Stereospecific induction of apoptosis in tumor cells via endogenous C16-ceramide and distinct transcripts

Concentration and distribution of individual endogenous ceramide species is crucial for apoptosis induction in response to various stimuli. Exogenous ceramide analogs induce apoptosis and can in turn modify the composition/concentrations of endogenous ceramide species and associated signaling. In this study, we show here that the elevation of endogenous C16-ceramide levels is a common feature of several known apoptosis-inducing triggers like mmLDL, TNF-alpha, H2O2 and exogenous C6-ceramide. Vice versa apoptosis requires elevation of endogenous C16-ceramide levels in cells. Enantiomers of a synthetic ceramide analog HPL-1RS36N have been developed as probes and vary in their capacity to inducing apoptosis in macrophages and HT-29 cells. Apoptosis induction by the two synthetic ceramide analogs HPL-39N and HPL-1R36N correlates with generation of cellular C16-ceramide concentration. In contrast to the S-enantiomer HPL-1S36N, the R-enantiomer HPL-1R36N shows significant effects on the expression of distinct genes known to be involved in cell cycle, cell growth and cell death (CXCL10, CCL5 and TNF-alpha), similarly on apoptosis induction. Enantioselective effects on transcription induced by metabolically stable synthetic probes provide clues on molecular mechanisms of ceramide-induced signaling, as well as leads for future anti-cancer agents.


INTRODUCTION
Ceramides with fatty acids of varying chain length bound as amides are components of the sphingomyelin (SM) cycle and well-established signaling molecules, since activation of sphingomyelinases (SMases) and subsequent ceramide generation is involved in signal transduction of the cellular stress response, in particular during stress-induced apoptosis. [1][2][3] The substrate of these pathways, SM, is an inert phospho-sphingolipid abundant in all eukaryotic cell types. On pro-inflammatory or proapoptotic stimulation, the SMases (phospholipase-like enzymes) catalyze its breakdown into ceramides which are highly bioactive lipid-mediators with two asymmetric carbons. 4,5 These molecules function as common, converging messengers in apoptosis, and are also generated by signaling through CD95-Fas/Apo1, tumor necrosis factor alpha (TNF-alpha), ionizing radiation or chemotherapeutic agents. [6][7][8] Moreover, in chronic atherosclerotic states, initiation of apoptosis following treatment of macrophages with oxidatively modified lipoproteins is mediated via an increase in the intracellular ceramide concentration. 9 Elucidation of distinct steps of apoptosis initiated by synthetic, metabolically stabilized and more rigid analogs of ceramide with defined stereochemical configuration offers a valuable strategy to define structure-activity relationships and may foster the development of pro-apoptotic anti-cancer agents or agents for enhancing or retaining the potency of well-known cancer drugs.

RESULTS
Effect of apoptogenic agents, C 6 -ceramide and synthetic, conformationally restricted ceramide analogs on C 16 -and C 24 -ceramides We have previously demonstrated that exogenous C 6 -ceramide provokes an increase in total ceramide levels (determined by diacylglycerolkinase-assay in fibroblasts, an effect inhibitable with our agent NB06. 10,11 Here we examined whether these effects accompany changes of endogenous C 16 -and C 24:1 -ceramides, especially when compared with the effect observed in response to C 6 -ceramide stimulation. To further characterize the molecular mechanism of apoptosis induction, we first analyzed changes in total cellular ceramide content, then endogenous C 16 -and C 24:1 -ceramide concentrations in response to known stimulatory agents such as mmLDL and TNF-alpha as references ( Figure 1). 12 We found that both, mmLDL and TNF-alpha, cause a significant rise in endogenous C 16 -ceramide (2), however, TNFalpha causes significant reduction of C 24:1 -ceramide concentrations in macrophages.
