The adipocyte enhancer‐binding protein (AEBP1) is a novel transcriptional repressor with carboxypeptidase activity. A two‐hybrid screen was conducted to identify components of AEBP1 that might be important in regulating its activity. The γ5 subunit of a heterotrimeric G protein was shown to bind specifically to AEBP1 and to attenuate its transcriptional repression activity. Adipogenic stimulation selectively decreased the Gγ5 level and enhanced the transcriptional repression activity of AEBP1 during mitotic clonal expansion at the onset of adipogenesis. Thus, the actions of Gγ5 and AEBP1 are directly linked, which could provide the basis for the regulation of transcription at the onset of differentiation. This report shows that a signal‐transducing molecule is involved, by direct protein–protein interaction, in the regulation of transcription during adipogenesis.
The stimulation of adipose differentiation is characterized by an increased lipogenic capacity, a change from a fibroblast morphology to the unilocular appearance of mature adipocytes and an alteration in the expression levels of hundreds of proteins (for a review, see Gregoire et al., 1998), including the adipose P2 gene (aP2) that encodes the adipocyte fatty acid‐binding protein (Hunt et al., 1986; Matarese and Bernlohr, 1988). In the aP2 promoter region, the AE‐1 sequence (nucleotides −159 to −125) functions as an enhancer element in the regulation of aP2 expression (Distel et al., 1987; Herrera et al., 1989), and trans‐acting factors, including the adipogenic transactivator C/EBPα (Christy et al., 1989; Herrera et al., 1989; Cao et al., 1991), human pre‐adipocyte factors (Ro and Roncari, 1991) and the transcriptional repressor adipocyte enhancer‐binding protein 1 (AEBP1) (He et al., 1995), have been defined. AEBP1 is a novel carboxypeptidase (CP), and its CP activity is vital for the transcriptional repression function. AEBP1 expression is down‐regulated during adipocyte differentiation (He et al., 1995).
To understand further the physiological role of AEBP1, we have used a two‐hybrid screen (Chien et al., 1991; Durfee et al., 1993; Harper et al., 1993) to search for proteins that interact with AEBP1. These yeast two‐hybrid studies revealed that AEBP1 interacts directly with the γ5 subunit (Fisher and Aronson, 1992) of a heterotrimeric G protein. Heterotrimeric G proteins are important mediators of signal transduction in eukaryotic cells, linking ligand‐bound seven‐transmembrane receptors (also called G protein‐coupled receptors) with the intracellular machinery such that signals delivered extracellularly result in an appropriate cellular response (for a review see Clapham and Neer, 1997). Upon receptor activation by an extracellular signal and interaction of the triggered receptors with G proteins, the α subunit releases GDP in exchange for GTP. The trimer then dissociates into α and βγ subunits that can interact independently with effectors. The Gβγ dimer plays a major part in signal transmission (for a review see Clapham and Neer, 1997). The physiological role of AEBP1 in this signal transduction presently is unknown.
Here we describe co‐immunoprecipitation, cell fractionation and immunofluorescent experiments which demonstrate that AEBP1 complexes with Gγ5 and Gβ, and that the G proteins localize to the nucleus in 3T3‐L1 but not in NIH 3T3 cells. We also have used ectopically expressed Gγ5 or recombinant Gγ5 protein to demonstrate that Gγ5 attenuates the transcriptional repression activity of AEBP1 in transiently transfected cells or in an in vitro transcription run‐off assay. Finally, we show that adipogenic stimulation selectively reduces the abundance of Gγ5 and enhances the repression activity of AEBP1 during the mitotic clonal expansion stage of adipogenesis. Hence, the actions of Gγ5 and AEBP1 are directly linked, which could provide the basis for the regulation of transcription at the onset of differentiation.
AEBP1 complexes with the γ5 and β subunits of a heterotrimeric G protein
We used co‐immunoprecipitation to establish that AEBP1 and Gγ5 associate in mammalian cells. Cell extracts were subjected to immunoprecipitation with antibody to either AEBP1 or Gγ5, and each immunoprecipitate was separated by SDS–PAGE and probed with antibody to either AEBP1 or Gγ5. As shown in Figure 1A (top panel), AEBP1 was detected in the Gγ5 immunoprecipitates (lane 2) as well as in the AEBP1 immunoprecipitate (lane 1). No such protein band was detected in the control immunoprecipitate (lane 4). The immunoprecipitated samples were also analyzed by immunoblotting with anti‐Gγ5 antibody. As shown in Figure 1A (bottom panel), Gγ5 was detected in the AEBP1 immunoprecipitate (lane 1) as well as in the Gγ5 immunoprecipitate (lane 2). No such protein band was detected in the control immunoprecipitate (lane 4).
