Frequent Silencing of Fragile Histidine Triad Gene(FHIT) in Burkitt’s Lymphoma Is Associated withAberrant Hypermethylation

Azhar Hussain,1 Marina I. Gutierrez,1* Georgina Timson,1 Abdul K. Siraj,1 Clara Deambrogi,2 Maha Al-Rasheed,1

Gianluca Gaidano,2 Ian Magrath,3 and Kishor Bhatia1,3*

1King Fahd National Centre for Children’s Cancer, KFSHRC, Riyadh, Saudi Arabia2Hematology Unit, Department of Medical Sciences & IRCAD, Amadeo Avogadro University of Eastern Piedmont, Novara, Italy3International Network for Cancer Treatment and Research, Brussels, Belgium

The fragile histidine triad (FHIT) gene, a potential tumor-suppressor gene, is frequently inactivated in multiple human cancers.However, the FHIT gene remains largely unexplored in Burkitt’s lymphoma (BL). Hence, we assessed whether loss of FHITexpression occurs in BL, and, if so, what is the mechanism of such loss. Lack of protein expression was observed in 50% ofBL cell lines. Methylation-specific polymerase chain reaction (MSP) showed that 45% of BL cell lines carried aberrantlymethylated FHIT alleles. Sequencing of bisulfite-treated DNA confirmed these data and indicated a very high density ofmethylation in all methylated alleles. Real-time, quantitative reverse-transcription PCR analysis indicated that attenuation offull-length FHIT transcription was correlated with methylation. Sequencing of transcripts illustrated that aberrant transcriptionresulting in loss of FHIT exons occurred more commonly in BL containing unmethylated FHIT genes. However, such transcriptsoften coexisted with full-length FHIT transcripts. Not surprisingly, therefore, loss of FHIT protein in BL correlated with CpGisland methylation, rather than with aberrant transcription. FHIT methylation also was detected in 31% (16 of 51) of theprimary BLs examined, including 2 samples whose derived cell lines also manifested FHIT hypermethylation. Aberrantmethylation can thus occur in vivo. In summary, this report provides evidence that epigenetic modification frequently resultsin loss of FHIT expression in BL. © 2004 Wiley-Liss, Inc.

INTRODUCTION

The FHIT (fragile histidine triad) enzyme is a147-amino-acid protein with diadenosine triphos-phate hydrolase activity that is involved in apopto-tic regulation and cell-cycle control (Barnes et al.,1996; Sard et al., 1999). The FHIT gene extendsmore than 1 Mb and encompasses 10 exons, ofwhich exons 5–9 encode the protein. The genomicregion spanning exons 3–5 includes the fragile siteFRA3B, on chromosome band 3p14.2 (Zimonjic etal., 1997). FRA3B is the most active aphidicolin-inducible common fragile site. It is also involved inthe t(3;8)(p14.2;q24) translocation associated with afamilial predisposition to renal carcinoma (Wilke etal., 1994; Paradee et al., 1996).

The instability of the FRA3B site results in fre-quent alterations in the expression of the FHITgene (Ohta et al., 1996; Huebner and Croce, 2001).Chromosomal deletions, translocations, and loss ofheterozygosity (LOH) affect the structure of FHITtranscripts (Druck et al., 1997). Aberrant splicingwithout intragenic deletion also has been de-scribed. Alteration of FHIT is a common occur-rence in a wide range of human malignancies, in-cluding hematopoietic neoplasms and lung, breast,

renal, cervical, head and neck, and esophageal car-cinomas (Negrini et al., 1996; Virgilio et al., 1996;Fong et al., 1997; Michael et al., 1997; Xiao et al.,1997; Huang et al., 2003; Ishii et al., 2003a). Inseveral types of tumors, genetic abnormalities in-volving FHIT have been associated with loss ofprotein expression, suggesting that these mecha-nisms mediate the silencing of FHIT.

The precise function(s) of the FHIT protein arestill unclear. In vitro, the protein catalyzes diade-nosine 5�,5��-p1,p3-triphosphate hydrolase activity,and it has been suggested that the dinucleosidepolyphosphate molecules act as signal molecules inbiological processes, including cell death (Trapassoet al., 2003). Loss of FHIT enzymatic activity re-sults in the accumulation of its substrates (Murphyet al., 2000), thereby preventing apoptosis.

*Correspondence to: Kishor Bhatia and Marina I. Gutierrez,KFNCCC&R, P.O. Box 3354, MBC #98-16, Riyadh, 11211, SaudiArabia. E-mail:gutierrez@kfshrc.edu.sa and dnadoc@hotmail.com

Received 25 March 2004; accepted 18 June 2004DOI 10.1002/gcc.20099Published online 21 September 2004 in

Wiley InterScience (www.interscience.wiley.com).

GENES, CHROMOSOMES & CANCER 41:321–329 (2004)

© 2004 Wiley-Liss, Inc.

Tumor-specific genomic alterations within theFHIT locus coupled with the apparent involve-ment of FHIT in apoptosis suggest that FHIT is atumor-suppressor gene (Siprashvili et al., 1997;Ishii et al., 2003b). Studies aimed at delineating thetumor-suppressor activity of FHIT have yieldedconflicting results, however, suggesting that its tu-mor-suppressor function may be dependent on celltype and/or gene dosage.

