Repression of adipogenesis by adenylyl cyclase stimulatory G-protein alpha subunit is expressed within region 146-220.

The heterotrimeric G-protein alpha subunit stimulatory with respect to adenylyl cyclase (Gsalpha) represses adipogenesis of 3T3-L1 mouse embryonic fibroblasts. Derepression occurs in response to inducers, to oligodeoxynucleotides antisense to Gsalpha, and to overexpression of heterotrimeric G-protein alpha subunit 2, inhibitory with respect to adenylyl cyclase (Gialpha2). Constitutive expression of Gsalpha blocks adipogenesis and was exploited as an assay, in which chimeras of Gialpha2 and Gsalpha were expressed stably in 3T3-L1 cells to define the region controlling adipogenesis. N-terminal analysis revealed region 146-220 of Gsalpha as a repressor of adipogenesis; substitution of Gialpha2 abolished the ability of the chimera to repress adipogenesis in response to inducers. Expression of a chimera in which the 146-235 region of Gsalpha was embedded in Gialpha2 fully repressed adipogenesis in response to the inducers. C-terminal analysis revealed no loss of function for truncated Gsalpha, lacking the terminal 38 residues. The repressor domain for adipogenesis maps to a region that includes switch domains I and II and is spatially distinct from the regions mapped for control of adenylyl cyclase.

The heterotrimeric G-protein ␣ subunit stimulatory with respect to adenylyl cyclase (G s␣ ) represses adipogenesis of 3T3-L1 mouse embryonic fibroblasts. Derepression occurs in response to inducers, to oligodeoxynucleotides antisense to G s␣ , and to overexpression of heterotrimeric G-protein ␣ subunit 2, inhibitory with respect to adenylyl cyclase (G i␣2 ). Constitutive expression of G s␣ blocks adipogenesis and was exploited as an assay, in which chimeras of G i␣2 and G s␣ were expressed stably in 3T3-L1 cells to define the region controlling adipogenesis. N-terminal analysis revealed region 146 -220 of G s␣ as a repressor of adipogenesis; substitution of G i␣2 abolished the ability of the chimera to repress adipogenesis in response to inducers. Expression of a chimera in which the 146 -235 region of G s␣ was embedded in G i␣2 fully repressed adipogenesis in response to the inducers. C-terminal analysis revealed no loss of function for truncated G s␣ , lacking the terminal 38 residues. The repressor domain for adipogenesis maps to a region that includes switch domains I and II and is spatially distinct from the regions mapped for control of adenylyl cyclase.
Mouse embryo fibroblast 3T3-L1 cells differentiate to adipocytes Kehinde, 1974, 1975;Green and Meuth, 1974), in response to inducers such as insulin (Russell and Ho, 1976) or dexamethasone and methylisobutylxanthine (DEXϩMIX) 1 in combination (see Wang et al., 1992, and Refs. therein). In addition to mediating signaling from a populous class of plasma membrane receptors to a less populous group of effectors, including adenylyl cyclases, phospholipase C␤, and various ion channels (Gilman, 1987;Birnbaumer et al., 1990;Bourne et al., 1990), G-proteins participate in more complex biological responses, including oncogenesis , early  and neonatal (Moxham et al., 1993) mouse development, and cellular differentiation (Strittmatter et al., 1994;Wang et al., 1992). The G-protein G s␣ is implicated as playing a critical role in adipogenesis based on the following observations: (i) G s␣ levels decline during differentiation; (ii) metabolic labeling with [ 35 S]methionine reveals a sharp decline in the rate of synthesis of G s␣ during differentiation; (iii) cholera toxin activation of G s␣ blocks differentiation; (iv) oligodeoxynucleotides antisense to G s␣ accelerate differentiation in response to dexamethasone and methylisobutylxanthine; (v) oligodeoxynucleotides antisense to, but neither missense nor sense to, G s␣ are themselves inducers of differentiation, (vi) overexpression of the counterregulatory Gprotein G i␣2 induces adipogenesis in the absence of inducers; and (vii) constitutive expression of G s␣ blocks the ability of inducers to promote adipogenesis (Watkins et al., 1987(Watkins et al., , 1989a(Watkins et al., , 1989bWang et al., 1992, Su et al., 1993. Cholera toxin blocks differentiation, yet agents that raise intracellular cyclic AMP levels such as forskolin or pertussis toxin, or addition of dibutyryl cyclic AMP fail to influence adipogenesis (Wang et al., 1992). These observations suggest that some effector other than adenylyl cyclase may be responsible for propagating the effects of G s␣ and G i␣2 on differentiation. In the current work, we make use of the ability of constitutive expression of G s␣ to block induction of adipogenesis, creating chimeras in which regions of G s␣ are replaced with G i␣2 and evaluating their ability to block differentiation. The results identify regions of G s␣ critical to regulation of differentiation, mapping to a region of the molecule including switch I and II regions (Lambright et al., 1994) but distinct from the domains regulating adenylyl cyclase (Coleman et al., 1994).

