Real-time dynamics of the F-actin cytoskeleton during secretion from chromaffin cells

Transmitted light images showed an intricate and dynamic cytoplasmic structural network in cultured bovine chromaffin cells observed under high magnification. These structures were sensitive to chemicals altering F-actin-myosin and colocalised with peripheral F-actin, β-actin and myosin II. Interestingly, secretagogues induced a Ca2+-dependent, rapid (>10 second) and transitory (60-second cycle) disassembling of these cortical structures. The simultaneous formation of channel-like structures perpendicular to the plasmalemma conducting vesicles to the cell limits and open spaces devoid of F-actin in the cytoplasm were also observed. Vesicles moved using F-actin pathways and avoided diffusion in open, empty zones. These reorganisations representing F-actin transfer from the cortical barrier to the adjacent cytoplasmic area have been also confirmed by studying fluorescence changes in cells expressing GFP-β-actin. Thus, these data support the function of F-actin-myosin II network acting simultaneously as a barrier and carrier system during secretion, and that transmitted light images could be used as an alternative to fluorescence in the study of cytoskeleton dynamics in neuroendocrine cells.

scinderin (Rodriguez del Castillo et al., 1990) or alternatively by network reorganisation based in the interaction between Factin and molecular motors such myosin (Nakanishi et al., 1989;Gutiérrez et al., 1989;Ñeco et al., 2002;Ñeco et al., 2003;Ñeco et al., 2004). More recently, this model of the Factin network as a passive barrier has been redefined by a number of contributions using the technique of total internal reflection fluorescence microscopy (TIRFM) allowing direct visualisation of vesicles within 300 nm of the cellular limits. These studies performed in living PC12 (Lang et al., 2000) and chromaffin cells (Steyer and Almers, 1999;Oheim and Stühmer, 2000;Johns et al., 2001), have suggested an active role for an actin-myosin transport system in propelling secretory vesicles through the subplasmalemmal zone.
In this study we address cytoskeleton dynamics during secretion using both transmitted light scanning microscopy and confocal fluorescence. The former is used as a new tool allowing the direct visualisation of complex cytoplasmic structures inside cell cytosol. These structures, with differential density to the passage of visible light, are strictly sensitive to actin and myosin inhibitors and present a distribution characteristic of the F-actin network. More interestingly, there is a clear association between the dynamics of both chromaffin granule movements and these cytoplasmic structures, which appears to suffer complex reorganisations in the subcortical and internal cytoplasmic regions during the secretory cycle of chromaffin cells. Taken together, these studies and the data gathered using fluorescent-tagged β-actin support an active 2871 Transmitted light images showed an intricate and dynamic cytoplasmic structural network in cultured bovine chromaffin cells observed under high magnification. These structures were sensitive to chemicals altering F-actinmyosin and colocalised with peripheral F-actin, β-actin and myosin II. Interestingly, secretagogues induced a Ca 2+dependent, rapid (>10 second) and transitory (60-second cycle) disassembling of these cortical structures. The simultaneous formation of channel-like structures perpendicular to the plasmalemma conducting vesicles to the cell limits and open spaces devoid of F-actin in the cytoplasm were also observed. Vesicles moved using F-actin pathways and avoided diffusion in open, empty zones. These reorganisations representing F-actin transfer from the cortical barrier to the adjacent cytoplasmic area have been also confirmed by studying fluorescence changes in cells expressing GFP-β-actin. Thus, these data support the function of F-actin-myosin II network acting simultaneously as a barrier and carrier system during secretion, and that transmitted light images could be used as an alternative to fluorescence in the study of cytoskeleton dynamics in neuroendocrine cells.
dual role of F-actin as a barrier and carrier system during secretion of neuroendocrine cells.