Apoptosis along with increased C 16 -ceramide concentrations The three compounds tested, C 2 -dihydroceramide (5), exogenous C 6 -ceramide (1) and HPL-39N (3), not only differ in their apoptosisinducing properties but also in their ability to change intracellular C 16 -ceramide concentration as determined by analyzing the lipid extract of macrophages. While exogenous C 6 -ceramide (1) and HPL-39N (3) induce a clear, statistically significant (P o0.05) increase in endogenous C 16 -ceramide concentration after 4 h, C 2 -dihydroceramide (5)-treated cells display no change (Figure 3a). Similarly, after 8 h (Figure 3b), a significant rise in the C 16 -ceramide concentration is noted, however, insignificant as for the C 2 -dihydroceramide (5)-induced effect (P o0.05). Compared with C 6 -ceramide, the effect of the racemic analogue HPL-39N (3) is significant as well but somewhat less pronounced (after 8 h: 2.19 pmol of C 16 -ceramide by C 6 -ceramide versus 2.74 pmol in response to HPL-39N (3)). Both the synthetic ceramide analogue HPL-39N (3) and its partial synthetic congener C 6 -ceramide (1) increase the C 24:1 -ceramide content in the cells after 8 h, whereas no significant change is observed after 4 h. This suggests that HPL-39 N (3) actually functions as a ceramide mimic and stimulates similar or identical mechanisms in the cell while C 2 -dihydroceramide (5) causes a drop in the C 24:1 -ceramide content (to 1.82 pmol C 24:1 -ceramide), it does not show a significant (Po 0.05) increase in the C 24:1 -ceramide content after 8 h. Obviously the 4,5-trans double bond in the sphingosine backbone is an essential element for raising the cellular C 24:1 -ceramide content.  Enantioselective effects of synthetic ceramide analogs HPL-1R36N (6) and HPL-1S36N (7) on endogenous ceramide species, caspase activation and apoptosis induction The ceramide analogue HPL-39 N (3) bears an asymmetrically substituted hydroxy group while, unlike ceramides, the amino group is integrated into an oxazole heterocycle. To probe for stereospecific effects in terms of apoptosis induction, as well as on distinct endogenous ceramide species concentrations, compounds HPL-38 N (4) and HPL-1RS36N (8) were tested as racemates/diastereomeric mixtures, as well as enantiomers (obtained by separation on chiral columns (for HPL-1R36N (6)/HPL-1S36N (7)), alternatively by enantioselective synthesis (HPL-38 N (4), see Supplementary Materials). In contrast to the 1RS-racemate HPL-38 N (4), HPL-39 N (3) shows a strong apoptosis-inducing activity. At equal concentrations, the pure 1R compound HPL-39 N (3) is significantly more active than the racemic mixture with regard to apoptosis induction. 13 The 1R compound HPL-39 N (3) (in comparison to exogenous C 6ceramide (1)) has significantly stronger effects on C 16   macrophages ( Figure 3); both HPL-39 N (3) and its natural reference C 6 -ceramide (1) cause a significant increase in C 16ceramide content of macrophages and in parallel corresponding rates of apoptosis.

Signaling mechanisms behind enantioselective apoptosis induction
Since apoptosis induction is of outstanding relevance to mechanisms of cancer treatment, we further examined agent-associated enantiospecific effects on the transcriptional level. Human HT-29cells (Caucasian colon adenocarcinoma grade II), a widely used cancer model, were treated with HPL-1R36N (6) and HPL-1S36 (7) and effects analyzed by spotted microarrays (Lab Arraytor 60-1 to 60-6 combi oligonucleotide chip (SIRS-Lab, Jena, Germany), Figures 5a and b). 17 Interestingly the enantiomers 1(R)-(E)-(2methyl-oxazol-4-yl)-hexadec-2-en-1-ol (6) and 1(S)-(E)-(2-methyl-oxazol-4-yl)-hexadec-2-en-1-ol (7) of the bioactive oxazol compound HPL-1RS36N (8), differed markedly as for time-and enantiomerspecific effects on transcription. On array analysis, 158 genes showed RNA-concentration changes ≥ factor 1.5 and were assigned as differentially regulated; 16 gene activities showing more than 2-fold changes were identified. Microarray analysis thus revealed that differences in molecular mechanisms of enantiomers can be directly detected at the transcriptional level. Notably key players of apoptotic pathways are affected differently and apparently reflect pro-apoptotic potency. As for synthetic ceramide analogs, the data indicate that tiny structural changes may be crucial and transcriptomics appears valuable for compound characterization and optimization of efficacy in this context.
