Transformation Blocks Differentiation-induced Inhibition of Serum Response Factor Interactions with Serum Response Elements¹ Free

Cellular proliferation and differentiation are coordinately regulated biological processes in most nontransformed cells. For example, cellular proliferative potential is repressed as basal epithelial keratinocytes differentiate (1), and differentiation of skeletal myoblasts into contractile myotubes results in their permanent withdrawal from the cell cycle (2). The differentiation of murine mesenchymal 3T3T stem cells into adipocytes also progressively represses cellular proliferative potential via a multistep process that ultimately leads to terminal differentiation.

Three distinct steps are involved in the differentiation of 3T3T cells into adipocytes: (a) PGA;³ (b) NTD; and (c) TD (3). Cells first undergo PGA associated with the expression of several transacting factors including PPAR-γ and C/EBP-γ and -β (4). In this state, cells show mitogenic responsiveness to 10 ng/ml PDGF and 5–10% serum. Subsequently, cells modulate C/EBP expression so that C/EBP-α is the dominant form (4) as they enter a nonterminal state of differentiation in which they lose their responsiveness to most mitogens except high concentrations of serum (5). Thereafter, cells lose their responsiveness to all mitogens as they undergo TD (3, 6). In contrast, the differentiation of SV40-transformed 3T3T cells (CSV3-1 cells) into adipocytes does not repress mitogenic responsiveness but rather increases it (7, 8). For example, at the NTD state, 5% serum is sufficient to induce the majority of CSV3-1 adipocytes to undergo DNA synthesis, whereas 30% serum is required for 3T3T cells. Therefore, CSV3-1 adipocytes cannot efficiently undergo TD.

The molecular mechanisms that mediate the progressive loss of serum-induced mitogenic responsiveness during adipocyte differentiation in nontransformed 3T3T cells but not in transformed cells has not been definitively established. We have shown that adipocyte differentiation represses AP-1 DNA binding activity and the serum-inducible transcription of AP-1 factors including junB and c-fos (9). In this regard, cellular senescence also decreases the inducibility of c-fos transcription by serum (10). Because the junB and c-fos promoters both contain SREs (11), it is possible that adipocyte differentiation represses proliferation by blocking the ability of the SRE to interact with the SRF that is required for its transactivation. The studies reported here test this possibility.

An intact SRE is necessary for the transcriptional activation of junB and c-fos expression by serum stimulation, and the SRE alone is sufficient to confer the serum-dependent transcriptional activation of heterologous promoters (12). The sequence -CC[A/T]₆GG-, which constitutes the core SRE domain, is the binding site for SRF (13). This M_r 67,000 nuclear phosphoprotein contains multiple domains: (a) a SRE DNA binding domain; (b) a transactivation domain; and (c) several phosphorylation domains (12). SRF binds to the SRE as a homodimer, and SRF can form a complex with TCFs, including Elk1, Sap1, Sap2, FLI1, and EWS-FLI1 in association with the Ets domain that commonly adjoins a SRE (12, 14, 15). X-ray crystallographic studies also suggest that the interaction of SRE and SRF involves mutual structural changes (16).

SRF transcription and expression are induced when quiescent undifferentiated cells are treated with serum because the SRF promoter contains a SRE, thus establishing the possibility of autoregulation (17). Serum treatment of cells can also induce SRF phosphorylation at sites near a unique SRF nuclear localization signal within its NH₂-terminal region (18). Control of SRF nuclear localization is also indirectly influenced by protein kinase A (19). SRF phosphorylation modulates the association versus dissociation characteristics of SRF with the SRE (20). Furthermore, repression of SRF DNA binding activity in senescent human fibroblasts is mediated by the hyperphosphorylation of SRF (21).

