Receptor Heterogeneity and Subtypes

Receptors within a given family generally occur in several molecular varieties, or subtypes, with similar architecture but significant differences in their sequences, and often in their pharmacological properties.⁶ Nicotinic acetylcholine receptors are typical in this respect; distinct subtypes occur in different brain regions (see Table 38.2), and these differ from the muscle receptor. Some of the known pharmacological differences (e.g. sensitivity to blocking agents) between muscle and brain acetylcholine receptors correlate with specific sequence differences; however, as far as we know, all nicotinic acetylcholine receptors respond to the same physiological mediator and produce the same kind of synaptic response, so why many variants should have evolved is still a puzzle.

Much of the sequence variation that accounts for receptor diversity arises at the genomic level, i.e. different genes give rise to distinct receptor subtypes. Additional variation arises from alternative mRNA splicing, which means that a single gene can give rise to more than one receptor isoform. After translation from genomic DNA, the mRNA normally contains non-coding regions (introns) that are excised by mRNA splicing before the message is translated into protein. Depending on the location of the splice sites, splicing can result in inclusion or deletion of one or more of the mRNA coding regions, giving rise to long or short forms of the protein. This is an important source of variation, particularly for GPCRs (see Kilpatrick et al., 1999), which produces receptors with different binding characteristics and different signal transduction mechanisms, although its pharmacological relevance remains to be clarified. Another process that can produce different receptors from the same gene is mRNA editing, which involves the mischievous substitution of one base in the mRNA for another, and hence a small variation in the amino acid sequence of the receptor.

Molecular heterogeneity of this kind is a feature of all kinds of receptors—indeed of functional proteins in general. New receptor subtypes and isoforms continue to be discovered, and regular updates of the catalogue are available (Alexander et al., 2009; IUPHAR Receptor Database and Channel Compendium). The problems of classification, nomenclature and taxonomy resulting from this flood of data have been mentioned earlier (p. 8). From the pharmacological viewpoint, where our concern is to understand individual drugs and what they do to living organisms, and to devise better ones, it is important that we keep molecular pharmacology in perspective. The ‘newer rite’ has proved revelatory in many ways, but the sheer complexity of the ways in which molecules behave means that we have a long way to go before reaching the reductionist Utopia that molecular biology promises. When we do, this book will get much shorter. In the meantime, we try to pick out the general principles without getting too bogged down in detail.

We will now describe the characteristics of each of the four receptor superfamilies.

Molecular Structure

These molecules have structural features in common with other ion channels, described on p. 45 (Ashcroft, 2000). The nicotinic acetylcholine receptor (Fig. 3.4), the first to be cloned, has been studied in great detail (see Karlin, 1993). It consists of a pentameric assembly of different subunits, of which there are four types, termed α, β, γ and δ, each of molecular weight (M_r) 40–58 kDa. The subunits show marked sequence homology, and each contains four membrane-spanning α-helices, inserted into the membrane as shown in Figure 3.4B. The pentameric structure (α₂, β, γ, δ) possesses two acetylcholine binding sites, each lying at the interface between one of the two α subunits and its neighbour. Both must bind acetylcholine molecules in order for the receptor to be activated. This receptor is sufficiently large to be seen in electron micrographs, and Figure 3.4B shows its structure, based mainly on a high-resolution electron diffraction study (Unwin, 1993, 1995; Miyazawa et al., 2003). Each subunit spans the membrane four times, so the channel comprises no fewer than 20 membrane-spanning helices surrounding a central pore.

Fig. 3.4 Structure of the nicotinic acetylcholine receptor (a typical ligand-gated ion channel).

[A] Schematic diagram in side view (upper) and plan view (lower). The five receptor subunits (α₂, β, γ, δ) form a cluster surrounding a central transmembrane pore, the lining of which is formed by the M₂ helical segments of each subunit. These contain a preponderance of negatively charged amino acids, which makes the pore cation selective. There are two acetylcholine binding sites in the extracellular portion of the receptor, at the interface between the α and the adjoining subunits. When acetylcholine binds, the kinked α-helices either straighten out or swing out of the way, thus opening the channel pore.

(Based on Unwin N 1993 Nicotinic acetylcholine receptor at 9A resolution. J Mol Biol 229: 1101–1124, and Unwin N 1995 Acetylcholine receptor channel imaged in the open state. Nature 373: 37–43.) [B] High-resolution image showing revised arrangement of intracellular domains.

(Reproduced with permission from Unwin N 2005 Refined structure of the nicotinic acetylcholine receptor at 4A resolution. J Mol Biol Mar 4;346(4):967–989.)

The two acetylcholine-binding sites lie on the extracellular parts of the two α subunits. One of the transmembrane helices (M₂) from each of the five subunits forms the lining of the ion channel (Fig. 3.4). The five M₂ helices that form the pore are sharply kinked inwards halfway through the membrane, forming a constriction. When acetylcholine molecules bind, a conformation change occurs in the extracellular part of the receptor (see review by Gay & Yakel, 2007), which twists the α subunits, causing the kinked M₂ segments to swivel out of the way, thus opening the channel (Miyazawa et al., 2003). The channel lining contains a series of anionic residues, making the channel selectively permeable to cations.

The use of site-directed mutagenesis, which enables short regions, or single residues, of the amino acid sequence to be altered, has shown that a mutation of a critical residue in the M₂ helix changes the channel from being cation selective (hence excitatory in the context of synaptic function) to being anion selective (typical of receptors for inhibitory transmitters such as GABA). Other mutations affect properties such as gating and desensitisation of ligand-gated channels.

Receptors for other fast transmitters, such as GABA_A receptors (Ch. 37), 5-HT (Ch. 15) and glycine receptors (Ch. 37), are built on the same five-subunit pattern, and form the group of cys-loop receptors. Other ligand-gated ion channels, such as glutamate receptors (see Ch. 37) and the ‘capsaicin receptor’ (TRPV1; see Ch. 41), whose structures are shown in Figure 3.18, have a different (P-loop) architecture, in which the pore is built from loops rather than transmembrane helices (see p. 45), in common with many other (non-ligand-gated) ion channels.

