In situ glucosylceramide synthesis and its pharmacological inhibition analysed in cells by 13C5‐sphingosine precursor feeding and mass spectrometry

Glycosphingolipids (GSLs) fulfil diverse functions in cells. Abnormalities in their metabolism are associated with specific pathologies and, consequently, the pharmacological modulation of GSLs is considered a therapeutic avenue. The accurate measurement of in situ metabolism of GSLs and the modulatory impact of drugs is warranted. Employing synthesised sphingosine and sphinganine containing 13C atoms, we developed a method to monitor the de novo synthesis of glucosylceramide, the precursor of complex GSLs, by the enzyme glucosylceramide synthase (GCS). We show that feeding cells with isotope‐labelled precursor combined with liquid chromatography–mass spectrometry (MS)/MS analysis allows accurate determination of the IC50 values of therapeutically considered inhibitors (iminosugars and ceramide mimics) of GCS in cultured cells. Acquired data were comparable to those obtained with an earlier method using artificial fluorescently labelled ceramide to feed cells.

Sphingolipids are structural components of the mammalian cell membrane. Glycosphingolipids (GSLs) are particularly present in the outer leaflet of the cell membrane and are unevenly distributed, generating lipid rafts (semi-ordered microdomains) together with cholesterol molecules [1]. Within these lipid rafts reside specific proteins involved in the interaction of cells with a variety of agents ranging from hormones to pathogens. The GSL composition of lipid rafts has been observed in various processes, a topic recently reviewed [2]. The metabolism of GSLs is complex [3][4][5][6]. The initial step in the formation of the majority of GSLs is the synthesis of glucosylceramide (GlcCer) by the enzyme glucosylceramide synthase (GCS) [7]. Using UDPglucose as a sugar donor, GCS transfers a glucose moiety to cytosolic ceramide to generate GlcCer. The ceramide is generated in two ways. Firstly, it may be derived from newly formed sphinganine generated from serine and palmitoyl-CoA [4]. Secondly, cytosolic ceramide is provided by the so-called salvage pathway in which sphingosine is formed by lysosomal degradation of ceramide, and upon release into the cytosol is employed for re-synthesis of ceramide [8,9]. Thus, as shown in Fig. 1A, both sphinganine (de novo synthesis) and sphingosine (salvage pathway) can act as precursors in the formation of ceramide and subsequent GlcCer.
To identify effective GCS inhibitors determination of the half-maximal inhibitory concentration (IC 50 ) values for GCS in intact cells is warranted. This is currently assessed by feeding cells, in the absence and presence of potential inhibitors, with N-[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl]-D-erythrosphingosine (NBD-ceramide) followed by detection of glucosylated NBD-ceramide by high-performance liquid chromatography [30]. This method is laborious and has intrinsic drawbacks. The presence of a fluorophore moiety in NBD-ceramide significantly alters the molecule and might alter metabolism compared to natural ceramide. We envisioned an alternative procedure: the use of natural, but 13 C isotope-encoded sphinganine or sphingosine to monitor GSL metabolism, in particular, the de novo formation of GlcCer by GCS. We previously reported the synthesis of 13 C-encoded lipids, including sphinganine and sphingosine [31][32][33]. Using natural precursors to assess in situ sphingolipid metabolism should render physiological data. Previous research has shown that sphingosine and sphinganine added to the cell medium are easily taken up by cells and enter sphingolipid metabolism [34]. Here, we demonstrate the use of 13 C-isotope encoded precursor lipids ( 13 C 5 -sphingosine and 13 C 5 -sphinganine) to study in situ GCS activity in living cells.

