ZFP36L2 Role in Thyroid Functionality

Thyroid hormone levels are usually genetically determined. Thyrocytes produce a unique set of enzymes that are dedicated to thyroid hormone synthesis. While thyroid transcriptional regulation is well-characterized, post-transcriptional mechanisms have been less investigated. Here, we describe the involvement of ZFP36L2, a protein that stimulates degradation of target mRNAs, in thyroid development and function, by in vivo and in vitro gene targeting in thyrocytes. Thyroid-specific Zfp36l2-/- females were hypothyroid, with reduced levels of circulating free Thyroxine (cfT4) and Triiodothyronine (cfT3). Their hypothyroidism was due to dyshormonogenesis, already evident one week after weaning, while thyroid development appeared normal. We observed decreases in several thyroid-specific transcripts and proteins, such as Nis and its transcriptional regulators (Pax8 and Nkx2.1), and increased apoptosis in Zfp36l2-/- thyroids. Nis, Pax8, and Nkx2.1 mRNAs were also reduced in Zfp36l2 knock-out thyrocytes in vitro (L2KO), in which we confirmed the increased apoptosis. Finally, in L2KO cells, we showed an altered response to TSH stimulation regarding both thyroid-specific gene expression and cell proliferation and survival. This result was supported by increases in P21/WAF1 and p-P38MAPK levels. Mechanistically, we confirmed Notch1 as a target of ZFP36L2 in the thyroid since its levels were increased in both in vitro and in vivo models. In both models, the levels of Id4 mRNA, a potential inhibitor of Pax8 activity, were increased. Overall, the data indicate that the regulation of mRNA stability by ZFP36L2 is a mechanism that controls the function and survival of thyrocytes.


