INTRODUCTION

Induction of cell proliferation by mitogen or growth factor stimulation leads to the specific and sequential expression of a large number of genes (1, 2, 3, 4, 5, 6, 7, 8): immediate early genes, induced rapidly and independently of de novo protein synthesis, delayed early genes whose transcriptional activation requires new protein synthesis, and others ending with the late G₁ genes, which are mainly enzymes involved in DNA synthesis. Conversely, the expression of other genes is also sequentially repressed after mitogenic stimulation (9, 10, 11). Many of these regulated genes are proto-oncogenes or anti-oncogenes. Most of the known ones belong to two major classes of mitogenic pathways: the growth factors receptors tyrosine protein kinases and the phorbol ester protein kinases C cascades. A third signaling pathway, the cyclic AMP-dependent cascade, considered for a long time to be a general inhibitor of proliferation, stimulates proliferation in several epithelial cell types and in yeast (reviewed in Ref. 12). In thyroid cells, this pathway is activated by thyrotropin (TSH),¹ which is the main physiological agent regulating the thyroid gland (13, 14). TSH and cyclic AMP promote cell proliferation, function, and differentiation while the mitogenic pathways elicited by EGF and tumor promoting phorbol esters are associated with the loss of expression of the differentiation specific genes (15, 16, 17, 18, 19). Several steps of the latter cascades are missing in the cyclic AMP pathway (14). The molecular mechanisms of this pathway are still largely unknown. We have therefore initiated a study aimed at the identification of genes that are regulated by cyclic AMP in thyroid cells.

To identify new genes that might be regulated by the thyroid mitogenic pathways, we prepared a cDNA library from the thyroid of a dog treated in vivo with methimazole and propylthiouracil. The inhibition by these drugs of thyroid hormones synthesis leads to a relief of the inhibitory control of these hormones on the hypophysis, to TSH secretion, and chronic stimulation of the thyroid. By differential screening, we identified 19 clones of which 13 have not been described to date. Among those, five corresponded to mRNA whose expression showed modulation after different mitogenic stimulations of thyroid cells in primary culture.

EXPERIMENTAL PROCEDURES

Animal tissues used for this work are the same as those of a previous study conducted in our laboratory (19). The dogs used in this study were used for cardiological studies, and their thyroids were resected before these experiments. Briefly, dogs were treated by oral administration for 4 weeks of methimazole (MM) (2 × 60 mg/day strumazol) and propylthiouracil (PTU) (2 × 150 mg/day propylthiurit) in order to increase the circulating TSH level. Triiodothyronine and thyroxine concentrations in the serum were followed by radioimmunoassay to ensure that the treatment had been effective. On the day of the experiment, 1 h prior to thyroid resection, the dogs received 50 mg/kg bromodeoxyuridine by intravenous injection. Bromodeoxyuridine labeling analyses were performed as described (20). Tissues for PCNA immunohistochemistry and RNA preparation were obtained from MM/PTU treated and untreated dogs. The immunohistochemical procedure for PCNA immunohistochemistry was used as described previously (21). The method involves methanol fixation of the tissue (previously snap-frozen in this case), followed by embedding in paraffin.

Dog thyroid follicles were obtained as detailed previously (22). The follicles were seeded in 100-mm Petri dishes and cultured in a control medium consisting of Dulbecco's minimal essential medium (Life Technologies, Inc.), Ham's F-12 (Life Technologies, Inc.), MCDB 104 (Life Technologies, Inc.) (2:1:1, v/v/v) supplemented with 1 mM sodium pyruvate, 5 µg/ml insulin (Sigma), 40 µg/ml ascorbic acid, 100 units/ml penicillin, 100 µg/ml streptomycin, and 2.5 µg/ml amphotericin B. Fetal calf serum 10% (FCS), TSH 1 milliunit/ml (Armour Pharmaceutical Co., Chicago, IL), forskolin 10⁵ M (Calbiochem), EGF 25 ng/ml (Sigma), and HGF 50 ng/ml (human recombinant HGF, kindly provided by T. Nakamura) were added when indicated.

