Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2023 Oct 25;95(45):16465–16473. doi: 10.1021/acs.analchem.3c00737

Defining the Ceramide Composition of Bovine and Human Milk Gangliosides by Direct Infusion ESI-CID Tandem Mass Spectrometry of Native and Permethylated Molecular Species

O Tara Liyanage †, Chaoshuang Xia †, Silvia Ringler ‡, Bernd Stahl †,‡,§, Catherine E Costello †,*
PMCID: PMC10652237  PMID: 37877731

Abstract

graphic file with name ac3c00737_0008.jpg

Gangliosides are glycosphingolipids composed of an oligosaccharide that contains one or more sialic acid residues and is linked to a ceramide, a lipid composed of a long chain base (LCB) that bears an amide-linked fatty acyl group (FA). The ceramide portions of gangliosides are embedded in cell membranes; the exposed glycans interact with the extracellular environment. Gangliosides play a myriad of roles in activities such as cell–cell communication, formation of lipid rafts, cellular adhesion, calcium homeostasis, host-pathogen interaction, and viral invasion. Although the epitopes responsible for the interactions of gangliosides are located in the glycan, the epitope presentation is strongly influenced by the orientation of the attached ceramide within the lipid membrane, a feature that depends on the details of its structure, that is, the specific LCB and FA. Since the identities of both the glycan and the ceramide affect the activity of gangliosides, it is important to characterize the individual intact molecular forms. We report here a mass spectrometry-based method that combines the information gained from low-energy collision-induced dissociation (CID) measurements for the determination of the glycan with tandem mass spectra obtained at stepped higher-energy CID for the detailed characterization of the LCB and FA components of intact gangliosides. We provide results from applications of this method to the analysis of gangliosides present in bovine and human milk in order to demonstrate the assignment of LCB and FA for intact gangliosides and differential detection of isomeric ceramide structures.

Gangliosides make up a class of highly complex glycosphingolipids that contain at least one sialic acid. They are present in cell membranes and are known to play a myriad of roles in cell–cell communication, host pathogen interaction, the formation of lipid rafts, cellular adhesion, and calcium homeostatsis.1−4 The highest abundance of this class of molecules is found in the brain, where they are considered vital in the formation of synapses and the transmission of nervous impulses.5 Gangliosides are designated according to the Svennerholm nomenclature, for example, GM3 and GD3, where the second letter (given here as M or D) indicates the number of sialic acid residues (here, one or two, respectively), and the number (given here as 3) is related to the length of the glycan backbone.6

Milk is a rich source of gangliosides; in bovine and human milk, GM3 and GD3 have the highest abundances.7,8 Due to the vital roles that gangliosides play in multiple biological processes, it seems intuitive that they are also crucial for infant growth and development. Human milk oligosaccharides and gangliosides can act as decoys to deflect harmful pathogens from binding to the receptors in the guts of infants, who do not yet have a fully developed immune system.9 Gangliosides have also been associated with the neurological development of infants.10 Although the epitopes responsible for the interactions of gangliosides with the extracellular environment are located in the glycan, the presentation of these epitopes is determined by the orientation of the attached ceramide within the lipid membrane, a feature that depends on the details of its structure, that is, the specific long chain base (LCB) and fatty acyl group (FA). Since the identities of both the glycan and the ceramide affect the activity of gangliosides, it is important to characterize the individual intact molecular forms, to understand their roles in biological processes, to evaluate the potential benefit to be gained by supplementing these in infant formulas, and to explore the possible use of gangliosides as therapeutics.

Gangliosides are amphiphilic: they contain a hydrophilic glycan headgroup linked to a hydrophobic ceramide. This property makes the characterization of intact gangliosides analytically challenging, and thus, many reports on the compositions of gangliosides extracted from biological sources have been based on independent determinations of the ceramide (FA and/or LCB) and glycan (GXn) distributions in the entire ganglioside pool. Most investigations that have utilized mass spectrometry-based methods to analyze intact gangliosides present in milk have employed conditions that primarily dissociate only glycosidic bonds and thereby characterize each linked ceramide component solely in terms of total lipid composition (i.e., total numbers of hydroxyl groups and of carbon atoms and degree of unsaturation). However, methods developed for analyzing the glycan and lipid groups separately and those that yield only total compositions of the ceramides do not fully define the distribution of species upon whose intact structures the biological activity depends. Therefore, we have undertaken the design of sensitive and specific methods for the full definition of intact gangliosides in bovine and human milk samples in order to elucidate the detailed patterns in bovine milk, particularly those used for infant formulas, and in human milk during different stages of lactation. The aim is to identify components that might contribute to the design of optimal infant formulas and to increase understanding of the biological roles of milk in infant development.