Materials and Methods
Isolation and culture of bovine chromaffin cells Chromaffin cells were prepared from bovine adrenal glands by collagenase digestion and further separated from debris and erythrocytes by centrifugation on Percoll gradients as described (Almazan et al., 1984;Gil et al., 1998). Cells were maintained in monolayer cultures using Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, 10 µM cytosine arabinoside, 10 µM 5-fluoro-2′-deoxyuridine, 50 IU/ml penicillin and 50 µg/ml streptomycin and were harvested in 35 mm Petri dishes (500,000 cells/dish, Corning, NY, USA). Cells were used between the second and fourth day after plating. All experiments were performed at room temperature (21-22°C). Chemicals were from Sigma (Madrid, Spain).

Dynamic confocal microscopy
Cells were loaded with quinacrine using a 4 µM concentration in culture medium for 10 minutes as described (Ñeco et al., 2002; Ñeco et al., 2003; Ñeco et al., 2004). Fluorescent emission from quinacrine trapped in the acidic vesicles was investigated using an Olympus Fluoview FV300 confocal laser system mounted in a BX-50 WI upright microscope incorporating a 100ϫ LUMPlan FI waterimmersion objective. Excitation was performed using Ar and HeNe visible light lasers. This system allows for Z-axis reconstruction (0.5-0.55 µm theoretical Z slice) and time-lapse dynamic studies with time resolutions ranging from 0.1 second for 200ϫ150 pixels image acquisition (adequate for region studies) to about 0.6 second for images of 400ϫ300 pixels (for visualisation of the entire cell). Simultaneous acquisition of transmitted light images was obtained using the channel implemented in the confocal microscope and using bright field optics (theoretical depth of field is 0.55-0.65 µm). The intensity of this channel was adjusted to avoid saturation of the subcortical region and visualise the lighter cytoplasmic structures. Distribution of the F-actin network was studied using rhodamine coupled to phalloidin in intact or permeabilised cells. As this chemical presents low permeability throughout the plasmalemma, intact cells in culture media were incubated with 1 µM rhodamine-phalloidin for 30 minutes at 37°C, in this way a low percentage (5-10%) of the cells, were labelled enabling Z-axis confocal reconstructions. Microtubules were visualised in intact cells using 1 µM taxol conjugated with Bodipy 564/570 (Molecular Probes, Eugene, OR). Incubation of this compound for 15-20 minutes at 37°C allowed efficient labelling of the total population of cultured chromaffin cells. Other actin and myosin affecting chemicals were used as described (Gil et al., 2000;Ñeco et al., 2002;Ñeco et al., 2003;Ñeco et al., 2004).
Production of GFP-actin and mutant and wild-type forms of RLC-GFP constructs A GFP-actin construct was obtained by NheI/MunI digestion of pEGFP-actin (Clontech, Palo Alto, CA) and inserted into an XbaI/EcoRI-digested pHSVpuc amplicon, upstream the IE 4/5 promoter. Packaging and amplification of this vector were performed using standard procedures (Lim et al., 1996). A plasmid encoding chicken gizzard smooth muscle myosin II regulatory light chain (RLC; a gift of Dr Kendrick-Jones, MRC, Cambridge, UK) was used for site-directed PCR cassette mutagenesis. The modifications were confirmed by sequencing in both directions. pRLC-wt-GFP and pRLC-T18A/S19A-GFP were generated using pEGFP-N1. The final construct was inserted in pHSVpuc and amplicons produced as indicated above. Primary cultures of chromaffin cells were infected with Herpes simplex virus (HSV-1) amplicon containing the constructs described above as described elsewhere (Gíl et al., 2002). Efficiency of virus infection was determined by fluorescent microscopy, using serial dilutions of purified virus. The dilution chosen for further experiments (20-30 µl virus per 1 ml of medium) produced 10-15% infection efficiency.