To purify the RNA, thyroid tissues were ground into a fine powder under liquid nitrogen and homogenized in 4 M guanidinium thiocyanate; thyroid cells were scraped from culture dishes in the same solution. Lysed cells were layered over a cushion of 5.7 M cesium chloride in an SW56 tube and centrifuged at 36,000 rpm at 15 °C for 17 h. Total RNA was extracted as described (23). After treatment with 60 µg/ml proteinase K in 0.3 M NaCl, 0.1% SDS and phenol/chloroform extractions, the RNA was precipitated and resuspended in water for spectrophotometric quantification. Polyadenylated RNA was purified by oligo(dT)-cellulose chromatography (Stratagene) or by the PolyATtract system (Promega) following the manufacturer's protocol. The quality of the poly(A)⁺ RNA prepared for the construction of the library and for the synthesis of the probes was confirmed by Northern blot analysis, using a thyroglobulin probe, detecting a 8.5-kb mRNA. For tissue distribution analysis, total RNA of various dog tissues was extracted with the TRIzol Reagent kit (Life Technologies, Inc., Paisley, UK) (24).

For the construction of a cDNA library in ZAP II, the poly(A)⁺ RNA prepared from the thyroid of a MM/PTU-treated dog was used for the construction of an oligo(dT)-primed, directional cDNA library with the ZAP-cDNA synthesis and cloning system of Stratagene, according to the manufacturer's protocol. The resultant library contained approximately 10⁷ primary recombinants with 0.7% non-recombinant phages. The library was amplified once by the plate lysate method to obtain 2.4 × 10¹⁰ plaque-forming units/ml. To assess size distribution of inserts in the cDNA library, individual plaques were randomly selected and plate lysate stocks were generated by infection of XL-1 blue Escherichia coli cells. Phage DNA was extracted and restricted by EcoRI/XhoI, and this was then electrophoresed on 1% agarose gels to visualize insert sizes. The mean size of the randomly selected cloned inserts was 2.6 kb.

Poly(A)⁺ RNA from in vivo quiescent or stimulated thyroid tissues was used as template for the synthesis of single-stranded digoxigenin-labeled cDNA probes using hexamers and Moloney murine leukemia virus reverse transcriptase following the protocol suggested by the manufacturer (Dog Labeling kit, Boehringer Mannheim). For differential screening, the library was plated at low density (approximately 6000 plaque-forming units/13-cm diameter dish), and duplicate nylon membranes (Qiabrane, Qiagen) lifts were made from each dish. Filters were treated with an alkaline solution, followed by neutralization and nucleic acid was fixed onto the membranes by UV cross-linking (25). Differential hybridization screening of the library was performed according to the procedure of the manufacturer (Boehringer Mannheim). The first filter was hybridized with the probe derived from quiescent thyroids, and the second one with the probe derived from stimulated thyroids. Following 4 h of prehybridization in 5 × SSC, 1% (w/v) blocking reagent for nucleic acid (Boehringer Mannheim), 0.1% N-lauroylsarcosine, 0.02% SDS, the two blots were incubated for 16 h at 68 °C in the hybridization solution containing the digoxigenin-labeled cDNA probes (25 ng/ml). Stringency washes were performed for 2 × 10 min at room temperature with 2 × SSC, 0.1% SDS, followed by 2 × 20 min at 65 °C with 2 × SSC, 0.1% SDS and 2 × 20 min at 65 °C with 0.1 × SSC, 0.1% SDS. The immunological detection of the digoxigenin-labeled probes was performed by a secondary alkaline phosphatase-labeled antibody complex in combination with a chemiluminescent substrate LumigenPPD. Filters were exposed to Hyperfilm MP (Amersham Corp.) for 1-4 h at room temperature. Clones that hybridized differentially with the two probes were replated at low density for a second screening. The remaining positives were picked out, plaque purified, and further analyzed by Northern blotting and sequencing.

