Alternatively spliced, truncated GCSF receptor promotes ... - Nature

Leukemia (2014) 28, 1041–1051 & 2014 Macmillan Publishers Limited All rights reserved 0887-6924/14 www.nature.com/leu

ORIGINAL ARTICLE

Alternatively spliced, truncated GCSF receptor promotes leukemogenic properties and sensitivity to JAK inhibition HM Mehta1,8, M Futami1,2,8, T Glaubach1, DW Lee3, JR Andolina1,4, Q Yang1, Z Whichard1, M Quinn1, HF Lu1, WM Kao5, B Przychodzen5, CA Sarkar6, A Minella7, JP Maciejewski5 and SJ Corey1 Granulocyte colony-stimulating factor (GCSF) drives the production of myeloid progenitor and precursor cells toward neutrophils via the GCSF receptor (GCSFR, gene name CSF3R). Children with severe congenital neutropenia chronically receive pharmacologic doses of GCSF, and B30% will develop myelodysplasia/acute myeloid leukemia (AML) associated with GCSFR truncation mutations. In addition to mutations, multiple isoforms of CSF3R have also been reported. We found elevated expression of the alternatively spliced isoform, class IV CSF3R in adult myelodysplastic syndrome/AML patients. Aside from its association with monosomy 7 and higher rates of relapse in pediatric AML patients, little is known about the biology of the class IV isoform. We found developmental regulation of CSF3R isoforms with the class IV expression more representative of a progenitor cell stage. Striking differences were found in phosphoprotein signaling involving Janus kinase (JAK)-signal transducer and activator of transcription (STAT) and cell cycle gene expression. Enhanced proliferation by class IV GCSFR was associated with diminished STAT3 and STAT5 activation, yet showed sensitivity to JAK2 inhibitors. Alterations in the C-terminal domain of the GCSFR result in leukemic properties of enhanced growth, impaired differentiation and resistance to apoptosis, suggesting that they can behave as oncogenic drivers, sensitive to JAK2 inhibition. Leukemia (2014) 28, 1041–1051; doi:10.1038/leu.2013.321 Keywords: GCSF; GCSFR; myelodysplastic syndromes; acute myeloid leukemia: JAK2

INTRODUCTION Gene ablation of granulocyte colony-stimulating factor (GCSF) or its receptor (GCSFR, gene name CSF3R) results in severe neutropenia, confirming their central importance in driving granulopoiesis.1,2 Nonsense mutations in the distal domain of the CSF3R, such as at codon 716 (d715), have been isolated from B80% of patients with severe congenital neutropenia (SCN) who developed secondary myelodysplastic syndrome (MDS) or AML.3–6 Truncation mutations of the CSF3R have also been isolated in chronic leukemias, such as an adult with Ph þ chronic myeloid leukemia as well as a child with SCN who developed chronic myelomonocytic leukemia and monosomy 7.7,8 A leukemogenic role for variant GCSFR is supported by the recent report that five distinct CSF3R mutations emerged over two decades in a patient with SCN who developed secondary MDS/AML.9 The GCSFR is expressed primarily on neutrophils and their precursor cells. Seven mRNA isoforms or classes encoding the GCSFR have been isolated.10,11 Only class I (the canonical type) and class IV (differentiation-defective) CSF3R isoforms are detectable in hematopoietic cells.12,13 As a result of alternatively splicing, the class IV GCSFR loses at position 725 the C-terminal 87 amino acids, which are replaced by a unique 34 amino-acid sequence.14 Increased expression of class IV CSF3R has been found in ex vivo culturing of

bone marrow cells from patients with monosomy 7.12,15 Children and adolescents with AML who overexpress the class IV CSF3R have a higher incidence of relapse.16 These findings underscore the antileukemic properties of the C-terminal region of the GCSFR. GCSF-induced GCSFR dimerization17 activates downstream signal transduction pathways involving Src kinases such as Lyn, Janus kinase (JAK)/signal transducer and activator of transcription (STAT), Ras/extracellular regulated kinase (ERK) and phosphatidylinositol 3-kinase/Akt (PKB).18 The cytoplasmic domain of GCSFR possesses four tyrosine residues (Y704, Y729, Y744, Y764), which serve as phospho-acceptor sites.19,20 SH2-containing proteins bind Y704 (STAT5 and STAT3) and Y764 (Grb2). Grb2 couples to both Gab2 and to SOS, permitting signaling diversification involving Ras/ERK, phosphatidylinositol 3-kinase/Akt and Src homology 2 phosphatase (SHP-2).21,22 Negative regulatory molecules, Src homology 2 domain containing inositol 5’-phosphatase and cytokine inducible Src homology 2 protein, are recruited to the GCSFR at residues 744 and 764.23 The class IV isoform lacks three of the four tyrosine residues (Y729, Y744, Y764) in the distal domain. We report that the class IV isoform, which is similar (Figure 1a) to the common nonsense mutations isolated from patients with SCN and MDS/AML, is elevated in a number of adults with AML/MDS. We further identified that there were pronounced

1 Department of Pediatrics (Hematology-Oncology) and Cell and Molecular Biology, Lurie Children’s Hospital of Chicago, Robert H Lurie Comprehensive Cancer Center, Northwestern University Feinberg School of Medicine, Chicago, IL, USA; 2Division of Molecular Therapy, Institute of Medical Science, University of Tokyo, Tokyo, Japan; 3Pediatric Oncology Branch, National Cancer Institute, Bethesda, MD, USA; 4Department of Pediatrics (Hematology-Oncology), University of Rochester School of Medicine, Rochester, NY, USA; 5Cleveland Clinic, Taussig Cancer Institute, Translational Hematology and Oncology Research, Cleveland, OH, USA; 6Department of Biomedical Engineering, University of Minnesota, MN, USA and 7Department of Medicine, Robert H Lurie Comprehensive Cancer Center, Northwestern University Feinberg School of Medicine, Chicago, IL, USA. Correspondence: Dr SJ Corey, Department of Pediatrics (Hematology-Oncology) and Cell and Molecular Biology, Lurie Children’s Hospital of Chicago, Robert H Lurie Comprehensive Cancer Center, Northwestern University Feinberg School of Medicine, 303 E. Superior Street, Chicago, IL 60611, USA. E-mail: [email protected] 8 These authors contributed equally to this work. Received 22 April 2013; revised 6 August 2013; accepted 18 September 2013; accepted article preview online 30 October 2013; advance online publication, 6 December 2013

GCSFR truncation promotes leukemic properties HM Mehta et al

1042 differences in growth, differentiation, proximal phosphoprotein signaling pathways, cell-cycle gene expression and sensitivity to JAK2 inhibition. Our data identify a critical region in the carboxylterminal domain of the GCSFR that confers anti-leukemogenic properties, which was one of the first properties attributed to human GCSF.24

MATERIALS AND METHODS cDNA and plasmid constructs CSF3R class I and class IV cDNA were cloned into the pcDNA3 vector. Another set of class I and class IV cDNA containing an internal HA tag sequence (located after the signal sequence) were cloned into the lentiviral transfer plasmid CSII-CMV-MCS-IRES2-Venus. The HA tag is inserted after the signal sequence, such that after the signal sequence is cleaved the HAtag presents itself at the N-terminus of the GCSFR. Chimeric growth hormone receptor (GHR)/GCSFR cDNA was constructed by overlap extension PCR method.25 cDNA of the extracellular domain of rabbit GHR (using RC2 as the template25) was amplified using primers A and B (Supplementary Table 1). The transmembrane plus intracellular domains of class I, class IV and d715 GCSFR were amplified by PCR using primer C as the common forward primer and primers D, E and F (Supplementary Table 1) as the specific reverse primers. PCR products were mixed and chimeric GHR/GCSFR and second round of PCR was performed using primer A as common forward primer and primer D, E and F as reverse primers for class I, class IV and d715, respectively. PCR products were cloned into CSII-CMV-MCS-IRES2-Venus.

