INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION SUMMARY APPENDIX A APPENDIX B APPENDIX C REFERENCES

Active cellular processes involving morphology, such as endo- and exocytosis, ruffling, lamellipodial extension, and cell motility, probably require reversibility in the binding events among submembrane structural proteins and lipids. However, few techniques are available for measuring or imaging in vivo binding/unbinding kinetic rates of proteins in cells. Standard nonimaging techniques for measuring kinetics rates in vitro, such as stop flow, concentration jumps, pressure jumps, and temperature jumps, are difficult to apply to single living cells. Total internal reflection/fluorescence recovery after photobleaching (TIR/FRAP), as described below, is shown in this study to measure and spatially map the varied submembrane binding kinetic rates of fluorescence-marked actin in living cells.

TIR fluorescence (TIRF) is an optical technique used to excite fluorophores within ~80 nm of a water-glass or cell-glass interface (Axelrod et al., 1992). The illumination intensity decays exponentially with distance from the surface with a characteristic length that depends on the illumination angle, indices of refraction at the interface, and illuminating wavelength. When applied to a living cell adhering to a coverslip and injected with a fluorescent actin marker, TIRF preferentially excites fluorophores at or near the plasma membrane at the cell-glass contact regions, while not exciting fluorophores deeper in the cytosol (see Fig. 1).

View larger version (117K): [in this window] [in a new window]   FIGURE 1   TIR versus EPI illumination on living BC3H1 cells injected with R-actin (a and b) or R-phalloidin (c- f). TIR illumination (a, c, and e) excites the fluorescence-marked actin selectively near the plasma membrane at cell-substrate contact regions. EPI illumination (b, d, and f) shown for the same cells excites the fluorescence-marked actin throughout the entire cell. Comparison of TIR versus EPI shows, for example, that the right side of the cell in c is not attached to the coverslip; a phase-contrast image (not shown) shows an adjacent nonfluorescent cell on the right, presumably intercolated into the substrate-proximal space under the fluorescent cell.

TIRF combined with FRAP can measure kinetic rates of fluorescent molecules in reversible equilibrium with binding sites at a surface (Thompson et al., 1981). TIR/FRAP previously has been used to measure unbinding rates (also called "off-rates") of cytoplasmic proteins (McKiernan et al., 1997) and immune system proteins (Hsieh and Thompson, 1995; Sheets et al., 1997; Gesty-Palmer and Thompson, 1997) from supported lipid bilayers, and the unbinding rates of hormones from biological cell surfaces (Hellen and Axelrod, 1991; Fulbright and Axelrod, 1993), all in nonliving systems and in a nonimaging mode.

In this paper, actin in living BC3H1 (mouse smooth muscle-like) cells is fluorescently labeled by microinjection of either rhodamine-labeled g-actin or rhodamine-labeled phalloidin (a toxin that binds strongly to f-actin). Actin kinetics are measured by TIR/FRAP in the presence or absence of cytochalasin B (which blocks monomer association/disassociation at the barbed (fast) end of actin polymers; MacLean-Fletcher and Pollard, 1980; Cooper, 1987), sodium azide and 2-deoxyglucose (which block ATP production), and unlabeled phalloidin. Rhodamine-phalloidin kinetics are also measured in the absence of further treatment. The spatially resolved kinetic data are displayed as spatial maps ("images") and analyzed by spatial autocorrelations, linear gradient fits, and dependence on initial intensity. We find that the average off-rate of actin is spatially nonuniform and sensitive to drug treatments. The optical and analysis techniques used here should be generally applicable to many other membrane or submembrane binding proteins.

Actin kinetics near the membrane may result from several phenomena: binding/unbinding of a polymerized f-actin filament side or tip to any of dozens of membrane-anchoring proteins or protein complexes in the membrane, binding of monomeric g-actin to specific anchoring proteins, binding of either type of actin directly to membrane lipids, or exchange of monomers at the filament tip. F-actin binding directly to lipid has been detected in vitro, using differential scanning calorimetry and electron microscopy, and in ionic conditions found in vivo (Gicquaud, 1993, 1995). Rates of turnover in actin filaments in the cytosol of Swiss 3T3 cells, fibroblast cell lines, and goldfish epithelial keratocytes have previously been measured by Wang (1987), Theriot and Mitchison (1992), and Tardy et al. (1995), using FRAP and fluorescent photoactivation techniques in non-TIR modes.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION SUMMARY APPENDIX A APPENDIX B APPENDIX C REFERENCES

Biological preparation

The cell chamber consisted of a 35-mm plastic culture dish with a 3/4" diameter hole drilled out and a cleaned coverslip glued over it onto the outside of the dish with Sylgard 182 epoxy resin (Dow Corning Corp., Midland, MI). Coverslips were previously cleaned by boiling in a porcelain holder in 5% Linbro 7× cleaning solution (Flow Laboratories, McLean, VA) for at least 45 min, followed by rinsing in running deionized water, dipping in acetone three times, and heating at 110-150°C to dryness for several hours.

