Aglycon diversity of brain sterylglucosides: structure determination of cholesteryl- and sitosterylglucoside[S]

To date, sterylglucosides have been reported to be present in various fungi, plants, and animals. In bacteria, such as Helicobacter pylori, proton NMR spectral analysis of isolated 1-O-cholesteryl-β-d-glucopyranoside (GlcChol) demonstrated the presence of an α-glucosidic linkage. By contrast, in animals, no detailed structural analysis of GlcChol has been reported, in part because animal-derived samples contain a high abundance of glucosylceramides (GlcCers)/galactosylceramides, which exhibit highly similar chromatographic behavior to GlcChol. A key step in vertebrate GlcChol biosynthesis is the transglucosylation reaction catalyzed by glucocerebrosidase (GBA)1 or GBA2, utilizing GlcCer as a glucose donor. These steps are expected to produce a β-glucosidic linkage. Impaired GBA1 and GBA2 function is associated with neurological disorders, such as cerebellar ataxia, spastic paraplegia, and Parkinson’s disease. Utilizing a novel three-step chromatographic procedure, we prepared highly enriched GlcChol from embryonic chicken brain, allowing complete structural confirmation of the β-glucosidic linkage by 1H-NMR analysis. Unexpectedly, during purification, two additional sterylglucoside fractions were isolated. NMR and GC/MS analyses confirmed that the plant-type sitosterylglucoside in vertebrate brain is present throughout embryonic development. The aglycon structure of the remaining sterylglucoside (GSX-2) remains elusive due to its low abundance. Together, our results uncovered unexpected aglycon heterogeneity of sterylglucosides in vertebrate brain.

rocking egg incubator at 38°C. Upon reaching the appropriate stage, the embryos were sacrificed, either heads without eyes (6 days old) or brains (8-18 days old) were quickly harvested and immediately frozen in liquid nitrogen (N 2 ). After lyophilization, samples were stored at 80°C until further use.

Isolation of sterylglucosides
A portion of the enriched sterylglucoside fraction was further purified by RP-HPLC, as described previously (19) with minor modifications. Briefly, the lipid film was resuspended in a small volume of mobile phase B (M:W at 85: 15, v/v), applied to an RP-HPLC Luna C18(2) column [4.6 mm inner diameter (i.d.) × 250 mm, particle size, 3 m; Phenomenex, Torrance, CA], and eluted with the following gradient of mobile phase A (pure methanol): 2 min, 0%; 13 min, 0-100% linear gradient; 40 min, 100% (washing step); and 15 min, 0% (equilibration). The flow rate was kept constant at 0.6 ml/ min, and the column was maintained at room temperature. Lipid detection with short wave UV light at 205 nm was disabled during elution to prevent potential damage of the UV-absorbing double bond at the C5-C6 position of the expected aglycon (see Fig. 2B). The eluent was manually collected into 14 fractions in the following volumes: fraction 1, 15 ml; fractions 2-13, 0.3-0.6 ml, and fraction 14, 22 ml. Subsequently, all fractions were subjected to LC-ESI-MS/ MS analysis, utilizing multiple reaction monitoring (MRM).

