Int J Hematol (2009) 89:173–187 DOI 10.1007/s12185-008-0242-9
ORIGINAL ARTICLE
Differential gene expression of bone marrow-derived CD34+ cells is associated with survival of patients suffering from myelodysplastic syndrome Wolf C. Prall Æ Akos Czibere Æ Franck Grall Æ Dimitrios Spentzos Æ Ulrich Steidl Æ Aristoteles Achilles Nikolaus Giagounidis Æ Andrea Kuendgen Æ Hasan Otu Æ Astrid Rong Æ Towia A. Libermann Æ Ulrich Germing Æ Norbert Gattermann Æ Rainer Haas Æ Manuel Aivado
Received: 7 July 2008 / Revised: 30 November 2008 / Accepted: 4 December 2008 / Published online: 20 January 2009 Ó The Japanese Society of Hematology 2009
Abstract One feature of the molecular pathology of myelodysplastic syndromes (MDS) is aberrant gene expression. Such aberrations may be related to patient survival, and may indicate to novel diagnostic and therapeutic targets. Therefore, we aimed at identifying aberrant gene expression that is associated with MDS and patient survival. Bone marrow-derived CD34? hematopoietic progenitor cells from six healthy persons and 16 patients with MDS were analyzed on cDNA macroarrays comprising 1,185 genes. Thereafter, our patients were followed-up for 54 months. We found differential expression of genes that were hitherto unrecognized in the context of MDS. Differential expression of 10 genes was confirmed by quantitative real-time RT-PCR. Hierarchical
W. C. Prall A. Czibere A. Kuendgen A. Rong U. Germing N. Gattermann R. Haas M. Aivado Department of Hematology, Oncology, and Clinical Immunology, Heinrich-Heine-University, Du¨sseldorf, Germany A. Czibere F. Grall D. Spentzos H. Otu T. A. Libermann M. Aivado Beth Israel Deaconess Medical Center and Harvard Medical School, Harvard Institutes of Medicine, Boston, MA, USA U. Steidl Department of Cell Biology, Albert Einstein Cancer Center, Albert Einstein College of Medicine, Bronx, NY, USA A. A. N. Giagounidis Department of Hematology, Oncology, and Clinical Immunology, St Johannes Hospital, Duisburg, Germany W. C. Prall (&) Experimental Surgery and Regenerative Medicine, Department of Surgery, Ludwig-Maximilians-University, Nussbaumstraße 20, 80336 Munich, Germany e-mail:
[email protected]
cluster analysis facilitated the separation of CD34? cells of normal donors from patients with MDS. More importantly, it also distinguished MDS-patients with short and long survival. Scrutinizing our cDNA macroarray data for genes that are associated with short survival, we found, among others, increased expression of six different genes that encode the proteasome subunits. On the other hand, the most differentially down-regulated gene was IEX-1, which encodes an antiapoptotic protein. We confirmed its decreased expression on RNA and protein level in an independent validation set of patient samples. The presented data broadens our notion about the molecular pathology of MDS and may lend itself to better identify patients with short survival. Furthermore, our findings may help to define new molecular targets for drug development and therapeutic approaches for patients with poor prognosis. Keywords cDNA array Gene expression MDS Survival IEX-1
1 Introduction Myelodysplastic syndromes (MDS) represent a heterogeneous group of clonal hematological stem cell disorders. In the defective stem cell, the finely tuned balance between self renewal, differentiation and apoptosis is disturbed. The bone marrow (BM) is usually hyper cellular due to enhanced proliferation of progenitor cells incapable of developing into mature blood cells [1, 2]. As a result, patients with MDS develop anemia are often associated with a varying degree of thrombocytopenia and leukocytopenia. In up to 23% of patients, the disease transforms into overt acute myeloid leukemia [3]. A broad variety of molecular changes like elevated Akt kinase phosphorylation [4], hyperactivation of the RAS
123
174
signaling pathway [5] and mutations of the PIG-A [6] or JAK2 gene [7] have been found in patients with MDS. Still, most of the molecular pathology of MDS is unknown. This lack of knowledge impedes the design of ‘‘targeted’’ therapeutic approaches such as imatinib therapy in patients with chronic myelomonocytic leukemia (CMML) who carry platelet-derived growth factor beta receptor fusion oncogenes [8]. It was one of the purpose of this study to identify molecular alterations that are pertinent to MDS. Such insights may serve to identify MDS as a candidate disease for testing investigational new drugs. Moreover, based on a patient follow-up of 54 months, we sought after transcriptional aberrations that may be associated with short survival. Such transcriptional aberrations may support the endeavor to better identify patients with short survival, and they may ultimately represent novel molecular targets for new treatment modalities of patients with an unfavorable prognosis.
2 Materials and methods 2.1 Samples and RNA preparation We obtained samples of normal BM from six persons (normal 1–6) who had a normal blood count and underwent surgical procedure (median age 84 years, range 76–87 years). An aliquot was used to prepare BM smears for a cytological examination of the BM. We also obtained BM samples from 16 patients with MDS (median age 67 years, range 18– 79 years). Later, a second independent set of 20 samples of CD34? cells were collected from peripheral blood (PB) for validation experiments. Among those 20 donors, there were 3 patients with cytopenia for reasons other than MDS (iron deficiency, osteomyelofibrosis), 6 healthy donors, and 11 patients with MDS. Patients’ characteristics including FAB-, WHO-classification, karyotypes, and IPSS scores are provided in Table 1. Highly purified CD34? hematopoietic progenitor cells were collected from mononuclear BM cells by 2-fold immunomagnetic separation, as previously described [9, 10]. RNA was extracted from CD34? cells with RNeasy Kit (Qiagen, Hilden, Germany). The study was approved by the institutional review board and patients gave their written informed consent. 2.2 cDNA synthesis and amplification cDNA first strand synthesis and amplification were performed as published elsewhere [9–12]. We incubated 300 ng RNA with 1 ll cDNA synthesis primer (10 lM), and 1 ll SMART II oligonucleotide (10 lM) at 70°C for 8 min. After short spinning, 2 min on ice, and 3 min at 42°C, we added a master mix consisting of 2 ll 59 first-strand buffer, 1 ll
123
W. C. Prall et al.
DTT (20 mM), 1 ll 509 dNTP mix (10 mM), and 0.5 ll MMLV RNase H- point mutant reverse transcriptase (Powerscript RT, BD Clontech, Palo Alto, CA). After incubation at 42°C for 1 h, cDNA was diluted with 40 ll TE buffer [10 mM Tris (pH 7.6), 1 mM EDTA], and 5 ll of this cDNA were subjected to a long-distance PCR protocol (BD Clontech, Palo Alto, CA). In brief, PCR was carried out with 5 ll 109 buffer, 1 ll 509 dNTP mix (10 mM), 1 ll PCR primer (10 lM), 1 ll advantage 2 polymerase mix, and 37 ll H2O. PCR conditions were 95°C for 1 min, followed by an amplification cycle with 95°C for 10 s, 65°C for 29 s, and 68°C for 3 min. The amplification cycle was repeated 15, 18, 21, and 24 times, and results were analyzed by agarose gel electrophoresis ensuring that cDNAs were obtained during the exponential phase of amplification. 2.3 cDNA macroarray hybridization We hybridized amplified cDNA with Atlas Human 1.2 I arrays (BD Clontech, Palo Alto, CA) as previously published [9, 10, 13–15]. In brief, after denaturing 500 ng of amplified cDNA, 1.5 ll CDS primer mix was added (BD Clontech, Palo Alto, CA), and samples were labelled by Klenow enzyme reaction mix supplemented with [a-32P]dATP (3,000 Ci/mmol; Amersham Pharmacia Biotech, Uppsala, Sweden). The probes were purified from unincorporated 32P-labeled nucleotides by column chromatography, and subjected to scintillation counting (LS 9000 Beckman Coulter, Beckman, Fullerton, CA). Macroarrays were prehybridized for 30 min at 68°C with 5 ml low-viscosity hybridization solution (ExpressHyb, BD Clontech, Palo Alto, CA), and 209 SSC. Next, purified probes were carefully added and hybridized at 68°C. After 18 h of incubation, a high stringency wash was performed. Macroarrays were then exposed to phosphorimager screens (FujiFilm, Du¨sseldorf, Germany), and kept for 36 h in a shielding lead box (Raytest, Straubenhardt, Germany) in order to minimize the influence of natural background radiation. 2.4 Image and data analysis Screens were scanned using a Phosphorimager (Fuji FLA3000, FujiFilm, Du¨sseldorf, Germany) controlled by the BASReader 3.01 Software (Raytest, Straubenhardt, Germany) at a resolution of 50 lm. Digital image analysis was performed with AIDA Image Analyzer 3.20 Software (Raytest, Straubenhardt, Germany), and subsequently, false positive signals were eliminated by visual inspection, as described [16]. We employed a standard global intensity-based normalization strategy, as recommended [12, 17, 18]. Gene expression data was normalized by determining the ratio of background-corrected dot intensities to the mean intensity of 20% highest expressed genes. Differential gene expression was analyzed by means of nearest neighborhood analysis
Differential gene expression in MDS
175
Table 1 Classification, clinical characteristics, and karyotypes of patients (a) with MDS who were analyzed by means of cDNA array analysis (b) from the independent validation set who were analyzed by means of quantitative real-time RT-PCR Patient No.
