Differential Nuclear and Cytoplasmic Expression of PTEN in Normal Thyroid Tissue, and Benign and Malignant Epithelial Thyroid Tumors

Thyroid Samples

Paraffin blocks from 139 unselected benign and malignant nonmedullary thyroid tumors were ascertained from Germany, Australia, and Switzerland. Histological classification of the thyroid tumors was in accordance with the World Health Organization.28 Of note, five papillary tumors were classified as follicular type (Lindsay tumor), and 13 tumors (seven FAs, five FTCs, one PTC) had a prominent granular eosinophilic-appearing cytoplasm (also known as oxyphilic or Hürthle cell).

Anti-PTEN Antibody Specificity

The monoclonal antibody 6H2.1 raised against the last 100 C-terminal amino acids of PTEN29 was used in all immunohistochemical analyses. As biochemical proof of antibody specificity for PTEN, total protein lysates were obtained10,15 from a series of thyroid cell lines for which PTEN status is known: NPA-87, K-1, FTC-133, and WRO-82–1 (gifts from D. V. Canlapan and D. Wynford-Thomas). Further, as an additional positive control, the wild-type full-length human PTEN cDNA sequence was cloned into the expression vector pcDNA3 and transfected into the PTEN null line FTC-133. Western blot analysis was performed as previously described10 except that 6H2.1 was used at a 1:250 dilution. Thyroid lines with endogenously expressing or exogenously introduced PTEN all demonstrated a single band at 55 kd, the molecular weight predicted for PTEN, although the PTEN null lines did not cross-react with 6H2.1. No other nonspecific bands were noted, thus proving antibody-specificity (Figure 1)▶ . Control antibody against α-tubulin (Sigma, St. Louis, MO), used at 1:10,000 dilution, immunoreacted evenly across protein lysates from all cell lines (Figure 1)▶ . The specificity of the antibody 6H2.1 and its suitability for immunohistochemistry in paraffin-embedded tissue has been demonstrated previously.29 In brief, we used the antibody against embedded PTEN-transfected U2OS cells, BALBc/3T3, Nalm6, and DU145 as positive controls; MDA-MB-468 with hemizygous deletion of PTEN and a truncation of the remaining allele; A172 which has loss of one PTEN allele, and a truncating mutation in exon 2 of the remaining allele; and PC3, which is null for PTEN.29 Further, commercially available peptide corresponding to PTEN has been used to successfully compete away 6H2.1 immunostaining in paraffin-embedded tissue (GL M, unpublished data).

Recently, it has been shown that a processed PTEN pseudogene (psiPTEN) can be transcribed in a number of cell lines and tissue types.30 For this reason, RNA in situ hybridization is not reliable. PsiPTEN, however, does not seem to be translated, at least in thyroid tumors, and therefore it would not be expected to complicate the analysis in this study.

Immunohistochemistry

The tissue samples were fixed by immersion in 10% buffered formalin and subsequently embedded in paraffin according to standard protocols. Four-mm sections were cut, mounted on Superfrost plus slides, and baked for 2 hours at 60°C. Subsequently, the sections were deparaffinized and rehydrated by passing through xylene and a graded series of ethanol solutions. Antigen retrieval was performed by boiling at 98°C in 0.01 mol/L sodium citrate buffer, pH 6.4, in a microwave oven for 20 minutes. To block endogenous peroxidase activity, the sections were incubated with 0.3% hydrogen peroxide in methanol for 30 minutes after cooling to room temperature. After blocking for 30 minutes in 0.75% horse serum, the sections were incubated with a PTEN monoclonal antibody 6H2.1 (dilution 1:100) for 1 hour at room temperature. Primary antibody binding was localized by using an avidin-biotin-peroxidase kit (Vector Laboratories, Burlingame, CA) according to the manufacturer’s instruction. The chromogenic reaction was carried out with 0.05% 3,3′diaminobenzidine (Sigma, St. Louis, MO) using nickel cobalt amplification which gives a black product.31 After counterstaining with Nuclear Fast Red (Rowley Biochemical Institute, Danvers, MA) and mounting, the slides were independently evaluated under a light microscope by two investigators (OG and AP) and randomly spot evaluated by a third investigator (CE). Intensity of staining was classified separately for the nucleus/nuclear membrane and the cytoplasm and graded strong (+++), moderate (++), weak (+), or absent (−). These independent assessments did not differ by more than one grading level.