Mass spectrometry and tandem mass spectrometric methodologies are indispensable tools for the analysis of milk gangliosides, but the full power of this approach has not been exploited. Early work noted above that separately characterized the ceramide and the glycan structures of milk gangliosides largely relied upon GC23 or GC–MS.27 In recent reports, LC–MS/MS methodologies have been applied to characterize and quantify milk gangliosides. However, these recent studies still focused on the analysis of the glycan headgroup and reported only the total composition of the ceramide, that is, the sum of the alkyl chains present in the LCB and FA.8,11 Collision-induced dissociation (CID) MS/MS in the range of 10–50 eV is the dissociation technique most widely employed on currently available instruments, but it produces only limited fragmentation of ceramides or the ceramide portion of glycosphingolipids. As noted above, the authors of most published reports on milk gangliosides have selected a single collision energy within this range and published spectra dominated by glycosidic fragments; in these, the contribution from fragmentation within the ceramide is minimal. We report here an electrospray ionization quadrupole time-of-flight (ESI-QTOF) MS/MS method that supplements low-energy CID measurements for the determination of the glycan with MS/MS spectra obtained at CID settings extended stepwise to voltages well above 100 V for further detailed characterization of the LCB and FA components of intact gangliosides. We have evaluated the use of a combination of collision energies over the range of 30–130 eV to define both the glycan and ceramide moieties of intact gangliosides. Our strategy utilizes both positive- and negative-ion modes to compile an array of ceramide fragments that enable confident assignment of the ceramide compositions of gangliosides in bovine and human milk. Among these are product ions that we had previously identified in the high-energy (5–10 keV) CID spectra of glycosphingolipids12 and/or that had been reported in the low-energy CID of ceramides13−15 but had not yet been documented in the CID spectra of glycosphingolipids. Herein, we provide representative results from our initial applications of this method to the analysis of bovine and human milks. We verify the key product ion assignments via parallel MS4 analyses performed on a tribrid orbitrap instrument. Within samples that contain a mixture of components, we have defined both the LCB and the FA moieties for multiple homologous peaks, and we include here examples of homologues that contain isomeric species. To the best of our knowledge, this is the first study to determine the detailed composition of the ceramides present in intact human milk gangliosides.

Experimental Section

Materials

C16 β-d-lactosyl ceramide (LacCer (18:1/16:0), C12 glucosyl(β) ceramide (Glc (d18:1/12:0)), C24:0 galactosyl(β) ceramide (Gal (d18:1/24:0)), and GM3 and GD3 ganglioside mixtures isolated from bovine milk were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). Gangliosides were also extracted from bovine milk and human milk samples donated to Danone Nutricia Research in accordance with the Helsinki Declaration of 1975, as revised in 1983. Sodium hydroxide (NaOH) pellets, dimethyl sulfoxide (DMSO), and methyl iodide (CH3I) were from Sigma-Aldrich (St. Louis, MO). All the solvents used for extraction steps and direct infusion experiments were LC–MS grade, except for chloroform (CHCl3), which was analytical grade (ethanol stabilized).

Permethylation

Gangliosides were permethylated according to the methods of Ciucanu and Kerek,16 Ciucanu and Costello,17 and Gunnarsson18 with a few modifications. Briefly, a solution containing 0.1–100 μg of a commercial ganglioside mixture or an extract from bovine or human milk was dried in a glass conical vial (DWK Life Sciences (Wheaton) no. 986254, Millville, NJ) using a speed vacuum concentrator. NaOH pellets (∼200 mg, crushed using a mortar and pestle) were added per mL of DMSO. The NaOH/DMSO mixture was vortexed, and the desired volumes were pipetted from the cloudy region. The dried oligosaccharides were resuspended in 100 μL of NaOH/DMSO, and the mixture was vortexed for a few seconds. One-half volume (50 μL) of CH3I was added, the solution was vortexed for 30 s, and the reaction was allowed to proceed for 1 h. An additional volume (100 μL) of NaOH/DMSO was then added, again followed by 50 μL of CH3I, and the reaction was allowed to continue for an additional hour. The process was repeated two more times. Thereafter, the reaction was terminated by the addition of 600 μL of CHCl3, followed by 600 μL of H2O (CHCl3 was added first because the addition of H2O dissolves the NaOH and increases the pH). The mixture was vortexed and centrifuged to facilitate the partitioning of the layers. The top (aqueous) layer was removed, and the bottom (organic) layer was retained. To the tube containing the organic layer, 400 μL of H2O was added, and the mixture was vortexed. The organic layer was retained, and the aqueous layer was discarded. This step was repeated until the aqueous layer was neutral upon testing with the pH strips. CHCl3 was evaporated using a speed vacuum concentrator, and the permethylated sample was reconstituted in 1:1 (v/v) MeOH/H2O.

Ganglioside Extraction from Bovine Milk or Human Milk

Gangliosides were extracted from bovine milk or human milk (0.5 mL) using a mixture of CHCl3, MeOH, and H2O, according to the extraction protocol of Svennerholm and Fredman,19 as described by Fong et al.20 with a slight modification. Briefly, 0.5 mL of H2O was added to 0.5 mL of milk in 12 mL glass tube (Kimax, DWK Life Sciences), and the mixture was vortexed. MeOH (2.7 mL) and CHCl3 (1.35 mL) were added, and the mixture was allowed to rock gently for 30 min before centrifugation for 30 min. The supernatant was transferred to a 15 mL Kimax tube, and the pellet was re-extracted using H2O (0.5 mL) and CHCl3/MeOH (1:2, 2 mL). After mixing and centrifugation, the supernatants were pooled, and the pellet was discarded. Solvent partition of the supernatant was achieved by the addition of 350 μL of 0.8 M KCl and 950 μL of H2O and gentle inversion three to four times, followed by centrifugation. The upper phase was passed through a C18 SPE cartridge (Waters, Milford, MA) for cleanup, and the lipid fractions were eluted with MeOH.

