| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Departments of
* Endocrinology and
Biostatistics, Genentech, Inc., South San Francisco, California 94080, USA; and
Department of Collaborative Research, CuraGen Corporation, New Haven, Connecticut 06511, USA
1Correspondence: Metabolic and Cardiovascular Disease Pharmacology Department, Novartis Pharmaceuticals Corporation, 556 Morris Ave., Summit, NJ 07901, USA. E-mail: sean.adams{at}pharma.novartis.com
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
twofold). It is proposed that temperature-sensitive regulation of distinct intracellular malonyl-CoA pool sizes plays an important role in driving this unique metabolic profile via maintenance of the lipogenic pool but diminution of the carnitine palmitoyltransferase 1 inhibitory pool under cold conditions.—Yu, X. X., Lewin, D. A., Forrest, W. F., Adams, S. H. Cold elicits the simultaneous induction of fatty acid synthesis and ß-oxidation in murine brown adipose tissue: prediction from differential gene expression and confirmation in vivo.
Key Words: PPAR • glucose transport • obesity • ACC
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
, 2
). The critical role for UCP1 in adaptive thermogenesis was further underscored by the cold intolerance displayed by UCP1 knockout mice (3)
. The transition of BAT to a more metabolically amplified state is largely driven by increased sympathetic signaling, which participates in stimulation of lipolysis and glucose uptake, increases in gene expression of UCP1 and other metabolically relevant genes, and ultimately enhancement of oxygen consumption.
Although UCP1-driven uncoupling is necessary for maximizing metabolic rate and associated heat production in BAT, additional systems cooperate in this process. For instance, maximal thermogenesis in rodent BAT appears to have an absolute reliance on maintenance of an adequate long chain fatty acid (LCFA) pool with coordinately active LCFA entry into mitochondria via carnitine palmitoyltransferase 1 (CPT1) activity. Indeed, inhibitors of CPT1 abolished catecholamine- or LCFA-driven stimulation of oxygen consumption and glucose uptake in brown adipocytes (4
5
6)
. A similar diminution of catecholamine-stimulated BAT metabolism has been observed in lipid-depleted BAT (7)
or situations where lipolysis is diminished (4)
. Furthermore, mice with genetic disruptions of either short or long chain fatty acyl-CoA dehydrogenase displayed marked cold intolerance (8)
. Despite accumulating evidence for a clear reliance on LCFA availability and ß-oxidation for driving BAT thermogenesis, our understanding of the molecular mechanisms regulating LCFA metabolism in this tissue remains incomplete.
BAT carbohydrate metabolism, like that of fatty acids, is clearly altered upon transition to the thermogenic state. BAT in cold-acclimated rodents is an active consumer of glucose (9
10
11
12)
. Stimulation of glycolysis may provide ATP via substrate-level phosphorylation of ADP in light of the poor coupling of mitochondrial oxidative phosphorylation in thermogenic BAT. Robust lipogenesis from glucose or other substrates has been reported in BAT (13
14
15
16
17
18)
, providing another potential sink for glucose carbon taken up by the tissue. Although some papers have reported cold-induced changes in enzyme activities associated with glucose flux and lipogenesis in BAT (12
, 15
, 18
19
20)
, the molecular events supporting this system require further evaluation.
The metabolic malleability of rodent BAT in response to alterations in ambient temperature makes this tissue a valuable site to explore the molecular underpinnings of intermediary metabolism. With this in mind, a comprehensive analysis focusing on temperature-dependent alterations in the expression of genes associated with carbohydrate and LCFA metabolism was carried out, using BAT from mice exposed to a range of environmental temperatures (cold, control, warm acclimated). The premise of such analyses was that temperature-sensitive directional changes in mRNAs corresponding to metabolically relevant genes may predict directional changes in associated metabolic pathways. Recently, Daikoku et al. analyzed the expression of 39 metabolically relevant genes in BAT from 48 h cold-exposed rats vs. controls (21)
. The data led the authors to speculate that whereas BAT glucose utilization and LCFA combustion are induced by cold in the rat, lipogenesis is diminished. Information regarding the time course of events and the effect of expression changes under warm conditions was not presented, and physiological or biochemical correlates to the expression patterns were not examined. Here we report temperature-sensitive gene expression profiles of 50 genes that led to a prediction of increased de novo lipogenesis, elevated glucose utilization, and active LCFA ß-oxidation in the BAT of cold-challenged mice. Such a unified hypothesis is consistent with the aggregate of information from the literature. Although paradoxical in light of the powerful inhibitory effect of the lipogenic intermediate malonyl-CoA on CPT1, experiments in vivo using radiolabeled tracers confirmed the model. Up-regulation of BAT malonyl-CoA decarboxylase (MCD) is proposed as a candidate regulatory mechanism to explain these results, based on the almost twofold induction of MCD mRNA levels in BAT samples derived from cold-challenged mice compared with controls.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
Animals
Animals used for the studies described here were male FVB-N mice purchased from Taconic (Germantown, NY) received at 3 wk of age and housed at two mice/cage until tissue harvest at 6 wk of age. Mice were fed standard rodent chow ad libitum (Chow 5010, Ralston Purina, St. Louis, MO) and were housed on a 12:12 light/dark cycle (lights on 06:00). Protocols were approved by the Genentech Animal Care and Use Committee.
