Thyroid Transcription Factor 1 Is Calcium Modulated and Coordinately Regulates Genes Involved in Calcium Homeostasis in C Cells

Abstract

Thyroid transcription factor 1 (TTF-1) was identified for its critical role in thyroid-specific gene expression; its level in the thyroid is regulated by thyrotropin-increased cyclic AMP levels. TTF-1 was subsequently found in lung tissue, where it regulates surfactant expression, and in certain neural tissues, where its function is unknown. Ligands or signals regulating TTF-1 levels in lung or neural tissue are unknown. We recently identified TTF-1 in rat parafollicular C cells and parathyroid cells. In this report, we show that TTF-1 is present in the parafollicular C cells of multiple species and that it interacts with specific elements on the 5′-flanking regions of the extracellular Ca²⁺-sensing receptor (CaSR), calmodulin, and calcitonin genes in C cells. When intracellular Ca²⁺ levels are increased or decreased in C cells, by the calcium ionophore A23187, by physiologic concentrations of the P₂ purinergic receptor ligand ATP, or by changes in extracellular Ca²⁺ levels, the promoter activity, RNA levels, and binding of TTF-1 to these genes are, respectively, decreased or increased. The changes in TTF-1 inversely alter CaSR gene and calcitonin gene expression. We show, therefore, that TTF-1 is a Ca²⁺-modulated transcription factor that coordinately regulates the activity of genes critical for Ca²⁺ homeostasis by parafollicular C cells. We hypothesize that TTF-1 similarly coordinates Ca²⁺-dependent gene expression in all cells in which TTF-1 and the CaSR are expressed, i.e., parathyroid cells, neural cells in the anterior pituitary or hippocampus, and keratinocytes.

Thyroid transcription factor 1 (TTF-1) is a homeodomain protein that was initially identified as a thyroid-specific factor responsible for thyroglobulin (TG) gene transcription (24, 42). It was later shown to be important in the thyroid for maximal and tissue-specific expression of the thyroperoxidase (TPO), thyrotropin (TSH) receptor (TSHR), and sodium iodide symporter (NIS) genes (14, 19, 21, 41, 45, 57), as well as for maximal expression of the major histocompatibility complex (MHC) class I gene (54). In the thyroid, TTF-1 is, therefore, a transcription factor which can regulate the expression of ubiquitous as well as tissue-specific genes. In the thyroid, TSH-cyclic AMP (cAMP) regulates TTF-1 levels and thereby coordinately regulates the above genes. This process decreases MHC class I levels and preserves self-tolerance in the presence of TSH-cAMP-induced increases in the thyroid-specific proteins important for thyrocyte function (30, 45, 54, 57).

TTF-1 was subsequently identified in lung epithelial cells and in neural cells of restricted regions of the brain during studies of thyroid development (34). It was then shown to be essential for the organogenesis of lung, ventral forebrain, and pituitary, as well as thyroid, tissues in knockout experiments (29). TTF-1 is now recognized as regulating surfactant expression in adult lung tissue (12, 67), but its role as a transcription factor regulating adult neural tissues remains unknown. Other than TSH-cAMP, which decreases TTF-1 gene expression in the thyroid (45, 54, 57), nothing is known about ligands or signals which might regulate TTF-1 levels in other tissues.

In the course of in situ hybridization studies examining the role of TTF-1 in thyroid function in adult rat thyroids in vivo, we identified TTF-1 mRNA in parafollicular C cells and in parathyroid cells (61). This observation had not been previously made in developmental or knockout studies (29, 34). We hypothesized that TTF-1 might be a transcription factor regulating genes involved in extracellular Ca²⁺ homeostasis, since this is the dominant function of these two types of cells.

In the present study, we validate this hypothesis by using a parafollicular C-cell model. We suggest that TTF-1 is a sensor of increased or decreased internal Ca²⁺ levels in C cells and that it provides a transcriptional mechanism to coordinately control, by a single transcription factor, multiple genes affecting Ca²⁺ homeostasis within the cell and in the organism. This mechanism is a hitherto-unrecognized transcriptional regulatory path which complements the secretory signal of high extracellular Ca²⁺ levels by enhancing the ability of C cells to detect changes in extracellular Ca²⁺ levels and by replenishing calcitonin stores used to lower extracellular Ca²⁺ levels, thereby rebalancing Ca²⁺ homeostasis (11, 22, 37, 40). This transcriptional regulatory action should, however, be common to all other cells that express both TTF-1 and the Ca²⁺-sensing receptor (CaSR) and whose growth, differentiation, or function is regulated by Ca²⁺, i.e., not only C and parathyroid cells but also keratinocytes, neural cells in the hippocampus, and anterior pituitary cells (8, 13, 18, 36, 61). We speculate that this may also be true of cells expressing TTF-1 but not the CaSR, i.e., thyroid cells, Purkinje cells, and neural cells in the retina. In sum, these data establish an additional and novel functional role for TTF-1.

MATERIALS AND METHODS

In situ hybridization and immunohistochemistry.

Nonisotopic in situ hybridization with deparaffinized tissue sections fixed in 4% paraformaldehyde was performed as described previously (59, 61). The riboprobe was a 756-bp SacII fragment (bp 149 to 905) of TTF-1 cDNA that had been subcloned into pBluescript SK(−) (Stratagene, La Jolla, Calif.). Antisense and sense riboprobes were transcribed in the presence of digoxigenin-11-UTP (Boehringer Mannheim Biochemica, Mannheim, Federal Republic of Germany) with T7 and T3 RNA polymerases, respectively. An indirect immunoperoxidase method was used to identify calcitonin-expressing C cells after TTF-1 in situ hybridization (61). In brief, sections were treated with 3% H₂O₂ and incubated with rabbit anticalcitonin sera (DAKO, Glostrup, Denmark). Preimmune antisera served as controls. After being washed with phosphate-buffered saline (PBS) (pH 7.5), sections were exposed to 1:100 dilutions of horseradish peroxidase-labeled swine anti-rabbit immunoglobulin G (DAKO) and rinsed with PBS before peroxidase activity was detected with 3,3′-diaminobenzidine solution containing 0.003% H₂O₂. All procedures were done at room temperature.

RNA isolation and Northern analysis.

RNA was prepared by use of RNeasy Mini Kits (Qiagen Inc.) and minor modifications of the manufacturer’s protocol as described previously (60, 61). RNA samples (20 μg, as measured by optical density) were run on denatured agarose gels, capillary blotted on Nytran membranes (11 by 14 cm; Schleicher & Schuell), UV cross-linked, and used for hybridization. Probes were labeled with [³²P]dCTP by use of a Multiprime DNA labeling system (Amersham Life Science Inc.). Membranes were hybridized and washed as described previously (59–61). The rat TTF-1 probe was a 465-bp SacII-BglII fragment excised from a pRcCMV vector (24). The rat β-actin probe was prepared by reverse transcription-PCR (RT-PCR) (see below) with FRTL-5 cell total RNA to amplify 220 bp of specific cDNA and then was purified from an agarose gel by use of a Jetsorb Kit (Genomed Inc.). Rat TG, TSHR, and TPO probes were described previously (27, 53). The glyceraldehyde phosphate dehydrogenase (GAPDH) probe was cut from a pTR1-GAPDH-Rat template (Ambion) and subcloned into a pBluescript SK(−) vector.