EXPERIMENTAL PROCEDURES
Mouse embryo fibroblast 3T3-L1 cells were obtained from the American Type Culture Collection (Rockville, MD). Cells were maintained in culture in 100-mm Petri dishes in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum. The protocols for stable transfection were described previously (Wang et al., 1992;. Transfectant clones selected in G418 were identified and replicate plated in 8-well chamber slides as well as 24-well plates that were sparsely seeded. Cells in 24-well plates were maintained for propagation. Protocols for indirect immunofluorescence and histochemical staining techniques are described elsewhere (Wang et al., 1992;Su et al., 1993). Construction of chimeras within expression vector pCW1 and structure-function analysis of control of adenylyl cyclase were as described earlier Gupta et al., 1990). The clones treated with DEXϩMIX were harvested at day 7. For clones not treated with DEXϩMIX, cultures were examined at day 7 but routinely harvested at day 18 to provide the fullest opportunity to detect adipogenic conversion in the absence of inducers. In all cases, the results obtained with clones not treated with DEXϩMIX were identical at days 7 and 18 (not shown). For Northern blot RNA hybridization, cells were harvested by phosphate-buffered saline and EDTA treatment and low speed centrifugation. Total RNA was extracted using RNA Stat-60 reagent (TEL-TEST B, Inc., Friendswood, TX). RNA (30 g/ lane) was separated on 1% agarose-formaldehyde gels and transferred to nylon membranes in the presence of 20 ϫ SSC. Membranes were hybridized for 20 h at 42°C with the aP2 cDNA, which was labeled by * This work was supported by United States Public Health Service Grant DK30111 from the NIDDK, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

RESULTS
Constitutive expression of G s␣ blocks the ability of dexamethasone and methylisobutylxanthine (DEXϩMIX) to induce adipogenesis, whereas expression of G i␣2 itself induces adipogenesis in the absence of DEXϩMIX. The strategy adopted to identify the domain(s) controlling adipogenesis was to express chimeras of G s␣ and G i␣2 and to evaluate their influence on adipogenesis in both the absence and presence of inducers. Chimeras with variable N-terminal substitution of G i␣2 for G s␣ , a C-terminal 38-residue substitution of G i␣2 for G s␣ , a C-terminal truncation of G s␣ lacking 38 residues, and constitutively activated mutants of G s␣ and G i␣2 with diminished intrinsic GTPase activities were prepared and used to stably transfect 3T3-L1 cell cultures (Fig. 1).
Stably transfected clones harboring the neomycin resistance gene were selected in the neomycin analog G418 and examined for G-protein subunit expression by indirect immunofluorescence microscopy ( Fig. 2A). In the absence of the primary (Ϫ1st Ab) or secondary (Ϫ2nd Ab) antibodies, no epifluorescence signal was detected. Immunostaining with a primary antibody raised against the C-terminal decapeptide of G i␣2 (CM-112) revealed faint staining of endogenous subunits in the cultures transfected with an empty expression vector alone, pCW1. Cultures stably transfected with pCW1 harboring the G s␣ (356)/ G i␣2 chimera displayed a strong epifluorescence signal following staining with the anti-G i␣2 antibodies. Expression of the G i␣2 (122)/G s␣ and G i␣2 (212)/G s␣ chimera, the Q227L constitutively activated mutant of G s␣ , and G s␣ itself was established by staining the cultures with anti-G s␣ antibody (CM-129) raised against the C-terminal decapeptide of the G-protein subunit (Fig. 2B). Expression of the C-terminal truncated mutant of G s␣ was established by staining with antibodies raised against the holoprotein rG s␣ . Immunoblotting of crude cell extracts confirmed these results, obtained by indirect immunofluorescence (Fig. 2C, a-c). Expression of G-protein ␣ subunits in the stable transfectant clones was approximately 1.5-fold over endogenous G s␣ levels. As shown earlier (Wang et al., 1992), endogenous levels of G s␣ decline during differentiation in response to DEXϩMIX, whereas levels of the vectordriven expression of ␣ subunits remains relatively constant (Fig. 2C, d).
