RAMP and MRAP accessory proteins have selective effects on expression and signalling of the CB1, CB2, GPR18 and GPR55 cannabinoid receptors

Receptor activity‐modifying proteins (RAMPs) and melanocortin receptor accessory proteins (MRAPs) modulate expression and signalling of calcitonin and melanocortin GPCRs. Interactions with other GPCRs have also been reported. The cannabinoid receptors, CB1 and CB2, and two putative cannabinoid receptors, GPR18 and GPR55, exhibit substantial intracellular expression and there are discrepancies in ligand responsiveness between studies. We investigated whether interactions with RAMPs or MRAPs could explain these phenomena.

Limited reports suggest the same for MRAP1, though a functional impact has not been reported to date (Chan et al., 2009;Kay et al., 2013).
The established cannabinoid receptors, CB 1 and CB 2 , are class A GPCRs.Named for their mediation of the effects of Cannabis sativa, these receptors play key roles in a variety of physiological processes, such as appetite, mood, memory, learning and inflammation and are attractive therapeutic targets in a range of disorders (Pacher et al., 2020).Two orphan receptors, GPR18 and GPR55, have been proposed as potential members of the cannabinoid receptor family, as they respond to some cannabinoid ligands (Henstridge et al., 2011;Morales et al., 2020).Here, we will refer to GPR18 and GPR55 as cannabinoid receptors.However, there are discrepancies in the published reports of the ligand responsiveness of GPR18 and GPR55, and the basis for these conflicting reports is largely unknown.
Cannabinoid receptors are located at the cell plasma membrane but also exhibit considerable intracellular expression.The specific subcellular localisation and characteristics of these intracellular populations are poorly defined.However, the lipophilic nature of cannabinoid ligands and a handful of studies on CB 1 and CB 2 receptors to date indicate that these intracellular cannabinoid receptors may contribute directly to cellular responses (Oyagawa & Grimsey, 2021;Whiting et al., 2022).The intracellular pool of CB 1 receptors, reported to be present in both over-expressing cells and in cell lines with endogenous CB 1 receptors, seems not to be formed by constitutive internalisation but may reside in endolysosomes from where they can reportedly signal (Brailoiu et al., 2011;Grimsey et al., 2010;Rozenfeld & Devi, 2008).Intracellular CB 2 receptors have been found in a range of cell types, including human primary B lymphocytes, T lymphocytes and monocytes, and this population has been suggested to signal via coupling to G proteins different from those used by cell surface CB 2 receptors (Brailoiu et al., 2014; What is already known • RAMP and MRAP accessory proteins modulate expression and signalling of some GPCRs.
• Cannabinoid receptors have pronounced intracellular expression, and there are published discrepancies regarding ligand engagement.

What does this study add
• MRAPs altered expression of most cannabinoid receptors, with subtle effects on CB 1 and GPR55 signalling.
• Some cannabinoid receptors enhanced total expression, but not surface expression of RAMP2 and/or RAMP3.

What is the clinical significance
• Cannabinoid receptor-MRAP interactions represent new potential mechanisms for modulating physiologies such as energy metabolism.
GPR18 has a predominantly intracellular distribution in transfected cells and in corneal epithelial cells with endogenous GPR18 expression, which may correspond with a high constitutive internalisation rate, though limited steady-state surface expression is present in some circumstances (Console-Bram et al., 2014;Finlay et al., 2016;Murataeva et al., 2019).GPR55 typically has robust cell surface expression, but predominant intracellular expression of GPR55 has also been observed, including in an endogenously expressing cell line and in human primary neutrophils (Balenga et al., 2011).
The expression and function of cannabinoid receptors overlap with RAMPs and MRAPs in several contexts.RAMP mRNA is present in most tissues in varying expression levels, and expression is altered in disease states such as heart disease, cancer, inflammation and diabetes (Hay et al., 2006;Jacob et al., 2012).RAMP1 is expressed widely in the brain, including in the hippocampus and cortex where CB 1 receptors and GPR55 are also expressed (Hay et al., 2006;Marichal-Cancino et al., 2017).RAMPs 1-3, CB 1 receptors and CB 2 receptors are all expressed in smooth muscle (Bermudez et al., 2017;Chauhan et al., 2015;Frayon et al., 2000).mRNA for RAMPs 1-3 is found in the lung, and based on single cell type mRNA expression, these RAMPs are expressed in alveolar type II cells along with CB 1 receptors, CB 2 receptors, and GPR18 (Karlsson et al., 2021).RAMPs 1-3 and GPR18 are also expressed in cardiomyocytes and spermatids/spermatozoa (Chiu et al., 2012;Flegel et al., 2016;Hay et al., 2006;Karlsson et al., 2021;Matouk et al., 2017).
CB 1 receptors and MRAP2 are both expressed in multiple brain regions, including the cerebellum and paraventricular nucleus of the hypothalamus (Arslan et al., 2000;Glass et al., 1997;Herkenham et al., 1991).
Given the notion that intracellular cannabinoid receptors may require the presence of a chaperoning protein to facilitate surface expression, and the feasibility of physiological co-expression between cannabinoid receptors and RAMPs or MRAPs, we hypothesised that accessory protein(s) might interact with cannabinoid receptors to influence their subcellular distribution and/or ligand responsiveness.We investigated this by determining the effects of RAMP and MRAP cotransfection on the cell surface and total expression of cannabinoid receptors, and vice versa, and measuring cannabinoid receptor signalling responses with and without introduction of accessory proteins.

| General experimental protocols and design
For all assays, the positions of conditions and/or treatments on and between plates were randomised between independent experiments to avoid potential influence of plate position or treatment order.All assays were measured in an automated manner and analysed with objective quantitation.Criteria for data exclusion are noted in the assay method where applicable.As no assays called for subjective sample acquisition or scoring, blinding was not utilised.
In each experiment, at least two technical replicates were performed and the mean of technical replicates taken as the datapoint for one independent experiment.the matched control condition for each independent experiment in order to display the trends in differences for the conditions of interest (which in some cases would not be discernible based on the collated raw data due to differences in raw readings between experiments).

