Impaired thermogenesis and sharp increases in plasma triglyceride levels in GPIHBP1-deficient mice during cold exposure

GPIHBP1, an endothelial cell protein, binds lipoprotein lipase (LPL) in the subendothelial spaces and transports it to the capillary lumen. In Gpihbp1 –/– mice, LPL remains stranded in the subendothelial spaces, causing hypertriglyceridemia, but how Gpihbp1 –/– mice respond to metabolic stress ( e.g., cold exposure) has never been studied. In wild-type mice, cold exposure increases LPL-mediated processing of triglyceride-rich lipoproteins (TRLs) in brown adipose tissue (BAT), providing fuel for thermogenesis and leading to lower plasma triglyceride levels. We suspected that defective TRL processing in Gpihbp1 –/– mice might impair thermogenesis and blunt the fall in plasma triglyceride levels. Indeed, Gpihbp1 –/– mice exhibited cold intolerance, but the effects on plasma triglyceride levels were paradoxical. Rather than falling, the plasma triglyceride levels increased sharply (from ~4,000 to ~15,000 mg/dl), likely because fatty acid release by peripheral tissues drives hepatic production of TRLs that cannot be processed. We predicted that the sharp increase in plasma triglyceride levels would not occur in Gpihbp1 −/− Angptl4 −/− mice, where LPL activity is higher and baseline plasma triglyceride levels are lower. Indeed, the plasma triglyceride levels in Gpihbp1 −/− Angptl4 −/− mice fell during cold exposure. Metabolic studies revealed increased levels of TRL processing in the BAT of Gpihbp1 −/− Angptl4 −/− mice. Statistical analyses were performed with GraphPad Prism software. Differences in triglyceride levels before and after cold exposure were analyzed by paired two-tailed t -tests. All other statistical calculations were performed by unpaired two-tailed t -tests.

INTRODUCTION GPIHBP1, a glycoprotein of capillary endothelial cells, plays three important functions in plasma triglyceride metabolism (1). First, GPIHBP1 captures lipoprotein lipase (LPL) in the interstitial spaces and transports it to its site of action in the capillary lumen (2,3). Second, GPIHBP1 expression is crucial for the margination of triglyceride-rich lipoproteins (TRLs) along capillaries (4), allowing LPL-mediated triglyceride hydrolysis to proceed. Finally, in vitro biochemical studies have demonstrated that the binding of LPL to GPIHBP1 stabilizes LPL's hydrolase domain and preserves catalytic activity-even in the presence of ANGPTL4, an LPL inhibitor protein (5,6).
In Gpihbp1-deficient mice (Gpihbp1 -/-), LPL is trapped in the interstitial spaces around parenchymal cells, causing impaired lipolytic processing of TRLs and severe hypertriglyceridemia ("chylomicronemia") (2,3). Human patients with GPIHBP1 loss-of-function mutations manifest chylomicronemia, with plasma triglyceride levels greater than 1,500 mg/dl (7-9). In Gpihbp1 -/mice, the plasma triglyceride levels on a chow diet are ~4,000 mg/dl (2,3,10). But while high plasma triglyceride levels in Gpihbp1 -/mice have been well documented (1), it is unclear how these mice respond to metabolic stress, such as that associated with exposure to the cold.
In mice, mutations that impair fatty acid utilization impair thermogenesis during cold exposure, compromising their ability to maintain a normal body temperature. For example, a deficiency in fatty acid binding protein 3 (FABP3) or a combined deficiency of FABP4 and FABP5 impairs fatty acid oxidation in brown adipose tissue (BAT), leading to cold intolerance (11,12). A deficiency of CD36, a putative fatty acid transporter, also causes cold intolerance (13,14).
However, no one has yet determined whether the processing of TRLs by the GPIHBP1-LPL complex is essential for thermogenesis. Normally, LPL-mediated processing of TRLs by BAT is robust during cold exposure, providing fatty acids for mitochondrial oxidation and leading to lower plasma triglyceride levels (15,16). The latter observations raise doubts about whether Gpihbp1 -/-by guest, on July 20, 2018 www.jlr.org Downloaded from mice, where triglyceride hydrolysis is clearly impaired, would remain euthermic during cold exposure.
In the current study, we had two goals. The first was to test whether intravascular lipolysis by the LPL-GPIHBP1 complex is crucial for thermogenesis during cold exposure. It seemed possible that the impaired intravascular lipolysis in Gpihbp1 -/mice would lead to defective thermogenesis and cold sensitivity, but our confidence in making this prediction was limited. Indeed, one could argue that the metabolic flux of triglycerides through the tissues of Gpihbp1 -/mice is normal (despite markedly elevated levels of triglycerides in the plasma compartment), and that this flux would provide the fuel required for thermogenesis. One could also argue that the fatty acids released by peripheral tissues during cold exposure (17)(18)(19) would be sufficient to fuel thermogenesis. Our second goal was to test whether plasma triglyceride levels in Gpihbp1 -/mice would fall during cold exposure-as they do in wild-type mice. Once again, it was difficult to make a firm prediction. On one hand, one could imagine that the defective lipolysis in Gpihbp1 -/mice would blunt the fall in plasma triglyceride levels. On the other hand, one could argue that the decline in plasma triglyceride levels during cold exposure might be fairly normal, simply because triglyceride levels fall sharply during cold exposure in Apoa5 -/mice (16), where impaired intravascular lipolysis leads to moderate to severe increases in plasma triglyceride levels (20)(21)(22).
In the current studies, we addressed both experimental goals, assessing cold sensitivity in Gpihbp1 -/mice as well as the plasma triglyceride response during cold exposure.

