Introduction

Glucocorticoids (GCs) are the cornerstone in the treatment of childhood leukemia. Especially, the initial acute lymphoblastic leukemias (iALL) are uniquely sensitive to the lytic actions of GCs.^1,2,3,4,5 However, in a significant proportion of these patients GC resistance is present or develops, which has an adverse effect on cure rates. In addition, acute myeloid leukemia (AML) and relapsed ALL (rALL) patients are highly resistant to GC-induced cell kill.^6,7

The cell lytic effect of GCs is mediated through the glucocorticoid receptor (GR). The amino-acid sequence of the GR was first described by Hollenberg et al⁸ in 1985. They identified two human GR isoforms termed GC receptor alpha (GR) and beta (GR). The analysis of the protein structure revealed that both receptor isoforms share the same amino-acid sequence up to amino acid 727, but diverge from this position with GR having an additional 50 amino acids and GR an additional nonhomologous 15 amino acids. Exons 1–8 of the GR gene contain the 5' noncoding and coding sequences common to the GR and GR. Alternative splicing of the large exon 9 generates the mRNA coding for the GR and GR.

The GR has been the primary focus of subsequent research, due to its predominant expression, ligand-binding properties and transcriptional activity. GR does not bind GCs and is transcriptionally inactive.⁸ The GR is a 97 Kd a protein that is organized into structurally and functionally defined domains: the highly variable region that includes a transactivation domain that is required for the regulation of gene expression, the DNA-binding domain crucial for the specific interaction between the receptor and the GC-responsive elements and the carboxyl-terminus involved in hormone binding, interaction with heat-shock proteins (hsp's), nuclear translocation, receptor dimerization and transactivation.^{9,10,11,12,13,14,15,16,17}

In the absence of hormone, the GR predominantly resides in the cytoplasm as a multiprotein complex consisting of the GR itself, two molecules of hsp90 and several other proteins.^18,19,20 Upon binding of a GC, the receptor complex dissociates and translocates to the nucleus. In the nucleus, the GR binds as a homodimer to GC-responsive elements, which are usually found in the promotor regions of GC-responsive genes.^21,22 The receptor then communicates with the basal transcription machinery resulting in enhancement or repression of gene transcription.²³ In addition, the receptor modulates gene transcription by physically interacting with other transcription factors like AP-1 and NF-B.^24,25

Many studies have been published describing the relation between clinical response and GR expression in childhood lymphoblastic leukemia, but receptor expression levels appeared to be of limited clinical usefulness.^26,27,28,29 These observations have led many researchers to believe that clinical resistance to GCs should be sought further downstream in the GC lytic pathway.

During the past decade, the GR has been increasingly studied. Oakley et al³⁰ demonstrated that GR, despite the fact that it is transcriptionally inactive, influences GC-mediated gene transcription. They showed that the GR was capable of binding and inactivating GR, thus inhibiting signal transduction. However, this was not always confirmed.³¹ Most clinical evidence that GR is capable of decreasing GC sensitivity of target cells has been obtained in respiratory medicine. Various authors showed that GR expression in airway cells and peripheral blood mononuclear cells obtained from patients with GC-resistant asthma was higher than in patients with GC-sensitive asthma or normal subjects.^32,33 So far, only one study has been published investigating GR and GR levels in pediatric leukemia samples.³⁴ Longui and co-workers found significantly lower GR/GR ratios in T-ALL (a subgroup relatively resistant to GCs) than in common/pre-B ALL. However, the number of patients studied was small (nine controls, 13 ALL patients). In addition, correlations with in vivo or in vitro treatment response were not studied.

More recently, a third splice variant of the GR was identified: the GR.³⁵ Owing to the use of an alternative splice donor site, three base pairs are retained between exons 3 and 4. As a consequence, an mRNA sequence is produced coding for an additional amino-acid (arginine) located in the region between the two zinc-fingers of the DNA-binding domain. Although steroid receptors show considerable conservation of amino-acid sequences in their DNA-binding domain, consistent with the functional importance of this region, amino-acid insertions at the same site in the GR have been described. Ray and co-workers described a small-cell carcinoma cell line that expressed a GR with a similar arginine insertion into the DNA-binding domain, although they assumed it to be the result of a mutation.³⁵ Interestingly, they suggested that the GC-resistant phenotype of this cell line might be explained by this amino-acid insertion as transfection studies demonstrated a strong reduction in the transcriptional activity of this receptor. The clinical significance of GR expression, however, is unknown.

The goal of the present study was to investigate the correlation between GR/GR/GR expression and in vitro GC resistance in childhood leukemia. This way, we tried to find out whether alternative splicing of the GR contributes to the differential GC-resistance profiles observed in childhood leukemia.

Materials and methods

Patient samples

Bone marrow or peripheral blood samples taken for routine diagnostic procedures were obtained from 43 iALL, 10 iAML, 11 (unpaired) rALL and one rAML patient with informed consent. Mononuclear samples were separated by sucrose density-gradient centrifugation (Ficoll Paque, density 1.077 g/ml; Pharmacia, Sweden). The percentage of leukemic cells was determined morphologically by May–Grünwald–Giemsa (Merck, Germany) staining of cytospin preparations. When necessary, the percentage of malignant cells was enriched to >80%, using monoclonal antibodies linked to magnetic beads (Dynabeads, Norway) to eliminate contaminating cells.³⁶

Immunophenotype was defined as precursor B-lineage (HLA-DR⁺/terminal deoxynucleotidyl transferase (TdT)⁺/CD19⁺) and T-lineage (TdT⁺/cytoplasmatic CD3⁺/CD7⁺). Precursor B lineage was further subdivided into pro-B (CD10^-, c^-, sIg^-), common (CD10⁺, c^-) and pre-B (CD10⁺ or CD10^-, c⁺, sIg^-). No B-ALL samples (CD10^-, c^-, sIg⁺) were included in this study.

