Distinct subcellular localization of calcium binding S100 proteins in human smooth muscle cells and their relocation in response to rises in intracellular calcium

Calcium ions acting as second messengers transduce extracellular signals into a wide variety of intracellular responses and thereby regulate many different biological processes such as secretion, proliferation, differentiation, transcription, apoptosis, and muscle contraction (for reviews see Parekh and Penner, 1997; Berridge, 1997). In recent years it has become clear that a number of human diseases, including cardiomyopathies, hypertension, neurodegenerative and neoplastic disorders are linked to altered Ca²⁺ handling mediated by Ca²⁺-binding proteins (Heizmann and Braun, 1992, 1995; van Eldik and Griffin, 1994; Richard et al., 1995; Polans et al., 1996; Heizmann, 1996).

The largest family of Ca²⁺-binding proteins shares a common structural motif, the EF-hand, named after the E- and F-helices of parvalbumin. The EF-hand domain consists of a helix-loop-helix motif that binds Ca²⁺selectively and with high affinity (Kretsinger, 1980; Nakayama and Kretsinger, 1994). Whereas calmodulin, the most prominent member of the EF-hand protein family, is ubiquitously expressed and multifunctional (Cohen and Klee, 1988; James et al., 1995), most of the other EF-hand proteins are expressed in a tissue- and cell-specific manner (Heizmann and Hunziker, 1991). This is also true for the S100 proteins representing the largest subfamily of EF-hand Ca²⁺-binding proteins (Donato, 1991; Hilt and Kligman, 1991; Fano et al., 1995; Schäfer and Heizmann, 1996).

S100 proteins are acidic proteins of low molecular mass (10-12 kDa), containing two distinct EF-hands with significantly different affinities for Ca²⁺. Both EF-hands are flanked by hydrophobic regions at either terminal and are separated by a central hinge region. The carboxy-terminal EF-hand contains the canonical Ca²⁺-binding loop encompassing 12 amino acids, whereas the amino-terminal loop consisting of 14 amino acids is specific for S100 proteins. To date, some 17 different proteins have been assigned to the S100 protein family that display various degrees of amino acid sequence homology in the range of 25% to 65%. At least thirteen S100 genes were found to be clustered on human chromosome 1q21 (Engelkamp et al., 1993; Wicki et al., 1996a,b), leading to the introduction of a new S100 nomenclature (Schäfer et al., 1995). The localization of the S100 gene cluster on human chromosome 1q21 is of special interest since a number of rearrangements (deletions, translocations, duplications) have been found to occur in this chromosomal region in cancer cells (Gendler et al., 1990; Hoggard et al., 1995; Forus et al., 1995; Bartoli et al., 1996; Weterman et al., 1996).

Alterations in S100 expression have been demonstrated in diseases such as Down’s syndrome (Allore et al., 1988), Alzheimer’s disease (van Eldik and Griffin, 1994), chronic inflammation (Rammes et al., 1997), and cardiomyopathies (Remppis et al., 1996).

S100 proteins are thought to exert their effect through Ca²⁺-regulated interactions with specific intracellular target proteins. Some intracellular S100 target proteins are myosin (Burgess et al., 1984), tubulin (Donato et al., 1989), microtubule-associated tau-proteins (Baudier and Cole, 1988), glial fibrillary acidic protein (Bianchi et al., 1996), tropomyosin (Gimona et al., 1997), phosphoglutamase (Landar et al., 1996), twitchin kinase (Heierhost et al., 1996), cytosolic phospholipase A2 (Wu et al., 1997), Ca²⁺ release channel (ryanodine receptor) of the sarcoplasmic reticulum (SR) (Treves et al., 1997), and annexins (Garbuglia et al., 1996).

