Blood online
Home About Blood Authors Subscriptions Permission Advertising Public Access contact us
 

 
Advanced
Current Issue
First Edition
Archives
Submit to Blood
Search
American Society of Hematology
Meeting Abstracts
Email Alerts

Blood, Vol. 107, Issue 5, 2061-2069, March 1, 2006
This Article
Right arrow Abstract
Right arrow Full Text
Services
Right arrow Email this article to a friend
Right arrow Alert me to new issues of the journal
Right arrow reprints & permissions
Right arrow Rights and Permissions
Citing Articles
Right arrow Citing Articles via CrossRef

Identification of glucocorticoid-response genes in children with acute lymphoblastic leukemia
Blood Schmidt et al. 107: 2061

Supplemental materials for: Schmidt et al

Abbreviations:
ALL: acute lymphoblastic leukemia
BCP-ALL: B-cell precursor acute lymphoblastic leukaemia
BFM: Berlin-Frankfurt-Münster (treatment protocol)
cDNA: complementary deoxyribonucleic acid
cRNA: complementary ribonucleic acid
CHX: cycloheximide
FACS: fluorescence activated cell sorter
FITC: fluorescein isothiocyanate
GC: glucocorticoid(s)
GR: glucocorticoid receptor
MTX: methotrexate
PI: propidium iodide

Section 1. Patients and other biologic systems

1.1. Patients data

Patients included in this study. Caucasian children with acute lymphoblastic leukemia (ALL) admitted to the Department of Pediatrics, Innsbruck Medical University, during September 2000 and December 2004, and treated according to BFM protocol 2000 (for protocol details, see: www.mh-hannover.de/kliniken/paed_haeminko/all/All_stu_2.htm), were enrolled in this study. The study was approved by the Ethics Committee of the Innsbruck Medical University (EK1-1193-172/35) and written informed consent was obtained from parents or custodians. Since previous studies suggested that malignant blasts purified from peripheral blood or bone marrow may not substantially differ1, we used peripheral lymphoblasts thereby saving the children the burden of repeated bone marrow puncture. The lymphoblasts were purified from peripheral blood by Lymphoprep® gradient and, if less than 90% pure, enriched by sorting procedures as detailed in section 2 (“Expanded Methods”). If samples did not reach ≥90% purity at all three time points (see treatment details below), the donor was excluded from expression profiling. Similarly, if any of the RNAs prepared from the purified lymphoblasts did not fulfill the quality requirements outlined in Section 4, the patient was excluded. For comparison purposes, a 72a-old Caucasian male with BCP-ALL treated at the Department of Hematology and Oncology of the Innsbruck Medical University was included after giving written informed consent.

Blood sampling and treatment details. EDTA blood was taken by venous puncture prior to, and at 6-8h and 24h after initiation of treatment with glucocorticoids (GC). To avoid tumor lysis syndrome, the daily GC dose was gradually increased over the first 3-4 days to reach 60mg prednisolone/m2/per day. Treatment was initiated with a single intravenous or oral application of 6-38% of the final dose depending on peripheral blast counts, T or B-cell phenotype and clinical conditions. On the second day, the children received 30-60% of the final GC dose distributed over 3 applications (morning, noon, evening). To account for treatment differences, GC bioactivity was determined in the sera (Table 1, main text). According to the BFM protocol, all children received an intrathecal injection of methotrexate (MTX, age-dependent dose: <1a, 6mg; 1a, 8mg; 2a, 10mg; ≥3a, 12mg) at the onset of treatment with GC. However, significant MTX leakage through the blood brain barrier within the first 24h was unlikely because the number of MTX response genes regulated in the children’s lymphoblasts did not differ from those of an adult patient who did not receive MTX (see Table S8 in Section 4, “Quality control”). This patient, a 72 year-old male with BCP-ALL (see Table 1 in the main text), received a single oral application of 20mg dexamethasone at treatment initiation on day 1 and another 12 mg in the morning of day 2.

Clinical signs of GC response. All patients responded to the treatment with a reduction of peripheral lymphoblasts within the first 24h. Annexin V staining (Annexin V-PE, Becton-Dickinson, Franklin Lakes, NJ) of peripheral blood cells was performed in 7 cases, however, the absence of significant numbers of Annexin V- positive cells suggested that such cells, if present, might be rapidly cleared from the circulation or lost during the Lymphoprep® cell preparation procedure. All children except BCP-ALL-24 scored as “prednisolone good responders” by day 8 as defined by the BFM protocol (<1000 blast/µl on day 8).

Ex vivo treatment. Sorted peripheral lymphoblasts from patient BCP-ALL-40 were cultured in RPMI 1640 (Bio Whitaker, Rockland, ME, USA) supplemented with 10% fetal calf serum (Sigma, Vienna, Austria), 100U/ml penicillin, 100µg/ml streptomycin and 2mM L-glutamine (GibcoBRL, Paisley, UK), at 37°C, 5% carbon-dioxide and saturated humidity in the presence of 10-7M dexamethasone or 0.1% ethanol as vehicle control for 6h and 24h.

