PKCϵ controls protection against TRAIL in erythroid progenitors

It is now well established that apoptosis plays a central role in the regulation of the size of the hematopoietic stem cell pool¹  as well as in the processes of cell differentiation along the various hematopoietic lineages. Fetal erythropoiesis can be negatively regulated by Fas ligand,^2,3  although bone marrow hematopoiesis does not appear to be affected by Fas deficiency.³  However, Fas expression on hematopoietic stem cells or hematopoietic progenitors can be induced by certain cytokines such as IFNγ or tumor necrosis factor α (TNFα),^3-5  reducing hematopoietic repopulating potential. Relatively little is known on the effects of other members of the TNF family on hematopoietic progenitors. We have demonstrated that TNF-related apoptosis-inducing ligand (TRAIL) acts as a negative regulator of adult erythropoiesis, selectively reducing the number of erythroblasts in liquid culture, as well as reducing the number and size of erythroid colonies in semisolid assays.⁶  Recently, Secchiero et al demonstrated that TRAIL inhibited the generation of mature erythroblasts in liquid culture through the activation of an ERK 1/2–mediated signaling pathway.⁷ 

TRAIL is a member of the TNF family of cytokines, which are structurally related proteins playing important roles in regulating cell death, immune response, and inflammation.^8,9  The unique feature of TRAIL, compared with other members of the TNF family, is its ability to induce apoptosis in a variety of malignant cells both in vitro and in vivo, displaying minimal toxicity on normal cells and tissues.^10,11  TRAIL interacts with 4 high-affinity transmembrane receptors belonging to the apoptosis-inducing TNF-receptor (R) family. TRAIL-R1 (DR4) and TRAIL-R2 (DR5) transduce apoptotic signals on binding of TRAIL, whereas TRAIL-R3 (DcR1) and TRAIL-R4 (DcR2) are homologs to DR4 and DR5, but they lack the intracellular death domain and apoptosis-inducing capability. It has been proposed that TRAIL-R3 and TRAIL-R4 function as decoy receptors protecting normal cells from apoptosis.^12,13 

Most normal human cell types tested to date, including bone, epithelial, endothelial, fibroblastic, and smooth muscle cells, are refractory to TRAIL. Nevertheless, TRAIL can induce hepatocyte apoptosis,¹⁴  as well as cell damage in the human prostate¹⁵  and brain.¹⁶  The earliest biochemical event following engagement of TRAIL death receptors by their ligand is the recruitment of proteins to the intracellular death domain of the receptor to form a structure known as the death-inducing signaling complex (DISC).¹⁷ 

Of the several known isoforms of protein kinase C, we have demonstrated that PKCϵ is selectively posttranscriptionally down-modulated in the EPO-dependent murine 32D-Epo.1 cells, while it is expressed in the parental cell line 32D as well as in the 32D-GM1 and 32D-G1 cells with granulomacrophagic and granulocytic phenotype, respectively. The subsequent observation that the pharmacologic inhibition of PKCϵ increased the number of erythroid colonies in vitro strongly suggested a relevant role for this isoform of PKC in erythropoiesis.¹⁸  Previous observations¹⁹  had already established a link between PKCϵ and apoptosis in different model systems. On the basis of the observation that PKCϵ up-regulation increased the formation and growth rate of tumors in nude mice,²⁰  Gubina et al²¹  demonstrated that PKCϵ prevents apoptosis of the factor-dependent TF-1 cells cultured in the absence of cytokines via Bcl-2 up-regulation.

Given this complex background, we decided to investigate the potential role of PKCϵ in the protection against TRAIL activity during human erythropoiesis.

Primary CD34⁺ cells were isolated from peripheral blood of healthy donors by immunomagnetic positive selection using the CD34⁺ cell isolation Kit (Miltenyi Biotech, Gladbach, Germany) in the magnetic field of a Vario-MACS apparatus (Miltenyi Biotech), according to the manufacturer's protocol. Purity of CD34⁺ cells was immediately checked by anti–CD34-PE mAb (Beckman Coulter, Miami, FL) and flow cytometry. Only samples exceeding 95% purity were used for subsequent experiments.