RNA extraction and reverse transcriptase reaction

Total RNA of 5 10⁶ cells was isolated using RNAzol (Campro Scientific, Veenendaal, The Netherlands) according to the manufacturer's protocol, with minor modifications: the chloroform–phenol extraction step was repeated and the precipitation step in isopropanol at -20°C was extended to overnight.

Next, total RNA was reverse transcribed by M-MLV-RT (Murine Leukemia Virus Reverse Transcriptase; Promega, Madison, WI, USA), including RNase inhibitor (0.6 IU/l), bovine serum albumin (0.2g/l), dNTPs (1 mM) and random hexamers (45 ng/l) in 50 l at 42°C for 2 h. After incubation, the mixture was heated to 95°C to denature M-MLV-RT and stored at –20°C.

Preparation of competitive templates standard mixtures

Primers to synthesize competitive templates (CTs) (F and R1R2) and for the amplification of the native target (NT) (F and R1) were selected by Oligo software as described previously.^37,38 The selection of the primers was based on an optimal annealing temperature of 58°C, the absence of hairpins and no tendency for primer–dimer formation. Selected primers are summarized in Table 1.

Table 1 - Primers used for the construction of CTs and for coamplification of the NT and the corresponding template.

Full table

CTs were synthesized by the amplification of a cDNA sample obtained from a CEM cell line, which expressed both GR and GR mRNA. In this PCR reaction, similar forward primers were used as for the amplification of the NT, but different (hybrid) reverse primers (R1R2). These hybrid primers consist of the reverse primers used for the amplification of the NT (R1) and a primer situated more upstream (R2). The thus obtained PCR products, approximately 75% the length of the NT, were pooled, separated on an agarose gel, gel purified and quantitated by coelectrophoresis with a size marker, as described previously.^37,38 Coamplification using primer F and R1 of a cDNA sample with known amounts of the CTs will result in two PCR products of different lengths, which can be separated by gel electrophoresis and quantified by densitometry after ethidium bromide staining.

CTs were mixed in combinations that were based on pilot experiments using 10 patients samples. For this study, several stocks were prepared containing a mixture of CTs for -actin, GR and GR in fixed ratios: (a) 10^-10/10^-13/10^-15, (b) 10^-10/10^-14/10^-16, and (c) 10^-10/10^-15/10^-17. These mixtures could subsequently be diluted further without disturbing the initial relationships of the CTs present.

Competitive template polymerase chain reaction

In order to normalize the expression of the target genes to the expression of -actin and in order to allow accurate determinations of GR/GR ratios, one single master mix was prepared for every cDNA sample containing PCR buffer (1 ), MgCl₂, dNTPs (200 M), sample cDNA and the appropriate CT mix in a total volume of 41 l. Primers (4 l, final concentration 4 ng/l) specific for the different targets (-actin, GR, GR) were added to separate aliquots of this master mixture. Reaction mixtures were overlaid with mineral oil (Sigma, St Louis, MO, USA) and cycled in an MJ Research PTC-200 (Biozym, Landgraaf, The Netherlands) with 1 min steps of denaturation at 94°C, primer annealing at 58°C and elongation at 72°C for 35 cycles. The first cycle included a 7 min denaturation step, whereas in the last cycle, the elongation period was extended to 7 min. In case of the GR amplification, the PCR was preceded by 17 annealings/elongation cycles during which the annealing temperature was decreased stepwise from 70°C with 0.5°C/cycle (touch down protocol) in order to improve sensitivity and specificity.

PCR products were separated by 130 V electrophoresis for 3 h on 2% agarose gels containing 0.1 mg/ml ethidium bromide. The results were analyzed by densitometry. Concentrations of NT molecules can be calculated using the NT/CT ratio and the molarity of the CT mixture. Pixel intensity (PI) staining has to be corrected for fragment size, as the intensity of PI staining is linearly related to the number of base pairs (bp). In addition, heterodimers (HD) consisting of the NT and CT molecules were taken into account (Table 2). The results were evaluable when the ratio NT/CT was between 0.1 and 10 for GR and -actin and between 0.1 and 6 for GR. All PCRs were performed in triplicate. The mean variation between triplicate experiments was 26% (13%) for GR normalized to -actin and 35% (18%) for GR normalized to -actin.

Table 2 - Quantitation of mRNA expression.

Full table

Quantitation of GR mRNA

The cDNA region spanning the exon 3/exon 4 boundary were: 5'-GGA GAT CTG GTT TTG TCA AG-3' (sense) and 5'-GAA GAC ATT TTC GAT AGC GG-3' (antisense). The conditions used for PCR were: 35 cycles of 94°C 15 s, 54°C for 39 s, 72°C for 40 s. PCR products were purified and digested with ACC I prior to analysis during 1 h at 37°C. In case the three base pair insertion is present, an ACC I restriction site is created (GTAGAC) and consequently the 670 bp product is cleaved into 573 and 97 bp products. Next, PCR products were separated by 100 V electrophoresis for 3 h on 2% agarose gels containing 0.1 mg/ml ethidium bromide. In all the samples tested, the 573 and 670 bp products could be visualized. The 97 bp product could only be visualized after increasing the amount of amplified cDNA loaded on gel (PI staining is linearly related to the number of bp and signal intensity was thus low). The results were analyzed by densitometry. The expression of GR was expressed as a percentage of the total GR message as follows:

All experiments were performed in triplicate. The mean variation between triplicate experiments was 28% (14%).

GR and GR protein detection

Cell lysates for GR detection were prepared as follows: a 1 ml cell suspension of 2–10 10⁶ cells/ml was washed three times in PBS. After the third wash, the pellet was resuspended and lysed in 80 l lysis buffer/1 10⁶ cells (5 M NaCl, 0.5 M Tris-HCl, 0.4% sodium dodecyl sulfate (SDS), 1% Triton-X, one tablet Complete™ (Boehringer Mannheim, protease inhibitor cocktail tablets: 1697498)/50 ml lysis buffer) during 1 h at 4°C. Lysates were clarified by microcentrifugation at 14 000 g during 10 min. Next, Laemli's sample buffer (Biorad 161-073) supplemented with 50 l of -mercaptoethanol/1 ml was added (lysis: Laemli=2:1) and lysates were boiled for 3 min.

