Journal of Leukocyte Biology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published online as doi:10.1189/jlb.0206112 on June 12, 2006

Published online before print June 12, 2006
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data: jlb.0206112, Brown et al
Right arrow All Versions of this Article:
jlb.0206112v1
80/2/433    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Brown, A. L.
Right arrow Articles by D’Andrea, R. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Brown, A. L.
Right arrow Articles by D’Andrea, R. J.
(Journal of Leukocyte Biology. 2006;80:433-447.)
© 2006 by Society for Leukocyte Biology

Genetic regulators of myelopoiesis and leukemic signaling identified by gene profiling and linear modeling

Anna L. Brown*,{dagger},{ddagger},1, Christopher R. Wilkinson*,{dagger},{ddagger},§,1, Scott R. Waterman, Chung H. Kok*,{dagger},{ddagger}, Diana G. Salerno*,{dagger},{ddagger}, Sonya M. Diakiw*,{dagger},{ddagger}, Brenton Reynolds*,{dagger},{ddagger}, Hamish S. Scott||, Anna Tsykin§,¶, Gary F. Glonek§, Gregory J. Goodall,**, Patty J. Solomon§, Thomas J. Gonda{dagger}{dagger} and Richard J. D’Andrea*,{dagger},{ddagger},2

* Haematology and Oncology Program, Child Health Research Institute, North Adelaide, South Australia;
{dagger} The Queen Elizabeth Hospital, Woodville, South Australia; Departments of
{ddagger} Paediatrics and
** Medicine and
§ School of Mathematical Sciences, University of Adelaide, South Australia;
The Division of Human Immunology and Hanson Institute, Institute of Medical and Veterinary Sciences, Adelaide, South Australia;
|| The Genetics and Bioinformatics Division, The Walter and Eliza Hall Institute of Medical Research, Melbourne, Victoria, Australia; and
{dagger}{dagger} Cancer Biology Program, Centre for Immunology and Cancer Research, Princess Alexandra Hospital, Woolloongabba, Queensland, Australia

2Correspondence: Child Health Research Institute, 7th Floor, Clarence Rieger Building, 72 King William Road, North Adelaide, South Australia 5006, Australia. E-mail: Richard.dandrea{at}adelaide.edu.au


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Mechanisms controlling the balance between proliferation and self-renewal versus growth suppression and differentiation during normal and leukemic myelopoiesis are not understood. We have used the bi-potent FDB1 myeloid cell line model, which is responsive to myelopoietic cytokines and activated mutants of the granulocyte macrophage-colony stimulating factor (GM-CSF) receptor, having differential signaling and leukemogenic activity. This model is suited to large-scale gene-profiling, and we have used a factorial time-course design to generate a substantial and powerful data set. Linear modeling was used to identify gene-expression changes associated with continued proliferation, differentiation, or leukemic receptor signaling. We focused on the changing transcription factor profile, defined a set of novel genes with potential to regulate myeloid growth and differentiation, and demonstrated that the FDB1 cell line model is responsive to forced expression of oncogenes identified in this study. We also identified gene-expression changes associated specifically with the leukemic GM-CSF receptor mutant, V449E. Signaling from this receptor mutant down-regulates CCAAT/enhancer-binding protein {alpha} (C/EBP{alpha}) target genes and generates changes characteristic of a specific acute myeloid leukemia signature, defined previously by gene-expression profiling and associated with C/EBP{alpha} mutations.

Key Words: myeloid • transcription factor • myeloid leukemia • microarray • gene expression


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
A detailed understanding of the molecular regulation of myelopoiesis is critical for developing new approaches to hematological therapy and for diagnosis and treatment of myeloid leukemia. A number of hematopoietic growth factors (HGFs), or cytokines, play a key role in regulating the myeloid lineage. Binding of the HGF ligand to their receptors results in activation of intracellular kinase activity and induction of multiple signaling pathways. This, in turn, mediates changes in cell behavior, ultimately through changes in gene expression associated with proliferation, survival, self-renewal, and differentiation. This response is mediated in part by modulation of the action of a number of lineage-specific transcription factors (TFs), which act by regulating key target genes such as cell cycle regulators, HGF receptors, and mature cell proteins that define particular cell types or lineages [1 , 2 ]. In addition, these TFs often autoregulate their own promoters and are able to inhibit alternative genetic programs, thus specifying lineage commitment [3 4 5 ]. In the granulocyte-macrophage (GM) lineage, key TFs are the CCAAT/enhancer-binding protein (C/EBP) family and the ets family member PU.1 (for a recent review, see Rosmarin et al. [6 ]), the ratios of which are critical for cell fate [7 ]. The nature of the links between myeloid HGFs and the action of TFs involved in the cellular response is still largely unclear. To address this important gap in our understanding of myeloid cell responses, we have been examining how intracellular signals initiated by the myeloid HGFs, GM-colony stimulating factor (CSF), and interleukin (IL)-3 impact on transcriptional programs.

Given the pivotal roles of HGF signaling in regulating hematopoiesis, it is not surprising that aberrant HGF receptor activation or activation of pathways downstream of HGF receptors is associated with leukemia. A key example of this is activation of fms-related tyrosine kinase 3 (FLT3), which represents the most commonly mutated gene in acute myeloid leukemia (AML) and is constitutively activated by acquired mutation (most commonly, internal tandem duplications) in 30–35% of AML cases [8 , 9 ]. Aberrant HGF signaling in AML cells can also result from constitutive activation of other receptors [10 11 12 ] or signaling molecules [13 14 15 ]. Autocrine production of GM-CSF and IL-3 also occur occasionally in AML [16 ] and chronic myeloid leukemia [17 ] and also result in constitutive activation of proliferation and survival pathways. The current model for pathogenesis of AML involves cooperation between these mutations and others, which leads to a block in differentiation or acquisition of self-renewal. This is associated with disruption of the normal transcriptional program, sometimes as a result of lesions involving key myeloid TFs such as C/EBP{alpha} and PU.1 [1 ] but commonly through the interference of translocation-derived fusion proteins [18 ]. There are overlaps between the effects of these two cooperating classes of mutation, as for example, FLT3 activation appears to contribute to a differentiation block in some contexts [19 ].

