Cellular program controlling the recovery of adipose tissue mass: An in vivo imaging approach

Abstract

The cellular program responsible for the restoration of adipose tissue mass after weight loss is largely uncharacterized. Leptin mRNA levels are highly correlated with adipose tissue mass, and leptin expression can thus be used as a surrogate for changes in the amount of adipose tissue. To further study the responses of adipocytes to changes in weight, we created a transgenic mouse expressing the luciferase reporter gene under the control of leptin regulatory sequences, which allows noninvasive imaging of the leptin expression of mice in vivo. We used these animals to show that weight loss induced by fasting or leptin treatment results in the retention of lipid-depleted adipocytes in adipose depots. To further study the cellular response to weight regain after leptin treatment, a leptin withdrawal protocol was used to induce a state of acute leptin deficiency in wild type mice. Acute leptin deficiency led to the transient deposition of large amounts of glycogen within pre-existing, lipid-depleted adipocytes. This was followed by rapid reaccumulation of lipid. Transcriptional profiling revealed that this cellular response was associated with induction of mRNAs for the entire pathway of enzymes necessary to convert glucose into acetyl-CoA and glycerol, key substrates for the synthesis of triglycerides.

Obesity is an important public health problem that is characterized by an excess of adipose tissue (1, 2). Adipose tissue mass is under homeostatic control and the depletion of adipose tissue by dieting, starvation, or leptin treatment results in a set of behavioral and metabolic responses that act with great potency to replete adipose tissue when the weight-reducing stimulus is removed (2–4). However, little is known about the molecular mechanisms in adipose tissue that control the repletion of adipose tissue mass when, for example, fasted or leptin-treated animals are re-fed.

To follow the fate of adipocytes before and after weight loss, we developed a BAC transgenic mouse strain expressing a luciferase reporter gene under the control of adipocyte-specific leptin regulatory sequences (5). Because the leptin gene is highly correlated with the amount of lipid stored, this construct allows us to use in vivo imaging to follow adipocytes (6–9). We demonstrate, in combination with standard histological methods, that substantial weight loss results in the depletion of lipid from adipocytes that continue to express adipocyte markers and that remain resident in the former sites of adipose tissue depots. Refeeding of fasted animals or cessation of leptin treatment is associated with the deposition of lipid in these lipid-depleted adipocytes within 6 h, by which time leptin gene expression returns to wild type levels. Having established this, we used a leptin withdrawal protocol to study the cellular changes and transcriptional program that is activated as weight is regained (3). We show that the repletion of lipid in adipocytes is associated with a coordinated induction of glycolytic genes and further suggest that fat is synthesized via a glycogen intermediate. These data have important implications for our understanding of the cellular mechanisms that regulate adipose tissue mass. These studies also provide a new transgenic mouse strain that can be used to image leptin expression in vivo in a noninvasive manner.

Results

Generation and Characterization of Leptin-Luciferase Transgenic Animals.

To follow the fate of adipose tissue under different conditions, we first generated a transgenic animal, in which luciferase is expressed under the control of leptin gene regulatory sequences. Leptin-luciferase transgenic animals were generated using a BAC, RP24-69D4, spanning leptin gene. This BAC contains 22 kb of 5′ sequence and 150 kb of 3′ sequence relative to the leptin transcription start site (Fig. 1A). The firefly luciferase-polyA (Promega) was inserted into the ATG site of the leptin BAC via homologous recombination, and transgenic mice were generated by pronuclear injection of the modified BAC. Tissue specific expression of luciferase in leptin-luciferase transgenic animals was confirmed using biochemical assays from lysed tissues and in vivo imaging. Luciferase activity, normalized to the amount of total protein, was present specifically in adipose tissue at high levels, with undetectable levels in other tissues, except for intestine (see Fig. 1A). It is not clear whether the small signal of luciferase expressed in this tissue is in epithelial cells or in resident adipocytes.

Fig. 1.

