A New Role for Lipocalin Prostaglandin D Synthase in the Regulation of Brown Adipose Tissue Substrate Utilization

Mice heterozygous for a disruption in exons II–V in the L-PGDS gene were crossed to produce L-PGDS KO mice and wild-type controls (18,19). KO mice for the gene for peroxisome proliferator–activated receptor γ coactivator PGC1β were generated as previously described (20). KO mice for the RIP140 were generated as previously described (21).

Mice were housed at a density of four animals per cage in a temperature-controlled room (20–22°C) with 12-h light/dark cycles. All animal protocols used in this study were approved by the U.K. Home Office and the University of Cambridge. Animals were fed a normal chow diet (D12450B; Research Diets) or a high-fat diet; either 45% or 60% calories-from-fat diets (D12451, D12492; Research Diets). For studies of cold acclimation, animals were housed as described previously (22). These experiments were approved by the North Stockholm Animal Ethics Committee.

The brown fat cell line IMBAT-1 was generated (23) and differentiated, as described previously (24,25). The RIPKO-1 cell line was generated by continuous culture of Rip140-null mouse embryonic fibroblasts, and differentiation of the cells was performed as previously described (26). Differentiated IMBAT-1 or RIPKO-1 adipocytes were infected with adenovirus vectors expressing Rip140, PGC1α, or green fluorescent protein gene (26). Equal levels of infection were confirmed by fluorescence microscopy for the coexpressed green fluorescent protein.

Eicosanoids were extracted from 50 to 100 mg of BAT by solid-phase extraction and quantified by liquid chromatography-mass spectrometry as described previously (27).

Glycogen levels were determined by the anthrone method (28).

Lipids were extracted from 100 mg of homogenized liver by heating to 85°C in 1 mL of 5% NP-40 in water. Triglyceride (TG) levels in extracts were assessed by enzymatic assay (Biovision) according to manufacturer instructions.

Enzymatic assay kits were used for determination of plasma free fatty acids (Roche) and TGs (Sigma-Aldrich, St. Louis, MO). Enzyme-linked immunosorbent assay kits were used for measurements of leptin (R & D Systems), insulin (DRG; Diagnostics International Limited), and adiponectin (B-Bridge International) according to manufacturer instructions. A Minispec LF TD-NMR (Brucker Optics) was used to measure body composition.

The lipid extract was derivatized by acid-catalyzed esterification (29) with 10% BF₃ in methanol (Sigma-Aldrich, Gillingham, U.K.), and the esterification was performed at 80°C for 90 min. Once cool, water and hexane were added, and the aqueous layer was discarded. The organic layer was dried and then reconstituted in hexane for gas chromatography–flame ionization detection analysis (GC-FID) with a Trace GC Ultra (ThermoScientific). Derivatized organic samples were injected onto a 30-m × 0.25-mm 70% cyanopropyl polysilphenylene-siloxane 0.25 μm TR-FAME stationary-phase column (ThermoScientific). The initial column temperature was 55°C for 2 min, increased by 15°C/minute to 150°C and then increased at a rate of 4°C/minute to 230°C, where it was held for 30 s.

Norepinephrine-induced thermogenesis was measured through oxygen consumption, as determined by the INCA system (Somedic, Hörby, Sweden) as described previously (30).

Mice were fasted overnight from 4 p.m. until 9 a.m. the next day. Mice were injected intraperitoneally with either glucose (1 g/kg) or pyruvate (1.3 g/kg), and blood glucose levels were measured with a One-Touch Ultra glucose meter (Lifescan, Milpitas, CA). Blood samples for insulin measurement were collected with glass capillary tubes.

Samples for histology were placed in 10% buffered formalin overnight before transfer to 70% ethanol and later embedding in paraffin. Multiple sections were stained with hematoxylin and eosin for morphological analysis. Percentages of lipid content and adipocyte area were determined by morphometric analysis. The average lipid area in BAT and cell size in WAT per mouse were calculated from at least 3 separate sections. Analysis was carried out with CELL^P analysis software (Olympus, Southend-on-Sea, U.K.).