In Vivo Deletion of Zfp36l2 Impairs Thyroid Function
We started the analysis of the role of ZFP36L2 in thyroid development and function by using a complete knockout mouse model of Zfp36l2 [29]. As previously reported, complete Zfp36l2 -/mice died within approximately three weeks of birth before weaning. ELISA measurement of circulating thyroid hormones in serum samples did not show any major differences in circulating free T4 (cfT4) levels between Zfp36l2 -/and control mice at postnatal day seven (PND 7, data not shown). cfT4 levels were reduced in Zfp36l2 -/mice at PND 21 (Figure 1a), possibly because the THs supplied by the mother, through lactation, were no longer sufficient. Since no males were available, we evaluated only females (n = 4 Zfp36l2 -/and n = 5 control mice). Thyroids appeared to develop normally and were not ectopic in the Zfp36l2 -/mice at PND 21. H&E staining of the sections of the thyroid gland showed a trend towards a decrease in the numbers of follicles together with an increased presence of fat in the thyroid of the Zfp36l2 -/mice ( Figure S1a, a 1 ) vs. controls ( Figure S1b, b 1 ).
To investigate the role of ZFP36L2 in thyroid function at later life stages, we developed mice on a C57/BL6 strain background with thyroid-specific inactivation of Zfp36l2 (t-Zfp36l2 -/-from now on). These mice were obtained by crossing the Zfp36l2 flox/flox mice [35] with the mice expressing a Pax8-Cre allele, driving gene inactivation mainly in the thyroid and kidney [36]. Mice were genotyped as described in M&M. As shown by semi-quantitative reverse transcription real-time PCR (RTqPCR) and Western blotting, Figure 1. Phenotype of constitutive and thyroid-specific Zfp36l2 -/mice. The levels of cfT4 were determined by ELISA in blood collected from the controls (n = 5) and conventional Zfp36l2 -/-(Zfp36l2 -/n = 4) female mice at PND 21 (a). The levels of cfT4 (b) and cfT3 (c) were determined by ELISA in blood collected from controls (n = 6) and t−Zfp36l2 -/-(n = 6) female mice at the PND 90. Body weight evaluation of the t−Zfp36l2 -/mice measurements were collected at PND 90 (d). Data are reported as mean ± standard deviation * p < 0.05; **** p < 0.0001.
To investigate the role of ZFP36L2 in thyroid function at later life stages, we developed mice on a C57/BL6 strain background with thyroid-specific inactivation of Zfp36l2 (t-Zfp36l2 -/from now on). These mice were obtained by crossing the Zfp36l2 flox/flox mice [35] with the mice expressing a Pax8-Cre allele, driving gene inactivation mainly in the thyroid and kidney [36]. Mice were genotyped as described in M&M. As shown by semi-quantitative reverse transcription real-time PCR (RTqPCR) and Western blotting, the deletion of the gene in the whole thyroid of the t-Zfp36l2 -/females at PND 90 was not complete, possibly reflecting continued expression in non-thyrocytes. Nevertheless, its mRNA ( Figure 2a) and protein levels (Figures 2b and S4a) were significantly reduced compared to control littermates (Zfp36l2 flox/flox C57/BL6). cfT4 and cfT3 were determined by ELISA at PND 30, 60, and 90. No major differences in cfT4 were detected in males compared to control littermates at any sampling time (data not shown). The t-Zfp36l2 -/females manifested a trend towards decreased cfT4 levels at PND 30 and PND 60 ( Figure S2), and the levels of cfT4 and cfT3 (Figure 1b,c, n = 6) were significantly reduced at PND 90. Concordantly with the hormone levels, the t-Zfp36l2 -/females had reduced body weights compared to the controls (Figure 1d). Considering these data and the fact that females are more susceptible to hypothyroidism [37], only this sex was investigated further. H&E staining of thyroid sections did not demonstrate any major defects, although slightly smaller follicles were detected in the t-Zfp36l2 -/females vs. controls ( Figure S3). the deletion of the gene in the whole thyroid of the t-Zfp36l2 -/-females at PND 90 was not complete, possibly reflecting continued expression in non-thyrocytes. Nevertheless, its mRNA ( Figure 2a) and protein levels ( Figure 2b, Figure S4a) were significantly reduced compared to control littermates (Zfp36l2 flox/flox C57/BL6). cfT4 and cfT3 were determined by ELISA at PND 30, 60, and 90. No major differences in cfT4 were detected in males compared to control littermates at any sampling time (data not shown). The t-Zfp36l2 -/females manifested a trend towards decreased cfT4 levels at PND 30 and PND 60 ( Figure  S2), and the levels of cfT4 and cfT3 (Figure 1b,c, n = 6) were significantly reduced at PND 90. Concordantly with the hormone levels, the t-Zfp36l2 -/-females had reduced body weights compared to the controls (Figure 1d). Considering these data and the fact that females are more susceptible to hypothyroidism [37], only this sex was investigated further. H&E staining of thyroid sections did not demonstrate any major defects, although slightly smaller follicles were detected in the t-Zfp36l2 -/-females vs. controls ( Figure S3). To more completely characterize the thyroid phenotype, thyrocyte proliferation and apoptosis were investigated. No major differences in proliferation levels (Ki67 staining) were observed between the t-Zfp36l2 -/-and control females (data not shown). Furthermore, in order to detect the possible presence of apoptosis, we performed Tunel assay staining. In this way, we detected a slight increase of apoptotic nuclei (Figure 3a-f) in thyroids from t-Zfp36l2 -/-vs. control females. These data were associated with an increase in the ratio of Bax over Bcl2 mRNA (Figure 3g).
We then investigated thyroid functionality by determining the expression levels of thyroid-specific transcripts (Nis, Tg, and Tpo) and transcripts for thyroid-enriched transcriptional factors (Pax8, Nkx2.1) by RTqPCR. These data demonstrated reductions of Pax8, Nkx2.1, Nis, Tg, and Tpo transcripts in t-Zfp36l2 -/-vs. controls (Figure 4a). The decreases in PAX8 and NIS levels were also demonstrated by Western blotting (Figure 4b, Figure S4b).
Collectively, the data show evidence of hypothyroidism in the t-Zfp36l2 -/-females and that the inactivation of Zfp36l2 in the thyroid affects the function of the gland in more ways than its development and morphology, except for the slight increase in apoptosis detected in the t-Zfp36l2 -/-females. To more completely characterize the thyroid phenotype, thyrocyte proliferation and apoptosis were investigated. No major differences in proliferation levels (Ki67 staining) were observed between the t-Zfp36l2 -/and control females (data not shown). Furthermore, in order to detect the possible presence of apoptosis, we performed Tunel assay staining. In this way, we detected a slight increase of apoptotic nuclei (Figure 3a-f) in thyroids from t-Zfp36l2 -/vs. control females. These data were associated with an increase in the ratio of Bax over Bcl2 mRNA (Figure 3g).
We then investigated thyroid functionality by determining the expression levels of thyroid-specific transcripts (Nis, Tg, and Tpo) and transcripts for thyroid-enriched transcriptional factors (Pax8, Nkx2.1) by RTqPCR. These data demonstrated reductions of Pax8, Nkx2.1, Nis, Tg, and Tpo transcripts in t-Zfp36l2 -/vs. controls (Figure 4a). The decreases in PAX8 and NIS levels were also demonstrated by Western blotting (Figures 4b and S4b).
Collectively, the data show evidence of hypothyroidism in the t-Zfp36l2 -/females and that the inactivation of Zfp36l2 in the thyroid affects the function of the gland in more ways than its development and morphology, except for the slight increase in apoptosis detected in the t-Zfp36l2 -/females.