| Receptor and accessory protein expression plasmids
Plasmids utilised in this study are summarised in Table 1.The 3HA-hCB 1 in pcDNA5/FRT and 3HA-hGPR55 in pcDNA5/FRT plasmids are reported here for the first time.The 3HA-hCB 1 in pcDNA5/FRT plasmid was generated by restriction digest (KpnI and XhoI) of 3HA-hCB 1 in pcDNA3.1+(cDNA Resource Center, Bloomsburg, PA, USA; www.cDNA.org,CNR010TN01) and ligation of the gel-extracted 3HA-hCB 1 sequence into digested and gel-extracted pcDNA5/FRT (Thermo Fisher Scientific, Waltham, MA, USA; V601020).To create the 3HA-hGPR55 in pcDNA5/FRT plasmid, the sequence for part of the 5 0 multiple cloning site and 3HA was amplified from an extant plasmid, and hGPR55 (cDNA Resource Center, GPR0550000) was amplified using primers to add a 5 0 overhang corresponding to the 3 0 end of the 3HA sequence, both using blunt-cloning KAPA HiFi polymerase.Amplicons were then combined in equal molar amounts and fused by overlap-extension PCR (KAPA HiFi LongRange polymerase).
Fusion products were gel-extracted and then ligated into pGEM-T Easy (Promega, Madison, WI, USA).After sequence verification, a 3HA-hGPR55 clone was then restriction digested from pGEM-T (KpnI and XhoI) and ligated into pcDNA5/FRT.Plasmids were transformed into and expanded via XL10-Gold E. coli, purified by miniprep kit, and preparations quantitated with a NanoDrop 1000 (Thermo Fisher Scientific).Plasmid coding sequences were verified by Sanger sequencing (Massey Genome Service, Massey University, New Zealand).Plasmid molecular weights were determined via https://molbiotools.com/dnacalculator.php.
In 'No Receptor' and 'No Accessory' experiment conditions, the receptor and/or accessory plasmids were substituted at equimolar concentration for a pcDNA5/FRT plasmid encoding chloramphenicol acetyl transferase (CAT) (Thermo Fisher Scientific).The bacterial CAT protein is expected to be functionally neutral in mammalian cells.

HEK-293S cells (generously provided by Professor David Poyner, Aston
Research Centre, RRID:CVCL_A784) (Wootten et al., 2013) were maintained in DMEM high glucose (Hyclone, Logan, UT, USA) with Lglutamine and sodium pyruvate with 8% FBS in a humidified atmosphere at 37 C with 5% CO 2 .Cells were trypsinised and seeded at 37,000 cells per well into a PDL-treated (0.05 mgÁml À1 incubated for at least 1 h at 37 C, followed by one wash with PBS) clear 96-well plate.
Transfection mixtures for each receptor and accessory plasmid of interest were diluted in DMEM high glucose with L-glutamine and sodium pyruvate (without serum).In 'No Receptor' and 'No Accessory' conditions, receptor and/or accessory plasmids were substituted for CAT in pcDNA5/FRT (described earlier).After diluting PEI MAX in DMEM, DNA mixtures were combined 1:1 with the diluted PEI.The final total transfection volume was 40 μl per well, with 0.36-μg PEI and 10 fmol of each plasmid of interest ($31-45 ng per plasmid, depending on the construct).Equimolar amounts of the plasmids were transfected to promote robust co-expression (Figure S2).Robust effects of accessory protein expression on the control receptors, calcitonin receptorlike receptor (CLR) and MC 2 , were also observed under these conditions (Figures 2a,5a,S3 and S4).Transfection mixtures were incubated at room temperature ($20 C) for 60 min and then dispensed on top of the culture medium.Cells were returned to the incubator.
Immunocytochemistry was carried out via similar methods as described previously (Grimsey et al., 2011).Forty to forty-eight hours after transfection, plates that were intended for cell surface labelling were placed on ice (to prevent plasma membrane trafficking), and media was replaced with primary antibody diluted in DMEM with 1 mgÁml À1 BSA-anti-HA 1:500 and anti-FLAG 1:750.Plates were incubated at room temperature ($20 C) for 30 min with gentle rocking.Note: pcDNA3, pcDNA3.1 and pcDNA5/FRT all utilise the cytomegalovirus (CMV) promoter, whereas pEF4a utilises the human elongation factor-1 alpha (EF-1α) promoter.'Native' refers to the natural endogenous signal peptide for the protein.For constructs with both signal peptides and tags, the tag is inserted between the signal peptide and the start of the receptor coding sequence.a pplss refers to the bovine preprolactin signal sequence (30 amino acids; UniProt P01239).b 3TCS refers to three thrombin cleavage sites.3TCS permits enzymatic (thrombin) cleavage of the HA tag and was incidental in this project.c 63R refers to the presence of arginine at position 63 of the hCB 2 coding sequence, a single nucleotide polymorphism (SNP) which is slightly more prevalent in the general population than glutamine at this position (63Q) (dbSNP rs2501432).
Cells were then washed twice with DMEM-BSA and fixed with 4% PFA for 10 min, followed by three washes in 1Â-PBS.Separate plates for total cell labelling were fixed in the same way and then incubated with primary antibody in immunobuffer (PBS with 1% normal goat serum, 0.2% Triton X-100 and 0.4 mgÁml À1 thiomersal) for either $3 h at room temperature, or overnight at 4 C, with gentle rocking-anti-HA 1:1000 and anti-FLAG 1:2500.Secondary antibodies were diluted 1:400 in immunobuffer and incubated for either $3 h at room temperature, or overnight at 4 C, with gentle rocking.Hoechst 33258 was incubated at 8 μgÁml À1 in PBS with 0.2% Triton X-100 (PBS-T) for 15-20 min at room temperature with gentle rocking.Between each labelling step, cells were washed with PBS-T.After the Hoechst stain, cells were washed an additional two times, with the final wash being incubated for at least 10 min.Cells were left in 50-μl PBS-T with 0.4 mgÁml À1 thiomersal for imaging.All stages involving secondary fluorescent antibodies or stains were performed under reduced light conditions and completely protected from light for incubations and storage.
In brief, four images per well were acquired using an ImageXpress Micro XLS microscope (Molecular Devices, Sunnyvale, CA, USA), with 10Â objective and filters appropriate for the nuclear stain and secondary antibodies utilised.Fluorescence intensity per cell above background threshold was determined via automated routines, which incorporated masks to exclude areas of the images not proximal to cell nuclei and to exclude areas of brightly fluorescent debris.Images with moderate intensity debris that was not removed by the automated mask, and/or with low cell number in comparison with other images, were excluded from subsequent analysis.Full wells or datasets were excluded in cases where there was known or suspected human error in the experimental set up or treatment, or overall poor signal to noise in the dataset was observed (generally indicating either failure of transfection or immunocytochemistry).To obtain relative expression (expressed as percentages) for presentation in figures, the data were normalised.For each independent experiment, the fluorescence intensity per cell above background was normalised to the control/comparator condition for that experiment set (100%).Statistical analysis was undertaken on raw nonnormalised data (further detail below).