Total RNA was isolated from cells and BAT fractions with RNeasy Mini columns (Qiagen). RNA was isolated from ground tissues with STAT-60 reagent (TEL-TEST), followed by chloroform extraction and isopropanol precipitation. The cDNA was generated with M-MLV reverse transcriptase (Promega) according to the manufacturer instructions. Real-time PCR was carried out with TaqMan primers and probes or SYBR green reagent (Applied Biosystems) with an ABI Prism 7900 sequence detection system (Applied Biosystems) according to the manufacturer instructions. Data were normalized to 18s rRNA. Primer sequences are available on request.

Mice were fasted overnight. Mice were subsequently anesthetized with 90 mg/kg sodium pentobarbital. After 10–15 min to allow full sedation, the tail vein was cannulated. A basal glucose sample was taken, and mice were injected with 0.2 MBq of 2-[¹⁴C]-deoxyglucose (2-[¹⁴C]-DG) (Perkin Elmer) and 1 mg/kg norepinephrine bistartrate. Glucose was measured at 10, 20, and 30 min with an AlphaTRAK (Abbott) glucose meter calibrated for rodents. Blood samples (30 μL) were collected at 10 and 20 min from the tail vein for measurement of serum DPM. At 30 min, animals were exsanguinated by cardiac puncture. Tissues were collected and frozen on dry ice. The 2-[¹⁴C]-DG uptake into tissues was quantified with the BaOH₂/ZnSO₄ precipitation method as described previously (31).

Results were expressed as mean ± SEM. Statistical analysis was performed with SPSS 17.0 (IBM, Armonk, NY). Pairwise comparisons were carried out by Student t test. For data with more than two groups, ANOVA was carried out, followed by Tukey post hoc test. For groups with multiple groups and conditions, two-way ANOVA was carried out.

Two paradigms of BAT activation were investigated: 1) cold exposure and 2) high-fat feeding. L-PGDS expression was physiologically upregulated in the BAT of mice after either cold acclimation (4°C) or high-fat feeding (45% calories from fat) relative to chow-fed mice housed at room temperature (22°C) (Fig. 1A). Conversely, mice housed at thermoneutrality (30°C) for 3 weeks had decreased L-PGDS expression relative to room temperature–housed control mice (Fig. 1B). To identify putative transcriptional regulators of L-PGDS, we focused on key factors implicated in BAT activation. Both PGC1α and PGC1β have been shown to be positive regulators of BAT function, whereas RIP140 is a negative regulator of BAT function (20,32). L-PGDS was found to be down regulated in mice lacking PGC1α (Fig. 1C) or PGC1β (Fig. 1E) but upregulated in mice lacking Rip140 (Fig. 1D). Finally, and consistent with a metabolic role for L-PGDS, L-PGDS was found to be predominantly expressed in mature brown adipocytes rather than the stromal vascular fraction (Fig. 1F).

To investigate further the reciprocal regulation of L-PGDS by PGCs and RIP140, cellular models were used. In a BAT cell line, ectopic overexpression of PGC1α caused an increase in endogenous L-PGDS mRNA levels (Fig. 2A). Conversely, increasing expression of Rip140 led to a decrease in L-PGDS expression levels (Fig. 2B). Finally, with Rip140 KO embryonic fibroblasts, we showed that titrating RIP140 back into these cells resulted in a proportional decrease in L-PGDS expression (Fig. 2C).

To test whether L-PGDS was required for appropriate BAT function, L-PGDS KO mice were sequentially acclimated to either 4°C or thermoneutrality (30°C) for a period of 3 weeks. At each temperature, both basal and maximal thermogenic capacity were measured. L-PGDS KO mice exhibited no alterations in their norepinephrine-stimulated (maximal) rates of oxygen consumption; however, L-PGDS KO mice did exhibit a small reduction in their basal energy expenditure after housing at 4°C (Fig. 3A).

Although changes in energy expenditure in L-PGDS KO mice were modest, there was a large alteration in their respiratory exchange ratio (RER) after norepinephrine stimulation at both 4°C and 30°C. Under unnstimulated conditions, L-PGDS KO mice only exhibited an increase in RER when they had been previously acclimated to 4°C (Fig. 3B). The elevated RER of the L-PGDS KO mice relative to wild-type littermates suggests that L-PGDS KO mice met their increased demand for energy at 4°C by utilizing more carbohydrate than did wild-type mice.

