Iodine Deficiency Increases Fat Contribution to Energy Expenditure in Male Mice

Abstract

More than a billion people worldwide are at risk of iodine deficiency (ID), with well-known consequences for development of the central nervous system. Furthermore, ID has also been associated with dyslipidemia and obesity in humans. To further understand the metabolic consequences of ID, here we kept 8-week-old C57/Bl6 mice at thermoneutrality (~28°C) while feeding them on a low iodine diet (LID). When compared with mice kept on control diet (LID + 0.71 μg/g iodine), the LID mice exhibited marked reduction in T4 and elevated plasma TSH, without changes in plasma T3 levels. LID mice grew normally, and had normal oxygen consumption, ambulatory activity, and heart expression of T3-responsive gene, confirming systemic euthyroidism. However, LID mice exhibited ~5% lower respiratory quotient (RQ), which reflected a ~2.3-fold higher contribution of fat to energy expenditure. LID mice also presented increased circulating levels of nonesterified fatty acids, ~60% smaller fat depots, and increased hepatic glycogen content, all indicative of accelerated lipolysis. LID mice responded much less to forced mobilization of energy substrates (50% food restriction for 3 days or starvation during 36 hours) because of limited size of the adipose depots. A 4-day treatment with T4 restored plasma T4 and TSH levels in LID mice and normalized RQ. We conclude that ID accelerates lipolysis and fatty acid oxidation, without affecting systemic thyroid hormone signaling. It is conceivable that the elevated plasma TSH levels trigger these changes by directly activating lipolysis in the adipose tissues.

More than a billion individuals live in geographical areas with insufficient availability of iodine, a condition that carries a significant health and socio-economic burden for individuals, families, and societies. Thanks to the efforts of governmental and many other agencies and organizations, multiple strategies to increase iodine availability have been developed over the 20th century and today iodine deficiency (ID) affects a much smaller number of individuals worldwide (1).

From a pathophysiological standpoint, the impact of ID is minimized through adaptive mechanisms triggered by the thyroid gland and the hypothalamus-pituitary-thyroid axis. With mild and moderate iodine deficiency, there is a drop in circulating levels of T4 that elevates TSH secretion, triggering thyroid growth and increased fractional iodine uptake. The relative T3 production is accelerated, both through an increase in thyroidal T3 output and as a result of faster extrathyroidal conversion of T4 to T3. As a consequence, most adult individuals with ID have normal serum T3 and are considered clinically euthyroid, despite an elevated serum TSH. However, the most devastating consequences of ID, even when it is mild or moderate, are seen when it happens in utero or during infancy and childhood, when the thyroid gland is smaller and has limited ability to compensate (2-4). In these cases, ID compromises the synthesis of thyroid hormones (TH), leading to an associated state of hypothyroidism.

Despite clinical euthyroidism, there is evidence that adult individuals with mild/moderate ID are at risk of developing metabolic disorders. For example, in US adults, low urinary iodine concentration was associated with increased odds for dyslipidemia, but a causal relationship has not been established (5). Moderate to severe ID in overweight women was found to be associated with a more atherogenic lipid profile; iodine supplementation in this group reduced the prevalence of hypercholesterolemia (6). Similarly, children with ID exhibit increased prevalence of overweight or obesity, hyperlipidemia, and type 2 diabetes. At the same time, iodine supplementation improved glucose and lipid metabolism in these children, and also lowered serum total and low-density lipoprotein cholesterol levels (7). Although the interpretation of these studies also requires rigorous nutritional assessment of these patients to eliminate compounding macronutrient malnutrition, that some of the metabolic abnormalities was restored by iodine supplementation indicates that ID does play an important metabolic role.

Here we used a mouse model to study the metabolic impact of moderate ID. Mice have a much higher surface/mass ratio than humans and rapidly adjust the activity of the sympathetic nervous system (SNS) to minimize changes in TH signaling (8, 9). Thus, to avoid these adjustments, all studies were conducted at thermoneutrality (~28°C). The ID mice remained euthyroid throughout the 6-week experimental period, but exhibited accelerated fat mobilization and oxidation, decreased adiposity, and increased hepatic glycogen storage. These findings confirm that in mice, ID can have metabolic consequences without disrupting systemic TH signaling.

