Fenretinide inhibited de novo ceramide synthesis and proinflammatory cytokines induced by Aggregatibacter actinomycetemcomitans.

Ceramides play an essential role in modulating immune signaling pathways and proinflammatory cytokine production in response to infectious pathogens, stress stimuli, or chemotherapeutic drugs. In this study, we demonstrated that Aggregatibacter actinomycetemcomitans, the pathogen for aggressive periodontitis, induced de novo synthesis of ceramide in Raw 264.7 cells. In addition, we identified that fenretinide, a synthetic retinoid, suppressed the de novo synthesis of ceramide induced by A. actinomycetemcomitans. Moreover, fenretinide attenuated interleukin (IL)-1β, IL-6, and cyclooxygenase-2 mRNA expression induced by A. actinomycetemcomitans. Fenretinide also decreased IL-1β, IL-6, and prostaglandin E2 proinflammatory cytokine levels in Raw 264.7 cells induced by A. actinomycetemcomitans. However, fenretinide had no significant effects on tumor necrosis factor alpha mRNA or protein levels. Furthermore, we showed that fenretinide inhibited the janus kinase-signal transducer and activator of transcription, phosphatidylinositol 3-kinase-Akt, protein kinase C, and nuclear factor-kappaB signaling pathways, whereas fenretinide up-regulated the mitogen-activated protein kinase signaling pathways after bacterial stimulation. This study emphasizes the de novo ceramide synthesis pathway in response to bacterial stimulation and demonstrates the anti-inflammatory role of fenretinide in the bacteria-induced immune response.


Mass spectrometry analyses for sphingolipids
Sphingolipids were extracted from the samples by the Lipidomics Core Facility using the Bligh Dyer technique. Sphingolipid analysis was performed using ESI-MS/MS on a Thermo Finnigan TSQ 7000 triple quadruple mass spectrometer operating in a multiple reaction monitoring positive ionization mode. This technique has been previously described by Bielawski et al. ( 22 ).

Quantitative real-time PCR
Total RNA was isolated from cells using TRIZOL (Invitrogen) according to the manufacturer's instructions and quantifi ed by a spectrophotometer. CDNA was synthesized by a TaqMan reverse transcription kit (Applied Biosystems, Foster City, CA) using 1,000 ng of total RNA. Real-time PCR was performed using a StepOnePlus Real-Time PCR System (Applied Biosystems) using conditions as follows: 50°C for 2 min, 95°C for 10 min, and 50 cycles of 95°C for 15 s, 60°C for 1 min. Amplicon primers of mouse IL-1 ␤ , mouse IL-6, mouse tumor necrosis factor alpha (TNF-␣ ), mouse cyclooxygenase 2 (COX-2), and mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were obtained from Applied Biosystems. Amplicon concentration was determined using threshold cycle values compared with standard curves for each primer. IL-1 ␤ , IL-6, TNF-␣ , and COX-2 mRNA levels were normalized to an endogenous control GAPDH expression and expressed as fold change compared with control groups. also serve as substrates for synthesis of SM by SM synthase (SMS) or be hydrolyzed by ceramidase to form Sph. Sph can be metabolized by Sph kinase to generate sphingosine 1-phosphate (S1P). Ceramides can be generated at high concentration within the cell membrane in response to various infectious pathogens, stress stimuli, or chemotherapeutic drugs ( 3,4 ). As "second messengers," ceramides can form large, ceramide-rich membrane platforms (also called lipid raft platforms). These lipid raft platforms have been shown to initiate and greatly amplify immune signaling transduction by recruiting transmembrane and intracellular signaling molecules ( 1,3,4 ).
