Single-cell force spectroscopy

A broad spectrum of biological processes requires controlled cell adhesion, including embryonic development, assembly of tissues and the nervous system, cellular communication, inflammation and wound healing, tumor metastasis, cell culturing, and viral and bacterial infection. Although much is known about cell adhesion, many questions remain unanswered owing to its multiple facets and complexity. Cell adhesion is commonly defined as the binding of a cell to a substrate, which can be another cell, a surface or an organic matrix. The process is regulated by specific cell-adhesion molecules (CAMs), which are typically transmembrane receptors that comprise an intracellular domain that interacts with cytoplasmic proteins, including the cytoskeleton, and an extracellular domain that specifically binds to adhesion partners (Kemler, 1992). Binding is commonly heterotypic, but it can be homotypic, such as that involving cadherins. The major classes of CAMs in mammals include cadherins, selectins, integrins and Ig-CAMs (cell-adhesion molecules of the immunoglobulin superfamily). Molecular and genetic approaches have identified the adhesion proteins and their ligand specificities, and have determined the processes in which they are involved. However, the molecular mechanisms by which CAMs work and how they regulate different types of adhesion are open debates (Morgan et al., 2007). For example, an extensive array of proteins is known to be involved in adhesive assemblies, i.e. focal adhesions [cell–extracellular-matrix (ECM) junctions], but the contributions of these proteins to the strength of adhesion are not quantitatively understood (Lo, 2006). To understand cell adhesion, therefore, the vast amount of qualitative data that is available must be augmented with quantitative data of the physics of adhesion.

Historically, the strength of the adhesion of a cell to a substrate has been studied using simple washing assays (Klebe, 1974)¹. Surprisingly, given the lack of standardization, washing assays have proven to be versatile and useful in identifying CAMs, important ECM components and other proteins that are involved in various forms of cell adhesion. To estimate the force to which cells are subjected, various assays that are based on the regulated flow of media have been implemented, including spinning-disk (Garcia et al., 1997) and flow-chamber (Kaplanski et al., 1993) assays. Unfortunately, the shear force that is exerted on the cells in these assays depends on parameters such as cell size, cell shape and how the cell is attached to the substrate, and can therefore only be estimated. For a more controlled and quantitative approach to measurements of adhesion strength, single-cell methods are needed.

Three types of single-cell force spectroscopy (SCFS) assays have been developed to measure the strength of cell adhesion down to single-molecule levels. All three assays use optical microscopes to observe the cell while force measurements are made, but differ in how cells are manipulated and forces are determined. The oldest method uses micropipettes to grasp and hold cells. The detachment force is measured using a bio-membrane force probe², which can gauge force between 10^–2 pN (pico-Newtons) and 100 pN (Evans et al., 1995). A second method uses a pipette to hold a cell while the strength of interactions between the cell and a functionalized bead are determined using a laser trap. The laser trap allows three-dimensional positioning of the bead with nanometer precision and force measurement from 10^–2 pN to 200 pN (Litvinov et al., 2002). The third method uses a cell that is attached to a cantilever of an atomic force microscope (Fig. 1). By combining atomic force microscopy (AFM) and optical microscopy, cells can be positioned to assess cellular interactions at a given location on a functionalized surface, tissue or on another cell (Benoit et al., 2000). The deflection of the cantilever is used to measure interaction forces. Among SCFS approaches, the AFM-based technique allows for the widest practical force range, from 10 pN to 10⁶ pN. This Commentary will be limited to the AFM-based method, which will henceforth be referred to as SCFS.

The capability of AFM to image cell topology or characterize cell-surface properties is outside the scope of this article, and readers are referred to other reviews (Radmacher, 2002; Dufrene, 2004). This Commentary will focus on the use of AFM to measure adhesion strength between a single cell and a substrate that is presented by a functionalized surface or by another cell. We will explain and demonstrate the capabilities of AFM and familiarize the reader with it benefits and limitations.

