|Home | About | Journals | Submit | Contact Us | Français|
Germline loss-of-function PTEN mutations cause 80% of Cowden syndrome (CS), an autosomal dominant disorder characterized by high risks of breast, thyroid, and other cancers. A large heterogenous group of CS-like (CSL) individuals, who have various combinations of CS features but who do not meet CS diagnostic criteria, have PTEN mutations <10% of the time, making molecular-diagnosis, prediction, genetic-counseling and risk-management challenging. Other mechanisms of loss-of-function such as hypermethylation, which should result in under-expression of PTEN, or of KILLIN, a novel tumor suppressor transcribed in the opposite direction, may account for the remainder of CS/CSL individuals. Screening for such promoter methylation, may improve the sensitivity of molecular diagnosis.
To determine whether germline methylation is found in CS/CSL individuals lacking germline PTEN mutations.
Nucleic-acids from prospective nested series of 123 CS/CSL patients and 50 unaffected individuals without PTEN germline variants analyzed for germline methylation and expression of PTEN and KILLIN (Cleveland Clinic, 2008/8–2010/6). Prevalence of component cancers between groups was compared with Fisher’s exact-test.
Frequency of germline methylation in PTEN mutation negative CS/CS-like individuals. Prevalence of component cancers in methylation-positive and PTEN mutation-positive patients.
Of 123 CS/CS-like patients, 45 (37%, 95%CI 29–45%) showed hypermethylation upstream of PTEN but no transcriptional repression. The germline methylation was found to transcriptionally downregulate KILLIN 250-fold (95%CI 45–14286, P=0.007); and exclusively disrupted p53-activation of KILLIN by 30% (95%CI 7–45% (P=0.008). Demethylation treatment increased only KILLIN expression 4.88-fold (95%CI 1.4–18.1, P<0.05). Individuals with KILLIN-promoter methylation had 3-fold increased prevalence of breast cancer (35/42 vs 24/64, P<0.0001) and a >2-fold increase of kidney cancer (4/45 vs 6/155, P=0.004) over those with germline PTEN mutations.
Germline KILLIN- methylation is common among CS/CSL patients, and associated with increased risks of breast and renal cancer over PTEN mutation-positive patients. These observations need to be replicated.
Germline mutations of PTEN (UCSCuc001kfb.2; RefSeqNM000314), encoding Phosphatase and TENsin homologue deleted on chromosome 101, cause 80% of autosomal-dominant Cowden syndrome (CS), and occurs 1:200,000 live-births. It is characterized by macrocephaly and typical mucocutaneous features (trichilemmomas, papillomatous papules), and hamartomas, with increased risk of various malignancies, ~10% lifetime risk for thyroid cancer and up to 50% lifetime risk for female breast cancer over the general population.1–6,7,8 However, only 5% of this heterogeneous group referred to as CS-like (CSL) individuals, who have some features of CS but do not meet diagnostic criteria, have germline PTEN mutations. In the absence of germline PTEN mutations, approximately 10% CS/CSL individuals harbor germline SDHB (UCSCuc001bae.2; RefSeq NM003000) and SDHD (UCSCuc001pmz.2; RefSeqNM 003002) variants.9 Overall, germline PTEN mutations/deletions and SDHx variants account for 82–85% of CS and 6–11% of individuals with CS-like phenotypic features.
CS is a great clinical mimic and is difficult to recognize because every patient shows variable expression and penetrance. Importantly, many individuals in the general population share one or a few features of CS but may not have CS and may not even harbor alterations in any predisposition genes. Many such patients present to primary care and other specialty clinicians who are called upon to recognize such individuals because individuals with specific gene mutations have increased risks of different spectra of neoplasias. In the context of a difficult-to-recognize syndrome, identification of additional cancer predisposition genes would facilitate molecular diagnosis, genotype-specific predictive testing of family members who are as yet clinically unaffected, genetic counseling, and medical management. Relevant to primary care, once a mutation or alteration is found, primary care physicians must have a basic understanding of gene-specific cancer risks as they do play and will increase their role as the coordinator of gene-specific personalized management, surveillance and so on.
