| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* Tyrolean Cancer Research Institute, Innsbruck, Austria;
Northern Institute for Cancer Research, Newcastle upon Tyne, United Kingdom;
Institute of Legal Medicine, Innsbruck Medical University, Innsbruck, Austria;
Division of Human Genetics, Innsbruck Medical University, Innsbruck, Austria;
|| Division of Molecular Pathophysiology, Biocenter, Innsbruck Medical University, Innsbruck, Austria
2Correspondence: Tyrolean Cancer Research Institute, Innrain 66, A-6020 Innsbruck, Austria. E-mail: stefan.schmidt{at}i-med.ac.at
ABSTRACT
Glucocorticoids (GCs) specifically induce apoptosis in malignant lymphoblasts and are thus pivotal in the treatment of acute lymphoblastic leukemia (ALL). However, GC-resistance is a therapeutic problem with an unclear molecular mechanism. We generated 
70 GC-resistant sublines from a GC-sensitive B- and a T-ALL cell line and investigated their mechanisms of resistance. In response to GCs, all GC-resistant subclones analyzed by real-time polymerase chain reaction (PCR) showed a deficient up-regulation of the GC-receptor (GR) and its downstream target, GC-induced leucine zipper. This deficiency in GR up-regulation was confirmed by Western blotting and on retroviral overexpression of GR in resistant subclones GC-sensitivity was restored. All GC-resistant subclones were screened for GR mutations using denaturing high-pressure liquid chromatography (DHPLC), DNA-fingerprinting, and fluorescence in situ hybridization (FISH). Among the identified mutations were some previously not associated with GC resistance: A484D, P515H, L756N, Y663H, L680P, and R714W. This approach revealed three genotypes, complete loss of functional GR in the mismatch repair deficient T-ALL model, apparently normal GR genes in B-ALLs, and heterozygosity in both. In the first genotype, deficiency in GR up-regulation was fully explained by mutational events, in the second by a putative regulatory defect, and in the third by a combination thereof. In all instances, GC-resistance occurred at the level of the GR in both models.—Schmidt, S., Irving, J. A. E., Minto, L., Matheson, E., Nicholson, L., Ploner, A., Parson, W., Kofler, A., Amort, M., Erdel, M., Hall, A., Kofler, R. Glucocorticoid resistance in two key models of acute lymphoblastic leukemia occurs at the level of the glucocorticoid receptor.
Key Words: apoptosis • mutation • mismatch repair • CCRF-CEM cell line • PreB 697 cell line
GLUCOCORTICOIDS (GC) PLAY an important role in the treatment of childhood acute lymphoblastic leukemia (ALL), as well as in other lymphoid malignancies due their effectiveness at inducing apoptosis in immature lymphoid cells (1)
. This effect is mediated by the GC receptor (GR), a ligand-activated transcription factor belonging to the nuclear receptor superfamily that resides in the cytoplasm and on ligand binding, translocates to the nucleus, and dimerizes and binds to GC response elements, thereby modulating gene expression (2)
. In addition to this, GR can also indirectly affect gene expression by protein-protein interactions with other transcription factors and has an inhibitory effect on activating protein AP-1 and NF-kappa B function (3
, 4)
.
GC sensitivity in ALL as judged by both in vitro assays and in vivo responsiveness after 8 d of GC monotherapy routinely administered in the course of the Berlin-Frankfurt-Münster (BFM) protocol, is highly variable. For example, in one in vitro study, LC50 assays of diagnostic leukemic blasts ranged from 0.05 to 2000 µg/ml and at relapse were 20–340 fold more resistant than a cohort of diagnostic blasts (5)
. In paired diagnostic/relapse samples, leukemic blasts with de novo sensitivity acquired GC resistance at relapse, while those with de novo GC resistance often acquired further resistance (5)
. Despite being used clinically for more than 50 yr, the molecular basis of GC sensitivity and resistance is not fully elucidated but is much debated since GC resistance is a continuing clinical problem (6
7
8
9
10
11
12
13
14)
. Investigations of GC resistance mechanisms in patients in vivo or ex vivo are demanding because of heterogeneity of the clinical material, infrequent occurrence of primary GC resistance, cotreatment with other antileukemic drugs in case of secondary GC resistance and technical difficulties in performing functional studies with primary leukemic cells. Thus leukemic cell line models have played a major role in GC resistance studies so far.