RT-PCR.

cDNA was synthesized by use of a First Strand cDNA Synthesis Kit (5 Prime → 3 Prime, Inc.). PCRs were performed by “touchdown” PCR (16) with a GeneAmp 9600 PCR apparatus (The Perkin-Elmer Corp., Norwalk, Conn.), Pfu DNA polymerase (Stratagene), and the following forward and reverse primer pairs (5′→3′): for TTF-1, ACCTTACCAGGACACCATGC and TACTTCTGCTGCTTGAAGCG; for CaSR, ACAAGCCATGATATTCGCC and TTGGATCACTTCGACCACC; for calmodulin III (CAM-III), CTGACTGAAGAGCAGATCGC and TCTGCTGCACTGATGTAGCC; for E-cadherin (E-CAD), AGAAGGAGGTGGAGAAGAAGACC and AGTGAGACCACTATGCAATCTGC; for calcitonin, CTCAGTGAAGAAGAAGCTCGC and GCATGCAGGTACTCAGATTCC; and for rat β-actin, AGCCATGTACGTAGCCATCC and TGTGGTGGTGAAGCTGTAGC. In each case, these sequences crossed an intron-exon boundary. PCRs without reverse transcriptase served as negative controls.

Nuclear extracts.

A method previously used to prepare nuclear extracts (26) was modified to prepare extracts from small numbers of cells. Cells were washed, scraped into 1 ml of PBS, pelleted in a microcentrifuge, and resuspended in 5 volumes of buffer A (10 mM HEPES-KOH [pH 7.9], 10 mM KCl, 1.5 mM MgCl₂, 0.1 mM EDTA) containing 0.3 M sucrose and 2% Tween 40. To release nuclei, cells were frozen and thawed once and then repetitively pipetted (50 to 100 times) by use of a micropipette with a yellow tip (200-μl capacity). Samples were overlaid on 1 ml of 1.5 M sucrose in buffer A and microcentrifuged for 10 min at 4°C. Pelleted nuclei were washed with 1 ml of buffer A, centrifuged for 30 s, and then resuspended in 50 μl of buffer B (20 mM HEPES-KOH [pH 7.9], 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 25% glycerol). Samples were placed on ice for 20 min with occasional vortexing and centrifuged for 20 min at 4°C. The supernatant fraction containing the nuclear protein was divided into aliquots and stored at −70°C. Buffers A and B also contained 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 2 ng of pepstatin A per ml, and 2 ng of leupeptin per ml. All procedures were performed on ice or at 4°C.

EMSA.

Oligonucleotides were labeled with [γ-³²P]ATP by use of T4 polynucleotide kinase and then purified on 8% native polyacrylamide gels (26, 57). Electrophoretic mobility shift analysis (EMSA) was performed as described previously (26, 57) with 3 μg of nuclear extract or 20 ng of recombinant TTF-1. In some applications, a 100-fold excess of unlabeled oligonucleotide or 1 μl of rabbit TTF-1 antiserum was added to the mixtures during the preincubation period. A radiolabeled double-stranded oligonucleotide probe (50,000 cpm) was added, and the incubation was continued for 20 min at 4°C. Mixtures were analyzed on 5% native polyacrylamide gels and autoradiographed.

Recombinant TTF-1 was produced by use of the pET system (Novagen, Madison, Wis.) as described by the manufacturer. TTF-1 cDNA was ligated to the EcoRI site of the expression vector pET-30(+); TTF-1 was recovered with elution buffer containing imidazole. The eluted fraction was dialyzed against 20 mM HEPES-KOH (pH 7.9)–100 mM KCl–0.1 mM EDTA–20% glycerol–0.5 mM dithiothreitol–0.5 mM phenylmethylsulfonyl fluoride–2 μg of leupeptin per ml–2 μg of pepstatin per ml and then concentrated in a Centricon 10 concentrator (Amicon, Beverly, Mass.) for use in EMSA.

Plasmids.

The TTF-1 expression vector pRcCMV-THA was constructed by ligating the rat TTF-1 coding sequence with the pRcCMV vector (24). Rat TTF-1 or thyroid enhancer binding protein (T/EBP) promoter-luciferase constructs were prepared by PCR amplification of TTF-1 (T/EBP) gene upstream sequences with a rat genomic clone as a template (28, 41). To clone the 5′-flanking region of the rat CaSR gene, we used the Promoter Finder DNA Walking Kit (Clontech, Palo Alto, Calif.) as described by the manufacture. Briefly, nested PCR was used with an adaptor-ligated rat genomic DNA library. The gene-specific primer for the first PCR was 5′-AATGCACTCCACAGACTCTGGTCTTG-3′; for the second PCR, it was 5′-AGCGTTGCCTTCTCTTCAGGGGACAG-3′. The successfully amplified clone was subcloned into the pCR2.1 TA cloning vector (Invitrogen, Carlsbad, Calif.) and sequenced, and the 1,095-bp fragment was subcloned into the pGL3-Promoter plasmid (Promega, Madison, Wis.) by use of a luciferase reporter gene (Promega).

To make plasmids with one or more copies of the TTF-1 C site on the CaSR, the 24-bp oligonucleotide used for gel shift analyses was ligated in the presence of T4 ligase by adding a complementary 3-bp sequence (TCA and TGA; 5′→3′) to each 5′ end, followed by blunt ending with the Klenow fragment. The mixture was cloned in the SmaI site of pBluescript SK(−); clones containing different multimers of the original oligonucleotide were selected by agarose gel electrophoresis and sequenced (55). Inserts were purified on agarose gels and then subcloned into the pGL3-Promoter vector containing a simian virus 40 (SV40) promoter and the luciferase reporter gene.

pSVGH, used to evaluate transfection efficiency, was a BamHI-EcoRI fragment containing the human growth hormone gene inserted into the BamHI-XbaI site of the pSG5 expression vector (Stratagene) (26). Plasmids were prepared by use of Plasmid Maxi Kits (Qiagen) as described by the manufacturer.

Cell culture.

Buffalo rat liver (BRL) cells (BRL 3A; ATCC CRL 1442), nonfunctioning rat FRT thyroid cells (3), and a fresh subclone (F₁) of functioning rat FRTL-5 thyroid cells (ATCC CRL 8305) (Interthyr Research Foundation, Baltimore, Md.), which had all the functional properties previously detailed (15, 26, 27, 30–32, 45, 53, 54, 57, 59, 60), were grown as described previously (26). Fresh medium was added every 2 or 3 days, and cells were passaged every 7 to 10 days. Rat medullary thyroid carcinoma (rMTC) cells (rMTC 6-23; ATCC CRL 1607) were maintained in Dulbecco’s modified Eagle’s medium supplemented with 2 mM glutamine, 1 mM nonessential amino acids, and 10% heat-inactivated horse serum. Medium was changed every other day, and cells were passaged every 6 to 8 days.

Transient expression analysis.

Lipofectamine (GIBCO BRL, Gaithersburg, Md.) was used to transfect promoter-reporter gene constructs into FRTL-5 or FRT cells. Briefly, cells were grown in six-well plastic plates to about 50% confluency and washed with 2 ml of warmed (37°C) serum-free culture medium (6H0 medium), and 1 ml of a premade plasmid-Lipofectamine mixture was added. The plasmid-Lipofectamine mixture was made by incubation of 1 μg of plasmid DNA with 10 μl of Lipofectamine and 200 μl of 6H0 medium for 30 min at room temperature and then dilution with 800 μl of 6H0 medium. Cells were incubated for 4 h at 37°C in a CO₂ incubator before 4 ml of complete medium with serum was added. Fresh medium was added after 24 h, with or without added ligands or reagents as noted in individual experiments, and reporter activity was measured 4 to 36 h later. To measure luciferase activity, cells were washed, scraped into 1 ml of Dulbecco’s PBS, pelleted in a microcentrifuge, resuspended, and lysed with 30 μl of 1× Reporter Lysis Buffer (Promega) by repetitive pipetting (30 times) by use of a micropipette with a yellow tip (200-μl capacity). The lysate was incubated for 30 min at room temperature, frozen on dry ice, thawed, and then centrifuged at 4°C for 10 min. Twenty microliters of the supernatant was mixed with 100 μl of luciferase assay reagent (Promega) and analyzed with a luminometer (Monolight 2090; Analytical Luminescence Laboratory). Two microliters of the supernatant was taken for protein measurements with bicinchoninic acid protein assay reagent (Pierce) as described by the manufacturer.