Cell culture
Murine fibroblasts NIH/3T3 were cultured in DMEM high glucose supplemented with 10% (v/v) FBS, 1% (w/v) GlutaMAX TM , and 1% (w/v) P/S. Primary human dermal fibroblasts were cultured in DMEM/F12 growth medium, supplemented with 10% (v/v) FBS and 1% (w/v) P/S. RAW 264.7 were cultured in RPMI growth medium, supplemented with 10% (v/v) FBS, 1% (w/v) Gluta-MAX TM and 1% (w/v) P/S; an additional 25 mM HEPES was added during incubations. All cell lines were incubated at 37°C and 5% CO 2 . Cells were seeded during passage and the incubations were started when cells were 80-100% confluent. In assays using CBE or the GBA2 inhibitor shown in Fig. 1, cells were pre-incubated with 300 lM CBE or 0.2 lM GBA2 inhibitor per well for 1 h prior to 13 C 5 -lipid incubation. Cells were incubated with 0.5 nmol 13 C 5 -sphingosine or 13 C 5 -sphinganine per well. After optimisation with 13 C 5 -sphingosine, where incubation times of 0-24 h were used, incubation times of 1 h for RAW cells and 3 h for fibroblasts were used for inhibitor assessment. After incubation, cells were washed five times with PBS and frozen in 150 lL H 2 O per well until lipid extraction.
All cell experiments were performed at least twice, with samples in duplicate or triplicate.
Zebrafish AB/TL zebrafish were kept in accordance with the directives of the local animal welfare committee (Instantie voor Dierwelzijn, IvD, Leiden, the Netherlands) and guidelines specified by the EU animal Protection Directive 2010/63/ EU. Embryos, derived from natural spawning, were kept in egg water (60 mgÁL À1 sea salt, 0.01 mgÁL À1 methylene blue) supplemented with 0.5% DMSO and 20 pmol 13 C 5sphingosine per fish for up to 5 days post-fertilisation (before the free-feeding stage).

Lipid extraction
Lipid extraction was performed through modified Bligh and Dyer extraction [37]. In short, after the addition of 500 pmol of both C 17 -sphinganine and C 17 -dihydroceramide as internal standard, lipids were separated into the lysolipids (upper phase) and neutral GSLs (lower phase), and calibration curve lipids were added to the appropriate samples. The lower phase lipids were deacylated through microwaveassisted hydrolysis [38] to yield their lysolipid counterparts. Samples were further cleaned through butanol extraction and, after drying, reconstituted in MS grade MeOH.
LC-MS/MS measurements were performed as described previously [33]. Compounds were separated using a BEH C18 column followed by detection on a Waters TQS micro with electrospray ionisation in positive mode (Waters, Milford, MA, USA). The isotope labelled lipids were identified through their distinct parent and daughter ions as follows: 13  Peak areas were determined using TargetLynx (Waters, Milford). After peak integration, results for both optimisation and determination of inhibitor IC 50 values were processed using GRAPHPAD PRISM version 8 (GraphPad Software Inc., San Diego, SA, USA). IC 50 values were estimated using the non-linear regression log(inhibitor) versus normalised response curve fit.

Results
Employing isotope labelled lipid feeding to monitor in situ glucosylceramide formation in cultured cells We first comparatively studied the suitability of 13 C 5sphinganine and 13 C 5 -sphingosine feeding to monitor GCS-driven GlcCer formation. Sphinganine feeding allows the monitoring of the de novo synthesis pathway, whilst sphingosine feeding allows the monitoring of the salvage pathway. The feeding of 13 C 5 -sphinganine led to the formation of 13 C 5 -dihydroceramide which was converted to 13 C 5 -ceramide by dihydroceramide desaturase. In contrast, fed 13 C 5 -sphingosine was directly converted to 13 C 5 -ceramide by ceramide synthase. At longer incubation time 13 C 5 -encoded GlcCer, lactosylceramide (LacCer), and globotriaosylceramide (Gb3) were increasingly detected in cells fed either with 13 C 5sphinganine or 13 C 5 -sphingosine. Examples of the resulting LC-MS/MS chromatograms can be found in Fig. S1. A direct comparison of feeding with both precursors can be found in Fig. S2. Since feeding with 13 C 5 -sphingosine led to the comparable generation of GSLs as feeding with 13 C 5 -sphinganine, in subsequent experiments we made use of 13 C 5 -sphingosine feeding.
To determine the optimal concentration of feeding, we incubated fibroblasts with a range of concentrations of 13 C 5 -sphingosine for 4 h. With increasing concentrations of fed 13 C 5 -sphingosine, as shown in Fig. 2, the intracellular labelled sphingosine and ceramide also increased. Labelled GlcCer and Gb3 both showed the most incorporation at 500 pmol 13 C 5sphingosine feeding. At 250 pmol or lower GlcCer and Gb3 are difficult to accurately quantify. Therefore, subsequent experiments were performed with 500 pmol, or 1 lM, 13 C 5 -sphingosine feeding per well.
To examine more closely the intracellular flux of the 13 C 5 -lipids we incubated the fibroblasts with 0.5 nmol 13 C 5 -sphingosine for a wider range of time points (Fig. 3). Cellular 13 C 5 -sphingosine was maximal after 1 h of incubation (Fig. 3A), and generation of 13 C 5ceramide occurred readily (Fig. 3B). The concentrations of 13 C 5 -ceramide reached a peak after 2 h of incubation. Figure 3C shows the formation of well-measurable quantities of 13 C 5 -GlcCer from 13 C 5 -ceramide. Concentrations of 13 C 5 -GlcCer increased over time of incubation. Further metabolism to 13 C 5 -LacCer and 13 C 5 -Gb3 was observed (Figs 3D,E, respectively). Interestingly, the concentration of 13 C 5 -Gb3 was higher than that of 13 C 5 -LacCer. The same trend is also observed in endogenous lipids. This observation is in accordance with earlier research showing higher concentrations of Gb3 than LacCer in cultured human fibroblasts [39]. After overnight incubation, intra-cellular 13 C 5sphingosine, 13 C 5 -ceramide, and 13 C 5 -GlcCer are each in equilibrium. We expect that at longer incubation times 13 C 5 -LacCer and 13 C 5 -Gb3 will also reach equilibrium. The purpose of the present investigation was the in situ measurement of de novo GlcCer formation by GCS. We therefore used a 13 C 5 -sphingosine incubation of cells for 3 h which allows optimal detection of the enzyme activity.

