Changes in the plasma membrane in metabolic disease: impact of the membrane environment on G protein‐coupled receptor structure and function

Abstract

Drug development targeting GPCRs often utilizes model heterologous cell expression systems, reflecting an implicit assumption that the membrane environment has little functional impact on these receptors or on their responsiveness to drugs. However, much recent data have illustrated that membrane components can have an important functional impact on intrinsic membrane proteins. This review is directed toward gaining a better understanding of the structure of the plasma membrane in health and disease, and how this organelle can influence GPCR structure, function and regulation. It is important to recognize that the membrane provides a potential mode of lateral allosteric regulation of GPCRs and can affect the effectiveness of drugs and their biological responses in various disease states, which can even vary among individuals across the population. The type 1 cholecystokinin receptor is reviewed as an exemplar of a class A GPCR that is affected in this way by changes in the plasma membrane.

Linked Articles

This article is part of a themed section on Molecular Pharmacology of GPCRs. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v175.21/issuetoc

Abbreviations

CCK

cholecystokinin

CCK₁ receptor

type 1 cholecystokinin receptor

CHS

cholesteryl hemisuccinate

CRAC

cholesterol recognition/interaction amino acid consensus

DHA

docosahexaenoic acid

ER

endoplasmic reticulum

PC

phosphatidylcholine

PE

phosphatidylethanolamine

PG

phosphatidylgycerol

PI

phosphatidylinositol

PS

phosphatidylserine

PUFA

polyunsaturated fatty acids

Scap

SREBP cleavage‐activating protein

SM

sphingomyelin

SREBP‐2

sterol regulatory element‐binding protein‐2

Introduction

GPCRs are present as intrinsic plasma membrane proteins in every excitable cell in the body, where they are ideally situated to be regulated by hormones and neurotransmitters and to initiate intracellular signalling events. These heptahelical proteins are known to change their shape (Kenakin and Miller, 2010) in response to binding of agonist ligands that typically approach from the extracellular milieu, association with heterotrimeric G proteins at their cytosolic face, and even lateral allosteric regulatory events, such as association with other receptors (oligomerization) or other membrane proteins or even lipids within the bilayer. While much is currently understood about inactive and active conformations of GPCRs, particularly class A GPCRs, including complexes with G proteins (Rasmussen et al., 2011), little is understood about the influence of the membrane micro‐environment on these receptors. Insights into the impact of lipid components of the plasma membrane come predominantly from investigating receptors reconstituted into artificial membranes (Zocher et al., 2012) or from receptor mutagenesis studies that typically involve loss‐of‐function that can be interpreted in several ways (Epand et al., 2006). In this review, we discuss the characteristics and composition of the plasma membrane in health and disease, particularly focusing on the types of changes that occur in metabolic diseases where GPCRs often represent potential targets for therapy. Furthermore, we discuss the effects these changes in the membrane, that can occur as part of this disease process, have on GPCRs.

Normal plasma membrane

The cell membrane represents a basic and essential structure that has enabled the evolution of the complex animal cell. It not only functions as a semi‐permeable barrier to protect the cell and allow two‐way transport of molecules between the interior of the cell and the extracellular environment, but it also represents a complex organelle organizing and regulating receptor molecules that are essential for initiation of signalling and for physiological responsiveness. These functions are made possible by the organization of the plasma membrane as a lipid bilayer that exhibits fluidity and supports lateral mobility of intrinsic proteins (Figure 1) (Singer and Nicolson, 1972). A change in membrane fluidity has been shown to alter the functions of integral membrane receptors, such as GPCRs (Prieto et al., 1990). Phospholipids are the predominant lipid component of the bilayer, having a hydrophilic head group and two hydrophobic fatty acid chains. The different phospholipids possess distinct head groups and distinct acyl chains, which contribute to different structural and functional properties. Sphingolipids, glycolipids and sterols are also present. The sphingolipids are a class of lipids containing a sphingoid base backbone, representing aliphatic amino alcohols such as sphingosine, which are O‐linked to a charged head group and can be amide‐linked to an acyl group, such as a fatty acid. Ceramides are a class of N‐acylated sphingoid bases that lack additional head groups. Ceramide can also be converted by metabolism into more complex forms such as sphingomyelin (SM), sphingosine and glycosphingolipids. Cholesterol is a unique and functionally very important sterol component of the plasma membrane. It has a small polar head and no acyl chain, thus it contributes less to biochemical variability than the phospholipids and sphingolipids.

Figure 1.

A graphic representation of the typical plasma membrane structure and composition of an animal cell. The top panel depicts the asymmetry of the bilayer, while the bottom panels depict the liquid‐disordered and liquid‐ordered phases. The predominant lipids in the bilayer are noted, and a representative GPCR is shown in tan that spans the membrane bilayer seven times, with its N‐terminal and C‐terminal domains located extracellularly and intracellularly respectively.

The composition of the plasma membrane bilayer is asymmetrical. The inner leaflet has most of the aminophospholipids, phosphatidylserine (PS) and phosphatidylethanolamine (PE), as well as phosphatidylinositol (PI), while the outer leaflet is enriched in phosphatidylcholine (PC) and SM (Noble et al., 1999; Ray et al., 2016) (Figure 1). Cholesterol can occupy either leaflet and can interact with the lipid and protein components. Additionally, it can exhibit tail‐to‐tail interactions with itself when it is present in both leaflets. Cholesterol exhibits preferential interactions, having stronger interactions with sphingolipids than with glycerophospholipids. This probably reflects the relative rigidity of the apolar region of the sphingolipids.

When the acyl chains of the phospholipids are packed tightly together and elongated, the arrangement is called a liquid‐ordered phase, which provides a strong permeability barrier, whereas a poorly ordered hydrophobic core results in a liquid‐disordered phase with greater permeability (Maxfield and Tabas, 2005) (Figure 1). Cholesterol is positioned in the lipid bilayer with its polar hydroxyl group facing the aqueous phase and interacting with the polar heads of the phospholipids and sphingholipids, and the hydrophobic steroid ring buried inside the membrane adjacent to the non‐polar acyl chains of the phospholipids. Due to its rigid tetracyclic structure, cholesterol can modify the order and thereby the fluidity of the membrane, dependent on its location. In the typical fluid lipid membrane, this has the effect of increasing rigidity and reducing fluidity. The homeostasis of membrane fluidity tends to be maintained by changes in cholesterol being compensated for by other lipids like phospholipids, as demonstrated in mammalian and insect cells (Dawaliby et al., 2016b).

Compartmentalization and organization of lipids and proteins in the plasma membrane contribute to microdomains that support specific functions. Lipid rafts are special liquid‐ordered domains that are rich in cholesterol and sphingholipids and are more rigid and resistant to detergents than the surrounding bilayer (Pike, 2003; Ray et al., 2016). These sites seem to have an avidity for proteins with signalling and regulatory functions (Pike, 2003; Ray et al., 2016); however, the spatiotemporal aspects of the dynamic life of the lipid rafts are unclear (Edidin, 2001; Munro, 2003). Caveolae are another cholesterol‐rich specialized lipid rafts that assume flask‐like invaginations of the cell membrane that are rich in cholesterol, SM and caveolin protein, and also contain an abundance of signalling proteins (Yamada, 1955; Rothberg et al., 1992; Okamoto et al., 1998; Michel and Bakovic, 2007; Ray et al., 2016). The functions of this organelle include trafficking, signalling and endocytosis of membrane proteins (Michel and Bakovic, 2007; Ray et al., 2016) and uptake of different products of glucose and lipid metabolism (Pol et al., 2001; Ortegren et al., 2007).

