EZH2 Mutations in Follicular Lymphoma from Different Ethnic Groups and Associated Gene Expression Alterations

Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-1597

Clinical Cancer Human Cancer Biology Research See related article by Amengual and O'Connor, p. 3047

Shuangping Guo1,6, John K.C. Chan2, Javeed Iqbal6, Timothy McKeithan6, Kai Fu6, Bin Meng3, Yi Pan4, Wah Cheuk2, Donglan Luo5, Ruian Wang1, Weiwei Zhang6, Timothy C. Greiner6, and Wing C. Chan6

Abstract Purpose: Gain-of-function mutations of enhancer of Zeste homolog 2 (EZH2) occur frequently in diffuse large B-cell lymphomas and in follicular lymphomas. However, the frequency of EZH2 mutation in Chinese follicular lymphomas and the potential targets affected by this mutation are unknown. Experimental Design: We determined EZH2 codon 641 mutations in Chinese follicular lymphomas (n ¼ 124) and compared them with Western follicular lymphomas (n ¼ 70) using a sensitive pyrosequencing assay. Gene expression profiling (GEP) was performed to determine differential gene expression between the mutated versus unmutated subgroups, and selected genes were validated using immunohistochemistry. Results: Our results showed similar frequencies of EZH2 codon 641 mutations in Chinese and Western follicular lymphoma cohorts (16.9% vs. 18.6%, c2 test, P ¼ 0.773), including all five reported mutation variants. We observed significant association of EZH2 mutation with low morphologic grade follicular lymphomas (grade 1–2, 23.6% vs. grade 3, 7.7%, c2 test, P ¼ 0.02). EZH2 mutations also showed significant association with BCL2 rearrangement in the Chinese cohort (26.8% vs. 8.8%, c2 test, P ¼ 0.008) and combined cohorts (26.3% vs. 9.1%, c2 test, P ¼ 0.002). GEP analysis identified several genes, including TCF4, FOXP1, TCL1A, BIK, and RASSF6P, with significantly lower mRNA expression (P < 0.01) in mutated cases, and the potential target TCL1A showed consistent results at the protein level. Conclusion: Similar prevalence of EZH2 mutation in two ethnic groups suggests shared pathogenetic mechanisms. The much lower frequency of EZH2 mutation in cases without BCL2 translocation suggests a different pattern of evolution of this subtype of follicular lymphoma. GEP studies showed a set of differentially expressed genes and suggested that EZH2 mutation may help to lock the tumor cells at the germinal center stage of differentiation. Clin Cancer Res; 20(12); 1–9. 2014 AACR.

Introduction BCL2 rearrangement in Chinese follicular lymphoma Follicular lymphoma is a common germinal center B-cell (58.5%) than in Western follicular lymphoma (80%– BCL2 (GCB)-derived non-Hodgkin lymphoma (NHL) exhibiting 90%; ref. 5). The rearrangement results in constitutive a spectrum of morphologies, immunophenotypes, and expression of the antiapoptotic BCL2 protein, which plays genetic aberrations. It is the second most common type of an important role in tumorigenesis of follicular lymphoma; BCL2 NHL representing about 30% of NHL in Western countries however, rearrangement alone is not sufﬁcient for (1, 2); however, the incidence of follicular lymphoma is malignant transformation of B cells. Multiple genetic abnor- lower in Asian countries (3, 4). A recent study of 98 Chinese malities, including mutation of enhancer of Zeste homolog EZH2 follicular lymphoma cases revealed a lower frequency of 2( ), have recently been reported in follicular lymphoma (6–9). The EZH2 gene encodes a histone methyl- transferase that constitutes the catalytic component of the Authors' Afﬁliations: 1Department of Pathology, Xi Jing Hospital, the polycomb repressive complex 2 (PRC2) and is involved in 2 Fourth Military Medical University, Xi'an, Shaan Xi Province; Department repressing gene expression through methylation of histone of Pathology, Queen Elizabeth Hospital, Hong Kong; 3Department of Pathology, Shandong University School of Medicine, Jinan, Shandong H3 on lysine 27 (H3K27). EZH2 codon 641 mutation, the Province; 4Department of Pathology, Tianjin Cancer Hospital; 5Department most common mutation hotspot, is a gain-of-function of Pathology, Guangdong General Hospital and Guangdong Academy of Medical Sciences, Guangzhou, China; and 6Department of Pathology and mutation leading to enhanced trimethylation of histone Microbiology, University of Nebraska Medical Center, Omaha, Nebraska H3K27 and plays an important role in the tumorigenesis of Note: Supplementary data for this article are available at Clinical Cancer GCB-type diffuse large B-cell lymphoma (DLBCL) and Research Online (http://clincancerres.aacrjournals.org/). follicular lymphoma (6–11). Interestingly, highly selective Corresponding Author: Wing C. Chan, City of Hope Medical Center, EZH2 inhibitors have now been developed which is a 1500E Duarte Rd, Duarte, CA 91010. Phone: 626-218-9437; Fax: 626- potential therapeutic strategy for lymphomas with EZH2 301-8842; E-mail: [email protected] activating mutations (12). However, it is unclear whether doi: 10.1158/1078-0432.CCR-13-1597 Chinese follicular lymphoma utilizes similar pathogenetic 2014 American Association for Cancer Research. pathways as their Western counterpart and the frequency of

www.aacrjournals.org OF1

Guo et al.