Enantioselectively inducible genes-stereospecificity of ceramide analogs Similar to HPL-39 N (3), HPL-1RS36N (8) also activates PKCalpha more strongly than C 6 -ceramide (1), and exhibits a larger proapoptotic activity in Renca cells. 13 Human colon carcinoma cells HT-29 are a well-established model to study ceramide-triggered effects on proliferation and induction of apoptosis. 17 In human HT-29 cells, the two enantiomers of HPL-1RS36N (8)-HPL-1R36N (6) and HPL-1S36N (7)-show distinct effects on gene expression. HT-29 cells, however, are adherent human colon adenocarcinoma cells, much less metabolically active compared with mononuclear leukocytes or monocytes. So it is hardly surprising that, ultimately, only a few differentially expressed genes were identified among the totally 593 human genes spotted on the microarray. Out of a total of 28 genes, 12 were found to be differentially expressed as analyzed by quantitative real-time PCR at one or more points in time: as for their biological function these can be classified into three groups: (I) transcription factors affecting cell cycle regulation (E2F1, E2F2, E2F5, E2F6), (II) enzymes and proteins involved in apoptosis (ASAH1, ITPK1, BCL2L11, TRAF4) and (III) chemo-and cytokines (CCL5, CXCL10, IL-1B, TNF-alpha), known to be involved in cancer development, cell death, cell growth, cellular development and cell cycle.

DISCUSSION
In light of the directly opposed effects of FKBP51 and FKBP52 inhibitors, 18 selective and metabolically stable synthetic probes are critical for mechanistic studies, as well as for assessing the pharmacological potential of ceramide mimics. Apart from corresponding structural functionalities, the synthetic ceramide analog HPL-1RS36N (8) shares similar alteration of target transcripts with ceramide (MB, MK and H-PD, unpublished experiments)). Since the R-enantiomer HPL-1R36N (6) is significantly more effective, the stereochemistry at the C1-atom, in addition to the 2,3-trans-double bond, proves to be an essential structural feature. We demonstrate that under our experimental setting, three genes showing differential regulation across all time points were identified by both hybridization experiments (CXCL10 and CCL5) and quantitative real-time PCR (TNF-alpha). All of these differentially regulated genes are known for their critical role in apoptosis and cell cycle control. CXCL10 belongs to the group of chemokines without ELR motif, and is part of the immune and lymphatic system and the immune response; it has signal transducing, receptor binding, chemokine and chemo-attractant activity and is only present extracellularly. CXCL10 is further involved in the cell cycle, inter-cellular communication, in cell growth, cell proliferation and movement, angiogenesis (preventing angiogenesis), apoptosis (detected in neurons) and carcinomas. 19 Our agents and structurally related compounds thus may target cells capable of producing CXCL10 such as adenocarcinoma cells. In human intestinal epithelial cells IFNgamma induces NF-κB, IL1 and TNF-alpha expression and secretion of CXCL10. CXCL10 itself, however, exhibits no direct effect on HT-29 cells since CXCL10 is not a ligand of the (only proven) CXCR4 receptor. The associated CXCL10 receptor CXCR3 appears to be as little expressed in HT-29 cells as the specific ligand for CXCR4, the CXCL12 receptor. 20 CXCL10, however, has been shown to induce caspase-3-dependent apoptosis in neurons, which is blocked irreversibly by the caspase-3 inhibitor DEVD and VAD. 21 If VAD is coupled with a fluorescent dye (e.g. CaspACE-FITC-VAD-FMK), activated caspases can be measured by flow cytometry. 22 In a flow cytometry experiment in CXCR3 receptor-positive cells, the two enantiomers HPL-1R36N (6) and HPL-1S36N (7) (Figure 4b) should accordingly exhibit differences in terms of activated caspases. A representative result for human macrophages (CXCR3 receptor positive) is shown in Figure 4b. Similar to CXCL10 expression, HPL-1R36N (6) is the more active enantiomer. When comparing differences in CXCL10 expression, caspase activation and apoptosis induction, the differences in caspase activation are relatively moderate (~1.4-fold). Accordingly, it is reasonable to assume that apoptosis-inducing effects in this context not only rely on the activation of caspases, but also on caspase-independent signaling, subject to future experimentation.
