Control of cell membrane tension by myosin-I

Abstract

All cell functions that involve membrane deformation or a change in cell shape (e.g., endocytosis, exocytosis, cell motility, and cytokinesis) are regulated by membrane tension. While molecular contacts between the plasma membrane and the underlying actin cytoskeleton are known to make significant contributions to membrane tension, little is known about the molecules that mediate these interactions. We used an optical trap to directly probe the molecular determinants of membrane tension in isolated organelles and in living cells. Here, we show that class I myosins, a family of membrane-binding, actin-based motor proteins, mediate membrane/cytoskeleton adhesion and thus, make major contributions to membrane tension. These studies show that class I myosins directly control the mechanical properties of the cell membrane; they also position these motor proteins as master regulators of cellular events involving membrane deformation.

Plasma membrane tension in eukaryotic cells is a master regulator of all cellular processes that involve membrane deformation including endocytosis (1), exocytosis (2), membrane repair (3), cell motility (4), and cell spreading (5). The total tension present in the plasma membrane (i.e., the “apparent” membrane tension, T_App) has a minor contribution from the surface tension of the lipid bilayer (T_M) and a substantial contribution from the molecular contacts that afford adhesion to the underlying actin cytoskeleton (γ) (6). To prevent large changes in tension, the plasma membrane must maintain continuous interactions with the cytoskeleton, despite the fact that this structure is highly dynamic and continuously remodeling. However, little is known about the molecules that mediate dynamic interactions between membrane and cytoskeleton, or the proteins that contribute directly to controlling membrane tension in cells.

A striking biological example of the complex mechanical interplay between the plasma membrane and the supporting cytoskeleton is provided by the brush border found on the apex of intestinal epithelial cells (7). The brush border functions as the primary site for nutrient absorption and consists of up to 1,000 microvilli, protrusions that increase apical membrane surface area and release vesicles into the intestinal lumen (7, 8). Each microvillus is supported by a parallel bundle of actin filaments that enables this structure to extend several microns from the cell surface (7). To stabilize this convoluted morphology, epithelial cells must furnish the brush border with high levels of membrane-cytoskeleton adhesion energy. One candidate molecule for carrying out this task is myosin-1a (Myo1a), a monomeric actin-based motor protein that is present at high levels in the microvillus and is known to bind directly to phospholipids by virtue of its basic C-terminal tail homology 1 (TH1) domain (9, 10). Myo1a KO mice exhibit large apical membrane herniations that are morphologically similar to “blebs” (11, 12). Because blebs represent complete delamination of membrane from the actin cytoskeleton, these results suggest that Myo1a makes a major contribution to membrane-cytoskeleton adhesion in the brush border. While previous studies have suggested a role for class I myosins in the regulation of cortical stiffness (i.e., whole cell deformability) of Dictyostelium discoideum cells (13), the role of class I myosins in the control of membrane tension has not been explored. Thus, the goal of the current study was to determine whether class I myosins function in controlling the mechanical properties of the plasma membrane.

Results

Probing Membrane Tension with an Optical Trap.

We sought to investigate the contribution of Myo1a and other class I myosins to plasma membrane tension in isolated organelles and living cells. To this end, we developed an optical trap assay that enabled us to measure the force exerted by a thin tubule or “tether” extracted from a membrane (14). In a typical tether force experiment, a concanavalin-A-coated 2.0 μm diameter microsphere was captured in the optical trap and then brought in contact with an isolated brush border or intact cell, which was firmly attached to a glass coverslip surface. Membrane tethers were then formed by translating the piezoelectric stage to move the sample away from the trapped bead. Forces exerted by membrane tethers on the bead were derived from microsphere position data (15), acquired at video rate using a CCD camera; position data were converted to force using the stiffness of the optical trap (k_Trap), which for our studies ranged from 0.1–0.4 pN/nm (Fig. S1).

Myo1a KO Brush Borders Demonstrate Defects in Membrane Force-Extension.

