Induction of apoptosis and potentiation of ceramide-mediated cytotoxicity by sphingoid bases in human myeloid leukemia cells.

Prior studies demonstrated that ceramide promotes apoptotic cell death in the human myeloid leukemia cell lines HL-60 and U937 (Jarvis, W. D., Kolesnick, R. N., Fornari, F. A., Jr., Traylor, R. S., Gewirtz, D. A., and Grant, S. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 73-77), and that this lethal process is potently suppressed by diglyceride (Jarvis, W. D., Fornari, F. A., Jr., Browning, J. L., Gewirtz, D. A., Kolesnick, R. N., and Grant, S. (1994) J. Biol. Chem. 269, 31685-31692). The present findings document the intrinsic ability of sphingoid bases to induce apoptosis in HL-60 and U937 cells. Exposure to either sphingosine or sphinganine (0. 001 10 microM) for 6 h promoted apoptotic degradation of genomic DNA as indicated by (a) electrophoretic resolution of 50-kilobase pair DNA loop fragments and 0.2-1.2-kilobase pair DNA fragment ladders on agarose gels, and (b) spectrofluorophotometric determination of the formation and release of double-stranded fragments and corresponding loss of integrity of bulk DNA. DNA damage correlated directly with reduced cloning efficiency and was associated with the appearance of apoptotic cytoarchitectural traits. At sublethal concentrations (</=750 nM), however, sphingoid bases synergistically augmented the apoptotic capacity of ceramide (10 microM), producing both a leftward shift in the ceramide concentration-response profile and a pronounced increase in the response to maximally effective levels of ceramide. Thus, sphingosine and sphinganine increased both the potency and efficacy of ceramide. The apoptotic capacity of bacterial sphingomyelinase (50 milliunits/ml) was similarly enhanced by either (a) acute co-exposure to highly selective pharmacological inhibitors of protein kinase C such as calphostin C and chelerythrine or (b) chronic pre-exposure to the non-tumor-promoting protein kinase C activator bryostatin 1, which completely down-modulated total assayable protein kinase C activity. These findings demonstrate that inhibition of protein kinase C by physiological or pharmacological agents potentiates the lethal actions of ceramide in human leukemia cells, providing further support for the emerging concept of a cytoprotective function of the protein kinase C isoenzyme family in the regulation of leukemic cell survival.

Recent investigation has examined the participation of sphingophospholipid-and glycerophospholipid-derived messengers in the regulation of leukemic cell survival. We (1,2) and others (3) have demonstrated that increased intracellular availability of ceramide induces programmed cell death or apoptosis in the human myeloid leukemia cell lines HL-60 and U937. Ceramide interacts with at least two distinct intracellular target enzymes, ceramide-activated protein kinase (4 -6) and ceramide-activated protein phosphatase (7)(8)(9). A cytotoxic role for ceramide-activated protein phosphatase and ceramideactivated protein kinase in ceramide action has been inferred, although the relative contributions of these enzymes to the initiation of apoptosis is presently uncertain (10,11). A contrasting cytoprotective function of diglyceride and, therefore, of one or more isoforms of protein kinase C (PKC) 1 is supported by several lines of evidence. Increased intracellular availability of diglyceride abrogates the initiation of apoptotic DNA damage by ceramide in both HL-60 and U937 cells (1,2); this effect is mimicked by such diverse pharmacological PKC activators as the stage 1 tumor promoters phorbol dibutyrate (2) and phorbol myristate acetate (2,3), the stage 2 tumor promoter mezerein (2), and the non-tumor-promoting macrocyclic lactone bryostatin 1 (2). Collectively, these findings have defined opposing cytotoxic and cytoprotective roles for ceramide and diglyceride and, by extension, for their respective target enzymes in the regulation of leukemic cell survival.
In further support of a central cytoprotective function for PKC, we have also described the induction of apoptosis in HL-60 cells by pharmacological agents that selectively inhibit activity of this isoenzyme family (e.g. calphostin C and chelerythrine) (12). The importance of sphingoid bases such as trans-4-sphingenine (sphingosine) and 4,5-dihydrosphingosine (sphinganine) as physiologically relevant inhibitors of PKC is well established (13). In addition, the cytotoxic properties of sphingoid bases and other, more complex, lysosphingolipids have been linked directly to inhibition of PKC (14). Both sphinganine and sphingosine have been shown to reduce proliferative capacity and long term viability in HL-60 cells (15). Ohta and co-workers recently examined the lethal actions of sphingosine within the context of cellular maturation and proposed that endogenous sphingosine mediates apoptotic cell death following phorboid-induced terminal differentiation in HL-60 cells (16). Apart from those studies, however, little information is presently available concerning the apoptotic influences of sphingoid bases in human leukemia cells.
The present report describes biochemical characterizations of direct and indirect apoptotic properties of sphingoid bases in undifferentiated HL-60 cells. These findings demonstrate that acute exposure to sphingosine and other sphingoid bases potently elicits apoptosis as assessed by multiple criteria, including the induction of double-stranded DNA damage, loss of clonogenic potential, and appearance of apoptotic morphology. These results additionally reveal that co-exposure to either sphingoid bases or selective pharmacological PKC inhibitors at sublethal concentrations augments the apoptotic capacity of the lethal lipid messenger ceramide. This interaction is mechanistically consistent with our previous observations that, conversely, ceramide-mediated cell death is suppressed by diglyceride and pharmacological PKC activators (1,2). Thus, it appears that the apoptotic response to ceramide is indirectly regulated by the combined actions of sphingosine and diglyceride, which respectively limit or extend the cytoprotective influence of PKC. Based upon these observations, we propose that the reciprocal influences of sphingoid bases and diglyceride on PKC coordinately modulate ceramide-mediated apoptosis in human myeloid leukemia cells.

