Altered sphingolipid biosynthetic flux and lipoprotein trafficking contribute to trans fat-induced atherosclerosis

Summary

Dietary fat drives the pathogenesis of atherosclerotic cardiovascular disease (ASCVD), particularly circulating cholesterol and triglyceride-rich lipoprotein remnants. Industrially produced trans-unsaturated fatty acids (TFAs) incorporated into food supplies significantly promoted ASCVD. However, the molecular trafficking of TFAs responsible for this association is not well understood. Here, we demonstrate that TFAs are preferentially incorporated into sphingolipids by serine palmitoyltransferase (SPT) and secreted from cells in vitro. Administering high-fat diets (HFD) enriched in TFAs to Ldlr^−/− mice accelerated hepatic very-low-density lipoprotein (VLDL) and sphingolipid secretion into circulation to promote atherogenesis compared to a cis-unsaturated fatty acid (CFA)-enriched HFD. SPT inhibition mitigated these phenotypes and reduced circulating atherogenic VLDL enriched in TFA-derived polyunsaturated sphingomyelin. Transcriptional analysis of human liver revealed distinct regulation of SPTLC2 versus SPTLC3 subunit expression consistent with human genetic correlations in ASCVD, further establishing sphingolipid metabolism as a critical node mediating the progression of ASCVD in response to specific dietary fats.

eTOC

Gengatharan et al. demonstrate that trans fatty acids (TFAs) are selectively incorporated by SPT to synthesize and secrete sphingolipids from cells. TFA-enriched high-fat diets accelerated hepatic sphingolipid and VLDL secretion to promote atherosclerosis in Ldlr^/− mice. SPT inhibition mitigated trans fat-driven atherogenesis while diminishing flux of sphingolipids into atherogenic VLDL.

Graphical Abstract

Introduction

Sphingolipids are a diverse class of bioactive lipids that consist of a long-chain base (LCB) with further modifications that can include an acyl chain or various head groups to construct more complex species. They play diverse roles in membrane biology^1–6 and signaling^7–11 and are implicated in numerous diseases including ASCVD^12,13, non-alcoholic fatty liver disease^14,15, obesity and associated insulin resistance disorders^16–18, diabetes^19–22, peripheral neuropathy^23,24, and neurodegeneration²⁵. Owing to their abundance and contribution to the blood lipid profile as components of VLDL, low-density lipoproteins (LDL), and high-density lipoproteins (HDL)²⁶, ceramides and other sphingolipid species are emerging as biomarkers for various diseases including ASCVD. Consequently, the Mayo Clinic (USA) has implemented a clinical test to measure plasma ceramides as a gauge of cardiovascular risk²⁷.

The initial rate-limiting enzyme of sphingolipid biosynthesis, serine palmitoyltransferase (SPT), canonically condenses the amino acid serine and palmitoyl-CoA to synthesize the LCB of sphingolipids. SPT is promiscuous for both amino acid and acyl-CoA substrates. For example, SPT will use alanine if serine levels are low^28,29 to produce 1-deoxysphingolipids³⁰. These non-canonical sphingolipids are also aberrantly produced by patients carrying SPTLC1 or SPTLC2 mutations implicated in hereditary sensory and autonomic neuropathy type 1 (HSAN 1)^31,32, and macular telangiectasia type 2^29,33. Similarly, while the canonical acyl-CoA substrate is the 16 carbon palmitoyl-CoA, SPT can incorporate acyl-CoAs from 10–26 carbons to make a diverse set of LCBs^34,35. The SPT enzyme complex comprises two primary subunits and several accessory proteins that influence substrate choice and enzyme activity^35–40. SPTLC1 encodes an obligate component of the enzyme with either SPTLC2 or SPTLC3 required for catalytic activity. While SPTLC1 and SPTLC2 expression are ubiquitous, SPTLC3 is selectively expressed across tissues including placenta, skin, glandular tissues, kidney, and liver^35,40–43.SPTLC3 variants have been associated with elevated LDL-C⁴⁴, myocardial infarction⁴⁵, and dyslipidemia⁴⁶. Additionally, SPTLC3 enhances LCB diversity by facilitating incorporation of short-chain acyl-CoAs³⁷ and the monomethyl branched-chain fatty acid (mmBCFA) anteiso-C17:040 into LCB pools. While saturated acyl-CoAs are commonly used by SPT, unsaturated acyl-CoAs also exhibit activity with SPT³⁴. However, the extent of their utilization by SPT and impact on downstream sphingolipid metabolism have not been thoroughly explored.

Dietary fat is absorbed and distributed throughout the body by the intestine and liver via chylomicrons and very-low-density lipoproteins (VLDL), respectively. In many cases, fatty acids are distinctly metabolized by cellular enzymes to influence their fate. Therefore, deciphering the molecular mechanisms of fatty acid trafficking through the lipidome and lipoproteins will improve our understanding of various diseases, including ASCVD where atherogenic lipoproteins and triglyceride-rich lipoprotein (TRL) remnants are retained in the endothelium and initiate atherosclerosis. Trans fatty acids (TFAs) are unsaturated fatty acids with at least one double bond in the trans conformation, which leads to a straight-chain structure similar to saturated fatty acids (SFAs) unlike the kinked structure of cis unsaturated fatty acids (CFAs). A large percentage of TFAs are industrially produced through partial hydrogenation of vegetable oils, of which the most abundant TFA is elaidate C18:1 (9E), the isomer of oleate C18:1 (9Z). Trans fats were once widespread in the food supply and drove a marked increase in ASCVD in the population, supported by clinical studies that linked TFAs to increased cardiovascular risk and an unfavorable plasma lipid profile including increased total cholesterol and LDL-cholesterol (LDL-C) as well as decreased HDL-cholesterol (HDL-C)⁴⁷. While dietary fatty acid composition influences the progression of various diseases, the contribution of fatty acid flux through downstream lipid pathways to their pathogenesis warrants further investigation.

Here, we comprehensively tracked the fate and diversity of fatty acids through the sphingolipid biosynthesis pathway to highlight how TFA incorporation by SPT promotes lipoprotein secretion and ASCVD. We designed HFDs with identical macronutrient compositions differing distinctly in their mono-unsaturated fatty acid species and used Ldlr^−/− mice⁴⁸ to assess their impact on ASCVD and associated pathologies. We found CFA diets induced poor glucose handling and greater adiposity, while diets including TFAs increased hepatic VLDL secretion and drove atherosclerotic plaque formation. Inhibiting SPT activity via dietary administration of myriocin, a SPT inhibitor, profoundly reduced TFA-induced atherosclerosis while revealing specific regulation of SPTLC2 and SPTLC3 subunit expression. Collectively, these findings highlight the distinct trafficking of fatty acids through the lipidome and further establish sphingolipid biosynthesis as a target for ASCVD.

Results

TFAs are preferentially metabolized by SPT over CFAs

The SPT complex catalyzes synthesis of LCBs and many studies^37,40 have highlighted its promiscuity with respect to amino acid and acyl-CoA substrate usage (Figure 1A). Consistent with these findings, Huh7 hepatocarcinoma cells treated with diverse saturated fatty acid species bound to albumin for 48 hours contained LCBs for each of the fatty acids tested including the canonical substrate C16:0 palmitate, C18:0 stearate, odd-chain fatty acid C17:0, and monomethyl branched chain fatty acids (mmBCFAs) iso-C16:0, anteiso-C17:0, or iso-C17:0 (Figure 1B). While saturated acyl-CoAs have been primarily explored as alternate substrates for SPT, monounsaturated fatty acids (MUFAs) are highly abundant in the diet and tissues, and microsomal enzyme activity assays have previously indicated that both cis-C18:1(9Z) oleate and trans-C18:1(9E) elaidate are metabolized by SPT to LCBs³⁴. To examine the extent of MUFA incorporation into LCBs and beyond, we cultured Huh7 cells with 100 μM of fatty acid bound to albumin for 48 hours and measured SPT activity through quantification of hydrolyzed sphingolipids to detect chromatographically-resolved, isobaric LCBs derived from oleate or elaidate (Figures 1C-D, Figure S1A). Elaidate was incorporated into its respective sphinganine d20:1 (SA d20:1) and sphingosine d20:2 (SO d20:2) LCBs at levels 8-fold and 15-fold more than oleate, respectively (Figure 1E). In contrast to oleate-derived LCBs, the elaidate-derived LCB SA d20:1(E) was produced at similar levels to other non-canonical LCBs, including those generated in cells treated with odd- and branched-chain fatty acids (Figure S1B). To examine the fates of oleate and elaidate through the lipidome more directly, we treated Huh7 cells with the ²H-labeled tracers oleate-d9 and elaidate-d17 and used multiple-reaction monitoring (MRM) to quantify SA d20:1 and SO d20:2 with mass shifts of 9 or 17, respectively (Table S1). Elaidate-d17 was incorporated into its respective SA LCB 5-fold and SO LCB 13-fold more than oleate-d9 (Figure 1F). When Huh7 cells were treated with an equimolar mix of both tracers in direct competition for SPT, we observed elaidate-d17 incorporation to be 17-fold higher for SA and 60-fold higher for SO as compared to oleate-d9 (Figure S1C). We additionally found that another trans fatty acid, trans-vaccenate, was incorporated into its respective sphinganine d20:1 (SA d20:1) and sphingosine d20:2 (SO d20:2) LCBs at levels 7-fold and 9-fold more than its isomer cis-vaccenate, respectively (Fig S1D). Trans-vaccenate-derived LCBs were lower than elaidate-derived LCBs, likely due to conversion of trans-vaccenate to conjugated linoleic acid (CLA) via stearoyl-CoA desaturase (SCD1)^49,50, which cannot desaturate elaidate. In summary, the straight-chain structure of trans fatty acids appears to be preferential to SPT as shown via incorporation of two trans fatty acids.

Figure 1. TFAs are preferentially metabolized by SPT over CFAs.

(A) Schematic depicting promiscuity of the initial rate-limiting enzyme of sphingolipid biosynthesis SPT in both amino acid and acyl-CoA substrates. Created with Biorender.com.

(B) Hydrolyzed LCBs synthesized from supplemented fatty acid treatments in Huh7 cells (n=3 per group).

(C) Hypothesized LCBs from oleate or elaidate incorporation by SPT. Oleate or elaidate would first form a SA 20:1 (9Z) or SA d20:1 (9E) LCB, respectively, after SPT and KDSR reactions and these LCBs would be maintained through CerS. They would next form a SO d20:2 (4E,11Z) or SO d20:2 (4E,11E) LCB, respectively, after CerS and DEGS reactions. The LCBs derived from oleate or elaidate will be noted by their unique double bond configuration as SA d20:1(Z) and SO d20:2(Z) from oleate and SA d20:1(E) and SO d20:2(E) from elaidate. Created with Biorender.com.

(D) After hydrolyzing sphingolipids from either oleate or elaidate treatment to their LCBs, unique peaks were identified via LC-MS in only cells treated with elaidate, corresponding to both the expected SA d20:1 (E) and SO d20:2 (E) LCBs. Another peak with the same MRM eluting prior to this novel peak was present at low levels in vehicle-treated cells and was increased upon oleate treatment. Through peak area increases corresponding to treatments of the different fatty acids, the first peak was identified as derived from oleate while the second peak was identified as derived from elaidate.

