Global approaches for the elucidation of phosphoinositide-binding proteins.

G Model CPL-4242; No. of Pages 10

ARTICLE IN PRESS Chemistry and Physics of Lipids xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Chemistry and Physics of Lipids journal homepage: www.elsevier.com/locate/chemphyslip

Global approaches for the elucidation of phosphoinositide-binding proteins Michael D. Best ∗ Department of Chemistry, The University of Tennessee, 1420 Circle Drive, Knoxville, TN 37996, United States

a r t i c l e

i n f o

Article history: Available online xxx Keywords: Phosphoinositides Protein–lipid binding Affinity enrichment Microarray Activity-based protein profiling Computational protein prediction

a b s t r a c t Phosphoinositide lipids (PIPn s) control numerous critical biological pathways, typically through the regulation of protein function driven by non-covalent protein–lipid binding interactions. Despite the importance of these systems, the unraveling of the full scope of protein–PIPn interactions has represented a significant challenge due to the massive complexity associated with these events, including the large number of diverse proteins that bind to these lipids, variations in the mechanisms by which proteins bind to lipids, and the presence of multiple distinct PIPn isomers. As a result of this complexity, global methods in which numerous proteins that bind PIPn s can be identified and characterized simultaneously from complex samples, which have been enabled by key technological advancements, have become popular as an efficient means for tackling this challenge. This review article provides an overview of advancements in large-scale methods for profiling protein–PIPn binding, including experimental methods, such as affinity enrichment, microarray analysis and activity-based protein profiling, as well as computational methods, and combined computational/experimental efforts. © 2013 Elsevier Ireland Ltd. All rights reserved.

1. Introduction The phosphatidylinositol polyphosphate lipids (phosphoinositides, PIPn s, PtdIns-Ps) have elicited considerable interest among a broad range of researchers since the signaling lipids that comprise this family control numerous critical biological processes. A primary mechanism by which PIPn s act involves their roles as sitespecific ligands that enforce the binding of peripheral proteins to membrane surfaces, which generally regulates the function of the protein target (Lemmon, 2008; Cho and Stahelin, 2005). In these instances, protein activity can either be modulated directly through lipid binding or indirectly by bringing the bound protein into proximity of enzymes or binding partners at the membrane surface (Cho and Stahelin, 2005). PIPn sub-cellular localization is also tightly controlled, which in turn dictates the location of bound proteins within the cell (Sprong et al., 2001). The PIPn s additionally play important roles in the biosynthesis of important molecules including diacylglycerol (DAG) and the soluble inositol phosphates (InsPs) via the production of inositol-(1,4,5)-triphosphate (Ins(1,4,5)P3 , IP3 ) (Streb et al., 1983; Carrasco and Merida, 2007; Gomez-Fernandez and Corbalan-Garcia, 2007; Sakane et al., 2007). As a result of these critical signaling activities, aberrant phosphoinositide activities have been linked to a number of diseases including cancer

∗ Tel.: +1 865 974 8658; fax: +1 865 974 9332. E-mail addresses: [email protected], [email protected]

and diabetes (Di Paolo and De Camilli, 2006; Pendaries et al., 2003; Vicinanza et al., 2008; Wymann and Schneiter, 2008). Despite the significance of the PIPn family, the quest for a complete understanding of the activities of these lipids is complicated by the highly complex nature associated with their biological functions. This starts with the PIPn structures themselves, which contain a conserved myo-inositol headgroup that can be phosphorylated at every permutation of the 3-, 4-, and 5-positions yielding seven isomers with independent biological activities (Fig. 1a–g). As a result of the sophisticated nature of these structures, considerable efforts have been invested in the synthesis of these compounds as well as functionalized derivatives for use as probes (Best et al., 2010; Conway and Miller, 2007). The complications related to studying the protein-binding properties and associated biological functions of PIPn s also result from the massive scope and intricate details associated with these molecular recognition events. There are numerous proteins that are now known to be bound and regulated by PIPn s, including proteins that may or may not contain conserved binding units such as the PH, PX, C2, FYVE, ENTH, ANTH, PROPPIN and tubby domains (Lemmon, 2003, 2007, 2008; Cho and Stahelin, 2005; Hurley, 2006; Hurley and Misra, 2000; Lemmon and Ferguson, 2000; McLaughlin et al., 2002). This broad scope is further complicated by details at the molecular level; for example not all members of these domain families exhibit PIPn -binding, and there is significant variation in PIPn binding specificities and affinities, even within individual domain families.

0009-3084/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.chemphyslip.2013.10.014

Please cite this article in press as: Best, M.D., Global approaches for the elucidation of phosphoinositide-binding proteins. Chem. Phys. Lipids (2013), http://dx.doi.org/10.1016/j.chemphyslip.2013.10.014

ARTICLE IN PRESS


M.D. Best / Chemistry and Physics of Lipids xxx (2013) xxx–xxx

2

OR1 R2O

OH

R3O OH

O O P O O O O

O O

Fig. 2. Schematic for the affinity chromatography approach for selective enrichment of PIPn -binding proteins.

1a: PIP3, R1, R2, R3 = PO3-2 1b: PI(3,4)P2, R1, R2 = PO3-2, R3 = H 2

3

1c: PI(4,5)P2, R , R =

PO3-2,

1

R =H

1d: PI(3,5)P2, R1, R3 = PO3-2, R2 = H 1e: PI(3)P, R1 = PO3-2, R2, R3 = H 1f: PI(4)P, R2 = PO3-2, R1, R3 = H 1g: PI(5)P, R3 = PO3-2, R1, R2 = H Fig. 1. Structures of the seven naturally occurring PIPn isomers, depicted with representative acyl chains.

As a result of the impressive scale and detail associated with proteins that bind PIPn s, global methods for elucidating these targets at the genomic scale are of considerable value as they allow for the simultaneous analysis of a large number of proteins (Cho et al., 2012; Scott et al., 2012). In addition, these techniques can circumvent challenges associated with the study of membraneinteracting proteins, such as the difficulties of overexpression and purification. Toward this end, recent advances in both experimental and computational approaches, as well as combinations thereof, have significantly impacted the understanding of these complex recognition events. This review article will provide an overview of advances in such large scale studies aimed at the elucidating PIPn binding proteins, and then conclude by discussing the benefits and complementarity of the various approaches as well as the prospects for future studies. 2. Experimental approaches for large scale identification of PIPn -binding partners 2.1. Affinity enrichment using PIPn -derivatized solid supports A prevalent approach to performing global identification of PIPn -binding proteins from complex samples has involved affinity enrichment. In this area, various moieties containing aspects of the PIPn structures have been attached onto solid support in order to separate cognate binding proteins based on non-covalent binding interactions with the resin-bound ligand. In this way, nonbinding proteins are first eluted off, followed by subsequent release of the solid-support bound target proteins, as depicted in Fig. 2. As

can be seen in the following cases, the affinity chromatography approach has significantly benefitted from recent advancements in bioanalytical tools, and particularly mass spectrometry (MS)-based proteomic protocols for the large-scale identification of protein components in complex mixtures (Cravatt et al., 2007). Affinity enrichment was initially reported for the isolation and identification of individual PIPn -binding proteins. For example, Hammonds-Odie and co-workers used a combination of affinity chromatography and photoaffinity labeling (Dorman and Prestwich, 2000) to identify and characterize centaurin-␣ as a PIPn binding protein from rat brain (Hammonds-Odie et al., 1996). In prior work involving the identification of proteins that bind the soluble Ins(1,3,4,5)P4 via an InsP affinity resin (Theibert et al., 1991, 1992), the authors observed a 46-kDa protein, but this protein was not labeled using an Ins(1,3,4,5)P4 probe bearing a phenylazide photoaffinity tag. Further studies revealed that this protein was successfully labeled with photoaffinity probe 2 (Fig. 3), which was attributed to the enhanced hydrophobicity and increased stability of the benzophenone affinity tag. In addition, this protein was most effectively competed off of affinity resin using PIP3 , after which it was demonstrated that cloned protein could be isolated using the Ins(1,3,4,5)P4 affinity resin. This work illustrates the tricky nature of differentiating PIPn - and InsP-binding proteins due to their similar phosphorylated myo-inositol moieties and since many PIPn -binding proteins bind to the headgroups of these lipids with reasonable affinities. A similar approach was used to characterize p42IP4 (Reiser et al., 1995; Stricker et al., 1995, 1997, 2003). Tanaka and co-workers synthesized PIP3 analog 3 bearing an aminophenyl group for conjugation onto affi-gel beads containing succinimidyl ester functionalities (Tanaka et al., 1997). Notably, derivative 3 includes a carbonate linkage at the 1-position of the myo-inositol headgroup in place of the natural phosphodiester linkage. The resulting support was employed to purify protein D89940, which was then termed PIP3 binding protein (PIP3BP), containing a zinc finger and two PH domains. Next, competition experiments for PIP3 -resin binding were used to show that the protein exhibited specificity for PIP3 over PI(4,5)P2 and PI(3,4)P2 , and mutagenesis studies demonstrated that mutations in either PH domain diminished binding. In a follow-up study, PIP3 -amine conjugate 4 was developed, which contained the natural phosphodiester linkage of the 1-position of the myo-inositol headgroup as well as a simplified


