A TSH-CREB1-microRNA Loop Is Required for Thyroid Cell Growth

Abstract

MicroRNA (miRNA or miR) are an important class of regulators that participate in such biological functions as development, cell proliferation, differentiation, and apoptosis. The aim of this study was to elucidate the role of miRNA in cell proliferation using a unique cell system, namely thyroid cells that require thyrotropin for their growth. Here, we report the identification of a set of five specific miRNA (miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a), whose down-regulation by thyrotropin is required for thyroid cell growth. In fact, overexpression of these miRNA negatively affects cell growth. We show that three of these miRNA target cAMP-responsive element binding protein (CREB)1, a thyrotropin-activated transcription factor, and that CREB1 binds the regulatory regions of the down-regulated miRNA. Hence, these data indicate that a synergistic loop involving thyrotropin, CREB1, and miRNA is required for thyroid cell proliferation.

MicroRNA (miRNA or miR) are small noncoding RNA capable of regulating gene expression at translational level. In fact, they repress target gene expression by binding to complementary sequences found in the 3′-untranslated regions (UTR) of target mRNA. Recent evidence suggests that a significant portion of the human genome is regulated by miRNA. One miRNA is capable of regulating several distinct mRNA, and all the human miRNA identified so far are believed to modulate more than one-third of the mRNA species encoded in the genome (1–3). Moreover, each gene may be regulated by more than one miRNA. Therefore, the potential regulatory circuitry afforded by miRNA is enormous.

Several studies have already shown that miRNA have a critical role in the proliferation and differentiation of several cell types (4–7). Thyroid cells are an excellent cellular system with which to study the role of miRNA in the control of hormone-regulated cell proliferation. Indeed, thyroid cells in culture, rendered quiescent by thyrotropin (TSH) deprivation, can be stimulated to undergo DNA synthesis in the absence of serum by the addition of thyrotropin (10 nm). The rat thyroid cell line PC Cl 3 has the following thyroid differentiated functions: dependence for growth on thyrotropin, which is necessary for normal growth and is able to stimulate quiescent cells to enter S phase, thyroglobulin synthesis and secretion in the culture medium, and ability to take up iodide from the culture medium (8). Without thyrotropin, thyroid cells remain attached to the plate, develop morphological alterations, and do not divide (8–11). Therefore, the thyroid cell system has the great advantage that its proliferation depends on a unique stimulating factor, namely thyrotropin, whose effector pathway is well defined. Moreover, thyroid differentiated cells are epithelial, that means the cell type from which the large majority of the human solid neoplasias originates.

The aim of our work was to identify the miRNA regulated by thyrotropin in rat thyroid cells and likely characterize their role in thyroid cell proliferation. In the present work, we have carried out miRNA expression profiling of untreated and thyrotropin-stimulated PC Cl 3 cells to identify the miRNA regulated by thyrotropin. Among the miRNA differentially expressed, we focused our attention on a subset of miRNA, including the miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a, down-regulated in thyrotropin-treated cells with respect to the untreated cells. Finally, we identified the cAMP-responsive element (CRE) binding protein (CREB)1 gene product as a target of miR-1, miR-28-A, and miR-296-3p, cyclin D2 (CCND2) as target of miR-297a, and junB as target of miR-296-3p and miR-297a. Moreover, we report that all these miRNA are able to inhibit thyroid cell growth.

Results

miRNA expression profile in thyrotropin-stimulated PC Cl 3 cells

We first extracted RNA from BSA-treated and thyrotropin-stimulated PC Cl 3 cells at various time points after exposure to thyrotropin (10 nm) (30 min, 2 h, 8 h, 16 h, 24 h, and 48 h), as described in Materials and Methods, and determined the miRNA expression profile with the miRNACHIP microarray (12). ANOVA analysis of the microarray resulted in a list of miRNA (P < 0.05) differentially expressed between thyrotropin-treated and BSA-treated PC Cl 3 cells (Table 1). At 30 min, 2 h, and 8 h, several miRNA were up-regulated or down-regulated with a ratio equal to or higher than two in thyrotropin-treated cells with respect to untreated cells. In this study, we focused on the miRNA down-regulated by thyrotropin. Therefore, we next validated the results of the miRNA chip analysis by real-time quantitative PCR (qRT-PCR) of five miRNA (miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a): the qRT-PCR data confirmed the down-regulation of the five miRNA in PC Cl 3 cells 30 min, 2 h, and 8 h after thyrotropin treatment (Fig. 1A). At 16 h, the levels of these miRNA started to increase, and at 24 h, they were expressed at a level comparable with that of PC Cl 3 cells treated with BSA. Similar results have been obtained also with FRTL5 cells, another rat thyroid cell line (13), treated for 1, 4, 8 h with thyrotropin (Supplemental Fig. 1A, published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org).