(E) Hydrolyzed SA and SO LCBs synthesized from 100 μM oleate or 100 μM elaidate, respectively, in Huh7 cells (n=3 per group).

(F) Hydrolyzed SA and SO LCBs synthesized from oleate-d9 or elaidate-d17, respectively, in Huh7 cells (n=3 per group).

Relative abundance is calculated by normalizing to internal standard specific to lipid class. Data are mean ± standard error of mean (SEM) and were analyzed using an independent t-test (E-F). *p < 0.05, **p < 0.01, *** p< 0.001, or **** p<0.0001 unless otherwise noted.

Molecular partitioning of CFAs and TFAs through sphingolipid metabolism

Next, we assessed MUFA incorporation into intact, cellular sphingolipids using specific MRMs (Table S1). Consistent with its high abundance in obesogenic lard diets, oleate-d9 was highly incorporated into the N-acyl chain of canonical ceramides only after elongation to C24:1-d9 or further, not as its original form C18:1-d9 (Figure 2A). On the other hand, elaidate was incorporated in a more versatile manner reminiscent of saturated acyl-CoAs (Figure 2A). We additionally used specific MRMs to quantify incorporation into the LCB of ceramides and glycosphingolipids (GSLs) (Figure 2B). While ceramides with an oleate-d9-derived LCB were below the detection limit, we found distinct ceramide (Cer), hexosyl-ceramide (Hex-Cer), and lactosyl-ceramide (Lac-Cer) species with elaidate-d17 incorporated into the LCB, suggesting that TFA-containing sphingolipids are processed throughout the pathway (Figure 2C). This ceramide N-acyl profile was similar to that obtained with palmitate or stearate treatments, indicating that ceramide synthases metabolize TFA-derived LCBs similarly to canonical LCBs (Figures S2A-B). Only total incorporation of tracers into sphingomyelin (SM) pools is observable with available MRMs (Table S1). Here, oleate-d9 was incorporated predominantly into SM 42:2-d9, presumably via the abundant, canonical Cer d18:1/24:1-d9. We detected no SM peaks with a mass shift of 18, indicating oleate-d9 is not incorporated into both the LCB and N-acyl at detectable levels (Figure 2D). In contrast, elaidate-d17 was incorporated into more diverse polyunsaturated SM species, including several species with a mass shift of 34 suggestive of elaidate incorporation into both the LCB and N-acyl. Overall, CFAs and TFAs appear to be molecularly partitioned in the sphingolipid biosynthetic pathway, as TFAs are structurally similar to canonical, saturated, SPT substrates and therefore preferentially incorporated in the LCB, while CFAs feed the ceramide pool through N-acylation (Figure 2E). Therefore, double bond configuration can affect flux through sphingolipid biosynthesis at different stages.

Figure 2. Molecular partitioning of CFAs and TFAs through sphingolipid metabolism.

(A) N-Acyl ²H labeling on canonical ceramides d18:1 from oleate-d9 or elaidate-d17 in Huh7 cells (n=3 per group).

(B) Schematic of elaidate-d17 incorporation into the LCB of sphingolipids. Created with Biorender.com.

(C) LCB ²H labeling from elaidate-d17 producing a d20:2-d17 LCB in ceramides, hexosyl-ceramides, and lactosyl-ceramides in Huh7 cells (n=3 per group).

(D) ²H labeling on sphingomyelin from oleate-d9 or elaidate-d17 in Huh7 cells (n=3 per group).

(E) Schematic of molecular partitioning of cis and trans fatty acids in the sphingolipid biosynthetic pathway. Created with Biorender.com.

Relative abundance is calculated by normalizing to internal standard specific to lipid class. Data are mean ± standard error of mean (SEM) and were analyzed using an independent t-test (A,D). *p < 0.05, **p < 0.01, *** p< 0.001, or **** p<0.0001 unless otherwise noted.

TFAs drive aberrant sphingomyelin secretion

We next assessed how these CFA and TFA treatments influence the broader lipidome, quantifying the flux of oleate-d9 and elaidate-d17 into phospholipids and neutral lipids in Huh7 cells. Oleate was preferentially utilized in phospholipids phosphatidylcholine (PC) and phosphatidylethanolamine (PE) while elaidate was more enriched in lysophospholipids (Figures 3A, S3A-F). This preferential utilization of oleate for phospholipids suggests that the kinked structure provided by the cis double bond may be crucial for membrane fluidity and that alterations in double bond configuration could lead to different downstream fates of lipid flux, consistent with previous findings highlighting differential lipidomic profiles and incorporation of CFAs and TFAs^51,52.

Figure 3. TFAs drive aberrant sphingomyelin secretion.

(A) Total abundances of lipids with ²H labeling from oleate-d9 or elaidate-d17 normalized to abundances following oleate-d9 supplementation in Huh7 cells (n=3 per group).

(B) Stacked plot of sphingomyelin (SM) secretory flux from Huh7 cells following fatty acid supplementation including oleate-d9 or elaidate-d17 (n=3 per group).

(C) Total sphingomyelin (SM) secretory flux from Huh7 cells with ²H labeling from oleate-d9 or elaidate-d17 (n=3 per group).

(D) Total phosphatidylcholine (PC) secretory flux from Huh7 cells with ²H labeling from oleate-d9 or elaidate-d17 (n=3 per group).

(E) Schematic depicting mechanism of action of SPT inhibitor Myriocin. Created with Biorender.com.

(F) LCB ²H labeling from elaidate-d17 producing a d20:2-d17 LCB in ceramides (Cer) in Huh7 cells treated with elaidate-d17 or elaidate-d17 and Myriocin (n=3 per group).

(G) ²H labeling on sphingomyelin from oleate-d9 or elaidate-d17 in Huh7 in the presence or absence of myriocin (n=3 per group).

(H) Total sphingomyelin (SM) secretory flux from Huh7 cells with ²H labeling from oleate-d9 or elaidate-d17 in the presence or absence of myriocin (n=3 per group).

Relative abundance is calculated by normalizing to internal standard specific to lipid class. Data are mean ± standard error of mean (SEM) and were analyzed using an independent t-test (A, C-D, F), one-way ANOVA (B), or two-way ANOVA with Fisher’s LSD post hoc test (G-H). Statistical analysis in (B) was performed using total abundance.*p < 0.05, **p < 0.01, *** p< 0.001, or **** p<0.0001 unless otherwise noted.

SM is the most abundant sphingolipid present in blood, predominantly in lipoproteins, and its presence in plasma correlates positively with ASCVD⁵¹. Therefore, we quantified secretion of SM from Huh7 cells into culture media under various fatty acid treatments, including oleate-d9 and elaidate-d17. Elaidate induced the highest total SM secretory flux from Huh7 cells compared to its cis isomer oleate or saturated palmitate and stearate (Figure 3B). This increase in SM secretion was significantly driven by SM with double incorporation of elaidate (d36:3-d34) (Figure 3B), such that nearly 10-fold more elaidate-incorporated SM was secreted compared to oleate-incorporated SM (Figure 3C). This trend was specific to sphingolipids, as oleate-incorporated PCs were secreted more than elaidate-incorporated PCs (Figure 3D), presumably due to the higher abundance of intracellular, oleate-derived PCs (Figure S3C). This preferential secretion of elaidate-derived SM further highlights the distinct metabolism of TFAs in the sphingolipid biosynthetic pathway.

To determine if synthesis of these elaidate-derived sphingolipids can be modulated, we treated cells with myriocin, a pharmacological inhibitor of SPT (Figure 3E). Myriocin reduced production of the LCBs SA d20:1 (Z)-d9 and SO d20:2 (Z)-d9 from oleate-d9 and SA d20:1 (E)-d17 and SO d20:2 (E)-d17 from elaidate-d17 following hydrolysis of sphingolipids (Figure S3G). In cells treated with elaidate-d17, myriocin also reduced the intracellular abundance of several Cer d20:2-d17, Hex-Cer d20:2-d17, Lac-Cer d20:2-d17, SM-d17, and SM-d34 species under elaidate-d17 treatment (Figures 3F-G, S3H-I). Additionally, myriocin suppressed the secretion of TFA-derived SM-d17 and SM-d34 polyunsaturated species, suggesting that SPT activity and fatty acid structure directly impact lipid secretion (Figure 3H).

SPT inhibition mitigates trans fat-driven dysregulation of hepatic lipid metabolism

Given the above findings as well as the strong correlations of circulating ceramides, SM, and cholesterol with ASCVD⁵³, we hypothesized that inhibition of sphingolipid biosynthesis could modulate lipoprotein metabolism and reduce aortic plaque deposition induced by TFAs^54–56. To test this in vivo, we utilized LDL receptor-deficient (Ldlr−/−) mice⁴⁸ to model the impact on hyperlipidemia and atherosclerosis. Ldlr^−/−mice develop accelerated atherosclerosis due to reduced clearance of circulating atherogenic LDL-C⁴⁸. We designed custom, high-fat diets (HFDs) with identical macronutrient composition (60% kcal fat) enriched in either 62% cis monounsaturated C18:1 fatty acids (Cis HFD) or a mixture of 28% trans monounsaturated C18:1 fatty acids and 34% cis monounsaturated C18:1 fatty acids (Trans HFD) (Figure 4A, Table S2). Each diet was also supplemented with myriocin, and we administered all four diets to Ldlr^−/− mice for 16 weeks to examine metabolic and physiological changes associated with hyperlipidemia, liver steatosis, and atherosclerosis. Cholesterol supplementation was not included in the diets to interrogate the dichotomy of CFA versus TFA ingestion without confounding dietary cholesterol intake⁵⁷. Mice fed Cis HFD gained the most weight, followed by mice fed Trans HFD, while myriocin attenuated body weight gain under both dietary fat compositions (Figure 4B). Food intake was reduced by 12% under Trans HFD + Myriocin but was not significantly different between the other diets (Figure S4A). Body weight differences largely reflected changes in adiposity (Figures 4C, S4B).

Figure 4. SPT inhibition mitigates trans fat-driven dysregulation of hepatic lipid metabolism.

(A) Fatty acid composition of Cis Unsaturated HFD derived from a combination of 34% lard and 66% olive oil and Trans Unsaturated HFD derived from 100% Primex.

(B) Body weight over course of 16 weeks in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=17 per group).

(C) Inguinal adipose tissue (iWAT) weight in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=15 per group).

(D) Representative images of hematoxylin and eosin (H&E) staining of the liver at 10x magnification highlighting hepatic steatosis in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=10 per group).

(E) Liver weight in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=15 per group).

(F) Hepatic de novo lipogenesis of palmitate from ²H₂O in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=5 per group).

(G) Hepatic diglyceride (DG) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=10 per group).

(H) Hepatic phosphatidylcholine (PC) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=10 per group).

(I) Hepatic phosphatidylethanolamine (PE) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=10 per group).

(J) Hepatic lysophosphatidylcholine (LPC) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=10 per group).

(K) Hepatic dihydroceramide (DHCer) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=9–10 per group).

(L) Hepatic ceramide (Cer) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=9–10 per group).

(M) Hepatic deoxydihydroceramide (doxDHCer) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=9–10 per group).