Characterization of leptin-luciferase animals. (A) (Upper) Expression of the luciferase protein in tissues from leptin-luciferase mouse. Luciferase activity was normalized by the amount of protein of indicated tissue lysate. (mean ± SEM) The right panel displays in vivo imaging of a representative leptin-luciferase transgenic animal showing adipose-specific luciferase activity. (Lower) A schematic map of the leptin BAC RP24-69D4. Firefly luciferase is inserted in the translation start site in the second exon. (B) Luciferase activity recapitulates regulation of leptin. The relative expression of luciferase versus leptin was compared among fed, fasted, and ob/ob animals. (Lower) Endogenous leptin expression levels from fasted, fed, and ob/ob animals were measured by Taqman real time PCR normalized to cyclophilin levels. (Upper) Luciferase activity of adipose tissue lysates from fed, fasted, and ob/ob animals normalized to protein content (mean ± SEM). (C) In vivo imaging of fed, fasted, and ob/ob transgenic animals. The graph on the left shows quantitation of in vivo luciferase activity using Living Image 3.0 software (mean ± SEM).

To confirm that the leptin-luciferase transgene was correctly regulated quantitatively, we next tested whether luciferase expression was correlated with leptin mRNA levels in obese and lean animals. Leptin levels are highly correlated with cellular lipid content and its mRNA levels can vary hundreds of folds among fasted versus obese animals (6). Relative expression of luciferase versus leptin was compared among fed, fasted, and ob/ob visceral fat pads. There was a 100-fold greater level of luciferase activity in extracts from ob/ob adipose tissue versus extracts from fasted animals, which fully recapitulated the mRNA levels for endogenous leptin expression (Fig. 1B). These results were also confirmed in vivo using a CCD camera to monitor luciferase activity. The data generated using in vivo imaging was highly correlative to the biochemical analysis (Fig. 1C). We further assayed luciferase expression in mice placed on a high-fat diet. In these animals, leptin mRNA was also highly correlated with luciferase levels, both in comparisons between groups and among individual animals within the high-fat fed and chow-fed animals. [supporting information (SI) Fig. S1].

Fasting and Refeeding Induce Dynamic Responses in Leptin-Luciferase Expression Levels in Adipocytes.

The level of luciferase was monitored in animals during a 3-day fast and after refeeding over time. Luciferase activity fell gradually over the course of the fast and by 3 days, there was an approximately six- to eightfold lower level of luciferase activity (Fig. 2A). The activity returned to baseline by 6 h after refeeding. Because 6 h is not enough time to regenerate the adipocyte population, this extremely rapid recovery of the leptin-luciferase signal after refeeding suggested the possibility that lipid-depleted adipocytes persist in adipose tissue after fasting.

Fig. 2.

Effects of fasting on adipose tissue and luciferase expression. (A) Time course luciferase activity in leptin-luciferase animals upon fasting and refeeding. The level of luciferase in animals was monitored during a 3-day fast and after refeeding. The data were collected for the same animal over time (n = 4). Luciferase activity fell gradually over the course of the fast. After refeeding, luciferase activity returned to baseline by 6 h. (Upper) Representative time-course images of fasted and re-fed animals are shown. (Lower) Quantitation of in vivo signals by Living Image 3.0 software (mean ± SEM). (B) Histological and immunohistochemical analysis of adipose tissue of fasted mouse. H&E staining of fed (Left) and fasted (Center) animals for comparison. Immunohistochemical staining of the adipocyte specific protein, aP2, in fasted animals (Right).

Consistent with this, histological analysis at the conclusion of the 3-day fast showed smaller adipocytes with a markedly lower cellular lipid content (Fig. 2B). The identity of these cells as adipocytes was established by immunohistochemical staining with an antibody to aP2, an adipocyte-specific marker (see Fig. 2B). However, we noted that there was residual lipid in some of the adipocytes after the fast, evident by significant luciferase activity, and thus sought to identify conditions under which there was near complete depletion of cellular lipid in adipocytes. We accomplished this by treating animals with leptin.

Leptin Treatment Induces near Complete Delipidation of Adipocytes, Conserving Adipocyte-Specific Protein Expression.