Cell lines and maintenance

Figure 1. Comparison of carboxyl-terminal region of the GCSFR in patients with myeloid leukemia. (a) Schematic representation of class I (wild type), class IV (alternatively spliced isoform), d725 (mutant) and d715 (mutant) variants of the GCSFR. We report a GCSFR nonsense mutation from a patient with chronic myelomonocytic leukemia that occurred at codon 726, resulting in a protein of 725 amino acids (the amino-acid numbering does not include the 23 amino-acid signal sequence). Alternatively, splicing of the GCSFR results in the class IV isoform, which retains the 725 amino acids. The d725 and class IV isoform are similar to the truncated d715 GCSFR, a nonsense mutant commonly observed in SCN patients that transition to AML. All GCSFR variants have no differences in their extracellular and juxtamembrane domains, but differ in their cytoplasmic domains. The cytoplasmic domain of GCSFR consists of conserved box 1 and 2 in the truncated forms, with box 3 included in the full-length class I GCSFR only. In the full-length form, there are four tyrosine residues (Y704, Y729, Y744 and Y764), however, only Y704 is conserved among the nonsense mutants and alternatively spliced isoform. (b) Box plot with whiskers showing maximum, minimum and a line for the median was used to represent the percentage of class IV CSF3R mRNA expressed in primary AML and MDS cells. mRNA was harvested from bone marrow mononuclear cells using deidentified samples from patients with either AML or MDS and then subjected to qPCR. Also shown human bone marrow mononuclear cells, human neutrophils and umbilical cord blood CD34 þ cells. Breaks are introduced in the y axis to give two segments covering 64% (lower segment) and 36% (upper segments) of the y axis. Lower segment shows 0–15% and the upper segment depicts 20–100%. Statistical significant differences (*Po0.05; **Po0.01) were calculated using Kruskal–Wallis test and Dunn’s multiple comparison post test. Leukemia (2014) 1041 – 1051

Murine interleukin (IL)-3-dependent cell line, Ba/F3, expressing either hGCSFRI (Ba/F3 GRI) or hGCSFRIV (Ba/F3 GRIV) were obtained by electroporation of either class I or class IV cDNA in pcDNA3 vector. N-terminal HA-tagged hGCSFRI (Ba/F3 HA-GRI) or hGCSFRIV (Ba/F3 HAGRIV) were obtained by lentiviral infection of Ba/F3 parent cells. Cells were maintained in RPMI containing glutamine and supplemented with 10% fetal bovine serum (FBS), 1% PenStrep (Invitrogen, Grand Island, NY, USA) and 2 ng/ml recombinant murine IL-3 (mIL-3, Peprotech, Rocky Hill, NJ, USA). For studying GCSF-induced proliferation, the mIL-3 was replaced with human GCSF (hGCSF, Peprotech). Murine IL-3-dependent myeloblastic cell line, 32D, expressing lentivirally infected rabbit GHR/GCSFRI or GHR/GCSFRIV or GHR/d715 chimeric receptors were grown in Stemline II serum-free hematopoietic stem cell expansion medium (Sigma, St Louis, MO, USA) supplemented with 1% PenStrep and 2 ng/ml mIL-3. To study GCSFR-induced proliferation and differentiation, mIL-3 was replaced with human growth hormone (hGH, AbD Serotec, Raleigh, NC, USA).

Differentiation of NB4 cells Human promyelocytic cell line NB4 was grown in RPMI, 10% FBS, 2 mM glutamine and 1% PenStrep. Differentiation was induced by adding 1 mM all-trans retinoic acid (Sigma). For control experiments only ethanol was added to a final concentration of 0.1%. Cells were harvested at 0, 3, 4, 5, 6 and 7 days. Differentiation was confirmed by microscopic study of WrightGiemsa staining and flow cytometric immunophenotyping.

Patient samples and isolation of CD34 þ cells from umbilical cord blood De-identified patient bone marrow mononuclear cells were obtained from the Leukemia Sample Bank at the University of Texas MD Anderson Cancer Center on institutional review board-approved protocol, and consent was obtained in accordance with the Declaration of Helsinki. Samples were analyzed under an institutional review board-approved laboratory protocol. Umbilical cord blood was obtained following institutional review board guidelines of Northwestern University and Prentice-Women’s Hospital (Chicago, IL, USA). Mononuclear cells were isolated using Ficoll separation and CD34 þ cells were further isolated by magnetic labeling as per the manufacturer’s protocol using the CD34 MicroBead kit and MACS MS column (Miltenyi Biotec, Bergisch Gladbach, Germany). Purity of the isolated CD34 þ cells was determined by flow cytometry. & 2014 Macmillan Publishers Limited


1043 Ex vivo differentiation of neutrophils from CD34 þ cells and staining Purified CD34 þ cells were induced to differentiate following the protocol reported elsewhere.26 Briefly, freshly isolated cells were grown for the first 7 days in serum-free hematopoietic stem cell media (StemSpan SFEM, Stemcell Technologies, Vancouver, BC, Canada) supplemented with 10% FBS, 1% PenStrep, 100 ng/ml of human stem cell factor (Peprotech) and 10 ng/ml each of human IL-3 (hIL-3, Peprotech) hGCSF and human thrombopoietin (Peprotech). After 7 days, the media was changed to StemSpan SFEM supplemented with 10% FBS, 1% PenStrep and 10 ng/ml of hGCSF. Fresh media was used every 2 days. Samples at 0, 4, 6, 8, 10, 12 and 14 days were used for staining and mRNA extraction. Differentiation was confirmed by Wright-Giemsa staining and immunophenotyping using flow cytometry.

Flow cytometry-based immunophenotyping Cells were labeled with appropriate fluorescence-conjugated antibodies (Supplementary Table 2) by incubating in dark for 60 min at 4 1C. Cells were washed once and run on a BD LSRII flow cytometer (Franklin Lakes, NJ, USA). Data were analyzed using FLOWJO software (Tree Star Inc., Ashland, OR, USA).

Internalization of GCSFR using flow cytometry Log-phase cells were washed twice with PBS and resuspended in RPMI containing 1% BSA at 1 106 cells/ml and incubated for 4 h at 37 1C. 1 106 cells were incubated with anti-human CD114 (GCSFR) antibody on ice for 45 min. Cells were washed once and hGCSF was added at 100 ng/ml. Incubation was carried out in 37 1C water bath for 0, 7.5, 15, 30 and 60 min. Internalization halt buffer (PBS, 1% FBS, 0.2% sodium azide) was added and put on ice. Cells were centrifuged at 300 g for 10 min at 4 1C. Supernatant was removed and labeled with phycoerythrin anti-mouse secondary antibody by incubating in dark for 60 min at 4 1C. Cells were washed, resuspended in ice-cold PBS containing 0.01% azide and run on a flow cytometer.

Cell proliferation assay Log-phase growing Ba/F3 HA-GRI, Ba/F3 HA-GRIV, 32D GHR/GRI or 32D GHR/GRIV cells were washed twice with PBS to remove IL-3 and resuspended in fresh medium without IL-3 at a concentration of 2 105 cells/ml. 100 ml of cells were added per well. hGCSF (Ba/F3 cells) or hGH (32D cells) was added to appropriate wells and incubation was carried out for 48 h at 37 1C and 5% CO2. MTT (3-(4,5-dimethylthiazolyl-2)-2,5diphenyltetrazolium bromide) reagent (American Type Culture Collection; ATCC, Manassas, VA, USA) was added and incubation was carried out for 4 h. Absorbance was read at 600 nm using a multi-well plate reader (Fluostar Optima, Ortenberg, Germany).