BC3H1 cells are flat, adherent, irregularly shaped cells on the order of 50 µm across that are a few microns thick at most in extranuclear regions. Injected BC3H1 cells were plated directly into the glass-bottomed dishes described above by standard procedures and incubated for at least 2 days at 37°C, 9% CO₂ in Dulbecco's modified Eagle's medium containing 10% heat-inactivated fetal bovine serum (GIBCO, Grand Island, NY). Before microinjection, the medium was replaced with room temperature Dulbecco's phosphate-buffered saline with 1 g/liter glucose (PBSg) (GIBCO).

Microinjection protocol

Cell microinjection was performed on a Leitz Diavert moving stage focus microscope; the same microscope was used for the subsequent optical observations. The entire microinjection positioner setup (as custom-assembled from standard precision translators) was firmly bolted to a microscope support column that moves vertically along with the stage focus mechanism. Starting from the bolting point, the setup consisted of a mechanical x-y-z translator (Line Tool Co., Allentown, PA) onto which was mounted a piezoelectric transducer (Physik Instruments, Costa Mesa, CA) driven by a standard 24-V power supply. To prevent sheer forces on the piezotranslator, the moving end of the piezoelectric transducer was glued to the moving platform of a one-dimensional miniature ball slider (model 0611; Daedel, Harrison City, PA), the frame of which was fixed to the x-y-z translator. A universal joint that held a micropipette holder/pressure tubing assembly (General Valve, Fairfield, NJ) was affixed to the ball slider's moving platform.

Microinjection pipettes (with filaments) were pulled with a Flaming Brown pipette puller (Sutter Instrument, Novato, CA) to a tip inside diameter of ~0.2 µm. Continuous rather than pulse pressure, at 5-20 psi, was used during injection to prevent pipette clogging. Also to prevent pipette clogging, the protein was prefiltered through a 0.22-µm centrifuge filter tube (SpinX; Costar, Cambridge, MA) at ~1000 × g for 5 min. The injection solution was delivered into about four pipettes (with the delivery close to the pipette tip to prevent clogging) with a 5-µl Hamilton syringe (Hamilton Company, Reno, NV). After each use, the syringe was cleaned with ethanol and blown out with nitrogen gas. After the pipette was swung into place on the universal joint and the tip was centered in the field of view, using a 3× objective and darkfield illumination, the pipette tip was brought down with the mechanical translator to touch the cell membrane. With a 40× phase-contrast objective, a dimple could be seen where the pipette touched the membrane. The piezoelectric translator was then transiently powered to give the pipette a ~2-µm fast poke. A ripple throughout the cell indicated that liquid was injected. A slow, contained ripple was found to correlate with probable cell survival, whereas a faster, more violent ripple was found to correlate with probable cell death. If the injection was judged as good, the cell coordinates were recorded. Fig. 2 shows phase-contrast images of several BC3H1 cells, two of which were microinjected. Actin is particularly difficult to microinject because it tends to polymerize where the injection solution contacts the cell buffer solution; after the pipette clogged, it was sometimes intentionally broken on the coverslip and further used at a lower pressure.

View larger version (94K): [in this window] [in a new window]   FIGURE 2   Phase-contrast images of BC3H1 cells. (a) A field showing a rhodamine-actin-injected cell (marked by a star), with several other uninjected cells nearby. The injected cell was among those used for data acquisition (cell a of Fig. 5). (b) Another rhodamine-actin-injected cell used for data acquisition (cell e of Fig. 5).

Injection solutions

Three different protein combinations were microinjected, which were prepared as follows.

