Cover Page

Cover Page

The handle http://hdl.handle.net/1887/62061 holds various files of this Leiden University dissertation

Author: Soethoudt, Marjolein

Title: Chemical tools to study the cannabinoid receptor type 2

Date: 2018-04-26

Chapter 4

Cannabinoid Receptor Ligand Profiling Reveals Biased Signaling and Off-target Activity

Published in Nature Communications, 2017, 8, 13958

60

Chapter 4

4.1 Introduction

Target validation is an essential element of pharmacological research and drug discovery (Chapter 1).

Pharmacological intervention using chemical probes provides a powerful means to assess the temporal consequences of acute modulation of protein function under both physiological and pathological conditions.

High selectivity and a well-defined molecular mode of action of chemical probes are essential to translate the preclinical studies on non-human species to the patient. However, this type of information is often lacking and reproducibility across different laboratories is sometimes difficult to obtain.

There is a great interest in the development of selective type-2 cannabinoid receptor (CB

R) agonists as potential drug candidates for various pathophysiological conditions,

which include chronic and inflammatory pain,

pruritus,

diabetic neuropathy and nephropathy,

liver cirrhosis,

and protective effects after ischemic-reperfusion injury.

CB

R belongs to the cannabinoid receptor family of G protein-coupled receptors, which also includes type-1 cannabinoid receptor (CB

R). Both CBRs are the biological target of Δ

- tetrahydrocannabinol (Δ

-THC), the main psychoactive component in cannabis.

CB

R and CB

R share an overall sequence homology of 44%, but the 7-transmembrane spanning region, which contains the ligand-binding domain, exhibits 68% similarity.

CB

R is predominantly expressed on immune cells and its expression level is believed to increase in tissues upon pathological stimuli,

whereas the CB

R is highly expressed in the brain.

Both receptors are coupled to G

proteins and modulate various intracellular signal transduction pathways, such as inhibition of cAMP-production, activation of pERK and G protein-coupled inwardly rectifying K

-channels (GIRKs), and recruitment of β-arrestin to the receptor.

It is currently unknown which signal transduction pathways (or combinations thereof) are relevant for therapeutic purposes. In addition, some compounds may act as biased and/or protean agonists,

and remarkable differences between rodent and human receptor orthologues have been noted, which are complicating the translation of results from preclinical animal models to human trials.

Different chemical classes have been described as CBR ligands (e.g., mixed CBR agonists: Δ

-THC (henceforth referred to as THC), CP55940, WIN55212-2, HU210, and the endogenous ligands 2-arachidonoyl glycerol (2-AG) and anandamide (AEA, N- arachidonoylethanolamine); CB

R antagonists: SR141716A (rimonabant), and AM251; CB

R agonists: HU308, HU910, Gp-1a, JWH015, JWH133 and AM1241; and CB

R antagonists:

AM630 and SR144528; see Figure 1 for structures).

These ligands are used to explore CBR biology and to obtain preclinical target validation of the CBR subtypes.

The high homology between the ligand binding domains of the two receptors and the overall higher tissue expression of CB

R pose challenges to develop selective ligands that target only CB

R. Yet, high selectivity is required to determine the exact role of each receptor in various (patho)physiological processes and to avoid CB

R-mediated (psychotropic) side effects caused by THC and other CB

R ligands.

Cannabinoid Receptor Ligand Profiling Reveals Biased Signaling and Off-target Activity

The need for highly selective CB

ligands is exemplified by the scientific dispute whether the CB

R plays an important role in normal brain function or not. This whole avenue of research is currently being hampered by possible bias of using non-selective pharmacological and immunological tools and has delayed the development of novel CB

R-based drugs.

Currently, most ligands are only characterized in a binding assay and/or in a limited set of functional assays using recombinant human receptors. The results are scattered among various publications and are derived from different experimental settings, which may have led to apparent contradictory results.

Conflicting results from in vivo models that employ some of the above mentioned ligands have also been described in the literature.

Often, information about potential off-targets and pharmacokinetics of ligands is also lacking.

This has complicated the comparison and interpretation of the data and led to confusion about which are the preferred ligands to be used for in vivo experiments aimed at validating the CB

receptor as a therapeutic target. Unfortunately, this situation, which has resulted in a loss of resources and unnecessary use of animals, is not unique to the CB

receptor field.

The US National Institutes of Health (NIH) shares these concerns from many scientists about the reproducibility issues in biomedical research and required action to counter this problem.

To improve target validation and to guide the selection of the best ligand for preclinical studies, a fully detailed profile of the current “gold standard” ligands is needed.

To provide important guidance for the field and to address potential species- dependent differences, the most widely used CB

R ligands were comprehensively profiled.