The Gating Mechanism

Receptors of this type control the fastest synaptic events in the nervous system, in which a neurotransmitter acts on the postsynaptic membrane of a nerve or muscle cell and transiently increases its permeability to particular ions. Most excitatory neurotransmitters, such as acetylcholine at the neuromuscular junction (Ch. 12) or glutamate in the central nervous system (Ch. 37), cause an increase in Na⁺ and K⁺ permeability. This results in a net inward current carried mainly by Na⁺, which depolarises the cell and increases the probability that it will generate an action potential. The action of the transmitter reaches a peak in a fraction of a millisecond, and usually decays within a few milliseconds. The sheer speed of this response implies that the coupling between the receptor and the ionic channel is a direct one, and the molecular structure of the receptor–channel complex (see above) agrees with this. In contrast to other receptor families (see below), no intermediate biochemical steps are involved in the transduction process.

A breakthrough by Katz and Miledi in 1972 made it possible for the first time to study the properties of individual ligand-gated channels by the use of noise analysis. Studying the action of acetylcholine at the motor endplate, they observed that small random fluctuations of membrane potential were superimposed on the steady depolarisation produced by acetylcholine (Fig. 3.5). These fluctuations arise because, in the presence of an agonist, there is a dynamic equilibrium between open and closed ion channels. In the steady state, the rate of opening balances the rate of closing, but from moment to moment the number of open channels will show random fluctuations about the mean. By measuring the amplitude of these fluctuations, the conductance of a single ion channel can be calculated, and by measuring their frequency (usually in the form of a spectrum in which the noise power of the signal is plotted as a function of frequency), the average duration for which a single channel stays open (mean open time) can be calculated. In the case of acetylcholine acting at the endplate, the channel conductance is about 20 picosiemens (pS), which is equivalent to an influx of about 10⁷ ions per second through a single channel under normal physiological conditions, and the mean open time is 1–2 ms. The magnitude of the single channel conductance confirms that permeation occurs through a physical pore through the membrane, because the ion flow is too large to be compatible with a carrier mechanism. The channel conductance produced by different acetylcholine-like agonists is the same, whereas the mean channel lifetime varies.

Fig. 3.5 Acetylcholine-induced noise at the frog motor endplate.

[A] Records of membrane current recorded at high gain under voltage clamp. The upper noise record was recorded during the application of acetylcholine (ACh) from a micropipette. The lower record was obtained in the absence of ACh, the blip in the middle being caused by the spontaneous release of a packet of ACh from the motor nerve. The steady (DC) component of the ACh signal has been removed by electronic filtering, leaving the high-frequency noise signal. [B] Power spectrum of ACh-induced noise recorded in a similar experiment to that shown above. The spectrum is calculated by Fourier analysis and fitted with a theoretical (Lorentzian) curve that corresponds to the expected behaviour of a single population of channels whose lifetime varies randomly. The cut-off frequency (at which the power is half of its limiting low-frequency value) enables the mean channel lifetime to be calculated.

(From [A] Anderson C R, Stevens C F 1973 J Physiol 235: 655; [B] Ogden D C et al. 1981 Nature 289: 596.)

The simple scheme shown in Fig. 2.1 is a useful model for ion channel gating. The conformation R*, representing the open state of the ion channel, is thought to be the same for all agonists, accounting for the finding that the channel conductance does not vary. Kinetically, the mean open time is determined mainly by the closing rate constant, α, and this varies from one drug to another. As explained in Chapter 2, an agonist of high efficacy that activates a large proportion of the receptors that it occupies will be characterised by β/α >> 1, whereas for a drug of low efficacy β/α has a lower value.

The patch clamp recording technique, devised by Neher and Sakmann, allows the very small current flowing through a single ionic channel to be measured directly (Fig. 3.6), and the results have fully confirmed the interpretation of channel properties based on noise analysis. This technique provides a view, unique in biology, of the physiological behaviour of individual protein molecules in real time, and has given many new insights into the gating reactions and permeability characteristics of both ligand-gated channels and voltage-gated channels (see p. 43). Single-channel recording has shown that many agonists cause individual channels to open to one or more of several distinct conductance levels. In the case of glutamate-activated channels, it appears that different agonists produce different receptor conformations associated with different channel conductances (Jin et al., 2003). Desensitisation of ligand-gated ion channels also involves one or more additional agonist-induced conformational states. These findings necessitate some elaboration of the simple scheme of Figure 2.1, in which only a single open state, R*, is represented, and are an example of the way in which the actual behaviour of receptors makes our theoretical models look a little threadbare.

Fig. 3.6 Single acetylcholine-operated ion channels at the frog motor endplate recorded by the patch clamp technique.

The pipette, which was applied tightly to the surface of the membrane, contained 10 µmol/l ACh. The downward deflections show the currents flowing through single ion channels in the small patch of membrane under the pipette tip. Towards the end of the record, two channels can be seen to open simultaneously. The conductance and mean lifetime of these channels agrees well with indirect estimates from noise analysis (see Fig. 3.5).

(Courtesy of D Colquhoun and D C Ogden.)

Molecular Structure

The first GPCR to be fully characterised was the β-adrenoceptor (Ch. 14), which was cloned in 1986. Molecular biology caught up very rapidly with pharmacology, and all of the receptors that had been identified by their pharmacological properties have now been cloned. What seemed revolutionary in 1986 is now commonplace, and nowadays any aspiring receptor has to be cloned before it is taken seriously.

G-protein-coupled receptors consist of a single polypeptide chain of up to 1100 residues whose general anatomy is shown in Figure 3.3B. Their characteristic structure comprises seven transmembrane α-helices, similar to those of the ion channels discussed above, with an extracellular N-terminal domain of varying length, and an intracellular C-terminal domain.

GPCRs are divided into three distinct families (see Schwartz, 1996). There is considerable sequence homology between the members of one family, but none between different families. They share the same seven-helix (heptahelical) structure, but differ in other respects, principally in the length of the extracellular N terminus and the location of the agonist binding domain (Table 3.3). Family A is by far the largest, comprising most monoamine, neuropeptide and chemokine receptors. Family B includes receptors for some other peptides, such as calcitonin and glucagon (see Ch. 19). Family C is the smallest, its main members being the metabotropic glutamate and GABA receptors (Ch. 37) and the Ca²⁺-sensing receptors⁸ (see Ch. 35).

Table 3.3 G-protein-coupled receptor families^a

^a A fourth distinct family includes many receptors for pheromones but no pharmacological receptors.

^b For full lists, see http://www.iuphar-db.org.