Cell lysates for GR were prepared as follows: a washed pellet of 10 10⁶ cells was lysed in 100 l lysis buffer as described above. After clearing the lysate of cellular debris by centrifugation at 14 000 g for 10 min, lysates were diluted with 900 l RIPA buffer (1xPBS, 0.1% SDS, 1% Igepal CA-630 (Sigma Chemicals), 0.5% sodium deoxycholate and one tablet of Complete™ (Boehringer Mannheim, protease inhibitor cocktail tablets: 1697498)/50 ml lysis buffer. Next, 0.6 g of anti-GR/ (Santa-Cruz; SC-1003) was added and incubated for 1 h at 4°C. Next, 20 l of resuspended agarose conjugate (Protein A–Agarose, Santa-Cruz: SC-2001) was added and incubated at 4°C overnight. Immunoprecipitates were collected by centrifugation at 1000 g for 5 min at 4°C. The pellet was washed 4x in RIPA buffer and after the final wash the pellet was resuspended in 20 l lysis buffer +10 l Laemli's sample buffer and boiled for 3 min.

Cell lysates (20 l: the equivalent of 2.5 10⁵ cells for whole-cell lysates or 7 10⁶ cells for immunoprecipitates) were separated by SDS-polyacrylamide gel electrophoresis (PAGE) (7.5–12.5%) and electroblotted (30 V, 12 h) onto nitrocellulose membranes (Biorad: 162-0115). Next membranes were washed 1 during 15 min in PBS, incubated for 1 h in blocking buffer (0.1% Tween-20, 5% nonfat dry milk blocking grade (Biorad) in PBS). Blots were incubated with commercially available antibodies against the GR (rabbit polyclonal, Santa-Cruz: SC-1002), GR (Affinity Bioreagents, PA3-514) and -actin (mouse monoclonal (clone c4), Boehringer Mannheim) for 1 h at room temperature in 1% blocking buffer (0.1% Tween-20, 1% nonfat dry milk (Biorad: 170-6404) in PBS). After four washes in PBS-T (0.1% Tween-20 in PBS), membranes were incubated with goat anti-rabbit-HRP (Santa-Cruz: SC-2004) or goat anti-mouse-HRP (DAKO: p0447) during 1 h. After four washes in PBS-T, peroxidase activity was visualized using an ECL kit (Amersham pharmacia biotech: RPN 2106). GR signal intensity was determined densitometrically and expressed as compared to the internal standard (an ALL patient sample included in all experiments whose values were arbitrarily set at 1). -actin was used for normalization. All experiments were performed in duplicate. The mean variation between duplicate experiments was 24% (12%). GR protein could only be visualized in the positive control, an immunoprecipitate of 10 10⁶ CEM-C7 cells.

Ligand binding (LB) assay

Specific ³H-DXM binding was determined using the whole-cell assay as described by Costlow et al³⁹ with minor modifications.^40.Briefly, cell suspensions of 1–4 10⁶/ml were incubated with a receptor saturating concentration (50 nM) of ³H-DXM (Amersham life sciences, TRK417; 41 10³ Ci/mol) in the presence (low specific activity) and absence (high specific activity) of a 500-fold excess of nonradioactive DXM during 45 min at room temperature in duplicate. Next, the cell suspensions were centrifuged and radioactive media were replaced by steroid-free media in order to reduce nonspecific binding. After 10 min incubation, the cell suspensions were centrifuged by three spins of 3 s each using an Eppendorf centrifuge. The cells were washed three times in ice-cold PBS containing 1% fetal calf serum. Before the last centrifugation step, a small sample was taken for cell counting. Cell number and volume were determined using an electronic cell counter (Casy^® 1 cellcounter, Schärfe Systems). Cells were then lysed in 0.1 N NaOH and the amount of radioactivity was determined by scintillation counting. Specific ³H-DXM binding was calculated as described previously.⁴⁰ By including the cell volume in the equation, the intracellular receptor concentration could be calculated (nM). The mean variation between duplicate experiments was 10% (4.5%).

Methyl-thiazol-tetrazolium salt assay

The assay conditions were essentially the same as described previously.^36,41 Briefly, aliquots of 80 l cell suspension (1-2 10⁶ cells/ml) were added to 96-well microculture plates containing 20 l aliquots of drug solutions. Leukemic cells were exposed to six concentrations of prednisolone (PRD) in duplicate for 4 days. Control leukemic cells were cultured in the absence of drugs. After 4 days of culturing, we added 10 l of 5 mg/ml MTT (Sigma) to each well. The microculture plates were shaken gently for 1 min and incubated for 6 h. The yellow tetrazolium salt MTT is reduced to dark-colored formazan by viable cells only. Formazan crystals were dissolved in 100 l acidified isopropanol. The optical density (OD) was measured at 565 nM with an EL-312 microplate reader (Biotek Instruments Inc., USA). The OD is linearly related to the number of viable cells.⁴¹ After correction for the OD of the culture medium, leukemic cell survival (LCS) was calculated as follows:

Only samples with more than 70% blasts at day 4 in the control wells and with a control OD higher than 0.050 U (adjusted for blank values), both required for reliable test results,³⁶ were used to calculate the LC₅₀ value, that is, the PRD concentration lethal to 50% of the blasts. Patients were classified as sensitive (LC₅₀ PRD 0.1 g/ml), intermediately resistant (LC₅₀ PRD<150g/ml, but >0.1 g/ml) or resistant (LC₅₀ PRD 150 g/ml) to GC-induced cell kill. This classification has been shown to correlate with clinical response in childhood Iall.^2,3

Statistics

The correlation between two variables was tested by the Spearman's rank-correlation test. The Mann–Whitney U-test was used for the analysis of unpaired observations, the Wilcoxon's signed-rank test for paired observations. A P-value 0.05 was considered to be statistically significant (two-tailed analysis).

