Mutations in the DNA mismatch repair (MMR) system lead to an instability of simple repetitive DNA sequences involved in several cancer types. This instability is reflected in a high mutation rate of microsatellites, and recent studies in colon cancer indicate that defects in MMR result in frequent frameshift mutations in mononucleotide repeats located in the coding regions of BAX and transforming growth factor-β (TGF-β) receptor genes. Circumstantial evidence suggests that the MMR defect may be involved in some lymphoid malignancies, although several allelotype analyses have concluded on the low level of microsatellite instability in acute lymphoblastic leukemias. To further evaluate the implication of MMR defects in leukemogenesis, we have studied a series of 98 children with acute lymphoblastic leukemia and 14 leukemic cell lines using several indicators of MMR defects. Microsatellite markers were compared between blast and normal DNA from the same patients and mutations were sought in mononucleotide repeat sequences of BAX and TGF-β receptor II (TGF-β RII). The absence of microsatellite instability (MI) and the absence of mutations in the genes examined from patient's leukemic cells contrasted with the observation that half of the cell lines displayed a high degree of MI and that three of seven of these mutator cell lines harbored mutations in BAX and/or TGF-β RII. From these results we conclude that MMR defects are very uncommon in freshly isolated blasts but are likely to be selected for during the establishment of cell lines.
THE ONCOGENETIC effect of mismatch repair (MMR) defects is well established since the description of its involvement in hereditary nonpolyposis colorectal cancer (HNPCC) in 1993.1 The MMR system provides normal cells with a 100 to 1,000-fold increase in level of protection against mutations arising during DNA replication by correcting nucleotide mispairs and loops in newly synthesized DNA.2 The MMR system contains several independent components and four of these genes (hMSH2, hGTBP, hMLH1, and hPMS2) have been shown to be inactivated in various cancer cell lines.3
Tumor cell lines with an inactivated MMR system often, though not always,4 exhibit marked microsatellite instability (MI).4-7 However, because most microsatellites are noncoding, mutations of such sequences are thought to reflect MMR defects rather than participate in tumor development. A direct oncogenetic effect of a high mutation rate in simple repeat sequences has recently been suggested for the type II transforming growth factor-β receptor (TGF-β RII) gene8-11 and the BAX gene.12
TGF-β is a potent inhibitor of proliferation of a range of epithelial and endothelial cells, and it suppresses the growth of certain cancer cell lines.11 The gene coding for an essential TGF-β receptor (TGF-β RII) has been shown to exhibit frameshift mutations within a coding stretch of 10 adenines in several human malignancies.8-10
Somatic frameshift mutations in the BAX gene may also be selected for in colonic cancers exhibiting MI. These mutations were found in a stretch of eight guanine repeats.12
Recent findings suggest that MMR defects may also be involved in some hematological malignancies. Mice deficient in MSH2 or PMS2 exhibit a marked MI and develop lymphomas at an early age.13,14 A recent study of 10 human lymphoblastic lymphomas showed two cases with mutations in the coding region of the hMSH2 gene.15Inactivating mutations of the hMLH1 gene were detected in 3 of 43 cell lines derived from lymphoid leukemias. Those cell lines that failed to express hMLH1 showed MI.16 Finally, MI was reported in a child with T-cell acute lymphoblastic leukemia (T-ALL) and determination of the allelic status at six MMR loci showed a hemizygous deletion in the gene coding for hPMS2.17 In addition to these studies indicating direct evidence for MMR inactivation, some studies have reported evidence of MI, although at low frequencies, in hematological neoplasms.17-22
To further evaluate the implication of MMR defects in leukemogenesis, we have studied a series of 98 children with acute lymphoblastic leukemia and 14 leukemic cell lines using several indicators of MMR defects. Microsatellite markers were compared between blast and normal DNA from the same patients and mutations were sought in mononucleotide repeat sequences of BAX and TGF-β RII. DNA was analyzed for MI using five microsatellites: two trinucleotide repeats that are known to be frequently mutated (AR and DM-1)23 and three tetranucleotide repeats that, in general, are thought to be more sensitive to replication errors (D5S1460, D11S1294, and D12S391).23 24 We also screened these samples and cell lines for frameshift mutations in the BAX and TGF-β RII genes.
MATERIALS AND METHODS
Cell lines.
We studied 14 cell lines: 5 B-lineage ALL cell lines (697, NALM6, REH, TOM-1, CCRF-SB), 5 T-leukemia cell lines (MKB1, MOLT-13, CEM, CCRF-HSB-2, CCRF-CEM), 2 Burkitt's lymphoma cell lines (DAUDI, RAJ1), 1 chronic myeloid cell line (K562), and 1 erythroleukemia cell line (HEL). The cell lines 697, NALM6, REH, MKB1, DAUDI, RAJ1, K562, and HEL were obtained by the Deutsche Sammlung Von Mikroorganismen und Zellkulturen (DSM, Braunschweig, Germany). CCRF-CEM, CCRF-HSB-2, and CCRF-SB were obtained from the American Type Culture Collection (ATCC; Rockville, MD). DNA was isolated using a DNA isolation kit (G Nome; Bio 101, Vista, CA) and stored at −20°C until analysis.