The most intensively studied human childhood ALL model for GC-induced leukemia apoptosis is the CCRF-CEM cell line, derived from a child with T-ALL at relapse (15)
. Earlier studies with descendents of this cell line have shown that acquired GC resistance may result from mutations in the GR gene (16
17
18)
. It is now known that all CCRF-CEM derivatives, even the GC-sensitive ones, are heterozygous for a mutation in the GR gene, i.e., L753F (16
, 19)
, which already occurred in vivo (20)
and results in an activation-labile receptor with diminished functional capability (21)
. In addition, the cell line carries a mutation in the MLH1 gene, an important component of the post-replicative mismatch repair (MMR) pathway (22)
. This mutation renders them functionally MMR deficient (23
, 24)
, a phenotype associated with an increased basal mutation rate because DNA polymerase errors occurring during normal DNA replication fail to be corrected. Thus the coexistence of a GR mutation concomitant with an increased mutation rate may create a biological environment that predisposes this cell line to mutational inactivation of the remaining wild-type (WT) GR allele, under GC selection pressure. Interestingly, several CCRF-CEM derivatives, including the CEM-C7H2 line used in this study, are pseudo-tetraploid, a feature that has also been observed in primary ALL cells, particularly of the T-ALL subtype (25)
. Thus, CEM-C7H2 cells carry two mutated and two WT GR alleles, allowing assessment of stepwise loss of GR alleles and its associated effects on GC sensitivity. We have recently developed a model system consisting of 40 GC-resistant and 20 GC-sensitive clonal derivatives of the CEM-C7H2 cell line and, based on DNA fingerprinting, provided evidence for selective loss of one of the two functional GR alleles in the resistant clones (26)
. However, whether loss of one allele suffices for GC resistance or whether other mechanisms contribute remained to be resolved.
Because MMR proficiency is highly variable in primary lymphoblasts, ranging from fully proficient to completely deficient (23)
, GC resistance mechanisms in the CCRF-CEM cell line may serve only as model for situations of leukemic blasts with reduced proficiency. We have therefore identified a MMR-proficient ALL cell line, PreB697 (27)
, which carries WT GR, generated GC resistance cell line clones and compared GC resistance mechanisms with those derived from the above CCRF-CEM-based system. We show that GC resistance occurs at the level of the GR and is due to mutational and/or regulatory mechanisms in both cell line models. However, regulatory rather than mutational changes predominate in the MMR-proficient cell line.
MATERIALS AND METHODS
Cell lines and tissue culture
CCRF-CEM-C7H2 was generated in our laboratory (18)
by limiting dilution subcloning of CEM-C7 cells kindly provided by E. B. Thompson (28)
. PreB697 (27)
(recently renamed EU-3 by the original author; harry_findley{at}oz.ped.emory.edu) was obtained from the German collection of microorganism and cell cultures (DSMZ, Braunschschweig, Germany). All cell lines were tested for, and found to be free of, mycoplasma infection and their authenticity was verified by DNA fingerprinting as detailed previously (26)
. Cells were grown in RPMI 1640 (BioWhitaker, Rockland, ME, USA) at 37°C, 5% carbon-dioxide and saturated humidity. Media were supplemented with 10% fetal calf serum (Sigma, Vienna, Austria), and 2 mM L-glutamine (Gibco BRL, Paisley, UK). The generation of GC-sensitive and GC-resistant descendants of the CEM-C7H2 has been described (26)
. GC-resistant subclones of the GC-sensitive PreB697 cell line were generated accordingly. In brief, 25,000 cells per microtiter well were cultured in the presence of 10–7M dexamethasone. After 3–4 wk of selection culture, individual clones were picked and further expanded. Several attempts to generate GC-sensitive subclones by limiting dilution in the absence of dexamethasone failed, hence the parental PreB697 cell line was used for comparisons.
Apoptosis determination
Apoptosis was determined by propidium iodide staining of nuclei (29)
as detailed previously (30)
. Briefly, cells were harvested prior to and 72 h after exposure to either 0.1% ethanol (as vehicle control) or 10–7M dexamethasone, resuspended in a buffer containing 0.1% TritonX-100, 0.1% sodium-citrate, and 0.005% propidium-iodide for 24 h and analyzed on a FACSCalibur FACS (Becton Dickinson, Schwechat, Austria) with Cell Quest Pro Software (Becton Dickinson) to assess percentage of nuclei with a reduced DNA content (sub-G1 peaks).