Here we demonstrate that smooth muscle cells represent a valuable model system to investigate the distinct subcellular localization and the functional role of S100 proteins since these cells co-express S100A1, S100A2, S100A4, and S100A6. S100A1, found to be expressed in muscle tissues (Engelkamp et al., 1992; Pedrocchi et al., 1993; Song and Zimmer, 1996; Remppis et al., 1996), was also shown to activate invertebrate giant protein kinases involved in the regulation of muscle function (Heierhorst et al., 1996), and modulates adenylate cyclase activity (Fano et al., 1989a) and the Ca²⁺-induced Ca²⁺ release in different types of muscle cells (Fano et al., 1989b; Treves et al., 1997). S100A2, expressed in a number of tissues, is of particular interest because it is down-regulated in tumor cells, suggesting it may be a candidate tumor suppressor gene (Lee et al., 1992; Wicki et al., 1997). S100A4 has been identified as playing a causal role in the metastatic behavior of different cancer cells in rodents (Davies et al., 1996). Moreover, it has been shown that S100A4 appears in association with tropomyosin as well as with actin stress fibers (Takenaga et al., 1994). Finally, S100A6 is expressed in the cytoplasm of epithelial cells, fibroblasts, neuronal and muscle cells, and is overexpressed in a number of tumor tissues (for review, see Schäfer and Heizmann, 1996), and it has been shown to specifically interact with annexins (Watanabe et al., 1993a), caldesmon (Skripnikova and Gusev, 1989), and other proteins (Filipek et al., 1996; Filipek and Wojda, 1996; Tokumitsu et al., 1992).

In the present work, we investigated the distinct subcellular localizations of S100 proteins. For this purpose, S100 proteins were visualized by immunofluorescence using confocal laser-scanning microscopy. Our results indicate that S100 Ca²⁺-binding proteins exhibit a distinct subcellular localization pattern in human vascular smooth muscle cell lines and in primary cell cultures derived from smooth muscle tissue. We provide evidence that the S100A1 protein specifically interacts with filamentous actin (F-actin) in smooth muscle cells, thus possibly representing a new regulatory system together with calponin and caldesmon in modulating smooth muscle contraction. In addition, they specifically translocate within these cells after an increase of the intracellular Ca²⁺ concentration. These experiments provide new insights into the function of S100 proteins through their interaction with different target proteins in the Ca²⁺-signal transduction cascade.

All experiments were performed with two stable smooth muscle cell lines derived from human aorta (HVSMC ATCC-13145 CRL-1999) and from human jejunum (HISM ATCC CRL-1692). Cells obtained from the American Type Culture collection (Maryland, USA) were grown in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 1% L-glutamine, 10% fetal bovine serum, 100 units/µl penicillin and streptomycin (all from Gibco BRL). The cells were incubated in a humidified incubator (Heraeus, Switzerland) at 5% CO₂and 37°C. Cells in passages 3 through 5 were used for analysis.

Primary cultures derived from human arterial smooth muscle tissue were a kind gift from Dr Z. Yang, University of Zurich, Switzerland. Briefly, cells were prepared from the left internal mammary artery of patients undergoing aorto-coronary bypass surgery. The work was performed in accordance with the requirements of the institutional review committee for the use of human material. Immediately upon tissue removal samples were taken to the laboratory and processed within 30 minutes. Following removal of the adventitia, tissues were minced into blocks of approximately 1 mm³. Tissue blocks were transferred to a sterile 25 cm² culture flask covered with fresh DMEM plus 10% fetal bovine serum, 1% L-glutamine, and 100 units/µl penicillin and streptomycin. Flasks were placed in a humidified incubator for 6 hours to allow time for the tissue blocks to attach to the culture surface. After 6 hours, fresh DMEM containing L-glutamine, antibiotics, and FBS was added and flasks were returned to the incubator for the next 48 hours. Confluent cells between passage 0 and 2 were used for experiments.

Polyclonal antisera against human recombinant S100A1, S100A4, and S100A6 proteins were raised in goats, and the polyclonal antiserum against human recombinant S100A2 in rabbits. These primary polyclonal antibodies are specific and only recognize the corresponding antigen. They do not cross-react either with other S100 or Ca²⁺-binding proteins or with other cellular proteins (Ilg et al., 1996). The mouse monoclonal antibody (mAb) G1/296 raised against a novel 63 kDa membrane protein (p63), which identifies the SR in the cytoplasm of primate cells (Schweizer et al., 1993), was a kind gift from Dr H. P. Hauri, Biocenter of the University of Basel.

