Delayed circadian phase of wheel running activity

We first determined if the mice used in the present study recapitulated that circadian alteration previously reported [21]. We compared circadian parameters using the recently developed statistical software CircaCompare [24]. Analysis of the running wheel results reveals statistical significance in differences in phase between Dio3+/+ and Dio3-/- groups of both sexes. Results of male have statistical significance in all three rhythmic parameters (mesor, amplitude and phase) between the two groups. The average circadian phase of Dio3-/- mice delayed from 1.08 (4.13 h, female) to1.33 (5.08 h, male) radians (Fig. 1). However, there are no significant differences in the circadian phase of energy expenditure results between the two groups when the running wheels were removed, although other two rhythmic parameters (mesor and amplitude) still have significance in differences (male: P-value for mesor difference < 0.001, P-value for amplitude difference < 0.001, P-value for difference in phase 0.37; female: P-value for mesor difference 0.40, P-value for amplitude difference < 0.01, P-value for difference in phase 0.057) between the two groups (Supplementary Fig. 1). These observations were consistent with the circadian phenotype previously observed in the Dio3-/- mice and supported the molecular studies included herein.

Fig. 1

Circadian parameters of running wheel data of adult Dio3-/- mice. Typical 48 h of data was analyzed using CircaCompare, according to Parsons et al., 2020 [23]. (a), male, n = 13,13; (b), female, n = 11,11

Circadian variations in serum thyroid hormones and corticosterone

We determined serum concentrations of thyroid hormones and corticosterone at four time points (ZTs, Zeitgeber time) (Fig. 2). Dio3+/+ male mice exhibit a marked circadian pattern of serum corticosterone, and this rhythm was not altered in Dio3-/- males (Fig. 2a, left). Concerning serum T4, a moderate circadian pattern was observed in Dio3+/+ males with a peak value around ZT6-ZT12. At ZT6, Dio3-/- males manifested significantly lower serum T4, while T4 was at same level at other ZTs (Fig. 2a, right).

Fig. 2

Circadian variation in serum corticosterone and T4 in adult mice. (a), Serum corticosterone and T4 in adult males. (b), Serum corticosterone and T4 in adult females. Values are expressed relative to values at ZT0 in wild type mice. *, **, ***, P < 0.05, 0.01 and 0.001, respectively, Dio3+/+ vs. Dio3-/-,as determined by two-way ANOVA and Tukey’s post hoc test (n = 4–5). (c), Immunofluorescence of cyp11b1 in female adrenal gland. Each picture set represents the results of three mice

A circadian pattern in serum corticosterone was also observed in Dio3+/+ females (Rhythmic, P = 0.044), with a peak value at ZT12-ZT18. However serum corticosterone in Dio3-/- females was significantly lower than in wild-type females at ZT 12 and ZT18 (Fig. 2b, left), and its rhythm was not significant (not rhythmic, P = 0.30, all rhythmic parameters can be accessible in supplementary rhythmic parameters file). Similarly, serum T4 in Dio3+/+ females changed along the ZTs with a peak value at ZT18, while T4 rhythm seems disturbed in Dio3-/- females, and were significantly lower than those in Dio3+/+ littermates at ZT0, ZT6 and ZT18 (Fig. 2b, right). Taken together, these observations suggest that DIO3-deficiency has significant effects on the circadian physiology of both the HPT and HPA axes, with a remarkable sexual dimorphism.

Adrenal gland expression (IF) of Cyp11b1

The alterations in serum corticosterone, especially in female Dio3-/- mice, suggested dramatic changes in its synthesis. Cyp11b1 (Steroid 11-Beta-Hydroxylase) is key enzyme for adrenal gland to produce corticosterone [25, 26], and we analyzed its expression in the adrenal gland by immunofluorescence (IF). IF signal of Cyp11b1 at ZT12 showed a substantial decrease in the female adrenal gland, suggesting a critical limitation in the production of corticosterone at the peak of its cycle.

