The Systematic Analysis of Protein

Nature Reviews Molecular Cell Biology | AOP, published online 28 October 2015; doi:10.1038/nrm4080 PERSPECTIVES

and sphingolipids also contain at least one TECHNOLOGY AND TECHNIQUES — INNOVATION fatty acid that can differ in length and/or in its level of unsaturation (which helps to The systematic analysis of protein– determine membrane thickness and fluid- ity), as well as a hydrophilic head group lipid interactions comes of age that extends out of the hydrophobic bilayer into the aqueous phase and can function as a signalling hub. The combination of these Antoine-Emmanuel Saliba, Ivana Vonkova and Anne-Claude Gavin various metabolic building blocks results Abstract | Lipids tailor membrane identities and function as molecular hubs in all in a diverse lipid landscape that is extensively exploited by proteins7. Proteins have cellular processes. However, the ways in which lipids modulate protein function adapted a variety of motifs and domains and structure are poorly understood and still require systematic investigation. In that can recognize and sense individual this Innovation article, we summarize pioneering technologies, including lipids and/or more general membrane lipid-overlay assays, lipid pull-down assays, affinity-purification lipidomics and the properties such as curvature, thickness or 7 liposome microarray-based assay (LiMA), that will enable protein–lipid interactions specialized microdomains . Furthermore, protein–lipid interactions can generally be to be deciphered on a systems level. We discuss how these technologies can be divided into three categories, according to applied to the charting of system-wide networks and to the development of new the mechanisms by which these molecules pharmaceutical strategies. recognize each other and the environment in which they are found (FIG. 1). Lipid localization is tightly regulated and charting protein–lipid interactions, and shapes the unique function and proper- a technological effort similar to the one Integral membrane proteins. An essential ties of every cellular membrane1,2. The made for DNA–protein and RNA–pro- type of protein–lipid interaction involves accumulation of discrete phosphoinositide, tein interactions is now needed for lipids. integral membrane proteins, which interact glycerophospholipid, sphingolipid or sterol We believe that a global atlas of protein– with specific lipids within the hydrophobic species defines the identity of the organelle lipid interactions will benefit biology and plane of the membrane. This type of inter- and locally regulates the organization of medicine by shedding light on the role of action is involved in targeting proteins to cellular membranes into microdomains. lipids with ‘orphan’ bioactive activity; by specific organelles8, and it can be used to Over the past few years, much effort has deciphering novel modes of action of lipids modulate protein structure, activity and been dedicated to cataloguing the ‘lipi- through their interactions with proteins; function9,10. For example, the protein p24, dome’, which is now estimated to contain and by understanding the misregulation of which is a component of the coat protein I more than 40,000 different lipid species lipids in disease. In this Innovation article, (COPI) machinery, has a transmembrane (see the LIPID MAPS Structure Database we summarize the recent technological segment that specifically and exclusively (LMSD))3; however, the biological role of advances that will allow global protein– binds to one sphingomyelin species, only a few of these lipids is known. lipid interactions to be mapped on a sys- SM18 (REF. 11). This interaction, which Lipids function predominantly through tems level and open up new pharmaceutical involves the head group and backbone of their interactions with proteins. Indeed, strategies. We also discuss the future needs the sphingolipid and a signature sequence protein–lipid interactions are involved in all and challenges involved in the integra- (VXXTLXXIY; where X indicates any biological processes and are of paramount tion of diverse, orthogonal approaches for amino acid) within the transmembrane interest in pharmaceutical discovery, as studying the regulation of proteins by lipids domain of p24, induces p24 oligomeriza- 60% of drug targets are located at the cell in biology and medicine. tion and activation to regulate COPI- surface or in other cellular membranes4,5. dependent vesicular transport. Integral Integral and peripheral membrane proteins The interaction landscape membrane proteins, however, are challeng- account for one-third of the full proteome6; Lipids are well known for their amphi- ing to study, and our current knowledge is however, research into the mechanisms that pathic nature and their capacity to form limited to a few prototypic examples that are used by lipids to regulate protein func- membrane bilayers. Their structures are have generally been investigated using tion and structure — and, ultimately, to diverse because of their distinct chemical advanced biochemical, biophysical or shape cell physiology — has been restricted backbones, which are composed of glycerol structural methods that cannot easily be to a small number of proteins, with a strong in the case of the glycerolipids and glyc- scaled up. More generic approaches — that focus on mechanistic insights. This gap in erophospholipids, sphingoid long-chain is, those that are applicable to all types of our knowledge is due to a lack of meth- bases for the sphingolipids and isoprene for protein–lipid interactions — are needed ods that are amenable to systematically the sterols (BOX 1). Glycerophospholipids (see below).

