Molecular mechanism of violacein-mediated human leukemia cell death

Violacein (Figure 1A), a purple-colored pigment produced by one of the strains of Chromobacterium violaceum found in the Amazon River, Brazil, is an indole derivative characterized as 3-(1,2-dihydro-5-(5-hydroxy-1H-indol-3-yl)-2-oxo-3H-pyrrol-3-ilydene)-1,3-dihydro-2H-indol-2-one.¹  The biosynthesis and biologic properties of this pigment have been extensively studied; in particular, its antitumoral, antibacterial, antiulcerogenic, antileishmanial, and antiviral activities are of interest.^1-6  Its activity on acute myeloblastic leukemia (HL60) cells is of special interest because the compound effectively induces cell death in these cells.

The development of violacein as a possible novel therapeutic avenue for the treatment of leukemia, however, depends on the establishment of the molecular basis for this cytotoxicity. Evidently, a possible future application of violacein as a therapeutic agent critically depends on its ability to target leukemia cells more effectively than normal blood cells. Hence, research comparing the cytotoxicity of violacein in transformed cells and the corresponding untransformed ones is urgently called for. In addition, identification of the molecular details mediating violacein effects on leukemic cells will provide insight into the possible benefit of this compound in comparison to existing therapeutics.

The above-mentioned considerations prompted us to investigate the effects of violacein in HL60 cells. This cell line is generally accepted as a valid model for studying myeloid leukemia biology.^7,8  Melo et al⁶  showed earlier that HL60 cells react to violacein with both increased cell death and diminished cellular proliferation. Hence, this cell type is attractive for studying the molecular mechanism of violacein, and we compared its effects to those observed in peripheral lymphocytes and monocytes of healthy donors and to its effects in cell types representing other types of leukemia. We used the histiocytic lymphoma cell line U937, which is less differentiated than HL60^9,10  and the K562 cell line which is representative of chronic myelogenic leukemia.^11,12  The results show that violacein mediates its cytotoxic effects on HL60 cells by way of activation of tumor necrosis factor (TNF) receptor signaling and that this effect is specific to these cells because it is not observed in U937 or K562 cell lines or in untransformed cells. Hence, violacein represents the first member of a novel class of cytotoxic drugs, which mediate apoptosis by ways of the specific activation of TNF receptor 1 signal transduction.

HL60 cells were purchased from the American Type Culture Collection (ATCC, Rockville, MD). The polyclonal antibodies against antiphospho-p38 mitogen-activated protein kinase (MAPK), antiphospho-p42/44, antiphospho–c-jun NH2-terminal protein kinase (JNK), antiphospho–MAPK/ERK kinase 1 (MEK1), CREB (cyclic adenosine 5′-monophosphate [AMP] response element binding protein), antirabbit and antimouse peroxidase-conjugated antibodies were from Cell Signaling Technology (Beverly, MA). Antibodies against c-Jun, nuclear factor κB (NF-κB) p50 and p65, caspase 8, caspase 3, p21, signal transducer and activator of transcription 1 (STAT1) and 2, inhibitor of apoptosis 1 (IAP1), TNF receptor 1 (TNFR1), TNF receptor-associated factor 1 (TRAF1) and 2, TNF-α receptor-associated death domain (TRADD) and TNFR1 were from Santa Cruz Biotechnology (Santa Cruz, CA). The polyclonal antibody against the p85 fragment of poly(adenosine 5′-diphosphate [ADP]–ribose) polymerase (PARP) was from Promega (Madison, WI). TNF-α was obtained from Biosource (Etten-Leur, The Netherlands). Infliximab was obtained from Centocor (Leiden, The Netherlands). Violacein was extracted from C violaceum (CCT 3496) and purified according to Rettori and Durán.² 

HL60, U937, and K562 cells were routinely grown in suspension in RPMI 1640 medium containing glutamine (0.200 g/L), antibiotics (penicillin 100 IU/mL; streptomycin 100 μg/mL) and supplemented with 10% heat-inactivated fetal bovine serum, in a 5% CO₂ humidified atmosphere at 37° C. In all experiments 3 × 10⁵ cells/mL were seeded, and after 72 hours the cells were treated with violacein for 24 hours. Violacein dissolved in dimethyl sulfoxide (DMSO) was added to the culture medium and adjusted to a final DMSO concentration of 0.1% (vol/vol).

The blood was collected from healthy donors, and lymphocytes and monocytes were isolated by Ficoll/Hypaque gradient centrifugation. Lymphocytes were cultured at the same conditions described for HL60 cells, the only difference was the addition of 5 μg/mL phytohemagglutinin in each well. Cells were plated at density of 1 × 10⁶ plating/mL in a 24-well plate. The medium was removed 48 hours after cell seeding and replaced with medium containing violacein. Monocytes were isolated from peripheral blood mononuclear cells by an adhesion step. After 4 hours of plating the cells were washed 5 times with RPMI medium and treated with violacein for 24 hours.

