Expression, regulation and function of phosphofructo-kinase/fructose-biphosphatases (PFKFBs) in glucocorticoid-induced apoptosis of acute lymphoblastic leukemia cells

The ALL children enrolled in this study were admitted to the Department of Pediatrics, Innsbruck Medical University, or the St. Anna Kinderspital in Vienna from September to December 2009, and treated according to the BFM protocol 2000 (for protocol details see: http://boriskononenko.files.wordpress.com/2009/09/1-protokoll-all-bfm-2000-mit-amendments-version-26-09-07-d0bad0bed0bfd0b8d18f1.pdf). To analyze the GC response in non-leukemic peripheral blood lymphocytes, Ficoll-purified peripheral blood mononuclear cells from 2 healthy volunteers and 3 children with epilepsia treated with a single injection of GC were included. The study was approved by the Ethics Committees of the Innsbruck Medical University (EK1-1193-172/35 for ALL children and UN2821 for epileptic children) and the Kinderspital in Vienna, and written informed consent was obtained from the patients and/or parents or custodians. The characteristics of 3 children with T-ALL and 10 with precursor B-ALL have been published previously [5], the remaining ALL children (3 T-ALLs and 15 precursor B-ALLs) showed similar clinical findings [Rainer et al. in preparation]. All 3 patients with epilepsia (E3, a 14 year-old girl; E4, a 6 year-old boy; E5, an 11 year-old boy) had an idiopathic generalized epilepsia without any evidence of metabolic, endocrinological, immunological, hematological or infectious disease. The 6 year old girl (epilepsia diagnosed at 4 years of age) was on Valproinate and Clobazam, the 9 year old girl (epilepsia diagnosed at 8 year of age) on Valproinate and Rufinamide, and E5 (epilepsia diagnosed at 2 years of age) on Valproinate and Sultiam. Because of therapy-refractory epilepsia, the 3 patients received their first GC pulse therapy with dexamethasone (Fortecortin, Merck, Vienna, Austria) 20 mg/m² administered intravenously over 1 hour. Blood samples were taken at 2, 6 and 24 hours after drug administration.

The T-ALL cell lines CCRF-CEM-C7H2 [18], 6 GC-sensitive and 6 GC-resistant derivatives CCRF-CEM-C7H2 [19], CEM-C7H2-2C8 [20], a CEM-C7H2 derivative with constitutive expression of the tetracycline-regulated reverse transactivator, rtTA [21], MOLT4 (CRL-1582, ATCC, Rockville, MD), and Jurkat (untransfected and a rat GR-transfected derivative [22]), the precursor B-cell lines 697/EU-3 (ACC 42, DSMZ, Braunschweig, Germany), NALM6 (ACC 128, DSMZ), RS4;11 (ACC 508, DSMZ) and AT-1 [23], and Burkitt's lymphoma Daudi (CCL-213, ATCC) were cultured in RPMI 1640 supplemented with 10% fetal calf serum and 2 mM L-glutamine at 37°C, 5% carbon-dioxide and saturated humidity. HEK293T packaging cells (ATCC, Manassas, VA) were cultured in DMEM supplemented as above. The cells were free of mycoplasma infection and their authenticity was verified by DNA fingerprinting, as detailed previously [24]. Doxycycline was dissolved (100 μg/ml) in phosphate-buffered saline and dexamethasone (10^-4M) in 100% ethanol. The final ethanol-concentration in the dexamethasone-treated and control cultures was maintained at 0.1%. All above reagents were from Sigma (Vienna, Austria).

Our procedure for generating mRNA expression profiles on Affymetrix HGU133 plus 2.0 microarrays has been detailed previously [5]. Briefly, RNA from peripheral blood lymphoblasts of patients prior to, or at 6-8 or 24 hours after initiation of systemic GC monotherapy (following the BFM therapy protocol recommendations), from healthy volunteers or epileptic children 6 hours after GC treatment, and GC-sensitive or GC-resistant cell lines, was converted into labeled target and hybridized to Affymetrix HGU133 plus 2.0 microarrays according to standard protocols. The microarray data from the 31 ALL-patients and 5 non-leukemic individuals were normalized separately from those of the 12 cell lines using GCRMA [25] after positive evaluation of comprehensive raw data quality assessment. The analysis was performed in R using packages from Bioconductor [26] version 2.5. Annotation of probe sets to genes based on Affymetrix NetAffx annotation database version 22 and location of probe sets relative to the 3' end of the transcripts were manually inspected using the Ensembl genome browser for genes PFKFB1, PFKFB2, PFKFB3 and PFKFB4. Microarray data for the samples of the 13 ALL children and 2 healthy donors were published previously [5] and are available at NCBI's GEO database (series GSE2677 and GSE2842). Microarray data for the 12 cell lines (GSE22152) and 4 additional non-leukemic individuals (GSE22779) have been deposited at GEO.

