The Diagnostic Significance of Myf-3 Hypermethylation in Malignant

Leukemia (2001) 15, 583–589  2001 Nature Publishing Group All rights reserved 0887-6924/01 $15.00 www.nature.com/leu The diagnostic signiﬁcance of Myf-3 hypermethylation in malignant lymphoproliferative disorders JME Taylor,1,2 PH Kay2 and DV Spagnolo1

1The Western Australian Centre for Pathology and Medical Research, Nedlands; and 2Department of Pathology, The University of Western Australia, Nedlands, Western Australia, Australia

Deregulated methylation of cytosine in DNA is a frequent find- tiation and proliferation by controlling gene expression.4 Cyto- ing in malignancy that is reflected by general genomic hypome- sine bases usually within a 5′ CpG 3′ dinucleotide are methyl- thylation and regional hypermethylation that includes the myogenic gene Myf-3. In this study of 198 DNA samples from ated by one or more of a series of 5-methyltransferase 186 patients with a wide range of lymphoproliferative disorders enzymes called DNMT1, 2, 3a and 3b. The activities of (LPD), the methylation status of Myf-3 was assessed to evalu- DNMT1, 3a and 3b for example, are increased in many types ate its significance in the diagnosis of malignant LPD. DNA was of neoplastic cells.5,6 Increased activity of the cytosine methyl- digested with the restriction endonucleases HpaII and MspI, ating systems in neoplastic cells is associated with hyperme- and using the Southern blot (SB) technique, the size and den- thylation of specific regions of the genome including the short sity of fragments that hybridized with a Myf-3 probe were used arm of chromosome 117 which includes the myogenic gene to assign the methylation status. None of the samples from 45 8 9 patients from a wide age range with benign LPDs had evidence Myf-3 (Myo-D1 ). of altered Myf-3 methylation and there was no age-related Myf-3 is rich in 5′ CpG 3′ dinucleotides and it is hypome- methylation change. By contrast, 115/123 (93%) of samples thylated in non-malignant cells. By contrast, in some types of from patients with non-Hodgkin lymphoma (NHL) or lymphoid neoplastic cells it has been shown to be hypermethylated. In leukemia had increased Myf-3 methylation. There was no our early studies we showed that Myf-3 is hypermethylated in methylation alteration in 22/24 (92%) of samples from patients DNA from a small number of malignant LPDs but not from with Hodgkin lymphoma (HL), nor in five of six samples from 10 LPDs that had atypical histopathologic features which were not benign LPDs. Moreover, our investigations into the methyl- diagnostic of lymphoma, while the remaining sample of atypi- ation status of Myf-3 in breast and colon neoplasms showed cal LPD had hypermethylated Myf-3 fragments. There was an that it may be useful as a prognostic indicator.11–13 association between increasing Myf-3 methylation and higher This study of the methylation status of Myf-3 in DNA from histopathologic grade of malignancy within specific lymphoma benign and malignant lymphoid tissues from a large number categories. It is concluded that the detection of increased Myf- 3 methylation is a sensitive and specific test of malignancy of patients with a variety of LPDs was undertaken to determine which may complement other molecular methods that are cur- its diagnostic significance in lymphoproliferative disorders. rently used for the assessment of clonality. It may be of parti- cular diagnostic use in natural killer (NK) and null cell malignancies for which other indicators of clonality are lacking. Materials and methods Furthermore, methylation status may prove to be of potential prognostic value. Leukemia (2001) 15, 583–589. Keywords: diagnosis; DNA methylation; leukemia; lymphoma; DNA isolation Myf-3 Cases for study were selected retrospectively from archival freshly frozen lymphoid tissues or from EDTA anticoagulated Introduction peripheral blood leukocytes and bone marrow aspirates. All solid tissues were stored at −80°C prior to DNA extraction. DNA was isolated from each sample using a standard The application of molecular techniques has greatly advanced 14 the ability to distinguish between malignant and non-malig- phenol/chloroform, ethanol precipitation method. DNA was nant lymphoproliferative disorders (LPDs). For example, extracted from 198 samples from 186 patients with benign or Southern blotting (SB) and polymerase chain reaction (PCR)- malignant LPDs (one patient had metachronous NHL and HL). based techniques are routinely used to detect monoclonal DNA samples were obtained from 134 lymph nodes, 11 rearrangements of immunoglobulin (Ig) and T cell receptor splenic samples, eight peripheral blood leukocyte specimens, (TCR) genes which in an appropriate clinical setting are seven tonsils, six bone marrow aspirates, and 32 other extra- characteristic of malignant LPDs.1–3 Furthermore, some lym- nodal tissues. phoma categories have specific chromosomal translocations that occur at high frequency. However, these current markers of monoclonality have limited prognostic value and cannot Case classification be used to assess the clonality of NK and null cell proliferations where neither Ig or TCR genes are rearranged. Three groups of well-characterised cases of benign, malignant and atypical LPDs were used. Classifications were confirmed Cytosine methylation is a part of the heritable epigenetic 1,2 3 machinery which is thought to influence cellular differen- immunophenotypically and genotypically by PCR and SB based methods as previously described. The diagnosis of malignant B and T cell LPDs was supported by the detection of clonal antigen receptor gene rearrangements. Malignant Correspondence: JME Taylor, Tissue Pathology Division, The Western LPDs were categorized according to the proposed WHO Australian Centre for Pathology and Medical Research (PathCentre), 15 Locked Bag 2009, Nedlands, Western Australia, Australia 6909; classification (Table 1). Fax: 08 9346 4122 Follicular lymphoma (FL) and diffuse large B cell lymphoma Received 20 October 2000; accepted 19 December 2000 (DLBCL) samples were sub-grouped according to cytologic Myf-3 hypermethylation in malignant LPD JME Taylor et al 584 Table 1 Myf-3 methylation status and classification of lymphoproliferative disorders in the study15