W.C. Chan, and B. Meng) and re-reviewed by W.C. Chan Translational Relevance and S. Guo, and classified on the basis of the World Health Follicular lymphoma is an indolent, but incurable Organization criteria. These cases showed at least 30% to subtype of non-Hodgkin lymphoma. These tumors har- 40% of the sectional areas consisting of malignant follicles bor t(14;18) translocation in at least 90% of patients. on hematoxylin and eosin (H&E) examination. This study Recently, activating enhancer of Zeste homolog 2 was approved by the Institution Review Boards of the (EZH2) mutations have been found in a significant respective institutions. number of patients with follicular lymphoma. However, the reported incidence has been quite variable and has Construction of tissue microarray and come almost exclusively in Western countries. We have immunohistochemistry developed a sensitive pyrosequencing assay for all muta- Representative areas of follicular lymphoma tissue blocks tion variants affecting codon 641. Our study has shown were selected to prepare the tissue microarray (TMA) by that Chinese and Western patients with follicular lym- examination of the corresponding H&E-stained slides. To phoma exhibit EZH2 Y641 mutation at similar frequen- ensure accurate representation, three areas of each tumor cies, and the mutation is far more frequently associated were selected for core preparation. Each target area on the with the BCL2 rearranged group in both populations. We selected blocks was punched to obtain a 1-mm diameter have also demonstrated by gene expression profiling that tissue core to be placed consecutively on the recipient follicular lymphoma cases with this mutation showed master blocks. TMAs containing representative tumor sec- a set of significantly downregulated genes in EZH2- tions from 124 Chinese and 70 Western follicular lympho- mutant cases. The gene signature provided some insight mas were used for this retrospective analysis. Sections from on the mechanisms of action of abnormally enhanced the TMAs were stained with anti-human BCL2 (clone 124), EZH2 function. EZH2 mutations may play a similar role CD20 (clone L26), BCL6 (clone PG-B6p), (Dako Corp), in lymphomagenesis in different ethnic groups and can and CD10 (clone 56C6), CD21 (clone EP3093), CD3 serve as a biomarker for potential EZH2 targeted therapy. (clone SP7), (Abcam), and anti-TCL1A antibody (clone 1–21, Santa Cruz Biotechnology Inc). Of note, 5 mm thick tissue sections were deparaffinized with xylene, washed with ethanol, and blocked with methanol with 3% hydro- EZH2 mutation in Chinese follicular lymphoma is not gen peroxide. Sections were incubated with the above known. The aim of the present study is to identify the antibodies using Bond polymer refine detection kit and frequency of EZH2 gene codon 641 mutation in Chinese bond IHC Stainer (Leica Microsystems Inc.). For TCL1A follicular lymphoma and the association with the status of staining, GCB cells and Na€ve B cells in the mantle zone of BCL2 expression and rearrangement. We also analyzed the reactive tonsil were used as positive controls. The per- gene expression profiling (GEP) data of follicular lympho- centage of positive tumor cells and the staining intensity mas with and without EZH2 gene mutation to identify were graded from 0 to 3, and the two scores were added to unique associations with EZH2 mutation in follicular form a final score to divide the cases from negative to lymphoma. strong TCL1A expression as described in Supplementary Table S1A.

Materials and Methods FISH Patient materials Full sections and/or TMAs of 124 Chinese follicular One hundred twenty-four Chinese follicular lymphomas lymphomas and 70 Western follicular lymphomas were diagnosed between January 1999 and May 2010 were assessed for the presence of BCL2 rearrangement on 4 mm obtained from Hong Kong and provinces of Shandong, formalin-fixed, paraffin-embedded (FFPE) sections using Tianjing, and Shaan Xi, China, among which 109 cases had the Vysis LSI BCL2 dual color break-apart rearrangement BCL2 rearrangement data detected by interphase FISH. We probe (Abbott Molecular). Cases were scored as BCL2 rear- also included 70 Western follicular lymphomas with BCL2 ranged on the basis of detecting unambiguous probe sep- rearrangement data detected by FISH from the University of aration above the set threshold. Nebraska Medical Center (UNMC; Omaha, NE) for com- parison. As follicular lymphoma with or without t(14;18) DNA extraction, PCR, and pyrosequencing may evolve through different molecular pathways, our Five to 10 mm thick tissue sections of representative initial hypothesis is that EZH2 mutation may be present FFPE tissue blocks were prepared. Genomic DNA was at different frequencies in follicular lymphoma with differ- extractedusingaQiagenDNAFFPETissueKit.Genomic ent t(14;18) status. We enriched both cohorts with a sub- DNA encompassing EZH2 codon 641 was PCR amplified stantial number of follicular lymphoma cases with no BCL2 using forward primer 50-GGCTGGGGGATTTTTATCAA-30 rearrangement (from previous cytogenetic and FISH stud- and biotinylated reverse primer 50- GTTATCAGTGCCT- ies) to test the hypothesis as such cases are uncommon TACCTCT CCA-30. The PCR products were sequenced on a particularly in Western follicular lymphoma. The cases Qiagen Pyromark Q24 instrument using a forward sequenc- were reviewed by expert hematopathologists (J.K.C. Chan, ing primer 50-GAAAAATGAATTCATCTCAG-30.Genomic