Rhodamine actin

Rhodamine phalloidin

Rhodamine actin and unlabeled phalloidin

Biochemical disruption treatments

Some of the rhodamine-actin-labeled cell dishes were subject to specific biochemical disruption. Disruptive treatments consisted of 50 µl of one of the following stock solutions added to the 2 ml of PBSg already on the cells and triturated ~20 times: 1) cytochalasin B stock solution (5 mg of cytochalasin B in 1.25 ml ethanol, stored at 20°C; 0.1 mg/ml final concentration of cytochalasin B, a concentration more than sufficient to totally disrupt actin filaments) or 2) sodium azide stock solution (5 ml of H₂O, 0.33 g of 2-deoxyglucose, 0.13 g of sodium azide, and 0.06 g of NaCl; 10 mM final concentration of both sodium azide and 2-deoxyglucose). For nondisrupted controls, the same solvents (ethanol or 100 mM NaCl) were added to cells in the same volume but without the cytochalasin or sodium azide solutes. No significant differences were seen between the ethanol-based and NaCl-based controls, so results for these two preparations are pooled where averaging is appropriate.

Postinjection protocol

After microinjection of many cells in one dish under dim epiillumination, the cell PBSg was replaced by 2 ml of fresh PBSg. A layer of mineral oil on top of the PBSg prevented evaporation. In this preparation, cells have been observed to live for more than 1 day. At least 15 min after injection, one injected cell was selected based on viable appearance in phase contrast and an acceptable amount of fluorescence gauged with dim epiillumination. In the case of rhodamine-actin-injected cells, a coin was flipped to randomly determine whether the cell dish would receive a disruptive treatment or control treatment as above. TIR/FRAP experiments were carried out at least 1 h after treatment with a computer-controlled laser/charge-coupled device (CCD) setup. All experiments were performed at 25°C. During or after the TIR/FRAP experiment, a phase-contrast time lapse movie (at 2 frames/min) of at least 15 min duration was taken to assay cell viability. A cell was defined to be "alive" if one or more lamellipodia were extended during the 15-min observation period.

The results are based on experiments on seven "control" rhodamine-actin-injected cells left untreated by disruptive agents, three rhodamine-actin-injected cells treated with cytochalasin B, three rhodamine-actin-injected cells treated with azide solution, five rhodamine-actin-injected cells treated with unlabeled phalloidin, and five rhodamine-phalloidin-injected cells. Of these, most were also used for spatial analysis, although some cells were too noisy or ill behaved for spatial analysis.

Cell fixation

For rhodamine-actin-labeled fixed cells, cells were microinjected and incubated for 1 h, then fixed according to the method of Bloch et al. (1989). Rhodamine-phalloidin-labeled fixed cells were fixed according to the method of Bloch et al., then labeled with 5 µl of rhodamine phalloidin methanol stock solution in 200 µl PBSg.

Optical setup

View larger version (29K): [in this window] [in a new window]   FIGURE 3   TIR optical setup and cell chamber. Our TIR optical setup uses two no. 2 coverslips and an inexpensive commercial 45-45-90° triangular prism (7-mm sides, refractive index = 1.52). The top coverslip, to which the cells are adhered, is glued to the bottom of a drilled-out 35-mm plastic tissue culture dish with Sylgard 182 encapsulating resin (Exxex Group, Wood Dale, IL). The bottom coverslip is in optical contact with both the top coverslip above and the prism below through thin layers of immersion oil. For searching the cell culture in the horizontal plane with the standard microscope stage translators, the dish-coverslip configuration slides freely with respect to the prism, which is prepositioned below the stage by a three-dimensional translator attached to the microscope but otherwise held fixed. (Another possibility is to keep the lower coverslip fixed relative to the prism and allow only the upper coverslip to slide horizontally; this would minimize oil smearing when the sample is moved.) The laser beam enters the coverslips through the prism at a ~75° incidence angle. Once trapped inside the coverslips, the beam totally internally reflects several times. The relatively thick double coverslip combination allows successive reflections to be well separated, but it is not so thick as to preclude the use of moderately short working distance objectives. The third or fourth reflection at the upper surface was used in these experiments. This setup requires an intermediate or long working distance water-immersion or air objective (an oil immersion lens would prevent TIR at the immersion-coverslip interface). (A Leitz achromat (50×, NA 1.00, free working distance 0.68 mm, 160-mm tube length) was used here.) In contrast to other inverted microscope prism-based TIR configurations, this TIR setup has the advantages of 1) providing complete access to the cell from above and 2) presenting the sample at the immediate far side of a coverslip (rather than through an additional layer of medium beyond the coverslip), which reduces optical aberrations.