Purified human CD34⁺ cells were cultured up to 18 days, at an optimal cell density of 1 × 10⁶ cell/mL, in serum-free X-vivo medium supplemented with 3 ng/mL recombinant human interleukin-3 (rIL-3) and 40 ng/mL stem cell factor (SCF), with or without 5 U/mL erythropoietin (EPO). Cytokines were readded every 3 days, up to 18 days.

HeL, K562, and TF-1 cell lines were grown in 10% FBS-enriched RPMI medium at the optimal density of 0.5 × 10⁶ cells/mL. TF-1 cells were maintained in the presence of 3 ng/mL IL-3.

IL-3, recombinant human SCF, and recombinant human EPO were all from PeproTech (London, United Kingdom). As a PKCϵ inhibitor, we used the translocation inhibitor H-Glu-Ala-Val-Ser-Leu-Lys-Pro-Thr-OH at 250 μg/mL for 48 hours (Calbiochem, San Diego, CA).

Aliquots of 0.3 × 10⁶ cells/experimental point were labeled by a panel of anti–TRAIL-R MoAbs (Alexis Biochemical, San Diego, CA). Expression of TRAIL-R1, TRAIL-R2, TRAIL-R3, and TRAIL-R4 was analyzed by indirect staining using 1 μg HS101 anti–human TRAIL-R1, HS201 anti–human TRAIL-R2, HS301 anti–human TRAIL-R3, and HS401 anti–human TRAIL-R4 monoclonal antibodies, followed by PE-labeled goat anti–mouse IgG (Beckman Coulter) as a second reagent. Analysis was performed by an Epics XL flow cytometer (Beckman Coulter) and Expo ADC software (Beckman Coulter).

Double-strand siRNAs (dsRNA) were designed to target sequences corresponding to nt's 223 to 244, 429 to 450, 942 to 963, and 1158 to 1179 on human PKCϵ mRNA (NM005400). The target sequences were as follows: 5′-AAGAT CAAAA TCTGC GAGGCC-3′, 5′-AAGAT CGAGC TGGCTG TCTTT-3′, 5′-AACTA CAAGG TCCCT ACCTTC-3′, and 5′-AAAAA GCTCA TTGCT GGTGCC-3′.

The respective sense and antisense RNA sequences were synthesized by Silencer siRNA Construction Kit (Ambion, Austin, TX).²²  Nonspecific siRNA duplexes containing the same nucleotides, but in irregular sequence (ie, scrambled PKCϵ siRNA), were prepared according to the manufacturer's protocol and used as controls.

The GFP-PKCϵ expression and control plasmid were kindly provided by Professor Peter Parker (Cancer Research UK, London Research Institute).

To maximize transfection efficiency, siRNAs (100 nM each) and GFP-PKCϵ plasmids (1 μg/transfection) were delivered using the Amaxa nucleofection technology (Amaxa, Koeln, Germany) according to the manufacturer's protocols.

Total RNA was isolated using the RNeasy mini kit (Qiagen, Hilden, Germany). Total RNA (1 μg) was reverse transcribed with Malone murine leukemia virus (MMV) reverse transcriptase, and progressive dilutions (1/10, 1/100, 1/1000, 1/10 000) were subjected to PCR amplification to detect β-actin and PKCϵ cDNA.

PCR was performed under the following reaction conditions: 95°C for 30 seconds, 56°C for 30 seconds, 72°C for 30 seconds, and a final extension at 72°C for 5 minutes. We used 35 cycles of amplification. The sequence of primers used for PCR was as follows: β-actin, 5′-TGACG GGGTC ACCCA CACTG TGCCC ATCTA-3′ (sense) and 5′-CTAGA AGCAT TTGCG GTGGA CGATG GAGGG-3′ (antisense); PKCϵ, 5′-CAATGGC CTTCTTAAG ATCAAAA-3′ (sense) and 5′-CCTGA GAGATC GATGATC ACATAC-3′ (antisense).