OF2 Clin Cancer Res; 20(12) June 15, 2014 Clinical Cancer Research

EZH2 Mutations in Follicular Lymphoma

Table 1. EZH2 codon 641 mutation variants by adding dNTPs in the designated order (G,A,A,T,C,G,A,T,C) during pyrosequencing

Amino acid change Nucleotide change Signal strength by nucleotide Y641 TAC W-AATACT W-AATACT___A4 T2 C0 G0 A2 T0 C2 Y641F TAC>TTC W-AATACT M-AATTCT___A4 T3 C1 G0 A1 T1 C1 Y641S TAC>TCC W-AATACT M-AATCCT___A4 T2 C2 G0 A1 T1 C1 Y641C TAC>TGC W-AATACT M-AATGCT___A4 T2 C0 G1 A1 T0 C2 Y641N TAC>AAC W-AATACT M-AAAACT___A6 T1 C1 G0 A1 T1 C1 Y641H TAC>CAC W-AATACT M-AACACT___ A4 T1 C1 G0 A2 T0 C2

Abbreviations: W, wild-type; M, mutant; Y, tyrosine.

DNA of cell line SU-DHL-6, harboring an EZH2 codon dataset] were chosen for analysis (Data available in the Y641N mutation, and normal tonsil lacking EZH2 muta- GEO database: GSE 55267.) tion, was mixed with SU-DHL-6 DNA constituting 5%, 10%, 20%, 30%, 50%, and 70% of the total DNA, respec- Statistical analysis tively. The mixture was used as template to determine Statistical significance of differences in EZH2 mutation the sensitivity of pyrosequencing to detect the mutant. frequency between lymphoma groups with or without The sensitivity of the assay was determined to be at 5% BCL2 rearrangement and expression was assessed by c2 test. of the genomic DNA of the cell line SU-DHL-6. According Statistical significance was defined as P < 0.05. to the principle of pyrosequencing, the light emission in the process of sequencing is proportional to the quantity Results of deoxynucleotide triphosphate (dNTP) incorporated Patient characteristics into the template. dNTPs were dispensed in a novel order One hundred and twenty-four Chinese follicular lym- of G, A, A, T, C, G, A, T, and C, based on the sequence of phoma patients were included in this study. The median age wild-type EZH2 codon 641 and five mutation variants. of these patients was 59.5 years (range, 26–89 years), and With this order of addition, at least one addition results in included 64 (51.6%) males and 60 (48.4%) females. These nucleotide incorporation for the mutant but not the wild- cases were enriched in BCL2 translocation negative cases so type allele, allowing all five mutations to be identified and that only 56 harbored a BCL2 rearrangement and 68 lacked detected with maximal sensitivity (Table 1). All detected a BCL2 rearrangement. Among the 70 Western follicular mutations were validated by repeating the assay. lymphomas, the median age was 60 (range, 35–88 years), and there were 33 (47.1%) male and 37 (52.9%) female GEP data analysis patients. This cohort was also enriched in BCL2 wild-type The GEP data from 68 cases of UNMC follicular lym- cases, 39 harbored a BCL2 rearrangement and 31 lacked a phoma were analyzed to find genes differentially expressed BCL2 rearrangement. There were no significant differences between follicular lymphomas with or without EZH2 muta- in clinical features between Chinese and Western follicular tion. We used HG-U133 plus 2 arrays (Affymetrix Inc.) for lymphomas. On histologic morphologic assessment, 72 of our analysis, and the methods for isolation and processing 124 (58%) in Chinese cohort and 39 of 70 cases (55.7%) in of RNA and acquisition of GEP raw data have been the Western cohort were of grade 1–2. described previously (13). The raw data were uploaded in BRB-Array Tools (version 3.7.0) for normalization, and the Frequencies of EZH2 codon 641 mutations analysis was supervised by the EZH2 gene mutation status. We observed EZH2 codon 641 mutations in 16.9% (21 of Genes were selected at a significance of P < 0.01 and >1.5- 124) Chinese follicular lymphoma and 18.6% (13 of 70) fold difference in expression levels between the two groups. Western follicular lymphomas using pyrosequencing and To enhance the identification of differential gene expression the results are summarized in Table 2. No significant dif- in follicular lymphoma tumor cells, only follicular lym- ference in the frequency of EZH2 mutation in follicular phoma cases with a high B-cell content [enriched B-cell lymphomas between Chinese and the Western patients was signature of >2-fold above the mean of the B-cell signature observed [16.9% (21/124) versus 18.6% (13/70), c2 test, (pan B-cell markers) in the entire follicular lymphoma P ¼ 0.773]. Representative results of pyrosequencing of

www.aacrjournals.org Clin Cancer Res; 20(12) June 15, 2014 OF3

Guo et al.