1.2. In vitro models of GC sensitivity, resistance and restored sensitivity

All cell lines were tested for, and found to be free of, mycoplasma infections and their authenticity was verified by DNA fingerprinting2. Cells were grown in RPMI 1640 (Bio Whitaker, Rockland, ME, USA) at 37°C, 5% carbon-dioxide and saturated humidity. Media were supplemented with 10% fetal calf serum (Sigma, Vienna, Austria), 100U/ml penicillin, 100µg/ml streptomycin and 2mM L-glutamine (GibcoBRL, Paisley, UK).

GC sensitivity models (“S-lines”): As in vitro models for GC-induced leukaemia apoptosis we used CCRF-CEM-C7H2 T-ALL cells3 and preB697 B-ALL cells4 (recently renamed EU-3 by the original author: harry_findley@oz.ped.emory.edu). Both cell lines undergo almost complete cell death after 48-72h incubation with 10-7M dexamethasone.

As GC resistance models (“R-lines”), CEM-C15, CEM-C7R16, CEM-C7R1low and PreB697-R4G4 were used. The GC-resistant CEM-C1 cell line has the same GR genotype as GC-sensitive CEM-C7H2 (GRwt/L753F) but the former expresses less GR and, contrary to the latter, does not increase GR expression upon GC exposure5,7. CEM-C7R1 are resistant because they lack functional GR (GRL753F/ΔT740)6. CEM-C7R1dim-low is a clonal derivative of CEM-C7R1 stably transfected with a vector constitutively expressing low levels of the human GR mutant A458T (for further technical details see Section 2). Although this cell line regulates considerably more probe sets than the parental C7R1 line (see the corresponding table in the internet database) it is still GC-resistant. PreB697-R4G4 is a GC-resistant subclone of PreB697 generated by selection culture in the presence of 10-7M dexamethasone similar as described for CEM-C7R16.

Restored GC sensitivity (converted or “C-lines”): GC sensitivity was restored in resistant CEM-C1 cells by stable, constitutive expression of rat GRwt (CEM-C1ratGR clone C1-4G4)5, and in resistant CEM-C7R1 by high level expression of human GRA458T (CEM-C7R1dim-high) as detailed in Section 2. CEM-C1ratGR and CEM-C7R1dim-high were included in this study because they regulate sets of genes that only partially overlap with the one regulated by human GRwt in C7H2 or PreB-697. Since all three GRs (human GRwt, ratGRwt and human GRdim at high levels) trigger cell death, we reasoned that co-regulation of a given gene in all three systems might support its possible functional role in cell death induction, whereas lack of regulation in any one system might argue against it.

1.3. Cycloheximide treatment

To assess whether gene regulations were dependent upon protein biosynthesis, we treated 2.5×106/ml CCRF-CEM-C7H2 cells with 10-7M dexamethasone or 0.1% ethanol as vehicle control in the presence of 10µg/ml cycloheximide (CHX) for 6h. This dose of CHX has been shown in preliminary experiments to completely inhibit protein biosynthesis as measured by 35S-methionine incorporation. The cells were recovered by centrifugation and snap frozen at -80°C for RNA preparation.

1.4. Mouse in vivo and in vitro models

For in vitro GC response, three 4-6 week-old male CD-1 mice were sacrificed by cervical dislocation, their thymuses pooled, single cell suspensions prepared and 5×106 cells per ml cultured in RPMI 1640 supplemented with 10% fetal calf serum, 100U/ml penicillin, 100µg/ml streptomycin and 2mM L-glutamine at 37°C, 5% carbon-dioxide and saturated humidity in the presence of either 10-7M dexamethasone or 0.1% ethanol as vehicle control for 4h.

To determine the in vivo response to GC, groups of three 4-6 week-old male CD-1 mice were injected intraperitoneally with 0.2mg dexamethasone per mouse or phosphate buffered saline, and sacrificed by cervical dislocation after 4h. One group of three mice was sacrificed immediately as an additional control. Single cell suspensions of the pooled thymuses of each group were prepared and snap frozen in liquid nitrogen for RNA preparation.

1.5. Healthy donors

Two healthy adult males (STS, 32a; and RPK, 55a) were, following informed consent, treated with dexamethasone according to a similar protocol as used for the children, i.e., injected intravenously with 20mg/m2 dexamethasone and 16h later received 13mg/m2 orally. Blood was taken immediately prior to, and 6h and 24h after, treatment initiation. Peripheral mononuclear cells were purified by Ficoll separation and the cells snap frozen in liquid nitrogen for RNA preparation.