Age/sex
FAB
IPSS
WHO
Karyotype
WBC (/ll)
Hb (g/dl)
Platelets (/ll)
1
61/F
RA
Int-1
RCMD
46, XX, 13q-
1,400
10.6
17,000
2
68/F
RA
Int-1
5q-
46, XX, 5q-, 12p-
1,900
9.3
350,000
3
62/M
RA
Int-1
RCMD
47, XY, ?8
1,900
9.4
132,000
4
50/M
RA
Int-1
RCMD
46, XY, t(X; 1)
2,700
7.9
91,000
5
18/F
RA
Low
PRA
46, XX
5,300
7.6
542,000
6 7
51/M 51/F
RA RA
Low Int-1
RCMD RCMD
46, XY 46, XX, 5q-, 12p-
2,400 4,700
11.0 7.8
60,000 182,000
8
75/M
RAEB
High
RAEB II
46, XY, 7q-
9
79/M
RAEB-t
High
sAML
Complex
10
68/M
RAEB
Int-2
RAEB II
46, XY
11
70/F
RAEB-t
High
sAML
12
52/M
RAEB-t
High
sAML
13
72/M
RAEB-t
High
14
75/M
RAEB
15
72/M
16
(a)
800
5.4
21,000
1,800
9.4
153,000
500
10.2
68,000
46, XX, 15q-
9,800
10.8
32,000
Complex
1,200
8.6
17,000
sAML
Complex
1,700
6.9
32,000
Int-1
RAEB I
46, XY
1,200
8.4
70,000
sAML/MDS
–
sAML
Complex
2,700
8.1
275,000
76/M
sAML/MDS
–
sAML
46, XY
28,000
7.2
5,000
101
56/F
RA
Low
RCMD
46, XX
3,600
9.8
344,000
102
69/M
RA
Low
RCMD
46, XY
3,400
10.0
70,000
103
71/F
RA
–
RCMD
46, XX
NA
104 105
64/M 79/F
RA RA
Int-1 Low
RCMD RA
47, XY, 5q-, 21-, ?8 46, XX
106
52/F
RA
Int-1
RCMD
107
70/M
RA
Low
RA
108
70/F
RAEB
Int-1
RAEB I
109
63/F
RAEB
Int-2
RCMD
110
70/M
RAEB
–
RAEB I
111
82/F
RAEB
Int-2
RSCMD
(b)
NA
NA
7,000 3,800
7.0 11.0
48,000 138,000
46, XX, t(1; 7)
3,200
12.0
57,000
46, XY
7,400
9.7
254,000
46, XX, 2-, del(5q)
4,600
10.0
279,000
45, XX, -7
3,800
9.4
26,000
ND
3,300
13.1
64,000
45, XX, -7
2,500
8.7
54,000
FAB French-American-British cooperative group classification, IPSS international prognostic scoring system, int intermediate, WHO World Health Organization classification, WBC white blood count, Hb haemoglobin, NA not available, ND not done, del deletion
using the software GeneCluster 2.1.6 (Whitehead Institute, Cambridge, MA). Each gene was scored with the signal-tonoise metric [19–22]. The signal-to-noise score equals: (lclass0 - lclass1/rclass0 ? rclass1) where l is the mean, and r the standard deviation for a given gene in all samples of one class. P values were generated via permutation testing with 20,000 random permutations of the class labels. Finally, gene expression data were clustered using the software dChip [23]. For hierarchical clustering, we used the 1-correlation distance metric and an average linkage algorithm. 2.5 Quantitative real-time RT-PCR Validation experiments were performed as previously published [10]. In brief, we chose 10 genes, and used LightCycler Probe Design Software Version 1.0 (Roche,
Mannheim, Germany) to design the following primer pairs: FLT3 F 50 -AAA CCT CAA GTG CTC G-30 , R 50 -GGA ACC CTT TTA TGG CT-30 ; ERCC3 F 50 -CCA TAC CAG CCA AGA T-30 , R 50 -ATT CAG GAG ACA TAG GG-30 ; NDPKB F 50 -AGC ACT ACA TTG ACC T-30 , R 50 -TTT TAC TGA ATC ACT GCC-30 ; GST-3 F 50 -CAA TAC CAT CCT GCG T-30 , R 50 -CCC ACA ATG AAG GTC T-30 ; IEX-1 F 50 -GAG TGG TGA GTA TCG C-30 , R 50 -GCT CCG AAG TCA GAT TA-30 ; EPA-1 F 50 -GTT CCA AGC CTT AGG G-30 , R 50 CCA ACA GTG TAG GTC T-30 ; MAD F 50 -GGA GAA GAA TAG ACG GGC TCA T-30 , R 50 -CTG AAG CTG GTC GAT TTG GT-30 ; Cath. L F 50 -TCA AAA CCG TAA GCC C-30 , R 50 -GTG AGA TAA GCC TCC C-30 ; TTP F 50 -AGA CTG AGC TAT GTC G-30 , R 50 -AGG GTT GTG GAT GAA G-30 and MCL-1 F 50 -GCC TTT GTG GCT AAA C-30 , R 50 -CTA CTC CAG CAA CAC C-30 . These were different
123
176
from the primers used for the synthesis of cDNA probes of the hybridization experiments, thus providing additional validation. The only results from normal and MDS samples that were measured in the same run were compared with each other. All analyses were done in duplicate. The procedure was performed using the LightCycler-FastStart Taq DNA Master SYBR green kit (Roche, Mannheim, Germany). One microliter of each primer (10 lM), 2.4 ll MgCl2 (4 mM), and 2 ll of the supplied enzyme mix containing reaction buffer, FastStart Taq DNA polymerase, and SYBR green I dye were added to a final volume of 20 ll. PCR was carried out as follows: 95°C for 8 min, then 45 cycles comprising 95°C for 15 s, 62°C for 5 s, and 72°C for 20 s. After RT-PCR, a melting curve analysis, and gel electrophoresis of every sample were performed in order to ensure specificity of the PCR. Crossing point (CP) values were determined using LightCycler 3.5 software and second derivative maximum method. The difference between the CPs of target gene and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (DCP) represents the relative transcription level of the target gene. Assuming an amplification efficacy of 1.9, a decrease of DCP by one cycle reflects a 1.9-fold increase of the target mRNA. 2.6 Western blot analysis CD34? cells were lyzed in a buffer consisting of 20 mM Tris pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM b-glycerolphosphate, 1 mM Na3VO4, 1 lg/ml leupeptin and 1 mM PMSF. Hundred microgram of protein were electrophoresed in a 10% acrylamide-SDS gel. Proteins were electroblotted onto a PVDF membrane in an electrophoresis buffer consisting of 50 mM Tris–base, 20% methanol, 40 mM glycine. Membranes were incubated in 5% non-fat dry milk in TBST (60 mM Tris–base, 120 mM NaCl, 0.2% Tween 20) for 1 h. Blots were probed with primary antibody overnight at 4°C in 2% BSA in TBST, and then incubated with a horseradish peroxidase-conjugate secondary antibody (Cell Signaling, Beverly, MA) in 5% dry milk in TBST for 1 h at room temperature. Bound antibodies were detected by chemoluminescence with ECL detection reagents (Amersham Pharmacia Biotech, Piscataway, NJ) and visualized by autoradiography. The primary antibody used for western blot analysis was anti-IEX-1 (C-20; Santa Cruz Biotechnology, Santa Cruz, CA). 2.7 Experimental quality assessment To assess the purity of BM-derived CD34? cells, 2 9 105 collected cells from three patients (MDS 5, 8, and 16) were
123
W. C. Prall et al.
analyzed by flow cytometry (FACScan, Becton Dickinson, Heidelberg, Germany). After incubation with a fluorescein isothiocyanate-conjugated (FITC) monoclonal CD34-antibody (clone 8G12, Becton Dickinson, Heidelberg, Germany), and an isotype-identical monoclonal FITC-labeled IgG1antibody (Becton Dickinson, Heidelberg, Germany) as control, we observed a purity of CD34? cells varying between 95 and 99% (Fig. 1a). Before hybridizing the samples to cDNA macroarrays, we checked for contaminating DNA and for sufficient length of amplified cDNA. To this end, a low-abundant ‘‘housekeeping’’ gene, p619, was amplified with five PCR primer pairs complementary to five different sites within the p619 gene, as described elsewhere (Inspector Kit, Sigma-Aldrich, Darmstadt, Germany). These primer pairs amplify sequences, which are located in the intervals of *1–2 kb along the 15 kb coding sequence of the p619 gene. The resulting PCR products are thus an indicative of the length of amplified cDNA. b-actin and GAPDH, both high-abundance ‘‘housekeeping’’ genes, were also PCR amplified and visualized together with the p619 gene PCR products using ethidium bromide stained 1.5% agarose gels. b-actin primers spanned one intron, thus producing a second PCR product if contaminating genomic DNA was present. In the present study, PCR amplification of the p619 gene showed complete cDNA amplification with sufficient cDNA length for hybridization experiments of all samples (Fig. 1b). To assess reproducibility, we split three RNA samples into two aliquots each. Aliquots were separately subjected to the entire experimental procedure. The resulting gene expression values were transformed logarithmically, plotted against each other, and Pearson’s correlation coefficients were calculated. These quality control analyses showed high reproducibility as reflected by Pearson’s correlation coefficient of 0.985 for intra-experimental control (data not shown), and 0.917–0.970 for inter-experimental control (Fig. 1c). In sum, our quality assessment confirms previous reports on the feasibility of such cDNA macroarray analyses [9, 10, 13–15].