After cold acclimation, L-PGDS KO mice had increased lipid content in BAT relative to wild-type mice (Fig. 3C). Lipids were extracted from wild-type and L-PGDS KO mice that had been acclimated to 4°C for 3 weeks and were analyzed by GC-FID. The lipid profile of the L-PGDS KO mice, when compared with wild-type controls, demonstrated a reduction in the proportion of dietary-derived essential fatty acids (C18:2 and C18:3), and an increase in the proportion of oleate (C18:1), which can be either derived from the diet or synthesized de novo (Fig. 3D). In accordance with elevated de novo lipogenesis, BAT from L-PGDS KO mice also had increased expression of the lipogenic genes for fatty acid synthase (FAS), stearoyl coenzyme A desaturase 1 (Scd1), and elongase 6 (Elovl6) (Fig. 4A). When compared with wild-type mice, L-PGDS KO mice had increased mRNA levels in BAT of the glucose transporter gene Glut4, α-enolase, and phosphofructokinase, genes involved in glucose uptake and glycolysis (Fig. 4B). Remarkably, the gene expression changes in both carbohydrate metabolism and fatty acid synthesis were highly specific. Although there was a small increase in UCP1 expression in BAT, there were no significant changes observed in genes regulating brown adipocyte function, fatty acid oxidation, lipid uptake, or brown adipocyte differentiation (Fig. 4C and Supplementary Fig. 1). Taken together, these data suggest that L-PGDS ablation results in active BAT that preferentially utilizes glucose, either directly as a substrate for thermogenesis or for de novo lipogenesis and subsequent β-oxidation.

Similar to the results observed in BAT, L-PGDS KO mice had increased levels of FAS, Elovl6, and Scd1 mRNA expression in liver when compared with wild-type mice (Fig. 4D–E). Despite these changes, no increase in hepatic TG level was detected (Supplementary Fig. 2A). Further to the changes in de novo lipogenesis, L-PGDS KO mice also exhibited a reduction in expression of the key gluconeogenic enzyme PEPCK (Fig. 4D), although no alterations in the glycogen biosynthetic program or glycogen levels were detected (Supplementary Fig. 2B–C). The change in PEPCK suggests that in the liver, products of the citric acid cycle were not driven toward production of glucose but instead were channeled to de novo lipogenesis. This hypothesis was also supported by the lower plasma glucose of cold-acclimated L-PGDS KO mice (Table 1) and by a small but significant reduction in glucose levels in response to a pyruvate tolerance test (Supplementary Fig. 2D).

The major sites of de novo lipogenesis in cold-acclimated mice are BAT and liver (14). Consistent with L-PGDS acting principally to regulate fuel availability to and utilization by thermogenic tissues, comparatively few changes in gene expression were found in muscle (Supplementary Fig. 1B–D), epididymal WAT, or subcutaneous WAT (Supplementary Fig. 3). Overall, under conditions of cold exposure, loss of L-PGDS appeared almost exclusively to affect BAT depots and liver, which in the cold acts as an important supplier of de novo synthesized lipids for oxidation by BAT.

Given that our data had indicated that loss of L-PGDS in cold-acclimated animals appeared to increase glucose utilization to meet energy demands, we next investigated whether a lack of L-PGDS could affect lipid and carbohydrate metabolism under standard laboratory conditions (24°C housing). In animals fed a high-fat diet, lack of L-PGDS caused a significant improvement in glucose tolerance (Fig. 5A). Surprisingly, this effect was observed despite apparently similar degrees of insulin resistance induced by high-fat diet in WT and L-PGDS KO mice, as indicated by markers of insulin sensitivity including serum insulin levels during the glucose tolerance test (Fig. 5B and Supplementary Fig. 4G); adiposity (Fig. 5C); adipocyte size (Supplementary Fig. 4A); markers of adipogenesis in BAT, subcutaneous WAT, and epididymal WAT (Supplementary Fig. 4B–D), tissue weights (Supplementary Fig. 4E–F); and fasted insulin, adiponectin, and leptin levels (Fig. 5E). Tissue markers of insulin sensitivity were also unchanged between wild-type and L-PGDS KO mice (Supplementary Fig. 5). Consistent with increased glucose utilization, mice lacking L-PGDS had lower fasting glucose levels (Fig. 5D). Because the changes in glucose tolerance and fasting glucose did not seem to be associated with markers of improved insulin sensitivity, we investigated whether there were any changes in other processes that mediate glucose uptake. In BAT norepinephrine has been shown to promote glucose uptake independently of insulin (33). Consistent with increased sympathetic tone or sensitivity to sympathetic tone, mice lacking L-PGDS had elevated thermogenic markers in BAT, and the constitutive glucose transporter gene Glut1 was also upregulated (Fig. 5F). To test whether response to sympathetic tone in terms of plasma glucose levels could be modulated by L-PGDS, we injected wild-type and L-PGDS KO mice with norepinephrine (1 mg/kg). As expected norepinephrine increased serum glucose levels as a result of its effects on hepatic glucose production (34,35); however, in L-PGDS KO mice the increase in plasma glucose levels was blunted relative to that in wild-type animals (Fig. 5G). To determine whether this was due to changes in glucose uptake as opposed to hepatic glucose uptake, we subsequently included 2-[¹⁴C]-DG as a tracer to measure tissue-specific glucose uptake. We detected elevated uptake of glucose into BAT in response to norepinephrine, with no changes in muscle and a tendency (P = 0.059) toward increased glucose disposal into WAT. Although hepatic glucose production could not be directly assessed in the same assay, the elevated accumulation of 2-deoxy-d-glucose in liver in L-PGDS KO mice was suggestive of decreased glucose output, because 2-deoxy-d-glucose accumulates in liver by mass action (through Glut2) but can be actively exported.

