All experiments were performed with permission of the Animal Welfare Officers at University Medical Center Hamburg-Eppendorf and Behörde für Gesundheit und Verbraucherschutz Hamburg. To induce Cre-loxP recombination, Cdh5-Cre/ERT2 x Lpl ^flox/flox (EndoLPLko) mice received three consecutive doses of 0.2 mg tamoxifen (Sigma-Aldrich. St. Louis, Missouri, USA) dissolved in 100 µl sunflower oil via gavage 1 week before the experiments. All mice had ad libitum access to food and water and were kept in a temperature-controlled room at given temperatures with a 12 h light: 12 h dark cycle. The mice were fed a chow diet (P1324, Altromin, Germany) or western type diet, ssniff Spezialdiäten GmbH, Germany, duration: 2 weeks before necropsy), as indicated in the text. For the experiments, age- and weight-matched male mice were used. For the combined oral glucose and fat tolerance test (OGFT), turnover study and indirect calorimetry, mice were first fed chow diet and then received western-type diet for 2 weeks. For OGFT and lipoprotein turnover studies body composition (fat and lean mass) was analyzed using a magnetic whole-body composition analyzer (EchoMRI™, Zinsser Analytic GmbH, Eschborn, Germany) 1 day prior to the experiments. For all terminal procedures, mice received 180 mg/kg ketamine and 24 mg/kg xylazine before necropsy.

For OGFT, mice were fasted for 2 h before receiving an oral gavage of 200 µl of a glucose-lipid emulsion containing 47 mg triglycerides/kg body weight and glucose (2 g/kg body weight). The gavage solution was labelled with ¹⁴C-triolein (0.15 MBq/kg body weight, Perkin Elmer, Waltham, MA, USA) and ³H-deoxyglucose (³H-DOG; 0.72 MBq/kg body weight, Hartmann Analytic, Braunschweig, Germany). Organs were harvested 2 h after gavage. For lipoprotein turnover studies, mice were fasted for 4 h and were subsequently tail vein-injected with 100 µl radiolabeled TRLs. Radiolabeled TRLs were prepared by extraction of rat chylomicrons (Skottova et al., 1995) using the method of Folch, addition of radiolabelled ¹⁴C-triolein and subsequent sonication. Solvents were evaporated, and labelled TRL were formed by addition of PBS and ¹⁴C-triolein (MBq/kg body weight) and subsequent sonication. ³H-DOG (MBq/kg body weight) was added to the emulsion to follow glucose uptake from the circulation without triggering glucose-stimulated insulin secretion. Organs were harvested 15 min after injection. For all experiments, organs from anesthetized mice were harvested after systemic perfusion with PBS-heparin (10 U/mL; Rotexmedica) via the left heart ventricle. Organs were homogenized using Solvable™ (Perkin Elmer) or frozen immediately in liquid nitrogen and stored at −80°C for further analysis. Radioactive counts were determined by scintillation counting using a Tricarb scintillation counter (Liquid Scintillation Analyzer Tri-Carb^®2810TR, Perkin Elmer).

For imaging of TRL disposal in BAT by confocal fluorescence microscopy, an immunofluorescence-based method was used. Briefly, BODIPY™ FL C 16 labeled TRL (BODIPY-TRL) were prepared using a similar approach to radiolabeled TRL. Briefly, 0.2 mg BODIPY™ FL C 16 (D3821, Thermofischer) dissolved in 1 ml Intralipid^® (CLINOLEIC 20%, Baxter S.A.) was applied intragastrically in rats and chylomicrons were obtained from cannulated lymphatic vessels. Lipids were extracted to obtain TRL particles as described above.

TRAP was performed as previously described (Long et al., 2014) with modifications. In brief, small pieces (50–100 mg) of frozen BAT were Dounce-homogenized in 4 ml homogenization buffer (50 mM Tris [pH 7.5], 12 mM MgCl2, 100 mM KCl, 1% NP-40, 100 μg/ml Cycloheximide, 1 mg/ml sodium heparin, 2 mM DTT, 0.2 U/μL RNasin, and 1x Complete EDTA-free protease inhibitor; Roche). After centrifugation at 13,000 rpm for 10 min, the lipid layer was removed and the supernatant was collected and incubated with anti-GFP antibody (5 μg/ml; Abcam, ab290) for 1 h at 4°C. Protein G dynabeads were washed twice in low-salt wash buffer (50 mM Tris [pH 7.5], 12 mM MgCl2, 100 mM KCl, 1% NP-40, 100 μg/ml cycloheximide, and 2 mM DTT), added to the homogenates with antibody, and subsequently incubated for 30 min. Dynabeads with immunoprecipitates were washed three times in high-salt wash buffer (50 mM Tris [pH 7.5], 12 mM MgCl2, 300 mM KCl, 1% NP-40, 100 μg/ml cycloheximide, and 2 mM DTT). Following the last wash, RLT buffer with β-mercaptoethanol was added to dynabeads, and RNA was extracted using a QIAGEN Micro RNeasy kit according to the manufacturer’s instructions. For input RNA, 5% of homogenates were mixed with TRIzol and processed according to the manufacturer’s instructions to extract total RNA. Isolated RNA was quantified by Qubit.