NATURE REVIEWS | MOLECULAR CELL BIOLOGY ADVANCE ONLINE PUBLICATION | 1

Box 1 | Liposomes: surrogates of biological membranes Peripheral membrane proteins. Another important type of interaction involves the Liposomes (also called phospholipid vesicles), which are closed vesicles comprising a bilayer of recruitment of soluble proteins to the cell 82,83 amphiphilic molecules, are in vitro mimics of natural membranes . A large variety of lipids are periphery of biological membranes. Several available from multiple suppliers (see Supplementary information S1 (table)), notably glycero domains recognize specific lipid head phospholipids, phosphoinositides, sphingolipids and sterols (the structures of the major lipid 12 species are shown in the figure, part a), although the commercially available lipid repertoire is still groups (for example, pleckstrin homology far from covering the range of natural lipids. Each lipid category is hierarchically organized around (PH) domains recognize phosphoinositide a core structure that is defined in a species on the basis of a modification of the head group phosphates and the C2 domain of lactadherin (for glycerophospholipids the modifications are labelled ‘R’ in the figure, part a) or a chemical recognizes phosphatidylserine) and/or spe- modification in the core structure (for example, in the sphingolipids and sterols). Lipids dissolved in cific membrane features (for example, epsin organic solvents can be stored in glass vials at -20 °C under argon or nitrogen, and the quality of the amino‑terminal homology (ENTH) and BAR initial lipid composition and any further degradation over time can be monitored, conveniently, domains recognize membrane curvature)12–14. by matrix-assisted laser desorption/ionization–time of flight (MALDI–TOF) mass spectrometry. Thus, lipids and other membrane attributes A thin-layer chromatography spotter can be used for the high-throughput and automatic handling function as second messengers to control the of lipids, because they are resistant to organic solvents16,39, but these robots are limited by poor spatiotemporal recruitment and activation of spatial resolution. 14 Liposomes are normally classified by their size and the number of bilayers that they contain specific protein effectors . A key example is (unilamellar liposomes have a single bilayer; multilamellar liposomes have multiple bilayers)82 and the Ser/Thr kinase AKT1, which is activated are usually designated as SUVs (small unilamellar vesicles, which have a diameter of <100 nm), LUVs by its recruitment to membranes through (large unilamellar vesicles, which have a diameter of ≤1 μm) or GUVs (giant unilamellar vesicles, which its specialized PH domain that recognizes have a diameter of >1 μm). The preparation of liposomes in general82, and of GUVs specifically83, phosphatidylinositol‑3,4,5‑trisphosphate has been reviewed elsewhere; several popular and successful methods are now used (see the figure, (PtdIns(3,4,5)P3) or PtdIns(3,4)P2; both lipids part b). These protocols are suitable for preparing a few liposome types, but they are difficult to scale are produced locally by phosphoinositol up because of major technical limitations: the protocols are time-consuming and require large 3‑kinase (PI3K)14. Domains involved in the amounts of starting material; each lipid mixture needs to be optimized; and the liposomes cannot be recruitment of proteins to the periphery of stored long-term. The introduction of a method that allows GUVs to be produced upon the 84 membranes were the focus of the first system- hydration of an agarose film has opened up new possibilities, as this protocol can be integrated 15 into a microarray format to form hundreds of liposome types in parallel39 and can be further atic protein–lipid interaction survey (see developed to create proteoliposome arrays42 (that is, liposomes into which membrane proteins have below). However, despite these pioneering been inserted). PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; efforts, many signalling lipids still have an PG, phosphatidyl glycerol; PI, phosphatidylinositol; PS, phosphatidylserine. elusive mode of action, for example, only a a Core structure Species handful of direct cellular effectors are known Glycerophospholipids PA: R = H PE: R= for sphingolipids and phosphatidylserine. CH NH2 O 3 In addition, the effects of mutations that O PC: R = N+ HO H P CH3 PG: R= OH have been observed in human biopsies, or of H C n O O O R CH 3 HO 3 post-translational modifications (for example H3C H PI: R = COOH O HO OH H n HO OH PS: R = phosphorylation or Lys acetylation), on the O OH NH2 lipid-binding affinity and specificity of pro- Sphingolipids H OH Ceramide H OH teins remain largely elusive. Another largely Sphingosine unexplored question is the role of ‘coinci- H27C13 OH OH H27C13 dence sensing’ (that is, the simultaneous H N H 2 H3C NH H n detection of several membrane properties, for O example, the simultaneous sensing of mem- H H Sterol H C CH Cholesterol H C CH brane curvature and the presence of phos- Cholestane 3 3 3 3 CH3 H CH3 H phatidylserine), or ‘cooperative mechanisms’ CH3 CH CH3 H CH3 H 3 (that is, when the binding affinity of a protein for a lipid X is changed upon the binding H H H H HO of a lipid Y), in the selective recruitment of peripheral proteins to specific membranes16,17. b Method to Size of Pros of method Cons of method Throughput Refs Soluble proteins. The third — and mostly produce liposome (per day) liposome overlooked — type of lipid–protein interaction involves soluble proteins that bind lipids Extrusion <500 nm Curvature Short storage time <10 82 control (<1 day) outside cellular membranes. These proteins bind lipids in a hydrophobic pocket and, by Electroformation >1 μm Well-established Formation in low <10 83 carrying their hydrophobic cargo through salt the aqueous phase of the cell, act as trans- Hydration on >1 μm Easy to set up, Polydispersity >1,000 39,84 porters, lipid chaperones (for example, pre- agarose-coated generic for any slide lipid mixture, senting lipids to metabolic enzymes) and/or‌ 18–20 physiological signalling factors . Two paramount examples are lipid-transfer proteins that transport Microﬂuidic >1 μm Size control, Diﬃcult to set up <10 85 jetting encapsulation various lipids between different organelles by a non-vesicular transfer mechanism2,

Nature Reviews | Molecular Cell Biology 2 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/molcellbio

and nuclear receptors21. This type of inter- have been designed to prepare mimetic model organisms such as S. cerevisiae16 or action has been the focus of surveys that membranes on sensor chips28. SPR studies Dictyostelium discoideum35 (FIG. 2). However, were limited to two protein families: the can provide both quantitative and qualita- a major drawback to these approaches is that kinases22 and the lipid-transfer proteins23 of tive data on molecular interactions — for the lipids are tested under highly artificial Saccharomyces cerevisiae. example, information on lipid specificity or conditions, for example, outside a membrane These different types of protein–lipid on the effects of mutations on binding affin- bilayer and at high, non-physiological con- interaction are not mutually exclusive. Often, ity — and this approach has been extensively centrations. Thus, they are prone to artefacts, the same protein can simultaneously sense used to study, for example, lipid-binding and interactions need to be verified using and bind to different membranes or lipids domains or proteins involved in signalling15, orthogonal assays26. — through distinct mechanisms — and this pore-forming peptides, coagulation factors creates molecular complexity and diversity. and enzymes28. Protein-array assays. In a protein-array For example, some lipid-transfer proteins An interesting and impressive newcomer assay (FIG. 1), proteins are immobilized on a also contain a PH domain (these include to the lipid field is native protein mass spec- solid support, and the binding of fluorescent oxysterol-binding protein (OSBP), OSBP- trometry. This method was first used to study liposomes that comprise a carrier lipid (usu- related protein 1 (ORP1L), ORP3, ORP4L, the dynamics and composition of soluble ally phosphatidylcholine) and a signalling ORP5, ORP6, ORP7, ORP8, ORP9L, ORP10 protein complexes29 but has been adapted for lipid is measured. The advantage of this and ORP11) or a transmembrane domain the study of integral membrane proteins that approach is that liposomes form a mem- (these include Niemann-Pick C1 protein are in complex with their lipid ligands25,30. brane bilayer and therefore closely mimic the (NPC1), ORP3, ORP5 and ORP8). Evidently, The challenge in adapting this method was to in vivo situation. So far, this assay has been then, proteins and lipids engage in pleio- preserve the native protein–membrane com- limited to two screens in S. cerevisiae (FIG. 2), tropic and multifaceted relationships that are plexes as they entered the gas phase. For this an organism for which a protein microarray based on various biophysical principles and purpose, innovative protocols — based on containing the majority of its proteome (80%; are traditionally studied by distinct scientific the use of detergent micelles31, bicelles or 5,800 proteins) is available37,38. The use of communities. The diversity of protein–lipid nanodiscs32 — were developed. This has this approach is also limited by the fact that interactions implies a need for appropri- enabled the identification of lipids that spe- the protocols designed to produce, handle ate, specialized assays that are optimized to cifically bind to large complexes of integral and store liposomes are labour-intensive and capture specific types of interactions (FIG. 1). membrane proteins, including ATP-binding difficult to scale up (BOX 1), thus precluding cassette (ABC) transporters33, mechano- the systematic analyses of different types Classic methods sensitive ion channels31 and two rotary of membrane. Several biochemical, biophysical and struc- ATPases/‌synthases34. These studies have tural methods are traditionally used to revealed complex stoichiometry, the identity Liposome microarray-based assay (LiMA). study biomolecular interactions24, and most of the bound lipid and the role of lipids in LiMA (FIG. 1) addresses many of the draw- of these have been adapted to the study of modulating protein folding and stability31. backs mentioned above39. This assay protein–lipid interactions, for example, measures the recruitment of proteins to X‑ray crystallography, nuclear magnetic Novel in vitro methods membranes in a quantitative, multiplexed resonance, isothermal titration calorimetry, Biochemical and biophysical assays based on and high-throughput manner. It integrates surface plasmon resonance (SPR) and micro- the use of artificial, surrogate membranes are biochemical principles — that is, the forma- scale thermophoresis (MST) (BOX 2). These popular (see above), as they allow lipids to tion of liposomes with diameters of >1 μm assays are generally low-throughput, but be studied in vitro in controlled and defined (BOX 1) on a thin agarose layer — with they have contributed important atomic or environments (that is, independently of the quantitative fluorescence microscopy-based mechanistic models for the different types of other attributes or constituents of cellular imaging and microfluidics. The procedure protein–lipid interactions, and they remain membranes). Some of these assays have been used to produce and array the liposomes is the methods of choice for studying integral adapted and applied to systematic analyses. generic, and hundreds of liposomes of vary- membrane proteins (reviewed in REF. 25). It is ing lipid composition (including the main beyond the scope of this article to discuss all Lipid-overlay and lipid pull-down assays. lipid classes) can be simultaneously produced of the classic assays in depth, as they have Lipid-overlay and lipid pull-down assays on a small (~1 cm2) microfluidic chip. The been reviewed elsewhere26,27; rather, we will (FIG. 1) are based on the same principle and recruitment of fluorescently labelled proteins focus on the most prominent and widely involve the immobilization of individual is measured by high-content, quantitative used method, SPR. lipids (or lipid head groups) on a solid sup- fluorescence microscopy40. Importantly, SPR allows the direct and rapid measure- port (for example, nitrocellulose membranes the assay is quantitative and can measure ment of lipid–protein association and dis- for lipid-overlay and magnetic beads for discrete changes in binding affinities that sociation rates without the need for protein lipid pull-down). The proteins that bind to are produced by protein mutations found or lipid labelling28. Interactions are studied the immobilized lipids can be characterized in human biopsies or by the presence of on the surface of ‘sensor chips’, which are using either immunodetection (for exam- additional lipids in the membrane (which glass slides that are coated with a very thin ple, with fluorescent or chemiluminescent trigger coincidence-sensing mechanisms)39,41. layer of gold, to which an artificial membrane antibodies) or mass spectrometry35,36. The LiMA is scalable to the proteome or lipidome or liposome (BOX 1) is attached. The recruit- lipid-overlay assay is available commercially levels and takes advantage of the available ment of an interacting protein to the artificial and is widely used to measure protein–lipid genome-wide collections of cell lines membrane or liposome changes the refrac- interactions in vitro (see Supplementary expressing GFP fusions. Recently, LiMA was tive index at the surface of the chip; this information S1 (table)). Both methods have applied to ~10,000 experiments designed to information is recorded. Many protocols been readily adapted to large surveys in quantify the role of phosphoinositides in the