LOH Analysis

In 95 samples, tumor tissue and blood or corresponding normal tissue (either normal thyroid tissue or adjacent muscle tissue) were available for extraction of paired somatic and germline DNA to study LOH. DNA extraction after microdissection was performed using standard protocols.32 All subsequent polymerase chain reactions were carried out using 0.6 μM each of forward and reverse primer in 1× polymerase chain reaction buffer (Qiagen, Valencia, CA), 4.5 mmol/L MgCl₂ (Qiagen), 1× Q-buffer (Qiagen), 2.5 U HotStarTaq polymerase (Qiagen), and 200 μmol/L dNTP (Gibco, Gaithersburg, MD) in a final volume of 50 μl. Reactions were subjected to 35 cycles of 94°C for 1 minute, 55°C to 60°C for 1 minute, and 72°C for 1 minute followed by 10 minutes at 72°C. Potential hemizygosity at the PTEN locus was assessed by screening for a T/G polymorphism within PTEN intron 8 (IVS8 + 32G/T) detected by differential digestion with the restriction endonuclease HincII as previously described25 except for using the primers PTEN-E8-F (5′-GCGTGCAGATAATGACAAGG-3′) and PTEN-I8-R (5′-TGTCAAGCAAGTTCTTCATCG-3′). If the result of the digestion was not informative, LOH analysis was performed using markers flanking PTEN, D10S541 (telomeric) and D10S579 (centromeric)1,16 as well as the marker D10S2491 that lies within PTEN.33 All forward primers were 5′-labeled with either HEX or 6-FAM fluorescent dye (Research Genetics, Huntsville, AL). Polymerase chain reactions were carried out as described above and separated by electrophoresis through 6% denaturing polyacrylamide gels using an Applied Biosystems model 377 automated DNA sequencer (Applied Biosystems, Perkin-Elmer Corp., Norwalk, CT). The results were analyzed by automated fluorescence detection using the GeneScan collection and analysis software (GeneScan, Applied Biosystems). Scoring of LOH was performed by inspection of the GeneScan analysis output. A double peak, observed in the microsatellite marker that was amplified from DNA extracted from the germline sample, indicated heterozygosity. A single peak in DNA extracted and amplified from the corresponding tumor sample indicated a loss of one allele. If normal cells were admixed with tumor cells, a ratio of 1.5:1 or greater of germline DNA peak to tumor DNA peak was also considered LOH.24

To examine the correlative trend between PTEN staining intensity and LOH, we performed a Mantel-Haenszel test34 for trend in the association between the row and column variables. A P < 0.05 was considered statistically significant.

PTEN Immunohistochemistry in Normal Thyroid and Primary Thyroid Tumors

Of the total 139 thyroid tumor samples examined for PTEN expression using the monoclonal antibody 6H2.1, 50 had accompanying normal thyroid tissue. Normal follicular thyroid cells showed a uniform strong (+++) to moderate (++) nuclear or nuclear membrane (hereafter referred to as nuclear) signal whereas the cytoplasmic staining was less strong, + to ++ (Figure 2A)▶ . Endothelial cells showed strong (+++) to moderate (++) PTEN expression with a nuclear predominance and were useful as internal positive controls (Figure 2▶ , B and E). In contrast, nuclear and cytoplasmic staining intensity of fibrocytes was very heterogeneous and varied from weak (Figure 2A)▶ to strong (Figure 2▶ F).

The quality and intensity of PTEN immunostaining in the nucleus and cytoplasm in 132 of 139 thyroid tumors was relatively uniform throughout each specimen. However, in seven carcinomas, PTEN expression differed significantly within different regions of each tumor (see below). Because PTEN expression in each of these seven tumors could not be classified into a single category, these different regions were classified separately as if they were two separate tumors, ie, PTEN expression was classified in 139 + 7 = 146 thyroid tumors. Hence, the intensity of PTEN staining in the nucleus and cytoplasm were assessed for 146 thyroid tumors (Figure 3)▶ for purposes of this study.