As a first step, we examined the force-extension properties of the apical membrane associated with brush borders isolated from WT or Myo1a KO mouse small intestine. Because isolated brush borders are prepared in the absence of ATP, tether formation in this case is an irreversible process. The first step in these experiments was to capture a ConA-coated bead with the optical trap and bring it in contact with a coverslip-adsorbed brush border. After waiting approximately 4 s to allow for bead binding to the apical membrane, tethers were pulled by translating samples away from the trapped bead at 1.0 μm/s (Fig. 1); recordings ended once the bead escaped from the trap. Under these conditions we observed that the slope of the membrane force-extension curve is significantly higher in WT brush borders relative to Myo1a KO samples (0.43 ± 0.01 vs. 0.05 ± 0.01 pN/nm, respectively; Fig. 1E). This approach also revealed that Myo1a limits the maximum length of membrane tether that we were able to extract during these experiments (WT, 3.5 ± 0.3 vs. KO, 7.8 ± 2.0 μm; Fig. 1F). Because these measurements were performed in the absence of active membrane trafficking or other potentially confounding subcellular activities, they clearly indicate that Myo1a makes a direct contribution to the mechanical stability of the brush border apical membrane.

Fig. 1.

(A) Confocal micrograph of a membrane tether pulled from a single isolated brush border labeled with Alexa488-Concanavalin A. Image is inverted and contrast enhanced to enable visualization of the extremely dim membrane tether. (B) Phalloidin signal from the isolated brush border shown in a reveals that the tether is devoid of F-actin. The position of the trapped bead and membrane tether are indicated; bars in A and B are 2 μm. (C) Merge of images from A and B (shown without contrast enhancement) demonstrates the colocalization of membrane (green) and F-actin (red) signals; panel on the right shows the orientation of microvillar actin bundles in this structure. (D) Cartoon depicting the polarity of microvillar actin bundles of the brush border imaged in A–C. (E) Force-extension records for membrane tethers extracted from WT (green) and Myo1a KO (red) brush borders. Linear fitting of raw data over the first μm of extension yields spring constants of 0.43 ± 0.01 pN/nm (n = 6, R² = 0.99) and 0.05 ± 0.01 pN/nm (n = 7, R² = 0.92) for WT and KO brush borders, respectively. (F) Box-plots of maximum tether lengths demonstrate that Myo1a KO brush borders released significantly longer tethers (7.8 ± 2.0 μm; red) relative to WT controls (3.50 ± 0.34 μm; green). *P < 0.05.

Myo1a Controls T_App in Living Epithelial Cells.

We next sought to determine whether Myo1a plays a role in regulating apical membrane mechanics in the context of living, polarized epithelial cells. In intact cells, the force exerted by a membrane tether held at constant length is directly related to the level of apparent membrane tension (6). Importantly, tether formation in this case is a reversible process; after forming a tether, release of the trapped microsphere allows the cell to resorb the extracted membrane. We carried out tether force measurements using the colonic adenocarcinoma cell line, NGI3 (16). Upon differentiation, NGI3 cells express endogenous Myo1a and build an elaborate brush border with densely packed microvilli. For these experiments, an optical trap was used to capture a ConA-coated bead and bring it into contact with a surface-adsorbed NGI3 cell to enable binding as described above. To form a membrane tether, the cell was translated away from the trapped bead at a constant rate of 1 μm/s. Tethers were pulled to a length of 5 μm as our initial experiments with NGI3 cells revealed that tether force is independent of length in this regime of extension (Fig. S2), in a manner similar to previous results with other cell lines (17). To probe the contribution of Myo1a to apparent membrane tension, we perturbed the endogenous population of this motor using 2 methods: (1) expression of an EGFP-tagged Myo1a-TH1 dominant negative construct, which disrupts the targeting of endogenous Myo1a and gives rise to cellular phenotypes similar to those observed in Myo1a KO mice (11, 18), and (2) expression of an EGFP-tagged full length Myo1a construct to supplement the population of endogenous Myo1a (19) (Fig. S3). Strikingly, expression of the EGFP-Myo1a-TH1 dominant negative in NGI3 cells dramatically reduced the force observed during the tethered phase of individual records, relative to cells expressing EGFP as a negative control (17.9 ± 5.1 vs. 32.1 ± 4.6 pN; Figs. 2 A and B). In contrast, expressing EGFP-Myo1a in NGI3 cells gave rise to an increase in tether force relative to control cells (42.5 ± 4.2 pN; Fig. 2 A and B). Thus, in the context of live epithelial cells, Myo1a enables the apical membrane to resist deformation (i.e., tether formation) and makes a substantial contribution to apparent membrane tension (T_App).