Mice were treated with high-dose leptin via s.c. pump for 8 days. Leptin treatment led to a depletion of all visible body fat, which was associated with the complete absence of luciferase expression (Fig. 3A). Histology of the adipose depot after 8 days of leptin treatment showed a dense population of cells with little or no detectable cellular lipid (Fig. 3B). However, these lipid-depleted cells still showed high levels of the adipocyte-specific protein aP2 expression, suggesting that after leptin treatment the adipose tissue was populated by lipid-depleted adipocytes (see Fig. 3B).

Fig. 3.

Adipose tissue response to leptin treatment and withdrawal. (A) In vivo imaging of leptin-luciferase animals upon leptin withdrawal. PBS (n = 6) or leptin at 2.5 μg/h (n = 5) was administered for 8 days by subcutaneously implanted osmotic pump. On day 8, leptin or PBS treatment was withdrawn by removing pumps under inhaled anesthesia. Animals were imaged before treatment, after 8 days of leptin treatment and daily for the 4 days following removal of leptin pumps. (Upper) Imaging results of representative animals form PBS and leptin-treated groups are shown. (Lower) Quantitation of in vivo signals by Living Image 3.0 software indicates a dramatic suppression of luciferase by exogenous leptin treatment (mean ± SEM). (B) Preservation of adipocyte identity during chronic leptin treatment. Leptin treatment results in near complete depletion of triglyceride droplets from white adipose tissue. White adipose tissue of mice treated for 8 days with leptin consists of dense islands of cells, as assessed by hematoxylin and eosin staining (Center). An identical magnification of control, PBS-treated white adipose tissue shows characteristic large, central lipid droplets and peripheral nuclei (Left). Note for comparison that nuclei are the same size. Immunohistochemical staining demonstrated high levels of the adipocyte-specific protein aP2 in the cytoplasm of high numbers of cells in delipidated white adipose tissue (Right, brown, peroxidase-positive material).

After 8 days, the leptin osmotic pumps were removed. Following withdrawal of exogenous leptin, animals fed ad libitum became markedly hyperphagic and body fat mass returned to pretreatment levels over the course of 4 days (Fig. S2). Luciferase levels were clearly detectable in animals after leptin withdrawal after 1 day; and after 4 days the luciferase signal returned to baseline levels (see Fig. 3A, bottom panel).

The data from leptin treated animals were consistent with data from fasted animals and suggested that under both conditions the adipose depot is composed of lipid-depleted adipocytes. The data further suggest that leptin treatment leads to the complete depletion of cellular lipid, providing a unique opportunity to study the molecular programs that are responsible for the repletion of cellular lipid.

Recovery from Hypoleptinemia Increases Cellular Glycogen Before the Reappearance of Cellular Lipid.

To study the cellular programs activated when fat mass is repleted, we prepared RNA from adipose tissue during leptin treatment, and after 1 to 4 days after leptin withdrawal for use with oligonucleotide microarrays to follow the patterns of gene expression after leptin withdrawal. While preparing the RNA from these samples, we noted the appearance of large amounts of a white flocculent material after ethanol precipitation. This material had the gross appearance of glycogen. To confirm this, we measured tissue glycogen levels biochemically. On the first day after leptin withdrawal there was a massive increase in glycogen content in the adipose depot, with a total glycogen content of 1% of the total wet weight of the tissue (Fig. 4A). Treatment of acutely leptin-deficient white adipose tissue lysates with amyloglucosidase (10) liberated a large amount of soluble glucose, confirming the presence of high levels of glycogen. Periodic acid-schiff (PAS) staining of tissue sections showed dense staining in cells containing small lipid droplets, confirming the presence of glycogen granules in adipocytes (see Fig. 3B). The PAS-stained material was further confirmed to be glycogen by predigesting tissue sections with α-amylase, which eliminated PAS-positive granules (Fig. 4B). Intense staining was visualized in a similar pattern to that shown previously for aP2 (see Fig. 4B). The total glycogen content decreased over the ensuing 3 days, by which time adipose tissue mass had returned to the same level as before treatment. Histological analysis over this period indicated enlarging adipocyte lipid droplets in the same cells that showed PAS staining for glycogen (see Fig. 4B).