Receptor binding of both human and mouse CB

R was investigated in several independent

academic and industry laboratories, as well as multiple signal transduction pathways

(GTPγS, cAMP, β-AR, pERK and GIRK). Selectivity of the ligands was determined towards a

customized panel of proteins associated with cannabinoid ligand pharmacology, which

includes the CB

R and the major proteins of the endocannabinoid system (ECS): N-Acyl

ethanolamines biosynthesizing enzyme N-acyl phosphatidylethanolamine-specific

phospholipase D (NAPE-PLD) and AEA hydrolyzing enzyme fatty acid amide hydrolase

(FAAH); 2-AG bio-synthesizing enzyme diacylglycerol lipase (DAGL) and -hydrolyzing

enzymes monoacylglycerol lipase (MAGL) and α/β-hydrolase domain 6 and 12 (ABHD6 and

ABHD12), as well as towards the putative endocannabinoid transporters; AEA and 2-AG-

binding transient receptor potential channels (TRP-channels, TRPV1-4, TRPM8, and

TRPA1).

In addition, off-target activity on GPR55, a receptor that binds CBR-type ligands,

and on COX-2, which oxygenates AEA and 2-AG, was also determined. Determination of the

selectivity of CB

R ligands over these other proteins and processes involved in the

endocannabinoid system, as well as over the TRP channels (which are involved in similar

biological processes as the CBRs) is essential for the development of selective CB

R ligands

and to avoid complications in the interpretation of the in vivo results obtained with these

compounds.

61

Chap ter 4

Chapter 4

4.1 Introduction

Target validation is an essential element of pharmacological research and drug discovery (Chapter 1).

Pharmacological intervention using chemical probes provides a powerful means to assess the temporal consequences of acute modulation of protein function under both physiological and pathological conditions.

High selectivity and a well-defined molecular mode of action of chemical probes are essential to translate the preclinical studies on non-human species to the patient. However, this type of information is often lacking and reproducibility across different laboratories is sometimes difficult to obtain.

There is a great interest in the development of selective type-2 cannabinoid receptor (CB

R) agonists as potential drug candidates for various pathophysiological conditions,

which include chronic and inflammatory pain,

pruritus,

diabetic neuropathy and nephropathy,

liver cirrhosis,

and protective effects after ischemic-reperfusion injury.

CB

R belongs to the cannabinoid receptor family of G protein-coupled receptors, which also includes type-1 cannabinoid receptor (CB

R). Both CBRs are the biological target of Δ

- tetrahydrocannabinol (Δ

-THC), the main psychoactive component in cannabis.

CB

R and CB

R share an overall sequence homology of 44%, but the 7-transmembrane spanning region, which contains the ligand-binding domain, exhibits 68% similarity.

CB

R is predominantly expressed on immune cells and its expression level is believed to increase in tissues upon pathological stimuli,

whereas the CB

R is highly expressed in the brain.

Both receptors are coupled to G

proteins and modulate various intracellular signal transduction pathways, such as inhibition of cAMP-production, activation of pERK and G protein-coupled inwardly rectifying K

-channels (GIRKs), and recruitment of β-arrestin to the receptor.

It is currently unknown which signal transduction pathways (or combinations thereof) are relevant for therapeutic purposes. In addition, some compounds may act as biased and/or protean agonists,

and remarkable differences between rodent and human receptor orthologues have been noted, which are complicating the translation of results from preclinical animal models to human trials.

Different chemical classes have been described as CBR ligands (e.g., mixed CBR agonists: Δ

-THC (henceforth referred to as THC), CP55940, WIN55212-2, HU210, and the endogenous ligands 2-arachidonoyl glycerol (2-AG) and anandamide (AEA, N- arachidonoylethanolamine); CB

R antagonists: SR141716A (rimonabant), and AM251; CB

R agonists: HU308, HU910, Gp-1a, JWH015, JWH133 and AM1241; and CB

R antagonists:

AM630 and SR144528; see Figure 1 for structures).

These ligands are used to explore CBR biology and to obtain preclinical target validation of the CBR subtypes.

The high homology between the ligand binding domains of the two receptors and the overall higher tissue expression of CB

R pose challenges to develop selective ligands that target only CB

R. Yet, high selectivity is required to determine the exact role of each receptor in various (patho)physiological processes and to avoid CB

R-mediated (psychotropic) side effects caused by THC and other CB

R ligands.

Cannabinoid Receptor Ligand Profiling Reveals Biased Signaling and Off-target Activity

The need for highly selective CB

ligands is exemplified by the scientific dispute whether the CB

R plays an important role in normal brain function or not. This whole avenue of research is currently being hampered by possible bias of using non-selective pharmacological and immunological tools and has delayed the development of novel CB

R-based drugs.

Currently, most ligands are only characterized in a binding assay and/or in a limited set of functional assays using recombinant human receptors. The results are scattered among various publications and are derived from different experimental settings, which may have led to apparent contradictory results.

Conflicting results from in vivo models that employ some of the above mentioned ligands have also been described in the literature.

Often, information about potential off-targets and pharmacokinetics of ligands is also lacking.

This has complicated the comparison and interpretation of the data and led to confusion about which are the preferred ligands to be used for in vivo experiments aimed at validating the CB

receptor as a therapeutic target. Unfortunately, this situation, which has resulted in a loss of resources and unnecessary use of animals, is not unique to the CB

receptor field.