The plasma membrane normally contains approximately 90% of the cholesterol in a cell (Lange et al., 1989) and 40–50% of all cellular lipids (Ray et al., 1969; Lange et al., 1989). It has recently been suggested that the cholesterol in the plasma membrane can be distributed into three pools: (i) the accessible pool that is preferentially depleted when cells are deprived of cholesterol (constitutes 16% of plasma membrane lipids); (ii) a second pool that is sequestered by SM and is sensitive to SM depletion, but not cholesterol depletion (constitutes a range from 10 to 23%, with a mean of 15%); and (iii) an inaccessible pool that represents the minimal amount of membrane cholesterol that is essential for cell health (constitutes 12% of plasma membrane lipids) (Das et al., 2014).

Cholesterol is a major regulator of lipid organization in the membrane, and its concentration is tightly regulated by feedback mechanisms. This involves its compartmentalization between the plasma membrane and the endoplasmic reticulum (ER), regulation of cholesterol uptake and the expression of transcription factors associated with de novo cholesterol synthesis and storage of excess cholesterol inside the ER as cholesteryl ester (Brown and Goldstein, 1986, 1997; Radhakrishnan et al., 2008). The key regulatory element in this homeostatic mechanism is sterol regulatory element‐binding protein‐2 (SREBP‐2), an ER membrane‐bound protein that is associated with a chaperone, SREBP cleavage‐activating protein (Scap) (Brown and Goldstein, 1997). When ER cholesterol falls below 5 mol% (Radhakrishnan et al., 2008), the SREBP‐2/Scap complex is transported to the Golgi where it undergoes a series of proteolytic cleavage events to release an active fragment that enters the nucleus and activates the transcription of genes involved in cholesterol and LDL synthesis (Brown and Goldstein, 1997). This migration of the SREBP‐2/Scap complex is negatively regulated by another ER protein, insulin‐induced gene 1, which prevents the transport of the complex to the Golgi body (Goldstein et al., 2006) when the ER cholesterol level rises above 5 mol% (Radhakrishnan et al., 2008). The LDL receptor mediates the uptake of cholesterol from the extracellular milieu via endocytosis and its movement into lysosomes, where the cholesterol is released and can be delivered to the plasma membrane and the ER (Goldstein et al., 2006). It is notable that the factor controlling the plasma membrane cholesterol content, SREBP‐2, is present in the ER (which contains ~1% of total cellular cholesterol; Lange and Steck, 1997). Recent findings using a unique cholesterol‐sensing probe have provided important insights into this phenomenon (Das et al., 2014). It has been shown that the LDL‐derived cholesterol liberated from the lysosomes first expands the accessible plasma membrane cholesterol pool and, after a brief interval, increases the ER cholesterol pool (Das et al., 2014). This suggests that the accessible pool in the plasma membrane is the predominant determinant of overall cholesterol homeostasis in a cell.

The importance of lipids in systems biology has been recognized by the introduction of a distinct discipline of lipidomics, encompassing the global analysis of all cellular lipids, their functions and information about their interacting partners (Watson, 2006). With advances in lipidomic techniques, it is becoming possible to elucidate the role of lipids in pathological states and to develop strategies to target them for therapeutic intervention (Watson, 2006; Pietilainen et al., 2011; Wu et al., 2011; Sato et al., 2012; Montastier et al., 2015; Zhang et al., 2016). Some examples of the application of this approach include observations of lipidome remodelling in the adipocyte membrane in obesity increasing the risk of inflammation (Pietilainen et al., 2011), identification of key biological and metabolic variables related to obesity, and the demonstration of tissue‐specific effects of weight loss present in adipose tissue (Montastier et al., 2015).

GPCRs and their dominant effectors, heterotrimeric G proteins

Sensitivity of GPCRs to their membrane environment

Membrane proteins such as GPCRs have evolved to function in complex micro‐environments, such as the plasma membrane. Biochemical and biophysical analyses, including high‐resolution crystal structures, have suggested many ways in which membrane lipids can tailor GPCR function to physiological needs. Receptor modulation by lipids can be divided into direct (physically interacting with the receptor) and indirect (changing the bulk properties of the membrane) interactions. The identification of distinct sites of lipid binding has been made possible by the resolution of crystal structures of membrane proteins and the recognition of sequence motifs (Li and Papadopoulos, 1998; Hanson et al., 2008). Indeed, cholesterol is present, tightly bound to GPCRs, in many crystal structures (Cherezov et al., 2007; Hanson et al., 2008; Liu et al., 2012; Wu et al., 2014; Zhang et al., 2014), and many GPCRs possess the characteristic amino acid motifs associated with the presence of cholesterol (Li and Papadopoulos, 1998; Hanson et al., 2008; Jafurulla et al., 2011; Potter et al., 2012). These consist of the cholesterol recognition/interaction amino acid consensus (CRAC) motif (Li and Papadopoulos, 1998) and the cholesterol consensus motif (Hanson et al., 2008). The presence of these motifs is suggestive, but not proof for the importance of sterols in the function and stability of specific GPCRs.

Of all the lipids, the role of cholesterol in GPCR regulation has been most extensively studied, with the effects of this lipid on various aspects of GPCR function reviewed in depth elsewhere (Paila and Chattopadhyay, 2010; Desai and Miller, 2012; Gimpl, 2016). Cholesterol can affect receptor ligand binding and signalling by altering receptor conformation, as well as affecting the lateral mobility of receptors within the bilayer that is critical for G protein coupling. Furthermore, it can also affect receptor trafficking events, including sequestration, internalization and recycling processes. Examples of receptors that are affected by changes in membrane cholesterol include the http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=29 (Hanson et al., 2008), rhodopsin (Niu et al., 2002), http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=13s (Bari et al., 2005), http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=14s (Nguyen and Taub, 2002), http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=76 (CCK₁ receptor) (Harikumar et al., 2005; Potter et al., 2012; Desai et al., 2014), http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=20s (Yu et al., 2004), http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=40s (Eroglu et al., 2003), http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=27s (Pang et al., 1999), http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=13&familyId=2&familyType=GPCR (Colozo et al., 2007), http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=360 (Monastyrskaya et al., 2005), http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=50s (Lagane et al., 2000), http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=369 (Gimpl et al., 1997) and http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=1 (Pucadyil and Chattopadhyay, 2004).

Cholesterol is particularly useful in the crystallization of GPCRs such as the β₂‐adrenoceptor (Cherezov et al., 2007), where it acts by mediating a parallel association of receptor molecules in the crystal lattice, and improving their thermal stability (Yao and Kobilka, 2005; Cherezov et al., 2007), similar to the effect of its water soluble analogue, cholesteryl hemisuccinate (CHS) (Yao and Kobilka, 2005). While much structural work has been achieved with the unique dominant retinal membrane protein, rhodopsin, the β₂‐adrenoceptor is the most extensively characterized non‐rhodopsin GPCR. The crystal structure of the β₂‐adrenoceptor shows that cholesterol fits into a shallow surface groove formed by TM1, 2, 3 and 4 (Hanson et al., 2008). Recent studies using dynamic single‐molecule force spectroscopy to examine the effect of CHS in liposomes have shown that cholesterol can allosterically alter the dynamic properties of the β₂‐adrenoceptor, significantly increasing the kinetic, energetic and mechanical stability of the receptor by increasing the free‐energy barriers stabilizing receptor segments against unfolding (Zocher et al., 2012). It is not clear how applicable this might be for other GPCRs. In another study using atomic‐scale molecular dynamic (MD) simulations, it was shown that cholesterol at ~10 mol% can constrain the distribution of β₂‐adrenoceptor conformations by binding specifically to the known high affinity binding sites and impeding movement of TM 5 and 6, and thereby preventing interchange between conformations (Manna et al., 2016).