Data acquisition

Data were collected with a CCD camera (Photometrics Star 1, Tucson, AZ). Data acquisition was automated and was controlled with a custom-written PC program and hardware consisting of a CTR-05 counter-timer board (Computer Boards, Inc.) interfaced to the acoustooptical modulator and laser shutter, and a GPIB board (National Instruments, Austin, TX) interfaced to the camera. The probe excitation light was provided as a flash of 1/60 s duration (to eliminate effects of 60-Hz noise in the laser power), synchronously with the CCD's mechanical shutter opening. Each probe exposure bleached no more than 0.5% of the surface-bound fluorescence per probe frame, as estimated by bleaching surface-bound rhodamine-bovine serum albumin (BSA) under the same conditions. The bleach was accomplished by increasing the illumination duration by a factor of 100-200 for a single flash, without changing the intensity. This protocol allows the use of simple on/off laser shuttering rather than graded analog control of laser intensity and is easier to implement where total laser output power is limited. The average bleach amplitude ranged from 0.5 to 11 counts per pixel in different runs, which gave adequate signal to noise when 5 × 5 pixel groups were fitted. Data were collected at 10 frames/min. The spatially resolved intensities were recorded in a stack of successive CCD image frames.

TIR/FRAP theory

(1)

(2)

View larger version (10K): [in this window] [in a new window]   FIGURE 4   Typical TIR/FRAP data with exponential fit (line) for a nondisrupted R-actin-labeled control cell. For this particular cell (shown in Fig. 5 a), the bleach fraction is 15%, the off-rate koff is 0.040 ± 0.006 s1, and the mobile fraction  is 0.9.

(3)

(4)

Parameter definitions

The "prebleach intensity," f(), is the average of ~10 prebleach frames at each pixel location or binned group. The bleach occurs at t = 0, the intensity just after the bleach is f(0₊), and the observed postbleach fluorescence for t > 0 is f(t). The fit parameters are defined as follows.

The "bleach amplitude" L is defined to be the amount of fluorescence lost as a result of the bleach pulse:

(5)

The "mobile fraction"  is the fraction of the bleach amplitude that recovers:

(6)

We calculated the mobile fraction from the right-side expression, without reference to fit parameter a, by estimating f as the average of f in the five last (i.e., largest t) frames to occur.

The "off-rate," k_off, is the exponential recovery rate, k, of the fit (Eq. 2). This identification of k as k_off depends on the process being in the "reaction limit" rather than the "diffusion limit" (see Appendix A) (Thompson et al., 1981).

The "average off-rate," k_off, is the product of the off-rate and the mobile fraction:

(7)

The parameters that are of most interest here are , k_off, k_off, and f(). The first three are "kinetic parameters" and should be independent of the illumination intensity, bleach intensity, bound fluorophore distance from the coverslip, or any other nonkinetic variable. One nonkinetic variable, the "bleach fraction" (defined as the normalized bleach amplitude L/f()) displays an interesting behavior in our data but is more difficult to interpret (see Appendix B).

Analysis methods

The spatially resolved kinetic data were analyzed in several different formats to emphasize particular features. All fitting, analysis, and display software was custom-written in Fortran.

Kinetic parameter spatial maps

Spatial autocorrelations

(8)

Gradient of kinetic parameters

(9)

Dependence of kinetic parameters on local brightness

Cross-correlations

(10)

Kinetics averaged over the whole cell

Estimated concentration of labeled protein in microinjected cells

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION SUMMARY APPENDIX A APPENDIX B APPENDIX C REFERENCES

Spatially resolved kinetics

Spatial maps

1

View larger version (56K): [in this window] [in a new window]   FIGURE 5   TIR prebleach intensity images of the seven R-actin-labeled cells used in nondisrupted control experiments (left) and the correspondingspatially resolved maps of koff (right), where each fit region is 4 × 4 pixels (1 µm × 1 µm). Filamentous structures are clearly seen in the TIRF images of cells a, d, and e. The bar represents 20 µm for all panels. The contrast scale for koff is dark gray = 0, white = 0.1 s1.

View larger version (81K): [in this window] [in a new window]   FIGURE 6   Images of the spatially resolved kinetic parameters for a R-actin-labeled cell (cell a in Fig. 5). (a) Prebleach fluorescence intensity f() (no binning). (b) Off-rate koff (grayscale range 0-0.1 s1). (c) Mobile fraction  (grayscale range 0-1). For the kinetic parameters in b and c, the pixels are binned 5 × 5, corresponding to 1.25 µm × 1.25 µm before the data are fit. The long white bar represents 20 µm. The gray bar in the lower right corner of each panel represents the relevant dcorr lengths for this cell. The dotted lines in each panel show the orientations of the maximum gradient of a planar fit to the data for f(), koff, and , respectively.

