Deoxysphingolipids, Novel Biomarkers for Type 2 Diabetes, Are Cytotoxic for Insulin-Producing Cells

In the past 3 decades, the prevalence of diabetes has risen worldwide at a dramatic rate, with incidence projections approaching 8% of the population by 2030 (1,2). This remarkable increase is largely due to the epidemic spread of type 2 diabetes (T2DM), which accounts for 90% of all cases of diabetes worldwide (3,4). Given the level of complexity associated with the pathophysiology of T2DM, understanding the mechanisms underlying this disease is necessary to design alternative strategies to limit its progression. Recently, substantial improvements occurred in the detection of early stage or undiagnosed T2DM, thus allowing appropriate treatments in high-risk populations. One of the latest biomarkers identified in patients with diabetes and metabolic syndromes are increased plasma levels of deoxysphingolipids (1-deoxySLs) (5,6), a type of sphingolipid characterized by an initial condensation of alanine or glycine instead of serine with palmitic acid and the resultant absence of the hydroxyl group in position C1. Consequently, although these deoxysphingoid bases can be acylated to deoxy-dihydroceramides, they cannot be further metabolized to complex sphingolipids or efficiently degraded by the canonical degradation pathway; thus, they tend to accumulate once produced. Importantly, 1-deoxySLs display toxic properties in vitro toward several cell lines (7–9), and in vivo, 1-deoxySLs are believed to impair neuronal functionality in patients with the hereditary sensory and autonomic neuropathy type I (10). In light of the increased plasma levels of 1-deoxySLs found in diabetic patients and of the reported cytotoxic effects associated with the exposure to increased 1-deoxySL concentrations, we investigated whether these atypical sphingolipids directly compromise pancreatic β-cells, the dysfunction of which plays an important role in the pathogenesis of both type 1 diabetes and T2DM.

Unless otherwise stated, all chemicals were purchased from Sigma and cell culture reagents from Gibco-BRL. Inhibitor stock solutions were freshly diluted to the concentrations required for the individual experiment indicated in the figure legends. Lipid stock solutions were prepared as previously described (10) as a bovine serum albumin (BSA) complex and added to the cells at the concentrations indicated in the figure legends. BSA was used as control.

The Ins-1 rat insulinoma cell clone 832/13, provided by C. Wollheim, was maintained in RPMI 1640 medium as previously described (11,12). Cell metabolic activity was tested with the 0.5% tetrazolium salt solution 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) or WST-1 (Roche) according to the manufacturer’s instructions. Cell death was quantified by trypan blue exclusion or lactate dehydrogenase (LDH) release in the medium (Roche). Cellular senescence was quantified with the β-galactosidase assay kit (Cell Biolabs). Adenovirus-expressing p21 (Adp21) (rat) and adenovirus-containing green fluorescent protein (AdGFP) were purchased from Vector Biolabs (Philadelphia, PA). Rac1 activity was measured with G-LISA Rac1 activation assay (Cytoskeleton, Denver, CO).

Wistar rats, leptin-deficient ob/ob mice on C57BL/6J background (B6.V-Lep/OlaHsd), and wild-type (WT) C57BL/6J (Harlan Laboratories) were kept under a light-dark regimen (16:8 h) at constant temperature and given free access to food and water. All animal experiments were performed in accordance with Swiss federal animal regulations and approved by the cantonal veterinary office of Zurich. Islets were harvested from pancreata of male Wistar rats (250–300 g) by collagenase (NB8 collagenase; Serva, Heidelberg, Germany) followed by trypsin digestion to dissociate them into single cells as previously described (13).

Dissociated islets cells were seeded in 12-well extracellular matrix (ECM)–coated plates (Novamed, Jerusalem, Israel); treated for 24 h with 5 μmol/L sphinganine, 1-deoxysphinganine, or BSA; and incubated in RPMI medium containing 3.3 mmol/L glucose for 1 h. Following sequential 1-h incubations with low (3.3 mmol/L), high (16.7 mmol/L), low (3.3 mmol/L) glucose concentrations, insulin secretion was measured by radioimmunoassay (Insulin-CT; CIS Bio International, Schering AG, Baar, Switzerland) according to the manufacturer’s instructions.

