1. Introduction
Selenoproteins are a small group of proteins containing the micronutrient selenium in their molecule. These proteins are involved in redox reactions that can reduce oxidative stress, activate thyroid hormones, and act on endoplasmic reticulum-associated degradation of misfolded proteins. There are only 24 and 25 selenoproteins in mice and humans, respectively. Nevertheless, selenoproteins exert a profound effect in several aspects of health, including the pathophysiology of different types of cancer [
1,
2,
3], autoimmune diseases [
4], irritable bowel syndrome [
5], type 2 diabetes [
6,
7], and thyroid disorders [
8,
9].
A hallmark feature of selenoproteins is the incorporation of selenium as the amino acid selenocysteine (Sec) in their primary structure. Sec is encoded by the UGA codon, also recognized as a stop codon. Circumvention of the stop recognition occurs due to a combination of
cis and
trans factors that have been extensively reviewed elsewhere [
10,
11,
12,
13,
14,
15,
16,
17,
18]. Notably, one of the key factors is the specific tRNA
[Ser]Sec, required for successful insertion of Sec during translation of selenoproteins, and encoded by the gene
Trsp.
Total loss of
Trsp in mice is embryonically lethal [
19], a finding that highlights how essential selenoproteins are for life. To demonstrate the role of selenoproteins in various organs, tissues and cell types, targeted deletion of
Trsp in specific cell types has been carried out for decades, with surprising consequences [
10,
20,
21]. However, the targeted deletion of
Trsp in brown adipocytes, a cell type critical for thermoregulation and energy expenditure in mammals, had not been attempted.
Brown adipocytes are the most abundant cell type found in the brown adipose tissue (BAT), the main site for adaptive thermogenesis in rodents, and subpopulations of brown adipocytes present distinct thermogenic potential [
22,
23]. Mammalian adaptive thermogenesis is classically activated upon exposure to cold or after caloric overload. BAT cold-induced adaptive thermogenesis relies primarily on the actions of uncoupling protein 1 (Ucp1), a proton pump localized in the inner membrane of mitochondria of brown adipocytes that dissipates energy from ATP synthesis as heat [
24,
25]. Ucp1 expression is synergistically upregulated after activation of beta-adrenergic signaling and by thyroid hormone 3,3′,5-triiodothyronine (T3) [
26]. Interestingly, T3 is mostly produced locally in cells, after the removal of one iodine from thyroxine (T4), the main prohormone from the thyroid gland. Conversion of T4 into active T3 in brown adipocytes is catalyzed by iodothyronine deiodinase type 2 (Dio2). Dio2 is a selenoprotein highly expressed in activated brown adipocytes and essential for cold-induced thermogenesis [
27], allowing for the increased Ucp1 expression [
28,
29]. Mice with a disruption of Dio2 cannot properly carry out adaptive thermogenesis, either by cold exposure or caloric overload at room temperature [
28,
30,
31]. Besides Dio2, glutathione peroxidases 1, 3, and 4 (Gpx1, Gpx3, and Gpx4, respectively), thioredoxin reductases 1, 2 and 3 (Txnrd1, Txnrd2, and Txnrd3, respectively), selenoproteins H, O, T and P (SelenoH, SelenoO, SelenoT, and SelenoP, respectively), and selenophosphate synthetase 2 (Sephs2) have also been found in BAT using a proteomics analysis [
32], but the role of these selenoproteins in BAT function is unknown.
We report the generation of a mouse model with a targeted deletion of the gene for Trsp in brown adipocytes, the Trspf/f-Ucp1-Cre+/− mice, to characterize the influence of selenoproteins in BAT function. These mice thermoregulate without changes in energy expenditure or Ucp1 expression, and surprisingly respond to acute cold exposure adequately. We also observed novel genes that respond to the lack of Trsp in brown adipocytes both at room temperature and after acute cold exposure, particularly those involved in methylation pathways and response to thyroid hormone levels.