To obtain SVF, interscapular BAT was minced and then digested with 1 mg/ml type II collagenase for 30 min at 37°C (Sigma Aldrich). The dissociated cells were passed through a 100 μm sieve to remove undigested particles. Centrifugation at 700 g for 10 min was then performed to separate the SVF pellet from the floating adipocytes. The resulting SVF pellet was dissolved in PBS and passed through a 40 µm sieve to achieve higher purity. For isolation of endothelial cells and brown adipocytes, the filtrate was centrifuged at 600 x g for 5 min, the cell pellet was resuspended and incubated with CD11b MicroBeads for depletion of the macrophage fraction (Miltenyi; 10 µl beads/107 cells). CD11b + cells were captured from the lysate using magnetic columns (Miltenyi). The flow through was centrifuged, the pellet was resuspended and incubated with CD31 MicroBeads (Miltenyi; 10 µl beads/107 cells) to isolate endothelial cells. The flow through, containing predominantly adipocytes, was collected. RNA was isolated from cells, tissue samples and SVF using TRIzol Reagent (ThermoFischer Scientific, Waltham, MA, USA) and NucleoSpin RNA/Protein kit (Macherey & Nagel, Düren, Germany) and used for cDNA preparation using the High Capacity cDNA Reverse Transcription kit with RNase Inhibitor (ThermoFischer Scientific) according to the manufacturer’s instructions. Real-time PCR using TaqMan Assay-on-Demand primer sets was performed on a QuantStudio 5 Real-Time-PCR System (ThermoFischer Scientific) and relative expression was normalized to the housekeeper Tbp. Taqman^® assays used in this study (assay IDs in brackets): Acaca (Mm01304285_m1), Angptl4 (Mm00480431_m1), Cd36 (Mm00432403_m1), Chrebpß (AIVI4CH), Dio2 (Mm00515664_m1), Elovl3 (Mm00468164_m1), Elovl6 (Mm00851223_s1), Fasn (Mm00662319_m1), Glut4 (Mm01245502_m1), Lipg (Mm00495368_m1), Ppargc1a (Mm00447183_m1), Scd1 (Mm00772290_m1), Srebp1c (AI89KJW), Tbp (Mm00446973_m1), Ucp1 (Mm00494069_m1).

Plasma triglyceride (Triglyceride FS Kit, DiaSys, Holzheim, Germany), cholesterol (Cholesterin FS Kit, DiaSys), and NEFA levels (NEFA-HR (2)-Kit, FUJIFILM Wako Chemicals, Neuss, Germany) were determined using commercial kits according to the manufacturer’s instructions. Blood glucose was determined by conventional test stripes (Accu-Chek, Roche).

A 5 µm thick hematoxylin and eosin-stained section were used for microscopy and size determination. The adipocyte diameter estimates were calculated by NIS-Elements from BAT of 5 independent mice per group.

Indirect calorimetry was performed in a TSE phenomaster (TSE Systems) in a temperature- and humidity-controlled chamber as described before (Heine et al., 2018). During the experiment, all mice were housed in single cages at a 12 h light: 12 h dark cycle and had ad libitum access to food and water. EndoLPLko and control mice were kept in the system at 22°C for 6d and for the analysis of energy expenditure, the temperature was decreased to 6°C for 7d. Oxygen consumption and carbon dioxide production were measured every 15 min. Body core temperature was continuously determined using an implantable mouse telemetry system based on G2 E- Mitter (Minimitter System by Starr Life Science) in the TSE phenomaster.

Murine brown adipocytes were seeded and differentiated in the lower chamber of a transwell system. On day seven of the brown adipocyte differentiation protocol (Haas et al., 2009), murine microvascular endothelial cells (InSCREENex GmbH) seeded in a 0.4 µM pore size TC-insert (Sarstedt) were added on the top of the adipocytes. Both brown adipocytes and endothelial cells were (co-)cultured for 4 hours in a media containing 50% differentiation media (DMEM Glutamax (ThermoFisher Scientific) supplemented with 10% FBS, 100 IU/ml penicillin, 100 μg/ml streptomycin, 20 nM insulin, 1 nM triiodothyronine) and 50% endothelial cell media (InSCREENex GmbH) with or without 10 µM CL316,243. RNA was isolated from brown adipocytes and endothelial cells separately.

