Mice Lacking Hepatic Lipase Are Lean and Protected against Diet-Induced Obesity and Hepatic Steatosis

Abstract

Hepatic lipase (HL)-mediated lipoprotein hydrolysis provides free fatty acids for energy, storage, and nutrient signaling and may play a role in energy homeostasis. Because HL-activity increases with increased visceral fat, we hypothesized that increased HL-activity favors weight gain and obesity and consequently, that HL deficiency would reduce body fat stores and protect against diet-induced obesity. To test this hypothesis, we compared wild-type mice (with endogenous HL) and mice genetically deficient in HL with respect to daily body weight and food intake, body composition, and adipocyte size on both chow and high-fat (HF) diets. Key determinants of energy expenditure, including rate of oxygen consumption, heat production, and locomotor activity, were measured by indirect calorimetry. HL-deficient mice exhibited reduced weight gain on both diets (by 32%, chow; by 50%, HF; both P < 0.0001, n = 6–7 per genotype), effects that were associated with reduced average daily food intake (by 22–30% on both diets, P < 0.0001) and a modest increase in the rate of oxygen consumption (by 25%, P < 0.003) during the light cycle. Moreover, in mice fed the HF diet, HL deficiency reduced both body fat (by 30%, P < 0.0001) and adipocyte size (by 53%, P < 0.01) and fully prevented the development of hepatic steatosis. Also, HL deficiency reduced adipose tissue macrophage content, consistent with reduced inflammation and a lean phenotype. Our results demonstrate that in mice, HL deficiency protects against diet-induced obesity and its hepatic sequelae. Inhibition of HL-activity may therefore have value in the prevention and/or treatment of obesity.

Hepatic lipase is required for the control of energy balance and body fat accumulation.

Hepatic lipase (HL) is an important lipid processing enzyme that influences inflammation (1), circulating lipoprotein levels, and susceptibility to atherosclerosis (2). A critical component of lipid homeostasis (2,3), HL hydrolyzes triacylglycerols (TAG) in plasma lipoproteins to form smaller particles for receptor-mediated lipoprotein uptake (3). In the process of lipoprotein hydrolysis, HL generates free fatty acids (FFAs) that may be oxidized for immediate energy needs in tissues such as muscle and heart, or reesterified to TAG for energy storage (3). Also, FFAs serve as nutrient signals to the central nervous system that can influence neurocircuits governing food intake (FI) and energy balance and thereby help to determine the defended level of body fat stores (4,5).

In humans, postheparin plasma HL activity increases with increasing visceral adiposity, as reflected by measures of intraabdominal fat (6,7), whereas conversely, loss of intraabdominal fat is associated with reduced HL activity (8). Although these findings are typically interpreted to suggest that obesity causes an increase of HL expression or activity, it remains possible that increased HL favors obesity. Consistent with the latter hypothesis, HL is linked to obesity risk in molecular genetic studies in mice. Specifically, quantitative trait loci on mouse chromosomes 2, 6, and 7 are coincident for percent body fat, HL activity, and cholesterol (9,10). Also, genetic crosses between the Western wild mouse, Mus musculus (includes the C57Bl6 strain) and the Algerian mouse, Mus spretus (11), two species that diverged genetically over 1 million years ago, suggest an influence of mouse HL alleles on obesity (12). If increased plasma HL activity favors weight gain, HL deficiency might be expected to reduce body fat stores and protect against the development of obesity. To test this hypothesis, we compared mice lacking endogenous HL (hl^−/−) to wild-type (WT) controls fed either a regular chow diet or an obesigenic high-fat (HF) diet. To better understand how HL deficiency impacts energy homeostasis, we measured daily and cumulative food intake as well as parameters of energy metabolism including oxygen consumption, respiratory quotient (RQ), heat production, and locomotor activity after 16 wk of either diet. The results of these studies, combined with measures of body composition, plasma glucose, liver fat content, and related parameters, suggest that HL deficiency reduces body fat mass and protects against diet-induced obesity (DIO) and its hepatic sequelae.