The US National Institutes of Health (NIH) shares these concerns from many scientists about the reproducibility issues in biomedical research and required action to counter this problem.

To improve target validation and to guide the selection of the best ligand for preclinical studies, a fully detailed profile of the current “gold standard” ligands is needed.

To provide important guidance for the field and to address potential species- dependent differences, the most widely used CB

R ligands were comprehensively profiled.

Receptor binding of both human and mouse CB

R was investigated in several independent

academic and industry laboratories, as well as multiple signal transduction pathways

(GTPγS, cAMP, β-AR, pERK and GIRK). Selectivity of the ligands was determined towards a

customized panel of proteins associated with cannabinoid ligand pharmacology, which

includes the CB

R and the major proteins of the endocannabinoid system (ECS): N-Acyl

ethanolamines biosynthesizing enzyme N-acyl phosphatidylethanolamine-specific

phospholipase D (NAPE-PLD) and AEA hydrolyzing enzyme fatty acid amide hydrolase

(FAAH); 2-AG bio-synthesizing enzyme diacylglycerol lipase (DAGL) and -hydrolyzing

enzymes monoacylglycerol lipase (MAGL) and α/β-hydrolase domain 6 and 12 (ABHD6 and

ABHD12), as well as towards the putative endocannabinoid transporters; AEA and 2-AG-

binding transient receptor potential channels (TRP-channels, TRPV1-4, TRPM8, and

TRPA1).

In addition, off-target activity on GPR55, a receptor that binds CBR-type ligands,

and on COX-2, which oxygenates AEA and 2-AG, was also determined. Determination of the

selectivity of CB

R ligands over these other proteins and processes involved in the

endocannabinoid system, as well as over the TRP channels (which are involved in similar

biological processes as the CBRs) is essential for the development of selective CB

R ligands

and to avoid complications in the interpretation of the in vivo results obtained with these

compounds.

62

Chapter 4

To assess which ligands are best suited for in vivo studies, all 18 compounds were profiled for their physico-chemical properties, in vitro ADME (absorption, distribution, metabolism, elimination) and pharmacokinetic parameters and cross-reactivity in the CEREP panel of 64 common off-targets. Commonly used non-selective ligands, including Δ9-THC and the endocannabinoids 2-AG and anandamide were also tested in vitro.

All ligands were high quality grade material, provided to each laboratory by the industry collaborator. The top three candidate CB2R agonists were further investigated at high doses in vivo to infer potential interactions with CNS CB1R. All data together results in the largest dataset generated so far under the same experimental conditions for all cannabinoid receptor ligands, leading to a consensus that HU910, HU308 and JWH133 possess the best CB2R agonist profiles among the ligands tested on the basis of selectivity, balanced signaling, pharmacokinetic profile and off-target activity, and may be considered

“golden standards” for CB2R validation studies in mice.

Figure 1. Structures of references ligands divided per class as described in the introduction

4.2 Results

4.2.1 Physico-chemical properties

The physico-chemical properties of the 18 compounds tested are listed in Table 1.

Cannabinoid Receptor Ligand Profiling Reveals Biased Signaling and Off-target Activity

Molecular weights span a range from 312 g/mol for JWH133 up to 555 g/mol for AM251 and the polar surface area values are overall very low (8 Å for JWH133 up to 63 Å for (S)- AM1241), due to a low number of heteroatoms present in the ligands. Importantly, all CBR ligands are very lipophilic molecules, which negatively affects their solubility, ADME- properties and off-target profile. Even the lowest lipophilicity value (clogP), calculated to be 4.9 for WIN55212-2, is relatively high. The most lipophilic CBR ligand is SR144528, which exhibits an extremely high clogP value of 9.2. Consequently, only CP55940 and (rac)- AM1241 were soluble in an aqueous phosphate buffer system (pH 6.5). Despite the fact that the membrane permeation coefficient (PAMPA) P

is low for several of the molecules, most compounds may be able to cross biological barriers as high percentages of the substances were found in membranes.

Table 1. Physicochemical properties of ligands

MW (g/mol) Polar surface areaa (Å) Kow clogPb logDc, j PAMPA Peffd, j [cm/s*10-6] Pct acceptore, j[%] Pct donorf, j[%] Pct membraneg, j[%] Kinetic solubilityh, k [µg/mL] Kinetic solubilityh, k [µmol/mL] Solubility in 15% DMSO, 85% PEG400i, l [µg/mL] Solubility in 15% DMSO, 85% PEG400i, l [µmol/mL]