Finally, to try to identify a mechanism for L-PGDS in the regulation of fuel utilization, we measured prostaglandin D₂ and E₂ in BAT from cold acclimated wild-type and L-PGDS KO mice (Fig. 5I). We detected no differences in prostaglandin levels, suggesting that L-PGDS may act as a lipocalin.

In this study, we demonstrated that L-PGDS contributes to the balance between carbohydrate and lipid utilization in vivo. We also demonstrated that the physiological levels of L-PGDS mRNA in BAT are strongly and positively correlated with activation of BAT metabolism. In vivo characterization of cold-exposed mice lacking L-PGDS demonstrated that lack of L-PGDS resulted in a modest impairment in metabolic rate but a substantial increase in RER. These changes in metabolic rate and RER suggested that the principal role of L-PGDS was to control the type of fuel utilized by BAT, rather than the maximal thermogenic capacity.

The first evidence for a role for L-PGDS in BAT metabolism came from its dynamic regulation in response to both physiological and genetic manipulations which affect BAT function. L-PGDS mRNA was upregulated in states of elevated BAT function, including cold acclimation and high-fat feeding, as well as in the Rip140 KO mouse model, which has substantially increased metabolic rate with elevated markers of brown adipocyte genes within WAT (32). Conversely, L-PGDS mRNA was downregulated by thermoneutral housing and in mouse models with reduced BAT function, including PGC1α and PGC1β KO mice. Notably, both Rip140 KO and PGC1β KO mice have RER values that inversely correlate with their L-PGDS levels (20,36), consistent with the elevated RER seen in L-PGDS KO mice.

Although differences in metabolic rate observed between L-PGDS KO mice and wild-type controls were modest, there were much more substantial changes in RER. An elevated RER would usually be considered a marker of increased carbohydrate utilization and decreased lipid utilization by an organism; however, conversion of carbohydrate to lipid and its subsequent oxidation has the same RER as oxidizing carbohydrate itself, suggesting that either increased direct metabolism of carbohydrate or increased de novo lipogenesis followed by fatty acid oxidation could explain the differences in RER between wild-type and KO mice. In agreement with this, molecular analysis of BAT from L-PGDS KO mice demonstrated an increase in the proportion of de novo synthesized lipid located within BAT, as well as increased expression of glycolytic and de novo lipogenic enzymes, when compared with wild-type controls. Under conditions of cold acclimation, the principal sites of de novo lipogenesis are BAT and liver (14). Similarly to BAT, the livers of L-PGDS KO mice also exhibited a gene expression profile consistent with converting carbohydrate to lipid but did not exhibit any increase in hepatic TG levels, suggesting that any increased TG synthesis was potentially being exported. The molecular prolipogenic changes in liver and BAT were supported by lower serum glucose levels. Overall, these data suggest that under conditions of 4°C housing, L-PGDS is required for appropriate utilization of dietary lipids. In the absence of L-PGDS, mice were still able to meet the thermogenic demands of cold exposure; however, this appeared to be at the expense of a substantial increase in carbohydrate utilization.