Quantitative reverse transcription PCR Based on manufacturer’s protocols, 5 105–107 cells were used for total RNA extraction using the Trizol reagent (Invitrogen) for NB4 cells and the RNeasy plus mini kit (Qiagen, Hilden, Germany) for cord blood CD34 þ cells and patient bone marrow mononuclear cells. cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA). Real-time quantitative PCR (qPCR) was performed using primers designed specifically for class I CSF3R, class IV CSF3R and Abl as a reference gene described by Sloand et al.15 or a second set of primers with actin as a reference gene (Supplementary Table 3). The qPCR was carried out using iQ SYBR Green Mix (Bio-Rad) in a MyIQ single color detection system (Bio-Rad). To study gene expression profiles of cell cycle proteins in response to GCSF in Ba/F3 HA-GCSFRI and Ba/F3 HA-GCSFRIV cells, the cells were serum starved for 8 h followed by stimulation with 100 ng/ml hGCSF. Cells were then harvested at 0, 30, 60, 120, 240 and 480 min. RNA purification, cDNA synthesis and qPCR using specific primers (Supplementary Table 3) were performed as described above for CD34 þ cells.

125

I labeling of GCSF

Iodination was performed using the Bolton–Hunter method.27,28 One mCi of 125 I Bolton–Hunter reagent (Perkin Elmer, Waltham, MA, USA) was incubated with fourfold molar excess of hGCSF for 2 h at room temperature (RT). Reaction was quenched using 0.2 M glycine buffer. The labeled protein was isolated from the free 125I Bolton–Hunter reagent using size exclusion chromatography. The reaction mixture was layered onto pre-packed PD10 columns (GE Lifesciences, Pittsburgh, PA, USA) and protein eluted using the gelatin elution buffer (0.25% gelatin in sodium phosphate buffer, pH7.2). Aliquots representing the first radioactive peak, measured using a gamma counter contained the labeled hGCSF. To determine specific activity, preand post-column samples were spotted on filter paper and placed in a beaker containing 0.9% NaCl solution. Solvent front was allowed to reach the top of the filter paper strip. Strip was cut into two halves; top half representing free 125I Bolton–Hunter reagent and bottom half representing labeled GCSF. Specific activities for three different labeling experiments were 51.62, 175.12 and 96.54 c.p.m./pg.

Equilibrium binding analysis Cells were washed with Dulbecco’s phosphate-buffered saline (PBS) and resuspended in the reaction buffer (Dulbecco’s PBS containing 1 mg/ml BSA and 0.2% sodium azide) at a density of 2 106 cells/ml. Different concentrations of 125I-labeled GCSF were added to the cells and incubated for 1 h at RT. Nonspecific binding for each concentration was determined by adding 400-fold excess of unlabeled GCSF in parallel. GCSF bound to cells was separated from free GCSF as described elsewhere.29 Binding affinity was calculated by nonlinear fitting of free GCSF concentration vs bound GCSF. The nonlinear fitting for single site binding was done using GraphPad Prism software (GraphPad Software Inc., LaJolla, CA, USA). & 2014 Macmillan Publishers Limited

Cell growth, viability and cell cycle analysis by flow cytometry Ba/F3 GRI and GRIV cells were harvested, washed with RPMI and serum starved in RPMI with 1% BSA and incubated at 37 1C and 5% CO2 for 0, 4, 8 and 16 h. Cell count and viability were determined using a hemocytometer and trypan blue staining. Starved cells were fixed in cold 0.5% paraformaldehyde in PBS, incubated for 15 min at RT, washed with phosphate buffered saline, 2 mM EDTA, 1% BSA (PEB) and permeablized with ice-cold 70% methanol for 1 h. Cells were washed with PEB and treated with 100 mg/ml of RNAse at 37 1C for 20 min. Cells were washed and resuspended in PEB containing 25 mg/ml propidium iodide (Invitrogen) for 20 min at RT in the dark, and immediately run on flow cytometer. Data were analyzed using FLOWJO.

Immunoprecipitation and western blot analysis For studying the time-dependent changes in GCSF-mediated phosphoprotein signaling, log-phase Ba/F3 HA-GRI and HA-GRIV were washed twice with PBS and serum starved in RPMI with 1% BSA and incubated overnight at 37 1C and 5% CO2. Cells were then centrifuged and resuspended to 1 107 cells/ml. The cells were treated with 100 ng/ml hGCSF for 0, 1, 5, 15, 30, 60 and 120 min at 37 1C and lysed in 1% NP-40 lysis buffer with protease inhibitor cocktail (Calbiochem, Durmstadt, Germany), 1 mM phenylmethylsulfonyl fluoride and 2 mM sodium vanadate for 30 min on ice. The lysates were prepared for western blotting by adding Laemelli buffer and heating it at 100 1C for 5 min. To study the effect of the JAK2 inhibitor on cell signaling, the Ba/F3 HA-GRI and HA-GRIV cells were washed twice with PBS and then resuspended in RPMI with 1% BSA to a density of 1 107 cells/ml. Various concentrations of the JAK2 inhibitor or dimethyl sulfoxide as diluent control were added to 1 107 cells and incubated for 1 h at 37 1C. hGCSF was added to a final concentration of 100 ng/ml and incubated for an additional 10 min at 37 1C. Cells were lysed as described above. The lysate was used for immunoprecipitation of JAK2 or prepared for western blotting as above. JAK2 was immunoprecipitated using rat anti-JAK2 polyclonal Ab (Supplementary Table 2) and protein A agarose beads.

RESULTS Aberrant GCSFR forms from adult patients with MDS/AML Children with SCN that progress to MDS/AML almost always display nonsense mutations affecting the carboxyl-terminus of the GCSFR. These patients are treated chronically with pharmacologic doses of GCSF, and thus there is a selection pressure for clonal outgrowth. We sequenced bone marrow cells from a group of 100 adult patients with MDS/AML and no prior GCSF administration, and we found one patient diagnosed with chronic myelomonocytic leukemia whose bone marrow cells harbored a nonsense Leukemia (2014) 1041 – 1051


1044 mutation at position 726 in the GCSFR gene (Figure 1a). The patient was a 68-year-old male presenting with a white blood count of 119 000/ml, hemoglobin 11.1 g/dl and platelet count 156 000/ml. The absolute neutrophil count was 65 590/ml. The cytogenetics was normal. The alleleic frequency of CSF3R mutation was 33%. We also analyzed a cohort of adult MDS/AML patients (Table 1) for expression of class IV CSF3R mRNA, which also encodes the canonical CSF3R through position 725 (Figure 1a). A box plot (Figure 1b) displays the median expression of class IV CSF3R expression in primary AML and MDS. Using Dunn’s multiple comparison test, statistically significant differences

Table 1.

in median class IV CSF3R mRNA expression (Po0.05) were observed for normal bone marrow mononuclear cells vs MDS mononuclear cells, normal neutrophils vs MDS mononuclear cells and normal CD34 þ cells vs normal neutrophils. Class I mRNA expression predominates with myeloid differentiation NB4 cells were used to model granulocyte differentiation. Over 7 days of treatment with all-trans retinoic acid, 70% of NB4 differentiated into bands or segmented forms (Figures 2a–c).

