Temperature Modulation of Integrin-Mediated Cell Adhesion

Abstract

In response to external stimuli, cells modulate their adhesive state by regulating the number and intrinsic affinity of receptor/ligand bonds. A number of studies have shown that cell adhesion is dramatically reduced at room or lower temperatures as compared with physiological temperature. However, the underlying mechanism that modulates adhesion is still unclear. Here, we investigated the adhesion of the monocytic cell line THP-1 to a surface coated with intercellular adhesion molecule-1 (ICAM-1) as a function of temperature. THP-1 cells express the integrin lymphocyte function-associated antigen-1 (LFA-1), a receptor for ICAM-1. Direct force measurements of cell adhesion and cell elasticity were carried out by atomic force microscopy. Force measurements revealed an increase of the work of de-adhesion with temperature that was coupled to a gradual decrease in cellular stiffness. Of interest, single-molecule measurements revealed that the rupture force of the LFA-1/ICAM-1 complex decreased with temperature. A detailed analysis of the force curves indicated that temperature-modulated cell adhesion was mainly due to the enhanced ability of cells to deform and to form a greater number of longer membrane tethers at physiological temperatures. Together, these results emphasize the importance of cell mechanics and membrane-cytoskeleton interaction on the modulation of cell adhesion.

Introduction

The modulation of cellular adhesion is a complex process that has been the subject of intense research and controversial debate due to its relevance in many cellular processes, including differentiation, migration, and division. In response to varying external biochemical and biophysical stimuli, cells regulate their adhesive state by modulating the number and binding capacity of their receptors to ligands (1). A paradigmatic example of cellular adhesion modulation is found in integrin-mediated leukocyte adhesion. Integrins are transmembrane proteins that are expressed on the surface of cells and have been shown to mediate leukocyte rolling, firm adhesion, and migration (2). Under pathological conditions, such as inflammation, leukocytes are activated by chemokines that induce changes in their adhesive state (3). This adhesion enhancement allows leukocytes to firmly adhere to the vascular endothelium and then migrate to the subendothelial tissue through extravasation. Another well-known case of cell adhesion modulation is temperature-modulated adhesion. Various studies have shown that temperature has a dramatic effect on the capacity of different types of cells to adhere (4–8). Moreover, the effect of temperature on cell adhesion appeared to be more pronounced within the first 15 min of cell contact (4,6). A recent study showed that within this timeframe, the early steps of a cell adhering to a surface (i.e., early cell spreading) could be explained by the viscoelastic properties of cells (9). Early studies by Waugh and Evans (10) showed that the deformability of cells increased with temperature. Moreover, the capacity of erythrocytes and leukocytes to flow through narrow capillaries has been shown to be reduced at low temperatures, indicating an increased resistance to deformation (i.e., increased stiffness) (11,12). Several recent works described both passive and active regulation of cell adhesion by mechanical triggers (13–18). For example, Caputo and Hammer (18) showed how microvillus deformability modulates rolling velocities, and Friedland and coworkers (13) showed how force can reinforce integrin-mediated adhesion. Thus, given the observed link between cell adhesion and cellular mechanics, we speculate that the viscoelastic properties of cells play a central role in the temperature modulation of cell adhesion.

The aim of this work was to investigate the molecular and biophysical determinants of cell adhesion modulation, using temperature-enhanced cell adhesion mediated by integrins as a model system. We used an atomic force microscope (AFM) to measure the effect of temperature on integrin-mediated adhesion and the elasticity of living monocytic cells. In AFM measurements, adhesion is probed by recording the forces that are necessary to completely detach a ligand-coated surface from the cell surface, whereas cell elasticity is determined from force-indentation curves analyzed using contact elastic theory (19–23). We used monocytic cells (THP-1) that express the integrin lymphocyte function-associated antigen-1 (LFA-1, α_Lβ₂), which is the most important receptor mediating adhesion to intercellular adhesion molecule 1 (ICAM-1) in leukocytes (24). An AFM tip coated with ICAM-1 was used to probe the effects of temperature on integrin-mediated adhesion and the elastic properties of the cells. The results obtained from a detailed analysis of force measurements revealed a dramatic increase in the work of deadhesion and cellular compliance from room temperatures to physiological temperatures. The mechanism underlying temperature-modulated adhesion was not explained by an important increase in the affinity of single integrin/ligand bonds, but rather by the modulation of cell elasticity and the cells' capacity to form a greater number of membrane tethers that were also longer in length.