Table 2. EZH2 mutation and BCL2 rearrangement and expression in follicular lymphomas

Classiﬁcation EZH2 mutation BCL2 rearranged BCL2 expression All Chinese FL (n ¼ 124) 21 (16.9%) 56 (45.2%) 76 (61.3%) Grade 1–2(n ¼ 72)a 17 (23.6%) 47 (65.3%) 57 (79.2%) Grade 3 (n ¼ 52) 4 (7.7%) 9 (17.3%) 19 (36.5%) All Western FL (n ¼ 70) 13 (18.6%) 39 (55.7%) 51 (72.9%) Grade 1–2(n ¼ 39)b 10 (25.6%) 30 (76.9%) 33 (84.6%) Grade 3 (n ¼ 31) 3 (9.7%) 9 (29%) 18 (58.1%)

Abbreviation: FL, follicular lymphoma. aIncludes grades 1 (n ¼ 33) and 2 (n ¼ 39). bIncludes grades 1 (n ¼ 17) and 2 (n ¼ 22).

wild-type EZH2 codon 641 and five mutant variants are Chinese follicular lymphoma with frequencies similar to shown in Fig. 1. The five mutation variants resulted in the those of Western follicular lymphoma. Y641F was the most amino acid change from tyrosine (Y) 641 to phenylalanine common mutation, accounting for 38% of mutants, with (F), serine (S), cysteine (C), asparagine (N), and histidine decreasing frequencies from Y641S (24%), Y641H (14%), (H) as shown in Table 1. These five variants were detected in Y641C (14%) to Y641N (10%).

A B Normal tonsil, wild-type EZH2 codon Y641 (A4 T2 C0 G0 A2 T0 C2) FL60, EZH2 codon Y641N (A6 T1 C1 G0 A1 T1 C1)

250 200 200 150 150 100 100 50 50 0 0

–50 –50 ESGAATCGATC ESGAATCGATC C 5 D FL51, EZH2 codon Y641F (A4 T3 C1 G0 A1 T1 C1) FL66, EZH2 codon Y641S (A4 T2 C2 G0 A1 T1 C1)

200 200 150 150

100 100 50 50 0 –50 0 ES GAA TCGA TC ESGAATCGATC E F FL45, EZH2 codon Y641C (A4 T2 C0 G1 A1 T0 C2) FL62, EZH2 codon Y641H (A4 T1 C1 G0 A2 T0 C2) 250

200 200

150 150

100 100 50 50 0 0 ES GAATCGATC ESGAATCGATC 5

Figure 1. Pyrosequencing of EZH2 codon 641 mutation in representative follicular lymphomas (FL). A, normal tonsil, wild-type EZH2 codon Y641 (A4, T2, C0, G0, A2, T0, C2). B–F, dotted red lines show the wild-type pattern. Filled arrows show dNTP additions that yield incorporation into the mutant but not the wild- type allele. B, FL60, EZH2 codon Y641N (A6, T1, C1, G0, A1, T1, C1). C, FL51, EZH2 codon Y641F (A4, T3, C1, G0, A1, T1, C1). D, FL66, EZH2 codon Y641S (A4, T2, C2, G0, A1, T1, C1). E, FL45, EZH2 codon Y641C (A4, T2, C0, G1, A1, T0, C2). F, FL62, EZH2 codon Y641H (A4, T1, C1, G0, A2, T0, C2).

OF4 Clin Cancer Res; 20(12) June 15, 2014 Clinical Cancer Research

EZH2 Mutations in Follicular Lymphoma

P = 0.008 30.% P = 0.163 P = 0.002 26.8% P = 0.155 26.3% P = 0.042 25.6% P = 0.007 25.% 23.5% 22.4% 22.8%

20.%

Figure 2. EZH2 mutation in follicular 15.% lymphomas (FL) by BCL2 status. EZH2 mutations were signiﬁcantly 9.7% more frequent in cases with 8.8% 9.1% 10.% 8.3% rearrangements by FISH and 7.5% expression of BCL2 by IHC in 5.3% Chinese follicular lymphoma and EZH2 mutation frequency 5.% combined Chinese and Western follicular lymphoma. 0.% Chinese FLs Western FLs Chinese and Western FLs (n = 124) (n = 70) (n = 194) BCL2 rearranged BCL2 wild-type BCL2-positive BCL2-negative

