Natural biased signaling of hydroxycarboxylic acid receptor 3 and G protein-coupled receptor 84

Background Medium-chain fatty acids and their 3-hydroxy derivatives are metabolites endogenously produced in humans, food-derived or originating from bacteria. They activate G protein-coupled receptors, including GPR84 and HCA3, which regulate metabolism and immune functions. Although both receptors are coupled to Gi proteins, share at least one agonist and show overlapping tissue expression, GPR84 exerts pro-inflammatory effects whereas HCA3 is involved in anti-inflammatory responses. Here, we analyzed signaling kinetics of both HCA3 and GPR84, to unravel signal transduction components that may explain their physiological differences. Methods To study the signaling kinetics and components involved in signal transduction of both receptors we applied the label-free dynamic mass redistribution technology in combination with classical cAMP, ERK signaling and β-arrestin-2 recruitment assays. For phenotypical analyses, we used spheroid cell culture models. Results We present strong evidence for a natural biased signaling of structurally highly similar agonists at HCA3 and GPR84. We show that HCA3 signaling and trafficking depends on dynamin-2 function. Activation of HCA3 by 3-hydroxyoctanoic acid but not 3-hydroxydecanoic acid leads to β-arrestin-2 recruitment, which is relevant for cell-cell adhesion. GPR84 stimulation with 3-hydroxydecanoic acid causes a sustained ERK activation but activation of GPR84 is not followed by β-arrestin-2 recruitment. Conclusions In summary, our results highlight that biased agonism is a physiological property of HCA3 and GPR84 with relevance for innate immune functions potentially to differentiate between endogenous, non-pathogenic compounds and compounds originating from e.g. pathogenic bacteria. Video Abstract. Graphical abstract


Figure S1
Figure S1 Basal activity and agonist-induced receptor internalization of HCA 3 and GPR84. CHO-K1 cells were transiently transfected with receptor constructs or empty vector (control). (A) Both, HCA 3 and GPR84, exhibited a basal cAMP inhibitory activity (cAMP level of empty vector-transfected forskolin-stimulated cells is set to 100 %) (B) The β 2 -adrenergic receptor (ADBR2) and V2 vasopressin receptor (V2R) served as control, which showed 22 % and 9 % reduction of cell surface expression upon stimulation with 10 µM isoprenalin and 10 µM arginine vasopressin (AVP), respectively. 100 µM 3HO but not 100 µM 3HDec induced a significant reduction in cell surface expression of HCA 3 . GPR84 cell surface expression levels were increased in presence of 100 µM C10 and 100 µM 3HDec. Cell surface expression levels of respective receptor construct in absence of agonist is set to 100 %. Data is shown as mean ± SEM of at least three independent experiments each carried out in triplicates. * P ≤ 0.05; ** P ≤ 0.01.

Figure S2
Figure S2 Differential, agonist-specific dynasore-sensitivity of HCA 3 and GPR84 in DMR analyses. CHO-K1 cells were transiently transfected with receptor constructs or empty vector, seeded in fibronectin-coated Epic plates and DMR responses were recorded. 3-hydroxyoctanoic acid (3HO) activated HCA 3 but not GPR84. Decanoic acid (C10) specifically activated GPR84 but not HCA 3 . 3-hydroxydecanoic acid (3HDec) activated both, HCA 3 and GPR84. No DMR response upon agonist stimulation was observed in cells transfected with empty vector. 80 µM dynasore was used. Shown is the agonist-induced wavelength shift in pm as mean ± SEM of three to five independent experiments, each carried out in triplicates. DMR at time points 10 min, 20 min and 40 min were extracted to generate Figure 1B and Figure S3.

Figure S3
Figure S3 Concentration-response curves derived from DMR analyses of HCA 3 and GPR84. CHO-K1 cells were transiently transfected with receptor constructs or empty vector, seeded in fibronectincoated Epic plates and DMR responses were recorded. Concentration-response curves in absence and presence of dynasore were derived from time points 10 min, 20 min and 40 min ( Figure S2). Shown is the agonist-induced wavelength shift in pm as mean ± SEM of three to five independent experiments, each carried out in triplicates.     3 and GPR84 showed a basal activity in HEK293-T cells (cAMP level of empty vectortransfected fsk-stimulated cells is set to 100 %). Both receptors were activated in a concentrationdependent manner by their respective agonists as determined by cAMP inhibition assays (cAMP level of HCA 3 and GPR84 transfected cells in absence of agonist is set to 100 %, respectively.) (B) 100 µM 3HO but not 100 µM 3HDec induced a significant reduction in cell surface expression of HCA 3 . No reduction of cell surface expression was measured upon stimulation of GPR84 with 100 µM 3HDec and 100 µM C10. Cell surface expression level in absence of agonist is set to 100 %. (C) In comparison to dyn-2 wt, cell surface expression of HCA 3 but not GPR84 was significantly reduced when K44A or R399A were co-transfected. Cell surface expression level of dyn-2 wt co-transfected cells is set to 100 %. (D) HEK293-T cells stably expressing β-arrestin-2-EA cells transiently transfected with GPR84 were stimulated with C10 and 3HDec. Quantification of β-arrestin-2 recruitment using the PathHunter βarrestin assay (Eurofins DiscoverX) showed no recruitment of β-arrestin-2 by GPR84 following agonist stimulation. Luminescence of GPR84 or empty vector transfected cells in absence of agonist is set 1, respectively. (E) Live-cell images of HEK293-T cells co-expressing GPR84-mRuby (red) and βarrestin-2-YFP (green) were acquired before stimulation and 30 min post-stimulation with 100 µM C10 or 100 µM 3HDec. (A-D) Data is given as mean ± SEM of at least three independent experiments each carried out in triplicates. ** P ≤ 0.01.

Supplementary Results and Discussion
The human HCA3 differs from gorilla HCA3 only in three amino acids and a C terminus extended by 24 amino acids. The orangutan HCA3 has only two differences compared to the gorilla HCA3 (Tyr 86 ; Trp 142 ) ( Figure S9D). We found that only for orangutan HCA3 we do not observe similar differences between 3HO-and 3HDec-induced DMR responses in presence of dynasore compared to human HCA3 ( Figure S9) has been shown to be crucial for nicotinic acid binding [3]. Tyr 86 of HCA3 might more likely play an indirect role for the interaction of HCA3 with dyn-2 whereas Trp 142 , located at the transition of ICL2 to TM4, could be relevant for a direct interaction. The finding that presence of dynasore causes a sustained signaling of the evolutionarily closest HCA3-relatives HCA1 (Arg 130 ) and HCA2 (Arg 142 ) further supports this hypothesis ( Figure S4). DMR analyses of the gorilla and orangutan HCA3 with D-PLA, ILA, D-Phe and L-PLA revealed that activation profiles and dynasore sensitivity of the DMR response are conserved between human and gorilla HCA3 ( Figure S9). In contrast, orangutan HCA3 is less potently activated by all agonists tested and presence of dynasore rather causes sustained or unchanged signaling than inhibition ( Figure S9C).
In summary, our analyses showed that dynasore affected the DMR response of human and gorilla but not orang utan HCA3 to D Phe and L-PLA in a similar manner like that to 3HDec, whereas the DMR response of D-PLA and ILA was similarly affected like that one to 3HO ( Figure S9). Thus, the crucial involvement of dyn-2 in HCA3 signaling and trafficking is a feature acquired during evolution before the split of the human/chimpanzee/gorilla lineage.