The role of membrane phospholipids in GPCR function has also been well demonstrated (Jastrzebska et al., 2009; Inagaki et al., 2012; Dawaliby et al., 2016a). Studies with rhodopsin have shown that PC, PE and PS enhance the formation, stability and function of activated rhodopsin complexes, with the negatively charged PS having the greatest effect (Jastrzebska et al., 2009). Structural studies utilizing NMR with magnetization transfer between rhodopsin and lipid in the rod outer segment disks and artificial membranes have shown that the surface of rhodopsin possesses specific sites of interaction with phospholipids (rank order of affinity PE > PS > PC) (Soubias et al., 2006). Negatively charged phospholipids, specifically phosphatidylgycerol (PG) in reconstituted nanodiscs, have been demonstrated to cause a change in G protein coupling without affecting the agonist binding by the neurotensin‐1 receptor (Inagaki et al., 2012). This could be a direct effect of the phospholipids on the receptor or the result of an effect on the G protein or a combination of both. The most recent evidence of phospholipid regulation of a GPCR was presented for the β₂‐adrenoceptor, where modern biochemical tools demonstrated that membrane phospholipids can act as direct allosteric modulators (Dawaliby et al., 2016a). Purified β₂‐adrenoceptors reconstituted in HDL lipoparticles or nanodiscs were used to study the modulation of ligand binding and receptor activation (measuring the outward movement of TM6, an essential step for GPCR activation; Rasmussen et al., 2011), in the presence of multiple phospholipids. Specifically, PG preferentially facilitated agonist binding and receptor activation, while maintaining the active conformation for a longer time when compared with PE, PS, PI and PC (PG represents 7% of all receptor‐bound phospholipids, whereas it constitutes only 0.1% of all lipids in the insect cell line used) (Dawaliby et al., 2016a). In contrast, PI favoured antagonist binding and stabilized the inactive state of the receptor (Dawaliby et al., 2016a). A key finding from this study was that these effects could be repeated in a concentration‐dependent manner even in the absence of ligand; hence, excluding the possibility that the phospholipids might have been acting directly on the ligand; as well as in the absence of a bilayer, thereby supporting the interpretation that phospholipids act as allosteric modulators of the receptor. It has been suggested that an ionic interaction occurs between lipid head groups and receptor side chains (Dawaliby et al., 2016a), although the precise interactions are not known. This has been difficult to determine, due to the inability to trap lipid‐induced intermediate states of the receptor in crystal structures (Dawaliby et al., 2016a).

Since the experimental approaches used to define the fine structural details of GPCR‐lipid interactions in biological membranes are quite limited, a pseudo‐atomic molecular modelling approach has recently emerged (reviewed in Periole, 2017). This approach utilizes the Martini coarse grain (CG) model to define the receptors, lipids and solvents in molecular dynamics. Simulations of the 5‐HT receptor, http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=1, in a 1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine membrane bilayer revealed that the cholesterol molecule initially explores the entire receptor surface, but spends the greatest time at the CRAC motif on TM5. The durations of these interactions range from nanoseconds to up to microseconds in the case of the preferred site of interaction (Sengupta and Chattopadhyay, 2012). With the β₂‐adrenoceptor, cholesterol caused a dose‐dependent change in the interface of the receptor assembled as an oligomer (Prasanna et al., 2014). The interface of the receptor assembly changed from the TM 4/5 interface in the absence of, or at lower concentrations of, cholesterol to a dominant TM 1/2 interface in the presence of up to 50% cholesterol. The authors propose that at higher concentrations, cholesterol occupies the interaction site at the TM 4/5 interface and hence prevents its use for oligomerization. However, these data differ from the observations from the 2.8 Å crystal structure of this receptor where cholesterol was observed to directly bind and stabilize the receptor dimer at the TM1/H8 interface (Hanson et al., 2008), instead of destabilizing the TM 4/5 interface. In another simulation study involving the 5‐HT_1A receptor, this trend was reversed, favouring TM 4/5 and TM 5/6 interfaces in the presence of higher cholesterol concentrations (Prasanna et al., 2016). If correct, these simulations may explain the ability of cholesterol to differentially modulate receptor activity as a function of its location in the cell membrane. Periole suggested that data from these studies should be interpreted with caution, since the protocols did not include structural restraints to maintain the ternary structures of the receptors that can be deformed due to the Martini CG force field in the absence of an elastic network, causing significant helix reorientation (Periole, 2017).

G protein interactions with the plasma membrane

The primary effectors of GPCRs, the heterotrimeric G proteins, can also be substantially affected by the lipid composition of the plasma membrane. The formation of stable or transient non‐lamellar regions in the membrane that can be enriched in some GPCRs can exhibit increased binding of the GDP‐bound inactive Gαβγ proteins to the membrane, as shown in liposomes where PE‐enriched non‐lamellar phases increased their binding (PE is predominantly present in the inner leaflet of plasma membrane) (Escriba et al., 1997; Vogler et al., 2004). This phenomenon can also be true for the binding of http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=286 to the membrane (Escriba et al., 1997). Of note, cholesterol has been shown to have a biphasic effect on Gα binding, where lower concentrations (25–50 μM, 5–10 mol%) can cause a modest increase, and increasing concentrations to up to 30–50 mol% can impair its binding by 44% (Escriba et al., 1997). Once activated, the Gα subunit tends to change its predominant localization from the non‐lamellar to the lamellar‐ordered phase, such as that present in lipid rafts. This can result in the dissociation of Gα from Gβγ and the GPCR and the initiation of downstream signalling events (Vogler et al., 2004). Hence, the specific properties of the membrane provide an important basis for alterations in the concentration and localization of G proteins, and the differences in the movement and function between the activated G protein subunits to stimulate distinct effectors. It will be exciting to explore whether this phenomenon can contribute to temporal changes in signalling patterns, as well as biased signalling.

Given the complexity of the effect of lipids on the function of GPCRs and G proteins, it is quite possible that lipids act as allosteric modulators in vivo and regulate the function of GPCRs in health and disease. Conceivably, allosteric modulation by membrane lipids may even be responsible for differences in pharmacological function of a single GPCR in different tissues.

The plasma membrane in various pathological states

Any changes in the composition or organization of biological membranes can have a substantial impact on cellular functions and, consequently, on physiology and pathophysiology. There are two main groups of factors that affect the plasma membrane: external factors such as diet and drugs that can change the lipid profile, and internal factors that can affect cellular metabolism, de novo lipid synthesis pathways and the organization of the plasma membrane through modification of the actin cytoskeleton.

