Oxidative Stress-Induced Sirtuin1 Downregulation Correlates to HIF-1α, GLUT-1, and VEGF-A Upregulation in Th1 Autoimmune Hashimoto’s Thyroiditis

In Hashimoto’s thyroiditis (HT), oxidative stress (OS) is driven by Th1 cytokines’ response interfering with the normal function of thyrocytes. OS results from an imbalance between an excessive production of reactive oxygen species (ROS) and a lowering of antioxidant production. Moreover, OS has been shown to inhibit Sirtuin 1 (SIRT1), which is able to prevent hypoxia-inducible factor (HIF)-1α stabilization. The aims of this study were to determine the involvement of NADPH-oxidases (NOX), SIRT1, and HIF-1α in HT pathophysiology as well as the status of antioxidant proteins such as peroxiredoxin 1 (PRDX1), catalase, and superoxide dismutase 1 (SOD1). The protein expressions of NOX2, NOX4, antioxidant enzymes, SIRT1, and HIF-1α, as well as glucose transporter-1 (GLUT-1) and vascular endothelial growth factor A (VEGF-A), were analyzed by Western blot in primary cultures of human thyrocytes that were or were not incubated with Th1 cytokines. The same proteins were also analyzed by immunohistochemistry in thyroid samples from control and HT patients. In human thyrocytes incubated with Th1 cytokines, NOX4 expression was increased whereas antioxidants, such as PRDX1, catalase, and SOD1, were reduced. Th1 cytokines also induced a significant decrease of SIRT1 protein expression associated with an upregulation of HIF-1α, GLUT-1, and VEGF-A proteins. With the exception of PRDX1 and SOD1, similar results were obtained in HT thyroids. OS due to an increase of ROS produced by NOX4 and a loss of antioxidant defenses (PRDX1, catalase, SOD1) correlates to a reduction of SIRT1 and an upregulation of HIF 1α, GLUT-1, and VEGF-A. Our study placed SIRT1 as a key regulator of OS and we, therefore, believe it could be considered as a potential therapeutic target in HT.


Introduction
Hashimoto's thyroiditis (HT) and Graves disease are autoimmune thyroid diseases, which comprise around 30% of all autoimmune conditions. HT is the principal cause of This study aimed to verify whether SIRT1 is involved in the modulation of oxidative stress and angiogenesis occurring in Hashimoto's disease. We performed in vitro experiment on human thyroid cells in the presence or absence of Th1 cytokines and on ex vivo controls as well as on HT human thyroid samples. We analyzed (1) the expression of NOX2 and NOX4 as sources of ROS, (2) the expression of antioxidants PRDX1, catalase, and SOD1, (3) the expression of SIRT1 protein in the context of the HT oxidative stress, and (4) the expression of HIF-1α and its targets, GLUT-1 and VEGF-A.

