FIAT Deletion Increases Bone Mass But Does Not Prevent High-Fat-Diet-Induced Metabolic Complications.

Factor inhibiting activating transcription factor 4 (ATF4)-mediated transcription (FIAT) interacts with ATF4 to repress its transcriptional activity. We performed a phenotypic analysis of Fiat-deficient male mice (Fiat-/Y) at 8 and 16 weeks of age. Microcomputed tomography analysis of the distal femur demonstrated 46% and 13% age-dependent increases in trabecular bone volume and thickness, respectively, in Fiat-/Y mice. Cortical bone measurements at the femoral midshaft revealed a substantial increase in cortical thickness in older Fiat-/Y mice. Bone gain was related to increased mineral apposition rate and increased osteoblast function. Femoral stiffness and strength were substantially increased in Fiat-/Y compared with wild-type (WT) mice. We also investigated whether FIAT contributes to metabolic function. When fed standard mouse chow, Fiat-/Y animals were glucose-tolerant. However, when fed a high-fat diet (HFD) for 8 weeks, Fiat-/Y mice gained more weight than control mice, with a specific increase in white adipose tissue fat mass. The increase in fat mass was due to reduced energy expenditure, which correlated with reduced fatty acid oxidation and lipolysis in the adipose tissue of mutant mice. The expression of the Scd1 gene, involved in lipogenesis, was upregulated in the subcutaneous adipose tissue of Fiat-/Y mice. Moreover, HFD-fed Fiat-/Y mice exhibited increased circulating leptin and insulin levels relative to WT mice, demonstrating that endocrine abnormalities are associated with the disturbance in energy balance. We conclude that Fiat-/Y mice exhibited an anabolic bone phenotype but displayed increased susceptibility to developing metabolic-related disorders when consuming an HFD.

Bone formation is a highly regulated process, and much progress has been made in understanding the growth factors, extracellular cues, signaling molecules, and transcription factors that control gene expression programs to regulate osteoblast proliferation, differentiation, and maturation [reviewed by Karsenty et al. (1)]. Among them are members of the basic domain-leucine zipper (bZIP) family such as Fra-1, ΔFosB, and activating transcription factor 4 (ATF4) (2–4). Gain- and loss-of-function studies in mice have demonstrated that these transcription factors control osteoblast differentiation and activity.

To elicit their transcriptional function, these factors need to interact with additional proteins such as Jun family members to form active homo- or heterodimers, which then bind to the promoter of target genes and initiate transcription (5). The activity of bZIP factors such as ATF4 and Fra-1 is under strict control in a number of cell types, including osteoblasts. Several levels of regulation have been identified that affect the activity of these molecules. These include transcriptional regulation (6), stability (7, 8), post-translational modifications such as phosphorylation (4, 9, 10), and interaction with dimerization partners (5, 11). The dimerization partner appears to influence specificity of DNA binding (12) and transcriptional activity (13, 14). Another mode of bZIP factor regulation involves interaction with leucine zipper partners to form inactive heterodimers. One such inhibitory leucine zipper molecule is FIAT (factor inhibiting ATF4-mediated transcription), whose name was created because of its ability to repress ATF4-dependent osteocalcin (Bglap) gene transcription (15).

FIAT is a 66-kDa protein containing 3 leucine zipper motifs, with the last zipper contained within the extended C-terminal coiled coil. In silico computer modeling of the FIAT primary sequence failed to identify a DNA-binding basic domain, suggesting that FIAT could not act as a sequence-specific bZIP transcriptional regulator. It was shown that although FIAT is not capable of homodimerization, it can instead form heterodimers with ATF4 in mammalian osteoblasts (15, 16). FIAT forms a dimer with ATF4 through its second zipper (16) to inhibit binding of ATF4 to its cognate DNA motif and thus inhibit osteoblast activity (15–17). In addition, FIAT has been shown to repress the transcriptional activity of the FRA-1/cJUN heterodimer without influencing homodimeric cJUN-dependent transcription (18). Therefore, FIAT acts as a negative regulator of osteoblast function by modulating early and late osteoblast activity by interacting with ATF4 and FRA-1, respectively.