Cultured cells were counted and 2 × 10⁶ cells were collected at specific time points, washed in PBS, and centrifuged at 200g for 10 minutes. Pellets were resuspended in a cell-lysis buffer (50 mM Tris-HCl, pH 7.4; 1% NP-40; 0.25% sodium deoxycholate; 150 mM NaCl; 1 mM EDTA; 1 mM PMSF; 1 mM Na₃VO₄; 1 mM NaF) supplemented with fresh protease inhibitors, and protein concentration was determined using BCA protein assay kit (Pierce, Rockford, IL). Proteins from each sample (14 μg) were then migrated in 5% SDS–acrylamide gels and blotted onto nitrocellulose filters.

Blotted filters were blocked and incubated with specific primary antibodies diluted as described in the manufacturers' protocols. Specifically, rabbit polyclonal anti-PKCϵ and anti–phospho-PKCϵ antibodies (Upstate, Lake Placid, NY) were used at the concentration of 1 μg/mL. MoAbs anti-PKCδ (Becton Dickinson, Heidelberg, Germany), anti–Bcl-2 (Santa Cruz Biotechnology, Santa Cruz, CA), anti–β-actin, and anti–β-tubulin (Sigma, St Louis, MO) were diluted 1:500, 1:50, 1:5000, and 1:20 000, respectively. Anti–Bax rabbit polyclonal antibody (Cell Signaling Technology, Beverly, MA) was diluted 1:1000 before use.

Filters were washed and further incubated for 1.5 hours at room temperature with 1:5000 peroxidase-conjugated anti–rabbit or with 1:2000 peroxidase-conjugated anti–mouse IgG (Pierce) in the primary antibody working solution at room temperature. Specific reactions were revealed with the ECL Supersignal West Pico Chemiluminescent Substrate detection system (Pierce).

Cell-culture viability was assessed by trypan blue exclusion. Apoptotic cells were identified by flow cytometry as subdiploid peaks generated either by DNA fragmentation or by annexin V/PI staining. Briefly, cells were permeabilized by ethanol in the presence of RNAse H buffer and stained with 50 μg/mL propidium iodide, or phosphatidylserine was stained by FITC conjugate annexin V (ACTIPLATE; Valter Occhiena, Torino, Italy) in Ca²⁺ and PI staining buffer, following the manufacturer's protocol.

The human erythromyeloid cell lines HeL, K562, and TF-1 all expressed at different levels surface glycophorin A. The phenotypic analysis with anti–TRAIL-receptor (TRAIL-R) moAbs revealed that the 3 cell lines express all the receptors, with a very neat expression of death receptors R1 and R2 in HeL and K562 cells, and a lower expression of R2 in TF-1 cells (Figure 1A). After 48 hours of treatment with 50 ng/mL TRAIL, apoptosis was induced in all 3 cell lines (HeL: 52.3% ± 9.1; K562: 49.6% ± 5.0; and TF-1: 50.2% ± 9.1; data are expressed as mean percentages of annexin V⁺ cells ± SD of 3 independent experiments) as detected by annexin V–FITC/PI staining and flow cytometry (Figure 1B).

On the basis of the known prosurvival effects of PKCϵ on TF-1 cells²¹  and its protection against TRAIL-mediated apoptosis in glioma cells,²³  we performed a series of experiments to investigate whether PKCϵ could modulate the sensitivity of erythromyeloid cell lines to the apoptogenic effects of TRAIL. For this purpose, we first overexpressed PKCϵ in our cell lines, using as negative control an inactive PKCϵ K522M mutated (PKCϵ_m) kinase.²⁴  As shown in Figure 2A, both PKCϵ and PKCϵ_m could be very well overexpressed in the 3 cell lines, and both the wild-type and mutated forms of the enzyme could be phosphorylated, as revealed by immunoblotting with anti–phospho-PKCϵ antibody. As PKCϵ overexpression did not modify the surface expression of TRAIL-Rs in either cell line (Figure 2B), we treated transfected and mock K562, HeL, and TF-1 cell lines with 50 ng/mL TRAIL for 48 hours. Figure 2C shows a significant reduction of TRAIL-mediated apoptosis in all 3 cell lines overexpressing PKCϵ, but not PKCϵ_m.

