The Adipose Tissue Phenotype of Hormone-Sensitive Lipase Deficiency in Mice

Authors


Medical Genetics Service, Sainte-Justine Hospital, 3175 Côte Ste-Catherine Road, Montréal, Québec, Canada H3T 1C5. E-mail mitchell@justine.umontreal.ca

Abstract

Objective: To directly ascertain the physiological roles in adipocytes of hormone-sensitive lipase (HSL; E.C. 3.1.1.3), a multifunctional hydrolase that can mediate triacylglycerol cleavage in adipocytes.

Research Methods and Procedures: We performed constitutive gene targeting of the mouse HSL gene (Lipe), subsequently studied the adipose tissue phenotype clinically and histologically, and measured lipolysis in isolated adipocytes.

Results: Homozygous HSL−/− mice have no detectable HSL peptide or cholesteryl esterase activity in adipose tissue, and heterozygous mice have intermediate levels with respect to wild-type and deficient littermates. HSL-deficient mice have normal body weight but reduced abdominal fat mass compared with normal littermates. Histologically, both white and brown adipose tissues in HSL−/− mice show marked heterogeneity in cell size, with markedly enlarged adipocytes juxtaposed to cells of normal morphology. In isolated HSL−/− adipocytes, lipolysis is not significantly increased by β3-adrenergic stimulation, but under basal conditions in the absence of added catecholamines, the lipolytic rate of isolated HSL−/− adipocytes is at least as high as that of cells from normal controls. Cold tolerance during a 48-hour period at 4 °C was similar in HSL−/− mice and controls. Overnight fasting was well-tolerated clinically by HSL−/− mice, but after fasting, liver triglyceride content was significantly lower in HSL−/− mice compared with wild-type controls.

Conclusions: In isolated fat cells, the lipolytic rate after β-adrenergic stimulation is mainly dependent on HSL. However, the observation of a normal rate of lipolysis in unstimulated HSL−/− adipocytes suggests that HSL-independent lipolytic pathway(s) exist in fat. Physiologically, HSL deficiency in mice has a modest effect under normal fed conditions and is compatible with normal maintenance of core body temperature during cold stress. However, the lipolytic response to overnight fasting is subnormal.

Introduction

Hormone-sensitive lipase (HSL) (E.C. 3.1.1.3) is an 84-kDa cytoplasmic protein of adipocytes, in which it catalyzes the hydrolysis of triacylglycerols. In adipocytes, HSL- mediated lipolysis is activated via a series of events, including β-adrenergic stimulation, a complex combination of cyclic adenosine monophosphate-dependent serine phosphorylations mediated by protein kinase A (1) (2), translocation of HSL from the adipocyte cytoplasm to the surface of the lipid droplet (3) and the reciprocal shift to the cytoplasm of perilipin A, a lipid droplet surface protein that is also phosphorylated by protein kinase A (4). HSL has also been reported to associate with a specific docking protein, lipotransin (5). The HSL gene (6) (7) (8) (locus LIPE in humans and Lipe in mice) maps to human chromosome 19 (9) and mouse chromosome 7 (10) (11). Exon 1, the longest coding exon, is present in all known HSL isoforms. For decades, attention has focused on HSL as the principal regulator of lipolysis in adipose tissue and as a potential candidate molecule for obesity (12). HSL is also expressed in other tissues (13) including skeletal muscle, myocardium, adrenal gland, macrophages (14) (15), pancreatic β cells (16), and male germ cells (17). A 120-kDa testis-specific HSL isoform also arises from the LIPE locus (18). In addition to triglycerides, HSL can cleave fatty acyl esters of cholesterol (19), retinoic acid (20), and steroid hormones (21).

Studies have shown an increase in adipose tissue HSL levels and activity under lipolytic conditions such as fasting (22) (23), pregnancy (24), and hibernation (25). To study the importance of HSL in fat and other tissues directly, we created HSL-deficient mice by gene targeting.

Research Methods and Procedures

Production of Targeting Constructs

We made a replacement vector using a 6741-base pair (bp) AflII/BamHI subclone of the previously described 129 strain mouse HSL genomic clone (7). The subclone extends from 763 bp upstream of the initiation ATG to exon 7. We created a 2009-bp deletion in the subclone, spanning the 3′ 513 bp of exon 1 plus 1494 bp of intron 1, and inserted in its place a promotorless lacZ gene in frame with mouse HSL exon 1. Downstream from this is a herpes simplex thymidine kinase promoter that drives a bacterial neomycin resistance (Neo) gene, derived from pMC1Neo poly(A) (Stratagene, La Jolla, CA). The final targeting construct has a 763-bp short arm and a 4082-bp long arm.

