GPCR oligomerization and “bivalent ligands”

Over a long period of time GPCRs were considered to act as monomeric entities in a 1:1:1 stoichiometry with the G-protein and the ligand.⁴³ However, experiments employing different techniques like cross-linking, immunoblotting and co-immunoprecipitation as well as FRET (fluorescence resonance energy transfer) and BRET (bioluminescence resonance energy transfer) investigations in living cells provided convincing evidence for GPCRs to form dimers or higher order oligomers.^{1, 44, 45} The existence of homodimers has been demonstrated for

R R*

AR AR*

ARG AR*G

R*G RG

receptor activation

agonist binding

G-protein coupling

R RG

ARG AR

R R*

AR*

AR AR*G

R*G A B

C

Introduction 7

several class A and C GPCRs including the dopamine D2 and D3 receptors,^{46, 47} the β₂-adrenoceptor,⁴⁸ the H₁Rs, H₂Rs and H₄Rs,^49-51 opioid receptors^52-54, the mGluRs^{55, 56} and CaRs⁵⁷. Besides homodimers also heterodimers like the δ/κ-opioid receptors,⁵³ the somatostatine SSTR_1B/dopamine D₂ receptors⁵⁸ and the GABA_B1/GABA_B2 receptors⁵⁹ have been identified. The GPCRs are believed to interact via their extracellular loops, trans-membrane helices and intracellular loops, forming covalent or non-covalent interactions.⁴⁴ Examples for covalently linked receptors are the mGluRs and CaRs providing dimers via disulfide bonds. Conversely, the GABA_B1R/GABA_B2R heterodimer forms non-covalent inter-actions through a C-terminal coiled-coil domain.⁹ Although, few is known about the physiological role of GPCR dimerization, several investigations suggest a crucial role in GPCR trafficking, folding, activation and internalization.⁴⁴ The GABA_B1R/GABA_B2R hetero-dimer is a demonstrative example: The presence of GABA_B2R is prerequisite for a proper transfer of GABA_B1R to the cell surface. Moreover, the GABA_B1R binds its agonist but does not couple to the G-protein, whereas the GABAB2R stimulates G-protein signaling but does not bind the ligand.^59-61 There is growing evidence that GPCR dimerization, in particular heterodimerization, results in complexes with modified ligand binding and signal transduction properties relative to the individual receptors.^{9, 54} Distinct characteristics arising from hetero-dimerization have been shown for the κ- and δ-opioid receptors⁵³, the μ- and δ-opioid receptors⁵⁴ or the angiotensin AT₁ and bradykinin B₂ receptors⁶². However, as many experiments have been performed in recombinant cell systems, the physiological relevance of GPCR oligomerization has to be further elucidated in ongoing studies.

The bivalent ligand approach in the design of ligands targeting GPCRs has proven to be promising to improve not only potency and selectivity but also the pharmacokinetic profile of compounds.⁶³ Usually bivalent ligands are characterized by a molecule containing two sets of pharmacophoric entities linked through a spacer. However, in the broader sense bivalent ligands can be divided in molecules comprising two sets of pharmacophoric groups or a single pharmacophore connected to a non-pharmacophoric recognition unit.^{64, 65} Two different binding modes for bivalent ligands at the receptor(s) are imaginable (Figure 1.4). If the spacer is of sufficient length the bivalent ligand may bridge two neighboring receptors, each pharmacophoric entitiy interacting with the binding site of one receptor. Molecular modeling studies with μ-opioid receptor dimers estimate a distance between the binding sites of approximately 27 Å.⁶⁵ For bivalent ligands with shorter linkers, next to the binding site an accessory recognition site at a single receptor is probable. The existence of accessory binding sites has been demonstrated for bivalent opioid receptor antagonists and explains increased affinities of bivalent ligands, that are not capable to link two receptors, relative to the monovalent counterparts.^63-66 According to Portoghese,⁶⁴ such bivalent ligands containing spacers of insufficient length for bridging two receptors fit to the “message-address” concept

proposed by Schwyzer⁶⁷. The pharmacophore can be considered as the “message” which is recognized by a family of receptors and, in case of agonists, is responsible for receptor activation, whereas the second pharmacophoric or non-pharmacophoric entity is considered as the “address” conferring additional affinity.

The increase in affinity of bivalent ligands relative to their monovalent analogs can be explained by the assumption that the bivalent ligands first bind in a univalent manner to the receptor (Figure 1.4). Thereby, the second recognition unit of the bivalent ligand achieves closer proximity to the second binding site (neighboring receptor or accessory binding site) corresponding to a high local concentration of the second recognition unit. This should afford an increase in affinity greater than expected from the sum of its two monovalent pharma-cophores.^{64, 65} The spacer length of the bivalent ligands plays a crucial role with respect to affinity as too short spacers prevent bridging of the binding sites, whereas too long linkers reduce the residence time of the recognition unit in vicinity to the binding site. Moreover, immobilization of the flexible linker of the bivalent ligand upon receptor binding results in a decrease in entropy. That means, the improved affinity of a bivalent ligand is enthalpy driven and to some extent compensated through the loss of entropy (ignoring a possible increase in entropy due to linker-mediated dehydration of a receptor surface).^{68, 69} For many bivalent ligands containing a linker of insufficient length to bridge neighboring receptors an increase in selectivity relative to the monovalent counterparts is observed.⁶³ This can be explained with an accessory binding site for the second pharmacophore (Figure 1.4 A) present only on one receptor subtype. The affinity of bivalent ligands can also be influenced by cooperativity effects.^{64, 65}

Figure 1.4. Bivalent ligand binding to A, a GPCR with an accessory binding site, or to B, a GPCR dimer. The bivalent ligand is believed to bind first in a univalent manner before addressing the second binding site. Concerning the receptor dimer (B) univalent binding of a second bivalent ligand is possible. However, bridging neighboring receptors via the bivalent ligand is favored. Adapted from Portoghese et al.⁶⁵

bivalent ligand A

B

Introduction 9

Taken together, the bivalent ligand approach is a promising way not just to design highly potent and selective compounds. The recent advances in studying GPCR dimerization and the rising evidence for heterodimers to form a kind of new receptor subtypes with modified pharmacological behavior will promote the demand for molecules addressing neighboring receptors simultaneously.^{9, 43} Such bivalent ligands are required as pharmacological tools to study receptor dimerization and the potential of GPCR dimers and higher order oligomers as drug targets.