Ageing has been shown to affect the plasma membrane of leukocytes, increasing PS expression and thereby altering the symmetry and fluidity of the membrane in older subjects (Noble et al., 1999). This phenomenon has been implicated in the age‐related decline in immunity and possibly contributes to an age‐related increase in malignancies. Perhaps the most studied disease associated with changes in cholesterol and lipid metabolism is atherosclerosis (Maxfield and Tabas, 2005), where membrane bilayer properties are affected in both early and late stages. The rate limiting step in early atherogenesis is the accumulation of cholesterol in the arterial wall in the form of cholesteryl fatty‐acyl esters in plasma lipoproteins (Williams and Tabas, 1998). These lipoproteins attract monocytes and T cells by causing local inflammation and lead to the secretion of chemokines and expression of adhesion molecules (Berliner et al., 2001; Hansson, 2005). The monocytes differentiate into macrophages and ingest lipoprotein particles (Berliner et al., 2001; Hansson, 2005), causing membrane rearrangement that further enhances the uptake of cholesterol into these cells (Maxfield and Tabas, 2005; Nagao et al., 2007). The incoming cholesterol is stored in the macrophages as cholesteryl esters that coalesce into membrane‐bound neutral lipid droplets in the cytoplasm, a feature that gives the cells a ‘foamy’ appearance in early atherosclerotic lesions (Gerrity, 1981). In advanced lesions, unesterified or free cholesterol accumulates, via a number of proposed mechanisms (Maxfield and Tabas, 2005), including apoptosis and necrosis of the macrophages, and their defective clearance, leading to the exposure of luminal blood to underlying plaque material, which promotes coagulation and thrombosis.

The function of the CNS is also quite dependent on lipids. The pathophysiology of Alzheimer's disease has been associated with alterations in membrane lipids. There is a substantial increase in the relative amounts of saturated fatty acids accompanied by a parallel decrease in polyunsaturated fatty acids (PUFA) (Soderberg et al., 1991) in cell membranes in this disease. A decline in the membrane levels of the most abundant fatty acid in neuronal membranes in the cerebral cortical grey matter, the polyunsaturated http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=1051 (DHA), has been associated with advanced age and the decline in memory that is associated with Alzheimer's disease, as studied in animal models (Favrelere et al., 2000). Indeed, treatment with the synthetic 2‐hydroxy‐DHA in mouse models of Alzheimer's disease has been found to be associated with reductions in http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4865 accumulation, lipid modification of raft‐like vesicles, improved cognitive scores and improvement in neurogenesis (Torres et al., 2014).

Cancer is still another example where the lipid bilayer is potentially important. Changes in expression of lipid metabolism genes have been found to correlate with tumourigenesis and with patient survival in breast, prostate, colorectal, ovarian and brain cancers (Madhavan et al., 2009; Hirsch et al., 2010; Hilvo et al., 2011; Crous‐Bou et al., 2012). Moreover, agents capable of modifying the altered membrane composition and organization in cancer cells, and consequently, the downstream signalling events are of great therapeutic interest in this group of diseases (Martinez et al., 2005; Teres et al., 2012; Escriba et al., 2015; Torgersen et al., 2016).

The plasma membrane in metabolic disease

The most common setting for metabolic disease is obesity, which has now reached epidemic proportions in the USA and around the world (Yach et al., 2006). Obesity is known to predispose an individual to type 2 diabetes mellitus and its associated co‐morbidities, such as cardiovascular disease. Metabolic syndrome (defined by the American Heart Association as the presence of three or more of the following risk factors: waist circumference (cm) ≥ 101.6(M), 88.9 (F); triglycerides (g^.L^‐1) ≥ 1.5, HDL (g^.L^‐1) < 0.4 (M); 50 (F); blood pressure (mmHg) ≥ 130/85, fasting glucose (g^.L^‐1) ≥ 1) is becoming more common and represents a state in which the plasma membrane is abnormal, affecting receptor function.

Most studies investigating the plasma membrane in metabolic syndrome have utilized accessible circulating cells, such as erythrocytes and leukocytes. Erythrocytes have been particularly useful as a model to study membrane composition and exchange of lipids between cell and plasma, due to absence of lipoprotein receptor, intracellular organelles and de novo cholesterol synthesis.

Table 1 summarizes the various changes reported to occur in the plasma membrane in metabolic disease. Changes in membrane fluidity due to abnormalities in membrane phospholipid composition have been associated with insulin sensitivity. In erythrocytes collected from obese, non‐diabetic women, it was shown that the membrane fluidity, as measured by anisotropy of fluorescent probes, was reduced due to increases in membrane SM, and this alteration was directly proportional to the degree of insulin resistance (Candiloros et al., 1996). This was also shown to be true in adipocytes where a positive correlation was shown to exist between membrane SM, cholesterol, PE and PC content with fasting plasma insulin concentrations in obese, non‐diabetic women (Zeghari et al., 2000a,b). There are reports suggesting that the changes in erythrocyte membrane properties depend on the metabolic phenotype of the patients, such as plasma triglycerides, HDL, ApoAI and insulinaemia (Candiloros et al., 1996). A multivariate analysis was performed on 23 women with metabolic syndrome, and it demonstrated that membrane viscosity was negatively correlated with waist circumference and positively correlated with systolic BP, whereas membrane hydrophobicity was negatively correlated with post‐load glucose levels (Anichkov et al., 2005).

Table 1.

Changes in plasma membrane in metabolic syndrome

Effect	References
↑ cholesterol	(Hodis et al., 1991; Peng et al., 1991; Zeghari et al., 2000a,b; Seres et al., 2005, 2006; Hahn‐Obercyger et al., 2009; Cazzola et al., 2011)
↑ sphingomyelin	(Candiloros et al., 1996; Zeghari et al., 2000a,b; Cazzola et al., 2011; Younsi et al., 2002)
↑ phosphatidylethanolamine	(Zeghari et al., 2000a,b; Younsi et al., 2002)
↑ phosphatidylcholine	(Zeghari et al., 2000a,b)
↑ ratio cholesterol:phospholipids	(Escriba et al., 2003; Alemany et al., 2007)
↓ n‐3 fatty acids	(Field et al., 1988; Shimomura et al., 1990; Borkman et al., 1993; Matsuo et al., 1995; Hulbert et al., 2005)
↑ oxidative injury to proteins and lipids (vulnerability to peroxidation, free radical damage)	(Hodis et al., 1991; Peng et al., 1991; Cazzola et al., 2004)
↓ in caveolae density	(Howitt et al., 2012; Hahn‐Obercyger et al., 2009)
↑ rigidity	(Paragh et al., 1999, 2002; Anichkov et al., 2005; Seres et al., 2005, 2006)
↓ fluidity	(Neufeld et al., 1986; Candiloros et al., 1996; Paragh et al., 1999; Younsi et al., 2002; Cazzola et al., 2004, 2011)

