Introduction

The gene coding for PTEN/MMAC1/TEP1 has recently been mapped to chromosome 10q22-23 (Steck et al., 1997) and identified as the gene responsible for susceptibility to Cowden (CD), Lhemitte-Duclos (LDD) and to Bannayan-Zonana syndromes (BZS) (Nelen et al., 1996; Liaw et al., 1997; Scala et al., 1998; Marsh et al., 1998). PTEN encodes a 403 amino acid dual-specificity phosphatase that has been shown to dephosphorylate a special class of lipids, the 3,4,5-triphosphate phosphatidylinositols (PtdIns(3,4,5)P3) (Maehama and Dixon, 1998; Stambolic et al., 1998; Haas-Kogan et al., 1998).

CD, also known as multiple hamartoma syndrome, is an autosomal dominant inherited cancer syndrome characterized by hamartomas of the skin, intestine, breast and thyroid and by increased risk of developing breast and thyroid tumors (Hanssen and Fryns, 1995; Eng, 1998). Benign and malignant thyroid abnormalities occur in almost 70% of CD patients. Benign lesions in CD individuals include hyperproliferative diseases such as thyroiditis, multinodular goiters and follicular adenomas. Development of malignant epithelial thyroid tumors, which are preferentially of the follicular histotype, is observed in almost 10% of CD patients (Longy and Lacombe, 1996).

Several evidences have accumulated indicating that PTEN is a tumor suppressor gene. In fact, homozygous mutations and/or deletions of PTEN have been observed in several sporadic tumors (Li et al., 1997; Steck et al., 1997; Rasheed et al., 1997); furthermore, PTEN is able to suppress cell proliferation both in vitro and in vivo (Furnari et al., 1997; Cheney et al., 1998); finally, mice heterozygous for PTEN knock-out (PTEN+/−) develop different tumors (Di Cristofano et al., 1998; Suzuki et al., 1998; Podsypanina et al., 1999).

On this basis, the PTEN gene represented an obvious candidate for the development of sporadic thyroid tumors. However, PTEN mutations have been detected at low frequency in sporadic thyroid carcinomas (Dahia et al., 1997; Halachmi et al., 1998). This notwithstanding, inactivation of PTEN function could play a role in thyroid carcinogenesis. In fact, recent studies have shown that mice heterozygous for PTEN knock-out (PTEN+/−) develop papillary thyroid carcinomas likely as a result of haploinsufficiency (Di Cristofano et al., 1998; Podsypanina et al., 1999).

In this study, we sought to determine whether PTEN might play a role in the pathogenesis of sporadic thyroid tumors. We investigated: (1) the existence of structural defects in the gene encoding PTEN, in eight tumor-derived cell lines and 26 surgically removed thyroid carcinomas; (2) the mRNA and protein expression of PTEN in the same thyroid tumor cell lines and carcinomas; (3) finally, the effects of PTEN re-expression in thyroid tumor cells. The findings reported in this study indicate that PTEN may act as a potential suppressor of thyroid tumorigenesis and that up-regulation of p27kip1 is required for PTEN-dependent growth suppression.

Results

Mutation analysis of PTEN gene in human thyroid carcinomas and tumor-derived cell lines

We have analysed 26 sporadic thyroid tumors and eight cell lines derived from thyroid carcinomas to detect mutations in the coding region of the PTEN gene by using single strand conformation polymorphism analysis (SSCP). Genomic DNA from exons 1 through 9 was amplified by PCR using intron-specific primers that flank each exon as previously described (Scala et al., 1998). Templates from the blood of two patients with wild type PTEN sequence were included as control. SSCP analysis followed by direct automated DNA sequencing demonstrated the existence of a silent heterozygous C→A transversion at the nucleotide 463 in exon 5 in one tumor DNA from a patient affected by a papillary carcinoma and a double mutation at the nucleotide 489 of exon 5 that caused a AAA→AAT transversion (K163→N) and at the nucleotide 497 on exon 6 that caused a GTA→GAA transversion (V166→E). Interestingly, sequence analysis of PCR products amplified from cDNA derived from this tumor indicated that these mutations were on the same allele (not shown). The double mutation was somatic since it was not detected in genomic DNA extracted from peripheral blood lymphocytes from the same patient. Such double mutation likely impairs PTEN function. In fact, the residues involved are located in a conserved alpha-helix (α7) that helps the formation of enzyme secondary structure; three different mutations that have been reported in this region involving G165, T167 and S170 resulted in loss of PTEN activity by impairment the proper folding of the alpha-helix (Myers et al., 1997). Thus our analysis confirmed that point mutations are infrequent in sporadic thyroid tumors.