NATURE REVIEWS | MOLECULAR CELL BIOLOGY ADVANCE ONLINE PUBLICATION | 3

recruitment of PH domains to membranes; to the periphery of membranes. Going Novel in vivo methods cooperativity between lipids was found to forward, protocols for the production of Recent technical advances have enabled the be a key mechanism for such recruitment41. proteoliposomes and for the study of integral study of protein–lipid interactions in vivo; the The assay is also adaptable to other readouts, membrane proteins in the context of different key functional assays are discussed below. such as those from mass spectrometry or lipids will need to be integrated42. advanced optical methods, including total Importantly, these in vitro assays produce In vivo functional assays and live-cell internal reflection fluorescence microscopy invaluable data sets on the ability of proteins imaging. Many in vivo functional assays have or two-photon excitation microscopy; the and/or different lipids or membrane environ- been developed to characterize the roles latter of these optical techniques enables ments to interact. The physiological rele- of intracellular membranes and individual equilibrium dissociation constants to be vance and functions of these interactions will lipids in tuning the behaviour and function determined. Currently, LiMA is best suited to need to be investigated through orthogonal of effector proteins. These cellular assays are studying the recruitment of soluble proteins and more physiological assays. based on similar general principles, and they

5 Readout Protein iden tiﬁc atio n (M S) Complex Lipid-binding assembly protein in vitro Bead Lipid R 4 Principle ad io ) a S c M t ( iv n i o ty ti a Cell o c r i iﬁ lysate m t Anti-tag n m e Tag u d antibody n i o in Tag d e Tag e t 3 Method Lip Tag t o id e r pull c Crosslinking -do t P w Tag i n o 2 Type n Chemically g in engineered er L e ip Spotted lipid in 1 Proteins id lipids g a - n s s o e id s v l p a e a i y r l la ic f l o a e y m r n ) g e e ra C h t b Plasma membrane L n C I m I T n Complex e r v Liposome m Lipid-binding i F o t assembly l r u o S proteins o in vivo s Nucleus r M c

( i e P

e r a n m

l c

o o

b r

Tag t i

a e

l y

Tag i

i c

Bead Lipid-binding s

ﬁ

g i

P proteins Spotted

e n

A n

n proteins a

e a l

Anti-tag r d

i b antibody

d m

i e p

i C I m

L n l a v a g iv r e o he d p A l ri ip Pe iM GFP Caged id L lipid s GFP Arrayed liposomes Ra F s ll l UV res -ce u a cue Live o ssa ing r y ag e im GFP l s a c n e Ras GFP ig n s c Lipid-binding t e GTP Fluorescent n m protein e Ras protein c ic s r GDP re o fusions o sc o ↓ Growth Expression lu p F y Ras-signalling- of Ras fusion deﬁcient cell Stimulation or Ras perturbation Ras GTP GDP Ras GTP y op Co ↑ Growth sc lon icro y gr ce m owth scen Fluore

Nature Reviews | Molecular Cell Biology

4 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/molcellbio

involve the systematic, targeted perturbation Box 2 | Classical methods for studying protein–lipid interactions of a membrane component (for example, the inhibition or mutation of enzymes in • Flotation assay: a method in which liposomes and proteins are mixed at the bottom of a sucrose lipid metabolic pathways) followed by the gradient and ultracentrifuged. If proteins and lipids interact, the complex floats in the upper 27 measurement of the effects of these perturba- fractions of the centrifugation tube . tions on specific phenotypic readouts (FIG. 1). • ITC (isothermal titration calorimetry): a biophysical method to obtain thermodynamic Among these functional assays, genetic parameters of the interaction between proteins and liposomes, which allows their molecular 86 interaction screening is probably the oldest affinities to be calculated . and most widely used technique for identify- • SPR (surface plasmon resonance): in this approach, a lipid membrane is formed or liposomes are ing functional relationships between cellular bound on top of a plasmon resonance sensor chip, and SPR is used to detect the binding and dissociation of proteins26,28. pathways that cannot be linked a priori43. The assay measures how two genes that are • MST (microscale thermophoresis): a method using a temperature field to perform biomolecular interaction studies, which can be applied to protein–lipid interaction studies87. deleted simultaneously perturb a phenotype, for example, cell growth or, spurred on by • Native protein mass spectrometry: the analysis of intact protein complexes by mass spectrometry. the recent advances in high-throughput live- cell imaging, cell morphology44. Interaction screens were used in S. cerevisiae to build recruiting PH domains to the membrane can necessarily measure direct physical associa- genetic interaction maps that are focused on be assessed by measuring the effect of tar- tions, because they also capture functional lipid biology (termed lipid E‑MAPs) and that geted perturbations of lipid metabolism (for links and indirect interactions. In addition,

