Widespread expression of erythropoietin receptor in brain and its induction by injury

Erythropoietin However, unequivocal demonstration of erythropoietin receptor (EPOR) expression in brain cells has remained difficult since previously available anti-EPOR antibodies (EPOR-AB) were unspecific. We report here a new, highly specific, polyclonal rabbit EPOR-AB directed against different epitopes in the cytoplasmic tail of human and murine EPOR and its characterization by mass spectrometric analysis of immunoprecipitated endogenous EPOR, Western blotting, immunostaining, and flow cytometry. Among others, we applied genetic strategies including overexpression, Lentivirus-mediated conditional knockout of EpoR , and tagged proteins, both on cultured cells and tissue sections, as well as intracortical implantation of EPOR -transduced cells to verify specificity. We show examples of EPOR expression in neurons, oligodendroglia, astrocytes, and microglia. Employing this new EPOR-AB with double-labelling strategies, we demonstrate membrane expression of EPOR as well as its localization in intracellular compartments such as the Golgi apparatus. Moreover, we show injury-induced expression of EPOR: In mice, a stereotactically applied stab wound to the motor cortex leads to distinct EpoR expression by reactive GFAP-expressing cells in the lesion vicinity. In a patient suffering from epilepsy, neurons and oligodendrocytes of the hippocampus strongly express EPOR. To conclude, this new analytical tool will allow neuroscientists to pinpoint EPOR expression in cells of the nervous system and to better understand its role in healthy conditions, including brain development, as well as under pathological circumstances, such as upregulation upon distress and injury.


INTRODUCTION
The growth factor erythropoietin (EPO) was named based on its first discovered effects on cells of the hematopoietic system. For >20 years it has been shown to act on other tissues, including the brain, e.g. (1)(2)(3)(4)(5). Its remarkable neuroprotective, neuroregenerative and procognitive effects make EPO an attractive candidate for treating human brain disease, e.g. (6)(7), and an important target of neuroscience research. In 1989, the EPO receptor (EpoR) has first been cloned in mouse (8), soon followed by cloning and characterization of the human EPOR gene (9)(10). A single EPO molecule binds to 2 specific cytokine type1 transmembrane receptor molecules, each with a calculated molecular mass of 59kDa, that together form the classical homodimeric EPOR (2,11). Binding of EPO to its receptor induces a conformational change, initiating EPOR-associated JAK2 transphosphorylation and multiple, celltype specific, downstream signal transduction cascades. These cascades include signal transducers and activators of transcription (STATs), phosphatidylinositol-3 kinase (PI3K)/AKT, RAS/extracellular signal-regulated kinase (ERK1/2), nuclear factor kappa B (NF-kappa B). Activation of these signaling cascades leads to further activation of anti-apoptotic factors and pathways, stimulation of cell differentiation, including induction of cellular shape-change and growth, or modulation of plasticity, in a cell-type and stimulation-dependent manner (5,(12)(13).
Antibodies against EPOR (EPOR-AB) have been widely used to characterize EPOR expression and localization, but cell surface EPOR expression is low, even in stimulated states, and most importantly, all commercially available EPOR-AB have been hampered by non-specific cross-reactivities, questioning the exclusively hereon based literature. This in turn raised discussions within the scientific community, questioning the expression of EPOR in extra-hematopoietic tissues (14)(15)(16). These discussions were likely nurtured by conflicts of interest, trying to 'restrict' the effects of EPO, a commercially for the anemia market highly attractive compound, to hematopoiesis. Nevertheless, they made it very obvious that the existing EPOR-AB were essentially unreliable, and that the production and thorough characterization of new and more specific EPOR-AB had to be seen as a major challenge for the future (14,(17)(18).