Th1 Cytokines Increase ROS Production and NOX4 Protein Expression in Human Thyroid Cells, without Changing NOX2 Protein Expression
ROS were detected by the 2′,7'-dichlorodihydrofluorescein diacetate (CM-H2DCFDA) method in human thyrocytes treated with IL-1α and IFNγ to mimic Hashimoto's thyroiditis (HT) and compared to non-treated cells. ROS were lightly marked by green fluorescence in the cytoplasm of untreated control cells ( Figure 1A). In Th1 cytokine-treated thyrocytes, ROS expression was highly increased ( Figure 1B). A significant rise of ROS was confirmed by the quantification of fluorescence normalized to the number of nuclei ( Figure 1C). ROS excess could be due to an increase of NOX4 protein known to generate H2O2 by 90% and superoxide by 10%. Th1 cytokines induced a significant increase of NOX4 protein expression measured by Western blotting ( Figure 1D). However, no change was observed for NOX2 protein expression ( Figure 1E). Figure 1. ROS production and NOX4 protein expression increased in primary cultures of human thyrocytes treated with Th1 cytokines, mimicking Hashimoto's thyroiditis, while NOX2 protein expression did not change. Human thyrocytes were treated for 24 h with Il-1α and IFNγ. Cells were immersed in a 1-µ M solution of CM-H2DCFDA for 30 min. Nuclei were stained by Hoescht. Images shown are representative of each condition. The picture on the left shows non-treated control thyrocytes (A) and Th1 cytokines-treated thyrocytes are displayed on the right (B). Scale bar = 10 µ m. ROS levels were increased in thyrocytes treated with Th1 cytokines compared to controls. Quantification of ROS expression was analyzed using ImageJ software (C). Fluorescence was quantified and normalized to the number of nuclei. Results are expressed as means ± standard error of the mean (SEM) from four experiments (n = 4). ** p < 0.01 compared to control. NOX4 (D) and NOX2 (E) protein expressions were quantified by Western blot technique in human thyro- Figure 1. ROS production and NOX4 protein expression increased in primary cultures of human thyrocytes treated with Th1 cytokines, mimicking Hashimoto's thyroiditis, while NOX2 protein expression did not change. Human thyrocytes were treated for 24 h with Il-1α and IFNγ. Cells were immersed in a 1-µM solution of CM-H2DCFDA for 30 min. Nuclei were stained by Hoescht. Images shown are representative of each condition. The picture on the left shows non-treated control thyrocytes (A) and Th1 cytokines-treated thyrocytes are displayed on the right (B). Scale bar = 10 µm. ROS levels were increased in thyrocytes treated with Th1 cytokines compared to controls. Quantification of ROS expression was analyzed using ImageJ software (C). Fluorescence was quantified and normalized to the number of nuclei. Results are expressed as means ± standard error of the mean (SEM) from four experiments (n = 4). ** p < 0.01 compared to control. NOX4 (D) and NOX2 (E) protein expressions were quantified by Western blot technique in human thyrocytes treated or not with Th1 cytokines. Th1 cytokines induced a significant increase of NOX4 expression while NOX2 did not change. Densitometric values were normalized against the β-actin level. Results are expressed as means ± SEM from three (NOX4) and five (NOX2) experiments (n = 3-5) at least in duplicate. * p < 0.05 compared to control. A representative blot is shown.

NOX4 Protein Expression Is Increased in HT Thyroid Samples without Any Change of NOX2 Protein Expression
Thyroid samples from both control and HT patients displayed considerable heterogeneity in expression of NOX2 and NOX4 quantified by Western blot. We noticed, however, a significant increase of NOX4 (Figure 2A), whereas NOX2 was unmodified as compared to controls ( Figure 2B).  (Figure 3A,B). In thyroids of HT patients, PRDX1 protein expression was unchanged ( Figure   Figure 2. NOX4 protein expression was increased in Hashimoto's thyroiditis (HT) thyroid samples while NOX2 did not change. Western blots performed on control and Hashimoto's thyroiditis (HT) thyroid samples revealed a significant increase of NOX4 protein in HT (A), but no change in NOX2 (B). Densitometric values were normalized against Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) level. Results are expressed as means ± SEM from five or six individual samples (n = 5-6). * p < 0.05 compared to control. Representative blots are shown. Immunohistochemistry on paranodular tissue obtained from multinodular goiter patients designated as controls (C,F) and thyroid tissues from HT patients (D,E,G,H). The pictures illustrate representative tissue samples of both conditions. In control thyroids, the expression of NOX4 (C) and NOX2 proteins (F) was more intense in active follicles (*) than in hypofunctional follicles ( †). In HT non-inflammatory zone, hypofunctional follicles ( †) lightly expressed NOX4 (D) and NOX2 (G). In contrast, thyrocytes of altered active-like follicles ($) expressed strongly NOX4 in their cytoplasm (D) while NOX2 was moderately expressed (G). In HT inflammatory zones, the numerous altered active-like follicles ($) expressed elevated levels of NOX4 in the cytoplasm of their thyrocytes (E). NOX2 was highly expressed in inflammatory cells (H). As previously shown by Gérard et al. (2002), a normal thyroid gland displays considerable morphological and functional heterogeneity among follicles, with hypofunctional follicles ( †) lined by flat thyrocytes and active follicles (*) recognized by their small size and their thickened epithelial layer [23]. In HT thyroids, the follicular heterogeneity was even more accentuated with hypofunctional follicles ( †) and altered active-like follicles ($), the latter presenting the same morphological aspects as those that were described by Marique et al. (2014) as having lost their capacity for normal hormonal synthesis and showing a high expression of 4-Hydroxynonenal (HNE), a marker of OS [8]. We confirmed in our samples that active-like follicles lost their capacity for T4 synthesis and strongly expressed HNE (Supplemental Figure S1).
In paranodular tissues of multinodular goiter considered as controls, NOX4 protein expression was low in hypofunctional follicles while the active follicles revealed greater staining ( Figure 2C). In HT samples, NOX4 staining was increased in the active-like follicles ( Figure 2D,E). NOX2 immunostaining was observed in active follicles from controls ( Figure 2F) and in active-like follicles from HT glands ( Figure 2G) but it was mainly increased in the numerous inflammatory cells infiltrating the thyroids of HT patients ( Figure 2H).

