N-(4-Hydroxyphenyl)retinamide increases dihydroceramide and synergizes with dimethylsphingosine to enhance cancer cell killing

Fenretinide [N-(4-hydroxyphenyl)retinamide (4-HPR)] is cytotoxic in many cancer cell types. Studies have shown that elevation of ceramide species plays a role in 4-HPR cytotoxicity. To determine 4-HPR activity in a multidrug-resistant cancer cell line as well as to study ceramide metabolism, MCF-7/AdrR cells (redesignated NCI/ADR-RES) were treated with 4-HPR and sphingolipids were analyzed. TLC analysis of cells radiolabeled with [³H]palmitic acid showed that 4-HPR elicited a dose-responsive increase in radioactivity migrating in the ceramide region of the chromatogram and a decrease in cell viability. Results from liquid chromatography/electrospray tandem mass spectrometry revealed large elevations in dihydroceramides (N-acylsphinganines), but not desaturated ceramides, and large increases in complex dihydrosphingolipids (dihydrosphingomyelins, monohexosyldihydroceramides), sphinganine, and sphinganine 1-phosphate. To test the hypothesis that elevation of sphinganine participates in the cytotoxicity of 4-HPR, cells were treated with the sphingosine kinase inhibitor d-erythro-N,N-dimethylsphingosine (DMS), with and without 4-HPR. After 24 h, the 4-HPR/DMS combination caused a 9-fold increase in sphinganine that was sustained through +48 hours, decreased sphinganine 1-phosphate, and increased cytotoxicity. Increased dihydrosphingolipids and sphinganine were also found in HL-60 leukemia cells and HT-29 colon cancer cells treated with 4-HPR. The 4-HPR/DMS combination elicited increased apoptosis in all three cell lines. We propose that a mechanism of 4-HPR–induced cytotoxicity involves increases in dihydrosphingolipids, and that the synergy between 4-HPR and DMS is associated with large increases in cellular sphinganine. These studies suggest that enhanced clinical efficacy of 4-HPR may be realized through regimens containing agents that modulate sphingoid base metabolism. [Mol Cancer Ther 2008;7(9):2967–76]

One of the newer strategies for cancer treatment, as well as chemoprevention, is to modulate key signaling mediators and pathways to control the behavior of cancer cells. Sphingolipids are emerging as promising targets because this diverse family of compounds includes many subspecies (ceramides, sphingosine 1-phosphate, gangliosides, among other things) that regulate cell proliferation, differentiation, migration, and programmed cell death by both apoptotic and autophagic pathways (1–4). In addition, recent findings about the regulation of ceramide formation and turnover have opened new doors to understanding not only the basis for the toxic effects of sphingolipids on cancer cells but also the relationship between ceramide metabolism and multidrug resistance (5–7).

Ceramide species, the neutral lipid backbone of sphingolipids, have been known for some time to play a role in signaling apoptotic cell death (5, 8, 9), which is thought to be a major cytotoxic mechanism for many anticancer drugs (10, 11). Ceramides seem also to participate in cell death via other pathways such as autophagy (3, 4, 12). Important anticancer drugs such as anthracyclines, Vinca alkaloids, paclitaxel (Taxol), etoposide, and fenretinide, also referred to as N-(4-hydroxyphenyl)retinamide (4-HPR), have been reported to elevate ceramides by a variety of mechanisms. These include inducing ceramide synthesis de novo, activating turnover of sphingomyelin, and/or suppression of the conversion of ceramides into more complex sphingolipids as illustrated in Fig. 1 (reviewed in refs. 1, 4, 5).