Section 2: Expanded Methods

2.1. Purification of peripheral lymphoblasts from patients and healthy donors

Peripheral blood was collected in EDTA tubes, diluted in equal volume of PBS/EDTA (2mM), layered on top of 5ml Lymphoprep® (Axis-shield PoC AS, Rodelokka, Norway), centrifuged at 800g for 20 minutes, the interphase washed and resuspended in PBS. The percentage of blasts in the sample was determined by FACS (FACS Calibur, Becton Dickinson, Franklin Lakes, NJ, USA) using mouse monoclonal antibodies against CD3, CD7, CD10, CD19, CD23 and, in some cases, CD34 (DAKO, Glostrup, Denmark). If purity was ≥90%, the cells were snap-frozen for RNA preparation (T-ALL-2 and -20). Since cell purity is essential for leukemia cell expression profiling8, samples that contained less than 90% of blasts were subjected to either depletion of non malignant cells (negative selection) or enrichment of malignant cells (positive selection). Both negative and positive selection procedures were based on magnetic field separation methods. For negative selection (BCP-ALL-13, -17, and T-ALL-25), a combination of mouse anti-CD3, -CD14, -CD16, -CD23 and -CD56 antibodies (DAKO, Glostrup, Denmark) was added and complemented with anti-mouse IgG-coupled magnetic beads (DYNAbeads, DYNAL Biotech, Oslo, Norway). Cells were separated in a magnetic field, their purity verified by FACS analysis, pelleted and snap-frozen for RNA preparation. For positive selection, the cells were incubated with anti-human CD10 (BCP-ALL-24, -31, -32, -33, -37, -38, -40, adult-BCP-ALL) or anti-human CD19 (BCP-ALL-43) FITC-labeled antibodies (DAKO, Glostrup, Denmark) followed by incubation with anti-FITC-magnetic beads (MACS, Miltenyi Biotech, Bergisch Gladbach, Germany) and magnetic field separation using MACS separation columns. Isolated cells were again analyzed by FACS for purity, pelleted and snap-frozen.

2.2. Generation and characterization of transfected cell lines

The generation of CEM-C1ratGR has been detailed previously5. For generation of CEM-C7R1dim-high and CEM-C7R1dim-low, we inserted the point mutation A458T into pEF-T-hGRwt, an expression vector for hGRwt 7 using the Stratagene QuickChange site-directed mutagenesis protocol (Stratagene, La Jolla, CA). Briefly, 100ng of pEF-T-hGRwt were amplified with 200nM of the forward primer:
5´-GCACAATTACCTATGTACTGGAAGGAATGATTG-3´
and the reverse primer
5´-CAATCATTCCTTCCAGTACATAGGTAATTGTGC-3´
using Pfu DNA polymerase. The wild-type template was digested with 10U DpnI (Promega, Madison,WI) for 1h and the mutated plasmid electroporated into E.coli. Plasmid DNA prepared from the resulting colonies was screened by restriction endonuclease digestion of the BseNI site created by the A458T mutation and positive plasmids subsequently verified by DNA sequencing (MWG Biotech, Ebersberg, Germany). The mutated GR sequences were amplified by PCR using Pfu DNA polymerase and the primers:
5´-TATACAATTGCCACCATGGACTCCAAAGAATCATTAAC-3´
and
5´-TATAGCGGCCGCTCACTTTTGATGAAACAGAAG-3´,
cut with MunI and NotI (Promega, Madison, WI) and inserted into EcoRI/NotI linearized dephosphorylated retroviral vectors pLIB-IGN and pLIB-IP resulting in pLib-IGN-hGRA458T and pLib-IP-hGRA458T. Plasmid pLib-IGN (a kind gift of R. Gräser, KTB, Freiburg i.Br., Germany) and pLib-IP are retroviral expression vectors coding for a bicistronic EMCV-IRES containing mRNA for the gene of interest and a GFP-Neomycin fusion gene9 or a puromycin resistance gene, respectively.

Two µg of pLib-IGN-hGRA458T or the vector control pLib-IGN were transfected into 1×106 Phoenix packaging cells together with 1µg of a plasmid coding for vesicular stomatitis virus protein VSV-G using 10µl Metafectene™ (Biontex Laboratories GmbH, Munich, Germany) according to the manufacturer’s instructions. The retrovirus-containing supernatants were filtered through 0.45µm syringe filters (Sartorius, Göttingen, Germany) 48h after transfection and centrifuged onto 1×106 CEM-C7R1 cells for 40 min at 700 g. After 48h, the infected cells were subjected to a selection culture (1mg G418/ml) for 7 days. Neomycin-resistant CEM-C7R1 cells were subsequently infected again with the second retroviral construct (pLib-IP-hGRA458T) and selected for puromycin (1µg/ml for 3 days). The resulting cell bulk was subsequently cloned by limiting dilution and individual clones screened for expression of GR, and the sensitivity to 10-7M dexamethasone determined. Clone CEM-C7R1dim-high showed high level GR expression and was sensitive to GC-induced apoptosis whereas CEM-C7R1dim-low expressed much lower levels and remained resistant (Figure S1).

