Learn more: PMC Disclaimer | PMC Copyright Notice
Epigenetic PTEN Silencing in Malignant Melanomas without PTEN Mutation
Abstract
A tumor suppressor gene at 10q 23.3, designated PTEN, encoding a dual specificity phosphatase with lipid and protein phosphatase activity, has been shown to play an important role in the pathogenesis of a variety of human cancers. Germline mutations in PTEN cause Cowden syndrome (CS), which is characterized by multiple hamartomas and a high risk of breast and thyroid cancers. Frequent loss of heterozygosity at 10q is found in both early and advanced-stage sporadic melanomas; however, mutations or deletions in PTEN are detected mainly in melanoma cell lines. In this study, we examined PTEN expression in 34 unselected sporadic melanomas (4 primary melanomas, 30 metastases) using immunohistochemistry and correlated this with the results of structural studies of this gene. Immunostaining of 34 melanoma samples revealed no PTEN expression in 5 (15%) and low PTEN expression in 17 (50%), whereas the rest of the tumors (35%) had high levels of expression. Hemizygous deletion was found in 32% of the tumors but neither intragenic PTEN mutation nor biallelic deletion was found in any of the samples. Of the 5 melanomas showing no PTEN expression, 4 had no mutation or deletion of PTEN. Of the 13 tumors having weak PTEN immunoreactivity and informative loss of heterozygosity results, 6 had evidence of hemizygous allelic loss of PTEN while the remaining 7 had intact PTEN. These results strongly support PTEN as a major tumor suppressor on 10q involved in melanoma tumorigenesis and suggest an epigenetic mechanism of biallelic functional inactivation not previously observed in other cancers where PTEN might be involved.
PTEN/MMAC1/TEP1 was recently identified as a tumor suppressor gene located at 10q23.3 and has been shown to be deleted or mutated in a variety of advanced tumors. 1-3 PTEN is a dual-specificity phosphatase with lipid and protein phosphatase activity. PTEN functions as a major lipid phosphatase, dephosphorylates PtdIns-3, 4, 5-triphosphate (PIP-3) and PtdIns-3, 4-diphosphate (PIP-2), which are required for the activation of PKB/Akt, an important mediator of cell survival protecting cells from apoptosis. 4-7 PTEN has been shown to be involved in cell migration, spreading, and focal adhesion formation through dephosphorylating focal adhesion kinase (FAK), presumably through its protein phosphatase activity. 8,9 Ectopic expression of PTEN results in cell cycle arrest at G1 and/or apoptosis in glioma and breast cancer cell line models. 5,10,11
Germline mutations of PTEN have been found in the autosomal dominant hamartoma syndrome, Cowden Syndrome (CS), which is characterized by multiple hamartomas and an increased risk of malignant and benign breast and thyroid tumors, and Bannayan-Riley-Ruvalcaba (BRR) syndrome, a related disorder characterized by neonatal onset macrocephaly, lipomatosis, hemangiomas, and speckled penis, as well as some cancer risk. 12-15 Recently, a Proteus-like syndrome was found to result from germline and germline mosaic PTEN mutations. 16
PTEN has been shown to be somatically deleted or mutated, to a greater or lesser extent, in a wide variety of sporadic advanced tumors, especially glioblastoma multiforme, endometrial carcinoma and advanced prostate cancers. 17-21 Most melanomas occur without any family history of melanocytic tumors and it is controversial whether melanomas are true component tumors of CS. Hereditary melanomas only account for approximately 10% of all clinical presentations; thus, the great majority are sporadic. 22 Involvement of genes that mediate growth arrest via cell cycle regulation, such as p16, has been found in a subset of hereditary and sporadic melanomas. 23 Loss of genetic material of the long arm of chromosome 10 has been detected in 30 to 50% of both early and advanced-stage sporadic melanomas, and has been associated with poor clinical outcome. 24,25 Given the frequent loss of heterozygosity (LOH) at 10q in melanomas, the 10q localization of the PTEN gene and the role of PTEN in cell cycle regulation and apoptosis, PTEN may be viewed as an excellent candidate melanoma growth suppressor to play a role in melanocytic oncogenesis. There are at least eight studies, comprising various numbers of tumors, examining the genomic status of PTEN in melanomas of various stages and principally, melanoma cell lines, with inconsistent results. 2,26-32 Despite the inconsistent results, it appeared that homozygous deletions and intragenic mutations of PTEN occurred, to an unknown extent, in the metastatic setting, especially in cell lines. Therefore, we sought to determine whether structural alterations in PTEN occurred with any frequency in noncultured cutaneous melanomas, especially in the metastatic setting, if loss of PTEN expression, detected by immunohistochemistry, is a mechanism of loss of function in melanomas, and if there is a correlation between PTEN protein expression levels and genomic structural alterations of this gene.