Differential gene expression analyses
The studies were designed to identify genes differentially expressed in mouse BAT with changes in environmental temperature. Mice were received at 3 wk of age and exposed to a variety of environmental temperatures or nutritional conditions until 6 wk of age before tissue harvest. Control and cold-challenged mice were housed at an ambient temperature (Ta) of 22°C and warm-acclimated mice were housed at 33°C, within their thermoneutral zone. Cold-challenged mice were transferred to a 4°C room for 48 h before tissue harvest. Warm acclimation led to a 50% reduction in food intake (not shown). This prompted the addition of a food-restricted control group to test whether warm-induced changes in gene expression were specific to temperature or diminished food intake. The food-restricted group was meal fed each morning (09:00) for 3 wk before tissue harvest, given an amount of chow equal to the mean intake of a cohort of warm-housed mice for the previous 24 h. Tissues were harvested from this group at 4–6 h postmeal feeding on the final day. After CO2-induced euthanasia in the afternoon, interscapular BAT was excised, cleaned of visible blood vessels, white adipose tissue (WAT), and connective tissue, and snap-frozen for subsequent RNA preparation. For differential expression studies, three independent BAT samples/treatment were generated for analysis; each sample was composed of BAT pooled from 10 mice. Follow-up studies examining the time course of cold inducibility of BAT genes were carried out on mice exposed to 4°C for various periods of time (1, 6, and 24 h) to augment data derived from the 48 h time point.
Initial analyses exploring condition-dependent gene expression profiles were carried out using quantitative expression analysis (QEA) or GeneCalling® (22)
. mRNA isolated from tissues treated under various conditions was reverse-transcribed using oligo-dT as primer. For each sample, 96 pools of variously sized fluorescently labeled fragments were generated by using linkers and primers matched to 96 pairs of restriction enzymes used to digest aliquots of the sample cDNA. Analyses were carried out in triplicate for each sample. Fragments were amplified by PCR and resolved using gel electrophoresis. GeneCalling software was used to make comparisons of traces derived from sample sets on which GeneCalling chemistry had been performed (see Fig. 1
). Sequence information and precise electrophoretic mobility were used to query a database of predicted fragments to initially identify genes whose expression was modulated by a given treatment. The identities of bands of interest, including those associated with genes presented here, were confirmed through the use of ‘poisoning’ PCR oligonucleotides (22)
and/or direct cloning and sequencing of the band excised from the gel, followed by poisoning (see Fig. 1
, inset). When necessary, contig analysis (Seqextend software, Genentech, and CuraToolsTM-SeqExtender software, CuraGen Corporation) was applied to partial length cDNAs to build consensus sequences of adequate length for identification using public databases and proprietary databases. Within each of the main comparisons below, 27,000 to 28,000 bands were assayed, estimated to represent at least 12,000 genes per the results of Shimkets et al. (22)
. The percentage of bands that differed at least twofold was 4.1% (cold vs. control; identities of a subset of 107 bands confirmed, representing 87 genes), 2.4% (warm vs. control; identities of a subset of 12 bands confirmed, representing 12 genes), and 4.5% (cold vs. warm; identities of a subset of 144 bands confirmed, representing 139 genes). These experiments focused on changes in the expression of metabolically relevant genes as a subset of the overall QEA data set. An illustrative example of QEA results associated with a gene (acyl-CoA synthetase 5) induced in BAT upon cold exposure and repressed on warm acclimation is depicted in Fig. 1
.
|
Real-time quantitative RT-PCR (TaqMan) quantitation of the abundance of mRNAs derived from metabolically relevant genes was carried out for a subset of genes emerging from the QEA analyses, plus additional candidates involved with lipid and glucose metabolism. The assay conditions described in detail elsewhere (23)
used standard reagents (PE Biosystems, Foster City, CA). Specific oligonucleotide sequences used to generate primers and probes are given in Table 1
. Each sample was assayed in duplicate, using three independent samples per treatment. All data were normalized to 18S mRNA levels to account for differences in total mRNA (18S primers/probes were purchased from PE Biosystems).
|
2-Deoxyglucose (2-DOG) uptake in BAT
The uptake of the nonmetabolizable glucose mimetic 2-[1,2-3H(N)]-deoxy-D-glucose into BAT after intraperitoneal (i.p.) injection of 2-DOG was considered to mirror uptake of glucose in this tissue. The experimental protocol and tissue 2-DOG uptake calculations were established by Hom et al. (24)
. At 15:00, control mice (22°C, n=7) or cold-challenged mice (4°C, 48 h, n=8) were injected i.p. with a 100 µl PBS solution containing 5 µCi [3H]-2-DOG and 2 µCi [U-14C]-sucrose. The latter was injected to accommodate corrections for [3H]-2-DOG radioactivity in the extracellular space (24)
. At 30 min postinjection, mice were anesthetized under CO2 and blood was collected by heart puncture. After death under CO2, the intrascapular BAT was removed, weighed, and placed in tubes on dry ice (note that cold-challenged mice were injected and remained in the cold until death). Entire BAT depots (100–150 mg) were transferred to glass scintillation vials containing 1 ml Solvable tissue solvent, placed in a 60°C water bath for 4 h with occasional swirling to solubilize the material, and partially decolorized by the addition of 200 µl 30% H2O2 with a return to the water bath for 30 min. On cooling, 10 ml Hionic Fluor was added, vials thoroughly mixed, and the samples analyzed for radioactive counts 30 min later (LS5000CE, Beckman Instruments, Palo Alto, CA). Dual-isotope counting with a quench curve was used to determine 3H and 14C radioactivity; under these conditions, radioactive counts in the 3H window were not compromised by 14C, and vice versa (not shown). Plasma samples (20 µl, in duplicate, combined with the reagents above) were also analyzed in this fashion.
