Abstract

   The Ser-Thr kinase CK2 plays important roles in sustaining cell
   survival and resistance to stress and these functions are exploited by
   different types of blood tumors. Yet, the physiological involvement of
   CK2 in normal blood cell development is poorly known. Here, we
   discovered that the β regulatory subunit of CK2 is critical for normal
   hematopoiesis in the mouse. Fetal livers of conditional CK2β knockout
   embryos showed increased numbers of hematopoietic stem cells associated
   to a higher proliferation rate compared to control animals. Both
   hematopoietic stem and progenitor cells (HSPCs) displayed alterations
   in the expression of transcription factors involved in cell quiescence,
   self-renewal, and lineage commitment. HSPCs lacking CK2β were
   functionally impaired in supporting both in vitro and in vivo
   hematopoiesis as demonstrated by transplantation assays. Furthermore,
   KO mice developed anemia due to a reduced number of mature erythroid
   cells. This compartment was characterized by dysplasia, proliferative
   defects at early precursor stage, and apoptosis at late-stage
   erythroblasts. Erythroid cells exhibited a marked compromise of
   signaling cascades downstream of the cKit and erythropoietin receptor,
   with a defective activation of ERK/JNK, JAK/STAT5, and PI3K/AKT
   pathways and perturbations of several transcriptional programs as
   demonstrated by RNA-Seq analysis. Moreover, we unraveled an unforeseen
   molecular mechanism whereby CK2 sustains GATA1 stability and
   transcriptional proficiency. Thus, our work demonstrates new and
   crucial functions of CK2 in HSPC biology and in erythropoiesis.
     __________________________________________________________________

   [68]graphic file with name hs9-7-e978-g001.jpg
   [69]Open in a new tab

INTRODUCTION

   Hematopoiesis is a finely tuned developmental process regulated by
   cell-intrinsic (epigenetic, gene transcription) and extrinsic
   (cytokines, chemokines, and growth factors)
   mechanisms/molecules.^[70]1,[71]2 Signals generated by Stem cell factor
   (Scf)-cKit axis^[72]3 and Erythropoietin-Receptor (EPO-R)^[73]4 are
   fundamental for hematopoietic stem/progenitor cell (HSPC) growth and
   for red blood cell (RBC) production, respectively, through the
   activation of shared pathways, such as ERK/MAP kinases, JAK/STAT and
   PI3K/AKT signaling cascades. The EPO-R signaling culminates in the
   activation of the transcription factor GATA1^[74]4–[75]6 which sustains
   erythroid cell viability,^[76]6,[77]7 the switching from
   embryonic/fetal to adult globin chains,^[78]8 heme biosynthesis^[79]9
   and supports also its own and Epo-R^[80]10 transcription.

   Protein kinase CK2 is a Ser/Thr kinase constituted by a tetramer of 2
   catalytic (α or α’)^[81]11 and 2 regulatory (β) subunits, with CK2β
   being essential for the assembly of the kinase in a tetramer and for
   the selection of phospho-targets.^[82]12 However, the α and β subunits
   exert also functions independently from the holoenzyme^[83]12 and,
   importantly, free β dimers can bind and stabilize other
   proteins.^[84]13 CK2 is involved in cell proliferation, resistance to
   stress, as well as many other processes.^[85]14 One main property of
   CK2 is to sustain cell viability, hampering apoptosis by inhibiting
   procaspase-2 and caspase 8 and regulating BCL-2 family members.^[86]15
   In particular, CK2 prevents the activation of pro-apoptotic BID and
   enhances the expression of anti-apoptotic BCL-XL.^[87]16 In addition,
   this kinase positively regulates survival pathways, such as NF-κB (in
   which it activates p65/RelA through phosphorylation in S529^[88]17 and
   enhances IκBα^[89]18 degradation and PI3K/AKT (through the inhibition
   of the phosphatase PTEN^[90]19 and the activation of
   AKT^[91]20,[92]21). We and others have demonstrated that CK2 is
   essential for acute myeloid leukemia (AML) and AML-derived leukemic
   stem cells,^[93]22,[94]23 multiple myeloma^[95]24–[96]26 and
   non-Hodgkin B cell lymphoma^[97]27–[98]30 cell survival, proliferation,
   and resistance to several intrinsic and extrinsic stresses.^[99]31

   Since literature has also widely reported the key role of CK2 in
   embryogenesis,^[100]32–[101]34 we hypothesized that this kinase might
   also be essential for the ontogenesis of blood cells. In a recent
   article published by our group, we have demonstrated the importance of
   CK2 in B cell commitment towards follicular fate and in the germinal
   center reaction using mice bearing loxP-flanked CK2β alleles and
   expressing the Cre recombinase under the control of the CD19
   promoter.^[102]35 In the present work, we crossed the same model of
   loxP mice with animals expressing the Cre recombinase under the control
   of the Vav1 promoter, thereby achieving CK2β KO in all the
   hematopoietic cells. We focused the analysis on the HSPC and erythroid
   compartments.

   CK2β KO during hematopoiesis dramatically affected mouse HSPC numbers
   and function ex vivo and in vivo. Mechanistically, we provided evidence
   that CK2β influences the expression of multiple transcription factors
   implicated in HSPC biology. Also, by exploiting an in vitro model of
   erythroid differentiation, we showed that CK2β is an essential
   stage-specific regulator of RBC differentiation, proliferation, and
   viability and we unraveled a molecular crosstalk between this kinase
   and the transcription factor GATA1.

METHODS

Mice

   C57BL/6 VAV1-CRE mice were purchased from Jackson Laboratory, C57BL/6
   Csnk2b 2loxP mice were described in,^[103]34 C57BL/6 LY5.1 (CD45.1)
   mice were purchased from Charles River. Mice were kept in a
   pathogen-free colony at the animal facility of the Veneto Institute of
   Molecular Medicine, Padova, Italy. The University of Padova
   organization in charge for animal wellness approved the experimentation
   and declared that it fulfilled the National and European rules for
   animal studies. All the experimental procedures were authorized by the
   Italian Ministry of Health (Protocol N° 126-2015 and 205/2022-PR). For
   details on mice breeding, genotyping, and sample processing, see
   [104]Suppl. Materials and Methods.

Repopulation assay

   In competitive transplantation, CD45.1 mice were used both as recipient
   and competitor cells. Females, in synchronized gestations, were
   sacrificed at 14.5dpc; mixtures of equal amounts (1x10^6 cells) of
   donor (either Vav1^+/Cre control or Csnk2b^fl/fl/Vav1^+/Cre KO from
   CD45.2^+ genetic background) and wild-type competitor (CD45.1^+) total
   fetal liver cells were intravenously injected in lethally γ-irradiated
   (900rad) CD45.1^+ recipient mice. In non-competitive transplantation
   assay, CD45.1 recipient mice were injected only with (1 × 10^6 cells)
   CD45.2^+ cells from either Vav1^+/Cre control or
   Csnk2b^fl/fl/Vav1^+/Cre KO. Engraftment was assessed at 13/14 days for
   both experimental assays and after 1 month for competitive experiments
   in PB, BM, Spleen, Thymus, and Lymph Nodes. Repopulation efficiency of
   donor and competitor populations was investigated through fluorescence
   activated cell sorting (FACS) analysis after staining of PB or smashed
   hematopoietic tissues with Fc Block, anti-CD45.1, and anti-CD45.2
   antibodies followed by erythrocyte cell lysis.

Cell culture and chemicals

   G1E-ER cells, a kind gift of Prof. M. Weiss (St. Jude Childrens’
   Hospital, Memphis, TN) were maintained as described in.^[105]36
   Erythropoietin and 1-Thioglycerol were purchased from Sigma (Mo), SCF
   from Peprotec (UK). For cell treatment, β-estradiol (Merck) was used at
   0.2 µM, CX-4945 (Activate Scientific GmbH) at 2.5 µM, Z-VAD-FMK
   (SelleckChem) caspase inhibitor at 5 µM, proteasome inhibitor
   bortezomib (SelleckChem) at 5 nM and cycloheximide (Merck) at 0.4
   µg/ml.

RNA-Seq

   The sequencing was performed on TER119^+ cells purified from 14.5 dpc
   fetal livers by FACS sorting. The complete RNA-Seq data are available
   in the Gene Expression Omnibus (GEO)
   ([106]http://www.ncbi.nlm.nih.gov/geo) under accession number
   [107]GSE158665. For the complete list of differently expressed genes
   see [108]Suppl. Tables S1 and [109]S2. For further details on the
   analysis methods, see [110]SDC file.

Statistical analysis

   Data were analyzed with the 2-tailed unpaired Student’s t-test. Values
   were considered statistically significant at P values below *0.05,
   **0.01, ***0.001.

   For further Materials and Methods, see the [111]Supplemental Digital
   Content file.

Data availability

   All data required to produce the results of this study are provided
   within the article, via supplemental files or with link to the dataset
   GEO with the accession number. Also included in supplemental file
   [112]Suppl. Tables S3 - S7 reporting in order: list of primers for
   genotyping, qRT-PCR primers, antibodies for WB and flow cytometry,
   hemochrome of adult mice.