Materials and Methods

Cell culture and reagents

The cell line THP-1 expressing integrin LFA-1 was maintained in continuous culture in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (Irvine Scientific, Santa Ana, CA), penicillin (50 U/mL; Gibco BRL, Grand Island, NY), and streptomycin (50 mg/mL; Gibco BRL). ICAM-1/Fc chimera and monoclonal antibody against ICAM-1 were purchased from R & D Systems (Minneapolis, MN). Stock solutions of cytochalasin D (cytD, 1000X; MP Biochemicals, Solon, OH) were prepared at 20 mM in DMSO.

AFM cantilevers preparation

AFM cantilevers were functionalized with human ICAM-1-Fc using a glutaraldehyde linkage to rule out possible destabilization by temperature. The cantilevers were initially silanized with 3-aminopropyltriethoxysilane. After incubation of the cantilevers with 0.1% glutaraldehyde for 30 min, ICAM-1-Fc (2.5 μg/mL) was coupled to the cantilever through the glutaraldehyde linker. Incubation for 1 h with 1% bovine serum albumin was used to block unspecific adhesion events.

AFM force measurements

Force measurements were carried out on a custom-built AFM with temperature control (25) using Si₃N₄ cantilevers with a nominal spring constant of k = 0.01 N/m (Veeco, Santa Barbara, CA). The spring constant of the cantilevers was calibrated by thermal fluctuation analysis (26) and did not vary significantly with temperature.

AFM force measurements were carried out on living THP-1 immobilized on dishes coated with poly-L-lysine (0.1 mg/mL, 20 min incubation) at 16°C, 24°C, and 37°C ± 0.5°C. The measurement buffer consisted of HEPES (10 mM) buffered RPMI culture medium containing 5 mM Mg²⁺, 1 mM EGTA, and 0.01% bovine serum albumin. After calibration, THP-1 cells were deposited on the dish and allowed to immobilize for 2 min. The ICAM-1 functionalized cantilever tip was then positioned on the center of a cell, and 5–10 force-distance (F-z) curves were acquired by approaching the tip to the cell at 3.75 μm/s, maintaining contact for 2 s, and subsequently retracting the cantilever at the same speed. The maximum indentation force was ∼400 pN. Curves were obtained on at least nine cells from a minimum of three independent experiments per temperature and condition. The experiments typically lasted 30 min and never longer than 1 h.

Single-molecule measurements were carried out on the same system by minimizing the indentation force (∼50 pN) and the contact time (∼20 ms) to reach an adhesion frequency of ∼30%. This ensured that ∼85% of the events were due to single LFA-1/ICAM-1 complexes (27).

Additional details regarding data processing and statistics are provided in the Supporting Material.

Results

Temperature-modulated cell adhesion

We quantified the effect of temperature on integrin-mediated cell adhesion by conducting a detailed analysis of the retraction force curves resulting from detaching ICAM-1-coated AFM tips from the surface of THP-1 cells expressing LFA-1. Measurements were carried out in the presence of 5 mM Mg²⁺/1 mM EGTA, which increases the binding affinity of LFA-1 but not of Mac-1 (α_Mβ₂) (28), to rule out the possibility of a major conformational change in the binding domain of LFA-1 (29). Representative examples of AFM force-distance curves obtained on living cells at the three tested temperatures are shown in Fig. 1 A. Each curve consisted of an approach (gray) and a retraction trace (from top to bottom for 37°C, 24°C, and 16°C). The approach trace presented a flat region of constant force where the tip was being lowered onto the surface but had not yet made contact. This was followed by a characteristic nonlinear force increase due to the continuously increasing contact area as the pyramidal tip indented the cell. After the maximum compression force was reached, a force drop was observed due to stress relaxation of the cell, which reflected the viscoelastic nature of cellular mechanics. The hysteresis observed between the approach and retract traces was also due in part to viscous dissipation. The retraction curve exhibited a similar nonlinear response followed by a pulling regime as receptor-ligand bonds were being stressed by the withdrawal of the cantilever from the cell substrate. The presence of a series of jumps was interpreted as the rupture of single or multiple bonds.

Figure 1.

