Int. J. Cancer: 66,358-366 (1996) 0 1996 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publicationde I'Union Internationale Contre le Cancer SPHINGOSINE AND ITS METHYLATED DERIVATIVE N,N-DIMETHYLSPHINGOSINE (DMS) INDUCE APOPTOSIS IN A VARIETY OF HUMAN CANCER CELL LINES Elizabeth A. SWEENEY',~, Chouhei SAKAKURA', Tsutomu SHIRAHAMA', Atsushi MASAMUNE~, Hideki OHTA~, Sen-itiroh H A K O M O Rand I ~ . Yasuyuki ~ IGARASH11,2,3 The Biomembrane Institute and Department of Pathobiology, University of Washington, Seattle, WA, USA. In the study of apoptosis initiated by various signals including induce apoptosis (Jarvis et al., 1994; Obeid et al., 1993). ligands binding to cell membrane receptors such as Fas and Previously, we have reported that human neutrophils treated TNFRl , the sphingomyelin pathway and its resulting metabo- with TNF-a (Ohta et al., 1995) and HL60 cells treated with lites, the sphingolipids, have been suggested t o be involved in phorbol myristate acetate (PMA Ohta et al., 1994) exhibited the signaling pathway. In earlier studies we presented data an increase in both Cer and Sph. Furthermore, we have shown which indicated that sphingosine (Sph) itself was increased that Sph exogenously added to neutrophils or HL60 cells during apoptosis induced by phorbol myristate acetate (PMA) in HL60 cells and tumor necrosis factor (TNF) in neutrophils, induces apoptosis in as little as 2 hr, whereas no effect was seen and when added exogenously was able t o induce apoptosis. We at that time point with similar concentrations of the cellpermeable analogue of Cer, Cs-ceramide (CsCer), or with report here that Sph and its methylated derivative N,N,dimethylsphingosine (DMS) are able t o induce apoptosis in Sph-1-phosphate (S-1-P; Ohta et al., 1995). cancer cells of both hematopoietic and carcinoma origin. In The current study examines the effect of Sph and its human leukemic cell lines CMK-7, HL6O and U937, treatment derivatives on apoptosis in a variety of human cancer cell lines with 20 pM Sph for 6 hr caused apoptosis in up t o 90% of cells. of both hernatopoietic and carcinoma origin. Derivatives Human colonic carcinoma cells HT29, HRTI8, MKN74 and COLO205 were shown t o be more susceptible to apoptosis upon examined included S-1-P, N,N,-dimethylsphingosine (DMS) addition of DMS (>SO%) than of Sph (<SO%), yet were weakly and N,N,N,-trimethylsphingosine (TMS). Unlike Sph, these or not sensitive t o N,N,N-trimethylysphingosine (TMS). Under methylated derivatives cannot be acylated to Cer or become phosphorylated to S-1-P. Our results were compared with the same conditions, in the presence of serum, neither Sph- I phosphate nor ceramide analogues Cz-, C6- or C8-ceramide effects of exogenously added analogues of Cer, which have were able t o induce apoptosis in any cell lines. However, in the previously been reported to cause apoptosis (Jarvis et al., 1994; absence of serum, ceramide analogues induced apoptosis in Obeid et al., 1993). leukemia cell lines after 18 hr, yet much less so than Sph or DMS. Furthermore, apoptosis inducedby Sph or DMS could not MATERIAL AND METHODS be inhibited by the ceramide synthase inhibitor fumonisin B I. Apoptosis was not induced by sphingolipids in primary culture Culture of human cells cells, such as HUVEC or rat mesangial cells, but was apparent in Cancer and transformed cell lines used for these experitransformed rat mesangial cells. Additionally, apoptosis induced by Sph, DMS or CzCerwas inhibited by protease inhibitors. Our ments were purchased from the ATCC (Rockville, MD), with data further support the evidence that the catabolic pathway of the exception of the CMK-7 cell line, which was kindly sphingomyelin involving Sph and other metabolites is an inte- provided by Dr. T. Sato (Sato et al., 1989), and HUVECs, gral part of the apoptosis pathway. which were purchased from Cell Systems (Kirland, WA). - o 1996 Wiley-Liss,Inc. Apoptosis is a form of programmed cell death, the term given to cellular death in which the cell itself is involved in its own demise in responsc to specific external or internal signals. Its hallmarks include nuclear compaction followed by internucleosomal degradation, cytoplasmic blebbing and condensation and, finally, disintegration into dense particles called apoptotic bodies, which are then phagocytosed by macrophages or neighboring cells (for review, see Martin et al., 1994). Apoptosis is known to be active during embryogenesis, morphogenesis, cell selection and at the end of the cell's pre-programmed life span, as in terminal differentiation. It can also be activated in cells whose DNA has been damaged beyond the cell's ability to repair it. One mechanistic pathway proposed to be involved in signaling apoptosis is the sphingomyelin pathway (Hannun, 1995; Hakomori and Igarashi, 1995). In this model, the enzyme sphingomyelinase (SMase) is activated to cleave sphingomyelin (SM) into ceramide (Cer) and phosphocholine (PC). The ceramide can be phosphorylated into Cer-1-phosphate or deacylated to sphingosine (Sph). Both Cer and Sph have been