Diet‐ and exercise‐induced weight loss in obese patients has been reported to have positive effects on membrane characteristics and insulin sensitivity, as reflected by an increase in membrane fluidity and an increase in insulin receptors present on monocytes (Neufeld et al., 1986). In a study investigating the membrane properties of erythrocytes from women exhibiting a loss in weight of 5% or more after an 8 week intervention, significant changes in membrane composition were observed. This was reflected as decreases in malondialdehyde, lipofuscin, cholesterol, SM (all ‘hardening’ factors), palmitic acid and nervonic acid (indicative of a reduction in SM) and increases in di‐homo‐γ‐linolenic acid, arachidonic acid and membrane fluidity (Cazzola et al., 2011). Another study involving obese, non‐diabetic women undergoing a longer intervention of 3 months for diet‐induced weight loss reported that the reduction in insulin resistance due to 5.7% weight loss correlated with a reduction in both SM and PE components of the erythrocyte membrane (Younsi et al., 2002). The altered membrane phospholipid composition in the diseased state may occur via various mechanisms, including alterations in the exchange of cholesterol and phospholipids between cell and plasma lipoproteins; the stimulation of the cellular import of phospholipids by external stimuli; and/or an effect on the synthesis of phospholipids (Younsi et al., 2002). The existence of a virtuous cycle has been proposed, where an increase in membrane fluidity initially caused by diet‐induced weight loss could give rise to a reduction in peripheral insulin resistance, which, in turn, could promote a series of metabolic events that contribute to an improvement in membrane fluidity and insulin sensitivity, resulting in an overall decrease in body weight and fat (Cazzola et al., 2011). Although this hypothesis remains untested, details of the metabolic and genetic mechanisms behind the diet‐induced changes in erythrocyte membrane composition also need to be investigated. The composition of diet including foods rich in antioxidants and absence of fried foods (Cazzola et al., 2011) may also have played a critical role in the results of this study. Of note, a weight reduction of less than 5% did not show any significant changes in erythrocyte membrane composition (Cazzola et al., 2011).

Fatty acids constitute important regulators of membrane function by contributing to the characteristics of membrane phospholipids, and much of their effect depends on the saturation of their acyl chains. Humans can synthesize saturated and monounsaturated fatty acids de novo; however, they rely on the diet to provide n‐3 and n‐6 PUFA. Although the membrane fatty acid composition is tightly regulated, it is still sensitive to changes in the fatty acid composition of the diet (Hulbert et al., 2005). Specifically, membrane changes are less affected by changes in saturated and monounsaturated fatty acids; however, they are sensitive to the changes in n‐3 and n‐6 PUFA in the diet (Hulbert et al., 2005). It was demonstrated that doubling the amount of n‐3 PUFA in the diet could yield a 24% change in n‐3 PUFA content in the composition of rat cardiac sarcolemma and a 26% change in the composition of the liver plasma membrane, reflected by a change in membrane lipid n‐3/n‐6 ratio (Hulbert et al., 2005).

http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=1800 activity has also been shown to be affected by the changes in fatty acid composition of adipocytes. When 3T3‐L1 adipocyte cells are grown in culture medium containing saturated fatty acids, insulin receptor binding and its action in vitro are decreased (Grunfeld et al., 1981). Also, a diet rich in PUFA, particularly n‐3 PUFA, has been shown to benefit insulin receptor function by increasing insulin binding and thereby increasing glucose transport (Field et al., 1988; Sohal et al., 1992; Clandinin et al., 1993), suggesting unsaturated fatty acids have a role in insulin's action. In animal studies examining the effects of a PUFA‐rich diet, there is evidence that the symptoms of diabetic animals are improved by inherently low levels of PUFA in adipocyte membranes (Field et al., 1988). Of note, the lipid‐lowering agent, bezafibrate, is linked to improved insulin sensitivity in dietary rat models of insulin resistance by causing an increase in unsaturated fatty acids and lowering the percentage of saturated and monounsaturated fatty acids of total triglycerides in skeletal muscle (Matsui et al., 1997). Studies have also shown that decreased insulin sensitivity is associated with decreased concentrations of PUFA in skeletal muscle phospholipids (Borkman et al., 1993), including changes in the skeletal muscle membrane lipid composition (Pan et al., 1995).

Animal studies have also shown that changing the fatty acid profile of the diet can alter overall energy expenditure. Rats fed n‐3 or n‐6 PUFA showed an increased metabolic rate and a lower fat content than their counterparts fed a saturated fat diet (Shimomura et al., 1990; Hulbert et al., 2005). Similarly, rats or baboons fed a saturated fat diet had the lowest metabolic rate, highest fat content and greatest weight gain when compared with those on an isocaloric diet containing n‐3 or n‐6 PUFA (Savage and Goldstone, 1965; Takeuchi et al., 1995). Similar studies in humans have shown that the PUFA‐rich diet was associated with a reduced body fat mass and increased metabolic rate (reviewed by Hulbert et al., 2005).

Additionally, a high‐fat diet has also been shown to induce chronic inflammation, so promoting insulin resistance and severe comorbidities (Olefsky and Glass, 2010; Ferrante, 2013). Recently, it was shown that endogenous fatty acid synthesis in macrophages alters the membrane lipid composition in a manner that leads to events responsible for high‐fat diet‐induced inflammation that promotes diabetes. Inactivating the fatty acid synthase enzyme in macrophages alters the membrane order and composition, impairing membrane cholesterol retention and inactivating Rho‐GTPase‐mediated cell adhesion, migration and activation, as seen in inflammation in obesity‐mediated diabetes (Wei et al., 2016). Another study utilized modern lipidomic approaches to study overall changes at different stages of acquired obesity and metabolic syndrome, illustrating adipose tissue lipid composition plays a role in the inflammation associated with obesity (Pietilainen et al., 2011). In this study, lipidomic profiling, molecular dynamics of lipid bilayers and in vitro studies confirmed that adipocyte membranes maintain their physical properties in spite of changes in membrane lipid composition during adipose expansion; however, this adaptation led to an increased vulnerability to inflammation in obesity (Pietilainen et al., 2011). Saturated fatty acid‐rich diets have been also shown to decrease β‐adrenoceptor binding by lowering the adipocyte cell membrane fluidity in rats, an effect related to decreased lipolytic activity in adipose tissue that may result in higher body fat accumulation (Matsuo et al., 1995).

Cholesterol‐rich plasma membrane domains in metabolic disease

Lipid rafts that harbour a variety of signalling molecules and also mediate the transport of substances such as glucose and fatty acids across the plasma membrane and into the cell are involved in the pathogenesis of metabolic disease. The insulin receptors and some of their signalling components are enriched in caveolae, while the absence of caveolae is correlated with insulin resistance (Stralfors, 2012). The activated insulin receptor interacts with caveolin‐1 and induces tyrosine phosphorylation of caveolin‐1 upon undergoing caveolin‐1‐dependent internalization (Liu et al., 2008; Fagerholm et al., 2009). In vitro cell studies have shown that a loss of function of caveolin‐1 negatively affects the efficacy of insulin responses (Gonzalez‐Munoz et al., 2009). Cavin peripheral membrane proteins are important components of the coat of caveolae and are critical for their biogenesis. Knocking out cavin in mouse models results in significant glucose intolerance, hyperglycaemia and hyperinsulinaemia (Liu et al., 2008). The interaction of the insulin receptor with caveolin‐1 is dependent on the lipid composition of caveolae, especially cholesterol, which is a rate‐limiting factor (Hahn‐Obercyger et al., 2009; Sanchez‐Wandelmer et al., 2009). Elevated membrane cholesterol levels in vitro have been linked to aberrant insulin receptor signalling, as indicated by impaired Akt activation and caveolin phosphorylation (Hahn‐Obercyger et al., 2009), whereas a high‐cholesterol diet has been shown to alter caveolin‐1 expression in vivo, which increases the localization of caveolae‐resident insulin receptors and insulin‐induced insulin receptor activation (Hahn‐Obercyger et al., 2009).

