Loss of Par-1a/MARK3/C-TAK1 Kinase Leads to Reduced Adiposity, Resistance to Hepatic Steatosis, and Defective Gluconeogenesis ^▿

Animal procedures.

The Washington University School of Medicine Animal Studies Committee approved all animal procedures. Unless stated otherwise, animals were fed a standard chow diet (Lab Diet 5063 containing 4.5% fat, 55% carbohydrate, 20% protein, and 4.7% fiber; Purina Mills, St. Louis, MO) and were housed with free access to food and water under a 12-h light/dark cycle. Unless indicated, only male mice were analyzed throughout this study.

Construction of the Par-1a _targeting construct.

A Par-1a genomic clone was isolated from a 129X1/SvJ mouse embryonic stem (ES) cell genomic library by hybridization with an isogenic 0.9-kb Par-1a-specific cDNA probe. The genomic organization of the mouse Par-1a gene was disrupted by replacing a 1.9-kb SacI/HpaI fragment with a 1.2-kb neomycin phosphotransferase cassette (Fig. 1 A). The replacement disrupts exon 2 encoding the ATP binding domain of Par-1a and also destroys the 5′ splice site between introns 2 and 3. The insertion of the neomycin selection cassette alters the PstI fragment from a single 13.8-kb fragment to two fragments of 7.5 and 5.9 kb, and this change was used to identify _targeted ES cell clones and knockout (KO) animals by Southern blotting (Fig. 1B) using the 5′ and 3′ probes indicated in Fig. 1A. Using mouse genomic DNA as the template, the 425-bp 5′ probe was amplified by PCR with primers of 5′-TGGTGGTACACGCCTGTATTC and 5′-CTGATCTACACAATTCCAGGAC. The 204 bp of 3′ probe was amplified by PCR with primers of 5′-GCATCTTGGGCTGTAAGTGATG and 5′-GGCTTACAACCATCTGTACAG.

PCR analysis.

PCR of tail DNA/tissues was achieved using a three-primer PCR with one 5′ primer from exon 2 (5′-GGAGACGGGAGGCAAGAAGTCA), a 3′ primer from the intronic sequence between exons 2 and 3 (5′-GGCTCATTCCCTTGGTGTACTT), and a 3′ primer from the neomycin cassette (5′GAGCAGCCGATTGTCTGTTGT), resulting in 400-bp (KO) or 300-bp (wild-type [WT]) bands, respectively (Fig. 1C).

Generation of mice harboring the Par-1a mutation.

RW4 ES cells were electroporated with linearized _targeting vector and selected with Geneticin (G418; Invitrogen) using established protocols developed in the Murine Embryonic Stem Cell Core of the Alvin J. Siteman Cancer Center at Washington University School of Medicine and Barnes-Jewish Hospital (details are available at http://escore.im.wustl.edu). A total of 96 G418-resistant ES cell clones were analyzed for homologous recombination by Southern blotting, and three clones were found to be positive. Positive ES clones were karyotyped and microinjected into 3.5-day postcoitus (dpc) C57BL/6 blastocysts, which subsequently were implanted into the uteri of pseudopregnant C57BL/6 × C3HF1 foster mothers. Male chimeras selected by the percentage of agouti color were mated to C57BL/6 females. Germ line transmission was determined by agouti coat color. F1 animals were tested for the _targeted Par-1a allele by Southern blotting and the PCR analysis of tail DNA. Heterozygous (HET) F1 males/females were interbred to generate the F2 littermates used for subsequent breeding and analysis and are referred to as wild-type, heterozygous, and null pups/mice.

Western blotting.

Tissues from wild-type and knockout animals were homogenized in 800 μl of mammalian cell lysis buffer (MCLB; 50 mM Tris-HCl, pH 8.0, 5 mM EDTA, 0.1 M NaCl, 0.5% NP-40, 2 mM dithiothreitol [DTT]) containing 1 μM microcystin, 1 mM sodium orthovanadate, 10 mM β-glycerol phosphate, 1 mM sodium fluoride, 2 μM phenylmethylsulfonyl fluoride (PMSF), and protease (Sigma, St. Louis, MO) and phosphatase inhibitor cocktails (Calbiochem, Gibbstown, NJ). Homogenates were rocked for 15 min at 4°C and then clarified twice by centrifugation. Clarified lysates were resolved by SDS-PAGE on a 10% SDS gel, and Western blotting was carried out with antibodies specific to actin (1:4,000; Sigma Chemical Co.), tubulin (1:1,000; Sigma), β-catenin (1:10,000; BD Biosciences/Pharmagen), Par-1a ascitic fluid (1:10,000) (45), glycogen synthase (1:1,000; Cell Signaling), phosphoglycogen synthase (1:1,000; Cell Signaling), and LC-3 (1:1,000; Cell Signaling). Par-1b was detected with ascites generated using a monoclonal antibody produced against keyhole limpet hemocyanin (KLH)-conjugated peptide corresponding to amino acids 288 to 307 (PDYKDPRRTELMVSMGYTRE) of human Par-1b. Bound primary antibodies were detected with horseradish peroxidase (HRP)-conjugated goat anti-mouse antibody (Jackson, West Grove, PA), and proteins were visualized by chemiluminescence.