Table 1.

miRNA differentially expressed between PC Cl 3 treated with TSH and PC Cl 3 treated with BSA

Unique ID	Ratio (treatment TSH vs. BSA)	P
mmu-mir-296-3p	0.07 (30 min)	2.86e-05
mmu-mir-290-5p	0.222 (30 min)	0.000106
mmu-mir-672-P	0.289 (30 min)	0.0003404
hsa-mir-768-3p-A	0.297 (30 min)	0.0003481
hsa-mir-28-A	0.2 (30 min)	0.0009455
mmu-mir-297a	0.184 (30 min)	0.0004057
mmu-mir-193-A	0.369 (30 min)	0.0006568
hsa-mir-1	0.555 (30 min)	0.0111224
mmu-mir-28-A	0.237 (30 min)	0.0045241
mmu-mir-340-P	0.459 (30 min)	0.0024481
mmu-mir-28-A	0.007 (2 h)	1.733 e-05
mmu-mir-20a-A	0.034 (2 h)	9.5e-05
mmu-mir-296-3p	0.073 (2 h)	5.62e-05
mmu-mir-290-5p	0.21 (2 h)	7.85e-05
mmu-mir-203-A	0.198 (2 h)	0.0011176
mmu-mir-15a-A	0.312 (2 h)	0.0005262
mmu-mir-193-A	0.369 (2 h)	0.0005866
hsa-mir-1	0.34 (2 h)	0.0008707
mmu-mir-93-A	0.42 (2 h)	0.0011828
mmu-mir-297a	0.292 (2 h)	0.0019717
mmu-mir-296-3p	0.033 (8 h)	8.96e-05
mmu-mir-290-5p	0.235 (8 h)	0.0001975
hsa-mir-28-A	0.185 (8 h)	0.0036819
hsa-mir-211-A	0.384 (8 h)	0.0019074
mmu-mir-291a-3p-A	0.235 (8 h)	0.002062
hsa-mir-219-1-A	0.406 (8 h)	0.0022649
mmu-mir-28-P	0.461 (8 h)	0.0038671
mmu-mir-721-A	0.419 (8 h)	0.0039114
hsa-mir-1	0.074 (8 h)	0.0039493
mmu-mir-297a	0.351 (8 h)	0.01037

Fig. 1.

miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a are down-regulated by thyrotropin via a cAMP/kinase A/CREB1 pathway. A, Validation of miRNA microarray data by qRT-PCR. RNA were extracted from PC Cl 3 cells treated with thyrotropin (10 nm) at the indicated times. The relative expression values indicate the relative change in the expression levels between PC Cl 3 cells treated with thyrotropin and PC Cl 3 cells treated with BSA, assuming that the value of the PC Cl 3 cells treated with BSA was equal to 1. Each bar represents the mean ± se from three independent experiments performed in triplicate. B, qRT-PCR analysis of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a was carried out on PC Cl 3 cells treated with forskolin (10 μm) at the indicated times. The relative expression values indicate the relative change in the expression levels between PC Cl 3 cells treated with forskolin and PC Cl 3 cells treated with dimethylsulfoxide (DMSO), assuming that the value of the PC Cl 3 cells treated with DMSO was equal to 1. Each bar represents the mean ± se from three independent experiments performed in triplicate. C, qRT-PCR analysis of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a was carried out on PC Cl 3 cells treated for 2 h with thyrotropin or thyrotropin plus H89 or H89 alone at the doses (1 and 10 μm). Each bar represents the mean ± se from three independent experiments performed in triplicate. D, qRT-PCR analysis of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a was carried out on PC Cl 3 cells treated for 2 h with thyrotropin or BSA after thyrotropin starvation in the presence or absence of shRNA for CREB1. The treatment with shRNA for CREB1 is of 48 h. The relative expression values indicate the relative change in the expression levels between PC Cl 3 cells in the presence or absence of shRNA for CREB1 (shCREB1) and treated with thyrotropin or BSA, assuming that the value of the PC Cl 3 cells treated with BSA was equal to 1. Western blot analysis (lower panel) of CREB1 proteins in total cell extracts from PC Cl 3 cells transiently transfected with shCREB1. α-Tubulin expression was used to normalize the amount of loaded proteins.

To verify that the down-regulation of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a was a direct effect of thyrotropin, we treated PC Cl 3 cells with forskolin (10 μm). This compound stimulates adenylate cyclase activity and increases intracellular cAMP level, thus activating cAMP-dependent protein kinase and other cAMP receptor proteins and thereby mimicking the effects of thyrotropin. Forskolin treatment resulted in a decrease in the expression of all five miRNA as early as 30 min after exposure to forskolin (Fig. 1B).

To determine whether the thyrotropin-induced down-regulation of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a is kinase A dependent, we measured their expression after treatment for 2 h with thyrotropin and the kinase A inhibitor H89 at two concentrations (1 and 10 μm) in comparison with the PC Cl 3 cells treated with BSA and H89 or the thyrotropin alone. As shown in Fig. 1C, treatment with H89 associated to thyrotropin led to the up-regulation of all miRNA, with the exception of only miR-1, vs. cells treated with thyrotropin alone. This effect was dependent on the H89 dosage, which indicates that the down-regulation of these miRNA is kinase A dependent. The treatment with H89 alone did not result in a significant change in the expression of all five miRNA (Fig. 1C).