Diagnostic category n Hypermethylation Minor change No change

Benign LPD: 45 samples from 45 patients 0 (0) 0 (0) 45 (100) Non-speciﬁc reactive hyperplasia 37 Granulomatous lymphadenitis 7 Lymphoid hamartoma 1

Atypical lymphoid hyperplasia: 6 samples from 6 patients 6 1 (17) 0 (0) 5 (83)

B cell neoplasms: 92 samples from 84 patients 71b (79) 16 (18) 3 (3) Precursor B lymphoblastic leukemia/lymphoma 6 B cell small lymphocytic lymphoma 10 B cell prolymphocytic leukemia 1 Splenic marginal zone B cell lymphoma 1 Plasma cell myeloma 2 Extra-nodal marginal zone B cell lymphoma of MALT type 7 Nodal marginal zone B cell lymphoma 1 Follicular lymphoma 34 [31] Mantle cell lymphoma 10 [9] Diffuse large B cell lymphoma 17 Atypical Burkitt lymphoma 3

T and NK cell neoplasms: 31 samples from 29 patients 19 (63) 6a (20) 5 (17) Precursor T lymphoblastic leukemia/lymphoma 7 T cell prolymphocytic leukemia 1 T cell granular lymphocytic leukemia 4 [3] Extra-nodal NK/T cell lymphoma, nasal type 2 Hepatosplenic γδ T cell lymphoma 2 [1] Sezary syndrome 1 Peripheral T cell lymphoma, not otherwise characterized 9 Anaplastic large cell lymphoma T/null cell, primary systemic type 4 Unclassiﬁed high grade cutaneous CD4 and CD56 positive NK 1 cell lymphoma

Hodgkin lymphoma: 24 samples from 23 patients 2 (8) 0 (0) 22 (92) Nodular lymphocyte predominance Hodgkin lymphoma 3 Nodular sclerosis Hodgkin lymphoma 11 Mixed cellularity Hodgkin lymphoma 10 [9]

n, sample number, where multiple samples were examined from the same patient, patient numbers are given in square brackets. Two non-speciﬁc reactive hyperplasia samples had evidence of Epstein–Barr virus (EBV) infection using an in situ hybridization method. A further patient whose biopsy showed borderline histopathologic features of malignancy without evidence of clonality, later developed a NHL that had hypermethylation of Myf-3 (later sample not included). Four patients each had two different metachronous B-NHL and another had B cell lymphoma and Hodgkin lymphoma in two separate biopsies. One FL sample with hypermethylated Myf-3 was originally diagnosed as atypical follicular hyperplasia but was later shown to have a clonal Ig heavy chain gene rearrangement. Percentage is shown in parenthesis (%). Symbols indicate 1 (a) or 2 (b) excluded repeat samples where the initial and repeat samples had identical maximum Myf-3 fragment sizes.