| cAMP assay
cAMP assays were performed using the CAMYEL BRET-based biosensor (ATCC MBA-277) with methodology similar to that published previously (Sharma et al., 2023;Singh et al., 2019).Cells were prepared and transfected as for receptor and accessory protein detection, except that cells were seeded into PDL-treated white plates (Corning, Corning, NY, USA), and the CAMYEL plasmid (7 fmol per well; 23.8 ng per well) was co-transfected.Forty to forty-eight hours later, cells were washed with assay medium (HBSS with 1 mgÁml À1 BSA) and then incubated with 80-μl assay medium in a 37 C incubator for 30 min.Equilibration for each plate was staggered to ensure matching conditions for each run.Drugs were prepared at 10Â concentration in assay medium and were moved into 1.2-ml strip tubes for dispensing.Final drug concentrations were as shown in the figures, with forskolin being included where indicated at 5 μM (approximate EC 50 for forskolin response), and all conditions being subject to the same vehicle concentration.Coelenterazine h was prepared in assay medium at 10Â concentration immediately before dispensing to cells (10 μl, to a final concentration in 100 μl of 5 μM).After the coelenterazine was dispensed, BRET1 output was recorded for $5 min (LUMIstar Omega, BMG Labtech, Ortenberg, Germany; using 475-30 and 535-30 nm filters, with simultaneous detection and interval time of 0.5 s per well).At the end of this baseline read, 10 μl of the drug mixtures was dispensed, and the plate reader was then re-started to read for 30 min.
Raw readings from the Rluc emission channel were plotted and inspected for any abnormalities (e.g., low raw Rluc could indicate failed transfection or incorrect volume of coelenterazine received).
The BRET ratio was calculated and plotted, with the baseline ratios expected to be equivalent between wells.After this verification, the mean BRET ratio for the readings taken over the 30-min drug stimulation was calculated and concentration-response curves plotted for the agonist response conditions (log[agonist] vs. response [three parameters]; GraphPad Prism, v8.1.1,GraphPad Software).Data were then normalised such that the vehicle control (without forskolin) was 0%, and the fitted response curve limit where there was no agonist response was 100% (i.e., approximately equivalent to forskolin without agonist).Concentration-response parameters pEC 50 (negative log EC 50 ) and span (i.e., difference between curve top and bottom) were obtained for each independent experiment for presentation and statistical analysis (see Sections 2.1 and 2.6).

| G protein dissociation assay
G protein dissociation assays were performed using the TRUPATH platform (a gift from Bryan Roth, Addgene kit #1000000163) (Olsen et al., 2020).Transfections were carried out with a similar method to that described previously (Patel et al., 2022) GraphPad Prism).Data were then normalised such that the 'No Receptor' control treated with vehicle was 0%, and the fitted response curve limit where there was no agonist response for the 'No Accessory' condition was 100% (i.e., approximately equivalent to receptor-transfected cells, treated with vehicle).Concentrationresponse parameters pEC 50 (negative log EC 50 ), span (i.e., difference between curve top and bottom) and curve tops and bottoms were obtained for each independent experiment for presentation and statistical analysis (see Sections 2.1 and 2.6).

| Data and statistical analysis
The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology (Curtis et al., 2022) 3 | RESULTS

| Comparative basal receptor expression
To (pplss) to promote increased expression for these two receptors (Finlay et al., 2017(Finlay et al., , 2016)).Signal peptides occur naturally in many plasma membrane proteins (including some GPCRs).These facilitate interaction with the signal recognition particle in the endoplasmic reticulum, enhancing translocation of peptides across the plasma membrane and therefore correct receptor folding and topology but are cleaved prior to cell surface delivery of the mature protein.Inclusion of the pplss signal sequence on 3HA-GPR18 enhanced cell surface and total expression slightly.pplss-3HA-hCB 1 receptors (expressed under the hEF-1α promoter) exhibited the greatest cell surface and total anti-HA labelling of all the constructs in the study.

| Cannabinoid receptor expression with and without RAMP co-expression
CLR and cannabinoid receptors were co-transfected with untagged human RAMPs 1, 2 or 3, or a plasmid encoding a benign protein ('No Accessory') and incubated for $48 h before measuring cell surface and total cellular receptor expression via anti-HA antibody labelling of N-terminal HA tags.
As expected, CLR cell surface expression was elevated by coexpression of untagged RAMPs 1, 2 and 3, as was total cellular expression (Figure 2a).Whereas RAMPs 1 and 2 did not influence the expression of any of the cannabinoid receptors, RAMP3 reduced cell surface and total expression of CB 1 receptors (without pplss) and CB 2 receptors by $25% (Figure 2b,c).GPR18 (without pplss) total expression was reduced to a similar degree, but without any apparent change in surface expression.Regarding detection of cell surface GPR18 (without pplss), note that although genuine receptor staining could be detected qualitatively, the low cell surface expression level resulted in the intensity of anti-HA labelling on the GPR18-transfected cells being in a similar range to, and only slightly greater than, the 'No Receptor' non-specific labelling control.

| RAMP expression with and without coexpression of cannabinoid receptors
In our next experiments, we monitored RAMP surface and total cellular expression.When expressed without receptor (transfected with plasmid encoding a benign protein), FLAG-tagged RAMPs were barely or not at all detectable at the cell surface by anti-FLAG labelling (Figure 3a,b).Co-expression of CLR markedly enhanced surface expression of RAMPs 1 and 2, with RAMP1 reaching the greatest detected expression level and RAMP2 slightly lower.Although RAMP3 surface detection was enhanced with CLR (Figure 3b), this remained at a low level compared with RAMPs 1 and 2 (Figure 3a).RAMP1 total cellular expression was readily detectable without receptor but was enhanced in the presence of CLR (Figure 3a,c).Total expression for RAMP2 and 3, as detected by anti-FLAG labelling, was low in comparison with RAMP1 (Figure 3a).RAMP2 total expression was unchanged by coexpression with CLR (Figure 3c).In contrast, RAMP3 total expression was markedly enhanced in the presence of CLR, having been essentially undetectable without receptor co-expression.
No effects of the cannabinoid receptors on the surface expression of FLAG-RAMP were detected (Figure 3b).Likewise, no effects of cannabinoid receptors on FLAG-RAMP1 total expression were observed (Figure 3c).However, four cannabinoid receptors enhanced FLAG-RAMP2 total expression (or detection thereof); pplss-CB 1 , CB 2 , GPR18 and GPR55.CB 2 receptors had the greatest effect, resulting in double the RAMP2 expression, in comparison with the expression when CLR was co-expressed.pplss-CB 1 receptors strikingly enhanced FLAG-RAMP3 total expression (or detection) to the equivalent of that for CLR co-expression.

| CB 1 and CB 2 receptor signalling with and without RAMP co-expression
Having observed the effects of RAMP3 on expression of CB 1 and CB 2 receptors and the effects of CB 1 and CB 2 receptors on total expression of RAMPs 2 and/or 3, we sought to determine whether the presence of RAMPs affected cAMP signalling from CB 1 , pplss-CB 1 and CB 2 receptors.
Firstly, we verified that our experimental paradigm could replicate earlier findings for RAMP modification of CLR-mediated stimulation of cAMP production (Hay et al., 2018).As expected, co-expression of CLR with untagged RAMP1 promoted high potency responsiveness to CGRP with lower potency responses in the presence of untagged RAMP3 and RAMP2.In response to human adrenomedullin 15-52, the order of highest to lowest potency for cAMP elevation was RAMP3 > RAMP2 > RAMP1 (Figure S3).Without RAMP co-expression, stimulation of cAMP production was essentially absent, with only small elevations evident at high agonist concentrations.
CB 1 , pplss-CB 1 and CB 2 receptors expressed with benign control plasmid ('No Accessory') all responded to the CB 1 /CB 2 receptor agonist CP55940, as expected from earlier reports.CB 1 and CB 2 receptors are primarily Gα i -coupled and therefore produce a reduction in cAMP on stimulation (Cawston et al., 2013;Oyagawa et al., 2018).However, when expressed at a high level, the response of CB 1 receptors can switch to stimulation of cAMP production (most likely via Gα s coupling), as we have observed for pplss-CB 1 receptors (Finlay et al., 2017).
We also applied two inverse agonists, SR141716A and SR144528, at a single concentration, as a proxy indicator of constitutive activity.
Inverse agonist treatment produced elevated cAMP with all three receptor constructs, which is consistent with predominantly Gα i -coupled constitutive activity.This is contrary to agonist-stimulated coupling for pplss-CB 1 receptors, but this observation is consistent with an earlier report (Finlay et al., 2017).The inverse agonism window for non-pplss CB 1 receptors was extremely small, increasing cAMP only $10% above forskolin-stimulated levels, which is consistent with the very low apparent expression level observed for this construct.
We did not observe any indication of alterations in agonistinduced nor constitutive cAMP signalling with RAMP co-expression, though these data should be considered exploratory due to only three independent experiment having been carried out (Figure 4a-c).

