Chelerythrin activates caspase-8, downregulates FLIP long and short, and overcomes resistance to tumour necrosis factor-related apoptosis-inducing ligand in KG1a cells

Authors


Uwe Platzbecker, MD, University Hospital Carl Gustav Carus Dresden, Medical Clinic and Polyclinic I, Fetscherstr. 74, Dresden, Germany. E-mail: platzbecker@mk1.med.tu-dresden.de

Abstract

Summary. KG1a cells (CD34+/38) express FAS and TRAIL (tumour-necrosis-factor-related apoptosis-inducing ligand) receptors but are resistant to FAS-ligand and TRAIL/APO2-L (apoptosis antigen-2 ligand)-induced apoptosis. KG1a cells are sensitized to FAS-induced apoptosis by chelerythrin, an inhibitor of protein kinase C (PKC). As cytoplasmatic adaptor molecules of FAS, e.g. FLIP {Fas-associated death domain protein (FADD)-like interleukin 1 beta-converting enzyme [FLICE (caspase-8)-inhibitory protein]}, also modulate TRAIL signals, we determined whether chelerythrin affected TRAIL-mediated apoptosis. Chelerythrin by itself induced apoptosis in KG1a cells, and apoptosis was associated with activation of caspase-8. While TRAIL alone failed to activate caspase-8 or induce apoptosis, the addition of TRAIL to chelerythrin-treated cells significantly enhanced cleavage of caspase-8 and apoptosis. Chelerythrin-pretreated KG1a cells showed decreased phosphorylation of protein kinase C (PKC)-ζ and downregulation of both FLIP long and FLIP short proteins. Downregulation of FLIP and induction of apoptosis were partially abrogated by pretreatment with the specific caspase-8 inhibitor, Z-IETD-FMK. The decrease in FLIP protein expression induced by chelerythrin was accompanied by a progressive increase in mRNA levels of both FLIP long and FLIP short. CD34+ precursors from normal human marrow were also sensitive to chelerythrin but, in contrast to KG1a cells, were not sensitized to TRAIL-mediated apoptosis. Thus, resistance to TRAIL-induced apoptosis in leukaemic KG1a cells but not in normal CD34+ precursors was overcome in the presence of chelerythrin. The mechanism appeared to involve inhibition of PKC. Central targets were FLIP long and FLIP short, and their interactions with caspase-8. Whether such a pathway can be exploited to selectively target leukaemic progenitor cells remains to be determined.

TRAIL [tumour necrosis factor (TNF)-related apoptosis-inducing ligand] is a member of the TNF family and has been shown to preferentially induce programmed cell death in transformed cells (Wiley et al, 1995; Pitti et al, 1996) via two death-domain-containing agonistic receptors, TRAIL R1 (death receptor-4, DR4) and TRAIL R2 (DR5) (Sheridan et al, 1997). Although most myeloid leukaemia cells express at least one agonistic TRAIL receptor, resistance to TRAIL is a common feature (Grundhoff et al, 1999). Chemotherapeutic agents have been shown to upregulate the expression of agonistic receptors, thereby enhancing TRAIL-induced apoptosis (Wen et al, 2000). Among the downstream adaptor molecules for TRAIL, FLIP {Fas-associated death domain protein (FADD)-like interleukin 1 beta-converting enzyme [FLICE (caspase-8)-inhibitory protein]} appears to have a decisive impact on the extent of TRAIL-mediated apoptosis (Schneider & Tschopp, 2000). Several known mRNA splicing variants have been identified, but only the products termed FLIP long and FLIP short are thought to be relevant as functional proteins. Whether these two FLIP isoforms act differently and how their expression is regulated in myeloid cells is not completely understood (Krueger et al, 2001).

Protein kinase C (PKC) comprises a family of serine/threonine kinases, which promote cell survival and protect against cell death. Various studies have implicated PKC-induced activation of phosphatidylinositol-3 kinase (Varadhachary et al, 1999) and the mitogen-activated protein (MAP)-kinase pathway (Holmstrom et al, 1998, 2000; Holmstrom et al, 1999; Ruiz-Ruiz et al, 1999; Varadhachary et al, 1999) as well as phosphorylation and, thus, inactivation of BAD (bcl-xL/bcl-2-associated death promoter) (Villalba et al, 2001) in the inhibition of FAS-mediated signalling. Whether PKC plays the same role in interfering with TRAIL-mediated signals is not well defined.