Data are presented as mean ± SEM. Two groups were compared by unpaired two-tailed Student’s t test, more than two groups by one-way or two-way ANOVA, as indicated in the figure legends. No method was used to determine whether the data met assumptions of either Student’s t test or ANOVA. The statistical parameters (i.e., p values, numbers of biological repeats) can be found in the figure legends. No exclusion or inclusion criteria were used for data analyses. Statistical analyses were conducted using Graph Pad software; p < 0.05 was considered significant.

Previously we described that Lpl is expressed in CD31-positive endothelial cells of BAT (Fischer et al., 2021). To confirm this finding, we employed NUTRAP mice, a Cre recombinase-dependent transgenic model that allows cell type-specific gene expression analysis by translating ribosome affinity purification (TRAP). This method employs pulldown of polyribosomes and thus does not require tissue disintegration (Roh et al., 2017). Endothelium-specific NuTRAP (EndoNuTRAP) mice were generated by crossing NuTRAP with VE-cadherin (Cdh5)-Cre mice. To analyze the temperature-dependence of Lpl expression, EndoNuTRAP mice were kept at room temperature (22 °C), or exposed to either cold (6°C) or 30°C for 3 days. Efficient enrichment of mRNA from endothelial cells in the BAT lysates was confirmed by high expression of the endothelial marker Gpihbp1 in the pulldown (TRAP) fraction as compared to whole tissue lysate (input) at all temperatures (Figure 1A). Of note, we observed substantial expression of Lpl in the endothelial cell fraction that increased with decreasing housing temperatures of the EndoNuTRAP mice (Figure 1B). To further investigate the expression of Lpl in endothelial cells, we established a transwell system for co-culturing murine microvascular endothelial cells with murine brown adipocytes (Figure 1C). To evaluate the potential mutual effects on gene expression, we cultured cells alone or in combination that were treated without or with the β3-adrenergic agonist CL316,243. Ucp1 expression was induced in CL316,243-treated brown adipocytes (Figure 1D), which confirms their thermogenic activation. Endothelial cells co-cultured with quiescent brown adipocytes showed a trend for increased Lpl expression in comparison to controls without adipocytes (Figure 1E). This effect was significant after additional incubation with the ß3-adrenergic receptor agonist CL316,243 (Figure 1E). These results suggested that activated brown adipocytes produce paracrine signals that lead to increased Lpl expression in endothelial cells. To estimate Lpl expression in endothelial cells versus adipocytes from BAT of cold-exposed mice, we employed antibody-based magnetic-activated cell sorting (Fischer et al., 2021). In this setup, Lpl mRNA levels were approximately 2.5-fold higher in thermogenic adipocytes compared endothelial cells (Figure 1F). Taken together, Lpl is expressed at appreciable levels in endothelial cells of cold-activated BAT and endothelial Lpl expression is stimulated by brown adipocytes.

To analyze the relevance of endothelial LPL for BAT energy metabolism, endothelial specific LPL knockout mice were created by crossing mice expressing a tamoxifen-inducible Cre recombinase under the control of the VE-cadherin (Cdh5) promoter with floxed Lpl mice to generate mice lacking LPL in endothelial cells (EndoLPLko) and control littermates. Cre recombinase expression was detected in stromal vascular fraction (SVF) isolated from BAT of EndoLPLko but not in control mice (Figure 2A). Moreover, significant reduced expression of Lpl in the SVF confirmed efficient knockdown (Figure 2A). Next, we assessed the effect of the endothelial cell-specific knockout on energy balance and metabolism. To create nutritional conditions typical for humans that go along with elevated lipoprotein turnover in adipose tissues (Worthmann et al., 2017), the study was performed with mice that were fed western type diet for 2 weeks. Mice were then housed at thermoneutrality (30°C), a state of thermogenic inactivity (Fischer et al., 2018), or at 6°C (cold) to strongly activate BAT. Under these conditions, body weight (Figure 2B), lean weight (Figure 2C), fat weight (Figure 2D), plasma cholesterol (Figure 2E), triglycerides (Figure 2F), non-esterified fatty acids (Figure 2G) and blood glucose (Figure 2H) were not different in EndoLPLko mice compared to control littermates. Except for plasma triglycerides, no significant effects of housing temperature on these parameters was observed. Furthermore, lack of LPL in endothelial cells had no effect on histological BAT appearance studied by hematoxylin and eosin staining (Figure 2I). As expected, the diameter of brown adipocytes was smaller in cold-exposed compared to thermoneutral-housed mice but we did not observe a major effect by the genotype (Figure 2J).