Image analysis and particle tracking Analysis of frames was performed using the program of public domain ImageJ with plug-ins for ROI measurement, image average and comparison of multiple channel images and 3D reconstruction (obtained from http://rsb.info.nih.gov/ij/). A multi-tracker plug-in was used to determine particle centroid in time-lapse studies as described earlier (Ñeco et al., 2003; Ñeco et al., 2004). The x-y coordinates determined for vesicles or threshold visible images (structure particles obtained after threshold of the transmitted light channel images) were transferred to Igor Pro, where specialised macros were used to calculate total lateral displacement and the mean square displacement (MSD) for any given time interval based in the equations defined by Qian et al. (Qian et al., 1991), where images were taken at intervals δt and the coordinates of a granule in one image were x(jδt) and y(jδt) and in other image x(jδt + nδt) and y(jδt + nδt). In each temporal sequence the displacement during the interval nδt can be measured for (N-n) intervals, where N is the total number of images acquired.
This equation is a good description of vesicle movement when subjected to a single coefficient of diffusion; the downward curvature of the experimental data indicated the diffusion of vesicles in a cage, which can be described by the approximate equation (Saxton and Jacobson, 1997), This equation represents the first two terms in an infinite series, where A 1 and A 2 are constants (0.99 and 0.85, respectively), D is the coefficient of confined diffusion, and r c is the radius of the theoretical circular cage. Therefore the MSD could be used to calculate r c when ∆t → ϱ and the data reach an asymptotic value. The Student's t-test for paired samples or the two-way ANOVA test was used to establish statistical significance among the different experimental data (samples were considered significantly different when P<0.05). All data were expressed as mean±s.e.m. of experiments performed in a number (n) of individual cells. The data presented represents experiments performed with cells from at least three different cultures.

Results
Transmitted light scanning microscopy allows observation of dynamic cytoplasmic structures in bovine chromaffin cells Recently, confocal microscopy analysis of cells labelled with 4 µM quinacrine has allowed us to make direct observation of vesicle movement throughout the entire cytoplasm of cultured bovine chromaffin cells (Ñeco et al., 2002; Ñeco et al., 2003; Ñeco et al., 2004). During these experiments, we were somewhat surprised by the complex images obtained by scanning microscopy of the transmitted light channel implemented in addition to fluorescence detection. These visible light images showed an intricate network of j=1 F-actin dynamics in chromaffin cells cytoplasmic structures forming polygonal cages as shown by the different optical density of its walls in relation to internal spaces. Furthermore, and apart from its non-confocal nature, it was also observed that the visible information changed notably in adjacent planes separated by 1 µm (Fig. 1). The density of these structures increased in the cytoplasmic region beneath the plasma membrane, forming a peripheral ring that was clearly shown when we focused the planes adjacent to the equatorial cell section (Fig. 1A,C). A density profile along the cell (Fig.  1C, line) shows that the majority of the vesicles appeared to be located within 1-2 µm of the internal limits of this peripheral ring (Fig. 1E). Furthermore, the visualisation of these structures in the y-z and x-z planes showed cytoskeletal cavities with vesicles located in their borders (Fig. 1F). The complexity of these cavities is best observed in 3D reconstructions (Fig.  1G) showing the continuity of these cages along the z-axis. Interestingly, the cytoplasmic structures are dynamic, as observed by time-lapse microscopy, and a close inspection of changes in a peripheral area showed variations in the density of the cortical structure with the infrequent formation of disruptions allowing vesicle penetration in the subplasmalemmal area as can be observed in for the structures ( In these measurements, vesicles tend to stay for periods of 2-7 seconds in the confocal plane presenting short cycle oscillations in fluorescence, whereas visible light intensity changed in cycles of time ranging from 10 to 20 seconds ( Fig.  2C), indicating the slower dynamics of these structures when viewed by transmitted light scanning microscopy. Further assessment of this difference in movement can be made by tracking the x-y positions of the vesicles and particle structures obtained after the threshold of the transmitted light images (Fig. 2D). In this way it was possible to observe an average displacement of 0.79±0.09 µm over 20 seconds (for 60 threshold particles in 16 cells) for the transmitted light structures and 3.75±0.50 µm/20 seconds (n=135 vesicles corresponding to 11 cells) for the vesicle displacement in the confocal plane. These data and the corresponding distributions ( Fig. 2D) indicated that on average transmitted light structures moved at a speed that was four-to fivefold lower than chromaffin granules in the cytoplasm of these cells. Indeed, using this type of data we obtained the mean square displacement (MSD) at different time intervals (Equation 1) and these values were averaged to generate the MSD versus time plot (Fig. 2E). Assuming that movement is governed by a single coefficient of diffusion, its value can be derived from the slope (slope = 4ϫD) (Qian et al., 1991). The diffusion coefficient was calculated to be 5.3±0.4ϫ10 -3 µm 2 /second for moving granules (n=27) and 2.11±0.06ϫ10 -3 µm 2 /second for structure movement (n=18). Both the movement of granules and structures were strictly calcium dependent, as incubation of chromaffin cells in a media lacking additional calcium and in the presence of 1 mM EGTA for 20 minutes drastically affected the diffusion coefficients calculated under these conditions which were at least 100 times lower than control values (3.8±0.6ϫ10 -5 µm 2 /second for moving granules, and 4.5±0.3ϫ10 -5 µm 2 /second for structure movement, n=19 in both cases), as can be observed in the accelerated video (Movie 2 in supplementary material).
