Transcriptomic and Phospho-Proteomic Analyses of Erythroblasts Expanded in vitro From Normal Donors and From Patients with Polycythemia Vera

Abstract

Erythropoiesis is a tightly regulated process which becomes decoupled from its normal differentiation program in patients with polycythemia vera (PV). Somatic mutations in JAK2 are commonly associated with this myeloid proliferative disorder. To gain insight into the molecular events that are required for abnormally developing erythroid cells to escape dependence on normal growth signals, we performed in vitro expansion of mature erythroblasts (ERY) from 7 normal healthy donors and from 7 polycythemic patients in the presence of IL3, EPO, SCF for 10, 11 or 13 days. Normal ERYs required exposure to the glucocorticoid dexamethasone (Dex) for expansion, while PV-derived ERYs expanded in the absence of dexamethasone. RNA expression profiling revealed enrichment of two known oncogenes, GPR56 and RAB4a, in PV-derived ERYs along with reduced expression levels of transcription factor TAL1 (ANOVA FDR < 0.05). While both normal and polycythemic-derived ERYs integrated signaling cascades for growth, they did so via different signaling pathways which are represented by their differential phospho-profiles. Our results show that normal ERYs displayed greater levels of phosphorylation of EGFR, PDGFRβ, TGFβ and cKit, while PV-derived ERYs were characterized by increased phosphorylation of cytoplasmic kinases in the JAK/STAT, PI3K and GATA1 pathways. Together these data suggest that PV erythroblast expansion and maturation may be maintained and enriched in the absence of dexamethasone through reduced TAL1 expression and by accessing additional signaling cascades. Members of this acquired repertoire may provide important insight into the pathogenesis of aberrant erythropoiesis in myeloproliferative neoplasms such as polycythemia vera.

Introduction

Polycythemia vera (PV) is an acquired clonal myeloid proliferative disorder characterized by abnormally high red blood cell counts, and is commonly accompanied by somatic, activating mutations in JAK2 [1, 2]. Sustained JAK2 activation has been shown to confer constitutive growth/survival signaling by activating specific cell signaling pathways, including STAT transcription factors and other second messenger proteins. Furthermore, JAK/STAT signaling has been shown to promote erythroid differentiation and self-renewal during normal erythropoiesis. For example, CD34+ cells isolated from cord blood, cultured on murine stromal cells, and transfected with an activating STAT5A construct demonstrate increased self-renewal and selective maturation toward the erythroid lineage[3]. Others have further developed in vitro models of normal erythroblast expansion and maturation. For example erythroid colony forming units obtained from CD34+ cells from normal healthy peripheral blood, when treated with erythropoietin (EPO) induce tyrosine phosphorylation of JAK2, STAT5A and STAT5B with subsequent nuclear translocation of the phosphorylated STAT5A[4], suggesting that JAK/STAT signaling is a second messenger system triggered by growth factor stimulation during normal erythropoiesis in vitro. Furthermore, optimized growth conditions that allow for erythroblast expansion require the addition of dexamethasone (Dex), the agonist for the glucocorticoid receptor (GR), to blood purified mono-nuclear cells (MNCs) which are also stimulated with the erythroid specific growth factors EPO, interleukin-3 (IL-3) and stem cell factor (SCF). Culturing of erythroblasts under these conditions results in a block in the proerythroblast maturation and sustained self-renewal[5–7].

These studies of normal erythroblast expansion and maturation have been applied to in vitro models of PV which revealed a maturation and expansion phenotype distinct from that observed during the normal erythroid differentiation program. Specifically experiments conducted on bone marrow-derived erythroblasts collected from PV patients have shown that proliferation and maturation of erythroid progenitors can occur via IL-3-dependent, EPO-independent signaling[11, 12]. Furthermore studies of erythroblasts expanded in vitro from PV-derived MNC cells have displayed hyper-proliferative growth, accelerated maturation, reduced apoptosis and expression of the dominant negative beta-isoform of the glucocorticoid receptor[8–10]. These previous studies suggest that abnormal erythropoiesis is conferred by acquiring sustained growth and survival signals downstream of their normally required membrane receptor-ligand complexes.

