Structure-dependent absorption of sphingolipid long-chain bases from the digestive tract into lymph


 Background:Dietary sphingolipids have various biofunctions, including skin barrier improvement, anti-inflammatory and anti-carcinoma properties. Long-chain bases (LCBs), the essential backbones of sphingolipids, are responsible for these bioactivities, and they vary structurally between species. Given these findings, however, the absorption dynamics of each LCB remain unclear.Methods:In this study, five structurally different LCBs were isolated and their absorption ratios and levels of their metabolites were analyzed using a lymph-duct-cannulated rat model by liquid chromatography tandem mass spectrometry (LC/MS/MS) with a novel multistage fragmentation method.Results:The five orally administered LCBs were absorbed and detected in lymph as free LCBs and several metabolites including ceramides, glucosylceramides, and sphingomyelins. The absorption rates of LCBs were 0.10-1.17% depending on their structure. The absorption rate of 4-trans,8-cis-sphingadienine was highest (1.17%), whereas that of 4-trans,8-trans,10-trans-sphingatrienine was lowest (0.10%). The amount of 4-trans,8-cis-sphingadienine-bound sphingomyelin in lymph was particularly higher than other four LCB-bound sphingomyelins.Conclusion:Structural differences among LCBs, particularly the geometric isomerism at the C8–C9 position, significant affect the absorption rate and amounts of metabolites. This is the first report revealed that the absorption rate and metabolism of sphingolipids are dependent on their LCB structure.

Orally ingested d18:1-bound glucosylceramide (GlcCer) and SM are degraded to ceramide and polar head groups by speci c enzymes in the digestive tract of rats, followed by the breakdown of ceramide to LCB and fatty acids [12][13][14][15][16][17]. Previous studies showed that maize-derived GlcCer and squid-derived ceramide 2-amynoethylphosphonate (CAEP), which have nonmammalian-type LCBs such as d18:2 and 9Me-d18:3, are also hydrolyzed in the digestive tract, and liberated LCBs are transferred to lymph as free LCBs and ceramides [18][19][20][21]. These ndings suggest that liberated LCBs are incorporated into small intestinal epithelial cells, reconstructed to sphingolipids from incorporated LCBs in the form of ceramide, and then absorbed into lymph in rats [18,22,23]. Since the structures of LCBs in plants, fungi, and marine organisms are different from those in mammals [1], the absorption rates of different LCBs into lymph are likely to differ depending on their structure.
Thus, to elucidate the bioactivities and mechanisms of each dietary sphingolipid, it is important to clarify the relationship between their individual biokinetics and structure. Using conventional MS analysis methods, one can analyze the LCB backbones of ceramides and GlcCer [24]. In addition, a novel analytical method for LCB backbone identi cation of SM was developed. To explore the relationship between LCB structures and absorption rate, as well as their metabolites, the amount of LCBs and their metabolites in lymph from rats administered structurally different pure LCBs were analyzed.

Lymph collection from rats administrated LCBs
Male Wistar/ST rats (Japan SLC, Inc., Shizuoka, Japan) aged 9 weeks were housed in individual stainless-steel cages and given a semipuri ed diet (AIN 93G formula) for a 5 day acclimation period. After overnight fasting, a vinyl catheter and a silicone catheter were implanted into the thoracic lymph duct and the duodenum, respectively, as described previously [28]. Test lipids (100 g/L) prepared in 1 mL of emulsi ed solution contained 100 mg of triolein (control group) or 90 mg of triolein plus 10 mg of each LCB (LCB-administered groups). Test lipids were emulsi ed with 10 mg of sodium taurocholate using a sonicator. After lymph collection for 30 min (initial lymph) on day 1 postoperation, the rats were infused with 1 mL of an emulsi ed test solution for 1 min, and the infusion of glucose-NaCl solution through the duodenal tube was continued at 1.8 mL/h until the end of the experiment. Lymph was collected in a test tube at 1 h intervals for 8 h following the administration of the test solution. The collected lymph was immediately frozen and stored at −80°C until subsequent analyses. Animal experiments were conducted in accordance with the guidelines of the Animal Committee of Iwate University (Authorization No. A201450).