CCL5 (RANTES) is another cytokine found to be differentially regulated by the two enantiomeric ceramide analogs (Figure 5c). It acts as an immunoregulant, chemotactic agent (inducing migration) by attracting T lymphocytes (T cells) of the Th1-and memory cell type on eosinophilic and basophilic granulocytes. It thus controls the recruitment of leukocytes into inflamed tissue. Apoptosis induction in peripheral blood T cells via CCR5 receptor occupation can be achieved with CCL5 concentrations in the micromolar range; signal transduction here is initiated by cytochrome c release via caspase-3-and caspase-9-and PARP-1dependent pathways. 23 However, in the absence of the CCR5 receptor-as in HT-29 cells-apoptosis cannot be induced. When interpreting the results of the flow cytometric experiments with the enantiomers (Figure 4b) in human macrophages it has to be considered that although these cells belong to the CCR5 receptorpositive ones, the pro-apoptotic effect of CCL5 is negligible due to the necessarily very high CCL5 concentration.
TNF-alpha (TNFSF2) is the third prominently altered cytokine transcript. In contrast to CXCL10 and CCL5, differential TNF-alpha expression, due to low concentration of its mRNA, has been detectable only by quantitative real-time PCR. TNF-alpha is a multifunctional pro-inflammatory cytokine belonging to the TNF super family playing a central role in a variety of biological processes such as apoptosis, induction of the expression of genes, lipid metabolism, secretion of cytokines, cell activation, cell death, cell adhesion, cell differentiation, cell stimulation and cell proliferation. 24 As a common feature, biological effects mediated by the three genes CXCL10, CCL5 and TNF-alpha are exclusively conveyed through binding to specific receptors. HT-29 cells express only TNFR1 and TNFR2 cytokine receptors for TNF-alpha and are therefore irresponsive to biological effects of CXCL10 and CCL5 including apoptosis induction. However, many cells in their natural tissue environment express these cytokine-specific receptors, and apoptosis can be initiated in response to secretion by, for example, adenocarcinoma cells. We thus have been able to identify several potential contributors to observed alterations in apoptosis along with raised concentration of C 16 -ceramide and have shown that caspase activation may be a critical component. An analysis of further details and (quantitative) degree of distinct contributions were beyond the scope of this report, however, will be addressed by future investigations.
Our data demonstrate that enantiospecific ceramide-like signaling and apoptosis of tumor cells can be induced by administration of synthetic probes with defined stereochemistry specifically raising the level of apoptogenic endogenous C 16ceramide. Enantioselective effects on transcription induced via these agents also demonstrate their utility as specific probes for elucidating molecular mechanisms of apoptosis signaled via distinct endogenous ceramide species. Further, this study is likely to stimulate further research toward structurally related antitumor agents, for example, capable of inducing CXCL10 in tumor cells (1).

Cell culture and cell culture experiments
Monocytes/macrophages. HMBC used in cell culture experiment were isolated from fresh buffy coats using Histopaque (Sigma-Aldrich, Deisenhofen, Germany) and cultured in six-well plates (9.6 cm 2 , Greiner Bio-One, Frickenhausen, Germany) six to seven days in a humidified incubator at 37°C and 5% CO 2 with RPMI-1640 medium (PAA, Cölbe, Germany) supplemented with mit 2 mM L-glutamine, 50 U/ml penicillin 50 μg/ml streptomycin (Sigma-Aldrich) and 10% FCS (Greiner Bio-One). Before and during incubation with compounds supplemented with RPMI-1640 Medium with a reduced content of 1% of FCS was used. Human colon carcinoma cells HT-29 (DSMZ, Braunschweig, Germany) were cultured in cell culture flasks (175 or 25 cm 2 (stock) in a humidified incubator at 37°C and 5% CO 2 with McCoy´s 5 A Medium (PromoCell, Heidelberg, Germany) supplemented with mit 2 mM L-glutamine, 50 U/ml penicillin, 50 μg/ml streptomycin and 10% FCS. Before experiments cells were seeded semi-confluent in 5.8 cm 2 Petri dishes (Greiner Bio-One), equilibrated at 37°C over night in McCoy´s 5 A Medium with 1% human AB serum and supplemented with 2 mM L-Glutamine, 50 U/ml Penicillin and 50 μg/ml Streptomycin.