Deletion analysis of PTEN gene in human thyroid tumor-derived cell lines

Previous studies have reported the occurrence of heterozygous deletions of PTEN in some thyroid tumors (Dahia et al., 1997; Halachmi et al., 1998). PTEN gene dosage in cultured cell lines was determined by Southern blot analysis as described (Whang et al., 1998). To this aim 10 μg of DNA from NPA, TPC-1, B-CPAP, WRO, ARO and FRO cell lines were digested with HindIII. The human primary thyrocytes HTC-2 cells were used as control. To discriminate between the pseudogene (Ψ) on chromosome 9 (band of 4 Kb) and the PTEN gene on chromosome 10q23 (band of 2.2 Kb) we used a probe derived from exon 5 of PTEN gene as described (Whang et al., 1998). As shown in Figure 1, the loss of one PTEN allele in B-CPAP cells was inferred by the finding of a 2 : 1 ratio between the 4-Kb band (Ψ) and the 2.2-Kb band (PTEN) as determined by densitometric analysis of films. Conversely, no gross alteration of the PTEN locus was observed in the remaining cell lines (at least in the region spanning exon 5).

Figure 1
figure 1

Southern blot analysis. HindIII-digested genomic DNA was separated on 0.8% agarose gel, transferred to nylon filters and probed with PTEN exon 5 DNA. The band representing the pseudogene is marked by Ψ. The band representing the functional PTEN gene is indicated by PTEN. Sources of HindIII-digested genomic DNAs were as indicated. Molecular weight markers are shown on the right

Full size image

Expression analysis of PTEN gene in thyroid tumor-derived cell lines

Northern blot

 To determine whether PTEN expression was down-regulated in thyroid tumors, Northern blot analysis was performed on a panel of eight human thyroid tumor-derived cell lines using a 1.3 Kb cDNA fragment encompassing the whole coding region as probe. Cell lines were as follows: four lines were derived from papillary carcinomas (NIM, B-CPAP, TPC-1 and NPA), one from follicular carcinoma (WRO) and three from anaplastic carcinomas (FRO, ARO and FB-1). Primary thyrocytes (HTC-2) were used as control. Results are shown in Figure 2. The level of PTEN mRNA was determined by densitometric analysis of films. All samples analysed showed expression of two major transcripts of 5.5 and 2.5 kb. The two major PTEN-specific transcripts were expressed at high level in normal thyrocytes (HTC-2) but their amount was reduced in some cell lines (Figure 2a). The GAPDH signal demonstrated that the differences observed could not be ascribed to unequal loading of samples.

Figure 2
figure 2

Expression of PTEN in normal and neoplastic thyroid cells. (a) Northern blot analysis. Total RNA (20 μg) were size-fractionated on denaturing gel, transferred to nitrocellulose and hybridized with PTEN cDNA. GAPDH was used as an internal control for uniform RNA loading. (b) Western blot analysis of PTEN expression in thyroid tumor-derived cell lines. Fifty micrograms of total proteins were resolved on 10% SDS–PAGE, transferred to nitrocellulose filters and probed with anti-PTEN antibodies. Antibodies to α-tubulin were used as loading control. (c) Protein extracts of TPC-1 cells preincubated with excess PTEN peptide before Western blot analysis

Full size image

Western blot

 We also investigated the level of PTEN protein in the same cell lines. The level of PTEN protein was determined by Western blot analysis. High levels of PTEN 60 kDa protein were detected in normal cultured thyrocytes (HTC-2). The levels of PTEN protein were consistently reduced in B-CPAP, NIM, FRO and FB-1 cells (Figure 2b). The finding that the level of PTEN protein correlated with the level of PTEN mRNA (except in WRO cells) suggests that in most cases PTEN expression may be down-regulated through mechanisms that involve reduced transcription rate from PTEN gene or decreased stability of the PTEN transcript. Antibodies to α-tubulin were used as loading control in all Western blot experiments (Figure 2b, lower panel). In Figure 2c, a competition experiment is shown where the 60 kDa PTEN band was efficiently competed by the corresponding peptide, thus demonstrating the specificity of the reaction.

Expression analysis of PTEN gene in human thyroid carcinomas

Northern blot

 Subsequently, we extended the study to human thyroid tumors. We have analysed 61 human fresh carcinomas of the thyroid gland for PTEN expression by Northern blot hybridization. As control, we used six normal or non-neoplastic thyroid tissues. The carcinomas used in this study were of different histological types (39 papillary carcinomas, 14 follicular carcinomas and eight anaplastic carcinomas). The results are summarized in Table 1. PTEN expression in thyroid tumors was quantified through Phosphorimager densitometric scanning of blots as described in the legend of Table 1. High basal levels of PTEN transcripts were found in normal thyroid tissue. Almost 60% of the tumors analysed expressed PTEN mRNA at levels similar to normal gland or slightly decreased. However, a 3–20-fold decrease in PTEN gene expression was observed in the remaining 40% of the cases. A representative experiment is shown in Figure 3a. These findings indicate that as an alternative to the structural mutations that occur in other models (gliomas, prostate cancer, endometrial cancer), downregulation of PTEN expression may represent a common pathway in the development of a subset of thyroid carcinomas.

