Hodgkin's lymphoma (HL) is characterized by the presence of neoplastic Hodgkin and Reed-Sternberg (HRS) cells scattered in a background of lymphocytes, histiocytes, plasma cells and polymorphonuclear leukocytes.1 Due to the scarcity of HRS cells in affected tissues, the genetic events involved in the development and progression of HL are still poorly understood.1 The identification of these events is essential for understanding that the pathogenesis of HL is of clinical and therapeutic importance. Indeed, most HL cases initially respond to radiotherapy and/or chemotherapy. However, 5–10% of early- and 20–40% of advanced-stage patients either progress during treatment or relapse following an initial remission.2 Although different clinical risk factors have been identified, including age, presence of B symptoms and stage, pretreatment identification of patients at high risk of relapse or treatment resistance remains difficult.3 Among the cytogenetic anomalies described in HL, changes within the 11q23 band have been frequently reported.4 Using microsatellite polymerase chain reaction (PCR) with DNA from microdissected HRS cells, Hasse et al.5 found a clonal allele loss at the 11q23 locus in 3 out of 5 HL cases examined. Interestingly, the 11q23 region contains a number of candidate genes that may be involved in lymphomagenesis, in particular the ATM gene.4 The ATM gene product is a serine-threonine kinase that plays a central role in the cellular response to DNA damage. Germline mutations in the ATM gene cause ataxia-telangiectasia (A-T), a rare autosomal recessive multisystem disorder associated with a high incidence of leukemias and malignant lymphomas including Hodgkin's lymphoma in individuals carrying 2 altered ATM alleles.6, 7 Somatic mutations in the ATM gene have been identified in some subtypes of T- and B-cell lymphomas/leukemias.8 Based on the published data available, we decided to explore the loss of heterozygosity (LOH) at the ATM gene locus (11q22-23) using 3 microsatellite markers located within and near the ATM gene.5, 9, 10 To overcome the difficulty related to the predominant population of nonneoplastic cells surrounding HRS cells, the latter cells were microdissected and analyzed by the single-cell PCR technique.11, 12 Fifteen cases were investigated and the results of LOH detection correlated with the expression of the ATM gene product by immunohistochemistry in 8 cases. In addition, the status of the ATM gene was examined in 2 cases showing LOH at the ATM gene locus using whole genome amplification and direct sequencing techniques on microdissected HRS cells.
Hodgkin's lymphoma (HL) is a lymphoid malignancy characterized by the presence of rare neoplastic cells, Hodgkin and Reed-Sternberg (HRS) cells, scattered among a predominant population of inflammatory cells. On the basis of previously reported cytogenetic analyses, the ATM (ataxia-telangiectasia mutated) gene at 11q22-23 has been implicated in the etiology of HL. We therefore developed a single-cell PCR approach to detect ATM loss of heterozygosity (LOH) in HRS cells. Three microsatellites were investigated; 1 localized inside the ATM gene and the remaining 2 in close proximity. In 2 of the 15 lymph node samples, an allelic loss of the ATM gene locus was detected. ATM protein expression was examined in 8 cases (including 1 of the 2 cases with LOH) by immunohistochemistry. In the case associated with an allelic loss, the ATM protein was absent in the HRS cells, whereas in the 7 remaining cases, without detectable LOH at the ATM locus, nuclear ATM expression was observed. In the 2 HL cases with LOH, the ATM gene was sequenced following whole genome amplification of DNA isolated from microdissected HRS cells. In 1 of these 2 cases, a splice site mutation in the second ATM allele was found. This mutation could generate a premature termination codon leading to a marked instability and a rapid degradation of the resulting ATM mRNA transcripts. This latter event could explain the loss of the expression of the ATM protein in HRS cells as detected by immunohistochemistry in this particular case. As previously reported in some B-cell lymphomas, our results suggest that ATM genetic anomalies could play a role in the pathogenesis of a subset of HL cases. © 2004 Wiley-Liss, Inc.
