Mutator Phenotype in Human Hematopoietic Neoplasms and Its Association With Deletions Disabling DNA Repair Genes and bcl-2 Rearrangements

We analyzed 66 samples of tumor DNA obtained from the Third Medical Clinic of the University of Heidelberg (Heidelberg, Germany) and from the Divisione di Ematologia of the University of Verona (Verona, Italy). These samples corresponded to 45 cases of leukemia (29 AML, 16 acute lymphoid leukemia [ALL]), and 21 of non-Hodgkin’s lymphoma (NHL), as detailed in Table 1. Tumor DNA was extracted from bone marrow biopsies or peripheral blood in the case of leukemias and from bone marrow or lymph nodes in the case of NHL. Control constitutional DNAs were extracted either from the leukocytes or bone marrow of the same patients during remission or from lymphoma-unaffected bone marrow samples of some NHL patients. The AMLs were classified morphologically according to the French-American-British classification,20 and the NHLs according to the Revised European-American Lymphoma (REAL) classification.21 ALLs were classified according to their immunophenotype.22 High-molecular-weight DNA was isolated by standard procedures.23 Cytogenetic analysis was performed on short-term cultures of bone marrow or peripheral blood cells. Chromosomes were analyzed using a modified GAG-banding technique, as described elsewhere.24 

DNA was amplified by polymerase chain reaction (PCR), as previously described,8 in a volume of 25 μL containing 0.1 μg of genomic DNA as template, 50 mmol/L KCl, 10 mmol/L Tris-HCl (pH 8.3), 1.5 mmol/L MgCl₂, 0.2 mmol/L of each deoxynucleoside triphosphate, 0.4 μmol/L of each sense and antisense primer, and 1 U of Taq Polymerase (Perkin Elmer, Norwalk, CT). The primer pairs listed in Table 2 were used. Forward primers were end-labeled with [γ³³P] adenosine triphosphate (ATP) for 1 hour at 37°C using T4 polynucleotide kinase (Amersham, Little Chalfont, UK). Reaction conditions consisted of 1 minute at 94°C, 30 seconds at 55°C, and 30 seconds at 72°C for 30 cycles. PCR products were electrophoresed on denaturing 6% polyacrylamide formamide-containing gels. After electrophoresis, the gel was dried and exposed to x-ray film at −70°C for 24 to 72 hours.

Ten micrograms of high-molecular-weight genomic DNA was digested independently with 2 restriction endonucleases (EcoRI andHindIII) under conditions recommended by the supplier (Boehringer Mannheim, Mannheim, Germany). After electrophoresis on 1% agarose gels, DNA was transferred to nylon membranes (Bio Rad, Munich, Germany) according to the manufacturer’s instructions. For the hybridization of Southern blots, denatured ³²P-labeled DNA probes were independently used for the major (MBR) and the minor (mcr) breakpoint cluster regions, PFL1 and PFL2, respectively,25,26 both kindly supplied by Dr M.L. Cleary (Stanford University School of Medicine, Stanford, CA). The labeling of the DNA probes by random priming was performed under conditions recommended by the supplier (Amersham). Autoradiography was performed at −70°C using x-ray film with double high-speed intensifying screens.

We performed LOH studies on the MLH1, MSH2, PMS1, and PMS2 DNA repair genes.3-5 Briefly, LOH was investigated by the same PCR assay described above, using the D3S1611, CA21, D2S117, and D7S517 microsatellite markers to study the MLH1, MSH2, PMS1, and PMS2 genes, respectively. The sequence of these primers has been reported elsewhere.27,28 The intensity of the bands corresponding to the 2 alleles of a given marker was quantified by analysis with Instantimager (Packard, Grove Hills, IL). In a set of experiments, the microsatellite markers D2S119, D2S136, D2S288, D2S378, D3S1561, D3S1298, D7S531, D7S481, and D7S503 were used28 to characterize the LOH involving the MLH1, MSH2, and PMS2 genes.