RESULTS

CK2β loss in the hematopoietic compartment is embryonic lethal and alters
hematopoietic cell number, viability, and proliferation

   We generated conditional CK2β-deficient mice in Vav1-expressing cells.
   In utero lethality between 19.5 and 20.5 days post conception (dpc) was
   observed in 100% of CK2β KO embryos (Csnk2b^fl/fl;Vav1^+/Cre). Control
   mice (Csnk2b^+/+;Vav1^+/Cre) and CK2β heterozygous mice
   (Csnk2b^+/fl;Vav1^+/Cre) were viable and fertile. We therefore analyzed
   fetuses at 12.5, 14.5, and 17.5 dpc. CK2β KO fetuses were pale and
   hemorrhagic, in some cases with the appearance of hydrops fetalis, more
   evident at 17.5 dpc (Figure [113]1A). Analysis of CK2β mRNA (Figure
   [114]1B) and protein levels (Figure [115]1C) in the fetal liver,
   primary site of hematopoiesis at this stage,^[116]37 and in the brain
   from 14.5 dpc fetuses showed complete Csnk2b deletion restricted to the
   hematopoietic tissue. No significant changes were observed in CK2α
   protein levels (Figure [117]1C). CK2β-null embryos displayed smaller
   fetal livers with reduced cellularity as compared to controls (Figure
   [118]1D). FACS analysis of the CD45^+ (leukocyte) and Ter119^+
   (erythroid) markers showed a reduction in the percentage of erythroid
   cells in the CK2β KO mice, while the percentage of leukocytes appeared
   slightly increased as compared to control littermates (Figure [119]1E).
   A deepened analysis of the white cell compartment revealed an
   accumulation in percentages of granulocytes and monocytes/macrophages
   in CK2β KO embryos (Figure [120]1F and [121]1G, left panels) that could
   account for the increased frequency of CD45^+ cells; however, the
   absolute counts of Mac1^+Gr1^+ and Mac^+F4/80^+ populations in total
   fetal livers resulted to be reduced in the KO samples compared to the
   control ones (Figure [122]1F and [123]1G, right panels). Moreover, also
   the B cell compartment was altered in CK2β KO mice with accumulation in
   percentages of immature B cells (B220^+ CD19^-) (Figure [124]1H, left
   panel) and expansion of pre/pro B cells (Figure [125]1I, left panel).
   Also in this case the counts in the total fetal livers show
   discrepancies compared to the frequencies due to the reduced global
   cellularity of livers in KO embryos; indeed, there were no significant
   changes in B220^+ CD19^- absolute numbers (Figure [126]1H, right panel)
   and there was a decrease of pre/pro cells in KO embryos (Figure
   [127]1I, right panel). Annexin V (AnnV) and 7-Aminoactinomycin D
   (7-AAD) staining revealed a higher percentage of apoptotic cells in
   CK2β KO fetal livers that became more evident at 17.5 dpc (Figure
   [128]1J). However, analysis of Ki-67 expression showed a mild increase
   in cell proliferation in CK2β KO compared to control mice (Figure
   [129]1K).

Figure 1.

   [130]Figure 1.
   [131]Open in a new tab

   KO of CK2β in the fetal hematopoietic system results in embryonic
   lethality with reduction in fetal liver cellularity and alteration in
   cell viability and proliferation. (A) Pictures of
   Csnk2b^fl/fl/Vav1^+/Cre (KO) and Csnk2b^+/+/Vav1^+/Cre (control)
   littermate fetuses at 14.5 and 17.5 days post conception (dpc). (B)
   Csnk2b mRNA levels evaluated by qRT-PCR at 14.5 dpc in control and KO
   fetal livers. Data are representative of 5 samples for each genotype.
   Gapdh was used as reference gene. (C) WB analysis of CK2β and CK2α
   protein expression at 14.5 dpc in fetal liver cells compared to
   non-hematopoietic tissue (brain); upper panel: WB image representative
   of 4 samples for each genotype; lower panel: densitometric analysis;
   GAPDH was used as loading control. (D) Upper level: representative
   image of control and KO fetal livers at 17.5 dpc; lower level: fetal
   liver cell count at different development time points (12.5 dpc:
   controls n = 9, KO n = 6; 14.5 dpc controls n = 17, KO n = 18; 17.5
   dpc: controls n = 18, KO n = 21). (E) FACS analysis of the white
   population (CD45^+) and erythroid cells (Ter119^+) in fetal livers at
   17.5 dpc: on the left representative contour plot, on the right
   histograms displaying the percentages of the 2 hematopoietic
   compartments (controls n = 8; KO n = 3). (F) Scatter plots representing
   percentages of cells (on the left; controls n = 11, KO n = 3) or
   absolute counts (on the right; controls n = 10, KO n = 6) after Gr1 and
   Mac1 staining and FACS analysis in 17.5dpc fetal livers. (G) FACS
   analysis of macrophages (Mac1^+F4/80^+) in 17.5 dpc fetal livers
   represented as percentage on the left (controls n = 3, KO n = 3) and as
   absolute counts on the right (controls n = 9, KO n = 6). (H) Graphs
   indicating the percentages (on the left) (controls n = 15, KO n = 6)
   and absolute counts (on the right) (controls n = 10, KO n = 6) of B
   cells in 17.5dpc fetal livers. (I) FACS analysis of precursors B cells
   in 17.5 dpc fetal livers: graphs indicating the percentages (on the
   left) and absolute counts (on the right) (controls n = 8, KO n = 3).
   (J) Evaluation of apoptosis at 14.5 and 17.5 dpc by FACS analysis,
   after staining with Annexin V (AnnV) and 7-AAD, performed on total
   liver population (at 14.5 dpc controls n = 4, KO n = 5; at 17.5 dpc
   controls n = 20, KO n = 6). (K) Analysis of Ki-67 expression in 14.5dpc
   total liver cells (controls n = 5, KO n = 3). In (C–K), graphs show
   means ± SD; Student’s t-test: *P < 0.05; **P < 0.01; ***P < 0.001. FACS
   = fluorescence activated cell sorting; KO = knockout; WB = western
   blot.

Deletion of CK2β affects hematopoietic stem and progenitor cell proliferation
kinetics and transcriptional programs

   Next, we analyzed the immature hematopoietic compartment represented by
   stem (HSC) and progenitor cells (HPCs) evaluating the effects of Csnk2b
   ablation. FACS analysis of HSCs^[132]38 and HPCs^[133]39 revealed a
   significant increase in the percentages of HSCs (Lin^- cKit^+ Sca-1^+
   CD150^+ CD48^-) (Figure [134]2A and [135]2D), MEPS (Lin^- cKit^+
   Sca-1^- CD34^- FcyRIII/II^-), CMPs (Lin^- cKit^+ Sca-1^- CD34^+
   FcyRIII/II^low) and GMPs (Lin^- cKit^+ Sca-1^- CD34^+ FcyRIII/II^high)
   (Figure [136]2B and [137]2D) in KO fetal livers compared to controls
   with no significant changes in CLPs (Lin^- IL-7R^+ cKit^+ Sca-1^-/low)
   (Figure [138]2C and [139]2D). However, analysis of absolute counts in
   the total fetal livers confirmed in the KO samples a significant
   expansion limited to the LSK compartment and a trend to higher counts
   for GMPs (Figure [140]2E). Notably, the intensity of cKit expression, a
   receptor essential for HSC and precursor survival and self-renewal, was
   found substantially reduced (mean fluorescence intensity, MFI) in CK2β
   KO HSPCs (Figure [141]2F).

Figure 2.

   [142]Figure 2.
   [143]Open in a new tab

   Absence of CK2β increases the amount of hematopoietic stem cells and
   precursors and perturbs their maturation program. Representative FACS
   dot plot of hematopoietic stem cells (HSCs) (A), common myeloid
   progenitors (CMPs), megakaryocyte/erythrocyte progenitors (MEPs),
   granulocyte/monocyte progenitors (GMPs) in 14.5 dpc fetal livers (B),
   common lymphoid progenitors (CLPs) (C); the representative dot plots
   show the percentage of cells; 1x10^6 and at least 4 × 10^5 events were
   acquired for LT-HSCs and the other progenitors respectively. (D)
   Histogram summarizing the percentages of each population (for HSC
   controls n = 5, KO n = 4; for the other precursors: controls n = 9, KO
   n = 6). (E) Graphs representing absolute cell counts of LSK, MEPs,
   CMPs, and GMPs populations referred to the total fetal liver (controls
   n = 3, KO n = 3). (F) Histograms exhibiting mean fluorescence intensity
   (MFI) in HSCs and each precursor sub-population (controls n = 9, KO n =
   6). (G) BrdU proliferation assay in 14.5 dpc fetal livers cells: cell
   cycle was analyzed inside the LSK (Lin^-Sca-1^+cKit^+) fraction after
   staining with anti-BrdU and 7-AAD. Left panel: representative contour
   plots and gating strategy; right panel: histograms summarizing the
   percentages of the cells in each stage (controls n = 4, KO n = 4). (H)
   qRT-PCR showing the relative mRNA expression of Scl/Tal1, Bmi-1, Runx1,
   Jun, Pu.1, Cebpa, Gata1, Gata2 genes in HSCs (controls n = 3, KO n =
   3), CMPs (controls n = 4, KO n = 5), GMPs (controls n = 4, KO n = 5)
   and MEPs (controls n = 4, KO n = 5) purified through sorting from
   14.5dpc fetal livers. Gapdh was used as reference gene. In all panels
   (D–H) the graphs report mean ± SD values; Student’s t-test: *P < 0.05;
   **P < 0.01; ***P < 0.001. CLPs = common lymphoid progenitors; CMPs =
   common myeloid progenitors; FACS = fluorescence activated cell sorting;
   GMPs = granulocyte/monocyte progenitors; HSCs = hematopoietic stem
   cells; KO = knockout; LSK = Lin-Kit+Sca1+; MEPs =
   megakaryocyte/erythrocyte progenitors.

   Since during blood cell development proliferation and differentiation
   must be tightly coordinated, we analyzed the cell cycle of LSK cells,
   using BrdU and 7-AAD staining and FACS analysis. As shown in Figure
   [144]2G, a higher percentage of cells was in G2/M, S, and sub-G[0]
   phases and displayed a greater DNA content (gate >2n) in KO mice
   compared to controls.