Lipid extraction
Lipids were extracted from recovered lymph samples as described previously [6] with slight modi cation. The collected lymph samples (0.5 mL) were transferred to glass-capped test tubes and mixed with chloroform (1 mL), methanol (2 mL), PBS (0.3 mL), and 1 nmol of internal standards (C17 ceramide, C12 glucosylceramide, d17:1, C17 sphingomyelin). After incubation at 37°C for 2 h, chloroform (1 mL) and water (1 mL) were added and the mixture was centrifuged at 1000´g for 5 min, and lower organic phases were collected into fresh test tubes. Chloroform (1 mL) was added to the upper phase and residual lipids were re-extracted. After centrifugation, the resulting lower organic phases were combined and dried with nitrogen gas. 0.4 M NaOH-methanol solution (1 mL) was added to the extracted lipids and incubated at 37°C for 2 h to saponify glycerolipids. Chloroform (1 mL) and water (1 mL) were added, and lipids were collected from lower organic phases and evaporated with nitrogen gas. Dried residues were treated with cold acetone (1 mL) and centrifuged at 1000´g at 4°C for 5 min. Supernatants were discarded, and the pellet was analyzed by LC-MS/MS.

Cultures of Caco-2 cells and LCB treatment
Caco-2 cells were cultured in DMEM supplemented with 10% FBS, NEAA, and P/S (100 U/mL penicillin, 0.1 mg/mL streptomycin) at 37°C in a humidi ed 5% CO 2 incubator. Caco-2 cells were seeded into sixwell cell culture inserts at a density of 1.5×10 5 cells/insert. To generate differentiated Caco-2 cells for use as a small intestinal epithelial cell model, the Caco-2 cells were cultured for another 21 days after reaching con uency. The medium was changed with a fresh one every other day. For LCB treatment, 10 mM d18:2 4t8c or d18:2 4t8t dissolved in serum-free DMEM was added to the cell culture insert and incubated for 24 h. The nal DMSO concentration was adjusted to 0.1%. The basal-side medium and cells were collected, and lipids were extracted as described above.

LC-MS/MS analysis of lymph lipids
Lymph lipids were analyzed with a TripleTOF 5600 LC-MS/MS system (AB SCIEX, Foster City, California, USA) equipped with an InertSustain NH2 column (diameter, 2.1 mm; length, 100 mm; particle size, 5 mm; GL Science) in the ESI positive mode as described previously with slight modi cation [6,24]. Sphingolipids extracted from lymph were dissolved in 200 mL of mobile phase A, and a 10 mL sample solution was used for sphingolipid analysis. Mobile phase A was acetonitrile:methanol:formic acid:1 M ammonium formate (95:5:0. ion source gas 1, 20 psi; ion source gas 2, 50 psi; curtain gas, 20 psi; TOF accumulation time, 0.2 s; product accumulation time, 0.05 s. To generate product ions, the collision energy and collision energy spread were set at 35 and 5 V, respectively. Data acquisition and analysis were respectively performed using Analyst TF 1.7.1 software and MultiQuant 3.0.1 software (AB SCIEX). The amounts of target analytes were calculated using individual internal standards.

SM analysis by in-source collision-induced dissociation/multiple reaction monitoring (CID/MRM)
To identify the LCB backbones of SM, the parameters for MS were as follows: negative ion mode for TOF scans; ion spray voltage oating, -4500 V; temperature, 300°C; declustering potential, -300 V; collision energy, -10 V; ion source gas 1, 20 psi; ion source gas 2, 50 psi; curtain gas, 20 psi; accumulation time, 0.2 s. The detected in-source fragmented [M-CH 3 ] -SM ions serving as precursor ions and [M-CH 3 -fatty acid]ions providing LCB backbone information were generated by collision-induced dissociation (CID). The collision energy and collision energy spread were set at -50 and 5 V, respectively. LC conditions were the same as those described above for the positive ion mode. LCB species for SM were identi ed from their typical product ions (d18:1-bound SM for m/z 449.31; d18:2-bound SM for m/z 447.30; d18:3-bound SM for m/z 445.28; 9Me-d18:2-bound SM for m/z 459.30; 9Me-d18:3-bound SM for m/z 457.28).