Cell viability was assessed by the trypan blue dye exclusion and the MTT reduction assays as previously reported.¹³ 

Cells (3 × 10⁷) were lysed in 200 μL cell lysis buffer (50 mM Tris [tris(hydroxymethyl)aminomethane]–HCl [pH 7.4], 1% Tween 20, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EGTA (ethylene glycol tetraacetic acid), 1 mM O-Vanadate, 1 mM NaF, and protease inhibitors [1 μg/mL aprotinin, 10 μg/mL leupeptin, and 1 mM 4-(2-amino-ethyl)-benzolsulfonyl-fluorid-hydrochloride]) for 2 hours on ice. Protein extracts were cleared by centrifugation, and the protein concentration was determined using Lowry method.¹⁴  An equal volume of 2 × sodium dodecyl sulfide (SDS) gel loading buffer (100 mM Tris-HCl [pH 6.8], 200 mM dithiothreitol [DTT], 4% SDS, 0.1% bromophenol blue, and 20% glycerol) was added to samples and boiled for 10 minutes. Cell extracts, corresponding to 3 × 10⁵ cells, were resolved by SDS-polyacrylamide gel (12%) electrophoresis (PAGE) and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blocked in 1% fat-free dried milk or bovine serum albumin (2%) in Tris-buffered saline (TBS)–Tween 20 (0.05%) and incubated overnight at 4° C with appropriate primary antibody at 1:1000 dilution. After washing in TBS-Tween 20 (0.05%), membranes were incubated with antirabbit or antimouse horseradish peroxidase–conjugated secondary antibodies, at 1:2000 dilutions (in all Western blotting assays), in blocking buffer for 1 hour. Detection was performed by using enhanced chemiluminescence (ECL).

Briefly, 2 × 10⁷ cells were harvested and washed twice with ice-cold phosphate-buffered saline (PBS) and resuspended in 0.2 mL ice-cold cell extract buffer (10 mM HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid]–KOH, pH 7.9, 1.5 mM MgCl₂, 10 mM KC1, 0.5 mM DTT, and 0.2 mM phenylmethysulfonyl fluoride [PMSF]). The cells were kept on ice for 10 minutes to allow them to swell, mixed by vortex for 10 seconds, and microfuged at 4° C at 14000g for 30 seconds. The supernatant was discarded, and the pellet was resuspended in 30 μL nuclear extraction buffer (20 mM HEPES-KOH, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgC1₂, 0.2 mM EDTA (ethylenediaminetetraacetic acid), 0.5 mM DTT, and 0.2 mM PMSF), placed on ice for 20 minutes, and centrifuged at 4° C at 14000g for 2 minutes. The supernatant was saved as the nuclear extract and used in Western blotting assay.

After treatment of the cells with 1000 nM violacein for 24 hours, whole-cell lysates were prepared with lysis buffer (20 mM HEPES, pH 7.7; 2.5 mM MgCl₂, 0.1 mM EDTA, 1% Nonidet-P40 [NP40], 20 mM p-nitrophenylphosphate, 1 mM O-Vanadate, 1 mM 4-(2-amino-ethyl)-benzolsulfonyl-fluorid-hydrochloride), 1 mM DTT, 10 μg/mL aprotinin, and 10 μg/mL leupeptin) and chilled on ice for 2 hours. After centrifugation, lysates were incubated overnight with anti-TNFR1 at 4° C and then rotated with Protein A-Sepharose at 4° C for 2 hours. The beads were washed 3 times with lysing buffer and twice with PBS. Next, the immunoprecipitates were resolved by SDS-PAGE and transferred to PVDF membrane as described in “Western blotting.”

Cytokine level was measured in the cell culture supernatants by using enzyme-linked immunosorbent assay (ELISA) kits from Biosource.

Control and violacein-treated cells were collected and resuspended in 1 × binding buffer (0.01 M HEPES/NaOH, pH 7.4, 0.14 mM NaCl, and 2.5 mM CaCl₂) at a concentration of 1 × 10⁶ cells/mL. Subsequently, 100 μL cell suspension was transferred to a 5-mL tube, and Annexin V fluorescein isothiocyanate (FITC; 5 μL) and 7-amino-actinomycin D (7-AAD; 10 μL) were added. The cells were incubated at room temperature for 15 minutes, after which 400 μL 1 × binding buffer was added, and apoptosis was analyzed by flow cytometry.

The Western blots represent 3 independent experiments. Cell viability and TNF quantification data were expressed as the means ± standard errors of 3 independent experiments run in triplicate. Data for each assay were analyzed statistically by analysis of variance (ANOVA).