In wild-type cells and cells stably transfected with the empty expression vector, no differentiation was observed in the absence of the inducers DEXϩMIX, as detected by staining for lipid with oil red O (Fig. 3). Nuclei were made visible by counterstaining with hematoxylin. When exposed to DEXϩMIX, cultures displayed nearly complete differentiation, as typified by lipid accumulation, the hallmark of adipocytes. Cultures stably transfected with the empty vector alone displayed robust differentiation in response to DEXϩMIX also. Expression of either wild-type G s␣ or the GTPase-deficient, activated Q227L mutant of G s␣ resulted in cultures that failed to respond to DEXϩMIX, unable to differentiate into adipocytes. Much like the short term effects of cholera toxin activating G s␣ (Wang et al., 1992), stable expression of G s␣ blocks adipogenesis.
The role of the N-terminal domain of G s␣ was analyzed using a series of chimeras with increasing degree of substitution with G i␣2 (Fig. 1). N-terminal substitutions of G s␣ to residue 145 with analogous regions of G i␣2 were largely unremarkable (Fig.  4). Stable expression of the G i␣2 (122)/G s␣ chimera or the G i␣2 (54)/G s␣ chimera (not shown) effectively blocked the ability of the inducers to provoke adipogenesis, much like constitutive expression of G s␣ . When the region of G i␣2 substitution for G s␣ increased from 122 to 212, corresponding to region 145-235 in G s␣ , the phenotype changed dramatically. Whereas expression of G s␣ or chimeras with N-terminal substitutions to 145 blocked DEXϩMIX-induced adipogenesis, expression of G i␣2 (212)/G s␣ chimera no longer blocked adipogenesis, relieving the repression of differentiation by G s␣ , much like oligodeoxynucleotides antisense to G s␣ (Wang et al., 1992). Constitutively expressed G i␣2 provoked adipogenesis in either the absence or presence of the inducers, whereas expression of the G i␣2 (212)/G s␣ chimera does not induce adipogenesis in the absence of DEXϩMIX. Thus, it is not the presence of the additional region of G i␣2 but, rather, the loss of the corresponding region of G s␣ that ablates the block of differentiation.
The role of the C-terminal domain of G s␣ was explored through comparison of the effects of wild-type compared with a mutant of G s␣ in which the C-terminal 38 residues have been removed. As shown in Fig. 5, constitutive expression of either the wild-type or truncated version of G s␣ blocks the ability of DEXϩMIX to induce adipogenesis. Although these data suggest no major role of the C-terminal 38 residues in the ability of G s␣ to repress adipogenesis, expression of a chimera in which the C-terminal 38 residues of G s␣ (356 -394) are replaced with the analogous region of G i␣2 (320 -355) reveal a more complex picture. Expression of the G s␣ (356)/G i␣2 chimera itself does not induce adipogenesis but does derepress the block of adipogenesis in response to DEXϩMIX in 20 -25% of the culture. Examination of the cultures for 4 -10 days after induction reveals this rather constant percentage of the cells progressing to adipocytes when the G s␣ (356)/G i␣2 chimera is stably expressed.