FIAT’s importance in bone physiology became apparent after characterization of the FIAT transgenic mice phenotype, which has a reduced bone mass and impaired bone rigidity. Reduction in osteoblast function was shown to be the cause of the low bone mass phenotype in these mice (15). Because transgenic mice overexpressing FIAT are osteopenic, we hypothesized that, inversely, inhibition of FIAT would augment ATF4 function and lead to increased osteoblast activity. This hypothesis was validated in tissue culture using small interfering RNA-mediated knockdown of FIAT, which resulted in increased osteocalcin (Bglap) gene transcription, type I collagen synthesis, and mineralization (17). To gain further evidence of the regulatory role of FIAT in vivo, we engineered mice deficient for Fiat. Because the Fiat gene (NM_178935.4) is located on the X chromosome, phenotypic manifestations were assessed in male progeny (test genotype, Fiat). We compared the skeletal phenotype of Fiat mice to control littermates as a function of age. Moreover, because increasing evidence has supported the existence of a reciprocal regulation between bone and energy metabolism (19) that could involve the targets of FIAT-mediated repression, ATF4 and FRA-1 (20–22), we investigated potential metabolic phenotypic manifestations in Fiat mice after dietary manipulation.

Materials and Methods

Fiat knockout mice

The Fiat-deficient mouse strain has been described previously (23). The McGill institutional animal care and use committee reviewed and approved all animal procedures, which followed the guidelines of the Canadian Council on Animal Care. The mice were kept in an environmentally controlled barrier animal facility with a 12-hour light, 12-hour dark cycle and fed either standard mouse chow (10% fat; Charles River Laboratories International, Wilmington, MA) or high-fat diet (HFD; 60% fat; Envigo, Indianapolis, IN) for 8 weeks, starting from 8 weeks of age for metabolic phenotyping. Water was provided ad libitum. DNA was prepared from tail snips using standard procedures and analyzed using polymerase chain reactions (PCRs) for the presence of the Fiat wild-type (WT) and targeted alleles, as previously described (23).

Reverse transcription quantitative PCR

Sixteen-week-old male mice were euthanized, and the tissues were rapidly dissected for RNA isolation. The epiphysis and all soft tissues were removed from the tibiae, and the marrow was flushed by centrifugation in nested microcentrifuge tubes. Both calvariae and tibiae samples were cut into small pieces with sterile scissors and homogenized in TRIzol reagent (Thermo Fisher Scientific Life Sciences, Waltham, MA). Primary osteoblasts and bone marrow stromal cells were also harvested and homogenized in TRIzol. Further steps were performed according to the manufacturer’s recommendations.

Liver and white and brown fat were isolated and homogenized using QIAzol reagent (Qiagen, Hilden, Germany) following standard protocol procedures.

We used 500 to 1000 ng of RNA for cDNA synthesis using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Reverse transcription quantitative PCR (qPCR) was performed using the TaqMan Kit (Applied Biosystems) and TaqMan assays (Applied Biosystems) for target genes, using β-microglobulin as a housekeeping reference. The reactions were run on a 7500 instrument (Applied Biosystems).

Osteoblast cultures

Primary osteoblasts were isolated from the long bones of WT control and Fiat-deficient mice, as described previously (24). The cells were plated in differentiation media containing 50 μg/mL ascorbic acid and 2 mM β-glycerophosphate. The medium was renewed 3 times each week.

Western blotting

Proteins were extracted from primary osteoblasts cultures in radioimmunoprecipitation assay buffer, separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and transferred to a nitrocellulose membrane (GE Healthcare, Little Chalfont, UK). For detection of FIAT, a polyclonal rabbit antibody directed against the N-terminus of FIAT was used (15). As a loading control, a monoclonal mouse antibody against β-actin (Sigma-Aldrich, St. Louis, MO) was used. Horseradish peroxidase-coupled secondary antibodies (GE Healthcare) and ECL reagent (GE Healthcare) were used for detection.

Microcomputed tomography

Femurs from 8- and 16-week-old mice were isolated and cleaned of soft tissue, fixed overnight in 4% paraformaldehyde, washed with 1× phosphate-buffered saline, and dehydrated in 50% and 70% ethanol. Samples were stored in 70% ethanol at 4°C until scanning. The femurs were imaged using a SkyScan 1172 μCT system (Bruker, Belgium). Imaging of cancellous bone was performed at the distal metaphysis, and imaging of the diaphyseal cortical bone was performed at the midpoint of the femur. The parameters were 5-μm pixel size, 50 kV, 194 μA, 0.5 mm Al filter, angular rotation step of 0.45°, and an exposure time of 500 ms, with a total scan duration of 45 minutes. Three-frame averaging was used to improve the signal-to-noise ratio. After scanning, 3-dimensional (3D) microstructural image data were reconstructed using the manufacturer’s software (Skyscan NRecon). The ring artifact correction was 4, and the beam hardening correction was set at 30%. The following parameters were measured: tissue volume (TV; in mm3), trabecular bone volume (BV; in mm3), BV-to-TV ratio (in percentages), trabecular number (in 1/mm), trabecular separation (in μm), and trabecular thickness (in μm). Cortical bone measurements consisted of the cross-sectional cortical BV and bone area, cortical thickness, percentage of total cortical porosity, and endosteal and periosteal diameter (in mm).