Fig. 4.

Accumulation of white adipose tissue glycogen during acute leptin deficiency. (A) White adipose tissue glycogen content rose 30- to 60-fold in leptin withdrawal animals (green and red bars) relative to PBS controls (blue bars) following withdrawal of exogenous leptin treatment, as measured by glucose liberated by amyloglucosidase (*, P < 0.05 vs. PBS; #, P < 0.065 vs. PBS). This increase occurred in both free-fed animals (Leptin-FF, green bars) and animals maintained at normocaloric intake levels (Leptin-NC, red bars), and increased glycogen content persisted for 3 and 4 days in these groups, respectively. (B) PAS staining of white adipose tissue on withdrawal day 1 and withdrawal day 3. PAS staining on withdrawal day 1 indicated large amounts of cytoplasmic glycogen content in leptin-FF and leptin-NC tissue sections (black arrows). This contrasted sharply with control PBS-treated white adipose tissue. On withdrawal day 3, glycogen accumulation decreased in parallel to the accumulation of lipid (red arrows). α-Amylase pretreatment of tissue sections eliminated PAS-positive granules in adipocyte cytoplasm, indicating specificity of staining for glycogen.

Controlling Food Intake Does Not Prevent Massive Accumulation of Adipocyte Glycogen.

Because animals became markedly hyperphagic and hyperinsulinemic after leptin withdrawal, we also assessed the possible contribution of hyperinsulinemia to the accumulation of glycogen in adipocytes by limiting food intake to pretreatment levels coincident with leptin withdrawal. We refer to these animals as normocaloric (NC) (Fig. S2). The free-fed hyperinsulinemic animals are referred to as FF. Under the NC conditions, where hyperphagia was prevented and insulin levels remained well below those of control animals (see Fig. S2D), there was still a massive accumulation of glycogen in the adipose as shown by PAS staining and treatment of sections with amylase (see Fig. 4B). Quantitatively, the amount of glycogen was similar to that seen in FF animals following withdrawal of leptin treatment (see Fig. 4B). In addition, glycogen persisted for relatively longer periods in the normocaloric versus the free-fed animals. This was associated with a slower time course for the repletion of adipocyte lipid in the NC animals, adding further evidence that increased cellular glycogen provides substrate for the synthesis of triglycerides in adipocytes after withdrawal of leptin treatment (see Fig. 4A).

The Transcriptional Program Associated with Recovery of Adipose Tissue Mass.

These data suggested that the repletion of cellular lipid content and adipose tissue is achieved via a glycogen intermediate in lipid-depleted adipocytes. To study this further, transcription profiles of adipose tissue were generated by analyzing RNAs at 6 or 8 days of leptin treatment and 1, 2, 3, and 4 days after leptin withdrawal. To refine the analysis, RNA was prepared both from hyperphagic animals fed ad libitum after leptin withdrawal (i.e., leptin-FF), as well as from animals whose food intake was restricted to that which the animals ate voluntarily before leptin withdrawal (i.e., leptin-NC animals; defined above). As an additional control, tissues from PBS-treated animals were analyzed both before (day 6) and after (day 1 after withdrawal) removal of the osmotic pumps. In all cases, RNA was labeled using biotin and hybridized to Affymetrix 11K oligonucleotide microarrays.

Selection and k-means cluster analysis of 683 genes changing significantly across experimental conditions identified 14 statistically distinct patterns of gene expression (Fig. S3) (11). In 8 of the 14 clusters, genes were induced (four clusters) or repressed (four clusters) in the leptin-FF group, the leptin-NC group, or in both (see Figs. S2 and S3). The six additional clusters corresponded to genes that were specifically induced or repressed by chronic leptin treatment with levels returning approximately to normal after leptin withdrawal. An additional three clusters of genes were induced only during acute leptin deficiency to approximately equivalent levels in both leptin-FF and leptin-NC mice. The fold-change data for all genes assessed by microarray analysis relative to control conditions are available in the supplemental data.