Inactivation of FHIT by mutation is a rare event(Druck et al., 1997; Gemma et al., 1997; Huebnerand Croce, 2001). Total loss of expression is usuallybrought about by other mechanisms, includingLOH and promoter hypermethylation (Esteller,2002). Indeed, aberrant methylation resulting inthe silencing of other proapoptotic genes has beenreported in several human malignancies (Harada etal., 2002; Hopkins-Donaldson et al., 2003; Rossi etal., 2003). Recent reports suggest that aberrant hy-permethylation of FHIT and its consequent tran-scriptional inactivation occur in breast, lung, esoph-ageal, cervical, prostate, and bladder cancer(Zochbauer-Muller et al., 2001; Noguchi et al.,2003; Uehara et al., 2003, Wu et al., 2003; Gutierrezet al., 2004). There is also evidence that, at least inbreast cancers, hypermethylation of one allele oc-curs in conjunction with LOH, and that these twoevents together constitute the “two hits” requiredfor the complete silencing of FHIT expression(Yang et al., 2002).

Burkitt’s lymphoma (BL) constitutes a histolog-ically and molecularly well-defined tumor entitywith a characteristic chromosomal translocationthat juxtaposes the MYC oncogene to one of theimmunoglobulin genes, usually IGH (Magrath,1997). There is little information on the role ofFHIT in BL, and, to the best of our knowledge, nodata exist on expression of FHIT. Ferrer et al.(1999) described aberrant FHIT transcription inseveral BL cell lines as a result of aberrant splicingevents that frequently skip exons 2–5. Because theFHIT protein is encoded by exons 5–9, not allaberrant transcripts cause loss of coding potential.Furthermore, most of the BLs that demonstratedaberrant FHIT transcripts also expressed the nor-mal transcript. In general, therefore, it seems thatthis mechanism is not sufficient to result in loss ofthe FHIT protein.

We report here the results of the current study,which showed loss of FHIT protein in BL cell linesand potential mechanisms of inactivation. We ob-served that aberrant hypermethylation and aber-rant transcription are both common events in BLcell lines, but that only the former was associated

with loss of protein expression. Furthermore, wedemonstrate that FHIT methylation also occurs inone third of primary tumors.

MATERIALS AND METHODS

Cell Lines and Primary Tumors

Cell lines were grown in RPMI1640 supple-mented with 10% fetal bovine serum in a 5% CO2

atmosphere at 37°C. We studied 18 BL cell lines—CA46, JLP119, JD38, LW878, Namalwa, Akata,BML895, Daudi, KK124, BL41, EW36, MC116,ST486, RAJI, Louckes, P3HR1, BL30, andRamos—and 13 AIDS-BL cell lines— AIDS-BLswere PA682PB, PA682BM, AS283A, BrgIgA,ESIII, LamC3�, LamG10, BrgIgM, HBL1,HBL2, HBL3, HBL4, and NC71 (the last 7 celllines were analyzed only for methylation)—previ-ously characterized for other genetic lesions(Shiramizu et al., 1991; Rossi et al., 2003). Fourlymphoblastoid cell lines and normal peripheralblood from 20 individuals (obtained after properconsent) were the controls.

DNA from 51 primary tumors (33 BL and 18AIDS-BL) was available from previous studies(Gutierrez et al., 1992; Rossi et al., 2003) and wassubjected to methylation analyses.

Western Blot Analysis

Five million cells from each cell line were usedto extract proteins by a standard lysis method. Afterelectrophoresis and electrotransfer, the membraneswere incubated with the primary antibody over-night at 4°C. We used rabbit anti-FHIT (ZymedLaboratories, San Francisco, CA) and anti-�-actin(Abcam Ltd., Cambridge, UK) to ensure equalloaded amounts. Immunoreactivity was assessedusing appropriate secondary antibodies (1:8,000 di-lution) from Amersham (Arlington Heights, IL).The blots were developed with an enhancedchemiluminescence kit (Amersham).

RNA Extraction, Reverse-TranscriptionPolymerase Chain Reaction, and Real-Time,Quantitative Reverse-Transcription PolymeraseChain Reaction

Total RNA was extracted from all cell lines withTrizol (Invitrogen Life Technologies, Carlsbad,CA) and was reverse-transcribed into cDNA withrandom hexamers. The cDNA was amplified withFHIT-specific primers (GAGAAATCCACTG-AGAACAG and GCAATAGCTCTTTTGCTG-GAA) spanning 680 bp from exon 2 to exon 10.PCR was done for 35 cycles at an annealing tem-

322 HUSSAIN ET AL.

perature of 56°C. Amplification of GAPD was usedas a control for the integrity and relative amount ofall cDNAs. PCR products were visualized afterelectrophoresis on 4% agarose gels.

To quantify the level of FHIT transcripts in BLcell lines, we developed real-time reverse-tran-scription polymerase chain reaction (RT-PCR)using primers in exon 1 (TTCCCAGCTGTC-AACATCCT) and exon 4 (CTCTTCGGA-GTCCTCAGTGG) that span 202 bp. We used aLightCycler (Roche, Mannheim, Germany) to am-plify this fragment using a touchdown annealingtemperature of 63°C–58°C in the presence of3 mM MgCl2 and Sybr green I dye. A characteristicand reproducible melting peak was visualized at83.33°C � 0.30°C, and serial dilutions of NamalwacDNA were used in order to plot a standard curvefor quantification. All estimations were relative tothe amount of GAPD transcripts calculated in asimilar experiment (ratio between the concentra-tions of FHIT transcripts and GAPD transcripts).

