PresentationCell Biology

Proteomic Analysis of Integrin Adhesion Complexes

See allHide authors and affiliations

Science Signaling  05 Apr 2011:
Vol. 4, Issue 167, pp. pt2
DOI: 10.1126/scisignal.2001827
A presentation from the 6th British Society for Proteome Research (BSPR)–European Bioinformatics Institute (EBI) Meeting “Multiscale Proteomics: From Cells to Organisms” at the Wellcome Trust Conference Centre, Cambridge, UK, 14 to 16 July 2009. The Presentation also complements the Science Signaling Research Article by Humphries et al. published 8 September 2009.


Integrin receptors regulate cell fate by coupling the binding of extracellular adhesion proteins to the assembly of intracellular cytoskeletal and signaling complexes. A detailed, integrative view of adhesion complexes will provide insight into the molecular mechanisms that control cell morphology, survival, movement, and differentiation. To date, membrane receptor–associated signaling complexes have been refractory to proteomic analysis because of their inherent lability and inaccessibility. We developed a methodology to isolate ligand-induced integrin adhesion complexes, and we used this technique to analyze the composition of complexes associated with multiple receptor–ligand pairs and define core and receptor-specific subnetworks. In particular, we identified regulator of chromosome condensation–2 (RCC2) as a component of fibronectin-activated signaling pathways that regulate directional cell movement. The development of this proteomics pipeline provides the means to investigate the molecular composition and function of various adhesion complexes.

Presentation Notes

Slide 1: Science Signaling logo

The slideshow and notes for this presentation are provided by Science Signaling (

Slide 2: Proteomic analysis of integrin adhesion complexes

This talk presents the development of a proteomic pipeline for the analysis of integrin adhesion complexes.

Slide 3: Integrins as adhesion receptors

Integrins are cell-surface adhesion receptors that mediate cell–cell and cell–extracellular matrix (ECM) interactions. These interactions are fundamental to multicellularity, and they control cell morphology and movement, differentiation, and tissue integrity. Consequently, dysregulation of these integrin-mediated controls can result in various disease processes, such as inflammation, atherosclerosis, and tumorigenesis. Sites of integrin–ligand engagement anchor the cell to the ECM or to adjacent cells through the recruitment of intracellular, membrane-associated signaling complexes. These complexes can contain vinculin, shown here in green in A375-SM human melanoma cells spreading on the ECM molecule fibronectin (FN), and mechanical connections to the actin cytoskeleton, shown here in red. This structural and functional integration of the cell with its environment is of central importance for the control of cell behavior and tissue organization. The mechanisms that regulate this integration, however, are not yet understood.

Slide 4: Integrins as bidirectional signalers

All integrins are heterodimeric molecules comprising noncovalently linked α and β subunits (blue and red, respectively). Each subunit contains a large extracellular domain, a transmembrane domain, and typically a small cytoplasmic domain. The extracellular domains cooperatively bind ligand, as shown in the crystal structure of the extracellular segment of αVβ3 integrin complexed with an Arg-Gly-Asp (RGD) ligand (top right) (1). The RGD peptide (yellow) binds to the interface between the α-subunit β-propeller domain (blue) and the β-subunit von Willebrand A (βA) domain (red). Amino acid residues in the intracellular tails of integrins are implicated in interactions at the α–β subunit interface and provide binding sites for cytoskeletal proteins such as talin, as shown in the nuclear magnetic resonance structures of the cytoplasmic domains of αIIbβ3 integrin (bottom right movie) (2) and the cytoplasmic domain of β3 (red ribbon in bottom right) complexed with the talin F3 domain (surface–electrostatic potential model in bottom right) (3). The integrin-binding F3 domain is located in the N-terminal head of talin, whereas the C-terminal flexible rod domain of talin contains a site that links it to F-actin, multiple sites for binding vinculin (which can, in turn, bind to F-actin), and a second integrin-binding site. Integrin–talin binding is key to regulating the affinity of integrin for ligand (3, 4). Talin–actin binding links the ECM and integrin receptors with the cytoskeleton and affects cytoskeletal organization, generation of traction forces on the ECM through actomyosin contraction, and cell migration (5, 6).