Analysis of these three clusters induced specifically by acute leptin deficiency revealed an enrichment of genes associated with gene ontology (GO) terms related to carbohydrate metabolism. Specifically, many genes induced in adipose tissue during weight gain play a role in glucose uptake, glycolysis, and acetyl-CoA and glycerol synthesis. An acute decrease in circulating leptin level induced nearly the entire complement of genes necessary to increase simple sugar flux into the adipocyte, and subsequently convert these sugars into cytosolic acetyl-CoA and glycerol for de novo fatty acid, triglyceride, and cholesterol biosynthesis (Fig. 5A). Average expression level for these mRNAs peaked on day 1 of acute leptin deficiency but remained elevated for 4 days following cessation of leptin treatment (Fig. 5B). Leptin withdrawal also induced the expression of genes required for the storage of cytoplasmic triglycerides, including genes involved in the synthesis of glycerol, such as glycerol-3-phosphate dehydrogenase mRNA (3- and 2.8-fold in leptin-FF and leptin-NC mice, respectively). Triglyceride storage is also augmented during acute leptin deficiency by a 70% and 30% induction of glycerol-3-phosphate acyltransferase in leptin-FF and leptin-NC animals, respectively.

Fig. 5.

Pathway of acetyl-CoA and glycerol generating enzymes up-regulated by acute leptin deficiency. (A) Acute leptin deficiency induced the pathway of genes (blue ovals) necessary to synthesize cytoplasmic acetyl-CoA and glycerol from simple sugar precursors. Fold change for each gene on day 1 following leptin withdrawal is shown below each enzyme name for leptin-FF (Left) and leptin-NC (Right) mice with the associated P value immediately below each fold-change value. Potential sources of carbon for acetate or glycerol synthesis are in blue text. BP, bisphosphate; CoA, CoA; DH, dehydrognease; DHAP, dihydroxyacetone phosphate; G, glucose; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; M, mannose; P, phosphate; PFK, phosphofructokinase; TPI, triose phosphate isomerase. (B) Average expression level for the genes in the cluster is shown in graphical format. Leptin-FF (green circles) and leptin-NC (red triangles) groups showed approximately equivalent induction relative to PBS controls (blue squares) following leptin withdrawal. Levels remained elevated for 4 days following cessation of exogenous leptin treatment.

Acute leptin deficiency specifically repressed four groups of genes that were induced or unchanged during chronic leptin treatment (see Figs. S3 and S4). These genes included both carnitine-palmitoyl transferase-I (CPT-I) and peroxisomal acyl-CoA oxidase (ACO), both of which are required for fatty acid oxidation. CPT-I, which enhances mitochondrial uptake of fatty acids for oxidation, is decreased in leptin-FF and leptin-NC mice 3.0 and 1.9-fold, respectively. ACO, which catalyzes enzymatic fatty acid breakdown, is 2.7-fold repressed in leptin-FF mice and 1.3-fold repressed in leptin-NC mice. White adipose phosphoenolpyruvate carboxykinase mRNA, which would detract from acetyl-CoA accumulation, was repressed 3.6- and 2.4-fold in leptin-FF and leptin-NC animals, respectively.

Leptin withdrawal did not regulate the mRNA of genes involved in glycogen synthesis or glycogenolysis, suggesting that the induction of glycogen synthesis is not regulated at the level of transcription. As discussed below, these data suggest that the genes required for glycogen synthesis continue to be expressed in adipocytes in the setting of weight loss, and thus allow the accumulation of glycogen when food is consumed. Lipid synthesis genes are repressed after a fast and this glycogen then serves as substrate for the biosynthesis of triglyceride when the genes required for lipid synthesis are induced.

Discussion

Adipose tissue mass is controlled by a negative feedback loop comprised of leptin, a hormonal signal made by adipocytes, integratory centers in the brain and perhaps elsewhere, and a set of efferent pathways from the CNS that modulate peripheral metabolism (2, 9, 12, 13). Although much is now known about the CNS pathways that respond to leptin, less is known about the end organ effects of leptin, in particular its effects on adipose tissue mass in vivo (2, 9). Similarly, the cellular and molecular programs responsible for the repletion of adipose tissue mass after weight loss have not been fully elucidated (14). In this article, we have addressed these issues using molecular genetics, physiologic, and computational tools.

