Senescence represses the nuclear localization of the serum response factor and differentiation regulates its nuclear localization with lineage specificity

The relationship between cell proliferation and differentiation has been extensively studied. The induction of differentiation typically is associated with the progressive loss of proliferative potential that leads to terminal differentiation (Potten and Lajtha, 1982; Till, 1982). The signaling pathways by which differentiation is induced and the factors that induce differentiation however can show lineage specificity. For example, in murine mesenchymal cells, serum induces proliferation, whereas plasma induces differentiation (Hoerl and Scott, 1989). In contrast, in human epithelial keratinocytes, serum induces differentiation rather than proliferation (Wille et al., 1984).

Cellular senescence resembles terminal differentiation in that both processes cause cells to irreversibly lose the ability to proliferate in response to mitogenic agents (Wier and Scott, 1986). Replicative senescence of normal human diploid cells occurs after a limited number of cell divisions (Hayflick and Moorhead, 1961; Hayflick, 1965; Goldstein, 1990). The number of cell divisions after which senescence occurs has a direct inverse correlation with the age of the donor. Cells from patients afflicted with premature aging syndromes also show a reduced in vitro lifespan (Stanulis-Praeger, 1987). Therefore, senescence in culture mimics aspects of the aging process in vivo. Of the many theories proposed to explain senescence, one that is believed to be most plausible proposes that telomere loss acts like a molecular clock that can keep track of cell division numbers (Campisi, 1997; Sedivy, 1998). Since telomeres shorten with each cell cycle in normal somatic cells, a critical shortening in the length of telomeric DNA is thought to signal irreversible growth arrest.

Once cells enter the senescent state, the expression of several classes of genes changes, including cell cycle regulatory cyclins, CDKs and CDK-inhibitors (Wong and Riabowol, 1996), the tumor suppressor genes Rb-1 and p53 (Jansen-Durr, 1998; Vaziri and Benchimol, 1999), and the serum-inducible gene c-fos. With regard to the current studies, it is important to emphasize that the serum-induced transcription of c-fos is specifically repressed by senescence (Seshadri and Campisi, 1990). Nevertheless, the SV40 large T antigen can induce both DNA synthesis and c-fos expression in senescent cells (Campisi, 1992). Overexpression of c-fos in senescent cells can also induce limited DNA synthesis (Phillips et al., 1992), while microinjection of anti-sense c-fos oligonucleotide into senescent cells can block the T antigen-induced DNA synthesis (Campisi, 1992). These results indicate that c-fos plays an important role in senescence. The repression of c-fos transcription also occurs during the adipocyte differentiation of 3T3T mesenchymal stem cells (Wang and Scott, 1994). More specifically, we reported that when 3T3T adiopcytes first develop a differentiated phenotype, three events happen. Serum and growth factor responsiveness is repressed, AP-1 DNA binding activity is repressed and the transcription of c-fos and junB is repressed.

The c-fos promoter has been studied extensively. Among several regulatory sequences on the c-fos promoter, the serum response element (SRE) appears to be most important in mediating the transcriptional induction of c-fos triggered by various growth factors, serum and stress stimuli (Treisman, 1990). The c-fos SRE is a 20 base pairs dyad symmetry element which is specifically recognized by the serum response factor (SRF), a 67 kDa phosphoprotein which binds as a homodimer to the SRE (Treisman, 1992). The activity of SRF is necessary for the ability of the c-fos SRE to respond to serum stimulation. Experiments have shown that removal of SRF from cell extracts reduces the c-fos transcription to basal levels, while transcription can be restored to maximal levels by addition of exogenous SRF (Norman and Treisman, 1988; Norman et al., 1988; Prywes et al., 1988). Moreover, in vivo depletion of SRF from cell nuclei following antibody microinjection abolishes the response of the SRE to serum stimulation (Gauthier-Rouviere et al., 1991). SRF can interact and cooperate with other protein factors, such as members of ternary complex factor (TCF) family, to optimize the transcriptional induction of c-fos (Price et al., 1996; Hill and Treisman, 1995).

