Eun Mi Je

Mutational analysis of MED12 exon 2 in uterine leiomyoma and other common tumors

Recurrent somatic mutations in MED12 exon 2 have recently been reported in uterine leiomyomas. The recurrent nature of the mutations strongly suggests that the mutations may play important roles in the pathogenesis of uterine leiomyomas. The aim of our study was to see whether MED12 exon 2 mutations occur in other human tumors besides uterine leiomyomas. We also attempted to confirm occurrence of the MED12 mutations in uterine leiomyomas of Korean patients. For this, we analyzed 1,862 tumor tissues, including a variety of carcinomas, leukemias and stromal tumors by single‐strand conformation polymorphism analysis. We found MED12 mutations in 35 uterine leiomyomas (35/67; 52.2%) and one colon carcinoma (0.3%), but none in other tumors. The MED12 mutations consisted of missense (77%) and inframe insertion‐deletion (23%) mutations, the pattern of which was similar to the earlier report. Our data indicate that MED12 exon 2 mutations may be tissue‐specific to uterine leiomyoma and rare in other tumors. Our study suggests that the MED12 mutations play unique roles in the pathogenesis of uterine leiomyomas and mutated MED12 could be therapeutically targeted in uterine leiomyomas.

Mutational analysis of splicing machinery genes SF3B1, U2AF1 and SRSF2 in myelodysplasia and other common tumors

Recurrent somatic mutations in splicing machinery components, including SF3B1, U2AF1 and SRSF2 genes have recently been reported in myelodysplastic syndromes (MDS). Such a recurrent nature strongly suggests that these mutations play important roles in tumor development. To see whether SF3B1, U2AF1 and SRSF2 mutations occur in other human tumors besides MDS, we analyzed the hotspot mutation regions of these genes in 2,345 tumor tissues from various origins (61 MDS, other 616 hematologic tumors, 1,421 epithelial tumors and 247 non‐epithelial stromal tumors) by single‐strand conformation polymorphism analysis. We found SF3B1, U2AF1 and SRSF2 mutations in 5 (8.2%), 12 (19.7%) and 8 (13.1%) of 61 MDS, respectively. We also confirmed these mutations in other myeloid neoplasia, including de novo acute myelogenous leukemia (AML), chronic myelomonocytic leukemia and MDS/myeloproliferative disorder. In addition, we discovered that the SRSF2 gene was mutated in two childhood acute lymphoblastic leukemias (childhood ALL) (1.5%). In solid tumors, we found SF3B1 mutations in gastric and prostate cancers, and U2AF1 mutation in a borderline mucinous tumor of ovary, but the overall incidences of the hotspot mutation regions were very low (0.2%). Our data suggest that SF3B1, U2AF1 and SRSF2 mutations occur not only in myeloid lineage tumors but also in lymphoid lineage tumors. The data suggest that the splicing gene mutations play important roles in the pathogenesis of hematologic tumors, but rarely in solid tumors.

Mutational and expressional analyses of SPOP, a candidate tumor suppressor gene, in prostate, gastric and colorectal cancers

Mounting evidence exists that alterations of ubiquitination processes are involved in cancer pathogenesis. Speckle‐type POZ protein (SPOP) is a key adaptor for Cul3‐based ubiquitination process. Recent studies reported that SPOP may be a tumor suppressor gene (TSG) and somatic mutation of SPOP was detected in prostate cancer (PCA). The aim of this study was to see whether alterations of SPOP protein expression and somatic mutation of SPOP gene are features of cancers. In this study, we analyzed SPOP somatic mutation in 45 gastric (GC), 45 colorectal cancer (CRC) and 45 PCA by single‐strand conformation polymorphism (SSCP). Also, we analyzed SPOP protein expression in 60 GC, 60 CRC and 60 PCA by immunohistochemistry. Overall, we detected three somatic missense mutations of SPOP gene in the coding sequences (p.Ser14Leu, p.Tyr87Cys and p.Phe133Leu). The mutations were observed in two PCA and one CRC. Of note, the p.Phe133Leu was a recurrent mutation reported in an earlier study. In the immunohistochemistry, SPOP protein was expressed in normal gastric, colonic and prostate epithelial cells, whereas it was lost in 30% of GC, 20% of CRC and 37% of PCA. Our data indicate that loss of SPOP expression was common in GC, CRC and PCA, but somatic mutation of SPOP in this study was rare in these tumors. Also, the data provide a possibility that loss of expression of SPOP gene might play a role in cancer pathogenesis by altering TSG functions of SPOP.