To determine further whether AEBP1 pulls down Gγ5 specifically, extracts from cells transfected with an expression plasmid (pJ3H‐AEBP1) encoding a hemagglutinin (HA)‐tagged version of AEBP1 (HA‐AEBP1) were subjected to immunoprecipitation with antibody directed against HA, Gγ5 or the related γ subunit Gγ7, and each immunoprecipitate was analyzed as above. As shown in Figure 1B, HA‐AEBP1 was detected in the Gγ5 immunoprecipitate (lane 7), but not in the Gγ7 immunoprecipitate (lane 8). HA‐AEBP1 was not detected in the Gγ5 immunoprecipitate from cells transfected with the control plasmid pJ3H‐AEBP1(−) (lane 2). Moreover, no HA‐AEBP1 was detected in control (anti‐IGFR) immunoprecipitates from cells transfected with either pJ3H‐AEBP1 (lane 10) or pJ3H‐AEBP1(−) (lane 5). These results strongly suggest that AEBP1 and the γ5 subunit are associated in mammalian cells. The interaction of AEBP1 with Gγ5 is specific since AEBP1 was not co‐precipitated with Gγ7 (lane 8), which implies that the lipid moiety normally present on Gγ subunits may not be responsible for the interaction.
Since G protein β and γ subunits are thought to be tightly associated in vivo and their regulatory effects are ascribed to the intact dimer (for a review see Clapham and Neer 1997), we examined whether Gβ is also present in the AEBP1–Gγ5 complex. Indeed, AEBP1 was detected in the Gβ immunoprecipitate (Figure 1A, top panel, lane 3). Similarly, HA‐AEBP1 was detected in the Gβ immunoprecipitate from cells transfected with pJ3H‐AEBP1 (Figure 1B, lane 9), but not that from cells transfected with pJ3H‐AEBP1(−) (lane 4). Conversely, Gβ was detected in the AEBP1 immunoprecipitate (Figure 1A, middle panel, lane 1) as well as in the Gβ immunoprecipitate (lane 3). As expected, Gβ was detected in the Gγ5 immunoprecipitate (lane 2). No such protein band was detected in the control immunoprecipitate (lane 4). Thus, in mammalian cells, AEBP1 and the Gβγ5 dimer may form a ternary complex. We do not exclude the possibility that other proteins associate in such a complex, or that smaller complexes of subsets of these proteins can form.
The Gγ5 and Gβ subunits localize in the nucleus in 3T3‐L1 cells
The AEBP1–Gγ5 interaction may occur only when Gγ5 localizes in the nucleus or when AEBP1 is associated with the plasma membrane. Gγ subunits are found mainly at the plasma membrane, where they remain associated by integration of the lipid moiety into the phospholipid bilayer. All 10 mammalian γ subunits, γ1, γ2, γ3, γ4, γ7, two γ8s, γ10, γ11 and γ12, have the C‐terminal CAAX motif (where C = cysteine, A = aliphatic amino acid, and X = any amino acid), which is a signal for the post‐translational modification (prenylation and carboxylmethylation) of the cysteine residue. The three C‐terminal residues are removed proteolytically, and the carboxyl group of the resulting prenylcysteine is carboxyl‐methyl esterified (for a review see Zhang and Casey, 1996). Gγ5 is unique in that it has a non‐aliphatic serine and an aromatic amino acid phenylalanine in the CAAX motif (CSFL). Recently, an alternative processing pattern, which retains the three terminal amino acids, for the γ5 subunit has been identified (Cook et al., 1998). This alternative pattern of processing might explain the unique association of γ5 with focal adhesion plaques in neonatal cardiac fibroblasts (Hansen et al., 1994) and in Swiss 3T3 and C6 glioma cells (Ueda et al., 1997). In yeast, disruption of the gene coding for the protease that removes the three C‐terminal amino acids of prenylated proteins results in the redistribution of Ras2p to internal membranes (Boyartchuk et al., 1997). Therefore, it is possible that γ5 targets to different locations in the cell.