WIN55212-2 523 36.28 4.685 3.67±0.03 0.25±0.22 0.5 50.6 49 <0.3 <0.0006 1.68 0.0032

CP55940 377 49.35 7.498 ND 0 0 53.2 46.7 1.4±0 0.037 ND ND

Δ⁹-THC 314 38.06 7.986 ND ND ND ND ND ND ND ND ND

HU-210 387 38.06 7.986 precipitation 0.22 0 23 76 <0.2 <0.0005 ND ND SR141716A 464 42.1 6.085 >3.335 0.27±0.11 0.3 25.2 74.3 <0.733333 <0.0016 1.23 0.0027 AM251 555 42.15 6.608 ND 0 0 10.3 89.3 <0.3 <0.0005 1.57 0.0028 JWH015 327 15.51 6.462 ND ND ND ND ND <0.2 <0.0006 1.2 0.0037 JWH133 312 7.91 8.458 >3

precipitation 1.86 2.3 36.1 61.7 <0.4 <0.0013 1.88 0.006 HU308 415 27.44 8.965 4.29 2.53 2.8 23.1 74.3 <0.75 <0.0018 2.06 0.005 Gp-1a 441 43.1 5.931 ND 0 0 29.7 70.3 <0.2 <0.0005 2.43 0.0055 HU910 415 31.35 9.001 out of range 0.45 1 42 57 <0.5 0.0012 ND ND (rac)-

AM1241 503 62.47 5.726 3.66±0.03 1.86±0.62 3.3. 41.3 55.7 25±2 0.0497±0.004 ND ND (R)-AM1241 503 58.26 5.726 3.66 1.22 2.1 38.3 59.7 12 0.0238 ND ND (S)-AM1241 503 62.65 5.726 3.64 1.41 2.2 34 64 6 0.0119 2.37 0.0047 AM630 504 39.77 4.862 3.82 0.2 0.37 35.4 64.3 <0.3 <0.0006 1.54 0.0031 SR144528 476 36.11 9.154 >3 0 0 47 54 <0.2 <0.0004 1.64 0.0034 Anandamide 348 41.91 6.31 out of range 0.26 1.2 96.3 2.7 <1.1 <0.0032 ND ND

2-AG 379 51.32 6.738 ND ND ND ND ND <0.5 <0.0013 ND ND

aSurface sum of all polar atoms in the molecule; ^bCalculated partition coefficient values (cLogP) from experimentally determined octanol- water partition coefficient values (Kow); ^cDistribution coefficient values; ^dParallel artificial membrane permeability assay (PAMPA) was used to determine membrane permeation coefficient values (Peff); ^ePercentage of molecule that is able to act as a hydrogen bond acceptor;

fPercentage of molecule that is able to act as a hydrogen bond donor; ^gPercentage of compounds found in membranes; ^hSolubility of the compound when diluted into aqueous environment from DMSO superstock; ⁱSolubility of the compound in the formulation used for in vivo administration of the ligands; ^jMean ± SD of three independent experiments; ^kMean of two independent experiments; ^lSingle experiment

4.2.2 Affinity and selectivity in CBR binding studies

To determine the affinity and selectivity of the 18 substances, [

H]-CP55940 displacement

assays using membrane fractions of Chinese Hamster Ovary (CHO) cells overexpressing

recombinant human CB

R and CB

R were performed in two independent laboratories. In

addition, mouse brain and spleen were used as source of mouse CB

R and CB

R,

respectively.

63

Chap ter 4

Chapter 4

To assess which ligands are best suited for in vivo studies, all 18 compounds were profiled for their physico-chemical properties, in vitro ADME (absorption, distribution, metabolism, elimination) and pharmacokinetic parameters and cross-reactivity in the CEREP panel of 64 common off-targets. Commonly used non-selective ligands, including Δ9-THC and the endocannabinoids 2-AG and anandamide were also tested in vitro.

All ligands were high quality grade material, provided to each laboratory by the industry collaborator. The top three candidate CB2R agonists were further investigated at high doses in vivo to infer potential interactions with CNS CB1R. All data together results in the largest dataset generated so far under the same experimental conditions for all cannabinoid receptor ligands, leading to a consensus that HU910, HU308 and JWH133 possess the best CB2R agonist profiles among the ligands tested on the basis of selectivity, balanced signaling, pharmacokinetic profile and off-target activity, and may be considered

“golden standards” for CB2R validation studies in mice.

Figure 1. Structures of references ligands divided per class as described in the introduction

4.2 Results

4.2.1 Physico-chemical properties

The physico-chemical properties of the 18 compounds tested are listed in Table 1.

Cannabinoid Receptor Ligand Profiling Reveals Biased Signaling and Off-target Activity

Molecular weights span a range from 312 g/mol for JWH133 up to 555 g/mol for AM251 and the polar surface area values are overall very low (8 Å for JWH133 up to 63 Å for (S)- AM1241), due to a low number of heteroatoms present in the ligands. Importantly, all CBR ligands are very lipophilic molecules, which negatively affects their solubility, ADME- properties and off-target profile. Even the lowest lipophilicity value (clogP), calculated to be 4.9 for WIN55212-2, is relatively high. The most lipophilic CBR ligand is SR144528, which exhibits an extremely high clogP value of 9.2. Consequently, only CP55940 and (rac)- AM1241 were soluble in an aqueous phosphate buffer system (pH 6.5). Despite the fact that the membrane permeation coefficient (PAMPA) P

is low for several of the molecules, most compounds may be able to cross biological barriers as high percentages of the substances were found in membranes.