(A) Representative force-distance curves from which the relevant parameters (maximum detachment force, work of de-adhesion (shaded area in the first curve), and maximum detachment length) are extracted. The outlines represent an immobilized cell expressing LFA-1 and a cantilever coated with ICAM-1 at different steps of a force curve (noncontact (I), indentation (II), and pulling (III). Inset images show lateral views of fluorescently labeled THP-1 cells at each temperature (bars = 10 μm). (B–E) Effect of temperature on the adhesion of THP-1 cells to ICAM-1: (B) work of de-adhesion, (C) apparent adhesion energy density (work of de-adhesion/area), (D) adhesion force per perimeter (maximum force/perimeter), and (E) maximum detachment distance.

We quantified the cellular adhesion from retraction force curves by measuring the work of de-adhesion, the detachment force, and the maximum detachment distance (Fig. 1 A). In addition, the known contact geometry between the AFM tip and the cell surface enabled us to calculate the apparent adhesion energy density (work of de-adhesion/area of contact) and the adhesion force per perimeter (detachment force/perimeter of contact; see Supporting Material, and inset in Fig. 4 A). The work of de-adhesion, calculated by integrating the force over distance, reflects the total work required to deform and completely detach the AFM tip from the cell surface, and showed a significant ∼8-fold increase from 16°C to 37°C (p < 0.0001; Fig. 1 B). In contrast, the apparent adhesion energy density presented no significant increase (Fig. 1 C). Of interest, the adhesion force per perimeter presented a slight insignificant, decreasing trend (Fig. 1 D). The maximum detachment distance required for complete detachment paralleled the increase found for the work of de-adhesion, presenting a significant ∼9-fold increase (p < 0.0001; Fig. 1 E). Similar results were observed in measurements carried out using tips coated with monoclonal antibody against LFA-1 (TS 1/22) instead of ICAM-1 (Fig. S1 in the Supporting Material). This antibody binds to an epitope that does not change after affinity modulation (30). Thus, these measurements suggest a mechanism other than intrinsic integrin affinity modulation. We tested the specificity of the interaction at 37°C by comparing the work of de-adhesion at 37°C with that obtained from measurements conducted in the presence of 5 mM EDTA, using uncoated tips or with ICAM-1 coated tips that had been blocked by prior incubation with anti-ICAM-1 antibody. In each condition, a decrease in the work of de-adhesion was observed with p-values of <0.001, 0.005, and 0.072, respectively (Fig. S2). These control measurements ensured that most of the unbinding events seen in the force curves were due to specific LFA-1/ICAM-1 unbinding.

Figure 4.

(A) Representative examples of approaching force traces (dotted lines) used to estimate cell elasticity with the corresponding best fits (solid lines) of the contact elastic model (Eq. S3) at 37°C, 24°C, and 16°C (from bottom to top). The sketch shows a schematic representation of a pyramidal cantilever indenting the cell body with the relevant parameters to determine the contact geometry: indentation (δ), depth of contact (h_c), and effective radius of contact (a). (B) Effect of temperature on cell elasticity (Young's modulus).

Increased temperature favors the extraction of long membrane tethers

Individual jumps preceded by force plateaus in the retraction curves were interpreted as membrane tethers, i.e., membrane tubes extracted from the cell surface and linked to the AFM tip through at least one LFA-1/ICAM-1 complex (31). The force jumps (tether forces) were interpreted as the friction force required to extract a tether at a constant velocity. Only jumps preceded by a force plateau (see Fig. 2 A and Materials and Methods for definition) were considered, and the tether forces and lifetimes were calculated from these jumps. The tether forces were significantly lower at the highest temperature (p < 0.0001; Fig. 2 B), whereas the tether lifetimes showed an opposite trend, with a ∼5-fold increase from 16°C to 37°C (p < 0.0001; Fig. 2 C). In similarity to the tether duration, the probability of tether extraction also increased (∼7% at 16°C, 23% at 24°C, and ∼35% at 37°C).

Figure 2.

(A) Enlargement of the curve at 37°C presented in Fig. 1A as a force-time plot showing tether events in the retraction trace (solid line) and the estimation of tether forces (arrows) and loading rates (dashed lines). The outlines represent two stages of the tether extraction process. Average results of the tether parameters (tether force (B) and tether lifetime (C)) at the three tested temperatures are shown. Data are shown as mean ± SE.