Drugs and Reagents
Synthetic preparations of D-erythro-sphingosine and D-erythrosphinganine were obtained from Biomol Research Laboratories, Inc. (Plymouth Meeting, PA). Other sphingosine derivatives (e.g. 3-keto-Derythro-sphingosine, N,N-dimethylsphingosine), and synthetic shortchain preparations of ceramide (N-octanoylsphingosine) and dihydroceramide (N-octanoylsphinganine) were also obtained from Biomol. Synthetic diglyceride analogs (1,2-dioctanoyl-sn-glycerol, 2,3-dioctanoyl-sn-glycerol, and 1,3-dioctanoyl-rac-glycerol) were obtained from Sigma. All lipids were initially dissolved in 100% ethanol and stored at Ϫ70°C. For experimental use, concentrated ethanol stocks of various sphingolipids were complexed at a 1:1 molar ratio with delipidated bovine serum albumin (fraction V; 2 mM in PBS) by vigorous mixing for 90 min at 37°C; stable protein-bound sphingolipid preparations were stored at Ϫ20°C. In contrast, glycerolipids were used directly as concentrated stocks in 100% ethanol. Bacterial preparations of sphingomyelinase (SMase; from Staphylococcus aureus) in a vehicle of 50% glycerol, 0.25 M Na 2 HPO 4 , pH 7.5, were obtained from Sigma or from Biomol and stored at 4°C. The selective PKC inhibitors calphostin C and chelerythrine (LC Services Corporation, Woburn, MA) were dissolved in sterile water, and stored at 4°C. The mycotoxin fumonisin B 1 (Sigma) was dissolved in 100% ethanol immediately before use. Bryostatin 1 was obtained in lyophilized preparations from (a) Dr. George R. Pettit (Arizona State University, Tempe, AZ) or from (b) the Cancer Treatment Evaluation Program of the National Cancer Institute; bryostatin 1 was dissolved in sterile Me 2 SO and stored at Ϫ20°C. All test reagents were diluted to final concentrations in complete medium at 37°C; the vehicles used were without discernible effect in HL-60 and U937 cells.

Preparation of Sphingosine Stereoisomers
Various sphingosine stereoisomers were synthesized and purified as described previously (17)(18)(19). D-erythro-sphingosine and L-erythrosphingosine were prepared, respectively, from tert-butoxycarbonyl-Lserine and tert-butoxycarbonyl-D-serine via coupling of the Garner aldehyde (17) with lithium pentadecyne in hexamethylphosphoramidetetrahydrofuran as described by Herold (18), followed by Birch reduction in lithium-ethylamine as described by Garner (19). D-and L-threo-isomers were prepared in a similar manner, with the exception that coupling reactions were performed with lithium pentadecyne in the presence of zinc bromide in diethyl ether (18). Stereoisomers were characterized as N-biphenyl-carboxamido derivatives of sphingosine by high performance liquid chromatography on the basis of elution from a chiral column in hexane/2-propanol (8:2, v/v).

Cell Culture
The human promyelocytic leukemic cell line HL-60 was derived from a patient with acute promyelocytic leukemia (20,21). The human monoblastic leukemia cell line U937 was derived from a patient with diffuse histiocytic lymphoma (22). Both cell lines were grown in complete RPMI 1640 medium (phenol red-free formulation, supplemented with 1.0% sodium pyruvate, non-essential amino acids, L-glutamine, penicillin, and streptomycin (all from Life Technologies, Inc.) and 10% heat-inactivated fetal bovine serum (HyClone Laboratories, Logan, UT). HL-60 and U937 cultures were passed twice weekly, both exhibiting characteristic doubling times of ϳ24 h. Cultures were maintained under a fully humidified atmosphere of 95% room air, 5% CO 2 at 37°C. Cell densities were determined by Coulter counter, and basal cell viability was assessed by trypan blue exclusion.

Test Exposures
All experimental incubations were performed as described previously (1,2,12). Cells in log-phase growth were pelleted, washed twice in complete medium, resuspended at a density of 4.25 ϫ 10 5 cells/ml), and maintained as indicated above; in view of the well established sensitivity of sphingoid base actions to surface dilution phenomena (13), cell density was carefully controlled in all test incubations. Cells were exposed to test agents for appropriate intervals in complete medium; loss of cells under these conditions due to either washing or cell adherence was negligible (Յ5%). Test incubations were terminated with gentle pelleting of the cells by centrifugation at 400 ϫ g for 10 min at 4°C; in most instances, aliquots of the medium were retained for direct assay of released DNA fragments. Following determination of cell density, the cells were pelleted and prepared as outlined below for agarose gel electrophoresis, spectrofluorophotometric assays of DNA fragments and DNA strand breaks, assay of cloning efficiency, or examination of cellular morphology.