Although ceramide produced for “signaling” involves the turnover of complex sphingolipids such as sphingomyelin, there is now a strong association between de novo ceramide biosynthesis and chemotherapy-induced apoptosis (1, 2, 13). The roles played by the enzymes that remove ceramides by metabolizing them into more complex sphingolipids were revealed during studies of cells refractory to chemotherapy (14–17) and led to the hypothesis that agents that modulate ceramide metabolism to sustain the apoptotic signal might improve the efficacy of chemotherapy (5). In this regard, blocking ceramide glycosylation with agents such as an inhibitor of glucosylceramide synthesis (1-phenyl-2-palmitoylamino-3-morpholino-1-propanol; Fig. 1) often enhances cytotoxicity (13–15, 18, 19). However, there is at least one exception that should be borne in mind: When ceramide is removed by hydrolysis to sphingosine and phosphorylated, this produces a compound, sphingosine 1-phosphate (S1P), which is associated with cell proliferation and survival (20).

In the present study, we describe findings with 4-HPR, an anticancer agent that has previously been shown to increase de novo sphingolipid biosynthesis by activation of serine palmitoyltransferase and dihydroceramide synthases (Fig. 1; ref. 21). 4-HPR imparts cytotoxic responses in many types of cancer cells (18) by increasing reactive oxygen species and ceramides and inducing apoptosis and/or mixed necrosis (22), although these may not be the only mechanisms of action of 4-HPR (23). By lipidomic analysis, using liquid chromatography/electrospray tandem mass spectrometry (LC MS/MS), of the subspecies of ceramides found in cells (i.e., variants in sphingoid base backbone and amide-linked fatty acids), as well as analysis of sphingolipids that are metabolically interconnected, we have determined that cells treated with 4-HPR displayed elevated dihydroceramide species rather than desaturated ceramides. In addition, there were substantial elevations in sphinganine and sphinganine 1-phosphate (Sa1-P). Due to the elevation in Sa1-P, which may be an antiapoptotic metabolite analogous to S1P, the combination of 4-HPR and the sphingosine kinase inhibitor d-erythro-N,N-dimethylsphingosine (DMS; Fig. 1) was tested. Experiments showed cytotoxic synergy when 4-HPR and DMS were combined at concentrations that were minimally cytotoxic as single agents. This cytotoxic synergy correlated with large increases in cellular sphinganine and dihydroceramides. These findings suggest that the efficiency of 4-HPR, and perhaps other agents that elevate dihydroceramides, may be enhanced by inhibiting sphingosine/sphinganine kinase activity to sustain elevation of sphingoid bases.

The ovarian carcinoma cell line MCF-7/AdrR, redesignated NCI/ADR-RES (24) and previously described as a breast cancer cell line, which is resistant to doxorubicin, was provided by Dr. Kenneth Cowan (University of Nebraska Medical Center Eppley Cancer Center, Omaha, NE) and Dr. Merrill Goldsmith (National Cancer Institute, Bethesda, MD). HT29, a human colon cancer cell line, and HL-60 cells, a human promyelocytic cell line, were obtained from the American Type Culture Collection. Cell culture media were from Invitrogen Corp., and characterized fetal bovine serum was from HyClone. C6-ceramide (N-hexanoylsphingosine), C6-dihydroceramide (N-hexanoyldihydro-sphingosine, d-erythro), DMS (d-erythro-N,N-dimethylsphingosine), sphinganine (d-erythro-dyhydrosphingosine), and myriocin were from BIOMOL. Myriocin was also purchased from Sigma Chemical and Calbiochem, and fumonisin B₁ (FB₁) was from Calbiochem. Ceramide and sphingomyelin (brain-derived) were from Avanti Polar Lipids, as was the internal standard cocktail for the mass spectrometric analysis. [9,10-³H(N)]Palmitic acid (50 Ci/mmol) was from Dupont/NEN. Silica Gel G TLC plates (prescored) were from Analtech. 4-HPR was kindly provided by R.W. Johnson Pharmaceuticals but was also purchased from Sigma. GT11 {N-C8:0-cyclopropenylceramide, N-[(1R,2S)-2-hydroxy-1-hydroxymethyl-2-(2-tridecyl-1-cyclopropenyl)ethyl]octanamide}, a dihydroceramide desaturase inhibitor, was purchased from Matreya.