Relative abundance is calculated by normalizing to internal standard specific to lipid class. Data are mean ± standard error of mean (SEM) and were analyzed using a two-way ANOVA with Fisher’s LSD post hoc test (B-C, E-M). *p < 0.05, **p < 0.01, *** p< 0.001, or **** p<0.0001 unless otherwise noted.

Cis HFD-fed mice exhibited elevated fasting blood glucose at 16 weeks (Figure S4C), and glucose tolerance was correspondingly compromised under the Cis HFD compared to all other diets (Figure S4D). However, Trans HFD-fed mice exhibited more advanced liver steatosis than Cis HFD-fed mice (Figure 4D), consistent with prior comparisons of TFAs, SFAs, and CFAs^58–60. Liver weight was elevated on the Trans HFD compared to the Cis HFD, while myriocin reduced liver weight and steatosis in both diets (Figure 4D-E). Expression of Col1a1, a marker of fibrosis, was also increased on the Trans HFD and reduced by myriocin, indicating a mitigation of preliminary fibrosis (Figure S4E). Stable isotope tracing via ²H₂O to evaluate de novo lipogenesis (DNL) revealed elevated palmitate synthesis on the Trans HFD compared to the Cis HFD (Figure 4F), which was attenuated by myriocin, consistent with previous observations of myriocin influencing SREBP1 mRNA and protein⁶¹. Myriocin additionally increased expression of several genes involved in fatty acid oxidation under the Trans HFD (Fig S4F), consistent with previous studies demonstrating this compound increases hepatic fatty acid oxidation to attenuate liver steatosis⁶².

To understand how these diets influenced molecular abundances across the lipidome, we quantified hepatic lipids using high-resolution mass spectrometry. Within the liver, diglycerides (DG) and total fatty acids were elevated under the Trans HFD compared to the Cis HFD, while these species along with neutral lipids triglycerides (TG) and cholesteryl esters (CE) were reduced by myriocin (Figures 4G, S4G-I). In contrast, hepatic PC and PE, as well as lysophosphatidylcholine (LPC) and lysophosphatidylethanolamine (LPE) were all elevated on the Trans HFD versus Cis HFD but further increased by myriocin (Figure 4H-J, Figure S4J). These results suggest myriocin reshapes lipid metabolism by diverting fatty acids away from sphingolipid pools and towards phospholipids rather than reducing lipid uptake overall.

Despite these changes in phospholipids and neutral lipids, hepatic sphingolipid content was not significantly altered between Cis and Trans HFD apart from deoxydihydroceramides (doxDHCer) (Figure 4K-M, S4K-M). Since 1-deoxysphingolipids cannot be phosphorylated and further degraded⁶³, this finding suggests that hepatic sphingolipid biosynthetic flux is increased under the Trans HFD. Myriocin effectively reduced total hepatic dihydroceramide (DHCer), ceramide (Cer), deoxysphinganine (doxSA), and doxDHCer while having no significant impact on more complex sphingolipids, suggesting that salvage pathways can compensate to maintain sphingolipid pools (Figure 4K-M, S4K-M). Therefore, rather than accumulate in the liver, we hypothesized these complex sphingolipids are secreted in lipoproteins at higher rates in Trans HFD- versus Cis HFD-fed mice.

Dietary trans fat induces hepatic sphingolipid and VLDL secretion

Consistent with the findings above, ²H₂O tracing and quantitation of enrichment in plasma cholesterol and palmitate pools revealed increased newly synthesized cholesterol and palmitate in mice fed a Trans HFD. These data suggest that increased lipid biosynthesis contributes to diet-induced changes in circulating lipids (Figures 5A-B). Myriocin attenuated this elevation in newly synthesized plasma cholesterol and palmitate (Figures 5A-B).

Figure 5. Dietary fat induces hepatic sphingolipid and VLDL secretion.

(A) Plasma total newly synthesized cholesterol from ²H₂O in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=5 per group).

(B) Plasma total newly synthesized palmitate from ²H₂O in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=5 per group).

(C) Plasma dihydroceramide (DHCer) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=9–10 per group).

(D) Plasma ceramide (Cer) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=9–10 per group).

(E) Plasma sphingomyelin (SM) abundance in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=9–10 per group).

(F) Plasma total newly synthesized Cer abundance from ²H₂O in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=5 per group).

(G) Plasma newly synthesized Cer profile from ²H₂O in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=5 per group).

(H) Plasma total newly synthesized SM abundance from ²H₂O in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=5 per group).

(I) Plasma newly synthesized SM profile from ²H₂O in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, or Trans HFD + Myriocin (n=5 per group).

(J) Schematic depicting mechanism of action of Tyloxapol, a lipoprotein lipase inhibitor, used to measure hepatic very-low-density lipoprotein (VLDL) secretion. Created with Biorender.com.

(K) Hepatic VLDL secretion after injection of Tyloxapol relative to t=0 in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=4–5 per group).

Relative abundance is calculated by normalizing to internal standard specific to lipid class. Data are mean ± standard error of mean (SEM) and were analyzed using a two-way ANOVA with Fisher’s LSD post hoc test (A-I, K). *p < 0.05, **p < 0.01, *** p< 0.001, or **** p<0.0001 unless otherwise noted.

To assess the impact of each diet on hepatic sphingolipid output, we quantified circulating sphingolipids, observing that the Trans HFD significantly increased total plasma DHCer (Figure 5C) and doxDHCer (Figure S5A), with a minor reduction in plasma S1P d18:1 levels (Figure S5D). There was no change in total plasma Cer, SM, Hex-Cer, and Lac-Cer abundances between the Cis and Trans HFD (Figures 5D-E, S5B-C). Myriocin effectively reduced circulating, DHCer, Cer, Hex-Cer, Lac-Cer, and SM under both HFDs (Figures 5C-E, S5A-C). To determine whether hepatic sphingolipid biosynthesis and secretion were influenced by Trans HFD feeding and myriocin, we quantified ²H-enrichment in sphingolipids from mice administered ²H₂O using high-resolution mass spectrometry, accounting for theoretical maximal labeling per molecule to determine fractional synthesis and newly synthesized relative abundances. The Trans HFD significantly increased newly synthesized DHCer, Cer, and SM present in plasma (Figures 5F-I, S5E-F). On the other hand, myriocin reduced the abundances of newly synthesized sphingolipids in circulation (Figures 5F-I, S5E-F), providing evidence that the Trans HFD increases hepatic SPT flux and secretion.

Next, we measured hepatic VLDL secretion in Ldlr^−/− mice fed each diet, since cholesterol, triglycerides, sphingolipids, and other lipid species are incorporated into VLDL by the liver. Specifically, we administered Tyloxapol to mice and quantified triglycerides in circulation over time that are representative of secreted VLDL (Figure 5J). Indeed, the Trans HFD greatly accelerated hepatic VLDL secretion compared to the Cis HFD (Figure 5K, S5G). We further hypothesized that SPT activity supports the assembly and secretion of VLDLs by the liver into blood. Incorporation of myriocin into diets significantly reduced VLDL secretion induced by Trans HFD (Figure 5K, S5G), highlighting functional links between dietary TFAs, sphingolipid biosynthesis, and atherogenic VLDLs.

Trans fat-derived VLDL-sphingolipids promote atherosclerosis

Dietary TFAs accelerate atherosclerosis compared to SFA, CFA or polyunsaturated fatty acid (PUFA)-enriched diets in Ldlr^−/− mice^54–56 and in the human population⁴⁷. Therefore, we next aimed to understand how SPT activity and TFAs influence the progression of ASCVD by measuring atherosclerotic lesions in the valves of the aortic root. The Trans HFD significantly increased atherosclerotic lesion area within the aortic root, characterized by intima thickening and cholesterol crystals compared to the fatty streaks found with the Cis HFD. Myriocin attenuated atherosclerotic lesion area under the Trans HFD (Figures 6A-B), consistent with previous findings in ApoE−/− mice fed long-term chow or a Western diet^64–66.

Figure 6. Trans fat-derived VLDL-sphingolipids promote atherosclerosis.

(A) Representative images of modified Van Gieson staining of the aortic root highlighting atherosclerotic lesions in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=10 per group).

(B) Atherosclerotic lesion quantitation in the aortic root including area under curve (AUC) of mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=9–10 per group).

(C) Lipoprotein analysis of plasma cholesterol from mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (3 pooled plasma per group from n=3 each). CR: chylomicron remnant, VLDL: very-low-density lipoprotein, IDL: intermediate-density lipoprotein, LDL: low-density lipoprotein, HDL: high-density lipoprotein.

(D) Total abundance of plasma polyunsaturated sphingomyelin (SM) containing >3 double bonds in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=9–10 per group).

(E) Profile of plasma di- and poly-unsaturated sphingomyelin (SM) containing >2 double bonds in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=9–10 per group).

(F) Lipoprotein analysis of plasma polyunsaturated sphingomyelin containing >3 double bonds from mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (3 pooled plasma per group from n=3 each). CR: chylomicron remnant, VLDL: very-low-density lipoprotein, IDL: intermediate-density lipoprotein, LDL: low-density lipoprotein, HDL: high-density lipoprotein.

(G) Hepatic mRNA expression of genes involved in sphingolipid metabolism in mice fed Cis HFD, Cis HFD + Myriocin, Trans HFD, Trans HFD + Myriocin (n=7–10 per group).

(H) SPTLC2 versus SPTLC3 correlation to genes involved in hepatic VLDL secretion, including MTTP and APOB in human liver.

Relative abundance is calculated by normalizing to internal standard specific to lipid class. Data are mean ± standard error of mean (SEM) and were analyzed using a two-way ANOVA with Fisher’s LSD post hoc test (B-G). *p < 0.05, **p < 0.01, *** p< 0.001, or **** p<0.0001 unless otherwise noted.

The primary lipids in human atherosclerotic lesions are cholesterol, glycerophospholipids, and sphingolipids, with SM accounting for approximately 60–70% of intimal phospholipids in advanced human lesions^12,67. ApoB lipoproteins such as VLDL and LDL, which contain more sphingolipids than HDL⁶⁸, accumulate in atherosclerotic lesions and pose an ASCVD risk^69,70. Consistent with our hypothesis, the plasma lipoprotein profile was altered in Trans HFD-fed mice, with increased VLDL-cholesterol (VLDL-C), VLDL-triglyceride (VLDL-TG), and VLDL-sphingomyelin (VLDL-SM) compared to mice fed a Cis HFD, and myriocin attenuated this Trans HFD-induced response (Figures 6C, S6A-B).

Targeted plasma lipidomics additionally highlighted that myriocin effectively reduced several lipid classes elevated on the Trans HFD, most notably the phospholipids PC and PE (Figure S6C) in addition to sphingolipids DHCer and doxDHCer (Figure 5C, S5A). We further dissected the plasma sphingomyelin profile to determine the contribution of TFA-derived polyunsaturated sphingolipids to dietary fat-driven atherosclerosis as we observed in vitro (Figure 3B,H). Plasma dihydrosphingomyelin (DHSM) and monounsaturated SM were unchanged between the Cis and Trans HFD (Figures S6D-E). Diunsaturated SM, which can include a variety of canonical LCB and N-acyl combinations in addition to TFA-derived LCBs, were slightly, but significantly, increased under the Trans HFD and attenuated by myriocin (Figures S6F). Consistent with stable isotope tracing in vitro (Figure 3B,H), Trans HFD significantly increased plasma levels of polyunsaturated SM species with 3 or more double bonds that presumably include both cis-sphingadienes derived from FADS3^71,72 and TFA-containing LCBs, while myriocin reduced these species (Figures 6D-E). These polyunsaturated SM species were particularly enriched in VLDL under the Trans HFD and reduced in abundance within VLDL by myriocin (Figure 6F). Collectively, these results suggest that TFA-driven SPT flux drives hepatic lipoprotein secretion and atherosclerosis in Ldlr^−/− mice.