Increasing temperature reduces the rupture force of individual LFA-1/ICAM-1 bonds

To determine how the rupture force of the LFA-1/ICAM-1 complex changes with temperature, we conducted single-molecule measurements on the same system. Sample force scans for the LFA-1/ICAM-1 single-bond interactions are shown in Fig. 3 A with adhesion in the second, fourth, and sixth traces. Of interest, the rupture forces measured at the same retraction velocity were significantly higher at low temperatures (Fig. 3 B, left), which may reflect a weakening of the interaction at 37°C. However, this may also be a result of the observed decrease in the corresponding loading rates, which dropped from ∼1500 pN/s at 16°C to ∼300 pN/s at 37°C (Fig. 3 B, right). Given previous results obtained from the LFA-1/ICAM-1 complex, we assumed that a single barrier on the dissociation pathway dominated the LFA-1/ICAM-1 interaction at the applied loading rates (30). Therefore, to better interpret these results, we extracted the relevant parameters of the interaction (i.e., the intrinsic lifetime and depth and width of the energy well) from the force measurements using a recently published formalism, which enabled us to transform rupture force distributions into lifetimes versus applied force τ(F) (32). The parameters obtained from fitting Eq. S2 are shown in Table 1. A decrease in the intrinsic lifetime and activation energy was observed with increasing temperature.

Figure 3.

(A) Six representative retraction curves of single-molecule measurements on cells obtained at 24°C. Only the second, fourth, and sixth curves exhibited rupture events. (B) Rupture force (left) and loading rate (right) distributions of individual LFA-1/ICAM-1 complexes measured at 16°C, 24°C, and 37°C (from top to bottom). (C) Force distributions at each temperature were transformed into lifetimes versus force plots using Eq. S1 (open symbols). Solid and dashed lines represent the best fits of Eq. S2 to the data and the upper and lower bounds of the 95% confidence intervals, respectively. Tether lifetimes at the corresponding tether forces for untreated (^∗) and cytD-treated (#) cells from Fig. 5, E and F, were also plotted.

Table 1.

Effect of temperature on the LFA-1/ICAM-1 interaction

Temperature (°C)	τ0 (s)	γ (nm)	ΔG‡ (kBTr)
16	${0.98}_{0.15}^{6.86}$	${0.46}_{0.33}^{0.59}$	${6.7}_{4.8}^{8.9}$
24	${0.68}_{0.13}^{3.49}$	${0.50}_{0.35}^{0.62}$	${6.4}_{4.2}^{8.5}$
37	${0.30}_{0.04}^{2.12}$	${0.23}_{0.06}^{0.51}$	${4.7}_{1.2}^{8.6}$

Interaction parameters (with 95% confidence intervals) were obtained by fitting Eq. S2 to lifetime versus force data (Fig. 3C). T_r is room temperature, k_B is the Boltzmann constant, and τ₀, γ, and ΔG^‡ are the intrinsic lifetime, width, and height of the interaction potential, respectively.

Cells are more compliant at high temperature

During AFM experiments, the applied indentation force is normally kept constant. Thus, a change in the mechanical properties of the cells will have an effect on the indentation and hence on the contact area as well. Such a change would affect the work of de-adhesion but not the apparent adhesion energy density, as we observed. Therefore, this led us to investigate the cell's mechanical properties. To estimate cellular stiffness, we analyzed approach force-distance curves at the different temperatures. Fig. 4 A shows representative approach curves on living cells at 16°C, 24°C, and 37°C, with their corresponding fits to the pyramidal contact model (Eq. S3). The cells were significantly softer at high temperature, as shown by the marked ∼14-fold decrease in the Young's modulus from 16°C to 37°C (p < 0.0001; Fig. 4 B).

Combined effect of temperature and actin cytoskeleton disruption

The observed increase in tether formation and compliance at 37°C may be explained by alterations in the cell's actin cytoskeleton. To investigate this hypothesis, we carried out similar measurements with cytD-treated cells. After treating the cells with 20 μM cytD, we observed a significant decrease in the apparent adhesion energy density (Fig. 5 B) and adhesion force per perimeter (Fig. 5 C), but, remarkably, no significant change in the work of de-adhesion (Fig. 5 A). Treatment with cytD had a significant effect on the detachment distance (Fig. 5 D), being higher at the lower temperatures. Of interest, whereas temperature had a dramatic effect on the work of deadhesion and detachment distance on resting cells, on the cytD-treated cells, temperature only affected the detachment distance, showing a moderate ∼2.5-fold increase from 16°C to 37°C (Fig. 5 D).

Figure 5.

