Leukemia (2008) 22, 370–377 & 2008 Nature Publishing Group All rights reserved 0887-6924/08 $30.00 www.nature.com/leu ORIGINAL ARTICLE The BCL2 rheostat in glucocorticoid-induced apoptosis of acute lymphoblastic leukemia C Ploner1, J Rainer2, H Niederegger2, M Eduardoff2, A Villunger3, S Geley1 and R Kofler1,2 1Division Molecular Pathophysiology, Department Biocenter, Medical University of Innsbruck, Innsbruck, Austria; 2Tyrolean Cancer Research Institute, Innsbruck, Austria and 3Division of Developmental Immunology, Department Biocenter, Medical University of Innsbruck, Innsbruck, Austria Glucocorticoid (GC)-induced apoptosis is essential in the GC might induce cell death by directly regulating genes treatment of acute lymphoblastic leukemia (ALL) and related controlling cell survival and apoptosis, or via (de)regulating malignancies. Pro- and anti-apoptotic members of the BCL2 genes or gene networks leading to cellular distress which, in family control many forms of apoptotic cell death, but the extent to which this survival ‘rheostat’ is involved in the beneficial turn, constitutes an apoptotic stimulus. In both scenarios, effects of GC therapy is not understood. We performed members of the large family of pro- and anti-apoptotic BCL2 9,10 systematic analyses of expression, GC regulation and function proteins, referred to as the ‘BCL2 rheostat’, might be of BCL2 molecules in primary ALL lymphoblasts and a involved either as direct GR targets or as sensors for potentially corresponding in vitro model. Affymetrix-based expression harmful GC effects.11 In addition, the status of the BCL2 profiling revealed that the response included regulations of rheostat, regardless of whether altered during GC exposure or pro-apoptotic and, surprisingly, anti-apoptotic BCL2 family members, and varied among patients, but was dominated by not, might define sensitivity to, and kinetics of, GC-induced cell induction of the BH3-only molecules BMF and BCL2L11/Bim death. The latter issue was addressed in great detail in mice and repression of PMAIP1/Noxa. Conditional lentiviral gene showing that GC-induced thymocyte apoptosis was impaired by overexpression and knock-down by RNA interference in the transgenic expression of anti-apoptotic or knockout of some pro- CCRF-CEM model revealed that induction of Bim, and to a apoptotic BCL2 family members (reviewed in Distelhorst,6 lesser extent that of BMF, was required and sufficient for Strasser,9 Ranger et al.10 and Labi et al.12). In human ALL, only apoptosis. Although anti-apoptotic BCL2 members were not a few of these genes have been functionally tested in this regulated consistently by GC in the various systems, their 13 overexpression delayed, whereas their knock-down acceler- respect. For instance, overexpression of BCL2 and knock- ated, GC-induced cell death. Thus, the combined clinical and down of the BCL2 homology domain 3 (BH3)-only molecule experimental data suggest that GCs induce both pro- and anti- BCL2L11/Bim14,15 interfered with, while overexpression of pro- apoptotic BCL2 family member-dependent pathways, with the apoptotic BAX16 and knock-down of anti-apoptotic MCL117 outcome depending on cellular context and additional signals sensitized for, GC-induced apoptosis in ALL cell lines (following feeding into the BCL2 rheostat. Leukemia (2008) 22, 370–377; doi:10.1038/sj.leu.2405039; a recommendation by the HUGO Gene Nomenclature Commit- published online 29 November 2007 tee, all official gene symbols are represented by uppercase letters Keywords: BCL2 rheostat; glucocorticoid-induced apoptosis; acute to distinguish them from their alternatives, for example, lymphoblastic leukemia BCL2L11 ¼ Bim). To what extent, if any, expression of BCL2 family proteins predicts GC sensitivity in patients with ALL is controversial (discussed in Schmidt3), although MCL1 has recently been suggested as major GC resistance gene that specifically protects ALL cells from GC-induced, but not chemotherapy- Introduction 17 induced, apoptosis. Taken together, the current evidence, strong in mice but less convincing in human systems, suggests that the Glucocorticoids (GCs) induce apoptosis in certain lymphoid status of the BCL2 rheostat influences GC sensitivity. cells and play an important role in the treatment of childhood Concerning the question of whether components of the BCL2 acute lymphoblastic leukemia (ALL) and other lymphoid rheostat might be regulated by GC, several BCL2 family malignancies.1 This effect is mediated by the GC receptor members responded to GC in numerous systems of GC-induced (GR), a ligand-activated transcription factor of the nuclear apoptosis, most notably Bim, which was induced in mouse receptor superfamily that resides in the cytoplasm and, upon 4,18 3,6,18 thymocytes, several leukemia cell lines, primary ligand binding, translocates into the nucleus, where it modulates 19 chronic lymphocytic leukemia cells and some patients with gene expression via binding to specific DNA response elements 4 ALL. Other reported regulations include BMF and