From this perspective, it is important to emphasize that two general classes of signaling mechanisms involving the SRF can regulate the activity of the SRE. A TCF-dependent pathway mediates the effects of PDGF, colony-stimulating factor 1, and 12-O-tetradecanoylphorbol-13-acetate on SRE via the ras → raf → mitogen-activated protein/extracellular signal-regulated kinase kinase → extracellular signal-regulated kinase cascade (22). Both phosphorylation of TCFs and the binding of TCFs to SRF are required for activation of the SRE by this pathway. A TCF-independent pathway mediates signals induced by serum to the SRE via the SRF, and these pathways involve the Rho family of GTPases (23). It currently appears that the SRF DNA binding activity is required for both of these pathway. For example, microinjection of anti-SRF antibodies into rat fibroblasts blocks the response of the SRE to serum and growth factor stimulation (12).

The studies reported here focus on the effects of adipocyte differentiation in nontransformed and transformed 3T3T cells on SRE-SRF interactions. The results show that adipocyte differentiation in nontransformed cells represses SRE-SRF interactions concomitant with the differentiation-induced repression of mitogenic responsiveness and the loss of junB and c-fos serum inducibility. The data further show that in neoplastically transformed cells, differentiation fails to repress mitogenic responsiveness, the inducibility of junB and c-fos, and SRE-SRF interactions. Our final discovery that differentiation represses SRF-SRE interactions by restricting SRF nuclear localization only in nontransformed cells suggests that neoplastic transformation can abrogate this key regulatory mechanism.

Nontransformed murine 3T3T mesenchymal stem cells, SV40 large T antigen-transformed 3T3T cells (designated CSV3-1 cells), and spontaneously transformed 3T3T cells (designated DW1 cells) were used in these experiments. They were routinely cultured at 37°C in 5% CO₂/95% air in DMEM supplemented with 10% BCS unless otherwise stated.

To induce quiescence and subsequent adipocyte differentiation, growing 3T3T or CSV3-1 cells were dissociated with 0.1% EDTA in PBS and plated onto 100-mm, ethylene oxide-sterilized, bacteriological Petri dishes at low density in heparinized DMEM containing 25% (v/v) human plasma. In this medium, 3T3T and CSV3-1 cells become quiescent within 3–4 days and subsequently express the nonterminal adipocyte phenotype between days 6 and 8. The terminal differentiation phenotype is expressed thereafter between days 10 and 15 only in 3T3T cells; under the same condition, most CSV3-1 cells retain their ability to undergo DNA synthesis when restimulated with serum (3, 9). The extent of differentiation in such cultures was routinely characterized by phase microscopic examination, and it exceeded 75%.

In selected experiments, quiescence in nontransformed 3T3T cells in an undifferentiated state was induced by culture in DMEM containing 0.5% BCS for 3–4 days. In all studies, quiescence was established by observation of secession of growth and/or by quantitation of ³[H]thymidine incorporation into DNA (3).

The plasmids pjunB SREtk/CAT and pc-fos SREtk/CAT were prepared by insertion of a junB SRE (5′-CTAGACTTCCTGTGCCCTAATATGGATGCTGGG-3′) or a c-fos SRE (5′-CTAGAGGATGTCCATATTAGGACATCTG-3′) into pBLCAT₂, which contains the thymidine kinase gene promoter upstream of the CAT gene (24). For transient transfection using Lipofectamine (Life Technologies, Inc.), murine 3T3T cells were grown in DMEM-10% BCS before transfection with 20 μg of pSREtk/CAT DNAs. Transfected cells were refed 24 h later with either DMEM-0.5% BCS (to induce quiescence) or heparinized DMEM-25% human plasma (to induce differentiation). Five days later, CAT assays were performed on cells before and after stimulation with 10% BCS for 7 h, which was predetermined to be the time required for maximum stimulation.

For CAT assays, cell lysates were incubated at 37°C for 15 h in 0.25 m Tris-HCl (pH 8.0), 1.5 μCi/ml [¹⁴C]chloramphenicol, and 500 μm acetyl-CoA. [¹⁴C]Chloramphenicol (Amersham) and its acetylated products were then separated by TLC, and the extent of conversion of chloramphenicol to its acetylated forms was quantitated using a radioanalytic imaging computer system.