We recently reported that adipocyte differentiation of nontransformed murine 3T3T cells represses the expression and nuclear localization of SRF without effecting SP-1 that was used as a control. Repression of SRF nuclear localization thus impaired the availability of SRF to bind to the SRE. We proposed that this mechanism can explain how adipocyte differentiation represses growth factor responsiveness (Ding et al., 1999). On the basis of those findings, this paper evaluates a series of questions. Is the inhibition of SRF nuclear localization during adipocyte differentiation unique or is it a general effect that involves multiple transcription factors? Does loss of proliferative capacity involving other biological processes, such as cellular senescence, also represses the nuclear localization of SRF? Does differentiation of all cell lineages change SRF nuclear localization in a similar manner?

Normal human fetal lung fibroblasts (strain WI-38) and murine 3T3T mesenchymal stem cells were used in these experiments. WI-38 fibroblasts were purchased from NIA aging cell culture repository (Coriell Institute) at population doubling levels (PDL) of 16 and 44, respectively (maximum lifespan of 50 population doublings for this culture was reported by the repository). WI-38 cells were cultured in minimal essential medium Eagle (MEM) containing Earle’s salts, supplemented with 2 mM glutamine, 20 mM NaHCO₃ and 10% fetal bovine serum at 37°C in 5% CO₂/95% air atmosphere. Early passage cells were defined as those that had progressed through 40% of their replicative life span (PDL of 20 to 24), whereas senescent cells were defined as those that had completed >95% of their life span (PDL of 48 to 50).

3T3T cells were routinely cultured at 37°C in 5% CO₂/95% air in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% bovine calf serum (BCS).

Specimens of normal human small bowel and bone marrow were obtained from the UT Surgical Pathology department from waste materials.

To induce predifferentiation growth arrest and subsequent adipocyte differentiation, growing 3T3T cells were dissociated with 0.1% ethylenediamine tetraacetic acid (EDTA) in phosphate-buffered saline (PBS) and plated onto 100-mm ethylene oxide-sterilized bacteriological petri dishes at low density in heparinized DMEM containing 25% (v/v) human plasma as previously described (Hoerl and Scott, 1989). In this medium, most cells become quiescent within 3 to 4 days and subsequently express the nonterminal adipocyte phenotype between days 6-8. For 3T3T cells, the terminal differentiation phenotype is thereafter expressed between days 10 to 15. The extent of differentiation in such cultures was routinely characterized by phase microscopic examination and exceeded 75%.

Nuclear proteins for western blotting analysis were isolated using the modified method of Dignam (Dignam et al., 1983). After washing cells twice with 4°C PBS (pH 7.4), the cells were harvested in 4°C PBS with a cell scraper. After centrifugation at 500 g for 5 minutes at 4°C, cells were resuspended and incubated for 10 minutes on ice in 4°C hypotonic buffer (1.5 mM MgCl₂, 10 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), 10 mM KCl, 0.2 mM phenylmethylsulfonyl fluoride (PMSF), 0.5 mM dithiothreitol (DTT), pH 7.9), the volume of the buffer was proportioned to the number of collected cells. After centrifugation at 25,000 g for 15 minutes at 4°C, pellets were collected.

Nuclear proteins were extracted from the pellets. Pellets were washed once with hypotonic buffer and centrifuged at 10,000 g for 15 minutes at 4°C, after that pellets were suspended in ice-cold low-salt buffer (25% glycerol, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.2 mM PMSF, 0.5 mM DTT, 20 mM KCl, 20 mM HEPES, pH 7.9). Nuclear protein were released by adding a high-salt buffer (25% glycerol, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.2 mM PMSF, 0.5 mM DTT, 1.2 M KCl, 20 mM HEPES, pH 7.9) drop by drop using the half of the volume of low-salt buffer. Samples were incubated on ice for 30 minutes, with frequent smooth mixing. After centrifugation at 25,000 g for 30 minutes at 4°C, the supernatant containing nuclear proteins was collected and stored at −80°C.

The total cellular proteins were prepared as previously described (Wang et al., 1996). Briefly, cell monolayers were rinsed twice with 4°C PBS and were then harvested in 4°C PBS. After centrifugation at 500 g for 5 minutes at 4°C, cell were lysed at 4°C for 10 minutes in lysis buffer that contain PBS (pH 7.4), 1% Triton X-100, 0.5% sodium fluoride and 10 mM sodium pyrophosphate. After centrifugation at 30,000 g at 4°C for 10 minutes, total cellular proteins were collected from the supernants and frozen at −80°C.