Materials and Methods

Mice

WT and hl^−/− mice (13) that were at least 94% backcrossed to the C57/Bl6 background were studied. Genetic knockout of HL was confirmed by PCR (14). All mice were female and were housed individually in microisolator cages in a modified barrier facility with a 12-h light, 12-h dark cycle. All animal studies were approved by the Institutional Animal Care and Use Committee of the University of Washington and were conducted in accordance with accepted guidelines published in the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Dietary intervention and metabolic studies

Body weight and FI were measured daily for 12 wk in groups of mice of each genotype (WT or hl^−/−; n = 4–7 per group) that were fed either regular mouse chow containing 4.5% fat (PicoLab Rodent Diet 20, no. 5053, PMI feeds; PicoLab, St. Louis, MO) or a HF diet containing 21% fat and 0.15% cholesterol (TD 88137; Harlan Teklad, Madison, WI). Body composition was determined in 21- to 23-wk-old mice by quantitative magnetic resonance (Echo Medical Systems LLC, Houston, TX), a technique that precisely and accurately measures fat mass, lean mass, and body fluids in unanesthetized animals (15). Rates of oxygen consumption (VO₂, normalized to lean body mass), CO₂ production, RQ, heat production, and locomotor activity were measured continuously by indirect calorimetry over 36 h using the Comprehensive Lab Animal Monitoring System (Columbus Instruments Co., Columbus, OH) located within the Animal Studies Core Laboratory of the Clinical Nutrition Research Unit at the University of Washington. Calorimetry data were obtained after 16 wk of either diet in both groups of mice (chow, n = 4 per group; HF, n = 5–7 per group), all mice were between 21 and 23 wk old at time of measurement.

Adipose tissue morphology

Formalin-fixed adipose tissue from four mice of each genotype was embedded in paraffin, cut into 5-μm sections, and stained with hematoxylin and eosin as described (16). Four noncontiguous sections were analyzed per mouse. Adipocyte cross-sectional area was determined using computer image analysis as described (17,18).

Mac2 immunostaining was used to visualize macrophages. Deparaffinized sections were incubated with antimouse-Mac2 antibody (1:5000; Cedarlane Labs, Burlington, NC) followed by biotinylated antirat IgG (H+L) antibody (1:500). Slides were incubated with horseradish peroxidase (Vectastain Elite ABC Kit), developed with a chromogen (Vector Substrate Kit, Vector Laboratories, Burlingame, CA), and counter stained with hematoxylin. Four noncontiguous sections were analyzed per mouse. The number of Mac2 stained macrophages per field was quantitated using Image Pro software, Media Cybernetics (Bethesda, MD).

Liver TAG content and morphology

Liver tissue from chow and HF fed mice of each genotype (chow: WT, 6; hl^−/−, 4; HF: WT, 5; hl^−/−, 4) was weighed and dissolved in potassium hydroxide and liberated glycerol (used to estimate TAG) measured in quadruplicate with the Triglyceride GPO-PAP kit (19,20,21). In some mice, we assessed hepatic lipid accumulation histochemically. Livers were flash-frozen in liquid nitrogen and embedded in OCT compound, and 10-μm sections were cut on a cryostat (Leica Microsystems, Bannockburn, IL). Tissue sections were stained with eosin, and nuclei were counterstained with hematoxylin. Stained sections were photographed using a Nikon Coolpix camera through an inverted microscope (Leica DMIL, Leica Camera AG, Solms, Germany).