Targeting and Culture of Embryonal Stem (ES) Cells

Electroporation was performed using a 10-μg linearized vector and 107 J1 ES cells (26) at 400 V and 25 μF in a 0.4-cm Gene Pulse cuvette (Bio-Rad, Hercules, CA). Cells were cultured for 24 hours in Dulbecco's modified Eagle's medium (Life Technologies, Gaithersburg, MD) with 0.1 mM nonessential amino acids (Life Technologies), 0.1 mM β-mercaptoethanol, 2 mM GlutaMax (Life Technologies), 15% ES cell qualified fetal calf serum, 50 U/mL penicillin, 50 μg/mL streptomycin, and 500 U/mL leukemia-inhibiting factor (Life Technologies) on EFneo cells. Selection was performed for 7 days in the same medium supplemented with 200 μg/mL G418 (Life Technologies).

HSL Genotyping by Polymerase Chain Reaction (PCR) and Southern Blotting

We screened for gene targeting using pools of four G418-resistant colonies. ES cell lysates were digested with proteinase K and directly used for PCR amplification. The sense primer LipM32 (5′-AGGCAGGAGAATGGAACAGT-3′), which is outside the targeting construct, corresponds to residues −814 to −834 upstream of ATG; the antisense primer LacZ-3 is complementary to nucleotides 167 to 189 of the β-galactosidase gene. PCR conditions were as described previously (27) except that the above primers were used. Amplification of a 1-kilobase (kb) fragment was diagnostic for the presence of a targeted allele.

We confirmed the genotypes of ES cell clones by Southern blots. Mouse genomic DNA (5 μg) was digested with BamHI and probed either with a 930-bp BamHI/AflII fragment that is located upstream of the short arm of the targeting vector or with a Neo cDNA. Using these probes, the targeted allele produces a 6.2-kb fragment and the normal allele (7.7 kb) (Figure 1). For genotyping of mice, tail DNA was similarly analyzed.

Figure 1.

Targeted disruption of the HSL gene. (a) Targeting scheme. (Top) genomic organization of the HSL locus with coding regions shaded. (Center) the targeting vector. (Bottom) the targeted allele, showing the diagnostic BamHI restriction fragments of the targeted (6.2 kb) and wild-type (7.7 kb) alleles. B, BamHI; LacZ, β-galactosidase cDNA; Neo, neomycin resistance cassette. (b) Genomic Southern blot analysis showing the diagnostic BamHI fragments in 5 μg of DNA from mice of each HSL genotype. (c) Northern blot analysis of adipose tissue and liver RNA. (Top) Northern blot using the HSL cDNA probe. Lanes contain 10 μg of whole-cell RNA from HSL+/+ and HSL+/− mice and 20 μg of whole-cell RNA from HSL−/− mice and from the liver (Liv), a nonexpressing tissue. Bottom, the same filter, stripped and rehybridized with a control probe, β-actin. (d) Western blot analysis of adipose HSL. In each lane, 30 μg of gonadal WAT protein is present. HSL genotypes are indicated above each lane.

Production and Analysis of HSL-Deficient Mice

Targeted ES cell clones were injected into BALB/c blastocysts and transferred to pseudopregnant recipients. Heterozygous F1 offspring of a male chimera with the targeted HSL allele were crossed with BALB/c females to produce the mice described. Mice were fed a regular chow diet and raised in a 12-hour light/dark cycle. They were weighed at 3 and 6 weeks and at 3, 4, 5, and 6 months. Serological testing for infections in the colony was negative. Mice were fasted overnight, anesthetized with methoxyflurane (Janssen Pharmaceuticals, Toronto, Canada), exsanguinated by cardiac puncture, and decapitated; their organs were rapidly removed for study. Except where indicated, mice were 6 months old at the time of metabolic studies and 4–6 months old when killed for histological studies.