Since JAK2 can signal through the nuclear compartment and directly activate the expression of JAK-STAT target genes, we performed RNA expression profiling of transcripts differentially expressed in ERYs derived from the peripheral blood of normal healthy volunteers and polycythemic patients. Our analysis provides further insight into the transcriptional and biochemical events which are specific to either the normal or abnormal development of the erythroblast lineage in vitro. RNA expression profiling of CD34+ cells isolated from cord blood of normal bone marrow and cultured under a range of supportive conditions has been described previously [13–18]. Additionally, others have directly compared the transcription profiles of normal and PV samples containing purified granulocytes, neutrophils or whole blood. Our study compares the RNA expression profiles of erythroid cells expanded from progenitors resident in peripheral blood of normal donors and PV patients. Analysis of the erythroid cells allows us to identify transcripts that may be important in the maintenance of erythroblast renewal and maturation during normal or abnormal erythroblast differentiation programs.

In order to better understand the cascade of signaling events required for specific expansion of erythroblasts in normal and polycythemic individuals, we also performed Reverse-Phase Protein Array (RPPA) analysis of protein lysates collected from erythroblasts obtained from normal healthy donors and from PV patients. One hundred eighty proteins were screened for differential phosphorylation or cleavage allowing us to observe the signaling events that occur during normal and abnormal erythropoietic differentiation programs in vitro. Of the 180 proteins analyzed, 39 showed differential phosphorylation/cleavage between normal and PV derived erythroblasts. Our results suggest that alternate signaling events distinct from those used during normal erythroid development are present during abnormal erythropoiesis and these events integrate self-renewal and maturation pathways which subsequently trigger the loss of the transcriptional machinery that programs normal erythroid development.

Methods

Subjects

Peripheral blood buffy coat was collected from 7 healthy donors (ND) and 7 polycythemia vera (PV) patients at the transfusion center of “La Sapienza” University (Rome, Italy) and from the tissue bank of the Myeloproliferative Disease Research Consortium (MPD-RC). Specimens were collected and provided for this study according to guidelines established by the institutional ethical committees and all diagnoses were established according to WHO criteria.

Liquid Culture of Human Erythroblasts

Mono-nuclear cells (MNCs) were separated by Ficoll-Hypaque density centrifugation (ρ < 1.077; Amersham Pharmacia Biotech, Uppsala, Sweden). MNCs (10⁶ cells/mL) were cultured under human erythroid massive amplification (HEMA) conditions stimulated with human stem cell factor (SCF 100 ng/mL, Sigma Aldrich), erythropoietin (EPO 5 U/mL, Neorecormon, Auckland, New Zealand), human interleukin-3 (IL-3, 1 ng/mL, Biosource, San Jose, CA, USA), estradiol (ES, 10⁻⁶M, Sigma, St. Louis, MO, USA) with or without Dexamethasone (Dex, 10⁻⁶M, Sigma, St. Louis, MO, USA), as described previously [19]. After 10, 11 or 13 days, cells were collected for further analyses as described below.

Morphological Evaluation

Erythroid maturation state was assessed according to standard morphological criteria by visual examination of cyto-centrifuged cell preparations (Cytospin 3, Shandon, Astmoor, England) which were air-dried, fixed in −20°C methanol and stained with May-Grünwald-Giemsa (Fisher Scientific, Pittsburgh, PA, USA).

Flow Cytometry

For each sample, 0.5 × 10⁵ – 1.0 × 10⁵ cells were suspended in Ca²⁺Mg²⁺-free phosphate buffered saline, supplemented with 0.5% BSA, and labeled with phycoerythrin (PE)-conjugated CD235a (anti-glycophorin A) and fluorescein isothiocyanate (FITC)-conjugated CD36 (anti-thrombospondin receptor) or CD45 or appropriate isotype controls (all from Becton Dickinson Biosciences, Franklin Lakes, NJ USA) while non-living cells were identified by propidium iodide staining (5 µg/mL, Sigma). All antibodies were incubated for 30 minutes in the dark and on ice. Non-living cells were excluded by SYTOX Blue (0.002 mM, Molecular Probes) staining. After staining, cells were assayed using BD FACS Aria (Becton Dickinson and Company, Franklin Lakes, NJ.) and the results were analyzed using BD FACS Diva software version 6.1.3.

