Special Reviews

Forces and Bond Dynamics in Cell Adhesion

See allHide authors and affiliations

Science  25 May 2007:
Vol. 316, Issue 5828, pp. 1148-1153
DOI: 10.1126/science.1137592


Adhesion of a biological cell to another cell or the extracellular matrix involves complex couplings between cell biochemistry, structural mechanics, and surface bonding. The interactions are dynamic and act through association and dissociation of bonds between very large molecules at rates that change considerably under stress. Combining molecular cell biology with single-molecule force spectroscopy provides a powerful tool for exploring the complexity of cell adhesion, that is, how cell signaling processes strengthen adhesion bonds and how forces applied to cell-surface bonds act on intracellular sites to catalyze chemical processes or switch molecular interactions on and off. Probing adhesion receptors on strategically engineered cells with force during functional stimulation can reveal key nodes of communication between the mechanical and chemical circuitry of a cell.

The physical role of a cell adhesion bond is to hold a cell to other cells or to tissue substrata while supporting the forces involved in cell function. Complicating this task, a single adhesion bond effectively resists force only for time periods less than that needed for its spontaneous dissociation under thermal activation. Thus, the diversity in the mechanochemistry of adhesion bonds reflects how mechanical force applied to a bond between a pair of interacting molecules alters activation energy barriers along kinetic pathways, or switches pathways, that lead to dissociation. Viewed ideally as illustrated by Fig. 1, applying adhesion stress through the local material structure to a bond is conceptually like pulling on the chemical interaction with a mechanical spring that mimics the compliance properties of structures attached to the binding site. Stretching this equivalent spring produces a force that lowers the chemical activation barrier to increase the frequency of bond dissociation while, at the same time, the spring potential defines an “energy well” that captures the dissociated states and regulates the likelihood of rebinding. Focusing our discussion on adhesive interactions in soft tissues and organs of eukaryotic cell systems, the important insights derived from the simple view in Fig. 1 are that bond survival effectively decreases exponentially with the level of pulling force and that deformations of soft structures even under small forces suppress rebinding after dissociation.

Fig. 1.

Conceptual view of force propagation to a bond and its impact on the chemical energy landscape governing bond kinetics. (A) Pulling on structural connections to a molecular bond creates a mechanical “springlike” potential (dashed blue curve) that alters the chemical energy of interaction or “landscape” (solid and dashed red curves) along the reaction coordinate defined by the pulling direction (37). The slope of the spring potential at the origin of interaction is the pulling force f, the product of the effective spring constant κs of the structural linkages with the increase in their separation xseparation under pulling. Of greatest impact on bond survival, the spring potential reduces the height of the activation energy barrier governing the off-rate kinetics located at xβ by ∼–fxβ. Brought to our attention years ago by Bell (38), the change in Arrhenius factor predicts a large exponential-like reduction in bond survival time, toff(f) ≈ toff° exp(–f/fβ), relative to its apparent unstressed lifetime toff° (39). The response scale for this accelerated dissociation is the level of force, fβ = kBT/xβ, that drops the barrier by one unit of thermal energy kBT (40). The reduction in bond lifetime with increase in force is illustrated in (B) by the behavior expected in a force-clamp test at different forces. When the anchored molecules unbind, the dissociated states are confined near the displaced minimum of the spring potential, as indicated in (C), from which strong thermal excitations can cause them to rebind. However, there is little likelihood of rebinding when the depth of the spring potential in (C) exceeds the binding energy Eo for forces ≥ (2κsEo)1/2. Consequently, molecules anchored by soft structures (small κs) rebind very infrequently even under low stress, which suggests why bond recruitment in cell adhesion usually involves bringing the constituents together by large-scale cytoskeletal movements as when forming the immunological synapse in T lymphocyte adhesion (41).

With a precipitous reduction in lifetime under stress and little likelihood of rebinding, bonds in cell adhesion are therefore being continually created, loaded over some period of time, then failing. Even for cells in tissues seemingly under static stress, forces build up transiently on the individual bonds that connect cells, albeit very slowly (maybe at miniscule rates of only ∼1 pN/s) until the bonds break and shift their loads to other bonds. The balance of stress is achieved through recruitment of new bonds driven by cytoskeletal movements, resulting in a “bubbly” dynamic process of bond loading, failure, and formation. By comparison, at the other extreme, the initial attachment of an immune system cell in the vasculature can apply force to bonds at an incredibly fast rate (for instance, ∼104 pN/s), which is then followed by quick release of the cell or rapid activation of new adhesive components to arrest the cell and enable its emigration into the surrounding tissue. Thus, if we ignore many complex features of the force responses of adhesive and structural bonds, the important mechanical property that characterizes these bonds is not static strength but, rather, dynamic strength. As illustrated in Fig. 2, mechanical strength emerges when the stress rate is sufficient to make the bond fail in less time than needed for its spontaneous dissociation and then rises sluggishly by an increment of force for each order of magnitude increase in the stress rate (that is, proportional to a logarithm of stress rate). Adhesion bonds that form “transient high-strength” connections are those that require fast loading to withstand force, whereas bonds that establish “persistent” connections are those that hold firm under conditions of slow loading as well as fast. Unexpectedly, these differences in stress-rate sensitivity cross structural family lines.