The basic experimental AFM-type set-up for SCFS is straightforward. An atomic force microscope that is fitted with a fluid chamber allows measurements to be made in aqueous environments under controlled temperatures. Suspended cells are added to the fluid chamber and allowed to settle. Thereafter, a single cell is captured by gently pressing a functionalized AFM cantilever onto it (Fig. 1). This converts the living cell into a probe, which is brought into contact with functionalized surfaces or other cells at a set force and for a specific adhesion time. Subsequently, the cantilever is withdrawn at a constant speed, detaching the cell from its binding place. During this separation process, the cantilever deflection, which is proportional to the vertical force that exists between the cell and substrate, is recorded in a force-distance curve (Fig. 2). This curve provides the signature of the cell adhesion. The challenge, however, lies in interpreting this signature, because various specific as well as unspecific adhesion processes can occur simultaneously.

The de-adhesion of a cell from a substrate that is described by the force-distance retrace curve can be broken into three phases (Fig. 2). During the initial phase (Fig. 2Ba), the retraction of the cantilever inverts the force that is acting on the cell from pushing to pulling. As the overall pulling force increases, the force that is acting at individual cell-substrate adhesion points increases. If many receptors act together, the applied detachment force will be sufficiently high to mechanically deform the cell cortex. The binding strengths of the receptors, as well as their number and geometric placement, determines at what force the cell will start to detach. The largest adhesion force that is recorded, the detachment force (F_detach), represents the maximum strength of cell-substrate binding. Because detachment of the cell is a complicated process, the maximum adhesion strength represents only a useful general measure. The work that is required to detach the cell can also be used to describe the adhesion strength of the cell. It is calculated from the area that is enclosed by the retraction-force–distance curve (Fig. 2B). Here, it is important to consider that the detachment force is a composition of many different properties of the cell (Bershadsky et al., 2006). These include cell elasticity, cortex tension, membrane properties, cell geometry and receptor properties such as binding strength, cooperativity and placement.

After the cell starts to detach from the substrate, individual force steps can be observed during the second phase (Fig. 2Bb). During this phase, the receptor(s) either detaches from the substrate surface or is pulled away from the cell cortex at the tip of a membrane tether. While parts of the cell cortex are in contact with the substrate, either of these processes can occur. During the final phase of detachment (Fig. 2Bc), the cell body is no longer in contact with the substrate and, thus, attachment is mediated exclusively by tethers (Sun et al., 2005a). The force that is required to extend a tether depends on the lipid composition of the cellular membrane and on the mechanical properties of the cell cortex. Thus, the lifetime of a membrane tether is dependent on the receptor-ligand interaction at its tip, whereas the force that is required to maintain and extend a tether is not (Marcus et al., 2004). Once initiated, this force is largely independent of tether length (Hochmuth et al., 1996). In cell-cell adhesion experiments, retraction distances that approach 100 μm are required, owing to tethers, to fully separate cells (Benoit and Gaub, 2002; Puech et al., 2006; Thie et al., 1998). There is ongoing research aimed at trying to use the mathematically tractable tethers to analyze receptor anchoring (Schmitz et al., 2008). Once all of the tethers have detached from the substrate, another cycle of adhesion and detachment is started after a short cell-recovery time.

A common variation of this method is an inversion of the set-up. Here, the cantilever of the atomic force microscope is functionalized with ECM proteins and used to probe immobile tissue-culture cells (Lehenkari and Horton, 1999). The experimental set-up is flexible and varies with the biological system. Many different cells and extracellular adhesion substrates have been used. It is also possible to apply this approach to detect the molecular adhesion events of microbial surfaces (Dufrene, 2004). This set-up can also be used to map the probe-binding properties of cell surfaces. However, SCFS that uses the cell as a probe has certain advantages – most importantly, cell-cell interactions can be probed and there is more freedom in which substrates, such as ECM components, are presented to the cell.