Clinical information for control and MDS/AML patients

Patient ID

Sex

FAB

Cytogenetics

WBC ( 1000)

PB blast (%)

BM blast (%)

Normal controls C1 M C2 F C3 F

Normal Normal Normal

Normal Normal Normal

11.8 7.4 2.6

0 0 0

1 1 3

MDS M3 M5 M7 M8 M36 M37 M39 M40 M41

M M M M M F F M F

RAEB RAEB RAEB RAEBT RA RAEBT RAEBT RAEB RAEBT

NA NA Normal NA 45,XY,-7,(2p-;3q þ ) 45,XX,-7 46,XX,-7, þ 8; 47–48,XX,-7, þ 8, þ mar[cp3]; 46,XX 45,XY,-7; 46,XY,-7,mar[cp4]; 46,XY 46,idem,-7, þ mar[2]; 47,idem with random changes

7.9 1.8 6.9 NA 19.6 NA 10.5 3.6 8.4

NA NA 0 NA 17 NA 18 2 7

12 13 6 8 18.8 NA 22 12 16

AML A6 A7 A8 A9 A10 A11 A12 A13 A14 A15

M F F M M M M M M M

M1 M1 M2 M2 M1 M2 M5 M5 M3 M3

46,XY,del(9)(q12q22)[5]; 46,XY,del(9)(q12q22),-21, þ mar[2]; 46,XY[13] Normal Normal 46,XY,t(6;9)(p23;q34)[16]; 46,XY[2] 45,XY,der(12)t(12;17)(p13;q11.2),-7[20] 47,XY, þ 21[14]; 46,XY[6] 47,XY, þ 8[15]; 46,XY[2] Normal 46,XY,del(7)(q22)[1] 43–47,XY,del(7)(q22),-8,-9, þ 1–2r[cp18]; 46,XY[1] 46,XY,t(15;17)(q22;q21)[10]; 45,XY,del(1)(q23),del(3)(p11),6,t(15;17)(q22;q21),add(19)(q13.4)[10]

10.6 64.3 45.8 10.7 47.7 32 200.5 29.1 7.3 1.4

81 98 66 45 88 62 63 22 1 0

83 93 63 62 93 74 86 80 60 7

Abbreviations: BM, bone marrow; F, female; FAB, French-American-British classification for hematologic diseases; M, male; MDS, myelodysplastic syndrome; NA, not available; PB, peripheral blood; RA, refractory anemia; RAEB, refractory anemia with excess blasts; RAEBT, refractory anemia with excess blasts in transformation; WBC, white blood cell.

Figure 2. Developmental expression pattern of class I and class IV GCSFR during differentiation. NB4 cells were induced to differentiate by treatment with 1 mM all-trans retinoic acid (ATRA). Samples were analyzed for differentiation and GCSFR isoform expression at 0, 3, 4, 5, 6 and 7 days. Umbilical cord blood-derived CD34 þ cells were induced to differentiate to neutrophils by growing them for 7 days in media containing a cocktail of cytokines (human stem cell factor, hGCSF, hIL-3 and human thrombopoietin, see Materials and methods) followed by a switch to media containing only hGCSF for another 7 days. Cells harvested at days 0, 4, 6, 8, 10, 12 and 14 were used to determine the differentiation stage and to determine class I and class IV expression. (a) Microscopic view of differentiation status of NB4 cells by Wright-Giemsa staining at 0, 120 and 168 h. (b) 200 Wright-Giemsa-stained NB4 cells from days 0, 3, 5, 6 and 7 were counted under a microscope and categorized into promyelocytes, metamylocytes, band cells and fully differentiated neutrophils. The cell types are represented as percent of total cells counted. (c) Flow cytometrybased immunophenotyping was used to determine differentiation by measuring expression of CD11b on NB4 cells after days 0, 3, 5 and 7 of ATRA treatment using APC-Cy7-conjugated anti-human CD11b antibody. (d) Ratio of class I:class IV CSF3R expression as measured by qPCR from NB4 cells harvested on days 0, 3, 5, 6 and 7. Data show class I:IV ratio for cells differentiated using ATRA and under control conditions (ethanol). (e) Microscope pictures ( 40) of Wright-Giemsa-stained cells showing progression of differentiation of CD34 þ cells, from myeloblast/ promyelocyte ¼ 4 myelocytes/metamyelocyte ¼ 4 bands/neutrophils. (f) 200 Wright-Giemsa stained of CD34 þ cells, at days 0, 2, 4, 6, 8, 10, 12 and 14 of differentiation treatment were counted under a microscope and categorized into myeloblast/promyelocyte, myelocytes/ metamyelocytes and bands/segmented cells. The cell types are represented as percent of total cells counted. (g) Flow cytometry-based immunophenotyping was used to determine differentiation of CD34 þ cells by measuring expression of CD34, CD11b, CD16b and CD66b on days 0, 4, 8 and 14 of differentiation treatment using PE-conjugated anti-human CD34 and CD16b, APC-Cy7-conjugated anti-human CD11b and FITC-conjugated anti-human CD66b antibody. Cells were labeled separately with each antibody, thus no compensation was required. Analysis was carried out using FLOWJO software. (h) Myeloperoxidase (MPO) heavy chain expression in of CD34 þ cells on day 4, 6, 10, 12 and 14 of differentiation treatment by western blot. 50 mg of protein was loaded per lane and MPO was detected using donkey anti-MPO (heavy chain) antibody. Actin was used as a loading control. (i) Ratio of class I:class IV GCSFR expression as measured by qPCR from CD34 þ cells, cells harvested on days 4, 6, 8, 10, 12 and 14 of differentiation treatment. (j) Class I and class IV expression as measured by qPCR and normalized to Abl from CD34 þ cells harvested on days 4, 6, 8, 10, 12 and 14 for differentiation treatment. Inset shows changes in expression of class IV on a smaller y axis scale; IB, immunoblot; PE, phycoerythrin. Leukemia (2014) 1041 – 1051

& 2014 Macmillan Publishers Limited


1045


Leukemia (2014) 1041 – 1051


1046 Quantitative reverse transcription PCR (qPCR) revealed a timedependent increase in the ratio of CSF3R class I:class IV mRNA in all-trans retinoic acid-treated cells (Figure 2d). To confirm these findings, we evaluated their dynamic expression during granulocytic differentiation of human CD34 þ cells. Treated for 7 days with a cytokine cocktail (SCF, IL-3, TPO and GCSF), CD34 þ cells were cultured for an additional 7 days in medium containing only GCSF. This resulted in terminal differentiation to neutrophils (Figures 2e–h). qPCR showed a gradual increase in class I:class IV ratio, primarily mediated by an increase in class I expression (Figures 2i and j). These data suggested that the class I isoform becomes the

Figure 3. Binding affinity and internalization of GCSFR isoforms. (a) Ba/F3 cells expressing class I or class IV cells were equilibrated with increasing concentrations of 125I-labeled hGCSF for 1 h at RT. Bound *GCSF was separated from the free. Data were plotted as bound vs free and the binding affinity was determined using nonlinear curve fitting based on a one site-specific binding model (Table 2). (b) Receptor internalization of class I, class IV and d715 GCSFR was determined by pre-equilibration with a mouse IgG1 antihuman CD114 Ab for 1 h at 4 1C followed by incubation with hGCSF at 37 1C for 0, 7.5, 15, 30 and 60 min. Ba/F3 parent cells were used to determine the background for each time point. Internalization at various times was determined by quantitating the number of surface receptors by flow cytometry, using a PE-conjugated rat anti-mouse IgG1 secondary antibody. The data are plotted as background subtracted geometric mean fluorescence instensity normalized to time 0, set at 100%, against time. Nonlinear curve fitting using a single phase decay was performed using GraphPad Prism v5.04 software. Statistical analysis was performed applying Bonferroni’s multiple comparison test after two-way analysis of variance. Statistically significant differences between class I and class IV at 7.5 and 15 min of internalization were observed and represented as *Po0.05 and **Po0.01, respectively. PE, phycoerythrin Leukemia (2014) 1041 – 1051