Material and Methods

All experiments were planned according to the American Thyroid Association Guide to investigating TH economy and action in rodent and cell models (10) and were approved by the Institutional Animal Care and Use Committee at University of Chicago.

Animals

Eight-week-old C57/Bl6 male mice were kept in single cages with a 12-hour dark/light cycle starting at 06:00 hours. To minimize interference from compensatory changes in the SNS (8), the present studies were conducted while mice were housed at thermoneutrality (~28°C). Mice were segregated in 2 main groups: low iodine diet (LID) (Low Iodine Diet, Envigo - Teklad Custom Diet; TD. 95007) or control diet (CD) (LID + KIO₃ [0.71 ug/g iodine]) (Envigo - Teklad Custom Diet; TD. 97350) (11). After 4 weeks of LID, a third group was established to receive 4 days of T4 IP (1.5 μg/100 gBW/d). Unless specified otherwise, all mice had free access to food and water. As indicated, after 4 weeks on the diets, some mice underwent 50% food restriction (FR) for 3 days, whereby they received 50% of food they consumed during the week previously; they were subsequently given food ad libitum. As indicated, some mice that had been on the diets up to 6 weeks (ad libitum) underwent a short-term starvation (S36h) (the food was removed in the evening and 36 hours later [2 darks and 1 day]) and subsequently returned to food ad libitum. All mice were killed by asphyxiation in a CO₂ chamber and the tissues were weighted and plasma collected and stored at –80°C for future analysis.

Indirect calorimetry

Eight-week-old mice were admitted to individual cages in a Comprehensive Lab Animal Monitoring System (Columbus Instruments), which were at thermoneutrality (28°C), and had free access to food and water (12). The 24-hour metabolic profiles were assessed through continuous indirect calorimetry: VO₂ and CO₂ are expressed as mL/kg BW/h and the relationship VCO₂/VO₂, known as the respiratory quotient (RQ), provides an indication of the makeup of the foodstuff combusted. RQ varies between 0.7 and 1.0 for the combustion of fats and carbohydrates, respectively. The percent of contribution of fat oxidation to energy expenditure (EE) was calculated as [468.6 × (1 − RQ)]/[5.047 × (RQ − 0.707) + 4.686 × (1 − RQ)] (described by Lusk, 1924) (8, 13, 14). Also, the system allows for continuous measurement of movement inside the cage that provides the ambulatory activity profiles (15). Mice were acclimatized to the new environment for 1 week before being given the experimental diets.

Plasma biochemistry

TSH plasma levels were measured using a kit developed for rats (16) (Millipore Corp., Billerica, MA); the assay sensitivity of overnight protocol (minimum detectable concentration + 2 SD) is 14 + 34 pg/mL and the inter- and intra-assay coefficient of variation is CV% <10% to 15%. Samples were read on MAGPIX, Luminex Xmap technology (Milliplex map, MA) (17). T4 and T3 plasma levels were measured with a mouse ELISA kit (18, 19) (Cusabio); the sensitivity of T4 and T3 ELISA kit is 20 ng/mL and 0.5 ng/mL; and both inter- and intra-assay coefficient of variation is CV% <15% (20). Plasma free fatty acid level (nonesterified fatty acids [NEFA]; Abcam 65341; sensitivity >2 µM), β-hydroxybutyrate levels (Abcam 83390, sensitivity >0.01 µM; RRID:AB_2868477), and glucose levels (Abcam 65333, sensitivity >1 µM) were measured by colorimetry and read on a plate reader (Cytation 5; Biotech Instruments Inc., VT).