Statistical analysis
Statistical signi cance was determined using Tukey's method and p < 0.05 was considered to indicate signi cant difference. Data are presented as mean ± standard deviation.

Recovery of free LCBs from lymph
To investigate the absorption of LCBs from the digestive tract to lymph, emulsions with 10 mg of puri ed LCBs were enterally administrated to a thoracic-duct-cannulated rat. Lymph was collected every hour up to 8 h after LCB administration. There were no signi cant differences in the amount of lymph output among rats administrated each LCB (Fig. S2), indicating that surgery and animal maintenance were appropriately carried out. First, free LCBs extracted from the collected lymph were analyzed by LC/MS/MS. HPLC retention times and exact masses of protonated ion signals ([M+H] + m/z 298.27, 298.27, 296.26, 312.29, and 310.27 for d18:2 4t8c , d18:2 4t8t , d18:3, 9Me-d18:2, and 9Me-d18:3, respectively) were used for the identi cation of each LCB species. Peak areas were integrated at m/z ± 0.05, and LCB amounts were determined by comparing [M+H] + ion signals of d18:2 4t8c , d18:2 4t8t , d18:3, 9Me-d18:2, and 9Me-d18:3 with the peak area of the d17:1 internal standard. The amounts of free LCBs in lymph at each timepoint are shown in Fig. 2. The amount of all LCBs, d18:2 4t8c , d18:2 4t8t , d18:3, 9Me-d18:2, and 9Me-d18:3, were elevated in the lymph of rats infused with the corresponding LCBs. The amount of d18:2 4t8c was highest at 3 h after administration (0.83 nmol). On the other hand, the amount of d18:2 4t8t , a geometrical isomer of d18:2 4t8c , was highest at 1 h after administration (4.38 nmol). The levels of both d18:2 4t8c and d18:2 4t8t at 8 h decreased to levels similar to those before administration.
The total amount of d18:2 4t8c up to 8 h after LCB administration was ~4.4-fold higher than that of d18:2 4t8t . In the case of other LCBs, the amount of d18:3 in lymph increased after 1-3 h, 9Me-d18:2 increased after 2-4 h, and 9Me-d18:3 peaked at 4 h after administration. This result suggests that the absorption rates of free LCBs into lymph differ with their structure, even among geometrical isomers.