RNA Isolation and Gene Expression Profiling

Total RNA was extracted separately from ERYs using Trizol (Invitrogen-Life Technologies, Paisley, UK) according to manufactures guidelines. Total RNA concentration was measured using NanoDrop (Thermo Scientific, Wilmington, DE, USA). 100 ng RNA was reverse transcribed using High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA, USA) and the subsequent cDNA was diluted with the HotStarTaq Master Mix (Qiagen, GmbH, Hilden, Germany). Labeling and hybridization of cRNA to HumanHT-12 V4 R2 Bead Chip arrays (Illumina, San Diego, CA) was carried out as described in [20, 21]. Raw data from each chip was exported from Genome Studio (Illumina, San Diego, CA), log₂ transformed, quantile normalized and differential expression of transcripts in PV and ND derived ERYs were determined using ANOVA with a false discovery rate cut off of 0.05. Hierarchical clustering of differentially expressed transcripts was carried out using the average linkage of their pair-wise Pearson’s dissimilarity and a heat map representing z-score transformed expression values was used to visually contrast the expression values across both ND and PV groups.

Protein Lysate Preparation

Day 10 erythroblasts were obtained from three PV patients and three normal adult donors. Cell pellets were lysed in buffer containing T-PER reagent (Thermo-Fisher Scientific, Waltham, MA), 300 mM NaCl, 1mM orthovanadate (Sigma, St. Louis, MO), 2 mM Pefabloc (Roche, Basel, Switzerland), 5 µg/mL aprotinin (Sigma, St. Louis, MO), 5 µg/mL pepstatin A (Sigma, St. Louis, MO) and 5 µg/mL leupeptin (Sigma, St. Louis, MO) and incubated on ice for 20 minutes. Samples were centrifuged at 9,300 × g for 5 minutes, supernatants were transferred to fresh tubes and total protein concentration was measured in a solution containing 1 µL of cell lysate in 1 mL 50% borate buffer saline preparation and 50% of Coomassie protein assay reagent (Thermo Fisher Scientific, Waltham, MA. Lysates were then diluted for printing in extraction buffer containing 50% T-PER (Thermo Fisher Scientific, Waltham, MA), 47.5% 2× SDS (Invitrogen, Carlsbad, CA) and 2.5% β-mercaptoethanol (Thermo Fisher Scientific, Waltham, MA) to final concentrations of 0.5 µg/µL and 0.125 µg/µL.

Reverse-Phase Protein Array (RPPA) Analysis

All samples were printed in triplicate spots on nitrocellulose-coated glass slides (GRACE Bio-Labs, Bend, OR) using an Aushon 2470 equipped with 185-µm pins (Aushon Biosystems, Billerica, MA) according to the manufacturer's instructions. Reference standard lysates, comprised of HeLa + Pervanadate (BD, Franklin Lakes, NJ), Jurkat + Etoposide (Cell Signaling, Danvers, MA) and Jurkat + Calyculin A (Cell Signaling), were printed in 14-point or 10-point dilution curves as procedural controls and as positive controls for antibody staining. Each reference standard curve was printed in triplicate at concentrations of 0.5 µg/µL and 0.125 µg/µL. A selected subset of the printed array slides were stained with Sypro Ruby Protein Blot Stain (Invitrogen, Carlsbad, CA) to estimate sample total protein concentration and the remaining slides were stored under desiccated conditions at −20°C. Immediately before antibody staining, printed slides were treated with 1× Reblot Mild Solution (Chemicon, Temecula, CA) for 15 minutes, washed 2 × 5 minutes with PBS (Invitrogen, Carlsbad, CA) and incubated for 1 hour in blocking solution (2% I-Block, (Applied Biosystems, Foster City, CA), 0.1% Tween-20 in PBS). Immunostaining was carried out using a signal amplification kit (DAKO, Carpinteria, CA). Arrays were probed with a library of 180 antibodies against antigens to total, cleaved and phosphorylated protein targets. Primary antibody binding was detected using a biotinylated goat anti-rabbit IgG H+L (1:7500) (Vector Laboratories, Burlingame, CA) or rabbit anti-mouse IgG (1:10) (DAKO) followed by streptavidin-conjugated IRDye680 fluorophore (LI-COR Biosciences, Lincoln, NE). Primary antibodies were extensively validated for single band specificity by western blot using complex cellular lysates. Negative control slides were incubated with secondary antibody only. All Sypro and immunostained slides were scanned using a Tecan power scanner™ (Tecan Group Ltd, Switzerland). Acquired images were analyzed with MicroVigene v5.0. (VigeneTech, Carlisle, MA) for spot detection, local background subtraction, negative control subtraction, replicate averaging and total protein normalization. The software package JMP v6 (SAS Institute, Cary, NC) was used to carry out internal standardization, two-way hierarchical clustering using Wards method, and two-groups Wilcoxon Rank-Test (significance cut off p ≤ 0.05).