Table 1 PTEN expression in thyroid tumors
Full size table
Figure 3
figure 3

Expression of PTEN in normal and neoplastic thyroid tissues. (a) Northern blot analysis. Total RNA (20 μg) were size-fractionated on denaturing gel, transferred to nitrocellulose and hybridized with PTEN cDNA. Lane TN, normal thyroid tissue. Lanes 1–18: thyroid carcinomas. (b) Western blot analysis of PTEN expression in thyroid tumor-derived cell lines. Fifty micrograms of total proteins were resolved on 10% SDS–PAGE, transferred to nitrocellulose filters and probed with anti-PTEN antibodies. Antibodies to α-tubulin were used as loading control. Normal thyroid tissue, lane TN; papillary carcinomas, lanes 2, 3, 5, 6 and 7; follicular carcinomas: lane 4; anaplastic carcinomas: lanes 8 and 9

Full size image

Western blot

 We have analysed the protein expression of PTEN by Western blot in 39 primary thyroid carcinomas (27 papillary, eight follicular, and four anaplastic carcinomas). As control, proteins extracted from four non-neoplastic thyroid glands were used. The results are summarized in Table 1. The amount of PTEN protein was quantified through densitometric scanning of films. PTEN protein was detected at high level in normal thyroid tissue whereas it was rather heterogeneous in carcinomas. In fact, PTEN protein was reduced in eight out of 27 papillary carcinomas (29%) in three out of eight follicular carcinomas (37%) and in two out of four anaplastic carcinomas (50%). A representative Western blot is shown in Figure 3b. PTEN is expressed at high level in normal thyroid gland (lane TN) and in some thyroid tumors (lanes 1, 2, 4, 6 and 7) but is markedly down-regulated in other tumors (lanes 3, 5, 8 and 9). Antibodies to α-tubulin were used as loading control in all Western blot experiments. In most clinical specimens there was a general correlation between PTEN mRNA and protein level, though not in all cases.

Constitutive PTEN expression suppresses colony outgrowth in thyroid tumor-derived cell lines

In order to investigate the effects of PTEN re-expression into malignant thyroid cells, we have cloned the PTEN full-length cDNA into the eukaryotic expression vector pCDNA-3 under the transcriptional control of the CMV promoter (pCMV-PTEN) (Figure 4). As cellular model systems we chose three cell lines derived from thyroid carcinomas with different endogenous PTEN levels: the papillary cell line TPC-1 and the anaplastic cell lines ARO and FB-1. Exponentially growing TPC-1, ARO and FB-1 cells were transfected with either pCMV-PTEN or with a control pCMV plasmid. Expression of the exogenous PTEN cDNA into transfected cells was detected by RT–PCR using SP6/PTEN internal primer pair on total RNA. Amplification of the pCMV-PTEN by SP6/PTEN internal primers demonstrated the presence of a band of the expected size (302 bp) that hybridized with PTEN cDNA only in ARO, TPC-1 or FB-1 cells that had been transfected with pCMV-PTEN but not in the same cells transfected by the backbone vector (Figure 4b, upper panel). Furthermore, to detect the total level of PTEN mRNA in ARO, TPC-1 or FB-1 cells, RT–PCR using internal primer pairs on the PTEN cDNA sequence was performed (Figure 5b, middle panel). In this case, RT–PCR performed on total RNA extracted from ARO, TPC-1 or FB-1 cells showed increased levels of a 320 bp band corresponding to the PTEN transcript in PTEN-transfected cells compared to pCMV-transfected cells. Aldolase primers were used as internal control (Figure 4b, lower panel). In parallel, transfected cells were subsequently selected for 10–15 days with G418, and stained with crystal violet (Figure 4c). Constitutive expression of PTEN into exponentially proliferating thyroid tumor cells strongly inhibited the outgrowth of colonies in ARO and FB-1 cells, as demonstrated by the finding that the number of colony/dish and the number of cells/colony in the PTEN-transfected cells was markedly decreased compared to the pCMV-transfected cells (see Table 2). PTEN-dependent inhibition of cell growth was less evident in TPC-1 cells. All the experiments described were performed three times in duplicate.