3 Results 3.1 Differential gene expression between CD34? cells from patients with MDS and normal controls Looking at median expression levels of all examined 1,185 genes, we observed similar percentages of genes with decreased (52.7%) and increased (47.3%) expression in MDS as compared to normal controls. In this study, we present the most differentially expressed genes as defined by P \ 0.01 and C2-fold increase (Table 2) or decrease
Differential gene expression in MDS
177
Fig. 1 Experimental quality assessment. a Representative dot plots from flow cytometry confirming purity of isolated CD34? cells to be C95% (patient MDS 16). The percentage of cells above the threshold intensity is given in the upper right corner. CD34? cells were incubated with an isotype-identical monoclonal FITC-labeled IgG1antibody or a FITC-labeled monoclonal CD34-antibody. b Representative agarose gel electrophoresis showing RT-PCR products from the amplification of GAPDH, b-actin, and five different p619 primer pairs in order to check for contaminating DNA and cDNA length (patient
MDS 7). Lane M 100 bp-ladder; Lane 1 b-actin; Lane 2 GAPDH; Lane 3 H2O; Lanes 4 to 8 different primer pairs for p619; lane 4 p619, nucleotides 12,984–13,892; lane 5 p619, nucleotides 9,406–10,202; lane 6 p619, nucleotides 8,290–8,998; lane 7 p619, nucleotides 5194– 5802; lane 8 p619, nucleotides 4,191–4,690. c Scatter plot and Spearman’s correlation coefficient from quality control experiments to assess inter-experimental variation. Two RNA aliquots from the same sample were subjected to cDNA macroarray analysis on different days
(Table 3) of median expression in CD34? cells from patients with MDS. In MDS-CD34? cells, we found significantly increased expression of genes that are involved in proliferation such as heat shock 90-kDa protein A (HSP90A), myeloproliferative leukemia oncogene (MPL), fms-related tyrosine kinase 3 (FLT3), proliferating cell nuclear antigen (PCNA), MCM2 DNA replication licensing factor (MCM2), cell division cycle 25B (CDC25B), and macrophage colony stimulating factor 1 (M-CSF-1). Likewise, there was an increased expression of pro-apoptotic genes [c-jun kinase 2 (c-JNK2), death-associated protein kinase 1 (DAPK1), caspase 2, caspase 3, n-SMase activation associated factor (NSAAF)], DNA repair-related genes [excision repair cross-complementing 3 gene (ERCC3), poly (ADP-ribose)polymerase (PARP), mutL (E. coli) homolog 1 (mutL-h1)], and detoxification-/stress-related genes [paraoxonase 1, glutathione S-transferase 12 (GST12), glutathione S-transferase 3 (GST-3)]. In contrast, MDS-CD34? cells showed a significantly decreased expression of seven anti-apoptotic genes [immediate-early response gene 1 (IEX-1), growth arrest
and DNA-damage-inducible beta (GADD45-b), serum/ glucocorticoid regulated kinase (SGRK), apoptosis inhibitor 2 (AI2), myeloid cell leukemia gene 1 (MCL-1), BCL2like 1, and endothelial TEK tyrosine kinase]. We also found decreased expression of 12 genes which encodes cytokines and inflammation-related proteins such as interleukin-8 (IL-8), formyl peptide receptor-like 1 (FPR-like 1), CXC chemokine ligand 2 (CXCL2), monocyte chemotactic protein 1 (MCP-1), interleukin-1 receptor type II (IL-1-R2), CD87, CD86 antigen, TNF-receptor 1B (TNFR-1B), CC chemokine receptor 1 (CCR-1), interferon gamma receptor 2 (IF-g-R2), interleukin-1 beta (IL-1-b), and interleukin-1 alpha (IL-1-a). Eventually, we also observed decreased expression of seven growth factors/receptors [vascular endothelial growth factor receptor 2 (VEGFR2), erythroid potentiating activity factor 1 (EPA1), fms-related tyrosine kinase 3 ligand (FLT3-L), acidic fibroblast growth factor 1 (FGF-1), vascular endothelial growth factor (VEGF), GM-CSF-2 receptor alpha, and vascular endothelial growth factor receptor 3 (VEGFR3)]. Our complete gene expression results are available as supplemental online information.
123
178
W. C. Prall et al.
Table 2 Genes showing increased expression in MDS-CD34? cells, ranked by fold change Gene
Factora
Scoreb
Pc
GenBank
Paraoxonase 1
9.7
0.82
\0.001
M63012
Serine protease inhibitor, Kazal type 2
7.3
1.03
\0.00001
M91438
6691
N-myc oncogene
7.0
0.89
\0.0001
M13228
4613
Heat shock 90-kDa protein A (HSP90A)
5.2
1.07
\0.00001
X07270
Dihydropyridine-sensitive CACNLB3
4.4
0.94
\0.00001
U07139
Integrin, alpha 6
4.2
1.16
\0.001
X53586
Protein kinase C, theta
4.0
1.27
\0.001
L07032
5588
Excision repair cross-complementing 3 gene (ERCC3)
3.9 (5.5)
1.69
\0.01
M31899
2071
Myeloproliferative leukemia oncogene (MPL)
3.8
1.04
\0.00001
U68162
4352
Locus link 5444
3655
Fms-related tyrosine kinase 3 (FLT3)
3.8 (8.4)
1.01
\0.00001
U02687
2322
Translin c-Jun kinase 2 (c-JNK2)
3.1 3.1
1.08 0.99
\0.00001 \0.00001
X78627 L31951
7247 5601
Death-associated protein kinase 1 (DAPK1)
3.1
0.81
\0.001
X76104
1612
Caspase 2
2.8
0.99
\0.00001
U13021
835
Tuberous sclerosis 2
2.8
0.87
\0.0001
X75621
7249
YY1 transcription factor
2.8
1.44
\0.001
M76541
7528
Proliferating cell nuclear antigen (PCNA)
2.8
0.8
\0.001
M15796
5111
Early response gene (ERG)
2.8
1.06
\0.00001
M21535
2078
Glutathione S-transferase 12 (GST-12)
2.7
1.38
\0.001
J03746
4257
Caspase 3
2.6
1.38
\0.001
U13737
836
Proteasome (prosome, macropain) subunit, alpha type, 2
2.6
0.88
\0.0001
D00760
5683
Bruton agammaglobulinemia tyrosine kinase
2.6
0.91
\0.00001
X58957
695
MCM2 DNA replication licensing factor (MCM2)
2.6
1.06
\0.00001
D21063
Guanine nucleotide binding protein, beta 5
2.5
0.89
\0.0001
AF017656
Poly (ADP-ribose)polymerase (PARP)
2.5
1.1
\0.0001
M18112
142
10681
Eph tyrosine kinase 1
2.4
1.14
\0.0001
M18391
2041
Janus kinase 2 (JAK2) Nucleoside diphosphate kinase B (NDPK-B)
2.4 2.4 (4.0)
1.16 1.14
\0.001 \0.0001
AF005216 L16785
3717 4831
Nuclear transcription factor, X-box binding 1
2.4
0.83
\0.001
U15306
4799
DNA polymerase gamma (POLG)
2.4
1.02
\0.00001
X98093
5428
Guanylate cyclase 1, soluble, beta 3
2.3
0.82
\0.001
X66533
2983
n-SMase activation associated factor (NSAAF)
2.3
0.86
\0.0001
X96586
8439
Zinc finger protein 162
2.2
0.93
\0.00001
D26120
7536
mutL (E. coli) homolog 1 (mutL-h1)
2.2
1.19
\0.001
U07418
4292
Acyl-Coenzyme A binding protein
2.2
0.98
\0.00001
M14200
1622
Cell division cycle 25B (CDC25B)
2.2
0.99
\0.00001
M81934
994
Tight junction protein 1 (zona occludens 1)
2.2
1.01
\0.00001
L14837
7082
Sarcoma amplified sequence
2.1
1.01
\0.00001
U01160
6302
Zinc finger protein 161
2.1
0.97
\0.00001
D28118
7716
Glutathione S-transferase 3 (GST-3)
2.1 (3.8)
1.37
\0.001
X15480
2950
Macrophage colony stimulating factor 1 (M-CSF-1)
2.0
1.11
\0.0001
M37435
1435
Scores and P values were determined by GeneCluster software. The values in brackets in the column ‘‘factor’’ were derived from the median expression value as determined by quantitative RT-PCR a
The factor was derived from the median expression values as determined by cDNA array analysis
b
The score represents the signal-to-noise score as determined by GeneCluster software
c
P values were obtained from the permutation test as implemented in the GeneCluster software
123
Differential gene expression in MDS
179
Table 3 Genes showing decreased expression in MDS-CD34? cells, ranked by fold change Gene
Factora
Scoreb
Pc
GenBank
Locus link
Immediate-early response gene (IEX-1)
37.5 (11.0)
2.06
\0.00001
AF071596
8870
Serine (or cysteine) proteinase inhibitor, clade A, member 1
25.0
1.34
\0.00001
X02920
5265
Interleukin-8 (IL-8)
24.0
1.57
\0.00001
Y00787
3576
Cathepsin L
21.9 (7.4)
1.75
\0.00001
X12451
1514
Formyl peptide receptor-like 1 (FPR-like 1)
18.3
1.02
\0.00001
L36645
2358
Solute carrier family 2, member 3
15.8
1.5
\0.00001
M20681
6515
CXC chemokine ligand 2 (CXCL2)
13.9
2.31
\0.00001
X53799
2920
Vascular endothelial growth factor receptor 2 (VEGFR2)
13.7
1.27
\0.00001
X61656
3791
Monocyte chemotactic protein 1 (MCP-1)
13.4
1.41
\0.00001
M24545
6347
Interleukin-1 receptor, type II (IL-1-R2)
12.2
1.5
\0.00001
X59770
7850
CD87 Serpine peptidase, clade B, member 2
11.9 10.9
1.25 1.61
\0.00001 \0.00001
U08839 M18082
5329 5055
