Interferon γ Induces Upregulation and Activation of Caspases 1, 3, and 8 to Produce Apoptosis in Human Erythroid Progenitor Cells

Four hundred milliliters of blood was obtained from normal donors after receiving informed consent approved by the Vanderbilt University and Department of Veterans Affairs Medical Center Institutional Review Boards. BFU-E were first isolated by sequential density gradient centrifugation, depletion of lymphocytes and adherent cells, and further negative selection of contaminant cells with CD2, CD11b, CD16, and CD45 monoclonal antibodies (MoAbs), as previously described.9 28 The BFU-E were suspended in 20 mL Iscove’s modified Dulbecco’s medium (IMDM) containing 20% fetal calf serum (FCS), 5% pooled human AB serum, 1% bovine serum albumin (BSA), 5 × 10⁻⁵ mol/L 2-mercaptoethanol (2-ME), 10 μg/mL insulin, 2 U/mL EPO (Amgen, Inc, Thousand Oaks, CA), 50 U/mL interleukin-3 (IL-3), 50 ng/mL SCF (Amgen Inc), penicillin at 500 U/mL, and streptomycin at 40 μg/mL in polystyrene flasks to generate ECFC. After incubation at 37°C in 5% CO₂/95% humidified air for 4 days (day-5 cells) or 5 days (day-6 cells), the cells were collected and further enriched by centrifugation through 10% BSA and over Ficoll-Hypaque, as necessary. The cells were then incubated in liquid culture containing 20% FCS, 5% pooled AB serum, 1% BSA, 10 μg/mL insulin, 2 U/mL EPO, and penicillin/streptomycin. IFNγ (4.75 × 10⁷ U/mg; Genzyme Corp, Cambridge, MA) and/or SCF were added, as further delineated, in some cultures. The cells were collected after incubation for the indicated times to perform protein or RNA extractions, and plasma clot cultures were made to measure the number of ECFC.

Cells were plated at a concentration of 10³/mL in an IMDM mixture containing 20% FCS, 5% pooled human AB serum, 1% deionized BSA, 10 μg/mL insulin, 2 U/mL EPO, penicillin, streptomycin, 2 mg/mL fibrinogen (Sigma, St Louis, MO), and 0.2 U/mL bovine thrombin (Parke-Davis, Morris Plains, NJ). In some experiments, IFNγ was added at the indicated concentrations. The clots were fixed on day 15 and stained with 3,3′ demethoxybenzidine and hematoxylin. Colonies of two or more hemoglobinized cells were scored as ECFC. ECFC purity was determined by the plating efficiency and data were expressed as the means ± SD with significance calculated using the t-test. The average purity of the ECFC in our experiments was 61% ± 8% (day 6), 68% ± 9% (day 7), 86% ± 5% (day 8), 90% ± 4% (day 9), and 89% ± 3% (day 10). Because the cell number and colony-forming capacity decreased after incubation of the cells with IFNγ for 72 hours or longer,9 we used an equal amount of RNA or protein from control and IFNγ-treated cells for RNA protection assays and Western blot analyses.

Day-5 cells were incubated with or without 500 U/mL of IFNγ for 72 hours and total RNA was isolated using ULTRASPEC (Bioticx Laboratories, Inc, Houston, TX). Twenty-microgram RNA samples were analyzed for the presence of transcripts of mRNAs related to apoptosis. An hApo-1 Multi-Probe Template Set including probes for caspases-1, -2, -5, -6, -7, -8, and -9 and granzyme B was purchased from PharMingen (San Diego, CA). L32 and GAPDH were included as internal controls. RNA protection assays were performed with the MAXIscript and RPA II Ribonuclease Protection Assay Kit (Ambion, Austin, TX), according to the manufacturer’s recommendations. Protected transcripts were separated by denaturing polyacrylamide gels and quantified by autoradiography.

