ReviewBiochemistry

Palmitoylation of Ligands, Receptors, and Intracellular Signaling Molecules

See allHide authors and affiliations

Science's STKE  31 Oct 2006:
Vol. 2006, Issue 359, pp. re14
DOI: 10.1126/stke.3592006re14

Abstract

Palmitate, a 16-carbon saturated fatty acid, is attached to more than 100 proteins. Modification of proteins by palmitate has pleiotropic effects on protein function. Palmitoylation can influence membrane binding and membrane targeting of the modified proteins. In particular, many palmitoylated proteins concentrate in lipid rafts, and enrichment in rafts is required for efficient signal transduction. This Review focuses on the multiple effects of palmitoylation on the localization and function of ligands, receptors, and intracellular signaling proteins. Palmitoylation regulates the trafficking and function of transmembrane proteins such as ion channels, neurotransmitter receptors, heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptors, and integrins. In addition, immune receptor signaling relies on protein palmitoylation at many levels, including palmitoylated co-receptors, Src family kinases, and adaptor or scaffolding proteins. The localization and signaling capacities of Ras and G proteins are modulated by dynamic protein palmitoylation. Cycles of palmitoylation and depalmitoylation allow H-Ras and G protein α subunits to reversibly bind to and signal from different intracellular cell membranes. Moreover, secreted ligands such as Hedgehog, Wingless, and Spitz use palmitoylation to regulate the extent of long- or short-range signaling. Finally, palmitoylation can alter signaling protein function by direct effects on enzymatic activity and substrate specificity. The identification of the palmitoyl acyltransferases has provided new insights into the biochemistry of this posttranslational process and permitted new substrates to be identified.

Introduction to Protein Palmitoylation

Palmitate is a 16-carbon saturated fatty acid that is covalently attached to a wide variety of cellular and viral proteins. This protein modification has piqued a great deal of interest in the signaling community because many signaling proteins are palmitoylated, and palmitoylation is important for their signaling function (15). For the vast majority of palmitoylated proteins, the fatty acid is attached posttranslationally to one or more cysteine residues through a thioester (S-acyl) linkage. "Palmitoylation" is actually a general term, as proteins can be S-acylated with fatty acids that are longer or shorter than palmitate. In addition to palmitate, proteins may be S-acylated with palmitoleate, stearate, and oleate, as well as with long-chain polyunsaturated fatty acids such as arachidonate and eicosapentanoate. Here, palmitoylation refers to the attachment of palmitate and other long-chain fatty acids to cysteine by a thioester linkage (Fig. 1). The likely donor for the reaction is intracellular palmitoyl-CoA (coenzyme A). One interesting feature of S-palmitoylation is its dynamic nature: The thioester bond between the palmitate and the protein is readily cleaved by palmitoyl thioesterases. Cycles of palmitoylation and depalmitoylation occur in a regulated manner for many proteins.

Fig. 1.

S-palmitoylation versus N-palmitoylation. (A) Chemical structures of thioester-linked (S-palmitoylation) and amide-linked (N-palmitoylation) palmitate attached to a protein. (B) Depalmitoylation of an S-palmitoylated protein leads to its release from the membrane bilayer into the cytosol. The fate of the remaining palmitoyl group is not known, but it is likely to be reacylated to form palmitoyl-CoA or another esterified protein or lipid species.

New insights into the mechanism of protein palmitoylation have been provided with the identification of palmitoyl acyl transferases (PATs), enzymes that catalyze the attachment of palmitate to protein substrates (Fig. 2). Several members of the DHHC-CRD (DHHC, Asp-His-His-Cys; CRD, cysteine-rich domain) family of proteins exhibit PAT activity toward target proteins in yeast and mammalian cells (Table 1) (612). A second set of proteins, members of the MBOAT (membrane-bound O-acyl transferase) family, function as PATs for secreted signaling proteins, such as Hedgehog (Hh), Spitz [an epidermal growth factor (EGF) molecule in flies], and Wingless (known as Wnt or Wg). For Hh and Spitz, in the mature protein the N-terminal cysteine is the site of modification; however, it is believed that the initial attachment of palmitate occurs through a thioester linkage, followed by bond rearrangement to generate an amide-linked palmitate (N-palmitoylation) (Fig. 1) (13).

Fig. 2.

Domain structure and topology of DHHC-CRD–containing PATs and MBOAT proteins. Shown schematically are a generic DHHC-CRD–containing PAT with the DHHC-CRD domain represented as a red oval (left) and a generic MBOAT protein with the MBOAT motif represented in blue. DHHC-CRD domain PATs contain 4 to 6 transmembrane domains, whereas MBOAT proteins contain 10 to 12 transmembrane domains. Four representative sequences for each type of protein are aligned. Residues identical in all four examples are in red. The cysteine residues in the DHHC-CRD domain–containing PATs are also in red, with the DHHC motif denoted by asterisks. [For a more complete list of DHHC-CRD–containing proteins, see (12).] The conserved histidine residue in MBOAT proteins that is presumed to form part of the active site is denoted by an asterisk. The protein sequences listed, along with their corresponding species, substrate(s), and GenBank accession numbers, are as follows: Erf2 (effect on Ras function), S. cerevisiae, palmitoylates yeast Ras, Q06551; Akr1 (ankyrin repeat-containing protein), S. cerevisiae, palmitoylates the yeast casein kinase Yck2, P39010; HIP14 (huntingtin interacting protein), also called DHHC17, H. sapiens, palmitoylates multiple neuronal proteins including huntingtin, Q8IUH5; DHHC9 (also called ZDHHC9 because it has a zinc finger), H. sapiens, palmitoylates H-Ras and N-Ras, Q9Y397; Rasp (named because mutant embryos resemble a coarse file), also called Ski (skinny hedgehog), Sit (sightless), or Cmn (central missing), D. melanogaster, palmitoylates Hh and Spitz, Q9VZU2; MART-2 (melanoma antigen recognized by T cells), H. sapiens, presumed human homolog of Rasp, Q5VTY9; Porc (porcupine), D. melanogaster, likely palmitoylates Wingless proteins, Q9VWV9; ACAT (acyl-coenzyme A:cholesterol acyltransferase), H. sapiens, generates cholesterol esters, NP003092.

Fig. 3.

