In the article “BMI-1 gene amplification and overexpression in hematological malignancies occur mainly in mantle cell lymphomas,” Beá et al. (1) report on expression of the BMI-1 PcG1 gene in various human hematological malignancies and carcinomas. BMI-1 expression was assessed by quantitative RT-PCR and by Western blot analysis. The authors found that some B-cell lymphomas representing mantle cell lymphomas expressed higher levels of BMI-1 mRNA and BMI-1 protein, which correlated with BMI-1 gene amplification in some cases. Other hematological malignancies and carcinomas also expressed BMI-1 but not at significantly enhanced levels. In general, expression of BMI-1 was low in large B-cell lymphomas and follicular lymphoma and enhanced in mantle cell lymphomas and chronic lymphocytic leukemia. The authors concluded that human B-cell lymphomas maintain in part the BMI-1 expression pattern of their normal counterparts, although BMI-1 gene alterations may contribute to lymphomagenesis in a subset of mantle cell lymphomas.
Over the past 2 years, we performed detailed studies of BMI-1 expression in human Reed-Sternberg cells of HD and various human B-NHL, in relation to their normal cellular counterparts in the germinal center. Contrary to Beá et al., we interpret BMI-1 expression patterns in HD and B-NHL as abnormal and suggestive of BMI-1 overexpression or failure to down-regulate the gene. This conclusion is based on expression analysis of human PcG complexes in normal lymphocytes and neoplastic cells.
PcG proteins function by forming PcG complexes that bind to chromatin. Two distinct complexes have been identified in humans; one complex contains the BMI-1 protein in combination with various other PcG proteins, and the second complex contains the EZH2 PcG protein. EZH2 does not interact with PcG proteins in the first complex and vice versa. Using immunohistochemical staining of BMI-1 and EZH2 proteins as representatives of the two PcG complexes, we found that normal germinal center B cells express the BMI-1 and EZH2 genes in a mutually exclusive pattern (2, 3). Resting B cells in the mantle zone and resting centrocytes in the follicle expressed BMI-1 in the absence of EZH2. By contrast, rapidly dividing centroblasts expressed EZH2 in the absence of BMI-1. Similarly, resting interfollicular T cells are BMI-1+/EZH2−, whereas dividing T cells in the lymph nodes are BMI-1−/EZH2+ (4).
The BMI-1 expression pattern in neoplastic cells of HD (3) and B-NHL (5, 6) was entirely different from that of healthy B cells, because BMI-1 was expressed in the presence of EZH2. We did not interpret EZH2 expression in neoplastic cells as abnormal, because healthy dividing B cells express EZH2 as well. However, BMI-1 expression in dividing neoplastic cells stands in sharp contrast to the BMI-1 expression profile of normal B cells, because these do not express BMI-1 when they are in cycle. From these observations we concluded that BMI-1/EZH2 coexpression in human B-cell lymphomas reflects abnormal expression of BMI-1.
The discrepancy between our conclusion and the interpretation of Beá et al. is probably caused by the fact that neither RT-PCR nor Western blot analysis on whole tissue differentiates between BMI-1 expression in neoplastic cells and BMI-1 expression by the infiltrate of healthy lymphocytes. Normal B and T cells surrounding tumor cells express BMI-1 (2, 3, 4, 5), and this probably explains why RT-PCR and western blot failed to detect major differences in BMI-1 expression in human tumors.
Our conclusion that altered expression patterns of BMI-1 in neoplastic cells of HD and B-NHL is related to lymphomagenesis in humans and is in line with induction of lymphomas in Bmi-1 transgenic mice (7). Collectively, these studies suggest that BMI-1 should be seriously considered as a human oncogene. As shown in the study by Beá et al., the underlying mechanism of BMI-1 expression in human lymphomas is unlikely to be related to BMI-1 gene amplification or translocation, because these rarely occur in human lymphoid malignancies. This points to a defect at the level of BMI-1 transcription, possibly related to a failure to down-regulate BMI-1 expression during up-regulation of EZH2 in dividing neoplastic cells.
The abbreviations used are: PcG, Polycomb group; RT-PCR, reverse transcription-PCR; HD, Hodgkin’s disease; B-NHL, non-Hodgkin’s lymphoma of B-cell origin; BMI-1, B-cell Moloney murine leukemia virus integration site 1.