criteria that reflected the histopathologic grade of malignancy. from a patient with lymphomatoid papulosis who had a Lower grade FL had follicular growth and comprised predomi- lymphoid infiltrate with malignant cytological features that nantly small cleaved cells (grade 1) or mixed small cleaved lacked immuno- or genotypic evidence of clonality. None of and large cells (grade 2), whereas higher grade 3 FL(l/d) had the atypical LPD samples had SB or PCR-based evidence of predominantly large cells with a follicular or follicular and clonal antigen receptor gene rearrangements and none had diffuse growth pattern. Lower grade DLBCL had unspecified clinical evidence of lymphoma subsequent to biopsy. cytologic features, whereas higher grade DLBCL(p/i) had a pleomorphic or immunoblastic morphology. Peripheral T cell lymphomas not otherwise characterized (PTL), were sub-grouped according to histopathologic grade Influence of age of malignancy based on cell size. Lower grade PTL samples were composed of medium or mixed medium and large size cells, whereas higher grade PTL(l) samples were comprised Several reports have shown that the methylation status of predominantly of large cells. some regions of the genome changes with age.16–19 Conse- The atypical LPD samples comprised five lymph nodes from quently, studies were undertaken to determine whether the five patients who had follicular hyperplasia with atypical his- increased Myf-3 methylation associated with different types of topathologic features that neither confirmed nor excluded a LPD was attributable to malignancy or to the age of the diagnosis of malignancy. In addition, there was a skin sample patient.

Leukemia Myf-3 hypermethylation in malignant LPD JME Taylor et al 585 Southern blotting assigned if there was an absent or weakly hybridizing 0.4 kb fragment relative to the density of the 0.6 kb fragment and/or The methylation status of Myf-3 was determined as previously by the presence of relatively dense fragments larger than described.10 Twelve µg of DNA was digested to completion 0.6 kb and up to 0.8 kb in size that were absent in DNA from according to the manufacturer’s instructions with 60 U of the benign lymphoid tissues. According to these criteria there was methylation sensitive and insensitive isoschizomers HpaII and no evidence of altered methylation in any of the benign LPD MspI, respectively. Following electrophoresis in a 1.8% aga- samples, including two hyperplastic EBV-positive DNA rose gel, the DNA was transferred to a positively charged samples which had Myf-3 fragments of size and relative den- nylon membrane using a BioRad (California, USA) vacuum sity that were indistinguishable from those of the other benign blotter according to the manufacturer’s instructions. DNA was samples. Moreover, there was no evidence of tissue-specific hybridized with a 0.8 kb cDNA probe specific for the 3′ methylation when the findings were compared of samples of untranslated region of Myf-3 (ATCC 61522), radiolabelled by peripheral blood leukocytes from healthy donors (results not random priming to high specific activity with dCTP 32P using shown), or lymph nodes, tonsils or spleens from patients with a Rediprime labelling kit (Amersham, Bucks, UK) according benign LPD. to the manufacturer’s instructions. Membranes were washed three times in a solution of 0.1% SDS and 0.1× SSC at 65°C for a total of 45–60 min. Autoradiograms were exposed at −80°C for a minimum of 2 days and fragment sizes were mea- LPD with atypical histopathologic features sured by reference to the electrophoretic mobility of DNA fragments obtained by the digestion of pUC 19 DNA with Of the five samples of DNA from lymph nodes of patients MspI and λ phage DNA with HindIII. diagnosed with atypical follicular hyperplasia, one sample The largest detected fragment size, irrespective of density, had an intense Myf-3 hybridizing fragment of 1.0 kb and a was used to assign the extent of Myf-3 methylation. weakly hybridizing Myf-3 fragment of 1.2 kb. The DNA sample of the skin biopsy from the patient who had lymphomatoid papulosis had no evidence of increased Myf-3 methylation. Results

The results are summarised in Table 1. Non-Hodgkin lymphoma and lymphoid leukemia