Total RNA was extracted from Ins-1 cells cultured in 1 μmol/L sphinganine or 1-deoxysphinganine for 24 h. Quality of RNA was assessed by a 2100 bioanalyzer (Agilent Technologies, Basel, Switzerland). cDNA was obtained with the RT2 First Strand Kit and profiled by the Rat Cell Death Pathway Finder PCR Array (both from SABiosciences, Hombrechtikon, Switzerland) according to the manufacturer’s instructions. Ceramide synthase (CerS) primers for SYBR green quantitative PCR are listed in the Supplementary Data.

Pancreas specimens were fixed in 4% formalin and paraffin embedded according to standard procedures (14). Ins-1 cells were fixed in 3.6% formaldehyde and permeabilized with 0.2% Triton X-100 in PBS. Primary antibodies used in this study are listed in the Supplementary Data. Apoptosis detection was performed with an ApopTag peroxidase kit (MP Biomedicals, Illkirch, France). Immunofluorescence analysis and image data collection were performed on a Zeiss Axioplan 2 imaging fluorescence microscope (Carl Zeiss Microimaging, Göttingen, Germany) or on a Leica SP2 AOBS confocal laser-scanning microscope (Leica Microsystems, Wetzlar, Germany) using a glycerol immersion objective lens (Leica, HCX PL APO CS 63× 1.3 Corr). Image z-stacks were collected with a pinhole setting of Airy 1 and twofold oversampling. Image stacks of optical sections were processed with the Huygens deconvolution software package version 2.7 (Scientific Volume Imaging, Hilversum, the Netherlands). Three-dimensional (3D) reconstruction, volume rendering, and quantification of signal overlap in the 3D volume model were done with the Imaris 7.2.1 software suite (Bitplane, Zurich, Switzerland). The degree of signal overlap in the 3D volume models is shown graphically as scatterplots by plotting the intensity of two fluorescent signals in each voxel of the 3D model. Voxels with similar signal intensity for both signals appear in the area of the diagonal. Single-cell quantification of stained cells by flow cytometry was performed with a FACSDiva flow cytometer (BD Biosciences, Allschwil, Switzerland).

Ins-1 cells cultured in 1 μmol/L sphinganine, 1-deoxysphinganine, or BSA for 24 h were lysed as previously described (15). Aliquots corresponding to 35 μg of proteins were separated by SDS-PAGE electrophoresis, blotted, and probed overnight at 4°C. Primary antibodies used in this study are listed in the Supplementary Data. Immunoreactive bands from at least three independent experiments were quantified by densitometry and normalized to β-actin or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels.

The sphingoid base profile was analyzed as described earlier (5). Ceramide species were extracted by adding 1 mL methanol/chloroform (2:1) [including the addition of 200 pmol C12 ceramide internal standard (Avanti Polar Lipids)] to 100 μL resuspended cells followed by 0.5 mL chloroform and 200 μL alkaline water (10). Interfering phospholipids were hydrolyzed by re-extracting the dried lipids with methanol-KOH:chloroform (4:1) as described earlier (10). Lipids were separated on a C18 column (Uptisphere 120 Å, 5 μm, 125 × 2 mm; Interchim, Montluçon, France) and analyzed on a TSQ Quantum Ultra mass spectrometer (Thermo Fisher Scientific, Reinach, Switzerland) using atmospheric pressure chemical ionization (10). Ceramides and deoxyceramides were identified by precursor ion scan (20 mV collision energy) with fragments of (m/z 264.3) and (m/z 268.3), respectively. Levels were normalized to ISTD and cell numbers.

Results are expressed as mean ± SEM. Significance was assessed with Student unpaired, two-tailed t tests or one-way ANOVA. P < 0.05 was considered significant. For overall P < 0.05, the Bonferroni multiple-comparison test was used to determine whether there was a significant difference between values of control (reference sample) and samples of interest.

