Antigen-Specific Ganglioside Serological Profile of Pancreatic and Gastric Cancer Patients by Multiple TLC Overlay Assay and IR-MALDI Mass Spectrometry

Simple Summary

The serological CD75s, CD15s, and iso-CD75s antigens in glycolipids play a significant role in pancreatic and gastric cancer and also differentiate the two cancer types. We evidenced that in both cancers, these antigens are preferentially and significantly expressed on long-chain neolacto-species, pointing toward specific glycosylation processing in immune evasion in these cancer types. The new findings of this study are enabled by the technology we developed here for serological antigen glycolipid profiling, which combines efficient micro-extraction of long-chain sialylated glycolipids with overlay antibody/lectin thin-layer chromatography staining and on-plate matrix-assisted laser desorption mass spectrometry analysis of specific antigen-containing glycolipids. This technology enabled us to explore, for the first time in cancer sera, the long-chain sialylated glycolipids of neolacto-series, which are primarily involved in peripheral immunity, along with ganglio-series, while also robustly exploring and differentiating pancreatic cancer and gastric cancer, including the different developmental stages, in a cohort of patient samples.

Abstract

Background: Altered glycosphingolipidome in cancerous tissues and cells reflects the circulatory glycosphingolipid (GSL) profiles, which is advantageous for establishing cancer biomarkers and/or unravelling GSL-associated mechanisms of immunity in cancer. Methods: Here, we combined a microscale extraction of GSLs with multiple overlay TLC assays and IR-MALDI-o-TOF MS and implemented it for the first time in serum analysis of CD75s-, CD15s-, and iso-CD75s-containing sialylated GSLs of ganglio- and neolacto-series. Results: This sensitive antigen-specific targeted GSL workflow enabled the identification of 80 sialylated GSLs containing the specific antigens in human sera and was applied for the investigation of clinical serum samples from gastric/stomach cancer patients (n = 40), pancreatic cancer patients (n = 40), and a cancer-free control group (n = 20). The CD75s-, CD15s-, and iso-CD75s-containing GSL series encompassing complex monosialylated and fucosylated GSLs of neolacto-series, with up to pentadecasaccharide chains, were detected in both cancer types, while differential semi-quantitative analysis indicates a tumor type-specific associated GSL profile. Both cancer types share a drop in the complex fucosylated neolacto-gangliosides during tumor progression, implying a decreased synthesis of long-chain neolacto-series. Conclusions: This drop suggesting a role of these highly polar complex ganglioside species in evading humoral tumor immune response in the early tumor stages.

1. Introduction

Pancreatic cancer (PC) is the third leading cause of cancer mortality in the EU, with a 3% rise in women and stable predictions in men in the year 2022, though other cancer types show positive trends. PC tops amongst “deathly” cancers, with a mortality/incidence ratio of 98% and a 5-year survival rate barely reaching 10% [1]. Another “silent killer”, gastric or stomach cancer (GC), ranks as the fourth leading cause of cancer death worldwide [1,2]. Although the incidence rates of GC have been declining in most countries, an unexpected increase in incidence in younger people (<45 years old) has been seen in high-income countries [3]. Unfortunately, over 80% of PC and GC patients have a late onset of mostly non-specific symptoms and are diagnosed to be in the advanced/unresectable disease stage [4]. Hence, the discovery, verification, and translation into clinical management of new early detection biomarkers is imperative to increase PC and GC patientsʼ survival time and rates [5]. Elucidation of the cancer-specific alterations/reprogramming of metabolic and signaling pathways at different stages of GC/PC progression is necessary to expedite the development of new therapeutic strategies [6]. Over 300 potential PC biomarkers of different molecular classes have been published, including serum carbohydrate antigens, (glyco)proteins, RNAs, circulating free DNA, circulating tumor cells, and exosomes. But the sought-after sensitivity and specificity have not been reached yet [4].