Statistical methods

Gene expression and protein arrays were analyzed according to specific statistical methods described in the appropriate methods sections. Amplifications levels and biochemical parameters (frequency of CD45 or CD235a positive cells, mRNA and protein cell contents) were compared using analyses of variance methods with Origin 3.5 software for Windows (Microcal Software Inc., Northampton, MA, USA).

Results and Discussion

Patient Clinical Data

Subjects included in this study were either normal donors (ND) or PV patients (PV). Samples collected from ND were provided as de-identified material, therefore the complete blood counts of these normal volunteers were not captured though their blood counts were within the ranges suitable for blood donation. PV patients were heterozygous (2 patients), homozygous (3 patients) or positive with unknown allele burden (1 patient) for the JAK2 V617F mutation or JAK2 wild-type (1 patient, lost to follow up). PV patients were analyzed either at diagnosis (1 patient) or had undergone no treatment (1 patient), phlebotomy +/− chemotherapy or +/− hydroxyurea (5 patients), or chemotherapy alone (1 patient). At the time of blood collection the age of PV donor patients ranged from 28 to 74 years with a average age of 54 years and had a mean 16.0 Hgb/dL (range {12.1, 20.8}, SD 2.8), 52.1% HCT (range {46.7,61.9}, SD 5.6), 12.0 × 10³ WBC/mL (range {6.5 × 10³,21.9 × 10³}, SD 6.5 × 10³) and 59.2 × 10⁶ PLT/mL (range {30.5 × 10⁶, 14.0 × 10⁷}, SD 41.4 × 10⁶).

Biological Properties of Erythroid Cells Cultured in vitro

MNCs from ND generated a greater number of cells after 10–13 days in culture stimulated in the presence of Dex than those without Dex [fold increases (FI) at day 10–12 were 3.5 vs. 0.6, p<0.05 (Table II)]. Additionally day 10–13 cultures free of Dex contained a greater percentage of cells expressing the non-erythroid marker CD45 (39% vs 8% respectively) and presented ERYs from each previously defined maturation stage. As seen previously by others, day 10–13 cultures from PV donors contained mostly immature ERYs (CD45pos cells 1–8%, Figure 2). As a result, the distribution of both the non-erythroid and erythroid cells and the respective maturation states of the ERYs studied were similar in both Dex-dependent normal donor derived and Dex-independent PV derived cultures.

Table II.

Amplification levels (as fold increase (FI) with respect to day 0) and cellular (frequency of cells positive for CD45 or CD235a) and biochemical (mRNA and protein content) characterization of erythroblasts expanded in vitro from Normal Donors (ND) and polycythemia vera patients (PV).

Donor	FI		CD45pos cells (%)		CD235apos cells (%)		Total RNA µg/106 cells		Total Protein µg/106 cells
	+Dex|	−Dex	+Dex	−Dex	+Dex	−Dex	+Dex	−Dex	+Dex	−Dex
ND (n=7) Mean (± SD)	3.5±2.4	0.6±0.5*	8.2±1.3	38.6±17.0*	87.2±4.6	57.4±16.0*	0.77±0.45	nd	48.2±21.8	24.4±16.8
PV (n=7) Mean (± SD)	20.6±16.8**	12.5±9.7	1.6±0.9	8.0±10.0**	96.6±2.6	90.111.0**	5.1±0.39**	5.1±3.8	68.0±5.0	65.1±26.4

^*p < 0.05 (paired t-test) when compared to the values obtained from cultures from the matched cultures treated with dexamethasone

^**p < 0.05 (t-test) when compared to the values obtained from the cultures from Normal Donors

Figure 2. Microarray analysis of erythroblasts expanded from mono-nuclear cells from normal donors and patients having polycythemia vera.