Material and methods
Fifteen cases of HL were randomly selected from the files of the Pathology Department of the Purpan Hospital in Toulouse, France. For all these cases, frozen and paraffin-embedded tissue samples were fixed in 10% neutral buffered formalin or Duboscq-Brasil (ethanol-based Bouin's fluid). The diagnosis of HL subtypes was made by established immunomorphologic criteria.13 Of the 15 cases studied, 8 were nodular sclerosis (NS) and 7 were mixed cellularity (MC). The phenotype, genotype and Epstein-Barr virus (EBV) status of these cases were determined as previously described (Table I).14, 15 In addition, 2 cases of mantle cell lymphoma (MCL) were also included in this study as a positive and negative control for the validation of the ATM allelotyping and protein expression studies. One sample had been previously shown to have an allele loss in the 11q22-23 region and lacked the expression of the ATM protein as detected respectively by comparative genomic hybridization (CGH) and Western blot analysis.16 The second case was chosen on the basis of a strong expression of the ATM protein as detected by immunohistochemical analysis.
Single-cell PCR amplification of microsatellites at ATM gene locus in HRS cells
Frozen sections of 8 μm were mounted on ChemMate slides (DakoCytomation, Trappes, France) and immunostained with monoclonal antibodies (mAbs) directed against the CD30 antigen (HSR4; dilution: 1/100; Beckman-Coulter, Marseilles, France) to visualize HRS cells. Antibody binding was detected by the APAAP technique.17 Due to the expression of the CD30 antigen by both HRS cells and reactive immunoblasts, only clearly atypical CD30+ cells were selected for further experiments. The HL cases positive for EBV were determined using an mAb directed against the latent membrane protein (LMP-1; anti-LMP-1 antibody CS1-4; dilution 1/25; DakoCytomation). Normal lymphocytes were immunostained with the anti-CD3/F7.2.38 mAb (DakoCytomation) and visualized using the same method. Tissue sections from the MCL cases were labeled with anti-CD20/L26 (dilution 1/50; DakoCytomation) and anti-CD3/F7.2.38 mAbs (hybridoma supernatant) for neoplastic and normal lymphoid cells, respectively. Once the frozen sections were covered with PBS buffer, single cells were picked under the microscope with a closed capillary glass using a hydraulic micromanipulator and then transferred by aspiration to an open capillary glass with another hydraulic micromanipulator.11 Each isolated cell was placed in a tube containing 12 μl extraction solution (50 mM Tris-HCl, pH 8, 10 mM EDTA, 100 mM NaCl, 200 μg/ml proteinase K, PCR-grade; Roche Diagnostics, Mannheim, Germany). After an overnight proteinase K digestion at 37°C followed by 10-min inactivation of the proteolytic enzyme at 95°C, the DNA was either stored at −20°C or directly used for PCR amplification. All cases were analyzed at least twice in completely independent cell isolation procedures and PCR assays. The buffers used to cover the tissue sections during the cell isolation process were used as negative controls in subsequent steps.
Loss of heterozygosity analysis at ATM gene locus
To evaluate LOH at the ATM locus, we used primers flanking dinucleotide (n = 2) and tetranucleotide (n = 1) microsatellite repeat polymorphisms located in the 11q22-23 chromosomal region (Table II). In addition to the microsatellite marker D11S1294,5 which is located approximately 200 Kb telomeric from the ATM gene,18 2 other microsatellites were examined: D11S2179, which is located between exons 62 and 63 in the ATM gene,9 and D11S1778, which is situated between the ATM gene and the D11S1294 microsatellite10 (Table II). The LOH status of the HRS and neoplastic mantle lymphoma cells was determined by comparing in each case the allelic profile of the malignant cells and normal lymphocytes.