Benign LPD Myf-3 was hypermethylated in 73 of the 92 (79%) samples of DNA from patients diagnosed with malignant B cell LPD. Myf- Following HpaII digestion of genomic DNA from benign 3 was also hypermethylated in 16/27 (59%) of DNA samples LPDs, the Myf-3 probe hybridized to two major fragments of from patients diagnosed with malignant T cell LPD and the approximately 0.6 kb and 0.4 kb. In most samples the density three DNA samples from patients with malignant NK/T cell of the 0.4 kb fragment was greater than or equal to the density LPD (Figures 2 and 3). Of the 120 DNA samples from patients of the 0.6 kb fragment. In nine of the 45 samples (20%), the who had malignant LPD (excludes three repeat DNA samples 0.6 kb fragment was slightly denser than the 0.4 kb fragment. that had the same maximum-sized Myf-3 hybridizing fragment Two additional, weakly hybridizing fragments of approxi- in DNA from the first and subsequent biopsies), 30 had no mately 0.2 kb were detected in all samples. MspI digests of evidence of Myf-3 hypermethylation. Of these, 22 DNA benign and malignant samples consistently produced the two samples which included the DNA sample from a patient diag- fragments of approximately 0.2 kb and the single band of nosed with null cell ALCL, had evidence of a minor increase 0.4 kb which were also detected in HpaII digests. As shown of Myf-3 methylation (Table 1). Comparison of the median in Figure 1, the size of each of the MspI digested Myf-3 Myf-3 hybridizing fragment size in samples of NHL/leukemia hybridizing fragments was predicted from the published (1.8 kb) vs benign LPD samples (0.6 kb) showed a statistically sequence of Myf-3.8 significant increased level of Myf-3 methylation in the malig- Based on these findings, hypermethylation of Myf-3 was nant samples (P , 0.0001). The number of Myf-3 fragments assigned by the presence of fragments larger than 0.8 kb in indicated that not all the cells within a clone had identical size. By contrast, a minor increase of Myf-3 methylation was sites of methylation and were polyclonal with respect to their level of increased methylation (Figure 3). This finding was in contrast to the evidence of monoclonal Ig or TCR gene rearrangements (Figure 4). Comparison of the median Myf-3 fragment size in DNA samples of FL, DLBCL and PTL that were sub-grouped according to low or high histopathologic grade of malignancy, showed a significantly greater level of methylation in FL and PTL lymphomas of higher histopathologic grade (Table 2 and Figure 1 The MspI restriction sites (M) and predicted fragment Figures 5 and 6). By contrast, there was no statistically signifi- sizes that hybridize to the Myf-3 probe are shown8 (GenBank cant methylation difference in the sub-groups of DLBCL. accession number AF027148). These fragment sizes were similar to No evidence of deregulated Myf-3 methylation was found the observed sizes in MspI digested DNA. Methylation at the indicated # in DNA samples from eight patients (all had a sufficient num- restriction site ( ) was predicted to result in a hybridizing fragment of ber of clonal cells to be detected by SB using antigen receptor approximately 0.6 kb in HpaII digested DNA. The cleavage site within ′ ′ ′ gene probes) (Figure 4). The overall failure rate to detect the 5 CCGG 3 recognition sequences M1 to M7, is immediately 3 of the cytosine indicated by the nucleotide sequence number 11917, increased methylation of Myf-3 in patients with NHL/leukemia 12099, 12488, 12498, 12533, 12583 and 12795, respectively. was 8/113 (7%).