Figure 4
figure 4

Re-expression of PTEN in thyroid tumor cells inhibits colony outgrowth. (a) The full-length PTEN cDNA was cloned into a pCMV vector. (b) Expression of exogenous PTEN cDNA into ARO, TPC-1 or FB-1 cells was detected by RT–PCR using SP6/1010 (upper panel) or 320/640 (middle panel) internal primer pairs on total RNA extracted from cells transiently transfected by pCMV (lanes 1, 3 and 5) or pCMV-PTEN (lanes 2, 4 and 6), respectively. Primers to amplify the aldolase transcript were used as internal control (lower panel). (c) Colony assay. Exponentially growing FB-1 cells were transfected either with the pCMV-PTEN or with a control pCMV plasmid, selected with G418, and stained with crystal violet

Full size image
Figure 5
figure 5

p27 up-regulation is necessary for PTEN-dependent growth arrest. (a) BrdU incorporation of pFLAG- or pFLAG-PTEN-transfected FB-1 cells. First column: one green pFLAG-transfected cell incorporates BrdU (yellow arrow). Second column: one green pFLAG-PTEN-expressing cell does not incorporate BrdU (red arrow). Third column: one green pFLAG-PTEN-expressing cell incorporates BrdU in the presence of antisense p27kip1 plasmid (yellow arrow). A 40×Neo-Achromat Zeiss objective was used. (b) Statistical analysis of BrdU incorporation in pEGF-PTEN-transfected FB-1 cells. (c) Western blot analysis with α-EGFP, α-Ser473-Akt, α-p27kip1 antibodies of transfected FB-1 cells. Lane 1, pEGFP-transfected cells; lane 2, pEGFP-PTEN-transfected cells; lane 3, pEGFP-PTEN-transfected cells in the presence of an excess antisense p27kip1 plasmid

Full size image
Table 2 Effects of PTEN re-expression in thyroid cancer cell lines
Full size table

Constitutive PTEN expression inhibits S phase entry in thyroid tumor-derived cell lines

Factors that induce cells to arrest proliferation may affect either cell growth or cell death. In particular, it was recently shown that PTEN negatively regulates cell survival conferring enhanced sensitivity to different apoptotic stimuli (Stambolic et al., 1998). To discriminate between these possibilities we made use of a PTEN-encoding construct in which the full-length cDNA coding for PTEN was fused in frame with the autofluorescent Eukaryotic Green Fluorescent Protein (EGFP) or with the N-terminal FLAG tag (Chalfie et al., 1994). All experiments described below were performed three times in duplicate in which at least 200 transfected cells were counted.

ARO or FB-1 cells were plated onto slides in 6-mm dishes and were subsequently transfected with pFLAG-PTEN, pEGFP-PTEN or control empty constructs, respectively. After 48 h post-transfection, cells were either collected for protein extraction or incubated with BrdU to determine growth rate. In the case of pEGFP-PTEN/pEGFP vectors, transfected cells were identified by green autofluorescence of the chimeric protein; in the case of pFLAG-PTEN/pFLAG vectors, transfected cells were identified by green fluorescence emitted by co-transfected EGFP. In both cases, cells incorporating BrdU were identified by use of anti-BrdU antibodies and counted. Results are summarized in Table 2. The inhibition of BrdU incorporation induced by PTEN in thyroid cells was evaluated as described in the legend to Table 2. We observed that both FLAG-PTEN and EGFP-PTEN induced a drastic reduction in the rate of BrdU uptake of ARO and FB-1 cells (35.6 and 86.1% of inhibition of BrdU uptake, respectively). Interestingly, growth inhibition exerted by PTEN re-expression was more effective in the cell line with lower endogenous PTEN levels (FB-1 cells) as compared with cell lines with higher endogenous protein (ARO) cells. A representative experiment on FB-1 cells is shown in Figure 5a.

On the other hand, PTEN-transfected cells showed the same viability as vector-transfected cells, since they were similar in number to vector-transfected cells. In addition, when stained with the nuclear dye Hoechst, PTEN-transfected cells failed to show cell blebbing typical of apoptotic cells (not shown), indicating that PTEN re-expression did not induce morphologically evident apoptosis. These results demonstrated that the expression of PTEN into thyroid tumor cells did not induce apoptosis but rather inhibited DNA synthesis. This result is in agreement with recent works in which PTEN expression was shown to induce G1 arrest but not apoptosis in tumor cells (Da-Ming et al., 1998).

p27 up-regulation is necessary for PTEN-dependent growth arrest in thyroid cancer cells

Finally we addressed the molecular mechanism by which PTEN inhibits S phase entry in thyroid carcinoma cells. In mammalian cells, cell cycle progression is regulated by activation of cyclin-dependent kinases (CDKs), which is controlled through multiple mechanisms (association with cyclin regulatory subunits, phosphorylation, binding to inhibitory proteins defined cyclin-dependent kinase inhibitors) (Sherr and Roberts, 1995). Up-regulation of the cyclin-dependent kinase inhibitor p27kip1 by PTEN has previously been reported to occur in different cell lines (Da-Ming and Hong, 1998; Cheney et al., 1999). We investigated whether PTEN was able to up-regulate p27kip1 expression in thyroid cancer cells by Western blot. As shown in Figure 5c, FLAG-PTEN-transfected FB-1 cells showed increased expression of p27kip1 protein compared to untransfected or vector-transfected cells. Also in ARO cells, expression of FLAG-PTEN induced up-regulation of p27kip1 expression (not shown). Same results with both cell lines were obtained when the pEGFP-PTEN/pEGFP vectors were used (not shown).