|Name||Repeat primer motif||Sequence 5′-3′1||Size (bp)2||Annealing temperature, °C||MgCl2/mM||References|
|NT033899 at UniSTS 1502403|
|2.2179 EXT||GAGTCCAGCCTAGGCAATACAGC||Personal design|
|3.1778 INT||CGTTTNAATGGGGTTCTGG||Personal design|
A nested PCR was carried out in a total volume of 50 μL with AmpliTaq gold buffer (Applied Biosystems, Courtaboeuf, France) supplemented with 2.5 or 3.5 mM MgCl2, 0.8 mM dNTP, 25 pM of each primer, 50 ng 5S ribosomal RNA from E. coli (Sigma-Aldrich, Poole, U.K.) and 1.25 U AmpliTaq gold DNA polymerase (Applied Biosystems). The first round of amplification was performed by using the 2 external primers EXT and CTT listed in Table II. For the second step, one external primer EXT was replaced by an internal primer INT, which was labeled at its 5′ end by a fluorophore (Fam, Tet, or Hex fluorescent dyes). The PCR conditions consisted of a denaturation step at 94°C for 1 min followed by an annealing step at variable temperature and variable time, depending on the set of primers used (Table II), and an extension step at 72°C for 1 min for 35 cycles. For all PCR amplifications, a pre-PCR heating step at 95°C for 10 min to activate the AmpliTaq Gold enzyme and to denature the DNA templates and a final period of 10 min at 72°C to complete the reaction were performed after the 35 cycles. For the second PCR, 5–10 μL of the PCR products were used as template under the same cycling conditions (Table II). PCR products were first checked by a 6–8% nondenaturing polyacrylamide gel electrophoresis and visualized by staining with ethidium bromide.14 The second PCR products were then separated on 5% denaturing polyacrylamide gels using the automated sequencer (ABI Prism 377; Applied Biosystems) and analyzed using the Genescan and Genotyper software programs (Applied Biosystems). Tamra- labeled Genescan 2500 molecular standard weight (Applied Biosystems) was included as internal size standards in each run. All samples in which 2 distinct alleles of approximately similar intensity were present in the normal DNA were considered to be informative. LOH was scored by visual detection of the complete absence of one tumor allele in heterozygous (i.e., informative) cases. For each case and each microsatellite, a total of 12 HRS single cells or 12 neoplastic mantle lymphoma cells and 12 single lymphocytes were analyzed and this analysis was repeated twice.
Detection of ATM gene product expression by immunohistochemistry
The ATM protein expression was studied with the Ab-2/ OP90 mAb (dilution 1/150) from Oncogene Research Products (France Biochem, Meudon, France) directed against an epitope within amino acids 368–380 of the human protein. This study was performed on 8 HL cases for which formalin-fixed and paraffin-embedded tissue sections were available; this antibody was not suitable for use on tissue sections fixed in Duboscq-Brasil or on frozen sections (data not shown). An antigen retrieval method was used and involved heating the tissue sections at 95°C in a water bath (DakoCytomation) with sodium bicarbonate 10 mM (pH 6) supplemented with 1% Tween 20 for 40 min. Staining was performed by the streptavidine-biotin-peroxidase complex (ABC) method using the streptABComplex/HRP duet (mouse/rabbit) kit (number K492; Dako, Glostrup, Denmark). Cells were considered positive when the staining was observed in the nuclei.
Whole genome amplification of DNA from isolated single cells
Whole genome amplification (WGA) was performed using the GenomiPhi WGA kit (Amersham Biosciences, Piscataway, NJ) according to the manufacturer's instructions with some modifications. Two HL cases showing LOH at or near the ATM gene locus (cases 3 and 6) were investigated. For each case, 20 neoplastic HRS cells and normal lymphocytes were microdissected and transferred into PCR tube containing 9 μl of the sample buffer with 200 μg/ml proteinase K for DNA isolation as described above. The resulting mixture was heat-denatured at 95°C for 3 min and then cooled in ice for an additional 5 min; 10 μl of the reaction mix (9 μl of reaction buffer and 1 μl of ϕ29 DNA polymerase) were added to each tube and incubated at 30°C for 18 hr. After heat inactivation of the enzyme at 65°C for 10 min, each WGA sample was either stored at −20°C or directly used for exon-specific PCR amplifications.
