SEARCH

SEARCH BY CITATION

Keywords:

  • MYO1A;
  • gastric cancer;
  • endometrial cancer;
  • MSI;
  • methylation

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Brush border Myosin Ia (MYO1A) has been shown to be frequently mutated in colorectal tumors with microsatellite instability (MSI) and to have tumor suppressor activity in intestinal tumors. Here, we investigated the frequency of frameshift mutations in the A8 microsatellite in exon 28 of MYO1A in MSI gastric and endometrial tumors and found a high mutation rate in gastric (22/47; 46.8%) but not endometrial (3/48; 6.2%) tumors. Using a regression model, we show that MYO1A mutations are likely to confer a growth advantage to gastric, but not endometrial tumors. The mutant MYO1A7A protein was shown to lose its membrane localization in gastric cancer cells and a cycloheximide-chase assay demonstrated that the mutant MYO1A7A protein has reduced stability compared to the wild type MYO1A. Frequent MYO1A promoter hypermethylation was also found in gastric tumors. Promoter methylation negatively correlates with MYO1A mRNA expression in a series of 58 non-MSI gastric primary tumors (Pearson's r = −0.46; p = 0.0003) but not in a cohort of 54 non-MSI endometrial tumors and treatment of gastric cancer cells showing high MYO1A promoter methylation with the demethylating agent 5-aza-2′-deoxycytidine, resulted in a significant increase of MYO1A mRNA levels. We found that normal gastric epithelial cells, but not normal endometrial cells, express high levels of MYO1A. Therefore, when considered together, our findings suggest that MYO1A has tumor suppressor activity in the normal gastric epithelium but not in the normal endometrium and inactivation of MYO1A either genetically or epigenetically may confer gastric epithelial cells a growth advantage.

Microsatellite instability (MSI) is a phenotype observed in a significant proportion of tumors mainly from colorectal, endometrial, and gastric origin.1, 2 MSI tumors are characterized by the presence of frequent insertion/deletion (indel) mutations in short tandem DNA repeats of 1–4 bp known as microsatellites.3–5 This phenotype is the result of a defective DNA mismatch repair system3–5 and patients with Lynch syndrome (also known as hereditary nonpolyposis colorectal cancer, HNPCC) inherit a mutant copy of a mismatch repair gene (mainly MLH1 and MSH2) and are predisposed to this type of tumors.6, 7 An MSI phenotype is observed in 8–25% and 20–30% of gastric8–11 and endometrial12–14 tumors, respectively. Gastric cancer is the fourth and fifth tumor type most frequent in males and females, respectively, and was responsible for 738,000 deaths worldwide in 2008. Endometrial cancer is the sixth tumor type most frequent in women worldwide, with 287,100 new cases estimated in 2008 and a relatively high survival rate when compared to other tumor types.15 The progression of both gastric and endometrial tumors is the result of the accumulation of genetic and epigenetic alterations. Gastric tumors are classified as diffuse or intestinal according to their histology and these subtypes are associated with different genetic abnormalities such as E-cadherin mutation/methylation or TP53 mutations, respectively.16 Most endometrial tumors (70–80%) are classified as Type I and follow the estrogen-related pathway with frequent alterations in PTEN, PIK3CA, KRAS and CTNNB1 (β-catenin), as well as an MSI phenotype. Type II endometrial tumors (estrogen-unrelated pathway) show frequent mutations in TP53 and chromosome instability.17, 18

In MSI tumors, some Indel mutations that occur in the coding sequence of tumor suppressor genes cause a frameshift that can result in a nonfunctional protein, conferring a growth advantage on cancer cells. These mutations are clonally selected and are observed in a significant proportion of the tumors, above the background mutation rate expected by chance in microsatellites of the same length. Over the last two decades a significant number of genes have been found to accumulate mostly inactivating mutations in a high proportion of MSI tumors.19–22 For instance, 60–90% of MSI colorectal tumors have mutations in Transforming Growth Factor β receptor type II (TGFBRII) and EPH receptor B2 (EPHB2) mutations are found in >40% of these tumors.21, 23–26 Mutations in these genes have been shown to confer tumor cells a significant growth advantage and the high frequency of these mutations highlights the relevance of the inactivation of these pathways in tumor progression.24–27 However, significant variability can be found in the mutation frequency of specific target genes in MSI tumors derived from different organs. For instance, although TGFBRII and EPHB2 mutations are common in colorectal and gastric tumors, they are not found at high frequency in MSI endometrial tumors.2, 19, 20, 28 These differences underscore the heterogeneity in the oncogenic process in different organs and provide a good opportunity to investigate such divergent tumorigenic pathways.