Association between EZH2 mutation and morphologic associated with BCL2 rearrangement in the combined Chi- follicular lymphoma grades nese and Western cases (26.3% vs. 9.1%, c2 test, P ¼ The EZH2 mutation frequency was related to the mor- 0.002; Fig. 2). EZH2 mutation was also significantly asso- phologic follicular lymphoma grades, with higher fre- ciated with the BCL2 protein expression in Chinese (22.4% quency in low morphologic grade follicular lymphomas vs. 8.3%, c2 test, P ¼ 0.042) and the combined follicular in both Chinese (grade 1–2, 23.6% vs. grade 3, 7.7%, c2 lymphoma cohorts (22.8% vs. 7.5%, c2 test, P ¼ 0.007, test, P ¼ 0.02) and Western population (grade 1–2, 25.6% respectively). The percentage of EZH2 mutation was vs. grade 3, 9.7%, c2 test, P ¼ 0.09; Table 2).The distri- enriched in BCL2 protein-positive cases in Western cohort; bution of EZH2 mutation in different morphologic grades however, it did not reach statistical significance (23.5% vs. of Western follicular lymphoma was similar to that of 5.3%, c2 test, P ¼ 0.155; Fig. 2). Chinese follicular lymphoma. The percentage of EZH2 mutation in grade 1 (7/33; 21.2%) and grade 2 (10/39; Differential gene expression analysis 25.6%) follicular lymphoma was quite similar in Chinese Our analysis focused on follicular lymphoma cases with follicular lymphomas, and similar trend was observed in high B-cell contents to facilitate the detection of changes Western follicular lymphomas. related to the tumor cells. A set of genes was significantly downregulated in mutated cases. Among these downregu- Association between EZH2 mutation and BCL2 lated genes, a number of them, such as TCF4, FOXP1, translocation TCL1A, BIK,andRASSF6P were also marked repressed As expected from our case selection, BCL2 rearrange- compared with normal centroblasts (Fig. 3). There was ments were detected in 56 of 124 (45.2%) and 39 of 70 also a set of genes such as PTPN22, which was upregulated (55.7%) in Chinese and Western follicular lymphomas, in mutant cases probably as secondary changes. We also respectively. BCL2 expressions were observed in 76 of compared our signature with the specific gene signature 124 (61.3%) and 51 of 70 (72.9%) in Chinese and Western derived previously using an EZH2-specific inhibitor follicular lymphomas, respectively. The positivity of BCL2 (GSK126) on GCB cell lines with EZH2 mutation (12). rearrangement and expression was higher in grade 1–2 than Notably, this signature was significantly (P < 0.02) down- in grade 3 follicular lymphomas in the two cohorts (Table regulated in cases with EZH2 mutation compared with 2). The frequency of EZH2 mutation was significantly wild-type cases (Fig. 3). higher in follicular lymphomas with a BCL2 rearrangement than in those lacking a BCL2 rearrangement in Chinese Downregulation of TCL1A in EZH2-mutated follicular follicular lymphoma (26.8% vs. 8.8%, c2 test, P ¼ 0.008). lymphomas However, the difference observed in Western follicular One of the target genes that were reported to be upregu- lymphoma did not reach statistical significance (25.6% vs. lated upon treatment of EZH2-mutant cell lines by a specific 9.7%, c2 test, P ¼ 0.163). EZH2 mutation was significantly inhibitor of EZH2-mutant protein (GSK126) is TCL1A (12).

www.aacrjournals.org Clin Cancer Res; 20(12) June 15, 2014 OF5

Guo et al.

CYSLTR1 TOX2 CCDC50 TCF4 Normal B-cell SMAD1 CCL17 Mean subsets PTPN1 FAIM3 LOC221442 Mutant Wild-type DOPEY2 MYO7A TFPI2

EZH2 EZH2 Mut WT CB CC Naïve CCL22 PLCL2 TRAM2 Figure 3. Differential gene P2RY10 TCF4 expression between EZH2 mutant RRAS2 TCF4 SC4MOL CNKSR2 CYP51A1 PVALB versus wild-type follicular PTPN22 ADD3 PTPN22 TFPI2 lymphoma cases (P < 0.01; SPAG1 LOC374443 DHCR24 FZD3 >1.5-fold change). Only cases with PTPN22 MYO1D TXLNB TCF4 IRAK2 TCF4 high B-cell content were included RANBP2 FOXP1 GRAMD1C TCF4 in the analysis. , EZH2 target gene SYT11 RASSF6 BTNL9 FOXP1 signature was derived in GCB cell ILDR1 RASGRP2 TESK2 IGHM BRIP1 TCL1A lines with mutant EZH2 gene GATM TCL1A KCNH8 BIK treated with EZH2-speciﬁc INPP5F E2F5 USP12 TAGAP HS2ST1 inhibitor (GSK126). The authors TEX9 INPP5F have included only upregulated INPP5F Pan B-cell signature (mean) RRAS2 genes (35 probe set) in at least four C2orf88 BFSP2 of ﬁve cell lines after inhibition of KIAA1274 EZH2 mRNA mutant EZH2 (12). CB, centroblast, EZH2 target gene signature (mean)* CC, centrocyte.