Recent data suggest that a specific inhibitor of PKC, chelerythrin (CHE), facilitates FAS-mediated apoptosis in the immature (CD34+/38) myeloid leukaemic cell line KG1a, which expresses FAS but is almost completely resistant to anti-FAS-induced apoptosis (de Thonel et al, 2001). KG1a cells express agonistic TRAIL receptor R2 and, to a lesser extent, TRAIL R1. They also show high levels of FLIP long and FLIP short expression. As agonistic TRAIL receptors and FAS share the same cytoplasmatic adaptor proteins to form the death-inducing signalling complex (DISC), we determined the effect and mechanism of CHE on TRAIL-induced apoptosis in KG1a cells.

Materials and methods

Cell cultures

KG1a, Jurkat and HL60 cells were obtained from American Type Culture Collection (ATCC, Rockville, MD, USA). Cells were maintained in Roswell Park Memorial Institute (RPMI)-1640 medium supplemented with 10% fetal calf serum, 1% glutamine and 1% sodium pyruvate. Cells were kept at 37°C in 5% CO2 and were plated in 24-well plates at a density of 0·5 × 106 cells/ml 24 h before an experiment. Bone marrow cells were obtained via aspiration from healthy volunteers who had given informed consent according to procedures approved by the Institutional Review Board of the Fred Hutchinson Cancer Research Center. Mononuclear cells were separated using the Ficoll separation technique, as described (Zang et al, 2001). CD34-positive selection was performed by magnetic-activated cell sorting (MACS), according to the manufacturer's protocol (Miltenyi Biotec, Auburn, CA,USA), to enrich for haemopoietic progenitor cells. A purity of more than 90% CD34+ cells was achieved in all experiments.

Reagents

Recombinant human TRAIL/APO-2 L (Wiley et al, 1995; Pitti et al, 1996) was provided by Genentech (South San Francisco, CA, USA), and CHE was purchased from Sigma (St Louis, MO, USA). The following antibodies were used: polyclonal (rabbit) anti-TRAIL R1, R2, R3 and R4 from Alexis (San Diego, CA, USA); monoclonal anti-TRAIL-R1 (M271)-IgG2a, anti-TRAIL-R2 (M413)-IgG1, anti-TRAIL-R3 (M430)-IgG1, and anti-TRAIL-R4 (M444)-IgG1, provided by Immunex (Seattle, WA, USA); the monoclonal antibodies against FLIP (NF6) and caspase-8 (C-15) were a kind gift from P. Krammer (Deutsches Krebsforschungszentrum, Heidelberg, Germany), anti-PKC-ζ antibodies were purchased from Santa Cruz (Santa Cruz, CA, USA) and from Cell Signalling Technology (Beverly, MA, USA); The caspase-8 inhibitor (Z-IETD-FMK) was purchased from Calbiochem (La Jolla, CA, USA).

Gamma-irradiation of cell cultures

KG1a cells (0·5 × 106/ml) were irradiated (10 and 30 Gy) with external gamma irradiation delivered by a sealed 137Caesium source (Gammacell 3000; MDS Nordion, Canada).

Determination of apoptosis

To determine apoptotic changes, cells were stained with fluorescein isothiocyante (FITC)-conjugated annexin V (Becton Dickinson, Franklin Lakes, NJ, USA), as suggested by the vendor. Stained cells were analysed on a FACScalibur (Becton Dickinson), and the results were analysed using cellquest software (Becton Dickinson).