Fig. 2.

Dynamic strength of a bond: the “nanorheology” of a molecular interaction. The important dynamical corollary to the Bell exponential model (Fig. 1) for the off rate of bonds under stress is that the strength and lifetime of a bond become interrelated properties governed by the stress rate (42), rf = Δft, and thereby the speed vpull at which the molecular linking structures are separated, that is, rf = κsvpull. When loaded by an increasing force, the failure rate of an idealized bond grows exponentially with time, predicting that the bond will break “most often” at a force f*, increasing by 1 unit of the thermal-activation force fβ for each e-fold exp (1) ∼2.72 times increase in the loading rate. The increase in strength and decrease in survival expected for bonds at different force rates (called force ramps) are sketched in (A). Derived from tests at several force rates, the most frequent rupture forces for the idealized bond follow a straight line when plotted against logarithms of the force rates as sketched in (B), that is, f* = fβ loge(rftoff/fβ). Key to the dynamics, the kinetic scale for force rate, rf° = fβ/toff°, defines the loading speed above which the bond is driven “far from equilibrium,” dissociating faster than its apparent spontaneous off rate 1/toff°, and thus resists force. The most frequent “lifetime” t* of the bond is precisely its strength divided by the loading rate, t*= f*/rf. So, while bond strength grows sluggishly with increased loading speed, bond survival falls extremely rapidly, as illustrated in (C).

Because of the rough topography of a cell membrane, cell adhesion ligands and their receptors in multicellular eukaryotic systems are generally large (30 to 50 nm) multidomain structures that project out from the lipid bilayer surface to overcome steric interference and enable connections between cells [smaller (∼7 to 8 nm) signaling molecules often mediate adhesion as well]. Representing a very large number of specific interactions, cell adhesion receptors and ligands are designed for one or more dynamic functions and often work cooperatively to achieve functional outputs (1). For many of these interactions, the receptors can be grouped into major families such as the selectins (members of the C-type lectin domain family), the cadherins [considered an offshoot of the immunoglobulin (Ig) superfamily], and the integrins. Although not all-inclusive, these three families serve to illustrate the broad range in mechanical performance among adhesion bonds.

Selectins, which appear on cell surfaces as homodimers of long chains, provide prominent examples of receptors producing transient high-strength connections (Fig. 3A). Ca2+-dependent binding of their outermost C-type lectin domain to cell-surface sialomucin proteins initiates capture of fast-moving white blood cells at vessel walls (2). To perform this important function, selectin adhesion bonds require fast loading to resist pulling, and they become very strong when subjected to the extreme force rates (∼104 pN/s) often experienced when a circulating cell first adheres to endothelium (3). Moreover, selectins exhibit a special “catch bond” quality (4) characterized by quick release under slow-loading and low-force conditions (Fig. 3A). By comparison, cadherins appear on cell surfaces as multidomain single chains and interact homophilically when Ca2+ is present (5). Hence, cadherins bind other cadherins on opposing cells to produce trans-bonded connections and may also bind their adjacent neighbors to produce cis-bonded lateral connections. Appearing to be capable of binding through a progression of overlapping antiparallel arrangements (6), the force responses of trans-bonded cadherins suggest a hierarchy of mechanical functionality. On the one hand, cadherins form short-lived connections and behave as transient connectors when binding just their outer-tip domains, which may be important for the dynamics of recognition and patterning of cells in development. On the other hand, deep-trans bonding of all domains produces strong attachments even under very slow loading, as demonstrated by the force responses of attachments between full-length cadherins (Fig. 3B). With little sensitivity to stress rate, such “persistent” connections may enable formation of durable structures like desmosomes in lateral junctions of epithelial cells, although opposing models based on dense networks of tip interactions have also been proposed (7). Last, integrins mediate perhaps the most diverse range of adhesive interactions in eukaryote biology, exhibiting widely different levels of attachment strength and lifetime. Representing one extreme of their dynamical response, many integrin interactions are long-lived and provide the persistent strength needed, for example, to hold together tissues, to transmit force during muscle contraction, and to arrest circulating immune cells on activated endothelium and enable their migration (8). A prominent example of persistent strength and the insensitivity to stress rate is demonstrated by the force responses of attachments to the integrin αLβ2 (Fig. 3B). Yet, representative of the opposite extreme, other integrin interactions are short-lived and behave as transient connectors that require fast loading for strength, as when integrins at the leading edge of a spreading cell form new attachments to an extracellular matrix (9) or when integrins initiate capture of lymphocytes in the systemic circulation (10), mimicking the response of a selectin as demonstrated by force responses of attachments to the integrin α4β1 (Fig. 3A). Because of the diversity in integrin mechanical response and the important role of their cytoskeletal connections in adhesion, we will center the remainder of our discussion around integrin bonds.