A fluorescent derivative of phalloidin, a mushroom toxin recognizing all isoforms of F-actin (FITC-phalloidin), was purchased from Sigma (Biosciences, USA). The secondary Cy3-coupled rabbit anti-goat and Cy3-coupled goat anti-rabbit antibodies were from Sigma (Biosciences, USA), and the donkey Cy2-coupled anti-mouse antibody was purchased from Amersham (Life Science, USA). CyDye is a mixture of fluorescent dyes (cyanine fluors) for labeling proteins.

HVSMC and freshly isolated arterial human smooth muscle cells were transferred into 24-well cell culture plates containing 1 ml/well of growth medium. For further processing, this medium was replaced by modified Ca²⁺-free Hanks’ buffer (MHB), containing 2 mM EGTA and 5 mM MES (2-morpholino-ethanesulfonic acid), pH 6.2 to 6.4 (Small and Celis, 1978), and then quickly substituted with ‘permeabilization buffer’, i.e. MHB containing 2% octyl-POE (n-octylpolyoxyethylene; ALEXIS Corporation, Switzerland) and 0.125% glutaraldehyde (Electron Microscopy Sciences, USA). Time from detachment to permeabilization was kept under 3 minutes. After 5 minutes of permeabilization and prefixation, cells were fixed for 20 minutes with MHB containing 1% glutaraldehyde. The HVSMC and the primary cell cultures were then washed 3 to 4 times with MHB. Aldehyde groups were reduced by treating the cells twice for 10 minutes each with NaHB₄ (0.5 mg/ml) in MHB on ice. Immunofluorescence labeling was carried out as described (Baschong et al., 1997) by incubating the cells with appropriate concentrations of primary and fluorochrome-conjugated secondary antibodies. Finally, the cells were washed with MHB, mounted bottom-up (i.e. inverted prior to mounting) in Mowiol 4-88 (Hoechst, Germany) containing 0.75% n-propyl-gallate as an anti-bleaching agent. Mounted slides were left to dry for 24 hours at room temperature in the dark, and then stored at 4°C in the dark until viewed.

For controls, the cells were incubated either with pre-immune sera or with polyclonal anti-S100 antibodies preabsorbed with the corresponding antigens. For this purpose 50 µg of recombinant S100 protein (Pedrocchi et al., 1994) was added to 100 µl of undiluted polyclonal anti-S100 antibodies and incubated for 10 hours at 4°C. After centrifugation at 14,000 g the supernatant was taken for control stainings.

A rise of intracellular [Ca²⁺] in smooth muscle cells was achieved by treatment with 1 µM thapsigargin (12 minutes at 37°C; Molecular Probes, USA). Cells were then washed 3 times with MHB for 2 minutes each, fixed, permeabilized, and immunolabeled as described above.

Micrographs were taken with a confocal microscope consisting of a Zeiss Axiovert fluorescence microscope with a Zeiss Plan Apo 63/1.4 oil objective lens and an Odyssey XL confocal laser-scanning unit (NORAN, USA), driven by the Intervision software package run on an INDY workstation (Silicon Graphics Inc., USA). The light source was an Argon laser tuned so that the exitation wave length for Cy3 was 529 nm, and that for FITC and Cy2 488 nm. The colocalization images were prepared with the ‘Imaris’, ‘Voxel Shop’, and ‘Colocalization’ software products (Messerli et al., 1993), available from Bitplane AG (Zürich, Switzerland).

The interaction of S100A1 and S100A6 with F-actin was evaluated by high-speed centrifugation. To this end, 5 µM G-actin (purified from rabbit skeletal muscle as described by Bremer et al., 1994) at a protein concentration of 1 mg/ml (i.e. 24 µM) was polymerized to F-actin in imidazole-buffer A (2.5 mM imidazole, 0.005% NaN₃, 0.2 mM CaCl₂, 0.2 mM ATP, pH 7) by adding KCl to 50 mM and 1 mM MgCl₂ (120 minutes, at room temperature). F-actin was then incubated for 120 minutes at room temperature with 25 µM of human recombinant S100A1 or S100A6 in buffer A containing either 1 µM Ca²⁺ or 1 µM EGTA. For controls, the proteins were incubated in buffer A, also containing either 1 µM Ca²⁺ or 1 µM EGTA. After incubation the mixtures were centrifuged at 100,000 g for 15 minutes. Supernatants and pellets were analyzed by SDS-PAGE.