PTEN is a well-characterized tumor suppressor phosphatase involved in cellular regulation10 via G1 cell cycle arrest and apoptosis.11 Interestingly, a novel gene, KILLIN (UCSCuc009xti.2; RefSeqNM 001126049), also residing in the 10q23.31 chromosomal region is involved in cell cycle arrest and is regulated by p53 (UCSCuc002gig.1; RefSeqNM 000546), similar to PTEN.12 PTEN and KILLIN share the same transcription start site but are transcribed in opposite directions. KILLIN has been shown to be necessary and sufficient for p53-induced apoptosis.12 This high-affinity DNA-binding protein inhibits eukaryotic DNA synthesis in vitro and causes S phase arrest before apoptosis in vivo.12 Because of similar function to PTEN, we investigated KILLIN as a predisposition gene in CS/CSL patients.
Epigenetic alterations play an important role in cancer progression through hypermethylation and silencing of tumor suppressor genes,13,14 and somatic PTEN hypermethylation has been recognized as a means of PTEN downregulation in a subset of malignancies. We sought to address the hypothesis that germline methylation of the 10q23.31 bidirectional promoter CpG island (a region of at least 200bp with a GC content of ≥50% and an observed/expected CpG ratio of >60%) silences PTEN, KILLIN, or both. This, consequently, would account for patients with CS or CS-like features but without germline PTEN mutations/deletions.
Between October 2005 and December 2009, 2000 CS or CS-like patients were prospectively enrolled mainly regionall and nationally by the Cleveland Clinic Genomic Medicine Institute in accordance with research protocol IRB8458-PTEN, approved by the respective Institutional Review Boards for Human Subjects Protection. All research participants provided written informed consent. To be enrolled in the IRB8458-PTEN, individuals are eligible if he/she meets the full CS diagnostic criteria established by the International Cowden Consortium (i.e. major criteria includes breast cancer, thyroid cancer, macrocephaly, endometrial carcinoma, Lhermitte-Duclos disease) according to version 2000 (Supplemental Table 1).18 Patients meeting the relaxed criteria are referred to as individuals with CS-like phenotypes or CSL. Of the 2,000 prospectively enrolled subjects meeting the criteria for protocol 8458-PTEN, <400 lacked germline PTEN pathogenic mutations, large deletions, variants of unknown significance, and polymorphisms by sequencing analysis of all 9 exons and the promoter. Of these 400, we selected a nested series of the most recent 123 participants who also were found not to have SDHB-D variation, regardless of family history status, comprising 48 CS, 75 CSL, and 50 unaffected individuals (population controls resident in the region), for the purposes of this study. Sample sizes were selected to ensure power (P>0.9) to detect a 5% prevalence of the methylation, as well as to detect a three-fold difference between cases and controls. All specimens from study participants and controls were prepared and analyzed within the Genomic Medicine Institute. The majority of the subjects were isolated cases, with the exception of 3 individuals each of whom had at least one family member who also agreed to be part of our study. All analyses were performed from August 2008 through June 2010.
The study population included 48 CS patients, 75 CSL patients, and 50 unaffected individuals as controls. The Combined Bisulfite Restriction Analysis (COBRA)19 and the bisulfite sequencing were performed as previously described.20 The bisulfite PCR primer sequences are shown in Supplemental Table 2. In order to provide a comprehensive analysis of the methylation status across the CpG islands upstream of PTEN, we screened four different regions (+400bp to +700bp; −188bp to −477bp; −425bp to −640bp; −806bp to −1043bp, all with respect to the PTEN translation start site).
The patient and control lymphoblastoid cell lines used in this study were generated from peripheral blood samples by the Genomic Medicine Biorepository (see http://www.lerner.ccf.org/gmi/gmb/methods.php) of the Cleveland Clinic Genomic Medicine Institute.
Promoter luciferase assay was performed in order to validate transcriptional repression by DNA promoter methylation and differential inhibition of p53 binding. Breast cancer cell line MDA-MB-453 (ATCC) was used for the luciferase assay. Lymphoblastoid cell lines and MDA-MB-453 were maintained in RPMI with 10% FBS and 2% antibiotics. The antibody used in the ChIP experiment (see below) was mouse monoclonal p53 (Santa Cruz; sc-126). The in vitro methylated constructs used for the luciferase assay were generated by first digesting 90µg of the original PTEN and KILLIN promoter constructs (containing 1 to 1344bp upstream of the PTEN translational start site cloned in either direction) with BglII (NEB) and BbvCI (NEB). The linearized, digested inserts and vectors were gel extracted. The insert DNAs, which contain the sequence that is methylated in vivo in CS/CSL patients, were then methylated with CpG SssI methylase (NEB) for 4 hours. Following in vitro methylation, the insert was re-ligated with its corresponding vector using a 3:1 insert to vector ratio with 2µg total DNA. For comparison, the unmethylated counterpart was digested and re-ligated in parallel.