| Expression of cannabinoid receptors with and without MRAP co-expression
In analogous design to the investigations described for RAMPs, MC 2 and cannabinoid receptors were simultaneously co-transfected with C-terminally FLAG-tagged human MRAPs 1α or 2 or a plasmid encoding a benign protein ('No Accessory') and incubated for $48 h prior to measuring cell surface and total cellular receptor expression via anti-HA antibody labelling of N-terminal HA tags.
MC 2 receptors were the first reported MRAP-interacting receptors and, as expected (Chan et al., 2009), cell surface expression (or detection) of these receptors was enhanced by co-expression of both MRAPs.MRAP1α also elevated total expression of MC 2 receptors, whereas MRAP2 did not (Figure 5a).
MRAP1α significantly reduced cell surface expression of CB 1 (without pplss) and CB 2 receptors, pplss-GPR18 and GPR55 (Figure 5b).The largest proportional changes were for CB 1 receptors and pplss-GPR18, with CB 1 receptors reduced to approximately half the expression without accessory, and pplss-GPR18 reduced to close to the background staining level.This finding with CB 1 receptors was mirrored by a reduction in total receptor expression.Although there was no analogous statistically significant reduction in total expression of pplss-GPR18, we noted trends toward this effect for both GPR18 and pplss-GPR18, as measurements were lower than the 'No Accessory' condition in all experiments (Figure 5c).consequently, had low statistical power to detect a potential change (Figure 5c).
MRAP2 induced greater GPR18 or pplss-GPR18 surface expression than the 'No Accessory' control in all experiments, though this potential change did not reach statistical significance (Figure 5b).
Regarding total cellular receptor expression, MRAP2 significantly enhanced expression of both GPR18 constructs.MRAP2 produced a modest reduction in the total expression of pplss-CB 1 receptors, without any effect on non-pplss CB 1 receptors.

| MRAP expression with and without cannabinoid receptor co-expression
We continued our investigation by studying MRAP-FLAG expression

| Cannabinoid receptor signalling with and without MRAP co-expression
As for the analogous study with RAMPs, we again measured cAMP signalling for CB 1 , pplss-CB 1 and CB 2 receptors and did not detect any apparent influence of MRAPs (though these data should be considered exploratory as only three independent experiments were carried out; Figure 7a-c).This was despite our assay setup faithfully reporting the expected effects of MRAPs on signalling from MC 2 receptors (Chan et al., 2009), with MRAP1α enabling high potency response to ACTH, MRAP2 facilitating a low potency response, and no response detected when no accessory protein was co-expressed (Figure S4).
Given the clear effects of both MRAPs on GPR55 expression and a reciprocal effect of GPR55 on MRAP1α, we were particularly motivated to study GPR55 signalling.GPR55 couples to Gα 13 (Henstridge et al., 2011), so we employed the TRUPATH G protein dissociation assay to investigate signalling in this pathway.As CB 1 receptors can also couple to Gα 13 (Avet et al., 2022), we included pplss-CB 1 receptors as a positive control in this experiment set.
Consistent with the lack of change in expression of pplss-CB 1 with MRAP1α (Figure 5b), we did not observe an effect of MRAP1α on the response magnitude nor potency of G protein dissociation induced by the CB 1 receptor agonist WIN (Figure 7d).Neither was there an effect on constitutive G protein dissociation, though the constitutive signalling window was extremely small (i.e., difference in assay signal between cells with and without expression of pplss-CB 1 ), and this resulted in considerable noise and large between-experiment variability in this dataset.Interestingly, MRAP2 reduced the potency of pplss-CB 1 response to WIN by a small but reproducible degree ($0.4 log units).
Lysophosphatidyl inositol (LPI) is a putative endogenous ligand for GPR55 (Henstridge et al., 2011), and we indeed observed dissociation of Gα 13 from β 3 γ 9 in response (Figure 7e) though with relatively low potency (mean raw pEC 50 for 'No Accessory', 6.3 ± 0.2).The potency and maximal extent of agonist-induced G protein dissociation were not affected by the presence of MRAPs.However, both MRAPs produced a small but statistically significant reduction in constitutive G protein dissociation.Given the maximal response extents (E max ) were equivalent between all conditions, the lower starting G protein association with MRAPs produced greater response windows (concentration-response curve 'spans') with agonist activation.
Although we were very interested to study GPR18 signalling, to date, we have been unable to detect GPR18-mediated activity in response to a range of putative GPR18 ligands in various pathways.
Though these data should only be considered preliminary, we did not detect any indication of response to putative GPR18 ligands, with or without co-transfection of accessory proteins, despite positive control receptors confirming the general ability to detect responses in the assays (Figure S5).