Human peripheral-blood–purified CD34 cells were first analyzed for the surface expression of TRAIL-Rs. Figure 3 shows that none of the 4 TRAIL-Rs is expressed on freshly purified CD34 cells. However, when cultured in serum-free medium in the presence of IL-3, SCF, and EPO, starting from day 3 CD34 cells showed a progressive increase of the surface expression of TRAIL-R1 and TRAIL-R2. On the contrary, TRAIL decoy receptors were hardly expressed at all, with the exception of a transient expression of TRAIL-R4 around day 13. Accordingly, when TRAIL was added to EPO-cultured CD34 cells, they were resistant to apoptosis at day 0. Starting at day 3, cells became sensitive to the apoptogenic effect of TRAIL (Figure 4A-B). Surprisingly, however, notwithstanding the stable surface expression of TRAIL death receptors (and in the virtual absence of TRAIL-decoy receptor expression), the sensitivity to TRAIL of EPO-differentiating CD34 cells decreased progressively from day 7 onward, and from day 10 erythroblasts became resistant to TRAIL (Figure 4B). Control cultures of purified CD34 cells in the absence of EPO were always resistant to the apoptogenic effect of TRAIL (Figure 4C), confirming that the sensitivity to TRAIL of normal CD34 cells between day 3 and day 7 was EPO dependent.

Our next question was therefore if PKCϵ could be implicated in the observed resistance to TRAIL of erythroblasts from day 10 onward. Figure 5A-B shows the kinetic of PKCϵ induction in our primary CD34 cell cultures in the presence or absence of EPO.

To formally prove that the induction of PKCϵ in EPO-cultured erythroblasts was responsible for their resistance to TRAIL, we first pharmacologically inhibited PKCϵ by the selective inhibitor of PKCϵ translocation.^6,18 Figure 5C shows that PKCϵ inhibition increases TRAIL-induced cell death both in cell lines and in day-10 CD34 cells. Subsequently, we designed and successfully transfected in HeL and K562 cells siRNAs targeting PKCϵ mRNA. Figure 5D shows a semiquantitative RT-PCR analysis of HeL cells treated with anti-PKCϵ siRNA, showing a more than 100-fold decrease of specific cDNA amplification. Specific targeting of PKCϵ mRNA with consequent protein synthesis inhibition (Figure 5E) sensitized both HeL and K562 cells to TRAIL-induced apoptosis (Figure 5F). Finally, we overexpressed PKCϵ in CD34 cells differentiated with EPO for 3 days, observing acquired resistance to the apoptogenic effect of TRAIL, while PKCϵ_m-transfected CD34 cells did not (Figure 5G).

Given that the general activation of PKC by PMA does not affect TRAIL-R aggregation at the cell surface,²⁵  and that overexpression of PKCϵ does not modulate TRAIL-R surface density, we looked at Bcl-2 levels as one possible antiapoptotic factor involved in the protection of erythroid progenitor cells.

Starting experiments on primary CD34-derived erythroblasts are shown in Figure 6A. Kinetic analysis of Bcl-2 expression during EPO-driven erythroid maturation showed an up-regulation of the protein expression at day 14, while Bax levels remained constant.

Since it had been previously demonstrated that Bcl-2 levels could be modulated by PKCϵ,²¹  we either inhibited or overexpressed PKCϵ in our cell lines and subsequently immunoblotted for Bcl-2. Results show that siRNA-mediated downmodulation of PKCϵ induces a reduction of Bcl-2 protein expression (Figure 6B), while PKCϵ, but not PKCϵ_m, overexpression increases Bcl-2 protein levels (Figure 6C).

Early erythroid progenitor cells are largely quiescent and require EPO to enter cell cycle, while late erythroid progenitors essentially require EPO as a survival factor that allows them to reach terminal differentiation. In vivo, physiologic EPO concentrations are suboptimal, causing apoptosis of about 20% of the bone marrow erythroid cells.²⁶  It is now well established that erythroid differentiation can be further negatively regulated by death-receptor ligands present in the marrow microenvironment, such as Fas and TRAIL.

