Abstract
The brown fat uncoupling protein-1 (ucp-1) gene is regulated by the sympathetic nervous system, and its transcription is stimulated by norepinephrine, mainly through cAMP-mediated pathways. Overexpression of the catalytic subunit of protein kinase A stimulated a chloramphenicol acetyltransferase expression vector driven by the 4.5-kb 5′-region of the rat ucp-1 gene. Mutant deletion analysis indicated the presence of the main cAMP-regulatory element (CRE) in the proximal region between −141 and −54. This region contains an element at −139/−122 able to confer enhancer and protein kinase A (PKA)-dependent activity to the basal thymidine kinase promoter. The potency of this element was much higher in differentiated than in nondifferentiated brown adipocytes. Gel shift analyses indicated that a complex array of proteins from brown fat nuclei bind to the −139/−122 element, among which CRE-binding protein (CREB) and Jun proteins were identified. In transfected brown adipocytes, c-Jun was a negative regulator of basal and PKA-induced transcription from the ucp-1 promoter acting through this proximal CRE region. A double-point mutation in the− 139/−122 element abolished both PKA- and c-Jun-dependent regulation through this site, and overexpression of CREB blocked c-Jun repression. Thus, an opposite action of these two transcription factors on the− 139/−122 CRE is proposed. c-Jun content in brown adipocytes differentiating in culture correlated negatively with both ucp-1 gene expression and the acquisition of the brown adipocyte morphology. These findings indicate that c-Jun provides a molecular mechanism to repress the basal and cAMP-mediated expression of the ucp-1 gene before the differentiation of the brown adipocyte.
INTRODUCTION
Brown adipose tissue is a major site for nonshivering thermogenesis in mammals. Its thermogenic function relies on a mitochondrial proton conductance pathway due to the presence of the uncoupling protein-1 (UCP-1), which releases as heat the energy from substrate oxidation normally conserved in the form of ATP (1). Transcription of the ucp-1 gene occurs only in the brown fat cell, and it is under a complex regulation by hormonal signals and during cell differentiation (2–6). The 5′-noncoding regions of the ucp-1 genes from rat and mouse contain most of the elements for transcriptional regulation assembled in two main regions. There is an upstream enhancer, involved in multihormonal regulation by retinoic acid, thyroid hormones, and agonists of peroxisome proliferator-activated receptor-γ (PPARγ) (6–9), as well as a proximal regulatory region including a silencer, CAAT/enhancer binding protein (C/EBP)-regulated sites, and basal promoter elements (3, 5). Although potential elements for differentiation-dependent ucp-1 gene expression, such as C/EBP (5) or PPARγ (9), have been reported, the regulatory elements for tissue-specific ucp-1 gene expression have not been identified. Studies performed so far using transgenic mice show brown fat-specific gene expression when both the enhancer and proximal regulatory regions are present in a ucp-1 construct transgene (7).
Control of brown fat thermogenesis in response to heat demands depends upon the release of norepinephrine from sympathetic terminals innervating the tissue. Norepinephrine activates transcription from the ucp-1 gene, and cAMP has been proposed as the main intracellular mediator of this action (2, 3). Furthermore, chronic cAMP-dependent protein kinase A (PKA) overactivity occurring in the brown fat of mice carrying a targeted disruption of the RIIβ subunit of PKA causes an enhanced expression of the ucp-1 gene (10). Attempts to define the regulatory sites responsible for norepinephrine stimulus of transcription in the mouse ucp-1 gene have yielded a complex pattern in which multiple putative cAMP-regulatory elements (CREs), widespread in the enhancer and proximal regions, appear to be involved (4). Neither the specific role of these CREs nor the trans-acting factors involved in the norepinephrine stimulus of ucp-1 gene transcription have yet been determined. On the other hand, there is a complex interaction between the transcriptional regulation of the ucp-1 gene due to brown adipocyte differentiation and in response to norepinephrine. Brown adipocyte differentiation is associated with a rise in basal ucp-1 gene expression and an increase in its responsiveness to norepinephrine action (11). Although changes in the abundance ofβ -adrenergic receptor subtype occur in association with the differentiation of the brown fat cell, intracellular cAMP generation in response to norepinephrine is similar before and after brown adipocyte is terminally differentiated (12, 13). Then, intracellular mechanisms must act in association with brown adipocyte differentiation to provide both high basal expression and full responsiveness of the ucp-1 gene to cAMP.
In recent years, a substantial increase in our understanding of the mechanisms of cAMP stimulation of mammalian gene transcription has been achieved. Initially, CREs were identified in several gene promoters and generally consist in small variations of a palindromic sequence TGACGTCA (14, 15). Subsequently, proteins that bound to these CREs were identified, and a whole family of transcription factors, the CREB/activating transcription factor (ATF) family, is now known to mediate most of the cAMP responsiveness of different genes. Several of these proteins are activated by PKA-dependent phosphorylation (16, 17). These CREB/ATF proteins are members of the larger basic-leucine zipper (bZIP) family of transcription factors (18). Other proteins from the bZIP family such as c-Jun and c-Fos or the C/EBP proteins, despite being known to mediate distinct biological signals in relation to cell proliferation or differentiation, also participate in the cAMP responsiveness of transcription of several genes (19–21). In fact, members of the bZIP family of transcription factors act as dimers, and a complex array of heterodimers may be formed between CREB/ATF proteins and other bZIP proteins (18).
c-Jun is a transcription factor that has a pivotal role in the mediation of signal transduction pathways by a wide variety of stimuli. In general, c-Jun promotes progression of the cell cycle (22) and mediates the proliferative response to growth factors in multiple cell types (23). c-Jun and c-Fos, the products of the c-jun and c-fos protooncogenes, interact with specific DNA sequences to modulate gene transcription. c-Jun/c-Jun homodimers and c-Jun/c-Fos heterodimers often act through a sequence element, named AP-1 site, that mediates transcriptional response to multiple signal transduction pathways (for review see Ref. 24). In addition, c-Jun is able to interact with other proteins, forming heterodimers with members of the CREB/ATF family such as ATF-2, ATF-3, and ATF-4 (25, 26). c-Jun homo- and heterodimers have been reported to interact with CREs, thereby affecting positively or negatively the cAMP responsiveness of several gene promoters (20, 27, 28).
