MOP and NOP receptor interaction: Studies with a dual expression system and bivalent peptide ligands

Opioids targeting mu;μ (MOP) receptors produce analgesia in the peri-operative period and palliative care. They also produce side effects including respiratory depression, tolerance/dependence and addiction. The N/OFQ opioid receptor (NOP) also produces analgesia but is devoid of the major MOP side effects. Evidence exists for MOP-NOP interaction and mixed MOP-NOP ligands produce analgesia with reduced side effects. We have generated a HEKMOP/NOP human expression system and used bivalent MOP-NOP and fluorescent ligands to (i) probe for receptor interaction and (ii) consequences of that interaction. We used HEKMOP/NOP cells and two bivalent ligands; Dermorphin-N/OFQ (MOP agonist-NOP agonist; DeNO) and Dermorphin-UFP101 (MOP agonist-NOP antagonist; De101). We have determined receptor binding profiles, GTPγ[35S] binding, cAMP formation and ERK1/2 activation. We have also probed MOP and NOP receptor interactions in HEK cells and hippocampal neurones using the novel MOP fluorescent ligand, DermorphinATTO488 and the NOP fluorescent ligand N/OFQATTO594. In HEKMOP/NOP MOP ligands displaced NOP binding and NOP ligands displaced MOP binding. Using fluorescent probes in HEKMOP/NOP cells we demonstrated MOP-NOP probe overlap and a FRET signal indicating co-localisation. MOP-NOP were also co-localised in hippocampal tissue. In GTPγ[35S] and cAMP assays NOP stimulation shifted the response to MOP rightwards. At ERK1/2 the response to bivalent ligands generally peaked later. We provide evidence for MOP-NOP interaction in recombinant and native tissue. NOP activation reduces responsiveness of MOP activation; this was shown with conventional and bivalent ligands.

Introduction While the effects of selective peptide agonists, such as DAMGO and N/OFQ [5,10,33,35], have been studied in co-expression systems, very little evidence exists demonstrating the activity of mixed ligands in a co-expression system. In order to assess the cellular aspects of mixed ligand interactions in a system co-expressing MOP and NOP, we have developed a human HEK co-expression system (HEK MOP/NOP ). We have used the bivalent MOP/NOP full agonist DeNo (15), in conjunction with the newly synthesised De101. De101 is a counterpoint to the ligand DeNO (S4 Fig in S1 File). Binding and functional activity assays were performed to determine the suitability of De101 as a test ligand. Furthermore, we visualised and determined co-expression using two novel fluorescent ligands, N/OFQ ATTO594 [14] for NOP and Dermorphin ATTO488 for MOP [39]. These imaging experiments were performed both using HEK MOP/ NOP and mouse CA1 hippocampal neurons to determine the potential for receptor co-localisation Ex Vivo. Mouse CA1 hippocampal neurones have been shown to express both MOP [40] and NOP receptors [41], however colocalisation/coexpression has not yet been definitively proven.
We hypothesise that co-expression of MOP and NOP will lead to measurable changes in downstream signalling pathways, such as GTPγ[ 35 S] binding and cAMP inhibition, as well as changes in the activation of the mitogen-activated protein kinase, ERK1/2.

Development of Co-expression system
Plasmids containing cDNA clones for human MOP or human NOP receptors were purchased from cDNA Resource Centre (www.cdna.org). Clones were individually transfected into HEK293 cells using Fugene HD (Promega, UK). Cell lines stably expressing the MOP receptor or NOP receptor were selected by adding geneticin (800μg/ml for MOP) or hygromycin B (600μg/ml for NOP) to cell culture medium. Following selection of preferred HEK MOP single expression clone, cells were co-transfected with NOP, with co-expression selected by coadministering geneticin (400μg.ml -1 ) and hygromycin B (200μg.ml -1 ). Surface expression of receptors for the single and co-expression systems was determined by using radioligand binding assays as described below (further cloning details can be found in Supplement section 1 in S1 File).

Synthesis of De101
The MOP/NOP ligand De101 was synthesized using a classical thiol-Michael reaction following the experimental conditions previously reported for the synthesis of DeNo using [Cys 18 ] UFP-101 in place of [Cys 18 ]N/OFQ-NH 2 . Structure and Chemistry can be found in S4-S6 Figs in S1 File.