The crystal structure of αVβ3 revealed a striking ~135° bend at the flexible “knees” of the extracellular “leg” domains (top right movie) (1, 7). The bent form of the integrin is thought to represent the conformation with low affinity for ligand. Allosteric changes, including unbending of the receptor, swinging of the β-subunit hybrid domain away from the α-subunit, and α-helical movements in the βA domain, lead to the acquisition of a high-affinity conformation and result from changes in metal coordination in the integrin “head” domain that occur upon ligand binding. In addition to causing receptor shape changes, the binding of multivalent ligands can induce integrin clustering and the accumulation of membrane-associated signaling complexes. The conformational changes to integrin can be induced from the inside of the cell, which increases the affinity of integrins for ligand (“inside-out” signaling), and from the outside, which elicits intracellular signals (“outside-in” signaling). The coupling of inside-out and outside-in signaling provides a mechanism for bidirectional signaling by integrins across the plasma membrane (4). However, the extent to which different ligands stabilize different integrin conformers to transduce specific signals is unknown.

Slide 5: Challenges in the field

There are 24 αβ heterodimers in the integrin family, and most of these receptors recognize a wide variety of ligands. Furthermore, many ECM macromolecules and counterreceptors bind to multiple integrins. This leads to a large number of potential integrin–ligand combinations that can occur in different cell types under different physiological conditions (left) (8). RGD-binding integrins, such as α5β1, recognize active sites containing the RGD motif in ligands such as FN. Leu-Asp-Val (LDV)–binding integrins, which include α4β1, recognize LDV-related sequences in ligands such as vascular cell adhesion molecule–1 (VCAM-1). The Asp40 residue in the LDV-related Ile-Asp-Ser-Pro sequence of VCAM-1 is essential for α4β1 binding.

No less complex is the molecular landscape of adhesions within the cell. Integrin–ligand binding induces the recruitment of supramolecular protein structures to the plasma membrane. More than 150 molecules have been reported to be either stably or transiently associated with sites of integrin-mediated cell adhesion, although the cell context and integrin specificity of this hypothetical “interactome” have not been defined (right) (9). Despite this complexity, key functional groups of proteins appear to be important for adhesion signaling. Adhesion receptors, actin regulators, and structural and adaptor proteins form the major physical structure of cell adhesions, linking the plasma membrane to the actin cytoskeleton (9). The majority of other proteins reported to be present in cell adhesions are enzymes, including tyrosine and serine-threonine phosphatases and kinases, guanosine triphosphatases (GTPases), and associated regulators, which together serve to control adhesion assembly, turnover, and signaling.

A key question in the biochemistry of cell adhesion is how different ligands can bind to certain integrins to elicit specific cellular responses. The signaling complexes recruited to ligand-engaged integrins control the cellular responses, so analysis of the composition and stoichiometry of specific adhesion complexes is essential for an understanding of integrin-mediated adhesion processes.

Slide 6: Aims

To understand how heterodimer-specific integrin-mediated signaling events regulate cell adhesion, we set out to isolate and systematically analyze ligand-induced integrin adhesion complexes. Mass spectrometry (MS) provides an attractive methodology for the investigation of complex protein mixtures isolated from cells because it can provide sensitive, rapid, non–candidate-based analyses and is readily automated (10). Long-standing limitations to proteomic analysis of cell adhesion signaling include the inherent instability and inaccessibility of receptor-associated complexes. That is, signaling complexes are usually maintained by low-affinity protein–protein interactions to enable efficient modulation of signaling, which makes them dynamic and labile structures. Signaling complexes linked to transmembrane receptors are often difficult to isolate because of incomplete solubility under standard detergent extraction conditions, and insoluble components are discarded in most affinity purification protocols. Furthermore, integrin activities are ligand-dependent, allostery-dependent, and mechanosensitive, thus their conformation and plasma-membrane environment probably determine the exact composition of recruited protein complexes. Thus, classical immunoprecipitation methods were not amenable to global analysis of specific integrin-associated complexes.