(A and B) Involvement of palmitoylated proteins in membrane-proximal events in TCR signaling. Phosphorylation of TCR subunits by myristoylated and palmitoylated SFKs generates a binding site for the SH2 domain of ZAP-70, which is recruited to the TCR. ZAP-70 is phosphorylated and activated by SFKs. Activated ZAP-70 phosphorylates palmitoylated LAT, generating binding sites for SH2 domains of other signaling proteins, including Grb2 and PLCγ.

Fig. 4.

Schematic representation of the intracellular sites of protein palmitoylation. Ras proteins (red) are palmitoylated at the cytoplasmic surface of the ER and then traffic through the secretory pathway to the inner leaflet of the plasma membrane, where they can be depalmitoylated. Depalmitoylated Ras becomes repalmitoylated at the Golgi surface. Hh also traffics through the secretory pathway. N-palmitoylation of Hh may occur within the lumen of the ER, the Golgi, or both. The dually lipidated (cholesterol + palmitate) Hh protein is then secreted from the cell. SFKs are cotranslationally N-myristoylated on soluble polysomes in the cytosol. SFKs such as Fyn appear to traffic directly to the plasma membrane, where palmitoylation occurs. Other SFKs such as Lck traffic through the secretory pathway. Palmitoylation could occur during biosynthetic transit or at the plasma membrane.

Table 1.

DHHC-CRD palmitoyl acyl transferases (PATs) and their substrates.

Palmitoylation is typically detected by monitoring incorporation of radiolabeled palmitate (3H-palmitate) or palmitate analogs (125I-IC16 palmitate) into the protein of interest (14, 15). Alternatively, chemical labeling or mass spectrometry can be used to identify palmitoylated proteins (16, 17). With the advent of proteomic approaches, it is now possible to identify all of the palmitoylated proteins in a particular organism. This was recently exploited in a study that identified 50 palmitoylated proteins in the yeast Saccharomyces cerevisiae and matched the substrates to their respective DHHC-CRD–containing enzymes (6).

Although there is no clear consensus sequence motif for palmitoylated proteins, educated guesses can be made on the basis of cysteines previously identified as palmitoylated in other proteins (1, 3). Confirmation that a particular cysteine residue is indeed modified by palmitate is usually obtained by mutating the sequence encoding the cysteine; palmitate incorporation should be reduced or eliminated in the mutant protein. A complementary approach is to treat cells with inhibitors of palmitoylation, such as 2-bromopalmitate (2-BP) or cerulenin. This obviates the need for mutagenesis and allows one to determine the effect of palmitoylation on the native protein (18, 19).

In cases where a PAT has been identified, knockdown or knockout of the PAT can be used to determine the effect of palmitoylation on a particular protein substrate. The yeast S. cerevisiae has seven DHHC-CRD proteins. Analyses of palmitoylated protein profiles from yeast strains with single or multiple DHHC-CRD gene deletions reveal that some yeast proteins do have a dedicated PAT (6). For example, palmitoylation of amino acid permeases is abolished when the gene encoding Pfa4, a DHHC-CRD PAT, is deleted. In contrast, palmitoylation of proteins such as Ras remains even when the putative PAT (Erf2) and five of the six other PATs are deleted. These findings imply that one or more of the following is true: (i) Enzymatic redundancy exists in the DHHC-CRD family; (ii) non–DHHC-CRD proteins may function as PATs in yeast; and (iii) nonenzymatic palmitoylation may occur.

Thioester-linked palmitate can be removed from palmitoylated proteins by palmitoyl thioesterases. Two enzymes that function in human lysosomes during protein degradation have been identified: PPT1 and PPT2 (3). A third enzyme, termed APT1, has been isolated from the cytosol and shown to remove palmitate from several signaling proteins, including Gαs [heterotrimeric guanine nucleotide–binding protein (G protein) α subunits that stimulate adenylate cyclases], endothelial nitric oxide synthase (eNOS), and H-Ras (20, 21). It is not known where in the cell APT1 functions and how it recognizes its substrates.

How does palmitoylation affect signaling protein function? The answers to this question are surprisingly diverse. Palmitoylation regulates the abilities of the modified proteins to bind to lipid bilayers (membrane binding), to move around the cell (membrane trafficking), and to associate with specific membranes (membrane targeting). In addition, palmitoylation enhances the interaction of the modified proteins with lipid rafts. Rafts are subdomains of the plasma membrane that are highly enriched in cholesterol, glycosphingolipids, and a large number of signaling molecules (22, 23). As described below, membrane binding and, in many cases, lipid raft association are critically important for signaling by palmitoylated proteins, which include ligands, receptors, and intracellular signaling proteins.

Palmitoylation of Secreted Ligands

Hh and Sonic Hedgehog (Shh) are members of a family of secreted signaling proteins that mediate growth and patterning during development. These proteins act as morphogens to form signaling gradients for long- and short-range interactions (24). In Drosophila, Hh mediates pattern formation in the wing and the eye. Vertebrates express three family members: Sonic, Indian, and Desert, of which Shh is the best studied (25). Shh plays a critical role in developmental patterning of the brain in mice and humans; lack of Shh induces holoprosencephaly and cyclopia (26, 27). Shh also regulates limb development as well as cellular proliferation and differentiation in both neuronal and non-neuronal cells. In addition, Shh signaling has been implicated in biogenesis of an increasing number of human cancers (28, 29).

Hh and Shh proteins are modified by a unique set of posttranslational processing reactions. Hh and Shh are synthesized as 45-kD precursors. After the signal sequence is cleaved, Hh and Shh undergo autocleavage to generate a 20-kD N-terminal signaling molecule (called HhN and ShhN). During this reaction, the C terminus of HhN is covalently modified by a molecule of cholesterol (29, 30). In addition, the N-terminal cysteine residues of Hh and Shh are modified by palmitoylation (13). Unlike nearly all other known palmitoylated proteins, in the mature protein palmitate is attached through an amide bond to the N terminus of Hh and Shh. This reaction likely occurs in the lumen of the secretory apparatus. It is not known how palmitoyl-CoA gains access to the lumen of the ER or the Golgi to serve as a substrate for N-palmitoylation. One possibility is that the palmitate moiety that is attached to Hh and Shh is derived from a fatty acylated donor different from palmitoyl-CoA.