Fig. 3.

Transient and persistent cell adhesion bonds. The distinction between these two types of adhesion bonds reflects major differences in their stress rate requirements rf° = fβ/toff°for onset of strength and their scales fβ for amplification of strength under increasing stress rate (Fig. 2). Consistent with the labels, the examples of transient connectors in (A) require fast loading for strength, whereas the examples of persistent connectors in (B) are strong even under slow loading. Also intriguing is that the transient connectors show much larger amplification of strength at high-loading speeds than persistent connectors, which suggests a fundamental feature of the way in which weak biomolecular bonds are chemically designed to achieve strength (43). Taken from in vitro tests of single recombinant receptor and ligand interactions when immobilized on a force probe and microsphere target (44), the responses for transient connectors in (A) are demonstrated by the dynamic strengths of P-selectin (PS) bonds to a reactive N-terminal segment of mucin PS glycoprotein ligand–1 (PSGL-1) (45) and by the dynamic strengths of integrin α4β1 bonds to a two-domain construct of VCAM-1 (46). Although similar to the β1-integrin interaction at high force rates, the strength of the PS interaction is switched on at a fast loading rate. Called a “catch bond” (4), this unusual response represents a mechanochemical switch triggered by force rate to turn off a fast dissociation pathway and lock in a slow dissociation pathway that resists force (45). Also taken from in vitro tests, the responses for “persistent connectors” in (B) are demonstrated by the dynamic strengths of homophilic full-length cadherin bonds (47, 48) and by the dynamic strengths of integrin αLβ2 bonds to ICAM-1 (46). Feedback from chemical pathways inside cells is known to reinforce cell adhesion mediated by integrin bonds. We found a significant parallel upward shift [*, blue-dashed linein(B)] in dynamic strengths of β2-integrin bonds when testing ICAM-1 bonds to the β2 integrin (LFA-1) in situ at the surface of cytokine-stimulated white blood cells; this suggests a range of affinity states for the β2 integrin and demonstrates the “inside-out” feedback at the single-molecule level. The units for force and force rate are pN and pN/s, respectively.

Integrin Bonds: The Archetype of Multifunctional Adhesive Design

Integrins are composed of noncovalently-associated α and β subunits. Each subunit is a type I transmembrane glycoprotein with a relatively large multidomain extracellular projection and a single membrane-spanning helix, usually ending with a short (20 to 70 amino acid), largely unstructured cytoplasmic tail (8). Humans produce 18 α and 8 β subunits that combine to form at least 24 different heterodimers, each of which binds to a specific overlapping repertoire of extracellular matrix ligands such as fibronectin, collagen, laminin, or fibrinogen and to cell surface counterreceptors like the Ig-superfamily proteins intercellular adhesion molecule–1 (ICAM-1) or vascular cell adhesion molecule–1 (VCAM-1) (8). Many of the advances in our understanding of the mechanisms by which integrins bind ligand have come from x-ray crystallography of the extracellular domains of αVβ3 and αIIβ3, as well as structures of complexes between ligands and isolated α subunit ligand-binding A domains [reviewed in (11, 12)]. These studies have revealed the architecture of integrin extracellular domains, explained the well-established requirement for divalent cations in integrin-ligand binding, and demonstrated that both the α and β subunits participate in binding ligands containing an Arg-Gly-Asp peptide or a related tripeptide motif. Very important in their function, conformational changes in the integrin extracellular domains play a major role in regulating the affinity of integrins for their extracellular ligands through a process termed integrin activation (8, 11, 12).