Since the nuclear and plasma membrane fractions were of major interest in this study, we performed subcellular fractionation in which nuclear (N), post‐nuclear soluble (S) and particulate (PP) [which contains mainly the plasma membrane rather than other membrane compartments, e.g. the endoplasmic reticulum (ER), lysosome and endosomes] fractions were prepared and analyzed by immunoblotting. Two isoforms of the acetyl‐CoA carboxylase (of 220 and 240 kDa), a cytosolic protein, are seen as one band on an 8.5% polyacrylamide gel, and the band can be seen almost exclusively in the S, but not in the N or PP, fraction (Figure 2A, top panel). Thus, the N fraction is largely free of contamination with intact cells or debris. As shown in Figure 2A, c‐Myc, a transcription factor, was localized exclusively in the N fraction. TFIIB, a general transcription factor, was also present mainly in the nuclear fraction, while a lesser amount (<10% of total signal) was observed in other fractions. On polyacrylamide gels, insulin‐like growth factor receptor (IGFR), a transmembrane protein, migrates as a 100 kDa band, and a precursor of the receptor (pIGFR) is also detected above 200 kDa. As shown in Figure 2A, these two forms of IGFR were primarily in the PP fraction. A band with an apparent mol. wt of ∼83 kDa in the N fraction may represent a non‐specific protein or an immunologically related protein. Therefore, our fractionation procedure is satisfactory.
To determine the subcellular distribution of AEBP1 and Gγ5, the amount of protein representing the cell equivalent of each fraction was analyzed by Western bloting. By this analysis, AEBP1 was observed at similar levels in the N and S fractions prepared from the pre‐adipocyte cell line 3T3‐L1 (Figure 2B). The AEBP1 band in the N fraction migrated slightly more slowly than that in the S fraction, which suggests that AEBP1 is modified post‐translationally and that the modification may determine the localization of AEBP1. Gγ subunits are expected to localize to the plasma membrane, i.e. in the PP fraction. Since Gγ5 and Gγ7 are the major Gγ subunits in fibroblasts such as 3T3‐L1 cells (Ueda et al., 1997), the localization patterns of these two Gγ subunits were determined. As expected, Gγ7 was localized almost exclusively to the PP fraction. In contrast, a substantial amount of Gγ5 was detected in the N fraction (Figure 2B). This is in sharp contrast to the pattern in NIH 3T3 cells, a mouse embryonic fibroblastic cell line that is unable to differentiate to adipocytes, in which Gγ5 was localized mainly to the PP fraction (Figure 2C). This finding indicates that Gγ5 specifically localizes to the nucleus in 3T3‐L1 cells, where it potentially could interact with AEBP1, or that Gγ5 may be transported to the nucleus in a complex with AEBP1. Since G protein β and γ subunits are thought to be tightly associated in vivo, it is likely that Gγ5 and Gβ localize together in the nucleus. Since Gγ5 is the most abundant Gγ subunit in fibroblasts, it is assumed that the majority of Gβ subunits would form dimers with Gγ5 and that their subcellular distribution patterns would be similar. Indeed, similar subcellular distributions of Gβ and Gγ5 were observed (Figure 2B). Since the Gβ antibody was raised against a peptide common to all six known Gβ subunits, the band probably represents a pool of Gβ subunits that interact with Gγ5. A weaker band was seen in the S fraction, which may represent Gβ in the ER and/or Golgi apparatus, or contamination from other fractions. Upon longer exposure, a smaller amount of Gγ5 was also seen in the S fraction. The distribution pattern in NIH 3T3 cells showed that both Gβ and Gγ5 subunits are localized mainly to the PP fraction (Figure 2C).