View larger version (15K): [in this window] [in a new window]   FIGURE 7   TIR/FRAP recoveries from the "fast" () and "slow" () ends of a particular R-actin-injected cell (cell a of Fig. 5), where the "fast" and "slow" ends are identified by a fit to Eq. 9. The bleach time is indicated by an arrow, and the smooth lines represent the computer fits to an inverted exponential. Note that the "slow" end appears to bleach less deeply than the fast end (see Appendix B). The noise in the recovery curves is due to both shot noise (random scatter) and possible small cell motions (infrequent step transitions).

Autocorrelations

View larger version (32K): [in this window] [in a new window]   FIGURE 8   Spatial autocorrelation functions G(r) for the average off-rate koff for the seven R-actin-injected cells depicted in Fig. 5.

Gradients of kinetic parameters

Cross-correlations

Repeatability of spatial patterns

Kinetics averaged over the whole cell

Kinetic rates with and without biochemical disruption

Calculation of C/A

8

Cell-to-cell variations

Ruling out possible artifacts in the recoveries

Diffusion of fluorophore in the bulk cytosol

Reversibility in bleaching or fluorophore attachment

Faster rates

1

1

1

1

Photochemical damage

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION SUMMARY APPENDIX A APPENDIX B APPENDIX C REFERENCES

Overview

This report describes the first application of TIR/FRAP in imaging mode to living cells, used here to measure actin kinetic unbinding rates ("off-rates") at the submembrane. Regardless of whether the actin in living BC3H1 cells is marked by injected rhodamine-labeled actin monomers or by injected rhodamine-phalloidin, the TIR/FRAP measurements show similar off-rates and fractions of marked actin that are reversibly bound ("mobile fraction"). The kinetic parameters derived from the TIR/FRAP results show considerable variations, both between different cells and spatially within a single cell. However, as could be seen from multiple TIR/FRAP experiments on the same cell, some spatial and cell-to-cell variations are repeatable and consistent features of the cells.

Although different parts of a typical cell show different kinetic rates, the kinetic rates are generally similar over distances on the order of 3 µm. This is quantitatively clear from the spatial autocorrelation functions for initial intensity, mobile fraction, off-rate, and average off-rate, which display a positive correlation over short distances, with a characteristic distance of 1-3 µm, and a negative correlation over larger distances greater than ~7-14 µm. Furthermore, cells exhibited significant cell-wide gradients of kinetic parameter polarity, with an apparent "fast" and "slow" end in the off-rate, mobile fraction, and average off-rate (and also bleach fraction). However, the prebleach fluorescence intensity exhibited somewhat smaller relative cell-wide gradients.

Some cells exhibited a dependence of the off-rate on initial intensity. However, the direction of the dependence was inconsistent from cell to cell. If we simplistically assume that the on-rate and local density of actin-binding sites are spatially invariant, then we might expect that areas with a faster off-rate would coincide with areas of lower prebleach intensity. However, the opposite was true in several cells: brighter locations coincided with faster off-rates. Therefore, the simplistic assumption cannot be true, and therefore the density of actin-binding sites and/or the on-rate must vary spatially in the cell.

In principle, recovery after TIR photobleaching could arise from nonkinetic processes, such as reversible photobleaching, recovery by diffusion, and recovery arising from membrane motion. The first two possibilities were refuted by control experiments. The third appears unlikely because the fluorescence at each pixel after bleach recovery is complete was found to be highly correlated ( = 0.94) with the prebleach fluorescence at that pixel. If recovery were due to membrane motion, we would expect little correlation between initial and final intensities at particular pixels.