The data shown in this study as well as others indicate that p27kip1 up-regulation is a common event associated with PTEN expression (Da-Ming and Hong, 1998). On this basis, we investigated whether PTEN-induced up-regulation of p27kip1 expression was necessary for PTEN inhibitory activity. To this aim, we transiently transfected the pFLAG-PTEN construct into FB-1 cells in the presence of an excess of a plasmid encoding the p27kip1 cDNA in the antisense orientation, and determined BrdU incorporation rate as a measure of S phase entry. When FB-1 cells were transfected with pFLAG control vector, BrdU incorporation rate was not grossly modified (26±3.5% of pFLAG-transfected versus 28% of untransfected cells, stained for BrdU). However, when pFLAG-PTEN was transfected, BrdU incorporation rate was consistently reduced to less than 4% (3.9±0.7). Such reduction in the rate of DNA synthesis induced by PTEN could be rescued completely in the presence of a fourfold excess antisense p27kip1 plasmid (Figure 5b). In this case, more than 26±3% of PTEN-transfected FB-1 cells incorporated BrdU. Similar results were observed if pFLAG-PTEN was transfected in the presence of 1 μM p27kip1 antisense oligonucleotides (not shown). Figure 5a shows a representative experiment: one pFLAG-transfected cell incorporates BrdU (yellow arrow) whereas none of the pFLAG-PTEN-transfected cells do (red arrow); however, when FB-1 cells were transfected with pFLAG-PTEN in the presence of a molar excess of p27kip1 antisense plasmid PTEN-transfected cells incorporate BrdU (yellow arrows). The Western blot shown in Figure 5c demonstrated that PTEN increased p27kip1 expression in FB-1 cells and that such increase can be prevented by use of antisense p27kip1 plasmid.

It has been shown that PTEN gene product regulates negatively the activity of protein kinase B/Akt (Stambolic et al., 1998). Such inhibition of Akt activity is relevant for PTEN growth-suppression. In fact, in several human tumors and cell lines, PTEN expression inversely correlates with Akt activation measured as increased phosphorylation at specific serine (ser473) and threonine (thr308) residues (Dahia et al., 1999; Haas-Kogan et al., 1998). Thus we determined whether growth inhibition induced by PTEN in FB-1 cells was accompanied by a reduction in the level of phosphorylated Akt. To this aim, we made use of an anti-Akt antibody that recognized specifically Ser473-phosphorylated Akt in immunoblot assays. Total levels of Akt protein in cell lysates were determined by use of a phosphorylation state-independent Akt antibody. As shown in Figure 5, PTEN transfection into FB-1 cells resulted in a reduction of the phosphorylation of the PTEN-regulated protein-kinase B/Akt (Figure 5c). Altogether, these results indicate that up-regulation of the cyclin-dependent inhibitor p27kip1 is required for PTEN growth-suppressing activity in thyroid tumor cells and that this likely occurs through reduction of Akt activity.

Correlation between PTEN expression, Akt phosphorylation and p27 expression

Several works have shown that PTEN expression inversely correlates with the phosphorylation status of Akt (Dahia et al., 1999; Haas-Kogan et al., 1998). However, there exists no information about the levels of p27kip1 relative to PTEN expression. Theoretically, if p27kip1 expression is dependent on intracellular PTEN levels, one would expect to observe low levels of p27kip1 as a consequence of the loss in PTEN expression. To address this issue we performed a Western blot analysis of the levels of PTEN, P-Akt and p27kip1 protein in eight tumor-derived cell lines and in a small panel of primary thyroid carcinomas. Results are shown in Figure 6b and a, respectively.

Figure 6
figure 6

Analysis of PTEN expression, p27 expression and Akt phosphorylation in human thyroid tumors and cell lines. (a) Western blot analysis of PTEN, p27kip1 and Akt expression and Ser474 phosphorylation in thyroid carcinomas. Fifty micrograms of total proteins were resolved on 10% SDS–PAGE, transferred to nitrocellulose filters and probed with anti-PTEN antibodies (first panel), anti-p27kip1 (second panel), anti-Ser473-specific antibodies (third panel) and anti-Akt antibodies (lower panel). (b) Western blot analysis of PTEN, p27kip1 and Akt expression and Ser474 phosphorylation in thyroid carcinoma-derived cell lines. Fifty micrograms of total proteins were resolved on 10% SDS–PAGE, transferred to nitrocellulose filters and probed with anti-PTEN (first panel), anti-p27kip1 antibodies (second panel), anti-Ser473-specific antibodies (middle panel) and anti-Akt antibody (lower panel)

