Afterwords, membranes were washed, incubated with secondary antibodies for 1 h, and visualized using the Odyssey imager system (LICOR). Bands were quantified and analyzed using ImageStudio software.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Data, Materials, and Software Availability

Sequencing FASTQ Files data have been deposited in Gene Expression Omnibus (GSE269144) (55).