Attachment of both cholesterol and palmitate is important for Hh and Shh function (31). Cholesterol serves to increase the anchorage of Hh to the cell membrane and thereby restricts its diffusion and spread (3234). However, cholesterol modification of Shh is also required for long-range signaling (35, 36). Regulation of Hh and Shh movement and signaling range is believed to occur through the association of cholesterol-modified Shh and Hh and palmitoylated Shh with large protein complexes containing lipophorins (33, 37). Palmitoylation of Hh and Shh is also important for signaling. Mutation of the Shh N-terminal Cys to Ser results in a mutant Shh (Cys25 → Ser) with reduced patterning activity in the mouse limb, whereas the equivalent Hh mutant (Cys85 → Ser) has no detectable activity in Drosophila (38). In vitro differentiation assays reveal that fatty acylated forms of Shh are far more active than nonacylated Shh (39, 40).

Three groups have identified a putative Hh PAT (the FlyBase nomenclature, Rasp, is used herein) (4143). Rasp is a multipass transmembrane protein that exhibits homology to MBOAT proteins, a family of acyltransferases that attach lipophilic moieties to other proteins (44). In Rasp mutants, Hh is synthesized but not N-palmitoylated and Hh signaling is defective. Moreover, Rasp mutant flies exhibit additional defects beyond Hh signaling, suggesting the existence of additional Rasp substrates. In fact, Rasp promotes N-palmitoylation of Spitz, a membrane-tethered Drosophila EGF ligand that is cleaved intracellularly and then secreted (45). Because the mature Spitz protein contains an N-terminal cysteine residue, it is likely that the palmitate on Spitz, like that on Hh and Shh, is linked through an amide bond. Two other Drosophila EGF-like ligands, Keren and Gurken, contain N-terminal cysteines and may also be substrates for Rasp.

N-palmitoylation of Spitz is required for its activity in vivo: A nonpalmitoylated Spitz mutant cannot rescue photoreceptor differentiation in Spitz mutant flies. Likewise, Rasp is required for EGF-mediated signaling in the Drosophila wing disc and ovary (45). N-palmitoylation of Spitz enhances its association with the plasma membrane and restricts its range of activity in vivo, thereby concentrating Spitz in a sharp gradient in the vicinity of the producing cell.

The third secreted ligand known to be palmitoylated is Wnt. Wnt proteins are members of a large family of morphogens that regulate pattern formation during development. In Drosophila, Wnt-1 and Wnt-3a are palmitoylated by Porcupine (Porc), another member of the MBOAT family of acyltransferases (46, 47). Inhibition of protein palmitoylation by 2-BP or loss of Porc have similar effects on Wnt-1: Lipidation, targeting to membrane rafts, and Wnt-1 secretion are all blocked (47). These findings have led to a model in which Porc-dependent palmitoylation of Wnt-1 in the endoplasmic reticulum (ER) targets the protein to raft-enriched domains in the Golgi, where it is then packaged into specialized transport vesicles for delivery to plasma membrane rafts (47). For Wnt-3a, the situation is slightly different. Nonpalmitoylated Wnt-3a is secreted from mammalian cells. Its signaling activity is compromised relative to wild-type Wnt-3a, but can be rescued by overexpression of the mutant protein (46). Thus, palmitoylation likely regulates Wnt function at multiple levels, with effects on protein trafficking and signaling activity.

Palmitoylation of Receptors

Plasma membrane receptors already have sufficient membrane-binding information in the form of single or multipass transmembrane (TM) segments. Yet many TM receptors are palmitoylated on one or several cysteine residues at or near the TM domain. What does palmitoylation do for these proteins? This section examines the diverse effects that palmitoylation exerts on signaling by neurotransmitter receptors and channels, immune cell receptors, and G protein–coupled receptors (GPCRs) as well as their associated intracellular scaffolding and signaling partners.

Palmitoylation of ion channels and neurotransmitter receptors

Palmitoylation plays an important role in regulating ion channel localization and activity (48). This is accomplished by effects of palmitoylation on the ion channels, many of which are also receptors, as well as on the scaffolding proteins that bind to the channels. For example, palmitoylation of voltage-gated Ca2+ channels and the Kv1.1 K+ channel regulates channel localization and ion sensing activity (4952). In addition, scaffolding proteins that anchor channels to the plasma membrane are palmitoylated. The adenosine kinase–associated protein AKAP15 is a myristoylated and palmitoylated scaffolding protein that anchors protein kinase A (PKA) and Ca2+ channels (53). Fatty acylation of AKAP15 is required for its membrane localization. Membrane-bound AKAP15 then promotes PKA-mediated phosphorylation of the CaV1.2 channel, thereby increasing Ca2+ currents. Another example is provided by KChIP2, a K+ channel–interacting protein. Palmitoylation is required for localizing both KChIP2 and its associated Kv4.3 voltage-gated K+ channel at the plasma membrane (54).

In the brain, excitatory signals occur when the neurotransmitter glutamate binds to its receptors on the postsynaptic cell. Members of the family of glutamate receptors include AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionate), kainate, and NMDA (N-methyl-D-aspartate) receptors. Studies of the AMPA receptor reveal the multiple levels of regulation by protein palmitoylation. First, AMPA receptor subunits are palmitoylated on two different cysteine residues. One cysteine, located at the end of TM domain 2, is palmitoylated by GODZ (Golgi apparatus specific protein with a DHHC zinc finger domain), a Golgi-associated PAT (55). This modification serves to induce accumulation of the receptor in the Golgi and inhibit cell surface expression of the AMPA receptor. Palmitoylation of the second cysteine, located in the C-terminal tail of the AMPA receptor, modulates the rate of agonist-induced receptor endocytosis. A second level of regulation is achieved by glutamate addition, which induces AMPA receptor depalmitoylation (55). Although the effects of agonist on the individual palmitoylation sites were not determined, this dynamic palmitoylation is likely to regulate AMPA receptor trafficking.

A third level of regulation is achieved by receptor-binding proteins. AMPA receptors cycle on and off the synaptic membrane and regulate synaptic strength. To efficiently transmit signals, the receptors need to be clustered in a region of the synapse termed the postsynaptic density (PSD). AMPA receptors are recruited to the synapse indirectly by PSD-95, a synaptic PDZ domain–containing protein that functions as a scaffold to concentrate receptors in the PSD (56). PSD-95 is dynamically palmitoylated (57). The half-life of the fatty acid is 2 hours; this turnover is regulated by glutamate receptor (GluR) activity and requires Ca2+ entry through the GluR. When palmitoylation of PSD-95 is blocked, synaptic clustering of PSD-95 and the AMPA receptor GluR1 subunit are both inhibited (57, 58). The functional consequence is fewer functional AMPA receptors at the synapse, resulting in decreases in the frequency and amplitude of synaptic signaling.