Membrane preparation
CHO OPIOID or HEK OPIOID cells were harvested, homogenised and resulting membrane fragments resuspended in wash buffer (50mM Tris-HCl and 5mM MgSO 4 , pH 7.4 with KOH) for saturation and displacement assays or a homogenisation buffer (50mM HEPES and 1mM EDTA, pH 7.4 with NaOH) for GTPγ[ 35 S] functional assays. Membrane suspensions were sedimented at 20,374g for 10 minutes at 4˚C, with the process being repeated three times. The pellet was suspended in the appropriate buffer at the desired volume before protein concentration was measured using a Lowry assay [15,46]. In both saturation binding and displacement binding studies, samples were incubated for 1 hr at room temperature before reactions were terminated by vacuum filtration onto polyethylenimine (PEI) soaked Whatman-GF/B filters, using a Brandel Harvester [14].

Cyclic adenosine monophosphate inhibition assays
HEK MOP , HEK NOP or HEK MOP/NOP cells were suspended in Krebs/HEPES buffer, containing isobutylmethylxanthine (1mM) and forskolin (1μM) as appropriate. Varying concentrations (0.1pM-1μM) of the test ligands were added prior to incubation. Following a 15 minute incubation at 37˚C, the reaction was terminated by addition of 10M HCl, following which 10M NaOH and Tris-HCl (1M, pH7.4) were added to neutralise the reaction followed by centrifugation (13,000g, 2 min). Binding protein from bovine adrenal cortex was used to measure cAMP collected from the supernatant as in [47].
In order to probe for loading controls, membranes were stripped using Restore Plus™ (ThermoFisher, UK) for 15 minutes. Membranes were thoroughly washed in TBS-T and blocked using the 5% milk/TBS-T for 1 hr at room temperature. Membranes were again probed overnight at 4˚C, using ERK1/2 (total) antibody (1:3000 dilution in TBS-T) following which detection of the immune-reactive band was achieved as previously described. Normalisation of total protein levels was achieved by representing levels of phosphor-ERK1/2 as a proportion of total ERK1/2 protein [15].

Mouse hippocampal brain slices
This study was conducted in accordance with the UK Animals Scientific Procedures Act, 1986 and following approval by the animal welfare and ethics committee of the University of Leicester. Pregnant C57/BL6 mice, at stage between E11-E13, were obtained from Charles River UK and delivered in house. Hippocampal slices were prepared from male and female C57/BL6 mouse pups aged between postnatal day (P) 6 to 9. Mice were housed in a 12 hour light/dark cycle and food and water was given ad libitum. Mice were humanely killed in accordance with home office guidelines by schedule 1 using cervical dislocation followed by swift decapitation.
Hippocampal slices were prepared by the methods denoted in [48], with further modifications to promote neural outgrowths. Dissection buffer (DB) utilised for dissections consisted of HBSS (Hank balanced salt solution), 4.5mg/ml glucose and 3.75μg/ml of amphotericin B. Culture media contained 50% MEM, (minimal essential medium), 25% HBSS, 25% heat inactivated horse serum, 4.5mg/ml glucose, 3.75μg/ml of amphotericin B and 0.5 mM glutamine. Following dissection, brains were transferred to a sterile petri dish on ice containing ice cold DB. After separation of the hemispheres with a scalpel, the hippocampi were isolated whilst the tissue was fully submerged in ice cold DB. Transverse hippocampal slices at 350μm were prepared using a tissue chopper (McIlwain tissue chopper). Using sterile syringe needles, slices were carefully separated and transferred into 6 well plates containing 28mm Menzel glaser #1-coverslips (Thermos Scientific, UK) with the prior addition of Celltak™ (1μg.mL1) (Sigma, UK). This promotes adhesion of slices and neuronal outgrowth. Slices were submerged in 1ml of pre-warmed media. Plates were placed in a humidified incubator at 37˚C and perfused with 95% O 2 and 5%CO 2 for a minimum of one week prior to inspection for neuronal outgrowths. Culture media was topped up every 2-3 days. Twenty-four hours prior to experimental use, cells were treated with a complex containing anti-Neun Ab attached to an anti-rabbit Alexa-fluor 405nm secondary Ab (UK) coated in 10μL Ab-Delivirin™ to allow identification and visualisation of neuronal cells [41].