Methylation-Specific Polymerase ChainReaction Analysis

For methylation-specific polymerase chain reac-tion (MSP) analysis, genomic DNA was extractedwith a Puregene kit (Gentra, Minneapolis, MN) orwas available from previous studies (Gutierrez etal., 1992; Rossi et al., 2003). One microgram ofgenomic DNA was denatured in 0.4 M NaOH andmodified with 3 M sodium bisufite and 10 mMhydroquinone at 55°C for 16 hr. After purificationwith a GeneCleanIII kit (Bio101, Vista, CA), theDNA was desulfonated in 0.4 M NaOH, precipi-tated in ethanol, and resuspended in dH20. Then200 ng was used as a template in MSP reactionswith 1.5 mM MgCl2 and 20 pmol of primers spe-cific for methylated (M) and unmethylated (U)forms (Zochbauer-Muller et al., 2001). The meth-ylated FHIT reaction consisted of 32 cycles of

touchdown PCR at annealing temperatures rangingfrom 71°C to 63°C with primers TGGGGC-GCGGGTTTGGGTTTTTACGC and CGT-AAACGACGCCGACCCCACTA. The unmethyl-ated FHIT reaction was done at 64°C for 33 cycleswith primers TTGGGGTGTGGGTTTGGG-TTTTTATG and CATAAACAACACCAACCC-CACTA, corresponding to nucleotides 189–301(GenBank accession number U76263). Each reac-tion was tested with untreated DNA to ensure lackof amplification, and three controls were includedto ensure specificity: (1) normal human DNA pre-viously treated with the CpG methylase SssI in thepresence of S-adenosylmethionine (in vitro meth-ylated DNA), (2) DNA from peripheral lympho-cytes from a healthy individual (normal control),and (3) no template (blank). PCR products wereanalyzed after electrophoresis on 4% agarose gelscontaining ethidium bromide.

Cloning and Sequencing of PCR Products

PCR products derived from bisulfite-treatedDNA amplified with primers outside the majorCpG-rich region in the first intron of FHIT(GGAGGTAAGTTTA-AGTGGAA and CCC-ACCCTAAAACCTCTTTT) and PCR productsderived from RT-PCR reactions (exons 2–10) werecloned into the pCR 2.1-TOPO vector (Invitrogen,Carlsbad, CA). From each PCR product, 10 posi-tive clones were selected and sequenced using anautomated DNA sequencer (Molecular Dynamics,Piscataway, NJ).

RESULTS

FHIT Expression and Methylation in BL Cell Lines

We determined the expression of FHIT proteinin a panel of 24 cell lines, 18 derived from BL and6 derived from AIDS-BL tumors. Western blotanalysis (Fig. 1) demonstrated unequivocally that

Figure 1. Western blot analysis of FHIT expression from 24 cell lines. The blots were incubated withanti-FHIT (top panel) or anti-�-actin (lower panel) antibodies. Molecular sizes of the proteins are indicatedon the left.

323METHYLATION AND LOSS OF FHIT EXPRESSION IN BL

FHIT protein was absent in 12 of the celllines (50%).

To utilize MSP to analyze the methylation statusof FHIT in the BL cell lines, we first checked themethylation status of FHIT in normal lymphoidpopulations by testing DNA from lymphocytes ob-tained from 20 healthy donors and from 4 lympho-blastoid cell lines. No methylation was detected inany of these nontumor samples (Fig. 2A). In con-

trast, 14 of the 31 cell lines (45%) demonstratedaberrant hypermethylation (Fig. 2B).

We then correlated Western blot data with meth-ylation data for 24 tumor cell lines (Table 1). Wefound that 8 of the 11 methylated cell lines had lostprotein expression, whereas only 4 of the 13 un-methylated cell lines failed to express FHIT. Al-though the sample sizes were too small for thisdifference to achieve statistical significance, statis-

Figure 2. Methylation-specific PCR showing re-sults from methylated (M) and unmethylated (U)reactions using DNA from normal lymphocytesfrom 9 donors (A, PBL), cell lines (B, CL) andprimary tumors (C, PRI). Amplified products of themethylated reaction were 112 bp in size, and thoseof the unmethylated reaction were 113 bp.

TABLE 1. Summary of FHIT Data from BL Cell Lines1

Cell lines Origin EBV Protein MSP M clones (%)Relative

FHIT level2

Transcripts

WT3 Aberrant

CA46 Sporadic � � M 100 0 � �AS283 AIDS-BL � � M 100 0 � �Ramos Sporadic � � M 100 NA4 � �Daudi Endemic � � M 40 0.056 � �Louckes Sporadic � � M 40 NA � �Akata Sporadic � � M NA 0 � �JLP119 Sporadic � � M NA 0.54 � �P3HRI Endemic � � M NA NA � �Namalwa Endemic � � M 90 0.58 � �BML895 Sporadic � � M 47 0.54 � �EW36 Sporadic � � M 20 NA � �LW878 Sporadic � � U 0 1.7 � �Raji Endemic � � U 0 1.1 � �KK124 Sporadic � � U 0 NA � �JD38 Sporadic � � U NA 1.5 � �MC116 Sporadic � � U NA 0.5 � �BL41 Sporadic � � U 0 0.9 � �PA682PB AIDS-BL � � U NA 0.6 � �ESIII AIDS-BL � � U NA 1.3 � �PA682BM AIDS-BL � � U NA NA � �ST486 Sporadic � � U 0 1.55 � �LamC3� AIDS-BL � � U 0 0.8 � �BrglgA AIDS-BL � � U 0 0 � �BL30 Sporadic � � U NA NA � �

1Seven additional AIDS-BL cell lines (LamG10, BrglgM, HBL1, HBL2, HBL3, HBL4, and NC71) were available as DNA samples and, thus, only wereincluded in analysis of methylation.2Relative FHIT level: relative transcripts quantified by RQ-RT-PCR.3WT: wild-type transcript detected by RT-PCR.4NA: not available.

324 HUSSAIN ET AL.

tical analysis showed a trend toward an association,suggesting that other mechanisms of inactivationalso operate in BL (see below).

To further assess the density and distribution ofmethylated alleles in BL, we used primers outsidethe first intronic CpG island that were specific forbisulfite-treated DNA to amplify, clone, and se-quence the resulting products. As can be seen inFigure 3, sequencing data confirmed a very highmethylation density at the 13 CpG sites analyzedin all methylated samples. However, the proportionof methylated clones varied from 20% to 100% indifferent cell lines (Table 1). None of the unmeth-ylated BLs or the normal control showed evidenceof any notable level of methylation.

To confirm that methylation of FHIT in BL canimpede or reduce its transcriptional activity, weused real-time RT-PCR to quantify the level ofFHIT transcripts and that of a reference gene,GAPD. We calculated relative FHIT transcripts af-ter normalization to GAPD and quantified the ex-pression in 7 methylated and 10 unmethylated celllines (Table 1). Expression levels ranged from 0 to0.58 (average, 0.24) in the methylated samples, incontrast to expression ranging from 0 to 1.3 (aver-age, 0.99) in the unmethylated samples (Fig. 4).

To detect aberrant transcripts, we used RT-PCRwith primers in exons 2 and 10, which encompassthe entire coding sequence and cover virtually allof the previously described abnormal messages.Aberrant transcripts coexpressed with or withoutthe wild-type transcript were detected in 15 and 4of the 24 cell lines, respectively. The remaining 5cell lines failed to express any transcript (Table 1).

As expected, in the latter, real-time, quantitativeRT-PCR showed the absence of transcripts andWestern blot analysis showed the absence of pro-tein.

To confirm the structure of these aberrant tran-scripts, we cloned and sequenced RT-PCR prod-ucts from 16 BLs with alternative transcripts. Fig-ure 5 has a summary of these results, which showedthat exons 5 and 6 were the most frequently lost,followed by exons 3 and 4 and an 11-bp deletion inexon 10 (Ferrer et al., 1999). We also correlatedaberrant FHIT transcription with lack of protein

Figure 4. Relative levels of FHIT transcripts quantified by real-timeRT-PCR plotted for methylated (M) and unmethylated (U) BLs. FHITtranscript levels were normalized to GAPD message levels. Cell linesexpressing protein are indicated by black squares; cell lines lacking FHITare indicated by white circles. The average expression level in eachgroup is indicated with a line. Three cell lines with certain levels oftranscript but lack of protein demonstrated mostly aberrant transcripts(loss of exons 8 and 3–6 and an 11-bp deletion), whereas only 20% ofclones had wild-type sequences.

Figure 3. Confirmation of methylation was as-certained by sequencing of bisulfite-treated DNAfrom 15 cell lines representing either those positivefor methylation by MSP (8) or those that werenegative (7). The map on top depicts the location ofthe PCR primers in intron 1 in the context of the 13CpG sites analyzed. Representative methylationpatterns (A–D) are shown underneath. Differentdensities of methylation at each CpG site are indi-cated in different shades. The proportion of clonesthat yielded a methylated sequence is given. All thenegative cell lines (7) yielded pattern A; positive celllines gave patterns B (4), C (3), or D.

325METHYLATION AND LOSS OF FHIT EXPRESSION IN BL

expression. Of the 19 BLs with evident aberranttranscripts, 12 retained FHIT expression (Table 1).No statistical association was found. However, ab-errant splicing may explain the lack of protein in 3of the 4 BLs with unmethylated FHIT describedearlier.

FHIT Hypermethylation in Primary Tumors

To determine whether FHIT methylation occursin vivo, we assessed the frequency of CpG islandmethylation in 51 primary BL tissues (33 BLs and18 AIDS-BLs) obtained at diagnosis and prior tochemotherapy. Methylation of FHIT by MSP wasshown in 12 BLs and 4 AIDS-BLs (Fig. 2C andTable 2). Overall, 31% were positive for hyper-methylation, including 2 samples for which thederived cell lines also displayed methylation. Tu-mors with methylated FHIT were distributedequally between those carrying mutant TP53 andthose carrying wild-type TP53. Among the non-African tumors, 10 of the 15 with methylated FHITwere EBV-associated.