Materials and Methods
Melanoma Samples
Thirty-four unselected sporadic primary or metastatic melanoma samples were obtained from cases undergoing surgery in the Division of Surgical Oncology, James Cancer Hospital and Solove Research Institute, Ohio State University Comprehensive Cancer Center (Columbus, OH) during the period from 1997 to 1999, in an anonymized fashion in accordance with an IRB-approved protocol. Of these 34 samples, 4 were primary cutaneous melanomas and 30 were metastatic to various sites, including lymph nodes and other soft tissues. Tumor tissues were snap-frozen and kept at −80°C until nucleic acid extraction. Normal tissue from unaffected surrounding fat or muscle was obtained from 28 of these patients during the surgical procedure and served as a source of germline DNA. DNA was extracted from frozen normal and tumor tissue using Qiagen DNA-Mini kit according to the manufacturer’s instructions (Qiagen, Valencia, CA).
Paraffin-embedded tissue blocks were available for all 34 melanomas. Four-micrometer sections were cut and mounted on Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA) for immunohistochemistry studies.
Immunohistochemistry
The specificity of PTEN monoclonal antibody 6H2.1 has been proven previously. 33,34 This antibody, raised against the last 100 C-terminal amino acids of human PTEN, was used essentially as previously described 33 with minor modifications. In brief, the sections were deparaffinized and hydrated by passing through xylene and a graded series of ethanol. Antigen retrieval was performed for 20 minutes at 98°C in 0.01 mol/L sodium citrate buffer, pH 6.4, in a microwave oven and incubating the sections in 0.3% hydrogen peroxide. After blocking for 30 minutes in 0.75% normal horse serum, the sections were incubated with 6H2.1 (dilution 1:100) overnight (or 16 hours) at 4°C. The sections were washed in PBS, pH 7.3, and then incubated with biotinylated horse anti-mouse IgG followed by avidin peroxidase using the Vectastain ABC elite kit (Vector Laboratories, Burlingame, CA). The chromogenic reaction was carried out with 3′−3′ diaminobenzidine using nickel cobalt amplification, which gives a black product. For sections with abundant melanin, 3′−3′−3′ triaminobenzidine was used, which produces a blue product. After counterstaining with nuclear Fast Red (Rowley Biochemical, Danvers, MA) and mounting, the slides were evaluated under a light microscope. The immunostaining patterns and intensities were determined by two independent observers (X.-P. Z. and C. E.) who each examined and independently scored the slides on two separate occasions. As previously described, 33,34 the vascular endothelium serves as an internal positive control, and the immunostaining of the endothelium is scored as ++. Levels of immunostaining in vascular endothelium are remarkably constant among various tissues, including breast, 33 thyroid, 34 pancreas (Perren A, Eng C, unpublished results) and colon (Zhou XP, Eng C, unpublished results). Immunostaining intensities equal to the vascular endothelium in a particular sample were scored as ++; weak or decreased staining intensity as +; and no immunostaining as −. Staining intensities greater than those of the internal control vascular endothelium were graded +++.
LOH Analysis
All 28 of the melanoma samples with matching germline DNA available were analyzed for LOH with markers D10S579, D10S2491, and D10S541, the former two of which are within 500 kb upstream of the transcriptional start site of PTEN and the latter within 300 kb of the translational stop site. In addition, LOH analysis was performed using intragenic intronic polymorphic markers IVS4 + 109 ins/del TCTTA and IVS8 + 32T/G within PTEN. Assessment of the status at IVS8 + 32T/G was performed by differential digestion with the restriction endonuclease HincII as described. 35 The status at IVS4 + 109 ins/delTCTTA was screened by PCR-based differential digestion with restriction endonuclease AflII following the manufacturer’s instructions (New England Biolabs, Beverly, MA). The primers used to amplify a 325-bp fragment containing exon 4 and this polymorphic site within intron 4 are E4-F (forward, 5′-CATTATAAAGATTCAGGCAAT-3′) and MMAC4-R (reverse, 5′-CTTTATGCAATACTTTTTCCTA-3′).