Plasma measurements

For mice of each genotype on the chow diet, fasting plasma was obtained at 5 wk of age for measurement of glucose and insulin (n = 4 per group), and at 21 wk of age for leptin (n = 6 per group). Glucose levels vary minimally over 52 wk in chow-fed female C57Bl/6 mice (Mouse Phenome Database available at http://www.jax.org/phenome). For mice on the HF diet, plasma was obtained at 25 wk of age after completion of cumulative FI studies (n = 5 per group). Blood glucose was measured using a hand-held glucometer (True Track Smart System; Home Diagnostics, Ft. Lauderdale, FL). Plasma insulin levels were measured using an ELISA immunoassay kit (ALPCO, Salem, NH) and leptin levels by RIA (Linco-Millipore, Billerica, MA), performed in the Laboratory Core of the Clinical Nutrition Research Unit at the University of Washington. Plasma cholesterol, triglyceride, and FFA concentrations were measured using standard colorimetric assays (cholesterol: Chol 7D62, Abbott Clinical Chemistry, Wiesbaden, Germany; TAG: TG kit, Roche/Hitachi, Roche Diagnostics, Mannheim, Germany; FFA: NEFA C kit, WAKO Chemicals, Richmond, VA). Plasma thyroid hormone (T₄) levels were measured by RIA at Phoenix Central Laboratory (Everett, WA).

Dynamic tests of glucose metabolism

Mice of each genotype were subjected to an ip glucose tolerance test (IPGTT) and an insulin tolerance test (ITT) between either 7–9 wk of age (among chow-fed mice; n = 4 per group) or 37–39 wk of age (among mice fed the HF diet, after 8–10 wk of the diet; n = 7 per group). The IPGTT was performed using a glucometer to measure blood glucose levels at 0, 30, 60, and 120 min (True Track Smart System Home Diagnostics, Ft. Lauderdale, FL) after ip injection of glucose (1 g/kg body weight, after an overnight fast). Area under the curve analysis of the glucose response to the IPGTT was determined using the trapezoidal rule (22). The ITT was performed by measuring blood glucose values at 0, 30, and 45 min after injection of insulin (1.3 U/kg ip, chow diet; n = 4 per group; 2 U/kg ip, HF diet; n = 7 per group; Humulin R; Eli Lilly, Indianapolis, IN) in mice fasted for 4 h.

Statistical analyses

All data are shown as mean ± sd unless stated otherwise. Comparisons of group mean values from mice of different genotypes on different diets were made by two-way ANOVA. For two group comparisons, a Student’s t test was used. Statistical significance was set at P < 0.05.

Results

Body weight changes in hl^−/− and WT mice fed low-fat and HF diets

To examine the effect of HL deficiency on body weight regulation, we studied mice of both genotypes during a 12-wk period on either a low-fat (chow) or a HF diet. Although body weight did not differ significantly by genotype at baseline (chow, hl^−/−, 16.4 ± 1.3 g vs. WT, 18.7 ± 1.1 g; P = not significant; HF, hl^−/−, 19.6 ± 1.3 vs. WT, 19.0 ± 2.1, P = not significant), weight gain over time was attenuated by 32 and 50% in hl^−/− mice fed the low-fat (chow) and HF diets, respectively, compared with WT controls (chow, hl^−/−, 4.8 ± 0.8 g vs. WT, 7.1 ± 0.9 g, and HF, hl^−/− 4.9 ± 1.4 g vs. WT 9.7 ± 3.0 g; Fig. 1, A–C). Two-way ANOVA demonstrated a major contribution of genotype to the difference in weight gain (P < 0.0005), irrespective of diet. Thus, hl^−/− mice exhibit reduced weight gain on both a standard chow and a HF diet, with the effect being somewhat greater in animals on the latter diet. Because there was no effect of genotype on body length, observed differences in weight were not attributable to differences in growth rate (Table 1).

Figure 1.

Weight gain in WT and hl^−/− female mice on chow (A and C) and HF (B and C) diets. WT mice (n = 6–7 per group) and hl^−/− mice (n = 5–6 per group) were fed either a chow (4.5% fat) diet or a HF (21% fat) diet for 12 wk. Cumulative weight gain is plotted in A and B. C, The weight gain is shown as the difference between the body weights on the first and last days of the 12-wk study; P < 0.0002 for contribution of genotype by two-way ANOVA on d 83 of diet. hl^−/−, HL-deficient mice.

