LDB1 Regulates Energy Homeostasis During Diet-Induced Obesity

Abstract

The broadly expressed transcriptional coregulator LDB1 is essential for β-cell development and glucose homeostasis. However, it is unclear whether LDB1 has metabolic roles beyond the β-cell, especially under metabolic stress. Global Ldb1 deletion results in early embryonic lethality; thus, we used global heterozygous Ldb1^+/− and inducible β-cell–specific Ldb1-deficient (Ldb1^Δβ-cell) mice. We assessed glucose and insulin tolerance, body composition, feeding, and energy expenditure during high-fat diet exposure. Brown adipose tissue (BAT) biology was evaluated by thermogenic gene expression and LDB1 chromatin immunoprecipitation analysis. We found that partial loss of Ldb1 does not impair the maintenance of glucose homeostasis; rather, we observed improved insulin sensitivity in these mice. Partial loss of Ldb1 also uncovered defects in energy expenditure in lean and diet-induced obese (DIO) mice. This decreased energy expenditure during DIO was associated with significantly altered BAT gene expression, specifically Cidea, Elovl3, Cox7a1, and Dio2. Remarkably, the observed changes in energy balance during DIO were absent in Ldb1^Δβ-cell mice, despite a similar reduction in plasma insulin, suggesting a role for LDB1 in BAT. Indeed, LDB1 is expressed in brown adipocytes and occupies a regulatory domain of Elovl3, a gene crucial to normal BAT function. We conclude that LDB1 regulates energy homeostasis, in part through transcriptional modulation of critical regulators in BAT function.

We report that the Ldb1 transcriptional coregulator has roles in energy homeostasis, in part through transcriptional modulation of critical regulators of brown adipose tissue function.

Obesity, imparted by reduced physical activity and a concomitant increase in high-fat food intake, has become a worldwide epidemic with life-altering comorbidities including type 2 diabetes and cardiovascular disease (1, 2). Of the strategies to combat weight gain and metabolic syndrome in humans, many seek to exploit the metabolic benefits of thermogenic brown adipocyte depots. As a result, elucidating the pathways that promote the development and activation of brown adipose tissue (BAT) is of particular interest.

Recently, efforts have revealed transcription factors and interacting coregulators that impart brown adipose cell identity and function (3, 4). In particular, the adipogenic master regulator peroxisome proliferator-activated receptor-γ (PPARγ) is required for the development of both the white and brown adipose lineages (5, 6), albeit insufficient to solely drive the BAT program. Notably, PPARγ positively regulates uncoupling protein 1 (Ucp1), a gene required for mitochondrial proton leak and heat generation in brown adipocytes (7). This regulation is facilitated by the zinc-finger methyltransferase PR domain-containing protein 16, a BAT-enriched coregulator that is necessary and sufficient in driving BAT-selective transcriptional programs (8). Of note, this occurs in part through physical interactions with PPARγ that regulate core BAT genes, including Ucp1, Diodinase type 2 (Dio2), and the cold-induced coregulator PPARγ-coactivator-1α. With increased focus on BAT biology, unique regulators required for brown adipose differentiation programs or energy expenditure (EE) and BAT-related thermogenesis are being uncovered.

We have been interested in the LIM (Lin11, Isl1, Mec3)-domain binding coregulator LDB1, especially in the context of pancreatic islet β-cell development and function (9). LDB1 is a scaffolding factor that avidly binds LIM-domains of the LIM-homeodomain (LIM-HD), and LIM-Only (LMO) classes of transcriptional regulators, which are broadly expressed throughout embryogenesis and in adult tissues. Additionally, LDB1-LMO complexes interact with basic helix-loop-helix and Gata classes of transcription factors, for example during erythropoiesis (10). Thus, our overarching hypothesis is that LDB1 integrates the function of LIM-domain transcription factor complexes throughout development and in the maintenance of adult tissue function.