In addition to PSD-95, other palmitoylated PDZ domain–containing proteins are involved in recruiting GluRs to the synapse. These include the AMPA receptor binding protein ABP and the glutamate receptor interacting protein GRIP1b (59, 60). Palmitoylation of ABP is required for localization of the protein to the plasma membrane, which is likely to be important for receptor recruitment to the synaptic membrane.

Palmitoylation of integrins and tetraspanins

Tetraspanins are a family of four TM domain proteins that regulate cell fusion, migration, and signaling. One of the notable biochemical characteristics of tetraspanins is that they associate with each other and with laminin-binding integrins to form large membrane-bound clusters termed the "tetraspanin web" or TEM (tetraspanin-enriched microdomains). Many tetraspanin proteins are palmitoylated at multiple cysteine residues; for example, CD151 has six cysteines that are sites of palmitoylation (61, 62). Likewise, integrins α3, α6, and β4 are palmitoylated; integrin β4 has seven potential cysteine palmitoylation sites (63, 64). Palmitoylation plays an important role in the ability of these two membrane-bound protein families to become incorporated into the TEM and to mediate adhesion-dependent signaling through the adaptor p130Cas and kinases of the Src family (63, 64).

Estrogen receptor palmitoylation

Steroid hormone receptors are typically viewed as transcriptional regulators functioning in the nucleus. However, a subpopulation of the estrogen receptor α (ERα) is localized at the plasma membrane. ERα is palmitoylated on a single cysteine. Palmitoylation is necessary for ERα association with the plasma membrane and for its ability to activate extracellular signal–regulated kinase (ERK) and Akt and thereby contribute to the regulation of cell proliferation (65, 66).

Palmitoylation of Immune Receptors: Influencing Signaling at Multiple Levels

One of the best examples of the importance of protein palmitoylation for cell signaling is typified by the T cell receptor (TCR). Although none of the TCR subunits themselves are palmitoylated, modification of co-receptors and downstream signaling molecules by palmitate is required for effective TCR signaling. A simplified view of how palmitoylated proteins participate in TCR signaling is presented in Fig. 3.

T cell co-receptors

CD4 and CD8 are transmembrane glycoproteins that increase T cell sensitivity to antigen. The extracellular domains of these co-receptors bind to major histocompatibility complex (MHC) class II (CD4) or class I (CD8) molecules on antigen-presenting cells, and the intracellular cytoplasmic tails bind to Lck, a kinase of the Src family. Two cysteines at the juxtamembrane region of CD4 are palmitoylated (67). Palmitoylation is required for localization of CD4 into lipid rafts, which in turn enhances raft aggregation and TCR-mediated tyrosine phosphorylation of downstream signaling proteins (68, 69).

CD8 is a heterodimer of α and β chains. The two chains can form αα homodimers [for example, in natural killer (NK) cells] or an αβ heterodimer (which occurs in thymus-derived T cells). CD8αβ can recognize antigen at lower concentrations than CD8αα. This enhanced co-receptor activity has been attributed to palmitoylation within the cytoplasmic tail of CD8αβ. Fatty acylation enables CD8αβ to partition into lipid rafts, where it activates Lck and enhances ligand-dependent TCR-mediated signaling (70, 71).

SFKs

Binding of antigen to the TCR triggers activation of the SFKs Lck and Fyn. The SFKs are membrane-bound tyrosine kinases that regulate multiple aspects of cell growth, motility, and signaling (72). After removal of the initiating methionine, all SFKs are cotranslationally modified by covalent attachment of the 14-carbon fatty acid myristate at their N termini. Because myristate alone is not sufficient to promote stable membrane binding (73), SFKs use a second signal in conjunction with myristate to achieve membrane association. For Src, the second signal is a polybasic cluster of amino acids (myristate + basic) (74). The other SFKs use myristate + palmitate, in which cysteines located at positions 3 and 5 or 6 are palmitoylated (7578).

Dual fatty acylation with myristate and palmitate is necessary and sufficient to promote localization of SFKs to the plasma membrane as well as to membrane rafts (7981). Inhibition of SFK palmitoylation, by mutation of the modified cysteine residues or by treatment with the palmitoylation inhibitor 2-BP, blocks TCR signaling (19, 82, 83). Moreover, Lck and Fyn molecules that are localized to the plasma membrane but not to membrane rafts cannot mediate signaling (19, 8385). These data establish that fatty acylation and raft association of Lck and Fyn are essential for their ability to function in TCR-mediated signaling (86).

Studies of SFKs have revealed an additional layer of complexity for protein fatty acylation. Use of radiolabeled fatty acids and matrix-assisted laser desorption/ionization–time-of-flight (MALDI-TOF) mass spectrometry revealed that, in addition to palmitate, Fyn can be heterogeneously S-acylated with palmitoleate, stearate, and oleate, and with long-chain polyunsaturated fatty acids such as arachidonate and eicosapentanoate (19, 85, 87). The functional importance of heterogeneous fatty acylation is evidenced by the finding that acylation with saturated or unsaturated fatty acids has different effects on Fyn localization and signaling. Attachment of saturated fatty acids, such as palmitate or stearate, promotes insertion of Fyn into lipid rafts. Because the lipids in rafts primarily contain saturated fatty acids and are organized in a liquid ordered phase (22, 23), it is energetically favorable for the saturated fatty acids attached to lipidated proteins to drive insertion of these modified proteins into rafts. In contrast, the presence of a cis double bond introduces a kink into the fatty acyl chain, and it is difficult to pack these unsaturated fatty acids into the highly ordered raft domain. Thus, SFKs modified by unsaturated fatty acids are excluded from rafts and are defective in mediating signaling events, such as increased Ca2+ influx and mitogen-activated protein kinase (MAPK) activation in T cells (19, 85).

Activation of both Fyn and Lck is required for propagation of TCR-mediated signaling. In resting T cells, Lck is bound to CD4 in nonraft domains, whereas Fyn is localized to membrane rafts. Engagement of the TCR and its association with CD4 leads to the following sequence of events: Lck is activated and translocates into lipid rafts, where it activates Fyn (88). If the critical event in this scenario is translocation of Lck into rafts, it should be possible to bypass the requirement for upstream TCR-CD4 interaction by directly targeting Lck to rafts (89). Indeed, a mutant Lck that cannot bind CD4 but constitutively targets to lipid rafts colocalizes with and activates Fyn. It is not clear why the CD4-Lck complex is not in rafts in resting cells, because both proteins are palmitoylated. One possibility is that regulated palmitoylation of CD4, Lck, or both occurs during T cell activation, thereby driving raft localization.