After in vivo excision of Bluescript phagemid from Uni-ZAP XR vector using the manufacturer's protocol, the resulting plasmid DNA was amplified and purified by the alkaline lysis method (25). Various restriction enzymes were used to map inserts and subfragments were further cloned in both orientations into M13 mp18 and M13 mp19 single-stranded DNA vectors. Sequence determination was performed by the dideoxy chain termination method (26) using Taq DNA polymerase and M13 fluorescent primers. Reactions were run on the Applied Biosystems 370A automated DNA sequencer. Approximately 300 base pairs of the extremities or the internal sequences of clones were compared with non-redundant European Molecular Biology Laboratory and GenBank data bases, using the BLAST Network Service. Sequences were also translated and used to search the protein data bases (Protein Identification Resource, SWISSProt, and GenPept) using the BLASTX sequence analysis program.

For Northern blot analysis, after denaturation using glyoxal and dimethyl sulfoxide, according to the procedure of McMaster and Carmichael (27), equal aliquots (8-10 µg) of total RNA were fractionated by electrophoresis on a 1% agarose gel in 10 mM phosphate buffer, pH 7. Acridine orange staining of the gel before blotting ascertained that the amounts of RNA were equal in all samples. Denaturated RNAs were blotted to a nylon membrane (Pall Biodyne B) using SSC × 20 (SSC × 1, 0.15 M NaCl, 0.015 M sodium citrate) as described (28). Prehybridization (4 h at 42 °C) and hybridization (overnight at 42 °C) were carried out in 50% formamide, 5 × Denhardt's (0.1% Ficoll, 0.1% polyvinylpyrrolidone), 5 × SSPE (20 × SSPE, 3.6 M NaCl, 0.2 M sodium phosphate, pH 8.3, 20 mM EDTA), 0.3% SDS, 250 µg/ml denatured salmon testis DNA, and 200 µg/ml bovine serum albumin. The hybridization solution contained in addition 10% dextran sulfate (w/v) and the heat-denatured probe. cDNA probes were -³²P-labeled by random primer extension to a specific activity of approximately 10⁹ cpm/µg (29). Filters were washed 4 times for 10 min in 2 × SSC, 0.1% SDS at room temperature and 4 times for 20 min in 0.1 × SSC, 0.1% SDS at 60 °C. They were then autoradiographed at 70 °C using Hyperfilm MP (Amersham Corp.) and Siemens intensifying screens.

RESULTS

To construct a TSH-stimulated thyroid cDNA library, a dog was treated with methimazole and propylthiouracil as described (see above). At the time of thyroid resection, the dog was biologically hypothyroid and its thyroid was thus chronically stimulated by TSH. Serum thyroxine and triiodothyronine levels were 0.5 µg/dl and <30 ng/dl, respectively, versus 1.1 ± 0.1 µg/dl and 37 ± 6 ng/dl in control dogs. Histological examination showed markedly a hyperplastic thyroid with almost total absence of colloid in the antithyroid drug-treated animal, and normal follicles containing colloid in the control thyroids. Bromodeoxyuridine labeling analysis demonstrated nuclear labeling in 8.1% of the thyroid cells of the treated dog, whereas the thyroid cells of control animals showed only 0.17 ± 0.07% labeled nuclei (mean ± S.E.).

Total RNA was extracted from the thyroid of the MM/PTU-treated dog and the poly(A)⁺ RNA fraction was isolated. The quality of the poly(A)⁺ RNA was confirmed by Northern blot analysis, using a thyroglobulin probe, which detected an intact 8.5-kb mRNA (data not shown). This poly(A)⁺ RNA was used as template for the construction of an oligo(dT)-primed cDNA library in the ZAP II phage vector.