Qualitative Analyses of DNA Damage
To assess both early and late aspects of ceramide-related DNA degradation, apoptotic DNA fragments of greatly differing sizes were resolved electrophoretically in parallel studies on both pulsed-field and static-field agarose gels as follows.
Pulsed-field Gel Electrophoresis-The formation of rosette (ϳ300kbp) and loop (ϳ50-kbp) DNA fragments was assessed by field-inversion gel electrophoresis as described previously (2). Pelleted cells were resuspended in PBS and mixed with molten 1.0% low melting-point agarose (In-Cert; FMC Corp. Bioproducts), yielding a final concentration of 2 ϫ 10 7 cells/ml; fractions of these mixtures (corresponding to ϳ2 ϫ 10 6 cells) were cast into precooled 85-l block molds and allowed to solidify at 4°C. The agarose-imbedded lysates were then treated with 250 mM EGTA, 250 mM EDTA, 1% N-lauroylsarcosine, pH 8.0 containing proteinase-K (200 g/ml; Sigma) at 55°C for 48 h. Deproteinated lysate plugs were rinsed in 250 mM EDTA, 250 mM EGTA, pH 8.0, and imbedded into 2.25% agarose gels (Sea-Kem Gold; FMC); high molecular weight DNA fragments were resolved by field-inversion electrophoresis at 6 V/cm for 24 -28 h in 0.5 ϫ Tris borate/EGTA buffer at 14°C; pulse intervals were ramped from t 1 ϭ 0.5 s to t 2 ϭ 50.0 s, with an F/R ratio of 3.0. Gels were stained for 6 h in 0.5 ϫ Tris borate/EGTA buffer containing ethidium bromide (0.5 g/ml), and DNA fragments were visualized by UV transillumination. DNA molecular weight reference preparations (48.6-kbp ladder; Life Technologies, Inc.) routinely were run in parallel to facilitate estimation of the size of rosette and loop DNA fragments.
Static-field Gel Electrophoresis-The formation of oligonucleosomal DNA fragments (ϳ0.2-1.2 kbp) was assessed by conventional agarose gel electrophoresis as described previously (2). Pelleted cells were resuspended in PBS and lysed by addition of 10 mM Tris-HCl, 15 mM EGTA, 15 mM EDTA, 0.1% Nonidet P-40, pH 7.4, yielding a final concentration of 4 ϫ 10 7 cells/ml; the lysates were then treated with proteinase-K (200 g/ml; Sigma) at 55°C for 24 h. The deproteinated extracts were centrifuged at 45,000 ϫ g for 75 min at 4°C, and the pellets were discarded; the supernatants were subsequently treated with ribonuclease-A (100 g/ml; Sigma) at 37°C for 18 h. Aliquots of final lysate preparations (corresponding to 2 ϫ 10 6 cells) were loaded into 2.25% agarose gels (Metaphor; FMC) impregnated with ethidium bromide (0.5 g/ml); low molecular weight DNA fragments were resolved by electrophoresis at 6 V/cm for 90 -180 min in 1 ϫ Tris acetate/ EGTA buffer at 10°C. DNA fragments were visualized by UV transillumination. DNA molecular weight reference preparations (100-bp ladder; Life Technologies, Inc.) were run in parallel to facilitate estimation of the size of oligonucleosomal DNA fragments.

Quantitative Analyses of DNA Damage
The formation and release of DNA fragments, and the corresponding breakage of bulk DNA were assessed as described previously (1,2). To measure intracellular DNA fragments, pelleted cells (4 ϫ 10 6 cells/ pellet in quadruplicate) were resuspended in PBS and lysed by addition of 5 mM Tris-HCl, 30 mM EGTA, 30 mM EDTA, 0.1% Triton X-100 (fully reduced), pH 8.0 (yielding a final density of 10 7 cells/ml), with gentle mechanical agitation. The lysates were centrifuged at 45,000 ϫ g at 4°C for 40 min; to measure extracellular DNA fragments, aliquots of incubation medium were adjusted to 5 mM Tris-HCl, 30 mM EGTA, 30 mM EDTA, pH 8.0, and centrifuged at 20,000 ϫ g at 4°C for 40 min. The pellets were discarded, and the presence of non-sedimenting DNA fragments in the supernatant from lysate and medium extracts was determined by dilution in modified Tris-sodium/EGTA buffer (3 mM NaCl, 10 mM Tris-HCl, 1 mM EGTA, pH 8.0) containing 1.0 g/ml bis-benzimide trihydrochloride (Hoechst 33258; Sigma), and monitoring net fluorescence in each sample ( ex ϭ 365, em ϭ 460). Final DNA values were calculated relative to highly purified calf thymus DNA calibration standard; values for all such responses are uniformly expressed as nanograms/micrograms DNA recovered or released from 10 6 cells, and reflect the absolute amount of non-sedimenting, low molecular weight fragments of DNA present in lysate and medium preparations. Corresponding loss of integrity of bulk DNA was determined by enhancedfluorescence alkaline unwinding analysis as described previously (1,2). Pelleted cells (8.25 ϫ 10 6 cells/pellet in quadruplicate) were resuspended in cold PBS and subjected to timed alkaline denaturation in 0.1 N NaOH; denaturation was terminated by neutralization in 0.1 N HCl. Cells were then further diluted in PBS and lysed by addition of 200 mM K 2 HPO 4 , 50 mM EDTA, 0.16% N-lauroylsarcosine with brief sonication. Damage to bulk DNA in cell lysates was quantified by spectrofluorophotometry in the presence of Hoechst 33258 ( ex ϭ 350, em ϭ 450); induction of strand breaks was demonstrated by reduction of net DNA fluorescence. Values were standardized against graded DNA strandbreakage induced by scaled irradiation from a [ 137 Cs] point source (30 -3000 rads), and are expressed as rad-equivalents.