LAT

Once activated, Lck phosphorylates tyrosine residues within conserved motifs known as ITAMs (immunoreceptor tyrosine-based activation motifs) on the TCR ζ subunit. These phosphotyrosine residues recruit ZAP70 through its SH2 (Src homology 2) domains. ZAP70 is another tyrosine kinase required for TCR signaling (Fig. 3). Anchoring of ZAP-70 on the ζ chain of the TCR enhances ZAP-70 phosphorylation by Lck, thereby activating ZAP-70 kinase activity. Activated ZAP70 then phosphorylates LAT (linker for activation of T cells). LAT is an integral membrane protein with a long cytoplasmic tail containing 10 tyrosine residues. Phosphorylation of these tyrosines recruits signaling molecules to LAT through direct [Grb2, phospholipase C–γ1 (PLCγ1)] and indirect binding [phosphatidylinositide 3-kinase (PI3K)] (90). These events are important for activating downstream TCR signaling pathways, including stimulation of Ras, Ca2+ fluxes, and interleukin-2 (IL-2) production (91).

LAT is palmitoylated at two cysteines within its cytoplasmic domain. Nonpalmitoylated LAT mutants are defective in raft localization and TCR-mediated signaling (92). On the basis of these findings, raft association of LAT has been proposed to be important for effective TCR signaling (93). However, this conclusion has recently been challenged. Zhang and colleagues generated a chimeric LAT that was not palmitoylated and not targeted to rafts. This chimeric protein was functionally indistinguishable from wild-type LAT with regard to T cell signaling and function (94). It appears that as long as LAT can recruit appropriate downstream signaling molecules, it can function in or out of rafts.

Once a T cell is primed, it may not respond again when antigen is re-added. This state of affairs, known as T cell anergy, results in defective TCR-mediated proliferation and IL-2 production. One of the pathways that is compromised in anergic T cells is LAT recruitment to rafts and the immunological synapse. The molecular basis for this effect is reduced LAT palmitoylation (95). Because the palmitoylation of Fyn was not altered, it is likely that a LAT-specific PAT or palmitoyl acyl thioesterase is involved in mediating anergy.

PAG

The activation of SFKs is negatively regulated by Csk, a tyrosine kinase that is structurally similar to SFKs but lacks the N-terminal SH4 membrane-binding motif (SH4 refers to a myristate + basic or myristate + palmitate motif at the N terminus of SFKs) (74). In resting T cells, Csk is recruited to the membrane by binding to PAG (phosphoprotein associated with glycosphingolipid-enriched microdomains), a transmembrane adaptor protein. PAG is palmitoylated within a Cys-Ser-Ser-Cys sequence adjacent to its transmembrane domain (96). It is likely that palmitoylation of PAG promotes its association with membrane rafts. This would allow Csk to access raft-associated SFKs and keep them in an inactive form. Upon TCR activation, PAG is dephosphorylated by phosphatases such as CD45 or Shp2, resulting in release of Csk and activation of SFKs (97, 98).

B cell receptor signaling

Many of the same molecular principles ascribed to TCR signaling also apply to B cell receptor (BCR) signal transduction (99). Antigen binding induces BCR clustering and translocation into lipid rafts, association with and phosphorylation by the raft-associated SFK Lyn, and activation of BCR-mediated downstream signaling (100). The BCR also associates with a co-receptor complex of three proteins: CD19, CD21 (the complement receptor), and CD81 (a tetraspanin protein). Coligation of the BCR with the CD19-CD21-CD81 complex occurs when complement-tagged antigen binds simultaneously to the BCR and to CD21. Under these conditions, CD81 becomes rapidly palmitoylated (101). Palmitoylation of CD81 stabilizes the association of the BCR with the CD19-CD21-CD81 co-receptor complex in rafts, thereby promoting raft clustering into larger domains with enhanced BCR-mediated signaling.

B cells also express a transmembrane adaptor protein, NTAL [non–T cell activation linker, also known as LAB (linker for activation of B cells)], that is structurally similar to LAT (102). NTAL contains a Cys-X-X-Cys potential palmitoylation site adjacent to its transmembrane domain, although the functional importance of this site and its potential acylation has not been explored.

Fc receptors

Binding of multivalent antigen to FcεRI, the high-affinity immunoglobulin E receptor, induces mast cell degranulation, a hallmark of allergic and inflammatory responses. Like other members of the multichain immune recognition receptor family (TCR, BCR), FcεRI is dependent on a SFK, Lyn, for signal transduction. The β and γ chains of FcεRI contain ITAM motifs that are phosphorylated by Lyn. Myristoylation and palmitoylation of Lyn are required for its ability to phosphorylate FcεRI, to associate with ligand-induced FcεRI aggregates, and to mediate signal transduction (103105). It is likely that some or all of these signaling events need to occur in membrane rafts, although there is some disagreement in the literature regarding this point (103105).

Another type of Fc receptor, FcγRIIB, is a single-chain transmembrane protein whose engagement by immunoglobulins leads to phagocytosis of pathogens in monocytes, macrophages, and neutrophils. FcγRIIB association with and phosphorylation by Lyn, as well as receptor-mediated signaling, are dependent on Lyn palmitoylation and raft localization (106). The FcγRIIA receptor isoform is also palmitoylated, and mutation of the palmitoylated cysteine residue inhibits FcγRIIA translocation to rafts and its ability to induce Ca2+ mobilization (107).