Based on the importance of LDB1 in pancreatic development and function(9, 11), we initially set out to investigate the role of LDB1 in adult glucose metabolism, especially with regard to pancreatic islet responses to diet-induced obesity (DIO). Hence, we used a global Ldb1 heterozygote (Ldb1^+/−) model challenged with a high-fat diet (HFD). These mice had an expectedly mild pancreatic phenotype, especially when compared with islet-specific constitutive or adult-inducible Ldb1 deficiency models (9, 11). Most notable was a reduction in plasma insulin levels that was likely a secondary response to increased insulin sensitivity. Surprisingly, the DIO Ldb1^+/− mice had reduced food intake coupled with normal body weight gain on HFD. Further, Ldb1^+/− mice had reduced EE regardless of diet, which was associated with impaired expression of thermogenic genes in the BAT of Ldb1^+/− mice. BAT-expressed LDB1 may contribute to these effects, as LDB1 is expressed in BAT and binds a key promoter domain of Elovl3, encoding an elongation enzyme required for the synthesis of very long-chain fatty acids in brown adipocytes, thus facilitating early BAT recruitment and function during cold stress (12). Strikingly, β-cell–specific LDB1 deficiency lacks this BAT phenotype, although plasma insulin is similarly reduced. Taken together, we found that LDB1 is required for normal EE and appears to be a unique regulator of BAT function.

Materials and Methods

Mouse models

Ldb1 null allele mice (Ldb1^-) were described (13). Male Ldb1^+/− and control Ldb1^+/+ littermates (indicated as CTL) were maintained on standard chow (4.7 kcal%, #7917, Envigo, Indianapolis, IN) until 13 weeks of age and then given ad libitum access to HFD for 8 weeks (58.0 kcal% fat, #D12331 Research Diets, New Brunswick, NJ). The conditional, inducible β-cell LDB1 knockout (Ldb1^Δβ-cell) was generated by crossing floxed Ldb1 (Ldb1^F/F) (14) and transgenic MIP-Cre^ERT (15) mice. Cre activity was induced via dosing tamoxifen (T5648; Sigma Aldrich, St. Louis, MO) at 150 μg/g body weight in 8- to 10-week male Ldb1^Δβ-cell mice and control Ldb1^F/F littermates (indicated as CTL). Tamoxifen was dissolved in corn oil and administered by gavage to all mice on 5 consecutive days followed by a 2-day rest, and then repeated for another five daily doses [as described Ediger et al.in (11)]. Mice were allowed a 2-week washout period prior to initiation of HF-feeding. Ldb1^Δβ-cell and control littermates were maintained on standard chow until 13 weeks of age when they were provided ad libitum access to HFD for 4 weeks. All mouse lines were maintained on a mixed (C57BL/6) background. Mice were single or group housed on a 12-hour:12-hour light:dark cycle (light on from 0600 to 1800 hours) at 25°C ± 1°C and constant humidity with free access to food and water except as noted. All studies were approved by and performed according to the guidelines of the University of Alabama at Birmingham Institutional Animal Care and Use Committee.

Body composition and indirect calorimetry

Weekly food intake and body weight measurements were collected. Body composition was measured immediately before and following 8-week (Ldb1^+/− and controls) or 4-week (Ldb1^Δβ-cell and controls) HF feeding using noninvasive nuclear magnetic resonance spectroscopy (EchoMRI; Echo Medical Systems) at the University of Alabama at Birmingham Nutrition Obesity Research Center Small Animal Phenotyping Core. EE and home cage activity were assessed using a combined indirect calorimetry system (Comprehensive Laboratory Animal Monitoring System; Columbus Instruments) as previously described (16). Volume of O₂ consumption and CO₂ production (VO₂ and VCO₂, respectively) were measured every 15 minutes to determine respiratory quotient (RQ) and EE. Home cage locomotor activity was determined using a multidimensional infrared light beam system.

Glucose and insulin tolerance tests

Intraperitoneal (IP) glucose and insulin tolerance tests were performed by injection of glucose (1.5 [HFD] or 2.0 [standard chow] g/kg, 20% w/v d-glucose [Sigma] in 0.9% w/v saline) or regular human insulin (0.75 U/kg [Lilly Humalog] in saline) to 5-hour fasted mice. Blood glucose was measured with a TheraSense Freestyle Glucometer immediately before and 15, 30, 45, 60, and 120 minutes after injection.