The screening of 250,000 plaques, with cDNA probes from control and stimulated thyroids, yielded 19 plaques which, after a second screening, showed consistent differential hybridization. Among these, 11 showed overexpression in the treated thyroid, and 8 showed underexpression. All cDNA inserts were partially sequenced and the sequences were compared to the GenBank and EMBL data bases, using the BLAST program for nucleic acids and proteins (Table I). Six clones were identified as previously characterized genes. These known clones coded for an inwardly rectifying potassium channel (IRK-2), nucleosome assembly protein-1 (NAP-1), ribosomal protein L7, elongation factor 1 (EF-1), non-muscle myosin light chain (MLC), and heat-shock protein-90 (hsp90). Four other clones (4, 5, 137, and 169) showed significant similarities to DNA sequences in the data bases but corresponded to proteins whose functions are unknown (accession numbers: 4, >gb/D14812/HUMORF16; 5, >emb/Z45305/HSC2MB121; 137, >emb/Z43809/HSC1LE021; 169, >gb/M85812/M85812). The remaining 9 clones did not match any known gene in the data bases.

Table I. Identification of 19 differentially expressed cDNAs Sequences of partial cDNA clones were compared with EMBL and GenBank data bases. (?, unknown gene) Clone Insert size Identification (kb) Overexpressed   3 3 ?   44 1.8 EF-1   51 0.7 ?   53 1.9 IRK-2   59 1.3 ?   101 2.8 ?   137 0.8 ?   164 0.7 MLC   165 0.6 ?   167 2 Ribosomal protein L7   169 1.4 ? Underexpressed   2 1.2 ?   4 0.8 ?   5 1 ?   16 3.2 ?   45 2.2 ?   134 2.5 NAP   143 1.5 Hsp 90   166 4.2 ?

Total RNA was extracted from the thyroids of several control or MM/PTU-treated dogs and subjected to Northern blot analysis. Acridine orange staining, rather than -actin, glyceraldehyde-3-phosphate dehydrogenase, or cyclophilin hybridization, was used to normalize the amount of total RNA transferred onto the filter. Indeed, these three genes are modulated in our system (data not shown). One representative experiment is illustrated in Fig. 1. Clone 3 mRNA is up-regulated in the stimulated dog thyroids. Clone 165 mRNA level was highly variable between dogs, enhanced in some animals or attenuated in others. It was therefore not further characterized. Clones 5, 4, 2, 45, 143 (hsp90), and 134 (NAP) transcript levels were all reduced in stimulated thyroids. No modulation was observed for clone 44 (EF-1) and clone 169, and no signal was detected for clones 101, 51, and 53 (IRK-2). Hybridization with clones 16, 59, 137, and 166 revealed only a smear, but no discrete transcript. All these data were confirmed in several dogs. Thus, these Northern studies demonstrate the modulation of five unknown genes in stimulated thyroids.

The following dog tissues were harvested shortly after sacrifice, snap-frozen in liquid nitrogen, and investigated by Northern blotting: thyroid, spleen, liver, lung, heart, skeletal muscle, kidney, adrenal, ovary, testis, stomach, pancreas, large intestine, lymph node, cerebrum, and cerebellum. Clones 2, 3, 4 (Fig. 2), 59, 165, and 169 (Table II) showed no particular tissue distribution pattern. They were detected in most of the tissues analyzed, with a particularly important expression in the adrenal for clone 3, and in the testis, thyroid, and brain for clone 4 (Fig. 2). Clone 5 showed strong thyroid specificity although it could also be clearly seen in both the ovary and cerebrum (Fig. 2). Clones 45 and 51 showed a much higher expression in the thyroid than in the other tissues. Clone 53 (IRK-2) was present only in skeletal muscle, cerebrum, and cerebellum, which was not unexpected (Table II). Clone 101, interestingly, appeared to be confined solely to the liver. To obtain qualitative estimates of the proliferative state of the tissues, those were stained with a monoclonal antibody (PC10) against proliferating cell nuclear antigen (PCNA). The number of PCNA positive nuclei in a tissue is supposed to be closely correlated to that tissue's growth fraction (21). However, the proliferative activity, qualitatively assessed by two observers, suggests no correlation between the tissue mRNA expression level and the level of proliferation (Fig. 2, Table II).