Our initial studies focused on two structurally related integrins, α4β1 and α5β1, and their ligands, VCAM-1 and FN, respectively. These integrins play distinct roles in cell behavior, with α4β1 integrin influencing rolling adhesion and cell motility of leukocytes and α5β1 integrin mediating mechanotransduction (11, 12).

Slide 7: Affinity purification assay

We chose K562 erythroleukemia cells for our work because this cell line has served as a prototype for many analyses of integrin-mediated adhesion and signaling. In particular, the integrin complement of K562 cells is dominated by α5β1, and therefore it is possible to pinpoint adhesion and signaling events to this specific integrin heterodimer. Stable transfection of a transgene encoding α4 integrin—yielding α5+4+ K562 cells—results in a similar level of cell attachment to VCAM-1 compared with that mediated by endogenous α5β1 binding to FN (13). Likewise, α5+4+ K562 cells demonstrate similar activation-state modulation of α4β1 to that observed in cells that endogenously express the receptor (13). Thus, K562 cells appear suitable for analyses of integrin adhesion and signaling events.

We overcame the challenges associated with the isolation of integrin adhesion complexes by developing a ligand affinity purification approach (14). Ligand-coated beads have been used previously to capture Fc receptor–linked complexes (15), and beads coated with integrin ligands have been used to mimic cell–substrate adhesion and to induce adhesion complex formation (1618). Our work aimed to provide a global analysis of ligand-induced adhesion complexes, and we adapted a microbead-based method initially described by Plopper and Ingber (16). Paramagnetic beads coated with integrin ligand (top left) were incubated with α5+4+ K562 cells to induce the formation of adhesion complexes in living cells (top right). Bead-bound cells were gently and rapidly isolated on a magnet, and a combination of sonication and non-ionic detergent extraction was used to purify integrin-associated protein complexes, which were then eluted from the beads with detergent and reducing agent. The isolated complexes were then subjected to biochemical or mass spectrometric analysis.

Slide 8: Specific protein recruitment

Because of the lability of integrin receptor–associated complexes, it was necessary to stabilize the complexes by using a chemical crosslinker before cell lysis. Use of the membrane-permeable, thiol-cleavable crosslinker dimethyl-3,3′-dithiobispropionimidate permitted the enrichment of integrin subunits and adhesion complex components, such as talin and paxillin, on integrin ligand–coated beads as compared with control beads, which were coated with an antibody that recognizes the transferrin receptor (TfR; left) or with a VCAM-1 Asp40→Ala (D40A) mutant that does not bind to integrins (control; right). These data demonstrate the specific and efficient isolation of stabilized α5β1–FN and α4β1–VCAM-1 adhesion complexes through use of our affinity purification approach (14).

Slide 9: Proteomic analysis workflow

We used MS-based proteomics to catalog the components of the isolated integrin-associated complexes. Purified proteins were resolved by means of sodium dodecylsulfate–polyacrylamide gel electrophoresis and visualized through Coomassie staining. Gel lanes were cut into 30 slices, and gel pieces were subjected to in-gel tryptic digestion (19). Extracted peptides were analyzed with liquid chromatography–tandem mass spectrometry (LC-MS/MS). LC-MS/MS data files for each slice of the gel lane were merged and used to search a human protein sequence database by using the Mascot search engine (20). To validate the proteomics data sets, multiple search engines and rigorous statistical algorithms at both the peptide and protein levels were used with Scaffold software (2123). Relative protein abundance was determined by using spectral counting, a method for protein quantification in which the number of tandem mass spectra detected and identified for a given protein is summed (2428).