Table 1. Physicochemical properties of ligands

MW (g/mol) Polar surface areaa (Å) Kow clogPb logDc, j PAMPA Peffd, j [cm/s*10-6] Pct acceptore, j[%] Pct donorf, j[%] Pct membraneg, j[%] Kinetic solubilityh, k [µg/mL] Kinetic solubilityh, k [µmol/mL] Solubility in 15% DMSO, 85% PEG400i, l [µg/mL] Solubility in 15% DMSO, 85% PEG400i, l [µmol/mL]

WIN55212-2 523 36.28 4.685 3.67±0.03 0.25±0.22 0.5 50.6 49 <0.3 <0.0006 1.68 0.0032

CP55940 377 49.35 7.498 ND 0 0 53.2 46.7 1.4±0 0.037 ND ND

Δ⁹-THC 314 38.06 7.986 ND ND ND ND ND ND ND ND ND

HU-210 387 38.06 7.986 precipitation 0.22 0 23 76 <0.2 <0.0005 ND ND SR141716A 464 42.1 6.085 >3.335 0.27±0.11 0.3 25.2 74.3 <0.733333 <0.0016 1.23 0.0027 AM251 555 42.15 6.608 ND 0 0 10.3 89.3 <0.3 <0.0005 1.57 0.0028 JWH015 327 15.51 6.462 ND ND ND ND ND <0.2 <0.0006 1.2 0.0037 JWH133 312 7.91 8.458 >3

precipitation 1.86 2.3 36.1 61.7 <0.4 <0.0013 1.88 0.006 HU308 415 27.44 8.965 4.29 2.53 2.8 23.1 74.3 <0.75 <0.0018 2.06 0.005 Gp-1a 441 43.1 5.931 ND 0 0 29.7 70.3 <0.2 <0.0005 2.43 0.0055 HU910 415 31.35 9.001 out of range 0.45 1 42 57 <0.5 0.0012 ND ND (rac)-

AM1241 503 62.47 5.726 3.66±0.03 1.86±0.62 3.3. 41.3 55.7 25±2 0.0497±0.004 ND ND (R)-AM1241 503 58.26 5.726 3.66 1.22 2.1 38.3 59.7 12 0.0238 ND ND (S)-AM1241 503 62.65 5.726 3.64 1.41 2.2 34 64 6 0.0119 2.37 0.0047 AM630 504 39.77 4.862 3.82 0.2 0.37 35.4 64.3 <0.3 <0.0006 1.54 0.0031 SR144528 476 36.11 9.154 >3 0 0 47 54 <0.2 <0.0004 1.64 0.0034 Anandamide 348 41.91 6.31 out of range 0.26 1.2 96.3 2.7 <1.1 <0.0032 ND ND

2-AG 379 51.32 6.738 ND ND ND ND ND <0.5 <0.0013 ND ND

aSurface sum of all polar atoms in the molecule; ^bCalculated partition coefficient values (cLogP) from experimentally determined octanol- water partition coefficient values (Kow); ^cDistribution coefficient values; ^dParallel artificial membrane permeability assay (PAMPA) was used to determine membrane permeation coefficient values (Peff); ^ePercentage of molecule that is able to act as a hydrogen bond acceptor;

fPercentage of molecule that is able to act as a hydrogen bond donor; ^gPercentage of compounds found in membranes; ^hSolubility of the compound when diluted into aqueous environment from DMSO superstock; ⁱSolubility of the compound in the formulation used for in vivo administration of the ligands; ^jMean ± SD of three independent experiments; ^kMean of two independent experiments; ^lSingle experiment

4.2.2 Affinity and selectivity in CBR binding studies

To determine the affinity and selectivity of the 18 substances, [

H]-CP55940 displacement

assays using membrane fractions of Chinese Hamster Ovary (CHO) cells overexpressing

recombinant human CB

R and CB

R were performed in two independent laboratories. In

addition, mouse brain and spleen were used as source of mouse CB

R and CB

R,

respectively.

64

Chapter 4

Using the Pearson correlation analysis, a statistically significant correlation was found between the binding affinities between the different labs (Pearson coefficient: 0.9304 (hCB

R), 0.6648 (hCB

R) and 0.7720 (mCB

R), see Figure 2).

Figure 2. Correlation of binding affinities between labs. The affinities of THC and HU210 could not be compared since these were not measured in all labs due to legal restrictions. A-B) Correlation of binding affinities of reference compounds separately determined by the labs of Roche and Leiden on hCB1R (A) and hCB2R (B). C) Correlation of binding affinities of reference compounds separately determined by the labs of Roche and Mauro Maccarronne on mCB2R. A-C) Statistics performed was two-tailed Pearson correlation analysis. All pKi values presented here are displayed as mean ± SEM of three independent experiments performed in duplicate (N=3, n=2).

