Germline epigenetic regulation of KILLIN in Cowden and Cowden-like syndrome

Abstract

Context: Germline loss-of-function phosphatase and tensin homolog gene (PTEN) mutations cause 80% of Cowden syndrome, a rare autosomal-dominant disorder (1 in 200,000 live births), characterized by high risks of breast, thyroid, and other cancers. A large heterogeneous group of individuals with Cowden-like syndrome, who have various combinations of Cowden syndrome features but who do not meet Cowden syndrome diagnostic criteria, have PTEN mutations less than 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 underexpression of PTEN or of KILLIN, a novel tumor suppressor transcribed in the opposite direction, may account for the remainder of Cowden syndrome and Cowden-like syndrome.

Objective: To determine whether germline methylation is found in Cowden syndrome or Cowden-like syndrome in individuals lacking germline PTEN mutations.

Design, setting, and participants: Nucleic acids from prospective nested series of 123 patients with Cowden syndrome or Cowden-like syndrome and 50 unaffected individuals without PTEN variants were analyzed for germline methylation and expression of PTEN and KILLIN at the Cleveland Clinic, August 2008-June 2010. Prevalence of component cancers between groups was compared using the Fisher exact test.

Main outcome measures: Frequency of germline methylation in PTEN mutation-negative Cowden syndrome and Cowden syndrome-like individuals. Prevalence of component cancers in methylation-positive and PTEN mutation-positive individuals.

Results: Of 123 patients with Cowden syndrome or Cowden-like syndrome, 45 (37%; 95% confidence interval [CI], 29%-45%) showed hypermethylation upstream of PTEN but no transcriptional repression. The germline methylation was found to transcriptionally down-regulate KILLIN by 250-fold (95% CI, 45-14 286; P = .007) and exclusively disrupted TP53 activation of KILLIN by 30% (95% CI, 7%-45%; P = .008). Demethylation treatment increased only KILLIN expression 4.88-fold (95% CI, 1.4-18.1; P = .05). Individuals with KILLIN -promoter methylation had a 3-fold increased prevalence of breast cancer (35/42 vs 24/64; P < .0001) and a greater than 2-fold increase of kidney cancer (4/45 vs 6/155; P = .004) over individuals with germline PTEN mutations.

Conclusions: Germline KILLIN methylation is common among patients with Cowden syndrome or Cowden-like syndrome and is associated with increased risks of breast and renal cancer over PTEN mutation-positive individuals. These observations need to be replicated.

Figure 1. Germline DNA methylation of PTEN and KILLIN in Cowden and Cowden-like syndrome patients

a) A schematic of the genomic structure of the PTEN and KILLIN genes in relationship to one another at 10q23. The region analyzed for DNA methylation is depicted by the orange bars, with the numbers showing the location of the bisulfite PCR product with respect to the translation start site of each gene. As depicted, the KILLIN promoter overlaps with the 5’UTR and coding region of PTEN . b) Example of Combined Bisulfite Restriction Analysis (COBRA) of DNA from a subset of controls and CS/CSL patients (numbers refer to patient IDs in eTable 1). The 0% and 100% are from peripheral blood DNA, the 100% having been in vitro methylated with SssI methylase. These serve as negative and positive controls for methylation pattern analysis of the patients. An increase in the intensity of smaller, digested bands compared to the adjacent normal indicate increased methylation in the tumor DNA. * = methylation in that patient sample. c) Results from bisulfite sequencing analysis of 8 of the patient samples. Each circle represents a CpG dinucleotide (black = methylated, white = unmethylated), with the percentage filled reflecting the percentage of methylation seen at that CpG. “M”=digested bands representing methylated DNA, “UM”=undigested PCR product representing unmethylated DNA.

Figure 2. Quantitative mRNA analysis of PTEN and KILLIN expression in 8 CS/CSL patients with germline methylation

qRT-PCR analysis of 4 controls and 8 patient samples. All samples were first normalized to their own internal control (GAPDH). The average of the controls, set to 1, was used for normalization for all samples. The top panel displays the expression for PTEN, which reveals significantly increased expression in three patient samples compared to the normals, while only one sample showed significantly decreased expression. P=0.013 for samples 21 and 397, P<0.0001 for samples 446 and 1350. The bottom panel reveals significantly decreased KILLIN expression in all patient samples analyzed. P=0.007 for sample 366, P<0.0001 for samples 21, 31, 32, 40, 397, 446, and 1350.

Figure 3. Quantitative mRNA analysis of PTEN and KILLIN expression in 8 CS/CSL patients with germline methylation, with and without demethylation and histone deacetlyase inhibition treatment

Eight patient cell lines with germline methylation in the region analyzed were subjected to 5-aza-2’-deoxycytidine for 96 hours at 0.5mM concentration and/or 200nM Trichostatin A. Quantitative RT-PCR analysis was performed on the cDNA from cells with and without drug treatment to detect changes in expression from the demethylation and histone deacetylase inhibition treatment. All values were first normalized to their internal control (GAPDH). The fold-increase or decrease in expression in the drug treated samples is derived by normalizing to its untreated counterpart, which was set as 1. PTEN expression is shown on the top panel and reveals a significant decrease in PTEN expression following demethylation in all but one cell line. * P=0.018, ** P=0.001, *** P<0.0001. Patient #397 that showed an increase in PTEN expression following demethylation treatment alone was not significant (P=0.42). The bottom panel shows KILLIN expression following demethylation and/or inhibition of histone deacetylation, which shows a significant increase in expression in 7/8 of the cell lines. *P=0.050, ** P=0.029, *** P=0.018, ****P=0.006, *****P<6.84 × 10⁻⁶.

Figure 4. Methylation selectively inhibits p53 binding and transcriptional activation of KILLIN while not affecting p53 binding for PTEN

a) Model depicts where the observed methylation (closed, small circles) resides with respect to both PTEN and KILLIN and shows the two regions where p53 binds for PTEN and is blocked from binding (swooped arrow) for KILLIN transcriptional activation. “PTEN site” and “KILLIN site” represent the PCR products amplified from the eluate to analyze for p53 binding following ChIP analysis. The thinner pink and blue boxes indicate the transcribed region that is not translated, whereas the wider pink and blue boxes indicate the translation start site for the given gene. b) Chromatin immunoprecipitation analysis of p53 pulldown of either KILLIN’s or PTEN’s p53 binding element in controls and Cowden syndrome patients. All samples are normalized to their negative control, IgG. Varied enrichment of the KILLIN and PTEN p53 binding sites was observed in the control samples, whereas a significantly greater amount of Region 1 (PTEN’s p53 binding site) was pulled down in 75% of the patient cell lines. * P=0.032, ** P=0.012, *** P=0.002. c) In vitro methylation with SssI methylase was performed for both the PTEN and KILLIN luciferase promoter constructs. The constructs contained the same promoter sequence (either in orientation for KILLIN or in the opposite orientation for PTEN), which includes -1 to -1344 of sequence upstream of the translation start site of PTEN (-745 to +600 in respect to the KILLIN translation start site). Luciferase promoter analysis of PTEN and KILLIN promoter activity was done using the MDA-MB-453 breast cancer cells in the absence (“WT”) or presence (“+ p53”) of p53 transfection. All values were first normalized to their internal control, Renilla luciferase. The fold increase in the samples with p53 overexpression was attained by normalization to those without p53 transfection, which was set as 1. The PTEN constructs showed significant activation by p53 regardless of methylation status (* P=0.012). However, the KILLIN methylated construct showed significantly less activation by p53 compared to the unmethylated KILLIN luciferase construct (** P=0.008).