The understanding of the function of receptors of this type owes much to studies of a closely related protein, rhodopsin, which is responsible for transduction in retinal rods. This protein is abundant in the retina, and much easier to study than receptor proteins (which are anything but abundant); it is built on an identical plan to that shown in Figure 3.3 and also produces a response in the rod (hyperpolarisation, associated with inhibition of Na⁺ conductance) through a mechanism involving a G-protein (see below). The most obvious difference is that a photon, rather than an agonist molecule, produces the response. In effect, rhodopsin can be regarded as incorporating its own inbuilt agonist molecule, namely retinal, which isomerises from the trans (inactive) to the cis (active) form when it absorbs a photon.

Site-directed mutagenesis experiments show that the long third cytoplasmic loop is the region of the molecule that couples to the G-protein, because deletion or modification of this section results in receptors that still bind ligands but cannot associate with G-proteins or produce responses. Usually, a particular receptor subtype couples selectively with a particular G-protein, and swapping parts of the cytoplasmic loop between different receptors alters their G-protein selectivity.

For small molecules, such as noradrenaline (norepinephrine), the ligand-binding domain of class A receptors is buried in the cleft between the α-helical segments within the membrane (Fig. 3.3B), similar to the slot occupied by retinal in the rhodopsin molecule. Peptide ligands, such as substance P (Ch. 19) bind more superficially to the extracellular loops, as shown in Figure 3.3B. By single-site mutagenesis experiments, it is possible to map the ligand-binding domain of these receptors, and the hope is that it may soon be possible to design synthetic ligands based on knowledge of the receptor site structure—an important milestone for the pharmaceutical industry, which has relied up to now mainly on the structure of endogenous mediators (such as histamine) or plant alkaloids (such as morphine) for its chemical inspiration.⁹ Recently, the difficulties of crystallising type A GPCRs have been overcome, allowing the use of the powerful technique of X-ray crystallography to study the molecular structure of these receptors in detail (see Weis & Kobilka, 2008). Also, fluorescence methods have been developed to study the kinetics of ligand binding and subsequent conformational changes associated with activation (see Lohse et al., 2008). From such studies we should gain a clearer picture of the mechanism of activation of GPCRs and the factors determining agonist efficacy, as well as having a better basis for designing new GPCR ligands.

Protease-activated receptors

Although activation of GPCRs is normally the consequence of a diffusible agonist, it can be the result of protease activation. Four types of protease-activated receptors (PARs), have been identified (see review by Ramachandran & Hollenberg, 2008). Many proteases, such as thrombin (a protease involved in the blood-clotting cascade; see Ch. 24), activate PARs by snipping off the end of the extracellular N-terminal tail of the receptor (Fig. 3.7) to expose five or six N-terminal residues that bind to receptor domains in the extracellular loops, functioning as a ‘tethered agonist’. Receptors of this type occur in many tissues (see Ramachandran & Hollenberg, 2008), and they appear to play a role in inflammation and other responses to tissue damage where tissue proteases are released. One of the family of PARs, PAR-2, is activated by a protease released from mast cells, and is expressed on sensory neurons. It is thought to play a role in inflammatory pain (see Ch. 41). A PAR molecule can be activated only once, because the cleavage cannot be reversed, so continuous resynthesis of receptor protein is necessary. Inactivation occurs by a further proteolytic cleavage that frees the tethered ligand, or by desensitisation, involving phosphorylation (see below), after which the receptor is internalised and degraded, to be replaced by newly synthesised protein.

Fig. 3.7 Activation of a protease-activated receptor by cleavage of the N-terminal extracellular domain.

Inactivation occurs by phosphorylation. Recovery requires resynthesis of the receptor.

page 30

G-protein-coupled receptors

• These are sometimes called metabotropic receptors.

• Structures comprise seven membrane-spanning α-helices, often linked as dimeric structures.

• One of the intracellular loops is larger than the others and interacts with the G-protein.

• The G-protein is a membrane protein comprising three subunits (α, β, γ), the α subunit possessing GTPase activity.

• When the trimer binds to an agonist-occupied receptor, the α subunit dissociates and is then free to activate an effector (a membrane enzyme or ion channel). In some cases, the βγ subunit is the activator species.

• Activation of the effector is terminated when the bound GTP molecule is hydrolysed, which allows the α subunit to recombine with βγ.

• There are several types of G-protein, which interact with different receptors and control different effectors.

• Examples include muscarinic acetylcholine receptors, adrenoceptors, neuropeptide and chemokine receptors, and protease-activated receptors.

G-Proteins and Their Role

G-proteins comprise a family of membrane-resident proteins whose function is to recognise activated GPCRs and pass on the message to the effector systems that generate a cellular response. They represent the level of middle management in the organisational hierarchy, intervening between the receptors—choosy mandarins alert to the faintest whiff of their preferred chemical—and the effector enzymes or ion channels—the blue-collar brigade that gets the job done without needing to know which hormone authorised the process. They are the go-between proteins, but were actually called G-proteins because of their interaction with the guanine nucleotides, GTP and GDP. For more detailed information on the structure and functions of G-proteins, see reviews by Milligan & Kostenis (2006) and Oldham & Hamm (2008). G-proteins consist of three subunits: α, β and γ (Fig. 3.8). Guanine nucleotides bind to the α subunit, which has enzymic activity, catalysing the conversion of GTP to GDP. The β and γ subunits remain together as a βγ complex. All three subunits are anchored to the membrane through a fatty acid chain, coupled to the G-protein through a reaction known as prenylation. G-proteins appear to be freely diffusible in the plane of the membrane, so a single pool of G-protein in a cell can interact with several different receptors and effectors in an essentially promiscuous fashion. In the ‘resting’ state (Fig. 3.8), the G-protein exists as an unattached αβγ trimer, with GDP occupying the site on the α subunit. When a GPCR is activated by an agonist molecule, a conformational change occurs, involving the cytoplasmic domain of the receptor (Fig. 3.3B), causing it to acquire high affinity for αβγ. Association of αβγ with the receptor occurs within about 50 ms, causing the bound GDP to dissociate and to be replaced with GTP (GDP–GTP exchange), which in turn causes dissociation of the G-protein trimer, releasing α-GTP and βγ subunits; these are the ‘active’ forms of the G-protein, which diffuse in the membrane and can associate with various enzymes and ion channels, causing activation of the target (Fig. 3.8). It was originally thought that only the α subunit had a signalling function, the βγ complex serving merely as a chaperone to keep the flighty α subunits out of range of the various effector proteins that they might otherwise excite. However, the βγ complexes actually make assignations of their own, and control effectors in much the same way as the α subunits (see Clapham & Neer, 1997). Association of α or βγ subunits with target enzymes or channels can cause either activation or inhibition, depending on which G-protein is involved (see Table 3.4).

