More than 25,000 colonies of our leukemia cDNA library were screened with cDNA probes made from pooled leukemia cell RNA from patients with chemosensitive (L1-3) AML and from patients with AML resistant/relapsed leukemia (R1–R5). We identified about 120 colonies with increased hybridization to cDNA probes from R-cells. They were subjected to a secondary screening, in which 19 putative genes were found to be upregulated more than 2-fold. The genes were identified by DNA sequencing (Table 1). One of the genes with increased (5-fold) expression in the R-cells has not been described earlier and consequently has not been associated with any function so far. We refer to its predicted product as

A

r

a

p

The screening result for the genes with highest relative expression in R-cells was validated by RT-PCR in the pooled cDNA probes prepared from RNA of L- and R-cells. Two sets of specific primers were used and rpP2 served as a control for equal loading. The novel gene ARAP was confirmed to be relatively overexpressed in the R-cells (Fig. 1). The LEDGF/p75 gene generates two transcripts by alternative RNA splicing, LEDGF/p75 and p52 mRNAs 101112. Both these transcripts showed increased levels in relapsed/chemoresistant AML, as did the transcripts for the ribosomal proteins L6 and S4 (Fig. 1). From these data, we conclude that the R-cells had increased expression of several genes associated with transcription, translation, mitochondrial function, and protein processing, suggesting a possible reprogramming of several pivotal cell functions of the blast cells from patients with chemoresistant AML.

We have recently found profound differences in phosphoprotein signaling cascades between AML cells from patients with low-risk and high-risk of relapse 13. Our L-cells were from low-risk patients, and their expression of 9 of the genes presented in Table 1 was studied in cells from individual patients. For comparison, we studied the expression of the same genes in cells from individual patients with high risk (H-cells) of relapse and the cells from patients with resistance (R-cells) to chemotherapy regimens (Table 1). It appeared that ARAP expression was enhanced in 4 of the 5 R-cell patients, average more than 4-fold, but not increased in the H-cells. High expression of this gene could therefore be a consequence of the chemotherapy. The genes encoding the transcription factors LEDGF/p75 and ERF-2 were also generally more upregulated in the R-cells than in the H-cells (Table 1). The other genes studied were upregulated in at least half of the patients whose R- and H-cells were studied. They were generally equally or more upregulated in the H-cells than in the R-cells, suggesting that previous chemotherapy was not required for the overexpression. We conclude that high-risk and low-risk AML cells differ not only with respect to phospho-signalomics 13, but also with respect to expression of the genes associated with AML relapse. Most R and H patients had increased expression of a number of the resistance-associated genes. We focused in more detail on the LEDGF/p75 because it had the highest relative expression in the R-cells relative to the L-cells among the differentially expressed genes tested (Table 1) and earlier studies indicating LEDGF/p75 to be involved in cell survival 1415.

Both LEDGF/p75 and p52 were associated with chemoresistance in native human AML cells (Fig. 1). In subsequent experiments, the full-length LEDGF/p75 and p52 cDNAs were cloned by RT-PCR from the promyelocytic leukemia NB4 cells. Additional novel spliced LEDGF/p75 mRNA transcripts were detected by RT-PCR of the overlapping region between LEDGF/p75 and p52, and these splicing patterns were confirmed by RT-PCR using primers derived from exon 3 and exon 8 (Fig. 2A). Subcloning and sequencing of p52 cDNAs derived from NB4 AML cells identified four previously unknown splice variants of the LEDGF/p75/p52 gene (summarized in Fig. 2B). One, p52b, had a p52-specific exon 9b. The putative amino acid sequence of this novel and well expressed (Fig. 2A) p52 mRNA variant was identical to p52 except for 25 extra amino acid residues at the C-terminal end due to an altered open reading frame. Three different low expression (Fig. 2A) variants without exon 6 are referred to as p52bΔE6 (Fig. 2B), p52Δ1, and p52bΔ1. The P52bΔE6 variant coded for a putative truncated protein due to a stop codon introduced into the P52b reading frame, whereas p52Δ1 and p52bΔ1 encoded proteins with gross deletions (lack of exons 6, 7 and large parts of 5 and 8).

LEDGF/p75 can be expressed in a number of splice variants in AML cells. We wanted to know the possible functional significance of the overexpression of LEDGF/p75 in AML cells from patients with chemoresistance, and the significance of the splice variants of p52 described. For this purpose, we first used Flp-In 293 cells transfected with the various genes and analyzed for induction of spontaneous apoptosis or for ability to modulate the apoptosis induction by daunorubicin.