The interaction of S100A1 with F-actin filaments was visualized by negative staining. Briefly, a 5 µl aliquot of the sample was absorbed for 1 minute onto glow-discharged (Aebi and Pollard, 1987) carboncoated, 400-mesh/inch copper grids. The grid was washed on two drops of deionized water and then placed on two drops of 0.75% uranyl formate (pH 4.25) for 15 seconds each. The specimens were inspected in a Zeiss 910 TEM operated at 100 kV, and electron micrographs were recorded on Kodak S0-163 electron image film (Eastman Kodak Co.) at a nominal magnification of ×50,000. The exact magnification was calibrated according to the method of Wrigley (1968).

To investigate the subcellular localization of human S100 proteins in HVSMC, immunofluorescence staining using S100-specific polyclonal antibodies and fluorescent secondary antibodies was applied to smooth muscle cells from aorta. The fluorescence micrographs in Fig. 1 display two different optical sections through a cell that was stained with an anti-S100A1 primary antibody. The section taken close to the tip of the cell (distal face of the cell in Fig. 1A) exhibits distinct reticular network-like staining around the cell nucleus. In contrast, in the optical section close to the glass coverslip (proximal face of the cell in Fig. 1B) S100A1-specific labeling is predominantly revealed in the peripheral regions of the cell, yielding a stress fiber-like pattern. No significant S100A1 staining was found in the cell nuclei. These initial results suggested a distinct localization of S100A1 in HVSMC, i.e. the protein was found exclusively in the cytoplasm of the cells where it was strongly associated with stress fibers and the SR. Fig. 2 shows the subcellular distribution of S100A4 in HVSMC. For the best possible comparison with S100A1, sections were recorded at the same cross-sectional level within the cells as in Fig. 1A and B. Accordingly, the subcellular localization of S100A4 appears to be very similar to that of S100A1. The anti-S100A4 antibody produced a stress fiber-like labeling in the cell periphery simultaneously with an extended SR-like membraneous network labeling pattern around the cell nucleus. These findings prompted us to make a detailed qualitative and quantitative analysis of the S100 protein colocalization patterns within distinct cell structures (see below).

Fig. 3 reveals two smooth muscle cells after immunofluorescence staining for S100A6. The optical sections (Fig. 3A and B) were recorded equivalent through two different cells. In 90% of the cells (Fig. 3A) the S100A6 antibodies labeled a reticular network-like structure around the cell nucleus similar to the S100A1 and S100A4 antibodies. In contrast to S100A1 and S100A4, S100A6 showed no stress fiber-like staining and was found in the cell nucleus in 10% of labeled cells (Fig. 3B). On the other hand, no stress fiber-like staining was revealed for S100A6.

When cells were immunofluorescence-stained for S100A2, yet another localization pattern was observed (Fig. 4). Different optical sections through the same cell (Fig. 4A,B) showed that S100A2 was found only in the cell nucleus. Reconstruction of the complete 3D distribution of S100A2 after recording of 30 optical sections through the labeled cells confirmed that localization of S100A2 was confined to the cell nucleus (data not shown). These staining patterns revealed a dispersed organization of punctate stained structures, which were prominent in the center of the nucleus.

The intracellular localization pattern of S100 proteins in HVSMC were identical to those found in human smooth muscle cells derived from jejunum (HISM) or human vascular smooth muscle tissue (data not shown).

In order to compare the specific localization patterns of S100 proteins in stable smooth muscle cell lines with the in vivo situation, freshly isolated human arterial smooth muscle cells were labeled with the S100-specific antibodies as already described. The fluorescence micrographs in Fig. 5 display the subcellular distribution of S100A1 and S100A4 in primary cultures of human smooth muscle cells. The reticular network-like staining around the cell nucleus (Fig. 5A and C) and the stress fiber-like structures in the periphery of the cells (Fig. 5B and D) are identical to those found in subcultured HVSMC. The staining pattern of S100A2 and S100A6 in primary cell cultures and cell lines is also identical (not shown).