Because 1-deoxySLs were found to be elevated in the plasma of diabetic patients in the low micromole range (5,6), we analyzed whether 1-deoxySLs can directly affect the viability of insulin-secreting cells. To this aim, the rat insulinoma cell line Ins-1 was treated at 50% confluence (Fig. 1A, where L indicates low density) for 24 h with 1-deoxysphinganine or sphinganine as control. 1-Deoxysphinganine incubation reduced both the metabolic activity, as measured by MTT (Fig. 1A) and WST-1 reduction (Supplementary Fig. 1A), and the number of live cells (Fig. 1B) in a dose-dependent manner. Treatment with 5 μmol/L caused cell roundup (Fig. 1C) and death, as shown by robust trypan blue inclusion (Fig. 1D) and LDH release (Fig. 1E). However, cells treated with 1 μmol/L 1-deoxysphinganine did not increase in number compared with the initial seeding but showed modest levels of lethality (Fig. 1D and E) or upregulation of genes involved in cell death pathways (Supplementary Fig. 1B–D), suggesting a cytostatic effect of the lipid at this concentration (Fig. 1B). When the lipid treatment was performed on 90% confluent cells (Fig. 1A, where H indicates high density), the metabolic activity was reduced only at the highest concentration of 1-deoxysphinganine tested, further indicating that 5 μmol/L 1-deoxysphinganine is cytotoxic for both dividing and quiescent cells, whereas lower lipid concentrations are cytostatic. In addition, treatment for only 1 h followed by washout and subsequent 23-h incubation was sufficient to reduce the metabolic activity of the cells comparably to a continuous 24-h incubation with 1-deoxysphinganine (Fig. 1F), suggesting a rapid effect of the lipid.

We then investigated whether the reduction of replication following low 1-deoxysphinganine concentration is mediated by induction of senescence. 1-Deoxysphinganine at 1 μmol/L increased β-galactosidase activity (Fig. 2A) and nuclear p21^WAF1/Cip1 expression (Fig. 2B and C), a hallmark (16) and inducer of senescence (17), respectively. To investigate whether increased p21 expression was sufficient to trigger the senescence pathway in Ins-1 cells, we used Adp21 or AdGFP as control (Fig. 2D). Adp21 infection decreased Ins-1 cell replication and increased β-galactosidase activity, whereas both parameters were unchanged following AdGFP incubation (Fig. 2E and F). In addition, the increased β-galactosidase activity after 1 μmol/L 1-deoxysphinganine treatment or Adp21 infection was accompanied by increased MTT reduction per cell (Supplementary Fig. 2), suggesting increased mitochondrial dehydrogenase activity, a parameter associated with cellular senescence (18). These data suggest that upregulation of p21 induced by 1 μmol/L 1-deoxysphinganine treatment contributes to the decreased Ins-1 replication by activating a senescence program.

Next, we further explored the cytotoxic effect of high 1-deoxysphinganine concentrations. The 5 μmol/L 1-deoxysphinganine abrogated the expression of p21 and induced cells with condensed pyknotic nuclei and high levels of activated caspase-3, hallmarks for the execution phase of apoptosis (Fig. 3A, Supplementary Fig. 3A). In addition, fluorescence-activated cell sorter (FACS) analyses of cells costained with annexin V and propidium iodide (PI) revealed that 5 μmol/L 1-deoxysphinganine treatment increased the number of both apoptotic and necrotic cells (Fig. 3B and C, Supplementary Fig. 3B) and induced the cells to arrest in the G0/G1 phase of the cell cycle (Supplementary Fig. 4), suggesting that the lipid triggers multiple cell death pathways in Ins-1 cells.