RNA isolation and cDNA synthesis

Total cellular RNA was isolated using the RNeasy® kit (Qiagen, Valencia, CA, USA), according to the manufacturer's instructions. Cells were lysed in buffer RLT and homogenized using a QIAshredder® spin column. After additional centrifugations using RNeasy spin columns, the RNA was eluted in diethyl pyrocarbonate (DEPC)-treated water. Concentration and purity of RNA were determined by measuring the absorbance at A260/A280.

cDNA was synthesized using a kit purchased from Gibco/BRL (Gaithersburg, MD, USA). RNA (500 ng) was incubated with 1 µl random primers (250 ng/µl), 1 µl NTP in a total volume of 12 µl and heated for 5 min at 65°C. After 5 min on ice, 7 µl of a mix containing 4 µl 5 x reverse transcriptase (RT) buffer, 2 µl 0·1 mmol/l dithiothreitol (DTT) and 1 µl RNAse inhibitor (Roche, Germany; 40 U/µl) was added. The whole mix was kept at 25°C for 10 min and then heated to 37°C. After 2 min at 37°C, 200 U Moloney murine leukaemia virus (M-MLV) reverse transcriptase (Gibco/BRL) was added and the mix was incubated at 37°C for an additional 48 min. Heating at 70°C for 15 min stopped the reaction and the samples were stored at −20°C.

Cloning of FLIP long and short

Cloning of FLIP long and short cDNA into a pcDNA 3·1 plasmid was performed using the Directional TOPO Expression kit®, according to the manufacturer's protocol (Invitrogen, Carlsbad, CA, USA). The following primers were used for FLIP: FLIP long and short forward primer, CAC CAT GTC TGC TGA AGT CAT C; FLIP long reverse primer, TTA TGT GTA GGA GAG GAT AAG TTT; FLIP short reverse primer, TCA CAT GGA ACA ATT TCC AAG AAT.

Taqman polymerase chain reaction (PCR)

The primers were chosen to hybridize to sequences at different exons to avoid amplification of genomic DNA. Probes labelled with 6-carboxy-fluorescein phosphoramidite (FAM) at the 5′ end and with 6-carboxy-tetramethyl-rhodamine (TAMRA) as quencher at the 3′ end were used for both target genes. The following primer/probe combinations were used:

FLIP long.  Forward, CCT AGG AAT CTG CCT GAT AAT CGA; probe, TTC GAG ACA CCT TCA CTT CCC TGG GCT A; reverse, TGG GAT ATA CCA TGC ATA CTG AGA TG.

FLIP short.  Forward, GCA GCA ATC CAA AAG AGT CTC A; probe, CCT TCA AAT AAC TTC AGG ATG ATA ACA CCC TAT GCC; reverse, ATT TCC AAG AAT TTT CAG ATC AGG A.

Equal amounts of cDNA for each sample were added to a prepared mastermix (Applied Biosystems, Foster City, CA, USA). All samples were analysed at least in duplicates. β-2 microglobulin was used as housekeeping reference gene and was purchased as a kit (VIC labelled) from Applied Biosystems. Reactions were performed in separate wells. The amounts of the FLIP long and short transcripts were determined by comparison with calibration curves obtained by serial dilution of plasmids (pcDNA 3·1) containing the full-length cDNA of each gene. The respective amount of FLIP of each sample was divided by the endogenous reference (β-2 microglobulin) amount to obtain a normalized target value. All real-time quantitative PCR reactions were performed on an ABI Prism 7700 sequence detection system (Applied Biosystems) that captured the fluorescent signal and generated a real-time amplification plot. Thermal cycling conditions comprised an initial uracil N-glycosylase (UNG) incubation at 50°C for 2 min, AmpliTaq Gold activation at 95°C for 10 min, 40 cycles of denaturation at 95°C for 15 s, and annealing and extension at 60°C for 1 min.

Western blot

Cells were lysed in protein lysis buffer containing 1% Triton-X, 150 mmol/l NaCl, 0·5% sodium deoxycholate, 0·1% sodium dodecyl sulphate (SDS), 50 mmol/l Tris, 0·25 mmol/l phenylmethlsulphonyl fluoride (PMSF) and proteinase inhibitors (Roche, Indianapolis, IN, USA). Lysed cells were centrifuged at 10 000 g for 3 min, and protein concentrations of the supernatants were determined using the Bradford assay (BIO-RAD Hercules, CA, USA). Equal amounts of protein were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidine difluoride membranes (PVDF; BIO-RAD Hercules). Dependent on the antibody, membranes were blocked with either 2% or 5% non-fat dry milk or 5% bovine serum albumin in phosphate-buffered saline, containing 0·05% Tween-20, at room temperature for 1 h, and were washed and incubated with the primary antibody overnight at 4°C. The blots were washed prior to and after a 1-h incubation with a horseradish-peroxidase-labelled anti-mouse or anti-rabbit secondary antibody (Pierce, Rockford, IL) and developed by enhanced chemiluminescence (Pierce).

