Variable Effects of Dietary Selenium in Mice That Spontaneously Develop a Spectrum of Thyroid Autoantibodies

Abstract

Selenium (Se) is a critical element in thyroid function, and variable dietary Se intake influences immunity. Consequently, dietary Se could influence development of thyroid autoimmunity and provide an adjunct to treat autoimmune thyroid dysfunction. Nonobese diabetic (NOD).H2^h4 mice spontaneously develop autoantibodies to thyroglobulin (Tg) and thyroid peroxidase (TPO). This mouse strain expressing a human thyroid-stimulating hormone receptor (TSHR) A-subunit transgene in the thyroid also develops pathogenic TSHR autoantibodies. In this report, we investigated whether dietary Se influences these immune processes. Male and female wild-type and transgenic NOD.H2^h4 mice were maintained on normal-, low-, or high-Se (0.1, 0, or 1.0 mg/kg) rodent diets. After 4 months, Se serum levels were extremely low or significantly increased on 0 or 1.0 mg/kg Se, respectively. Varying Se intake affected Tg antibody (TgAb) levels after 2 (but not 4) months; conversely, TPO antibody (TPOAb) levels were altered by dietary Se after 4 (but not 2) months. These data correspond to the earlier development of TgAb than TPOAb in NOD.H2^h4 mice. In males, TgAb levels were enhanced by high Se and in females by low Se intake. Se intake had no effect on pathogenic TSHR autoantibodies in TSHR transgenic NOD.H2^h4 females. In conclusion, in susceptible NOD.H2^h4 mice, we found no evidence that a higher dietary Se intake ameliorates thyroid autoimmunity by reducing autoantibodies to Tg, TPO, or the TSHR. Instead, our finding that low dietary Se potentiates the development of autoantibodies to Tg and TPO in females is consistent with reports in humans of an increased prevalence of autoimmune thyroiditis in low-Se regions.

In NOD.H2^h4 mice, low dietary Se potentiates the development of thyroid autoantibodies in females, consistent with an increased prevalence of autoimmune thyroiditis in low selenium regions.

Selenium (Se) is a critical element for normal thyroid function, and variability in dietary Se influences immune responses [reviewed in (1–5)]. Consequently, Se intake has the potential to affect thyroid autoimmunity in humans both before disease manifestation and as a possible adjunct to therapy. Serum levels of Se are low in some newly diagnosed patients who have Graves disease (6). Similarly, low Se intake was associated with an increased prevalence of thyroiditis in a large group of Chinese patients (7). In the reverse direction, increased dietary Se was associated with decreased thyroid autoantibody levels in some investigations but was without effect in other studies (8). However, in a recent meta-analysis, increased Se intake reduced autoantibodies to thyroid peroxidase (TPO) for up to 12 months when combined with l-thyroxine (T4) but for only 3 months without l-T4 (9).

In mice, numerous studies have investigated the outcome of variable Se dietary intake on immune responses. For example, nonautoimmune-prone mice (C57BL/6 strain) infected with Listeria and maintained on a Se-deficient diet produced less interferon-γ; tumor necrosis factor-α; and interleukins 6, 10, and 18 than mice of the same strain on a control-Se diet (10). Similarly, production of interferon-γ and interleukin 6 was defective in FVB/N mice on a Se-deficient diet (11). In the nonobese diabetic (NOD).H2^h4 strain in which spontaneous thyroiditis is enhanced by dietary iodine (12–14), Se supplementation increased regulatory T cells and caused a small (but significant) decrease in autoantibodies to thyroglobulin (Tg) (15, 16).

Recently, we developed a mouse strain that spontaneously develops pathogenic antibodies to the thyrotropin receptor (TSHR) (17). This novel TSHR/NOD.H2^h4 strain was generated by transferring the transgene for the human thyroid-stimulating hormone receptor (TSHR) A-subunit targeted to the thyroid from BALB/c mice (18, 19) to nontransgenic NOD.H2^h4 recipients. As we and others have shown, the TSHR A-subunit shed after cleavage of the membrane bound TSHR is the target of the autoimmune response in Graves disease (20–22). Unlike nontransgenic NOD.H2^h4 mice, which require immunization to develop TSHR antibody (TSHRAb), mice of the TSHR/NOD.H2^h4 strain develop pathogenic TSHRAbs spontaneously (17). In addition, transgenic TSHR/NOD.H2^h4 mice develop Tg antibodies (TgAbs) and TPO antibodies (TPOAbs), like their nontransgenic littermates (12–14).