ARTICLE IN PRESS


M.D. Best / Chemistry and Physics of Lipids xxx (2013) xxx–xxx

-2 -2

OH

O3PO

O3PO -2

-2 -2

H N

-2

OH

O3PO

NH2 O

O H R2 HN

1

P O

O

O

5a: R1 = PO3-2, R2 = C17H35 5b: R1 = H, R2 = C5H11

3 O O

-2

O3PO

OH O

O3PO -2

O3PO

P O

OH

-2

NH2

O3PO -2

O O

P O

O O

H N

N 6H

O

OH

O3PO

O

7

O

HN

4 R1O

-2

OH

O3PO

O

O

O

-2

O O

-2

O

N H

OH O

O3PO R2O

H NH

OH

R O

O

O

O O

O3PO

O

OH

O3PO

OH

O3PO

-2

S 4

5

O

T

O

H N

O

T

2

O3PO -2

O

O O P O O O

OH

O3PO

3

O

OH

P

O O

O O

6a: R1, R2 = PO3-2 6b: R1 = H, R2 = PO3-2

O

O O

N 5H

S

H N

S O

O N H

S 4 H

H NH N H

O

6c: R1 = PO3-2, R2 = H Fig. 3. Examples of synthetic PIPn probes employed for global identification studies including affinity chromatography and activity-based protein profiling.

glycerolipid backbone containing an amine nucleophile (Shirai et al., 1998). This probe was found to be more effective at binding target proteins and led to the enrichment of protein targets Tec, Gap1m and Akt. In a somewhat different approach, Klarlund and co-workers screened cDNA expression libraries with 32 P-labeled PIPn s to identify GRP1 as a PIP3 -binding protein (Klarlund et al., 1997). Rao and co-workers employed synthetic biotinylated analogs of both PIP3 (5a) and PI(3,4)P2 (5b) for immobilization onto streptavidin-coated beads to pursue affinity enrichment (Rao et al., 1999). This approach was combined with libraries of radiolabeled proteins produced through coupled in vitro transcription/translation to identify protein binding targets. For proof-of-concept, [35 S]methionine-labeled Akt was studied, which was captured by beads containing 5a and 5b using both homogeneous protein sample as well as when mixed with a small library of 100 cloned proteins. Next, resin beads were applied to screen 500 pools of the library, after which PIP3BP, PDK1, and a novel protein termed PHISH, were obtained. Wang and co-workers also exploited a biotinylated PIP3 analog for affinity enrichment, in this case for the isolation of recombinant proteins p85-SH2 and PLC␥-PH from E. Coli (Wang et al., 2000). In addition, SWAP-70 was identified through affinity purification from bovine brain (Shinohara et al., 2002) and the design and synthesis of other synthetic PIPn affinity matrices were reported around this time as well (Painter et al., 2001; Lim et al., 2002). Krugmann and co-workers provided an important advancement by utilizing trypsin digestion and MS detection to identify proteins enriched by affinity chromatography (Krugmann et al.,

2002). Following optimization using Akt, this approach facilitated the identification of 21 proteins from pig leukocyte cytosol using PIP3 , PI(3,4)P2 , PI(3,5)P2 and PI(3)P affinity matrices, including 14 previously known proteins and 7 novel targets. It was reported to be more difficult to identify PI(3)P-binding proteins due to a high background of non-specific binding to the beads. In addition, prefractionation of cell extracts by ion exchange chromatography was used to decrease the sample complexity. From the identified proteins, the authors focused on ARAP3 for a detailed analysis, which was purified using PIP3 -beads and could be competed off with PIP3 . While this protein contains five predicted PH domains, mutation of the N-terminal PH domain was found to abrogate PIP3 binding. Further studies demonstrated that GFP-tagged ARAP3 translocated to membranes upon EGF stimulation of PIP3 production, ARAP3 acted as a specific Arf6 GAP that was activated by PIP3 , and ARAP3 overexpression led to significant changes in cell morphology. This work illustrates the power of global methods utilizing MS for identifying larger numbers of protein targets, but then focusing on specific examples to elucidate key biochemical pathways. Pasquali and co-workers employed affinity chromatography with tandem MS-based detection in conjunction with cleavable linker technology to identify 10 previously known and 11 putative PIPn -binding proteins from mouse bone marrow macrophages (Pasquali et al., 2007). This approach benefited from efficacy on the analytical scale, thus circumventing the need for large quantities of biological material. Here, probes 6a–c, consisting of either PIP3 , PI(4,5)P2 or PI(3,4)P2 analogs linked to biotin affinity anchors via a cleavable disulfide linker, were exploited. In initial optimization studies using HEK-293 cells with overexpressed Akt, the cleavable


G Model CPL-4242; No. of Pages 10 4

ARTICLE IN PRESS M.D. Best / Chemistry and Physics of Lipids xxx (2013) xxx–xxx

linker approach was found to be advantageous by enhancing Akt detection compared to a comparable probe lacking the disulfide due to the detachment of bound protein from the affinity resin. Of the 11 novel proteins, 3 were selected for further validation studies by PIP stripsTM and overlay using 32 P-labeled PIPn s, and PI(3,4)P2 binding specificity was confirmed for PRDX1 and CAP1. The latter radiolabeling approach was also developed for identifying complexes with protein kinases. In 2008, Catimel and co-workers reported systematic efforts to define the interactomes of PI(3,5)P2 and PI(4,5)P2 from colorectal carcinoma cells (Catimel et al., 2008). In this case, to maximize the coverage of protein targets, experiments were performed using both synthetic PIPn analogs covalently attached to resin beads for affinity enrichment as well as PIPn -containing liposomes, for which enrichment had previously been demonstrated (Knodler and Mayinger, 2005) and samples with bound proteins were separated by size exclusion chromatography. This combination was performed based on the knowledge that proteins exhibit variability in their ability to bind individual synthetic PIPn motifs and liposomes presenting these lipids. In both cases, enriched proteins were separated via SDS-PAGE and excised bands were analyzed using LC/MS/MS. Overall, of the 388 target proteins that were identified, 105 were isolated via only PI(3,5)P2 experiments, 187 resulted from experiments involving PI(4,5)P2 , and 96 were identified from both. In addition, the combined liposome/PIPn -resin approach paid off as these two types of samples resulted in differing, although overlapping, proteins in both cases. In addition to the extensive list of putative PIPn -binding proteins, these results provided insights into binding domain targets, such as the presence of 14 proteins with HEAT domains, and biological activities, represented by the identification of 72 actin-binding proteins. This report was followed a year later by a similar effort to elucidate the PIP3 interactome, from which 282 proteins were elucidated, including 82 using liposome separation, 175 via PIP3 beads and 25 that overlapped between these (Catimel et al., 2009; Conway et al., 2010). Zhang and co-workers used PIP3 -based affinity chromatography to identify 9 protein binding targets in Dictyostelium discoideum (Zhang et al., 2010). This included 3 myosin proteins, acetyl-CoA decarboxylase, and 5 PH domain-containing proteins, both known (Akt and PhdA) and unknown (PhdB, PhdG, and PhdI). All three of the latter proteins exhibited specific PIP3 -binding in vitro using lipid blotting and in vivo using localization studies via fluorescence microscopy-based detection of GFP-fusions, although PhdG exhibited diminished membrane binding in response to PIP3 . In addition, PhdB was the lone protein in which membrane association was not disrupted through the use of the PI3K inhibitor LY294002, and membrane binding also went unchanged in pten cells. Also, cAMP stimulation of PIP3 was observed to trigger the translocation of PhdB and PhdI, but not PhdG. In migrating cells, PhdI localized preferentially to the leading edge, PhdB prefered the lagging edge, and PhdG did not exhibit any particular enrichment. In mutagenesis studies, mutation of the PH domain of PhdI abrogated binding, only the first of two PH domains was found to be necessary for PhdG binding, and PhdB exhibited a complex result in which PH domain mutation was detrimental for binding in vitro, but not in vivo. Dixon and co-workers implemented a multi-stage approach to enhance the enrichment and detection of PIPn -binding proteins from 1321N1 Astrocytoma cells (Dixon et al., 2011). First, cells were treated with vanadate to increase PI(3,4)P2 concentration due to the activation of class I PI3K and SHIP2 and the simultaneous inhibition of PI(3,4)P2 4-phosphatases (Batty et al., 2007). This was performed to promote translocation of target proteins to membrane fractions as an initial phase of enrichment, and was validated through the tracking of the known protein target TAPP-1. Next, biotinylated PI(3,4)P2 was implemented for further affinitybased enrichment of protein binding partners. In initial MS based

