10
How lipids and proteins interact in a membrane: a molecular approach Anthony G. Lee Received 31st March 2005, Accepted 4th July 2005 First published as an Advance Article on the web 14th July 2005 DOI: 10.1039/b504527d Membrane proteins in a biological membrane are surrounded by a shell or annulus of ‘solvent’ lipid molecules. These lipid molecules in general interact rather non-specifically with the protein molecules, although a few ‘hot-spots’ may be present on the protein where anionic lipids bind with high affinity. Because of the low structural specificity of most of the annular sites, the composition of the lipid annulus will be rather similar to the bulk lipid composition of the membrane. The structures of the solvent lipid molecules are important in determining the conformational state of a membrane protein, and hence its activity, through charge and hydrogen bonding interactions between the lipid headgroups and residues in the protein, and through hydrophobic matching between the protein and the surrounding lipid bilayer. Evidence is also accumulating for the presence of ‘co-factor’ lipid molecules binding with high specificity to membrane proteins, often between transmembrane a-helices, and often being essential for activity. Intrinsic membrane proteins The membrane surrounding a biological cell mediates all interactions between the cell and its environment; it therefore contains a variety of proteins involved in transporting ions and molecules across the membrane and in sending signals across the membrane, to and from the cell. The internal compart- ments present in eukaryotic cells are also each surrounded by their own membranes, each containing their own sets of membrane proteins. Finally, many bacteria have an outer membrane surrounding the inner cell membrane, providing protection from the environment; the composition of the bacterial outer membrane is distinctly different from that of the inner membrane. The importance of membranes for the function of a cell is shown by the fact that about 30% of the genome codes for membrane proteins. The membrane is now a major focus for activity not only in fundamental research but also in the applied pharmaceutical sector where most drugs and the greatest proportion of research funds are targeted to membrane components. The most fundamental of the roles of a biological membrane is as a permeability barrier, this barrier being provided by the lipid bilayer component of the membrane, into which the membrane proteins are inserted. Insertion of membrane proteins into the lipid bilayer cannot be allowed to destroy the permeability barrier properties of the lipid bilayer and the membrane proteins must be stable and functional in the environment provided by the lipid bilayer; this means that membrane proteins must have co-evolved with the lipid component of the membrane. Understanding the membrane as a biological system therefore requires an understanding of how lipid and protein molecules interact in a membrane. Intrinsic membrane proteins are those that span the hydrophobic core of the lipid bilayer component of the membrane. The cost of transferring a peptide bond from water into a non-polar environment has been estimated to be about 25 kJ mol 21 when not hydrogen bonded but only about 2.4 kJ mol 21 when hydrogen bonded. 1 Membrane spanning regions of a protein therefore adopt either a-helical or b-sheet structures since these two structures allow the formation of the maximum number of peptide hydrogen bonds. b-sheet structures are found in proteins in bacterial outer membranes, in the form of b-barrels such as that adopted by the porin OmpF 2 shown in Fig. 1. All other intrinsic membrane proteins are based on the a-helix, with transmembrane regions containing one or more hydrophobic a-helices, as shown for the mechanosensitive channel MscL 3 and for the potassium channel KcsA 4 in Fig. 1. The nature of the problem One of the puzzles of membrane biochemistry is that the lipid composition of the membrane is very complex, a typical membrane containing hundreds to thousands of chemically distinct species of lipid molecule, differing in the structures of the fatty acyl chains and lipid headgroups. Of course, not all School of Biological Sciences, University of Southampton, Southampton, UK, SO16 7PX. E-mail: [email protected]; Tel: 44 (0)23 8059 4331 Anthony Lee Anthony Lee gained BSc and PhD degrees in Chemistry from the Universities of London and Cambridge, respectively. Following a period as a Molecular Pharmacologist in Cambridge, supported by Fellowships from the Salters’ Company and from King’s College, and then at the National Institute for Medical Research, he moved to the University of Southampton as a Biochemist where he now holds a Chair in Biochemistry in the School of Biological Sciences. REVIEW www.rsc.org/molecularbiosystems | Molecular BioSystems This journal is ß The Royal Society of Chemistry 2005 Mol. BioSyst., 2005, 1, 203–212 | 203 Downloaded by Indiana University - Purdue University at Indianapolis on 08 September 2012 Published on 14 July 2005 on http://pubs.rsc.org | doi:10.1039/B504527D View Online / Journal Homepage / Table of Contents for this issue

How lipids and proteins interact in a membrane: a molecular approach

Embed Size (px)

Citation preview

How lipids and proteins interact in a membrane: a molecular approach

Anthony G. Lee

Received 31st March 2005, Accepted 4th July 2005

First published as an Advance Article on the web 14th July 2005

DOI: 10.1039/b504527d

Membrane proteins in a biological membrane are surrounded by a shell or annulus of ‘solvent’

lipid molecules. These lipid molecules in general interact rather non-specifically with the protein

molecules, although a few ‘hot-spots’ may be present on the protein where anionic lipids bind with

high affinity. Because of the low structural specificity of most of the annular sites, the composition

of the lipid annulus will be rather similar to the bulk lipid composition of the membrane. The

structures of the solvent lipid molecules are important in determining the conformational state of

a membrane protein, and hence its activity, through charge and hydrogen bonding interactions

between the lipid headgroups and residues in the protein, and through hydrophobic matching

between the protein and the surrounding lipid bilayer. Evidence is also accumulating for the

presence of ‘co-factor’ lipid molecules binding with high specificity to membrane proteins, often

between transmembrane a-helices, and often being essential for activity.

Intrinsic membrane proteins

The membrane surrounding a biological cell mediates all

interactions between the cell and its environment; it therefore

contains a variety of proteins involved in transporting ions and

molecules across the membrane and in sending signals across

the membrane, to and from the cell. The internal compart-

ments present in eukaryotic cells are also each surrounded by

their own membranes, each containing their own sets of

membrane proteins. Finally, many bacteria have an outer

membrane surrounding the inner cell membrane, providing

protection from the environment; the composition of the

bacterial outer membrane is distinctly different from that of

the inner membrane. The importance of membranes for the

function of a cell is shown by the fact that about 30% of the

genome codes for membrane proteins. The membrane is now a

major focus for activity not only in fundamental research but

also in the applied pharmaceutical sector where most drugs

and the greatest proportion of research funds are targeted to

membrane components.

The most fundamental of the roles of a biological membrane

is as a permeability barrier, this barrier being provided by the

lipid bilayer component of the membrane, into which the

membrane proteins are inserted. Insertion of membrane

proteins into the lipid bilayer cannot be allowed to destroy

the permeability barrier properties of the lipid bilayer and the

membrane proteins must be stable and functional in the

environment provided by the lipid bilayer; this means that

membrane proteins must have co-evolved with the lipid

component of the membrane. Understanding the membrane

as a biological system therefore requires an understanding of

how lipid and protein molecules interact in a membrane.

Intrinsic membrane proteins are those that span the

hydrophobic core of the lipid bilayer component of the

membrane. The cost of transferring a peptide bond from

water into a non-polar environment has been estimated to be

about 25 kJ mol21 when not hydrogen bonded but only about

2.4 kJ mol21 when hydrogen bonded.1 Membrane spanning

regions of a protein therefore adopt either a-helical or b-sheet

structures since these two structures allow the formation of

the maximum number of peptide hydrogen bonds. b-sheet

structures are found in proteins in bacterial outer membranes,

in the form of b-barrels such as that adopted by the porin

OmpF2 shown in Fig. 1. All other intrinsic membrane proteins

are based on the a-helix, with transmembrane regions

containing one or more hydrophobic a-helices, as shown for

the mechanosensitive channel MscL3 and for the potassium

channel KcsA4 in Fig. 1.