Assessment of de novo fatty acid synthesis
The protocol used was patterned after established methods that track the incorporation of 3H2O into saponifiable tissue lipids (25)
. Control (n=7) or cold-challenged mice (4°C, 48 h, n=8) were injected i.p. at 15:00 with 3 mCi/mouse 3H2O in 100 µl PBS. At 1 h postinjection, blood was collected by heart puncture after CO2 anesthesia for subsequent plasma analyses. BAT was excised, rinsed in fresh PBS, blotted, weighed, and frozen in tubes on dry ice. Lipids were extracted from whole BAT depots using the following procedure: 1) samples were spiked with a trace amount of [1-14C]-palmitate (6000 cpm) as internal standard to follow recovery of LCFA, 2) 300 µl 30% (w/v) KOH was added and samples were placed at 70°C 15 min, after which 300 µl ethanol was added and additional heating used for 2 h with occasional mixing, 3) samples were acidified with 300 µl 9M H2SO4 and extracted three times using 1 ml light petroleum ether with vortexing, 4) the ether fractions (total volume 3 ml) were combined, extracted three times using 1 ml H2O, then transferred to scintillation vials and dried, and 5) radioactivity was quantified using dual-isotope counting after addition of 10 ml Econofluor scintillant with vortexing. The specific activity of plasma water was determined through measurement of tritium radioactivity in plasma samples (10 µl) and using a value of 53M for H2O (17)
. All values reported have been corrected for the recovery of palmitate internal standard, which averaged 75% across all samples.
ß-Oxidation in BAT
A direct quantitative measure of ß-oxidation in mouse BAT in vivo requires analysis of the flux rate of LCFA toward metabolic end products, including flow of LCFA carbon to CO2 and water-soluble products (acetyl-CoA, acetate, acetyl-carnitine, Krebs cycle intermediates, etc.). This approach poses many technical challenges, such as the need to follow LCFA tracer kinetics across the BAT in situ. Thus, we used a less complicated approach to assess the tissue-specific activity of ß-oxidation by analyzing the amount of [1-14C]-palmitate carboxyl-carbon present in the water-soluble fraction of BAT after an injection of a tracer amount in vivo. Tail vein injections of a radiolabeled palmitate solution (palmitate:BSA ratio 5:1; 5 µCi/mouse) in 50 µl PBS were administered to control (22°C, n=8) or cold-challenged (4°C, 48 h, n=8) mice in the afternoon. This procedure was facilitated by 3 min exposure of cold mice to a warm heat lamp before injection, after which they were returned to the cold room until tissue harvest. At 60 min postinjection, mice were killed under CO2 and the intrascapular BAT removed, weighed, and placed in tubes on dry ice. This time was chosen based on pilot studies in a separate cohort of mice indicating that the water-soluble radioactivity per milligram BAT was stable between 60 and 90 min postinjection. Keeping samples on ice throughout, acid-soluble products (ASP, containing water-soluble metabolites) were extracted as follows: 1) 250 µl 6% (w/v) HClO4 was added to a given sample in a microcentrifuge tube and the tissue was minced using scissors, 2) minced samples were homogenized thoroughly using a disposable TeflonTM pestle and a hand-held motorized Kontes homogenizer and spiked with [3H]-glucose (4500 cpm) tracer to track recovery of water-soluble metabolites, 3) an additional 250 µl HClO4 was added with vortexing and samples were centrifuged 10 min at 12,000 g (4°C), 4) the clarified infranatant containing ASP was transferred to a 2 ml microcentrifuge tube (samples generally displayed a fatty layer above the infranatant, which was readily punctured for ASP removal by using a syringe and needle) and volume was brought to 500 µl using HClO4, 5) after addition of 300 µl ethanol, samples were extracted three times using 1 ml petroleum ether, resulting in the final ASP preparation. Although the extractions were designed to remove any residual [1-14C]-palmitate, negligible radioactivity was detected in the petroleum ether fraction, with virtually all of the LCFA remaining in the pelleted residue (see below). The volume of ASP was calculated by noting the weights of final ASP samples coupled to the specific gravity of representative ASP aliquots. Radioactivity contained in 600 µl aliquots was quantified using dual isotope counting, and data were expressed as total ASP dpm/100 mg BAT. All values reported have been corrected for the recovery of [3H]-glucose internal standard, which averaged 67% across all samples.