For single-strand conformation polymorphism (SSCP) analysis, exons 9 and 16 of MLH1 and exons 5 and 13 of MSH2 were amplified by PCR according to previously reported procedures,29 using exon-specific oligonucleotide primers reported by others.30,31 Briefly, 25 PCR cycles were run in 50 μL standard reaction mixture containing 0.2 μg of DNA, 0.4 μmol/L of each primer, and 2 U Taq polymerase (Perkin Elmer) as follows: 1 minute denaturation at 94°C, 30 seconds annealing at 55°C, and 30 seconds extension at 72°C in a DNA thermal-cycler (Perkin Elmer). A total of 10 μL of the amplified products was mixed with 0.1 μCi α-³³P-dATP and 0.1 U Taq polymerase, and 5 additional cycles were run. To detect DNA mutations, PCR product samples were heated at 95°C for 5 minutes and then electrophoresed through a 6% acrylamide gel containing 5% glycerol; the gel was then dried, and exposed to x-ray film at −70°C for 24 to 72 hours.

P53 gene mutations were detected by PCR-SSCP analysis of exons 5 to 8 with specific primers as described above.32 

We looked for microsatellite alterations in 10 microsatellite markers scattered on 7 different chromosomes, as detailed in Table 2 and in Materials and Methods. MSI was defined as a gain of new alleles at a given locus; LOH was a significant attenuation (>50%), or loss of 1 normal allele in the tumor, compared with constitutional DNA; some representative alterations found in leukemias are shown in Fig 1. The bands indicating MSI, in some cases, were not equimolar to the germ line bands, likely due to contaminating normal cells, as also suggested by cytological analysis (data not shown), or heterogeneity within the leukemic poulation. PCR analysis disclosed molecular changes in only 3 of the 29 AML patients analyzed (10%), thus confirming the scarcity of MSI in adult myeloid leukemia.8-10 Interestingly, however, these alterations were detected in 10 of 37 lymphoid tumors (27%), including 6 of 16 ALL patients (37%) and 4 of 21 NHL patients (19%). Thus, in our study population, MSI was more frequently observed in ALL than AML patients (P < .05, Fisher’s exact test). Molecular alterations, listed in Table 3, were found on 1 to 3 loci in 9 of 13 cases; on the other hand, patient 213 (AML), and patients 32, 509, and VR5 (all with ALL) presented widespread genetic instability involving most of the loci analyzed and consisting of both MSI and LOH. Overall, MSI was more frequently detected, accounting for 33 of 44 molecular alterations found, while the remaining 11 were LOH at defined loci (Table 3). MSI was found in tumor DNA obtained at diagnosis in 6 leukemia patients (702, 739, 619, 32, 509, and VR5), and 2 NHL patients (11, 23), thus indirectly suggesting that alterations in mismatch repair genes might already occur in the early steps of the malignancy (Table 4). In the remaining 5 patients, alterations were detected in tumor DNA obtained at relapse, after chemotherapy. Overall, MSI in leukemia was detected in 3 of 9 patients studied at relapse (33%) and in 6 of 36 patients studied at diagnosis (16%) (Tables 4 and 5). However, because DNA samples at diagnosis were not available, it could not be established whether the same alterations were already present at disease onset or had developed during tumor progression.

  Given the relatively frequent occurrence of microsatellite alterations in ALL, it would be important to establish whether they might be useful as molecular markers of disease. To address this point, serial DNAs collected from a B-ALL patient (patient 74) over a 14-month follow-up were analyzed by PCR. Alterations in marker D15S161, consisting of LOH, were first shown in DNA collected during a relapse that occurred in June 1990 (Fig 2). As expected, LOH disappeared when this patient achieved remission (August 1990), and was no longer detected at different time points when the patient was still in remission (Fig 2). However, this same molecular alteration was again detected during a second relapse, which occurred about 1 year later and then disappeared after effective therapy and achievement of remission (Fig 2). Marker D3S1611, which is closely associated with mismatch repair gene MLH1, showed the same behavior (Fig 2).