With the advent of new, gene-profiling technologies, it has become possible to investigate the molecular basis of myeloid leukemias using approaches that measure global gene-expression changes. Over recent years, there has been a large mass of data generated with respect to the gene-expression profiles of AML patient samples [20 21 22 23 ], sets of genes downstream of leukemogenic TFs [24 , 25 ] and associated with FLT3 mutations [19 , 26 27 28 ]. Such profiling needs to be considered together with the gene-expression patterns available for myeloid cell line models undergoing directed differentiation [29 30 31 32 ] and granulocytic populations at different stages [33 ]. In many of these analyses, gene-expression changes cannot generally be associated, a priori, with a specific cellular process (e.g., mitogenesis, promoting, or blocking differentiation, survival, self-renewal), as in these systems, many processes are occurring simultaneously. Dissection of gene-expression changes associated with each cellular outcome requires differential activation of these processes in parallel cell systems, such that gene-expression profiles can be monitored simultaneously. With such a system, a linear modeling approach permits identification of different classes of genes based on such a complex set of conditions. With this in mind, we turned to a bipotential cell line model of myeloid differentiation, which displays differential responses to GM-CSF, IL-3, and activated GM-CSF receptor mutants. The FDB1 cell line is strictly growth factor-dependent for survival; however, cells proliferate continuously in the presence of IL-3 with only a minimal amount of spontaneous differentiation to neutrophils, monocytes, and megakaryocytes. In the presence of GM-CSF, FDB1 cells differentiate synchronously along the neutrophil and monocyte lineages with complete differentiation after 5–7 days [34 ]. This ability to uncouple mitogenesis and self-renewal from differentiation while still using physiological stimuli allows a dissection of these processes, not readily achieved in primary cells, which undergo simultaneous proliferation and differentiation and eventual growth arrest and cell death. To increase the power of this study, we included two activated mutants of the GM-CSF receptor, which have differential leukemogenic activity. These two mutants, FI{Delta} and V449E, are derived from the common signaling subunit (hßc) for GM-CSF, IL-3, and IL-5 (reviewed in Gonda and D’Andrea [35 ] and D’Andrea and Gonda [36 ]) and represent two distinct classes of activated receptor [extracellular (EC); transmembrane (TM). The two mutants display overlapping, biochemical responses compared with the GM-CSF and IL-3 receptor complexes [37 ] and most likely, represent alternative receptor configurations [35 ]. In vivo, V449E induces a myeloid leukemia consistent with its ability to support the generation of immature myeloid cell lines in vitro [38 ], and the EC mutant FI{Delta} induces cytokine-independent formation of CFU-GM [colony-forming units-GM] and erythroid progenitor colonies and leads to a myeloproliferative disease in murine models [38 , 39 ]. It is important that the FDB1 cell line also responds differentially to these activated GM-CSF receptor mutants. V449E induces factor-independent proliferation of FDB1 cells and is able to block GM-CSF-induced differentiation, properties that mimic the ability of this mutant to induce AML in vivo. FI{Delta} induces factor-independent GM differentiation, reflective of the signals that give rise to myeloproliferative disorder in vivo [34 , 40 ]. Thus, FDB1 provides a manipulable model with which to study downstream signaling and transcriptional responses from cytokines and activated receptors. Here, we have used this model, time-course gene-expression profiling, and linear modeling to dissect the molecular mechanisms underlying the differential activities associated with the myeloid growth factors GM-CSF and IL-3 and the activated mutants of the GM-CSF receptor. With a view to understanding the molecular control of myeloid differentiation, we focused our downstream and clustering analysis on identification of candidate regulatory genes encoding TFs and TF-associated products. In addition, we have used comparative analysis between our data and other studies focused on C/EBP{alpha} and AML to examine further the mechanism of leukemic receptor signaling. Here, we provide evidence for a mechanism of leukemic induction by activated growth factor receptor mutants, which may involve modulation of C/EBP{alpha} activity.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Cytokines and antibodies
Recombinant murine (m)IL-3 and GM-CSF were produced from baculovirus vectors supplied by Dr. Andrew Hapel (John Curtin School of Medical Research, Canberra, Australia).

Culture and analysis of FDB1 cells
FDB1 cells were maintained as described previously [34 ]. Infected pools were maintained as described plus 1 µg/mL Puromycin (Sigma Chemical Co., St. Louis, MO). Receptor expression (FI{Delta} or V449E) was confirmed by staining with a murine anti-FLAGTM antibody (Sigma Chemical Co.) followed by an anti-mouse fluorescein isothiocyanate-conjugated antibody (Silenus, Chemicon International, Temecula, CA). Cells were analyzed by flow cytometry using an Epics Elite ESP (Coulter Electronics, UK). If necessary, cells were sorted for expression as described previously [41 ]. For the time courses, cells were washed three times in Iscove’s modified Dulbecco’s medium and cultured in 500 BM U/mL mIL-3 or mGM-CSF or without growth factor {1 BM unit is the amount that gives 50% maximal colony formation (CFU-GM) using bone marrow cells; see ref. [42 ]}. To assess differentiation, cells were centrifuged onto slides and stained with May-Grünwald-Giemsa, and the proportion of differentiated cells was determined microscopically. To assess cell viability, the percentage of cells excluding trypan blue was determined using a hemocytometer.

Experimental design
The experiment is arranged as a factorial design studying cell lines and time. For time-course information, each cell population was sampled at six time-points (0, 6, 12, 24, 48, and 72 h), and the time zero (T0) point was used as a reference to obtain relative expression levels following IL-3 withdrawal. We also included the parental FDB1 cell line maintained in IL-3 to establish any baseline differences at T0 as a result of expression of the activated receptors. Matched time comparisons were also performed at all six time-points between the two conditions supporting differentiation (GM-CSF and FI{Delta}) and between the two cell populations expressing activated receptor mutants (FI{Delta} vs. V449E). Two biological replicate samples were used for each cell population, and a dye-swap replicate was performed for each comparison.

Hybridization and data analysis
Total cellular RNA was harvested with Trizol (Invitrogen, Carlsbad, CA) and further purified with the RNeasy RNA purification kit (Qiagen, Valencia, CA). Aliquots of RNA were kept to perform quantitative reverse transcriptase-polymerase chain reaction (QRT-PCR) analysis. cDNA was generated using 50 µg total RNA as a template and primed with PolyT(V)N (4.0 µg) and random hexamers (1.0 µg, Amersham, UK). cDNA was labeled with Cy5 or Cy3 dyes using the Cy-Scribe post-labeling kit (Amersham) as per the instructions. Labeled cDNA was hybridized to microarray slides printed with the CompuGen Mouse OligoLibrary (v2.0, 21,997 65-mers comprising 21,587 unique genes) by the Adelaide Microarray Facility (Australia). Slides were scanned using a GenePix 3000B scanner (Axon Instruments, Sunnydale, CA), and the Spot package (CSIRO, Australia) was used to identify spots and estimate fore- and background intensities (using a morphological opening background estimator) [43 , 44 ]. Data analysis was performed in R (www.r-project.org) using the Limma package of Bioconductor [45 , 46 ] and in-house scripts. Arrays were normalized using intensity-dependent spatial normalization and scale normalization [47 ]. Spatial loess was also applied to a subset of arrays [48 ]. Linear modeling and F-test-based classifications were performed with the Limma package of bioconductor [45 ]. We estimated, effects for baseline (T0) differences between cell lines change over time in FI{Delta} (0, 72 h) and sample by time-interaction parameters between FI{Delta} and V449E and between FI{Delta} and GM-CSF. To examine the distribution of differentially expressed genes for a given comparison of interest, we constructed a volcano plot, in which we plotted log2(fold change) on the x-axis and the –log10[false discovery rate (FDR)-adjusted P value] on the y-axis. This generates a volcano-like shape, in which genes at low fold changes are typically not differentially expressed (P values close to 1; –log(p) approaching zero), and those with strong evidence for differential expression typically have high fold-change values. Gene ontology (GO) over-representation analysis was performed using GO-STAT using the CompuGen Library as the comparison set and applying a FDR P value adjustment [49 ]. Promoter analysis was performed using the CUREOS database (http://cureos.wehi.edu.au) using Transfac matrix M00770 (V$CEBP_Q3, www.biobase.de) and requiring mouse and human homology. A {chi}2test was used to test for over-representation of consensus-binding sites. Additional information may be found under Gene Expression Omnibus (GEO) Accession GSE3333. We have followed the guidelines set out by the Microarray Gene Expression Data Society (www.mged.org/miame).