Figure 3 depicts the selectivity of the ligands for the CB2

R versus CB

R. HU210 > CP55940, WIN55212-2 >

-THC were found as the highest affinity non-selective human CBR ligands.

Figure 3. CB2R selectivity of cannabinoid reference ligands on mouse and human CBRs. CB2R selectivity for all cannabinoid receptor reference ligands are presented as the difference in mean pKi values between CB2R and CB1R for both the human (black bars, Leiden data) and mouse (white bars, Maccarrone data) orthologues.

From left to right: ligands with decreasing CB2R selectivity (from HU308 to Gp-1a), nonselective ligands (from HU210 to CP55940), ligands with CB1R selectivity (AM251 and SR141716A).

Cannabinoid Receptor Ligand Profiling Reveals Biased Signaling and Off-target Activity Conversely, HU308, HU910 and JWH133 were the most selective human CB

R ligands (Table

R than for CB

R.

Notably, JWH015 and Gp-1a were less than 30-fold selective for CB

R. Importantly, the binding selectivity of the ligands for mouse CB

R over mouse CB

R appeared to be greatly reduced for all ligands (<100 fold), except AM630 and SR144528, which are actually more selective on mCB

R than on hCB

R. The most selective agonists on mCB

R were (rac)- AM1241 (66-fold), JWH133 (40-fold) and Gp-1a (20-fold). As expected, AEA and 2-AG, the endogenous ligands of CB

R and CB

R, were non-selective and showed moderate binding affinities towards both receptors (pKi ~7).

Table 2. Binding affinity and selectivity of reference ligands on human and mouse cannabinoid receptors.

hCB2R hCB1R CB2R Selectivity* mCB2R mCB1R CB2R Selectivity*

WIN55212-2 8.57 ± 0.16 8.72 ± 0.24 1 7.28 ± 0.17 7.81 ± 0.15 0.3

CP55940 8.44 ± 0.18 9.26 ± 0.12 0.2 9.22 ± 0.26 8.8 ± 0.03 3

Δ⁹-THC 8.16 ± 0.17 8.48 ± 0.08 0.5 ND ND

HU210 9.78 ± 0.04^a 9.55 ± 0.06^a 2 9.27 ± 0.38 8.96 ± 0.27 2

SR141716A 5.60 ± 0.61 9.19 ± 0.10 0.0003 6.95 ± 0.2 8.35 ± 0.06 0.04

AM251 6.92 ± 0.24 9.58 ± 0.21 0.002 5.34 ± 0.8 8.62 ± 0.07 0.001

JWH015 7.92 ± 0.26 6.47 ± 0.09 28 6.63 ± 0.21 5.94 ± 0.15 5

JWH133 7.18 ± 0.34 <5 153 7.69 ± 0.23 6.09 ± 0.13 40

HU308 7.44 ± 0.12 <5 278 7.15 ± 0.21 6.08 ± 0.14 12

Gp-1a 7.66 ± 0.11 6.97 ± 0.35 5 7.68 ± 0.23 6.37 ± 0.08 20

HU910 7.22 ± 0.31 <5 166 6.88 ± 0.17 6.14 ± 0.13 6

(rac)-AM1241 8.39 ± 0.10 6.89 ± 0.31 32 7.73 ± 0.25 5.91 ± 0.16 66

(R)-AM1241 8.44 ± 0.25 6.86 ± 0.46 38 7.59 ± 0.11 6.36 ± 0.13 17

(S)-AM1241 7.02 ± 0.15 <5 105 6.74 ± 0.2 6.48 ± 0.08 2

AM630 7.39 ± 0.02 6.51 ± 0.25 8 7.66 ± 0.23 5.6 ± 0.298 115

SR144528 7.88 ± 0.06^a 5.77 ± 0.09^a 129 10.7 ± 0.16 6.92 ± 0.2 6026

Anandamide 6.91 ± 0.28 7.04 ± 0.28# 1 6.46 ± 0.18 7.13 ± 0.06 0.2

2-AG 6.94 ± 0.43 7.15 ± 0.47# 1 7.53 ± 0.22 6.99 ± 0.29 3

pKi values are presented as the mean ± SEM (N=3, n=2), unless stated otherwise; Data sets from Leiden (human) and Maccarrone (mouse), unless stated otherwise; *CB2R selectivity was calculated as follows: 10^(pKi CB2R-pKi CB1R); ^#Mean ± SEM of 4 independent experiments performed in duplicate; ^a Dataset from Roche

4.2.3 Functional activity and selectivity of CBR signaling pathways

To determine the functional activity and selectivity (towards CB

R over CB

R) of the ligands five different assays (GTPγS, cAMP, β-AR, pERK and GIRK) were performed on both human CB

R and human CB

R (Tables 3-7). All ligands were tested on cAMP signaling on both mouse CBRs and HU910, HU308 and JWH133 were tested on G protein activation and β- Arrestin recruitment on mCB

R, to determine interspecies behaviour of the ligands. Efficacy of the ligands is normalized to the effect produced by CP55940 (10 µM) in all assays.