Workflow for the Characterization of Ceramide Compositions of Gangliosides in Human and Bovine Milk

A table of all possible ceramide fragments was created using fragment ion assignments obtained from our analyses of neutral glycolipid standards, as well as information available in the LIPID Metabolites and Structure Strategy database (LIPID MAPS, www.lipidmaps.org), previous literature, and theoretical values calculated for novel gangliosides based on the fragmentation patterns of standards. The nomenclature originally proposed by Domon and Costello12 supplemented by additions from other authors13−15,21 was used to designate the MS/MS fragments. The m/z values for MSn ceramide-derived fragment ions obtained experimentally were manually searched against the assembled tables of all of the potential ceramide fragments. The product ion assignments were matched within ±10 ppm and confirmed by manual inspection

Results and Discussion

MS/MS Data from the Neutral Glycolipid Standards

In this study, native and permethylated neutral glycolipid standards with known LCB and FA groups were analyzed to determine the types of ceramide-derived fragments that can be obtained under the instrumental conditions evaluated during our method development. The fragmentation patterns are instrument- and parameter-dependent, yet there were few relevant literature sources for ceramide fragmentation under the chosen conditions; indeed, it has been reported that obtaining a variety of ceramide fragments from glycolipids by MS/MS in the negative mode is “challenging.”22 Therefore, it was necessary for us to first build a library of the diagnostic m/z values found in MSn spectra recorded over a range of collision voltage settings in order to determine the optimum conditions for the generation of LCB- and FA-specific fragments from neutral glycosphingolipids with defined chemical structures. Using a representative example, Figure 1 summarizes the fragmentation pathways and designations used for product ions obtained in these modes. Supporting Information Table 1 lists the calculated values for product ions labeled in all of the figures. We observed a wide array of ceramide fragments for neutral glycolipids under the instrumental conditions utilized in our study. These were quite similar to the most abundant fragments we had reported for spectra recorded at keV dissociation energies.12 Representative MS2 spectra obtained in the negative-ion mode are shown in Figure 2. Dissociation of the [M – H]− species generated for GlcCer (d18:1/12:0) at m/z 642.495 and GalCer (d18:1/24:0) at m/z 810.680 both produced P(d18:1) fragments at m/z 237.222 and R(d18:1) fragments at m/z 263.237, which defined the LCB. Evidence for the FA chain (12:0) in the MS2 spectrum of the GlcCer standard shown in Figure 2a included the S ion at m/z 224.201, the T ion at m/z 240.198, the U ion at m/z 198.186, and the V ion at m/z 181.159, in addition to the RCOO– ion at m/z 199.170. The presence of the fatty acyl chain 24:0 in the GalCer (d18:1/24:0) standard was consistent with the observation of the characteristic product ions labeled in the MS2 spectrum (Figure 2b): S at m/z 392.389, T at m/z 408.384, V ion at 349.347, and RCOO– at m/z 367.358. In the MS2 spectrum recorded for LacCer(d18:1/16:0) (Figure 2c), the product ion P(d18:1) observed at m/z 237.222 originated from the LCB; the product ions S at m/z 280.2644 and RCOO– at m/z 255.231 originated from the FA chain (16:0). It is interesting to note that the formation of the RCOO– species observed in these and other negative-ion MS/MS spectra included in this report requires a rearrangement of the acyl group.14

Figure 1.

Figure 1

Open in a new tab

Scheme of ceramide fragments from (a) native glycolipids analyzed in the negative-ion mode. (1,2) (b) Permethylated glycolipids analyzed in the positive-ion mode. LCB = long chain base; FA = fatty acyl group. The example used here illustrates the structures and product ion values calculated for ceramide [d18:1/12:0]. Variations in the LCB and FA induce diagnostic m/z shifts.

Figure 2.

Figure 2

Open in a new tab

ESI-QTOF CID MS/MS (CE 40–50 eV) spectra of native neutral glycosphingolipids recorded in the negative-ion mode. (a) GlcCer (d18:1/12:0), [M – H]−m/z 642.495, (b) GalCer (d18:1/24:0) [M – H]−m/z 810.680, and (c) LacCer (d18:1/16:0) [M – H]−m/z 860.609. For product ion assignments, see Figure S1b.

CID MS/MS analysis of the [M + Na]+ ions from the permethylated derivatives of these same neutral glycolipid standards generated the MS2 spectra shown in Figure 3a–c. Since all contained the d18:1 LCB, they all produced the W ion at m/z 310.311, which underwent loss of methanol to generate the very abundant W′ ion at m/z 278.284. The only other abundant product ions in these MS2 spectra were C- or Z-types that arise via glycosidic cleavages; the Z0 fragments define the total composition of the ceramide, whereas the W and W′ fragments specify the LCB. The difference between the compositions of the W- and Z0-type fragments allows calculation of the FA length and its level of unsaturation.

Figure 3.

Figure 3

Open in a new tab

ESI-QTOF CID-MS/MS (CE 70–80 eV) spectra of permethylated glycolipids recorded in the positive-ion mode (a) GlcCer (d18:1/12:0), [M + Na]+m/z 750.587, (b) GalCer (d18:1/24:0) [M + Na]+m/z 918.773, and (c) LacCer (d18:1/16:0), [M + Na]+m/z 1010.748. For product ion assignments, see Figure S1c.