Normal healthy derived erythroblasts expanded from mono-nuclear cells stimulated for 11–13 days with 10 ng/mL SCF, 3 Units/mL EPO and 1 ng/mL IL-3 and 10 uM Dexamethasone (Dex) and Polycythemia vera derived erythroblasts expanded from mono-nuclear cells stimulated with 10 ng/mL SCF, 3 Units/mL EPO and 1 ng/mL IL-3 without dexamethasone. Hypothesis testing using ANOVA with a false-discovery threshold of 0.05 reveals 55 probe sets which are differentially expressed between erythroblasts derived from normal healthy donors and polycythemic patients as shown by hierarchical clustering.

RNA Expression Analysis

Gene expression analysis was performed on erythroblasts from 8 subjects, including 4 normal healthy JAK2 WT donors and 4 PV patients (JAK2 V617F heterozygous 1, homozygous 2, wild type 1). Additionally, 1 technical replicate sample from patient A was included. Erythroblasts from each group were expanded from MNCs for 11 or 13 days in the same manner as those used for the proteomic analysis in a manner described in methods[25]. One hundred nanograms of RNA was reverse transcribed, labeled and hybridized to the Illumina HumanHT-12 V4 R2 Bead Chip array and raw gene level data was log2 transformed, quantile normalized and tested for significantly different expression levels between ND and PV derived ERYs using ANOVA with a false-discovery rate threshold of 0.05.

Hierarchical clustering of the 55 probe sets differentially expressed in ND and PV-derived ERYs (ANOVA FDR < 0.05) showed that the ND ERYs clustered separately from the PV samples. Of the 55 differentially expressed transcripts, 29 were more highly expressed in PV derived ERYs and 26 were more highly expressed in ND ERYs. Differentially expressed transcripts represented proteins involved in cholesterol and divalent ion transport, GPCR-GTPase signaling, mitochondrial signaling/metabolism and regulation of transcription. The cholesterol efflux component ABCA1, which has been demonstrated to mobilize HSPCs into the livers and spleens of Abca1−/− mice [22], is significantly down-regulated in PV-derived ERYs, suggesting that PV derived MNCs may be affected in a similar manner. ND-derived ERYs expressed higher levels of the magnesium ion transporters MAGT1, KCNMB2, HSD18B7 while ND ERYs expressed higher levels of KCNK5. Furthermore ND-derived ERYs expressed higher levels of the adenosine-A1 GPCR ADORA1 and the GTPase EVI5, while PV-derived ERYs expressed higher levels of the adhesion GPCR GPR56 and the known oncogenic GTPase RAB4A. Both GPR56 and RAB4 have been shown to contribute to cellular transformation in different tumor contexts [23, 24] and to regulate the recycling of cell membrane receptors including PDGFR via interactions with the endoplasmic reticulum [25]. This suggests that specific oncogenic effectors are differentially expressed in PV erythroid cells and that these effectors may play a role in maintaining self-renewal during ERY expansion. Mitochondrial genes HMGCL and UCP2 were significantly enriched in ND-derived ERY samples while their organelle-localized counterparts OXCT1, PDSS1, APIP and PRDX3 were more highly expressed in PV-derived samples. Loss of PRDX3 has been observed in Fanconi anemia and implicated in the loss of structural integrity of mitochondria, which display decreased sensitivity to oxidative stress [26]. Notably, TAL1 and ZBTB3 were the only two transcription factors displaying differential expression in our experimental contrast and were both expressed at higher levels in ND ERYs. TAL1 has been shown to act as a positive regulator of erythroid differentiation and is targeted and activated by GATA1 [27, 28]. The observed decrease in TAL1 expression in PV-derived ERYs suggests that EPO signaling for growth and differentiation diverges in PV-derived ERYs, and that decreasing levels of TAL1 mRNA may require PV-derived ERYs to recruit compensatory factors for survival. Non-protein encoding RNAs, including the lncRNA DLEU1 and psuedogenes CYCSL1 and HNRPNA1PA, were significantly enriched in ERYs derived from PV patients. Interestingly, erythroid progenitors from CD34+ bone marrow cells collected from chronic phase CML patients have shown increased DLEU1 expression when compared to normal CD34+ bone marrow controls [29] suggesting DLEU1 expression may have functional links to hematopoietic programming events and that DLEU1 may be recruited by PV-derived ERYs for survival. The complete table of differentially expressed probe sets and their respective significance and fold change distances can be found in Table III.