R-Actin kinetics

Theriot and Mitchison (1991, 1992) measured the dissipation times of fluorescence-photoactivated actin from actin filaments in lamellipodia as observed by epiillumination. In two different fibroblast cell lines, actin dissipation times were found to be 55 ± 28 s and 181 ± 99 s. In highly motile goldfish epithelial keratocytes, the actin dissipation time was observed to be 23 s. These dissipation times were reasonably interpreted as the average time in which a marked actin remains incorporated in an actin polymer filament, i.e., the characteristic time for depolymerization. Our TIR/FRAP measurements yielded an average residency time of 31 ± 7 s for actin to remain at or near the cytoplasmic membrane in BC3H1 smooth muscle cells. The characteristic times provided by these two very different techniques might be viewed as "consistent" with each other (although it is surprising that the turnover rates in the keratocytes are not significantly faster than in our less mobile BC3H1 cells). However, the cell types were different (keratocytes, for example, do not have stress fibers), and the optics were different: dissipation of photoactivated fluorescence preferentially probes actin depolymerization in lamellipodia at the cell periphery, whereas TIR/FRAP can probe cell-substrate contact regions underlying even rather thick central parts of a cell.

The time scales of actin kinetics as seen by TIR/FRAP can be compared to time scales of cellular motion. If, for example, actin kinetic time scales are much longer than time scales of cellular motion, then actin binding and unbinding at the cytoplasmic submembrane are probably not involved in cell motility. On the other hand, if the time scales of actin kinetics are short compared to cell motility, then there is ample time for the actin cytoskeleton to reshuffle during the morphological alterations of motility. Of course, the time scale for cellular motion can be a somewhat ambiguous concept because it would depend on an arbitrary assignment of a characteristic distance. We estimated the time scales of cellular motion by an autocorrelation technique that reports the average time that the local cell boundary remains within ~0.4 µm of its original position, about one pixel here (see Appendix C). Although arbitrary, this characteristic distance is clearly much smaller than any dimension of the cell. In several experiments with four cells, time scales for cell morphological alterations ranged from 50 to 200 s for local motions of <0.4 µm. These time scales are somewhat longer than the actin off-rate, measured as 31 s, thereby allowing for the possibility that the actin cytoskeleton does reorganize with binding/unbinding events during cell motion.

Treatment of rhodamine-actin-injected cells with sodium azide results in a reduction in mobile fraction by a factor of 10. Furthermore, TIR/FRAP on a "dead" cell with no motile activity (shown by a phase-contrast movie) shows zero average kinetic rates. These results confirm that the reversible chemistry of actin at the cell submembrane as detected by TIR/FRAP is an active, energy-dependent process. Patterson and Spudich (1995) have suggested (for a very different cell type, Dictyostelium) that azide may disrupt some actin involved in cellular spreading and flattening. In general, stress on the cell (including the response to microinjection itself) may affect the actin network. Treatment with cytochalasin B also results in a 10-fold reduction in the average mobile fraction of injected rhodamine-actin. Cytochalasin B (at much lower concentrations than used here) is known to block polymerization of actin into a filament (at the barbed end) and to decrease interfilament interactions (MacLean-Fletcher and Pollard, 1980). If cytochalasin B decreases the polymerization "on-rate" of monomeric actin onto filaments but the filaments do not change their length by a significant factor (MacLean-Fletcher and Pollard, 1980), then perhaps the off-rate is also reduced. TIR/FRAP directly measures "off-rates"either marked actin departing from filaments or marked filaments leaving the submembrane under illumination. At our concentrations, it is likely that actin filaments are more extensively disrupted. The cytochalasin B effect seen here may suggest that cytochalasin B at high concentration inhibits the motion of whole filaments.

R-phalloidin kinetics

Phalloidin is a toxin that binds very strongly to polymerized actin, and one might predict that it would inhibit actin binding/unbinding kinetics. According to the results presented here, such is not the case; phalloidin (in either unlabeled or labeled form) does not alter the kinetic off-rate or mobile fraction, relative to rhodamine-actin. Although phalloidin slows locomotion and growth in tissue culture cells in a dose-dependent manner (Wehland et al, 1977; Wehland and Weber, 1981), cells injected with phalloidin still divide and move (Wang, 1987), and phalloidin-labeled f-actin does translocate in a living cell (Wehland et al., 1980). A related actin marker, phallacidin, does not inhibit cytoplasmic streaming in algae (Barak et al., 1980). In our experiments, the injected phalloidin concentration is estimated to be approximately the same as that used by Wang (1987). The similarity of the TIR/FRAP results for fluorescent actin and fluorescent phalloidin suggests that phalloidin can be a good actin marker not just for structure but also for dynamics at the submembrane of living cells.