Patients.
A total of 98 patients was studied. Blast samples were obtained either at diagnosis (63 B-lineage ALLs, 19 T-ALLs) or in relapse (13 B-lineage ALLs, 3 T-lineage ALLs). Patients were classified on the basis of their lymphoid morphology and their immunophenotype as assessed by flow cytometry.
Sample collection and preparation of mononuclear cell lysates.
Bone marrow was collected on EDTA before induction therapy, during clinical remission, and, if necessary, before induction therapy in relapse cases according to the European Organisation for Research and Treatment of Cancer (EORTC) 58881 treatment protocol. Mononuclear cells were separated by gradient centrifugation over Ficoll (Lymphoprep; Pharmacia, Uppsala, Sweden) and lysed as previously described.25 The lysates were stored at −20°C until analysis.
Microsatellite typing.
For each patient we compared tumor DNA with DNA obtained in complete remission. The first screening in 63 B-ALL patients and 3 B-ALL cell lines was conducted with 247 CA microsatellite markers from Genethon (Evry, France),26 evenly distributed over the genome. The complete list of markers is available on request. Polymerase chain reaction (PCR) and microsatellite analysis were performed at Genethon using primers labeled with fluorescent dyes and an automated DNA laser sequencer (Applied Biosystems, Foster City, CA). A second group of 55 ALL patients and 11 cell lines was analyzed with five markers.23,24 Two trinucleotide repeat markers (AR, DM-1) and 3 tetranucleotide repeat markers (D5S1460, D11S1294, D12S391) were obtained from Genset (Paris, France) according to sequences available from the Genome data base (http://gdb.infobiogen.fr). One of each primer was 5′-fluorescein labeled. PCR was performed on 5 μL of lysates of cell lines and bone marrow cells containing 30 ng of DNA, and PCR products were analyzed using a fluorescent automated laser DNA sequencer (A.L.F.; Pharmacia) as described.27
BAX and TGF-β RII gene analysis.
A 101-bp region encompassing the (G)8 tract in the BAX gene was amplified by PCR with primers 5′-TTC ATC CAG GAT CGA GCA GGG CG-3′ (5′-fluorescein–labeled) and 5′-GAC ACT CGC TCA GCT TCT TGG TG-3′. An 86-bp region encompassing the (A)10 tract in the TGF-β RII gene was amplified with primers 5′-ATG CTG CTT CTC CAA AGT GCA TTA-3′ (5′-fluorescein–labeled) and 5′-GCA CTC ATC AGA GCT ACA GGA ACA-3′. PCR consisted of 35 cycles (20 seconds at 94°C, 20 seconds at 60°C), and a final incubation was performed at 70°C for 1 hour to allow for the addition of an extra deoxyadenosine to all amplified chains. Before loading on the ALF DNA sequencer (Pharmacia), a 120-bp fluorescent marker was added to all samples for proper alignment.
DNA sequencing.
All samples exhibiting an abnormal size of the BAX or TGF-β RII gene fragments were sequenced. Unlabeled primers with the same sequence as described above were used to amplify the fragments of the BAX and TGF-β RII genes. PCR products were ligated into a pGEM-T vector (Promega, Paris, France) and subcloned using DH5α competent bacteria (GIBCO-BRL, Gaithersburg, MD) to be able to separately analyze the product from both alleles. Plasmid DNA was sequenced by the dideoxy method using an autoread sequencing kit (Biotech, Pharmacia, Uppsala, Sweden) and the ALF DNA sequencer.
RESULTS
Microsatellite instability.
In a first set of experiments, we compared the allelic profile of 247 dinucleotide repeat markers in blast cells isolated from bone marrow samples and in normal DNA from 63 patients with B-lineage ALL in a genome-wide allelotyping. Using microsatellite typing (14,820 genotypes at 247 loci), differences between blast and normal DNA were detected only in four patients. Three cases showed an extra allele at one locus in the blast samples and one case showed one extra allele at two loci. Surprisingly, two (NALM6 and REH) of three B-ALL cell lines that we included in this screening exhibited multiple alleles for about 30% of the markers. To get a more complete overview of the frequency of MI in leukemic cells and cell lines, we studied 11 additional cell lines and 55 tumor and remission sample pairs derived from B-ALL and T-ALL patients either at diagnosis (20 B-ALLs and 19 T-ALLs) or at relapse (13 B-lineage and 3 T-lineage ALLs) using five highly polymorphic microsatellite markers: 2 trinucleotide and 3 tetranucleotide repeats that may be more sensitive to MI than dinucleotide repeats.23 24 The 20 B-ALLs studied at diagnosis in this second set of experiments were randomly chosen from the series already studied with dinucleotide repeat markers. Five of these 11 cell lines (CEM, CCRF-CEM, MKB1, MOLT13, CCRF-HSB1) showed microsatellite alterations (Table 1) for one or several markers. MI usually appeared as multiple (four or more) alleles in these cell lines. Microsatellite alterations were extremely common in T-ALL cell lines (five of five). In contrast, none of the 55 tumor samples showed microsatellite alterations.
