Mechanisms of deoxysphinganine cytotoxicity in lymphoma cells Authors: Fabiola N Velazquez, Karissa J Lopez, Bahman Khalvatifahlylani, Daniel Canals, Yusuf A Hannun Published: 2026 ABSTRACT AbstractDeoxysphingolipids (dSLs) are atypical sphingolipids that accumulate in several pathological settings, yet their impact on hematologic malignancies is poorly understood. Here, we investigate the pathways and mechanisms of deoxysphinganine (dSA) cytotoxicity in lymphoma cells and its potential as a therapeutic agent. dSA exhibited markedly greater cytotoxicity than canonical sphingoid bases in lymphoma cell lines yet induced only cytostatic effects in normal human T cells, indicating a therapeutically exploitable window. Inhibition of ceramide synthase blocked the generation of deoxy(dihydro)ceramides, prevented mitochondrial depolarization, caspase activation, ER stress, and DNA damage, establishing CerS-dependent deoxysphingolipids as essential mediators of dSA-induced death. Mechanistically, dSA engaged a mitochondrial apoptotic pathway, with DNA damage occurring downstream of mitochondrial permeabilization and caspase activation, while PERK-driven ER stress occurred in parallel and was dispensable for cytotoxicity. Subtype-specific engagement of ER stress and DNA damage further suggests that dSL signaling is shaped by lineage context. The differential sensitivity between malignant lymphoid cells and normal T cells, together with the central role of CerS-derived deoxy(dihydro)ceramides, highlights deoxysphingolipid metabolism as a druggable vulnerability in lymphoma. These findings support further exploration of dSA-based strategies and targeted modulation of dSL synthesis as a novel therapeutic avenue for non-solid hematologic malignancies.Supplementary key words: sphingolipids, apoptosis, mitochondria, lymphocytes, ceramides, deoxysphingolipids, ceramide synthase, deoxysphinganine, ER stress Deoxysphingolipids (dSLs) are an atypical class of sphingolipids (SLs) produced when serine palmitoyltransferase (SPT) incorporates L-alanine instead of L-serine in the initial step of SL biosynthesis (1). The absence of a C1 hydroxyl group prevents their further conversion into complex SL species and precludes degradation through the canonical sphingosine-1-phosphate (S1P) lyase pathway (2). Under physiological conditions, circulating dSL levels are low; however, elevated dSL concentrations have been reported in several pathological conditions, including reduced L-serine availability (3, 4), increased plasma L-alanine (5), impaired glucose metabolism (6), exposure to chemotherapeutic agents (7), and SPT mutations that enhance affinity for L-alanine relative to the wild-type enzyme (8). Deoxysphinganine (dSA), the first intermediate in the synthesis of dSLs, was originally isolated from the marine clam Spisula polynyma and designated spisulosine (ES-285) (9). Spisulosine exhibits potent antiproliferative and pro-apoptotic activity in a broad range of solid tumor cell lines and has been evaluated in early-phase clinical trials as an experimental anticancer agent (10, 11). Exogenous dSA has therefore been used as a tool compound to dissect the mechanisms underlying dSL-induced cytotoxicity. In hepatoma cells, dSA triggers caspase activation preserving mitochondrial membrane integrity (12), while in prostate cancer cells dSA promotes the accumulation of de novo–synthesized ceramides that can activate atypical protein kinase C isoforms such as PKCζ (13). In several non-tumor and solid cancer cell models, dSLs have been shown to perturb membrane organization resulting in mitochondrial injury (14, 15, 16, 17) and endoplasmic reticulum (ER) stress (18). Although the precise mechanism of action of dSA differs between cellular contexts, N-acylation of dSA by ceramide synthases and the subsequent formation of deoxy(dihydro)ceramides are required for full dSL cytotoxicity (2, 19). Very recent genetic and metabolic studies further demonstrate that the synthesis of very-long-chain (C22–C26) deoxy(dihydro)ceramides is critical for dSL-mediated toxicity not only in solid cancer cells (17) but also in neuronal cell models (20), where these species localize to mitochondria, disrupt organelle function, and contribute to neurotoxicity (8, 16). Although multiple studies have begun to elucidate the cytotoxic actions of deoxysphingolipids, their impact on non-solid malignancies remains largely unexplored. Building on this gap, we investigate the mechanisms of deoxysphinganine-induced cytotoxicity in lymphoma cells, with the goal of defining the primary cellular targets and downstream death pathways engaged by this atypical sphingoid base and determining whether deoxysphingolipids can be harnessed as a therapeutic strategy in lymphoma. Material and methods Cell culture Human T lymphoblastic cell lines and MCF7 cells were grown in RPMI 1640 supplemented with 10% FBS (Fetal Bovine Serum), 100 units/ml penicillin, and 100 μg/ml streptomycin. T98G cells were grown in DMEM supplemented with 10% FBS (Fetal Bovine Serum), 100 units/ml penicillin, and 100 μg/ml streptomycin. Jurkat gRNA V2, BakA, and BakB cells were previously generated as described (21). Human normal T peripheral blood CD3+ Pan T cells were purchased from Human Cells Bioscience and culture in RPMI 1640 supplemented with 10% FBS, 10 ng/ml of human IL-2, 100 units/ml penicillin and 100 μg/ml streptomycin. Cells were evaluated for Mycoplasma contamination (Lonza Cat# #: LT07–710, MycoAlert PLUS Mycoplasma Detection Kit) before using them for analytical purposes. Deoxysphinganine solution preparation and cell treatment Deoxysphinganine (m18:0) (legacy code 860493, Avanti) was dissolved in methanol to prepare a 1 mM stock solution, which was aliquoted and stored at −80 °C. On the day of treatment, a working solution was prepared by diluting the stock 1:10 in complete culture medium (RPMI supplemented with 10% FBS), yielding a 100 μM dSA solution. This working solution was then added to the cells to achieve the indicated final dSA concentrations. Control cells were treated with an equivalent volume of vehicle (methanol diluted 1:10 in complete medium). Electroporation Knockdown of individual ceramide synthase isoforms was achieved using the Neon NxT Electroporation System (Thermo Fisher Scientific) following the manufacturer’s protocol. For each transfection, 2 × 106 cells were resuspended in Neon NxT Resuspension Buffer R and electroporated using a single 400 V pulse with a 30 s pulse width. Immediately after electroporation, cells were dispensed into culture vessels containing prewarmed complete growth medium and allowed to recover at 37 °C. Previously validated siRNA oligonucleotides targeting CERS1, CERS4, CERS5, CERS6, and a non-targeting control were used as described by Mullen et al. (22). For CERS2, 20 nM siRNA was