Attenuation of EPO-dependent erythroblast formation by death-associated protein kinase-2

Erythroid progenitor cell survival is regulated by unique networked mechanisms. As cell-extrinsic factors, FASL and TRAIL can induce apoptosis, while erythropoietin (EPO) provides essential survival signals via JAK2/STAT5, PI3K/AKT, and RAS/MEK/ERK1,2 routes.¹  Cell-intrinsic regulators include BCL-X and NIX, plus GATA1 as a caspase-3 target.^2-4  To define new potential key survival-regulating factors, we presently profiled differentially staged primary murine splenic erythroblasts. One discovered erythroid-predominant factor was death-associated protein kinase-2 (DAPK2). Among 3 DAPK serine/threonine kinases,^5,6  DAPK1 first was identified as an IFN-γ-induced factor which facilitates cell death initiated by IFN-γ, TNF-α, FAS, or oncogene expression.⁷  DAPK1 possesses a DAP kinase upper-lobe signature, CaM regulatory domain, ankyrin repeats, and a C-terminal death domain.^5,6  ZIPK (zipper-interacting protein kinase/DAPK3/DLK) lacks a CaM domain, but possesses leucine zipper and nuclear localization motifs.^5,6,8  DAPK2 retains a related CaM domain, but possesses a unique C-terminal region.^5,6,9  Such structural differences further suggest that DAPK1, ZIP-K and DAPK2 likely play unique biologic roles.

DAPK1 can exert proapoptotic effects, potentially via caspase-independent type-II mechanisms.^5-7  Gene disruption experiments further place DAPK1 upstream of p53,⁹  and decreased DAPK1 expression is linked to multiple myeloma, ALL, and colon, breast and lung cancers.¹⁰  For DAPK2, overexpression studies in cell lines also point to potential proapoptotic effects,^11,12  but to date this has not been investigated in primary cells which normally express, and regulate endogenous DAPK2. Via transcriptome analyses, transgenic mouse experiments, and analyses in EPO-dependent erythroid progenitor cells, we now demonstrate predominant DAPK2 expression in erythroid progenitors; characterize DAPK2's proapoptotic capacity to sharply limit erythropoiesis during hemolytic anemia; and initially associate DAPK2's activity to EPO's actions as a candidate negative feedback factor.

pA2gata1-EE-T-Y343 mice expressing an “EE” tag¹³  were treated with thiamphenicol for 5 days.¹⁴  At 80, 100, or 120 hours after withdrawal, splenocytes were isolated, incubated in 50 U/mL DNase-I, 0.5 U/mL dispase-I, and purified by magnetic-activated cell sorting (MACS).¹³  For DAPK2 transgenics, a flag-hDAPK2 cDNA was cloned to Gata1-IE3.9int,¹⁵  and injected (8 kb Sal-I fragment) into FVB pronuclei. Phenylhydrazine dosing was at 52.5 mg/kg. Blood cell counts were via an ADVIA-120 (Bayer, Tarrytown, NY). Hematocrits and reticulocytes also were assayed by microcentrifugation and thiazole-orange staining.¹⁶  For mouse models used herein, all studies and protocols received Institutional Animal Care and Use Committee review and approval from all participating institutions.

RNA was isolated, and used to prepare biotin-cRNA.¹⁷  Hybridizations were to MG-U74Av2 arrays, and were analyzed using Genechip 5.1 software. Reverse-transcription (RT) and quantitative polymerase chain reaction (qPCR) were as described.¹⁸ 

Pre/pro-B cells, granulocytes/monocytes, and mast cells were expanded in BIT9500 medium plus recombinant cytokines as indicated. Erythroblasts were expanded in SP34-EX medium.^17-19  NIH-3T3, OP9, G1E/JC4, and EML, UT7epo cells were cultured as detailed in legends.

Flow cytometric analyses for KIT, CD71, Ter119, EE-T-Y343, and annexin-V were as described.^17-19  Tissues were ground in liquid nitrogen (LN2), and homogenized in an Igepal lysis buffer.^17-19  These samples, and cultured cell extracts, then were prepared for Western blotting as previously described.^17-19 

(Flag)DAPK2 and siRNAs were expressed using CAD G Whiz (CGW) vectors.^20,21  Lentiviruses were prepared in 293-FT cells, and concentrated. Transductions used polybrene (8 μg/mL) and limiting multiplicity of infection (MOIs). GFP^pos cells were isolated by fluorescence-activated cell sorting (FACS).

DAPK2 activity was assayed after immunoprecipitation using myosin light chain (MLC).²²  Phospho-Ser318-DAPK2 was assayed via Western blotting.