The fate of injected [1-14C]-C16:0 taken up into BAT was presumed to mimic the metabolism of endogenous LCFA in the tissue. Thus, an accounting of the tissue LCFA specific activity (SA) was performed through analysis of the pelleted residue and fatty layer derived from the initial HClO4 extraction (see above): 1) 400 µl chloroform/methanol (2:1) was added to tubes containing the residues, with subsequent dissociation of the material via mincing with scissors, 2) minced samples were homogenized using a disposable TeflonTM pestle and a hand-held motorized Kontes homogenizer, 3) samples were extracted overnight at room temperature after an additional 800 µl chloroform/methanol was added, 4) contents were vortexed and poured into 2 ml microcentrifuge tubes through a cell strainer cap (derived from 12x75 mm Falcon tubes, Cat. #352235; Becton Dickinson, Franklin Lakes, NJ) to remove debris, 5) the total volume of chloroform/methanol was brought to 1500 µl, 6) radioactivity in 200 µl aliquots was quantified in duplicate using 15 ml Econofluor, and 7) LCFA concentrations in the chloroform/methanol fractions were determined in triplicate through the use of a standard kit (Wako NEFA-C, Neuss, Germany) miniaturized to use 5 µl sample. The volumes of the final chloroform/methanol preparations were calculated by noting the weights of the final sample coupled to the specific gravity of representative chloroform/methanol aliquots. Specific activities as 14C dpm/nmol LCFA carboxyl-carbon, in combination with total ASP dpm (above), allowed for calculation of the nmol LCFA carboxyl-carbon residing in sample ASP pools at time of harvest: 1/SA * dpm in ASP pool = nmol LCFA carboxyl carbon in ASP pool.
Total tissue LCFA content was calculated from LCFA concentration data and total chloroform/methanol volume and expressed as nmol LCFA/100 mg BAT. Recovery of LCFA was presumed to be equivalent between treatments.
Statistical analyses
Treatment-related differences in mean expression values of genes analyzed by quantitative real-time RT-PCR were assessed statistically by a Dunnett’s test. Treatment means were compared against control values within each gene, using a pooled standard error of the mean derived from data across all genes and treatments. A Student’s t test was used to assess differences between cold vs. control BAT 2-DOG uptake, lipogenesis, and LCFA ß-oxidation.
Data emerging from quantitative real-time RT-PCR were analyzed using a mathematical clustering algorithm in order to determine which genes displayed similar expression patterns. Distance between the treatment averages was defined by the Manhattan distance (26)
, which sums the six absolute differences from the six paired treatment groups for two genes to determine the distance between them. All computations were completed using software from SPLUS version 3.4 (Insightful, Seattle, WA) and SAS version 6.12 (SAS Institute, Cary, NC). Data were placed into six clusters for this analysis.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
, pyruvate kinase, LDH, and glycerol-3-phosphate dehydrogenase (cytosolic, NADH-linked) appeared elevated in cold-challenged mouse BAT compared with control samples in the QEA analysis (Table 1)
. In contrast, QEA indicated a cold-related diminution of phosphoenolpyruvate carboxykinase mRNA level in BAT (Table 1)
. Patterns initially emerging from the QEA results were confirmed in follow-up experiments using quantitative real-time RT-PCR (TaqMan), in that cold induced the expression of a range of genes associated with carbohydrate utilization in BAT. Under cold conditions, for instance, the abundance of mRNAs encoding phosphofructokinase C (PFK-C), PFK-L, and PDH were significantly increased by cold exposure (Fig. 2
A). Remarkably, acute cold exposure (1–24 h) strongly attenuated gene expression for GLUT1, GLUT2, and GLUT3, with a qualitatively similar but less pronounced pattern for the newly described carrier GLUT8 (Fig. 2B
). In contrast, a significant rise in GLUT4 mRNA was observed under these conditions (Fig. 2B
). Many genes whose expression levels were induced by cold were also increased by the food restriction/meal feeding paradigm (see Table 2
and Fig. 2
).
|
|
TaqMan analyses clearly indicated a significant two- to threefold induction of aconitase mRNA under cold or food-restricted conditions compared with controls (Fig. 2A
). A similar analysis for isocitrate dehydrogenase (ICDH) indicated that compared with controls, the various treatments yielded no significant alteration of its mRNA levels (not shown).
Differential expression of genes associated with lipogenesis, malonyl-CoA generation, and fatty acid desaturation
QEA patterns indicated an increase in the expression of the lipogenesis-associated genes ATP-citrate lyase, fatty acid synthase (FAS), and glycerol-3-phosphate acyltransferase (GPAT) in response to cold exposure (Table 1)
. Consistent with this, TaqMan analysis indicated that levels of ATP-citrate lyase mRNA rose twofold in the BAT of cold-challenged mice (Fig. 3
). Expression was markedly depressed in the course of warm acclimation (Fig. 3)
. By TaqMan evaluation, FAS mRNA levels were also decreased by warm acclimation but, in contrast to QEA profiles, were not found to be increased by cold exposure (Fig. 3)
. The same pattern held for GPAT mRNA changes (not shown). Food restriction/meal feeding significantly increased ATP-citrate lyase mRNA levels, with a trend (1.5-fold) toward higher expression for FAS (Table 2
and Fig. 3
).
|
Cold elicited a significant two- to threefold rise in levels of mRNAs encoding acetyl-CoA carboxylase 1 (ACC1) and ACC2 in mouse BAT (Fig. 3)
. ACC2 mRNA was also increased under food-restricted/meal feeding conditions but was attenuated under warm conditions (Fig. 3)
. The QEA profiles for stearoyl-CoA desaturase 1 (SCD1) and SCD2 indicated cold induction of these genes in BAT, with a down-regulation of SCD1 expression under warm acclimation (Table 2)
. By quantitative TaqMan analysis, both genes were down-regulated significantly under warm conditions (Fig. 3)
. Although SCD2 mRNA abundance displayed a robust cold induction, SCD1 mRNA levels were unchanged, in contrast to the QEA (Fig. 3)
. Both mRNAs were increased by the food restriction/meal feeding regimen (Fig. 3)
.