used (Cat#: L-010282-00-0005, Dharmacon). For CERS2 knockdown, cells were electroporated with 20 nM ON-TARGET plus SMARTpool siRNA (Dharmacon, Cat# L-010282-00-0005). Knockdown efficiency was evaluated 48 hs after electroporation. Real-time PCR Total RNA was isolated from cells, and RNA integrity was confirmed prior to reverse transcription. For each sample, 1 μg of total RNA was reverse-transcribed using the SuperScript First-Strand Synthesis System for RT-PCR (Thermo Fisher Scientific) according to the manufacturer’s instructions. Quantitative real-time PCR was performed on a QuantStudio 3 Real-Time PCR System (Applied Biosystems) using TaqMan Gene Expression Assays. Reactions were run in technical triplicates. Gene expression was normalized to ACTB as the endogenous control. Viability assay Viability of Molt4 and Jurkat cells was measured by CCK-8 assay (cat#96992, Sigma-Aldrich) according to the manufacturer’s instructions. Comparison of sensitivity between adherent and non-adherent cell lines against dSA treatment was performed using alamarBlue Cell viability reagent (Invitrogen). Cells were incubated with the reagent alamarBlue (10 μl) dissolved in 90 μl of DMEM supplemented with 10% of FBS (final volume: 100 μl in 96 multiwell plate), during 4 h at 37°C. Fluorescence measurements were done using a SpectraMax Id3 microplate reader with an excitation wavelength of 540–570 nm and the fluorescence emission reads at 580–610 nm. Viability of Molt4 and Jurkat cells was measured by CCK-8 assay (cat#96992, Sigma-Aldrich) according to the manufacturer’s instructions. Viability of different human T lymphoblastic cell lines was measured by MMT assay as described (23). Cells were incubated with MTT 1 mg/ml during 2 h and then cells lysed in 10% SDS + 0.01 N HCl solution. LDH release assay LDH release was measured using the LDH Cytotoxicity Assay Kit (ab65393; Abcam) according to the manufacturer's protocol. Cells were seeded in 24-well plates at a density of 5 × 104 cells per well in complete medium, and treatments were applied the following day. After 24 h, cells were collected, centrifuged at 600 × g for 5 min, and LDH activity was evaluated in the supernatant. Western blot Primary antibodies used for Western blot included: Parp1 (1:2000, cat#9542, Cell Signaling Technology), cleaved caspase 3 (1:1000, cat#9661, Cell Signaling Technology), CHOP (1:1000, cat#2895, Cell Signaling Technology), eIF2α (1:1000, cat#9722, Cell Signaling Technology), phopho-eIF2α (1:1000, cat#9721, Cell Signaling Technology), p-γH2A.X (1:1000, ab11174, Abcam), ATF6 (1:1000, cat#853101, Biolegends), ATF4 (1:1000, cat#693901, Biolegends), and Beta-Actin (1:25,000, Cat. #A5441, Sigma-Aldrich). After incubation with the appropriate secondary antibody, signal was detected using enhanced chemiluminescence or Supersignal West Dura Extended Duration Substrate (Life Technologies) according to the manufacturer’s instructions. LC–MS/MS sphingolipid analysis Cells were collected by centrifugation at 800 xg for 5 min. Then, cells were washed once in cold PBS, centrifuged at 800 xg for 5 min, and the cellular pellet was collected in 2 ml of cell extraction solvent (isopropanol: water: ethyl acetate (30:10:60; v:v:v). An internal standard mix was added to the sample at this step. Samples were kept at −80°C until lipid extraction. Lipids were extracted by vortexing, sonication for 30s, and centrifugation at 3000 ×g for 10 min. The supernatant was transferred to a new tube, and the pellet was re-extracted. Supernatants were aliquoted into 1 ml for inorganic lipid phosphate determination, and the remainder for lipid measurements. Extracts were then analyzed by the Lipidomic Core Facility at Stony Brook University Medical Center, as described previously (24). Lipid levels were normalized to total inorganic lipid phosphate. Cell counting Cell suspension samples were diluted (1:1 dilution) in 0.4% trypan blue solution (ThermoFisher Scientific). Neubauer chambers were loaded with the stained cell suspension, and cells (non-viable (blue) and viable (unstained)) were counted under the microscope. For the proliferation assay, the number of viable cells over time was counted. For the cell death assay, the percentage of trypan blue positive cells over the total number of cells (at a specific time point) was plotted. Mitochondria membrane potential measurement Mitochondrial membrane potential was assessed using TMRE (tetramethylrhodamine ethyl ester). Cells were harvested after treatment and centrifuged at 500 × g for 5 min, and the pellet was resuspended in RPMI containing 2% FBS and 20 nM TMRE. Cells were incubated for 45 min at 37°C in a humidified cell culture incubator protected from light, washed twice with 1× PBS, and finally resuspended in 1× PBS. TMRE fluorescence was measured by flow cytometry (BDAccuri C6 Plus) using a 488 nm excitation laser and a 570 ± 10 nm bandpass emission filter. Statistical analysis Statistical analyses were performed using GraphPad Prism (GraphPad Software). Data are presented as mean ± SEM and were analyzed by one-way ANOVA with a Dunn’s Multiple Comparison Test, two-way ANOVA or by unpaired Student’s t test. P < 0.05 was considered statistically significant. Results dSA is a highly toxic sphingoid base against lymphoma cells Deoxysphinganine (dSA) has shown a strong anti-tumor effect against different solid cancer cell lines. However, its efficacy against non-solid cancer cells has not yet been explored. To compare the efficacy of dSA in adherent versus non-adherent cancer cell lines, the IC50 for dSA was determined in MCF7 (breast cancer) and T98G (glioblastoma), representatives of adherent cancer cells, and on Molt4 and Jurkat cells as examples of non-adherent cancer cell lines. Analysis of cell viability by alamarBlue assay showed that Jurkat and Molt4 cells were more susceptible to dSA (IC50 = 512 nM and 483 nM, respectively) compared to MCF7 and T98G cells (IC50 = 1233 nM and 1616 nM, respectively) (Fig. 1A), suggesting a stronger anti-tumor effect for dSA against non-adherent cancer cells. We then compared the toxic effect of dSA against other sphingoid bases in Molt4 cells (Fig. 1B). While dSA treatment resulted in cell death, sphingoid bases such as sphinganine, L-erythro sphingosine and deoxysphingosine did not have any effect. D-erythro sphingosine, a canonical sphingoid base strongly associated with cell death and cell cycle arrest (25, 26), also resulted in cell death but its potency was lower compared to dSA. At 1.5 μM, dSA caused ∼82% cell death, whereas D-erythro-sphingosine resulted in ∼23% cell death. These results were confirmed using cck-8 viability assay, 0.75 μM dSA significantly decreases cell viability in Molt4 and Jurkat cells, while D-erythro-sph required higher concentrations to start having an effect (Fig. 1C). The cell death effect of