HSL Assay and Western Blotting

HSL was assayed as neutral cholesteryl esterase (13). Adipocyte proteins for Western blotting were prepared as follows. A solution of 10 mM tris(hydroxymethyl)aminomethane-HCl (pH 7.4), 1 mM EDTA, 20 μg/mL leupeptin (L-069; Sigma, St. Louis, MO), 2 μg/mL antipain (A-6191; Sigma), and 1 μg/mL pepstatin A (P-5318; Sigma) was added to adipose tissue in a ratio of 1:15 (w/v) and homogenized vigorously (Wheaton Tissue Grinder; VWR Conlab, Mississauga, Canada). Five volumes of chloroform:methanol (2:1) were then added. After vigorous stirring, the mixture was incubated on ice for 1 hour and then centrifuged at 5000 rpm for 10 minutes. After removal of the liquid phases, the interface region was allowed to dry. A solution of 50 mM tris(hydroxymethyl)aminomethane-HCl (pH 6.8), 1.6% bromophenol Blue, 8% glycerol, 100 mM dithiothreitol, and 0.5% sodium dodecyl sulfate was added. After an overnight incubation at 4 °C, samples were used for sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Western blotting was performed using a 1/60 dilution of an affinity-purified polyclonal anti-rat HSL raised in rabbits, kindly provided by C. Londos (National Institutes of Health, Bethesda, MD).

RNA Isolation, Northern Blot Analysis, and RNA Quantitation

After RNA isolation from frozen adipose tissue (28), Northern blotting was performed as described previously (29) using a mouse HSL cDNA probe (residues 545 to 869). This fragment was obtained by reverse transcription-PCR using primers LipM35 (5′-CAGACTCTCCTCCATCGGGCT-3′), corresponding to residues 545 to 565 of the mouse HSL cDNA, and LipM36 (5′-TGAGATGGTAACTGTGAGCC-3′), complementary to residues 849 to 869. HSL mRNA was quantitated with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA), using mouse β-actin cDNA as a control probe.

Metabolite and Hormone Measurements

Metabolites were measured using the following kits: nonesterified fatty acids (kit no. 994-75409E; Wako Chemicals, Neuss, Germany), glucose (Medisense Blood Glucose Meter, Bedford, MA), triglycerides and glycerol (1488872; Roche Diagnostics, Laval, Canada), cholesterol (1489437; Roche Diagnostics), and leptin (RI-82K; Linco Research Inc., St. Charles, MO). Liver triglyceride content was measured after Folch extraction (30).

Isolation of Adipose Cells

Adipocytes were isolated as described previously (31). Briefly, perigonadal fat pads were excised, finely chopped in 20 mL of warm α-minimum essential medium (Life Technologies, Burlington, Canada) containing 1 mg/mL collagenase (type II; Sigma Chemicals, St. Louis, MO) and 1% bovine serum albumin (Boehringer Mannheim, Mannheim, Germany), and allowed to digest for 1 hour in a shaking water bath at 37 oC. The cell suspension was filtered through a 250-μm mesh Nitex filter (Thompson, Inc., Montreal, Canada) and centrifuged at 500 rpm for 5 minutes. The cells were washed once with 30 mM Krebs–Ringer–HEPES buffer, pH 7.4 (32).

Lipolysis Assay

Lipolysis was measured as glycerol release from an adipocyte suspension. In each experiment, adipocytes were pooled from perigonadal fat of two to four 6-month-old mice of the same gender and genotype. One hundred microliters of adipocyte suspension were incubated for 2 hours in a final volume of 500 μL of Krebs–Ringer–HEPES (30 mM) buffer (pH 7.4) supplemented with 2.5% bovine serum albumin, 2.5% N6-[R-(-)-1-methyl-2-phenyl] adenosine (10 μM), and 1 U/mL adenosine deaminase. In some tubes, the lipolytic β3-adrenergic agonist CL316,243 (10 μM) (Wyeth-Ayerst Research Laboratories, Princeton, NJ) was added. After a 2-hour incubation, the reaction was stopped by freezing on dry ice. The mixture was extracted with chloroform to remove triglycerides, and glycerol content was then determined as described above.

We calculated total adipocyte number as follows. After fixation in 4% glutaraldehyde, the diameters of 200 adipocytes in each sample were measured using Sigma Scan Image measurement software (Jandel Corp., Chicago, IL). Mean adipocyte lipid content was calculated as described previously (33). The total triglyceride content of a 100-μl aliquot of adipocyte suspension was measured after Folch extraction (30), hydrolyzed, and assayed for glycerol as described above. Adipocyte number was calculated by dividing total lipid weight by mean adipocyte lipid content; adipocyte surface area was calculated as described previously (33).