ATM exon PCR and TA cloning
PCR conditions were established to amplify the 2 promoter regions and individually 65 of the 66 exons of the ATM gene and their flanking intronic sequences. PCR amplifications were performed in 50 μl of reaction mixture containing 1 μl of the WGA amplified product, 0.2 mM each dNTP, 2.5 mM MgCl2 and the reaction buffer A (Promega, Charbonnières, France). Primer sequences for promoters 1 and 2 and for exons 1a, 1b, 2, 51, 53, 54 and 60 were those described by Castellvi-Bel et al.19 and for the other exons by Sandoval et al.20 Each was used at 0.5 μM. Forty PCR cycles were performed and consisted of 1-min denaturation at 94°C, 1-min annealing at 54 or 59°C, depending on the used primer pair,19, 20 and 1-min elongation at 72°C steps. For all PCR amplifications, a pre-PCR heating step at 95°C for 5 min before adding the 2 U Taq DNA polymerase (Promega) and a final period of 10 min at 72°C at the end of the 40 cycles to complete the PCR reaction were performed. Each product was first purified with Wizard PCR Preps DNA purification system (Promega) and then cloned in pGEM-T easy vector as described by the manufacturer (Promega), which was used to transform UltraMAX DH5α-FT competent cells (Invitrogen, Cergy Pontoise, France). The selected cell colonies for each exon-specific vector were amplified by PCR using universal primers T7 and SP6. PCR amplification products were checked by separation on 1.5% agarose gel electrophoresis. The bacterial cellular colony, which corresponded to the anticipated size on agarose gel, was expanded and the plasmids were purified using Plasmid Midi Kit (Qiagen, Courtaboeuf, France).
ATM sequence analysis
Direct sequencing of purified plasmids was performed by cycle sequencing with ABI Prism BigDye Terminator Chemistry V3.1 (Perkin Elmer, Oak Brook, IL) using the universal primers T7 and SP6 followed by electrophoresis on an automated Applied Biosystems 3100 genetic analyzer. The sequence obtained was compared with published germline sequence of ATM (Genbank U82828) using the Sequence Navigator software program.
Loss of heterozygosity at ATM gene locus
The analysis of the 3 microsatellite profiles in the 15 HL cases studied showed an LOH in the region of the ATM gene in 2 cases (cases 3 and 6). In one case (case 3), LOH was demonstrated in the D11S1778 and D11S2179 microsatellites, whereas the third microsatellite (D11S1294) showed a normal biallelic profile. In case 6, the LOH was located at the D11S1294 microsatellite (Fig. 1a), whereas the D11S2179 showed a normal biallelic profile and the D11S1788 was noninformative; thus, the exact limits of the LOH cannot be determined but could include the ATM gene itself. All microsatellite profiles were normal in the 13 remaining cases (Fig. 1c). As expected, an allelic loss was found in the D11S2179 micrcosatellite in the DNA from the patient with mantle cell lymphoma, which had been previously shown to carry an allelic loss in the 11q22-23 locus by CGH (Fig. 2a). In the DNA from the patient with mantle cell lymphoma, 2 other microsatellites were homozygous, i.e., noninformative for the study of LOH. In the second mantle cell lymphoma sample, the 3 microsatellites showed either a normal/biallelic [D11S2179 (Fig. 2c) and D11S1294 microsatellites] or noninformative/homozygous (D11S1778) profile. Of note is that in the samples that were heterozygous for certain markers, the 2 alleles were not present in all PCR products. Indeed, it was noted that in some PCR products, 1 of the 2 alleles, never the same one, was not amplified from the HRS cells and/or the lymphocytes. Some microsatellites produced stutter/shadow bands, which are well-known PCR artifact.21 Analysis of the microsatellites D11S2179 and D11S1778, which were 2 bp repeats, gave results that were sometimes difficult to interpret due to the formation of such bands (Figs. 1c and 2a and c). The longer 4 bp repeat marker (D11S1294) was less prone to this artifact and thus the alleles were much easier to distinguish (Fig. 1a).22
Correlation between LOH at ATM gene locus and ATM protein expression
ATM protein expression analysis was performed in 8 cases (including 1 case with LOH) for which formalin-fixed and paraffin-embedded tissue sections were available. In all cases, small lymphocytes were strongly positive and used as positive internal control (Fig. 1b and 1d). In 6 cases, the HRS cells showed a strong nuclear ATM staining using the Ab-2/OP90 mAb (Fig. 1d). No LOH was detected in any of these cases. In one case (case 6), which had associated LOH, no nuclear ATM staining was seen in the HRS cells while a positive nuclear staining in the surrounding small lymphocytes was observed (Fig. 1b). In the remaining case (case 15), the nuclear staining of HRS cells was weak and associated with a weak to moderate cytoplasmic immunostaining. No LOH was demonstrated in this case. A positive nuclear staining in the surrounding small lymphocytes was observed in this sample. As expected, the case of mantle cell lymphoma associated with allelic loss in the ATM locus showed lymphoma cells negative for the Ab-2/OP90 mAb (Fig. 2b). By contrast, in the mantle cell lymphoma case negative for LOH detection, neoplastic cells were strongly positive with the latter antibody (Fig. 2d).