predominant CSF3R mRNA during differentiation, consistent with a positive feedback loop. Class IV receptor demonstrates decreased internalization without change in binding affinity To identify the biochemical consequences of class I and IV GCSFR isoforms, we determined if they displayed different ligand affinity. With radiolabeled GCSF, we performed steady-state binding analysis on Ba/F3 cells expressing either class I or class IV GCSFR (Figure 3a). Cell lines were sorted for comparable levels of receptor isoform expression. Binding analysis confirmed similar levels of receptor on the cell surface and no significant difference in binding affinities between the GCSFR isoforms (Table 2). Because the class IV lacks the dileucyl motif that facilitates internalization,30–32 we next determined internalization by flow cytometry. Decreased internalization was observed for class IV isoform compared with that for class I at 7.5 min and 15 min (Figure 3b). Slower internalization of the truncated receptors would permit prolonged receptor signaling. Although class IV and d715 are similar in their truncation of the carboxyl-terminal domain, the class IV does have additional 34 amino-acid sequence, which may contribute to its reduced internalization even in comparison to d715. Class IV receptor demonstrates increased proliferation and survival but decreased differentiation We found that as the GCSF concentration increased, the class IV isoform induced greater proliferation compared with class I (Figure 4a) at high GCSF concentration, however, at low GCSF concentrations (o5 ng/ml), class I GCSFR displayed higher proliferation. These results confirmed the observations derived from monosomy 7 patients, which showed enhanced proliferation at higher doses of GCSF.15 To assay for differentiation, we studied 32D cells as an experimental cell model. Although White et al. showed that the transfection of 32D cells with class IV GCSFR isoform impaired myeloid differentiation, they saw only a small difference between cells expressing the canonical or class IV isoform.13 However, these results were likely confounded by the presence of endogenous murine GCSFR in 32D cells. To circumvent this, we constructed chimeric receptors consisting of the extracellular domain of rabbit GHR and the transmembrane and cytoplasmic domains of the GCSFR (Figure 4b). The GHR also belongs to the hematopoietin/cytokine receptor superfamily. Upon ligand binding, the receptor undergoes homodimerization and triggers changes in cellular protein phosphotyrosine content.25 Because 32D cells do not express endogenous GHR, we could stimulate chimeric receptors with GH. We established 32D cells with three different chimeric receptors: GHR/GRI (class I

Table 2.

Summary of binding characteristics of class I and class IV

isoforms Ba/F3 G-CSFRI

Average

Ba/F3 G-CSFRIV

Kd (nM)

Number of sites/cell

Kd (nM)

Number of sites/cell

0.58±0.02

10 267±834

0.53±0.04

9996±1812

Abbreviation: GCSFR, granulocyte colony-stimulating factor receptor. Binding affinities (Kd) and receptor numbers (sites/cell)±s.e.m. were determined from independent steady-state binding experiments, using radiolabeled hGCSF and Ba/F3 cells expressing either class I or class IV GCSF receptor. The values were determined as an average of values obtained from three independent experiments and by using nonlinear regression analysis as done in Figure 3a.



1047

Figure 4. Differential proliferation and differentiation response induced by class I and IV GCSFR. (a) GCSF-induced proliferation of Ba/F3 cells expressing either class I or IV cells was determined at 48 h using MTT (3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide) assay. Proliferation is quantitated as absorbance at 600 nm. Data are plotted as absorbance vs hGCSF concentration. (b) Comparison of GCSFR isoforms and growth hormone receptor (GHR)/GCSFR chimeric receptors. Three different forms of GCSFR are shown: class I, class IV and the d715 truncation mutant of GCSFR. The three GCSFR receptors differ from each other in their cytoplasmic domain. The full-length receptor has four tyrosine residues, of which the last three are missing in the class IV isoform and d715 mutant. The class IV isoform differs from the d715 mutant in that it has a novel 34 aa C-terminal region containing a single tyrosine. Chimeric receptors were created that have the extracellular domain of the rabbit GHR and intracellular domain of either class I, class IV or d715. The figure shows a schematic representation of the structure of the GHR and GCSFR and the generated chimeric receptors. (c) Proliferation of 32D cells expressing the chimeric receptors GHR/ GRI and GHR/GRIV was induced by increasing concentrations of growth hormone (GH). Proliferation was determined using the MTT assay, by quantitating absorbance at 600 nm. Data are plotted as absorbance vs hGH concentration. (d) 32D cells expressing either the chimeric GHR/ GRI (class I), GHR/GRIV (class IV) and GHR/d715 (d715, truncation mutant) cells were grown in medium containing 100 ng/ml hGH. Cytospin samples (5 104 cells per condition) were stained using Wright-Giemsa staining. After 6 days or later, class I cells differentiated into mature granulocytes with segmented nuclei and secondary granules. Class IV cells showed dysplasia with ring-shaped nuclei and lacked secondary granules. d715 cells did not show dysplasia, but differentiation was inhibited. (e) Total 200 cells were counted under a microscope and categorized into myeloblast/promyelocyte, myelocytes/metamyelocytes and band/neutrophils. The cell types are represented as percent of total cells counted. (f ) Immunophenotyping was carried out as a measure of differentiation. Ly6G and 6C (Gr-1) expression on surface were determined using PE-conjugated rat anti-mouse Ly6G and 6C antibody on cells treated with 100 ng/ml hGH for days 0, 3, 6 and 9. Cells were incubated with the antibody for 1 h at 4 1C in dark. Samples were run on a flow cytometer to determine PE-labeled cells. Analysis was carried out using FlowJo software. PE, phycoerythrin; TM, transmembrane.


Leukemia (2014) 1041 – 1051


1048 GCSFR), GHR/GRIV (class IV GCSFR) and GHR/d715 (d715 GCSFR). We observed that treatment of these cell lines with 100 ng/ml human GH resulted in proliferation, with a comparable dose response effect for the isoforms as observed with Ba/F3 cells (Figures 4a and c). When cultured with human GH for 9 days, we found impaired differentiation in cells expressing class IV isoform or d715 mutant (Figures 4d–f). To determine the effect of cytokine-stimulation on intracellular signaling pathways promoting survival, we performed a time-course study of cell growth and survival. For BaF3 GRI and GRIV cells, the total cell count and viability were similar through 8 h of starvation. However, a significant difference in cell count and viability between the BaF3 GRI cells, GRIV cells and d715 cells arose at 16 h, with complete cell death observed at 24 h for class I and d715 with B20% viability remaining for class IV cells (Figures 5a and b). Flow cytometric analysis demonstrated a G0/G1 cell cycle arrest at 8 h for both BaF3 GRI and GRIV cells, with a similar magnitude of effect (approximately 15% increase in number of events for both cell lines when compared with the 4 h starvation; Figure 5c). There was also a concomitant 13–15% decrease in the number of cells in S/M/G2 for both BaF3 GRI and GRIV cells without an increase in cell death. However, prolonged starvation of 16 h resulted in increased cell death much more pronounced in the BaF3 GRI cells, with the BaF3 GRIV cells demonstrating resistance to starvationinduced apoptosis.

Class IV receptor shows increased cell cycle protein expression and atypical signaling pathways Having established differences in developmental expression, internalization rates, proliferation, differentiation and resistance to starvation-induced apoptosis, we next sought to identify the mechanistic basis for increased class IV-mediated proliferation by examining induced cell cycle protein expression and changes in signal transduction pathways in comparison with class I. Cell cycle gene expression using qPCR identified a similar trend in the expression of E2F target cell cycle proteins in both class I (GCSFRI) and class IV (GCSFRIV) for PolA, CCNA1 and CCNE1 (Figure 6a) with upregulation in expression after 240 min of GCSF stimulation. A statistically significant increase was observed for CCND1 expression with class IV at 240 min (Figure 6a). Increased levels of phospho-STAT3, phopho-STAT5 and phopho-ERK1/2 are seen in class I-expressing cells compared with class IV cells (Figure 6b). Although no major differences were found for phospho-Akt, higher levels of activation were observed for Src family kinases, such as Lyn, in class IV-expressing cells (Figure 6b). To further correlate growth responses with signaling pathways, we studied the effects of dasatinib (a Src/Abl tyrosine kinase inhibitor) and selective JAK2-specific inhibitor NVP-BSK805.33,34 Dasatinib, as great as 10–6 M, did not inhibit proliferation (data not shown). Surprisingly, treatment with the JAK2 inhibitor reduced maximally the class IV GCSFR (GI50 ¼ 28.5 nM)-mediated proliferation, followed by d715 GCSFR (GI50 ¼ 547 nM) and class I

Figure 5. Relative resistance to apoptosis in serum- and cytokine-starved Ba/F3 cells. Ba/F3 cells expressing class I or class IV were serum starved for a total of 0, 4, 8, 16 and 24 h. (a) Cell count and (b) viability by trypan blue staining was performed on a hemocytometer. (c) Ba/F3 cells starved for 0, 4, 8 and 16 h were stained with propidium iodide to determine cell cycle distribution by flow cytometry. The data were analyzed using FlowJo software. Leukemia (2014) 1041 – 1051