We have recently identified mutations in an A8 repeat track in the last exon (exon 28) of Myosin Ia (MYO1A) in 32% (37 of 116) of colorectal MSI tumors.29 MYO1A is one of eight membrane binding class I myosins expressed in vertebrates.30 Class I myosins are monomeric motors with the capacity for binding multiple calmodulin light chains in a Ca2+ sensitive manner and the ability to interact directly with membranes by virtue of highly basic regions (TH1) in their C-terminal tail domains.31, 32 In the intestine, MYO1A is the most abundant actin-based motor protein found in the microvillar compartment. It has long been known to interact with the apical membrane of the microvilli providing a physical link between the plasma membrane and the underlying actin bundle.21, 22 Using large human tumor collections and animal models we have shown that MYO1A is genetically and epigenetically inactivated in both MSI and non-MSI colorectal tumors and that the loss of a functional MYO1A protein significantly contributes to intestinal tumor progression.29

Here, we show that MYO1A is expressed at high levels in the normal gastric mucosa but not in the normal endometrium. This is consistent with the high mutation frequency observed in MSI gastric, but not endometrial, tumors. We also demonstrate that the frequently observed mutant MYO1A7A protein loses its normal membrane localization in gastric cancer cells and has reduced stability compared to the wild type protein. Moreover, MYO1A was also found to be frequently inactivated by aberrant hypermethylation of the proximal promoter region in non-MSI gastric tumors.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Patient samples and tumor cell lines

A series of 47 MSI gastric tumors and the matched normal tissue was obtained from Sapporo Medical University (Sapporo, Japan). A total of 48 MSI endometrial tumors were obtained from the Vall d'Hebron University Hospital (Barcelona, Spain) and the University Medical Center Groningen (Groningen, The Netherlands). The matched endometrial normal samples were not available. Written informed consent was obtained from all patients in the study according to the Clinical Research Ethical Committee in the different institutions. MSI was determined in these samples by PCR amplification and sequencing of the five microsatellite markers (BAT26, BAT25, D5S346, D2S123 and D17S250) recommended by the NCI guidelines.33 Tumors with two or more of the five markers displaying instability were defined as MSI. MYO1A mRNA levels in human normal and tumor samples were obtained from a collection of 667 normal human samples from different tissues (Gene Expression Omnibus: GSE7307) and 10,000 normal and tumor samples from genesapiens.org.34 MYO1A methylation and mRNA levels in the same series of 58 primary gastric tumors and 54 primary endometrial tumors, were obtained from The Cancer Genome Atlas (http://cancergenome.nih.gov/). Immunohistochemical staining with the monoclonal anti-MYO1A antibody HPA041633 (Atlas Antibodies) is from Proteinatlas.org.35

The gastric cancer cell lines AGS, MKN45 and KATO-III were maintained on Dulbecco's modified Eagle's medium (DMEM; PAA Laboratories) containing 10% fetal bovine serum (FBS; PAA Laboratories) and 1× antibiotic-antimycotic (Invitrogen; 10,000 U penicillin, 10,000 μg of streptomycin and 25 μg/ml of amphotericin B) at 37°C/5% CO2. Cell lines were tested to be free of mycoplasma contamination. KATO-III cells were a kind gift of Dr. Carme de Bolos (Programa de Recerca en Càncer, IMIM-Hospital del Mar, PRBB, 08003 Barcelona, Spain). MKN45 cells were obtained from the American Type Culture Collection (ATCC). The ID of all the cell lines was confirmed by STR profiling.