Biomechanical parameters

To assess mechanical integrity, right femora were loaded to failure in a 3-point bending assay using an Instron model 5943 single column table frame machine (Instron, High Wycombe, UK). The tests were performed in the posterior-to-anterior direction at a loading rate of 0.05 mm/s and a support distance of 7 mm. From the resulting load vs displacement curves, the structural properties were calculated to describe the specific aspects of the tissue’s mechanical behavior under load. Those properties included stiffness (N/mm), maximum load at failure (N), and energy to failure (N × mm). Stiffness was defined as the slope of the elastic portion of the load displacement curve.

Serum parameter measurements

Blood was collected from the mice, and serum was isolated by centrifugation and stored at −80°C until further analysis. Bone resorption markers, C-terminal telopeptide (CTX; Immunodiagnostic Systems, The Boldens, UK) and tartrate-resistant acid phosphatase 5b (TRACP 5b; Immunodiagnostic Systems) were measured using enzyme-linked immunosorbent assay kits. Leptin and insulin were measured using enzyme-linked immunosorbent assay kits (EMD Millipore, Billerica, MA).

Bone formation analysis

For trabecular bone formation, the mice were injected with calcein (20 mg/kg) 10 and 3 days before euthanization. Tibial bone samples were dissected, fixed in 4% paraformaldehyde (Sigma-Aldrich), dehydrated in ascending concentrations of ethanol (70% to 100%), and embedded in methyl methacrylate (Fisher Scientific) using standard procedures for nondecalcified bone. Consecutive longitudinal sections (5 μm thickness) of tibiae were cut using a saw microtome (Leica RM2255; Leica Biosystems, Nussloch, Germany). Standard bone histomorphometry was performed using commercial software (BIOQUANT OSTEO, Nashville, TN). Based on histomorphometry standards, we determined the percentage of single-labeled bone surface and double-labeled bone surface, mineralizing surface, mineral apposition rate, and bone formation rate (BFR) per unit of bone surface on the trabecular, periosteal, and endocortical surfaces.

In vitro mineralization

After 14 and 24 days of culture, primary osteoblasts were washed with phosphate-buffered saline and fixed with 10% paraformaldehyde (Sigma-Aldrich) for 15 minutes at room temperature. After washing, the cells were stained with 2% Alizarin S red staining (pH, 4.2; Sigma-Aldrich) and incubated for 20 minutes at room temperature with gentle shaking. After aspiration of the dye, the wells were washed repeatedly with distilled water while shaking for 5 minutes. For quantification of staining, 10% (volume-to-volume ratio [v/v]) acetic acid was added to each well, and the plates were incubated at room temperature for 30 minutes while shaking. The cells were then scraped from the plate with a cell scraper (Fisher Life Sciences, Waltham, MA) and transferred with 10% v/v acetic acid to a 1.5-mL microcentrifuge tube. The slurry was vortexed for 30 seconds, heated to 85°C for 10 minutes, and transferred to ice for 5 minutes. Once the slurry had cooled, it was centrifuged at 20,000g for 15 minutes, and the supernatant was removed to a new 1.5-mL microcentrifuge tube. Next, 10% v/v ammonium hydroxide was added to neutralize the acid. Aliquots of the supernatant were read at 405 nm in 96-well plates.

Isolation and adipogenic differentiation of bone marrow stromal cells

Primary bone marrow stromal cells (BMSCs) were isolated from the femurs and tibia of 8-week-old WT and mutant littermate mice. In brief, the bone marrow was flushed with α-minimal essential medium supplemented with 20% fetal bovine serum (FBS) and 1% penicillin-streptomycin. The adherent cell fraction was allowed to expand until 80% confluent and then replated for differentiation assays.

For adipogenic differentiation, BMSCs were plated at a density of 20,000 cells/cm2 in Dulbecco’s modified Eagle medium/20% FBS, and differentiation was induced 24 hours after plating with a cocktail containing 1 µM dexamethasone (Sigma-Aldrich), 5 µg/mL insulin (Sigma-Aldrich), 0.5 mM isobutyl methylxanthine (Sigma-Aldrich), and 1 µM rosiglitazone (Sigma-Aldrich). Two days later, the media were changed to adipogenic maintenance medium consisting of Dulbecco’s modified Eagle medium/20% FBS plus 5 µg/mL insulin for the remainder of the experiment.

Adipocyte lipid droplets were visualized by staining with 0.2% Oil Red O (Sigma-Aldrich) in 60% isopropanol after fixing the cells in 10% neutral-buffered formalin.