ChIP analysis was performed in order to validate that p53 binding is differentially affected by DNA methylation and performed as previously described,19 according to the Upstate Cell Signaling Solutions protocol. ChIP analysis utilized 2 controls and 4 patients (patients 21, 31, 32, and 40 from eTable 1) that were selected based on methylation status, representation of CS and CSL, and similar levels of KILLIN mRNA downreglation. Sequences of the primers used for the quantitative ChIP PCRs can be found in Supplemental Table 2.
Luciferase assays were performed as previously described using MDA-MB453 cells.21
The quantitative RT-PCRs were performed as previously described.19 The study population included 4 controls and 8 patients (patients 21, 31, 32, 40, 366, 397, 446, and 1350 from eTable 1) that were selected based on confirmed methylation status by bisulfite sequencing analysis and representation of CS and CSL condition.
The study population included 8 patients (patients 21, 31, 32, 40, 366, 397, 446, and 1350 from eTable 1) that were selected based on confirmed methylation status by bisulfite sequencing analysis and representation of CS and CSL condition. Demethylation treatment was performed with a cytosine analog, 5-aza-2’deoxycytidine (5-aza; Sigma), for 96 hours at 0.5 µM concentration with ~40% confluent suspension lymphoblastoid cells. Inhibition of histone deacetylation was performed with 200 nM concentration of Trichostatin A (TSA; Sigma) with ~40% confluent suspension lymphoblastoid cells for 48 hours, with or without 0.5 µM 5-aza. The drug was changed daily, and the cells collected for RNA isolation.
The statistical significance of the results from qRT-PCR and luciferase assays was calculated by unpaired Student’s t test, with P<0.05 being considered statistically significant, using the Microsoft Excel version 12.2.5. The prevalence of component malignancies between KILLIN promoter methylation positive patients and germline pathogenic PTEN mutation positive patients was compared using the Fisher’s 2-tailed exact test with P<0.05 considered to be significant.
We analyzed germline genomic DNA from CS/CSL patients and from population controls for methylation upstream of PTEN using COBRA. Differential germline methylation was detected between 188 and 477bp upstream of the translation start site for PTEN (Figure 1A). All normal controls showed no methylation (Figure 1B). Among the 123 CS/CSL patient samples analyzed, 45 (37%) were hypermethylated compared to all 50 controls (Figure 1B). Twenty of the 48 (42%) classic CS patients without germline PTEN mutations showed germline hypermethylation. Of the 75 PTEN mutation negative CSL patients, 25 (33%) were found to have germline hypermethylation. Bisulfite sequencing analysis confirmed these differences in a set of CS and CSL patients (Figure 1C).
We then investigated whether methylation segregates with disease in family members of a proband with germline methylation. Of the 45 participants with methylation, only one proband (#616, eTable 1) had >1 affected family member and >1 unaffected family member who agreed to enroll in our study. We found germline methylation in 4/6 of the family members, and three of these four had documented CS/CSL features (with one unknown phenotype). The two remaining unaffected family members did not have germline methylation.
Promoter methylation should result in decreased expression of the relevant gene. In order to validate the pathogenic relevance of this methylation, the expression of PTEN was analyzed in 4 control and 8 patient cell lines as proof-of-principle. PTEN expression in the methylated patient samples was surprisingly not decreased, and instead, increased PTEN expression was noted (Figure 2, top panel). The PTEN 5’UTR and coding region analyzed for methylation overlaps with the putative promoter for KILLIN, a newly characterized tumor suppressor gene (Figure 1A).12 Therefore, in order to address our hypothesis that germline methylation upstream of PTEN may, instead, be silencing KILLIN, we then analyzed KILLIN expression in the patient samples that showed germline methylation. In the methylated patient samples tested, significant under-expression of KILLIN was observed compared to the controls (Figure 2, bottom panel; ** P=0.007).