3. Discussion
We have reported the development of a novel transgenic mouse model, the Trspf/f-Ucp1-Cre+/− mice. Our model was characterized by targeted deletion of the gene Trsp in brown adipocytes, which are the primary cell type of BAT. The product of the Trsp gene, tRNA[Ser}Sec, is responsible for carrying out the synthesis of selenoproteins, including Dio2, a regulator of cold-induced adaptive thermogenesis in BAT.
Dio2 upregulation is promoted by thyroid hormone and is considered a primary indicator of activation of adaptive thermogenesis in cold exposure [
44]. In males,
Dio2 expression was unchanged between
Trspf/f and
Trspf/f-Ucp1-Cre+/− mice, despite a reduction in circulating T3. In addition, circulating T4, the substrate used by Dio2 to generate T3 that, in turn, activates thermogenesis, was unchanged. In BAT, Ucp1 levels were also unchanged. Nevertheless, BAT of male
Trspf/f-Ucp1-Cre+/− mice had a distinct tissue morphology resembling more a white adipocyte with increased lipid accumulation, typical of under-stimulated brown adipocytes [
45]. Together, these results suggest that, despite T4 levels not being affected by the loss of
Trsp in male
Trspf/f-Ucp1-Cre+/− mice, the physiology of their brown adipocytes are indeed impaired. This impairment in the morphology of BAT in males possibly occurs since early development, because mice lacking
Dio2 show permanent defect in adaptive thermogenesis in the embryonic BAT [
46].
Murine selenoproteins are expressed in a sexually dimorphic manner in other organs, such as liver and kidneys [
47,
48]. A possible explanation for the sexual dimorphism we observed in BAT of male
Trspf/f-Ucp1-Cre+/− mice could be alterations in the sympathetic tone in our mouse model. However, one of the regulators of the sympathetic tone of BAT is T4 via central and peripheral mechanisms [
49]. However, the
Dio2 knockout mouse model does not present sexual dimorphism in their thermogenic responses after a caloric overload at room temperature [
31]. Moreover, T4 is not altered in males, making it unlikely that the sympathetic tone is diminished in BAT of these mice. An alternative to be explored in future studies is whether another selenoprotein, another factor involved in selenium metabolism, or another selenium-related mechanism plays a role in regulating the sympathetic tone in BAT according to sex.
Interestingly, elevated circulating TSH levels in male
Trspf/f-Ucp1-Cre+/− mice at room temperature suggest either impaired thyroid stimulation to produce T4—which does not occur as T4 levels are maintained—or a direct effect of TSH via its receptor TSHR in brown adipocytes, which could explain the morphological changes of BAT. Such direct effect of TSH has been demonstrated in rats to positively modulate oxygen consumption and inversely regulate
Dio2 mRNA [
50]. However, this observation is discrepant with our results herein in the male
Trspf/f-Ucp1-Cre+/− mice, as circulating TSH were higher but effects on neither oxygen consumption nor
Dio2 mRNA were observed. It is possible that other mechanisms modulated by TSH alone, such as activation of Erk or Akt phosphorylation, are recruited [
50], reducing the expected responses cited above.
Acute cold exposure leads to intense lipolysis, adipocyte autophagy, and remodeling of BAT, using the lipid deposits available to increase the mitochondrial oxidative cycle. This, in turn, could produce heat, maintaining the animal’s core body temperature and avoiding hypothermia. Extensive use of lipid deposits leads to activation of a lipogenic phase with enhanced lipid uptake in brown adipocytes, guaranteeing fuel for thermogenesis maintenance as cold exposure continues [
51]. Despite BAT undergoing intense remodeling in the first hours of cold exposure, it is intriguing that male
Trspf/f-Ucp1-Cre+/− mice do not present alterations in the autophagy flux; levels of autophagy factor p62/SQSTM1 were sustained in male
Trspf/f-Ucp1-Cre+/− mice, while autophagy genes were upregulated in our microarray analysis. This discrepancy suggests either a reduction in translational efficiency of autophagic transcripts or a temporal dissonance in autophagy activation in male
Trspf/f-Ucp1-Cre+/− mice acutely exposed to cold. The varying length of cold exposure has been considered as a source of conflicting data for autophagy response in adaptive thermogenesis ([
52] citing [
53,
54,
55]). These studies did not expose mice for 4 h in the cold; therefore, it is also possible that we uncovered a transitional moment for autophagy. However, male
Trspf/f-Ucp1-Cre+/− mice accumulated more triglycerides than control mice, despite similar levels of mitochondrial markers at room temperature. These combined findings suggest that male
Trspf/f-Ucp1-Cre+/− mice maintained mitochondrial capacity with excess resources for the initial lipolytic phase of adaptive thermogenesis. It is possible that during cold exposure, as brown adipocytes switch from lipolysis to lipogenesis, the advantage provided by additional fuel is counteracted by deficiencies in thermogenic responses that are dependent on other factors, such as thyroid hormone activation or sympathetic tone [
56].