Hypothalamic gene expression

We then measured hypothalamic expression of clock-related genes at different ZTs. Clock genes were particularly impacted by DIO3 deficiency at the start of the dark cycle (ZT12), with Dio3-/- males exhibiting significant decreases in the expression of Dbp, Per1, Per2, Clock, Tbx3, Rora and Cry2. The expression of Per2 was increased at ZT0 and ZT6 in the hypothalamus of Dio3-/- males. We observed no significant differences in the expression of Bmal1. (Fig. 3a)

Fig. 3

Hypothalamic expression of genes related to the circadian clock in male (A) and female (B) mice. Data represent the mean ± SEM of 4–5 mice per experimental group and ZT. *P < 0.05, **P < 0.01, ***, P < 0.001, as determined by two-way ANOVA and Tukey’s post hoc test

In Dio3-/- females, hypothalamic expression of clock genes was less affected than in males. Dio3-/- females showed increased expression of Dbp and Tbx3 at ZT0 and ZT18, of Rora and Per2 at ZT0). They also manifested decreased expression of Rora at ZT18 and of Cry2 at ZT12. We observed no significant differences in the expression of Clock, Bmal1 and Nr1d1. (Fig. 3b)

We also investigated the expression of genes related to T3 action, including thyroid hormone receptors alpha and beta (Tra and Trb, respectively), T3-regulated genes Hr and Klf9, and thyroid hormone deiodinases Dio2 and Dio3. (We can measure Dio3 mRNA in Dio3-/- mice since these mice carry a triple point mutation in the Dio3 coding region that renders the enzyme fully inactive [22], but leaves the mRNA largely intact). In Dio3+/+ male mice, Dio2 expression showed a pattern with higher expression at ZT12 and ZT18 compared to ZT0 and ZT6. This pattern was disrupted in Dio3-/- mice, showing a substantial reduction in Dio2 expression at ZT12 (Fig. 4a). Hypothalamic Dio3 mRNA did not vary across the ZTs in Dio3+/+ males, but it was markedly elevated in Dio3-/- at ZT6, ZT12 and ZT18 (Fig. 4a). In contrast to males, Dio2 expression was significantly increased at ZT0 in Dio3-/- females versus controls (Fig. 4b). In Dio3+/+ females, Dio3 mRNA showed a moderate circadian pattern, with highest expression at ZT18. Similarly to males, Dio3 expression in Dio3-/- females was substantially increased at all ZTs (Fig. 4b).

Fig. 4

Hypothalamic expression of genes related to thyroid hormone action. (a, b) Hypothalamic gene expression in male (a) and female (b) mice. Data represent the mean ± SEM of 4–5 mice per experimental group and ZT. *P < 0.05, **P < 0.01, ***, P < 0.001, as determined by two-way ANOVA and Tukey’s post hoc test

Concerning the hypothalamic expression of the T3 receptors, Dio3-/- females showed no differences in the expression of Tra or Trb at any ZT, but their expression, especially that of Trb, showed a noticeable circadian profile, with a trough at ZT18 (Fig. 4b). In Dio3+/+ males, both Tra and Trb showed a circadian profile with a trough at ZT12, while Dio3-/- males showed a significant decreased in Tra and Trb expression at ZT0 and ZT12, respectively(Fig. 4a).

Expression of T3-regulated genes Hr and Klf9 revealed consistent circadian patterns same as Dio2 or Dio3 in Dio3+/+ males or females, respectively (Fig. 4), but we observed significant differences in Dio3-/- mice. In Dio3-/- females, Hr expression was increased at ZT0 and ZT18 in Dio3-/-, while Klf9 expression was increased at ZT18 (Fig. 4b). In Dio3-/- males, there was no significant differences in the expression of Klf9, but Hr mRNA was increased at ZT0 and decreased at ZT12 (Fig. 4a). These results suggest T3 drove the phase shift in gene expression through T3-responsive genes.

We analyzed the phase shift in all the hypothalamic gene expression results with CircaCompare. It gives a Presence of rhythmicity (P-value) 0.0014 for Dio3-/- and 0.0015 for Dio3+/+, φ1 = 0.408 (1.56 h) for the males. However both groups of data are arrhythmic for the females: presence of rhythmicity (P-value) 0.26 for Dio3-/- and 0.11 for Dio3+/+, φ1= -0.832 (-3.18 h). These data suggest sex-biased alterations in the hypothalamic circadian rhythm of Dio3-/- mice.

Hepatic gene expression

We examined gene expression in the liver. In this tissue, clock-related genes exhibited circadian patterns in wild type mice of both sexes (Supplementary Figs. 2, 3). In Dio3-/- males, expression of clock-related genes showed no remarkable differences (Supplementary Fig. 2). Dio3-/- females showed limited differences in the hepatic expression of clock genes, although decreased expression of Per1 at ZT0 and Tbx3 at ZT18 seems to exacerbate the circadian profile of these genes (Supplementary Fig. 3).