Slide 10: Hierarchical clustering analysis

To interrogate the multiple proteomics data sets, hierarchical clustering was performed on the quantitative data in which spectral counts were normalized to the total number of spectra identified in the sample. Heat maps were used to visualize the clustering outputs; each block of color represents an individual protein, with red indicating high spectral count, blue indicating low spectral count, and black indicating that the protein was not detected in the sample. Clustering using a Pearson correlation–based distance metric identified clusters of proteins enriched with one or both integrin ligands, or enriched in all samples, including the control. Importantly, α5 integrin and α4 integrin were only detected in FN and VCAM-1 samples, respectively (left, arrowheads). In addition to highlighting proteins that were specifically enriched in only one integrin–ligand complex (left), hierarchical clustering identified proteins that were found in both complexes (such as β1 integrin and the well-studied adhesion proteins kindlin-3, talin-1, and zyxin) (right, arrowheads), which represent core components shared by the two complexes. Furthermore, hierarchical clustering dissected subclusters of proteins that were more highly enriched with one ligand as compared with the other. For example, filamin, Rap1b, and vinculin were found in greater abundance in FN samples than in VCAM-1 samples (top right, arrowheads).

Slide 11: Ligand-induced interaction networks

We used interaction network analysis to examine the molecular landscape of the isolated adhesion complexes in the context of currently known protein–protein interactions. Proteins specifically enriched in integrin–ligand complexes—those enriched at least two-fold compared with the control—were mapped onto a human interactome consisting of the union of four molecular interaction databases and three published human protein–protein interaction data sets (29) and a set of integrin adhesion complex components curated from published studies (9). We mapped 386 (95%) of the 406 proteins specifically enriched in α5β1–FN complexes and 181 (98%) of the 185 proteins specifically enriched in α4β1–VCAM-1 complexes onto this interactome. To assess the relative abundance changes of proteins recruited to α5β1–FN and α4β1–VCAM-1 complexes, spectral count data were normalized to the spectral count of β1 integrin in each data set, and relative enrichment was mapped onto each protein identification as a node color in the network, with red indicating enrichment in FN samples and blue indicating enrichment in VCAM-1 samples.

Examination of proteins identified within two path lengths (two protein–protein interactions) of β1 integrin (the center node in each interaction network) from the α5β1–FN (left) and α4β1–VCAM-1 (right) complexes revealed that the FN-induced network was more expansive than the VCAM-1–induced network. Furthermore, the VCAM-1–induced network mainly contained proteins that were detected in both integrin–ligand complexes, as indicated by black outlining of the nodes, whereas the FN-induced network contained a large proportion of proteins that were specifically enriched in α5β1–FN complexes, as indicated by the absence of black outlines around these nodes. These findings are consistent with a role for α4β1 in motile cell behavior, such as the extravasation and migration of leukocytes at sites of inflammation, which have previously been postulated to form weaker cytoskeletal interactions because of the formation of less extensive adhesion complexes (11, 30). Conversely, α5β1–FN adhesions often occur where there may be a requirement for more robust cell–ECM interactions or ECM organization, such as in endothelial cells involved in vascular development (31, 32). The different types of heterodimer-specific β1 integrin complexes observed here revealed a surprising difference in scale and composition and suggest that such global analyses of adhesion complexes can provide valuable insight into the nature of cell adhesion.

Slide 12: Links to GTPase signaling modules

Interaction network analysis revealed numerous highly connected signaling subnetworks, which highlighted proteins in key positions of intersection. Regulator of chromosome condensation–2 (RCC2), a molecule not previously implicated in integrin signaling, was specifically detected in FN samples (red node in black ring). To assess its potential association with integrin, we analyzed the molecular networks around RCC2. Strikingly, RCC2 was located in a subnetwork comprising the direct interactors (1-hop neighborhoods) of β1 integrin (green dashed line) and the small GTPases Rac1 (blue dashed line) and Arf6 (red dashed line) in the α5β1–FN complex. This observation was intriguing because it provided a potential link between adhesion signaling molecules and Rac1 and Arf6, which control cell membrane protrusion and membrane trafficking, respectively (33). The detection of RCC2 with LC-MS/MS in FN but not VCAM-1 or control samples was verified through immunoblotting (top right).