Fig. 3.8 The function of the G-protein.

The G-protein consists of three subunits (α, β, γ), which are anchored to the membrane through attached lipid residues. Coupling of the α subunit to an agonist-occupied receptor causes the bound GDP to exchange with intracellular GTP; the α–GTP complex then dissociates from the receptor and from the βγ complex, and interacts with a target protein (target 1, which may be an enzyme, such as adenylyl cyclase, or an ion channel). The βγ complex may also activate a target protein (target 2). The GTPase activity of the α subunit is increased when the target protein is bound, leading to hydrolysis of the bound GTP to GDP, whereupon the α subunit reunites with βγ.

Table 3.4 The main G-protein subtypes and their functions^a

Signalling is terminated when the hydrolysis of GTP to GDP occurs through the GTPase activity of the α subunit. The resulting α–GDP then dissociates from the effector, and reunites with βγ, completing the cycle. Attachment of the α subunit to an effector molecule actually increases its GTPase activity, the magnitude of this increase being different for different types of effector. Because GTP hydrolysis is the step that terminates the ability of the α subunit to produce its effect, regulation of its GTPase activity by the effector protein means that the activation of the effector tends to be self-limiting. The mechanism results in amplification because a single agonist–receptor complex can activate several G-protein molecules in turn, and each of these can remain associated with the effector enzyme for long enough to produce many molecules of product. The product (see below) is often a ‘second messenger’, and further amplification occurs before the final cellular response is produced.

How is specificity achieved so that each kind of receptor produces a distinct pattern of cellular responses? With a common pool of promiscuous G-proteins linking the various receptors and effector systems in a cell, it might seem that all specificity would be lost, but this is clearly not the case. For example, mAChRs and β-adrenoceptors, both of which occur in cardiac muscle cells, produce opposite functional effects (Chs 13 and 14). The main reason is molecular variation within the α subunits, of which more than 20 subtypes have been identified¹⁰ (see Wess, 1998; Table 3.4). Four main classes of G-protein (G_s, G_i, G_o and G_q) are of pharmacological importance. As summarised in Table 3.4, they show selectivity with respect to both the receptors and the effectors with which they couple, having specific recognition domains in their structure complementary to specific G-protein-binding domains in the receptor and effector molecules. G_s and G_i produce, respectively, stimulation and inhibition of the enzyme adenylyl cyclase (Fig. 3.9).

Fig. 3.9 Bidirectional control of a target enzyme, such as adenylate cyclase by G_s and G_i.

Heterogeneity of G-proteins allows different receptors to exert opposite effects on a target enzyme.

The α subunits of these G-proteins differ in structure. One functional difference that has been useful as an experimental tool to distinguish which type of G-protein is involved in different situations concerns the action of two bacterial toxins, cholera toxin and pertussis toxin (see Table 3.4). These toxins, which are enzymes, catalyse a conjugation reaction (ADP ribosylation) on the α subunit of G-proteins. Cholera toxin acts only on G_s, and it causes persistent activation. Many of the symptoms of cholera, such as the excessive secretion of fluid from the gastrointestinal epithelium, are due to the uncontrolled activation of adenylate cyclase that occurs. Pertussis toxin specifically blocks G_i and G_o by preventing dissociation of the G-protein trimer.

Targets for G-Proteins

The main targets for G-proteins, through which GPCRs control different aspects of cell function (see Milligan, 1995; Nahorski, 2006; Table 3.4), are:

The adenylyl cyclase/cAMP system

The discovery by Sutherland and his colleagues of the role of cAMP (cyclic 3′,5′-adenosine monophosphate) as an intracellular mediator demolished at a stroke the barriers that existed between biochemistry and pharmacology, and introduced the concept of second messengers in signal transduction. cAMP is a nucleotide synthesised within the cell from ATP by the action of a membrane-bound enzyme, adenylyl cyclase. It is produced continuously and inactivated by hydrolysis to 5′-AMP by the action of a family of enzymes known as phosphodiesterases (PDEs). Many different drugs, hormones and neurotransmitters act on GPCRs and produce their effects by increasing or decreasing the catalytic activity of adenylyl cyclase, thus raising or lowering the concentration of cAMP within the cell. There are nine different molecular isoforms of the enzyme, some of which respond selectively to Gα_s or Gα_i (see Simonds, 1999).

Cyclic AMP regulates many aspects of cellular function including, for example, enzymes involved in energy metabolism, cell division and cell differentiation, ion transport, ion channels, and the contractile proteins in smooth muscle. These varied effects are, however, all brought about by a common mechanism, namely the activation of protein kinases by cAMP. Protein kinases regulate the function of many different cellular proteins by controlling protein phosphorylation (see p. 39) Figure 3.10 shows how increased cAMP production in response to β-adrenoceptor activation affects enzymes involved in glycogen and fat metabolism in liver, fat and muscle cells. The result is a coordinated response in which stored energy in the form of glycogen and fat is made available as glucose to fuel muscle contraction.

Fig. 3.10 Regulation of energy metabolism by cAMP.

AC, adenylyl cyclase.

Other examples of regulation by cAMP-dependent protein kinases include the increased activity of voltage-gated calcium channels in heart muscle cells (see Ch. 21). Phosphorylation of these channels increases the amount of Ca²⁺ entering the cell during the action potential, and thus increases the force of contraction of the heart.

In smooth muscle, cAMP-dependent protein kinase phosphorylates (thereby inactivating) another enzyme, myosin-light-chain kinase, which is required for contraction. This accounts for the smooth muscle relaxation produced by many drugs that increase cAMP production in smooth muscle (see Ch. 4).

As mentioned above, receptors linked to G_i rather than G_s inhibit adenylyl cyclase, and thus reduce cAMP formation. Examples include certain types of mAChR (e.g. the M₂ receptor of cardiac muscle; see Ch. 13), α₂ adrenoceptors in smooth muscle (Ch. 14) and opioid receptors (see Ch. 41). Adenylyl cyclase can be activated directly by certain agents, including forskolin and fluoride ions, agents that are used experimentally to study the role of the cAMP system.