Fenretinide, also called n-4-hydroxyphenyl retinamide, is a synthetic retinoid. Fenretinide has been widely used in clinical trials for breast cancer chemoprevention ( 5 ) and for treatment of neuroblastoma and Ewing sarcoma ( 6 ). It has been shown that fenretinide inhibits the proliferation of many tumor cell lines (7)(8)(9)(10)(11), whereas it manifests minimal cytotoxicity to nonmalignant cells ( 7,12 ). Early studies of fenretinide supported the hypothesis that fenretinide increased the de novo synthesis of ceramide by stimulating SPT (13)(14)(15). However, later studies ( 16,17 ) confi rmed that fenretinide increased dhCer, not ceramide levels. The later studies demonstrated that fenretinide was a potent inhibitor of DEGS ( 16,17 ). In addition, fenretinide has been shown to have an anti-infl ammatory effect. Fenretinide inhibited chemokine ( 18 ) and chemokine receptor expression ( 19 ) in vitro. In animal studies, fenretinide suppressed chronic arthritis induced by administration of streptococcal cell wall ( 20 ) and decreased the mRNA levels of proinfl ammatory mediators in the spinal cord after a spinal cord injury ( 21 ). However, the underlying mechanisms related to the anti-infl ammatory effect of fenretinide were not elucidated. In this study, we report that fenretinide had an anti-infl ammatory effect in response to Aggregatibacter actinomycetemcomitans ( A . a ), the pathogen for aggressive periodontitis. Additionally, we elucidate signifi cant difference of total SM levels in Raw 264.7 cells was observed before or after A . a stimulation. This result indicates that the SMase pathway might not play a significant role in ceramide accumulation in Raw 264.7 cells after A . a stimulation. Because SM is synthesized from ceramide by SM synthase ( Fig. 1 ), the mild increases in total SM might be caused by the increased levels of ceramide in Raw 264.7 cells after bacterial stimulation. A.a mainly increased long-chain (C16) and very-long-chain (C22, C24, and C24:1) dhCer species ( Fig. 2G-J ) and ceramide species ( Fig. 2K-N ).

Fenretinide inhibited the de novo synthesis of ceramide after A . a stimulation
In previous studies, fenretinide has been confi rmed as an agonist for SPT (13)(14)(15) but also as a potent inhibitor for DEGS ( 16,17 ). Because SPT and DEGS are critical in the de novo pathway for ceramide biosynthesis, we hypothesized that fenretinide would interfere with ceramide synthesis induced by A . a. To test our hypothesis, Raw 264.7 cells were treated with fenretinide (5 M) or control vehicle ethanol for 4 h. Cells were unstimulated or stimulated with A . a (100 CFUs) for an additional 4 h. Various kinds of sphingolipid levels, including dhSph, dhCer, ceramide, Sph, S1P, and SM, were evaluated in cell extracts by mass spectrometry analysis. In cells without bacterial stimulation, there were signifi cant increases of dhSph ( Fig. 3A ; 2.5-fold) and total dhCer ( Fig. 3B ; 19.5-fold) in fenretinide-treated cells compared with sphingolipid levels in vehicle-treated cells. In contrast, there was a signifi cant 36% decrease of the total ceramide in fenretinide-treated cells compared with vehicle-treated cells ( Fig. 3C ). There was no signifi cant difference of Sph, S1P, and total SM levels between the two groups of cells ( Fig. 3D-F ). Fenretinide mainly increased longchain (C16) and very-long-chain (C22, C24, and C24:1) dh-Cer species ( Fig. 3G ), whereas it signifi cantly decreased very-long-chain (C22, C24, and C24:1) ceramide species ( Fig. 3H ). These data support the results from previous studies, which showed that fenretinide is a stimulator of SPT ( 13-15 ) but an inhibitor of DEGS ( 16,17 ).
Previously we have shown that dhSph, dhCer, ceramide, Sph, and S1P levels signifi cantly increased in Raw 264.7 cells after short-term (30-120 min) stimulation by A . a ( Fig. 2 ). In vehicle-treated cells, we observed signifi cant increases of dhSph ( In contrast, there were no signifi cant differences of S1P and total SM levels in vehicle-treated cells before or after bacterial stimulation ( Fig. 4E, F ). A . a signifi cantly induced the production of very-long-chain (C24 and C24:1) dhCer species ( Fig. 4G ). In addition, A . a signifi cantly increased the production of long-chain (C16) and verylong-chain (C22, C24, and C24:1) ceramide species ( Fig. 4H ).