MYO1A mutations in MSI gastric and endometrial tumors

MYO1A mutations were screened by direct sequencing of PCR amplified genomic DNA (ABI Prism 3100 sequencer; Applied Biosystem). Tumors were scored as mutant when the height of the mutant peak in the position of the last A of the A8 track in exon 28 of MYO1A was >20% of the wild type allele in the electropherograms (see Fig. 1a–1d). The primers used were MYO1A-EX28-F AGGGGATGGGCACTAGACTT and MYO1A-EX28-R CAAGGAGCTTGAGGAGGAAA for DNA extracted from fresh frozen samples and MYO1A-EX28-FFPE-F CCTCACTGCACAGTCACCTC and MYO1A-EX28-FFPE-R CCCTCTCCAGGTTCTCAGTG for DNAs extracted from formalin-fixed, paraffin-embedded samples. The annealing temperature for both primer sets was 65°C. Mutation frequency data of more than 3,500 coding and noncoding mononucleotide repeat tracts (MNRs) from MSI gastric and endometrial tumors was used to perform a nonlinear regression analysis as previously reported.19 Statistical analysis was performed using the R software environment version 2.1.1 (www.R-project.org) in combination with the software library nls2 (version 2003.1) for nonlinear regression.20, 36

thumbnail image

Figure 1. MYO1A mutations in gastric and endometrial tumors. Panels a and b show a chromatogram of the MYO1A region of exon 28 containing the A8 microsatellite in a wild type normal gastric mucosa sample and the matching mutant MSI tumor, respectively. (c, d) MYO1A chromatograms of a wild type and a mutant MSI endometrial tumor, respectively. The arrows in panel b and d indicate the heterozygous deletion of one A in the A8 track. Panels e and f show the mutation frequency of MYO1A A8 track in MSI gastric and endometrial tumors, respectively (filled back circle). Mutation frequency data of more than 3,500 coding (gray filled circles) and noncoding (open diamonds) mononucleotide repeat tracts (MNRs) from MSI gastric (e) and endometrial (f) tumors was used to perform a nonlinear regression analysis. The X-axis represents the MNR length and the Y-axis the mutation frequency. The mean mutation frequency (black solid line) and two prediction lines (two-fold standard error, dashed lines) are shown. Repeats outside the region delimited by the two prediction lines are likely to represent genes under positive or negative selection pressure. The A8 MYO1A repeat (filled back circle) was above the upper prediction line for MSI gastric cancer (e) but not for MSI endometrial cancer (f). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Western blotting

The stomach and the uterus of wild type and Myo1a KO mice were dissected, longitudinally opened and washed with ice-cold phosphate buffered saline. The stomach mucosa and the endometrial layer of the uterus were scraped with a scalpel blade in an ice bath, snap-frozen in liquid nitrogen and stored at −80°C. Cell lines were scraped with a rubber policeman on ice. Tissue/cell pellets were resuspended in 50 μl of RIPA lysis buffer (0.1% SDS; 1% NP40; 0.5% Na-deoxycholate in PBS) containing proteases inhibitor (Pepstatine 5 μg/μl; PMSF 0.3 mM; Aprotinine 1 μg/μl; Sodium orthovanadate 100 μM). After a 30-min incubation on ice, cells were sonicated for 10 sec three times at 20–50 kHz on ice, then the lysate was centrifuged for 20 min at 16000g at 4°C and the supernatant was transferred into a new microtube and stored at −80°C. Protein concentrations were quantified with a BCA Protein Assay Kit (Thermo Scientific). Lysates (100 μg of total protein) were subjected to SDS–PAGE on a 10% gel and transferred to PVDF membranes as described before37 and hybridized using antibodies against human MYO1A38 (1:1000), tubulin (1:1000; clone 6C5; Santa Cruz) or GFP39 (1:5; clone JFP-J1, RIKEN BioResource Center). The membrane was incubated for 1 hr at room temperature with a secondary antibody conjugated with horseradish peroxidase. Secondary antibodies were used at a dilution of 1:5000 in 5% skimmed milk in PBS-0.1%Tween-20 (all from Dako). The membrane was washed 3 × 10 min in PBS-0.1%Tween-20 and signal detection was achieved by incubating the membranes with Enhanced Chemiluminescence substrate (ECL, Healthcare) or Immobilion Western ECL substrate (Millipore). The levels of Tubulin were used as a loading control.