In the current study, we used NOD.H2^h4 mice with and without the TSHR A-subunit transgene to address the question of whether long-term dietary intake of Se influences, on the one hand, the spontaneous development of autoantibodies to Tg and TPO and, on the other hand, pathogenic autoantibodies to the TSHR.

Methods

Mice studied

NOD.H2^h4 mice (originally from The Jackson Laboratory, Bar Harbor, ME) and transgenic TSHR/NOD.H2^h4 mice (17) (which express low levels of the human TSHR A-subunit in the thyroid and thymus) were bred at Cedars-Sinai Medical Center. Mice of the TSHR/NOD.H2^h4 strain have been cryopreserved by the Mutant Mouse Regional Resource Center under the designation NOD.Cg-Tg(TG-TSHR)51.9Smcl/Mmmh (MMRRC:037586-MU). Beginning at 8 weeks of age, all mice were provided with drinking water containing 0.05% sodium iodide (NaI). At the same time and continuing until the end of the study, different groups of NOD.H2^h4 and TSHR/NOD.H2^h4 mice (similar numbers of males and females) were fed custom diets containing various amounts of Se (see later). Blood was drawn 2 months after starting the Se diets together with NaI water. After 4 months on the diets and NaI water, mice were euthanized to harvest blood and thyroid glands. All mouse studies were approved by the Institutional Animal Care and Use Committee at Cedars-Sinai Medical Center and conducted in accordance with mandated standards of humane animal care.

Custom Se diets

The following custom diets were obtained from Envigo (Madison WI): Se deficient (Lo; 0 mg/kg TD.92163), Se control (Con; 0.1 mg/kg TD.96363), and Se enriched (Hi; 1.0 mg/kg TD.120305). These diets had previously been used to investigate the role of Se in preeclampsia in rats and rhabdomyolysis in rabbits (23, 24).

Se content of custom diets and mouse sera

Sera from mice exposed for 4 months to diets with varying Se content were pooled (four to 10 mice) to provide duplicate aliquots of 220 μL per mouse group. We also analyzed ∼100 mg of each diet for Se content. After digestion with concentrated acid (see Supplemental Methods), samples were analyzed by inductively coupled plasma mass spectrometry (Agilent 9000c; Agilent Technologies, Wilmington, DE) with argon plasma. Se data are presented for sera as μg/L and for custom diets as mg/g dry weight.

Autoantibodies to Tg and TPO (TgAb and TPOAb)

Mouse Tg was isolated from murine thyroid glands as previously described (25). Enzyme-linked immunosorbent assay (ELISA) wells (Immulon 4HBX; Thermo Scientific, Rochester, NY) were coated with mouse Tg (1.5 μg/ml) and incubated with test sera (1:100 dilution) in duplicate. Antibody binding was detected with horseradish peroxidase–conjugated goat anti-mouse immunoglobulin G (IgG; catalog no. A3673; Sigma Chemical Co., St. Louis, MO; Table 1), the signal developed with o-phenylenediamine (Sigma Chemical Co.) and the reaction stopped with 20% (volume-to-volume ratio) H₂SO₄. The negative control was serum from 8-week-old NOD.H2^h4 mice on regular water; the positive control was serum from BALB/c mice immunized with mouse Tg and complete Freund adjuvant (26). TgAb values are presented as the optical density (OD) at 490 nm.

Table 1.