In addition to affecting the insulin receptor, caveolae also regulate the function of the glucose transporter, http://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=165. GLUT4 alters the blood glucose level by governing insulin‐mediated glucose uptake and is trafficked from intracellular vesicles to lipid rafts in the plasma membrane upon insulin receptor activation (Ros‐Baro et al., 2001; Karlsson et al., 2002). GLUT4 co‐localizes with lipid raft markers in muscle cells (Fecchi et al., 2006). Decreased GLUT4 expression and reduced insulin receptor expression have been observed in caveolin‐1‐deficient adipocytes (Gonzalez‐Munoz et al., 2009). Furthermore, these cells exhibited decreased insulin‐stimulated glucose uptake and GLUT4 translocation to the cell surface (Gonzalez‐Munoz et al., 2009). Taken together, the normal glucose transportation regulated by GLUT4 is highly dependent on the integrity of lipid rafts, and any disturbances in caveolae may result in hyperglycaemia.

Caveolae are also important for the ability of insulin‐sensitive adipocytes to convert fatty acids into triglycerides and to store them in central lipid droplets (Ost et al., 2005; Meshulam et al., 2011). Endogenous fatty acid production is limited in adipocytes. Adipocyte membranes are rich in caveolae invaginations that enlarge the membrane area by at least 50% (Thorn et al., 2003) and thereby provide a route for fatty acid entry and exit. Long‐chained fatty acids have been shown to physically interact with caveolae in lipid rafts of adipocytes (Pohl et al., 2004). Adipocytes mediate the rapid conversion of fatty acids into triglycerides, since excess fatty acid accumulation can cause cell lysis (Stralfors, 1990). In primary rat adipocytes, fatty acids are taken up and converted to triglycerides at the plasma membrane in a class of caveolae that are rich in the protein, perilipin (Ost et al., 2005). Furthermore, triglyceride and free fatty acid levels are significantly elevated in caveolin‐1‐null mice and this is accompanied by a reduction in the size of fat cells (Meshulam et al., 2011). In adipocytes isolated from these mice, their cellular integrity is negatively affected and this is accompanied by a modest increase in fatty acid‐induced lipolysis (Meshulam et al., 2011). In rats with diet‐induced obesity, the number and density of caveolae are altered, with a decreased number in vascular endothelium and an increased number at the ends of smooth muscle cells, phenomena linked with the abnormal endothelium‐dependent vasodilatation observed in these animals (Howitt et al., 2012).

Caveolae have also been associated with various regulators of blood pressure (Zhang, 2014). The production of nitric oxide from endothelial nitric oxide synthase is important for vascular function, and a decrease nitric oxide production results in hypertension (Zhang et al., 2008). Endothelial nitric oxide synthase is functional only when it is present on the cell surface within caveolae (Zhang et al., 2008). Caveolae are critical in endothelial shear stress‐induced vasodilatation as they mediate the production of vasodilators including nitric oxide, prostaglandins and epoxyeicosatrienoic acids (Chai et al., 2013). Caveolae are also involved in the regulation of α₁‐adrenoreceptor signalling by compartmentalizing http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2387, a substrate for receptor‐activated PLCβ1 (Morris et al., 2006). Furthermore, caveolae are involved in http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=35 (AT₂ receptor) internalization, a phenomenon postulated to be responsible for tachyphylaxis of the AT₂ receptor‐mediated contractions (Linder et al., 2007).

Oxidative stress on the membrane in metabolic disease

The production of reactive oxygen species and other radicals generated by oxidation can damage macromolecules such as proteins and lipids. There is a positive correlation between obesity and susceptibility of membrane lipoproteins to undergo peroxidation, and erythrocyte peroxidation has also been suggested to have a role in the pathogenesis of obesity‐related pathologies that involve tissue‐specific hypoxic centres such as in atherosclerosis. The intrinsic potential of erythrocytes for free radical generation makes them sensitive to oxidative stress. Oxidative injury of these cells can change their size and shape, which can lead to tissue hypoxia. Indeed, a study comparing erythrocytes from lean, overweight and obese women demonstrated that the vulnerability of erythrocytes to peroxidation and free radical‐induced damage was increased in overweight and obese subjects (BMI between 25 and 33 kg·m⁻²) compared with lean subjects (BMI < 25 kg·m⁻²) (Cazzola et al., 2004). Additionally, erythrocyte membrane fluidity was decreased, with an increase in the ratio of cholesterol to phospholipids and a decrease in the content of n‐3 fatty acids and phospholipids (Cazzola et al., 2004). An enhanced production of oxygen‐free radicals has also been observed in hypercholesterolaemia, as reflected by an increase in the cholesterol content of erythrocytes, platelets, leukocytes and endothelial cells (Hodis et al., 1991; Peng et al., 1991).

Changes in leukocyte membrane composition and rigidity have also been associated with oxidative stress in metabolic syndrome (Paragh et al., 1999; Seres et al., 2005, 2006). Increases in membrane cholesterol levels in circulating leukocytes by approximately twofold, leading to an increase in membrane rigidity, have been associated with http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=1022‐ and AT₂ receptor‐mediated oxidative burst and superoxide anion production in hypercholesterolaemia, type 2 diabetes and obesity (Seres et al., 2005, 2006).

Therapies directed at membrane lipids

As discussed, the importance of membrane properties and composition is being progressively recognized in various diseases, such as metabolic syndrome. Membrane lipid therapy is an emerging field based on the idea of designing drugs that can alter the plasma membrane properties for the treatment of diseases where the lipid bilayer plays a role in its pathogenesis. Five categories of interventions have been proposed that form the molecular basis of membrane lipid therapy (Escriba et al., 2015). These include targeting different types of regulatory effects involving the cell membrane: (i) direct binding of compounds to the membrane, thereby altering its structure; (ii) modification of activities of enzymes responsible for the synthesis or degradation of membrane components; (iii) regulation of expression of genes encoding factors that affect membrane composition; (iv) modification of protein–protein interactions involving specific membrane components and microdomains; and (v) modification of protein–lipid interactions (Escriba et al., 2015).

The use of membrane lipid therapy in oncology has been studied for chemotherapeutic agents that act via direct modulation of the plasma membrane (Triton and Yee, 1982). The investigational drug, Minerval®, a synthetic derivative of 2‐hydroxyoleic acid, is an anti‐tumour agent that has been shown to act specifically on cancer cells by normalizing the abnormal membrane lipid composition that is related to the activation of the proliferative Ras signalling events. Minerval® acts by changing the lipid order, enhancing the order of lipid rafts and up‐regulating sphingomyelin (SM) synthase 1 activity leading to increased membrane SM, which subsequently activates a cascade of signalling effectors responsible for cell cycle arrest and cell death (Martinez et al., 2005; Teres et al., 2012; Torgersen et al., 2016).