Puma mRNA or by protein–protein interactions with other transcription 4,20 2 induction in mouse thymocytes or BCL2 and Bcl-XL protein factors. A large number of genes have been identified that are 21 repression in children with ALL. However, in a recent study regulated by GC in experimental systems of GC-induced with primary ALL cells from children treated with GC ex vivo, apoptosis3 and related clinical samples,4,5 but the genes neither Bim nor any other BCL2 family member was signifi- responsible for cell death induction remain controversial (for 5 cantly regulated. The most critical question, that is, to what recent reviews see Schmidt et al.,3 Distelhorst,6 Schaaf and extent the BCL2 rheostat responds to GC treatment in patients Cidlowski7 and Planey and Litwack8). in vivo, has not been thoroughly addressed. In this study, we performed a systematic analysis of mRNA Correspondence: Professor R Kofler, Division of Molecular Patho- expression and GC regulation of all BCL2 family members physiology, Department Biocenter, Medical University of Innsbruck, during the early phase of systemic GC mono-therapy in children Fritz-Pregl-Str. 3, Innsbruck, Tyrol A-6020, Austria. E-mail: Reinhard.Kofl[email protected] with ALL and complemented it with extended functional Received 11 September 2007; revised 22 October 2007; accepted 23 analyses in CCRF-CEM cells, a widely used model for GC- October 2007; published online 29 November 2007 induced apoptosis in childhood ALL. The BCL2 rheostat in glucocorticoid-induced apoptosis C Ploner et al 371 Materials and methods procedure. Second, we checked all probe sets (collection of eleven 25-mer oligonucleotides on the array that recognize a The details of this section are available online as Supplementary given transcript) assigned to the 21 genes of the BCL2 family9,10 Information. for correct annotation and position along the transcripts and found proper probe sets (that is, probe sets that mapped within B600 bp of the 30 end of the respective transcript) for all the 21 Results genes and almost all of their 42 RefSeq transcripts. The exceptions were one variant each of BCL2L14/Bcl-G and The BCL2 rheostat in the initial phase of GC treatment BCL2L11/Bim and three variants of the C1orf178/Bfk gene To assess expression and possible regulation of the BCL2 (Supplementary Table S1). Prior to treatment, MCL1 and BNIP3L rheostat in response to initial GC therapy in patients with ALL, were strongly expressed in all 13 childhood ALL samples, we utilized our recently established expression profiles (Affyme- whereas BCL2L10/Boo/Diva, BOK and Bcl-G were not detect- trix, U133 plus 2.0) of lymphoblasts from 13 ALL children prior able in any of them (Figure 1a). The mRNA expression pattern to, and 6–8 and 24 h after, initiation of GC treatment.4 First, all seen in the children was largely maintained in the CCRF-CEM arrays were re-normalized using GCRMA (robust multi-array and PreB697 ALL in vitro models. The mean regulations (M- average with background adjustment using sequence informa- values, log2-fold changes) of the rheostat components in the tion), which has been shown to be superior compared to the childhood ALL samples along with their significance (pBH, previously used RMA (robust multiarray analysis) normalization P-values adjusted for multiple hypothesis testing according to Figure 1 Expression and regulation of BCL2 family members in ALL cells. (a) Expression (U133 plus 2.0-derived E-values, log 2 scale) of BCL2 family members in untreated malignant lymphoblasts from 13 children and 2 in vitro models (CEM-C7H2, PreB697). An intensity scale is indicated below the graph. E-values and probe sets for this graph are depicted in Supplementary Table S3. (b) mRNA regulations of BCL2 family members in peripheral blasts from 13 ALL children at 6–8 and 24 h of systemic GC monotherapy.4 Extent of regulation (mean M) was plotted against significance (Benjamini–Hochberg adjusted P-values, expressed as negative logarithm to the power of 10). The dotted lines indicate significance of pBH ¼ 0.05 and regulation of M ¼ ±1. Genes with pBH values p0.05 are indicated. M-values and probe sets for the two ‘volcano plots’ are depicted in Supplementary Table S4. ALL, acute lymphoblastic leukaemia. Leukemia The BCL2 rheostat in glucocorticoid-induced apoptosis C Ploner et al 372 Benjamini and Hochberg) are depicted in the ‘volcano plots’ in Figure 1b, which visualize regulations on the x axis and significance of these regulations on the y axis. At the early time point (6–8 h after treatment initiation), the BH3-only molecule BMF was the only significantly regulated BCL2 family member (pBH ¼ 0.004, mean M ¼ 0.8), but four additional genes came close to the significance cutoff pBH ¼ 0.05, including the pro- survival gene BCL2A1/A1 (pBH ¼ 0.057) and the pro-apoptotic HRK (pBH ¼ 0.057) and PMAIP1/Noxa (pBH ¼ 0.055) genes. Interestingly, A1 (mean M ¼ 0.7) was induced and HRK (mean M ¼À0.3) and Noxa (mean M ¼À1.0) were repressed, pre- sumably reflecting GC-mediated pro-survival signals.