Six subjects with heterozygous mutations in ABCA1, one compound heterozygote ABCA1 and eight unaffected members of a Dutch family were studied. Subject description, sample collection, ethics approval and plasma lipids/apolipoprotein quantification have been reported previously (Karuna et al., 2010; Karuna et al., 2011). ApoB-depleted human plasma was generated by precipitation with dextran sulfate/Mg²⁺ as described previously (Karuna et al., 2011).

Abca1 HSKO mice were generated as described previously (Timmins et al., 2005). Mice were housed in the Wake Forest University School of Medicine animal facility with a 12 h light/12 h dark cycle and fed a commercial chow diet ad libitum. All animal procedures were approved by the Institutional Animal Care and Use Committee of Wake Forest University School of Medicine. Abca1 HSKO and Abca1 fl/flWT littermates (hereafter referred to as WT) were derived from heterozygous mating. Seventeen week-old mice were sacrificed after a 4 h fast or in the non-fasting state.

Mouse plasma S1P were analysed as described previously (Karuna et al., 2011).

Primary hepatocytes were isolated as described previously (Chung et al., 2010) with minor modifications. After isolation from liver, hepatocytes were centrifuged at 50 × g for 5 min in a 50% Percoll-containing Williams’ media E to pellet live cells, which were then washed with Williams’ media E before seeding in 35 mm dishes at a density of 0.4 × 10⁶ cells per dish.

Hepatocytes were cultured for 2 h in Williams’ E media and then switched to MEM media for 2 h. For labelling hepatocytes were incubated with MEM for 24 h containing 1 mM d3-serine and 2 mM C13-alanine (Cambridge Isotope Laboratories) with or without myriocin (1 μg/mL in MetOH, Sigma). Cells were pelleted and lysed with extraction buffer (50 mM HEPES, pH 7.4, 1 mM EDTA, 0.2% Triton X100) and protein concentrations measured by BCA. 90 µg of protein in 100 µL of extraction buffer were used for lipid extraction and sphingoid base analysis.

The sphingoid base profile was analysed as described before (Othman et al., 2012). Tissues were homogenized in PBS [pH 7.4) with 0.2% Triton X-100 (vol/vol)] using a Precellys 24 tissue homogenizer (Bertin Technologies). 100 µL plasma, 80 µg tissue homogenate or 90 μg cell lysates were extracted for lipid analysis. Methanol (0.5 mL) spiked with 200 pmol of the internal standards d7-sphingosine and d7-sphinganine (d7SA, d7SO; Avanti Polar Lipids) were added to the aliquot of plasma or tissue homogenate.

The sphingoid bases were separated on a C₁₈ column (Uptispere 120 Å, 5 µm, 125 mm × 2 mm, Interchim) and analysed on a TSQ Quantum Ultra MS (Thermo) as described before (Mwinyi et al., 2017; Lone et al., 2020). Intra- and inter-assay variation (CV %) of the method was between 5% and 20%.

14 different sphingoid base backbones were analysed (C₁₆SO, C₁₆SA, C₁₇SO, C₁₇SA, C₁₈SO, C₁₈SA, C₁₈SAdiene, C₁₉SO, C₂₀SO, C₂₀SA, deoxysphinganine (1-deoxySA), deoxysphingosine (1-deoxySO), deoxymethylsphinganine (doxmethSA), deoxymethylsphingosine (doxmethSO) sphingoid bases were reported.

Plasma from WT and Abca1 HSKO mice was subjected to a 12% SDS-PAGE and transferred to a nitrocellulose (Schleicher and Schuell). Membranes were blocked with 5% milk in TBST buffer, incubated with primary antibody at 4°C overnight, washed three times, and then incubated with secondary antibody for 1 h at room temperature. Blots were exposed using Supersignal West Pico chemiluminescence substrate (Pierce) and visualized with a Fujifilm LAS-3000 camera. Antibodies for Western blots included: anti-mouse apoM antibody (LifeSpan), anti-mouse apoA-I primary antibody (Biodesign), and anti-rabbit secondary antibody (GE Healthcare). Quantification of band intensity was performed using Multi Gauge software.

Two-tailed Student’s t-test was used to compare results for human ABCA1 mutations (normal vs. heterozygotes) and mouse genotypes (WT vs. Abca1 HSKO). p-value < 0.05 was considered statistically significant.

The plasma of a single Tangier patient with compound heterozygous mutations in ABCA1 showed a drastic reduction in all plasma sphingoid bases. The relative decrease was about 70% for the most abundant sphingoid bases (C18SO and C16SO), but less pronounced for the minor species. In contrast, heterozygous (Het) mutations in ABCA1 show reduced plasma HDL-cholesterol, apoA-I, and S1P compared to their family controls (WT) (Karuna et al., 2010; Karuna et al., 2011), but did not show a significant difference in total plasma SL levels compared to the unaffected family members (Figures 1A, B). However, in apoB-depleted plasma, which lacks VLDL and LDL but still contains HDL and albumin, C18SO and C16SO were almost absent in the Tangier patient plasma and C18SO was also significantly reduced in the Het carriers (Figures 2A, B). Collectively, our data suggest a decrease in HDL sphingolipids in humans with ABCA1 mutations, indicating that ABCA1 activity determines sphingolipid levels, most likely by maintaining HDL particle concentration.