Interestingly, the MSD versus time plot for control granules showed a downward curvature and fitted better to an equation representing vesicle movements inside a cage (see Equation 2). The best fit for these data (R 2 = 0.96 compared to 0.91 for lineal regression) gave a value of the theoretical radius of the cage confining the vesicles (r c ) of 0.86 µm. In order to estimate whether the dimensions of the polygonal structures visualised with transmitted light scanning microscopy could be limiting such vesicle movements, we measured the wall to wall maximal X and Y distances for an high number of such cages (n=272 measurements in five cells) and obtained an average value of 1.5±0.7 µm (mean±s.d.), that closely matches the diameter of 1.72 µm for the theoretical circular cage imprisoning vesicle movement.
Chemicals affecting F-actin alter the structure and dynamics of the cytoplasmic structures visualised with transmitted light What is the molecular nature of these visible structures? Based on its distribution accumulating in the cell periphery, our first guess would be the cytoskeletal network of F-actin forming a subplasmalemmal barrier in chromaffin cells (Trifaró et al, 1984;Perrin and Aunis, 1985). In consequence, we treated chromaffin cultures with specific agents affecting such filamentous protein. Treatment of cultured cells with 1 µM latrunculin A, a chemical stabilizing monomeric actin (Spector et al., 1983), for 15 minutes resulted in drastic changes in the characteristics of these visible structures. First, there was an obvious alteration of the cell shape that was accompanied by a reduction of the intensity of the peripheral cortical network (Fig. 3). Quantification by measurement of the density of the cell cortex showed a 60% decrease in the average optical intensity of this cortical structure in measurements performed in nine cells before and after latrunculin treatment (Fig. 3C).
Moreover, the incubation of chromaffin cells with latrunculin A also affected the internal network of polygonal cages (Fig.  3D), which continuously decreased its optical density during incubation with this chemical (Fig. 3E).
Furthermore, the F-actin stabiliser phalloidin coupled to the fluorophore rhodamine could be used to study the colocalisation of this filamentous protein with the structures observed by transmitted light scanning microscopy. Using a 1 µM concentration of this chemical incubated for 30 minutes (Gil et al., 2000), we observed a spatial coincidence of both the peripheral transmitted light structures (Fig. 4A) and the cortical F-actin labelling ( Fig. 4B), that presented significant overlap (Fig. 4C). In some cells intracellular phalloidinrhodamine labelling of cytoplasmic structures can be observed colocalising extensively with that observed using transmitted light ( Fig. S1 in supplementary material). Interestingly, treatment of the cells with this F-actin stabiliser drastically affected both the dynamics of vesicle mobility and transmitted light structures, as under these conditions vesicles were immobilised in contact with the visible cytoplasmic structures that become static upon stabilisation (Fig. 4D,E). These experiments also proved that the transmitted light image was enriched with the visual information coming from the plane studied, otherwise, there was little probability of coincidence between vesicle and structure location. Taken together, these experiments showed that F-actin activity governs the structure and motility of the network observed by transmitted light scanning microscopy.