revealed new components of the eisosome example, PH domains that bind PtdIns(4,5)‌P2 the effects of metabolic perturbations on the complex (a protein complex that marks the will not be recruited to yeast membranes if lipidome of specific subcellular membranes

site of endocytosis at the plasma membrane the enzyme that produces PtdIns(4,5)P‌ 2, the remain difficult to predict because of, among in some eukaryotes), as well as novel molecu- PtdIns(4)P 5‑kinase, is mutated). Similarly, other factors, the so‑called ‘ripple effect’ lar mechanisms linking protein degradation live-cell imaging (FIG. 1) has been used to (REF. 50) (in which perturbing the level of one and fatty acid desaturation45,46. observe the translocation of proteins or lipid induces indirect changes in the levels of The Ras rescue assay (FIG. 1) is another protein domains fused to a fluorescent many other lipids). In short, the approaches in vivo functional assay that was developed tracer (usually GFP) to cellular membranes discussed in this subsection are not usu- in yeast. It is based on rescuing the growth of following cell stimulation or metabolic per- ally designed to capture the mechanisms a thermosensitive mutant of the yeast Cdc25 turbations15,16,49. Generally, these cell-based of biomolecular recognition; this requires (which is a membrane-bound guanine nucle- assays are generic and compatible with the integration of orthogonal methods, for otide exchange factor for Ras). Membrane- large-scale and systematic surveys. They are example, biophysical, structural or large-scale binding domains or proteins fused to a very powerful tools for inferring functional biochemistry (see above). constitutively active Ha‑Ras mutant can tar- relationships between proteins or measur- get active Ras to the membrane and thus res- ing membrane binding directly inside cells. Affinity-purification lipidomics. A series cue growth47,48. The role of specific lipids in Their main drawback is that they do not of new physiological assays have recently been developed that alleviate some of the drawbacks of the in vivo functional assays ◀ Figure 1 | System-wide methods for capturing protein–lipid interactions. Protein–lipid inter discussed above. An example is affinity- actions can be divided into three categories according to the type of protein involved in the interac- purification lipidomics, which relies on the tion: peripheral membrane proteins, soluble proteins and integral membrane proteins (curve 1). homologous expression of affinity-tagged Each protein–lipid interaction can be captured by specialized in vitro assays (using artificial mem- protein fusions to systematically purify branes) or in vivo assays (using biological membranes) (curve 2). Each method (curve 3) and its associ- and characterize protein–lipid complexes ated principle (curve 4) is schematically shown. Finally, each method is associated with specific readout that have been assembled in vivo22,23 (FIG. 1). techniques (curve 5). To capture direct protein–lipid interactions of peripheral membrane proteins, The main advantage of this method is that four in vitro-based methods are available. These techniques are based on the immobilization of either endogenously expressed and native proteins the lipid or the liposome (lipid pull-down, lipid-overlay assay, liposome microarray-based assay (LiMA)) or of the protein (protein array) on a solid surface, followed by probing with the other ligand (protein are retrieved from cells or tissues, closely or lipid or liposome) in solution. Direct protein–lipid interactions of peripheral membrane proteins can mirroring physiological conditions. The also be captured using in vivo methods. Live-cell imaging involves the observation of the recruitment protein–lipid complexes are kept in their of GFP-fused proteins to the cellular plasma membrane. Another in vivo method, caged lipids, which native states throughout the purification is based on a chemical biology approach, relies on the enrichment of a particular lipid in cellular mem- process, remain functional and are amenable branes by ‘uncaging’ the chemically modified lipid using ultraviolet (UV) light; the recruitment of the to in vitro activity-based assays or structural protein to the membrane is then followed with fluorescence microscopy. Finally, among in vivo meth- analyses23. The co‑eluting proteins and lipids ods devoted to peripheral membrane proteins, the Ras rescue assay is based on a mutant yeast strain are analysed by mass spectrometry and lipid that is deficient in the Ras signalling pathway and has a growth deficiency as a result. Growth is res- omics methods. This protocol is not limited cued only when a membrane-targeted protein fused to constitutively active Ras protein is expressed. to one cell type and can, in principle, be The second interaction mode between lipids and proteins involves soluble proteins, that is, those present in the cytoplasm that can tightly bind lipids. Such interactions are analysed by protein–lipid used to quantify changes in complex com- co‑purification; this requires the purification of in vivo assembled protein–lipid complexes from a position between cell lines or following cell whole-cell lysate and analysis of the associated lipids with mass spectrometry (MS) or thin-layer chro- stimulations; it also benefits from being able matography (TLC). Finally, to capture direct interactions between lipids and an integral membrane to use the many available proteome-wide protein, a crosslinking strategy is used that creates a covalent link between the protein and the lipids, collections of cell lines that express a tagged forming complexes that can be purified and analysed by mass spectrometry. AP, affinity purification. fusion protein51,52. The assay is also readily

NATURE REVIEWS | MOLECULAR CELL BIOLOGY ADVANCE ONLINE PUBLICATION | 5

Interaction screen testing the ability of cell lysate to interact 13 lipid-transfer proteins tested 35 with PtdIns(3,4,5)P3 using lipid pull-down (D. discoideum) for their ability to interact with lipids using affinity-purification Arachidonoyl 374 genes implicated in plasma membrane lipidomics (yeast)23 lipid-interacting biology by genetic interaction assays (yeast)45 proteins identified Cholesterol-interacting proteins using chemically Interaction screen using 33 PH Interaction screen using 172 proteins and identified using bifunctional engineered lipids domains and 10 lipids in lipid-overlay, 56 lipids in a lipid-overlay assay (yeast)16 lipid proteomics (human)55 (human)60 Ras-rescue and SPR assays (yeast)15 Interaction screen testing the 103 kinase–lipid Interaction screen using 91 PH Interaction screen using 5,800 ability of 132 PH domains to interactions tested using domains and 122 different types proteins and 6 types of liposome interact with PtdIns(3,4,5)P3 affinity-purification of liposomes in LiMA (yeast and in a protein array assay (yeast)37 using live-cell imaging (human)49 lipidomics (yeast)22 C. thermophilum)41