aP2 is an early marker expressed when the embryonic fibroblasts commit to the adipocyte phenotype. In the 3T3-L1 cells,  (Su et al., 1993) using antibodies specific for the C terminus of G s␣ (CM-129; NULL (Ϫ2nd Ab), NULL, ␣s, ␣sQ227L, ␣i2(122)/s, and ␣i2(212)/s), the C terminus of G i␣ (CM-112; NULL and ␣s(356)/i), and recombinant G s␣ (CM-474, data not shown). Clones were selected on the basis of expression relative to wild type. The results from the indirect immunofluorescence the expression of aP2 mRNA was analyzed by RNA blotting (Fig. 6A). Within 48 h of exposure to DEXϩMIX, aP2 mRNA expression increases dramatically. By day 3, aP2 mRNA levels have peaked and are sustained in the mature phenotype. The expression of this early marker for differentiation was analyzed in cells stably expressing the chimera and mutant forms of G s␣ and G i␣2 (Fig. 6B). Expression of G s␣ , two GTPase-deficient mutants of G s␣ (G225T and Q227L), the G i␣2 (122)/G s␣ chimera, and G s␣ with the truncated C terminus all displayed no significant induction of aP2 mRNA in response to DEXϩMIX. Exwere confirmed independently by immunoblotting of whole cell extracts prepared from the clones selected for further study (not shown). In each case, three to five independent clones were propagated for study, all yielding the same phenotype with respect to adipogenenic conversion. PC, phase-contrast microscopy; EPI, epifluorescence microscopy. C, immunoblots of crude cell membranes (a and b) and whole cell extracts (c and d) of stable transfectant clones. Samples (50 g protein/lane) were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to nitrocellulose, probed with the anti-subunit-specific rabbit antisera (1:200 dilution) indicated, and stained with calf alkaline phosphatase-linked goat anti-rabbit IgG (Watkins et al., 1987). The levels of expression in the stable transfectants relative to endogenous G s␣ levels (set at 1.0) were as follows: empty vector, 1.0; G s␣ , 1.6; G s␣ Q227L, 1.7; G s␣ G225T, 1.24; G i␣2 (122)/G s␣ , 1.3; G i␣2 (212)/G s␣ , 1.4; G i␣2 (122)/G s␣ (235)/G i␣2 , 1.7; G s␣ (356) truncate (TRUN), 1.25; and G s␣ (356)/G i␣2 , 2.1. FIG. 3. Expression of G s␣ , constitutively active mutants of G s␣ blocks induction of adipogenesis: analysis by staining of lipid with oil red O. Wild-type 3T3-L1 cells (NULL) and clones transfected with empty vector (VECTOR), as well as those transfectants stably expressing G s␣ (␣s) and G s␣ Q227L transfectants (␣sQ227L) were plated on coverslips and propagated in 24-well culture plates. At confluence (day 0), one set of cells were treated with dexamethasone and methylisobutylxanthine (ϩINDUCERS). DEXϩMIX were removed after incubation for 2 days, and the cells were maintained in DMEM containing 10% fetal bovine serum for 7 days. A second replicate set of cells was maintained in DMEM containing 10% fetal bovine serum in the absence of DEXϩMIX. At day 7, cells were fixed by 3% paraformaldehyde for 5 min and stained with oil red O for 10 min. Hematoxylin (1%) was used to stain nuclei. Adipogenesis was examined under a Zeiss Axiophot microscope. The darkly stained bodies of the cytosol are oil droplets. Bar, 100 m.

FIG. 4. Expression of G s␣ and the Gi␣2(122)/s but not the Gi␣2(212)/s, chimeras blocks induction of adipogenesis: analysis by oil red O staining.
Stably transfected clones expressing G s␣ , G s␣ / G i␣2 chimeras, and G i␣2 were plated on coverslips and propagated in 24-well culture plates. At confluence (day 0), one set of cells were treated with dexamethasone and methylisobutylxanthine (ϩINDUC-ERS). DEXϩMIX were removed after incubation for 2 days, and the cells were maintained in DMEM containing 10% fetal bovine serum for 7 days. A second replicate set of cells was maintained in DMEM containing 10% fetal bovine serum in the absence of DEXϩMIX. At day 7, cells were fixed by 3% paraformaldehyde for 5 min and stained with oil red O for 10 min. Hematoxylin (1%) was used to stain nuclei. Adipogenesis was examined under a Zeiss Axiophot microscope. The darkly stained bodies of the cytosol are oil droplets. Bar, 100 m.
pression of the G i␣2 (212)/G s␣ and G s␣ (356)/G i␣2 chimeras, in sharp contrast, removed the block of DEXϩMIX-induced adipogenesis caused by G s␣ , as depicted by the dramatic increase in aP2 mRNA.