Nuclear extracts were prepared as described previously (9). 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 min at 4°C, cell pellets were resuspended and incubated on ice for 5 min in 5 ml of 4°C STM buffer [20 mm Tris-HCl (pH 7.85), 250 mm sucrose, 1.1 mm MgCl₂, and 0.2% Triton X-100]. Nuclei were then sedimented by centrifugation at 800 × g for 10 min, washed once with STM buffer, and washed once with STM buffer lacking Triton X-100. Nuclei were then resuspended in STM buffer lacking Triton X-100, 0.4 m KCl, and 5 mm β-mercaptoethanol and incubated on ice for 10 min to facilitate nuclear lysis. After centrifugation at 2000 × g for 10 min at 4°C, the supernatant containing nuclear proteins was collected, divided into aliquots, and frozen at −80°C for subsequent assays. The protein concentrations were determined using a protein assay kit (Bio-Rad Laboratories) according to the supplier’s instructions.

Multiple oligonucleotide pairs were used in these experiments: (a) 5′-CTTCCTGTGCCCTAATATGGATGCTGG-3′ and its complimentary partner that contains the junB SRE (Biosynthesis); (b) 5′-GGATGTCCATATTAGGACATC-3′ and its complimentary partner that contains the c-fos SRE (Santa Cruz Biotechnology, Inc.); (c) 5′-GGATGTCCATATTATTACATC-3′ and its complimentary partner that contains a mutant c-fos SRE (Santa Cruz Biotechnology, Inc.); (d) 5′-CTTCCTGTGCCCTTATATGGATGCTGG-3′ and its complimentary partner that contains the β-actin SRE; (e) 5′-CTTCCTGTGCCCTTTTTTGGATGCTGG-3′ and its complimentary partner that contains the EGR2 SRE; (f) 5′-CTTCCTGTGCCCTTATTTGGATGCTGG-3′ and its complimentary partner that contains the thrombospondin-1 SRE; and (g) 5′-CTTCCTGTGCCCTATTATGGATGCTGG-3′ and its complimentary partner that contains the dystrophin SRE.

The latter four preparations were synthesized by the St. Jude Children’s Research Hospital Center for Biotechnology. One oligonucleotide from each pair was radioactively end-labeled using [γ-³²P]ATP (3000 Ci/mmol; Amersham) and T4 polynucleotide kinase and annealed to its complimentary partner. Binding reactions were carried out by mixing 10 μg of nuclear proteins and 0.035 pmol (in 1 μl) of ³²P-labeled oligonucleotide pairs with 1 μg of poly(deoxyinosinic-deoxycytidylic acid) in a total volume of 25 μl of binding buffer [2 mm Tris-HCl (pH 7.5), 8 mm NaCl, 0.2 mm EDTA, 0.2 mm β-mercaptoethanol, and 0.8% glycerol]. The binding reactions were allowed to proceed at room temperature for 20 min. Thereafter, 2 μl of 0.1% bromphenol blue were added, and the reaction mixture was subjected to electrophoresis on nondenaturing 5% polyacrylamide gels followed by visualization using autoradiography as described previously (9).

For gel supershift assays, 1 μg (in 1 μl) of specific supershift antibodies against SRF (Santa Cruz Biotechnology, Inc.) or of other control antibodies was added to the reaction mixture containing the nuclear protein extract and incubated for an additional 30 min at room temperature before the performance of mobility shift assays (9).

For indirect immunofluorescence, undifferentiated or differentiated cells were allowed to attach to slides overnight at 37°C in culture media. Cells were rinsed briefly with PBS and then fixed for 20 min with 4% (w/v) paraformaldehyde in 100 mm sodium phosphate (pH 7.4). Fixed cells were washed three times in PBS containing 10 mm glycine for 5 min each, permeablized for 5 min with 1% NP40 in PBS and glycine, and washed as described previously. Slides were exposed to various dilutions (1:25 to 1:200) of rabbit anti-SRF or anti-SP-1 antibody (Santa Cruz Biotechnology, Inc.) for 60 min. Cells then were rinsed as described previously and incubated with fluorescein-conjugated goat anti-rabbit IgG diluted 1:200 (Santa Cruz Biotechnology, Inc.) for 45 min. Slides were washed with three changes of PBS, and then coverslips were mounted with mounting medium containing 4′,6-diamidino-2-phenylindole (Vector Laboratories, Inc.). The cells were examined using a Zeiss fluorescence microscope with a Plano ×40 objective and photographed using Kodak black and white ASA 200 film.