Confocal microscopy
Confluent HEK MOP , HEK NOP and HEK MOP/NOP cells (and hippocampal slices) were grown on ethanol-sterilised coverslips (28mm Menzel glaser #1), incubated for 24h before being transferred to a Harvard PDMI-2 peltier unit and continually perfused with ice-cold Krebs buffer, pH 7.4, maintaining a constant temperature of 4˚C.
N/OFQ ATTO594 and Derm ATTO488 were injected either separately or together at a concentration of 100nM following which images were acquired using a Nikon C1Si confocal microscope (60X oil immersion objective). N/OFQ ATTO594 (excitation 594 nm; emission 620 nm), and/or Derm ATTO488 (excitation 488 nm; emission 520 nm) were allowed to incubate for 5 min, before coverslips were washed with the ice-cold Krebs. Following wash-off, cells were imaged at the desired wavelength (Derm ATTO488 -488nm, image capture at 520 nm; N/OFQ ATTO594 -594 nm, image captured at 620nm), with the images collected by Nikon C1Si software [14,39]. As both probes are agonists all experiments were undertaken at 4˚C to prevent receptor activation and internalisation.
FRET studies were undertaken by measuring the binding of both 100nM Derm ATTO488 and 100nM N/OFQ ATTO594 on HEK MOP/NOP cells. Derm ATTO488 was stimulated by the 488nm laser (laser power: 20%; Gain: 6.85) with measurements assessed in both the green and red channel [49]. To confirm FRET-pairing, several controls were undertaken (see Fig 7). N/ OFQ ATTO594 was incubated on HEK MOP/NOP cells alone and measured with the 488 nm laser, while photobleaching of N/OFQ ATTO594 was undertaken using 594nm laser (laser power: 50% Gain: 7.10) in the presence of Derm ATTO594 to measure its relative fluorescence pre and post photobleaching [50]. All FRET experiments were undertaken at 4˚C to prevent receptor activation and internalisation.

Statistical analysis
For radioligand assays and Western blot techniques all data are expressed as the mean ± SEM (n). GraphPad Prism V6.05 (San Diego, USA) was used for curve fitting and analysis. For displacement binding studies, the concentration which caused 50% displacement of the radioligand (IC 50 ) was corrected for the competing mass of radioligand by use of the Cheng and Prusoff equation using the pK d values produced from saturation binding experiments (S2, S4 and S6 Tables in S1 File) [51]. GTPγ[ 35 S] binding results are expressed as a stimulation factor (agonist stimulated specific binding/basal specific binding) [15]. In cAMP inhibition studies, results are expressed as percentage inhibition of forskolin stimulation. For ERK1/2 activity, following normalisation of the protein bands, ERK1/2 activity is measured as activity compared to basal levels. Statistical analysis was performed using t-test or one way ANOVA with Bonferroni correction, as described in figure and table legends. For Confocal microscopy, FIJI was used to analyse colocalization using the Ezcolocalization plugin to determine Pearson correlation coefficient [52]. This software measures relative overlapping fluorescence intensities to produce statistical data (Pearson coefficient correlation) to determine levels of co-expression. Changes in fluorescence during photobleaching experiments was measured as previously described using corrected total cell fluorescence [Corrected total cell fluorescence = Integrated density − (Area of selected cell × Mean fluorescence of background readings)] [53].

Characterisation of De101
In displacement binding studies at CHO hMOP , Dermorphin and De101 displaced the binding of [ 3 H]-DPN in a concentration dependent and saturable manner ( Table 1
In HEK MOP/NOP cells, both [ 3 H]-DPN and [ 3 H]-N/OFQ were able to bind and both were displaced by naloxone, N/OFQ or a combination of both (Fig 3C and 3D). Tritiated-DPN, in combination with naloxone, produced a B max of 464 and pK d of 8.72. Tritiated-DPN, with N/ OFQ as the NSB, produced a B max of 169 and pK d of 8.59. When using [ 3 H]-N/OFQ and N/ OFQ to determine non-specific binding, a B max of 689 and pK d of 9.51 was produced. Tritiated-N/OFQ, with naloxone as NSB, produced a B max of 228 and pK d of 10.05 . Cross displacement, in terms of NSB determination, suggests an interaction; this was probed further in a series of displacement experiments.
