A novel membrane pool of protein kinase c and its role in mammalian cell signaling

1.1.2.1.2 Nuclear receptors Hydrophobic signaling molecules can diffuse directly across the plasma membrane of target cells and bind to their intracellular receptors.. All of the known

2004

Often, the signaling cell and target cell are of different cell types Sometimes cells can also send signals to themselves and other cells of the same type Extracellular signals can act on target cells over either short or long distance There are several different modes of signaling

In contact-dependent signaling, signals are exposed to the extracellular space while still tightly bound to the signaling cell’s surface In paracrine signaling, signals are secreted from the signaling cell and diffuse in the immediate environment of the signaling cell In synaptic signaling, when the dendrites of neurons receive signals, the neuron sends electrical impulses rapidly along the axon of the neuron to the end of the axon to elicit the release of neurotransmitters into the synapse and the signals are finally delivered to the postsynaptic target cell In endocrine signaling, the endocrine cells secrete signal molecules (hormones) to the blood stream that are delivered to different parts of the multicellular organism Some extracellular signals can act on themselves or other cells of the same type In autocrine signaling, the cells that secrete the signals and the cells that accept the signals are identical

Gap junctions are specific cell-to-cell junctions that can connect the cytoplasms of adjacent cells via water-filled channels

1.1.2 Cell signaling machinery

1.1.2.1 Receptors of the signals

The receptors of signals can localize in different compartments in cells and the same receptors situated in different locale may have different biological functions

1.1.2.1.1 Cell surface receptors

Hydrophilic signal molecules must bind to receptors on the cell surface, which transduce the signals Most cell surface receptors belong to one of three classes of receptors: ion-channel-linked receptors, G-protein-linked receptors, and enzyme-linked receptors1

1.1.2.1.1.1 Ion-channel-linked receptors

Ion-channel-linked receptors are involved in synaptic signaling The receptors for some transmitters themselves are ion channels Alternatively, the receptors can link with isolated ion channels When the transmitters reach the receptors in the postsynaptic membrane, the conformation of ion-channel/receptors changes, which in turn influences the flowing of ions such as Ca2+, Na+, or K+ in the postsynaptic membrane If the receptors are not ion channels, the conformational change of the receptor can affect the conformation of the linked ion channel, resulting in alteration of the flow of ions With the change of flowing of ions, the potential of target cells subsequently changes, so the signals can be transduced by ion-channel-linked receptors2

1.1.2.1.1.2 G-protein-linked receptors

All of the G-protein-linked receptors belong to a large family of homologous, seven-pass transmembrane proteins G-protein-linked receptors indirectly regulate the activity of a separate protein which is also membrane-bound The intermediate of the regulation is a third protein, called a trimeric GTP-binding protein (G-protein) The membrane-bound proteins are either ion channels or enzymes When the signals bind to the receptors in the target cells and activate the target proteins, the conformation of the enzymes or the permeability of the ion

channels are changed Subsequently, signaling proteins in cells can relay the signals to different compartments in cells3,4

1.1.2.1.1.3 Enzyme-linked receptors

There are five classes of enzyme-linked receptors; (1) receptor tyrosine kinases, (2) tyrosine kinase-associated receptors, (3) receptor serine/threonine kinases, (4) transmembrane guanylyl cyclases, and (5) histidine-kinase-associated receptors The first two classes are the most abundant in cells Enzyme-linked receptors are single-pass transmembrane proteins with an extracellular ligand-binding domain and an intracellular catalytic or enzyme-binding domain The great majority of the receptors are themselves protein kinases or are associated with kinases Binding of agonists to receptors induces a conformational change of the receptors Such conformational change leads to the activation of kinases that are either intrinsic to the receptor or associated with the receptor Thus the signals are transduced from the extracellular to the intracellular environment5,6

Besides the three groups of cell surface receptors, there are some other cell surface receptors that do not fit into the three classes There is one group of cell surface receptors that activate the signaling pathways depending on proteolysis For example, the receptor protein Notch is activated by cleavage In vertebrates, the majority of the Notch protein is targeted to the cell surface after processing by a furin-like convertase (FLC) at cleavage site 17,8 The S1 cleavage divides Notch into two polypeptides: one contains almost the entire extracellular domain and the other contains a short ectodomain stub followed by a transmembrane domain and a long cytoplasmic tail After ligand binding, a change occurs that renders the region just proximal to the membrane susceptible to cleavage by metalloproteases of the ADAM family (a disintegrin and metalloprotease), especially at a position 12 amino acids from the membrane This site is referred to as the S2 site After ligand-induced ectodomain shedding, Notch undergoes cleavage at a third site (S3) located within the transmembrane domain by an unusual enzyme called γ-secretase The freed intracellular domain enters the nucleus, where

it switches a DNA-bound corepressor complex into an activating complex, leading to activation of selected target genes 9,10

1.1.2.1.2 Nuclear receptors

Hydrophobic signaling molecules can diffuse directly across the plasma membrane of target cells and bind to their intracellular receptors These agonists include steroid hormones, thyroid hormones, retinoids, and vitamin D Upon binding to the receptors, they activate the receptors that can bind to specific DNA sequences adjacent to the target genes to regulate the transcription of such genes The receptors are structurally related and are part of the nuclear receptor family Some receptors that are activated by intracellular metabolites are also included in this family Some nuclear receptors, such as those for cortisol, are located in the cytoplasm, and translocate to the nucleus after ligand binding, while other nuclear receptors, those for thyroid hormone and retinoids, are bound to DNA even in the absence of ligands After the binding of hormones to the receptors, the inhibitory protein complexes bound to the inactive receptors dissociate and the receptors bound to coactivator proteins induce gene transcription11

1.1.2.1.3 Intracellular enzymes as receptors

Nitric oxide gas (NO) and carbon monoxide (CO) can rapidly diffuse across the membrane and bind to iron in the active form of guanylyl cyclase, which stimulates this enzyme to produce the small intracellular mediator cyclic GMP The cyclic GMP can induce responses

in target cells, for example, keeping blood vessels relaxed12

1.1.2.2 Intracellular signaling molecules

After extracellular signaling molecules bind to the receptors, the signals are relayed into the cell interior by a combination of small and large intracellular signaling molecules The former are second messengers; the latter are called intracellular signaling proteins including

G proteins, small GTPase and protein kinases13-15

1.2 Protein kinases

1.2.1 History and Definition

The presence of phosphorus in proteins has been known since the early part of the last century The function of phosphoproteins was considered to be mainly nutritional16 In the early 1940s, Cori and his colleagues showed that the enzyme glycogen phosphorylase exists

in two forms: an inactive form called phosphorylase b and an active form termed phosphorylase a These two forms of the enzyme are interconvertible in intact cells17,18 It was not until the 1950s that the mechanism of this interconversion was discovered when two independent groups reported that enzymes of phosphorylation-dephosphorylation were involved19,20 Based on their discoveries, the inactive phosphorylase b can be activated by phosphorylation catalyzed by a newly isolated enzyme, phosphorylase kinase (a protein kinase) Phosphorylase a can be inactivated by dephosphorylation catalyzed by a specific phosphatase Many kinases have been discovered in the past 50 years It is now universally accepted that the reversible phosphorylation of proteins regulates most aspects of cell life21-26 Protein kinases and protein phosphatases provide the basis of a network that allows extracellular signals to coordinate biochemical functions within the target cells27-30 It is therefore not surprising that a large number of protein kinases exist It seems likely that in mammalian cells over 500 different kinase molecules have evolved to form the requisite signal transducing networks31 In contrast, a limited number of protein phosphatases are present in the cell With the full appreciation that protein kinases and protein phosphatases are equally important in signal transduction and integration within the cells, however, this research focuses on protein kinases

Protein kinases are defined as enzymes that transfer a phosphoryl group from a phosphate donor onto an acceptor amino acid in a substrate protein Most kinases use ATP as the phosphoryl group donor although several kinases also utilize GTP The protein kinases, which catalyze the phosphorylation step in various phosphorylation-dephosphorylation systems, can be divided into two main classes, the serine-threonine protein kinases and the

tyrosine protein kinases based on the phosphorylated residues in the substrate protein Each class can be subdivided into groups or entities depending on the nature of the agents that regulate activity