Western immunoblotting procedures were performed as described previously (9). Nuclear proteins were mixed with SDS sample buffer, boiled for 5 min, separated by electrophoresis in a 7.5% SDS-polyacrylamide gel, transferred to nitrocellulose, and immunoblotted with rabbit anti-SRF or anti-SP-1 antibodies (Santa Cruz Biotechnology, Inc.). Bound antibody was detected using goat anti-rabbit IgG conjugated to peroxidase (Sigma) and visualized using enhanced chemiluminescence Western blotting analysis kit (Amersham).

To initiate our investigation of the mechanisms by which adipocyte differentiation represses the transcription of junB and c-fos, a 5300-bp junB promoter fused to the bacterial CAT gene was stably transfected into 3T3T cells, and the effect of serum induction was tested. Initial studies showed that serum induced CAT activity in quiescent undifferentiated cells, but not in differentiated cells (data not shown). This suggested that differentiation exerts its inhibitory activity on junB transcription via an effect on the junB promoter. Because SREs exist in both the junB and c-fos promoters, we hypothesized that they might represent the target domain for such repression.

To determine whether the function of the junB and c-fos SREs is specifically repressed by adipocyte differentiation, oligonucleotides corresponding to the junB SRE and the c-fos SRE were synthesized and ligated upstream of the herpes simplex virus thymidine kinase promoter in the CAT reporter plasmid pBLCAT2 (24). When they were transiently transfected into murine 3T3T cells, both reporter genes were established to be serum inducible. Table 1 shows that CAT activity regulated by either the junB SRE or the c-fos SRE was induced >10-fold when quiescent undifferentiated cells were stimulated with 10% BCS. In contrast, serum treatment of NTD adipocytes induced only minimal CAT activity using either the pjunB SREtk/CAT or pc-fosSREtk/CAT reporter plasmids. This indicates that the function of both junB and c-fos SREs is repressed by adipocyte differentiation. Because neither SRE in these plasmids contains an intact TCF Ets binding domain, adipocyte differentiation must repress SRE activity in a TCF-independent manner.

The binding characteristics of the SRF transcription factor to the junB and c-fos SREs were investigated by electrophoretic mobility shift assays using junB SRE oligonucleotides or c-fos SRE oligonucleotides as binding targets. First, nuclear proteins extracted from rapidly growing 3T3T cells were incubated with labeled oligonucleotides, and the identity and specificity of shifted complexes were determined by competition assays using different unlabeled DNA competitors and by supershift assays using SRF supershift antibodies.

Fig. 1 shows that a comparable SRF band was evident in lanes containing junB or c-fos SRE probes when nuclear extracts of undifferentiated cells were used. This complex was established to contain both SRE and SRF using several approaches. The binding is specifically eliminated by the addition of a 50-fold excess of unlabeled junB or c-fos SRE oligonucleotides, whereas no inhibition was detected when an oligonucleotide with point mutations in the GG sites of the c-fos SRE, which abolishes its binding with SRF, was used in competition assays. Moreover, this band is supershifted by a SRF supershift antibody, but not by a JunD supershift antibody.

In contrast, when junB SRE and c-fos SRE oligonucleotides were incubated with proteins extracted from the nuclei of differentiated cells, mobility shift analysis showed a >90% reduction in the binding activity of SRF to both junB and c-fos SREs (Fig. 2). Repression of SRF binding activity to the SRE is specific in that the binding of the SP-1 transcription factor to its DNA domain shows no repression by differentiation. It is important to note that differentiation induces a general 35–40% decrease in total protein expression, including both SP-1 and SRF, in whole cell extract preparations. Therefore, decreased binding activity of the SRF to the junB and c-fos SREs correlates with the previously described loss of transcriptional activation function of these elements during adipocyte differentiation (9).

