Loss of PTEN tumor suppressor function is observed in tumors of breast, prostate, thyroid, and endometrial origin. Allelic losses in the proximity of the PTEN locus (10q23) also occur in sporadic colorectal cancers (CRCs), but biallelic inactivation of this site has not been frequently demonstrated. We hypothesized that alternative mechanisms of PTEN allelic inactivation, such as promoter hypermethylation, might be operative in CRC and that PTEN inactivation may be related to recognized forms of genomic instability. We characterized a cohort of 273 sporadic CRCs by determining their microsatellite instability (MSI) status. Of these, 146 cancers were examined for PTEN promoter methylation by methylation-specific PCR. Mutations at the poly(A)6 repeat sequences in PTEN exons 7 and 8 and deletions at the 10q23 locus were also identified using microsatellite analysis. The presence of PTEN protein was determined by immunostaining, and the results were correlated with the promoter methylation status. We observed that PTEN promoter hypermethylation was a frequent occurrence in MSI-high (MSI-H) tumors (19.1% of MSI-H versus 2.2% of MSI-low/microsatellite stable tumors; P = 0.002). A PTEN mutation or a deletion event was present in 60% of the tumors with promoter region hypermethylation. Hypermethylation of the PTEN promoter correlated significantly with either decreased or complete loss of PTEN protein expression (P = 0.004). This is the first demonstration of PTEN inactivation as a result of promoter hypermethylation in MSI-H sporadic CRCs. These data suggest that this silencing mechanism plays a major role in PTEN inactivation and, in colon cancer, may be more important than either allelic losses or inactivating mutations. The significant correlation of PTEN hypermethylation with MSI-H tumors further suggests that PTEN is an additional important “target” of methylation along with the hMLH1 gene in the evolution of MSI-H CRCs and also confers the “second hit” in the biallelic inactivation mechanism for some proportion of tumors.
The novel candidate tumor suppressor PTEN (phosphatase and tensin homolog deleted on chromosome 10) gene (also known as MMAC/TEP1) has been recently identified and mapped to chromosome 10q23.3 (1 , 2) . PTEN is a dual phosphatase, acting at both serine-threonine and tyrosine sites. PTEN inhibits the activation of Akt/protein kinase B by phosphatidylinositol 3,4,5-trisphosphate (3, 4, 5) , thus modulating a major pathway controlling cell proliferation and survival (6) . The COOH-terminal region of the PTEN protein is required for its tumor suppressor function (5) . Mutations in PTEN exons 7, 8, and 9 produce truncated proteins lacking the COOH-terminal region. These truncated products are rapidly degraded, resulting in loss of PTEN protein expression.
Genetic alterations at the PTEN locus have been described in a variety of neoplasms, including tumors of the central nervous system, thyroid, breast, prostate, and bladder and of endometrial origin. Germ-line mutations of PTEN have been detected in 80% of patients with Bannayan-Zonana syndrome, Cowden disease, and juvenile polyposis (7 , 8) , suggesting that PTEN is also an inhibitor of intestinal polyposis. PTEN mutations in sporadic colorectal cancers (CRCs) are uncommon (9 , 10) . The coding region of PTEN contains several repeat sequences, including two poly(A)6 tracts in exons 7 and 8. Recently, mutations at the (A)6 repeat of PTEN were reported in approximately 18% of patients with colorectal tumors showing microsatellite instability (MSI), suggesting that PTEN might be a target of defective mismatch repair function in colorectal carcinogenesis (11 , 12) . Alterations of PTEN in the form of allelic losses have also been observed in some colorectal tumors (13) . An increasing amount of evidence suggests that PTEN may be inactivated by mechanisms other than mutations and/or deletions (14 , 15) . Studies of prostate cancer (15) and leukemia and lymphoma cell lines (14) suggest that epigenetic mechanisms may account for cases in which PTEN expression is down-regulated or even totally ablated in the absence of a detectable mutation. Recently, PTEN promoter methylation was identified as a mechanism of functional loss in some endometrial cancers (16) . We hypothesized that alternative mechanisms of allelic inactivation might also be operative during colon carcinogenesis. We determined the frequency of PTEN promoter hypermethylation in colon cancer and investigated the relationship between this characteristic and tumor MSI. We also examined the extent to which PTEN promoter hypermethylation was either a biallelic event or constituted the “second hit” in a two-hit inactivation mechanism.