The protein concentrations were measured using the Protein Assay Kit of Bio-Rad and their assay protocol.

Undifferentiated or differentiated 3T3T cells, or WI-38 fibroblasts (young or senescent) were allowed to attach to slides overnight at 37°C in culture medium. Cells were briefly rinsed with PBS and then fixed for 20 minutes with 4% (w/v) paraformaldehyde in 100 mM sodium phosphate (pH 7.4). Fixed cells were washed in PBS containing 10 mM glycine three times for 5 minutes each, then permeablized for 5 minutes with 1% Nonidet P-40 in PBS plus glycine, and washed as before. Slides were exposed to various dilutions (1:50 to 1:200) of rabbit anti-SRF antibody (Santa Cruz Biotechnology, Inc.) for 60 minutes. Cells were rinsed as before and incubated with fluorescein-conjugated goat anti-rabbit IgG diluted 1:200 (Santa Cruz Biotechnology, Inc.) for 45 minutes. Slides were washed with three changes of PBS, then coverslips were mounted with mounting medium containing 4′,6-diamidino-2-pheylindole (Vector Laboratories, Inc.). The stained cells were examined and photographed using a confocal laser scanning fluorescence microscope (Zeiss LSM 510).

Normal human bone marrow slides were stained following the same procedure, except that slides were further incubated with 100 μg/ml RNAase and 0.05 μg/ml propidium iodide in PBS for 10 minutes before final washes.

Western immunoblotting procedures were performed as previously described (Wang and Scott, 1994). The primary antibodies used in these experiments include rabbit antibodies against SRF, YY1, C/EBPα, CREB-1, AP-2α, N-Myc and p53, from Santa Cruz Biotechnology, Inc. Protein samples were mixed with sodium dodecyl sulfate (SDS) sample buffer, boiled for 5 minutes, and then separated by electrophoresis in 7.5% to 12% SDS-polyacrylamide gel. After the transfer of proteins onto nitrocellulose membranes, they were incubated for 1 hour at room temperature in the western blocking buffer containing 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 0.5% (v/v) Tween-20, and 1% (w/v) nonfat dry milk. Immunological evaluation was then performed overnight at 4°C in the western blocking buffer containing 1:300 dilution of primary antibodies. The membranes were subsequently washed with western blocking buffer and incubated for 1 hour at room temperature with a peroxidase-conjugated secondary antibody (1:2000 dilution, Sigma). After extensive washing with western blocking buffer, the immunocomplexes on the membrane were visualized using ECL western blotting analysis kit (Amersham Corp.).

The formaldehyde-fixed, paraffin-embedded human tissue sections were dewaxed by incubating in xylenes and subsequently rehydrated by processing through a series of washes with decreasing ethanol solution (100%, 95%, 70% and 50%) and then H₂O. To quench endogenous peroxidase activity, the sections were incubated in methanol containing 0.3% (v/v) hydrogen peroxide for 30 minutes. Then, the sections were blocked by incubation in PBS containing 3% normal goat serum for 60 minutes at room temperature to prevent non-specific binding of the antibodies to the tissue. Thereafter, the sections were incubated with primary rabbit anti-SRF antibody (1:100 dilution in PBS containing 1.5% goat serum, Santa Cruz) overnight at 4°C. After this incubation period, the section slides were rinsed with three changes of PBS for 5 minutes each, then exposed to 1:300 dilution of peroxidase-conjugated goat anti-rabbit IgG (Sigma) for 90 minutes at room temperature. After having been rinsed as before, the peroxidase label was developed by exposure to 0.05% (w/v) diaminobenzidine in phosphate buffer containing 0.01% H₂O₂ for 3 minutes. The slides were washed in PBS, counterstained by hematoxylin (Sigma), and dehydrated by an ethanol/xylene series before coverslips were finally mounted. The specimens were viewed using a standard light microscope (Carl Zeiss).

We recently reported that the nuclear localization of SRF is dramatically decreased during the process of adipocyte differentiation of murine 3T3T cells (Ding et al., 1999). This paper expands that analysis to determine whether adipocyte differentiation represses the nuclear localization of multiple transcription factors or just SRF. We specifically tested the intracellular localization of six additional transcription factors including YY1, C/EBPα, CREB-1, AP-2α, N-Myc and p53. In undifferentiated proliferating cells, all these factors are predominantly localized in the nucleus. Next, our studies asked if the relative amount of individual transcription factors differs in nuclear extracts of undifferentiated versus differentiated 3T3T cells.