The Expression of PRDX1 and Catalase Is Decreased by Th1 Cytokines in Human Thyroid
Cells and the Expression of Catalase Is Lowered in HT Thyroid Samples PRDX1 and catalase protein expression quantified by Western blotting were significantly decreased in primary cultures of human thyrocytes treated with Th1 cytokines ( Figure 3A,B). In thyroids of HT patients, PRDX1 protein expression was unchanged ( Figure 3C) but catalase expression was significantly decreased as compared to controls ( Figure 3D). In control thyroids, PRDX1 ( Figure 3E) and catalase ( Figure 3H) were detected in both hypofunctional and active follicles although the latter showed a higher degree of immunostaining. In HT non-inflammatory zone, the expression of PRDX1 ( Figure 3F) and catalase ( Figure 3I) was reduced compared to controls. In HT inflammatory zones, the expression of PRDX1 ( Figure 3G) and catalase ( Figure 3J) was high in active-like follicles in which the OS is maximal.  Results are expressed as means ± SEM from three experiments (n = 3) at least in duplicate. In HT thyroid samples, PRDX1 protein expression (C) was unchanged as compared to controls, while catalase protein expression (D) was significantly reduced. Densitometric values were normalized against GAPDH level. Values are expressed as means ± SEM from five or six individual samples (n = 5-6). * p < 0.05, compared to controls. Representative blots are shown. Immunohistochemically stained sections of paranodular tissue from multinodular goiter patients designated as controls (E,H) and HT thyroid tissues (F,G,I,J). In control glands, Figure 3. Antioxidant proteins, PRDX1, and catalase, were decreased by Th1 cytokines and catalase was significantly decreased in thyroid samples. PRDX1 and catalase protein expression were quantified by Western blot technique. Primary cultures of human thyrocytes treated with Th1 cytokines had significantly reduced PRDX1 (A) and catalase (B) expression. Densitometric values were normalized against the β-actin level. Results are expressed as means ± SEM from three experiments (n = 3) at least in duplicate. In HT thyroid samples, PRDX1 protein expression (C) was unchanged as compared to controls, while catalase protein expression (D) was significantly reduced. Densitometric values were normalized against GAPDH level. Values are expressed as means ± SEM from five or six individual samples (n = 5-6). * p < 0.05, compared to controls. Representative blots are shown. Immunohistochemically stained sections of paranodular tissue from multinodular goiter patients designated as controls (E,H) and HT thyroid tissues (F,G,I,J). In control glands, hypofunctional follicles ( †) expressed low levels of PRDX1 and active follicles (*) showed an increased PRDX1 (E) and catalase (H) marking. In HT thyroids, hypofunctional ( †) and altered active-like ($) follicles in noninflammatory zone presented, respectively, no and low staining of PRDX1 (F) and of catalase (I). In inflammatory zones, active-like follicles ($) strongly expressed PRDX1 (G) and catalase (J). Illustrated images are representative of both conditions. (E-J) Scale bar = 50 µm. (E1-J2) Scale bar = 20 µm.
In control thyroids, PRDX1 ( Figure 3E) and catalase ( Figure 3H) were detected in both hypofunctional and active follicles although the latter showed a higher degree of immunostaining. In HT non-inflammatory zone, the expression of PRDX1 ( Figure 3F) and catalase ( Figure 3I) was reduced compared to controls. In HT inflammatory zones, the expression of PRDX1 ( Figure 3G) and catalase ( Figure 3J) was high in active-like follicles in which the OS is maximal.