Confocal microscopy

AGS cells were transfected (Lipofectamine 2000; Invitrogen) with the vectors EGFP-MYO1A or EGFP-MYO1A7A and seeded on gelatine-coated glass coverslips. Cells were then fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 and stained with rhodamine-phalloidin (0.1 μM; Cytoskeleton) and DAPI. Images were captured with a confocal microscope (FV1000 Olympus).

Effects of MYO1A7A mutations in protein stability

HEK-293T cells were seeded in 6-well plates (500,000 cells/well) and allowed to attach to the substrate for 24 hr in DMEM (PAA Laboratories) supplemented with 10% FBS and 1× antibiotic antimycotic (Invitrogen; 10,000 U penicillin, 10,000 μg of streptomycin and 25 μg/ml of amphotericin B) at 37°C and 5% CO2. Cells were transfected with pEGFP-MYO1AWT or pEGFP-MYO1AA7 with calcium phosphate as previously described.40 Twenty-four hours after transfection, cycloheximide was added to the cells at a final concentration of 50 μg/ml. Cells were collected at the indicated timepoints and processed for Western blotting as described above. A monoclonal antibody against GFP39 was used (1:5; clone JFP-J1, RIKEN BioResource Center). The EGFP-MYO1A and tubulin bands were quantified using GeneTools software (Syngene, Cambridge, United Kingdom), and the average of two independent experiments was plotted using Prism (GraphPad Software).

MYO1A promoter methylation in non-MSI gastric and endometrial tumors

DNA methylation in the CpG dinucleotide located in position −154 bp was assessed in MKN45 and KATO-III gastric cancer cell lines using bisulfite sequencing. Briefly, genomic DNA was isolated from each line using the phenol/chloroform method as described before.41 DNA was treated with EZ DNA Methylation-Gold Kit (Zymo Research) following the manufacturer's instructions and then PCR-amplified with the following conditions: 94°C for 10 min, then 45 cycles of 94°C for 30 sec, 59.1°C for 30 sec and 72°C for 30 sec and then 72°C for 10 min. Primers for DNA amplification and sequencing were designed using MethPrimer 1.1 software (MYO1A-BISULF-CG154-F TTTAAATTTGGGAGATAATGGAGTAAG and MYO1A-BISULF-CG154-R AATCAACACAAAATCCAAACTATTC). The 174 bp-product of the PCR amplification of bisulfite-converted DNA was cloned using the CloneJET PCR Cloning Kit (Thermo Scientific) according to manufacturer's instructions and a minimum of eight clones were sequenced using BigDye Terminator 1.1 and an ABI PRISM 3,100 sequencer (Applied Biosystems, Foster City, CA).

Genome-wide methylation data are available for 58 gastric and 54 endometrial non-MSI tumors through the public repository of The Cancer Genome Atlas (http://cancergenome.nih.gov/). The Infinium quantitative methylation assay (Illumina) was used by the TCGA Research Network to assess the levels of methylation of 27,578 CpG sites located within the proximal promoter regions of transcription start sites of 14,475 consensus coding sequencing (CCDS) using HumanMethylation27 Bead Chips. For MYO1A, two CG dinucleotides located −154 bp and +271 bp relative to the transcription start site are interrogated. Messenger RNA expression data are available for the same tumor samples using Gene Expression Microarray, 1 × 244 K (G4502A, Agilent; endometrial tumors) or RNA sequencing (Illumina Genome Analyzer RNA Sequencing; gastric tumors).

5-Aza-2′-deoxycytidine treatment and quantitative RT-PCR

Cells were grown in complete DMEM medium containing 5-aza-2′-deoxycytidine (0, 2.5 or 10 μM; Sigma) for 72 hr. Medium was replaced every 24 hr. Total RNA was extracted from control and 5-aza-2′-deoxycytidine treated cells, using TRI Reagent (Molecular Research Center) according to the manufacturer's instructions. Total RNA (500 ng) was reverse transcribed using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems) and relative MYO1A mRNA levels were assessed by Real-Time PCR using SYBR Green Master Mix (Applied Biosystems, Branchburg, NJ). We used GAPDH as a standardization control for the 2−ΔΔCt method as described before.42 The primers used were MYO1A-qPCR-F TTCCTACTGGGGCTGAAGAA, MYO1A-qPCR-R CTCCTGATTTGCTGTGCTGA, GAPDH-qPCR-F ACCCACTCCTCCACCTTTGAC and GAPDH-qPCR-R CATACCAGGAAATGAGCTTGACAA.