Tissue distribution.

The distribution of Par-1a and Par-1b in various tissues was examined using a modified Western blot protocol. In brief, mouse tissues harvested from ∼6-month-old mice were homogenized in 0.8 ml of mammalian cell lysis buffer (as described above). Lysates were rocked for 15 min at 4°C, followed by centrifugation for 10 min at 20,800 × g. Clarified supernatants were transferred to prechilled tubes. Proteins were resolved by SDS-PAGE on 7.5% gels until the 50-kDa molecular mass marker was at the bottom of the gel. Proteins were transferred to nitrocellulose membranes in CAPS buffer [0.1 M 3-(cyclohexylamino)-1-propanesulfonic acid and 10% methanol (C2632; Sigma, St. Louis, MO)] at 1.2 mA for 1.5 h. Membranes were blocked in 1× TBST (100 mM Tris, pH 8.0, 1.5 M sodium chloride, 2% Tween 20) containing 5% milk for 1 h. Membranes were probed with Par-1a ascites (1:10,000), Par-1b ascites (1:1,000), actin (1:4,000; Sigma Chemical Co.), or β-catenin antibody (1:30,000) by being rocked overnight at 4°C. Membranes were washed four times for 13 min each in 1× TBST. Membranes then were incubated with a goat anti-mouse antibody (dilution, 1:30,000; Jackson, West Grove, PA) for 1 h at room temperature. Membranes then were washed four times in 1× TBST for 13 min each. Blots were developed using ECL detection reagent (GE Healthcare, Piscataway, NJ).

Breeding, animal weights, and growth measurements.

Embryos and pups generated from heterozygous crosses were weighed and genotyped by PCR analysis. Plug observation was recorded as 0.5 dpc; embryos were harvested, rinsed, and weighed. Mean body weights at each time point were compiled from at least 6 mice per genotype. Breeding schemes followed standard protocols, and for double-knockout (DKO) experiments genotypes are abbreviated by a two-letter code (the first letter indicates Par-1b status, and the second letter indicates Par-1a status: W, +/+; H, +/−; K, −/−; e.g., HK). Binning and comparison within and between genotypes via univariate and multivariate analyses tested trends over time. We assessed the number and genotype of litters in 10 superovulated timed pregnancies at 8.5 and 10.5 dpc. Offspring ratios for Par-1a^−/−, Par-1b^−/−, and all genotypes in DKO experiments were recorded as pups/litter and total numbers per genotype. The significance of deviations from expected Mendelian ratios were tested using chi-square statistics with 2 (WT, HET, and KO) or 8 degrees of freedom (KK, WW, WK, KW, WH, HW, KH, HK, and HH) in Par-1a null and DKO experiments, respectively (two-tailed P values are provided).

Metabolic measurements (indirect calorimetry).

Metabolic measurements were determined using previously established protocols (8, 43). Briefly, metabolic characterization included metabolic rate, energy expenditure, activity, food consumption (in g/day), and body temperature. During data analysis, energy expenditure and metabolic rate were normalized with respect to body weight. Energy expenditure and respiratory quotient (RQ) were calculated from the gas exchange data. RQ was the ratio of the rate of the elimination of carbon dioxide (VCO₂) to VO₂ and the equation for energy expenditure (or heat) was (3.815 + 1.232 × RQ) × VO₂. Activity was measured on x and z axes using infrared photobeams to count the number of beam breaks.

HFD studies.

HFD studies followed previously published protocols (43) and consisted of a total of 55 mice. In 35 3-week-old mice (WT, Par-1a^−/−, Par-1b^−/−), the regular rodent chow diet (CD) was replaced with an adjusted-calories diet (42% fat) from Harlan Teklad (88137; Western diet; Madison, WI). The weight of the HFD chow was recorded, and food consumption as well as mouse weight were determined for individual mice on a weekly basis. Although resistance to weight gain on HFD has been reported previously for Par-1b null mice (43), for the direct comparison and determination of specific organ uptake values, the HFD experiment included a new group of nine Par-1b null mice and therefore consisted of six different trial groups: WT-CD, WT-HFD, Par-1a^−/−-CD, Par-1a^−/−-HFD, Par-1b^−/−-CD, and Par-1b^−/−-HFD. Metabolic profiling after a total diet time of 16 weeks consisted of (i) fasting (n = 18 CD, 35 HFD) and blood glucose levels after feeding (n = 6 per genotype, randomly selected), (ii) glucose tolerance testing (GTT) of 18 CD mice (6 per genotype, randomly selected) and all HFD mice, (iii) fludeoxyglucose ([¹⁸F]FDG) biodistribution experiments (n = 19 CD, 31 HFD), (iv) organ weights, and (v) histologic examination. The HFD trial was designed to allow at least 24 h between GTT and [¹⁸F]FDG biodistribution studies, during which animals were kept on HFD.

Serum factor quantification.

Serum levels of insulin, triglycerides, adiponectin, leptin, and cholesterol were determined by Ani Lytics Incorporated (Gaithersburg, MD; n ≥ 11 animals per genotype and sex); glucagon and free fatty acids were determined according to previously established protocols (18).