Antibodies Used to Detect Mouse IgG Class (or IgG Subclass) Antibodies Binding in ELISA or in Flow Cytometry

Peptide/Protein Target	Antigen Sequence (if Known)	Name of Antibody	Catalog No., Manufacturer	Species Raised in; Monoclonal or Polyclonal	Dilution Used	RRID
Mouse IgG		Horseradish peroxidase-conjugated anti-mouse IgG	A3673, Sigma Chemical Co.	Polyclonal; goat	1:1000	AB_258099
Mouse IgG		Goat anti-mouse IgG (H+L) secondary antibody FITC	A16073, Thermo Fisher	Polyclonal; goat	10 μg/mL	AB_253476
Mouse T reg		Mouse regulatory T cell staining kit #1	88-8111, Affymetrix	Monoclonal; rat	As per kit	AB_2575187
Mouse IgG1		Anti-mouse IgG1 polyclonal antibody, biotin conjugated	10708, Southern Biotech	Polyclonal; goat	1:1000	AB_609714
Mouse IgG2a		Anti-mouse IgG2b: biotin conjugated	10908, Southern Biotech	Polyclonal; goat	1:1000	AB_609729

Abbreviation: RRID, Research Resource Identifier.

TPOAb were measured using Chinese hamster ovary (CHO) cells stably expressing murine TPO (25). Sera (1:50 dilution) were incubated with mouse TPO-CHO cells and binding was detected with goat anti-mouse IgG conjugated to fluoroscein-isothiocyanate (catalog no. A16073; Thermo Fisher, Waltham, MA; Table 1). Cells staining with propidium iodide (1 μg/mL) were excluded from analysis. The negative control for IgG class antibody binding to mouse TPO-CHO cells was serum from young NOD.H2^h4 mice (8 weeks old) on control water. Positive controls were mouse monoclonal antibodies 15 and 64 to human TPO (27), provided to us by Dr. J. Ruf (Marseille, France), which recognize mouse TPO (25, 27). Flow cytometry was performed (10,000 events) using a BD FACScanto II with FACSDiva Software (Becton Dickinson, San Jose, CA) for collection and FloJo v.X.07 software (FloJo, Ashland, OR) for analysis. Data are reported as the geometric mean.

IgG subclasses of TgAb

The distribution of TgAb between subclass IgG2b and IgG1 was measured using a modification of the TgAb ELISA described previously. One set of duplicate mouse Tg coated wells was exposed to biotinylated anti-IgG2b and another set to biotinylated anti-IgG1 (both antibodies from Southern Biotech, Birmingham AL; Table 1). After washing, wells were incubated with streptavidin peroxidase (BD Biosciences, San Jose, CA) and washed again, and the signal developed with o-phenylenediamine and the reaction stopped with 20% (volume-to-volume ratio) H₂SO₄. IgG subclass data are reported for individual mice as the ratio of OD values obtained using anti-IgG2b to the OD values obtained using anti-IgG1. Sera were studied after 4 months on the diet (insufficient serum was available for separate analysis of males and females after 2 months).

Assay for pathogenic TSHR antibodies

TSHR antibodies were measured by thyroid-stimulating hormone (TSH)–binding inhibition (TBI) in 25μL mouse serum using a clinical assay kit (Kronus Inc, Star ID). The data are reported as the percent inhibition of ¹²⁵I-TSH binding to the TSH holoreceptor.

Regulatory T cells in splenocytes

At euthanasia, spleens were aseptically collected, washed and disrupted manually to release mononuclear cells. Red blood cells were lysed with ammonium chloride potassium lysing buffer. Viability determined with Trypan blue (Sigma Chemical Co.) was >90%. Splenocytes (10⁷ cells) were stained with mouse regulatory T cell staining kit 1 (Cleveland, OH) following the manufacturer’s instructions: splenocytes were first incubated with anti-mouse CD4 and anti-mouse CD25; subsequently, the cells were permeabilized and incubated with anti-rat Foxp3 (or isotype control antibody). Stained cells were analyzed with a BD FACScanto II and FACSDiva Software (Becton Dickinson). Cells in the lymphocyte gate were used for analysis. Data are reported as percentage of FoxP3 positive (+ve) CD25 +ve T cells.