Table 1.

Body length, body weight, percent body fat, and liver weight in WT and hl^−/− female mice on chow and HF diets

Diet	Chow		HF
	WT	hl−/−	WT	hl−/−
Length (cm)a	10.0 ± 0.2	9.9 ± 0.2	10.6 ± 0.2b	10.5 ± 0.3b
Body weight (g)
Baseline	18.7 ± 1.1	16.4 ± 2.0	19.0 ± 2.2	19.6 ± 1.3
End	25.8 ± 1.0	21.4 ± 1.7c	28.7 ± 4.1	24.5 ± 2.2c
Δ Body weight	7.1 ± 0.9	4.5 ± 0.3	9.7 ± 3.0d	4.9 ± 1.4d
Body fat (%)e	15.8 ± 1.2	15.6 ± 1.8	34.0 ± 6.1f	23.4 ± 3.0f
Liver weight (g)g	1.2 ± 0.1	1.0 ± 0.1	1.6 ± 0.5h	1.2 ± 0.2h
Liver TAG (%)	19.6 ± 1.5	19.0 ± 2.1	40.3 ± 3.5i	20.2 ± 4.7i

^aChow: WT, n = 6, hl^−/−, n = 4; all 25 weeks; HF: WT, n = 5, hl^−/−, n = 4; all 31–33 wk; 

^bP < 0.001 for diet; 

^cP < 0.001 for genotype and <0.01 for diet; 

^dP < 0.0002 for genotype; 

^eChow: WT, n = 6, hl^−/−, n = 4; all 21–23 wk; HF: WT, n = 7, hl^−/−, n = 4; all 21–23 wk; 

^fP < 0.05 for genotype; 

^gAs in ^a except for Chow: WT, n = 5; 

^hP < 0.05 for genotype; 

ⁱP < 0.01 for genotype and diet. 

Body adiposity in hl^−/− and WT mice on low-fat and HF diets

To determine whether the reduced weight gain of hl^−/− mice was associated with reduced fat or lean mass, we measured body composition after 16 wk on either diet. Although the latter diet increased percent body fat content significantly in both genotypes, the magnitude of this effect was reduced by nearly 30% in mice with HL deficiency compared with controls (hl^−/−, 23.8 ± 6% vs. WT, 33.4 ± 6.6% body weight; P < 0.001). By comparison, genotype differences in body fat content did not achieve statistical significance in mice fed the chow diet (Table 1).

Adipose tissue cell size and macrophage content

To further characterize the body adiposity phenotype of hl^−/− mice, we assessed the size of gonadal fat pads in hl^−/− and control mice fed the HF diet. Consistent with the lean phenotype of hl^−/− mice, the size of adipocytes in their gonadal fat pads was reduced by 53% compared with WT controls (hl^−/−, 2996 ± 1271 vs. WT, 6404 ± 1292 μm²; P < 0.01), as shown histologically in Fig. 2, A and B, and quantitatively in Fig. 2E. Interestingly, Mac 2 staining demonstrated a marked decrease in the number of adipose tissue macrophages (ATMs) in gonadal adipose tissue from hl^−/−, vs. WT mice, (hl^−/−, 0.7 ± 0.5 macrophages per field vs. WT, 4.5 ± 1.4 macrophages per field; P < 0.005; Fig 2, C, D, and F).

Figure 2.

Adipose tissue characteristics in WT and hl^−/− mice. A and B, Photomicrographs of H&E stained gonadal adipose tissue from HF fed mice. Size bar, 100 μm. Note reduced cell size, increased number of nuclei in hl^−/− mice. E, Quantitation of adipocyte size, P < 0.01 vs. WT mice. ATMs in WT and hl^−/− mice. C and D, Photomicrographs of Mac2 stained gonadal adipose tissue from HF fed mice. Size bar, 100 μm. Note increased number of macrophages in WT mice. F, Quantitation of macrophages, P = 0.005 vs. WT mice.