Cell lines were cultured using conditions previously described. In brief, HL-60 (25) and MCF-7/AdrR cells (26) were cultured in RPMI 1640, and HT-29 cells were grown in DMEM (27). The media additionally contained 10% fetal bovine serum, 100 units/mL penicillin, 100 μg/mL streptomycin, and 584 mg/L l-glutamine. Cells were grown in a humidified 5% CO₂ tissue culture incubator at 37°C, and the adherent cell lines (all but HL-60) were released from the dishes for subculturing using 0.05% trypsin/0.53 mmol/L EDTA (Invitrogen).

Cells were radiolabeled in mid-log phase growth by addition of [³H]palmitic acid (1.0 μCi/mL medium) for the times specified in the figure legends. After removal of the culture medium, cells were washed with PBS and the lipids were recovered from the dish by addition of ice-cold 2% acetic acid in methanol, followed by scraping of the cells from the dish with a plastic scraper. The mixture was transferred to 1-dram glass vials with Teflon screw caps for lipid extraction by the method of Bligh and Dyer (28). After vortex mixing and centrifuging, the organic lower phase was removed and evaporated under a stream of nitrogen. The lipids were dissolved in chloroform/methanol (2:1), spotted on Silica Gel G TLC plates, and developed with a solvent system containing chloroform/acetic acid (90:10, v/v). A commercial ceramide standard was co-chromatographed and visualized with iodine vapor. The corresponding region of the TLC plate (composed of both [³H]dihydroceramides and [³H]ceramides) was scraped and transferred to scintillation vials with 0.5 mL of water and 4.5 mL of EcoLume. Quantitation of tritium was by liquid scintillation counting with correction for quenching (which was minimal).

The sphingolipids were extracted and analyzed by the method of Merrill et al. (29) as briefly described here. After transfer of the cells to 13 × 100 mm borosilicate test tubes, a cocktail of sphingolipid internal standards (C17-sphingosine, C17-sphinganine, C17-sphingosine 1-phosphate, and C17-sphinganine 1-phosphate, C12-ceramide, C12-ceramide 1-phosphate, C12-glucosylceramide, C12-lactosylceramide, and C12-sphingomyelin) from Avanti Polar Lipids was added, and the lipids were extracted and divided into two fractions that were applied to reverse-phase LC (for sphingoid bases) and normal-phase LC (for complex sphingolipids) coupled with electrospray MS/MS. The reverse-phase LC used a Supelco 2.1 mm i.d. × 5 cm Discovery C18 column and a flow rate of 1 mL/min. Mobile phase A consisted of CH₃OH/H₂O/HCOOH (58:41:1; all mobile phase solvents are given in v/v); mobile phase B consisted of CH₃OH/HCOOH (99:1); both also contained 5 mmol/L ammonium formate. For each analysis, the column was equilibrated with A/B (60:40) for 0.4 min; the sample was injected (50 μL by a Perkin-Elmer Series 200 Autosampler); A/B (60:40) was continued for 0.5 min, followed by a 1.8- min linear gradient to 100% B, which was held for 5.3 min; and then the column was reequilibrated at initial conditions for 0.5 min. The normal-phase LC used a Supelco 2.1 mm i.d. × 5 cm LC-NH₂ column and a flow rate of 1.0 mL/min. Mobile phase A consisted of CH₃CN/CH₃OH/CH₃COOH (97:2:1); mobile phase B consisted of CH₃OH/H₂O/CH₃(CH₂)₃OH/CH₃COOH (64:15:20:1); both also contained 5 mmol/L ammonium acetate. For each analysis, the column was equilibrated with A/B (98:2) for 0.5 min; the sample was injected; and A/B (98:2) was continued for 1.1 min, followed by a 0.2-min linear gradient to 82% A. This state is held for 0.4 min, followed by a 0.8-min linear gradient to 100% B and reequilibration of the column at initial conditions for 0.5 min.