implicated as possible second-messengcr molecules in intracellular signaling. Cer has been suggested to be involved in signaling apoptosis induced by the addition of extracellular agents such tumor necrosis factor (TNF)-a (Dbaibo et al., 1993; Kolesnick and Golde, 1994) or anti-Fas antibody (Cifone et al., 1994), and analogues of Cer have been reported to Suspension cell lines CMK-7, HL60 and U937 were grown in RPMI 1640 media, and all others were grown in DMEM supplemented with sodium pyruvate, sodium glutamate, penicillin-streptomycin and, unless noted otherwise, 10% heatinactivated (50°C for 30 min) FBS (Hyclone, Logan, UT). For experiments performed in the absence of serum, hematopoietic cells were adapted to HL-1 serum-free media (Hycor, Irvine, CA) supplemented with sodium glutamate by slowly decreasing the amount of serum-containing RPMI and increasing the amount of HL-1 media over 1 week, until cells were growing well in HL-1 media alone. Isolation of glomeruli and culture of mesangial cells were performed and the specificity of cell type confirmed as reported by Mackazy et al. (1988). Treatment of cells with agents DMS, TMS, S-1-P and CsCer were prepared in house (Igarashi et al., 1989; Ruan et al., 1992; Vunnam and Radir, 1979). All other chemicals were purchased from Sigma (St. Louis, MO) unless otherwise indicated. All lipids were dis7To whom correspondence and reprint requests should be sent, at The Biomembrane Institute, 201 Elliott Ave. W., Suite 305, Seattle, WA 98119, USA. Fax: (206) 281-9899. Received: September 20, 1995 and in revised form December 12, 1995. SPHINGOSINE AND DMS CAUSE APOPTOSIS solved in ethanol/water (50/50) mixture as a stock solution, then added directly to cell culture media. Cell cultures were diluted with fresh media and incubated overnight to insure that cells were in log phase before treatment. Protease inhibitors were prepared in stock solutions as follows: leupeptin 5 m M in distilled water, TLCK 20 mM and ALLnM 10 mM in DMSO and PMSF 10 mM and TPCK 20 mM and 1 mM in ethanol. Fumonisin B1was diluted in PBS to 1mM. Final dilutions were added directly to cell cultures prior 359 to treatment with 20 p M Sph or DMS for 5-6 h r or, in cells grown in HL-1 media, with 10 K M CzCer for 18 hr. Analysis of DNA distribution by flow cytomety The flow cytometric analysis of Nicoletti et al. (1991) was modified for use in this study. Briefly, suspension cultures were centrifuged, washed in PBS and resuspended in lysis buffer (100 mM Na citrate with 0.1% Triton XlOO) containing 50 Kg/ml propidium iodide, then stored overnight at 4°C in the dark. For adherent cell lines, detached cells in the culture supernatant, washes and adherent cells which were detached using trypsin/EDTA were collected by centrifugation, washed DNA Content FIGURE1 - Flow cytometry of sphingolipid-induced apoptosis in hematopoietic cells. CMK-7, HL60 and U937 cells, grown in the presence of 10% FCS, were untreated (Cont; U F C ) or treated with 20 pM of DMS (d-f, sphingosine (SPH; g i ) or Cz-ceramide (CzCer;j-Z) for 6 hr. Sub-diploid peaks representative of apoptotic DNA were apparent in cells treated with sphingosine or DMS but not in ceramide analogues or untreated cells. Bar 1 measures hypodiploid cells; bar 2, cells in G,/G1 phase of the cell cycle; bar 3, cells in G2/Mphase; and bar 4, total cells analyzed. FIGURE2 - Morphology of sphingolipid-induced apoptosis in hematopoietic cells stained with GiemsaIWright. CMK-7, HL60 and U937 cells, grown in the presence of 10% FCS, were untreated (CONT) or treated with 20 p.M sphingosine (SPN) or DMS for 4 hr. Untreated cells exhibited a normal nucleus to cytoplasm ratio, but condensed and fragmented DNA, apoptotic bodies and membrane blebbing were observed in treated cells. TABLE I - PERCENTAGE OF APOPTOTIC CELLS IN LEUKEMIC CELL LINES TREATED FOR 6 OR 18 HR WITH 20 pM SPHINGOLIPIDS IN THE PRESENCE OR ABSENCE OF SERUM Cell line 6 hr CMK-7 CMK-7(SF) HL60 HL60 (SF) U937 U937 (SF) 18 hr CMK-7 CMK-7(SF) HL60 HL60(SF) u937 U937 (SF) Sphingolipids s-1-P - + ++ - SPN DMS +++++ +++++ ++++ ++++ ++++ ++++ +++++ ++++ ++ + ++++ ++++ ++++ ++++ +++ + ++ ND -I+ ++ + ++ +++++ +++++ +++ +++ -/+ - + ++ + Data expressed using the following scale: -/+, <S%; TMS - ++ +++ +- ND -I+ ND CzCer CnCer CsCer - - -I+ - - -I+ - - ++ + - -/+ - - - - - +++ ++ + ++++ -I+ ++++ ++ - / + -/+ ND +++ ++ ND +, 10-25%; ++, 2545%; +++, 45-65%; + + + +, 6545%; + + + + +, > 85%. ND, not donc; SF, cells grown in serum-free media. SWEENEY E T A L 360 with PBS and processed as above. Cells were analyzed using the EPICS profile (Coulter, Hialeah, FL). Analysis of DNA fragmentation by electrophoresis To examine DNA, cells were treated and collected as above, then resuspended in a TTE buffer (10 mM Tris HC1, 10 mM EDTA, 0.1% Triton Xl00) containing 100 pg/ml proteinase K and incubated for 18 hr at 50°C. RNAase was then added at 100 yg/ml and the mixture incubated for 1 hr at 37°C. DNA of adherent cells was further processed by extraction with phenol/ chloroform/isoamyl alcohol followed by ethanol