We previously used transiently transfected HEK293 cells to show that Bcl-2 protects against daunorubicin-induced, but not okadaic acid induced apoptosis 16. In the present study, we compared the ability of transfected LEDGF/p75 and Bcl-2 to protect against daunorubicin induced apoptosis, using either a moderate drug concentration (0.5 μM) for an extended period of time (16 h) or a high concentration (5 μM) for a short period (1 h). In either case, the protective effect of LEDGF/p75 was comparable with that of Bcl-2, transfected under similar conditions (Fig. 3). We found significantly increased survival, as compared to cells transfected with empty vector, for cells transfected with LEDGF/p75 (p < 0.01). We conclude that LEDGF7p75 consistently overexpressed in resistant and high-risk AML patient cells were indeed able to protect against a first line anti-leukemic drug.

We investigated next the effect of LEDGF/p75 splice variants on in vitro apoptosis of Flp-In 293 cells. First, the effect of the splice variants was investigated for spontaneous apoptosis in cells cultured in medium alone. A decreased number of viable cells was observed for cells transfected with the three splice variants lacking exon 6 (p52Δ1, p52bΔ1, p52bΔE6, see Fig. 2) compared with cells transfected with the control vector (Fig. 4). A similar decrease in viability was also observed when testing an artificial LEDGF/p75_317–530 gene segment encoding only the 317–530 amino acids of the C-terminal domain and thereby lacking overlap with p52 (Fig. 4). In contrast, a high viability was observed both when testing cells transfected with the control vector and with the p52b splice variant containing exon 6. The non-viable cells had morphology consistent with apoptosis in all these cultures. Thus, the 3 splice variants without exon 6 seemed to mediate proapoptotic effects. Secondly, a similar difference between splice variants without exon 6 and the p52b variant was observed for daunorubicin-induced apoptosis (0.5 μM daunorubicin, 16 hrs exposure), but the differences were smaller than for spontaneous apoptosis (Data not shown). Finally, the effect of the four p52 splice variants was also determined for Flp-In 293 cells transfected with LEDGF/p75 and exposed to daunorubicin as described above. LEDGF/p75 antagonizes daunorubicin-induced apoptosis, and the number of viable cells did not differ between the four p52 splice variants in these experiments.

Since the LEDGF/p75 overexpression was found in resistant AML cells, it was important to know whether overexpression of LEDGFp75 and p52b also could protect AML cells. For this, we chose the highly anthracyclin-sensitive IPC-81 rat AML cell line 17, which originates from the BNML rat model of leukemia, known to be a reliable predictor of treatment success for chemotherapy agents in current clinical use against AML 18. These cells are also highly sensitive to apoptosis induction by activation of the cAMP-dependent protein kinase 1920. In order to achieve efficient gene transfer, the LEDGF/p75 and p52b constructs were delivered via a retroviral system. In order to visualize transduced cells, the LEDGF/p75 genes were cloned in tandem with a GFP gene governed by an internal ribosome entry site. We found that cells with enforced expression of either LEDGF/p75 or p52b were significantly protected against apoptosis induced by either daunorubicin or cAMP analog (Fig. 5). We conclude that both LEDGF/p75 and p52b are able to protect against AML death in vitro, suggesting that their consistent overexpression in AML cells from treatment – resistant AML is not coincidental.

The regulation of LEDGF/p75 expression was investigated in NB4 human AML cells. We analyzed LEDGF/p75 mRNA expression in serum-starved NB4 cells, and significantly increased LEDGF/p75 mRNA levels were observed after 24 and 48 hours (Fig. 6A). We then investigated the effect of daunorubicin on expression of LEDGF/p75 mRNA and its protein. rpP2 cDNA as a standard reference in the RNA preparations. LEDGF/p75 mRNA was significantly increased in cells treated with 30 nM and 90 nM daunorubicin for 18 hours (<10% and 40–50% apoptosis, respectively), whereas LEDGF/p75 mRNA levels were reduced and a majority of cells were apoptotic (approx. 90%) when cells were exposed to 180 nM daunorubicin (Fig. 6B). The levels of LEDGF/p75 protein in total protein extracts from NB4 cells were further analyzed by western blot analysis using anti-LEDGF/p75 antibodies. A 75 kDa band specific of LEDGF/p75 increased in the presence of 30 and 90 nM daunorubicin (Fig. 6C), whereas after exposure to 180 nM daunorubicin, degradation of LEDGF/p75 was observed with the formation of a novel protein band of approximately 58 kDa. This new protein seems to be a relatively stable LEDGF/p75 degradation product generated by caspase 3 mediated cleavage during apoptosis 21.