Impact of catabolic enzymes on inhibition of GCS
In the past, a variety of cell lines from various species have been used to determine in situ IC 50 values (concentration resulting in a 50% reduction of activity). To assess the importance of cell type, we compared the effect of Eliglustat in murine RAW 264.7 cells, NHD fibroblasts, and the murine fibroblast cell line NIH/3T3. When measuring GCS activity by monitoring the formation of 13 C 5 -GlcCer, interference by other enzymes metabolising GlcCer needs to be ruled out or prevented. Such interfering enzymes are the lysosomal GBA1 and the cytosol-facing GBA2. For GBA1 and GBA2 specific cell-permeable inhibitors are available, respectively CBE and N-[5-(3,3-dimethyl-2-butoxy)pentyl]-1-deoxynojirimycin (GBA2 inhibitor; Fig. 1). It is known that fibroblasts hardly express GBA2 and therefore measurement of GCS activity only requires inhibition of possibly interfering GBA1. We incubated cells with or without the various inhibitors prior to feeding with 13 C 5 -sphingosine in the presence of a variable concentration of the GCS inhibitor Eliglustat. The presence of the GBA1 inhibitor CBE, and that of the GBA2 inhibitor in the case of RAW 264.7, did not significantly impact on the observed 13 C 5 -GlcCer formation (see Table 1). Of note, quite similar results were obtained with the various cells tested, suggesting broad applicability. Furthermore, at the used experimental conditions (3 h of 500 pmol 13 C 5 -sphingosine feeding) when using fibroblasts or RAW 264.7 cells, the presence of GBA1 and GBA2 inhibitors is not required and therefore their addition was therefore omitted in subsequent experiments with these cells.
Inhibitor assessment using NBD-ceramide and 13 C 5 -sphingosine feeding of cultured cells Previously GCS activity in intact cells was studied by the feeding of NBD-ceramide and analysis of its conversion to NBD-GlcCer. This NBD-ceramide substrate assay has been applied in combination with GCS inhibitors and thus has generated IC 50 values for these compounds. We determined the IC 50 values of several inhibitors of GCS using the established 13 C 5sphingosine method and compared these to values previously obtained using NBD-labelled ceramide as substrate. As a comparison, we examined established GCS inhibitors: Eliglustat, Miglustat, Ibiglustat, and the iminosugar AMP-DNM ( Table 2).

Application of the assay of in situ GCS activity in intact zebrafish larvae
Genetically modified zebrafish with impaired GSL metabolism have become available in recent years. For example, fish lacking GBA1 have been developed allowing investigations on various aspects of GD [40,41]. Measurement of the metabolic flux of GSLs in full organisms, in addition to the steady-state lipid levels, is of heightened interest. We therefore tested the applicability of 13 C 5 -sphingosine feeding, through addition to the egg water, of zebrafish larvae (up to 5 days post-fertilisation, dpf) to measure the in situ metabolism of GSL, in particular the formation of GlcCer. Individual fish were fed at 8 h postfertilisation with 20 pmol 13 C 5 -sphingosine in 100 lL egg water. Three pools of five fish were harvested at daily intervals, lipids were extracted and analysed by LC-MS/MS. As seen in Fig. 4, the 13 C 5 -sphingosine is readily taken up by the larvae and metabolised to both ceramide and GlcCer by 5 dpf. Thus, the method is applicable in the fish model and could be used to test the efficacy of drug interventions aiming to reduce GlcCer biosynthesis.