Glucose and insulin measurements and tolerance tests.

Randomly fed or fasted mice were analyzed, and blood glucose levels were determined using a β-glucose photometer and β-glucose cuvettes (HemoCue AB, Angelholm, Sweden) or Accucheck advantage glucometer (Roche Diagnostics Corp., Indianapolis, IN). Normoglycemia was defined as blood glucose values between 90 to 130 mg/dl. Intraperitoneal (i.p.) glucose tolerance tests were performed on fasted (12 h) animals on CD and all HFD trial mice. Mice were injected i.p. with d-glucose (20% solution; 1 g/kg of body weight), and blood glucose levels were determined at 15, 30, 60, 90, 120 min, and >24 h postinjection. In addition, insulin levels were determined during GTT immediately before and 15, 30, and 60 min after glucose injection. Therefore, ∼20 to 30 μl of blood was collected using the Microvette 200 capillary blood collection system (Sarstedt, Newton, NC) and centrifuged at 30,000 × g for 5 min. Fasting serum insulin levels were determined for 50 male mice on CD (n = 24 WT, 8 Par-1a^−/−, 18 Par-1b^−/−) and 18 randomly selected mice on HFD (n = 6 per genotype) using rat insulin as a standard and enzyme-linked immunosorbent assays (ELISA) as specified by the manufacturer (Crystal Chem, Inc., IL). Insulin tolerance tests (ITT) were performed on 6-h-fasted male mice (n = 9 Par-1a^+/+, 9 Par-1a^−/−). Insulin sensitivity in Par-1b null mice has been reported (43) and was not repeated. Blood glucose values were measured immediately before and at 15-min intervals for a total of 90 min after the i.p. injection of insulin (0.30 IU/kg HumulinR; Eli Lilly and Company, Indianapolis, IN).

MicroPET studies.

Micro-positron emission tomography (microPET) studies entailed the imaging of cohorts of male WT and Par-1a null mice that were repetitively imaged once a week for four consecutive weeks, each time under a different metabolic condition. Brown adipose tissue (BAT), skeletal muscle (SM), heart, liver, and brain were examined under the following four conditions: (i) fasted with saline, (ii) fasted with insulin, (iii) nonfasted with saline, or (iv) nonfasted with insulin. The next morning, either 0.5 U/kg insulin (Humulin N; Eli Lilly, Indianapolis, IN) or saline was administered to mice by i.p. injection. Thirty minutes after the injection, mice were lightly anesthetized with isoflurane, followed by a tail vein injection of [¹⁸F]FDG (∼200 to 500 μCi in saline). Immediately after the injection of radiotracer, mice were placed supine in the microPET scanner (Focus120 or Focus220 scanner; Concorde MicroSystems, Knoxville, TN) and imaged (acquisition time, ∼10 min; one bed position; ordered set expectation maximum reconstruction). Mice were allowed to recover and then were anesthetized and imaged again at 1 and 2 h after the injection of radiotracer. The Par-1b null and WT mice used for comparison were age-matched males and treated as described above, except that they were imaged on an R4 scanner (43). MicroPET images were corrected for decay, but not attenuation or scatter, and then stacked regions of interest (ROI) of relevant tissues and organs were analyzed with AnalyzePC 6.0 software. ROI counts were converted to counts per gram of tissue (nCi/g), assuming a tissue density of 1 g/ml. Data for the accumulation of [¹⁸F]FDG on microPET images were expressed as standard uptake values (SUV), representing counts per gram of tissue divided by the injected dose of radioactivity per gram of animal weight (87, 88). After statistical analysis using a one-way analysis of variance (ANOVA) and Student's t tests, comparison between genotypes and experiments was enabled by the normalization of the SUV to the corresponding WT littermates, with data for Par-1a- and Par-1b-disrupted mice against their littermate WT as the normalized average SUV ± standard errors of the means (SEM); errors were propagated (see below).

[¹⁸F]FDG biodistribution experiments.

As previously reported, white adipose tissue (WAT) cannot be analyzed by noninvasive microPET imaging (43). As an alternative, we examined the glucose uptake of WAT and other organs by traditional [¹⁸F]FDG biodistribution (12, 20, 53, 88). Mice were anesthetized with isoflurane (2% vaporized in oxygen), followed by the tail vein injection of [¹⁸F]FDG (20 μCi in 100 μl saline). Sixty minutes after [¹⁸F]FDG injections, mice were sacrificed by cervical dislocation (under anesthesia). Blood, liver, heart, brain, muscle, WAT, and BAT were rapidly removed and weighed, and their radioactivity was measured using a Beckman 8000 gamma counter. The percent injected dose per gram of tissue (% ID/g) was calculated [(tissue μCi) (injected Ci)⁻¹ (g tissue)⁻¹ × 100] and subsequently compared among tissues, genotypes, and diets. Biodistribution analysis was performed on 31 HFD mice (17 WT, 8 Par-1a^−/−, 6 Par-1b^−/−) and 19 mice from the CD arm (age-matched control group; total n = 50). Tissue samples from these mice were fresh-frozen and stored at −80°C. Subsequent cryostat sections confirmed tissues and allowed histomorphological/histochemical comparison to tissues obtained from animals that did not undergo [¹⁸F]FDG uptake (n = 5; no differences observed) or HFD (see Results). For these HFD trial biodistribution experiments, the 50 mice were prospectively assigned to one of nine experimental days with an average of six animals per experiment (range, 3 to 10 mice/day). Excluding the 16 weeks on HFD (as described above), these experiments were performed during a 10-month period.