Triglyceride and apoB measurements
Blood was collected from anesthetized mice by retro-orbital puncture with heparinized capillary tubes (Kimble Chase). Plasma was separated from blood cells by centrifugation (14,000  g for 30 sec) and stored at -80°C. Plasma triglycerides were quantified with a commercial kit (Sigma, TR0100). To assess relative levels of apoB proteins, plasma samples were delipidated with acetone:ethanol (1:1) and then size-fractionated by SDS-PAGE. ApoB was detected by western blotting with an apoB-specific monoclonal antibody (2G11, 10 g/ml) (28) and IRDye680-or IRDye800-labeled secondary antibodies (LI-COR; 1:2,000). Antibody binding was detected with an Odyssey infrared scanner (LI-COR).

Immunohistochemistry
Brown adipose tissue was collected from mice that had been perfused with 15 ml of PBS containing 5 mM EDTA, followed by 5 ml of freshly prepared 3% paraformaldehyde in PBS. Within each experiment, exposure conditions were identical.

Downloaded from
Statistical analyses were performed with GraphPad Prism software. Differences in triglyceride levels before and after cold exposure were analyzed by paired two-tailed t-tests. All other statistical calculations were performed by unpaired two-tailed t-tests.

RESULTS
To assess cold sensitivity in the setting of GPIHBP1 deficiency, we housed Gpihbp1 −/− and littermate wild-type mice in the cold (4°C) for 6 h. Three of seven Gpihbp1 −/− mice, but no wildtype mice, developed hypothermia ( Fig. 1A-B). We also housed Gpihbp1 −/− and littermate wildtype mice at 12°C for two days and then at 6°C for two additional days. After reducing the temperature to 6°C, the body temperature of Gpihbp1 −/− mice was significantly lower than in wildtype mice (Fig. 1C).
When wild-type mice were housed at 12°C for two days and then 6°C for two additional days, the plasma triglyceride levels fell (Fig. 1D), consistent with earlier studies (16). Plasma triglyceride levels also fell in Gpihbp1 +/− mice, where the plasma triglyceride levels are normal (2) (Supplemental Figure S1). The plasma triglyceride response in Gpihbp1 −/− mice during cold exposure was paradoxical. Rather than declining, the triglyceride levels increased sharply (from 3,000-5,000 to 13,000−19,000 mg/dl) (Fig. 1D). As expected, the food intake during cold exposure increased in both wild-type and Gpihbp1 −/− mice-and to a similar degree (Fig. 1E).
Body weight during cold exposure was stable in both groups of mice (Fig. 1F).
To be confident that the higher plasma triglyceride levels in Gpihbp1 −/− mice during cold exposure were not due to increased food consumption, we performed a follow-up study in which food consumption was restricted to 3 g/day (the average amount consumed by Gpihbp1 −/− mice at room temperature). Despite the restricted intake of food, the plasma triglyceride levels in Gpihbp1 −/− mice increased sharply during cold exposure ( Fig. 2A). The plasma triglyceride levels in Gpihbp1 −/− mice also increased when the mice were fed a fat-free diet (Fig. 2B). We strongly suspect that the higher plasma triglyceride levels in Gpihbp1 −/− mice during cold exposure were due to increased hepatic production of TRLs-simply because the release of fatty acids from adipose tissue during the cold would be expected to drive hepatic TRL production (30) and because TRL processing in Gpihbp1 −/− mice is defective (1). Of note, biomedical scientists often infer changes in TRL production when plasma triglyceride levels change after inhibiting LPL activity by guest, on July 20, 2018 www.jlr.org