DISCUSSION

The FHIT gene has been postulated to be atumor-suppressor gene (Siprashvili et al., 1997;Ishii et al., 2003a). This was supported by theresults of in vivo studies showing that several tu-mors, including hematologic neoplasms and lung,breast, prostate, and bladder carcinomas, carry aninactivated FHIT gene (Negrini et al., 1996; Vir-gilio et al., 1996; Fong et al., 1997; Michael et al.,1997; Xiao et al., 1997; Huang et al., 2003; Ishii etal., 2003a), as well as of in vitro studies and animalmodels demonstrating that FHIT induces apopto-sis and reduces tumorigenicity (Siprashvili et al.,1997; Sard et al., 1999; Roz et al., 2002; Ishii et al.,2003b; Sevignani et al., 2003; Trapasso et al., 2003).

Furthermore, FHIT knockout mice are prone tospontaneous and induced tumor development (Du-mon et al., 2001; Zanesi et al., 2001). However, therole of FHIT in BL remains unknown.

Previous studies demonstrated that inactivatingpoint mutations are not common in the FHIT gene(Druck et al., 1997; Gemma et al., 1997; Huebnerand Croce, 2001, 2003). Instead, LOH and othergenomic instabilities at the 3p14.2 locus (FRA3B)that cause loss of FHIT coding sequences havebeen seen frequently in various tumor types (Ne-grini et al., 1996; Virgilio et al., 1996; Fong et al.,1997; Michael et al., 1997; Xiao et al., 1997; Huanget al., 2003; Ishii et al., 2003a). Aberrant methyl-ation of FHIT also has been described in a propor-tion of carcinomas (Zochbauer-Muller et al., 2001;Noguchi et al., 2003; Uehara et al., 2003; Gutierrezet al., 2004).

The current study demonstrated that loss ofFHIT protein is a common event in BL/AIDS-BL(50%, Fig. 1), with two potential mechanisms rel-evant to FHIT inactivation assessed. We observedFHIT methylation in 36% of BL/AIDS-BL(Fig. 2B and C). Absence of methylation in normalDNA was confirmed by MSP analyses of normalperipheral blood lymphocytes (Fig. 2A). Becausecell lines (45%, Table 1) as well as primary tumors(31%, Table 2) demonstrated methylation, thisphenomenon occurs in vivo. This conclusion wasfurther strengthened by the detection of aberrantmethylation in two independent paired samples(fresh tumors and derived cell lines). Bisulfite se-quencing not only confirmed these results but alsoindicated that some BLs yielded only highly meth-ylated clones, suggesting that these have biallelicmethylation or, alternatively, monoallelic methyl-ation and LOH. However, other BLs had 20%–

Figure 5. Structure of the FHIT gene (in map ontop). Coding exons are represented by gray boxesand noncoding exons by dotted boxes. The initia-tion and end of translation sites are also shown.Aberrant FHIT transcripts from BL cell lines ob-tained by RT-PCR were sequenced. The bar graphshows the frequency of loss of each exon and thefrequency of an 11-bp deletion at the beginning ofexon 10 in the aberrant transcripts.

326 HUSSAIN ET AL.

50% methylation (Fig. 3), suggesting a monoallelicevent or methylation in a subpopulation of cells.

Correlation was found between FHIT methyl-ation and loss of protein in BL (Table 1). Whereas73% (8 of 11) of methylated cases lacked FHIT,only 30% (4 of 13) of unmethylated cases fell intothis category. Statistical analyses confirmed thistrend, although a larger series would be needed toachieve significance. These data also suggest thatother inactivating mechanisms are operating in BL.Bisulfite sequencing showed that whenever meth-ylated alleles were present, all 13 CpG sites ana-lyzed were densely methylated. To analyze theinfluence of methylation on transcription, we usedreal-time RT-PCR to estimate the relative numberof FHIT transcripts (Fig. 4). Methylation was suf-ficient to silence FHIT in some cell lines or tosignificantly reduce its transcriptional activity inothers, which had partial methylation (Table 1).

We then examined what role aberrant transcriptsmight have in influencing FHIT expression. All ofthe BL/AIDS-BL cell lines that expressed FHITtranscripts showed multiple aberrant splicingevents (Fig. 5), but aberrant transcripts usuallycoexisted with the wild-type message. Thus, it wasnot surprising that no statistical association wasfound between the presence of aberrant transcriptsand loss of protein, a finding also reported previ-ously in acute lymphoblastic leukemia (Hallas etal., 1999). Aberrant transcription not only is insuf-ficient to silence FHIT but also does not collabo-rate with methylation. Indeed, there was a trendtoward an association between the presence of ab-errant transcripts and lack of methylation (Ta-ble 1). In contrast, methylation of CDKN2A (p16) inhepatocellular carcinoma induced its aberrant tran-scription (Suh et al., 2000). Furthermore, aberranttranscripts may not be tumor-specific, as they havebeen detected in normal cells (Ferrer et al., 1999;Peters et al., 1999).

Finally, we correlated FHIT methylation in pri-mary tumors with clinicopathologic characteristics.No association was found between methylatedFHIT and TP53 (p53) mutation (Table 2). A similarlack of association was reported for lung and breastcancer (Zochbauer-Muller et al., 2001). We alsocorrelated the present data with Epstein–Barr virus(EBV) positivity in 49 sporadic BLs. The majorityof cases with methylation of FHIT (67%) wereEBV-positive. Furthermore, FHIT methylationwas more common in BL cell lines and primarytumors (22 of 51, 43%) than in AIDS-BL cell linesand biopsy specimens (8 of 31, 26%).

