Analysis of CALM/AF10 expression by RT-PCR and expression profiling of CALM/AF10 positive patients using microarray technology

We were able to analyse samples from 17 CALM/AF10–positive leukemic patients. The patients were children or young adults, between 5 and 46 (average = 22) years old, a similar age distribution was reported in other studies for CALM/AF10 patients (Narita et al., 1999; Salmon-Nguyen et al., 2000; Bohlander et al., 2000). All leukemias were acute leukemias, undifferentiated or from the lymphoid or myeloid lineage. We did not see any obvious correlation between age, diagnosis and the location of the breakpoints in CALM or AF10 in our patients.

Among a series of 29 patients with T-ALL with T-cell receptor gd rearrangement, four (14%) were found positive for CALM/AF10 transcripts, confirming a previous report of the high incidence of CALM/AF10 among this group of leukemias (Asnafi et al., 2003).

The reciprocal AF10/CALM transcript was identified in 7 of the 17 patients. The observation that the AF10/CALM transcripts can only be found in some patients has also been reported previously (Bohlander et al., 2000; Carlson et al., 2000), suporting the notion that CALM/AF10 is the relevant fusion protein in these leukemias.

Gene expression profiling is becoming an important tool in the analysis of leukemias (Andersson et al., 2005; Haferlach et al., 2005). ALL with chromosomal rearrangements involving the MLL gene can be distinguished from ALL without MLL involvement. Three cytogenetically defined subtypes of AML: t(15;17), t(8;21) and inv(16) can also be recognized based on the gene expression profile (Schoch et al., 2002).

Among our 17 CALM/AF10 patients, we were able to analyze the gene expression profile for 10 of them. Hierarchical clustering analysis was performed using in addition the gene expression profile of 20 AML patients with MLL rearrangements and 20 patients with t(8;21) positive AML. The hierarchical clustering analysis was restricted to the HOX A cluster genes and MEIS1. Furthermore, the gene expression profile of the CALM/AF10 positive patients was compared to the profile of groups of 15 patients from each of 9 other leukemia subtypes: T-ALL, ALL with MLL rearrangements, AML with MLL rearrangements, AML with inversion of the chromosome 16, AML showing normal karyotype, AML t(8;21), AML with acute promyelocytic leukemia, AML with complex aberrant karyotype, as well as from 15 normal bone marrow control samples.

The hierarchical clustering revealed high expression levels of the homeobox gene MEIS1, and the HOXA cluster genes HOXA1, HOXA4, HOXA5, HOXA7, HOXA9, HOXA10. The CALM/AF10 patients had a very similar expression pattern to patients with MLL rearrangements as far as these genes were concerned. Curiously, from the ten CALM/AF10 patients analyzed, only two had a diagnosis of AML and one of an acute biphenotypic leukemia. The other seven patients had been diagnosed as ALL (1), T-ALL (2), Pro T-ALL (2) and Pre T-ALL (2). These results led to the conclusion of a common expression pattern between CALM/AF10 patients independently of the type of leukemia (lymphoid or myeloid), considering the expression of the above-cited genes.

When looking for differentially expressed genes that were specific for CALM/AF10 leukemias, only three genes (DNAJC1, COMMD3 and BMI1) were found to be consistenly higher expressed in CALM/AF10 patients compared to all other leukemia groups analyzed. Since these three genes are located downstream (centromeric) to the 10p21 breakpoint, a position effect caused by the juxtapposition of the strong CALM promoter or enhancer in the 10;11 translocation can be assumed. This phenomenon has also been observed by Dik and colleagues, who showed overexpression of HOXA genes and BMI1 and the two other genes (DNAJC1 and COMMD3) which lie in the vicinity of the breakpoint on 10p12 (Dik et al., 2005). They analyzed six CALM/AF10 T-ALL patients and compared their expression profiles to 17 T-ALL CALM/AF10 negative patients. However, they did not see higher expression levels of HOXA7.

The involvement of HOX genes in acute leukemias has been widely reported (Blatt et al., 1988; Kroon et al., 1998; Thorsteinsdottir et al., 2001; Ayton and Cleary, 2003; Soulier et al., 2005). Ayton and Cleary (2003) showed the dependence of MLL fusion proteins on Hoxa9 and Hoxa7 to transform myeloid progenitors in a mouse transplantation model. The mechanism by which MLL fusion genes can induce leukemia is probably related to chromatin structure. MLL has intrinsic histone H3, lysine 4 methyltransferase activity encoded by its SET (Su(var)3-9, Enhancer-of-zeste, Thithorax) domain and in addition recruits the SWI/SNF chromatin remodelling complex. These domains are lost in MLL fusions so that the chromatin-associated functions might come from the fusion partner (Zeisig et al., 2005). Yi Zhang’s group (Okada et al. 2005) found an interaction between AF10 and MLL/AF10 and the human histone methyltransferase hDOT1L (disruptor of telomeric silencing 1). hDOT1L was shown to be able to transform myeloid progenitors when fused to MLL. It was also shown that the octapeptide-leucine zipper motif of AF10 is necessary and sufficient for the interaction with hDOT1L. MLL/AF10 associated myeloid transformation was accompanied by up-regulation of Hoxa9 and Hoxa7. In addition, AF10 also interacts with GAS41, a protein which interacts with the SWI/SNF complex. Soulier et al. (2005) observed upregulation of all HOXA genes in T-ALL patients, suggesting that HOXA up-regulation as a common feature of these leukemias. The TALE (three amino acid loop extension) homeobox cofactor Meis1 was shown to collaborate with HOXB3 in leukemogenesis and accelerate the onset of HOXA-induced