In cells stimulated with A . a , we observed signifi cant increases in dhSph ( Fig. 5A ; 2.2-fold) and total dhCer ( Fig. 5B; 14.4-fold) in fenretinide-treated cells compared ELISA IL-1 ␤ , IL-6, and TNF-␣ ELISA kits were purchased from R&D systems (Minneapolis, MN). The prostaglandin (PG)E2 ELISA kit was purchased from Cayman Chemical Co. (Ann Arbor, MI). Protein concentration in cell lysate was determined by a DC Protein Assay Kit (Bio-Rad Laboratories). The concentration of cytokines was normalized by total protein concentration in cell lysate.

Western blot assays
Total proteins were extracted by RIPA buffer (Cell Signaling Technology, Beverly, MA) supplemented with a proteinase inhibitor cocktail and a phosphatase inhibitor cocktail (Roche Diagnostics, Indianapolis, IN). Proteins were loaded on 10% Tris-HCl Ready Gel (Bio-Rad Laboratories, Hercules, CA) and electrotransferred to nitrocellulose membranes, blocked in 5% milk, and incubated overnight at 4°C with primary antibodies. P-STAT1, p-phosphatidylinositol 3-kinases (PI3K), p-Akt, p-protein kinase C (PKC), p-inhibitor of nuclear factor kappa-B kinase subunit alpha I B ␣ ), p-nuclear factor-kappaB (NF-B) p65, p-Jun N-terminal kinase (JNK), p-ERK, p-p38, and GAPDH were purchased from Cell Signaling Technology. The presence of the primary antibodies was detected on radiographic fi lm by using HRP-conjugated secondary antibodies (Cell Signaling Technology) and SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientifi c, Rockford, IL). Digitalized images of the radiographic fi lms were obtained in a gel documentation system (Bio-Rad Laboratories). Protein density/mm 2 was analyzed by Quantity One software and normalized by GAPDH protein.

Statistical analysis
Data were analyzed by unpaired two-tailed Student's t -test with Welch's correction for unequal variances. All statistical tests were performed using GraphPad Prism software (GraphPad Software Inc., La Jolla CA). Values are expressed as mean ± SD. A P value of р 0.05 was considered signifi cant.

A . a stimulated the de novo synthesis of ceramide
Previous studies have shown that bacteria induced ceramide generation by the SMase pathway (23)(24)(25), but the effect of A . a on sphingolipid metabolism has not been determined. Because ceramides serve as "second messengers" modulating immune signaling transduction, it is important to determine the effect of A . a on sphingolipid metabolisms, especially on ceramide synthesis. To address this, we treated Raw 264.7 cells with A . a (100 CFUs) for various times (30-120 min) and analyzed various sphingolipid levels (including dhSph, dhCer, ceramide, Sph, S1P, and SM) in cell extracts by mass spectrometry analysis. The dhSph and total dhCer levels signifi cantly increased after A . a stimulation ( Fig. 2 A, B ). The total ceramide level also signifi cantly increased 120 min after A . a stimulation ( Fig.  2C ). These data support the hypothesis that A . a induced the generation of ceramide by the de novo pathway. In addition, the Sph and S1P levels signifi cantly increased 60 min and 120 min after A . a stimulation ( Fig. 2D, E ). Furthermore, we observed mild increases of total SM 30 min to 120 min after A.a stimulation ( Fig. 2F ). However, no with vehicle-treated cells ( Fig. 5A, B ). There was a 63% decrease of the total ceramide in fenretinide-treated cells compared with the total ceramide level in vehicle-treated cells ( Fig. 5C ) and a 21% decrease in the Sph level in fenretinide-treated cells compared with the Sph level in vehicletreated cells ( Fig. 5D ). There was no signifi cant difference of S1P between the two groups of cells after bacterial stimulation. However, we observed a 21% decrease of total anti-infl ammatory role against A . a stimulation. Although fenretinide has inhibited the proliferation of many tumor cell lines (7)(8)(9)(10)(11), it has been shown to have minimal cytotoxicity to nonmalignant cells ( 7,12 ). In our study, fenretinide (2.5-5 M) did not induce toxicity in Raw 264.7 cells after up to 16 h treatment (n = 4; P > 0.05; data not shown), which confi rmed that fenretinide had minimal cytotoxicity. To study the effect of fenretinide on proinfl ammatory cytokine expression induced by A . a , Raw 264.7 cells were treated with fenretinide (2.5-5 M) or control vehicle ethanol for 4 h. Then the cells were unstimulated or stimulated with A . a (100 CFUs) for 4 h. Fenretinide