Real-time PCR
Total RNA was prepared from 5 x 106 cells exposed either to 0.1% ethanol (as vehicle control) or 10–7M dexamethasone for 0, 6, and 24 h using TriReagent (MRC, Cincinnati, OH, USA) according to the manufacturer’s protocol. RNA integrity was confirmed by agarose gel electrophoresis. RNA (1 ìg) was reverse-transcribed using Superscript II (Invitrogen-GmbH, Lofer, Austria) and random hexamer primers following the manufacturer’s instructions. Expressions of GR-, GC-induced leucine zipper (GILZ) (31
, 32)
, and TATA box-binding protein (TBP) were determined using gene specific primers located in adjacent exons, a Taqman probe spanning the respective exon-exon boundary and 20 ng of cDNA per reaction. Sequences of primers and probes are listed below: Primers and probe targeting GR exon 2-exon 3: Forward primer: 5'-GAACTTCCCTGGTCGAACAGTT-3'; reverse primer: 5'GAGCTGGATGGAGGAGAGCTT-3, Taqman probe: 5'FAM-TGGCTATTCAAGCCCCAGCATGAGA-3'TAMRA; Primers and probe targeting GILZ exon 1 – exon 2: forward primer 5'CTTCTCTTCTCTGCTTGGAGGG-3, reverse primer: CGATCTTGTTGTCTATGGCCAC, Taqman probe: 5'FAM-CTGGACAACAGTGCCTCCGGAGC-3'DABCYL; primers and probe targeting TBP exon 5-exon 6: forward primer: GCCCGAAACGCCGAATA, reverse primer: CGTGGCTCTCTTATCCTCATGAT, Taqman probe: 5'FAM-ATCCCAAGCGGTTTGCTGCGGT-3'Dabcyl.
All CEM-C7H2 subclones were analyzed on an ABI 7900HT instrument using TaqMan universal PCR MasterMix (Applied Biosystems, Darmstadt, Germany). The PreB697 subclones were assayed on an icycler instrument (Bio-Rad, Vienna, Austria) with a Brilliant qPCR core kit (Stratagene, Amsterdam, The Netherlands). All experiments were performed 2–3 times each in duplicate or triplicate. Target gene computed tomography (CT) values were normalized to the corresponding CT values of TBP using the delta-computed tomography method as described in "User Bulletin #2" by Applied Biosystems available at www.appliedbiosystems.com.
Generation of GC-resistant CEM-C7H2 subclones with transgenic GR expression
For generation of subclones with transgenic GR, human GR
cDNA was amplified from pEF
T-hGR (33)
with primers introducing a 5'MunI and a 3'NotI site and inserted into EcoRI/NotI restricted pLIB-MCS-iresPuro. The latter was generated by inserting a multiple cloning site followed by an internal ribosomal entry site (IRES)-driven puromycin resistance gene into the retroviral expression vector pLIB (Invitrogen, Lofer, Austria). The resulting plasmid (pLIB-hGR-IRES-Puro) was transfected into Phoenix packaging cells (34)
(kindly provided by G. Nolan), along with a plasmid coding for vesicular stomatitis virus protein VSV-G using 12 µl MetafecteneTM (Biontex Laboratories GmbH, Munich, Germany). The infectious supernatant was used to transduce GC-resistant CEM-C7H2 subclones R11G5 and R5C3. The bulk cultures were subcloned by limiting dilution in the presence of 2 µg/ml puromycin.
DNA extraction, amplification, and detection of DNA fingerprints
DNA extraction, amplification, and DNA fingerprinting have been performed as detailed previously (26)
. Briefly, 15 short tandem repeat (STR) loci and the gender specific locus amelogenin were amplified using the Geneprint PowerPlex 16 System (Promega, Madison, WI, USA) and subjected to electrophoresis on an ABI PRISM 3100 Genetic Analyzer. Data were analyzed according to the international nomenclature (35)
using GeneScan Analysis (Versions 2.1) and Genotyper (Version 2.5).
Mutational screening
GR exons 3–9 and flanking intronic sequences were amplified from 100–200 ng genomic DNA from all cell lines using PCR and were mutationally screened by denaturing HPLC using a Wave 3500 system (Transgenomic, Crewe, UK). Primers, annealing temperatures, and DHPLC parameters were identical to those described previously (36)
.
Sequencing
Direct sequencing was performed by purifying 100 µl of PCR product using a Qiaquick PCR purification kit (Qiagen Ltd., Crawley, Sussex, UK) with a final elution vol of 30 µl and then sequenced using both forward and reverse primers with the ABI Version 3 BigDye terminator cycle sequencing kit and analyzed on an ABI prism DNA sequencer (Applied Biosystems, Warrington, UK). To characterize minority PCR species, PCR products were subcloned into the Pgem-T-Easy plasmid (Promega, Southampton, UK) according to the manufacturer’s instructions and transformed into E. coli (JM109), and isolated plasmid DNA from positive transformants was then sequenced with pUC/M13 forward and reverse primers (Promega). Sequence alignments were carried out using Omiga software (Accelrys, Cambridge, UK).