Journal of Cell Science 118 (13)  (n=18). Also plotted were the best linear fits for the different data, used to estimate the diffusion coefficient according to Qian et al. (Qian et al., 1991) (continuous lines), and the best fit to the equation MSD(∆t) = r c 2 [1 -A 1 exp (-4 A 2 D ∆t /r c GFP-β-actin and the regulatory light chain of myosin II colocalise with the transmitted light network within the cortical region We have shown in a recent publication that myosin II is associated with the F-actin network in the cytoplasm and periphery of chromaffin cells (Ñeco et al., 2002; Ñeco et al., 2004). Based on this, we conceived experiments expressing both the β-actin and regulatory light chain (RLC) of myosin II linked to enhanced green florescence protein (GFP) by their Nterminal and C-terminal domains respectively, using an amplicon-based strategy that has proven successful for the expression of other proteins in chromaffin cells in culture (Gíl et al., 2002;Ñeco et al., 2004). GFP-β-actin was heavily expressed in the cytoplasm and the cortical region of chromaffin cells (Fig. 4F), and colocalised with the cortical density visualised with transmitted light (Fig. 4G,H), whereas in the cytoplasm there was a partial alignment. In addition, when we expressed RLC-GFP and specially an inactive unphosphorylatable form, T18A/S19A RLC-GFP (Komatsu et al., 2000;Ñeco et al., 2004) in chromaffin cells, it could be observed in the cytoplasm forming a network that matched that visualised with transmitted light (Fig. 4I,K). In addition, if granules were labelled with quinacrine, its brilliant fluorescence could be observed in association with both the transmitted light structures and the GFP fluorescence associated with T18A/S19A RLC-GFP (Fig. 4K). Interestingly, the expression of this inactive form of RLC yielded a static network of transmitted light structures whereas the wild-type form of RLC sustained both vesicle (Ñeco et al., 2004) and structure mobility (D.G., unpublished observation).
Structure mobility was affected by chemicals affecting actin-myosin but not by microtubule inhibitors If these structures visualised by transmitted light are composed by Factin-myosin II, their dynamics should also be altered by 2,3-butanodione monoxime (BDM), which affects myosin ATPase activity (Herrmann et al., 1992); or wortmannin, which alters the activity of myosin light chain kinase (MLCK) (Nakanishi et al., 1992). Incubation of chromaffin cells with concentrations of 10 mM BDM, or 1 µM wortmannin for 15 minutes, clearly affects the dynamics of the F-actin visible light structures, producing stabilisation similar to that caused by 1 µM phalloidin or 10 µM jasplakinolide. Under these conditions, transmitted light particle tracking measurements indicated a 50-65% reduction in the mobility of these structures using chemicals affecting actin-myosin in numerous measurements performed in more than 40 threshold density particles in 5-20 cells under each experimental condition (Fig. S2 in supplementary  material).
Similar experiments were conceived to test whether or not microtubules control the activity and dynamics of this network. Cells were incubated for 15 minutes with 10 µM vinblastine or 1 µM taxol labelled with Bodipy 564/570. None of these compounds were able to modify the dynamics of this network observed in the transmitted light channel (Fig. S2 in supplementary material).