precipitation and resuspended in TE buffer. DNA was then separated by electrophoresis in agarose gel containing 1 pg/ml ethidium bromide. Gels were examined and photographed under UV light. Examination of cellular morphology To examine cellular morphology, cells were treated and collected as above. Slides were prepared using a Cytospin (Shandon, Pittsburgh, PA) and stained with Giemsa-Wright. Cells were viewed and photographed using a Nikon inverted microscope fit with a Nikon camera. RESULTS FIGURE 3 - DNA fragmentation in sphingolipid-treated hematopoietic cells. CMK-7, HL60 and U937 cells were treated for 4 or 6 hr in the presence of 10% FCS, and DNA samples were prepared as described in “Material and Methods.” Treatments are marked as untreated (C), sphingosine 20 I*.M(S20), DMS 20 I*.M(D20) and DMS 10 pM (D10). No fragmented DNA was apparent in the controls, but nucleosomal fragmentation associated with apoptosis is clearly discernible in all 3 cell lines treated with either sphingosine or DMS. Sph and DMS induce apoptosis in hematopoietic cells Hematopoietic cell populations routinely undergo apoptosis during cell differentiation and selection. Consequently, leukemia cell lines, such as HL60 and U937, are commonly used in the study of apoptosis. We examined the ability of sphingolipids and their derivatives to induce apoptosis in 3 human leukemia cell lines, the promyelocytic cell line HL60, the histiocytic lymphoma cell line U937 and the megakaryoblastic cell line CMK-7. Initial studies employed the staining of cellular DNA and distribution analysis by flow cytometry. Cells were treated with Sph, its methylated derivatives DMS and TMS, S-1-P or ceramide analogues C2Cer, C6Cer or C8Cer, then stained and analyzed. In the resulting histograms, shown in Figure 1, cells are gated for normal size in untreated populations. Necrotic cells are much larger and are thus excluded from analysis. Apoptotic cells appear as sub-diploid peaks (represented here by reference line 1). In human leukemic cell lines CMK-7, HL60 and U937, treatment with 20 p M Sph for 6 hr caused apoptosis in 85%, 75% and 36%, respectively. DMS under the same conditions caused less apoptosis, 65%, in CMK-7 cells but more apoptosis in HL60 (85%) and U937 (50%) cells. These responses were dose- and time-dependent (not shown). Treatment with TMS at 20 pM for 6 hr, however, showed apoptosis in only HL60 cells (60%), not in the other 2 lines in the presence of FCS (though apoptosis was induced in a low percentage of CMK-7 cells grown in the absence of serum). Similarly, no apoptosis was apparent after treatment for 6 hr with any of the ceramide TABLE I1 - PERCENTAGE OF APOPTOTIC CELLS IN CARCINOMA, PRIMARY CULTURED AND TRANSFORMED CELL LINES TREATED WITH 20 pM SPHINGOLIPIDS FOR 6 HR IN THE PRESENCE OF SERUM Cell line A431 MKN74 HT29 HRTl8 COL0205 HUVEC Mesengial SV40Mes13 Sphingolipid s-I-P SPN DMS ND ++ ++ +++ +++ ++ +++ ++ - ND -!+ - - -I+ + +++ -I+ - +++ Data presented according to the scale used in Table I. TMS -I+ - -!+ -!+ ND ND CzCer ChCer C&er SPHINGOSINE AND DMS CAUSE APOPTOSIS A431 MKN74 COL0205 361 HT29 HRTl8 E 4 DNA Conknt FIGURE 4 - Flow cytometry of sphingolipid-induced apoptosis in human carcinoma cell lines. A431, MKN74, COLO205, HT29 and HRTl8 cells, grown in the presence of 10% FCS, were untreated (Cont; a-e) or treated with 20 p M of DMS (f-j),sphingosine (SPN; k+j or C2 -ceramide (C2Cer; pt) for 6 hr. Sub-diploid peaks representative of apoptotic DNA were apparent in cells treated with sphingosine or DMS but not in those treated with ceramide analogues or untreated. Phases of the cell cycle are labeled as in Figure 1. FIGURE5 - Morphology of sphingolipid-induced apoptosis in human carcinoma cell lines stained with Giemsa-Wright. MKN74, HRT18, Col0205, A431 and HT29 cells were untreated (CONT) or treated for 6 hr in the presence of 10% FCS with 20 pM sphingosine (SPN) or DMS. Untreated cells appeared normal, but DNA condensation, mebrane blebbing and apoptotic bodies were observed in all treated cell lines. 362 SWEENEY ETAL. plasm, thus illustrating a range of apoptotic stages. Apoptotic bodies were apparent and increased in number with longer treatment. Membrane blebbing was also apparent. To observe if nucleosomal DNA fragmentation was also induced by the lipids, hematopoietic cells were treated with 10-20 FM Sph or DMS for 4 or 6 hr, as indicated in Figure 3. DNA was then extracted and separated on a 1.5% agarose gel. No fragmentation was seen in controls for all 3 cell lines, whereas in Sph- or DMS-treated cells the DNA ladder indicating internucleosomal fragmentation typical of apoptosis was quite clear at 4 hr, and by 6 hr nearly all DNA had fragmented. Apoptosis induction in serum-free cultures Most published experiments to date on Cer-induced apoptosis were conducted in the absence of serum. For comparison, then, hematopoietic cells were adapted to grow in HL-1 serum-free media and tested with agents as above. In comparison with those seen in Figure 1, responses to Sph and DMS remained the same in the absence of serum