Nonmuscle tropomyosin-4 requires coexpression with other low molecular weight isoforms for binding to thin filaments in cardiomyocytes

Tropomyosins (TMs) are a family of actin-filament binding proteins that are expressed in most eukaryotic cells. In vertebrates, different isoforms are expressed in muscle (skeletal, cardiac and smooth) and nonmuscle cells. TMs are encoded by multiple genes but diversity is also generated by alternative RNA splicing. At least 18 different TM isoforms are expressed from four separate genes in vertebrates by a combination of alternative RNA splicing and alternative promoter usage (reviewed in Pittenger et al., 1994; Lin et al., 1997). Furthermore, alternative RNA processing for the generation of TM isoform diversity is a common mechanism found in a number of organisms including nematodes, flies, frogs, birds and mammals, demonstrating that alternative splicing and TM isoform diversity arose fairly early in the evolution of multicellular organisms. The complexity of TM isoform expression is probably not due to functional redundancy as this complexity has been largely maintained through evolution, suggesting important functions that have been conserved through selective pressures. Despite great advances in our understanding of TM isoform diversity at the molecular level, the function of most of the isoforms still remains unresolved.

The function of TM in skeletal and cardiac muscles is, in association with the troponin complex (troponin-I, -T and -C), to regulate the calcium-sensitive interaction of actin and myosin (Ebashi et al., 1969; Greene and Eisenberg, 1980; Farah and Reinach, 1995). In contrast, the functions of the TM isoforms expressed in nonmuscle and smooth muscle cells are not clearly understood. These cells are devoid of a troponin complex, and the two calcium-sensitive regulatory mechanisms controlling the interaction of actin and myosin in smooth muscle and nonmuscle cells involve myosin light chain kinase and caldesmon (Adelstein and Eisenberg, 1980; Yamashiro-Matsumura and Matsumura, 1988). These differences in the regulation of the contractile system between various muscle and nonmuscle cell types appear to require structurally as well as functionally distinct isoforms of TM. Comparative amino acid and nucleic acid analyses of TMs from skeletal muscle, cardiac muscle, smooth muscle and nonmuscle cells show these proteins to be highly conserved. Nevertheless, structural differences do exist among the various protein isoforms. The regions that differ between isoforms correspond in large part to sequences encoded by the alternatively spliced exons located in the amino-terminal region, an internal exon and a carboxy-terminal exon. These regions appear to encode functional domains involved in head-to-tail binding, coiled-coil interactions (homodimers versus heterodimers), as well as actin, caldesmon, tropomodulin and troponin-T binding (reviewed in Pittenger et al., 1994; Lin et al., 1997). In addition, most nonmuscle cells, such as fibroblasts, express a greater number of TM isoforms (6-8) than do skeletal muscle (2 isoforms), cardiac muscle (1 isoform) and smooth muscle (2 isoforms) cells. Thus nonmuscle cells may have adapted TM isoforms to serve distinct functions.