1.2.2 Classification and superfamily of protein kinases

The eukaryotic protein kinases make up a large superfamily of homologous proteins The kinase domains that define this group of enzymes contain 12 conserved subdomains that fold into a common catalytic core structure32 In the mid’s of 1990, phylogenetic trees derived from an alignment of kinase domain amino acid sequences served as the basis for the classification by Hanks and Hunter Protein kinases can be divided into 4 groups: (1) the AGC group; (2) the CaMK group; (3) the CMGC group; (4) conventional protein tyrosine kinase (RTK) group Recently, the near completion of the human genome sequence allows the identification of almost all human protein kinases32 The classification of protein kinases has been extended from the Hanks and Hunter’s four broad groups, 44 families and 51 subfamilies to adding four new groups including the STE group, CK1 group, TTBK group and TKL group encompassing 90 families and 145 subfamilies31 In humans, a total of 518 protein kinases have been identified including 478 human eukaryotic protein kinases and 40 atypical protein kinases The protein kinases constitute about 1.7 % of all human genes31 According to the type of amino acid phosphorylated, the protein kinases can be divided into two subgroups: protein serine/threonine kinases and protein tyrosine kinases

Serine/threonine protein kinases were the first discovered kinases that catalyze the phosphorylation of hydroxyamino acids All of the known protein kinases including tyrosine kinases share a related catalytic domain of about 260 amino acids and are distinguished by their unique regulatory domains32

Tyrosine kinases are the other major kinases in the protein kinase family Protein tyrosine phosphorylation, first encountered in 1979 as an activity of viral transforming gene products33, was quickly recognized to play an important role in transducing growth factor receptor signals across the plasma membrane With time it has become clear that tyrosine

phosphorylation also regulates protein function in many other cellular processes including the cell cycle, transcription, and synaptic transmission Tyrosine kinases also have a catalytic domain of about 260 amino acids just like serine/threonine kinases A number of residues within this domain are highly conserved between both types of protein kinases, but the tyrosine kinases and serine/threonine kinases are distinguished by specific signature motifs34 The constitution of major serine/threonine kinases and tyrosine kinases is shown in Figure1.1

Figure 1.1 Major members of the serine/threonine kinase family and tyrosine kinase family (adapted

from Hanks S.K, et al.34 )

In addition to serine/threonine kinases and tyrosine kinases, there are reports on other types

of protein kinases that phosphorylate proteins on histidine35,36 or on arginines37

1.3 Protein kinase C superfamily

The protein kinase C superfamily belongs to the AGC family of serine-threonine kinases38,39 PKC was originally discovered by Nishizuka and coworkers as a proteolytically activated kinase40,41 However, it was subsequently demonstrated that PKC was reversibly activated by

a lipid-soluble, membrane-associated factor in the presence of Ca2+42,43 Diacylglycerol (DAG) was later shown to be the neutral lipid that activated PKC in a phosphatidylserine (PS)- and

Ca2+-dependent manner44-46 This discovery linked PKC to major signal transduction pathways involving phosphatidylinositol turnover PKC is now known to be an increasingly large family with multiple subspecies, which represents one of the major classes of downstream targets for lipid second messengers PKCs are involved in the regulation of cell proliferation, differentiation, and immunity and are implicated in the development of diseases such as cancer, neurodegenerative diseases, heart attack and diabetes47

1.3.1 PKC family members

After 27 years of study, exhaustive genetic screening has defined a superfamily of mammalian PKC of at least 14 isozymes The mammalian PKC isotypes have been grouped into 4 subfamilies on the basis of their biochemical properties and sequence homology: (1) conventional PKCs (cPKC) containing α, βІ, βII, γ48; (2) novel PKCs (nPKC) including ε, η,

δ, θ49; (3) atypical PKCs containing ι/λ, ζ (aPKC)50, and (4) protein kinase C-related kinases (PRKs) 1-3 Sequence comparisons place PKC isotypes into 5 subgroups (Figure 1.2) including the newly discovered PKCζII which does not contain a catalytic domain51 These groupings are almost identical to the classification based on biochemical properties, with the only exception being the splitting of the novel PKCs into two pairs of very closely related kinases, namely δ, θ and ε, η Closer examination of protein sequence alignments between PKC isotypes reveals the presence of blocks of homology between family members These conserved regions confer a specific localization and/or activation input to different isozymes

Figure 1.2 Dendrogram based on a sequence comparison of the PKC superfamily Protein sequences

of the fully-cloned members of the human PKC superfamily were compared using Clustal V software with PAM 250 residue tables ( adapted from Ponting, C.P & Parker, P.J 52 ).

1.3.2 Structure of PKCs

All PKC isozymes, except for PKCζII, consist of a single polypeptide chain with an terminal regulatory domain and a C-terminal catalytic domain interspersed by a hinge region (the V3 domain)50,53 The carboxyl-terminal domains of all these PKC molecules are the catalytic domains containing consensus sequences for ATP- and substrate-binding sites34,54 The amino-terminal domains of PKC are quite diverse and are referred to as regulatory

N-domains of PKCs containing Ca2+ (where applicable), phospholipid and DAG/phorbol ester binding sites The primary structure of PKCs, on the basis of amino acid sequence homology, consists of variable regions, V1-V5, interspersed by conserved domains, C1-C4 For the cPKCs, they have classical C1 and C2 domains55 However, for nPKCs and aPKCs, their C1 and C2 domains have either different sequences or different orders of these two domains (Figure 1.3) cPKCs and nPKCs contain a C1 domain that is defined by the presence of two repeated zinc-finger motifs, C1A and C1B55

Figure 1.3 Domain structures of the PKC subfamilies The Figure shows a comparison of the protein architecture of the various subgroups of the PKC superfamily The C1 domain of the aPKCs is

smaller than that of the cPKCs and nPKCs as it contains only one copy of the zinc-finger motif

(adapted from Ponting, C.P & Parker, P.J 52 )

Mutational and deletion analysis has provided evidence that the C1 domain is the binding site for phorbol ester49,56, and this has been confirmed by the solution of the crystal structure of

the second zinc-finger motif of PKCδ complexed with PMA/TPA57 Binding studies have shown that DAG competes with PMA for binding to PKC and the two molecules are therefore assumed to interact with PKC at the same site58 Only the C1B domain binds phorbol ester in the case of PKCδ59; however, both the C1A and C1B domains bind phorbol

esters in vitro in the case of PKCγ60 The C1 domain is absent from PRK1 and aPKCs contain only a single zinc-finger motif so that they are not responsive to phorbol ester The C1 domain also contains a pseudosubstrate sequence, which presumably is responsible for maintaining the enzyme in an inactive form in the absence of activators61 The pseudosubstrate sequence resembles substrate recognition motifs of the same kinase The pseudosubstrate sequence can interact directly with the substrate binding site in the catalytic domain on the same PKC molecule, thereby sterically blocking protein substrate access to the active site Activation of protein kinases occurs by dissociation of the regulatory domain from the catalytic domain For PRKs, which lack the C1 domain, the HR1a motif is proposed

to act as a pseudosubstrate site to inhibit kinase activity62 The C2 domain is found in the cPKCs immediately C-terminal to the C1 domain C2 domains are present in many other proteins, including the synaptotagmins, rabphilin-3A, phospholipases and GAPs52 As many

of these proteins bind phospholipids in a Ca2+-dependent manner, it has been assumed that the C2 domain confers calcium/phosphatidylserine binding to the cPKCs The classical C2 domain is missing from the Ca2+-independent PKCs (nPKCs and aPKCs), but the V0 domain

of nPKCs and aPKCs and the HR2 domain of the PRKs are related to the C2 domain through the investigation of the sequence of the proteins Due to the differences in structural identity and composition of C1 and C2 domains, cPKCs can be regulated by both diacylglycerol and calcium, while nPKCs can only be regulated by diacylglycerol; aPKCs and PRKs can be regulated neither by calcium nor diacylglycerol Furthermore, the primary structure of PKCs also includes other important domains, for example, HR1 (homology region-1) domain, V5 domain These domains are related to the stability, modulation of activity and subcellular localization of PKCs The HR1 domain, initially identified as a region of homology between