Palmitoylation in GPCR and G Protein Signaling

GPCRs

The GPCRs constitute one of the largest mammalian protein families, with nearly 1000 members in the human genome. This superfamily of receptors contains seven TM domains and is activated by a wide variety of stimuli, including neurotransmitters, peptide hormones, lipids, ions, and light. Their roles in multiple aspects of human disease have made the GPCRs popular drug targets. Several GPCRs have been shown to be palmitoylated at cysteines within the C-terminal tail (108). For rhodopsin, there is structural evidence to indicate that palmitoylation results in the formation of a fourth cytoplasmic loop (109). There are two interesting features of GPCR palmitoylation. First, palmitoylation is often required for efficient delivery of the protein to the cell surface. Some nonpalmitoylated GPCR mutants are retained intracellularly, primarily in the ER (110113). This finding is consistent with the notion that GPCR palmitoylation occurs in the ER or in the ER Golgi intermediate compartment (ERGIC) (112, 114). An alternative fate for other nonpalmitoylated GPCRs is degradation, which is observed for the chemokine receptor CCR5 and the A1 adenosine receptor (110, 115). How might palmitoylation regulate protein trafficking? Tagging a receptor with palmitate may help target the modified protein to specific sites or vesicles that mediate plasma membrane targeting. It is also possible that nonpalmitoylated receptors with free cysteine sulfhydryl groups may misfold in the lumen of organelles during protein synthesis or sorting and thereby be targeted for degradation.

A second notable feature is that agonist stimulates palmitate turnover on many GPCRs. This is evidenced by an increased rate of radiolabeled palmitate incorporation that results from an increased loss of nonradioactive palmitate from the protein. Activation-induced palmitoylation and depalmitoylation cycles have been documented for the β2-adrenergic, α2A-adrenergic, m2 muscarinic acetylcholine, 5-HT4a serotonin, and V1a vasopressin receptors (108, 116120). The functional consequences of increased palmitate turnover depend on the relative rates of depalmitoylation and repalmitoylation. For some receptors, agonist treatment results in decreased overall palmitoylation levels, whereas in other cases, increased levels have been documented (108). However, the observation of increased turnover of protein-bound palmitate implies that there is dynamic generation of a pool of depalmitoylated receptors during agonist stimulation.

How does palmitoylation regulate GPCR signaling? When GPCR palmitoylation is blocked, coupling to G proteins is defective for some receptors but not others (118, 121, 122). Defective signaling may be a consequence of impaired delivery of nonpalmitoylated receptor to the cell surface. However, an additional feature of GPCR biochemistry is the ability of the receptors to become desensitized as a result of phosphorylation. For the β2-adrenergic receptor, palmitoylation appears to regulate the accessibility of PKA sites that mediate receptor desensitization; the depalmitoylated receptor is a better substrate for PKA-mediated phosphorylation (121). Likewise, phosphorylation of the bradykinin B receptor within a tyrosine-based internalization motif is mutually exclusive with palmitoylation of a cysteine residue located four amino acids away (123). Thus, it is tempting to speculate that depalmitoylation promotes GPCR phosphorylation, leading to internalization or desensitization (or both) and to decreased signaling capacity.

G protein–coupled receptor kinases (GRKs), a family of kinases that recognize agonist-bound activated receptors, can also phosphorylate GPCRs. GPCRs phosphorylated by GRKs bind to arrestin proteins, which leads to uncoupling from G proteins and receptor internalization. Palmitoylation regulates the GRK-GPCR interaction. For example, a nonpalmitoylated mutant of the thyrotropin-releasing hormone receptor exhibits reduced association with arrestin and decreased internalization in response to agonist (124). In addition, two members of the GRK family, GRK4 and GRK6, are palmitoylated. Palmitoylation of GRK6 serves two functions. First, it promotes GRK association with the membrane. Second, palmitoylation enhances kinase activity: Palmitoylated GRK6 is 10 times as active as the nonpalmitoylated protein when assayed for β2-adrenergic receptor phosphorylation in vitro (125).

Palmitoylation of G proteins and their regulators

G proteins transmit signals between GPCRs and their effectors. Signaling is dependent on localization of G protein α and βγ subunits to the cytoplasmic face of the plasma membrane. Different types of lipid modifications regulate association of the subunits with the membrane. βγ is targeted to the membrane by prenylation of the γ subunit. The α subunits achieve membrane binding by a combination of myristoylation, palmitoylation, and association with βγ. All Gα subunits except transducin are S-palmitoylated (126). In addition, Gαi,o and Gαz are N-myristoylated. There are several similarities between N-myristoylated α subunits and SFKs. For both sets of proteins, N-myristoylation is required for subsequent S-palmitoylation, and the combination of myristate + palmitate drives membrane binding of the dually modified proteins. However, several lines of evidence indicate that signals from βγ also play a role in palmitoylated α subunit localization. For example, expression of βγ can rescue membrane binding of nonpalmitoylated αz, αi, and αq (127130). Mutations that prevent the α subunit from binding to βγ inhibit membrane binding (130), and artificial targeting of βγ to the mitochondrial outer membrane causes α subunits to relocalize there as well (131). These data imply that βγ provides membrane binding and targeting information for the heterotrimeric complex.

The functional interactions among α, β, and γ subunits during G protein biosynthesis have been elucidated (132). Prenylation of γ occurs in the cytosol. βγ binds to the ER, where the C-terminal Cys-Ala-Ala-X (CAAX) motif of γ is further modified by proteolytic cleavage and carboxymethylation. The βγ dimer then proceeds to the Golgi. At the Golgi, βγ meets up with the α subunit, which is palmitoylated there. Formation of the prenylated and palmitoylated heterotrimeric αβγ complex is necessary for efficient delivery to the plasma membrane. A nonpalmitoylated α subunit can still associate with βγ in the Golgi, but the complex is not targeted to the plasma membrane. Thus, the story of G protein trafficking illustrates three important principles regarding palmitoylated proteins: (i) Two signals are required for membrane targeting (prenyl + palmitate); (ii) the signals can be delivered in trans (prenyl modification on the γ subunit, palmitate on α); and (iii) as described below, this trafficking pathway exhibits many similarities to monomeric small guanosine triphosphatases (GTPases) such as H-Ras.

In addition to enhancing membrane-binding affinity and plasma membrane targeting, palmitoylation regulates α subunit activity at multiple levels. For example, palmitoylation increases the affinity of α subunits for βγ (133). Moreover, the ability of a constitutively activated form of Gα12 to induce cell transformation is blocked when palmitoylation is inhibited (134). An additional level of complexity has been uncovered for Gαs. This subunit was originally thought to contain only one lipid modification: S-palmitoylation at Cys3. However, the N terminus, Gly2, of Gαs is also N-palmitoylated (135). N-palmitoylation not only provides additional membrane-binding affinity for Gαs, but also provides a factor of 60 to 200 higher affinity of the protein for binding to and activating adenylate cyclase. The palmitoyl acyltransferase that catalyzes N-palmitoylation of Gαs is not known, and the N-terminal sequence of Gαs does not exhibit any similarities to those of Hh or Spitz.