Relative level of expression (× median value)

0.25 0.5 1 2 4

Consistent with this observation, we observed that this gene used. Pyrosequencing was used in the present study because was downregulated in follicular lymphoma cases carrying it is more sensitive than conventional Sanger sequencing EZH2 Y641 mutation compared with our wild-type. There and is widely used clinically to detect point mutations in have been reports suggesting variable expression of TCL1A FFPE tissues. The assay was able to detect 5% of the mutant in follicular lymphoma. We evaluated TCL1A expression at genomic DNA of the cell line SU-DHL-6, which harbors an the protein level and observed that B cells in the mantle EZH2 codon Y641N mutation. Assays with lower sensitivity zone of reactive tonsil showed strong expression of TCL1A, may miss the mutation in cases with low tumor content, whereas GCB cells showed moderate expression in both which are not uncommon in follicular lymphomas. We cytoplasm and nucleus. Of the follicular lymphoma cases detected a similar frequency of EZH2 codon 641 mutations without EZH2 Y641 mutation, the majority of the follicular in Chinese follicular lymphoma [21 of 124 (16.9%)] and in lymphoma cases (67.2%, 39 of 58) showed strong homog- Western follicular lymphoma [13 of 70 (18.6%)]. All of the enous expression of TCL1A, whereas 10.3% (6 of 58) were five reported mutations, Y641F, Y641S, Y641C, Y641H, moderate, and 22.4% (13 of 58) were either weak or and Y641N, were detected in Chinese follicular lympho- negative. Interestingly, none of the follicular lymphoma ma with frequencies similar to those previously reported cases with EZH2 Y641 mutation showed strong expression; in Western follicular lymphoma (7, 9–11). Y641F was the it was generally variable from moderate (6 cases) to weak (1 most common mutation, accounting for 38% of mutants. case), or negative (3 cases) in 10 mutant cases. The results It has been reported that all of the five mutant proteins are summarized in Supplementary Table S1B, and repre- confer a gain-of-function phenotype and that Y641N is sentative images are shown in Fig. 4. the most potent at generating H3K27me3, whereas Y641H is the least potent (8). The BCL2 rearrangement is a genetic hallmark of follic- Discussion ular lymphomas, but a small percentage of follicular lym- EZH2 is the catalytic subunit of the PRC2 and is involved phoma cases lack this rearrangement; the latter are much in repressing gene expression through methylation of lysine more frequent in the Chinese population (5). It is possible 27 of histone H3. The EZH2 codon 641 mutation is a gain- that cases with or without BCL2 rearrangement may have of-function mutation leading to enhanced trimethylation different pathogenetic mechanisms that may be reflected in of histone H3K27 (6). EZH2-mediated epigenetic silencing different profiles of genetic mutations. A recent study in in GCB cells contributes to proliferation and lymphoma- Western patients indeed found a marked difference in EZH2 genesis (14). mutation frequency between the two subgroups (7). We The frequency of EZH2 mutations varied from 7.2% to investigated the association of EZH2 mutation and the 22% (Supplementary Table S2) in prior studies (7, 9–10); status of BCL2 rearrangement in follicular lymphoma, the variation is probably due to the different assay methods including a substantial number of cases with no BCL2

OF6 Clin Cancer Res; 20(12) June 15, 2014 Clinical Cancer Research

EZH2 Mutations in Follicular Lymphoma

Figure 4. Expression of TCL1A in follicular lymphomas. A, expression of TCL1A in GC and mantle zone cells in a reactive tonsil (40 magnification). B, a follicular lymphoma case without EZH2 Y641 mutation showing strong TCL1A expression (40 magnification). C, follicular lymphoma cases with EZH2 Y641 mutation showing moderate expression of TCL1A (40 magnification). D, mutant with weak TCL1A expression (20 magnification). E, mutant negative for TCL1A, scattered T cell and mantle zone B cells were strongly positive (40 magnification). rearrangement. The frequency of EZH2 mutation was sig- tion in cases negative for BCL2 rearrangement (7) is prob- nificantly associated with BCL2 rearrangement in Chinese ably spurious and due to the low number of cases investi- follicular lymphoma, but did not reach statistical signifi- gated. The low frequency of EZH2 mutation in follicular cance in Western follicular lymphoma, probably due to the lymphomas without BCL2 rearrangement and in grade 3 small number of cases. However, overall (combining both follicular lymphoma suggests the preferential use of differ- cohorts) EZH2 mutation was significantly associated with ent pathways to promote the malignant transformation of BCL2 rearrangement. The frequency of EZH2 codon 641 GCB lymphocytes to follicular lymphoma. mutation in Chinese follicular lymphoma with BCL2 rear- Although it has been demonstrated that the EZH2 codon rangement was very close to that reported in a previously 641 mutation enhances trimethylation of histone H3K27, study (28%), using a sensitive detection technique (7). The the genomic targets that are affected by the mutation and mutation frequency in Chinese follicular lymphoma lack- thus promote lymphomagenesis have not been identified. ing BCL2 rearrangement was low and comparable with Our GEP data analysis revealed a set of differentially similar patients in Western countries but far higher than expressed genes; among them TCF4, FOXP1, RASSF6, reported previously (7). As EZH2 codon 641 mutation was TCL1A, and BIK were downregulated, compared with both first identified in a grade 1 follicular lymphoma lacking wild-type follicular lymphoma and normal centroblast. BCL2 rearrangement (9), the reported lack of EZH2 muta- Both TCF4 and FOXP1 were consistently upregulated in

www.aacrjournals.org Clin Cancer Res; 20(12) June 15, 2014 OF7

Guo et al.