UNCORRECTED PROOF
Molecular Medicine www.molmed.org Independent of work based on EPOR-AB, genetically altered mice helped to demonstrate that EPOR signalling is necessary for normal brain development (19) and that it has a distinct function in neurogenesis (20). In addition, EPO and EPOR mRNA are expressed in brain tissue (21), and specific binding sites for EPO in brain have been demonstrated in mouse and human by means of radiolabelled EPO (22)(23). In cell culture, mRNA expression combined with functional assays, e.g. altered phosphorylation of second messenger pathways induced by EPO in microglia, served to prove specific EPOR expression in the absence of reliable EPOR-AB (24).
The fact that cellular EPOR protein expression has been difficult to assess, strongly limited the in-depth investigation of the EPO/EPOR system. Particularly in the human brain, the study of its (patho-) physiological role has been highly constrained since additional means of verification as used in experimental animals and cell cultures are naturally excluded. Recognizing this critical issue in EPO/EPOR research, we aimed at generating reliable EPOR-AB. We present here the comprehensive characterization of a novel, highly specific EPOR-AB, using an array of state-of-theart technologies. This new AB tool may help overcome the described obstacles and lead to revisiting some of the reported data.

Polyclonal AB
Two rabbits were immunized with a purified recombinant protein corresponding to amino acids (AA) 273-508 (intracellular C-terminus) of the unprocessed human EPOR. The coding sequence was generated by gene synthesis (Geneart, Regensburg, Germany) and ligated via EcoRI and HindIII into the bacterial expression vector pASK-IBA37+ (IBA-Lifescience, Göttingen, Germany). The recombinant His-Tag fusion protein was expressed and purified by Ni-NTA affinity chromatography according to the manufacturer's manual. Crude antiserum SA7378 was affinity purified with the immunogen coupled to CNBr Sepharose (GE-UNCORRECTED PROOF Molecular Medicine www.molmed.org Healthcare, Freiburg, Germany). The AB is referred to as ctEPOR-AB in this manuscript.

Monoclonal AB
AB producing hybridomas were generated by Synaptic Systems (Göttingen; see also http://www.sysy.com/mabservice.html) as follows: Three 8-10 weeks old BALB/c female mice were subcutaneously immunized with a synthetic peptide corresponding to AA 25-39 (extracellular N-terminus) of unprocessed human EPOR precursor coupled to KLH via a C-terminal cysteine over a period of 75 days. Cells from the knee lymph nodes were fused with the mouse myeloma cell line P3X63Ag8.653 (ATCC CRL-1580). Resulting hybridomas were screened by direct ELISA against the immunogen and immunofluorescence on 3T3 NIH fibroblasts overexpressing fulllength human EPOR. Clone 45A3, used in this study, was re-cloned two times by limiting dilution and the immunoglobulin subclass was determined (IgG2b). The AB is referred to as ntEPOR-AB in this manuscript.

Cell lines
The following human cell lines were used:

Human IPS cells
Human material was used in accordance with ethical guidelines and the Helsinki Declaration. Subjects gave informed consent regarding generation and use of IPS cells or scientific investigation of brain samples. Human fibroblasts were reprogrammed using a non-integrative RNA-based virus to induce the expression of 4 reprogramming factors: OCT4, SOX2, KLF4 and c-MYC (CytoTune®-iPS 2.0 Sendai Reprogramming Kit, Life Technologies GmbH). After transduction, IPS cells (clones isAu1-3; isAu3-2) were adapted to a feeder-free culture system (Matrigel® matrix, Corning, Wiesbaden, Germany) and cultured in TeSR™-E8™ medium (Stem Cell TechnologiesTM, Cologne, Germany).

Primary mouse cell culture
The preparation and culture conditions of primary mouse oligodendrocytes and microglia are described in detail elsewhere (24)(25). In brief, oligodendrocytes were prepared from the forebrains of newborn P1-2 NMRI mice. After differentiation, oligodendrocyte precursors were shaken off from a bottom layer of astrocytes and seeded in Super-Sato medium (DMEM with high glucose supplemented with B-27 supplement, 2mM GlutaMAX, 1mM sodium pyruvate, 1% horse serum (HS), 50U/mL penicillin and 50µg/mL streptomycin, all from Life Technologies GmbH) and 0.5mM triiodothyronine, and 0.52mM L-thyroxine both Merck, Darmstadt, Germany). For primary microglia, newborn C57BL6 mice (P0-P1) were used. The cell suspension derived from their forebrains was seeded in high glucose DMEM medium with 10% HS, 1% GlutaMAX supplement, 50U/mL penicillin and 50µg/ml streptomycin (all from Life Technologies GmbH). Half of the microglia-conditioned medium was exchanged by fresh medium 3-4 days later, and at day 7, the medium was partially replaced by L929-conditioned medium. Primary microglia were detached by shaking of flasks and seeded in serum-free microglial growth medium (high glucose DMEM with 1mM sodium pyruvate, 1.5g/L sodium bicarbonate, 100U/mL penicillin and 100µg/mL streptomycin, all from Life Technologies GmbH).