Downloaded from
with Triton WR-1339 (31,32). LPL-mediated TRL processing is profoundly defective in Gpihbp1 −/− mice at baseline; hence, studies with Triton WR-1339 would make little sense. We suspected that increased TRL production in cold-exposed Gpihbp1 −/− mice would lead to higher plasma apoB levels. Indeed, this was the case (Fig. 2C, Supplemental Figure S2).
We also examined the relevance of apoC-III to the plasma triglyceride levels during cold exposure. ApoC-III retards LPL-mediated TRL processing, limiting access of LPL to TRL substrates (33). In transgenic mice that overexpress human apoC-III, TRL processing by the LPL-GPIHBP1 complex is impaired (34), causing hypertriglyceridemia. When human apoC-III transgenic mice were exposed to the cold, the plasma triglyceride levels fell sharply (Fig. 3A), suggesting that the higher LPL activity in BAT during cold exposure (35,36) outweighs the ability of apoC-III to limit LPL accessibility to triglyceride substrates (33,34). We also examined mice deficient in both apoC-III and GPIHBP1 (Gpihbp1 −/− Apoc3 −/− ). In those mice, the baseline plasma triglyceride levels on a chow diet (1248 ± 424 mg/dl) were lower than in Gpihbp1 −/− mice (3966 ± 576 mg/dl). Nevertheless, the plasma triglyceride levels in Gpihbp1 −/− Apoc3 −/− mice increased sharply during cold exposure (Fig. 3B), as in Gpihbp1 −/− mice (Fig. 3C). Thus, apoC-III deficiency has little effect on the plasma triglyceride response of Gpihbp1 −/− mice during cold exposure.
We suspected that Angptl4 deficiency, which stabilizes LPL (5, 6) and increases LPL activity in BAT (35,36), might increase the processing of TRLs in cold-exposed Gpihbp1 −/− mice and prevent the spike in plasma triglyceride levels. Earlier studies showed that plasma triglyceride levels in Gpihbp1 −/− Angptl4 −/− mice are only about 10% of those in mice lacking Gpihbp1 alone (4,37). We confirmed that finding and went on to show that the plasma triglycerides in Gpihbp1 −/− Angptl4 −/− mice fall during cold exposure (Fig. 3D). Because LPL activity in BAT increases during cold exposure (36) and because plasma triglyceride levels in Gpihbp1 −/− Angptl4 −/− mice, housed either at room temperature or 6C. We also tested Gpihbp1 −/− mice that had been maintained on a fat-free diet for 3 weeks; the baseline plasma triglyceride levels in that group of Gpihbp1 −/− mice (866 ± 334 mg/dl) were more comparable to those in Gpihbp1 −/− Angptl4 −/− mice (408 ± 170 mg/dl). We observed more 14 C in BAT in the Gpihbp1 −/− Angptl4 −/− mice than in Gpihbp1 −/− mice (Fig. 4A). Also, the amount of 14 C-labeled fatty acids remaining in the plasma was lower in the Gpihbp1 −/− Angptl4 −/− mice than in Gpihbp1 −/− mice (Fig. 4A). There were no differences in 14 C uptake by the liver (Fig. 4A). In our experiments, some of the 14 C-labeled fatty acids would have been oxidized and eliminated as carbon dioxide. The recovery of the 14 C was lower in cold-exposed Gpihbp1 −/− Angptl4 −/− mice than in Gpihbp1 −/− Angptl4 −/− mice housed at room temperature, consistent with more TRL processing and fatty acid oxidation (Supplemental Figure S3). Together, our findings support the idea of increased TRL processing in Gpihbp1 −/− Angptl4 −/− mice, particularly during cold exposure. We found increased [ 3 H]retinol uptake in the BAT of cold-exposed (Fig. 4B), consistent with an earlier study (16) suggesting that TRL particles can be taken up by BAT during cold exposure.
Why is there more TRL processing in Gpihbp1 −/− Angptl4 −/− mice, given that GPIHBP1 deficiency impairs transport of LPL to the capillary lumen? One possibility is that, despite an absence of GPIHBP1, small amounts of LPL reach the capillary lumen in Gpihbp1 −/− Angptl4 −/− mice. To test that idea, we performed immunohistochemistry on BAT, focusing on whether small amounts of LPL reach the luminal surface of capillary endothelial cells in the absence of GPIHBP1. The presence (or absence) of LPL in the capillary lumen was judged by imaging capillaries containing endothelial cell nuclei, where the basolateral and luminal surfaces of capillaries are separated and therefore easily visualized by confocal microscopy (3,29). Nearly all of the LPL in the BAT from wild-type mice was bound to GPIHBP1 on capillary endothelial cells-and was clearly present along the capillary lumen (Fig. 5A-B). In Gpihbp1 −/− mice, the LPL was mislocalized to the interstitial spaces surrounding adipocytes and endothelial cells ( 5A), and no LPL could be detected in the capillary lumen (Fig. 5B). In Gpihbp1 −/− Angptl4 −/− mice, the LPL was mislocalized to the interstitial spaces, and as expected the LPL signal was more intense than in Gpihbp1 −/− mice (Fig. 5A). In contrast to findings in Gpihbp1 −/− mice, small amounts of LPL were detectable along the capillary lumen in the BAT of Gpihbp1 −/− Angptl4 −/− mice (Fig. 5B). In these experiments, we also examined, as an experimental control, the BAT of Lpl knockout mice carrying a skeletal muscle-specific human LPL transgene (Fig. 5A-B). In those mice, we could not detect mouse LPL in brown adipose tissue, either in the interstitial spaces or along the capillary lumen. The absence of mouse LPL in these control studies increased our confidence that the small amounts of LPL along the capillary lumen in Gpihbp1 −/− Angptl4 −/− mice were not an artifact.