Frameshift mutations of axon guidance genes ROBO1 and ROBO2 in gastric and colorectal cancers with microsatellite instability

Aims: Several lines of evidence indicate that axon guidance genes are involved not only in neural development but also in cancer development. ROBO1 and ROBO2, crucial regulators of axon guidance, are considered potential tumour suppressor genes. The aim of this study was to explore whether ROBO1 and ROBO2 genes are somatically mutated and expressionally altered in gastric (GC) and colorectal cancers (CRC). Methods: In a public database, we observed that both ROBO1 and ROBO2 had mononucleotide repeats in their coding exons that could be mutation targets in cancers with microsatellite instability (MSI). We analysed mutations of these repeats in 77 GC and 88 CRC either with high MSI (MSI-H) or low MSI/microsatellite stability (MSI-L/MSS) by single-strand conformation polymorphism (SSCP) and DNA sequencing. We analysed ROBO1 and ROBO2 expressions in GC and CRC by immunohistochemistry as well. Results: Overall, we found five ROBO1 and five ROBO2 frameshift mutations in the repeats. They were detected exclusively in the cancers with MSI-H (10/70, 14.2%), but not in MSI-L/MSS (0/95, 0%) (p = 0.018). In the immunohistochemistry, loss of ROBO2 expression was identified in 22 (29%) and 17 (19%) of GC and CRC, respectively, while increased expression of ROBO2 was found in 15 (20%) and 22 (25%) of GC and CRC, respectively. There were co-occurrences of mutation and loss of expression in both ROBO1 (4/5, 80% mutated cases, p < 0.001) and ROBO2 (5/5, 100% mutated cases, p < 0.05) genes. Conclusion: This is the first report of ROBO1 and ROBO2 frameshift mutations in GC and CRC. Frameshift mutations of ROBO1 and ROBO2 genes and alteration of ROBO2 expression in GC and CRC suggest that both genes might play roles in the pathogenesis of GC and CRC.

Mutational and expressional analyses of MYD88 gene in common solid cancers

AIMS AND BACKGROUND Myeloid differentiation primary response gene 88 (MYD88) is a protein involved in hematopoietic differentiation and innate immunity. Recent studies revealed MYD88 mutation in hematological malignancies and MYD88 overexpression in some solid cancers. The aim of this study was to see whether alterations of MYD88 protein expression and somatic mutation of MYD88 gene are features of common solid cancers. METHODS We analyzed MYD88 mutation in 45 gastric, 45 colorectal, 45 breast, 45 hepatocellular, 45 prostate and 45 lung carcinomas by single-strand conformation polymorphism (SSCP). We also analyzed MYD88 protein expression in 60 gastric, 60 coloretal and 107 prostate carcinomas by immunohistochemistry. RESULTS In the immunohistochemistry results, MYD88 protein was highly expressed in gastric (75%), colorectal (80%) and prostate (83%) cancers. However, MYD88 expression was significantly different among normal tissues (gastric: 58%, colon: 100%, prostate: 86%). MYD88 expression was significantly increased in gastric cancer cells compared with normal cells, whereas it was decreased in colorectal cancer cells compared with normal cells. There were no somatic mutations of the MYD88 gene in gastric, colorectal, breast, hepatocellular, prostate and lung carcinomas. CONCLUSIONS Our data indicate that MYD88 overexpression might be a feature of many solid cancers, but MYD88 expression in normal cells differs depending on the organs. The data suggest that a gain of MYD88 expression in gastric cancers might play a role in cancer pathogenesis by activating oncogenic functions of MYD88.

Mutational analysis of PIK3CA,JAK2,BRAF,FOXL2,IDH1,AKT1 and EZH2 oncogenes in sarcomas

Recent studies have revealed several recurrent mutations in oncogenes that could not only be underlying mechanisms of tumorigenesis, but also be potential targets for cancer therapies. Compared to carcinomas, genetic alterations of sarcomas are relatively unknown. To see whether recurrent oncogenes discovered in non‐sarcomatous malignancies are present in sarcomas as well, we analyzed oncogenes with known mutations in various types of sarcomas. We performed mutational analysis of recurrent mutation sites of PIK3CA (exons 9 and 20), JAK2 (exon 14), BRAF (exon 15), FOXL2 (exon 1), IDH1 (exon 4), AKT1 (exon 3), and EZH2 (exon 16) genes in 108 sarcomas by single‐ strand conformation polymorphism and DNA sequencing. The sarcomas consisted of malignant fibrous histiocytomas, rhabdomyosarcomas, osteosarcomas, malignant peripheral nerve sheath tumors, leiomyosarcomas, synovial sarcomas, liposarcomas, angiosarcomas, chondrosarcomas, and Ewing sarcomas. Overall, we detected the two PIK3CA mutations and one JAK2 mutation (total: 3/108: 2.8%). Two rhabdomyosarcomas (16.7%) and one angiosarcoma (16.7%) harbored the mutations, whereas other sarcomas harbored none. The PIK3CA mutations were novel missense mutations that had not been detected in other cancers. The JAK2 mutation was an intron mutation. This study demonstrated that the somatic mutations of PIK3CA and JAK2 occurred in a small fraction of the sarcomas and that these mutations may not play a principal role in the development of sarcomas.