Confocal laser microscopic analysis (Hansen et al., 1994) confirmed the nuclear localization of Gγ5 and Gβ in 3T3‐L1 cells (Figure 2D, panels A and E). Consistent with the cell fractionation data, Gγ5 and Gβ were not detected in the nuclei of NIH 3T3 cells (panels B and F). As in the case of the cell fractionation analysis, Gγ7 was also not detected in the nuclei of either cell type (panels C and D). Therefore, it appears that the nuclear localization of the G protein subunits is cell type specific. Nuclear localization of several Gα (e.g. Gsα, Giα2 and Giα3) and Giβ subunits has been reported (Crouch, 1991; Bégin‐Heick et al., 1997; Crouch and Simson, 1997). The physiological significance of the nuclear localization of these G protein subunits is not clearly understood. The nuclear localization of Gγ5 in certain cell types implies a specific physiological role for Gγ5 in such cell types.
Gγ5 regulates transcriptional repression activity of AEBP1
We investigated the biological significance of the association between Gγ5 and AEBP1 by determining AEBP1 transcriptional repression activity in vivo. We have shown previously that the reporter plasmid paP2(−168)CAT, which contains the aP2 promoter region (−168 to +21) fused to the reporter gene chloramphenicol acetyltransferase (CAT), could be used to assess AEBP1 transcriptional repression activity through its binding site (AE‐1) in the proximal promoter region. When an expression plasmid encoding AEBP1 (pSVAEBP1) was co‐transfected with paP2(−168)CAT, CAT activity was significantly decreased (He et al., 1995). We transfected an expression plasmid encoding Gγ5 (pRc/CMVγ5) along with both pSVAEBP1 and paP2(−168)CAT, and determined the repression activity of AEBP1 (He et al., 1995). As shown in Figure 3A, the repression activity of AEBP1 was attenuated by ectopic overexpression of Gγ5, and this inhibition by Gγ5 of repression activity was dose responsive. Transfection of a plasmid containing the Gγ5 cDNA in the opposite orientation [pRc/CMVγ5(−)] along with the two aforementioned plasmids had no effect on CAT expression (Figure 3A). Co‐transfection of pRc/CMVγ5 with paP2(−168)CAT alone had no effect on CAT expression (data not shown). Therefore, the inhibition by Gγ5 of repression activity was not due to non‐specific up‐regulation of CAT expression.
Since it was unclear whether Gγ5 inhibits AEBP1 repression activity directly or via intermediates, we developed an in vitro transcription system utilizing bacterially produced recombinant proteins. The transcription template pCMV/AE‐1 contains a human cytomegalovirus (CMV) gene fragment (−829 to +363) with three copies of the AE‐1 sequence inserted upstream of the CMV promoter (Figure 3B). We tested transcription activity by using an in vitro run‐off transcription assay utilizing HeLa cell nuclear extracts. When recombinant AEBP1 was added to the nuclear run‐off transcription reaction along with the pCMV/AE‐1 template, transcription activity was significantly decreased (Figure 3C, lanes 1 and 2), but when tested with the template pCMV, without AEBP1‐binding sites, recombinant AEBP1 protein had no effect on transcription (lanes 3 and 4). These results demonstrate that, as seen in vivo (He et al., 1995), the repression activity of AEBP1 requires localization to a specific promoter region, and is not due to transcriptional ‘squelching’ or to a non‐specific shutdown of the RNA polymerase II machinery. Next, we tested whether recombinant Gγ5 protein, which would not have the lipid moiety, has the ability to inhibit the transcriptional repression activity of AEBP1. When recombinant Gγ5 protein was added to the nuclear run‐off reaction, the repression activity of AEBP1 was virtually abolished (Figure 3D, lanes 4–6). This derepression effect of recombinant Gγ5 protein was not due to a non‐specific transcriptional enhancement, since the transcription level did not increase when the recombinant Gγ5 protein was added to a transcription reaction with the pCMV/AE‐1 template but without AEBP1 (lane 3). Furthermore, the recombinant Gγ5 protein had no significant effect on transcription of the pCMV template (lanes 7–10). Therefore, these results demonstrate that the lipid moiety may not be necessary for the inhibitory interaction, and that Gγ5 may inhibit the repression activity of AEBP1 directly.