| DISCUSSION
The purpose of our study was to determine whether subcellular distribution and signalling of cannabinoid receptors could be influenced by one or more of the RAMP and MRAP accessory proteins.
GPR18 was the only cannabinoid receptor found to undergo apparent chaperoning by an accessory protein, with MRAP2 coexpression enhancing GPR18 cell surface and total cellular expression.
pplss-GPR18 also enhanced MRAP2 cell surface expression (and/or detection).These reciprocal effects between GPR18 and MRAP2 on expression and/or detection indicate strong potential for direct interaction between these proteins.Indeed, these effects were similar to those observed for MC 2 receptors which unequivocally interact with MRAP2, though it was particularly interesting that co-expression of pplss-GPR18 enhanced MRAP2 surface detection, when coexpression of MC 2 did not.It is interesting to speculate that the overall extremely low immunocytochemical signal for cell surface MRAP2 we observed and/or some instances of apparent lack of change of MRAP2 surface expression may relate to MRAP2 membrane topology (see further discussion below).Given the documented difficulties in reproducing GPR18 G protein coupling and ligand engagement between studies and laboratories, it is an exciting prospect that MRAP2 may have potential for modifying GPR18 signalling.Frustratingly, we have not been able to detect a constitutive or a liganddriven, GPR18-mediated, signalling response to date.It would be illuminating to perform a wider screen of GPR18 G protein coupling in the presence of MRAP2 with a range of putative GPR18 ligands.This prospect would be feasible with the use of a G protein activation biosensor platform, such as TRUPATH.
Our data indicate cross-influences between GPR55 and both MRAPs.MRAPs 1α and 2 produced a robust reduction in GPR55 surface and total expression, implying that MRAP promoted constitutive degradation of GPR55.Such a finding is not unprecedented; for example, MRAP1 can reduce the surface expression of MC 4 and MC 5 receptors (Chan et al., 2009).The reduction in GPR55 expression correlated with a reduction in constitutive activity in the Gα 13 dissociation assay and, assuming GPR55 is constitutively active in vivo, has potential to be physiologically relevant.However, there was no indication of an effect of either MRAP on the potential for LPI-induced signalling, with all conditions reaching equivalent maximal G protein dissociation, despite starting from different constitutive activation levels.It is plausible that the MRAPs might not affect GPR55 signalling.Indeed, separate MRAP domains and post-translational modifications have been implicated in GPCR interaction, compared with those involved in ligand engagement (Maben et al., 2016).However, we also note that receptor-accessory protein combinations can exhibit heterogeneity in that some, but not all, ligand effects are influenced by presence of the accessory protein (Hay et al., 2018;Wootten et al., 2013).
MRAPs might therefore modulate GPR55 signalling with other agonists and/or signalling pathways.
While GPR55 did not produce a reciprocal effect on MRAP2 expression, GPR55 reduced the apparent cell surface expression of MRAP1α.The equivalent effect was also evident for prototypically interacting MC 2 receptor, as well as pplss-CB 1 and CB 2 receptors.At face value, this finding seems illogical, particularly as MC 2 was chaperoned to the cell surface by MRAP1α and most likely requires continued interaction with MRAP1α at the cell surface in order to produce signalling responses.We speculate that this effect might result from a reduction in the ability to detect the C-terminal FLAG tag on MRAP1α, rather than reflect a genuine reduction in cell surface expression.
Potential explanations for this include altered post-translational modification and/or conformation that reduces the anti-FLAG antibody affinity for or access to the FLAG epitope or a change in the MRAP1α proportional antiparallel dimer topology.Regarding the former possibility, MRAPs are targets for glycosylation (Sebag & Hinkle, 2007).Further, post-translational modification (sulfation) of the FLAG tag in response to GPCR oligomerisation has been reported (Hunter et al., 2016).In the present study, we utilised the anti-FLAG antibody that was found to be less sensitive to (though was still somewhat inhibited by) sulfation of the FLAG tag.Still, the potential for an influence of this or other post-translational modifications cannot be eliminated.To the MRAP1α dimer topology, MRAPs can exist as monomers but primarily form anti-parallel homodimers, which are required for interaction with MC 2 receptors (Sebag & Hinkle, 2007).In the case here with MRAP1α, we would speculate that the receptors producing the apparent decrease likely reduce cell surface expression of the extracellular C-terminus topology that is 'visible' to our cell surface labelling paradigm, but any potential effect on surface expression of the extracellular N-terminus topology could not be monitored.Interestingly, RAMP3 can shift the proportional topology of MRAP2 to favour the extracellular N-terminus (Chen et al., 2020).In further study, the hypothesis that MRAP1α and/or MRAP2 predominant membrane topology is influenced by interactions with cannabinoid receptors could be tested by comparison to N-terminally tagged MRAPs.A further possible explanation is competition with an endogenously-expressed natural interacting partner for MRAP1α.
As well as producing the aforementioned reduction in MRAP1α surface expression, expression of CB 1 , pplss-CB 1 and CB 2 receptors were all themselves altered by MRAP1α or MRAP2 in one or more contexts.The pplss-CB 1 receptors provided an interesting case in that MRAP2 modestly reduced the potency of pplss-CB 1 -mediated G protein dissociation without any apparent change in expression of cell surface pplss-CB 1 receptors.This MRAP2 finding exemplifies that accessory proteins have the potential to influence signalling independently of receptor surface expression (Chan et al., 2009;Wootten et al., 2013).In terms of MRAP2 affecting the Gα 13 dissociation assay with WIN stimulation, but not the cAMP (Gα i ) assay with CP55,940 stimulation, we cannot clarify whether this difference was due to selectivity in how MRAP2 influences CB 1 G protein coupling or ligand engagement or perhaps a reflection of assay sensitivities.Of course, investigating other G protein coupling and signalling pathways for CB 1 and CB 2 receptors could be illuminating.
We have largely provided evidence against interactions between cannabinoid receptors and RAMPs, in terms of both expression and receptor signalling.Based on obtaining the expected CLR responses and different patterns of effects (or lack thereof) for the same receptors with different accessory proteins (or vice versa), we feel that our data are robust and sufficiently sensitive to detect genuine interactions.
Despite largely negative data for the RAMPs, we did identify some interesting phenomena.First, RAMP3 had a similar effect on CB 1 , CB 2 receptors and GPR18 (but not pplss-CB 1 or pplss-GPR18).Given that RAMP3 did not alter GPR55 expression, the effects on the other cannabinoid receptors seem specific.It is interesting that the pplss-containing constructs seem to be able to bypass the influence of RAMP3.For pplss-CB 1 receptors, this could simply be due to the extremely high receptor expression level overwhelming any RAMP3 effect.However, this is not the case for pplss-GPR18 which had approximately the same basal surface and total expression as non-pplss CB 1 receptors.
Conversely, pplss-CB 1 receptors enhanced the total expression of RAMP3 substantially but without any hint of enabling RAMP3 cell surface expression.RAMP2 total (but not surface) expression was also increased by pplss-CB 1 (but not CB 1 ), as well as CB 2 , GPR18 (but not pplss-GPR18) and GPR55.For the pplss-CB 1 effects, the exceptionally high expression level of pplss-CB 1 receptors makes it difficult to rule out a potential artefact or 'bystander' effect that might increase RAMP2 and 3 expression without being a meaningful interaction.The presence of signal sequences on both RAMPs and pplss-CB 1 also implies the potential for a bottleneck at the machinery required for signal peptide recognition and cleavage (Barbash et al., 2019), though this was not an issue for CLR, which also has a signal sequence.Furthermore, the total cellular expression of RAMP1 and MRAP1α was not affected by co-expression of pplss-CB 1 , so if a bystander effect is at play this at least implies that pplss-CB 1 and RAMPs 2 and 3 'cross paths' at some point in their life cycle, distinct from RAMP1 and MRAP1α.Similarly for CB 2 , GPR18 and GPR55, the effects of these receptors on RAMP2 total expression are unique within our dataset, so it seems unlikely these findings are artefacts.Speculating instead that there are genuine interactions with pplss-CB 1 , CB 2 , GPR18 and GPR55, we wonder whether these receptors could act in either competition or synergy with natural partners for RAMP2/3, altering cellular responses via either sequestering RAMPs or contributing to hetero-oligomeric receptor-accessory complexes.The same could be possible for the cannabinoid receptor and MRAP interactions discussed above, as both MRAPs can form multimers (Chen et al., 2020;Malik et al., 2015).
Our investigation is by no means exhaustive, and there are a number of avenues for potential future study.An important aspect to investigate will be whether the cannabinoid receptors and accessory protein interactions indicated represent physical interactions or are produced via indirect mechanisms.Techniques such as co-immunoprecipitation, BRET or BiFC, potentially also utilising mutagenesis, could be employed to further validate and clarify the nature of these interactions (Oyagawa & Grimsey, 2021).A considerable limitation is that we have only studied one cell line, which is not a natural host for the receptors and accessories studied.If more than the transfected pair of proteins is required for functional interaction, such relationships would likely have been missed in our study leading to a false-negative result.Even if an additional requisite protein partner was endogenously expressed in our cells, the expression level was likely to be considerably lower than the over-expressed receptor and accessory.Furthermore, there is also potential for receptor signalling outputs other than those we have studied to be specifically influenced by the presence of an accessory protein.We have also only studied one ratio between receptor and accessory protein expression.However, given that most receptors were expressed at a similar or lower level than the benchmark receptors CLR and MC 2 , it seems likely that sufficient accessory proteins would have been expressed to be able to produce an effect-for all except perhaps pplss-CB 1 with its extremely high expression level.All constructs in this study were expressed from constitutive promoters, which would have prevented any potential cross-influences on transcription via feedback mechanisms such as changes in basal signalling.An obvious next step in validating the physiological relevance of these putative interactions will be to evaluate these relationships in models with endogenous expression of the partners.Another potentially consequential aspect to investigate will be comparison of wild type with disease-causing mutated RAMPs and MRAPs (Gong et al., 2019;Ramachandrappa et al., 2013).