Table II. Summary of RNA tissue distribution of 5 clones Total RNA was extracted from dog tissues and subjected to Northern blot analysis (, no visible expression; +, barely to moderate expression; ++, moderate to strong expression). For PCNA immunohistochemistry, the proliferative activity was reported using the following ascending scale: , ±, ++, +++, where represents no positive nuclei and +++ represents the positive control level (mouse small intestine). Clone 53 Clone 59 Clone 101 Clone 165 Clone 169 PCNA labeling Thyroid       + +   Spleen       +   + Liver     ++     ± Lung       +     Heart   +     +   Skeletal muscle ++ +   +     Kidney   +   + + + Adrenal         + ++ Ovary         + ++ Testis   ++   + ++ ++ Stomach         ++ ++ Pancreas             Large intestine   +     ++ +++ Lymph node           + Cerebrum ++ +     ++   Cerebellum ++ +     ++

In primary cultures of dog thyroid cells, the thyrocytes were cultured for 4 days without mitogenic agents to allow cells to spread and reach a quiescent state. In the absence of stimulating agents, virtually no labeling of nuclei was observed during the 48 h after the addition of [³H]thymidine to the culture medium. Addition of TSH (1 milliunit/ml), forskolin (10⁵ M), an adenylate cyclase activator, EGF (25 ng/ml), and FCS (10%) at day 4 induced, after a lag of about 18 h, a progressive increase in the fraction of [³H]thymidine incorporating nuclei (30). The results of Northern blot hybridization analyses are presented Fig. 3, 4, 5, 6, and in Table III.

Table III. Summary of in vitro mRNA regulation studies Dog thyrocytes in primary culture were submitted to acute (1 to 48 h) or chronic (3 days or more) stimulation with TSH (1 milliunits/ml) or EGF (25 ng/ml). Total RNA was extracted and subjected to Northern blot analysis. (, no modulation; , increased expression; , decreased expression; , increase followed by decrease). Clone mRNA kb TSH EGF Chronic Acute Chronic Acute 5 3.7         3 4         2 1.5         4 1.4 ; 1.8 NTa     NT 45 3     NT   Hsp-90 1.9 ; 2.7       NT NAP-1 <1     NT NT 169 5 mRNA       NT EF-1 1.8       NT a  NT, not tested.

The clone 5 mRNA levels peaked after 20 h incubation in the presence of TSH and decreased after 48 h (Fig. 3). EGF also increased clone 5 mRNA levels after 6-24 h (Table III). Chronic stimulation by EGF and FCS did not change the expression of this clone (Fig. 3).

Clone 3 mRNA was down-regulated by forskolin (Fig. 4A) and TSH (not shown) for the first 24 h. It was increased thereafter, but never exceeded the control levels, even after a 7 days stimulation (not shown). EGF led to an up-regulation, with acute (Fig. 4A) as well as with chronic stimulation (Table III). The mRNA content was already enhanced after 8 h of EGF stimulation, and increased further with longer incubation times. The negative effect of forskolin or TSH on mRNA levels was also observed on the EGF stimulation, since the simultaneous addition of EGF and TSH for 48 h led to a lower mRNA level than when EGF was added alone (Fig. 4B). The effects of EGF and EGF + TSH on clone 3 mRNA content were reproduced by HGF (50 ng/ml), another potent mitogenic agent for thyrocytes (31), and HGF + TSH. However, the increase observed with HGF was less pronounced, and the TSH inhibition was smaller (Fig. 4B).