Ceramide quantification. Macrophages were treated 4 h with mmLDL (54 μg/ml) and TNF-alpha (3 ng/ml) (Figure 1), 4 or 8 h with 10 μM exogene C 6 -ceramide (1), 10 μM HPL-39 N (3), 10 μM C 2 -dihydroceramide (5) (Figure 3) or treated 4 or 7 h with 10 μM HPL-1R36N (6) and HPL-1S36N (7) (Figure 4). All organic compounds were dissolved in a 10 mM ethanol/DMSO stock solution; the same concentrations of solvents were present in control samples. Appropriate dilutions in ethanol/cell culture medium and volumes were applied to the cells at the beginning of incubation. Lipid extraction, purification and ceramide quantification were performed by HPLC-analysis as described in section ceramide quantification using DECCA (7-diethylamino)coumarin-3-carbonyl azide) for ceramide labeling and 100 pmol C 8 -ceramide as internal standard. 25 Oligonucleotide array/quantitative real-time PCR. HT-29 cells were quantification is performed according to the book by Jenkins and Hannun. 27 To separate ceramides from other (sphingo-)lipids, the crude extract is separated on a 0.2 mm silica gel 60 F254-coated TLC plate (Merck) using a 6 : 4 : 1 (v/v) chloroform/methanol/water mixture as mobile phase. Exogenous ceramides are used as TLC plate marker, lanes separated from cellular lipids and detected separately in an iodine chamber. Ceramide bands (cellular lipids) are scraped from the plate and transferred to Poly-Prep Columns (Bio-Rad, Munich, Germany), eluted with chloroform (3 × 250 μl) and finally with 250 μl methanol. Glycerolipid/monoacylglycerol impurities are removed by alkaline hydrolysis in 0.03 M NaOH in 90% ethanol at 37°C for 30 min. After neutralization a second Bligh and Dyer lipid extraction is performed. The organic phase is dried over water-free sodium sulfate and evaporated to dryness.
Ceramide labeling and HPLC separation. An optimized HPLC method based on conditions similar to those by Moersel and Balestrieri was used to achieve best performance in terms of sensitivity and ceramide species separation. 25,28,29 Solutions containing 100 pmol reference C 8 -ceramide (for calibration) and ceramides obtained from cellular lipid extracts (equivalent to 5 to 10 nmol lipid phosphate or (5,10,25,50,100,250 and 500 pmol of C 16  Total RNA sample preparation Monolayers were washed with ice-cold PBS and cells were scraped in 3 ml lysis buffer and total RNA from HT-29 cells was extracted with RNeasy Mini Spin Columns (Qiagen, Hilden, Germany) or TriPure Isolation Reagent (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's instructions. RNA yields were determined spectrophotometrically on a NanoDrop ND-1000 Spectro-Photometer (NanoDrop Technologies, Wilmington, USA) by measuring the absorbance at 260 and 280 nm. All RNA samples used for microarray analysis were analyzed by ethidium bromidestained RNA agarose gel, to confirm purity and integrity of RNA, and a Bioanalyzer 2100 (Agilent Technologies, Boeblingen, Germany).

Oligonucleotide microarray hybridization
Experiments were performed using the Lab-Arraytor-60-1 to 60-6 combi oligonucleotide microarray (SIRS-Lab), comprising 539 probes (each made as triple replicates) addressing 519 transcripts corresponding to inflammation, as well as 22 reliable control probes (see Supplementary Notes: 'Microarray Experiment Description' online according to ref. 30 About 10 μg of total RNA were reverse transcribed using Superscript-II reverse transcriptase from Invitrogen in the presence of aminoallyl-dUTP from Sigma (Taufkirchen, Germany) and labeled using the Dyomics DY-648-S-NHS/DY-548-S-NHS labeling system (Dyomics, Jena, Germany). DY-548-S-NHS-labeled cDNA from HPL-1S36N (7)-treated cells were co-hybridized with DY-648-S-NHS-labeled cDNA obtained from the same amount of total RNA isolated from cells treated with HPL-1R36N (6). After incubation in a hybridization apparatus (HS 400, TECAN, Crailsheim, Germany, for 10 h at 42°C, formamide-based hybridization buffer system) arrays were washed according to the manufacturer's instructions, dried and hybridization signal intensities were measured immediately using an Axon 4000B scanner (Axon Instruments, Foster City, CA, USA). Microarray data preprocessing of hybridization signals included (i) spot detection and background subtraction, (ii) spot flagging according to defined signal-tonoise threshold values and (iii) normalization and transformation of the signals obtained from different channels. For the former two steps, the GenePix 5.0 Analysis Software 5.0 (Axon Instruments) was used; for the third step we applied the approach from the study by Huber et al. 31,32 Quantitative real-time PCR cDNA synthesis. First-strand complementary DNA synthesis was performed with 2 μg of isolated total RNA from a HT-29 cells (also used for hybridization experiments) according to the manufacturer's instructions. After adjusting total RNA volume to 11 μl, 1 μl (2.5 μg/μl oligo-d(T)12-18primer solution is added, denatured at 70°C in a PTC-200 DNA Engine (Bio-Rad) for 10 min and chilled on ice for 5 min. About 9 μl of RT-Mix (4 μl reaction-buffer 5 × , 2 μl DTT 100 mM, 0.5 μl RiboLock RNAse-inhibitor 40 U/ml,1 μl Revert Aid Reverse Transcriptase (RT)), 10 μl 10 mM dNTP stock solution (10 mM dATP, 10 mM dGTP, 10 mM dCTP, 10 mM dTTP) (Thermo Scientific Molecular Biology, St. Leon-Rot, Germany) added, mixed and incubated 60 min at 42°C. The enzyme was then inactivated by incubation at 70°C for 5 min.