The sequence motif CC[A/T]₆GG, which is the binding site for SRF, is also called the CArG box (12). It is found not only in the promoters of serum-inducible genes involved in controlling proliferation, but also in the promoter regions of some muscle-specific genes. We therefore investigated whether the repression of SRF binding during adipocyte differentiation is characteristic of only a unique subset of SREs or of a broader array of SREs. We synthesized four additional SRE-containing oligonucleotides characteristic of the EGR2, β-actin, thrombospondin-1, and dystrophin promoters and used them as the binding targets for mobility shift assays. It is noteworthy that the [A/T]₆ sequence in each of the six SREs is different (25, 26, 27, 28).

When mobility shift assays were performed using these SRE-oligonucleotides and nuclear proteins extracted from growing 3T3T cells, a common SRF complex formed with each SRE, and the yield of the complex was reduced comparably by excess, unlabeled c-fos SRE, but not by a c-fos SRE mutant. The complex was also supershifted by SRF supershift antibody. These results suggest that all four SREs are recognized by SRF and have binding activities similar to that of the c-fos SRE (Fig. 3).

Mobility shift assays were next performed to determine whether adipocyte differentiation represses the binding activity of SRF to the β-actin, EGR2, thrombospondin-1, and dystrophin SREs in a manner comparable to the effect of differentiation on the binding of SRF to the junB and c-fos SREs. Fig. 4 shows that differentiation indeed represses SRF binding to all tested SREs. This suggests that repression of SRF binding activity by adipocyte differentiation is not limited to the c-fos and junB SREs but rather involves a global effect on multiple SRF binding sites. These results further suggest that the [A/T]₆ portion of the CC[A/T]₆GG domain is not critically important to the effects of differentiation because the [A/T]₆ sequence differs in each of the six SREs tested.

Because adipocyte differentiation is a complex process that involves the sequential modulation of many transacting factors that regulate many genes leading to irreversible TD, studies next characterized the kinetics of repression of SRF and SRE interactions by differentiation. It has been well documented that after the initial induction of the adipocyte differentiation pathway, 10–15 days are typically required for TD to occur (3, 5). The first step in the differentiation process involves the establishment of PGA, which requires 3–4 days after the addition of differentiation-promoting medium to growing cells. Thereafter, between days 6 and 8, expression of the nonterminal adipocyte phenotype and repression of serum inducibility of junB and c-fos develop. Although they are highly differentiated, these adipocytes, retain mitogenic responsiveness to selected combinations of mitogens (3, 5). Continuous culture of such adipocytes in differentiation medium to days 10–15 results in the expression of the TD phenotype wherein adipocytes are unresponsive to all physiological mitogens.

Accordingly, Fig. 5 presents the results of kinetic analysis of the repression of SRF binding to the SRE during the process of adipocyte differentiation. Repression of SRF binding to the SRE is not obvious within the first 4 days after the addition of differentiation-promoting medium, when the majority of cells reside in the PGA state. Only after the expression of the nonterminal adipocyte phenotype on days 6–8 is a dramatic repression of SRF binding to the SRE evident. This correlates with the expression of an adipocyte differentiation phenotype that is associated with repression of mitogenic responsiveness to serum and repression of the ability of serum to induce the transcription of junB and c-fos.

The CSV3-1 transformed cell line serves as an outstanding murine model for human cancer because it expresses multiple defects in the multistep process of differentiation. A most significant defect is the inability of differentiation to repress mitogenic responsiveness that is required for TD. In this regard, we have demonstrated that in CSV3-1 cells, differentiation not only fails to repress mitogenic responsiveness to serum, but it also fails to repress the inducibility of the AP-1 factors by serum (7, 8).

We therefore performed a series of experiments on transformed CSV3-1 cells and a spontaneously transformed 3T3T cell line designated DW1 to analyze SRE-SRF interactions. To assure that differentiation-inducing culture conditions per se do not affect the interactions of the SRF and SRE in transformed cells, we first studied the DW1 transformed 3T3T cell line that cannot differentiate. After such cells are cultured in differentiation medium for more than 8 days, the majority of the cells remain undifferentiated. Nuclear extract prepared from these cells was used in mobility shift assays, and the result shows no reduction of SRF binding activity to the junB or c-fos SREs (Fig. 6). This result rules out the possibility that some factors in the differentiation-promoting medium influence SRF binding to the SRE.