Tissue isolation

Ad libitum-fed control and Ldb1^+/− mice were sedated with pentobarbital and decapitated. Plasma was collected from trunk blood, and tissues were harvested and immediately stored at –80°C until analysis.

Immunofluorescence, immunohistochemistry, and immunoblot analysis

Pancreata were prepared for immunofluorescence staining as described (9). Primary antibodies included goat α-LDB1 (1:1000 Santa Cruz) and guinea pig α-Insulin (1:1000 Dako) (Table 1). Cy-2 and Cy-3 conjugated α-goat and α-guinea pig secondary antibodies (1:500; Jackson ImmunoResearch) were used for detection. Images were generated using an Olympus IX81 and processed by CellSens Dimensions (Olympus) or ImageJ (National Institutes of Health) software. Western blotting was performed as described (9). Interscapular brown adipose depots from C57BL/6 mice were dissected, fixed in 4% paraformaldehyde for 24 hours at 4°C, and embedded in paraffin. Sections were rehydrated, and antigen retrieval was performed in 0.01M sodium citrate buffer using a pressure cooker. Sections were blocked with CAS-Block (Invitrogen), primary antibodies (goat α-Ldb1, 1:250; rabbit α-PPARγ, 1:500 Santa Cruz; mouse α-ISL1, 1:1000 DSHB) were added overnight (4°C), and α-goat and α-rabbit secondary antibodies were applied for 40 minutes (37°C). Vectastain Elite ABC Kit (Vector) with Peroxidase Substrate Kit 3,3′-diaminobenzidine (Vector) was used for signal detection, and nuclei were counterstained with hematoxylin.

Table 1.

Antibodies Used in This Study

Peptide/Protein Target	Name of Antibody	Manufacturer, Catalog No., and/or Name of Individual Providing the Antibody	Species Raised in; Monoclonal or Polyclonal	Dilution Used
LDB1	CLIM-2	Santa Cruz	Goat; polyclonal	1:250–1:1000
β-Actin		Cell Signaling	Rabbit; monoclonal	1:1000
PPARγ		Santa Cruz	Rabbit; polyclonal	1:500
Islet-1	39.4D5	DSHB	Mouse; monoclonal	1:1000
Elovl3		Santa Cruz	Goat; polyclonal	1:1000
Insulin		Dako	Guinea pig; polyclonal	1:1000

Quantitative real-time PCR

RNA was isolated using the RNeasy Lipid Mini Kit (Qiagen) according to the manufacturer’s instructions. Complementary DNA was synthesized by reverse transcription polymerase chain reaction (PCR) using SuperScript III, DNase treatment, and anti-RNase treatment (Invitrogen). Single-gene quantitative PCR was performed as previously described (16). Data were normalized to housekeeping genes Gapdh (islet) and Hprt or 36B4 (BAT). See Supplemental Table 1 ^{(61.8KB, pdf)} for list of probes and primer sets.

ChIP

Chromatin immunoprecipitation (ChIP) was performed as described (9, 17). Briefly, brown adipose lobes were dissected from C57Bl/6 wild-type mice, phosphate-buffered saline washed, minced, fixed with 1% formaldehyde, and glycine quenched. Protein-DNA fragments were prepared by sonication (Bioruptor XL; Diagenode), precleared, and then incubated overnight at 4°C with α-LDB1 (Santa Cruz) or control goat immunoglobulin G (IgG). Chromatin complexes were precipitated with Protein G Dynabeads (Life Technologies) at 4°C for 4 hours, washed, and eluted, and crosslinks were reversed. Quantitative PCR was performed on the IP DNA using SYBR Green PCR master mix (Bio-Rad) and a LightCycler 480 II (Roche). Enriched ChIP DNA was normalized to Albumin promoter DNA enrichment (18) and calculated relative to IgG, set as onefold (ΔΔCt). ChIP experiments were performed four times with independent chromatin preparations.