Since AEBP1 binding to the AE‐1 sequence is crucial for transcriptional repression by AEBP1, we determined whether the effect of Gγ5 was due to an interaction with AEBP1 that prevents binding to the AE‐1 site. This possibility was examined by electrophoretic mobility shift assays (EMSAs). We have shown using EMSAs that recombinant AEBP1 protein specifically binds the AE‐1 sequence (He et al., 1995). As shown in Figure 3E, the recombinant Gγ5 protein was able to prevent binding of recombinant AEBP1 protein to the AE‐1 sequence (lanes 10–12). A control recombinant protein had no effect on DNA binding by AEBP1 (lanes 2–4). Moreover, the recombinant Gγ5 protein had no effect on DNA binding by other transcription factors (data not shown). Thus, Gγ5 may regulate transcriptional repression by AEBP1 by preventing the tethering of AEBP1 to the promoter. The Gγ5 protein may directly mask the DNA‐binding domain of AEBP1; alternatively, the interaction may cause a conformational change in AEBP1 which renders it unable to bind the AE‐1 sequence.
Gγ5 level decreases selectively at the onset of adipogenesis
The above results demonstrated that Gγ5 can regulate AEBP1 transcriptional repression activity. We next asked whether the actions of Gγ5 are biologically relevant in adipogenesis. Adipogenesis is a complex process affected, in a positive or negative manner, by various hormones and growth factors (for a review see Gregoire et al., 1998). G proteins also play a prominent role in adipogenesis. The levels of Gsα and Giα2, the G proteins that mediate activation and inhibition of adenylyl cyclase activity respectively, have been shown to modulate the rate of adipogenesis in 3T3‐L1 cells (Wang et al., 1992; Su et al., 1993). These studies indicated that Gsα and Giα2 are acting in a counter‐regulatory manner, and that they mediate adipocyte differentiation in a manner apparently independent of adenylyl cyclase. Since these G proteins are localized to the nucleus in response to growth factors and hormones and play a prominent role in adipogenesis, one exciting possibility is that Gγ5 may have a direct physiological role in adipogenesis through its actions. To examine this possibility, we determined the subcellular distribution pattern of γ5 and β subunits upon adipogenic stimulation, and found that their subcellular distributions were not changed significantly upon induction (data not shown). However, the abundance of Gγ5 was decreased selectively during the mitotic clonal expansion phase of adipocyte differentiation at days 1 and 2 (Figure 4A). Upon induction, post‐confluent pre‐adipocytes undergo several required rounds of mitotic clonal expansion prior to terminal differentiation (for a review, see Gregoire et al., 1998). It was estimated that Gγ5 was decreased to ∼10 and 50% of the initial level at days 1 and 2, respectively. Right after this period, the amount of Gγ5 returned to the day 0 level. No significant changes in the abundances of Gβ, Gγ7, ERKs, TFIIB and AEBP1 were observed over this period (Figure 4A and B; data not shown).
Adipogenic stimulation enhances the repression activity of AEBP1 at the onset of adipogenesis
The selective decrease in Gγ5 level during a specific stage of adipogenesis and its AEBP1‐inhibitory activity suggest the possibility that these actions are linked directly and are physiologically relevant in adipogenesis. To test this hypothesis, we examined the transcriptional repression activity of AEBP1 upon adipogenic stimulation. The cells co‐transfected with pSVAEBP1 along with paP2(−168)CAT showed ∼35 and 70% of the CAT activity of cells transfected with the control plasmid pSVAEBP1(−) for 1 and 2 days, respectively (Figure 4C, top panel). When the cells transfected with pSVAEBP1 were treated with the adipogenic inducers (IDM), CAT activity was decreased further to ∼20 and 50% of the CAT activity from cells transfected with pSV(−)AEBP1 for 1 and 2 days, respectively (Figure 4C, top panel). This enhancement of repression activity upon adipogenic stimulation is specific for the aP2 promoter, since no significant change in CAT activity was observed for cells transfected with a reporter plasmid (pGALTKCAT) containing a heterologous promoter (He et al., 1995) (Figure 4C, low panel). The increased repression activity is likely to be due to the selective decrease in Gγ5 level mediated by the adipogenic stimulation (Figures 4A and B) because AEBP1 levels did not change during this period (data not shown). Finally, the adipogenic stimulation had no effect on the activity of another transcriptional repressor (AEBP2; He et al., 1999) that does not interact with Gγ5 (data not shown). These results suggest that Gγ5 may be an important mediator that regulates AEBP1 transcriptional repression activity at the onset of adipogenesis.