1998 2001 2002 2004 2008 2009 2010 2012 2013 2014 2015

Ras rescue assay47 Lipid-overlay Native protein mass spectrometry PtdIns(3)P caged lipid58 Liposome microarray-based Screen assay36 of integral membrane proteins with assay (LiMA)39 Method their lipid interaction partners30 Figure 2 | Timeline of key events in protein–lipid interaction research. the timeline. The species in which the screen was carried out is shown in The development of key methods and their application to systematic parentheses. C. thermophilum, ChaetomiumNature Reviews thermophilum | Molecular; D. discoideum Cell Biology, protein–lipid interaction screens are listed here in chronological order. Dictyostelium discoideum; LiMA, liposome microarray-based assay; The systematic screens that have been conducted so far are shown above PH, pleckstrin homology; PtdIns(3)P, phosphatidylinositol‑3‑phosphate;

the timeline. The development of key techniques that are suitable for the PtdIns(3,4,5)‌P3, phosphatidylinositol‑3,4,5‑trisphosphate; SPR, surface large-scale detection of protein–lipid interactions are indicated below plasmon resonance.

scalable and has been applied to systematic that allow the selective coupling of two allow lipids and the composition of cellular surveys of yeast kinases22 and lipid-transfer functional groups in biological samples54) membranes to be precisely manipulated proteins, which revealed new mechanisms to a reporter molecule, such as a fluorescent in both time and space and — coupled for the non-vesicular transport of phos- dye for visualization or biotin for affinity with some of the in vivo assays described phatidylserine between organelles23 (FIG. 2). purification54–57. In combination with high- above, such as live-cell imaging — hold Another important outcome of these pilot resolution mass spectrometry, the use of great promise. studies is that they demonstrate the feasibil- bifunctional lipids can provide global maps ity of affinity-purification lipidomics and the of protein–lipid interactions directly in liv- Towards multi-omics integration need for broader, proteome-wide efforts for ing cells or organisms. The list of available Pioneering protein–lipid interaction studying protein–lipid interactions in higher bifunctional lipids is continuously increas- screens were based on early technologies eukaryotes. However, this approach remains ing and spans all of the main lipid classes, and focused on the study of soluble proteins limited to the characterization of relatively and the approach is becoming generic54. or of specific lipid-binding domains15,16. stable and soluble protein–lipid complexes. The main drawback of this method is that Since then, more generic and physiologi- the modified lipids are generally metabo- cal methods have been developed. These Chemical engineering of lipids. The chemi- lized in living cells, which generates various hold great potential for the next genera- cal engineering of lipids (FIG. 1) contributes bifunctional products that can, in principle, tion of screens, which will embrace entire invaluable tools for measuring, visualizing also react with proteins. As for the genetic proteomes and lipidomes in a range of and studying lipids inside living cells, and methods described above, the characteriza- cell types and in various physiological these are broadly applicable to all types of tion of the precise lipid or lipids involved or pathological states (FIG. 3). We expect protein–lipid interactions. For example, requires the integration of orthogonal that novel screens will be used to discover photoactivatable lipids have been devel- methods such as biophysical, structural or novel modes of protein–lipid interaction oped that contain a chemical group (such large‑scale biochemistry approaches. that involve multiple lipids, lipids and as a diazirine ring) that enables them to Other interesting advances include the proteins and, potentially, lipids and glyco form crosslinks (FIG. 1) with proteins that recent development of caged lipids (FIG. 1), lipids (FIG. 3a). Although a few examples of they interact with in response to ultra- which bear a photoactivatable protective lipid-binding domains that can recognize violet light11,53 (reviewed in REF. 54). These group that is intended to keep the lipid several lipids simultaneously have been approaches, however, are limited in their (whether it is a signalling or structural lipid) described12,16,17, a full repertoire of such scope, as they do not allow for affinity inactive until a light pulse removes the ‘cage’ interactions is not known. Also, other enrichment of the crosslinked protein–lipid (REFS 58–60). Also, alkyne-tagged lipids aspects of membrane biophysics, such as complexes. This issue was recently addressed combined with Raman-scattering imaging membrane electrostatics and bending, and with the development of bifunctional lipids (which is a method used to measure the how they are influenced by protein–lipid that, in addition to the photoreactive moiety, specific vibrational signatures of chemi- interactions, must be integrated into the contain an alkyne group. This alkyne group cal bonds) offer a promising approach for design of future screens63 (FIG. 3b). enables the lipid to be coupled through click directly imaging the metabolic route of lipids As our understanding of the molecular chemistry (that is, bio-orthogonal reactions in live cells61,62. These in vivo techniques and functional circuitry between proteins

6 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/molcellbio

and lipids unfolds, we must extend and com- was demonstrated, using membrane-mimics Translational medicine challenges plete these networks through the integration (BOX 1), that cholesterol-dependent raft-like Understanding the molecular and functional of orthogonal ‘omics’ data sets and develop lipid domains have a pivotal role in the circuitry linking proteins and lipids in comprehensive, multilevel models that fusion of the human immunodeficiency human cells and how they are affected in dis- depict the roles of lipids in cell biology and virus with membranes; this illustrates the ease states will be an invaluable tool for med- physiology50. There is, for example, a press- power of in vitro approaches in reconstitut- icine69. Mutations in lipid-binding domains ing need for a more systematic functional ing complex protein–lipid interactions68. In that cause discrete changes in their affinity interpretation of the multitude of protein– this regard, LiMA holds great promise, and and specificity for lipids frequently affect cell lipid interactions that have been detected, we envision that in the future this micro- fate and the physiology of whole organisms. which can be addressed through the in silico fluidic platform will support new assays, Prominent examples include point mutations integration of complementary data sets including enzymatic assays (for example, in the PtdIns(3,4,5)P‌ 3- and PtdIns(3,4)P‌ 2- derived from both in vivo and in vitro for the detection of kinase, phosphatase and specific PH domain of AKT1, which result biochemical assays. phospholipase activity), as well as assays for in diseases such as colorectal cancers70, Another important advance is the abil- the activity of transport proteins, and fluo- Proteus syndrome (which is characterized ity to experimentally measure the func- rescence resonance energy transfer (FRET), by overgrowth of skin and skeleton)71 or tion of lipid–protein interactions through atomic force, super-resolution and electron metabolic disorders (such as hypoglycae- the integration of different assays, so that microscopy (FIG. 3c). Finally, the develop- mia)72; mutations in the histone fold of Son binding events can be directly linked to, ment of new chemical tools54 and probes of sevenless homologue 1 (SOS1; a guanine for example, their subcellular location, the to measure protein and lipid binding and nucleotide exchange factor for RAS), which enzymatic activity64,65 or oligomerization activity, or specific membrane attributes cause Noonan syndrome73; and mutations in state of the protein, or whether the local such as thickness, inside living cells will be the PX domain of p40phox (also known as membrane is organized into nanodomains instrumental in the move towards NCF4; a subunit of NADPH oxidase), which or microdomains66,67 (FIG. 3b). Recently, it multi-omics integration. underlie chronic granulomatous disease74.