To determine whether microsatellite alterations might be a marker of impending relapse, 4 consecutive DNA samples from another ALL patient (patient 63) were analyzed for MSI involving markers vWA and D2S123 (Fig 3). MSI could be shown in both markers during impending and full-blown relapse and consisted of faint bands of increased size, compared with the alleles detected at remission (Fig3). Conventional bone marrow cytology of this patient showed hypercellularity with 50% blasts at initial relapse and 90% blasts at full relapse. In view of these findings, we attempted to measure the sensitivity of the PCR assay used to detect MSI. To this end, genomic DNAs from 2 healthy individuals showing a different allele for marker D18S61 were mixed in different proportions, amplified with D18S61-specific primers, and analyzed on acrylamide gel. As shown in Fig 4, the B-specific allele of 1 individual could be detected by this method when his DNA represented at least 3% of the total DNA used as template. According to our estimate, therefore, 3% leukemic cells in the bone marrow sample represent our current limit of detection in disease affected tissues. Even though these observations are limited to 2 cases, they suggest that the assay might potentially be applied to monitor minimal residual disease and disease progression in those leukemia patients who present with microsatellite alterations at diagnosis.

In an attempt to correlate MSI with genetic alterations of known oncogenes, the DNA from all 21 NHL patients was analyzed by Southern blotting, as detailed in Materials and Methods, for rearrangements of the bcl-2 gene, whose involvement in lymphomagenesis is well established.25,26,33 Strikingly, bcl-2 rearrangements were detected in 4 of 4 patients with MSI, but only in 3 of 17 patients without MSI (Fig 5 and Table 6). This difference, which is significant (P < .05, Yate’s corrected χ²test) cannot be accounted for by differences in lymphoma histotype between MSI⁺ and MSI⁻ patients (Table 6) and suggests an association between MSI and bcl-2 rearrangements in NHL. Finally, in view of the possibility that lymphoma cells might carry multiple genetic alterations, we also addressed the presence of mutations involving the p53 gene. As previous reports showed that most p53 mutations in NHL cluster in exons 5 to 8,34,35 these exons were analyzed by reverse transcriptase (RT)-PCR in those NHL patients with MSI; no mutations were found in these patients: genomic DNA from colon cancer patients, with known mutations for the exons considered, was also analyzed and served as a positive control for the SSCP assay. In our patients, therefore, the presence of MSI and bcl-2 rearrangements did not correlate with p53 abnormalities.

Because MSI in solid tumors is linked to the functional inactivation of mismatch repair genes, including MLH1, MSH2, PMS1, PMS2, and, recently, MSH6,3-6 we investigated whether this might also be true for MSI in leukemias and lymphomas. We looked for mutations affecting the MLH1 and MSH2 genes; given the complexity of these genes, we focused on exons 9 and 16 of MLH1, and exons 5 and 13 of MSH2, all of which have been reported to be mutated in some solid and also lymphoid tumors.18,36 We examined samples from all the patients who were positive for MSI and were unable to demonstrate tumor-specific mutations by SSCP analysis. Genomic DNA from a healthy donor, containing a known mutation in the second domain of the CD4 gene,29 was also simultaneously analyzed by SSCP and served as a control for our SSCP experimental conditions (data not shown).

As chromosomal rearrangements in hematologic neoplasms are a major cause of gene disruption,37 we advanced that deletions affecting chromosomes 2, 3, and 7 carrying the MSH2 and PMS1, MLH1, and PMS2 genes, respectively, might occur, and lead to their inactivation in some patients. To test this hypothesis, genomic DNA from patients showing MSI was amplified with primer pairs for D3S1611, CA21, D2S117, and D7S517, which are microsatellite markers closely linked to MLH1, MSH2, PMS1, and PMS2, respectively.27 All mutated NHL and leukemia patients except patients 213 and VR5 were informative for D3S1611 and CA21; all patients, except VR5, were informative for D2S117, and all except patients 619, 52, and 298 for D7S517. LOH of markers closely linked to mismatch repair genes MLH1, MSH2, and PMS2 was demonstrated in 4 of 6 ALLs and 1 of 3 AMLs with microsatellite alterations. In particular, among the ALL patients, patient 32 presented LOH at the D3S1611 site (MLH1), patient 63 at the CA21 site (MSH2), patient 74 at the D3S1611 and D7S517 sites (MLH1 and PMS2), and patient 509 at the D2S117 (PMS1) site; AML patient 213 presented LOH at the D7S517 site (PMS2) (Fig 6). On the other hand, neither 6 RER⁻ ALLs, nor 6 RER⁻ AMLs and 4 RER⁺ NHL presented LOH at the same chromosomal loci (data not shown). To evaluate whether the observed LOH was confined to a particular marker or was indicative of larger choromosomal deletions, LOH at mismatch repair gene loci was analyzed in further detail in 3 patients (63, 74, and 213). To this end, microsatellite markers mapping close to D3S1611, CA21, and D7S517 were chosen for amplification, as reported in Materials and Methods. In the case of patient 74, this analysis disclosed that LOH extended to markers located both centromeric (D3S1561 and D7S531) and telomeric (D3S1298 and D7S503) to D3S1611 and D7S512 (Fig 7); marker D7S481 was not informative in patient 74. On the other hand, in the case of patients 63 and 213, no additional LOH was demonstrated at these loci. These findings suggest that in some leukemia patients MSI is linked to inactivation of DNA repair genes by chromosomal deletions.