Concerning the expression of genes involved in T3 signaling, hepatic Dio3 showed a mild circadian profile in males, with lowest expression at ZT18. This profile was ameliorated in Dio3−/− males due to significantly increased Dio3 expression at ZT12 and ZT18 (Fig. 5a). T3-regulated genes Dio1, Klf9 and Hr showed rhythmic expression profiles, and the expression of T3-regulated genes Dio1 and Hr was significantly lower at ZT18 (Fig. 5a), suggesting lower T3 action at this ZT. But we observed no circadian expression pattern in Tra or Trb.

Fig. 5

Hepatic expression of genes related to thyroid hormone action in male (a) and female (b) mice. Data represent the mean ± SEM of 4–5 mice per experimental group and ZT. *P < 0.05, **P < 0.01, ***, P < 0.001, as determined by two-way ANOVA and Tukey’s post hoc test

In females, we observed a mild circadian pattern in the expression of Dio3, Tra and Trb, with lowest expression at ZT18 for all three genes (Fig. 5b). In Dio3-/- females, this pattern was partially disrupted for Dio3 and Tra due to significantly decreased expression of these genes at ZT0. Interestingly, the hepatic expression of T3-regulated genes showed a more pronounced circadian pattern in females with highest expression at ZT18 for Dio1 and Hr, and at ZT6 for Klf9. However, this circadian pattern was completely ablated for Dio1 due to significant reductions in expression at ZT12 and ZT18 in Dio3-/- females (Fig. 5b).

We also analyzed the circadian phase shift in the hepatic gene expression results. The results from male mice indicate a presence of rhythmicity (P-value) 0.068 for Dio3-/- and 0.063 for Dio3+/+, φ1 = 0.045 (0.17 h). Both groups of data are more arrhythmic for the females: presence of rhythmicity (P-value) 0.11 for Dio3-/- and 0.13 for Dio3+/+, φ1 = 0.044 (0.168 h). These data indicate no significant effect on hepatic circadian rhythm in Dio3-/- of either sex.

Gene expression in white adipose tissue (WAT)

The WAT expression of most clock-related genes exhibited stronger circadian profiles in females of both genotypes (Supplementary Fig. 5) than in males (Supplementary Fig. 4). The loss of Dio3 function did not erase the rhythm of gene expression in WAT, although we noted a significant increase in the expression of Nr1d1 at ZT0 in the Dio3-/- male WAT. (Supplementary Fig. 4). Similarly, although significant differences were observed in the expression of Per1 at ZT0, Bmal1 at ZT12, and Nr1d1 at ZT18, these differences did not translate into a major disruption of the overall circadian profiles of those genes in the WAT of females (Supplementary Fig. 5).

Concerning genes related to T3 signaling, we observed that Dio3 did not show a circadian pattern of expression in the WAT of Dio3+/+ males, but its expression was significantly increased in Dio3-/- male mice at all ZTs (Fig. 6a). Interestingly, Dio2 expression, despite the variability in the data, suggests a circadian pattern of expression with highest levels at ZT18 (Fig. 6a). This Dio2 rhythm was ablated in Dio3-/- males. In addition, our data show that the Dio2 expression curve matches the trend of Trb, and the Dio3 expression curve matches the trend of Tra and Klf9 (Fig. 6a). We also measure the expression of Lep, Lpl and Mest, which are related to adiposity [27,28,29]. Interestingly, the expression of Lep Lpl and Mest showed a peak at ZT18, and this pattern was disrupted in the WAT of Dio3−/− males (Fig. 7a). The Lep expression was significantly decreased in Dio3-/- males at ZT18. On the contrary, the Lpl expression was significantly increased in Dio3-/- males at ZT0 and ZT18 (Fig. 7a).

Fig. 6

Expression of genes related to thyroid hormone action in male (a) and female (b) mice WAT. Data represent the mean ± SEM of 4–6 mice per experimental group and ZT. *P < 0.05, **P < 0.01, as determined by two-way ANOVA and Tukey’s post hoc test

Fig. 7

Expression of genes related to adiposity in male (a) and female (b) mice WAT. Data represent the mean ± SEM of 4–6 mice per group and ZT. *P < 0.05, **P < 0.01, as determined by two-way ANOVA and Tukey’s post hoc test

We obtained different results in the WAT of females. In contrast to males, the expression curves all match one similar trend in the WAT of females, which showed peak expressions at ZT6. This pattern was disrupted in Dio3-/- females largely as a result of a marked reduction in expression at ZT6 (Fig. 6b). We did not observe significant alterations in Dio3-/- females in the expression of Tra, Trb, Klf9, Lep, Lpl or Mest (Figs. 6b and 7b). However, in the case of Lep, Lpl and Mest, the trend was similar to that of males, with peak expression at ZT18 (Fig. 7b).