Full size image

We found that Akt phosphorylation was increased in five out of 13 tumors compared with normal thyroid tissue; moreover, PTEN expression inversely correlated with the presence of phosphorylated Akt in a small panel of thyroid tumors with different PTEN expression (eight papillary, four follicular and one anaplastic carcinomas). In eight tumors that presented high PTEN expression (Figure 6a, lanes 2, 3, 5, 8, 10–13) low levels of phosphorylated Akt was found; only in one case (lane 2) we observed a high degree of Akt phosphorylation in the presence of PTEN expression. On the contrary, in five tumors in which PTEN expression was reduced consistently (Figure 6a, lanes 1, 4, 6, 7 and 9) we observed a high degree of phosphorylated Akt in three cases (Figure 6a, lanes 1, 7 and 9) and a moderate degree of phosphorylated Akt in two other cases (Figure 6a, lanes 4 and 6, respectively). When p27kip1 and PTEN expression in the same thyroid tumors were matched, we observed a striking correlation (compare first and second panels in Figure 6a). In fact, out of eight tumors in which PTEN expression was high (Figure 6a, lanes 2, 3, 5, 8, 10–13), p27kip1 levels were elevated in five (Figure 6a, lanes 2, 8, 11–13), intermediate in two (Figure 6a, lanes 3 and 5) and absent only in one case (Figure 6a, lane 10). Conversely, p27kip1 was completely absent in tumors that presented low PTEN expression (Figure 6a, lanes 1, 6, 7 and 9).

Similar results were obtained when we analysed thyroid tumor-derived cell lines: Akt phosphorylation was increased in four out of eight cell lines compared with primary normal thyrocytes; in most cell lines PTEN and P-Akt presented an inverted pattern of expression (except B-CPAP cells) which could not be accounted for by reduced Akt levels (Figure 6b); p27kip1 expression was higher in those cell lines that retained some PTEN expression.

Thus it appears that in the process of thyroid carcinogenesis, the loss of PTEN in a subset of tumors may result in reduced p27kip1 expression, possibly due to the activation of the protein kinase B/Akt pathway.

Discussion

PTEN is a newly identified tumor suppressor gene that has been implicated in a wide variety of cancers (Steck et al., 1997; Li et al., 1997). Since germline mutations in the PTEN gene are responsible for a dominantly-inherited autosomal disorder characterized by increased susceptibility to malignancies of the thyroid gland, it was an obvious candidate for the development of the corresponding sporadic cancer. The results presented in this study indicate that although the frequency of PTEN mutations is low in sporadic thyroid carcinomas, PTEN inactivation may represent a critical step in the development and/or progression of thyroid cancer. In fact, we observed that the primary mode of PTEN inactivation in tumors of the thyroid gland does not involve gene mutation and/or deletion but occurs through down-regulation of its expression. The expression of PTEN mRNA and protein is reduced in several thyroid tumor-derived cell lines and in almost 40% of clinical specimen. Because it seems that the reduction of PTEN expression is observed both in well differentiated and in poorly differentiated tumors, from our data it is not clear whether PTEN down-regulation occurs early or late in disease progression. However, it is of note that anaplastic carcinomas present a slightly higher frequency of reduction of PTEN expression compared with well differentiated carcinomas (50 versus 37%), in agreement with previous reports in which PTEN gene has been found mutated and/or deleted in advanced cancers and hence defined as MMAC1 (Mutated in Multiple Advanced Cancer) (Steck et al., 1997).

As to the mechanisms whereby PTEN expression is down-regulated in thyroid carcinomas, our data support the existence of different mechanisms. In a small number of carcinomas (six samples), where protein expression was reduced despite high levels of mRNA, a mechanism involving enhanced degradation rate of the protein can be envisaged. However, in most cases we found a good correlation between mRNA and protein levels in most cell lines and surgically removed carcinomas. In this case, PTEN downregulation could be due to different molecular mechanisms (i.e. deletion of the gene, transcriptional silencing etc.). Although previous studies have reported that monoallelic deletion of the PTEN gene at 10q23 occurs in approximately 3–30% of tumors analysed (Dahia et al., 1997; Halachmi et al., 1998), several considerations suggest that monoallelic deletion of PTEN can not account completely for the reduced PTEN mRNA expression reported in this study. First, the frequency of PTEN gene alterations reported in the literature (Dahia et al., 1997; Halachmi et al., 1998) is somewhat lower compared with the frequency of PTEN-deficient tumors observed in this study. Second, although most tumor cell lines presented reduced PTEN expression, only one (B-CPAP cells) had a heterozygous alteration spanning exon 5, which indicates that gene deletion may not be a major cause of the reduction of PTEN expression in these cells. Finally, in hematological malignancies, monoallelic deletion of the PTEN gene does not result in a marked reduction or loss of PTEN expression (Dahia et al., 1999). Possible alternative molecular mechanisms that may account for the low mRNA levels observed in thyroid tumor cells are the presence of mutations in the promoter of the gene and/or in the 3′ untranslated region of the transcript (that give rise to an unstable transcript) or methylation of CG dinucleotides at the promoter/enhancer region as in prostate cancer (Whang et al., 1998). Further studies are necessary to discriminate between these possibilities.