Secretagogues induce changes in the density and distribution of the visualised transmitted light structures
The cellular distribution, colocalisation with phalloidin, GFPβ-actin and RLC-GFP and its sensitivity to chemicals affecting actin-myosin strongly suggest that the structure visualised by transmitted light scanning microscopy might be composed of the chromaffin cell cytoskeleton complex of F-actin modulated by the activity of structural myosin II (Kumakura et al., 1994;Ñeco et al., 2002;Ñeco et al., 2004). Thus, transmitted light scanning microscopy could be used to study the cytoskeletal dynamics that accompanies secretion, obtaining a dynamic picture of the complex rearrangements of cytoskeletal elements occurring during the secretory cycle. The stimulation of the cells with 10 µM acetylcholine resulted in a 2-4% increase in the cell perimeter as a consequence of exocytotic membrane incorporation and the simultaneous formation of evident discontinuities in the subcortical structures ( Fig. 5A and Movie 3 in supplementary material). Measurements of optical density in a ROI located over one of the detected disruptions, such as those depicted in Fig. 5B at 2-second intervals, showed that the formation of these subcortical patches within a few seconds was fast enough to encompass secretion (Fig. 5D). In contrast with these rapid changes occurring in the cell periphery, we also observed the reorganisation of intracellular structures forming polygons, increasing the open space in their interior (Fig. 5C). This process, however, takes tens of seconds to fully develop as shown in the images of this panel taken at 10-second intervals. These changes were observed in dozens of cells stimulated with either physiological agonist or by sustained superfusion with high potassium stimulatory solution and they were strictly dependent of the presence of calcium in the stimulatory medium. In similar experiments, stimulation for 1 minute with acetylcholine in a medium lacking Ca 2+ (and in the presence of 1 mM EGTA) produced no apparent changes in the cytoskeletal peripheral integrity compared with resting conditions (Fig. S3 in supplementary material), whereas the posterior perfusion of this cell with this secretagogue in a medium containing 2 mM CaCl 2 induced rapid and drastic changes in the cytoskeletal structure.
In another series of experiments, cells were depolarised briefly with a 10-second superfusion with 59 mM KCl. Under these conditions cells secrete catecholamines for 10-12 seconds as measured using single cell amperometry (Gil et al., 1998;Ñeco et al., 2003), and the cortical cytoskeleton disassembles in parallel (Fig. 6A,B   slowly increased its optical density to recover the initial value 80-100 seconds after the initiation of the transient stimulus (Fig. 6B). Interestingly, disruption of the cytoskeletal F-actin structure was a process involving transfer of material from the subcortical area to the adjacent cytoplasmic regions, as ROI measurements within this zone showed that, with a few seconds delay, increases were detected in the optical density encompassing the parallel decrease of the peripheral barrier (Fig. 6A,B, comparison of the optical density in cortical and interior ROIs). The concept of cortical F-actin transfer rather than its destruction was also supported when the average transmitted light intensity of the overall region studied in Fig.  6 did not change significantly during secretion (Fig. 6B). We observed in many stimulated cells, that the reorganisation of F-actin implied the formation of channel-like structures perpendicular to the membrane plane (see second and third frame of Fig. 6A). A detailed observation of the granule positions during secretion showed that quinacrine-loaded vesicles accessed the cortical region through the newly opened disruptions and they were frequently found in the narrow space left by the channel-forming structures. Finally, they increased their presence in the subplasmalemmal regions 10-15 seconds after disruption of the cortical barrier as detected by measuring quinacrine fluorescence within a ROI located in the exterior of the cortical structure (Fig. 6C).

Vesicles moved along F-actin during stimulus-dependent cytoskeletal reorganisations
During cell stimulation, F-actin reorganised and formed disruptions and the open cytoplasmic spaces as described above, but could the vesicles use such spaces to diffuse and access the subplasmalemmal area? A simple, yet accurate way to observe vesicle positions is to obtain the cumulative image of vesicles in relation to cytoskeletal structures during secretion or even under resting conditions. Cumulative images obtained over 15 seconds in non-stimulated cells show that vesicles moved in the internal face of the cytoskeletal barrier and appeared to avoid the dark area located in the centre of polygons formed by the F-actin network (Fig. 6D, arrows). After stimulation, most of the vesicles either were released or moved to other cell areas but the vesicles observed moved in contact with the cytoskeletal network and again avoided the use of open spaces (arrows in Fig. 6E). These observations were made in dozens of stimulated cells and indicated that during rest conditions or stimulatory activity, chromaffin granules moved preferentially along F-actin structures avoiding access to bare cytoplasmic zones.