Many aspects of adhesion can be examined using AFM-based SCFS, ranging from cell-cell to single-molecule experiments, and there are no restrictions that govern which CAMs can be studied or which cells can be used. Initial experiments, which were performed a decade ago, studied cell-cell adhesion of Escherichiacoli and mammalian cells that were grown on cantilevers, but these studies were not at the single-cell level (Razatos et al., 1998; Thie et al., 1998). Shortly thereafter, measurements of the binding strength between RGD peptides (synthetic peptides that contain the RGD integrin-binding motif) and osteoclasts indicated the viability of using living cells and AFM to study single-molecule binding properties (Lehenkari and Horton, 1999). Since then, a wide variety of adhesive interactions have been studied using many types of cells. Table 1 lists the combination of single-molecule receptor-ligand interactions that have been studied using SCFS. The various cell types that have been used are also shown in the table.

Although the atomic force microscope is a high-precision force-measuring tool, it is versatile. Using piezoelectric actuators, the atomic force microscope probe can be positioned with sub-nanometer accuracy at relatively high speeds (>100 μm/second). With SCFS, cell-substrate contact times can range from milliseconds to tens of minutes. The imaging capability of the atomic force microscope can also be used to characterize the adhesion substrate at a spatial resolution that approaches 2 nm, which clearly exceeds that of light microscopy (Cisneros et al., 2006; Franz and Muller, 2005; Taubenberger et al., 2007). Thus, for example, AFM images of cells adhering to aligned collagen matrices revealed cell-induced rearrangements of individual collagen fibrils (Friedrichs et al., 2007a). In addition, AFM can be combined with most modern optical techniques, such as fluorescence-correlation spectroscopy, wide-field fluorescence, total internal-reflection fluorescence and confocal microscopy (Chiantia et al., 2007; Franz and Muller, 2005; Puech et al., 2006; Trache and Meininger, 2005). Commercial atomic force microscopes that can be integrated into standard and modern inverted and transmission optical microscopes are available³. In addition, the flexibility and ease of AFM-based SCFS measurements extends its use from quantitatively characterizing whole-cell adhesion down to single receptor-ligand interactions.

As one would expect, the strength of adhesion increases with the length of time a cell is allowed to adhere to a substrate or another cell. Initially, single receptor-ligand pairs anchor the cell. These quickly increase in number and undergo modifications to greatly increase the total strength of adhesion (Friedrichs et al., 2007b; Taubenberger et al., 2007; Thie et al., 1998). Thus, by simply varying the cell-substrate contact time during SCFS, both the adhesion properties of single molecules and whole cells can be quantified.

Since its inception (Binnig et al., 1986), AFM has been used to study molecular interactions (Butt, 1991; Ducker et al., 1991). However, it took 8 years until the first set-ups that could measure discrete interactions between single molecules were designed (Lee et al., 1994; Moy et al., 1994). Since then, this technique, termed single-molecule force microscopy (SMFS), has been applied to characterize the binding behavior of many different oligosaccharides, nucleic acids and proteins (Hansma et al., 2004; Kedrov et al., 2007; Zhuang and Rief, 2003). Receptor-ligand interactions are examined by measuring the unbinding (or rupture) forces between receptors (or ligands) that are attached to the stylus of the cantilever of the atomic force microscope and ligands (or receptors) that are immobilized to a surface. Dynamic SMFS probes these rupture forces at different loading rates (that is, applied force versus time) to determine the properties of the receptor-ligand energy landscape (Evans, 1998; Evans and Calderwood, 2007). These properties typically include the free energy (ΔG) that separates the bound state from the transition state, the distance that separates the bound state from the transition state (x_u) and the lifetime of the bound state at equilibrium (k_off). Although these in vitro measurements provide insights into the behavior of receptor-ligand interactions, they have limitations.