Gangliosides from Bovine Milk

Since gangliosides contain sialic acid, they efficiently form stable species in the negative ionization mode; this mode is particularly advantageous for the analysis of anionic glycosphingolipids. GM3 and GD3 have been reported to be the most abundant gangliosides in bovine milk.8 We analyzed bovine GM3 and GD3 from a commercial source and found that both contain mixtures of many homologues of GM3 and GD3; although the distribution of homologues varied among samples, we detected similar ganglioside mixtures in our extractions from bovine milk. Consistent with previous reports,8,23 the negative-ion mode MS spectra of bovine GM3 and GD3 included molecular ion species corresponding to gangliosides having ceramides with both odd and even numbers of carbon atoms. The peaks corresponding to the [M – 2H]2– and [M – H]− species in the ESI-MS spectra recorded for both GM3 and GD3, as illustrated in Figures 4a and 5a, indicated that the most abundant ceramide species were d39:1, d40:1, d41:1, and d42:1. For the longer-chain species, we noted an increase in the contribution from components containing a second site of unsaturation. Particularly for the shorter-chain gangliosides, we also observed peaks corresponding to gangliosides containing ceramides that had fully saturated alkyl chains in both LCB and FA. Accurate mass measurements recorded on the tribrid orbitrap instrument verified that these peaks corresponded to components with a saturated LCB rather than hydroxylated forms of the next lower homologue; the mass difference for these alternative assignments corresponds to the value for O (15.994915 u) vs CH4 (16.031300 u) or 0.036385 u, a distinction that can be clearly made using data recorded at high resolution/high mass accuracy. This is the first report of bovine milk gangliosides containing saturated LCB.

Figure 4.

Figure 4

Open in a new tab

ESI-QTOF MS and MS/MS spectra recorded for native GM3 derived from bovine milk. (a) [M – H]− region in the negative-ion mode MS spectrum. (b) Negative-ion mode CID MS/MS of the [M – H]−m/z 1235.798 assigned as GM3 (d40:1). (c) [M + Na]+ region in the positive-ion mode MS1 spectrum, obtained after permethylation of the sample. (d) CID MS/MS for the [M + Na]+m/z 1456.017 of permethylated GM3(d40:1). For product ion assignments, see Figure S1d,e.

Figure 5.

Figure 5

Open in a new tab

ESI-QTOF MS and MS/MS spectra recorded for native GD3 derived from bovine milk (a) the [M – 2H]2– region in the negative-ion MS of native GD3 from bovine milk. (b) Low m/z region in the CID MS/MS spectrum of the [M – 2H]2–m/z 769.949, assigned as GD3 (d41:1). (c) [M + Na]+ region in the positive-ion mode MS1 spectrum, obtained after permethylation of the sample. (d) Low m/z region in the CID MS/MS spectrum of [M + Na]+m/z 1831.207 of permethylated GD3(d41:1). For product ion assignments, see Figure S1f,g.

Although multiple groups have reported the characterization and quantification of GM3 and GD3 in bovine milk, these reports all listed the total composition of the total ceramide rather than individual values for the carbon numbers and degree(s) of unsaturation of the LCB and FA.8,11,24,25 A limited number of studies have determined the range of LCB and FA groups present in bovine milk. However, these values were obtained by first releasing the glycan headgroup and then analyzing the fatty acyl and LCB moieties separately.23,26,27 As noted above, analysis of the intact molecules should be more informative because it provides the means to obtain the most biologically relevant information about the combinations of LCB and FA groups present in a given sample. The UVPD MS/MS fragmentation of a few abundant molecular species of GM3 derived from bovine milk has been reported,22 but, to the best of our knowledge, there has not been a comprehensive analysis, by tandem MS or other analytical methods, of the diversity of ceramide compositions (in terms of LCB and FA) present in the individual intact molecular species of bovine or human milk gangliosides.

It is evident from the previous reports cited herein that the glycan headgroup of gangliosides in milk can be reliably determined on the basis of established MS/MS methodologies. In the present study, we therefore placed our focus on obtaining and assigning diagnostic ceramide fragments from the intact gangliosides. We confirmed that, at lower collision energies, the fragments generated by glycosidic cleavages within the glycan headgroup can be easily detected and indicate the glycan topology (see Supporting Information Figure S1 for glycan product ion nomenclature), but this outcome is incomplete since no products of fragmentation within the ceramide could be observed for CID performed at ≤60 eV. For example, in the CID MS/MS spectrum of GM3 (d34:1), shown in Supporting Information Figure S2a, recorded at 60 eV collision energy, Y0, Y1, and Y2 products resulting from glycosidic cleavages are observed, but there are not ceramide fragments in the corresponding low mass regions shown in Supporting Information Figure S2d. As the collision energy is increased, very low mass fragments originating via extensive disruption of the glycan headgroup remain, but ceramide fragments can be clearly observed, as shown in Supporting Information Figure S2b,e and S2c/f. The spectrum recorded at 80 eV collision energy includes the P(d18:1) m/z 237.221 and R(d18:1) m/z 263.236 fragments from the LCB and the S(16:0) m/z 280.263 and U(16:0) m/z 254.247 fragments from the FA moiety. The spectrum recorded at 110 eV also includes the fatty acyloxy fragment RCOO–(16:0) that is the product of a rearrangement, as noted above.