Finally, to gain insights into the molecular regulation of sphingolipid homeostasis, we quantified the expression of various sphingolipid pathway enzymes in the livers of mice fed each diet. While several biosynthetic and catabolic enzymes were downregulated under the Trans HFD compared to Cis HFD, the most dramatic change we observed across all diets was downregulation of the Sptlc3 expression upon long term dietary myriocin treatment (Figure 6G). In contrast, prolonged reduction of SPT activity by dietary myriocin slightly increased Sptlc1 expression and had no effect on Sptlc2 expression, suggesting distinct regulation and function of these subunits (Figure 6G).

SPTLC3 variants are associated with elevated LDL-C⁴⁴, myocardial infarction⁴⁵, and dyslipidemia⁴⁶. Furthermore, this subunit is primarily expressed in cells that mediate lipid processing and/or serve as epithelial barriers including liver, intestine, and skin^43,73. Given the above regulation by myriocin (Figure 6G), we hypothesized that SPTLC3 has specialized function and regulation associated with vesicular processing of lipids, which is critical for lipoprotein secretion. Targeting of SPTLC3 in Huh7 cells via lentiviral delivery of CRISPR-Cas9 and sgRNA significantly reduced SPTLC3 protein levels (Figure S6G). On the other hand, we failed to appreciably reduce expression of SPTLC2 in Huh7 cells (Figure S6G), potentially due to the lower transcription of this subunit compared to SPTLC1 and SPTLC3 (Figure S6H). Therefore, targeting SPTLC3 may be more critical and relevant in Huh7 cells. Flux measurements in media containing [¹³C]serine and [¹³C]glycine indicated that targeting of SPTLC3 effectively reduced the synthesis of intracellular DHCer, Cer, Hex-Cer, Lac-Cer, and SM as well as SM secretory flux (Figures S6I-N), providing evidence that targeting SPTLC3 reduces sphingolipid biosynthesis and secretion in some cells.

To better understand the functional importance of SPTLC3 expression in the liver, we used Correlation AnalyzeR⁷⁴ to identify genes co-regulated with SPTLC3 and SPTLC2 in publicly available human transcriptional datasets. We identified that SPTLC3 has a strong positive correlation with numerous VLDL secretion genes, including MTTP and APOB as well as other APO genes in human liver. In contrast, SPTLC2 has a strong negative correlation with MTTP, APOB, and other APO genes, establishing these genes among the top differentially correlated genes between SPTLC2 and SPTLC3 (Figure 6H, Table S3). These differential correlations suggest opposing functions of SPTLC2 and SPTLC3 in regulating hepatic lipid secretion that correspond to the demonstrated promiscuity of SPTLC3^37,40. Consistent with human disease correlations, SPTLC3 may act to bridge hepatic sphingolipid biosynthetic flux to lipid secretion via lipoproteins, which ultimately leads to atherosclerosis.

Discussion

Here, using a combination of in vitro and in vivo metabolic tracing, dietary manipulations, pharmacological interventions, and physiological analyses, we have highlighted a functional role for SPT in hepatic lipoprotein metabolism and ASCVD progression. We specifically characterized distinct TFA-derived sphingolipids that are selectively secreted from cells, promote VLDL secretion from liver, and accelerate deposition of atherosclerotic plaques in Ldlr^−/− mice.

While cholesterol metabolism has been identified as a major driver of trans fat-induced ASCVD, there are many underlying mechanisms controlling the early pathogenesis that have not been thoroughly explored. Though therapeutic interventions have targeted reducing LDL-C due its positive correlation with ASCVD risk, patients with significant attenuation of LDL-C have a residual ASCVD risk^75,76. Plasma TRL remnant levels are an independent risk factor for ASCVD due to their size and corresponding propensity to carry more cholesterol than LDLs^77–81. However, since triglycerides do not accumulate within atherosclerotic plaques, other lipid species enriched in TRLs may contribute to this residual ASCVD risk.

Ceramides in circulation are linked to ASCVD pathogenesis⁸², and our results highlight specific benefits in reducing their production and dissemination as sphingomyelin in lipoproteins. Arterial wall SMase is thought to hydrolyze sphingomyelin to ceramides, and this increase in ceramides stimulates lipoprotein aggregation and subendothelial retention in artery walls^83,84, thereby mediating the initial stages of atherosclerosis. We demonstrate that pharmacological inhibition of sphingolipid biosynthesis in vivo reduces de novo lipogenesis to diminish TFA-induced VLDL secretion, which delivers sphingolipids, cholesterol, and triglycerides to blood. This reduction in circulating atherogenic lipids including sphingomyelin on VLDL and subsequent transfer to LDL is sufficient to restrict the progression of atherosclerosis induced by dietary TFAs. Notably, SPT activity in enterocytes may also contribute to circulating atherogenic sphingolipids as dietary fatty acids are initially packaged into chylomicrons, so further studies are needed to understand contribution from the intestine.

Although myriocin has previously been shown to mitigate progression of atherosclerosis, most of these studies included cholesterol in the diet^64,66, which is known to advance atherosclerosis. Additionally, the most studies were performed in Apoe−/− mice^64–66, which develop accelerated atherosclerosis on atherogenic diets and carry the majority of cholesterol via VLDL unlike LDL in humans⁸⁵. Here, we directly interrogated the role of different MUFAs in atherosclerosis progression with specifically customized diets varying in CFA versus TFA content without the influence of dietary cholesterol, using Ldlr^−/− mice that have a similar lipoprotein profile to humans. Myriocin can induce gastrointestinal dysfunction⁸⁶ and limit intestinal cholesterol absorption⁸⁷. A minor reduction in food intake was noted only in mice fed Trans HFD + Myriocin in certain weeks. We also observed increased hepatic phospholipid abundances under the Trans HFD + Myriocin diet, demonstrating that dietary fats are absorbed and processed into complex lipids within tissues. This remodeling indicates that diverting fatty acids away from SPT may be beneficial on these HFDs. Therefore, the absence of dietary cholesterol, increased abundance of hepatic phospholipids, and reduction of circulating lipoproteins suggest that a reducing SPT activity can mitigate diet-induced liver steatosis and atherosclerosis. Importantly, these findings are not specific to TFAs, as myriocin also influences the hepatic metabolism and trafficking of CFA and SFA-enriched diets.

Our studies also highlight key differences in the fate of TFAs from CFAs, predominantly oleate, which are readily incorporated into phospholipids and sphingolipids by ceramide synthases, promote more adiposity, yet induce far less VLDL secretion and atherosclerotic plaque formation compared to TFAs. We hypothesize that TFAs are more atherogenic due to their structural mimicry of SFAs and reduced ability to be desaturated. Unlike SFAs that have mechanisms such as stearoyl-CoA desaturase (SCD1) and fatty acid desaturase (FADS3) to be converted to CFAs or sphingadienes^71,72, respectively, TFAs may be ineffective substrates for these desaturases, leading to greater accumulation and toxicity in ASCVD compared to SFAs as has been reported⁴⁷. The FADS3 gene has similarly been implicated in ASCVD through human genetics⁴⁵, and SCD1 mediates saturated fat-induced atherosclerosis⁸⁸. This dichotomy between CFAs and TFAs provides molecular evidence for the benefit of Mediterranean diets enriched in CFAs compared to high SFA diets.

Characterization of SPT subunit expression in response to myriocin also suggested distinct regulation. In Huh7 cells, a pooled genetic knockout of SPTLC3 was sufficient to decrease sphingolipid biosynthesis and secretory flux of SM. Furthermore, we identified a strong positive correlation between SPTLC3 and hepatic VLDL secretion genes in publicly available human liver transcriptional datasets. These findings suggest distinct functions associated with lipid secretion encoded by SPTLC3, a gene linked to ASCVD by variants in human populations demonstrated in genome-wide association studies (GWAS)^45,46,44 and positively correlated with NAFLD severity.⁸⁹ Therefore, SPTLC3 may be a more direct target for reducing hepatic VLDL secretion and circulating lipoproteins that typically carry sphingolipids with a propensity to aggregate and initiate atherosclerosis.

While the World Health Organization (WHO) announced in 2018 a global plan to eliminate TFAs from food supplies by the end of 2023, WHO reported in June 2024 only 53 countries follow best practice policies for removing trans fats from their food supplies, leaving more than 4 billion people at risk for the atherogenic effects of dietary trans fats⁹⁰. Despite this ban, TFAs are still acceptable at levels below 0.5 g per serving in the USA and naturally present in dairy and meat products, so understanding their impact on lipid physiology remains important. We have described a biochemical pathway where TFAs mimic SFAs to drive SPT flux and synergize with hepatic lipoprotein secretion to deliver atherogenic sphingolipids and cholesterol into circulation. Thus, SPT may serve as a potential therapeutic target for mitigating TFA- and SFA-induced ASCVD.

Limitations of the study

Polyunsaturated SM were elevated upon TFA treatment in vitro and Trans HFD administration in vivo. These species may serve as biomarkers of aberrant TFA flux through SPT and sphingolipid metabolism. Stable isotope tracing with the ²H-labeled TFA elaidate revealed the polyunsaturated SM were derived from elaidate. However, this distinction could not clearly be made in vivo using MRMs that identify SM as the sum of their acyl chains. Therefore, the presence of various isobaric SM species can complicate the interpretation of elevated levels of polyunsaturated SM in contexts other than dietary trans fat. Another limitation of these studies is the use of the Ldlr^−/− mice, as the the absence of this receptor could impact our results since dietary TFAs reduce LDLR activity^91,92. However, Ldlr^−/− mice are necessary as a model of accelerated atherosclerosis via accumulation of circulating atherogenic LDL-C as mice typically carry their cholesterol in HDL and do not typically form atherosclerotic lesions. We have also speculated about the role of SPTLC3 in hepatic lipoprotein secretion, as SPTLC3 is positively correlated to VLDL secretion genes in human liver. While we have shown a CRISPR-Cas9-mediated pooled knockout of SPTLC3 in Huh7 cells reduces secretory flux of SM, further comprehensive studies in additional cell lines and in vivo models will be critical to fully characterize this relationship. Nevertheless, our results confirm that some liver-derived cells depend on SPTLC3 for SM secretion.

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Christian Metallo (metallo@salk.edu).

Materials Availability

Plasmids generated in this study for SPTLC3 knockout are available upon request. No other unique materials were generated in this study.