   We next inquired if CK2β loss could affect the expression of
   transcription factors important for HSC and HPC early lineage
   commitment. By qRT-PCR we evaluated the expression of specific mRNAs
   coding for transcription factors involved in cell maturation
   (Gata1,^[145]5,[146]40 Jun^[147]41) or in the maintenance of stem cell
   quiescence and self-renewal (Scl/Tal1,^[148]42 Bmi-1,^[149]43
   Runx1,^[150]44,[151]45 Pu.1,^[152]46 Cebpa,^[153]47 Gata2^[154]48), in
   FACS-sorted LSKs, CMPs, GMPs and MEPs (Figure [155]2H). In CK2β KO
   embryos, the LSK compartment exhibited a reduced expression of Scl/Tal1
   and Pu.1 with no significant changes in Bmi-1 and Runx1 mRNA levels
   compared to controls. Loss of CK2β caused a decrease of Runx.1, Pu.1
   and Cepba expression in CMPs and a decrease of Cebpa and Gata1 in GMPs.
   MEPs from KO mice showed alterations in almost all the genes analyzed
   with a significant reduction of Scl/Tal1, Bmi-1, Runx1, and Gata1 and
   an increase of Jun expression.

   In summary, lack of CK2β causes an accumulation and increased
   proliferation of HSCs and an altered expression of transcription
   factors involved in the regulation of HSPC quiescence and commitment.

CK2β is essential for hematopoietic reconstitution capability in vitro and in
vivo

   To assess the differentiative potential of progenitors, we performed
   methylcellulose-based colony-forming assays of fetal liver HSPCs.
   Strikingly, KO cells displayed an impairment to generate erythroid and
   myeloid colonies at 3 and 6 days ([156]Suppl. Figures S1A and S1B). In
   vivo competitive repopulation assays of 14.5 dpc fetal liver cell
   transplanted in C57BL/6 congenic mice expressing the differential
   leukocyte marker CD45.1 were then performed (Figure
   [157]3A).^[158]49,[159]50 As summarized in Table [160]1, an almost
   complete restoration of hematopoiesis was obtained as early as 2 weeks
   after transplantation with comparable blood cell percentages between
   mice injected with both kinds of cell mix. To note, FACS analysis of
   peripheral blood revealed an equal contribution to hematopoiesis when
   control CD45.2^+ and competitor CD45.1^+ donor cells were injected,
   while the injection of CK2β KO CD45.2^+ did not contribute to mature
   blood cells (Figure [161]3B). The same phenotype was observed after 4
   weeks in all the hemo-lymphopoietic organs (Figure [162]3B–[163]3D). We
   also evaluated the contribution to the generation of LSK and HSCs from
   CD45.1^+ and CD45.2^+ cell populations. Remarkably, in recipient mice
   injected with the competitive KO transplant there was a complete
   absence of CD45.2 expression both in LSK (Figure [164]3E) and in HSCs
   (CD48^-CD150^+ in LSK gated cells)^[165]51 (Figure [166]3F). Since the
   lack of engraftment in the BM could be linked to a problem of homing of
   HSCs, we evaluated by FACS the expression of CXCR4 (a known key
   mediator of HSC homing to the bone marrow^[167]52) in LSK compartment
   of control and CK2β KO fetal livers (Figure [168]3G). Unexpectedly,
   CK2β KO embryos showed higher percentages of CXCR4^+ LSK cells and
   higher MFI of this marker as compared to control counterpart,
   suggesting that the observed phenotype is likely not dependent on this
   molecule. To exclude the possibility that the lack of KO cell
   engraftment was due to the inability of CD45.2^+ CK2β null LSK to
   compete with CD45.1 cells for the BM niches, we also performed
   non-competitive transplantation assays. As described in [169]Suppl.
   Figure S1C, CD45.1 adult mice were irradiated and injected with only
   CD45.2 cells from control or KO fetal livers; 2 out of 5 mice injected
   with CK2β null CD45.2^+ cells died at day 12, thus we anticipated the
   experimental end point at 13 days. In all the compartments analyzed
   (SP, BM, and spleen) we evidenced almost the complete absence of CD45.2
   cells from KO samples ([170]Suppl. Figure S1D). Moreover, analysis of
   CD45.2^+ LSK frequencies and events in the BM ([171]Suppl. Figure S1E)
   underlined that CK2β KO progenitors and stem cells were unable to
   engraft and sustain hematopoiesis.

Figure 3.

   [172]Figure 3.
   [173]Open in a new tab

   Fetal liver cell transplantation from KO fetuses in recipient adult
   mice results in the lack of hematopoietic reconstitution. (A)
   Transplant workflow: a mixed population with equal proportions of
   CD45.2 (donor from control or KO) and CD45.1 (competitor) fetal liver
   cells were injected in recipient CD45.1 adult mice after lethal total
   body irradiation (900 rads). The transplantation with CD45.2 cells from
   control embryos plus CD45.1 competitor cells is indicated as Mix
   Control; when CD45.2 cells were obtained from CK2β-null embryos plus
   CD45.1 competitor cells, the injected population is called Mix KO.
   Hematopoietic restoration was assessed at 2 and 4 weeks from
   irradiation. Injection of PBS was used as control of irradiation
   efficacy. Sample size: 7 mice injected with mix control, 7 with mix KO,
   and 2 with PBS. (B–F) Flow cytometry analysis of CD45.2 and CD45.1
   relative amounts in (B) peripheral blood (PB), (C) bone marrow (BM),
   spleen (SP), and (D) lymph nodes and thymus, (E) inside the gate of LSK
   and (F) inside long term HSCs. On the left of each figure are depicted
   the representative dot plots and on the right are the histograms with
   CD45.1 and CD45.2; (G) Flow cytometry analysis of CXCR4 expression in
   14.5 dpc fetal livers; upper panel: representative dot plots; lower
   panel: histograms representing the percentages and mean fluorescence
   intensities (controls n = 5, KO n = 3). Data in the graphs (B–G) are
   shown as mean ± SD, Student’s t-test: *P < 0.05, **P < 0.01, ***P <
   0.001 comparing donor CD45.2 vs competitor CD45.1; °°P < 0.01, °°°P <
   0.001 comparing CD45.1 population from KO mix vs that one of control
   mix; #P < 0.05, ##P < 0.01, ###P < 0.001 comparing CD45.2 population
   from KO mix vs that one of control mix. BM = bone marrow; HSCs =
   hematopoietic stem cells; KO = knockout; LSK = Lin-Kit+Sca1+; PB =
   peripheral blood; PBS = phosphate buffered saline; SP = spleen.

Table 1.

   Hemochrome of Transplanted Mice
   Parameters 2 wks 4 wks
   CTRL KO CTRL KO
   WBC (10^3/µL) 2.87 ± 0.70 2.60 ± 0.35 9.83 ± 1.53 7.17 ± 1.24[174]^**
   RBC (10^6/µL) 10.12 ± 0.32 8.11 ± 1.73 12.00 ± 1.87 10.80 ± 0.67
   HGB (g/dL) 15.00 ± 0.53 11.80 ± 3.12 15.87 ± 2.13 14.73 ± 0.86
   HCT (%) 42.33 ± 1.62 34.67 ± 9.27 49.30 ± 7.75 44.77 ± 2.78
   MCV (fL) 41.90 ± 0.92 42.40 ± 3.00 41.05 ± 0.28 41.43 ± 0.29[175]^*
   PLT (10^3/µL) 1062.00 ± 419.24 650.00 ± 68.79 1842.67 ± 152.39
   1764.00 ± 151.61
   MPV (fL) 7.00 ± 0.66 8.90 ± 1.10 5.07 ± 0.38 5.07 ± 0.38
   LYM% 51.73 ± 8.35 61.03 ± 3.65 75.80 ± 5.36 70.85 ± 5.59
   MID% 10.07 ± 1.79 6.73 ± 1.94 9.55 ± 0.94 12.27 ± 2.37[176]^*
   GRA% 38.20 ± 6.56 32.23 ± 2.40 14.65 ± 5.99 16.88 ± 3.46
   [177]Open in a new tab
   ^*

   P < 0.05;
   ^**

   P < 0.01 compared to mice injected with cells from control embryos.

   CTRL = controls; GRA = granulocytes; HCT = hematocrit; HGB =
   hemoglobin; KO = knockout; LYM = lymphocytes; MCV = mean corpuscolar
   volume; MID = monocytes; MPV = mean platelet volume; PLT = platelets;
   RBC = red blood cell; WBC = white blood cell.