Quanti cation of reconstituted ceramide and GlcCer in lymph
Since the LCBs absorbed from the digestive tract were predicted to be metabolized to ceramide and complex sphingolipids in intestinal epithelial cells, these LCB metabolites in lymph were also analyzed. As expected, a portion of each administrated LCB was processed into ceramide ( Fig. 3A-D), and more than 80% was bound to the fatty acid C16:0, followed by C24:0 and C23:0 (Fig. S3), and the bound fatty acid species did not depend on the structure of the administrated LCB (Fig. S3). Similar to the results of free LCB absorption, the amount of d18:2 4t8t ceramide in lymph peaked at 1 h after administration and then gradually decreased to only trace amounts at 8 h (Fig. 3A). The rates of d18:2-bound ceramide absorption into the lymph of groups administrated d18:2 4t8c and d18:2 4t8t were compared. As in the case of free LCBs, ceramides containing d18:2 4t8c were absorbed more slowly and to a lesser extent than d18:2 4t8t ceramides (Fig. 3A). In the case of other LCBs, the amount of administrated LCB-bound ceramides in the lymph increased at 2-4 h after administration of d18:3, at 2-6 h after administration of 9Me-d18:2, and at 3-7 h after administration of 9Me-d18:3 ( Fig. 3B-D). The absorption of 9Me-d18:3bound ceramides were the slowest among the LCBs tested in this study, and the maximum amount of ceramides detected in lymph recovered at 6 h after LCB administration (Fig. 3D). The total absorption rates of d18:2 4t8c , d18:2 4t8t , d18:3, 9Me-d18:2, and 9Me-d18:3 into lymph as ceramide were 0.095% ± 0.008%, 0.143% ± 0.023%, 0.030% ± 0.008%, 0.054% ± 0.019%, and 0.088% ± 0.023%, respectively (Fig.  6).
GlcCers metabolized from the administrated LCB in the collected lymph were identi ed. Interestingly, d18:2 4t8t and 9Me-d18:2 4t8t were the only LCBs detected in lymph that were converted to GlcCer (Fig. 4A,  C), which was unlike in the case of free LCBs or ceramide. GlcCer of other LCBs was present only at trace levels ( Fig. 4B, D). Only C16:0-bound GlcCers of d18:2 4t8t and 9Me-d18:2 4t8t were detected, and no other fatty acids such as C24:0 and C23:0 were detected in lymph. Changes in the amount of GlcCer in lymph showed a similar trend. The amount of reconstructed GlcCer in lymph was highest at 3 h after administration of d18:2 4t8t , and at 4 h after administration of 9Me-d18:2 4t8t (Fig. 4A, C). Both d18:2 4t8t and 9Me-d18:2 4t8t bound to GlcCer were decreased to trace levels at 6 h after administration (Fig. 4A, C). In addition, the absorption rates of GlcCers as the LCB were lower than that of ceramides ( Fig. 3A-D, 4A, C). The biosynthesis rate of GlcCer is estimated to be lower than the ceramide synthesis rate because it requires an additional glycosylation step. Additionally, it is possible that the transport of reconstituted GlcCer from the small intestine to lymph may be slower than that of ceramide. The absorption rates of d18:2 4t8t and 9Me-d18:2 into lymph as GlcCer were 0.011% ± 0.004% and 0.017% ± 0.005%, respectively (Fig. 6).
Uptake and transport of d18:2 geometrical isomers in differentiated Caco-2 cells In rat lymph, the levels of reconstituted GlcCers and SMs clearly varied with the geometric isomerism of the 8-position of d18:2 (Fig. 4A, 5A). Thus, the absorption behaviors of d18:2 4t8c and d18:2 4t8t were analyzed using differentiated Caco-2 cells, an intestinal epithelial transport system model. LCBs (10 mM) were added to the apical side of Caco-2 cells, which were cultured on a cell culture insert for 21 days and differentiated into intestinal epithelium-like cells, and incubated for 24 h. Lipids were extracted from the medium on the basolateral side and from cells. Free LCBs, reconstituted ceramide, GlcCer, and SM were analyzed by LC/MS/MS (Fig. 7A-D). There were no signi cant differences in the amounts of free d18:2and d18:2-bound ceramides in the medium on the basolateral side between treatments with d18:2 4t8c and d18:2 4t8t (Fig. 7A, B). Levels of d18:2-GlcCer were higher in d18:2 4t8t -treated cells, and the amounts of d18:2-SMs in d18:2 4t8c -treated cells were larger than that in d18:2 4t8t -treated cells, which consistent with the rat lymph results. Quantitative analysis revealed that the amount of intracellular free d18:2 in the d18:2 4t8c -treated cells were signi cantly higher than that in the control cells, but there was no signi cant difference between the d18:2 4t8t -treated cells and the control cells (Fig. 7E). The amount of intracellular d18:2-bound ceramides signi cantly increased only in the d18:2 4t8t -treated cells compared with that in the control cells (Fig. 7F), but there was no signi cant difference between the d18:2 4t8c -treated cells and the control cells. The amount of intracellular d18:2-bound GlcCers signi cantly increased in the d18:2 4t8ttreated cells compared with that n the control cells, but there was no change in the amount of d18:2bound GlcCers in the d18:2 4t8c -treated cells compared with that in the control cells (Fig. 7G). These results followed a similar trend to those of the experiment on LCB absorption in rat lymph. However, unlike in the rat experiment, the amount of intracellular d18:2-bound SMs of both d18:2 4t8c -and d18:2 4t8ttreated cells signi cantly increased compared with that in the control cells, and there were no signi cant differences in the amount of intracellular d18:2-bound SMs between d18:2 4t8c -and d18:2 4t8t -treated cells (Fig. 7H).