(Pro)erythroblasts at 3 developmental stages first were prepared, and profiled. Specifically, thiamphenicol was used to limit BFUe formation in pA2gata1-EE-T-Y343 mice.^13,14  Upon thiamphenicol withdrawal, splenic erythropoiesis initiated synchronously and at 80, 100 and 120 hours, Kit^highCD71⁺Ter119⁻; Kit^lowCD71⁺Ter119⁺; and Kit⁻CD71⁺Ter119^high cells (stages I, II and III) were generated at high frequencies (Figure 1A). This approach also enabled efficient erythroblast purification via MACS (Figure 1A and Figure S1A, available on the Blood website; see the Supplemental Materials link at the top of the online article).

Stage I-III erythroblasts next were analyzed for stage-to-stage transcriptome modulations. Robust profiling outcomes (Figure S1B) allowed for functional categorizations. For 2 broad sets, transcription plus chromatin factors and signal transduction plus cell cycle factors, profiles are presented in Figures S2, S3, and serve to further characterize erythroblast stagedness.

Attention next focused on candidate (anti)apoptotic factors. Twelve genes encoding (anti)apoptotic factors were stage-modulated (Figure 1B). At stage II, 6 were up-modulated, including Bcl-xL, Nix, Pim1, p85alpha, Gata1, and Dapk2. In contrast, Bax, Bak, Caspase-3, Aif, Ak-2 and Vdac were down-modulated 2- to 4-fold. With the exception of Dapk2, these factors have been studied previously in primary cell systems and mouse models. Dapk2 therefore was investigated further. RT-qPCR first confirmed high-level expression in maturing splenic erythroblasts, and this also was observed in bone marrow erythroblasts (Figure 1C,D). DAPK2 levels in early, mid and late stage erythroblasts from both the spleens of phenylhydrazine-treated mice, and bone marrow also were analyzed via Western blotting, with similar results. Dapk2 and Dapk1 levels were further analyzed in several hematopoietic lineages, and tissues. As analyzed among hematopoietic cells, Dapk2's expression was selectively high in bone marrow and in late-stage erythroblasts. Dapk1 levels, in contrast, were relatively low in all hematopoietic lineages assayed (Figure 1D). These results, and additional supporting Western blot experiments (Figure S4), suggested possible erythroid-predominant roles for DAPK2.

In a transgenic approach, a Gata1-IE3.9int vector¹⁵  next was used to reinforce DAPK2 expression (Figure 2A). In Gata1-IE3.9int-DAPK2 mice, peripheral blood cell levels for lymphoid, myeloid, granulocytic lineages were normal (data not shown), and only modest effects on reticulocytes (approximately 1.03% ± 0.30% increase) and hematocrits (approximately 2.33 ± 0.52 point decrease) were observed (Figure S5A). Splenic CD71^highTer119⁺ erythroblasts also were somewhat elevated as 2.75% plus or minus 0.15% of total splenocytes in Gata1-IE3.9int-DAPK2 mice versus 1.25% plus or minus 0.05% in wild-type littermates (Figure S5B). After phenylhydrazine dosing, however, Gata1-IE3.9int-DAPK2 mice exhibited severe anemia. In controls, hematocrits fell to approximately 21.1%, and recovered by approximately day 8. In Gata1-IE3.9int-DAPK2 mice, hematocrits fell to a nadir of approximately 14.3%, and remained comparably low through day 8 (Figure 2B). Furthermore, 30% of Gata1-IE3.9int-DAPK2 mice failed to reverse anemia and died by day 5 (Figure S5C). EPO responses also were examined. In controls, EPO (1000 U/kg) increased hematocrits by 5.6 points. In Gata1-IE3.9int-DAPK2 mice, this response was limited to a 1.7 point increase (ie, 31.3% of control response values; Figure 2C). Responses to 5FU-induced anemia also were tested at a myeloablative dose (150 mg/kg, IP). At days 5 through 15, Gata1-IE3.9int-DAPK2 mice again exhibited more severe anemia (P = .001 to .078 for mean hematocrits) as well as somewhat delayed recovery (see Figure S5D).