The functional significance of TM isoform diversity remains to be determined. Evidence for distinct roles of the different nonmuscle isoforms comes from biochemical studies, which demonstrated differences among TM isoforms in their relative affinity for actin, ability to protect F-actin from the severing actions of gelsolin or villin, ability of caldesmon to promote their interaction with F-actin, and their effects on myosins I and II (reviewed in Pittenger et al., 1994; Lin et al., 1997). Additional evidence for the distinct role of TMs in the stabilization of actin filaments has been inferred from studies in transformed cells, where the loss of microfilament bundles has been correlated with the decreased expression of specific isoforms of TM (reviewed in Pittenger et al., 1994; Lin et al, 1997). Genetic studies in lower eukaryotes have also demonstrated a role for TMs in cytokinesis (Balasubramanian et al., 1992), vesicular transport (Liu and Bretscher, 1992) and localization of mRNA (Erdelyi et al., 1995; Tetzlaff et al., 1996). A number of studies have addressed the function of TM in vertebrate cells and have implicated a role for TM in intracellular granule movement (Hegmann et al., 1989; Pelham et al., 1996) and mitosis (Warren et al., 1995). In addition, mutations in TM have been associated with familial hypertrophic cardiomyopathies (FHC) and autosomal dominant nemaline myopathy (Thierfelder et al., 1994; Laing et al., 1995). Although it is known that TMs are expressed in a cell-type-specific manner, it remains to be determined if specific isoforms are critical for the assembly, integration and regulation of distinct actin-containing structures, such as sarcomeres of cardiac muscle or microfilaments of fibroblasts. TMs from striated, smooth muscle and brain cells are able to integrate into the microfilaments of fibroblasts (Wehland and Weber, 1980; Gimona et al., 1995; Dome et al., 1988). Whether fibroblast isoforms will be able to integrate into the thin filaments of skeletal muscle and cardiac muscle is not known. In one study, a mixture of TMs isolated from brain were found to incorporate into the I-bands of cardiac myocytes (Dome et al., 1988). However, brain contains a multiplicity of TM isoforms and the identity of the specific isoform(s) which incorporated was not known. In addition, immunofluorescence studies in developing skeletal and cardiac muscles revealed the presence of nonmuscle TMs associated with thin filaments, although the identity of the isoforms involved was not known (Wang et al., 1990).

In the present study the ability of different muscle (skeletal muscle α-TM) and nonmuscle (fibroblast TM-1, -2, -3, -4, -5, -5a or -5b) isoforms of tropomyosin to incorporate into neonatal rat cardiomyocytes (NRCs) was studied using GFP- and viral epitope (HA or VSV)-tagged TMs. The neonatal rat cardiomyocytes offer a unique experimental system to study intracompartmental sorting of muscle and nonmuscle cytoskeletal proteins, such as myosin light chains (Soldati and Perriard, 1991). All isoforms, except fibroblast TM-4, were able to incorporate into the I-band of NRCs. In addition, cells that incorporated the muscle or nonmuscle GFP-TMs into their myofibrils continued to beat and exhibit sarcomere shortening. Co-transfection of TM-4 with other low molecular weight isoforms of TM (TM-5, -5a or -5b) resulted in incorporation of TM-4 into the sarcomeres of NRCs. These results demonstrate that the ability of TM-4 to bind to actin filaments can be influenced by its interactions with other low molecular weight, but not high molecular weight TM isoforms, possibly by cooperative interactions via head-to-tail binding or formation of heterodimers between different TM subunits. The results have important implications for understanding the role of TM isoform diversity and their function in actin filament dynamics. Furthermore, these studies demonstrate that GFP-TM can be used to study thin-filament dynamics in muscle cells and actin filament dynamics in nonmuscle cells.

For the construction of GFP-TM-fusions, the ATG at the 5´ end of the cDNA sequences for fibroblast TM-1, -2, -3, -4, -5(NM1), -5a, -5b and skeletal muscle α-TM were fused in-frame with the 3´ end of GFP using the EGFP-C1 expression vector (Clontech). The cDNAs for these constructs have been described in detail elsewhere (Helfman et al., 1986; Yamawaki-Kataoka and Helfman, 1987; Goodwin et al., 1990; Temm-Grove et al., 1996), as well as the construction of HA- and VSV-TMs (Gimona et al., 1995; Temm-Grove et al., 1996).

Primary cultures of neonatal rat cardiomyocytes were prepared as described previously (Sen et al., 1988). Cells were seeded at a density of 0.4×10⁶ cells per 30 mm dish in 2 ml culture medium. The culture medium consisted of 68% Dulbecco’s MEM (Animed AG, Basel, Switzerland), 17% medium 199 (Animed), 10% horse serum (Gibco Laboratories, Grand Island, NY), 5% fetal calf serum (Gibco), 200 mM glutamine (Animed) and 1% penicillin-streptomycin (Animed). Prior to transfection, the cells were grown for 24 hours in 10% CO_2. Transfection was carried out using a modified CaPO₄ transfection method (Chen and Okayama, 1987). Briefly, the culture medium was changed to maintenance medium (Dulbecco’s MEM, 19% medium 199, 4% horse serum, 200 mM glutamine, 1% penicillin-streptomycin and 0.1 M phenylephrine) 1-4 hours prior to transfection. Approximately 3 µg of DNA were used per 30 mm dish. The precipitate was added to the medium and cells were incubated for 4-6 hours at 3% CO₂. After incubation, the cells were washed twice with Tris-buffered saline and incubated in maintenance medium at 10% CO₂. Fixation and immunofluorescence was carried out 24 or 72 hours after transfection.