Differential expression of genes associated with fatty acid availability and catabolism
By QEA analysis, the expression of many genes associated with lipid metabolism were induced in mouse BAT under cold conditions. For example, QEA suggested cold-induced increases in the mRNAs corresponding to lipoprotein lipase (LPL), monoglyceride lipase (MG-lipase), very long chain acyl-CoA dehydrogenase (VLCAD), long chain CAD (LCAD), medium chain CAD (MCAD), acyl-CoA synthetase 5 (ACS5), carnitine/acylcarnitine translocase (CAC), and MCD (Table 2)
. TaqMan analysis concurred with these patterns, as the BAT abundance of LPL, MG-lipase, CAC, and MCD mRNAs were increased two- to threefold in the cold (Fig. 4
). Levels of these mRNAs were not affected by food restriction/meal feeding. Consistent with QEA data (Table 2)
, TaqMan analysis of BAT CPT2 mRNA indicated that transcript levels for this gene were attenuated significantly after warm acclimation and depressed with food restriction/meal feeding. M-CPT1 mRNA levels were induced twofold under cold conditions (Fig. 4)
and were unaffected by food restriction/meal feeding (Fig. 4)
. Although relatively low amounts of LCPT1 mRNA could be detected in BAT by quantitative real-time RT-PCR, no treatment effect was observed (not shown). As expected, the abundance for this species in BAT was only 5% of the level detected in liver and was present in BAT at only 3% of the amount of MCPT1 transcripts (not shown).
|
Differential expression of ß-adrenergic receptors and peroxisome proliferator-activated receptors (PPARs)
The mRNA abundance corresponding to the ß1-adrenergic receptor (ß1-AR) was largely unaffected by the experimental conditions used, with the exception of a significant drop in expression under food-restricted/meal-fed conditions (Fig. 5
). In contrast, ß2-AR mRNA levels fell quickly in response to cold or food-restricted/meal-fed conditions and recovered somewhat at 48 h vs. other time points in the cold (Fig. 5)
. After a transient induction of its mRNA abundance in the cold, expression of ß3-AR was significantly diminished by 48 h cold exposure to a level equivalent to that observed in warm-acclimated mice (Fig. 5)
. Expression of PPAR in BAT was strongly suppressed by food restriction/meal feeding and increased by warm acclimation (Fig. 5)
. Exposure to cold appeared to transiently decrease PPAR mRNA levels at 6 h whereas subsequent levels were not significantly different from controls. PPAR
mRNA levels were also attenuated with food restriction/meal feeding and were diminished significantly at all time points examined after cold exposure (Fig. 5)
. Warm acclimation did not affect expression of the gene.
|
Gene expression cluster analysis
As shown in Fig. 6
, BAT expression data for genes examined using quantitative real-time RT-PCR were used to generate six expression cohorts using a mathematical clustering algorithm to minimize investigator bias with respect to grouping (see Materials and Methods). Values associated with a given treatment (i.e., food-restricted, warm-acclimated, or a particular time point in the cold) were generated from BAT derived from separate cohorts of mice within each treatment and compared with mRNA levels observed in the BAT from a set of control mice. Examination of the results of such an analysis revealed distinct patterns of gene expression as follows. Cluster 1 represented genes whose expression was strongly inhibited by acute (
24 h) cold exposure or food restriction/meal feeding but recovered after 48 h cold exposure and was largely unaffected by warm acclimation. Cluster 2 includes genes whose expression increased in response to cold by 6–24 h and was largely unaffected by warm acclimation (UCP1 excepted), with variable response to food restriction/meal feeding. Cluster 3 included genes whose expression was largely unaffected by cold exposure but whose mRNA abundance dropped under warm acclimation conditions, with no effect or a slight increase with food restriction/meal feeding. Cluster 4 represented genes whose mRNA levels were depressed by cold exposure or food restriction/meal feeding and largely unaffected by warm conditions (PPAR excepted). Cluster 5 contains genes induced by cold (with an early trend upward, but transient lowering of mRNA level at the 6 h time point) and repressed by warm conditions, with a variable expression in response to food restriction/meal feeding. Finally, cluster 6 are those genes whose corresponding mRNA levels are induced by 24–48 h in the cold and are largely unaffected by warm conditions or food restriction/meal feeding.
|
Assessments of cold-related changes in glucose uptake, de novo lipogenesis, and LCFA ß-oxidation in BAT
Uptake of the nonmetabolizable glucose analog 2-DOG into the BAT of 48 h cold-challenged mice was 2.7-fold greater than controls (Fig. 7
A). A 2.8-fold stimulation of BAT de novo lipogenesis was also observed in mice exposed to cold 48 h (Fig. 7B
). Resident radioactivity in the water-soluble fraction of BAT after an intravenous dose of [1-14C]-C16:0 was considered a reflection of the activity of the LCFA ß-oxidation apparatus in BAT (see Materials and Methods). This parameter was increased twofold in the BAT of cold-challenged mice vs. controls (Fig. 8
A). The nonesterified LCFA content of BAT was significantly elevated in cold-exposed mice (786±41 nmol/100 mg BAT) vs. controls (476±24 nmol/100 mg BAT)(P<0.001). On taking the tissue LCFA specific activity into account and assuming the fate of the carboxyl-carbon of injected tracer [1-14C]-C16:0 tracks that of endogenous LCFA carboxyl-carbon, it was estimated that the quantity of LCFA carboxyl-carbon present in BAT water-soluble metabolites at the time of harvest was 1.7-fold higher in samples from cold-challenged animals (Fig. 8B
).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
and brown fat-inducible thioesterase (BFIT) (28)
. The current study focused on metabolically relevant genes in murine BAT, with temperature-dependent expression patterns used to generate a predictive model of tissue metabolic status and regulation of BAT intermediary metabolism. This integrative approach based on mRNA abundance changes is somewhat analogous to the use of classic ‘crossover’ analysis, which examines alterations in tissue metabolite levels to model metabolic control loci. The initial gene expression patterns observed via GeneCalling QEA led to a prediction of coordinated up-regulation of glucose uptake, LCFA synthesis, and LCFA combustion in the BAT from cold-challenged mice, a model further solidified through patterns emerging from follow-up experiments using more quantitative real-time RT-PCR analyses. The current studies focused on gene expression changes in whole BAT and were not designed to dissect expression patterns specific to each of the various cell types comprising the tissue. Hence, certain gene expression changes we observed in response to ambient temperature and nutrition may have been manifested in cells other than mature brown adipocytes. Nevertheless, metabolic predictions made through analysis of whole BAT mRNA were confirmed through experiments examining whole BAT metabolism in vivo.