Data and Code Availability

• All raw mass spectrometry data is uploaded to Metabolomics Workbench (doi: 10.21228/M86V58)

• Values used to create all graphs in the paper and uncropped images are available in Data S1.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR Methods

Experimental Model Details

Cell culture experiments

Huh7 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (P/S). Media was supplemented with 100 μM fatty acid treatments conjugated with bovine serum albumin (BSA) for 48 hours. For ²H flux measurements of intact sphingolipids and broader lipidome, cells were treated with elaidic acid-d17 (Cayman Chemical, 27715) or oleic acid-d9 (Avanti Polar Lipids, 861809) in DMEM supplemented with 10% delipidated FBS and 1% P/S. FBS was delipidated using fumed silica. To test direct competition of both fatty acids, Huh7 cells were treated with an equimolar mix of elaidic acid-d17 and oleic acid-d9 at concentrations of 50 μM each. For hydrolysis assays measuring the LCB derived from oleic acid-d9 and elaidic acid-d17, standard FBS was utilized. To inhibit SPT flux in specific experiments, myriocin was used at a concentration of 100 nM in DMSO and added in combination with the fatty acid treatment for 48 hours. To calculate secretory flux, 0.75 ml of fresh media or 1.5 ml of spent media was evaporated under vacuum at 4°C and resuspended into 0.1 mL of H₂O. Concentrated media was extracted as described below. All cells were maintained at 37°C and 5% CO₂ and periodically tested for mycoplasma.

CRISPR-Cas9-engineered knockout cell lines

The online CRISPR guide tool CRISPick⁹³ (https://portals.broadinstitute.org/gppx/crispick/public) developed by the Zhang lab at the Broad Institute was used to design single guide RNAs (sgRNAs) to target human AAVS1 (sgRNA sequence: GGGCCACTAGGGACAGGAT) as a control, SPTLC3 (sgRNA sequence: CCCCAAGCACAAGCGATGTG and CATACTTGGCTGCAAGACCA), and SPTLC2 (sgRNA sequence: GGTTGCTGTGCTCACGTATG and GTAGAAGCTGCTATGGCGTA). These sgRNA duplexes were cloned into the lentiCRISPRv2 plasmid (Addgene #52961)⁹⁴. In one well of a 6 well plate per plasmid, human embryonic kidney HEK-293FT cells at 50% confluency were transfected with 1 μg of the lentiCRISPRv2 plasmid with sgRNA duplex, 750 ng psPAX2 and 250 ng pMD2.G using 6 μl and 74 μl Opti-MEM with a total volume of 100 μl. Medium containing viral particles was harvested at 48 and 72 hours after transfection, sterile filtered via 0.45 μm filters, followed by addition of polybrene at 6 μg/ml. Huh7 cells were infected with 0.5 ml lentivirus containing the AAVS1-, SPTLC3-, or SPTLC2-specific sgRNA to generate pooled knockouts followed by puromycin selection at 3 ug/ml. The pooled SPTLC3 knockout cell line was validated by Western blot. We did not achieve a pooled SPTLC2 knockout as shown by Western Blot.

Animal experiments

Experimental protocols were approved and performed according to the Institutional Animal Care and Use Committee (IACUC) of the Salk Institute for Biological Studies. Four-five-week-old Ldlr^−/− C57BL/6J male and female mice (JAX# 002207) were fed with irradiated 60% high fat diets (HFD) prepared by Dyets for 16 weeks. These diets include Cis Unsaturated HFD (105063GI), Cis Unsaturated HFD with 2.2 mg/kg Myriocin Added (105064GI), Trans Unsaturated HFD (105061GI), and Trans Unsaturated HFD with 2.2 mg/kg Myriocin Added (105062GI). The Trans Unsaturated HFD was designed with 100% Primex, a partially hydrogenated vegetable oil, and the Cis Unsaturated HFD was designed with 34% lard and 66% olive oil. Dietary fatty acid composition is detailed in Table S2. Mice were housed at room temperature (21°C) in the same room with a 12 hour light: dark cycle. Tissues were collected after mice were fasted for 6 hours. Mice were anesthetized with isoflurane and tissues were freeze-clamped immediately using Wollenberger clamps pre-cooled to the temperature of liquid nitrogen and stored at −80°C until analysis. Liver, epididymal white adipose tissue (eWAT), and inguinal white adipose tissue (iWAT) were weighed prior to being frozen. Blood was collected in EDTA-coated tubes (Sarstedt Inc.) and centrifuged at 2000g for 5 min. The plasma supernatant was transferred to a new Eppendorf tube and stored at −80°C until analysis.

Method Details

Glucose tolerance test

Ldlr^−/− C57BL/6J male mice (n=10) fed the diets for 15 weeks were fasted overnight with water provided ad libitum. Mice were weighed the following morning and baseline fasting blood glucose was measured via tail bleed with a Contour Next glucometer (Bayer). Mice received an intraperitoneal bolus injection of 2 g glucose/kg body weight. Blood glucose was measured via tail bleed at 15, 30, 60, 120, and 180 min post-injection.

Hepatic VLDL-TG secretion

Ldlr^−/− C57BL/6J mice (n=5: 3 female, 2 male) fed the diets for 16 weeks were fasted for 5 hours with water provided ad libitum. Mice were weighed and baseline fasting blood was collected via tail bleed. Tyloxapol (10% w/v in H₂O) at a dose of 0.5 g/kg body weight was injected via tail vein after mice were anesthetized. Plasma was collected via tail vein at 15 min, 30 min, 60 min, and 120 min post-injection. Triglyceride and cholesterol levels were quantified via enzymatic kits (234–60, 236–60, SE-035, Sekisui).

Fast protein liquid chromatography

Plasma from Ldlr^−/− C57BL/6J male mice (n=9) per diet were combined into 3 pooled plasma samples and separated via gel-filtration fast-protein liquid chromatography (FPLC) as previously described⁹⁵. Samples were run on a GE Superose 6 10/30 GL column in 0.15 M sodium chloride containing 1 mM ethylenediaminetetraacetic acid and 0.02% sodium azide with pH 7.4. Fractions of 0.5 mL were collected at a flow rate of 0.5 mL/min. Triglyceride and cholesterol levels were quantified via enzymatic kits (Sekisui, 234–60, 236–60, SE-035). Sphingomyelin was measured using 100 μl of each fraction as described below with 0.5 μl of UltimateSPLASH One Mix (Avanti Polar Lipids, Cat #330820) used per sample as internal standards for quantification.

Histology

Liver sections and the top half of the heart were fixed overnight in 10% neutral buffered formalin to perform histology as previously described^96,97. Fixed liver was washed with PBS and stored in 70% ethanol until sectioning. Fixed heart for aortic root analysis was washed with PBS and stored in PBS with 30% sucrose and 0.01% sodium azide until sectioning. Liver sections were stained with H&E (hematoxylin & eosin) to visualize hepatic steatosis. Aortic root sections were stained with modified Van Gieson to visualize atherosclerotic lesions.

Atherosclerotic plaque quantitation

Serial 5 μm sections of the aortic root from Ldlr^−/− C57BL/6J male mice (n=9–10) were stained with modified Van Gieson to measure atherosclerotic lesion sizes sequentially from the beginning of the aortic valves and area under the curve as previously described⁹⁶. Aortic root cross-sectional atherosclerotic lesion size was quantified via QuPath.

Long-chain base hydrolysis

Long-chain bases were hydrolyzed as previously described⁹⁸. Cells (~400,000) were spiked with internal standards sphinganine-d7 (Avanti Polar Lipids, Cat# 860658) and sphingosine-d7 (Avanti Polar Lipids, Cat# 860657) and were scraped with 0.5 mL methanol. Homogenate aliquot of 50 μL was taken to determine protein content using the BCA protein assay (Thermo Fisher Scientific). Samples were placed on a mixer for 1 hr at 37°C and then centrifuged for 5 min at 16,000g. The MeOH was supernatant transferred to a new Eppendorf tube and hydrolyzed for 16 hr at 65°C. 100 μL 10M KOH, 625 μL chloroform, 100 μL 2N NH₄OH, and 500 μL alkaline water were added to samples followed by vortexing for 5 min and centrifugation for 5 min at 16,000g. The lower organic phase was washed 3 times with alkaline water and dried under air. After dried extracts were resuspended in 60 μl Buffer B, 5 μL of sample was injected. Quantification of hydrolyzed long chain bases was performed on an Agilent 6460 QQQ LC-MS/MS. Metabolite separation was achieved with a C18 column (Hypersil GOLD aQ C18 100 × 2.1 mm, 1.9 μm particle size, Thermo Fisher Scientific). Mobile phase A was composed of a 60:40 ratio of methanol:water containing 0.1% formic acid and 5 mM ammonium formate. Mobile phase B consisted of 100% methanol containing 0.1% formic acid and 5 mM ammonium formate. The gradient elution program consisted of a flow rate of 0.2 ml/min with the following profile: 0 min, 40%B; 0.5 min, 40%B; 16 min, 100%B; 25.5 min, 100%B; 26 min, 40%B. The capillary voltage was set to 3.5 kV, the drying gas temperature was 350 °C, the drying gas flow rate was 10 L/min, and the nebulizer pressure was 60 psi. Long-chain bases were analyzed by multiple reaction monitoring (MRM) of the transition from precursor to product ions at associated optimized collision energies and fragmentor voltages (Table S1). Relative abundance of long-chain bases was calculated by normalizing to internal standard specific to long-chain base species and to protein levels.

Targeted sphingolipid quantification

Frozen liver (20–30 mg), plasma (100 μl), cells (~400,000 cells), or media (concentrated as described above from 0.75 ml fresh media or 1.5 ml spent media) were spiked with the following internal standards: 20 pmol sphinganine-d7 (Avanti Polar Lipids, Cat# 860658), deoxysphinganine-d3 (Avanti Polar Lipids, Cat# 860474), 100 pmol d18:0-d7/13:0 dihydroceramide (Avanti Polar Lipids, Cat# 330726), 200 pmol d18:1-d7/15:0 ceramide (Avanti Polar Lipids, Cat# 860681), 100 pmol d18:1-d7/15:0 glucosylceramide (Avanti Polar Lipids, Cat# 330729), 100 pmol d18:1-d7/15:0 lactosylceramide (Avanti Polar Lipids, Cat# 330727), 200 pmol sphingosine-d7 (Avanti Polar Lipids, Cat# 860657), and 200 pmol d18:1/18:1-d9 sphingomyelin (Avanti Polar Lipids, Cat# 791649). Sphingomyelin (d18:1/18:1)-d9 (Avanti Polar Lipids, Cat#860740) was used instead of sphingomyelin d18:1/18:1-d9 when cells were treated with oleic acid-d9 to avoid the same MRM if oleate-d9 was incorporated into SM d36:2 (Table S1). Tissue was homogenized and cells were scraped with 0.5 mL methanol and 0.5 mL H₂O. Homogenate aliquot of 100 μL was taken to determine protein content using the BCA protein assay (Thermo Fisher Scientific). The remaining homogenate was transferred to a new Eppendorf tube and 1 mL chloroform was added. For plasma or media, 0.5 mL methanol, 0.5 mL H₂O, and 1 mL chloroform were added directly. Samples were vortexed for 5 min and centrifuged for 5 min at 4 °C at 15,000g. The organic phase was collected and 2 μL of formic acid was added to the remaining polar phase which was re-extracted with 1 mL of chloroform. Combined organic phases were dried under nitrogen.