did not induce the mRNA production of IL-1 ␤ , IL-6, COX-2, and TNF-␣ in Raw 264.7 cells without bacterial stimulation ( Fig. 6A-D ). In contrast, compared with control cells stimulated with A . a , the mRNA levels of IL-1 ␤ , IL-6, and COX-2 in fenretinide-treated cells were signifi cantly reduced in a dose-dependent manner after bacterial stimulation ( Fig.  6A-C ). However, there was no signifi cant difference of TNF-␣ mRNA levels among treatment groups after bacterial stimulation ( Fig. 6D ). Accordingly, fenretinide did not induce the cytokine levels of IL-1 ␤ , IL-6, PGE2, and TNF-␣ in Raw 264.7 cells without bacterial stimulation SM in fenretinide-treated cells compared with vehicle-treated cells ( Fig. 5F ). Because SM is generated from ceramide by SM synthase, the decrease of total ceramide level in fenretinide-treated cells might further inhibit the synthesis of SM, resulting in the lower total SM levels seen in fenretinide-treated cells compared with controls. Fenretinide mainly enhanced long-chain (C16) and very-longchain (C22, C24, and C24:1) dhCer species ( Fig. 5G ). In contrast, fenretinide signifi cantly inhibited long-chain (C16) and very-long-chain (C22, C24, and C24:1) ceramide species induced by A . a ( Fig. 5H ). In conclusion, these data support out hypothesis that fenretinide promoted the biosynthesis of dhSph and dhCer but it inhibited the de novo synthesis of ceramide induced by A.a . In addition, fenretinide might play a role in suppression of SM synthesis.

Fenretinide inhibited janus kinase-signal transducer and activator of transcription, PI3K-Akt, PKC, and NF-B signaling pathways in Raw 264.7 cells induced by A . a
Because ceramides serve as "second messengers," which regulate immune signaling transduction ( 1, 3, 4 ), we hypothesized that inhibition of the de novo synthesis of ceramide by fenretinide would affect immune signaling pathways, thus infl uencing downstream proinfl ammatory cytokine production. Many immune signaling pathways regulate proinfl ammatory cytokine production associated with bacterial stimulation, including the janus kinase-signal transducer and activator of transcription (JAK-STAT) pathway ( 26,27 ), The data are representative of two separate experiments. Data are plotted as mean ± SD (n = 3; * P < 0.05, ** P < 0.01, *** P < 0.001). ( 42 ). In addition, Sims et al. (43) reported that Kdo 2 -lipid A, a TLR4-specifi c agonist, induced de novo sphingolipid biosynthesis in Raw 264.7 cells, which is essential for the induction of autophagy. In animal studies, inhibition of de novo ceramide synthesis prevented alveolar cell apoptosis, oxidative stress, and emphysema induced by SU5416, a specifi c inhibitor of vascular endothelial growth factor receptor ( 44 ). In addition, inhibition of de novo ceramide synthesis alleviated infl ammatory cytokines in the spinal cord induced by administration of morphine ( 45 ). Furthermore, inhibition of SPT with myriocin reduced radiation-induced pulmonary infl ammation and delayed the onset of radiation-induced pulmonary fi brosis in animals ( 46 ). In this study, we demonstrated that the de novo pathway contributed to the immune response to bacterial stimulus. As shown in Fig. 2 and Fig. 4, A .a induced the de novo synthesis of dhSph, dhCer, and ceramide. Previous studies have shown that Pseudomonas aeruginosa ( 23,24 ) and Neisseria gonorrheae ( 25 ) activated acid SMase at early time points (5-45 min). Our previous study also demonstrated that P . aeruginosa activated acid SMase up to 300 min after bacterial stimulation ( 24 ). In the current study, we did not observe a signifi cant reduction of total SM at an stimulation compared with controls. The p-JNK and p-ERK levels increased in fenretinide-treated cells after bacterial stimulation to protein levels similar to those seen in vehicle-treated cells after bacterial stimulation ( Fig. 8A, H, I ). Fenretinide-treated cells exhibited an increased p-p38 level compared with control before bacterial stimulation. The p-p38 level was even higher after bacterial stimulation in fenretinide-treated cells compared with controls ( Fig.  8A, J ). In conclusion, our data demonstrate that fenretinide exhibited signifi cant effects on the attenuation of JAK-STAT, PI3K-Akt, PKC, and NF-B signaling pathways induced by A . a , whereas it played a role in up-regulation of the MAPK signaling pathways after bacterial stimulation.