Physiological measurements

Energy expenditure parameters were measured using a 4-chamber Oxymax system (Columbus Instruments, Columbus, OH). After a 24-hour acclimation period, oxygen consumption and carbon dioxide production data were collected for 24 hours. Heat production was calculated using the following formula: heat = (calorific value × oxygen consumption)/body weight.

Glucose tolerance test

Sixteen-week-old male mice were fasted overnight and administered 2 mg of d-glucose (Sigma-Aldrich) per gram of body weight by intraperitoneal injection. The blood glucose levels were measured in blood samples from tail vein puncture using the OneTouch Ultra2 glucose meter (LifeScan Europe, Zug, Switzerland) at 15, 30, 60, and 120 minutes after glucose injection.

Statistical analysis

Kolmogorov-Smirnov tests were used to determine whether the distribution of the variables was normal for each data set. The results are expressed as the mean ± standard error of the mean. Statistically significant differences for the 2-group comparisons were calculated using an unpaired Student’s t test. For blood biochemistry data, 2-way analysis of variance with the Bonferroni post hoc test was used to determine the effects of diet and genotype. A P value < 0.05 was used as the criterion for statistical significance.

Results

Fiat gene inactivation and gross phenotype

The Fiat gene (NM_178935.4) is located on the X chromosome; therefore, the effect of Fiat inactivation was examined by comparing WT (+/Y) to mutant (−/Y) male littermates. The efficiency of gene inactivation was verified using reverse transcription qPCR and Western blot analysis. As shown in Figure 1(A), Fiat messenger RNA (mRNA) was not detectable in the calvaria from mutant mice. Fiat mRNA from total RNA extraction of spleen, brain, and heart tissue also was not detectable (data not shown), indicating that we generated global Fiat-deficient mice. We also isolated primary osteoblasts to determine FIAT protein expression and measured undetectable levels in mutant mice [Fig. 1(B)]. Mutant mice are viable and fertile without obvious morphological abnormalities and found at the expected Mendelian ratios (data not shown), suggesting that FIAT is not necessary for embryonic development. In addition, no substantial difference in bodyweight between WT and Fiat mice at 8 weeks [Fig. 1(C)] and 16 weeks [Fig. 1(D)] of age was observed when the mice were fed standard mouse chow.

Figure 1.

Fiat expression and weight gain in FIAT-deficient mice. (A) Fiat mRNA and (B) FIAT protein were not detectable in knock-out male (Fiat) mice. β-Actin was used as a protein loading control; the arrowhead points to the FIAT protein in the Western blot. Bodyweight (BW) was comparable between genotypes at (C) 8 weeks and (D) 16 weeks of age.

Skeletal phenotype of Fiat mice

To assess cancellous bone mass in Fiat-deficient mice, the femurs from male mice at 8 and 16 weeks of age were examined using μCT. A comparison of representative 3D reconstruction images of distal femurs of 16-week-old mice showed clear differences in trabecular bone mass [Fig. 2(A) and 2(B)]. The cancellous fractional bone volume (bone volume/tissue volume [BV/TV]), assessed by μCT at the metaphyseal region of distal femur tended to increase in Fiat mice compared with WT at 8 weeks of age and the increase persisted significantly until 16 weeks [Fig. 2(C)]. The increase in BV/TV was associated with an increase in trabecular thickness [Fig. 2(D)]. The trabecular number [Fig. 2(E)] and tissue volume [Fig. 2(G)] exhibited upward trends in both 8- and 16-week-old Fiat mice; however, the trabecular separation remained the same between the WT and Fiat mice [Fig. 2(F)]. In addition, the structure model index (SMI), which quantifies the 3D structure for the relative amount of plates (SMI of 0 indicates strong bone) and rods (SMI of 3 indicates fragile bone), was lower in the Fiat mice compared with the WT mice. However, the difference was not statistically significant [Fig. 2(H)].

Figure 2.

Long bone trabecular phenotype of Fiat-null mice. (A, B) Three-dimensional reconstruction of 25-μm trabecular bone obtained by μCT from distal femur of WT and Fiat 16-week-old male littermates. μCT scans were used to measure (C) BV over TV (BV/TV), (D) trabecular thickness, (E) trabecular number, (F) trabecular separation, (G) tissue volume, and (H) SMI as a function of age by analysis of 250-μm of trabecular bone under the primary spongiosa of the distal femur. Analyses were performed on 9 to 14 mice of each genotype at 8 and 16 weeks of age. **P < 0.01.

Cortical bone from Fiat mice was also assessed using μCT at the femoral midshaft in 8- and 16-week-old mice. The cortical bone volume [Fig. 3(A)], bone area [Fig. 3(B)], bone thickness [Fig. 3(C)], and periosteal diameter [Fig. 3(D)] were increased in older (16 weeks) Fiat male mice. The difference in cortical porosity was not statistically significant between the genotypes at either age tested [Fig. 3(E)].