The nature of the problem

One of the puzzles of membrane biochemistry is that the

lipid composition of the membrane is very complex, a typical

membrane containing hundreds to thousands of chemically

distinct species of lipid molecule, differing in the structures of

the fatty acyl chains and lipid headgroups. Of course, not all

School of Biological Sciences, University of Southampton, Southampton,UK, SO16 7PX. E-mail: [email protected]; Tel: 44 (0)23 8059 4331

Anthony Lee

Anthony Lee gained BSc andPhD degrees in Chemistry fromthe Universities of Londonand Cambridge, respectively.Following a period as aMolecular Pharmacologist inCambridge, supported byFellowships from the Salters’Company and from King’sCollege, and then at theNational Institute for MedicalResearch, he moved to theUniversity of Southampton asa Biochemist where he now holdsa Chair in Biochemistry in theSchool of Biological Sciences.

REVIEW www.rsc.org/molecularbiosystems | Molecular BioSystems

This journal is � The Royal Society of Chemistry 2005 Mol. BioSyst., 2005, 1, 203–212 | 203

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online / Journal Homepage / Table of Contents for this issue

this complexity is required for proper function of the

membrane; the costs involved in controlling the lipid

composition of a membrane in which each and every species

of lipid had its own distinct function would be prohibitive.

Indeed, the fact that, for example, the fatty acyl chain

composition of the membrane lipids can be changed signifi-

cantly by changing diet, with no observable effects on function

(see, for example, ref. 5), shows that the exact lipid com-

position is not critical for function. Nevertheless, properties

of the lipids such as the average chain length are kept fairly

constant, this being achieved by the normal processes of

metabolic control in a cell, that is, by the selectivity shown by

the enzymes involved in lipid synthesis and lipid turnover.6

What is an acceptable lipid composition for a membrane

and what is not will be determined in part by the requirements

of the intrinsic membrane proteins, but other factors will also

be important, including the requirements for particular lipid

headgroups for interaction with extrinsic membrane proteins,

the role of the membrane lipids as a store for components to

be used elsewhere in the cell, the role of the lipid bilayer in

events such as membrane fusion and fission, and the role of

lipids in determining bulk properties of the membrane such as

elasticity, important in allowing deformation of the membrane

without damage. Some of the functions of a biological

membrane are also likely to depend on the mixing properties

of the lipid molecules in the membrane. For example, in mix-

tures of phosphatidylcholines, cholesterol and sphingolipids,

strong association between the cholesterol and sphingolipid

molecules can cause domains in a liquid ordered state,

enriched in cholesterol and sphingomyelin, to separate from

liquid disordered domains, enriched in phosphatidylcholine.7

The liquid ordered domains have been referred to as rafts. The

line tension present at the interfaces between the domains of

liquid ordered and liquid disordered lipid can lead to breaking

of the membrane at the interface with the formation of

separate vesicles enriched in phosphatidylcholine, a process

that could be important in the budding of vesicles in cells.8 The

existence of liquid ordered domains in a membrane raises

an interesting question about the localization of intrinsic

membrane proteins in the membrane. Integral membrane

proteins and model transmembrane a-helices are excluded

from gel phase domains9–12 and transmembrane a-helices are

excluded from liquid ordered domains13 probably due to poor

packing of the helices with closely packed lipid molecules in

the gel or liquid ordered phases. However, it is currently

unclear how far these results can be extrapolated to the

biological membrane; do large domains of liquid ordered lipid

exist in biological membranes and are these domains denuded

of intrinsic membrane proteins?

There is no doubt that changing the chemical composition

of the lipid bilayer surrounding an integral membrane protein

can affect the activity of the protein. For example, Fig. 2 shows

the effect of changing phospholipid fatty acyl chain length on

the activities of bacterial diacylglycerol kinase, which uses

ATP to convert diacylglycerol to phosphatidic acid, and the

Ca2+-ATPase that transports Ca2+ across the membrane of the

sarcoplasmic reticulum in skeletal muscle;14,15 in both cases

Fig. 1 Side views of the bacterial b-barrel protein OmpF (A) and of

two a-helical membrane proteins, the homopentameric mechanosensi-

tive channel of large conductance, MscL (C) and the homotetrameric

potassium channel KcsA (D). In (C) and (D) one subunit in each

oligomeric structure is shown in blue. In (D) potassium ions are shown

in purple and the lipid molecule modelled as a diacylglycerol (DAG) is

shown in space-fill format bound at one of the monomer–monomer

interfaces. (B) shows a surface plot of the bacteriorhodopsin trimer;

each monomer in the trimer is shown in a different colour. Lipid

molecules bound to the surface are shown in space-fill format; the

lipid headgroups are not resolved in the crystal structure and the lipid

molecules have therefore been modelled as 2,3-di-O-phytanyl-sn-

propane. (PDB files 1OPF, 1QHJ, 1K4C and 1MSL, respectively).

Fig. 2 The effect of fatty acyl chain length on enzyme activity in

bilayers of phosphatidylcholine in the liquid crystalline phase. Ca2+-

ATPase (%; right hand axis) or diacylglycerol kinase (#; left hand

axis) were reconstituted into phosphatidylcholines containing mono-

unsaturated fatty acyl chains of the given chain lengths. ATPase

activities were determined at 25 uC. For diacylglycerol kinase the

substrate was dihexanoylglycerol.14,15

204 | Mol. BioSyst., 2005, 1, 203–212 This journal is � The Royal Society of Chemistry 2005

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

there is an optimum fatty acyl chain length for activity in

the range C16–C20 that matches well the average fatty acyl

chain length found in a biological membrane. Similarly,

changing lipid headgroup structure leads to changes in enzyme

activity; for example, Table 1 shows that, for these two

proteins, phosphatidylcholine supports a higher activity

than phosphatidylethanolamine, and that the anionic phos-

pholipids phosphatidic acid and phosphatidylserine support

very low activities.15–17

Explanations for the observed effects of lipid structure on

membrane protein function fall into two broad classes.

Explanations at the microscopic level seek an explanation in

terms of the chemical interactions (hydrogen bonding, charge,

hydrophobicity and size) between the lipid and the protein.

Explanations at the macroscopic level seek an explanation in

terms of changes in bulk properties of the lipid bilayer such as

viscosity, internal pressure, and stored curvature elastic stress.