Leukemia Myf-3 hypermethylation in malignant LPD JME Taylor et al 586

Figure 2 Southern blot showing the size of fragments hybridizing with the Myf-3 probe following digestion of DNA from a series of benign and malignant LPDs with MspIorHpaII. All MspI digests of DNA from benign or malignant LPDs have a 0.4 kb fragment. In the HpaII digests of DNA from the reactive lymph nodes, there is an additional 0.6 kb fragment, generally of lesser density than the 0.4 kb Figure 3 Southern blot showing the size of fragments hybridizing fragment, and sometimes accompanied by very faint, slightly larger with the Myf-3 probe following digestion of DNA from a series of fragments as in lanes 1, 3, 5 and 12. The lanes containing DNA from malignant LPD with HpaII. Myf-3 in DNA from NHLs is hypermethyl- malignant LPDs (lanes 7–11, 14) show either minor Myf-3 methylation ated (lanes 1–7, 9) or has a minor increase of Myf-3 methylation (lane increase (lanes 7, 9, 10) or Myf-3 hypermethylation (lanes 11, 14). 8). Multiple large-sized fragments with a varying density indicate that DNA from the B cell small lymphocytic lymphoma in lane 10 has a not all the cells of a clone have identical sites of Myf-3 methylation very dense 0.6 kb fragment relative to the 0.4 kb fragment, a profile (lanes 1, 2, 4–7). The relatively faint 0.4 kb fragment shows that a high clearly distinct from reactive hyperplasia and is evidence of minor proportion of malignant cells is present in the samples, most evident in increase of Myf-3 methylation. By contrast, DNA from the splenic lanes 3–5. Hybridization with a kappa gene probe confirmed the high marginal zone B cell lymphoma in lane 13 which had evidence of B proportion of clonal cells in the sample represented in lane 4 cell clonality in an antigen receptor SB (Figure 4), does not have such (Figure 4). The methylation profile of Myf-3 varied in three biopsies a distinct profile and was not regarded as unequivocal evidence of a taken at intervals of 1 and 3 years from the same patient (lanes 5–7). minor increase of methylation. Myf-3 hypermethylation is evident in The first two samples (lanes 5, 6) have fragments of common size but DNA from the extranodal NK cell lymphoma of nasal type (lane 14) different relative density which indicates there are different pro- which had no evidence of clonal antigen receptor gene rearrange- portions of lymphocyte sub-clones each identified by its level of Myf- ments. HpaII digests: lanes 1, 3, 5, 7, 9–14; MspI digests: lanes 2, 4, 3 hypermethylation. The third sample has large fragments that are 6, 8; non-specific reactive hyperplasia: lanes 1–6, 12; B cell small absent in the earlier samples and is consistent with the proliferation lymphocytic lymphoma: lane 10; follicular lymphoma grade 1: lanes of lymphocyte sub-clones that have highly methylated Myf-3. The 7, 8; T cell granular lymphocytic leukemia: lane 11; diffuse large B maximum fragment size increased in a case of FL-mixed small cleaved cell lymphoma: lane 9; splenic marginal zone B cell lymphoma: lane and large cell (grade 2), that progressed during 3 years to FL-large cell 13; and extranodal NK cell lymphoma, nasal type: lane 14. (grade 3) with diffuse growth (lanes 8, 9, respectively). Faint bands of hybridized fragments indicate hypermethylation of a small proportion of cells in 2 biopsies from the patient with mixed cellularity HL taken Hodgkin lymphoma 2 years apart, the later biopsy has a greater fragment size (lanes 10, 11 respectively). Follicular lymphoma (FL)-grade 1: lanes 1, 5–7; diffuse large B cell lymphoma-pleomorphic cell: lane 4; FL grade 3 with There was no evidence of altered Myf-3 methylation in DNA diffuse growth: lanes 2, 9; FL grade 2: lane 8; Precursor T lymphoblas- samples from 22/23 patients (including one sample with evi- tic lymphoma: lane 3; and mixed cellularity Hodgkin lymphoma: dence of EBV infection using an in situ hybridization lanes 10, 11. technique). Hypermethylated Myf-3 fragments were seen in two metachronous DNA samples from a patient with mixed cellularity HL in which the later sample had evidence of a of a wide age range that were diagnosed with benign LPD. clonal T cell population. The earlier DNA sample had a The maximum fragment size ranged from 0.5 kb to less than maximum-sized Myf-3 fragment of 1.2 kb and the later 3.1 kb. 0.8 kb throughout the age groups sampled and there was no These fragments were faint by comparison with the 0.4 kb increase in the methylation status of Myf-3 with age. band and indicated that only a small proportion of cells had hypermethylated Myf-3 (Figure 3). Discussion

Inﬂuence of age on Myf-3 methylation Deregulated DNA methylation is a frequent ﬁnding in malignancy and previous studies of DNA from colon, breast and In order to determine whether the methylation status of Myf- lymphoid tissues showed that the Myf-3 gene is hypermethyl- 3 in DNA from lymphoid cells varied with age, the maximum ated in malignant but not benign tissues.10–13 Myf-3 has not extent of Myf-3 methylation was compared in DNA from been shown to be expressed in normal, benign or malignant tissue samples of lymph node, tonsil and spleen from patients lymphoid tissues and there is no in vivo evidence to indicate