Fig. 1 clearly shows that among these transcription factors, SRF is the only factor whose nuclear content is dramatically decreased in differentiated 3T3T adipocytes. In regard to other factors, the amount of YY1, CREB-1, AP-2α, N-Myc and p53 in the nucleus shows no significant difference between differentiated and undifferentiated cells, while that of C/EBPα shows a obvious increase when cells differentiate. This latter finding is consistent with previous reports that the expression of C/EBPα is induced in the late stage of adipocyte differentiation (Cao et al., 1991).

Table 1 quantitates and summarizes the data presented in Fig. 1. On the basis of the expression of these factors relative to loading controls in various experiments, the data confirm that in differentiated adipocytes, SRF exhibits a 95% reduction in expression, whereas the other tested factors show significantly less relative reduction in expression in association with differentiation. This suggests that repression of SRF nuclear localization is specific in differentiated adipocytes.

Cellular senescence resembles the state of terminal differentiation, since in both processes, cell lose their mitogenic responsiveness to growth factor stimulation (Goldstein, 1990). Previous studies on the mechanisms of senescence used gel mobility-shift assays to show that the ability of SRF to bind to the SRE is greatly decreased in senescent human fibroblasts (Atadja et al., 1994). Therefore, we tested whether cellular senescence also represses the nuclear localization of SRF. Human WI-38 fibroblasts were continuously grown until senescence, and assays were performed on cell specimens at different population doublings.

Indirect immunofluorescence was first used since senescent WI-38 cells showed comparable fluorescence background as growing cells. Using confocal microscope, the results in Fig. 2 clearly show that the nuclear staining of SRF is evident in proliferating WI-38 cells, however, only minimal nuclear staining is detectable in senescent cells.

These results were further confirmed by western blotting analysis. The relative amount of SRF in total whole cell extracts versus nuclear extracts of young and senescent cells was characterized (Fig. 3). The data show that senescent cells show only a slight decrease in SRF expression whereas the localization of SRF in the nucleus is dramatically decreased by senescence. To evaluate the results in Fig. 3, it is essential to realize that equal amounts of total cell protein from young and senescent cells was added to each well but that a different yet equal amount of nuclear protein from young and senescent cells was used. This approach was used to assure comparable band intensity in the two groups since SRF is usually concentrated in the nucleus. Therefore, cellular senescence, like adipocyte differentiation, represses the nuclear localization of SRF and thus represents a mechanism that can explain how repression of SRF-SRE interactions occur during senescence (Atadja et al., 1994).

All the experiments described previously concerning the nuclear localization of SRF employed cultured murine 3T3T cells or human fibroblasts. Studies were therefore next performed to determine if differentiation in vivo influences the expression and/or nuclear localization of SRF in other cell lineages in the same or in a different manner. Immunofluorescence studies were first performed using normal human bone marrow specimens because such hematopoietic cells show a vast range of differentiation characteristics and because such cells are of mesenchymal origin as are 3T3T cells and WI-38 cells. The results of such studies, however, showed that no significant staining of such cells with anti-SRF antibodies could be detected in either undifferentiated or differentiated cells. Therefore, we concluded this was not a worthwhile in vivo tissue to evaluate.

Immunohistochemical studies next focused on analysis of SRF expression and nuclear localization in the human small bowel mucosa cells that represent an epithelial lineage that shows well characterized differentiation features. In contrast to data from studies on 3T3T adipocytes and senescent fibroblasts, Fig. 4 demonstrates an increased expression and nuclear staining of SRF in cells at the surface tip of the villus, where differentiated epithelial cells are located. Only minimal SRF nuclear staining is detected in cells at the base of crypt where most of the undifferentiated and proliferative cells reside. These results suggest that differentiation can effect SRF expression and nuclear localization with lineage specificity.

We previously reported that the nuclear localization of SRF is repressed during adipocyte differentiation of nontransformed murine 3T3T cells, but not in differentiated transformed cells (Ding et al., 1999). The present study expands those findings and shows that adipocyte differentiation selectively represses the nuclear localization of SRF, but not the nuclear localization of a series of other transcription factors, including YY1, C/EBPα, CREB-1, AP-2α, N-Myc and p53. This suggests that adipocyte differentiation may induce a specific regulatory mechanism to control SRF nuclear localization.