dSA started at ∼0.75 μM (Fig. 1D). These results demonstrate a strong cytotoxicity for dSA against non-solid cancer cells at low micromolar concentrations. dSA is a highly toxic sphingoid base against lymphoma cells. A: IC50 determination for dSA on adherent and non-adherent cancer cell lines. Cells were treated with different concentrations of dSA and cell viability determined by alamarBlue assay after 48 hs of treatment. Viability of vehicle-treated cells (0 μM dSA) is defined as 100%, and all other values are expressed relative to this control. IC50 for each cell line was calculated from 3 independent experiments. B: Molt4 cells were treated with different concentration of the following spingoid bases: dSA, D-erythro sphingosine (D-Sph), dihydrosphingosine (dhSph), 1-deoxysphingosine (dSO), L-erythro sphingosine (L-Sph). Hexadecyl-amine is a long chain alkylamine (C16H35N) that was also included in the analysis. Cell death was quantified 48 hs after treatment by determining the percentage of trypan blue positive cells. C: Molt4 and Jurkat cells were treated with different concentrations of dSA and D-Sph. Cell viability was measured after 48 hs of treatment using cell counting kit 8 and absorbance measured at 450 nm. D: Molt4 cells were treated with different concentrations of dSA. Cell death was determined after 24 hs of treatment by trypan blue staining (upper panel) and LDH release assay (lower panel). All results represent the mean ± SD from 3 independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗∗P ≤ 0.01; ∗∗∗∗P ≤ 0.0001. Ceramide synthase activity is essential for the toxic effect of dSA on lymphoma cells Previous studies have shown that ceramide synthase (CerS) activity is needed for the biological effect of dSA in non-tumoral (27) and solid cancer cells (28), therefore we evaluated if this was also the case in non-solid cancer cells. Fumonisin B1 (FB1) was used as an inhibitor of CerS activity. Pre-treatment with FB1 completely protected against cell death caused by dSA in Molt4 cells, determined by trypan blue staining (Fig. 2A) and LDH release assay (Fig. 2B). Measurement of sphingolipid levels by LC/MS determined that FB1 significantly prevented the increase of different deoxydihydroceramides and deoxyceramides 14Z species after dSA treatment in Molt4 cells (Fig. 2C), pointing to these deoxysphingolipid species as possible mediators for the toxic effect observed after dSA treatment. A similar protective effect of FB1 on LDH release (Supplementary Fig. S1A) and deoxydihydroceramides/deoxyceramides 14Z accumulation (Supplementary Fig. S1B) was observed in Jurkat cells, indicating that CerS-dependent deoxysphingolipid synthesis is broadly required for dSA cytotoxicity in T-cell lymphoma models. Sphingosine and canonical ceramides, two classical SL associated with cell death, did not change significantly after dSA treatment (Supplementary Fig. S1, C and appsec1D). In contrast, treatment with 1 μM sphingosine markedly increased cellular sphingosine and ceramide levels (data not shown) but did not induce cytotoxicity (Fig. 1C). These findings support the conclusion that dSA-induced cell death is primarily attributable to dSL accumulation rather than alterations in canonical ceramides. The analysis of the different deoxydihydroceramides/deoxyceramides 14Z species revealed that dSA induces a broad elevation of multiple species, with the largest absolute increase in C16 species, but with substantial accumulation of very-long-chain species as well. To identify the specific CerS isoform responsible for dSA toxicity, we performed targeted individual CERS knockdown in Jurkat cells. Silencing CERS1 (Fig. 2D), CERS4 (Fig. 2E), CERS5 (Fig. 2F) or CERS6 (Fig. 2G) did not significantly alter LDH release after dSA treatment, suggesting that long-chain ceramide synthesis is dispensable for this effect. In contrast, CERS2 knockdown (Fig. 2H and Supplementary Fig. S2A) significantly protected Jurkat cells from dSA-induced cell death (Fig. 2I). Measurement of classical SLs showed that siRNA against CERS2 did not cause a change in total ceramide, sphingosine, or S1P levels while specifically reducing C24-ceramide (Supplementary Fig. S2B), confirming the specificity of the manipulation. Importantly, CERS2 knockdown abolished dSA-induced accumulation of very long chain deoxydihydroceramides/deoxyceramides 14Z, and also reduced C16 species (Fig. 2J). Despite the reduction in C16-deoxydihydroceramides/deoxyceramides 14Z, individual knockdown of CERS5 and CERS6 did not significantly impact LDH release, arguing that CerS2-derived very-long-chain deoxydihydroceramides/deoxyceramides 14Z, rather than CerS5/6-derived C16 species, are the critical toxic metabolites. Our findings are consistent with recent reports in HeLa cells (17) and neurons (20), where loss of CERS2 or other components of the very-long-chain ceramides synthesis pathway conferred strong protection against dSA-induced cytotoxicity. Taken together, these results demonstrate that targeting CerS2-mediated deoxysphingolipid synthesis can effectively uncouple dSA exposure from downstream cell death in lymphoma cells. Ceramide synthase activity is essential for the toxic effect of dSA. Molt4 cells were pre-treated with 25 μM FB1 for 1h, and then different concentrations of dSA were added. Cell death was quantified after 24 hs of dSA treatment by trypan blue staining (A) and LDH release assay (B). Results are the mean ± SD from 3 independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. C: Deoxysphingolipid species were measured by LC/MS in cell extracts from Molt4 cells after 24 hs of treatment with dSA in the presence or absence of 25 μM FB1. Results represent total pmol of lipid/nmol of inorganic lipid phosphate (Pi). 3 independent samples per each condition were analyzed. Ordinary one-way ANOVA. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. D-G: Jurkat cells were transfected with control siRNA (AS) and siRNA targeting CERS1 (D), CERS4 (E), CERS5 (F) and CERS6 (G). The upper panels show the mean normalized expression of the corresponding CERS gene. Statistical analysis: Unpaired t test with Welch’s correction; ∗P ≤ 0.05. The lower panels show cell death in AS- and siCERS- transfected cells, determined by LDH release assay 24 h after dSA treatment. Statistical analysis: Two-way ANOVA followed by Šídák's multiple comparisons test. H: CERS2 mean normalized expression in Jurkat cells transfected with control siRNA (AS) and siRNA against CERS2. Statistical analysis: Unpaired t test with Welch’s correction. ∗∗∗∗P ≤ 0.0001. I: Cell death in AS and siCerS2 Jurkat cells determined by LDH release assay 24 hs after dSA treatment. Statistical analysis: Two-way ANOVA, Šídák's multiple comparisons test. ∗∗∗P ≤ 0.001. J: Deoxysphingolipid species were measured by LC/MS in cell extracts from Jurkat cells transfected with AS or siCERS2 after 24 hs of treatment with vehicle or 0.75 μM dSA. Results represent total pmol of lipid/nmol of inorganic lipid (Pi). 