As documented in Figs 1 and 2, S100A1 and S100A4 appear to be associated with stress fiber- and SR-like structures. To positively identify these structures, colocalization studies were performed. A specifically designed ‘colocalization’ software for confocal images (Bitplane, AG; Switzerland) was applied for qualitative and quantitative analysis of the degree of colocalization of two individually labeled proteins in the cell. For double immunofluorescence, first the respective S100 protein was labeled with the primary and secondary Cy3-coupled antibody. Second, actin filaments were visualized by FITC-conjugated phalloidin, known to recognize F-actin such as in stress fiber-like structures, independently of the actin isotype. Fig. 6 illustrates the specific colocalization of S100A1 with F-actin. Fig. 6B reveals an optical section of a double-stained cell close to the substrate. The fluorescence signal from the green channel (FITC) indicates actin staining and that from the red channel (Cy3) S100A1 labeling. The overlap of both fluorescence signals produces a yellow signal, demonstrating a high degree of colocalization between F-actin and S100A1. Fig. 6A represents a 2-D histogram of the simultaneously recorded green and red channels. To establish this 2-D histogram, a collection of voxels (volume elements defined by the confocal scanning procedures) representing the two fluorescence channels, was analyzed. For each voxel a point was generated in the 2-D histogram with the x-coordinate representing the intensity of the voxel in the green channel (for actin-staining) and the y-coordinate representing the intensity of the voxel in the red channel (for S100A1 staining). A region of interest excluding the background staining was defined (yellow frame), and the selected voxels were used for further statistical analysis and visualization. Starting from this data set, all voxels with identical intensity for both fluorescence signals were identified. Hence, this finite set of voxels comprises only signals arising from image elements that contain information from both fluorescence channels. The true colocalization (i.e. inters off all volume elements harboring the same intensity fluorescence signal in the two channels) of the labeled proteins can in this fashion be reliably demonstrated. As a result, a colocalization image consisting of S100A1- and actin-containing structures is generated in each of the observed optical sections (Fig. 6C). The statistical analysis of the colocalization image demonstrates a significant colocalization between S100A1 and actin filaments in this optical section.

In order to investigate the complete 3-D association between S100A1 and actin, 20 optical sections through several stained cells were recorded at an interval of 0.12 µm in the z-direction, and 20 colocalization data sets for each cell were generated (data not shown). Table 1A represents the colocalization measurements obtained in a single cell. S100A1 and F-actin appeared with different degrees of association between each other, due to the irregular distribution of the two proteins in the cells. The tips of the cells contain no stress fibers and therefore the integrated spatial analysis indicates very low colocalization ratios for the layers in these regions (layers 12-20).

Fig. 6 (D-F) shows the colocalization data of S100A4 with F-actin. This is very similar to the data obtained with S100A1. The results demonstrate that there is significant colocalization in the spatial arrangement of S100A4 and actin filaments in the peripheral regions of cells (for details see Table 1b). Colocalization is highest close to the substrate (i.e. near the bottom of the cell layers 1-7), again reflecting the intracellular distribution of actin and S100A4. Taken together, these results suggest that S100A4, like S100A1, is closely associated with the actin-stress fiber systems in cultured vascular smooth muscle cells.

The ability of some S100 proteins to bind directly to F-actin was further investigated using a sedimentation assay (Fig. 7). S100A1 in F-actin filament promoting buffer (see Materials and Methods) by itself did not sediment upon centrifugation at 100,000 g for 30 minutes in the presence or absence of 1 µM Ca²⁺ (Fig. 7A). However, when S100A1 and highly purified F-actin from rabbit skeletal muscle were mixed at appropriate ratios in the presence or absence of Ca²⁺ for 120 minutes, Ca²⁺-dependent co-sedimentation after high-speed centrifugation was observed (Fig. 7B, lane 4), indicating a direct interaction of S100A1 with F-actin. For comparison, the same sedimentation assay was also performed with F-actin and S100A6, the subcellular localization of which is distinct from that of actin stress fibers. As expected, S100A6 remained in the supernatant both in the presence and absence of Ca²⁺ (Fig. 7C, lanes 1-4).