Serum T4 and TSH and thyroid histology

T4 levels were measured (10 μL aliquots) by ELISA (mouse/rat T4 ELISA; Calbiotech, El Cajon, CA); data are reported as μg T4/dL. TSH was measured by radioimmunoassay (28) (Dr. S. Refetoff; University of Chicago; fee for service) and reported as mU/L. Thyroid glands were preserved in zinc fixative (BD Pharmingen, San Diego, CA) and paraffin-embedded, and serial sections were stained with hematoxylin and eosin (IDEXX BioResearch Laboratory Animal and Biological Materials Diagnostic Testing, Columbia, MO). The percentage of the thyroid gland infiltrated with lymphocytes (the extent of thyroiditis) was scored “blind” by two researchers (B.B. and S.M.M.) and used to calculate the mean values.

Statistics

Significant differences between responses in different groups were determined by the Mann–Whitney rank sum test or, when normally distributed, by Student t test. Multiple comparisons were made using analysis of variance (ANOVA). Tests were performed using SigmaStat (Jandel Scientific Software, San Rafael, CA).

Results

In NOD.H2^h4 mice, beginning at 8 weeks of age, we assessed the effect of varying Se dietary intake (Con, Lo, and Hi), provided for up to 4 months, on the spontaneous development of autoimmune thyroiditis (TgAb, TPOAb, and lymphocytic infiltration). Also, we examined NOD.H2^h4 mice transgenic for the human TSHR A-subunit that, in addition to TgAb and TPOAb, spontaneously develops TSHR antibodies (17). Dietary iodine supplementation, the standard protocol used in the NOD.H2^h4 mouse model to accelerate the autoimmune process (12–14), remained constant throughout the period of varying dietary Se intake. We studied males and females separately in both wild-type and transgenic NOD.H2^h4 mice because sexual dimorphism is well recognized in autoimmunity [reviewed in (29)], and serum TSH levels are significantly higher in male than in female NOD.H2^h4 (17).

Two critical time points were selected to study the effects of varying Se intake: (1) after 2 months of exposure, when TgAb are detectable in most mice, and (2) after 4 months, when TPOAb are detectable (25) and TgAb levels are significantly higher than after 2 months (30).

Thyroglobulin antibodies

As occurs in NOD.H2^h4 mice fed regular chow, TgAb levels generally increased from the 2-month to the 4-month time point independent of the dietary Se content or the presence or absence of the human TSHR A-subunit transgene (Fig. 1). However, there were significant differences between subgroups and at different times.

Figure 1.

TgAb levels in NOD.H2^h4 mice are affected after 2 (but not 4) months’ exposure to variable Se dietary intake. Males and females respond differently. Rodent diets indicated as Lo, Con, or Hi contained 0.0, 0.1, or 1.0 mg/kg Se, respectively, together with iodide-supplemented water. Nontransgenic NOD.H2^h4 males and females after (A) 2 months and (B) 4 months on the diets Transgenic TSHR/NOD.H2^h4 males and females after (C) 2 months and (D) 4 months on the diets. Data are shown as the mean + standard error of the mean values for TgAb (OD 490 nm in ELISA); the number of mice in each group is given in parentheses. Statistically significant differences: (A) males ^#P = 0.036 (ANOVA); females *P = 0.045 (rank sum test); (C) males: ^#P = 0.029 (rank sum test). The dashed lines indicate the mean + 2 standard deviations of TgAb observed in negative controls (see Methods).

After 2 months on varying Se dietary intake, TgAb levels in NOD.H2^h4 mice were higher in males on Hi-Se vs Con-Se diet (P = 0.036) as opposed to females in which the Lo-Se diet significantly increased TgAb levels (P = 0.045; Fig. 1A). Remarkably, we observed the reverse outcome with the male transgenic TSHR/NOD.H2^h4 mice in that the Lo-Se (not the Hi-Se) diet increased TgAb levels (P = 0.029). Varying Se intake was without effect on the female transgenics (Fig. 1C).

After 4 months on diets with different Se contents, at which time point TgAb levels were generally much higher, no significant differences were observed for any group of NOD.H2^h4 mice, male or female, wild-type (Fig. 1B), or TSHR A-subunit transgenics (Fig. 1D).