Cumulative FI

To establish whether the reduced weight gain in hl^−/− mice resulted from reduced energy intake, we measured daily FI in both genotypes of mice over a 12-wk period on both chow and HF diets. As illustrated in Fig. 3, A and B, HL deficiency was associated with a 23–33% reduction of cumulative FI at 12 wk (d 83) on both diets. Two-way ANOVA demonstrated a major contribution of genotype (P < 0.0001) and a minor contribution of diet (P < 0.05) to the reduction in cumulative FI. Average daily FI was reduced comparably on both diets in hl^−/− mice (average daily FI decreased by 30% on the chow diet and by 22% on the HF diet; P < 0.0001; Fig. 3C).

Figure 3.

Cumulative FI in WT and hl^−/− female mice on chow and HF diets. Daily food intake was measured for 12 wk during (A) chow diet (WT, n = 4; hl^−/−, n = 6) and (B) HF diet (WT, n = 7; hl^−/−, n = 5). Results are reported as group means ± se for each day and is expressed as kilocalories. Two-way ANOVA on d 83 of each diet indicates a major contribution of genotype to the food intake phenotype; *, P < 0.0001. C, Average daily food intake in WT and hl^−/− female mice on chow and HF diets. Two-way ANOVA demonstrates a significant contribution of genotype to the daily FI phenotype; *, P < 0.0001. All measurements were obtained between 0830 and 1030 h. The FI is expressed as kilocalories consumed per day.

Energy metabolism

To investigate whether increased energy expenditure also contributes to reduced weight gain in mice with HL deficiency, we measured rates of oxygen consumption (VO₂ normalized to lean body mass), RQ, heat production, as well as locomotor activity in WT and hl^−/− mice on each of the two diets. Interestingly, HL deficiency was associated with a moderate and significant increase in rates of oxygen consumption particularly during the light cycle (VO₂: hl^−/−, 5124 ± 194 vs. WT, 4075 ± 157 ml/kg lean body mass per hour) with strong effects of genotype (P = 0.002) and diet (P = 0.003) (Fig. 4, A and B, and supplemental Table S1, published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org). The average RQ was also increased in hl^−/− mice during the light cycle (hl^−/−: 0.966 ± 0.064, WT: 0.913 ± 0.047), with a modest contribution of genotype (P < 0.03) and a substantial contribution of diet (P < 0.0001; Fig. 4, C and D, and supplemental Table S1). The increased VO₂ and RQ are consistent with increased energy expenditure, mainly as a result of increased glucose oxidation. Because there was a trend toward reduced locomotor activity in the HL-deficient mice (on chow but not on HF diet), the increased energy expenditure exhibited by these animals was not due to increased physical activity (supplemental Table S1).

Figure 4.

Oxygen consumption and RQ in WT and hl^−/− mice during dark (DC1, DC2) and light (LC) cycles. A and C, Chow diet. B and D, HF diet; two-way ANOVA for average VO₂ yields *, P < 0.005 for the contribution of genotype during LC, and *, P < 0.04 on DC2; **, P < 0.005 for the contribution of diet during LC and DC2. Two-way ANOVA for RQ reveals a *, **, P < 0.03 for the contribution of genotype during LC and P < 0.0001 for the contribution of diet during LC and DC2. E–H, Reduced diurnal variation in oxygen consumption (E), RQ (F), heat production (G), and locomotor activity (H) in WT and hl^−/− female mice on chow and HF diets. Diurnal variation is expressed as the ratio of light cycle and the mean of the two dark cycles [DC = (DC1 + DC2)/2] measurements of VO₂, RQ, heat production, and locomotor activity. A ratio of one indicates complete lack of diurnal variation (chow: WT, n = 6, and hl^−/−, n = 4, HF: WT, n = 7; hl^−/−, n = 5). On the chow diet, two-way ANOVA demonstrates a significant contribution of genotype to the diurnal variation in oxygen consumption and RQ (P = 0.002) as well as a significant contribution of genotype to the diurnal variation in heat expenditure and locomotor activity (P < 0.03).