Cyclic AMP is hydrolysed within cells by phosphodiesterases (PDES), an important and ubiquitous family of enzymes (see Beavo, 1995, for review). Eleven PDE subtypes exist, of which some (e.g. PDE₃ and PDE₄) are cAMP selective, while others (e.g. PDE₅) are cGMP selective. Most are weakly inhibited by drugs such as methylxanthines (e.g. theophylline and caffeine; see Chs 27 and 47). Rolipram (used to treat asthma; Ch. 27) is selective for PDE₄, expressed in inflammatory cells; milrinone (used to treat heart failure; Ch. 21) is selective for PDE₃, which is expressed in heart muscle; sildenafil (better known as Viagra; Ch. 34) is selective for PDE₅, and consequently enhances the vasodilator effects of nitrous oxide (NO) and drugs that release NO, whose effects are mediated by cGMP (see Ch. 20). The similarity of some of the actions of these drugs to those of sympathomimetic amines (Ch. 14) probably reflects their common property of increasing the intracellular concentration of cAMP. Selective inhibitors of the various PDEs are being developed, mainly to treat cardiovascular and respiratory diseases.

The phospholipase C/inositol phosphate system

The phosphoinositide system, an important intracellular second messenger system, was first discovered in the 1950s by Hokin and Hokin, whose recondite interests centred on the mechanism of salt secretion by the nasal glands of seabirds. They found that secretion was accompanied by increased turnover of a minor class of membrane phospholipids known as phosphoinositides (collectively known as PIs; Fig. 3.11). Subsequently, Michell and Berridge found that many hormones that produce an increase in free intracellular Ca²⁺ concentration (which include, for example, muscarinic agonists and α-adrenoceptor agonists acting on smooth muscle and salivary glands, and vasopressin acting on liver cells) also increase PI turnover. Subsequently, it was found that one particular member of the PI family, namely phosphatidylinositol (4,5) bisphosphate (PIP₂), which has additional phosphate groups attached to the inositol ring, plays a key role. PIP₂ is the substrate for a membrane-bound enzyme, phospholipase Cβ (PLCβ), which splits it into diacylglycerol (DAG) and inositol (1,4,5) trisphosphate (IP₃; Fig. 3.12), both of which function as second messengers as discussed below. The activation of PLCβ by various agonists is mediated through a G-protein (G_q, see Table 3.4). After cleavage of PIP₂, the status quo is restored as shown in Figure 3.12, DAG being phosphorylated to form phosphatidic acid (PA), while the IP₃ is dephosphorylated and then recoupled with PA to form PIP₂ once again.¹¹ Lithium, an agent used in psychiatry (see Ch. 46), blocks this recycling pathway (see Fig. 3.12).

Fig. 3.11 Structure of phosphatidylinositol bisphosphate (PIP₂), showing sites of cleavage by different phospholipases to produce active mediators.

Cleavage by phospholipase A₂ (PLA₂) yields arachidonic acid. Cleavage by phospholipase C (PLC) yields inositol trisphosphate (I(1,4,5)P₃) and diacylglycerol (DAG). PA, phosphatidic acid; PLD, phospholipase D.

Fig. 3.12 The phosphatidylinositol (PI) cycle.

Receptor-mediated activation of phospholipase C results in the cleavage of phosphatidylinositol bisphosphate (PIP₂), forming diacylglycerol (DAG) (which activates protein kinase C) and inositol trisphosphate (IP₃) (which releases intracellular Ca²⁺). The role of inositol tetraphosphate (IP₄), which is formed from IP₃ and other inositol phosphates, is unclear, but it may facilitate Ca²⁺ entry through the plasma membrane. IP₃ is inactivated by dephosphorylation to inositol. DAG is converted to phosphatidic acid, and these two products are used to regenerate PI and PIP₂.

Inositol phosphates and intracellular calcium

Inositol (1,4,5) trisphosphate (IP₃) is a water-soluble mediator that is released into the cytosol and acts on a specific receptor—the IP₃ receptor—which is a ligand-gated calcium channel present on the membrane of the endoplasmic reticulum. The main role of IP₃, described in more detail in Chapter 4, is to control the release of Ca²⁺ from intracellular stores. Because many drug and hormone effects involve intracellular Ca²⁺, this pathway is particularly important. IP₃ is converted inside the cell to the (1,3,4,5) tetraphosphate, IP₄, by a specific kinase. The exact role of IP₄ remains unclear, but recent evidence suggests that it, and also higher inositol phosphates, plays a role in controlling gene expression.

page 34

Diacylglycerol and protein kinase C

Diacylglycerol is produced as well as IP₃ whenever receptor-induced PI hydrolysis occurs. The main effect of DAG is to activate a membrane-bound protein kinase, protein kinase C (PKC), which catalyses the phosphorylation of a variety of intracellular proteins (see Nishizuka, 1988; Walaas & Greengard, 1991). DAG, unlike the inositol phosphates, is highly lipophilic and remains within the membrane. It binds to a specific site on the PKC molecule, which migrates from the cytosol to the cell membrane in the presence of DAG, thereby becoming activated. There are 10 different mammalian PKC subtypes, which have distinct cellular distributions and phosphorylate different proteins. Most are activated by DAG and raised intracellular Ca²⁺, both of which are produced by activation of GPCRs. PKCs are also activated by phorbol esters (highly irritant, tumour-promoting compounds produced by certain plants), which have been extremely useful in studying the functions of PKC. One of the subtypes is activated by the lipid mediator arachidonic acid (see Ch. 17) generated by the action of phospholipase A₂ on membrane phospholipids, so PKC activation can also occur with agonists that activate this enzyme. The various PKC isoforms, like the tyrosine kinases discussed below (p. 37), act on many different functional proteins, such as ion channels, receptors, enzymes (including other kinases), transcription factors and cytoskeletal proteins. Kinases in general play a central role in signal transduction, and control many different aspects of cell function. The DAG–PKC link provides a channel whereby GPCRs can mobilise this army of control freaks.

The MAP kinase system

This signal transduction pathway (see below and Fig. 3.15) is activated not only by various cytokines and growth factors acting on kinase-linked receptors (see p. 37), but also by GPCR ligands. It controls many processes involved in cell division, apoptosis and tissue regeneration.

page 35

The main postulated roles of GPCRs in controlling enzymes and ion channels are summarised in Figure 3.13.