Thyroid peroxidase antibodies

In agreement with our previous study (25), TPOAb in nontransgenic and transgenic hTSHR/NOD.H2^h4 mice developed later than TgAb, being very low or absent at the 2-month time point and increasing substantially after 4 months (Fig. 2). Given the lack of TPOAb levels at the 2-month time point, different Se intakes were without effect on this parameter in all four groups studied, namely, male and female nontransgenic (Fig. 2A) and transgenic (Fig. 2C) NOD.H2^h4 mice.

Figure 2.

TPO levels in NOD.H2^h4 female mice are affected after 4 months (but not 2 months) exposure to variable Se diets. Rodent diets indicated as Lo, Con, or Hi contained 0.0, 0.1, or 1.0 mg/kg Se, respectively, together with iodide supplemented water. Nontransgenic NOD.H2^h4 males and females after (A) 2 months and (B) 4 months. Transgenic TSHR/NOD.H2^h4 males and females after (C) 2 months and (D) 4 months. Dashed line: Mean + 2 standard deviation for TPOAb in BALB/c mice after 16 weeks. Data are shown as the mean + standard error of the geometric mean values for TPOAb; the number of mice in each group is given in parentheses. Statistically significant differences: (B) females *P < 0.05 (ANOVA); (D) females *P = 0.035 (t test).

When TPOAb levels were clearly evident at the 4-month time point, different Se intakes had an effect on this parameter, but only in female mice. Wild-type NOD.H2^h4 females fed either a Hi- or Lo-Se diet developed higher TPOAb levels than animals maintained on a Con-Se diet (Fig. 2B). In transgenic hTSHR/NOD.H2^h4 females, only a Hi-Se intake enhanced TPOAb levels (Fig. 2D).

Histological evidence of thyroiditis, the extent of lymphocytic infiltration, studied at euthanasia (the 4-month time point) was variable and was not significantly different between groups (Supplemental Fig. 1 ^{(34.4KB, pdf)} ).

TSHR antibodies

Unlike TgAb and TPOAb, pathogenic TSHRAb do not develop spontaneously in wild-type NOD.H2^h4 mice, but only in this strain with the human TSHR A-subunit transgene expressed in the thyroid. Because TSHRAb detected by ELISA are nonfunctional, we explored the effect of different dietary Se intake on these antibodies using a clinically relevant TBI assay. Functional pathogenic TSHRAb can only be assayed in sera from female NOD.H2^h4 mice because high TSH levels in males produce spurious values in TBI or bioassays (17). After 2 months, TSHRAb were not evident in female wild-type or transgenic NOD.H2^h4 mice (Fig. 3A). At the 4-month time point, TBI positivity was detected in three of seven female TSHR A-subunit transgenics on the Con-Se diet, with similar small numbers positive on the Lo-Se (two of eight) and Hi-Se (two of nine) diets (Fig. 3B).

Figure 3.

(A) TBI in female nontransgenic NOD.H2^h4 and (B) transgenic TSHR/NOD.H2^h4 mice exposed to rodent diet containing Lo, Con, or Hi levels of Se for 2 or 4 months together with iodide-supplemented water. Rodent diets indicated as Lo, Con, or Hi contained 0.0, 0.1, or 1.0 mg/kg Se, respectively. Data are shown for individual mice after exposure for 2 months and 4 months. Dashed line: mean + 2 standard deviation values in nontransgenic mice. The number of mice in each group is given in parentheses.

T4 and body weight

Variation in dietary Se intake had little effect on serum T4 levels, with a small but significant difference only evident after 4 months in male nontransgenic NOD.H2^h4 mice on a Lo-Se diet (P < 0.05; Fig. 4B). Body weights were also lower after 2 months on the Lo-Se diet in male nontransgenics (P < 0.05) and transgenics (P < 0.05), but only in female transgenics (P < 0.001; Supplemental Fig. 2A and 2B ^{(34.4KB, pdf)} ).

Figure 4.