Fig. 2.

(A) Representative tether force records from NGI3 cells transfected with EGFP (green), EGFP-Myo1a-TH1 (red), or EGFP-Myo1a (blue). The region of the record marked as “Tethered” corresponds to the quasi-stable equilibrium force achieved after piezo-stage translation has stopped and tether formation is complete. (B) Bar graphs of the average force observed during the tethered phase of individual records reveal that EGFP-Myo1a-TH1 dominant-negative expression lowers tether force (17.9 ± 5.1 pN, n = 14), while expression of EGFP-Myo1a increases tether force (42.5 ± 4.2 pN, n = 11) relative to EGFP-expressing control cells (32.1 ± 4.6 pN, n = 15). *P < 0.05, **P < 0.005. (C) Model representation of the correspondence between tether force measurements and the molecular level perturbations induced by expressing EGFP-Myo1a or EGFP-Myo1a-TH1, relative to the EGFP negative control.

Myo1a Contributes to Membrane-Cytoskeleton Adhesion (γ) in Living Epithelial Cells and Fibroblasts.

The results observed in NGI3 cells could be explained in 1 of 2 ways: (1) Myo1a could have a direct impact on membrane surface tension (T_M), or (2) Myo1a could make a significant contribution to membrane-cytoskeleton adhesion energy (γ) by physically linking the membrane to the actin cytoskeleton. With its combined membrane- and actin-binding activities, Myo1a is ideally suited for mediating to membrane-cytoskeleton adhesion; decreasing or increasing the number of functional Myo1a molecules per unit area of membrane with the expression of EGFP-Myo1a-TH1 or EGFP-Myo1a, respectively, would give rise to a corresponding reduction or elevation in adhesion energy, and thus apparent membrane tension (Fig. 2C). To further test this hypothesis, we performed tether force measurements under conditions that favored the formation of multiple membrane tethers (20, 21). When multiple membrane tethers are simultaneously pulled from the same local region of membrane, they demonstrate a tendency to coalesce into a single tether in the absence of “pinning” forces (22), for example, forces provided by molecular links to the underlying cytoskeleton. Thus, assaying the number of tethers formed from a single microsphere/cell encounter provides a read-out on the density of molecular contacts between the membrane and cytoskeleton. Increasing the contact area between the microsphere and cell, or increasing the loading rate (by raising trap stiffness) biased events toward the formation of multiple tethers (23) (Fig. S4A and B). Multiple tether formation was confirmed using DIC microscopy (Fig. S5A). The presence of multiple tethers was also indicated by the appearance of “staircases” during the tethered phase of force records (Fig. 3A). The rapid drops or “steps” between discrete plateaus in force are the result of adjacent membrane tether coalescence (22, 24). Similar to previous studies of multiple tether mechanics (20, 25), observed force steps appear as integer multiples of the force measured for single tethers (Fig. S5B and C). Intriguingly, visual inspection of force records revealed that expression of the EGFP-Myo1a-TH1 dominant negative significantly impaired the cell's ability to stabilize multiple tethers. This was indicated by 2 important observations. First, the tethered phase of records from TH1-expressing cells appeared to start at a lower level of force (Fig. 3A). We confirmed this by calculating force-time integrals across the tethered phase of individual records. EGFP-Myo1a-TH1 expression significantly reduced the mean force-time integral, suggesting a reduction in the total number of tethers that contribute to force during the tethered phase (Fig. 3D). Second, when the number of observable steps (i.e., coalescence events) was tallied from force records, TH1-expressing cells demonstrated a significantly reduced average number of events per record (Fig. 3 B and C).

Fig. 3.