The aim of this study was to establish the main regulatory elements and transcription factors that determine the cAMP responsiveness of the rat ucp-1 gene. The main CRE was recognized in the proximal region of the rat ucp-1 promoter (at −139/−122), and its activity was shown to be dependent upon the stage of brown adipocyte differentiation. CREB and Jun proteins from brown fat nuclei were identified as binding UCP-CRE. c-Jun was found to be a powerful repressor of the basal and PKA-induced transcription of the ucp-1 gene, acting through this CRE. Furthermore, c-Jun expression was found to correlate negatively with ucp-1 gene expression in differentiating brown fat cells. We propose a dominant negative role for c-Jun in the molecular mechanisms that mediate adrenergic responsiveness of ucp-1 gene expression linked to brown fat differentiation.
RESULTS
The Rat ucp-1 Gene Is Activated by PKA due to a Major Responsive Site in the Proximal 5′-Noncoding Region
Primary cultures of differentiated brown adipocytes were transfected with (−4551)UCP-CAT and then incubated with norepinephrine or cAMP effectors before harvesting and analysis of chloramphenicol acetyl transferase (CAT) activity (see Fig. 1A). Norepinephrine (0.1 μm) caused an almost 4-fold increase in the expression of (−4551)UCP-CAT. Forskolin (10 μm), a direct activator of the adenylate cyclase catalytic subunit, also elicited a marked stimulation. 3-Isobutyl-1-methylxanthine (IBMX) (0.5 mm), a phosphodiesterase inhibitor, had a similar effect. Furthermore, 1 mm 8-bromo-cAMP, a nonmetabolizable cAMP derivative, caused a 3-fold rise of the transfected (−4551)UCP-CAT activity. To determine whether PKA, whose cAMP-mediated activation is elicited by all these agents, is a major effector of (−4551)UCP-CAT activity, brown adipocytes were cotransfected with SRα-PKA, a vector driving the expression of the catalytic subunit of PKA. This resulted in a significant (P < 0.001) increase in (−4551)UCP-CAT expression similar to that achieved by the previously tested agents. These results indicate that the 4.5-kb 5′-noncoding region of the rat ucp-1 gene contains PKA-responsive elements.
Effects of Norepinephrine, cAMP Effectors, and Overexpression of the Catalytic Subunit of PKA on the Expression of (−4551)UCP-CAT. Deletion Mutant Analysis of the PKA Effect A, Primary brown adipocytes differentiated in culture were transfected with 15 μg/plate of the (−4551)UCP-CAT plasmid and exposed or not to 0.1 μm norepinephrine (NE), 10 μm forskolin, 0.5 mm IBMX, or 1 mm 8-Br-cAMP. Other plates included in the transfection 3 μg of the SRα-PKA, the expression vector for the catalytic subunit of PKA. Results are expressed as -fold induction relative to untreated cells. B, Brown adipocytes were transiently transfected with 15 μg/plate of (−4551)UCP-CAT or equivalent amounts of the deletion mutants illustrated on the left. Transfections included (PKA) or not (Basal) 3 μg/plate of the SRα-PKA expression vector. Results are expressed as the relative CAT activity with respect to the basal value for (−4551)UCP-CAT, which is set to 1. Bars are means of three independent experiments, each performed in triplicate. sems did not exceed 10% of the means.
To determine the sites in the 5′-region of the rat ucp-1 gene responsible for this PKA-mediated stimulation, the effect of the transfected plasmid SRα-PKA was determined on deletion mutants of the (−4551)UCP-CAT transiently transfected into primary cultures of brown adipocytes. The study of mutants with progressively longer 5′-deletions of (−4551)UCP-CAT indicated that PKA responsiveness was lost only when the proximal region between −141 and −54 was suppressed (Fig. 1B). The presence of sequences between −2494 and −172 in the absence of the −141/−54 region was also able to support a 2-fold responsiveness to PKA, markedly lower than that elicited when this proximal region was present. These data indicate the presence of multiple PKA responsive elements in the rat ucp-1 gene but also the greater response due to the proximal −141/−54 region. On the other hand, all the constructs in which the −141/−54 region was deleted showed lowered basal expression with respect to others in which this region was present, even though the activity of those constructs was always significantly (P < 0.05) higher than the background activities elicited by a transfected promoterless CAT vector (not shown).
The −139/−122 Element of the Rat ucp-1 Gene Has Enhancer Activity and Confers PKA Responsiveness to a Heterologous Promoter
Our previous analysis on the DNA protein-binding domains in the proximal region of the rat ucp-1 gene identified a major DNaseI footprint site protected by rat brown fat nuclear proteins at− 139/−122, within the PKA-responsive region (29). A similar observation has been reported using nuclear extracts from Syrian hamster brown adipose tissue (30). To test the ability of the− 139/−122 sequence from the ucp-1 gene promoter to function as PKA-response element, a series of heterologous CAT vectors were generated by ligating one, two, or three copies of this sequence (UCP-CRE oligonucleotide) to the herpes simplex virus (HSV) thymidine kinase promoter. As shown in Fig. 2, one copy of the UCP-CRE confers a 4-fold induction by PKA to the unresponsive tk-CAT gene. Addition of more copies of this UCP-CRE resulted in a higher responsiveness to PKA (a 12-fold and a 32-fold induction for the (UCP-CRE)2-tk-CAT and (UCP-CRE)3-tk-CAT, respectively). Furthermore, basal activity of these last plasmids indicates that UCP-CRE also shows enhancer properties in differentiated brown adipocytes.
Basal and PKA-Induced Expression of Chimeric Constructs Containing Multiple Copies of the −139/−122 Element Upstream from the Basal Thymidine Kinase Promoter Primary brown adipocytes differentiated in culture were transfected with 5 μg/plate of the chimeric CAT constructs depicted on the left with (PKA) or without (Basal) 3 μg/plate of SRα-PKA. The UCP-CRE oligonucleotide corresponds to the −139/−122 sequence of the rat ucp-1 gene (see Materials and Methods). Results are expressed as the fold-induction of activity with respect to the basal activity of the thymidine kinase (tk)-CAT construct, which is set to 1. Bars are means of two independent experiments, each performed in triplicate.