Membrane lipid therapy has also been utilized for metabolic diseases. Consumption of different fatty acids in the diet can alter the membrane lipid composition, which can be associated with lower body mass index values (Vogler et al., 2008; Lopez‐Miranda et al., 2010). High oleic acid intake has been correlated with reductions in saturated fatty acid levels and improvement in the glycaemic status of older diabetic patients (Perona et al., 2007). This has been associated with changes in the membrane lipid packing by increasing the propensity for the non‐lamellar phase, hence regulating the interaction of some G proteins relevant for metabolic signalling (Yang et al., 2005; Perona et al., 2007). Furthermore, synthetic hydroxyl‐fatty acid, 2‐hydroxyoleic acid, has been shown to reduce blood pressure in animal models, due to an increase in expression of Gs, Gq and PKC proteins, and an increase in β₂‐adrenoceptor‐mediated cAMP responses (Alemany et al., 2004). Orlistat, a derivative of the potent pancreatic lipase inhibitor lipstatin, is a fatty acid synthase inhibitor that has been shown to be effective when accompanied by lifestyle changes for the treatment of obesity, and prevention of type 2 diabetes (Torgerson et al., 2004). Cholesterol‐lowering drugs, statins, which are known to inhibit 3‐hydroxy‐3‐methyl‐glutaryl‐coenzyme A reductase, key for the synthesis of cholesterol, can also affect membrane receptors, such as up‐regulating LDL‐1 receptor activity and down‐regulation of AT₂ receptor expression (Liao and Laufs, 2005; Matarazzo et al., 2012). The reduced expression of stress proteins such as heat shock protein‐72 (HSP‐72) in patients with type 2 diabetes has been correlated with insulin resistance, and the protective effects of HSP‐72 have been documented in humans and animals (Chung et al., 2008). Hydroxamic acid and its derivatives have been shown to restore the impaired expression of stress proteins seen in type 2 diabetes (Vigh et al., 1997). BGP‐15 is an hydroxamic acid derivative, which activates the stress signal pathway by remodelling membrane rafts (Gombos et al., 2011), that was recently studied in phase 2 clinical trials for the treatment of type 2 diabetes; however, study results are not yet known.

GPCR function in abnormal plasma membrane environments in metabolic disease

Changes in membrane lipid composition and structure have been described in various pathological states that have altered GPCR‐mediated signalling. In the case of cancer, it has been shown that the anthracycline, daunorubicin (daunomycin), which is used in the treatment of solid tumours and leukaemias, reduces the formation of the non‐lamellar phase and thereby reduces the levels of membrane‐associated G proteins and PKC, as shown in heart and brain plasma membranes, and consequently disrupts oncogenic signalling (Escriba et al., 1995).

Age‐related changes in plasma membrane composition of cholesterol and phospholipid, distribution of phospholipid and fluidity can also regulate the function of certain GPCRs. These include receptors involved in growth, neurotransmission, blood pressure and metabolic health. A study utilizing erythrocytes from elderly normotensive and hypertensive subjects found that a 1.2‐fold increase in the membrane cholesterol/phospholipid ratio in hypertensive subjects is associated with a decrease in G protein and PKC levels (Escriba et al., 2003). Specifically, the membrane‐associated Gαi and Gαo levels were observed to be decreased by 31 and 38%, respectively, compared with normotensive controls, whereas Gβ and PKCα levels were reduced by 21 and 32% respectively (Escriba et al., 2003). Since it has been suggested that the Gi‐coupled α₂‐adrenoceptors control the blood pressure centrally, whereas the Gs and Gq‐coupled α₁ and β‐adrenoceptors control the blood pressure peripherally, the reduction in the G protein effector levels relevant to these receptors in hypertensive subjects could contribute to impaired adrenoceptor signalling in hypertension (Feldman, 1990; Feldman et al., 1995; Feldman and Gros, 1998; Escriba et al., 2003). Moreover, prolonged intake of oleic acid in elderly hypertensive patients has been shown to result in a reduction in blood pressure, accompanied by a reduction in erythrocyte membrane cholesterol/phospholipid ratio (Alemany et al., 2007), thereby increasing the membrane fluidity that is directly proportional to the α₂‐adrenoceptor activity (Yang et al., 2005).

The G protein‐coupled formyl‐Met‐Leu‐Phe receptor is present on the cell surface of leukocytes and recognizes the chemotactic peptide N‐formyl‐Met‐Leu‐Phe, facilitating the migration and action of these cells to neutralize invading pathogens. In patients with hypercholesterolaemia, the function of this receptor has been shown to be altered, resulting in enhanced generation of superoxides associated with the pathogenesis of atherosclerotic plaque formation (Paragh et al., 1999). In normal conditions, the signalling of this receptor on granulocytes is mediated via PLC‐induced IP₃ production and intracellular calcium elevation. However, in patients with hypercholesterolaemia, granulocytes exhibit an approximately twofold increase in membrane cholesterol, along with a decrease in membrane fluidity, which is associated with altered signalling. Here, the IP₃‐induced intracellular calcium responses were observed to be diminished, and an increase in extracellular calcium influx was observed with increased PKC activity and activation of the arachidonic acid cascade, leading to superoxide generation (Paragh et al., 1999).

Angiotensin has also been demonstrated to have a role in atherosclerosis; it stimulates plaque formation in the intima of blood vessels, as well as enhancing local endothelial inflammation and injury (Ohishi et al., 1997; Paragh et al., 2002). It was shown that the signalling mediated by AT₂ receptors in neutrophils from patients with hypercholesterolaemia is altered; the pertussis toxin‐sensitive Gi protein‐mediated IP₃ activation and calcium signalling are impaired, leading to increased production of superoxide anions (Paragh et al., 2002). Interestingly, in controls, the predominant responding receptor was found to be the http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=34, whereas the AT₂ receptor responded more strongly in hypercholesterolaemia. However, possible differences in receptor numbers were not considered. Indeed, another study by the same group, which also included patients with other components of metabolic syndrome, obesity and diabetes, reported that the membrane rigidity of neutrophils positively correlated with the AT₂ receptor‐mediated superoxide generation independently of intracellular cholesterol homeostasis (Seres et al., 2006). In this study, the membrane cholesterol of neutrophils was observed to be significantly higher only in patients with hypercholesterolaemia (~2.1‐fold), whereas the membrane rigidity was significantly increased compared to control in all three groups. Additionally, the concentration of membrane‐bound saturated fatty acids was higher by 1.17–1.25‐fold, whereas the content of the polyunsaturated fatty acid was also lower 1.78–1.88‐fold in neutrophils from all three groups when compared with control, with an overall decrease in the ratio of membrane bound‐unsaturated to saturated fatty acids. Additionally, lipid oxidation was increased in the cells from all three groups. Furthermore, a 6 week treatment with fluvastatin in clinical trials counteracted the AT₂ receptor‐mediated superoxide production in neutrophils and partially restored G protein signalling, along with decreasing the membrane cholesterol by ~0.7‐fold, without affecting the increased membrane rigidity seen in hypercholesterolaemia (Seres et al., 2005). All this evidence supports the importance of the membrane micro‐environment for the function of GPCRs in the pathogenesis of various components of metabolic syndrome.

The CCK₁ receptor as an exemplar of a metabolically important GPCR sensitive to its membrane micro‐environment

The CCK₁ receptor, a class A GPCR, is a key mediator in an important servomechanism regulating appetite. It is activated by CCK, a gastrointestinal hormone that is synthesized in the I‐cells of the small intestine and released in response to a meal (Gibbs et al., 1973; Chandra and Liddle, 2007). The role of CCK to reduce meal size has been demonstrated in animal models (Gibbs et al., 1973; Smith et al., 1981), as well as in humans (Kissileff et al., 1981; Smith and Gibbs, 1985), supporting the CCK₁ receptor as a potential target for the treatment of obesity (Bignon et al., 1999; Castillo et al., 2004; Berger et al., 2008; Elliott et al., 2010; Cameron et al., 2012). Although most studies have utilized http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=864, a form of this hormone that reduces meal size (satiation) without a consistent effect on meal frequency (satiety), more recent studies have shown that the predominant form of this hormone in the circulation is http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=3552, which reduces both meal size and meal frequency (Reeve et al., 2003; Sayegh et al., 2014). The signal for satiety is transmitted via the CCK₁ receptor present on vagal afferent neurons in the gut to the centres in the hypothalamus responsible for regulating energy balance and food intake (Smith et al., 1981; Li and Owyang, 1994; Lateef et al., 2012).