It is puzzling that the loss of
Trsp, i.e., selenoprotein synthesis capacity, does not lead to a more dramatic impact in gene expression, particularly of selenoproteins after acute cold exposure. As mentioned in the previous paragraph, the late lipogenic phase of chronic cold exposure depends on the proper transcriptional activation of T3-responsive genes encoding key proteins involved in lipogenesis early on, which will later allow for withstanding the constant temperature stress. Our finding that the
Thrsp transcript, which encodes a coactivator of lipogenic genes via interaction with the thyroid hormone receptor [
40,
41], was downregulated in
Trspf/f-Ucp1-Cre+/− mice suggests lower availability of thyroid hormones locally in the early stages of cold exposure. It is plausible to infer that these mice, upon chronic exposure to cold, may diminish their thermogenic capacity due to impaired lipid uptake, because the proteins required for lipogenesis could be at lower levels. Prospective studies examining the chronic activation of BAT adaptive thermogenesis, either in a more physiological manner after chronic cold exposure or in a pathological context, after a high-fat diet, may consolidate whether observed changes in thyroid hormone response of the
Trspf/f-Ucp1-Cre+/− mice are sustained. Moreover, the participation of additional gut hormones in concert with the sympathetic response may mediate chronic thermogenic responses. A chronic paradigm for adaptive thermogenesis activation, particularly in a pathological context after a hypercaloric overload, may allow for display of a worsened metabolic dysfunction. We have observed a moderate impairment of Akt activation, i.e., in insulin response, in the gWAT of
Trspf/f-Ucp1-Cre+/− mice. Therefore, it is likely that the feeding of a high-fat diet will accelerate the development of a metabolic dysfunction in a
Trspf/f-Ucp1-Cre+/− mouse.
An interesting aspect found by our microarray analysis was the upregulation of genes involved in methylation. As occurs with other tRNAs [
57], a portion of the tRNA
[Ser]Sec population is methylated at the uridine position 34 (Um34) yielding a specific subpopulation [
58]. In fact, differential methylation of Um34 leads to the existence of two possible molecular conformations in relation to the anticodon sequence [
59]. Consequently, two subsets of selenoproteins, deemed stress-related and housekeeping, occur, with tRNA
[Ser]Sec with Um34 supporting primarily translation of stress-related selenoproteins. Members of this class of selenoproteins are sensitive to selenium status [
20,
21], which is likely the case for Dio2 in the activated BAT. The specific methyltransferase responsible for the methylation of Um34 has not been identified, although one that is sensitive to S-adenosyl-homocysteine accumulation has been indicated as a promising candidate [
60]. In our mouse model, loss of tRNA
[Ser]Sec in brown adipocytes led to upregulation of
Mthfd2, which encodes an enzyme from folate one-carbon metabolism. This enzyme also binds to several ribonucleoproteins and is involved in protein translation [
61], controlling the methylation of N
6-methyladenosine (m
6a) of mRNAs [
42]. It is tempting, at this point, to suggest that this enzyme may be an additional promising candidate for at least controlling the Um34 methylation of tRNA
[Ser]Sec.