DISCUSSION
De novo sphingolipid synthesis plays an important role in regulating immune response. Previous studies have shown that de novo synthesis of sphingolipids is critical for regulating ceramide and other sphingolipid levels in response to chemotherapeutic and proapoptotic agents (37)(38)(39)(40), heat stress ( 41 ), and infl ammatory cytokines difference among groups of cells ( Fig. 5E ). One of the reasons is that S1P is a single-chain lipid that has suffi cient aqueous solubility to pass across cell membrane ( 2 ). It has been shown that S1P has low nanomolar concentrations in the cells but higher nanomolar concentrations in serum, where it is associated with lipoproteins and albumin ( 2,50 ). In addition, S1P can be further irreversibly degraded by S1P lyase to release ethanolamine phosphate and hexadecenal ( 51 ). The lack of signifi cant difference of S1P 4 h after bacterial stimulation among all groups of cells might be caused by the quick extracellular release and degradation by S1P lyase.
In this study, although there were high levels of dhSph and dhCer in Raw 264.7 treated with fenretinide before bacterial stimulation ( Fig. 3A, B ), those sphingolipids did not induce proinfl ammatory cytokine mRNA or protein expression before bacterial stimulation ( Fig. 6 and 7 ); nor did they activate immune signaling pathways before bacterial stimulation ( Fig. 8 ). This phenomenon might be caused by those sphingolipids' hydrophobic property and their restricted subcellular localization ( 2,51 ). It is known that the de novo synthesis of sphingolipids occurs in the ER and possibly at ER-associated membranes, such as the perinuclear and mitochondria-associated membranes ( 2,51 ). The dhSph and dhCer might be restricted in the ER and possibly the ER-associated membrane compartments and therefore might not interact with receptors on the plasma membrane, which would prevent them from initiating early time point (15 min, data not shown) and 30 to 120 min after A . a stimulation ( Fig. 2F ). One of the causes of discrepancy might be the different strain of bacteria used. Some bacteria, including pseudomonas, possess bacterial SMase ( 47,48 ), which can hydrolyze SM, resulting in the generation of ceramide. A . a might lack bacterial SMase. Another possible reason for the discrepancy is that we used paraformaldehyde-fi xed dead bacteria. The paraformaldehyde fi xation process might inactivate bacterial SMase. Further study is needed to determine if live A . a possess bacterial SMase and could hydrolyze SM.
Although the total ceramide level in fenretinide-treated cells with bacterial stimulation ( Fig. 5C ) was signifi cantly reduced to levels similar to those without bacterial stimulation ( Fig. 3C ), the Sph level was signifi cantly enhanced in fenretinide-treated cells with bacterial stimulation ( Fig. 5D ; P < 0.01) compared with fenretinide-treated cells without bacterial stimulation ( Fig. 3D ). These results suggest that bacterial stimulation might activate the salvage pathway, which would explain the accumulation of Sph after bacterial stimulation. It is estimated that the salvage pathway contributes to about 50% to 90% of sphingolipid biosynthesis and turnover ( 49 ); therefore, it is not surprising that the salvage pathway could contribute to ceramide generation induced by bacterial stimulation.