The −139/−122 Element in the Rat ucp-1 Gene Binds CREB and c-Jun Proteins from Brown Adipose Tissue Nuclei
Electrophoretic mobility shift assays were undertaken using the UCP-CRE oligonucleotide as a labeled probe and protein extracts from rat interscapular brown fat nuclei. A complex pattern of DNA-protein binding complexes was formed as revealed by the presence of at least four retarded bands (named A, B1, B2, and C) (Fig. 3A). The appearance of B1 and B2 as two separate bands was particularly evident in long time runs of the electrophoreses (see Fig. 3, B, C, and D). Excess of the unlabeled UCP-CRE probe suppressed the appearance of the C and B bands but not the A band, which was therefore considered as nonspecific. The C and B bands also disappeared when incubations included as competitor an excess of an oligonucleotide corresponding to a well characterized CRE such as the CRE-I (−94/−77) in the rat phosphoenolpyruvate carboxykinase (PEPCK) gene (see Fig. 3A). A similar result was obtained using the rat somatostatin gene CRE (S-CRE) oligonucleotide as competitor (not shown). An excess of an unrelated oligonucleotide (the GA-rich sequence from the stromelysin ETS-binding site, see Materials and Methods), used as negative control, was without effect. This indicated that CRE-binding proteins were the main proteins binding specifically to the −139/−122 element in the rat ucp-1 gene. To further assess this, the UCP-CRE-labeled probe was incubated with partially purified recombinant CREB-1 (CREB), a representative member of the CREB/ATF family of transcription factors mediating cAMP stimulation of gene transcription (see Fig. 3B). This resulted in the appearance of a single retarded band in the gel-shift assay. The specificity of binding was shown by the efficient competition of an excess of either the rat PEPCK-CRE-I or rat S-CRE. The mobility of the complex formed by CREB was identical to the C band formed by the brown fat nuclear extract. The identity of CREB was confirmed by incubating the brown fat nuclear extract and the UCP-CRE probe with an antibody against CREB (not shown).
Electrophoretic Mobility Shift Assays of the Nuclear Proteins that Interact with the −139/−122 Element in the Rat ucp-1 Gene A double-stranded oligonucleotide corresponding to the −139/−122 region of the rat ucp-1 gene (UCP-CRE) was used as labeled probe. A, The probe was incubated with 5 μg of nuclear protein extracts from rat brown adipose tissue (BAT), and competitors were added at a 100-fold molar excess relative to probe concentration. PEPCK-CRE-I is an oligonucleotide corresponding to the −94/−77 site in the rat PEPCK gene, and ETS is a GA-rich oligonucleotide used as negative control (see Materials and Methods). Arrows indicate the mobilities of the major protein-DNA complexes formed. B, Gel-mobility shift assay of the interaction of the− 139/−122 UCP-CRE probe with CREB. The labeled probe was incubated with 1 μg of partially purified E. coli-expressed CREB as described (45 ). Competitors were PEPCK-CRE-I as in panel A or S-CRE, an oligonucleotide corresponding to the −60/−29 site in the rat somatostatin gene. C, Gel-mobility shift assays of the interaction of the −139/−122 UCP-CRE probe with c-Jun. The labeled probe was incubated with 1 fpu (footprint protection unit) of purified recombinant c-Jun. Competitors were as in panel B. D, Effects of anti-Jun serum on the protein-DNA complexes formed between brown fat nuclear proteins (BAT) and the −139/−122 UCP-CRE probe. Brown fat nuclear protein extract (5 μg) was incubated with 0.2 μl or 1 μl of anti-Jun (56 ) or preimmune (control) serum before incubation with the labeled probe.
To investigate whether other bZIP transcription factors not belonging to the CREB/ATF family but potentially related to cAMP-signaling pathways could bind the −139/−122 UCP-CRE, the labeled probe was incubated with purified C/EBPβ (21, 31). Results indicated that the UCP-CRE was unable to bind C/EBPβ efficiently (not shown). The amount of C/EBPβ used in the assays in which binding did not occur was enough to induce at least a 30% retardation of labeled probes such as the −457/−440 and −335/−318 C/EBP binding sites in the ucp-1 gene (5) or the −94/−77 PEPCK-CRE-I (21). Further experiments were performed to assess whether c-Jun proteins could bind to the −139/−122 UCP-CRE. Incubation of the −139/−122 probe with recombinant purified c-Jun showed significant binding, indicated by the appearance of a retarded band with similar mobility to the B1 band (see Fig. 3C). Binding was specific as shown by the competition of an excess of PEPCK-CRE-I, which is known to bind c-Jun homodimers (20), but not by competition with the negative control ETS. When brown fat nuclear extracts were incubated with antibodies specific to Jun proteins, both B bands disappeared specifically. A nonspecific increase in the intensity of retardation at levels close to the mobility of the nonspecific A band was observed in the presence of either the anti-Jun serum or the preimmune (control) serum (see Fig. 3D). This result indicates that Jun proteins are components of the protein complexes binding the UCP-CRE and originating the B bands.
c-Jun, but not c-Fos, Represses Basal and PKA-Induced ucp-1 Gene Promoter Activity
To examine the functional significance of the binding of c-Jun to the main proximal CRE in the ucp-1 gene, transient cotransfection analysis of the CAT-driven ucp-1 promoter was undertaken. However, as recently pointed out (32), plasmid constructs derived from pUC contain an artifactual AP-1 binding site that hampers their use for studying c-Jun or c-Fos action. As this was the case for the (−4551)UCP-CAT and derived mutant deletion series, new constructs were obtained by deleting this AP-1 site in the former (−2494)UCP-CAT (see Materials and Methods for details). The construct in which the −2494/+110 fragment of the ucp-1 gene drives CAT expression and has the AP-1 site deleted showed a similar response to PKA (>6-fold stimulation) to that of the former version (Fig. 4). Cotransfection into primary brown adipocytes of 3 μg of an expression vector in which the entire open reading frame of c-Jun is transcribed from the cytomegalovirus (CMV) promoter significantly (P < 0.01) diminishedΔ AP1(−2494)UCP-CAT expression. Moreover, the ability of PKA to stimulate ΔAP1(−2494)UCP-CAT was almost completely suppressed by CMV-c-Jun cotransfection (Fig. 4). Lower amounts of CMV-c-Jun resulted in a weaker inhibitory effect and a 10-fold lower amount of cotransfected expression vector (0.3 μg) reduced by only 20% the basal and PKA-induced activity of ΔAP1(−2494)UCP-CAT (not shown). Parallel experiments using the CMV-c-Fos vector showed no effect on the basal expression of the ucp-1 promoter, and this vector also failed to suppress the PKA-induced expression of the ucp-1 promoter. The effects of an equivalent mixture of transfected CMV-c-Jun and CMV-c-Fos were essentially indistinguishable from the action of the corresponding amount (one half) of the single CMV-c-Jun expression vector.