Lentiviral transduction of primary cells
EpoR conditional mouse mutants with floxed exons 3-6 were generated on the  (26). The cells were infected with lentiviruses at day1 in vitro. The viral constructs contained either a cassette for GFP only (control) or a cassette for GFP and Crerecombinase. Protein was extracted on day10 in vitro.

STAT5 phosphorylation assay
This assay is described in detail elsewhere (17). In brief, UT-7 or OCIM-1 cells were serum-and EPO-deprived overnight (1% FBS in IMDM, both Life Technologies GmbH). On the next day, they were incubated with different concentrations of recombinant human EPO (rhEPO, NeoRecormon, Roche) or control solution for 15min, followed by protein extraction for Western blotting. Immunodetection was done with anti-phosphorylated STAT5 (1:500, Cell Signaling, Danvers, MA, USA) and GAPDH (1:5000, Enzo Life Sciences, Farmingdale, NY, USA); 20µg of protein was loaded for SDS-PAGE. www.molmed.org motor cortex. After implantation, the needle was left in place for 2min, then slowly withdrawn from the brain and the skin incision closed with sterile suture. Directly after the skin incision was closed, the animals received carprofen i.p. for pain treatment which was repeated every 6-8h (5mg/kg Rimadyl®, Pfizer). At 24h after surgery, animals were anesthetized i.p. (0.276mg/g Tribromoethanol; Sigma-Aldrich) and
Before IP, lysates were diluted 1:1 with IP buffer to obtain 0.5% Triton X-100. For the EPOR IP, protein-G sepharose beads were covalently linked to ctEPOR-AB (Synaptic Systems) with 40mmol/L dimethyl-pimelimidate (Sigma-Aldrich). Bead slurry (200µL; Thermo Scientific, Waltham, MA, USA) was cross-linked with 400µg ctEPOR-AB or 400µg of the IgG fraction from the same rabbit before immunization.
For EPOR IP from UT-7 protein lysates, 9µg of ctEPOR-AB coupled to protein-G sepharose per 1mg protein lysate were incubated for 2h at 4°C. Afterwards, beads were centrifuged and washed. For immunoblot analysis, EPOR was eluted from the beads by repeated boiling in Laemmli buffer at 95°C. For mass spectrometric protein identification, EPOR was eluted as before but with a non-reducing SDS-buffer (without β-mercaptoethanol) to prevent masking of the EPOR by excess AB heavy chains in the subsequent gel electrophoresis. To increase the efficiency of EPOR capture from UT-7 protein lysates, 2 consecutive IP with fresh beads were performed in a way that the flow-through of the first IP was used as input for the second. Eluted proteins were precipitated by methanol/chloroform treatment (30). Pellets were UNCORRECTED PROOF Molecular Medicine www.molmed.org solubilized in reducing sample buffer and pooled prior to electrophoresis. As starting material for mass spectrometry 4mg protein lysate from UT-7 cells was used.

SDS-PAGE and Western blots
SDS-PAGE was performed with self-made 10% SDS-polyacrylamide gels. As protein ladder we used PageRulerTM Plus pre-stained and SeeBlue Plus2 pre-stained in this gel system (both Life Technologies GmbH). For all Western blotting, the following protein amounts were loaded: [1] 15 µg for lysates derived from cell lines; [2] 20 µg for lysates derived from primary cultures; [3] 50 µg for lysates derived from tissue.
Afterwards, proteins were transferred to a nitrocellulose membrane and blocked with 4% milk powder and 4% HS in Tris buffered saline with 0.05% Tween 20.