We next asked whether the down-regulation of the five miRNA was dependent on the activation of the CREB1 transcription factor. Indeed, through activation of its receptor, thyrotropin stimulates the cAMP pathway, eventually leading to CREB1 phosphorylation and activation of CREB1 transcriptional activity. Therefore, the PC Cl 3 cells were treated with short hairpin RNA (shRNA) for CREB1 for 48 h before thyrotropin stimulation, and we measured the expression of the five miRNA in PC Cl 3 cells treated with thyrotropin (2 h) in the presence of shRNA for CREB1 or scrambled oligonucleotides. As shown in Fig. 1D, in PC Cl 3 cells in which shRNA significantly reduced CREB1 protein levels, thyrotropin-dependent miRNA were not down-regulated, whereas they were dramatically down-regulated in the PC Cl 3 cells treated with thyrotropin in the presence of the scrambled oligonucleotides. A weak down-regulation of miR-1 was observed in the cells where CREB1 expression was blocked. These results would suggest that miR-1 is regulated by thyrotropin through mechanisms that are independent of PKA activation and only partially of thyrotropin-CREB1 pathway.

To further confirm the role of CREB1 in down-regulation of miRNA, we have analyzed the expression of the thyrotropin-down-regulated miRNA in PC Cl 3 cells overexpressing CREB1 proteins. As shown in Supplemental Fig. 1B, the expression of these miRNA is drastically reduced by CREB1 overexpression. Taken together, these results indicate the presence of a thyrotropin-cAMP-CREB1 signaling pathway that induces miRNA down-regulation, with the exception of miR-1, which is likely regulated by other mechanisms.

CREB1 protein directly binds the miRNA regions located upstream or downstream the miRNA sequences

To determine the molecular mechanisms underlying the negative regulation of miRNA expression by the thyrotropin-CREB pathway, we examined the region encompassing 3000 bp upstream and downstream of the sequence of the thyrotropin-down-regulated miRNA. We identified putative CREB recognition sites in all five miRNA (miR-1 at −2400 bp, miR-28-A at −852 bp, miR-290-5p at +540 bp, miR-296-3p at −1140 bp, and miR-297a at −2640 bp) (Fig. 2A). Subsequently, to evaluate whether the CREB1 protein is able to bind to these miRNA regulatory regions in vitro, we performed a chromatin immunoprecipitation (ChIP) analysis in PC Cl 3 cells treated for 1 h with forskolin (10 μm) and immunoprecipitated chromatin with anti-CREB1 antibodies or rabbit IgG used as control. The results reported in Fig. 2B show that the CREB1 protein binds to these sequences. In fact, the miRNA regions located upstream or downstream the miRNA sequences were amplified from the DNA recovered with anti-CREB1 antibody in PC Cl 3 cells when analyzed by quantitative PCR using specific primers spanning these regions. Conversely, no amplification was observed with anti-IgG precipitates and with primers for the regulatory region of the miR-548c, that is not regulated by thyrotropin (data not shown) and does not have a CRE element (Fig. 2B). Interestingly, the amount of CREB1 protein bound to these putative regulatory regions was much more abundant in cells treated with thyrotropin or forskolin than in untreated cells (data not shown). No amplification was observed in cells treated with CREB1 shRNA (Fig. 2C) or with thyrotropin for 24 h (data not shown).

Fig. 2.

CREB1 binds to the promoters of the miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a genes. A, Schematic representation of the CRE element in the regions upstream or downstream of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a. B, ChIP assay was carried out on PC Cl 3 cells treated with forskolin (10 μm) for 1 h. C, ChIP assay was carried out on PC Cl 3 cells treated with CREB1 shRNA for 48 h. The precipitated DNA for CREB1 antibodies was used as a template for qRT-PCR with primers that amplify the upstream or downstream regions of miRNA containing the CRE element and with primers that amplify the region of miR-548c, which did not contain the CRE element. The percentage of chromatin indicates the change in the immunoprecipitated chromatin by CREB1 antibody with respect to IgG used as control.

miR-1, miR-28-A, and miR-296-3p target CREB1, miR-297a targets CCND2, and miR-296-3p and miR-297a target junB

Because miRNA are able to modulate gene expression at posttranscriptional level by directly cleaving mRNA or repressing mRNA translation (14), we sought to identify the targets of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a. Using bioinformatic tools, we identified several genes that could be targeted by these miRNA. The candidate target of three of the thyrotropin-down-regulated miRNA (miR-1, miR-28-A, and miR-296-3p) was the CREB1 gene, which encodes the CREB1 protein that plays a relevant role in thyroid cell proliferation (15). For each seed sequence of these three miRNA, there was a matching sequence in the 3′-UTR of the CREB1 gene (Fig. 3A). To determine the effect of miR-1, miR-28-A, and miR-296-3p on the CREB1 target, we transfected these miRNA into the PC Cl 3 cells and used Western blot analysis to look for changes in CREB1 protein levels. Overexpression of miR-1, miR-28-A, and miR-296-3p in the PC Cl 3 cells decreased CREB1 protein levels (Fig. 3B) in comparison with the cell transfected with the scrambled oligonucleotide or the miR-548c that does not have matching sequences in the 3′-UTR of the CREB1 gene. The cells transfected with these miRNA in combination show very low CREB1 levels, indicating a synergistic effect of these miRNA on CREB1 expression. There were no significant changes in the CREB1 mRNA levels in cells transfected with miR-548c, miR-1, miR-296-3p, or the scrambled oligonucleotide (Fig. 3C). This result is consistent with posttranscriptional regulation of the CREB1 protein by miR-1 and miR-296-3p and demonstrates that these miRNA are not involved in CREB1 mRNA degradation. Conversely, CREB1 mRNA levels decreased after transfection with miR-28-A. It is likely that this miRNA reduces CREB1 protein level also by affecting CREB1 mRNA stability.