PRK1 and PRK2, is composed of three repeat motifs63 A single copy of this repeat is found

in two other RhoA-binding proteins, rhophilin64 and rhotekin65 The first HR1 repeat in PRK1 (HR1a) binds to the active GTP-RhoA complex, but not to the inactive GDP-RhoA66 The second HR1 motif also binds to RhoA, but weakly and not strictly to the active RhoA It has also been reported that the HR1b of PRK1 can weakly interact with RhoB67 The HR1c repeat does not bind to Rho68 and its function is unknown The V5 region of PKC plays an important regulatory role in kinase function For PKCβI and βII, the V5 region plays a critical role in subcellular localization69 For cPKCs and other PKCs, there are two phosphorylation sites in the V5 domain which need to be phosphorylated At least one of these sites of phosphorylation has been confirmed to be a site of autophosphorylation The binding partners for the different domains in PKCs can control subcellular localization, the activity and the function, of the kinases For example, the C1 domain can bind to diacylglycerol, leading to the release of the pseudosubstrate domain and subsequent activation of the cPKC and nPKC57 As for the C2 domain, it prefers anionic membranes to zwitterionic ones The role of calcium is to provide a bridge between the C2 domain and anionic phospholipids70 and to induce intra- or interdomain conformational changes71 The members of the PKC superfamily have similar as well as distinct structural features that dictate their subcellular localization and functions in mammalian cells

1.3.3 Biochemical properties of PKC

Certain features of the PKC isozymes can be deduced from the protein structures derived from the sequence of the cloned PKC cDNAs However, the greater body of information has been accumulated by detailed biochemical analyses of naturally occurring and recombinant forms of individual isozymes The biochemical properties of the different PKC isozymes

have been investigated in detail in both in vivo and in vitro systems with respect to activation,

autophosphorylation, substrate specificity and proteolytic degradation

1.3.3.1 Activation

1.3.3.1.1 PKC activation in vivo by membrane translocation

PKC is readily recovered as a soluble protein from resting cells homogenized in the presence

of EDTA72 Upon treatment of cells with agonists leading to the stimulation of phosphatidylinositol-specific phospholipase C activity, a redistribution of PKC from cytosol

to the plasma membrane is observed Treatment of parietal yolk sac cells with phorbol ester elicits a rapid decrease in cytosolic PKC that is accompanied by a significant increase in the level of plasma membrane-associated enzyme73 A rise in either internal Ca2+ or DAG can cause cytosol-membrane translocation of PKC74,75 These results established a temporal correlation between the intracellular distribution of PKC and its putative activation state Phorbol esters are potent activators of PKCs They can mimic the effects of the natural activator of PKCs, diacylglycerol76 It is widely recognized that phorbol esters are promoters

of skin tumors, with pleiotropic and often tissue specific cellular responses such as proliferation77, differentiation78, apoptosis79 and gene expression80 PKC has been identified

as the high-affinity intracellular receptor of phorbol esters Phorbol ester activates PKC in

binding to PKC competitively, indicating that the sites of interaction for these activators are similar if not identical58 Since phorbol esters have more than 1000-fold higher affinity than DAG for PKC in the presence of lipid82, they have been used extensively in many PKC assays

1.3.3.1.2 Lipid-induced PKC activation

Pseudosubstrate sequences within the intramolecular regulatory domain have been identified

in PKCs These sequences provide a means of autoinhibitory activity of proteins by interacting with the catalytic domain83,84 The interaction between specific lipids or other activators and the regulatory domain of the PKCs leads to conformational changes that release autoinhibitory restraints imposed by the pseudosubstrate domain and induce PKC

activation85 Some lipid activators of PKC are illustrated in the following segments including diacylglycerol, phosphatidyl-L-serine, and fatty acids

1.3.3.1.2.1 Diacylglycerol (DAG)

cPKCs and nPKCs are activated by DAG and phorbol esters in the presence of phosphatidylserine (PS)49,86 Activation of PKC by DAG analogues is very stereo-specific:

the sn-1,2-DAGs, but not the 1,3- or 2,3-DAGs are effective in assays using either lipid

vesicles or mixed micelles87,88 Both the oxygen esters (1,2-position) and the primary hydroxyl group (3-position) are essential for DAG function The acyl chain composition does not appear to be critical although a preference for unsaturated side chains has been reported42,44,89 The major requirement with respect to acyl chains appears to be their length

In vitro studies showed that DAG with both acyl chains containing six or more carbon atoms

induces maximal kinase activity, indicating that DAGs must be sufficiently hydrophobic to associate with membranes or micelles90-92 Activation of PKC by DAGs is believed to result from two effects elicited by DAGs First, the affinity of PKC for PS increases in the presence

of DAG and strengthens PKC association with the membrane93,94 Secondly, DAG together with PS induces the conformational changes that expose the pseudosubstrate domain95

1.3.3.1.2.2 Phosphatidyl-L-Serine (PS)

It has been almost 20 years since Nishizuka and coworkers reported that the pro-enzyme of the protease-activated kinase (PKC) was activated by PS96 PS is the phospholipid which is most effectively able to activate PKC97,98 and is thought to represent the principal physiological phospholipid cofactor The high affinity of PKC for PS in mixed micelles requires DAG DAG can increase the affinity of PKC for PS 93 Similarly, Ca2+ increases the affinity of PKC for PS 93 The activation response curve of PKC by PS is sigmoidal PS concentrations of 15 mol % of total lipid in the plasma membrane are sufficient to totally activate PKC in the presence of saturating amounts of Ca2+ and DAG PKC interacts with multiple phosphatidylserine molecules The actual number of PS molecules that interact with

PKC has recently been calculated to be on the order of 8 Thus, PKC interaction with PS can cause extensive segregation of acidic phospholipids99

1.3.3.1.2.3 Other phospholipids

Besides PS, several other negatively charged phospholipids such as phosphatidylinositol (PI), phosphatidic acid (PA), phosphatidylglycerol (PG) and cardiolipin (CL) are also capable of supporting PKC activity although they are less effective43,92 In contrast to the highly specific interaction between PKC and PS in the presence of DAG, the interaction between PKC and other phospholipid effectors appears to be due to less specific electrostatic effects, reflecting the capacity of acidic lipids in general to bind to PKCs100 Various PKC isozymes have distinguishable responses to these phospholipids in some instances For example, cardiolipin

is a strong activator of PKCγ and PKCε101,102 The precursors of DAG, 4,5-biphosphate (PIP2), phosphatidylinositol-3,4,5-triphosphate (PIP3) and phosphatidylinositol-3, 4-biphosphate have been reported to be potent and selective activators of PKCδ, ε, η103,104 PIP3 was also implicated in the activation of PKCζ105 Recent evidence demonstrates that PKCζ can be activated by ceramide, which is produced by

phosphatidylinositol-stimulation of sphingomyelin hydrolysis catalyzed by sphingomyelinase, both in vitro and in vivo106 These data suggest that individual PKC isozymes can be regulated by more than one lipid and they may have separate and unique functions in the cell

A2, while DAG is produced by phospholipase C-catalyzed hydrolysis of phosphatidylinositol

and phosphatidylcholine or via prior formation of phosphatidic acid through phospholipase D-catalyzed hydrolysis114-116 Three major groups of phospholipase C, PLCβ, PLCγ, and PLCδ, are identified in mammalian tissues, and each group consists of more than one isoform The enzymes are activated upon cell stimulation and cleave inositol phospholipids to produce inositol phosphates and DAG needed for PKC activation At a later phase of the cellular response, phospholipase D plays a major role in the production of DAG from PC to prolong PKC activation117 Lysophosphatidylcholine (lysoPC) is the product of PC hydrolysis by PLA2 LysoPC together with DAG can significantly strengthen subsequent cellular responses118 DAG produced from PI hydrolysis is rapidly converted to phosphatidic acid (PA) by the action of DAG kinase, and further to PI, PIP, and PIP2 PA produced from PC is converted to DAG by the action of phosphomonoesterase116,119,120 On the other hand, the DAG produced from PC hydrolysis is a poor substrate for DAG kinase, and is slowly degraded by DAG lipase121

In summary, PKC activation is a convergent point for signal-induced, membrane phospholipid hydrolysis products formed by the activation of phospholipase A2, C and D122and in specialized cases may also be regulated by sphingolipid metabolites106,123 These phospholipases may be activated by different stimuli at different locations in cells