The down-regulation of the PTEN expression observed in a subset of thyroid tumors is a novel finding, and suggest that reduced PTEN expression may be involved in the development of thyroid cancer. Several lines of evidence support the concept that even slight changes in the level of PTEN expression is critical to its tumor suppressor activity. Twofold overexpression of PTEN mRNA significantly inhibits cell migration, spreading and formation of focal adhesion plaques in vitro (Tamura et al., 1998). Heterozygous PTEN+/− mice develop both follicular and papillary thyroid carcinomas, probably as a consequence of haploinsufficiency, thus providing an in vivo evidence that the level of PTEN expression is important for its tumor suppressing activity, particularly in thyrocytes (Di Cristofano et al., 1998; Podsypanina et al., 1999).

Recent papers have demonstrated that PTEN is involved in the regulation of several cellular processes such as growth and survival through negative control of the protein kinase B/Akt pathway (Bellacosa et al., 1991; Stambolic et al., 1998; Haas-Kogan et al., 1998). The protein kinase B/Akt pathway is activated by phosphoinositide-3-kinase (PI3K), which generates 3-phosphate phosphatidylinositols, that binding to the PH domain, recruit Akt to the plasma membrane where it is activated by phosphorylation of Ser473 and Thr308 (Marte and Downward, 1997; Downward, 1998). In turn, PI3K is activated by tyrosine kinases via its regulatory p85 subunit or by ras via its catalytic p110 subunit. By removing the phosphate group on the D3 position of the inositol ring of 3-phosphate phosphatidylinositols, PTEN negatively regulates Akt activity (Stambolic et al., 1998; Maehama and Dixon, 1998; Myers et al., 1998).

Although the exact role of PI3K and Akt in thyroid tumorigenesis is yet to be defined, it appears that molecules that lead to the activation of this pathway are involved in the development of thyroid tumors (i.e. RTKs, ras etc.), indicating that activation of the PI3K/Akt pathway may represent a crucial event during thyroid cancerogenesis. In fact, ras mutations occur in 50% of follicular carcinomas and activation through chromosomic rearrangement of the tyrosine kinase receptor TRK-A, and RET/PTC genes occurs in approximately 40% of papillary carcinomas (Santoro et al., 1992). The findings reported in this work add new insights into the molecular biology of thyroid tumors, since it suggests that in addition to the constitutive activation of stimulatory molecules (ras, RET/PTC, TRK-A), the PI3K/Akt pathway could be activated through loss of expression and/or function of an upstream inhibitory molecule such as PTEN. Accordingly, we have observed that in a small set of carcinomas and cell lines, decreased PTEN expression correlated with increased phosphorylation of the PTEN-regulated protein kinase Akt/PKB.

Our results also demonstrate that restoration of PTEN expression in thyroid tumor cells suppresses growth and colony formation in vitro. In particular, it appears that PTEN re-expression in thyroid tumor cells does not induce apoptosis but rather it blocks progression through cell cycle by impairing entrance into S phase. Furthermore, this study provides straightforward evidence that the cyclin-dependent kinase inhibitor p27kip1 mediates, at least in part, the effects exerted by PTEN on cell cycle progression. In fact, constitutive PTEN expression induces up-regulation of p27kip1 in thyroid cancer cells (FB-1 and ARO), and antisense p27kip1 plasmid or oligonucleotides prevent PTEN-imposed growth arrest (FB-1). The observation that PTEN induces up-regulation of p27kip1 also in cell lines of non-thyroid origin as embryonal kidney or glioblastoma cells (Bosc23 cells, not shown; Da-Ming and Hong, 1998) strengthens the hypothesis that the activities of these two proteins are interconnected in most cells. Intriguingly, we report that in a subset of thyroid carcinomas, PTEN loss is associated with reduced p27kip1 protein expression, thus providing support to the concept that p27kip1 represents one of the relevant downstream molecular targets of PTEN signaling in vivo.

As to the mechanism involved, our data suggest that the chronic activation of the PI3K/Akt pathway (due to the absence of PTEN expression) may account for the observed decrease in p27kip1 expression in most cases. However, the finding that p27kip1 expression is retained in a few tumors with highly activated Akt, and conversely, that its expression is lost in one or two tumors in which Akt is not active, suggest that different pathways are involved in the regulation of p27kip1 expression in thyroid cancer cells. Since p27kip1 phosphorylation leads to its turnover, one possibility is that, by directly or indirectly increasing the rate of p27kip1 phosphorylation, the PI3K/Akt pathway targets it to degradation and that PTEN would decrease the rate of p27kip1 turnover by preventing activation of this pathway. Accordingly, we have recently shown that p27kip1 expression is lost in approximately 40% of primary thyroid carcinomas (Baldassarre et al., 1999).