Fig. 3.

CREB1 is a target of miR-1, miR-28-A, and miR-296-3p. A, Schematic representation of the 3′-UTR sites of the CREB1 gene targeted by miR-1, miR-28-A, and miR-296-3p. B, Immunoblots of the CREB1 and vinculin proteins, used as loading control. One representative out of three independent experiments. The lower panel represents the mean densitometric analysis of three independent experiments. Proteins were extracted from PC Cl 3 cells transfected with scrambled oligonucleotide, miR-548c, miR-1, miR-28-A, and miR-296-3p (50 nmol/ml) alone or in combination (mix miR) and collected after 48 h. C, qRT-PCR analysis of CREB1 mRNA in the same samples shown in B. Relative expression values indicate the relative change in CREB1 mRNA expression levels between miR-treated and scrambled oligonucleotide-treated cells, normalized with glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The bars represent the mean ± se (n = 3); P < 0.05 vs. scrambled control. D, Relative luciferase activity in PC Cl 3 cells transiently transfected with wild-type and mutant constructs for miRNA seed sequences and with miR-1, miR-28-A, and miR-296-3p oligonucleotide and a control nontargeting scrambled oligonucleotide. The relative activity of firefly luciferase expression was standardized to a transfection control, using Renilla luciferase. The bars represent the mean ± se (n = 3); P < 0.05 vs. the scrambled oligonucleotide. E, qRT-PCR analysis of AREG, NR4A2, and CCNA2 specific mRNA in PC Cl 3 cells transfected with miR-1, miR-28-A, miR-296-3p, or the scrambled oligonucleotide (50 nmol/ml). Relative expression values indicate the relative change in gene expression levels between miR-treated and scrambled oligonucleotide-treated cells, normalized with GAPDH. The bars represent the mean ± se (n = 3); P < 0.05 vs. scrambled control.

Interestingly no changes in the expression of other members of the CREB1 family, such as cAMP responsive element modulator and activating transcription factor 1 was observed after miR-1, miR-28-A, and miR-296-3p overexpression (Supplemental Fig. 2A).

Most miRNA are thought to control gene expression by base pairing with the miRNA-recognizing elements located in their messenger target. To determine whether direct interaction between miRNA and CREB1 mRNA resulted in the decreased expression of the CREB1 protein, we inserted 963 bp (21–984 bp) of the 3′-UTR of the CREB1 mRNA downstream of the luciferase open reading frame. This reporter vector was transfected into the PC Cl 3 cells together with miR-1, miR-28-A, and miR-296-3p. Luciferase activity was much lower after miR-1, miR-28-A, and miR-296-3p transfection (Fig. 3D) than after transfection with the scrambled oligonucleotide. A further decrease of the luciferase activity was obtained when the same miRNA were used in combination (Supplemental Fig. 2B). This conclusion is further supported by similar experiments, in which we used as a reporter construct the same vector of the previous experiments but carrying the respective target sites modified by introducing a mutation in them (Fig. 3D). These reporter vectors carrying mutations in the target sites were insensitive to the effect of miR-1, miR-28-A, and miR-296-3p, proving that the modification of the target sites of CREB1 3′-UTR is able to block the function of these miRNA (Fig. 3D).

To verify that CREB1 is targeted by miR-1, miR-28A, and miR-296-3p, we measured the expression level of the CREB1 targets amphiregulin (AREG) (16), nuclear receptor subfamily 4, group A, member 2 (NR4A2) (17), and cyclin A2 (18), in PC Cl 3 cells transiently expressing these miRNA. As shown in Fig. 3E, these CREB1 target genes were significantly lower in the miRNA-transfected cells than in cells transfected with a scrambled oligonucleotide.

We identified CCND2 as a candidate target of miR-297a. Indeed, one site in the 3′-UTR of the CCND2 gene that matches the miR-297a seed sequences was predicted. The experiments shown in Fig. 4, left panels, confirm that the CCND2 gene is a target of miR-297a. In fact, Western blot analysis (Fig. 4C) shows that CCND2 protein levels are lower in the miR-297a-transfected cells than in cells transfected with miR-548c or the scrambled oligonucleotide, whereas CCND2 mRNA expression levels are unchanged (Fig. 4E). Moreover, miR-297a transfection down-regulated the luciferase activity of a reporter construct carrying the 3′-UTR CCND2 sequence but not of a reporter construct carrying the CCND2 3′-UTR mutated in the seed sequence for miR-297a (Fig. 4G).

Fig. 4.