1.3.3.2 Posttranslational processing and maturation

All known PKC isozymes can autophosphorylate86,124-131 Detailed studies of the role of autophosphorylation have mostly been carried out with cPKC isozymes and have revealed that it is an intramolecular reaction at both serine and threonine residues on both regulatory and catalytic domains132-134 Autophosphorylation of PKC has a Km value for ATP about 10-fold lower than that for substrate phosphorylation132 and shows a dependence on PS and DAG similar to that observed for phosphorylation of exogenous substrates135,136 The last step for PKC maturation is the autophosphorylation of serine/threonine residues in the turn motif and hydrophobic motif After this step, are the PKCs ready to be activated by activators such

as DAG137 The competence of members of the PKC family to respond to lipid second

messengers depends on a series of ordered phosphorylation events138,139 The kinase domain

of PKCs contains three conserved phosphorylation sites: the activation loop (XTXTFCG) which is phosphorylated by phosphoinositide-dependent kinase (PDK1)140,141, and two carboxyl-terminal sites, the turn motif and hydrophobic FXXFS/TF/Y motif which are regulated by autophosphorylation in conventional PKCs142 Phosphate at the activation loop unmasks the substrate binding cavity and structures the enzyme for catalysis, with phosphates

at the two carboxyl-terminal positions then locking the enzyme in this catalytically competent conformation Newly synthesized PKCs are in a conformation that promotes binding of PDK-1 to its carboxyl terminus and in which the activation loop is exposed for phosphorylation Thus, the primary regulation of this phosphorylation event is the direct interaction of PKC with PDK1 Although phosphorylation at the activation loop is required for the maturation of PKC, once the enzyme is phosphorylated at the C-terminal sites, the activation loop phosphate is dispensable143 The fully phosphorylated species of PKC is the mature form that is competent to respond to lipid second messengers

It has been reported that protamine can stimulate autophosphorylation of PKC144,145, which is consistent with the role of protamine as an unconventional activator of PKC41 Protamine is thought to induce a conformational change that releases the pseudosubstrate domain from the active site, thereby leading to PKC activation85 Protamine sulfate phosphorylation occurs in the absence of phosphatidylserine, diacylglycerol, and calcium96 Protamine sulfate alters the structure of PKC in the same way as lipid activators, thus effecting the same change in function: activation Protamine sulfate can expose Arg19 of PKCβII’s pseudosubstrate domain

to proteolysis In summary, autophosphorylation, as one of the immediate early events of PKC activation, can be used as a potential measure of activation146,147

1.3.3.3 Substrate specificities

Numerous substrate specificity studies on PKCs have revealed that PKCs have a broad

substrate specificity in vitro118,137 The use of synthetic peptide substrates has indicated that all PKC preparations require basic residues surrounding the target serine or threonine61,148,149

with the phosphorylation site motif being R-X-X-S/T-X-R150 A comparison of PKC

isozymes and their activities towards distinct substrates in vitro has revealed that they also

show differences in substrate phosphorylation, in addition to differences in cofactor/activator dependence For cPKC, PKCα and PKCβ have a very similar pattern of specificity, while PKCγ exhibits a strong preference for substrates with a C-terminal basic amino acid151 For PKCδ, the substrate preference is MBP4-14 peptide ≈ EGFR peptide>>peptide ε in the presence of cardiolipin, whereas in the presence of PS, the preferences are peptide ε>>MBP4-14 peptide ≈ EGFR peptide152 These data clearly indicate that the substrate specificities of PKC isozymes can possibly be modulated by the binding of different cofactors/activators to their regulatory domain and that the choice of substrates in combination with phospholipids can also alter the way in which the enzyme responds to Ca2+ The variations in substrate specificity and the different activation requirements of PKC family members suggest distinct regulatory functions of these enzymes An important factor that underlies isotype selectivity appears to be the ability of the substrate to compete with the pseudosubstrate for occupation of the substrate binding pocket The catalytic cleft is lined with residues that are variable in the various PKCs, suggesting differences between isotypes

in the strength of the pseudosubstrate inhibition which in turn may lead to differences in substrate accessibility A considerable amount of experimental data indicates that PKC

isotypes differ in their substrate preferences, the bulk of which relies on in vitro evidence These are not sufficient to explain the observed in vivo substrate specificity, because in cells

the localization of the kinases and substrates sometimes can decide the accessibility of the kinases to substrates, and thus decide the substrate specificity

1.3.3.4 Proteolytic activation and degradation

PKC was first discovered as a protease-activated protein kinase40,153 The inactive PKCs can

be converted to an active form by limited proteolysis by calcium-dependent protease Ca2+dependent proteolysis of PKCs is thought to be catalyzed by calpains74,154 Calpains cleave PKC in the V3 hinge region and produce two distinct fragments, a regulatory fragment and a

-catalytically active fragment, known as PKM153 It is believed that in vivo activation of PKC

and translocation to the cell membrane is a prerequisite for proteolytic cleavage However, it

is not yet clear that the final destiny of the cleavage is to initiate further degradation or to release the active fragment to the cytosol155 The other model of degradation is the vesicle-associated model in which certain PKC isotypes phosphorylate an element of the endocytic machinery, leading to the stimulation of endocytosis In this condition, membrane-associated PKCs remain attached to vesicles and are transported to the cell’s general degradative compartments (lysosomes or proteosomes), where destruction occurs156 In many cell types, prolonged treatment with phorbol esters results in complete depletion of cellular PKC, so called down-regulation157 Down-regulation is the prevention of permanent kinase activity It

is not yet clear whether substrate phosphorylation or autophosphorylation is involved in the process of down-regulation It is possible that PKC isozymes in different cells may be subjected to differential down-regulation because of the difference in their susceptibilities to proteolysis and unique subcellular localizations158

1.3.3.5 Tissue and subcellular distribution of PKCs

PKC isozymes have markedly different distributions in tissues and cells In one given cell type, there is more than one isozyme of PKC155,159 PKCα, βI, βII, δ, ε and ζ are ubiquitously distributed49,76,160-162 PKCγ is exclusively expressed in brain and spinal cord161 PKCη, θ and

λ are most abundant in skin and lung130,163, skeletal muscle131 and ovary and testis164 Such differences in tissue distribution suggest a divergence in function among the PKC isozymes Subcellular localization of the various PKC isozymes appears to be dynamic and depends on the activation state of the cell155,159 cPKC isozymes are present mainly in the cytosolic fraction Upon stimulation by a variety of extracellular ligands, the membrane-associated enzyme increases at the expense of the cytosolic one73,165 Various PKC isozymes have been found to be associated with the nucleus, some of which can translocate to this organelle depending on the cell type and activators166 Recombinant PKCα mutants that are devoid of the regulatory domain or C-terminal parts of the catalytic domain have been shown to

translocate to the nucleus of cells167 It was suggested that a nuclear targeting sequence of PKCα is present in the hinge and N-terminus of the catalytic domain The nuclear targeting sequence of PKCα seems to be exposed upon treatment with phorbol ester, which results in the translocation of PKCα to the nucleus167 It has been demonstrated that during liver regeneration, nuclear PKCα levels declined, whereas PKCδ levels increased168 The presence

of cPKC in the nucleus suggests that both the Ca2+-dependent and –independent activation of PKC in the nucleus may be critical for the regulation of gene expression

1.3.3.6 Biological function of PKCs

Several approaches have been used to investigate the function of PKC isozymes One approach is to use specific antibodies to detect the activated (phosphorylated) form of PKC isozymes169 Another approach is to relate the expression pattern of PKCs to cell phenotype The third approach is to use specific activators and inhibitors to study the functions of PKCs

in mammalian cells170-172 The last and more integrated approach is the use of isolated organ preparations and whole animal models The availability of PKC gene knockout models can help to address the question of which PKC isozyme is responsible for certain phenotypes173,174 Of the 12 PKC isozymes identified thus far, the genes for PKCγ, PKCβ, PKCε, and PKCθ have been mutated to generate null mice by homologous recombination175-