Immunoprecipitation

Cell lysates (10 × 106) were prepared in lysis buffer (20 mmol/l HEPES, 2 mmol/l EDTA, 125 mmol/l NaCl, 0·1% NP40, 2 µg/ml aprotinin, 2 µg/ml leupeptin, 1 mmol/l PMSF, 1 mmol/l DTT) for 30 min on ice followed by centrifugation (3 min, 10 000 g). Supernatants were normalized for protein concentration. Each sample was immunoprecipitated with a polyclonal anti-PKC-ζ antibody (Santa Cruz) and collected by absorption to protein G-Sepharose. The bound immunocomplexes were washed three times in lysis buffer. After adding loading buffer, the samples were boiled at 96°C for 5 min and analysed on 12% SDS gels. The PVDF membrane was incubated with an antiphospho-PKC-ζ antibody (Cell Signalling Technology, Beverly, MA, USA) overnight at 4°C and developed as described for Western blots.

Statistics

A student's t-test was used for statistical comparison of results.

Results

KG1a cells express both agonistic and decoy TRAIL receptors and high levels of FLIP long and short

To exclude the absence of agonistic TRAIL receptors as the underlying mechanism of TRAIL resistance in KG1a cells, expression of TRAIL receptors in KG1a cells was compared with HL60 cells, known to express all four receptors. As shown in Fig 1A, KG1a cells expressed very low levels of TRAIL R1, moderate levels of R2 and R3, and high levels of TRAIL R4. TRAIL exposure, however, resulted in only a minor increase in annexin-positive cells after 3 h (data not shown) and 6 h (Fig 2A). As shown in Fig 1B, KG1a cells contained high levels of FLIP long and FLIP short as compared with the TRAIL-sensitive Jurkat cell line.

Figure 1.

(A) TRAIL-R1, R2, R3 and R4 expression in KG1a and HL60 cells. Western blots were performed as described in Materials and methods. Equal amounts of protein (30 µg) were loaded for all samples. (B) Expression of FLIP long (55 kDa) and short (25 kDa) in KG1a cells (TRAIL resistant) and Jurkat cells (TRAIL sensitive). Western blots were performed as described in Materials and methods. Equal amounts of protein (20 µg) were loaded for all samples.

Figure 2.

CHE-mediated sensitization to TRAIL-mediated apoptosis: (A) KG1a cells; (B) unfractionated normal marrow cells (NBM); (C) granulocytes (gran); (D) lymphocytes/progenitors (lymph/prog); and (E) purified normal CD34+ marrow cells. Cells were preincubated with CHE for 1 h. TRAIL was added at 300 ng/ml. After 6 h the proportion of annexin-positive cells was determined by flow cytometry. Each data point represents the mean of three independent determinations ± standard deviations.

CHE activates caspase-8 and sensitizes KG1a cells to TRAIL-induced apoptosis

CHE can induce DNA fragmentation (Freemerman et al, 1996) and apoptosis (Jarvis et al, 1994; Sweeney et al, 2000; Yamamoto et al, 2001). As shown in Fig 2A, incubation of KG1a cells with CHE led to a dose-dependent increase in annexin-positive cells. Apoptosis induction was associated with cleavage of pro-caspase-8 (Fig 3). Of note, the addition of Z-IETD-FMK, a caspase-8 inhibitor, significantly (P = 0·002) reduced the extent of CHE-induced apoptosis (Fig 4). These data suggest that CHE-induced apoptosis was only in part caspase-8 dependent, and an additional pathway was involved.

Figure 3.

Expression of pro-caspase-8 (p53/p55) and cleavage products (p41/p43) in KG1a cells exposed to TRAIL. Cells were exposed to TRAIL (300 ng/ml) with or without CHE preincubation at concentrations of 3, 4 and 5 µmol/l, and harvested at 6 h for Western blots. Equal amounts of protein (10 µg) were loaded for all samples.

Figure 4.

