Regulation and dynamics of force transmission at individual cell-matrix adhesion bonds

Abstract

Integrin-based adhesion complexes link the cytoskeleton to the extracellular matrix (ECM) and are central to the construction of multicellular animal tissues. How biological function emerges from the tens to thousands of proteins present within a single adhesion complex remains unclear. We used fluorescent molecular tension sensors to visualize force transmission by individual integrins in living cells. These measurements revealed an underlying functional modularity in which integrin class controlled adhesion size and ECM ligand specificity, while the number and type of connections between integrins and F-actin determined the force per individual integrin. In addition, we found that most integrins existed in a state of near-mechanical equilibrium, a result not predicted by existing models of cytoskeletal force transduction. A revised model that includes reversible cross-links within the F-actin network can account for this result and suggests one means by which cellular mechanical homeostasis can arise at the molecular level.

Fig. 1. Single-molecule tension measurements in living cells reveal distinct subpopulations of load-bearing integrins.

(A) FRET-based MTSs. MTSs are attached to the coverslip surface via the HaloTag domain. (B) FRET-force calibration curves for MTS_low (blue) and MTS_high (purple) (16, 51). (C) Representative images showing green fluorescent protein (eGFP) (left), FRET donor (middle), and acceptor channels (right) for HFFs adhering to a surface functionalized with MTS_low. Scale bar, 5 μm; inset scale bar, 2 μm. (D) Example intensity traces (left) for the FRET donor (green) and acceptor (orange). Vertical dashed lines delineate frames during which the acceptor dye was directly excited with 633-nm light; arrows mark acceptor or donor bleaching; horizontal gray dashed lines indicate upper and lower force measurement limits. Right: Corresponding load time series before acceptor photobleaching (light blue). Intensity, arbitrary units (a.u.). (E) Single-molecule load distributions for MTS_high underneath cells, within adhesions, and outside adhesions. N = number of cells, n = number of sensors. (F) Combined single-molecule load distributions for MTS_low and MTS_high sensors underneath cells, within adhesions, and outside adhesions for MTS_low [blue; data from (14)] and MTS_high (purple).

Fig. 2. Integrin class does not determine the force per integrin but affects ligand specificity.

(A) Images of eGFP-paxillin and ensemble FRET maps for pKO-α_v, pKO-α_v/β₁, and pKO-β₁ cells adhering to MTS_low (top), MTS_high (middle), and MTS_FN9–10 (bottom). Insets show corresponding bright-field images for pKO-β₁ cells, which rarely spread on surfaces functionalized with MTS_low and MTS_high. Scale bars, 10 μm. (B) Ensemble quantification of pKO-αv, pKO-αv/β1, and pKO-β1 cells adhering to MTS_low, MTS_high, and MTS_FN9–10. When adhering to MTS_low, pKO-α_v/β₁ cells exert more integrated traction compared to pKO-α_v cells (pKO-α_v: 67 cells, mean: 5.5 nN; pKO-α_v/β₁: 43 cells, mean: 9.8 nN) (***P = 3 × 10⁻⁴). When adhering to MTS_high, pKO-α_v/β₁ and pKO-α_v cells produce comparable traction overall (pKO-α_v: 77 cells, mean: 2.6 nN; pKO-α_v/β₁: 22 cells, mean: 7 nN). For MTS_FN9–10, pKO-α_v/β₁ and pKO-β₁ cells exert a higher integrated traction as compared to pKO-α_v cells (pKO-α_v: 13 cells, mean: 7.9 nN; pKO-α_v/β₁: 12 cells, mean: 26.9 nN; pKO-α_v/β₁: 12 cells, mean: 23.8 nN) (***P < 10 to 3). (C) Single-molecule load distributions for pKO cell lines adhering to MTS_FN9–10. Black bars indicate unbound molecules. (D) Adhesion area measured for pKO-α_v, pKO-α_v/β₁, and pKO-β₁ cells adhering to MTS_FN9–10. Areas were calculated from the thresholded eGFP-paxillin signal. Differences in adhesion area were not significant between the three cell types.

Fig. 3. Vinculin is required for higher forces per integrin.

(A) Ensemble FRET maps for WT and vin^−/− MEFs transfected with eGFP-paxillin and seeded on coverslips functionalized with MTS_low and MTS_high sensors. Scale bar, 10 μm. (B) Total integrated traction per cell for forces <7 pN measured with MTS_low. Open circles indicate the mean value. (WT: 96 cells, mean: 3.6 nN; vin^−/−: 89 cells, mean: 4.4 nN.) (C) Total integrated traction per cell for forces between 7 and 11 pN measured with MTS_high. Open circles indicate the mean. (WT: 71 cells, mean: 8.7 nN; vin^−/−: 99 cells, mean: 1.5 nN.) (D) Histograms of the single-molecule load measurements for WT and vin^−/− MEFs measured for cells adhering to MTS_low for sensors outside adhesions (left) and within adhesions (right). ***P < 0.001 using two-sided Wilcoxon rank sum test.

Fig. 4. Dynamic transitions in load constitute a minority of sensor measurements.

(A) Representative traces showing step (left) and gradual ramp (right) load transitions (FRET donor: green; FRET acceptor: orange; load: blue) for HFFs adhering MTS_low. Black arrows mark acceptor or donor bleaching; dashed black lines indicate direct excitation of the FRET acceptor. Horizontal gray dashed lines indicate upper and lower force measurement limits for MTS_low. (B) Percentage low force (defined as <2 for MTS_low or < 7 pN for MTS_high) (blue), higher force but static (green), and dynamic (hashed; subset of loaded integrins) sensors for a variety of cell types adhering to different MTSs. (C) Percent of dynamic sensors with step (magenta) and ramp (purple) transitions. U2OS cells had no observable dynamic events.

Fig. 5. A modified model of cytoskeletal force transduction yields mechanical equilibrium at individual integrins.

(A) Simplified cartoon of a FA: Nonmuscle myosin II pulls on reversibly cross-linked actin filaments, which are linked to integrins by vinculin and talin. (B) Cytoskeletal dynamics model: F-actin filaments bind to anchors (blue) and are linked by cross-linking proteins (green). (C) An example force trace of the standard clutch model and possible clutch model extensions that account for multivalent clutch connections, viscous relaxation, or reversible cross-links. Reversible cross-links allow for stable force plateaus as well as sporadic ramp and step events. The dashed gray lines indicate zero force. (D) Calculated energy dissipation from simulations with irreversible (top) and reversible (bottom) cross-links. (E) Force distribution for simulated anchors with reversible cross-linking (k_x,off = 20 s⁻¹).