MATERIALS AND METHODS
Tumor specimens were obtained from a total of 273 patients with sporadic CRC treated by physicians associated with either the Cancer and Leukemia Group B (CALGB) or the University of California San Diego. Clinical data were available on 267 of these patients. Institutional review board approval was granted for this study. The CALGB cohort with clinical data consisted of 168 individuals who received chemotherapy for colon cancer as part of CALGB Protocol 8896 (Intergroup 0089). These patients underwent surgical resection of an adenocarcinoma of the colon and were determined to have a high risk of tumor recurrence based on regional nodal disease (127 patients, stage III) or local extension of tumor with obstruction or perforation due to tumor (41 patients, stage II). All of the CALGB patients received 5-fluorouracil-based adjuvant chemotherapy. Median clinical follow-up for the CALGB cohort, by the Kaplan-Meier method, is 8.25 years. The University of California San Diego cohort contained 99 individuals with clinical data treated for colon cancer between January 1983 and November 1993. Of these 99 patients, 41 (41.8%) were stage I, 8 (8.2%) were stage II, and 33 (33.7%) were stage III, and 16 (16.3%) were stage IV. Stage could not be determined for one individual. Adjuvant chemotherapy was administered to 20.4% of the University of California San Diego patients, with one stage II (12.5%) and nine stage III (27.3%) patients receiving this treatment. The median clinical follow-up for the University of California San Diego patients is 8.0 years. Available median clinical follow-up for the combined patient population is 8.25 years.
Matching tumor and normal tissue specimens were obtained from all patients from a site distant from the target lesion. All specimens were reviewed by a single pathologist, and the histological type and grade of tumors were classified according to the established criteria. Paraffin-embedded primary tissue and control samples were prepared from H&E-stained 5-μm section slides. Genomic DNA from tumor and matched normal tissues was extracted by microdissection. The microdissected tissues were hydrated, digested in proteinase K, and then processed using Wizard DNA Clean-up System (Promega, Madison, WI) using standard protocols described previously (17) .
Microsatellite analysis was performed on tumor and corresponding normal DNA using a panel of five National Cancer Institute workshop-recommended markers including two mononucleotide (BAT25 and BAT26) and three dinucleotide repeat sequences [D2S123, D5S346, and D17S250 (18)] . PCR was performed using 32P-labeled primers and subsequent electrophoresis on 8% polyacrylamide gels (17) . Tumors exhibiting shifts in electrophoretic mobility at two or more of the five loci analyzed were classified as MSI-high (MSI-H), whereas those showing a shift in one locus were classified as MSI-low (MSI-L). The remaining tumors, lacking MSI events, were determined to be microsatellite stable (MSS). The presence of mutation of the poly(A)10 tract of transforming growth factor β receptor II was investigated by PCR followed by PAGE as described above. The presence of bandshifts or an additional band was interpreted as a mutation.
Bisulfite Modification and Methylation-Specific PCR Assay.
Genomic DNA (2 μg) was denatured with NaOH. Bisulfite treatment, during which methylated DNA is protected, and unmethylated cytosine is converted to uracil, was carried out for 16 h at 50°C on denatured genomic DNA as described previously (19) . DNA samples were then purified using the Wizard DNA Clean-up System (Promega), followed by treatment with NaOH, ethanol precipitation, and resuspension in water.