Zfp36l2 Gene Inactivation Affects Expression of Key Thyroid Genes in Immortalized Thyrocytes.
To investigate the molecular mechanisms underlying the effects of Zfp36l2 inactivation in thyrocytes, we generated a cellular model by knocking out Zfp36l2 in immortalized rat thyroid cells (FRTL5). FRTL5 is a well-characterized in vitro model widely used to study thyroid function since it retains the ability to concentrate iodine, produce thyroglobulin, and respond to TSH stimulation [38]. The Zfp36l2 knock-out FRTL5 cells (L2KO, from now on) were obtained by lentiviral delivery of the Crispr/Cas9 components, as detailed in M&M, whereas control cells were infected with an empty vector (EV, from now on). We determined the levels of the Zfp36l2 mRNA and protein both in the bulk culture and in selected clones. A representative result is reported in Figure  5a, showing that Zfp36l2 mRNA was strongly reduced and the protein was not detectable

Zfp36l2 Gene Inactivation Affects Expression of Key Thyroid Genes in Immortalized Thyrocytes
To investigate the molecular mechanisms underlying the effects of Zfp36l2 inactivation in thyrocytes, we generated a cellular model by knocking out Zfp36l2 in immortalized rat thyroid cells (FRTL5). FRTL5 is a well-characterized in vitro model widely used to study thyroid function since it retains the ability to concentrate iodine, produce thyroglobulin, and respond to TSH stimulation [38]. The Zfp36l2 knock-out FRTL5 cells (L2KO, from now on) were obtained by lentiviral delivery of the Crispr/Cas9 components, as detailed in M&M, whereas control cells were infected with an empty vector (EV, from now on). We determined the levels of the Zfp36l2 mRNA and protein both in the bulk culture and in selected clones. A representative result is reported in Figure 5a, showing that Zfp36l2 mRNA was strongly reduced and the protein was not detectable in the clone; whereas, they were both significantly reduced in the bulk culture (Figure 5a,b). We chose to work on the bulk culture as being more representative of the heterogeneity found in the in vivo model. 5a,b). We chose to work on the bulk culture as being more representative of the heterogeneity found in the in vivo model. L2KO cells did not show any major changes in cellular morphology vs. EV cells. Since transcripts playing a role in the regulation of the cell cycle and apoptosis are targeted by ZFP36L2 in other cell types [39], we tested both cellular activities. MTT assays did not show any major differences in the proliferation of exponentially growing EV and L2KO cells (Figure 5c). Despite that, we found an increased expression of P21/Waf1 (Figure 5d), suggesting an impairment of the cell cycle. Furthermore, the increased ratio of Bax/Bcl2 transcripts pointed to a possible increase in apoptosis ( Figure 5e).  L2KO cells did not show any major changes in cellular morphology vs. EV cells. Since transcripts playing a role in the regulation of the cell cycle and apoptosis are targeted by ZFP36L2 in other cell types [39], we tested both cellular activities. MTT assays did not show any major differences in the proliferation of exponentially growing EV and L2KO cells (Figure 5c). Despite that, we found an increased expression of P21/Waf1 (Figure 5d), suggesting an impairment of the cell cycle. Furthermore, the increased ratio of Bax/Bcl2 transcripts pointed to a possible increase in apoptosis (Figure 5e).
The molecular characterization of thyroid-specific or enriched transcripts and proteins conducted in exponentially growing EV and L2KO cells mainly agreed with what we found in mice. Indeed, Pax8, Nkx2.1, and Nis mRNAs were significantly reduced (Figure 6a), as were their encoded proteins (Figures 6b-d and S5) in L2KO vs. EV. In contrast, no differences were observed for Tg and Tpo in the L2KO cells (Figure 6a) or for Foxe1/TTF2 (Figures 6a,d and S5). As members of the ZFP36 family might have redundant functions, we analyzed the cellular content of ZFP36L1 and ZFP36 in the L2KO cells. Levels of ZFP36L1 were reduced in L2KO vs. EV (Figures 6e and S5), whereas Zfp36 mRNA showed a slight trend towards a decrease (data not shown).
Overall, the data suggest a role for ZFP36L2 in thyrocyte functionality in rat immortalized thyrocytes. The molecular characterization of thyroid-specific or enriched transcripts and proteins conducted in exponentially growing EV and L2KO cells mainly agreed with what we found in mice. Indeed, Pax8, Nkx2.1, and Nis mRNAs were significantly reduced (Figure 6a), as were their encoded proteins (Figure 6b-d, Figure S5) in L2KO vs. EV. In contrast, no differences were observed for Tg and Tpo in the L2KO cells (Figure 6a) or for Foxe1/TTF2 (Figure 6a,d, Figure S5). As members of the ZFP36 family might have redundant functions, we analyzed the cellular content of ZFP36L1 and ZFP36 in the L2KO cells. Levels of ZFP36L1 were reduced in L2KO vs. EV (Figure 6e, Figure S5), whereas Zfp36 mRNA showed a slight trend towards a decrease (data not shown).
Overall, the data suggest a role for ZFP36L2 in thyrocyte functionality in rat immortalized thyrocytes.