Cold Stress Testing

Mice were housed individually in an empty cage at 4 °C with free access to food and water. Activity level was assessed visually and rectal temperature was assessed hourly for 8 hours and then at 24, 32, and 48 hours.

Statistical Methods

Groups were compared using the unpaired two-tailed Student's t test, except where otherwise stated.

Results

Creation of HSL-Deficient Mice

Two of 1500 G418-resistant clones (0.1%) contained targeted HSL alleles. One targeted clone was used to create chimeras, 8 of 10 of which had >80% coat color chimerism. Male chimeras were crossed with both 129 and BALB/c females. F1 offspring were obtained only from BALB/c crosses. The mice described in this paper are offspring of F1 × F1 matings.

HSL Activity and mRNA in Gene-Targeted Mice

HSL activity, measured as neutral cholesteryl esterase, was undetectable in perinephric adipose tissue homogenates of homozygous mutant mice (0.20 ± 0.42 μmol/min per milligram of protein, median ± SD, n = 10), compared with 24.3 ± 4.0 (n = 10) for heterozygotes and 40.0 ± 11.2 (n = 10) for wild-type mice.

On Northern blots of adipose tissue RNA (Figure 1c), HSL mRNA was undetectable for HSL−/− mice and was estimated as ∼63% of normal for HSL+/− mice. On long exposure of overloaded gels, a faint hybridizing band was detectable in the HSL−/− mice, estimated as 0.9% of normal intensity and of smaller size than normal HSL mRNA. 5′ Rapid amplification of cDNA ends PCR was performed using total adipose tissue RNA from an HSL−/− mouse. We amplified, cloned, and sequenced an abnormal HSL cDNA derived from alternate splicing between mouse HSL exon 2 and exon A. Exon A is an alternate upstream exon corresponding to positions −7048 to −6872 of the mouse HSL gene and initiates <10% of HSL transcripts in normal mouse adipose tissue (34). The exon A reading frame differs from that of exon 2. The first in-frame methionine codon of this abnormal mRNA corresponds to position 256 of normal human HSL. On Western blots of fat tissue extracts from HSL−/− mice, no immunoreactive HSL was detectable (Figure 1).

Adipocyte Phenotype of HSL-Deficient Mice

Clinically, both homozygous and heterozygous HSL- deficient mice were viable and grossly normal. HSL−/− mice tolerated overnight fasting and exposure to a temperature of 4 °C for 48 hours as well as controls. Weight curves (Figure 2) did not reveal significant differences related to HSL genotype. However, perigonadal, perinephric, and mesenteric fat depots were smallest in HSL−/− mice and largest in HSL+/− mice. Weights of other organs were similar (Table 1). The difference in total abdominal fat mass reached marginal significance (p < 0.05) when compared between HSL-deficient mice and heterozygotes. The other parameters in Table 1 were compared between mice of each genotype but no statistically significant differences were obtained. Histologically, white adipose tissue (WAT) from four HSL-deficient mice showed marked heterogeneity in adipocyte size (Figure 3), with adipocytes of both large and small diameter being present. A similar variation in adipocyte diameter is present in the subcutaneous inguinal WAT of HSL−/− mice (data not shown). WAT histology in three heterozygotes was indistinguishable from that of controls. Two small regions of multilocular cells were observed in the abdominal WAT of mice with normal or partial HSL activity, one in a wild-type mouse and one a heterozygote. Similar findings have been reported in normal mouse adipose tissue (35).

Figure 2.

Weight curves of HSL-deficient mice. Values shown are mean ± SEM. Mice were included only if weights were available at all indicated time points. Symbols are as follows: diamond, heterozygote; square, wild-type; circle, HSL-deficient. The following numbers of mice were used: −/−, 20 females, 8 males; +/−, 25 females, 12 males; +/+, 16 females, 8 males.

Table 1.  Organ weights in HSL-deficient mice*
 Abdominal fatOther organs 
 TotalPerinephricMesentericInguinalLiverHeartSpleenBrainKidneyn
  • *

    Mean ± SEM.

  • Total abdominal fat mass is the sum of the perinephric, mesenteric, and inguinal masses.

  • p < 0.05 with respect to HSL+/− mice.

  • §

    n = 4 mice.

  • n = 18 mice.