Erythropoiesis is defective in CK2β KO mice

   Deletion of CK2β in purified Ter119^+ mature erythroid cells was
   complete (Figure [178]4A). The levels of the catalytic α subunit were
   found decreased and a CK2-specific kinase assay showed a 60% residual
   enzymatic activity (Figure [179]4A), likely attributable to the
   remaining α dimers. FACS analysis of CD71 and Ter119 markers in fetal
   livers at 14.5 and 17.5 dpc^[180]53,[181]54 revealed that CK2β KO
   embryos had unbalanced frequencies of erythroid subfractions compared
   to control mice (Figure [182]4B). KO samples displayed accumulation of
   both CD71^low/- Ter119^- and CD71^high Ter119^- immature populations
   and, particularly at 17.5 dpc, depletion of the most mature Ter119^+
   CD71^- fraction (Figure [183]4B, left and middle panels). Analysis on
   pellets of peripheral blood obtained from 17.5 dpc embryos confirmed a
   reduced amount of mature hemoglobinized cells in CK2β KO samples
   (Figure [184]4B). Furthermore, KO fetal liver cells presented a reduced
   expression of both Ter119 and CD71 (Figure [185]4B, right panels).
   Similarly to the HSPCs, the MFI of cKit expression was found reduced in
   CK2β-deficient CD71^+Ter119^-/low population mainly composed of CFU-E,
   BFU-E and pro-Erythroblasts (Figure [186]4C). Complete blood cell count
   analysis performed on peripheral blood of 17.5 dpc embryos pointed out
   a strong reduction of RBC counts in KO samples (Figure [187]4D). To
   further support the evidence of terminal erythroid maturation
   impairment, since the extrusion of the nucleus is one central step of
   erythrocyte production, we evaluated the amount of nucleated and
   anucleated erythroid cells within the gate of Ter119^+ FSC^low
   cells.^[188]55,[189]56 The nuclear staining through Syto-16 tracker
   evidenced a higher percentage of nucleated cells (Syto-16^+) with
   parallel reduction of enucleated cells (Syto-16^-) in KO samples
   compared to controls (Figure [190]4E). Cytospin preparations of
   Ter119^+ fetal liver cells (Figure [191]4F, upper panel) revealed
   dysplastic features with membrane irregularities and blebs (blue
   arrows) of immature erythroblasts in KO mice. Reticulocytes looked
   irregularly shaped with intracellular inclusions resembling in size and
   color (but not in number) the Howell-Jolly bodies (red arrows). Also,
   peripheral blood smears (Figure [192]4F, bottom panel) showed larger
   erythrocytes, which in some instances appeared swollen and vacuolized.
   Since these morphological alterations could rely on defects of red cell
   membrane properties, such as the ability to change shape under the
   effect of mechanical forces, we evaluated erythrocyte deformability
   capacity in 17.5 dpc embryos by means of an ektacytometer. As shown in
   Figure [193]4G, erythrocytes from peripheral blood of KO animals
   presented a significant reduction in the elongation index, thus less
   deformability, when they were exposed to a range of applied stress from
   0.3 to 30 Pa.

Figure 4.

   [194]Figure 4.
   [195]Open in a new tab

   Ablation of Csnk2b in the hematopoietic compartment causes ineffective
   erythropoiesis with red blood cell morphological alteration, impairment
   of cell viability, and unbalance in cell proliferation. (A) Top: WB
   analysis of CK2β and CK2α expression; bottom: CK2 kinase activity
   displayed as count per minute (CMP) in Ter119^+ cells from 17.5 dpc
   fetal livers. (B) Study of erythroid maturation stages by FACS at 14.5
   and 17.5 dpc, analyzing transferrin receptor (CD71) and
   glycophorin-associated protein Ter119. In detail: (B) left panel:
   representative dot plot with the gating strategy and sub-populations 1
   (Ter119^-CD71^-/low), 2 (Ter119^-CD71^high), 3 (Ter119^+CD71^high) 4
   (Ter119^+CD71^-/low) and representative pellets of peripheral blood
   samples from 17.5dpc embryos; middle panel: histograms summarizing
   percentages of each erythroid sub-population; right panel: peaks and
   graphs showing Ter119 and CD71 expression represented as mean intensity
   fluorescence (MFI). Sample size: n = 19 controls and 9 KO at 14.5 dpc;
   n = 20 controls and 6 KO at 17.5 dpc. (C) MFI of cKIT in the gate of
   Ter119^-/low CD71^+ cells (14.5dpc fetal livers; n = 16 controls and n
   = 7 KO). (D) Red blood cell counts (RBC) in the peripheral blood of
   17.5 dpc embryos (n = 7 controls, n = 3 KO). (E) Evaluation of
   nucleated and enucleated erythroid cells through Syto-16 staining.
   Left, contour, and dot plots showing the gating strategy; right, graphs
   summarizing the percentages of nucleated (Syto-16^+) and enucleated
   (Syto-16^-) cells in control and KO erythroid cells. Sample size: n = 3
   controls and n = 3 KO. (F) Upper panel: cytospin of Ter119^+ cells
   purified from fetal livers; lower panel peripheral blood smears from
   17.5 dpc fetuses. Samples stained with May-Grunwald Giemsa. Red arrows
   indicate cells with membrane blebs, and blue arrows indicate abnormal
   reticulocytes. Sample size: data are representative of 4 controls and 4
   KO, at least 3 images were acquired for each sample. (G) Histogram
   representing the deformability of red blood cells in 17.5 embryos; test
   performed using Lorca ektacytometer on whole peripheral blood. Rigidity
   and plasticity are evaluated as Elongation Index (EI); deformability
   was assessed over a range of shear stresses between 0.3 and 30 Pa.
   Sample size: data are representative of 14 controls and 6 KO. Results
   were expressed as means ± SEM. Multiple t-tests were corrected with
   Holm-Sidak method: *P < 0.05, **P < 0.01, ***P < 0.001. (H) Evaluation
   of apoptosis at 14.5 and 17.5 dpc by flow cytometry analysis, after
   staining with Annexin V (AnnV) and 7-AAD; analysis was performed on
   erythroid sub-populations (CD71^+Ter119^-) and (Ter119^+); top panel:
   representative dot plot, bottom panel: histograms showing cell
   percentages. Sample size: controls n = 8, KO n = 4 at 14.5 dpc;
   controls n = 9, KO n = 4 at 17.5 dpc. (I) WB of CASPASE-3, PARP, the
   anti-apoptotic BCL-xL, and the P53 downstream target P21 levels. Sample
   size: n = 4 controls and 4 KO at 14.5 dpc; 12 controls and 13 KO at
   17.5 dpc). For BAX, MCL-1, and PUMA the sample size was composed of n =
   3 controls and 3 KO at 14.5 dpc; n = 6 controls and 5 KO at 17.5 dpc.
   (J) Analysis of cell cycle stages in erythroid sub-populations after
   anti-BrdU and 7-AAD staining combined with Ter119/CD71 markers. Sample
   size: n = 6 controls and 5 KO. In (A–E) and (H–J), graphs report means
   ± SD. In (A) and (I) densitometric analysis, data were normalized over
   GAPDH and over the mean value of controls. Student’s t test: *P < 0.05,
   **P < 0.01, ***P < 0.001. EI = elongation index; FACS = fluorescence
   activated cell sorting; KO = knockout; RBC = red blood cell counts; WB
   = western blot.

   Both fetal and adult CK2β heterozygous (het) mice did not suffer an
   anemic condition: for detailed description of their erythroid
   phenotype, see [196]Suppl. Figure S2, S Results and [197]Suppl. Table
   S7 that reports the hemochrome values.

   Thus, loss of CK2β results in abnormal erythroid maturation, with
   reduction of terminally differentiated cells.

CK2β regulates erythroid cell survival and proliferation by impinging on
multiple intracellular pathways

   To understand the mechanism underlying deficient erythroid maturation,
   we evaluated apoptosis by AnnV and 7-AAD staining (Figure [198]4H). The
   mature Ter119^+ population from KO mice presented, as early as 14.5 dpc
   and to a higher extent at 17.5 dpc, an increase of cells in late-stage
   apoptosis (AnnV^+/7-AAD^+) and necrosis (AnnV^-/7-AAD^+) as compared to
   controls. On the contrary, the immature CD71^+Ter119^- fraction did not
   present significant alterations in cell viability. WB analysis of
   Ter119^+ purified cells (Figure [199]4I) revealed in CK2β-deficient
   cells a significant decrease of PRO-CASPASE-3 with accumulation of its
   active cleaved form, an increase of PARP cleavage and a reduction of
   BCL-xL, a BCL-2 family member pivotal for the survival of late stages
   erythroblasts. Investigation of other BCL-2 family members highlighted
   an up-regulation of the pro-apoptotic proteins BAX and PUMA in KO
   samples at both 14.5 and 17.5 dpc and the same trend was also observed
   for the anti-apoptotic factor MCL-1 that is reported be essential in
   immature erythroblasts and erythroid progenitors.^[200]57 Moreover,
   14.5 and 17.5 dpc fetal liver KO cells expressed markedly higher levels
   of the TP53 downstream target P21 (Figure [201]4I).

   To assess the proliferation kinetics, we performed an in vivo BrdU
   incorporation assay. FACS analysis showed a reduction of CD71^+
   Ter119^- immature cells in the S phase and an increase of Ter119^+
   mature cells in the subG[0]/G[1] phase without significant changes in S
   and G[2]/M phases, in KO mice compared to CTRL (Figure [202]4J).

   Next, we carried out RNA-Sequencing in Ter119^+ fetal liver cells. As
   many as 1143 differentially expressed genes were identified in
   CK2β-null mice as compared to controls (See [203]Suppl. Tables S1 and
   S2 for the complete list of up and down-modulated genes respectively).
   Pathway enrichment analysis revealed in KO samples (Figure [204]5A) an
   up-regulation of some P53-targets, such as Cdkn1a (P21), Zmat, and
   Gadd45, a deregulation of genes associated with the JAK/STAT pathway,
   in particular a down-modulation of Stat5, essential for erythroid cell
   viability,^[205]58 and overexpression of Socs2, coding for an inhibitor
   of this signaling cascade.^[206]59 Moreover, we observed a
   down-regulation of Kit and of genes associated to the PI3K/AKT pathway
   such as Pi3kr1 and Gab1. Interestingly, there was the up-regulation of
   Jun, co-factor of PU.1, known to inhibit erythroid differentiation
   counteracting the transcriptional capability of GATA1, the master
   regulator of terminal erythroid maturation.^[207]5 Another dysregulated
   gene category was the group of solute carrier transporters (Slc)
   implicated in ion homeostasis.^[208]60 Intriguingly, mRNA expression of
   Siglec1, Vcam1, and Selplg were decreased in KO samples. These are
   markers expressed by macrophages of the erythroblastic islands^[209]61
   and it was described in a recent work that a contamination of these
   cells is frequently observed in purified Ter119 fraction because of the
   tight connection between erythroid and macrophage populations.^[210]62
   We therefore validated by qRT-PCR the expression of some of the
   deregulated genes identified by RNA-Seq (genes highlighted in yellow).
   Analysis was performed at 17.5 dpc (Figure [211]5B) and 14.5 dpc
   ([212]Suppl. Figure S3A) on Ter119^+ fetal liver cells. A complementary
   analysis obtained interrogating RNA-seq data with Gene Ontology terms
   of our interest (see [213]Supplemental material for details on the
   analysis) showed in KO samples the significant deregulation of other
   114 genes ([214]Suppl. Figure S3B) involved in erythroid cell membrane
   stability (Ebp4.1-pink highlighted), water homeostasis (Aqp1-green
   highlighted^[215]63), heme biosynthesis (Tmem14c-violet highlighted),
   and in particular genes related to the mitotic process (blue
   highlighted genes). Furthermore, we evaluated the deregulation in
   expression of genes coding for kinases, confirming the down-regulation
   of Csnk2b in KO samples ([216]Suppl. Figure S3C, top), and of genes
   coding for transcription factors, showing a reduction in KO cells of
   Ikzf1 (Ikaros1) and Klf13, both involved in erythroid maturation
   ([217]Suppl. Figure S3C, bottom).^[218]64,[219]65

Figure 5.