Considering the impairment of selenoprotein synthesis in the brown adipocyte, it is intriguing that a general decline in selenoprotein gene expression, particularly
Dio2, is not observed. Stress-related selenoprotein mRNAs, such as
Dio2, are primary targets for nonsense-mediated decay (NMD) degradation, a mechanism that misreads the in-frame UGA codon of selenoproteins as a nonsense codon [
62]. Selenium deficiency further activates the NMD machinery to target selenoprotein mRNAs for degradation [
63].
Trspf/f-Ucp1-Cre+/− mice were fed diets with adequate selenium, which likely provided enough selenium to inhibit NMD machinery, allowing selenoprotein mRNAs to circumvent NMD degradation.
Moderate dietary selenium intake in
Trspf/f-Ucp1-Cre+/− mice also raises a striking possibility regarding the fate of selenium in brown adipocytes. With the impairment of selenoprotein synthesis, available selenium may be either released back into circulation or to neighboring cell types, or redirected within the brown adipocyte for use in additional pathways. Jedrychowski et al. [
32] have demonstrated that excess selenium in brown adipocytes impacts adaptive thermogenesis via incorporation of Sec in the place of cysteine in several non-selenoproteins, i.e., facultative selenation. Notably, cysteine residue 253 of Ucp1 was identified to be selenated. Selenation of Ucp1 increases its redox sensitivity, enhancing energy expenditure, activating adaptive thermogenic responses, and protecting from high-fat diet-induced obesity. It is possible that selenium not used in selenoprotein synthesis by the
Trspf/f-Ucp1-Cre+/− mice enhances facultative selenation. Enhanced facultative selenation, particularly of Ucp1, could explain the mild effects of acute cold exposure observed in our mouse model, with the lack of
Trsp and consequent selenoprotein synthesis freeing selenium to be redirected towards selenation in brown adipocytes. It is unknown whether facultative selenation in brown adipocytes requires tRNA
[Ser]Sec. We believe that the
Trspf/f-Ucp1-Cre+/− mouse model may be fitting to test this hypothesis in the future.
A limitation of this study encompasses the inability to breed or maintain mice at their thermoneutrality (~28–30 °C) due to the lack of this capacity at our animal facilities. Mice maintenance at temperatures slightly lower than thermoneutrality could have primed their BAT to activate thermogenic mechanisms. Such prior activation could diminish the impact of acute cold exposure, and this effect was observed in a mouse model lacking
Dio2 [
64]. Despite this limitation, we still observed subtle changes at the transcriptional level after acute cold exposure, particularly in genes responding to thyroid hormone and involved in methylation pathways. The fact that we unveiled changes in gene expression between the
Trspf/f-Ucp1-Cre+/− and
Trspf/f control mice indicates that the effects of this temperature priming were potentially minimized.
In summary, we developed a novel mouse model lacking the Trsp gene in brown adipocytes and demonstrated that loss of Trsp in these cells leads to a modest impairment of acute cold exposure responses in male Trspf/f-Ucp1-Cre+/− mice, but not in females. We also observed subtle changes in thyroid hormone responsiveness and genes involved in methylation pathways. Future studies moving towards a chronic cold exposure may help refine our knowledge of the role of selenoproteins in rodent adaptive thermogenesis.
4. Materials and Methods
4.1. Chemicals and Antibodies
All chemicals used in experiments were from Fisher Scientific (Hampton, NH, USA) or Sigma-Aldrich (St. Louis, MO, USA), unless specified. Primary antibodies included anti-β-actin (1:3000; cat. #A2228; Sigma-Aldrich), anti-Ucp1 (1:1000; cat. #ab10983, Abcam; Cambridge, MA, USA), anti-PGC1α (1:750; cat. #AB3242, MilliporeSigma; Burlington, MA, USA), anti-ATPB (1:1000; cat. #ab14730; Abcam), anti-Akt (1:1000; cat. #9272, Cell Signaling Technology, Beverly, MA, USA); anti-phospho-Akt-Ser473 (1:750; cat#9271, Cell Signaling Technology), and anti-p62/SQSTM1 (1:750; cat. #5114T, Cell Signaling Technology).