Enzymatic deglucosylation of sterylglucosides
Aglycon release was effected by suspending a portion of the enriched sterylglucoside fraction in a total volume of 20 or 40 l reaction buffer [50 mM citrate-phosphate buffer (pH 5.3), 0.25% Triton X-100, 0.6% sodium taurocholate, 1-2 l of 100 g/l Cerezyme ® in PBS] and incubation at 37°C for 16-20 h. The reaction was terminated by addition of 2 ml of C:M (2:1, v/v) and 460 or 480 l of water [adjusted to a total volume of 500 l (including the reaction mixture)] to facilitate lipid extraction after Folch's partition. The organic layer was separated, dried, and the sterol fraction was purified by TLC on silica gel 60 using hexane:diethyl ether:acetic acid (80:20:1, v/v/v) as an eluent. Lipids were stained by primuline reagent (0.01% primuline, 80% acetone) and visualized by long wave UV detection. The band comigrating with standard Chol (Wako, Osaka, Japan) was collected and extracted by Folch's partition. The organic layer containing the released aglycons was separated and dried under a stream of N 2 gas. neurological disorders, such as cerebellar ataxia and spastic paraplegia (13)(14)(15). However, the exact molecular mechanism and specific contributions of GBA1 and GBA2 dysfunction to neuronal malfunction and degeneration are not fully elucidated. Nevertheless, the transferase activity of GBA1 and GBA2 represents an intriguing intersection between two major lipid metabolic pathways, namely sphingolipids and sterols. Consequently, sphingolipid-sterol cross-talk may be important in maintaining neuronal homeostasis, all in the context that its deregulation plays a crucial role in the pathogenesis of neurodegenerative disorders, such as GD and Parkinson's disease.
While the presence of GlcChol in human and mouse tissue, such as brain, has been inferred by LC-ESI-MS/MS analyses (5), its purification and complete structural analysis based on GlcChol-containing fractions are yet to be reported. Here, we describe the purification of the sterylglucoside fraction from embryonic chicken brain. Embryonic chicken brains were selected as starting material due to the well-investigated and comparatively discernable composition of their glycolipid fraction (16,17). During development of the purification procedure, we placed special emphasis on removing the large excess of chromatographically similar galactosylceramide (GalCer) known to be present in the CNS of vertebrates. Structural analysis of the isolated sterylglucoside fraction revealed the presence of a variety of aglycons. In addition to the major component featuring the expected cholesteryl aglycon, at least two more steryl aglycons were encountered, including the plant-type sitosteryl. Our work described here is the first report to describe the complete structure of GlcChol and 1-O-sitosteryl--d-glucopyranoside [(GlcSito), also known as glucosyl--d-sitosterol] derived from vertebrate brain and to demonstrate that sterylglucosides have a heterogeneous aglycon composition.

Animals
Fertilized Boris Brown chicken eggs were purchased from Inoue Poultry Farm (Sagamihara, Japan) and maintained in a Da, calculated as loss of hexose and NH 3 , as rationalized in supplemental Fig. S1; scan speed, 200 Da/s; collision energy, 15 eV.

NMR spectroscopy
Highly purified sterylglucoside fractions and authentic standards were dissolved in CDCl 3 containing tetramethylsilane as an internal chemical shift reference. One-dimensional 1 H-NMR and two-dimensional double quantum filtered correlation spectroscopy (DQF-COSY) and homonuclear Hartmann-Hahn (HOHAHA) spectra, as well as 1 H- 13 C multiplicity-edited heteronuclear single quantum coherence (HSQC) spectra were recorded on a DRX-500 spectrometer (Bruker BioSpin, Yokohama, Japan) equipped with a TXI cryogenic probe. Probe temperature was set at 25°C. The NMR data were processed with XWIN-NMR (version 3.5) and the spectra were displayed using XWIN-PLOT (version 3.5).

Synthesis of deuterium-labeled cholesteryl--d-glucoside
The synthetic route to deuterium-labeled cholesteryl--dglucoside (GlcChol-d7) is shown in supplemental Fig. S2. Unless stated otherwise, reactions were performed under argon. All solvents and chemicals were purchased as reagent grade from commercial suppliers and used without further purification, unless stated otherwise. Dry solvents were purchased from Kanto Chemical Co. and used as supplied. Analytical TLC and flash column chromatography were performed using the indicated solvent systems on Merck silica gel 60 F256 plates and on Kanto Chemical Co. silica gel 60 N (40-100 mesh), respectively. Low-resolution mass spectra (LRMS) were recorded on an SCIEX 4000 QTRAP mass spectrometer. NMR spectra were obtained on a JEOL ECA-500 spectrometer ( 1 H at 500, 13 C at 125 MHz) in the indicated solvents, with chemical shift referenced to residual nondeuterated solvent.

GC/MS analysis
The lipid containing released aglycons was suspended in 25 l of TMS at room temperature for 30 min. The resulting trimethylsilylated material was subjected to GC/MS analysis on a GCMS-QP2010 Ultra (Shimadzu, Kyoto, Japan) equipped with an Ultra1 capillary column (25 m × 0.2 mm, film thickness of 0.33 m; Agilent Technologies Inc., Santa Clara, CA). We employed the following temperature gradient: from 180 to 250°C at a heating rate of 20°C/min and from 250 to 300°C at a heating rate of 5°C/min.