The regulation of a transcription factor by protein association is not unprecedented. A protein complex may prevent a transcription factor from binding to a specific DNA sequence or from interacting with basal or other transcription factors (for a review see Cowell, 1994). Apparently, the repression function of AEBP1 is regulated by preventing AEBP1 binding to DNA. Remarkably, the protein that prevents AEBP1 from repressing transcription is the γ5 subunit of a heterotrimeric G protein. This is the first report of a signal‐transducing molecule being directly involved, by protein–protein interaction, in transcriptional regulation during adipogenesis. In view of the tight association of the β and γ subunits in vivo, the inhibitory effect of Gγ5 may result from ternary complex formation between the Gβγ5 dimer and AEBP1. However, it is conceivable that Gγ5 may also modulate the transcriptional repression function of AEBP1 directly, as seen in the in vitro experiments. It has been suggested that an intrinsic Gγ monomer may exist in certain systems (Higgins and Casey, 1994; Morishita et al., 1994; Herlitze et al., 1996).
Gγ5 appears to function as a regulator for the transcriptional repression activity of AEBP1 during adipocyte differentiation. We speculate that these actions of Gγ5 and AEBP1 might provide a basis for the regulation of transcription at the onset of differentiation. In this regard, the selective decrease in Gγ5 level upon adipogenic stimulation seems to induce the DNA‐binding and repression activity of AEBP1 during clonal expansion. We speculate that AEBP1 might regulate a certain gene(s) whose expression is altered transiently during clonal expansion. Recent studies suggest that growth‐promoting transcriptional activities and a critical tyrosine dephosphorylation event are involved during mitotic clonal expansion at the onset of adipogenesis (Liao and Lane, 1995; Shugart et al., 1995; Richon et al., 1997). AEBP1 might also regulate the adipogenic gene encoding the transactivator C/EBPα. Expression of antisense AEBP1 RNA causes a marked induction of C/EBPα expression prior to adipocyte differentiation (S.‐W.Kim and H.‐S.Ro, unpublished). There are at least three C/EBP isoforms (α, β and δ) which are expressed during different stages of adipocyte differentiation. These proteins play vital roles in the initiation of adipogenesis and throughout the differentiation process. Typically, expression of C/EBPα occurs after the differentiation process has begun (for a review see Mandrup and Lane, 1997). Expression of C/EBPα at higher than physiological levels also induces adipogenesis in fibroblasts (Freytag et al., 1994), whereas expression of antisense C/EBPα RNA suppresses adipogenesis in 3T3‐L1 cells (Lin et al., 1992). These results have suggested that C/EBPα is a key regulator and plays a vital role in a mechanism which results in adipogenesis. The transcriptional factor AP‐2α/CUP (C/EBPα‐undifferentiated protein) was shown to be involved in repressing the C/EBPα gene in pre‐adipocytes and, upon induction of differentiation, AP‐2α/CUP expression is down‐regulated, allowing expression of C/EBPα (Vasseur‐Cognet and Lane, 1993; Tang et al., 1997; Jiang et al., 1998). AP‐2α/CUP expression, which persists during clonal expansion, is abolished at day 4 post‐induction when the expression of the C/EBPα gene is induced. Since C/EBPα has been shown to induce a set of genes involved in growth arrest (Constance et al., 1996), which is required for establishing the differentiation stage, repression of the C/EBPα gene during clonal expansion appears to be crucial. AEBP1 may be involved in the repression of the C/EBPα gene. The enhanced repression activity of AEBP1 during clonal expansion would ensure tight repression of the C/EBPα gene during this phase. Determination of the mechanism of repression of the C/EBPα gene by AEBP1 and assessment of its effects on the differentiation program are important future studies.
Materials and methods
Protein A–agarose and fluorescein isothiocyanate (FITC)‐conjugated goat anti‐mouse or anti‐rabbit antibodies were purchased from Santa Cruz Biotechnology, as were polyclonal antibodies directed against Gγ5, Gγ7, Gβ, PKA and IGFR. Monoclonal antibody directed against the HA epitope was purchased from Boehringer Mannheim. Dulbecco's modified Eagle's medium (DMEM) and bovine calf serum were purchased from Gibco‐BRL. The cosmic calf serum was purchased from HyClone Laboratories. The HeLa nuclear extract was purchased from Promega.