Starvation experiments.

Starvation experiments consisted of the determination of blood glucose as well as ketone levels using a MediSense Precision Xtra diabetes glucose and ketone monitor system (Abbott, Alameda, CA). After overnight fasting (20 h), measurements were repeated; one group of mice was subjected to extended fasting (an additional 4 h), and the other group of mice was given an i.p. injection of 2 mg/g glucose for 2 h. Tissues were collected at all time points and samples separated for Western blotting, routine histology, electron microscopy (EM), and glycogen determination.

Dissection and histology.

Organs were excised, weighed, rinsed in phosphate-buffered saline (PBS), fixed in 10% neutral-buffered formalin, rinsed in PBS, and stored in 70% ethanol. Formalin-fixed tissues were paraffin embedded (FFPE) and sectioned (∼5 μm) using standard procedures. For morphological evaluation and fat content, hematoxylin-and-eosin (H&E) and oil red O staining was performed on FFPE and fresh-frozen sections (stored at −80°C), respectively. Morphological analysis was performed using multiple independent samples per site/organ (13 to 26 samples per genotype) as well as all 55 animals in the HFD trial. We quantified BAT and WAT samples taken from the interscapular and gonadal fat pads, respectively. Quantification was performed using digital image processing via ImageJ (v. 1.37; http://rsb.info.nih.gov/ij/) and Photoshop CS3 (Adobe Systems) according to previously established semiautomatic protocols (16, 39). Specifically, oil red O positivity was quantified from ≥20 high-power fields, cell numbers or lipid vacuoles were quantified in 6 to 20 randomly chosen images per sample, and periodic acid Schiff (PAS) staining was quantified using integrated density measurements (IntDen; equivalent to the product of the area and mean gray value, with gray value being the luminescence value between 0 [black] and 255 [white]).

Glycogen determination.

Fifty milligrams of liver and skeletal muscle samples were extracted with 0.9 N perchloric acid and 99% ethanol to precipitate glycogen. The glycogen from the pellet was dialyzed and digested to free glucose with amyloglucosidase. Glucose concentrations were measured by the glucose oxidase method (glucose analyzer II; Beckman Instruments, Fullerton, CA).

EM.

EM was performed according to established protocols (85). Tissues were fixed overnight at 4°C in modified Karnovsky's fixative containing 3% glutaraldehyde and 1% paraformaldehyde in sodium cacodylate buffer, pH 7.4. After being rinsed in sodium cacodylate buffer, samples were postfixed in phosphate cacodylate-buffered 21% OsO₄ for 1 h, dehydrated in graded ethanols with a final dehydration in propylene oxide, and embedded in Embed-812 (Electron Microscopy Sciences, Hatfield, PA). One-micron-thick plastic sections were examined by light microscopy after being stained with toluidine blue. Ultrathin sections (90 nm thick) of individual samples were cut onto formvar-coated slot grids, which permit the visualization of larger cross-sections. Sections were poststained with uranyl acetate and Venable's lead citrate and viewed with a JEOL model 1200EX electron microscope (JEOL, Tokyo, Japan). Digital images were acquired using the AMT advantage HR (Advanced Microscopy Techniques, Danvers, MA). Plastic sections as well as an ultrastructural review were performed blinded to the genotype.

Digitization and image analysis.

Analyses were performed using an Olympus DP70 digital camera (Olympus, Tokyo, Japan) connected to an Olympus BX51 light microscope or a Scanscope XT whole-slide scanner/ImageScope v10.0.36.1805 (Aperio, Vista, CA). Image and pixel quantification employed established semiautomatic threshold algorithms (60) using the software package ImageJ (http://rsb.info.nih.gov/ij) (1). Statistical testing between groups was performed using Student's t test and the Wilcoxon two-sample test.

Statistics.

Chi-squared t tests (two-tailed), the nonparametric alternative Mann-Whitney U test, one-way and two-way ANOVA, and Bonferroni posttests to compare replicate means were used when appropriate. In experiments with small numbers of samples (defined as n ≤ 12), the D'Agostino-Pearson omnibus K2 normality test (24) was used to approximate the Gaussian distribution of values. P values of <0.05 indicate non-Gaussian distribution, and we assessed these experiments with both parametric and nonparametric tests; both P values are provided when applicable. In all studies, the accepted level of significance was P < 0.05, and data are reported as means ± SEM. In selected experiments and for comparison between strains and experiments, values were normalized to the corresponding WT and errors were propagated according to the formula σ_x̄WT/x̄_WT)² + (σ_x̄KO/x̄_KO)²] ^1/2 × x̄_NKO, where σ̄ is the SEM, x̄ is the average, and N is the normalized average. N̄KO was adapted from reference 66.