TABLE 2. Summary of Data from Primary BL Tumors

BL Origin EBV FHIT TP53a

2218 Endemic � M NAb

OJI Sporadic � M MUTAA Sporadic � M WTVA Sporadic � M WTFB Sporadic � M WTRPF Sporadic � M WT3792 Sporadic � M NAAC Sporadic � M NAAG Sporadic � M MUTBD Sporadic � M MUTBE Sporadic � M MUTTD Sporadic � M NA2207 Endemic � U NAVGO Sporadic � U MUTCV Sporadic � U MUTFNR Sporadic � U WTSG Sporadic � U WTMP Sporadic � U WTAF Sporadic � U WTCCH Sporadic � U WTMD Sporadic � U WTSCL Sporadic � U WTDW6 Sporadic � U NASRC Sporadic � U MUTIN Sporadic � U MUTJD Sporadic � U MUTHOL Sporadic � U MUTJRI Sporadic � U MUTRJ Sporadic � U WTRJR Sporadic � U WTND Sporadic � U NAAS Sporadic � U NAME Sporadic � U NA

AIDS-BL EBV FHIT TP53

1907 � M MUT2415 � M WT1246 � M WT1220 � M WT3683 � U MUT2407 � U MUTPA � U WT1249 � U WT3685 � U WT3680 � U NA1143 � U MUT1248 � U MUT642 � U WT2276 � U WT1142 � U WT1247 � U WT1909 � U WT3653 NA U NA

aTP53: wild-type (WT) or mutant (MUT) TP53 gene.bNA: not available.

327METHYLATION AND LOSS OF FHIT EXPRESSION IN BL

Interestingly, a recent report suggests that loss ofFHIT protein correlates significantly with a worseprognosis in diffuse large B-cell lymphomas (Chenet al., 2004). Clinical data available from a subset of16 lymphomas analyzed in this report suggested asimilar trend: 4 of the 6 lymphomas that relapsedhad methylation of FHIT at diagnosis. Larger stud-ies are needed to elucidate further the significanceof this observation. Genetic and epigeneticchanges not only contribute to the malignant phe-notype, but they may also collaborate in influenc-ing the clinical behavior of the tumor.

These observations demonstrate that loss ofFHIT function frequently occurs in vivo in BL andmay play a role in lymphomagenesis. Althoughaberrant transcription occurs, it is not sufficient toimpede FHIT expression. Therefore, an epigeneticmechanism is more important than a genetic mech-anism for inactivation of FHIT in BL.

ACKNOWLEDGMENT

The authors acknowledge the technical assis-tance of Muna Ibrahim in the cloning of PCRproducts.

REFERENCES

Barnes LD, Garrison PN, Siprashvili Z, Guranowski A, RobinsonAK, Ingram SW, Croce CM, Ohta M, Huebner K. 1996. Fhit, aputative tumor suppressor in humans, is a dinucleoside 5�,5��-P1,P3-triphosphate hydrolase. Biochemistry 35:11529–11535.

Chen PM, Yang MH, Hsiao LT, Yu IT, Chu CJ, Chao TC, Yen CC,Wang WS, Chiou TJ, Liu JH. 2004. Decreased FHIT proteinexpression correlates with a worse prognosis in patients withdiffuse large B-cell lymphoma. Oncol Rep 11:349–356.

Druck T, Hadaczek P, Fu TB, Ohta M, Siprashvili Z, Baffa R,Negrini M, Kastury K, Veronese ML, Rosen D, Rothstein J,McCue P, Cotticelli MG, Inoue H, Croce CM, Huebner K. 1997.Structure and expression of the human FHIT gene in normal andtumor cells. Cancer Res 57:504–512.

Dumon KR, Ishii H, Fong LY, Zanesi N, Fidanza V, Mancini R,Vecchione A, Baffa R, Trapasso F, During MJ, Huebner K, CroceCM. 2001. FHIT gene therapy prevents tumor development inFhit-deficient mice. Proc Natl Acad Sci USA 98:3346–3351.

Esteller M. 2002. CpG island hypermethylation and tumor suppres-sor genes: a booming present, a brighter future. Oncogene 21:5427–5440.

Ferrer M, Lopez-Borges S, Lazo P. 1999. Expression of aberrantfunctional and nonfunctional transcripts of the FHIT gene inBurkitt’s lymphomas. Mol Carcinogenesis 25:55–63.

Fong KM, Biesterveld EJ, Virmani A, Wistuba I, Sekido Y, BaderSA, Ahmadian M, Ong ST, Rassool FV, Zimmerman PV, Giac-cone G, Gazdar AF, Minna JD. 1997. FHIT and FRA3B 3p14.2allele loss are common in lung cancer and preneoplastic bronchiallesions and are associated with cancer-related FHIT cDNA splic-ing aberrations. Cancer Res 57:2256–2267.

Gemma A, Hagiwara K, Ke Y, Burke LM, Khan MA, Nagashima M,Bennett WP, Harris CC. 1997. FHIT mutations in human primarygastric cancer. Cancer Res 57:1435–1437.