In addition to cold exposure, high-fat feeding has been shown to promote BAT activation. After feeding with a very high-fat diet (60% calories from fat), L-PGDS KO mice had improved glucose tolerance relative to wild-type mice. Although euglycemic-hyperinsulinemic clamp data would be needed to formally confirm an absence of alterations in insulin sensitivity, there were no molecular markers of insulin resistance that differed between WT and L-PGDS KO mice, suggesting that the improved glucose tolerance of the L-PGDS KO mice could be driven by mechanisms not insulin dependent. Consistent with this concept, the thermogenic program in BAT and the constitutive glucose transporter Glut1 were upregulated in mice lacking L-PGDS. To test the concept that the sympathetic nervous system could regulate glucose levels in L-PGDS KO mice, we injected mice lacking L-PGDS with norepinephrine in combination with 2-[¹⁴C]DG. Mice lacking L-PGDS exhibited a 60% increase in glucose disposal to BAT relative to wild-type controls, supporting the concept of sympathetically mediated glucose uptake into BAT being elevated in the absence of L-PGDS. BAT exhibits very high rates of glucose uptake in response to either norepinephrine or insulin, and although the exact contribution of glucose as a substrate for thermogenesis remains debatable, it seems likely that it does contribute significantly (37). Several publications have demonstrated sympathetic control of glucose uptake into BAT (38–42) and that this uptake is dependent on UCP1 (39). Insulin stimulated uptake of glucose to BAT is not dependent on it being thermogenically functional (40); however, cold exposure does increase the rate of glucose disposal into BAT in response to insulin (42). Overall, these results suggest that BAT can have a major impact on systemic glucose levels and, importantly, that it can do so independent of the actions of insulin.

The finding that L-PGDS can mediate fuel provision to BAT is a potentially important one given the renewed excitement in the potential of BAT to treat metabolic disease after its rediscovery in adult humans (8,12,43). Although much research into BAT has focused on its recruitment and molecular activation, relatively less research has gone into the supply of nutrients to BAT. Recent articles regarding angiogenesis in BAT (44) and the control of lipid uptake into BAT (45) have helped to further our understanding of fuel supply to BAT; however, if BAT is ever to be used to treat human obesity, then it must be both fully activated and appropriately supplied with nutrients. The ability of L-PGDS to control in part the balance of lipid and carbohydrate utilization by BAT highlights another potential layer of regulation of BAT activity.

To try to address in part how L-PGDS may affect BAT function on a molecular level, we measured prostaglandin levels in cold-acclimated BAT and detected no differences. Because L-PGDS is a bifunctional molecule capable of acting both as a carrier of lipophilic molecules (46) and as a prostaglandin synthase (15), the absence of changes in prostaglandins suggested that L-PGDS may act primarily as a lipocalin in terms of its regulatory role in BAT. What L-PGDS binds to and how that molecule may regulate BAT function will be an area for future study.

In summary, L-PGDS appears to regulate the balance between carbohydrate and lipid metabolism in BAT. The association between metabolic health and an ability to switch between carbohydrate and lipid metabolism is already established; however, the ability of L-PGDS to control the balance between lipid and carbohydrate utilization, with only minor effects on metabolic rate, highlights a new type of control mechanism for whole organism substrate handling that could potentially be used to bypass states of nutritionally induced or genetic insulin resistance.

Diabetes U.K., EU HEPADIP FP6, The Medical Research Council Centre for Obesity Related Disorders, and the Medical Research Council funded this work. Studies in Stockholm were supported by the Swedish Research Council and ADAPT FP7 and the BBSRC.

C.L. is employed by AstraZeneca, Mölndal, Sweden. No other potential conflicts of interest relevant to this article were reported.

S.V., H.F., M.Ch., M.Ca., M.D., C.L., M.M., P.V., and K.B. performed experiments. S.V., C.Y.T., M.M., J.L.G., H.F., and B.C. analyzed data. S.V., M.P., P.V., J.K.S., J.L.G., B.C., and A.V.-P. designed experiments. S.V. and A.V.-P. wrote the manuscript. Y.U. and N.E. provided essential reagents. S.V. and A.V.-P. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Helen Westby, Dan Hart, Agnes Lukasik, Sylvia Osborne, and Sarah Grocott, all from the University of Cambridge, for technical assistance.