Measurement of intracellular calcium levels.

FRTL-5 cells were grown to 60% confluency on a glass disk and exposed to 4 mM Fura-2 (Molecular Probes, Eugene, Oreg.) for 1 h. Cells were then washed, and ATP, ADP, or CTP was added. Specific absorbances at 340 and 380 nm were measured simultaneously with an Axiovert 135 microscope (Carl Zeiss, Oberkochen, Germany) and an Attofluor digital fluorescence microscopy system (Atto Instruments, Rockville, Md.) by a previously described method (33).

Statistical significance.

All experiments were repeated at least three times with different batches of cells. Values are the means ± standard deviations (SD) for these experiments. Significance between experimental values was determined by two-way analysis of variance; significance was set at a P value of <0.05.

RESULTS

TTF-1 mRNA exists in parafollicular C cells of multiple species.

Because the percentage and location of C cells within the thyroid vary widely among species, we initially measured TTF-1 in C cells of multiple species to ensure that the study of TTF-1 function in C cells was broadly relevant. TTF-1 mRNA was visualized by in situ hybridization of rat thyrocytes as a black coloration in the perinuclear region of anticalcitonin serum-stained cells (Fig. 1A). Probe specificity was established by showing that the sense probe did not detect TTF-1 mRNA by in situ hybridization (Fig. 1B) and by Northern analysis (61). Additionally, the antisense TTF-1 riboprobe did not detect TTF-1 mRNA in total RNA preparations from BRL cells or nonfunctioning FRT thyroid cells, both of which lack TTF-1 (Fig. 2A) (45, 57). In the rat, TTF-1 mRNA expression was much stronger in C cells than in thyrocytes (Fig. 1A). The relatively low level of TTF-1 mRNA expression in adult rat thyrocytes in vivo has been explained by data showing that TSH-cAMP and thyroglobulin within the follicular lumen are transcriptional suppressors of TTF-1 gene expression (45, 57, 59, 60). This characteristic may be a major difference between adult thyroid and fetal thyroid (34).

FIG. 1.

TTF-1 RNA is present in rat, dog, and human parafollicular C cells. In situ hybridization with an antisense TTF-1 riboprobe was done with normal rat and dog thyroids as well as human thyroid medullary carcinoma tissue (A, C, and D, respectively). The normal rat and dog thyroids were immunostained with anticalcitonin sera to identify C cells. As a control, the rat thyroid was hybridized with a sense TTF-1 riboprobe and then immunostained with anticalcitonin sera (B). TTF-1 mRNA (black coloration) is observed in the perinuclear area of thyroid follicular epithelium (thin arrows) and parafollicular C cells (thick arrows); calcitonin protein is seen as a brown coloration (arrowheads) of the cytoplasm of parafollicular C cells. Tissues were counterstained with methyl green. Bars, 25 μm.

FIG. 2.

rMTC cells express TTF-1 RNA but not thyroid-specific or restricted genes such as the TG or TSHR genes; TTF-1 in nuclear extracts of rMTC C cells and rat FRTL-5 thyroid cells has the same DNA binding properties. (A) Northern analysis was performed with functioning rat FRTL-5 thyroid cells as a positive control and nonfunctioning rat FRT thyroid cells as well as BRL cells as negative controls. Blots were sequentially hybridized with radiolabeled rat TTF-1, TSHR, TG, and β-actin probes. (B) A radiolabeled oligonucleotide containing the upstream TTF-1 binding element within the TSHR 5′-flanking region was incubated with nuclear extracts from FRTL-5 or rMTC cells and subjected to EMSA. Incubations in lanes 3 and 4 included a 100-fold molar excess of unlabeled oligo C, the TTF-1 or Pax-8 binding site within the TG promoter (24). (C) EMSA shows that anti-TTF-1 inhibits complex formation by C-cell and FRTL-5 cell nuclear extracts. Lanes 1 to 3 contained 3, 1, and 0.1 μg, respectively, of nuclear extract. Lanes 5 to 7 and 9 to 11 contained the same respective amounts of the extract plus 1 μl of immune serum directed against a synthetic TTF-1 peptide. Lanes 4 and 8 contained 3 μg of extract plus 1 μl of preimmune serum. Lane 12 contained an incubation mixture of the radiolabeled probe with 20 ng of recombinant TTF-1 (rTTF-1).

In the dog thyroid, TTF-1 mRNA colocalized in a large number of calcitonin-expressing C cells within interfollicular spaces (Fig. 1C). The number of C cells was far larger than that in the rat thyroid (Fig. 1C versus 1A). Although C cells are very rare in the normal human thyroid and difficult to see within a single tissue section, almost all human medullary thyroid carcinoma cells, which originate from parafollicular C cells of the thyroid, expressed TTF-1 mRNA (Fig. 1D). The presence of TTF-1 in parafollicular C cells is not, therefore, a species-specific phenomenon.

We examined the levels of TTF-1 in continuously cultured rMTC cells, since these cells serve as a potential C-cell model. The level and appearance of TTF-1 RNA in rMTC cells were comparable to those in rat FRTL-5 thyroid cells (Fig. 2). The existence in rMTC cells of high levels of TTF-1 RNA but not of TSHR or TG RNA (Fig. 2) raises the obvious question of the role of TTF-1 in C cells. Thus, rMTC cells do not express the array of genes which are associated with the tissue-specific functions of the thyroid or lung and which are TTF-1 regulated: TG, TPO, TSHR, NIS, or surfactants.

TTF-1 protein is expressed in C cells and has properties of TTF-1 protein in thyroid cells.

To establish that TTF-1 RNA is functionally expressed as TTF-1 protein in C cells, we performed EMSA. We used nuclear extracts and an oligonucleotide containing the upstream TTF-1 binding element of the TSHR as a probe (45). The upstream as well as downstream TTF-1 binding sites in the TSHR are TTF-1 specific and do not act with Pax-8 (45). Nuclear protein from rMTC cells formed a DNA-protein complex which had the same mobility and intensity as that formed by nuclear extract from FRTL-5 cells (Fig. 2B, lane 1 versus lane 2).

To unequivocally establish that the complex was a TTF-1–DNA complex, we first showed that the formation of the complex was inhibited by an excess of an unlabeled oligonucleotide from the TG promoter, oligonucleotide C (oligo C), which also bound TTF-1 (Fig. 2B, lanes 3 and 4 versus lanes 1 and 2). Second, we showed that an antiserum made against a TTF-1-specific peptide (anti-TTF-1) caused the same supershift in FRTL-5 and rMTC cells (Fig. 2C, lanes 5 to 7 versus lanes 9 to 11) and that the complex was not supershifted by control preimmune serum (Fig. 2C, lanes 4 and 8). Smaller amounts of nuclear extract resulted in decreased amounts of the TTF-1–DNA complex (Fig. 2C, lanes 1 to 3), decreased amounts of the anti-TTF-1-supershifted complex (Fig. 2C, lanes 5 to 7 versus lanes 9 to 11), and a nearly total loss of the TTF-1–DNA complex in the presence of anti-TTF-1 (Fig. 2C, lanes 6, 7, 10, and 11 versus lanes 2 and 3). Finally, we showed that recombinant TTF-1 (Fig. 2C, lane 12) and the radiolabeled probe formed a complex with the same mobility as the complex formed by nuclear extracts from rMTC parafollicular C cells and FRTL-5 thyroid cells.

Thus, TTF-1 protein, as well as RNA, appears to exist at similar levels in rMTC and rat FRTL-5 thyroid cells. Moreover, the protein forms similar levels of complexes with a representative TTF-1 element.

TTF-1 regulatory elements exist in C-cell genes important for calcium homeostasis.