Results

Validation of CT-RT-PCR

Coamplification of the CT and NT (ie GR/) showed that both fragments were amplified with equal efficiency; the ratio NT/CT was similar before and after amplification. The slopes obtained by plotting the ratio of CT/NT initially present (varying from 0.1 to 10 for GR and from 0.1 to 6 for GR) vs the ratio obtained after amplification was 1.1 for GR and 1.0 for GR. The mean fit coefficients (r²) for GR and GR were, respectively, 0.98 and 0.97. Sequence analysis confirmed that the amplified fragments belonged to GR and GR.

GR mRNA expression vs protein expression

GR expression was determined at the mRNA level competitive template reverse transcriptase polymorase chain reaction (CT-RT-PCR) and at the protein level (Western blotting (WB) and (LB) assay). Receptor expression determined by WB and LB correlated strongly with each other. This was observed in each leukemia subgroup separately: iALL =0.77, P<0.0001, n=35; rALL =0.85, P=0.004, n=9; iAML =0.94, P=0.005, n=6), and also after combining all leukemia samples (including one rAML sample; =0.73, P<0.0001, n=51).

GR mRNA expression correlated with protein expression determined by LB (=0.50, P=0.0001, n=56; Figure 1), and by WB (=0.39, P=0.003, n=54) in the study population as a whole. In each subgroup separately, only in iALL samples a statistically significant correlation between mRNA and protein expression was found: =0.56, P=0.0005, n=38 when determined by LB, =0.39, P=0.02, n=35 when determined by WB. The group of rALL and iAML patients, however, was small.

Figure 1.

Correlation between GR mRNA and protein expression as determined by LB assay. The former was normalized with -actin mRNA. The latter was expressed in nanomolars (nanomol GR/l cell volume)

Full figure and legend (22K)

GR// expression in childhood leukemia

The results of GR// expression in the different leukemia subtypes are summarized in Table 3. GR mRNA was median 137x lower than GR mRNA (range 26–1805; n=63) and correlated with GR expression (Figure 2). At the protein level, GR could be visualized in an immunoprecipitate of CEM-C7, but not in any of the patient samples (Figure 3). The 3 bp insertion typical for GR occurred in median 2.8% of all GR mRNA (range 0.95–7.4%; n=47).

Figure 2.

Correlation between GR and GR mRNA expression in 63 leukemia patients as determined by CT-RT-PCR. GR and GR mRNA levels are expressed as compared to -actin mRNA levels.

Full figure and legend (20K)

Figure 3.

Western blot analysis of GR expression in childhood ALL (lane 3–6) and AML samples (lane 7–10). As a positive control an immunoprecipitate of CEM-C7 was used (lanes 1 and 2). GR was not detectable in any of the patient samples. The light (b) and heavy (a) chains of SC-1003 (anti-GR/ IgG) are also visualized.

Full figure and legend (48K)

Table 3 - GR// expression in the different leukemia subtypes.

Full table

Within iALL samples, T-ALL (n=11) expressed significantly (P<0.05) lower GR levels than common/pre-B samples (n=26) both at the mRNA (median 1.7 10^-3 vs 9.5 10^-3 GR mRNA/-actin mRNA (P=0.01)) and protein level (median 0.46 vs 1.3 arbitrary units determined by WB (P=0.004) and median 28 vs 66 nM (P<0.001) determined by LB). Also, GR mRNA levels were significantly lower in T-ALL than in common/pre-B ALL: median 1.2 10^-5 vs 4.5 10^-5 GR mRNA/-actin mRNA (P<0.05). GR/GR ratios did not differ significantly: 154 vs 166 (P=0.48). The percentage of GR did not differ between T-ALL (n=17) and common/pre-B ALL (n=8): median 2.8 vs 2.4% (P=0.44). Two pro-B and four iALL samples with unknown immunophenotype were excluded from these analyses.

GR mRNA levels (expressed as compared to mRNA levels of -actin) were 3.8-fold higher in rALL than in iALL samples (P=0.006). At the protein level, rALL and iALL samples expressed similar levels of GR, both when determined by WB and LB (Table 3). Consequently, rALL samples had a significantly lower protein/mRNA ratio than iALL samples (median 3608 vs 7624 nM/mRNA GR (normalized to -actin) when protein was determined by LB, or median 80 vs 185 GR protein (arbitrary units)/ mRNA GR (normalized to -actin) when protein was determined by WB (P<0.05). rALL samples expressed 3.5-fold higher GR mRNA levels than iALL samples (P<0.01). GR/GR ratios were similar in rALL and iALL (median 207 vs 154; P=0.80). GR levels did not differ significantly between iALL and rALL (2.6 vs 2.8%, P=0.70).

AML samples expressed over two-fold lower levels of GR mRNA than the other leukemia subgroups (P<0.05). GR protein expression determined by LB was two-fold lower in iAML than in iALL samples (P=0.03), which is in line with their lower mRNA expression levels. A similar difference was found for GR protein expression determined by WB (median 0.72 versus 1.1 arbitrary units), but this difference did not reach statistical significance (P=0.28). GR levels were similar in iALL and iAML (0.037 vs 0.032 GR mRNA/-actin mRNA). iAML showed a significantly lower GR/GR mRNA ratio (median 87) than the other leukemia subgroups (P<0.05). GR expression was similar in iALL and iAML (2.6 vs 3.0%, P=0.83).