Gene expression analysis

Total tissue RNA was extracted from liver or heart using the RNeasy Lipid tissue Mini Kit (Qiagen) or RNeasy Fibrous tissue Mini Kit (Qiagen), following the manufacturer’s instructions including DNase treatment for 30 minutes. RNA was quantified with a NanoDrop spectrophotometer, and 1.0 μg total RNA was used to produce cDNA using the Transcriptor First Strand cDNA Synthesis Kit (Roche). Genes of interest were measured by RT-PCR (StepOnePlus real-time PCR system; Applied Biosystems) using PowerUp SYBR Green Master Mix (Applied Biosystem). Standard curves consisting of 4 to 5 points of a serially diluted mixture of experimental and control cDNA; the amplification efficiency was 80% to 110% with a coefficient of correlation consistently >0.98 The melting curve protocol was used to verify the specificity of the amplicon generation. Cyclophilin B (CycloB) was used as a housekeeping internal control gene; its expression level was similar between groups. Results were expressed as the ratio of test mRNA to CycloB mRNA levels (17). The target genes studied were: glycogen synthase 2 (Gys2); liver glycogen phosphorylase (Pygl); and Sarco/endoplasmic reticulum Ca2+-ATPase (Serca2). The pair of primers were mGys2: sense 5′-ACCAAGGCCAAAACGACAG-3′ and antisense 5′-GGGCTCACATTGTTCTACTTGA-3′. mPygl: sense 5′-GAGAAGCGACGGCAGATCAG-3′ and antisense 5′-CTTGACCAGAGTGAAGTGCAG-3′. mSerca2: sense 5′- ATGACAATGGCACTTTCTGTTCTA-3′ and antisense 5′- AAGGAGATTTTCAGCACCATCAG-3′. mCycloB: sense 5′-GGAGATGGCACAGGAGGAA-3′ and antisense 5′-GCCCGTAGTGCTTCAGCTT- 3′.

Histology

After dissection, liver and epididymal white adipose tissue (eWAT) were immersed in buffered formalin (10%) and fixed for 24 hours. Paraffin-embedded tissues were sectioned and processed as described (21) for staining with (i) hematoxylin and eosin (H&E); (ii) periodic acid–Schiff (to indicate glycogen content) (22, 23), and (iii) Oil-red O (to indicate lipid content) (24); in the latter case, liver samples were frozen in optimal cutting temperature media and sectioned. The data are representative of 4 to 6 mice of each group and 3 to 4 sections of each animal. The histology slides were scanned by the CRi Pannoramic MIDI 20x whole slide scanner at Integrated Light Microscopy Core Facility from The University of Chicago and photos and analyses were performed by CaseViewer software. For the adipocyte area, 20 adipocytes form each animal was measured by CaseViewer software. All animals’ sections are present in Figure S2 and S3 (25). Original magnification was ×10 and ×20 for each slide.

Statistical analyses

All data were expressed as mean ± SEM or box-and-whiskers plot indicating median and quartiles and analyzed using PRISM software (GraphPad Software, Inc., San Diego, CA). One-way ANOVA was used to compare more than 2 groups, followed by the Student-Newman-Keuls test to detect differences between 2 groups. The Student t test was used to compare differences between two groups. P < 0.05 was used to reject the null hypothesis.

Results

To investigate the metabolic modifications occurring during ID, mice were placed on a LID or CD and housed at thermoneutrality. A question that will permeate the interpretation of the results is whether TH signaling in the different tissues was preserved in the LID mice. The LID regimen markedly reduced plasma T4 and elevated plasma TSH levels, without affecting plasma T3 levels (Table 1). This is similar to what has been shown in mice (26) and rats (27). TH signaling was not compromised in LID mice as assessed through sustained weight gain, growth (tibial length), VO₂, ambulatory activity, and the cardiac expression of a well know T3-responsive gene (Serca2); all parameters were similar to CD mice (Table 2 and S1, Fig. 1A–E and S1) (25) growth delay or arrest a hallmark of hypothyroidism in small rodents (10).

Table 1.