analysis, detection of target proteins proved problematic due to low amounts. To circumvent this problem, stable isotope labeling with amino acids in cell culture (SILAC) was exploited to enhance detection by using control- and test-populations with light (12 C) and heavy (13 C) atom-labeled amino acids, respectively. This led to the identification of ∼80–85 proteins with isotopic label ratios of 1:1.5 as putative protein targets, including previously known and unknown protein binding partners. Next, the authors focused on the validation of IQGAP1 due to its key roles in multiple signaling pathways and its borderline detection ratio in analysis. Here, SPR analysis and lipid overlays were used to show that IQGAP1 binds to PIP3 but not PI(3,4)P2 or PI(4,5)P2 . 2.2. Microarray analysis and high-throughput screening Microarray methodology has additionally been developed for the identification and characterization of protein–PIPn to take advantage of the high-throughput nature of this technique. In general microarray analysis, one component of a recognition event is immobilized onto a surface, such as microarray slides, in order to detect non-covalent binding interactions with partners contained in incubated solutions. Thus, for microarray analysis of protein–lipid binding events, either the lipid or protein could be immobilized onto the surface, both of which approaches have been undertaken to elucidate binding (Fig. 4) (Feng, 2005; Bally et al., 2010). In addition, the lipid component in these studies can once again range from individual synthetic lipid analogs to liposomal samples. Finally, microarray has also been reported as a means of characterizing protein–lipid binding interactions, both through the immobilization of individual PIPn (Gong et al., 2009a; Rowland et al., 2012) motifs as well as whole liposomes (Losey et al., 2009; Smith and Best, 2011). In an early example, Zhu and co-workers developed a proteome chip in which they cloned, overexpressed, and purified 5800 proteins from yeast (approximately 80% of the yeast proteome) that were then printed onto slides for analysis (Zhu et al., 2001). This protein array was then incubated with liposomes composed of phosphatidylcholine (PC) or PC with 5% of PI(3)P, PI(4)P, PI(3,4)P2 , PI(4,5)P2 or PIP3 , and 1% of a biotin–PE conjugate to enable detection of bound liposomes using Cy3-streptavidin. This led to the identification of 150 protein binding targets, 49 of which bound preferentially to one or more of the evaluated PIPn isomers. Later, it was pointed out that many of the proteins identified by this screen are not known to possess lipid-binding domains (Gallego et al., 2010). Kanter and co-workers developed lipid microarrays to study Tcell and autoantibody binding of lipids associated with multiple sclerosis (Kanter et al., 2006). This involved the spraying of lipids including gangliosides, sulfatides, cerebrosides, sphingomyelin and other brain lipids onto microarray slides, and led to the conclusion that autoimmune responses to sulfatides and other lipids may contribute to pathogenesis. Gallego and co-workers extended this approach by immobilizing 51 lipids and metabolic intermediates onto nitrocellulose for analysis, including PIP3 , PI(4,5)P2 , PI(3,4)P2 , PI(4)P, and PI(3)P and PI as well as other glycerophospholipids, sphingolipids, fatty acids, glycerolipids, sterols and prenols (Gallego et al., 2010). This library was then screened using 172 soluble proteins expressed with TAP tags, including 91 proteins containing lipid-binding domains, and bound proteins were detected using an anti-TAP antibody. After data filtering to remove potential false positives and negatives relating to promiscuous and water-soluble lipids, 530 interactions were identified corresponding to combinations of 124 proteins with 30 lipids, with the PIPn s representing the most common targets. The lipid array was found to identify 60% of previously known interactions, with many missed interactions attributed to additional binding requirements, and 68% of detected




5

B.Protein Immobilization

A.Lipid Immobilization

BoundPIP n lipidmotif

PIPn-binding protein

non-binding proteins

wash Immobilized PIPn lipidmotif

Immobilized PIPn-binding protein

Fig. 4. Cartoon depicting the approaches developed for microarray analysis. (A) Surface immobilization of PIPn lipid motifs allows for the binding and detection of target proteins. (B) Alternatively, potential target proteins can be attached to the surface and treated with lipids, such as in liposome formulations, to detect binding.

interactions were novel. Subsequently, 10 proteins representing 34 protein–lipid pairs were selected for validation via binding to liposomes and biological membranes, and 24 interactions involving 8 proteins were verified. Additionally, several novel interactions with sphingolipids, and ligand targets of PH domains were characterized. Recently, Lu and co-workers optimized the signal of fluorescence-tagged liposomes in order to screen a yeast proteome microarray to characterize PI(3,5)P2 -binding proteins (Lu et al., 2012). First, to circumvent the issue of the quenching of liposomal fluorophores, the amounts of encapsulated sulforhodamine B and lissamine rhodamine-phosphatidylethanolamine (PE) incorporated in the lipid bilayer were optimized to maximize fluorescence signal. Next, yeast proteome chips were screened for the binding of fluorescent liposomes containing 0.4% PI(3,5)P2 , and using criteria that the authors considered stringent due to high standard deviations in analysis, 162 protein targets were implicated. This list included previously reported effectors such as Sro77p, Pkc1p and Abp1p. These results were then validated by capture of GST-fusions of the protein in question onto glutathione beads, treatment with quenched liposomes, removal of unbound liposomes and lysis to release the fluorophore of bound liposomes. Subsequently, a consensus sequence was discovered in 8 of 22 selected proteins using a MEME motif search that was found to be statistically significant in the list of 162 putative targets, which may or may not be the binding site. Finally, docking studies using this sequence indicated high confidence for the binding of the PI(3,5)P2 headgroup. 2.3. Activity-based lipid probes The strategy of activity-based protein profiling (ABPP) has advanced the annotation of protein function through the use of probes that selectively label the active site of target proteins in complex mixtures, enabling the collective purification, identification and characterization of enzyme–substrate and protein–ligand interactions (Cravatt et al., 2008). Furthermore, the incorporation of click chemistry into the ABPP strategy has proven fruitful by minimizing the modifications made in probe structures for use as activity-based probes (ABPs) (Speers and Cravatt, 2004; Best, 2009). ABPP shares some characteristics with affinity chromatography in that protein targets are selectively enriched from complex biological samples. However, ABPs contain modifications that enable the covalent labeling of target proteins, such as the introduction of chemically reactive groups (typically for substrate analogs) or photoaffinity tags (commonly used for ligand analogs). Benefits of ABPP include that covalent labeling of targets can circumvent complex