4.2. Animals
Animal procedures were approved by the University of Hawaii Office of Research Compliance Animal Welfare Program, protocol #17-2521, first approved by the Institutional Animal Care and Use Committee on 19 October 2017. Mice were held in our vivarium and used for experiments in minimal numbers needed to provide significant results. B6.FVB-Tg(Ucp1-cre)1Evdr/J (MGI J:206508; shortened here as
Ucp1-
Cre) mice on a C57BL6/J background were purchased from The Jackson Laboratory (cat. #024670; Bar Harbor, ME, USA).
Trsp-
LoxP were generated as previously described [
65]. To generate a strain with a targeted deletion of
Trsp in brown adipocytes, the
Trspf/f-Ucp1-Cre+/− mice,
Trspf/+ mice were crossed with
Ucp1-
Cre mice. Homozygote
Trspf/f littermate controls were used to compare with
Trspf/f-Ucp1-Cre+/− mice after genotyping confirmation by PCR. Mice were in a 12 h light–dark cycle, fed
ad libitum with chow containing ~0.25 ppm sodium selenite, and group-caged until exposure to cold. Euthanasia occurred at 12-weeks old by CO
2 asphyxiation, and all mice were euthanized ~2 pm, to avoid circadian influences in results. After euthanasia, blood and interscapular BAT, which comprises approximately 60% of all BAT depots and considered the most significant in mice [
66], were removed immediately, snap-frozen into liquid nitrogen, perfused with formalin, or incubated in RNALater Stabilization Solution (Thermo Fisher Scientific; Waltham, MA, USA).
4.3. Energy Expenditure Assessment
Oxygen consumption (VO
2), respiratory quotient (RQ) and energy expenditure (EE) were assessed for 24 h with the PanLab OxyLet
ProTM System (Harvard Apparatus, Barcelona, Spain) in 11-week-old mice. Mice were habituated in individual cages for 24 h prior to running experiments, as previously described [
67]. Data were analyzed using PanLab Metabolism (Prague, Czech Republic) software.
4.4. Cold Exposure and Core Body Temperature Assessment
A G2 emitter thermal probe (Starr Life Sciences Corp.; Oakmont, PA, USA) was surgically inserted into the peritoneal cavity of mice one week prior to exposure to cold. Only mice that adequately recovered from survival surgery were used in experiments. On the day of experiment, 12-week-old mice were transferred to individual cages with only water and minimal bedding, transported to a cold room set at 4 °C between 09:00 and 10:00 a.m., and each cage placed on a receiver telemetry platform (Starr Life Sciences Corp.). Mice remained in the cold room uninterrupted for 4 h and were euthanized ~2 pm. Core body temperature was recorded every 5 min by the receiver using Vital View Legacy Version 5.1 software (Starr Life Sciences Corp.).
4.5. Triglyceride Content
Triglyceride content of BAT was measured using the Triglyceride Quantification Assay Kit (Abcam) for colorimetric detection, following the manufacturer’s protocol.
4.6. RNA Extraction
Mouse interscapular BAT was collected at time of euthanasia (~2pm) either in RNALater (Thermo Fisher Scientific) or snap-frozen, according to final use. For quantitative PCR (qPCR), total RNA was extracted from snap-frozen tissues using a TissueRuptor (Qiagen; Germantown, MD, USA) with disposable probes, and isolated using the EZNA total RNA kit I (Omega Biotek; Norcross, GA, USA). For microarray, total RNA was extracted from tissues in RNALater, processed with TissueRuptor and isolated using the Qiagen RNA extraction kit (Qiagen).
4.7. Microarray Analysis
RNA quality was assessed in an Agilent 2100 BioAnalyzer (Agilent Technologies; Santa Clara, CA, USA) and samples with RIN > 5.5 were deemed satisfactory for microarray. Mouse ClariomS Array gene hybridization (Affymetrix—Thermo Fisher Scientific) was carried out with 100 ng of total RNA, run and scanned in GeneChip Fluidics Station 450 and a GeneChip Scanner (Affymetrix—Thermo Fisher Scientific) in the Genomics and Bioinformatics Shared Resource facility at the University of Hawaii Cancer Center. The CEL files were processed and analyzed using the Transcriptome Analysis Console (TAC) 4.0 from Thermo Fisher Scientific. Genes with a fold change greater than 2 and
p-value less than 0.05 were considered differentially expressed. Canonical pathway and network analysis of differentially expressed genes was performed using Qiagen’s Ingenuity Pathway Analysis (IPA; Qiagen;
https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis).