Erythroid potentiating activity factor 1 (EPA-1)
10.0 (8.7)
Matrix metalloproteinase 11 (MMP-11)
9.8
1.38
\0.00001
X03124
7076
1.54
\0.00001
X57766
4320
Myelin oligodendrocyte glycoprotein
8.6
1.35
\0.00001
U18840
4340
G-protein coupled receptor 109 B
8.3
1.79
\0.00001
D10923
8843
MAX dimerization protein (MAD)
7.7 (8.4)
1.54
\0.00001
L06895
4084
Growth arrest and DNA-damage-inducible, beta (GADD45-b)
7.0
1.83
\0.00001
AF078077
4616
EGF-response factor 1
6.7
1.46
\0.00001
X79067
677
Tissue inhibitor of metalloproteinase 2 (TIMP-2)
6.6
1.41
\0.00001
J05593
7077
Fructose-1,6-bisphosphatase 1
6.2
1.36
\0.00001
M19922
2203
CD86 antigen
6.2
1.12
\0.00001
L25259
942
Heme oxygenase (decycling) 1
5.6
1.16
\0.00001
X06985
3162
SKI-like
5.6
1.1
\0.00001
X15219
6498
Fms-related tyrosine kinase 3 ligand (FLT3-L)
5.2
1.46
\0.00001
U04806
2323
Serum/glucocorticoid regulated kinase (SGRK)
5.2
1.85
\0.00001
AJ000512
6446
TNF-receptor 1B (TNF-R-1B) G-protein, beta polypeptide 1
5.2 5.0
1.18 1.42
\0.00001 \0.00001
M32315 M36430
7133 2782
Acidic fibroblast growth factor 1 (FGF-1)
5.0
1.34
\0.00001
X51943
2246
CC chemokine receptor 1 (CCR-1)
4.7
1.14
\0.00001
D10925
1230
Apoptosis inhibitor 2 (AI2)
4.7
1.46
\0.00001
U45878
330
Tristetraproline (TTP)
4.6 (7.2)
1.56
\0.00001
M92843
7538 7422
Vascular endothelial growth factor (VEGF)
4.4
2.03
\0.00001
M32977
GM-CSF-2 receptor, alpha
4.3
1
\0.00001
X17648
1438
Polycystic kidney disease 2 (autosomal dominant)
4.2
1.92
\0.00001
U50928
5311
FGR oncogene
4.1
1.22
\0.00001
M19722
2268
Glucagon-like peptide
3.8
1.69
\0.00001
J04040
2641
Human embryo kinase 2
3.6
1.04
\0.00001
X75208
2049
Ras-related associated with diabetes
3.6
1.24
\0.00001
L24564
6236
Interferon gamma receptor 2 (IF-g-R2)
3.1
1.32
\0.00001
U05875
3460
Myeloid cell leukemia gene 1 (MCL-1)
3.1 (2.8)
1.57
\0.00001
L08246
4170
Cathepsin H Interleukin-1, beta (IL-1-b)
3.0 2.7
1.44 1.12
\0.00001 \0.00001
X07549 K02770
1512 3553
Phospholipase C, beta 2
2.6
0.98
\0.00001
M95678
5330
FOS-like antigen-1
2.5
1.07
\0.00001
X16707
8061
Interleukin-1, alpha (IL-1a)
2.5
1.15
\0.00001
X02851
3552
BCL2-like 1
2.5
1.32
\0.00001
Z23115
598
PTK7 protein tyrosine kinase 7
2.2
1.01
\0.00001
U33635
5754
123
180
W. C. Prall et al.
Table 3 continued Gene
Factora
Scoreb
Pc
GenBank
Locus link
TEK tyrosine kinase, endothelial
2.2
1.03
\0.00001
L06139
7010
Vascular endothelial growth factor receptor 3 (VEGFR3)
2.1
1.14
\0.00001
X68203
2324
Scores and P values were determined by GeneCluster software. The values in brackets in the column ‘‘factor’’ were derived from the median expression value as determined by quantitative RT-PCR a
The factor was derived from the median expression values as determined by cDNA array analysis
b
The score represents the signal-to-noise score as determined by GeneCluster software P values were obtained from the permutation test as implemented in the GeneCluster software
c
3.2 Validation of selected genes We used quantitative real-time RT-PCR to confirm differential expression of 10 genes as determined by cDNA macroarray analysis. Genes were selected due to a prominent-fold change, statistical significance, signal-to-noise ratio or biological interest. Quantitative RT-PCR analyses were performed with the same patient samples that had been subjected to cDNA macroarrays, but using different primer sequences than those present on the cDNA macroarray platform. In CD34? cells from patients with MDS, we confirmed increased gene expression of FLT3 (8.4-fold), ERCC3 (5.5-fold), nucleoside diphosphate kinase B (NDPK-B; 4.0fold), and GST-3 (3.8-fold) (Table 2). cDNA macroarray analyses had suggested that matrix metalloproteinase-11 (MMP-11) is decreased in MDS, whereas we did not find significantly different expression levels of MMP-11 by quantitative real-time RT-PCR. In contrast, we successfully validated the decreased expression of IEX-1 (11.0fold), EPA-1 (8.7-fold), MAX dimerization protein (MAD; 8.4-fold), cathepsin L (7.4-fold), tristetraproline (TTP; 7.2fold), and MCL1 (2.8-fold) (Table 3). 3.3 Hierarchical clustering and survival We used the complete cDNA macroarray expression values to build a dendrogram including all samples under investigation (Fig. 2). Unsupervised hierarchical clustering showed a clear separation of normal samples from MDS. The MDS cluster consisted of two branches, each including seven and nine patients with MDS, respectively. We examined whether this segregation of MDS-patients into two groups was associated with leukocyte or platelet count, hemoglobin level, age, gender, MDS-type according to FAB or WHO, IPSS, karyotype or survival. To this end, we determined correlation coefficients and found ‘‘survival’’ to be the parameter with the strongest statistical association with the segregation of MDS-patients into two subclusters (Spearman = 0.697, P = 0.003). The WHO type was less significantly associated (Spearman = 0.537, P = 0.032),
123
Fig. 2 The entire cDNA macroarray data distinguishes normal CD34? cells from MDS-CD34? cells as well as CD34? cells from patients with short and long survival. The dendrogram was derived from hierarchical cluster analysis using the 1-correlation distance metric and an average linkage algorithm. Survival is represented by bars, stars indicate patient that are still alive
and the FAB type did not reach statistical significance (Spearman = 0.492, P = 0.053). The median survival of patients in one cluster was 8 months (range 1–11 months) in comparison to 48 months (range 5–54 months) for patients within the second cluster (Fig. 2). This difference was statistically significant as determined by Log-Rank test (P = 0.0016). 3.4 Gene expression of CD34? cells from MDS-patients with short survival Given the association between cDNA macroarray data and patients’ survival, we repeated nearest neighborhood analysis, again using the entire cDNA macroarray results. We compared those patients from the ‘‘short survival’’ cluster with normal donors in order to identify genes that are associated with MDS having unfavorable prognosis. Again considering only differentially expressed genes as defined by P \ 0.01 and C2-fold change, we found 42 and
Differential gene expression in MDS
181
Table 4 Genes showing increased expression in CD34? cells of patients from the ‘‘short survival’’ cluster, as compared with normal donors Gene
Factora
Scoreb
Pc
GenBank
Locus link
Serine protease inhibitor, Kazal type, 2
8.65
1.06
\0.0005
M91438
Cyclin D2
6.24
1.65
\0.0005
D13639
Cyclin E2
5.72
1.06
\0.0005
AF091433
Dihydropyridine-sensitive L-type calcium channel beta-3
5.50
1.11
\0.0005
U07139
Myeloproliferative leukemia oncogene
4.48
1.03
\0.0005
U68162
4352
Fms-related tyrosine kinase 3 (FLT3)
4.38
1.39
\0.0005
U02687
2322
Integrin, alpha 6
4.33
1.04
\0.0005
X53586
3655
Caspase 2
4.33
1.88
\0.001
U13021
835
Budding uninhibited by benzimidazoles 1 (yeast homolog)
4.12
1.14
\0.0005
AF053305
699
Protein kinase C, theta
4.10
1.12
\0.0005
L07032
5588
Excision repair cross-complementing 3 gene (ERCC3) PR domain containing 2, with ZNF domain
4.09 4.01
1.64 1.18
\0.0005 \0.0005
M31899 D45132
2071 7799
6691 894 9134
Transcriptional intermediary factor 1
3.90
1.04
\0.0005
AF009353
8805
Translin
3.79
1.31
\0.0005
X78627
7247
Early response gene (EGR)
3.69
1.32
\0.0005
M21535
2078
Proteasome (prosome, macropain) subunit, alpha type, 2
3.47
1.15
\0.0005
D00760
5683
MCM2 DNA replication licensing factor (MCM2)
3.34
1.22
\0.0005
D21063
Poly (ADP-ribose) polymerase (PARP)
3.33
1.58
\0.0005
M18112
142
Ligase IV, DNA, ATP-dependent
3.30
1.14
\0.0005
X83441
3981
Bruton agammaglobulinemia tyrosine kinase
3.23
1.23
\0.0005
X58957
695
Proteasome (prosome, macropain) subunit, alpha type, 1
3.21
1.15
\0.0005
D00759
5682
Protein kinase, cAMP-dependent, catalytic, beta
3.20
1.29
\0.0005
M34181
5567
N-SMase activation associated factor (NSAAF)
3.10
1.17
\0.0005
X96586
8439
Nuclear transcription factor, X-box binding 1
3.05
1.14
\0.0005
U15306
4799
Minichromosome maintenance deficient (S. cerevisiae) 4
3.01
1.41