Quantitative real-time PCR. Real-Time PCR was performed in a Bio-Rad iQ-cycler (Bio-Rad) equipped with skirted Micro seal 96-Well PCR-plates covered with Microseal 'B' adhesive foils (Bio-Rad). Reaction volume (20 μl, containing 30 ng cDNA) consists of 4 μl diluted cDNA-solution (7.5 ng/μl), 2 μl 0.1 μM gene-specific forward and reverse primer solution (2 pMol each; Biomers, Ulm, Germany), 4 μl DEPC-treated water and 10 μl RT 2 Real-Time SYBR-Green/Fluorescein-PCR-Master-Mix (SA Bioscience, Hilden, Germany/ Qiagen). According to manufacturer recommendations, a cycling program with an initial activation step (94°C, 3 min) followed by 45 cycles of denaturation (94°C, 30 s), annealing (60°C, 30 s), elongation (72°C, 30 s) and a final elongation (72°C, 30 s) is executed. Fluorescence acquisitions in the SYBR green and ROX (internal reference dye) channels were performed at the end of the annealing step. A melting protocol ranged from 94 to 59°C following a stepwise increment of 0.5°C held for 3 s. Each sample as well as a negative template control (NTC) was amplified in triplet for each of the primer pairs assayed. Raw data (ct-values) were extracted using iCycler iQsoftware (version 3.1, Bio-Rad) running on the Bio-Rad iQ-cycler.

Statistics
Preprocessing and statistical analysis of microarray gene expression data were performed using the statistical software R in combination with Bioconductor. 33,34 qBasePlus-Software (Version 1.3; Biogazelle, Gent, Belgium) was used for RT-qPCR data. Statistical significance was investigated using one-way ANOVA in combination with post hoc t-tests and Bonferroni's correction for multiple testing, Po0.0001 (one-way ANOVA); significant difference was considered if Po 0.05. For more details we refer to the Supplementary Material. Primer design, synthesis, data read out and sequences Gene specific primers (18-22-bp length) were designed by use of Primer 3 software version 0.4.0 (MIT Center for Genome Research; http://frodo.wi. mit.edu/cgi-bin/primer3/primer3_www.cgi) to obtain an annealing temperature of 57°C and an amplicon length between 50 and 150 bp (up to 250 bp if necessary). 35 Gene and species specificity was tested using NCBI nucleotide database, nucleotide blast and interrogation mode 'blastn' (the National Center for Biotechnology Information, Bethesda, MD, USA).
Raw data extraction, normalization software/reference genes Relative gene expression for each investigated gene was calculated using qBasePlus-Software (Version 1.3, Biogazelle) or the method of Pfaffl. 36 Primer efficiency was determined for each primer by a cleaned up PCRproduct dilution series (QIAquick PCR Purification Kit, Qiagen). HPRT1 was chosen as a reference gene for normalization of relative gene expression of each gene. It was selected as the most stable gene of all tested genes by 'NormFinder'-algorithm. 37 Stability value was calculated as 0.200 ± 0.094.