Apoptosis was determined by FACS analysis of propidium iodide (PI)-treated permeabilized cells [27] as previously detailed [28]. Briefly, cells were analyzed with a FACScan cytometer (Becton Dickinson Biosciences, San Jose, CA) in combination with CellQuest Pro software (Becton Dickinson Biosciences) acquiring forward scatter/sideward scatter, FL-2 H (log), and FL-3 H (linear). In FL-2 H, the percentage of nuclei with reduced DNA-content (SubG1 peak) was assessed.

Our immunoblotting procedure has been detailed recently [29]. Briefly, proteins were extracted from 5 × 10⁶ cells in 100 μl RIPA-buffer, quantified by Bradford analysis, mixed with 40 μl loading buffer (4 × SSB, 5% β-mercaptoethanol), denatured, fractionated on a 12.5% SDS-PAGE and electroblotted onto nitrocellulose. The membranes were incubated overnight with rabbit polyclonal antibodies against PFKFB2 (N-term, AP8146a, Abgent), PFKFB2 phosphorylated Ser466 (University of Dundee, UK) or mouse monoclonal antibodies against α-TUBULIN (DM1A, CalBiochem, Nottingham, UK) as a loading control. Specifically bound antibodies were detected with anti-rabbit, anti-sheep or anti-mouse horseradish-peroxidase-conjugated secondary antibodies (Amersham Pharmacia Biotech, Uppsala, Sweden) and visualized by chemiluminescence (ECL, Amersham) and subsequent exposure to AGFA Curix X-ray films for 1 second to 30 minutes.

The lentiviral conditional expression constructs pHR-tetCMV-PFKFB2-15A and -15B (U264 and U265) were generated using the GATEWAY™technology (Invitrogen, Carlsbad, CA). The details of this procedure and the generation of stable clonal cell lines with tetracycline-regulated expression of cDNAs cloned into such constructs has been described previously [30]. In brief, human sequences coding for PFKFB2-15A and -15B mRNA were PCR-amplified using unique SacI-flanked forward primer 5'-GAGCTCTGTGCTCGACGAGCTCGT-3' and XbaI-flanked isoform specific reverse primers 5'-TCTAGATGGGTCTTCGGCTAGT-3' (for PFKFB2-15A) and 5'-TCTAGAGTAGATCCCAGTCGT-3' (for PFKFB2-15B) and cDNA from CEM-C7H2 cells as template. The purified PCR product was cloned into the SacI-XbaI site of pENTR-MCS-deltaNcoI (U243), thereby generating pENTR207-PFKFB2-15A (U260) and pENTR207-PFKFB2-15B (U263). Constructs were sequence-verified and subsequently recombined into the "destination vector" pHR-tetCMV-Dest-IRES-GFP (U192) to generate pHR-tetCMV-PFKFB2-15A-ires-GFP (U264) and pHR-tetCMV-PFKFB2-15B-ires-GFP (U265), as previously described [30]. The lentiviral plasmids were transfected into 293T packaging cells together with pVSV-G and pSPAX (kindly provided by Didier Trono), and the lentivirus-containing supernatants were used to transduce CEM-C7H2-2C8, which constitutively express rtTA [20]. After limiting dilution cloning, three clonal cell lines expressing the PFKFB2-15A isoform, termed CEM-PFKFB2-15A#C3, #D6 and #E8, and three expressing PFKFB2-15B named CEM-PFKFB2-15B#65, #66 and #95, were selected for further experiments.

For high through-put real time RT-PCR, 50 μl of diluted cDNA (2 ng/μl) were added to 50 μl of TaqMan Universal MasterMix (Applied Biosystems, Foster City, CA) and introduced into microfluidic cards containing real time RT-PCR mixes for human PFKFB2 (Hs01015408_m1, Applied Biosystems) and TATA box-binding protein for mRNA normalization (TBP, Hs00427620_m1) according to the manufacturer's guidelines. After equilibration at RT, the channels were filled with 100 μl 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. Similarly, but on conventional 96 well plates, mRNA encoding the 2 PFKFB2 splice variants 15A and 15B were quantified (Hs01015410_m1 and Hs01016554_m1, Applied Biosystems, for PFKFB2-15A and PFKFB2-15B, respectively). Primary real-time PCR data analysis was performed with SDS software version 2.2.1, and further analysis was performed in R (version 2.8). Data from 3 technically replicated measurements were averaged and normalized to the internal TBP control. Log2 fold change values (M values) were calculated for 3 biological replicates by comparing normalized real-time PCR data from GC-treated samples against data from the corresponding control samples. M values were averaged for the 3 biological replicates and p-values were calculated (Student's t-test) to test against the null hypothesis of no differential expression (mean M = 0).