Error propagation and statistical tests in general were chosen assuming the most pessimistic situation (80). All data were analyzed using Prism 5.0b (GraphPad Software Inc., La Jolla, CA) or Microsoft Excel 2008 (version 12.1.9; Microsoft Corporation, Redmond WA).

Par-1a^−/− mice are hypermetabolic and show 9% reduction in body weight.

When Par-1a heterozygotes were intercrossed, the offspring contained WT, HET, and KO pups, although genotype proportions deviated significantly from expected Mendelian ratios with the loss of homozygous KO mice (n = 750; WT, 30%; HET, 52%; KO, 17%; χ² = 27.864, P < 0.0001). All pups developed without any visible dysmorphisms or obvious external pathology, and animals survived to adulthood. Par-1a null mice exhibited significant weight differences compared to their wild-type littermates (e.g., at 32 weeks, WT mice [n = 6] were 33.4 ± 0.94 g and Par-1a^−/− mice [n = 6] were 25.5 ± 0.42 g; P < 0.0001, t test). In contrast to Par-1b null mice that show reduced weights at 6 weeks of age (9, 42), in Par-1a null mice the difference was not discernible this early (Fig. 2 A and B).

Body weight represents a net balance of food intake and energy expenditure. Based on the decreased weight in Par-1a null mice, we measured total energy expenditure (EE), total metabolic rate (MR), and respiratory quotients of male mice on a regular chow diet. Par-1a null mice exhibited higher O₂ consumption (MR showed an ∼9% increase) (Fig. 2C) and higher EE (10% increase) (Fig. 2D). We also analyzed the respiratory quotient (RQ = VCO₂/VO₂) as a measure of fuel-partitioning patterns, and RQ ranged from 0.76 to 0.78 (P = 0.32, t test) (Fig. 2E). Food intake in Par-1a null mice was slightly less than that of WT littermates, although this difference was not statistically significant (WT, 3.15 ± 0.09; Par-1a^−/−, 2.94 ± 0.04; P = 0.16, t test) (Fig. 2F). Thus, unlike Par-1b null mice, Par-1a null mice are not hyperphagic (43). The lack of effect on RQ suggests that there was no difference in fuel selection between carbohydrates and lipids, leading us to explore whether the increase in energy expenditure occurs secondarily to increased thermogenesis. However, body temperatures of Par-1a null mice were similar to those of WT littermates (WT, 34.3 ± 0.5; Par-1a^−/−, 34.1 ± 0.8; P = 0.8, t test), and the expression of uncoupling proteins in selected tissues showed no significant difference (not shown). Lastly, Par-1a null mice showed similar levels of activity (compared to that of WT mice; P = 0.96), and basic sensorimotor testing as well as open-field tests showed no significant difference (P = 0.64 [not shown]).

Par-1a^−/− mice are normoglycemic and normoinsulinemic and show no insulin sensitivity.

Serum levels of insulin, glucagon, leptin, adiponectin, triglycerides, cholesterol, and free fatty acids in Par-1a null mice were measured and compared to values from WT and Par-1b null mice (20-week-old mice on regular chow diet) (Table 3). Male Par-1a null mice exhibited levels of adiponectin that were 1.6-fold higher than those of WT littermates. Par-1b null mice showed a significant decrease in female adiponectin values. In contrast to Par-1b null mice, Par-1a null mice showed no difference in serum insulin levels (Table 3), and blood glucose was normal (WT, 152 ± 18 mg/dl; Par-1a^−/−, 147 ± 5 mg/dl; P = 0.216, t test). Par-1a null mice on CD exhibited ITT curves identical to those of WT littermates (Fig. 2G; P = 0.87, one-way ANOVA). On CD, Par-1a null mice showed no significant difference in GTT relative to that of WT littermates (Fig. 3 C) (P = 0.15, t test; and P = 0.181, two-way ANOVA for the entire time course; Par-1b^−/− data not shown).

Par-1a^−/− mice show delayed weight gain on high-fat diet.

The striking resistance to HFD-induced weight gain in Par-1b null mice (43) prompted us to challenge Par-1a null mice with a similar HFD for 16 weeks (Fig. 3A and B). The main finding was the reduced weight gain of Par-1a null mice compared to that of WT littermates on HFD (Fig. 3A). Specifically, while Par-1b null mice on HFD showed a statistically significant lack of weight gain after 7 weeks [designated t(7w); t(7w) = 3.297; P < 0.05; t(17w) = 7.234; P < 0.001, two-way ANOVA with Bonferroni posttests; weight gain, +5%; weight on CD, 19 ± 0.75 g; weight on HFD, 20.3 ± 1.34; P = 0.4097, t test], Par-1a null mice on HFD continued to gain weight during the entire experiment (Par-1a^−/− weight gain, +39%; CD weight, 23.47 ± 1.6 g; HFD weight, 32.4 ± 1.47; P = 0.004, t test), albeit at a reduced rate and significantly different from that of WT littermates (+70% weight gain; CD weight, 27.12 ± 0.5 g; HFD weight, 46 ± 0.9; P = 0.001, t test). HFD food intake of Par-1a null mice was similar to that of WT littermates (P = 0.39, t test). While the Par-1a phenotype with a reduced rate of weight gain is, at first glance, less dramatic than that observed in Par-1b null mice [Par-1a^−/− versus WT: t(all) = 2.779, P < 0.05, one-way ANOVA with Bonferroni's multiple comparison test], it is a highly significant finding (P = 0.001) (Fig. 3B). Par-1a null mice required 7 additional weeks of HFD to reach the original weight of their WT littermates.