Discussion
To explore the mechanism underlying the biological effects of dietary sphingolipids, in this study, the absorption and metabolism of nonmammalian-type LCBs from the digestive tract to lymph in rats were evaluated. It has been reported that ~ 0.3% of nonmammalian-type sphingolipids are absorbed into the lymph in rats as free LCBs or ceramides, and the amounts of free LCBs in lymph are ~ 20 times smaller than those of other sphingolipid classes such as ceramide [19,21]. In addition, the rate of absorption of sphingosine, a major mammalian LCB, into the lymph is estimated to be three times higher than that of d18:2, a plant-derived LCB [19]. In this study, to exclude the e ciency of the degradation of sphingolipids in the gastrointestinal tract, the absorption of puri ed free LCBs in rats cannulated with thoracic lymph was examined, and free LCBs and reconstituted sphingolipids in lymph were analyzed by LC/MS/MS. The amount of free d18:2 4t8t in lymph, which has a double bond at C8-C9 in this trans form of d18:2, clearly increased at 1 h after oral administration of d18:2 4t8t , and the level was higher than that for d18:2 4t8c , which has a double bond at the 8-position in this cis form ( Fig. 2A). On the other hand, d18:3, 9Me-d18:2, and 9Me-d18:3, which have the same C8-C9 double bond as d18:2 4t8t in the trans form, appear to be absorbed more slowly in the lymph than d18:2 4t8t (Fig. 2). These ndings suggest that the absorption rate of LCBs varied depending on the geometrical isomerism of C-C double bonds and the presence of the C9 methyl group. The uptake rate of LCBs into the epithelial cells may depend on the structure of the LCBs.
Previous studies showed that about 0.2−0.3% of GlcCers from maize, whose major LCB is d18:2 4t8c , was absorbed into the lymph of rats [19]. The present analysis showed that the rate of absorption of d18:2 4t8c from the gastrointestinal tract into lymph was ~ 1.2% (Fig. 6). Previous studies showed that it is also necessary to consider the e ciency of degradation by sphingolipid digestive enzymes in the intestine [18,19]. The absorption rate calculated in the present work was higher than that in a previous report because puri ed LCBs were used to the experiments [19].
Orally administrated sphingosine is converted to ceramide and incorporated into lymph [22]. Similarly to sphingosine, d18:2, d18:3, 9Me-d18:2, and 9Me-d18:3 are believed to be mainly converted into ceramide after absorption from the gastrointestinal tract and then transferred to the lymph. d18:2-, d18:3-, 9Me-d18:2-, and 9Me-d18:3-bound ceramides were also analyzed by LC/MS/MS. As expected, all of the LCBs in this study were detected as ceramides in the rat lymph (Fig. 3). The relationship between the absorption rate and the LCB structure of ceramides displayed the same tendency as that of free LCBs.
Since the absorption rates of free LCBs and ceramides were similar among LCBs, it is speculated that the conversion of incorporated LCBs to ceramides occurs rapidly, and all LCBs are likely to be substrates of ceramide synthase (CerS). In each LCB-administrated group, C16:0-bound ceramides were the predominant species in lymph, which are nonmammalian-type LCBs (Fig. S3). In addition, most GlcCer and SM molecular species in lymph were also C16:0. These ndings suggest that following administration, LCBs were metabolized to ceramides, and these re-synthesized ceramides were used for GlcCer and SM synthesis in intestinal cells. CerS has six isozymes with different substrate speci cities; CerS6 has high substrate speci city for C16:0-CoA and is the major CerS in the mouse intestine [29], and it is also believed to be the major CerS family member in the rat small intestine.
When SM was analyzed in the ESI positive mode, only one fragmentation product ion (m/z 184.07) derived from a phosphocholine moiety was detected (Fig. S4A). SM compounds with the same molecular weight, such as d18:1/C24:1 and d18:2/C24:0, were not separated by normal phase HPLC. To determine the LCB backbones of SMs, an ion trap mass spectrometer (IT-MS) in the ESI negative mode was employed [30]. Alternatively, the product ion of ceramide generated from SM by in-source fragmentation can be decomposed by post fragmentation using an atmospheric pressure chemical ionization (APCI) probe to obtain information on LCBs [31]. Addition of alkali metals to the HPLC mobile phase can also yield product ions derived from LCBs by CID of