Clonogenic Assay
Because we have found that HL-60 cells resist uptake of trypan blue even in advanced stages of apoptosis, the use of dye exclusion was precluded in these studies as a valid index of diminished viability; proliferative capacity was instead assessed in terms of clonogenic potential. Pelleted cells were washed extensively and prepared for softagar cloning as described previously (1,2,12). Cells were resuspended in cold PBS and seeded in 35-mm culture plates at a fixed density (400 cell/ml/well) in complete RPMI 1640 medium containing 20% fetal calf serum, 10% 5637-CM, and 0.3% Bacto agar. Cultures were maintained for 10 -12 days, and formation of colonies (defined as groups of Ն50 cells) was scored using an inverted microscope.

Cytology
Pelleted cells were resuspended in PBS, fixed in conventional cytocentrifuge preparations, stained with 20% Wright-Giemsa stain, and reviewed by light microscopy. The occurrence and mode of cell death in each treatment group were determined based on morphological criteria outlined previously (1, 2, 12). At least 3 fields of 100 cells each were scored for each treatment by assessing the expression of cytoarchitectural characteristics of either apoptosis (cell shrinkage, condensation of nucleoplasm and cytoplasm, formation of membrane blebs and apoptotic bodies) or necrosis (cell swelling, nuclear expansion, deterioration of organellar membranes, gross cytolysis).

In Vitro Assay of Total Cellular Protein Kinase C Activity
Assay of total protein kinase C activity in crude cell homogenates was performed as described previously (58). Briefly, preparations of cell lysates were transferred to acetylated filter discs and added to reactions mixtures containing lysis buffer (20 mM Tris-HCl, 500 M EDTA, 500 M EGTA, pH 7.5), synthetic phospholipid, phorbol 12-myristate 13acetate, and synthetic substrate (acetylated myelin basic protein Nterminal peptide AcMBP 4 -14 ). The reaction was initiated by the addition of 25 Ci of [␥-32 P]ATP, 20 M non-isotopic ATP, allowed to proceed for 5 min at 30°C, and terminated by addition of cold ortho-phosphoric acid (1%, v/v). The filters were washed and radioactivity determined by conventional liquid scintillometry.