Plasma hormone analyses

Plasma insulin was quantified via enzyme-linked immunosorbent assay assay according to the manufacturer’s instructions (Crystal Chem). Plasma analyte levels were determined using Bio-Plex Pro Mouse Diabetes 8-Plex Assay (Bio-Rad Laboratories). All other analytes were measured from terminal plasma collection per the manufacturer’s instructions. Analytical variation of assays was assumed as reported by the manufacturer.

Statistics

All data are represented as mean ± standard error of the mean (SEM). Glucose disappearance rate (k_g) was defined as Δ blood glucose/min. Statistical significance was determined using unpaired Student t tests or, where appropriate, one- and two-way analysis of variance with multiple-comparison Tukey and Sidak post-tests, respectively, which was completed using GraphPad Prism version 6.0 (GraphPad Software). EE data were analyzed via analysis of covariants (SAS v9.4) as recommended by Tschöp et al. (19) and Friedman’s nonparametric 2-way analysis of variance for repeated measures. Statistical significance was assigned when P < 0.05.

Results

LDB1 heterozygosity does not impair glucose homeostasis under high-fat feeding

Because LDB1 is required for β-cell development and subsequent postnatal function (9, 11) and Ldb1 null mice (i.e., Ldb1^−/−) are uninformative due to early embryonic lethality (13), we assessed glucose homeostasis in a phenotypically mild global Ldb1 heterozygous mouse (Ldb1^+/− )(13) challenged by chronic HF feeding. To corroborate that LDB1 heterozygosity translated to reduced protein and messenger RNA (mRNA) levels, we quantified pancreatic Ldb1 expression and confirmed a 50% reduction at the transcript and protein levels in Ldb1^+/− mice [Fig. 1(a) and 1(b)]. However, despite reduced LDB1, we found that low- or high-fat diet does not significantly affect fasting glycemia in Ldb1^+/− and control littermates. Furthermore, when challenged with IP glucose, glucose tolerance in Ldb1^+/− and controls was not statistically different on chow diet or following 10-week HFD [Fig. 1(c–e)]. These results suggest that maintaining one Ldb1 allele is sufficient for glucose homeostasis in healthy and DIO animals.

Figure 1.

Regulation of glucose metabolism and insulin action in Ldb1^+/− and control (CTL) littermates. (a, b) Islet LDB1 mRNA and protein levels following 14-week HF feeding. mRNA levels were normalized to Gapdh. Western blotting of nuclear extracts isolated from total pancreas. LDB1 band densitometry is included above each lane, relative to Actin loading control. (c) Fasting blood glucose on chow and following 10 weeks of HFD. Blood glucose excursion following an IP bolus of glucose (1.5 g/kg) on (d) chow diet and (e) after 8 weeks of HFD. Data represented as mean ± SEM. **P < 0.01; n = 10 to 14 mice/group.

We next assessed the impact of Ldb1 on insulin action by challenging HF-fed Ldb1^+/− and control littermates with an IP insulin bolus. Ldb1^+/− mice displayed a subtle enhancement in insulin action as indicated by increased blood-glucose disappearance rate (k_g) from 0 to 45 minutes [Fig. 2(a) and 1(b)]. Enhanced insulin sensitivity is further supported by the trend toward reduced ad libitum circulating insulin levels [P = 0.07; Fig. 2(c)]. We also observed a mild, yet significant decrease in islet mRNA levels of Insulin 2 (Ins2) and Slc2a2 (encoding the glucose transporter Glut2) in the Ldb1^+/− mice following 14 weeks of HF feeding (Supplemental Fig. 1 ^{(33.4MB, tiff)} ). This is consistent with our previous findings in pancreas-specific LDB1 deficiency models (9, 11), but could also be secondary to an appropriate adaptive response to the observed improvement in insulin sensitivity. Collectively, these data confirm our prior findings (9) and indicate that Ldb1 mediates systemic insulin action but is haplosufficient for the maintenance of glucose homeostasis.

Figure 2.