dissociation during processing, and that this approach allows for labeling in live cells and organisms. Recently, ABPP has begun to impact lipid studies (Best et al., 2011). Gubbens and co-workers developed a range of PC ABPs containing both a photoaffinity label and click chemistry tags for the proteomic elucidation of PCbinding proteins (Gubbens and de Kroon, 2010; Gubbens et al., 2007, 2009). In addition, enzymes that manipulate PC have been labeled using ABPs containing a fluorophosphonate reactive group in place of one acyl chain (Tully and Cravatt, 2010). Rowland and co-workers developed and applied PIPn activity probes for the elucidation of cognate binding partners in human melanoma cells (Rowland et al., 2011). Here, PI(3,4,5)P3 probe 7 was initially designed and synthesized, which contains both a benzophenone photoaffinity tag as well an alkynyl secondary label linked via a Y-shaped lysine linker. While this probe contains the PIP3 headgroup, the glycerolipid backbone is simplified by using an aminohexyl chain to provide hydrophobicity and also allow for modular synthesis (Gong et al., 2009b). The alkynyl tag provides versatility in analysis, as a fluorescent dye can be clicked on for initial validation and optimization studies using gel electrophoresis, while alternatively a biotin moiety can be introduced via click chemistry for affinity enrichment of protein targets (Fig. 5). In the validation stage, probe 7 yielded robust labeling of the PH domain of Akt in fluorescence-based gel studies, for which the signal was abrogated by controls involving no irradiation, heat-denatured protein, and competition with an unlabeled PIP3 headgroup as a competitor. Additionally, a probe containing a longer linker between the headgroup and labeling unit was found to be ineffective for labeling of purified Akt or cell extracts, indicating the importance of the placement of the photoaffinity tag. Finally, enrichment was followed by tandem MS, which resulted in 265 putative protein targets, at least 44 of which had been previously identified using affinity enrichment. 3. Computational methods for predicting PIPn -binding proteins Computational methods have additionally proven invaluable for predicting, identifying and characterizing PIPn targets, and have followed a similar path involving a significant enhancement in scope as a result of technological advancements and progress in the understanding of the details of protein–lipid recognition events. While domain families that include multiple PIPn -binding proteins can be annotated using sequence homology searching (DiNitto and Lambright, 2006), since membrane-binding abilities and specificities vary significantly within these families, sequence searches are




Fig. 5. Strategy for the labeling and identification of target proteins using PIPn activity probes. Probes are incubated with cell extracts to enable protein binding (Step 1), after which irradiation is used to photo-cross-link probes to bound proteins (Step 2). Subsequent tagged cross-linked proteins are then analyzed by using click chemistry to introduce a fluorescent label for in-gel detection (Step 3) or via biotinylation (Step 4), streptavidin enrichment and mass spectrometry-based proteomics for structure identification (Step 5).

not always effective at predicting membrane association. Therefore, computational efforts have utilized combinations of structural and sequence-based searches to predict membrane binding activity (Cho et al., 2012; Scott et al., 2012). In addition, many of the subsequent examples have taken a combined computational and experimental approach to predict, characterize and validate protein–PIPn binding interactions (Fig. 6).

For example, Isakoff and co-workers identified a consensus sequence of PH domains through alignment of the sequences of domain family members known to bind PIPn s along with comparison to non-binding members (Isakoff et al., 1998). Following the identification of conserved motifs, the authors searched for other PH members containing this sequence, and identified putative targets such as Gab1, Dos, and myosinX. Next, an in vivo assay in

Fig. 6. Overview of computational approaches to the large scale identification of PIPn -binding proteins. Strategies including predictive algorithms, consensus sequence identification, and conclusions derived from structural analysis of known protein targets have been used to search protein databanks to predict novel PIPn -binding proteins. Following this, in vitro and in vivo binding assays have been used to validate results and refine prediction methods for further analysis.




yeast was developed to validate binding properties. This utilized a mutant temperature-sensitive Ras exchange factor (cdc25ts) that could be rescued by membrane targeting. To do so in a PIPn -binding dependent manner, an activated farnesylated p110 domain of PI3K was expressed in yeast along with a fusion construct of cdc25ts attached to the PH domain to be studied. In this way, the production of 3-phosphorylated PIPn s by p110 would result in the recruitment of cdc25ts to membranes via the fused PH domain, which would allow yeast growth, thus confirming PH domain binding properties. In studies, Gab1, Dos and myosinX all rescued yeast growth, as did 3 out of 4 additional PH domains predicted to bind (EST810295, EST2301443 and EST684797). Mutations to PH domains, and domains lacking the conserved sequence were not found to rescue yeast growth. Finally, PIP3 binding of EST684797 was confirmed by fluorescence microscopy and isothermal titration calorimetry (ITC). The identification of the consensus sequence led to further characterization of the PIPn -binding properties of target proteins (Dowler et al., 2000). Bhardwaj and co-workers developed a structural bioinformatics model for predicting general membrane-binding proteins (Bhardwaj et al., 2006). In doing so, a support vector machine (SVM) was trained using a machine learning algorithm to discern the difference between membrane-binding and non-membranebinding proteins on the basis of three characteristics: (1) net charge, assigned using CHARMM force-field parameters; (2) distribution of surface cationic charges; and (3) amino acid composition, which was separated into overall and surface composition. From data sets, it was determined that membrane-binding proteins were best identified by enhanced positive charge, and sizes of the largest cationic patches, while the total cationic patch size showed less correlation. In addition, correlation was found for proximity of cationic patches and membrane-binding surfaces, as well as the presence of aliphatic (such as Val, Leu, Ile, and Met), aromatic (Trp), cationic (Lys, Arg) and Cys residues on the membrane-binding surface, although overall amino acid composition was not found to be an indicator. After these features were fed to the SVM and accuracy was optimized to 93.7%, this system was applied to study the membrane-binding activities of multiple isoforms of the protein kinase C (PKC) family, which possess C2 binding domains. Despite these isoforms having >50% homology, the prediction was made that PKC␪-C2 was membrane-binding while PKC␦-C2, PKC␧-C2, and PKC␩-C2 were not. This result was validated using surface plasmon resonance (SPR) binding studies using liposomes containing PC, phosphatidylserine (PS) and PI-(4,5)-P2 . This work was later extended through the use of PU-learning (Bhardwaj et al., 2010) as well as an alternating decision tree to aid in the differentiation of binding and non-binding proteins within the same domain family (Kallberg et al., 2012). Zhang and Arnold also analyzed C2 domains by using sequence profile searches, phylogenic and phyletic pattern analysis and structure prediction to categorize these into distinct families of C2 domains (Zhang and Aravind, 2010). Also reported in 2006, Lin and co-workers developed and applied an SVM prediction system to search a large number of proteins from the Swiss-Prot database for proteins involved in lipid degradation, lipid metabolism, lipid synthesis, lipid transport, lipid binding, lipopolysaccharide biosynthesis, and protein lipidation (Lin et al., 2006). This SVM was developed using 14,776 lipid-binding and 133,441 non-lipid binding proteins and evaluated using an independent set of 6768 binding and 64,761 non-bonding proteins. Analysis resulted in predicted sensitivities (accuracy of predicting binding proteins) ranging from 78.9 to 90.6% and predicted specificity values (accuracy of predicting non-binding proteins) of 97 to 99.9%. This SVM also predicted 76 lipid-binding proteins that do not possess conserved sequences associated with lipid recognition.

7

Park and co-workers used a combination of experimental and computational approaches to identify PIP3 -binding PH domains and applied these results to advance prediction methods (Park et al., 2008). Initially, 130 PH domains from mouse were conjugated to yellow fluorescent protein (YFP) and analyzed for PIP3 binding through confocal imaging in response to PDGF stimulation. Following a series of controls involving PI3K inhibition and mutation, 26 PH domains that target PIP3 were identified. In applying these results for developing a prediction tool, a sequence-based approach was initially pursued, but was abandoned due to low sequence homology, poor alignment, and lack of other criteria that correctly implicated the 26 identified domains. Subsequently, a recursivelearning strategy was undertaken by calculating a two-dimensional sequence profile matrix for the identified PIP3 -binding PH domains. This system was applied to analyze an independent set of PH domains from humans, mice, Caenorhabditis elegans, Drosophila, Saccharomyces cerevisiae, and Schizosaccharomyces pombe. Overall, of 36 PH domains studied theoretically and experimentally, 14 were predicted to bind PIP3 , and 13 were validated through microscopy. These results were then used to improve the algorithm and elucidate the most important factors for PIP3 -binding. Silkov and co-workers developed a high-throughput homology modeling including in-depth analysis of biophysical properties to study the ANTH domain family at the genomic scale (Silkov et al., 2011). In doing so, SkyLine (Mirkovic et al., 2007) was employed to collect sequences similar to ANTH domains that have a high probability of adopting ANTH-like protein folds, which provided 246 results. To predict membrane binding among these, structures were evaluated based on the presence of PI(4,5)P2 conserved binding sequences from the ANTH domain of CALM, surface electrostatic properties, and the availability of an amphiphilic ␣-helix or exposed hydrophobic residue for membrane penetration. One observation made during searches was the presence of “enhanced” and “super-enhanced” binding domains containing extra amino acids compared to the “classic” PI(4,5)P2 binding sequence. Overall, only 66% of the identified ANTH domains contained the “classic” conserved binding domain. For validation, a representative member (CAP8 ARATH, AAL11587) was studied via high quality composite modeling, SPR analysis indicating a preference for PIP3 binding over PI(4,5)P2 and PI(3,4)P2 , lipid monolayer assay to evaluate the capability of membrane penetration, and vesicle tabulation assay to assess membrane deformation. Chen et al. implemented a combination of experimental and computational approaches to characterize the lipid-binding properties of PDZ domains, of interest due to their complex binding properties in which they target both lipids and other proteins (Chen et al., 2012). To do so, 70 primarily uncharacterized PDZ domains were first selected and analyzed for the binding of liposomes with compositions similar to the inner-plasma membrane through SPR. Of these, 27 domains exhibited sub-micromolar Kd values, most of which did not show selectivity among the PIPn s. These results were then used to build a computational model (Bhardwaj et al., 2006), which was applied to screen 2000 PDZ domains from 20 species, with 30% of the targets predicted to bind membranes. For validation, 25 mostly uncharacterized PDZ domains were selected for SPR analysis, 6 of which were predicted to bind membranes, with 3 ultimately exhibiting Kd values below 1 ␮M, resulting in an overall accuracy of 90%. Modeling studies were also used to categorize PDZ domains into distinct groups based on the proximity of cationic patches to the peptide-binding domain, including class A in which these are distal, and class B in which the two domains are proximal. For experimental studies, class A example SAP102-PDZ3 was selected, which had a prominent cationic patch that formed a groove, was found to exhibit selectivity for PI(4,5)P2 and PIP3 over other PIPn s, and the presence of liposomes was not found to impact