Par-1a^−/− mice show resistance to glucose intolerance on high-fat diet.

Par-1a null mice fed an HFD were subjected to glucose tolerance testing (Fig. 3C). As expected, WT mice on HFD exhibited glucose intolerance relative to that of WT mice on CD, with a prolonged elevation of blood glucose of ∼400% from 15 min onwards (t value = 3.645, P < 0.01, two-way ANOVA). Strikingly, Par-1a null mice exhibited improved glucose tolerance on HFD relative to that of WT controls on the same diet (P = 0.041, t test [two genotypes]; t value = 3.38, P < 0.01, one-way ANOVA [all genotypes]) (Fig. 3D). GTT response in Par-1a null mice under HFD conditions was essentially identical to that of WT and Par-1a null mice on CD (P = 0.4781, t test [two genotypes]; t value = 0.346, P > 0.05, one-way ANOVA [all genotypes]) (Fig. 3D and data not shown). We conclude that Par-1a disruption leads to the maintenance of normal glucose tolerance even under HFD conditions; accordingly, Par-1a null mice on HFD were normoglycemic (P = 0.3, t test). We next examined serum insulin levels in HFD-fed mice during the first 60 min of the GTT (Fig. 3D) and found significantly lower levels in Par-1a null mice (t value = 3.16, P = 0.0194, t test [two genotypes]; F [mean square between groups over the mean square within groups] = 6.454, P = 0.018, one-way ANOVA). Although we did not formally assess insulin tolerance in Par-1a null mice on HFD, the combination of improved glucose tolerance and reduced insulin secretion during the GTT is indicative of peripheral insulin hypersensitivity. Thus, Par-1a null mice demonstrate normal insulin and glucose sensitivity unless challenged by HFD. In contrast, Par-1b null mice exhibited insulin hypersensitivity even on CD (43). The direct assessment of pancreatic islets has proven difficult with Par-1 mice (38); however, morphological and morphometric differences were not observed in the islet organ of Par-1a null mice (not shown).

Par-1a^−/− mice show diet-dependent reduced adiposity.

We determined whether the reduced body weight of Par-1a null mice was reflected disproportionately in different tissues (Fig. 4 A). Measurements of tissue weights in Par-1a null mice indicated that decreased weights of most organs were proportional to total body weight and consistent with a decrease in overall body mass rather than in specific organs. However, two tissues did not follow this rule and were disproportionately smaller: white adipose tissue (WAT from gonadal fat pads) and brown adipose tissue (BAT). We also performed the dissection of tissues in the mice on HFD and, as expected, the energy excess in WT mice led to a disproportionate increase in overall adiposity (BAT-CD versus BAT-HFD, P = 0.001; WAT-CD versus WAT-HFD, P = 2.5e⁻⁵; t tests) (Fig. 4A). Par-1a null mice showed a similar disproportionate increase in overall adiposity (BAT-CD versus BAT-HFD, P = 0.007; WAT-CD versus WAT-HFD, P = 1.4e⁻⁹; t tests) with a weight increase in BAT and gonadal WAT that eliminated the reduced adiposity observed under chow diet (Fig. 4A, columns a versus b and c versus d, both P = 0.08, t tests).

The dissection of the gonadal fat pads (Fig. 4B) representative of the WAT showed a disproportionate decrease in the fat mass of Par-1a null mice relative to that of WT mice on CD (Fig. 4C; raw weight values provided). The weight difference corresponds to ∼48% (Par-1a^−/−) and ∼36% (Par-1b^−/−) of the WT-WAT weight (Fig. 4C). Although we did not quantitatively assess abdominal fat, during dissection the dimensions of mesenteric root, omental, and retroperitoneal fat content were strikingly lower in Par-1a null mice, and this difference is not accounted for in the displayed values of gonadal WAT (Fig. 4C).

The dissection of the interscapular fat depot, representative of BAT, is shown in Fig. 4E. The weight difference corresponds to ∼52% (Par-1a^−/−) and ∼38% (Par-1b^−/−) of the WT-BAT weigh (Fig. 4F). For both adipose tissues, the difference significantly exceeded the overall ∼9% weight difference observed in Par-1a null mice. Adipocyte counts from six independent samples of WAT and BAT indicated no differences in adipocyte size from WT and Par-1a null mice (for WAT, WT mice showed 436 ± 100 adipocytes per mm² and Par-1a^−/− mice showed 419 ± 74; P = 0.137, t test; for BAT, WT mice showed 1,894 ± 349 and Par-1a^−/− mice showed 1,385 ± 434; P = 0.382, t test). Thus, we conclude that the observed decrease in the adipose tissue of Par-1a null mice is due to decreases in total adipocyte cell number, not cell size, findings similar to those made for Par-1b null mice (43).