(A) Representative multiple tether force records from NGI3 cells transfected with EGFP (green) or EGFP-Myo1a-TH1 (red). Arrows indicate the beginning of the tethered phase in each case. Rapid drops in force during the tethered phase correspond to tether coalescence events. (B) Multiple representative examples of the tethered phases from records of NGI3 cells expressing EGFP (green) and EGFP-Myo1a-TH1 (red). Data were plotted with a fixed offset of 25 pN between all records. Visual inspection reveals that EGFP-Myo1a-TH1 records contain fewer observable coalescence events relative to EGFP control records of the same duration. (C) Box plots show the mean number of coalescence events or “steps” observed in EGFP or EGFP-Myo1a-TH1 records; cells expressing the TH1 dominant negative demonstrate significantly fewer coalescence events, relative to control cells (*P < 0.05). (D) Box plots of the force-time integrals calculated from the tethered phase of individual records; EGFP-Myo1a-TH1 expressing cells demonstrate significantly lower force-time integrals relative to EGFP-expressing controls (**P < 0.001). (E) Ensemble averages of tethered phase data from NGI3 cells transfected with EGFP (green) and EGFP-Myo1a-TH1 (red); open circles show average values at 0.3-s intervals; solid lines are single exponential fits to averaged data. Similar force decay kinetics were observed for EGFP (0.12 ± 0.01 s⁻¹, n = 28, R² = 0.99) and EGFP-Myo1a-TH1 (0.13 ± 0.01 s⁻¹, n = 30, R² = 0.98), respectively. (F) Model of the mechanism underlying the formation of multiple tethers in cells expressing EGFP (negative control) or EGFP-Myo1a-TH1 (dominant negative). These cartoons represent “snap shots” in the records shown in A, taken at the beginning of the tethered phases at the time point indicated by the black arrows.

As discussed above, the force steps observed during multiple tether events are integer multiples of a unitary tether force (∼30 pN in the case of NGI3 cells, Fig. 2 and Fig. S5B), suggesting that each step represents the same underlying physical process, that is, tether coalescence. Because the coalescence of membrane tethers is due to the failure of bonds that link the membrane and underlying cytoskeleton, we expect the irreversible transition from n tethers to n-1 tethers to proceed as a first-order process with a rate equal to approximately 1/lifetime of a single tether (26). Indeed, ensemble averaging of the tethered phases from individual EGFP and EGFP-Myo1a-TH1 records revealed single-exponential decays in force for both data sets (Fig. 3E). Fits to these data revealed comparable kinetics (EGFP, 0.12 ± 0.01 s⁻¹; EGFP-Myo1a-TH1, 0.13 ± 0.01 s⁻¹) indicating that a similar molecular process controls the rate of tether coalescence in EGFP and EGFP-Myo1a-TH1-expressing cells. We propose that Myo1a plays a role in stabilizing multiple membrane tethers in both cases. However, in cells expressing the dominant negative, a large fraction of the endogenous Myo1a population is displaced from the plasma membrane; this reduces its effective contribution to membrane-cytoskeleton adhesion and ultimately impairs the cell's ability to form and stabilize multiple tethers (Fig. 3F).

Based on the model outlined above, one prediction is that the expression of EGFP-Myo1a in NGI3 cells should enhance the cell's capacity for stabilizing multiple tethers. While we attempted multiple tether experiments with NGI3 cells expressing EGFP-Myo1a, our ability to extract tethers from these cells was dramatically decreased. If Myo1a does contribute to membrane-cytoskeleton adhesion (γ), then this observation might reflect the expected outcome. Because NGI3 cells express endogenous Myo1a, transfection with EGFP-Myo1a creates an over-expression scenario that could give rise to exaggerated membrane-cytoskeleton adhesion. We suspect that the extraction of multiple tethers is precluded in this case, as the forces required are beyond the upper limits of our optical trap. In light of this possibility, we examined the impact of EGFP-Myo1a expression on multiple tether formation from cells that express low levels of endogenous Myo1a. For these experiments, we used NIH 3T3 fibroblasts, which do not express detectable Myo1a and exhibit a low apparent membrane tension (∼25% of T_App for NGI3, Fig. S4C) (5). Multiple tether force measurements in NIH 3T3 cells produced step-containing records as observed in NGI3 cells described above. Inspection of raw data and analysis of force-time integrals revealed that expression of EGFP-Myo1a in NIH 3T3 cells produces a modest increase in the force observed during the tethered phase (Fig. 4, royal blue data). These results show that Myo1a expression is capable of increasing membrane-cytoskeleton adhesion outside the context of the polarized cytoskeleton found in NGI3 cells.