Effects of c-Jun, c-Fos, and Mutations of c-Jun on the Basal and PKA-Induced Expression of (−2494)UCP-CAT Primary brown adipocytes differentiated in culture were transfected with 10 μg/plate of the ΔAP1(−2494)UCP-CAT. Transfections included (PKA) or not (Basal) 3 μg of SRα-PKA. Jun and Fos are CMV-driven expression vectors for the rat cDNAs of c-Jun and c-Fos, respectively. Jun ΔL3 is a CMV-driven expression vector for a mutant form of c-Jun that contains leucine-to- valine substitution at the leucine 3 within the leucine zipper domain. Jun ΔBR is a CMV-driven mutant form of c-Jun lacking amino acids 260–266, within the DNA-binding domain of c-Jun (50 ). Each expression vector (3 μg/plate) for the wild-type or mutant forms of c-Jun and c-Fos was included in the transfections except in the Jun + Fos experiments, in which 1.5 μg/plate of each expression vector were cotransfected. Results are shown as relative to the basal expression of ΔAP1(−2494)UCP-CAT, which is set to 1. Bars are means of three independent transfection experiments, each performed in triplicate. sems did not exceed 10% of the means.
The domains of c-Jun required for repression of the ucp-1 gene transcription were examined using mutant forms of rat c-Jun cDNA with mutations in known functional domains. The mutant JunΔL3 encodes a c-Jun protein defective in dimerization due to a leucine-to-valine substitution at leucine 3 within the leucine zipper domain. The JunΔ BR is a deletion mutant lacking amino acids 260–266 within the DNA-binding domain (see Materials and Methods). Both mutations significantly (P < 0.05) attenuated the ability of CMV-c-Jun to inhibit basal and PKA-induced expression of the ucp-1 promoter expression (Fig. 4).
The Proximal CRE Is Required for the Inhibitory Effect of c-Jun on the Rat ucp-1 Gene Promoter
Deletion mutants from ΔAP1(−2494)UCP-CAT were obtained and transiently transfected into brown adipocytes differentiated in culture to assess whether the proximal CRE region that binds Jun proteins was responsible for the repressing effect of c-Jun on the ucp-1 gene promoter. Deletion of most of the 5′-noncoding region of the ucp-1 gene did not affect the ability of c-Jun to inhibit basal and PKA-induced expression of the ucp-1 gene promoter (Fig. 5). The presence of 141 bp upstream from the transcription start site was enough to retain the inhibition by c-Jun of the ucp-1 gene promoter expression. Conversely, an internal deletion in which only the −172/−54 region ofΔ AP1(−2494)UCP-CAT had been eliminated was enough to suppress any inhibitory action of c-Jun, despite a previous report on an AP-1 binding site at −2422 (7). Neither the basal nor the PKA-stimulated expression present in this construct caused by sequences upstream from− 172 was affected by c-Jun expression. The construct in which the whole region upstream from −54 had been deleted was insensitive to the inhibitory action of c-Jun. These results indicate that, indeed, the− 141/−54 region, in which the Jun binding site is present, is required for the c-Jun action inhibiting ucp-1 gene promoter expression.
Deletion Mutant Analysis of the Inhibitory Action of c-Jun on the (−2494)UCP-CAT Primary brown adipocytes differentiated in culture were transfected with 10 μg/plate of ΔAP1(−2494)UCP-CAT or equivalent amounts of the deletion mutant derivatives depicted in the figure. Transfections included (+) or not (−) 3 μg of the SRα-PKA expression vector (PKA) and/or 3 μg of the CMV-c-Jun expression vector (Jun). Results are shown as relative to the basal expression of ΔAP1(−2494)UCP-CAT, which is set to 1. Bars are means of three independent transfection experiments, each performed in triplicate. sems did not exceed 10% of the means.
To determine whether the isolated −139/−122 UCP-CRE can confer c-Jun-dependent repression, the (UCP-CRE)2-tk-CAT plasmid was transfected into differentiated brown adipocytes (see Fig. 6C). Both basal and PKA-stimulated expression of this heterologous construct were significantly (P < 0.01) blocked by c-Jun expression. In contrast, two copies of a double-point mutant version of the UCP-CRE (mutUCP-CRE, see Fig. 6A) were unable to confer either PKA- or c-Jun-dependent responsiveness to the neutral thymidine kinase promoter (Fig. 6C). These results are consistent with the lack of capacity of mutUCP-CRE to bind either CREB or Jun proteins present in protein extracts from brown fat nuclei as shown in Fig. 6B. The only protein-(mutUCP-CRE) binding complex formed was the nonspecific A band. When used as unlabeled competitor, the mutUCP-CRE oligonucleotide was also unable to compete for the B and C bands formed with UCP-CRE as labeled probe (not shown).
Expression of Chimeric Constructs Containing Either the −139/−122 UCP-CRE or a Double Point Mutant Version Upstream from the Thymidine Kinase Promoter in Differentiated Brown Adipocytes. Effects of Cotransfection of c-Jun or CREB Expression Vectors upon Basal and PKA Responsiveness A, Sequence of the double-stranded oligonucleotides corresponding to the −139/−122 region of the rat ucp-1 gene (UCP-CRE) or to its double-point mutant derivative version (mutUCP-CRE) in which the two changed bases are shown in lowercase. B, Electrophoretic mobility shift assay. Wild-type and mutated forms of UCP-CRE were used as labeled probes, and an equal amount of each one was incubated with 5 μg of nuclear protein extracts from rat brown adipose tissue. Arrows indicate the mobilities of the major protein-DNA complexes formed. C, Brown adipocytes differentiated in culture were transiently transfected with 5 μg/plate of the (UCP-CRE)2-tk-CAT or the (mutUCP-CRE)2-tk-CAT plasmids, in which two copies of either the −139/−122 UCP-CRE or its mutant derivative were placed upstream from the thymidine kinase promoter in a modified (ΔAP1)pBLCAT2 vector (see Materials and Methods). Transfections included (+) or not (−) 3 μg of the SRα-PKA expression vector (PKA), and/or 3 μg of the CMV-c-Jun expression vector (Jun), and/or 3 μg of the RSV-CREB expression vector (CREB). Results are shown relative to the basal expression of (UCP-CRE)2-tk-CAT, which is set to 1. Bars are means of two to three independent transfection experiments, each performed in triplicate. sems did not exceed 10% of the means.
