Signalling bias at GPR84

With the discovery of GPR84 agonist MCFAs, lipid mimetics, and other natural products, functional assays began to uncover signalling bias. This began with the exploration of structure-activity relationships (SAR) around DIM derivatives with regard to the canonical G-protein and β-arrestin pathways. Most of the compounds were biased towards the G-protein pathway, exemplified by PSB-16671 (Fig 2), while only a few compounds were found to have limited bias towards the β-arrestin pathway (Pillaiyar, Köse et al. , 2017). These derivatives are also known to act allosterically (Mahmud, Jenkins et al. , 2017), adding further texture to the downstream signalling of GPR84. Further SAR around 6-OAU lipid mimetics highlighted the variability of cAMP versus β-arrestin signalling, ranging from unbiased compounds such as PSB-1584, to compounds such as PSB-16434 (Fig 2) which had a 79-fold pathway selectivity towards cAMP (Pillaiyar, Köse et al. , 2018). Additionally, a cyclopropane-containing MCFA isolated from the marine bacterium Labrenzia sp. 011 has been shown to recruit β-arrestin without affecting cAMP production in GPR84 stable cell lines, demonstrating bias towards β-arrestin from natural product MCFAs (Amiri Moghaddam, Dávila-Céspedes et al. , 2018). Biased signalling resulting from natural agonists has been shown to underlie context-specific signalling in other inflammation associated GPCRs, which are exemplified by the 20 chemokine receptors and their distinct responses to the 50 described chemokines (Eiger, Boldizsar et al. , 2021).
A ligand-based virtual screen and subsequent hit optimisation led to the discovery of DL-175, which is a potent and selective G-protein biased agonist with no detectable recruitment of β-arrestin (Lucy, Purviset al. , 2019). As an in vitro tool this compound has found use in dissecting G-protein versus β-arrestin pathway effects in macrophages and neutrophils (Fredriksson, Holdfeldt et al. , 2022; Lucy, Purvis et al. , 2019; Mårtensson, Sundqvist et al. , 2021). In its initial biological characterisation, it was found that DL-175 confers distinct functional responses in macrophages when compared to the balanced agonist 6-OAU (Lucy, Purvis et al. , 2019).
The currently described range of GPR84 agonists point to a marked system bias of this receptor in favour of Gi-mediated pathways over the β-arrestin pathway (Fig 3). We and others have observed trends that favour the G-protein pathway within the chemical series of the orthosteric agonists 6-OAU (Pillaiyar, Köse et al. , 2018), DL-175, and some MCFAs (Lucy, Purvis et al. , 2019; Mikkelsen, Arora et al. , 2022; Peters, Rabe et al. , 2020). Indeed, it has been suggested that this system bias is a physiological property of GPR84 in its proposed function in innate immunity (Peters, Rabe et al. , 2020).
In contrast, compounds with bias towards the β-arrestin pathway are more commonly seen within the DIM class of allosteric agonists, although these are also micromolar potency and low affinity compounds (Köse, Pillaiyar et al. , 2020; Pillaiyar, Köse et al. , 2017). The absence of a physiological agonist with which to set as a reference ligand when determining bias factors remains a challenge to medicinal chemistry projects, though the evidence to date indicates that GPR84 is inherently poorly coupled to the β-arrestin pathway. With few exceptions, GPR84 agonists with sub-micromolar cAMP potencies are also biased towards cAMP (Fig 3). Furthermore, activity in the β-arrestin pathway is also highly variable, as analogues within the major ligand classes of MCFAs, 6-OAU derivatives, and DIM derivatives have been shown to have greater correlation between binding and cAMP potency than between binding and β-arrestin potency (Köse, Pillaiyar et al. , 2020). Nonetheless, multiple distinct methods have shown that GPR84 can indeed couple to β-arrestins and it remains an important pathway to investigate. This system bias is another mechanism by which cellular contexts dictate the resulting signals and physiological responses and have been observed in other inflammatory GPCR pairings such as CXCR4 and the arrestin-coupled ACKR3, CCR2 and the arrestin-coupled D6R, and the C5a1 receptor and the arrestin-coupled C5a2 receptor (Pandey, Kumariet al. , 2021; Yen, Schafer et al. , 2022).