With this more aggressive approach, we have determined that, in some cases, gangliosides with isomeric ceramides are present. Figure 4a displays the [M – H]− region in the MS1 spectrum obtained for native bovine GM3, and Figure 4c shows the [M + Na]+ region in the MS2 spectrum of the permethylated sample. The low mass region of the MS2 spectrum obtained upon dissociation of [M – H]−m/z 1235.798, the peak assigned to an abundant homologue of GM3 (d40:1), is shown in Figure 4b; it contains fragments that arise via cleavage within the ceramide. This spectrum includes P(d18:1) m/z 237.220 and R(d18:1) m/z 263.237, originating from the LCB. FA product ions S(22:0) at m/z 364.357, U(22:0) at m/z 338.343, and V(22:0) at m/z 321.315 are generated from the dominant component, while additional FA fragment ions S(23:0) at m/z 378.373, S(24:0) at m/z 392.387, and V(24:0) at m/z 349.348 are also present. These assignments are supported by the positive-ion MS2 spectrum of the corresponding peak at m/z 1456.017 in the MS1 spectrum of the permethylated sample, labeled as GM3(40:1) in Figure 4c. In the MS2 spectral segment shown in Figure 4d, generated via CID of the peak labeled as GM3(d40:1) m/z 1235.798 in Figure 4c, the signal assigned to W′(d18:1) at m/z 278.284, has the highest intensity, followed by W(d18:1) m/z 310.310, W′(d16:1) m/z 250.253, W′(d17:1) m/z 264.268, W(d16:1) m/z 282.279, and W(d17:1) m/z 296.293, the last of which are observed at much lower intensities. These W and W′ ions recorded for GM3(d40:1) confirm that GM3(d18:1/22:0) is the most abundant isomeric species and GM3(d16:1/24:0) and GM3(d17:1/23:0) are the minor species. When previously reported analyses of the gangliosides used LC–MS/MS, the negative-ion CID measurements were performed at a single optimized collision energy. The data that we present here, obtained by direct infusion, utilizing multiple collision energies, and combining information from multiple spectra, demonstrate that employing different collision energies and a longer observation time has the potential to provide more complete ceramide composition information. This finding encourages the development of LC–MS conditions that allow the acquisition of more MS/MS spectra for each molecular ion value. When achieved at the higher CID voltage settings, fragmentation within the ceramide seems to be independent of the glycan; quite similar MS/MS spectra were obtained for the CID of the GD3 species assigned to the d40:1 homologue. These results are provided in Supporting Information Figure S3.

GD3 ceramide fragments in the MS/MS spectra can be assigned in a manner similar to those found for the GM3 species. For disialogangliosides, the presence of a second carboxyl group supports efficient formation of both [M – H]− and [M – 2H]2– species. The [M – 2H]2– region of the MS1 spectrum of commercial GD3 is shown in Figure 5a. For the MS2 spectrum shown in Figure 5b, CID of GD3(d41:1) [M – 2H]2–m/z 769.949, the ceramide fragment P(d18:1) m/z 237.220, originates from the dominant LCB, while the abundant product ion S(23:0) m/z 378.372 originates from the complementary FA chain. Much lower abundance of FA-specific products S(22:0) at m/z 364.356 and S(24:0) at m/z 392.386 indicate the presence of minor amounts of GD3(d17:1/24:0) and GD3(d19:1/22:0).

The [M + Na]+ molecular ion region of the MS1 spectrum obtained for permethylated bovine GD3 is shown in Figure 5c. Product ions present in the low mass region of the MS/MS spectrum shown in Figure 5d, which was recorded for the peak labeled as GD3(d41:1) [M + Na]+m/z 1831.207, are in agreement with the assignments made on the basis of the negative-ion MS/MS spectrum of this component. Abundant W(d18:0) and W′(d18:0) product ions can be observed at m/z 310.310 and m/z 278.284, respectively, in this spectrum; these fragments indicate that the major component is GD3(d18:1/23:0). Low abundance product ions corresponding to W′(d17:1) at m/z 264.270 and W′(d19:1) at m/z 292.302 indicate the presence of minor amounts of the molecular species GD3 (d17:1/24:0) and GD3(d19:1/22:0). These results mirror the MS2 spectra provided in Supporting Information Figure S4, obtained for CID of the MS1 peaks corresponding to native and permethylated GM3(d41:1). Low-abundance isomers, detected during the analyses of both GM3 and GD3, could inadvertently be overlooked, yet there remains the possibility that such components could have biological activity; therefore, it is important that the analytical method has a wide dynamic range for detection. The assignments for the isomers of GD3(d41:1) in the bovine milk sample were further validated via an MS4 analysis of the peak corresponding to the [M – 2H]2– of the native molecular species, observed at m/z 769.9522 in the MS1 mass spectra shown in Figure 6a–d that were measured with the Thermo Fisher Fusion Lumos Tribrid Orbitrap instrument. The Y1 product ion at m/z 796.6665 in the MS2 spectrum shown in Figure 6b was selected for MS3, and the MS3 product ion Y0 (m/z 634.6135 in Figure 6c) was dissociated to produce the MS4 spectrum shown in Figure 6d. The MS4 spectrum contained diagnostic peaks that could be assigned as P(d18:1) m/z 237.2223, R(d18:1) m/z 263.2379, S(23:0) m/z 378.3736, T(23:0) m/z 394.3685, U(23:0) m/z 352.3580, and V(23:0) m/z 335.3315, thereby confirming that GD3(d18:1/23:0) is the major isomer. Very low abundance fatty acyl fragments S(22:0) m/z 364.3579 and S(24:0) m/z 392.3891 were also observed, in keeping with the QTOF MS data. Additional MS4 results are shown in Supporting Information Figure S5.

Figure 6.

Figure 6

Open in a new tab

ESI-Tribrid Orbitrap mass spectra recorded for native GD3 derived from bovine milk. (a) [M – 2H]2– region in the negative-ion MS1 spectrum. (b) CID MS2 spectrum recorded for the dissociation of m/z 769.9522(2-), assigned as GD3(d41:1). (c) CID MS3 spectrum recorded for dissociation of the MS2 product ion at m/z 796.6665(1-) assigned as Y1. (d) CID MS4 spectrum recorded for dissociation of the product ion at m/z 634.614(1-), assigned as Y0, i.e., corresponding to the unmodified ceramide (d41:1). CID took place in the linear ion trap; all spectra were measured in the orbitrap. For product ion assignments, see Figure S1h.