The mass spectrometry data were collected using a PE Sciex ApI 3000 triple quadrupole mass spectrometer equipped with a turbo ion-spray source. Dry N₂ was used as the nebulizing gas at a flow rate of 6 L/min. The ionspray needle was held at 5,500 V, and the orifice and ring voltages were kept low (30–40 and 180–220 V, respectively) to minimize collisional decomposition of molecular ions before entry into the first quadrupole, and the N₂ drying gas temperature was set to 500°C. N₂ was used to collisionally induce dissociations in Q2, which was offset from Q1 by 30 to 40 V. Q3 was then set to pass molecularly distinctive product ions.

As the first step, each cell type was examined by product ion scans for m/z 264.4 and 266.4 through the (dihydro)ceramide and monohexosyl(dihydro)ceramide range and by precursor ion scans of m/z 184.4 for (dihydro)sphingomyelins to identify the major subspecies. With this information, a multiple reaction monitoring method has been developed for the fatty acyl chain length variants where each transition has a dwell time of ∼25 ms. After building the multiple reaction monitoring protocol, the intensities of the signals for the different sphingolipid subspecies can be compared with the internal standards to arrive at estimations of the mass that are usually within ±10%.

For the analysis of the MCF-7/AdrR cells, dishes (10 cm) of cells at mid-log phase growth were treated for 24 and 48 h with 4-HPR (5.0 μmol/L), DMS (5.0 μmol/L), 4-HPR plus DMS, or vehicle (ethanol) in medium containing 5% fetal bovine serum. After rinsing in ice-cold PBS, cells were harvested with trypsin-EDTA and dispersed in serum-free medium for counting by hemocytometer or analysis of protein amount, then transferred to 13 × 100 mm borosilicate glass tubes and frozen at −80°C until analysis. HT-29 and HL-60 cells were treated with 10 μmol/L 4-HPR (or vehicle) and collected after 24 h.

Cytotoxicity was determined after 72-h exposures using a fluorescence-based, digital imaging microscopy method (DIMSCAN) as described (18). Drug-induced cytotoxic synergy was analyzed by CalcuSyn software from Biosoft (18). By this method, a combination index equal to 1 indicates an additive effect, and combination indices of 0.7 to 0.85 and 0.3 to 0.7 indicate moderate synergism and synergism, respectively.

Apoptosis was evaluated on intact cultured cells after staining with SYTO-13 (Invitrogen) and propidium iodide. Cells were observed under an inverted fluorescent microscope (Olympus 1X70). With SYTO-13, normal nuclei exhibit a loose chromatin colored in green, and apoptotic nuclei exhibit condensed green-colored chromatin and/or fragmentation. Postapoptotic necrosis is denoted by condensed and/or fragmented nuclei, but orange colored.

Initial experiments assessed the effect of 4-HPR on the biosynthesis of total ceramides, including dihydroceramides, and consequent cytotoxicity. As shown in Fig. 2A, when the cells were incubated with increasing concentrations of 4-HPR, there was a dose-dependent increase in the levels of [³H](dihydro)ceramides generated. For example, [³H](dihydro)ceramides were 0.47 ± 0.09% of the total ³H-lipids in the control cells, whereas the percentage increased to 0.76 ± 0.1% (1.6-fold) and 1.27 ± 0.03% (2.7-fold) in cells treated with 5.0 and 10 μmol/L of 4-HPR, respectively. We next evaluated 4-HPR cytotoxicity and determined whether DMS, a known sphingosine kinase inhibitor (20), would enhance cytotoxicity. The data of Fig. 2B and C show that 4-HPR cytotoxicity was enhanced by the inclusion of DMS, particularly at higher 4-HPR concentrations that increased (dihydro)ceramides. In MCF-7/AdrR cells (Fig. 2B), with 3.0 μmol/L 4-HPR and 1.5 μmol/L DMS, cell killing was additive (combination index, 1.2); however, at higher concentrations (e.g., 6.0 μmol/L 4-HPR and 3.0 μmol/L DMS), cell killing was synergistic (combination index, 0.6). The addition of DMS, which was minimally cytotoxic at the concentrations tested, also enhanced the cytotoxicity of 4-HPR in HL-60 cells (Fig. 2C). For example, the addition of DMS (1.25 μmol/L) to 4-HPR (2.5 μmol/L) enhanced cell killing synergistically (combination index, 0.5). To determine what aspects of sphingolipid metabolism correlated with cell killing, we investigated pathways involved in ceramide generation and conducted detailed lipid analyses.