Correspondence re: S. Beá et al., BMI-1 Gene Amplification and Overexpression in Hematological Malignancies Occur Mainly in Mantle Cell Lymphomas. Cancer Res., 61: 2409–2412, 2001
In their letter, Raaphorst et al. summarize their observations on BMI-1 protein expression in NHL1 (1) and compare these results with our previous studies published in “Cancer Research” (2). Unfortunately, we could not discuss these findings in our article, because their study was not yet published at the time of our publication. These two studies were performed using two different technical approaches. Raaphorst et al. analyzed the protein expression of BMI-1 and EZH2, another member of the Polycomb-group genes, using immunohistochemical staining of formalin fixed-paraffin embedded tissues, whereas we examined BMI-1 gene alterations by Southern blot and mRNA and protein expression using real-time quantitative PCR and Western blot, respectively. Contrary to the opinion of Raaphorst et al. we think these two studies provide complementary information on the alterations of BMI-1 in these tumors and its potential role in the pathogenesis of NHL. The apparent contradictory interpretations indicated by Raaphorst et al. are probably attributable to misunderstanding of some of our results.
Initial studies using in vitro and animal models had identified BMI-1 as a gene with oncogenic potential, particularly in the development of murine and feline lymphomas (3,4,5). Our study showed for the first time that BMI-1 gene is amplified and overexpressed in a subset of NHLs, particularly of mantle cell type, suggesting that this gene may be an oncogenic target of the chromosome 10p13 alterations observed in different human tumors (6,7,8,9,10). Therefore, our findings do support the oncogenic potential of BMI-1 in human neoplasms.
Previous studies by Raaphorst et al. (11) have indicated that BMI-1 is highly expressed in normal resting B lymphocytes localized in the mantle zone of the lymphoid follicles, whereas it was down-regulated in proliferating centroblasts of the germinal centers. In our study, we showed that CLLs and MCLs had significantly higher mRNA and protein levels than FL and diffuse LCLs. MCL and CLL are tumors mainly derived from naive pregerminal center cells localized in the mantel zone of the follicle, and FL and at least a subgroup of LCL are considered tumors derived from germinal center cells. The parallel pattern of BMI expression between these lymphomas and their respective normal cell counterpart in the lymphoid follicle made us suggest that the different NHLs maintain in part the BMI-1 expression profile of their normal cell counterpart. In their letter, Raaphorst et al. suggest that our real-time quantitative reverse transcription-PCR and Western blot techniques failed to detect major differences in BMI-1 expression in human lymphomas, probably attributable to the fact that these techniques using whole tissue did not distinguish between BMI-1 expression in tumor and reactive cells. This comment indicates a real misunderstanding of our results because our study did show major differences in the expression between different types of tumors. Particularly, the mRNA levels in CLL and MCL were significantly higher than in FL and LCL (P < 0.01). In addition, it would be difficult to interpret these differences based only in the presence of reactive cells, because the number of reactive T cells in FL is usually higher than in CLL and MCL.
The immunohistochemical studies performed by Raaphorst et al. showed that BMI-1 protein expression was detected in dividing centroblasts of LCL. This pattern was considered as abnormal, because BMI-1 is not expressed in proliferating centroblasts of normal germinal center, suggesting that the failure to down-regulate BMI-1 in proliferating neoplastic cells may play a role in the pathogenesis of the tumors. This observation may not be totally in contradiction with our findings. Although our quantitative study did show BMI-1 mRNA down-regulation in LCL (mean 0.6 relative units, SD 0.4) comparing to CLL (mean 2.2 relative units, SD 1.3) and MCL (mean 2.5 relative units, SD 2.3), BMI mRNA levels showed variation between tumors with a certain overlap between LCL and MCL or CLL. It may be possible that the BMI-1 expression observed by Raaphorst et al. may correspond to some of the cases with a relatively higher level detected in our study. Immunohistochemistry is a very useful technique to correlate expression and morphology, but it is not a reliable technique to quantify protein levels. Some studies have quantified the immunohistochemical staining using scores, including the intensity of the reaction and/or the number of positive cells. Unfortunately, Raaphorst et al. did not use any system to quantify their results, and therefore, they are difficult to fully compare with our findings. On the other hand, they only examined 6 LCLs, 6 MCLs and 5 CLLs, whereas our study included the quantitative measurements of 22 LCLs, 27 MCLs, and 10 CLLs.
In summary, these studies clearly indicate that BMI-1 expression is deregulated in NHL and may play a role in the pathogenesis of these tumors. The observations in both studies may allow to hypothesize that the mechanisms leading to this deregulation vary in different lymphomas. Thus, in tumors such as MCL, derived from cells with a relative high steady-state level of BMI-1 expression, oncogenic activation may occur by gene amplification or transcriptional deregulation leading to significantly higher levels of gene expression. In addition, the failure to silence completely BMI-1 expression in tumors derived from dividing germinal center cells, which normally do not express BMI-1, may also contribute to the pathogenesis of these tumors.
The abbreviations used are: NHL, non-Hodgkin’s lymphoma; CLL, chronic lymphocytic leukemia; MCL, mantle cell lymphoma; FL, follicular; LCL, large B-cell lymphoma.