Fig. 3.13 G-protein and second messenger control of cellular effector systems.

AA, arachidonic acid; DAG, diacylglycerol; IP₃, inositol trisphosphate. Not shown in this diagram are signalling pathways where arrestins, rather than G-proteins, link GPCRs to downstream events (see text).

Desensitisation

As described in Chapter 2, desensitisation is a feature of all GPCRs, and the mechanisms underlying it have been extensively studied. Two main processes are involved (see Koenig & Edwardson, 1997; Ferguson, 2001; Kelly et al., 2008):

The sequence of GPCRs includes certain residues (serine and threonine), mainly in the C-terminal cytoplasmic tail, which can be phosphorylated by kinases such as protein kinase A (PKA), PKC and specific membrane-bound GPCR kinases (GRKs).

Phosphorylation by PKA and PKC, which are activated by many GPCRs, generally leads to impaired coupling between the activated receptor and the G-protein, so the agonist effect is reduced. These kinases are not very selective, so receptors other than that for the desensitising agonist will also be affected. This effect, whereby one agonist can desensitise other receptors, is known as heterologous desensitisation, and is generally weak and short lasting (see Fig. 3.14).

Fig. 3.14 Desensitisation of G-protein-coupled receptors (GPCRs).

Homologous (agonist-specific) desensitisation involves phosphorylation of the activated receptor by a specific kinase (GPCR kinase, GRK). The phosphorylated receptor (P-R) then binds to arrestin, causing it to lose its ability to associate with a G-protein, and to undergo endocytosis, which removes the receptor from the membrane. Heterologous (cross-) desensitisation occurs as a result of phosphorylation of one type of receptor as a result of activation of kinases by another. PKA and PKC, protein kinase A and C, respectively.

Phosphorylation by GRKs (Fig. 3.14) is receptor-specific to a greater or lesser degree, and affects mainly receptors in their activated (i.e. agonist-bound) state, resulting in homologous desensitisation. The residues that GRKS phosphorylate are different from those targeted by other kinases, and the phosphorylated receptor serves as a binding site for β-arrestins, intracellular proteins that block the interaction with G-proteins and also target the receptor for endocytosis, producing a more profound and long-lasting desensitisation. The first GRK to be identified was the β-adrenoceptor kinase, BARK, but several others have since been discovered, and this type of desensitisation seems to occur with most GPCRs. Besides initiating desensitisation, GRKs and arrestins are also involved as intermediates in various other GPCR-mediated signalling pathways that are distinct from those involving G-proteins (see Reiter & Lefkowitz, 2006). For example, binding of β-arrestin to GPCRs can recruit Src proteins, which in turn activate the MAP kinase cascade, which plays an important role in controlling cell division (see p. 39).

Further Developments in Gpcr Biology

By the early 1990s, we thought we had more or less got the measure of GPCR function, as described above. Since then, the plot has thickened, and recent developments (see review by Pierce et al., 2002) have necessitated a substantial overhaul of the basic model, whose implications for pharmacology in the future are not yet clear. Those wishing to stick to the basic story of GPCR function can safely skip this section.

Protein Phosphorylation and Kinase Cascade Mechanisms

One of the major principles to emerge over the last 10–20 years (see Cohen, 2002) is that protein phosphorylation is a key mechanism for controlling the function of proteins (e.g. enzymes, ion channels, receptors, transport proteins) involved in regulating cellular processes. Phosphorylation and dephosphorylation are accomplished by kinases and phosphatases, respectively—enzymes of which several hundred subtypes are represented in the human genome—which are themselves subject to regulation dependent on their phosphorylation status. Much effort is currently being invested in mapping the complex interactions between signalling molecules that are involved in drug effects and pathophysiological processes such as oncogenesis, neurodegeneration, inflammation and much else. Here we can present only a few pharmacologically relevant aspects of what has become an enormous subject.

In many cases, ligand binding to the receptor leads to dimerisation. The association of the two intracellular kinase domains allows a mutual autophosphorylation of intracellular tyrosine residues to occur. The phosphorylated tyrosine residues then serve as high-affinity docking sites for other intracellular proteins that form the next stage in the signal transduction cascade. One important group of such ‘adapter’ proteins is known as the SH2 domain proteins (standing for Src homology, because it was first identified in the Src oncogene product). These possess a highly conserved sequence of about 100 amino acids, forming a recognition site for the phosphotyrosine residues of the receptor. Individual SH2 domain proteins, of which many are now known, bind selectively to particular receptors, so the pattern of events triggered by particular growth factors is highly specific. The mechanism is summarised in Figure 3.15.

What happens when the SH2 domain protein binds to the phosphorylated receptor varies greatly according to the receptor that is involved; many SH2 domain proteins are enzymes, such as protein kinases or phospholipases. Some growth factors activate a specific subtype of phospholipase C (PLCγ), thereby causing phospholipid breakdown, IP₃ formation and Ca²⁺ release (see above). Other SH2-containing proteins couple phosphotyrosine-containing proteins with a variety of other functional proteins, including many that are involved in the control of cell division and differentiation. The end result is to activate or inhibit, by phosphorylation, a variety of transcription factors that migrate to the nucleus and suppress or induce the expression of particular genes. For more detail, see Pawson (2002). Nuclear factor kappa B (NFκB) is a transcription factor that plays a key role in inflammatory responses (see Ch. 17; Karin et al., 2004). It is normally present in the cytosol complexed with an inhibitor (IκB). Phosphorylation of IκB occurs when a specific kinase (IKK) is activated in response to various inflammatory cytokines and GPCR agonists. This results in dissociation of IκB from NFκB and migration of NFκB to the nucleus, where it switches on a wide variety of proinflammatory genes.