Gangliosides from Human Milk

As had been reported previously, we observed that even-numbered carbon chain lengths are dominant for ceramides of GM3 derived from human milk. Results presented in Figure 7 that were obtained for GM3 from a human milk sample indicate that the most abundant molecular species were GM3(d36:1), (d38:1), and (d40:1). For GM3(d36:1), the ceramide fragments S(18:0) m/z 308.292, T(18:0) m/z 324.291, and S(20:0) m/z 336.326 were observed, indicating the presence of the isomeric molecular species GM3 (d18:1/18:0) and GM3(d16:1/20:0). During analysis of the d36:1 molecular species of permethylated human GM3, [M + Na]+m/z 1399.950, the corresponding distribution of W and W′ ions was observed, with W′(d18:1) m/z 278.284 being the most abundant fragment ion and W′ (d16:1) m/z 250.251, W′(18:0) m/z 280.300, and W (d18:1) m/z 310.310 product ions being observed at much lower intensities. For GM3 (d36:1), GM3(d18:1/18:0) was the most abundant isomer; GM3(d18:0/18:1) and GM3(d16:1/20:0) were present at only threshold levels. GD3 levels are reported to be much lower in human milk than in bovine milk.8−11 To date, we have not encountered measurable levels of GD3 in the human milk samples provided in our ongoing study, but we anticipate that the application of our methodology during our planned survey of milk collected at various lactation time points will identify a number of samples containing this ganglioside.

Figure 7.

Figure 7

Open in a new tab

GM3 ESI-QTOF MS and MS/MS spectra. (a) [M – H]− region in the negative-ion MS of native GM3 from human milk. (b) CID MS/MS of [M – H]−m/z 1179.736 of GM3(d36:1). (c) [M + Na]+ region in the positive-ion mode MS of GM3 derived from human milk after permethylation (d) CID MS/MS for the [M + Na]+m/z 1399.950 of permethylated GM3(d36:1). For product ion assignments, see Figure S1i,j.

Comparison of Ceramide Compositions of Human and Bovine Milk LCB and FA Chains

The LCB distribution d18:1 > d16:1 > d18:0 was observed, in the same proportions, in the major GM3 molecular species of human milk, whereas the LCB and FA in bovine milk showed much higher variability with respect to both the proportions and chain lengths (Table 1). The most abundant LCB values observed for bovine milk-derived GM3 were d16:1 and d18:1. The most abundant FA chains for GM3 derived from human milk were C20:0 and C22:0; C24:1 was the most abundant unsaturated FA moiety. For bovine milk, the GM3 FA groups were almost exclusively saturated, the most abundant being C22:0 and C23:0. Odd-chain-length FAs were highly represented. N-Acetyl neuraminic acid was determined to be the dominant form of sialic acid in the milk from both sources; no signals corresponding to the presence of N-glycolyl neuraminic acid were detected in the commercial bovine gangliosides or the extracts that we prepared from either bovine milk or human milk.

Table 1. LCB and FAs Found in Milk Gangliosides.

bovine milk gangliosides
GM3 isomer 1 isomer 2 isomer 3
d34:1 d18:1/16:0    
d36:1 d18:1/18:0 d16:1/20:0  
d38:1 d16:1/22:0 ad18:1/20:0  
d39:1 d16:1/23:0 d17:1/22:0 ad18:1/21:0
d40:1 d18:1/22:0 d16:1/24:0 d17:1/23:0
d41:1 d18:1/23:0 d17:1/24:0 d19:1/22:0
d42:1 d18:1/24:0 d19:1/23:0  
d43:1 d19:1/24:0 d20:1/23:0 ad18:1/25:0
GD3 isomer 1 isomer 2 isomer 3
d34:1 d18:1/16:0    
d36:1 d18:1/18:0 d16:1/20:0  
d38:1 d16:1/22:0    
d39:1 d16:1/23:0 d17:1/22:0 ad18:1/21:0
d40:1 d18:1/22:0 d16:1/24:0 d17:1/23:0
d41:1 d18:1/23:0 d17:1/24:0 d19:1/22:0
d42:1 d18:1/24:0 d19:1/23:0  
d43:1 d19:1/24:0 d20:1/23:0 d18:1/25:0
human mi1k gangliosides
GM3 isomer 1 isomer 2 isomer 3
d34:1 d18:1/16:0 d16:1/18:0 d18:0/16:1
d36:1 d18:1/18:0 d16:1/20:0 d18:0/18:1
d38:1 d18:1/20:0 d16:1/22:0 d18:0/20:1
d39:1 d18:1/21:0 d17:1/22:0  
d40:1 d18:1/22:0 d16:1/24:0 d18:0/22:1
d42:2 d18:1/24:1    
Open in a new tab
a

Isomers detected only in the mass spectra of permethylated derivatives.