The modified DNA was used as a template for methylation-specific PCR using primers specific for either the methylated or modified unmethylated sequences (19) . Appropriate negative and positive controls were included in each PCR reaction. Primer sequences used to amplify a 173-bp unmethylated product were 5′-TGGGTTTTGGAGGTTGTTGGT-3′ (sense) and 5′-ACTTAACTCTAAACCACAACCA-3′ (antisense), which amplify a 173-bp product, and primer sequences for the methylated reaction were 5′-GGTTTCGGAGGTCGTCGGC-3′ (sense) and 5′-CAACCGAATAATAACTACTACGACG-3′ (antisense), generating a 155-bp product (20) . Step-down PCR reactions were performed in a 25-μl reaction volume containing 1× PCR buffer (Invitrogen Life Technologies, Inc., Carlsbad, CA), 2.5 mm MgCl2, 200 μm deoxynucleoside triphosphates, 0.5 μm of each PCR primer, 0.75 unit of AmpliTaq polymerase, and approximately 25 ng of bisulfite-modified DNA, as described previously (19) . Reactions were hot-started at 95°C for 5 min. This was followed by 33 cycles at 95°C for 45 s, 57°C for 30 s, and 72°C for 30 s, followed by a 10-min extension at 72°C in a PTC 200 DNA Engine Thermocyler (MJ Research, Inc., Waltham, MA). Human lymphocytic DNA artificially methylated by SssI methylase was used as a positive control for the methylated primer set in each PCR reaction, whereas untreated genomic DNA was used as a positive control for the unmethylated reaction. A water blank was used as a negative control with every PCR amplification. The amplification products were separated on a 3% agarose gel and visualized by ethidium bromide staining and UV transillumination. The results obtained by methylation-specific PCR of PTEN promoter were subsequently confirmed by bisulfite sequencing.
Mutations in PTEN Coding Sequences.
Mutations of PTEN at repetitive sequences within its coding region were identified using a PCR-based assay. A 229-bp region encompassing the PTEN poly(A)6 repeat beginning at nucleotide 795 of exon 7 and a 200-bp region encompassing the poly(A)6 repeat beginning at position 963 of exon 8 were amplified using previously published primers (21) . PCR was carried out for 30 cycles, and each cycle consisted of denaturation for 30 s at 94°C, annealing for 60 s at 58°C, and extension for 45 s at 72°C. PCR products were electrophoresed on 8% polyacrylamide gels, and the shifts in bands were scored for insertion-deletion mutations.
Loss of Heterozygosity (LOH) Determination.
Colon cancers in which both tumor and corresponding normal DNA were available were examined to determine the presence of allelic losses at the 10q23 locus. PTEN lies between D10S1765 and D10S541, a genetic distance of 1 cM, but a physical distance of only several hundred kbp. D10S1765 is within 500 kb upstream of the transcriptional start site, and D10S541 is within 300 kb of the transcriptional stop site. We analyzed the three most commonly used polymorphic markers (D10S541, D10S215, and D10S1765) flanking and within PTEN to study LOH events at the 10q23 locus in all tumors showing PTEN promoter methylation or mutations. Additionally, 10q23 LOH status was determined for 16 patients whose tumors contained unmethylated PTEN. LOH was defined by ≥50% loss of intensity in one or more bands in the tumor DNA compared with the allelic pattern in the normal DNA. All three markers were screened as documented previously (22) . PCR conditions for these markers have been described elsewhere (23 , 24) .
PTEN Expression in Tumor Specimens.
The monoclonal antibody 6H2.1, raised against the last 100 COOH-terminal amino acids of PTEN (Cascade Biosciences, Winchester, MA; Ref. 25 ), was used to identify the presence of PTEN in tumor samples. Staining was performed on 5-μm, formalin-fixed, paraffin-embedded sections on all cases containing a methylated PTEN promoter. Briefly, the sections were deparaffinized and hydrated by passing through xylene and a graded series of ethanol. Microwave antigen retrieval was performed for 20 min at 98°C in 0.01 m sodium citrate buffer (pH 6.4). To block endogenous peroxidase activity, the sections were incubated in 0.3% hydrogen peroxide for 30 min. After blocking for 30 min in 5% normal serum, the sections were incubated with PTEN antibody overnight at 4°C, washed in PBS, and then incubated in a biotinylated second antibody followed by avidin peroxidase using the Vectastain ABC Elite kit (Vector Laboratories, Burlingame, CA). The chromogenic reaction was performed with the Vector NovaRED substrate kit, and sections were counterstained with Lerner’s hematoxylin. The immunostaining patterns and intensities were evaluated independently by two investigators using endometrial stroma and/or normal colonic epithelium as an internal positive control. Cytoplasmic immunostaining intensities for PTEN equal to colonic stroma and/or normal colonic epithelium in a sample were scored as ++; weak or decreased staining intensity was scored as +; and no immunostaining was scored as −, as described previously for colon and other tissues (26) .
