RHEX, a novel regulator of human erythroid progenitor cell expansion and erythroblast development

EPO ligation of dimeric EPOR complexes activates JAK2 kinase and the phosphorylation of nine EPOR cytoplasmic PY motifs (Fig. 1 A). Certain signal transduction factors (STFs) that dock at EPOR PY sites are well defined (e.g., p85α at PY480, STAT5 at PY344) and some are also direct JAK2 targets (e.g., STAT5, phospholipase Cγ), whereas others couple to downstream signaling modules (Wojchowski et al., 2010; Watowich, 2011; Broxmeyer, 2013). To seek novel EPOR STFs, a PY-phosphoproteomic approach was applied using UT7epo cells as a human EPC model (Komatsu et al., 1993). This involved hematopoietic growth factor withdrawal with or without EPO challenge, tryptic digests of lysates, and PY peptide isolation plus liquid chromatography-tandem mass spectrometry (LC-MS/MS) identification (Fig. 1 B). Overall, 54 unique EPO/EPOR-modulated PY proteins were identified with ≥2.5–82.6-fold modulation (Table S1). These included known STFs modulated at known PY sites, known STFs modulated at novel sites, known proteins not previously associated with EPO/EPOR signaling, and novel targets encoded by predicted ORFs. As one major functional subset, 11 targets proved to be molecular adaptors. GAB1-3, SHC1, IRS2, DOK1, and SPRY1 comprise known (and validating) targets (Fig. 1 C), whereas SH2D2A, STAM2, and SPRY2, together with a C1ORF186 product, represent novel candidate EPO/EPOR STFs (Fig. 1 D). C1ORF186 (designated as RHEX) was up-modulated by EPO ≥20-fold in its phosphorylation at PY132 and PY141 sites (single PY peptide; Fig. 1 E) and is this report’s prime focus.

RHEX is encoded at a six-exon C1ORF186 locus (Fig. 2 A) that generates a singular predicted 1.6 kb nt coding transcript. Northern blotting detected major 1.6 kb, and minor <0.5 kb nt transcripts (Fig. 2 B). Interestingly, RHEX proved to be well conserved in Homo sapiens and primates (99% nt conservation) but was not detected in rat, mouse, or lower vertebrate genomes. RHEX transcript expression among tissues and blood cells was also investigated and was relatively high level in primary human EPCs and kidney (Fig. 2, C and D). RNA-Seq also indicated elevated RHEX levels in CFUe as compared with CD34^pos progenitors (Fig. 2 E). At a protein level, RHEX’s predicted domains included an amino-terminal (NT) hydrophobic region and two carboxy-terminal candidate GRB2 binding sites (Neumann et al., 2009; Fig. 2, F and G). RHEX, however, is unique and exhibits homology only with limited residues of a recently reported erythrocytic spectrin (NP_003117.2). Basic assessments of RHEX levels among human hematopoietic cell lines (and 293 cells) using polyclonal antiserum to RHEX further revealed expression only in erythroid UT7epo cells (Fig. 2 H).

To analyze RHEX’s subcellular localization and actions, a (PY)RHEX reactive monoclonal antibody was next generated and was used in UT7epo cells to first validate rapid EPO induction of PY-RHEX (Fig. 3 A and not depicted). Human SCF, IL3, GMCSF, TPO, Flt3L, or serum, in contrast, did not detectably stimulate RHEX’s PY phosphorylation (unpublished data). RHEX’s hydrophobic NT region prompted subcellular localization analyses. As indicated by CD71 and wheat germ agglutinin (WGA) markers, (PY)RHEX resided at the plasma membrane (Fig. 3 B). Experiments using JAK2 and SRC kinase inhibitors (TG101348 and Dasatinib, respectively) pointed to RHEX as a JAK2 target (Fig. 3 C). Beyond this, coimmunoprecipitation experiments using an RHEX-flag construct indicated co-association with EPOR and JAK2 complexes (Fig. 3 D).

To assess function, RHEX knockdown experiments were performed. Three assessed shRNAs each efficiently inhibited RHEX expression (Fig. 4 A and not depicted). As assayed via clonal colony formation, RHEX knockdown limited UT7epo cell growth ≥3-fold (Fig. 4 B, top). Analyses in liquid culture confirmed effects on growth (with no significant effects on survival), and a parallel attenuation of ERK1,2 activity was also observed (Fig. 4, B and C). Possible effects of RHEX on murine EPCs were also studied via lentiviral expression in G1E-ER4 cells (Welch et al., 2004). In this proerythroblast model, mEPOR ligation did not effectively induce PY-RHEX formation (Fig. 4 D). RHEX, furthermore, was unstable and no reproducible effects on proliferation or on Gata1-ER–induced late-stage differentiation were observed (unpublished data).