Sometimes these two classes of explanation are merely

different ways of saying the same thing. Sometimes, however,

there are important differences between the two classes of

explanation. An example is provided by the various explana-

tions that have been presented for the effects of phosphatidyl-

ethanolamines compared to those of phosphatidylcholines. A

microscopic explanation would seek to explain any observed

effects in terms of the different sizes and different potentials

for hydrogen bonding of the phosphatidylethanolamine

and phosphatidylcholine headgroups. For example, the phos-

phatidylcholine headgroup can act as a hydrogen bond

acceptor through its phosphate group whereas the phos-

phatidylethanolamine headgroup can be both a hydrogen

bond acceptor through its phosphate group and a hydrogen

bond donor through its –NH3+ group. Further, interactions

of the phosphatidylcholine and phosphatidylethanolamine

headgroups with the surrounding water molecules will be

very different; molecular dynamic simulations show that

whereas the hydrophobic –NMe3+ group induces formation

of a clathrate-like hydration shell around the headgroups

in order to optimise inter-water hydrogen bonding, direct

hydrogen bonds are formed between the –NH3+ group and the

water molecules.18,19

In contrast to explanations at the microscopic level, a

macroscopic explanation would seek to explain any observed

effects of lipid composition in terms of parameters such as

stored curvature elastic stress.20 Phosphatidylethanolamines

tend, when isolated, to form hexagonal HII phases rather

than bilayer phases (Fig. 3).21 In a biological membrane, the

presence of both the intrinsic membrane proteins and the

bilayer-preferring lipids will, however, force the phosphatidyl-

ethanolamine to adopt a bilayer structure,22 which will

therefore be in a state of curvature stress. It has been suggested

that this curvature elastic energy could enhance binding of

extrinsic membrane proteins to the surface of the membrane,

since insertion of a protein into the bilayer surface will reduce

the curvature stress (Fig. 4).23 It has also been suggested that

curvature elastic stress could shift the equilibrium between

conformational states of an intrinsic membrane protein to

favour that with the greatest hydrophobic thickness (Fig. 4).24

However, it is questionable whether curvature elastic stress

could be important in biological membranes. It is important in

a complex membrane containing many different membrane

proteins that, unless there is a specific requirement otherwise,

each protein should be unaffected by what its neighbours are

doing; the design of a robust membrane system requires that

cross-talk between proteins in the membrane be minimized.

Membrane proteins should be ‘independent demons’ in the

membrane, affected only by their immediate environments, but

macroscopic properties of the membrane, as properties of the

whole membrane, would tend, if important, to link together all

the proteins in the membrane. For example, if stored elastic

energy, a thermodynamic property of the whole membrane,

Table 1 Effects of lipid headgroup structure on membrane proteinactivitya

Headgroup

Relative activity

Ca2+-ATPase Diacylglycerol kinase

Phosphatidylcholine 1 1Phosphatidylethanolamine 0.5b 0.44c

Phosphatidic acid 0 0Phosphatidylserine 0.32 0.07a Proteins were reconstituted into bilayers containing the givenphospholipids in which both fatty acyl chains were oleic acid.Activities were measured at 25 uC. Data from refs. 15–17.b Although the temperature is above that of the bilayer-hexagonalHII phase transition for the phosphatidylethanolamine, the lipid islikely to be a bilayer phase because membrane proteins stronglystabilize the bilayer phase.22 c In a mixture containing 80 mol%phosphatidylethanolamine and 20 mol% phosphatidylcholine.

Fig. 3 (Top) The structures of the zwitterionic glycerophospholipids

phosphatidylcholine and phosphatidylethanolamine. (Bottom) The

structures adopted by lipids when dispersed in water depend on their

‘shape’, lipids with an overall cylindrical shape adopting a bilayer

structure whereas lipids with a conical shape can adopt non-bilayer

phases such as the hexagonal HII phase.

This journal is � The Royal Society of Chemistry 2005 Mol. BioSyst., 2005, 1, 203–212 | 205

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

was important, binding of an extrinsic membrane protein, by

discharging some of the stored elastic energy, would alter the

conformational states of any intrinsic membrane proteins that

also depended on stored elastic energy (Fig. 4), which would

clearly be undesirable, making the behaviour of the membrane

difficult to control. Perhaps intrinsic membrane proteins

have evolved so that they are not sensitive to the kinds of

change in membrane bulk properties to which they are likely to

be exposed. If membrane proteins are affected largely by

interactions with their neighbouring lipid molecules then it

becomes important to have techniques for studying lipid–

protein interactions.

How to study lipid–protein interactions

X-ray crystallography. The most direct method for observing

lipid–protein interactions is by X-ray crystallography.

Unfortunately, crystals of membrane proteins are grown from

detergent solutions, and most lipid molecules are either lost in

the detergent solution or are not sufficiently ordered in the

protein crystal to be observable. Even when lipid molecules are

observed, electron density maps often only correspond to

partial lipid molecules and there is then a real possibility for

confusion between molecules of lipid and molecules of

detergent. There is also the problem that many of the lipid

molecules reported in crystal structures have unusual con-

formations, raising the possibility of inadequate refinement.25

Nevertheless, the observation of lipid molecules in high

resolution structures of membrane proteins has already told

us a lot about the ways in which lipid molecules can bind to

membrane proteins.14

The crystal structure of bacteriorhodopsin is unique in

including a large number of lipid molecules,26,27 these

molecules being retained presumably because of the special

properties of bacteriorhodopsin: bacteriorhodopsin occurs as a

trimer in the quasi-crystalline purple membrane of the

bacterium Halobacterium salinarum, where the molar ratio of

lipid to protein is unusually low, with about 30 lipid molecules

per bacteriorhodopsin trimer.28 Some of these lipid molecules

can be seen in the crystal structure (Fig. 1) forming a shell

or annulus around the protein; in the native membrane

the whole of the hydrophobic surface of the trimer would, of

course, be covered by lipid molecules. Many of the lipid fatty

acyl chains are located in distinct grooves on the surface of the

protein, but the lipid headgroups are not resolved in the

structure,26,27 suggesting that there is considerable disorder in

the headgroup region. These lipid molecules, interacting

relatively non-specifically with the protein, act as ‘solvent’

for the protein and are referred to as boundary or annular

phospholipids.

More generally, few if any lipid molecules are resolved

in crystal structures of membrane proteins, those that are

observed presumably being unusually tightly bound. These

retained lipid molecules are often located at protein–protein

interfaces in oligomeric proteins, and are often essential for

activity; they have been referred to as non-annular or ‘co-

factor’ lipids.14 Typical of a non-annular lipid is the

phosphatidylethanolamine molecule bound to the photo-

synthetic reaction centre from Thermochromatium tepidum29

shown in Fig. 5. The conformation adopted by the lipid

headgroup is very different to that adopted in a crystal of the

lipid alone,30 the amine group of the lipid in the T. tepidum

structure being folded down, allowing the phosphate group to

interact with adjacent Lys and Arg groups. The selectivity of

binding at a site of this type will depend on the size of the

headgroup, on its charge and on its ability to take part in

hydrogen bonding. The functional importance of the bound

phosphatidylethanolamine molecule does not appear to have

been studied for the reaction centre of T. tepidum. However,

the presence of a cardiolipin molecule resolved in the

crystal structure of the photosynthetic reaction centre from

Rhodobacter sphaeroides has been shown to be important for

the thermal stability of the protein.31 Similarly, mutation of

the Lys residues involved in binding cardiolipin to the

cytochrome bc1 complex of yeast leads to reduced levels of

the bc1 complex in the membrane, suggesting either that

binding of cardiolipin results in increased thermal stability for

the complex, or that it is required for the proper assembly of

the complex.32

Fig. 5 Structures of phosphatidylethanolamine free (A) and bound to

the photosynthetic reaction centre from T. tepidum (B). (B) also shows

protein residues interacting with the lipid headgroup (PDB file 1EYS).