The main function of C cells is to sense increases in extracellular Ca²⁺ and secrete calcitonin. The critical gene product involved in the sensing process is the CaSR (11, 22, 40). Calcitonin decreases extracellular Ca²⁺ by its action on bone and kidney (5, 37). Although an oversimplification, the function of parathyroid cells counterbalances the action of C cells. Parathyroid cells sense decreases in extracellular Ca²⁺ and secrete parathyroid hormone, which increases extracellular Ca²⁺ by its action on bone and kidney (5, 10, 50). As in C cells, the CaSR is the critical extracellular Ca²⁺-sensing component of parathyroid cells (11). Since C cells and parathyroid cells function coordinately to maintain Ca²⁺ homeostasis throughout the body and since both contain TTF-1, we questioned whether TTF-1 might regulate genes important for Ca²⁺ homeostasis in C or parathyroid cells.

Using rMTC cells as our C-cell model, we examined whether the CaSR and calcitonin genes existed in the cells, whether they had putative TTF-1 elements, and whether any of these putative TTF-1 elements might be functionally important. RT-PCR (Fig. 3) revealed that rMTC cells contained RNA encoding the CaSR, calcitonin/calcitonin gene-related peptide (CT/CGRP), and calmodulin III (CAM-III) genes. After binding calcium, CAM-III activates a multiplicity of proteins in the cell and is a mediator of the intracellular Ca²⁺ signal; it is ubiquitous in many cells within the organism. Expression of the RNA for both the CaSR and CT/CGRP genes did not correlate with the presence of TTF-1 in the cells (Fig. 3). Both were not detected in BRL cells or rat FRT thyroid cells, which have no TTF-1. The CaSR and CT/CGRP genes were, however, also not detected in functioning rat FRTL-5 thyroid cells, which have TTF-1 (42, 45, 57). Of note, the E-CAD gene, another calcium regulatory gene, was expressed in FRT cells but not in FRTL-5 or rMTC cells (Fig. 3).

FIG. 3.

Expression of genes important for calcium homeostasis in rMTC cells. RT-PCR was carried out with total RNA and the primers described in Materials and Methods. The marker used was a φX174 DNA HaeIII digest. The sizes of the PCR products anticipated for the CaSR, CAM-III, E-CAD, and calcitonin (CT/CGRP) genes, based on the primers used, are, respectively, 568, 302, 515, and 157 bp.

Using the sequence of a consensus TTF-1 binding site (Table 1), which we derived from TTF-1 binding sites in the TG, TPO, and TSHR genes (21, 24, 45), we identified several putative TTF-1 binding elements in the published 5′-flanking regions of the rat CaSR (52), rat CT/CGRP (58), rat CAM-III (39, 44), and mouse E-CAD (6) genes (Table 2). Since some variation exists in the consensus binding sites that were used in this analysis (Table 1), we performed the same search on other genes whose activity or differentiated expression was not intimately linked to calcium levels. We did not identify putative TTF-1 binding sites in the GenBank published 5′-flanking regions of the rat β-actin, GAPDH, albumin, histone H1d, histone H10, β-fibrinogen, multidrug resistance, P-glycoprotein, alpha-1-interferon, pit-1, insulin, cyclin B1, neutral cholesterol ester hydrolase, high-affinity Na⁺/glucose cotransporter, aldolase, and tryptophan oxygenase genes or in mouse Sox-4 and Wnt-4. Interestingly, we also did not identify putative TTF-1 binding sites in the published 5′-flanking regions of the rat phospholipase A2, CAM-I, CAM-II, calmodulin kinase, RK8-13 Ca²⁺ pump, and Ca²⁺ ATPase genes, in the chicken caldesmon gene, or in the mouse calcineurin gene. It seemed reasonable, therefore, to question the possibility that one or more of the putative TTF-1 binding elements in the CaSR, CT/CGRP, CAM-III, and E-CAD 5′-flanking regions might be functional.

TABLE 1.

Consensus TTF-1 binding site sequences derived from TTF-1-specific binding sites in the rat TG, TPO, and TSHR genes^a

Gene	Site	Core sequence
TG	A	TACTCAAGTA
TG	B	GACTCAAGTA
TPO	A	CACTCATAGA
TPO	B	CTGCCAAGTG
TSHR	Upstream	CACTTAAGTG
TSHR	Downstream	CACTTGAGAG
Consensus		CACTCAAGTG
		GTGCTGTAGA
		T A

^aKnown DNA core sequences from elements which bind TTF-1 specifically were taken from earlier reports (21, 24, 45). 

TABLE 2.

Putative TTF-1 binding sites in the CaSR, CAM-III, CT/CGRP, and E-CAD genes^a

Species	Gene	Region	Sequence
Rat	CaSR	A (−551 to −528)	GACTGTAGGT
Rat	CaSR	B (−256 to −233)	GACTCCAGAA
Rat	CaSR	C (−208 to −185)	GACTCGAGGA
Rat	CAM-III	A (−1041 to −1018)	GAGCCAAGTC
Rat	CAM-III	B (−714 to −691)	TGCTCAAGGG
Rat	CAM-III	C (−571 to −548)	CACACAGGTC
Mouse	E-CAD	A (−581 to −558)	GACTCTTGAA
Mouse	E-CAD	B (−446 to −423)	AACTCATCCA
Rat	CT/CGRP	A (−919 to −896)	CACTGGAGAC
Rat	CT/CGRP	B (−425 to −402)	CACTAAAGTC
Rat	CT/CGRP	C (−335 to −332)	TACTTGAGAT

^aDNA sequences were identified based on their similarity to the known TTF-1-binding site sequences shown in Table 1. A, B, and C denote different putative binding sites within the same gene. Their positions in the 5′-flanking region are numbered with the ATG start nucleotide as +1 for the CaSR, CAM-III, and E-CAD genes but with the 5′ G of the CAP site as +1 for the CT/CGRP gene (58). The numbers represent the 24-bp oligonucleotides which include the sequences shown in the right-hand column as their core sequences. These oligonucleotides were synthesized and used for binding studies (see Fig. 4). 

We synthesized 24-bp oligonucleotides containing the putative TTF-1 binding elements listed in Table 2, plus 7 bp on their 5′ and 3′ ends, and then labeled each to use as probes for EMSA (Fig. 4A). Oligonucleotides CaSR-C, CAM-III-A, CAM-III-B, E-CAD-A, and CT/CGRP-C formed protein-DNA complexes with nuclear extracts from FRTL-5 and rMTC cells (Fig. 4A, lanes 4, 5, 6, 8, and 12). Because the mobility of the protein-DNA complex in each case was the same as the mobility of a complex formed with a radiolabeled oligonucleotide probe containing the downstream TTF-1 binding element of the TSHR (Fig. 4A, lane 1), we tentatively identified the complexes as TTF-1 complexes. The following results supported this conclusion.

FIG. 4.