Fig. 4.

(A) Representative tether force records for NIH 3T3 fibroblasts expressing EGFP (green) or 1 of 5 different EGFP-tagged class I myosins (as labeled on the plot). (B) Box plots of force-time integrals calculated over the tethered phases of individual records from NIH 3T3 cells expressing the various class I myosin constructs (*P < 0.05, **P < 0.001). Cells expressing myosin-I constructs uniformly demonstrate higher force-time integrals relative to EGFP expressing controls. (C) Ensemble averages of tethered phases from records of NIH 3T3 cells transfected with various class I myosin constructs (open circles); solid lines show single exponential fits to the data. Rate constants obtained from exponential fits are as follows: EGFP, 0.04 ± 0.01 s⁻¹, n = 23, R² = 0.97; EGFP-Myo1a, 0.17 ± 0.01 s⁻¹, n = 21, R² = 0.99; EGFP-Myo1b, 0.06 ± 0.01 s⁻¹, n = 15, R² = 0.97; EGFP-Myo1c, 0.13 ± 0.01 s⁻¹, n = 10, R² = 0.91; EGFP-Myo1d, 0.20 ± 0.01 s⁻¹, n = 9, R² = 0.96; EGFP-Myo1e, 0.28 ± 0.01 s⁻¹, n = 5, R² = 0.98.

Interestingly, the force decay kinetics in NIH 3T3 cells expressing EGFP-Myo1a were comparable to those observed in experiments with NGI3 cells (Fig. 4C vs. Fig. 3E; 0.17 ± 0.01 s⁻¹ vs. 0.12 ± 0.01 s⁻¹, respectively), which normally express significant levels of endogenous Myo1a. This kinetic similarity suggests that our tether force measurements are in fact probing the mechanical contributions made by Myo1a in the case of both experiments.

Control of Membrane Mechanics May Be a Class-Wide Function for Class I Myosins.

The results summarized to this point indicate that Myo1a controls apparent membrane tension (T_App) by contributing to membrane-cytoskeleton adhesion (γ). Given that all 8 vertebrate class I myosins contain a basic TH1 domain (27), which mediates interactions with cellular membranes (28), we sought to determine whether other myosin-I isoforms could also contribute to membrane-cytoskeleton adhesion. To this end, we expressed EGFP-tagged versions of 3 other short-tailed (Myo1b, Myo1c, and Myo1d) and 1 long-tailed (Myo1e) class I myosins in NIH 3T3 cells (Fig. S6) and probed their impact on multiple tether formation as described above. Inspection of raw data records qualitatively revealed that regardless of the isoform, class I myosin expression increased the force measured during the tethered phase relative to EGFP-expressing controls (Fig. 4A). This was confirmed through quantitative analysis of multiple tether records by calculating average force-time integrals (Fig. 4B) and producing ensemble averages of data (Fig. 4C) for each construct as outlined above (see Fig. 3). Of note here is the fact that Myo1e, the only long-tailed isoform included in our analysis, had the most dramatic impact on multiple tether formation. The uniform increase in multiple tether formation observed with the expression of different myosin I isoforms, suggests that membrane-cytoskeleton adhesion and the control of plasma membrane tension may be general functions for all vertebrate class I myosins.