Fig. 4 Possible effects of stored curvature elastic energy on the

function of membrane proteins. In (A) the binding of an extrinsic

membrane protein is increased in the presence of lipids favouring the

hexagonal HII phase, as the fatty acyl chains of neighbouring lipid

molecules can distort to fill the free volume created by insertion of

the extrinsic membrane protein into the lipid headgroup region. In (B)

the presence of lipids favouring the hexagonal HII phase shifts the

conformational equilibrium of an intrinsic membrane protein towards

the conformation with the greatest hydrophobic thickness.

206 | Mol. BioSyst., 2005, 1, 203–212 This journal is � The Royal Society of Chemistry 2005

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

Electron spin resonance

A disadvantage of X-ray crystallography as a way of reporting

on lipid–protein interactions is that X-ray crystallography does

not report on the protein in its native lipid bilayer environ-

ment. For that, electron spin resonance (ESR) and fluores-

cence spectroscopy have proved to be highly informative. ESR

studies make use of phospholipid molecules with nitroxide spin

labels attached to selected positions in the fatty acyl chains.

ESR spectra of spin labelled lipids in native membranes or

reconstituted lipid–protein systems show the presence of a

subpopulation of highly immobilized spin labels, not present in

protein-free membranes.33–35 This subpopulation corresponds

to lipid molecules whose rotational mobility is impeded by

interaction with the protein. The term immobilized is used to

indicate that the ESR spectrum is that which would be seen in

a powder; that is, it corresponds to a random array of spin

labels moving only slowly. The presence of the rigid protein

surface reduces the extent of the motional fluctuations of the

lipid fatty acyl chains and the chains have to tilt and become

conformationally disordered to pack well with the surface of

the protein; good packing between the lipid and protein

molecules in the membrane is important because poor packing

would prevent the membrane from being an effective perme-

ability barrier.

The ESR approach can be used to estimate the number of

lipid molecules binding to the surface of a membrane

protein.34,35 In a series of studies, Marsh and others have

shown that the number of bound lipid molecules fits reason-

ably well to the expected circumference of the transmembrane

region of the protein,25,33 providing strong evidence for the

presence of a distinct annular shell of lipid molecules around

each membrane protein.

ESR studies also report on the length of time that a lipid

molecule remains in the annular shell. To observe two distinct

environments for a lipid in a membrane on the ESR timescale

requires that the time taken for a lipid molecule to exchange

between the annular shell and the bulk phase be long on the

ESR timescale, which is about 1028 s. This requirement is met

at low temperatures, but, at temperatures closer to physio-

logical temperatures, rates of exchange become appreciable

and have characteristic effects on the ESR spectra that can be

used to obtain on and off rate constants at the protein surface;

the on rate constant is diffusion controlled and the off rate

constant reflects any specificity in binding.33,36,37 Off rates

for phosphatidylcholines are typically about 1–2 6 107 s21 at

30 uC.33 This is significantly slower than the rate of exchange

of two lipid molecules in the bulk phase resulting from trans-

lational diffusion in the membrane (ca. 8 6 107 s21 at 30 uC).

Thus it appears that the off rate is lowered by a slightly

more favourable lipid–protein interaction than lipid–lipid

interaction. The differences are, however, relatively small,

suggesting that the lipid–protein interaction is a non-sticky

one, consistent with the observation that lipid–protein

binding constants depend rather weakly on lipid structure, as

described later.

It has sometimes been suggested that annular lipid could

only affect the function of a membrane protein if the lifetime

of a lipid molecule in the annular shell around the protein is

long compared to the timescale for the functionally important

conformational changes of the protein. However, this is not so;

it does not matter which particular lipid molecule is in the

annular shell around a protein; swopping one molecule of a

lipid for another molecule of the same lipid will not affect

the function of the protein. Rapid exchange of the lipids can

average the environment sensed by the lipid but will not

average the environment sensed by the protein; the environ-

ment sensed by the protein (the annular lipid) is the same

however fast the lipids exchange.

Fluorescence spectroscopy

A disadvantage of the ESR technique with spin labelled

lipids is that it gives only an averaged picture of the interaction

between a membrane protein and its surrounding lipid

molecules and will not detect any heterogeneity in binding.

Here fluorescence spectroscopy has an advantage. The

fluorescence method measures the quenching of the fluores-

cence of Trp residues in a membrane protein caused by

phospholipids containing nitroxide-labelled fatty acyl chains

or brominated fatty acyl chains. Phospholipids containing

brominated fatty acyl chains are easily prepared by addition

of bromine across the cis-double bonds in a phospholipid

containing two mono-unsaturated fatty acyl chains;38 phos-

pholipids containing brominated fatty acyl chains behave

much like conventional phospholipids with unsaturated fatty

acyl chains because the bulky bromine atoms have effects on

lipid packing that are similar to those of a cis double bond.10

Quenching of the fluorescence of a Trp residue in a membrane

protein by a brominated phospholipid requires the brominated

chains to be close to the Trp residue in the protein. The

mechanism of quenching of Trp fluorescence by brominated

molecules is not clear, and could either be by heavy atom

quenching, which requires contact between the Trp and

bromine, or by fluorescence energy transfer, with a distance

Ro for which transfer is 50% efficient of ca. 8 A.39 Bolen and

Holloway40 showed that quenching fitted to a sixth-power

dependence on the distance of separation between the Trp

residue and the bromine, as in Forster energy transfer, but

Lodokhin41 showed that a similar distance dependence would

be expected for a collisional model for quenching when

account was taken of the depth distributions of the fluoro-

phore and quencher in the membrane.

Because the fluorescence lifetime for tryptophan is con-

siderably less than the time for two lipids to exchange position

in a bilayer (see above), fluorescence quenching gives an

essentially static picture of the membrane. Therefore, if an

intrinsic membrane protein is considered to have a set of

annular binding sites located around its circumference (Fig. 6),

the level of fluorescence quenching for a membrane protein

reconstituted into a bilayer containing brominated phos-

pholipid molecules will be proportional to the probability that

a brominated lipid molecule occupies a site close enough to a

Trp residue to cause quenching.42,43 Given that the distribu-

tion of lipid molecules in a mixture of two species of lipid in

the liquid crystalline phase is close to random,44,45 the level of

quenching observed for a Trp-containing protein in a mixture

of a non-brominated lipid and a brominated lipid will depend

This journal is � The Royal Society of Chemistry 2005 Mol. BioSyst., 2005, 1, 203–212 | 207

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

on the binding constant for the brominated lipid compared to

that for the non-brominated lipid.42,43 The method relies, of

course, on having Trp residues in the transmembrane region of

the membrane protein, but this will usually be the case since

Trp is a hydrophobic residue.