Mismatch repair assay
Cytosolic extracts were prepared from CCRF-CEM and PreB 697 cell lines using a minimum of 1 x 108 cells. Cells were washed in ice-cold isotonic buffer (20 mM HEPES, pH 7.9, 5 mM KCl, 1.5 mM MgCl2, 0.2 mM PMSF, 1 mM DTT, 250 mM sucrose, 1 complete EDTA-free protease inhibitor tablet/50 ml buffer (Boehringer Mannheim), followed by ice-cold hypotonic buffer (as above but without the sucrose) then resuspended in twice the pellet vol of ice-cold hypotonic buffer. Cells were lysed mechanically using a BioSpec Mini-Beadbeater (Stratech Scientific, Luton, UK), checking the progression of the lysis microscopically. When around 85% of the cells were lysed, nuclei were pelleted and the supernatant was cleared by centrifugation at 12,000 g for 7 min. Aliquots of the lysate were stored at –80°C prior to analysis. Cytosolic protein concentration was measured using the bicinchoninic acid (BCA) protein assay kit (Pierce & Warriner, Chester, UK).
The ability of cytosolic extracts to repair DNA mismatches was assessed as described previously (37)
. Briefly, 5 ng of an M13mp2 phage heteroduplex substrate, containing a G.T mispair and a nick in the strand containing the T, was prepared and incubated with 50 µg of cell lysate and the repaired/unrepaired heteroduplex substrate was purified and electroporated into MMR-deficient E. coli. Transformed bacteria were plated out with the -complementation E. coli strain CSH50 onto minimal agar plates supplemented with isopropyl ß -D-thiogalactopyranoside and X-Gal (Sigma, Dorset, UK). The resulting plaques were scored as pure blue, mixed burst, or clear. Reduction in the percentage of mixed plaques and increase in pure blue or clear plaques is indicative of MMR. Repair efficiency was calculated as 1 – (% mixed plaques in test sample divided by % mixed plaques in a water only control). The M13mp2 phage derivatives and E. coli strains were kindly provided by Dr. Thomas Kunkel (National Institute of Environmental Health Sciences, Research Triangle Park, NC, USA).
Western blotting
Western blotting was performed using standard techniques with anti-MLH1, anti-PMS2, or anti-MSH2 mouse monoclonal antibodies at 2 µg/ml (BD Pharmingen, Oxford, UK), or with anti-MSH6 rabbit polyclonal antibody (pAb) at 1:3000 (a kind gift from Prof. Jiricny, Zurich) or anti-human GR rabbit pAb (E-20, Santa Cruz Laboratories, Santa Cruz, CA, USA). Each blot was also probed using either polyclonal antisera against human actin (Oncogene, CN Biosciences, Nottingham, UK) or human glyceraldehyde phosphate dehydrogenase (GAPDH, Abcam, Cambridge, UK) as a protein-loading control. Secondary antibodies were horseradish peroxidase conjugates, and immune complexes were visualized using an enhanced chemiluminescence (ECL) reagent, ECL, or ECL-Plus (Amersham Pharmacia Biotech) according to the manufacturer’s instructions.
Fluorescent in situ hybridization
Subconfluent, actively proliferating cell cultures were incubated with colcemid for 2–3 h without any pretreatment for synchronization. Hypotonic treatment, fixation, and slide making were performed using standard protocols (38)
. Fluorescence in situ hybridization (FISH) detection of complete or partial copy number changes of chromosome 5 was performed with the use of the direct-labeled, dual-color LSI EGR1 (5q31)/D5S23,D5S721 (5p15.2) probe (Vysis, Downers Grove, IL, USA) by following the manufacturer’s instructions. At least 100 interphase nuclei were scored on a Zeiss Axioplan 2 microscope using appropriate single and dual band-pass filter sets to detect the amount of cells with chromosome 5 gains or losses.