   [220]Figure 5.
   [221]Open in a new tab

   Lack of CK2β affects erythroid-associated signaling pathways. (A)
   RNA-Seq analysis on RNA extracted from sorted Ter119^+ control and KO
   fetal liver cells (14.5 dpc): heat maps of the genes significantly up
   or down-modulated after pathway enrichment analysis. Values are
   quantified by the scale bar that visualizes differences in the z score
   (expression difference/SD) relative to the mean. (B–G) Images were
   produced using Ter119^+ cells purified with beads from fetal livers. In
   detail: (B) qRT-PCR performed to validate the different expression of
   some genes of the heat maps (highlighted in yellow) (cKit, Cdkn1a, Jun)
   (n = 4 for control and KO samples). (C) Elisa test, evaluation of EPO
   plasma levels in 17.5 dpc embryos (controls n = 5, KO n = 3). (D) WB
   displaying the expression of EPO-R at 17.5 dpc; top level:
   representative dot plot; bottom: densitometric analysis (controls n =
   3, KO n = 3); (E) WB showing the expression and activation of molecules
   downstream EPO-R: left panel: representative images from 14.5 and 17.5
   dpc; right panel: densitometric values of 17.5 dpc sample. For total
   proteins, the graphs report densitometric signal normalized over GAPDH
   and then over the mean of values in control samples. For the
   phosphorylated forms, densitometric analysis reports the values as
   fraction of the phospho over the corresponding total protein, followed
   by normalization over the mean values of control samples. (F)
   Assessment of GATA1 transcriptional capability analyzing the expression
   of its targets by qRT-PCR and on other key genes involved in erythroid
   differentiation (n = 4 for control and KO samples) and (G) WB. (H)
   Evaluation by WB of ERK and AKT activation on sorted CD71^+Ter119^low/-
   populations. On the left, representative dot plot with the gating
   strategy for sorting; in the middle, representative WB of AKT, ERK,
   CK2α and CK2β expression after stimulation with SCF for 15’; on the
   right, graphs representing AKT and ERK activating phosphorylation:
   densitometric values are reported as phospho/total protein levels
   normalized on the mean of controls. Samples size: n = 4 pools of
   control samples, each one composed of 3 fetal livers, and n = 3 pools
   of KO, each one composed of 2 fetal livers. In (F), Gapdh was chosen as
   reference gene, (G) and (H) was used as loading control. In the graphs
   (B–H), data are shown as mean ± SD; statistical significance was
   analyzed by Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001. AKT
   = serine-threonine kinase; EPO = erythropoietin; EPO-R =
   erythropoietin-receptor; ERK = extracellular signal-regulated kinase;
   KO = knockout; WB = western blot.

CK2β-deficient erythroid cells show deregulated cell signaling and GATA1
activity

   Since EPO-R is essential in driving erythroid maturation from
   pro-erythroblast to late stages, we investigated the signaling upstream
   and downstream of this receptor. ELISA performed on plasma obtained
   from 17.5 dpc embryos showed a trend for lower EPO levels in CK2β KO
   samples (Figure [222]5C). EPO-R protein levels on Ter119^+ purified
   cells resulted unaltered in KO mice compared to controls (Figure
   [223]5D). Next, we analyzed downstream JAK/STAT, ERK, PTEN, and
   PI3K/AKT signaling cascades,^[224]4 this latter culminating in the
   AKT-directed phosphorylation of GATA1 on Ser310 that empowers its
   transcriptional activity.^[225]66 Immunoblot analysis on Ter119^+ fetal
   liver cells (Figure [226]5E) revealed markedly decreased levels of
   total AKT and GATA1 at both 14.5 and 17.5 dpc and of STAT5, ERK1/2
   (p44-42) at 17.5 dpc. However, their phosphorylation at CK2-independent
   or indirectly regulated (ie, AKT S473) sites was not significantly
   affected when the quantification was normalized against the total
   protein level, as opposed to the phosphorylation of residue AKT
   S129^[227]67 (direct CK2 target) which is clearly reduced in KO cells.
   PTEN showed a significant decrease of the inhibitory phosphorylation on
   Ser380/Thr382/383, both at 14.5 and 17.5 dpc and a depletion in total
   protein levels at later time points in CK2β KO mature erythroid cells.
   Moreover, KO samples displayed a higher phosphorylation, thus
   activation, of SAPK/JNK kinases, which are involved in stress signaling
   and in the activation of the transcription factor c-JUN ([228]Suppl.
   Figure S3D). Consistently, KO mice exhibited reduced mRNA levels of the
   GATA1 target genes β-globin,^[229]8 Lrf/Zbtb7a^7, and Klf^[230]5
   transcription factors and of Alas2,^[231]9 which encodes for
   5’-aminolevulinate synthase 2, the enzyme catalyzing the first reaction
   in heme biosynthesis (Figures [232]5F and [233]S3E). As a further
   expected consequence of GATA1 down-modulation, we observed the
   up-regulation of the ε embryonic (β-like) globin chain and of Gata2
   transcription factor (Figures [234]5F and [235]S3E). Validation of
   these findings was also performed at the protein level for ALAS2 and
   LRF (Figure [236]5G). Moreover, the low levels of GATA1 could also
   explain the deregulated expression of Slc genes observed in the RNA-seq
   analysis since they are reported to be controlled by this transcription
   factor.^[237]60,[238]63

   Importantly, in KO samples also FOG1, a key co-factor of GATA1
   essential for both erythropoiesis and megakaryopoiesis,^[239]63,[240]68
   was down-regulated (Figure [241]5G). Additionally, Prussian blue
   staining on fetal liver sections clearly showed the presence of
   reactive ferric iron deposits in CK2β-null mice, likely due to Alas2
   deficiency^[242]69 ([243]Suppl. Figure S3F).

   cKit receptor is known to sustain immature erythroid cells, such as
   BFU-E, and to cooperate with EPO-R in maintaining CFU-E and
   pro-erythroblasts. Since we demonstrated that cKit is down-regulated
   also in these populations (included in Ter119^-/lowCD71^+ gate) (Figure
   [244]4C), we asked whether erythroid linage impairment could be partly
   due to a less effective cKit signaling. CD71^+Ter119^-/low fraction was
   isolated through sorting from 14.5 dpc pools of control or KO fetal
   livers (Figure [245]5H left panel, gating strategy). Then, cells were
   treated with SCF for 15 minutes to evaluate the activation of 2 key
   effectors downstream Kit, that is, ERK and AKT. WB analysis revealed a
   severe reduction of both ERK and AKT phosphorylation in KO samples
   compared to controls (Figure [246]5H, middle and right panels).

   Altogether, these results clearly suggest that CK2 regulates
   erythropoiesis by controlling signaling pathways downstream EPO-R, and
   in particular regulating GATA1 levels and transcriptional proficiency,
   as well as the modulation of cKit expression and its downstream
   targets.

CK2 regulates GATA1 expression, localization, and turnover during erythroid
differentiation

   To understand the CK2-mediated mechanism of signaling and GATA1
   regulation, we used the G1E-ER cell line, a model of
   β-estradiol-inducible GATA1-dependent erythroid
   differentiation.^[247]36 G1E-ER cells were cultured with or without
   β-estradiol in the presence of CX-4945 (silmitasertib), a selective
   inhibitor of CK2 kinase activity.^[248]70 Exposure to β-estradiol led
   to erythroid differentiation as indicated by the appearance of Ter119^+
   CD71^+ cells (Figure [249]6A). G1E-ER cells can differentiate up to the
   orthochromatic erythroblast stage at most and after 72 hours of culture
   with β-estradiol they undergo apoptosis.^[250]6,[251]71–[252]73 To
   note, the addition of silmitasertib caused a significant reduction of
   Ter119^+ cells and a decrease of CD71 expression. Importantly, cell
   viability remained comparable in treated and untreated samples up to 24
   hours ([253]Suppl. Figure S4A), while at 48 hours, only in the
   condition where silmitasertib was used as single treatment, cell
   viability was slightly reduced. Therefore, apoptosis was unlikely the
   cause of the reduction of differentiated cells in the combined
   treatment. Tetramethylbenzidine test ([254]Suppl. Figure S4B) performed
   at 48 hours indicated a decreased content of hemoglobin in the combined
   treatment confirming the impairment of erythroid maturation. The same
   analysis was performed upon Csnk2b or Csnk2a knockdown by siRNA (Figure
   [255]6B). The silencing of both subunits, when combined with
   β-estradiol, produced a similar impairment of erythroid
   differentiation, while the decrease of CD71 MFI was observed also in
   the absence of β-estradiol (Figure [256]6B). AnnV analysis did not
   highlight significant changes in cell viability except for the
   condition of Csnk2a siRNA without β-estradiol, in which a mild increase
   in apoptosis was seen ([257]Suppl. Figure S4C).