Glucose uptake and catabolism
It is well established that cold exposure and sympathetic activation stimulate the uptake of glucose into rodent BAT (5
, 6
, 9
10
11
12
, 29
, 30)
, consistent in the current model in which 2-DOG uptake into BAT was strongly induced (Fig. 7A
). This phenomenon is likely driven to some extent simply through the enhanced metabolic rate of the tissue and hence a greater flux rate of metabolizable substrates. However, a greater degree of complexity is evident from the current study, which illustrates that enhancement of the expression (and presumably translation) of some genes involved with glucose uptake and glycolysis supports the cold-stimulated utilization of glucose by BAT (Table 2
and Fig. 2
). Such findings generally concur with BAT gene expression data reported by Daikoku et al. (21)
. Expression of genes encoding enzymes associated with glucose utilization was not altered in BAT under thermoneutral warm conditions (Fig. 2A, B
), suggesting that these pathways are not down-regulated in this tissue under this condition. Changes in gene expression may augment more acute events such as movement of glucose transporters to the cell surface in response to insulin and/or catecholamine stimulation of BAT (6
, 12
, 29
, 31)
.
Interactions between signals derived from insulin, catecholamines, and LCFA are important in modulating BAT glucose uptake. For instance, cold exposure appears to enhance the insulin sensitivity of a variety of rodent tissues, including BAT (9
, 10)
, and increases the number of glucose transporters present at the cell surface (12)
. Although the molecular mechanisms underlying these events are not completely defined, intracellular signals emanating from sympathetic stimulation of BAT under cold conditions appear to be important. Intravenous norepinephrine infusion or artificial stimulation of the sympathetic nerves communicating with BAT, for example, significantly enhanced BAT glucose uptake in rats (11
, 12)
. Surgical denervation of these nerves attenuated the cold induction of glucose uptake in BAT (11
, 12
, 30)
. Furthermore, direct application of norepinephrine to isolated rodent brown adipocytes stimulated glucose uptake (5
, 6
, 29)
. Signals associated with BAT LCFA metabolism appear critical to this process. First, blockade of brown adipocyte CPT1 using pharmaceutical antagonism in vitro prevents the stimulation of glucose uptake (5
, 6)
and metabolic rate (4
, 5)
by norepinephrine. Second, exposure of brown adipocytes to LCFA independently stimulates glucose uptake (5
, 6)
and cellular respiration (4
5
6)
. It remains to be shown whether the primary glucose uptake effects of LCFA provision or CPT1 blockade in BAT are due to changes in mitochondrial LCFA combustion/thermogenesis or are associated with signaling events tied to alterations in the cellular LCFA/LCFA-CoA profile.
It is notable that the stimulation of glucose uptake in BAT by insulin, norepinephrine, or LCFA is antagonized in the presence of cytochalasin B, indicating their effects on glucose uptake are via glucose transporter activity (6)
. Our gene expression data indicate a transient but marked diminution of mRNAs corresponding to GLUT1, GLUT2, and GLUT3 immediately after cold exposure, with increased expression of GLUT4 over time in the cold (Fig. 2B
). The latter event corroborates results seen in the rat (21
, 30)
. Thus, it is postulated that transporter-dependent glucose uptake in the early hours of cold exposure in our model would be explained largely by the activity of cell-surface GLUT4, whereas by 48 h multiple transporters may be engaged to support elevated glucose flux into mouse BAT.
Lipogenesis
Patterns of gene expression were consistent with a prediction of elevated lipogenesis in the BAT of cold-challenged mice, with apparent increases in expression of many genes involved with this pathway (Table 2
and Fig. 3
). Robust de novo lipogenesis in rodent BAT is known to occur and has been reported to increase in response to short- or long-term cold exposure (13
14
15
16
17
18)
. In fact, cold induced a large increase in BAT de novo lipogenesis in our model (Fig. 7B
), an event likely supported by coordinated up-regulation of lipogenic genes such as ATP-citrate lyase and ACC1. Furthermore, a general down-regulation of expression of lipogenic genes (Fig. 3
; see Fig. 6
for GPAT) was observed under warm conditions (Fig. 3)
, when BAT lipogenesis has been reported to be attenuated in mice (17)
. Although likely contributing to determination of the metabolic state of cold-challenged BAT, regulation beyond changes in mRNA levels for genes encoding lipogenic proteins must occur. For example, a delayed cold-induced rise in FAS and GPAT enzyme activities in BAT has been observed in mice (18)
and rats (19
, 20)
, but FAS and GPAT mRNAs were not increased in the cold by quantitative RT-PCR analysis in our mice (Fig. 3
; see Fig. 6
for GPAT). Furthermore, phosphorylation via AMP-kinase is known to powerfully inhibit, whereas citrate activates, ACC1 enzyme activity.