Associations between PTEN promoter methylation and protein expression status and clinical characteristics such as tumor stage, lymph node metastasis, and tumor grade were tested using the χ2 test or Fisher’s exact test, as appropriate. Associations with survival were tested using the log-rank test. Survival was measured from the date of diagnosis until death due to colon cancer (cause-specific death). Patients who died of other causes were censored at the date of death. The Kaplan-Meier method was used to estimate survival curves. Data were analyzed using SAS and S-Plus statistical software.
MSI in Sporadic CRCs.
Using a panel of five National Cancer Institute workshop-recommended markers, 267 sporadic CRCs were examined for the presence of MSI (18) . Informative results, defined as successful amplification of at least four of the five markers, were obtained for all of the samples studied. Of the 267 CRCs examined, 42 (15.7%) harbored altered allelic lengths in two or more markers, which were defined as being positive for MSI-H. As expected, a majority of these tumors revealed MSI at the BAT26 marker, which has been suggested to be highly specific for determining the MSI-H phenotype in sporadic colon cancers (27) . The remaining 225 tumors showed altered allelic lengths at one marker and were defined as MSI-L (n = 53; 19.8%), or they were not mutated at any of the five microsatellite markers and were classified as MSS (n = 172; 64.4%).
Frequent Hypermethylation of the PTEN Promoter in Colon Cancers.
The CpG island of PTEN spans approximately 406 bp, and the sequences immediately 5′ to the transcription start site contain multiple putative binding sites for transcription factors, including AP2, AP4, E2F, and SP1, which are sensitive to methylation. The region between nucleotides −405 and −104 from the transcription site of PTEN has a 72% GC content, which fulfills the criteria for a CpG island (16) . Of the 267 cancers included in our study cohort, 146 tumors were selected for PTEN promoter methylation analysis. These tumors included all MSI-H and MSI-L cancers, as well as a comparable number of randomly chosen MSS tumors for comparison. Informative results were obtained from 130 cancers. Ten of 132 informative tumors (7.6% of this selected population) showed hypermethylation of the PTEN promoter (Table 1) ⇓ . Fig. 1A ⇓ shows the results of amplification using methylated and unmethylated specific primers. The 173-bp band represents unmethylated PTEN and is present in all tumors because, despite microdissection, these samples still contain normal cells as well as tumor cells. The 155-bp band represents the methylated PTEN promoter. To investigate the possibility that PTEN promoter methylation is a cancer-specific (28) event, we studied normal colonic cell DNA from an additional 20 individuals with no evidence of tumor (data not shown). None of these normal samples showed aberrant methylation of the PTEN promoter, indicating that the observed methylation is cancer specific and might have played a role in the progression of these neoplasms (Fig. 1B) ⇓ .
In this study, we grouped MSI-L and MSS tumors together for comparison purposes because both have similar molecular and clinical features and do not differ in clinical outcome (29) . PTEN promoter hypermethylation was significantly associated with MSI-H tumors (19.1% of MSI-H versus 2.2% of MSI-L/MSS tumors; Fisher’s P = 0.002), as depicted in Table 1 ⇓ . In keeping with this, 70% of tumors with PTEN promoter hypermethylation also contained a mutation in the poly(A)10 tract of transforming growth factor β receptor II, a defect found in only 30% of tumors without PTEN promoter hypermethylation (Fisher’s P = 0.0003; Table 2 ⇓ ).
The methylation-specific PCR data were further confirmed by performing bisulfite sequencing on some of the sporadic colon cancers methylated for the PTEN promoter. These data clearly showed that after bisulfite sequencing, each of the four cases showed sequences specific to the PTEN gene and not the pseudogene (Fig. 2A) ⇓ . As shown in this figure, there are only two critical nucleotides in the PTEN gene and the pseudogene that can discriminate between the two sequences, and in each of these cases the sequences obtained were specific for the PTEN gene and not the pseudogene. These data also clearly showed that there was heavy methylation of the PTEN gene promoter because the majority of the individual CpG sites were methylated in each of these cases (Fig. 2B) ⇓ .
