Файл: Extracellular and Intracellular Signaling (книга).pdf

involving native tissues that some GPCRs form homodimers and some form heterodimers in their functional forms.

A GPCR class C receptor mGlu5 has been shown to be a homodimer stabilized by a disulfide bridge in the extracellular domain.81 The extracellular VFT domain of mGlu1 receptor has been crystallized as a dimer with a disulfide bridge in di erent relative orientations of the monomers depending on the presence or absence of the agonist glutamate,82,83 suggesting a possible mechanism of activation involving the relative change in the orientation of the monomeric TM bundles in the homodimer.84

Another class C GPCR provided the first conclusive evidence of a functional heterodimer. The g-aminobutyric acid-binding (GABAB) receptor is only functional as a heterodimer between GABAB1 and GABAB2 monomeric subunits, where each subunit plays a distinct role: GABAB1 subunit binds to the agonists and GABAB2 subunit couples to the G protein.85 Sweet and umami taste receptors provide other class C examples of heterodimerization. Of the three genes encoding these receptors (T1R1, T1R2 and T1R3), T1R2-T1R3 heterodimer results in a sweet receptor and the T1R1-T1R3 heterodimer results in an umami receptor.14 The three monomeric receptors don’t display the functional behavior if they are expressed alone.

There is also plenty of evidence now in favor of the presence of functional homodimers and heterodimers in class A GPCRs.80 Here we will mention a more recent example involving the dopamine D2 receptor, which has been shown to form homodimers out of monomers in functionally di erent states; maximal activation was observed when one monomer was bound to the agonist (was active) and the other monomer was bound to an inverse agonist (was inactive).86

Interaction of GPCRs with other membrane and cytoplasmic proteins has been known for a while and has also been reviewed.87 The physiological implications of these interactions are slowly being uncovered as only some of these interactions have been amenable to detailed experimental investigations. A discussion of these interactions is beyond the scope of this chapter.

11.3.3Functional Control of GPCRs by Ligands

As mentioned earlier, GPCRs are pleiotropic in terms of the multiple intracellular signaling cascades they can a ect upon binding to an agonist through both G protein-coupled and b-arrestin coupled pathways. Experimentally, a single functional assay (usually by definition) cannot see all the signaling e ects of a ligand. It is now evident that agonist-bound GPCRs exist in multiple distinct conformations, where each conformation can potentially activate a di erent signaling pathway. From a therapeutic perspective, this may not be desirable if, of the multiple signaling pathways activated by a drug-molecule, one pathway may be mainly responsible for the desirable therapeutic benefit and another may be causing unwanted side-e ects. This opens at least two distinct possibilities of controlling the functional consequences of GPCR activation: biased agonists and allosteric modulators. Each of these possibilities will be

binding to GPCRs can induce a cascade of processes, some through G protein coupling and some through b-arrestin coupling. As shown before, b-arrestin coupling can lead to specific signaling events and receptor endocytosis, which can further lead to either recycling of the receptor or its degradation. Figure 11.16 also shows the e ect of biased agonists A and B that can either exist or be designed to activate a subset of possible signaling pathways.

Classically, the GPCR agonists have been characterized as such by their e ect on G protein-coupled pathways. Experiments measuring b-arrestin signaling are becoming more commonplace so the e ects of classical agonists need to be evaluated and should lead to a detailed characterization of these ligands. The parathyroid hormone (PTH), for example, can activate extracellular signal-related kinase using distinct G protein-dependent and G proteinindependent pathways. The PTH analogs, however, can use the same PTH receptor and separately use either G protein-dependent or -independent pathways. An example is that [Trp1]PTHrp-(1-36) stimulates ERK1/2 via G protein pathway, whereas PTH-1A [[D-Trp12,Tyr34]PTH-(7-34)] stimulates the same via b-arrestin pathway in a G protein-independent manner.78 This can have direct therapeutic consequences because PTH regulates calcium homeostasis as well as bone metabolism and utilization of b-arrestin 2 pathway is critical for this benefit,88 so b-arrestin biased PTH analogs mentioned above provide potentially improved therapy for osteoporosis. Another example is that nicotinic acid is therapeutically very beneficial as an anti-lypolytic agent (via G protein-mediated pathways), but causes cutaneous flushing as a major side-e ect, which has been directly linked to the activation of b-arrestin 1 pathways.89 An analog of this molecule that doesn’t a ect the G protein pathways but blocks the b-arrestin 1 pathways will be highly desirable. This also necessitates new characterization of classical agonists, e.g. a ligand that blocks G protein pathways but uses b-arrestin pathways may have been classified before as an antagonist (or inverse agonist) but now should be more accurately described as a b-arrestin biased agonist.

These studies are also increasing our understanding of the relationship between various signaling pathways and previously unexplained side-e ects of drug molecules. The use of knowledge about biased signaling during the drug design phase has the potential to generate multiple novel ways to control and hopefully cure many ailments with minimal side-e ects.