Variant receptor expression vs GC resistance in iALL

GR expression did not correlate with GC resistance, neither when determined at the mRNA (=0.11; P=0.50) nor protein level (=0.06 (WB) and -0.05 (LB), respectively; P>0.05). GR mRNA expression was not associated with GC resistance (=-0.07, P=0.66). There was no relationship between in vitro GC resistance and GR/ ratios (=0.22, P=0.20). GR/ ratios did not differ significantly between sensitive, intermediately resistant and resistant patients (median sensitive 121, intermediately resistant 195 and resistant 226, P>0.05; Figure 4).

Figure 4.

GR/GR mRNA ratios in GC-sensitive (LC₅₀ PRD 0.1 g/ml) intermediately resistant (LC₅₀ PRD>0.1 but < 150 g/ml) and resistant patients (LC₅₀ PRD 150 g/ml). Median ratios are depicted as horizontal lines (sensitive 121, intermediately resistant 195 and resistant 226, P>0.05).

Full figure and legend (16K)

GR mRNA expression correlated with GC resistance (=0.52, P=0.007, n=26). Resistant and intermediately resistant patients expressed significantly higher levels than sensitive iALL patients (median 4.2 and 2.7% vs 2.0%, P<0.05; Figure 5).

Figure 5.

GR mRNA expression in sensitive (LC₅₀ PRD 0.1 g/ml) intermediately resistant (LC₅₀ PRD>0.1 but <150 g/ml) and resistant patients (LC50 PRD 150 g/ml). Resistant and intermediately resistant patients expressed significantly higher levels than sensitive iALL patients (median 4.2 and 2.7% vs 2.0%, P< 0.05).

Full figure and legend (15K)

Meaningful correlations between GR// and GC resistance could not be established in iAML and rALL because of the relatively small number of patients (n=10 and 11, respectively) and the high incidence of highly resistant patients (median LC₅₀ PRD in AML >250 g/ml, in rALL 31 g/ml vs iALL 0.37 g/ml).

Discussion

The cause of GC resistance in childhood leukemia is largely unknown. The GR has been directly linked to GC resistance. Altered expression levels, genetic alterations (polymorphisms), alternative splicing (GR///P) and secondary modifications of this protein (eg mono-ADP ribosylation) have been suggested to influence cell sensitivity to GC treatment.^{35,41,42,43,44,45,46} In this study, we focused on the expression levels of the GR splice variants GR, GR, and GR in relation to in vitro GC resistance in childhood leukemia.

GR expressions (at both the protein and mRNA level) were highly variable in all leukemia subgroups, with levels being lower in T-ALL and iAML samples, confirming the results previously obtained by our group⁴⁰ and others.^26,39,47

We did not find an association between GR expression and in vitro GC resistance in iALL, which contrasts with the data previously reported by our group.⁴⁰ This difference may be explained by the fact that the number of GC-resistant patients classified as T-ALL, a subgroup with comparatively low GR levels, differs; in the latter study, approximately 6/10 resistant iALL samples were T-ALL, whereas in this study only 3/11 resistant samples were classified as such.

GR mRNA expression correlated with protein expression. However, in rALL we observed a decreased protein to mRNA ratio, suggesting that a less efficient translation of mRNA into protein occurs, or that protein is more rapidly degraded. This phenomenon may be of importance as various studies have shown that for GC-induced cell lysis, autoinduction of the GR is required.^48,49 This upregulation may be hampered by a less efficient mRNA translation or increased protein degradation. Consequently, the GR/GC-mediated signal may not be sufficient to induce apoptosis. However, it still remains to be established whether upregulation of GR expression is required for GC-induced apoptosis in rALL. The fact that decreased protein to mRNA ratios were observed in rALL, a leukemia subgroup over 75-fold more resistant to GC-induced cell lysis,^6,7 suggests a possible link between this process and the acquisition of GC resistance.

This is the first study quantitating GR/GR mRNA levels using a CT-PCR. In the leukemia samples tested, we found a median 85- to 200-fold lower GR mRNA than GR mRNA expression, which is slightly higher than the results previously reported by Oakley et al³⁰ (0.22% in CEM) and Honda et al⁵⁰ (0.16% in normal lymphocytes). We were not able to detect GR at the protein level. This may be caused by low (or absent) levels of the GR protein. As we did not include a known amount of purified GR in the experiment, we cannot exclude the possibility that failure to detect the GR is the result of a comparatively low affinity of the GR antisera used. However, based on the low mRNA levels it seems likely that protein levels are also low or absent.

Transfection studies have demonstrated that a GR to GR ratio of 10:1 is required for significant inhibition of GR function.⁴³ The relatively low GR mRNA levels that were found in this study (median 85- to 200-fold lower than GR) indicate that it is unlikely that the GR is capable of inhibiting GR-mediated cell lysis in childhood leukemia. In accordance with this, we were not able to demonstrate a significant correlation between GR mRNA expression or GR/GR mRNA ratios and in vitro GC resistance.

The 3-bp insertion typical for GR was found in approximately 3% of the total GR message, which is in line with the results previously published by Rivers et al.³⁵ Of particular interest is the fact that GR mRNA expression correlated with in vitro GC resistance in iALL. Given the multifactorial nature of GC resistance and the heterogeneity of clinically obtained specimens, this correlation (=0.52) is relatively strong. In line with our data, preliminary results obtained by Gerdes and co-workers demonstrated a possible link between GR expression and a poor in vivo response to PRD monotherapy in pediatric iALL (Gerdes et al. Quantification of the glucocorticoid receptor and its splice variant gamma in childhood acute lymphoblastic leukemia was carried out using real-time PCR. Blood 2001; 98: 113a; abstract). Based on what is known about the GR so far, it is difficult to explain the relation between GR expression and GC resistance in iALL. Ray et al⁵¹ showed that a 3-bp insertion at the same location in the GR of a carcinoma cell line resulted in an approximately 50% reduction of transcriptional activity.⁵¹ In addition, Kasai et al⁵² previously reported a GR with (also) an arginine inserted at the same site resulting in a similar reduction in transcriptional activity. It seems unlikely that a 50% reduction in transcriptional activity in up to 7.4% of the GR proteins (assuming that the mRNA transcript is as efficiently translated as GR) results in a GC-resistant phenotype.