Results

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Frequent MYO1A mutations in MSI gastric but not in MSI endometrial tumors

We have recently reported frequent frameshift mutations in the A8 track in exon 28 of MYO1A both in MSI colorectal cancer cell lines (44.4%; 16/36) and primary tumors (31.3%; 42/134). Because gastric and endometrial tumors frequently show a microsatellite instable phenotype, we investigated the incidence of MYO1A7A mutations in MSI tumors derived from these organs. Mutations in this A8 microsatellite were frequent in gastric MSI tumors (22/47; 46.8%) but not in endometrial MSI tumors (3/48; 6.2%; Fig. 1a–1d). All the mutations found appeared to be heterozygous as the wild type allele was always visible (Fig. 1b and 1d). Twenty-four of the 25 mutations observed (96%) in both gastric and endometrial cases were A8−>A7 mutations and one gastric case showed an A8−>A6 mutation. No mutations were observed in the DNA of the matching normal tissue from the gastric cancer cases (Fig. 1a). The normal samples from the three mutant endometrial cases were not available for analysis. To further investigate the potential of MYO1A mutations as a driver of the tumorigenic process, we used a regression model where the frequency of MYO1A mutations is plotted against the mutation rates of >3,500 coding and noncoding repeats.20 As observed for colorectal MSI tumors,29 gastric tumors showed a mutation rate that is significantly above what would be expected by chance, suggesting that these mutations confer a growth advantage on gastric tumors (Fig. 1e). Mutation frequency in endometrial MSI tumors, however, falls within the values expected due to a stochastic process, suggesting no selective pressure in this tumor type (Fig. 1f). No significant associations were found between MYO1A mutations and clinicopathological characteristics of the gastric MSI cases in the study (Supporting Information Table 1).

MYO1A7A mutations disrupt membrane localization and reduce protein stability

The A8->A7 MYO1A frameshift mutation changes the last eleven amino acids of the protein, and we have previously shown that the MYO1A7A mutant fails to correctly localize to the cytoplasmic membrane in colon cancer cells.29 Here, we investigated the effect of this frameshift mutation in gastric cancer cells. Transfection of gastric cancer AGS cells with the wild type EGFP-MYO1A protein demonstrated its plasma membrane localization (Fig. 2a). However, the mutant EGFP-MYO1A7A protein showed a diffuse cytoplasmic localization (Fig. 2e). The same expression pattern of the wild type and A8−>A7 mutant was observed in LLC-PK1/CL4 and HeLa cells, a swine kidney proximal tubule cell line and a cervical cancer cell line, respectively (not shown), demonstrating that the mutant MYO1A7A protein losses the membrane localization independently of the cell type of origin.

thumbnail image

Figure 2. The A8−>A7 mutation disrupts the membrane localization of MYO1A. Transfection of wild type EGFP-MYO1A into AGS gastric cancer cells showed the expected membrane localization (a). Panels b and c show staining with Rhodamine-phalloidin (actin) and DAPI (nuclear staining), respectively. The merged image is shown in d. However, EGFP-MYO1A7A showed a diffuse cytoplasmic staining (e). Actin and nuclear staining as well as the merge of all three channels are shown in panels fh, respectively. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Using in vitro systems with forced overexpression of the mutant MYO1A7A, we have previously shown that this mutant form of MYO1A has a dominant negative activity and significantly interferes with the capacity of MYO1A to regulate the differentiation and polarization of colon cancer cells.29 However, the dominant negative activity in primary tumors would depend on the steady state levels of the mutant MYO1A7A protein, and the effects of this frameshift mutation on protein stability have not been directly investigated. Transfection of wild type EGFP-MYO1A or mutant EGFP-MYO1A7A protein into HEK293 cells followed by inhibition of protein synthesis with cycloheximide treatment revealed that the mutant MYO1A7A protein has reduced stability compared with the wild type MYO1A protein (Fig. 3). This observation may have important implications regarding the biological relevance of the dominant negative effects of the mutant protein in primary tumors and will be discussed below.