Protein identification
Gel regions of interest were identified by overlaying images from colloidal Coomassie staining and immunodetection in the Delta 2D image analysis software (Decodon, Greifswald, Germany). Gel bands were excised manually and subjected to automated in-gel digestion with trypsin as described previously (31).

Immunohistochemistry on paraffin embedded human brain sections
Brain slices of 1-3µm thickness from formalin-fixed and paraffin-embedded tissue blocks were deparaffinized. Endogenous peroxidases were blocked with 3% H 2 O 2 in PBS for 20min followed by epitope blocking with 0.02% casein in PBS for 15min.

Generation of EPOR-AB
The aim of this work was to generate specific and sensitive EPOR-AB and to investigate EPOR expression in the central nervous system (CNS). Therefore, polyclonal rabbit EPOR-AB and monoclonal mouse EPOR-AB were produced and tested. After extensive characterization of a whole panel of AB (data not shown), 2 highly promising candidates were selected and validated for several research UNCORRECTED PROOF Molecular Medicine www.molmed.org purposes: (I) The polyclonal rabbit EPOR-AB SA7378, directed against the Cterminus (here always referred to as ctEPOR-AB) and (II) the monoclonal mouse EPOR-AB 45A3, directed against the N-terminus (here referred to as ntEPOR-AB).
The present study is mainly built on ctEPOR-AB because of its high specificity and broad spectrum of applications in human and mouse.

Functional EPOR validation in the test systems
As prerequisite of testing EPOR-AB in the cell lines used here, we functionally validated their EPOR expression. In the EPO-dependent megakaryoblastic leukaemia cell line UT-7, incubation with different concentrations of rhEPO led to STAT-5 phosphorylation (Fig.1A). Also, in the erythroleukaemia cell line OCIM-1, incubation with rhEPO induced STAT-5 phosphorylation (Fig.1B). UT-7 cells only proliferated and survived in the presence of rhEPO in the medium (Fig.1C-D). In the mouse microglia cell line EOC-20, stably transduced with N-terminally HA-tagged human EPOR, rhEPO administration activated MAPK phosphorylation in a concentration-dependent manner (Fig.1E). We also confirmed EPOR mRNA expression in all of these cell lines by qPCR (normalized to GAPDH as housekeeping gene -data not shown).

Detection of EPOR/EpoR by Western blotting
To confirm reliable detection of EPOR by Western blotting, we transfected HEK293 FT cells with different EPOR expression vectors. The polyclonal rabbit ctEPOR-AB detected full-length human EPOR and its degradation product specifically while the C-terminally truncated mutant of EPOR was not detected by this AB (Figs.1F-G).

Immunoblots of lysates from transduced EOC-20 cells (N-terminally HA-tagged
human EPOR) and respective controls showed specific detection of full-length human EPOR by ctEPOR-AB (Fig.1H). This was validated with an HA immunoblot of the same lysates (Fig.1I). In protein lysates from UT-7 and OCIM-1 cells, ctEPOR-AB detected bands of the same molecular weight (Fig.1J). As shown in Fig.1F and J, in UT-7 cells and HEK293 FT cells (only when transfected with the full length human EPOR) a specific degradation product was additionally detected at ~40kDa by the ctEPOR-AB. This degradation product of EPOR has been described earlier (11). with ctEPOR-AB (Fig.1K). EPOR was further recognized by ctEPOR-AB in human placenta and fetal brain extracts (Fig.1L). In addition to human EPOR, ctEPOR-AB detected murine EpoR in transfected HEK293 FT cells, mouse fetal liver and lysates from cultured primary mouse oligodendrocytes (Fig.1M). To validate the specificity of murine EpoR detection, EpoR was knocked out in primary astrocytes derived from EpoR-fl/fl mice. In fact, ctEPOR-AB recognized a doublet of bands (EPOR with or without N-glycosylation, resulting in a difference of ~3kDa) with a molecular weight of around 65kDa (Fig.1N, control transduction). Shown is a clear reduction upon expression of Cre-recombinase. The residual expression of the protein is likely due to a slow turnover of EpoR, slow kinetics of Cre-recombination of the floxed EpoR allele, or incomplete infection of the lentivirus. In any case, Cre-dependent reduction of the signal led us to conclude that ctEPOR-AB specifically detects EpoR. Together, these results indicate specific EPOR/EpoR detection with ctEPOR-AB in human and murine cell and tissue extracts.