MALDI-Spiral TOF/TOF analysis
The highly purified sterylglucoside fraction and authentic GlcChol standard were each dissolved in C:M (1:1, v/v) at a concentration of 1 g/l, mixed with MALDI matrix A [10 g/l of 2,5-dihydroxybenzoic acid in C:M (1:1, v/v)] and matrix B (1 g/l LiCl in water) at a ratio of 1:1:1 (v/v/v). From the resulting mixture, 0.6-1.5 l were spotted onto a MTP 384-hole mirror finish stainless steel plate (JEOL Ltd., Tokyo, Japan) and dried. The samples were analyzed with a JMS-S3000 Spiral TOF (JEOL Ltd., Akishima, Japan) equipped with the TOF/TOF option (20). A Nd-YLF laser pulse of 349 nm was operated at 250 Hz. For product-ion mass spectrum acquisition, helium collision gas was introduced. The collision energy was 20 keV to induce high-energy collision-induced dissociation (CID).

Quantification of sterylglucosides in animal tissue
The lyophilized embryonic chicken tissue (110 mg) at the desired developmental stage was homogenized and total lipids were extracted with a C:M (2:1, v/v, 5-10 ml) mixture spiked with 5 pmol/mg lyophilized tissue of GalCer (d18:1-C12:0) and GlcChol-d7, each as internal standards. Extracts were dried under a flow of N 2 gas and hydrolyzed for 2 h at room temperature in C:M (2:1, v/v, 2 ml) containing 0.1 M KOH. After neutralization with 1 M acetic acid (100 l), the reaction mixture was subjected to Folch's partition and the lower phase was dried under a flow of N 2 gas. The resulting lipid film was suspended in C:M (2:1, v/v) at a concentration of 200 g lyophilized tissue/l, diluted 10-fold with mobile phase B or A, and aliquots (10 l) were subjected to RPLC-ESI-MS/MS or HILIC-ESI-MS/MS analysis, respectively. Peak areas were integrated and quantified relative to the associated internal standard.

Statistical analysis
Statistical analyses were performed using GraphPad Prism version 5.0 for Windows (GraphPad Software, San Diego, CA) to calculate mean ± SEM values.