To further address the role of endothelial cell-specific LPL expression for whole body energy balance in adaptation to cold ambient temperature, EndoLPLko and control mice were subjected to indirect calorimetry. Energy expenditure increased in both genotypes when the mice were exposed to cold temperature (Figure 3A). Furthermore, respiratory exchange rate was slightly lower under these conditions, indicating higher lipid compared to glucose oxidation (Figure 3B). However, the EndoLPLko mice exhibited no significant difference in energy expenditure or respiratory exchange rate compared to the Cre-controls, indicating preserved systemic energy homeostasis in the absence of endothelial LPL.

To study whether lack of endothelial LPL affects organ-specific energy handling in the postprandial state, EndoLPLko and control mice kept on a western-type diet received a combined fat and glucose gavage containing ³H-deoxyglucose (³H-DOG) and ¹⁴C-triolein as radiolabeled tracers. In line with previous data (Bartelt et al., 2011), uptake of ¹⁴C-triolein into BAT was increased in cold-exposed mice whereas lipid uptake into the liver was diminished (Figure 4A). No significant differences were detected in the other organs investigated. Of note, no effect of endothelial-specific LPL knockdown on ¹⁴C-triolein uptake in any organ could be detected, except for a significantly lower hepatic lipid uptake in warm-housed mice (Figure 4A). As expected, ³H-DOG uptake into BAT was higher in cold-exposed mice, and this was additionally observed for heart, inguinal WAT and liver (Figure 4B). Again, no effect of the different genotypes was observed. Plasma analysis demonstrated that the concentrations of triglycerides (Figure 4C), cholesterol (Figure 4D) and non-esterified fatty acids (Figure 4E) in these mice were not different between the experimental groups.

These data indicate that TRL and glucose disposal in BAT is independent of endothelial LPL expression in the postprandial state. Next, we tested whether this, and the lacking effect of endothelial LPL knockout on energy homeostasis (Figure 3), might be due to compensation by altered BAT expression of genes critical for thermogenesis (Figure 4F), lipoprotein and lipid processing (Figure 4G) or glucose handling (Figure 4H). As predicted, cold exposure led to the induction of Ucp1 (Figure 4F) (Cannon and Nedergaard, 2004), endothelial lipase (Lipg) (Schaltenberg et al., 2021) and fatty acid elongase 3 (Elovl3) (Jakobsson et al., 2005) (Figure 4G). Also in line with previous work (Sanchez-Gurmaches et al., 2018), glucose-regulated DNL genes including carbohydrate response element-binding protein-β (Chrebpβ), Elovl6, acetyl-CoA carboxylase-α (Acaca) and fatty acid synthase (Fasn) were induced by cold (Figures 4G,H). However, at a given housing temperature expression of these genes was not different in EndoLPLko mice compared to littermates, suggesting that endothelial LPL does not influence postprandial BAT energy and lipid homeostasis.

Insulin action on brown adipocytes has been shown to increase LPL-dependent lipoprotein disposal in BAT (Heine et al., 2018), and thus high insulin levels in the postprandial state may mask potential effects of LPL produced in endothelial cells. To investigate the role of endothelial cell LPL in BAT lipid disposal under low insulin conditions, we intravenously injected fluorescently-labeled TRL into mice fasted for 4 h. Interscapular BAT was studied ex vivo by confocal microscopy 15 min after the injection. Brown adipocytes from control animals (wild type) showed a strong accumulation of BODIPY-labelled fatty acids within their lipid droplets and the same was observed for EndoLPLko mice (Figure 5A). In contrast, mice lacking LPL selectively in brown adipocyte (BAT-LPLko) showed little accumulation of BODIPY fatty acids in BAT. Thus, LPL in endothelial cells, unlike that in brown adipocytes, is not essential for uptake of TRL-derived fatty acids in the fasted state. To further address this notion in a more quantitative fashion, we intravenously injected TRL labeled with ¹⁴C-triolein together with ³H-DOG in tracer amounts into fasted mice that were kept on a western type diet and either housed at 30°C (warm) or at 6°C (cold). Notably, we observed that the cold-dependent increase in BAT ¹⁴C-triolein uptake was moderately but significantly higher in EndoLPLko mice compared to controls (Figure 5B). In contrast, other organs exhibited no genotype effect. In contrast to lipid uptake, ³H-DOG uptake into BAT was not influenced by the absence of LPL in endothelial cells (Figure 5C). Of note, further BAT analysis revealed that cold-exposed EndoLPLko mice compared to controls exhibited increased expression of the thermogenic markers Ucp1 and deiodinase-2 (Dio2) (Figure 5D), and of the lipases Lpl, Lipg as well as the fatty acid elongases Elovl3 and Elovl6 (Figure 5E). In line with the lack of genotype effect on glucose handling, no effect was observed for glucose-handling genes including DNL enzymes in cold-exposed EndoLPLko versus control mice (Figure 5F). Altogether, the lack of endothelial LPL resulted in higher lipid disposal by activated BAT, which may be explained by the compensatory induction in the expression of adipocyte LPL and endothelial LIPG.