Glycosphingolipids (GSLs) are ubiquitous components of mammalian cell membranes, with remarkable structural diversity, consisting of a hydrophobic ceramide of different lengths, saturations, and hydroxylation degrees anchored in the outer membrane leaflet and a hydrophilic glycan chain of various structural architectures protruding toward the extracellular space. GSLs can be neutral or acidic, with the latter carrying one or more sialic acid (Sia/Neu5Ac) residues, commonly known as gangliosides, and carrying terminal antigenic structures serving as immune mediators and/or determinants for cell–cell communication. GSLs have a bioactive role in modulating crucial cellular functions, including cell growth, adhesion/motility, apoptotic death or survival, and cell–cell recognition and interaction [7]. Their cellular expression pattern is highly variable and cell-type-specific due to the interplay of cell-specific glycosyltransferases and glycosidases expression and function [8]. Aberrantly expressed GSLs have been a hallmark of oncogenic transformation and reprogramming, including promoting oncogenesis [9], tumor growth rate and invasiveness [10], metastasis [11], and the cancer cell signaling network [12]. This high GSL expression on the cell surface of cancer cells masks and hence hinders recognition by immune cells [13]. GSLs are secreted or shed by various tumors into the bloodstream in free soluble form or packed in vesicles (exosomes, lipoproteins, micelles), enabling investigation of cancer GSL fingerprints [13]. For example, elevated total serum levels of gangliosides have been reported in PC patients and in PC tissues, elevated GM3, and an altered sialylated neolacto (nLc) profile associated with tumor progression. Elevated GM3 levels were also evidenced in the plasma of patients with a progressive stage of GC [14,15,16]. We previously showed, using nanoESI-QTOF-MS and nanoLC/QTOF MS, a distinctive ceramide pattern of GM3 in PC and GC [17].

Therefore, it stands to reason that extended coverage of the glycosphingolipidome in cancers could be leveraged for early biomarker and therapeutic target discovery. Moreover, unraveling the expression of specific GSL immuno-epitopes in cancers can give unique insights into the immune processes mediating different cancer types. In this context, it is relevant that dysregulation of fucosylated epitopes and CD15s gangliosides in surgical specimens of GC and PC tissues has been reported, and the extent of cellular α2,3/α2,6-sialylated antigens (CD75s and iso-CD75s) was evidenced as a hallmark of neoplastic growth and metastasis [18,19,20,21]. The GSL analysis is particularly challenged by their high structural diversity in biological matrices, due to the high level of linkage- and monosaccharide unit-derived isomers with subsequent functionally distinct biological roles [22]. Alone, the extraction of GSLs and specific epitope-containing sub-glycosphingolipidome requires days to weeks of multi-step procedures for subsequent structural and quantitative assessment by mass spectrometry (MS) or other bioanalytical tools. We and others developed several MS protocols for serum, cellular, and tissue GSL profiling. A broad coverage of ganglio- and nLc series in serum/plasma is rarely reported. We previously profiled serum GSLs using nanoESI-QTOF MS and nanoLC/QTOF MS. More recently, using four-dimensional (4D) glycolipidomics analysis, we elucidated up to 376 serum GSLs [17,23].

Here, we investigated, in PC and GC cohorts, the serological profile of the complex, sialylated GSLs containing the immune glycan epitopes, also known as clusters of differentiation (CD): (i) CD75s: Neu5Acα2,6Galβ1,4GlcNAc; (ii) CD15s, also known as sialyl-Lewis X (sLe^x) determinant, that is, Neu5Acα2,3Galβ1,4(Fucα1,3)GlcNAc; and (iii) iso-CD75s: Neu5Acα2,3Galβ1,4GlcNAc.

We combined, for the first time to our knowledge, microscale extraction of serum GSL with multiple antibody/lectin overlay TLC assay and IR-MALDI-o-TOF MS for serological profiling of specific epitope-containing GSLs of ganglio- and neolacto-series (nLc). We unraveled the distinctive changes in the expression patterns of CD75s-, iso-CD75s-, and CD15s-containing GSLs in the serum of patients with PC and GC versus cancer-free controls (Ctrl) and the discriminative GSL pattern of PC versus GC. This study lays the foundation for expanding the elucidation of the immune metabolism of specific antigenic lipids in cancer.