Cell extracts were prepared by lysing 10⁷ cells in 100 μL lysis buffer containing 1% Triton X-100, 20 mmol/L Tris-HCl, pH 7.5, 10% glycerol, 140 mmol/L NaCl, 100 mmol/L sodium fluoride, 10 mmol/L EDTA, 2 mmol/L vanadate, 0.2 mmol/L phenylmethylsulfonyl fluoride, and 0.15 U/mL aprotinin at 4°C. Insoluble materials were removed by centrifugation for 20 minutes at 14,000g and 4°C.29 The samples were quantitated using a Bio-Rad protein assay kit II (Bio-Rad, Hercules, CA) and were boiled for 5 minutes in a sodium dodecyl sulfate (SDS) sample buffer before 8% or 13% SDS-polyacrylamide gel electrophoresis (SDS-PAGE). The concentration of protein per cell was nearly identical between normal and IFNγ-treated cells. After SDS-PAGE, the proteins were transferred to nitrocellulose. The membranes were blocked by incubation in Tris-buffered saline with 0.05% Tween-20 (TBST) and 5% nonfat milk powder at 4°C overnight. After a brief rinse, the blots were incubated for 2 hours at room temperature in TBST-5% milk powder with one of the following antibodies.

Primary antibodies were diluted as follows: MoAb for FasL (Transduction Laboratories, Lexington, KY) had a dilution of 1:1,000; MoAb for FADD (Transduction Laboratories) had a dilution of 1:250; MoAb for caspase-3 (Transduction Laboratories) had a dilution of 1:1,500; polyclonal rabbit antibody for caspase-3 (a gift from Dr Donald Nicholson, Merck Frosst Canada Inc, Quebec, Canada), which can recognize both the proenzyme and its fragments, had a dilution of 1:10,000; polyclonal goat antibodies for caspase-8, p10, and p20 (Santa Cruz Inc, Santa Cruz, CA) had dilutions of 1:200; polyclonal rabbit antibody for caspase-1 and p10 (Santa Cruz) had a dilution of 1:200; and MoAb for PARP (Enzyme Systems Products, Livermore, CA) had a dilution of 1:10,000.

Immunoblotting was followed by washing in TBST and 2 hours of incubation at 26°C with antimouse (Amersham Corp, Arlington Heights, IL), antirabbit, (Amersham), or antigoat (Santa Cruz) IgG-peroxidase conjugate (1:2,000) in TBST-5% milk powder. After washing 5 times for 5 minutes each time, the immunoblots were developed using the ECL method (Amersham). Some blots were stripped by incubation in a buffer containing 2% SDS, 62.5 mmol/L Tris-HCl for 30 minutes at 50°C (and were then reblocked and reprobed). Density scanning was performed using an LKB 2222-010 UltraScan XL Laser Densitometer (LKB Produkter AB, Bromma, Sweden). Caspase-3 inhibitor DEVD-fmk and caspase-8 inhibitor IETD-fmk were purchased from Enzyme Systems Products, whereas YVAD-cho and DEVD-cho were purchased from BIOMOL (Plymouth Meeting, PA).

Day-5 cells were treated with or without 500 U/mL of IFNγ for 72 hours and the total RNAs were isolated. RNase protection assays were performed using a multiprobe set including 7 members of the caspase family and granzyme B, as well as two housekeeping genes, L32 and GAPDH, which delineate RNA loading on the gels. Two separate experiments demonstrated that the expressions of caspase-1, -2, -6, -8, and -9 mRNAs were significantly increased in the cells incubated with IFNγ. Caspase-5 and -7 mRNAs were not changed by incubation with IFNγ (Fig 1).

Because FADD and caspase-8, the most receptor-proximal ICE-like protease, are associated with the CD95 DISC and the DISC is essential for induction of apoptosis by Fas,22 we studied the effect of IFNγ on FADD and caspase-8. Human day-6 ECFC were treated with IFNγ (500 U/mL) for 24 to 96 hours and immunoblots were made of the cell protein extracts. This concentration of IFNγ has previously been shown to be optimum in this system.9,14 These studies showed that FADD is constitutively expressed in human erythroid progenitors from day 6 to day 10 and that the amount of FADD is not altered by incubation with IFNγ (Fig 2). FasL expression in the ECFC was similarly examined and also demonstrated constitutive expression without any quantitative change after incubation with IFNγ, similar to the pattern previously demonstrated by flow cytometric analysis.14 In contrast, procaspase-8 was upregulated 30% to 150% by IFNγ after as little as 24 hours of incubation and remained upregulated through 96 hours (Fig 3). Proteolytically cleaved forms of caspase-8 (p31, p20) were clearly observed in our cells after IFNγ treatment, and this cleavage occurred also at and after 24 hours of incubation.