Fig. 6 shows lipid binding constants measured in this way

for OmpF and MscL, demonstrating that lipid binding

constants vary with fatty acyl chain length and so with the

hydrophobic thickness of the lipid bilayer. The hydrophobic

thickness of a membrane protein would be expected to match

that of the surrounding lipid bilayer because the cost of

exposing either fatty acyl chains or hydrophobic amino acids

to water is very high. Any potential mismatch between the

hydrophobic thicknesses of the lipid bilayer and the protein

will lead to distortion of the lipid bilayer, or the protein, or

both, to minimize the mismatch. The high efficiency of

hydrophobic matching between a membrane protein and the

surrounding lipid bilayer has been demonstrated experimen-

tally for the potassium channel KcsA where varying the fatty

acyl chain length for the surrounding phospholipids from

C12 to C24 results in no change in the environment of the Trp

residues located at the ends of the transmembrane a-helices.46

Distortion of either the lipid or the protein to achieve

hydrophobic matching will require work. A lipid that has to

distort in order to bind to a protein will show a lower binding

constant than a lipid that can bind to the protein without

distortion. The lipid showing strongest binding to a particular

membrane protein would therefore be expected to be that

giving a bilayer with a hydrophobic thickness equal to the

hydrophobic thickness of the protein. The fact that the optimal

chain length for binding to OmpF is shorter than that for

binding to MscL (Fig. 6) is consistent with the observation

that the bacterial outer membrane is thinner than the bacterial

inner membrane.47

The energetics of distortion of a lipid bilayer around a rigid

membrane protein have been analysed in terms of the bulk

bending properties of the lipid bilayer, with stretching of the

lipid chains being required for hydrophobic matching when

the lipid bilayer is too thin and compression being required

when the bilayer is too thick.48–50 A comparison of relative

lipid binding constants estimated from the results of such a

theoretical calculation48 with experimental data shows that

agreement is reasonable for moderate levels of mismatch for

the b-barrel protein OmpF but is poor for the a-helical protein

MscL39,47 (Fig. 6). This suggests that the b-barrel structure of

OmpF is relatively rigid so that distortion of the lipid bilayer

to provide hydrophobic matching is less costly than distortion

of the protein. In contrast, MscL is less rigid, with distortion of

both the lipid bilayer and the protein occurring to produce

hydrophobic matching,39 because the cost of distorting a lipid

bilayer is relatively high.51 Distortion of a-helical membrane

proteins explains the marked dependence of the activities of

a-helical membrane proteins on bilayer thickness illustrated

in Fig. 2.22

Relative lipid binding constants also depend on lipid

headgroup structure, although again effects are rather small,

as shown in Table 2 for binding of lipids to Ca2+-ATPase.10

Binding of a phosphatidylethanolamine is a factor of 2 weaker

than binding of the equivalent phosphatidylcholine and the

anionic lipid phosphatidylserine binds as strongly to Ca2+-

ATPase as phosphatidylcholine in the absence of Ca2+, but a

factor of two weaker in the presence of Ca2+. In part, weak

binding of phosphatidylserine in the presence of Ca2+ is

because, in the presence of Ca2+, phosphatidylserines form

gel-like domains of (PS)2Ca;52 Ca2+-ATPase interacts weakly

with lipid in the gel-phase (Table 2), presumably due to poor

packing between the rough surface of the protein and the rigid

fatty acyl chains.

The fact that lipid binding constants vary rather little with

headgroup structure suggest that the lipid–protein interaction

is a non-specific one, involving the same kinds of charge

and hydrogen-bonding interactions that are important

for lipid–lipid interactions. The importance of hydrogen

bonding for lipid–protein interactions is shown in mole-

cular dynamics simulations comparing MscL in bilayers of

Fig. 6 Relative lipid binding constants determined using fluorescence

quenching by a brominated phospholipid. (A). Lipid binding sites on

the transmembrane surface of a membrane protein. Two lipid

molecules, A and B, are shown exchanging at one ‘site’. In this case,

two lipid binding sites are close enough to a Trp residue in the protein

to result in quenching when occupied by brominated lipid molecules.

(B). The dependence of relative lipid binding constant K on chain

length. The chain length dependencies of the binding constants K for

phosphatidylcholines relative to the strongest binding lipid are plotted

for MscL (#) and the b-barrel protein OmpF (%). The dotted line

shows the theoretical dependence of lipid binding constant on chain

length calculated from the data of Fattal and Ben-Shaul48 for a protein

of hydrophobic thickness 30 A, as described in Powl et al.39 Data from

refs. 39 and 47.

Table 2 Relative lipid binding constants of phospholipids toCa2+-ATPase

LipidRelative associationconstanta,b

Phosphatidylethanolamine 0.45Phosphatidylserine 1.0Phosphatidylserine + 10 mM Ca2+ 0.45Dimyristoylphosphatidylcholine in gel phase 0.04a Measured relative to dioleoylphosphatidylcholine b Data fromref. 10.

208 | Mol. BioSyst., 2005, 1, 203–212 This journal is � The Royal Society of Chemistry 2005

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

phosphatidylethanolamine and phosphatidylcholine; fewer

hydrogen bonds are formed with phosphatidylcholine than

with phosphatidylethanolamine, the decrease in the number of

hydrogen bonding interactions being compensated for by a

conformational change in the C-terminal region of MscL

in bilayers of phosphtaidylcholine, bringing the C-terminal

region closer to the membrane, leading to stronger interactions

with the membrane.53

Although most of the annular sites around a membrane

protein show relatively little specificity for lipid type, the

possibility exists that a small number of sites may show greater

specificity. When a membrane protein contains a large number

of Trp residues distributed around the circumference of the

protein, the fluorescence quenching method will give an

averaged lipid binding constant for all the annular sites.

However, the use of site directed mutagenesis allows the

generation of proteins containing single Trp residues, so that

lipid binding constants can be measured at particular locations

around the circumference, defined by the position of the Trp

residue in the protein. MscL has the advantage for such studies

that the native protein contains no Trp residues; mutating

Leu-69 to Trp then allows the determination of lipid binding

constants on the extracellular side of the membrane and

mutating Tyr-87 to Trp allows the determination of lipid

binding constants on the intracellular side of the membrane

(Fig. 7).54 These experiments show that on the extracellular

side of the membrane anionic and zwitterionic lipids bind

with equal affinity, consistent with the lack of charged

residues on the protein in a position able to interact with the

lipid headgroups. However, on the intracellular side of the

membrane, anionic lipids bind close to residue 87 with

an affinity much higher than that of zwitterionic lipids, this

‘hot-spot’ for anionic lipid binding corresponding to a cluster

of three positively charged residues, Arg-98, Lys-99 and

Lys-100 (Fig. 7).

Fluorescence quenching methods, combined with site

directed mutagenesis, can also be used to study the binding

of lipids at non-annular sites on the protein. As shown in

Fig. 1, KcsA binds one anionic lipid molecule at each of the

four protein–protein interfaces in the tetrameric structure.

KcsA contains five Trp residues per monomer, two on the

intracellular side of the membrane and three on the extra-

cellular side, where the non-annular binding site is located.55

To study lipid binding at the non-annular site the two Trp

residues on the intracellular side were mutated to Leu, leaving

just the three Trp residues on the extracellular side (Marius,

Alvis, East and Lee, unpublished) (Fig. 8). Of these, Trp-87 is

close to the boundary of the protein and so will be quenched

by brominated lipid bound to the annular sites but is too far

from the non-annular site to be quenched by a brominated

lipid binding to the non-annular site, Trp-67 is close to the

non-annular site and so will be quenched from here, but not

from the annular sites, and Trp-68 is too far from both the

non-annular site and from the annular sites to be quenched

from either. The level of fluorescence quenching observed

with brominated anionic lipid is double that observed with

brominated phosphatidylcholine, showing that whereas phos-

phatidylcholines can only bind to annular sites and quench the

fluorescence of Trp-87, brominated anionic lipids can bind to

both annular and non-annular sites and thus quench the

fluorescence of both Trp-87 and Trp-68; these results show

that the non-annular binding site is specific for anionic lipid.