However, it should be noted that efficacy is relative by definition, and is dependent on the

reference ligand used as well as the assay conditions. For both human and mouse CB

R, the

potency of the ligands correlated with their binding affinity in most assays, except for β-AR

and GIRK signaling (Figure 4).

65

Chap ter 4

Chapter 4

Using the Pearson correlation analysis, a statistically significant correlation was found between the binding affinities between the different labs (Pearson coefficient: 0.9304 (hCB

R), 0.6648 (hCB

R) and 0.7720 (mCB

R), see Figure 2).

Figure 2. Correlation of binding affinities between labs. The affinities of THC and HU210 could not be compared since these were not measured in all labs due to legal restrictions. A-B) Correlation of binding affinities of reference compounds separately determined by the labs of Roche and Leiden on hCB1R (A) and hCB2R (B). C) Correlation of binding affinities of reference compounds separately determined by the labs of Roche and Mauro Maccarronne on mCB2R. A-C) Statistics performed was two-tailed Pearson correlation analysis. All pKi values presented here are displayed as mean ± SEM of three independent experiments performed in duplicate (N=3, n=2).

Figure 3 depicts the selectivity of the ligands for the CB2

R versus CB

R. HU210 > CP55940, WIN55212-2 >

-THC were found as the highest affinity non-selective human CBR ligands.

Figure 3. CB2R selectivity of cannabinoid reference ligands on mouse and human CBRs. CB2R selectivity for all cannabinoid receptor reference ligands are presented as the difference in mean pKi values between CB2R and CB1R for both the human (black bars, Leiden data) and mouse (white bars, Maccarrone data) orthologues.

From left to right: ligands with decreasing CB2R selectivity (from HU308 to Gp-1a), nonselective ligands (from HU210 to CP55940), ligands with CB1R selectivity (AM251 and SR141716A).

Cannabinoid Receptor Ligand Profiling Reveals Biased Signaling and Off-target Activity Conversely, HU308, HU910 and JWH133 were the most selective human CB

R ligands (Table

R than for CB

R.

Notably, JWH015 and Gp-1a were less than 30-fold selective for CB

R. Importantly, the binding selectivity of the ligands for mouse CB

R over mouse CB

R appeared to be greatly reduced for all ligands (<100 fold), except AM630 and SR144528, which are actually more selective on mCB

R than on hCB

R. The most selective agonists on mCB

R were (rac)- AM1241 (66-fold), JWH133 (40-fold) and Gp-1a (20-fold). As expected, AEA and 2-AG, the endogenous ligands of CB

R and CB

R, were non-selective and showed moderate binding affinities towards both receptors (pKi ~7).

Table 2. Binding affinity and selectivity of reference ligands on human and mouse cannabinoid receptors.

hCB2R hCB1R CB2R Selectivity* mCB2R mCB1R CB2R Selectivity*

WIN55212-2 8.57 ± 0.16 8.72 ± 0.24 1 7.28 ± 0.17 7.81 ± 0.15 0.3

CP55940 8.44 ± 0.18 9.26 ± 0.12 0.2 9.22 ± 0.26 8.8 ± 0.03 3

Δ⁹-THC 8.16 ± 0.17 8.48 ± 0.08 0.5 ND ND

HU210 9.78 ± 0.04^a 9.55 ± 0.06^a 2 9.27 ± 0.38 8.96 ± 0.27 2

SR141716A 5.60 ± 0.61 9.19 ± 0.10 0.0003 6.95 ± 0.2 8.35 ± 0.06 0.04

AM251 6.92 ± 0.24 9.58 ± 0.21 0.002 5.34 ± 0.8 8.62 ± 0.07 0.001

JWH015 7.92 ± 0.26 6.47 ± 0.09 28 6.63 ± 0.21 5.94 ± 0.15 5

JWH133 7.18 ± 0.34 <5 153 7.69 ± 0.23 6.09 ± 0.13 40

HU308 7.44 ± 0.12 <5 278 7.15 ± 0.21 6.08 ± 0.14 12

Gp-1a 7.66 ± 0.11 6.97 ± 0.35 5 7.68 ± 0.23 6.37 ± 0.08 20

HU910 7.22 ± 0.31 <5 166 6.88 ± 0.17 6.14 ± 0.13 6

(rac)-AM1241 8.39 ± 0.10 6.89 ± 0.31 32 7.73 ± 0.25 5.91 ± 0.16 66

(R)-AM1241 8.44 ± 0.25 6.86 ± 0.46 38 7.59 ± 0.11 6.36 ± 0.13 17

(S)-AM1241 7.02 ± 0.15 <5 105 6.74 ± 0.2 6.48 ± 0.08 2

AM630 7.39 ± 0.02 6.51 ± 0.25 8 7.66 ± 0.23 5.6 ± 0.298 115

SR144528 7.88 ± 0.06^a 5.77 ± 0.09^a 129 10.7 ± 0.16 6.92 ± 0.2 6026

Anandamide 6.91 ± 0.28 7.04 ± 0.28# 1 6.46 ± 0.18 7.13 ± 0.06 0.2

2-AG 6.94 ± 0.43 7.15 ± 0.47# 1 7.53 ± 0.22 6.99 ± 0.29 3

pKi values are presented as the mean ± SEM (N=3, n=2), unless stated otherwise; Data sets from Leiden (human) and Maccarrone (mouse), unless stated otherwise; *CB2R selectivity was calculated as follows: 10^(pKi CB2R-pKi CB1R); ^#Mean ± SEM of 4 independent experiments performed in duplicate; ^a Dataset from Roche