Several alternative hypotheses can be proposed. First GR overexpression may be related to GC resistance due to confounding bias; that is, the expression of the GR may be related to other (unkown) factors that induce GC resistance in pediatric iALL. This factor may for example favor directly or indirectly the use of the alternative splice donor site between exons 3 and 4, but induce GC resistance through different pathways. Second, apart from reduced transactivating activity, the arginine insertion in the DNA-binding domain may influence, which GC-responsive genes are transcribed. Different cell types within one organism respond differently to GC exposure varying from induction to inhibition of apoptosis, or even induction of cell proliferation.^53,54,55,56 Very little is known about the mechanisms that determine which GC-responsive genes are transactivated or transrepressed upon the activation of the GR. It does not seem unlikely that structural changes in the DNA-binding domain influence this process. The effect of an activated GR on cellular apoptosis may thus differ from (or even oppose) GR. Future studies are warranted to clarify these aspects of GR function.

In conclusion, the low expression levels of GR mRNA and the lack of association with in vitro GC resistance challenge the concept that the GR acts as a significant inhibitor of GR activity in childhood leukemia. The association between GR expression and in vitro GC resistance in iALL warrants further investigation. The fact that decreased protein to mRNA ratios were observed in rALL, a leukemia subgroup over 75-fold more resistant to GC-induced cell lysis, suggests a possible link between this process and the acquisition of GC resistance.