5

Clone 2 mRNA levels were slightly increased after TSH treatment whereas a 3-day EGF + FCS treatment led to a clear accumulation (Fig. 5, Table III). Acute stimulation by EGF also enhanced clone 2 mRNA levels (Table III).

The mRNA levels of two identified clones (hsp90 and NAP) were also studied (Fig. 6). Treatment with forskolin led first to a decrease of hsp90 and NAP-1 mRNA levels, then after 20 h to an increase. Hsp90 mRNA control levels increased in parallel with the age of the culture; they were further enhanced by chronic treatment with EGF (Table III).

5

Analysis of the data obtained for the other clones presented in Table III shows that clone 4 mRNA expression was enhanced by TSH and EGF, and clone 45 mRNA amounts were increased by TSH, but remained equal after EGF treatment. As in vivo, no modulation was observed for EF-1 and clone 169, no signal was detected for clones 101, 51, and IRK-2, and hybridization with clones 16, 59, 137, and 166 was uninterpretable.

Four clones were selected for further characterization: clones 5 and 3 (based on their modulated mRNA levels), clones 45 and 51 (based on their tissue distribution). The complete sequences of clones 5 and 3 were determined. Hydropathicity profile of the predicted 343-amino acid sequence of clone 5 according to the analysis of Kyte and Doolittle (48) revealed the presence of a stretch of 20 hydrophobic amino acids, which might represent a potential transmembrane region. Protein sequence comparison of clone 5 with the data bases showed 30% identity with the protein kinase family (serine-threonine kinases). However, among the 12 well conserved subdomains of the kinase catalytic domain (32), only two are present in our clone: subdomains VIII and IX, with an homology of 80% (Fig. 7a). Clone 3 encodes a 659-amino acid protein containing 6 ankyrin-like repeats (Fig. 7b). The DNA library accession numbers of clones 3 and 5 are, X99145[GenBank] and X99144[GenBank], respectively.

DISCUSSION

To identify new genes modulated by the TSH-dependent mitogenic pathway, we performed a differential screening of a thyroid cDNA library from a dog treated in vivo with MM/PTU. The use of an in vivo chronically stimulated thyroid insured, with no a priori assumption, that the real in vivo growth pathway was investigated. This led to the identification of 11 clones showing overexpression and 8 showing underexpression. While 6 clones encode known proteins, the 13 others have not been described to date. Among them, 4 (5, 4, 137, and 169) showed significant similarities to DNA sequences in the data bases but corresponded to proteins whose function is unknown. In vivo mRNA regulation was assessed by Northern blotting to confirm the results of the differential screening. Among the unknown candidates, clone 3 mRNA was up-regulated and clones 5, 4, 2, and 45 mRNAs were down-regulated after antithyroid drug administration. mRNA tissue distribution analysis showed neither a particular tissue distribution for most of the clones studied, nor a parallelism between mRNA expression and proliferation rate. Clones 5 and 45 are more abundantly expressed in the thyroid. Clone 51 appeared to be exclusively present in this tissue, but the signal was very faint.

Northern blotting showed that 5 unidentified cDNAs (2, 3, 4, 5, and 45) out of 13, corresponded to mRNA of which levels were modulated by the cAMP-dependent and -independent mitogenic pathways in dog thyrocytes in primary cultures. This could result from modified transcription, degradation, or both. Remarkably, clone 3 mRNA levels were increased in vivo but not in vitro by TSH. This discrepancy suggests that the in vivo situation does not necessarily match the in vitro stimulation. Whereas in vitro the cAMP-dependent pathway alone is directly activated by TSH or forskolin, the situation in vivo appears to be more complex. The presence of EGF and IGF-1 in the thyroid has indeed been described (14), and the TSH stimulation induced in vivo by the treatment with the antithyroid drugs could indirectly involve these growth factors or even other, undescribed factors. In other words, the effects observed in vivo could be the result of different interacting signaling components, which is not the case in vitro. To confirm clone 3 mRNA up-regulation by agents acting via tyrosine kinase receptors, we investigated the action of HGF, which is the most potent mitogen for our cells (31). HGF reproduced the effects of EGF, but the increase was less pronounced. As HGF has a weaker dedifferentiating action on thyrocytes than EGF, these results may suggest that clone 3 mRNA expression is inversely related to the differentiation state of the thyrocytes.