AML blasts were collected from the peripheral blood of 13 patients. Blasts were collected from 3 patients with chemo-sensitive AML at the time of diagnosis ("low risk", L1-3). All had complete hematological remission after the first induction cycle. Two are still alive 93 and 109 months after diagnosis, respectively; the third had 44 months of disease-free survival even though consolidation therapy could not be completed due to serious side effects. All three had low/intermediate risk of relapse 431 judged from cytogenetics (two inv(16), one normal) and no genetic Flt3 abnormality. AML cells were collected from five patients who had undergone previous intensive therapy, (i) two patients had relapse after intensive chemotherapy and allotransplantation, respectively, and (ii) three patients had primary resistant disease and did not reach remission after three induction cycles. For comparison, cells were collected at the time of primary diagnosis from patients with high risk of resistance/relapse 313233. AML ("high-risk", H1-5). Two patients (H1, 2) had AML secondary to primary myelodysplastic syndrome (MDS), and treatment was not attempted, whereas the three last patients (H3-5) later showed resistance to at least two different induction regimens. All human samples were collected in accordance with the Helsinki Declaration.

Leukemic peripheral blood mononuclear cells (PBMC) were collected from peripheral blood, and isolated by density gradient separation (Ficoll-Hypaque; NyCoMed, Oslo, Norway; specific density 1.077) 3435. Cells were stored in liquid nitrogen until use. The percentage of blasts among leukemic PBMC exceeded 95% for all patients, as judged by light microscopy of May-Grünwald-Giemsa stained smears.

Flp-In 293 cells (Invitrogen) were cultured in DMEM medium supplemented with 10% heat-inactivated fetal bovine serum. The p52 specific splice variants were cloned into pcDNA3.1/V5-His TOPO (Invitrogen). LEDGF/p75 in pcDNA3.1/His was a generous gift from C. A. Casiano (St. Loma Linda University, USA). Cells were cultured and transfected using FuGENE-6 (Boehringer Mannheim) according to the manufacturer's protocol. After 24 h, cells were diluted to desired number and seeded into collagen coated 12-well plates. Cells were fixed by 4% buffered paraformaldehyde after DNR treatments. Cells were examined for daunorubicin-induced apoptosis by fluorescence microscopy after DNA staining 1634. For estimation of statistical significance of values for percentage of apoptosis, we used the non-parametric Wilcoxon paired comparison test.

The promyelocytic leukemia cell line IPC-81 and NB4 cells were generous gifts from Dr M Lanotte (INSERM U-496, Centre G Hayem, Hospital St Louis, Paris, France). The cells were cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen) under standard conditions supplemented with 10% heat-inactivated horse serum (Invitrogen), streptomycin (5 μg/ml) and penicillin (5 U/ml). Cells were kept in logarithmic growth, and as a standard procedure, cell density was adjusted to 0.2 × 10⁶ cells/ml before addition of drugs. The cyclic AMP analog 8-chlorophenylthio-cAMP (8-CPT-cAMP) was purchased from Biolog Life Science, Bremen, Germany. Daunorubicin (DNR) was purchased from Sigma-Aldrich Inc. (St Louis, MO, USA).

Phoenix-Eco virus producer cells and CRU5-IRES-GFP retrovirus vector were kindly provided by Dr. J. Lorens (Department of Biomedicine, University of Bergen) and the cells were maintained in DMEM supplemented with 10% heat-inactivated fetal calf serum (Invitrogen). The full-length cDNAs of LEDGF/p75 and p52b with Bst XI linker were cloned into CRU5-IRES-GFP.

Retrovirus were pseudotyped with the vesicular stomatitis virus glycoprotein (VSVG) and concentrated by ultracentrifugation essentially as described 36. Briefly, 1.5 × 10⁶ Phoenix cells plated the day prior to transfection using the calcium phosphate method, 2 μg retroviral plasmid and 2 μg VSV-G. After 48 hours, filtered supernatants were concentrated by centrifugation for 3 hours at 50 000 g, 4°C. IPC-81 cells (0.5 × 10⁶) were transduced by spin infection in which a mixture of virus and cells was centrifuged at 1200 g for 90 minutes at room temperature in the presence of protamine sulphate (5 mg/ml, Sigma-Aldrich). Routinely, 90–95% of the cells showed fluorescence 48 hours after infection. After treatment, cells were fixed in phosphate-buffered saline containing 2% formaldehyde. Screening for apoptosis was done blindly by two independent experienced evaluators using phase contrast microscopy to study surface morphology and nuclear morphology as visualised by the DNA-specific stain bisbenzimide, Hoechst 33342 (Sigma-Aldrich). Three different fields were randomly selected for counting at least 300 cells. All experiments were repeated at least five times. Percent apoptosis is defined as the ratio between cells with both membrane budding and chromatin condensation and the total of cells counted.