after 6 hr of treatment, but CzCer showed a small increase in the percentage of apoptosis (<20%) in HL60 and CMK-7 cells and slightly higher in U937 cells (> 2S%), as summarized in Table I. Cells were further treated for 18 hr and analyzed. As shown in Table I, little change is seen from 6 to 18 hr of treatment in the presence of serum. However, in the absence of serum, apoptosis in CMK-7 or HL60 cells treated with C2Cer increased from less than 20% at 6 hr to 50% for CMK-7 cells and more than 60% in HL60 at 18 hr. After 6 hr of treatment with CdCer or C8Cer, no apoptosis was seen in these 2 cell lines. At 18 hr, apoptosis in cells treated with C&er increased to above 25% and a small percentage of apoptosis was seen with C8Cer. In U937 cells treated for 6 hr with CfCer in the absence of serum, apoptosis was observed in 25% of cells and, for C6Cer, in 20%. These values also increased slightly after 18 hr. Sph and DMS induce apoptosis in human carcinoma cell lines As stated earlier, apoptosis is most widely tested in leukemia cell lines. Apoptosis, however, is known to occur in most cell types. Therefore, human colonic carcinoma cell lines HT29, HRT18 and COLOZOS; stomach carcinoma cell line MKN74 and epidermoid carcinoma cell line A431 were also tested for sensitivity to Sph and its derivatives. A summary of the flow cytometry data on apoptosis induced by sphingolipids in cell lines derived from solid tumors is presented in Table I1 FIGURE6 - DNA fragmentation in human tumor cell lines together with that of primary and transformed cell lines. All grown in the presence of 10% FCS and treated with sphingolipids. cancer cell lines tested, except Colo205, proved to be more Fragmentation is not observed in untreated (C) MKN74 cells, but susceptible to apoptosis upon addition of DMS ( > 50%) than in those treated for 6 hr with 20 p M sphingosine or DMS a ladder Sph (25-50%), yet were weakly or not sensitive to TMS or pattern indicative of inter-nucleosomalfragmentation is observed. S-1-P (Fig. 4). All results were dose- and time-dependent (data Untreated A431 cells (C) and those treated with 5 pM of not shown). C2-ceramide(Cer) for 12 hr also showed no fragmentation, but We also tested the 3 Cer analogues under the same cells treated with 5 pM sphingosine (SPN) or DMS for 12 hr conditions as Sph and DMS (in 10% FCS), but in all cell lines showed definite inter-nucleosomalfragmentation. only a very weak response was seen after treatment with 20 pM for 6 hr (Table 11). HT29 and HRT18 responded very slightly analogues tested at concentrations from 2 to 20 KM in the to C2Cer (Fig. 4p-t) and ChCer. MKN74 and COL0205 presence of serum (Fig. l),nor with S-1-P in CMK-7 or HL60 responded just slightly to C&er only. No apoptosis was seen cells (Table I). with CsCer in any cell line. This response did not increase after To examine cells for changes in morphology associated with incubation for 18 hr (data not shown). apoptosis, treated and untreated hematopoietic cells were To confirm the apoptosis observed by flow cytometry, cells stained with Giemsa-Wright and observed under magnifica- were treated for 6 hr with Sph or DMS and stained with tion. Figure 2 shows rcprcscntativc photographs of CMK-7, Giemsa-Wright. Untreated cells showed normal nuclei to HL60 and U937 cells untreated or treated with 20 pM Sph or cytoplasm ratios and occasional cell division. Cells treated for DMS for 4 hr. In untreated cultures, the majority of cells were 6 hr with 20 pM Sph or DMS showed darkly stained, in GI phase with normal DNA to cytoplasm ratios. Occasional condensed nuclei corresponding to apoptosis with varying cells in various stages of mitosis were evident. After treatment amounts of cytoplasm still associated (Fig. 5). Membrane with Sph or DMS, cell nuclei had become condensed and blebbing was apparent in cells in early apoptosis, and apoptotic fragmented and in some cases had lost all associated cyto- bodies were abundant in most cultures. SPHINGOSINE AND DMS CAUSE APOPTOSIS 363 8 A HUVEC Mesangial SV40 Mes 13 4 za v) DNA Content FIGURE 7 - Effect of sphingolipidson apoptosis in primary cultured and transformed cell lines grown in the presence of 10% FCS. (a) No apoptosis was apparent in HUVECs untreated (Cont) or treated for 6 hr with 20 pM sphingosine (SPN) or DMS. (6) Apoptosis is not seen in normal rat mesangial cells, untreated (Cont) or treated for 6 hr with 20 p M sphingosine (SPN) or DMS, as indicated, but is apparent in sphingosine- or DMS-treated SV40 Mes 13-transformed rat mesangial cells. (c) No inter-nucleosomal fragmentation is apparent in DNA samples prepared from normal rat mesangial cells (lanes 5-7) or in SV40 Mes 13-transformed rat mesangial cells untreated (lane 1) or treated with 20 pM C,-ceramide (lane 2). SV40 Mes 13 cells treated with 20 pM sphingosine (lane 3) or DMS (lane 4) exhibited ladder formation indicative of inter-nucleosomal fragmentation associated with apoptosis. DNA samples from A431 and MKN74 cells were further examined for the internucleosomal DNA fragmentation pattern often accompanying