We next investigated the effects of differentiation of CSV3-1 cells on the binding of SRF to junB and c-fos SREs. Fig. 7 presents the data showing that differentiation in CSV3-1 cells fails to repress SRF binding to the junB and c-fos SREs. For these assays, which were reproduced multiple times, CSV3-1 cell specimens showed >75% adipocyte differentiation. Therefore, aberrant differentiation as expressed in CSV3-1 cells fails to repress (a) mitogenic responsiveness; (b) the serum-dependent inducibility of Jun and Fos expression; and (c) the interaction of SRF and SRE.

There are many possible molecular mechanisms that could explain how differentiation represses SRF-SRE interactions. However, our results suggest that adipocyte differentiation restricts SRF-SRE interactions by blocking the nuclear localization of SRF in 3T3T cells, but not in CSV3-1 cells. Fig. 8 documents this fact using immunofluorescence methods. The data specifically show that SRF is clearly localized to the nucleus in undifferentiated 3T3T cells, but not in differentiated 3T3T cells. Differentiation-induced inhibition of SRF nuclear localization is not a generalized phenomenon because Fig. 8 also shows that differentiation does not block the nuclear localization of the SP-1 transacting factor.

Because the staining of differentiated 3T3T adipocytes with the anti-SRF antibody shows a high immunofluorescence background that makes it difficult to illustrate the clear lack of nuclear staining, Western blotting methods were also used to characterize the relative amount of SRF in whole cell versus nuclear preparations of undifferentiated and differentiated nontransformed 3T3T cells and neoplastically transformed CSV3-1 cells. Fig. 9 clearly shows that differentiation of nontransformed 3T3T cells blocks the nuclear localization of SRE, whereas differentiation fails to block SRF nuclear localization in transformed CSV3-1 cells.

These results strongly suggest that differentiation represses SRF-SRE interactions in nontransformed 3T3T cells by blocking the ability of SRF to localize to the nucleus. In addition, these results explain the differences between the effect of differentiation on SRF-SRE interactions of nontransformed and transformed cells.

The multistep process of adipocyte differentiation in nontransformed 3T3T cells progressively represses mitogenic responsiveness (3, 5, 9). NTD adipocytes specifically show a significant repression in their mitogenic responsiveness to serum and PDGF. Whereas 5–10% serum induces >95% of quiescent undifferentiated cells to undergo mitogenesis, <20% of NTD cells are induced to proliferate under these conditions. Similarly, treatment with 10 ng/ml PDGF and 50 μg/ml insulin induces >70% of quiescent undifferentiated cells to grow, but <15% of NTD cells are induced to undergo DNA synthesis. Nevertheless, NTD adipocytes can be efficiently induced to proliferate when they are treated with high concentrations of combinations of mitogens. Therefore, reduced mitogenic responsiveness can be used to clearly distinguish NTD cells from TD cells, which, by definition, cannot be induced by mitogens to proliferate.

A major research focus of our laboratory for the past several years has been to establish the molecular mechanisms that mediate the progressive repression of mitogenic responsiveness by differentiation. We established previously that adipocyte differentiation repressed the AP-1 DNA binding activity of the Jun and Fos transcription factors and that differentiation represses the serum-inducible transcription of junB and c-fos genes (9). Previous reports also established that neoplastic transformation blocks the ability of differentiation to repress mitogenic responsiveness (7, 8).