Figure 6.

   [258]Figure 6.
   [259]Open in a new tab

   CK2 inhibition and Csnk2b knockdown impair erythroid differentiation of
   G1E-ER cell line through down-modulation of STAT5, GATA1, and FOG1
   molecules. G1E-ER cells were induced to differentiate with β-estradiol
   (0.2 µM) in presence or absence of sub-apoptotic concentration of
   silmitasertib (CX-4945 2.5 µM). Cells were collected at 24 and 48 hours
   from transfection. For Csnk2b and Csnk2a down-modulation, cells were
   electroporated with 300pmol of siRNA or scrambled oligos, 24 hours
   after cells were treated with β-estradiol (0.2 µM) and collected at 48
   and 72 hours. (A) Evaluation of G1E-ER differentiation upon exposure to
   β-estradiol and CX-4945 or (B) after Csnk2b or Csnk2a gene silencing.
   In both (A) and (B), on the left, representative dot plots after
   staining with AnnV and anti-CD71/Ter119 antibodies. The represented
   percentages of differentiated cells (Ter119^+ CD71^+) refer to the gate
   of viable cells; on the right histograms show the percentages of
   differentiated cells and CD71 MFI. (C) and (D) Representative WB of
   expression and activation of molecules downstream EPO-R upon exposure
   to β-estradiol and CX-4945 for 24h or (D) Csnk2b silencing for 48 hours
   (n = at least 6 independent experiments). (E) Gata1 mRNA levels and
   expression of its target genes after CK2 inhibition or (F) Csnk2b
   down-modulation. GAPDH was used as loading control. (G)
   Immunofluorescence using anti-GATA1 antibody after treatment with
   CX-4945 (top panel) or Csnk2b siRNA (bottom panel). In red GATA1, in
   blue DAPI. Images were collected by confocal microscope, acquiring 10
   sections along the z-stack for a total thickness of 10 µm; images are
   shown as maximum projection. Scale bar: 10 µm. Green squares represent
   higher magnification of a single cell. (H) Graphs representing GATA1
   mean fluorescence intensity values of IF experiments performed in
   G1E-ER cells after CK2 inhibition (top panel) or Csnk2b siRNA (bottom
   panel); quantification was performed with ImageJ subtracting background
   signal, at least 3 images for each treatment were analyzed; each dot
   represents a cell. In (A), (B), and (E–H), graphs represent the mean
   values ± SD. Statistical significance was analyzed by Student’s t-test:
   *P < 0.05, **P < 0.01, ***P < 0.001; *significant data compared to
   untreated samples; # significant data compared to β-estradiol alone; °
   significant data compared to CX-4945 alone. EPO-R =
   erythropoietin-receptor; WB = western blot.

   WB analysis showed that silmitasertib used alone or with β-estradiol
   did not cause any significant change in EPO-R levels, whilst the
   combined treatments produced a significant decrease of STAT5 and GATA1
   total protein levels and a trend towards reduction in FOG1, thus
   mirroring the changes detected in fetal liver KO erythroid cells,
   except for AKT and ERK levels that remained unaffected (Figures [260]6C
   and [261]S4E). Silmitasertib also caused a decrease of PTEN inhibitory
   phosphorylation (S380/T382/383) and of CK2α-dependent phosphorylation
   of AKT on S129 but not of phospho-GATA1 S301, phospho-AKT S473 and
   phospho-ERK T202/Y204 when the phosphorylated forms were weighted on
   the total protein levels (Figures [262]6C and [263]S4E). In the
   experiments of Csnk2b and Csnk2a silencing we confirmed the effective
   down-modulation of both subunits through WB analysis (Figures [264]6D
   and [265]S4F); to note, it was already reported that the
   down-modulation of the catalytic subunit led also to a decrease of the
   regulatory protein^[266]22 ([267]Suppl. Figure S4D). Csnk2b silencing,
   even when combined with β-estradiol, did not impact on EPO-R
   expression, however, it caused a decline in GATA1, FOG1, STAT5, and
   PTEN total protein levels without significant changes in their
   phosphorylation sites (Figures [268]6D and [269]S4F). Unlike Csnk2b
   down-modulation, Csnk2a silencing, when combined with β-estradiol, led
   only to reduction of GATA1 expression leaving unchanged FOG1 levels
   ([270]Suppl. Figure S4D). Analyzing some GATA1 transcriptional targets,
   we discovered that silmitasertib in combination with β-estradiol,
   caused a drop in the mRNA expression of β-Globin, Alas2, and Lrf/Zbtb7a
   as well as of Gata1 itself (Figure [271]6E), while Csnk2b siRNA
   produced effects mostly on Alas2 gene (Figure [272]6F). IF showed that
   GATA1-ER protein accumulated exclusively in a nuclear-speckled pattern
   upon differentiation. Strikingly, CK2 inhibition or CK2β
   down-modulation, especially when combined with β-estradiol, produced a
   remarkable reduction in nuclear GATA1 and a redistribution in more
   diffuse puncta (Figure [273]6G). We quantified GATA1 immunofluorescence
   signal confirming WB results for both CK2 inhibition and Csnk2b
   silencing (Figure [274]6H). To test if GATA1 protein was degraded
   through the proteasome, we performed Csnk2b silencing in the presence
   of β-estradiol and the proteasome inhibitor bortezomib (Bz).
   Remarkably, WB analyses showed that Bz was effective in recovering
   GATA1 levels, especially upon Csnk2b siRNA in combination with
   β-estradiol ([275]Suppl. Figure S4G). Since GATA1 degradation by
   proteasome is preceded by its inactivation through caspase
   cleavage,^[276]74 we evaluated whether the blockade of caspases could
   produce similar results in GATA1 protein level restoration. As
   expected, the effects produced by the pan-caspase inhibitor
   Z-Vad-FMK^[277]75 and silmitasertib (data not shown) instead of Csnk2b
   siRNA ([278]Suppl. Figure S4H and S4I) were comparable to the previous
   ones obtained with Bz. The blockade of protein synthesis with
   cycloheximide in combination with β-estradiol and silmitasertib
   produced a higher decrease of GATA1 levels compared to cycloheximide
   and β-estradiol combo, indicating a faster turnover of this protein
   when CK2 is inhibited ([279]Suppl. Figure S4J).

   Summarizing, these data demonstrate that CK2 regulates GATA1 turnover,
   intracellular distribution, and transcriptional activity during
   erythroid differentiation.

CK2 regulates HSP70-GATA1 association

   The chaperone HSP70 binds and protects GATA1 in the nucleus from the
   premature cleavage by caspase-3 in erythroid precursors.^[280]76
   Therefore, we evaluated HSP70 protein levels in Ter119^+ cells purified
   from fetal livers of KO and control mice. WB showed a trend in decrease
   of HSP70 expression in 14.5 dpc KO samples that became significant at
   17.5 dpc (Figure [281]7A). A similar down-modulation of HSP70 protein
   levels was observed in G1E-ER cell line upon Csnk2b silencing plus
   β-estradiol (Figure [282]7B). However, the reduction of HSP70 was only
   at the protein level; indeed, we detected an opposite trend analyzing
   Hsp70 mRNA expression in CK2β KO embryos or after its down-modulation
   in G1E-ER cells ([283]Suppl. Figure S5A upper and lower panels
   respectively). Unlike CK2β depletion, CK2 inhibition did not produce a
   decrease of HSP70 protein amount (Figure [284]7C), prompting us to
   investigate the precise mechanisms of HSP70 regulation. Since HSP70
   chaperoning activity is regulated by C-terminal phosphorylation
   mediated by CK2 and other kinases,^[285]77 we checked HSP70 and
   phospho-HSP70 expression upon CK2 chemical inhibition with
   silmitasertib in G1E-ER cells. Unexpectedly, we did not detect any
   significant change in phosphorylation performing WB analysis in both
   total cell lysates (Figure [286]7C) and in the nuclear fraction
   ([287]Suppl. Figure S5B). Moreover, investigation of HSP70 localization
   by WB in nuclear and cytoplasm extracts ([288]Suppl. Figure S5C) as
   well as by IF ([289]Suppl. Figure S5D) did not evidence significant
   changes in its distribution upon 24 hours of treatment with
   β-estradiol, CX-4945 or the combination of the 2 compounds. We next
   checked if CK2 inactivation could affect the GATA1/HSP70 binding by
   performing proximity ligation assay (PLA) (Figure [290]7D). HSP70-GATA1
   interaction was confirmed in G1E-ER cell line in untreated condition or
   after β-estradiol alone. However, proximity was significantly reduced
   after exposure for 24 hours to silmitasertib with or without
   β-estradiol. Although at 24 hours the protein levels of GATA1 in the
   combo treatment were already reduced and could partially account for
   the decrease of PLA dots, the results upon single treatment with
   CX-4945, which did not produce changes in GATA1 expression, indicate a
   real decrease in GATA1/HSP70 interactions. Analysis performed at an
   earlier time point of 6 hours, when GATA1 protein levels are still not
   altered by CK2 inhibition, showed a significant reduction of PLA dots
   in the combo treatment when compared with untreated condition (Figure
   [291]7D).

Figure 7.