Measurements of the level of quenching in mixtures of

brominated anionic lipid and non-brominated phosphatidyl-

choline give the binding constant for the anionic lipid at the

non-annular site (Fig. 8).56 The presence of anionic lipid is

essential for the function of KcsA; KcsA only opens in the

presence of anionic lipid.57,58

Molecular dynamics simulations

Molecular motions in a lipid bilayer extend in time from the

picosecond range characteristic of C–C bond rotations in lipid

fatty acyl chains to the microseconds range and slower

required to describe the diffusion of lipids in the plane of the

membrane. The extent of motion varies from a few A to a

fraction of a micrometer, required to describe collective

Fig. 7 The structure of the MscL pentamer. (A) Shown in space-fill

format are the residues Leu-69, Phe-80 and Tyr-87 mutated to Trp

residues, and the charge residues Arg-98, Lys-99 and Lys-100. The

side chain of Lys-100 is not resolved in the crystal structure, and has

been modelled in for illustrative purposes. The horizontal lines

represent the probable location of the hydrocarbon core of the

surrounding lipid bilayer defined by the experiments shown in B. (B).

Trp fluorescence emission maxima (nm) for Trp mutants are plotted as

a function of position, for MscL reconstituted in di(C18:1)PC. The

dotted line at 332.6 nm marks the expected fluorescence emission

maximum for a Trp residue immediately below the glycerol backbone

region of the bilayer.

This journal is � The Royal Society of Chemistry 2005 Mol. BioSyst., 2005, 1, 203–212 | 209

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

motions of large numbers of lipid molecules leading to

undulations of the bilayer. Simulations of small bilayer patches

over a period of a few tens of nanoseconds will therefore not be

able to provide a full description of motion in a lipid bilayer.

Simulations of lipid–protein systems also have to solve the

problem of how best to insert a membrane protein into a

lipid bilayer.59,60 Nevertheless, molecular dynamics simula-

tions can provide an insight into the behaviour of the system

not available in any other way.

As expected, molecular dynamics simulations show that

insertion of a membrane protein into a lipid bilayer influences

the bilayer structure, and that the observed effects vary with

the properties of the inserted protein. In some systems addition

of protein leads to an increase in order parameters for the lipid

fatty acyl chains, in some cases to a decrease. The simulations

also show that effects are generally restricted to the layer of

lipid immediately around the protein, consistent with the

idea of a lipid annulus. Gramicidin is a particularly popular

protein for study because of its relative simplicity. Gramicidin

dimerises to form a channel across the membrane.61 Experi-

ments show that gramicidin disorders lipid fatty acyl chains in

the gel phase but increases order in the liquid crystalline

phase,62,63 increased order for the chains in the liquid

crystalline phase following from a restriction of the ampli-

tude of motion for the chains.64 These simple statements,

however, hide a considerable complexity, demonstrated in

molecular dynamics simulations of gramicidin in liquid

crystalline dimyristoylphosphatidylcholine.65,66 In simulations

at a 1 : 50 molar ratio of gramicidin to lipid, it was possible to

distinguish between phosphatidylcholine molecules next to the

channel (phospholipids that were either hydrogen bonded

directly to the channel or via one intervening water molecule)

and the ‘bulk’ phospholipid, not hydrogen bonded to the

channel.66 The presence of the channel was found to have

no effect on the properties of the bulk phospholipid but

increased the order parameters for the fatty acyl chains of the

phospholipids bound to the channel. The increase in order

parameter for the bound phospholipids corresponded to an

increase in the trans–gauche ratio by 27% and 7% for the sn-1

and sn-2 chains, respectively. The average hydrophobic

thickness for the bilayer next to the channel, defined as the

carbonyl–carbonyl distance across the bilayer, was 8% greater

than that for the bulk lipid, this increase in thickness following,

of course, from the increase in order parameter.

One important result from the simulations was that the

effects of the channel on the lipid bilayer were short range,

affecting only those phospholipids bound to the channel.66

Another important result that emerged from these simulations

was that the range of interaction energies between the bound

phospholipids and the channel was very broad; the energies of

individual phospholipid–protein interactions fluctuated over a

very wide range on a timescale of 50–500 ps.65 The fluctuations

arose because the total interaction energy between the

phospholipid and gramicidin molecules was the sum of many

weak van der Waals and electrostatic interactions; there was

no single deep energy well into which the phospholipid fell to

give a single favoured conformation so that lipid molecules

were not frozen in a single long-lived conformation on the

protein surface.

Rather similar conclusions can be drawn from simulations

of other membrane proteins. A simulation of the KcsA

tetramer in bilayers of 1-palmitoyl-2-oleoylphosphatidylcho-

line showed that about 30 lipid headgroups made contact with

the protein whereas about 40 lipid molecules made contact

through their fatty acyl chains.67 From the size of the KcsA

tetramer it can be estimated that about 26 lipid molecules

would be required to form a complete annular shell around the

protein.46 This is in good agreement with the number of lipid

headgroups in contact with KcsA estimated in the molecular

dynamics simulation. The greater number of lipid molecules

contacting KcsA through their fatty acyl chains than through

their headgroups is consistent with the decrease in order down

the fatty acyl chain, giving a picture where the headgroup and

the glycerol backbone region of the lipid are located snugly

against the protein surface, with the ends of the chains being

able to stray away from the protein surface, being replaced by

chains from lipid molecules whose headgroups are not bound

to the protein. Interactions of the lipid headgroup and glycerol

backbone regions with KcsA were dominated by hydrogen

bonding interactions between the acyl carbonyl groups and the

headgroup phosphate group, some of these hydrogen bonds

Fig. 8 Measuring the binding constant at the non-annular site on

KcsA. (Top) Location of Trp residues on the extracellular side of

KcsA. The cross-hatched surface is the view of the KcsA from the

extracellular side. Shown are the locations of the Trp residues W67,

W68 and W87 around one of the non-annular binding sites; the lipid

molecule occupying the site has been modelled as a diacylglycerol

(DAG) since the headgroup of the lipid is not resolved. (PDB file

1K4C). (Bottom) Fluorescence quenching curves for the Trp-mutant

of KcsA reconstituted into mixtures of phosphatidylcholine with

brominated phosphatidylcholine (#) and with brominated phospha-

tidylglycerol ($). The solid lines show fits to binding equations, with a

binding constant at the non-annular site for phosphatidylglycerol of

3.0 ¡ 0.7 mole fraction21 (Marius, Alvis, East and Lee, unpublished).

210 | Mol. BioSyst., 2005, 1, 203–212 This journal is � The Royal Society of Chemistry 2005

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

being relatively long lived (2–5 ns or longer) but some being

transient with a lifetime of 0.1 ns or less.67 The diffusion

coefficient describing lateral mobility for the bound lipids was

estimated to be about half that for the bulk lipids67 in good

agreement with the ESR results described above.