Although we observed a signifi cant reduction of ceramide level in fenretinide-treated cells 4 h after bacterial stimulation ( Fig. 5C ), the S1P did not manifest a signifi cant fenretinide. Previous work has shown that dhCers are involved in cell cycle arrest, apoptosis, autophagy, and oxidative stress ( 54 ), which could affect immune signaling transduction. There was an up-regulation of MAPK in fenretinide-treated cells after bacterial stimulation; especially the p-p38 and p-JNK levels were higher at 60 min in fenretinide-treated cells after bacterial stimulation compared with controls ( Fig. 8H, J ). Previous studies have demonstrated that fenretinide plays a role in up-regulation of MAPK signals ( 55,56 ), which is dependent on the reactive oxygen species induced by fenretinide ( 55 ). We speculate that the up-regulation of protein kinase signals after bacterial stimulation in fenretinide-treated cells might be associated with the high level of dhCer, which further enhanced the oxidative stress induced by bacterial stimulation. The TNF-␣ production might be regulated by MAPK and NF-B signals. The up-regulation of MAPK signals after bacterial stimulation offset the down-regulation of NF-B signals, which resulted in no signifi cant difference of TNF-␣ levels among all treatment groups after bacterial stimulation. On the other hand, IL-1 ␤ , IL-6, and PGE2 might be mainly regulated by the NF-B signaling pathway and less affected by the oxidative stress; therefore, we observed signifi cant reductions of those cytokine levels in fenretinide-treated cells after bacterial stimulation.
A previous study ( 21 ) demonstrated that long-term (18 h) treatment with fenretinide inhibited TNF-␣ in microglial cell culture induced by LPS stimulation. In our study, Raw 264.7 cells pretreated with fenretinide for 16 h and stimulated with immune signaling transduction. In contrast, ceramide, Sph, and S1P reside on the plasma membrane ( 2,51 ). Ceramides are neutral hydrophobic molecules, which are able to flip-flop across membranes ( 52 ). On the other hand, Sph, and S1P are single-chain lipids, which have suffi cient aqueous solubility to move across membranes ( 2 ). Therefore, ceramide, Sph, and S1P are called "bioactive sphingolipids" ( 2 ) and are capable of interacting with receptors on the plasma membrane and initiating immune signaling transduction. We speculate that the signifi cant reductions of proinfl ammatory cytokines in fenretinide-treated cells with bacterial stimulation might mainly be caused by the low level of ceramide in this study.
dhCer induced by fenretinide might play a minor role in regulation of cytokine production. Previously, Cuzzocrea et al. ( 53 ) showed that TNF-␣ and IL-1 ␤ levels were decreased by inhibition of ceramide biosynthesis by fumonisin B1, tyclodecan-9-xanthogenate, and (3-carbazol-9yl-propyl)-[2-(3, 4-dimethoxy-phenyl)-ethyl]methylamine (NB6), inhibitors of ceramide synthase, acid sphingomyelinase, and the secretory form of acid sphingomyelinase. In addition, Ndengele et al. ( 45 ) demonstrated that inhibition of ceramide biosynthesis by administration of fumonisin B1 in animals signifi cantly attenuated IL-1 ␤ , IL-6, and TNF-␣ levels in spinal cord tissues induced by morphine. In contrast, we observed signifi cant reductions of IL-1 ␤ , IL-6, and PGE2 but not in TNF-␣ in fenretinidetreated cells with bacterial stimulation. This discrepancy might be caused by the enhanced dhCer level induced by  inner membrane potential ( 1 ). In contrast, short-chain ceramides, particularly C2-ceramide and C6-ceramide, are water soluble and therefore membrane permeable. Therefore, using exogenous short-chain ceramides as mimics of endogenous long-chain ceramides is, in general, not justifi ed ( 1 ). In addition, previous studies have shown that the PI3K pathway plays a dual role in modulating the immune response ( 36,69 ). PI3K serves as an essential proinfl ammatory signaling molecule activating the NF-B signals ( 34 ). On the other hand, PI3K also has an anti-infl ammatory role, which negatively regulates the p38 MAPK signal pathways ( 69 ). That might explain why we observed an up-regulation of p-p38 in fenretinide-treated cells without bacterial stimulation compared with control ( Fig. 8A, J ).