177 Neural, immunologic, and endocrine phenotypes have been reported For example, in PKCγ-null mice, there is a deficit in spatial learning, impairment in contextual fear, decreasing hyperalgesia after mechanical or inflammatory peripheral nerve injury175,178 Also, PKCγ-null mice show a decreased sensitivity to ethanol-induced hypothermia and to the sedative effects of ethanol as measured by the duration of drug-induced loss of righting reflex (LORR)179 PKCβ-null mice were initially found to demonstrate deficiencies in B-cell function and impaired hormonal immune response Furthermore, PKCβ-null mice show defects in mast cell degranulation and interleukin-6 production180 PKCθ-null mice show abnormalities in adult T-cell signaling, particularly in T-cell receptor-initiated activation of

nuclear factor-κB174 PKCε-null mice show decreased nociceptor sensitization after local injection of epinephrine and supersensitivity to the sedative effects of ethanol176,181

1.3.4 Protein Kinase C Related Kinase 1 (PRK1)

1.3.4.1 History and structure of PRK1

Protein kinase C-related kinases (PRKs) were first identified as protease-activated serine/threonine kinases via conventional protein chemistry and were originally named protease-activated kinases (PAK) It was isolated from rat liver and was shown to be able to phosphorylate S6 peptide and its analogue182-184 Later, PRK cDNAs were isolated from a human hippocampal cDNA library by low stringency plaque hybridization using a cDNA fragment for the catalytic domain of PKCβII as a probe185 ,or by PCR186 ,or with specific oligonucleotide primers187 PRKs have 50 % sequence similarity in the catalytic domain to protein kinase C186,188 The best characterized member of PRKs is PRK1, also known as PKN (protein kinase N)185 or PAK1 (protease-activated protein kinase 1)183 There are two reasons that PRK1 but not PKN is adopted as the preferred nomenclature First, PKN means novel protein kinase which is too general to represent the characteristics of this kinase, while the name of PRK1 states clearly that this kinase is homologous to PKCs Second, before Mukai and Ono used PKN to describe this kinase, another kinase which is a basic 45-47 kDa serine/threonine protein kinase activated by NGF and several other factors in PC12 cells and other cell types had already been named PKN189 In order to avoid confusion, I adopt the term

of PRK1 that was proposed by Peter Parker to refer to this kinase The amino acid sequence

of PRK1 is highly conserved in higher eukaryotes190 With the sequencing of the S cerevisiae

genome now complete191, it is clear that Pkc1 represents the sole PKC in this organism Like the PRKs, Pkc1 has been shown to be a downstream effecter of Rho GTPase, both

genetically and biochemically Rho1, the S cerevisae RhoGTPase192, directly activates

purified Pkc1 in vitro193 Purified Pkc1 is not activated by a combination of PS and DAG or

PMA in vitro194 The S pombe enzyme Pkc2 has similar biochemical features to Pkc1 In Drosophila, only one type of PKC exists, prk1 It seems, in terms of ontology, PRK1 is the

oldest isotype in the PKC superfamily The detailed structure of PRK1 (rat PRK1) is shown

in Figure 1.4 The C-terminal region of PRK1 contains the ser/thr type protein kinase domain, which is homologous to that of PKC family members The N-terminal region of PRK1 contains three homologous stretches of approximately 70 amino acids, each containing a region relatively rich in charged residues, followed by a leucine zipper-like sequence which form an ACC finger and also bind to RhoA These three domains are HR1 domains or CZ regions186 A stretch of about 130 amino acids corresponding to the HR2 domain between HR1 and catalytic domains has weak similarity to the C2 domain of PKCε The C-terminal part of the C2-like region functions as an autoinhibitory region, which is sensitive to arachidonic acid195 Also in the regulatory domain, there are pseudosubstrate and RhoB binding domains PRK1 is expressed ubiquitously in human and rat tissues196

Figure 1.4 Domain structure of PRK1 The figure shows the domains of PRK1 that interact with various molecules (adapted from several reviews and papers 186,188,195,196 )

with histone H1 and EGFR-(650-658) peptide substrate compared with brain PKCs61 Hence, PRK1 can phosphorylate most substrates of PKCs, but with different efficiency

PRK1 is activated by phospholipids, but not by Ca2+, diacylglycerol (DAG), phorbol ester or

sphorylated in the activation loop by phosphoinositide-dependent

phosphatidylserine63,199,200 Unsaturated fatty acids such as arachidonic and linoleic acids were identified as potential activators of PRK1200 Cardiolipin and phospholipids such as PI4,5-P2 and PI3,4,5-P3 and lysophospholipids such as lysophosphatidic acid (LPA) and lysophosphatidylinositol can also stimulate the protein kinase activity of PRK1187 In 1996, it was found that the small GTPase RhoA binds to PRK1 in a GTP-dependent manner201 In addition, PRK1 can interact with RhoB and RhoC201 PRK1 has been reported to interact very weakly with Rac by which PRK1 can be activated64 In the N-terminus of PRK1, there is a pseudosubstrate domain (amino acids 39-53) which can autoinhibit the activity of PRK1 The minimum region required for RhoA binding was found to consist of residues 33-111 which contains the HR1a domain and pseudosubstrate sites Since this pseudosubstrate domain overlaps with the first HR1 domain, an activation model was presented whereby RhoA binding to the HR1 domain could produce an unmasked and active catalytic domain of PRK1202 However, data from later work showed that this hypothesis was not entirely correct195 The mechanism by which activated RhoA induces conformational changes in PRK1 remains unknown

PRK1 is reported to be pho

protein kinase 1 (PDK1)203,204 Activated Rho binds to PRK1 and induces a conformational change that is permissive for PRK1 to bind to PDK1139 Rho binding disrupts the autoinhibitory effect mediated by a pseudosubstrate domain in PRKs62 The endogenous GTPase activity and the p122 Rho GAP-stimulated GTPase activity of RhoA are inhibited by the interaction with PRK1, suggesting the presence of a regulatory mechanism that sustains the GTP-bound active form of RhoA66 In PDK1 knock-out embryonic stem cells, no PRK1 protein can be found, although PRK1 mRNA is readily detected205

1.3.4.3 Biological function of PRK1-effector of Rho

rendipitously from Aplysia, and found to

level of active Rho protein: exchange factors, which

1.3.4.3.1 The structure and role of Rho proteins

The first Rho gene to be identified was cloned se

show homology to Ras in the mid 1980s206 Subsequently three human isoforms, RhoA, RhoB and RhoC, were cloned207 The Rho proteins are highly conserved during evolution In common with other Ras-related GTP-binding proteins, Rho proteins bind GTP and GDP, and hydrolyse GTP yielding GDP and free phosphate They also bind Mg2+, which is essential for GTP hydrolysis208,209 Mutation of certain residues, including G14V and Q63L, inhibits both intrinsic and GTP-stimulated GTP hydrolysis by Rho proteins, creating constitutively active proteins All three Rho proteins are post-translationally processed at the carboxyl terminus in three steps207 First they are isoprenylated at the conserved cysteine residue four amino acids from the C-terminus; second, the proteins are carboxymethylated; third, the last three amino acids are removed by proteolysis RhoA has been reported to be phosphorylated on serine

188 which may regulate its activity210

Three types of proteins can regulate the

catalyze the release of guanine nucleotide and therefore allow binding to GTP; GTPase activating proteins (GAPs), which stimulate the intrinsic rate of GTP hydrolysis, thereby converting the protein to the inactive form; and guanine nucleotide dissociation inhibitors (GDIs), which bind tightly to the proteins in the cytosol and prevent their interaction with membrane or GAPs207,211 In the basal state, most cellular RhoA is found in the cytoplasm and kept in an inactive form via GDI-dependent and possibly –independent mechanisms35,212 Upon stimulation by Rho activators, such as lysophosphatidic acid (LPA), activated Rho translocates to the plasma membrane and activates specific effector pathways35,213 The Rho proteins are activated not through tyrosine kinases but through GPCR agonists such as lysophosphatidic acid (LPA), thrombin, bombesin, and endothelin Figure 1.5 shows possible signaling pathways for GPCR activation of Rho

Figure 1.5 Proposed signaling pathways for GPCR activation of Rho (modified from review 214 )

Conventionally, G-protein (small GTPase) interacting proteins are classically defined as effectors if they bind selectively the GTP-bound form (active) of the small GTPase A number of potential downstream effector proteins for Rho have been identified through their ability to interact with Rho215 These include the protein serine/threonine kinases ROK/Rho-kinase and PRK1 and several adaptor-type proteins with no apparent enzymatic activities215 These kinases have been postulated to be principal mediators of Rho functions215,216 Rho-

mediated cellular responses include modulation of cytoskeletal dynamics (stimulating actin stress fiber formation and focal adhesion complex assembly)217, phospholipids metabolism, smooth muscle contraction, cell migration and tumor cell invasion, MAP kinase activation, gene transcription218,219, cell growth and survival response Other known Rho functions include regulation of endocytosis, exocytosis, glucose transport, and regulation of ion channels220,221