apoptosis. Both cell lines demonstrated some ladder-type DNA fragmentation after treatment with DMS. Additionally, A431 cells showed similar fragmentation after treatment with 5 pM Sph for 12 hr but none after CzCer (Fig. 6). MKN74 cells exhibited some fragmentation in control DNA; however, no ladder appearance indicative of nucleosomal fragmentation was observed and the control culture appeared quite healthy, so this was determined to be degradation after processing. transformed rat mesangial cell line, SV40 Mes 13, Sph caused apoptosis in SO% of cells and DMS in 40% (Fig. 7b). These results were confirmed by fragmentation studies of DNA samples prepared from primary cultured and transformed rat mesangial cells untreated or treated with 20 pM Sph, DMS or CzCer for 6 hr. No fragmentation was seen in control or treated normal cells (lanes 5-7). Control transformed cells (lane 1), as well as those treated with C2Cer (lane 2), also exhibited no fragmentation. However, ladder formation is apparent in transformed cells treated with Sph or DMS (lanes 3 and 4, respectively). Apoptosis induction in primary cultures and transformed cell lines To examine the effect of Sph on normal cell lines, HUVEC cultures at passages 5-7 were treated with sphingolipids and observed for evidence of apoptosis. Almost no apoptosis was seen after 6 hr treatment with any sphingolipid (Fig. 7a). After 18 hr, however, some apoptosis was apparent in DMS-treated cells (data not shown). Primary cultures of rat mesangial cells were also examined for sensitivity to sphingolipids. Cells were isolated and cultured as indicated in “Material and Methods” and examined morphologically and histochemically to confirm cell type (Mackazy et al., 1988). Cells were examined for evidence of apoptosis after treatment for 6 hr with 20 pM sphingolipids. Similar to results seen with HUVECs, no apoptosis was apparent in any of the treated cells (Fig. 7b). However, in a Inhibition of sphingolipid-induced apoptosis by protease inhibitors It is accepted that the apoptotic process involves a protease similar to those of the family known as ICE-like proteases. Inhibitors of these enzymes inhibit apoptosis (Kumar, 1995). We tested HL60 cells treated with 20 pM Sph or 20 pM DMS for 5.5 hr or 10 pM CzCer for 18 hr in the presence of protease inhibitors and analyzed them by flow cytometry. Figure 8a summarizes the results of experiments using 5 different protease inhibitors, tosyl-L-lysine chloromethyl ketone (TLCK), tosyl-L-phenylalanine chloromethyl ketone (TPCK), leupeptin (Leu), phenylmethyl fluoro phosphate (PMSF) and acetylleucyl-leycyl-normethionel (ALLnM). The cysteine protease inhibitor Leu at 25 pM inhibited apoptosis induced by the sphingolipids by a minimum of 20%. Concentrations of up to 100 pM for the less specific cysteine protease inhibitors 364 SWEENEY ETAL. ALLnM and PMSF failed to inhibit apoptosis. The serine protease inhibitor TPCK greatly inhibited apoptosis induced by all 3 agents, though DMS-induced apoptosis was only partially inhibited. TLCK, also a serine protease inhibitor, inhibited Sph-induced apoptosis completely and DMSinduced apoptosis by up to 70%, but at much higher concentrations. Figure 8b illustrates the ability of the 3 inhibitors to block apoptosis in a dose-dependent manner. C2Cer-induced apoptosis was the most sensitive to inhibition by the cysteine inhibitor Leu, with 50 PM blocking 69% of the apoptosis. The A 1 0SPN 20 uM loo e G n 80 - DMS 20 uM = c C, Cer 10pM 0 60m ._ 40 - c Q 2 20- ALLnM 100gM TLCK TPCK ImM 100pM Serine Protease Inhibitors Leu 25pM PMSF 100pM Cysteine Protease Inhibitors B Leupeptin 1oc 80 60 40 20 0 I I I I 0 10 20 30 I 40 I same concentration was able to inhibit DMS-induced apoptosis only by 53% and that of Sph-induced by 40%. However, TPCK was able to completely inhibit apoptosis induced by Sph at the lowest concentration tested, 25 yM, yet DMS was inhibited by 45% at o r above 50 p M ( p < 0.05). At 50 FM, TPCK inhibited C2Cer by 85% ( p < 0.01); this dropped to 60% at 25 yM. At its effective concentrations, TLCK inhibited DMS-induced apoptosis more effectively than TPCK, by up to 6 5 7 0 % ( p < 0.05). Sph-induced apoptosis was also inhibited by more than 90% ( p < 0.01). These results support the conclusion that these sphingolipids do indeed induce apoptosis and are possible messcngers in apoptotic signaling. Furthermore, their apoptotic signal is apparently up-stream from the protease(s) involved in apoptosis. The fact that CzCer is more sensitive to the cysteine protease inhibitor and Sph and DMS to the serine protease inhibitors also suggests that these 2 compounds are not affecting the pathway at the same point or in the same manner. DMS and Sph, however, responded similarly to serine protease inhibitors, yet DMS-induced apoptosis was never completely blocked. This, together with the results that DMS is nearly always more effective in inducing apoptosis, implies that DMS is the more potent inducer. Sphingolipid-induced apoptosis in the presence of fumonisin BI The conversion of Sph to Cer