Purification of sterylglucosides
A multi-step chromatographic purification protocol was developed to isolate the sterylglucoside fraction from CNS tissue in order to facilitate a complete structure analysis. After hydrolysis of the glycerophospholipid fraction, sterylglucosides were enriched by normal phase chromatography, utilizing a C:M gradient during elution, together with the large excess of GalCer known to be present in vertebrate brain. Removal of the large excess of GalCer was achieved by RP chromatography utilizing a M:W gradient (Fig. 1A).
follows. Cholesterol-d7 (100 mg, 0.25 mmol) and 2,3,4,6-tetra-O-acetyl--d-glucopyranosyl 2,2,2-trichloroacetimidate (150 mg, 0.3 mmol) were dissolved in dry dichloromethane (5 ml) at 40°C and stirred for 10 min. The reaction was initiated by addition of trimethylsilyl trifluoromethanesulfonate (5 l) and stirred for 2 h at 40°C. Subsequently, the reaction was quenched with triethylamine (1 ml), and the volume was increased with dichloromethane (5 ml) prior to extraction against water and brine. The organic layer was dried over Na 2 SO 4 and concentrated in vacuo. The residue was subjected to flash chromatography on silica gel (gradient of hexane:ethyl acetate at 10:1 to 1:1, v/v); product elution was monitored by TLC (hexane:ethyl acetate at 3:1, v/v), to give compound 3 as a white amorphous solid (112.5 mg, 0.16 mmol, 63% yield).  Fig. S2) was synthesized as follows. Dry methanol (9 ml) and dry dioxane (1 ml) were mixed and treated with sodium (5 mg). After the initial reaction subsided and the mixture cooled to room temperature, it was transferred under Schlenk conditions to a new reaction vessel charged with compound 3 (100 mg, 0.14 mmol). Reaction progress was monitored by TLC (C:M at 9:1, v/v) and, after 2 h stirring at room temperature, the reaction mixture was dried in vacuo. The residue was subjected to flash chromatography on silica gel (gradient of C:M at 50:1 to 5:1, v/v) to yield compound 4 as a white amorphous solid (33 mg, 0.06 mmol, 43% yield). Finally, a portion was subjected to RP-HPLC purification as detailed above (see Isolation of sterylglucosides) prior to its application as an internal standard for MS-based sterylglucoside quantification.  with tetrabutylammonium cation and its intrinsic positive charge facilitated excellent ion yield in positive mode. The presence of the prominent ion in all tested samples at similar ion yield suggests that this contaminant was instrumentally entered into the final purification step. The predicted lithium adduct of GlcChol (C 33 H 56 O 6 , 548.41) at [M + Li] + m/z 555.5 was subjected to high-energy CID. The product ion spectrum (Fig. 2C)  . The charge remote fragmentation pattern matched the fingerprint of the Chol standard (23), with the expected constant shift of 162 Da (Fig. 2D) supporting the presence of a sterylhexoside.
NMR analysis of GSX-1 revealed a low-field pattern in the range of 4.5-3.3 ppm (Fig. 3A), which is typically associated with glucopyranosides. The 1 H chemical shifts of the hexoside portion match well with those of the GlcChol reference ( Table 2). The large splitting (7.8 Hz) of the anomeric proton (Glc H-1) at 4.44 ppm is archetypical for an axialaxial J coupling consistent with -glycosidic linkage. In the low-field region, the aglycon exhibits a deshielded vinyl proton at 5.38 ppm (Chol H-6) consistent with the presence of a Chol C5-C6 double bond and the bridging proton at 3.58 ppm (Chol H-3). Complete superposition (Fig. 3D) of the Monitoring by boronated TLC revealed the successful separation of the chromatographically similar GalCer fraction. Fractions (13)(14)(15)(16)(17) corresponding to the standard GlcChol in the Rf value were pooled for further treatment and were designated as GSX.
To confirm the presence of glycosylated sterols, GSX was subjected to RPLC-ESI-MS/MS analysis by neutral-loss scan. Fragmentation under CID conditions favored elimination of the carbohydrate moiety, together with the cationic site. This is elucidated through formation of a resonance-stabilized homo-allylic carbocation of the aglycon (supplemental Fig. S1). Surprisingly, three distinct peaks were detected, indicating the presence of multiple compounds (Fig. 1B). Each major peak corresponded to the individual molecular-related ion [M + NH 4 ] + at m/z 566, 580, and 594, respectively. Utilizing a large-scale column with similar chromatographic conditions, GSX was further separated into three major fractions, termed GSX-1, GSX-2, and GSX-3. Fractions were monitored by RPLC-ESI-MS/MS (Fig. 1C) and corresponding fractions were pooled as indicated.
The prominent ion at m/z 242.3 ( Fig. 2A) was identified the low-field region, the HOHAHA spectrum (Fig. 4A) confirmed the presence of glucopyranoside as the sole carbohydrate. The most notable HSQC cross-peak (Fig. 4B), not corresponding to the Chol aglycon of GSX-1, is located at 13 C:  = 45.9 ppm/ C HSQC correlation spectra of GSX-1 (Fig. 3B) with an authentic GlcChol sample (Fig.  3C) confirms the presence of the expected -cholesteryl aglycon. This concludes the complete structure determination of natural GlcChol from embryonic chicken brain.