8

peptide-binding properties. For class B, Rhophilin 2-PDZ, containing cationic residues clustered proximal to the peptide-binding domain, was selected, which did not exhibit PIPn specificity, and the presence of liposomes resulted in a 2-fold increase in peptide binding affinity. These results demonstrate the power of computational methods to not only predict binding properties, but also indicate the molecular level properties that result in variations among similar proteins and within domain families. In addition to these reports, many of which focused on PIPn binding, other efforts have been made to predict lipid-binding proteins more generally. For example, an algorithm developed to predict lipid-binding interfaces based on the presence of hydrophobic or amphipathic sequences implicated proteins including ␣-actinin, Arp2, CapZ, talin and vinculin, which were validated using FTIR spectroscopy, film balance, T-jump, CD spectroscopy and calorimetry studies (Tempel et al., 1995; Scott et al., 2006). In addition, an SVM was used to predict lipid-binding proteins by selecting amino acids within 3.5 A˚ of the lipid in crystal structures of complexes obtained from the Protein Data Bank (PDB). The amino acids were then characterized based on side chain pKa value, hydrophobicity and molecular mass (Irausquin and Wang, 2007). The results of this study indicated enhanced frequencies of phenylalanine and tyrosine at lipid-binding sites, and values for sensitivity (52.78%) and specificity (70.84%) were reported. An SVM based on position-specific scoring matrix (PSSM) was used to analyze 1262 X-ray crystal structures, which contained 536 lipidbinding proteins, and an accuracy of 80.86% was reported (Xiong et al., 2010).

4. Conclusions As can be seen from this review, a wide variety of approaches have been reported in recent years to unravel the complex details of protein-PIPn binding interactions. Further, each of these strategies has seen significant enhancement in the scope of studies that can be performed efficiently as a result of technological advancements in analysis and computational prediction. Thus, while these approaches were initially successful for the analysis and identification of a few proteins, the scope has quickly been ramped up to the elucidation of hundreds of target proteins simultaneously. From the results that have been obtained using these varying techniques, it is clear that these methods are complementary. In particular, when comparing the lists of proteins obtained from different types of studies, it is common to observe differing but overlapping results. For example, details such as the choice of PIPn moiety (synthetic lipid analogs versus liposomal formulations) affect the proteins that are detected. This suggests that there may not be a single method that will be effective at analyzing all PIPn -binding proteins, which is perhaps not surprising due to the massive scope of these interactions, and since there is widespread variation in the structures of target proteins and in the mechanisms and molecular level details associated with these binding interactions. As can be seen with the probes depicted in Fig. 3, studies have employed diverse PIPn analogs, with structures varying in terms of the linkage and location of reporter tags and the extent of the glycerolipid backbone that is included. Often, this variation arises from choices made during the design process that are affected by considerations such as synthetic accessibility. While it is challenging to deconvolute how the structures of these probes affect results, the pursuit of varying avenues can be seen as beneficial for ultimately providing a broader picture of PIPn -binding proteins. Furthermore, the complex nature of the environment in which these events occur also presents a significant challenge for

elucidation and leads to advantages and disadvantages for each of the described methods. Affinity chromatography benefits from the efficiency of enriching large numbers of PIPn -binding proteins directly from complex samples, but studies are generally performed using extracts, and less abundant proteins and those that are more weakly bound may be lost during processing. Activity-based protein profiling may circumvent the dissociation of the complex through covalent labeling and enables studies in the native environment of live cells and organisms, but the efficacy of covalent labeling depends on the placement of the reactive group and may vary from protein to protein based on the particular 3-dimensional structure and the availability of proximal residues for cross-linking. Microarray analysis likely benefits from simplifying analysis to the direct detection of individual binding events in a spatially resolved manner outside of the complexity of cellular extracts, but the scope can be limited by the need to produce large numbers of samples of pure protein targets. Finally, while computational methods have advanced well beyond specific consensus sequences to the extraction of complex data sets that predict membrane-binding properties, the broad variation in the structure and mechanism of PIPn -binding proteins still hinders global prediction. Thus, the availability of a number of complementary methods is quite appropriate to tackle the grand challenge of understanding protein–PIPn interactions. Otherwise, due to the intricate nature of protein–PIPn binding interactions, it is important to carefully consider experimental conditions to maximize the identification of target proteins while limiting false positives. For example, solution pH and salt concentrations are critical since these factors control the complex ionization states of the PIPn ligands as well as the target proteins, and thus can modulate the affinity and specificity of proteins toward the different PIPn isomers (Kooijman et al., 2009). The choice of the particular ions that are included is also an important consideration since cations such as calcium can play important roles in binding via bridges between the protein and lipid recognition elements (Cho and Stahelin, 2005). Furthermore, for approaches that involve the use of lipid mixtures, such as liposomal samples, the choice of lipid components is key since many PIPn binding proteins have additional lipid-binding domains, including both cytosolic and integral membrane proteins. For example, the PH domain of Akt has been shown to bind to PS in addition to PIP3 (Huang et al., 2011) and phospholipase D is known to prefer membranes containing both PI(4,5)P2 and phosphatidylethanolamine (PE) (Brown et al., 1993; Hodgkin et al., 2000). Thus, it would be ideal to analyze liposomes containing different combinations of complementary lipids and see how this affects protein labeling studies. Finally, while the strategies described in this review have been effective for elucidating PIPn -binding proteins, the list of proteins that result from these analyses should nevertheless best be viewed as putative targets until binding interactions are validated using multiple assays. Fortuitously, a wide range of techniques for characterizing the details of these interactions at the molecular level have been developed, which have not been the focus of this review (Cho et al., 2001, 2012; Narayan and Lemmon, 2006). However, the rapid increase of the scale at which studies are performed makes it all the more challenging to systematically go through and validate long lists of putative protein targets, and thus the lipid community may be busy for quite some time characterizing the details of these interactions. This is exacerbated by the number of different organisms that need to be analyzed as well as each of the different PIPn isomers in order to uncover variations in PIPn signaling. Nevertheless, the efficiency of these global methods for the identification of numerous protein targets renders them as powerful tools for overcoming the longstanding challenges of elucidating critical protein–PIPn binding interactions.