Ldb1^+/− mice have enhanced insulin action during HF feeding compared with control (CTL) littermates. (a) Blood glucose excursion following an IP bolus of insulin (0.75 U/kg) after 8 weeks of HFD. (b) k_g between time point 0 (time of injection) and time point 45 minutes (nadir of blood glucose curve) during the insulin tolerance test. (c) Ad libitum plasma insulin levels after 10 weeks on HFD. Data represented as mean ± SEM. ***P < 0.001; n = 10 to 14 mice/group. (d) Relative Ldb1 expression across various tissues in 8-week-old wild-type C57BL/6 male mice (n = 3). mRNA levels were normalized to Gapdh, and the lowest expressing tissue (colon) was set to onefold. a.u., arbitrary units.

LDB1 is broadly expressed across metabolic tissues

To assess potential metabolic roles for LDB1 beyond the pancreas, we first performed an mRNA expression screen across various tissues. Ldb1 mRNA was found in many tissues required for metabolic action, including white and brown adipose, liver, skeletal muscle, and brain [Fig. 2(d)]. The low level of Ldb1 mRNA in the total pancreas sample is likely due to reduced expression in acinar cells (9) and not low template quality, as Insulin I was robustly expressed in these samples (Supplemental Fig. 2 ^{(33.8MB, tiff)} ).

LDB1 regulates energy balance in DIO mice

In light of our evidence that LDB1 is expressed across multiple metabolic tissues and influences insulin sensitivity [Fig. 1(a) and Fig. 2(a) and 2(b)], we hypothesized that LDB1 has roles beyond the β-cell to impact metabolism. To investigate this hypothesis, we assessed components of energy balance in Ldb1^+/− mice during chronic HF feeding. Although Ldb1^+/− mice were slightly smaller at the study onset, they displayed similar diet-induced body and fat mass accrual [Fig. 3(a–c)] and similar lean mass (before and after HF feeding) when compared with controls [Fig. 3(d)]. In spite of normal obesogenic response, Ldb1^+/− mice were characterized by reduced average weekly food intake (P = 0.0395), resulting in diminished cumulative consumption by week 6 of HFD [Fig. 3(e) and 3(f)].

Figure 3.

Energy balance during HF feeding in Ldb1^+/− and control (CTL) littermates. (a) Average weekly body weight. (b) Change in body weight from the onset of HF feeding. (c) Fat mass. (d) Lean mass. (e) Average weekly food intake. (f) Cumulative food intake throughout the study period. Data represented as mean ± SEM. *P < 0.05; ***P < 0.001; n = 10 to 14 mice/group.

Because of the apparent energy imbalance, we hypothesized that EE is impaired in these mice. Using indirect calorimetry, we identified reduced EE in Ldb1^+/− mice that could not be rescued by decreasing the ambient temperature [Fig. 4(A) and 4(B); Supplemental Fig. 3(a–c) ^{(68.1MB, tiff)} ]. Analysis of EE by analysis of covariants (19) identified an expected correlation with body weight (Supplemental Fig. 4 ^{(42.2MB, tiff)} ); however, based on our sample size, we also analyzed this data utilizing nonparametric methodology and uncovered an association (P = 0.0498) between genotype and EE when adjusted for body weight (Supplemental Fig. 4 ^{(42.2MB, tiff)} ). Importantly, reduced EE was also observed during HF feeding [Fig. 5(a)]. Although EE was suppressed, substrate utilization (RQ) and locomotor activity were similar in control and Ldb1^+/− mice, regardless of diet [Fig. 4(c) and 4(d) and Fig. 5(b) and 5(c); Supplemental Fig. 3(d) and 3(e) ^{(68.1MB, tiff)} ]. In contrast to our data obtained from long-term HF feeding, food intake during this 1-week study was comparable between control and Ldb1^+/− mice [Fig. 4(e) and 4(f) and Fig. 5(d); Supplemental Fig. 3(f) ^{(68.1MB, tiff)} ]. It is important to note that beyond the contracted duration of HF feeding, the indirect calorimetry study also differed in that all mice were single housed, an experimental condition known to influence EE and food intake (19).