The last few years have witnessed major advances in the decoding of the molecular basis of acute leukemias; indeed, in many cases (up to 50%) a specific chromosomal translocation, often involving genes that encode transcription factors, is involved in the pathogenesis of the disease.13,37 In the other cases, however, translocations are random or do not occur at all, thus leaving the origin of the malignancy unexplained. Therefore, other types of genetic damage, for example those affecting tumor suppressor genes and DNA repair genes, might occur and participate in the leukemic transformation. In view of this likelihood, a number of recent studies addressed the occurrence of microsatellite alterations in hematologic malignancies as indirect evidence of disruption of DNA repair genes.7-12,38 The overall conclusion reached by most of these studies was that, unlike findings in some hereditary and sporadic solid tumors, MSI is uncommon in human leukemia and is detected only in a minority of cases, with the possible exception of therapy-related leukemia.39Remarkably, AML was the predominant type of leukemia investigated in these studies; less is known about MSI occurrence in lymphoid tumors.

We compared the prevalence of microsatellite alterations in lymphoid versus myeloid tumors and found a higher prevalence of MSI in ALL than in AML. Although this finding was expected on the basis of previous observations, an explanation for it is presently not forthcoming. Indeed, it is known that knockout mice lacking the murine equivalent of mismatch repair genes MSH2 and MSH6 develop lymphoid malignancies resembling human lymphoblastic lymphoma with high frequency.14-16,18 Furthermore, a screening of human cell lines established from leukemia/lymphoma for MSI disclosed genetic instability exclusively in lymphoid cell lines.36 Finally, it is also known that HNPCC kindreds develop an excess of lymphoid tumors.40,41 Overall, these observations and our findings suggest a connection between inactivation of mismatch repair genes and subsequent MSI on 1 side and lymphoid tumorigenesis on the other.

In contrast to the results of Pabst et al,11 we detected MSI much more frequently than LOH; this discrepancy might probably be explained, at least partially, by differences in the microsatellite markers used, as well as the patient population. Interestingly, 2 of 6 ALLs with MSI belong to the rare immunophenotypic group of mature B-cell ALL, which constitute about 4% of ALLs: this might indicate a nonrandom association between the 2. Clearly, only investigation of MSI prevalence in this rare subset of ALLs and in the closely related B-lymphoblastic lymphoma will definitively resolve this issue. In our patients, MSI seemed to be an adverse prognostic factor. Two of the 6 leukemic patients with MSI detected at diagnosis (patients 702 and 619) relapsed within 15 months, and the relapse was refractory to chemotherapy. In patient 702, the leukemic cells harbored a translocation t(8;21) that normally associates with a favorable outcome.42 Furthermore, the 4 NHL patients with MSI also had an overall poor response to chemotherapy. This contrasts with findings in colorectal cancer, where MSI seems to be a good prognostic indicator2; 1 possible explanation for this discrepancy might be that MSI in leukemia is correlated with resistance to chemotherapy, which usually is the first-line therapy for hematologic neoplasms. Intriguingly, human cell lines with mutator phenotype carrying MLH1 mutations are tolerant to some DNA-alkylating agents, thus suggesting that a functioning mismatch repair system might favor the cytotoxic effects of these drugs.43 A future area of investigation will try to address whether MSI detected at diagnosis might represent a negative prognostic indicator for the outcome of chemotherapy.