Cells were transfected with various GFP-TM plasmids and observed after 24 to 72 hours incubation. In some cases the cells were recorded by video-microscopy. All cells used for live observations were plated on culture dishes with glass surfaces inserted into the bottom (MaTek Corp., Ashland MA, USA) and were analyzed live.

The immunofluorescence labeling was carried out as already described (Messerli et al., 1993a,b; Komiyama et al., 1996). The polyclonal anti-VSV-G epitope #49 antibody (Soldati and Perriard, 1991), as well as the monoclonal antibody B4 recognizing the M band of protein myomesin (Grove et al., 1984) were raised in our laboratory. The monoclonal antibody 12CA5 recognizing the HA epitope has been described (Niman et al., 1983). The F-actin-specific reagent phalloidin-TRITC, recognizing all isoforms of actin, was purchased from Molecular Probes (Molecular Probes Inc., Eugene, OR, USE). The secondary antibodies FITC-conjugated anti-mouse IgG+IgM, FITC- and TRITC-conjugated anti-rabbit IgG, as well FITC- and TRITC-conjugated anti-mouse IgG, were purchased from Cappel and Pierce (Westchester, PA, USA).

For confocal microscopy, the imaging system consisted of a Zeiss Axiophot fluorescence microscope, a Biorad MRC-600 confocal scanner (Biorad Lasersharp Ltd, Oxfordshire, England) and a Silicon Graphics Workstation (Silicon Graphics, Inc., Mountain View, CA, USA). The images were recorded using a Zeiss Neofluar 100×/1.3 NA objective or a Zeiss Neofluar 100×/1.4 NA objective. The system was equipped with an argon/krypton mixed gas laser. Image processing was done on the Silicon Graphics Workstation using ‘Imaris’ (Bitplane AG, Zurich, Switzerland), a 3-D multi-channel image processing software specialized for confocal microscopic images (Messerli et al., 1993), which is marketed by Bitplane AG (Technoparkstrasse 1, 8004 Zürich, Switzerland).

Live pictures were taken on a Zeiss Axiovert microscope with a Zeiss Plan Neofluar objective (63×/1.30) and a CCD camera (Kappa CF8/1 FMCC) and directly transferred to the computer using Neotech’s Image Grabber software, or alternatively recorded directly with a video recorder.

In order to follow the incorporation of different TM isoforms in living cells, the ATG methionine at the 5´ end of the cDNA coding sequences for skeletal muscle α-TM and fibroblast TM-1, -2, -3, -4, -5(NM1), -5a or -5b were fused in-frame with the 3´ end of the pEGFP protein. These isoforms are expressed from four different TM genes via alternative RNA splicing and, in some cases, the use of alternative promoters. A schematic diagram of the four genes and the various isoforms they express is shown in Fig. 1. Skeletal muscle α-TM and TM-1, -2 and -3 all contain 284 amino acids and are representative of high molecular weight (HMW) TM isoforms. TM-4, -5(NM1), -5a and -5b all contain 248 amino acids and are representative of low molecular weight (LMW) TMs.