Although ligand binding is mediated by the large extracellular domains, the integrin cytoplasmic tails play a key role in cellular control of their adhesive interactions and the subsequent dynamic cellular responses such as cell spreading or migration. Interactions of the short cytoplasmic tails, and of the β tails in particular, with intracellular cytoskeletal and signaling proteins figure prominently in the regulation of integrin activation (13). Furthermore, after the binding of an extracellular ligand, complex multiprotein assemblies of cytoskeletal, scaffolding, and signaling proteins are recruited to the integrin cytoplasmic face, where they both link integrins to the actin cytoskeleton and convey signals into the cell (14, 15). Hence, by binding both extracellular and intracellular ligands, integrins provide a transmembrane conduit for the bidirectional transmission of mechanical force and biochemical signals across the plasma membrane to regulate cell adhesion, migration, proliferation, and death.

Integrin Anchoring in Cell Adhesion

The mechanical properties of adhesive attachments to cells are most often attributed to the ligand/receptor interaction. However, formation, strength, and survival of a cell adhesive attachment also depend on how molecular connections below the membrane surface—those anchoring the receptor to the cell cytostructure—respond to force. Integrins generally function in specialized complexes involving assemblies of many adhesion molecules and cytoskeletal and signaling adaptors (15). These integrin clusters come in various forms, for example, focal adhesions, focal complexes, fibrillar adhesions, or podosomes, which are defined according to their size, shape, subcellular localization, molecular constituents, and organization (15). While some clusters are widespread, others, for example, the immunological synapse or costamers, show strict cell-type specificity. The differences in size and composition of adhesion sites presumably reflect the link to the cytoskeleton and integrin signaling. Nonetheless, despite their various specialized roles in mediating transient or stable adhesion, reorganizing the extracellular matrix, and activating specific signaling pathways, these adhesions share a number of common features. They are sites at which integrins connect intracellular actomyosin-generated cytoskeletal contractility to extracellular ligands and where external forces can be transmitted to the cytoskeleton, for example, sites where they can initiate biochemical signals. Like the exterior ligand/receptor interaction, intracellular molecular bonds are also time-dependent connections whose formation and persistence change considerably with application of force. As illustrated in Fig. 4, the abrupt reduction in interfacial stiffness and the onset of fluid-like tether flow often observed when pulling on an integrin bond suggest that the pulling force can disrupt the molecular-scale complex anchoring the integrin tails to the cytoskeleton.

Fig. 4.

Receptor unbinding from the cytoskeleton: cohesive failure. When pulling a cell-surface bond with a probe, force usually builds up steadily until either the adhesion bond fails and the probe recoils [as in (A) when probing the β2-integrin LFA-1 on a blood granulocyte (46)] or the rise in force slows abruptly for a period of time before failure of the adhesion bond [as in (B) when probing the β1-integrin VLA-4 on a B-lymphocyte (46)]. As sketched schematically in (C), pulling on the receptor initially deforms the cell surface into a nanoscale “pucker” that increases its extension in proportion to the force, where the elastic-like response reflects the level of tension-like stress in the cortical cytoskeleton (49). However, as seen in (B), the elastic-like response can end prematurely with the onset of a fluid-like tether flow enabled by release of the receptor from the cytoskeleton. As sketched in (D), continued probe extrusion of this membrane nanotube (tether) quickly distances the receptor from the cell cytostructure (50, 51) and eventually ends when the adhesion bond breaks. The question mark in (D) indicates that the components that stay attached to the receptor tail domain remain to be established.

The many molecules present in adhesion sites imply many, potentially parallel, mechanisms for linking an integrin to the cytoskeleton (14, 15). Nonetheless, several proteins have been identified as prime candidates for direct integrin-actin linkages, including talin, filamin, α-actinin, and tensin [which, along with integrin-associated adaptor and signaling molecules such as vinculin, paxillin, focal adhesion kinase, and Src-family kinases, activate and/or respond to kinase, phosphatase, and small guanosine triphosphatase signaling cascades (16, 17)]. Each of these large actin-binding proteins also contains a binding site for integrin β subunit cytoplasmic tails (15, 16, 18, 19). Of these, talin, an antiparallel homodimer composed of 250-kD subunits, has received the most attention. Acting as a “hub” in the linkage between integrin β tails and the cytoskeleton, talin interacts with a constellation of focal adhesion proteins—including the integrin β-chain tail, vinculin, focal adhesion kinase, phosphatidylinositol phosphate kinase type 1 γ, and F-actin (20)—and plays important roles in activation of integrin receptors (21), in formation of the initial linkage between ligand-occupied receptors and the cytoskeleton (9), and in the subsequent reinforcement of the linkage (22). As discussed below, the use of cell lines deficient in these linker proteins (9, 19) or expressing mutations that selectively disrupt an integrin-linker, a linker-linker, or a linker-cytoskeletal interaction (9, 19, 21, 23) holds considerable promise for elucidating the roles of specific molecular interactions during cellular response to force.