2.3. RNA preparation and characterization

For total RNA isolation, a combination of TRIreagent (MRC Inc., Cincinnati, OH) and RNAeasy spin columns (Quiagene, Valencia, CA) was used according to the manufacturers’ protocols. Briefly, up to 1×107 cells per ml were lysed with 1ml TRIreagent, 200µl of chloroform were added, the mixture vortexed, centrifuged, and the RNA from the aqueous phase precipitated with isopropanol. The pelleted RNA was washed in 70% ethanol in DEPC-water, resuspended in 100µl nuclease-free water, and 350µl of RLT buffer containing 1% -mercaptoethanol and 250µl of 100% ethanol were added. The solution was centrifuged through a silica-gel column (RNAeasy) for 15 seconds at 8000g. After several washing steps, the RNA was eluted in 14µl nuclease free water. RNA quantity and purity was determined by optical density measurements (OD260/280) and RNA integrity by using the 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Only high quality RNA was further processed.

2.4. Microarray analysis

Target labelling and hybridization. For Affymetrix GeneChip analysis, 1.5µg total RNA was processed to generate a biotinylated hybridization target using “One Cycle cDNA Synthesis” and “One Cycle Target Labelling” kits from Affymetrix (Affymetrix, Santa Clara, CA). All procedures were performed according to the manufacturer’s protocols (Gene Expression Analysis, Technical Manual, Revison #4; Eukaryotic Target Preparation, Revison #3). In brief, total RNA was reverse- transcribed into cDNA using an anchored oligo-dT-T7-Primer, converted into double-stranded cDNA and purified with the “Affymetrix Sample Clean-up Kit” according to the manufacturer’s protocol. Thereafter, cRNA was generated by T7 polymerase-mediated in vitro transcription including a modified nucleotide for subsequent biotinylation. Following RNA purification, 20µg of cRNA were fragmented at 95°C using the Affymetrix fragmentation buffer, mixed with 200µl hybridization buffer containing hybridization controls and hybridized to U133 Plus 2.0 microarrays. The arrays were stained and washed in an Affymetrix fluidic station 450 following the EukGE-ws2v4 protocol. Fluorescence signals were recorded by an Affymetrix scanner 3000 and image analysis performed with the GCOS software (version 1.2).

Data processing and presentation. Data processing and analysis was performed in R using Bioconductor11 version 1.5. GeneChip raw expression values were normalized and summarized using the robust multiarray analysis (RMA) method proposed by R. Irizarry12,13. The annotation of the Affymetrix probe sets was carried out using Bioconductors hgu133plus2 annotation package version 1.6.8. Normalized expression values were inserted into a database employing Bioconductor’s developmental package “maDB” to facilitate later analysis. Regulation values (M-values) were calculated using the formula M=log2(R/G) where R and G are the normalized expression values from a sample and the according control sample respectively.

2.5. Real time RT-PCR

For cDNA synthesis, 500ng total RNA was reverse transcribed using Superscript II (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. Sufficient amounts of RNA were available from all time points (0h, 6h and 24h) from patients BCP-ALL-13,-17,-31,-32,-33,-38,-39,-40,-43 and from T-ALL patients -20 and -25. In T-ALL-2, BCP-ALL-13 and BCP-ALL-37, RNA was only available from time points 0h and 6h. In brief, 250ng hexamer random primers were added to 500ng RNA, incubated at 70°C for 10 minutes, placed on ice and 4ul 5× first strand buffer, 2µl 0.1M DTT and 1µl of 10mM dNTP mix were added. The reaction was incubated at RT for 10 minutes, at 42°C for 2 minutes, and (after addition of 1µl Superscript) at 42°C for another 50 minutes, and was terminated by incubation at 70°C for a final 15 minutes.

For real time PCR, 50µl of diluted cDNA (2ng/µl) were added to 50µl of TaqMan Universal MasterMix (Applied Biosystems, Foster City, CA) and introduced into microfluidic cards according to the manufacturer’s guidelines: after equilibration at RT, the canals were filled with 100µl of reaction mix and centrifuged two times for 1 minute at 1000 rpm. Thereafter, the cards were sealed, loaded into the HT7900 real time machine (Applied Biosystems), and run with a 2-step PCR thermo-protocol that included an initial 94.5°C step for 10 min followed by 40 cycles of 97°C for 15 sec alternating with 60°C for 1 min. Fluorescence signal intensities were read during the 60°C temperature step. Real time PCR data were analyzed with SDS software version 2.2.1.

In addition to the genes on the microfluidic card (tabulated in Section 4.3), the following genes were analyzed by real time RT-PCR using ABI pre-manufactured assays as described above, but using 96- well plates rather than fluidic cards on a BioRad i-cycler (BioRad, Hercules, CA): c-myc (Hs.00153408_m1), BTNL9 (Hs.00537320_m1), SNF1LK (Hs.00545020_m1), GZMA (Hs.00196206_m1) and GPR65/TGAG-8 (Hs.00269247_s1). The latter was used to compare fluidic card data with conventional PCR and resulted in a good correlation (R2 = 0.6488). The results of the above analyses are summarized in Section 4 (Quality control).