The Expression of SOD1 Is Decreased by Th1 Cytokines in Human Thyroid Cells and Is Unchanged in HT Thyroid Samples
Western blotting quantification revealed that SOD1 protein expression was significantly diminished by Th1 cytokines' treatment ( Figure 4A) but SOD1 was not modified in HT thyroid samples compared to controls ( Figure 4B). In control, as well as in HT thyroids, SOD1 protein expression was detected by immunohistochemistry (IHC) in hypofunctional follicles and in active or active-like follicles in which the staining was intense ( Figure 4C-E).
Western blotting quantification revealed that SOD1 protein expression was signifi cantly diminished by Th1 cytokines' treatment ( Figure 4A) but SOD1 was not modified in HT thyroid samples compared to controls ( Figure 4B). In control, as well as in HT thy roids, SOD1 protein expression was detected by immunohistochemistry (IHC) in hypo functional follicles and in active or active-like follicles in which the staining was intens ( Figure 4C-E).  2.5. In Th1 Cytokines-Treated Human Thyroid Cells, as Well as in HT Thyroid Samples, the Expression of SIRT1 Is Decreased While the Expression of HIF-1α Is Increased A significant decrease of SIRT1 protein expression was observed in primary cell cultures of human thyrocytes treated with Th1 cytokines ( Figure 5A). Thyrocytes treated with IL-1α and IFNγ showed a highly increased HIF-1α protein expression compared to non-treated cells ( Figure 5B). In whole thyroid samples of HT patients, Western blotting quantification revealed reduced SIRT1 protein expression ( Figure 5C) and upregulated HIF-1α protein expression ( Figure 5D) compared to controls. non-treated cells ( Figure 5B). In whole thyroid samples of HT patients, Western blotting quantification revealed reduced SIRT1 protein expression ( Figure 5C) and upregulated HIF-1α protein expression ( Figure 5D) compared to controls.
Immunostaining showed that SIRT1 protein expression was high in active follicles of control glands ( Figure 5E) and slightly decreased in active-like follicles from HT thyroids ( Figure 5F,G). The staining of HIF-1α was higher in HT thyroids ( Figure 5I,J) than in controls ( Figure 5H).

Figure 5.
In Th1-HT context, SIRT1 was downregulated while HIF-α was upregulated. Addition of Th1 cytokines to primary cultures of human thyrocytes induced a significant reduction of SIRT1 (A) and an increase of HIF-1α protein (B). Densitometric values were normalized against the βactin level. Results are expressed as means ± SEM from seven (SIRT1) and four (HIF-1α) experiments (n = 4-7) at least in duplicate. SIRT1 protein expression (C) was significantly decreased in HT patients while HIF-1α (D) was increased. Densitometric values were normalized against the GAPDH level. Results are expressed as means ± SEM from five or six individual samples (n = 5-6). * p < 0.05, *** p < 0.005 compared to controls. Representative blots are shown. Paranodular tissue Figure 5. In Th1-HT context, SIRT1 was downregulated while HIF-α was upregulated. Addition of Th1 cytokines to primary cultures of human thyrocytes induced a significant reduction of SIRT1 (A) and an increase of HIF-1α protein (B). Densitometric values were normalized against the β-actin level. Results are expressed as means ± SEM from seven (SIRT1) and four (HIF-1α) experiments (n = 4-7) at least in duplicate. SIRT1 protein expression (C) was significantly decreased in HT patients while HIF-1α (D) was increased. Densitometric values were normalized against the GAPDH level. Results are expressed as means ± SEM from five or six individual samples (n = 5-6). * p < 0.05, *** p < 0.005 compared to controls. Representative blots are shown. Paranodular tissue from multinodular goiter patients designated as controls (E,H) and HT thyroid samples (F,G,I,J) were used to perform IHC with SIRT1 and HIF-1α specific antibodies. In controls (E), SIRT1 was detected in thyrocytes of hypofunctional follicles ( †) and more intensively in active follicles (*). In HT (F,G), staining was less intense in altered active-like follicles ($). In control thyroids (H), HIF-1α staining was observed in the cytoplasm of the thyrocytes ( † and *). In HT thyroids (I,J), non-inflammatory and inflammatory zones displayed hypofunctional ( †) and Immunostaining showed that SIRT1 protein expression was high in active follicles of control glands ( Figure 5E) and slightly decreased in active-like follicles from HT thyroids ( Figure 5F,G). The staining of HIF-1α was higher in HT thyroids ( Figure 5I,J) than in controls ( Figure 5H).