4-HPR has been shown to increase ceramides by the de novo pathway (18, 21, 30, 31). We confirmed this in MCF-7/AdrR cells by using myriocin and FB₁, inhibitors of early enzymes of sphingolipid biosynthesis, serine palmitoyltransferase and (dihydro)ceramide synthases, respectively. Either agent (FB₁, 50 μmol/L; myriocin, 0.25 μmol/L), when introduced simultaneously with 4-HPR and [³H]palmitate, blocked 4-HPR–induced [³H](dihydro)ceramide increase (data not shown). Next, we determined the influence of 4-HPR and DMS on sphingolipid metabolism using LC MS/MS methods for quantitative analysis. Total ceramides, without regard to molecular species, sphinganine, sphingosine, and S-1-P were measured (Table 1) after +24 and +48 hours of exposures. 4-HPR alone, compared with controls, enhanced total ceramide levels dose-responsively by 2.8- and 7.7-fold at 24 and 48 hours, respectively. The 4-HPR/DMS regimen, compared with controls, elevated total cellular ceramide levels by 4.4- and 17-fold at +24 and +48 hours, respectively. Sphinganine, an early product in the de novo pathway, increased additively on exposure to 4-HPR, DMS, and 4-HPR/DMS. For example, at +24 hours, 4-HPR, DMS, and 4-HPR/DMS increased sphinganine levels 6-, 2.3-, and 8.7-fold, respectively, and at +48 hours, increased sphinganine levels 3.0-, 1.6-, and 8.0-fold over controls, respectively. Importantly, the 4-HPR/DMS combination promoted a more than additive increase in sphinganine at +48 hours. The levels of sphingosine were not dramatically altered by 4-HPR treatment, whereas DMS diminished sphingosine levels by 8.7-fold (11% of control) and 10.3-fold (10% of control) at +24 and +48 hours, respectively. At +24 hours, 4-HPR treatment diminished S1P levels by ∼60%; however, by +48 hours levels were similar to controls. DMS treatment decreased relative S1P levels by ∼60% at +24 hours, but S1P returned to 80% of control levels at +48 hours. The 4-HPR/DMS regimen reduced S1P levels by 68% and 63% at +24 and +48 hours, respectively, compared with control.

We next determined the molecular subspecies of sphingolipids in MCF-7/AdrR cells treated with the 4-HPR/DMS combination. After first scanning the lipid extracts to determine the species that were present, a multiple reaction monitoring protocol was used to quantify these species by comparison with internal standards (29). Desaturated ceramides were not elevated by 4-HPR. Instead, there was notable elevation in dihydroceramides, particularly the 16- and 18-carbon chain species (Fig. 3A). This was accompanied by reductions in desaturated ceramide–containing sphingomyelin (Fig. 3B) and monohexosylceramides (Fig. 3C) and elevations in the dihydroceramide backbone versions of these complex sphingolipids. It may be noteworthy that the decreases in desaturated ceramide backbone–containing sphingomyelin were more than compensated for by increases in dihydrosphingomyelin, whereas the elevation in monohexosyl-dihydroceramides did not compensate for the decreases in monohexosylceramides (compare Fig. 3B and C). Another interesting change in the MCF-7/AdrR cells treated with 4-HPR was a marked increase in sphinganine and sphinganine 1-phosphate (Sa1P), but not sphingosine or S1P (Fig. 3D, front two rows). These findings are intriguing because sphinganine is known to be cytotoxic (32). We speculate that Sa1P might protect from cytotoxicity when generated endogenously, although it does not seem to be protective when added exogenously although it is recognized by at least some S1P receptors (33). To explore a possible role of these sphingoid bases in the cytotoxicity of 4-HPR, we characterized the effects of DMS on S1P and Sa1P (Fig. 3D, back two rows). In cells treated with DMS alone, there was a small decrease in S1P that was accompanied by decrease in sphingosine; sphinganine and Sa1P were little affected by DMS. However, in the presence of DMS, 4-HPR caused a nearly 3-fold greater increase in sphinganine (for an ∼9-fold increase versus the untreated control), whereas the Sa1P increase was blunted (Fig. 3D, rows 2 and 4). In summary, 4-HPR treatment alone caused increases in Sa1P > sphinganine and no increases in sphingosine or S1P, and in the presence of DMS, 4-HPR increased sphinganine > Sa1P and both sphingosine and S1P decreased.