Clustering of TFs
GO terms were obtained via the SOURCE website (http://source.stanford.edu, UniGene build 139) or via direct query of the National Center for Biotechnology Information gene database. To increase sensitivity further to potential TFs, we used homologene to extracted GO terms for human homologs. We selected all genes, which were children of the terms transcription (GO:0006350), nucleoplasm (GO:0005654), DNA binding (GO:003677), or transcription regulator activity (GO:0030258) or contained the terms "transcription," "histone," or "chromatin" in their GenBank Description. A total of 2338 genes was selected as transcription factor-associated. This was reduced down to 340 after filtering out genes with nonsignificant F-test values (1x10–5). Clustering was performed using the Diana algorithm (R cluster package). We evaluated the distance between two profiles using a measure based on that of Bar-Joseph [50 ] for comparing two time profiles. Briefly, for a gene i, a smoothed spline curve C was fitted to each of the five time-course and matched time data sets. The distance between two genes (i and j) for time course p was calculated as Formula, and the total distance was then calculated as Formula. The dendrogram was then cut at a height of 1.25 to define 14 clusters. Full annotation information of genes in each cluster is listed in Supplementary Table 1.

QRT-PCR
Aliquots of RNA prepared for the microarray analysis were subjected to real-time RT-PCR analysis. For this, RNA was treated with DNase (Ambion, Austin, TX), reverse-transcribed with Oligo-dT (Ambion) using Omniscript RT (Qiagen). The sequences of the oligonucleotides used for PCR are listed in Supplementary Table 2. Gene-specific PCR reactions were performed for 38 cycles using Amplitaq Gold (Perkin Elmer, Wellesley, MA) and recommended conditions. SYBR green (10x; Molecular Probes, Eugene, OR) was added to a final concentration of 0.6x per reaction, which was performed on the Rotor-Gene 3000 and related software used for data collection and to determine mean expression values relative to Cyclophilin A (Corbett Research, Version 5.0). Amplification products were analyzed by melt curve and resolved on 2% agarose to confirm specificity. Analysis was performed using a mixed-effects linear model to estimate T0 differences between each cell line and mean changes within cell lines over time, treating PCR run as a blocking variable.

Retroviral transduction and functional analysis in the FDB1 cell line
For functional analysis, the full-length c-Myb cDNA was cloned into the murine stem cell virus (MSCV)-internal ribosome entry site (IRES)-green fluorescence protein (GFP) retroviral vector and FDB1 cells infected by cocultivation as described previously [40 ]. Following infection with virus-producing cells, GFP+ FDB1 cells were isolated by flow cytometry and expanded in mIL-3. For testing, cells were withdrawn from IL-3 and monitored for morphological differentiation in the presence and absence of GM-CSF for 5 days, as described previously [40 ]. Changes in cell morphology were recorded following cytocentrifugation of cells onto a glass slide and staining with May-Grünwald-Giemsa.

Association of the V449E gene set with leukemia subsets identified by gene profiling
We downloaded the (MAS5) normalized data used by Valk et al. [51 ] from GEO (GSE1159-21979). The normalized data were loaded into R for analysis with the Limma package, and the patients were divided into the 16 clusters (groups) as defined in ref. [51 ]. For each cluster, we compared expression in patients against normal controls, and for each gene obtained, FDR adjusted P value. We then mapped the V449E-associated gene set (see Go Go Go Table 4 ) to the probes on the HG-U133A GeneChip. After cross-species mapping, the V449E-associated data set reduced from 44 probes to 30. Then, for each of the 16 clusters identified in this study, we looked for association between our V449E gene set and genes differentially expressed in this cluster. We ranked all genes on the array on the basis of FDR-adjusted P values and then examined the distribution of ranks of the 30 Affymetrix probes homologous to our V449E gene set. Significance was assessed using a Wilcoxon rank sum test as implemented in R.


View this table:
[in this window]
[in a new window]
  		Table 1. Genes Associated with Granulocyte and Macrophage Differentiation (Full Annotation and Linear Modeling Results May Be Found in Supplementary Table 3)

 

View this table:
[in this window]
[in a new window]
  		Table 2. Known Myeloid Differentiation and Proliferation-Associated Genes1

 

View this table:
[in this window]
[in a new window]
  		Table 3. Genes Associated with Myeloid Cell Proliferation and Self-Renewal (Full Annotation and Linear Modeling Results May Be Found in Supplementary Table 4)

 

View this table:
[in this window]
[in a new window]
  		Table 4. V449E Down-Regulated Genes (Full Annotation and Linear Modeling Results May Be Found in Supplementary Table 6)

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
A model system of myeloid cell growth and differentiation
We have previously outlined the use of retroviral infection to generate FDB1 cell populations expressing the activated GM-CSF receptor mutants FI{Delta} and V449E [40 ]. A similar level of expression of the hßc mutant in each population was confirmed by staining for the FLAG epitope fused to the N terminus of the mutant ß subunit (Fig. 1A ). For each condition, changes in morphology, growth, and viability were quantitated (see Fig. 1B and 1C ). The FDB1 cells behaved as described previously [34 , 40 ], and complete differentiation to granulocytes and macrophages (approximately equal ratio) occurred over 5 days in GM-CSF and in response to the FI{Delta} signal. The switch from IL-3 to V449E signaling did not result in any discernible change to morphology, growth, or viability of the FDB1 cells.


Figure 1
View larger version (51K):
[in this window]
[in a new window]
  		Figure 1. FDB1 experimental system. (A) Cell surface expression of the hßc-activated mutants on FDB1 cells as determined by flow cytometry. Transduced cells (solid peaks) were stained with an anti-FLAGTM monoclonal antibody and compared with identically stained, untransduced cells (open peaks). (B) Photomicrographs of cells induced to self-renew or differentiate over 5 days in the presence of the indicated growth factors or through the activity of the hßc-activated mutants. Original magnification, 400x. (C) Differential counts performed with the cell populations used for RNA preparation. Black, Blast cells; light gray, granulocytes; dark gray, monocytes; white, intermediately differentiated cells.

 
Experimental design and data generation
For dissection of signaling pathways, we performed a time-course study using FDB1 cells switched from a continuous growth signal (IL-3) to signaling via the leukemic receptor, V449E, or to conditions permitting synchronous monocytic and granulocytic differentiation (GM-CSF or FI{Delta}). This allowed a parallel time-course comparison to reveal gene-expression changes associated with the switch to alternative differentiation-inducing signals (GM-CSF or FI{Delta}) or signaling via the leukemia-inducing V449E mutant. In addition, we performed comparisons between cell populations at matched time-points to reveal differences in gene expression between cells undergoing differentiation or proliferation/self-renewal. The overall experimental design, incorporating matched-time and time-course comparisons (along with dye swaps and biological replicates), is shown schematically in Figure 2 and was based on admissible design criteria [52 ]. A major strength of this design is that it allows estimation of genes with differential expression in the initial cell populations (i.e., T0 differences) and time-related changes within a cell population (main effects). It is important that this design also permits identification of genes with significantly different expression profiles between cell populations over time (interaction effects).