Slide 13: RCC2 regulates Rac1 and Arf6 activity and FN-dependent adhesion complex formation

To test the role of RCC2 in regulating Rac1 and Arf6 activities, we used RNA interference to reduce the expression of RCC2. Cells spreading on FN exhibited a wave of Rac1 and Arf6 activation over 90 min. Intriguingly, knockdown (KD) of RCC2 expression to less than 20% of endogenous levels (top left) enhanced both Rac1 activity (top middle) and Arf6 activity (top right) during cell spreading on FN. Although RCC2 has previously been shown to bind the nucleotide-free form of Rac1 and has been suggested to be a guanine nucleotide exchange factor and, thus, an activator of GTPase activity (34), our data suggest that RCC2 limits Rac1 and Arf6 activation during cell spreading. Depletion of RCC2 also increased the total area of cellular adhesions in cells spread on FN, as shown by vinculin staining (bottom). Augmented adhesion was most marked at 30 to 90 min of cell spreading (bottom right), which coincided with peaks of Rac1 and Arf6 activity, highlighting the role of RCC2 in coordinating two adhesion signaling pathways that separately regulate membrane protrusion (Rac1) and membrane delivery (Arf6).

Slide 14: RCC2 modulates directional cell migration

We hypothesized that a molecule that regulates both Rac1 and Arf6 would play a role in cell motility. To assess the functional consequences of dysregulated GTPase signaling upon RCC2 knockdown, cell migration on preformed cell-derived matrices was examined. Cells migrate with greater persistence on a cell-derived matrix than on a two-dimensional FN substrate because matrix-fiber engagement localizes and restricts GTPase activity, which controls the formation of off-axial lamellae (35). Thus, cell-derived matrices were used here to provide a more physiologically relevant substrate for measuring cell migration. Migration paths of cells were tracked after live imaging, and persistence was determined by dividing the linear displacement (Euclidean distance) of a cell after 12 hours by the total distance moved (accumulated distance). Persistence values range from 0 to 1, with cells plated on FN-coated plastic (top right, gray bars) migrating randomly and showing a persistence value of ~0.25. Depletion of RCC2 reduced nonrandom, persistent cell migration along these FN fibers by 50%, but the total distance moved by the cells was not affected (right). These data demonstrate that RCC2 is an ECM-dependent regulator of directional migration, which, through precise restriction of GTPases, allows cells to respond to the spatial organization of their environment.

Slide 15: Summary

We have developed a method for the isolation and proteomic analysis of integrin adhesion complexes. We have cataloged the first experimentally defined proteomes of two integrin–ligand complexes (α5β1–FN and α4β1–VCAM-1) and observed a surprising difference in their scale and composition. Furthermore, we have identified RCC2 as a previously unknown dual regulator of Rac1 and Arf6 activation and a control node of directional cell migration associated with the α5β1–FN complex.

There is currently a paucity of data on transmembrane receptor proteomes, largely because these complexes are labile and physically inaccessible. This workflow now permits the proteomic analysis of functionally relevant protein complexes that mediate adhesion signaling and have essential roles in a diverse range of biological functions. Moreover, this methodology could also be applied to other fields, such as growth factor, cytokine, and channel signaling, thus providing a means by which to attain a systems-level understanding of cell surface receptor signaling.

Slide 16: Acknowledgments

This work was supported by the Wellcome Trust (grants 045225 and 074941 to M.J.H.) and by a Biotechnology and Biological Sciences Research Council (BBSRC) Collaborative Awards in Science and Engineering Ph.D. studentship, sponsored by GlaxoSmithKline (A.B.). The Biomolecular Analysis Facility mass spectrometer and Bioimaging Facility microscopes used in this work were purchased with grants from the BBSRC, Wellcome Trust, and the University of Manchester Strategic Fund.

Editor’s Note: This contribution is not intended to be equivalent to an original research paper. Note, in particular, that the text and associated slides have not been peer-reviewed.


View Abstract

Navigate This Article