Effect of temperature on the adhesive and elastic properties of untreated (open bars) and cytD-treated (solid bars) cells (16°C, 24°C, and 37°C, from left to right). (A) Work of de-adhesion. (B) Apparent adhesion energy density. (C) Adhesion force per perimeter. (D) Detachment distance. (E) Tether force. (F) Tether lifetime (mean ± SE). (G) Young's modulus. Statistically significant differences due to temperature are only shown (if found) for cytD-treated cells. The abscissa in G is valid for all the plots.

The analysis of individual tethers revealed that treatment with cytD significantly decreased tether forces, as compared to untreated cells, and increased tether lifetimes at the two lower temperatures but not at 37°C (Fig. 5, E and F). Moreover, cytD-treated cells were more prone to form long membrane tethers, as the frequency of tether extraction increased to ∼40% at all three temperatures. Of interest, for cytD-treated cells, temperature had a significant effect on tether lifetimes but not on tether forces.

As expected, the Young's modulus of THP-1 cells was significantly affected by treatment with cytD and dramatically dropped by as much as ∼26-fold at 16°C, although it dropped by only ∼2-fold at 37°C (Fig. 5 G). Temperature had no significant effect on the elasticity of cells treated with cytD.

Discussion

The use of temperature to modulate cell adhesion capacity enabled us to study the biophysical mechanisms underlying cell adhesion without any biochemical intervention. We observed that the work required to detach an AFM cantilever tip from the cells' surface decreased dramatically from 37°C to 16°C, confirming the results obtained from conventional adhesion assays (4,6,7). For example, in the work by Juliano and Gagalang (6), the percentage of cells that remained attached after 15 min incubation shifted from ∼80% at 35°C to ∼10% at 15°C, in agreement with the change in the work of de-adhesion reported here. In the studies cited above, the authors observed an abrupt change between 24°C and 37°C, centered at ∼30°C. The similarity of this abrupt increase to a phase transition led to the hypothesis that cell adhesion is governed by the fluidity of the plasma membrane, as temperature is known to increase membrane fluidity (33,34). However, it was found that the temperature profile for membrane fluidity presented no discontinuities, and adhesion assays using membrane-fluidizing agents showed no obvious correlation between cell adhesion and membrane fluidity (35). However, it is possible that increased membrane fluidity would have an effect on the observed enhanced tether formation and the decreased tether forces at 37°C. The effect of temperature on cell adhesion has been shown to be rapid, on the timescale of 5 min (6). Our measurements required ∼30 min from start to finish. Thus, it is unlikely that the suppression in the work of de-adhesion with declining temperature was due to a sequestering or downregulation of membrane integrins in this short timescale, as showed from flow cytometry analysis of cell surface expression of LFA-1 (Fig. S6).

The work of de-adhesion obtained from AFM force measurements has been shown to be an excellent estimate for the quantification of the adhesive state of living cells, and our results further confirm that observation (16,17). However, it does not provide us with an intrinsic parameter to define the quality of adhesion, as it depends on experimental conditions such as the area of contact (36). A more appropriate parameter is the apparent adhesion energy density, i.e., the energy per unit area required to separate two surfaces (19). This parameter depends mainly on the density and binding affinity of the formed bonds, and not on the mechanical properties of the two surfaces. As shown in Fig. 1 B, temperature had no apparent effect on the apparent adhesion energy density, which suggests that the number of receptors per unit area and their binding capacity were not the main cause of temperature-enhanced adhesion. The adhesion force per perimeter (Fig. 1 B) showed a tendency to diminish, although this was not statistically significant. When pulling from an elastic body, the stress is concentrated at the periphery of the contact zone. Therefore, the adhesion force per perimeter was interpreted as the maximum force (normalized by the perimeter) supported by the bonds in the outermost rim of contact (19,37). It was thus a rough estimate of the ligand/receptor interaction. The slight decrease in the adhesion force per perimeter suggests that temperature may have a small effect on the binding strength of the LFA-1/ICAM-1 interaction.