Identification of minor sterylglucosides
The low content of the highly purified fractions GSX-2 and GSX-3 precluded their NMR spectroscopic analysis. However, to gain initial structural insight, we subjected the partially purified fraction GSX to NMR analysis (Fig. 4). In ( Table 3) with authentic GlcSito revealed good agreement, further supporting the presence of a sitosteryl aglycon.
Next, the GSX fraction was subjected to enzymatic deglycosylation, followed by GC/MS analysis of the liberated aglycons (Fig. 5). As expected, Chol eluted at 19.3 min (peak 1) and represented the major sterol. Additionally, two more sterols eluted at 21.4 and 23.5 min, termed peak 2 and peak 3, respectively. The elution time and fragmentation pattern of Fig. 5A peak 3 (Fig. 5C) coincided well with authentic sitosterol (supplemental Fig. S3B), strongly supporting the presence of GlcSito. The elution time and molecular ion [M] +· of Fig. 5A peak 2 (Fig. 5B) corresponded to authentic campesterol (supplemental Fig.  S3A), although their fragmentation pattern matched poorly. For example, the intensity ratio between m/z 129 and its complementary fragment [M  129] +· m/z 343 differed greatly between authentic campesterol and peak 2 (supplemental Fig. S3A, Fig. 5B). Moreover, fragment m/z 192.3 was only present in peak 2, while m/z 457, which is typically associated with loss of a methyl radical, was absent. However, the lack of any diagnostic NMR signal associated with campesterol or its C-24 epimer, dihydrobrassicasterol, might be attributed to its low abundance in the GSX fraction.

Developmentally dependent presence of sterylglucosides
To allow accurate quantification of sterylglucosides, a deuterated GlcChol (GlcChol-d7) derivative was prepared, as detailed in the Materials and Methods. Lipid extracts of embryonic chicken brains aged from E6 (heads without eyes) to E18, spiked with the internal standards, GlcChol-d7 and GalCer (d18:1-C12:0), were subjected to LC-ESI-MS/MS analysis. MRM monitoring of sterylglucosides, specifically GlcChol, GlcSito, and GSX-2 (Fig. 6), revealed the continued presence of all three sterylglucosides during embryonic development. GlcCer was monitored alongside the sterylglucosides. The abundance of GlcChol detected in stage E6 embryonic chicken brain was comparable to GlcCer (d18:1-C16:0). In contrast to the continued decrease of GlcCer (d18:1-C16:0) during development, GlcChol content decreased only until stage E10 and subsequently steadily increased until E18, resulting in about a three times excess of GlcChol compared with GlcCer (d18:1-C16:0). Similarly, GSX-2 and GlcSito continuously decreased until stage E10 and subsequently stabilized to a low amount. At all time points analyzed, GSX-2 and GlcSito fractions were digested completely by Cerezyme ® . By contrast, at stages E16 and E18, about 10% of the GlcChol fraction was Cerezyme ® resistant, consistent with previous reports (5).