1049 GCSFR (GI50 ¼ 2141 nM; Figure 6c). Similar results were obtained using a myeloid cell line, 32D cells expressing the chimeric receptors GHR/GRI and GHR/GRIV (data not shown). Using another

JAK2 inhibitor, ruxolitinib also showed similar pattern as NVPBSK805 with class IV showing maximum sensitivity (GI50 ¼ 5.2 nM) followed by d715 (GI50 ¼ 94.4 nM) and class I (GI50 ¼ 486 nM). However, ruxolitinib demonstrated approximately four- to sixfold stronger inhibition of proliferation for all three receptor types when compared with NVP-BSK805. To understand the signaling events perturbed in growth inhibition, we looked at phosphorylation of key signaling molecules. We performed studies using NVP-BSK805, as it is selective for JAK2 as compared with ruxolitinib, which is equipotent for JAK1 and JAK2.35 A similar pattern of inhibition of JAK2 phosphorylation was observed for both class I- and class IV-expressing cells, with 50% inhibition observed at B10 nM (Figure 6d). As JAK2 phosphorylates STAT3 and STAT5, we examined their phosphorylation status. However, inhibition of phosphorylation of STAT3 and STAT5 occurred only at the highest concentration of 500 nM NVP-BSK805 (Figure 6d). Thus, there may be off-target effects of NVP-BSK805 not yet identified. We also examined the phosphorylation state of SHP-2, ERK1/2 and Src family of kinases. Similar to the STAT proteins, inhibition of ERK1/2 phosphorylation was observed at the highest concentration of 500 nM (Figure 6d). SHP2 showed a dose-dependent decrease in phosphorylation, however, unlike the JAK2 inhibition the decrease was limited to the class IVexpressing cells mimicking the cell proliferation response. The data suggest a possible role for SHP-2 in class IV-mediated survival via JAK2 activation. NVP-BSK805 may thus serve as a useful chemical probe to interrogate the critical role of JAK2 and SHP-2 in enhanced class IV isoform-mediated growth.

DISCUSSION Nonsense mutations affecting the GCSFR have been reported primarily in patients with SCN who have been chronically treated with pharmacologic doses of GCSF. Here, we report another case of a nonsense mutation in GCSFR occurring in an adult with Figure 6. Signaling mechanisms associated with class IV GCSFR expression. (a) Quantitative expression by qPCR of cell cycle genes. Ba/F3 cells expressing either GCSFRI or GCSFRIV were serum starved for 8 h and then stimulated with 100 ng/ml hGCSF for 0, 30, 60, 120, 240 and 480 min. Ribosomal RNA normalized data were further normalized to time 0, to represent fold change in expression of individual proteins upon stimulation with hGCSF. Error bars represent s.e.m. for three independent experiments. Statistical analysis was performed applying Bonferroni’s multiple comparison test after two-way analysis of variance. Statistically significant differences between GCSFRI and GCSFRIV were observed for the CCND1 gene and represented as **Po0.01, respectively. (b) Western blot analysis of GCSF-induced class I and class IV GCSFR signaling. Ba/F3 expressing either class I or class IV GCSFR were starved overnight and treated with 100 ng/ml GCSF for 0–120 min. Cells were lysed and 50 mg of protein were analyzed by western blot analysis for the specific antibodies shown (Supplementary Table 2). (c) Effect of JAK2 inhibitors on cell proliferation and phosphosignaling. Growth inhibition curves of Ba/F3 cells expressing class I, class IV or d715 GCSFR and growing in the presence of 100 ng/ml of hGCSF were obtained for increasing concentrations of JAK2 inhibitors ruxolitinib or NVP-BSK805. Proliferation was determined using the MTT (3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide) assay. Data representative of three independent experiments are presented as values normalized to 3-(4, 5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide values for cells growing in the absence of inhibitor (untreated), but in the presence of hGCSF. Curve fitting was performed using GraphPad Prism v5.04 software. (d) Ba/F3 cells expressing class I or class IV GCSFR were washed to remove IL-3 and preincubated for 1 h with the inhibitor, followed by addition of 100 ng/ml of hGCSF and additional incubation for 10 min. Cells were lysed and 50 mg of protein were western blotted with specific antibodies. JAK2 was immunoprecipitated from 500 mg of lysate as described in the Methods. IB, immunoblot; IP, immunopreciptate. & 2014 Macmillan Publishers Limited

Leukemia (2014) 1041 – 1051


1050 myeloid leukemia without exposure to GCSF. The resulting truncated receptor is similar to that found in the alternatively spliced class IV isoform, which has been associated with increased relapse rates in AML.16 We found elevated levels of class IV at diagnosis in adult patients with MDS/AML. Interestingly, statistical differences were observed between normal and MDS groups, but not for the AML group, however, there was one AML patient, with elevated class IV expression. Increased class IV expression was also observed in CD34 þ cells compared with neutrophils and bone marrow mononuclear cells, suggesting class IV expression is higher in blood stem cells. Thus, increased class IV expression may either reflect a state of stemness or contribute to impaired differentiation. Ehlers et al. reported higher relapse rates in AML patients treated with GCSF and increased class IV expression. Thus, higher class IV expression may be associated with leukemic properties or chemoresistance. Here, we report that the class IV isoform demonstrated induced a greater proliferative response in a dose-dependent manner, failed to induce myeloid differentiation, demonstrated resistance to starvation-induced apoptosis and transduced a distinct phosphoprotein signaling profile. Whether it is due to nonsense mutations or alternative splicing, a truncated GCSFR promotes leukemogenic properties. Using NB4 cells, we determined the expression of class I and class IV relative to each other and found an increase in class I:class IV ratio (Figure 2d). The ratio could increase in one of two ways: class I increases or class IV decreases over time. We therefore used primary human hematopoietic stem cells to understand the changes in class I:class IV ratio. The CD34 þ differentiation data (Figure 2f) were analyzed for specific isoform expression, as normalized to c-Abl, there was negligible change in class IV expression when compared with the marked upregulation (B60-fold) for class I (Figure 2j). Because the increase in the I:IV ratio was due to increased class I CSF3R expression, there is a formal positive feedback loop for class I expression during differentiation. Biochemical analysis of ligand-receptor binding did not identify any differences between class I and IV GCSFR isoforms, but differences were observed in the receptor internalization that point to a slower internalization of class IV GCSFR. Previous reports have identified the C-terminus of class I GCSFR, which are critical for internalization.31,36 These regions are missing in class IV and d715 GCSFR. Thus, both class IV and d715 are expected to demonstrate reduced internalization. As expected class IV internalization was slower and reduced, but not for d715, however, reduced internalization using radiolabeled ligand has been observed with d715 compared with class I GCSFR.32,36 Thus, reduced internalization of class IV as with d715 results in ineffective downmodulation of surface receptors, thereby increasing its signaling duration. In addition, truncated receptors act in a dominant negative manner by preventing internalization of wt GCSFR.32 One striking difference between class I and class IV is the absence of the distal three tyrosine residues, which serve as docking sites for SH2-containing proteins, in the class IV isoform or those truncated receptors due to nonsense mutations such as d715 or d725. The intracellular signaling in class IV-expressing cells showed reduced and/or delayed activation of STAT3, STAT5 and ERK1/2 but tonic activity of Src family kinase Lyn (Figure 6b). Previous studies on d715 GCSFR showed that maximal proliferation correlated with sustained STAT5 activation.37 Tight regulation in the expansion of granulocyte progenitors could also be due to growth control associated with a JAK-STAT-SOCS3-negative feedback loop or ubiquitination and receptor degradation.38,39 The class IV GCSFR might promote a greater degree of proliferation by failure to initiate feedback inhibition by SOCS3 or signals for differentation.40 To evaluate this scenario, we treated the three cell types with JAK2 inhibitors, ruxolitinib and NVPBSK805. This resulted in a significant decrease in proliferation of the class IV cells only (Figure 6c), suggesting different sets of signaling pathways were transducing proliferation. The data suggest that class IV-mediated growth is dependent on JAK2, Leukemia (2014) 1041 – 1051