Discussion

Experiments performed in multiple eukaryotic model systems have implicated class I myosins in various aspects of membrane-related events including phagocytosis (29–33), endocytosis (34–37), exocytosis (38, 39), and membrane recycling (40). Although our current data set does not allow us to rule out the possibility that perturbations in membrane trafficking may contribute to the changes in membrane tension observed in our experiments, our results do provide strong support for a model where class I myosins play a direct role in the control of membrane tension, by contributing to adhesion between the plasma membrane and underlying actin cytoskeleton. Indeed, mechanical measurements with isolated brush borders revealed that Myo1a increases apical membrane force-extension stiffness approximately 10-fold (Fig. 1E). Because these experiments were performed in the absence of ATP, potentially confounding processes such as membrane trafficking were not active during these in vitro measurements and had little impact on the observed mechanical responses. This model finds additional support when we consider that disrupting Myo1a function in polarized epithelial (NGI3) cells reduced the mean tether force by approximately 50% (Fig. 2 A and B). Apparent membrane tension (T_App), considered to be the sum of in-plane tension (T_m) and membrane/cytoskeleton adhesion (γ), is directly related to the tether force (F_Tether): T_App = T_m + γ = (F_Tether)²/(8Bπ²), where B is the membrane bending stiffness (41). Thus, perturbation of Myo1a reduced apparent membrane tension by approximately 70%. This value approaches previously published estimates that attribute over 75% of apparent membrane tension to membrane-cytoskeleton adhesion (6). Finally, analysis of multiple tether formation provides some of the most direct support for this model. Expression of the Myo1a TH1 dominant negative decreased the ability of NGI3 cells to support and stabilize multiple membrane tethers, whereas over-expression of Myo1a or other class I myosins stabilized multiple membrane tethers (Figs. 3 and 4). Because the ability to form multiple tethers is directly linked to the density of molecular contacts between the membrane and cytoskeleton, these results tell us that class I myosins are important players in mediating these interactions. Thus, the results presented here strongly support a model where class I myosins play a direct role in the control of membrane tension, by contributing to adhesion between the plasma membrane and underlying actin cytoskeleton.

The mechanical measurements presented here provide a physical explanation for the phenotypes observed in the Myo1a KO mouse (11). Among the most striking defects observed in this model are herniations of apical membrane that extend from the apical surface of KO enterocytes. In most cell types cytosolic fluid pressure, created by myosin-II powered contractility in the cell cortex, exerts a positive (i.e., outward) force on the plasma membrane (42). In the enterocyte, the high levels of membrane-cytoskeleton adhesion provided by the microvillar population of Myo1a function to counter cytosolic pressure so that the brush border can stabilize the enormous quantity of plasma membrane packed into this domain.

In addition to providing access to information about membrane-cytoskeleton adhesion, the multiple tether experiments described here may provide important mechanistic information on the formation of “tethers” under normal physiological conditions. As an example, leukocytes rolling along endothelium extrude multiple membrane tethers to stabilize their rolling velocities, ultimately enabling arrest and extravasation (25). Thus, one goal for future studies will be to determine whether the class I myosins expressed in leukocytes, play a role in the formation and stabilization of these important membrane structures.

While the importance of the actin cytoskeleton in shaping the plasma membrane and its mechanical properties is well established (14), the results described here show that actin-based motors, and specifically class I myosins, play a role in controlling the mechanical interactions between these 2 systems. Class I molecules are ideal candidates for fulfilling this function within cells. Cryo-electron microscopy studies have established that Myo1a is an extended molecule arranged with the actin-binding motor domain at 1 end, and the putative lipid interacting domain at the other (43). This domain arrangement is well suited for the bivalent crosslinking of plasma membrane to actin filaments, while maintaining an approximate 15-nm (projected length of Myo1a) gap between these 2 compartments. In addition to domain organization, specific mechanochemical properties appear to be tuned to enable these motors to contribute to membrane tension (44). Recent single molecule studies indicate that the activity of myosin-1b is exquisitely sensitive to opposing external load (45). These studies show that the rate of ADP release and thus detachment from actin slows down approximately 100-fold in response to loads as small as 2 pN (45). Load-dependent kinetics may enable class I myosins to function in membrane/cytoskeleton adhesion by allowing them to remain strongly bound to F-actin for long periods, without hydrolyzing ATP.