To establish the specificity of violacein for leukemic cells as compared with untransformed cells, we determined in parallel the effects of violacein on cell viability for HL60, U937, and K562 cells and for human peripheral blood cells (lymphocytes and monocytes) using a MTT assay. In agreement with Melo et al,⁶  who observed earlier that violacein is proapoptotic for HL60 cells, we established that after 24 hours of incubation violacein strongly reduced mitochondrial activity in this cell type, displaying an IC50 (concentration that inhibits 50%) value around 700 nM (Figure 1B). Effects of violacein on untransformed cells and on U937 and K562 leukemia cells were much less pronounced. Thus, the cytotoxicity effects of violacein are more specific for HL60 cells in this assay.

To obtain more insight into the molecular mechanisms mediating violacein effects on HL60 cells, we examined the phosphorylation state of a panel of signal transduction mediators in response to violacein in concentrations up to 1000 nM. The phosphorylation state remained unchanged of p42/p44 MAP kinase and its upstream activator, the mitogen-activated protein kinase 1, of JNK and its substrate c-jun as well as of cyclic adenosine 5′-monophosphate (AMP) response element binding protein (CREB) (Figure 2A-B). Thus, these elements are likely not involved in violacein cytotoxicity to HL60 cells. In contrast, p38 MAP kinase phosphorylation was strongly up-regulated in response to violacein (Figure 2A). This indicates that violacein action in HL60 cells is accompanied by the induction of a specific set of signal transduction events.

The effects of violacein on the expression of a selected set of proteins were investigated by Western blotting. In accordance with the described cell cycle arrest induced by violacein, we observed a marked up-regulation of p21 protein expression in HL60 cells after 24 hours of incubation (Figure 2C). Concomitantly, c-jun levels slightly decreased, but expression of signal transducers and activators of transcription (STAT1 and STAT2) was augmented (Figure 2B-C).

Importantly, a set of genes recognized as being downstream of NF-κB was up-regulated, including the p50 and p65 subunits of NF-κB, as well as TRAF2 (Figure 3A). The possible activation of NF-κB was confirmed by direct observation of the nuclear translocation of NF-κB as determined by Western blot analysis of nuclear extracts of HL60 cells (Figure 3B). Furthermore, in HL60 cells violacein led to the production of TNF-α, an event generally associated with the activation of NF-κB. Analogous to the cytotoxic effect, violacein-mediated NF-κB activation was specific to HL60 cells as evident by the lack of TNF-α production and up-regulation of NF-κB observed in the other cell types investigated in this study (Figures 3A-C). We concluded that the effects of violacein on gene expression resemble those induced by the activation of proinflammatory receptors in HL60 cells.

As the effects of violacein on HL60 cells (NF-κB activation and cell death) resemble those of TNF receptor 1 activation in these cells, we directly assessed the activation state of this receptor by using a coimmunoprecipitation assay. To this end, the TNF receptor 1 was immunoprecipitated, and, subsequently, the association of TRAF2 to the receptor was studied by using Western blotting. As TRAF2 association to the receptor is dependent on its activation, this assay allows direct assessment of TNF receptor 1 activation. As evident from Figure 4A, treatment of HL60 cells with violacein activated the TNF receptor. In agreement, caspase 8, which is a prominent effector of TNF receptor 1, was strongly activated by violacein (Figure 4B). Finally, preincubation for 1 hour with 100 μg/mL infliximab, an inhibitor of human TNF signaling,¹⁵  strongly reduced the violacein-induced cytotoxicity in HL60 cells (Figure 5A-B). Thus, effects of violacein on HL60 cells involve activation of the TNF receptor 1.

To confirm that the activation of the TNF signaling pathway by violacein culminates in apoptosis of HL60 cells, we treated these cells with violacein (1000 nM), TNF (100 pg/mL), and their combinations with IFX (100 μg/mL) for 24 hours. The activation of caspase 3 was observed when the cells were treated with violacein or TNF, and this effect was prevented in the presence of infliximab (Figure 5C).

In addition, apoptosis induced by violacein was also detected through phosphatidylserine exposure, fragmentation of PARP, and down-regulation of IAP1 (Figure 6A-B), without accompanying necrosis. Hence, the specific induction of TNFR-mediated apoptosis is the predominant molecular mechanism in violacein-induced cytotoxicity.