Our calculation of the adhesion force per perimeter and apparent adhesion energy density made use of some assumptions about the cells' geometry and the contact elastic model applied to estimate the contact area that should be addressed. First, various groups have demonstrated the advantage of the Hertzian contact model over the liquid droplet model to estimate the viscoelasticity of leukocytic cells using AFM (23,38). The dependence of the Young's modulus on indentation (Fig. S5) further confirms these results. Second, the AFM tip was assumed to indent a flat half space. Obviously, our cells were not perfectly flat but spherical, with a radius of ∼10 μm that did not change with temperature or cytD treatment, as the cells were not allowed to fully spread (Fig. 1, Fig. S4, and Table S1). An extension of Hertz's contact theory for a cone indenting a sphere was recently published (39). In our particular case, with cells of radius 10 μm and indentations <2.5 μm, the overestimation of the contact radius by assuming a flat surface is <5%, which falls well within the experimental error. Third, we assumed a perfectly smooth surface. Cells are not perfectly smooth; rather, they are rough, with microvilli (∼0.35 long in the case of neutrophils) extruding from them (40). The cell surface can then be approximated by a rough surface with a Gaussian distribution of microvilli lengths. The effect of such a rough surface on the contact of two elastic curved surfaces was studied previously (37), and the results showed that for a roughness (microvilli) distribution with a standard deviation of 10% of the indentation (∼0.13 μm for neutrophils), the effect on the effective contact radius was <7% compared to smooth surfaces. Thus, we would expect a similar bias in our calculations. Finally, as reflected by the hysteresis between the approach and retract curves, the cells were not perfectly elastic, but had a viscous contribution. The calculation of a purely elastic modulus is thus a first estimate of the viscoelastic properties of the cell. However, the mechanical response of neutrophils has been shown to be dominated by elastic stresses, with a complex elastic modulus weakly increasing with frequency (23). Thus, we would expect our Young's modulus to slightly increase with the probing velocity.

To study the effect of temperature on the adhesion strength of single LFA-1/ICAM-1 bonds, we carried out single-molecule measurements on the same system by minimizing the contact force and time. The force distributions (Fig. 3 B) were in good agreement with those reported recently on monomeric ICAM-1 at similar loading rates, ruling out the possibility of cooperative unbinding of the bonds (41). However, it is remarkable that, in contrast to the work of de-adhesion, the individual rupture forces measured at the same retraction velocity decreased with temperature. This result was expected, since it has been observed in other systems (25,42). This decrease, though, was induced by a pronounced decrease in the effective loading rate as a result of the decreased cellular stiffness. The parameters obtained from fitting Eq. S2 to lifetimes versus force data enabled us to interpret the effect of temperature on the LFA-1/ICAM-1 interaction (Table 1). The main change was observed at 37°C, where there was a reduction of the potential width and height. Given the combination of polar and hydrophobic interactions between the binding site of LFA-1 and ICAM-1, a combined destabilization of hydrogen bonds with stabilization of the hydrophobic interactions by temperature may explain the observed change (42,43). Thus, temperature affected the interaction, leading to a lower potential barrier that was also less affected by force. To estimate the capacity of the bond to support force, we defined the stiffness of the interaction (D) as the slope of the free-energy landscape (42). Because we assumed a harmonic potential, this would have the form G(x) = 1/2Dx², where G is the free-energy and x is the reaction coordinate. From this definition and the parameters in Table 1, we found D-values of 260, 210, and 728 pN/nm at 16°C, 24°C, and 37°C, respectively. The corresponding maximum forces that the complex can support (Dγ) are 120, 105 and 167 pN, respectively. Thus, although the LFA-1/ICAM-1 interaction can potentially support a higher force at the highest temperature, the <50% increase in the strength of the complex cannot account for the eightfold increase in the work of de-adhesion.