Mutation Analysis
PTEN mutation analysis of all 9 coding exons, exon-intron junctions, and flanking intronic sequences was performed using PCR-based denaturing gradient gel electrophoresis and semi-automated sequencing as previously described. 12,14,36
Results
PTEN Immunohistochemistry in Primary and Metastatic Melanomas
The expression of PTEN in 30 metastatic melanomas and 4 primary melanomas was evaluated by immunohistochemical analysis. All 34 melanoma sections had accompanying vascular endothelial cells present, which showed strong PTEN immunostaining in the cytoplasm and the nucleus, were graded ++, and served as internal positive controls as described previously. 33,34 Interestingly, the endothelial cells showed strong (++) PTEN immunostaining with a nuclear predominance (Figure 1) ▶ . Nuclear and cytoplasmic staining intensity of fibrocytes varied from very strong (+++) to weak (+).
PTEN immunohistochemistry in melanomas. Positive staining (++) of vascular endothelial cells serves as an internal positive control (arrows). A: Melanoma exhibiting positive staining (++) for PTEN in all tumor cells. B: Melanoma exhibiting weak staining (+) in majority of tumor cells. C: Melanoma with negative staining (−) for PTEN in all tumor cells. D: Melanoma exhibiting absent staining for PTEN in tumor cells with abundant melanin. Original magnifications, ×20.
Overall, among all 34 melanomas, the neoplastic nuclei had weak (+) or no (−) PTEN immunostaining compared to the nuclear staining (++) of their respective vascular endothelial cells. Nearly two-thirds of the melanomas, 22 of 34 (65%), had weak (+) or absent (−) cytoplasmic PTEN staining (Figure 1, B and C ▶ ; Table 1 ▶ ). Five (15%) melanomas lost all PTEN immunoreactivity (−) and 17 (50%) showed weak (+) cytoplasmic PTEN immunostaining. In general, the staining intensity or lack thereof within a sample was relatively uniform.
Table 1.
Summary of PTEN Immunostaining and LOH Data
PTEN staining | |||
---|---|---|---|
++ | + | − | |
LOH | 2 | 6 | 1 |
ROH | 8 | 7 | 4 |
Total | 10 | 13 | 5 |
LOH, loss of heterozygosity; ROH, retention of heterozygosity.
The remaining one-third of melanomas (12 of 34; 35%) showed moderate (++) cytoplasmic immunostaining and weak (+) nuclear staining compared to that of their respective vascular endothelial cells (Figure 1A ▶ ; Table 1 ▶ ). However, two distinct staining patterns were noted among this group of 12 melanomas. Of these 12, 8 had uniform immunostaining of all tumor cells. The remaining 4 from this group had non-uniform staining of the tumor cells.
There were 8 tumors (7 metastatic; one primary) with abundant melanin in the cytoplasm of some or most tumor cells. Interestingly, most of the tumor cells with abundant melanin in their cytoplasm had weak (+) or no (−) immunohistochemical evidence of PTEN expression (Figure 1D) ▶ . Of these 8 tumors with abundant melanin, 5 had weak immunostaining and 3 had no PTEN expression.
LOH and PTEN Mutation Analysis
PCR-based denaturing gradient gel electrophoresis revealed no intragenic PTEN mutations in any of the 34 melanomas (data not shown).
LOH analysis was performed on the 28 melanomas with corresponding germline DNA available using three microsatellite markers flanking PTEN (D10S579, D10S2491, and D10S541), and two polymorphic markers within PTEN (IVS4 + 109ins/delTCTTA and IVS8 + 32G/T). All 28 pairs were informative at a minimum of one of these markers and 26 of these pairs were informative at 2 or more loci (Figure 2) ▶ . LOH at 10q23 was scored when one or more of the panel of 5 polymorphic loci showed LOH. Using these criteria, 10q23 LOH was seen in 9 tumors (32%; Figure 2 ▶ ). However, if the criteria were made more stringent, ie, LOH was scored only if 1 of the 2 intragenic PTEN polymorphic markers (IVS4 + 109ins/delTCTTA and/or IVS8 + 32G/T) showed LOH, then we found evidence for PTEN hemizygous deletion in only 5 tumors (18% of total, 31% of 16 cases informative for at least 1 of the 2 intragenic loci).