Natural products have been regarded as important sources of potential chemotherapeutic agents.^16-19  Violacein could be an interesting candidate for cancer treatment because of its cytotoxic effect at low concentration on different transformed cell lines, and it seems especially promising with respect to leukemia.^1,6  The development of violacein to treat cancerous disease is limited, however, by a lack of rational insight into the molecular mechanisms by which violacein affects leukemic cell physiology as well as the specificity of these mechanisms to HL60 cells in comparison to K562, U937, and untransformed cells. In an effort to determine the molecular mechanisms underlying the effects of violacein, we analyzed a variety of signal transduction events. The most prominent effects were the activation of p38 MAP kinase, the activation of caspase 8, and the increased levels of proinflammatory products of the NF-κB pathway. All these events are closely associated with the activation of TNF receptor 1 (Figure 7). Accordingly, direct assessment of the activation status of this receptor revealed that violacein resulted in activation of TNF receptor 1. Subsequent inhibition studies, using the clinical human TNF signaling inhibitor infliximab, showed that activation of TNF receptor 1 mediates at least in part violacein cytotoxicity in HL60 cells. We here demonstrate for the first time, that the mechanism of violacein cytotoxicity in HL60 cells is completely different when compared with other leukemia cell lines (U937 and K562) and with normal human blood cells (lymphocytes and monocytes). These data suggest that the induction of TNF production is a specific response of HL60 cells to violacein.

TNF receptor 1 signaling can simultaneously activate caspase 8 and the transcription factor NF-κB. Activation of caspase 8 is required for TNF-α–induced apoptosis, whereas NF-κB activation is associated with cell survival.^20,21  Thus, violacein-dependent TNF receptor 1 activation will provide 2 opposing signals to the cell with respect to the decision to undergo programmed cell death (Figure 7). This may partly be explained by the fact that in leukemia cells various antiapoptotic pathways are constitutively activated. It has been demonstrated that the activation of IAP, X-linked inhibitor of apoptosis protein (XIAP), and Bcl-2 by NF-κB mediates resistance to several apoptotic stimuli in some cell types, and aberrant activation of these proteins has been described for leukemic cells.^22-24  In HL60 cells, the net effect is clearly proapoptotic as evident from the strong effects of violacein on cell death. Hence, the activation of TNF-dependent proapoptotic mechanisms may gain dominance in relation to the TNF receptor 1-induced antiapoptotic pathways (Figure 7). The dominance of the proapoptotic pathway induced by TNF was confirmed by detecting caspase 3 activation when the cells were treated with violacein or TNF. In addition, this activity was abolished in the presence of infliximab. Therefore, the slight increase of activated caspase 3 caused by the treatment with violacein for 24 hours confirmed previous results, suggesting the early apoptosis induction by this pigment. In other words, the production of TNF stimulated by violacein may amplify the effect of this pigment in neighboring cells. In addition to the increase of activated caspase 3, apoptosis induction by treatment with violacein for 24 hours was demonstrated by down-regulation of IAP1, PARP cleavage, and exposure of phosphatidylserine. Interestingly, no accompanying necrosis was detected, highlighting that induction of TNFR signaling is the major effector of violacein signaling in leukemia cytotoxicity.

Our data also suggest that violacein-induced activation of p38 and TNF in HL60 cells leads to growth arrest. Violacein caused decrease of c-jun expression, which is implicated in the regulation of cell proliferation and cell cycle progression. C-Jun being part of activated protein 1 (AP-1) transcription factor complex functions as a proliferation-promoting gene and is involved in cell cycle progression.²⁵  Consequently, the expression of c-Jun would be expected to change in response to a decreased proliferation and growth arrest of HL60 cells on action of violacein. In addition, we demonstrated a strong increase in p21 protein level. This protein is a typical inhibitor of cell cycle progression. Because HL60 cells are p53 (one of the activators of p21 transcription) negative of homozygous deletions,²⁴  it is conceivable that p21 induction by violacein is p53 independent. Taken together, our data suggest that the increase in p38 activity induced by violacein accounts for the activation of p21 and/or induction of cell cycle arrest at G₂/M. This hypothesis is supported by a recently published report demonstrating the requirement of p38 activation in G₂/M arrest and an increase in p21 at G₁/S phase by low doses of decitabine.²⁶  In addition, it has been demonstrated that the p21 expression can be regulated through STATs, which mechanism is p53 independent and can occur in response to TNFR1 signaling.²⁷  According to this observation and our results, we suggest that the p21 up-regulation in response to violacein could be also dependent on STATs.

An important question concerns the difference of violacein effects in the HL60, K562, U937, and the untransformed peripheral blood cells. We did not detect the same level of violacein-induced cytotoxicity or inflammatory gene transcription in the latter cell types. Although, obviously, this observation argues well for possible future clinical application of violacein, the mechanistic explanation for this observation is still unclear. Possible explanations include differences in levels of TNF receptor expression or different activity of the feedback mechanisms induced after TNF receptor activation. Alternatively, the violacein-dependent cell death of HL60 cells may be a consequence of violacein-induced TNF release and difference between the untransformed and transformed cells lies in their ability to react to violacein with TNF production. Future work is essential to distinguish between these various possibilities.

We thank Angelique Groot and Dennis Van Der Coelen for their assistance with the cytokine quantification and Jennie M. Pater for FACS analysis.