Figure 2
View larger version (19K):
[in this window]
[in a new window]
  		Figure 2. Experimental design. Parental FDB1 cells grown in IL-3 were washed and cultured for up to 3 days in the presence of GM-CSF. In parallel, FDB1 cell populations expressing FI{Delta} or V449E were washed and cultured in the absence of growth factor. RNA was harvested from these populations at 0, 6, 12, 24, 48, and 72 h. Comparisons performed are represented by arrows (double-headed arrow indicates the use of dye-swap comparisons). All comparisons presented were performed four times (i.e., two replicate cultures, two dye swaps, total slides 112). Differential cellular outcomes associated with each condition are indicated on the right. P, Proliferation; S, survival; D, differentiation; L, leukemic signaling.

 
Linear modeling to identify differentially expressed genes
The linear modeling approach is particularly well-suited to factorial experiments comparing one parameter (cell population) against another (time) and is more powerful than the alternative approach of analyzing the data set as a series of smaller experiments with the same total number of slides [45 ]. We used a linear modeling approach that allowed us to combine arrays from different cell populations and use an F-test-based approach to select genes with complex expression profiles of potential biological interest [45 , 46 ]. This allowed us to identify common gene-expression changes between FDB1 cell populations undergoing one or more fates in common as shown in Figure 2 . Thus, we classed differentiation-associated genes as those that increased in expression over 72 h in the FI{Delta} and/or the GM-CSF cell population and for which this change was significantly greater than the change (if any) in the V449E cell population. The criteria used to select proliferation and self-renewal-associated genes were identical, except that we required decreases in gene expression rather than increases. Using an F-test with a FDR-adjusted P value cut-off of 1 x 10–5 and a minimum change of 1.4-fold, we identified 205 differentiation-associated genes and 175 proliferation or self-renewal-associated genes. These gene sets are represented in volcano plots in Figure 3 , which provide a visual indication of the relative evidence for differential expression of a given gene. Figure 3A shows a clear increase in the number of genes displaying differential expression over time, based on FDR-adjusted P value, in the parental FDB1 population shifted to GM-CSF and the FI{Delta} population. Although few significant changes are observed at the 6-h time-point, clear groups of differentiation-associated (red) and proliferation-associated (green) genes are identified by this analysis over the 72-h time-course. To examine V449E signaling as a model of leukemic-activated cytokine receptor signaling, we compared the gene-expression profiles of the parental FDB1 cells and the FI{Delta} population (in IL-3) with the V449E cell population under the same conditions. Using an F-test with a FDR-adjusted P value cut-off of 1 x 10–5 and a minimum change of 1.4-fold, we identified 44 genes specifically down-regulated by the V449E mutant (shown in purple in Fig. 3B ). It is interesting that this gene set is strongly differentiation-associated, as indicated by the increased expression of these genes over time in the parental cells responding to GM-CSF and in the FI{Delta} population (Fig. 3A) . This is most likely reflective of the ability of V449E to block differentiation of myeloid cells, a property that is crucial for induction of AML in vivo.


Figure 3
View larger version (31K):
[in this window]
[in a new window]
  		Figure 3. Volcano plots of evidence for differential expression. Plot shows –log10(FDR-adjusted P value) versus log2 (fold change). (A) Distribution of genes for each cell population at 6, 24, and 72 h (each compared with 0 h). (B) Baseline (0 h, IL-3) differences between each cell population. In each plot, the set of 205 differentiation-associated genes is represented by red points, the 175 proliferation and self-renewal-associated genes by green points, and the 44 genes specifically down-regulated in V449E at 0 h by purple points. The dashed gray line represents a FDR-adjusted P value cut-off of 0.01. We see that in FI{Delta} and GM-CSF populations, many of the highlighted genes show early evidence of change, and changes become more significant over time. Typically, the genes down-regulated in V449E at T0 stay down-regulated at subsequent time-points, and they display increased expression over time in the FI{Delta} and GM-CSF FDB1 populations (A).

 
After examining profiles of genes with significant gene-expression changes under various conditions, we selected 10 genes of interest and performed QRT-PCR to provide validation for the microarray results (Fig. 4 ). For each gene, we then made eight comparisons and compared significance level and direction between microarray and QRT-PCR-based estimates. For the 80 comparisons made, we observed 72.5% concordance on significance level and direction. In general, microarray-based estimates of fold changes were smaller than those from QRT-PCR.


Figure 4
View larger version (44K):
[in this window]
[in a new window]
  		Figure 4. Comparisons of QRT-PCR and microarray data. For each gene, comparisons were made at 0, 24, and 72 h. In each plot, microarray results are represented by smoothed spline profiles with V449E by a green, dashed line; FI{Delta} by a blue-dot, dashed line; and GM-CSF by a pink, long, dashed line. Baseline (T0) comparisons are represented by points as follows: cyan points, V449E compared with parental FDB1; purple squares, FI{Delta} compared with parental FDB1. RT-PCR results are represented by points with bars representing 95% confidence intervals. Dark green circles, V449E; navy triangles, FI{Delta}; red diamonds, GM-CSF. Light blue circles, V449E compared with parental FDB1 at T0; dark purple circles, FI{Delta} compared with parental FDB1. (A) Genes associated with GM differentiation; (B) genes associated with proliferation and self-renewal. Primer sequences used are presented (see Supplementary Table 5).

 
Validation of the model system—measuring gene-expression changes and function in the FDB1 system
As discussed above, we identified differentiation-associated genes on the basis of elevated expression in the FI{Delta} and/or the GM-CSF population and stable expression in the V449E population. Table 1 lists all 205 differentiation-associated genes together with their fold-changes and FDR P values over 72 h in the FI{Delta} cell population (changes in GM-CSF were similar and are supplied along with full gene annotation information in Supplementary Table 3). Example gene-expression profiles are shown in Figure 5A . GO analysis of this gene set indicated significant over-representation of terms such as defense response (GO: 0006952, P=2x10–8), response to wounding (GO: 0009611, P=5x10–6), and chemotaxis (GO: 0006935, P=5x10–5), consistent with up-regulation of genes involved in mature granulocytes and macrophages. Many of the genes identified in this analysis are known to be associated with myeloid differentiation (Table 2 ), thus providing an important validation of this cell-line model of myeloid differentiation and confirming the hypothesis that genes displaying concordant regulation in response to the FI{Delta} mutant and GM-CSF will be of most relevance to GM differentiation in vivo.


Figure 5
View larger version (41K):
[in this window]
[in a new window]
  		Figure 5. Examples of proliferation and differentiation-associated gene-expression profiles. (A) Genes associated with GM differentiation. (B) Genes associated with proliferation and self-renewal. In each plot, points represent normalized microarray fold-change values (one point per array), and lines represent weighted, smoothed spline estimates of the change over 72 h. Upper plots: green circles and dashed line, V449E profile; blue triangles and a dot-dash line, FI{Delta} profile; red squares and a solid line, V449E versus FI{Delta} direct comparisons. Cyan points, V449E compared with parental FDB1 at T0 (i.e., in IL-3). Lower plots: blue triangles and a dot-dash line, FI{Delta}; pink diamonds and long, dashed line, GM-CSF; purple squares and a solid line, FI{Delta} versus GM-CSF direct comparisons.