Ability of putative TTF-1 binding sequences found in the CaSR, CAM-III, E-CAD, and calcitonin (CT/CGRP) genes to form a TTF-1–DNA complex with rMTC or FRTL-5 thyroid cell nuclear extracts. (A) Oligonucleotides containing the putative TTF-1 binding sequences shown in Table 2 were synthesized, radiolabeled, and incubated with nuclear extracts from FRTL-5 cells (top) or rMTC cells (bottom). Migration was compared to that of complexes formed between the same extracts and a radiolabeled oligonucleotide with the sequence of the downstream TTF-1 binding site of the TSHR (TSHR DS) (45). (B) Formation of protein-DNA complexes between nuclear extracts from rMTC cells and radiolabeled oligonucleotides containing putative TTF-1 binding sites, i.e., CaSR-C, CAM-III-A, CAM-III-B, E-CAD-A, and CT/CGRP-C (lanes 3 to 12), was evaluated in the presence of a 100-fold molar excess of unlabeled oligonucleotide containing the TTF-1 binding element (oligo C) or TTF-2 binding element (oligonucleotide K) of the TG promoter (24, 56). An oligonucleotide with the sequence of the downstream TTF-1 binding site of the TSHR (TSHR DS) was the control (lanes 1 and 2). (C) Protein-DNA complexes formed between nuclear extracts from rMTC cells and radiolabeled CaSR-C, CAM-III-A, CAM-III-B, E-CAD-A, or CT/CGRP-C (lanes 3 to 12) were compared to those formed with the downstream TTF-1 binding site of the TSHR (TSHR DS) (lanes 1 and 2) in the presence of an antiserum to a specific TTF-1 peptide or its preimmune control. The arrow denotes the supershifted complex in the presence of the immune serum. (D) Recombinant TTF-1 protein (20 ng) was incubated with radiolabeled oligonucleotides containing putative TTF-1 binding sites from the CaSR, CAM-III, E-CAD, or CT/CGRP genes (lanes 2 to 12) or the downstream TTF-1 binding site of the TSHR (TSHR DS) (lane 1). The presence of a multiplicity of complexes with recombinant TTF-1 is related to oligomerization of the TTF-1 protein during the incubation (4).

First, formation of the complex with rMTC nuclear extracts in each case was prevented by a 100-fold excess of oligo C of the TG gene, which contains the TTF-1 binding site (Fig. 4B), but not by oligonucleotide K of the TG gene, which binds to TTF-2 (24, 56). Second, formation of the complex in each case was decreased and supershifted by anti-TTF-1 but not by preimmune serum (Fig. 4C). Finally, each of the oligonucleotides CaSR-C, CAM-III-A, CAM-III-B, E-CAD-A, and CT/CGRP-C could bind recombinant TTF-1 protein, whereas the oligonucleotides CaSR-A, CaSR-B, CAM-III-C, CT/CGRP-A, and CT/CGRP-B, which did not form complexes with rMTC nuclear extracts (Fig. 4A), did not interact with recombinant TTF-1 protein (Fig. 4D). The complexes formed with recombinant TTF-1 in this experiment had the same mobility as a complex formed with the downstream TTF-1 site of the TSHR (Fig. 4D, lane 1). Moreover, they were prevented from forming by oligo C of the TG gene (data not shown) and were supershifted by anti-TTF-1 but not by preimmune serum (data not shown). In sum, at least one of the putative TTF-1 binding elements found in the 5′-flanking regions of the CaSR, CAM-III, and CT/CGRP genes, as well as the E-CAD gene, is a functional element, as evidenced by its ability to bind TTF-1.

We next questioned whether the TTF-1 site which binds TTF-1 in these genes is functional with respect to gene expression. We chose the CaSR as our primary test model, since it is present in parafollicular C cells and parathyroid cells and since its Ca²⁺-sensing role is critical for the calcium homeostasis of each (10, 11, 22, 40).

TTF-1 regulates transcription of the CaSR gene.

We obtained a 1,095-bp fragment of the 5′-flanking region of the rat CaSR by PCR of rat genomic DNA (Fig. 5). The portion between bp −243 and −1 of the 5′-flanking region (Fig. 5) is identical to the published untranslated region of the rat CaSR cDNA sequence (52) and contains the putative TTF-1 site termed CaSR-C. The remainder of the PCR-derived genomic 1,095-bp fragment is not identical to the sequence of the untranslated region of the rat CaSR cDNA and does not contain the CaSR-A and CaSR-B sites (Fig. 5).

FIG. 5.

Structure (A) and sequence (B) of a 5′-flanking region of the rat CaSR by comparison to the untranslated region (UTR) of the rat CaSR. The sequence of 1,095 bp of the 5′-flanking region of the rat CaSR was obtained by nested PCR. The sequence of the untranslated region of the rat CaSR cDNA is from Riccardi et al. (52). The boxed region in panel B denotes a region of sequence identity between the two sequences. Thick underlining between bp −201 and −192 denotes the putative TTF-1 CaSR-C site. Thin underlining at bp −544 to −535 and bp −249 to −240 denotes the putative TTF-1 CaSR-A and CaSR-B sites, respectively.

We ligated the 1,095-bp PCR-derived fragment of the rat genomic CaSR to a luciferase reporter gene and transfected it into FRTL-5 thyroid cells. Rat FRTL-5 thyroid cells grown in the absence of TSH have high levels of TTF-1; those grown in the presence of TSH have low levels of TTF-1 (45, 57). They do not contain the CaSR under either condition and are readily transfected by other TTF-1-sensitive genes, which could serve as positive controls. In this experiment, the positive control was a 199-bp construct of the TSHR, pTR(−199)CAT, which contains the downstream TTF-1 site (45, 57). The pGL3-Promoter vector with no CaSR insert was the negative control.

Expression of the wild-type CaSR construct, pGL3-CaSR, was significantly higher in cells maintained in the presence of TSH, whereas that of pTR(−199)CAT was significantly lower and that of the pGL3-Promoter control vector was unchanged (Fig. 6A). These results suggested that TTF-1 was a negative regulator of the CaSR, whereas it was a positive regulator of the TSHR. To confirm the former point, we cotransfected FRT cells, which have no TTF-1 (Fig. 2) (42, 45, 57), with an expression vector containing TTF-1 cDNA; we showed that it decreased pGL3-CaSR activity but not the activity of the pGL3-Promoter control vector (Fig. 6B).

FIG. 6.

TTF-1 suppresses native CaSR promoter activity when transfected into FRTL-5 cells and is a suppressor of the TTF-1 binding site on the CaSR 5′-flanking region when CaSR-C is ligated to an SV40 promoter-driven luciferase reporter gene and transfected into FRTL-5 or FRT thyroid cells. (A) FRTL-5 cells were grown in the presence (+) or absence (−) of TSH, which decreases endogenous TTF-1 levels (57). Cells were transfected with pGL3-CaSR, which contains 1,095 kb of the 5′-flanking region of the CaSR linked to a luciferase reporter gene in the pGL3-Promoter vector; transfected pGL3-Promoter (pGL3-Prom.) vector alone was the control. The promoter activity of a chloramphenicol acetyltransferase chimera containing 199 bp of the minimal TSHR promoter, pTR(−199)CAT, was the positive control [pTR(−199)]; its counterpart vector with no TSHR insert was the negative control (p8CAT). The activity of pTR(−199)CAT, which contains the downstream TTF-1 binding site of the TSHR, was measured as described previously (57). (B) FRT cells, which contain no endogenous TTF-1 (Fig. 2A), were cotransfected with pGL3-CaSR or the pGL3-Promoter vector plus an expression vector containing TTF-1, pRcCMV-THA (+), or the control vector pRcCMV containing no TTF-1 (−). The activity of the pGL3-Promoter vector cotransfected with the pRcCMV vector was arbitrarily set at 1. (C) One, two, or seven copies of the CaSR-C oligonucleotide (Table 2) were tandemly ligated and inserted in either the positive or the negative orientation into the pGL3-Promoter vector (arrows). Each was transfected into FRTL-5 cells, which contain endogenous TTF-1, and luciferase activity was measured 48 h after transfection. Results in panels A to C are expressed as the mean ± SD luciferase activity of each construct relative to that of the respective control vector with no insert (data shown without error bars). (D) FRT cells, which contain no endogenous TTF-1 (Fig. 2A), were transfected with TTF-1 cDNA, pRcCMV-THA (+), or the control vector pRcCMV (−) plus the SV40 promoter-driven plasmid containing one, two, or seven copies of the CaSR-C oligonucleotide (arrows) ligated and inserted in a positive orientation, as represented. The control was the pGL3-Promoter vector with an SV40 promoter but no CaSR-C oligonucleotide insert. Data are expressed as the mean ± SD of the ratio of activities in the presence and absence of TTF-1 cotransfection. In all panels, data are from at least three different experiments with three different batches of transfected cells. A statistically significant change from the control with a P value of <0.05, <0.01, or <0.001 is denoted, respectively, by one, two, or three asterisks.