requires acylation of the compound by the enzyme Cer synthase. The mycotoxin fumonisin B1 has been shown to completely inhibit this synthase (Merrill et al., 1993), thereby blocking the formation of Cer. To confirm that Sph was not causing apoptosis by being converted to Cer, the induction of apoptosis by Sph was examined in the presence of fumonisin B1. HL60 cells were treated with 20 p M Sph or DMS in the presence of 25 y M fumonisin B1 for 5 hr and analyzed for apoptosis by flow cytometry. In Figure 9, the DNA distributions clearly demonstrate that apoptosis induced by Sph was not affected by the fumonisin B,. Fumonisin B1 itself had no effect on the cells, and the apoptosis induced by DMS, which cannot be acylated, was not affected by the mycotoxin. Thcse results indicate that Sph is not being converted to Cer prior to inducing apoptosis but is acting as the inducer itself, further supporting the role of Sph or DMS as messengers in apoptosis. DISCUSSION 50 In the study of apoptosis through receptors such as Fas and TNF, the sphingomyelin pathway has been implicated to be involved in signal transduction. The data presented here 100 80 60 40 20 0 50 0 100 150 200 100 1. TLCK 80 / 60 40 20 0 J 0 I I I I I 200 400 600 800 I000 Concentration (uM) FIGURE 8 - Inhibition of sphingolipid-inducedapoptosis by protease inhibitors. (a) HL60 cells were treated for 5.5 hr with 20 FM sphingosine (open bar) or DMS (hatched bar) or for 18 hr with 10 FM C2-ceramide(solid bar) in the presence of protease inhibitors as indicated and analyzed by flow cytometry. The cysteine protease inhibitor leupeptin inhibited sphingolipid-inducedapoptosis, especially C2-ceramide,but the less specific inhibitors ALLnM and PMSF did not. TPCK and TLCK, serine protease inhibitors, were able to completely block apoptosis induced by sphingosine and to partially block that induced by DMS. (b) HL60 cells were treated for 5.5 hr with 20 yM sphingosine (0)or DMS (W) or for 18 hr with 10 yM C2-ceramide (A)in the presence of the indicated concentrations of leupeptin, TPCK or TLCK, then analyzed by flow cytometry. Leupeptin inhibited C2-ceramide-inducedapoptosis in a dose-dependent manner but weakly inhibited apoptosis induced by sphingosine or DMS. Sphingosine- and C2-ceramideinduced apoptosis was completely inhibited by TPCK, yet DMSinduced apoptosis was only partially inhibited. TLCK effectively inhibited both sphingosine- and DMS-induced apoptosis. Treated cells in the absence of inhibitors were used as control, and the total apoptosis for each treatment was assigned a 100% value (*p < 0.01, +*p < 0.05). SPHINGOSINE AND DMS CAUSE APOPTOSIS Media SPh 365 DMS - m LL I z m' LL + DNA Content FIGURE 9 - Sphingolipid-inducedapoptosis in HL60 cells in the presence of fumonisin B1. HL60 cells grown in 10% FCS were treated for 5 hr with 20 pM sphingosine (b,e) or DMS (c,f) in the absence (a-c) or presence (d-f) of the ceramide synthase inhibitor fumonisin B1(25 p,M) and analyzed by flow cytometry. No inhibition of apoptosis was seen in the presence of the inhibitor. further support the evidence that this pathway, which is known to involve precursors and catabolites of Sph, plays an integral role in the mechanism of apoptosis. These results clearly show that Sph and DMS, in a dose- and time-dependent manner, cause apoptosis in some percentage of cells in all cancer cell lines tested under conditions in which similar concentrations of 3 Cer analogues, CzCer, C&er and C8Cer, were not able to cause apoptosis. In comparison, Cer analogues did cause apoptosis after 18 hr in hematopoietic cell lines grown in the absence of serum. Sphinganine, which differs from Sph only in the lack of the double bond, also caused apoptosis in the cells tested (data not shown). However, S-1-P, which is a next step in the metabolic pathway of Sph, did not cause apoptosis in most cells tested. Furthermore, the methylated derivative DMS, which is not further catabolized within the cell to either Cer or S-1-P, induced apoptosis in all cancer cells tested. It appeared to be a stronger inducer than Sph as it induced more apoptosis and was less susceptible to inhibition by protease inhibitors. Finally, the Cer synthase inhibitor fumonisin B1was not able to inhibit apoptosis induced by Sph. These data suggest that Sph and/or DMS may be a messenger in the apoptotic pathway distinct from, and in addition to, Cer. Our studies of Sph and DMS are performed in the presence of 10% FCS, the normal growth conditions for cell culture, and with the cells in logarithmic phase. Nearly all other studies published on exogenously added ceramides use cells abruptly removed from serum. In our experience, abrupt removal of serum and the growth factors it contains can cause cell death, even apoptosis, in a matter of hours, and it has even been reported to increase cellular Cer levels (Jayadev et al., 1995). Furthermore, Smith and Merrill (1995) have reported that adding fresh media (without serum) to cells causes significant changes in the levels of sphingolipids. To compare our results with those