To further analyze the functional interaction between the bZIP proteins found to bind the UCP-CRE, the expression vectors for CREB and/or c-Jun were cotransfected with the (UCP-CRE)2-tk-CAT plasmid. As depicted in Fig. 6C, although transfection of Rous sarcoma virus (RSV)-CREB alone was without effect, when cotransfected with CMV-c-Jun, CREB was able to block repression by c-Jun of (UCP-CRE)2-tk-CAT activity. The ability of CREB to antagonize the inhibitory effect of c-Jun was even higher when the influence of both on the PKA-induced expression of the plasmid was analyzed. These results point to a functional competitive interaction of CREB and c-Jun upon the UCP-CRE.
Basal Enhancer and PKA Responsiveness Conferred by the −139/−122 UCP-CRE Depends on Brown Adipocyte Differentiation
Brown adipose tissue precursor cells were cultured for 7 days in conditions leading to differentiated brown adipocytes or in a hormone-depleted culture medium known to impair adipocyte differentiation (6). Only the former led to the appearance of the brown adipocyte morphology, characterized by rounding up of the cells and accumulation of lipid droplets (see Fig. 7A). High levels of UCP1 mRNA expression were also present as a phenotypic feature of the differentiated cells with respect to the nondifferentiated cells (Fig. 8B).
Expression and PKA Responsiveness of the Chimeric Construct (UCP-CRE)2-tk-CAT in Differentiated and Nondifferentiated Cultured Brown Adipocytes Primary brown adipocyte precursor cells were grown in culture for 4 days and treated thereafter with either regular differentiating medium or nondifferentiating medium, as described in Materials and Methods. A, Microphotographs of the cells on day 7 after being cultured in the different media. Magnification 40×. B, Differentiated or nondifferentiated brown adipocytes were transiently transfected with 5 μg/plate of the (UCP-CRE)2-tk-CAT including (+) or not (−) 3 μg of the SRα-PKA expression vector (PKA). Results are expressed as the fold-induction of activity with respect to the basal activity of the empty thymidine kinase-CAT vector (ΔAP1)pBLCAT2 in each cell type, which is set to 1. Bars are means of two independent experiments, each performed in triplicate.
Electrophoretic Mobility Shift Assay of the− 139/−122 Region of the ucp-1 Gene with Nuclear Protein Extracts from Differentiated or Nondifferentiated Cultured Brown Adipocytes. Northern and Western Blot Analyses of c-Jun Expression in Cultured Brown Adipocytes Cells were grown in culture as described in the legend of Fig. 7. A, The −139/−122 UCP-CRE probe was incubated with 5 μg of nuclear protein extracts from either differentiated or nondifferentiated brown adipocytes. Arrows indicate the mobilities of the major protein-DNA complexes formed. B, Northern blot analysis of 15 μg of total RNA from differentiated or nondifferentiated brown adipocytes. Northern blots were probed with the rat UCP1 and the rat c-Jun cDNA probes. The sizes of the detected transcripts are depicted on the right. C, Nuclear protein extracts (5 μg) from either differentiated or nondifferentiated brown adipocytes were analyzed by Western blot using specific antisera for CREB (Santa Cruz Biochemicals) or Jun (56 ) proteins. The sizes of the detected proteins are indicated on the right.
To analyze whether the differentiation state of brown adipocytes can affect either enhancer or PKA-responsive activity of the −139/−122 element of the ucp-1 gene promoter, we transfected (UCP-CRE)2-tk-CAT into either differentiated or nondifferentiated primary brown adipocytes. As depicted in Fig. 7B, UCP-CRE did not show enhancer activity in nondifferentiated cells (only 1.5-fold with respect to the empty vector (ΔAP1)pBLCAT2) in contrast to its ability to confer a 4-fold induction of activity to basal thymidine kinase in differentiated cells. The PKA-responsiveness conferred by the UCP-CRE showed dramatic differences when assessed in nondifferentiated (a 3-fold induction by PKA) as compared with differentiated cells (12-fold). Therefore, it is concluded that the activity of the UCP-CRE is dependent upon the stage of brown adipocyte differentiation.
Negative Correlation of c-Jun Abundance with Respect to the Expression of the ucp-1 Gene and the Differentiation of the Brown Adipocyte
To analyze whether endogenous c-Jun abundance could be involved in determining the differentiation-dependent activity of UCP-CRE, we performed gel-shift analysis using nuclear protein extracts from either differentiated or nondifferentiated brown adipocytes. As shown in Fig. 8A, the c-Jun-related B bands were more intense in the gel shift when extracts from nondifferentiated cells were tested while the C band predominated in extracts from differentiated brown adipocytes.
Likewise the expression of c-Jun was assessed in both differentiated and nondifferentiated cells. As shown in Fig. 8B, c-Jun mRNA showed a pattern of expression similar to that found in other murine cells, i.e. two mRNA species of 3.2 and 2.7 kb (33). c-Jun mRNA levels in nondifferentiated brown fat cells were 4-fold those in differentiated brown adipocytes. Accordingly, c-Jun protein content was higher (5-fold) in nondifferentiated than in differentiated brown adipocytes (Fig. 8C). In contrast, CREB abundance in differentiated brown adipocytes was 2-fold that in nondifferentiated cells. Thus, the changes in the relative abundance of c-Jun and CREB during the differentiation of brown fat cells in culture may account for the differentiation-dependent basal and PKA-inducible ucp-1 gene transcriptional activity found in these cells.