In FA, the neoplastic nuclei had less intense PTEN immunostaining (+ to ++) compared to normal follicular epithelium whereas the cytoplasmic PTEN staining intensity did not differ significantly from that observed in normal follicular cells (Figures 2B and 3)▶ ▶ . In thyroid carcinomas (FTC, PTC, and undifferentiated thyroid carcinoma [UTC]) as a group, nuclear PTEN immunostaining was mostly weak in comparison with normal thyroid follicular cells and FAs. Among the three classes of carcinomas, FTCs had the strongest immunostaining in both the nucleus and cytoplasm and UTCs the weakest (Figures 2, C–E, and 3)▶ ▶ . A few carcinomas, in particular UTCs, showed no PTEN staining in the nucleus (FTC, 3 of 28; PTC, 3 of 37; UTC, 8 of 26) or in the cytoplasm (one FTC, two PTCs, five UTCs; Figure 2▶ E); eight carcinomas (one FTC, two PTCs, five UTCs) had no immunoreactivity in the nucleus or in the cytoplasm. In almost half of all thyroid carcinomas (FTC, 46%; PTC, 49%; UTC, 35%), the cytoplasmic staining was more intense than the nuclear staining (Figures 2, F and G, and 3)▶ ▶ . In contrast, the more intense cytoplasmic over nuclear staining was observed in only 7% of FAs. In other words, the stepwise decrease in PTEN immunoreactivity in the nucleus seemed to precede that in the cytoplasm from normal thyroid tissue to FA to carcinomas and finally to UTC (Figure 3)▶ . There was no obvious difference in PTEN staining pattern and intensity in Hürthle cell tumors compared to non-Hürthle cell tumors. Fibrocytes seemed to stain a little more intensely in tumor stroma (Figure 2▶ F) compared to normal thyroid stroma (Figure 2A)▶ .

Seven carcinomas (four UTCs, two PTCs, one FTC), but no adenomas, showed dichotomous regional PTEN staining within each sample. These were characterized either by islands of strongly immunopositive cells among sheets of cells staining more weakly (Figure 2▶ , H and I) or by single cells staining weakly randomly distributed among cells staining strongly. In one UTC, the positive cells (graded +) were small and more regular whereas the PTEN-negative cells were larger, pleiomorphic, and had a more undifferentiated appearance. This correlation, however, was not seen in the other three UTCs. This pattern of immunostaining in the former UTC was replicated several times, thus indicating that this was not an artifact.

Comparison of Immunohistochemical and LOH Data

In 95 tumors, normal and tumor tissue samples were available for LOH analysis. Similar to the immunohistochemical analysis, the four available carcinomas with dichotomous staining intensity within each sample, which had paired normal and tumor tissue, were considered as eight samples. Therefore, 99 total samples were considered. For purposes of this study, which was to compare the immunohistochemical data to the LOH data, one copy of PTEN was considered to be physically deleted only when one or more of the markers, which lie within or closely flanking the gene, showed LOH. Using this definition for monoallelic PTEN deletion, LOH data from 92 tumors were informative. Of the 92 informative tumors, 27 tumors (29%) were shown to have loss of one allele of PTEN and 65 tumors were classified as having no LOH (Table 1)▶ .

LOH analysis was performed separately for regions with different staining intensity in those four carcinomas with dichotomous staining intensity and available paired normal and thyroid tissue. One of these four tumors showed LOH whereas the other three tumors did not. Interestingly, in this tumor, LOH was identified in both immunoreactive-positive and -negative regions.

Among the 92 informative tumors assessed for LOH and PTEN immunostaining, there seemed to be an associative trend between decreased or absent staining and 10q23 LOH (Table 2▶ , Figure 3▶ ). The proportion of tumors that had LOH steadily increased from tumors with +++ nuclear staining (0% with LOH), ++ staining (21% LOH), + staining (32% LOH) to no (−) staining (75% LOH) (P = 0.003). This trend was also mirrored if we considered only cytoplasmic staining and LOH status (++ cytoplasmic staining [18% with LOH], + staining [36% LOH], to no [−] staining [67% LOH]) (Table 2▶ ; Figure 3▶ ; P = 0.008). Two samples showed no PTEN staining without evidence for LOH. One of the samples was heterozygous for IVS8 + 32G/T whereas the other markers were not informative. The other sample showed regions of heterozygosity at D10S579 and was not informative for the other markers.