We also determined the weight of muscle samples (tibialis anterior, gastrocnemius, plantaris, diaphragm, and extensor digitorum longus) typically used for the assessment of myogenic phenotypes and fiber type composition (22, 84), which, under HFD conditions, showed significantly decreased weights (WT, P = 3.7e⁻⁶; Par-1^−/−, P = 0.006; both t tests) (Fig. 4A, column e). Skeletal muscle plays an integral role in the coordination of fuel homeostasis (41) and is the most important site of insulin-stimulated glucose disposal (26). We therefore performed functional (motor, biodistribution, and microPET analysis), biochemical (glycogen quantification), and morphological (routine, PAS, fiber type analysis, and ultrastructure) screening of skeletal muscle but observed no differences between wild-type and Par-1a null mice (not shown). Thus, the reduced body weight of Par-1a mice on HFD likely is accounted for by a combination of reduced muscle mass (3, 92) and reduced adiposity, although the latter is not reflected in the gonadal WAT measurements (Fig. 4A, columns c and d).

Par-1a loss causes decreased WAT glucose uptake in the absence of BAT phenotype.

MicroPET imaging of WAT was performed (not shown); however, the reduced size of Par-1 null mice and high tracer concentrations in kidneys, bladder, and skeletal muscle precluded the meaningful quantification of uptake in retroperitoneal, gonadal, and calf fat deposits, respectively. We therefore applied invasive biodistribution analysis, and in contrast to Par-1b null mice, WAT of Par-1a null mice showed significantly decreased glucose uptake under normal dietary conditions (WT, 0.52 ± 0.08%ID/g [n = 9]; Par-1a^−/−, 0.32 ± 0.05%ID/g [n = 8]; U [sum of signed ranks] = 14; P = 0.036, Mann-Whitney U test; P = 0.05, t test) (Fig. 4D). When challenged with HFD, Par-1a null mice showed a statistically significant increase in [¹⁸F]FDG uptake compared to that of Par-1a null mice on chow diet that eliminated the difference of glucose uptake from that of WT mice on HFD (Fig. 4D).

We next examined glucose uptake in BAT of CD-fed Par-1a null mice by microPET analyses of the interscapular BAT (Fig. 4G). We evaluated four different conditions at 1 h after [¹⁸F]FDG injection: the fed and fasted state using acute insulin challenge versus saline administration. As part of the HFD trial, we also examined BAT glucose uptake via biodistribution analysis in all genotypes. We found that the loss of Par-1a did not significantly alter BAT glucose uptake in any of these six tested conditions (not shown).

Although Par-1a null mice exhibited reduced fat mass on chow diet and delayed weight gain relative to that of WT mice on HFD, sampled gonadal adipose mass in Par-1a null mice was not significantly different from that of WT mice after 16 weeks of HFD conditions (P = 0.12 and 0.13, both t tests; for WAT/BAT, n = 8 Par-1a^−/− mice and 17 WT mice) (Fig. 4C and F). In contrast, the BAT/WAT mass in Par-1b null mice remained reduced under HFD conditions compared to that of WT mice on the same diet (P = 0.002 and 0.04, t tests [WAT/BAT]; n = 6 for Par-1b^−/− and 17 for WT) (Fig. 4C and F).

Par-1a^−/− mice show resistance to hepatic steatosis.

The chronic exposure of mice to HFD causes liver injury via the accumulation of lipids, leading to fatty liver (steatosis) (13, 79). The absence of steatosis in Par-1a null mice was apparent from the gross examination of the liver, which was notably darker than that of the WT littermates (not shown), and histologic examination showed abundant panlobular, macrovesicular steatosis in WT (Fig. 5 A, image b, inset) but not Par-1a null livers (Fig. 5A, image d). Remarkably, with respect to hepatic steatosis, Par-1a null mice on HFD were no different than WT mice on CD (Fig. 5A, compare image d to image a). Based on previous findings (43), we quantified hepatic lipid content in fresh-frozen sections stained with oil red O and found 14.6 to 25% in WT versus 6.4 to 7.7% in Par-1a null mice when expressed as percent fat per high-power field (Fig. 5B). The quantitative difference in the WT on chow diet (Fig. 5B) corresponds to an absence of centrilobular (zone III) lipid accentuation in Par-1a null mice (Fig. 5A, compare image a to image c). In addition, biodistribution studies demonstrated identical [¹⁸F]FDG uptake in livers of Par-1a null and WT mice on CD and HFD (not shown). Similarly, the microPET examination of the liver using four different metabolic conditions (see Materials and Methods) did not show significant differences (P value range, 0.12 to 0.71; t tests). These findings indicate that Par-1a null mice show resistance to steatosis in the absence of altered hepatic glucose uptake under both normal and HFD conditions.

Decreased glycogen deposition in Par-1a^−/− livers leads to hypoketotic hypoglycemia.