Somatic mutation of H3F3A, a chromatin remodeling gene, is rare in acute leukemias and non‐Hodgkin lymphoma

To the Editor: Chromatin remodeling is a dynamic modification of chromatin architecture to allow access of genomic DNA to the regulatory transcription machineries, and thereby control gene expression (1). Chromatin remodeling provides wellorchestrated regulation of crucial cell growth and division steps and therefore involved in tumorigenesis (1). Histones are chief components of chromatin, acting as spools around which DNA winds, and play a role in gene regulation. Histones H2A, H2B, H3, and H4 are known as core histones, while histones H1 and H5 are known as the linker histones. Recent studies found that the gene encoding H3 histone family 3A (H3F3A) was somatically mutated in glioblastoma (2–4). Of note, such mutations were recurrent in two amino acid residues (p.K27M, p.G34R/p.G34V) within the histone tails. Mutations of chromatin remodeling genes, including ATRX, UTX, MLL, EP300, ARID1A, and CHD6, have been also identified in solid cancers (5–7). Thus, it is interesting to know whether H3F3A mutation occurs in tumors from hematopoietic system as well. However, to date, the data on H3F3A somatic mutation in other human tumors besides glioblastomas are lacking. For this, we analyzed somatic mutations of H3F3A gene in fresh bone marrow aspirates of 911 hematologic tumors (acute myelogenous leukemias (AML), acute lymphoblastic leukemias (ALL), multiple myelomas, and myelodysplastic syndromes) (Table 1) by polymerase chain reaction (PCR) and single-strand conformation polymorphism (SSCP) assay. Also, we analyzed the gene in paraffin-embedded tissues of 93 non-Hodgkin lymphomas (NHL). Approval was obtained from the Catholic University of Korea, College of Medicine’s institutional review board for this study. Genomic DNA each from tumor cells and normal cells (remission bone marrow cells in the cases of leukemias) were used in this study. For NHL, malignant cells and normal cells were selectively procured from hematoxylin and eosin-stained slides by microdissection as described previously (8). Up to now, all of the H3F3A somatic mutations have been detected at nucleotide sequences encoding amino acids 27 and 34 in exon 2 (2–4). Thus, we analyzed a part of the exon 2 in H3F3A gene by polymerase chain reaction (PCR)-based single-strand conformation polymorphism (SSCP) analysis. Genomic DNA each from tumor cells and normal cells of the same patients were amplified by PCR with a primer pair (5′CAC CCAGGAAGCAACTG -3′ and 5′TTCCTGTTATCCAT CTTTTTGT-3′; product size: 164 base pairs). Radioisotope ([P]dCTP) was incorporated into the PCR products for detection by autoradiogram. Other procedures of the PCRSSCP and DNA sequencing were described in our previous studies (8). SSCP analysis of aberrantly migrating bands on the SSCP led to the identification of one H3F3A somatic mutation in a childhood precursor B-ALL, but no H3F3A mutation was detected in the other tumors. However, the H3F3A mutation detected was a mutation in the intron (c.128 + 30_31delAA) that would not alter the coding sequences. To confirm the SSCP results, we repeated our experiments twice, including PCR and SSCP and to ensure the specificity of the results, and found that the data were consistent. A main concern in cancer research is to identify whether any mutation found in a cancer type is common in other cancer types as well. Because H3F3A is involved in chromatin remodeling that is essential not only for glial cells but also for hematopoietic cells, we analyze H3F3A somatic mutation in hematologic neoplasia. However, the present study detected somatic mutation of H3F3A only in a BALL, but its incidence was very low (0.5% in the childhood ALL). Moreover, this mutation was an intron mutation and not overlapped with the known H3F3A mutations. Taken together, our results suggest that the H3F3A codon 27 and 34 mutations may be a very rare event in human hematopoietic tumors. The discovery of the H3F3A mutation offered Table 1 H3F3A exon 2 mutation analyzed in hematopoietic tumors

Loss of ARID1A expression is uncommon in gastric, colorectal, and prostate cancers.