In order to study the ability of GFP-tagged TMs to incorporate into neonatal rat cardiomyocytes, we first studied GFP-skeletal muscle α-TM, since skeletal muscle α-TM is the naturally occurring isoform expressed in cardiac muscle (Wieczorek et al., 1988). NRCs were transfected with either the control pEGFP plasmid or the pEGFP construct containing the coding sequence for skeletal muscle α-TM. Following the transfection, living cells were observed for 24-72 hours. Cells transfected with the control plasmid exhibited diffuse fluorescence (Fig. 2a). By contrast, cells expressing the GFP-skeletal muscle α-TM fusion exhibited a sarcomeric distribution of incorporation. In addition, the incorporation of the GFP-TM protein did not interfere with sarcomere shortening (Fig. 2b,c). The GFP-skeletal muscle α-TM fusion was observed to exhibit a sarcomeric pattern of incorporation as early as 24 hours following transfection, which was seen as long as 5 days. A video sequence of images of the GFP-skeletal muscle α-TM in living NRCs can be accessed at http://www1.cell.biol.ethz.ch/members/leu/gfp_home.htm1.

To verify that the GFP-TM was incorporated into the I-band of NRCs, cells transfected with control and GFP-TM plasmids were fixed and stained for filamentous actin using phalloidin and for the M-band using anti-myomesin antibody (Fig. 3). Cells expressing GFP alone exhibited a mainly diffuse pattern of staining, although there was some association of free-GFP with the Z-line in fixed cells (Fig. 3c). This association was not observed in living cells. In addition, cells expressing GFP did not exhibit any identifiable changes in the I-band (Fig. 3a) or M-band (Fig. 3b). By contrast, the GFP-skeletal muscle α-TM fusion protein was localized to the I-band of NRCs, and absent from the M-band (Fig. 3d-f), mimicking the localization of endogenous skeletal muscle α-TM. These results demonstrated that GFP fused to a sarcomeric TM was localized correctly.

In striated muscle cells, sarcomeric TM is precisely organized in the thin filaments of myofibrils. One question is whether specific isoforms of TM are required for assembly into these structures. It was therefore of interest to determine if different nonmuscle isoforms of TM would be able to properly incorporate into sarcomeres of NRCs. Accordingly, plasmids containing various constructs of nonmuscle GFP-TM were transfected into NRCs and analyzed in living cells 24-72 hours post-transfection. The GFP-TM fusions were observed to exhibit a sarcomeric pattern of incorporation as early as 24 hours following transfection. All isoforms were found to exhibit a sarcomeric pattern of incorporation with the exception of fibroblast TM-4, which had a diffuse pattern of staining similar to the results obtained using the control pEGFP plasmid (Fig. 4). In addition, the incorporation of the various nonmuscle GFP-TM fusion proteins did not interfere with sarcomere shortening (data not shown), indicating that incorporation of nonmuscle TMs did not have a dominant negative effect of sarcomeric function. However, the positive immunostaining of endogenous sarcomeric TM, which was colocalized in the sarcomeres with the exogenous nonmuscle TMs, revealed that incorporated nonmuscle TMs did not replace, at least not completely, the endogenous sarcomeric TM molecules coating the thin filaments.

Transfected cells expressing the various GFP-tagged nonmuscle isoforms of TM were fixed and stained for filamentous actin and for myomesin to verify that the GFP-TMs were incorporated into the myofibrillar I-band of NRCs. With the exception of TM-4, all of the GFP-nonmuscle-TM fusions were localized to the I-band of NRCs, and absent from the M-band (data not shown). These results demonstrated that GFP fused to a nonmuscle TM was localized correctly to the thin filaments of myofibfrils.

The various GFP-TM plasmids were also transfected into primary rat fibroblasts to determine if they would incorporate into actin filaments. All isoforms tested, including GFP-TM-4, were able to incorporate into microfilaments as judged by co-localization with phalloidin (data not shown).