The storage and utilization of glucose is one of the main contributors to the pathophysiologic changes in diabetes (21, 90). PAS staining was used to detect glycogen content (Fig. 5C). Fasted WT liver showed partially depleted liver glycogen stores (Fig. 5C, image a), whereas liver sections from fasted Par-1a null mice exhibited the complete depletion of hepatic glycogen content (PAS quantification [fasted]: WT, 23.6 ± 8.5; Par-1a^−/−, 0.17 ± 0.13; P = 0.01, t test) (Fig. 5C, image c). This depletion was not observed in Par-1b null mice (26.68 ± 2.8; P = 0.73, t test) (Fig. 5C, image e). Two hours after glucose administration, glycogen levels were substantially elevated and visible and confluently PAS positive in livers of WT mice (Fig. 5C, image b). In contrast, Par-1a null mice showed the absence of the restoration of hepatic glycogen content with only spotty PAS-positive islands (Fig. 5C, image d). The quantitative enzymatic analysis of glycogen was consistent with the PAS staining such that fasted and glucose-stimulated liver glycogen was significantly decreased in Par-1a null animals (Fig. 5E) with no significant difference in Par-1b null mice. After mice were fed, glycogen levels were normal and no significant histological differences were observed between groups (not shown).

To test the physiologic significance of hepatic glycogen depletion in Par-1a null mice, we performed an extended fasting experiment and observed significantly decreased glucose levels in Par-1a null mice after overnight fasting (20 h) compared to that of WT littermates (Fig. 5D). After four additional hours without external energy sources, Par-1a null animal blood glucose levels further declined (P = 0.04 compared to results for the WT mice; t test), and the mice developed hypoketotic hypoglycemia (Fig. 5D). In contrast, Par-1b null mice showed an absence of physiologic decline of blood glucose levels with appropriately low ketone levels (Fig. 5D).

To test whether the tissue differences in glycogen levels relate to glycogen synthesis, we determined total and phosphoglycogen synthase (pGS) levels (Fig. 5F) and found increased levels of GS and pGS in the livers of Par-1a null mice. Since Par-1a null mice demonstrate resistance to steatosis as well as defective gluconeogenesis, we probed for UCP2 in fasted and nonfasted liver samples but found similar protein levels (not shown). Most recently, autophagy has been linked to lipid (91) as well as glycogen/glucose homeostasis (55, 56). Therefore, we monitored LC-3 as a marker for membrane formation and early stages of autophagy but observed no differences (not shown). The electron microscopic examination of livers from fasted WT mice demonstrated irregularly shaped glycogen islands (Fig. 5G, images a and b) and variably sized lipid vacuoles (5G, image b) in close proximity to mitochondria with normal morphology (Fig. 5G, image c). In striking contrast, livers from fasted Par-1a null mice did not have lipid vacuoles, were completely depleted of glycogen (Fig. 5G, images d and e), and had abundant autophagocytic vacuoles throughout the hepatocytes (Fig. 5G, images e and f). These autophagic vacuoles were normally configured and frequently associated with mitochondria (Fig. 5G, image f). Our ultrastructural findings with WT mice indicate a catabolic state, whereas the findings in Par-1a null liver are indicative of more extreme energy deprivation with the self digestion of organelles (Fig. 5G, image f).

Par-1b/Par-1a double mutants are not viable, and at least one allele is necessary for embryonic survival.

Cross-breeding experiments were performed to obtain mice disrupted for both Par-1a and Par-1b. After an initial experiment where 24 intercrosses were performed and only six of the nine possible genotypes were obtained, genotype frequencies in ongoing HH × HH crosses were monitored. Eventually two of the three missing genotypes were obtained (Fig. 6A) but at a substantial deviation from expected Mendelian ratio (probability of 2.176911e⁻¹⁹⁸ using the chi-square test with 8 degrees of freedom) and with a complete absence of mice lacking both Par-1a and Par-1b (KK). In addition, there was a drastic reduction in the numbers of KH and HK mice (0.5 and 2.7%, respectively). Instead of the expected 12.5% (corresponding to an estimated 308 mice; gray background in Fig. 6A), we obtained a total of 14 KH and 69 HK mice (out of 2,466). Pregnant females that had been superovulated from HH intercrosses were sacrificed, and fetuses at 10.5 (n = 4) or 8.5 (n = 4) dpc were isolated. None of the obtained sets of embryos (total n = 31) included the KK genotype (expected frequency, ∼2 mice). All other genotypes were obtained (not shown). These results illustrate that Par-1b/Par-1a double mutants are not viable and that at least one allele of either Par-1a or Par-1b is necessary for viability and likely embryonic survival.

When normalized to age-matched WT controls, Par-1b null mice are ∼20% reduced in body weight (43) and Par-1a null mice are ∼9% reduced in body weight (this study). The loss of one Par-1a allele in Par-1b null mice (KH) led to a more-pronounced reduction in body weight (30%) than that observed in Par-1a null mice containing a single allele of Par-1b (HK, 23%) (Fig. 6B). HK and KH mice were not healthy, and a high proportion of these animals died shortly after birth (not shown). The severity of phenotypes observed in each genotype followed the order KK > KH > HK > KW > WK > WW and suggests an intriguing gene dosage effect at the organism level.