Diurnal variation in hl^−/− mice

Because the increased rates of oxygen consumption and RQ in HL-deficient mice was most prominent during the light cycle (when physical activity, feeding and O₂ consumption are relatively low in normal mice), we reasoned that diurnal variation in this parameter might also be disrupted by HL deficiency. To investigate this possibility, we determined ratios of light-cycle to dark-cycle values of VO₂, RQ, heat production, and locomotor activity from both genotypes of mice on each of the two diets. In accordance with our reasoning, hl^−/− mice displayed higher light-dark cycle ratios (e.g. less of a reduction during the light cycle) for each of these parameters, but only while fed the chow diet (Fig. 4, E–H) .

Plasma measurements

Next, we examined whether genotype differences in energy balance were accompanied by changes in plasma levels of glucose, FFAs, insulin, or leptin. Consistent with their decreased body adiposity, leptin levels were reduced in HL-deficient mice, an effect that was particularly pronounced on the HF diet (hl^−/−, 8.5 ± 2.8 vs. WT, 26.5 ± 5.4 ng/ml; P < 0.03) (Table 2). FFA concentrations were significantly increased in fasting plasma of chow-fed HL-deficient mice (hl^−/−, 0.61 ± 0.21 vs. WT, 0.27 ± 0.11 mEq/ml; P < 0.04) (Table 2) consistent with reduced fatty acid oxidation under these conditions.

Table 2.

Plasma metabolic measurements in WT and hl^−/− female mice

Diet	Chow		HF
	WTa	hl−/−a	WTb	hl−/−b
T4(μg/dl)	3.3 ± 0.9	3.4 ± 1	3.3 ± 0.5	4.1 ± 0.7
Glucose (mm)	7.8 ± 0.6	6.6 ± 0.8	9.7 ± 3.1c,e	9.3 ± 1.3d,e
Insulin (pmol/liter)	47.9 ± 4.9	20.8 ± 4.9	36.1 ± 18.6	41.0 ± 25.7
Leptin (ng/ml)	2.8 ± 1.9c	1.1 ± 0.6c,f	26.5 ± 12.1	8.4 ± 6.8f
FFA, fasting (mEq/ml)	0.27 ± 0.11	0.61 ± 0.21g	0.36 ± 0.06	0.29 ± 0.21
FFA, nonfasting (mEq/ml)	1.44 ± 0.34	0.55 ± 0.26h	N.A.	N.A.
Cholesterol (mm)	2.3 ± 1.4	3.1 ± 1.1	4.3 ± 1.5	4.9 ± 0.9
TAG (mm)	0.8 ± 0.2	0.7 ± 0.1	0.4 ± 0.1	0.2 ± 0

N.A., Not available. 

^an = 4; 

^bn = 5; 

^cn = 6; 

^dn = 7; 

^eP < 0.02 vs. mice on chow diet; 

^fP < 0.003 vs. WT mice on both diets; 

^gP < 0.04 vs. fasting WT mice on chow diet; 

^hP < 0.01 vs. nonfasting WT mice on chow diet. 

Despite clear differences in body weight and fat mass, differences in fasting plasma glucose levels were not detected between genotypes on either chow or the HF diet (Table 2). As expected, HF feeding increased fasting glucose levels significantly in both genotypes and, although hl^−/− mice tended to have lower plasma glucose levels on both diets, this effect did not achieve statistical significance. Similarly, plasma TAG levels did not differ significantly between genotypes on either diet, and the expected reduction of plasma TAG levels during HF feeding was similar between hl^−/− mice and controls. By comparison, fasting plasma insulin levels were reduced in hl^−/− mice on the chow diet, but this reduction did not persist on the HF diet, despite reduced body fat mass and plasma leptin levels (Table 2). Because of the prominent role of thyroid hormone in control of metabolic rate, we also investigated whether increased thyroid hormone levels (T4) contributed to the hypermetabolic phenotype in hl^−/− mice. No significant differences in plasma T4 levels were detected between the genotypes (Table 2).