Figure 3.

Cortical phenotype of Fiat-deficient mice. (A) Cortical bone volume, (B) area, (C) thickness, (D) periosteal diameter, and (E) cortical porosity were measured by μCT analysis in 8-week-old (n = 11 to 12) and 16-week-old (n = 9 to 14) WT or Fiat male mice. *P < 0.05, **P < 0.01.

Mechanical testing

The substantial increase in the cortical parameters of 16-week-old Fiat mice suggested a mechanically stronger cortical architecture in these mice. The results of 3-point bending tests at the femur diaphysis indicated a substantial increase in stiffness [Fig. 4(A)], maximum load [strength; Fig. 4(B)], and energy to failure [Fig. 4(C)] in 16-week-old Fiat mice compared with age- and sex-matched WT littermates.

Figure 4.

Bone biomechanical properties of 16-week-old Fiat-null mice. Three-point bending tests revealed substantial changes in (A) stiffness, (B) maximum load, and (C) energy to failure between genotypes. *P < 0.05.

Increased osteoblast activity in the absence of FIAT

We next investigated whether changes in the Fiat mice bone mass were accompanied by changes in the markers of bone formation or bone resorption. To address this, the expression of genes from total calvarial RNA isolated from 16-week-old WT and Fiat mice was evaluated by qPCR. The mRNA level of osteocalcin (Bglap) was significantly increased [Fig. 5(A)]. Additionally, an increase in Bglap expression was measured in RNA isolated from adult mice tibiae (data not shown). In contrast, expression of Rankl [Tnfsf11; Fig. 5(B)] and serum levels of TRACP 5b and CTX did not change in Fiat mice relative to WT mice [Fig. 5(C) and 5(D), respectively], indicating that resorption is not different between the 2 genotypes.

Figure 5.

Fiat-deficient mice have enhanced osteoblast activity. qPCR of (A) osteocalcin (Bglap) and (B) Rankl (Tnfsf11) transcript levels in 16-week-old mice calvaria, normalized to β2-microglobulin. Measurement of the level of circulating (C) TRACP 5b and (D) CTX in the serum of 16-week-old Fiat and WT males (n = 9 to 13). (E) The mineral apposition rate (MAR), (F) bone formation rate per unit of tissue volume of trabecular bone (BFR/TV), and (G) bone formation rate per unit of bone surface (BFR/BS) were measured by double calcein staining (WT, n = 9; Fiat, n = 12). *P < 0.05; **P < 0.01.

To further determine the underlying mechanism for the elevated cancellous bone mass observed in Fiat-deficient mice, we performed dynamic histomorphometry in 16-week-old animals. Fiat−/Y mice displayed a substantial increase in mineral apposition rates in metaphyseal cancellous bone compared with age-matched WT controls [Fig. 5(E)]. The BFR using different measures such as TV and bone surface has been shown to measure different bone formation properties (25). Although the BFR per unit of tissue volume increased 52% (P = 0.06) in the Fiat−/Y mice [Fig. 5(F)], the BFR adjusted for bone surface showed a lower increasing trend in the Fiat−/Y mice compared with the controls [Fig. 5(G)].

To confirm that the increased bone formation phenotype of Fiat-deficient mice was caused by increased osteoblast activity, primary osteoblasts were isolated from the long bones of WT and Fiat mice, cultured in differentiation media for 14 and 24 days, and stained for mineralized matrix formation. Fiat-deficient osteoblasts demonstrated increased nodule formation [Fig. 6(A)] and alizarin red staining under osteogenic conditions [Fig. 6(B)]. Next, we examined the effect of Fiat ablation on the expression of osteogenic marker genes in the primary cultures. The expression of alkaline phosphatase [Alpl; Fig. 6(C)], Bglap [Fig. 6(D)], and Runx2 [Fig. 6(E)] was strongly increased. Collectively, these data suggest that FIAT plays a role as a negative regulator of osteoblast differentiation and mineralization through regulation of osteoblast-specific genes.

Figure 6.

Mesenchymal lineages differentiation of Fiat-deficient cells. (A) Formation of mineralized matrix was examined by alizarin red S staining of cultured primary osteoblasts isolated from long bone of WT (top) and Fiat (bottom) mice. (B) Cells were fixed after 24 days of culture in osteogenic medium, and alizarin red S staining was quantified by absorbance at 405 nm. Expression of osteoblast markers (C) Alpl, (D) Bglap, and (E) Runx2 in primary osteoblasts cultured in osteogenic medium for 14 and 24 days was measured by reverse transcription qPCR analysis, and the values were normalized to β2-microglobulin mRNA expression. (F) BMSCs from WT and Fiat mice (n = 3 per genotype) were induced to undergo adipogenic differentiation with staining for Oil Red O after 6 days. (G) mRNA levels for adipogenic markers Paprγ, adiponectin (Adipoq), CebpA, CebpB, and Fabp4 were measured by reverse transcription qPCR. *P < 0.05; **P < 0.01; ***P < 0.001.