Because treatment with exogenous 1-deoxy-dihydroceramides (m18:0,24:1 and m18:0,16:0) and 1-deoxy-methylsphinganine, where alanine is replaced by glycine, reduced the cell replication similarly to 1-deoxysphinganine treatment (Fig. 4A), we tested whether exogenous 1-deoxysphinganine was also metabolized by the cells to the deoxy forms of ceramide. Incubation with 1-deoxysphinganine significantly increased the cellular levels of deoxy-dihydroceramide with various acyl chain lengths (Fig. 4B) without altering normal ceramide levels (Supplementary Fig. 5A) and selectively upregulated the expression of ceramide synthase 5 (CerS5) (Fig. 4C), suggesting that 1-deoxysphinganine is readily metabolized in Ins-1 cells and accumulates in the acylated form. In addition, pretreatment of Ins-1 cells with the class 2 amphiphile U18666A, a well-established inhibitor of NPC1 (Niemann-Pick type C) protein that prevents intracellular trafficking of sphingolipids and cholesterol (19,20), partially rescued 1-deoxysphinganine-mediated cytotoxicity (Fig. 4D, Supplementary Fig. 5B). Moreover, pharmacological inhibition of ceramide synthesis with fumonisin B1 (FB1) was unique in reducing the toxicity of 5 μmol/L 1-deoxysphinganine (Fig. 4D, Supplementary Fig. 5C), whereas inhibition of either the first step of sphingolipid synthesis with myriocin or glucosylceramide synthesis with PPMP (1-phenyl-2-palmitoylamino-3-morpholino-1-propanol) had no effect on cell viability (data not shown). Collectively, these data indicate that intracellular uptake followed by metabolic conversion to 1-deoxy-dihydroceramides is responsible, at least in part, for 1-deoxysphinganine toxicity.

To further characterize the biochemical components of 1-deoxysphinganine-induced cytotoxicity, we analyzed the phosphorylated status of key proteins involved in major signaling pathways. Western blot quantification revealed increased phosphorylation of Jun N-terminal kinase (JNK) and p38 mitogen-activated protein kinase (MAPK) in 1-deoxysphinganine-treated Ins-1 cells. In addition, the ratio of AKT phosphorylation was unchanged with lipid treatment, but the total protein amount increased in the presence of 1-deoxysphinganine (Fig. 5A–C). Because kinase activation by phosphorylation suggests a possible role of these proteins in the 1-deoxysphinganine-mediated phenotype, we tested this hypothesis by treating the cells with specific kinase inhibitors. Pretreatment with the p38 MAPK inhibitor Birb796 partially rescued cell replication (Fig. 5D) and reduced 1-deoxysphinganine-induced senescence (Fig. 5E). Conversely, the JNK inhibitor SP600125 potentiated cytotoxicity and senescence (Fig. 5D and E), and the combined treatment partially rescued the JNK inhibitor effect at 1 μmol/L 1-deoxysphinganine incubation. These data suggest that p38 MAPK activation is an effector of 1-deoxysphinganine cytotoxicity and senescence, whereas JNK activation plays a protective role in Ins-1 cells.

The observed changes in cell morphology observed following 1-deoxysphinganine treatment (Fig. 1C) prompted us to further analyze whether the lipid induced cytoskeletal alterations. To detect early cytoskeletal rearrangements, cells were imaged after 5 h of treatment. Phalloidin-stained actin filaments were mainly concentrated in the cortical area and in filopodia following control BSA or sphinganine treatment. However, on 1-deoxysphinganine incubation, actin staining accumulated in punctated structures mainly located in the perinuclear area of the cells. Different from the actin phenotype, tubulin staining showed a similar microtubule pattern in all treatments, suggesting that 1-deoxysphinganine preferentially interferes with the organization of actin cytoskeleton in Ins-1 cells (Fig. 6A). Quantitative analysis of members of the Rho family of guanosine triphosphatases (Rho-GTPases) that regulate intracellular actin dynamics (21) showed increased levels of Rac1 activity (Fig. 6B) and expression of Rac1 and RhoA, albeit the latter did not reach significance upon 1-deoxysphinganine treatment (Fig. 6C), suggesting that Rho-GTPases may be involved in 1-deoxysphinganine-mediated changes in actin cytoskeleton reorganization. Of note, actin alterations were not accompanied by apoptosis at this time point (data not shown). In addition, pretreatment with JNK, p38 MAPK, and CerS inhibitors did not prevent actin remodeling (Supplementary Fig. 6A), suggesting that cytoskeletal and cell cycle effects are modulated by different signaling pathways.