Yeast two‐hybrid screening
Yeast strain Y153 harboring pGBT‐AEBP1, an expression plasmid encoding full‐length AEBP1 fused to the Gal4 DNA‐binding domain, was transformed with a human cDNA library that was constructed in the vector pACT. A positive clone (pACT‐Gγ5) encoding the γ5 subunit of a heterotrimeric G protein fused to the Gal4 activation domain was obtained from ∼7×106 transformants. Interaction between fusion proteins expressed by the specific plasmids was indicated by the growth of yeast transformants on selective medium (SC‐Leu, Trp and His) containing 60 mM 3‐aminotriazole. His+ colonies were analyzed for β‐galactosidase activity using a filter lift procedure (Durfee et al., 1993).
3T3‐L1 cells were collected in cold RIPA buffer containing 1 mM phenylmethylsulfonyl fluoride (PMSF), 10 μg aprotinin/ml, 10 μg leupeptin/ml, 5 mM EDTA and 5 mM EGTA. The cell lysates were incubated with protein A–agarose for 1 h, then the beads were discarded and the supernatants were incubated with specific antibodies (1 μg/ml anti‐Gγ5 or anti‐Gβ, 2 μg/ml anti‐HA, anti‐PKA or anti‐IGFR, or 2.5 μg/ml of either affinity‐purified anti‐AEBP1 antibodies or IgG from normal serum) for 1 h and then overnight with protein A–agarose. Samples were collected and washed four times with RIPA buffer. The precipitated samples were resolved by SDS–PAGE and analyzed by Amersham's ECL blotting system.
Fractionation was performed on ice or at 4°C as described (Oláh et al., 1994) with modifications. Cells were collected into 500 μl of fractionation buffer (2 mM EDTA, 2 mM EGTA, 20 mM Tris–HCl pH 7.5) supplemented with 25 μg leupeptin/ml, 25 μg aprotinin/ml and 1 mM PMSF, and then disrupted by 25 passages through a needle (25 gauge) on ice. Cells were centrifuged for 5 min at 1000 g, and the pellet, which represents the nuclear fraction (N), was washed and centrifuged twice with fractionation buffer. The nuclear fraction was examined under a light microscope and found to contain intact nuclei without any unlysed cells. The supernatant was centrifuged for 1 h at 19 000 g to separate the plasma membrane‐enriched fraction (PP) and soluble cytosolic fraction (S). Equal amounts of these fractions were mixed with sample buffer and resolved by 8.5 or 15% SDS–PAGE, then transferred to nitrocellulose membrane on ice or at 70°C (Gγ analysis; Robishaw and Balcueva, 1993) and analyzed by Amersham's ECL or ECL‐plus blotting system. To assess the degree of contamination by cytosolic proteins of the N fraction, a known cytosolic protein, acetyl‐CoA carboxylase (AC), was used as a marker protein. Since AC is biotinylated in cells (Geelan et al., 1997), the immunoblotting was carried out with horseradish peroxidase(HRP)‐conjugated avidin (Haneji et al., 1993). As nuclear markers, the c‐Myc transcription factor and the general transcription factor TFIIB, that binds to DNA along with TFIID near the TATA box of promoters, were used. IGFR, a transmembrane protein, was used as a plasma membrane marker.
Immunofluorescent confocal laser microscopy
Cells were grown on 18×18 mm No. 1 coverslips, washed twice with phosphate‐buffered saline (PBS) and fixed with 4% formaldehyde in PBS for 10 min at room temperature. Cells were permeabilized with 0.01% saponin (Sigma) in PBS for preservation of G protein subunits in the plasma membrane (Hansen et al., 1994), washed with PBS then blocked by incubation with 5% fat‐free milk powder in PBS for 30 min at room temperature. Samples were incubated with primary antibodies (2 μg/ml for Gβ and Gγ5, 4 μg/ml for Gγ7 and AEBP1) overnight at 4°C and with FITC‐conjugated secondary antibody (3.5 μg/ml) for 1 h at room temperature. Samples were examined with confocal laser microscopy (Carl Zeiss) to photograph the nuclear sections at 3 μm from the bottom of the cells.
