Unique Gene Expression Profiles of AML Patients with CEBPA Mutations.

The transcription factor CCAAT/enhancer binding protein alpha (C/EBPalpha) is critical for granulopoiesis. 5–10% of patients with acute myeloid leukemia (AML) carry mutations in the coding region of the CEBPA gene. In a gene expression profiling study of 285 de novo AML patients we previously identified sixteen distinct clusters of AML (Valk et al, N Engl J Med 2004). Eighteen patients (6.3%) carried mutations in CEBPA, and 17 of them were found in two clusters (clusters #4 and #15), indicating that patients with CEBPA mutations exhibit unique gene expression profiles. In cluster #15, all specimens (n=8) carried CEBPA mutations, whereas in cluster #4 CEBPA mutations were found in 9 out of 15 cases. The other 6 cases in this subgroup showed low or no CEBPA mRNA expression, and in 4 of the latter the gene appeared to be switched off by CpG-hypermethylation. We sought to understand why CEBPA mutations were found in two separate clusters, and asked whether we could identify differences between the two clusters. We found no difference when analyzing CEBPA mutation types as most specimens in both clusters carried both an N-terminal truncation and a C-terminal in-frame insertion mutation. Morphologically, specimens in cluster #4 appeared to be less differentiated as compared to patients in cluster #15 (predominant FAB-types being M1 and M2, respectively). With respect to overall survival, patients in cluster #15 tend to have a slightly worse prognosis than patients with mutations in cluster #4 (Kaplan-Meier method, log-rank test, p=0.03). Although two separate clusters were formed, we felt that genes present in expression profiles of both cluster #4 and #15 could be potentially interesting as they could be linked to defective C/EBPalpha functioning. Strikingly, out of the 22 genes differentially expressed in cluster #15, 12 were also differentially expressed in cluster #4, including CTNNA1, TUBB-5, NDFIP1, SFXN3, KIAA0746 and TENS1. Interestingly, all 12 genes were significantly downregulated, suggesting that they could be targets of wild type C/EBPalpha and/or downregulated by mutated C/EBPalpha. To test this hypothesis, we introduced either wild type or mutant CEBPA-ER into 32Dcl1, a cell line model constitutively expressing the human G-CSFR. In line with previous reports, we found that activation of C/EBPalpha by addition of beta-estradiol resulted in proliferation arrest and differentiation of these cells within two days, even in the presence of IL-3. Expression levels of the C/EBPalpha target gene CSF3R increased drastically (12-fold after 24 hours, 53-fold after 48 hours) upon stimulation with beta-estradiol as compared to unstimulated or empty vector control clones. Experiments with clones expressing a C-terminal mutant carrying an 18-nt insertion in the bZIP region showed that proliferation was only modestly inhibited and that differentiation was severely impaired both in the presence of IL-3 or G-CSF. Interestingly, no upregulation of the CSF3R gene was observed following beta-estradiol stimulation of mutant CEBPA-ER in the presence of IL-3. Moreover, activation of mutant C/EBPalpha counteracted the induction of CSF3R expression observed following G-CSF activation. These findings suggest that C-terminal C/EBPalpha mutants can have a dominant negative role in AML. Our studies demonstrate that 32Dcl1-CEBPA-ER cells provide a useful model to further elucidate the possible relationships of C/EBPalpha and C/EBPalpha mutants with differentially expressed genes identified in the gene expression studies.