We next tested whether FIAT deficiency affected >1 lineage derived from mesenchymal stromal cells. To investigate the adipogenic potential of mesenchymal progenitors in Fiat mice, we isolated BMSCs and subjected them to adipogenic differentiation. We found a substantial reduction in adipogenic differentiation of Fiat BMSCs, as indicated by Oil Red O staining [Fig. 6(F)]. The mRNA levels of the adipogenic differentiation markers Pparγ, adiponectin (Adipoq), CebpA, and Fabp4 were decreased in BMSCs from Fiat mice compared with WT mice [Fig. 6(G)].

Increased body weight and fat mass in Fiat-deficient mice

To assess a putative role of FIAT in energy homeostasis, Fiat and WT control mice were fed an HFD for 8 weeks, starting at 8 weeks of age. Before the start of the dietary manipulation, no genotype differences in body weight had been observed between the 8-week-old mice. During the 8 weeks of the HFD, we compared the weight of the Fiat mice with that of their WT littermates and observed a progressive increase in body weight in the Fiat-deficient male mice [Fig. 7(A)]. At study termination, the subcutaneous and epididymal fat pads were dissected and weighed. A nonstatistically significant tendency toward increased epididymal fat (epididymal white adipose tissue [WAT]) in Fiat mice was observed [Fig. 7(B)]; however, subcutaneous fat (subcutaneous WAT [sWAT]) was significantly greater in the Fiat mice than in the WT controls [Fig. 7(C)]. In addition to these WATs, brown adipose tissue (BAT) is present in mammals and regulates energy expenditure and thermogenesis (26). When analyzing the expression levels of markers of BAT, Pgc1α and Ucp1, we observed that these markers were unaffected by the HFD treatment between genotypes (data not shown). This suggests that BAT was not the mediator of the effect of Fiat deletion on energy expenditure during HFD consumption.

Figure 7.

HFD-fed Fiat mice had substantial increases in body weight and fat mass. (A) Body weight growth curves for Fiat mice (n = 8) and WT mice (n = 6) fed an HFD for 8 weeks, starting at 8 weeks of age. (B) Epididymal white adipose tissue (eWAT) and (C) sWAT were weighed in Fiat mice and control littermates. *P < 0.05; **P < 0.01; ***P < 0.001.

Decreased oxygen consumption in Fiat mice

To determine whether the increased fat mass of Fiat mice could be a consequence of a decrease in whole body energy expenditure, we measured the energy expenditure in the mice of both genotypes using indirect calorimetry. Fiat males had decreased oxygen consumption [Fig. 8(A)], lower carbon dioxide production [Fig. 8(B)], and reduced heat generation [Fig. 8(C)] compared with WT control mice. The decrease in energy expenditure measured in mutant mice was independent of the circadian rhythm [Fig. 8(A–C)].

Figure 8.

Fiat-deficiency decreased energy expenditure in mice fed a high-fat diet. (A, B) Metabolic rates and (C) heat production after 8 weeks of an HFD during dark and light 12-hour phases. (A) Volume of oxygen consumed (VO2). (B) Volume of carbon dioxide produced (VCO2). *P < 0.05; **P < 0.01.

Repressed lipolysis and enhanced lipogenesis in WAT of Fiat mice

The increased fat mass in the Fiat mice prompted us to examine the expression levels of genes involved in lipogenesis, lipolysis, and fatty acid β-oxidation in white adipose tissue of 16-week-old mice. The mRNA levels of the key gene for lipolysis (5′-AMP-activated protein kinase catalytic subunit α2 [Prkaa2]) was decreased in the sWAT of the Fiat mice compared with the WT mice [Fig. 9(A)]. Gene expression of Scd1 (stearoyl CoA desaturase 1), a lipogenic marker, was increased in sWAT [Fig. 9(B)]. Both Prkaa2 gene expression [Fig. 9(C)] and Pgc1α [peroxisome proliferator-activated receptor-γ coactivator 1-α; Fig. 9(D)], a key regulator of mitochondrial biogenesis, were decreased in the epididymal adipose tissue of Fiat mice.

Figure 9.

Expression of genes involved in lipogenesis, lipolysis, and mitochondrial biogenesis. Reverse transcription qPCR of mRNA levels in WATs obtained from WT and Fiat mice after 8 weeks of HFD feeding. mRNA levels of (A) Prkaa2 and (B) Scd1 measured in sWAT and (C) Prkaa2 and (D) Pgc1α measured in epididymal WAT (eWAT). *P < 0.05.