Unfortunately, sedimentation experiments could not be performed with S100A4, which is known to form oligomers (Ilg et al., 1996) and was found to aggregate by itself in F-actin forming buffer.

The direct interaction of F-actin (Fig. 8A) with S100A1 was further confirmed by electron microscopy of negatively stained specimens. As documented in Fig. 8B, S100A1 was associated with F-actin filaments in the presence of 1 µM Ca²⁺. S100A1/F-actin complexes were visualized as partially decorated filaments with a low background, representing unbound S100A1. Since S100A1 has a low molecular mass (11 kDa), it is not surprising that the decoration pattern of the actin filaments is not immediately obvious when compared to native F-actin. When the same experiments were performed in the absence of Ca²⁺, S100A1 was unable to bind to actin filaments and remained as monomers in the background (Fig. 8C), consistent with the results obtained above. These data indicate that S100A1 has the intrinsic capacity to directly interact with F-actin.

A reticular network-like distribution of S100A1, S100A4, and S100A6 proteins around the cell nuclei was demonstrated in HVSMC, indicating a possible association of these proteins with the SR in addition to their association with actin stress fibers. To date it is known that Ca²⁺ homeostasis in smooth muscle cells is regulated by two membrane systems: the plasma membrane and the SR. Since S100 proteins have been found to stimulate adenylate cyclase in the SR (Fano et al., 1989a) and Ca²⁺-induced Ca²⁺ release from the SR in muscle cells (Fano et al., 1989b; Treves et al., 1997), it has been speculated that they accumulate in the SR of smooth muscle cells. To investigate the association between S100 proteins and the SR in more details, a colocalization approach was used as described above. Visualization of the SR was achieved by staining the cells with a primary monoclonal antibody directed against a novel 63 kDa SR-membrane bound protein (p63), which recognizes the SR membranes in human cells (Schweizer et al., 1993). Fig. 9 shows the colocalization of S100A1, S100A4, and S100A6 with the SR-membrane bound p63 in HVSMC. The optical sections displayed here depict a central region around the cell nucleus. Data concerning the spatial relationship of the S100 proteins and the SR were obtained by acquiring a series of 20 optical sections of each cell spaced at 0.16 µm in the z-direction. Statistical analysis of these optical sections revealed a similar, high degree of colocalization of S100A1, S100A4, and S100A6 with the SR. The section by section numeric evaluation from this quantitative colocalization analysis is presented in Table 2. The colocalization ratio here indicates that for all these S100 proteins the highest degree of colocalization is observed in the middle of the cell (for S100A1 in layers 9-16; for S100A4 in layers 4-12, and for S100A6 in layers 2-14). These findings suggest that the reticular-like organization of these S100 proteins is due to their association with the SR surrounding the cell nucleus.

In summary, we conclude that S100A1, S100A4, and S100A6 are specifically located at the SR of human vascular smooth muscle cells.

In order to investigate whether a rise of intracellular [Ca²⁺] affects the subcellular localization of S100 proteins, HVSMC were treated with thapsigargin (a sesquiterpine lactone) known to elevate cytosolic Ca²⁺ by blocking the SR-Ca²⁺-ATPase in many cells, including vascular smooth muscle cells (Thastrup et al., 1990).

Fig. 10 shows relocation of S100A1, S100A4, and S100A6 but not of nuclear S100A2 upon thapsigargin exposure.When the cytosolic Ca²⁺ concentration rose, the SR-associated S100A1 (Fig. 10A), S100A4 (Fig. 10B), and S100A6 (Fig. 10C) relocated to vesicle-like structures in the central region around the cell nucleus. The intracellular localization pattern of S100A2 (Fig. 10D), the nuclear localization pattern of S100A6, and the S100A1 and S100A4 associated with F-actin (data not shown), however, were unaffected. Individual vesicles with high content of S100A1, S100A4, and S100A6 proteins could be clearly distinguished by serial sectioning along the z-axis through the cells (data not shown).

From these observations we conclude that local Ca²⁺release leads to relocation only of the SR-associated S100 protein moiety.