Such "submembrane dynamics" (as measured by the postbleach fluorescence recovery of R-phalloidin) could arise, in principle, from several phenomena falling into two distinct classes: 1) "actin dynamics" involving either exchange of whole fluorescence-marked actin filaments in the evanescent field or depolymerization of R-phalloidin-bound monomers from actin polymers or 2) "phalloidin dynamics" involving reversible dissociation of R-phalloidin from its actin filament (or other) binding sites. To determine whether our measured results arise from actin dynamics rather than phalloidin dynamics, we can compare our recovery rates to previous measurements of phalloidin association and disassociation rates from filamentous rabbit skeletal muscle actin, reported as 2.9 × 10⁴/M/s and 2.6 × 10⁴/s, respectively (de la Cruz and Pollard, 1996). Based on these figures, phalloidin dynamics would have a characteristic recovery time of ~3000 s, or ~100 times longer than our observed times. We can conclude that phalloidin does not bind to different actin molecules during the course of our experiment, and it is unlikely that the kinetics we observe are phalloidin-actin dissociation dynamics. The kinetic process we observe experimentally with TIR/FRAP on R-phalloidin-injected cells probably results from the motion of entire filaments or protofilaments.

Because phalloidin binds predominately to polymerized actin, the similarity in the time scales of TIR/FRAP results for R-phalloidin and R-actin cells supports this inference that the kinetics (and intensities) provided by the TIR images arise mainly from the motion of entire filaments or protofilaments. This is not to say that monomeric actin does not engage in significant reversible binding activities with nonactin sites at the submembrane, but only that the effects are dominated by polymerized actin, at least in the cells used here.

Using standard spot FRAP on filaments in the cytoplasm of phalloidin-injected Swiss 3T3 cells, Wang (1987) measured the actin recovery time to be 500 ± 65 s. Based on FRAP diffusion measurements with an epifluorescent focused laser spot, Wang (1987) attributed this process to attachment/detachment of small actin filaments to larger bundles of filamentous actin. Because our measured characteristic time (the reciprocal of the k_off) of actin dynamics at or near the cytoplasmic membrane in phalloidin-injected cells was only 30 ± 12 s, the dynamic rates of phalloidin-labeled actin located deep within the cytosol in Swiss 3T3 cells appear to be an order of magnitude slower than the corresponding rates near the membrane of our BC3H1 cells.

Imaging TIR/FRAP as a technique

In addition to producing spatially resolved data, an imaging mode for TIR/FRAP has several advantages over a nonimaging mode. Imaging mode TIR/FRAP automatically documents the absence of common problems, such as beam drift, focus drift, or changing interference fringes in the illuminating beam, thereby affording higher confidence in the data. The imaging mode also permits both better background subtraction and possible correction of temporal illumination fluctuations by calibration with the background emission intensity in off-cell regions. The imaging mode is also useful for assaying the cellular state during or after experimentation.

There are also some disadvantages to using the imaging mode. The CCD has significant readout noise per pixel for each image, somewhat higher than the noise of a single-channel photodetector (usually a photomultiplier or avalanche photodiode) typically used in nonimaging mode. The second disadvantage of imaging is slower speed; nonimaging mode TIR/FRAP has been used previously at sampling rates as high as 30,000 data bins/s (McKiernan et al., 1997). Both of these problems can be partly ameliorated with newer commercial cooled CCD cameras, which are both quieter and faster than the one used here.

	SUMMARY

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION SUMMARY APPENDIX A APPENDIX B APPENDIX C REFERENCES

This paper demonstrates that spatially resolved kinetics can be measured using TIR/FRAP with imaging at the submembrane of living cells. Use of a CCD detector yields orders of magnitude more data than conventional single-channel detectors, from which can be extracted spatial maps and information about cell-wide gradients, dependence of kinetics on initial intensity, and correlations of kinetic parameters with fluorescence structures. This project demonstrates that significant and repeatable variations occur in the kinetic parameters over the surface of the cell and from cell to cell.

Possibile future applications of TIR/FRAP on living cells include the correlation of cell-wide kinetic parameter gradients to external stimulation or cell motion (such as in fast-moving keratocytes); measurements of kinetics of some of the many other submembrane proteins; and comparison of cell-to-cell kinetic variations with the stage of cell division. In particular, TIR/FRAP studies on other proteins crucial to the motile and mechanical properties of the cell surface (such as annexin), but with a less complex (and thereby more interpretable) chemistry or a smaller range of binding sites, appear worthwhile. In addition, the microinjection step may be circumvented in the future by use of cells transfected with genes for specific cytoskeletal protein-green fluorescent protein hybrids.