however, class I may require both JAK1 and JAK2. The d715 GCSFR showed intermediate sensitivity between class IV and class I GCSFR, suggesting that truncation of GCSFR makes it more sensitive to JAK inhibition. Class IV GCSFR and d715 share the membrane proximal cytoplasmic region, but class IV possesses an extra distal 44 amino acids (including a novel 34 amino-acid region). This region may be involved in downmodulation of the proliferative response mediated by the membrane proximal region, which may explain higher class IV sensitivity to JAK inhibition. Our phosphoprotein signaling data imply that a JAK pathway independent of STAT promotes class IV proliferation. Interestingly, we identified SHP-2 as a potential regulator of class IV-mediated proliferation. However, of the signaling pathways investigated following treatment with NVP-BSK805, only the SHP-2 Tyr 580 phosphorylation pattern correlated strongly with growth (Figures 6c and d). The precise role for SHP-2 in GCSFR signaling is not fully understood. Lyn, JAK1 and JAK2 phosphorylate SHP-2.22,41,42 The JAK2 inhibitor inhibited JAK2 phosphorylation at similar concentrations in both class I and IV cells. Of note, it inhibited GCSF-induced SHP-2 phosphorylation at a lower concentration in class IV cells than class I cells. Thus, the greater SHP-2 inhibition might explain the greater cell cycle arrest effect seen in class IV cells. SHP-2 can associate with Y704, Y729 and Y764 either directly or indirectly and thus loss of the latter two tyrosine residues limits the binding sites on the class IV isoform.38,43 Association of SHP-2 with Y704 is spatially favorable for its activation by JAK proteins compared with the distal tyrosine residues. Because SHP-2 has fewer binding sites on the class IV isoform, cells that express that isoform are likely to be more sensitive to JAK2 inhibitors when it comes to SHP-2. SHP-2 binding to GCSFR requires the region between residues 715–735, and thus GCSFR d715 is unable to bind SHP-2.38 SHP-2 dephosphorylates STAT5, which is lost in d715 leading to enhanced STAT5 signaling by d715. Class IV retains part of the 715–735 region, which might explain the reduced STAT5 phosphorylation by class IV (Figure 6b). Thus, SHP-2 controlled cellular signaling is critical for class IV, but not d715. These differences in class IV and d715 explain their differential sensitivity to JAK inhibitors. We recently reported that Gab2 forms a complex with Lyn and after GCSF stimulation, Gab2 recruits SHP-2, which dephosphorylates phospho-Lyn Tyr507, leading to Lyn activation, which promotes proliferation associated with the class I isoform.22 Thus, the intracellular proliferative signaling system consisting of Lyn-Gab2-SHP-2 and JAK-STAT is profoundly perturbed by the loss of the distal three tyrosine residues and an additional signaling pathway may be operative. JAK2 inhibitors will be useful as a chemical probe to identify that cytokine-sensitive pathway. Furthermore, JAK2 inhibitor such as ruxolitinib, already approved for myelofibrosis,34 might be clinically effective to treat acute myeloid leukemias where the class IV receptor is overexpressed.16 In support of JAK2 inhibitor therapy for GCSFR-mediated leukemogenesis, we recently reported that another mutation of GCSFR (T618I), which is observed in over 50% patients with chronic neutrophilic leukemia and atypical chronic myelogenous leukemia,44 is sensitive to ruxolitinib but not to dasatinib.45

CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS We thank Dr Shigekazu Nagata for providing the cDNA for rabbit GHR and Dr John Crispino for providing the JAK2 inhibitors. Funding to SJC from NIH Independent Scientist Award KO2-HL03794, RO1-CA108992, JP McCarthy Foundation, NIH PO1CA55164, Leukemia SPORE CA100632 and AA/MDS International Foundation; to MF from a New Investigator Award from the AA/MDS International Foundation; and to TG and JRA from NIH T32CA079447.



1051 REFERENCES 1 Seymour JF, Lieschke GJ, Grail D, Quilici C, Hodgson G, Dunn AR. Mice lacking both granulocyte colony-stimulating factor (CSF) and granulocyte-macrophage CSF have impaired reproductive capacity, perturbed neonatal granulopoiesis, lung disease, amyloidosis, and reduced long-term survival. Blood 1997; 90: 3037–3049. 2 Liu F, Poursine-Laurent J, Link DC. The granulocyte colony-stimulating factor receptor is required for the mobilization of murine hematopoietic progenitors into peripheral blood by cyclophosphamide or interleukin-8 but not flt-3 ligand. Blood 1997; 90: 2522–2528. 3 Dong F, Hoefsloot LH, Schelen AM, Broeders CA, Meijer Y, Veerman AJ et al. Identification of a nonsense mutation in the granulocyte-colony-stimulating factor receptor in severe congenital neutropenia. Proc Natl Acad Sci USA 1994; 91: 4480–4484. 4 Dong F, Brynes RK, Tidow N, Welte K, Lowenberg B, Touw IP. Mutations in the gene for the granulocyte colony-stimulating-factor receptor in patients with acute myeloid leukemia preceded by severe congenital neutropenia. N Engl J Med 1995; 333: 487–493. 5 Germeshausen M, Ballmaier M, Welte K. Incidence of CSF3R mutations in severe congenital neutropenia and relevance for leukemogenesis: Results of a long-term survey. Blood 2007; 109: 93–99. 6 Dong F, Dale DC, Bonilla MA, Freedman M, Fasth A, Neijens HJ et al. Mutations in the granulocyte colony-stimulating factor receptor gene in patients with severe congenital neutropenia. Leukemia 1997; 11: 120–125. 7 Germeshausen M, Schulze H, Kratz C, Wilkens L, Repp R, Shannon K et al. An acquired G-CSF receptor mutation results in increased proliferation of CMML cells from a patient with severe congenital neutropenia. Leukemia 2005; 19: 611–617. 8 Carapeti M, Soede-Bobok A, Hochhaus A, Sill H, Touw IP, Goldman JM et al. 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 1997; 11: 1005–1008. 9 Beekman R, Valkhof MG, Sanders MA, van Strien PM, Haanstra JR, Broeders L et al. Sequential gain of mutations in severe congenital neutropenia progressing to acute myeloid leukemia. Blood 2012; 119: 5071–5077. 10 Fukunaga R, Seto Y, Mizushima S, Nagata S. Three different mRNAs encoding human granulocyte colony-stimulating factor receptor. Proc Natl Acad Sci USA 1990; 87: 8702–8706. 11 Fukunaga R, Ishizaka-Ikeda E, Seto Y, Nagata S. Expression cloning of a receptor for murine granulocyte colony-stimulating factor. Cell 1990; 61: 341–350. 12 White SM, Ball ED, Ehmann WC, Rao AS, Tweardy DJ. Increased expression of the differentiation-defective granulocyte colony-stimulating factor receptor mRNA isoform in acute myelogenous leukemia. Leukemia 1998; 12: 899–906. 13 White SM, Alarcon MH, Tweardy DJ. Inhibition of granulocyte colony-stimulating factor-mediated myeloid maturation by low level expression of the differentiationdefective class IV granulocyte colony-stimulating factor receptor isoform. Blood 2000; 95: 3335–3340. 14 Fukunaga R, Ishizaka-Ikeda E, Nagata S. Growth and differentiation signals mediated by different regions in the cytoplasmic domain of granulocyte colonystimulating factor receptor. Cell 1993; 74: 1079–1087. 15 Sloand EM, Yong AS, Ramkissoon S, Solomou E, Bruno TC, Kim S et al. Granulocyte colony-stimulating factor preferentially stimulates proliferation of monosomy 7 cells bearing the isoform IV receptor. Proc Natl Acad Sci USA 2006; 103: 14483–14488. 16 Ehlers S, Herbst C, Zimmermann M, Scharn N, Germeshausen M, von Neuhoff N et al. Granulocyte colony-stimulating factor (G-CSF) treatment of childhood acute myeloid leukemias that overexpress the differentiation-defective G-CSF receptor isoform IV is associated with a higher incidence of relapse. J Clin Oncol 2010; 28: 2591–2597. 17 Horan T, Wen J, Narhi L, Parker V, Garcia A, Arakawa T et al. Dimerization of the extracellular domain of granuloycte-colony stimulating factor receptor by ligand binding: a monovalent ligand induces 2:2 complexes. Biochemistry 1996; 35: 4886–4896. 18 Touw IP, van de Geijn GJ. Granulocyte colony-stimulating factor and its receptor in normal myeloid cell development, leukemia and related blood cell disorders. Front Biosci 2007; 12: 800–815. 19 Hermans MH, van de Geijn GJ, Antonissen C, Gits J, van Leeuwen D, Ward AC et al. Signaling mechanisms coupled to tyrosines in the granulocyte colony-stimulating factor receptor orchestrate G-CSF-induced expansion of myeloid progenitor cells. Blood 2003; 101: 2584–2590. 20 Akbarzadeh S, Ward AC, McPhee DO, Alexander WS, Lieschke GJ, Layton JE. Tyrosine residues of the granulocyte colony-stimulating factor receptor transmit proliferation and differentiation signals in murine bone marrow cells. Blood 2002; 99: 879–887. 21 Zhu QS, Robinson LJ, Roginskaya V, Corey SJ. G-CSF-induced tyrosine phosphorylation of Gab2 is Lyn kinase dependent and associated with enhanced Akt and differentiative, not proliferative, responses. Blood 2004; 103: 3305–3312.