Displacement binding assays. Dermorphin displaced [ 3 H]-DPN in HEK MOP cells (pK i : 8.32) and in HEK MOP/NOP cells (pK i : 7.86), with a small (but not statistically significant)  Table 2). Dermorphin was unable to displace [ 3 H]-N/OFQ in HEK NOP cells, but was able to displace this radioligand in the co-expression system (pK i : 7.91) (Fig 4 and Table 2) displaying similar affinity when displacing [ 3 H]-DPN in this cell line.
N/OFQ displaced [ 3 H]-N/OFQ in HEK NOP cell membranes (pK i : 9.39) and in the HEK-MOP/NOP co-expression system with a pK i of 9.11 (Fig 4 and Table 2). N/OFQ was unable to displace [ 3 H]-DPN in HEK MOP cell membranes but did displace this radioligand in the coexpression system with a pK i of 8.75; displaying similar affinity to that achieved in the coexpression system when using [ 3 H]-N/OFQ. Both the bivalent ligands, DeNo and De101, displayed affinity across all three cell lines. In HEK MOP cells, DeNo displaced [ 3 H]-DPN with a pK i of 9.00, DeNo demonstrated a statistically significant reduction in affinity for the MOP receptor in the co-expression system (pK i : 8.44) (Fig 5 and Fig 5), this was statistically   Confocal microscopy. Co-localisation of MOP and NOP was determined through coincubation with N/OFQ ATTO594 (NOP) and Dermorphin ATTO488 (MOP) (Fig 6A-6C). In HEK MOP/NOP cells, binding of Dermorphin ATTO488 and N/OFQ ATTO594 was co-localized with a Pearson correlation coefficient of 0.91 suggesting that the two ligands, and hence receptors, were in close proximity to each other. Binding of N/OFQ ATTO594 in HEK MOP/NOP was reversed in the presence of 10 μM SB-612111, but not the MOP antagonist Naloxone (10μM) (Fig 6D  and 6E). Binding of Derm ATTO488 was reversed in the presence of 10 μM Naloxone in HEK-MOP/NOP , but not 10 μM SB-612111 (Fig 6F and 6G).
In order to assess the potential for receptor interaction in non-recombinant systems, hippocampal (CA1) slices from mouse brain were incubated with 100 nM each of Dermorphin ATTO488 and N/OFQ ATTO594 (Fig 6H-6K). Hippocampal slice tissue opioid receptor mRNA expression profile is shown in S7 Table in S1 File. The anti-NeuN antibody was used to select neuronal processes. Both fluorescent opioid ligands bound in similar regions (Pearson correlation coefficient 0.83), further suggesting receptor interaction in native tissue (Fig 6K). Importantly, binding of Derm ATTO488 was fully inhibited by 10 μM of the MOP selective antagonist CTOP indicating the probe bound fully to MOP, while CTOP did not inhibit the binding of N/OFQ ATTO594 (Fig 6M and 6O). Administration of 10 μM of SB-612111 inhibited the binding of N/OFQ ATTO594 , but not Derm ATTO488 , again demonstrating selectivity for their respective receptors ex vivo (Fig 6L and 6N).