DISCUSSION
To date, the evidence that GlcChol is present in vertebrate brain was based on LC-ESI-MS/MS data (5). Isolation of sufficient GlcChol from brain for mandatory NMR structural analysis has been hampered by the abundant copresence of GalCer, which exhibits highly similar chromatographic behavior to GlcChol. In the present study, we achieved separation of GlcChol from GalCer using a newly established two-step chromatographic setup. Untargeted LC-ESI-MS/MS analysis of the putative GalCer-free GlcCholrich fraction (GSX) isolated from embryonic chicken brain revealed the presence of at least three distinct sterylglucosides. Fragment ions of the main component corresponded well to the expected GlcChol, while the two minor components exhibited an increased mass, greater by 14 and 28 Da. Charge-mediated as well as chargeremote fragmentation (24) of the further-purified main component and Chol standard by MALDI-Spiral TOF/ TOF and Fourier transform ion cyclotron resonance mass spectrometry (23,25), together with GC/MS analysis of the liberated aglycon, confirmed the presence of a -cholesteryl residue. Complete superposition of the 2D NMR spectral data from natural and authentic GlcChol proves that a glucopyranoside residue is present and is linked via a -glycosidic linkage to a -cholesteryl aglycon.
The minor component, with a mass difference of +28 Da compared with GlcChol, is consistent with the presence of two additional methylene groups. Indeed, 2D NMR and GC/MS spectral data corresponded well to authentic sitosteryl--d-glucopyranoside. The minor component (GSX-2), with a mass difference of +14 Da compared with GlcChol, suggests either the presence of an additional methylene or keto group. The GC elution profile of GSX-2 is consistent with sterols featuring an additional methylene group compared with Chol like campesterol, but its EI MS fingerprint did not match well with authentic campesterol (Fig. 5B, supplementary Fig. S3A). Specifically, the intensity ratio between m/z 129.1 and its complementary fragment [M-129] +· m/z 343.2 generated during EI-induced retro-Diels-Alder reaction has been suggested to reflect the relative stability of the formed oxonium and C-4 carbonium ion, respectively (26). Elevated intensity of [M-129] +· is thus associated with resonance stabilization of the C-4 carbonium ion via the adjacent double bond or by electron-donating C-4 methylation. The prominent intensity of m/z 129 in GSX-2 thus suggests reduced C-4 carbonium ion stability compared with campesterol, possibly due to an altered position of the double bond. Nevertheless, fragments m/z 254 and 213 indicate the presence of a double bond in the B or C ring, while suggesting that the putative additional methylene or keto group is associated with the sterol side chain. expression of CerS1. The same trends of CerS6 and CerS1 expression levels have been reported during postnatal rat brain development (P1 to P21). The late stage increase of GlcCer (d18:1-C24:1) could be associated with CerS2 [a key enzyme in myelin biosynthesis (28)] expression levels and coincides well with the known onset of myelination in late stage chicken embryos. In contrast, the blood-brain barrier development occurs between P1 and P30 in chicken and is likely not associated with the observed GlcCer and sterylglycoside embryonic development profile. The developmental profile of each GlcCer species in embryonic chicken brain differed markedly. While the metabolic importance of this fatty acid-specific profile of GlcCers is currently unclear, its origin might be related to the strong correlation observed between ceramide species and the expression levels of their respective ceramide synthases, as reported during postnatal rat brain development (27). The steady decline of GlcCer (d18:1-C16:0) might indicate a steady reduction of ceramide synthase (CerS)6 expression levels, while the increase in GlcCer (d18:1-C18:0) could correspond to increased the presence of Chol, campesterol, and sitosterol was detected in both embryonic chicken brain and egg yolk (data not shown), while the presence or absence of the free unknown GSX-2 aglycon in egg yolk and embryonic chicken brain could not be confirmed.
These findings have functional implications. The consumption of cycad (the seed of Cycas circinalis), a traditional indigenous food source in Guam, is one of the strongest epidemiological links to the Guamanian neurological disorder, amyotrophic lateral sclerosis-parkinsonism dementia complex (ALS-PDC) (31). Sterylglucosides in cycad seeds and flour, such as GlcSito (main sterylglucoside component), Throughout embryonic development, GlcChol levels in the chicken brain exceeded GlcChol levels in egg yolk (data not shown). This is in line with previous reports showing that GlcChol can be synthesized via transglucosylation from GlcCer by GBA1 or GBA2 in the lysosomal compartments or at the cytosolic surface of the endoplasmic reticulum and/or Golgi, respectively (5,12). By contrast, the content of GlcSito and GSX-2 in egg yolk exceeded their abundance in embryonic chicken brain at all stages analyzed. The presence of diet-derived plant sterols, such as campesterol and sitosterol, in murine brain has been reported previously (29,30). Consistent with these reports, campesteryl-/dihydrobrassicasterylglucoside, and stigmasterylglucoside, have been demonstrated to be neurotoxins in vitro (31,32) and in vivo (32)(33)(34). Rodents fed with washed cycad flour or GlcSito recapitulated multiple key features of ALS-PDC, suggesting that plant-type sterylglucosides are a potential factor involved in the development of neurodegenerative disorders in vivo. Taking into consideration that Marques et al. (5) demonstrated the presence of GlcChol in human plasma, it would be of great interest to quantify sterylglucoside levels in the plasma of ALS-PDC patients in the future, based on the above detailed LC-MS/ MS method. Nevertheless, direct uptake and incorporation into brain of plant-type sterylglucosides have not been demonstrated. Similarly, acceptor preference of GBA1 and GBA2 toward other sterols, including plant sterols or even polyphenols, have not been reported to date. It remains to be determined whether GlcSito and GSX-2 are derived from egg yolk or are de novo synthesized in embryonic chicken brain. Fig. 6. Concentrations of sterylglucosides and selected GlcCer species in embryonic chicken brain at indicated developmental stage. Numerical values were averaged over three to four experiments. Data (mean ± SEM) were analyzed using an unpaired t-test. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with the concentration of each lipid in E6 heads without eyes.