3 independent samples per each condition were analyzed. Statistical analysis: Two-way ANOVA followed by Šídák's multiple comparisons test. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. dSA triggers a mitochondrial apoptotic cell death Subcellular tracking of alkyne-dSA has shown that exogenous dSA primarily localized to mitochondria (14), hence we decided to evaluate mitochondrial integrity after dSA treatment. Treatment with dSA resulted in a decrease in mitochondrial membrane potential in Molt4 (Fig. 3A) and Jurkat cells (Supplementary Fig. S3). Activation of the intrinsic apoptotic pathway was measured by activation/cleavage of caspase 3 and cleavage of PARP (Fig. 3B). Pre-treatment with the pan-caspase inhibitor ZVAD completely protected against caspase activation and partially prevented cell death (LDH release) caused by dSA in Molt4 cells (Fig. 3C). The importance of mitochondria for dSA cytotoxicity was confirmed in Jurkat cells with knockout of the pore-forming protein Bak generated in-house (21). Here, Jurkat sgRNA Bak cells were protected against mitochondrial membrane depolarization (Fig. 3D) and caspase activation (Fig. 3E) after dSA treatment, demonstrating a requirement for the mitochondrial pathway of apoptosis. dSA induced-cell death is partially dependent on the mitochondrial membrane depolarization and the subsequent activation of caspases. A: Mitochondrial membrane depolarization was evaluated by flow cytometry in Molt4 cells treated with 0.75 μM dSA. The right panel shows the percentage of TMRE-low cells, indicative of reduced mitochondrial membrane potential (ΔΨm). Results are the quantification from three independent experiments. Statistical analysis: Unpaired t test with Welch’s correction; ∗∗P ≤ 0.01. B: Western blot analysis for the detection of cleaved Parp1 (cParp1) and cleaved caspase 3 (cCasp3) was performed in protein extracts from Molt4 cells treated with 0.75 μM dSA at the indicated time points. Actin was used as a loading control. C: Molt4 cells were treated with 0.75 μM dSA in the presence or absence of 40 μM ZVAD. The upper panel shows Western blot analysis of cParp1 and cCasp3. Quantification for these markers from three independent experiments is shown in the panel on the right. Cell death was assessed by LDH release assay (lower panel). Statistical analysis: Ordinary one-way ANOVA. ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. D: Mitochondrial membrane depolarization was evaluated by flow cytometry in Jurkat sgRNA V2 (control cells), Jurkat sgRNA BakA and Jurkat sgRNA BakB cells treated with 0.75 μM dSA. The right panel shows the quantification from three independent experiments. Statistical analysis: Unpaired t test with Welch’s correction; ∗∗∗∗P ≤ 0.0001. E: cParp1 and cCasp3 were evaluated by Western blot analysis in protein extracts from Jurkat sgRNA V2 (control cells), Jurkat sgRNA BakA and Jurkat sgRNA BakB cells treated with vehicle (ctr) or 0.75 μM dSA. Actin was used as a loading control. Quantification for cParp1 and cCasp3 from three independent experiments is shown on the right panel. Statistical analysis: Two-way ANOVA followed by Šídák's multiple comparisons test. ∗P ≤ 0.05; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. dSA results in PERK activation independent of its cytotoxic effect but dependent on ceramide synthase activity In MEF cells and photoreceptor cells, dSLs have been shown to localize to the ER and induce the ER stress response (18, 29). In non-solid cancer cells, dSA treatment led to an increase of CHOP protein, denoting induction of ER stress (Fig. 4A). There are three main ER sensors whose activation can drive to ER stress: PERK, IRE1, and ATF6 (30). Analysis of protein expression demonstrated that dSA treatment triggered PERK activation (Fig. 4B, Supplementary Fig. S1E), while the remaining two UPR activator proteins, IRE1 (Fig. 4C) and ATF6 (Fig. 4D), were not affected. Activation of PERK by dSA was confirmed by using the PERK inhibitor GSK2606414. Pre-treatment with GSK2606414 impeded the increase in p-eIF2α and ATF4 (downstream targets of PERK activation) after dSA treatment (Fig. 4E). Importantly, pretreatment with GSK2606414 did not protect against caspase activation (Fig. 4F) and LDH release (Fig. 4G) in Molt4 cells, suggesting that PERK activation is not involved in the cytotoxic effect of dSA. dSA treatment results in ER stress dependent on PERK activation and ceramide synthase activity but independent of its mitochondrial effect. A: Expression of CHOP was detected by Western blot analysis in protein extracts from Molt4 cells after 24 hs of treatment with dSA 0.75 μM. Quantification for relative CHOP expression from three independent experiments is shown on the right panel. Statistical analysis: Unpaired t test with Welch’s correction; ∗∗P ≤ 0.01. B-D: ER stress markers were evaluated by Western blot analysis in protein extracts from Molt4 cells treated with 0.75 μM dSA. Treatment with 0.5 μM thapsigargin was used as positive control of ER stress activation. PERK (B) and Ire1α (C) activation were analyzed at different time points. ATF6 cleavage (D) was evaluated after 16 hs of treatment with dSA. E-G. Molt4 cells were treated with 0.75 μM dSA in the presence or absence of PERK inhibitor (GSK2606414). Expression of ATF4, phospho-eIF2α (E) and cCasp3 (F) were evaluated by Western blot analysis. Quantification of these markers from three independent experiments is shown on the panels at the bottom. Cell death was assessed by LDH release assay 24 hs after dSA treatment (G). Statistical analysis: Ordinary one-way ANOVA. ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001. H-J: Representative blots of markers of PERK activation in protein lysates from Jurkat sgRNA V2 (control cells), Jurkat sgRNA BakA and Jurkat sgRNA BakB cells treated with 0.75 μM dSA (H) and from Molt4 cells treated with 0.75 μM dSA in the presence or absence of 40 μM ZVAD (I) or 25 μM FB1 (J). Actin was used as a loading control. Bottom panels show the relative expression of the indicated proteins quantified from three independent experiments. Statistical analysis: Two-way ANOVA followed by Šídák's multiple comparisons test for panel H and ordinary one-way ANOVA for panels I-J. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001. ER stress and mitochondrial dysfunction are closely interconnected (31). The lack of protection against cell death caused by dSA upon PERK inhibition suggested that ER stress is not upstream of the mitochondrial apoptotic cell death. Nevertheless, different findings have shown that mitochondria dysfunction could trigger ER stress. To determine if mitochondria membrane depolarization caused by dSA was upstream of ER stress, we evaluated PERK activation in Jurkat sgRNA Bak cells (Fig. 4H) and in the presence of caspases inhibition (Fig. 4I). Analysis of p-eIF2α and ATF4 showed that PERK activation by dSA was independent of mitochondria membrane depolarization and the subsequent caspases activation. Since ER stress observed after dSA treatment is not involved in its apoptotic effect, we next tested if CerS activity was also required for this effect. Similar to the protection for cell death (Fig. 2A, B), pre-treatment with FB1 prevented the increase in p-eIF2α and ATF4 caused by dSA treatment (Fig. 4J). These results suggest that the impact on mitochondria and ER membranes caused by dSA is uncoupled, but both require CerS activity. dSA results in DNA damage downstream of mitochondrial membrane depolarization and caspase activation Extensive evidence has shown that DNA damage can induce caspase activation and apoptosis by engaging the mitochondrial pathway (32); conversely, caspase activation has been shown to promote DNA damage through the activation of caspase-activated DNase (CAD) (33). Since dSA treatment triggered mitochondrial apoptotic cell death, we examined whether DNA damage was involved in dSA cytotoxicity. In both Molt4 (Fig. 5A) and Jurkat (Supplementary Fig. S1E) cells, dSA treatment resulted in DNA damage detected by an increase in p-γH2A.X. Results showed that the Jurkat sgRNA Bak cells were resistant to dSA (Fig. 5B), demonstrating that the DNA damage caused by dSA treatment is downstream of its mitochondrial membrane depolarization effect. Moreover, pre-treatment with the pan-caspase inhibitor ZVAD completely protected against DNA damage (Fig. 5C), whereas PERK inhibition did not have an effect. Consistent with a requirement for CerS activity in mediating the mitochondrial response to dSA, pharmacological CerS inhibition in MOLT4 cells markedly reduced dSA-induced PARP and caspase-3 cleavage (Fig. 5D) and attenuated the associated DNA damage (Fig. 5E). A similar protection from dSA-induced PARP and caspase-3 cleavage was observed in Jurkat cells (Supplementary Fig. S1E), further supporting the role of CerS in these mitochondrial events preceding DNA damage. These results suggest that dSA causes DNA damage in lymphoma cells downstream of the mitochondria membrane depolarization and caspases activation and independent of ER stress. dSA treatment results in DNA damage downstream of its mitochondrial effect but independent of ER stress. A-B: Analysis of the DNA damage marker p-γH2A.X in Molt4 cells (A) and in Jurkat sgRNA V2 (control cells), Jurkat sgRNA BakA, and Jurkat sgRNA BakB cells treated with 0.75 μM dSA (B). The right panel in B shows p-γH2A.X expression relative to actin quantified from three independent experiments. Statistical analysis: Two-way ANOVA followed by Šídák's multiple comparisons test. ∗P ≤ 0.05; ∗∗P ≤ 0.01. C: Molt4 cells were pre-treated for 1h with 40 μM ZVAD or 0.5 μM GSK2606414 and then treated with 0.75 μM dSA for 20 hs. Protein samples were extracted and evaluated by Western blot for the indicated markers. Right panel shows p-γH2A.X expression relative to actin quantified from three independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. D-E: Representative Western blot of cell death markers (D) and the DNA damage marker p-γH2A.X (E) in Molt4 cells treated with 0.75 μM dSA in the presence or absence of 25 μM FB1. The right panels show quantification of the indicated markers from three independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. dSA induces cell death in lymphoma cells with no significant effect on normal human T cells To determine the potential use of dSA for the treatment of non-solid tumors, the viability of different non-solid cancer cell lines was measured after dSA treatment. The malignant DB, SUPT1 and HL60 cells showed a significant decrease in cell viability after 24 hs of treatment with 0.5 μM dSA (Fig. 6A). Analysis of protein samples suggested that Molt4 and DB cells were more sensitive to the activation of cell death and DNA damage by dSA than Jurkat and SUPT1 cells. Interestingly, HL60 cells, a promyelocytic cell line, was less sensitive to caspases activation and DNA damage while not showing ER stress upon dSA treatment (Fig. 6B). These results suggest that dSA decreases cell viability and activates caspases in different nonsolid cancer cells lines; however, the induction of ER stress and DNA damage seems to be dependent on the cell subtypes. dSA causes cell death in non-solid cancer cell lines with no significant effect on normal human T cells. A: MTT viability assay was performed in different nonsolid cancer cell lines after 24 hs of treatment with dSA. Results are the mean ± SD from 3 independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001. B: Markers of cell death, ER stress and DNA damage were analyzed by Western blot in protein extracts from different nonsolid cancer cell lines after 20 hs of treatment with 0.75 μM dSA. Control samples were treated with vehicle. Actin was used as a loading control. The table at the bottom summarizes the markers detected in the different cancer cell lines. The panels at the right show the quantification of the indicated markers from three independent experiments. Statistical analysis: Unpaired t test with Welch’s correction. ∗P ≤ 0.05; ∗∗P ≤ 0.01. C-D: Normal human T cells from two different healthy patients (#1 and #2) were treated with dSA at the indicated concentrations. The number of cells was quantified at day 1, 3 and 5 after treatment (C). Cell death was determined by LDH release assay 24 hs after dSA treatment (D). E: Mitochondrial membrane potential in normal human T cells from two different healthy patients (#1 and #3) treated with dSA was assessed by flow cytometric analysis of TMRE staining. Results are presented as the percentage of TMRE-low cells, indicative of reduced mitochondrial membrane potential (ΔΨm). Statistical analysis: Unpaired t test with Welch’s correction. F: Deoxysphingolipid species were measured by LC/MS in cell extracts from normal human T cells from patient #3 following 24 hs of treatment with 0.75 μM dSA. Results are expressed as pmol lipid per nmol inorganic lipid phosphate (Pi). Three independent samples were analyzed per each condition. Statistical analysis: Unpaired t test with Welch’s correction for deoxysphinganine, deoxysphingosine and total deoxy(dihydro)ceramides, and two-way ANOVA followed by Šídák's multiple comparisons test for individual deoxy(dihydro)ceramide species. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. G: Proposed mechanisms of cellular action for dSA in lymphoma cells. An important aspect of therapy, especially in cancer treatment, is the effect of the drug on healthy cells. To analyze the effect of dSA on non-cancer cells, normal human T cells from three different patients were evaluated. The results showed that 0.75 μM dSA decreased the number of cells over time (Fig. 6C); however, at this same concentration dSA did not cause cell death (Fig. 6D) or alter mitochondrial membrane potential (Fig. 6E), suggesting a cytostatic but not a cytotoxic effect for dSA on normal T cells. Importantly, treatment of normal T cells with dSA increases deoxydihydroceramides/deoxyceramides 14Z (Fig. 6F and Supplementary Fig. S4) resulting in a similar profile at what was observed in Jurkat cells (Fig. 2J). These findings indicate that, despite similar dSL accumulation, primary T cells are relatively resistant to dSA-induced mitochondrial dysfunction, underscoring a selective susceptibility of malignant hematopoietic cells to dSL-driven cytotoxicity. Discussion Deoxysphingolipids (dSLs) are increasingly recognized as bioactive lipids that accumulate in a range of pathological diseases (34, 35, 36), yet their impact on hematologic malignancies has been not determined. Here, we identify deoxysphinganine (dSA) as a highly potent cytotoxic sphingoid base against lymphoma and other non-solid cancer cells and delineate a CerS- and mitochondria-dependent cell death pathway. In line with previous work in solid tumors (37) and neuronal models (20), dSA displayed stronger cytotoxicity than canonical sphingoid bases, including D-erythro-sphingosine, and its toxicity was selective for malignant lymphoid cells over normal T lymphocytes. Together, these findings position dSA-derived deoxy(dihydro)ceramides as key drivers of lymphoma cell death and support the concept that dSL metabolism may be therapeutically exploitable in non-solid malignancies. Our data extend prior observations that N-acylation of dSA by ceramide synthases is required for dSL toxicity (38) by demonstrating an essential role for CerS activity in lymphoma cells. Pharmacologic CerS inhibition with fumonisin B1 fully prevented dSA-induced cell death, blocked accumulation of deoxy(dihydro)ceramides, and abrogated downstream ER stress and DNA damage, implicating these atypical ceramides as central mediators rather than dSA itself. Consistent with recent reports in HeLa cells (17), knockdown of CERS2 significantly protected Jurkat cells from dSA-induced death, indicating that very-long-chain deoxy(dihydro)ceramides are particularly important for the pro-death response in non-solid cancer cells. Similarly in neuronal models, CERS2 drives the synthesis of very-long-chain deoxyceramides that results in ER stress and neurotoxicity (20). These results suggest that the chain-length–specific ceramide repertoire generated by different CerS isoforms critically shapes the cytotoxic versus adaptive outcomes of dSL accumulation and highlight CerS2-dependent very-long-chain deoxy(dihydro)ceramides as prime effectors in lymphoma. Mechanistically, our study supports a model in which the mitochondrial apoptotic pathway acts as a central effector of dSA cytotoxicity in lymphoma cells, rather than establishing mitochondria as the primary upstream target. dSA treatment led to mitochondrial membrane depolarization, activation of caspase-3, PARP cleavage, and caspase-dependent cell death. The requirement for pore-forming Bcl2 proteins underscores the importance of mitochondrial outer membrane permeabilization in the lethal response to dSA and suggests that deoxy(dihydro)ceramides could directly or indirectly perturb mitochondrial membranes to favor Bcl-2 family–driven pore formation. Given that alkyne-dSA has been shown to localize to mitochondria (14), it is tempting to speculate that very-long-chain deoxy(dihydro)ceramides may accumulate at mitochondrial contact sites, where their atypical biophysical properties could destabilize membranes, alter cristae architecture, or modulate the activity of mitochondrial death regulators. Future studies using organelle-resolved lipidomics and reconstitution approaches will help to dissect how dSLs remodel specific mitochondrial subdomains to trigger apoptosis. Importantly, in our system, ZVAD pretreatment reduced dSA-induced cell death by ∼50%. This incomplete protection suggests the presence of a caspase-independent death component. Such caspase-independent mechanism may reflect the biophysical integration of deoxy(dihydro)ceramides into cellular membranes, leading to loss of membrane integrity and organelle dysfunction, and could account for the residual cell death observed in the presence of ZVAD. However, this has to be evaluated in light of the specificity for dSA/deoxy(dihydro)ceramides as higher levels of sph/(dh)Ceramides did not reproduce these effects. Future studies dissecting alternative regulated cell death pathways (e.g. necroptosis or ferroptosis) will be important to fully define the non-apoptotic mechanism engaged by dSL accumulation. Our findings also clarify the relationship between dSA-induced mitochondrial injury and nuclear DNA damage. dSA treatment increased γH2A.X phosphorylation in Molt4 and Jurkat cells, and this response was abolished in Bak-deficient cells and by pan-caspase inhibition but remained unaffected by PERK blockade. These results place DNA damage downstream of mitochondrial membrane depolarization and caspase activation, consistent with caspase-activated DNase driving DNA fragmentation during apoptosis (33) rather than primary genotoxic stress initiating the death cascade. Although our current data do not exclude the possibility that dSA or its metabolites might also interfere with DNA replication or repair under certain conditions, they strongly support a model in which mitochondrial apoptosis is the primary initiating event, with DNA damage serving as a downstream hallmark of cell demise. In contrast to the clear requirement for mitochondria, our data indicate that PERK-dependent ER stress is a context-dependent but non-essential response to dSA in lymphoma cells. dSA treatment induced CHOP expression and selectively activated PERK, without engagement of IRE1 or ATF6, mirroring previous observations that dSLs can elicit a partial unfolded protein response (18). However, pharmacological PERK inhibition efficiently blocked eIF2α phosphorylation and ATF4 upregulation but failed to protect from caspase activation or loss of viability, arguing against a major contribution of PERK signaling to the execution of cell death. Furthermore, PERK activation persisted in Bak-deficient cells and under caspase inhibition, demonstrating that ER stress neither lies upstream of the mitochondrial apoptotic pathway nor is solely a consequence of apoptosis. Instead, PERK activation was abolished by CerS inhibition, indicating that deoxy(dihydro)ceramides perturb ER homeostasis in parallel to mitochondria, likely by altering membrane composition or protein/lipid trafficking. A recent study in retinal