In conclusion, our results support that de novo synthesis of ceramide plays an important role in host immune response to A . a . Fenretinide inhibited the de novo synthesis of ceramide and suppressed JAK-STAT, PI3K-Akt, PKC, and downstream NF-B signaling pathways. Furthermore, it attenuated IL-1 ␤ , IL-6, and PGE2 proinfl ammatory cytokine expression induced by A . a .
A . a for an additional 4 h signifi cantly decreased TNF-␣ expression induced by A.a compared with controls (data not shown). However, pretreatment with fenretinide for 4 h had no signifi cant effect on alleviating TNF-␣ expression induced by A. a . Moreover, we observed even higher TNF-␣ mRNA and protein levels in Raw 264.7 cells treated with a high dose (10 M) of fenretinide with bacterial stimulation compared with controls (data not shown). This discrepancy might be caused by the activities of dhCer, and possibly dhSph, involved at various times. The impact of various activities associated with dhCer ( 54 ), such as cell cycle arrest, apoptosis, autophagy, and oxidative stress, might be variable during different time points. In the long-term (18 h) treatment study of fenretinide, the antiproliferative and autophagy roles associated with dhCer might have a signifi cant impact on cells, leading to cell growth arrest or cell death and thus less TNF-␣ production. In contrast, in this short-term (4 h) treatment study, the antiproliferative and autophagy roles associated with dhCer might only have a minor effect on cells, whereas the oxidative stress induced by both bacterial stimulus and dhCer might have a major impact on cells, which could cause an up-regulation of MAPK after bacterial stimulation.
In this study, inhibition ceramide de novo synthesis by fenretinide signifi cantly attenuated p-PI3K, p-Akt, and p-PKC expression after bacterial stimulation compared with control. Our data show that ceramide played an essential role in regulating the PI3K-Akt and PI3K-PKC signaling pathways. Previous studies have shown that Akt was recruited to the ceramide-rich microdomains after PI3K activation ( 57,58 ). However, these studies also found confl icting results regarding how ceramide regulated the PI3K-Akt and PKC signaling pathways. Monick et al. ( 59 ) demonstrated that LPS induced the generation of ceramide and the activation of PI3K-Akt in human alveolar macrophages. They also showed that blocking ceramide synthesis by D609 blocked LPS-induced PI3K-Akt activation. D609 is a potent inhibitor of the phosphatidylcholine-specifi c phospholipase C, an enzyme known to be involved in acid SMase activation ( 25,60 ). In addition, Barsacchi et al. ( 60 ) showed that activation of neutral SMase by TNF-␣ induced the generation of ceramide and the activation of the PI3K-Akt signaling pathway. Our results are in line with the fi ndings by Monick et al. (59) and Barsacchi et al. (60) that ceramide plays a role in activation of the PI3K-Akt signaling pathway. However, there was also evidence that synthetic exogenous ceramide inhibited the PI3K-Akt and PKC signaling pathways. Exogenous C6-ceramide inhibited PKC ␣ activity ( 61 ) and suppressed Akt activation ( 62,63 ). In addition, exogenous C2-ceramide led to dephosphorylation of Akt (64)(65)(66)(67)(68). The inhibition of Akt activity by C2-ceramide or C6-ceramide had no effect on PI3K activity ( 63,64 ) or was associated with down-regulation of PI3K activity ( 66,68 ). It is important that natural, long-chain ceramides have a signifi cant difference from the synthetic short chain ceramides. The long-chain ceramides are not water soluble and are therefore unable to dissipate the mitochondrial