Induction of Apoptosis by Sphingosine and Sphinganine-
Apoptotic cell death in HL-60 cells is characterized by loss of proliferative capacity, double-stranded degradation of genomic DNA, and profound alterations of cellular morphology. Highly selective pharmacological PKC inhibitors induce apoptosis in HL-60 cells (12), raising the possibility that sphingoid bases mediate similar lethal influences in the physiological regulation of cell death. The capacity of sphingoid bases to promote apoptotic cell death therefore was examined in these cells. Exposure of HL-60 cells to synthetic preparations of sphingosine or sphinganine at a fixed concentration (10 M) for 6 h potently induced apoptosis. Qualitative assessment of DNA damage on agarose gels demonstrated electrophoretic patterns of DNA fragments formed by internucleosomal hydrolysis of static chromatin in response to either lipid (Fig. 1); these included both high molecular weight loop fragments (appearing as single truncated bands of ϳ50 kbp on pulsed-field gels; Fig.  1A) and low molecular weight oligonucleosomal fragments (appearing as "ladders" of ϳ0.2-1.2 kbp on static-field gels; Fig.  1B). In related studies, quantitative assessment of DNA damage by spectrofluorophotometry demonstrated extensive degradation of genomic DNA in response to sphingosine and sphin- ganine (Fig. 2). Both lipids comparably promoted the formation and release of double-stranded DNA fragments ( Fig. 2A). Under basal conditions, such fragments were present in intracellular levels of Յ225 ng/10 6 cells and in extracellular levels of Յ55 ng/10 6 cells. Sphingosine (10 M) and sphinganine (10 M) increased the total accumulation of apoptotic DNA fragments respectively to 2206 Ϯ 280 ng/10 6 cells (p Ͻ 0.001) and 2070 Ϯ 158 ng/10 6 cells (p Ͻ 0.001). In addition, both lipids promoted extensive breakage of bulk DNA (Fig. 2B). Spontaneous breakage of bulk DNA was detected at a level of Յ155 rad equivalents. Sphingosine (10 M) and sphinganine (10 M) increased the extent of bulk DNA breakage, respectively, to 5387 Ϯ 195 rad equivalents and to 5009 Ϯ 145 rad equivalents. Sphingosine (10 M) and sphinganine (10 M) also substantially suppressed the proliferative capacity of HL-60 cells, such that colony formation was reduced by 87% and 82%, respectively (data not shown). In addition, while apoptotic morphology was discernible in Յ1.5% of vehicle-treated cells, both lipids elicited the expression of morphological features classically associated with apoptosis, including condensed nucleoplasm and cytoplasm, formation of membrane blebs and apoptotic bodies, fragmentation of the nucleus, and overall cell shrinkage. Exposure to sphingosine (10 M) or sphinganine (10 M) for 6 h increased the fraction of cells exhibiting apoptotic traits to 94% and 91%, respectively (Fig. 3). Comparable apoptotic responses to sphingosine and sphinganine were also observed in parallel studies with U937 cells (data not shown).
The apoptotic responses of HL-60 cells to sphingosine and sphinganine were equivalent, consistent with similar efficacies reported for these lipids with respect to inhibition of PKC (14). Conversely, the corresponding N-acyl derivatives ceramide and dihydroceramide differed markedly in apoptotic capacity (Table I), in that ceramide potently induced DNA fragmentation, whereas dihydroceramide was ineffective. Thus, while the effects of sphingosine have been attributed to conversion to ceramide in some settings (23), the identical responses to sphingosine and sphinganine indicates that the lethal actions of sphingoid bases do not reflect artifactual accumulation of ceramide. This was confirmed in related studies involving the mycotoxin fumonisin B 1 , which prevents N-acylation of sphingoid bases by inhibition of ceramide synthase (24). There was no evidence of apoptotic DNA damage following exposure of HL-60 cells to fumonisin B 1 (100 M) for 6 h; moreover, the extent of DNA fragmentation elicited by exposure to sphingosine (10 M) or sphinganine (10 M) for 6 h was not attenuated in the presence of fumonisin B 1 (Table I), confirming that sphingoid base-related cell death was not mediated by ceramide.
The apoptotic capacity of sphingosine did not exhibit stereospecificity (Table II), consistent with a specific involvement of PKC. Direct comparison of D-erythro-sphingosine with Lerythro-sphingosine and the corresponding enantiomer pair Lthreo-sphingosine and D-threo-sphingosine revealed similar efficacies with respect to induction of apoptotic DNA damage. Both the accumulation of DNA fragments and breakage of bulk DNA in response to each isomer were equivalent, although the L-threo isomer frequently exhibited a slightly higher efficacy for this response (ϳ15%). Structurally related sphingoid bases were also screened for potential apoptotic capacity in HL-60 cells (data not shown). For example, the methylated derivative N,N-dimethylsphingosine was somewhat more potent than sphingosine in the induction of apoptotic DNA damage (e.g. by 28%), whereas 3-ketosphingosine was essentially ineffective at promoting apoptosis.
Concentration-response Characteristics of Sphingosine-induced Apoptosis-The concentration-response characteristics of sphingosine action were determined in subsequent studies (Fig. 4). Exposure of HL-60 cells to sphingosine over a broad range of concentrations (0.001-100 M) for 6 h deceased clonogenicity and increased expression of apoptotic cytoarchitecture in a concentration-dependent manner (Fig. 4A). These responses were inversely correlated (R 2 ϭ 0.988). Both the loss of clonogenicity and the appearance of apoptotic morphology were evident at 1 M and maximal at 10 M, with respective EC 50  (Fig. 4B) and the corresponding breakage of bulk DNA (Fig. 4C), exhibited divergent concentration-response profiles. The concentration-response profile for sphingosine-induced DNA fragmentation was distinctly biphasic, reminiscent of the apoptotic responses to selective pharmacological PKC inhibitors reported previously (12). Significant (p Ͻ 0.01) accumulation of DNA fragments was discernible at 1 M and maximal at 10 M; above 10 M, the generation of DNA fragments declined progressively (to ϳ40% of the maximal fragmentation observed at 10 M). As the total number of cells recovered at the end of the exposure interval was not appreciably diminished, the apparent reduction in the extent of DNA damage associated with exposure to sphingosine at higher levels (i.e. Ͼ10 M) could not be attributed to release of DNA upon physical dissolution of dead cells. In contrast, the concentration-response profile for sphingosine-induced breakage of bulk DNA was linear, rather than biphasic. Significant (pϽ0.01) breakage of bulk DNA was detected at 0.1 M and maximal at 25 M; above 25 M, however, DNA breakage appeared to remain constant. The disparate concentration-response profiles for sphingosine provided by these separate assays suggested a fundamental change in the nature of DNA damage at higher sphingosine levels, a supposition that was confirmed in electrophoretic analyses of apoptotic DNA fragments (Fig. 5). Concentrationrelated changes in the appearance of apoptotic DNA fragments were demonstrated on both pulsed-field gels (Fig. 5A) and static-field gels (Fig. 5B). DNA loop fragments were initially observed at 1 M and persisted throughout the range of concentrations tested; these bands increased in intensity and assumed a more compact appearance as the sphingosine concentration was escalated to 10 and 100 M. In contrast, oligonucleosomal DNA fragment ladders were observed exclusively at 10 M, and were replaced by a very faint continuous streak of DNA at 100 M. Virtually identical electrophoretic profiles of apoptotic DNA fragments were obtained from HL-60 cells in response to sphinganine (not shown).

Potentiation of Ceramide-induced Apoptosis by Sphingoid Bases and Pharmacological Inhibitors of PKC-Previous inves-
tigations have demonstrated that ceramide mediates the induction of apoptotic cell death in HL-60 and U937 cells (1)(2)(3). In addition, we have shown that the apoptotic response to ceramide in these cells is attenuated or abolished by diglyceride and a variety of pharmacological PKC activators (1,2). To test the converse possibility that ceramide-related apoptosis is potentiated by sphingoid bases, additional studies were conducted to assess the apoptotic capacity of ceramide in the absence or presence of sphingosine and sphinganine (Fig. 6). Exposure of HL-60 cells to ceramide at a maximally effective concentration (10 M) for 6 h potently induced apoptotic DNA damage, increasing the net (i.e. intracellular and extracellular) accumulation of double-stranded DNA fragments to 1635 Ϯ 245 ng/10 6 cells and bulk DNA breakage to 3315 Ϯ 325 rad equivalents. Ceramide exposure also increased the fraction of cells expressing apoptotic traits to 36% (data not shown). Co-exposure to ceramide (10 M) and either sphingosine or sphinganine at a sublethal concentration (750 nM) significantly (p Ͻ 0.001) enhanced ceramide action, as reflected by both the net accumu-