PTEN Mutations in Sporadic CRCs with Hypermethylation.
PTEN mutation is an infrequent finding in colon cancer, but it may be more common in MSI-H tumors because mutations at the (A)6 repeat of PTEN were reported recently in 18–21% of colorectal tumors with MSI-H (11 , 12) . To determine whether PTEN promoter hypermethylation might function as a “second hit” in a biallelic inactivation mechanism, mutations at the poly(A)6 repeat sequences in exons 7 and 8 were determined in all tumors (n = 10) that were hypermethylated at the PTEN promoter (Table 3) ⇓ . For comparison, 16 tumors randomly chosen from a group of 111 tumors with unmethylated PTEN promoters were also analyzed (Table 3) ⇓ . Seven of the 26 tumors (26.9%) harbored frameshift mutations in the coding poly(A)6 repeats within PTEN, with a predictive effect of early stop codon insertion and a resultant failure to translate the PTEN protein. Mutation rates at the (A)6 repeat of PTEN tended to be different between methylated (40%; 4 of 10) and unmethylated (19%; 3 of 16) sporadic cancers (Fisher’s P = 0.37). The frequency of frameshift mutations was equally distributed between both exons of PTEN, with four tumors showing mutations in exon 7, and three tumors bearing mutations in the exon 8. All mutational events were independent in every tumor, and none of the cancers showed evidence of homozygous mutations or the presence of multiple mutations at exon 7 and 8 repeat sequences.
LOH Analysis at the 10q23 Locus in Colon Cancers Methylated at the PTEN Promoter.
Allelic losses at the 10q23 locus are a possible mechanism of PTEN inactivation in sporadic colon cancers containing PTEN promoter methylation but no evidence for mutations in the poly(A)6 coding tract of PTEN. LOH at 10q23 was observed in 23% (6 of 26) of the tumors, with at least one LOH event at one of the three markers analyzed (Table 3) ⇓ . We did not detect allelic imbalances in any of the normal specimens using these markers. Of all of the cases showing deletions, we observed LOH events in 83% (5 of 6) of tumors at the D10S1765 locus, 33.3% (2 of 6) of tumors at the D10S541 locus, and 16.6% (1 of 6) of tumors at the D10S215 locus. Two tumors showed allelic losses at both the D10S541 and D10S1765 markers. Interestingly, none of the tumors with frameshift mutations at the coding poly(A)6 repeat showed allelic loss at the 10q23 locus, suggesting that either a mutation or deletion might inactivate the second allele in tumors containing promoter region methylation.
PTEN Expression Correlates with the Promoter Methylation.
PTEN expression was determined by immunohistochemistry of tissue specimens from all sporadic CRCs whose PTEN promoter methylation, PTEN mutation, and 10q23 LOH status were known. Many of the tumor sections also contained stroma and/or normal colonic epithelium that showed strong cytoplasmic and nuclear PTEN immunostaining, providing internal positive controls (graded ++; Ref. 26 ). If blood vessels were present, their endothelium also expressed PTEN strongly (graded ++). In other cases, a separate section of normal tissue from the same patient was used as a control. The majority of the colon cancers with methylated PTEN promoters [8 of 10 (80%)] demonstrated weak (+) or absent (−) cytoplasmic PTEN staining (Fig. 3 ⇓ and Table 3 ⇓ ). Of the 10 tumors with methylated PTEN promoter regions, 3 tumors (30%) showed no PTEN protein immunoreactivity (−), and 5 tumors (50%) demonstrated weak (+) cytoplasmic PTEN immunostaining. The remaining two tumors (20%) showed either normal (++) cytoplasmic immunostaining or had mixed tumor cell populations showing weak/absent as well as normal immunostaining intensity. None of the unmethylated sporadic colon cancers showed complete loss of PTEN expression, whereas 19% (3 of 16) had weak (+) staining, and the majority (81%) of these cancers showed normal cytoplasmic expression. All three tumors with less than ++ staining were also positive for MSI or LOH, implying that PTEN promoter methylation is not a necessary event and that additional genetic events may lead to decreased protein expression.