To the Editor: The SWI–SNF chromatin-remodeling complex regulates gene expression patterns and plays critical roles in development, differentiation, and proliferation of cells (1). AT-rich interactive domain-containing protein 1A (ARID1A) is a large subunit of the SWI–SNF complex and is considered to perform a tumor suppressor function (1). Initially, frequent somatic mutation of ARID1A gene was reported in clear cell and endometrioid carcinomas of ovary (2, 3). In nonovarian tumors, gastric (GC), colorectal (CRC), urinary bladder, and other cancers [pancreas, breast, lung, prostate (PCA), and brain cancers] have been reported to harbor ARID1A mutations (4–6). In the GC andCRC cancers, approximately 50% of the mutations were detected in cancers with microsatellite instability (MSI) (4, 6). In contrast, in pancreas, breast, lung, PCA, and brain cancers, all except one mutation were detected in cancers with stable MSI (MSS) (6). Wang et al. (4) analyzed ARID1A expression in GC and found that loss of ARID1A expression was observed in 83% of the GC with ARID1A mutations, but the loss of expression was also observed in 19% of those without ARID1A mutation. In breast cancers, low expression of ARID1A was observed in 56% of the cases (7). Together, these data suggest that loss of ARID1A expression in cancers may vary depending on tissue types andMSI status. To see whether loss of ARID1A expression is present and dependent on MSI status, we analyzed GC, CRC, and PCA with high MSI (MSI-H) and those withMSS in the present study. There is a mononucleotide repeat (G7 in exon 20) in the coding sequences of human ARID1A gene (http://genome.cse.ucsc.edu/). Frameshift mutation of genes at mononucleotide repeats is a feature of cancers with MSI, and most of the ARID1A mutations in cancers with MSI have been detected in this repeat (4, 6). We analyzed this repeat in ARID1A gene in 91 GC, 100 CRC, and 90 PCA by polymerase chain reaction (PCR) and single-strand conformation polymorphism (SSCP) assay as described previously (8, 9). The GC consisted of 32 GC with high MSI (MSI-H), 14 GC with low MSI (MSI-L), 45 GC with stable MSI (MSS), 40 CRC with MSI-H, 15 GC with MSI-L, and 45 CRC with MSS. All of the PCA were MSS. Genomic DNA each from tumor cells and corresponding normal cells were amplified with seven primer pairs for by PCR. Radioisotope ([P]dCTP) was incorporated into the PCR products for detection by SSCP autoradiogram. PCR-SSCP and DNA sequencing analysis identified aberrant bands in eight of the GC with MSI-H (8/32; 25%), one of the GC with MSI-L (1/14; 7%), and four of the CRC with MSI-H (4/40; 10%), but neither in those with MSS (GC, CRC, and PCA) and the CRC with MSI-L. All the mutations were detected in the G7 repeat and were frameshift mutations (c.5548delG and c.5548dupG). DNA from normal tissue showed no shifts in SSCP, indicating the mutations had risen somatically. To see whether ARID1A protein is altered in the cancers, we used tissue microarray (TMA) blocks that contained paraffin-embedded GC (N = 100), CRC (N = 100), and PCA (N = 100) tissues. Each case has cores representing cancers as well as those representing corresponding normal epithelial tissues. The TMA included GC with MSI-H (N = 20) and CRC with MSI-H (N = 20), which showed ARID1A mutations in three and two cancers, respectively. All other GC, CRC, and PCA were MSS and did not harbor ARID1A mutation. For immunohistochemistry, we used DAKO REAL EnVision System (DAKO, Glostrup, Denmark) with a rabbit polyclonal antibody against human ARID1A (Sigma, Saint Louis, MO, USA; dilution 1/200). Other procedures for immunohistochemistry were described in our previous reports (9). Positive ARID1A immunostaining was observed in 91%, 92%, and 95% of the GC, CRC, and PCA, respectively (Fig. 1). Normal mucosal epithelial cells of stomach and colon, and normal glandular cells of prostate showed ARID1A immunopositivity in all cases. The ARID1A immunostaining, when present, was

Somatic mutation of a tumor suppressor gene BAP1 is rare in breast, prostate, gastric and colorectal cancers