Serum T4 levels after exposure for 2 or 4 months in NOD.H2^h4 mice to rodent diets containing Lo, Con, or Hi Se (0.0, 0.1, or 1.0 mg/kg Se, respectively) together with iodide-supplemented water. Nontransgenic NOD.H2^h4 males and females after (A) 2 months and (B) 4 months. Transgenic TSHR/NOD.H2^h4 males and females after (C) 2 months and (D) 4 months. Data are shown as the mean + standard error of the mean values for T4 (mg/dL); the number of mice in each group is given in parentheses. Statistically significant differences: (B) males *P < 0.05 (ANOVA).

Relationship between thyroid autoantibodies and TSH

We addressed the issue of whether serum TSH levels at an early stage in the development of thyroiditis (2-month exposure) could influence the subsequent autoantibody response at a later stage (4-month exposure). At the 2-month time point, we measured serum TSH levels in the four groups of mice (male and female wild-type and transgenic NOD.H2^h4 strains), each group fed Con-, Lo-, or Hi-Se diets. The high cost (fee for service) limited our sampling to three mice in each category (36 samples). Because varying dietary Se intake revealed no suggestive effect on TSH (not shown), we pooled the data for mice on all three diets. TSH levels in males, whether wild-type or transgenic, were significantly higher than in females (Fig. 5).

Figure 5.

Serum TSH levels in NOD.H2^h4 and TSHR/NOD.H2^h4 transgenic males and females mice. For each group, the data are pooled for three mice each on Con-, Lo-, and Hi-Se diets (n = 9) for 4 months. Statistically significant differences *P < 0.05 (ANOVA).

Regulatory T cells and TgAb IgG subclasses

To provide potential insight into mechanisms associated with the effects of varying Se intake on thyroid autoimmunity, we examined two other parameters in wild-type and transgenic NOD.H2^h4 strains mice after 4 months on different Se dietary exposure. First, regulatory T cells were quantified by flow cytometry as splenocytes positive for the cell surface markers CD4 and CD25 (31) and subsequently for the intracellular marker FoxP3 (32). No differences were observed in the percentage of FoxP3+ CD25+ T cells in nontransgenic NOD.H2^h4 males or females on varying Se intake (Fig. 6A) or in male TSHR/NOD.H2^h4 mice (Fig. 6B, left panel). However, in female TSHR/NOD.H2^h4 mice, the percentage of FoxP3+ CD25+ T cells was significantly lower on a Lo-Se diet relative to the Con-Se diet (Fig. 6B, right panel; P < 0.05, ANOVA). Of interest, regulatory T cells levels were significantly higher in female than male wild-type, but not transgenic, NOD.H2^h4 mice when data were pooled for animals on Con-, Lo-, and Hi-Se diets (Fig. 6C vs 6D).

Figure 6.

Regulatory T cells in (A) nontransgenic NOD.H2^h4 and (B) transgenic TSHR/NOD.H2^h4 mice on rodent diet containing Lo, Con, or Hi levels of Se (together with iodide-supplemented water) for 4 months. Regulatory T cells were quantified by flow cytometry to identify FoxP3 +ve CD25 +ve splenic T cells. Total numbers of regulatory T cells in (C) nontransgenic NOD.H2^h4 vs (D) transgenic TSHR/NOD.H2^h4 mice [data pooled from mice on different diets in panels (A) and (B)]. Statistically significant differences: (B) *P < 0.05 (ANOVA); (C) *P = 0.004 (t test).

Second, we examined the distribution of TgAb among the IgG subclasses. In mice, antibodies of subclass IgG2b reflect T-helper type 1 cells and TgAb of subclass IgG1 reflect cytokines from T-helper type 2 cells [e.g., (33, 34)]. No significant differences were observed for the TgAb IgG2a:IgG1 ratio in nontransgenic or transgenic TSHR/NOD.H2^h4 mice on varying Se intake (Supplemental Fig. 3 ^{(34.4KB, pdf)} ). These data do not indicate a switch between cytokines from T-helper type 1 or type 2 cells.