Section 3: Expanded Results

3.1. ALL subgroup classification by gene clustering (Figure S2)

3.2. Performance of candidates identified in experimental systems

Previous expression profiling analyses in experimental systems of GC-induced apoptosis (human and mouse ALL and myeloma cell lines, mouse thymocytes) suggested a number of candidate genes. Our recent review compiled the most frequently GC-regulated genes in these systems (see Table 1 in reference)15. One-hundred and six probe sets corresponding to the 31 genes of this list were identified on the U133 plus 2.0 microarray (either via a BLAST search against the GenBank data-base or directly by using the UniGene numbers) and the corresponding M-values of all children at both time points derived from our data base. Redundant probe sets were condensed by using the strongest regulation for each data point and the data entered into Table S1. The exceptions were the 8 probe sets for the BCL2L11 locus: here we show the 3 regulated probe sets individually, because they may correspond to different transcripts: Probe set 1555372_s_at (BCL2L11-1 in Table 1) matches the 3´ end of a multiple myeloma-derived cDNA referred to as “Bam”16; 225606_at (BCL2L11-2 in Table 1) is located about 1kb downstream of the currently known 3´ end of Bim transcripts, and 1558143_s_at (BCL2L11-3 in Table 1) recognizes the reported 3´ end of the major Bim isoforms (including BimEL, BimL, BimS)17-19. Interestingly, 208536_at that recognizes the BH3 domain as well as surrounding mRNA sequences present in BimEL, BimL, and BimS (but not Bam), was not regulated in any of the children but was induced in CCRF-CEM cells and in the adult patient. The remaining 4 probe sets corresponding to the BCL2L11 locus were either unregulated in all systems tested (1561844_s_at), or regulated in some of the additional systems but not in children (1553088_s_at, 222343_at, 1553096_s_at). LDH-A and MAP2K3 are not among the genes frequently found in experimental systems15. The former was included because its regulation in CCRF-CEM-C7H2 cells formed the basis for suggesting that GC might cause apoptosis via its effects on metabolism20. Regulation of MAP2K3 has recently been implied in GC-induced apoptosis in S49 and CCRF-CEM cells21. For this gene, there were 3 probe sets on the array. As with the other genes represented by more than 1 probe set, the M-value from the probe set with the strongest regulation was entered into Table 1. Nevertheless, with the exception of BCP-ALL-13, no patient revealed an even marginal (M-value 0.7 to 0.9) regulation.

3.3. Defining an initial list of regulated probe sets

Table 2 was generated by compiling probe sets that were regulated with an M-value of ≥1 (2-fold and more) in 7 or more of 13 ALL children at 6-8h and/or 24h. The resulting 128 probe sets were divided into 2 parts. The probe sets in “Part A” (Table 2a and b) fulfilled an additional requirement, i.e., they were regulated in 6 or more patients at 6-8h with M≥0.7 and were thus considered to encompass probe sets corresponding to primary GC response genes. Probe sets in “Part B” may include late response genes. The M-values for all 128 probe sets can be derived from the M-value table in the internet (www.ncbi.nlm.nih.gov/geo), the M-values of the top 62 probe sets are depicted in Table S3.

3.4. Comparative expression profiling – rationale and performance of the “top 62” probe sets

Rationale. Expression profiling within a single biologic system, such as childhood ALL, generates lists of most frequently regulated genes but their function in the investigated response remains unclear. The golden standard to address this question is gene over-expression or knock-out/down in model systems. In the case of GC-induced leukemia apoptosis this approach is, however, met with considerable difficulties. As will be shown below, most of the strong human candidate genes are not regulated in the mouse and cannot be tested by knock-out studies in this species. The 2 top candidates that were also regulated in the mouse (FKBP51 and DDIT4/Dig2) have already been tested functionally in human ALL cell lines and found to confer GC resistance22,23. Thus, either the 2 genes do not play a functional role in GC-induced apoptosis in patients or functional testing in cell lines did not reflect the clinical situation. The latter possibility was supported by the observation that genes whose regulation was considered responsible for GC-induced apoptosis by functional analyses in ALL cell lines were regulated in only a few, if any, patients (see Table 2 in the main text). Given the problems with current functional analysis systems, we addressed the significance of individual genes by assessing their regulations in a variety of biologic systems (comparative expression profiling). Like the above functional analyses, the comparative approach does not lead to definitive proof. However, unlike classical functional analyses that can only investigate a limited number of genes at a time, our approach provides complete data for all genes simultaneously. To address functional significance more directly, we are currently developing lentiviral systems for gene over-expression and knock-down that will allow functional testing of primary ALL cells ex vivo.

Performance of the top 62 probe sets in additional systems (Table S4). The M-values of the above 62 probe sets in the additional systems (Table 1 of the main text) were compiled in the Table S4. The mouse orthologues for the human probe sets were determined by using the “Gene” (formerly called “LocusLink”) and “HomoloGene” NCBI databases. For genes where the mouse contained more probe sets than the human array, the mouse probe set(s) with the best regulation was included. When there were fewer probe sets in the mouse than in the human, NA (not applicable) was entered in the respective row in Table 4.