To further characterize whether aberrantly localized actin may intersect with the secretory apparatus of the cells, we performed costaining for actin and insulin. FACS-based single-cell quantification revealed that the lipid treatment did not alter the cellular content of actin and insulin (Supplementary Fig. 6B). However, confocal analyses showed that filamentous actin accumulated intracellularly in proximity to and partially colocalized with insulin-containing vesicles (Fig. 6D).

To confirm the relevance of the results in primary cells, we tested whether 1-deoxysphinganine affects the functionality of isolated islets. Lipid delivery to the cells was improved by dissociation of 1,400 islets from six Wistar rats into single cells; these were seeded in ECM-coated plates and treated for 24 h with 5 μmol/L sphinganine, 1-deoxysphinganine, or BSA as control. Of note, dissociated islets plated on ECM are virtually quiescent (R.A.Z., personal communication). Under these experimental conditions, 1-deoxysphinganine treatment induced cellular vacuolization (Fig. 7A) and reduced the metabolic activity (Fig. 7B) and, to a lesser extent, the number of live cells (Fig. 7C). However, senescence was not observed, as indicated by comparable β-galactosidase activities (Supplementary Fig. 7A) and absence of p21 staining (data not shown). Similarly to the effect in Ins-1 cells, 1-deoxysphinganine treatment induced rearrangement of actin cytoskeleton with the resulting accumulation of actin in intracellular punctated structures (Fig. 7D). 1-Deoxysphinganine did not alter the cellular content of insulin (Supplementary Fig. 7B) or selectively reduce the number of insulin-producing cells (Supplementary Fig. 7C). However, after incubation with the lipid, isolated β-cells were unable to regulate insulin secretion in response to glucose stimulation (Fig. 7E, Supplementary Fig. 7D). Collectively, these data indicate that 1-deoxysphinganine is cytotoxic to isolated primary islets and compromises both functionality and cytoarchitecture of β-cells.

The in vitro results showed 1-deoxySL-mediated cytotoxicity in insulin-producing cells, suggesting that raised levels of these lipids may contribute to the failure of β-cells during the development of diabetes. However, elevated 1-deoxySL levels were also found in the plasma of patients with metabolic syndrome who did not present with hyperglycemia and overt diabetes (5), raising the question that an increased amount of 1-deoxySLs is not the sole cause of β-cell toxicity and that additional factors contribute to the diabetic phenotype. To further investigate the causal network of atypical sphingolipids and β-cell toxicity, we analyzed pancreatic islets in leptin-deficient ob/ob mice. Kept on a normal chow diet, these mice develop obesity and a mild hyperglycemia that reverts with aging because pancreatic β-cell compensation occurs and increased insulin levels improve glucose homeostasis (22,23). Sixty-week-old ob/ob mice had only slightly higher plasma glucose levels than age-matched WT animals but a robust increase in plasma HDL, cholesterol, and alanine aminotransferase levels (Supplementary Fig. 8A), the latter reflecting steatotic liver damage (Supplementary Fig. 8B). As previously shown (23), ob/ob pancreata were characterized by pronounced islet hyperplasia, vascularization, and robust insulin production (Supplementary Fig. 8B), suggesting that β-cells could compensate for the increased insulin demand without reaching exhaustion. Despite the evident hyperplasia, islets did not show active replication at the analyzed age (Supplementary Fig. 8B). Of note, quantification of the plasma lipid profiles in ob/ob mice revealed a moderate, but significant increase of 1-deoxysphinganine levels (Fig. 8A), which was not associated with increased senescence or apoptosis (Supplementary Fig. 8B). In addition, like in human samples (5), sphingosine was the most abundant species in the mouse plasma and significantly increased in ob/ob mice (Supplementary Fig. 8C). Although plasma 1-deoxySLs likely did not reach a critical concentration threshold to induce β-cell failure, absence of other pathological parameters, including hyperglycemia, may account for the normal viability of β-cells. To test this hypothesis, we incubated Ins-1 cells with 1-deoxysphinganine in the presence and absence of glucose. Treatment with 30 mmol/L glucose for 24 h potentiated 1-deoxysphinganine-induced toxicity (Fig. 8B, Supplementary Fig. 8D), suggesting that hyperglycemia and 1-deoxySLs synergize to induce glucolipotoxicity in insulin-producing cells.