Two groups have reported that STAT1 is required for efficient constitutive expression of caspase-1 in some human cell lines and that caspase-1 is essential for IFNγ-induced apoptosis.30 31Previous experiments in this laboratory have shown that IFNγ induces a strong STAT1 activation (C.H. Dai, unpublished data). To observe the expression and activation of caspase-1 in ECFC, the cells were cultured in medium with or without IFNγ for 24 to 96 hours and the total cell lysates from each time point were prepared for immunoblot analysis using anti–caspase-1 antibody. The results are shown in Fig 4 and demonstrate that procaspase-1 expression significantly increased after incubation with IFNγ for 24 hours. Two additional experiments, with lesser exposure, clearly confirmed this (data not shown). In addition, the cleavage of caspase-1 was also increased by 24 hours of incubation with IFNγ and persisted for 96 hours. A slight decrease in control caspase-1 with time of incubation also was evident.

Because the activation of caspase-3 to either of its catalytically active p17 or p12 subunits has been demonstrated in cells undergoing apoptosis,32-34 we analyzed caspase-3 expression in human erythroid progenitors at various times after IFNγ treatment. These experiments indicate that caspase-3 is very highly expressed in human erythroid progenitors and increases from day 7 to day 10 of cell incubation with IFNγ (Fig 5). An increase of the proenzyme level of caspase-3 was observed after 24 hours of IFNγ treatment using two different anti-CPP32 antibodies. The mean increase in concentration of procaspase-3 at 24, 48, 72, and 96 hours of incubation with IFNγ, compared with the controls, in four experiments, as detected by scanning, was 15%, 31%, 145%, and 134%, respectively. Activation of caspase-3, identified by the appearance of the 17-kD peptide fragment, occurred at and after 48 hours of IFNγ stimulation (Fig 5).

PARP, one of the proteolytic substrates of the caspases during apoptosis, has been reported to be cleaved to 24- and 89-kD fragments, which represent the N-terminal DNA binding domain and the C-terminal catalytic domain of the enzyme, respectively.35 Whereas most of the caspases can cleave PARP at high concentrations in vitro, it appears that caspase-3 and -7 are primarily responsible for PARP cleavage during apoptosis.36 Because caspase-3 was activated in the presence of IFNγ, we further investigated whether this activation, as well as the activation of other caspases, would lead to the cleavage of PARP. As shown in Fig 6, an increase of the 89-kD PARP cleavage product was observed at 48 hours of IFNγ incubation, approximately the same time as caspase-3 activation. A comparison of the intact PARP band in control and IFNγ-treated samples showed that a decrease of the inactive form of PARP of up to 60% accompanied the formation of the activation fragment after 48 to 96 hours of incubation with IFNγ. In the control cultures not treated with IFNγ, a smaller cleavage of PARP also was noted.

To obtain further evidence for the involvement of the caspase system in IFNγ-induced apoptosis of human ECFC, we tested the capacities of peptide inhibitors of the caspase family to inhibit caspase activities in vitro and reduce apoptosis. DEVD is a relatively potent inhibitor of caspases-1, -3, and -8, whereas YVAD is a potent and selective inhibitor of caspase-1 and IETD is known to inhibit caspase-8.37 38 Day-6 cells were treated with or without the inhibitor(s), in the absence or presence of IFNγ, for 96 hours and then cell lysates were prepared for immunoblot analysis. As shown in Fig 7, either IETD-fmk or DEVD-fmk effectively inhibited the cleavage of caspase-3 and -8 as well as the PARP substrate.

To observe the effects of the inhibitor on the reduction of cell growth induced by IFNγ, two classes of caspase inhibitors, -fmk and -cho, were used. These experiments indicated that the addition of either DEVD or IETD to IFNγ-treated ECFC for 96 hours partially blocked the reduction in cell growth induced by IFNγ (Table 1). The addition of similar concentrations of IETD-fmk or DEVD-fmk to IFNγ-treated plasma clot cultures of day-6 ECFC reduced the number of erythroid colonies with cellular nuclear fragmentation, indicative of apoptosis, by 33% and 38%, respectively (data not shown).

Previous experiments have demonstrated that IFNγ produces apoptosis of erythroblasts and reduces the number and size of erythroid colonies as well as the degree of differentiation.9 IFNγ downregulates the EPO and SCF receptors of ECFC and upregulates Fas mRNA and protein, whereas FasL remains constitutively produced without apparent change as the cells mature.14 To further investigate the manner in which the Fas/FasL pathway is involved in human erythroid progenitor cell apoptosis induced by IFNγ, we examined the expression and activation of several members of the caspase family that have been reported to mediate apoptosis in many cells of different origin.