Association of PTEN Promoter Methylation with Clinical and Pathological Data.
The relationship of PTEN promoter hypermethylation to clinical and histopathological variables was examined (Table 4) ⇓ . There was no significant correlation between PTEN status and any of the variables examined, although there was a trend toward a higher incidence of PTEN promoter methylation in tumors that were poorly differentiated and arose in a proximal location. Methylation of the PTEN promoter region did not influence survival probability in univariate analysis (Fig. 4) ⇓ . Numbers are too small to compare survival within tumor stage, although results are suggestive of poorer prognosis in stage 3 patients with methylated tumors (data not shown).
PTEN is a candidate tumor suppressor that may have a multifunctional role in regulation of cell proliferation, migration, and invasion (3 , 7) . Recent reports suggest that PTEN exerts an important tumor suppressor role in colorectal carcinogenesis (7 , 8) , but the extent and nature of PTEN loss during CRC development have not been fully characterized. Allelic deletions in PTEN in CRCs have been reported (13) , but somatic mutations are rarely found when unselected cohorts are examined (30) . Frameshift mutations at the (A)6 repeats in the coding sequences of PTEN have been described in 20% of patients with MSI-H sporadic CRCs (11 , 12) . Loss of PTEN function has been related to weak or absent protein expression, but the majority of patients with this phenotype harbored monoallelic mutations and/or deletions at the PTEN locus (31) . There has been no evidence to date of biallelic inactivation of PTEN during colon carcinogenesis, suggesting that additional mechanisms of gene inactivation might exist during the formation of these tumors.
Epigenetic silencing of PTEN by promoter methylation was initially postulated for a subset of prostate cancer cell lines and also identified as a mechanism in melanoma progression (31) . The genomic sequence of PTEN is 98% identical to a highly conserved processed PTEN pseudogene (psiPTEN), and this sequence identity extends 841 bp into the promoter region. As a consequence, it was recently emphasized that extreme caution needs to be exercised in selecting primer sets that can exploit the critical nucleotide differences in the sequences between PTEN and its pseudogene when analyzing the PTEN promoter region methylation status (20) . Following these recommendations, in the present investigation, we report that PTEN promoter methylation is a frequent event in MSI-H sporadic CRCs and may constitute an important epigenetic mechanism of PTEN inactivation. We also found that PTEN mutations were predominantly observed in MSI-H cancers that comprise ∼10–15% of all sporadic cancers. Methylation of the hMLH1 promoter is a mechanism of mismatch repair loss in a subset of sporadic MSI-H CRCs (32) . The predicted overall frequency of PTEN mutations in unselected sporadic cancers is 4–5%, which is less than the rate of PTEN promoter methylation observed in MSI-H tumors in our study (11 , 12) . In addition, we found a high degree of association between PTEN promoter hypermethylation and mutation of the (A)10 region of transforming growth factor β receptor II. These observations suggest that these a “methylator” phenotype can produce mismatch repair deficiency, transforming growth factor β receptor II loss, and PTEN methylation in a subset of sporadic CRCs (28 , 32) .
Although our data suggest that PTEN may be a tumor suppressor gene target in the presence of defective mismatch repair function (11 , 12) , these tumors also show heterozygous PTEN mutations. We therefore questioned whether PTEN promoter hypermethylation might constitute the “second hit” in a biallelic inactivation mechanism of colon carcinogenesis. We therefore examined the relationship among PTEN promoter hypermethylation, PTEN mutations, and PTEN LOH events in sporadic colon cancers. Interestingly, 50% of tumors with PTEN promoter hypermethylation also showed either frameshift mutations at the poly(A)6 repeat sequences or LOH events at regions flanking or within the 10q23 locus. The observed frequency of PTEN somatic frameshift mutations in MSI-H cancers (27%) was in accordance with the existing literature (11 , 12) . A LOH event in at least one of the two informative markers (D10S541 and D10S1765) was observed in 20% of tumors with PTEN promoter hypermethylation. Comparable rates of LOH have been cited for these markers in prostate, ovarian, and breast cancers (33) . None of the tumors with PTEN promoter hypermethylation contained both a mutation and a deletion. Taken together, these findings imply that, during carcinogenesis in MSI-H tumors, PTEN function is lost through a process whereby one allele is silenced by a methylation event, and the second allele is lost by a mutational event. PTEN mutations might also occur in sequences outside the (A)6 repeats investigated in our study and thus are not detected by MSI analysis. This has to be taken into account for tumors with weak PTEN expression and no LOH or MSI events.