TABLE II
Non-stereoselective induction of DNA damage by sphingosine HL-60 cells were exposed to synthetic preparations of D-erythrosphingosine (D-e-So), L-erythro-sphingosine (L-e-So), D-threo-sphingosine (D-t-So), L-threo-sphingosine (L-t-So) at equimolar concentrations (10 M lation of DNA fragments and the breakage of bulk DNA. In fact, the extent of ceramide-induced DNA damage was augmented by approximately 89 -96% in the presence of either sphingoid base. Moreover, sphingosine and sphinganine increased the fraction of cells exhibiting apoptotic morphology in response to ceramide to 64% and 61%, respectively (data not shown). Ceramide-induced apoptosis was comparably enhanced by sphingosine and sphinganine in U937 cells (data not shown). In other studies, HL-60 cells were exposed to ceramide over a broad range of concentrations (0.0001-100 M) in the absence or presence of sphingosine at a fixed (subeffective) concentration (750 nM) for 9 h (Fig. 7). As reported previously (1), ceramide produced a linear concentration-related increase in the accumulation of double-stranded DNA fragments. Sphingosine produced a distinct increase in the response to ceramide, consisting of both (a) a marked leftward shift in the ceramide concentration response profile and (b) a substantial increase in the effectiveness of maximal ceramide levels; this interaction thus appeared to entail increases in both the potency and efficacy of ceramide action. In other trials, stereochemical aspects of the reciprocal modulation of ceramide-mediated DNA degradation by sphingosine and diglyceride were examined. HL-60 cells were exposed to ceramide at a maximal concentra-tion (10 M) for 6 h in the absence or presence of various isomers of sphingosine (750 nM) or diacylglycerol (10 M). Consistent with the direct induction of apoptosis by sphingosine described above, potentiation of ceramide-related apoptosis by sphingosine was not stereoselective (Table III). When compared directly, both the D-and L-forms of erythro and threo enantiomers of sphingosine equivalently augmented the extent of DNA damage obtained following 6-h exposure to ceramide. As we have noted previously, however (2), the ability of diglyceride to limit ceramide-induced DNA damage was exclusively associated with the 1,2-sn-substituted isomer, whereas the 2,3sn-substituted and 1,3-rac-substituted species were ineffective (Table IV). These steric aspects of the reciprocal modulation of ceramide action by sphingosine and diglyceride are consistent with alterations in PKC activity.
Effects of Pharmacological Protein Kinase C Modulators on Ceramide-induced DNA Fragmentation-In other studies, potentiation of ceramide-induced apoptosis was also observed following treatment with bacterial sphingomyelinase (SMase; 0.001-100 milliunits/ml) by either (a) acute co-exposure to highly selective pharmacological PKC inhibitors such as calphostin C or chelerythrine at concentrations previously shown to be sub-effective in the induction of DNA fragmentation (12), or (b) following chronic pre-exposure to a non-differentiationinducing pharmacological PKC activator such as bryostatin 1 at a concentration sufficient to promote extensive down-modulation of assayable PKC activity (58) (Figs. 8 and 9). As we have described in previous reports (1,2), treatment of HL-60 cells with SMase induced the accumulation of double-stranded DNA fragments in a concentration-dependent manner.
The response to SMase was markedly augmented by calphostin C, which acts at the enzyme's regulatory domain (10 nM; Fig. 8A) or chelerythrine, which acts at the enzyme's catalytic site (1 M; Fig. 8B). Calphostin C and chelerythrine both produced marked leftward shifts in the concentration-response profile to SMase. The potentiative actions of these compounds differed in other respects, however, inasmuch as the magnitude of the response to SMase at maximal concentrations was significantly (p Ͻ 0.001) enhanced by calphostin C, but not by chelerythrine. Thus, whereas both agents increased SMase potency, only calphostin increased SMase efficacy.
Furthermore, the induction of DNA fragmentation by synthetic ceramide was sharply potentiated by chronic (i.e. 24 h) pre-exposure to the non-tumor-promoting PKC activator bryostatin 1 (250 nM; Fig. 9), enhancing the response to ceramide by 89%. These interactions were accompanied by extensive downmodulation of total assayable PKC activity in crude cell lysates  7. Potentiation of ceramide-induced apoptosis by sphingosine. HL-60 cells were exposed to ceramide (0.0001 to 100 M) in the absence (ࡗ) or presence (f) of sphingosine (750 nM) for 6 h. The total accumulation of apoptotic DNA fragments was then assessed by quantitative spectrofluorophotometry as before; values are expressed as micrograms of DNA/10 6 cells. Data shown are from a representative study performed four times with comparable results. All values reflect mean Ϯ S.E. of triplicate determinations.

TABLE III
Potentiation of ceramide-induced DNA damage by sphingosine HL-60 cells were exposed to ceramide (C 8 Cer; 5 M) for 6 h in the absence or presence of D-and L-forms of both erythro-sphingosine (D-e-So, L-e-So) and threo-sphingosine (D-t-So, L-t-So) at equimolar concentrations (750 nM). Apoptotic DNA degradation was quantified by spectrofluorophotometry. Values reflect mean Ϯ S.E. of triplicate determinations. Data shown are from a representative study performed three times with comparable results. ( Fig. 9, inset). PKC down-modulation was confirmed in parallel studies in which expression of cPKC␣, the predominant species of the enzyme present in HL-60 cells, was monitored by conventional Western analysis (data not shown); 24-h pre-exposure to bryostatin 1 virtually eliminated the presence of immunoreactive cPKC␣. Neither sphingosine nor sphinganine produced an additional augmentation of ceramide-related DNA damage in HL-60 cells down-modulated for PKC activity by bryostatin 1 pretreatment, however (data not shown), consistent with the position that PKC represents the primary subcellular target for sphingoid bases in the potentiation of ceramide action.