Impairment of Ca²⁺ homeostasis and altered expression of Ca²⁺-binding proteins have been associated with a number of muscle disorders. Members of the S100 protein family deserve particular scrutiny because of their cell-specific expression pattern and their association with a variety of proteins known to modulate muscle contraction. Previous investigations into the physiological role of S100 proteins and their intracellular localization in muscle cells have been hampered by the lack of an appropriate in vitro model system. The present study: (i) demonstrates that smooth muscle cells (cell lines and primary cultures) are a suitable model for investigating intracellular distribution of co-expressed S100 proteins, and (ii) it provides new information regarding the specific interaction of S100 proteins with distinct intracellular structures and their physiological functions.

First, the distinct intracellular localization of S100 protein in smooth muscle cells and their association with specific intracellular structures is reported. Specifically, S100A1 and S100A4 were found only in the cytoplasm, where they were strongly associated with the SR and with actin stress fibers. S100A6 was also localized in the SR, but in addition, it also resided in the cell nucleus. S100A2 was located only in the cell nucleus. To the best of our knowledge, these data are the first to demonstrate a differential localization pattern of S100 proteins in smooth muscle cells.

S100 proteins have been reported to interact with numerous cytoskeletal proteins. Among others, tubulin, desmin, and junctional membrane protein (Donato, 1991; Garbuglia et al., 1996) have been identified as targets of S100 proteins. Recent studies have also demonstrated distinct localization patterns of S100 proteins in muscle cells (Haimoto and Kato, 1988; Zimmer and Landar, 1995). However, these localization and interaction experiments were performed with protein samples that were extracted from whole organs, thus containing different types of S100 proteins according to tissue distribution in the corresponding organ. As was demonstrated by Ilg et al. (1996), antibodies raised against these proteins as well as the commercially available anti-S100 antibodies crossreact with other family members. Hence, the use of such isolated and purified proteins and antibodies for interaction and localization studies limits interpretation of these previously published results.

Only the recent development of human recombinant S100 proteins together with the generation of specific anti-S100 antibodies allowed us to demonstrate unambiguously the distinct intracellular distribution patterns of different S100 proteins in human smooth muscle cells.

Both single- and double-fluorescence experiments combined with an integrated colocalization analysis clearly demonstrated that S100A1 and S100A4 appear to be strongly associated with actin stress fibers (see Fig. 6). The functional significance of the association of S100 proteins with actin filaments, however, remains elusive, although several possibilities are being evaluated. It has been suggested that interaction of S100 proteins with caldesmon and calponin modulates their interaction with F-actin and therefore smooth muscle contraction (Pritchard and Marston, 1991; Fuji et al., 1994), but direct evidence for such interactions in living cells has not been reported.

Another intriguing piece of evidence underscoring the observed in situ association between S100 proteins and the actin cytoskeleton using immunofluorescence was the ability of S100 proteins to directly bind to F-actin in vitro, as assessed by a sedimentation assay and subsequent electron microscopy (Figs 7, 8). This set of data clearly demonstrates that S100A1 can interact directly with F-actin filaments in a Ca²⁺-dependent manner with a molar ratio of approximately five S100A1 molecules per one actin molecule. This finding is in line with the results of Watanabe et al. (1993b), who have previously described a Ca²⁺-dependent interaction of F-actin with calvasculin, a protein that is highly homologous to S100A4.

Our present results demonstrating the association of S100A1 and S100A4 with actin stress fibers in vivo strengthen the notion that members of the S100 protein family are involved in the regulation of actin filament polymerization. Takenaga et al. (1994) described a direct association of S100A4 with tropomyosin, thus proposing a direct regulatory function of S100A4 in the tropomyosin-actin interaction. Heierhorst et al. (1996) reported that a myosin-associated giant protein kinase, present in striated and smooth muscle along with sarcomeric proteins, is activated by S100 in a Ca²⁺-dependent manner, thereby possibly constituting a novel, third level of Ca²⁺ regulation in muscle. Taking these results together with our present findings, we propose that S100 proteins play an important role in the organization of the actin cytoskeleton via a Ca²⁺-sensitive interaction with F-actin and actin-associated proteins. In vitro reconstitution experiments with native muscle thin filaments are in progress to systematically investigate this possibility.

