In acute lymphoblastic leukemia (ALL) cure rates have reached >90%, however, therapy resistance and relapses occur. Hence, the major interest in leukemia research is the understanding of ALL biology and, based on that, the development of new targeted therapies. As patient-derived xenograft (PDX) mouse models, such as the NOD/SCID/human ALL xenotransplantation model, are systems resembling the human disease with leukemia engraftment and manifestation in bone marrow (BM), central nervous system (CNS), and spleen, they provide reliable models to study ALL biology and preclinically evaluate new treatment strategies.
MicroRNAs (miRNAs) are described to post-transcriptionally regulate the expression of genes coding for critical regulators of leukemogenesis either functioning as tumor-suppressors or oncomirs. Moreover, ALL miRNA profiles can serve as biomarkers indicating disease stages and distribution and can be used for patient risk stratification. To reliably compare miRNA data across different samples and, most importantly, across various cohorts, data normalization with a constantly expressed miRNA reference in the respective samples is required.
We aimed at identifying a reliable miRNA reference which shows a stable and constant expression in different ALL samples in a series of cell lines and PDX ALL specimens derived from different organ compartments.
We analyzed the expression of 6 frequently used miRNA references (RNU6, RNU1A1, SNORD44, 5sRNA, miR-103a-3p and miR-532-5p) by real-time quantitative PCR. All together 9 ALL cell lines (NALM-6, REH, RS4;11, KOPN-8, UoCB6, RCH-ACV, MHH-CALL4, EU3, and HAL-01), 22 PDX specimens including CNS-, BM-, and spleen-derived ALL cells, and peripheral blood mononuclear cells of 6 healthy controls (HC-PBMCs) were investigated. Expression stabilities of the respective miRNA references were assessed applying four different algorithms (geNorm, BestKeeper, Normfinder, ∆CT). MiRNA scores were ranked according to their performance using RefFinder. Finally, the impact of using various miRNA references on data normalization was evaluated in different ALL and control samples.
Analyzing the stability of miRNAs within the ALL cohort, we identified miR-532-5p in combination with miR-103a-3p (geNorm), miR-532-5p (BestKeeper and NormFinder), and miR-103a-3p followed by miR-532-5p (∆CT; mean standard deviation of miR-103a-3p and miR-532-5p: 2.215 and 2.220, respectively) as the most stably expressed miRNAs. Importantly, the RefFinder algorithm identified miR-532-5p as the most reliable reference.
Next, we assessed the stability of miRNA expression in HC-PBMCs, and identified SNORD44 (NormFinder, BestKeeper, ∆CT) or SNORD44 in combination with RNU6 (geNorm) as the miRNAs showing the lowest variation. These data were further substantiated by using RefFinder which identified SNORD44 as the most reliable miRNA reference in HC-PBMCs. Interestingly, when we combined the cohorts of ALL and HC samples and applied RefFinder, miR-532-5p in combination with miR-103a-3p were identified as the most stable miRNAs.
Last, we analyzed the influence of using expression levels of different miRNAs for data normalization. We assessed the expression of miR-181a-5p, a miRNA described to be frequently overexpressed in ALL cells as compared to HC-PBMCs and used expression data of different miRNAs for normalization: miR-532-5p (most stably expressed in ALL cohort), miR-532-5p/miR-103a-3p (combining the most stable miRNA references in ALL cohort and HCs), or SNORD44 (most stably expressed in HC-PBMCs). Interestingly and in line with its described upregulation in ALL, we identified a significant overexpression of miR-181a-5p in BM-derived PDX samples as compared to HC-PBMCs when using miR-532-5p or miR-532-5p/miR-103a-3p for data normalization. However, when SNORD44 was used as a reference, no differential expression of miR-181a-5p in HC-PBMCs and BM-ALL cells was found indicating a critical impact of the normalization reference.
In summary, we identified miR-532-5p to be constantly expressed across different leukemia samples providing a reliable miRNA reference for data normalization in ALL miRNA expression studies. Moreover, these data clearly emphasize that the chosen reference critically depends on the cellular background of the specific samples and requires careful selection and verification within the respective sample types.
Disclosures
No relevant conflicts of interest to declare.