Plasma TSH and TH Levels, Intermediary Metabolites, and Specific Liver mRNA Levels in Mice Kept on CD or LID

Parameter	CD Mice	LID Mice
TSH (ng/mL)	1.1 ± 0.45	70 ± 20***
T4 (ng/mL)	45 ± 5.6	22 ± 1.1***
T3 (ng/mL)	0.46 ± 0.03	0.48 ± 0.01
NEFA (nmol/μL/g)	13.9 ± 5.5	19.4 ± 6.4#
β-HB (nmol/μL)	0.80 ± 0.48	0.47 ± 0.07*
Glucose (nmol/μL)	8.4 ± 1.1	7.4 ± 0.5
Gys2 mRNA	1 ± 0.36	1.5 ± 0.46**
Pygl mRNA	1 ± 0.39	0.93 ± 0.16

Data were pooled from mice kept on CD or LID for up to 4 to 6 weeks; NEFA was divided by (total adipose tissue (g)/body weight(g)); mRNA levels of the target gene were measured by RT-quantitative PCR and normalized by the housekeeping gene CycloB. The following genes were studied: Gys2: Glycogen synthase 2; Pygl: liver glycogen phosphorylase. Values are mean ± SD. Data from 2 independent experiments were pooled and analyzed by Student t test; n = 4-11 animals per group. ^#P = 0.07; *P < 0.05, **P < 0.01, ***P < 0.001 vs CD. Abbreviations: AT, adipose tissue; CD, control diet; LID, low iodine diet; NEFA, nonesterified fatty acids; β-HB: β-hydroxybutyrate.

Table 2.

Body Weight and Relative Organ Weight in Mice Kept on CD or LID

Tissue Parameter	CD Mice	LID Mice
BW initial (g)	24 ± 1.6	24 ± 0.9
BW final (g)	28 ± 3.1	27 ± 2.0
Food intake (g/d)	3.2 ± 0.4	3.9 ± 0.7
Tibia (mm)	18 ± 0.15	18 ± 0.49
Thyroida	0.06 ± 0.01	0.19 ± 0.01b
Livera	49 ± 4	48 ± 10
WATa	32 ± 3	18 ± 4b
RPa	11 ± 1.8	5.0 ± 1.8b
SUBa	13 ± 2	10 ± 1c
Hearta	5.2 ± 0.9	5.1 ± 1
Gastroa	5.0 ± 0.5	5.0 ± 0.7
Kidneya	7.0 ± 0.7	6.0 ± 1

Data obtained from mice kept on CD or LID for 6 weeks. Values are mean ± SD. Data from 2 independent experiments were pooled and analyzed by Student t test; n = 4-6 animals per group.

^a values are /g BW ×1000; food intake were measured 1 week before food restriction.

^b P < 0.001 vs CD.

^c P < 0.05 vs CD.

Abbreviations: BW, body weight; Gastro, gastrocnemius; RP, retroperitoneal fat; SUB, subcutaneous fat; WAT, white adipose tissue.

Figure 1.

Metabolic phenotype of mice kept on CD or LID or LID+T4. (A-E) VO₂ of first through fifth weeks during the light and dark cycle (–); (F-J) Same as A, except that RQ is shown. Values are the mean ± SEM or box-and-whiskers plot indicating median and quartiles (n = 4); area under the curve (AUC) was calculated during light and dark cycle for each individual animal. The difference between both cycles are that the dark cycle it is when the animal is actively eating and light cycle it is when the animal is resting. Student t test or 1-way ANOVA was used to compare all groups. *P < 0.05; **P < 0. 01; ***P < 0.001 vs CD. CD, control diet; EE, energy expenditure; LID, low iodine diet; RQ, respiratory quotient.

Metabolic assessment during LID

During the 4-week period, LID mice had a ~30% higher food intake compared with CD mice but maintained similar body weight (Table 2 and S1) (25). Indirect calorimetry revealed that 24-hour VO₂ during the 4-week period (average light-dark: 2100-2800 mL/kg/h) was similar in both groups (Fig. 1A–D); CD and LID mice presented similar movement profiles (Figure S1) (25). There was a tendency for lower 24-hour RQ in LID mice by the second week of iodine deficiency, which became more pronounced after 4 weeks (average light-dark: 0.95-0.99 vs 0.83-0.87, LID vs CD, respectively) (Fig. 1F-I). As a result, the calculated overall fat contribution to the 24-hour EE increased progressively from the second week on to the fourth week of LID (Fig. 2A–D).

Figure 2.