A unique feature of the CCK₁ receptor is its sensitivity to membrane cholesterol content, as described in a review by Desai and Miller (2012). In brief, this phenomenon was first recognized in patients with cholesterol gallstones, where the impaired responsiveness of the gallbladder to CCK was shown to be a function of increased membrane cholesterol in gallbladder muscularis cells, presumably caused by the transfer of cholesterol from lithogenic bile (Yu et al., 1995; Chen et al., 1997, 1999; Xiao et al., 1999, 2000, 2005). This effect has been studied in animal models (Yu et al., 1996) and humans (Yu et al., 1995; Xiao et al., 2000, 2005), where physical removal of excess cholesterol ex vivo corrected the receptor dysfunction. These observations were also replicated in vitro by modifying the membrane cholesterol content of mammalian cells using chemical and metabolic means (Harikumar et al., 2005; Potter et al., 2012; Desai et al., 2014), as well as in cell lines with mutations in cholesterol synthesis machinery (Harikumar et al., 2013).

Of note, the http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=77 that is structurally highly similar to the CCK₁ receptor and possesses the same cholesterol recognition motifs is functionally insensitive to changes in cholesterol (Potter et al., 2012; Desai et al., 2014). This has been particularly useful in creating chimeric receptor constructs and mutations to identify the structural basis of the sensitivity of CCK₁ receptors to cholesterol, and it has been localized to a CRAC motif at the cytosolic end of TM3 (Potter et al., 2012; Desai et al., 2014). Mutation of the tyrosine in position 140 of CCK₁ receptor to alanine has resulted in loss of sensitivity to cholesterol (Potter et al., 2012; Desai et al., 2014). Furthermore, this mutant has also been shown to mimic the behaviour and possibly the conformation of the wild‐type CCK₁ receptor present in a high‐cholesterol environment (Desai et al., 2014). Interestingly, structural analogues of cholesterol such as bile acids and sitosterol also affect the function of the CCK₁ receptor via this site by competing with cholesterol's action (Desai et al., 2015a, 2016). In terms of other lipids, the effect of SM on CCK₁ receptor function has also been documented (Harikumar et al., 2005), where a decrease in SM did not change CCK binding affinity or biological potency, but significantly inhibited receptor internalization and trafficking (Harikumar et al., 2005). These studies indicate that cholesterol most likely affects CCK₁ receptor function by a direct interaction, and both cholesterol and SM alter the conformation of the receptor differentially to regulate specific functions.

Since most studies related the sensitivity of the CCK₁ receptor to cholesterol have been performed using gallbladder muscularis cells, or cell lines, it has been unclear whether this effect might also be present at other sites where the CCK₁ receptor is naturally expressed, such as the vagal afferent neurons in the gut that transmit CCK‐induced satiety signals to the hypothalamus (Chaudhri et al., 2008; Dockray, 2009; Sam et al., 2012). Circulating leukocytes of a limited number of patients with components of metabolic syndrome have been demonstrated, by a single group, to contain excess membrane cholesterol (Paragh et al., 1999, 2002; Seres et al., 2005, 2006). If this phenomenon also affects vagal afferent neurons expressing the CCK₁ receptor, this could theoretically induce reduced CCK responsiveness in the satiety pathway. However, the study of CCK₁ receptors present on vagal afferent neurons in live humans presents a real challenge.

In a recent study (Desai et al., 2017), we have developed a technique that enables the study of CCK₁ receptor function in a patient's own cell membrane micro‐environment. In this work, the CCK₁ receptor was expressed ex vivo in leukocytes collected from patients using an adenoviral construct, in order to measure its function. This was achieved within 24 h of sample collection without any changes occurring in the natural membrane micro‐environment (Desai et al., 2017). Applying this technique to samples collected from 112 patients with a high incidence of obesity and metabolic disease revealed a wide variation in CCK responsiveness. Reduced CCK responsiveness correlated with higher cellular cholesterol levels in this unselected population. Interestingly, this correlation was particularly evident in subsets of the population who might be candidates for obesity pharmacotherapy, including those who were obese, those with components of metabolic syndrome and those with frank diabetes mellitus.

However, in this study (Desai et al., 2017), no standard clinical, biochemical or morphometric parameters, alone or in combination, were found to be adequate to identify those subjects exhibiting this defect in CCK responsiveness. Another key finding was that the CCK sensitivity was best correlated with different types of lipids in patients with different BMIs. In the case of the lean patients, hypertriglyceridaemia was most predictive of abnormal CCK responsiveness, whereas in the case of obese and diabetic patients, the cholesterol parameters were most predictive. Additionally, in diabetics, the strongest indicator for reduced CCK responsiveness was poor control of the disease, as reflected by elevated HbA1C levels. These observations have broad implications, suggesting that membrane components other than just cholesterol may have important regulatory effects on this GPCR.

These observations may explain why the previous clinical trials using full agonists of the CCK₁ receptor for weight loss did not reach their primary end points. Specifically, low responsiveness to CCK in some patients being studied could have been a function of this lateral allosteric modulation by membrane cholesterol or other membrane components that could have led to diminished responses to potent agonists, consequently preventing the anticipated satiety responses. Also, since the CCK₁ receptor is a pleiotropic GPCR activating multiple signalling proteins, changes in membrane lipids could also cause changes in the overall signalling profile (Miller and Desai, 2016).

A powerful approach has been proposed, which may be classified under membrane lipid therapy, is the development of positive allosteric modulators at the CCK₁ receptor that can correct the proposed defective conformation of this receptor caused by high cholesterol (Desai et al., 2015b,c; Miller and Desai, 2016). It will be important to minimize the intrinsic agonist activity of such agents in order to reduce the possibilities of side effects and a potential trophic effect associated with potent and prolonged stimulation of the CCK receptor (Dawra et al., 1993; Hoshi and Logsdon, 1993). Such a drug could potentially occupy the abnormal conformation of the receptor in a manner that allows the defective biological response of CCK in obesity to be recalibrated and could, theoretically, act only during the brief half‐life of the hormone in the circulation. Using this strategy, it might be possible to target the membrane‐induced defective conformations at different stages of the disease to safely and effectively increase CCK function mediating satiety without prolonging its effect, down‐regulating the receptor or contributing to side effects and toxicity. These agents can also be considered for use in combination with other approaches of membrane lipid therapy, as well as with pharmacological approaches involving other targets for obesity.

Conclusions

The plasma membrane can provide lateral allosteric modulation of GPCR structure and function. Multiple lipid components of the plasma membrane can affect different properties of a receptor, explaining possible differences in ligand recognition and biological responses in different disease states and in individuals across the population. It is becoming clear that the membrane composition in which a receptor is expressed should be considered when developing new drugs targeting that receptor.

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Southan et al., 2016), and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 (Alexander et al., 2015a,b,c,d).

Conflict of interest

The authors declare no conflicts of interest.

Desai, A. J. , and Miller, L. J. (2018) Changes in the plasma membrane in metabolic disease: impact of the membrane environment on G protein‐coupled receptor structure and function. British Journal of Pharmacology, 175: 4009–4025. 10.1111/bph.13943.