Comparison of Immunohistochemical and Structural Mutation Data
There were a total of 28 paired samples where both immunohistochemical data and LOH data are available. All but one were informative at 2 or more polymorphic loci within or flanking PTEN (Figure 2) ▶ . Ten of the 28 samples were graded ++ for PTEN expressioin by immunohistochemistry, and 6 of these 10 did not show evidence of 10q23 or PTEN deletion. Two samples (Figure 2 ▶ , samples 2 and 9, and Table 1 ▶ ) might be viewed as apparently discordant when LOH data are compared to those obtained from immunohistochemistry. Sample 2 showed LOH at D10S541 and retention of both alleles at the remaining 4 loci, notably those within PTEN. Thus, although this sample had LOH 3′ of PTEN, it is doubtful if it is deleted within PTEN. Hence, these data are concordant. Sample 9, on the other hand, showed LOH at all loci except for D10S541 where it is not informative (Figure 2) ▶ . This melanoma showed a patchy, non-uniform staining pattern with weak (+) staining in some regions of the tumor section, although strong (++) staining was predominant.
Five melanomas were immunostain negative (−). Among these, 4 had no evidence of PTEN allele loss and 1 showed LOH involving 4 of the 5 markers (Figure 2) ▶ .
Thirteen melanomas were immunostained weakly (+) and were informative at at least one polymorphic locus (Figure 2) ▶ . Of these 13, 6 might be classified as having LOH representing hemizygous PTEN allelic loss that could correspond to the diminished immunostaining (Figure 2 ▶ , samples 5, 8, 16, 22, 26, and 27). The remaining 7 with weak immunoreactivity retained heterozygosity at their respective informative loci.
In summary, the presence of LOH always correlated with decreased PTEN expression. In contrast, neither the stringent nor more relaxed criteria for scoring LOH in PTEN deletion revealed a correlation between the absence of allelic losses and PTEN expression.
Correlation between PTEN Immunohistochemistry and Clinicopathological Parameters
With a relatively small series comprising 30 metastatic tumors and 4 primary melanomas, we could not perform a rigorous statistical analysis of PTEN immunohistochemistry and clinicopathological features. However, it might be worth noting that all of the 5 melanomas with complete loss of PTEN expression (−) were metastatic melanomas. Among the only 4 primary tumors, 2 had hemizygous PTEN deletion and showed weak PTEN expression (+), whereas the remaining 2 had strong immunostaining (++).
Discussion
That loss of heterozygosity of markers along the long arm of chromosome 10 occurs in primary and metastatic melanomas is beyond doubt, however, to what extent PTEN, on 10q23.3, plays a role was questionable based on previous studies using mainly melanoma cell lines and some noncultured primary and metastatic melanomas. 2,26-32 In this study, we have found that somatic intragenic mutations of PTEN do not occur to a significant degree in non-cell line primary and metastatic melanomas. Power calculations suggest that if PTEN mutations occur at a frequency of 10% or more, the sample size of 34 used in this study should have a >95% likelihood of detecting at least one such mutation. Given that no mutations were noted, even in metastatic tumors, it might be reasonable to claim that intragenic PTEN mutation does not occur in more than 10% of melanomas and hence, is not considered a major mechanism of inactivation in this tumor type, at least in the midwestern United States population.
A relatively large study based on European melanoma samples 32 and a small United States study 28 have shown biallelic structural inactivation occurs by either homozygous deletion at 10q23 or somatic intragenic PTEN mutation plus loss of the remaining wild-type allele. In contrast, we have found that biallelic structural alteration does not occur in melanomas, even in the metastatic setting. Instead, even in the primary setting, hemizygous deletion at 10q23, encompassing PTEN, can occur with some frequency (50% of a small number) as an early event. In the metastatic setting, partial or complete expressional loss at the protein level was shown to occur in two-thirds of the melanomas, 75% of which are due to epigenetic silencing of PTEN (Figure 2) ▶ . Indeed, of the 5 metastatic tumors with no PTEN expression, 4 had complete silencing of both PTEN alleles via mechanisms beyond structural alteration and one experienced loss of one allele and silencing of the remaining wild-type allele by epigenetic means (Table 1 ▶ and Figure 2 ▶ ). Similarly, among the 13 tumors (2 primary) with weak (+) PTEN immunostaining, 6 (2 primary) had hemizygous allelic loss that could account for the decreased protein expression. However, the remaining 7 had neither PTEN mutation nor deletion. Thus, these 7 might have monoallelic silencing via mechanisms other than structural alteration (mutation/deletion) of PTEN, which would also account for the decreased expression. All 10 melanomas with strong immunostaining (++, Figure 2 ▶ ) have intact biallelic PTEN structure, although tumor 9 is worthy of comment. Tumor 9 has LOH involving all of PTEN. However, immunostaining is patchy: the majority of the tumor in the section had strong (++) immunostaining, but there were small areas of weak (+) staining. It is possible that the portion of this tumor used for LOH analysis is reflected by the areas of weak staining.