3.5. Identification and characteristics of cell cycle genes (Table S5A and Table S5B)

Among the repressed 37 probe sets in Table S2, expression of a subgroup of 34 probe sets was significantly higher in cell lines than in the other investigated systems (Tables S5A and S5B). These probe sets corresponded to 27 genes known to be regulated during progression through the S-, G2- and M-phases of the cell division cycle24,25. Among primary cells, lymphoblasts from the ALL patients had higher levels than peripheral lymphocytes from 2 healthy donors, probably reflecting the higher proliferative state of the former. Interestingly, T-ALL-2 showed expression (E)-values only slightly higher than those seen in the two healthy donors, which might partly explain the absence of repression of the cell cycle genes (in fact there was a small induction) in this child.

Section 4: Quality control and data validation

4.1. Individual microarray performance

Expression analysis quality assessment parameters included 3´ to 5´ ratio of house-keeping control genes and percentage of detection calls. The mean value for 3´ to 5´ ratios for GAPDH was 1.4±0.66 SD and that for percent of present called genes was 44.4±2.96 SD. The individual performance of all U133 2.0 plus microarrays is given in Table S2. The elevated 3´ to 5´ ratio in patient 24 was caused by the fact that 2nd round amplification had to be performed for all 3 time points in this patient because of insufficient amounts of RNA for the 24h time point. This technique generally leads to less optimal 3´ to 5´ ratios. However, even the worst value (5.99 for the 24h time point) remained within reasonable limits and the percent present call was not negatively affected at all. Hence the patient was not excluded from the study.

4.2. Data reproducibility, effect of cell sorting

Numerous publications 14,26-29 have shown the high degree of reproducibility of data generated by the Affymetrix system and this reproducibility has even been improved in the most recent version used in this study. In full agreement, when the same RNA was used twice for target preparation (technical replicate of BCP-ALL-32-0h), only 32 of the 54000 probe sets differed by M≥1. In another technical replicate, performed with 1µg and 5µg starting RNA, respectively (replicate of T-ALL-25-6h), 25 of 54000 probe sets appeared differentially expressed. Finally, although comparisons were generally made between equally treated cells (i.e., either unsorted at all time points or sorted at all time points) to avoid possible gene regulations introduced by the sorting procedure, we included a comparison between sorted and unsorted cells. For this purpose, we used a bone marrow sample with over 90% malignant lymphoblasts prior to and after sorting (BCP-ALL-37-BM—sorted versus —unsorted). In this setting, 268 probe sets were differentially expressed, several of which were related to genes expressed in erythrocytes and reticulocytes. This might be explained by the fact that sorting removes erythrocytes/reticulocytes that may still be present after Lymphoprep®-purification. Most importantly however, none of the regulated probe sets coincided with any of the GC-regulated genes depicted in Table 3 of the main text.

4.3. Data validation by TaqMan real time RT-PCR

For verification of gene regulations detected by microarray analysis, a panel of 28 genes was analyzed by real time RT-PCR as detailed in the Supplemental Methods section (2.5.). On the fluidic card (24 genes, see below summary table), all patient samples, including the adult, were analyzed (except BCP-ALL-24 and the 24h values for BCP-ALL-13 and T-ALL-2 due to insufficient amounts of RNA) as well as 3 cell lines (CEM-C7H2, CEM-C1, CEM-C1ratGR). For SNF1LK, BTNL9, c-myc, GZMA and TGAG-8 individual RT-PCRs were performed for all three time points (0h, 6h, 24h) for the adult patient, BCP-ALL-17, 33, 40, 43, and T-ALL 20 and 25. Sufficient material for 2 time points remained for BCP-ALL-13 (0h/8h), -31 (0h/24h), -37 (0h, 6h), and T-ALL-2 (0h/8h). After removing duplicates that differed by CT values ≥0.5 (3.77% of the data points), regulations were calculated for each gene and each patient (0h compared to 6-8h or 24h) and cell line (6h exposure to dexamethasone compared to vehicle control). Subsequently, the ABI-derived regulations were correlated with the data for the corresponding probe set (and in instances with more than 1 probe set per gene, the best matching probe set) obtained with the Affymetrix system. Correlation coefficients of the actual regulation values for all 25 regulated genes are summarized in Table S7.

As exemplified in Figure 4 for the most promising candidate genes listed in Table 3 of the main text, there was a good correlation between these two mRNA quantification techniques in the majority of cases. The exception was the highly regulated probe set 208078_s_at that was originally assigned to TCF8. The corresponding real time RT-PCR did, however, not reveal any regulation (data not shown). In search for an explanation, we blasted 208078_s_at against the genome and discovered that it matched the SNF1LK, but not the TCF8, locus (meanwhile, Affymetrix changed the assignment of 208078_s_at from TCF8 to SNF1LK). When we repeated the experiments with an SNF1LK-specific real time RT-PCR, we found the expected strong correlation (R2 = 0.921, Figure 4). Another important evidence derived from the real time RT-PCR experiments: regulations obtained with the unassigned probe set 228854_at that mapped about 4kb downstream of the proposed polyadenylation site of ZBTB16 (but not to any known ZBTB16 mRNAs) showed strong correlation (R2 = 0.8266) with the ZBTB16 real time RT-PCR data (Figure 4), suggesting that an unknown, alternatively polyadenylated, ZBTB16 mRNA was expressed (and regulated) in the investigated lymphoblasts. Thus, although the real time RT-PCR experiments supported the general quality of microarray data, they also underscored the importance of data validation for each individual case.