published, the hematopoietic cells used in this study were gradually adapted to grow in HL-1 medium. This medium contains essential growth factors but no serum and therefore eliminates some lipid-binding serum proteins. Overall, Sph and DMS studies did not differ greatly in cells grown in serum vs. those grown in HL-1, though the cells in the absence of serum are slightly more fragile. However, abrupt withdrawal of serum caused significant changes in all 3 cell lines, both morphological and in the cell cycle, within 4 hr (data not shown). Apoptosis assays are treated for a minimum of 2 hr, and 6 hr or more is the usual treatment time. As incubations for signal transduction assays can be as short as a minute, this dilemma might not always apply, yet for observing cell responses in culture for apoptosis, differentiation or prolifera- tion, this could be a crucial detail. In some cells, for example HUVECs, removal from serum for 6 hr causes arrest in Go or GI, with some apoptosis. It is probable, then, that the studies done on serum-starved cells represent mixed effects of treatment and withdrawal of growth factors. It is, in the least, an additional variable to be considered. The effect of FCS on Cer-induced apoptosis may simply reflect the presence of lipid-binding proteins on Cer analogues or their ability to cross the cell membrane since significantly higher concentrations of Cer can induce apoptosis even in the presence of serum (though longer treatment times do not increase apoptosis in the presence of serum). Yet, it is conceivable that removal of serum and growth factors changes the susceptibility of the cell, thus lowering some threshold for the cell to suicide. The idea of a threshold for apoptosis is considered by Fisher (1994), who discusses the concept that different apoptosis thresholds exist for different cell types, as in normal vs. malignant cells. Agents which cause apoptosis in malignant cells cause only reversible cell cycle arrest in normal cells, a fact exploited in many conventional anti-cancer therapies. Paradoxically, it appears that the same oncogenes that transform cells actually sensitize them to apoptosis induced by an exogenous agent, 2 of these genes being bcl-2 and c-myc, which have been implicated as targets for the pathways involving the SM pathway. Perhaps a disturbance in the natural cell damage inspection and repair-or-apoptosis pathway allows uncontrolled proliferation but confers greater susceptibility to apoptosis once triggered. The theory of differing thresholds is supported by the data presented here on HUVECs and on the normal and transformed mesangial cells. Normal cells were not as susceptible to apoptosis induced by sphingolipids as transformed cells or cancer cell lines, yet after longer treatments some apoptosis was apparent. Furthermore, we have evidence that Sph and DMS inhibit MAP kinase activity, which is known to be involved in many signaling pathways, and that the endogenous level of this enzymatic activity in malignant cells might correlate with susceptibility to apoptosis (data not shown). ICE-like proteases, including apopain/CPP32, have been implicated in the apoptosis pathway(s) in many cell systems (Kumar, 1995; Tewari et al., 1995). Specific inhibitors of these proteases are effectively able to block apoptosis induced by ligands binding to the receptors TNFRl or Fas. Apoptosis induced by sphingolipids can be inhibited by the serine protease inhibitors TPCK and TLCK and, to some degree, by the peptide inhibitor of cysteine proteases Leu, but not by the 366 SWEENEY ETAL. less specific ALLnM or PMSF. This supports our theory that Sph and DMS elicit a physiological signal on the cells and that one or both may be signal messengers in apoptosis. It further suggests that this messenger acts up-stream in the pathway from one or more cellular proteases. It is also significant that DMS-induced apoptosis is less inhibited than that of Sph by both TLCK and TPCK. DMS may act more quickly or bind more tightly with its target than Sph. It is unclear at present, however, whether DMS and Sph are acting as different messengers or if DMS is merely mimicking Sph. Finally, addition of Sph has been shown to elicit similar cellular responses as Cer, including regulation of protein kinases (Pushkareva et al., 1992). Spiegel and co-workers have also shown that Sph and S-1-P induce large increases in Ca2+ concentrations in cells (Zhang et al., 1990). Our studies using HL60 cells and neutrophils show that TNF induces apoptosis and raises cellular Cer levels and, interestingly, cellular Sph levels after the rise in Cer (Ohta et al., 1994). Okazaki et al. (1995) have reported differences in c-jun message induction in response to Sph and Cer. That evidence, together with the data reported here, supports the theory that Sph and/or DMS may act as a messenger in the apoptosis pathway distinct from Cer and that more than one pathway may be involved. ACKNOWLEDGEMENTS This work was supported in part by funds from The Biomembrane Institute, in part under a research contract with Otsuka