Acknowledgment This material is based upon work supported by the National Science Foundation under Grant Number CHE-0954297. References Bally, M., Bailey, K., Sugihara, K., Grieshaber, D., Voros, J., Stadler, B., 2010. Liposome and lipid bilayer arrays towards biosensing applications. Small 6, 2481–2497. Batty, I.H., van der Kaay, J., Gray, A., Telfer, J.F., Dixon, M.J., Downes, C.P., 2007. The control of phosphatidylinositol 3,4-bisphosphate concentrations by activation of the Src homology 2 domain containing inositol polyphosphate 5-phosphatase 2, SHIP2. Biochem. J. 407, 255–266. Best, M.D., 2009. Click chemistry and bioorthogonal reactions: unprecedented selectivity in the labeling of biological molecules. Biochemistry 48, 6571–6584. Best, M.D., Zhang, H.L., Prestwich, G.D., 2010. Inositol polyphosphates, diphosphoinositol polyphosphates and phosphatidylinositol polyphosphate lipids: structure, synthesis, and development of probes for studying biological activity. Nat. Prod. Rep. 27, 1403–1430. Best, M.D., Rowland, M.M., Bostic, H.E., 2011. Exploiting bioorthogonal chemistry to elucidate protein–lipid binding interactions and other biological roles of phospholipids. Acc. Chem. Res. 44, 686–698. Bhardwaj, N., Stahelin, R.V., Langlois, R.E., Cho, W.W., Lu, H., 2006. Structural bioinformatics prediction of membrane-binding proteins. J. Mol. Biol. 359, 486–495. Bhardwaj, N., Gerstein, M., Lu, H., 2010. Genome-wide sequence-based prediction of peripheral proteins using a novel semi-supervised learning technique. BMC Bioinformatics 11, S1–S6. Brown, H.A., Gutowski, S., Moomaw, C.R., Slaughter, C., Sternweis, P.C., 1993. ADP-ribosylation factor, a small GTP-dependent regulatory protein, stimulates phospholipase-D activity. Cell 75, 1137–1144. Carrasco, S., Merida, I., 2007. Diacylglycerol, when simplicity becomes complex. Trends Biochem. Sci. 32, 27–36. Catimel, B., Schleber, C., Condron, M., Patsiouras, H., Connolly, L., Catimel, J., Nice, E.C., Burgess, A.W., Holmes, A.B., 2008. The PI(3,5)P2 and PI(4,5)P2 Interactomes. J. Proteome Res. 7, 5295–5313. Catimel, B., Yin, M.X., Schieber, C., Condron, M., Patsiouras, H., Catimel, J., Robinson, D., Wong, L.S.M., Nice, E.C., Holmes, A.B., Burgess, A.W., 2009. PI(3,4,5)P3 interactome. J. Proteome Res. 8, 3712–3726. Chen, Y., Sheng, R., Kallberg, M., Silkov, A., Tun, M.P., Bhardwaj, N., Kurilova, S., Hall, R.A., Honig, B., Lu, H., Cho, W., 2012. Genome-wide functional annotation of dualspecificity protein- and lipid-binding modules that regulate protein interactions. Mol. Cell 46, 226–237. Cho, W., Stahelin, R.V., 2005. Membrane-protein interactions in cell signaling and membrane trafficking. Annu. Rev. Biophys. Biomol. Struct. 34, 119–151. Cho, W.W., Bittova, L., Stahelin, R.V., 2001. Membrane binding assays for peripheral proteins. Anal. Biochem. 296, 153–161. Cho, H., Wu, M., Bilgin, B., Walton, S.P., Chan, C., 2012. Latest developments in experimental and computational approaches to characterize protein–lipid interactions. Proteomics 12, 3273–3285. Conway, S.J., Miller, G.J., 2007. Biology-enabling inositol phosphates, phosphatidylinositol phosphates and derivatives. Nat. Prod. Rep. 24, 687–707. Conway, S.J., Gardiner, J., Grove, S.J.A., Johns, M.K., Lim, Z.-Y., Painter, G.F., Robinson, D.E.J.E., Schieber, C., Thuring, J.W., Wong, L.S.-M., Yin, M.-X., Burgess, A.W., Catimel, B., Hawkins, P.T., Ktistakis, N.T., Stephens, L., Holmes, A.B., 2010. Synthesis and biological investigation of phosphatidylinositol phosphate affinity probes. Org. Biomol. Chem. 8, 66–76. Cravatt, B.F., Simon, G.M., Yates, J.R., 2007. The biological impact of massspectrometry-based proteomics. Nature 450, 991–1000. Cravatt, B.F., Wright, A.T., Kozarich, J.W., 2008. Activity-based protein profiling: from enzyme chemistry. Annu. Rev. Biochem. 77, 383–414. Di Paolo, G., De Camilli, P., 2006. Phosphoinositides in cell regulation and membrane dynamics. Nature 443, 651–657. DiNitto, J.P., Lambright, D.G., 2006. Membrane and juxtamembrane targeting by PH and PTB domains. Biochim. Biophys. Acta 1761, 850–867. Dixon, M.J., Gray, A., Boisvert, F.M., Agacan, M., Morrice, N.A., Gourlay, R., Leslie, N.R., Downes, C.P., Batty, I.H., 2011. A screen for novel phosphoinositide 3-kinase effector proteins. Mol. Cell. Proteom. 10, M110.003178. Dorman, G., Prestwich, G.D., 2000. Using photolabile ligands in drug discovery and development. Trends Biotechnol. 18, 64–77. Dowler, S., Currie, R.A., Campbell, D.G., Deak, M., Kular, G., Downes, C.P., Alessi, D.R., 2000. Identification of pleckstrin-homology-domain-containing proteins with novel phosphoinositide-binding specificities. Biochem. J. 351, 19–31. Feng, L., 2005. Probing lipid–protein interactions using lipid microarrays. Prostaglandins Other Lipid Mediat. 77, 158–167. Gallego, O., Betts, M.J., Gvozdenovic-Jeremic, J., Maeda, K., Matetzki, C., AguilarGurrieri, C., Beltran-Alvarez, P., Bonn, S., Fernandez-Tornero, C., Jensen, L.J., Kuhn, M., Trott, J., Rybin, V., Muller, C.W., Bork, P., Kaksonen, M., Russell, R.B., Gavin, A.C., 2010. A systematic screen for protein–lipid interactions in Saccharomyces cerevisiae. Mol. Syst. Biol. 6, 430. Gomez-Fernandez, J.C., Corbalan-Garcia, S., 2007. Diacylglycerols, multivalent membrane modulators. Chem. Phys. Lipids 148, 1–25. Gong, D., Smith, M.D., Manna, D., Bostic, H.E., Cho, W., Best, M.D., 2009a. Microplatebased characterization of protein-phosphoinositide binding interactions using a synthetic biotinylated headgroup analogue. Bioconjugate Chem. 20, 310–316.