We then analyzed the circadian phase shift in the WAT gene expression results with CircaCompare. The analyses showed a presence of rhythmicity (P-value) 0.34 for Dio3-/- and 0.13 for Dio3+/+, φ1 = 0.178 (0.68 h) for the males. Both groups of data are more rhythmic for the females: presence of rhythmicity (P-value) 0.11 for Dio3-/- and 0.055 for Dio3+/+, φ1 = 0.234. (0.89 h). These data suggest Dio3-/- affects WAT circadian rhythm in both sexes of mice.

Gene expression in male brown adipose tissue (BAT)

We extended our investigations of circadian gene expression to male BAT. In this tissue, most clock-related genes showed a strong circadian pattern of expression. We observed significant differences in the expression of certain genes (Dbp, Nr1d1) at particular ZTs (Supplementary Fig. 6), but these changes did not substantially disrupt the circadian pattern except for Cry2, whose expression was significantly reduced at ZT6 and markedly increased at ZT18 (Supplementary Fig. 6).

Concerning T3-signaling, both Dio2 and Dio3 expression showed a strong circadian pattern of expression in wild type mice, with peak expression at ZT6. This pattern was disrupted in Dio3-/- mice due to a sharp reduction in expression precisely at ZT6 for both Dio2 and Dio3, and to a significant expression increase at ZT18 for Dio3 (Fig. 8). Both Tra and Trb exhibited an expression pattern in males, with the lowest expression at ZT18, but the overall pattern was not dramatically disrupted despite small but statistically significant expression differences (Fig. 8). We observed no changes in the circadian expression of Klf9, although Hr mRNA exhibit a circadian profile with lowest expression at ZT18, a time in which the expression was increased in Dio3-/- mice (Fig. 8).

Fig. 8

Expression of genes related to thyroid hormone action in BAT of male mice. Data represent the mean ± SEM of 4–5 mice per experimental group and ZT. *P < 0.05, **P < 0.01, ***, P < 0.001, as determined by two-way ANOVA and Tukey’s post hoc test

We also analyzed the expression of genes of relevance to BAT physiology. Pronounced circadian patterns were observed in Dio3+/+ mice in the expression of Pgc1a, Prdm16 and Elovl3, with peak expression at ZT6 (Pgc1a and Prdm16) or ZT18 (Elovl3) (Fig. 9a). For Pgc1a, the expression pattern was significantly disrupted in Dio3-/- due to decreased expression at ZT6. In the case of Elovl3, marked increases (ZT0 and ZT12) or decreases (ZT18) in expression changed its rhythmic pattern from the control (Fig. 9a). Adiponectin, Lpl, and Pparg all exhibited significant expression pattern changes in mutant mice (Fig. 9a). In contrast to WAT, the BAT expression of Lep and Mest showed more significant increases in mutant mice (Fig. 9a). The expression of Lep in Dio3-/- mice was reduced at ZT12 and markedly elevated at ZT18. The notable increased expression of both Lep and Mest in the BAT of mutant mice at ZT18 stands in sharp contrast with the results from WAT, in which we observed the opposite regulation (Fig. 7a). Although we noted no significant change in Ucp1 expression, the expression curve of mutant mice did show a more flattened pattern (Fig. 9a).

Fig. 9

Expression of genes relevant to BAT physiology. (a), real-time PCR data of lipid metabolism genes in BAT. Data represent the mean ± SEM of 4–5 mice per experimental group and ZT. *, **, ***, **** indicate P < 0.05, 0.01, 0.001, 0.0001, as determined by two-way ANOVA and Tukey’s post hoc test. (b), plot of the real-time PCR data using circadian analyzing software CircaCompare

We also analyzed the circadian phase shift in BAT gene expression data. We obtained a presence of rhythmicity (P-value) 0.04 for Dio3-/- and 0.01 for Dio3+/+, φ1 = 0.552 (2.11 h) for the circadian clock genes (Fig. 9b). Interestingly, there is a significant change in the rhythmicity of the lipid metabolism genes of the Dio3-/- BAT. The analysis of the lipid metabolism genes shows the circadian rhythm was significant in the Dio3+/+ BAT and erased (Fig. 9b) in the Dio3-/- BAT: presence of rhythmicity (P-value) 0.54 for Dio3-/- and 0.02 for Dio3+/+, φ1 = 1.211. (4.63 h). These data suggest that loss of DIO3 function in male mice eliminates the circadian rhythm of lipid metabolism in BAT, independently of clock genes, in addition to its phase shift effect on circadian genes.