Two well-defined signal transduction pathways are summarised in Figure 3.15. The Ras/Raf pathway (Fig. 3.15A) mediates the effect of many growth factors and mitogens. Ras, which is a proto-oncogene product, functions like a G-protein, and conveys the signal (by GDP/GTP exchange) from the SH2-domain protein, Grb, which is phosphorylated by the RTK. Activation of Ras in turn activates Raf, which is the first of a sequence of three serine/threonine kinases, each of which phosphorylates, and activates, the next in line. The last of these, mitogen-activated protein (MAP) kinase, (which is also activated by GPCRs, see above), phosphorylates one or more transcription factors that initiate gene expression, resulting in a variety of cellular responses, including cell division. This three-tiered MAP kinase cascade forms part of many intracellular signalling pathways involved in a wide variety of disease processes, including malignancy, inflammation, neurodegeneration, atherosclerosis and much else. The kinases form a large family, with different subtypes serving specific roles. They are thought to represent an important target for future therapeutic drugs. Many cancers are associated with mutations in the genes coding for proteins involved in this cascade, leading to activation of the cascade in the absence of the growth factor signal (see Chs 5 and 55). For more details, see reviews by Marshall (1996), Schenk & Snaar-Jakelska (1999), Avruch (2007).

A second pathway, the Jak/Stat pathway (Fig. 3.15B) is involved in responses to many cytokines. Dimerisation of these receptors occurs when the cytokine binds, and this attracts a cytosolic tyrosine kinase unit (Jak) to associate with, and phosphorylate, the receptor dimer. Jaks belong to a family of proteins, different members having specificity for different cytokine receptors. Among the targets for phosphorylation by Jak are a family of transcription factors (Stats). These are SH2-domain proteins that bind to the phosphotyrosine groups on the receptor–Jak complex, and are themselves phosphorylated. Thus activated, Stat migrates to the nucleus and activates gene expression (see Ihle, 1995).

Other important mechanisms centre on phosphaditylinositol-3-kinase (PI₃ kinases, see Vanhaesebroeck et al., 1997), a ubiquitous enzyme family that is activated both by GPCRs and RTKs and attaches a phosphate group to position 3 of PIP₂ to form PIP₃. Other kinases, particularly protein kinase B (PKB, also known as Akt), have recognition sites for PIP₃ and are thus activated, controlling a wide variety of cellular functions, including apoptosis, differentiation, proliferation and trafficking. Akt also causes nitric oxide synthase activation in the vascular endothelium (see Ch. 20).

Recent work on signal transduction pathways has produced a bewildering profusion of molecular detail, often couched in a jargon that is apt to deter the faint-hearted. Perseverance will be rewarded, however, for there is no doubt that important new drugs, particularly in the areas of inflammation, immunology and cancer, will come from the targeting of these proteins (see Cohen, 2002). A recent breakthrough in the treatment of chronic myeloid leukaemia was achieved with the introduction of the first specific kinase inhibitor, imatinib, a drug that inhibits a specific tyrosine kinase involved in the pathogenesis of the disease (see Ch. 55).

The membrane-bound form of guanylyl cyclase, the enzyme responsible for generating the second messenger cGMP in response to the binding of natriuretic peptides (see Chs 19 and 21), resembles the tyrosine kinase family and is activated in a similar way by dimerisation when the agonist is bound (see Lucas et al., 2000).

Figure 3.16 illustrates the central role of protein kinases in signal transduction pathways in a highly simplified and schematic way. Many, if not all, of the proteins involved, including the receptors and the kinases themselves, are substrates for kinases, so there are many mechanisms for feedback and cross-talk between the various signalling pathways. Given that there are over 500 protein kinases, and similarly large numbers of receptors and other signalling molecules, the network of interactions can look bewilderingly complex. Dissecting out the details has become a major theme in cell biology. For pharmacologists, the idea of a simple connection between receptor and response, which guided thinking throughout the 20th century, is undoubtedly crumbling, although it will take some time before the complexities of signalling pathways are assimilated into a new way of thinking about drug action.

Fig. 3.16 Central role of kinase cascades in signal transduction.

Kinase cascades (e.g. those shown in Fig. 3.15) are activated by GPCRs, either directly or via different second messengers, by receptors that generate cGMP, or by kinase-linked receptors. The kinase cascades regulate various target proteins, which in turn produce a wide variety of short- and long-term effects. CaM kinase, Ca²⁺/calmodulin-dependent kinase; DAG, diacylglycerol; GC, guanylyl cyclase; GRK, GPCR kinase; IP₃, inositol trisphosphate; PKA, cAMP-dependent protein kinase; PKC, protein kinase C; PKG, cGMP-dependent protein kinase.

Structure of Nuclear Receptors

All NRs are monomeric proteins that share a broadly similar structural design (see Fig. 3.17 and Bourguet et al., 2000, for further details). The N-terminal domain displays the most heterogeneity. It harbours the AF1 (activation function 1) site that binds to other cell-specific transcription factors in a ligand-independent way and modifies the binding or activity of the receptor itself. Alternative splicing of genes may yield several receptor isoforms each with slightly different N-terminal regions. The core domain of the receptor is highly conserved and consists of the structure responsible for DNA recognition and binding. At the molecular level, this comprises two zinc fingers—cysteine- (or cystine-/histidine-) rich loops in the amino acid chain that are held in a particular conformation by zinc ions. The main function of this portion of the molecule is to recognise and bind to the hormone response elements located in genes that are regulated by this family of receptors, but it also plays a part in regulating receptor dimerisation as well.

Fig. 3.17 Nuclear receptors.

[A] Structure of a nuclear receptor, showing the different domains. [B] The two main classes of nuclear receptors. ER, oestrogen receptor; FXR, farnesoid receptor; GR, glucocorticoid receptor; HSP, heat shock protein; LXR, liver oxysterol receptor; MR, mineralocorticoid receptor; PPAR, peroxisome proliferator receptor; PR, prolactin receptor; RXR, retinoid receptor; TR, thyroid receptor; VDR, vitamin D receptor.

It is the highly flexible hinge region in the molecule that allows it to dimerise with other NRs and also to exhibit DNA binding in a variety of configurations. Finally, the C-terminal domain contains the ligand-binding module and is specific to each class of receptor. A highly conserved AF2 region is important in ligand-dependent activation. Also located near the C-terminal are motifs that contain nuclear localisation signals and others that may, in the case of some receptors, bind accessory heat shock and other proteins.

