Protein Expression in Acute Myelogenous Leukemia Stem Cells Compared to Bulk AML Cells Using Reverse Phase Proteins Arrays (RPPA)

The majority of studies of Acute Myelogenous Leukemia (AML) cell biology that are performed on patient derived samples utilize the more readily available bulk mononuclear leukemia cells (BULK) instead of rare stem cells (SC), defined as CD34+CD38− for this analysis. However, resistance of SC to therapy is thought to be more clinically relevant with SC leading to relapse and mortality. If the cellular machinery within the SC compartment is significantly different from that of the Bulk cells then these studies could generate results of uncertain veracity arising from studying the wrong cell. The ability to measure gene and protein expression in SC would answer this question but to date conventional methodology has required too many cells to permit the analysis of protein expression and activation patterns within SC. Using RPPA methodology on blood (PB) or marrow (BM) derived AML cells we have previously demonstrated that recurrent protein expression signatures with prognostic implications exist in AML. Our RPPA methodology requires only 333 cells per slide making the analysis of protein expression in stem cells possible for the first time. We therefore set out to determine whether protein expression in AML SC differed from BULK cells. Leukemia enriched (CD3/CD19 depleted) mononuclear cell fractions from 103 AML samples from 87 patients (including 13 paired PB and BM and 5 paired diagnosis and relapse samples) were separated in to CD34− and CD34+ fractions, followed by separation of the CD34+ fraction into CD34+CD38+ and CD34+CD38− fractions using magnetic antibody selection. Among the 103 samples 8 had only CD34+/− samples, 31 had BULK/CD34+/CD34−, 18 had BULK/CD34+/CD34−/CD34+CD38+, and 45 had all 5 samples. Whole cell lysates and total RNA preparations were made from these fractions. RPPA were printed using 5 serial 1:2 dilutions, printed in replicate. Slides were stained with 127 antibodies (predominantly apoptosis, cell cycle and STP regulating proteins) detecting 87 total, 37 phopho and 3 caspase cleavage sites. Spot intensities were quantified using MicroVigene software. Data was analyzed using R, with loading control and topographical background normalization being utilized. Analysis was performed using a mixed-effects model and β-uniform model. We compared expression between the BULK, and SC fractions, between CD34+ and CD34− fractions and between CD34+CD38+ and C34+CD38−(SC) fractions. In univariate analysis, protein expression in SC was higher for Caspase8, ZNF342, α-Catenin, and TRIM24, and lower for, Bax, pP70S6K and pS6RP across the entire population when compared to protein from BULK leukemic cells. The fold differences were subtle ranging from 15 to 27% increases and 13 to 38% decreases in expression in the SC relative to the BULK cells. There was no statistically significant evidence for global differences in individual protein expression for the CD34+/CD34− or CD34+CD38+/CD34+CD38− comparisons. In multivariate analysis, when patterns of expression of all 127 proteins were compared, the majority of samples from individual patients showed close correlation regardless of whether BULK/SC, CD34+/CD34− or CD34+CD38+/CD34+CD38− were compared. Gene expression profiling of these same samples using mRNA amplification techniques is underway. Comparison of differences in GEP and RPPA will be performed at a later date. In summary, the RPPA technique has enabled the analysis of expression of numerous proteins and their activation state in AML stem cells for the first time. We observed that protein expression in SC generally is similar to that of BULK cells. This increases confidence in the ability to generalize the results obtained in studies of bulk AML cells. A few proteins did show consistent differences in expression between stem and bulk cells. Novel agents targeting these should be investigated for the ability to individually or synergistically target AML SC.