The maximum distance at which the tip and the surface detach followed a marked increase with temperature (Fig. 1 E), parallel to that observed in the work of de-adhesion. This longer detachment distance was mainly due to the extraction of membrane tethers from the cell surface (31). During the initial pulling regime of the retraction curves, the force was mainly distributed among the bonds formed at the rim of the tip/cell contact area (37). After one bond of the rim failed, the rest were not able to withstand the force and broke simultaneously. After that point, mainly tethered bonds remained attached and the total force was equal to the number of tethers times the force required to extract a tether at the applied retraction speed. This led to the low variability in the magnitude of the jumps in force (tether forces) preceded by a force plateau (Fig. 2). It has been suggested that tether extraction in leukocytes requires the release of the adhesion receptor from the cytoskeleton (40,44). We observed that the probability of tether extraction was ∼5-fold higher at 37°C than at 16°C, suggesting that LFA-1 interaction with the cytoskeleton was weaker at higher temperature. Thus, at 37°C, cells were more prone to form tethers, which were also greater in length. To investigate the mechanism responsible for this behavior, we analyzed the lifetime and force steps of individual tethers. Tether forces are independent of the specific binding between the AFM tip and the membrane, and have been described as being mainly due to the friction of the membrane with membrane and lipid-binding proteins anchored to the cytoskeleton, which hinder the flow of lipids (45,46). Thus, a change in the membrane-cytoskeleton interaction, either through direct adhesion or through membrane proteins, would induce a change in the tether forces. It is likely that the known change in membrane rigidity by temperature (34) leads to modulation of the interaction between the membrane and the cytoskeleton, decreasing friction and subsequently the tether forces. The tether lifetime is governed by the receptor-ligand interaction at the force supported by the bond (14,47). As observed in Fig. 2 B, tether forces significantly decreased with temperature, whereas tether lifetimes had an opposite trend, being markedly higher at 37°C (Fig. 2 C). The decrease in the tether force suggests that temperature weakened the interaction between the membrane and the cytoskeleton (48). During tether extraction, the tether force was applied to the LFA-1/ICAM-1 bonds that linked the tether to the AFM tip. As a consequence of the lower tether force at high temperature, and given the dependence of bond lifetime on applied force (Eq. S2), the tethers lasted longer at 37°C. Shao and coworkers (40) showed that for tether extraction to occur, a certain force level must be overcome. This is assumed to depend on the interaction between the membrane receptor and the cytoskeleton, which in turns depends on the applied loading rate (44). Our results suggest, therefore, that more tethers were extracted at 37°C (i.e., the threshold force was lower) because the applied loading rate also decreased. The overall effect of tether force decrease and lifetime increase favored the total work of de-adhesion. The average lifetimes extracted from the individual tethers represent a measure of the bond lifetime at the applied (tether) force. Thus, we included the tether lifetimes measured at the corresponding tether forces (Fig. 5, E and F) in the lifetime versus force plots in Fig. 3 C. It is interesting to note that the tether lifetimes were systematically higher than those predicted from single-molecule measurements at the corresponding temperatures. This discrepancy may be due to several factors, such as multiple bonds per tether, additional transition states along the dissociation pathway, rebinding, cooperativity, loading rate history, or changes in the effective spring constant of the system during tethering (14,49,50). In any case, tether extraction appears to be an important mechanism by which cell adhesion is enhanced by temperature.

The loading rate measured from single-molecule measurements decreased markedly with temperature (Fig. 3 B, right). Since the retraction velocity and the spring constant of the cantilever were kept constant, the observed decrease must have been due to reduced stiffness of the cell. Indeed, the Young's modulus computed from the approaching force curves confirmed this supposition (Fig. 4), showing a dramatic ∼14-fold decrease from 16°C to 37°C. Previous studies observed similar elastic behavior using a number of techniques on different cell types (11,51,52). Of interest, other studies indicated an opposite effect of temperature in adherent cells grown on culture dishes (53–55). In those studies, the authors found an increase in the Young's modulus of the cells at higher temperatures. However, it should be noted that these measurements were performed on firmly adhered cells, in which a more active cytoskeletal machinery may be able to increase cellular prestress and thus cellular stiffness (56). This was not the case in our measurements, where cells were only slightly immobilized and conserved their round geometry (Fig. 1, Fig. S4, and Table S1).

It is worth mentioning the expected effect of increasing the retraction velocity to the adhesion parameters. First, adhesion forces per perimeter are expected to increase logarithmically at the range of loading rates applied in our measurements (57). Second, tether forces are known to increase linearly with extraction velocity. Third, at higher tether forces, tether lifetimes would decrease exponentially (Eq. S2), leading to a decrease in the maximum detachment distance. The combination of these responses would lead to a moderate increase of the apparent adhesion energy density.

Based on the above results, we concluded that temperature-enhanced cell adhesion was due to two main factors: reduced cell stiffness and lower linkage between the plasma membrane and the cytoskeleton at higher temperatures. The former will enhance the number of bonds formed, whereas the latter will favor the extraction of long membrane tethers, and both effects will increase the work required to detach the two surfaces. To confirm this mechanistic view, we conducted the same measurements on cytD-treated cells. CytD is a drug that inhibits polymerization of actin, the main component of the cytoskeleton that is responsible for the structural stability and viscoelastic properties of cells. Leukocytes have an actin cortex that interacts with the plasma membrane via membrane and lipid-binding proteins (58). Thus, the effect of actin cytoskeleton destabilization by cytD was expected to have two important repercussions in cell adhesion. First, it would induce softening of the cells, enhancing their deformability and increasing the contact area between the adhesive surfaces. Second, it would limit membrane-cytoskeleton interaction, favoring the formation of membrane tethers. In fact, the Young's modulus of cytD-treated cells dropped at the three tested temperatures (Fig. 5 G) and, of interest, remained unaffected by temperature. As expected, the membrane-cytoskeleton interaction was also reduced, as revealed by a significant drop in tether forces, which was more pronounced at the two lower temperatures (Fig. 5 E). As a result, tether lifetimes significantly increased at the two lowest temperatures, but not at 37°C (Fig. 5 F). This suggests that at 37°C, the membrane is already loosely linked to the cytoskeleton, which will favor the known tethering and rolling behavior of leukocytes (59).