To examine possible stage-specific effects, an ex vivo expansion system was used to analyze bone marrow erythroblast development. Gata1-IE3.9int-DAPK2 Kit⁺CD71^highTer119⁻ (pro)erythroblasts formed at near wild-type frequencies (data not shown). Kit⁻CD71^highTer119⁻ progeny, however, were somewhat under-represented. At a Kit⁻CD71^highTer119⁺ stage, this deficit was even more marked (Figure 2D). Results were reproducible, as also assessed during days 1 to 4 of culture (Figure S6D). Gata1-IE3.9int-DAPK2 erythroblast survival potentials also were studied. Here, Kit⁺CD71^highTer119⁻ erythroblasts were isolated at day 3, and replated in EPO at limiting doses. At 16 hours, annexin V⁺ cells were assayed. In Gata1-IE3.9int-DAPK2 erythroblasts, apoptosis occurred at increased frequencies, especially among Kit⁻CD71^highTer119⁻ cells (Figure 2E and Figure S6A-C).

In UT7epo cells, effects of siRNA-mediated DAPK2 knock-down on erythroid progenitor survival were studied using a DAPK2 siRNA lentivirus. UT7epo-siDAPK2 and control UT7epo-siLuc cells were prepared, and were plated in EPO at a limiting dose (0.2 U/mL). At the indicated intervals, frequencies of viable cells were determined. The knock-down of DAPK2 afforded substantial survival advantages (up to 42.8% over controls; Figure 2F). The observed comparably extended time-course for UT7epo cell death is attributed to the inclusion of FBS, and UT7epo's nature as an EPO-dependent yet permanent cell line.^23-25 

Possible EPO modulation of DAPK2 also was investigated. UT7epo-DAPK2 cells were cultured in the absence of cytokines, and EPO-exposed. (Flag)DAPK2 then was immunoprecipitated and assayed using an in vitro kinase assay. EPO-exposure stimulated DAPK2 activity approximately 200% (Figure 2G). DAPK2 Ser318 phosphorylation also was analyzed, and EPO was observed to also limit this inhibitory event (Figure 2G). EPO is known to stimulate several inhibitory factors in negative feedback routes including CIS,²⁶  SOCS3²⁷  and a β-Trcp ubiquitin ligase.²³  Apparent EPO stimulation of DAPK2 at least suggests a similar negative feedback role. Gata1-IE3.9int-DAPK2 (pro)erythroblasts, in addition, appeared to be attenuated in their transition from Kit⁻CD71^highTer119⁻ to Kit⁻CD71^highTer119⁺ stages (see Figure 2D and Figure S6D). This finding raises the interesting additional possibility that DAPK2 also may exert regulatory effects on proerythroblast development.

Several findings indicate that DAPK2 may enforce significant negative regulation over (pro)erythroblast formation. (1) DAPK2's expression among hematopoietic cells is predominantly erythroid, and up-modulated during erythroblast development. (2) Gata1-IE3.9int-enforced DAPK2 expression compromises (pro)erythroblast survival and production, especially during anemia. This at least suggests anemia and/or stress-specific activation of DAPK2, and DAPKs interestingly can be up-modulated by hypoxia.²⁸  (3) DAPK2's proapoptotic effects were exerted in a stage-selective fashion within EPO-dependent Kit⁻CD71^highTer119⁻ erythroblasts. (4) As analyzed in primary erythroblasts ex vivo, DAPK2 also may act to attenuate proerythro-blast development specifically at a Kit⁻CD71^highTer119⁻ to Kit⁻CD71^highTer119⁺ transition. (5) In addition, DAPK2 may be EPO-modulated, and its knock-down enhances EPO-dependent erythroid progenitor cell survival. DAPK2 therefore comprises a new candidate attenuator of erythropoiesis which may network with EPO and BCL-XL/NIX pathways to control survival potentials, possibly via caspase-independent type II mechanisms.^5-7,29  The nature of anemia- and/or stress-induced pathways that engage DAPK2 under these conditions, but not at steady state, also will be of significant interest to discover in future, and ongoing investigations.

This work was supported by National Institutes of Health (NIH) grant R01HL44491. Core facility support also was provided via NIH P20 RR18789.

National Institutes of Health

Contribution: J.F., D.Z., and M.M. performed the bulk of thiamphenicol-based splenocyte preparations and analyses, including transcriptome profiling (together with L.O. and D.W.); D.W. and E.H. worked with J.F. to prepare transgenic expression constructs and mice; E.H., D.W., and L.O. contributed to investigations using ex vivo expansions and flow cytometry; M.T. and B.T. contributed to the design, construction, and preparation of high titer lentiviral vectors. All authors contributed in meaningful ways to data interpretation, figure construction. and manuscript writing and preparation.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Don M. Wojchowski, Director, Stem and Progenitor Cell Biology Program, Maine Medical Center Research Institute, 81 Research Drive, Scarborough, ME 04074; e-mail: wojchd@mmc.org.