We have presented evidence for the existence of a variety of influences on expression and signalling between cannabinoid receptors and the MRAP and RAMP families of accessory proteins.There are many organs and cell types where cannabinoid receptors and these accessory proteins are co-expressed, and it will be very Figure legends indicate the number of independent experiments undertaken.The majority of figures show means from independent experiments with symbols (scatterplot) and overall mean and SEM shown with bars and lines.In a few instances, data are presented from one representative experiment showing mean and technical error as SD, as indicated in the figure legends.Data are often normalised to T A B L E 1 Plasmids used in this study.
. Briefly, HEK-293 cells (ATCC CRL-1573, RRID:CVCL_0045) were lifted with trypsin and seeded in complete medium (DMEM high glucose; Thermo Fisher Scientific, with 10% FBS) in six-well polystyrene cell culture plates at either 500,000 or 200,000 cells per well (for transfection after 24 h or 48 h, respectively).Cells were cultured to reach approximately 50% confluence at transfection.Transfection mixtures comprising 150 ng of each TRUPATH component plasmid (Gα 13 -Rluc8 pcDNA5/FRT/T0, Gβ 3 pcDNA3.1 and Gγ 9 -GFP2 pcDNA3.1)and 35 fmol of each receptor and MRAP plasmid of interest ($110-156 ng per plasmid, depending on the construct) were diluted in OptiMEM.In 'No Receptor' and 'No Accessory' conditions, receptor and/or accessory plasmids were substituted for CAT in pcDNA5/FRT (described earlier).Total DNA per well was $750 ng.DNA mixtures were combined with 6.75 μg of PEI MAX in OptiMEM, to a total transfection volume of 174 μl per well.Transfection mixtures were incubated at room temperature for 20 min and then dispensed dropwise on top of the culture medium.Cells were returned to the incubator overnight.The next day, transfected cells were lifted from six-well dishes, and seeded into PDL-treated white plates (Perkin Elmer, Waltham, MA, USA) in complete medium at 50,000 cells per well.Cells were cultured for a further 24 h prior to assay.On stimulation day, assay medium equilibration for each half-plate assay run was staggered to ensure matching conditions for each run.Cells were washed once with warm PBS, and then equilibrated for 30 min in an incubator in 80 μl per well of warm assay medium (DMEM high glucose phenol red-free, 25-mM HEPES, 1 mgÁml À1 BSA).Drugs were prepared at 10Â concentration in assay medium and were moved into 96-well polypropylene V-bottom plates for dispensing.Final drug concentrations were as shown in the figures and all conditions being subject to the same vehicle concentration.Coelenterazine 400a was prepared in assay medium at 10Â concentration immediately before dispensing to cells (10 μl, to a final concentration in 100 μl of 5 μM).After the coelenterazine was dispensed, BRET2 output was recorded for 5 min in the plate reader (LUMIstar Omega, BMG Labtech, 410-80 and 515-30 nm filters, with simultaneous detection and interval time of 0.5 s per well).At the end of this baseline read, 10 μl of the drug mixtures were dispensed, and the plate reader was then re-started to read for 25 min.Similar to cAMP assay analysis, raw readings from the Rluc emission channel were plotted and inspected for any abnormalities.After this verification, the mean BRET ratio for the readings taken over the first 10 min of drug stimulation was calculated, and concentrationresponse curves plotted (log[agonist] vs. response [three parameters]; . Data were graphed in Graph-Pad Prism.Statistical analysis was performed with SigmaPlot (v14.0,Systat Software).Statistical tests were undertaken on raw data (not normalised), utilising one-way ANOVA with repeated measures (raw data value ranges tended to vary between independent experiments).Data were first tested for adherence to assumptions of normality (Shapiro-Wilk) and equality of variance (Brown-Forsythe).If one or both tests did not pass, data were log transformed and the majority of datasets then passed the assumptions for parametric testing (exception noted below).For the receptor and accessory detection datasets, multiple raw untransformed datasets failed the normality and equal variance tests.We opted to log transform all the raw data for these experiments so that datasets were treated equivalently.If a result of a significant difference was found in the initial ANOVA test, a Holm-Šídák post hoc test with correction for multiple comparisons was performed with comparison to the control dataset.For five datasets that did not pass the normality test even after log transforming the data (RAMP1 and RAMP2 surface expression and RAMP3, MRAP1α and MRAP2 total expression), a non-parametric Kruskal-Wallis ANOVA on ranks was performed on the normalised data, with Dunn's post hoc test for comparison to the 'No Receptor' control.P values <0.05 were considered to indicate statistically significant differences.2.7 | MaterialsSuppliers of materials are shown in parentheses after the item(s) supplied.Molecular biology: XL10-Gold ultracompetent E. coli (Agilent), QIAprep Miniprep kit (QIAGEN), KAPA HiFi HotStart and LongRange Polymerases (KAPA Biosystems).Cell culture and transfection: culture flasks and general plasticware (Corning), Dulbecco's modified Eagle medium (DMEM) high glucose with L-glutamine and sodium pyruvate (Hyclone SH30243; Thermo Fisher Scientific 11995073), fetal bovine serum (FBS, NZorigin, Moregate), Trypsin 0.05% with EDTA and phenol red (Gibco, Billings, MT, USA and Thermo Fisher Scientific), poly-D-lysine (PDL, Sigma P1149), polyethylenimine (PEI) MAX (Polysciences, Warrington, PA, USA; 24765) and OptiMEM (Gibco, Thermo Fisher Scientific).Immunocytochemistry: clear 96-well plates (Nunc, Thermo Fisher Scientific), normal goat serum (Sigma, St Louis, MO, USA), anti-HA raised in mouse (Biolegend, San Diego, CA, USA; 901503, RRID:AB_ 2565005), anti-FLAG raised in rabbit (Sigma F7425, RRID:AB_439687), anti-mouse and anti-rabbit raised in goat Alexa Fluor 488 and 594 secondary antibodies (Thermo Fisher Scientific A11029, A11034, A11032 and A11037) and Hoechst 33258 (Applichem, Chicago, IL, USA).Signalling Assays: white 96-well plates (Costar Corning; CulturPlate Perkin Elmer), Hanks' Balanced Salt Solution with calcium and magnesium (HBSS; Gibco 14025, Thermo Fisher Scientific), DMEM high glucose phenol red-free (Gibco 21063029, Thermo Fisher Scientific), bovine serum albumin (BSA) low endotoxin and low free fatty acid (ICP Bio, Auckland, New Zealand), coelenterazine h (Nanolight #301), coelenterazine 400a (Nanolight #340), forskolin (Tocris, Bristol, UK), CP55,940 (Cayman, Ann Arbor, MI, USA), SR141716A (a generous gift from Roche, Switzerland), SR144528 (Abcam, Cambridge, UK), WIN55,212-2 (WIN; Cayman), L-α-lysophosphatidylinositol (LPI; Merck, Darmstadt, Germany), human adrenomedullin 15-52 (Bachem AG, Bubendorf, Switzerland), human α-calcitonin gene-related peptide (CGRP; Bachem AG), ACTH 1-24 (Bachem AG), N-arachidonoyl glycine (NAG; Cayman and ENZO, Farmingdale, NY, USA), O-1602 (O16; Cayman), Δ 9 -tetrahydrocannabinol (THC; THC Pharm GmbH, Frankfurt, Germany), 2-arachidonoyl glycerol (2-AG; Cayman) and 1-μM cannabidiol (CBD; THC Pharm GmbH).2.8 | Nomenclature of targets and ligandsKey protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, and are permanently archived in the Concise Guide to PHARMACOLOGY 2021/22(Alexander et al., 2021).
understand the relative starting expression of the cannabinoid receptors under investigation, we first compared cell surface and total expression of human HA-tagged receptors, transiently expressed in HEK-293S cells, without introducing accessory proteins (Figures1 and S1).Note that some expression constructs differed in the number of N-terminal HA tags, as indicated.Most constructs contained the same constitutive promoter (CMV), but pplss-CB 1 was expressed with the hEF-1α promoter.We selected the HEK-293S line because of the reported lack of endogenous RAMP expression(Qi et al., 2013;Wootten et al., 2013).Consistent with this, we could not detect CLR or MC 2 receptor-derived cAMP signalling in the absence of accessory protein co-transfection (FiguresS3 and S4), indicating that RAMPs or MRAPs are either not endogenously expressed in the cell line or only at an extremely low level in comparison to the introduced receptor.Expression of CLR was readily observable, and we utilised this receptor as a reference point for relative comparison.Cell surface and total cellular anti-HA labelling for 3HA-tagged MC 2 receptors were equivalent to 1HA-CLR, indicating that MC 2 receptors were likely to be expressed at a lower overall level than CLR, but with a similar surface:total expression ratio.1HA-CB 2 receptor surface expression was equivalent to 1HA-CLR, but total expression was $40% of 1HA-CLR, while 3HA-GPR55 had equivalent total HA staining to 1HA-CLR but greater cell surface labelling ($170%).This indicated that both CB 2 receptors and GPR55 had a proportionally greater cell surface to total expression ratio than CLR.Both cell surface and total cellular HA labelling for 3HA-CB 1 receptors and 3HA-GPR18 were extremely low compared with 1HA-CLR (<5%) but were detectable relative to the no receptor control (Figure1c).Due to the extremely low expression level of CB 1 receptors and GPR18, we decided to also utilise receptor constructs that incorporated a preprolactin signal sequence/peptide F I G U R E 1 Receptor expression in HEK-293S cells without accessory protein co-transfection.(a, b) anti-HA labelling of intact cells to label cell surface HA epitopes (a) and permeabilised cells to measure total cellular HA epitopes (b), normalised to that for CLR in matched experiments.Note that some receptor constructs have one HA tag and others three (as indicated), so in some cases, relative intensity of staining is not directly indicative of relative underlying receptor expression.Receptors were expressed from a CMV promoter, except pplss-CB 1 , which was expressed from hEF-1α.Data are measurements from independent experiments (n = 5), shown as individual values with means ± SEM.(c) Representative widefield images of total cellular anti-HA labelling for the indicated receptors.Presented image intensity demonstrates that CB 1 receptor and GPR18 staining is discernible from the 'No Receptor' negative control.The CLR image is saturated at this intensity due to high relative staining.Scale bar, 50 μm.

