Expression of BEX1 in acute myeloid leukemia with MLL rearrangements

TO THE EDITOR

For many years, cytomorphology and cytochemistry provided the basis for the classification of leukemias. More recently, immunophenotyping, cytogenetics and molecular genetics have contributed to tumor subclassification. Gene expression profiling promises to further improve tumor diagnosis. Determination of the expression of a limited number of genes may assist safe discrimination between tumor types. In recent years, it has been shown that not only acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML), but also distinct subtypes of both diseases may be identified by the array technology.^{1, 2, 3}

Continuously growing cell lines are useful tools for basic and applied aspects of cell biology — once their applicability as model systems for the respective primary tumor cells is demonstrated. Recently, we have shown the suitability, as in vitro models, of ALL-derived cell lines carrying genetic alterations of the ‘mixed lineage leukemia’ (MLL) gene: like the respective primary tumor cells, these cell lines can be classified on the basis of MEIS1, HOXA9 and FLT3 gene expression.⁴ However, the genes that allowed discrimination of MLL mutant (MLLmu) and MLL wild-type (MLLwt) ALL cells did not permit AML cells with or without MLL mutations to be distinguished.⁴ Here, we set out to find genes that would allow one to recognize MLLmu and MLLwt AML cells.

In a first step, we performed gene expression array analysis with pooled RNAs of eight MLLmu and eight MLLwt AML cell lines, respectively, using Affymetrix HG-U133A chips (Santa Clara, CA, USA). We observed that ‘brain expressed X-linked 1’ (BEX1) was overexpressed 17-fold in the MLLmu panel (Figure 1). It has recently been reported that human BEX1 was expressed in various tissues like the brain, pancreas, testis, kidney, liver, spleen and adrenal gland, but not in peripheral blood leukocytes, lymph nodes and bone marrow.⁵ In that study, a probe covering the complete BEX1 coding region had been used.

Figure 1

BEX1 expression in MLLmu and MLLwt AML cell lines. (a) Expression array and RT-PCR analysis: expression array analysis was performed with pooled RNAs of eight MLLmu and eight MLLwt AML cell lines. Results are shown on the right side of Figure 1 (mu/wt array). Expression of the seven most differently expressed genes were retested by RT-PCR analysis for each cell line individually (cDNAs from the pooled RNAs of expression array analysis are named ‘mix’). Six of seven genes were myeloid differentiation and activation markers. RT-PCR analysis showed that they were not expressed uniquely and consistently in one type of cell lines only. BEX1 was the gene shortlisted for the recognition of MLLmu AML, being overexpressed in the MLLmu group only (7/8 cell lines). B) Northern blot analysis: BEX1 is strongly expressed in MLLmu AML cell lines only (7/8). Note that the MLLwt cell line U-937 expresses BEX1 only weakly.

As BEX1 has a very high-sequence homology to BEX2, we set out to verify these results using a human multiple tissue expression array (BD Biosciences Clontech, Heidelberg, Germany) with BEX1- and BEX2-specific oligonucleotide probes: both genes were strongly expressed in brain-derived tissues. The exclusive expression of BEX1 was seen in the pancreas and testis, and the exclusive expression of BEX2 in the kidney, liver and adrenal gland. Most importantly, however, BEX1 and BEX2 were not expressed in normal cells of the hematopoietic system. Furthermore, reverse transcriptase polymerase chain reaction (RT-PCR) analysis showed that BEX1 was also not expressed in hematopoietic cell lines other than AML (0/32), including cell lines derived from Hodgkin lymphoma, anaplastic large cell lymphoma and ALL. BEX1 was expressed in 82% of the MLLmu AML cell lines (9/11) and in 18% (2/11) of the MLLwt AML cell lines (partly shown in Figure 1a). The BEX1-positive MLLwt cell lines showed only very weak BEX1 signals, supporting the idea of a positive correlation between MLL gene aberration and high expression levels of BEX1 in AML (Figure 1a and b). Even MLLmu ALL cell lines tested BEX1 negative (0/5), suggesting that BEX1 might be a true marker for the identification of MLLmu AML cells. With respect to the absence of BEX1 expression in MLLmu ALL cell lines, one might speculate that BEX1 expression depends on the type of MLL rearrangement or on the histological background of the cells.

To test whether the positive correlation of MLL aberration and BEX1 expression, as observed in AML cell lines, also exists in primary cells, we performed quantitative RT-PCR analysis for leukemic blast cells from four patients with AML (2 MLLmu, 2 MLLwt samples). BEX1 expression levels in the two MLLmu samples were similar to those in the MLLmu cell line THP-1, while BEX1 levels in the MLLwt samples were hardly detectable (Figure 2).

Figure 2

BEX1 expression in MLLmu and MLLwt primary AML cells. Quantitative RT-PCR expression of BEX1 transcript was performed for leukemic blast samples from four patients with AML: two MLLmu samples (both FAB M5), two MLLwt samples (one PML-RARA positive FAB M3, one AML1-ETO positive FAB M2). The BEX1 expressing MLLmu cell line THP-1 was used as control in the ‘delta-delta CT’ (δ-δCt) quantification method (threshold cycle: Ct). Quantitative RT-PCR was performed on the ABI Prism 5700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA).

Our results suggest that BEX1 is a candidate gene for the diagnosis of MLLmu AML and raise the question of whether BEX1 plays a role in leukemogenesis. BEX family members interact with neuronal proteins like the p75 neurotrophin receptor and the olfactory marker protein.^{6, 7} Both proteins are not expressed in AML-derived cell lines (unpublished own data). Thus, it will be of interest to identify a BEX1 partner in myeloid cells, and to elucidate the possible role of BEX1 for cell signalling processes in MLLmu AML.