The increase of cytotoxicity of the 4-HPR/DMS regimen in the multidrug resistant MCF-7/AdrR cells compared with single-agent treatment was striking (see Fig. 2B). This cytotoxicity increase seemed to correlate with sphinganine levels (Fig. 3D) but was not obviously associated with changes in the amounts and types of more complex sphingolipids in the MCF-7/AdrR cells, as these were not found to be substantially different in cells treated with 4-HPR alone versus 4-HPR/DMS. For example, analysis of (dihydro)ceramides, (dihydro)sphingomyelin, and monohexosyl(dihydro)ceramides by LC MS/MS showed little change in the amounts of these complex lipids in cells exposed to either DMS or 4-HPR/DMS, compared with 4-HPR (data not shown; experiment conducted as described in Fig. 3D). However, a limitation of the approach used is that experiments measured steady-state sphingolipid levels at +48 hours and not the flux of total sphingolipid mass through the various pathways.

We have recently reported that 4-HPR elevates dihydroceramide in DU-145 prostate cancer cells (4) and in MOLT-4 ALL leukemia cells (19); therefore, this may be a common cancer cell response to 4-HPR. To explore this further, the effects of 4-HPR were examined using two additional human cancer cell lines, HT-29 colon cancer cells and HL-60 human promyelocytic leukemia cells. As shown in Fig. 4, for both of these cell lines, 4-HPR treatment had the same general effects on sphingolipid composition as was seen in MCF-7/AdrR cells, notably dramatic increases in dihydroceramides (shown for C16-dihydroceramide in Fig. 4A and B), as well as the presence of the dihydroceramide backbone in sphingomyelins and monohexosyl-glycosphingolipids. Decreases in C16-ceramides and certain other chain length ceramides and substantial increases in sphinganine and Sa1P (Fig. 4A and B, insets) were also determined.

Sphinganine has been reported to be cytotoxic for many transformed cells (32), but to the best of our knowledge, this has not been tested previously with multidrug-resistant cancer cells. Consistent with other types of transformed cells, sphinganine was cytotoxic in MCF-7/AdrR cells (Fig. 5A). Sphinganine is converted to dihydroceramides by dihydroceramide synthases (see Fig. 1). Interestingly, the addition of GT11, a dihydroceramide desaturase inhibitor (34), which mimicked the effect of 4-HPR on dihydroceramide desaturase, enhanced sphinganine cytotoxicity (Fig. 5A). A similar experiment conducted in HL-60 cells showed that whereas these cells were largely refractory to sphinganine, both GT11 and DMS enhanced sphinganine cytotoxicity (Fig. 5B). Sphinganine combined with DMS at 6 and 4 μmol/L, respectively, induced 98% cell kill, and the combination was synergistic (combination index, 0.6). The combination of sphinganine + GT11 + DMS showed the highest toxicity in both cell lines.