To further demonstrate the close proximity of MOP and NOP in HEK MOP/NOP cells, FRET experiments were performed using Derm ATTO488 and N/OFQ ATTO594 (Fig 7A-7D). Derm ATTO488 was incubated with HEK MOP/NOP cells (100nM) and fluorescence was measured by 488 nm laser in both green (A) and red (B) filter channels. Fluorescence was detected in the green channel after stimulation by 488 nm laser, but not in the red channel. Subsequently, 100 nM N/OFQ ATTO594 was added and measured using the 488nm laser which does not activate N/OFQ ATTO594 alone. Fluorescence was detected in the red channel (Fig 7C), indicating FRET and therefore close proximity (Fig 7D) of Derm ATTO488 and N/OFQ ATTO594 and by inference of their receptors. Fig 7E-7M demonstrate the relevant controls to demonstrate FRET.  red channel (B). Following the addition of N/OFQ ATTO594 , fluorescence is now detected by FRET in the red channel following stimulation with the 488nm laser (C). There is significant overlap with the green channel (D), shows merged image (Derm ATTO 488 excited by 488nm wavelength, N/OFQ ATTO594 by 594nm wavelength) demonstrating colocalisation (Pearson Correlation Coefficient = 0.89) further indicating a close proximity between Derm ATTO488 and N/OFQ ATTO594 . In order to demonstrate FRET-pairing of Derm ATTO488 and N/OFQ ATTO594 , a series of control experiments were performed. N/OFQ ATTO594 (100nM) was incubated with HEK MOP/NOP cells and stimulated by 488nm laser (E) and the 594nm laser (F). No fluorescence emission was demonstrated by stimulation with the 488nm laser, while N/OFQ ATTO594 was activated by the 594nm laser. A further confirmation of FRET pairing is through photobleaching of the acceptor molecule. HEK MOP/NOP cells were co-incubated with 100nM Derm ATTO488 (G) and N/OFQ ATTO594 (H) with co-localisation demonstrated in (I). N/OFQ ATTO594 was exposed to 594nm until loss of fluorescence (i.e photobleaching) was seen (J). At this point, Derm ATTO488 fluorescence was measured (K) following which a heatmap (L) was generated, which demonstrated several areas of increased fluorescence (orange). The graph (M) demonstrates levels of fluorescence produced by Derm ATTO488 in HEK MOP/NOP pre and post photobleaching (p<0.05; student's t-test). Data are the mean±SEM of five experiment. Scale bar (white) represents 20μm.
https://doi.org/10.1371/journal.pone.0260880.g007 50 in the coexpression system (Fig 8 and Table 3). N/OFQ stimulated a response in both HEK NOP and HEK MOP/NOP cell lines; in this case there was no significant change in agonist potency. DeNo produced a response in both MOP and NOP single expression and also in the co-expression system However, pEC 50 values (pEC 50 :7.63; E max :1.26). in the co-expression system were significantly lower than those in both HEK MOP and HEK NOP cell membranes. The value for DeNO was not significantly different from that demonstrated by Dermorphin in HEK MOP/ NOP . De101 produced a response in HEK MOP cells but failed to produce a response in HEK NOP cells. De101 produced a pK b of 8.67±10 at 100 nM in antagonist assays when co-incubated with N/OFQ (Fig 10. De101 stimulated binding of GTPγ[ 35 S] in HEK MOP/NOP cell membranes producing a pEC 50 of 8.67 and E max of 1.34 (Fig 8 and Table 3). De101 demonstrated a significant increase in pEC 50 when compared to Dermorphin in the co-expression system. These data seem to indicate that where NOP is stimulated the response to MOP is shifted rightwards.

GTPγ[ 35 S] assays. Dermorphin stimulated the binding of GTPγ[ 35 S] in both HEK MOP and HEK MOP/NOP cell lines, with a statistically significant decrease in the pEC
It has previously been reported that linker length in bivalent pharmacophores can have an effect on receptor binding leading to changes in potency and efficacy. In order to determine whether such effects were occurring in DeNo and De101, Dermorphin and N/OFQ were co- incubated (as individual unlinked peptides) in the co-expression system. This combination of ligands produced a response (pEC 50 :7.70; E max :1.28), which was not significantly different from the pEC 50 for DeNo in the co-expression system, but was significantly different from their respective pEC 50 values in single expression systems (S7A Fig in S1 File and Table 3). Co-incubation of Dermorphin and UFP-101 (pEC 50 : 8.46; E max :1.33) produced a response similar to De101 in the co-expression system. These data suggest that linkage per se was unimportant.
Cyclic Adenosine Monophosphate (cAMP) assay. cAMP inhibition assays demonstrated a similar trend in function for both the monovalent and bivalent ligands. Dermorphin produced a concentration-dependent inhibition of forskolin-stimulated cAMP formation in HEK MOP and HEK MOP/NOP whole cells. Dermorphin pEC 50 was significantly reduced in the co-expression system (Table 3 and Fig 9). N/OFQ produced concentration-dependent inhibition of forskolin stimulated cAMP formation in HEK NOP and HEK MOP/NOP cells. There was no significant difference in the values obtained for N/OFQ in both cell lines (Table 3 and Fig 9).