DISCUSSION
Norepinephrine has been classically recognized as the main inducer of brown fat thermogenesis; it acts mainly by stimulating expression of the ucp-1 gene at the transcriptional level (2, 3). The use of three different agents to increase the level of cAMP as well as cotransfection of a catalytic subunit of PKA expression vector in brown adipocytes resulted in a similar induction of transcription from the 4.5-kb ucp-1 gene promoter to that seen with norepinephrine. Thus, we show for the first time that the adrenergic regulation of the ucp-1 gene transcription via norepinephrine is mediated by cAMP stimulation of PKA. This is consistent with the enhanced expression of ucp-1 observed in the brown fat of transgenic mice with chronic PKA overactivity due to the targeted disruption of the RIIβ subunit of PKA (10). Furthermore, using a series of deletion mutants of the ucp-1 promoter, we have identified a major PKA-responsive region in the ucp-1 gene promoter located at− 141/−54. This region contains a CRE motif at −139/−122, able to bind CREB, that is sufficient to confer PKA responsiveness to a neutral promoter.
We have also found that the −141 to −54 proximal region of the rat ucp-1 gene is crucial for the basal transcriptional activity of the ucp-1 promoter. This is consistent with the presence of deoxyribonuclease I (DNase I) hypersensitivity in this region (34). Furthermore, the −139/−122 CRE has enhancer properties in differentiated brown adipocytes (where the endogenous ucp-1 gene is highly expressed) but not in nondifferentiated cells (which have a much lower expression of the endogenous ucp-1 gene). Therefore, in addition to being the main cAMP-responsive element identified in the rat ucp-1 gene, the −139/−122 element probably accounts for the basal promoter activity of the gene characteristic of the differentiated brown adipocyte. In the mouse ucp-1 gene, the presence of several putative CRE sequences along the gene has been proposed (4). The 5′-GCGCGTCA-3′ core of the− 139/−122 UCP-CRE (antisense strand) is identical to a CRE sequence in the proximal region of the mouse gene that was also claimed to be important for basal expression (4). Present data fully establish the relevance of this proximal CRE in both basal and cAMP stimulation of transcription from the ucp-1 gene promoter. On the other hand, although our present results demonstrate a major role for this proximal CRE in the rat ucp-1 gene, the presence of other CRE sequences upstream from the ucp-1 gene promoter should be considered. In the mouse gene, a distal CRE located in the enhancer region was proposed to confer most of the norepinephrine responsiveness (4). The fact that the distal CRE in the mouse and the corresponding sequence in the rat are less conserved than the proximal CRE may explain the differences in the relative roles of these sites in these species.
Present results demonstrate that c-Jun expression in brown adipocytes represses basal transcription from the ucp-1 gene promoter. Furthermore, c-Jun completely blocks the induction of transcription from the ucp-1 promoter by the catalytic subunit of PKA. The− 141/−54 region responsible for repression by c-Jun colocalizes with the main cAMP-responsive region in the rat ucp-1 gene promoter. Furthermore, the isolated UCP-CRE confers c-Jun-dependent repression to the neutral tk gene promoter. Both the DNA-binding and the leucine zipper domains of c-Jun are required to mediate its inhibitory effect on ucp-1 gene transcription. These findings, along with c-Jun binding to the UCP-CRE, are consistent with a model in which c-Jun blocks ucp-1 transcription by interacting directly with the ucp-1 promoter via formation of functional transcriptional complexes either alone or with other(s) member(s) of the CREB/ATF family. Involvement of c-Jun/c-Fos heterodimers in mediating this effect is not likely because of the lack of effect of c-Fos cotransfection on ucp-1 gene transcription. Any potential effect mediated by c-Jun/c-Fos heterodimers would be enhanced by cotransfection of both expression vectors, which does not appear to be the case for the repression of c-Jun upon ucp-1 gene transcription. Although the involvement of other Jun-related proteins, such as Jun-B or Jun-D, cannot be ruled out, preliminary data indicate that Jun-B is uneffective in repressing ucp-1 gene transcription (P. Yubero, F. Villarroya, and M. Giralt, unpublished observations). Also, our present findings rule out C/EBPs as components of the brown fat nuclear protein complexes interacting with the −139/−122 CRE, even though C/EBPβ is overexpressed in rat brown fat under noradrenergic stimulus (35), and C/EBPβ binds and transactivates several CRE reporter gene constructs (31, 36). On the other hand, neither the two C/EBP responsive elements present in the proximal regulatory region of the ucp-1 gene (5) nor the cis-acting sequences in the enhancer region of the gene (6–9) are required for the negative regulation by c-Jun. However, whether this negative regulation by c-Jun through the UCP-CRE affects the stimulation of the ucp-1 gene transcription by retinoic acid (6, 7), thyroid hormones (8), or agonists of PPARγ (9) remains to be determined.
From our present results, a model in which an opposite action of CREB and c-Jun in regulating basal and PKA responsiveness of transcription from the ucp-1 gene through direct competition by binding to the −139/−122 CRE is proposed. Support for this hypothesis comes from several lines of evidence. Both PKA and c-Jun-dependent regulation are reproduced by the isolated UCP-CRE but are lost when a double-point mutation that abolishes binding is introduced in this element. Furthermore, overexpression of CREB blocks c-Jun repression consistently with a direct competition for binding to the same site in the UCP-CRE. Effects of CREB alone on PKA responsiveness were hardly observed in cotransfection assays, probably because of the high constitutive expression of CREB in brown fat cells (Ref. 37 and Fig. 8C), similarly to what has been described for CREB-responsive genes in other cell types (38). In an analogous manner to the present findings on the ucp-1 gene, other studies have shown that c-Jun is able to repress transcription via CRE sites, as for instance in the human insulin gene (27) and the α- and β-subunit genes for human CG (28). In contrast, the c-jun gene promoter is negatively regulated by CREB and positively regulated by c-Jun (39). Thus, a single cis-element provides positive or negative regulation depending upon the relative abundance of active transcription factors binding to this site.