b Measuring the eﬀect of protein–lipid binding on membranes Membrane proteins Membrane curvature changes

c Studying the structure of protein–lipid membrane interactions Protein structure assemblies at membrane interface using AFM

AFM cantilever Lipid clustering induced d Perturbing protein–lipid interactions Targeting LBDs Drug Lipid micro- or nanodomains

a Systematic screening of protein–lipid interactions Targeting integral membrane protein–lipid interactions Novel modes of protein–lipid interaction Multiple-lipid Lipid and Lipid and recognition protein glycolipid by one LBD co-recognition co-recognition Lipid-speciﬁc membrane destabilization drugs

Liposome microarray Figure 3 | Future scientific and technological challenges for lipo- c | Whereas this technology hasNature been usedReviews so far| Molecular in combination Cell Biology with some microarray-based assays (LiMA). a | In the future, LiMA will be fluorescence microscopy39, further development will allow protein struc- used to investigate novel modes of protein–membrane interactions that ture assemblies to be investigated at the membrane interface using tech- imply multiple partners, for example, the coincident sensing of multiple niques such as atomic force microscopy (AFM). d | LiMA is also an lipids (left) and the co‑recognition of lipid and proteins (middle) or lipids attractive method for developing novel pharmaceutical compounds that and glycolipids (right). b | Another field of investigation is the measure- perturb protein–lipid interactions. Such compounds might target the ment of the effect of protein–lipid interactions on the biophysical proper- lipid binding domains (LBDs) of peripheral membrane proteins (top), inte- ties of membranes, notably the ability of proteins to bend membranes gral membrane proteins (middle) or compounds that kill pathogens by (top) or to generate microdomains or nanodomains of lipids (bottom). destabilizing membranes (bottom).