We then made a series of chimeric reporter gene constructs having one or more repeats of the CaSR-C oligonucleotide. These were placed in a sense or antisense direction 5′ to an SV40 promoter that was linked 3′ to a luciferase reporter gene (pGL3-Promoter) (Fig. 6C). We transfected each into FRTL-5 cells maintained without TSH, i.e., with high levels of TTF-1 (45, 57). Luciferase activity decreased with increasing copy number of the CaSR-C oligonucleotide regardless of orientation (Fig. 6C). Mutation of the TTF-1 element between bp −208 and −185 from GACTCGAGGA to GGCCCGTGGA (changes in bold) resulted in a loss of the decrease in luciferase activity expressed by the different constructs (data not shown).

Finally, we cotransfected FRT thyroid cells, which as noted above have no TTF-1 (Fig. 2) (42, 45, 57), with a TTF-1 expression vector and a luciferase-linked SV40 promoter construct having one or more copies of the CaSR-C oligonucleotide in a sense orientation (Fig. 6D). A negative control was the SV40 promoter-luciferase vector with no CaSR-C oligonucleotide insert. TTF-1 caused a significant decrease in promoter activity in the constructs with but not without the CaSR-C oligonucleotide (Fig. 6C). TTF-1 did not decrease promoter activity in a construct with a mutated CaSR-C site (data not shown). We suggest that the significant TTF-1-induced decrease exhibited by a single element in Fig. 6D versus the absence of a decrease in Fig. 6C is explained by the much higher levels of TTF-1 resulting from transient transfection (Fig. 6D) compared to the endogenous TTF-1 levels in FRTL-5 cells (Fig. 6C). TTF-1 levels measured as RNA or protein (antibody or gel shifts) are over 10-fold higher in the transient transfection situation (data not shown).

These results indicated that the CaSR-C element which binds TTF-1 is functionally responsive to TTF-1. Because activity is expressed independent of direction and is proportional to copy number in the presence of TTF-1, it has the properties of a silencer (35).

Expression of TTF-1 and CaSR or CT/CGRP genes is inversely regulated by the intracellular calcium signal.

Altered extracellular Ca²⁺ levels signal C cells by modulating the inositol phosphate (IP) signal system and thereby changing intracellular Ca²⁺ levels (11, 22, 40). We evaluated the effect on TTF-1 and CaSR gene expression of several agents known to increase the internal Ca²⁺ signal. We initially used FRTL-5 thyrocytes, since the effects of these agents are well described for these cells, since FRTL-5 cells contain endogenous TTF-1, and since FRTL-5 cells can be readily transfected by both exogenous TTF-1 and the CaSR (see above).

Treatment of FRTL-5 thyroid cells with the calcium ionophore A23187, which induces a concentration-dependent increase in cytosolic Ca²⁺ levels (9, 15, 53), caused a concentration-dependent decrease in TTF-1 RNA levels (Fig. 7A, panel B) as well as in TSHR gene expression (data not shown) (see reference 53). Similarly, ATP, which acts through the P₂ purinergic receptor and the IP signal system, also caused a concentration-dependent decrease in TTF-1 RNA levels (Fig. 7A, panel c) as well as a concentration-dependent increase in internal Ca²⁺ levels (data not shown) (see references 2, 46 to 48, and 63) and TSHR gene expression (data not shown) (see reference 53). Activation of kinase C, mimicked by the phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA), had a minimal effect on TTF-1 RNA levels at the TPA concentrations tested (Fig. 7A, panel a). However, TPA was synergistic with A23187 in its ability to decrease TTF-1 RNA levels when it was tested at the lowest level used, which had no significant effect alone (Fig. 7A, panel d).

FIG. 7.

TTF-1 RNA levels (A) and TTF-1 binding to the CaSR-C TTF-1 element (B) are decreased by the ATP- or A23187-increased the internal Ca²⁺ signal, and kinase C activation can synergistically increase the Ca²⁺ signal effect. (A) FRTL-5 cells were grown to 70% confluency in complete medium containing TSH and then maintained in medium with no TSH plus 0.2% calf serum for 6 days. TPA (1.0, 10, or 100 nM), A23187 (0.1, 1.0, or 10 μM), ATP (0.1 or 1.0 mM), or 1 nM TPA plus A23187 (0.1, 1.0, or 10 μM) was then added to the medium (panels a to d, filled bars, respectively). After 4 h, total RNA was recovered and subjected to Northern analysis with radiolabeled cDNAs encoding TTF-1 and β-actin. The TTF-1/β-actin ratio was calculated after densitometric quantitation of autoradiograms of the blots. Data (filled bars) are expressed relative to the TTF-1/β-actin ratio in cells maintained with basal medium with no addition, which was arbitrarily set at one (open bars). There was no change in β-actin levels with any treatment. Data are the mean ± SD for four separate experiments with different batches of cells. A statistically significant (P < 0.05) decrease, compared to the control, caused by one agent is denoted by a single asterisk. A statistically significant (P < 0.01 or P < 0.001) additive or synergistic effect of TPA and A231871 is denoted by two or three asterisks, respectively. (B) FRTL-5 cells were identically grown and treated with TPA (1.0, 10, or 100 nM; lanes 2 to 4, respectively), A23187 (0.1, 1.0, or 10 μM; lanes 5 to 7, respectively), ATP (0.1 or 1.0 mM; lanes 8 and 9, respectively), or 1 nM TPA plus A23187 (0.1 or 1.0 μM, lanes 10 and 11, respectively). Lane 1, basal activity. Nuclear protein was isolated 24 h after treatment. EMSA was performed with a radiolabeled oligonucleotide having the sequence of the TTF-1 element of the CaSR-C oligonucleotide.

The increases in the internal Ca²⁺ signal and the decreased TTF-1 RNA levels in FRTL-5 cells were associated with reduced binding of TTF-1 to its DNA elements. Thus, by using the CaSR-C oligonucleotide (Fig. 7B), the TTF-1 binding element within the downstream TSHR minimal promoter (data not shown), or oligonucleotide C of the TG promoter, which binds TTF-1 or Pax-8 (data not shown), we could show that A23187 or ATP treatment of FRTL-5 cells caused cell extracts to exhibit a significant decrease in TTF-1–DNA complex formation, in comparison with extracts from untreated cells (Fig. 7B, lanes 5, 7, 8, and 9 versus lane 1). As was the case for RNA levels, TPA caused a minimal decrease in complex formation (Fig. 7B, lanes 2 to 4 versus lane 1) but was synergistic with A23187 when tested at the lowest concentration, which had no measurable effect alone (Fig. 7B, lanes 10 and 11 versus lanes 5 and 6). Thus, an increase in the Ca²⁺ signal can downregulate TTF-1 complex formation with functional TTF-1 elements in multiple genes, including the CaSR gene. The Ca²⁺ signal action does not appear to be mediated by kinase C activation; rather, activation of the kinase C signal can enhance the activity of the Ca²⁺ signal.

ATP, as well as ionophores, can increase internal Ca²⁺ levels in rMTC cells (20). It was thus not surprising that the A23187- and ATP-induced decrease in TTF-1 RNA levels was duplicated in rMTC cells, as measured by Northern analysis (Fig. 8A and B, respectively). A23187 increased CaSR and CT/CGRP RNA levels within 4 h of treatment, coincident with the decrease in TTF-1 RNA levels (Fig. 8A). ATP also caused a prominent increase in CaSR and CT/CGRP RNA levels and decreased TTF-1 RNA levels (Fig. 8B). However, the ATP effect on TTF-1 RNA was evident within 2 h of treatment, and the increase in both CaSR and CT/CGRP RNA levels was better measured at 4 h (Fig. 8B). Thus, agents which increased internal calcium levels in C cells (see below) could decrease TTF-1 RNA levels exactly as in FRTL-5 cells; further, there was an associated increase in CaSR and CT/CGRP RNA levels. These data (Figs. 8A and B) suggest that the TTF-1 site on the CT/CGRP gene is likely to be a silencer, as is the case for the CaSR gene, since increased CT/CGRP gene expression was associated with decreased TTF-1 binding (Fig. 4).