In a previous study, we showed that PFKFB2 is regulated by GC in 11/13 children suffering from T-ALL and preB-ALL [5]. To elucidate whether GC exclusively regulated PFKFB2 in lymphoid cells, or whether other PFKFB-isoenzymes are susceptible to GC-treatment as well, we re-analyzed this data using the more robust GCRMA normalization and included 18 additional ALL patients undergoing GC therapy as well as 3 non-leukemic children who received GC as part of their epilepsia treatment, and 6 GC-sensitive and 6 GC-resistant derivatives of the CEM-C7H2 T-ALL cell line. Additional file 1, Table S3 summarizes mean expression levels (mE-value) and mean GC regulation (mM-value) of the 3 isoforms, i.e. PFKFB1, PFKFB3 and PFKFB4, in these samples, organized in the following groups: T-ALLs, precursor B-ALLs [as one group and as 3 separate groups based on molecular defects: hyperdiploid, ETV6/RUNX1 (previously TEL/AML1) translocation, and a heterogeneous group termed "others"], non-leukemic donors and GC-sensitive and -resistant CCRF-CEM derivates. Data for individual patients and cell lines are depicted in Additional file 1, Tables S1 and S2. PFKFB1 and PFKFB4 isoenzymes were neither detectably expressed in any of the investigated systems nor regulated by GC. The brain/placenta- and tumor-specific PFKFB3 isoenzyme showed expression levels comparable to those of PFKFB2 (for comparisons see Table 1), however, in contrast to PFKFB2, its expression was not upregulated by GC. Interestingly, expression of PFKFB3 was also observed in peripheral blood lymphocytes from non-leukemic donors, indicating that this isoenzyme is not restricted to placenta and brain or malignant tissues. In conclusion, the PFKFB1 and 4 isoenzymes were neither expressed nor regulated by GC and, hence, might not play a role in the anti-leukemic GC effects. PFKFB3 is expressed, but shows no significant GC-regulation, excluding it from the list of likely candidate genes.

Next, we analyzed expression and GC regulation of the 2 splice variants of PFKFB2 in the above biological systems. As summarized in Table 1, basal expression of both PFKFB2 isoforms was variable: almost all T-ALLs (known to show particularly strong response to GC) and ETV6/RUNX1 positive ALLs (which have a good prognosis) showed intermediate, all other subgroups, no or very low expression. In all groups with detectable expression of PFKFB2, mRNA levels (as measured by signal intensity) of PFKFB2-15A were ~2-fold higher than those of PFKFB2-15B. Concerning regulation, T-ALL patients showed upregulation of both splice variants at both time points, with mean M values between 1.7 and 2.7. This was also true for precursor B-ALL patients, although the extent of regulation was considerably less (mean M values between 0.6 and 1.1).

On an individual basis (Additional file 1, Table S1), the T-ALL group appeared quite homogeneous, with 5/6 T-ALL children showing induction (as defined by an M-value >1, i.e., >2-fold) after 6 and/or 24 hours of GC-treatment. In hyperdiploid precursor B-ALL children, isoform PFKFB2-15A was upregulated in 5/7 cases, with one child presenting a tendency for downregulation (M-value = -0.6), while PFKFB2-15B induction was present only in 1 of the 7 children after 6 hours. The subgroup of children with the ETV6/RUNX1 translocation showed induction of PFKFB2-15A in 4/10 children, 2 showed a slight downregulation, and 4 lacked PFKFB2-15A regulation. A more homogeneous situation was observed for the PFKFB2-15B isoform, where induction was detected in 8/10 children, with the remaining 2 showing a tendency to downregulate PFKFB2-15B after the first time point. The last subgroup [low hyperdiploid, E2A/PBX1 translocation, t(8;4) and no chromosomal abnormalities] showed upregulated PFKFB2-15A in 7/8 cases and PFKFB2-15B in 4/8, and no regulation in the others. In contrast to the ALL samples where PFKFB2 induction was a frequent event, this was seen only once and only for PFKFB2-15A in the non-leukemic donors. Concerning the in vitro cell lines systems, all 6 GC-sensitive CEM-C7H2 subclones showed clear induction of PFKFB2-15A and -15B, with M values ranging from 1.6 to 4.0 (15A) and 1.1 to 2.3 (15B). The 6 R-lines resulted in a remarkably heterogeneous picture, with M-values ranging from 0.0 to 5.1 (15A) and 0.5 to 3.5 (15B). Thus, despite the complexity of the expression profiling data, PFKFB2 induction appeared to be a frequent feature in the GC response of T-lineage ALL and also occurred in precursor B-ALL, albeit with lower frequency. Moreover, PFKFB2 induction was absent or reduced in systems resistant to GC-induced apoptosis (non-leukemic donors, CEM-C7H2-R-lines).