thumbnail image

Figure 3. Stability of MYO1A7A mutant protein. (a) HEK293 cells were transfected with vectors expressing the wild type (EGFP-MYO1AWT) or mutant EGFP-MYO1A (EGFP-MYO1A7A) and the levels of EGFP-MYO1A protein were assessed by Western blot following cycloheximide (CHX) treatment for the indicated time. (b) EGFP-MYO1A and Tubulin protein levels were quantified by densitometry in two independent experiments and the mean (±SEM) background-corrected ratio was plotted as a function of time. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

MYO1A promoter methylation in gastric and endometrial cancer

We have previously demonstrated that although the promoter of MYO1A does not have a dense CpG island, methylation of two CpG dinucleotides located in position −154 bp and +271 bp relative to the transcription start site of MYO1A regulate gene expression.29 As previously shown for colon cancer cells, we also observed frequent methylation of these two CpG sites in primary non-MSI gastric tumors (14 of 58 tumors −24.1%, had methylation levels >50%; Fig. 4a). Moreover, we observed a significant inverse correlation between the levels of MYO1A mRNA expression and methylation of the MYO1A promoter in this series of 58 primary gastric tumors (Pearson's r = −0.46; p = 0.0003; Fig. 4a). However, no correlation was observed between MYO1A mRNA and promoter methylation levels in a series of 54 primary non-MSI endometrial tumors (Pearson's r = 0.02; p = 0.87; Fig. 4b). The presence of MYO1A methylation was confirmed by bisulfite sequencing in gastric cancer cell lines (Fig. 4c–4e). Moreover, inhibition of DNA methyltransferases using 5-aza-2′-deoxycytidine in gastric cancer cells (MKN45) that showed high levels of methylation (Fig. 4c, 4d and 4e) resulted in a significant increase in the levels of MYO1A mRNA expression (Fig. 4f). As a control, we showed that 5-aza-2′-deoxycytidine treatment of gastric cancer cells with low methylation levels of the MYO1A promoter (KATO-III; Fig. 4c 4d and 4e) did not result in higher levels of MYO1A mRNA (Fig. 4g). These results demonstrate that aberrant methylation of the promoter of MYO1A is a common mechanism of inactivation in gastric tumors.

thumbnail image

Figure 4. MYO1A promoter methylation in gastric cancer. Correlation between mRNA levels and methylation levels in 58 gastric non-MSI tumors (a; Pearson's r = −0.46; p = 0.0003) and 54 non-MSI endometrial tumors (b; Pearson's r = −0.02; p = 0.87). Panel c shows the location of the CpG in position −154 bp relative to the transcription start site (TSS) of MYO1A (vertical lines: CpG dinucleotides). (d) Bisulfite treated genomic DNA containing the −154 bp MYO1A CpG from MKN45 and KATOIII cells was amplified and cloned as described in the Material and Methods section. At least eight individual clones from each cell line were sequenced. White and gray boxes represent unmethylated and methylated sequences, respectively. Panel (e) shows representative methylated (MKN45) and unmethylated (KATOIII) MYO1A promoter sequences. The C in position −154 bp relative to the TSS is highlighted in gray. (f) Treatment of MKN45 gastric cancer cells with high methylation levels (70%) with 5-aza-2′-deoxycytidine resulted in a significant increase in the levels of mRNA expression of MYO1A (Real-Time RT-PCR). (g) As a control, we used KATOIII cells showing low levels of methylation of the MYO1A promoter (10%). No changes in mRNA levels were observed following treatment with 5-aza-2′-deoxycytidine in this cell line. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