Zfp36l2 Inactivation Impairs The FRTL5 Response to TSH
Since TSH stimulates both cellular proliferation and the expression of thyroid-specific or enriched genes in FRTL5 cells [40], we investigated whether the loss of ZFP36L2 could affect these cells' response to TSH. To test this, EV and L2KO cells were deprived of TSH by growing them in a medium without TSH (5H, from now) for three days. Such a procedure results in FRTL5 cell synchronization in the G0/G1 phase ( Figure S6). After starvation, the cells were stimulated by adding a fresh complete growing medium (6H from now, containing 1 mU/mL TSH) for 24 and 48 h. We chose these time points since FRTL5 doubling time is reported to be 36 h [41]. First, we tested whether ZFP36L2 levels could be directly regulated by TSH. We found that ZFP36L2 protein levels were reduced after TSH starvation and increased after TSH stimulation (Figures 7a and S7a). Interestingly, the levels of ZFP36L2 protein seemed to be increased in starved L2KO cells vs. EV (Figures 7 and S7). The same behavior was observed with ZFP36L1 (Figures 7a and S7b). Furthermore, TSH appeared to promote ZFP36L2 protein phosphorylation since treatment of the extracts with calf intestinal phosphatase resulted in the appearance of a single band of immunoreactive protein (Figure 7b). To confirm TSH signaling, we analyzed NIS protein levels in L2KO and EV cells. As shown in Figure 7a, NIS was downregulated in starved L2KO cells compared to EV cells, the increase after TSH stimulation was lower, and it was delayed in L2KO cells ( Figure S7c). Overall, these data suggested that the response to TSH was impaired in the absence of ZFP36L2.
Since TSH controls both proliferation and survival of thyrocytes, we investigated both processes in L2KO and EV cells treated as above. Cell counts showed a reduced number of L2KO vs. EV cells after 48 h of TSH stimulation (Figure 7c (Figures 7f and S7g). Interestingly, the levels of total P38α protein showed a reduction in L2KO vs. EV (Figures 7f and S7h). The increased p-P38 in L2KO cells vs. EV was already apparent after shorter periods of TSH stimulation ( Figure S8).