We previously reported that hepatic ABCA1 activity contributes ∼80% of the plasma HDL-cholesterol pool in chow-fed mice (Timmins et al., 2005). To determine whether hepatocyte ABCA1 is also a major contributor to plasma HDL sphingolipids, we compared the sphingoid base profile between Abca1 HSKO mice and WT littermate controls (Abca1 fl/fl). As shown in Figure 3, there was a significant decrease in plasma levels of C₁₈SO, C₁₈SA, C₁₈SAdiene C₁₆SO, and C₁₇SO but not of C₂₀SO in Abca1 HSKO mice Because plasma triglyceride concentrations are elevated in Tangier disease subjects (Clee et al., 2000; Kolovou et al., 2003) and non-fasted Abca1 HSKO mice (Chung et al., 2010), we performed a similar sphingoid base profiling experiment with non-fasted Abca1 HSKO mouse plasma, which gave similar results (Supplementary Figure S1). Collectively, our data showed that hepatocyte ABCA1 deficiency leads to significantly reduced HDL plasma sphingolipid levels.

We next asked whether the lack of HDL as a lipid transporter is the primary cause for the reduction in plasma sphingolipids or whether ABCA1 deficiency in Abca1 HSKO is also associated with lower cellular levels and an altered SL synthesis. To address these questions, we analyzed total SL levels in hepatic tissue of the mice and found a significant reduction in the major (C₁₆SO, C₁₈SO, S₁₈SA, and C₁₈SAdiene), but not in the minor sphingoid bases (C₁₇SO, C₁₉SO, C₂₀SO, C₂₀SA, and doxSA) (Figure 4).

To determine whether sphingolipid synthesis is altered in hepatocytes lacking ABCA1, we measured the de novo SL synthesis in cultured primary hepatocytes by the incorporation of stable isotope labelled d3N15-serine into de novo formed sphingoid bases. Myriocin (myr), a specific inhibitor of sphingolipid de novo synthesis was added as negative control. Surprisingly, the total amount of newly synthesized SLs was slightly higher in Abca1 HSKO hepatocytes (Figure 5A) compared to WT controls. A similar picture was seen when analyzing the export of the labelled SL into the media, which was also higher for SO by Abca1 HSKO hepatocytes (Figure 5B). This was specific as Myriocin completely blocked sphingolipid de novo synthesis in both wild type and Abca1 HSKO cells.

About 60% of the plasma S1P is transported by HDL and about 30% bound to albumin. The binding to HDL is mediated by apoM (Christoffersen et al., 2011). Previously, we reported reduced plasma S1P in subjects with ABCA1 mutations (Karuna et al., 2011). Thus, we analyzed plasma S1P and apoM levels in Abca1 HSKO mice. Both, S1P and apoM concentrations were significantly lower in plasma of Abca1 HSKO mice (Figures 6A–C).

We previously reported a significant reduction in plasma S1P and apoM in apoB-depleted plasma (i.e., HDL) of humans with heterozygous ABCA1 mutations (Karuna et al., 2011). In agreement with the human data, we also found significant reductions in plasma S1P and apoM levels in Abca1 HSKO mice, suggesting that hepatocyte ABCA1 is a major contributor to plasma HDL-S1P levels. Liver and hepatocytes can produce S1P (Kurano et al., 2013) but whether hepatocyte ABCA1 is directly involved in S1P transport from liver to plasma is unknown. It was suggested earlier that ABCA1 mediates S1P transport from erythrocytes (Kobayashi et al., 2009), which are the major source for plasma S1P (Pappu et al., 2007) and express significant amounts of ABCA1 (Ohkawa et al., 2020).

HDL stimulates secretion of apoM (Ahnstrom et al., 2010), which is the major carrier for plasma S1P (Christoffersen et al., 2011). Thus, plasma HDL reduction due to ABCA1 deficiency results in less plasma apoM to transport S1P. Regardless, ABCA1 regulation of apoM and S1P merits further investigation considering the atheroprotective role of apoM/S1P-containing HDL (Egom et al., 2013).

Abca1 HSKO mice had attenuated PI3 kinase activation in liver in response to insulin, but the mechanism is still unclear (Chung et al., 2010). S1P activates PI3 kinases in liver and hepatocytes (Osawa et al., 2010; Studer et al., 2012), suggesting that attenuated PI3 kinase activation in Abca1 HSKO mice may be partially due to reduced plasma and HDL-S1P (Key et al., 2017).

In conclusion, our results demonstrate that hepatocyte ABCA1 is a major contributor in maintaining normal plasma sphingolipids, including SM and S1P. However, further research is necessary to find out whether the altered SL metabolism is a direct consequence of ABCA1 deficiency or secondary to HDL deficiency caused by the lack of ABCA1.