Gutierrez MI, Bhatia K, Barriga F, Diez B, Sackmann Muriel F, deAndreas ML, Eppelman S, Risueno C, Magrath IT. 1992. Molec-ular epidemiology of Burkitt’s lymphoma from South America:Differences in breakpoint locations and EBV association fromtumors in other world regions. Blood 79:3261–3266.

Gutierrez MI, Siraj AK, Khaled H, Koon N, El-Rifai W, Bhatia K.2004. CpG island methylation in Schistosoma- and non-Schisto-soma-associated bladder cancer. Mod Pathol [Epub ahead of print21 May 2004].

Hallas C, Albitar M, Letofsky J, Keating MJ, Huebner K, Croce CM.1999. Loss of FHIT expression in acute lymphoblastic leukemia.Clin Cancer Res 5:2409–2414.

Harada K, Toyooka S, Shivapurkar N, Maitra A, Reddy JL, Matta H,Miyajima K, Timmons CF, Tomlinson GE, Mastrangelo D, HayRJ, Chaudhary PM, Gazdar AF. 2002. Deregulation of caspase 8and 10 expression in pediatric tumors and cell lines. Cancer Res62:5897–5901.

Hopkins-Donaldson S, Ziegler A, Kurtz S, Bigosch C, Kandioler D,Ludwig C, Zangemeister-Wittke U, Stahel R. 2003. Silencing ofdeath receptor and caspase-8 expression in small cell lung carci-noma cell lines and tumors by DNA methylation. Cell DeathDiffer 10:356–364.

Huang LW, Chao SL, Chen TJ. 2003. Reduced Fhit expression incervical carcinoma: correlation with tumor progression and poorprognosis. Gynecol Oncol 90:331–337.

Huebner K, Croce CM. 2001. FRA3B and other common fragilesites: the weakest links. Nature Rev Cancer 1:214–221.

Huebner K, Croce CM. 2003. Cancer and the FRA3B/FHIT fragilelocus: it’s a HIT. Br J Cancer 88:1501–1506.

Ishii H, Vecchione A, Furukawa Y, Sutheesophon K, Han SY, DruckT, Kuroki T, Trapasso F, Nishimura M, Saito Y, Ozawa K, CroceCM, Huebner K, Furukawa Y. 2003a. Expression of FRA16D/WWOX and FRA3B/FHIT genes in hematopoietic malignancies.Mol Cancer Res 1:940–947.

Ishii H, Zanesi N, Vecchione A, Trapasso F, Yendamuri S, Sarti M,Baffa R, During MJ, Huebner K, Fong LY, Croce CM. 2003b.Regression of upper gastric cancer in mice by FHIT gene deliv-ery. FASEB J 17:1768–1770.

Magrath IT. 1997. Non-Hodgkin’s lymphomas: epidemiology andtreatment. Ann NY Acad Sci 824:91–106.

Michael D, Beer DG, Wilke CW, Miller DE, Glover TW. 1997.Frequent deletions of FHIT and FRA3B in Barrett’s metaplasiaand esophageal adenocarcinomas. Oncogene 15:1653–1659.

Murphy GA, Halliday D, McLennan AG. 2000. The Fhit tumorsuppressor protein regulates the intracellular concentration ofdiadenosine triphosphate but not diadenosine tetraphosphate.Cancer Res 60:2342–2344.

Negrini M, Monaco C, Vorechovsky I, Ohta M, Druck T, Baffa R,Huebner K, Croce CM. 1996. The FHIT gene at 3p14.2 isabnormal in breast carcinomas. Cancer Res 56:3173–3179.

Noguchi T, Takeno S, Kimura Y, Uchida Y, Daa T, Yokoyama S,Gabbert HE, Mueller W. 2003. FHIT expression and hypermeth-ylation in esophageal squamous cell carcinoma. Int J Mol Med11:441–447.

Ohta M, Inoue H, Cotticelli MG, Kastury K, Baffa R, Palazzo J,Siprashvili Z, Mori M, McCue P, Druck T, et al. 1996. The FHITgene, spanning the chromosome 3p14.2 fragile site and renalcarcinoma-associated t(3;8) breakpoint, is abnormal in digestivetract cancers. Cell 84:587–597.

Paradee W, Wilke CM, Wang L, Shridhar R, Mullins CM, Hoge A,Glover TW, Smith DI. 1996. A 350-kb cosmid contig in 3p14.2that crosses the t(3;8) hereditary renal cell carcinoma translocationbreakpoint and 17 aphidicolin-induced FRA3B breakpoints.Genomics 35:87–93.

Peters UR, Hasse U, Oppkieger E, Tschan M, Ong ST, Rassool FV,Borisch B, Tobler A, Fet MF. 1999. Aberrant FHIT mRNAtranscripts are present in malignant and normal hematopoiesis,but absence of FHIT protein is restricted to leukemia. Oncogene18:79–85.

Rossi D, Gaidano G, Gloghini A, Deambrogi C, Franceschetti S,Berra E, Cerri M, Vendramin C, Conconi A, Viglio A, Muti G,Oreste P, Morra E, Paulli M, Capello D, Carbone A. 2003. Fre-quent aberrant promoter hypermethylation of O6-methylguanine-DNA methyltransferase and death-associated protein kinasegenes in immunodeficiency-related lymphomas. Br J Haematol123:475–478.