The Expression of GLUT-1 and VEGF-A Is Increased by Th1 Cytokines in Primary Cultures of Human Thyrocytes and in HT Thyroid Samples
Thyrocytes treated with Th1 cytokines had higher GLUT-1 and VEGF-A protein expressions than non-treated thyrocytes ( Figure 6A,B). Western blotting quantification revealed that in HT patients, GLUT-1 and VEGF-A protein expression was also significantly increased compared to controls ( Figure 6C,D).

The Expression of GLUT-1 and VEGF-A Is Increased by Th1 Cytokines in Primary Cultures of Human Thyrocytes and in HT Thyroid Samples
Thyrocytes treated with Th1 cytokines had higher GLUT-1 and VEGF-A protein expressions than non-treated thyrocytes ( Figure 6A,B). Western blotting quantification revealed that in HT patients, GLUT-1 and VEGF-A protein expression was also significantly increased compared to controls ( Figure 6C,D).
Compared to controls ( Figure 6E), a higher immunohistochemical staining of GLUT-1 was observed in HT thyroids ( Figure 6F,G). Similar results were observed for VEGF-A ( Figure 6 H-J). Once again, active and active-like follicles showed more intense cytoplasmic staining than hypofunctional follicles.  GLUT-1 was highly increased in HT patients compared to controls (C). A significant increase of VEGF-A protein expression was also observed in HT thyroids compared to controls (D). Densitometric values were normalized against the GAPDH level. Results are expressed as means ± SEM from five or six individual samples (n = 5-6). * p < 0.05 compared to controls. Representative blots are shown. Faint GLUT-1 immunolabelling was detected in control thyroids (E) and the staining was more pronounced in active follicles (*). In non-inflammatory zones of HT thyroids, thyrocytes of altered active-like follicles ($) and hypofunctional follicles ( †) showed considerable, diffuse, and cytoplasmic staining of GLUT-1 (F). In inflammatory zones, GLUT-1 expression was still more pronounced in active-like follicles ($) (G). In control thyroids (H), VEGF-A staining was low in hypofunctional follicles ( †) and was moderate in active follicles (*). Both hypofunctional ( †) and altered active- Compared to controls ( Figure 6E), a higher immunohistochemical staining of GLUT-1 was observed in HT thyroids ( Figure 6F,G). Similar results were observed for VEGF-A ( Figure 6H-J). Once again, active and active-like follicles showed more intense cytoplasmic staining than hypofunctional follicles.