Elevated levels of 1-deoxySLs have been found in the blood of patients with metabolic syndrome and diabetes (5,6), raising the question of the role of atypical lipids in the development of these pathologies. The present results show that exposure to 1-deoxySLs in the low micromole range compromised insulin secretion and triggered senescence and cell death in insulin-producing cells, indicating that 1-deoxySLs are indeed toxic for these cells. In our experimental approach, we elucidated 1-deoxysphinganine toxicity at three distinct levels, namely, 1) changes in cellular structure, 2) engagement of signaling molecules, and 3) activation of effector proteins. A major finding of this study is that 1-deoxysphinganine-mediated toxicity is a complex phenomenon and triggers multiple pathways, including cytoskeletal remodeling, senescence, necrosis, and apoptosis.

1-Deoxysphinganine treatment selectively altered cytoskeleton organization both in Ins-1 cells and in primary islets, inducing accumulation of filamentous actin in intracellular punctated structures juxtaposed to insulin-containing vesicles. Similar but transient actin fiber alterations have been observed previously in 1-deoxysphinganine-treated Vero cells (7). In addition, 1-deoxysphinganine impaired cytoskeleton dynamics in sensory and motoneurons without inducing cell death (10). The present study shows not only that the lipid induced actin rearrangements before the appearance of apoptotic markers but also that inhibitors shown to mitigate cytotoxicity did not prevent cytoskeletal remodeling. These data suggest that cytoskeleton rearrangements are a direct effect of 1-deoxysphinganine incubation rather than a consequence of cell lethality and that cytoskeleton and cell cycle effects are regulated by distinct signaling pathways. In this context, whether alterations in Rho-GTPase activation, reported by Cuadros et al. (7) and in the present study, are the key conserved molecular mechanisms also behind cytoskeletal alteration in neurons is currently under investigation.

In the case of β-cells, the early cellular morphological alterations may impair cellular functionality, including insulin secretion. Indeed, earlier studies demonstrated that insulin secretion in pancreatic β-cells is coupled with reorganization of the filamentous actin web located beneath the plasma membrane, thus allowing docking of insulin-containing granules to the cell membrane and consequent secretion. Importantly, glucose stimulation directly induces rearrangement of the actin web (24,25) to which insulin-containing granules are in tight contact (26). Thus, impaired remodeling of actin cytoskeleton resulting from 1-deoxysphinganine treatment may contribute to the defective insulin secretion observed in the present primary β-cell cultures. However, we cannot exclude that additional lipid-induced changes, including altered gene transcription, may contribute to the phenotype.

Ceramide is a key intracellular signaling molecule involved in several cellular functions, including cell death (27). Importantly, both cell-permeant analogs of ceramide (28) and de novo ceramide synthesis (29) impair insulin secretion and mitogenesis in pancreatic β-cells and induce apoptosis (30), supporting a critical regulatory role for ceramide in the metabolic dysfunction of these cells. In the search for the molecular signaling generated by 1-deoxysphinganine treatment, we explored the hypothesis that 1-deoxysphinganine uptake and conversion to 1-deoxy-dihydroceramide is necessary to exert its toxicity. Our mass spectrometry and inhibitor analyses support this hypothesis. In addition, 1-deoxysphinganine increased the expression of CerS5, whereas β-cell lipotoxicity resulting from palmitate treatment stimulated the synthesis of CerS4 (31), suggesting that different lipids stimulate specific CerS isoforms. Collectively, the results suggest that some of the cytotoxic effects of 1-deoxysphinganine occur after its intracellular uptake and metabolism to 1-deoxy-dihydroceramide. However, we cannot exclude that 1-deoxysphinganine triggers death receptors on the cell surface. Given the similarity of the molecular structure of 1-deoxysphinganine and sphingosine, it is worthy of further investigation to determine whether 1-deoxysphinganine can engage or antagonize the same membrane receptors, as previously suggested (8).