EPOR protein identification
To test whether the specific band detected in Western blots is indeed EPOR, we performed immunoprecipitations (IP) from UT-7 lysates and subsequent mass spectrometric protein identification. We used covalently immobilized ctEPOR-AB in combination with non-reducing elution conditions to minimize the masking effect of excess antibody heavy chains, which have an apparent electrophoretic mobility similar to the EPOR. After IP with ctEPOR-AB, eluted proteins from ctEPOR-AB protein-G sepharose beads and respective control beads were separated by SDS-PAGE and visualized by colloidal Coomassie staining or immunoblotting. The overlay of the 2 gel images ( Fig.2A) was used to identify the region of the Coomassie-stained gel potentially containing the EPOR protein. Identical gel regions from ctEPOR-AB IP and the control IP were excised and subjected to tryptic digestion followed by liquid chromatography coupled to mass spectrometry (LC-MS). Against a common background mainly consisting of chaperone proteins, human EPOR protein was detected in eluates from ctEPOR-AB beads, but not from control beads. The identification of 11 EPOR derived peptides with high mass accuracy at both precursor and fragment ion level resulted in sequence coverage of 22.6% (Fig.2B-C), basically in line with recent LC-MS data on EPOR immunoprecipitates (37). Taken together, UNCORRECTED PROOF Molecular Medicine www.molmed.org our results show that ctEPOR-AB indeed binds specifically to full-length EPOR. Also, in reducing conditions we could effectively elute EPOR after IP (Fig.2D). Noteworthy, we detected the EPOR protein in Western blots between 59-68 kDa, depending on the gel system and protein molecular weight marker used (see Fig.1, 2A and D and material and methods).

EPOR/EpoR detection by immunocytochemistry
To validate the specificity of ctEPOR-AB on formaldehyde fixed cells, we stained the same antigen with AB directed against different epitopes (38). In EOC-20 cells transduced with an N-terminally HA-tagged human EPOR, anti-HA and ctEPOR-AB double-staining showed almost complete co-localization with most of the signal located intracellularly (Fig.3A). Control EOC-20 cells were negative (Fig.3A). In UT-7 cells, ctEPOR-AB revealed a similar staining pattern (Fig.3B). EPOR staining was colocalized with Golgi staining, indicating detection of a membrane protein (Fig.3C). In addition, ntEPOR-AB and ctEPOR-AB double-stained EPOR in UT-7 cells (Fig.3D).

EpoR detection in the brain of healthy mice
Using ctEPOR-AB on frozen brain sections of healthy young mice, we found EpoR expression mainly in a subpopulation of cells of the oligodendrocyte lineage (Fig.4A).

To get better insight at which stages cells of the oligodendrocyte lineage express
EpoR, we labelled oligodendrocyte precursor cells by tamoxifen injections in NG2-CreERT2 mice (28). At 72h after the second tamoxifen injection, we identified precursors double-stained for GFP and ctEPOR-AB (GFP+/EPOR+), as well as GFP+/EPOR+ cells with clear morphology (processes with parallel myelin bundles) of already differentiated oligodendrocytes (Fig.4B-B''). Moreover, GFAP+/ EPOR+ cells were seen in postnatal neurogenesis areas such as dentate gyrus (Fig.4C)

EPOR/EpoR detection in the injured CNS of mice
Next, we stereotactically injected EPOR-transduced EOC-20 cells or medium only ('stab wound' analogue) in the motor cortex of adult mice (Fig.4D). This experiment served 2 purposes: [1] to recover defined cells that carry human EPOR in brain sections; [2] to confirm injury-induced endogenous EpoR expression, since in earlier work, we had proposed upregulation of EPOR upon injury (6). At 24h after injection of medium only ('stab wound'), we saw cells with strong ctEPOR-AB signal in close proximity to the injection site (Fig.4E). Many of these cells were GFAP+/EPOR+ (Fig.4F). On the contralateral site, no GFAP+/EPOR+ cells were seen (Fig.4G). In the motor cortex of mice injected with transduced EOC-20 cells (murine microglia cell line), double-labelling with HA-AB and ctEPOR-AB confirmed specific recovery of these cells in frozen sections of paraformaldehyde-perfused mice. Also in this condition, endogenous cells with high EpoR expression were observed in close proximity of the injection site (Fig.4H-H''). These results confirm the pronounced upregulation of EpoR in cells reacting to injury, provoked here by an experimental stab wound.