CHE-induced apoptosis in KG1a cells. The y-axis shows the percentage of annexin-positive cells after 6 h of incubation. The extent of apoptosis was reduced (P = 0·002) in the presence of the specific caspase-8 inhibitor (Z-IETD-FMK, 50 µmol/l preincubation for 1 h). Each data point represents the mean of three independent determinations ± standard deviations.

We next determined whether pretreatment with CHE facilitated apoptosis induction by TRAIL. As shown in Fig 2A, KG1a cells preincubated with CHE for 1 h were sensitive to TRAIL-induced apoptosis. The increase in apoptotic cells by TRAIL after CHE preincubation was accompanied by a decrease in the total amount of pro-caspase-8 and an increase in the cleavage products of pro-caspase-8 (Fig 3).

Normal marrow cells express low levels of TRAIL receptors and are resistant to TRAIL (Zang et al, 2001). As shown in Fig 2B, normal marrow cells were sensitive to CHE at lower concentrations than observed with KG1a cells. However, in contrast to KG1a cells, CHE pretreatment resulted in only minimal enhancement of TRAIL sensitivity (Fig 2B). Examination of marrow subpopulations showed that the increase in annexin-positive cells was almost exclusively due to apoptotic granulocytes (Fig 2C), whereas lymphocytes and haemopoietic progenitor cells, while sensitive to CHE, remained resistant to TRAIL (Fig 2D). Furthermore, normal marrow cells enriched for CD34+ cells (purity > 90%) were not sensitized to TRAIL in the presence of CHE at any concentration (Fig 2E).

Reduced expression of FLIP long and FLIP short in KG1a cells in response to CHE

A decrease in FLIP expression may permit TRAIL-mediated apoptosis in otherwise resistant cells. As shown in Fig 5, KG1a cells showed a dose-dependent decrease in FLIP long and FLIP short proteins for at least 6 h after CHE exposure. Different doses of gamma-irradiation with 10 and 30 Gy induced an increase in annexin-positive cells comparable to 4 and 5 µmol/l CHE respectively. However, FLIP long and short expression levels did not change (data not shown). These observations suggest that the CHE-mediated decrease of FLIP protein expression was specific and not a general result of apoptosis. The changes in FLIP protein expression were followed, however, by increasing mRNA levels as determined by quantitative Taqman-PCR (Fig 6A), resulting in significant increases in FLIP long and FLIP short mRNA after 6 h as compared with controls (Fig 6B and C). These data indicate that the CHE-mediated decrease of FLIP protein expression was counter-regulated by an increase in transcription of both splicing variants.

Figure 5.

FLIP long (55 kDa) and FLIP short (25 kDa) expression in KG1a cells exposed to CHE. Cells were incubated with CHE at 1, 2, 3, 4 and 5 µmol/l, and harvested at 1 (A), 3 (B) and 6 h (C) for Western blots. Equal amounts of protein (10 µg) were loaded for all samples.

Figure 6.

Taqman-PCR for FLIP long and FLIP short in KG1a cells. (A) Equal amounts of serial dilutions of plasmids (pcDNA3·1) containing 109−104 copies of either full-length FLIP long or FLIP short transcripts (not shown) were added to a prepared mastermix and run over 40 cycles on a ABI Prism 7700 (Applied Biosystems). A standard curve was calculated for each transcript and served as a reference for the calculation of the amount of transcripts in the sample (unknowns). (B) FLIP long and (C) FLIP short mRNA expression in KG1a cells at 1, 3 and 6 h was determined. Cells were exposed to CHE at 4 µmol/l and harvested at 1, 3 and 6 h. The mRNA was extracted and converted to cDNA by RT-PCR. The reference value is 1 for untreated cells at each time point. There was a significant increase in FLIP long and FLIP short expression (P < 0·05) at 6 h of CHE. Data represent the mean (standard deviation for FLIP long 1 h: ± 0·02, 3 h: ± 0·29, 6 h: ± 0·04, and FLIP short 1 h: ± 0·17, 3 h: ± 0·06, 6 h: ± 0·11) of three independent experiments.