CCND2 is a target of miR-297a, and junB is a target of miR-296-3p and miR-297a. Schematic representation of CCND2 gene 3′-UTR target site for miR-297a (A) and of junB gene 3′-UTR target sites for miR-296-3p and miR-297a (B). Immunoblots for CCND2 (C), junB (D), and vinculin protein, as loading control. One representative out of three independent experiments. The lower panel represents the mean densitometric analysis of three independent experiments. Proteins were extracted from scrambled oligonucleotide, miR-548c and miR-297a sense-transfected PC Cl 3 cells (C), or miR-296-3p and miR-297a-transfected PC Cl 3 cells (D) 48 h after transfection. qRT-PCR analysis of CCND2 (E) and of junB (F) mRNA in the same samples shown in B and D. Relative expression values indicate the relative change in CCND2 and junB mRNA expression levels between miR-treated and scrambled oligonucleotide-treated cells, normalized with glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The bars represent the mean ± se (n = 3); P < 0.05 vs. scrambled control. The 3′-UTR of CCND2 gene enables miR-297a regulation (G), and the 3′-UTR of junB gene enables miR-296-3p and miR-297a regulation (H). Relative luciferase activity in PC Cl 3 cells transiently transfected with wild-type and mutant constructs for miRNA seed sequences and with miR-297a or miR-296-3p oligonucleotide and a control nontargeting scrambled oligonucleotide. Relative activity of firefly luciferase expression was normalized to a transfection control, using Renilla luciferase. The bars represent the mean ± se (n = 3); P < 0.05 compared with the scrambled oligonucleotide.

Finally, we identified junB as a candidate target of miR-297a and miR-296-3p. In fact, two sites in the 3′-UTR of the junB gene were predicted to match the miR-297a and miR-296-3p seed sequences, respectively. The experiments shown in Fig. 4, right panels, confirm that the junB gene is a target of miR-296-3p and miR-297a. In fact, Western blot analysis shows decreased junB protein levels in the miR-296-3p- or miR-297a-transfected cells compared with cells transfected with a scrambled or miR-548c oligonucleotide (Fig. 4D). Moreover, miR-297a and miR-296-3p transfection down-regulated the luciferase activity of a reporter construct carrying the 3′-UTR junB sequences but not of a reporter construct carrying the junB 3′-UTR mutated in the seed sequence for miR-296-3p and miR-297a (Fig. 4H).

Because a significant decrease in junB mRNA level was observed with miR-297a, but not with miR-296-3p (Fig. 4F), we can assume that miR-297a affects junB protein synthesis by facilitating the specific degradation of junB mRNA.

The regulation of CREB1, CCND2, and junB protein levels by these miRNA is supported by Western blot analysis, showing that CREB1, CCND2, and junB levels are inversely correlated with miR expression in rat thyroid cells stimulated to proliferate by thyrotropin (Fig. 5).

Fig. 5.

CREB1, CCND2, and junB protein expression is inversely correlated with the thyrotropin-down-regulated miR expression. Western blot analysis of the expression of CREB1 (A), CCND2 (B), and junB (C) proteins in PC Cl 3 cells treated with thyrotropin. The level of vinculin protein served as loading control. One representative out of three independent experiments. The table under the Western blot represents the mean densitometric analysis of three independent experiments. qRT-PCR analysis of miR expression is shown in each lower panel.

miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a overexpression inhibits proliferation of rat thyroid cells

To determine whether the thyrotropin-induced down-regulation of the five miRNA analyzed affects thyroid cell growth, we evaluated the growth potential of PC Cl 3 cells transiently expressing the thyrotropin-down-regulated miRNA using an XTT assay 96 h after seeding. As shown in Fig. 6A, the growth rate of PC Cl 3 cells expressing each miRNA (miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a) was significantly lower than in cells treated with a scrambled oligonucleotide. We carried out a colony-forming assay of PC Cl 3 cells after transfection with vectors carrying miR-1, miR-28A, miR-290-5p, miR-296-3p, miR-297a, or the backbone vector. As shown in Fig. 6B, thyroid cells transfected with each of the five miRNA genes generated fewer colonies than cells transfected with the backbone vector. We next investigated the cell cycle phase distribution of the miR-transfected cells. As shown in Fig. 6C, the S phase population was lower in PC Cl 3 cells expressing miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a than in PC Cl 3 cells treated with a scrambled oligonucleotide. Conversely, the pool population in the sub-G₁ phase of the cell cycle, which corresponds to apoptotic cells, was dramatically higher in miRNA-transfected cells (apart from those transfected with miR-297a). This indicates that overexpression of these miRNA exerts an apoptotic effect, probably consequent to the induced inability to overcome G₁ phase. In the case of miR-297a-transfected cells, the number of cells in sub-G₁ phase remained unchanged, whereas there was an increase of cells in G₁. Taken together, these data demonstrate that overexpression of thyrotropin-down-regulated miRNA exerts a negative effect on thyroid. Interestingly, PC Cl 3 cells transfected with the down-regulated miRNA in combination show a decrease in cell growth (50%) (Fig. 6A) and a reduction in the number of colonies (90%) (Fig. 6B) higher than that obtained by using a single miRNA.

Fig. 6.