These studies made use of a unique BAC transgenic mouse line that expresses luciferase under the control of leptin promoter elements, making it possible to assess the activity of the leptin gene in vitro biochemically, and in vivo by imaging of luciferase activity (15). The leptin BAC clone that was used conferred fat-specific expression of leptin, and its quantitative expression was highly correlated with adipocyte lipid content in fasted, fed, high-fat fed, and ob/ob mice. Leptin-luciferase animals also showed an increase in reporter expression at 8 days postnatally, which recapitulated postnatal leptin surge (K.B., unpublished work) (16). All together, these data indicate that the DNA elements required for qualitative and quantitative expression of the leptin gene reside between −22 and + 150 kB of the transcription start site of the leptin gene. Because leptin levels are highly correlated with adipose tissue mass (6–9), the leptin-luciferase animal model also provides a unique means for assessing adipose tissue mass noninvasively in vivo. In this article, we have used this strain to study the response of adipose tissue before and after either food restriction or leptin treatment.

Leptin expression is markedly reduced after a 3-day fast and its level of expression is restored nearly to baseline levels within 6 h after refeeding. Repletion of leptin expression is extremely rapid, suggesting that there is unlikely to be de novo adipogenesis. Rather, these data suggested that food restriction results in the depletion of lipid within adipocytes. This conclusion is supported by data from immunohistochemistry of adipose tissue, showing that aP2 (17) is still expressed at high levels in adipose tissue after a fast, at a time when there are markedly reduced amounts of lipid. Eight days of leptin treatment caused an even more dramatic reduction of adiposity, and histological analysis of adipose tissue of leptin-treated animals showed dense nests of aP2-expressing cells that were lipid depleted. Data from the transcriptional profiles further indicated that the cells in the adipose tissue depot after fasting or leptin treatment continue to express RNAs for a number of adipocyte-specific/enriched mRNAs, such as peroxisome proliferator-activated receptor-gamma and CCAAT enhancer binding protein-alpha. In addition, mitoses were not visible in tissue sections from adipose tissue after leptin withdrawal during times when adipose mass is being rapidly repleted, and genes that are involved in the clonal expansion phases of adipogenesis in vitro remained unchanged or decreased in abundance during acute leptin deficiency. (Dataset S1) Previous studies have shown that adipose tissue from animals treated with leptin contains the same DNA content as control samples (18) and, in contrast to some other previous reports (19), we did not observe evidence for adipocyte apopotosis after leptin treatment (Dataset S1 and data not shown). We conclude that the restoration of fat mass after fasting or leptin treatment is the result of the activation of a program of lipid reaccumulation in lipid-depleted adipocytes and does not require de novo adipogenesis.

The reaccumulation of fat mass following weight loss is the result of the activation of a different gene-expression program that is activated by acute hypoleptinemia. The initial phase of this cellular program is marked by an increase in adipocyte glycogen content. The glycogen content of white adipose tissue is normally exceedingly low (20) and previous studies have shown that fasting reduces glycogen levels even further, while subsequent refeeding transiently increases adipose tissue glycogen content (21–23). Because glycogen content increases after leptin withdrawal, even in the absence of hyperinsulinemia, we conclude that the observed increase in glycogen synthesis in white adipose tissue after refeeding is induced by low leptin levels (24).