In HEK MOP , HEK NOP and HEK MOP/NOP cells, DeNO produced a concentration-dependent inhibition of forskolin stimulated cAMP formation. The pEC 50 value obtained by DeNo in HEK MOP/NOP cells was significantly lower than both values obtained in HEK MOP or HEK NOP cells (Table 3 and Fig 9). De101 produced a concentration-dependent inhibition of forskolin stimulated cAMP formation in HEK MOP and HEK MOP/NOP cells while being inactive in HEK NOP cells. In HEK MOP/NOP cells, De101 produced a pEC 50 of 8.81 and E max of 85.99%, which was significantly higher than Dermorphin in this cell line. (Table 3 and Fig 9).  Table 3). Dermorphin and N/OFQ produced a pEC 50 of 8.04, similar to both Dermorphin alone, or DeNO in the co-expression system but significantly lower than N/OFQ alone in HEK MOP/NOP and HEK NOP (Table 3). Dermorphin co-administered with UFP-101 produced a pEC 50 of 8.95, which was significantly greater than Dermorphin in the co-expression system, but similar to the potency of Dermorphin in HEK MOP cell line (Table 3).
A further demonstration of De101 antagonist activity was measured in the cyclic AMP inhibition assay. In these experiments De101 produced a pK b of 8.77 (±0.18) (Fig 10B). As in GTPγ[ 35 S] assays, in the co-expression system NOP activation led to a rightward shift in the curve of MOP agonists. The results would indicate activation of NOP leads to an inhibitory action on MOP receptor ligand activation when these two receptors are co-expressed. ERK1/2 activity. Dermorphin (1 μM) produced a time-dependent increase in pERK1/2 in both HEK MOP and HEK MOP/NOP cells (Fig 11). In HEK MOP cells this peaked at 10 min (maximum fold phosphorylation 7.21), which subsequently returned to basal levels after 15 min. In HEK MOP/NOP cells, Dermorphin peaked at 10 minutes (maximum fold phosphorylation 11.82) and remained elevated for the duration of the time course (30 min).
N/OFQ (1 μM) produced a time-dependent increase in pERK1/2 in both HEK NOP and HEK MOP/NOP cells (Fig 11). In HEK NOP cells, N/OFQ produced a biphasic stimulation of ERK1/2 activity which peaked at 5 (3.91 Fold) and 10 min (5.89 fold), and returned to basal after 15 min. In HEK MOP/NOP cells, N/OFQ-stimulated pERK1/2 activity reached a peak at 7.5-10 min (~5 fold). ERK1/2 activity reduced from 15 min and remained above basal activity for the duration of the time course.
Full uncropped blots are available as a supplementary document.

Discussion
We have generated a MOP/NOP co-expression system which we have used to determine (i) how targeting of two opioid receptors affects cellular signalling cascades and (ii) evidence for receptor interaction. Our HEK MOP/NOP expressed similar numbers of NOP (689 fmol/mg protein) and MOP (464 fmol/mg protein) receptors. Similar levels of expression of the two receptors of interest is important such that the potential to create receptor and coupling reserves is equal. Using our novel fluorescent ligands [14,39], we aimed to determine whether MOP and NOP receptors were expressed in close proximity to each other in our co-expression system. A significant overlap in binding of both fluorescent probes on the HEK MOP/NOP cell surface was detected, producing a Pearson coefficient correlation of 0.91 (High level of colocalization). Furthermore, FRET experiments demonstrate the close proximity of the fluorescent ligands and, by extension, MOP and NOP receptors. For FRET to occur ligands must be within 10 nm of each other [54] thereby indicating that these two receptors are close enough to potentially form a structural interaction [55]. However, this is not a natural system with both receptors expressed due to cloning techniques. In order to demonstrate the potential for co-expression in a native system, we assessed the binding of N/OFQ ATTO594 and Derm ATTO488 in ex vivo CA1 hippocampal neuronal processes. The NOP selective antagonist SB-612111 and the MOP-selective antagonist CTOP were used to demonstrate selectivity of binding of the fluorescent ligands in the native system and, fluorescent antibodies to NeuN were used to identify  neuronal outgrowths. A Pearson coefficient correlation of 0.83 again indicates a high level of co-expression and, therefore, co-expression in the same cell. As a reminder and a note of caution with these experiments; the ligands are both agonists so to prevent internalisation imaging was performed at 4˚C and this may limit receptor movement in the membrane. If there is a constitutive interaction then there are no issues but if an interaction is ligand driven low temperature may reduce or slow this interaction. Experiments using other methods to reduce internalisation such as high sucrose could address this issue. Taken as a whole this data provides evidence of MOP/NOP interaction also in native tissue.