At this point, it is worth mentioning the possible effect of a change in membrane properties on tether extraction. Hackl and coworkers (34) reported decreased membrane rigidity and increased access area and flickering of giant vesicles with increasing temperature. Our observation that temperature had a small effect on tether extraction in cytD-treated cells may suggest that the contribution of the membrane elastic properties was small. However, a softer membrane with higher flickering amplitude may modulate the membrane-cytoskeleton interaction, which would explain the enhanced tether formation and decreased tether forces on untreated cells at 37°C. Thus, further disruption of the cytoskeleton by cytD would have a small effect on a soft membrane that is already loosely linked to the cytoskeleton at 37°C, but an important effect on a rigid and tightly linked membrane at lower temperatures, as we observed. Nevertheless, further measurements are required to assess the contribution of membrane elastic properties to tether formation.

Given these results, we expected to observe a significant increase in the work of de-adhesion on cytD-treated cells, especially at the lower temperatures. After treatment with cytD, the work of de-adhesion increased by ∼2-fold at 16°C and 24°C, where cytD had greater effect on cell elasticity and tether force, whereas an opposite effect was observed at 37°C. The overall effect of cytD on the work of de-adhesion did not reach statistical significance (p = 0.76). However, cytD is known to inhibit cellular adhesion (6), as confirmed by the dramatic drop found in the apparent adhesion energy density and adhesion force per perimeter (Fig. 5 A). This effect on the apparent adhesion energy density suggests that cytD may impair the adhesion capacity of integrins. This hypothesis was corroborated by the observation that a significant percentage of cells (10–15%) showed no adhesion after treatment with cytD. This impaired binding may be due to a possible change in the distribution of integrins within the cell surface, reduced diffusion, or possible changes in cell surface topography (59,60). Nevertheless, we concluded from these measurements that temperature had a much less pronounced effect on the cell adhesion parameters of the cytD-treated cells, because both cellular elasticity and membrane/cytoskeleton binding remained unaffected.

Conventional adhesion assays normally involve the application of a constant force to allow for the initial adhesion of suspended cells onto a protein-coated substrate. The cells are given a certain amount of time to adhere, and then an opposite force is applied to induce detachment. The adhesion is then quantified from the relative number of cells that remained attached. Thus, the observed decrease in cellular elasticity at 37°C would enhance the area of contact between cells and surface, increasing the number of bonds formed and augmenting the initial adhesion capacity (9). In addition, the higher compliance of the cells would induce a lower loading rate, allowing the cells to deform more and remain in contact longer, whereas a lower membrane-cytoskeleton interaction would increase the probability of forming longer tethers. This would prolong the time the cells remain in contact with the surface, favoring reformation of bonds. As an overall result, more cells would remain attached to the surface at higher temperatures. Caputo and Hammer (18) studied the contribution of microvilli deformability to cellular rolling. They found that tether viscosity had an optimal value to favor rolling, and that more deformable microvilli enhanced adhesion. Although our stiffness measurements involved large-scale deformations, they are in agreement with Caputo and Hammer's conclusions regarding the importance of cellular mechanics and membrane-cytoskeleton interactions for the modulation of cell adhesion.

In summary, temperature-modulated cell adhesion is governed by the regulation of cellular elasticity and membrane tether formation. Both mechanisms play a crucial role at different stages of the process. Upon binding, the reduced cell stiffness enables the formation of more bonds by increasing the area of contact between the surfaces. Upon detachment, lower loading rates and enhanced membrane tether formation prolong the time the two surfaces remain in contact and increase the energy required for detachment. Alternatively, we conclude that a reduction of temperature suppresses cell adhesion and consequently immune functions. Our study thus provides insight into the biophysical mechanisms by which reduced temperatures contribute to controlling swelling and inflammation.