RESULTS
Two in vitro models for GC resistance in ALL
To systematically study development of GC resistance in ALL, two GC resistant models were employed. The first was generated from CCRF-CEM-C7H2 childhood T-ALL cells (26)
; the other one, using the B-cell precursor-ALL cell line PreB697 (this study) by selection culture in GC containing media and resulted in 41 GC-resistant CEM-C7H2 and 30 GC-resistant PreB697 descendents. For comparison, 13 GC-sensitive CEM-C7H2 descendents were generated by limiting dilution in the absence of GC. Due to lack of clonability, no GC-sensitive PreB697 subclones could be derived, thus the parental PreB697 cell line served as control. To confirm their GC-resistant phenotype, cells were treated with 0.1 µM dexamethasone or vehicle control and assessed for the induction of apoptosis at 72 h by flow cytometric evaluation following PI staining. Data for all cell lines are summarized in Fig. 1
. The percentage of apoptotic cells induced by GC in sensitive CEM-C7H2 subclones was 74.7% ± 10.8 in response to dexamethasone compared to 6.2% ± 5.4 in GC resistant subclones (mean±SD). Vehicle control values were 12.2% ± 5.7 and 6.8% ± 5.8 for sensitive and resistant subclones, respectively. Similarly, the percentage of apoptotic cells for the PreB697 GC resistant subclones was 16.0% ± 13.7, compared with 67.1% ± 13.2 for the parental, GC sensitive cell line, with vehicle control values of 4.3% ± 3.0 and 1.7% ± 0.4, respectively. Thus, the data verified the GC-resistant phenotype of all subclones although the preB697 descendants were somewhat less resistant than the corresponding CEM-C7H2 derivatives.
|
Mismatch repair status
Previous studies have shown that CCRF-CEM cells are MMR-deficient (22
, 23)
. Since this is critical information for a resistance model, we determined the MMR status of the PreB697 parental cell lines, using an in vitro functional assay, based on the ability of cellular protein lysates to repair heteroduplex substrates containing GT mismatches. This assay has been well validated by several laboratories, including our own (23
, 24)
, and was also used to reveal the MMR-deficient status of the CCRF-CEM parent cell line. This deficiency was shown to be attributable to a splice site mutation in the MLH1 gene, leading to loss of MLH1 expression, and functionality could be restored by the addition of exogenous MLH1 protein (23)
. In the present study, in vitro analysis of G.T MMR showed that PreB697 cells were fully proficient, showing 83% of repair compared to 7% for CCRF-CEM (Fig. 2
a). In confirmation of this, PreB697 were shown to express a full complement of MMR proteins, including MLH1, MSH2, MSH6, MSH3, and PMS2 as judged by Western blotting, contrasting with the absent MLH1 and its heterodimeric partner PMS2 in the CCRF-CEM parent line (Fig. 2b
). Thus, our models enable assessment of GC resistance mechanisms in ALL in a MMR-proficient and -deficient background.
|
GC resistance is associated with reduced induction of GR and its downstream target GILZ
As reviewed recently (7
, 10
, 11
, 39)
, there are numerous possibilities how a cell might acquire GC resistance, including qualitative or quantitative defects at the GR level, increased elimination or metabolism of GC, or interference with downstream components of the GC-induced death signaling pathway. We initiated our search by scrutinizing the most apical component of the death pathway, i.e., the GR. In particular, we assessed basal GR expression levels, the functionality of the GR by quantifying its downstream target GILZ, and finally GR autoinduction, a phenomenon that is critical for GC-induced apoptosis in CCRF-CEM (40
41
42)
and perhaps other systems (33
, 43)
. Thus, quantitative analyses of GR and GILZ mRNA expression were assessed by quantitative real-time reverse-transcriptase PCR (RT-PCR) in both cell line systems using TBP as reference gene for normalization (Fig. 3
). Levels of expression were measured under basal conditions and after 6 and 24 h exposure to 0.1 µM dexamethasone or vehicle control. In both GC-sensitive and the GC-resistant subclones, basal levels of expression of GR and GILZ were similar. However, after 6 h of GC exposure, the GC-sensitive CEM subclones, as well as the GC-sensitive parental PreB697 line, showed an increased expression of both GR and GILZ when compared with their corresponding resistant subclones. At 24 h, GR and GILZ expression increased further and the difference between GC-sensitive and -resistant groups was even more pronounced with no overlap. These findings, namely an increase in sensitive and no increase—or a marginal one—in GC-resistant ALL cells, could be further confirmed at the protein level by Western blotting (Fig. 3B
). In both, the T- and the B-ALL model, GC-resistance—although indistinguishable by GR expression levels of untreated samples—is associated with a deficient GR up-regulation on GC exposure.