Liver TAG content

Because DIO in C57/Bl6 mice is accompanied by hepatic TAG accumulation (steatosis), we sought to evaluate whether deficiency of HL confers protection against this effect. We observed a striking effect of HL deficiency to reduce liver TAG content on the HF diet, compared with WT mice (WT, 404 ± 35 mg/g liver vs. hl^−/−, 202 ± 47 mg/g liver), whereas genotype differences in liver TAG levels were less striking on the chow diet (WT, 198 ± 12 mg/g liver vs. hl^−/−, 190 ± 21 mg/g liver; Fig. 5, right panel). Two-way ANOVA revealed P < 0.003 for both genotype and diet. Findings from biochemical analysis of liver TAG content were confirmed by histochemical evidence of large vacuoles consistent with steatosis in WT (Fig. 5, left upper panels), but not hl^−/− mice (Fig. 5, left lower panels) fed the HF diet.

Figure 5.

Left panel, Hepatic steatosis in WT but not in hl^−/− female mice (n = 3 per group) on HF diet. Liquid nitrogen-fixed liver tissues were embedded in OCT compound, and 10-μm sections were cut and stained with hematoxylin and eosin. Note the large vacuoles representing fat in the WT but not in the hl^−/− liver tissue. CV, central vein. Size bar represents 100 μm. Right panel, Liver TAG content in WT and hl^−/− female mice on chow and HF diets. Liver tissue was weighed and dissolved with 3 m potassium hydroxide in 65% ethanol. Liberated glycerol (to estimate TAG) was measured in quadruplicate with the Triglyceride GPO-PAP kit. (chow: WT, n = 6, hl^−/−, n = 4; HF: WT, n = 5, hl^−/−, n = 4). Two-way ANOVA demonstrates major contributions of both genotype and diet (both *, P < 0.003) to the liver TAG phenotype.

Dynamic tests of glucose metabolism

To determine whether HL deficiency also has beneficial effects on insulin sensitivity and glucose metabolism, we performed glucose and insulin tolerance tests in both genotypes of mice on both diets. Despite their decreased body weight and adiposity, genotype differences were not detected in either the IPGTT (confirmed by comparison of area under the curve glucose measurements) or the ITT in mice on either diet, although mutant mice displayed a trend toward a greater blood glucose reduction during the ITT compared with control mice on the HF diet (supplemental Fig. S1). Thus, reduced body fat mass is not associated with clear improvement in insulin sensitivity or glucose tolerance in hl^−/− mice.

Discussion

Although HL is clearly implicated in the control of lipoprotein metabolism, a role in energy homeostasis has not previously been described. Here, we report that although HL-deficient mice exhibit normal body weight at a young age, over time they gain less weight and body fat than controls on a standard chow diet, and this effect is more prominent when mice are fed an obesigenic HF diet. On the latter diet, reductions of body weight in hl^−/− mice were associated with significant decreases of percent body fat and fat cell size compared with WT controls, indicating that HL deficiency protects against DIO. The mechanism underlying observed differences of body weight and fat mass involve both a modest increase of energy expenditure and a more substantial decrease of FI. Whereas HL deficiency strongly protected against obesity-induced hepatic steatosis, it had surprisingly little impact on measures of glucose metabolism. We conclude that HL is required for the control of energy balance and body fat accumulation. Furthermore, our results suggest that the effect of DIO to cause hepatic steatosis, but not insulin resistance or glucose intolerance, is dependent on HL.