Control of the localization of macromolecules into nucleus of eukaryotic cell is a very complex process. In principle, regulation of the nuclear localization of transcription factors can involve the activity of nuclear localization signals (NLSs), cytoplasmic retention signals and nuclear import receptors (Kaffman and O’Shea, 1999). No well-characterized case of the import receptors has yet been reported for SRF and SRF has not been found to contain cytoplasmic retention signals. Concerning nuclear localization signals for SRF, our preliminary data show that when a yellow fluorescent protein which contain two copies of SRF-NLS was introduced into 3T3T cells, the yellow fluorescence signals were expressed at the same level in the nuclei of both growing undifferentiated cells and differentiated adipocytes (data not shown). This result suggests that adipocyte differentiation does not inhibit the function of SRF-NLS. Therefore, it remains to be determined how nuclear import of SRF is regulated in differentiated adipocytes.

It is also possible that different mechanisms control SRF nuclear localization in different cell lineages. The fact that in human tissues differentiation regulates SRF nuclear localization with lineage specificity supports this idea. More specifically, nuclear SRF immunostaining is most evident in differentiated epithelial cells located at the tip of small bowel mucosa villus, while SRF is barely detected in undifferentiated cells at the base of crypts. This finding probably results from both an increased expression of SRF in differentiated small bowel epithelial cells and increased nuclear localization of SRF. Considering that it has been demonstrated that serum induces normal epithelial keratinocyte differentiation rather than proliferation (Lechner et al., 1982; Wille et al., 1984), it is possible that SRF-target elements are present on some epithelial-specific genes that are required during differentiation. The nuclear localization of SRF in differentiated epithelial cells therefore could reflect the involvement of SRF in promoting the serum-induced differentiation of epithelial cells.

There are other lines of evidence that suggest that the function of SRF in regulating cell differentiation may be cell-lineage specific. First, it is known that SRF is essential for some cell-type-specific gene regulation, such as neuronal (Ghosh and Greenberg, 1995) and muscle-specific gene expression (Vandromme et al., 1992; Soulez et al., 1996). Second, studies on SRF localization during vertebrate embryogenesis and development reported that enriched SRF expression is mainly found in specific mesodermal and neuroectodermal tissues (Croissant et al., 1996; Belaguli et al., 1997). Third, SRF is required for myogenic differentiation in vitro and mesoderm formation in vivo during mouse embryogenesis (Arsenian et al., 1998; Soulez et al., 1996; Vandromme et al., 1992).

It is on the basis of this information that an answer to the question of why SRF nuclear localization differs so drastically in small bowel mucosa cells relative to 3T3T adipocytes and senescent fibroblasts can be proposed. Perhaps the serum response element that binds SRF exist in the promoter of specific differentiation genes of the small bowel epithelium so that SRF expression and nuclear localization is essential to induce differentiation of such cells as they migrate to the tip of the villus. In contrast, in mesenchymal 3T3T cells and fibroblasts, it appears that the SRE is primarily used in the promoters of proliferation control genes so that when terminal differentiation or senescence is to occur, mechanisms are activated that repress SRF nuclear localization and to a lesser extent SRF expression.

Concerning the role of SRF in cell proliferation, SRF regulates transcriptional induction of several immediate-early genes (IEGs) (Spencer and Misra, 1996), but SRF is not essential for cell proliferation per se. SRF null mutant cells do not show significant proliferative defects even though they show impaired serum induction of SRE-containing IEGs (Arsenian et al., 1998). This probably reflects the existence of compensatory mechanisms to control cell growth. For example, the TCF transcription factor also has been shown to be able to respond to serum/growth factor stimulation (Shaw et al., 1989). This could explain why some types of cells can proliferate even though they lack nuclear SRF. From this perspective, we found that SRF is barely detected in the nuclei of most mesenchymal bone marrow cells regardless of their state of differentiation.