Contribution of fat oxidation to EE of CD or LID mice kept on CD or LID or LID+T4. (A-E) Contribution of fat contribution to EE (%) of first through fifth weeks during the light and dark cycle (–). Values are shown in box-and-whiskers plot indicating median and quartiles (n = 6); Student t test or 1-way ANOVA was used to compare all groups for the light or dark cycle. *P < 0.05; **P < 0. 01; ***P < 0.001 vs CD. CD, control diet; EE, energy expenditure; LID, low iodine diet.

At the end of the fourth week of LID, half of animals received treatment with T4 (1.5 μg/100 g BW/d) for 4 days. This was sufficient to normalize plasma TSH without affecting T3 levels; plasma T4 levels were slightly elevated (Table S1) (25). Although treatment with T4 did not affect 24 h-VO₂ (average light-dark: 2000-2500 mL/kg/h; Fig. 1E), it promptly restored the RQ (average light-dark: 0.86-0.89 vs 0.97-0.98, LID+T4 vs CD, respectively), the contribution of fat to EE (Fig. 1J and Fig. 2E) and the weight of the adipose tissues (Table S1) (25) to CD levels.

While we established that the drop in RQ associated to LID was well defined by the 4th week, we next studied a series of histological and biochemical parameters in mice that were kept on LID for 6 weeks. At the end of the experimental period, mice were killed and their thyroids were found to be increased by ~3-fold, confirming a state of ID (Table 2 and S1) (25). The H&E staining of liver sections revealed the presence of clear intracellular areas resembling storage vacuoles in the CD liver (Fig. 3A; S2) (25), which were increased markedly in the LID liver (Fig. 3B; S2) (25). The nature of these vacuoles was investigated through periodic acid–Schiff staining, which turns purple in the presence of glycogen. Indeed, CD mice exhibit deposits of glycogen widespread throughout the liver parenchyma (Fig. 3C; S2) (25), which were dramatically increased in the LID mice (Fig. 3D; S2) (25). It is notable that the liver of LID mice also showed a ~1.5-fold increase in Gys2 mRNA levels, whereas Pygl mRNA levels remained unaffected (Table 1).

Figure 3.

Representative histology of liver and eWAT sections from mice kept on CD or LID. Histology of liver stained (A-B) with hematoxylin and eosin; (C-D) with periodic acid–Schiff (PAS); (E-F) with oil red-O. (G-H) Histology of eWAT stained with hematoxylin and eosin; (I) adipocyte area (μm²). (A, C, E, G) CD mice; (B, D, F, H) LID mice. The sections shown are representative of 4 to 6 mice in each group and 3 to 4 sections of each animal; original magnifications are ×10 and ×20. Student t test was used to compare CD and LID groups. ***P < 0.001 vs CD. CD, control diet; eWAT, epididymal white adipose tissue; LID, low iodine diet.

The weight of different fat depots was ~60% reduced in LID mice, including epididymal, retroperitoneal, and subcutaneous adipose tissue (Table 2). This was likely the result of accelerated lipolysis in LID mice given the reduced adipocyte area of eWAT (Fig. 3G–I; S3) (25), higher NEFA levels in the circulation, and greater contribution of fat to 24 hours (Table 1). Moreover, lipid deposits in the liver were reduced in LID mice, as assessed through staining with oil red-O (Fig. 3E, F; S2) (25), as well as β-hydroxybutyrate levels were in the circulation (Table 1). Nonetheless, LID feeding did not affect glucose levels in the circulation (Table 1).

Three-day food restriction

To force mobilization of energy substrates, mice that had been on CD or LID for 4 weeks underwent a 50% FR for 3 days. Although both groups entered FR with similar body weights (CD: 26.7 ± 2.2 g vs LID: 26.1 ± 1.6 g), CD mice experienced a ~7% drop in body weight, whereas the LID mice lost only ~2.5% (Figure S4). The 3-day VO₂ dropped in both groups by 7% to 8% (Fig. 4A–C); only in the CD mice did the light-RQ drop to ~0.81 (Fig. 4D, E). There was only a trend for the dark-RQ to drop as well (Fig. 4D and F). This difference between light and dark is likely to reflect lower insulin levels during the light period, which is when the mice eat much less.