HFD modified glucose metabolism in Fiat mice

We searched for an effect of Fiat deficiency on the ability to clear or metabolize circulating glucose. Intraperitoneal glucose tolerance tests revealed that Fiat male mice managed a glucose load significantly better than did control animals when fed standard chow [Fig. 10(A)]. However, this metabolic advantage was lost when the Fiat mice were fed an HFD. Under those circumstances, the rate of glucose metabolism was lessened in the Fiat mice, such that it was nearly identical to that of the control mice [Fig. 10(B)]. The expression of the glucose transporter 4 (Glut4) in adipose tissue was increased in the Fiat mice fed a standard chow diet [Fig. 10(C)], indicating the importance of the role of GLUT4 in adipose tissue on favorable systemic glucose tolerance [Fig. 10(A)]. The HFD-induced change in glucose tolerance was associated with a marked increase in serum insulin level, suggesting that HFD-fed Fiat mice are more resistant to insulin than their WT counterparts [Fig. 10(D)]. In addition, as expected from the increased WAT mass, the concentration of the adipose tissue-derived hormone leptin was significantly increased in the serum of Fiat mice at the end of the HFD treatment [Fig. 10(E)]. No change was found in the blood concentrations of insulin [Fig. 10(D)] and leptin [Fig. 10(E)] between the WT and Fiat mice when fed the standard diet.

Figure 10.

HFD feeding deteriorated energy metabolism in Fiat-deficient mice. (A) Glucose tolerance tests were performed in WT and Fiat mice under standard diet (SD) and (B) HFD. Glucose was administered to overnight fasted mice, and the blood glucose levels were sampled from venous tail blood at the indicated time points. (C) Expression of Glut4 in adipose tissue was assessed by reverse transcription qPCR. Levels of (D) circulating insulin and (E) leptin were measured in SD- and HFD-fed mice. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Discussion

To better understand the physiological relevance of FIAT, we used a mouse model in which the Fiat gene had been globally deleted through the use of Cre-LoxP recombination driven by the cytomegalovirus promoter (27). This genetic strategy was selected, because initial attempts at allele inactivation using the Col1a1-Cre driver strain (28) yielded poor penetrance (data not shown), and the Osx-Cre strain was reported not to be restricted to osteoblasts postnatally (29). The approach also seemed appropriate because the ATF4 protein, the main dimerization partner of FIAT, accumulates predominantly in osteoblasts (8). The skeletal phenotype of Fiat-deficient mice was shown to be associated with an increase in osteoblast activity (Figs. 5 and 6).

In the Fiat mice, μCT data showed an age-dependent increase in bone volume throughout the skeleton, affecting the trabecular and cortical bone. This increase was measured as statistically significant at 16 weeks of age. The mineral apposition rate was markedly increased in the Fiat mice for trabecular bone, indicating an increase in the amount of bone matrix deposited per active osteoblast cluster. Analysis of the biomechanical properties by mechanical testing of 16-week-old control and Fiat-deficient male mice demonstrated that the increase in bone mass and microarchitecture after Fiat deletion translated into substantial increases in bone strength and stiffness. To unravel the cellular mechanisms that account for the high bone mass, we undertook differentiation studies of cultured primary osteoblasts derived from mice long bones. Consistent with the in vitro phenotype in which FIAT expression was inhibited using short hairpin RNA-mediated knockdown in MC3T3-E1 osteoblastic cells (17), primary osteoblast cultures obtained from Fiat mice demonstrated increased mineralization and increased expression of both early and late osteoblast-specific markers (Runx2, Alpl, and Bglap). Reciprocally, adipogenic differentiation of Fiat-deficient BMSCs was reduced. Thus, we concluded that FIAT normally inhibits osteogenic differentiation.

The mRNA levels or circulating concentrations of osteoclast markers (Rankl, TRACP 5b, and CTX) remained unchanged. Therefore, inactivation of Fiat seems to result in an absence of the coupling that typically exists between osteoblasts and osteoclasts, leading to a high bone mass phenotype.

The effects of Fiat deficiency on bone observed in the Fiat mice are the opposite of, and biologically consistent with, the effects described for Fiat overexpression in transgenic mice in which excess FIAT production resulted in decreased bone mineral density, bone volume, bone formation rate, and bone strength (15). Our in vitro and in vivo studies have demonstrated that FIAT regulates the bone mass through disruption of most ATF4 functions. Therefore, we surmised that the transcription of ATF4 target genes, such as Chop, Tnfsf11, and Bglap, could be modulated through FIAT-mediated signaling. However, we only found a relevant change in the expression level of Bglap, and the levels of Chop and Tnfsf11 were not affected by the loss or gain of Fiat (data not shown), suggesting transcription regulation through other pathways.