As β-arrestins are part of the canonical desensitisation and internalisation pathways of GPCRs, an inherent system bias away from this pathway has broad implications for receptor regulation and drug design. The lack of efficacy of DL-175 in the GPR84-β-arrestin pathway has been demonstrated using tagged receptor and arrestin systems by chemiluminescent enzyme fragment complementation (Lucy, Purvis et al. , 2019; Mårtensson, Sundqvist et al. , 2021) and BRET (Marsango, Ward et al. , 2022), as well as by measuring arrestin translocation to the membrane in enhanced bystander BRET assays (Fredriksson, Holdfeldt et al. , 2022). It is now also known that two key threonine residues in intracellular loop 3 are phosphorylated by GRK2/3 following agonist stimulation with ZQ-16, but not DL-175 (Marsango, Ward et al. , 2022). Phosphorylation of these residues, Thr263 and Thr264, allow for subsequent interactions with β-arrestin and β-arrestin-2 (Marsango, Wardet al. , 2022). This is consistent with the idea that bias is orchestrated by GRK proteins in response to certain agonist-induced receptor conformations, which then influence β-arrestin interactions and signalling, rather than directly favouring or disfavouring interactions with β-arrestin itself (Choi, Staus et al. , 2018; Zidar, Violinet al. , 2009).
Functional results in primary cells using the GRK2/3 inhibitor cmpd101 support the physiological involvement of GRK2/3 in GPR84 desensitisation. The label-free impedance response of BMDMs stimulated with DL-175 is more prolonged than the response to 6-OAU, but interestingly pre-treatment with cmpd101 results in a 6-OAU response that mirrors DL-175 alone (Lucy, Purvis et al. , 2019). In neutrophils, the ROS response induced by ZQ-16 was also prolonged by pre-treatment with cmpd101, whereas the response of DL-175 was not affected (Fredriksson, Holdfeldt et al. , 2022). Further work is needed to determine the involvement of GRK2/3 on the kinetics of these responses and direct readouts such as live-cell cAMP or β-arrestin recruitment would be especially valuable. However, both the impedance response of BMDMs and ROS production in neutrophils when stimulated with DL-175 were terminated in relatively short timeframes, suggesting an alternative mechanism of desensitisation. By comparison to FPR2, a direct coupling of GPR84 to the actin cytoskeleton has been hypothesised (Fredriksson, Holdfeldt et al. , 2022). The importance of desensitisation in drug design can be highlighted by the use of S1P1-desensitising agonists for the treatment of multiple sclerosis. Agonists causing persistent S1P1signalling are sought for their endothelial protective properties (Grailhe, Boutarfa-Madec et al. , 2020). Likewise, in infection research, agonist-induced internalisation of the HIV coreceptor CCR5 is an effective strategy for viral entry inhibition, and a further desirable effect is to achieve this with minimal receptor activation and the concomitant inflammatory response (Kazmierski, Bifulco et al. , 2003).
GPR84 has been shown to activate transducer proteins including Gi/o, G12/13, G15, GRK2/3, and β-arrestin1/2 (Gaidarov, Anthony et al. , 2018; Marsango, Ward et al. , 2022; Peters, Rabe et al. , 2022; J. Wang, Wu et al. , 2006) which couple to effector molecules including β-catenin, DOK3, NLRP3, and phospholipase C (Dietrich, Yanget al. , 2014; Gao, Qu et al. , 2020; Peters, Rabe et al. , 2022; Zhang, Chen et al. , 2022), small GTPases such as ras/rho and dynamin (Peters, Rabe et al. , 2022), kinases PI3K, Akt, ERK1/2, JNK, and p38 (Gao, Qu et al. , 2020; Meng, Zhanget al. , 2017; Park, Yoon et al. , 2018; Recio, Lucyet al. , 2018), and the transcription factors NF-κB and STAT3 (Recio, Lucy et al. , 2018; Yin, Cheng et al. , 2020). Further investigation into the effector proteins downstream of GPR84 may be important when assessing physiological efficacy.
For example, positive allosteric modulators at the Gq-coupled M1 muscarinic acetylcholine receptor with similar selectivity profiles and similar effects on the binding and calcium responses of acetylcholine were still found to differentially potentiate, i.e. bias, receptor coupling to phospholipases C and D (Marlo, Niswender et al. , 2009). Further evaluation revealed that PLD activity is necessary for M1-dependent long-term depression in the prefrontal cortex, an effect that is implicated in targeting M1 for the treatment of cognitive deficits in schizophrenia and Alzheimer’s disease (Moran, Xiang et al. , 2019).
In the case of GPR84, the best evidence to the importance of bias is the suggestion that the Gi-based agonist DL-175 promotes equivalent levels of phagocytosis but not chemotaxis as 6-OAU (Lucy, Purvis et al. , 2019). However, given the differences in assay time points, cell types, and concentration-response profiles between these assays and those for cAMP production and β-arrestin recruitment, it remains a possibility that other effectors are involved in the distinct responses between phagocytosis and chemotaxis. For example, it has recently been shown that GPR84 couples to Gα15 which results in phospholipase C activity, ERK phosphorylation, and calcium signalling (Peters, Rabe et al. , 2022). Given that GPCRs can not only promiscuously couple to a number of G-proteins, but can also ‘switch’ in coupling preferences over a signalling time course (Cawston, Redmond et al. , 2013), the Gαi and Gα15 pathways may be important pharmacological descriptors to monitor in future experiments.