ZFP36L2 Modulates Thyrocyte Function and Apoptosis via the Notch1 Pathway In Vitro and In Vivo
As described above, Zfp36l2 gene inactivation reduces the levels of thyroid-specific transcripts as well as those of Zfp36l1 both in vivo and in vitro. This observation suggests that thyroid-specific transcripts are not direct targets of ZFP36 protein activity. Another potential mechanism by which ZFP36L2 could modulate the levels of thyroid-specific transcripts was described in T-lymphoblastic leukemia, in which ZFP36L2 and L1 were able to modulate the cellular content of Notch1 mRNA and protein [42], a signaling pathway involved in thyrocyte function [15,43,44]. Therefore, we assayed the levels of the Notch1 gene in our experimental settings in vitro and in vivo. Notch1 gene expression was increased in exponentially growing L2KO vs. EV cells (Figure 8a, protein is shown in Supplementary Figure S9a), as well as in TSH-starved cells (Figure 8b, protein is shown in Supplementary Figure S9b). Furthermore, TSH stimulation increased its levels in EV after 48 h, whereas it had little effect in L2KO cells (Figure 8b, protein is shown in Supplementary  Figure S9b). This was also in agreement with a reduced ARE binding activity of ZFP36L2 when phosphorylated [45]. This result was confirmed in thyroids from t-Zfp36l2 -/mice.

Discussion
Tissue-specific gene expression results from the interplay between transcriptional and post-transcriptional mechanisms. The former has been extensively investigated; the latter is almost unexplored in thyroid development and function. Studies of posttranscriptional regulation in the thyroid could improve our knowledge of the genes altered in CH, whose mechanisms of action are often unexplained. It is noteworthy that several recent papers have described the role of splicing modulation in thyroid development in vivo [49], suggesting this mode of post-transcriptional regulation as a mechanism involved in thyroid development, function, carcinogenesis, and thyroid NOTCH1 and HES1 are positive regulators of the expression of thyroid-specific genes that decreased following Zfp36l2 gene inactivation. We looked for other NOTCH1 targets that could act as transcriptional repressors. Since ID4, a transcriptional repressor, is a target of NOTCH1 [47] and has been reported as a negative regulator of PAX8 transcriptional activity [48], we attempted to verify its increase in our models. We found, by RTqPCR, that levels of the Id4 transcript were increased in L2KO vs. EV cells (Figure 8e) as well in thyroids from t-Zfp36l2 -/mice vs. controls (Figure 8f).
Collectively, those data suggest that ZFP36L2 might exert its effects by deregulating NOTCH1 signaling and promoting an increase of HES1 and Id4 levels in thyrocytes. This imbalance might damage thyrocyte function and survival.