 
We also examined genes for which higher levels of expression are associated with proliferation and self-renewal. We identified these on the basis of their maintained expression at 72 h in the V449E-expressing FDB1 population and their down-regulation in the FI{Delta} population or in cells responding to GM-CSF. Table 3 lists the 175 proliferation-associated genes (for full gene annotations, see Supplementary Table 4). Several of the genes in Table 3 have known ability to block differentiation or have a reported association with an immature hematopoietic cell phenotype (Table 2) . Example gene-expression profiles are presented in Figure 5B . In particular, the proto-oncogene c-Myb has known roles in regulation of myeloid cell proliferation (for review, see ref. [53 ]), and the microarray profile for c-Myb was verified using Northern analysis (see Supplementary Fig. 1). As a representative of this class of genes and to test the capacity of the FDB1 cell line to respond to a known myeloid oncogene, we used a retroviral vector to enforce expression of c-Myb in FDB1 cells. Figure 6 shows that FDB1 cells infected with a MSCV-IRES-GFP retroviral vector expressing c-Myb are blocked in differentiation in the presence of GM-CSF. In addition, we have shown that these cells can be maintained in GM-CSF long-term, consistent with overexpression of c-Myb generating a complete block in myeloid differentiation. Although this is not unanticipated, given the transforming capacity of Myb [54 ], nevertheless, it highlights the ability of our expression analysis to identify genes that function as hematopoietic regulators and the ability of such genes to perturb growth and differentiation in the FDB1 model system.


Figure 6
View larger version (82K):
[in this window]
[in a new window]
  		Figure 6. Growth and differentiation effects of enforced c-Myb expression in FDB1 cells, which were transduced with a MSCV-IRES-GFP retrovirus encoding c-Myb or with control retrovirus and selected on the basis of GFP expression. Cells were washed to remove growth factor and then cultured in mIL-3 or GM-CSF for 5 days.

 
We conclude from this analysis that the FDB1 model displays appropriate gene regulation for myeloid differentiation and that genes identified by this approach are therefore highly likely to have a role in these processes in vivo. The functional testing with c-Myb demonstrates that this cell line will be an extremely useful model of myeloid growth and differentiation, facilitating further analysis of genes identified in this study.

GO and divisive clustering to identify novel candidate myeloid transcriptional regulators
We next used this comprehensive data set to perform a level of gene prioritization aiming to highlight key genes with potential regulatory function. This process greatly facilitates the future functional analysis required to establish cause and effect for any genes of interest. With a view to identification of key myeloid regulators, we used GO terms to identify 340 TFs and TF-associated proteins displaying differential regulation consistent with a potential regulatory role (see Materials and Methods for details). Proteins in these categories can influence cell fate and the switch between proliferation and differentiation [55 ] and are often involved in leukemic translocations resulting in fusion proteins that have the capacity to reprogram myeloid cell behavior [1 ]. As TFs with regulatory roles in hematopoiesis often have well-defined patterns of expression, we used divisive, hierarchical clustering on gene-expression profiles under all conditions to group together genes that display coordinate regulation. The clustering approach identified eight clusters (237 genes) associated with proliferation (down-regulated during differentiation) and six clusters (103 genes) associated with differentiation (see Supplementary Table 1 for full gene annotation information). Within these divisions, clusters grouped together genes that had a similar rate or magnitude of gene-expression change. A heat map showing all clusters is presented in Figure 7 . To select genes of particular interest from this group, we focused on genes that clustered with genes known to have roles in myelopoiesis or leukemogenesis. This information was taken together with evidence available from the literature (Entrez PubMed, iHOP [56 ]) to generate a group of genes with a high probability of having regulatory capacity in the myeloid lineage. These genes are discussed briefly below. Relevant references are summarized in Supplementary Table 5.


Figure 7
View larger version (75K):
[in this window]
[in a new window]
  		Figure 7. Heat map displaying TF-associated clusters. Heat map showing mean expression level for each comparison performed for 340 genes, significantly, differentially expressed at 0, 6, or 72 h. These genes were identified as TF-associated, based on GO classification or gene description. The tree obtained from divisive clustering is presented on the left-hand side of the plot. Fourteen clusters were defined by cutting the tree at a height of 1.25 and are indicated using the colored bands on the left-hand side. Numbering of the clusters starts from the bottom (red) cluster and proceeds upward (Cluster 6, dark blue; Cluster 9, chartreuse; Cluster 10, medium blue; Cluster 12, light blue). The 340 rows represent individual probes on the array, and the 28 columns represent the points in the time-course analysis. Heat map values are the expression values estimated using a linear modeling approach scaled across rows, and red indicates up-regulation, and green indicates down-regulation. The size of each band gives an indication of the magnitude of the change, and color gives an indication of the direction and magnitude of the change. (Full annotation details may be found in Supplementary Table 1.) Top horizontal band indicates the microarray comparison (V/F indicates matched time comparisons of the V449E cell population against the FI{Delta} population; G/F indicates matched time comparison of the GM-CSF treated cells against the FI{Delta} cell population; IL-3 indicates the comparison of the V449E cell population with parental cells in IL-3). Bottom band indicated the time point.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Novel regulators of myeloid proliferation and self-renewal
The proliferation-associated cluster (#6) contains the proto-oncogene c-Myb, discussed above, and a number of other genes involved in the pathogenesis of leukemia through translocations and/or overexpression, including Runx1 [57 ], Scl [58 ], Calreticulin [59 ], Gata2 [60 ], and ETO2 [61 ], which is highly related to ETO, the founding member of the ETO gene family, and is a partner of AML1/Runx1 in the common t(8;21) translocation in AML. ETO2 itself has also been found to form a translocation fusion product with AML1 in therapy-induced t(16;21) myeloid leukemias; it may affect transcription of several target genes via interaction with the corepressor N-CoR. It is also a target of retroviral insertion in multiple hematopoietic tumors. A recent report describing ETO2 down-regulation during myeloid differentiation agrees with our finding (see Fig. 4 ) [62 ]. The Ikaros family of TFs was also of particular interest; Eos is present in Cluster 6, and Helios is the sole member of Cluster 8. Helios is known to share some of the functions of Ikaros, with which it can dimerize, but unlike Ikaros, its roles in different hematopoietic lineages are largely undefined [63 ]. Interrogation of the retroviral-tagged cancer gene database [64 ] (http://rtcgd.ncifcrf.gov/), using the transcriptional regulators in Cluster 6, revealed Sox4 as a novel gene of interest, as it is the target of 63 retroviral insertions in a range of hematopoietic tumors, including myeloid tumors. A role for Sox4 in myeloid leukemogenesis has been confirmed recently by retroviral-mediated, forced expression analysis in an animal model [65 ]. Cluster 6 also contained the Emx2 homeobox gene, which has been shown to induce alterations in cell fate or changes in behavior of neural cells but has no known role in the myeloid lineage [66 ]. The transcriptional changes for Helios, Sox4, and Emx2 were confirmed by QRT-PCR, and the results can be seen in Figure 4 .

Novel regulators of myeloid differentiation

We also studied a number of transcription-associated factors, which display increased levels of expression, concomitant with growth arrest and onset of morphological differentiation. These have the potential to be important initiators of myeloid differentiation. Clusters 9 and 10 were closely related and contained genes with gradual increases in expression over time in cells undergoing GM differentiation in the GM-CSF and FI{Delta} conditions. These clusters included genes known to be involved in inducing myeloid differentiation such as PU.1 [4 ],Notch1 [67 ],Cdkn2a [68 , 69 ], and Egr2 [70 ]. As would be predicted for this class of genes, a number of them (e.g., PU.1 and Cdkn2a) can act as myeloid tumor suppressors. Of novel interest to the myeloid system was the circadian rhythm gene, Per2 [71 ], which functions as a tumor suppressor. It is important that a recent study identified Per2 as a C/EBP{alpha} target gene in myeloid cell lines and showed that enforced Per2 expression induced tumor suppressor activity in leukemic cell lines [72 ]. Other genes of interest in Clusters 9 and 10 are Btg2 [73 , 74 ] and Tcf7l2 [75 ].Btg2 is an antiproliferative, p53-dependent transcriptional coregulator, which may also be a target of c-Myb repression in myeloid cells. The increase in Btg2 gene expression was also confirmed by QRT-PCR (Fig. 4) .