The Future: Mapping the Communication Between Mechanical and Chemical Circuitry of a Cell

All cells sense and respond to applied forces in a cell-type–specific manner to regulate a broad range of processes from cell migration to stem cell differentiation, tissue formation, and tumorigenesis (24, 25). Although a variety of systems are employed to sense force and convert it into biochemical signals, adhesion molecules (and integrins in particular) are known to play an important role in this mechanosensory process and in how cells respond to the applied stress (26, 27). Although other adhesion receptors, such as the selectins and cadherins, are also regulated through cytoskeletal interactions, a defining feature of integrins is that integrin/ligand and integrin/cytoplasmic connections transmit and receive feedback (through conformational changes) to enhance or reduce their strengths of attachment (12, 28, 29). Along with the prominent clustering of receptors [increasing what is referred to as “avidity” (30)], feedback from inside the cell also acts directly on the integrin adhesion bond, greatly amplifying its mechanical strength as demonstrated by the force responses of attachments to the β2-integrin LFA-1 on the surface of a cytokine-stimulated white blood cell (Fig. 3B). Because many of the enzymes and signaling molecules involved in this feedback are closely associated with the receptor-cytoskeletal linkage, the effect of pulling on, or even detaching, a receptor from the cytoskeleton as described in Fig. 4 is likely to alter interactions among these proteins, which suggests a physical mechanism for communication between the mechanical (stress-bearing) circuitry and chemical circuitry of a cell. Pulling forces can catalyze cellular processes in many ways (26, 29), including (i) conformational transitions (from limited to full denaturation and unfolding) that expose otherwise cryptic sites to promote new protein-protein interactions or that expose sites with specific posttranslational modification, (ii) reorganization or segregation of specific molecules in an adhesive complex, and (iii) even liberation of a constituent so that it can interact with more distant complexes.

Because of the numerous molecules present in adhesion complexes, the experimental challenge is to sort through the many (possibly parallel) intracellular signaling pathways that likely emanate from integrin linkages to the cytoskeleton. Current investigations with engineered knockout or overexpressing cells are making substantial progress in identifying the proteins important for integrin-mediated responses to force (22, 31), which are enhanced by innovative in vitro assays suggesting ways that the integrin-associated proteins can act as force sensors (32). Moreover, potentially aiding in this quest, probing individual cell adhesion complexes with ultrasensitive force techniques (Fig. 4) provides an unexpected opportunity to assay the kinetics of molecular connections (Fig. 5) hidden beneath the cell membrane (maybe even deep in the cell, if linked to a long structural filament). Taking advantage of structural and functional assays that identify key mutations selectively targeting integrin-cytoskeletal linkages and signaling pathways, the exciting prospect is to use cell-surface force spectroscopy and engineered cell lines as a material science tool to explore and characterize key nodes in the “mechanical circuitry” that connect receptor tails to the cytostructure and to examine how forces applied to these nodes communicate physical cues from outside the cell to catalyze or trigger specific steps in cell signaling and regulation inside the cell. Even bolder, the next step should be to integrate precision techniques like single-molecule force spectroscopy with high-resolution optical techniques like single-molecule fluorescence (33) or novel methods that image the real-time dynamics of coupling between integrins, actin, and other components of adhesion and signaling at the cellular level (34, 35). Together, such integrated approaches can provide access to the molecular machinery by which adhesion molecules transmit force and biochemical signals into and out of the cell during cell migration, tissue remodeling, and differentiation.

Fig. 5.

Receptor-cytoskeletal anchoring strengths. Although only a few cases have been examined in detail (46, 52), the forces for receptor release from the cell cytoskeleton at the onset of tether flow have been found to increase with pulling speed and produce histograms that agree with the idealized bond kinetics described in Figs. 1 and 2. Plotted versus the logarithm of the force rate measured during the elastic-like response before the onset of tether flow, examples are shown of the most frequent forces observed at release of PSGL-1 from the cytostructure of a blood granulocyte [when attached to a PS probe (52)] and observed at release of the VLA-4 from the cytostructure of a lymphocyte [when attached to a VCAM-1 probe (46)], which provide direct assays of the kinetic stability and rate of failure for the molecular complexes anchoring these receptors beneath the cell surfaces.

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article