We evaluated the influence of a six-carbon cell-permeable analogue of dihydroceramide on MCF-7/AdrR cells and compared it to C6-ceramide. Cells were exposed in 96-well plates for 72 hours. Whereas C6-ceramide reduced cell viability with an EC₅₀ of ∼13 μmol/L, no reduction in cell viability was observed with C6-dihydroceramide at concentrations as high as 20 μmol/L (data not shown), consistent with previous reports of the minimal cytotoxicity of short-chain dihydroceramides. The specific cytotoxicity of long-chain dihydroceramides such as those increased by 4-HPR is under investigation.

Lastly, we evaluated the ability of 4-HPR and DMS to induce apoptosis in MCF-7/AdrR, HT-29, and HL-60 cells. For this, SYTO-13 and propidium iodide, two vital fluorescent dyes, were used. With these reagents, normal nuclei exhibit loose green-colored chromatin; apoptotic nuclei exhibit condensed green-colored chromatin and/or fragmentation (punctate). Postapoptotic necrosis is denoted by nuclei exhibiting the same apoptotic morphologic features, but orange colored. Apoptosis, a well-known cell death mechanism of 4-HPR, was evident in MCF-7/AdrR, HT-29, and HC-60 cells after 48-hour exposure (Fig. 6). As noted by increased nuclear fragmentation, the addition of DMS to 4-HPR greatly enhanced apoptosis in all cell lines, thus showing the potency and versatility of this binary combination. In HL-60 cells, DMS alone elicited nuclear morphology characteristic to apoptosis.

The goal of the present study was to assess the mechanism(s) of action of 4-HPR with respect to the role of sphingolipids in the cytotoxic response using human cancer cell models. Our hope is that this information will lead to strategies to enhance the single-agent efficacy of 4-HPR and/or facilitate the design of multiagent drug combinations that contain 4-HPR as a sphingolipid species modulator. Clinical trials have shown that both acute and chronic administrations of 4-HPR to cancer patients are well tolerated (35–37). New oral and i.v. formulations of 4-HPR specifically designed to increase plasma levels and dosing intensity are currently in phase I clinical trials supported by the National Cancer Institute Rapid Access to Intervention Development program, with encouraging results.⁵

In one study of clinically achievable doses in ovarian cancer patients, 4-HPR steady-state plasma levels ranged from 3.1 to 12.5 μmol/L (4-HPR capsules, 900 mg/m² twice daily for 7 days; ref. 38). Overall survival with plasma levels equal to 9 μmol/L was significantly higher than for those with 4-HPR levels <9 μmol/L. In a phase I study of a 4-HPR i.v. emulsion in hematologic malignancies, the emulsion showed a linear relationship of 4-HPR dose to 4-HPR plasma level obtained, with total steady-state 4-HPR plasma levels of >50 μmol/L at >905 mg/m²/d, although free 4-HPR plasma drug levels and 4-HPR tissue levels remain to be determined (39). These studies indicate that doses evaluated in vitro are achievable clinically.

One mechanism whereby single agent 4-HPR has been reported to induce apoptosis and/or necrosis in cancer cells is via the increase of ceramides (21, 22, 30, 31, 40, 41). However, because many studies used less precise TLC methods that could not provide in-depth structural information, it was not known which particular sphingolipid subspecies were elevated. This could be an important factor because certain chain length ceramides (i.e., C16 and C18) have been more often associated with inhibition of cell growth and induction of apoptosis (1). The specific nature of the sphingoid base backbone is also important because exogenously delivered short-chain dihydroceramides are generally held to be less bioactive than ceramides in tumor cells and in normal cells (41–43). Hence, it was of considerable surprise for us to find, using LC MS/MS to analyze the sphingolipids of MCF-7/AdrR, HL-60, and HT-29 cells, as well as DU-145 prostate and MOLT-4 ALL cancer cells, in related studies (4, 19), that 4-HPR caused large increases in dihydroceramides and more complex sphingolipids containing these saturated backbones. The elevation in dihydroceramides is likely due to stimulation of de novo ceramide synthesis with concurrent inhibition of dihydroceramide desaturase because studies by our lab (4) and others (44) have found that this enzymatic activity is significantly decreased by addition of 4-HPR to intact cell or in vitro activity assays. Therefore, because 4-HPR stimulates serine palmitoyltransferase and dihydroceramide synthase (21), inhibition of the desaturase would be predicted to cause large increases in dihydroceramides (Fig. 1).