Changes in the fluorescence of GFP-β-actin-expressing cells support the F-actin dynamics observed using transmitted light scanning microscopy In order to confirm the observations made with transmitted light scanning microscopy, we studied the cytoskeletal dynamics in chromaffin cells expressing GFP-β-actin construct, where fluorescence is distributed in the cell periphery and also in cytoplasmic patches produced by overexpression (Fig. 7A). These patches corresponded to Factin aggregates as they were labelled by phalloidin (Fig.  7B,C), which also stained the peripheral ring corresponding to the cortical F-actin barrier. Additional proof of the accumulation of GFP-β-actin in the F-actin cortical network, was obtained by treatment with 1 µM latrunculin A. This substance affected the integrity of the peripheral fluorescence that decreased in the cortical area and continuously increased in parallel in the internal cytoplasm as shown in the ROI determinations (Fig. 7D,E), thus validating the use of this fluorescent construct to study of cortical F-actin dynamics in chromaffin cells.
Interestingly, cells overexpressing GFP-β-actin manifested fluorescent changes encompassing those occurring in the transmitted light channel upon cell stimulation by depolarisation. In these experiments a temporal decrease of fluorescence integrated in the cortical zone and the parallel increase in the fluorescence of the cytoplasmic subcortical area was observed and similar results were obtained by analysing the transmitted light images. Furthermore, the formation of cortical disruptions during stimulation and F-actin transfer has also been observed in cells expressing the GFP-β-actin construct (Fig. 8) for a depolarised cell. These experiments proved simultaneous detection of these disruptions in the confocal images of GFP-β-actin fluorescence (Fig. 8A) and in the transmitted light channel images (Fig. 8B). Determination of fluorescence intensity at different distances from the cell limits demonstrated GFP-β-actin transfer from the external to the internal zones during the formation of such disruptions (Fig. 8C). Taking these observations from both transmitted light and fluorescence GFP-β-actin images together, it is clear that during stimulation F-actin redistributed to leave open space in the peripheral barrier and to form the trails for vesicle movement.

Discussion
Transmitted light scanning microscopy provides a new alternative to studying dynamic changes in the F-actin cytoskeleton Although confocal fluorescence microscopy has found widespread use in cell biology, the use of transmitted light scanning microscopy implemented in confocal microscopes is not as widespread. In this work we have shown, using a variety of pharmacological and molecular biology tools in chromaffin cells, that the complex images obtained by transmitted scanning microscopy probably representing variations in cytoplasmic viscosity, are closely associated with the activity of the network of F-actin regulated by interaction with myosin II. Moreover, the study of fluorescence changes after Journal of Cell Science 118 (13) Fig. 7. GFP-β-actin can be used to study cortical F-actin dynamics in chromaffin cells. Cells infected with amplicons containing the GFPβ-actin expressing vector were visualised by confocal fluorescence microscopy. (A) Cells expressing GFP-β-actin were fixed, permeabilised and incubated with phalloidin-rhodamine as indicated in methods. Depicted are confocal images of GFP-β-actin (A), phalloidin-rhodamine (B) and both channels showing evident overlapping (C). (D) GFP-β-actin-expressing cell treated with 1 µM latrunculin A for 30 minutes. Images were acquired at 1 Hz, and frames were separated by 5-minute intervals. (E) Fluorescence in the interior or cortical region was determined and plotted against time after latrunculin A addition. Bar, 10 µm. expression of GFP-β-actin in these cells, support the results obtained using transmitted light, although partial labelling of F-actin structures in the form of β-actin patches, clearly provided a lower level of observation of the subtle structural reorganisations taking place after cell stimulation. By contrast, transmitted light images were highly detailed, and clearly showed for example, immobilised vesicles along the stabilised filaments of F-actin after treatment with phalloidin, latrunculin A or myosin inhibitors or even channel-like structures conducting vesicles one by one to the cell limits (see Fig. 6). After observation of experiments in dozens of chromaffin cells labelled with quinacrine there seems to be a clear relationship between vesicle and F-actin mobility, vesicles moved on the Factin structures and avoided crossing the cytoplasm 'dark' space visualised in the transmitted light images. In addition, virtually any experimental condition affecting actin-myosin produced a decrease in both the mobility of the cytoplasm structures visualised by transmitted scanning microscopy and vesicles. It is interesting to note that by analysing modifications in image intensity and even tracking vesicles and threshold transmitted light images, we have estimated that vesicles move at a four to five times faster speed than the changes observed in the F-actin network. In consequence, even if vesicle mobility is dependent on F-actin-myosin II structural activity to create the vectors for directing movements, vesicle mobility along the dynamic tracks may have a fast component dependent on the activity of molecular motors such myosin V (Rosé et al., 2003;Rudolf et al., 2003).