Figure 4.

Indirect calorimetry analysis of Ldb1^+/− and control (CTL) littermates. (a, b) EE (kcal/h) at 26°C, 24°C, and 22°C in chow-fed mice. (c) Diurnal patterns of substrate utilization [RQ (VCO₂/VO₂)] at 26°C. (d) Locomotor activity (counts) at 26°C. (e) Cumulative food intake (g/mouse) at 26°C, 24°C, and 22°C. (f) Diurnal patterns of average food intake (g/mouse/15 min) at 26°C. Data represented as mean ± SEM. *P < 0.05; **P < 0.01; n = 5 mice/group.

Figure 5.

Indirect calorimetry analysis of Ldb1^+/− and control (CTL) littermates on HFD at 26°C. (a) Circadian EE (kcal/h). (b) RQ (VCO₂/VO₂). (c) Total locomotor activity (counts). (d) Food intake (g/mouse). Data represented as mean ± SEM (n = 5 mice/group).

LDB1 regulates BAT gene expression

The indirect calorimetry results suggested LDB1 regulates not only EE, but also the thermogenic response to cold exposure. In rodents, BAT has important functions in thermo-regulation and energy homeostasis (20, 21). Therefore, we analyzed mRNA levels of BAT thermogenic genes to determine whether HF-fed Ldb1^+/− mice have reduced EE potential. Expression of several BAT markers and genes relating to BAT oxidative metabolism and mitochondrial biogenesis were decreased in Ldb1^+/− mice, including cell death activator CIDE-A (Cidea), Elongase of very long-chain fatty acids (Elovl3), and Cytochrome c oxidase 7a1 (Cox7a1) [Fig. 6(a)], suggesting that the BAT program may be sensitive to Ldb1 gene dosage. We also observed a 50% reduction in BAT expression of Dio2 in HF-fed Ldb1^+/− mice [Fig. 6(a)], an enzyme crucial in BAT response to sympathetic tone (22).

Figure 6.

BAT-expressed LDB1 regulates key thermogenic factors. BAT expression analysis of gluco- and lipo-regulatory genes and thermogenic genes in (a) Ldb1^+/− (n = 10 to 14 mice/group) and (b) Ldb1^Δβ-cell (n= 6 to 13 mice/group) compared with their respective control (CTL) littermates after HF feeding. BAT mRNA levels were normalized to Hprt or 36B4. (c) Immunohistochemical staining images for LDB1, PPARγ, and ISL1 in BAT from chow-fed CTL C57BL/6 mice. (d) Schematic of the mouse Elovl3 5′ sequence demonstrating several putative core homeodomain-binding elements that may associate with LDB1 complexes. (e) LDB1 ChIP using chromatin isolated from chow-fed CTL mouse BAT demonstrates LDB1 enrichment of Elovl3 5′ sequences containing a homeodomain binding element (site 3 in d; n = 4). (f) Representative Western-blot image of LDB1 and Elovl3 protein in BAT of chow-fed CTL and Ldb1^+/− mice (n = 3 for each genotype). Actin is included as loading control. Protein markers (in kilodaltons) are shown. Data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ****P < 0.0001. a.u., arbitrary units.

We recognized the possibility that changes in BAT mRNA levels could be an indirect effect of the reduced insulin in the Ldb1^+/− mice. To assess this, we performed a parallel analysis of the impact of HF feeding on mice in which LDB1 was inducibly deleted in postnatal pancreatic β-cells [termed Ldb1^Δβ-cell; Supplemental Fig. 5(a) ^{(72.2MB, tiff)} ]. In agreement with the findings of Ediger et al. (11), Ldb1^Δβ-cell mice exhibit glucose intolerance (data not shown) and markedly impaired circulating insulin levels [Supplemental Fig. 5(b) ^{(72.2MB, tiff)} ] (11). Strikingly, Ldb1^Δβ-cell mice exhibit normal food intake and energy balance [Supplemental Fig. 5(c–e) ^{(72.2MB, tiff)} and Supplemental Fig. 6 ^{(39.3MB, tiff)} ], and analysis of BAT gene expression in Ldb1^Δβ-cell mice identified no overt changes in the expression of critical thermogenic genes [Fig. 6(b)]. Together these data strongly argue that the alterations in BAT gene expression in the global Ldb1^+/− mice are not a consequence of reduced plasma insulin.