\0.0005
X74794
4173
Guanylate cyclase 1, soluble, beta 3
2.94
1.15
\0.0005
X66533
2983
YY1 transcription factor Mitochondrial transcription factor 1-like
2.88 2.77
1.76 1.17
\0.0005 \0.0005
M76541 M62810
7528 6930
Janus kinase 2 (JAK2)
2.67
1.52
\0.0005
AF005216
3717
Adenylate cyclase 7
2.57
1.07
\0.0005
D25538
113
Minichromosome maintenance deficient (mis5, S. pombe) 6
2.51
1.2
\0.0005
D84557
4175
Glutathione S-transferase 12
2.50
1.37
\0.0005
J03746
4257 5728
Mutated in multiple advanced cancers 1
2.46
1.01
\0.0005
U92436
Microsomal glutathione S-transferase 2
2.40
1.23
\0.0005
U77604
4258
RNA binding motif protein 5
2.38
1.28
\0.0005
U23946
10181
Tripartite motif-containing 22
2.37
1.08
\0.0005
X82200
10346
Histone deacetylase 1
2.35
1.39
\0.0005
D50405
3065
CCAAT-box-binding transcription factor
2.34
1.08
\0.0005
M37197
10153
Ligase III, DNA, ATP-dependent
2.22
1.05
\0.0005
X84740
3980
Zinc finger protein 161
2.20
1.41
\0.0005
D28118
7716
Proteasome (prosome, macropain) subunit, alpha type, 3
2.20
1.02
\0.0005
D00762
5684
NCK adaptor protein 1
2.05
1.37
\0.0005
X17576
4690
Genes were ranked according to the fold change, scores and P values were determined by GeneCluster software a
The factor was derived from the median expression values as determined by cDNA array analysis
b
The score represents the signal-to-noise score as determined by GeneCluster software
c
P values were obtained from the permutation test as implemented in the GeneCluster software
123
182
W. C. Prall et al.
Table 5 Genes showing decreased expression in CD34? cells of patients from the ‘‘short survival’’ cluster as compared with normal donors Gene
Factora
Scoreb
Pc
GenBank
Locus link
Immediate-early response gene 1 (IEX-1)
58.18
2.43
\0.0005
AF071596
8870
Monocyte chemotactic protein 1
51.73
1.62
\0.0005
M24545
6347
Serine/cysteine proteinase inhibitor, clade A, member 1
42.50
2.02
\0.0005
X02920
5265
Cathepsin L
29.89
1.73
\0.0005
X12451
1514
CD87
25.18
1.48
\0.0005
U08839
5329
Solute carrier family 2, member 3
23.83
1.98
\0.0005
M20681
6515
Interleukin-8 (IL-8)
23.62
1.55
\0.0005
Y00787
3576
Formyl peptide receptor-like 1 (FPR-like 1)
23.41
1.23
\0.0005
L36645
2358
Interleukin-1 receptor, type II (IL-1-R2)
22.34
2.06
\0.0005
X59770
7850
CXC chemokine ligand 2 (CXCL2)
16.61
2.79
\0.0005
X53799
2920
Vascular endothelial growth factor receptor 2 (VEGFR2) Matrix metalloproteinase 11 (MMP-11)
14.70 13.81
1.31 2.13
\0.0005 \0.0005
X61656 X57766
3791 4320
Small inducible cytokine A4
13.63
1.28
\0.0005
J04130
6351
Serine/cysteine proteinase inhibitor, clade B, member 2
12.36
1.92
\0.0005
M18082
5055
Erythroid potentiating activity factor 1 (EPA-1)
12.31
1.85
\0.0005
X03124
7076
G-protein coupled receptor 109 B
11.94
1.88
\0.0005
D10923
8843
Myelin oligodendrocyte glycoprotein
10.64
1.42
\0.0005
U18840
4340
MAX dimerization protein (MAD)
9.11
1.46
\0.0005
L06895
4084
Growth arrest and DNA-damage-inducible beta (GADD45-b)
8.81
2.3
\0.0005
AF078077
4616
Interleukin-11 (IL-11)
8.55
1.69
\0.0005
M57765
3589
Acidic fibroblast growth factor 1
8.39
2.43
\0.0005
X51943
2246
Cyclin-dependent kinase inhibitor 1 (CDKN1A)
8.28
2.19
\0.0005
U09579
Solute carrier family 6, member 6
8.09
1.32
\0.0005
Z18956
6533
CD86 antigen
8.00
1.81
\0.0005
L25259
942
EGF-response factor 1
7.40
1.25
\0.0005
X79067
677
Peripheral myelin protein 22
7.36
1.68
\0.0005
D11428
5376
Heme oxygenase (decycling) 1 Fms-related tyrosine kinase 3 ligand (FLT3L)
7.32 7.19
1.26 1.98
\0.0005 \0.0005
X06985 U04806
3162 2323
CC chemokine receptor 1 (CCR-1)
7.05
1.85
\0.0005
D10925
1230
Tumor necrosis factor receptor 1B (TNF-R1B)
6.97
2.61
\0.0005
M32315
7133
Tissue inhibitor of metalloproteinase 2
6.62
1.87
\0.0005
J05593
7077
Glucagons
5.94
2.75
\0.0005
J04040
2641
Fructose-1,6-bisphosphatase 1
5.90
1.3
\0.0005
M19922
2203
Guanine nucleotide binding protein, beta polypeptide 1
5.81
1.51
\0.0005
M36430
2782
Tumor necrosis factor ligand, member 6
5.72
1.34
\0.0005
D38122
356
Serum/glucocorticoid regulated kinase (SGRK)
5.68
1.96
\0.0005
AJ000512
Apoptosis inhibitor 2 (AI2)
5.31
1.42
\0.0005
U45878
330
Tristetraproline (TTP)
5.23
1.89
\0.0005
M92843
7538
Vascular endothelial growth factor (VEGF)
4.84
2.62
\0.0005
M32977
7422
Polycystic kidney disease 2 (autosomal dominant)
4.17
2.07
\0.0005
U50928
5311
Ras-related associated with diabetes
4.04
1.33
\0.0005
L24564
6236
Inhibitor of DNA binding 2 FGR oncogene
3.75 3.43
1.38 1.54
\0.0005 \0.0005
M97796 M19722
3398 2268
6446
Mal, T cell differentiation protein
3.32
1.23
\0.0005
M15800
4118
Cathepsin H
3.14
2.02
\0.0005
X07549
1512
Interferon gamma receptor 2 (IFN-g-R2)
3.08
1.41
\0.0005
U05875
3460
Myeloid cell leukemia gene 1 (MCL-1)
2.93
1.51
\0.0005
L08246
4170
Interleukin-1, alpha (IL-1a)
2.69
1.66
\0.0005
X02851
3552
123
Differential gene expression in MDS
183
Table 5 continued Gene
Factora
Scoreb
Pc
GenBank
BCL2-like 1
2.29
1.24
\0.0005
Z23115
Interleukin-6 receptor alpha subunit precursor (IL-6-R a)
2.28
1.23
\0.0005
M20566
Locus link 598
Genes were ranked according to the fold change, scores and P values were determined by GeneCluster software a
The factor was derived from the median expression values as determined by cDNA array analysis
b
The score represents the signal-to-noise score as determined by GeneCluster software
c
P values were obtained from the permutation test as implemented in the GeneCluster software
50 genes with increased and decreased expression, respectively. Less dramatic changes were observed among genes with significantly increased expression in CD34? cells from patients with short survival. Most prominent increases were observed for genes that are associated with cell cycle and proliferation such as cyclin D2 (6.2-fold), cyclin E2 (5.7fold), MPL (4.5-fold), and FLT3 (4.4-fold). Interestingly, among those genes with increased expression, three genes emerged that encode the proteasome subunits alpha type 1, 2, and 3. These genes had 3.2-, 3.5-, and 2.2-fold higher median expression levels in MDS-CD34? cells from the ‘‘short survival’’ group. This finding prompted the question about the expression levels of those proteasome-related genes that had not been identified to be differentially expressed according to nearest neighborhood analysis. We, therefore, retrieved the expression data from the remaining proteasome genes on our cDNA macroarrays, and also found increased expression of proteasome activator subunit 2 (PA28 beta) (2.0-fold), proteasome 26S subunit ATPase 3 (1.9-fold), and proteasome subunit beta type 1 (3.1-fold), albeit these changes did not reach statistical significance. Among genes with significantly decreased expression in CD34? cells from patients with short survival, we found that 20% of those genes encode cytokines and cytokine receptors. IEX-1 was the most differentially expressed gene in this analysis (58.2-fold), again, and was therefore, further scrutinized. The entire list of genes with differential expression in CD34? cells from patients with MDS who belong to the ‘‘short survival’’ cluster can be viewed in Tables 4 and 5. 3.5 IEX-1 cDNA macroarray analyses revealed that IEX-1 had a 37.5fold lower expression in MDS-CD34? cells overall, and a 58.2-fold lower expression in CD34? cells from MDSpatients with short survival as compared to normal CD34? cells. This prompted our particular interest in IEX-1, and we, therefore, examined the differential expression of IEX-1 in a prospectively collected independent validation set of samples from patients with MDS (n = 11), patients
with cytopenia for reasons other than MDS (n = 3), and healthy donors (n = 6) (Table 1b). Since IEX-1 expression does not significantly differ between PB-CD34? and BM-CD34? cells [10], we used PB-CD34? cells for this validation experiment. Quantitative real-time RT-PCR analyses on CD34? cell samples from our independent validation set revealed a 4.7-fold lower expression of IEX1 in patients with MDS-type refractory anemia (RA) as compared to patients with cytopenia for reasons other than MDS. Furthermore, there was a 10.3-fold lower expression of IEX-1 in patients with MDS-type refractory anemia with excess of blasts (RAEB) as compared to normal donors (Fig. 3b). To investigate whether decreased transcription translated into decreased protein expression of IEX-1, we examined CD34? cell lysates from four patients with MDS and one normal control. Using western blotting analysis, we showed a lack of IEX-1 protein expression in MDSCD34? cells (Fig. 3c).