F
I G U R E 2 Receptor expression with RAMP co-transfection.Cell surface and total receptor expression (as detected by anti-HA labelling) with co-transfection of plasmids encoding untagged RAMPs 1, 2 or 3, normalised to co-transfection of a benign plasmid control ('No Accessory'; 'No Acc.') in matched experiments.(a) CLR cell surface and total cellular expression.(b) Cannabinoid receptor cell surface expression.(c) Cannabinoid receptor total cellular expression.The 'No Receptor' ('No Rec.') control indicates the relative level of non-specific staining detected.Data are measurements from independent experiments (n = 5), shown as individual values with means ± SEM. *P<0.05,significantly different from 'No Accessory' control.Statistical tests were performed on raw, not normalised, data; 'No Receptor' controls were not included in the statistical tests.
MRAP1α and MRAP2 had similar effects as each other on both CB 2 receptors and GPR55.Detection of cell surface CB 2 receptors and GPR55 reduced by approximately 30-40% with MRAP1α or MRAP2 co-expression (Figure5b).GPR55 total expression was also reduced and to a similar degree by both MRAPs.Total expression of CB 2 receptors was not altered by either MRAP, though we noted relatively large variability in results for MRAP2 co-expression and, F I G U R E 3 RAMP expression with receptor co-transfection.Cell surface and total RAMP expression (as detected by anti-FLAG labelling) with co-transfection of HA-tagged receptors or benign plasmid control ('No Receptor'; 'No Rec.').(a) Normalised to RAMP1 with CLR co-expression in matched experiments.(b, c) Normalised to co-transfection with CLR in matched experiments.The 'No Accessory' ('No Acc.') control indicates the relative level of non-specific staining detected.Data are measurements from independent experiments (n = 5), shown as individual values with means ± SEM. *P<0.05,significantly different from 'No Receptor' control (NB: statistical tests were performed on raw, not normalised, data).'No Accessory' controls were not included in the statistical tests.
with and without receptor co-transfection.MRAP1α was expressed robustly at the cell surface regardless of receptor co-expression but, interestingly, introduction of MC 2 receptors reduced detection of cell MRAP1α (Figure6a,b).Co-expression of MC 2 receptors did not significantly alter the total expression of MRAP1α (Figure6a,c).Cell surface F I G U R E 4 cAMP levels following CB 1 , pplss-CB 1 and CB 2 receptor agonist and inverse-agonist stimulation with RAMP co-transfection.cAMP levels (as detected by CAMYEL biosensor) for cells expressing (a) CB 1 , (b) pplss-CB 1 and (c) CB 2 receptors with co-transfection of RAMPs (R1/R2/R3) or benign plasmid control ('No Accessory'; 'No Acc.').Cells stimulated with forskolin plus 1-μM SR141716A (SR1), 1-μM SR144528 (SR2), varying concentrations of CP55,940 (CP) or matched vehicle (V), as indicated.Panel on left is data from one representative experiment (showing mean and technical error as SD), normalised to the 'top' or 'bottom' of the fitted curve as 100% (i.e., approximately forskolin without agonist) and the vehicle control (without forskolin) as 0%.Right-hand panels are data from independent experiments (n = 3) with mean and SEM, presented as normalised to (CP curve 'Span' and SR1/SR2 responses) or as difference from (ΔpEC 50 ) the 'No Accessory' control ('No Acc.') in matched experiments.Negative ΔpEC 50 indicates lower potency than the matched 'No Accessory' control.As only three independent experiments were carried out, these data should be considered exploratory and no statistical comparisons have been undertaken.F I G U R E 5 Receptor expression with MRAP co-transfection.Cell surface and total receptor expression (as detected by anti-HA labelling) with co-transfection of plasmids encoding FLAG-tagged MRAPs 1α or 2, normalised to co-transfection of a benign plasmid control ('No Accessory'; 'No Acc.') in matched experiments.(a) Cell surface and total cellular expression of MC 2 receptors.(b) Expression of cannabinoid receptors on cell surface.(c) Total cellular expression of cannabinoid receptors.The 'No Receptor' ('No Rec.') control indicates the relative level of non-specific staining detected.Data are measurements from independent experiments (n = 5), shown as individual values with means ± SEM. *P<0.05,significantly different from 'No Accessory' control (NB: statistical tests were performed on raw, not normalised, data).'No Receptor' controls were not included in the statistical tests.stainingfor MRAP2-FLAG was very low ($4%) in comparison with MRAP1α, though detection of total cell MRAP2 was more robust, at around 35% of that for MRAP1α.Neither MRAP2 surface nor total staining were modified detectably by MC 2 receptor co-expression.Three cannabinoid receptor constructs reduced MRAP1α surface detection in a manner equivalent to MC 2 , pplss-CB 1 , CB 2 receptors and GPR55 (Figure6b).None had any influence on MRAP1α total expression (Figure6c).Detection of cell surface MRAP2 was enhanced approximately threefold by co-expression of pplss-GPR18 in comparison with MC 2 receptor co-expression.pplss-GPR18 did not significantly affect MRAP2 total expression relative to the 'No Receptor' control but did increase MRAP2 total expression in comparison with MC 2 co-expression in all experiments.Although not producing a statistically significant difference from the 'No Receptor' control, coexpression of pplss-CB 1 enhanced MRAP2 surface detection in comparison with MC 2 co-expression in all experiments.Co-expression of pplss-CB 1 also produced a clear increase in MRAP2 total cellular expression.