References

1. Kaspers GJL, Pieters R, Van Zantwijk CH, Van Wering ER, Van der Does-van den Berg A, Veerman AJP. Prednisolone resistance in childhood acute lymphoblastic leukemia: vitro-vivo correlations and cross-resistance to other drugs. Blood 1998; 92: 259–266. | PubMed | ISI | ChemPort |
2. Kaspers GJL, Veerman AJP, Pieters R, Van Zantwijk CH, Smets LA, Van Wering ER et al. In vitro cellular drug resistance and prognosis in newly diagnosed childhood acute lymphoblastic leukemia. Blood 1997; 90: 2723–2729. | PubMed | ISI | ChemPort |
3. Pieters R, Huismans DR, Loonen AH, Hählen K, Van der Does-van den Berg A, Van Wering ER et al. Relation of cellular drug resistance to long-term clinical outcome in childhood acute lymphoblastic leukaemia. Lancet 1991; 338: 399–403. | Article | PubMed | ChemPort |
4. Hongo T, Yajima S, Sakurai M, Horikoshi Y, Hanada R. In vitro drug sensitivity testing can predict induction failure and early relapse of childhood acute lymphoblastic leukemia. Blood 1997; 89: 2959–2965. | PubMed | ISI | ChemPort |
5. Riehm H, Reiter A, Schrappe M, Berthold F, Dopfer R, Gerein V et al. Corticosteroid-dependent reduction of leukocyte count in blood as a prognostic factor in acute lymphoblastic leukemia in childhood (therapy study ALL-BFM 83). Klin Padiatr 1987; 199: 151–160. | PubMed | ISI | ChemPort |
6. Zwaan CM, Kaspers GJL, Pieters R, Ramakers-van Woerden NL, Den Boer ML, Wünsche R et al. Cellular drug resistance profiles in childhood acute myeloid leukemia: differences between FAB types and comparison with acute lymphoblastic leukemia. Blood 2000; 96: 2879–2886. | PubMed | ISI | ChemPort |
7. Klumper E, Pieters R, Veerman AJP, Huismans DR, Loonen AH, Hählen K et al. In vitro cellular drug resistance in children with relapsed/refractory acute lymphoblastic leukemia. Blood 1995; 86: 3861–3868. | PubMed | ISI | ChemPort |
8. Hollenberg SM, Weinberger C, Ong ES, Cerelli G, Oro A, Lebo R et al. Primary structure and expression of a functional human glucocorticoid receptor cDNA. Nature 1985; 318: 635–641. | Article | PubMed | ISI | ChemPort |
9. Wright AP, Zilliacus J, McEwan IJ, Dahlman-Wright K, Almlof T, Carlstedt-Duke J et al. Structure and function of the glucocorticoid receptor. J Steroid Biochem Mol Biol 1993; 47: 11–19. | Article | PubMed | ISI | ChemPort |
10. Westphal HM, Mugele K, Beato M, Gehring U. Immunochemical characterization of wild-type and variant glucocorticoid receptors by monoclonal antibodies. EMBO J 1984; 3: 1493–1498. | PubMed | ISI | ChemPort |
11. Eriksson P, Wrange O. Protein–protein contacts in the glucocorticoid receptor homodimer influence its DNA binding properties. J Biol Chem 1990; 265: 3535–3542. | PubMed | ISI | ChemPort |
12. Danielsen M, Hinck L, Ringold GM. Two amino acids within the knuckle of the first zinc finger specify DNA response element activation by the glucocorticoid receptor. Cell 1989; 57: 1131–1138. | Article | PubMed | ISI | ChemPort |
13. Beato M, Chalepakis G, Schauer M, Slater EP. DNA regulatory elements for steroid hormones. J Steroid Biochem 1989; 32: 737–747. | Article | PubMed | ChemPort |
14. Miesfeld R, Godowski PJ, Maler BA, Yamamoto KR. Glucocorticoid receptor mutants that define a small region sufficient for enhancer activation. Science 1987; 236: 423–427. | PubMed | ChemPort |
15. Miller NR, Simons Jr SS. Steroid binding to hepatoma tissue culture cell glucocorticoid receptors involves at least two sulfhydryl groups. J Biol Chem 1988; 263: 15217–15225. | PubMed |
16. Simons Jr SS, Chakraborti PK, Cavanaugh AH. Arsenite and cadmium(II) as probes of glucocorticoid receptor structure and function. J Biol Chem 1990; 265: 1938–1945. | PubMed |
17. Gehring U, Segnitz B. Modulation of DNA binding of glucocorticoid receptors. Mol Cell Endocrinol 1988; 56: 245–254. | Article | PubMed |
18. Prima V, Depoix C, Masselot B, Formstecher P, Lefebvre P. Alteration of the glucocorticoid receptor subcellular localization by non steroidal compounds. J Steroid Biochem Mol Biol 2000; 72: 1–12. | Article | PubMed |
19. Morishima Y, Kanelakis KC, Silverstein AM, Dittmar KD, Estrada L, Pratt WB. The Hsp organizer protein hop enhances the rate of but is not essential for glucocorticoid receptor folding by the multiprotein Hsp90-based chaperone system. J Biol Chem 2000; 275: 6894–6900. | Article | PubMed | ISI | ChemPort |
20. Miyata Y, Yahara I. Cytoplasmic 8 S glucocorticoid receptor binds to actin filaments through the 90-kDa heat shock protein moiety. J Biol Chem 1991; 266: 8779–8783. | PubMed | ISI | ChemPort |
21. Lamberts SW, Huizenga AT, de Lange P, de Jong FH, Koper JW. Clinical aspects of glucocorticoid sensitivity. Steroids 1996; 61: 157–160. | Article | PubMed |
22. Chrousos GP, Castro M, Leung DY, Webster E, Kino T, Bamberger C et al. Molecular mechanisms of glucocorticoid resistance/ hypersensitivity. Potential clinical implications. Am J Respir Crit Care Med 1996; 154: S39–S43.
23. Yamamoto KR, Alberts BM. Steroid receptors: elements for modulation of eukaryotic transcription. Annu Rev Biochem 1976; 45: 721–746. | Article | PubMed | ChemPort |
24. Ray A, Prefontaine KE. Physical association and functional antagonism between the p65 subunit of transcription factor NF-kappa B and the glucocorticoid receptor. Proc Natl Acad Sci USA 1994; 91: 752–756. | Article | PubMed | ChemPort |
25. Schule R, Rangarajan P, Kliewer S, Ransone LJ, Bolado J, Yang N et al. Functional antagonism between oncoprotein c-Jun and the glucocorticoid receptor. Cell 1990; 62: 1217–1226. | Article | PubMed | ISI | ChemPort |
26. Quddus FF, Leventhal BG, Boyett JM, Pullen DJ, Crist WM, Borowitz MJ. Glucocorticoid receptors in immunological subtypes of childhood acute lymphocytic leukemia cells: a Pediatric Oncology Group Study. Cancer Res 1985; 45: 6482–6486. | PubMed |
27. Pui CH, Ochs J, Kalwinsky DK, Costlow ME. Impact of treatment efficacy on the prognostic value of glucocorticoid receptor levels in childhood acute lymphoblastic leukemia. Leukemia Res 1984; 8: 345–350.
28. Pui CH, Costlow ME. Sequential studies of lymphoblast glucocorticoid receptor levels at diagnosis and relapse in childhood leukemia: an update. Leukemia Res 1986; 10: 227–229.
29. Ho GA, Odenwald E, Reiter A, Sauter S, Riehm H. Lack of correlation between glucocorticoid receptor levels, response to prednisone monotherapy and relapse-free survival in childhood leukemia. Proc Am Soc Clin Oncol 1991; 10: 238.
30. Oakley RH, Sar M, Cidlowski JA. The human glucocorticoid receptor beta isoform. Expression, biochemical properties, and putative function. J Biol Chem 1996; 271: 9550–9559.
