GPCR activation in U2OS cells by selective agonists

(A) Dose-dependent cAMP accumulation after ADORA1 activation with agonist, CCPA, or inhibition with antagonist DPCPX (5μM). (B) Dose-dependent IP1 accumulation after PEA stimulation of HRH1 and inhibition by Cetirizine (1μM). (C) ALP activity after stimulation of FZD4 by Wnt3a. (D) The ratio of phosphorylated ERK1/2 (pERK1/2) by total ERK 1/2 (tERK1/2) after stimulation of ACKR3 by SDF-1α. For all graphs, data are shown as mean ± SD, with n=3 independent repeats, each having duplicate determinants, ns=not significant, *p < 0.05, **p < 0.01, ***p < 0.001 vs vehicle control (VC). Statistical significances were determined by one way or two-way ANOVA of receptors by agonists or antagonists and post-hoc Tukey or Sidak testing for multiple comparisons.

Characterization of EVs isolated from U2OS cell culture media

(A) Representative immunoblots of EVs isolated by SEC from U2OS cell culture media (U2) but not from media control (MC) show detection of EV markers including CD9, CD63, CD81, flotillin, and syntenin, in EV fractions 7-9. Endoplasmic reticulum marker, calnexin, was not detected in the EV fractions. (B) Representative images from transmission electron microscopy of isolated EVs (red arrows) from pooled EV-enriched fractions 7-9. (C) Size distribution of pooled EV fractions 7-9 isolated from MC and U2. (D) Quantification of pooled EV fractions 7-9 of MC and U2 measured by f-NTA. Data are shown as mean ± SEM, n=7-8. A Student’s t-test determined significant differences in the EV concentration between the MC and U2, *p < 0.05.

Differentially expressed EV miRNAs in response to GPCR activation

(A) The heatmap showed unsupervised hierarchical clustering of miRNA expression (row) following GPCR stimulation (column). The relative abundance of miRNAs is represented in Z-score value (z-transformed fold changes), blue, below-mean expression; red, above-mean expression. (B-E) Volcano plots displayed the analysis of EV miRNAs after ADORA1 (B), HRH1 (C), FZD4 (D), and ACKR3 (E) activation. On the x-axis, the dotted line indicates miRNAs that satisfied the |log2 fold change|≥ 1.5 cut-off, while the dotted line on the y-axis indicates miRNA that met Skillings-Mack p-value < 0.2. (F) Venn diagram showed the miRNAs of interest with at least 1.5-fold change across four GPCR groups following treatment. n=5-6 replicates per GPCR group.

EV miRNAs differentially expressed in 4 GPCR groups.

Differentially expressed miRNAs (p < 0.05) were identified in each treatment group compared to its vehicle control, n=5-6 samples of each group.

Pathway analysis of the differentially expressed EV miRNA after GPCR activation

The bubble plots showed the top 25 significantly enriched KEGG pathways for the miRNAs (≥ 1.5-fold change) of individual GPCR. (A) ADORA1, (B) HRH1, (C) FZD4, (D) ACKR3. The dot size represents the number of enriched gene targets, and the color shows the p-value of the enrichment. For the enrichment analysis, cut-off criteria were p-value (FDR) < 0.05 and gene count > 2. KEGG terms: * Endocrine and other factor-regulated calcium reabsorption, ^ Progesterone-mediated oocyte maturation.

Sensitivity analysis across alternative estimates of treatment effects.

(A) Scatterplot matrix showing pairwise concordance of alternative treatment-effect estimators, including all per-miRNA estimates from all receptors combined, with text labels (in quotes) indexed to the table below; note that the perfect correlation in the “individual” and “matched” approaches is expected, and the “pooled” and “batched” approaches only apply to the receptors ADORA1 and H1R. All estimators not based on matched pairs demonstrate bias in the form of exaggeration of effect size on average, and estimators employing a higher degree of aggregation (either at the treatment group or batch level) show worse correlation with the unbiased matched-pairs estimator (an example of ecological bias); it is clear that process variance brings considerable risk of bias in experiments of this type, and that great care should be taken to provide adequate representation of process variance via controls across experiments. (B) Comparison table detailing the differences between alternative estimators, including advantages and disadvantages of each approach. Our preferred approach is “matched”, on both theoretical and empirical grounds. Briefly, the six combinations were: 1) treatment effect estimated as difference of group medians with permutation-based p-value; 2) difference of group medians with p-value from Welch’s robust t-test; 3) (applies to batched receptors only) treatment coefficient from median regression of pooled-batch data with p-value from the rank-based Mann-Whitney test; 4) (applies to batched receptors only) sample-weighted average of treatment coefficients from separate median regressions of data from each batch with p-value as the harmonic mean of Mann-Whitney p-values calculated separately on each batch; 5) mean matched-pairs estimator (simple average of all within-pair differences between agonist and vehicle control) with permutation-based p-value; and 6) regression matched-pairs estimator (treatment coefficient from a within-pairs linear regression) with p-value from the unbalanced-robust rank-based Skillings-Mack test. Note that (1) was the only approach where we applied group-based normalization (for all others we used run-based), and that the treatment effects in (5) and (6) are not strictly the same except in the absence of missing data but they are operationally 100% correlated.

The analysis of GPCR gene expression across various cell lines, as sourced from the Human Protein Atlas.

Representative immunoblots showed stimulation of ACKR3 with SDF-1α-induced phosphorylation of ERK1/2 (pERK1/2) compared to total ERK1/2 (tERK1/2).

β-actin was used as a loading control.

Quantification of EVs in U2OS media in response to receptor activation.

U2OS cells were incubated with vehicle control (VC) or a selective agonist CCPA (A), PEA (B), Wnt3a (C), or SDF-1α (D) for ADORA1, HRH1, FZD4, and ACKR3, respectively. EVs were isolated and the concentration was normalized as a percentage of the media control (MC). EV size distribution following agonist or VC treatment in ADORA1 (E), HRH1 (F), FZD4 (G), and ACKR3 (H). Data are shown as mean ± SEM, n=3. A t-test determined no significant differences in EV concentration or size distribution between the VC and agonist stimulation in all GPCRs, p >0.05.

miRNAs detected in media control.

Differentially expressed miRNAs in the isolated EVs following GPCR activation.

All differentially expressed miRNAs (meeting p < 0.2) after stimulation were listed.

The top 10 hub genes identified in miRNA targets (≥ 1.5-fold change) PPI network.

Pathway analysis of miR-502-3p and miR-137 targets.

The bubble plots showed the top 25 significantly enriched KEGG pathways for the targets of (A) miR-502-3p and (B) miR-137. The dot size represents the number of enriched gene targets, and the color shows the p-value of the enrichment. For the enrichment analysis, the cut-off criteria were p-value (FDR) < 0.05 and gene count > 2. KEGG term abbreviations: ** Aldosterone-regulated sodium reabsorption.