4 Discussion We observed a gene signature in MDS reflecting increased proliferation and apoptosis, which goes well in line with the results from previous gene profiling studies in MDS [24–29]. We also found increased expression of genes that indicate enhanced DNA repair and oxidative stress. Furthermore, we found decreased expression of numerous genes encoding cytokines, inflammation-related proteins, growth factors/-receptors, and anti-apoptotic proteins. These transcriptional alterations not only correspond with established concepts of the molecular pathology of MDS [30–32], but also include genes that were hitherto unrecognized in the context of MDS. Combining cDNA macroarray and quantitative real-time RT-PCR analyses, we identified and confirmed differential expression of three genes that were already known to have altered expression levels in MDS: FLT3 [33], GST-3 [34], and EPA-1 [35]. The latter, also known as tissue inhibitor of metalloproteinase-1, was found to be expressed 10.0-fold lower in MDS (P \ 0.00001). Interestingly, EPA-1 encodes a glycoprotein that stimulates erythroid burst-forming and
123
184
Fig. 3 Differential IEX-1 expression in CD34? cells. a Quantitative real-time RT-PCR curves of IEX-1 and the GAPDH control of one representative experiment are shown. b Average differences of crossing points (DCP) of IEX-1 and GAPDH for all examined CD34? cells from normal donors (n = 6), patients with cytopenia for reasons other than MDS (CROM; n = 3), patients with MDS type refractory anemia (RA; n = 7) and MDS type refractory anemia with excess of blasts (RAEB; n = 4) samples are displayed. The lines in each bar represent the standard deviation. The average difference of DCPs (DDCP) of CD34? cells from control patients with cytopenia for reasons other than MDS and CD34? cells from patients with MDS type refractory anemia (RA) is indicated by the double-headed arrow. c Decreased IEX-1 protein expression in MDS-CD34? cells. Western blotting analysis using CD34? cell lysates from patients with MDS and a healthy donor demonstrate lower IEX-1 protein levels in MDS
123
W. C. Prall et al.
erythroid colony-forming units [36]. Its mode of action is unknown, but suppression of apoptosis is supposedly involved. This was concluded from in vitro studies showing an inhibition of apoptosis after retroviral induction of EPA-1 in Burkitt’s lymphoma cell lines [37]. The reduced expression of EPA-1 in CD34? cells from patients with MDS might therefore contribute to both impaired maturation and increased apoptosis of erythroid cells. In our view, this observation warrants further research on the potential influence of recombinant EPA-1 on MDS cells. We identified and confirmed differential expression of two genes that are related to c-myc. First, in MDS-CD34? cells, we detected decreased expression of MAD, which promotes cellular differentiation by antagonizing the c-myc oncogene, and which is associated with growth suppression [38]. Second, we found increased expression of NDPK-B in MDS-CD34? cells. NDPK-B is a transcriptional activator that binds to the c-myc promoter [39]. In sum, this data suggests a role for c-myc in the molecular pathology of MDS. We also found increased expression of ERCC3, a DNA repair helicase with essential endonuclease activity for nucleotide excision repair. This may reflect increased DNA damage in MDS, thus leading to a greater need for repair [40]. Two more differential genes were identified that had decreased expression levels in MDS-CD34? cells: cathepsin L, which is involved in the intracellular protein catabolism and in matrix degradation [41], and TTP. The latter is a transcription factor [42], capable of destabilizing TNF-a-mRNA by binding to AU-rich elements of TNF-amRNA, thus reducing TNF-a levels [43]. If low TTP transcription in MDS translates into low TTP protein expression then this may explain the elevated level of TNF-a that is frequently seen in these patients [44]. Survival had a strong impact on our cluster analysis as it was the parameter that had the strongest statistical association with the segregation of our MDS-patients into two subclusters (Fig. 2). Therefore, we scrutinized our gene signatures, and found distinct transcriptional alterations in patients with short survival. There was decreased expression of VEGF-related genes and no aberrant increase of EGFR-related genes, thus abating the potential hopes raised by the advent of bevacizumab and cetuximab. However, overall six different proteasome-related genes showed an increased expression in patients from the ‘‘short survival’’ cluster. This data should encourage clinical phase with the proteasome-inhibitor bortezomib for the treatment of MDS, in particular for patients with an unfavorable prognosis. The most prominent differential gene from our cDNA analyses, which may also represent a therapeutic target, was IEX-1. IEX-1 showed a dramatically decreased
Differential gene expression in MDS
185
Fig. 4 The IEX-1 gene clusters into a group of 14 genes, which includes 4 more genes [Bcl-like 2, apoptotic protease activating factor (APAF), endothelial TEK tyrosine kinase, and apoptosis inhibitor 2 (AI2)] that are known to inhibit apoptosis. Using the entire cDNA array data, we built a dendrogram by hierarchical cluster analysis using the 1-correlation distance metric and an average linkage algorithm. Each column represents one individual, either normal donor or patient with MDS. PT protein tyrosine, RTK receptor tyrosine kinase
expression in MDS-CD34? cells and an even more striking down-regulation in patients with short survival. Its decreased expression in MDS has independently been described by others [25]. We further focused on IEX-1 and confirmed its decreased expression by real-time RT-PCR and western blot in an independent patient set, thereby underscoring its potential relevance in the molecular pathology of MDS. It has been speculated that IEX-1 not only suppresses, but also induces apoptosis in a cell- and stimulus-dependent manner [45]. We, therefore, utilized our gene dendrogram from hierarchical clustering and examined the function of those genes that were included in the IEX-1 subcluster (Fig. 4). This cluster consisted of another 13 genes, thereof which 4 genes [Bcl-like 2, apoptotic protease activating factor (APAF), endothelial TEK tyrosine kinase, and apoptosis inhibitor 2 (AI2)] are known to inhibit apoptosis. This indicates that IEX-1 is a mediator of anti-apoptosis in MDS-CD34? cells, but additional research is needed to address this question. C-myc, a repressor of IEX-1 has been shown to be increased in MDS [46], whereas C/EBP, an inducer of IEX1 was found to be mutated in some cases of MDS [47]. These are two potential explanations for reduced IEX-1 levels in MDS cells. Recently, a decreased phosphorylation of extracellular signal-regulated kinase (ERK) was found in MDS [48]. This might, perhaps, be also explained by our finding of decreased IEX-1 in MDS as IEX-1 is capable of reversing ERK dephosphorylation [49]. Furthermore, IEX-1 has been shown to directly interact with MCL-1 [50, 51]. MCL1 is also an anti-apoptotic protein with a function analogous to Bcl-2 [52]. The significantly decreased expression of this IEX-1-interacting