DISCUSSION
Sphingoid bases represent a versatile class of endogenous inhibitory effectors of the PKC isoenzyme family (13,14), and thus have been found to suppress or attenuate numerous PKCdependent aspects of leukemic cell survival. In HL-60 cells, sphingosine and sphinganine markedly limit proliferative ca-pacity and viability (15), and recent evidence has suggested that this response involves the induction of apoptosis (16). Monocytoid differentiation in HL-60 cells is sustained by PKC activity (reviewed in Ref. 25), a well defined process elicited by prolonged treatment with synthetic diglyceride (26), bacterial phospholipase C (27), or tumor-promoting phorboids (28 -32). These responses are potently antagonized by sphingoid bases. Induction of HL-60 cell differentiation by synthetic diglyceride is abolished by sphinganine (33). Phorboid-related maturation in these cells is similarly attenuated by both sphinganine (33,34) and sphingosine (35), as well as by such diverse pharmacological inhibitors of PKC as isoquinoline derivatives (e.g. H7) (36), fungal metabolites (e.g. staurosporine), and acylcarnitines (e.g. palmitoylcarnitine) (37). Moreover, terminal monocytoid differentiation of HL-60 cells ultimately culminates in apoptotic cell death (38,39). This process reportedly results from progressive, age-related increases in the intracellular availability of sphingosine, the apparent consequence of an augmented capacity to deacylate endogenous ceramide (16). Whether such alterations in sphingolipid metabolism represent an intrinsic feature of cellular maturation, or instead reflect a generalized feedback response to the sustained PKC activity necessary to support terminal differentiation, remains to be determined.
The present results demonstrate that sphingoid bases exert both direct and indirect apoptotic influences in myeloid leukemia cells. Acute exposure to sphingosine or sphinganine were found to (a) induce double-stranded degradation of genomic DNA, (b) suppress proliferative capacity, and (c) promote apoptotic cytoarchitectural changes. These findings are in agreement with qualitative characterizations of sphingosine-related apoptosis in HL-60 cells within the context of terminal differentiation described by Ohta and co-workers (16). The apoptotic actions of sphingosine and sphinganine exhibited essentially identical concentration-response profiles. A fundamental  change in DNA damage was noted at high sphingoid base concentrations (i.e. 10 -25 M), however. Specifically, whereas bulk chromatin was continuously cleaved into large (ϳ50-kbp) DNA fragments, subsequent degradation of this high molecular weight material into small (ϳ0.2-2.0-kbp) oligonucleosomal fragments was arrested. This phenomenon presumably reflects selective, concentration-related inhibition of the subtype(s) of apoptotic endonuclease responsible for internucleosomal hydrolysis of 50-kbp fragments. While such an underlying mechanism has yet to be demonstrated conclusively, this observation is consistent with reports suggesting that very early genomic lesions such as the initial breakage of static chromatin into high molecular weight (i.e. 300-and 50-kbp) DNA fragments are more central to the apoptotic process than the subsequent formation of low molecular weight DNA cleavage products (i.e. 0.2-2.0-kbp oligonucleosomal ladders) (40,41). Moreover, it is noteworthy that a similar concentration-dependent change in the character of apoptotic DNA damage has been previously documented in HL-60 cells following exposure to highly selective PKC inhibitors such as calphostin C and chelerythrine (12).
The intrinsic capacity of sphingoid bases to initiate apoptosis is directly consistent with the central cytoprotective role for the PKC isoenzyme family in the regulation of leukemic cell survival proposed in previous studies (1,2,12). Nonetheless, these findings must be interpreted with caution because the cellular concentrations of these lipids required for maximal inhibition of PKC activity may not be realized in living systems, an issue that has received comment from other investigators (14,42). The additional finding that sphingoid bases markedly potentiate the induction of apoptosis by ceramide when present at sublethal levels may therefore have considerable physiological significance. From a mechanistic standpoint, this interaction is consistent with the reciprocal ability of diglyceride to attenuate ceramide action that we have described previously (2). Taken together, these findings raise the possibility that inhibition of PKC activity by endogenous sphingoid bases contributes to the regulation of apoptosis, not by initiating cell death directly, but rather by sensitizing the intracellular signaling systems that govern cell survival to the actions of a primary lethal messenger such as ceramide. Susceptibility to the apoptotic influence of ceramide thus may represent a function of the relative intracellular availability of sphingoid bases and diradylglycerols. Studies designed to evaluate the impact of this concentration ratio on the apoptotic efficacy of ceramide are currently under way in our laboratory.
Although the PKC isoenzyme family represents a principal intracellular target for sphingoid bases (13,14), there is ample evidence to indicate that the bioeffector properties of these lipids may involve the modulation of additional regulatory systems. For example, recent investigation in other laboratories has demonstrated the existence of a novel family of sphingosine-activated protein kinases (43,44); these isoenzymes reportedly are (a) stimulated by sphingosine in a highly stereospecific manner (with a marked preference for the D-erythro species), but (b) completely insensitive to sphinganine. Similarly, pronounced stereoselectivity is also associated with other biological actions of sphingosine, including dephosphorylation of pRb (45,46), and the inhibition of the c-Src/v-Src protein kinases (47) and a variety of enzymatic activities that require calmodulin for optimal function (e.g. the multifunctional Ca 2ϩ -/calmodulin-dependent protein kinase) (48). Nonetheless, the stereoselectivity of sphingoid base action described in these studies is most consistent the established lipid sensitivity of PKC, strongly suggesting that the apoptotic properties of sphingosine and sphinganine derive from inhibition of PKC.