9

Gong, D., Bostic, H.E., Smith, M.D., Best, M.D., 2009b. Synthesis of modular headgroup conjugates corresponding to all seven phosphatidylinositol polyphosphate isomers for convenient probe generation. Eur. J. Org. Chem. 24, 4170–4179. Gubbens, J., de Kroon, A., 2010. Proteome-wide detection of phospholipid–protein interactions in mitochondria by photocrosslinking and click chemistry. Mol. Biosyst. 6, 1751–1759. Gubbens, J., Vader, P., Damen, J.M.A., O’Flaherty, M.C., Slijper, M., de Kruijff, B., de Kroon, A., 2007. Probing the membrane interface-interacting proteome using photoactivatable lipid cross-linkers. J. Proteome Res. 6, 1951–1962. Gubbens, J., Ruijter, E., de Fays, L.E.V., Damen, J.M.A., de Kruijff, B., Slijper, M., Rijkers, D.T.S., Liskamp, R.M.J., de Kroon, A., 2009. Photocrosslinking and click chemistry enable the specific detection of proteins interacting with phospholipids at the membrane interface. Chem. Biol. 16, 3–14. Hammonds-Odie, L.P., Jackson, T.R., Profit, A.A., Blader, I.J., Turck, C.W., Prestwich, G.D., Theibert, A.B., 1996. Identification and cloning of centaurin-alpha – a novel phosphatidylinositol 3,4,5-trisphosphate-binding protein from rat brain. J. Biol. Chem. 271, 18859–18868. Hodgkin, M.N., Masson, M.R., Powner, D., Saqib, K.M., Ponting, C.P., Wakelam, M.J.O., 2000. Phospholipase D regulation and localisation is dependent upon a phosphatidylinositol 4,5-bisphosphate-specific PH domain. Curr. Biol. 10, 43–46. Huang, B.X., Akbar, M., Kevala, K., Kim, H.Y., 2011. Phosphatidylserine is a critical modulator for Akt activation. J. Cell Biol. 192, 979–992. Hurley, J.H., 2006. Membrane binding domains. Biochim. Biophys. Acta 1761, 805–811. Hurley, J.H., Misra, S., 2000. Signaling and subcellular targeting by membranebinding domains. Annu. Rev. Biophys. Biomol. Struct. 29, 49–79. Irausquin, S.J., Wang, L.J., 2007. A machine learning approach for prediction of lipidinteracting residues in amino acid sequences. In: IEEE, New York. Isakoff, S.J., Cardozo, T., Andreev, J., Li, Z., Ferguson, K.M., Abagyan, R., Lemmon, M.A., Aronheim, A., Skolnik, E.Y., 1998. Identification and analysis of PH domaincontaining targets of phosphatidylinositol 3-kinase using a novel in vivo assay in yeast. Embo J. 17, 5374–5387. Kallberg, M., Bhardwaj, N., Langlois, R., Lu, H., 2012. A structure-based protocol for learning the family-specific mechanisms of membrane-binding domains. Bioinformatics 28, I431–I437. Kanter, J.L., Narayana, S., Ho, P.P., Catz, I., Warren, K.G., Sobel, R.A., Steinman, L., Robinson, W.H., 2006. Lipid microarrays identify key mediators of autoimmune brain inflammation. Nat. Med. 12, 138–143. Klarlund, J.K., Guilherme, A., Holik, J.J., Virbasius, J.V., Chawla, A., Czech, M.P., 1997. Signaling by phosphoinositide-3,4,5-trisphosphate through proteins containing pleckstrin and Sec7 homology domains. Science 275, 1927–1930. Knodler, A., Mayinger, P., 2005. Analysis of phosphoinositide-binding proteins using liposomes as an affinity matrix. Biotechniques 38, 858–862. Kooijman, E.E., King, K.E., Gangoda, M., Gericke, A., 2009. Ionization properties of phosphatidylinositol polyphosphates in mixed model membranes. Biochemistry 48, 9360–9371. Krugmann, S., Anderson, K.E., Ridley, S.H., Risso, N., McGregor, A., Coadwell, J., Davidson, K., Eguinoa, A., Ellson, C.D., Lipp, P., Manifava, M., Ktistakis, N., Painter, G., Thuring, J.W., Cooper, M.A., Lim, Z.Y., Holmes, A.B., Dove, S.K., Michell, R.H., Grewal, A., Nazarian, A., Erdjument-Bromage, H., Tempst, P., Stephens, L.R., Hawkins, P.T., 2002. Identification of ARAP3, a novel PI3K effector regulating both Arf and Rho GTPases, by selective capture on phosphoinositide affinity matrices. Mol. Cell 9, 95–108. Lemmon, M.A., 2003. Phosphoinositide recognition domains. Traffic 4, 201–213. Lemmon, M.A., 2007. Pleckstrin homology (PH) domains and phosphoinositides. In: Wakelam, M.J.O. (Ed.), Cell Biology of Inositol Lipids and Phosphates. Portland Press Ltd., London, pp. 81–93. Lemmon, M.A., 2008. Membrane recognition by phospholipid-binding domains. Nat. Rev. Mol. Cell Biol. 9, 99–111. Lemmon, M.A., Ferguson, K.M., 2000. Signal-dependent membrane targeting by pleckstrin homology (PH) domains. Biochem. J. 350, 1–18. Lim, Z.Y., Thuring, J.W., Holmes, A.B., Manifava, M., Ktistakis, N.T., 2002. Synthesis and biological evaluation of a PtdIns(4,5)P-2 and a phosphatidic acid affinity matrix. J. Chem. Soc. Perkin Trans. 1, 1067–1075. Lin, H.H., Han, L.Y., Zhang, H.L., Zheng, C.J., Xie, B., Chen, Y.Z., 2006. Prediction of the functional class of lipid binding proteins from sequence-derived properties irrespective of sequence similarity. J. Lipid Res. 47, 824–831. Losey, E.A., Smith, M.D., Meng, M., Best, M.D., 2009. Microplate-based analysis of protein-membrane interactions via immobilization of whole liposomes containing a biotinylated anchor. Bioconjugate Chem. 20, 376–383. Lu, K.Y., Tao, S.C., Yang, T.C., Ho, Y.H., Lee, C.H., Lin, C.C., Juan, H.F., Huang, H.C., Yang, C.Y., Chen, M.S., Lin, Y.Y., Lu, J.Y., Zhu, H., Chen, C.S., 2012. Profiling lipid–protein interactions using nonquenched fluorescent liposomal nanovesicles and proteome microarrays. Mol. Cell. Proteom. 11, 1177–1190. McLaughlin, S., Wang, J.Y., Gambhir, A., Murray, D., 2002. PIP2 and proteins: interactions, organization, and information flow. Annu. Rev. Biophys. Biomol. Struct. 31, 151–175. Mirkovic, N., Li, Z.H., Parnassa, A., Murray, D., 2007. Strategies for high-throughput comparative modeling: applications to leverage analysis in structural genomics and protein family organization. Proteins 66, 766–777. Narayan, K., Lemmon, M.A., 2006. Determining selectivity of phosphoinositidebinding domains. Methods 39, 122–133. Painter, G.F., Thuring, J.W., Lim, Z.Y., Holmes, A.B., Hawkins, P.T., Stephens, L.R., 2001. Synthesis and biological evaluation of a PtdIns(3,4,5)P-3 affinity matrix. Chem. Commun. 37, 645–646.




Park, W.S., Do Heo, W., Whalen, J.H., O’Rourke, N.A., Bryan, H.M., Meyer, T., Teruel, M.N., 2008. Comprehensive identification of PIP3-regulated PH domains from C. elegans to H. sapiens by model prediction and live imaging. Mol. Cell 30, 381–392. Pasquali, C., Bertschy-Meier, D., Chabert, C., Curchod, M.L., Arod, C., Booth, R., Mechtler, K., Vilbois, F., Xenarios, I., Ferguson, C.G., Prestwich, G.D., Camps, M., Rommel, C., 2007. A chemical proteomics approach to phosphatidylinositol 3kinase signaling in macrophages. Mol. Cell. Proteom. 6, 1829–1841. Pendaries, C., Tronchere, H., Plantavid, M., Payrastre, B., 2003. Phosphoinositide signaling disorders in human diseases. FEBS Lett. 546, 25–31. Rao, V.R., Corradetti, M.N., Chen, J.A., Peng, J.R., Yuan, J.Y., Prestwich, G.D., Brugge, J.S., 1999. Expression cloning of protein targets for 3-phosphorylated phosphoinositides. J. Biol. Chem. 274, 37893–37900. Reiser, G., Kunzelmann, U., Hulser, E., Stricker, R., Hoppe, J., Lottspeich, F., Kalbacher, H., 1995. Peptide-specific antibodies indicate species heterogeneity of a 42 kDa high-affinity inositol 1,3,4,5-tetrakisphosphate receptor protein from brain. Biochem. Biophys. Res. Commun. 214, 20–27. Rowland, M.M., Bostic, H.E., Gong, D., Speers, A.E., Lucas, N., Cho, W., Cravatt, B.F., Best, M.D., 2011. Phosphatidylinositol 3,4,5-trisphosphate activity probes for the labeling and proteomic characterization of protein binding partners. Biochemistry 50, 11143–11161. Rowland, M.M., Bostic, H.E., Gong, D., Lucas, N., Cho, W., Best, M.D., 2012. Microarray analysis of Akt PH domain binding employing synthetic biotinylated analogs of all seven phosphoinositide headgroup isomers. Chem. Phys. Lipids 165, 207–215. Sakane, F., Imai, S., Kai, M., Yasuda, S., Kanoh, H., 2007. Diacylglycerol kinases: why so many of them? Biochim. Biophys. Acta 1771, 793–806. Scott, D.L., Diez, G., Goldmann, W.H., 2006. Protein–lipid interactions: correlation of a predictive algorithm for lipid-binding sites with three-dimensional structural data. Theor. Biol. Med. Mod. 3, 3. Scott, J.L., Musselman, C.A., Adu-Gyamfi, E., Kutateladze, T.G., Stahelin, R.V., 2012. Emerging methodologies to investigate lipid–protein interactions. Integr. Biol. 4, 247–258. Shinohara, M., Terada, Y., Iwamatsu, A., Shinohara, A., Mochizuki, N., Higuchi, M., Gotoh, Y., Ihara, S., Nagata, S., Itoh, H., Fukui, Y., Jessberger, R., 2002. SWAP-70 is a guanine-nucleotide-exchange factor that mediates signalling of membrane ruffling. Nature 416, 759–763. Shirai, T., Tanaka, K., Terada, Y., Sawada, T., Shirai, R., Hashimoto, Y., Nagata, S., Iwamatsu, A., Okawa, K., Li, S.W., Hattori, S., Mano, H., Fukui, Y., 1998. Specific detection of phosphatidylinositol 3,4,5-trisphosphate binding proteins by the PIP3 analogue beads: an application for rapid purification of the PIP3 binding proteins. Biochim. Biophys. Acta 1402, 292–302. Silkov, A., Yoon, Y., Lee, H., Gokhale, N., Adu-Gyamfi, E., Stahelin, R.V., Cho, W., Murray, D., 2011. Genome-wide structural analysis reveals novel membrane binding properties of AP180 N-terminal homology (ANTH) domains. J. Biol. Chem. 286, 34155–34163. Smith, M.D., Best, M.D., 2011. Characterization of protein-membrane binding interactions via a microplate assay employing whole liposome immobilization. In: Mark, S.S. (Ed.), Bioconjugation Protocols: Strategies and Methods. , 2nd ed. Humana Press Inc., Totowa, pp. 477–489. Speers, A.E., Cravatt, B.F., 2004. Profiling enzyme activities in vivo using click chemistry methods. Chem. Biol. 11, 535–546. Sprong, H., van der Sluijs, P., van Meer, G., 2001. How proteins move lipids and lipids move proteins. Nat. Rev. Mol. Cell Biol. 2, 698-698.