Although several models for apoptosis-induced by IFNγ through the Fas/FasL system have been proposed,39 the data presented here seem most consistent with the model presented by Fraser and Evan.40 First, ligation of Fas recruits other proteins, such as FADD, and forms the CD95 DISC.19-21 Caspase-8/FLICE is a third component of the DISC and is the first caspase in a cascade of ICE-like proteases activated by CD95. Upon binding to FADD, through the DED, caspase-8 is converted to its active subunits, which are released to the cytosol and can activate a cascade of ICE-like preteases.22 Hence, caspase-8 is called an initiator protease. Overexpression of caspase-8 resulting in apoptosis has been reported.24 When we measured FADD levels, no major change was evident in IFNγ-treated human ECFC. In contrast, procaspase-8 was upregulated and cleavage of caspase-8 was detected in the cells by 24 hours of incubation with IFNγ. The kinetics of caspase-8 expression were similar to the kinetics of Fas upregulation that occurs after incubation of ECFC with IFNγ.14 

Although the precise role of caspase-1/ICE in apoptosis is still uncertain, ectopic expression of ICE in some cells has resulted in apoptosis41 and a reduction of apoptosis in ICE^−/− cells provides evidence for a requirement of ICE in IFNγ-induced cell death. Because ICE appears to be activated near the apex of an ICE-related protease cascade, ICE has been called an amplifier protease, which amplifies the initial caspase-8 activation signal.21 40 Both increases in the expression of ICE and the cleavage of ICE, demonstrated by the appearance of activation fragments, were detected at and after 24 hours of incubation with IFNγ. These observations indicate that ICE is activated in IFNγ-treated cells and that activation of ICE may precede activation of other caspase, such as caspase-3.

Caspase-3/CPP32/apopain has been called an executioner protease, and it appears to have a critical role during apoptosis. It has been shown to be responsible either partially or totally for the proteolytic cleavage of many key proteins, such as PARP, DNA-PK, and U1-70 kD.24Activation of caspase-3 has been reported in human hematopoietic cells during apoptosis induced by EPO deprivation as well as IL-3 deprivation.42 43 In this study, an increased expression of intact caspase-3 was observed in the human ECFC at and after 24 hours of treatment with IFNγ. A distinct activation of caspase-3 and its substrate, PARP, demonstrated by the appearance of proteolytically fragments, was detected 24 hours later, as compared with caspase-8 and caspase-1. This correlates with the appearance of apoptotic changes induced by IFNγ and occurs later than the increase in FAS, caspase-8, and caspase-1.

It has been reported that IFNγ enhanced the activity of several apoptosis-related genes in HT-29 cells, a human colon adenocarcinoma cell line, including Fas, and caspases-1, -3, -4, -7, -8, and -10, but not caspases-2 and -6.44 Some gene mRNAs were induced or upregulated by 1 hour of IFNγ treatment, such as those for Fas and caspases-7, -8, and -10. Caspase-3 and -4 mRNAs were weakly induced, whereas others, such as those for Fas and caspases-1, -8, and -10, were strongly induced by IFNγ.42 However, no studies were reported on the presence, activation, or changes in concentration of the caspase proteins.

To further determine whether caspases had an intermediary role in the induction of apoptosis in human ECFC by IFNγ, two classes of caspase inhibitors were tested. Both prevented at least 90% of the IFNγ induced cleavage of caspases-8 and -3, as well as PARP. However, addition of the inhibitors to the IFNγ-treated cells only partially prevented apoptosis. This suggests that other, alternative pathways for apoptosis also may be activated by IFNγ, or that the inhibitors may be active for a limited time during the 96-hour cell incubations.

In conclusion, FADD and FasL were constitutively present in human ECFC, whereas Fas expression was strongly upregulated by IFNγ. Caspase-1, -3, and -8 were not only upregulated but also activated when these cells were induced to undergo apoptosis by IFNγ. Cleavage of PARP appeared at nearly the same time as caspase-3 activation and both were delayed compared with the activation times for caspase-8 and -1. The caspase inhibitors, IETD and DEVD, very efficiently blocked the activation of caspase-8 plus -3 and the cleavage of PARP, but only reversed the inhibition of cell growth induced by IFNγ by approximately 30%. Therefore, the Fas/FasL pathway may be only partially responsible for production of ECFC apoptosis by IFNγ.

The authors are very grateful to Dr Donald Nicholson and Merck Frosst Canada, Inc for their gift of polyclonal antibody to caspase-3.