In addition to the elevation in dihydrosphingolipids, our analyses revealed that 4-HPR also caused large increases in sphinganine and Sa1P. This is intriguing because elevation of sphinganine is known to be cytotoxic (27, 45), whereas sphingoid base 1-phosphates are mitogenic and inhibit apoptosis (20). Therefore, in addition to dihydrosphingolipids, 4-HPR increased the amount of both pro– and anti–cell death mediators. To test the hypothesis that Sa1P might interfere with the cytotoxicity of 4-HPR, DMS was used to decrease the amount of Sa1P and also decrease S1P; this resulted in additional elevation of the amount of sphinganine (Fig. 3D). The 4-HPR/DMS combination was synergistically cytotoxic in MCF-7/AdrR and HL-60 cells, and the combination elicited apoptosis in all three cell lines (Fig. 6). This suggests that sphingoid bases are among the mediators of the cellular effects of single agent 4-HPR and especially of the 4-HPR/DMS combination, particularly considering the potent cytotoxicity of exogenously added sphinganine for MCF-7/AdrR cells (Fig. 5A). Our previous studies on synergistic cytotoxicity between 4-HPR and safingol, a putative sphingosine kinase inhibitor, in solid tumor cell lines (18) support our findings in 4-HPR and DMS. Further, that HL-60 cells were largely refractory to sphinganine but vulnerable to the sphinganine/DMS combination (Fig. 5B) lends support to the antagonistic roles of sphinganine and Sa1P in the cytotoxic response to 4-HPR.

Our study, like those by other investigators using varying types of short-chain ceramide analogues, found that exogenously delivered short-chain C6-dihydroceramide was not cytotoxic. However, the biological equivalency of cell-penetrant, short-chain dihydroceramides versus native long-chain dihydroceramides is not clear. Indeed, short-chain desaturated ceramides undergo cellular uptake, hydrolysis, and reacylation to long-chain ceramides (46), and in breast cancer cell lines [¹⁴C]C6-ceramide was shown to be largely converted to glucosyl[¹⁴C]C6-ceramide (47). To gain insight into the cytotoxic potential of long-chain dihydroceramides, we used GT11, an inhibitor of dihydroceramide desaturase (34), in cells exposed to sphinganine, a dihydroceramide precursor (Fig. 5A and B). In both MCF-7/AdrR and HL-60 cells, the combination of sphinganine and GT11 was more cytotoxic than either agent alone. It is interesting to note that the three-drug combination of sphinganine + GT11 + DMS, which would block sphinganine phosphorylation and retard conversion of long-chain dihydroceramides to ceramides, was the most cytotoxic. These data suggest that long-chain dihydroceramides may have direct cytotoxic properties. Dihydroceramides are also known inducers of autophagy (4), a process that is both protective and lethal for cancer cells, depending on various factors (48–50). Indeed, production of S1P by sphingosine kinase is one of the factors that have been found to be important in noncytotoxic autophagy (3, 50); therefore, suppression of 4-HPR–induced Sa1P might alter the balance between “safe” and “lethal” autophagy and play a role in the synergy between 4-HPR and DMS. The advent of new formulations of fenretinide that achieve higher 4-HPR plasma levels and, presumably, increased tumor bed levels suggests that combination chemotherapies that directly manipulate sphingolipid levels to enhance cancer-cell–specific cytotoxicity may soon be clinically achievable. Further preclinical investigations of these sphingolipid pathways are, therefore, indicted.

No potential conflicts of interest were disclosed.

We thank June Kawahara, Michelle Idler, Eric Morrison, Vicky Norton, Rose Ostrander, and Jessica Alvarez for compiling the typescript; and Adam Blackstone and Karen Hirsch, Ph.D., for assistance with the figures.