MYO1A is expressed in the normal stomach but not in the normal endometrium

To understand better why gastric, but not endometrial tumors accumulate high frequency of MYO1A mutations in MSI cases and frequent promoter methylation, we investigated the levels of expression of MYO1A in normal endometrium and stomach as well as tumors derived from these organs. Western blotting analysis demonstrated high levels of expression of Myo1a in stomach epithelial cells while endometrial samples showed undetectable levels in Myo1a wild type mice (Fig. 5a). As expected, both stomach and endometrial normal samples from Myo1a KO mice showed no expression of Myo1a, demonstrating the specificity of the antibody (Fig. 5a). Moreover, immunohistochemical staining with an anti-MYO1A antibody revealed high expression levels in human normal gastric samples (Fig. 5c and 5f) and undetectable levels of MYO1A in human normal endometrium (Fig. 5d and 5g). In good agreement, the levels of MYO1A mRNA in human normal stomach samples were significantly higher than in normal endometrial samples (Student's t-test p = 0.0005; Fig. 5h). Moreover, endometrial MYO1A levels did not differ from the levels observed in other human normal tissues that are expected to have low/absent MYO1A expression (myometrium, prostate and ovary; Fig. 5h). In addition, MYO1A mRNA expression in human gastric tumors was comparable to intestinal tumors and significantly higher than in other tumor types investigated, including endometrial tumors (uterine adenocarcinomas; Fig. 5i).

thumbnail image

Figure 5. MYO1A expression in gastric and endometrial normal and tumor tissues. (a) The levels of MYO1A protein were assessed by Western blot analysis in samples enriched in epithelial cells of the gastric mucosa or the endometrium of Myo1a wild type and KO mice. Lysates of Caco2 colon cancer cells after 0 or 21 days of confluent culture (d0 and d21, respectively) were used as a positive control. (b) Immunohistochemistry with an anti-MYO1A antibody revealed high levels of expression in the brush border of human small intestinal samples demonstrating the specificity of the antibody. High MYO1A was also observed in the epithelial cells of the stomach (c) but not in glandular endometrial cells (d). Higher magnification of the indicated areas is shown in the lower panels (eg). (h) Relative MYO1A mRNA levels in human normal tissues. (i) Relative MYO1A mRNA levels in human tumors of different origin. Filled gray bars are gastrointestinal carcinomas and in black are endometrial tumors (uterine adenocarcinomas). Box-whisker plot of the gene's expression in cancer tissues. The bottom of the box is the 25th percentile of the data, the top of the box is the 75th percentile and the vertical red line is the median. The whiskers extend to 1.5 times the interquartile range from the edges of the box and any data points beyond this are considered outliers, marked by hollow circles. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

High levels of expression of MYO1A are observed in the epithelial layer covering the luminal surface of the intestinal tract.29, 38 We have recently shown that MYO1A suppresses the tumorigenic process in normal intestinal epithelium and the loss of this Myosin is important for the adenoma-to-carcinoma transition of colon tumors.29 Here, we show that MYO1A is also expressed in the normal gastric mucosa but no detectable levels were observed in the endometrial layer of the uterus in mice or humans. Given the polarized structure of epithelial cells of the gastric mucosa and the presence of abundant microvilli, it is possible that, as observed in the intestine, MYO1A has a tumor suppressor effect in the stomach.

We found mutations in the A8 track in exon 28 of MYO1A in MSI gastric tumors at a frequency higher than expected given the background mutation rates in these tumors. Indeed, MYO1A is among the most frequently mutated genes in MSI gastric tumors (>46%). Only TGFBRII has been reported to be mutated at higher frequency that MYO1A in human MSI gastric tumors (60–90%). The observed MYO1A mutation incidence was similar to the frequency of mutations reported for a noncoding A13 track in the 3′-UTR of EGFR and is above the observed mutation frequency in the G8 track of BAX or the A8 microsatellite of MSH3 (25–40%), two well established mutational targets with known tumor suppressor role in this tumor type.2, 23, 28, 43 This observation is consistent with the notion that MYO1A is expressed at high levels in normal gastric epithelial cells and suppresses tumor progression. Here, we show that the mutant MYO1A7A protein loses its membrane localization in gastric cancer cells. Moreover, it has previously been shown that MYO1A7A interferes with the activity of the wild type MYO1A protein.29 Therefore, the MYO1A7A mutation would confer the tumor cells a significant growth advantage and would be clonally selected, explaining the high proportion of gastric tumors with A8−>A7 mutations. Endometrial tumors are derived from the normal epithelial cells in the inner layer of the uterus. Since normal endometrial cells do not have high levels of expression of MYO1A, there would be no selective pressure for the accumulation of mutations in the A8 repeat of MYO1A in MSI endometrial tumors. This is consistent with our results showing that the MYO1A mutation frequency observed in MSI endometrial tumors is at levels expected by chance, indicating no selective pressure for these mutations.