31. Hecht K, Carlstedt-Duke J, Stierna P, Gustafsson J, Bronnegard M, Wikstrom AC. Evidence that the beta-isoform of the human glucocorticoid receptor does not act as a physiologically significant repressor. J Biol Chem 1997; 272: 26659–26664. | Article | PubMed | ISI | ChemPort |
32. Leung DY, Hamid Q, Vottero A, Szefler SJ, Surs W, Minshall E et al. Association of glucocorticoid insensitivity with increased expression of glucocorticoid receptor beta. J Exp Med 1997; 186: 1567–1574. | Article | PubMed | ISI | ChemPort |
33. Hamid QA, Wenzel SE, Hauk PJ, Tsicopoulos A, Wallaert B, Lafitte JJ et al. Increased glucocorticoid receptor beta in airway cells of glucocorticoid-insensitive asthma. Am J Respir Crit Care Med 1999; 159: 1600–1604. | PubMed |
34. Longui CA, Vottero A, Adamson PC, Cole DE, Kino T, Monte O et al. Low glucocorticoid receptor alpha/beta ratio in T-cell lymphoblastic leukemia. Horm Metab Res 2000; 32: 401–406. | PubMed | ISI | ChemPort |
35. Rivers C, Levy A, Hancock J, Lightman S, Norman M. Insertion of an amino acid in the DNA-binding domain of the glucocorticoid receptor as a result of alternative splicing. J Clin Endocrinol Metab 1999; 84: 4283–4286. | PubMed |
36. Kaspers GJL, Veerman AJP, Pieters R, Broekema GJ, Huismans DR, Kazemier KM et al. Mononuclear cells contaminating acute lymphoblastic leukaemic samples tested for cellular drug resistance using the methyl-thiazol-tetrazolium assay. Br J Cancer 1994; 70: 1047–1052. | PubMed | ISI | ChemPort |
37. Rots MG, Willey JC, Jansen G, Van Zantwijk CH, Noordhuis P, DeMuth JP et al. mRNA expression levels of methotrexate resistance-related proteins in childhood leukemia as determined by a standardized competitive template-based RT-PCR method. Leukemia 2000; 14: 2166–2175. | Article | PubMed | ISI | ChemPort |
38. Willey JC, Crawford EL, Jackson CM, Weaver DA, Hoban JC, Khuder SA et al. Expression measurement of many genes simultaneously by quantitative RT-PCR using standardized mixtures of competitive templates. Am J Respir Cell Mol Biol 1998; 19: 6–17. | PubMed | ISI | ChemPort |
39. Costlow ME, Pui CH, Dahl GV. Glucocorticoid receptors in childhood acute lymphocytic leukemia. Cancer Res 1982; 42: 4801–4806. | PubMed |
40. Haarman EG, Kaspers GJ, Pieters R, Rottier MM, Den Boer ML, Janka-Schaub GE et al. In vitro glucocorticoid resistance in childhood leukemia correlates with receptor affinity determined at 37 degrees C, but not with affinity determined at room temperature. Leukemia 2002; 16: 1882–1884. | Article | PubMed |
41. Pieters R, Huismans DR, Leyva A, Veerman AJP. Adaptation of the rapid automated tetrazolium dye based (MTT) assay for chemosensitivity testing in childhood leukemia. Cancer Lett 1988; 41: 323–332. | PubMed |
42. Moalli PA, Rosen ST. Glucocorticoid receptors and resistance to glucocorticoids in hematologic malignancies. Leukemia Lymphoma 1994; 15: 363–374. | PubMed |
43. Oakley RH, Jewell CM, Yudt MR, Bofetiado DM, Cidlowski JA. The dominant negative activity of the human glucocorticoid receptor beta isoform Specificity and mechanisms of action. J Biol Chem 1999; 274: 27857–27866. | Article | PubMed | ISI | ChemPort |
44. Smets L, Metwally EA, Knol E, Martens M. Potentiation of glucocorticoid-induced lysis in refractory and resistant leukemia cells by inhibitors of ADP-ribosylation. Leukemia Res 1988; 12: 737–743.
45. Hurley DM, Accili D, Stratakis CA, Karl M, Vamvakopoulos N, Rorer E et al. Point mutation causing a single amino acid substitution in the hormone binding domain of the glucocorticoid receptor in familial glucocorticoid resistance. J Clin Invest 1991; 87: 680–686. | PubMed | ISI | ChemPort |
46. Koper JW, Stolk RP, de Lange P, Huizenga NA, Molijn GJ, Pols HA et al. Lack of association between five polymorphisms in the human glucocorticoid receptor gene and glucocorticoid resistance. Hum Genet 1997; 99: 663–668. | Article | PubMed | ISI | ChemPort |
47. Yarbro GS, Lippman ME, Johnson GE, Leventhal BG. Glucocorticoid receptors in subpopulations of childhood acute lymphocytic leukemia. Cancer Res 1977; 37: 2688–2695. | PubMed |
48. Kofler R. The molecular basis of glucocorticoid-induced apoptosis of lymphoblastic leukemia cells. Histochem Cell Biol 2000; 114: 1–7. | PubMed | ISI | ChemPort |
49. Ramdas J, Liu W, Harmon JM. Glucocorticoid-induced cell death requires autoinduction of glucocorticoid receptor expression in human leukemic T cells. Cancer Res 1999; 59: 1378–1385. | PubMed | ISI | ChemPort |
50. Honda M, Orii F, Ayabe T, Imai S, Ashida T, Obara T et al. Expression of glucocorticoid receptor beta in lymphocytes of patients with glucocorticoid-resistant ulcerative colitis. Gastroenterology 2000; 118: 859–866. | Article | PubMed | ISI | ChemPort |
51. Ray DW, Davis JR, White A, Clark AJ. Glucocorticoid receptor structure and function in glucocorticoid-resistant small cell lung carcinoma cells. Cancer Res 1996; 56: 3276–3280. | PubMed | ISI | ChemPort |
52. Kasai Y. Two naturally-occurring isoforms and their expression of a glucocorticoid receptor gene from an androgen-dependent mouse tumor. FEBS Lett 1990; 274: 99–102. | Article | PubMed |
53. Messmer UK, Pereda-Fernandez C, Manderscheid M, Pfeilschifter J. Dexamethasone inhibits TNF-alpha-induced apoptosis and IAP protein downregulation in MCF-7 cells. Br J Pharmacol 2001; 133: 467–476. | Article | PubMed | ISI | ChemPort |
54. Kaspers GJL, Zwaan CM, Pieters R, Veerman AJP. Cellular drug resistance in childhood acute myeloid leukemia. A mini-review with emphasis on cell culture assays. Adv Exp Med Biol 1999; 457: 415–421. | PubMed |
55. Iacobelli S, Ranelletti FO, Longo P, Riccardi R, Mastrangelo R. Discrepancies between in vivo and in vitro effects of glucorticoids in myelomonocytic leukemic cells with steroid receptors. Cancer Res 1978; 38: 4257–4262. | PubMed |
56. Haarman EG, Kaspers GJ, Veerman AJ. Glucocorticoid resistance in childhood leukaemia: mechanisms and modulation. Br J Haematol 2003; 120: 919–929. | Article | PubMed | ISI | ChemPort |

Acknowledgements

This work was supported by a grant from the Dutch Cancer Society (VU 97-1564) and the European Commission (EUGIA). We are especially grateful to Dr LJ van't Veer from the Netherlands Cancer Institute (Amsterdam) and Dr. JP Meijerink from the Sophia Children's Hospital (Rotterdam) for their many helpful suggestions. In addition, we would like express our gratitude to Dr JC Willey from the medical college of Ohio (USA) for his help in the design of the primers for the CT-RT-PCR. We also thank AJF Broekhuizen, DR Huismans, AH Loonen and CH van Zantwijk for their excellent technical assistance. Finally, we thank the members of the board of the DCOG: H van den Berg, JPM Bökkerink, SSN de Graaf, B Granzen, PM Hoogerbrugge, WA Kamps, FAE Nabben, R Pieters, JA Rammeloo, T Révész and AJP Veerman.