alkali metal adduct ions of SM (Fig. S4B) [32]. To analyze the LCB backbone of SM using a conventional ESI-LC/MS/MS system, a method using a combination of in-source fragmentation and post-source fragmentation in the ESI negative mode was established. Since [SM-CH 3 ] − ions can be observed by IT-MS [30], these ions were selected as precursor ions. The CID fragment ion of [SM-CH 3 ] − was [SM-CH 3 -fatty acid] − , which provides information on the LCB structure of SM. This method (in-source CID/MRM) is useful for quantifying the molecular species of SM; hence, the SM species in lymph and cultured cells were analyzed by in-source CID/MRM (Fig. 5, 7, S5). Comparison of the absorption behavior of LCB metabolites revealed that reconstituted GlcCer and SM were absorbed slower than ceramide (Fig. 3−5). Ceramide is de novo synthesized in the endoplasmic reticulum (ER) and transported to the Golgi by the ceramide transport protein (CERT), and transported ceramide is further metabolized to SM in the Golgi by sphingomyelin synthase (SMS) [33]. GlcCer synthesis requires the transfer of glucose from UDP-glucose to ceramide [34]. Therefore, it was suggested that differences in the absorption time between GlcCer/SM and free LCB/ceramide into lymph could be caused by enzymatic reactions. However, it is also possible that differences in absorption time are due to different mechanisms of transport to lymph.
Compared with the amount of d18:2-GlcCer in Caco-2 cells, the amount in d18:2 4t8t -treated cells was 1.37-fold higher than that in d18:2 4t8c -treated cells (Fig. 7G). Since d18:2 4t8t is more easily metabolized to d18:2-GlcCer than d18:2 4t8c , it is presumed that d18:2-GlcCer accumulated more readily in Caco-2 cells when d18:2 4t8t was added than when d18:2 4t8c was added. In the case of d18:2 4t8t treatment, which resulted in signi cantly higher accumulation of GlcCer in Caco-2 cells, a larger amount of GlcCer was released into the medium than in the case of d18:2 4t8c treatment. On the other hand, d18:2-SM levels were signi cantly higher only when d18:2 4t8c was added, than in the control group (Fig. 7D). Although the amount of d18:2-SM in Caco-2 cells was similar for cells treated with both d18:2 4t8c and d18:2 4t8t (Fig. 7H), the amount of d18:2-SMs released into the medium were greater for the d18:2 4t8c -treated group than for the d18:2 4t8t -treated group (Fig. 7D). There was no signi cant difference in the amount of d18:2-SM between the d18:2 4t8t -treated group and the control group (Fig. 7D). Therefore, it is assumed that d18:2 4t8c -SMs were more easily transferred to the medium than d18:2 4t8t -SMs. The migration of SMs from cells to the basolateral medium may be related to the structure of the LCB portion of SM. The amounts of intracellular d18:2-GlcCer and d18:2-SM were increased by d18:2 4t8c or d18:2 4t8t treatment, which indicated that the level of d18:2-SM was over 100 times greater than that of d18:2-GlcCer (Fig. 7G,  H). The migration behaviors of d18:2-GlcCer and d18:2-SM from Caco-2 cells into the medium were similar to the incorporation behaviors of d18:2-GlcCer and d18:2-SM into the lymph (Fig. 4A, 5A). In rat lymphatic cannulation experiments, although d18:2 was more easily metabolized to d18:2-GlcCer in the 8-trans form than in the 8-cis form, d18:2 was predominantly metabolized to d18:2-SM. Additionally, d18:2-SM with the 8-cis form of d18:2 more easily migrated to the lymph than d18:2-SM with the 8-trans form of d18:2, resulting in a greater amount of d18:2 4t8c being absorbed into the lymph uid than d18:2 4t8t . The double bond at C8-C9 of d18:2 4t8t is a trans form, which makes this LCB molecule linear, whereas the double bond at C8-C9 of d18:2 4t8c is a cis form, which gives this molecule a bent form. Therefore, it can be concluded that the geometric isomer form of LCBs is a critical factor for their absorption and metabolism.
Ethics approval and consent to participate Animal experiment was approved by the Animal Committee of Iwate University.

Consent for publication
Not applicable.

Availability of data and materials
The dataset supporting the conclusions of this article is available upon request for corresponding author after publication.

Competing interests
The authors declare no competing interests.   Values are mean ± SD (n = 4 5).

Figure 6
Calculated recovery rate and sphingolipid class ratio for each long-chain base. Values for the control group were subtracted.