Se in mouse sera and custom diets

Se was measured in nontransgenic and transgenic NOD.H2^h4 mice exposed for 4 months to Lo-, Con-, and Hi-Se diets. In males and females of both strains on the Lo-Se diet, serum Se levels were extremely low (at the limits of assay detection) and substantially higher in mice on either Con- or Hi-Se diet (Fig. 7A and 7B). Combining the values for all mice in each dietary group, serum Se levels were slightly but significantly higher in mice on the Hi-Se diet than on the Con-Se diet (Fig. 7C). The Se contents of the custom diets were also tested (Fig. 7D). It is notable that the magnitude of the Se increase in sera from mice on the Hi diet relative to the Con diet was disproportionately lower (nearly 10-fold) than the Se content of the Hi vs the Con diet (25% in serum vs 224% in the diet).

Figure 7.

Se content of mouse sera and custom diets. (A–C) Se was analyzed (by inductively coupled plasma mass spectrometry) in sera from mice exposed for 4 months to custom Se diets containing Lo, Con, or Hi levels (together with iodide supplemented water). (A) Nontransgenic NOD.H2^h4 males and females and (B) transgenic TSHR/NOD.H2^h4 males and females; Se data are shown as μg/L [mean + standard deviation (SD) for duplicate aliquots]. (C) Combined Se values for all mice on the three custom diets (mg/L, mean + standard error of the mean, n = 8 measurements). *P = 0.029, Rank sum tests. (D) Se content of custom diets (ng/g diet sample, mean + SD of duplicates). The dashed lines in (A–C) represent the mean + 2 SDs of Se values in all mice on Lo Se.

Discussion

NOD.H2^h4 mice spontaneously develop autoantibodies to Tg and TPO (12–14), whereas the same strain transgenic for the human TSHR A-subunit targeted to the thyroid, in addition, develop pathogenic autoantibodies to the TSHR (17). We asked the question whether long-term dietary Se intake influences the development of these autoantibodies in both mouse strains. Male and female mice, 8 weeks old, were maintained for 4 months on rodent pellets with a Con-, Lo-, or Hi-Se content (0.1, 0.0, or 1.0 ppm Se). As noted previously (17) and confirmed in this study, serum TSH levels were far higher in male than in female NOD.H2^h4 and TSHR transgenic NOD.H2^h4 mice, independent of dietary Se intake. Sexual dimorphism is well recognized in autoimmunity, due (for example) to the effects of sex hormones and sex specific gut microbiota [reviewed in (29)]. Therefore, we were not surprised to observe some differences between the responses of male and female NOD.H2^h4 mice on variable Se intake.

We observed three major effects of dietary Se. First, changes occurred earlier for TgAb (after 2 months) than for TPOAb (4-month exposure), consistent with the much earlier spontaneous appearance of TgAb than of TPOAb in NOD.H2^h4 mice (25). Second, the number of TSHR A-subunit transgenic female mice that developed pathogenic TSHR antibodies (measured by TBI) was unchanged by variable Se dietary intake. Third, there was no effect of dietary Se intake on the proportion of regulatory T cells (FoxP3+, CD25+) in wild-type NOD.H2^h4 and male transgenic mice. Only in transgenic female NOD.H2^h4 mice did Lo-Se diets reduce the proportion of regulatory T cells cells.

Another point to be noted is a difference between the current study and previous reports from our laboratory on TgAb levels. After the 2-month exposure, TgAb levels were comparable in wild-type NOD.H2^h4 and transgenic TSHR/NOD.H2^h4 mice maintained on regular chow (17, 35) and the Con-Se diet (present study). In contrast, we now find that variable dietary Se exposes significant differences in TgAb levels between wild-type and transgenic strains at this early time point. These differences are no longer evident later (4-month time point) when the TgAb response is more robust (30). The data suggest that human TSHR A-subunit protein expression in the mouse thyroid influences the effect of dietary Se on TgAb generation.