4.8. Real-Time qPCR
One microgram of total RNA extracted from the interscapular BAT was reverse transcribed using the High Capacity kit (Applied Biosystems—Thermo Fisher Scientific). 10 ng of cDNA were used in real-time qPCR reactions with the PerfeCTa SYBR Green SuperMix (Quantabio; Beverly, MA) and specific primers, listed in
Supplementary Table S4. Relative quantification of target gene expression was calculated based on the Δ
Ct method, normalized to the expression of housekeeping genes for
18s or
Gapdh, and plotted as fold change relative to
Trspf/f at room temperature values.
4.9. Immunohistochemistry
Mice acclimated to room temperature had their BAT extracted after perfusion with formalin and were paraffin-embedded for histological analysis. Immunohistochemistry was performed in slides using citrate buffer in a pressure cooker. Slides were labeled with primary antibody (anti-Ucp1; 1:100), Mouse on Mouse Basic kits with Vectastain ABC, and diaminobenzidine peroxidase substrate kits (Vector Labs; Burlingame, CA, USA) as previously described [
68]. Slides were counter-stained with hematoxylin to visualize adipocyte nuclei.
Four brightfield images were acquired from a single section of BAT from each subject at 40× magnification using the simple random sampling workflow in Stereo Investigator (MBF Bioscience; Williston, VT, USA). To compare Ucp1 immunoreactivity, the optical density of each image was quantified using ImageJ (public domain) and averaged for each subject. To calculate Ucp1 optical density, images were converted to black-and-white, inverted, and the mean pixel value for the entire image recorded. Total nuclei were counted using ImageJ Cell Counter plugin (public domain) and used to normalize the mean pixel value for each image.
4.10. Western Blot
BAT and gWAT protein were extracted using RIPA Lysis and Extraction buffer containing protease and phosphatase inhibitors (Thermo Scientific) followed by sonication and two rounds of centrifugation. 10–20 μg of total protein was loaded into 4–20% TGX SDS-PAGE (Bio-Rad; Hercules, CA, USA) and wet-transferred overnight in Tris-glycine buffer containing 9% methanol to an Immobilon-FL® membrane (MilliporeSigma) using the Criterion Blotter (Bio-Rad) apparatus. Primary and secondary antibodies were incubated for 1 h at room temperature with rotation, and secondary antibodies were IRDyes (Li-Cor Biosciences; Lincoln, NE) for use at an Odyssey CTx infra-red scanner (Li-Cor Biosciences). All blots were used only once, except the measurement of phosphorylated Akt, performed on the same membrane after confirmation of complete stripping of previous antibodies using NewBlot PVDF Stripping Buffer (Li-Cor Biosciences). Analysis of the Western blot was conducted with the Image Studio Lite version 5.2.5. software (Li-Cor Biosciences).
4.11. Thyroid Function
Thyroid function was inferred from analyses of total T4, total T3, and TSH levels in the serum. Twenty microliters of serum were used to assay for TSH with a Milliplex MAP Mouse Pituitary Magnetic Bead Panel (Millipore-Sigma) on a Luminex 200 platform (Luminex; Austin, TX, USA). Twenty-five microliters of serum were used to assay for total T4 and total T3 using the AccuDiagTM ELISA—T4 kit (Diagnostic Automation; Woodland Hills, CA, USA) and the T3 (Total) ELISA kit (Abnova; Taipei, Taiwan), respectively, according to the manufacturer’s instructions.
4.12. Statistical Analysis
Results were plotted and graphed using Graphpad Prism v. 8.0 (San Diego, CA), and an alpha value of 0.05 was adopted. Unpaired Student’s t-test or two-way analysis of variance (2WA) with Bonferroni’s ad hoc post-test were used according to the number of variables analyzed in each experiment.