Discussion
Tissue-specific gene expression results from the interplay between transcriptional and post-transcriptional mechanisms. The former has been extensively investigated; the latter is almost unexplored in thyroid development and function. Studies of post-transcriptional regulation in the thyroid could improve our knowledge of the genes altered in CH, whose mechanisms of action are often unexplained. It is noteworthy that several recent papers have described the role of splicing modulation in thyroid development in vivo [49], suggesting this mode of post-transcriptional regulation as a mechanism involved in thyroid development, function, carcinogenesis, and thyroid hormone signaling [19]. Hence, the search for novel alleles of CH candidate genes continues since their interaction with already characterized genes involved in CH, and with environmental factors, could play a role in the pathogenesis of CH.
In our previous work, we observed that levels of the mRNA of ZFP36L2, a protein that can promote RNA degradation [50], were increased during thyroid development [32] and reduced during hypothyroidism promoted by environmental stressors [34], as well as during thyroid carcinogenesis [33,41]. Its role in thyroid development and function has not yet been characterized.
Overall, the results reported here confirm the role of the ZFP36L2 protein in thyroid function. Indeed, we report that Zfp36l2 gene inactivation promotes thyroid dyshormonogenesis, as suggested by reduced levels of cfT4 in the mouse model and normal thyroid development and morphology. We also report a slight impact on the survival of thyrocytes in vitro and in vivo. This effect appears to be mediated by the altered expression of thyroidspecific enzymes and their transcriptional regulators observed both in vivo and in vitro. Mice knocked-out for Zfp36l2 -/generally die before weaning and exhibit a marked reduction of fetal liver hematopoietic progenitors [29]. In our previous work, we suggested that the ZFP36L2 reduction promoted by environmental factors might contribute to the observed pancytopenia by promoting hypothyroidism [34]. This suggestion is supported here by the reduction of cFT4 in the Zfp36l2 -/mice. It is noteworthy that decreases in cfT4 and cFT3 levels have been considered determinants of the defect in erythropoiesis detected in TRα knockout mice at later life stages [51]. Furthermore, the detected hypothyroidism might also contribute to ZFP36L2 s influence on ovulation and oocyte maturation. Indeed, hypothyroidism might affect both processes [52]. Therefore, the hypothyroidism detected in the Zfp36l2 -/females might be a causative factor in several of the phenotypes associated with ZFP36L2 activity loss.
In oocytes, it was shown that PKA inhibition significantly rescued the effects associated with Zfp36l2 inactivation [53]. Indeed, PKA, as well as other signaling pathways, are activated by TSH in the thyroid [54]. In agreement with these previously shown data, we describe different responses to TSH in EV and L2KO cells. Indeed, ZFP36L2 is apparently phosphorylated after TSH stimulation, and its levels are increased. Along with the inhibition of its activity, we also report an increase in cleaved NOTCH1 protein, one of its targets in other tissues [45]. We focused our attention on the Notch1 pathway since its role in thyroid development and function has been reported previously. For example, it has been shown that Notch1 pathway activation strongly influences thyroid development if induced before thyroid cell specification in zebrafish [15]. In our in vivo models, such an effect was only partially confirmed, given that no defects were detected in gland development [15]. We suggest that this might depend on the later expression of Pax8-Cre, which could drive Zfp36l2 gene inactivation at later stages of thyroid development.
The NOTCH1 downstream target Hes1 has also been implicated in thyroid development and function [43] since its genetic inactivation resulted in the reduction of NKX2.1 positive cells during thyroid development and a reduced expression of thyroid-specific genes [44]. We demonstrated increases in cleaved-NOTCH1 in the t-L2KO thyroids, but also an impairment of thyroid function like that described in Hes1 -/mice. This discrepancy might be due to the window of analysis since Hes1 -/mice died at embryonic stage E. 16.5; thus, the effects at later life stages might be different. Furthermore, NOTCH1 targets several genes that might play other roles in regulating gene expression in the thyroid, meaning that some of them could affect the expression of thyroid-specific enzymes, while others could influence the expression of proteins able to modulate proliferation and survival of thyrocytes [44]. Indeed, activated Notch1 pathways have been implicated in the delayed cell cycle progression and apoptosis seen in FRTL5 [42]. Thus, its activation could be involved in the reduction of cell proliferation seen in L2KO with the increase in apoptosis seen in thyroids from Zfp36l2 -/female mice. Furthermore, NOTCH1 pathway activation has been associated with the increase of P21 in thyroid cancer models [55]. The expression profile of thyroid-specific transcripts is in contrast with previously published data demonstrating that NOTCH1 was their positive transcriptional regulator. This implies that NOTCH1 activation along the ZFP36L2-NOTCH1 axis in the thyroid might involve other players. We propose that one of them could be the ID4 protein, a transcriptional repressor since it is induced by NOTCH1 and could inhibit PAX8-regulated gene expression by interacting with PAX8 protein [48]. Consistently, we detected Id4 mRNA increases in thyrocytes carrying inactivated Zfp36l2 and decreases in Pax8 transcript levels since PAX8 can also regulate its own promoter [48]. Overall, the data suggest that different NOTCH1 targets are involved in the control of cell proliferation (HES1) and thyroid-specific gene expression (ID4). Furthermore, considering the modulation of epigenetic mechanisms by ZFP36L2 [35], we cannot rule out their involvement. Indeed, a Notch Repressive Complex (NRC) containing NuRD and PRC1 has been described. PRC1 is a target of ZFP36 [56,57] and could increase following the deletion of Zfp36l2 and inhibit the expression of Pax8 and Nkx2.1, as reported in other systems [58].
Interestingly, we also found evidence for the hyper-activation of the P38-MAPK pathway after TSH stimulation in L2KO vs. EV cells. This might depend on the accumulation of MKK6 protein (its upstream activator), whose transcript levels are modulated by mRNA decay mechanisms [59]. This suggestion would agree with the increase of Nis mRNA in L2KO, already described as dependent on CHOP activation in thyroid cells [60]. The increased activation of the P38-MAPK pathway might contribute to slowing down the G2/M transition in L2KO and to the increased apoptosis in vivo since it is inhibited in thyroid hyperplastic and neoplastic tissues [61]. In our models, this event is driven by ZFP36L2 inactivation in association with the reduced expression of thyroid-specific transcripts. This last is a hallmark of the transformation of thyrocytes [33]. Finally, apoptosis is an early and pivotal step in thyroid carcinogenesis, and ZFP36L2 inactivation results in its increase. Hence, ZFP36L2 might participate in the generation of surviving thyrocytes that are more prone to transformation [24,62].
Overall, the data suggest that mRNA decay is a mechanism for regulating thyroid function by modulating thyroid-specific gene expression. The silencing of ZFP36L2 by gene inactivation indirectly modulates the expression of thyroid-specific enzymes and their transcriptional regulators. These events promote hypothyroidism in vivo, beginning at the early phases of post-natal life. The data support the hypothesis that ZFP36L2 exerts this activity by regulating the NOTCH1 pathway, including its targets HES1, active on modulation of proliferation and apoptosis, and ID4, modulating thyroid-specific gene expression. Further studies are needed to characterize both mechanisms. Finally, the similarity of the in vivo and in vitro results confirm the suitability of the FRTL5 cells as an appropriate in vitro model to investigate the mechanisms regulating thyrocyte function by in vitro gene-targeting approaches.