Overexpression of miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a inhibits thyroid cell proliferation. A, Cell proliferation assay of PC Cl 3 cells transfected with miR-1, miR-28-A, miR-290-5p, miR-296-3p, miR-297a alone or in combination (mix miR), or the scrambled oligonucleotide (50 nmol/ml) and seeded in 96-well plates at 1 × 10⁴ cells/well. The transfection efficiency is about 85%. After 96 h, 20 μl of Promega's CellTiter 96 AQueous One Solution were dispensed into each well and absorbance measured at 595 nm to evaluate cell viability. Each bar represents the mean ± se from three independent experiments performed in triplicate; P < 0.05 compared with the scrambled oligonucleotide. B, Colony-forming assay of PC Cl 3 cells transfected with a vector expressing miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a (SBI's miRNA Precursor Constructs) under the transcriptional control of the CMV promoter. The empty vector served as a control. The same miR were also transfected in combination (mix miR). Transfections were performed in triplicate, and the results are reported as the mean ± se of these experiments; P < 0.05 compared with the scrambled oligonucleotide. The transfection efficiency is about 90%. C, Flow cytometric analysis of PC Cl 3 cells transfected with miR-1, miR-28-A, miR-290-5p, miR-296-3p, miR-297a, or the scrambled oligonucleotide. After transfection, the DNA of the transfected PC Cl 3 cells was analyzed 72 h later by flow cytometry after propidium iodine staining. The transfection efficiency is about 85%. Each bar represents the mean ± se from three independent experiments performed in triplicate; P < 0.05 compared with the scrambled oligonucleotide.

Discussion

In this study, we examined the role of miRNA expression in the proliferation of rat thyroid cells. We show that treatment of thyroid cells with thyrotropin, which induces entry into S phase of the cell cycle, is associated with the down-regulation of at least five miRNA. The expression level of these miRNA decreases as early as 30 min after thyrotropin stimulation and lasts for at least 8 h before returning to the levels observed in quiescent cells. Our data indicate that thyrotropin exerts this effect by activating the cAMP pathway, because similar results were obtained by treating the cells with forskolin, a cAMP synthesis stimulator, and by blocking CREB1 protein synthesis that inhibits thyrotropin-induced down-regulation of these miRNA.

Three of the miRNA studied (miR-1, miR-28-A, and miR-296-3p) target the CREB1 transcription factor that has a relevant role in thyroid cell proliferation. In fact, by activating its specific receptor in thyroid cells, thyrotropin induces the cAMP pathway and, consequently, activates the CREB transcription factor through its phosphorylation. Importantly, all the miRNA down-regulated by thyrotropin are located near CRE-responsive elements to which CREB1 is able to bind in vivo as demonstrated by ChIP analysis (Fig. 2). Therefore, our results suggest the existence of a regulatory synergistic loop that involves thyrotropin, CREB1, and the miR-1, miR-28-A, miR-290-5p, miR-296-3p, and miR-297a and eventually leads to increased expression of CREB1-target genes.

We also show that the CCND2 gene, which encodes the G₁/S-specific CCND2 protein, is a target of miR-297a. This cyclin forms a complex with and functions as a regulatory subunit of cyclin-dependent kinase 4 or cyclin-dependent kinase 6; the activity of these complexes is required for the G₁/S cell cycle transition. The CCND2 protein is involved in the phosphorylation of the tumor suppressor retinoblastoma protein. The CCND2 gene is overexpressed in ovarian and testicular tumors (19). Another target of the miR-296-3p and miR-297a is the gene coding for the junB protein that, in complex with the proteins of the fos family, forms the activator protein-1 transcription factor whose induction is critical for cell proliferation (20). The up-regulation of CCND2 and junB by mir-297a and miR-296-3p down-regulation surely contributes to thyroid cell proliferation induced by thyrotropin, even though it is likely that CREB1 up-regulation plays a critical role in the regulation of thyroid cell growth. Therefore, down-regulation of the thyrotropin-regulated miRNA potentiates the thyrotropin effect on thyroid cell proliferation, which also supports the presence of a thyrotropin-CREB1-miRNA loop that acts synergistically. Of course, this mechanism does not exclude a role of the thyrotropin pathway on the same genes at transcriptional level, as already published (21). Therefore, the down-regulation of specific miR, acting also at translational level, would potentiate the thyrotropin effects.

Finally, we demonstrate that all the thyrotropin-down-regulated miRNA inhibit thyroid cell growth by reducing their ability to proceed to S phase of the cell cycle. Moreover, overexpression of four of these miRNA induces PC Cl 3 cell apoptosis; the inability of miR-transfected cells to enter S phase might account for the induced cell death. These results suggest that these miRNA might be down-regulated in thyroid carcinoma, and maybe their expression could have a role also in thyroid differentiation. Indeed, our preliminary data indicate that miR-1, miR-28-A, and miR-297a are down-regulated in papillary thyroid carcinomas (Leone, V., and D'Angelo, D. unpublished results) and that miR-1 and miR-28-A expression affects the expression of the main thyroid markers (data not shown). It is likely that this inhibitory effect on thyroid differentiation is mediated by CREB1 down-regulation.

It is noteworthy that this analysis of the miRNA expression profile in thyrotropin-stimulated PC Cl 3 cells also shows that the up-regulation of a set of miRNA (i.e. miR-130b, miR-24, miR-29b, and miR-23) has as candidate or validated target genes phosphatase and tensin homolog, p21, p63, and p53, which negatively affect the regulation of cell proliferation (data not shown). Therefore, the positive and negative regulation of miRNA expression by thyrotropin is critical for the thyrotropin stimulatory effect. However, the analysis and characterization of the thyrotropin-up-regulated miRNA will be subject of further study.

In conclusion, this study demonstrates that down-regulation of a set of specific miRNA is a critical event for thyroid cell proliferation. This finding supports the idea that the regulation of miRNA expression synergizes with the traditional proliferation pathways in promoting cell growth.