22 Futami M, Zhu QS, Whichard ZL, Xia L, Ke Y, Neel BG et al. G-CSF receptor activation of the Src kinase Lyn is mediated by Gab2 recruitment of the Shp2 phosphatase. Blood 2011; 118: 1077–1086. 23 Hunter MG, Jacob A, O’Donnell LC, Agler A, Druhan LJ, Coggeshall KM et al. Loss of SHIP and CIS recruitment to the granulocyte colony-stimulating factor receptor contribute to hyperproliferative responses in severe congenital neutropenia/acute myelogenous leukemia. J Immunol 2004; 173: 5036–5045. 24 Nicola NA, Begley CG, Metcalf D. Identification of the human analogue of a regulator that induces differentiation in murine leukaemic cells. Nature 1985; 314: 625–628. 25 Ishizaka-Ikeda E, Fukunaga R, Wood WI, Goeddel DV, Nagata S. Signal transduction mediated by growth hormone receptor and its chimeric molecules with the granulocyte colony-stimulating factor receptor. Proc Natl Acad Sci USA 1993; 90: 123–127. 26 Dick EP, Prince LR, Sabroe I. Ex vivo-expanded bone marrow CD34 þ derived neutrophils have limited bactericidal ability. Stem Cells 2008; 26: 2552–2563. 27 True CD. Preparation of an iodinated radioligand. Curr Protoc Neurosci 2001, Appendix 3: Appendix 3A. 28 Bolton AE, Hunter WM. The labelling of proteins to high specific radioactivities by conjugation to a 125I-containing acylating agent. Biochem J 1973; 133: 529–539. 29 Mehta HM, Woo SB, Neet KE. Comparison of nerve growth factor receptor binding models using heterodimeric muteins. J Neurosci Res 2012; 90: 2259–2271. 30 Hermans MH, Antonissen C, Ward AC, Mayen AE, Ploemacher RE, Touw IP. Sustained receptor activation and hyperproliferation in response to granulocyte colony-stimulating factor (G-CSF) in mice with a severe congenital neutropenia/ acute myeloid leukemia-derived mutation in the G-CSF receptor gene. J Exp Med 1999; 189: 683–692. 31 Aarts LH, Roovers O, Ward AC, Touw IP. Receptor activation and 2 distinct COOHterminal motifs control G-CSF receptor distribution and internalization kinetics. Blood 2004; 103: 571–579. 32 Hunter MG, Avalos BR. Deletion of a critical internalization domain in the G-CSFR in acute myelogenous leukemia preceded by severe congenital neutropenia. Blood 1999; 93: 440–446. 33 Baffert F, Regnier CH, De Pover A, Pissot-Soldermann C, Tavares GA, Blasco F et al. Potent and selective inhibition of polycythemia by the quinoxaline JAK2 inhibitor NVP-BSK805. Mol Cancer Ther 2010; 9: 1945–1955. 34 Verstovsek S, Kantarjian H, Mesa RA, Pardanani AD, Cortes-Franco J, Thomas DA et al. Safety and Efficacy of INCB018424, a JAK1 and JAK2 Inhibitor, in Myelofibrosis. N Engl J Med 2010; 363: 1117–1127. 35 Quinta´s-Cardama A, Vaddi K, Liu P, Manshouri T, Li J, Scherle PA et al. Preclinical characterization of the selective JAK1/2 inhibitor INCB018424: therapeutic implications for the treatment of myeloproliferative neoplasms. Blood 2010; 115: 3109–3117. 36 Ward AC, van Aesch YM, Schelen AM, Touw IP. Defective internalization and sustained activation of truncated granulocyte colony-stimulating factor receptor found in severe congenital neutropenia/acute myeloid leukemia. Blood 1999; 93: 447–458. 37 Gits J, van Leeuwen D, Carroll HP, Touw IP, Ward AC. Multiple pathways contribute to the hyperproliferative responses from truncated granulocyte colony-stimulating factor receptors. Leukemia 2006; 20: 2111–2118. 38 van de Geijn G-JM, Gits J, Aarts LHJ, Heijmans-Antonissen C, Touw IP. G-CSF receptor truncations found in SCN/AML relieve SOCS3-controlled inhibition of STAT5 but leave suppression of STAT3 intact. Blood 2004; 104: 667–674. 39 Irandoust MI, Aarts LHJ, Roovers O, Gits J, Erkeland SJ, Touw IP. Suppressor of cytokine signaling 3 controls lysosomal routing of G-CSF receptor. EMBO J 2007; 26: 1782–1793. 40 Ward AC, Loeb DM, Soede-Bobok AA, Touw IP, Friedman AD. Regulation of granulopoiesis by transcription factors and cytokine signals. Leukemia 2000; 14: 973–990. 41 Yin T, Shen R, Feng G-S, Yang Y-C. Molecular characterization of specific interactions between SHP-2 phosphatase and JAK tyrosine kinases. J Biol Chem 1997; 272: 1032–1037. 42 Wang L, Xue J, Zadorozny EV, Robinson LJ. G-CSF stimulates Jak2-dependent Gab2 phosphorylation leading to Erk1/2 activation and cell proliferation. Cell Signal 2008; 20: 1890–1899. 43 Ward AC, Smith L, de Koning JP, van Aesch Y, Touw IP. Multiple signals mediate proliferation, differentiation, and survival from the granulocyte colony-stimulating factor receptor in myeloid 32D cells. J Biol Chem 1999; 274: 14956–14962. 44 Maxson JE, Gotlib J, Pollyea DA, Fleischman AG, Agarwal A, Eide CA et al. Oncogenic CSF3R mutations in chronic neutrophilic leukemia and atypical CML. N Engl J Med 2013; 368: 1781–1790. 45 Mehta HM, Glaubach T, Long A, Lu H, Przychodzen B, Makishima H et al. Granulocyte colony-stimulating factor receptor T595I (T618I) mutation confers ligand independence and enhanced signaling. Leukemia 2013; 27: 2407–2410.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu) & 2014 Macmillan Publishers Limited

Leukemia (2014) 1041 – 1051