FIG. 8.

Effect of A23187 (A) or ATP (B) on TTF-1, CaSR, or CT/CGRP RNA levels in rMTC cells and the decrease in TTF-1 promoter activity caused by ATP treatment of cells (C). (A) rMTC cells were grown to 70% confluency and then exposed to (+) or not exposed to (−) 1.0 μM A23187 for 4 h. Total RNA was recovered and subjected to Northern analysis (top). After densitometric quantitation of autoradiograms or after quantitation of the blots with a BAS 1500 phosphorimager, the TTF-1/GAPDH, CaSR/GAPDH, or CT/CGRP/GAPDH ratios were calculated. (B) rMTC cells were identically grown and treated with 1.0 mM ATP for 2 or 4 h, at which times total RNA was recovered and subjected to Northern analysis as in panel A. Data are compared to results from cells maintained without ATP. (C) FRTL-5 cells were grown to 70% confluency in complete medium containing TSH and then maintained in medium with no TSH plus 0.2% calf serum for 6 days. Cells were transfected with the indicated TTF-1–luciferase chimeras containing different lengths of the 5′-flanking region. ATP (1 mM) was added to the medium 24 h after transfection, and 4 h later reporter gene activity was measured and normalized to data obtained with the pSV control after correction for transfection efficiency. Data in all panels are the means ± SD for three different experiments. One, two, and three asterisks represent significant changes from the control with P values of <0.05, <0.01, and <0.001, respectively.

The regulatory mechanism by which the Ca²⁺ signal decreases TTF-1 RNA levels is mediated transcriptionally. Thus, TTF-1 promoter activity, measured with TTF-1 promoter-luciferase reporter gene chimeras, was decreased by treating FRTL-5 cells with ATP, and the activity was localized within 1.15 kb of the 5′-flanking region (Fig. 8C). The ATP-induced decrease in TTF-1 promoter activity was not duplicated by the same concentration of ADP or CTP (Fig. 9A), was associated with a coincident increase in the activity of the SV40 promoter-luciferase chimera containing seven copies of the CaSR-C oligonucleotide (Fig. 9B), and was associated with an increase in internal Ca²⁺ levels not duplicated by ADP or CTP (Fig. 9C). How Ca²⁺ might modulate an element on the TTF-1 5′-flanking region within this 1.15-kb segment is not known at this point.

FIG. 9.

Effect of ATP on TTF-1 (A) and CaSR (B) promoter activity is specific and associated with increased internal Ca²⁺ levels (C), and suppression of the TTF-1 promoter induced by ATP is reversed by decreasing extracellular Ca²⁺ levels (D). (A and B) FRTL-5 cells were grown to 70% confluency in complete medium containing TSH and then maintained in medium with no TSH plus 0.2% calf serum for 6 days. The 5.18-kb fragment of the TTF-1 5′-flanking region linked to a luciferase reporter gene was transfected into one set of cells (A). A duplicate set of cells (B) was transfected with a plasmid containing seven copies of the native CaSR-C oligonucleotide ligated and inserted in a positive orientation in the pGL3-Promoter vector (Fig. 6). ATP, ADP, or CTP (1.0 mM) was added to the medium 24 h after transfection, and reporter gene activity was measured 4 h later. Data are expressed relative to the promoter activity of the control, with no addition. (C) The ability of each nucleotide to increase intracellular calcium levels in duplicate batches of transfected cells was measured. Results are the superimposed kinetics of 30 different cells as a function of time. (D) The 5.18-kb fragment of the TTF-1 5′-flanking region linked to a luciferase reporter gene was transfected into cells exactly as in panel A. At 24 h after transfection, cells were treated with 1.0 mM ATP as in panel A. Four hours after the ATP addition, one set of cells was washed with PBS without Ca²⁺ and incubated for the noted times with Ca²⁺-free Dulbecco’s modified Eagle’s medium containing 10 mM HEPES, 1.8 mM MgSO₄, glutamine, 0.2% calf serum, hormones other than TSH used to grow FRTL-5 cells (32), and 2 mM EGTA (tEGTA). A second set of cells was washed in the same way and incubated with the same medium except that no EGTA was added (−EGTA). Cells were harvested to measure luciferase activity. In panels A, B, and D, data are the means ± SD for at least three experiments, each done in duplicate with different batches of cells. A statistically significant change from the control with a P value of <0.01 or <0.001 is denoted by two or three asterisks, respectively.

The ATP-induced decrease in TTF-1 promoter activity (Fig. 9C) and the resulting increase in CaSR promoter expression (data not shown) could be reversed by washing FRTL-5 cells free of ATP and adding medium free of Ca²⁺ plus 2 mM EGTA (Fig. 9D). Similarly, the ATP-induced decrease in TTF-1 RNA levels and the increase in CaSR or calcitonin RNA levels in rMTC cells could be reversed by decreasing extracellular Ca²⁺ in exactly the same way (data not shown).

These results clearly indicate that TTF-1 promoter activity is negatively regulated by high intracellular Ca²⁺ concentrations, that this effect can be reversed by reducing intracellular Ca²⁺ concentrations, and that changes in TTF-1 levels are inversely related to CaSR and CT/CGRP gene expression.

DISCUSSION

When we identified TTF-1 mRNA in rat parafollicular C cells in the thyroid and in chief cells in the parathyroid (61), we hypothesized that TTF-1 might play a role in the expression of genes important for calcium regulation in cells and/or calcium homeostasis in the organism. In the present study, we showed that TTF-1 RNA exists in C cells of diverse species. Its levels in C cells in vivo may in fact be higher than that in thyroid cells in vivo, possibly reflecting the fact that follicular TG and TSH additively suppress TTF-1 levels in thyroid cells (45, 57, 59, 60). We examined the question of function in a parafollicular C-cell model, rMTC cells in continuous culture. We showed that TTF-1 RNA in rMTC cells results in the expression of a functional TTF-1 protein, based on the ability of TTF-1 protein to form protein-DNA complexes with TTF-1 binding sites in the TSHR and TG 5′-flanking regions. We showed that rMTC cells express the CaSR, calmodulin, and calcitonin genes, that there are putative TTF-1 binding sequences on each of these genes, and that at least one site on each gene is able to bind TTF-1. We used the CaSR gene as a representative calcium-sensitive gene to show that the TTF-1 binding site functions as a silencer in the presence of TTF-1 and that changes in internal Ca²⁺ levels reciprocally regulate TTF-1 and CaSR gene expression at the transcriptional level. We showed that the potential Ca²⁺ modulation site of the TTF-1 gene exists within 1.15 kb of its 5′-flanking region, although we do not know how Ca²⁺ might regulate the putative element, e.g., via modulation of a kinase activity, a binding protein, a coregulatory molecule, or a combination of these activities. Nevertheless, we showed that increased intracellular Ca²⁺ in C cells decreases TTF-1 and increases CaSR levels; conversely, decreased Ca²⁺ increases TTF-1 and decreases CaSR levels. We showed that the expression of the CT/CGRP gene is coincidentally and similarly regulated by Ca²⁺ and TTF-1.