partner in MDS-CD34? cells was revealed by cDNA array analyses and corroborated by quantitative real-time RTPCR. In MDS, TNF-a represents one of the most important mediators of apoptosis and its levels are known to be increased in the BM and serum from patients with MDS [53]. Interestingly, IEX-1 inhibits TNF-a-induced apoptosis, and decreased IEX-1 may therefore render MDS-CD34? cells with more sensitive to TNF-a mediated apoptosis [54]. In summary, these observations indicate to a tangible involvement of IEX-1 in the molecular pathology of MDS. Apoptosis is a known feature in MDS but its exact origin and relevance remain elusive [32, 53]. Accordingly, both pro-apoptotic and anti-apoptotic therapies have been proposed [55–57] and both exhibit some efficiency in subsets of patients with MDS. However, if we want to target ‘‘apoptosis’’ more efficiently, we need a more comprehensive understanding of its mechanisms in MDS. Therefore, IEX-1 as a potentially new mediator of apoptosis may represent an important insight into the molecular pathology of MDS. Retinoic acids and hydroxytamoxifen have been shown to induce IEX-1 expression in leukemic and breast cancer cells, respectively [58, 59]. It remains to be seen if these or other agents are capable of efficient IEX-1 induction in MDS, and whether its induction may translate into inhibition of apoptosis, and ultimately improve blood cell counts of patients with MDS. The presented expression data broadens our notion about the molecular pathology of MDS, and if further tested, it may lend itself to better identify patients with short survival. The described molecular alterations may serve to identify MDS as a candidate disease for testing investigational new drugs. The observation of increased
123
186
expression of proteasome-related genes should encourage investigators to explore proteasome-inhibiting therapies in patients with MDS. Importantly, recent pre-clinical data [60] and a first case report [61] support our proposal to examine proteasomal inhibition as a therapeutic concept in MDS. Likewise, drugs which enhance the expression of IEX-1 could theoretically be considered for the treatment of MDS. However, it shall be emphasized that further translational and clinical research would be warranted to corroborate the IEX-1 and other novel molecular pathology findings outlined in this paper. Acknowledgments We are indebted to our colleagues Christoph Maintz, Ulrich Grabenhorst, and Christoph Losem for their support, and to Ba¨rbel Junge, Maria Wolf, Anke Boekmann, Elke RosenbaumKo¨nig, and Hildegard Gaussmann for cell separations. Further, we are grateful to Annette Schreiber, and Claudia Aivado for the visual assessment of arrays. Finally, I thank Else-Marie Meyer for her inspiring lessons in molecular genetics, and Carlo Aul for his cordial support. This work was supported by the Bundesministerium fu¨r Bildung und Forschung (BMBF), Kompetenznetz ‘‘Akute und Chronische Leuka¨mien’’ (No. TP16), and the Leuka¨mie-Liga Du¨sseldorf, e.V.
References 1. Koeffler HP, Golde DW. Cellular maturation in human preleukemia. Blood. 1978;52:355–61. 2. Raza A, Alvi S, Borok RZ, et al. Excessive proliferation matched by excessive apoptosis in myelodysplastic syndromes: the causeeffect relationship. Leuk Lymphoma. 1997;27:111–8. 3. Germing U, Hildebrandt B, Pfeilstocker M, et al. Refinement of the international prognostic scoring system (IPSS) by including LDH as an additional prognostic variable to improve risk assessment in patients with primary myelodysplastic syndromes (MDS). Leukemia. 2005;19:2223–31. doi:10.1038/sj.leu.2403963. 4. Nyakern M, Tazzari PL, Finelli C, et al. Frequent elevation of Akt kinase phosphorylation in blood marrow and peripheral blood mononuclear cells from high-risk myelodysplastic syndrome patients. Leukemia. 2006;20:230–8. doi:10.1038/sj.leu. 2404057. 5. Niimi H, Harada H, Harada Y, et al. Hyperactivation of the RAS signaling pathway in myelodysplastic syndrome with AML1/ RUNX1 point mutations. Leukemia. 2006;20:635–44. doi: 10.1038/sj.leu.2404136. 6. Okamoto M, Shichishima T, Noji H, et al. High frequency of several PIG-A mutations in patients with aplastic anemia and myelodysplastic syndrome. Leukemia. 2006;20:627–34. doi: 10.1038/sj.leu.2404135. 7. Ingram W, Lea NC, Cervera J, et al. The JAK2 V617F mutation identifies a subgroup of MDS patients with isolated deletion 5q and a proliferative bone marrow. Leukemia. 2006;20:1319–21. doi:10.1038/sj.leu.2404215. 8. Magnusson MK, Meade KE, Nakamura R, Barrett J, Dunbar CE. Activity of STI571 in chronic myelomonocytic leukemia with a platelet-derived growth factor beta receptor fusion oncogene. Blood. 2002;100:1088–91. doi:10.1182/blood-2002-01-0165. 9. Prall WC, Czibere A, Jager M, et al. Age-related transcription levels of KU70, MGST1 and BIK in CD34? hematopoietic stem and progenitor cells. Mech Ageing Dev. 2007;128:503–10. doi: 10.1016/j.mad.2007.06.008.
123
W. C. Prall et al. 10. Steidl U, Kronenwett R, Rohr UP, et al. Gene expression profiling identifies significant differences between the molecular phenotypes of bone marrow-derived and circulating human CD34? hematopoietic stem cells. Blood. 2002;99:2037–44. doi:10.1182/ blood.V99.6.2037. 11. Herrler M. Use of SMART-generated cDNA for differential gene expression studies. J Mol Med. 2000;78:B23. 12. Steidl U, Bork S, Schaub S, et al. Primary human CD34? hematopoietic stem and progenitor cells express functionally active receptors of neuromediators. Blood. 2004;104:81–8. doi: 10.1182/blood-2004-01-0373. 13. Kang HL, Benzer S, Min KT. Life extension in Drosophila by feeding a drug. Proc Natl Acad Sci USA. 2002;99:838–43. doi: 10.1073/pnas.022631999. 14. Modlich O, Prisack HB, Munnes M, Audretsch W, Bojar H. Immediate gene expression changes after the first course of neoadjuvant chemotherapy in patients with primary breast cancer disease. Clin Cancer Res. 2004;10:6418–31. doi:10.1158/10780432.CCR-04-1031. 15. Prosniak M, Hooper DC, Dietzschold B, Koprowski H. Effect of rabies virus infection on gene expression in mouse brain. Proc Natl Acad Sci USA. 2001;98:2758–63. doi:10.1073/pnas. 051630298. 16. Huang H, Colella S, Kurrer M, Yonekawa Y, Kleihues P, Ohgaki H. Gene expression profiling of low-grade diffuse astrocytomas by cDNA arrays. Cancer Res. 2000;60:6868–74. 17. Beissbarth T, Fellenberg K, Brors B, et al. Processing and quality control of DNA array hybridization data. Bioinformatics. 2000;16:1014–22. doi:10.1093/bioinformatics/16.11.1014. 18. Quackenbush J. Computational analysis of microarray data. Nat Rev Genet. 2001;2:418–27. doi:10.1038/35076576. 19. Armstrong SA, Staunton JE, Silverman LB, et al. MLL translocations specify a distinct gene expression profile that distinguishes a unique leukemia. Nat Genet. 2002;30:41–7. doi: 10.1038/ng765. 20. Golub TR, Slonim DK, Tamayo P, et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science. 1999;286:531–7. doi:10.1126/science.286. 5439.531. 21. Spentzos D, Levine DA, Ramoni MF, et al. Gene expression signature with independent prognostic significance in epithelial ovarian cancer. J Clin Oncol. 2004;22:4700–10. doi:10.1200/ JCO.2004.04.070. 22. Wu B, Abbott T, Fishman D, et al. Comparison of statistical methods for classification of ovarian cancer using mass spectrometry data. Bioinformatics. 2003;19:1636–43. doi:10.1093/ bioinformatics/btg210. 23. Li C, Wong WH. Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc Natl Acad Sci USA. 2001;98:31–6. doi:10.1073/pnas.011404098. 24. Chen G, Zeng W, Miyazato A, et al. Distinctive gene expression profiles of CD34 cells from patients with myelodysplastic syndrome characterized by specific chromosomal abnormalities. Blood. 2004;104:4210–8. doi:10.1182/blood-2004-01-0103. 25. Hofmann WK, de Vos S, Komor M, Hoelzer D, Wachsman W, Koeffler HP. Characterization of gene expression of CD34? cells from normal and myelodysplastic bone marrow. Blood. 2002;100:3553–60. doi:10.1182/blood.V100.10.3553. 26. Mano H. DNA micro-array analysis of myelodysplastic syndrome. Leuk Lymphoma. 2006;47:9–14. doi:10.1080/ 10428190500264231. 27. Miyazato A, Ueno S, Ohmine K, et al. Identification of myelodysplastic syndrome-specific genes by DNA microarray analysis with purified hematopoietic stem cell fraction. Blood. 2001;98: 422–7. doi:10.1182/blood.V98.2.422.