NATURE REVIEWS | MOLECULAR CELL BIOLOGY ADVANCE ONLINE PUBLICATION | 7

Globally, however, little is known about the 1. van Meer, G., Voelker, D. R. & Feigenson, G. W. 30. Barrera, N. P. et al. Mass spectrometry of membrane Membrane lipids: where they are and how they transporters reveals subunit stoichiometry and functional consequences of single-nucleotide behave. Nat. Rev. Mol. Cell. Biol. 9, 112–124 (2008). interactions. Nat. Methods 6, 585–587 (2009). polymorphisms on lipid–protein interaction 2. Holthuis, J. C. & Menon, A. K. Lipid landscapes and 31. Laganowsky, A. et al. Membrane proteins bind lipids pipelines in membrane homeostasis. Nature 510, selectively to modulate their structure and function. networks, and the recently developed high- 48–57 (2014). Nature 510, 172–175 (2014). throughput in vitro binding assays are likely 3. Fahy, E. et al. Update of the LIPID MAPS 32. Hopper, J. T. et al. Detergent-free mass spectrometry comprehensive classification system for lipids. J. Lipid of membrane protein complexes. Nat. Methods 10, to contribute to filling this gap. Res. 50, S9–S14 (2009). 1206–1208 (2013). Lipids are also taking centre stage in the 4. Overington, J. P., Al‑Lazikani, B. & Hopkins, A. L. How 33. Bechara, C. et al. A subset of annular lipids is linked to many drug targets are there? Nat. Rev. Drug Discov. the flippase activity of an ABC transporter. Nat. Chem. field of infectious diseases, and pharmaceuti- 5, 993–996 (2006). 7, 255–262 (2015). cal strategies targeting fungal or bacterial lipid 5. Wymann, M. P. & Schneiter, R. Lipid signalling in 34. Zhou, M. et al. Mass spectrometry of intact V‑type disease. Nat. Rev. Mol. Cell. Biol. 9, 162–176 (2008). ATPases reveals bound lipids and the effects of metabolism or membranes are being pro- 6. Babu, M. et al. Interaction landscape of membrane- nucleotide binding. Science 334, 380–385 (2011). posed to counteract antibiotic resistance75,76. protein complexes in Saccharomyces cerevisiae. 35. Zhang, P., Wang, Y., Sesaki, H. & Iijima, M. Proteomic Nature 489, 585–589 (2012). identification of phosphatidylinositol (3,4,5) In addition, intracellular pathogens such as 7. Capelluto, D. G. Lipid-mediated Protein Signaling triphosphate-binding proteins in Dictyostelium bacteria77, viruses78 and parasites79 can hijack (Springer, 2013). discoideum. Proc. Natl Acad. Sci. USA 107, 8. Sharpe, H. J., Stevens, T. J. & Munro, S. A 11829–11834 (2010). the host cell machineries that are responsi- comprehensive comparison of transmembrane 36. Dowler, S., Kular, G. & Alessi, D. R. Protein lipid ble for lipid homeostasis, and targeting the domains reveals organelle-specific properties. overlay assay. Sci. STKE 2002, pl6 (2002). Cell 142, 158–169 (2010). 37. Zhu, H. et al. Global analysis of protein activities mechanisms involved could lead to new 9. Nury, H. et al. Structural basis for lipid-mediated using proteome chips. Science 293, 2101–2105 therapeutic strategies. interactions between mitochondrial ADP/ATP carrier (2001). monomers. FEBS Lett. 579, 6031–6036 (2005). 38. Lu, K. Y. et al. Profiling lipid–protein interactions using Lipid metabolism and lipid-binding 10. Whorton, M. R. & MacKinnon, R. Crystal structure of nonquenched fluorescent liposomal nanovesicles and proteins have thus emerged as promising the mammalian GIRK2 K+ channel and gating proteome microarrays. Mol. Cell. Proteomics 11, regulation by G proteins, PIP2, and sodium. Cell 147, 1177–1190 (2012). new targets for pharmacological interven- 199–208 (2011). 39. Saliba, A. E. et al. A quantitative liposome microarray tion in various human diseases80,81. In addi- 11. Contreras, F. X. et al. Molecular recognition of a single to systematically characterize protein-lipid sphingolipid species by a protein’s transmembrane interactions. Nat. Methods 11, 47–50 (2014). tion, many of the new assays described here domain. Nature 481, 525–529 (2012). 40. Pepperkok, R. & Ellenberg, J. High-throughput should enable the disruption of protein–lipid 12. Lemmon, M. A. Membrane recognition by fluorescence microscopy for systems biology. Nat. Rev. phospholipid-binding domains. Nat. Rev. Mol. Cell. Mol. Cell. Biol. 7, 690–696 (2006). interactions by small molecules to be studied Biol. 9, 99–111 (2008). 41. Vonkova, I. et al. Lipid cooperativity as a general using, for example, LiMA (FIG. 3d). 13. Antonny, B. Mechanisms of membrane curvature membrane-recruitment principle for PH domains. sensing. Annu. Rev. Biochem. 80, 101–123 (2011). Cell Rep. 12, 1519–1530 (2015). 14. Scott, J. D. & Pawson, T. Cell signaling in space and 42. Hansen, J. S., Thompson, J. R., Helix-Nielsen, C. & Conclusion time: where proteins come together and when they’re Malmstadt, N. Lipid directed intrinsic membrane apart. Science 326, 1220–1224 (2009). protein segregation. J. Am. Chem. Soc. 135, In this Innovation article, we have summa 15. Yu, J. W. et al. Genome-wide analysis of membrane 17294–17297 (2013). rized and illustrated the new approaches for targeting by S. cerevisiae pleckstrin homology 43. Costanzo, M. et al. The genetic landscape of a cell. domains. Mol. Cell 13, 677–688 (2004). Science 327, 425–431 (2010). investigating protein–lipid interactions on a 16. Gallego, O. et al. A systematic screen for protein–lipid 44. Laufer, C., Fischer, B., Huber, W. & Boutros, M. large scale. The vast majority of lipids have not interactions in Saccharomyces cerevisiae. Mol. Syst. Measuring genetic interactions in human cells by Biol. 6, 430 (2010). RNAi and imaging. Nat. Protoc. 9, 2341–2353 been assigned a functional biological role, thus 17. Moravcevic, K., Oxley, C. L. & Lemmon, M. A. (2014). these methods might allow novel biological Conditional peripheral membrane proteins: facing up 45. Aguilar, P. S. et al. A plasma-membrane E‑MAP to limited specificity. Structure 20, 15–27 (2012). reveals links of the eisosome with sphingolipid insights into lipid biology. In vitro approaches 18. Lev, S. Non-vesicular lipid transport by lipid-transfer metabolism and endosomal trafficking. Nat. Struct. are constantly progressing towards allowing proteins and beyond. Nat. Rev. Mol. Cell. Biol. 11, Mol. Biol. 17, 901–908 (2010). 739–750 (2010). 46. Surma, M. A. et al. A lipid E‑MAP identifies Ubx2 as a the reconstitution of cellular membrane com- 19. Forman, B. M., Chen, J. & Evans, R. M. Hypolipidemic critical regulator of lipid saturation and lipid bilayer plexity, and they will allow the exact mecha- drugs, polyunsaturated fatty acids, and eicosanoids stress. Mol. Cell 51, 519–530 (2013). are ligands for peroxisome proliferator-activated 47. Isakoff, S. J. et al. Identification and analysis of PH nisms by which proteins target membranes to receptors α and δ. Proc. Natl Acad. Sci. USA 94, domain-containing targets of phosphatidylinositol be deciphered. In addition, the development 4312–4317 (1997). 3‑kinase using a novel in vivo assay in yeast. 20. Grabon, A., Khan, D. & Bankaitis, V. A. EMBO J. 17, 5374–5387 (1998). of affinity- and chemical biology-based meth- Phosphatidylinositol transfer proteins and instructive 48. Moravcevic, K. et al. Kinase associated‑1 domains ods will shed light on the in vivo roles of lipids. regulation of lipid kinase biology. Biochim. Biophys. drive MARK/PAR1 kinases to membrane targets by Acta 1851, 724–735 (2015). binding acidic phospholipids. Cell 143, 966–977 More generally, these methods illustrate the 21. Musille, P. M., Kohn, J. A. & Ortlund, E. A. (2010). value of multidisciplinary approaches that Phospholipid-driven gene regulation. FEBS Lett. 587, 49. Park, W. S. et al. Comprehensive identification of 1238–1246 (2013). PIP3‑regulated PH domains from C. elegans to integrate biology, chemistry, engineering and 22. Li, X., Gianoulis, T. A., Yip, K. Y., Gerstein, M. & H. sapiens by model prediction and live imaging. the physical sciences. Snyder, M. Extensive in vivo metabolite-protein Mol. Cell 30, 381–392 (2008). interactions revealed by large-scale systematic 50. Cowart, L. A. et al. Revealing a signaling role of phytosphingosine‑1‑phosphate in yeast. Mol. Syst. Antoine-Emmanuel Saliba is at the Institute for analyses. Cell 143, 639–650 (2010). 23. Maeda, K. et al. Interactome map uncovers Biol. 6, 349 (2010). Molecular Infection Biology and Core Unit Systems phosphatidylserine transport by oxysterol-binding 51. Gavin, A. C. et al. Proteome survey reveals modularity Medicine, University of Würzburg, Josef-Schneider- proteins. Nature 501, 257–261 (2013). of the yeast cell machinery. Nature 440, 631–636 Strasse 2, D-97080 Würzburg, Germany. 24. Charbonnier, S., Gallego, O. & Gavin, A. C. The social (2006). network of a cell: recent advances in interactome 52. Huh, W. K. et al. Global analysis of protein Ivana Vonkova is at the European Molecular Biology mapping. Biotechnol. Annu. Rev. 14, 1–28 (2008). localization in budding yeast. Nature 425, 686–691 Laboratory (EMBL), Structural and Computational 25. Barrera, N. P., Zhou, M. & Robinson, C. V. The role of (2003). Biology Unit, Meyerhofstrasse 1, lipids in defining membrane protein interactions: 53. Thiele, C., Hannah, M. J., Fahrenholz, F. & Huttner, W. B. Cholesterol binds to synaptophysin and D-69117 Heidelberg, Germany. insights from mass spectrometry. Trends Cell Biol. 23, 1–8 (2013). is required for biogenesis of synaptic vesicles. Nat. Cell Anne-Claude Gavin is at the European Molecular 26. Narayan, K. & Lemmon, M. A. Determining selectivity Biol. 2, 42–49 (2000). of phosphoinositide-binding domains. Methods 39, 54. Haberkant, P. & Holthuis, J. C. Fat and fabulous: Biology Laboratory (EMBL), Structural and 122–133 (2006). bifunctional lipids in the spotlight. Biochim. Biophys. Computational Biology Unit and Molecular Medicine 27. Zhao, H. & Lappalainen, P. A simple guide to Acta 1841, 1022–1030 (2014). Partnership Unit, Meyerhofstrasse 1, biochemical approaches for analyzing protein-lipid 55. Hulce, J. J., Cognetta, A. B., Niphakis, M. J., Tully, S. E. D-69117 Heidelberg, Germany. interactions. Mol. Biol. Cell 23, 2823–2830 (2012). & Cravatt, B. F. Proteome-wide mapping of 28. Besenicar, M., Macek, P., Lakey, J. H. & Anderluh, G. cholesterol-interacting proteins in mammalian cells. Correspondence to A-C.G. Surface plasmon resonance in protein-membrane Nat. Methods 10, 259–264 (2013). e‑mail: [email protected] interactions. Chem. Phys. Lipids 141, 169–178 (2006). 56. Gubbens, J. et al. Photocrosslinking and click 29. Zhang, Z., Wu, S., Stenoien, D. L. & Pasa-Tolic, L. High- chemistry enable the specific detection of proteins doi:10.1038/nrm4080 throughput proteomics. Annu. Rev. Anal. Chem. 7, interacting with phospholipids at the membrane Published online 28 October 2015 427–454 (2014). interface. Chem. Biol. 16, 3–14 (2009).