Conclusions

This work addresses the need for determining the ceramide compositions of the gangliosides present in bovine and human milk. Significant differences are observed in terms of both the LCB and the FA chains in bovine and human milk gangliosides between the species and within the molecular weight distributions of each. We demonstrate here the capability to determine the detailed ceramide compositions via complementary negative-ion mode and positive-ion mode CID MS/MS analyses of bovine and human milk. Subsequent studies will examine the individual-specific differences and changes during the lactation period. Additionally, it will be important to couple these MS approaches to online liquid chromatography to facilitate the detection of low-abundance molecular species and the quantification of the different isomeric species. These will be the next steps in the research project. Changes in ceramide composition are likely to have biological effects since the presentation of the glycan headgroup, which acts as a receptor, is affected by changes in the ceramide structure, and thus, the functions of individual ganglioside species depend on the structures of both glycan and ceramide. Biological studies of the roles of gangliosides in bovine milk and human milk (and milks from other sources) that are supported by advances in analytics will improve our understanding of the physiological ramifications of ceramide structural differences.

Acknowledgments

The authors thank Peter Chen for the careful review of all assignments and the manuscript draft. We also thank Kelly Dingess for her assistance with the preparation of the final copy. This research is supported by NIH grants R24 GM134210, S10 OD010724, and S10 OD021728 and a cooperative research agreement between Boston University and Danone Nutricia Research. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies. The presentation of experimental details complies with MIRAGE guidelines.28

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.3c00737.

  • Structures of GM3 and GD3 gangliosides showing desigations for glycosidic cleavages at C1 of each glycan residue and ESI-QTOF CID tandem mass spectra of additional molecular species of GD3 (from bovine milk) and GM3 (from bovine and human milks) (PDF)

Author Present Address

∥ Amgen, Cambridge, MA 02142 United States

Author Contributions

O.T.L. isolated gangliosides from bovine and human milk, permethylated the standards and samples, and performed the MS and tandem MS analyses. C.X. provided assistance with analyses performed on the Fusion Lumos Tribrid Orbitrap MS. S.R. and B.S. selected and prepared milk samples and provided background on milk science. C.E.C. designed and guided the study and provided MS instrumentation and expertise. The manuscript was written by O.T.L. and C.E.C. with contributions from all authors. All authors have given approval to the final version of the manuscript.

The authors declare the following competing financial interest(s): This project has received funding from a cooperative research agreement between Boston University and Danone Nutricia Research.

Supplementary Material

ac3c00737_si_001.pdf (2.3MB, pdf)