Figure 4.

Metabolic phenotype of CD or LID mice undergoing 3 days of 50% FR. (A) VO₂ on day 1 with food ad libitum, days 2 through 4 with FR (50%), day 5 and 6 refeeding during light and dark cycle (–); (B) average of VO₂ during light cycle; (C) average of VO₂ during dark cycle (–); (D) same as in A, except that what is shown RQ; (E) same as in B, except that what is shown is RQ; (F) same as in C, except that what is shown is RQ. Values are the mean ± SEM (n = 6); Student t test was used to compare CD and LID groups. ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs respective group day 1; *P < 0.05, **P < 0.01, ***P < 0.001 vs CD same day. CD, control diet; FR, food restriction; LID, low iodine diet; RQ, respiratory quotient.

In contrast, the 24-hour RQ in the LID mice remained steady at 0.86 to 0.96 (light-dark) throughout the 3-day FR period (Fig. 4D–F). The changes in RQ values indicate that FR increased fat contribution to the light-EE in CD mice by 45% to 50%, to levels above the LID mice. However, fat contribution remained steady in the LID mice (Fig. 5A).

Figure 5.

Contribution of fat oxidation to EE of CD or LID mice undergoing 3 days of 50% FR. (A) Contribution of fat contribution to EE (%) during FR of light and dark cycle (–); (B) same as in G, except that what is shown is during refeeding. Values are shown in box-and-whiskers plot indicating median and quartiles (n = 6); Student t test was used to compare CD and LID groups during light or dark cycle. **P < 0.01, ***P < 0.001 vs CD. CD, control diet; EE, energy expenditure; FR, food restriction; LID, low iodine diet.

After 2 days of recovery, during which food was given ad libitum, body weight was restored and once again no differences between groups was observed (CD: 27.0 ± 2.1 g vs LID: 26.3 ± 1.7 g); likewise, 24-hour VO₂ went back to normal in both groups (Fig. 4A–C). The CD mice in particular exhibited a marked increase in 24-hour RQ from 0.97 to 1.0 (light-dark), but in the LID mice 24-hour RQ only increased slightly from 0.89 to 0.97 (light-dark) (Fig. 4D–F). By the end of the recovery period, fat contribution to 24- hour EE in CD mice was back to being lower than in LID mice (Fig. 5B).

36-hour starvation

Next, mobilization of energy substrates in mice that had been on CD or LID for 6 weeks was forced using 36-hour starvation. Although both groups entered S36h with similar body weights (CD: 27.4 ± 3.0 g vs LID: 26.4 ± 1.7 g), during the subsequent S36h period, both CD and LID mice lost ~10% body weight (Figure S4). In addition, the VO₂ (light-dark: 1440-1910 mL/kg/h) dropped to lower levels in the LID mice than in CD mice (P = 0.08) (Fig. 6A-C), but the drop in RQ was more pronounced in the CD mice (dark-light: 0.78-0.77 vs 0.74-0.74; P = 0.15, LID vs CD, respectively) (Fig. 6D-F). Based on the changes in RQ, S36h increased fat contribution to 24-hour EE in CD mice by 2.5- to 5.9-fold (light-dark), whereas in the LID mice the increase was limited to 0.6- to 3.4-fold (light-dark) (Fig. 7A).

Figure 6.

Metabolic phenotype of CD or LID mice submitted to S36h. (A) VO₂ on day 1 with food ad libitum, days 2 and 3 with S36h, days 4 and 5 refeeding during light and dark cycle (–); (B) average of VO₂ during light cycle; (C) average of VO₂ during dark cycle (–); (D) same as in A, except that what is shown is RQ; (E) same as in B, except that what is shown is RQ; (F) same as in C, except that what is shown is RQ. Values are the mean ± SEM (n = 6); Student t test was used to compare CD and LID groups. Values are the mean ± SEM (n = 6); Student t test was used to compare CD and LID groups. ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs respective group day-1; *P < 0.05, **P < 0.01, ***P < 0.001 vs CD same day. CD, control diet; LID, low iodine diet; RQ, respiratory quotient; S36h, short-term starvation.