8 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/molcellbio

57. Peng, T. & Hang, H. C. Bifunctional fatty acid chemical 70. Carpten, J. D. et al. A transforming mutation in the 83. Walde, P., Cosentino, K., Engel, H. & Stano, P. reporter for analyzing S‑palmitoylated membrane pleckstrin homology domain of AKT1 in cancer. Nature Giant vesicles: preparations and applications. protein–protein interactions in mammalian cells. 448, 439–444 (2007). ChemBioChem 11, 848–865 (2010). J. Am. Chem. Soc. 137, 556–559 (2015). 71. Lindhurst, M. J. et al. A mosaic activating mutation in 84. Horger, K. S., Estes, D. J., Capone, R. & Mayer, M. 58. Subramanian, D. et al. Activation of membrane- AKT1 associated with the Proteus syndrome. N. Engl. Films of agarose enable rapid formation of giant permeant caged PtdIns(3)P induces endosomal fusion J. Med. 365, 611–619 (2011). liposomes in solutions of physiologic ionic strength. in cells. Nat. Chem. Biol. 6, 324–326 (2010). 72. Arya, V. B. et al. Activating AKT2 mutation: J. Am. Chem. Soc. 131, 1810–1819 (2009). 59. Hoglinger, D., Nadler, A. & Schultz, C. Caged lipids as hypoinsulinemic hypoketotic hypoglycemia. J. Clin. 85. Stachowiak, J. C. et al. Unilamellar vesicle tools for investigating cellular signaling. Biochim. Endocrinol. Metab. 99, 391–394 (2014). formation and encapsulation by microfluidic Biophys. Acta 1841, 1085–1096 (2014). 73. Yadav, K. K. & Bar-Sagi, D. Allosteric gating of Son of jetting. Proc. Natl Acad. Sci. USA 105, 4697–4702 60. Niphakis, M. J. et al. A global map of lipid-binding sevenless activity by the histone domain. Proc. Natl (2008). proteins and their ligandability in cells. Cell 161, Acad. Sci. USA 107, 3436–3440 (2010). 86. Lemmon, M. A., Ferguson, K. M., O’Brien, R., 1668–1680 (2015). 74. Matute, J. D. et al. A new genetic subgroup of chronic Sigler, P. B. & Schlessinger, J. Specific and high-affinity 61. Wei, L. et al. Live-cell imaging of alkyne-tagged small granulomatous disease with autosomal recessive binding of inositol phosphates to an isolated pleckstrin biomolecules by stimulated Raman scattering. mutations in p40 phox and selective defects in homology domain. Proc. Natl Acad. Sci. USA 92, Nat. Methods 11, 410–412 (2014). neutrophil NADPH oxidase activity. Blood 114, 10472–10476 (1995). 62. Folick, A., Min, W. & Wang, M. C. Label-free imaging 3309–3315 (2009). 87. Jerabek-Willemsen, M., Wienken, C. J., Braun, D., of lipid dynamics using coherent anti-stokes Raman 75. Ling, L. L. et al. A new antibiotic kills pathogens Baaske, P. & Duhr, S. Molecular interaction studies scattering (CARS) and stimulated Raman scattering without detectable resistance. Nature 517, 455–459 using microscale thermophoresis. Assay Drug Dev. (SRS) microscopy. Curr. Opin. Genet. Dev. 21, (2015). Technol. 9, 342–353 (2011). 585–590 (2011). 76. Anderson, T. M. et al. Amphotericin forms an 63. McLaughlin, S. & Murray, D. Plasma membrane extramembranous and fungicidal sterol sponge. Acknowledgements phosphoinositide organization by protein Nat. Chem. Biol. 10, 400–406 (2014). The authors thank all members of A.-C.G.’s groups for con‑ electrostatics. Nature 438, 605–611 (2005). 77. Salomon, D. et al. Effectors of animal and plant tinuous discussions and support. This work was partially 64. Mouchlis, V. D., Bucher, D., McCammon, J. A. & pathogens use a common domain to bind host funded by the German Federal Ministry of Education and Dennis, E. A. Membranes serve as allosteric activators phosphoinositides. Nat. Commun. 4, 2973 Research (BMBF; 01GS0865) in the framework of the of phospholipase A2, enabling it to extract, bind, and (2013). IG‑Cellular System Genomics project and the CellNetworks hydrolyze phospholipid substrates. Proc. Natl Acad. 78. Hsu, N. Y. et al. Viral reorganization of the secretory (Excellence Initiative of the University of Heidelberg) com‑ Sci. USA 112, E516–E525 (2015). pathway generates distinct organelles for RNA munity to A.-C.G. A.-E.S. was supported by the European 65. Leonard, T. A., Rozycki, B., Saidi, L. F., Hummer, G. & replication. Cell 141, 799–811 (2010). Molecular Biology Laboratory and the EU Marie Curie Hurley, J. H. Crystal structure and allosteric activation 79. Bhattacharjee, S., Stahelin, R. V. & Haldar, K. Host Actions Interdisciplinary Postdoctoral cofund Programme. of protein kinase C βII. Cell 144, 55–66 (2011). targeting of virulence determinants and 66. Groves, J. T. & Kuriyan, J. Molecular mechanisms in phosphoinositides in blood stage malaria parasites. Competing interests statement signal transduction at the membrane. Nat. Struct. Trends Parasitol. 28, 555–562 (2012). The authors declare competing interests: see Web version Mol. Biol. 17, 659–665 (2010). 80. Miao, B. et al. Small molecule inhibition of for details. 67. van den Bogaart, G. et al. Membrane protein phosphatidylinositol‑3,4,5‑triphosphate (PIP3) sequestering by ionic protein-lipid interactions. Nature binding to pleckstrin homology domains. Proc. Natl 479, 552–555 (2011). Acad. Sci. USA 107, 20126–20131 (2010). FURTHER INFORMATION 68. Yang, S. T., Kiessling, V., Simmons, J. A., White, J. M. 81. Jo, H. et al. Deactivation of Akt by a small molecule LIPID MAPS Structure Database: & Tamm, L. K. HIV gp41‑mediated membrane fusion inhibitor targeting pleckstrin homology domain and http://www.lipidmaps.org occurs at edges of cholesterol-rich lipid domains. Nat. facilitating Akt ubiquitination. Proc. Natl Acad. Sci. Chem. Biol. 11, 424–431 (2015). USA 108, 6486–6491 (2011). SUPPLEMENTARY INFORMATION 69. Koberlin, M. S. et al. A conserved circular network of 82. Jesorka, A. & Orwar, O. Liposomes: technologies and See online article: S1 (table) coregulated lipids modulates innate immune analytical applications. Annu. Rev. Anal. Chem.1, ALL LINKS ARE ACTIVE IN THE ONLINE PDF responses. Cell 162, 170–183 (2015). 801–832 (2008).

NATURE REVIEWS | MOLECULAR CELL BIOLOGY ADVANCE ONLINE PUBLICATION | 9