Cellular differentiation and senescence are two similar biological processes, because both can lead to irreversible growth arrest. More specifically, both differentiated and senescent cells are growth-arrested with a G₁ DNA content and neither can be stimulated to enter the S phase of the cell cycle by any known combination of physiological mitogens even though such cells can be maintained in this viable, albeit non-replicating state for very long periods of time (Hayflick and Moorhead, 1961; Hayflick, 1965; Plisko and Gilchrest, 1983; Phillips et al., 1984; Pignolo et al., 1998). At the molecular level, both cell differentiation and senescence are associated with an altered pattern of gene expression, much of which is transcriptionally based (Cristofalo et al., 1992). For example, in both systems, downregulation of the DNA-binding of the AP-1 and SRF, and transcriptional repression of the serum-induction of c-fos occurs (Campisi, 1997; Dimri and Campisi, 1994; Wang and Scott, 1994; Ding et al., 1999). In 3T3T cells, we proposed that this repression of the c-fos transcription and SRF-SRE interactions results from blocking the nuclear localization of SRF (Ding et al., 1999).

It is therefore significant that in this paper, we demonstrated that the nuclear exclusion of SRF also occurs during human fibroblast senescence. Although it was previously reported that hyperphosphorylation of SRF in senescent human Hs68 cells might contribute to decreased SRF DNA binding activity (Atadja et al., 1994), we did not observe any migration-shift changes of SRF in senescent cells or differentiated cells by western blotting. Considering that small molecular mass changes may not be distinguished in one-dimensional gel electrophoretic analysis, we can not rule out the possibility that hyperphosphorylation of SRF did occur in our senescent cell studies. Future experiments will need to determine whether decreased SRF-SRE interactions result from combined effects of hyperphosphorylation and nuclear exclusion of SRF or if hyperphosphorylation is the regulatory mechanism that mediates the inhibition of nuclear localization of SRF in senescent human cells. In the latter case, phosphorylation is known to be a mechanism that regulates nuclear import (Jans and Hubner, 1996). For example, phosphorylation by cyclin-dependent kinase cdc2 usually inhibits nuclear transport of protein factors, such as SV40 T antigen (Jans, 1995). However, what kinase is responsible for the possible hyperphosphorylation of SRF is unknown.

The discovery that repression of SRF nuclear localization occurs during both mouse adipocyte differentiation and human fibroblast senescence indicates that this might be a common mechanism in one of pathways that repress serum/growth factor responsiveness in mesenchymal cells. It may be important that both differentiation and senescence are thought to represent tumor suppressive mechanisms (Koeffler, 1983; Reiss et al., 1986; Wynford-Thomas, 1997; Campisi, 1997; Wangenheim and Peterson, 1998). There are several lines of evidence support this conclusion. First, clinical and experimental investigation indicates that in general, differentiation and malignancy can be inversely correlated in specific tissues (Gabbert et al., 1985; Eccles, 1983). Second, nonterminal differentiation induces 3T3T cells to become resistant to transformation by physical or chemical carcinogens or oncogene products and nonterminal differentiation induces transformed cells to revert to a benign state and become resistant to retransformation (Maercklein et al., 1990; Tzen et al., 1990). Third, proliferative senescence is associated with increased activities of several tumor suppressors including Rb-1, p53 and p21 thus blocking the passage of senescent cells through the cell cycle (Garkavtsev et al., 1998; Vaziri and Benchimol, 1996). This pathway is often impaired in the generation of immortal cancer cells (Sedivy, 1998).

The discovery that transformed mesenchymal cells do not necessarily repress the nuclear localization of SRF also raises the possibility that repression of the expression and nuclear localization of SRF may be one of molecular mechanisms mediating tumor suppressor activity in such cells. To test this hypothesis, future studies on SRF-deficient cells will be useful to determine if they show resistance to transformation. Another possibility, is that transformed cells may express lesions in the molecular mechanisms regulating the nuclear transport of many transcription factors. In support of this possibility, it has already been reported that p53, BRCA1, Rel, PEBPβ and β-catenin show aberrant intracellular localization in neoplastically transformed cells (Zaika et al., 1999; Chen et al., 1995; Gilmore et al., 1996; Sachdev and Hannink, 1998; Tanaka et al., 1998; Sheng et al., 1998). Therefore, the development of defects in regulation of the expression and/or distribution of SRF in cells may mediate important aspects of lineage specific carcinogenesis.

The authors acknowledge support of the Muirhead chair of excellence endowment to R.E.S. The data in this publication represent part of the PhD thesis of W.D.