MTT Assay and Cell Counts
MTT assays were performed as previously described [67]. Briefly, 5 × 10 3 cells were plated in 96 multi-well plates. The cells were allowed to grow for up to 7 days. At the indicated time points, 500 µg/mL MTT was added to each well for 1 h. After incubation, samples were resuspended in 100% DMSO, and absorbance was read at 570 nm using an Envision 2103 Multilabel Reader (Perkin Elmer, Waltham, MA, USA).
For cell counts, 1 × 10 5 cells were plated in 12 multi-well plates. One day later, cells were starved using a medium lacking TSH (5 hormones media, 5H) for 3 days. Next, the 6H medium was added. Cells were harvested at the indicated time points and counted using a TC20 cell counter (Biorad, Hercules, CA, USA) and trypan blue exclusion (seeding was considered as Time 0).

Cell Cycle Analysis, Real-Time PCR, Western Blotting, and CIP Treatment
Cells were plated in plates (10 cm) at 3.5 × 10 4 cells/cm 2 the day before the experiments. The next day, 5H medium was added to the cell for TSH deprivation. TSH stimulation was achieved by adding cell culture medium containing 1 mU/mL TSH (6 hormones media, 6 h). Cells were harvested before replacement of the medium (time 0) and after 24 and 48 h. The collected cells were in part fixed in 70% ethanol for PI staining and FACS analysis [53], in part lysed in TRIZOL reagent (Thermo Fisher Scientific) for analysis of mRNA levels, and, finally, in part lysed in RIPA buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 2% NP-40, 1% sodium deoxycholate, complete mini EDTA-free protease inhibitor cocktail-Roche-and PhosSTOP phosphatase inhibitor cocktail-Roche) for Western blotting analysis.
Reverse transcription-quantitative PCR (RTqPCR) experiments were conducted as previously reported [68]. First, RNA was converted to cDNA using Quantitect Reverse Transcription Kit (Qiagen, Hilden, Germany) [68]. Subsequently, mRNA levels were evaluated as previously described [69], using the PowerUP SYBR green Master Mix (Thermo Fisher) and primers listed in table S1.
Whole-cell lysates for Western blotting analysis were prepared and quantified using Bradford reagent (Biorad). After BIS-TRIS SDS-PAGE, proteins were transferred to 0.2 µm PVDF membranes (Thermo Fisher Scientific). Membranes were blocked with 5% nonfat dry milk (Biorad) and incubated overnight at 4 • C with the indicated antibodies. Signals were revealed using ImmunoCruz Western blotting luminol reagent (Santa Cruz Biotechnology, Dallas, TX, USA) or Immobilon ECL Ultra (Merk) and acquired using a Molecular Imager Chemidoc XRS+ and Image Lab software (Biorad), or Kodak Medical X-ray Processor 2000 and Fujifilm Super RX-N films. Protein signal quantification was performed using Fiji NIH Image J software. For the analysis of ZFP36L2 phosphorylation, whole-cell lysates (30 µg) were treated with 30 units of Calf intestinal phosphatase (CIP, New England Biolabs, Ipswich, MA, USA) for 30 min at 37 • C before being loaded on gels for SDS-PAGE and Western blotting.