Table III.

Ilumina gene level probesets differentially expressed in erythroblasts expanded from normal donors and from polycythemia vera patients (ANOVA, FDR ≤ 0.05).

Gene Symbol	p-value	Fold Change	Increased Expression In
TPRKB	2.77E-06	1.51838	PV
PDSS1	8.68E-06	1.78142	PV
ADCY7	1.95E-06	1.4921	PV
PRDX3	8.75E-06	3.30933	PV
UBE3C	9.89E-06	1.40177	PV
NUP54	1.40E-05	1.46017	PV
C5ORF23	1.50E-05	1.03373	PV
RBM3	1.53E-05	1.46075	PV
GPAM	2.03E-05	1.37906	PV
KCNK5	2.15E-05	1.3157	PV
CYCSL1	2.39E-05	1.76543	PV
DPM1	2.56E-05	1.60963	PV
COMMD8	2.88E-05	1.7569	PV
LOC653226	2.96E-05	1.73931	PV
LOC653381	3.53E-05	1.39892	PV
MEX3B	3.71E-05	1.81592	PV
DLEU1	4.20E-05	1.39051	PV
SSH3	4.34E-05	1.1476	PV
RAB4A	4.47E-05	1.3749	PV
OXCT1	4.50E-05	1.51226	PV
HNRPA1P4	4.63E-05	1.86112	PV
GPR56	4.66E-05	1.67202	PV
NCBP2	5.43E-05	1.27028	PV
C5ORF35	5.59E-05	1.36357	PV
MRS2	6.17E-05	1.49052	PV
PRKACB	6.19E-05	1.13546	PV
MGC39900	7.30E-05	1.22888	PV
APIP	7.73E-05	1.73765	PV
TCEAL8	7.88E-05	1.39981	PV
LOC388588	1.01E-07	2.46922	ND
NDST2	6.12E-07	1.32647	ND
UCP2	1.41E-06	2.28552	ND
ADORA1	1.49E-06	1.74723	ND
KIAA1191	2.91E-06	1.4101	ND
LOC650803	2.93E-06	1.62872	ND
TAL1	3.82E-06	1.94648	ND
LOC652968	7.74E-06	1.23378	ND
DCAF10	1.10E-05	1.58601	ND
WDR13	1.54E-05	1.6463	ND
HSD17B7	1.55E-05	1.494	ND
ABCA1	1.71E-05	2.18257	ND
ZBTB3	2.15E-05	1.31531	ND
HMGCL	2.15E-05	1.55359	ND
CA1	3.03E-05	6.29227	ND
CA2	3.19E-05	3.21449	ND
EVI5	3.78E-05	1.34896	ND
SMOX	4.08E-05	4.03663	ND
VPS11	4.11E-05	1.3152	ND
MAGT1	4.36E-05	1.7831	ND
KCNMB2	5.08E-05	1.13657	ND
FKBP8	5.33E-05	1.53016	ND
EPB42	7.09E-05	2.56824	ND
LOC645284	7.34E-05	1.12603	ND
HBBP1	7.63E-05	2.97898	ND
HS.551143	7.77E-05	1.36584	ND