2. Materials and Methods

2.1. Materials

All solvents for the lipid extraction and MS analysis were of HPLC grade. Methanol, water, and chloroform were purchased from Sigma-Aldrich (St. Louis, MO, USA) and Merck (Darmstadt, Germany). Water of high purity (MilliQ, Synergy UV, Millipore, Billerica, MA, USA) was used for the buffer preparation. In-house distilled chloroform and methanol were used for the TLC analysis. Pooled human serum from male AB blood (Sigma-Aldrich, St. Louis, MO, USA) was used for the method development and evaluation.

2.2. Clinical Samples

Preoperative sera of patients suffering from GC (n = 40) and PC (n = 40), along with Ctrl sera from tumor-free individuals (n = 20), were procured from RNTech (Paris, France) and RNTech SRL (Bucharest, Romania).

2.3. Reference Gangliosides

As reference GSL mixture and positive Ctrl for TLC and MS structural assignment, a polar ganglioside fraction derived from human granulocytes, with well-known ganglioside composition, containing mainly neolacto (nLc)-series gangliosides with up to nLc14 glycan chain, was used [24,25].

2.4. GSL Extraction

Serum GSL extraction was performed following an earlier protocol [17]. Serum aliquots of 300 µL were lyophilized in glass vials. The lyophilizate was extracted two times with 1.3 mL chloroform/methanol/water (30/60/8, v/v/v), followed by mixing, supersonic treatment for 30 min, and centrifugation for 4 min at maximum speed (Hettich universal, rotor 1323, Bäch, Switzerland). The supernatants were pooled and dried under nitrogen. The samples were reconstituted in 600 µL digest buffer containing 10 mM Tris buffer, with a pH of 7.5 and 10 mM CaCl₂, and the mixture was emulsified by mild supersonic treatment, followed by addition of 2.5 U Bacillus cereus phospholipase C (PLC) (Sigma Aldrich, Steinheim, Germany). Samples were digested for 16 h in a water bath shaker at 37 °C. Then, a two-step solid-phase extraction was applied: reverse-phase (RP) chromatography on C18 Sep-Pak columns and normal-phase (NP) fractionation on silica columns. PLC digests were desalted and fractionated in polar and neutral lipids on C18 Sep-Pak. A total of 100 mg of C18 material was packed in the glass cartridge, equilibrated with 6 column volumes (CV) of chloroform/methanol (2:1, v/v), methanol, and high-purity water; the salty digest was then directly loaded, and the flow-through fraction was reapplied twice. The column was washed with 6 CV of the high-purity water, and the polar lipid fraction (containing GSLs) was eluted using 3 CV of neat methanol and then evaporated under nitrogen. The dried eluate was reconstituted in 600 µL chloroform, applied on a glass cartridge packed with silica beads (100 mg Iatrobeads per column), and pre-conditioned with 6 CV of chloroform, chloroform/methanol (1:1), methanol, and chloroform, respectively. The cartridges were washed with 3 CV of chloroform and chloroform/methanol (9/1, v/v), followed by elution in four steps using (i) 3 CV of chloroform/methanol (2/1, v/v), (ii) 3 CV of chloroform/methanol (1/2, v/v), (iii) 3 CV of chloroform/methanol (1/4, v/v), and (iv) a combined step of 3 CV methanol and 3 CV methanol/water (4/1). Collected fractions were evaporated and reconstituted in methanol. Fractions eluted in the third and fourth steps, predominantly containing highly polar gangliosides, were further used for the following analyses: TLC and direct infusion nanoESI-QTOF MS/MS analysis.

2.5. Thin-Layer Chromatography (TLC)

Highly polar ganglioside mixtures eluted from the silica column were applied to glass-backed silica gel 60 precoated high-performance TLC plates (HPTLC plates, 10 × 10 cm², 0.2 mm thickness, No. 5633; Merck, Darmstadt, Germany) with an automatic applicator (Linomat IV, CAMAG, Muttenz, Switzerland) and separated in a solvent system composed of chloroform/methanol/water (50/47/14, each by volume) supplemented with 2 mM CaCl₂. The detailed experimental protocol for TLC can be found elsewhere [26]. The run time was invariably fixed to 25 min. Gangliosides on TLC plates were detected by the multiple overlay assay.