An unexpected finding of these experiments was that GFP-TM-4 was unable to incorporate into thin filaments in NRCs, although it was able to localize to stress fibers in primary fibroblasts. A number of possibilities could account for the inability of this isoform to bind to actin filaments in NRCs. First, it was possible that TM-4 has a preference for nonmuscle actin, and therefore might interact poorly with the sarcomeric actin expressed in NRCs. Secondly, we recently reported that TM-4 is able to form coiled-coil heterodimers with other LMW isoforms of TM (Temm-Grove et al., 1996). Thus it was possible that heterodimers of TM-4 with TM-5, TM-5a or TM-5b would be the preferred binding configuration compared to a homodimer comprising two TM-4 subunits. Heart cells, unlike fibroblasts, do not express low molecular weight TM isoforms, hence the binding observed in fibroblasts. In addition, it is possible that different nonmuscle TMs could bind cooperatively, for example via head-to-tail overlap, and thus TM-4 might require other nonmuscle isoforms for binding along actin filaments.

In order to determine if binding of TM-4 could be improved by the presence of other LMW TM isoforms, GFP-TM-4 was co-transfected into NRCs with untagged and HA-or VSV-tagged TM-4, TM-5, TM-5a or TM-5b. Untagged and epitope-tagged TM-4 were also included to test whether the inability of incorporation of GFP-TM-4 was due to the formation in GFP-TM-4 monotransfected cells of a homodimer of GFP-tagged TM-4, containing two molecules of GFP, which would be sterically unfavorable for the binding to thin filaments. NRCs were transfected with different combinations of GFP-TM-4 and other LMW TM isoforms and tested for incorporation into sarcomeres. Co-expression of GFP-TM-4 with VSV-TM-4 did not result in a sarcomeric pattern of incorporation (Fig. 5a). By contrast, co-transfections of GFP-TM-4 with VSV-TM-5, VSV-TM5a, or VSV-TM5b led to a sarcomeric incorporation of GFP-TM-4 (Fig. 5b-d). Identical results were obtained when untagged TMs were co-expressed in NRCs (data not shown).

To verify that the GFP-TM-4 incorporation was due to co-expression of other low molecular weight isoforms, cells co-transfected with different combinations of plasmids were fixed and stained for the VSV-epitope tag and filamentous actin using TRITC-phalloidin. Cells co-expressing GFP-TM-4 and VSV-TM-4 exhibited a diffuse pattern of staining, although there was some association of free-GFP with the Z-line in fixed cells (data not shown). This association could also be observed in living cells (Fig. 5a). In addition, cells expressing TM-4 did not exhibit any identifiable changes in the I-band (Fig. 6a). In agreement with results obtained in living cells (Fig. 5), cells co-expressing VSV-TM-5, VSV-TM-5a, or VSV-TM-5b show incorporation of GFP-TM-4 into I bands (Fig. 6d-l).

We also considered the possibility that a homodimer of two GFP-TM-4 subunits might not be able to bind to actin because the presence of two GFP moieties sterically interfered with the ability of TM-4 to bind to actin. Accordingly, NRCs were transfected with HA-TM-4 and VSV-tagged TM-4, TM-5, TM-5a or TM-5b. In agreement with the results obtained with GFP-TMs, cells co-transfected with plasmids expressing HA-TM-4 and VSV-TM-4 did not exhibit incorporation of TM-4 into thin filaments (data not shown). Identical results were obtained when cells were singly transfected with plasmids expressing VSV-TM-4 or HA-TM-4 (data not shown). On the other hand, co-transfection of NRCs with HA-TM-4 and VSV-TM-5, VSV-TM-5a, VSV-TM-5b now showed incorporation of TM-4 into thin filaments of NRCs (data not shown).

Since nonmuscle cells such as fibroblasts express multiple isoforms of TM, it was conceivable that cooperative interactions could occur between HMW and LMW isoforms via head-to-tail overlaps, or perhaps through some other mechanisms, and promote actin-binding. Thus the lack of expression of other nonmuscle isoforms in NRCs could affect the ability of TM-4 to incorporate into sarcomeric thin filaments. Another possibility was simply that the co-expression of any TM that incorporates into thin filaments of NRCs would promote the binding of TM-4 to actin filaments. To determine if this was the case, plasmids expressing two high molecular weight isoforms of nonmuscle TM, fibroblast TM-1 and TM-2, as well as skeletal muscle α-TM, were co-transfected with TM-4 to determine if co-expression of these HMW TMs would promote the incorporation of TM-4. Co-expression of either HMW TM failed to enhance the binding of TM-4 (data not shown), indicating that the ability of enhancing TM-4 localization was dependent on the expression of another LMW TM isoform.