The results from our colocalization experiments with a novel marker protein for human SR (Schweizer et al., 1993) indicate that S100A1, S100A4, and S100A6 are also closely associated with the SR of human smooth muscle cells. These findings correlate with previously described interactions of S100 proteins with contraction-modulating enzymes located in the SR of muscle cells. For example, Fano et al. (1989a,b) and Treves et al. (1997) reported that S100 proteins stimulate the Ca²⁺-induced release of Ca²⁺ from the SR as well as the basal Mg²⁺-activated SR-associated adenylate cyclase. Both Ca²⁺-dependent intracellular effects play a central role in the pathophysiology of muscle disorders with an altered contractile activity. Whether S100A1, S100A4, and S100A6 interact with modulators of muscle contraction, and whether these proteins regulate the activity of such modulators, requires further examination. Nevertheless, at this stage we can only speculate that the observed distribution of S100 proteins in human smooth muscle cells suggests their participation in fundamental Ca²⁺-dependent cell activities.

In contrast to the distribution of S100A1 and S100A4, a completely different localization pattern was revealed for S100A6, and particularly, for S100A2. These two proteins were found to be localized in the cell nucleus. There is now compelling evidence that Ca²⁺ ions play an important role in the regulation of several nuclear functions (for review, see Gilchrist et al., 1994; Bachs et al., 1994). This regulation system acts through activation of Ca²⁺-binding proteins, located in the cell nucleus. However, another important function of Ca²⁺-binding proteins may be the regulation of intracellular Ca²⁺ concentrations via perinuclear Ca²⁺ stores residing in the lumen of the nuclear envelope and the SR. Whether S100 proteins are directly involved in these events in the cell nucleus and the nuclear envelope is still not known. However, the distinct nuclear localization pattern of S100A2 and S100A6 in smooth muscle cells implies that these proteins do take part in fundamental nuclear functions.

It is widely believed that cell function and survival are critically dependent on the precise regulation of the intracellular Ca²⁺content. This intracellular Ca²⁺regulation in vascular smooth muscle cells is crucial because it is a primary factor in the regulation of muscle contraction via activation of Ca²⁺-binding proteins. However, as yet the exact nature of the activation mechanism for S100 proteins is only poorly understood. Therefore, we investigated whether treatment of vascular smooth muscle cells with thapsigargin, which is known to increase the cytosolic [Ca²⁺] within the physiological range, affects the intracellular localization of S100 proteins. In the course of these investigations, we found that prolonged (12 minutes) rise in the intracellular [Ca²⁺] causes distinct relocation of the cytoplasmic but not of the nuclear S100 proteins, thereby yielding vesicles with high S100 content in the region of the SR. This observation is in agreement with previous studies by Subramanian and Meyer (1997) who also demonstrated that the structural integrity of the SR remained preserved during short Ca²⁺ transients or Ca²⁺ store depletion. In contrast, persistent Ca²⁺ increase lasting longer than 10 minutes leads to fragmentation of the SR into vesicles. Such persistent Ca²⁺ overload in muscle cells is thought to be the basic pathophysiological phenomenon in a variety of cardiac and smooth muscle disorders related to altered contraction. Therefore, we speculate that specific relocation of S100 proteins in response to rise of intracellular [Ca²⁺] is involved in the alterations of fundamental cellular functions in smooth muscle.

We are very much indebted to Dr M. Steinmetz for teaching us the actin-binding assay. A. Hefti is gratefully acknowledged for providing help with electron-microscopy. We are also indebted to Dr H. P. Hauri for kindly providing antibodies against p63, and to Dr A. Rowlerson and M. Killen for critical reading of the manuscript. This study was supported by the Swiss National Science Foundation (grant no. 31-50510.97 to C.W.H. and no. 32-49126.96 to D.A.); BIOMED 2 (European Union grant no. BMH4CT950319/BBW grant no. 95.0215-1, Switzerland); the Swiss Heart Foundation; the Swiss Society of Cardiology; the Ciba-Geigy Foundation; and the Sandoz Foundation. D.A. is a recipient of a career development award from the Foundation Professor Max Cloëtta, Zurich.