1.3.4.3.2 Biological functions of PRK1

Coexpression of wild-type PRK1 with RhoB mildly potentiates the effects of RhoB on EGF receptor trafficking, whereas coexpression of RhoB with the kinase dead PRK1-K644M mutant completely blocks these effects Thus, the first description of a biological function for PRK1 in mammalian cells was that PRK1 regulates the kinetics of EGF receptor trafficking via RhoB222 In Drosophila, PRK1 is involved in the regulation of dorsal closure during

embryogenesis as knockout of the PRK1 gene in the fly is embryonic lethal223 It has also reported that ectopically expressed PRK1 stimulates actin stress fiber depolymerization and membrane ruffling in 3T3 L1 cells204 In in vitro experiments, PRK1 was shown to directly

phosphorylate myosin and CPI-17224,225 Furthermore, PRK1 has been shown to interact with

a number of cytoskeletal proteins and thus may be involved in the modulation of cytoskeletal dynamics For example, PRK1 can directly bind to and phosphorylate the head-rod region of the intermediate filament proteins and inhibits the polymerization of many proteins related to intermediate filament formation226,227 PRK1 can also bind to the region containing EF-hand-like motifs of non-skeletal muscle type α-actinin in a Ca2+-sensitive manner, and to that of skeletal muscle type α-actinin in a Ca2+-insensitive manner PI-4,5P2 regulates the F-actin-

gelating activity of α-actinin in vitro, and also activates the protein kinase activity of PRK1 in

to be enriched in a subset of endoplasmic reticulum (ER) and ER-derived vesicles by electron microscopy229 PRK1 may also be involved in the transport of glucose in an over-expression system230 GLUT4 (glucose transporter 4) is the principal transporter for glucose uptake In

the resting cell, GLUT4 vesicles are tethered to actin Since PRK1 can regulate actin reorganization, PRK1 might affect the translocation of Glut4 indirectly It remains to be determined whether PRK1 is involved in the transport of glucose under physiological conditions PRK1 may also be involved in the apoptotic process PRK1 can be cleaved at specific sites in some apoptotic cells231 This cleavage seems to be catalyzed by caspase-3 or

a related protease The major proteolysis takes place between the N-terminal regulatory domain and the C-terminal catalytic domain, and generates a constitutively active kinase fragment which may contribute to some signal transduction, eventually leading to apoptosis Finally PRK1 may play some roles in the development of cancers, because PRK1 can interact with the E6 oncoprotein and phosphorylate E6232

1.3.5 Protein kinase C-alpha (PKCα)

PKCα was the first cloned and is the best-studied member of the PKC superfamily It can phosphorylate different substrates depending on the types of cells and stimuli In addition to several commonly used substrates, for instance MARCKS, S6 peptides and MBP, PKCα can also phosphorylate some proteins that are key regulators of cell signaling pathways Raf-1 is phosphorylated by PKCα at multiple sites233 The Raf-1 kinase inhibitor protein (RKIP) Ser-

153 can be phosphorylated by PKCα and phosphorylation of Ser-153 relieves inhibition of the Raf/MAP kinase signaling cascade 234 Rho protein is also a downstream target of PKCα

in the formation of lamellipodia 235

PKCα is involved in a variety of biological responses PKCα can regulate the proliferation of cells due to activation of the extracellular signal-regulated kinase/mitogen-activated protein kinase (ERK/MAPK) cascade initiated by Raf-1233,234,236 Alternatively, PKCα can induce the upregulation of p21waf1/Cip1 237 in the IEC-18 intestinal crypt cell line In MCF-10 human mammary epithelial cell line, overexpressed PKCα results in cell cycle arrest and inhibits cell proliferation by phosphorylating retinoblastoma protein (R6)238

The effect of PKCα on apoptosis depends on the cell type in question For some cell types, PKCα exhibits an anti-apoptotic effect Activation of ERK/MAPK and phosphorylation of anti-apoptotic protein Bcl2 by PKCα233,236,239 mediates the anti-apoptotic effect of PKCα in NIH3T3 cells, COS7 cells and HL60 cells On the other hand, in some cell types, the effect

of PKCα is pro-apoptotic Thus, PKCα can mediate the phosphorylation and solubilization of lamin B in HL60 cells to alter nuclear and chromatin structure to induce apoptosis240

PKCα can also influence cell differentiation The inhibition of cell cycle progression and expression of cell-specific functions are the characteristics of cell differentiation PKCα is known to upregulate some proteins which affect cyclin and cyclin-dependent kinase to arrest the cell cycle to facilitate differentiation in MCF-10 human mammary epithelial cells238

Furthermore, PKCα is involved in cell migration and adhesion For example, PKCα is found

to translocate and accumulate in the focal contacts241; PKCα can also phosphorylate ERM proteins (ezrin, radixin, and moesin) to mediate cell migration242

PKCα is important in carcinogenesis Overexpression of PKCα in human breast cancer cells results in a more aggressive and metastatic phenotype with anchorage-independent growth in soft-agar and tumorgenicity in nude mice235,243 PKCα can participate in the invasion, morphological change and drug-resistance of human breast tumors243 Mutations in PKCα have been found in different types of cancers For example, PKCα-D294G, a point mutation

in the V3 domain of PKCα, was identified in human pituitary and thyroid cancer240

Moreover, PKCα has been found to be involved in cardiac hypertrophy and angiogenesis Overexpression of wild-type PKCα induced hypertrophic growth of neonatal cardiomyocytes244 PKCα mediates hypertrophic growth of neonatal cardiomyocytes partially via activation of the ERK1/2-dependent signaling pathway244 Overexpression of PKCα in rat capillary endothelial cells enhanced their migration in response to hepatocyte growth factor245 PKCα is also implicated in migration of vascular smooth muscle cells246

PKCα is also implicated in regulation of the inflammatory response PKCα can increase the production of nitric oxide (NO)247; overexpression of PKCα in the epidermis of transgenic mice by keratin 5 promoter results in the expression of cyclooxygenase-2 and tumor necrosis factor248 However, the function of PKCα seems to depend on tissue and/or cell type Thus, PKCα can inhibit the secretory responses and release of arachidonic acid metabolites in mast cells249 and induce the expression of IκB to inhibit the inflammation induced by NFκB250 in HepG2 cells

1.4 The composition and structure of the plasma membrane

Communication among cells is important for differentiation in multicellular organisms Many

of the extracellular signals need to bind to receptors on the cell surface to relay signals and signaling complexes often form in proximity to the plasma membrane Thus, the plasma membrane provides an ideal matrix for organizing signal transduction to achieve specificity

and efficiency

1.4.1 Structure and composition of the plasma membrane

The lipids in the membrane are mainly phospholipids which prefer to form bilayer structures

in aqueous solution In 1972, S J Singer and G L Nicolson proposed the fluid mosaic model for membrane structure251 This model suggests that membranes are dynamic structures composed of protein and phospholipids In this model, the lipid bilayer is a fluid matrix, a two-dimensional solvent for proteins Both lipids and proteins in the membrane can only move rotationally and/or laterally Singer and Nicolson also defined two classes of membrane proteins, one is peripheral membrane proteins, the other is integral membrane proteins

1.4.2 Lipids in the membrane

1.4.2.1 The classes of lipids

There are many different types of lipids in the plasma membrane and the composition of lipids in the membrane of various cell types is different For example, the percentage of total lipid by weight in the plasma membrane of a rat liver cell is as follows: 17% cholesterol, 7%

phosphatidylethanolamine, 4% phosphatidylserine, 24% phosphatidylcholine, 19% sphingomyelin, 7% glycolipids and 22% other lipids The corresponding lipids found in erythrocyte membranes are 23%, 18%, 7%, 17%, 18%, 3% and 13%, respectively252 In mouse liver cells, the plasma membrane contains 54% (dry weight) lipid with a protein-to-lipid ratio of 0.85; while in human erythrocytes, the membranes contain 43% lipid with a protein-to-lipid ratio of 1.1253 There is significant difference in the lipid composition of the plasma membrane compared with that of endomembranes For example, rat liver nuclear membrane contains 35% lipid with a protein-to-lipid-ratio of 1.6 and mitochondrial inner membrane has a lipid composition of only 24% and an exceptionally high protein-to-lipid-ratio of 3.2 Such differences underlie different biological functions performed by different membranes within the same cell