   [292]Figure 7.
   [293]Open in a new tab

   CK2 is involved in the control of GATA1 stability. (A) HSP70 expression
   by WB in Ter119^+ cells purified from fetal liver at 14.5 and 17.5 dpc
   and (B) in G1E-ER cells upon silencing of Csnk2b and treatment with
   β-estradiol (0.2 µM). In (A) number of mice is indicated with dots
   (control) or squares (KO); in (B) n = 9 independent experiments. (C)
   Evaluation of HSP70 protein levels through WB upon CK2 inhibition with
   CX-4945 (2.5 µM) and induction of differentiation with β-estradiol (0.2
   µM) (n = 16 independent experiments). In (A–C), GAPDH was used as
   loading control. (D) PLA for HSP70 and GATA1 proteins performed in
   G1E-ER cells after treatment with CX-4945 and β-estradiol for 6 and 24
   hours. Red dots indicate interactions, in blue the nuclei; on the
   right; graph showing the average number of dots/nuclei ± SD. Objective
   63×, zoom 1×, acquiring 10 slices along the z-stack; the represented
   images are the result of maximum projection. The green squares show
   higher magnifications. Images are representative of 3 independent
   experiments. (E) Immunofluorescence of GATA1 (red) and CK2β (green) in
   liver cells from 14.5 dpc fetuses and (F) in G1E-ER cells after
   treatment with CX-4945 and β-estradiol for 6 hours. In blue the nuclei.
   Images are represented as a single plane; scale bar: 10 µm. (G) IF for
   CK2β (red) and HSP70 (green) localization in untreated G1E-ER cells and
   (H) proximity, through PLA assay, in the same cell line (top) or in
   14.5 dpc total fetal livers of control embryos (bottom). In both (G)
   and (H) nuclei are stained with DAPI (blue) and images are represented
   as maximum projection of 10 section z-stack. 63× objective, zoom 2×.
   Scale bar: 10µm. In (A–D), graphs represent mean ± SD. Statistical
   significance was analyzed by Student’s t test: *P < 0.05, **P < 0.01,
   ***P < 0.001; *significant compared to untreated samples; # significant
   compared to β-estradiol alone. KO = knockout; PLA = proximity ligation
   assay; WB = western blot.

   Next, we observed in fetal liver cells of 14.5 dpc control mice, that
   GATA1 and CK2β displayed partial co-localization in the nuclear
   compartment (Figure [294]7E). In untreated G1E-ER cells both at 6h
   (Figure [295]7F) and 24h after plating ([296]Suppl. Figure S5E), CK2β
   was distributed diffusely and partially co-localized with GATA1 in the
   nucleus. Co-localization persisted after β-estradiol addition with an
   enrichment of the signals of both proteins in the nuclei ([297]Suppl.
   Figure S5E and S5F). Intriguingly, CK2 inhibition caused a
   redistribution of CK2β in a diffuse pattern much less overlapping with
   GATA1 (Figure [298]7F). In addition, immunofluorescence (IF)
   experiments showed that CK2α and β subunits co-localized in G1E-ER
   cells in untreated conditions and after 24 hours of β-estradiol
   exposure with accumulation in the nuclear compartment, thus indicating
   the presence of CK2 holoenzyme during erythroid differentiation
   ([299]Suppl. Figure S5G). Intriguingly, silmitasertib in combination
   with β-estradiol produced the reduction of the nuclear amount of both α
   and β subunits ([300]Suppl. Figure S5G). We also checked CK2β and HSP70
   relationship. IF performed on G1E-ER cells at basal condition revealed
   the absence of co-localization in the nuclear compartment (Figure
   [301]7G), data confirmed also through PLA assay in this cell line and
   in fetal liver cells (unsorted) from control embryos (Figure [302]7H).
   Altogether, these results suggest that CK2 might control GATA1 through
   a dual mechanism: CK2β is involved in sustaining HSP70 protein levels,
   while CK2α-dependent kinase activity is important for the maintenance
   of HSP70-GATA1 interactions, likely independent on HSP70
   phosphorylation.

   A schematic molecular model of our findings is summarized in Figure
   [303]8.

Figure 8.

   [304]Figure 8.
   [305]Open in a new tab

   Visual summary of molecular changes upon CK2β KO/down-modulation or CK2
   inhibition. (A) Molecular phenotype of control vs (B) CK2β KO embryos
   focusing on cKIT and EPO-R signaling pathways and the effects of their
   alterations on HSPC and erythroid cell fate. KO embryos are pale with a
   smaller liver, and present a lower expression of cKIT receptor but
   normal levels of EPO-R with reduced levels of plasma EPO; downstream
   both receptors in KO samples AKT and ERK effectors are less activated
   and reduced in total amount; there is also depletion of STAT5 and GATA1
   transcription factors. Deregulation of these pathways lead to expansion
   of LSK population with likely a reduction in self-renewal capability,
   and a partial impairment of cell viability and differentiation
   potential. On erythroid compartment, loss of CK2β causes an
   accumulation of immature Ter119^- erythroblasts with reduction of their
   proliferation rate and a partial apoptosis; in Ter119^+ fraction CK2β
   loss leads to the impairment of final differentiation with reduction of
   mature erythrocytes also due to increased apoptosis. Moreover, more
   mature erythroid cells show morphologic alterations with increased
   dimensions, membrane irregular shape, and less deformability. (C)
   Proposed model of CK2-dependent control of GATA1 stability. The dashed
   arrow indicates a still unknown mechanism. (1-top) lack of CK2β impairs
   GATA1 stability and enhances its degradation; CK2β absence causes also
   HSP70 protein level reduction, likely compromising its synthesis or
   increasing its turnover, and this event might accelerate the
   degradation of the transcription factor. (2-down) treatment with
   CX-4945 during erythroid differentiation causes delocalization of CK2
   from the nuclear compartment and in turn, destabilizes HSP70/GATA1
   complex. HSP70 levels are not decreased and HSP70 is not de-localized
   into cytoplasm after treatment with CX-4945; however, there are less
   interactions between the 2 proteins, making GATA1 more prone for
   degradation. In both conditions, 1 and 2, GATA1 turnover is increased
   by means of proteasome and caspase activation; the reduction of GATA1
   levels leads to the decrease of its transcription activity. AKT =
   serine-threonine kinase; EPO-R = erythropoietin-receptor; ERK =
   extracellular signal-regulated kinase; HSCs = hematopoietic stem cells;
   KO = knockout.

DISCUSSION

   We have herein described the phenotype of mice lacking CK2β in the
   hematopoietic compartment and provided substantial evidence of a key
   role of this protein in 2 processes: (1) HSPCs progression towards more
   differentiated precursors; (2) regulation of erythropoiesis, exerted by
   controlling proliferation, apoptosis and GATA1 function.

   We discovered that CK2β -deficient HSCs are expanded, due to a higher
   proliferation rate, but at the same time an amount of these cells
   undergoes apoptosis (Figure [306]2). This is in line with our previous
   findings in leukemia stem cells^[307]23 and other studies that
   demonstrate the importance of CK2 in sustaining cell viability and the
   quite different roles of the 2 CK2 subunits in the control of cell
   cycle: while CK2α is more associated to cell cycle
   progression,^[308]78–[309]80 free CK2β is also involved in cell cycle
   inhibition in particular in the regulation of the checkpoint between
   G2/M phases.^[310]81,[311]82 Moreover, HSPCs exhibited alterations in
   the expression of transcription factors important for the maintenance
   of stemness properties as well as for lineage
   commitment^[312]40,[313]42,[314]44,[315]46,[316]47 (Figure [317]2H).
   Thus, a combination of different problems such as the unbalanced
   expression of these factors, the higher mitotic rate, and reduced cell
   viability could lead to the loss of quiescence and subsequent
   exhaustion of HSPC pools coupled with reduction of their
   differentiation potential. Altogether, these several elements would
   also explain, at least in part, the impairment of HSPC engraftment in
   competitive (Figure [318]3) and non-competitive ([319]Suppl. Figure
   S1C-E) transplantation experiments and the reduced capability to form
   colonies in a selective environment such as in methylcellulose cultures
   ([320]Suppl. Figure S1A and S1B). Although we did not observe a
   reduction in CXCR4 expression in CK2β KO HSPCs (Fig. [321]3G), we do
   not completely exclude reduced capabilities in homing process to the BM
   of these cells. Moreover, alteration in transcription factors could
   favor the commitment towards preferred populations such as
   monocyte/macrophages and granulocytes observed in KO embryos. Thus,
   these data suggest that CK2β might tune the HSPCs’ balance between
   proliferation and differentiation. FACS analysis pointed out that CK2β
   KO embryos were able to produce some mature hematopoietic cells (Figure
   [322]1) of the myeloid lineage that likely undergo apoptosis at later
   time points. Thus, a combination of alterations in cell viability and
   in differentiation programs could determine the CK2β KO phenotype such
   as the reduced cellularity of fetal livers. To note, KO embryos also
   displayed reduction in B mature (B220^+CD19^+) cells with expansion in
   percentages of immature pre/pro B population, data that support our
   previous finding obtained in the Csnk2b^fl/fl/CD19^+/Cre KO mouse
   model.^[323]35 The depletion of mature erythroid compartment (Ter119^+)
   in CK2β KO embryos is another cause of reduced fetal liver cellularity.
   We discovered stage-dependent erythroid functions for CK2β. CK2β
   promotes cell cycle progression in immature erythroblasts
   (Ter119^-CD71^+) protects from apoptosis and supports differentiation
   in mature Ter119^+ cells.