To further address whether PFKFB2 regulation was related to GC-sensitivity and/or preB-ALL and T-ALL origin, we treated 9 leukemia cell lines with 10^-7M dexamethasone for 2, 6, 12 and 24 hours in biological triplicates and performed real time RT-PCR using primers covering the exon-3/exon-4 boundary present in both splice variants (Figure 1A). To assess the contribution of the individual splice variants, the 24 hour samples were further analyzed using primers specific for PFKFB2-15A and -15B (Figure 1B and 1C). As we previously reported [31], the extent of GC sensitivity in these cell lines was markedly different, i.e., untransfected Jurkat and MOLT4 T-ALL cell lines as well as AT-1 precursor B-ALL and Daudi Burkitt lymphoma cells were resistant to GC-induced apoptosis, while all others were GC-sensitive, although with varying kinetics (Additional file 1, Figure S1). Effects of GC on cell cycle progression were also determined [31] (Additional file 1, Figure S1) and showed that GC-induced apoptosis was frequently preceded or accompanied by an increase of cells in the G1 phase of the cell division cycle, whereas the cell lines resistant to GC-induced apoptosis were also resistant to the GC effects on the cell cycle. The notable exception was AT-1, in which G1 cell cycle arrest was observed in the complete absence of apoptosis. As shown in Figure 1A, PFKFB2 induction was a frequent event in these cell lines starting as soon as 2 hours after initial GC exposure and increasing up to 24 hours. In those instances where PFKFB2 induction was observed, both splice variants appeared to be similarly regulated (Figure 1B). There was no apparent correlation between PFKFB2 induction and GC sensitivity or T/B-lineage origin of the leukemic cell lines. Concerning basal levels, the 2 splice variants showed co-expression across the panel of leukemia cell lines (Figure 1C), with PFKFB2-15A levels slightly exceeding those of -15B, thus resembling the findings on the microarrays. In general, B-lineage cells revealed lower expression levels than T-lineage cells (again resembling the situation in patients), but there was no apparent correlation between basal PFKFB2 levels and GC-sensitivity. For instance, GC-sensitive C7H2 T-ALL cells had similar PFKFB2 levels as GC-resistant MOLT4 T-ALL. The same was true for GC-sensitive Jurkat^GR and their parental GC-resistant Jurkat cell line.

Bild vergrößern

Similar to the situation in patients, GC-dependent induction was more pronounced in T-ALL cells than in B-lineage leukemias, but there was no obvious correlation with GC sensitivity across all samples. In the Jurkat T-ALL system, however, induction of PFKFB2 correlated with GC sensitivity, as we have similarly seen in the CCRF-CEM system using microarray technology. Although we were unable to definitively prove regulation at the protein level (for reasons explained in Additional file 1), the combined clinical and experimental mRNA data were compatible with the notion that induction of PFKFB2 contributes to GC-induced apoptosis, particularly in T-ALL cells.

To evaluate a possible contribution of increased PFKFB2 levels to the anti-leukemic effects of GC, we generated cell lines conditionally overexpressing either PFKFB2-15A or -15B in a doxycycline-dependent manner. Three clones for each isoform (termed CEM-PFKFB2-15A#C3, #D6, #E8 and CEM-PFKFB2-15B#65, #66, #95, respectively) were analyzed for regulation of the transgene by quantitative RT-PCR and immunoblotting (Figure 2 and Additional file 1, Figure S2 and S3). In all instances we obtained doxycycline dose- and time-dependent induction of the PFKFB2 splice variants to levels as high as, or even exceeding, those obtained after GC treatment. Nevertheless, no effect was observed on cell viability, suggesting that induction of either PFKFB2 splice variant is not sufficient to explain apoptosis seen after GC exposure (Figure 3). In addition, the FACS data of PI-stained nuclei strongly suggested that there was no effect on cell cycle progression or GC-induced cell cycle arrest (data not shown). To investigate whether transgenic expression of PFKFB2 resulted in altered GC susceptibility, we treated the above T-ALL cell lines with 10^-7M dexamethasone in the presence and absence of doxycycline (i.e., with and without transgenic PFKFB2 induction) and analyzed the extent of apoptosis (as measured by FACS analysis of propidium iodide incorporation) at various time points (Figure 3). Neither PFKFB2-15A nor -15B over-expression significantly changed GC sensitivity in these cells, although both isoforms were clearly detectable by immunoblotting (Figure 3C). Moreover, in the case of PFKFB2-15A where corresponding analyses can be performed, we observed that the activation-specific Ser466 [11] was phosphorylated, suggesting that the transgenic protein was not only sufficiently well expressed, but also active as a kinase. In conclusion, the data show that, at least in the investigated model system, neither PFKFB2-15A nor -15B over-expression mimics the anti-leukemic effects of GC, nor does it alter GC sensitivity.

Bild vergrößern

Bild vergrößern