DISCUSSION
We assessed the effects of Gpihbp1 deficiency on body temperature and plasma triglyceride levels during cold exposure. When Gpihbp1 −/− and wild-type mice were housed at 4C, nearly half of the Gpihbp1 −/− mice, but none of the wild-type mice, developed hypothermia. When the two groups of mice were tested in a milder cold-exposure protocol (two days at 12C followed by two additional days at 6C), the body temperature in the Gpihbp1 −/− mice was lower than in wild-type mice. These findings show that the defect in intravascular lipolysis associated with Gpihbp1 deficiency results in cold intolerance. Neither the metabolic flux of triglycerides through tissues of Gpihbp1 −/− mice nor the utilization of plasma fatty acids or glucose are sufficient for maintaining body temperature. Our findings provide support for the notion that the TRL-mediated delivery of fatty acid fuel to BAT is important for thermogenesis. It is now clear that defective intravascular lipolysis (e.g., GPIHBP1 deficiency), defective entry of fatty acids into mitochondria (e.g., CPT-1 deficiency) (38), defects in intracellular fatty acid trafficking (e.g., FABP3 deficiency) (11), and deficiencies in mitochondrial enzymes that metabolize fatty acids (e.g., LCAD) (39) cause cold intolerance. A deficiency of CD36, a putative fatty acid transporter, also leads to cold intolerance (13,14). In the current study, we had hoped to determine whether a combined deficiency of GPIHBP1 and CD36 would cause an exaggerated cold intolerance phenotype. Unfortunately, despite exhaustive breeding efforts, only one Gpihbp1 −/− Cd36 −/− mouse survived to weaning, and that mouse was very small and died before any experiments could be performed.
When we embarked on our studies, we suspected that the plasma triglyceride levels in Gpihbp1 −/− mice during cold exposure might fall but that the extent of the fall would be blunted.
To our surprise, we observed a sharp increase in plasma triglyceride levels in Gpihbp1 −/− mice during cold exposure. In hindsight, this finding made perfect sense. Exposure to the cold stimulates lipolysis in adipose tissue (17)(18)(19), releasing fatty acids for the production of hepatic TRLs (which in the setting of GPIHBP1 deficiency cannot undergo normal processing). As noted earlier, by guest, on July 20, 2018 www.jlr.org Downloaded from biomedical scientists who study lipid metabolism in mice often infer changes in hepatic TRL production rates based on changes in plasma triglyceride levels when LPL activity is blocked (31,32). In Gpihbp1 −/− mice, we suspect that increased hepatic TRL production, combined with the block in TRL processing (2), explains the sharp increase in plasma triglyceride levels during cold exposure.
The baseline plasma triglyceride levels in Gpihbp1 −/− Apoc3 −/− mice were lower than in days (open symbols), after which the mice were exposed to cold and food was restricted to 3 g/day.
The plasma triglyceride levels (closed symbols) increased sharply during cold exposure (two days at 12°C followed by two days at 6°C). (B) Plasma triglyceride levels in Gpihbp1 −/− mice (n = 9) maintained on a fat-free diet, before and after cold exposure (two days at 12°C followed by two days at 6°C). (C) Bar graph comparing relative amounts of apoB in the plasma before and after cold exposure (using the same mice as in panel B). *p < 0.05; ***p < 0.001