Female          
−/−1.57 ± 0.220.35 ± 0.050.26 ± 0.040.59 ± 0.150.96 ± 0.080.13 ± 0.010.10 ± 0.010.49 ± 0.010.33 ± 0.0116
+/−2.69 ± 0.440.63 ± 0.120.46 ± 0.081.60 ± 0.250.99 ± 0.050.13 ± 0.010.09 ± 0.010.47 ± 0.010.32 ± 0.0119
+/+2.07 ± 0.440.49 ± 0.130.26 ± 0.041.32 ± 0.290.90 ± 0.040.13 ± 0.010.08 ± 0.010.48 ± 0.010.30 ± 0.0110
Male          
−/−0.62 ± 0.050.16 ± 0.030.18 ± 0.020.29 ± 0.091.24 ± 0.120.16 ± 0.010.09 ± 0.010.49 ± 0.010.56 ± 0.066
+/−1.92 ± 0.510.39 ± 0.130.38 ± 0.081.15 ± 0.321.36 ± 0.150.18 ± 0.010.07 ± 0.010.50 ± 0.010.52 ± 0.037
+/+1.39 ± 0.380.26 ± 0.080.31 ± 0.090.83 ± 0.211.21 ± 0.16§0.18 ± 0.020.10 ± 0.010.47 ± 0.010.48 ± 0.065
Figure 3.

Histology of WAT and BAT. Representative fields from perirenal WAT and interscapular BAT depots are shown: (a) WAT, −/−; (b) WAT, +/+; (c) BAT, −/−; (d) BAT, +/+. Scale bars, 100 μm.

The weight of interscapular brown adipose fat pads from HSL−/− mice ranged from 210 to 370 mg (270 ± 30 mg, mean ± SEM, n = 5), compared with 70 to 200 mg in control littermates (130 ± 50 mg, n = 5; p = 0.005). Heterogeneity was also seen in adipocyte size in interscapular brown adipose tissue (BAT). Most adipocytes contained large lipid droplets, but some had the typical appearance of multilocular brown adipocytes (Figure 3). In the two heterozygotes studied, brown fat histology was normal. Preliminary studies suggest that HSL-deficient mice are tolerant of cold stress. After housing for 48 hours at 4 °C, two HSL−/− mice were as active as control and heterozygote mice and their rectal temperatures were similar: −/−, 35.9 °C; +/−, 36.2 °C; +/+, 36.2 °C (mean of two observations in each case).

Figure 4 shows levels of circulating metabolites, leptin, and liver triglyceride concentration. Nonesterified fatty acids were lowest in HSL−/− mice. In HSL−/− males, this reached statistical significance (p = 0.03) in comparison with wild-type littermates. After an overnight fast, livers of HSL−/− mice were distinguishable grossly by lack of the yellowish steatosis found in HSL+/− and HSL+/+ mice. The hepatic triglyceride content of HSL−/− mice was less than in other mice (Figure 4). When HSL−/− and HSL+/− females were compared, mean glucose levels were higher and leptin levels were lower in HSL-deficient mice. However, this finding was absent in males, and its significance, if any, is uncertain. Of note, cholesterol was significantly increased in HSL-deficient mice of both genders in comparison with normal control littermates.

Figure 4.

Fasting plasma metabolite and hormone levels. Each pair of panels shows the mean and SE for one parameter in female (left) and male (right) mice. Black bars, HSL −/−; gray bars, +/−; white bars, +/+. The numbers of mice studied are shown above each column, except if indicated otherwise by a number above or within the data panels. Results of different genotypes were compared using the two-tailed Student's t test. *p < 0.05; **p < 0.01.

In adipocytes isolated from perigonadal fat pads of 6-month-old mice, the mean diameter was less in mutants than in wild-type littermates (Table 2). Because lipolysis under basal conditions and after β3-adrenergic stimulation was comparable in adipocytes from male and female mice, we pooled the results from five lipolysis experiments (three with female adipocytes and two with male adipocytes) (Table 2). Basal lipolysis in mutant adipocytes was greater than that in controls (p = 0.03). When expressed in terms of cell number, a similar pattern was seen, but the differences were not statistically significant: HSL−/− adipocytes produced 5.55 ± 0.45 nmol of glycerol per 106 adipocytes per hour (median ± SEM) compared with 5.19 ± 1.13 for heterozygotes and 4.41 ± 0.52 for normal controls (p > 0.05 for all comparisons).