Phosphorylation/Cleavage Profiling by Reverse Phase Protein Array Analysis

Both PV and Dex-treated ND-derived ERYs exhibited activation of multiple canonical proliferation and survival signaling pathways as can be seen in the 39 endpoints which were differentially phosphorylated/cleaved between the two treatment groups (Wilcoxon Rank-Test p < 0.05). These endpoints include cell membrane receptors, cytoplasmic kinases, cleaved caspases, transcriptional effectors and cell cycle regulators. ND-derived ERYs were characterized by phospho-enrichment of multiple cell surface receptors, which in some cases are also activated in AML[31, 32]. For example, EGFR phosphorylation at Tyr1068 was greater in ND ERYs, which suggests that Dex-dependent growth of myeloid progenitors signals through EGFR. Phosphorylation of PDGFRβ Tyr716 and cKit Tyr719 was also significantly greater in ND ERYs. Furthermore, the cKit receptor associated proteins CD63 and CD9, which have the ability to form membrane bound complexes and display altered sensitivity to extracellular growth signaling ligands in the growth factor dependent myeloid cell line MO7e[33] are enriched in ND-derived ERYs. TGFβ and its direct downstream target SMAD2 also showed increased activation levels in ND ERYs. For this study it is our assumption that these signaling events convey the internal representations for growth, survival and expansion of normal healthy ERYs in vitro.

By contrast, PV-derived ERYs displayed increased phosphorylation of Ser345 on Chk1, which is a critical cell cycle regulator during hematopoiesis. Chk1 haplo-insufficient mice phenotypically display anemia, subverted organization of bone marrow architecture and decreased numbers of erythroid progenitors[34]. Notch1 and phsopho-IGF1R Tyr1131 were also present at higher levels in PV-derived ERYs. Notch1 acts as an IGF1R target in T cell acute lymphoblastic leukemias which evade apoptotic events via AKT signaling and PTEN inhibition[35]. Interestingly, phospho-enrichment of PTEN at Ser380 and an mTOR partner PRAS40 at Thr246 were also observed in PV-derived ERYs. Both of which are known inhibitory post-translational modifications [36–37]. PRAS40 was originally identified as a target of AKT phosphorylation and has recently been shown to partner with the mTOR complex in several cellular processes including survival and proliferation[37–39]. In addition to this functional cluster of signaling proteins with elevated phosphorylation in PV-derived ERYs, we also observed decreased phosphorylation of the PTEN inhibitor GSK3α/β at Ser279/216 in erythroblasts cultured from PV donors. These data suggest that loss of normal Dex-dependent signaling in abnormally developing erythroblasts may be compensated for by alternative survival signaling through both IGF1R-targeted activation of Notch1, loss of the inhibitory phosphorylation of PTEN by GSK3β, and by mTOR activity via phosphorylation and inhibition of PRAS40 by AKT.

We also observed a switch in the phospho-activation profile of the alpha and delta isoforms of Protein Kinase C between ND-derived and PV ERYs respectively. ND ERYs displayed increased phosphorylation of the classic Ca²⁺ dependent alpha isoform PKCα Ser657 [40] while increased phospho-enrichment of both PKCδ Thr505, which is activated via Ca²⁺-independent signals such as diacylglycerol, and its target MARCKs is shown in PV-derived ERYs. This differential signaling through classic and novel PKC isoforms has been described previously to effect EPOR-mediated signaling of differentiation and self-renewal pathways in bone marrow-derived progenitors [41] suggesting that PV-derived ERYs may signal through calcium-independent mechanisms and preferentially recruit PKCδ signaling during in vitro expansion.

Furthermore, two known c-AMP dependent proliferation CREB activators, MSK1 and RSK3 [42–44] were significantly phospho-enriched at Thr356/Ser306 and Ser360 respectively in PV derived ERYs. CREB has been shown to activate GATA1 and is required for erythroid differentiation [45]. Our data suggests that signaling through MSK1 and/or RSK3 may provide alternate signals for growth and differentiation in PV derived ERYs. Additionally we observed increased phosphorylation of STAT2 Tyr690/STAT3 Tyr705/STAT5 Tyr694 in PV-derived ERYs. Crkl is an SH2/SH3 adapter [46] that co-immunoprecipitates with active STATs upon EPO-dependent expansion of normal CD34+ progenitors and is more highly enriched in ND ERYs suggesting that Crkl may be an important interface for JAK/STAT signaling during normal erythroid development (Fig. 3). These data suggest that growth factor requirements for normal erythroid expansion and maturation may be bypassed in PV via activation of multiple growth signals located in the cytoplasm. Signals hyper activated in PV ERYs include those mediated by JAK/STAT proteins through multiple pathways such as i) those that intersect with PKCδ, ii) CREB-GATA1 activators such as MSK1 and RSK3, and iii) through IGF1R-Notch1 mediated governance of PTEN’s control on AKT with downstream association of phosphorylated PRAS40 with mTORC1.