Materials and Methods

Cell lines and transfections

PC Cl 3 and FRTL5 cells are described elsewhere (13, 22). They were grown in Ham's F-12 medium, Coon's modification (Sigma-Aldrich, Milan, Italy) supplemented with 5% calf serum (Life Technologies, Inc., Paisley, PA) in the presence of a mix containing six growth factors (10 nm thyrotropin, 10 nm hydrocortisone, 100 nm insulin, 5 μg/ml transferrin, 5 nm somatostatin, and 20 μg/ml glycyl-histidyl-lysine). After 24 h, the medium was changed, and cells were made quiescent by incubation in the presence of medium plus 0.23% BSA for 72 h. Then, thyrotropin was added to this culture medium, and RNA were extracted from thyrotropin-treated and BSA-treated cells at the various time points. Triplicate cultures were assayed for each time point.

For transfection of miRNA oligonucleotide, 2.5 × 10⁵ cells seeded in six-well plates were transfected with 50 nmol/ml pre-miR miRNA precursor or a control nontargeting scrambled oligonucleotide (Ambion, Austin, TX) using siPORT neoFX Transfection Agent (Ambion) according to the manufacturer's instructions. The transfection efficiency was about 85% evaluated by transfection of AllStars Neg. small interfering RNA Alexa Fluor 488 (QIAGEN, Valencia, CA). H89 was obtained from Calbiochem (San Diego, CA). Forskolin was obtained from Sigma (St. Louis, MO).

RNA extraction and qRT-PCR

Total RNA was isolated from cells with Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. The integrity of the RNA was assessed by denaturing agarose gel electrophoresis (virtual presence of sharp 28S and 18S bands). RT-PCR and qRT-PCR for mature miRNA were carried out according to the manufacturer's instructions of the miScript System kits (QIAGEN, Hilden, Germany). Reactions contained miScript Primer Sets (QIAGEN), specific for miR-1, miR-28-A, miR-290-5p, miR-296-3p, miR-297a, and U6 (used to normalize RNA levels). qRT-PCR analysis for CREB1, CCND2 and junB, CCNA2, AREG, and NR4A2 were performed as previously described (23). Each reaction was carried out in duplicate. To calculate the relative expression levels, we used the 2^−ΔΔCT method (24). The sequences of the primers we used are:

• Ccnd2 forward (F) CACCGACAACTCTGTGAAGC,

• Ccnd2 reverse (R) CCACTTCAGCTTACCCAACAC;

• CREB1 F CTAGTGCCCAGCAACCAAGT,

• CREB1 R GGAGGACGCCATAACAACTC;

• junB F GCCACCGAGACCGTAAAG,

• junB R CTGTGCGAGCTGGTATGAGT;

• CCNA2 F AGTGCCGCTGTCTCTTTACC,

• CCNA2 R CATAGCATGGGGTGATTCAA;

• AREG F GGTGAATGCAGATACATCGAGA,

• AREG R CGTTCGCCAAAGTAATCCTG;

• NR4A2 F CCACGTCGACTCCAATCC,

• NR4A2 R TAGTCAGGGTTTGCCTGGAA.

miRNACHIP microarray

Microarray experimental procedures were performed as previously described (25). The experiment has been performed twice. Each sample has been analyzed for miRNA expression profile three times.

Plasmids and constructs

MiRNA Precursor Constructs expressing miR-1, miR-28-A, and miR-290-5p under the transcriptional control of the cytomegalovirus (CMV) promoter were from SBI System Biosciences (Mountain View, CA). pEGP-miR expression vector expressing miR-296-3p and miR-297a were from Cell Biolabs (San Diego, CA).

Short hairpin RNA expression constructs specific for rat CREB1 and for scrambled control were from GeneCopoeia (Rockville, MD). The 3′-UTR region of the CREB1 gene, including binding sites for miR-1, miR-28-A, and miR-296-3p, the 3′-UTR CCND2 region, including binding sites for miR-297a, and the 3′-UTR of the junB gene, including binding sites for miR-296-3p and miR-297a, were amplified by PCR from rat genomic DNA by using the primers:

• 3′-UTR-CREB1 F AATTTCTAGACTGTTAAGGTGGAAAATGGACTG,

• 3′-UTR-CREB1 R AATTTCTAGAGCTTTCTTGGTGGTGGTATGTAA;

• 3′-UTR-cycD2 F AATTTCTAGAGCCTGCAATATGGGAACAAA,

• 3′-UTR-cycD2 R AATTTCTAGACCATCCCAAATGGTTCAATC;

• 3′-UTR-junB F AATTTCTAGAGAGCCTCCCTTGCTCCATAC,

• 3′-UTR-junB R AATTTCTAGATGCGTGTTTCTTCTCCACAG.

The amplified fragments were cloned into pGL3-Control Firefly luciferase vector (Promega, Madison, WI) at the XbaI site immediately downstream from the stop codon of luciferase.

Mutation into miR-1, miR-296-3p, and miR-28-A binding sites of the 3′-UTR of CREB1 gene, into miR-296-3p and miR-297a binding sites of the 3′-UTR of junB gene, and into miR-297a binding sites of the 3′-UTR of CCND2 gene were introduced by using QuikChange Site-directed Mutagenesis kit (Stratagene, La Jolla, CA) following the manufacturer's instructions.