pigment ephitelial cells used trace amounts of metabolically labeled dSA and showed that dSL are retained at ER exit points resulting in reduction of membrane fluidity and changes in cargo trafficking (39). The differential engagement of ER stress and DNA damage across lymphoma subtypes suggests that cell-type–specific thresholds for ER stress and DNA repair shape how dSLs are sensed and processed in distinct hematologic malignancies. In this context, it is noteworthy that dSA elicited qualitatively similar apoptotic and stress responses in both Jurkat and Molt4 cells, but with quantitatively stronger effects in Molt4. These differences likely reflect intrinsic variation in stress-response capacity or sphingolipid metabolism between T-cell leukemia subtypes rather than distinct mechanisms of dSA action. Accordingly, our conclusions focus on the shared ceramide-dependent mitochondrial pathway observed in both cell lines, while acknowledging that cell-line-specific factors modulate the magnitude of the response. A related limitation is that our isoform-specific analysis of CerS function was performed in Jurkat cells only. siRNA transfection was substantially more efficient and reproducible in Jurkat than in Molt4 cells. As a consequence, we cannot exclude differences in CerS2 or other CerS contributions to dSA toxicity in Molt4 or additional lymphoma subtypes. Future work using optimized genetic tools or alternative perturbation strategies will be required to determine how lineage-specific CERS expression and activity fine-tune dSL-induced cell death. An important aspect of our study is the demonstration that dSA exerts potent cytotoxic effects across a panel of malignant lymphoid cell lines but not in normal human T cells. At concentrations that efficiently killed Jurkat, Molt4, DB, SUPT1, and HL-60 cells, dSA reduced the number of normal T cells over time without inducing significant loss of viability, indicating a predominantly cytostatic effect in non-transformed lymphocytes. Importantly, dSA treatment of normal human T cells increased deoxydihydroceramides/deoxyceramides 14Z to levels comparable to those observed in Jurkat cells, yet without measurable loss of mitochondrial membrane potential, underscoring a selective susceptibility of non-solid cancer cells that may reflect differences in, mitochondrial apoptotic priming or stress-response pathways between malignant and normal T cells (40, 41), although our current data do not yet define the mechanistic basis for this differential sensitivity. Importantly, dSL are well established mediators of neurotoxicity in hereditary sensory neuropathy and chemotherapy-induced peripheral neuropathy, and clinical evaluation of the deoxysphingoid-based agent ES-285 was limited by off-target liver and nervous system toxicity. Thus, our in vitro observations demonstrate a favorable window between malignant and normal T cells under the conditions tested, but they do not establish a systemic therapeutic window or exclude toxicities in other vulnerable cell types such as neurons or hepatocytes. The micromolar dSA concentrations used in our in vitro assays exceed steady-state plasma dSL levels reported in patients and are intended to model acute, mechanism-oriented toxicity rather than recapitulate circulating exposures in vivo. Defining the metabolic and signaling determinants of this apparent therapeutic window, and understanding whether CerS-dependent mitochondrial apoptosis can be preferentially engaged in malignant cells without recapitulating known dSL toxicities, will be critical for rationally exploiting dSL pathways in lymphoma. Taken together, our work expands the spectrum of dSL-responsive malignancies to include non-solid tumors and provides a mechanistic framework for how dSA kills lymphoma cells. We propose a model in which CerS-dependent synthesis of deoxy(dihydro)ceramides constitutes the central toxic event, leading to mitochondrial permeabilization, caspase activation, and secondary DNA damage, while PERK-driven ER stress represents a parallel, largely non-lethal adaptive response (Fig. 6G). The main limitations of this study are that it relies primarily on in vitro models and pharmacologic inhibition; in vivo validation in lymphoma xenografts or genetically engineered mouse models, as well as more selective tools to manipulate dSL synthesis, will be necessary to determine whether this pathway can be targeted at exposures that spare non-malignant tissues and avoid the dose-limiting neurotoxicity and hepatotoxicity observed with dSA (ES-285). Altered amino acid availability (28) or specific point mutations in SPTLC subunits (42) provide alternative means to manipulate dSL synthesis; however, a key limitation of these strategies is that the resulting cellular dSL levels remain substantially lower than those achieved by direct exposure to dSA at cytotoxic concentrations (unpublished results). Nonetheless, the pronounced sensitivity of lymphoma cells—coupled with the relative resistance of normal T cells—supports the concept that controlled modulation of dSL metabolism might offer a novel therapeutic avenue for hematologic malignancies, while highlighting the need for future in vivo studies to rigorously define efficacy, toxicity, and true therapeutic window. Data availability Data that support the findings of this study are included within the article and its supplementary materials. Supplemental data This article contains supplemental data. Conflict of interest The authors declare that they have no conflicts of interest with the contents of this article. Acknowledgments The authors would like to acknowledge the Stony Brook Cancer Center Biological Mass Spectrometry Shared Resource for expert assistance with Lipidomics Analysis. We are also grateful to the members of the Lipid Cancer lab at Stony Brook University for helpful feedback and discussions throughout the duration of this study. Author contributions K. J. L., F. N. V., D. C., B. K., and Y. A. H. writing – review & editing; K. J. L., D. C., B. K. methodology; K. J. L., B. K. formal analysis; F. N. V. writing – original draft; F. N. V., D. C., and Y. A. H. supervision; F. N. V. and Y. A. H. funding acquisition; F. N. V., D. C. data curation; F. N. V. and Y. A. H. conceptualization; Y. A. H. resources. Funding and additional information This work was supported by National Cancer Institute Grant P01-CA097132 (YAH) and by Institutional Research Grant 21-143-01-IRG (FNV) from the American Cancer Society. Contributor Information Fabiola N. Velazquez, Email: FabiolaNoelia.Velazquez@stonybrookmedicine.edu. Yusuf A. Hannun, Email: Yusuf.Hannun@stonybrookmedicine.edu. Supplemental data Supplementary Fig S1 Ceramide synthase activity is essential for the toxic effect of dSA. A. Jurkat cells were treated with 0.75 μM dSA under the presence of different concentrations of FB1. Cell death was quantified 24 hs after dSA treatment using LDH assay. Results are the mean ± SD from 3 independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗∗P ≤ 0.01; ∗∗∗∗P ≤ 0.0001. B. Deoxysphingolipids species were measured by LC/MS in cell extracts from Jurkat cells after 24 hs of treatment with 0.75 μM dSA in the presence or absence of 15 μM FB1. Results are represented as total pmol of lipid/nmol of inorganic lipid phosphate (Pi). Three independent samples per each condition were analyzed. Ordinary one-way ANOVA. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. C-D. Sphingosine, sphingosine-1P and total ceramide were measured by LC/MS in cell extracts from Molt4 (C) and Jurkat (D) cells after 24 hs of treatment with 0.75 μM dSA in the presence or absence of FB1. Three independent samples per each condition were analyzed. Ordinary one-way ANOVA. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. E. Expression of cell death markers (cleaved Parp1 and caspase 3), ER stress marker (p-eIF2α) and DNA damaging marker (p-γH2A.X) were determined by western blot analysis in protein extracts from Jurkat cells treated 0.75 μM dSA in the presence or absence of 15 μM FB1. Actin was used as a loading control. Bottom panel shows the quantification of the indicated markers from three independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗P ≤ 0.05. Supplementary Fig S2 CERS2 knockdown in Jurkat cells. A. Representative western blot of CerS2 in protein extracts from Jurkat cells transfect with control siRNA (AS) or siRNA targeting CERS2. B. Canonical sphingolipids were quantified by LC/MS in cells extracts from Jurkat cells transfected with control siRNA (AS) or siRNA targeting CERS2. Results are expressed as pmol lipid per nmol inorganic lipid phosphate (Pi). Three independent samples per each condition were analyzed. Statistical analysis: Unpaired t test with Welch’s correction for sphingosine, sphingosine-1P and total ceramides, and two-way ANOVA followed by Šídák's multiple comparisons test for individual ceramide species. ∗P ≤ 0.05. Supplementary Fig S3 Mitochondrial membrane potential in Jurkat cells treated with dSA. Mitochondrial membrane potential was assessed by flow cytometric analysis of TMRE staining. Results are presented as the percentage of TMRE-low cells, indicative of reduced mitochondrial membrane potential (ΔΨm). Three independent samples were analyzed for each condition. Statistical analysis: Unpaired t test with Welch’s correction. ∗∗P ≤ 0.01. Supplementary Fig S4 Deoxysphingolipid levels in normal human T cells treated with dSA. Deoxysphingolipid species were quantified by LC/MS in cell extracts from normal human T cells from patient #1 following 24 hs of treatment with 0.75 μM dSA. Results are expressed as pmol lipid per nmol inorganic lipid phosphate (Pi). Three independent samples were analyzed for each condition. Statistical analysis: Unpaired t test with Welch’s correction for deoxysphinganine, deoxysphingosine and total deoxy(dihydro)ceramides, and two-way ANOVA followed by Šídák's multiple comparisons test for individual deoxy(dihydro)ceramide species. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗∗P ≤ 0.0001. References Associated Data Supplementary Materials Supplementary Fig S1 Ceramide synthase activity is essential for the toxic effect of dSA. A. Jurkat cells were treated with 0.75 μM dSA under the presence of different concentrations of FB1. Cell death was quantified 24 hs after dSA treatment using LDH assay. Results are the mean ± SD from 3 independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗∗P ≤ 0.01; ∗∗∗∗P ≤ 0.0001. B. Deoxysphingolipids species were measured by LC/MS in cell extracts from Jurkat cells after 24 hs of treatment with 0.75 μM dSA in the presence or absence of 15 μM FB1. Results are represented as total pmol of lipid/nmol of inorganic lipid phosphate (Pi). Three independent samples per each condition were analyzed. Ordinary one-way ANOVA. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. C-D. Sphingosine, sphingosine-1P and total ceramide were measured by LC/MS in cell extracts from Molt4 (C) and Jurkat (D) cells after 24 hs of treatment with 0.75 μM dSA in the presence or absence of FB1. Three independent samples per each condition were analyzed. Ordinary one-way ANOVA. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001; ∗∗∗∗P ≤ 0.0001. E. Expression of cell death markers (cleaved Parp1 and caspase 3), ER stress marker (p-eIF2α) and DNA damaging marker (p-γH2A.X) were determined by western blot analysis in protein extracts from Jurkat cells treated 0.75 μM dSA in the presence or absence of 15 μM FB1. Actin was used as a loading control. Bottom panel shows the quantification of the indicated markers from three independent experiments. Statistical analysis: Ordinary one-way ANOVA. ∗P ≤ 0.05. Supplementary Fig S2 CERS2 knockdown in Jurkat cells. A. Representative western blot of CerS2 in protein extracts from Jurkat cells transfect with control siRNA (AS) or siRNA targeting CERS2. B. Canonical sphingolipids were quantified by LC/MS in cells extracts from Jurkat cells transfected with control siRNA (AS) or siRNA targeting CERS2. Results are expressed as pmol lipid per nmol inorganic lipid phosphate (Pi). Three independent samples per each condition were analyzed. Statistical analysis: Unpaired t test with Welch’s correction for sphingosine, sphingosine-1P and total ceramides, and two-way ANOVA followed by Šídák's multiple comparisons test for individual ceramide species. ∗P ≤ 0.05. Supplementary Fig S3 Mitochondrial membrane potential in Jurkat cells treated with dSA. Mitochondrial membrane potential was assessed by flow cytometric analysis of TMRE staining. Results are presented as the percentage of TMRE-low cells, indicative of reduced mitochondrial membrane potential (ΔΨm). Three independent samples were analyzed for each condition. Statistical analysis: Unpaired t test with Welch’s correction. ∗∗P ≤ 0.01. Supplementary Fig S4 Deoxysphingolipid levels in normal human T cells treated with dSA. Deoxysphingolipid species were quantified by LC/MS in cell extracts from normal human T cells from patient #1 following 24 hs of treatment with 0.75 μM dSA. Results are expressed as pmol lipid per nmol inorganic lipid phosphate (Pi). Three independent samples were analyzed for each condition. Statistical analysis: Unpaired t test with Welch’s correction for deoxysphinganine, deoxysphingosine and total deoxy(dihydro)ceramides, and two-way ANOVA followed by Šídák's multiple comparisons test for individual deoxy(dihydro)ceramide species. ∗P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗∗P ≤ 0.0001. Data Availability Statement Data that support the findings of this study are included within the article and its supplementary materials.