leukemias as well as to reduce the latency of leukemias in mice transplanted with cells retrovirally-expressing the fusion gene NUP98-HOXD13 (Thorsteinsdottir et al., 2001; Pineault et al., 2003).

Considering the similar overexpression pattern of HOX genes observed in MLL- and in CALM/AF10-induced leukemias, the leukemogenic mechanisms in both entities could have a common pathway.

When the gene expression profiles of the CALM/AF10 patients was compared to the expression profiles of other leukemia patients or normal bone marrow it became apparent that more than 90% of the top 100 differentially regulated genes were down-regulated in the CALM/AF10 patients. Some of the more interesting down-regulated genes (Table 21) are discussed in the following paragraphs.

RAB34: was the gene with the strongest repression in the CALM/AF10 samples. The Rab family is associated with vesicle membrane compartment functions and is important in vesicle trafficking (Sun et al., 2003).

TRIO (Triple fuctional domain) protein gene, which is involved in phagocytosis of apoptotic cells was found expressed at a high frequence in oral carcinomas (de Bakker et al., 2004; Baldwin et al., 2005).

The CEBPD (CAAT/enhancer binding protein d) gene belongs to the C/EBP family of transcription factors and is important in proliferation, differentiation and apoptosis. CEBPD is a putative tumor suppressor gene whose deficiency in mouse fibroblasts cells caused impaired growth control and was associated with chromosomal instability (Huang et al., 2004). Mutations in C/EBPD gene are common in human breast cancer (Tang and DeWille, 2002). ST7 is a tumor suppressor gene which is downregulated by promoter methylation mediated by the PRMT5 (protein arginine methyltransferase), a protein associated with the human SWI/SNF complex (Pal et al., 2004).

The expression of H2AFJ (histone family member J) was correlated with human melanocytic tumour (de Wit et al., 2005).

JAG1 (jagged1) is a NOTCH ligand, whose expression is associated with bad prognosis in human breast cancer (Reedijk et al, 2005).

HCK (hematopoietic cell kinase) is a src kinase family member, expressed in hematopoietic cells including myeloid cells and to a minor extent in B-cells (Quintrel et al., 1987). The inactivation of the HCK gene results in defects in phagocytosis in macrophages (Hauses et al., 1998).

TEM7R (tumor endothelial marker 7) is abundantly expressed in tumor endothelial cells (Nanda et al., 2004).

CDH15 (Cadherin 15): Cadherins are a family of cell surface molecules involved in cellular structural and functional organization. The loss of heterozigosity of cadherin gene found in solid tumors makes these family of genes candidate tumor supressor genes (Kremmidiotis et al., 1998; Cool and Jolicoer, 1999).

CTSZ (Cystein Protease Cathepsin Z) is expressed in pancreatic cell lines (Mahlamäki et al., 2002).

CTNNA1(aE-catenin) can suppress tumor invasion in human cancer (Vermeulen et al., 1999). TKR2 (Human toll-like receptor 2) is a membrane receptor playing a role in innate immunity which activates NF-kB (Yang et al., 1999).

HOXB4 is a member of the HOX B gene cluster. HOXB4 expression has been shown to increase the self-renewal of HSCs without disrupting differentiation (Abramovich et al., 2005).

USP3 (ubiquitin specific protease 3): functions in deubiquitination processes (Sloper-Mould et al., 1999).

ZNF185 (zink finger protein 185) was identified by expression profiling in prostate cancer. The methylation of the ZNF185 promoter transcriptional silencing of this gene is implicated in prostate tumorigenesis (Vanaja et al., 2003).

The transcriptional repression of these genes of which several have important functions in membrane trafficking, cellular growth, self-renewal, differentiation, angiogenesis and tumor suppression, could very well contribute to the transforming and leukemogenic potential of CALM/AF10. However, more detailed functional studies as well as the proof that these genes are direct targets of CALM/AF10-mediated gene regulation (see also next paragraph) are required to corroborate these assumptions.

6.3. Inducible expression of CALM/AF10 in cell lines