The multiplicity of TM isoforms in different cell types raises the possibility that specific associations of given isoforms are required for distinct actin structures. The recent development of green fluorescent protein (GFP) gene fusions to follow the dynamic distribution of a protein of interest in living cells offers a powerful alternative to standard tagging methods (Chalfie et al., 1994). In the present study we have used GFP-tagged isoforms in order to investigate the in vivo utilization of the individual muscle and nonmuscle TM isoforms. One question concerning the function of different isoforms of TM is whether the different isoforms will exhibit cell-type-specific preference for certain actin-containing structures. Here we show that specific isoforms characteristic of nonmuscle cells can incorporate into the myofibrillar apparatus of cardiomyocytes. Although it remains to be determined if nonmuscle isoforms can substitute functionally for sarcomeric proteins, these studies demonstrate that the presence of the GFP moiety does not prevent the incorporation of the tagged TMs into sarcomeres and this incorporation does not lead to apparent perturbations of the cytoarchitecture of NRCs. This study demonstrated that GFP-TMs could be incorporated into actin filaments of muscle cells and thus will provide a powerful tool to study their dynamics. In addition, incorporation of GFP-TMs into the myofibers still exhibited sarcomeric shortening and cell beating. Thus, TM can be added to the expanding list of cytoskeletal proteins for which GFP tagging is proving useful in the study of their dynamic distribution, including MAP4 (Olson et al., 1995), actin (Doyle and Botsein, 1996), myosin light chain and myomesin fragments (Auerbach et al., 1997) and α-actinin (Dabiri et al., 1997).

During myogenesis the patterns of TM isoform expression change, where the skeletal muscle α and β TMs are expressed and the nonmuscle isoforms of TM are repressed. Whether this change in the pattern of isoform expression reflects a requirement of specific isoforms for incorporation into sarcomeric structures was not known. Using cultured neonatal rat cardiomyocytes we were able to show that both skeletal muscle α-TM and most nonmuscle isoforms could incorporate into the thin filaments of sarcomeric assemblies. In rodents, skeletal muscle α-TM is the major isoform of TM expressed in cardiac muscle (Wieczorek et al., 1988). We found that with the exception of fibroblast TM-4, all nonmuscle isoforms were able to incorporate into sarcomeric structures. In this regard it is worth noting that in skeletal and smooth muscle, TM exists as a coiled-coil heterodimer of one α and one β subunit. In fibroblasts the HMW TM isoforms (TM-1, 2 and 3) exist as homodimers, where LMW can exist as heterodimers (Gimona et al., 1995; Temm-Grove et al., 1996). Although the HMW fibroblast type isoforms are able to incorporate into the thin filaments, it remains to be determined if they do so as homodimers or if they heterodimerize with the endogenous muscle isoforms, since TM-1, -2 and -3 are capable of forming heterodimers with skeletal α-TM in vivo (Gimona et al., 1995). However, since NRCs express no endogenous LMW TMs, incorporation of LMW TM-5(NM1), TM-5a and TM-5b must occur as homodimers.