No homozygous MYO1A7A mutations were observed in MSI gastric tumors. This is consistent with our earlier observation in colorectal tumors29 and suggests a lack of selective pressure to acquire mutations in the second allele of MYO1A. The insufficiency of a single functional copy of MYO1A to maintain normal function (haploinsufficiency) could explain the absence of homozygous mutations in MYO1A. About 300 genes have been reported to be haploinsufficient in the human genome.44 For example, frequent monoallelic mutations are found in flap endonuclease (FEN1) and Beclin 1 (BECN1) and have been shown to be haploinsufficient for their tumor suppressor activity.45, 46 We have previously shown that monoallelic inactivation of Myo1a in a mouse model results in an increased number of intestinal tumors and shorter animal survival following tumor initiation with heterozygous Apc mutations,29 indicating that MYO1A is haploinsufficient for tumor suppression in the gastrointestinal tract. In addition, the dominant negative activity of the mutant protein could contribute to the absence of homozygous MYO1A7A mutations. Using forced expression of ectopic mutant MYO1A7A in vitro, we have shown that the mutant protein has a dominant negative activity causing the loss of polarization and differentiation and conferring a growth advantage to tumor cells that express moderate/high endogenous levels of wild type MYO1A.29 However, we show here that the mutant MYO1A7A protein has reduced stability compared with the wild type protein and this probably limits the contribution of the dominant negative activity towards tumor progression in primary gastrointestinal tumors.

Promoter hypermethylation is well documented to act as an epigenetic mechanism of regulation of gene expression and the role of promoter methylation on the transcriptional silencing of tumor suppressor genes during the oncogenic process has been reported in most tumor types.47 The regulatory role of CpG methylation has been mostly studied in the context of CpG islands containing a high density of these dinucleotides. However, more recently, it has been shown that for genes with CpG islands in their promoters, CpG methylation in regions outside the CpG island is most important for gene expression regulation.48 Moreover, CpG methylation in promoters that do not have a conventionally defined CpG island is emerging as an important mechanism of tumor suppressor gene silencing that could account for >20% of the genes regulated by this epigenetic mechanism.48–50 Although the promoter of MYO1A does not contain a dense CpG island, we have previously shown that the expression of MYO1A is regulated by aberrant methylation in colorectal tumors.29 Here, we found frequent methylation in the promoter of MYO1A in non-MSI gastric cancer cell lines and primary tumors. An inverse correlation was observed between the expression of MYO1A and the levels of methylation in a series of 58 primary gastric tumors, indicating that promoter methylation may also regulate the expression of MYO1A in this tumor type. Moreover, treatment with the methyltransferase inhibitor 5-aza-2′-deoxycytidine resulted in a significant increase in the levels of MYO1A mRNA expression in gastric cancer cells that have high levels of promoter methylation (MKN45) but not in cells with low levels of promoter methylation (KATO-III). Collectively these results indicate that promoter hypermethylation is an alternative mechanism of inactivation of MYO1A in gastric tumors.

In conclusion, in this study we identified MYO1A as one of the main mutation targets in MSI gastric tumors, while background mutation rates were observed in endometrial MSI tumors. Using a regression model, we demonstrate that MYO1A mutations are clonally selected and may confer a growth advantage to gastric cancer cells. We also demonstrate that the frequently observed A8−>A7 mutations in the microsatellite in exon 28 of MYO1A lead to the loss of membrane localization and reduced stability of the mutant protein. Moreover, we have identified promoter hypermethylation as an alternative mechanism of inactivation of MYO1A in gastric tumors. These results further demonstrate the importance of MYO1A inactivation in gastrointestinal tumors and highlight the profound differences in the molecular mechanism underlying the tumorigenic process of colon, gastric and endometrial tumors.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

The JFP-J1 anti-GFP antibody used was obtained from RIKEN BioResource Center and is the property of Mitsubishi Kagaku Institute of Life Sciences (MITILS) and Mitsubishi Chemical Corporation (MCC).

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
IJC_27856_sm_SuppInfo.docx18KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.