Effect of inhibition of caspase-8 by Z-IETD-FMK on changes of FLIP long and FLIP short expression induced by CHE

As CHE activated caspase-8, we were interested in determining whether the inhibition of caspase-8 would influence the downregulation of both FLIP proteins observed in the presence of CHE. When KG1a cells were preincubated with the specific caspase-8 inhibitor Z-IETD-FMK (50 µmol/l; 1 h) before the addition of CHE, downregulation of both FLIP forms was partially abrogated (Fig 7A). The levels of pro-caspase-8 remained unchanged (Fig 7B). These observations suggest the possibility that CHE-mediated decreases in FLIP expression were caspase-8 dependent. The nature of this interaction remains to be determined.

Figure 7.

Expression of FLIP long and FLIP short (A) and pro-caspase-8 (B) in KG1a cells. Cells were exposed to CHE at 1, 4 and 5 µmol/l with (+) or without (–) preincubation with a caspase-8 inhibitor (Z-IETD-FMK) at 50 µmol/l, and harvested at 1 h. Equal amounts of protein (10 µg) were loaded for all samples. CHE = chelerythrin; C-8-inh. = caspase-8 inhibitor.

Effect of CHE on phosphorylation of PKC-ζ

We next determined whether the effect of CHE involved the atypical PKC-ζ form, which has been shown to play a crucial role in the FAS pathway (de Thonel et al, 2001). Atypical PKCs are directly activated by phosphorylation so that the phosphorylation state is a sensitive readout of the activation of the enzyme (Newton, 2002). Starting at 1 h with a maximum at 6 h of incubation with 5 µmol/l CHE, there was a decrease in the phosphorylation of PKC-ζ at Thr 410 in KG1a cells (Fig 8). These data suggest that PKC-ζ was one target through which CHE sensitized KG1a cells to TRAIL-induced apoptosis.

Figure 8.

Phosphorylation of PKC-ζ (76 kDa) in KG1a cells. Shown are results in an untreated control (left) and in cells exposed to chelerythrin (CHE) at 5 µmol/l and harvested at 1, 3 and 6 h, and lysed. Immunoprecipitation was performed with a polyclonal anti-PKC-ζ antibody (Santa Cruz) as described in Materials and methods. Samples were electrophoresed on SDS-PAGE, transferred to a PVDF membrane and blotted with an antiphospho-PKC-ζ antibody (Cell Signalling Technology).

Discussion

The aim of the present study was to further characterize mechanisms of resistance to TRAIL-mediated apoptosis downstream of the receptor, specifically in regards to the role of FLIP long and FLIP short. FLIP modulates both TRAIL receptor and FAS-mediated signals, and a recent study has shown that FAS-initiated apoptosis in ‘resistant’ KG1a cells is facilitated by CHE (de Thonel et al, 2001), a specific inhibitor of PKC (Herbert et al, 1990).

Our data show that CHE also sensitized KG1a cells to TRAIL-induced apoptosis. CHE alone induced apoptosis in a dose-dependent fashion, which was partially abrogated by a caspase-8 inhibitor. The extent of cell death progressively increased in the presence of TRAIL. Sensitization of KG1a cells to TRAIL-induced apoptosis by CHE was associated with a significant increase in cleavage of pro-caspase-8 and a downregulation of both FLIP long and FLIP short. This is in agreement with the notion that resistance to TRAIL was mediated downstream of the receptors. Normal marrow cells also underwent apoptosis in response to CHE and did so at lower concentrations than required for KG1a cells. However, a substantial increase in apoptosis with the addition of TRAIL was observed only in granulocytes. Normal CD34+ haemopoietic precursors remained resistant to TRAIL even in the presence of CHE, suggesting the presence of additional ‘safety’ checkpoints in these cells. Such a pattern is also consistent with the data on apoptosis resistance in CD34+ cells reported by Kim H. et al (2002).

CHE is a broad inhibitor of all members of the PKC family, which is composed of ‘classical’, ‘novel’ and ‘atypical’ isoforms. CHE was able to interfere with the phosphorylation of PKC-ζ, which has been shown to play an important role in the FAS-mediated apoptosis (de Thonel et al, 2001). The present results in a myeloid leukaemic cell line and normal granulocytes are consistent with previous reports on the involvement of PKC in modulating response to TRAIL in Jurkat cells and solid tumours (Guo & Xu, 2001; Sarker et al, 2001; Shinohara et al, 2001; Trauzold et al, 2001), but apparently not in early haemopoietic precursors. These data urge caution with the extrapolation of results from tumour cell lines to primary cells.