Quantification of sphingolipids was performed on an Agilent 6460 QQQ LC-MS/MS as previously described⁹⁹. Sphingolipid species were separated on a C8 column (Spectra 3 μm C8SR 150 × 3 mm inner diameter, Peeke Scientific). After dried extracts for cells and media were resuspended in 60 μl Buffer B and dried extracts for liver and plasma were resuspended in 100 μl, 5 μL of sample was injected. Mobile phase A was composed of 100% HPLC-grade water containing 2 mM ammonium formate and 0.2% formic acid, and mobile phase B consisted of 100% methanol containing 0.2% formic acid and 1 mM ammonium formate. The gradient elution program consisted of a flow rate of 0.5 ml/min with the following profile: 0 min, 82% B; 3 min, 82% B; 4 min, 90% B, 18 min, 99% B; 25 min, 99% B; 27 min; 82% B; 30 min, 82% B. Column re-equilibration followed each sample and lasted 10 min. The capillary voltage was set to 3.5 kV, the drying gas temperature was 350 °C, the drying gas flow rate was 10 L/min, and the nebulizer pressure was 60 psi. Sphingolipid species were analyzed by multiple reaction monitoring (MRM) of the transition from precursor to product ions at associated optimized collision energies and fragmentor voltages (Table S1). MRMs for sphinganine (SA), sphingosine (SO), and sphingosine-1-phosphate (S1P) were scheduled from 0–10 min. MRMs for dihydroceramide (DHCer), ceramide (Cer), hexosyl-ceramide (Hex-Cer), lactosyl-ceramide (Lac-Cer), and sphingomyelin (SM) were scheduled from 10–30 min. Relative abundance of sphingolipids was calculated by normalizing to internal standard specific to sphingolipid class and to protein levels.

Targeted lipid quantification

Cells (~400,000 cells) or media (concentrated as described above from 0.75 ml fresh media or 1.5 ml spent media) were spiked with 1 μg of each of the following internal standards: 15:0–18:1(d7) phosphatidylcholine (Avanti Polar Lipids, Cat #791637), 15:0–18:1(d7) phosphatidylethanolamine (Avanti Polar Lipids, Cat #791638), 18:1(d7) lysophosphatidylcholine (Avanti Polar Lipids, Cat#791643), 18:1(d7) lysophosphatidylethanolamine (Avanti Polar Lipids, Cat #791644), 15:0–18:1(d7) diacylglycerol (Avanti Polar Lipids, Cat #791647), 15:0–18:1(d7)-15:0 triacylglycerol (Avanti Polar Lipids, Cat #791648). Cells were scraped with 0.5 mL methanol and 0.5 mL H₂O. Homogenate aliquot of 100 μL was taken to determine protein content using the BCA protein assay (Thermo Fisher Scientific). The remaining homogenate was transferred to a new Eppendorf tube and 1 mL chloroform was added. For media, 0.5 mL methanol, 0.5 mL H₂O, and 1 mL chloroform were added directly. Samples were vortexed for 5 min and centrifuged for 5 min at 4 °C at 15,000g. The organic phase was collected and 2 μL of formic acid was added to the remaining polar phase which was re-extracted with 1 mL of chloroform. Combined organic phases were dried under nitrogen.

Targeted lipids were quantified on an Agilent 6460 QQQ LC-MS/MS equipped with an Accucore C30, 150 × 2.1 mm, 2.6 μm particle (Thermo Fisher Scientific) column at 40°C. After dried extracts were resuspended in 60 μl 65:30:5 ACN: IPA:H₂O, 5 μL of sample was injected. Mobile phase A was composed of a 60:40 ratio of acetonitrile:water containing 10 mM ammonium formate and 0.1% formic acid and mobile phase B consisted of a 90:10 ratio of isopropanol:acetonitrile with 10 mM ammonium formate and 0.1% formic acid. The liquid chromatography gradient ran with the following profile: 0 min, 30% B; 3 min, 30% B; 8 min, 43% B; 9 min, 50% B; 18 min, 90% B; 26 min, 99% B; 30 min, 99%B; 36 min, 30% B. The flow rate was 0.1 ml/min for the first 8 min and 0.2 ml/min for the remainder of the method. Column re-equilibration followed each sample and lasted 10 min. The capillary voltage was set to 3.5 kV, the drying gas temperature was 350 °C, the drying gas flow rate was 10 L/min, and the nebulizer pressure was 60 psi. Lipids were analyzed by multiple reaction monitoring (MRM) of the transition from precursor to product ions at associated optimized collision energies and fragmentor voltages (Table S1). Relative abundance of lipids was calculated by normalizing to internal standard specific to lipid class and to protein levels.

General lipidomics quantification

Frozen liver (10–20 mg) or plasma (10 μl) was spiked with 2 ug (for liver) or 1 ug (for plasma) of each of the following internal standards: 18:1-d7 cholesteryl ester (Avanti Polar Lipids, Cat #791645), 15:0–18:1(d7) phosphatidylcholine (Avanti Polar Lipids, Cat #791637), 15:0–18:1(d7) phosphatidylethanolamine (Avanti Polar Lipids, Cat #791638), 18:1(d7) lysophosphatidylcholine (Avanti Polar Lipids, Cat#791643), 18:1(d7) lysophosphatidylethanolamine (Avanti Polar Lipids, Cat #791644), 15:0–18:1(d7) diacylglycerol (Avanti Polar Lipids, Cat #791647), 15:0–18:1(d7)-15:0 triacylglycerol (Avanti Polar Lipids, Cat #791648). Tissue was homogenized with 0.5 mL methanol and 0.5 mL H₂O. Homogenate aliquot of 100 μL was taken to determine protein content using the BCA protein assay (Thermo Scientific). The remaining homogenate was transferred to a new Eppendorf tube and 1 mL chloroform was added. For plasma, 0.5 mL methanol, 0.5 mL H₂O, and 1 mL chloroform were added directly. Samples were vortexed for 5 min, centrifuged for 5 min at 4 °C at 15,000g. The organic phase was collected and 2 μL of formic acid was added to the remaining polar phase which was re-extracted with 1 mL of chloroform. Combined organic phases were dried under nitrogen.

Chromatographic separation and lipid species identification was performed using Q Exactive orbitrap mass spectrometer with a Vanquish Flex Binary UHPLC system (Thermo Scientific) equipped with an Accucore C30, 150 × 2.1 mm, 2.6 μm particle (Thermo) column at 40 °C. After dried extracts for liver were resuspended in 100 μl and dried extracts for plasma were resuspended in 60 μl 65:30:5 ACN: IPA:H₂O, 5 μL of sample was injected. Chromatography was performed using a gradient of 40:60 v/v water: acetonitrile with 10 mM ammonium formate and 0.1% formic acid (mobile phase A) and 10:90 v/v acetonitrile: propan-2-ol with 10 mM ammonium formate and 0.1% formic acid (mobile phase B), both at a flow rate of 0.2 mL/min. The liquid chromatography gradient ran with the following profile: 0 min, 30% B; 3 min, 30% B; 8 min, 43% B; 9 min, 50% B; 18 min, 90% B; 26 min, 99% B; 30 min, 99%B; 36 min, 30% B.Lipids were analyzed in positive mode using spray voltage 3.2 kV. Sweep gas flow was 1 arbitrary units, auxiliary gas flow 2 arbitrary units and sheath gas flow 40 arbitrary units, with a capillary temperature of 325 °C. Full mass spectrometry (scan range 200–2,000 m/z) was used at 70,000 resolution with 10E6 automatic gain control and a maximum injection time of 100 ms. Data dependent MS2 (Top 6) mode at 17,500 resolution with automatic gain control set at 105 with a maximum injection time of 50 ms was used. Lipids were identified in EL-MAVEN using exact mass of precursor ion in MS1 chromatogram and product ion in MS2 spectra (Table S4). Lipids were quantified using peak areas in MS1 chromatogram. Relative abundance of lipids was calculated by normalizing to internal standard specific to lipid class and to protein levels or volume plasma.

In vitro sphingolipid biosynthetic flux

To measure intracellular sphingolipid biosynthetic flux in Huh7 control and SPTLC3 KO cell lines, we treated cells with 100 μM BSA-palmitate and performed stable isotope tracing with a combination [¹³C₃]serine and [¹³C₂]glycine for 48 hours in lipidated media. To measure sphingomyelin secretory flux, cells were similarly treated with 100 μM BSA-palmitate and traced with [¹³C₃]serine and [¹³C₂]glycine for 48 hours in delipidated media to increase the percentage of labeled lipids in spent media.

Cells (~400,000 cells), or media (concentrated as described above from 0.75 ml fresh media or 1.5 ml spent media) were spiked with the following internal standards: 100 pmol d18:0-d7/13:0 dihydroceramide (Avanti Polar Lipids, Cat# 330726), 200 pmol d18:1-d7/15:0 ceramide (Avanti Polar Lipids, Cat# 860681), 100 pmol d18:1-d7/15:0 glucosylceramide (Avanti Polar Lipids, Cat# 330729), 100 pmol d18:1-d7/15:0 lactosylceramide (Avanti Polar Lipids, Cat# 330727), 200 pmol sphingosine-d7 (Avanti Polar Lipids, Cat# 860657), and 200 pmol d18:1/18:1-d9 sphingomyelin (Avanti Polar Lipids, Cat# 791649). Cells were scraped with 0.5 mL methanol and 0.5 mL H₂O. Homogenate aliquot of 100 μL was taken to determine protein content using the BCA protein assay (Thermo Fisher Scientific). The remaining homogenate was transferred to a new Eppendorf tube and 1 mL chloroform was added. For media, 0.5 mL methanol, 0.5 mL H₂O, and 1 mL chloroform were added directly. Samples were vortexed for 5 min and centrifuged for 5 min at 4 °C at 15,000g. The organic phase was collected and 2 μL of formic acid was added to the remaining polar phase which was re-extracted with 1 mL of chloroform. Combined organic phases were dried under nitrogen.

Chromatographic separation and lipid species identification was performed using Q Exactive orbitrap mass spectrometer with a Vanquish Flex Binary UHPLC system (Thermo Fisher Scientific) equipped with an Kinetex C18 column, 100 × 2.1 mm, 1.7 μm particle (Phenomenex) column at 40 °C. After dried extracts for cells or media were resuspended in 60 μl MeOH, 5 μL of sample was injected. Chromatography was performed using a gradient of 98:2 v/v water: methanol with 5 mM ammonium acetate (mobile phase A) and 50:50 v/v methanol: isopropanol with 5 mM ammonium acetate (mobile phase B), both at a flow rate of 0.2 mL/min. The liquid chromatography gradient ran with the following profile: 0 min, 30%B; 1 min, 30%B; 2 min, 70%B; 11 min, 95%B; 17 min, 30%B; 21.5 min, 30%B; 27 min, 30%B. Lipids were analyzed in positive mode using spray voltage 3.2 kV. Sweep gas flow was 1 arbitrary units, auxiliary gas flow 2 arbitrary units and sheath gas flow 40 arbitrary units, with a capillary temperature of 325 °C. Full mass spectrometry (scan range 200–2,000 m/z) was used at 140,000 resolution with 10E6 automatic gain control and a maximum injection time of 100 ms. Data dependent MS2 (Top 12) mode at 17,500 resolution with automatic gain control set at 10E6 with a maximum injection time of 50 ms was used. Lipids were identified and quantified using EL-MAVEN using exact mass of precursor ion in MS1 chromatogram and product ion in MS2 spectra (Table S4). Relative abundance of lipids was calculated by normalizing to internal standard specific to lipid class and to protein levels. The percent isotopologue distribution of each sphingolipid was determined and corrected for natural ¹³C abundance using in-house algorithms adapted from a previous report¹⁰⁰.