Correlation between LOH at ATM gene locus and clinical data
Clinical details were available for 10 of the HL patients, including the 2 cases with LOH. The allelotyping results and immunohistochemical findings were analyzed with respect to stage, response to treatment and relapse. No significant difference between these 2 cases with LOH and the remaining 8 cases of HL was observed.
ATM sequence analysis
In case 6, a splice site mutation IVS61-2A→C was identified in intron 61 in the proximity of exon 61 encoding for a part of the PI-3 kinase domain of the ATM protein. This sequence variant IVS61-2A→C modified the invariant splice acceptor site consensus sequence AG (Fig. 3a). In case 3, no mutation was found but we noted the presence of an ATM sequence variant F570S, described previously in a diffuse large B-cell lymphoma.23
Recurrent chromosomal anomalies play an important role in many human lymphoid tumors and are often responsible for the deregulation of genes involved in the control of cell survival and proliferation.24 In HL, cytogenetic analysis has yielded conflicting results probably due to the small number of neoplastic HRS cells, showing rare mitosis in a large background of nonneoplastic lymphocytes and inflammatory cells.25, 26 In this context, genetic studies using microsatellite markers may be very helpful to demonstrate chromosomal anomalies that are not detected by standard cytogenetic techniques.27, 28 In the present study, using HRS cells isolated through micromanipulation, samples from 15 cases of HL were analyzed for LOH with a panel of 3 microsatellite markers located at the ATM locus. Single-cell PCR has been shown to be a powerful tool for clarifying the cellular origin and clonality of HRS cells.29, 30 Using this technique, LOH was demonstrated in 2 cases (13.5% of cases studied). In addition to technical problems related to the formation of stutter bands, we noted other technical problems that are now well known in PCR-based DNA analysis of single cells such as allele dropout (ADO) and preferential amplification of shorter alleles, which may lead to misinterpretation of PCR results.31, 32 In addition, by using frozen sections of 8 μm in thickness, PCR products are not always representative of the total DNA in the cell. This finding can be explained by the fact that in many instances part of the cell nucleus is missing and this occurs more commonly in large cells, such as HRS cells compared with small cells such as lymphocytes.29 This latter artifact and ADO are thought to be random processes affecting either one of the alleles and by carrying out the analysis of several single cells from the same tissue genetic status can be determined.29, 33, 34
The results of the present study add to the growing literature addressing genetic and clinical aspects of HL. LOH at the ATM locus suggests that ATM gene anomalies could contribute to the development of a subset of HL cases. In 2 studies, HRS cells and bystander lymphocytes were laser-microdissected, pooled from tissue sections or cytospins5, 35 and subsequently studied for LOH using microsatellite markers either scattered over the entire genome35 or localized at candidate loci.5 These studies indicate a relatively high frequency of clonal chromosomal deletions in HRS cells at certain loci such as 6q25,35 1q42, 4q26, 9p23 and 11q22-23.5 However, it should be noted that the ATM gene was not specifically investigated in these studies as the markers used were not located with the gene itself. In a recent study by Liberzon et al.,36 tumors from 23 children with Hodgkin disease were screened for ATM mutations and LOH. Four out of 12 informative patients exhibited LOH centomeric to the ATM gene, with LOH covering the whole ATM region in 1 patient. Thus, it would appear that while LOH is found in a subset of patients with Hodgkin's disease, it may not be as common an event as in other lymphoid malignancies,37 such as B-cell chronic lymphocytic leukemia and mantle cell lymphoma, in which ATM deletions have been detected in 20% and 50% of cases, respectively.38, 39 In B-cell chronic lymphocytic leukemia, the 11q22-23 deletion seems to be associated with a more aggressive disease and poor survival.40