RHEX’s actions were next studied in human CD34^pos progenitor-derived EPCs. Endogenous RHEX transcript levels became elevated at days 5–7 of culture (pro- to basophilic erythroblasts; Fig. 5, A and B). In developmentally staged EPCs as isolated via FACS, RHEX was elevated in proerythroblasts (Fig. 5 C and not depicted). In shRNA lentivirus transduction experiments, when RHEX was knocked down, EPC expansion was attenuated (without significant effects on viability; Fig. 5, D and E). Beyond this, glycophorin-A (GPA) levels were skewed and increased when RHEX expression was inhibited (Fig. 5 F). Cytospin morphologies also revealed apparent effects of RHEX knockdown on erythroid development, including a delayed formation of maturing normoblasts (Fig. 5 G). By inspection of cell pellets, Western blotting, and RT-PCR, hemoglobinization was also attenuated, which is further consistent with a role for RHEX in supporting erythroblast development at a normoblast stage (Fig. 5, H and I). Tests of GRB2 as a candidate RHEX partner were also advanced. As analyzed in UT7epo cells, RHEX-Flag and GRB2 coimmunoprecipitated (Fig. 5 J; as did these endogenous STFs, Fig. 5 K). Here, EPO exposure moderately increased levels of GRB2 plus RHEX coimmunoprecipitation.

Recent discoveries of novel EPOR pathways (Broxmeyer, 2013) and new EPOR agonists (Drüeke, 2013) have heightened interest in transducers of EPO’s effects. Our PY-phosphoproteomic analyses first reveal several novel candidate EPOR targets as molecular adaptors, S/T kinases, tyrosine phosphatases, ubiquitin factors, and cell cycle regulators (Supplemental Table S1). A focus on RHEX was prompted by its sharp EPO modulation, ORF novelty, conserved representation in H. sapiens and primates, and stage-specific up-modulation in developing erythroid cells. RHEX proved to be plasma membrane–associated, to coimmunoprecipitate with the EPOR, and to comprise a likely JAK2 PY target. EPO-regulated sites Y132 and/or Y141 were also implicated to bind GRB2. Via SOS plus RAS, GRB2 may mediate RHEX effects on ERK1/2 activity, but GRB2 itself is a versatile adaptor protein that can also couple to SHIP, SHP2, and up to 90 interacting proteins (Bisson et al., 2011). GRB2 can also preassemble with RTKs and repress basal signaling (Lin et al., 2012). Merit therefore exists for future investigations of implicated RHEX PY132, PY141, and GRB2 engaged pathways.

In primary EPCs, loss of function studies confirmed RHEX effects on proliferation but also indicated additional roles during normoblast development in that RHEX knockdown led to persistent erythroblastic features, high-level GPA expression, and decreased HBB transcript plus globin expression. Although this might relate indirectly to observed effects of RHEX on growth, effectors that reinforce EPC growth (e.g., via RAS; Zhang and Lodish, 2007; Blanc et al., 2012) often attenuate erythroid differentiation. Apparent effects of RHEX on normoblast development, therefore, may involve alternative mechanisms (and may point to novel EPO/EPOR effects on erythroid differentiation). RHEX-regulated routes that support erythroblast development, therefore, should also be of significant interest to define during myeloproliferative disease (Barbui et al., 2013) and the ineffective erythropoiesis of thalassemia (Rivella, 2012).

UT7epo EPC (Komatsu et al., 1993), 293, JURKAT, RAJI, K562, HL60, and HEL cell lines were maintained as previously described (Singh et al., 2012b). In cytokine withdrawal experiments, UT7epo cells were washed three times and cultured for 20 h in 0.2% BSA, 10 µg/ml holo-transferrin, 0.1 mM 2-mercaptoethanol, and IMDM (Singh et al., 2012b). Subsequent EPO (and cytokine) challenges were at the concentrations and time intervals indicated. Primary human hematopoietic cells used in RT-PCR analyses included monocytes, neutrophils, T cells, and platelets (AllCells). Murine G1E-ER4 cells (M. Weiss, Children’s Hospital of Philadelphia, Philadelphia, PA) were maintained in EPO plus mSCF as per Welch et al. (2004).

Primary EPCs were generated ex vivo from GCSF mobilized human CD34^pos cells. CD34^pos progenitors were cultured for 48 h in X-vivo10 media (Lonza) supplemented with 50 ng/ml each of rhSCF, rhIL3, rhFLT3, and rhTPO (PeproTech). For erythroid cell development, expanded progenitors were plated (at 10⁶ cells/ml) in StemSpan medium (STEMCELL Technologies) supplemented with 100 ng/ml rhSCF, 2 U/ml EPO, 100 µg/ml holo-transferrin, 1.5 µM β-estradiol, 0.5 µM dexamethasone, and 0.1 mM β-mercaptoethanol. On days 2, 6, and 10, cultures received 0.6 volume of medium. On days 4, 8, and 12, cultures were replated in fresh medium (10⁶ cells/ml). Erythroid development was assessed via flow cytometry, cytospin, and RT-PCR analyses.