²H₂O Lipid biosynthesis measurements

Ldlr^−/− C57BL/6J male mice (n=5) fed the diets for 16 weeks were administered ²H₂O in 0.9% NaCl at a dose of 0.027 mL/g body weight via intraperitoneal injection. Drinking water was replaced with 8% ²H₂O drinking water for 30 hours. 24 hours post-injection, mice were fasted for 6 hours with 8% ²H₂O drinking water provided ad libitum. Tissues and blood were collected as described above.

²H₂O enrichment in plasma from samples or standards was measured via deuterium-acetone exchange. 5 μL of sample or standard was reacted with 4 μL of 10N NaOH and 4 μL of 5% solution of acetone in acetonitrile for 24 hours. Acetone was extracted after addition of 500 mg of Na₂SO₄ and 600 μL of chloroform. After 2 min centrifugation at 3000 g, 80 μL was transferred in triplicate into a GC-MS vial and plasma ²H₂O enrichment was quantified from external standard curve on an Agilent DB-35MS column (30 m x 0.25 mm i.d. × 0.25 μm, Agilent J&W Scientific) installed in an Agilent 7890 A gas chromatograph (GC) interfaced with an Agilent 5975 C mass spectrometer (MS) with the following temperature program: 60 °C initial, increase by 20 °C/min to 100 °C, increase by 50 °C/min to 220 °C, and hold for 1 min.

De novo lipogenesis

De novo lipogenesis via ²H₂O enrichment in liver and plasma was quantified by spiking 10–20 mg of frozen liver or 10 μL of plasma with internal standards 15 nmol palmitate-d31 (Cambridge Isotope Laboratories) and 15 μg coprostanol (Sigma, Cat# 7578) and homogenizing with 250 μL methanol and 250 μL water. Homogenate aliquot of 50 μL was taken to determine protein content using the BCA protein assay (Thermo Fisher Scientific). 500 μL chloroform was added to the remaining homogenate, then samples were vortexed for 5 min and centrifuged for 5 min at 4 °C and 15 000g. The chloroform phase was collected, dried, and resuspended with 500 μL 2% H₂SO₄ in methanol for 2 hours at 50 °C. 100 μL of saturated NaCl and 500 μL of hexane were added, sample were vortexed, and the upper hexane phase containing fatty acid methyl esters (FAMEs) was collected and transferred into a GC-MS vial. FAMES were analyzed using a Select FAME column (100 m × 0.25 mm i.d.) installed in an Agilent 7890 A GC interfaced with an Agilent 5975 C MS using the following temperature program: 80 °C initial, increase by 20 °C/min to 170 °C, increase by 1 °C/min to 204 °C, then 20 °C/min to 250 °C and hold for 10 min. The percent isotopologue distribution of each fatty acid was determined and corrected for natural abundance using in-house algorithms adapted from a previous report¹⁰⁰. Relative abundance is calculated by normalizing to respective internal standards (palmitate-d31 for fatty acids, coprostanol for cholesterol) and to protein levels. Fractional synthesis of newly synthesized palmitate and cholesterol was calculated based on a previous method¹⁰¹.

Sphingolipid biosynthesis

Sphingolipid biosynthesis via ²H₂O enrichment in plasma was quantified by spiking 10 μL of plasma with internal standards 100 pmol d18:0-d7/13:0 dihydroceramide (Avanti Polar Lipids, Cat# 330726), 200 pmol d18:1-d7/15:0 ceramide (Avanti Polar Lipids, Cat# 860681), and 1 ug 15:0–18:1(d7) phosphatidylcholine (Avanti Polar Lipids, Cat #791637). 0.5 mL methanol, 0.5 mL H₂O, and 1 mL chloroform were added directly. Samples were vortexed for 5 min, centrifuged for 5 min at 4 °C at 15,000g. The organic phase was collected and 2 μL of formic acid was added to the remaining polar phase which was re-extracted with 1 mL of chloroform. Combined organic phases were dried under nitrogen.

Chromatographic separation and lipid species identification was performed using Q Exactive orbitrap mass spectrometer with a Vanquish Flex Binary UHPLC system (Thermo Fisher Scientific) equipped with an Kinetex C18 column, 100 × 2.1 mm, 1.7 μm particle (Phenomenex) column at 40 °C. After dried extracts for plasma were resuspended in 60 μl MeOH, 5 μL of sample was injected. Chromatography was performed using a gradient of 98:2 v/v water: methanol with 5 mM ammonium acetate (mobile phase A) and 50:50 v/v methanol: isopropanol with 5 mM ammonium acetate (mobile phase B), both at a flow rate of 0.2 mL/min. The liquid chromatography gradient ran with the following profile: 0 min, 30%B; 1 min, 30%B; 2 min, 70%B; 11 min, 95%B; 17 min, 30%B; 21.5 min, 30%B; 27 min, 30%B. held at 30% B for 10 min. Lipids were analyzed in positive mode using spray voltage 3.2 kV. Sweep gas flow was 1 arbitrary units, auxiliary gas flow 2 arbitrary units and sheath gas flow 40 arbitrary units, with a capillary temperature of 325 °C. Full mass spectrometry (scan range 200–2,000 m/z) was used at 140,000 resolution with 10E6 automatic gain control and a maximum injection time of 100 ms. Data dependent MS2 (Top 12) mode at 17,500 resolution with automatic gain control set at 10E6 with a maximum injection time of 50 ms was used. Relative abundance of lipids was calculated by normalizing to internal standard specific to lipid class and volume of plasma. Sphingomyelin was normalized to 15:0–18:1(d7) phosphatidylcholine as a deuterated sphingomyelin standard would have interfered with the mass isotopologue distribution of specific sphingomyelin species Lipids were identified and quantified using EL-MAVEN using exact mass of precursor ion in MS1 chromatogram and product ion in MS2 spectra (Table S4). The percent isotopologue distribution of each sphingolipid was determined and corrected for natural ¹³C abundance, the most abundant natural stable isotope, using in-house algorithms adapted from a previous report¹⁰⁰.

Fractional synthesis of newly synthesized sphingolipids was adapted from previous methods^101,102 using the following equation:

$$FNS=\frac{ME}{n\times p}$$

ME is the average number of deuterium atoms incorporated per molecule$\left(ME=1\times M1+2\times M2+3\times M3+\cdots \right)$, p is the deuterium enrichment in body water, and n is the maximum number of hydrogen atoms from water incorporated per molecule. The theoretical n was calculated by following deuterium incorporation from water, NADPH, NADH, or acetyl-CoA in the synthesis of precursors (serine and palmitate) and directly in sphingolipid biosynthesis. We assume serine can incorporate 2 deuterium atoms through phosphoserine phosphatase (PSPH) at the beta hydroxy group and at the N terminus through protein hydrolysis. We assume palmitate can incorporate 22 deuterium atoms through de novo lipogenesis¹⁰¹. Following condensation by SPT, KDSR can incorporate 2 additional deuterium atoms through NADPH + H⁺ to synthesize a d18:0 LCB, which can incorporate 26 deuterium atoms. We assume any desaturation step during fatty acid synthesis or via DEGS1/FADS3 loses 1 deuterium atom. For example, a d18:1 LCB can incorporate 25 deuterium atoms. We assume fatty acids elongated from palmitate incorporate 3 additional deuterium atoms per acetyl-CoA addition. Limitations of this approach include that synthesis of elongated fatty acids may be underestimated as it does not correct for elongation of unlabeled fatty acids. Synthesis of unsaturated fatty acids may also be underestimated as it similarly does not correct for desaturation of unlabeled fatty acids¹⁰³. Theoretical n values for sphingolipids are reported in Table S5.

RNA isolation and quantitative RT-PCR

RNA was extracted from 10–20 mgs liver or ~400,000 cells using Direct-zol RNA kit (Direct-Zol RNA Miniprep Plus kit, Zymo Research). cDNA synthesis was performed using iScript Reverse Transcription Supermix for RT-PCR (iScript Reverse Transcription Supermix, Bio-Rad) with the following thermocycler protocol: 5 min at 25°C, 20 min 46°C, 1 min 95°C. PCR reactions were carried out using 96-well plates on an Applied Biosystems ViiA 7 Real-Time PCR System using the following parameters: 95°C for 20 s, 40 cycles of 95°C for 1 s, and 60°C for 20 s. The final volume (10 μL) of PCR SYBR-Green reaction consisted of 5 μL fast SYBR-Green Master Mix (Bio-Rad), 2 μL cDNA, 1 μL of 10 μM forward and reverse primers, and 2 μL of water. Primers are noted in Table S6.

Western blot

Huh7 AAVS1, SPTLC3 KO, and SPTLC2 KO cells were lysed in Radio-Immunoprecipitation Assay (RIPA) lysis buffer (Thermo Fisher Scientific, J62524.AD) with Halt^™ Protease Inhibitor Cocktail (1x) and 5 mM EDTA solution (Thermo Fisher Scientific, 78429). Protein concentrations were determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23227). 20 μg of protein and PageRuler Plus Prestained Protein Ladder (Thermo Fisher Scientific, 26619) were loaded and separated using 4 to 15% Mini-PROTEAN TGX Precast SDS-PAGE Gels (Bio-Rad, #4561086). Protein samples were then transferred to a nitrocellulose membrane. Membranes were blotted for Vinculin (7F9) (Santa Cruz Biotechnology, sc-73614), SPTLC3 (Abcam, ab237532), and SPTLC2 (Proteintech, 51012–2-AP). Anti-mouse (Cell Signaling Technology, 7076S) and anti-rabbit (Cell Signaling Technology, 7074S) horseradish peroxidase-conjugated antibodies were used along with Clarity Western ECL Substrate (Bio-Rad, #170–5061) for imaging.

Gene co-expression correlation analysis

Co-expression analysis of SPTLC2 and SPTLC3 in human liver was performed using the online tool CorrelationAnalyzeR⁷⁰. SPTLC2 and SPTLC3 were selected to be analyzed in normal liver tissue in the gene versus gene comparison platform. The top 100 differentially correlated genes with SPTLC2 and SPTLC3, as determined by their Pearson correlation coefficients (r), are listed in Table S3. These correlations coefficients against SPTLC2 and SPTLC3 for each gene are plotted in Figure 5J.

Quantification and Statistical Analysis

Data are presented as mean ± standard error of mean (SEM) of at least three biological replicates as indicated in figure legends. Statistical analysis was performed with GraphPad Prism 10.2.1 using two-tailed independent t-test to compare two groups, one-way ANOVA with Fisher’s least significant difference (LSD) post hoc test to compare more than two groups, two-way ANOVA with Fisher’s LSD post hoc test to compare two-factor study designs, and Pearson correlation coefficient (r) for gene co-expression correlation analysis. For all tests, p<0.05 was considered significant with *p < 0.05, **p < 0.01, *** p< 0.001, or **** p<0.0001 unless otherwise noted.