The primers used are:

• 3′UTR-CREB1-1MutF AAGACAAAATAAACATTTTATTTTCTAAACATTTCTTTTT,

• 3′UTR-CREB1-1MutR AAAAAGAAATGTTTAGAA AATAAAATGTTTATTTTGTCTT;

• 3′UTR-CREB1-28MutF ATTTCAATTTGCTGGATTGCCTTAGGGACAGAATTACCCCAG,

• 3′UTR-CREB1-28MutR CTGGGGTAATTCTGTCCCTAAGGCAATCCAGCAAATTGAAAT;

• 3′UTR-CREB1-296MutF TCTAATTTTGTTGCCTTGCTTCACCACCAAGAAAGC,

• 3′UTR-CREB1-296MutR GCTTTCTTGGTGGTGAAGCAAGGCAACAAAATTAGA;

• 3′UTR-jun297MutF CCCTGGCCTGCCTCTCTCAATCATCTGTGTACATATATTTTTTTT,

• 3′UTR-jun297MutR AAAAAAAATATATGTACACAGATGATTGAGAGAGGCAGGCCA;

• 3′UTR-jun296MutF CAGGCGATAAAAACTCGTTTAGAGGTCTGGGGCGCAGCTCAC,

• 3′UTR-jun296MutR GTGAGCTGCGCAAAAGACCTTTTTTCGAGTGCCCATCGCCTG;

• 3′UTR-ccnd2MutF GGAGACTTTTTTTTTTCTCATTGTTCGCTAGCACATACACCC,

• 3′UTR-ccnd2MutR GGGTGTATGTGCTAGCGAACAATGAGAAAAAAAAAAGTCTCC.

Luciferase target assays

1 × 10⁵ PC Cl 3 cells were cotransfected using siPORT neoFX Transfection Agent (Ambion) in 12-well plates with the modified firefly luciferase vector (0,2 μg) described above, the Renilla luciferase reporter plasmid (pRL-CMV; Promega) (30 ng), and the RNA oligonucleotides (50 nmol/ml). Firefly and Renilla luciferase activities were measured 24 h after transfection with the Dual-Luciferase Reporter Assay System (Promega). Firefly activity was normalized to Renilla activity as control of transfection efficiency.

ChIP assay

Chromatin samples were processed for ChIP experiments as reported elsewhere (26). Samples were subjected to immunoprecipitation with the specific α-CREB1 antibody. To calculate the percentage of total chromatin, we used the 2^−ΔΔCT method (24). The sequences of the primers we used are:

• F-cre1r ATGGTCCCAACACTGCTTTG,

• R-cre1r GAGACACGTCTTTATTTGGCAGT;

• F-cre28r CGTCAGTGATTCTTGAGGATAAGA,

• R-cre28r TCATCAACCTCCTTACTGACTCTTT;

• F-cre290r TCGGTGTCTCTTATTTCACCCTA,

• R-cre290r AGCTAGGCTTAGGGTTTCTATCCT;

• F-cre296r AAGAGGCTAGAAGAGTACAGGTTCC,

• R-cre296r GACGTCTTACAATCGTCCTGTG;

• F-cre297r CAGTGCCCATGAGGATCAG,

• R-cre297r AGATGTCTCCGCTTTCTTTAGC;

• F-548r GCCCAGGCTGGAGAGCAATG,

• R-548r CAGGCACGGTGGCTCACACC.

Protein extraction, Western blotting, and antibodies

Cells were lysed in lysis buffer containing 1% Nonidet P-40, 1 mm EDTA, 50 mm Tris-HCl (pH 7.5), and 150 mm NaCl, supplemented with complete protease inhibitors mixture (Roche Diagnostic, Mannheim, Germany). Total proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes (Amersham, Rainham, UK). Membranes were blocked with 5% nonfat milk and incubated with antibodies against CREB1 (sc-58), junB (sc-46), CCND2 (sc-754), activating transcription factor 1, and CREM (Santa Cruz Biotechnology Inc., Santa Cruz, CA).

Flow cytometry

After trypsinization, cells were washed once in PBS and fixed in 70% ethanol overnight. Staining for DNA content was performed with 2 mg/ml propidium iodide and 20 mg/ml ribonuclease A for 30 min. We used a FACScan flow cytometer (Becton Dickinson, San Jose, CA) that was interfaced with a Hewlett-Packard computer (Palo Alto, CA). Cell cycle data were analyzed with the CELL-FIT program (Becton Dickinson).

XTT assay

Transfected cells using siPORT neoFX Transfection Agent (Ambion) were seeded in 96-well plates at 1 × 10⁴ cells/well. After 96 h, 20 μl of Promega's CellTiter 96 AQueous One Solution was dispensed into each well and absorbance measured at 595 nm to evaluate cell viability.

Colony forming assay

Cells, plated at a density of 90% in 100-mm dishes, were transfected using lipofectamine 2000 with 5 μg of MiRNA Precursor Constructs expressing miR-1, miR-28-A and miR-290-5p, miR-296-3p, miR-297a, or scramble and supplemented with drug geneticin (G418) 24 h later. Two weeks after the onset of drug selection, the cells were fixed and stained with crystal violet (0.1% crystal violet in 20% methanol).