Roz L, Gramegna M, Ishii H, Croce CM, Sozzi G. 2002. Restorationof fragile histidine triad (FHIT) expression induces apoptosis andsuppresses tumorigenicity in lung and cervical cancer cell lines.Proc Natl Acad Sci USA 99:3615–3620.

Sard L, Accornero P, Tornielli S, Delia D, Bunone G, Campiglio M,Colombo MP, Gramegna M, Croce CM, Pierotti MA, Sozzi G.1999. The tumor-suppressor gene FHIT is involved in the regu-lation of apoptosis and in cell cycle control. Proc Natl Acad SciUSA 96:8489–8492.

Sevignani C, Calin GA, Cesari R, Sarti M, Ishii H, Yendamuri S,Vecchione A, Trapasso F, Croce CM. 2003. Restoration of fragilehistidine triad (FHIT) expression induces apoptosis and sup-

328 HUSSAIN ET AL.

presses tumorigenicity in breast cancer cell lines. Cancer Res63:1183–1187.

Shiramizu B, Barriga F, Neequaye J, Jafri A, Dalla-Favera R, Neri A,Gutierrez M, Levine P, Magrath I. 1991. Patterns of chromosomalbreakpoint locations in Burkitt’s lymphoma: Relevance to geog-raphy and EBV association. Blood 77:1516–1526.

Siprashvili Z, Sozzi G, Barnes LD, McCue P, Robinson AK, Eryo-min V, Sard L, Tagliabue E, Greco A, Fusetti L, Schwartz G,Pierotti MA, Croce CM, Huebner K. 1997. Replacement of Fhitin cancer cells suppresses tumorigenicity. Proc Natl Acad Sci USA94:13771–13776.

Suh SI, Pyun HY, Cho JW, Baek WK, Park JB, Kwon T, Park JW,Suh MH, Carson DA. 2000. 5-Aza-2�-deoxycytidine leads todown-regulation of aberrant p16INK4A RNA transcripts and re-stores the functional retinoblastoma protein pathway in hepato-cellular carcinoma cell lines. Cancer Lett 160:81–88.

Trapasso F, Krakowiak A, Cesari R, Arkles J, Yendamuri S, Ishii H,Vecchione A, Kuroki T, Bieganowski P, Pace HC, Huebner K,Croce CM, Brenner C. 2003. Designed FHIT alleles establishthat Fhit-induced apoptosis in cancer cells is limited by substratebinding. Proc Natl Acad Sci USA 100:1592–1597.

Uehara E, Takeuchi S, Tasaka T, Matsuhashi Y, Yang Y, Fujita M,Tamura T, Nagai M, Koeffler HP. 2003. Aberrant methylation inpromoter-associated CpG islands of multiple genes in therapy-related leukemia. Int J Oncol 23:693–696.

Virgilio L, Shuster M, Gollin SM, Veronese ML, Ohta M, HuebnerK, Croce CM. 1996. FHIT gene alterations in head and necksquamous cell carcinomas. Proc Natl Acad Sci USA 93:9770–9775.

Wilke CM, Guo SW, Hall BK, Boldog F, Gemmill RM, Chan-

drasekharappa SC, Barcroft CL, Drabkin HA, Glover TW. 1994.Multicolor FISH mapping of YAC clones in 3p14 and identifica-tion of a YAC spanning both FRA3B and the t(3;8) associated withhereditary renal cell carcinoma. Genomics 22:319–326.

Wu Q, Shi H, Suo Z, Nesland JM. 2003. 5�-CpG island methylationof the FHIT gene is associated with reduced protein expressionand higher clinical stage in cervical carcinomas. Ultrastruct Pathol27:417–422.

Xiao GH, Jin F, Klein-Szanto AJ, Goodrow TL, Linehan MW,Yeung RS. 1997. The FHIT gene product is highly expressed inthe cytoplasm of renal tubular epithelium and is down-regulatedin kidney cancers. Am J Pathol 151:1541–1547.

Yang Q, Nakamura M, Nakamura Y, Yoshimura G, Suzuma T,Umemura T, Shimizu Y, Mori I, Sakurai T, Kakudo K. 2002.Two-hit inactivation of FHIT by loss of heterozygosity and hy-permethylation in breast cancer. Clin Cancer Res 8:2890–2893.

Zanesi N, Fidanza V, Fong LY, Mancini R, Druck T, Valtieri M,Rudiger T, McCue PA, Croce CM, Huebner K. 2001. The tumorspectrum in FHIT-deficient mice. Proc Natl Acad Sci USA 98:10250–10255.

Zimonjic DB, Druck T, Ohta M, Kastury K, Croce CM, PopescuNC, Huebner K. 1997. Positions of chromosome 3p14.2 fragilesites (FRA3B) within the FHIT gene. Cancer Res 57:1166–1170.

Zochbauer-Muller S, Fong KM, Maitra A, Lam S, Geradts J, AshfaqR, Virmani AK, Milchgrub S, Gazdar AF, Minna JD. 2001. 5�CpGisland methylation of the FHIT gene is correlated with loss ofgene expression in lung and breast cancer. Cancer Res 61:3581–3585.

329METHYLATION AND LOSS OF FHIT EXPRESSION IN BL