Pharmaceutical Co. and Seikagaku Co. and by National Institute of Health Outstanding Investigator Grant CA-42505 (to S. Hakomori). REFERENCES C.M., KAROLAK, L.A. and HANNUN, Y.A., CIFONE,M.G., Dk MARIA,R., RONCAIOLI, P., RIPPO,M.R., AZUMA, OBEID,L.M., LINARDIC, M., LANIER,L.L., SANTONI, A. and TESTI,R., Apoptotic signaling Programmed cell death induced by ceramide. Science, 259,1769-1771 (1993). through CD9S (FasiApo-1) activates an acidic sphingomyelinase. J. exp. Med., 177,1547-1552 (1994). S. and IGARAOHTA,H., SWEENEY, E.A., MASAMUNE, A,, HAKOMORI, DBAIBO,G.S., OBEID,L.M. and HANNUN, Y.A., Tumor necrosis SHI,Y., Induction of apoptosis by sphingosine in human leukemia factor-a (TNF-a) signal transduction through ceramide. J. biol. Chem., HL60 cells: a possible endogenous modulator of apoptotic DNA 268,17762-17766 (1993). fragmentation occurring during phorbol ester-induced differentiation. Cancer Res., 55,691-697 (1995). FISHER, D.E., Apoptosis in cancer therapy: crossing the threshold. Cell, 78,539-542 (1994). OHTA,H., Y A T O M I , ~ SWEENEY, ., E.A., MASAMUNE, A,, HAKOMORI, S. HAKOMORI, S . and IGARASHI, Y., Glycosphingolipids and sphingolipids and IGARASHI, Y., A possible role of sphingosine in induction of closely associated with or causing apoptosis. Actu Histochem. Cyto- apoptosis by tumor necrosis factor-a in human neutrophils. FEBS chem., 28,77-82 (1995). Lett., 355,267-270 (1994). HANNUN, Y.A., The sphingomyelin cycle and the second messenger OKAZAKI, T., SAWAI,H., TASHIMA, M., SAWADA, H. and DOMAE,N., function of ceramide. J. biol. Chem., 269,3125-3128 (1995). The difference of signal transduction between ceramide and sphingoIGARASHI, Y., HAKOMORI, S., TOYOKUNI, T., DEAN,B., FUJITA,S . , sine-induced apoptosis in HL-60 cells [Abstract]. GlycoconjuguteJ., 12, SUGIMOTO, M., OGAWA, T., EL-GHENDY, K. and RACKER, E., Effect of 579 (1995). chemically well-defined sphingosine and its N-methyl derivatives on PUSHKAREVA, M.Y., KHAN,W.A., ALLESSENKO, A.V., SAHYOUN, N. protein kinase C and src kinase activities. Biochemistry, 28, 6796-6800 and HANNUN, Y.A., Sphingosine activation of protein kinases in Jurkat (1989). JARVIS,W.D., KOLESNICK, R.N., FORNARI,F.A., TRAYLOR, R.S., T cells: in vitro phosphorylation of endogenous protein substrates and GEWIRTZ, D.A. and GRANT,S., Induction of apoptotic DNA damage specificity of action. J. biol. Chem., 267,15246-15251 (1992). and cell death by activation of the sphingomyelin pathway. Proc. nut. RUAN,F., SADAHIRA, Y., HAKOMORI, S. and IGARASHI, Y., Chemical Acad. Sci. (Wash.), 91,73-77 (1994). and its inhibitory efsynthesis of D-eryrro-sphingosine-1-phosphate, JAYADEV, S., LIU, B., BIELAWSKA, A.E., LEE, J.Y., NAZAIRE,F., fect on cell motility. Biuorg. med. chem. Lett., 2,973-978 (1992). PUSHKAREVA, M.Y., OBEID, L.M., and HANNUN, Y.A., Role for SATO,T. and 13 OTHERS, Establishment of a human leukaemic cell line ceramide in cell cycle arrest. J. biol. Chem., 270,2047-2052 (1995). (CMK) with megakaryocytic characteristics from a Down’s syndrome KOLESNICK, R.N. and GOLDE,D.W., The sphingornyelin pathway in patient with acute megakaryoblastic leukaemia. Brit. J. Huematol., 72, tumor necrosis factor and interleukin-1 signaling. Cell, 77, 325-328 184-190 (1989). (1994). SMITH,E.R. and MERRILL,A.H., Differential roles of de novo S . , ICE-like proteases in apoptosis. Trends biochem. Sci., 20, sphingolipid biosynthesis and turnover in the “burst” of free sphingoKUMAR, 198-203 (1995). sine and sphinganine and their 1-phosphates and N-acyl derivatives, C.A., BRINSTER, that occurs upon changing the medium of cells in culture. J. biol. MACKAZY,K., STRIKER,L.J., ELLIOI.,S., PINKERT, E., Glomerular epithelial, mesangial, and endothe- Chem., 270,18749-18758 (1995). R.L. and STRIKER, lial cell lines from transgenic mice. Kidney Znt., 33,677-684 (1988). TEWARI, M., QUAN,L.T., O’ROURKE, K., DESNOYERS, S., ZENG,Z., MARTIN,S.J., GREEN,D.R. and COTTER,T.G., Dicing with death: BEIDLER,D.R., POIRIER,G.G., SALVESEN, G.S. and DIXIT,V.M., dissecting the components of the apoptosis machinery. Trends bio- Yama/CPP32p, a mammalian homolog of CED-3, is a crmAchem. Sci., 19,26-30 (1994). inhibitable protease that cleaves the death substrate poly(ADPMERRILL,A.H., VAN ECHTEN,G., WANG,E. and SANDHOFF, K., ribose) polymerase. Cdl, 81,801-809 (1995). Fumonisin B, inhibits sphingosine (sphingdnine) N-acyltransferase VUNNAM, R.R. and RADIN,N.S., Short chain ceramides as substrates and novo de sphingolipid biosynthesis in cultured neurons in sifu. J. for glucocerebroside synthetase. Differences between liver and brain biol. Chem., 268,27299-27306 (1993). enzymes. Biochim. biophys. Actu, 573,7342 (1979). NrcoLETTl, I., MIGIJORATI, G., PALIACCI, M.C., GRIGNANI, F. and RICCARDI, C., A rapid and simple method for measuring thymocyte ZHANG,H., BUCKLEY, N.E., GIBSON, K. and SPIEGEL, S., Sphingosine apoptosis by propidium iodide staining and flow cytometry. J. immunol. stimulates cellular proliferation via a protein kinase C-independent Methods, 139,271-279 (1991). pathway. J. biol. Chem., 265,76-81 (1990).