ABC-DLBCL compared with GBC-DLBCL (15). This obser- mutation is significantly higher than in normal B cells vation suggests that one of the functions of EZH2 mutation (na€ve B cells, centrocytes, and centroblasts) and in follic- may be locking the cells at the GCB cell stage of differen- ular lymphomas without EZH2 mutation. PTPN22 is con- tiation by downregulating the expression of crucial genes sidered to be one of the principal negative regulators of for further differentiation. TCL1A is also of particularly antigen receptor signaling in normal B and T lymphocytes, interest; it was initially identified as an oncogene in T-cell but it can also positively regulate the kinase AKT, providing prolymphocytic leukemias, but may also be expressed in B- a powerful survival signal. B-cell receptor signaling in fol- cell lymphomas, including follicular lymphoma (16). licular lymphoma may need to be finely tuned to avoid TCL1A serves as a coactivator of the AKT/mTOR pathway further differentiation, and simultaneous activating the AKT and has also been reported as an inhibitor of de novo DNA pathway may optimize cell survival, perhaps compensating methylation in B-cell chronic lymphocytic leukemia (17), for the effects of TCL1 downregulation on the AKT/mTOR so downregulation of TCL1A may have diverse effects, pathway (20). including epigenetic alterations. We further confirmed the In conclusion, EZH2 codon 641 mutations occur at frequent expression of TCL1A protein in follicular lympho- similar frequency in Western and Chinese follicular lym- mas. It was revealed that follicular lymphomas displayed phoma and more often associated with low-grade cases. The striking variability in the intensity of TCL1A staining and mutations are far more frequent in the BCL2 rearranged there is a correlation with EZH2 Y641 mutation. Thirty-nine group. This mutation, probably plays a similar role in the cases showing strong positivity for TCL1A harbored no pathogenesis of follicular lymphomas in both Chinese and EZH2 Y641 mutation, whereas the expression of TCL1A in Western patients and may serve as a marker for potential the mutant cases varied from moderate to weak, or negative, EZH2 targeted therapy. and none showed strong positivity. The immunohistochemical results are in keeping with the GEP analysis and Disclosure of Potential Conflicts of Interest may explain the reported variability of TCL1A expression in No potential conflicts of interest were disclosed. follicular lymphoma. Interestingly, it was also noted that the expression of TCL1A increased in DLBCL cell lines Authors' Contributions harboring EZH2 mutation when treated with the EZH2- Conception and design: J. Iqbal, W.C. Chan Development of methodology: S. Guo, T. McKeithan, T.C. Greiner, W.C. specific inhibitor GSK126 (12). This observation is consis- Chan tent with our finding and suggests that TCL1A is likely Acquisition of data (provided animals, acquired and managed patients, targeted by EZH2 mutation in follicular lymphoma. We provided facilities, etc.): J.K.C. Chan, J. Iqbal, K. Fu, B. Meng, Y. Pan, W. Cheuk, D. Luo, T.C. Greiner also observed that specific gene signature generated by Analysis and interpretation of data (e.g., statistical analysis): S. Guo, EZH2-mutant inhibitor (GSK126; ref. 12) showed expected J.K.C. Chan, J. Iqbal, K. Fu, R. Wang, W. Zhang, T.C. Greiner correlation in mutant follicular lymphoma cases, suggesting Writing, review, and/or revision of the manuscript: S. Guo, J.K.C. Chan, J. Iqbal, T. McKeithan, K. Fu, T.C. Greiner, W.C. Chan that such inhibitors may provide a rational and viable Administrative, technical, or material support (i.e., reporting or orga- therapeutic approach for these patients. Downregulation nizing data, constructing databases): B. Meng, Y. Pan, D. Luo, W.C. Chan of BIK (Bcl2-interacting killer), a proapoptotic gene, is also of interest as 40% of cases of multiple myeloma were Acknowledgments reported to have a methylated BIK CpG island (18). Another The authors thank Martin Bast for the clinical data collection and Debra Lytle for technical assistance. downregulated gene is RASSF6, a Ras-association family The costs of publication of this article were defrayed in part by the (RASSF) member that has been shown to be frequently payment of page charges. This article must therefore be hereby marked epigenetically inactivated by promoter CpG island hyper- advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate methylation in childhood leukemias (19). this fact. PTPN22 Among the upregulated genes, is intriguing. The Received June 14, 2013; revised January 8, 2014; accepted February 2, expression of PTPN22 in follicular lymphomas with EZH2 2014; published OnlineFirst March 14, 2014.