4.4. Analysis of methotrexate response genes

Following the BFM protocol, children received an intrathecal injection of methotrexate in an age-dependent dose (<1a, 6mg; 1a, 8mg; 2a, 10mg; ≥3a, 12mg) concomitant with the initiation of the 1-week GC monotherapy. To assess whether leakage of the drug through the blood brain barrier might have occurred within the first 24 h to an extent that altered gene expression, we tested possible regulation of known methotrexate in vivo response genes26. The corresponding list of 97 probe sets was kindly provided by Drs. Evans and Cheok (St. Jude Chidren’s Hospital) and used to determine the corresponding probe sets present on the U133 plus 2.0 array by annotation of the original probe sets on the U95A array to “Gene” (formerly “LocusLink”) identifiers and subsequently assigning the respective identifiers to Affymetrix U133 probe sets. The response of all children (and, as control, the adult patient who did not receive methotrexate) to the resulting 244 probes sets was tabulated and the redundant probe sets condensed to 1 probe set per gene by using the probe set that gave the highest regulation (M-value) in a given patient. Thirty-eight of the resulting 83 methotrexate response genes were unregulated in all 13 children, another 26 were regulated in ≤ 3 children. The 45 genes that were regulated in one or more children are summarized in Table S8. On an individual basis, 11 ± 4.8 genes (mean of all 13 children ±SD) were regulated per child compared to 10 genes in the adult not treated with methotrexate. Thus, we concluded that intrathecal methotrexate did not significantly affect gene expression outside the blood brain barrier.

Section 5: Guide to complete data set access via the internet

All primary data of this publication, i.e., the 87 original cel-files corresponding to the 87 U133 plus 2.0 arrays listed in Table S6 (Section 4), can be accessed via the internet (www.ncbi.nlm.nih.gov/geo; GEO accessions: GSE2677, GSE2842, GSE2843). The files are named according to array designations in Table S6 followed by the extension “.cel” (except that the prefix “BCP”-ALL was changed to “B”-ALL for brevity). Although all other data can be generated from these 87 files, we provide a number of additional files for the reader’s convenience. These include 55 Excel files containing all probe sets differentially expressed (M≥1) in the corresponding comparisons (26 for ALL children, 26 for additional systems, and 3 for controls). The files are named by the designations of the compared arrays separated by vs. (versus). Positive regulation values (M-values) indicate higher expression in the first array in the file name, negative M-values the converse. Further, we provide 3 tables of all probe sets with their respective M-values and another 3 for their expression (E) - values, 1 containing all 26 comparisons of ALL children, 1 for the additional systems except mouse thymocytes, and 1 for mouse thymocytes in vivo and in vitro. These lists enable easy analysis of potential regulation or expression of essentially any known gene in all systems investigated in this study.