These findings raised the intriguing possibility that LDB1, either directly or secondary to its role in other tissues, regulates BAT gene programming. Consistent with a cell-autonomous role, we observed Ldb1 transcript and Ldb1 immunoreactivity in BAT [Fig. 2(d) and Fig. 6(a) and 6(c)]. In agreement with our expression analysis, we detected nuclear LDB1 signal in cells throughout the PPARɣ⁺ interscapular brown adipose depot [Fig. 6(c)]. Moreover, signal for ISL1, a direct-binding partner of LDB1, was absent [Fig. 6(c)], suggesting LDB1 is likely partnering with another LIM-HD or LMO factor in this tissue. Further, we performed an LDB1 ChIP using chromatin isolated from chow-fed C57BL/6 mouse BAT and then quantified enrichment of upstream sequences from the most affected mRNA in the Ldb1^+/− mice, Elovl3 [Fig. 6(a) and 6(d)]. The sequence examined is located within chr19:46130819-46132067 and bears several TAAT/ATTA elements, suggestive of binding by LIM-HD transcription factors, one class of LDB1-interacting partners. As compared with IgG control, LDB1 significantly enriched for upstream 5′ Elovl3 sequences [Fig. 6(e)] in a promoter region also bound by PPARγ (23, 24). This indicates that BAT-expressed LDB1 directly regulates a metabolic gene pivotal in brown adipocyte function. In further support of LDB1-mediated regulation of Elovl3, we observed reduced Elovl3 protein levels after western-blotting BAT whole-cell extracts from chow-fed Ldb1^+/− mice, as compared with controls [Fig. 6(f)]. Taken together, our findings strongly argue that LDB1 directly impacts BAT gene expression.

Discussion

The transcriptional coregulator LDB1 is known for its role in pancreatic islet development and function (9, 11), erythropoiesis (10), and spinal cord neural subtype specification (25). However, metabolic function in nonpancreatic tissues has yet to be elucidated. Moreover, LDB1 is broadly expressed, notably in various tissues involved in glucose and lipid homeostasis. Thus, our studies encompassed a global investigation of LDB1 to determine its impact on various aspects of metabolism. In summary, we uncovered LDB1 impacts on EE regulation of energy balance, insulin action, and excitingly unique roles in BAT. Although the increase in insulin sensitivity and decreased BAT function may initially appear inconsistent, given the broad expression of LDB1 in multiple metabolic tissues (e.g., BAT, white adipose tissue, liver, skeletal muscle, and brain), we hypothesize that LDB1 has tissue-specific roles that may differentially impact (for example) insulin sensitivity in liver and muscle, while decreasing activity in BAT.

Ldb1^+/− mice subjected to indirect calorimetry analysis revealed an overall reduced EE. Notably, during this EE analysis, we did not observe differences in food intake. Although this may be initially interpreted as an inconsistency between the studies, there are key variables that must be considered. First, our initial observations involved mice during chronic HF feeding, although the latter were made in a study lasting just 3 days. Secondly, mice in the chronic HF feeding study were group housed, while feeding measured during the calorimetry study occurred during single housing. With group-housed mice, confounding factors such as food spilling, social dominance, and aggression can dramatically influence traits linked to energy balance (19). Additionally, group-housed mice often huddle, which considerably reduces thermoregulatory requirements and thus EE and intake (19). Findings published during study describe the expression of human growth hormone (hGH) in the MIP-Cre^ERT model (26) and bring to question the role growth hormone has on EE and its implications in the interpretation of our data. However, Heffernan et al. reported that inducing pharmacological levels of hGH failed to alter EE in lean or obese mice (27). Our findings are further bolstered by the observation that the hGH produced by a MIP-FoxM1-hGH model was undetectable in plasma (28), and together suggest that MIP-Cre^ERT-derived hGH has no impact on EE.