Table 2.  Diameter and lipolysis of isolated adipocytes
inline image

In all five lipolysis experiments, HSL-deficient adipocytes showed minimal or no β3-adrenergic responsiveness(Table 2). The mean values of basal and stimulated lipolysis in HSL-deficient adipocytes were not significantly different when compared for each of the five experiments using the paired one-tailed Student's t test (p = 0.13) in contrast to the 4-fold to 9-fold increases in adipocytes from wild-type littermates (p = 6.9 × 10−5) and intermediate increases in heterozygotes (p = 0.0028).

Discussion

HSL-deficient mice provide several insights into adipocyte physiology. HSL-deficient adipocytes, as expected, show little or no lipolytic response to β3-adrenergic stimuli. In contrast, there was a normal, several-fold increase in wild-type cells, comparable with normal values from the literature (36), and an intermediate response in heterozygous cells. The small increase in the mean values of lipolysis after adrenergic stimulation of HSL−/− adipocytes (Table 2) may be confirmed as a real phenomenon in future experiments. Current theories of lipolysis hold that activation of HSL is only one component in the regulation of lipolysis, with important roles being played by other proteins such as perilipin A (4), which coats the lipid droplet and potentially may regulate the access of HSL to its triglyceride substrate. Lipid droplet surface proteins may also prove to be important in controlling the access of non-HSL lipase(s) to the lipid droplet.

In contrast to adrenergic-stimulated lipolysis, under basal conditions, adipocytes isolated from HSL−/− mice are at least as lipolytically active as normal adipocytes. Because the diameters of the isolated adipocytes were smaller than normal (Table 2), differences in lipolysis between HSL-deficient and normal adipocytes are progressively more striking if lipolysis is calculated in terms of adipocyte number, adipocyte surface area, or adipocyte mass. It will be challenging to determine whether HSL−/− cells are smaller because of a physiologically important increase in basal lipolysis and to what extent the high calculated lipolytic rate depends simply on their reduced cell dimensions.

Histologically, in contrast to the small cell diameter of isolated HSL-deficient adipocytes, many very large adipocytes are present(Figure 3). These findings can be reconciled if the large adipocytes were retained on the filter or disrupted during isolation. Large adipocytes are known to be particularly fragile during isolation (37).

The presence of large fat-engorged adipocytes is predicted from the known function of HSL. However, there is no obvious explanation for the small or normal-sized adipocytes in HSL-deficient mice. We speculate that the adipocytes of normal size are either resistant to lipid accumulation by mechanisms that are currently unclear or that they represent young cells that have not yet accumulated large amounts of triglycerides.

Taken together, the lipolysis data in mice of different HSL genotypes in the presence and absence of adrenergic stimulation strongly suggest the presence of at least two lipolytic pathways in fat cells, one that is mediated by HSL and is sensitive to β3-adrenergic stimuli, plus an HSL-independent (basal) pathway(s). Of note, this theory does not necessary imply that HSL is quiescent in the nonstressed state in vivo, because basal adrenergic tone in the living animal could provide a constitutive level of HSL-dependent lipolysis. The observation of marked histological changes in nonstressed HSL-deficient animals is consistent with the notion that HSL is physiologically important in adipose tissue even in nonstressed conditions.

Abdominal fat depots were smallest in HSL-deficient mice, in contrast to our initial expectations (Table 1), and largest in HSL+/− mice. This can account for most of the differences in mean body weight after 4 months of age (Figure 2), and correlates with mean leptin levels (Figure 4). It is not obvious how HSL gene dosage would produce a nonlinear effect on fat mass, and these results require confirmation in a larger number of animals. However, they are noteworthy because human conditions such as familial combined hyperlipidemia, type 2 diabetes, and insulin resistance have been associated with abnormal lipolysis (38); also, there is some evidence from gene-mapping studies that genetic variation in or near the HSL locus is associated with metabolic syndrome (39). The finding of hypercholesterolemia in HSL deficiency (Figure 4) may be pertinent in this context.

Other examples exist of gene-targeted animals for which nonlinear, apparently unrelated phenotypes are observed in heterozygous and deficient animals. For instance, the peroxisome proliferator-activated receptor-γ knockout mouse shows prenatal lethality in homozygotes (40), whereas the heterozygote phenotype is a reduced tendency to the development insulin resistance after challenge with a high-fat diet (41).

In brown fat, HSL deficiency results in hypertrophy with the presence of fat-engorged adipocytes histologically, but does not cause marked cold intolerance. Fatty acids are thought to play a central role in thermogenesis (42) (43). Our observations suggest either that BAT cells have a non-HSL-dependent supply of fatty acids for thermogenesis or that fatty acids are not essential for thermogenesis.