Leukemia Myf-3 hypermethylation in malignant LPD JME Taylor et al 587

Figure 4 Southern blot showing clonally rearranged fragments of DNA from NHL hybridizing with antigen receptor gene probes. An immunoglobulin heavy chain joining region probe was hybridized to EcoR1 (lane 1) and HindIII (lane 2) digested DNA from the patient with splenic marginal zone B cell lymphoma without unequivocal evidence of increased Myf-3 methylation (Figure 2, lane 13). Both digests clearly indicate a monoclonal cell population which has rearranged both alleles. Comparison of germline and rearranged fragment densities indicates that the proportion of clonal cells is less than 50%. A kappa light chain joining region probe was hybridized to EcoRI (lanes 3, 4) and HindIII (lanes 5, 6) digested DNA from a patient with diffuse large B cell lymphoma having pleomorphic cytology (lanes 4, 6) and placenta from a healthy donor (lanes 3, 5). Both kappa alleles of the lymphoma sample are rearranged and the relative density of the rearranged and germline fragments indicates a very high proportion of clonal cells which is consistent with the relative density of the hypermethylated Myf-3 fragments in the Myf-3 Southern blot (Figure 3, lane 4). Bars indicate germline fragments and arrows indicate clonally rearranged fragments. Figure 5 Plot depicting the maximum Myf-3 fragment size in HpaII digests of DNA from B cell malignancies using Southern blotting. The wide range of values within many categories reflects their hetero- Table 2 Comparison of maximum Myf-3/HpaII fragment sizes geneous nature. Hypermethylated fragments were absent in some between categories of non-Hodgkin lymphoma/leukemia having dif- samples; however, the majority of these had a minor increase of ferent histopathologic grades of malignancy methylation that was absent in benign LPDs. FL and DLBCL were sub- grouped according to histopathologic criteria that reflected malig- Lower grade n Median Higher grade n Median P nancy grade, those FL comprised of predominantly large cells (grade category (kb) category (kb) 3) with or without diffuse growth more frequently had more extensive Myf-3 methylation than those without. Consistent with its low grade, DNA from samples of B-SLL had a uniformly low level of increased FL 26 1.7 FL (l/d) 6 5.4 .014 methylation which was distinct from that seen in benign LPDs. B-SLL: FL 26 1.7 DLBCL (p/i) 8 4.4 .028 B cell small lymphocytic lymphoma (10); PC: plasma cell myeloma PTL 4 0.6 PTL (l) 5 2.2 .048 (2); B-PLL: B cell prolymphocytic leukemia (1); MCL: mantle cell lym- malignant T/NK 23 1.0 T-LBL 7 5.5 .001 phoma (10); aBL: atypical Burkitt lymphoma (3); FL: follicular lym- LPD excluding phoma; small cleaved cell or mixed small and large cell with follicular T-LBL growth (26); FL(l/d): follicular lymphoma, grade 3 with or without diffuse growth (6); DLBCL: diffuse large B cell lymphoma, unspecified n, sample number, excluding repeat samples (3) having the same cytology (9); DLBCL(p/i): diffuse large B cell lymphoma, pleomorphic maximum fragment size. or immunoblastic cytology (8); MZL: marginal zone lymphoma, MALT Fisher’s exact test (two-sided) was used to compare the frequency type (7), splenic type (1), nodal type (1); and B-LBL: precursor B lym- of fragments .0.8 kb in the PTL and PTL(l) samples. All other stat- phoblastic leukemia/lymphoma (6). The number of samples is indi- istical analyses used the Mann–Whitney test (two-sided) to com- cated and excludes two repeat samples of FL from one patient in pare the median Myf-3 fragment sizes. which all samples had an identical maximum fragment size. that Myf-3 methylation contributes to lymphomagenesis, thus with their clinically and histopathologically heterogeneous the biological significance of Myf-3 hypermethylation in nature (Figures 5 and 6). Nevertheless, some categories had a malignant LPDs remains unclear. However, the chromosomal level of increased methylation consistent with their histopa- location and the density of CpG dinucleotides in Myf-3 indi- thologic grade of malignancy which might indicate a potential cated that it may have the potential to be a sensitive marker value for Myf-3 methylation status in prognostication. In of deregulated DNA methylation in lymphomas. addition, the high frequency of increased Myf-3 methylation The finding of increased Myf-3 methylation in samples from in NHL/leukemia and its absence in benign LPDs supports the 93% of patients with NHL/leukemia but not in any samples notion that deregulated methylation is more likely to be an from patients with benign LPDs indicates that the methylation early event in lymphomagenesis. status of Myf-3 is a sensitive and specific marker of malig- By contrast to the findings in NHL, increased Myf-3 methyl- nancy. Importantly, it may serve as a surrogate marker of clon- ation was infrequently detected in DNA from the samples ality in NK and null cell proliferations for which the assess- harboring HL. This may be due to the generally small pro- ment of antigen receptor gene rearrangements is portion of malignant cells in HL that is below the limit of uninformative. detection using the Southern blot technique, or alternatively, The highly variable level of increased Myf-3 methylation increased methylation may not be a feature of the malignant found in many of the NHL/leukemia categories was consistent cells of HL. The significance of the coincidental findings of