   The significant reduction of cKit expression in KO samples might
   contribute to the observed alterations in HSPCs and erythroblasts
   (Figures [324]2F and [325]4C). Indeed, at the BFU-E stage, cells are
   mainly dependent on cKit receptor for proliferation and
   differentiation,^[326]83 while in CFU-E and in pro-erythroblasts stages
   cKit is co-expressed with EPO-R and is able to transactivate EPO-R
   intracellular domain.^[327]3,[328]58,[329]83 We demonstrated that in
   Ter119^-/lowCD71^+ population sorted from CK2β KO fetal livers, which
   contains CFU-E, BFU-E, and pro-E, the key downstream effectors, such as
   ERK and AKT, are less activated (Fig. [330]5H). Thus, it is conceivable
   that an impairment of cKit and EPO-R signaling accounts for the
   proliferative and differentiation defects. Although in CK2β KO more
   mature erythroid cells (Ter119^+) we did not detect a depletion in
   EPO-R protein levels (Figure [331]5D), the trend of reduction in EPO
   concentration could suggest a less effective stimulation upstream of
   this receptor (Figure [332]5C). Since during fetal life EPO is mainly
   produced by hepatocytes and macrophages of the blood islands,^[333]84
   we cannot exclude that macrophage production of EPO is impaired in CK2β
   KO embryos. Moreover, we observed in KO embryos the down-modulation of
   Siglec1, Vacam1, and Selplg (Figure [334]5A) encoding for the adhesion
   molecules essential for the blood island formation.^[335]61 These data
   may suggest that there are looser interactions between nurse
   macrophages and erythroblast cells with potential negative impact on
   erythroid maturation and viability. Biochemical and RNA-seq experiments
   confirmed the ineffective activation of cKit and EPO-R downstream
   cascades (Figure [336]5A and [337]5E). For instance, CK2β was found
   important to positively regulate STAT5 (Figure [338]5A and [339]5E)
   that has a key role in EPO-dependent erythropoiesis being mainly
   involved in the up-regulation of BCL-xL expression.^[340]85,[341]86 The
   inhibitory effect on BCL-XL seems to be specific to this BCL-2 family
   member; indeed, focusing on other factors belonging to the same protein
   family, we observed an increase of the anti-apoptotic factor MCL-1 in
   Ter119^+ CK2β KO samples, likely due to an (ineffective) attempt to
   sustain cell viability. While BCL-XL is a key survival factor in more
   mature erythroid cells, MCL-1 is reported to be more important for
   erythroid precursors.^[342]57 Ter119^+ cells from CK2β-null samples
   evidenced also the up-regulation of the pro-apoptotic proteins BAX and
   PUMA, showing overall a clear unbalance versus cell death signals. We
   discovered that CK2β sustains 2 other key erythroid transcription
   factors, GATA1 and FOG1, whose reduction in KO samples was associated
   to the consequential down-modulation of Lrf, Klf, β-Globin, and Alas2
   target genes. To further support our data, in particular the impairment
   of β-Globin production, Ouyang et al^[343]87 demonstrated that CK2 is
   involved in the positive regulation of EKLF transcriptional activity
   through direct phosphorylation of its interaction domain. Altogether,
   this transcriptional dysfunction might underlie the phenotypic changes
   of higher apoptosis, accumulation of immature erythroblasts, presence
   of non-heme iron deposits ([344]Suppl. Figure S3F), and abnormal shape
   likely due to deregulation of adhesion molecules and ion/water-channels
   as suggested by alteration of Aqp1 and Slc gene expression.^[345]63
   Moreover, we confirmed through ektacytometer analysis less
   deformability of CK2β KO erythrocytes (Figure [346]4G), underlying a
   higher risk of hemolysis when these cells are exposed to mechanical
   stresses in the blood stream.

   In CK2β KO mice the effect on erythroid lineage is likely due to both
   lack of β chains and a decrease of CK2 enzymatic activity; depletion of
   β subunits inhibits holoenzyme formation, thus partially impinging on
   kinase efficiency, but it might also make the α subunits more unstable
   and be degraded as suggested by the reduction of CK2α protein levels in
   CK2β-null Ter119^+ cells (Figure [347]4A). Heterozygous mice are not
   anemic although adult animals present some alterations in bone marrow
   and spleen; this mild phenotype could be explained since β subunits are
   produced in large excess compared to catalytic counterparts,^[348]12
   thus the half-dose of CK2β is likely sufficient to keep CK2 kinase
   activity at normal levels ([349]Suppl. Figure S2B) and to preserve CK2β
   independent functions without reducing cell viability or influencing
   dramatically erythroid differentiation.

   The dysfunctional GATA1/FOG1 activity, by impinging on platelet
   formation, could also explain the presence of hemorrhages in KO embryos
   (Figure [350]1A). Indeed, aberrant platelet development and function
   have been recently demonstrated in the conditional CK2β^fl/flPF4-Cre
   mouse model in megakaryocytes.^[351]88 We can suppose that the
   ineffective hemostasis in the CK2β KO embryos might concur with the
   impaired erythropoiesis to premature lethality. Moreover, given the
   broad functions of this kinase, it is conceivable that the erythroid
   phenotype in KO mice could be the result of the simultaneous impairment
   of other transcription factors (Ikzf1, Klf13, Smad3) as indicated by
   RNA-seq data (Figures [352]5 and [353]S3B,
   [354]S3C).^[355]64,[356]65,[357]89

   We have provided substantial evidence that CK2 could control GATA1
   mainly through post-translational mechanisms. To note, RNA-seq results
   suggest a widespread control by CK2β of genes that encode for ubiquitin
   ligases (Trim^[358]90 and Rnf^[359]91 family genes) and deubiquitinases
   (Usp genes) ([360]Suppl. Tables S1 and [361]S2). Thus, the extensive
   signaling impairment downstream EPO-R could be linked to a shared
   CK2β-regulated mechanism, whereby this protein would be essential to
   avoid an excessively fast protein turnover. To suggest such a possible
   mechanism, we exploited G1E-ER cell model of erythroid differentiation.
   We confirmed in these cells, upon CK2 inhibition or Csnk2b silencing
   and induction of differentiation, the molecular alterations obtained in
   vivo for the signal cascades downstream EPO-R. However, some
   discrepancies in the results obtained between the in vivo and in vitro
   systems could be due to the lack of terminal erythroid differentiation
   capability of G1E-ER cells^[362]6,[363]71–[364]73 and the residual
   expression of CK2β in silencing experiments. Moreover, some differences
   between CK2 inhibitor and Csnk2b silencing effects are reasonable
   considering that the free amount of β dimers^[365]12 could still work
   as chaperone when the alpha kinase activity is inhibited. With this
   cell line model, we partially elucidated the molecular mechanisms of
   CK2-dependent GATA1 regulation: CK2 maintains the nuclear localization
   and distribution of GATA1 and upholds its levels, likely by
   potentiating the activity of HSP70, which is central in regulating
   GATA1 protein degradation rate preventing its cleavage by
   caspases.^[366]76 CK2β sustains HSP70 levels in erythroid cells through
   a mechanism independent from the phosphorylation of the C-terminal
   region (Figure [367]7C), known to contain target residues of CK2, CK1,
   and GSK3 kinases.^[368]77 Evidence suggests that while CK2β might keep
   high the levels of HSP70, CK2α might regulate the stability of the
   HSP70- GATA1 association (Figure [369]7D). Although CK2 co-localizes
   with GATA1 in the nuclear compartment at basal condition and during
   differentiation (Figures [370]7E, [371]7F, and [372]S5E), the exact
   mechanism exploited by CK2 to control HSP70 abundance, thus influencing
   GATA1 stability, has not been elucidated yet. Considering the lack of
   co-localization in IF and signal in PLA assays performed with CK2β and
   HSP70 probes (Figure [373]7G and [374]7H), a direct interaction between
   these 2 proteins is unlikely.

   Also, HSP70 abundance may not be essential in the CK2-mediated
   stabilization of GATA1 at mid-gestation, since the reduction of GATA1
   in fetal liver KO Ter119^+ cells occurred earlier (14.5 dpc) than that
   of HSP70 (17.5 dpc) (Figures [375]5E and [376]7A). We demonstrated that
   inhibition of caspases with ZVAD ([377]Suppl. Figures S4H and S4I) or
   blockade of proteasomes, through bortezomib ([378]Suppl. Figure S4G),
   can rescue GATA1 nuclear expression upon CK2β down-modulation or CK2
   chemical inhibition. These data indicate that in our experimental
   conditions, both mechanisms are involved in GATA1 degradation. Caspase
   activity seems to be more linked to CK2-dependent regulation of HSP70,
   whereas degradation through proteasome might be driven by other
   mechanisms that have not been elucidated yet: (1) CK2-dependent control
   of ubiquitin ligases or deubiquitinases, as suggested by RNA-seq data,
   or (2) CK2 possible influence on other axis of degradation such as the
   one mediated by phosphorylation and acetylation of GATA1 followed by
   its interaction with co-chaperone Hsp27.^[379]92 Moreover, the kinase
   activity was found to be required to keep CK2β and CK2α in the nucleus
   closer to the GATA1/HSP70 complex (Figures [380]7F and [381]S5D) but it
   does not seem to influence HSP70 intracellular localization
   ([382]Suppl. Figures S5C and S5D). To this regard, Filhol O. et
   al.^[383]93 postulated that nuclear localization may sustain the
   function of CK2 and it is known that CK2β autophosphorylation on Ser2
   and Ser3 relies on to the formation of a CK2 multimeric
   structure^[384]94 that in turn might reduce the nuclear-cytoplasmic
   shuttling of the kinase. It has also been reported that
   autophosphorylation on Ser2, Ser3, and that on Ser209 (this latter
   indirectly mediated by CK2 through the intermediate effector CDK1
   kinase) could be important in the regulation of CK2β
   degradation.^[385]95,[386]96 Thus, the loss of CK2 might affect GATA1
   levels in 2 sequential steps: (1) upon the reduction of CK2β and of
   CK2α kinase activity an increased rate of GATA1 degradation occurs when
   differentiation is induced; (2) the subsequent dissociation of the
   HSP70-GATA1 complex or depletion of HSP70 causes a further degradation
   of the transcription factor. Since CK2α and CK2β could also work
   independently from each other, we can hypothesize that they might
   influence HSP70 in different ways, one playing on HSP70 levels and the
   other one on HSP70 ability to interact with other proteins.

   In conclusion, the described novel functions for CK2β in HSPC biology
   and erythropoiesis and in the EPO-R and GATA1 signaling, indicate a key
   role for this kinase as a master regulator of blood cell development.

ACKNOWLEDGMENTS