In a simulation of bacteriorhodopsin in bilayers of phos-

phatidylcholine the lipid molecules were seen to tilt and

become conformationally disordered to allow them to nestle

against the rough surface of the protein.68 The seven helices in

bacteriorhodopsin were found to have distinctly different

energies of interaction with the surrounding lipid bilayer, the

second and sixth helices interacting especially strongly with the

lipid molecules because of electrostatic interactions between

positively charged amino acid resides in these helices and

the negatively charged phosphate in the lipid headgroup.69 In

other words, the lipid annulus is not homogeneous, the

molecules of phosphatidylcholine interacting more strongly

at some sites than at others. A simulation of the related seven-

helix protein rhodopsin in bilayers of dioleoylphosphatidyl-

choline again showed marked differences between the energies

of interaction of the seven helices with the lipid bilayer.70 As

with the simulation of gramicidin described above, the range of

helix-lipid interaction energies was very broad, fluctuating

markedly with time.70 A simulation of rhodopsin in bilayers of

1-palmitoyl-2-oleoylphosphatidylcholine found that the pre-

sence of the protein resulted in a decrease in order parameters

for the lipid palmitoyl chain71 although a simulation of the

effects of a bundle of five transmembrane a-helices in a bilayer

of dimyristoylphosphatidylcholine found that the presence

of the protein resulted in an increase in fatty acyl chain

order parameters.72 The rhodopsin simulation showed

that the bilayer did not have a uniform thickness around the

circumference of the rhodopsin molecule, the thickness of the

hydrophobic core of the bilayer being about 3 A greater close

to the second transmembrane a-helix than to the sixth or

seventh helices.71 It was also observed that the ends of the

transmembrane a-helices were generally located in the lipid

headgroup region so that the loops connecting the transmem-

brane a-helices were located outside the membrane.71

A simulation of rhodopsin in bilayers of 1-stearoyl-2-

docosohexaenoyl-phosphatidylcholine showed that the poly-

unsaturated docosohexaenoyl (DHA) chain bound more

deeply in the protein surface than did the saturated stearoyl

chain.73,74 The greater energy of interaction with the DHA

chain followed from better van der Waals contact with the

surface, resulting from the low rotational barriers to isomer-

ization around the methylene groups connecting the vinyl

groups in the DHA chain.73 It is not yet known whether

preferential solvation by polyunsaturated fatty acyl chains is

unique to rhodopsin or will be found to be a feature of all

membrane proteins.

Membrane proteins in their membrane environments

A final problem concerns how a membrane protein ‘sits’ in the

surrounding lipid bilayer. This cannot be determined directly

from the crystal structure since crystal structures report on

membrane proteins in a detergent environment rather than in a

lipid bilayer. One way to identify the position of a lipid bilayer

around a membrane protein is by fluorescence spectroscopy,

making use of the environmental sensitivity of Trp fluores-

cence emission. The method is illustrated in Fig. 7 for MscL.75

Trp residues were introduced into each of the lipid-exposed

residues in the second transmembrane a-helix of MscL and

the fluorescence emission maxima were determined for the

mutants reconstituted into bilayers of dioleoylphosphatidyl-

choline. From experiments with KcsA where the Trp residues

are located at the ends of the transmembrane a-helices, it is

known that a Trp residue located in the glycerol backbone

region of the bilayer emits at 332.6 nm46 so that the interfacial

residues (the residues located close to the glycerol backbone

regions on the two sides of the lipid bilayer) in MscL can be

read off as Leu-69 on the extracellular side and Leu-92 on the

intracellular side (Fig. 7). Leu-69 is close to Asp-68, suggesting

that the carboxyl oxygens of Asp-68 are located close to the

glycerol backbone region of the bilayer and are responsible for

determining the position of the bilayer on the extracellular

side; an analysis of a number of membrane crystal structures

suggests that carboxyl groups might often be located at the

interface.14 The interface on the intracellular side is determined

by the positions of Arg-11 and Asp-16; the hydrophobic

thickness of MscL determined in this way is ca. 25 A, in good

agreement with the observation that the phosphatidylcholine

that binds most strongly to MscL is that with a chain length of

C16 (Fig. 6), a chain length that gives a bilayer of hydrophobic

thickness of ca. 24 A.75

Summary

In summary, a combination of X-ray crystallography, electron

spin resonance and fluorescence spectroscopies, and molecular

dynamics simulations is starting to clarify the interactions

between intrinsic membrane proteins and the lipid bilayer. The

bulk of the lipids surrounding a membrane protein interact

with the protein rather non-specifically but, nevertheless, are

important in determining the structure and function of

the protein, in this way acting like a typical ‘solvent’ for the

protein. Because these sites are relatively non-specific, the

composition of the annulus around a membrane protein will

be similar to the bulk lipid composition of the membrane.

However, evidence is now emerging for the existence of

‘hot-spots’ on the surface of membrane proteins showing

marked selectivity for anionic phospholipids, and for selective

lipid binding sites located between transmembrane a-helices

where binding of lipids can have marked effects on the

function of the protein. The ability to manipulate these

interactions using molecular biological approaches means that

we should soon be in a position to understand the importance

of these interactions for the function of membrane proteins in

real biological membranes.

References

1 M. A. Roseman, J. Mol. Biol., 1988, 201, 621.2 S. W. Cowan, T. Schrimer, G. Rummel, M. Steiert, R. Ghosh,

R. A. Pauptit, J. N. Jansonius and J. P. Rosenbusch, Nature, 1992,358, 727.

3 G. Chang, R. H. Spencer, A. T. Lee, M. T. Barclay and D. C. Rees,Science, 1998, 282, 2220.

This journal is � The Royal Society of Chemistry 2005 Mol. BioSyst., 2005, 1, 203–212 | 211

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online

4 Y. Zhou, J. H. Morals-Cabral, A. Kaufman and R. Mackinnon,Nature, 2001, 414, 43.

5 G. W. Gould, J. M. McWhirter, J. M. East and A. G. Lee,Biochem. J., 1987, 245, 751.

6 A. G. Lee, Prog. Lipid Res., 1991, 30, 323.7 E. London, Curr. Opin. Struct. Biol., 2002, 12, 480.8 A. G. Lee, Curr. Biol., 2005, 15, R421.9 E. London and G. W. Feigenson, Biochemistry, 1981, 20, 1939.

10 J. M. East and A. G. Lee, Biochemistry, 1982, 21, 4144.11 S. Mall, R. Broadbridge, R. P. Sharma, A. G. Lee and J. M. East,

Biochemistry, 2000, 39, 2071.12 S. Mall, R. Broadbridge, R. P. Sharma, J. M. East and A. G. Lee,

Biochemistry, 2001, 40, 12379.13 M. E. Fastenberg, H. Shogomori, X. L. Xu, D. A. Brown and

E. London, Biochemistry, 2003, 42, 12376.14 A. G. Lee, Biochim. Biophys. Acta, 2003, 1612, 1.15 J. D. Pilot, J. M. East and A. G. Lee, Biochemistry, 2001, 40,

14891.16 A. P. Starling, K. A. Dalton, J. M. East, S. Oliver and A. G. Lee,

Biochem. J., 1996, 320, 309.17 K. A. Dalton, J. M. East, S. Mall, S. Oliver, A. P. Starling and

A. G. Lee, Biochem. J., 1998, 329, 637.18 F. Zhou and K. Schulten, J. Phys. Chem., 1995, 99, 2194.19 K. V. Damodaran and K. M. Merz, Biophys. J., 1994, 66, 1076.20 S. M. Gruner, Proc. Natl. Acad. Sci. U. S. A., 1985, 82, 3665.21 P. R. Cullis and B. de Kruijff, Biochim. Biophys. Acta, 1979, 559,