Leukemia Myf-3 hypermethylation in malignant LPD JME Taylor et al 588 that had no phenotypic or genotypic evidence of clonality. The DNA sample from the patient who had lymphomatoid papulosis did not have evidence of increased Myf-3 methylation and was consistent with the lack of detectable clonal gene rearrangements but not with the malignant cytological appearance of the abnormal lymphocytes. Interestingly, a proportion of patients with lymphomatoid papulosis have disease progression to cutaneous ALCL and in three of four ALCL samples there was increased Myf-3 methylation (Figure 6). These ﬁndings raise the possibility that deregulation of the DNA methylating machinery is a contributory factor in disease progression. The effect of EBV infection on Myf-3 methylation in LPDs remains to be studied. However, in the three EBV infected samples tested, there was no evidence of increased Myf-3 methylation suggesting that EBV infection in the absence of malignancy is not sufﬁcient to induce increased methylation of Myf-3. In conclusion, this study of the level of Myf-3 methylation in a large number of benign and malignant LPDs shows that it is a sensitive diagnostic test for the detection of malignant lymphoid cells and is especially useful in those cases in which current molecular diagnostic methods are unhelpful. Further- more, in some categories of NHL/leukemia the association of the level of Myf-3 methylation with the histopathologic grade of malignancy has potential value as an aid in prognostication.

Figure 6 Plot depicting the maximum Myf-3 fragment size in HpaII Acknowledgements digests of DNA from T and NK/T cell malignancies using Southern blotting. Similar to the findings in B cell malignancies there is a wide We wish to thank S Cairns and I van Bruggen for expert tech- range of values (Figure 5). DNA from samples of T-LBL had a greater nical assistance, and M Bulsara for the statistical analysis. level of hypermethylation when compared with other malignancies and reflected the highly malignant nature of this disease. DNA from samples of PTL(l) had greater hypermethylation than PTL and is consistent with the relatively higher histopathologic grade of malignancy References of PTL that are composed of large cells. The two samples of HSL were from one patient, an initial sample of splenic tissue had a minor 1 McCarthy KP, Sloane JP, Kabarowski JHS, Matutes E, Wiedeman increase of Myf-3 methylation, by comparison, a sample of leukemic LM. A simplified method of detection of clonal rearrangements lymphocytes taken 4 years later had hypermethylation of Myf-3 which of the T-cell receptor-γ chain gene. Diagn Mol Pathol 1992; 1: is consistent with de novo methylation. Similar to the B cell malig- 173–179. nancies, DNA from some T and T/NK malignancies did not have hyp- 2 Trainor KJ, Brisco MJ, Story CJ, Morley A. Monoclonality in B- ermethylated Myf-3. T-PLL: T cell prolymphocytic leukemia (1); T- lymphoproliferative disorders detected at the DNA level. Blood GLL: T cell granular lymphocytic leukemia (3); NK/T, n: extranodal 1990; 75: 2220–2222. NK/T cell lymphoma, nasal type (2); UNK: unclassified high grade 3 Spagnolo DV, Taylor J, Carrello S, Saueracker E, Kay PH. Southern cutaneous NK cell lymphoma (1); HSL: hepatosplenic γδ T cell lym- blot analysis of lymphoproliferative disorders: use and limitations phoma (2); SS: Sezary syndrome (1); PTL: peripheral T cell lymphoma, in routine surgical pathology. Pathology 1994; 26: 268–275. not otherwise characterized, medium cell or mixed medium and large 4 Holliday R. The inheritance of epigenetic defects. Science 1987; cell (4); PTL(l): peripheral T cell lymphoma, not otherwise charac- 238: 163–169. terized, large cell (5); ALCL: anaplastic large cell lymphoma, primary 5 el-Deiry WS, Nelkin BD, Celano P, Yen RW, Falco JP, Hamilton systemic type, CD30 positive (4); and T-LBL: precursor T lymphoblas- SR, Baylin SB. High expression of the methyltransferase gene tic leukemia/lymphoma (7). The number of samples is indicated and characterizes human neoplastic cells and progression stages of excludes a repeat sample from one patient with T-GLL that had a colon cancer. Proc Natl Acad Sci USA 1991; 88: 3470–3474. constant maximum fragment size. 6 Robertson KD, Uzvolgyi E, Liang G, Talmadge C, Sumegi J, Gon- zales FA, Jones PA. The human DNA methyltransferases (DNMTs) 1, 3a and 3b: coordinate mRNA expression in normal tissues and Myf-3 hypermethylation and T cell clonality in one sample of overexpression in tumours. Nucleic Acids Res 1999; 27: 2291– HL is unclear. Clonal TCRγ gene rearrangements have been 2298. reported rarely in HL but without clear evidence to identify 7 de Bustros A, Nelkin BD, Silverman A, Ehlich G, Poiesz B, Baylin the malignant cells of HL as having the clonal rearrange- SB. The short arm of chromosome 11 is a ‘hot-spot’ for hyperme- 20 thylation in human neoplasia. Proc Natl Acad Sci USA 1988; 85: ment. One possibility is that this sample had clonal cells 5693–5697. with both hypermethylated Myf-3 and clonally rearranged 8 Chen B, Dias P, Jenkins JJ, Savell VH, Parham DM. Methylation TCRγ genes and that these cells were in sufficient number for alterations of the MyoD1 upstream region are predictive of sub- the hypermethylated Myf-3 to be detected by the Southern classification of human rhabdomyosarcomas. Am J Pathol 1998; blot method. These clonal cells might be the malignant cells 152: 1071–1079. of HL or a predominant clone of benign, reactive T lympho- 9 Braun T, Bober E, Buschhausen-Denker G, Kotz S, Grzeschik K- H, Arnold HH. Differential expression of myogenic determination cytes with uncharacteristic Myf-3 hypermethylation. genes in muscle cells: possible autoactivation by the Myf gene The finding of hypermethylated Myf-3 fragments was unex- products. EMBO J 1989; 8: 3617–3625. pected in a sample from a patient who had an atypical LPD 10 Kay PH, Taylor J, Kees UR, Spagnolo D. Myf-3 hypermethylation