Primary pharmacological support for the cellular interaction of MOP and NOP comes from the observation that Dermorphin was able to displace [ 3 H]-N/OFQ in the co-expression system, and conversely N/OFQ was able to displace [ 3 H]-DPN, characteristics not demonstrated by these ligands in single expression systems. Moreover, all ligands tested demonstrated a loss of affinity at the MOP receptor in the co-expression system.
The ability of high affinity MOP receptor ligands to displace [ 3 H]-N/OFQ has been demonstrated previously [34]. However, to the best of our knowledge the effect of NOP ligands on MOP receptor binding has not been previously demonstrated. In this paper, we show that displacement of radioligand binding by MOP or NOP selective ligands (as demonstrated in single receptor systems) is bidirectional in the co-expression system. Both Dermorphin and N/OFQ fail to produce 100% displacement of these respective radioligands in HEK MOP/NOP cell membranes. For both DeNo and De101, we demonstrate that these bivalent ligands are able to displace 140% and 120% of [ 3 H]-DPN in HEK MOP/NOP membranes, which is significantly higher than that of the single pharmacophore ligands. This effect is not seen with in [ 3 H]-N/OFQ displacement assays. We have no obvious explanation for this phenomenon but modification of the binding pocket(s) in a potential dimeric conformation might modify the way bivalent ligands interact.
If drugs are to be developed targeting the MOP/NOP heterodimer it is essential to understand any potential changes in downstream signalling. The first pathway investigated in this study was the initial G-protein activation, through measurement of GTPγ 35 S binding. The first evidence for changes in signalling due to direct interaction of MOP and NOP is seen with the peptide Dermorphin, with a loss of potency in the co-expression system when compared to the HEK MOP single-expression system. This is comparable with loss of potency seen in DAMGO by Wang and colleagues (2005) in their MOP/NOP co-expression system [35]. There is limited research to determine the effects of co-administration of MOP and NOP ligands, or drugs developed to target both receptors simultaneously, an area this study seeks to address. From the perspective of the NOP receptor, N/OFQ potency increases in the coexpression system. Of more interest is the activity of the bivalent pharmacophore, DeNO. DeNo demonstrates a significant loss of potency when compared to results in both HEK MOP and HEK NOP . DeNo demonstrated similar potencies to Dermorphin in previous studies, so to determine whether the loss of potency was as a result of cellular interaction of MOP and NOP, Dermorphin and N/OFQ were administered in HEK MOP/NOP cells. The results obtained demonstrated a similar potency for Dermorphin in the co-expression system (significantly lower than in single expression system), but also demonstrated a loss of potency for N/OFQ when co-administered with Dermorphin. The MOP agonist-NOP antagonist ligand De101, demonstrated no changes in potency when administered in the co-expression system when compared to single expression systems. Furthermore, the co-administration of Dermorphin and the NOP antagonist UFP-101 produced a potency similar to that produced in HEK MOP cells by Dermorphin alone.
The results seen in GTPγ[ 35 S] functional assays were mirrored by those seen in cAMP assays. Dermorphin, DeNO and a combination of Dermorphin and N/OFQ all demonstrated a reduction in potency in HEK MOP/NOP cells. N/OFQ, administered alone, produced a higher potency in the co-expression system, while De101 demonstrated unchanged potency values in the co-expression system when compared to HEK MOP . These results further support the suggestion of a MOP/NOP heterodimer that, when formed, leads to changes in signalling.