On the basis of these results, we propose that the expression of the ucp-1 gene and its responsiveness to cAMP are modulated by the relative abundance of c-Jun in the brown adipocyte. We report here that c-Jun content correlates inversely with the acquisition of the terminally differentiated brown adipocyte phenotype, as assessed by both cell morphology and ucp-1 gene expression. In differentiating white adipocytes, a transient increase in the expression of c-jun occurs in association with the mitotic clonal expansion before terminal differentiation (40). Furthermore, exposure of 3T3-L1 adipocytes to agents that inhibit their differentiation process, such as tumor necrosis factor-α or retinoic acid, results in a persistent rise in the expression of c-Jun mRNA (41, 42). The higher expression of c-Jun mRNA and protein in nondifferentiated compared with differentiated brown adipocytes suggests a parallel role for c-Jun in brown adipocytes, although the cell cycle status during their differentiation is unknown. Further research is also necessary to identify the specific mechanisms down-regulating c-Jun abundance, and perhaps c-Jun activity, in association with brown adipocyte differentiation.
The induction of the expression of the ucp-1 gene, which is repressed in nondifferentiated cells, is the key event in the acquisition of the differentiated phenotype of the brown fat cell both during development (43) and in cell culture (11, 44). Furthermore, terminal brown adipocyte differentiation is associated with an enhancement in the responsiveness of ucp-1 gene expression to the adrenergic stimulus both in vivo (43) and in primary cultured brown adipocytes (11). This differentiation-dependent modulation of the ucp-1 gene transcription is reproduced when analyzing the activity of the UCP-CRE, thus indicating that this element can support these regulatory effects. Furthermore, the action of c-Jun as a dominant inhibitor of basal and cAMP-induced transcription from the ucp-1 promoter provides the first evidence of a molecular mechanism by which expression of the ucp-1 gene can be differentially regulated by norepinephrine in undifferentiated vs. terminally differentiated brown adipocyte cells. Although the involvement of the upstream enhancer-regulatory region cannot be ruled out, a schematic overview of differences in the transcriptional regulation of the ucp-1 gene associated with brown adipocyte differentiation is proposed on the basis of the present findings on the proximal regulatory region (see Fig. 9).
Model for the Differentiation-Dependent Regulation of the ucp-1 Gene Transcription through Its Proximal Regulatory Region The proximal regulatory region of the rat ucp-1 gene contains the CRE at −139/−122 and two C/EBP-responsive elements at− 457/−440 and −335/−318 (5 ). In nondifferentiated cells, low levels of basal and cAMP-induced ucp-1 gene expression are explained by c-Jun repression together with low transactivating activity by C/EBP proteins due to its low expression in nondifferentiated brown adipocytes when compared with the differentiated cells (6 ). In contrast, differentiated brown adipocytes have high levels of basal ucp-1 gene expression, and the gene is fully responsive to the noradrenergic stimulus. At the transcriptional regulatory level, this can be explained by C/EBP transactivation and the release of c-Jun inhibition allowing CREB-dependent regulation of the UCP-CRE.
MATERIALS AND METHODS
Materials
DNA-modifying enzymes and poly(deoxyinosinic-deoxycytidylic)acid were purchased from Boehringer Mannheim (Indianapolis, IN) or Promega (Madison, WI). [α-32P]dCTP was from Amersham (Arlington Heights, IL) and d-threo-[1,2-14C]chloramphenicol was from ICN (Cleveland, OH). Tissue culture media and FCS were obtained from Biowhittaker (Verviers, Belgium). T3, insulin, norepinephrine (arterenol bitartrate), 8-bromo-cAMP, forskolin, and IBMX were from Sigma (St. Louis, MO).
Oligonucleotides and Plasmids
Oligonucleotides were chemically synthesized by Boehringer Mannheim. The UCP-CRE double-stranded oligonucleotide corresponds to positions −139 to −122 of the rat ucp-1 gene, and its sequence is 5′-GGGAGTGACGCGCGTCTG-3′, flanked by XbaI ends. The mutated version mutUCP-CRE corresponds to the sequence 5′-GGGAGTGTGGCGCGT-CTG-3′ also flanked by XbaI ends. The PEPCK-CRE-I and S-CRE are double-stranded oligonucleotides corresponding to the −94/−77 and −60/−29 CREs of the rat phosphoenolpyruvate carboxykinase (45) and rat somatostatin (46) gene promoters, respectively. ETS is an oligonucleotide corresponding to the −208/−192 GA-rich region in the stromelysin promoter used as negative control in the DNA binding experiments (47).
(−4551)UCP-CAT, a pSP73-derived plasmid containing the region −4551 to +110 of the rat ucp-1 gene driving the promoterless CAT gene, was kindly provided by Dr. D. Ricquier (3). The plasmids (−2494)UCP-CAT, (−896)UCP-CAT, (−141)UCP-CAT and (−54)UCP-CAT were constructed using the internal restriction sites AatII, HindIII, BstXI, and NaeI in (−4551)UCP-CAT, respectively. The internal deletions between nucleotides −172/−54 and −2469/−54 were carried out by digesting with SpeI/NaeI and BclI/NaeI, respectively.
Plasmids derived from (−4551)UCP-CAT but lacking the artifactual AP-1 site (32) present upstream from the polylinker of pSP73 (position 8449 in (−4551)UCP-CAT)(3) were constructed as follows:Δ AP1(−2494)UCP-CAT was obtained by eliminating the fragment between 8348 and 2058 in (−4551)UCP-CAT by digestion and further religation using the AatII sites at those positions. The plasmid (Δ−172/−54)(−2494)UCP-CAT containing the internal deletion between nucleotides −172 and −54 was obtained using the unique SpeI and NaeI sites in ΔAP1(−2494)UCP-CAT.Δ AP1(−141)UCP-CAT and ΔAP1(−54)UCP-CAT were obtained by digestion and further religation of the original plasmids using the AatII and BglII sites corresponding to the former 8348 and 8690 positions in (−4551)UCP-CAT.
The heterologous (UCP-CRE)-tk-CAT vectors in which copies of the− 139/−122 sequence of the ucp-1 gene are placed upstream from the HSV thymidine kinase promoter were generated by cloning one, two, or three copies (direct repeats) of the synthetic double-stranded oligonucleotide UCP-CRE into the XbaI site of a version of pBLCAT2 in which the artifactual AP-1 site (position 32 in pBLCAT2) (48) had been previously deleted by an AatII/HindIII digestion and further religation. The mutant version (mutUCP-CRE)-tk-CAT was generated by cloning two copies of the mutUCP-CRE double-stranded oligonucleotide as a direct repeat into the XbaI site of the (ΔAP1)pBLCAT2 plasmid.