Discussion
Our results demonstrated that, in primary cultures of human thyroid cells, Th1 cytokines induce an increase in NOX4 known to generate H 2 O 2 and superoxide as well as a decrease in antioxidant proteins, PRDX1, catalase, and SOD1, responsible for OS. This is associated with a downregulation of SIRT1 protein expression and an upregulation of HIF-1α, GLUT-1, and VEGF-A.
In Hashimoto's thyroid samples, the proteins studied followed the same trends. However, follicular heterogeneity was apparent, even in normal thyroids, including hypofunctional follicles and active follicles, with the latter being characterized by their small size and their thick thyrocytes [23]. The intraglandular heterogeneity was even more accentuated in HT with hypofunctional follicles and active-like follicles that, as described by Marique et al. [8], may be considered as altered follicles having lost their function and being the targets of OS, as shown by HNE immunostaining.
Together, with the large follicular heterogeneity, there were also regional differences in the inflammation within the glands of HT patients. Of course, this is a limitation of our study in which whole glands were homogenized to perform Western blots. However, except for PRDX1 and SOD1, we observed significant differences between controls and HT patients.
Our study shows that NOX4 was increased in human thyroid cells treated with Th1 cytokines, as well as in HT thyroid samples, while NOX2 expression was not modified. This is in contrast to the results reported by Colin et al. (2014) [10] and this could be explained by the fact that we used primary cultures of human thyroid cells rather than PCCL3 cell lines. Colin et al. (2014) [10] analyzed the NOX2 protein level by immunohistochemistry on two HT patients without making the distinction between the different types of follicles. As illustrated in this paper, these proteins revealed considerable heterogeneity in their expression. NOX4 was mainly increased in active-like follicles, whereas NOX2 was strongly expressed in inflammatory cells. NOX4 expression was shown to be regulated in vascular cells by SIRT1 [24,25], which also reduced OS by increasing antioxidant enzymes such as SOD1 and catalase [26][27][28].
Our results showed a decrease of SIRT1 in Th1-treated human thyroid cells and in Hashimoto's thyroids, suggesting that a significant interaction between ROS and SIRT1 exists in HT. H 2 O 2 could hold a key role in this interaction between OS and SIRT1. Indeed, this ROS could be produced in Th1 cytokines-treated human thyrocytes and in HT by a decrease of Cav-1 [8] or PPARγ [9] and also by a NOX4 overexpression [11]. However, H 2 O 2 is most likely not the only ROS implicated and ROS are not the only explanation for a SIRT1 decrease. Several authors showed that post-translational modifications, inflammatory cytokines, and miRNA could also explain SIRT1 protein decrease in an OS environment [18,29]. SIRT1 is also known to interact with HIF-1α [30] as it inhibits its cellular accumulation via deacetylation, leading to an impairment of the resulting signaling pathways [31][32][33]. In consequence, hypoxia-induced suppression of SIRT1 may be responsible for an improvement of HIF-1α signaling [31]. Although the exact mechanism remained to be elucidated, our results showed a correlation between SIRT1 downregulation and HIF-1α upregulation in Th1 cytokines-treated human thyroid cells and in HT patients.
In normoxic conditions, HIF-1α has a cytoplasmic localization and is rapidly degraded by the proteasomes before being able to act in its role as transcription factor in the nucleus. Under hypoxic conditions, HIF-1α is stabilized and translocated to the nucleus where HIF-1α/HIF-1β binding occurs and gene transcription can be initiated [34]. Hypoxia can stabilize HIF-1α but nuclear factor (NF)-κB factors and cytokines can also lead to a HIF-1α signaling [35,36]. Consequently, HIF-1α appears to be overexpressed in many malignant tumor types, whereas in normal tissues, lower protein levels were observed [34,[37][38][39].
Furthermore, it is known that HIF-1α acts as a GLUT-1 transcription factor [14,40]. Indeed, aerobic glycolysis is another well-known and major process observed in cancer metabolism [14,41] but our work also highlighted a similar upregulation in HT tissues. The upregulation of VEGF-A that we observed in Th1 cytokines-treated human thyrocytes and in HT thyroid samples could also be induced by HIF-1α. Indeed, VEGF is another well-known target of HIF-1α in thyroid cells [15,16] and it is known to be increased in HT [42] and in thyroid cancers [17].
To date, it is increasingly accepted that thyroid cancers develop in patients with autoimmune thyroid disease [43][44][45] and preferentially in patients with HT [46]. Our results support this observation by showing that, as in papillary thyroid carcinoma (PTC), NOX4, HIF-1α, GLUT-1, and VEGF-A are increased in Th1 cytokines-treated human thyroid cells and in Hashimoto's thyroid samples. The similar expression pattern between HT and PTC may suggest that the increased metabolism associated with a disturbed redox balance in HT thyrocytes could potentially be a first step toward tumorigenesis.
In conclusion, as illustrated in Figure 7, we propose a pathogenic pathway for HT development centered on ROS overproduction and SIRT1 downregulation. Intracellular increase of ROS could result from an upregulation of NOX4 induced by Th1 cytokines. Our results also suggest that the element linking the OS and the HIF-1α/GLUT-1 and HIF-1α/VEGF-A pathways may involve SIRT1. An OS-induced downexpression of SIRT1 may lead to increased HIF-1α activity followed by overexpression of GLUT-1 and VEGF-A. Moreover, the diminished SIRT1 expression induced a reduction of antioxidant defenses due to PRDX1, catalase, and SOD1, further aggravating OS. Interestingly, in accordance with our observations, it was shown that SIRT1 regulates NOX4 expression [24,25], which is a protein considered essential for OS in thyrocytes. Finally, SIRT1, as an antioxidative stress molecule, could be considered as an optimal therapeutic target for HT.
Furthermore, it is known that HIF-1α acts as a GLUT-1 transcription factor [1 Indeed, aerobic glycolysis is another well-known and major process observed in c metabolism [14,41] but our work also highlighted a similar upregulation in HT ti The upregulation of VEGF-A that we observed in Th1 cytokines-treated human thyr and in HT thyroid samples could also be induced by HIF-1α. Indeed, VEGF is an well-known target of HIF-1α in thyroid cells [15,16] and it is known to be increased [42] and in thyroid cancers [17].
To date, it is increasingly accepted that thyroid cancers develop in patients wi toimmune thyroid disease [43][44][45] and preferentially in patients with HT [46]. Our r support this observation by showing that, as in papillary thyroid carcinoma (PTC), N HIF-1α, GLUT-1, and VEGF-A are increased in Th1 cytokines-treated human thyroi and in Hashimoto's thyroid samples. The similar expression pattern between HT an may suggest that the increased metabolism associated with a disturbed redox bala HT thyrocytes could potentially be a first step toward tumorigenesis.
In conclusion, as illustrated in Figure 7, we propose a pathogenic pathway f development centered on ROS overproduction and SIRT1 downregulation. Intrace increase of ROS could result from an upregulation of NOX4 induced by Th1 cyto Our results also suggest that the element linking the OS and the HIF-1α/GLUT-1 and 1α/VEGF-A pathways may involve SIRT1. An OS-induced downexpression of SIRT lead to increased HIF-1α activity followed by overexpression of GLUT-1 and VE Moreover, the diminished SIRT1 expression induced a reduction of antioxidant de due to PRDX1, catalase, and SOD1, further aggravating OS. Interestingly, in accor with our observations, it was shown that SIRT1 regulates NOX4 expression [24,25], is a protein considered essential for OS in thyrocytes. Finally, SIRT1, as an antioxid stress molecule, could be considered as an optimal therapeutic target for HT.  Hashimoto's thyroiditis. In normal thyroid, the redox balance is properly controlled and SIRT1 is expressed in the thyrocytes' cytoplasm, mainly in active follicles. In Hashimoto's thyroiditis, Th1 cytokines increase ROS production by upregulating NOX4 and decreasing the antioxidant enzymes such as PRDX1, catalase, and SOD1. The disruption of the redox balance induces an oxidative stress, particularly in altered active-like follicles. HIF-1α and its targets, VEGF-A and GLUT-1, are overexpressed by Th1 cytokines and in Hashimoto's patients. The downregulation of SIRT1 by Th1 cytokines and in Hashimoto's patients is a key event in the thyroiditis pathogenesis since SIRT1 has been shown in other cell types to be able to modulate oxidative stress by regulating NOX4 and HIF-1α. (T) inhibitor arrow and (↑) activator arrow.