In conclusion we report that inactivation of the PTEN function is a critical step in the development and/or progression of thyroid cancer. In fact, although we detected few mutations in the PTEN gene in thyroid tumors and cell lines, we found that most cell lines and approximately 40% of primary tumors presented reduced or loss of PTEN expression. Furthermore, it appears that PTEN may act as a potential suppressor of thyroid tumorigenesis since PTEN re-expression into cell lines markedly inhibited cell growth, likely through up-regulation of expression of the cyclin-dependent kinase inhibitor p27kip1.

Materials and methods

Cell lines and tissue samples

The human thyroid carcinoma cell lines used in this study have been previously described (Baldassarre et al., 1999). All cell lines were grown in Dulbecco's modified eagle medium (DMEM) containing 10% fetal calf serum. Tumor samples were obtained at the National Cancer Institute ‘Fondazione Pascale’, Naples, Italy and at the Laboratoire d'Histologie et de Cytologie, Centre Hospitalier, Lyon Sud, France.

PCR–SSCP analysis and DNA sequencing analysis

PCR amplifications and SSCP analysis of exons 1–9 of the gene encoding PTEN were carried out in patients and control genomic DNAs as described (Scala et al., 1998). Sequencing of the PCR product was performed by cycle sequencing on an Applied Biosystem 377 automated DNA sequencer, using the PRISM AmpliTaq FS Ready Reaction Dye Terminator sequencing kit (Applied Biosystem and Perkin-Elmer Cetus). PCR products from patients were cloned into the pCR2.1 TA vector (Invitrogen) and five different recombinant clones from each affected patient were sequenced with M13 universal and reverse primers.

RNA extraction, Northern blotting and hybridization

Extraction of total cellular RNA and Northern blot hybridization was performed essentially as described (Sambrook et al., 1989). All cDNA probes were radio-labeled with a random priming synthesis kit (Amersham Inc.). The PTEN probe used in this study was the complete coding sequence of human cDNA.

Protein extraction, Western blotting and antibodies

All the antibodies used for Western blot were from Santa Cruz. Cells were scraped in ice-cold PBS and lysed in Nonidet-P40 lysis buffer (0.5% NP-40, 50 mM HEPES pH 7, 250 mM NaCl, 5 mM EDTA supplemented with NaF, Na3VO4, PMSF, aprotinin and leupeptin). Proteins were separated on polyacrylamide gel, transferred to nitrocellulose filter membranes (Hybond C, Amersham Inc.), blocked in 5% non-fat dry milk, incubated with primary antibodies for 1 h at room temperature and revealed by enhanced chemiluminescence (ECL, Amersham Inc.).

Southern blotting and hybridization

For Southern analysis 10 μg of genomic DNA was isolated from cultured cell lines digested with HindIII, separated by agarose gel electrophoresis and transferred to nylon Hybond-N+ membranes (Amersham Inc.). The PTEN probe used for Southern blot was a 300 bp fragment spanning exon 5 which allowed discrimination between PTEN and pseudogene sequences.

Vectors

The PTEN cDNA was obtained from normal thyroid tissue by RT–PCR performed according to manufacturer's conditions (Perkin Elmer-Cetus). Amplified DNA was cloned into pCRII vector (Invitrogen Inc.) and sequenced. PTEN-specific oligonucleotide primer sequences were chosen in order to amplify full length coding sequence according to cDNA sequence; forward primer (nucleotides 977–1004): 5′-CCACCAGCAGCTTCTGCCATCTCTCTCC-3′; reverse primer (nucleotides 2262–2287) 5′-TTTATTTTCATGGTGTTTTATCCCTC-3′. Subsequently, PTEN cDNA was cloned into pcDNA 3 expression vector (Invitrogen Inc.) under the control of CMV promoter (pCMV-PTEN) or in frame into the EcoRI–BamHI sites of pEGFP-C (Clontech Inc.) or pFLAG (Sigma). In all cases the correct cloning of PTEN constructs were confirmed by DNA sequencing.

Colony assays

Ten million cells were transfected in duplicate by electroporation and subsequently plated in duplicate in 10-mm dishes. After 24 h, selection was started by adding G418 to the culture medium. Cells were selected for 10–14 days and subsequently stained with crystal violet. The number of colonies and the average number of cells/colony was subsequently determined.

Immunofluorescence analysis

BrdU incorporation assay was performed as follows: 5×105 cells were transfected with 6 μg each of control empty vector of the PTEN-encoding construct, respectively. Labeling procedure was carried out as recommended by manufacturer (Boehringer Mannheim Biochemicals). Fluorescence was visualized with Zeiss 140 epifluorescent microscope equipped with filters allowing discrimination between Texas Red and Fluorescein.