2.6. Multiple Antibody/Lectin Overlay TLC Assay

For the overlay assay, we combined immunostaining using antibodies specific to glycan epitopes, CD15s and CD75s, with lectin assay specific to the glycan epitope iso-CD75s. Serum gangliosides were detected using a sequential multiple overlay assay protocol [26,27], adapted here for the ganglioside analysis of human serum. Three types of gangliosides were detected on a single TLC plate by consecutive overlay assays: (i) α2,6-sialylated gangliosides with CD75s (Neu5Acα2,6Galβ1,4GlcNAc) epitope, (ii) CD15s (Neu5Acα2,3Galβ1,4(Fucα1,3)GlcNAc), and (iii) iso-CD75s (Neu5Acα2,3Galβ1,4GlcNAc) containing species. To differentiate the individual ganglioside types, various substrate systems that gave differently colored precipitates were used for enzyme-amplified detection and visualization. To immunodetect CD75s gangliosides on TLC plates, the following procedure was applied: Firstly, the plates were fixed by chromatography with 0.5% polyisobutylmethacrylate (Plexigum P28, Röhm, Darmstadt, Germany) [26,27], then incubated overnight in phosphate-buffered saline (PBS) and blocked by dipping the plates in 1% bovine serum albumin in PBS (solution A) for 30 min. A primary polyclonal chicken anti-CD75s antibody was applied for 1 h. The plates were washed three times with PBS before a secondary, rabbit anti-chicken IgY antibody, conjugated with alkaline phosphatase (AP) (Dianova, Hamburg, Germany), was applied for 1 h. Both antibodies were diluted to 1:2000 in solution A. Following the washing step (three times with PBS), the plates were equilibrated in 200 mM Tris buffer, pH 9.1, and incubated in Fast Red substrate solution. The latter was obtained by mixing equal volumes of 2 mg/mL naphthol AS-MX phosphate (Serva Feinbiochemica GmbH & Co KG, Heidelberg, Germany) and 2 mg/mL Fast Red TR salt solution (Fluka Chemie GmbH, Buchs, Switzerland), both dissolved in 200 mM Tris, pH 9.1. The bands on TLC plates where AP-catalyzed reaction yielded a red precipitate indicated the presence of CD75s gangliosides.

CD15s gangliosides were immunodetected using monoclonal CSLEX-1 antibody (supernatant of a CSLEX-1 mouse hybridoma cell line culture, also commercially available as a mouse anti-CD15s/sLe^x monoclonal antibody, unconjugated, Clone CSLEX1, BD Pharmingen, San Diego, CA, USA) as the primary antibody and AP-conjugated anti-mouse IgM as the secondary antibody. Both antibodies were diluted to 1:1000 in solvent A. For the visualization of immune complexes, the plates were immersed in a 0.05% (w/v) solution of 5-bromo-4-chloro-3-indolylphosphate p-toluidine salt (BCIP; Roth, Karlsruhe, Germany) in glycine buffer (0.1 M glycine, 1 mM ZnCl₂, 1 mM MgCl₂, pH 10.4). The presence of CD15s gangliosides in TLC bands was visualized by developing a blue precipitate.

The presence of iso-CD75s gangliosides was interrogated by a lectin overlay assay. Following an overnight incubation of TLC plates in PBS, they were blocked with solution A for 30 min and equilibrated in buffer 1 with 0.1 M Tris (pH 7.5) containing 0.15 M NaCl, along with 1 mM MgCl₂, MnCl₂, and CaCl₂ for each. Biotinylated plant lectin—Maackia amurensis agglutinin (MAA or MAL) I (2 µg/mL in buffer 1)—was applied for 1 h, followed by 1 h incubation in streptavidin-AP solution (1 µg/mL in solution A). MAA I and streptavidin-AP conjugate were from Vector Laboratories, Burlingame, CA, USA. Starting with the MAA application, all incubation steps were followed by a washing step (3× with PBS). Finally, to visualize iso-CD75s ganglioside bands, the plates were first equilibrated in 0.1 M Tris buffer, pH 9.5, supplemented with 0.1 M NaCl and 0.05 M MgCl₂ and then incubated with NBT/BCIP substrate solution containing 75 µg/mL NBT and 37.5 µg/mL BCIP in the same buffer. The combined NBT/BCIP substrate system gives a grey/purple precipitate.