Our confirmation of a cold-related increase in de novo lipogenesis and coordinately elevated lipogenic gene expression in mouse BAT is consistent with the literature, but our conclusions differ from those expressed by Daikoku et al. (21)
. Based on changes in mRNA levels, the authors speculated that BAT lipogenesis drops in response to 48 h cold exposure in rats. For instance, they reported a lack of induction of FAS or ACC1 mRNA levels, with diminished ACC2 expression. However, it appears that compared with ACC1, ACC2 activity plays a negligible role in lipogenesis (32)
. A close inspection of the data indicates that BAT ACC1 mRNA was in fact increased by cold in their study (21)
.
Regulation of ß-oxidation in BAT
Prevalent views of BAT metabolism emphasize the importance of LCFA availability to support thermogenesis in response to cold exposure (1
, 2)
, supported by at least three main lines of evidence. First, activation of lipolysis in BAT in response to sympathetic nervous signals is a well-described event, illustrated by the rapid drop in BAT triglyceride stores with acute cold exposure in rodents (14)
. Consistent with this scenario, we observed that BAT mRNA encoding MG-lipase increased significantly in the cold (Table 2
and Fig. 4
). Second, in vitro manipulations of LCFA availability in BAT strongly affect tissue metabolic rate and glucose uptake. The cold-induced increase in BAT mRNAs encoding LPL (Fig. 4
; also ref 21
) and certain fatty acid transport proteins (21)
may facilitate a continued availability of LCFA from the circulation and from cellular stores. Third, LCFA stimulate uncoupling via UCP1, though the mechanism of this effect remains open to debate (33)
. Detailed mechanisms by which LCFA availability supports the thermogenic status of BAT and the means by which LCFA combustion is regulated in this tissue remain to be clarified. To our knowledge, no in vivo assessment of LCFA ß-oxidation in BAT has been reported.
Consistent with the hypothesis that LCFA metabolism plays a central role in support of BAT thermogenesis, many genes encoding proteins involved with LCFA combustion were increased in BAT upon cold exposure (Table 2
and Fig. 4
). Particularly notable is the cold induction of the levels of M-CPT1 and CAC mRNAs whose translation would act to facilitate LCFA import into mitochondria. MCPT1 (and possibly CAC) mRNA levels were also increased in the BAT of cold-challenged rats (21)
. Although the function of ACS5 remains to be established, it is notable that mRNA for this enzyme in murine BAT was induced significantly in the cold. If protein translation occurred in parallel, this finding supports the idea that a rise in BAT ACS5 activity plays a part in adaptive thermogenesis and BAT fatty acid metabolism. Follow-up studies indicated a greater enrichment of LCFA metabolites in the water-soluble fraction of BAT from cold-challenged mice in vivo (Fig. 8)
, consistent with accelerated ß-oxidation in this tissue under these conditions.
Simultaneous induction of de novo lipogenesis and LCFA combustion in the BAT of cold-exposed rodents presents a fascinating paradox with respect to metabolic regulation. Current concepts regarding control of mammalian ß-oxidation center on the critical role of CPT1, whose activity modulates entry of LCFA into mitochondria for subsequent catabolism to CO2 and chain-shortened metabolites. CPT1 activity (and hence LCFA combustion) is inhibited by malonyl-CoA, the product of ACC activity (34)
. In rodent liver, for instance, it is clear that a concomitant increase in malonyl-CoA generation upon induction of lipogenesis participates in the attenuation of LCFA ß-oxidation, whereas diminished lipogenesis and lower malonyl-CoA content enable robust ß-oxidation and ketogenesis (34)
. Our results in BAT indicate that these events are not mutually exclusive in this tissue, a point noted by Nicholls and Locke in an early review of BAT biology (1)
and subsequently considered by Buckley and Rath (18)
.
We suspect that coordinated induction of lipogenesis and LCFA ß-oxidation in thermogenic BAT may be explained through differential regulation of separate malonyl-CoA pools in the brown adipocyte. At least two extramitochondrial malonyl-CoA pools likely exist in many cells: a lipogenic pool generated by ACC1 activity and an ‘inhibitory’ pool generated by ACC2 (see ref 32
). Recent evidence from ACC2 knockout mice strongly supports this assertion in that LCFA ß-oxidation is activated in these mice with no effect on hepatic lipogenesis (32)
. Thus, we postulate that in mouse BAT, maintenance of the ACC1-generated lipogenic malonyl-CoA pool is facilitated through increases in gene expression and activities of ACC1 and other lipogenic proteins (e.g., Fig. 3
). In parallel, BAT LCFA combustion may be maintained in the cold via regulation of the ACC2-generated malonyl-CoA pool at levels yielding submaximal inhibition of CPT1. Consistent with this hypothesis, total BAT malonyl-CoA content was reported to drop by 50% in 48 h cold-exposed adult mice (18)
, albeit changes in its level within specific pools are not known. We have observed that mRNA encoding MCD, an enzyme with malonyl-CoA degrading activity, is significantly induced by cold exposure in BAT (Fig. 4)
. Should MCD activity and protein mimic mRNA changes, our results indicate that modulation of the inhibitory malonyl-CoA pool via MCD activity may support the thermogenic character of BAT. This simple view must be tempered, however, by the added complexity introduced by observations of abundant ACC2 mRNA (ref 21
and this study), protein (35)
, and a cold-induced increase in ACC2 mRNA observed in rodent BAT (Fig. 3
; in contrast to a drop reported in rats, see ref 21
). Parallel induction of MCD and ACC2 mRNAs in the cold may point to amplified turnover of the inhibitory pool of malonyl-CoA in BAT. Indeed, a short half-life (<20 s) for total cellular malonyl-CoA in liver has been reported (35a)
, but this parameter has not been measured in BAT. Finally, the potential effects of MCD phosphorylation on its enzyme activity (36)
in BAT will also require additional research.