In UT7epo cell PY-phosphoproteomic studies, cytokines were withdrawn for 20 h. Cells were then exposed to EPO (±4 U/ml for 15 min), and 9 M urea lysates were prepared and used to generate tryptic digests (Sathyanarayana et al., 2012). For duplicate samples, PY-phosphoproteomic LC-MS/MS analyses were then applied using a PhosphoScan approach (Stokes et al., 2012). Gene and transcript analyses used NCBI programs Gene, BLAST, Homologene, Gene Expression Omnibus, and Ensembl Genome browser. RHEX’s candidate structural features were assessed using BLAST Conserved Domain plus Cobalt (NCBI), ProSite (ExPASy), TMHMM Server (v 2.0), and TMpmed (chEMB Net).

For RT-PCR analyses, EPCs (including FACS or MACS isolated subpopulations) were lysed in TRIZOL reagent (Invitrogen), and RNA was prepared (RNeasy; QIAGEN). Reverse transcription (Superscriptase III; Invitrogen) and real-time quantitative polymerase chain reactions (iQ SYBR Green, i-Cycler; Bio-Rad Laboratories) were performed as previously described (Dev et al., 2013). The following primer pairs were from Invitrogen: hRHEX/C1ORF186, hALAS2, hHBB, and hBETA-ACTIN. Northern blotting was performed as detailed by Pircher et al. (2001) using a randomly primed full-length ³²P-RHEX cDNA probe.

Rabbit antibodies to RHEX were generated by immunizations with KLH-coupled peptides and immunogen boosts. Responses were evaluated by ELISA (phosphorylated and nonphosphorylated peptides) and by Western blotting (UT7epo and HL60 cell lysates). Additional antibodies were to (P)ERK1,2, GRB2, β-TUBULIN (Cell Signaling Technology), Hemoglobin (Santa Cruz Biotechnology, Inc.), and the EPOR (Singh et al., 2012b). Western blotting was as described previously (Singh et al., 2012b). In chemiluminescence, HRP-conjugated antibodies (Jackson ImmunoResearch Laboratories, Inc.) and Super-Signal West-Dura reagent were used. In immunoprecipitations, 0.4% Igepal lysates were prepared as per Singh et al. (2012b). Antibodies used (4 µg per immunoprecipitation) were to a RHEX-Flag epitope (Sigma-Aldrich) and the EPOR (Singh et al., 2012b). Immune complexes were retrieved using protein A/G particles (Thermo Fisher Scientific). In analyses of RHEX PY phosphorylation, cell lysates were exposed to Lambda phosphatase before SDS denaturation and Western blotting. In JAK2 and SRC inhibitor studies, UT7epo cells were cultured for 20 h with TG101348 or Dasatinib at the indicated concentrations (0.5% DMSO, solvent).

UT7epo cells were stained with FITC–anti-CD71, collected onto poly-L-lysine slides, exposed to 4% formaldehyde in PBS (20 min, 23°C), and then to 90% methanol (10 min, −20°C). Anti-(PY)RHEX antibody was used at 4 µg/ml (4°C, 60 min) and was detected using an Alexa Fluor 647 anti–rabbit second antibody. Alternatively, surface staining was with 5 µg/ml AF555-WGA for 10 min at 37°C (Life Technologies). Nuclear staining was with 1 µM ToPro-3 iodide for 20 min (Invitrogen). Preparations were washed three times in PBS and coverslipped (1.5 mm, Corning) in VECTASHIELD (Vector Laboratories). Images were acquired and analyzed using a confocal microscope system (SP1; Leica; Dev et al., 2013).

shRNAs (as designed with SBI) were cloned to a pGreenPuro vector modified to include an EF1α promoter for GFP plus puromycin resistance cDNA expression. shRNA sequences were as follows: shRNA-RHEX-UTR-#1, 5′-GATCCGGAAGAACTTTCAGGTAAACTTCCTGTCAGCCTTCTTGAAAGTCCATTTTTTTTGAATT-3′; shRNA-RHEX-UTR-#2, 5′-GATCCGGAGGTAAAGTATGAGAACTACTTCCTGTCAGTCCATTTCATACTCTTGATTTTTTGAATT-3′; shRNA-RHEX-UTR-#3, 5′-GATCCGGAGGAAGAACTTTCAGGTAACTTCCTGTCAGTCCTTCTTGAAAGTCCATTTTTTTGAATT-3′; shRNA-NT, 5′-GATCCCTAAGGTTAAGTCGCCCTCGCTCTAGCGAGGGCGACTTAACCTTAGGTTTTTGAATT-3′. Packaged lentiviruses (prepared using SBI-NT 293 cells) were recovered at 50 h after transduction, 0.4 µm filtered, concentrated (PEG-IT reagent, SBI), titered, aliquoted, and stored at −80°C. UT7epo and G1E-ER4 cell transductions involved plating cells at 3 × 10⁵ cells/ml, and at 20 h replating to 12-well plates at 2 × 10⁵ cells/ml (0.4 ml/well). At 4 h of culture, 4 µg/ml polybrene was added, followed by 40 µl shRNA-encoding lentivirus in IMDM (at defined MOIs). At 20 h after transduction, 0.5 ml of culture medium was added. At 40 h after transduction, 0.5 ml of cells were selected in 2.5 µg/ml puromycin or plated in methylcellulose for colony forming assays (see below).