Genes in Cluster 12 generally displayed early and large increases in gene expression (especially in FI{Delta}-induced differentiation). This cluster contained six genes, and of these, Fos [76 ], Egr1 [55 ], and Hlx1 [77 ] are known inducers of myeloid differentiation, making the other three genes (Egr3 [78 ], Klf5 [79 , 80 ], and an uncharacterized expressed sequence tag clone) of potential interest. Egr3 has not been reported to be important in the myeloid lineage but has known roles in the T cell lineage. Klf5 has been reported to be a direct target of C/EBP{alpha} in myeloid cells and is involved in the C/EBP cascade that regulates adipocyte differentiation [80 ]. The transcriptional regulation of Klf5 was confirmed by QRT-PCR and can be seen in Figure 4 . A key to understanding the role of these potential novel regulators in the switch between myeloid proliferation and differentiation will be functional analysis approaches in the FDB1 cell line as described above, prior to studies in primary cell and in vivo systems.

A common gene-expression signature between the V449E-activated mutant and human AML: involvement of C/EBP{alpha}
Given the selective leukemogenic capacity of the V449E mutant, we wished to identify gene-expression patterns associated specifically with signaling from this activated mutant. As shown in Figure 3B , there are a striking number of gene-expression differences unique to the V449E cell population. In particular, a total of 44 genes was down-regulated 1.4-fold or more (maximum, eightfold) in the V449E population, relative to the FI{Delta} and parental populations in IL-3 (F-test P value 1x10–5). Fold changes for these genes relative to parental FDB1 cells in IL-3 are shown in Table 4 (for full gene annotations, see Supplementary Table 6). Closer inspection of the V449E-down-regulated gene list revealed that 10 of the 44 genes are known C/EBP{alpha} and/or C/EBP{epsilon} target genes (marked with asterisks in Table 4 ). To investigate this potential link with C/EBP activity further, we examined promoter regions of this gene set using the Cureos TF/promoter database (http://cureos.wehi.edu.au). A total of 15 genes contained high-quality C/EBP consensus-binding sequences conserved between mouse and human matches (Transfac matrix V$CEBP_Q3). Given that 4706 of the 17,731 unique genes on the array contained consensus sites, this suggests an over-representation of CEBP-binding sites ({chi}2 test P value of 0.25). To investigate this link further, we generated a larger gene set by relaxing the requirement that genes must be down-regulated by more than 1.4-fold compared with the parental and FI{Delta} population in IL-3 (but retaining the F-test, P value of 1x10–5). Using this gene set, we observed that 80 of the 203 down-regulated genes contained consensus-binding sites, which represents a significant enhancement ({chi}2 test P value of 3x10–5). This suggests that the V449E down-regulated group of genes may contain many novel C/EBP targets in addition to those known targets already identified, and it suggests a model in which V449E signaling affects C/EBP activity, which contributes to a block in differentiation of FDB1 cells [34 ] and contributing to AML in vivo.

To test whether this V449E-specific gene signature was related to the changes observed in primary human leukemias, we compared the V449E gene set (Table 4) with a number of AML signatures identified using gene-expression profiling [51 ]. For this, we based our analysis on the gene set-enrichment approach of Mootha et al. [87 ] but adapted the approach for across the microarray platform use by using the Wilcoxon Rank Sum tests. This analysis revealed a highly significant association (P=0.0006) with a specific AML cluster (Cluster 4 in Valk et al. [51 ]; Table 5 ). Eight of the 15 leukemias in this cluster carried mutations in the C/EBP{alpha} gene, suggesting that a loss of C/EBP{alpha} activity is central to this leukemic phenotype. Taken together, these findings provide further support for a model in which V449E signaling affects C/EBP activity, which is important for is leukemic properties. The FDB1 cell line model provides a system with manipulability, which can now be used to study the mechanism of dysregulation of C/EBP{alpha} by activated receptor signaling and the target genes that may be central to AML pathogenesis.


View this table:
[in this window]
[in a new window]
  		Table 5. Association of the V449E-Associated Gene Set with AML Subtypes Defined by Valk et al. [51 ]

 
In summary, the combination of the FDB1 model of myeloid differentiation and sophisticated array design and analysis tools has identified several genes of interest with regard to regulation of myeloid growth and differentiation and the changes that are associated with myeloid leukemogeneis. The ability of the FDB1 cell line to switch between continued growth and differentiation makes it an excellent primary, functional screening tool, and we have validated this system for functional studies using forced overexpression of the c-Myb proto-oncogene. By using approaches to compare this data set with a myeloid leukemia gene set, we have been able to generalize findings beyond this system. With a view to linking key signaling events to transcriptional programs, we are also investigating the nature of signaling from these activated receptors and differences in gene activation as a result of second-site receptor mutations or treatment with signaling pathway inhibitors, which affect selective cellular outcomes [40 ]. This type of analysis provides an opportunity to further correlate specific genes and clusters with myeloid cell proliferation, survival, differentiation, and leukemic receptor signaling and to determine which regions of the GM-CSF receptor and its associated signaling pathways induce these transcriptional changes.


   ACKNOWLEDGEMENTS
 
This work was supported by the U.S. National Institutes of Health Award R01 HL60657. R. J. D. was supported by the Peter Nelson Leukaemia Research Fellowship. A. L. B. is a fellow of the Leukaemia Foundation of Australia. We are grateful to Dr. Simon Barry for his critical input into this project. We also thank Mrs. Silvia Nobbs for her help with flow cytometry and Mrs. Michelle Perugini for assistance with figure preparation. The data set from this study is available at GEO GSE3333.