The color development was stopped for each overlay assay when the background staining became apparent. The plates were washed with PBS and air-dried before the densitometry (see below). After each overlay assay, an inactivation step in the form of heat treatment at 140 °C for 30 min in an oven (type ULE 500, Memmert GmbH, Schwabach, Germany) was performed [27].

2.7. Densitometry

After each overlay assay, the plates were subjected to densitometric analysis using a CD60 TLC plate scanner (Desaga, Heidelberg, Germany, software ProQuantR, version 1.06.000). Fast Red- and BCIP-stained ganglioside bands were quantified by scanning at 520 nm and 650 nm, respectively. For the NBT/BCIP staining detection system, the output signal was obtained by subtracting scans at 610 nm recorded before the staining from those recorded after staining with NBT/BCIP. Data (Supplementary Data S1) were processed using Microsoft Excel and Origin 7.0 software. The peak area for each ganglioside was a measure of its expression level.

2.8. IR-MALDI-o-TOF MS

Gangliosides detected by the multiple overlay TLC assay were further analyzed by Infrared Matrix-Assisted Laser Desorption/Ionization orthogonal Time-of-Flight Mass Spectrometry (IR-MALDI-o-TOF MS) according to the previously described protocol [26,27]. Briefly, the plates were immersed in 10 mM ammonium acetate, pH 7, and left for 2 h, then air dried. Stained ganglioside spots were marked with a soft graphite pencil prior to the extraction of the fixative by repeated dipping in distilled chloroform. The plates were cut into pieces of a maximum of 30 × 40 mm and fixed onto a custom-made target plate using double-sided adhesive pads. Glycerol served as a liquid matrix for the IR-MALDI process and was spotted on stained lanes and soaked up by the porous silica gel [26]. MS analysis was performed on a modified o-TOF mass spectrometer (similar to proTOF, MDS Sciex, Concord, Ontario, Canada) equipped with an Er: YAG laser (Bioscope, BioOptics Laser Systems AG, Berlin, Germany) emitting at wavelength of 2.94 µm. The IR laser pulse duration was ~100 ns with a repetition rate of 2 Hz, and the spot size on target was ~250 nm. Measurements were performed in the negative ion mode with a TOF acceleration voltage of 10 kV and 1.2 mbar nitrogen pressure in the ion source. Generally, a mass resolution of ~10,000 and a mass accuracy of <25 ppm were achieved. The acquired mass spectra were evaluated using MoverZ software (version 2001.02.13, Genomic Solutions, Ann Arbor, MI, USA). The IR-MALDI-o-TOF MS data of the reference human granulocyte extract, used for annotation reference, is listed in Table S1.

2.9. NanoESI-QTOF MS

NanoESI-QTOF MS/MS experiments were performed to acquire a sensitive and comprehensive GSL profile in human sera of patients that can serve as a reference and complementary structural profiling to multiple overlay TLC assays. Highly polar ganglioside fractions eluted from the silica column were dissolved in 30 µL methanol at a concentration corresponding to 10 µL serum/µL and then pooled. They were analyzed by direct infusion nanoESI MS using a QTOF I mass spectrometer (Micromass, Manchester, UK). Using a custom-made pipette puller, capillaries were made in-house from borosilicate glass (Hilgenberg, Malsfeld, Germany). Collision-induced dissociation (CID) was performed using argon as a collision gas, set to 15 psi. For CID experiments, the selected precursors were accelerated with 40 to 120 V. Data were acquired using MassLynx software (version 3.5, Micromass, Manchester, UK).