A model to understand the double role of F-actin as both a barrier and a support for vesicle access to the subplasmalemmal area in chromaffin cells Even if the combination of pharmacological tools and the observed distribution and colocalisation of the transmitted light structures with phalloidin, GFP-β-actin, and myosin II RLC labelling may constitute convincing evidence of its F-actin identity, the complexity of the changes occurring in these structures during cell stimulation is additional and important proof of its nature. As shown in Figs 5-8, using both transmitted light and confocal observation of GFP-β-actin, three types of major change were characterised in the cytoplasmic F-actin distribution after secretory stimulation: fast and discrete disruptions in the cortical peripheral barrier, formation of channel-like structures perpendicular to the membrane plane, and slower changes in the interior of the cytoplasm with the appearance of the formation of wider open spaces devoid of F-actin. The formation of cortical disruptions in response to different secretagogues was the key element supporting the role of peripheral F-actin as an effective barrier avoiding vesicle access to secretory active sites (Aunis and Bader, 1988;Vitale et al., 1995). Our experiments are the first to show the real-time dynamics of this process, in resting conditions, vesicles tend to locate within 200-400 nm of the membrane, following secretagogue superfusion and within 5-10 seconds, F-actin reorganises itself forming cortical disruptions of about 0.5-2 µm and the vesicles penetrate to previously forbidden areas in the cell limits. After 10-20 seconds of stimulation the general disposition of F-actin, that was parallel to the membrane, changed to a perpendicular alignment increasing its concentration in deeper subcortical regions, thus forming frequently channel-like open spaces of about 300-400 nm used by granules to access the membrane. A static view of these structures was previously shown when the cortical region of rhodamine-phalloidin-labelled chromaffin cells was studied by z-reconstruction of confocal images (Ñeco et al., 2003). Vesicles using these channel-like structures will be accessing docking sites 10-20 seconds after initiation of secretion and therefore will refill the pools of docked vesicles released during sustained stimulation. Finally, after 60-100 seconds, F-actin reorganisation in response to transient stimuli recovered the original distribution of parallel to the plasma membrane and again vesicles were restricted in their access to active sites. This reorganisation of F-actin is of extreme importance as it provides the basis to understanding the double nature of the cytoskeletal network of F-actin as a dense barrier that does not allow vesicle access to the docking sites and also, after stimulation and reorganisation, as providing the system to transport vesicles from the cytoplasmic pool to specific points of the peripheral area. In contrast to these rapid changes, continuous cell stimulation also induced the slow formation of intracellular spaces lacking F-actin with the aspect of empty polygons. We do not know the functional relevance of this process as vesicles do not seem to access the interior of these zones as they only move only along the F-actin walls. The formation of these polygons during secretion is supported by electron microscopy studies in fixed chromaffin cells (Tchakarov et al., 1998).
Taking these observations together, an active role as both a barrier and a carrier for F-actin-myosin is suggested in this neuroendocrine model, but our studies do not exclude another scaffold function as has been proposed in presynaptic terminals (Sankaranarayanan et al., 2003). We observed that these transmitted light structures are also easily visible and dynamic in the cell body of motoneurons and hippocampus neurons, therefore the possible application of transmitted light scanning microscopy to the study of secretion in neuronal and other cells could open new alternatives for the study of transport and secretion processes in living cells.