   FOOTNOTES
 
1 These authors contributed equally to this work. Back

Received February 23, 2006; revised March 20, 2006; accepted March 23, 2006.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Tenen, D. G. (2003) Disruption of differentiation in human cancer: AML shows the way Nat. Rev. Cancer 3,89-101[CrossRef][Medline]
  2. Zhu, J., Emerson, S. G. (2002) Hematopoietic cytokines, transcription factors and lineage commitment Oncogene 21,3295-3313[CrossRef][Medline]
  3. Chen, H., Ray-Gallet, D., Zhang, P., Hetherington, C. J., Gonzalez, D. A., Zhang, D. E., Moreau-Gachelin, F., Tenen, D. G. (1995) PU.1 (Spi-1) autoregulates its expression in myeloid cells Oncogene 11,1549-1560[Medline]
  4. Nerlov, C., Graf, T. (1998) PU.1 induces myeloid lineage commitment in multipotent hematopoietic progenitors Genes Dev. 12,2403-2412[Abstract/Free Full Text]
  5. Radomska, H. S., Huettner, C. S., Zhang, P., Cheng, T., Scadden, D. T., Tenen, D. G. (1998) CCAAT/enhancer binding protein {alpha} is a regulatory switch sufficient for induction of granulocytic development from bipotential myeloid progenitors Mol. Cell. Biol. 18,4301-4314[Abstract/Free Full Text]
  6. Rosmarin, A. G., Yang, Z., Resendes, K. K. (2005) Transcriptional regulation in myelopoiesis: hematopoietic fate choice, myeloid differentiation, and leukemogenesis Exp. Hematol. 33,131-143[CrossRef][Medline]
  7. Kondo, M., Scherer, D. C., Miyamoto, T., King, A. G., Akashi, K., Sugamura, K., Weissman, I. L. (2000) Cell-fate conversion of lymphoid-committed progenitors by instructive actions of cytokines Nature 407,383-386[CrossRef][Medline]
  8. Gilliland, D. G., Jordan, C. T., Felix, C. A. (2004) The molecular basis of leukemia Hematology (Am. Soc. Hematol. Educ. Program) ,80-97
  9. Gilliland, D. G., Griffin, J. D. (2002) The roles of FLT3 in hematopoiesis and leukemia Blood 100,1532-1542[Abstract/Free Full Text]
  10. Forbes, L. V., Gale, R. E., Pizzey, A., Pouwels, K., Nathwani, A., Linch, D. C. (2002) An activating mutation in the transmembrane domain of the granulocyte colony-stimulating factor receptor in patients with acute myeloid leukemia Oncogene 21,5981-5989[CrossRef][Medline]
  11. Carapeti, M., Soede-Bobok, A., Hochhaus, A., Sill, H., Touw, I. P., Goldman, J. M., Cross, N. C. (1997) Rarity of dominant-negative mutations of the G-CSF receptor in patients with blast crisis of chronic myeloid leukemia or de novo acute leukemia Leukemia 11,1005-1008[CrossRef][Medline]
  12. Gari, M., Goodeve, A., Wilson, G., Winship, P., Langabeer, S., Linch, D., Vandenberghe, E., Peake, I., Reilly, J. (1999) c-kit proto-oncogene exon 8 in-frame deletion plus insertion mutations in acute myeloid leukaemia Br. J. Haematol. 105,894-900[CrossRef][Medline]
  13. Benekli, M., Baer, M. R., Baumann, H., Wetzler, M. (2003) Signal transducer and activator of transcription proteins in leukemias Blood 101,2940-2954[Abstract/Free Full Text]
  14. Bentires-Alj, M., Paez, J. G., David, F. S., Keilhack, H., Halmos, B., Naoki, K., Maris, J. M., Richardson, A., Bardelli, A., Sugarbaker, D. J., et al (2004) Activating mutations of the noonan syndrome-associated SHP2/PTPN11 gene in human solid tumors and adult acute myelogenous leukemia Cancer Res. 64,8816-8820[Abstract/Free Full Text]
  15. Radich, J. P., Kopecky, K. J., Willman, C. L., Weick, J., Head, D., Appelbaum, F., Collins, S. J. (1990) N-ras mutations in adult de novo acute myelogenous leukemia: prevalence and clinical significance Blood 76,801-807[Abstract/Free Full Text]
  16. Young, D. C., Griffin, J. D. (1986) Autocrine secretion of GM-CSF in acute myeloblastic leukemia Blood 68,1178-1181[Abstract/Free Full Text]
  17. Jiang, X., Lopez, A., Holyoake, T., Eaves, A., Eaves, C. (1999) Autocrine production and action of IL-3 and granulocyte colony- stimulating factor in chronic myeloid leukemia Proc. Natl. Acad. Sci. USA 96,12804-12809[Abstract/Free Full Text]
  18. Rowley, J. D. (1998) The critical role of chromosome translocations in human leukemias Annu. Rev. Genet. 32,495-519[CrossRef][Medline]
  19. Mizuki, M., Schwable, J., Steur, C., Choudhary, C., Agrawal, S., Sargin, B., Steffen, B., Matsumura, I., Kanakura, Y., Bohmer, F. D., Muller-Tidow, C., Berdel, W. E., Serve, H. (2003) Suppression of myeloid transcription factors and induction of STAT response genes by AML-specific Flt3 mutations Blood 101,3164-3173[Abstract/Free Full Text]
  20. Dugas, M., Schoch, C., Schnittger, S., Haferlach, T., Danhauser-Riedl, S., Hiddemann, W., Messerer, D., Uberla, K. (2001) A comprehensive leukemia database: integration of cytogenetics, molecular genetics and microarray data with clinical information, cytomorphology and immunophenotyping Leukemia 15,1805-1810[Medline]
  21. Schoch, C., Kohlmann, A., Schnittger, S., Brors, B., Dugas, M., Mergenthaler, S., Kern, W., Hiddemann, W., Eils, R., Haferlach, T. (2002) Acute myeloid leukemias with reciprocal rearrangements can be distinguished by specific gene expression profiles Proc. Natl. Acad. Sci. USA 99,10008-10013[Abstract/Free Full Text]
  22. Virtaneva, K., Wright, F. A., Tanner, S. M., Yuan, B., Lemon, W. J., Caligiuri, M. A., Bloomfield, C. D., de la Chapelle, A., Krahe, R. (2001) Expression profiling reveals fundamental biological differences in acute myeloid leukemia with isolated trisomy 8 and normal cytogenetics Proc. Natl. Acad. Sci. USA 98,1124-1129[Abstract/Free Full Text]
  23. Armstrong, S. A., Staunton, J. E., Silverman, L. B., Pieters, R., den Boer, M. L., Minden, M. D., Sallan, S. E., Lander, E. S., Golub, T. R., Korsmeyer, S. J. (2002) MLL translocations specify a distinct gene expression profile that distinguishes a unique leukemia Nat. Genet. 30,41-47[CrossRef][Medline]
  24. Alcalay, M., Meani, N., Gelmetti, V., Fantozzi, A., Fagioli, M., Orleth, A., Riganelli, D., Sebastiani, C., Cappelli, E., Casciari, C., et al (2003) Acute myeloid leukemia fusion proteins deregulate genes involved in stem cell maintenance and DNA repair J. Clin. Invest. 112,1751-1761[CrossRef][Medline]
  25. Dorsam, S. T., Ferrell, C. M., Dorsam, G. P., Derynck, M. K., Vijapurkar, U., Khodabakhsh, D., Pau, B., Bernstein, H., Haqq, C. M., Largman, C., Lawrence, H. J. (2004) The transcriptome of the leukemogenic homeoprotein HOXA9 in human hematopoietic cells Blood 103,1676-1684[Abstract/Free Full Text]
  26. Neben, K., Schnittger, S., Brors, B., Tews, B., Kokocinski, F., Haferlach, T., Muller, J., Hahn, M., Hiddemann, W., Eils, R., Lichter, P., Schoch, C. (2005) Distinct gene expression patterns associated with FLT3- and NRAS-activating mutations in acute myeloid leukemia with normal karyotype Oncogene 24,1580-1588[CrossRef][Medline]