Supplementary Material

1

Data S1. Unprocessed data underlying display items in the manuscript, related to Figures 1–6 and S1-6.

2

Table S1. Lipid multiple reaction monitoring (MRMs), collision energies, and fragmentor voltages for LC-MS, related to Figures 1–2 and STAR Methods

3

Table S3. Top 100 differentially correlated genes with SPTLC2 and SPTLC3 in human liver, related to Figure 6

4

Table S4. Orbitrap high-resolution QE mass spectrometry lipid analysis, related to STAR Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Vinculin (7F9) antibody	Santa Cruz Biotechnology	Cat# sc-73614; RRID:AB_1131294
SPTLC3 antibody	Abcam	Cat# ab237532
SPTLC2 antibody	Proteintech	Cat#51012–2-AP; RRID:AB_2195870
Anti-mouse IgG, HRP-linked antibody	Cell Signaling Technology	Cat#7076S
Anti-rabbit IgG, HRP-linked antibody	Cell Signaling Technology	Cat#7074S
Bacterial and virus strains
One Shot Stbl3 E. Coli	Thermo Fisher Scientific	Cat#C7373
Biological samples
Mouse liver	This paper	N/A
Mouse plasma	This paper	N/A
Mouse aorta	This paper	N/A
Chemicals, peptides, and recombinant proteins
DMEM	Thermo Fisher Scientific	Cat#11965–092
Fetal bovine serum	Thermo Fisher Scientific	Cat#16000–044
Penicillin-streptomycin	Thermo Fisher Scientific	Cat#15140–122
DPBS	Thermo Fisher Scientific	Cat#14190–144
Bovine serum albumin	Sigma-Aldrich	Cat#A6003
Silica, fumed	Sigma-Aldrich	Cat#S5130
Ethyl alcohol, pure	Sigma-Aldrich	Cat#E7023
Palmitic acid	Sigma-Aldrich	Cat#P0500
Iso-palmitic acid	Larodan	Cat#11–1514-13
Heptadecanoic acid	Sigma-Aldrich	Cat#H3500
Iso-heptadecanoic acid	Larodan	Cat#11–1615-13
Anteiso-heptadecanoic acid	Larodan	Cat#11–1614-13
Stearic acid	Sigma-Aldrich	Cat#S4751
Oleic acid	Sigma-Aldrich	Cat#O1008
Oleic acid-d9	Avanti Polar Lipids	Cat#861809
Elaidic acid	Sigma-Aldrich	Cat#E4637
Elaidic acid-d17	Cayman Chemical	Cat#27715
DMSO	Thermo Fisher Scientific	Cat#BP231–100
Myriocin	Cayman Chemical	Cat#63150
D-(+)-Glucose	Sigma-Aldrich	Cat#G8270
Tyloxapol	Santa Cruz Biotechnology	Cat#255711
Sodium Chloride	Sigma-Aldrich	Cat#S9888
Sodium azide, 0.5% w/v	Thermo Fisher Scientific	Cat#LC229401
10% Neutral buffered formalin	Sigma-Aldrich	Cat#HT501128
MSTFA Silylation Reagent	Macherey-Nagel	Cat# 701270.510
D2O	Sigma-Aldrich	Cat#151882
Sodium hydroxide	Sigma-Aldrich	Cat#S5881
Acetone	Sigma-Aldrich	Cat#650501
Acetonitrile	Sigma-Aldrich	Cat#34851
Sodium sulfate	Sigma-Aldrich	Cat#238597
Palmitate-d31	Cambridge Isotope Laboratories	Cat#DLM-215–1
Coprostanol	Cayman Chemical	Cat#26764
Methanol	Sigma-Aldrich	Cat#34860
Chloroform	Sigma-Aldrich	Cat#366927
Water	Sigma-Aldrich	Cat#270733
2-propanol	Sigma-Aldrich	Cat#34863
LiChrosolv 2-propanol	Sigma-Aldrich	Cat#1.02781
LiChrosolv methanol	Sigma-Aldrich	Cat#1.06035
LiChrosolv acetonitrile	Sigma-Aldrich	Cat#1.00029
Sulfuric acid	Sigma-Aldrich	Cat#258105
Hexane	Sigma-Aldrich	Cat#293253
Potassium hydroxide	Sigma-Aldrich	Cat#484016
Ammonium hydroxide	Sigma-Aldrich	Cat#AX1303
Formic acid	Sigma-Aldrich	Cat#5.43804
Ammonium formate	Sigma-Aldrich	Cat#70221
Ammonium acetate	Sigma-Aldrich	Cat#73594
Sphinganine-d7	Avanti Polar Lipids	Cat#860658
Deoxysphinganine-d3	Avanti Polar Lipids	Cat#860474
d18:0-d7/13:0 dihydroceramide	Avanti Polar Lipids	Cat#330726
d18:1-d7/15:0 ceramide	Avanti Polar Lipids	Cat#860681
d18:1-d7/15:0 glucosylceramide	Avanti Polar Lipids	Cat#330729
d18:1-d7/15:0 lactosylceramide	Avanti Polar Lipids	Cat#330727
Sphingosine-d7	Avanti Polar Lipids	Cat# 860657
d18:1/18:1-d9 sphingomyelin	Avanti Polar Lipids	Cat#791649
(d18:1/18:1)-d9 sphingomyelin	Avanti Polar Lipids	Cat#860740
15:0–18:1(d7) phosphatidylcholine	Avanti Polar Lipids	Cat#791637
15:0–18:1(d7) phosphatidylethanolamine	Avanti Polar Lipids	Cat#791638
18:1(d7) lysophosphatidylcholine	Avanti Polar Lipids	Cat#791643
18:1(d7) lysophosphatidylethanolamine	Avanti Polar Lipids	Cat#791644
15:0–18:1(d7) diacylglycerol	Avanti Polar Lipids	Cat#791647
15:0–18:1(d7)-15:0 triacylglycerol	Avanti Polar Lipids	Cat#791648
UltimateSPLASH One Mix	Avanti Polar Lipids	Cat#330820
iScript Reverse Transcription Supermix	Bio-Rad	Cat#1708840
SYBR-Green Master Mix	Bio-Rad	Cat#1725124
Radio-Immunoprecipitation Assay (RIPA) lysis buffer	Thermo Fisher Scientific	Cat#J62524.AD
Halt™ Protease Inhibitor Cocktail	Thermo Fisher Scientific	Cat#78429
[13C3]serine	Cambridge Isotope Laboratories	Cat#CLM-1574-H
[13C2]glycine	Cambridge Isotope Laboratories	Cat#CLM-1017
Puromycin	GoldBio	Cat#P600
Opti-MEM	Thermo Fisher Scientific	Cat#31985–062
Fugene 6 Transfection REagent	Promega	Cat#E2691
Clarity Western ECL Substrate	Bio-Rad	Cat#170–5061
Novex Tris-Glycine SDS Running Buffer (10X)	Thermo Fisher Scientific	Cat#LC2675–5
20X TBS Tween-20 Buffer	Thermo Fisher Scientific	Cat#28360
NuPage Transfer Buffer (20X)	Thermo Fisher Scientific	Cat#NP0006–1
Critical commercial assays
Triglyceride Kit	Sekisui	Cat#236–60
Cholesterol Kit	Sekisui	Cat#234–60
DC-Cal Multi-Analyte Calibrator	Sekisui	Cat#SE-035
Pierce BCA Kit	Thermo Fisher Scientific	Cat#23227
Direct-zol RNA MiniPrep Plus kit	Zymo Research	Cat#R2072
Plasmid Mini Kit	Qiagen	Cat#12123
QIAquick Gel Extraction Kit	Qiagen	Cat#28704
HiSpeed Plasmid Midi Kit	Qiagen	Cat#12643
Deposited data
Mass spectrometry data (Project ID: PR002085)	Metabolomics Workbench	doi: 10.21228/M83R6P
Data S1	This paper	N/A
Experimental models: Cell lines
Huh7 cells	Provided by M. Hermann, MIT, Cambridge, MA, USA	RRID:CVCL_0336
HEK 293FT	Thermo Fisher Scientific	Cat#R70007; RRID:CVCL_6911
Experimental models: Organisms/strains
Ldlr−/− C57BL/6J	Jackson Laboratories	Strain #:002207; RRID:IMSR_JAX:002207
Oligonucleotides
Primers for mouse sphingolipid genes, see Table S6	This paper	N/A
sgRNA to target human AAVS1: : GGGCCACTAGGGACAGGAT	This paper	N/A
sgRNA to target SPTLC3: CCCCAAGCACAAGCGATGTG and CATACTTGGCTGCAAGACCA	This paper	N/A
sgRNA to target SPTLC2: GGTTGCTGTGCTCACGTATG and GTAGAAGCTGCTATGGCGTA	This paper	N/A
Recombinant DNA
Lenti CRISPR v2 plasmid	Addgene	Cat#52961
psPAX2		Cat#12260
pMD2.G		Cat#12259
Software and algorithms
MATLAB R2022a	MathWorks	https://www.mathworks.com/products/matlab.html
GraphPad Prism 10.2.1	GraphPad	https://www.graphpad.com/scientific-software/prism/
EL-MAVEN v0.11.0	Elucidata	https://docs.polly.elucidata.io/Apps/Metabolomic%20Data/El-MAVEN.html
CRISPick	Doench et all.92	https://portals.broadinstitute.org/gppx/crispick/public)
QuPath 0.4.3	QuPath	https://qupath.github.io/
Other
DB-35 MS column, 30 m x 0.25 mm i.d. × 0.25 μm	Agilent	Cat#122–3832UI
Select FAME column, 100 m × 0.25 mm i.d.	Agilent	Cat#CP7420
Hypersil GOLD aQ C18 100 × 2.1 mm, 1.9 μm particle size	Thermo Fisher Scientific	Cat#25002–102130
Spectra 3 μm C8SR 150 × 3 mm inner diameter	Peeke Scientific	Cat# S-3C8SR-FJ
Accucore C30, 150 × 2.1 mm, 2.6 μm particle	Thermo Fisher Scientific	Cat# 27826–152130
Kinetex C18 100 × 2.1 mM, 1.7 μm particle	Phenomenex	Cat#00D-4475-AN
4 to 15% Mini-PROTEAN TGX Precast SDS-PAGE Gels	Bio-Rad	Cat#4561086
Cis Unsaturated HFD	Dyets	Cat#105063GI
Cis Unsaturated HFD + Myriocin	Dyets	Cat#105064GI
Trans Unsaturated HFD	Dyets	Cat#105061GI
Trans Unsaturated HFD + Myriocin	Dyets	Cat#105062GI

Highlights.

• TFAs are preferred over CFAs by SPT to synthesize long-chain sphingoid bases

• Trans fat-derived sphingomyelin are highly secreted from cells

• TFA-enriched HFD promotes hepatic SPT flux, VLDL secretion, and atherosclerosis

• SPT inhibition reduces circulating VLDLs and atherosclerosis in Ldlr^/− mice