In addition to cytoskeletal remodeling, 1-deoxysphinganine treatment induced a complex dose-dependent pattern of toxicity in Ins-1 cells characterized by the appearance of p21-induced senescence at low doses and apoptosis and necrosis at high doses. Of note, the senescence growth arrest was limited to replicating cells, as demonstrated previously (32–34), and quiescent primary islets were devoid of senescence markers upon lipid treatment. These data imply that 1-deoxysphinganine triggers multiple signaling pathways. Indeed, the lipid was shown to selectively activate JNK, MAPK, extracellular signal–related kinase 1/2, and protein kinase C but not AKT in NIH-3T3, RH-7777, PC-3, and LNCaP cell lines (8,9), whereas we show that JNK, p38 MAPK phosphorylation, and AKT levels increased in Ins-1 cells, indicating that the intracellular signaling effectors stimulated by 1-deoxysphinganine depend on the cellular context. Because JNK and MAPK are known to be activated by intracellular ceramide (35), it is possible that the increased 1-deoxy-dihydroceramide synthesis observed in Ins-1 cells activates these kinases. In addition, the fact that the CerS inhibitor FB1 was the only compound able to rescue the toxicity of high 1-deoxysphinganine doses suggests that increased ceramide (or its deoxy form) synthesis is upstream of or a prerequisite for the activation of different signaling effectors.

Although a more in-depth analysis of the set of kinases activated by 1-deoxysphinganine is needed to elucidate the precise downstream signaling cascade, the present inhibitor studies reveal an intriguing antagonistic role of JNK and p38 MAPK in the context of 1-deoxysphinganine-induced senescence. Of note, a similar dose-dependent senescence–apoptosis transition and opposite effect of p38 MAPK and JNK on senescence has been reported in endothelial progenitor cells upon doxorubicin treatment (36), further confirming that different kinases activated by the same stimulus may exert opposite cellular effects.

In conclusion, this work shows that 1-deoxysphinganine treatment and its conversion to 1-deoxy-dihydroceramide compromises the viability of insulin-producing cells through multiple pathways, indicating that similar to free fatty acids (30), 1-deoxySLs induce lipotoxicity but with higher efficiency [low micromoles per liter for 1-deoxySLs vs. low millimoles per liter for free fatty acids (31,37,38)]. However, the present in vivo study, together with the fact that patients with hereditary sensory and autonomic neuropathy type I and metabolic syndrome have elevated 1-deoxysphinganine levels without overt diabetes, suggests that the raised amount of 1-deoxysphinganine observed in vivo is not sufficient to directly induce β-cell failure but would require additional pathological parameters, such as an established chronic hyperglycemic state, to promote β-cell toxicity (39). In this context, targeting 1-deoxySL synthesis as a combination therapeutic strategy for T2DM warrants further investigation.

Acknowledgments. The authors thank Heidi Seiler for technical assistance and Amedeo Caflisch for providing the Birb796 inhibitor.

Funding. This research was supported by the Zurich Center for Integrative Human Physiology and Rare Disease Initiative Zurich, Clinical Research Priority Program for Rare Diseases, University of Zurich.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.A.Z., T.H., A.O., A.B.H., H.B., T.G., O.O.O., E.S., K.G., J.-H.J., U.U., Y.W., A.v.E., R.G., and S.S. contributed to the study design; data acquisition, analysis, and interpretation; and drafting and critical revision of the manuscript. R.A.Z., T.H., A.O., A.B.H., H.B., T.G., O.O.O., E.S., K.G., J.-H.J., U.U., and Y.W. researched data and reviewed and edited the manuscript. A.v.E. and R.G. contributed to the discussion and reviewed and edited the manuscript. S.S. researched data and wrote the manuscript. S.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 45th Annual Meeting of the European Pancreatic Club, Zurich, Switzerland, 26–29 June 2013.