Classification of Nuclear Receptors

The NR superfamily consists of two main classes (I and II), together with a third that shares some of the characteristics of both (see Fig. 3.17 and Germain et al., 2006, for further details). Class I consists largely of receptors for the steroid hormones, including the glucocorticoid and mineralocorticoid receptors (GR and MR), as well as the oestrogen, progesterone and androgen receptors (ER, PR and AR, respectively). These receptors generally recognise hormones (e.g. glucocorticoids) that act in a negative feedback fashion to control biological events (see Ch. 32 for more details). In the absence of their ligand, these NRs are predominantly located in the cytoplasm, complexed with heat shock and other proteins and possibly reversibly attached to the cytoskeleton or other structures. Following diffusion (or possibly transportation) from the blood into the cell, their ligand partner binds to their NR with high affinity. These liganded receptors generally form homodimers and translocate to the nucleus, where they can transactivate or transrepress genes by binding to ‘positive’ or ‘negative’ hormone response elements (see below). Large numbers of genes can be regulated in this way by a single ligand. For example, it is estimated that the activated GR itself can regulate transcription of ~1% of the genome either directly or indirectly.

Class II NRs function in a slightly different way. Their ligands are generally lipids already present to some extent within the cell. This group includes the peroxisome proliferator-activated receptor (PPAR) that recognises fatty acids; the liver oxysterol receptor (LXR) that recognises and acts as a cholesterol sensor, the farnesoid (bile acid) receptor (FXR), a xenobiotic receptor (SXR; in rodents the PXR) that recognises a great many foreign substances, including therapeutic drugs, and the constitutive androstane receptor (CAR), which not only recognises the steroid androstane but also some drugs such as phenobarbital (see Ch. 43). These latter NRs are akin to airport security guards who alert the bomb disposal squad when suspicious luggage is found. They induce drug-metabolising enzymes such as CYP3A (which is responsible for metabolising about 60% of all prescription drugs; see Ch. 9 and Synold et al., 2001), and also bind some prostaglandins and non-steroidal drugs, as well as the antidiabetic thiazolidinediones (see Ch. 30) and fibrates (see Ch. 23). Unlike the receptors in class I, these NRs almost always operate as heterodimers together with the retinoid receptor (RXR). They tend to mediate positive feedback effects (e.g. occupation of the receptor amplifies rather than inhibits a particular biological event). When class II monomeric receptors bind to RXR, two types of heterodimer may be formed: a non-permissive heterodimer, which can be activated only by the RXR ligand itself, and the permissive heterodimer, which can be activated either by retinoic acid itself or by its partner’s ligand.

A third group of NRs is really a subgroup of class II in the sense that they form obligate heterodimers with RXR, but rather than sensing lipids, they play a part in endocrine signalling. The group includes the thyroid hormone receptor (TR), the vitamin D receptor (VDR) and the retinoic acid receptor (RAR).

Control of Gene Transcription

Hormone response elements are the short (four or five base pairs) sequences of DNA to which the NRs bind to modify gene transcription. They are usually present symmetrically in pairs or half sites, although these may be arranged together in different ways (e.g. simple repeats or inverted repeats). Each NR exhibits a preference for a particular consensus sequence but because of the family homology, there is a close similarity between these sequences.

Once in the nucleus, the ligand-bound receptor recruits further proteins including co-activators or co-repressors to modify gene expression through its AF1 and AF2 domains. Some of these co-activators are enzymes involved in chromatin remodelling such as histone acetylase/deacetylase which, together with other enzymes, regulate the unravelling of the DNA to facilitate access by polymerase enzymes and hence gene transcription. Co-repressor complexes are recruited by some receptors and comprise histone deacetylase and other factors that cause the chromatin to become tightly packed, preventing further transcriptional activation. Some unliganded class II receptors such as TR and VDR are constitutively bound to these repressor complexes in the nucleus, thus ‘silencing’ the gene. The complex dissociates on ligand binding, permitting an activator complex to bind. The case of CAR is particularly interesting; like some types of G-proteins described earlier in this chapter, CAR also forms a constitutively active complex that is terminated when it binds its ligand.

The discussion here must be taken only as a broad guide to the action of NRs, as many other types of interaction have also been discovered. For example, some receptors may bring about non-genomic actions by directly interacting with factors in the cytosol, or they may be covalently modified by phosphorylation or by protein–protein interactions with other transcription factors such that their function is altered (see Falkenstein et al., 2000). In addition, there is good evidence for separate membrane and other types of receptor that can bind some steroid hormones such as oestrogen (see Walters & Nemere, 2004). This intricate network of receptors and their nuclear and cytosolic interactions serves as a subtle regulator of blood lipids as well as transducing the effects of hormones that have arrived from distant tissues. Much remains to be discovered about this interesting and complex family of receptor proteins.

Voltage-Gated Channels

These channels open when the cell membrane is depolarised. They form a very important group because they underlie the mechanism of membrane excitability (see Ch. 4). The most important channels in this group are selective sodium, potassium or calcium channels.

Commonly, the channel opening (activation) induced by membrane depolarisation is short lasting, even if the depolarisation is maintained. This is because, with some channels, the initial activation of the channels is followed by a slower process of inactivation.

The role of voltage-gated channels in the generation of action potentials and in controlling other cell functions is described in Chapter 4.

Ligand-Gated Channels

These (see above) are activated by binding of a chemical ligand to a site on the channel molecule. Fast neurotransmitters, such as glutamate, acetylcholine, GABA and ATP (see Chs 13, 16 and 37) act in this way, binding to sites on the outside of the membrane. The vanilloid receptor TRPV1 mediates the pain-producing effect of capsaicin on sensory nerves (as well as responding to low pH and heat; see Ch. 41).

Some ligand-gated channels in the plasma membrane respond to intracellular rather than extracellular signals, the most important being the following:

Other examples of channels that respond to intracellular ligands include arachidonic acid-sensitive potassium channels and DAG-sensitive calcium channels, whose functions are not well understood.

Calcium Release Channels

These are present on the endoplasmic or sarcoplasmic reticulum rather than the plasma membrane. The main ones, IP₃ and ryanodine receptors (see Ch. 4) are a special class of ligand-gated calcium channels that control the release of Ca²⁺ from intracellular stores.

Store-Operated Calcium Channels

When the intracellular Ca²⁺ stores are depleted, ‘store-operated’ channels (SOCs) in the plasma membrane open to allow Ca²⁺ entry. The mechanism by which this linkage occurs involves interaction of a Ca²⁺-sensor protein in the endoplasmic reticulum membrane with a dedicated Ca²⁺ channel in the plasma membrane (see Potier & Trebak, 2008). In response to GPCRs that elicit Ca²⁺ release, the opening of these channels allows [Ca²⁺]_i to remain elevated even when the stores are running low, and also provides a route through which the stores can be replenished (see Ch. 4).