F
I G U R E 6 MRAP expression with receptor co-transfection.Cell surface and total expression of MRAP (as detected by anti-FLAG labelling) with co-transfection of HA-tagged receptors or benign plasmid control ('No Receptor'; 'No Rec.').(a) Normalised to MRAP1α with co-expression of MC 2 receptors in matched experiments.(b, c) Normalised to co-transfection with MC 2 receptors in matched experiments.The 'No Accessory' ('No Acc.') control indicates the relative level of non-specific staining detected.Data are measurements from independent experiments (n = 5), shown as individual values with means ± SEM. *P<0.05,significantly different from 'No Receptor' control (NB: statistical tests were performed on raw, not normalised, data).'No Accessory' controls were not included in the statistical tests.

F
I G U R E 7 cAMP levels following CB 1 , pplss-CB 1 and CB 2 receptor agonist, and inverse-agonist stimulation and Gα 13 -β 3 γ 9 dissociation following pplss-CB 1 and GPR55 agonist stimulation, with MRAP co-transfection.(a-c) cAMP levels (as detected by CAMYEL biosensor) for cells expressing (a) CB 1 , (b) pplss-CB 1 and (c) CB 2 receptors with co-transfection of MRAPs or benign plasmid control ('No Accessory'; 'No Acc.').Cells stimulated with forskolin plus 1 μM SR141716A ('SR1'), 1 μM SR144528 ('SR2'), varying concentrations of CP55,940 ('CP') or matched vehicle ('V'), as indicated.Panel on left is data from one representative experiment (showing mean and technical error as SD), normalised to the 'top' or 'bottom' of the fitted CP response curve as 100% (i.e., approximately forskolin without agonist) and the vehicle control (without forskolin) as 0%.Right-hand panels are data from independent experiments (n = 3) with mean and SEM, presented as normalised to (CP curve 'Span' and SR1/SR2 responses) or as difference from (ΔpEC 50 ) the 'No Accessory' control ('No Acc.') in matched experiments.Negative ΔpEC 50 indicates lower potency than the matched 'No Accessory' control.As only three independent experiments were carried out, these data should be considered exploratory and no statistical comparisons have been undertaken.NB: 'No Accessory' control data are the same as in Figure 4 as these experiments were conducted concurrently.(d, e) Dissociation of Gα 13 from β 3 γ 9 (as detected by TRUPATH biosensor) for cells expressing (d) pplss-CB 1 or (e) GPR55 with co-transfection of MRAPs or benign plasmid control ('No Accessory'; 'No Acc.').'No Receptor' controls (transfected only with benign plasmid) were also included for both sets.Cells stimulated with varying concentrations of WIN55,212-2 ('WIN'), Lα-lysophosphatidylinositol (LPI), or matched vehicle ('V'), as indicated.Panel on left is data from one representative experiment (showing mean and technical error as SD), normalised to the 'bottom' of the fitted agonist (WIN or LPI) response curve for the 'No Accessory' condition as 100% (i.e., approximately vehicle), and the 'No Receptor' control incubated with vehicle as 0%.Right-hand panels are data from independent experiments (n = 5), shown as individual values with means ± SEM and presented as normalised to (agonist response curve 'Span', 'E max ' and constitutive activity) or as difference from (ΔpEC 50 ) the 'No Accessory' control ('No Acc.') in matched experiments.Negative ΔpEC 50 indicates lower potency than the matched 'No Accessory' control.*P<0.05,significantly different from 'No Accessory' control (NB: statistical tests were performed on raw, not normalised, data).
interesting to investigate these interactions further and determine their relevance in physiology and/or disease processes.Even in cases where alterations to cannabinoid receptors or accessory protein expression did not result in modulation of signalling (so far as we detected), cannabinoid receptor interactions with and/or influences on expression of accessory proteins could be a mechanism for crosstalk with receptors whose functions are heavily influenced by accessory protein expression.Our findings support earlier hypotheses that RAMP and MRAP accessory proteins can interact with GPCRs outside the calcitonin and melanocortin receptor families (respectively), and open new avenues for exploration of the complexities of the endocannabinoid system.