Differential gene expression in MDS 28. Pradhan A, Mijovic A, Mills K, et al. Differentially expressed genes in adult familial myelodysplastic syndromes. Leukemia. 2004;18:449–59. doi:10.1038/sj.leu.2403265. 29. Ueda M, Ota J, Yamashita Y, et al. DNA microarray analysis of stage progression mechanism in myelodysplastic syndrome. Br J Haematol. 2003;123:288–96. doi:10.1046/j.1365-2141.2003. 04601.x. 30. Pellagatti A, Esoof N, Watkins F, et al. Gene expression profiling in the myelodysplastic syndromes using cDNA microarray technology. Br J Haematol. 2004;125:576–83. doi:10.1111/j.13652141.2004.04958.x. 31. Pellagatti A, Fidler C, Wainscoat JS, Boultwood J. Gene expression profiling in the myelodysplastic syndromes. Hematology. 2005;10:281–7. doi:10.1080/10245330500065680. 32. Steensma DP, Tefferi A. The myelodysplastic syndrome(s): a perspective and review highlighting current controversies. Leuk Res. 2003;27:95–120. doi:10.1016/S0145-2126(02)00098-X. 33. Allampallam K, Shetty V, Hussaini S, et al. Measurement of mRNA expression for a variety of cytokines and its receptors in bone marrows of patients with myelodysplastic syndromes. Anticancer Res. 1999;19:5323–8. 34. Holmes J, Wareing C, Jacobs A, Hayes JD, Padua RA, Wolf CR. Glutathione-s-transferase pi expression in leukaemia: a comparative analysis with mdr-1 data. Br J Cancer. 1990;62:209–12. 35. Ries C, Loher F, Zang C, Ismair MG, Petrides PE. Matrix metalloproteinase production by bone marrow mononuclear cells from normal individuals and patients with acute and chronic myeloid leukemia or myelodysplastic syndromes. Clin Cancer Res. 1999;5:1115–24. 36. Gasson JC, Golde DW, Kaufman SE, et al. Molecular characterization and expression of the gene encoding human erythroid-potentiating activity. Nature. 1985;315:768–71. doi:10.1038/315768a0. 37. Guedez L, Stetler-Stevenson WG, Wolff L, et al. In vitro suppression of programmed cell death of B cells by tissue inhibitor of metalloproteinases-1. J Clin Invest. 1998;102:2002–10. doi: 10.1172/JCI2881. 38. Chen J, Willingham T, Margraf LR, Schreiber-Agus N, DePinho RA, Nisen PD. Effects of the MYC oncogene antagonist, MAD, on proliferation, cell cycling and the malignant phenotype of human brain tumour cells. Nat Med. 1995;1:638–43. doi:10.1038/nm0795-638. 39. Postel EH, Berberich SJ, Flint SJ, Ferrone CA. Human c-myc transcription factor PuF identified as nm23-H2 nucleoside diphosphate kinase, a candidate suppressor of tumor metastasis. Science. 1993;261:478–80. doi:10.1126/science.8392752. 40. Mitra S, Boldogh I, Izumi T, Hazra TK. Complexities of the DNA base excision repair pathway for repair of oxidative DNA damage. Environ Mol Mutagen. 2001;38:180–90. doi:10.1002/em.1070. 41. Kane SE, Gottesman MM. The role of cathepsin L in malignant transformation. Semin Cancer Biol. 1990;1:127–36. 42. Taylor GA, Thompson MJ, Lai WS, Blackshear PJ. Phosphorylation of tristetraprolin, a potential zinc finger transcription factor, by mitogen stimulation in intact cells and by mitogen-activated protein kinase in vitro. J Biol Chem. 1995;270:13341–7. doi: 10.1074/jbc.270.22.13341. 43. Carballo E, Lai WS, Blackshear PJ. Feedback inhibition of macrophage tumor necrosis factor-alpha production by tristetraprolin. Science. 1998;281:1001–5. doi:10.1126/science.281. 5379.1001. 44. Kitagawa M, Saito I, Kuwata T, et al. Overexpression of tumor necrosis factor (TNF)-alpha and interferon (IFN)-gamma by bone marrow cells from patients with myelodysplastic syndromes. Leukemia. 1997;11:2049–54. doi:10.1038/sj.leu.2400844. 45. Huang YH, Wu JY, Zhang Y, Wu MX. Synergistic and opposing regulation of the stress-responsive gene IEX-1 by p53, c-Myc,
187
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
and multiple NF-kappaB/rel complexes. Oncogene. 2002;21: 6819–28. doi:10.1038/sj.onc.1205854. Rajapaksa R, Ginzton N, Rott LS, Greenberg PL. Altered oncoprotein expression and apoptosis in myelodysplastic syndrome marrow cells. Blood. 1996;88:4275–87. Kaeferstein A, Krug U, Tiesmeier J, et al. The emergence of a C/ EBPalpha mutation in the clonal evolution of MDS towards secondary AML. Leukemia. 2003;17:343–9. doi:10.1038/sj.leu. 2402805. Fuhler GM, Drayer AL, Vellenga E. Decreased phosphorylation of protein kinase B and extracellular signal-regulated kinase in neutrophils from patients with myelodysplasia. Blood. 2003;101:1172–80. doi:10.1182/blood.V101.3.1172. Letourneux C, Rocher G, Porteu F. B56-containing PP2A dephosphorylate ERK and their activity is controlled by the early gene IEX-1 and ERK. EMBO J. 2006;25:727–38. doi:10.1038/ sj.emboj.7600980. Kruse ML, Arlt A, Sieke A, et al. Immediate early gene X1 (IEX1) is organized in subnuclear structures and partially co-localizes with promyelocytic leukemia protein in HeLa cells. J Biol Chem. 2005;280:24849–56. doi:10.1074/jbc.M501571200. Kumar R, Lutz W, Frank E, Im HJ. Immediate early gene X-1 interacts with proteins that modulate apoptosis. Biochem Biophys Res Commun. 2004;323:1293–8. doi:10.1016/j.bbrc.2004.09.006. Wang X, Studzinski GP. Antiapoptotic action of 1, 25-dihydroxyvitamin D3 is associated with increased mitochondrial MCL-1 and RAF-1 proteins and reduced release of cytochrome c. Exp Cell Res. 1997;235:210–7. doi:10.1006/excr.1997.3667. Parker JE, Mufti GJ. Excessive apoptosis in low risk myelodysplastic syndromes (MDS). Leuk Lymphoma. 2000;40:1–24. doi: 10.3109/10428190009054877. Wu MX, Ao Z, Prasad KV, Wu R, Schlossman SF. IEX-1L, an apoptosis inhibitor involved in NF-kappaB-mediated cell survival. Science. 1998;281:998–1001. doi:10.1126/science.281. 5379.998. Czibere A, Prall WC, Zerbini LF, et al. The nonsteroidal antiinflammatory drug exisulind selectively induces apoptosis via JNK in secondary acute myeloid leukemia after myelodysplastic syndrome. Cell Cycle. 2005;4:812–7. Czibere A, Prall WC, Zerbini LF, et al. Exisulind induces apoptosis in advanced myelodysplastic syndrome (MDS) and acute myeloid leukaemia/MDS. Br J Haematol. 2006;135:355–7. doi:10.1111/j.1365-2141.2006.06298.x. List A, Beran M, DiPersio J, et al. Opportunities for Trisenox (arsenic trioxide) in the treatment of myelodysplastic syndromes. Leukemia. 2003;17:1499–507. doi:10.1038/sj.leu.2403021. Arlt A, Minkenberg J, Kocs B, et al. The expression of immediate early gene X-1 (IEX-1) is differentially induced by retinoic acids in NB4 and KG1 cells: possible implication in the distinct phenotype of retinoic acid-responsive and -resistant leukemic cells. Leukemia. 2004;18:1646–55. doi:10.1038/sj.leu.2403481. Semlali A, Oliva J, Badia E, Pons M, Duchesne MJ. Immediate early gene X-1 (IEX-1), a hydroxytamoxifen regulated gene with increased stimulation in MCF-7 derived resistant breast cancer cells. J Steroid Biochem Mol Biol. 2004;88:247–59. doi:10.1016/ j.jsbmb.2003.12.005. Braun T, Carvalho G, Coquelle A, Vozenin MC, Lepelley P, Hirsch F, et al. NF-kappaB constitutes a potential therapeutic target in high-risk myelodysplastic syndrome. Blood. 2006;107: 1156–65. doi:10.1182/blood-2005-05-1989. Terpos E, Verrou E, Banti A, Kaloutsi V, Lazaridou A, Zervas K. Bortezomib is an effective agent for MDS/MPD syndrome with 5q- anomaly and thrombocytosis. Leuk Res. 2007;31:559–62. doi:10.1016/j.leukres.2006.05.018.
123