Discussion
When using isotope labelled lipids, both the de novo and the salvage pathway of ceramide synthesis can be monitored. Feeding of both 13 C 5 -sphingosine and 13 C 5 -sphinganine allowed us to track the flux of lipids through the metabolic pathways, and 13 C 5 -sphingosine was chosen as the feeding substrate for further work. The 13 C 5 -sphingosine is taken up well by the fibroblasts. Levels of 13 C 5 -sphingosine increase to a peak after 1 h of incubation. 13 C 5 -ceramide shows a peak after 2 h of incubation and, following that, the glycosylated forms increase. After overnight incubation, levels of 13 C 5 -sphingosine, -ceramide, and -GlcCer seem to reach an equilibrium. A further increase in incubation times will likely lead to an equilibrium in the further metabolites as well, in addition to allowing the analysis of the half-life time of the labelled lipids in the fibroblasts.
We applied the 13 C 5 -sphingosine feeding to evaluate the inhibitory properties of several known GCS inhibitors. To do so, we first examined the influence of catabolic enzymes on the assay results by employing inhibitors of both GBA1 and GBA2. The presence of active GBA1 appeared to have no impact on the found values, for which the short incubation times are likely responsible. For GBA1 to have an effect, the newly synthesised glycolipids need to reach the lysosome. In the short incubation times used here, it is unlikely that the 13 C 5 glycolipids have already been reabsorbed, and an extension of the incubation times will probably show an influence of a GBA1 inhibitor on the results. A far more likely candidate for interference in our values is GBA2, as this is found on the cytosolic leaflet of the plasma membrane. While fibroblasts contain almost no GBA2, RAW 264.7 cells have a much higher expression. Therefore, we studied the effect of the GBA2 inhibitor on apparent IC 50 values for GCS in RAW 264.7 cells. This analysis revealed that the apparent IC 50 values of GCS inhibitors are similar with or without GBA2 inhibitors. Thus, under the assay conditions used there is no influence of both GBA1 and GBA2 on the assessed IC 50 values of compounds for GCS. We compared the IC 50 values found using the isotope method to those earlier obtained using NBDceramide as substrate. Though the two methods are fundamentally distinct, the established IC 50 values are pretty similar. Besides the method of lipid measurement as such, the two protocols also differ in other aspects. Firstly, we here use a natural sphingosine precursor allowing the swift formation of ceramide in vivo. Secondly, uptake of NBD-ceramide by cells from the medium has a limited capacity and the unnatural chemical structure of NBD-ceramide might impact on its cellular fate.
The novel method appears particularly suited to studying at organismal level GSL metabolism in zebrafish. Pilot experiments have shown that the 13 C 5sphingosine is taken up and tolerated by fish larvae up to 5 days post-fertilisation. Delivery through both feeding in the egg water and injection into the yolk (data not shown) resulted in the incorporation of 13 C 5sphingosine into the backbone of higher GSLs.
Another elegant approach to studying in situ GSLs metabolism has been developed by Hannun and coworkers [42] They make use of odd-chain lipids to distinguish newly synthesised molecules derived from those. It was demonstrated that the de novo GSL metabolism can be examined through feeding cells with C 17 sphinganine. As C 17 sphinganine does not naturally occur, the altered length of the chain allows the distinguishing of newly synthesised (glyco)sphingolipids by means of mass spectrometry. While both Snider et al. [42] and our study examined the shortterm incorporation of labelled lipids in the GSL metabolism, with longer incubation times, other steps in GSL metabolism downstream from GCS can a priori also be monitored. Such future studies can elegantly be assisted by the availability of specific inhibitors of lysosomal glycosidases involved in the degradation of GSLs, for example, GBA1 [43,44].
In conclusion, isotope labelled lipids can be exploited to study in situ GSL metabolism in intact cells. The 13 C 5 -sphingosine is readily taken up by various cell types, and the generation of GSL metabolites can be detected after 3 h. This can be applied for the determination of the IC 50 of various inhibitors of GCS. Our novel method rendered similar IC 50 values of SRT drugs as earlier established with NBDceramide as the substrate for GCS. The use of isotope labelled precursors has the potential to find broader applications than that presented in this study. Downstream pathways from GCS could be analysed, and fluxes of GSLs could be measured, even in intact zebrafish. The method may assist in further optimisation of GCS inhibitors for therapeutic purposes.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. The LC-MS/MS chromatograms of 13 C 5sphingosine, and deacylated 13 C 5 -ceramide, glucosylceramide, lactosylceramide, and globotriaosylceramide. Fig. S2. A comparison of 0.5 nmol 13 C 5 -sphingosine (light bars) and -sphinganine (dark bars) feeding with 2 hour and 4 hours incubation time.