If, in fact, germline methylation downregulates KILLIN expression, then demethylation should restore KILLIN expression. DNA methylation and histone deacetylation of the promoter often work together to achieve gene silencing,22 and histone acetylation has previously been shown to be transcriptionally relevant in the vicinity of the PTEN-KILLIN bidirectional promoter.23 Therefore, we investigated whether reversal of these epigenetic modifications, via demethylation and/or inhibition of histone deacetylation, would restore only KILLIN expression. KILLIN methylated patient lymphoblastoid cell lines were treated with the demethylating drug 5-aza-2’-deoxycytidine and/or the histone deacetylase inhibitor Trichostatin A (TSA). Demethylation and/or inhibition of histone deacetylation led to a significant decrease in PTEN expression for 7 of the 8 (88%) patient cell lines (Figure 3, top panel). In contrast to PTEN, KILLIN expression was restored in 88% (7 of the 8) analyzed patient cell lines following exposure to 5-aza-2’-deoxycytidine and/or TSA (Figure 3, bottom panel).
Because the methylation of the shared bidirectional promoter had a differential impact on transcription for these two genes, we sought to mechanistically explain what might account for the differential epigenetic control. Both genes are transcriptionally regulated by p53, and there appears to be two distinct p53-binding sites - one for KILLIN and the other for PTEN. The p53-binding site for transcriptional activation of PTEN lies outside of our germline methylated region (Figure 4A),27 whereas, the putative p53 binding site for KILLIN lies within the methylated region identified in this study. Therefore, if we are correct that the methylation downregulates only KILLIN expression, then the methylation should exclusively inhibit p53 binding and activation for KILLIN alone, without affecting PTEN transcription (Figure 4A).
One powerful way to interrogate this is by ChIP analysis: if there is no methylation “blocking” the relevant p53-binding sites, then ChIP should reveal the p53-associated regions of DNA by “pulling down” the sites bound by p53 protein via the use of a p53 antibody. Accordingly, we utilized 4 lines from patients (patients #21, 31, 32, and 40, eTable 1) who exhibited germline methylation of the KILLIN promoter and found that in 3, p53 bound more strongly to its PTEN binding site and relatively poorly to its KILLIN binding site, which was blocked by methylation (Figure 4B). As controls, ChIP analysis revealed no difference of p53 binding to both the PTEN and KILLIN p53-binding sites in the unmethylated control cell lines tested (Figure 4B).
To further address whether the differential p53 binding of these two regions is due to methylation seen in the patient samples, we artificially and purposefully methylated the same CpG region in a PTEN or KILLIN promoter construct. By overexpressing p53 in these cells, we observed a significant increase in PTEN promoter activity, without significant differences in the level of activation between the unmethylated and methylated PTEN constructs (Figure 4C). Although both the unmethylated and methylated KILLIN constructs also provided an increase in transcriptional activation with p53 overexpression, the methylated KILLIN construct showed significantly less transcriptional activation by p53 compared to the unmethylated KILLIN construct or to the PTEN constructs (Figure 4C; P=0.008).
We then turned our attention to the prevalence of component malignancies in those with germline KILLIN promoter methylation and those with proven pathogenic germline PTEN mutations. We found a significant association between the KILLIN methylation status and prevalence of female breast cancer. In our 42 women with methylation, 35 had invasive breast cancers compared to 24 of 64 women (from the same IRB8458-PTEN series) with germline PTEN pathogenic mutations (P<0.0001). Renal cell carcinoma was over-represented in the methylation positive subjects over PTEN mutation positive individuals (4/45 vs 6/155, P=0.004). However, no differences in prevalent thyroid cancers or endometrial cancers were found between the two groups (P=0.2 and 0.4, respectively). Among the 12 epithelial thyroid carcinomas in KILLIN methylation positive CS/CSL individuals, 7 are classic papillary thyroid carcinomas (cPTC), compared to the 5 cPTC to 10 FTC/FvPTC ratio seen in PTEN mutation positive individuals.