References 1. Laurini JA, Perry AM, Boilesen E, Diebold J, Maclennan KA, Muller-€ 5. Pan Y, Meng B, Sun B, Guan B, Liang Y, Wang H, et al. Frequencies of Hermelink HK, et al. Classiﬁcation of non-Hodgkin lymphoma in BCL2 and BCL6 translocations in representative chinese follicular Central and South America: a review of 1028 cases. Blood 2012; lymphoma patients: morphologic, immunohistochemical, and FISH 120:4795–801. analyses. Diagn Mol Pathol 2012;21:234–40. 2. Anderson JR, Armitage JO, Weisenburger DD. Epidemiology of the 6. Yap DB, Chu J, Berg T, Schapira M, Cheng SW, Moradian A, et al. non-Hodgkin's lymphomas: distributions of the major subtypes differ Somatic mutations at EZH2 Y641 act dominantly through a mecha- by geographic locations. Non-Hodgkin's Lymphoma Classiﬁcation nism of selectively altered PRC2 catalytic activity, to increase H3K27 Project. Ann Oncol 1998;9:717–20. trimethylation. Blood 2010;117:2451–9. 3. Aoki R, Karube K, Sugita Y, Nomura Y, Shimizu K, Kimura Y, et al. 7. Ryan RJ, Nitta M, Borger D, Zukerberg LR, Ferry JA, Harris NL, et al. Distribution of malignant lymphoma in Japan: analysis of 2260 cases, EZH2 codon 641 mutations are common in BCL2-rearranged germinal 2001–2006. Pathol Int 2008;58:174–82. center B cell lymphomas. PLOS ONE 2011;6:e28585. 4. Huh J. Epidemiologic overview of malignant lymphoma. Korean 8. Sneeringer CJ, Scott MP, Kuntz KW, Knutson SK, Pollock RM, Richon J Hematol 2012;47:92–104. VM, et al. Coordinated activities of wild-type plus mutant EZH2 drive

OF8 Clin Cancer Res; 20(12) June 15, 2014 Clinical Cancer Research

EZH2 Mutations in Follicular Lymphoma

tumor-associated hypertrimethylation of lysine 27 on histone H3 15. Care MA, Barrans S, Worrillow L, Jack A, Westhead DR, Tooze RM. A (H3K27) in human B-cell lymphomas. PNAS 2010;107:20980–85. Microarray platform-independent classiﬁcation tool for cell of origin 9. Morin RD, Johnson NA, Severson TM, Mungall AJ, An J, Goya R, et al. class allows comparative analysis of gene expression in diffuse large Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphoma. PLOS ONE 2013;8:e55895. B-cell lymphomas of germinal-center origin. Nat Genet 2010;42: 16. Aggarwal M, Villuendas R, Gomez G, Rodriguez-Pinilla SM, Sanchez- 181–5. Beato M, Alvarez D, et al. TCL1A expression delineates biological and 10. Bod€ or€ C, O'Riain C, Wrench D, Matthews J, Iyengar S, Tayyib H, et al. clinical variability in B-cell lymphoma. Modern Pathol 2009;22:206–15. EZH2 Y641 mutations in follicular lymphoma. Leukemia 2011;25: 17. Palamarchuk A, Yan PS, Zanesi N, Wang L, Rodrigues B, Murphy M, 726–29. et al. Tcl1 protein functions as an inhibitor of de nove DNA methylation 11. Morin RD, Mendez-Lago M, Mungall AJ, Goya R, Mungall KL, Corbett in B-cell chronic lymphocytic leukemia (CLL). PNAS 2012;109: RD, et al. Frequent mutation of histone-modifying genes in non- 2555–60. Hodgkin lymphoma. Nature 2011;476:298–303. 18. Hatzimichael E, Dasoula A, Kounnis V, Benetatos L, Lo Nigro C, 12. McCabe MT, Ott HM, Ganji G, Korenchuk S, Thompson C, Van Aller Lattanzio L, et al. Bcl2-interacting killer CpG methylation in multiple GS, et al. EZH2 inhibition as a therapeutic strategy for lymphoma with myeloma: a potential predictor of relapsed/refractory diseaswith ther- EZH2-activating mutations. Nature 2012;492:108–12. apeutic implications. Leuk lymphoma 2012;53:1709–13. 13. Dybkaer K, Iqbal J, Zhou G, Geng H, Xiao L, Schmitz A, et al. Genome 19. Hesson LB, Dunwell TL, Cooper WN, Catchpoole D, Brini AT, Chiar- wide transcriptional analysis of resting and IL2 activated human natural amonte R, et al. The novel RASSF6 and RASSF10 candidate tumour killer cells: gene expression signatures indicative of novel molecular suppressor genes are frequently epigenetically inactivated in child- signaling pathways. BMC Genomics 2007;8:230. hood leukaemias. Mol Cancer 2009;8:42. 14. Velichutina I, Shaknovich R, Geng H, Johnson NA, Gascoyne RD, 20. Negro R, Gobessi S, Longo PG, He Y, Zhang ZY, Laurenti L, et al. Melnick AM, et al. EZH2-mediated epigenetic silencing in germinal Overexpression of the autoimmunity-associated phosphatase center B cells contributes to proliferation and lymphomagenesis. PTPN22 promotes survival of antigen-stimulated CLL cells by selec- Blood 2010;116:5247–55. tively activating AKT. Blood 2012;119:6278–87.

www.aacrjournals.org Clin Cancer Res; 20(12) June 15, 2014 OF9

EZH2 Mutations in Follicular Lymphoma from Different Ethnic Groups and Associated Gene Expression Alterations

Shuangping Guo, John K.C. Chan, Javeed Iqbal, et al.

Clin Cancer Res Published OnlineFirst March 14, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-13-1597

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2014/03/14/1078-0432.CCR-13-1597.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2014/05/22/1078-0432.CCR-13-1597. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.