Microarray-based analysis and clinical validation identify ubiquitin-conjugating enzyme E2E1 (UBE2E1) as a prognostic factor in acute myeloid leukemia

In this study, we performed a genome-wide screening to identify gene expression that correlate with adult AML OS. The gene expression profiles (GEPs) from 2 independent datasets of patient samples were used in our analysis. The correlation of each gene expression and AML OS was calculated by a program coded by R software. Only gene identified with statistical significance in both datasets was considered as positive results for further test. By this strategy of analysis, we identified 8 AML prognostic genes. Next, we tested our findings using an independent cohort of 50 AML samples. Our result suggested that although several genes, such as HBD and ACOT11, had trend correlation, only one gene, UBE2E1, was statistically correlated with AML OS in our validation cohort. The negative findings of other 7 genes in our validation cohort might be caused by the relatively small number of patients. In addition, we noticed that the patients in microarray-testing cohort and our validation cohort had different ethnic backgrounds. Further studies might be necessary to draw a more confirmative conclusion.

Mounting evidence has shown that AML is highly heterogeneous and dynamic [8]. The heterogeneous entity of AML emerges from the disease genetic basis, leukemogenesis, pathophysiology, and prognosis. However, cluster of gene expression signature [5], or even single gene expression [4], has been shown to correlate with AML prognosis. Therefore, what is the interpretation of prognostic single gene expression, such as UBE2E1 and LEF1, in AML? We hypothesized that different subgroups of AML, with discrete driver mutations, might have similar epigenetic effectors’ upregulation/downregulation, which correlate with patient’s survival. We also hypothesized that in different prognosis-relevant AML subgroups, the effectors have patterned expression. To test these hypotheses, we performed another microarray-based analysis for UBE2E1 expression in AML with complex karyotype vs. normal karyotype, FLT3 mutation vs. wildtypeFLT3, and NPM1 mutation vs. wildtypeNPM1. We selected those genetic abnormalities because they have high frequency of occurrence in AML and correlate with the patient clinical outcome: patients with complex karyotype or FLT3 mutation had poor treatment outcome, while patients with NPM1 mutation had good treatment outcome [8]. As shown in Additional file 1: Table S1, complex karyotype or FLT3-mutated AML had relatively high UBE2E1 expression, compared with normal karyotype or wildtype FLT3 AML, respectively. NPM1-mutated AML had relatively low UBE2E1 expression. These preliminary findings might indicate that UBE2E1 have patterned expressions, which was well matched with AML classification despite of different genetic basis.

Protein ubiquitination was accomplished by sequential action of enzymes E1, E2, and E3. Specifically, E2 transferred E1-activated ubiquitin to E3, an ubiquitin ligase, and formed an isopeptide bond between ubiquitin and protein substrate. UBE2E1 was a member of ubiquitin-conjugating enzyme E2 class. Zhu et al. showed that UBE2E1 regulated HOX gene expression by ubiquitinating histones [9]. Thus, UBE2E1 might play a regulatory role in cell by selectively ubiquitinating target proteins. The function of UBE2E1 in cell signaling is still largely unknown. However, the regulation of UBE2E1 on HOX gene might be a key to understand the prognostic role of UBE2E1 in AML. HOX gene is a family of highly conserved homeodomain transcription factor genes [10]. There are 39 HOX genes, belonging to 4 gene clusters, in human. Previous work has shown that HOX genes are aberrantly expressed in AML [11, 12]. Animal study indicated that overexpression HOX gene, HOXA10 and HOXA9, promoted AML leukemogenesis [13, 14]. To identify potential UBE2E1 downstream HOX genes, we started with microarray datasets. We found co-expression of UBE2E1 with HOXA11 in AML. We also examined HOXA11 and UBE2E1 co-expression in our validation cohort (Additional file 1: Figure S3). Interestingly, a recent publication suggested that HOXA11 expression correlated with glioblastoma patient treatment responses and prognosis [15]. Thus, it is highly possible that UBE2E1 regulates AML chemoresistance through HOXA11.

In our study, we found co-expression of UBE2E1 with HOX family gene, HOXA11, in AML. Therefore, we hypothesized that UBE2E1 regulates HOXA11 gene expression in AML, and HOXA11 transcription factor level might be relevant to AML treatment resistance. We are actively conducting more mechanistic studies to demonstrate the role of UBE2E1 in AML.