The extent of differentiation was quantified by scoring either positive or negative for lipid accumulation individual cells in large fields from cultures treated with DEXϩMIX and stained with oil red O (Fig. 7). The results show dramatically that constitutive expression of G s␣ , its activated mutants, or the C-terminal truncate effectively represses the induction of adipogenesis by DEXϩMIX. Expression of G i␣2 or its GTPasedeficient, activated mutant, in contrast, provoked adipogenesis, even in the absence of DEXϩMIX (Fig. 4). The ability of G s␣ to repress adipogenesis is not diminished by substitution of its N-terminal 145 residues with the analogous 122 N-terminal region of G i␣2 . Further extension of the substitution from 145 to residue 235 of G s␣ with G i␣2 largely relieves the block of adipogenesis caused by G s␣ , identifying the intervening region as a primary repressor of adipogenesis. The C-terminal truncate (G s␣ (1-356)) represses adipogenesis as well as full-length G s␣ . Addition of the C-terminal region of G i␣2 to the G s␣ truncate resulted in a modest derepression of induction of adipogenesis by DEXϩMIX.
To define further the nature of the repressor domain of G s␣ for adipogenesis, additional chimeras were constructed and stably expressed in the 3T3-L1 cells (Fig. 8A). Because expression of G s␣ (1-394), G s␣ (235)/G i␣2 , and G i␣2 (122)/G s␣ chimeras blocks adipogenesis in response to inducers, but G i␣2 (212)/G s␣ does not, we explored whether the 145-235 region of G s␣ embedded within G i␣2 would repress or permit induction of adipogenesis. As shown by oil red O staining of the clones (Fig.  8B), expression of the G i␣2 (122)/G s␣ (235)/G i␣2 chimera suppresses the adipogenic response to DEXϩMIX. Expression of the G i␣2 (212)/G s␣ chimera devoid of the 145-235 region permits a robust adipogenic response, as made more clear by color (Fig.  8B) compared with black-and-white (Fig. 4) images of the oil red O staining. The fact that expression of the G s␣ (235)/G i␣2 chimera fully represses DEXϩMIX-induced adipogenesis suggests no simple interpretation of the inability of the G s␣ (356)/ G i␣2 to block adipogenesis completely (Fig. 7). DISCUSSION G-proteins now are recognized as critical elements in a variety of complex biological processes. G-proteins have been shown to be key regulators of oncogenesis (Pace, et al., 1991;Voyno-Yasenetskaya et al., 1994;Wong et al., 1995), neonatal development (Moxham et al., 1993), early mouse development , and cellular differentiation (Strittmatter et al., 1990(Strittmatter et al., , 1994Wang et al., 1992;Su et al., 1993). Activating mutants of G s␣ and G i␣2 are associated with a number of cancers arising in endocrine tissues such as the thyroid, pituitary, and ovary (Landis et al., 1989;Lyons et al., 1990). In neonatal development, targeted elimination of G i␣2 in liver and adipose tissue at birth generates transgenic mice with a runted phenotype (Moxham et al., 1993). In early mouse development, probed in totipotent mouse F9 teratocarcinoma stem cells, the morphogen retinoic acid induces a sharp reduction in G i␣2 as the cells commit to primitive endoderm (Galvin-Parton et al.,  1990). Suppression of G i␣2 by constitutive expression of RNA FIG. 5. Role of the C-terminal domain of G s␣ in repressing adipogenesis: analysis by oil red O staining. Stably transfected clones expressing ␣s, ␣s truncate (TRUN), and ␣s(356)/i chimeras were plated on coverslips and propagated in 24-well culture plates. At confluence (day 0), one set of cells were treated with dexamethasone and methylisobutylxanthine (ϩINDUCERS). DEXϩMIX were removed after incubation for 2 days, and the cells were maintained in DMEM containing 10% fetal bovine serum for 7 days. A second replicate set of cells were maintained in DMEM containing 10% fetal bovine serum in the absence of DEXϩMIX. At day 7, cells were fixed by 3% paraformaldehyde for 5 min and stained with oil red O for 10 min. Hematoxylin (1%) was used to stain nuclei. Adipogenesis was examined under a Zeiss Axiophot microscope. The dark stained bodies of the cytosol are oil droplets. Bar, 100 m.