Although the detailed kinetics of myosin-I/actin interactions are in many cases well-characterized (44), less is known about the mechanism underlying myosin-I membrane binding. While it was established many years ago that Myo1a binds to acidic phospholipids via its basic TH1 domain (10), more recent studies have revealed that vertebrate and Acanthamoeba class I myosin TH1 domains contain a PH motif able to bind tightly to highly charged, acidic phosphoinositides such as PIP₂ (46, 47). These studies may help explain earlier biophysical data implicating PIP₂ in membrane/cytoskeleton adhesion (48). However, PIP₂ is estimated to comprise <1% of total phospholipid found in the inner leaflet of the plasma membrane (49). Thus, in domains such as the brush border, where membrane/cytoskeleton adhesion must be high to maintain a complex morphology, higher abundance lipids (e.g., phosphatidylserine) and alternate, unexplored lipid binding motifs within the TH1 domain are likely to play a role. Finally, because TH1 domains have been identified in myosin-I genes from the earliest eukaryotes (27), the control of membrane tension may represent an ancient and conserved function for these molecular motors.

Materials and Methods

Optical Trap Instrumentation.

Our optical trap is built around a single-mode diode-pumped solid-state Nd:YVO₄ laser (LG Laser Technologies; TEM₀₀, 3 W, λ = 1064 nm) that is coupled to a Nikon TE-2000-U inverted light microscope via optics that are housed in central unit from Molecular Machines & Industries. In addition to the laser head, the central unit contains beam-conditioning optics, z axis focusing lenses, and 2 galvanometer-mounted mirrors for the control of beam position. We used a Nikon PlanFluor 100x/1.3 lens (72% transmission in the IR) to focus the laser and form a trap at the focal place. A motorized X-Y scanning stage (Marzhauser) was used for course control of sample position. A piezoelectric stage insert (Mad City Labs) provided high-resolution position control with subnanometer accuracy. Scanning stage position, laser power, and laser focal depth (z axis trap position) were all under computer via software provided by MMI. Images of trapped particles were captured with a Sony Exwave HAD color CCD using a National Instruments PCI-1410 image acquisition card with custom software written in LabVIEW 8.5. Time-lapse images acquired at video rate were used to obtain the position of trapped particles with software developed by Carter et al. (15). Trap stiffness calibration was performed by measuring the excursion of a trapped bead in response to different viscous drag forces applied by moving the flow chamber with the piezoelectric stage. Stokes' law (6πηrv = k_Trap x; where v is solution velocity, η is coefficient of solution viscosity, r is the radius of the microsphere, and x is microsphere displacement from trap center) was used to calculate the value of k_Trap for our experiments (Fig. S1).

Brush Border Isolation and Manipulation.

Brush borders were isolated using previously described protocol (50). All procedures involving animals were performed under the protocols prescribed by the Vanderbilt University Medical Center Institutional Animal Care and Use Committee. For membrane tether studies, brush borders were typically transferred into a flow cell assembled with a glass slide, a coverslip and 2 pieces of doubled-sided tape. Several volumes of buffer were applied to the flow cell to flush out brush borders that were only loosely anchored to the glass; tether experiments were then carried out in 75 mM KCl, 10 mM imidazole, 1 mM EGTA, 2.5 mM MgCl₂, and 0.01% Na-Azide, pH 7.2.

Cell Culture and Transfections.

NGI3 and NIH 3T3 cells were cultured on coverslips in 6-well plates at 37°C, with 5% CO₂. Culture medium consisted of DMEM (Invitrogen) supplemented with 10% fetal bovine serum (HyClone) and 2 mM glutamine (GibcoBRL). Transfections were carried out in 6-well plates with 6–8 μg DNA per 2.5 mL plating media using the reagent Lipofectamine2000 (Invitrogen). Cells were assayed 2–3 days post transfection. In a typical experiment, cell-coated coverslips were assembled into a flow chamber using double-sided tape and a glass slide. For all experiments, cells were incubated in CO₂-independent medium (Invitrogen) at 37°C with an objective heater controlled by a TC-124 temperature controller (Warner Instruments).