Figure 7.

Contribution of fat oxidation to EE of CD or LID mice submitted to S36h. (A) Contribution of fat contribution to EE (%) during FR of light and dark cycle (–); (B) Same as in G, except that what is shown is during refeeding. Values are shown in box-and-whiskers plot indicating median and quartiles (n = 6); Student t test was used to compare CD and LID groups during light or dark cycle. **P < 0.01, ***P < 0.001 vs CD. CD, control diet; EE, energy expenditure; FR, food restriction; LID, low iodine diet; S36h, short-term starvation.

After 2 days of being switched to food ad libitum, all mice rapidly regained body weight (CD: 28.2 ± 3.0 g vs LID: 27 ± 2.0 g), and accelerated 24-hour VO₂ with no differences between groups (Fig. 6A–C). The CD mice in particular exhibited a marked increase in 24-hour RQ from 0.99 to 1.0 (light-dark) but in the LID mice 24-hour RQ only increased slightly from 0.93 to 0.98 (light-dark) (Fig. 6D–F). The fat contribution to EE decreased in both groups, and the LID mice returned to have a higher fat contribution to EE vs CD mice (Fig. 7B). This is more dramatic during the light period, likely as a result of lower insulin levels.

Discussion

The present studies revealed that feeding mice on LID results in substantial metabolic alterations that are unlikely to result from changes in thyroid status. Mice that were kept on LID for 4 to 6 weeks gained weight normally, exhibited normal tibial growth, normal rate of metabolism, similar movement profile, and no change in the expression of a typical T3-responsive gene in the heart (Table 2; Fig. 1; S1) (25). In addition, plasma T3 levels were not affected in these mice. All of these parameters indicate that LID mice remained systemically euthyroid (Table 1). At the same time, LID mice exhibited several metabolic changes that were compatible with accelerated lipolysis: (i) higher circulating NEFA levels (Table 1); (ii) reduced adiposity (Table 2 and Fig. 3G–I); (iii) elevated contribution of fat oxidation to the EE (Fig. 2); and (iv) elevated hepatic storage of glycogen (Fig. 3). When challenged with food deprivation, CD mice exhibited the expected drop in body weight and RQ, whereas LID mice lost less body weight and did not accelerate further fat oxidation (Fig. 4 and 5; S4) (25), a scenario that is compatible with a limited ability of LID mice to further mobilize fatty acids because of marked reduction in adiposity.

It is unclear what the mechanisms triggering these changes were, given the apparent state of euthyroidism detected in the LID mice. The SNS interacts with TH signaling and could have been playing a role. However, all mice were kept at thermoneutrality to avoid adjustments in the SNS signal. Hence, 2 not mutually exclusive possibilities could be involved.

First, it is conceivable that the markedly elevated plasma TSH observed in the LID mice (Table 1) acted directly in the adipose tissue and/or liver, accelerating lipolysis and thus fat oxidation. Adipose tissues are known for expressing the TSH receptor (TSHr), which is coupled to the adenylate cyclase and was historically used to assay circulating levels of thyroid-stimulating immunoglobulins (28). Indeed, mice with adipose tissue-specific inactivation of TSHr (AT-TSHr-KO) exhibit reduced lipolytic rates and increase adipocyte size, which are compatible with impaired lipolysis (29, 30). Furthermore, this AT-TSHr-KO mouse exhibits a reduction in the expression of lipolytic genes, including Adrβ1, Adrβ3, and Lipe, the latter being the rate-limiting enzyme in lipolysis (31). Thus, it makes sense that the markedly elevated circulating levels of TSH observed in the LID mice was responsible for accelerating lipolysis and the oxidation of fatty acids (Table 1). That this was likely the case is supported by the rapid normalization of adiposity, RQ, and fat contribution to EE observed in the LID briefly treated with T4, which normalized plasma TSH levels (Fig. 1-2 and Table S1) (25). Of course, T4 could be playing a direct role as well. However, TH accelerates lipolysis, making it unlikely that it could be involved in this setting, even if nongenomic effects are considered (32).