Patients
Thyroid tissues from HT patients (n = at least 5, Table 1) were surgically obtained for Western blotting (WB) and immunohistochemistry (IHC). Paranodular tissue from patients with multinodular goiter was used as control samples (n = at least 5) and used to perform primary cell cultures. All thyroid samples were obtained after patients gave their informed consent (4 October 2017/466). Samples were obtained from the Biolibrary of the Cliniques universitaires Saint-Luc, referenced as BB190044, member of the Biothèque Wallonie Bruxelles (BWB) and of biobanking and biomolecular resources research infrastructure in Belgium (BBMRI.be).

Primary Cell Cultures
As previously described [16], thyrocytes were isolated from paranodular tissue from patients operated on for multinodular goiter. Briefly, thyroid samples were cut into pieces and incubated several times in collagenase type II (Worthington, Lakewood, CA, USA) solution until entirely digested. Isolated cells were filtered and cultured in modified Earle's medium (MEM) (ThermoFischer Scientific, Waltham, MA, USA) to which 5% newborn calf serum (ThermoFischer Scientific, Waltham, MA, USA), 1% penicillin-streptomycin (Ther-moFischer Scientific, Waltham, MA, USA), 1% amphotericin B (ThermoFischer Scientific, Waltham, MA, USA), and 1 mU/mL TSH (Sigma-Aldrich ® , St. Louis, MO, USA) at 37 • C, 5% CO 2 were added. After five days, cells were starved for one day in medium containing 0.5% newborn calf serum and then incubated for one additional day with recombinant human IL-1α (2 ng/mL, PeproTech, Minneapolis, MN, USA) and recombinant human IFNγ (10 ng/mL, PeproTech, Minneapolis, MN, USA) diluted in MEM containing 0.5% newborn calf serum. Cells were then harvested for further analyses. Primary cell cultures were realized at least in duplicate and repeated at least three times.