The construction of a leukemia cDNA library and hybridization was as described previously 37. The SMART system (Clontech) was used to construct a cDNA library from the pooled RNA collected from AML patient cells (Table 1; S1 to S3 and R1 to R5). The UV-cross linked replica membranes with total over 25,000 colonies were subjected to primary screening by radio-labeled first strand cDNA from the pooled RNA of either "low risk" (L1-3) or "resistant" (R1-5) AML cells. Colonies with different signal intensity towards cDNA derived from resistant and sensitive AML were subjected to a second round of screening. DNA sequence reactions were carried out by the BigDye terminator method and analyzed by automatic sequencing (ABI PRISM). The BLAST search program was used for homology search (National Center for Biotechnology Information, Bethesda, MD).

Gene fragments from selected differentially expressed genes were amplified from plasmids by PCR using T7 promoter and M13 reverse primers. The amplified specific PCR fragments were used in dot-blot analysis using ³²P-labeled cDNA probes from RNA of L1, 2, 3; R1, 2, 3, 4, 5 and H1, 2, 3, 4, and 5. The membranes were analyzed by phospho-imaging, and the hybridization signal normalized to the mRNA expression level of the ribosomal protein P2 (rpP2) gene for sample from each AML patient.

The 5'-end of ARAP mRNA was obtained by using 5'Generacer kit (Invitrogen) with primers described in the kit and antisense primers designed from the cDNA sequence. The antisense primers, 5'-GTCATCATCTTCTGAGTCTGACTC-3' and 5'-CACCATCATCACCTTGTTTTTGCC-3', were used for PCR amplification.

Total RNA from either NB4 cells or from patient AML blasts pooled from S1-3 or R1-5 was used for ss cDNA synthesis using oligo-(dT)-primer and superscript II reverse transcriptase (Invitrogen). The PCR reaction (50 μl) contained 1 μl cDNA and specific 5'-and 3'-primers. The PCR products were collected after 21–24 cycles of amplification. The primers were: rpL6, 5'-ATGGCGGGTGAAAAAGTTGAGAAGC-3' and 5'-GTAGAACACCAATTTGTGAGGA-3'; rpS4, 5'-ATGGCTCGTGGTCCCAAGAAGCATC-3' and 5'-GCACCCACTGCTCTGTTTGGCCGCC-3'; ARAP, 5'-CAAAAACAAGGTGATGATGATGG-3' and 5'-CCAGAGGGGCGGCAGCAGTCC-3'; Primers for full-length ARAP, 5'-ATGCCCAAATCCAAGCGCGACAA-3' and 5'-GTCATCATCTTCTGAGTCTGACTC-3'; LEDGF/p75, 5'-ATGACTCGCGATTTCAAACCTGGA-3' and 5'-GTTATCTAGTGTAGAATCCTTCAGA-3'; p52, 5'-ATGACTCGCGATTTCAAACCTGGA-3' and 5'-GTCCAATGAGTCTGTATCAAGATC-3'; p75-1, 5'-ATGACTCGCGATTTCAAACCTGGA-3' and 5'-GATCTTCATCTCTTGTTTGCTCCAC-3'; P75-2, 5'-AAGCCACCCACAAACAAACTACCC-3' and 5'-CCCTCCTTTTCTCTTCTTTTCTCC-3'; rpP2, 5'-ATGCGCTACGTCGCC-3' and 5'-TTAATCAAAAAGGCCAAATCCCAT-3'.

Cell pellets were lyzed in ice-cold 10 mM Tris pH 7.5 with 0.5% NP-40, 1 mM MgCl₂, 1 mM DTT, 10 mM KCl, 0.4 M NaCl for 5 min, and centrifuged at 11,000× g for 5 min. 50 μg supernatant protein was separated by 12% SDS-polyacrylamide gel electrophoresis under reducing conditions and transferred to a nitrocellulose filter. The membrane was probed with anti-LEDGF/p75 polyclonal antibodies (kindly provided by C. A. Casiano) and the membrane was developed using the enhanced ECL detection method (Amersham Biosciences, Uppsala, Sweden).