FIG. 6. Expression of G s␣ , constitutively active mutants of G s␣ , and the G i␣2 (122)/s but not the G i␣2 (212)/s chimeras blocks induction of early differentiation marker gene aP2: analysis by Northern blot hybridization. RNA (30 g/lane) was analyzed on Northern blots probed with 32 P-labeled aP2 cDNA. A, RNA was extracted from 3T3-L1 clones treated with DEXϩMIX from day 0 to day 7. B, RNA was extracted from 3T3-L1 wild-type cells without (Ϫ) or with (ϩ) DEXϩMIX (D/M) treatment and from clones stably transfected and expressing the indicated G-protein ␣ subunit also treated with DEXϩMIX.
antisense to G i␣2 mRNA mimics the decline observed in response to retinoic acid and induces differentiation to primitive endoderm in the absence of the morphogen (Gao and Malbon, 1996).
3T3-L1 embryonic fibroblasts are a useful model of cellular differentiation, in which inducers such as dexamethasone in combination with methylisobutylxanthine promote a highly differentiated state in which the cells accumulate lipid. The adipocyte phenotype requires the activation of an array of genes necessary for lipid synthesis. The capacity to stain the cells for lipid with oil red O provides a facile determination of differentiation. Earlier it was shown that cholera toxin activation of G s␣ effectively blocks induced differentiation and that G s␣ levels decline precipitously during induced differentiation (Wang et al., 1992). Oligodeoxynucleotides antisense, but not sense or missense, to G s␣ mimicked the decline in G s␣ and the adipogenic conversion of the cultures (Wang et al., 1992). Exploitation of the antagonistic G s␣ /G i␣2 axis shown for adenylyl cyclase revealed that overexpression of G i␣2 as well as the constitutively active Q205L mutant of G i␣2 would also induce differentiation in the absence of dexamethasone and methylisobutylxanthine (Su et al., 1993). Although these data would provide support for an intimate role of adenylyl cyclase and cAMP in this process, this is not the case. Elevation of intracellular cAMP levels by pertussis toxin, forskolin, or addition of the dibutyryl or chlorophenylthio derivatives of cAMP fail to affect adipogenesis (Wang et al., 1992). These data implicate some effector other than adenylyl cyclase in the control of adipogenesis (Wang and Malbon, 1996).
The G s␣ /G i␣2 axis controlling adipogenesis was exploited in the current work to identify regions of the G s␣ molecule that are involved in this regulation. Expression of wild-type G s␣ effectively blocks the induction of differentiation. Constitutively active mutants of G s␣ G225T and Q227L both blocked the ability of dexamethasone and methylisobutylxanthine to induce differentiation. Thus, constitutively active mutants of G s␣ block and those of G i␣2 induce differentiation. Mutations of this nature have been associated with tumorigenesis of several endocrine tissues . Substitution of sequences of the N terminus of G s␣ with analogous ones of G i␣2 proved revealing with respect to the domain(s) involved in controlling adipogenesis. Whereas substitution of the N-terminal 145 amino acids of G s␣ with the first 122 amino acids of G i␣2 produced no obvious effects on the ability of chimera to suppress differentiation, substitution of the N-terminal 235 residues of G s␣ with the first 212 residues of G i␣2 had a profound effect. Expression of the G i␣2 (212)/G s␣ construct abolishes the ability of the chimera to suppress induction of adipogenesis in response to DEXϩMIX, whereas expression of a chimera in which the G s␣ 146 -235 region is embedded in G i␣2 effectively blocks DEXϩMIX-induced adipogenesis. These data clearly demarcate region 146 -235 of G s␣ as critical to its ability to repress adipogenesis.