Figure 3. Reverse Phase Protein Array of erythroblast protein lysates collected from normal donors and patients with Jak2 V617F positive polycythemia vera.

0.125–0.5 µg/µl of lysates were printed onto nitrocellulose-coated glass slides in triplicate and probed with a library of approximately 180 antibodies recognizing total, cleaved and phospho-protein endpoints. Primary antibody binding was detected using a biotinylated goat anti-rabbit IgG H+L (1:7500) or rabbit anti-mouse IgG (1:10) followed by streptavidin-conjugated IRDye680 fluorophore reaction. Differentially enriched signal intensities (Wilcoxon Rank-Test p < 0.05) of 39 features involved in multiple proliferation and survival pathways are shown.

Taken together, our results delineate differences in gene expression and in signaling pathways that occur in PV erythroid cells compared to normal erythroid cells. These data suggest that PV erythroblast expansion and maturation may be maintained and enriched in the absence of dexamethasone through activation of PV-specific signaling pathways and through transcription of PV-specific target genes, which are distinct from those accessed by normal healthy hematopoietic cells differentiating into the erythroid lineage. Importantly, these results suggest that there are PV-specific pathways that contribute to abnormal erythropoiesis in PV patients. Functional studies are needed to delineate which of these pathways is downstream of mutations in the JAK-STAT pathway and in epigenetic modifiers in PV patients, and to determine if any of these might represent novel targets for therapies that could inhibit abnormal erythropoiesis in PV.

Figure 1. Characterization of erythroid cells generated by day 10–13 in cultures of normal donors (ND) or polycythemia vera (PV) patients.

Log₁₀ scatter plots of CD235a with either CD36 or CD45 of cells obtained at day 10, 11 and 13 after culture of mono-nuclear cells from normal healthy donors or PV patients stimulated with 100 ng/mL SCF, 3 Units/mL EPO and 1 ng/mL IL-3 in the absence of Dexamethasone. May-Grunwald staining of representative cytospun day 10 cells is also presented. These data show the greater percentage of ERY expressing CD36 but not CD235a and the marginally detectable levels of polychromatophilic/orthochromatic ERY in cultures of ND with Dex than in those without Dex. By contrast the frequency of ERY expressing CD36/CD235a was high and of polychromatophilic/orthochromatic ERY were barely detectable in cultures of PV cells without Dex. Black, blue, green and red arrows/arrowheads indicate pro-erythroblasts and basophilic, polychromatophilic and orthocromatic erythroblasts respectively. n.d. = not done.

Table I.

Clinical data on the PV patients included in cell and biochemical studies.

Patients	JAK2V617F	Age/ Sex	WBC/ml	Hbg/dL	Hct%	plt(×105)/ mL	Duration (months)	Therapy
A	Homozygous	74/F	9300	16.5	56.8	425	2	Phlebotomy + CHT
B	Hetrozygous	28/M	8700	20.8	61.9	358	0	Unknown
C	Homozygous	60/M	21900	14.4	47.3	862	180	HU and Phlebotomy
D	WT	54/M	11900	17.9	53.7	305	0	Phlebotomy
E	Pos*	49/F	13040	12.1	46.6	1402	24	Aspirin/Ranitidina
F	Hetrozygous	46/M	12300	14.8	49.2	625	72	Phlebotomy
G	Homozygous	65/M	6500	16.1	49.7	317	36	HU

Abbreviations: CHT=not otherwise specified chemotherapy, Hbg=hemoglobin, Hct=hematocrit, HU=hydroxyurea, Plt=Platelets, WBC=white blood cells,

^*Allele burden not avail. Lost to follow up.