To the Editor, Tumor suppressor genes (TSGs) encode proteins that suppress cell proliferation and growth in normal cells. However, inactivation of TSG by mutations or other alterations disrupts the functions and may lead to cancers (1). BRCA1-associated protein-1 (BAP1), a deubiquitinating enzyme, has been suggested to be a TSG with a role that inhibits cell proliferation and growth via the BRCA1-related growth control pathway (2–4). BAP1 interacts not only with BRCA1, but also with HCF-1 that binds with histone-modifying complex during cell division. Germline mutation of BAP1 predisposes to uveal melanomas, lung cancers, and meningiomas (5). Frequent somatic mutations of BAP1 gene have been reported in melanomas and malignant mesotheliomas (6, 7). Human BAP1 gene resides at chromosome 3p21 that is frequently deleted in many carcinomas (6). However, to data, the data on BAP1 somatic mutation in other common cancers are lacking. To observe whether BAP1 somatic mutations occur in common human carcinomas as well, we analyzed the entire coding region (exon 1–17) and spicing sites of human BAP1 gene by polymerase chain reaction (PCR) and single-strand conformation polymorphism (SSCP) using 19 primer pairs. For this, we used in methacarn-fixed tissues of 45 breast, 45 prostate, 45 gastric, and 45 colorectal carcinomas from Korean patients. In cancer tissues, malignant cells and normal cells were selectively procured from hematoxylin and eosinstained slides using a hypodermic needle affixed to a micromanipulator (8, 9). Radioisotope ([P]dCTP) was incorporated into the PCR products for detection by autoradiogram. The PCR products were subsequently displayed in SSCP gels. After SSCP, direct DNA sequencing reactions were performed in the cancers with mobility shifts in SSCP. Single-strand conformation polymorphism from the carcinomas showed aberrant bands in one colon carcinoma compared with wildtype bands from normal tissues (Fig. 1). Direct DNA sequencing analysis led to identification of a BAP1 somatic mutation (1/45 colon carcinomas; 2.2%). The BAP1 mutation was a missense mutation (c.179G > A), which would result in substitution of Arg by Glu at amino acid residue 60 (p.Arg60Glu) (Fig. 1). This mutation was a novel mutation that had not been detected in earlier studies (6, 7). The colon carcinoma with this mutation was a usual adenocarcinoma with high microsatellite instability (MSI-H) in ascending colon. There was no mutation in the other carcinomas. To confirm the SSCP results, we repeated the experiments twice, including tissue microdissection, PCR and SSCP to ensure specificity of the results, and found that the data were consistent. As a possible inactivation mechanism for BAP1 TSG in human cancers, we analyzed somatic mutation of BAP1 gene in gastric,

Absence of MYD88 gene mutation in acute leukemias and multiple myelomas

To the Editor: Myeloid differentiation primary response gene 88 (MYD88), a myeloid differentiation marker, plays a role in the early genetic responses of myeloid cells to various differentiation and growth inhibitory stimuli (1). MYD88 is an adapter protein for Toll-like receptors that activates NF-jB signaling (2). Recently, Ngo et al. (3) sequenced MYD88 gene in diffuse large B cell lymphoma (DLBCL) and detected somatic mutations of MYD88 gene in 39% of activated B cell-like (ABC) DLBCL. The MYD88 mutations were rare or absent in other DLBCL subtypes, but were detected in 9% of mucosa-associated lymphoid tissue lymphomas (3). Of note, 29% of ABC DLBCL harbored a recurrent mutation (p.L265P) in the MYD88 Toll receptor domain (3). Experimentally, the p.L265P mutant enhanced cell survival by assembling a protein complex of IRAKs, leading to activation of IRAK and NF-jB signaling (3). Other studies also identified the p.L265P mutation in ABC DLBCL and chronic lymphocytic leukemia (4, 5). These data suggest that the MYD88 mutation may be causally implicated in human lymphoid malignancies. However, MYD88 mutation has not been studied in other lymphoid malignancies, including acute lymphoblastic leukemias. As the nomenclature for MYD88 is originated from its functions in myeloid differentiation, it is also possible that MYD88 mutation is related to myeloid-lineage malignancies. For this, we analyzed somatic mutations of MYD88 gene in fresh bone marrow aspirates of 220 acute leukemias (146 acute adulthood leukemias and 74 acute childhood leukemias) and 31 multiple myelomas by polymerase chain reaction (PCR) with specific primers (Table 1) and single-strand conformation polymorphism (SSCP) assay. The adulthood leukemias consisted of 106 AML and 40 acute lymphoblastic leukemias (ALL; 36 pre-B-ALL and 4 T-ALL). The childhood leukemias consisted of 21 AML and 53 ALL (43 pre-B-ALL and 10 T-ALL) (Table 2). The DNA samples from the same patients were extracted from the bone marrows after