Is there any evidence for an effect of the TSHR A-subunit transgene on the immune response? The low-expressor human TSHR A-subunit transgene in NOD.H2^h4 mice used in this study is located on chromosome 1 (approximately 33 Mb) (36). This location excludes genes that control self-tolerance by their effects on thymic expression of tissue restricted antigens (37–40), genes encoding the TSHR and Tg (chromosomes 12 and 15), and genes encoding immune molecules associated with susceptibility to thyroid autoimmunity in humans (41–44). Overall, this information does not suggest any effect of the transgene on the development of thyroid autoantibodies in transgenic NOD.H2^h4 mice.

Turning to other parameters studied, serum T4 was significantly reduced (unrelated to TSH levels) in male wild-type NOD.H2^h4 mice on Lo-Se diet. Varying Se intake had no effect in TSHR A-subunit NOD.H2^h4 mice. Body weight was significantly reduced in some NOD.H2^h4 mice, both wild-type and transgenic, particularly in males on the Lo-Se diet. The latter findings parallel observations in humans of weight loss in men with a Lo-Se dietary intake (45).

Levels of serum Se in wild-type NOD.H2^h4 and transgenic TSHR/NOD.H2^h4 mice, exposed for 4 months to custom diets, provide insight into some of our findings. The most marked changes were the reduced serum Se levels in all mice on Lo-Se diet. To our knowledge, the effects of Lo Se on thyroid autoimmunity in mice have not previously been studied. We observed that Lo-Se intake was associated with increased TgAb and TPOAb in female NOD.H2^h4 mice and with low body weight in both nontransgenic and transgenic TSHR/NOD.H2^h4 mice.

Turning to Hi-Se diet, in both sexes, serum Se levels were raised by about 25%, comparable with the increases previously observed in male wild-type NOD.H2^h4 exposed to sodium-selenite-supplemented drinking water (15, 16). The increased magnitude of these serum Se levels is less than the much higher Se levels in the diet or sodium-selenate water. The Hi-Se diet (1 mg/kg) did not exceed the recommended maximum for animal experimentation (2 mg/kg) (46). However, on Hi-Se intake, the Se content of red blood cells is a better index of Se status of an animal than plasma (or serum) (46).

How do our immunologic data compare with other findings in mice? We did not observe the small decrease in TgAb levels reported by Wang and colleagues (16) in male wild-type NOD.H2^h4 mice whose drinking water was supplemented with sodium selenite for similar time intervals as in our study. In addition, contrary to a previous report, and despite a comparable increase in serum Se, we did not observe that regulatory T cells were increased by Hi-Se intake in NOD.H2^h4 mice (15). However, the experimental protocol used in the earlier studies differed from that in our study: mice were younger than ours at the start (4 weeks vs 8 weeks), the iodide concentration was lower than in our study (0.005% vs 0.05%), and Se supplementation was added after 8 weeks on iodide (15, 16). Consequently, in the earlier studies, the effects of Se were tested in established and ongoing thyroid autoimmunity. In contrast, we investigated the effects of variable Se intake on the development of thyroid autoantibodies and other parameters in NOD.H2^h4 mice.

Compared with studies in humans, our data show similarities and differences which may be related (at least in part) to gender differences as well as the time of appearance of thyroid autoantibodies. The increase in TgAb at the early time interval in NOD.H2^h4 females on a Lo diet is consistent with data from a large study of Chinese patients (>6000 individuals) that showed increased thyroiditis in patients in from an area with a Lo- vs Con-Se intake (7). However, our observation that the spontaneous development of pathogenic TSHR antibodies in transgenic female mice is unaffected by variable Se dietary intake differs from the report of low serum Se levels in some (but not all) newly diagnosed patients who have Graves disease (6). Finally, the absence of an effect of high Se levels on thyroid autoantibodies differs from studies in humans showing a reduction in thyroid autoantibodies (particularly TPOAb) in individuals on diets supplemented with Se [e.g., (8)].

In conclusion, using autoimmune-prone NOD.H2h4 mice exposed to variable Se dietary intake, we found no evidence that higher Se intake ameliorates thyroid autoimmunity. On the other hand, our data support the findings in humans that Lo-Se intake may potentiate the development of thyroid autoimmunity, at least in females.