The final signalling pathway investigated was the ERK1/2 MAPK pathway, activated by both canonical (G-protein) and non-canonical signalling (Arrestin) pathways. The ERK1/2 pathway is involved in numerous cellular functions including proliferation, transcription activation and cell death, all of which are controlled by spatio-temporal activation of the protein itself, with activation at MOP beginning around 2-5 minutes post drug administration peaking from 7.5-10 minutes post administration [56]. The bivalent pharmacophores, DeNo and De101, both demonstrated a significant delay in activation of ERK1/2 in the HEK MOP/NOP line when compared to Dermorphin and N/OFQ. More interestingly, the peak activation of ERK1/ 2 by DeNO occurred at the latest time point in our study. The results again suggest the interaction of MOP and NOP may lead to significant changes in opioid ligand signalling, whether this be through structural interaction or changes in recruitment of G-protein receptor kinase (GRK) and arrestin requires further investigation. As previously demonstrated by Hawes and colleagues, both NOP and MOP activate MAPK via G i beta/gamma. Moreover, in CHO cells expressing both receptors pre-treatment with N/OFQ reduced NOP and MOP activation (with DAMGO) of MAPK. These data indicate NOP is modulating MOP signalling [57].
While we provide strong evidence (structural and functional) for direct interaction between MOP and NOP, we cannot confirm heterodimerisation. The overall effect of targeting both MOP and NOP simultaneously appears to be negative with respect to MOP. An important question arising from these findings is does this negative effect carry forward to physiological action with regards to analgesia? The results provided by mixed MOP/NOP agonists such as buprenorphine [16] and cebranopadol [17] wouold suggest that the overall effect is beneficial, as these drugs d display analgesia with reduced side-effects. That said this may be due to targeting MOP and NOP on different cells/pathways or on the same cell if structural receptor interaction is a driver. Future work could include disruption of the heterodimer using single transmembrane domains such as the example used by He and colleagues, whereby the introduction of MOR TM1 -TAT lead to the disruption of the MOP-DOP heterodimer [58].
There are several potential limitations to our work probing MOP and NOP interactions. Firstly, the use of transfected cells which can produce significantly higher numbers of receptors than seen native tissue; we have tried to control for this by selecting clones with similar levels of expression for MOP and NOP in single expression systems and both MOP/NOP in the double expression system. Expression differences/overexpression could lead to 'forced' interactions between the receptors, to manifest as co-localisation or changes in signalling pathways due to receptor competition for G proteins and or GRKs. In order to assess whether the close proximity of MOP and NOP is due simply to high expression, we extended our study to include co-expression in native tissue and provide significant evidence that interaction of the type seen in recombinants also occurs in a native tissue; mouse hippocampal neurites. Secondly, signalling differences seen in the co-expression system such as reduction in MOP agonist potency may be due to either competition at downstream signalling pathways or, due to the decrease in receptor expression of MOP in the co-expression system when compared to single expression system, a lack of receptor reserves (this could potentially be probed for MOP with β-funaltrexamine) and therefore decreased potency. We believe this is not the case, as both De101 and dermorphin co-incubated with UFP-101 demonstrate similar or higher potency to the single expression MOP system. If receptor competition for downstream signalling pathways was occurring, this reversal would only be possible if UFP-101 was acting as an inverse agonist, for which we have no evidence.
What does this data mean for development of bivalent opioids and, by extension, bifunctional opioids? Work with cebranopadol and other mixed opioids is now maturing and, in general, these produce good analgesia with reduced side effects. MOP-morphine like molecules are currently seen as the 'enemy'; driven by the opioid epidemic. That said MOP analgesia and analgesics used in the right setting is/are good [59]. Attempts to design out the troublesome side effect profile have been met with variable success so a different approach with mixed ligands is worthy of consideration. A mixed ligand (bivalent or bifunctional) can potentially reduce the 'amount' of MOP activation by adding in NOP (or other targets) with the NOP component producing analgesia in its own right along with reducing the adverse effects of MOP. There is an excellent study in non-human primates from Ko and Naughton [24] showing that largely ineffective doses of i.t morphine and N/OFQ synergise to produce good quality antinociception. Add into the mix partial agonists/biased agonists (for example oliceridine) and allosteric modulators and the potential for analgesic design increases [60,61].