Palmitoylation of α subunits is dynamic and is regulated by GPCR agonists (136139). Before activation, binding of βγ to Gαs protects the α subunit from depalmitoylation. Upon ligand binding to the GPCR, α dissociates from βγ and is depalmitoylated by a palmitoyl thioesterase (20, 133). At least for Gαs, receptor activation increases palmitate turnover but does not change the stoichiometry of palmitoylation at steady state (140). What then is the importance of transient depalmitoylation? Two views have been proposed. One model postulates that depalmitoylation of Gαs leads to release of the activated subunit from the membrane to the cytosol, thereby turning off signaling (137, 141). However, another study observed depalmitoylated Gαs at the plasma membrane, concentrated in unidentified membrane subdomains (142).

G protein catalytic activity is regulated by RGS (regulator of G protein signaling) proteins, which serve as GTPase-activating proteins (GAPs) and negatively regulate G protein signaling. Palmitoylation of Gαz subunits reduces their affinity for RGS proteins and consequently reduces their rates of GTPase hydrolysis (143). Several members of the RGS family are palmitoylated (144, 145). Surprisingly, palmitoylation is not required for RGS binding to membranes (146, 147), although it differentially regulates the abilities of the various RGS proteins to inhibit G protein signaling (147, 148). Palmitoylation can exert both inhibitory and stimulatory effects on RGS4 and RGS10 GAP activity, depending on the particular palmitoylation site (149). For RGS16, palmitoylation within an N-terminal region enhances GAP activity and promotes additional palmitoylation of an internal cysteine within the conserved RGS box of RGS16 (150, 151). Palmitoylation of the RGS-box cysteine further potentiates GAP activity and the ability of RGS16 to inhibit Gαi-coupled signaling pathways.

Palmitoylation plays an additional role in RGS targeting. In an interesting study, RGS7 was shown to shuttle between the plasma membrane and the nucleus. It does so by binding to a palmitoylated shuttle protein, R7BP (152, 153). Depalmitoylation of R7BP results in the release of a R7BP-RGS7-Gβ5 complex from the plasma membrane and translocation to the nucleus, where additional signaling functions are postulated to occur.

Palmitoylated Ras Proteins: On and Off and All Around the Cell

Ras proteins are small GTPases that regulate normal and malignant cell growth and differentiation. The three major Ras isoforms H-Ras, N-Ras, and K-Ras4B are all posttranslationally modified in the cytosol by farnesylation within the C-terminal CAAX box (154). Proteolytic cleavage of the AAX sequence and carboxymethylation then occur at the surface of the ER (155). Farnesylation alone is not sufficient to anchor Ras proteins in a lipid bilayer, and thus a second membrane-binding or membrane-targeting signal is required (156, 157). For K-Ras4B, a polybasic motif serves to target the protein to the plasma membrane. In contrast, H-Ras and N-Ras are palmitoylated in the Golgi and traffic to the plasma membrane through the secretory pathway (155, 158). Ras proteins are palmitoylated by a heterodimeric complex consisting of a DHHC-CRD–containing protein complexed with either Erf4 (yeast) or GCP16 (humans) (9, 11). The lateral distribution of palmitoylated Ras proteins within the plane of the plasma membrane bilayer is regulated by the activation state of Ras. Inactive, GDP-bound Ras proteins tend to segregate in lipid rafts. Ras activation leads to redistribution of GTP-bound H-Ras into nonraft subdomains of the plasma membrane, and this event is critical for Ras-mediated signaling (159161).

Two studies have provided insights into the regulation of Ras localization and palmitoylation (162, 163). These studies showed that plasma membrane H-Ras is depalmitoylated and internalized through a nonvesicular transport mechanism back to the Golgi, where it is repalmitoylated. To test the functional importance of Ras depalmitoylation, Rocks et al. attached a fatty acid to Ras through a noncleavable thioether bond; this Ras mutant cannot be depalmitoylated (163). Curiously, the stably palmitoylated Ras protein was mislocalized and was found associated with nearly all internal cell membranes. These data argue that the reversible cycle of palmitoylation and depalmitoylation is crucial for allowing H-Ras to be properly localized in the cell. Moreover, these findings support the "kinetic bilayer trapping" hypothesis proposed by Silvius and co-workers, which postulates that the localization of the PAT determines the final localization of proteins containing myristate + palmitate or prenyl + palmitate (164166). An additional function for palmitate cycling is that it provides a constant pool of Ras on the Golgi. Golgi-associated Ras actively signals through pathways distinct from those activated by plasma membrane–bound Ras (167, 168).

Can H-Ras function with just one lipid modification—that is, either farnesyl or palmitate? For farnesyl alone, the answer is yes, but not as well. Nonpalmitoylated, prenylated H-Ras can still induce cell transformation and signaling, but at a lower efficiency relative to dually modified H-Ras (156, 157, 169). Interestingly, palmitoylation, along with a polybasic cluster, can provide full functionality to H-Ras in the absence of a farnesyl moiety. Buss and colleagues generated an activated H-Ras mutant where the C-terminal X amino acid in the CAAX motif was replaced with six lysines (170). This construct (Ext61L) was palmitoylated but not farnesylated. Ext61L H-Ras induced NIH 3T3 cell transformation as well as native, activated H-Ras did. Thus, H-Ras can function as a transforming protein with a polybasic + palmitate motif instead of prenyl + palmitate.

Why are H-Ras and N-Ras modified by farnesyl + palmitate, whereas K-Ras4B uses a farnesyl + polybasic motif? Activated H-Ras and K-Ras4B localize in different nonraft microdomains of the plasma membrane (160), and this localization likely accounts for the different signaling outputs of the two proteins. H-Ras is a better activator of PI3K than is K-Ras4B, and conversely, K-Ras4B activates Raf-1 and Rac to a greater extent than does H-Ras (171173). These differences are entirely due to the differences in the C-terminal hypervariable domains of H-Ras and K-Ras4B. Thus, the use of different membrane-targeting motifs influences the signaling specificity of the various Ras isoforms. Moreover, reversible membrane association of different Ras isoforms is regulated by different mechanisms. H-Ras is released from the membrane by depalmitoylation. In contrast, disruption of electrostatic interactions leads to dissociation of K-Ras4B from the membrane. This is accomplished by binding of calmodulin to the K-Ras4B polybasic motif or by protein kinase C–mediated phosphorylation within the K-Ras4B polybasic motif (174, 175).