First, whereas only the D-erythro species occurs naturally in mammalian systems (49), the four isomers are equipotent in the inhibition PKC activity in vitro (35), and we observed a complete lack of stereoselectivity in the capacity of sphingosine to initiate apoptosis. Second, sphingosine and sphinganine are equivalent inhibitors of PKC (suggesting that the trans-4 double bond is not essential for inhibition of PKC activity by sphingoid bases) (35), and we noted essentially identical apoptotic responses to sphingosine and sphinganine. An analogous relationship has been noted with respect to the physiological activation of PKC by diglycerides, in that 1,2-diradyl-sn-glycerols are stimulatory, whereas 1,3-rac-substituted species are inactive (50,51). The application of such steric influences as criteria for implicating PKC in the mechanism of action of sphingoid bases and diradylglycerols also appears to be relevant in considering the modulation of ceramide action. Thus, the findings that the apoptotic capacity of ceramide was (a) comparably augmented by both D-and L-forms of erythrosphingosine and threo-sphingosine, but (b) selectively abolished by sn-1,2-substituted (but not sn-2,3-substituted or rac-1,3-substituted) forms of diglyceride additionally supports an involvement of PKC activity in the reciprocal modulation of ceramide action by sphingosine and diglyceride. Also consistent with an involvement of PKC in the apoptotic properties of sphingoid bases, down-modulation of PKC by chronic pre-exposure to bryostatin 1 potentiated ceramide-induced apoptosis to essentially the same extent as did acute inhibition of PKC by sphingoid bases. In this regard, it is significant that the potentiated response to ceramide noted in PKC-down-modulated cells could not be further augmented in the presence of sphingosine or sphinganine.
While the biological actions of sphingosine have been attributed, under some circumstances, to N-acylation of sphingosine to form ceramide via the ceramide synthase pathway (23), the cytotoxic properties of sphingosine described in this report are unlikely to stem from such a process. First, as already noted, sphingosine and sphinganine exhibited equivalent potency and efficacy in both the direct induction of apoptosis and the potentiation of ceramide-dependent cell death. Conversion of sphingosine and sphinganine (i.e. dihydrosphingosine) to the corresponding N-acylated derivatives (i.e. ceramide and dihydroceramide, respectively) yields metabolites with disparate biological actions because the established bioeffector properties of ceramide, including the capacity to induce apoptosis, reportedly are not associated with dihydroceramide (3,52). Second, and more significantly, both direct and indirect apoptotic influences of sphingosine were unaffected by the mycotoxin fumonisin B 1 . Because this toxin prevents the acylation of sphingosine to ceramide though competitive inhibition of ceramide synthase (Refs. 53 and 54; reviewed in Ref. 24), the actions of sphingosine described above more likely to reflect a direct action of sphingosine, rather than the artifactual accumulation of ceramide. Finally, it should be noted that, whereas a recent report describes transcriptional repression of multiple PKC isoforms in CV-1 monkey kidney cells following chronic treatment with fumonisin B 1 (55), we found no evidence that acute (i.e. 6-h) exposure to 100 M fumonisin B 1 induced apoptosis in HL-60 cells.
The capacity of sphingoid bases to induce apoptosis is consistent with previous findings from this and other laboratories demonstrating that diverse exogenous inhibitors of PKC alone initiate this process (12,56,57). These results are also compatible with other studies indicating that the apoptotic efficacy of the potent antileukemic agent 1-[␤-D-arabinofuranosyl]cytosine is augmented by manipulations that reduce cellular PKC activity, including both (a) down-modulation of PKC by chronic exposure to pharmacological PKC activators (58) and (b) inhibition of PKC by acute exposure to pharmacological PKC inhibitors (59). Furthermore, preliminary observations indicate that the ability of 1-[␤-D-arabinofuranosyl]cytosine to induce apoptosis in HL-60 cells is also subject to reciprocal modulation by diglyceride and sphingosine. 2 Collectively, these findings have potentially important implications for targeting PKC in the development of novel antileukemic strategies. Indeed, the potential utility of sphingoid bases as antineoplastic agents has been noted by other investigators (60). Antitumor actions of sphingosine and structurally related compounds have been documented in numerous cell types (reviewed in Ref. 61). For example, sphingosine and other sphingoid bases profoundly reduce tumor cell number in vitro (62) and restrict tumor growth and metastasis in vivo (63). Similarly, synthetic structural analogs of sphingoid bases (e.g. stearylamine) have been found to inhibit the activity of PKC in purified preparations (33), and to exert potent antitumor influences both in vitro (64) and in vivo (65). Furthermore, recent observations by Schwartz and co-workers indicate that safingol (referred to elsewhere as SPC-100270), a synthetic preparation of L-threosphinganine, potently limits the extent of tumor cell invasiveness (66) and substantially augments the antineoplastic actions of such diverse agents as doxorubicin and mitomycin (67). Whether these interactions stem from potentiation of tumor cell apoptosis remains to be established.
In conclusion, these observations demonstrate that sphingoid bases promote apoptotic cell death in human myeloid leukemia cells through both direct and indirect mechanisms. Within the context of physiological regulation of apoptosis, the potentiation of ceramide-induced cell death by sphingoid bases directly complements our previous observations that diglyceride opposes ceramide action. On the basis of these findings, therefore, it is proposed that (a) the regulation of leukemic cell survival depends upon a balance between ceramide-driven systems (e.g. ceramide-activated protein kinase) and PKC, and that (b) the cytoprotective influence of PKC is modulated by the reciprocal actions of sphingoid bases and diradylglycerols.