11.3.3.2Allosteric Ligands and Signal Modulation

Orthosteric ligands bind to GPCRs in regions that fully or partially overlap with that of the endogenous ligand(s), thereby sterically excluding the possibility of both occupying the GPCR at the same time. Allosteric ligands bind to GPCRs in regions that don’t overlap with the endogenous ligand binding site, so both can occupy the receptor at the same time. This can have important signaling consequences because, as mentioned before, agonists induce an ensemble of GPCR conformations with a range of functional implications and allosteric ligands can dramatically modulate those conformations (e.g. by

antagonist. The biased antagonist N-a-tosyltryptophan (N-a-T) selectively blocks the G protein coupled pathways allowing the agonist PGD2 to continue to activate the b-arrestin pathways.92

There is plenty of evidence in favor of di erent allosteric modulators stabilizing a di erent subset of GPCR conformations, one being antibody binding profiles of Ab45531 and Ab45523 for the chemokine receptor CCR5, which is a coreceptor for HIV-1 entry. These antibody binding profiles di er in the presence of allosteric HIV-1 entry inhibitors like aplaviroc, TAK-779 and SCH-C.65

The muscarinic acetylcholine receptors (mAChRs) are class A GPCRs that

provide a good pharmacological example for the need of selective allosteric modulators.93,94 There are five mAChR subtypes implicated in many physio-

logical pathways. The acetylcholine binding site residues are highly conserved across these five receptor subtypes, making it di cult to design orthosteric agonists or antagonists. It has been shown that these receptors have one or two allosteric binding sites available for allosteric modulators to regulate the e ect of orthosteric ligands.93 This provides for a general strategy for therapies targeting receptor subtype selectivity as the allosteric sites will be much less conserved than the orthosteric sites. In addition, a new class of modulators

has emerged called ‘‘bitopic’’ that interact with both the allosteric and the orthosteric sites for self-modulation of their activity.95,96

The studies of these modulators really adds to the diverse ways in which GPCR conformational flexibility can be exploited for innate signaling as well as for therapeutic applications where target receptor subtype selectivity is highly desirable or activation of specific pathways causes undesirable side-e ects.

11.3.4Challenges in GPCR Targeted Drug Design

The lack of 3D structures for most human GPCRs (none were available until 2007) has led to the almost exclusive use of high-throughput screening (HTS) and virtual ligand screening (VLS) techniques in drug design, which do not require protein structures as input. Even though the number of compounds being screened by HTS and VLS techniques is higher than ever, the number of new approved drugs is on a decline. A large proportion (43%) of drug candidates in clinical trials fail due to lack of e cacy and a significant one-third fail due to toxicity and side-e ects.97 Some of these undesirable e ects result from drugs hitting other GPCRs or even other subtypes of the same receptor, e.g. many dopamine D2 receptor agonists used for treating Parkinson’s disease produce behavioral side-e ects (e.g. compulsive gambling even in nongamblers)98 because they actually bind with higher a nity to the dopamine D3 receptor which has been associated with emotion, reward and addiction. Such side-e ects can be e ectively minimized by designing D2 selective agonists, which requires atomic-level structures for both D2 and D3 receptors. As the theme of this chapter suggests that multiple GPCR conformations and their signaling consequences need to be accounted for, it is a daunting challenge. Concerted e orts are underway to crystallize a number of GPCRs with agonists, antagonists and inverse agonists, which will slowly provide valuable

information. Parallel to these e orts, validated structure prediction methods that can generate an ensemble of GPCR structures that can be validated by mutagenesis as well as functional/binding assays and provide avenues for rational structure-based drug design are highly desirable.

Identification of the molecular basis of toxicity and other side-e ects has been possible for a very few cases, some of which have been mentioned in this chapter. A recent pertinent example is the anti-obesity target, the cannabinoid CB1 receptor, which is a GPCR stimulated by cannabis, generating feelings of euphoria. Application of two promising anti-obesity CB1 antagonists, rimonabant (marketed, and then pulled from the European Union) and taranabant, resulted in adverse psychiatric symptoms like depression, anxiety and suicidal ideation. This has led many big pharmaceutical companies to abandon drug development programs on CB1 target, as it is not clear whether side- e ects are due to o -target interactions (i.e. not related to CB1) or due to CB1 antagonism.

This leads to two possible scenarios. Firstly, if the side-e ects are confirmed to be o -target and it becomes known which receptors need to be avoided (hereafter called anti-targets), the structure-based rational drug design process can potentially use the structures of the target receptor and all the anti-target receptors to tailor the drug towards the target and away from the anti-targets, taking into account any information about signaling pathways that might be available. Secondly, if the side-e ects are confirmed to be due to the target receptor, then it is most likely due to one of the multiple signaling pathways activated by the receptor. This will require a detailed therapeutic characterization of conformation-specific signaling pathways for the target receptor, so that biased agonists or allosteric modulators can be rationally designed using the therapeutically beneficial pathways.

Identification of o -targets for side-e ects of drugs is an area of active research. A recent study looked at 3665 FDA approved and investigational drugs and their chemical similarities against 200,000 ligands that have been organized into families based on the proteins they target.99 This identified thousands of unanticipated o -targets for the approved drugs. Some of these are shown in Figure 11.18 using red arrows. Based on these predictions, 30 o -target associations were tested experimentally and 23 were confirmed, 5 of which were potent (o100 nM). Key unknown associations uncovered were that transporter inhibitor Prozac antagonizes the b1 adrenergic receptor, the ion-channel drug Vadilex inhibits the 5-hydroxytryptamine (5-HT) transporter and the enzyme inhibitor Rescriptor antagonizes the histamine H4 receptor. Prozac’s adverse e ects in terms of SSRI-discontinuation syndrome and sexual dysfunction can be explained by its b1 blocking ability as b blockers have those e ects. Vadilex and Rescriptor are displaying polypharmacology as they are hitting very unrelated receptors, confirming that structurally diverse proteins can bind to structurally similar ligands. Such analysis needs to become routine to help minimize potential side-e ects of the drugs being developed.

Experimental studies are uncovering new signaling mechanisms and computational studies can complement them through their use in rational drug