Upregulation of EPOR in the hippocampal formation of a patient suffering from temporo-mesial complex-focal epilepsy
Formalin-fixed, paraffin-embedded tissue from a patient, who underwent selective unilateral hippocampectomy, was used for immunohistochemical detection of EPOR upregulation under these conditions (Fig.5A). The patient had been suffering from pharmaco-resistant complex-focal seizures of temporo-mesial origin for more than 10 years. Neuropathological analysis of the surgery material revealed hippocampal sclerosis stage Wyler III. EPOR was upregulated in several but not all remaining neurons of CA1 (Fig.5B), of CA4 (Fig.5C) and of the dentate gyrus (Fig.5D), as well as in oligodendrocytes and endothelial cells of capillaries in the adjacent white matter (Fig.5E). Without primary AB, no staining could be detected (Fig.5F). This suggests upregulation of EPOR upon severe chronic distress in different cell types of the human CNS.

DISCUSSION
In the present work, we took the challenge requested for a long time by the scientific community (14,(17)(18) to generate a specific AB for valid detection of EPOR in human and murine cells and tissues, with particular focus on the brain. We present here a highly specific polyclonal rabbit AB directed against the intracellular Cterminus of the human EPOR. This AB, referred to as ctEPOR-AB, specifically recognizes EPOR, as proven by mass spectrometry, and has a broad range of documented applications in both human and murine cells and tissues, ranging from Western blotting, flow cytometry, and immunoprecipitation to immunocytochemistry and immunohistochemistry on frozen as well as paraffin-embedded sections.
Importantly, by employing this AB, we were able to confirm expression of EPOR in brain cells and its upregulation upon injury (39). Our comprehensive in vitro and in vivo data clearly reject earlier claims, solely based on in vitro studies, that EPOR expression and EPO function outside the hematopoietic system does not exist (15).
The great need for specific EPOR-AB in the field is also reflected by a very recent study of Drorit Neumann and colleagues (37). This group of authors published specific mouse and rat monoclonal EPOR-AB that detect EPOR expression in human cancer cells and tissues (37). Complementary to this approach and with particular focus on the brain, we have developed a highly specific and sensitive polyclonal rabbit AB, ctEPOR-AB, suitable for applications not only in human but also in murine material. Since the polyclonal nature of this AB has limitations, not only due to the restricted life time of a rabbit, we are currently working on further exploitation of the herewith acquired knowledge. Preliminary results of epitope mapping with this polyclonal ctEPOR-AB revealed only few strongly recognized epitopes. These epitopes are presently used for generating specific mouse monoclonal AB. They will be tested alone or in the form of potentially more sensitive 'cocktails', with collective properties similar to the here reported ctEPOR-AB.
With the examples of EPOR expression in the brain shown here, we confirmed earlier work of ourselves and others which needed in the past additional methods for validation, and still left doubts in the scientific community due to the non-specificity of UNCORRECTED PROOF Molecular Medicine www.molmed.org previous EPOR-AB. For instance, GFAP positive stem cells in the adult neurogenic niches showed EpoR immunoreactivity here, which is in line with reports demonstrating distinct effects of EPO on adult neural stem cells (40)(41). Also, studies identifying EPO as inducer of oligodendrocyte precursor cell differentiation (42)(43) are now further supported by the detection of specific EPOR binding sites in culture and brain sections. Importantly, the role of the EPO/EPOR system in response to brain injury (5-6) is confirmed with our stab wound approach.
To conclude, based on the here reported novel tool, it will now be possible to investigate the role of EPOR in the intact and injured human and murine brain in more detail. This, in turn, will facilitate the development of EPO for therapeutic use outside the hematopoietic system.