Gene regulation
Remarkable patterns of coordinated BAT gene expression became evident from mathematical clustering of those genes assayed by quantitative real-time RT-PCR (Fig. 6)
, indicative of shared regulatory elements within the various cohorts of genes. For instance, polyunsaturated fatty acids (PUFAs) and PUFA derivatives are known to depress the expression of the SCD-1 and FAS genes, likely via interaction with PUFA response elements and diminution of sterol regulatory binding protein (SREBP1) binding in the promoter regions of the genes (37
, 38)
. Based on clustering results, similar elements in the promoter of GPAT would be expected to exist, and it may be postulated that under warm conditions, genes in this cohort are down-regulated through increased PUFA/PUFA derivative availability and/or a lowering of active SREBP1 in BAT. PUFA responsiveness may also be present in the SCD2 and ATP-citrate lyase promoters (37
, 39)
, and the latter gene is known to respond positively to insulin and glucose in WAT and liver (39)
. Thus, by extension, the promoters of BFIT, ACC2, and MG-lipase likely contain regulatory elements with high similarity to those in SCD2 and ATP-citrate lyase. Induction of these genes in BAT by cold exposure may indicate that under these conditions, repression by PUFA/PUFA-derivatives is lifted, elevated glucose metabolism activates carbohydrate-responsive elements in the genes, and/or insulin-derived signals are increased. Certainly, assertions as to common trans-activating factors and promoter elements within a particular clustering cohort require experimental confirmation. However, this experiment highlights the value of cluster analysis to generate hypotheses regarding regulatory components affecting the expression of genes whose promoters are not defined vis a vis BFIT, GLUT8, and others.
Gene regulatory pathways involving PPARs may be involved with the changes observed in gene expression under different thermal conditions. PPAR, for instance, has been shown to interact with PPAR coactivator 1 (PGC1) to induce some genes (M-CPT1, MCAD, LCAD) involved with LCFA ß-oxidation in 3T3-L1 preadipocytes (40)
, each of which displayed cold induction in BAT (Table 2
and Fig. 4
). mRNA abundance for PPAR and PPAR is diminished by cold in mice (Fig. 5)
and rats (41)
, indicating that PPAR mRNA levels alone are unlikely to explain PPAR-mediated effects in this tissue. Elevated sympathetic tone in BAT during cold exposure is known to facilitate induction of certain genes including UCP1 and likely supports metabolic gene regulation through the PGC1 pathway (42)
. One ramification of increased sympathetic stimulation is an apparent down-regulation of ß2 (43)
and ß3 (43
, 44)
-adrenergic receptors, with little effect on the ß1-adrenergic receptor profile (43
, 44)
. Consistent with this, we observed diminished mRNAs encoding ß2-AR and ß3-AR in the BAT of cold-challenged mice, with no effect of cold on ß1-AR mRNA abundance (Fig. 5)
. The cause of diminution of ß1-AR and ß2-AR mRNAs by food restriction/meal feeding and of the drop in ß3-AR mRNA on warm exposure are not known.
It is notable that the BAT expression of many genes associated with carbohydrate metabolism and lipogenesis responded to the food restriction/meal feeding regimen in a manner similar to cold exposure (Table 2
; Figs. 2
, 3
, and 6
). This indicates that at least some common regulatory factors are shared under these conditions. It may be argued that at the time of tissue collection in our food-restricted group/meal-fed group, BAT metabolism may be in a lipogenic state, analogous to the BAT from cold-exposed mice and similar to what is observed in the liver upon refeeding after a period of starvation. In addition, chronically meal-fed mice display daily torpor; anticipatory arousal and activated BAT thermogenesis occur just before the presentation of the daily meal (45)
. Thus, similarities in gene expression patterns between meal-fed and cold-exposed mice in the current experiments may be related to some degree of adaptive thermogenesis in the BAT of our meal-fed mice at the time of tissue sampling.
|
SUMMARY AND CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
) may apply in BAT. Other postulated cold-related energy-consuming events such as increased turnover of malonyl-CoA via elevated ACC2/MCD activities are worthy of further study. The current series of experiments underscore the remarkable ability of BAT to amplify metabolic pathways in response to cold exposure and points to unique mechanisms by which LCFA combustion and synthesis occur in parallel to support the established UCP1-mediated heat-generating futile cycle.
|
ACKNOWLEDGMENTS |
|---|
Received for publication July 10, 2001.
Revision received October 10, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
3')