1.4.2.1.1 Fatty acids

Fatty acids are either saturated or unsaturated In the membrane, stearic acid and palmitic acid are the most common saturated fatty acids254 The most common unsaturated fatty acid is oleic acid Arachidonic acid can be synthesized from linoleic acid which is important for the activation of PRK1 Triacylglycerols are a major energy reserve and the principal neutral derivatives of glycerol found in animals255

1.4.2.1.2 Phospholipids

Phosphatidic acid is found in most natural systems and is an important intermediate in the biosynthesis of glycerophospholipids256 In these compounds, various polar groups are esterified to the phosphoric acid moiety of the molecule The phosphate and the esterified entities are referred as a “head” group Phosphatides with choline or ethanolamine are referred to as phosphatidylcholine (lecithin) or phosphatidylethanolamine, respectively These phosphatides are two of the most common constituents of biological membranes257 Other common head groups found in phosphatides include serine and inositol

Figure 1.6 The structure of phospholipids (adapted from Albert, B et al.258 )

1.4.2.1.3 Sphingolipids

Sphingosine is the backbone of these lipids and a fatty acid is joined to a sphingosine to form

a ceramide Sphingomyelins represent a phosphorus-containing subclass of sphingolipids The glycosphingolipids consist of a ceramide with one or more sugar residues in a β-glycosidic linkage at the 1-hydroxyl moiety which appear to determine certain elements of tissue and organ specificity, cell-cell recognition and tissue immunity259

1.4.2.1.4 Steroids

Cholesterol is a principal component of animal plasma membranes—there is approximately one cholesterol molecule for every phospholipid molecule It is an important determinant of membrane properties In certain animal cells it may constitute up to 50% of the lipid molecules in the plasma membrane260,261 It is noteworthy that cholesterol occurs in lesser amounts in the membranes of animal intracellular organelles Cholesterol is absent from the plasma membrane of most plant and all bacterial cells Unlike the phosphoglycerides and sphingomyelins, cholesterol is too hydrophobic to form the bilayer structure, it can only mix

with phospholipids to participate in the formation and functioning of the bilayer Within the membrane, it is oriented with its small hydrophilic end facing the external surface of the bilayer and the bulk of its structure packed in among the fatty acid tails of the phospholipids The relatively rigid fused ring system of cholesterol and the weakly polar alcohol group at the C-3 position have important consequences for the properties of the plasma membranes.The placement of the cholesterol molecules interferes with the tight packing of the phospholipids, which tends to increase the fluidity of the bilayer By decreasing the mobility of the first few CH2 groups of the hydrocarbon chains of the phospholipid molecules, cholesterol molecules make the bilayer less deformable and thus decrease the permeability of the bilayer to small water-soluble molecules It also prevents the hydrocarbon chains from coming together and crystallizing and thereby inhibits possible phase transition

1.4.2.2 Biological membranes are asymmetric structures

1.4.2.2.1 Lipids exhibit lateral membrane asymmetry

Experimentally, the biochemical and biophysical properties of membranes are often studied

in model membrane systems in which mixtures of lipids are allowed to swell in an aqueous environment to form liposomes at thermodynamic equilibrium262,263

The existence of lateral domains is a shared property for both biological membranes and membranes in model systems There are at least three main differences between lateral domains in biological membranes and those in giant liposomes262 First, the lipid composition of the two leaflets in the model systems but not biological membranes are identical Second, the size of the lateral domains in biological membranes is much smaller than that in the model systems (100-300 nm compared with μm scales in the liposome); and the lateral domains in biological membranes are more dynamic than their counterparts in the model systems (the life-time being measured in seconds compared with that of minutes in the liposome264-267) Third, biological membranes contain proteins that would influence the composition, size and maybe overall properties of the lateral domains268,269

Lipid rafts are microdomains within the plasma membrane that are enriched in cholesterol, sphingomyelin, and glycolipids 270,271 Sphingolipids differ from most biological phospholipids in containing long, largely saturated acyl chains, which allow them to readily pack tightly together272-274 Cholesterol has important effects on phase behavior The addition

of cholesterol gives the membrane properties intermediate between the gel and lipid-disorder phases Cholesterol is thought to contribute to the tight packing of lipids in lipid-ordered microdomains by filling interstitial spaces between other lipid molecules272,275 The combined characteristics of these two types of lipids form the basis for the formation of microdomains (lipid rafts) in the plasma membrane, where the concentration of lipids and proteins as well as their physicochemical properties are different from the surrounding environment Such microdomains provide a structural basis for heterogeneity of the plasma membrane and are thought to be essential for signal transduction

1.4.2.2.2 Lipids exhibit transverse membrane asymmetry

Phospholipids are also distributed asymmetrically across many membranes The lipids have been found to be asymmetrically distributed between the inner and outer monolayers For example, in human erythrocyte membranes, almost all of the lipid molecules that have choline in their head group (phosphatidylcholine and sphingomyelin) are located in the outer monolayer, while almost all of the phospholipid molecules that contain a terminal primary amino group (phosphatidylethanolamine and phosphatidylserine) are found in the inner monolayer Because the negatively charged phosphatidylserine is located in the inner monolayer, there is a significant difference in charge between the two halves of the bilayer Some integral membranes prefer to bind to particular lipid classes in the inner and outer monolayers to affect signal transduction273 For example, the activator of PKCs, phosphatidylserine exists in the inner leaflet of the plasma membrane under physiological conditions Flippases can flip phosphatidylserine from the inner leaflet to the outer leaflet of the plasma membrane in the condition of apoptosis276

1.4.3 Membrane proteins

Membrane proteins carry out essentially all of the active functions of the membrane, including transport activities, receptor functions, and other related processes Membrane proteins are classified as peripheral and integral membrane proteins

The terms of “integral membrane proteins” and “peripheral membrane proteins” were originally proposed by S.J Singer277 and later elaborated by S.J Singer and G.L Nicolson in

1972251 In the last 33 years, the definition of integral membrane protein has not changed: an integral membrane protein requires disruption of the lipid bilayer in order to be released from the membrane It is noteworthy that the distinction between an integral membrane protein and a peripheral membrane protein is not based on the mode of their attachment to the lipid bilayer but rather on their relative strength of attachment251,278,279

1.4.3.1 Peripheral membrane proteins

Peripheral membrane proteins are proteins that weakly associate with the membrane mainly through electrostatic and hydrogen-bonding interactions with integral membrane proteins or with phospholipids in the membrane A peripheral membrane protein can be released from the membrane by relatively gentle treatment that leaves the lipid bilayer intact, such as extraction with solutions of very high or low ionic strength, solutions of extreme pH or chaotropic agents278-280 The binding of proteins to membranes involves different types of interactions These include interactions between cationic residues of proteins and anionic lipids281, interactions between aromatic residues of proteins and zwitterionic lipid282, hydrophobic interactions involving aliphatic residues (and Phe) that are exposed by conformational changes of proteins283 and interactions between proteins and a lipid messenger such as DAG284 Specific domains in proteins have been identified to be involved

in the interaction with cellular membranes, such as the C1 domain, C2 domain, PH domain and FYVE domain

1.4.3.2 Integral membrane proteins

An integral membrane protein is a membrane protein that requires disruption of the lipid bilayer in order to be released from the membrane251 It is important to appreciate that a protein does not have to possess a transmembrane domain in order to be an integral

membrane protein There are many bona fide integral membrane proteins that do not have a

transmembrane domain at all Examples are cytochrome b5285, flotillin-1286 and K-Ras and H-Ras287 Another class of well characterized (with their crystallographic structures solved) integral membrane proteins is monotopic integral membrane proteins that only penetrate one leaflet of the lipid bilayer278 (e.g., cyclooxygenase-1 and -2288, squalene-hopene cyclase289and fatty acid amide hydrolase290) Based on the original definition of Singer and Nicolson, the theme that an integral membrane protein does not necessarily possess a transmembrane domain prevails258,278,291-293