Immunohistochemistry
Immunostaining was performed as previously described [16]. Briefly, human thyroid samples were fixed in 4% paraformaldehyde and then embedded in paraffin. After rehydration, paraffin sections underwent antigen retrieval by heating in citrate buffer (not done for HIF-1α staining) and endogenous peroxidase activity was inhibited with 1% H 2 O 2 solution for 20 min. Paraffin sections were blocked with a goat serum (1/50, Sigma-Aldrich ® , St. Louis, MO, USA) and Bovine serum albumin (BSA) (Millipore, Merck, Darmstadt, Germany) 5% solution for 30 min at room temperature (RT). Sections were then incubated with primary antibody. Incubation time and dilution varied depending on the primary antibody used, as described in Table 2. A secondary antibody, Envision (Dako, Carpinteria, CA, USA), coupled to peroxidase and specific for the primary antibody (Table 2) was applied to the slides for 1 h at RT. Peroxidase activity was revealed with 3-3 diaminobenzidine tetrahydrochloride (DAB) (Vector, Burlingame, CA, USA). Sections were then counterstained with Mayer's hematoxylin and fixed in Entellan mounting medium (Millipore, Merck, Darmstadt, Germany).

Western Blotting
The thyroid samples or cultured cells were homogenized in radioimmunoprecipitation assay (RIPA) buffer containing a protease inhibitor cocktail and PhoSTOP (Roche, Basel, Switzerland). Protein concentration was determined using a bicinchoninic assay protein assay kit (ThermoFischer Scientific, Waltham, MA, USA). For Western blotting [47], 10 µg of proteins were loaded per well on a SDS-PAGE gel and migration was performed in migration buffer (Tris-EDTA, SDS, glycine) under 120 V for 90 min. Proteins were transferred onto nitrocellulose blotting membranes (GE Healthcare, Chicago, IL, USA) with a constant amperage of 350 mA for 90 min in transfer buffer (Tris-base, glycine, SDS, methanol). Membranes were blocked for 1 h at RT with TBS (Tris buffer Saline) containing 0.1% Tween and 5% BSA. Primary antibodies (Table 2) were then incubated on membranes overnight at 4 • C for immune detection of targeted proteins. Membranes were washed in TBS-Tween and incubated for 1 h at RT with the peroxidase-conjugated secondary antibody. Finally, membranes were developed on CL-X Posure ® films (ThermoFischer Scientific, Waltham, MA, USA) using ECL TM Western Blotting Detection Reagents (GE Healthcare, Chicago, IL, USA). Protein levels were normalized to β-actin for primary cultures of thyroid cells. The high number of inflammatory cells and connective tissue in human samples forced us to use GAPDH as normalizer for these blots as β-actin is more highly expressed in inflammatory cells in the fibrotic state [48,49]. Films were scanned and quantified by densitometry using ImageJ software (Version 1.52A) [50].

Quantification of ROS Levels Using CM-H2DCFDA
Levels of ROS in primary cultures of human thyrocytes, treated or not with Th1 cytokines, were assessed using CM-H 2 DCFDA technology. Briefly, human primary thyrocytes were cultured on glass coverslips and treated during 24 h with Th1 cytokines (IL-1α and IFNγ) after six days of culture including 24 h of starvation. At the day of analysis, cells were washed with Hank's Balanced Salt Solution (HBSS, ThermoFisher Scientific, Waltham, MA, USA) and incubated in a 10-µM solution of 6-chloromethyl-2 ,7dichlorodihydrofluorescein diacetate (CM-H 2 DCFDA, ThermoFisher Scientific, Waltham, MA, USA) at 37 • C in the dark for 30 min. Fluorescence was visualized with a fluorescence microscope by excitation around 488 nm. Images were quantified using the ImageJ program (Version 1.52a) [50].

Statistical Analysis
All experiments were reproduced at least three times. All values are expressed as means ± SEM. Shapiro-Wilk normality test was performed, and statistically significant differences were determined using a student t-test or by using a Mann-Whitney test for non-parametric data. p < 0.05 was considered statistically significant.