We demonstrate that, in differentiating erythroid progenitors, PKCϵ levels are regulated by EPO and control the protection against the apoptogenic effect of TRAIL. PKCϵ is important for cytokine and growth factor receptor–mediated signaling. This has been demonstrated by the observations that loss of PKCϵ results in a severely attenuated response of macrophages to LPS and IFNγ in PKCϵ^–/– animals, as well as in impaired PDGF-receptor signaling.^27,28  It has also been demonstrated that PKCϵ acts as a link between the integrin and IFNγ signaling pathways.²⁹  We had previously demonstrated, indeed, that PKCϵ was one of the PKC isoforms modulated during human and mouse erythroid-cell differentiation¹⁸  and that TRAIL, as one of the TNF family members, could negatively regulate erythropoiesis.⁶  The data that we describe in this paper demonstrate that EPO promotes in human CD34 cells a series of events that, being finely regulated in their kinetics, determine the sensitivity of these cells to TRAIL. In fact at day 0, CD34 cells are insensitive to TRAIL because they do not express death receptors on the cell surface. PKCϵ is virtually absent at this early stage of differentiation. From day 3 onward, differentiating erythroid progenitors express surface death receptors TRAIL-R1 and TRAIL-R2, while PKCϵ levels are still undetectable. Cells become sensitive to TRAIL, and overexpression of PKCϵ at this stage abrogates TRAIL-induced apoptosis. Around day 7/8, PKCϵ levels increase rapidly, while surface death-receptor expression remains stable. PKCϵ induction is EPO dependent, since control CD34 cells cultured in the presence of SCF + IL-3 do not up-regulate PKCϵ synthesis. The induction of PKCϵ levels confers to erythroid cells resistance to TRAIL, notwithstanding the surface expression of death receptors and the virtual absence of surface decoy receptors. The resistance to apoptosis is dependent on PKCϵ induction, since both silencing of PKCϵ mRNA with consequent downmodulation of the protein synthesis and pharmacologic inhibition of PKCϵ activation are able to restore cell sensitivity to TRAIL. Very recently, Gillespie et al³⁰  have shown that PKCϵ levels are correlated with the sensitivity of melanoma cell lines to TRAIL, suggesting that PKCϵ might have a general role in the fine tuning of the signaling emanating from death-receptor triggering.

Our data show that the antiapoptotic molecule Bcl-2 is a candidate mediator of erythroid-cell resistance to TRAIL. Bcl-2 levels are in fact up-regulated in erythroid progenitors with a kinetic that is compatible with that of EPO-driven PKCϵ induction. The observation of the reciprocal effects of PKCϵ downmodulation or overexpression on Bcl-2 levels in our model system demonstrates that Bcl-2 is modulated by PKCϵ levels. Indeed, these data parallel those from Gubina et al²¹  that demonstrated that the overexpression of PKCϵ in the TF-1 cell line was able to induce Bcl-2 expression. However, we cannot exclude that other antiapoptotic intermediates could counteract the effects of TRAIL. For example, although in our experiments Bax levels remained constant, McJilton et al³¹  demonstrated that in the human prostate cancer cell line LNCaP, PKCϵ could interact with Bax blocking its conformational changes required for the mitochondrial death-signaling pathway.

Altogether, our data prompt us to hypothesize a progression of the differentiating erythroid progenitor through an initial phase of resistance to TRAIL (due to the lack of specific surface-receptor expression), followed by a period of sensitivity that ends around day 7, due to the EPO-driven up-regulation of PKCϵ with downstream positive effects on Bcl-2 (Figure 7).

PKCϵ therefore is a novel quantitative regulator of EPO-dependent cell expansion and survival in erythropoiesis, similarly to what was recently suggested by Schmidt et al³²  for Btk.

It will be therefore important to re-examine the pathophysiology of diseases such as myelodysplastic syndrome (MDS), characterized by an enhanced response to death ligands, in the light of PKCϵ levels as key regulators of the physiologic cellular response to TRAIL.

We are grateful to Instrumentation Laboratory, Italy, for technical support and to Vincenzo Palermo, Domenico Manfredi, and Davide Dallatana for technical support.