BAX and TGF-β RII frameshift mutations.
Subsequent screening for frameshift mutations in the BAX and TGF-β RII genes was performed with primers encompassing the stretch of mononucleotide repeats in each gene.13 28 Analysis showed PCR products of normal size in the 55 bone marrow samples. In three cell lines, fragments of altered size were detected. Two cell lines (NALM-6, MKB-1) had frameshift mutations in BAX and TGF-β RII, while a third cell line (REH) was mutated in the TGF-β RII gene only. All PCR fragments with altered mobility were cloned and sequenced to confirm the presence and type of mutation. All mutations consisted of one or two nucleotide deletions in the mononucleotide repetitive sequences (Table 1). NALM-6 was mutated on both alleles of the BAX and TGF-β RII genes. MKB-1 and REH had a mono-allelic deletion in the TGF-β RII gene. The BAX gene in MKB-1 contained two different deletions and a normal allele. It is noteworthy that this cell line is near tetraploid. All these mutations were confirmed in at least two bacterial clones.
DISCUSSION
Numerous studies have documented different forms of genomic instability in cancer cells. MMR defects underlie a very common form of instability that was recently characterized following the observation of microsatellite alterations in tumor samples. In contrast to mutations in classical tumor suppressor genes (TSG), such as retinoblastoma, whose germline transmission predisposes to familial cancers, MMR gene defects do not provide mutated cells with a direct selective advantage but rather increase the probability of mutations in other genes that otherwise would occur at low frequency. Theoretical considerations as well as current observations suggest that MMR defects, like other forms of genome instability, may not always be required for tumor initiation or progression but increase the evolution rate of tumor cells and consequently are often selected for a certain stage of development of many tumors.2,7 This selection is only indirect and results from the fact that MMR defects have, in turn, led to the inactivation of TSG and this contributes to tumor development during clonal evolution. TSG containing simple repeat sequences are thought to be frequently inactivated in MMR-defective tumors because defects in the MMR system result in a high rate of genetic alterations in simple repeat sequences like those in microsatellites.9 12
Since their description in HNPCC, MMR defects have been studied in several other human cancers,2 and recent reports indicate a possible involvement in hematological malignancies.13-17 In a genome-wide screening of 63 B-ALL patients, and 3 B-lineage leukemic cell lines with 247 microsatellite markers, we detected a high rate of microsatellite alterations (30%) in two out of three cell lines. However, in the cohort of B-ALL patients we found only three patients with one microsatellite alteration and one patient with two MI. This frequency (0.34 × 10−3) is even lower than that observed in parent to offspring transmission studies (1.2 × 10−3).24 Not only the number, but also the pattern, of alterations in microsatellites differed between cell lines and patient samples. Cell lines exhibited multiple alleles in comparison with only one extra allele in the bone marrow samples. Therefore, it is most likely that the rare observation of microsatellite mutations in freshly isolated blasts rather reflects the clonal nature of the cell population than instability of the microsatellites. The subsequent screening of 11 additional cell lines and 55 ALL patients with five microsatellite markers sensitive to MI23,24 showed no alterations in any of the 55 bone marrow samples but clear alterations in 5 of 11 cell lines. BAX and TGF-β RII frameshift mutations were present in three of the seven cell lines exhibiting MI, whereas neither the other cell lines nor any of the patients had frameshift mutations in one of these genes. From this we conclude that MMR defects are present in a subset of ALL cell lines but are very uncommon in vivo. Thus, the involvement of MMR deficiency in ALLs may have been previously overestimated because cell lines were often used as experimental material. Another confusing factor has been the interpretation of rare microsatellite alterations as being MI caused by MMR defects.21
The discordance between freshly isolated blasts and leukemic cell lines is puzzling. In a population of tumor cells, as in a bacterial population, the overall selective pressure on genome stability results from a balance between the advantages of a potential for rapid evolution when confronted by a changing environment and the detrimental effects of a high mutation rate on cell physiology. In this regard, it is likely that the establishment of cell lines represents a dramatic change in the environment of a tumor cell. Consequently, it is reasonable to postulate that a small fraction of MMR-defective cells that remains undetected among the leukemic population, because it is usually not selected for in vivo, will acquire a selective advantage in vitro from its ability to quickly generate mutants that adjust to the new environment. It should be noted that if the freshly isolated lymphoblastic cells are heterogeneous, a mutation that is present in less than 10% of the population of cells will be likely to remain undetected. Our present finding that an MI phenotype is very uncommon in childhood ALLs in vivo does not exclude the involvement of MMR defects in rare occasions of ALL as described by Baccichet et al17 for a T-ALL patient.
Supported in part by grants from the Ligue Nationale Contre le Cancer, the Fondation de France, the Association pour la Recherche contre le Cancer, and the Délégation à la recherche Clinique de l'Assistance Publique, Hôpitaux de Paris.
Address reprint requests to Bernard Grandchamp, MD, PhD, INSERM U409 Faculté de Médecine Xavier Bichat, BP 416, 75870 Paris Cedex 18, France; e-mail: bgrandch@bichat.inserm.fr.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" is accordance with 18 U.S.C. section 1734 solely to indicate this fact.