A TSH-mediated increase in lipolysis can also explain our findings of increased hepatic glycogen content, and increased mRNA levels for Gys2 in the LID mice, without alterations in glucose plasma levels (Fig. 3A-F and Table 1). Lipolysis releases glycerol from adipose tissue, which is taken up by the liver. There, glycerol stimulates gluconeogenesis thus increasing the amount of intracellular glucose 6-phosphate (33). The latter is a key signal molecule that stimulates hepatic glycogen synthase and glycogen synthesis (34-37). Indeed, there is evidence in humans that glycerol can potently suppress hepatic glycogenolysis by ~45% (38). Furthermore, TSHr is also expressed in the liver and, conceivably, TSH signaling could also be playing a direct role in this organ. However, the studies available about the role played by TSH in the liver are difficult to reconcile. On one hand, mice with liver-specific-TSHr- KO exhibited no alterations in 24-hour VO2, 24-hour RQ, or 24-hour EE or no changes in H&E liver histology, although a reduced lipid content was observed (39). On the other hand, an abnormal glucose metabolism was observed in mice with subclinical hypothyroidism (normal serum T3 and T4 and elevated serum TSH). In these studies, TSH was found to directly stimulate gluconeogenesis via cAMP/protein kinase A pathway (40). Therefore, it is unclear that by acting directly in the liver, TSH could explain all/part of the phenotype developed by our mice.

Second, based on studies of LID rats (41), it is conceivable that LID mice exhibited liver-specific reduction in TH signaling, despite all the other indications that they remained systemically euthyroid. These studies in rats revealed that after a 5-week period on LID, rats exhibited a reduction hepatic nuclear T3 content and in α-glycerophosphate dehydrogenase and malic enzyme activities (41). This would also make sense given that rats with systemic hypothyroidism have increased hepatic glycogen depots because of decreased gluconeogenesis and glycogenolysis (42-44). Nonetheless, it is challenging in this setting to define liver-specific TH signaling given that all known T3-responsive genes in the liver are metabolic genes that are also responsive to general metabolic stimuli, such as insulin, glucagon, glucose, and fatty acids levels. As we have observed in this model, the LID mice developed profound metabolic alterations which are likely to affect the expression of typical T3-responsive genes in the liver per se.

The present findings in mice are in line with the studies of humans with ID, which show an increased risk of developing metabolic disorders (i.e., dyslipidemia with hypercholesterolemia and glucose intolerance) (5-7). However, a full reconciliation of the results is difficult given, among other things, on how the ID evolves in each model. In the mouse model, the control diet is identical to the ID diet except for the iodine content. In humans, ID might be associated with malnutrition and/or an unbalanced diet, compounding factors that contribute to the final metabolic phenotype.

In conclusion, the present results indicate that LID mice exhibit important metabolic alterations that are unlikely the result of changes in systemic thyroid status. These changes include reduced adiposity, increased fat contribution to EE, and increased liver glycogen depots, which can all be explained by accelerated lipolysis. Given that these animals were kept at thermoneutrality, a logical explanation is that lipolysis was activated directly by the elevated plasma TSH levels seen in the LID mice. To our knowledge, these are the first descriptions of this phenotype in LID mice and therefore need to be confirmed. They provide new insights about the relationship between iodine deficiency and energy metabolism.

Glossary

Abbreviations

CD

control diet

EE

energy expenditure

eWAT

epididymal white adipose tissue

FR

food restriction

Gys2

glycogen synthase 2

H&E

hematoxylin and eosin

ID

iodine deficiency

LID

low iodine diet

NEFA

nonesterified fatty acid

Pygl

liver glycogen phosphorylase

RQ

respiratory quotient

S36h

short-term starvation

Serca2

Sarco/endoplasmic reticulum Ca2+-ATPase

SNS

sympathetic nervous system

TH

thyroid hormone

THSr

TSH receptor

Additional Information

Disclosure Summary: A.B. is a consultant for Allergan Inc, Synthonics Inc and BLA Technology LLC; the other authors have nothing to disclose.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories listed in Reference (25).