The increased pool of glycogen in white adipose tissue during fat regeneration appears to provide the substrate for expansion of adipose triglyceride mass. This conclusion is consistent with previous findings showing that adipose triglyceride stores are primarily synthesized within adipose tissue, rather than being imported into fat (22). Adipose tissue from animals re-fed after a 48 h fast was previously shown to have a respiratory quotient of >1.0, indicative of a high rate of glucose metabolism and not fatty acids (23, 25). In re-fed animals, glycogen disappears from adipose tissue at the same time as fatty acid synthesis is maximal, and PAS-positive material and lipid droplets can be easily seen in the same cells during the recovery from weight loss (see Fig. 4B) (22). Finally, glycogen is an efficient precursor of both components of adipose tissue triglycerides, the fatty acid, and glycerol moieties (21). In aggregate, these data suggest that the reaccumulation of fat mass following leptin withdrawal, or re-feeding following a fast, involves (a) the increased glycogen synthesis, which is likely to be substrate driven because mRNAs for the genes involved in glycogen synthesis and breakdown do not fall during leptin treatment or subsequent withdrawal (26); and (b) transcriptional induction of genes required for lipid biosynthesis from monosaccharides. Glycogen synthesis is often regulated by changes in the sizes of substrate pools and is posttranslational of enzymatic modification, and not at the level of gene expression (26).

The observation that mRNAs involved in glycogen metabolism do not change after fasting or leptin treatment is in contrast to the enzymes regulating glycolysis and the enzymatic pathway that generates acetyl-CoA from simple sugars, which are markedly induced under these conditions. This increase in gene expression is in contrast to other conditions, in which these same genes are regulated by posttranscriptional mechanisms (11, 27). In addition, these enzymatic genes are not repressed by leptin treatment, indicating that the response to leptin withdrawal or refeeding is not the result of an induction of the same genes repressed during leptin treatment, and instead constitutes a unique gene expression program (28, 29).

In summary, we have developed a unique transgenic line that allows the noninvasive analysis of leptin expression, a robust surrogate for adipose tissue mass. We have used this line to study the mechanisms by which adipose tissue mass is recovered after weight loss and show that weight loss by fasting or leptin treatment results in the persistence of lipid-depleted adipocytes in the adipose depot. We further show that hypoleptinemia stimulates the recovery of adipose tissue mass in part by activating a unique cellular and gene expression program that facilitates the conversion of a glycogen intermediate into triglycerides.

Methods

Animal Experiments.

Individually caged 7- to 8-week-old C57BL/J6 female mice were purchased from the Jackson Laboratory. PBS (PBS control group) or leptin at 2.5 μg/h (Leptin-FF and Leptin-NC groups) was administered for 8 days by subcutaneously implanted osmotic pumps (Alzet). Serum leptin (R&D Systems) and insulin (Crystal Chem) were measured by ELISA according to the manufacturers' protocols.

BAC Modifications and in Vivo Imaging.

Modification of BAC number RP24-69D4 has been performed as described in ref. 30. Generation of transgenic mice following microinjection was based on standard techniques described by Hogan and colleagues (31). Luciferase activity in tissues was measured using Luciferase Assay Kit (Promega) and normalized by protein content. (Pierce BCA kit). In vivo imaging of transgenic animals were performed using the Xenogen IVIS Lumina imaging system (Caliper). Anesthetised animals were injected intraperitonally with luciferin (200 μl of stock 15 mg/ml in PBS). After 15 to 20 min, the animals were imaged in an imaging chamber and the photon image was analyzed by Living Image 3.0 software (Xenogen).

White Adipose Tissue Histology and Glycogen Quantitation.

For PAS staining, small pieces of tissue (50 mg) were incubated overnight in alcoholic fixative (90% ethanol, 10% formaldehyde). Tissue fragments were embedded in paraffin wax (Paraplast Plus, Fisher Scientific). Sections were PAS stained and hemaxoxylin counterstained according to standard protocol (Sigma). To demonstrate specificity of glycogen staining, several sections were preincubated for 5 min at 37°C in 5-mg/ml α-amylase (Sigma) after rehydration before PAS staining. Glycogen was quantitated with the amyloglucosidase method (10).

Leptin Withdrawal Oligonucleotide Microarray Analysis and k-Means Clustering.

Samples were prepared for Murine 11k microarrays from 10 μg of total RNA as outlined in the Affymetrix technical bulletin and as previously described (28). For cluster analysis, genes were included if greater than fourfold and >250 average difference change units (abundance measurement), or twofold and >500 average difference change units, increased or decreased in one time point relative to day 6 treated PBS control. Of the genes examined, 683 met these criteria and were clustered according to fold-change value using the modified k-means clustering algorithm with a dot product metric (28).