|
Transgenic GR restores GC sensitivity
To test whether the downstream components of the GC-induced cell death pathway are intact, we transduced CEM-C7H2 subclones R11G5 and R5C3 with a retroviral construct expressing human WT GR. As shown in the Western blots in Fig. 4
, transgenic GR was expressed at levels resembling those seen after 24 h GC treatment in the GC-sensitive CEM-C7H2 cell line and restored GC sensitivity, i.e., the transfected cells underwent typical GC-induced cell death. These data strongly suggest that the downstream components of the GC-induced death pathway are intact in these cells and that apoptosis resistance occurred at the level of the GR.
|
GC-resistant CCRF-CEM, but not preB697, show high frequency of GR mutations
To address whether deficiency in GR autoinduction might be caused by mutations in the GR gene, mutational screening of GR exons encoding the DNA binding and ligand binding domains was performed in all GC-resistant PreB697 and CEM-C7H2 GC-resistant and GC-sensitive subclones. DHPLC revealed a number of altered chromatogram patterns indicative of DNA alterations. By DNA sequencing, a number of novel insertions, deletions, and point mutations were identified (see
Fig. 8
in the discussion). For the PreB697-resistant clones, 3 from 30 were found to have acquired a GR mutation under GC selection: Two clones (R4B11 and R3G3) having the same IVSC-2 A > G splice site mutation, preceding exon 4 and clone R3E3, an insertion of C in exon 9 (2447 insC) that leads to a frameshift and replacement of the last 6 amino acids of the GR with 26 novel residues. Chromatograms of GR exon 5 showed a heteroduplex pattern in all PreB697 subclones, except clones R3C11 and R4C3. DNA sequencing showed that the heteroduplex pattern was due to the presence of a common intronic polymorphism, IVSD-16 G 
T and that loss of the heteroduplex pattern in clones R3C11 and R4C3 was due to LOH at the GR locus.
|
|
|
|
For the CEM-C7H2, all sublines derived from the parental line bear the L753F mutation. No additional mutations were found in the 11 GC-sensitive subclones screened. However, 11 mutations were detected in 10 of 37 G-resistant clones. They included two nonsense mutations (G506X, Q713X); two frameshift mutations, i.e., A1615, which leads to a premature stop codon, and 2446insA, which creates a novel C terminus similar to 2447ins C above; one donor splice site mutation preceding exon 5 and 6 missense mutations, one in the DNA binding domain (A484D), one in the hinge region (P515H); and the others in the ligand-binding domain (L756N, Y663H, L680P, R714W). GC-resistant cell line R19E5 acquired both the P515H and L680P mutations. Again, chromatograms of exon 9 invariably showed heteroduplexes, but profiles attributable to the sole presence of the L753F differed from those bearing additional mutations in this exon (see Fig. 5
). Subclone R29C7 gave a homoduplex pattern for exon 9 and after sequencing was shown to be due to complete LOH at this locus. Cloning and sequencing of exon 9 amplicons from R28G4 revealed either the 753F or the 756N mutation in 7 of 7 bacterial clones. Hence, the 756N mutation had occurred on the WT allele. Similarly, 8 of 9 exon 9 clones from R30F4 showed either the 753F mutation or an insertion of A at 2488, however, 1 clone was found to be WT. Given that the DNA fingerprint profile of subclone R30F4 indicated loss of one WT allele, the 2488insA mutation had to have occurred on the remaining one. Any explanation for the occurrence of all three observed alleles requires several mutational events, on the contrary a polymerase error during PCR amplification at this position consisting of a mononucleotide string of As could account as well for this single incidence. Thus GC resistance due to GR mutations occurs in MMR proficient and deficient leukemic cell lines but is more common in the latter. Nevertheless, the majority of GC-resistant subclones had no detectable GR gene mutation and almost all of them had a remaining WT allele raising the question as to how the deficient GR autoinduction might be molecularly explained. A partial answer to this question came from DNA fingerprinting analyses that were originally designed to further address the MMR status of our cell lines by analyzing micro satellite stability.