We used FRTL-5 thyroid cells to complement studies with rMTC cells, since FRTL-5 cells contain TTF-1, since intracellular Ca²⁺ is an important signal within the cells, and since the cells do not contain either the CaSR or the CT/CGRP genes. FRTL-5 cells are, therefore, convenient “null” cells with which to examine the regulation of transfected CaSR or CT/CGRP genes. Using FRTL-5 cells, we showed that TPA alone causes a minimal decrease in TTF-1 RNA levels and TTF-1 complex formation with the TTF-1 binding element of the CaSR but that TPA acts synergistically with A23187 to decrease TTF-1 RNA levels and TTF-1 complex formation with the CaSR. Since TPA acts synergistically with A23187 to stimulate calcitonin secretion from rMTC cells (25), it is reasonable to conclude that data derived from FRTL-5 cell experiments are consistent with and applicable to C cells. Moreover, it seems clear that the Ca²⁺-regulated, TTF-1-mediated changes in CaSR and CT/CGRP gene expression described here complement and are consistent with the secretory process by which calcitonin is released from C cells by altered internal Ca²⁺.

In sum, TTF-1 levels are transcriptionally regulated by changes in intracellular Ca²⁺ levels which are brought about by changes in extracellular Ca²⁺ levels or by stimulation of cells with a physiologically active ligand that interacts with the cells and modulates the IP-Ca²⁺ signal system, i.e., ATP. TTF-1 regulatory elements exist on tissue-restricted or -specific genes associated with calcium homeostasis in C cells just as they exist on tissue-restricted or -specific genes associated with the formation and secretion of thyroid hormones in thyrocytes. In C cells, the IP-calcium signal is the coordinator; in the thyroid, the cAMP signal has this function. The relationship between the two control mechanisms for TTF-1 levels and function in both C cells and thyroid cells will, however, become an interesting question, since the two signals are functionally interactive in both types of cells (17, 25, 31, 62).

The intracellular Ca²⁺ signal is important in almost all cell types, including neurons. The extracellular Ca²⁺ level is one means of regulating the intracellular Ca²⁺ signal in neurons and is known to be important for the hormone secretory activity of or neuropeptide synthesis in the pituitary as well as the parathyroid and C cells (1, 10, 23, 43, 65, 66, 68). One means to sense the extracellular Ca²⁺ level involves Ca²⁺ channels. The pituitary, the parathyroid, and parafollicular C cells use, instead, the CaSR. The CaSR recognizes high extracellular Ca²⁺ levels, initiates a Gq protein-mediated increase in IP and intracellular Ca²⁺ levels, and thereby induces the secretion of the hormones (11, 18, 22, 40, 49). Using a pituitary cell model, however, Emanuel et al. (18) showed that, whereas acute increases in Ca²⁺ levels in the medium increased IP, cytosolic Ca²⁺, and cAMP levels and the associated secretion process, similar changes in extracellular Ca²⁺ levels for a period of 24 h increased CaSR RNA levels two- to fourfold. It is reasonable to presume that the change in CaSR RNA levels reflects the effect of internal Ca²⁺ on TTF-1 levels and that the change in TTF-1 regulates CaSR gene expression, in analogy to the C-cell model described here.

The following model emerges (Fig. 10). The CaSR senses acute increases in extracellular Ca²⁺ levels. This change induces acute increases in IP and intracellular Ca²⁺ levels and initiates a secretory response in which secretory vesicles release calcitonin (Fig. 10, secretory phase). If persistent, the change in internal Ca²⁺ will regulate TTF-1 levels, and the altered TTF-1 levels will coordinately regulate the transcription of multiple genes; i.e., increased Ca²⁺ will decrease TTF-1 and decrease the silencing activity of TTF-1 on the CaSR and CT/CGRP genes (Fig. 10, synthesis phase). Increased CaSR levels should enhance the responsiveness of the cell to high extracellular Ca²⁺; increased calcitonin levels would replenish the secreted calcitonin which was used to lower serum Ca²⁺. The priming signal of the synthesis phase (Fig. 10) would appear to be increased intracellular Ca²⁺ levels; however, increased kinase C activity may synergistically increase the functional response. If extracellular Ca²⁺ levels were decreased, a suppression phase would ensue (Fig. 10). Internal Ca²⁺ levels would be decreased by the CaSR, TTF-1 levels would be increased, and CaSR or CT/CGRP gene expression would be decreased. Synergistic activity by kinase C would also decrease. The suppression phase would complement a decrease in secretory activity.

FIG. 10.

Proposed model of extracellular calcium-altered, TTF-1-mediated regulation of the expression of genes controlling calcium homeostasis in C cells. The CaSR on C cells senses acute changes in extracellular Ca²⁺ levels ([Ca²⁺]_o). The calcium-CaSR interaction causes an acute increase in IP and intracellular Ca²⁺ levels ([Ca²⁺]i). This change initiates the secretory response in which calcitonin (CT) is released from secretory vesicles into the circulation (secretion phase). If persistent, however, the change in internal Ca²⁺ levels mediated by the CaSR will regulate TTF-1 levels, and altered TTF-1 levels will coordinately regulate genes important for calcium homeostasis, i.e., the CaSR and CT/CGRP genes (synthesis and suppression phases). Both the synthesis and suppression phases of gene expression would complement changes in secretory activity by altering the ability of the cell to sense extracellular Ca²⁺ or its ability to replenish CT in secretory vesicles. We suggest that the model is applicable to the secretion of hormones by other cells that contain the CaSR and TTF-1, i.e., pituitary trophic hormones in anterior pituitary cells or parathyroid hormone in parathyroid cells. Solid arrows indicate positive signals, T-bars indicate suppression, and broken arrows indicate the translocation of RNA and protein. This model focuses only on the potential role of TTF-1, although we recognize that other factors regulate these synthesis and secretion in C cells (see Discussion and references 7, 17, 62, and 68). PKC, protein kinase C.

This model does not integrate the Ca²⁺ TTF-1 transcriptional signal mechanism with effects of glucocorticoids, serotonin type 1 agonists, or cAMP on the transcription of the noted genes or on secretion (7, 17, 62, 68). Integration with the activity of Ras-responsive elements is also not considered. In part, this is the result of our desire to simplify and focus the model only on TTF-1; in part, however, integration of the activities of multiple promoter elements is best presented after their direct evaluation in the C-cell system used here, since conflicting data are obtained for several of these agents when different systems are used (7, 17).

TTF-1 may have a role in regulating the transcriptional expression of calcium-responsive genes in many cells, and the model may be generally applicative. TTF-1 exists in keratinocytes, Purkinje cells of the cerebellum, and neurons of the hippocampus and retina (61). Keratinocytes and hippocampus neurons contain the CaSR in addition to TTF-1. Keratinocytes proliferate and form a differentiated squamous cell layer when they are cultured in physiological concentrations of Ca²⁺ (36, 51). We hypothesize, therefore, that TTF-1 may be an important Ca²⁺-sensing factor which responds to altered intracellular Ca²⁺ levels and coordinately regulates the gene expression responsible for calcium homeostasis in each of these types of cells.

Finally, two additional points should be noted. First, TTF-1 elements can be positive regulatory elements, as in the case of the TSHR, TG, TPO, NIS, or MHC genes (14, 19, 21, 24, 41, 42, 45, 54, 57), or negative regulatory elements, as unequivocally shown here for the CaSR gene. In the case of the parathyroid hormone gene, a TTF-1 element would be expected to function in opposition to the TTF-1 element on the CaSR or CT/CGRP gene, i.e., be an enhancer responding to high TTF-1 levels induced by low extracellular Ca²⁺ levels. Second, TTF-1 data from whole thyroids or primary thyroid cell cultures must be interpreted cautiously because TTF-1 is present and functional in more than one cell type. In the dog, particularly, there are large numbers of C cells, and these cells express much higher levels of TTF-1 RNA in vivo than thyroid follicular epithelia (61). This fact may explain why TSH-cAMP does not appear to suppress TSHR gene expression in primary cultures of dog thyrocytes and may contribute to apparent species specificity (38, 64). The functional role of TTF-1 is, therefore, best studied in cell lines FRTL-5 or rMTC, where no ambiguity exists as to cell type.