Leukemia Myf-3 hypermethylation in malignant LPD JME Taylor et al 589 in malignant lymphoproliferative disorders. Int J Cancer 1995; 62: Clinical Advisory Committee Meeting, Airlie House, Virginia, 797–799. November 1997. Mod Pathol 2000; 13: 193–207. 11 Ha¨hnel R, Harvey J, Kay PH. Hypermethylation of the myogenic 16 Ahuja N, Li Q, Mohan AL, Baylin SB, Issa J-PJ. Aging and DNA gene Myf-3 in human breast carcinomas. Anticancer Res 1996; methylation in colorectal mucosa and cancer. Cancer Res 1998; 16: 2111–2115. 58: 5489–5494. 12 Iacopetta BJ, Harmon D, Spagnolo DV, House AK, Kay PH. Hyper- 17 Golbus J, Palella TD, Richardson BC. Quantitative changes in T methylation of the Myf-3 gene in human colorectal cancer. cell DNA methylation occur during differentiation and ageing. Eur Anticancer Res 1997; 17: 429–432. J Immunol 1990; 20: 1869–1872. 13 Shannon B, Kay P, House A, Iacopettta B. Hypermethylation of the 18 Issa J-P J, Ottaviano YL, Celano P, Hamilton SR, Davidson NE, Myf-3 gene in colorectal cancers: associations with pathological Baylin SB. Methylation of the oestrogen receptor CpG island links features and with microsatellite instability. Int J Cancer 1999; 84: ageing and neoplasia in human colon. Nat Genet 1994; 7: 109–113. 536–540. 14 Maniatis R, Fritsch EF, Sambrook J. Molecular Cloning: a Labora- 19 Uehara Y, Ono T, Kurishita A, Kokuryu H, Okada S. Age-depen- tory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Labora- dent and tissue specific changes of DNA methylation within and tory, 1982. around the c-fos gene in mice. Oncogene 1989; 4: 1023–1028. 15 Harris NL, Jaffe ES, Diebold J, Flandrin G, Muller-Hermelink HK, 20 Strickler JG, Burgart LJ, Weiss LM. Classical Hodgkin’s disease. In: Vardiman J, Lister TA, Bloomfield CD. The World Health Organis- Weiss LM (ed.). Pathology Of Lymph Nodes, Churchill ation classification of hematological malignancies report of the Livingstone: New York, 1996, pp 169–214.

Leukemia