Among the 19 clones which were apparently differentially expressed in chronically stimulated thyroids, 7 (2, 3, 4, 5, 45, hsp90, and NAP-1) were regulated in vivo, as suggested by the screening, and in vitro by the various mitogenic cascades; three (101, 51, and IRK-2) were not detected by Northern blot analysis, four (16, 59, 137, and 166) led to uninterpretable results, and two (EF-1, and 169) were not regulated.

In our thyrocytes, hsp90 mRNA expression was modulated by chronic treatments: positively by EGF, negatively by TSH and forskolin. As both these agents are mitogenic, it is difficult to relate this expression to proliferation as in some other systems (33, 34, 35, 36, 37).

NAP-1 assembles the nucleosomes by mediating the formation of a histone octamer and transferring it to DNA (38, 39). In our cells, NAP-1 mRNA varied in a cyclical manner in response to TSH. This could correspond to the modulation of the histones, which are well known cell cycle-dependent genes. During the cell cycle of Balb/c 3T3 cells, the content of NAP-1 was found to change in parallel with histone synthesis (40). The modulation of its expression in our cells is compatible with this concept. The expression of a human NAP-related gene also increases rapidly in T cells after induction of cell proliferation, and mitogenesis was inhibited in cells treated with the corresponding antisense oligonucleotide (41).

On the basis of the in vitro mRNA regulation results, we selected first clones 5 and 3 for further characterization. Both were completely sequenced. Clone 5 protein sequence comparison with the data bases revealed 30% identity with the protein kinase family. This large superfamily of homologous proteins contains a 250-300-amino acid kinase domain, composed of 12 conserved subdomains that fold into a common catalytic core structure (32). However, our clone lacks most of these consensus motifs: indeed, only subdomains VIII and IX are present (with an identity of 80%). Subdomain VIII appears to play a major role in the recognition of peptide substrates, and contains several autophosphorylation sites, required for maximal kinase activity in many protein kinases. No clear ATP-binding site is identified as in the Bcr protein serine/threonine kinase (42) and in the human A6 tyrosine kinase (43). Clone 5 could represent a new example of these atypical kinases, and work is currently in progress to test this hypothesis.

Clone 3 protein sequence comparison with the data bases showed that this protein contains 6 ankyrin-like repeats. The ankyrin-like repeat is a 31-33-amino acid motif which consists of a TPLHLA core sequence and 8-10 other well conserved residues. These motifs are found in ankyrin and several other proteins, very heterogeneous with respect to their origin, function, intracellular localization, and number of repeats. These proteins include cell cycle control proteins (e.g. cdc10 gene product of Schizosaccharomyces pombe), putative integral membrane proteins (e.g. Notch gene product of Drosophila melanogaster, involved in cellular differentiation), transcription factors (e.g. NF-B p105 precursor), and others (e.g. IB-like molecules, virally-encoded proteins) (44). More recently, ankyrin repeats have been discovered in INK4 proteins, a family of cell cycle regulators characterized by their selective inhibition of cdk4 and cdk6 (45, 46, 47). Ankyrin-like repeats are involved both in interactions between distinct proteins and within a single molecule, and this can serve a variety of functions. The presence of 6 of such motifs in clone 3 protein suggests that it might interact with other proteins, and these putative partners will be sought. We are now studying the possible role of clone 3 and clone 5 proteins in the proliferation and differentiation of the thyroid cell.