PTEN expression has been detected in the majority of human tissues, but its distribution is quite variable (34) . The staining pattern of PTEN in the colon has not been well studied, although one report suggests that PTEN expression in the colon is very similar to that found in prostate, with punctuate, granular staining throughout the mucosal and epithelial layers (34) . In the present investigation, we found that 80% of the methylated CRCs showed reduced or lost PTEN staining. We observed complete loss of staining in three cancers, all containing PTEN mutation and/or deletion along with promoter hypermethylation. In sporadic breast and cervical cancers, the D10S541 marker, which is located just 3′ of PTEN, shows a high rate of LOH and correlates well with the loss of PTEN protein expression (35) . In these colon cancers, we observed a somewhat lower rate of LOH at D10S541 when compared with D10S1765, which is located at 5′ of PTEN. PTEN protein expression, however, correlated strongly with loss at both D10S541 and D10S1765. Although two colon tumors with methylated PTEN promoters showed normal levels of PTEN expression, these tumors did not demonstrate either mutations of the (A)6 regions of PTEN exons 7 or 8 or 10q23 LOH, suggesting that probably only one allele may have been methylated and that the remaining wild-type allele was responsible for the normal PTEN expression. All of the sporadic cancers containing unmethylated PTEN promoters showed PTEN expression, with 81% of these exhibiting normal (++) staining levels. Our data therefore suggest that loss of PTEN expression occurs primarily as a result of PTEN promoter hypermethylation combined with a PTEN mutation.
We were unable to identify significant correlations between PTEN status and any of the clinical variables studied in this investigation, although our data suggest that CRCs with PTEN promoter methylation, such as MSI-H tumors, tend to be proximally located and poorly differentiated. The cohort examined here is too small to allow a definitive conclusion concerning PTEN protein or PTEN promoter methylation status and clinical outcome independent of MSI status.
To the best of our knowledge, this is the first report describing epigenetic silencing of PTEN through hypermethylation of its promoter in a large group of sporadic CRCs. We found that epigenetic inactivation of PTEN strongly correlated with the presence of high levels of MSI, suggesting that PTEN may be targeted by the same processes creating the MSI phenotype, such as hMLH1 inactivation. Whereas many aspects of PTEN function and regulation have been identified, how the stability of PTEN protein is regulated in certain tumors without PTEN mutations, but with low or absent protein levels, remains to be established. Studies are currently under way to address these issues.
We thank Dr. Nissi Varki (Director, Histopathology Core at the University of California San Diego) for assistance with the immunostaining method and its interpretation. The institutions that participated in this study are listed in Table 5 ⇓ .
Grant support: The research for Cancer and Leukemia Group B 9865 was supported, in part, by National Cancer Institute Grant CA31946 to the Cancer and Leukemia Group B (Richard L. Schilsky, Chairman). This work was also sponsored by NIH grants RO1-CA 72851 (C. R. Boland) and RO1-CA90231 (J.M. Carethers) and in part by an American Cancer Society-IRG award (A. Goel) C. N. Arnold was funded by a grant from the Dr. Mildred-Scheel-Stiftung, Germany.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: A. Goel and C. N. Arnold contributed equally to this work.
Requests for reprints: C. Richard Boland, Baylor University Medical Center, Hoblitzelle Suite H-250, 3500 Gaston Avenue, Dallas, TX 75246. Phone: (214) 820-2692; Fax: (214) 818-9292; E-mail:
- Received August 21, 2002.
- Revision received January 20, 2004.
- Accepted March 3, 2004.
- ©2004 American Association for Cancer Research.