Section 6: References

1. Yeoh EJ, Ross ME, Shurtleff SA et al. Classification, subtype discovery, and prediction of outcome in pediatric acute lymphoblastic leukemia by gene expression profiling. Cancer Cell. 2002;1:133-143.
2. Parson W, Kirchebner R, Mühlmann R et al. Leukemia cell line identification by short tandem repeat profiling: power and limitations. FASEB J. 2005;19:434-436.
3. Strasser-Wozak EMC, Hattmannstorfer R, Hála M et al. Splice site mutation in the glucocorticoid receptor gene causes resistance to glucocorticoid-induced apoptosis in a human acute leukemic cell line. Cancer Res. 1995;55:348-353.
4. Findley HW, Jr., Cooper MD, Kim TH, Alvarado C, Ragab AH. Two new acute lymphoblastic leukemia cell lines with early B-cell phenotypes. Blood. 1982;60:1305-1309.
5. Geley S, Hartmann BL, Hala M et al. Resistance to glucocorticoid-induced apoptosis in human T-cell acute lymphoblastic leukemia CEM-C1 cells is due to insufficient glucocorticoid receptor expression. Cancer Res. 1996;56:5033-5038.
6. Hala M, Hartmann BL, Böck G, Geley S, Kofler R. Glucocorticoid receptor gene defects and resistance to glucocorticoid-induced apoptosis in human leukemic cell lines. Int J Cancer. 1996;68:663-668.
7. Riml S, Schmidt S, Ausserlechner MJ, Geley S, Kofler R. Glucocorticoid receptor heterozygosity combined with lack of receptor auto-induction causes glucocorticoid resistance in Jurkat acute lymphoblastic leukemia cells. Cell Death Differ. 2004;11 Suppl 1:S65-S72.
8. de Ridder D, van der Linden CE, Schonewille T et al. Purity for clarity: the need for purification of tumor cells in DNA microarray studies. Leukemia. 2005;19:618-627.
9. Geley S, Kramer E, Gieffers C et al. Anaphase-promoting complex/cyclosome-dependent proteolysis of human cyclin A starts at the beginning of mitosis and is not subject to the spindle assembly checkpoint. J Cell Biol. 2001;153:137-148.
10. Martin SJ, Reutelingsperger CPM, Green DR. Annexin V: a specific probe for apoptotic cells. In: Cotter TG, Martin SJ, eds. Techniques in apoptosis. London: Portland Press Ltd; 1996:107-120.
11. Gentleman RC, Carey VJ, Bates DM et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol. 2004;5:R80.
12. Irizarry RA, Bolstad BM, Collin F et al. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res. 2003;31:e15.
13. Irizarry RA, Hobbs B, Collin F et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003;4:249-264.
14. Ross ME, Zhou X, Song G et al. Classification of pediatric acute lymphoblastic leukemia by gene expression profiling. Blood. 2003;102:2951-2959.
15. Schmidt S, Rainer J, Ploner C et al. Glucocorticoid-induced apoptosis and glucocorticoid resistance: Molecular mechanisms and clinical relevance. Cell Death Differ. 2004;11 Suppl 1:S45-S55.
16. Claudio JO, Masih-Khan E, Tang H et al. A molecular compendium of genes expressed in multiple myeloma. Blood. 2002;100:2175-2186.
17. U M, Miyashita T, Shikama Y, Tadokoro K, Yamada M. Molecular cloning and characterization of six novel isoforms of human Bim, a member of the proapoptotic Bcl-2 family. FEBS Lett. 2001;509:135-141.
18. Abrams MT, Robertson NM, Yoon K, Wickstrom E. Inhibition of glucocorticoid-induced apoptosis by targeting the major splice variants of BIM with siRNA and shRNA. J Biol Chem. 2004;279:55809-55817.
19. Bachmann PS, Gorman R, MacKenzie KL, Lutze-Mann L, Lock RB. Dexamethasone resistance in B-cell precursor childhood acute lymphoblastic leukemia occurs downstream of ligand-induced nuclear translocation of the glucocorticoid receptor. Blood. 2005;105:2519-2526.
20. Tonko M, Ausserlechner MJ, Bernhard D, Helmberg A, Kofler R. Gene expression profiles of proliferating versus G1/G0 arrested human leukemia cells suggest a mechanism for glucocorticoid-induced apoptosis. FASEB J. 2001;15:693-699.
21. Miller AL, Webb MS, Copik AJ et al. p38 MAP Kinase is a Key Mediator in Glucocorticoid-induced Apoptosis of Lymphoid Cells: Correlation Between p38 MAPK Activation and Site-specific Phosphorylation of the Human Glucocorticoid Receptor at Serine 211. Mol Endocrinol. 2005;19:1569-1583.
22. Wochnik GM, Ruegg J, Abel GA et al. FKBP51 and FKBP52 differentially regulate dynein interaction and nuclear translocation of the glucocorticoid receptor in mammalian cells. J Biol Chem. 2004;280:4609-4616.
23. Wang Z, Malone MH, Thomenius MJ et al. Dexamethasone-induced gene 2 (dig2) is a novel pro-survival stress gene induced rapidly by diverse apoptotic signals. J Biol Chem. 2003;278:27053-27058.
24. Whitfield ML, Sherlock G, Saldanha AJ et al. Identification of genes periodically expressed in the human cell cycle and their expression in tumors. Mol Biol Cell. 2002;13:1977-2000.
25. Fontijn RD, Goud B, Echard A et al. The human kinesin-like protein RB6K is under tight cell cycle control and is essential for cytokinesis. Mol Cell Biol. 2001;21:2944-2955.
26. Cheok MH, Yang W, Pui CH et al. Treatment-specific changes in gene expression discriminate in vivo drug response in human leukemia cells. Nat Genet. 2003;34:85-90.
27. Holleman A, Cheok MH, den Boer ML et al. Gene-expression patterns in drug-resistant acute lymphoblastic leukemia cells and response to treatment. N Engl J Med. 2004;351:533-542.
28. Ferrando AA, Neuberg DS, Staunton J et al. Gene expression signatures define novel oncogenic pathways in T cell acute lymphoblastic leukemia. Cancer Cell. 2002;1:75-87.
29. Kohlmann A, Schoch C, Schnittger S et al. Pediatric acute lymphoblastic leukemia (ALL) gene expression signatures classify an independent cohort of adult ALL patients. Leukemia. 2004;18:63-71.

Files in this Data Supplement:



This Article
Right arrow Abstract
Right arrow Full Text
Services
Right arrow Email this article to a friend
Right arrow Alert me to new issues of the journal
Right arrow reprints & permissions
Right arrow Rights and Permissions
Citing Articles
Right arrow Citing Articles via CrossRef

click for free articles
home about blood authors subscriptions permissions advertising public access contact us
  Copyright © 2008 by American Society of Hematology         Online ISSN: 1528-0020