Because BAT thermogenesis is a primary component of EE in mice (29, 30), we hypothesized that LDB1 is also a critical regulator of BAT function. In support of this hypothesis, essential BAT mRNAs, including those related to mitochondrial and BAT function, were suppressed. Specifically, we observed reduced mRNA in DIO Ldb1^+/− mice encoding for the lipid droplet protein Cidea (an established marker of BAT in rodents) (31), the oxidative protein Cox7a1, and Elovl3, which is necessary for synthesis of very long-chain fatty acids and full metabolic capacity in brown adipocytes (11). The reduction of these gene products in this study does not appear to be linked to expression of the transcriptional coactivator Pgc1α (32), as mRNA levels were unchanged. We also found Ldb1 occupying the 5′ promoter region of Elovl3 by ChIP, providing evidence for direct LDB1 action in BAT.

We also identified a reduction in the mRNA encoding DIO2, an enzyme required for rodent BAT EE (22). Dio2 induction in BAT stimulates the conversion of intracellular thyroxine (T4) to 3,3′,5-triiodothyronine (T3), enhancing thermogenic responses (33). Thus, reduced DIO2 might indicate that BAT is less responsive to the major activating signals for thermogenesis. BAT is densely innervated and regulated by the sympathetic nervous system (34, 35). Sympathetic outflow to BAT increases norepinephrine release from postganglionic nerve terminals that bind to brown adipocyte β₃-adrenoceptors (36), ultimately activating UCP1 activity, EE, and thermogenic gene expression (20). Taken together with evidence demonstrating that DIO2 is pivotal in increasing adrenergic responsiveness (37, 38), our findings suggest that reduced expression of Dio2 in Ldb1^+/− mice may interfere with sympathetic nervous system–induced activation of BAT EE. Plasma analyses revealed significantly elevated T3 and T4 in the Ldb1^+/− DIO cohort, as compared with DIO controls (Supplemental Table 2 ^{(29.5KB, pdf)} ). Although a reduction in DIO2 may account for increased T4, there is no clear mechanism by which it would also increase T3. Thus, although it is tempting to associate this with changes in BAT Dio2 mRNA, we do not yet know if the Ldb1^+/− mice also have defects in the hypothalamic-pituitary-thyroid axis or governing feedback mechanisms. It is interesting to note that LDB1 has roles in the pituitary (39) and human genome-wide association studies have linked elevated T4 and thyroid-stimulating hormone to the pituitary LIM-HD transcription factor (and Ldb1 partner) LHX3 (39). Future studies will dissect the contributions of BAT- vs hypothalamic-pituitary-thyroid–expressed LDB1 in EE control.

LDB1 is a scaffolding coregulator lacking DNA-binding and direct gene activation/repression capacity (40). Rather it has high avidity for the LIM-domain family of proteins (including LIM-HD and LMO factors) (41, 42), including the ISL1 transcription factor. ISL1 is required for islet development and postnatal β-cell function (43–45); however, it is unclear which LIM factors mediate activity in other LDB1-sensitive tissues such as BAT, as revealed in our study. We did not observe ISL1 expression in BAT, raising the likelihood that other LMO or LIM-HD partners mediate the effect of LDB1 on BAT gene expression. ChIP revealed LDB1 enrichment of the Elovl3 promoter in a domain enriched in LIM-HD binding TAAT/ATTA elements. The related LIM-HD factor LHX8 is 1 candidate for mediating LDB1 effects in the BAT, as it is a brown adipose–specific marker (8, 46). However, we cannot exclude the possibility that LDB1 interacts with LMO-mediated Gata and basic helix-loop-helix transcription factor complexes in this tissue.

Overall, our study strongly supports insulin-independent roles for LDB1 in the regulation of energy homeostasis during DIO, in part through direct effects on BAT gene expression. Future studies will aim to identify the tissue-specific action of LDB1 in regulation of energy homeostasis.