For example, receptors must be purified, which means removing them from their biological context, so one cannot be certain of their functional state. This is of particular concern with integrins that are known to have several substrate-binding affinity states. Furthermore, transmembrane receptors are purified in truncated forms, and therefore consist of only extracellular domains. This is problematic because it is known that some receptors are regulated through interactions with cytoplasmic factors, such as the regulation of integrins by paxillin (Rose et al., 2007).

In contrast to SMFS, SCFS enables single receptor-ligand interactions to be examined in their cellular environment. Using a living cell as a probe ensures that the receptors are native. In Table 1, we list the unbinding forces of receptor-ligand interactions that have been measured using SCFS. Several of these receptor-ligand pairs have also been studied using SMFS on solely purified proteins. The rupture forces that are measured by SCFS and SMFS are generally in agreement, but some show considerable deviation from each other. This might indicate that the strength of receptor-ligand interactions depends on the experimental conditions. Unfortunately, experimental conditions have not been standardized and are often not sufficiently documented to allow for rigorous comparisons, i.e. experiments are performed at different temperatures or the manner in which the bond is stressed is not clear. Moreover, it is possible that, although the binding strength that is measured by the two techniques is similar, the energy barrier that is crossed to break the bond differs. The external force that is applied to a bond defines the coordinate along which the bond is forced to break. In SMFS, receptor-ligand pairs are isolated, truncated and attached to solid surfaces, whereas in SCFS the receptors are in their cellular environment. Thus, bonds might break across different paths within the energy landscape of the binding interaction. Future dynamic SCFS experiments might show to what extent the biological context of receptors influences the energy barriers that separate bound and unbound states.

SCFS is not without limitations – the fact that SCFS uses a cell as a probe can also complicate certain aspects of single-molecule measurements. In contrast to the rigid stylus of the atomic force microscope that is used in SMFS, the applied forces in SCFS cause the cell to stretch and deform. In addition, because the mechanical response of cells to deformation is not necessarily linear, dynamic SCFS measurements are not as straightforward as SMFS measurements (Evans and Calderwood, 2007). Perhaps more crucial is the multitude of possible specific and unspecific cell-surface interactions. Therefore, special care must be taken to ensure that the interactions that are recorded occur predominately, if not exclusively, between the receptor and ligand of interest. To this end, purified substrates and blocked surfaces are used. Cells can also be genetically modified to limit the number of possible receptors that are expressed, and rigorous control experiments that demonstrate the specificity of the interactions observed must be performed.

The use of a living cell to study single-molecule interactions at the cell surface has proved fruitful; however, it was shown early on that SCFS could be used to measure the dynamic-force spectrum of binding interactions (Chen and Moy, 2000). Making use of the advantages of living cells, Moy and co-workers demonstrated that the activation of leukocytes induced changes in the unbinding-energy landscape of the integrin LFA-1 (ITGAL) from its ligand ICAM1 (Zhang et al., 2002). Others have since studied the changes in binding dynamics of various integrins upon activation by antibodies (Li et al., 2003; Zhang et al., 2004) and magnesium (Wojcikiewicz et al., 2006). The properties of genetically modified receptors have also been studied (Alon et al., 2005).

Cellular process, as opposed to single molecules, can be studied by increasing the cell-substrate contact time⁴. Not surprisingly, in most SCFS studies the strength of the adhesion between two cells or a cell and a substrate increases with contact time and contact area. Generally, retraction curves show that higher detachment forces are the result of increased numbers of adhesive interactions. However, significant cell-induced deviations are seen. For example, the high early adhesion forces that occur between cells that express the surface receptor Notch and its ligand Delta diminish as the receptors are cleaved and internalized as part of the signaling pathway (Ahimou et al., 2004). By contrast, Chinese hamster ovary (CHO) cells that express integrin α2β1 and are in contact with aligned type I collagen switch to an activated adhesion state (Taubenberger et al., 2007). This probably occurs as a result of the clustering of receptors into load-sharing entities and not because of the activation of individual integrin molecules. This phenomenon was observed for integrin-α2β1-mediated binding of Madin-Darby canine kidney (MDCK) cells to both type I and type IV collagen (J. Friedrichs, A. Manninen, D.J.M. and J.H., unpublished). These papers demonstrate that SCFS can be used to study dynamic and regulated adhesion processes that occur at the cellular level.