Thyroid Tissue Preparation
Thyroids from 6 female mice, aged 3 months, of each genotype (Zfp36l2 flox/flox and Pax8cre-Zfp36l2 flox/flox ) were sampled by dividing the right and left lobes. Next, samples were divided as follows: 6 lobes were formalin-fixed and paraffin-embedded; 3 lobes were lysed in TRIZOL reagent for gene expression analysis, as described before; and 3 lobes were lysed in RIPA buffer for Western blotting analysis, as described before. Lysis of the tissue samples was performed using a Misonix XL2020 sonicator, at the power setting 4, for 15 s, on ice.

Measurement of fT3 and fT4 Hormones
FT4 and fT3 levels were measured in 6 serum samples from 3-month-old female animals of each genotype by using an ELISA kit (Diametra, Roma, Italy) following the manufacturer's instructions, as previously described [68], and reading at the Biotek ELX800 multi-plate reader.

Histology and Tunel Assay
Hematoxylin and Eosin staining were performed on 5 µm tissue slices prepared from 6 samples of Formalin-fixed and Paraffin-embedded thyroids from 3-month-old female animals of each genotype, as previously described [33].
Tunel assays were performed on 5 µm tissue slices prepared from 3 samples of Formalin-fixed and Paraffin-embedded thyroids from 3-month-old female animals of each genotype using In Situ Cell Death Detection Kit, Fluorescein (Roche, Basel, Switzerland), following the manufacturer's instructions. Images were acquired using a Zeiss Axioplan 2 microscope at 10X magnification and ZEN 3.1 software.

Statistical Analysis
Graphs and statistical analysis were performed using GraphPad 7 (Prism). Unpaired Student's t-test (for direct comparisons) or Two-way ANOVA (for multiple comparisons) were performed to evaluate the significance of differences.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/ijms22179379/s1, Table S1. List of primers used for RTqPCR. Figure S1. Hematoxylin and Eosin staining of the thyroid of Zfp36l2 -/mice (N= 4, a,a 1 ) and controls (N = 5, b,b 1 ) at PND21. a,b images are 10× magnification, a 1 ,b 1 images are 40X magnification. Figure S2. ELISA quantification of cfT3 and cfT4 in 1-month-and 2-month-old mice. Controls (N = 6) and t-Zfp36l2 -/-(N = 6) mice were used. Figure S3. Haematoxylin and eosin staining of the controls (N = 3) and t-Zfp36l2 -/-(N = 3) thyroids. No appreciable morphological changes are evident (10× magnification). Figure S4. Densitometric analysis of Western blotting experiments shown in Figure 3. Figure S5. Densitometric analysis of Western blotting experiments shown in Figure 5. Figure S6. Cell cycle analysis of L2KO vs. EV cells to verify the efficiency of 3 days of TSH deprivation. Cells were starved as detailed in M&M. The data show that about 90% of L2KO and EV cells were in the G0/G1 phase, as expected. Figure S7. Densitometric analysis of Western blotting experiments shown in Figure 6. Figure S8. Analysis by Western blotting of the activation of P38-MAPK pathways at early times of TSH stimulation. EV and L2KO were starved and stimulated with TSH as described in M&M. At the reported time points; the cells were collected, lysed in RIPA buffer, and analyzed by Western blotting (a). Densitometric analysis (b). Figure S9. Western blot analysis of cleaved-Notch1 in experimental points showed in Figure 8. Figure S10. Densitometric analysis of western blotting experiments showed in Figure S9.

Conflicts of Interest:
The authors declare no conflict of interest.