Cell culture and differentiation
3T3‐L1 cells were cultured in DMEM containing 10% bovine calf serum until confluent, referred to as day 0, then switched to differentiation medium containing 10% cosmic calf serum, 10 μg of insulin/ml, 1 μM dexamethasone and 0.5 mM 1‐methyl‐3‐isobutylxanthine for 2 days (Green and Kehinde, 1975; Russell and Ho, 1976). Differentiating cells were maintained in DMEM containing 10% cosmic calf serum and 10 μg of insulin/ml with a change of medium at an interval of 3 days.
Transfection and CAT assay
3T3‐L1, NIH 3T3 and COS‐7 cells were cultured in 90 mm dishes and transfected at 80% confluence by the polybrene procedure as described (He et al., 1995). Detailed descriptions of plasmid constructions (pJ3H‐AEBP1, pRc/CMVγ5 and pCMV/AE‐1) are available from the authors on request. Transfection was performed with 2.5 μg of the reporter plasmid paP2(−168)CAT and 5 μg of the AEBP1 expression plasmid pSVAEBP1 [or the control plasmid pSVAEBP1(−)] along with various amounts of the expression plasmid encoding the γ5 subunit (pRc/CMVγ5). The γ5 cDNA was also cloned into expression vector pRc/CMV (Invitrogen) in the opposite orientation [pRc/CMVγ5(−)] to serve as a negative control. All transfections also included 1 μg of pHermes‐lacZ. β‐galactosidase activity was assayed 48 h after transfection to normalize transfection efficiency, and CAT activity was assayed as described (He et al., 1995).
In vitro transcription assay
Nuclear run‐off experiments were carried out using HeLa nuclear extract according to the manufacturer's instructions with pCMV or pCMV/AE‐1 templates (see Figure 3B). Histidine‐tagged recombinant AEBP1 protein (He et al. 1995) (200 ng) was pre‐incubated with various amounts of histidine‐tagged recombinant γ5 protein in a buffer containing 1× HeLa nuclear extract buffer [20 mM HEPES (pH 7.9), 100 mM KCl, 0.2 mM EDTA, 0.5 mM dithiothreitol (DTT), 20% glycerol], 3 mM MgCl2, 0.4 mM each of ATP, CTP and UTP, 0.016 mM GTP, 40 U of RNase inhibitor, 10 μCi of [α‐32P]GTP and 10 ng of template DNA for 30 min on ice; then 8 U of HeLa nuclear extract was added to start transcription (25 μl in total volume) and incubation was continued at 30°C for 1 h. The reaction was stopped by adding 175 μl of stop mixture (300 mM Tris–HCl pH 7.4, 300 mM NaOAc, 0.5% SDS, 2 mM EDTA, 3 μg of tRNA/ml). The product was resolved by electrophoresis on a 6% denaturing polyacrylamide gel and exposed to X‐ray film overnight at −70°C with an intensifying screen.
EMSA for AEBP1–Gγ5 interaction
Recombinant AEBP1 protein was pre‐incubated with recombinant γ5 or control (recombinant Rab11) proteins for 10 min at room temperature in binding buffer (100 mM Tris pH 7.5, 100 mM KCl, 50 mM MgCl2, 10 mM DTT and 25% glycerol). 32P‐Labeled AE‐1 oligonucleotides were added and incubated for a further 20 min at room temperature. For competition studies (Figure 3E, lane 8), an ∼200‐fold molar excess of cold unlabeled AE‐1 oligonucleotides was added to the binding reaction.
We thank R.A.Singer and N.D.Ridgway for discussion and comments on the manuscript; S.J.Elledge for the human library in pACT vector; J.Chernoff for plasmid pJ3H; P.P.Poon for the recombinant Rab11 protein; all the members of the Yeasty Group for help in the yeast two‐hybrid screening; J.Shi for technical assistance; and the Department of Biology for immunofluorescent confocal laser microscopy. This work was supported by grants from the Heart and Stroke Foundation (Nova Scotia) of Canada, the Canadian Diabetes Association and NSERC to H.‐S.R. We acknowledge the support of a Walter C.Sumner Memorial Fellowship to A.M. and an HSFC Research Scholarship to H.‐S.R.
- Copyright © 1999 European Molecular Biology Organization