In contrast to a recent report on FAS-mediated apoptosis in KG1a cells (de Thonel et al, 2001), the present results showed a definite involvement of both FLIP isoforms in PKC-mediated sensitization to TRAIL-induced apoptosis in KG1a cells. FLIP shows homology to caspase-8 and blocks the early events of signal transduction via FAS and TRAIL receptors by inhibiting the recruitment and cleavage of caspase-8 to FADD (Irmler et al, 1997; Schneider & Tschopp, 2000). We observed high levels of FLIP long and FLIP short proteins in KG1a cells, and a dose-dependent downregulation after incubation with CHE. Data in the literature are inconsistent regarding the role of PKC in FLIP expression, and generally refer only to FLIP long. The agent Gö6976, as an inhibitor of classical PKC (Martiny-Baron et al, 1993), had no effect on FLIP expression in Jurkat cells (Gomez-Angelats & Cidlowski, 2001). Bisindolylmaleimide, however, another broad inhibitor of PKC, including the atypical forms (Li et al, 1999; Toullec et al, 1991), lowered FLIP long levels in dendritic (Willems et al, 2000) as well as in myeloma cells (Mitsiades et al, 2002). While the published data on the specificity of CHE for PKC are conflicting (Lee et al, 1998; Davies et al, 2000), the results presented here suggest that the expression of FLIP in KG1a cells was PKC dependent, consistent with the fact that KG1a cells have rather high PKC activity (Humbert et al, 2002).

An effect of CHE on caspase-8 has not been described previously. However, such a pattern would explain activation of caspase-3 by CHE as described in granulocytes (Sweeney et al, 2000) and our observation of a sensitization to TRAIL in this subpopulation. Given the specificity of CHE, one possibility is that the activation state (non-activation) of pro-caspase-8 is PKC dependent. As a consequence, inhibition of PKC by CHE would result in pro-caspase-8 activation. A recent report supports this notion by showing that the inhibition of PKC with bisindolylmaleimide VIII was followed by activation of caspase-8, possibly via JNK/p38 kinase (Ohtsuka & Zhou, 2002). Another mechanism might be the inhibition of PKC-mediated phosphorylation of FADD and the known effects of PKC on DISC formation (Gomez-Angelats & Cidlowski, 2001; de Thonel et al, 2001). As apoptosis induction mediated by CHE was only partially abrogated by preincubation with a caspase-8 inhibitor, a second pro-apoptotic pathway is likely to be involved.

Of note was the observation that FLIP downregulation by CHE was reversible with a caspase-8 inhibitor, conceivably related to decreased FLIP consumption with decreased caspase-8 activity. It is known that FLIP can be rapidly degraded through the ubiquitin–proteasome pathway (Fukazawa et al, 2001; Kim Y. et al, 2002). Further, it has recently been shown that FLIP long serves as a dual-function regulator for caspase-8 activation (Chang et al, 2002). Thus, dependent upon the level of expression, FLIP long may have inhibitory or facilitating effects. Finally, the fact that a decrease in FLIP protein expression was followed by an increase in FLIP mRNA levels suggests that the CHE-induced interference with PKC was short lived or that counter-regulatory mechanisms were activated.

These observations provide evidence for the regulation of TRAIL resistance in myeloid leukaemic cells via PKC, caspase-8 and FLIP. As normal marrow CD34+ cells were not sensitized to TRAIL by CHE, additional pathways seem to be involved in the protection against TRAIL-induced apoptosis. Whether PKC inhibitors can be used therapeutically, for example to purge the bone marrow of leukaemic cells, remains to be determined.

Acknowledgments

We appreciate the help of Bonnie Larson and Helen Crawford with manuscript preparation. Many thanks also to Heather-Marie Wilson, Martin Bornhäuser, Heiner Renneberg and Peter Horn for contributing discussions and practical help, and to P. Krammer and A. Krueger for supplying the anticaspase-8 and anti-FLIP antibodies. We appreciate also the support of David Yadock and Shelly Heimfeld. This work was supported in part by PHS grants HL36444, CA18029 and CA87948. U.P. was supported by the Alexander von Humboldt Stiftung (Feodor Lynen Program).

Ancillary