Streb, H., Irvine, R.F., Berridge, M.J., Schulz, I., 1983. Release of Ca2+ from a nonmitochondrial intracellular store in pancreatic acinar-cells by inositol-1,4,5triphosphate. Nature 306, 67–69. Stricker, R., Kalbacher, H., Lottspeich, F., Reiser, G., 1995. Characterization of a highaffinity Ins-P-4 (inositol 1,3,4,5-tetrakisphosphate) receptor from brain by an antipeptide antiserum. FEBS Lett. 370, 236–240. Stricker, R., Hulser, E., Fischer, J., Jarchau, T., Walter, U., Lottspeich, F., Reiser, G., 1997. cDNA cloning of porcine p42(IP4) a membrane-associated and cytosolic 42 kDa inositol(1,3,4,5)tetrakisphosphate receptor from pig brain with similarly high affinity for phosphatidylinositol (3,4,5)P-3. FEBS Lett. 405, 229–236. Stricker, R., Vandekerckhove, J., Krishna, M.U., Falck, J.R., Hanck, T., Reiser, G., 2003. Oligomerization controls in tissue-specific manner ligand binding of native, affinity-purified p42(IP4)/centaurin alpha 1 and cytohesins – proteins with high affinity for the messengers d-inositol 1,3,4,5tetrakisphosphate/phosphatidylinositol 3,4,5-trisphosphate. Biochim. Biophys. Acta 1651, 102–115. Tanaka, K., ImajohOhmi, S., Sawada, T., Shirai, R., Hashimoto, Y., Iwasaki, S., Kaibuchi, K., Kanaho, Y., Shirai, T., Terada, Y., Kimura, K., Nagata, S., Fukui, Y., 1997. A target of phosphatidylinositol 3,4,5-trisphosphate with a zinc finger motif similar to that of the ADP-ribosylation-factor GTPase-activating protein and two pleckstrin homology domains. Eur. J. Biochem. 245, 512–519. Tempel, M., Goldmann, W.H., Isenberg, G., Sackmann, E., 1995. Interaction of the 47kDa talin fragment and the 32-kDa vinculin fragment with acidic phospholipids: a computer analysis. Biophys. J. 69, 228–241. Theibert, A.B., Estevez, V.A., Ferris, C.D., Danoff, S.K., Barrow, R.K., Prestwich, G.D., Snyder, S.H., 1991. Inositol 1,3,4,5-tetrakisphosphate and inositol hexakisphosphate receptor proteins – isolation and characterization from rat-brain. Proc. Natl. Acad. Sci. U.S.A. 88, 3165–3169. Theibert, A.B., Estevez, V.A., Mourey, R.J., Marecek, J.F., Barrow, R.K., Prestwich, G.D., Snyder, S.H., 1992. Photoaffinity-labeling and characterization of isolated insositol 1,3,4,5-tetrakisphosphate-binding and insotiol hexakisphosphate-binding proteins. J. Biol. Chem. 267, 9071–9079. Tully, S.E., Cravatt, B.F., 2010. Activity-based probes that target functional subclasses of phospholipases in proteomes. J. Am. Chem. Soc. 132, 3264–3265. Vicinanza, M., D’Angelo, G., Di Campli, A., De Matteis, M.A., 2008. Phosphoinositides as regulators of membrane trafficking in health and disease. Cell. Mol. Life Sci. 65, 2833–2841. Wang, D.S., Ching, T.T., St Pyrek, J., Chen, C.S., 2000. Biotinylated phosphatidylinositol 3,4,5-trisphosphate as affinity ligand. Anal. Biochem. 280, 301–307. Wymann, M.P., Schneiter, R., 2008. Lipid signalling in disease. Nat. Rev. Mol. Cell Biol. 9, 162–176. Xiong, W.J., Guo, Y.Z., Li, M.L., 2010. Prediction of lipid-binding sites based on support vector machine and position specific scoring matrix. Protein J. 29, 427–431. Zhang, D.P., Aravind, L., 2010. Identification of novel families and classification of the C2 domain superfamily elucidate the origin and evolution of membrane targeting activities in eukaryotes. Gene 469, 18–30. Zhang, P.B., Wang, Y., Sesaki, H., Iijima, M., 2010. Proteomic identification of phosphatidylinositol (3,4,5) triphosphate-binding proteins in Dictyostelium discoideum. Proc. Natl. Acad. Sci. U.S.A. 107, 11829–11834. Zhu, H., Bilgin, M., Bangham, R., Hall, D., Casamayor, A., Bertone, P., Lan, N., Jansen, R., Bidlingmaier, S., Houfek, T., Mitchell, T., Miller, P., Dean, R.A., Gerstein, M., Snyder, M., 2001. Global analysis of protein activities using proteome chips. Science 293, 2101–2105.


Global market for dairy proteins.

Elucidation of the primary structures of proteins by mass spectrometry.

[Chemical biology for pharmaceutical science--integrative approaches for elucidation of biological phenomena].

High-Throughput Proteomic Approaches to the Elucidation of Potential Biomarkers of Chronic Allograft Injury (CAI).

In vitro approaches to the elucidation of mechanisms of chemical teratogenesis.

Computational approaches for rational design of proteins with novel functionalities.

Seeking local approaches to global disease threats.

Worldly approaches to global health: 1851 to the present.

Elucidation of the anti-autophagy mechanism of the Legionella effector RavZ using semisynthetic LC3 proteins.

Antisense approaches to the function of glial cell proteins.

Integrated Proteomic and Transcriptomic-Based Approaches to Identifying Signature Biomarkers and Pathways for Elucidation of Daoy and UW228 Subtypes.

Novel ncRNAs transcribed by Pol III and elucidation of their functional relevance by biophysical approaches.

In silico approaches for the identification of virulence candidates amongst hypothetical proteins of Mycoplasma pneumoniae 309.

Bacteriophages and phage-derived proteins--application approaches.

Elucidation of the Burkholderia cenocepacia hopanoid biosynthesis pathway uncovers functions for conserved proteins in hopanoid-producing bacteria.

Proteome analysis for the global proteins in the jejunum tissues of enterotoxigenic Escherichia coli -infected piglets.

Analytical approaches for the characterization and identification of olive (Olea europaea) oil proteins.

Therapeutic approaches targeted at the amyloid proteins in Alzheimer's disease.

Novel approaches to address challenges in global drug development.

Systems approaches: a global and historical perspective on integrative medicine.

Comparative and Quantitative Global Proteomics Approaches: An Overview.

Elucidation of the Teixobactin Pharmacophore.

Proteomics analysis and its role in elucidation of functionally significant proteins in Mycoplasma bovis.

Ubiquitylation as a Rheostat for TCR Signaling: From Targeted Approaches Toward Global Profiling.