GC-resistant CCRF-CEM, but not preB697, show microsatellite instability and a high frequency of GR allele loss
Using DNA fingerprinting at 16 loci, we have previously reported that CEM-C7H2 cells show a high degree of microsatellite instability, a phenotype closely associated with MMR deficiency (26)
. While both GC-sensitive and GC-resistant subclones revealed this phenotype to a similar extent, a remarkable difference was observed in the D5S818 locus that maps 20 Mbp centromeric to the GR locus on chromosome 5. For this locus, 2 alleles (D5S818–12 and -13) were detected in the parental CEM-C7H2 line at equal ratio. However, 37 of 41 resistant subclones showed 50% reduction of one of the 2 alleles (D5S818–13). In contrast, only 2 of 17 of the sensitive subclones revealed this phenomenon. This finding suggested that one of the GR alleles was selectively lost during culture in the presence of dexamethasone. To further scrutinize whether this DNA fingerprint pattern indeed indicated loss of one GR allele, we performed FISH analysis on CEM-C7H2 and one subclone, both of which had equal ratios of D5S818–12 and -13, and on three other clones with altered ratios by using a green probe for the short arm (D5S23, D5S721) and a red probe for the long arm (EGFR-1) of chromosome 5. As depicted in Fig. 6
, the two cell lines (CEM-C7H2, C7H2-R9E9) with equal ratio of both fingerprinting markers showed four copies of both 5p and 5q, whereas the three subclones with a reduction of one of these alleles (R28G4, R21D3, R1E5) showed 4 copies of 5p, but only three copies of 5q. Thus, FISH—although of course not differentiating between WT and mutated allele—confirmed that the DNA fingerprinting pattern at the D5S818 STR reflected GR copy number in all 5 instances tested. To assess whether GC-resistant preB697 subclones might show similar evidence for GR allele loss, we performed DNA fingerprinting on all 30 GC-resistant subclones. Again, heterozygosity was apparent at the D5S818 locus, with the presence of D5S818–11 and D5S818–13 alleles. However, in contrast to GC-resistant CEM-C7H2 subclones, peak height ratios were similar in all subclones, bar one (R3C11), which showed a reduction in the peak height of the D5S818–13 allele, indicative of LOH. Moreover, PreB697 and its GC-resistant subclones showed microsatellite stability at all other loci, consistent with MMR proficiency.
While 11/37 resistant CEM-C7H2 subclones were shown to be affected by both mutations of the GR and loss of one allele, thereby leaving the cell with no functional GR at all, none of the resistant PreB subclones revealed complete loss of functional GR. Moreover, LOH only occurred in 29/37 resistant CEM-C7H2 subclones compared to 2/30 instances in the resistant PreB 697 lines. Similarly, 3/30 PreB 697 lines carried GR mutations of one allele, resulting in only one WT GR allele. Thus, while in most PreB 697 subclones (25/30) neither mutations nor LOH could account for the deficient GR up-regulation, the same applied only to 1/41 resistant CEM-C7H2 subclones. Hence, although deficiency in GR autoinduction was a conserved feature in all GC-resistant subclones from both ALL models, different molecular mechanisms resulted in this phenomenon (see Fig. 7
).
DISCUSSION
In this study we present two models of resistance to GC-induced apoptosis in childhood ALL. Although resistance may arise at numerous levels in this death pathway via a plethora of molecular mechanisms, in the two models investigated in this study, GC resistance appeared to occur exclusively at the level of the GR. In the CCRF-CEM model, this might not be surprising because of a loss of function mutation (21)
in 2 of the 4 GR alleles in this tetraploid cell thus predisposing it to mechanisms acting at the GR level. In addition, the MMR deficiency of CCRF-CEMs (22)
facilitates point mutations, further explaining that 1/3 of the investigated GC-resistant subclones had acquired an additional GR gene mutation (for a discussion of these mutations see below). While the more frequent mechanism of GR loss, i.e., large deletions in, or complete loss of, the long arm of chromosome 5, is not typically associated with a MMR-deficient phenotype, recent evidence suggests that it may be associated with increased chromosomal instability (44)
. Because the two events in combination leave the cells without functional GR, they fully explain GC resistance. In the remaining 2/3 of the cells, haploinsufficiency might account for GC resistance and its associated loss of GR autoinduction. Haploinsufficiency on its own is, however, not very likely because one of the GC-sensitive subclones (C7H2-S11) shows fingerprinting evidence for loss of one functional GR allele, yet is still able to autoinduce the GR and drive cells into apoptosis. More likely, there may be an as-yet-undefined mechanism that prevents GR autoinduction and thus GC-induced apoptosis. This, or a similar, mechanism might also be operative in GC-resistant PreB697 subclones and precipitate GC resistance. In this model, the majority of subclones (25/30) might entirely rely on this mechanism to escape the lethal GC effect while the remainder (5/30) may use it in combination with structural defects (mutations, deletions) in the GR gene.