Individuals with heritable syndromes, such as hereditary nonpolyposis colorectal cancer, that are negative for mutations in the known predisposition genes have rarely been shown to have heritable hypermethylation (also known as epimutation) of the respective promoters of these genes.29, 30 This guided our initial hypothesis that a subset of CS/CSL patients without PTEN mutations would possibly have PTEN promoter hypermethylation. Instead, our alternative hypothesis was proven correct, resulting in our uncovering a novel CS/CSL predisposition gene, KILLIN, and a new mechanism of epimutation that contributes to the pathogenesis of CS/CSL in those without germline PTEN mutations. The bidirectional promoter is affected by the distinct mechanism of exclusive disruption of p53 binding and activation of KILLIN, while p53 regulation of PTEN (latter is outside of the methylated region) remains unaffected.
The germline KILLIN promoter epigenetic modification mechanism described here accounts for 1/3 of germline PTEN mutation negative CS and of those whose phenotypic features resemble CS, prominently those with breast and thyroid disease. In our current series, >40% of PTEN mutation negative classic CS and 33% of mutation negative CSL patients have germline epigenetic inactivation of the KILLIN promoter. If these data can be, and must be, replicated independently, then a hypothetical schema for prioritizing gene testing could be as follows. Individuals with classic CS should be offered PTEN testing first; those found not to have germline PTEN mutations should then be offered KILLIN epigenetic analysis, in the setting of genetic counseling. Finally, classic CS individuals without germline PTEN mutation (80% are mutation positive) and without KILLIN epigenetic inactivation (half of the 20% should have KILLIN epigenetic inactivation) should then be offered SDHB/D testing (10% of the 20% should have SDHB/D mutation). Altogether, therefore, PTEN, KILLIN, and SDHB/D should then account for 92% of all classic CS. Patients with CS-like features, especially where breast cancer and/or renal carcinomas are present in the individual and/or family, should be offered KILLIN methylation analysis first because it accounts for 30% of such patients compared to PTEN mutations which only accounts for 5–10% of such individuals.
By discovering another cancer predisposition gene, we have added to the sensitivity of molecular diagnosis, and predictive testing becomes possible. Importantly, genetic counseling and gene-informed risk assessment and management become evidence-based. In contrast to germline PTEN mutations, germline methylation of the KILLIN promoter confers a significantly higher prevalence of female invasive breast cancer and renal cell carcinomas. The current national practice guidelines for those with PTEN germline mutations includes heightened surveillance of the female breasts and thyroid, but do not have awareness of renal cancer risk. If our observations of two- to three-fold increased risks of renal and/or breast cancer with KILLIN germline methylation over those of PTEN mutation holds, then extra vigilance for the organs-at-risk, breast and kidneys, is warranted. The KILLIN-associated breast cancer risks would parallel those conferred by germline BRCA1/2 mutations.
Among CS/CSL patients, presence of germline KILLIN gene promoter hypermethylation was common and was associated with increased risk of breast and renal cancer compared with PTEN mutation-positive patients. Two limitations of this study must be considered. First, the relatively small sample size may result in type II error. Second, this study is preliminary in nature and the assumption that KILLIN surveillance will improve CS/CSL diagnosis may be overly optimistic until further validation in a larger patient set can be performed.
We thank the patients who participated in this study and whose contribution made this work possible. We are grateful to members of the Eng lab for thoughtful discussions. CE is the Sondra J. and Stephen R. Hardis Chair of Cancer Genomic Medicine at the Cleveland Clinic, was a Doris Duke Distinguished Clinical Scientist, and is an American Cancer Society Clinical Research Professor, generously funded, in part, by the F.M. Kirby Foundation.
Funding/Support: This study was funded, in part, by the Breast Cancer Research Foundation, the William Randolph Hearst Foundations and P01CA124570 from the National Cancer Institute (all to CE).
Role of Sponsor: The funding organizations and sponsors had no role in the design and conduct of the study, in the collection, management, analysis, and interpretation of the data, or in the preparation of the manuscript. All analytical and interpretative work of the manuscript occurred under the direction of Dr. Eng at the Cleveland Clinic Genomic Medicine Institute.
Author Contributions: CE conceptualized and directed the study. KLB and CE designed the study. KLB acquired and analyzed the data. JM obtained family histories, collected phenotypic documentation, and reviewed all clinical information. KLB and CE interpreted the data and drafted the manuscript. All authors reviewed and critically revised the manuscript, and approved the final manuscript. CE had access to all the data and is responsible for the conduct and content of the study.
Financial Disclosures: None reported.
Supplementary Information: The eTable and Supplemental Tables are available online at http://www.jama.com
No potential conflict of interest relevant to this article was reported.