In the lentiviral knockdown of RHEX in primary human EPCs, pre-expanded cells were transduced with pGreenPuro lentiviruses encoding RHEX-targeting or nontargeting (NT) shRNAs. Specifically, nontreated 12-well tissue culture plates (BD), were coated with 20 µg/ml Retronectin (Takara Bio Inc.), washed, and used to preadsorb lentiviruses at 37°C for 1.5 h. This was followed by centrifugation at 800 g for 25 min at 4°C. Pre-expanded hCD34^pos cells were plated at 0.5 × 10⁶ cells/ml in Xvivo 10 medium (as described above) and were transduced using 3 µg/ml polybrene (30 min incubation at 37°C) and spinoculation for 30 min at 400 g at 25°C. After culture for 20 h, transduced cells were then plated in StemSpan medium (STEMCELL Technologies) supplemented with 100 ng/ml rhSCF, 100 µg/ml holo-transferrin, 2 U/ml EPO, 1.5 µM β-estradiol, 0.5 µM dexamethasone, and 0.1 mM β-mercaptoethanol. Puromycin was included at 2.5 µg/ml. At the indicated time points, erythroid cell formation was assayed via flow cytometry (c-KIT, TFR1, GPA), cytospin morphologies, and RT-PCR. Viable cell counts (ViCell system; Beckman Coulter) and YOPRO3 viability analyses (Molecular Probes) were also performed. Lentiviral expression vectors also were prepared including those expressing RHEX-flag (EF1α) plus puromycin resistance cDNAs (PGK), and RHEX plus GFP cDNAs. Each template, together with empty vector negative control templates, was packaged as above and was titered using NIH3T3 cells.

Flow cytometry analyses of erythroid development used APC–anti-CD117, FITC–anti-CD71, and PE–anti-CD235a antibodies (BD), a FACSCalibur cytometer (BD), and CellQuest Pro software. In colony-forming unit assays, 0.5 × 10⁴ transduced UT7epo cells were plated in MethoCult H4434 medium (STEMCELL Technologies) supplemented with 3 U/ml EPO and 4.0 µg/ml puromycin. Colonies were analyzed using a STEMvision system (STEMCELL Technologies), differential interference fluorescence microscopy (DMI6000 B; Leica), and EVOS fluorescence microscopy. Cytospin analyses (10⁵ cells) involved poly-L-lysine slide centrifugation for 15 min at 100 g (Hettich Universal-16A cyto-centrifuge) and May-Grunwald Giemsa staining (Dev et al., 2013).

Table S1 shows EPO/EPOR PY-regulated targets as interrogated in UT7epo cells via PhosphoScan LC-MS/MS PY-proteomics.

Phosphoproteomic studies were advanced collaboratively via Phosphoscan LC-MS/MS with Cell Signaling Technology, Inc. Expert technical contributions of research staff (D.M. Wojchowski laboratory) are also gratefully acknowledged for human CD34 cell culture (R. Asch), RHEX antibody characterization (A. Pradeep), lentivirus construction (A. Johnson), and hEPC CD marker analyses plus FACS (E. Jachimowicz and N. Rainville). For RNA-Seq analyses of RHEX and ALAS2 in CD34^pos progenitors and CFUe, the authors thank Drs. M. Narla and X. An (New York Blood Center), as well as Dr. M. Weiss (Children’s Hospital of Philadelphia) for the generous provision of G1E-ER4 cells.

Primary support was provided by NIH R01-HL44491 (D.M. Wojchowski), with additional partial support via NIH R01 DK89439 (D.M. Wojchowski). Core facility contributions (flow cytometry, quantitative RT-PCR via Progenitor Cell Analysis and Molecular Phenotyping Cores) were supported in part via NIH/NIGMS P30GM106391 (D.M. Wojchowski). Confocal microscopy was supported in part by the Cell Imaging Core facility of COBRE Center Award P30GM103392. Supporting contributions by Affymax, Inc. and Takeda Global Research and Development Center, Inc. (Deerfield, IL) also are gratefully acknowledged.

The authors declare no competing financial interests.