For the assessment of adhesion processes at a level that mimics the in vivo state, isolated primary cells have been used. For example, cells that were isolated from zebrafish embryos were used to examine the importance of non-canonical Wnt signaling for cell adhesion in early development, and the authors used SCFS to determine the specific adhesion of different types of primary cells to functionalized substrates as well as to other cells (Puech et al., 2005; Ulrich et al., 2005). Moreover, using the same system, the specific contributions of cell adhesion versus cell-cortex tension to cell sorting during zebrafish gastrulation have been clarified (Krieg et al., 2008).

Force-distance curves that trace cell detachment reveal the unbinding of individual receptor-ligand interactions and, thus, binding frequency (Gilbert et al., 2007). By simultaneously measuring the contact area using light microscopy, the number of active receptors per cell-surface area can be estimated and receptor-ligand attachment rates found (Gilbert et al., 2007). If combined with other techniques, such as FACS to determine the number of particular cell-surface receptors, SCFS allows the fraction of activated and inactivated receptors to be estimated. A search for functional states or environmental conditions that tune receptors might be possible with this combination. Recent SCFS studies on the adhesion of leukemic cells to bone-marrow stromal cells showed that the myeloid leukemia fusion protein BCR-ABL increased integrin β1 expression, which, in turn, increased the strength of the adhesion between the two cell types (Fierro et al., 2008). The addition of the anti-cancer drug imatinib mesylate suppressed the integrin-β1-dependent adhesion to the level of control cells.

Current SCFS set-ups do have some limitations. Adhesion measurements that use single cells are time consuming because only one cell can be characterized at a time. For statistical reasons, many detachment-force–distance curves must be recorded, which limits the length of the contact times that can reasonably be assayed. Furthermore, the almost unavoidable thermal drift in AFM complicates long-contact-time experiments (>20 minutes) and the tight adhesion of cells after longer contact times (>1 hour) exceeds the capability of the system⁵. Thus, SCFS is currently restricted to short contact times that range from milliseconds to ∼20 minutes. There is also a high cost associated with SCFS; fortunately, inexpensive atomic force microscopes that have been specifically developed for SCFS should soon become available.

As with most new techniques, SCFS needs to mature. Presently, enthusiasm to publish comes, at times, second to rigorous examination of the data. Of particular concern are: (1) the need to establish controls that demonstrate the specificity of the molecular interaction being studied; (2) the temptation to over-interpret numerical data that are gleaned from unverified mathematical models; and (3) the difficulty in appreciating the complexity of both the physics and biology of the systems studied. This situation will improve as familiarity with the technique increases and standard experimental procedures and data-analysis norms are adopted.

The use of SCFS is still in its infancy and has much potential for development. This potential is based on the versatility of SCFS and the enormous variety of cell biological and medical applications to which it can be applied. Here, we have demonstrated that SCFS provides a `force signature' of the cell-adhesion process and have shown how this tool has been used to study CAMs and the dynamics of regulated adhesion processes in living cells. In practice, all possible forms of cell adhesion can be studied, with limitations that include the restriction of experiments to short contact times and the high associated cost, as discussed above. The combination of AFM with advanced light-microscopy imaging has yet to be applied to its full advantage, and studies in which force measurements are correlated with changes in cell shape and structure will emerge in the foreseeable future. The development of single-cell force spectroscopes that have standardized cell-handling and analysis routines will also provide a possibility to expand the experimental parameters that can be addressed with this innovative technique.

We thank A. Taubenberger, C. Franz, J. Friedrichs, M. Krieg and K. Simons for stimulating discussions. The Deutsche Forschungsgemeinschaft, the European Union, the Munich Center for Integrated Protein Science and the Volkswagenstiftung supported this work.