The current studies were designed to test the hypothesis that an essential mechanism underlying the ability of differentiation of nontransformed 3T3T cells to repress mitogenic responsiveness and the serum-inducible transcription of junB and c-fos factors involves modulation of the interaction of the SRE and the SRF. We first tested whether adipocyte differentiation represses the activity of SREs in transient transfection assays using 3T3T cells and showed that adipocyte differentiation markedly represses the serum inducibility of pjunB SREtk/CAT and pc-fos SREtk/CAT activities. We next established that adipocyte differentiation represses the junB and c-fos SRE-SRF interactions using mobility shift and gel supershift assays. The effects of differentiation on SRE-SRF interactions were also established not to reflect a global repression of DNA-binding factors because differentiation does not significantly alter the interaction of SP-1 with its DNA binding site. Furthermore, the ability of differentiation to repress SRE-SRF interaction correlates with the kinetics of the transcriptional repression of the serum inducibility of junB and c-fos by differentiation. This suggests that the disruption of SRF-SRE interactions accounts, at least in part, for the differentiation-induced transcriptional repression of growth response genes and might explain the loss of mitogenic responsiveness that is seen in differentiated cells. This conclusion is supported by the fact that differentiation repressed the ability of SRF to bind to the SREs from three other immediate early genes: (a) EGR2; (b) thrombospondin-1; and (c) β-actin. Because many additional immediate early genes carry SREs within their regulatory regions, such as EGR1, cyr61, pip92, and zif-268 (29), their serum inducibility might also be repressed by adipocyte differentiation. In addition to its role in regulating the activation of many immediate early genes, some studies suggest that SRF-SRE interaction might be important and necessary in regulating cell cycle progression (30). It has been shown that inhibiting the interaction of SRF-SRE through microinjection of anti-SRF antibodies or injection of SRE oligonucleotides efficiently blocks the entry of stimulated fibroblasts into S phase (30). In fact, DNA synthesis is inhibited even when cells are injected 8–15 h after serum stimulation. This suggests that SRF-SRE interactions are continuously involved and required in the proliferation pathway. SRF-SRE interactions are also involved in regulating the expression of several muscle-specific genes. Therefore, the data showing the repression of SRF binding to the muscle-specific dystrophin gene SRE by adipocyte differentiation is not surprising because myogenic differentiation and adipocyte differentiation are two highly distinct biological processes.

How does nonterminal adipocyte differentiation repress SRE-SRF interactions? A series of possibilities exists. Differentiation could modify SRF by phosphorylation to restrict SRE-SRF interactions, as has been demonstrated in other biological systems (18, 20, 21). Next, differentiation could induce the expression of a SRE-binding factor to compete with SRF for its DNA binding site. Such factors could include YY1, myogenin, Nkx-2.5, or E12 (31, 32, 33, 34). Differentiation could also repress SRF transcription or translation so that it would not be available to bind SRE. In addition, differentiation could restrict the nuclear localization of SRF so that it could not physically bind to SREs. In this regard, cell differentiation has been previously shown to modulate subcellular localization of N-myc; it is usually localized to the nucleus of embryonic neural tissues, but it accumulates in the cytoplasm upon differentiation of specific classes of neurons (35). Similar effects have been reported for human D4S234 (36). The data in this study further suggest that adipocyte differentiation in 3T3T cells represses SRF-SRE interactions by blocking the nuclear localization of SRF. How this differentiation-dependent effect is mediated remains unknown, but it might involve modification of the nuclear localization signal sequence that exists within SRF (37).

Our discovery that differentiation of transformed CSV3-1 cells cannot repress SRF-SRE interactions has the potential to be an important discovery, especially from the perspective of previously published data showing that differentiation of CSV3-1 cells fails to repress the serum inducibility of AP-1 factors (38), whereas treatment of CSV3-1 cells with atypical mitogens can induce junB expression (39). The discovery that differentiation of transformed CSV3-1 cells does not block the nuclear localization of SRF as occurs in nontransformed cells suggests that neoplastically transformed cells may express lesions in the molecular mechanisms that regulate the nuclear transport of critical regulatory factors. Our findings are in accord with other reports showing that the nuclear localization of other transacting factors can be modified during carcinogenesis. These include the glucocorticoid receptor, ISGF-3, nuclear factor κB, and AP-1 factors (37). It has also been reported that the tumor suppressor effects of wild-type p53 can be inactivated by blocking its nuclear localization in selected cases of breast cancer and neuroblastoma (40). We therefore favor the concept that many cancers may express defects in the ability of the nuclear membrane and/or nuclear pores to correctly control and maintain the critical balance of key regulatory factors in the nucleus.

We thank Wanda Patrick and Joyce Baker for assistance in typing the manuscript and preparing the figures and Robin Cox for technical advice.