Most integral membrane proteins fall into two classes although there are diverse integral membrane proteins in cells In the first class of integral membrane proteins, only a small hydrophobic segment attaches or anchors to the membrane, while most of the protein extends out into the aqueous solvent In the second class of integral membrane proteins, most of the protein is embedded in the membrane only a small surface of the protein extrudes outside the membrane Proteins can integrate into the cellular membrane in various ways as discussed below

1.4.3.2.1 A protein with transmembrane segments

For the proteins that are anchored by a small hydrophobic polypeptide segment, that segment often takes the form of a single α-helix Usually an alpha helix of about 20 residues is sufficiently long to span a lipid bilayer if the helix is oriented perpendicular to the bilayer plane294 Some membrane proteins take on a more globular shape and are often involved in transport activities and other functions Such functions require a substantial portion of the peptide to be embedded in the membrane In these proteins, the primary sequence may consist of several hydrophobic α-helical segments joined by hinge regions so that the protein

winds in a zig-zag pattern back and forth across the membrane such as G-protein-coupled receptors3,295 Major efforts have been made to identify membrane-spanning alpha helices in the study of integral membrane proteins278,291

1.4.3.2.2 Proteins with β-barrel

Besides the high helical content of many membrane proteins, β-turns and β-strands also exist

in integral membrane proteins296 β-strands are found in the outer membranes of negative bacteria, forming rigid pores known as β-barrels297 In each subunit of porin, 16 β strands form a barrel-shaped structure with a pore in the center In a porin monomer298, the outward-facing side groups on each of the β strands are hydrophobic and form a nonpolar ribbonlike band that encircles the outside of the barrel The side groups facing the inside of the bilayer are predominantly hydrophilic

Gram-1.4.3.2.3 Lipid-anchored membrane proteins

Certain proteins are found to be covalently linked to lipid molecules In these lipid-anchored proteins, the lipid hydrocarbon chains are embedded in the bilayer, but the protein itself does not enter the bilayer Four different types of lipid anchoring motifs have been found 299 These are amide-linked myristoyl anchors, thioester-linked fatty acyl anchors, thioether-linked prenyl anchors, and amide-linked glycosylphosphatidylinositol anchors Each of those anchors exhibits a characteristic pattern of structural requirement

1.4.3.2.3.1 Amide-linked myristoyl anchors

Myristic acid can be linked via the α-amino group of the N-terminal glycine residue of the protein via an amide bond The reaction is referred to as N-myristoylation300 Retention of such proteins at the membrane by the N-terminal anchor may play an important role in a membrane associated function, but a single myristate anchor to a protein never gives rise to

an integral membrane protein

1.4.3.2.3.2 Thioester-linked fatty acyl anchors

A variety of proteins contain fatty acids covalently bound via ester linkages to the side chains

of cysteine and sometimes to serine or threonine residues within a polypeptide chain301

Myristate, palmitate, stearate, and oleate can be esterified in this way, with the C16 and C18 chain lengths being most commonly found Palmitoylated proteins can be categorized into four general classes 302 One group consists of transmembrane proteins that are acylated on a cysteine residue at or near the transmembrane domain303,304 In the second group, the palmitoylation occurs in the C-terminal region and depends on the prior prenylation of the cysteine residue in the “CAAX” motif305 Some proteins are palmitoylated at one or more cysteines within the first 10-20 amino acids306,307 Some other membrane proteins are dually acylated and contain a “consensus” sequence: Met-Gly-Cys at the N-terminus308-310 Certain proteins have dual modification by myristoylation/palmitoylation or prenylation/palmitoylation to become integral membrane proteins For example, some kinases in the Src family containing the N-terminal sequence Met-Gly-Cys are modified by two fatty acids: myristate and palmitate308,309, which makes these proteins integrate into the cellular membrane

1.4.3.2.3.3 Thioether-linked prenyl anchors

Prenylated proteins are modified by attachment of either the 15-carbon farnesyl or 20-carbon geranylgeranyl moiety The modified residue is a cysteine at or near the C-terminus These proteins for prenylation contain a C-terminal “CAAX” box consensus sequence Prenylation

of the cysteine in the CAAX box triggers further modification in which the last three amino acids are proteolytically removed and the resultant C-terminal prenylated cysteine is carboxy-methylated311

1.4.3.2.3.4 Glycosylphosphatidylinositol anchors

Glycosylphosphatidylinositol, or GPI, modifies the carboxy-terminal amino acid of a target protein via an ethanolamine residue linked to the oligosaccharide, which is linked in turn to the inositol moiety of a phosphatidylinositol312 The inositol moiety can also be modified by

an additional fatty acid, and a variety of fatty acyl groups are found linked to the glycerol group GPI-linked proteins are a major class of integral membrane proteins313 Well over 100 GPI-linked proteins have been identified in cells, where they perform numerous functions

including acting as enzymes and receptors GPI-linked proteins diffuse in membranes much faster than transmembrane proteins

1.4.3.2.4 Topology of the integral membrane proteins

The word ‘topology’ is used to refer to the arrangement of membrane polypeptides relative to the two sides of a bilayer Topology of proteins emphasizes the number and orientation of the membrane-spanning segments278 The classical approaches to the determination of

membrane protein topology are vectorial labeling and in situ proteolysis; more recently there

have been advances in the use of immunological methods for the localization of exposed portions of the membrane proteins278 Other methods to study the topology of membrane proteins include constructing fusion proteins with marker enzymes that are active only on one side of the membrane and predicting the topology of membrane proteins by computational methods from the protein sequence Because of the difficulties in determining the high-resolution crystal structures of integral membrane proteins, most of the topological information about membrane proteins is of two-dimensional low resolution and the exact structure-function relations are only testable hypotheses314-317

1.4.3.3 Experimental approaches to distinguish peripheral membrane proteins from integral membrane proteins

A protein is classified as an integral membrane protein when it can only be extracted from cellular membranes by methods that disrupt the integrity of the lipid bilayer251 A variety of methods can be used to extract integral membrane proteins Non-ionic detergent such as Triton X-100 is often employed Detergents are small amphipathic molecules that interact with both nonpolar and polar environments318 Detergents disrupt membranes319 They bind

to hydrophobic regions of proteins such as transmembrane domains, thereby replacing the unfavorable contacts between hydrophobic protein regions and water with more favorable

hydrophilic domains of the detergent In most in vitro assays, membrane proteins are

solubilized using detergent (e.g., 1 % Triton X-100) and subjected to various analyses Several strategies are commonly employed to examine whether a protein in question is

tightly membrane-bound and thus an integral membrane protein Chaotropic agents are generally large molecular ions such as thiocyanate, perchlorate, and trichloroacetate They enhance the transfer of nonpolar molecules to aqueous environments by their disrupting influence on water structure Therefore, a chaotropic agent will strip peripheral membrane proteins off biological membranes without disrupting the lipid bilayer The peripheral membrane proteins can be removed from membranes by other extraction regimens that disrupt their weak association with membranes but leave the lipid bilayer intact For example, exposure to organic solvents and low salt concentration reduces hydrophobic interactions High salt extractions decrease the Debye-Huckel screening length and coulombic attraction, leading to diminished hydrophobic interactions Treatment of membranes with a high concentration of urea or guanidine disrupts hydrogen bonds

1.5 The prevailing translocation dogma in cell signaling

Most extracellular signals bind to receptors in the plasma membrane to initiate the signal transduction cascade inside the cell It is generally thought that signaling proteins in the cytosol need to translocate to the plasma membrane as peripheral membrane proteins to relay the signals by random diffusion The translocations of PKCs, PKB and Ras/MAPK are typical examples of this dogma137,320-322

The process of translocation includes two aspects: (1) the generation of new protein-protein binding interactions, for example, phosphorylation of protein, GDP to GTP exchange and conformational change and (2) second messenger-mediated interactions between signaling proteins and lipid partners in the plasma membrane, for example, Ca2+, DAG and phosphoinositide derivatives322 By these two processes, cells can rapidly and reversibly induce new binding sites for signaling proteins and thereby change the localization state of a large class of continuously diffusing signaling proteins322

1.5.1 Translocation of peripheral membrane protein

In the basal state, most of the peripheral membrane proteins randomly diffuse via Brownian motion in the cytosol Upon engaging agonists, the receptors on the cell surface undergo a