SRα-PKA is an expression vector for the catalytic subunit of PKA transcribed from the SRα promoter (49). Construction of pRSV-CREB, the mammalian expression vector for full-length CREB-1, has been described (21). Expression plasmids driving c-Fos and various forms of c-Jun were kindly provided by Dr. T. Curran. pCMV-c-Jun and pCMV-c-Fos are mammalian expression vectors that contain the rat cDNAs of c-Jun and c-Fos, respectively, driven by the cytomegalovirus promoter (50, 51). pCMV-JunΔL3 is a CMV-driven expression vector containing a leucine-to-valine amino acid substitution at leucine 3 within the leucine zipper domain, which disrupts dimerization (50). pCMV-Jun ΔBR contains amino acid 260–266 deletion within the DNA-binding domain that disrupts DNA binding (50).
Cell Culture and Transfection Assays
Isolation and culture of brown preadipocytes was performed as described (6, 11). Three-week-old Swiss mice were killed and interscapular, cervical, and axillary depots of brown fat were removed. Precursor cells were isolated, plated on 60-mm petri dishes (7500 cells/cm2), and grown in 5 ml DMEM-Ham’s F12 medium (1:1) supplemented with 10% FCS, 20 nm insulin, 2 nm T3, and 100 μm ascorbate (regular differentiating medium). When indicated, cells were grown in a hormone-depleted (nondifferentiating) medium containing 10% charcoal-treated FCS (6).
Murine primary brown adipocytes differentiated in culture were transiently transfected by the calcium phosphate precipitation method on day 7 of culture, when 80–90% of cells had already differentiated (5). Each transfection contained between 5 and 15 μg of UCP-CAT vectors and included or not the indicated amounts of expression vectors. When indicated 0.1 μm norepinephrine, 1 mm 8-Br-cAMP, 0.5 mm IBMX, or 10μ m forskolin were added after transfection. RSV-β-galactosidase (2 μg) was included in all the experiments to assess the efficiency of separate transfections. The cells were incubated for 24 h and, for each condition, at least three plates were pooled. The experiments were performed at least twice using independent DNA preparations of each construct. Analysis of CAT activity was carried out as described (52, 53). Acetylation of[ 14C]chloramphenicol was determined by TLC and quantified by radioactivity counting (AMBIS, San Diego, CA). The CAT activity was normalized for variation in transfection efficiency using theβ -galactosidase activity measured for each sample as a standard.
DNA Binding Experiments
Nuclear proteins were isolated from rat brown adipose tissue as reported previously (5). Protein extracts from differentiated and nondifferentiated brown adipocyte nuclei were prepared as described (54). Protein concentration was determined by the micromethod of Bio-Rad (Richmond, CA) using BSA as standard. Partially purified bacterially expressed CREB was a kind gift from Dr. R. Hanson (45). C/EBPβ is an 11-kDa polypeptide form of C/EBPβ (95% pure), kindly provided by Dr. S. McKnight (55). Recombinant purified c-Jun was from Promega.
For the gel retardation assays, the UCP-CRE or mut-UCP-CRE oligonucleotides were end-labeled using [α-32P]dCTP and Klenow enzyme. The DNA probe (10,000–20,000 cpm) was incubated for 30 min at 25 C with 5 μg of brown adipose tissue nuclear protein extract or purified CREB, C/EBPβ, or c-Jun proteins. Reactions were carried out in a final volume of 20 μl containing 20 mm HEPES (pH 7.6), 0.1 mm EDTA, 1 mm dithiothreitol, 50 mm NaCl, 10% glycerol, and 2 μg (nuclear extracts) or 0.5 μg (purified proteins) of poly(deoxyinosinic-deoxycytidylic)acid. Samples were analyzed by electrophoresis at 4 C for 60–80 min in nondenaturing 5% polyacrylamide gels in 0.5× TBE (44.5 mm Tris, 44.5 mm borate, 1 mm EDTA). Gels were analyzed by autoradiography. In the competition experiments, 100-fold molar excess of unlabeled oligonucleotide was included in each respective binding reaction. When indicated, 0.2 μl or 1 μl of rabbit antiserum against Jun proteins, kindly provided by Dr. R. Bravo (56), or 1 μl of an antiserum against CREB (Santa Cruz Biochemicals, Santa Cruz, CA), or equivalent amounts of preimmune (control) serum were incubated with the brown adipose tissue nuclear extracts for 2 h at 4 C before incubation with the labeled probe.
RNA Isolation and Northern Blot Analysis
Total RNA was extracted from cultured brown adipocytes by a single-step method using guanidine hydrochloride (57). For Northern blot analysis, 15 μg of total RNA were denatured, electrophoresed on 1.5% formaldehyde/agarose gels, and transferred to nylon membranes (Hybond N, Amersham). Ethidium bromide (0.2 μg/ml) was added to RNA samples to check equal loading of gels and transfer efficiency (58). Hybridization and washing were carried out as reported (43). Blots were hybridized to DNA probes corresponding to the full-length cDNA for rat UCP-1 (59) or rat c-Jun (50). The cDNA probes were labeled with[α -32P]dCTP using the random oligonucleotide-primer method. Autoradiographs were quantified by densitometric scanning (LKB Instruments, Rockville, MD).
Immunoblot Analysis
Samples containing equal amounts of nuclear protein extracts from differentiated or nondifferentiated brown adipocytes were electrophoresed on 0.1% SDS/12% polyacrylamide gels. Proteins were transferred to polyvylidene difluoride membranes (Millipore, Bedford, MA) and probed with the antisera against CREB (Santa Cruz Biochemicals) or Jun (56). Immunoreactive material was detected by the enhanced chemiluminescence (ECL) detection system (Amersham). The sizes of the proteins detected were estimated by using protein molecular mass standards (Bio-Rad).
Statistical Analysis
Where appropriate, statistical analysis was performed by Student’s t test and significance is indicated in the text.
We thank Drs. D. Ricquier, T. Curran, R. Hanson, S. McKnight, M. Muramatsu, and R. Bravo for the generous gifts of plasmids, purified proteins, and/or antisera.
This work was supported by Grant PB95–09695 from DGICyT, Ministerio de Educación y Cultura, Spain, and by Grant 1995SGR-00096 from Generalitat de Catalunya.
These authors made equal contributions to this work.