We have established the role of FIAT in bone mass control through the study of in vivo mouse models. The recent emergence of the importance of bone as an endocrine organ led us to investigate whether FIAT also has a functional role in bone as a regulator of energy metabolism. In the present study, we examined the phenotype of mice lacking Fiat and challenged with an HFD. Fiat mice exhibited increased body weight and white fat mass caused by reduced energy expenditure in response to the HFD challenge. Energy expenditure could result from changes in thermogenesis, locomotor activity, and/or skeletal muscle mitochondrial function (30). In the present study, we did not find a change in the levels of the BAT genes, Ucp1 and Pgc1, implicated in thermogenesis function, suggesting that the decreased in energy expenditure was not due to alterations in thermogenesis. However, the exact mechanism by which physical activity and/or skeletal muscle mitochondria regulate changes in Fiat mice energy expenditure remains to be determined.

Our results suggest that the increased body weight in Fiat mice was caused by an increase in fat mass. This must involve increased differentiation of adipose-derived vascular stromal cells, which are a distinct source of adipocyte progenitors from BMSCs (31, 32). We hypothesized that changes in fat mass involve modulation of lipogenesis and/or lipolysis. Consistent with this possibility, expression of the lipogenic marker gene Scd1 was shown to be upregulated in sWAT. The lipolysis regulatory gene Prkaa2 was reduced in WATs, and the fatty acid β-oxidative related gene, Pgc1α, showed a decrease in epididymal WAT. Therefore, increased lipogenesis and decreased lipolysis together resulted in the WAT mass increase. Not surprisingly, because the WAT fat mass and body weight were not altered with the standard diet, no change occurred in the lipogenic/lipolysis marker gene expression between the Fiat and WT mice when fed standard mouse chow (data not shown).

It is known that ATF4 regulates lipid metabolism. Studies using Atf4-deficient mice revealed increased energy expenditure, as measured by oxygen consumption, increased expression of lipolysis and β-oxidation marker genes, and decreased expression of lipogenic markers in WATs (21). Remarkably, Fiat mice displayed a metabolic phenotype that is the opposite of the phenotypic effects reported in Atf4 null mice, suggesting that FIAT could affect lipid metabolism through an ATF4-dependent mechanism. To further support this notion at the genetic level, the lipid metabolism could be examined in compound mutant mice lacking 1 allele of Atf4 and 1 allele of Fiat.

Increased body weight is often associated with impaired glucose control and increased insulin secretion. However, when comparing the standard diet-fed Fiat mice with their WT littermates, the Fiat mice exhibited significantly faster glucose clearance after a glucose challenge. Thus, the measured glucose clearance in the Fiat mice after 8 weeks of the HFD, which was nearly indistinguishable from that of the WT controls, was most likely due to the increased body weight gain in these mice. Insulin-mediated glucose homeostasis correlates with the expression of glucose transporter GLUT4. Mice with fat-specific knockout of Glut4 are glucose intolerant and hyperinsulinemic (33), demonstrating that GLUT4 expression in adipose tissue affects metabolic control. In the present study, Fiat mice fed a standard diet developed a threefold increase in the expression of Glut4 in WAT [Fig. 10(C)], indicating that overexpression of GLUT4 seemed sufficient to increase glucose metabolism.

Translocation of GLUT4-containing vesicles to the cell surface is part of the underlying molecular mechanism regulating insulin-mediated increased glucose transport activity (34). Studies have shown that this translocation is mediated by formation of functional SNARE complexes containing syntaxin 4, SNAP23, and VAMP2 (35–37). Nogami et al. (38) have demonstrated that FIAT (which they named γ-taxilin) specifically interacts with syntaxin 4. Thus, it remains pertinent to identify the function of FIAT in molecular events involved in insulin-stimulated GLUT4 translocation.

In summary, the present study identified FIAT as a negative regulator of bone mass acting distinctly on bone formation without alteration in resorption and affecting both the trabecular and the cortical compartments. Fed a standard diet, Fiat-deficient mice did not develop any metabolic abnormalities, although the genetic disruption of Fiat caused increased body weight gain when the mice were exposed to an HFD. This supports the idea that the dietary stressor is responsible for the progression of the metabolic phenotype. We demonstrated that the increased fat mass in Fiat mice is caused by augmented fatty acid synthesis and suppressed fat lipolysis in WAT and by decreased energy expenditure with reduced oxygen consumption. Taken together, our study has identified a possible new function for FIAT in regulating lipid metabolism and energy expenditure.