  27. Lacayo, N. J., Meshinchi, S., Kinnunen, P., Yu, R., Wang, Y., Stuber, C. M., Douglas, L., Wahab, R., Becton, D. L., Weinstein, H., et al (2004) Gene expression profiles at diagnosis in de novo childhood AML patients identify FLT3 mutations with good clinical outcomes Blood 104,2646-2654[Abstract/Free Full Text]
  28. Furumoto, Y., Gonzalez-Espinosa, C., Gomez, G., Kovarova, M., Odom, S., Parravicini, V., Ryana, J. J., Rivera, J. (2004) Rethinking the role of Src family protein tyrosine kinases in the allergic response: new insights on the functional coupling of the high affinity IgE receptor Immunol. Res. 30,241-253[CrossRef][Medline]
  29. Zheng, X., Ravatn, R., Lin, Y., Shih, W. C., Rabson, A., Strair, R., Huberman, E., Conney, A., Chin, K. V. (2002) Gene expression of TPA induced differentiation in HL-60 cells by DNA microarray analysis Nucleic Acids Res. 30,4489-4499[Abstract/Free Full Text]
  30. Song, J. H., Kim, J. M., Kim, S. H., Kim, H. J., Lee, J. J., Sung, M. H., Hwang, S. Y., Kim, T. S. (2003) Comparison of the gene expression profiles of monocytic versus granulocytic lineages of HL-60 leukemia cell differentiation by DNA microarray analysis Life Sci. 73,1705-1719[CrossRef][Medline]
  31. Bruno, L., Hoffmann, R., McBlane, F., Brown, J., Gupta, R., Joshi, C., Pearson, S., Seidl, T., Heyworth, C., Enver, T. (2004) Molecular signatures of self-renewal, differentiation, and lineage choice in multipotential hemopoietic progenitor cells in vitro Mol. Cell. Biol. 24,741-756[Abstract/Free Full Text]
  32. Juan, H. F., Lin, J. Y., Chang, W. H., Wu, C. Y., Pan, T. L., Tseng, M. J., Khoo, K. H., Chen, S. T. (2002) Biomic study of human myeloid leukemia cells differentiation to macrophages using DNA array, proteomic, and bioinformatic analytical methods Electrophoresis 23,2490-2504[CrossRef][Medline]
  33. Theilgaard-Monch, K., Jacobsen, L. C., Borup, R., Rasmussen, T., Bjerregaard, M. D., Nielsen, F. C., Cowland, J. B., Borregaard, N. (2005) The transcriptional program of terminal granulocytic differentiation Blood 105,1785-1796[Abstract/Free Full Text]
  34. McCormack, M. P., Gonda, T. J. (2000) Novel murine myeloid cell lines which exhibit a differentiation switch in response to IL-3 or GM-CSF, or to different constitutively active mutants of the GM-CSF receptor {alpha} subunit Blood 95,120-127[Abstract/Free Full Text]
  35. Gonda, T. J., D’Andrea, R. J. (1997) Activating mutations in cytokine receptors: implications for receptor function and role in disease Blood 89,355-369[Free Full Text]
  36. D’Andrea, R. J., Gonda, T. J. (2000) A model for assembly and activation of the GM-CSF, IL-3 and IL-5 receptors. Insights from activated mutants of the common ß subunit Exp. Hematol. 28,231-243[CrossRef][Medline]
  37. Jenkins, B. J., Blake, T. J., Gonda, T. J. (1998) Saturation mutagenesis of the ß subunit of the human granulocyte-macrophage colony-stimulating factor receptor shows clustering of constitutive mutations, activation of ERK MAP kinase and STAT pathways, and differential ß subunit tyrosine phosphorylation Blood 92,1989-2002[Abstract/Free Full Text]
  38. McCormack, M. P., Gonda, T. J. (1999) Myeloproliferative disorder and leukaemia induced in mice by different classes of constitutive mutants of the human IL-3/IL-5/GM-CSF receptor common ß subunit Oncogene 18,7190-7199[CrossRef][Medline]
  39. D’Andrea, R. J., Harrison-Findik, D., Butcher, C. M., Finnie, J., Blumbergs, P., Bartley, P., McCormack, M., Jones, K., Rowland, R., Gonda, T. J., Vadas, M. A. (1998) Dysregulated hematopoiesis and a progressive neurological disorder induced by expression of an activated form of the human common ß chain in transgenic mice J. Clin. Invest. 102,1951-1960[Medline]
  40. Brown, A. L., Peters, M., D’Andrea, R. J., Gonda, T. J. (2004) Constitutive mutants of the GM-CSF receptor reveal multiple pathways leading to myeloid cell survival, proliferation, and granulocyte-macrophage differentiation Blood 103,507-516[Abstract/Free Full Text]
  41. Jenkins, B. J., D’Andrea, R., Gonda, T. J. (1995) Activating point mutations in the common ß subunit of the human GM-CSF, IL-3 and IL-5 receptors suggest the involvement of ß subunit dimerization and cell type-specific molecules in signaling EMBO J. 14,4276-4287[Medline]
  42. Metcalf, D. (1984) The Haemopoietic Colony Stimulating Factors Elsevier Amsterdam.
  43. Yang, Y. H., Buckley, M. J., Speed, T. P. (2001) Analysis of cDNA microarray images Brief. Bioinform. 2,341-349[Abstract/Free Full Text]
  44. Ritchie, M. E. (2004) Quantitative Quality Control and Background Correction for Two-Color Microarray Data ,1-139 University of Melbourne Melbourne. Ph.D. thesis
  45. Smyth, G. K. (2004) Linear models and empirical Bayes methods for assessing differential expression in microarray experiments Stat. Appl. Genet. Mol. Biol. 3Article 3.
  46. Gentleman, R. C., Carey, V. J., Bates, D. M., Bolstad, B., Dettling, M., Dudoit, S., Ellis, B., Gautier, L., Ge, Y., Gentry, J., et al (2004) Bioconductor: open software development for computational biology and bioinformatics Genome Biol. 5,R80[CrossRef][Medline]
  47. Yang, Y. H., Dudoit, S., Luu, P., Lin, D. M., Peng, V., Ngai, J., Speed, T. P. (2002) Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation Nucleic Acids Res. 30,e15[Abstract/Free Full Text]
  48. Cui, X., Kerr, M. K., Churchill, G. A. (2003) Transformations for cDNA microarray data Stat. Appl. Genet. Mol. Biol. 2Article 4.
  49. Beissbarth, T., Speed, T. P. (2004) GOstat: find statistically overrepresented gene ontologies within a group of genes Bioinformatics 20,1464-1465[Abstract/Free Full Text]
  50. Bar-Joseph, Z., Gerber, G., Simon, I., Gifford, D. K., Jaakkola, T. S. (2003) Comparing the continuous representation of time-series expression profiles to identify differentially expressed genes Proc. Natl. Acad. Sci. USA 100,10146-10151[Abstract/Free Full Text]
  51. Valk, P. J., Verhaak, R. G., Beijen, M. A., Erpelinck, C. A., Barjesteh van Waalwijk van Doorn-Khosrovani, S., Boer, J. M., Beverloo, H. B., Moorhouse, M. J., van der Spek, P. J., Lowenberg, B., Delwel, R. (2004) Prognostically useful gene-expression profiles in acute myeloid leukemia N. Engl. J. Med. 350,1617-1628[Abstract/Free Full Text]
  52. Glonek, G. F. V., Solomon, P. J. (2004) Factorial and time course designs for cDNA microarray experiments Biostatistics 5,89-111[Abstract]
  53. Gonda, T. J. (1998) The c-Myb oncoprotein Int. J. Biochem. Cell Biol. 30,547-551