One unexpected result of the present studies was the requirement of TM-4 to be co-expressed with other LMW TM isoforms for incorporation into actin filaments of NRCs. That the incorporation of TM-4 is dependent on the presence of other LMW TM isoforms has important implications in understanding the regulation of TM binding and actin filament dynamics in nonmuscle cells. We previously found that in vitro TM-5, TM-5a and TM-5b have affinities for F-actin similar to skeletal muscle TMs (Pittenger and Helfman, 1992; TemmGrove et al., 1996). In addition, we have found that recombinant TM-4 prepared in insect cells using the baculovirus expression system bound relatively poorly to actin (M. F. Pittenger and D. M. Helfman, unpublished observations). These studies thus demonstrate that among the LMW isoforms, TM-4 has a relatively weak apparent affinity for F-actin. The fact that TM-4 is able to readily bind to the microfilaments of fibroblasts suggests that there is a fundamental difference in the cellular environment of neonatal rat cardiomyocytes and fibroblasts (Gimona et al., 1995; our unpublished observations). Such differences could be due to actin isoforms, the presence of different regulatory proteins and cooperative interactions among different LMW TMs. Among the six vertebrate actin isoforms, cardiac muscle expresses α-cardiac actin and fibroblasts express β- and γ-cytoplasmic isoforms (Vandekerckhove and Weber, 1981; Otey et al., 1986). Whether TM-4 exhibits a preference for nonmuscle actins compared to sarcomeric actins remains to be determined. It is also possible that TM-4 requires the presence of a regulatory protein for binding to actin filaments. For example, the binding of a mixture of LMW isoforms (TM-4, TM-5(NM1), TM-5a and TM-5b) to F-actin was enhanced by caldesmon (Yamashiro-Matsumura and Matsumura, 1988). However, it is not presently known if this is the case for the binding of individual isoforms. Finally, it is possible that the actin-binding of TM-4 is dependent on the cooperation of other LMW isoforms and occurs via the formation of heterodimers with these LMW TM subunits (Temm-Grove et al., 1996). Since cardiac cells do not express LMW isoforms, each isoform presumably exists as a homodimer when singly transfected into NRCs. These data thus suggest that, in NRCs, TM-5, TM-5a and TM-5b can bind to actin in vivo as homodimers, whereas the affinity of TM-4 for actin filaments is enhanced when it is in a heterodimeric state with another TM subunit. This is important to the understanding of actin filament dynamics in fibroblasts and other nonmuscle cells, in which multiple isoforms of LMW isoforms are expressed.

Although it is known that fibroblasts contain multiple forms of TM, relatively little is known about the function of this isoform diversity. Most microfilaments have TM bound along their length. TMs are believed to function, in part, to stabilize actin filaments and protect filaments from the actions of severing proteins such as gelsolin (reviewed in Pittenger et al., 1994; Lin et al., 1997). Amongst the six TM isoforms expressed by normal rat fibroblasts are three TMs of major abundance (TM-1, -2 and -4), and three relatively minor tropomyosins (TM-3, -5(NM1) and -5a) (Matsumura et al., 1983; Goodwin et al., 1990; Temm-Grove et al., 1996). In a normal fibroblast, TM-4 is expressed in relatively higher levels than the other LMW isoforms and the majority of this isoform would likely exist in the homodimeric state. The assembly and disassembly of microfilaments is a dynamic process in locomoting and dividing fibroblasts. How the binding of specific dimers of TM isoforms along actin filaments is regulated during these dynamic events and affects the stability of actin filaments still remains to be determined. In addition, the binding of homodimers of TM-4 along actin filaments could be dependent on proteins such as caldesmon, whereas heterodimers of TM-4 and other LMW isoforms would exhibit intrinsically stronger interactions, and therefore be less dependent on the actions of caldesmon. Furthermore, the levels of different TM isoforms change during development and are altered in transformed cells (reviewed in Pittenger et al., 1994; Lin et al., 1997). Thus it is likely that different ratios of homodimers and heterodimers will be present in normal and transformed cells (Temm-Grove et al., 1996). Such changes in the levels of different LMW isoforms could effect their interactions with actin filaments, and thereby alter filament dynamics and cell motility. How the different combinations of subunits affect the relative affinity of TM for actin, head-to-tail overlap and interactions of other proteins such as myosins I and II is currently being investigated. These studies will no doubt provide important new insights into the function of TM isoform diversity and regulation of actin filament dynamics.

The work was supported by grant 31.277556/89 and 31.37537/93 to J.-C. P. from the Swiss National Science Foundation. C. B. received a postdoctoral fellowship from the American Heart Association, New York State Affiliate. D. M. H. is supported by a grant from the National Cancer Institute (Grant CA58607) and was the recipient of a Visiting Professorship from the Swiss Federal Institute of Technology.