Careful comparison of the aligned protein sequences of G s␣ and G i␣2 reveals the control domain to be restricted to region 146 -220, as the 221-235 region of G s␣ includes only one conservative substitution (R231K) and one nonconservative substitution (D229S). Analysis by mutagenesis revealed that this (D229S) substitution is not critical to repressor activity (not shown). Analysis of the C-terminal region of G s␣ was approached using truncation of the final 38 residues, which in this model of differentiation had equal capability as full-length G s␣ or constitutively activated G s␣ mutants to suppress 3T3-L1 cell differentiation. Interestingly, the substitution of the Cterminal 36 residues of G i␣2 for the corresponding 38 residues of G s␣ , chimera G s␣ (356)/G i␣2 , resulted in a small, but significant, release from the repression observed with constitutive expression of native G s␣ or the truncated 1-356 version. The ability of the G s␣ (235)/G i␣2 chimera to fully repress the adipogenic response suggests that the effects of the G s␣ (356)/G i␣2 chimera are not simply related to substitution of the G i␣2 sequence into the C terminus of G s␣ .
With structural features of the ␣ subunits of several heterotrimeric G-proteins deduced from x-ray diffraction, it is now possible to speculate as to the regions of G s␣ involved in effector regulation (Lambright et al., 1994). Projections of G s␣ and G i␣2 sequences on the deduced structure of the GTP-liganded form of G i␣1 (Coleman et al., 1994) (red) were used to identify the regions implicated in the control of adenylyl cyclase (yellow) in the space-filling model displayed in Fig. 9. The GTP molecule is rendered in white. Highlighted in blue is the region of G i␣2 (122-212), which is corresponding to the amino acid sequence 145-235 of G s␣ , shown to be critical to the ability of the chimera to repress adipogenesis. Interestingly, this region of G s␣ includes switch domains I and II (green), but not III (cyan), which change conformation on activation of the molecule by GTP binding. The colorization of the sequences displays not only the overlap of the control domain with switches I and II, but also the spatial separation of the domain from that implicated in adenylyl cyclase regulation.
It is the loss of the controlling domain of G s␣ or the addition of the analogous region of G i␣2 that is responsible for the effects on adipogenesis? The fact that expression of native or consti- FIG. 7. Expression of G s␣ , constitutively active mutants of G s␣ , and the G i␣2 (122)/s but not the G i␣2 (212)/s chimeras blocks induction of adipogenesis: quantitation. Wild-type 3T3-L1 cells (Null) and clones transfected with empty vector (Vector alone), as well as those transfectants stably expressing various ␣ subunits were plated on coverslips and propagated in 24-well culture plates. At confluence (day 0), one set of cells was treated with dexamethasone and methylisobutylxanthine (ϩ D/M). DEXϩMIX were removed after incubation for 2 days, and the cells were maintained in DMEM containing 10% fetal bovine serum for 7 days. A second replicate set of cells was maintained in DMEM containing 10% fetal bovine serum in the absence of DEXϩMIX. At day 7, cells were fixed by 3% paraformaldehyde for 5 min and stained with oil red O for 10 min. Hematoxylin (1%) was used to stain nuclei. Adipogenesis was examined under a Zeiss Axiophot microscope. Fields of 100 -200 cells were scored either positive or negative for lipid accumulation. Percentages are mean values from at least six individual experiments, each with independent clones quantified from multiple fields. Bars, S.E.
FIG. 8. Expression of G s␣ , G i␣2 (122)/ G s␣ , and the G i␣2 (122)/G s␣ (235)/G i␣2 but not the G i␣2 (212)/G s␣ chimeras blocks induction of adipogenesis: staining of lipid accumulation. Cloning, induction of adipogenesis, and staining of lipid accumulation with oil red O were performed as described in the legend to Fig. 4. A, constructs analyzed. B, oil red O staining of clones. ‫,ء‬ adipogenesis in response to DEXϩMIX. tutively active G i␣2 induces differentiation in the absence of dexamethasone and methylisobutylxanthine (Su et al., 1993) and the inability of the G i␣2 (212)/s chimera to alter differentiation in the absence of the inducers, suggests the former rather than the latter interpretation. Thus, we have identified a domain of G s␣ requisite for regulation of adipogenesis as separate and distinct from those controlling adenylyl cyclase, consistent with the inability of cAMP to influence this differentiation process (Wang and Malbon, 1996). Identification of the region controlling adipogenesis provides the boundaries for more detailed mutagenesis and a basis from which to initiate the search for the effector(s) through which G s␣ exerts its control of adipogenesis.