GR mutations
Altogether this study has identified 13 novel mutations in the GR, which are associated with acquired GC resistance in ALL and together with previous studies make a total of 21 (Fig. 8
). They include splice site, frameshift, and nonsense and missense mutations and are predominantly found in the ligand binding domain. Mutation of the arginine residue at position 732 in the rat, which corresponds to R714 in human GR, was associated with a loss of ligand-binding activity; but in the rat study, the amino acid substitution was glutamine and in this study was tryptophan (45)
. In addition, a PCR expression mutagenesis study showed that mutation of the same residue resulted in a reduced affinity for coactivators (46)
. While no functional data exist regarding the other missense mutations detected in this study, i.e., L680P, A484D, Y663H, and P515H, the former three residues are highly conserved among species, while the P515H, residing in the hinge region, is less so. Mutations in the hinge region have not previously been associated with GC resistance; however recent evidence suggests that sequence alterations of this region can influence receptor activity (47)
. Interestingly, the 2 GR frameshift mutations acquired in the CEM-C7H2 GC-resistant clones were found in two separate, coding, mononucleotide repeats. Instability of these error prone microsatellite regions is a phenotype closely associated with MMR deficiency. In contrast, the frameshift mutation found in one of the PreB 697 clones was just outside the mononucleotide region, but the translational consequence is similar, i.e., the formation of GR protein with an aberrant C terminal end.
Possible relevance for the in vivo situation
Relative few studies have addressed mechanisms of GC resistance in clinical samples. Acquired somatic mutations of the GR have been detected in clinical samples of children with ALL; one a point mutation (L753F) and the other a 29 bp deletion (702), both mutations affecting the ligand-binding domain (20
, 36)
. However, in a large mutational screen of samples taken at ALL relapse, after prolonged combination therapy containing GC, only 1 from 50 patients had acquired a GR mutation (36)
. Similarly, no somatic mutations were identified in diagnostic ALL samples in which prednisone resistance was assessed both in vitro and in vivo (48)
. Thus acquired somatic mutations of GR are not a predominant mechanism of GC resistance in clinical samples.
Whether or not deficiency in GR autoinduction plays a role in GC resistance in clinical samples is, however, less clear. In a primary childhood ALL-NOD/SCID xenograft model, Bachmann et al. (49)
showed that GC resistance was not associated with alterations in GR expression levels or nuclear translocation of the GR. Moreover, in four of four investigated xenografts (two GC-resistant and two GC-sensitive) dexamethasone treatment caused GR down-regulation, suggesting that GR autoinduction does not occur in GC-sensitive cells and hence its loss cannot account for resistance. This observation, however, contrasts with a study by Tissing et al. (50)
, which addresses GR regulation in GC-sensitive and -resistant primary leukemic blasts treated with GC in vitro. In this study every single in vitro GC exposed sample of 10 GC-resistant and 12 GC-sensitive patients was found to up-regulate the GR. These contradictory results leave unresolved which of these models, if any, allows extending conclusions to the clinical situation. At present, there is only one report where GC response was analyzed in vivo (i.e., in ALL children undergoing GC-monotherapy) (51)
, and this study showed neither consistent GR up-regulation nor receptor down-regulation. Instead GR induction was detected in all 3 T-ALLs and 2 of 10 preB-ALLs samples, while no regulation was observed in the others. Thus, given their controversial outcome and their poor concordance with the observed in vivo GC response, it seems questionable whether these studies allow final conclusions on the relevance of GR regulation in clinical GC resistance.
In conclusion, we have shown that mutations in the GR gene and, perhaps more importantly, deficiency in GR autoinduction appear to account for GC resistance in two independent ALL models. While mutations in patients are rare, GR autoinduction occurs in a number of children with ALL during GC monotherapy (52)
. Whether GR up-regulation defects might lead to GC resistance in such patients needs to be evaluated in samples from GC-sensitive and -resistant patients treated in vivo.
ACKNOWLEDGMENTS
We thank Dr. S. Geley for stimulating discussions, Dr. Ernst Werner for valuable advice in setting up real-time PCR profiling and S. Jesacher for technical assistance. Supported by grants from the Austrian Science Fund (SFB-F021, P18571), the Austrian Ministry for Education, Science and Culture (GENAU-CHILD), and the United Kingdom Leukemia Research Fund. The Tyrolean Cancer Research Institute is supported by the "Tiroler Landeskrankenanstalten Ges.m.b.H. (TILAK)," the "Tyrolean Cancer Society," various businesses, financial institutions and the People of Tyrol.
FOOTNOTES
1 These authors contributed equally to this work. 
Received for publication April 28, 2006. Accepted for publication July 11, 2006.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