399.22 A. G. Lee, Biochim. Biophys. Acta, 2004, 1666, 62.23 G. S. Attard, R. H. Templer, W. S. Smith, A. N. Hunt and

S. Jackowski, Proc. Nat. Acad. Sci. U. S. A., 2000, 97, 9032.24 A. V. Botelho, N. J. Gibson, R. L. Thurmond, Y. Wand and

M. F. Brown, Biochemistry, 2002, 41, 6354.25 D. Marsh and T. Pali, Biochim. Biophys. Acta, 2004, 1666, 118.26 H. Belrhali, P. Nollert, A. Royant, C. Menzel, J. P. Rosenbusch,

E. M. Landau and E. Pebay-Peyroula, Structure, 1999, 7, 909.27 H. Luecke, B. Schobert, H. T. Richter, J. P. Cartailler and

J. K. Lanyi, J. Mol. Biol., 1999, 291, 899.28 A. Corcelli, V. M. T. Lattanzio, G. Mascolo, P. Papadia and

F. Fanizzi, J. Lipid Res., 2002, 43, 132.29 T. Nogi, I. Fathir, M. Kobayashi, T. Nozawa and K. Miki, Proc.

Nat. Acad. Sci. U. S. A., 2000, 97, 13561.30 I. Pascher, M. Lundmark, P. G. Nyholm and S. Sundell, Biochim.

Biophys. Acta, 1992, 1113, 339.31 P. K. Fyfe, N. W. Isaacs, R. J. Cogdell and M. R. Jones, Biochim.

Biophys. Acta, 2004, 1608, 11.32 C. Lange, J. H. Nett, B. L. Trumpower and C. Hunte, EMBO J.,

2001, 20, 6591.33 D. Marsh and L. I. Horvath, Biochim. Biophys. Acta, 1998, 1376,

267.34 P. C. Jost and O. H. Griffith, Methods Enzymol., 1978, 49, 369.35 J. K. Brotherus, O. H. Griffith, M. O. Brotherus, P. C. Jost and

J. R. Silvius, Biochemistry, 1981, 20, 5261.36 J. Davoust and P. F. Devaux, J. Magn. Reson., 1982, 48, 475.37 J. M. East, D. Melville and A. G. Lee, Biochemistry, 1985, 24,

2615.38 J. Carney, J. M. East, S. Mall, P. Marius, A. M. Powl, J. N. Wright

and A. G. Lee, Current Protocols in Protein Science, ed. J. Coligan,B. Dunn, H. Ploegh, D. Speicher, and P. Wingfield, John Wiley,in press.

39 A. M. Powl, J. M. East and A. G. Lee, Biochemistry, 2003, 42,14306.

40 E. J. Bolen and P. W. Holloway, Biochemistry, 1990, 29, 9638.

41 A. S. Ladokhin, Biophys. J., 1999, 76, 946.42 E. London and G. W. Feigenson, Biochemistry, 1981, 20, 1939.43 S. Mall, R. P. Sharma, J. M. East and A. G. Lee, Faraday Discuss.,

1998, 111, 127.44 A. G. Lee, Biochim. Biophys. Acta, 1977, 472, 285.45 D. Marsh, CRC Handbook of Lipid Bilayers. CRC Press, Boca

Raton, 1990.46 I. M. Williamson, S. J. Alvis, J. M. East and A. G. Lee, Biophys. J.,

2002, 83, 2026.47 A. H. O’Keeffe, J. M. East and A. G. Lee, Biophys. J., 2000, 79,

2066.48 D. R. Fattal and A. Ben-Shaul, Biophys. J., 1993, 65, 1795.49 C. Nielsen, M. Goulian and O. S. Andersen, Biophys. J., 1998, 74,

1966.50 O. G. Mouritsen and M. Bloom, Biophys. J., 1984, 46, 141.51 J. A. Lundbaek, P. Birn, A. J. Hansen, R. Sogaard, C. Nielsen,

J. Girshman, M. J. Bruno, S. E. Tape, J. Egebjerg, D.V. Greathouse, G. L. Mattice, R. E. Koeppe and O. S. Andersen,J. Gen. Physiol., 2004, 123, 599.

52 H. Hauser and G. G. Shipley, Biochemistry, 1984, 23, 34.53 D. E. Elmore and D. A. Dougherty, Biophys. J., 2003, 85, 1512.54 A. M. Powl, J. M. East and A. G. Lee, Biochemistry, 2005, 44,

5873.55 I. M. Williamson, S. J. Alvis, J. M. East and A. G. Lee, Cell. Mol.

Life Sci., 2003, 60, 1581.56 S. J. Alvis, I. M. Williamson, J. M. East and A. G. Lee, Biophys. J,

2003, 85, 1.57 L. Heginbotham, L. Kolmakova-Partensky and C. Miller, J. Gen.

Physiol., 1998, 111, 741.58 I. Valiyaveetil, Y. Zhou and R. Mackinnon, Biochemistry, 2002,

41, 10771.59 D. P. Tieleman and H. J. C. Berendsen, Biophys. J., 1998, 74, 2786.60 J. D. FaraldoGomez, G. R. Smith and M. S. P. Sansom, Eur.

Biophys. J., 2002, 31, 217.61 D. A. Doyle and B. A. Wallace, in Biomembranes. Volume 6

Transmembrane receptors and channels, ed. A. G. Lee, JAI Press,Greenwich, Connecticut, 1997, pp. 327–359.

62 J. A. Killian, Biochim. Biophys. Acta, 1992, 1113, 391.63 D. Rice and E. Oldfield, Biochemistry, 1979, 18, 3272.64 M. Ge and J. H. Freed, Biophys. J., 1993, 65, 2106.65 T. B. Woolf and B. Roux, Proteins: Struct., Funct., Genet., 1996,

24, 92.66 S. W. Chiu, S. Subramaniam and E. Jakobsson, Biophys. J., 1999,

76, 1929.67 S. S. Deol, P. J. Bond, C. Domene and M. S. P. Sansom, Biophys.

J., 2004, 87, 3737.68 O. Edholm, O. Berger and F. Jahnig, J. Mol. Biol., 1995, 250, 94.69 H. Jang, P. S. Crozier, M. J. Stevens and T. B. Woolf, Biophys. J.,

2004, 87, 129.70 P. S. Crozier, M. J. Stevens, L. R. Forrest and T. B. Woolf, J. Mol.

Biol., 2003, 333, 493.71 T. Huber, A. V. Botelho, K. Beyer and M. F. Brown, Biophys. J.,

2004, 86, 2078.72 L. Saiz, S. Bandyopadhyay and M. L. Klein, J. Phys. Chem. B,

2004, 108, 2608.73 S. E. Feller, K. Gawrisch and T. B. Woolf, J. Am. Chem Soc., 2003,

125, 4434.74 M. C. Pitman, A. Grossfield, F. Suits and S. E. Feller, J. Am.

Chem. Soc., 2005, 127, 4576.75 M. Powl, J. N. Wright, J. M. East and A. G. Lee, Biochemistry,

2005, 44, 5713.

212 | Mol. BioSyst., 2005, 1, 203–212 This journal is � The Royal Society of Chemistry 2005

Dow

nloa

ded

by I

ndia

na U

nive

rsity

- P

urdu

e U

nive

rsity

at I

ndia

napo

lis o

n 08

Sep

tem

ber

2012

Publ

ishe

d on

14

July

200

5 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/B

5045

27D

View Online