To date we have mainly studied the effect of HSL deficiency in nonstressed mice but preliminary results suggest a role for HSL in lipolytic stress.

After a short overnight fast, circulating fatty acid levels and liver triglyceride content are low in HSL-deficient mice compared with wild-type littermates, suggesting that HSL may play an adaptive role in prolonged fasting.

During preparation of this article, Osuga et al. (44) reported an independently derived gene-targeted HSL-deficient mouse strain. These authors also documented a reduced lipolytic response to β-adrenergic stimulation in isolated adipocytes and found size heterogeneity of adipocytes. One possible difference in the results from the two groups was the finding by Osuga et al. (44) of a mild lipolytic response to β-adrenergic stimuli. They reported a modest but statistically significant increase in fatty acid release, with a similar but not significant increase in glycerol release. However, the difference in glycerol release before and after adrenergic stimulation was of the same order as in our mice. Another possible difference between the two strains is that Osuga et al. (44) did not observe the increase in basal lipolysis seen in our mice.

The two reports differ in several technical respects. First, to measure lipolysis, we used a 2-hour incubation with a β3-adrenergic agonist, whereas Osuga et al. (44) used a 15-minute incubation in the presence of isoproterenol, a general β-adrenergic agonist. There may be differences in the intensity or the specific signaling pathways of general and β3-selective β-adrenergic stimulation. Second, neither mouse strain had been bred to a pure background. Both strains segregated genes from 129 Sv-derived ES cells and from BALB/c as well as C57BL/6. Random epigenic background effects or differences in the environments to which the different strains were exposed may have contributed to observed differences. Finally, the two targeting constructs differed. Our construct removed the 3′ end of exon 1, which would disrupt all known HSL transcripts in adipose tissue. In contrast, that of Osuga et al. (44) lacked exon 6, which contains the catalytic site of HSL. Neither mouse has evidence of detectable HSL peptide or of HSL activity, but this has not been exhaustively documented in either case. In theory, small levels of aberrant peptides with structural similarity to HSL could be generated in either strain. Although there is no evidence for this, it cannot be formally ruled out that such peptides may contribute to the observed differences between strains by influencing the structure of the lipolytic protein complexes at the lipid droplet surface, or in the case of our mice, by providing a small amount of HSL activity that is below the limit of detection of our assay. As discussed above, lipolysis may be regulated in part by lipid droplet surface proteins such as perilipin A. If this is the case, it would not be necessary to evoke the presence of undetectable HSL-related peptides.

In whole animal studies comparing 9- to 14-week-old HSL−/− and HSL+/+ mice, Osuga et al. (44) did not detect a significant difference in abdominal fat mass, although heterozygote values were not provided. Nor did they detect a statistically significant difference in respiratory quotient in fed or fasted mice. In fact, the mean value for HSL−/− mice was somewhat lower than that for HSL+/+ mice, suggesting that oxidative fat metabolism was at least as active in the HSL-deficient mice. In contrast, our data are consistent with a role for HSL in the determination of adipose tissue mass and an important role for HSL at times of active lipid energy metabolism such as fasting. These possible discrepancies define areas for future study in a larger series of mice.

Both studies strongly suggest that there are at least two lipolytic pathways in adipocytes. From our results with isolated adipocytes, we are currently exploring the hypothesis that the HSL-independent pathway(s) mediates basal lipolysis while an HSL-dependent pathway is responsive to lipolytic stresses but quiescent at other times.

The role of HSL in nonadipose tissues also deserves scrutiny. For example, Osuga (44) and we (manuscript in preparation) observed that male HSL-deficient mice are sterile, suggesting an important role for HSL in male fertility. HSL-deficient mice will be useful for the study of other tissues that normally express HSL, including muscle, adrenal gland, macrophages, and pancreatic β cells.

Acknowledgments

This project originated as a Canadian Genetic Diseases Network collaboration between G.A.M. and M.A.R and was supported by Medical Research Council of Canada Grants MA-12625 (to G.A.M.) and MT-0857 to (J.H.H.). We thank Yves Théorêt, Michel Tremblay, Linda May, Gillian Shilabeer, and David Lau for technical discussions, Constantine Londos for supplying anti-HSL antibody and for thoughtful comments about lipolysis measurement and other techniques, and Ginette Richard for secretarial assistance.

Ancillary