4.2.3 Functional activity and selectivity of CBR signaling pathways

To determine the functional activity and selectivity (towards CB

R over CB

R) of the ligands five different assays (GTPγS, cAMP, β-AR, pERK and GIRK) were performed on both human CB

R and human CB

R (Tables 3-7). All ligands were tested on cAMP signaling on both mouse CBRs and HU910, HU308 and JWH133 were tested on G protein activation and β- Arrestin recruitment on mCB

R, to determine interspecies behaviour of the ligands. Efficacy of the ligands is normalized to the effect produced by CP55940 (10 µM) in all assays.

However, it should be noted that efficacy is relative by definition, and is dependent on the

reference ligand used as well as the assay conditions. For both human and mouse CB

R, the

potency of the ligands correlated with their binding affinity in most assays, except for β-AR

and GIRK signaling (Figure 4).

66

Chapter 4

Figure 4. Correlation between binding affinity and functional potency. (A-E) Correlation between hCB2R affinity (Leiden dataset) of the reference compounds with their functional potencies in the different assays. F) Correlation between mCB2R affinity of the reference compounds with their functional potencies in the cAMP assay. Maccarrone dataset (P-value 0.0285, Pearson coefficient 0.5305); Roche dataset (P-value 0.3049, Pearson coefficient 0.2645. After exclusion of compounds (Rac)-AM1241 and (R)-AM1241, which seem to be inactive in their mouse cAMP assay: P-value 0.0069, Pearson coefficient 0.6647). Statistics performed was two- tailed Pearson correlation analysis. All values are presented as mean ± SEM (N=3, n=2).

Graphs showing the pEC

values of the reference ligands on CB

R in all assays are shown in

R in the GTPγS assay (Table 3), while WIN55212-2 acted as a partial agonist. HU910 behaved as a partial CB

R agonist as well, but was, together with HU308, the most selective for CB

R in this assay (185- and 193-fold, respectively). Of note, JWH133 was considered functionally inactive on hCB

R, because its maximal effect was only 20% at 10 µM. On mCB

R, both HU308 and JWH133 were full agonists, but HU910 remained partially active. The potency of all three ligands was similar for human and mouse receptors. In contrast to previous reports,

Gp- 1a acted as an inverse agonist on CB

R, but was inactive at CB

R. Both THC and the endocannabinoids AEA and 2-AG acted as partial agonists on both receptors with similar potency.

Figure 5. CB2R potencies of cannabinoid receptor reference ligands across different assays. A) ‘Mixed’ CBR agonists, B) selective CB2R agonists, C) antagonists, D) endocannabinoids.

67

Chap ter 4

Chapter 4

Figure 4. Correlation between binding affinity and functional potency. (A-E) Correlation between hCB2R affinity (Leiden dataset) of the reference compounds with their functional potencies in the different assays. F) Correlation between mCB2R affinity of the reference compounds with their functional potencies in the cAMP assay. Maccarrone dataset (P-value 0.0285, Pearson coefficient 0.5305); Roche dataset (P-value 0.3049, Pearson coefficient 0.2645. After exclusion of compounds (Rac)-AM1241 and (R)-AM1241, which seem to be inactive in their mouse cAMP assay: P-value 0.0069, Pearson coefficient 0.6647). Statistics performed was two- tailed Pearson correlation analysis. All values are presented as mean ± SEM (N=3, n=2).

Graphs showing the pEC

values of the reference ligands on CB

R in all assays are shown in

R in the GTPγS assay (Table 3), while WIN55212-2 acted as a partial agonist. HU910 behaved as a partial CB

R agonist as well, but was, together with HU308, the most selective for CB

R in this assay (185- and 193-fold, respectively). Of note, JWH133 was considered functionally inactive on hCB

R, because its maximal effect was only 20% at 10 µM. On mCB

R, both HU308 and JWH133 were full agonists, but HU910 remained partially active. The potency of all three ligands was similar for human and mouse receptors. In contrast to previous reports,

Gp- 1a acted as an inverse agonist on CB

R, but was inactive at CB

R. Both THC and the endocannabinoids AEA and 2-AG acted as partial agonists on both receptors with similar potency.

Figure 5. CB2R potencies of cannabinoid receptor reference ligands across different assays. A) ‘Mixed’ CBR agonists, B) selective CB2R agonists, C) antagonists, D) endocannabinoids.