References

  1. Hakomori S. Structure, organization, and function of glycosphingolipids in membrane. Curr. Opin. Hematol. 2003, 10, 16–24. 10.1097/00062752-200301000-00004. [DOI] [PubMed] [Google Scholar]
  2. Ledeen R. W.; Wu G. Ganglioside Function in Calcium Homeostasis and Signaling. Neurochem. Res. 2002, 27, 637–647. 10.1023/A:1020224016830. [DOI] [PubMed] [Google Scholar]
  3. Yu R. K.; Tsai Y. T.; Ariga T.; Yanagisawa M. Structures, Biosynthesis, and Functions of Gangliosides–an Overview. J. Oleo Sci. 2011, 60, 537–544. 10.5650/jos.60.537. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Sonnino S.; Mauri L.; Chigorno V.; Prinetti A. Gangliosides as Components of Lipid Membrane Domains. Glycobiology 2007, 17, 1R–13R. 10.1093/glycob/cwl052. [DOI] [PubMed] [Google Scholar]
  5. Sipione S.; Monyror J.; Galleguillos D.; Steinberg N.; Kadam V. Gangliosides in the Brain: Physiology, Pathophysiology and Therapeutic Applications. Front. Neurosci. 2020, 14 (1004), 572965. 10.3389/fnins.2020.572965. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Svennerholm L. The Gangliosides. J. Lipid Res. 1964, 5, 145–155. 10.1016/S0022-2275(20)40231-7. [DOI] [PubMed] [Google Scholar]
  7. Giuffrida F.; Elmelegy I. M.; Thakkar S. K.; Marmet C.; Destaillats F. Longitudinal Evolution of the Concentration of Gangliosides GM3 and GD3 in Human Milk. Lipids 2014, 49, 997–1004. 10.1007/s11745-014-3943-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Lee H.; An H. J.; Lerno Jr L. A.; German J. B.; Lebrilla C. B. Rapid Profiling of Bovine and Human Milk Gangliosides by Matrix-Assisted Laser Desorption/Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Int. J. Mass Spectrom. 2011, 305, 138–150. 10.1016/j.ijms.2010.10.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Peterson R.; Cheah W. Y.; Grinyer J.; Packer N. Glycoconjugates in Human Milk: Protecting Infants from Disease. Glycobiology 2013, 23, 1425–1438. 10.1093/glycob/cwt072. [DOI] [PubMed] [Google Scholar]
  10. Ali A. H.; Wei W.; Wang X. A Review of Milk Gangliosides: Occurrence, Biosynthesis, Identification, and Nutritional and Functional Significance. Int. J. Dairy Technol. 2022, 75, 21–45. 10.1111/1471-0307.12816. [DOI] [Google Scholar]
  11. Liu Z.; Li C.; Pryce J.; Rochfort S. Comprehensive Characterization of Bovine Milk Lipids: Phospholipids, Sphingolipids, Glycolipids, and Ceramides. J. Agric. Food Chem. 2020, 68, 6726–6738. 10.1021/acs.jafc.0c01604. [DOI] [PubMed] [Google Scholar]
  12. Domon B.; Costello C. E. Structure Elucidation of Glycosphingolipids and Gangliosides Using High-Performance Tandem Mass Spectrometry. Biochemistry 1988, 27, 1534–1543. 10.1021/bi00405a021. [DOI] [PubMed] [Google Scholar]
  13. Lee M. H.; Lee G. H.; Yoo J. S. Analysis of Ceramides in Cosmetics by Reversed-Phase Liquid Chromatography/Electrospray Ionization Mass Spectrometry with Collision-Induced Dissociation. Rapid Commun. Mass Spectrom. 2003, 17, 64–75. 10.1002/rcm.878. [DOI] [PubMed] [Google Scholar]
  14. Hsu F. F.; Turk J. Characterization of Ceramides by Low Energy Collisional-Activated Dissociation Tandem Mass Spectrometry with Negative-Ion Electrospray Ionization. J. Am. Soc. Mass Spectrom. 2002, 13, 558–570. 10.1016/S1044-0305(02)00358-6. [DOI] [PubMed] [Google Scholar]
  15. Hsu F. F. Complete Structural Characterization of Ceramides as [M-H](−) Ions by Multiple-Stage Linear Ion Trap Mass Spectrometry. Biochimie 2016, 130, 63–75. 10.1016/j.biochi.2016.07.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Ciucanu I.; Kerek F. A Simple and Rapid Method for the Permethylation of Carbohydrates. Carbohydr. Res. 1984, 131, 209–217. 10.1016/0008-6215(84)85242-8. [DOI] [Google Scholar]
  17. Ciucanu I.; Costello C. E. Elimination of Oxidative Degradation During the Per-O-Methylation of Carbohydrates. J. Am. Chem. Soc. 2003, 125, 16213–16219. 10.1021/ja035660t. [DOI] [PubMed] [Google Scholar]
  18. Gunnarsson A. N-andO-alkylation of glycoconjugates and polysaccharides by solid base in dimethyl sulphoxide/alkyl lodide. Glycoconjugate J. 1987, 4, 239–245. 10.1007/bf01048429. [DOI] [Google Scholar]
  19. Svennerholm L.; Fredman P. A Procedure for the Quantitative Isolation of Brain Gangliosides. Biochim. Biophys. Acta 1980, 617, 97–109. 10.1016/0005-2760(80)90227-1. [DOI] [PubMed] [Google Scholar]
  20. Fong B.; Norris C.; Lowe E.; McJarrow P. Liquid Chromatography–High-Resolution Mass Spectrometry for Quantitative Analysis of Gangliosides. Lipids 2009, 44, 867–874. 10.1007/s11745-009-3327-1. [DOI] [PubMed] [Google Scholar]
  21. Adams J.; Ann Q. Structure Determination of Sphingolipids by Mass Spectrometry. Mass Spectrom. Rev. 1993, 12, 51–85. 10.1002/mas.1280120103. [DOI] [Google Scholar]
  22. O’Brien J. P.; Brodbelt J. S. Structural Characterization of Gangliosides and Glycolipids Via Ultraviolet Photodissociation Mass Spectrometry. Anal. Chem. 2013, 85, 10399–10407. 10.1021/ac402379y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Bode L.; Beermann C.; Mank M.; Kohn G.; Boehm G. Human and Bovine Milk Gangliosides Differ in Their Fatty Acid Composition. J. Nutr. 2004, 134, 3016–3020. 10.1093/jn/134.11.3016. [DOI] [PubMed] [Google Scholar]
  24. Sørensen L. K. A Liquid Chromatography/Tandem Mass Spectrometric Approach for the Determination of Gangliosides GD3 and GM3 in Bovine Milk and Infant Formulae. Rapid Commun. Mass Spectrom. 2006, 20, 3625–3633. 10.1002/rcm.2775. [DOI] [PubMed] [Google Scholar]
  25. Lee H.; German J. B.; Kjelden R.; Lebrilla C. B.; Barile D. Quantitative Analysis of Gangliosides in Bovine Milk and Colostrum-Based Dairy Products by Ultrahigh Performance LC-Tandem Mass Spectrometry. J. Agric. Food Chem. 2013, 61, 9689–9696. 10.1021/jf402255g. [DOI] [PubMed] [Google Scholar]
  26. Chakraberty R.; Reiz B.; Cairo C. W. Profiling of Glycosphingolipids with SCDase Digestion and HPLC-FLD-MS. Anal. Biochem. 2021, 631, 114361. 10.1016/j.ab.2021.114361. [DOI] [PubMed] [Google Scholar]
  27. Martín M.-J.; Martín-Sosa S.; Hueso P. Bovine Milk Gangliosides: Changes in Ceramide Moiety with Stage of Lactation. Lipids 2001, 36, 291–298. 10.1007/s11745-001-0720-x. [DOI] [PubMed] [Google Scholar]
  28. Struwe W. B.; Agravat S.; Aoki-Kinoshita K. F.; Campbell M. P.; Costello C. E.; Dell A.; Feizi T.; Haslam S. M.; Karlsson N. G.; Khoo K.-H.; Kolarich D.; Liu Y.; McBride R.; Novotny M. V.; Packer N. H.; Paulson J. C.; Rapp E.; Ranzinger R.; Rudd P. M.; Smith D. F.; Tiemeyer M.; Wells L.; York W. S.; Zaia J.; Kettner C. The Minimum Information Required for a Glycomics Experiment (MIRAGE) Project: Sample Preparation Guidelines for Reliable Reporting of Glycomics Datasets. Glycobiology 2016, 26, 907–910. 10.1093/glycob/cww082. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ac3c00737_si_001.pdf (2.3MB, pdf)

Articles from Analytical Chemistry are provided here courtesy of American Chemical Society

RESOURCES