Of particular relevance is the finding that in NHL patients MSI is associated with rearrangements of bcl-2, a proto-oncogene whose product is involved in inhibition of apoptosis.44 Chromosomal translocations resulting in high-level bcl-2 expression have been detected in the majority of follicular neoplasms, in about one third of diffuse large B-cell lymphomas and less commonly in other NHL.33 In agreement with this, we did not detect MSI in those lymphoma types, which are not commonly associated with bcl-2 rearrangements (Table 6). In view of the pathogenic relevance of bcl-2, it is tempting to speculate that an increased cell survival conferred by bcl-2 overexpression and genomic instability due to DNA repair gene-defects might cooperate in facilitating the inclusion of new mutations in the genome, possibly increasing the malignancy of the tumor. We also looked for other associated genetic abnormalities and could not detect p53 mutations in the RER⁺ NHL patients. This finding, however, was not completely unexpected, as a negative association between MSI and p53 gene mutation in colorectal cancer was previously reported.45 

Given the relative frequency of MSI in ALL, we believe that microsatellite analysis might be useful in monitoring the disease course in some leukemia patients. Leukemias constitute a unique opportunity to study tumor progression because the neoplastic cells are easily accessible, and the bone marrow is routinely sampled for conventional cytologic evaluation. Although observed in only a few patients, our study shows that the same microsatellite alterations can be consistently detected during follow-up and might be predictive of impending relapse. The advantage of an early detection of leukemia relapse by microsatellite analysis would be to begin chemotherapy at an early stage, when the tumor burden is still low. However, the limited sensitivity of this technology is a critical factor; some investigators suggest that it might be feasible to increase it to detect 1 neoplastic cell among 500 normal cells.46 Compared with some other tumor-specific molecular markers, such as K-ras mutations (detected in 1 of 10,000 cells), this is still low, but might be sufficient in many situations. While specific genetic changes are very sensitive tumor markers, including the BCR-ABL, MLL-AF4, E2A-PBX1, and TEL-AML1 fusion genes in ALL, and AML1-ETO, CBFβ-MYH11, and PML-RARα in AML,13 unfortunately they are not always observed. Microsatellite analysis is easy to perform and also reproducible by nonradioactive methods47; these features should facilitate its application in future studies aimed at assessing the relevance of microsatellite alterations as a tumor marker in lymphoid malignancies.

Although many studies describe MSI in hematologic neoplasms, its relationship to genetic alterations of DNA repair genes has only been marginally explored. To address this issue, we looked for point mutations in a limited number of exons of MLH1 and MSH2, but found none. Admittedly, this was not surprising given the complexity of these genes and the reported absence of clusters of mutations in RER⁺ tumors.30,31 To fully characterize mutations of DNA repair genes in human hematologic malignancies, it would be necessary to analyze all of the exons of MLH1 and MSH2 and perhaps extend these investigations to the PMS1 and PMS2 genes. On the other hand, our analysis of the chromosomal loci of these genes using closely linked microsatellite markers showed LOH in 4 of 6 RER⁺ ALLs, and 1 of 3 RER⁺ AMLs, thus showing that chromosomal deletions might contribute to inactivate mismatch repair genes, at least in some leukemia cases. As defects in mismatch repair genes are apparent only in completely knocked-out cells,27,48 it could be speculated that the remaining allele in leukemias positive for LOH carries disabling mutations. We advance that chromosomal deletions might also contribute to leukemogenesis by targeting and inactivating genes involved in DNA repair.

We are grateful to Dr R. Zamarchi for statistical analysis, P. Gallo for artwork, and P. Segato for precious help in the preparation of the manuscript. We also thank Dr Fabrizio Vinante and the Divisione di Ematologia of the University of Verona, Italy, for bone marrow samples from ALL patients.