How Palmitoylation Alters Protein Signaling Function: Beyond the Membrane

Three recent studies provide examples of the versatility of palmitoylation in modifying protein function beyond membrane binding and targeting.

Example 1: RPE65

The visual cycle relies on signaling by a GPCR (rhodopsin) and G proteins. Light absorption triggers isomerization of the rhodopsin-bound chromophore 11-cis-retinal to all-trans-retinal and thereby initiates visual signaling. Resynthesis of 11-cis-retinal begins with conversion of all-trans-retinal to all-trans-retinol (vitamin A). The next step is esterification of all-trans-retinol with palmitate to generate an all-trans-retinyl ester. In this reaction, the retinal protein LRAT (lecithin retinol acyltransferase) functions as a palmitoyl acyl transferase. In addition, the reaction requires RPE65 (retinal pigment epithelium 65), a protein that exists in a membrane-bound, palmitoylated form (mRPE65) and a soluble form. Rando and colleagues showed that mRPE65 functions as the palmitate donor (176). Moreover, palmitoylation generates a switch that alters ligand binding specificity. The membrane-bound, palmitoylated RPE65 binds all-trans-retinyl esters, whereas the soluble form binds all-trans-retinol. Thus, ligand binding selectivity is regulated during cycles of light and dark by protein palmitoylation.

Example 2: Htt

The Huntington protein, Htt, is a membrane-bound protein that is palmitoylated by the DHHC-CRD–containing PAT HIP14 (177, 178). Expansion of a polyglutamine tract within Htt is the cause of Huntington’s disease. Palmitoylation of Htt is regulated by the length of the polyglutamine tract: Mutant Htt containing an expanded polyglutamine tract exhibits decreased interaction with HIP14 and reduced palmitoylation. When Htt palmitoylation is inhibited (by mutation of the modified cysteine, treatment with the palmitoylation inhibitor 2-BP, or knockdown of HIP14 expression), Htt proteins aggregate and form neurotoxic inclusions. Conversely, overexpression of HIP14 reduces inclusion formation. This study illustrates the ability of palmitoylation to alter protein-protein interactions.

Example 3: Vac8

A study of Vac8, a myristoylated and palmitoylated protein, provides an example of how palmitoylation can regulate protein function (179). Vac8 is a multifunctional protein: It mediates vacuole fusion, cytosol-to-vacuole protein transport, and vacuolar inheritance in yeast. There are three palmitoylated cysteine residues within the N terminus, but they appear not to be functionally equivalent. Palmitoylation of either Cys4 or Cys5 confers membrane binding and vacuole fusion, whereas palmitoylation of Cys7 does not. Simple binding to the vacuolar membrane was not sufficient for Vac8 to function properly. Replacement of the N-terminal myristate + palmitate motif of Vac8 with the myristate + basic motif of Src resulted in a chimera that localized to the vacuolar membrane in a manner similar to wild-type Vac8. However, the Src-Vac8 fusion protein was defective in promoting vacuolar fusion and vacuolar inheritance. These activities could be rescued by reintroducing a cysteine palmitoylation site within the Src sequence. Thus, being in the right place at the right time is not sufficient for Vac8 function; it needs an additional helping hand from palmitate.

Conclusions and Perspectives

Protein palmitoylation can clearly contribute to membrane binding, membrane targeting, and membrane trafficking. These effects can be explained by multiple molecular mechanisms (Fig. 4). Membrane binding is enhanced by hydrophobic insertion of the acyl chain into the lipid bilayer. Membrane targeting may occur by kinetic trapping: For some proteins, once the protein is modified by a membrane-bound PAT, the palmitoylated protein remains trapped in that particular membrane because of the strength of the hydrophobic interaction between the palmitate moiety and the bilayer (164, 165). Thus, in some cases, the localization of the PAT determines the initial site of localization of the modified protein. Targeting of palmitoylated proteins into membrane rafts can be accounted for by the saturated nature of the fatty acyl chain. Saturated fatty acids prefer to insert into liquid ordered raft domains rather than the bulk plasma membrane (19, 85, 87).

How do we explain the ability of palmitoylation to promote protein-protein interactions? Even if two proteins interact only weakly in solution, the reduced dimensionality that results from localizing these proteins to the same membrane microdomain increases their propensity to interact. Protein-protein interactions could also occur outside of the bilayer as long as there is a mechanism to sequester the palmitate moiety. One way to accomplish this is to have a binding partner that has a palmitate binding site. For example, serum albumin contains at least seven fatty acid binding sites; myristate and palmitate bind within long hydrophobic pockets in the protein (180). Perhaps there are binding partners for palmitoylated proteins inside the cell that bind and sequester the fatty acid moiety. There is precedence for this type of interaction for prenylated proteins. Binding of geranylgeranylated Rho to Rho guanine dissociation inhibitor (GDI) involves sequestration of the Rho prenyl group within a hydrophobic pocket of RhoGDI; this interaction serves to maintain the Rho-RhoGDI complex in the cytosol (181).

Finally, it is intriguing to speculate whether palmitate exposure can be regulated. Several N-myristoylated proteins undergo a "myristoyl switch": Myristate is sequestered within a hydrophobic pocket of the protein in one state and then flipped out when a stimulus is applied (for example, ligand binding) (1, 182). To date, there are no known examples of proteins that undergo a "palmitoyl switch." However, the crystal structure of BET3, a protein involved in tethering transport vesicles to the Golgi, reveals a molecule of palmitate completely buried within a hydrophobic channel of the protein (183). This raises the possibility that a palmitoyl switch might be triggered to extrude the palmitate moiety from the protein in a regulated manner. As more palmitoylated proteins and PATs are identified, knockdown and knockout techniques will complement the biochemical studies to help elucidate the functions of palmitoylation. Moreover, the use of proteomic and mass spectrometric methods to validate protein palmitoylation of entire organisms (6) will shed additional light on this growing family of lipid-modified proteins.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
  172. 172.
  173. 173.
  174. 174.
  175. 175.
  176. 176.
  177. 177.
  178. 178.
  179. 179.
  180. 180.
  181. 181.
  182. 182.
  183. 183.
  184. 184.
  185. 185.
View Abstract

Navigate This Article