The lean phenotype of the HL-deficient mice may relate to the role of HL in inflammation (1). HL is present in macrophages (23), inflammatory cells that infiltrate adipose tissue during adipocyte enlargement such as occurs with HF diet treatment. One could envision that HL activity provided by macrophages liberates FFAs from plasma lipoproteins that can be taken up by nearby adipocytes, where they can be reesterified and stored. In the absence of macrophages and HL activity, there would be less FFA available, thus reducing adipocyte growth. Our findings of decreased ATM infiltration along with resistance to DIO are similar to observations in HF fed mice deficient in phosphodiesterase 4B (PDE4B) (24). PDE4B belongs to a family of phosphodiesterases that hydrolyze cAMP and regulate the majority of cellular processes. Recent data indicate that PDE4B is the major isoform that mediates inflammatory responses. Similar to hl^−/− mice, PDE4B null mice display a lean phenotype, reduced adipocyte size, and reduced macrophage infiltration in adipose tissue (24). In view of the emerging roles of both HL and PDE4B in inflammation, it is reasonable to hypothesize that the lean phenotype and protection against DIO in HL- and PDE4B-deficient mice may be a consequence of reduced inflammation, as suggested by the reduced number of ATMs in both models.

However, a lean phenotype and protection against the development of DIO has also been observed in several other mutant mouse models in which a key molecule involved in lipid metabolism is deleted. These include the very low density lipoprotein receptor (25,26), acetyl-CoA carboxylase-2 (27,28), steaoryl-CoA desaturase-1 (29,30,31), acylCoA-diacyl glycerol acyltransferase-1 (28), and hormone sensitive lipase (32). Each of these mouse models exhibits a lean, hypermetabolic phenotype with DIO resistance and, although these various molecules contribute in their own distinct way to the control of lipid metabolism and fat deposition, in no case is the basis for the energy homeostasis phenotype fully understood. The very low density lipoprotein receptor transfers exogenous FFAs generated by intravascular hydrolysis to the adipocyte to provide substrates for TAG synthesis (25,26), whereas acetyl-CoA carboxylase-2 directs FA to TAG synthesis and away from mitochondrial oxidation by generating malonyl-CoA, a precursor for fatty acid synthesis that also inhibits carnitine palmitoyltransferase-1 (27). Steaoryl-CoA desaturase-1 provides one of the major substrates for TAG synthesis by unsaturating steaoryl-CoA to generate oleoyl-CoA (30,31), whereas acylCoA-diacyl glycerol acyltransferase-1 catalyzes the final step in TAG synthesis (28) and hormone sensitive lipase catalyzes the hydrolysis of intracellular TAG in response to hormonal and other stimuli (32).

Although our findings suggest that HL should now be added to the list of proteins involved in lipid metabolism that are required for DIO, mice with HL deficiency display several features not observed in these other mouse models. For one, reduced FI is a predominant feature in hl^−/− mice irrespective of diet, which is not the case in other models, and the increase of metabolic rate in hl^−/− mice is small in comparison with the other models (26,27,28,29,31,32). Similarly, increased insulin sensitivity is a feature of each of the other models, but was not evident in hl^−/− mice. Indeed, the finding that during HF feeding, HL-deficient mice have reduced body fat and yet are not more insulin sensitive or glucose tolerant than WT controls raises the possibility that HL is a determinant of insulin sensitivity and that during DIO, increased levels of activity of HL exerts an insulin-sensitizing effect in peripheral tissues, favoring fat accumulation. The combination of reduced FI and low plasma leptin levels in hl^−/− mice raises the possibility that the lean phenotype of these mice involves an increase of leptin sensitivity. Whether this mechanism might also explain our observation of reduced diurnal variation in oxygen consumption, RQ, heat production, and locomotor activity awaits additional study.

In summary, our studies demonstrate that HL deficiency reduces body fat mass and protects against the development of obesity primarily by reducing FI, but also by preventing the decrease of energy expenditure that normally accompanies reduced fat stores. These findings may result in part from a reduction in inflammation associated with HL deficiency. Based on these results, we propose that modalities for down-regulating HL activity may have beneficial effects not only on lipoprotein metabolism, but on energy homeostasis as well.