Loss of heterozygosity (LOH) studies are useful to detect tumor suppressor genes involved in human cancers. In childhood acute lymphoblastic leukemia (ALL), LOH has been reported for many chromosomal regions, including 11q23, close to the MLLgene. Previously in this journal, Webb and colleagues1 reported LOH of a polymorphic trinucleotide repeat (mllGAAn) in intron 6 of the MLL gene. The heterozygosity index of the repeat was reported to be 0.54, and LOH was found in 6 (40%) of 15 of the informative ALL cases. TheMLL gene is flanked by 2 polymorphic microsatellite markers (D11S1356 and D11S1364) that have been used to determine LOH. Surprisingly, Webb and colleagues found no LOH at these 2 markers. Other studies have reported LOH of D11S1356 in 4%-16% of ALL cases.2-4 Löchner and colleagues5showed LOH in exon 8 of MLL in 3 (4%) of 74 T-lineage ALL cases.
We attempted to assess the prevalence of LOH within the MLLgene, using 40 cases of ALL.
Children with ALL were recruited at pediatric oncology clinics throughout New Zealand. DNA was extracted from unfixed diagnostic and remission bone marrow slides as published.6 Parental DNA was extracted from peripheral blood leukocytes. LOH at mllGAAn was assessed using polymerase chain reaction (PCR) amplification (forward primer: TCCCCGCCCAAGTATCCCTGTAAAA, reverse primer: GCTGCGCCTTGCCAAGCCTAAAT, 30 cycles, annealing temperature 58°C). The α-P32dCTP-labeled PCR products were electrophoresed on 6% denaturing polyacrylamide sequencing gel.
Initially, the mllGAAn polymorphism was reassessed among unrelated individuals. Surprisingly, only 2 of 36 unrelated individuals were heterozygous. Then, 40 ALL cases were tested for LOH and microsatellite instability (MSI) at the mllGAAn locus. Only 1 case was informative for the study of LOH, and in this case the leukemic DNA showed retention of heterozygosity. In 18 cases the leukemic DNA was of sufficient quality to assess MSI. MSI was not detected in any of these cases.
Our results contrast with those reported by Webb and colleagues.1 They found a high heterozygosity index (0.54), an LOH rate of 40% (6 of 15 informative cases), and an MSI rate of 10% (3 of 29 cases). In other reports, MSI on 11q was reported in 10 (8.8%) of 114 cases at D11S13563 and in none of 40 cases at 8 different markers.7
We noted that PCR primers were difficult to design, because the mllGAAn polymorphism was located within an Alu repeat element. Our primers were designed to avoid this repeat, but we note that the primers of Webb and colleagues are situated within the repeat element. Furthermore, their forward primer binds in numerous sites throughout the genome. As a consequence we doubt that the results of Webb and colleagues reflect LOH or MSI of the MLL gene.
Examination of 13 independent sequences from National Center for Biotechnology Information (NCBI) GenBank confirms that length variation does occur at the mllGAAn locus. Additional research with a larger number of ALL cases will be required to determine the true frequency of intragenic LOH of the MLLgene.
Loss of heterozygosity at the MLL locus
The letter from van Schooten et al is provocative as it seems to contradict data that we have previously published in this journal.1-1 The issue centers around the level of heterozygosity at the mllGAAn microsatellite, and the authors suggest that poor experimental design on our behalf means that our results are doubtful. We are confident that our findings are accurate and wish to respond to these comments.
The criticism is that one of the primers is located within Alu sequences. We were aware of this at the time and tried a number of different pairs of primers but found the one used the most efficient. The polymerase chain reaction (PCR) was optimized carefully to avoid mispriming, and only the non-Alu primer was end-labeled. To confirm that the alleles observed were genuine, we analyzed the polymorphism in 2 families and found no discrepancies. The result from one of these families is shown in lanes 1, 2, and 3 of Figure 2 from the original paper.1(p286) To further confirm that these amplification products were polymorphic alleles, we directly sequenced from the PCR products of the 2 most common alleles (J.C.W., unpublished work, August 1998). We did not sequence the rarer allele, as there were no homozygous samples. We also confirmed that the polymorphic alleles were in Hardy-Weinberg equilibrium, and it would be unlikely that we would see this ratio if we were amplifying random bits of the genome. Finally, repeat remission samples for some of the patients existed and, where possible, these also were amplified (unpublished work). Again, we found no discrepancies. For these reasons we believe that our data are sound.
van Schooten et al do not suggest any alternative explanations for the discrepancies between the 2 sets of data. Racial differences in the normal population groups or sample size may account for some of the inconsistency. An alternative explanation is that van Schooten et al have not resolved the PCR products sufficiently. This would result in the 2 most common alleles appearing to be the same size and only the larger allele distinguishable as a polymorphism. Consequently, the heterozygosity frequency would appear much lower than it actually is, since the larger allele is much rarer. In our hands the frequency of the rarer allele was 0.07, and this is consistent with the frequency of 0.06 observed in van Schooten et al's control samples. Poor resolution of the PCR products may also explain why no microsatellite instability was observed in the acute lymphoblastic leukemia samples.
The authors also suggest that the loss of heterozygosity (LOH) we saw was purely artifactual. It is worth pointing out that we were able to confirm LOH in 2 of these samples by fluorescence in situ hybridization analysis. Since then, there have been a number of papers describing LOH at the MLL gene locus using various different methods.1,2 van Schooten et al's suggestion that there is no LOH at this polymorphism is not backed up by their data: they see very little heterozygosity, and it is not possible to detect loss of a homozygous allele when using only a nonquantitative PCR-based technique.
References
Supported by a grant from the Cancer Society of New Zealand.