Epigenetic mechanisms of inactivation of PTEN were initially postulated for a subset of prostate cancer lines. 37 In this instance, hypermethylation of the promoter was shown to be the mechanism. Subsequently, epigenetic PTEN silencing was shown to be a major mechanism in hematological malignancy cell lines, where approximately 30 to 40% of these lines had a genomic alteration (deletion or mutation), 50% had no transcript, and up to 70% had no PTEN protein. 7 Thus, in these cell lines, transcriptional silencing, likely secondary to promoter methylation, as well as translational and post-translational mechanisms are pertinent.
High levels of PTEN expression, as demonstrated by immunohistochemistry, have been found in the developing human neural crest (graded +++), which is the precursor of melanocytes, and to a lesser extent (++) melanoblasts and normal adult melanocytes 34 (Zhou X-P, Eng C, unpublished findings). PTEN plays a role in G1 arrest and/or apoptosis (see Introduction). Conceivably, absent or decreased levels of such a molecule in melanocytes could lead to inability to cell cycle arrest and inability to undergo programmed cell death, thereby leading to melanoma formation. If PTEN′s role in cell adhesion and migration can be confirmed, then absent or decreased PTEN could also lead to increased ability to metastasize.
With the accumulating knowledge of PTEN inactivation, it would appear that there are several mechanisms which lead to PTEN inactivation, and these might operate in a tissue-specific manner. Initial tumor cell line work almost uniformly suggested that PTEN intragenic mutations, homozygous deletions, and the two structural hits would be the rule across a large variety of tumors. 1,2,27 However, in retrospect, it would appear that these sorts of PTEN defects, which result in inability to arrest at G1 and/or inability to undergo apoptosis, are selected for in cell lines. Among noncultured neoplasias, in contrast, the high frequency of PTEN intragenic mutations and homozygous deletions have not been found in all tumor types, as suggested by early cell line work. Glioblastoma multiforme and endometrial adenocarcinomas seem to have a high frequency of intragenic PTEN mutations and deletions, such that in these two tumor types, the two structural hits can be observed. 17-20 Primary breast carcinomas have a relatively low (≪5%) frequency of somatic intragenic PTEN mutation and a hemizygous deletion frequency of 30 to 40%. 38-40 However, we have shown that the second hit in primary breast carcinomas can be epigenetic as well. 33 Early work on sporadic noncultured nonmedullary thyroid tumors revealed a 10 to 25% LOH frequency, without any homozygous deletions, and a somatic frameshift mutation within PTEN in a single papillary thyroid carcinoma. 35,41 Recently, we have shown that although an epigenetic second hit can occur to functionally silence PTEN protein expression in thyroid tumors, differential subcellular compartmentalization between nucleus and cytoplasm might represent a novel mechanism of functional inactivation or modulation in nonmedullary thyroid cancers. 34 Noncultured malignant melanomas might be somewhat unique: though PTEN inactivation is seen to occur by one structural hit (LOH) followed by the second epigenetic hit, PTEN protein expression in melanomas can be biallelically silenced by epigenetic phenomena as well. Although numbers are relatively small, our study suggests that the latter might be a major mechanism of PTEN inactivation in the pathogenesis of melanomas.
In summary, we have found weak or absent PTEN protein expression with relatively high frequency in malignant melanomas. Our data not only strongly support PTEN as a major tumor suppressor on 10q involved in melanoma tumorgenesis, but also suggest a unique mechanism of biallelic functional inactivation not previously observed in other cancers where PTEN might be involved.
Acknowledgments
We thank Jacqueline A. Lees for providing the antibody 6H2.1.
Footnotes
Address reprint requests to Charis Eng, Human Cancer Genetics Program, The Ohio State University, 420 W. 12th Avenue, Suite 690 MRF, Columbus, OH 43210. E-mail: .ude.uso.rtcdem@1-gne
Supported in part by P30CA16058 from the National Cancer Institute, Bethesda, MD (as a seed grant from the Ohio State University Comprehensive Cancer Center to M. J. W. and C. E.) and the American Cancer Society (RPG98-211-01-CCE to C. E.).