ATM protein expression has been examined in relatively few HL cases. In the present study, nuclear ATM expression in the HRS cells was detected in 6 of the 8 cases examined, suggesting that loss of ATM expression is not a frequent event. Starczynski et al.,41 using an immunohistochemical approach, have examined variations in ATM protein expression during normal lymphoid development and among B-cell-derived neoplasias. They found that ATM expression was modulated during lymphoid development with variations in expression between lymphoid subsets of different maturity. In 5/25 classic HD cases examined, ATM expression was noted in the HRS cells. Interestingly, there was an apparent correlation between ATM expression and the stage of differentiation; all the ATM-positive tumors expressed a post-GC phenotype as assessed by their expression of CD138, whereas the majority of ATM-negative tumors were either negative or showed a low CD138 expression. However, they did find a single tumor that strongly expressed CD138 but lacked ATM expression, which may represent a post-GC subgroup that carry inactivating ATM mutations. The presence of molecular variants in the ATM gene in individuals with Hodgkin's lymphoma has been examined by Offit et al.,42 Takagi et al.43 and Liberzon et al.36 In the study by Liberzon et al.,36 missense variants were detected in 2 out of 23 childhood Hodgkin tumors. These variants were found in the germline and the 2 patients had a more aggressive clinical course of the disease. Offit et al.42 observed multiple germline variants in patients with HD compared with the normal population and Takagi et al.43 identified single nucleotide polymorphisms of the ATM gene in 5 of 14 cases of pediatric HD. Two of these variants altered ATM function and support the involvement of ATM variants in the pathogenesis of HD.
In the present study, the concomitant loss of the ATM protein expression in the one case showing LOH suggests a biallelic inactivation of the ATM gene. Similar findings have been described in some B-cell lymphomas.16, 44 In the current study, the sequence analysis of the remaining ATM allele in the 2 cases of HL with LOH at the ATM gene locus enabled us to detect in 1 case a splice site mutation in the intronic region adjacent to exon 61 that codes for the PI-3 kinase domain of the ATM protein. Splicing mutations are a prevalent and important class of mutations in A-T patients (see, for example, Teraoka et al.45 and http://www.benaroyaresearch.org/bri_investigators/atm.htm). Without analyzing the ATM mRNA/cDNA, the direct consequences of these mutations are difficult to predict46 and such an analysis could not be performed on these samples because of the lack of frozen tissue specimens for these specific samples. However, 2 hypotheses could be considered on the effect of this mutation on the ATM mRNA transcripts and/or protein. One possibility is that the IVS61-2A→C change is responsible for the skipping of exon 61 (Fig. 3b). The outcome of such an event would be the transcription of an mRNA lacking the exon 61 and the synthesis of truncated nonfunctional protein. This abnormal ATM protein could be unstable and quickly targeted for degradation via the proteasome pathway.47, 48 Alternatively, the IVS61-2A→C mutation could induce a deletion of several nucleotidic sequences leading to the activation of 1 of the 8 potential cryptic splice sites localized at nucleotide positions 8612, 8621, 8639, 8642, 8646, 8649, 8664 and 8670 downstream of this point mutation as previously described by Eng et al.46 (Fig. 3c). In this context, the potential activation of the first 3 of these cryptic sites results in premature termination codon (PTC) in exon 61. The resulting mRNAs could be detected and destroyed by the mRNA surveillance pathway (i.e., the nonsense-mediated decay)49 leading to the absence of the ATM protein. In conclusion, the demonstration of ATM gene anomalies in a subset of HL cases is in agreement with previous reports concerning other B-cell lymphoproliferative disorders. However, further studies are necessary to determine the actual role of ATM gene anomalies in the pathogenesis of HL as well as their prognostic value.
The authors thank Mrs. Marie-Andrée Daussion, Mr. Michel March, Mr. Daniel Roda and Mr. Xavier Sicard for their excellent technical assistance and Dr. Sandra Angèle for helpful discussions.