2.10. Statistical Analysis

Statistical tests were performed using SPSS Statistics 17 software (SPSS Inc., Chicago, IL, USA). The measured ganglioside expression levels often were not normally distributed, so we employed a nonparametric Mann–Whitney U (MWU) test for the comparisons of two samples. Thus, by MWU, we compared the ganglioside expression levels: GC versus the Ctrl group, PC versus the Ctrl group, and GC versus PC. For the MWU test, the exact p-values were calculated. Kendall’s τ was calculated to test the correlation of parameters, such as the expression of individual gangliosides with the clinical (histopathological) parameters. All tests used were two-tailed, and the level of significance was set to τ = 0.05. Boxplots were generated using GraphPad Prism version 9.5.1 to visualize alterations in serum ganglioside levels associated with GC and PC. Ganglioside peak areas from densitometry (Supplementary Data S1) were scaled to unit variance prior to multivariate analysis. Principal component analysis (PCA) was performed in R (version 4.5.2), https://www.r-project.org accessed on 17 December 2025, and eigenvalues and variance explained by each principal component were extracted using the factoextra package (version 1.0.7). Supervised partial least squares–discriminant analysis (PLS-DA) was conducted using the ropls package (version 1.42.0). PCA and PLS-DA score plots (Supplementary Data S2 and S3) were generated using ggplot2 (version 4.0.1), and three-dimensional score plots were visualized using the rgl package (version 1.3.31) in R.

3. Results

3.1. Direct nanoESI-QTOF MS Screening of Serum Gangliosides

In this study, a previously developed microliter-scale serum GSL protocol (Supplementary Figure S1), amenable to long-chain serum GSLs, was successfully combined with a multiple overlay TLC assay and subsequent IR-MALDI-o-TOF MS profiling to increase the sensitivity of the epitope-specific GSL profiling [17]. Sialylated, long-chain GSLs were present in fractions 3 and 4 of the extraction procedure, with partial overlapping observed by nanoESI-QTOF MS (Figure 1). Although the normal-phase fractionation has substantially improved the GSL detection by direct infusion nanoESI MS profiling, it was not required for the more robust TLC analysis.

Figure 1.

(-) NanoESI-QTOF mass spectrum of human serum gangliosides from the pool of highly polar ganglioside fractions, obtained by the normal phase fractioning (Iatrobead silica column) of polar lipid extracts from human serum. A list of all detected ganglioside molecular ions is presented in Table 1.

Using direct infusion nanoESI-QTOF MS/MS, structural elucidation of gangliosides eluting in fractions 3 and 4 (Supplementary Figure S1) from sera of cancer patients and cancer-free Ctrls was achieved, which corroborates the gangliosidome compositional pattern subsequently obtained by multiple overlay TLC and IR-MALDI-o-TOF MS (Figure 2, Table 1). Here, 55 GSLs were identified (Table 1), containing GM3, GM2, GD3, GD1, and GT1 species of the ganglio-series and sialylated nLc4, nLc6, nLc8, and nLc10 of the nLc series exhibiting various ceramide compositions. As expected, GM3 species dominated the spectrum (Figure 1), showing the considerable heterogeneity of their fatty acyl chains (Table 1). MS/MS analysis indicated the presence of nLc-type gangliosides (Figure 2). They share the common core oligosaccharide structure, Galβ1,4GlcNAcβ1,3Galβ1,4Glc, known as nLc4, which can be extended by repeating the type 2 core structure, Galβ1,4GlcNAc, to nLc6 (Galβ1,4GlcNAcβ1,3Galβ1,4GlcNAcβ1,3Galβ1,4Glc), nLc8, nLc10, nLc12, etc., species (Table 1). Aside from our recent reports [17,23], such long-chain sialylated nLc gangliosides have hitherto not been evidenced in human serum or plasma. This repertoire of nLc gangliosides was evidenced both in the sera from cancer patients (PC and GC) and in Ctrls, indicating that these cancer types do not uniquely synthesize these species (Supplementary Figure S2). Rather, changes in the relative quantitative pattern of serological ganglio- and nLc series are expected to occur with these cancer types.

Figure 2.

Representative CID mass spectra of the most abundant serum monosialogangliosides of the neolacto-series. * indicates diagnostic fragment ions derived from the sphingoid base (18:1;O2) of the ceramide moiety.

Table 1.

Serum ganglioside ions detected by nanoESI-QTOF MS (compare Figure 1) and multiple TLC overlay assay followed by direct IR-MALDI-o-TOF MS (compare Figure 3 and Figure 4). ✓ indicates detection by the specified method; a, CD15s epitope; b, CD75s epitope; c, iso-CD75s epitope (assigned by overlay TLC assays).

Proposed Structure	N-Acylgroup	Monoisotopic Mass	nanoESI-QTOF *	TLC-MS *
GM3	C14:1	1122.67	✓	✓
GM3	C14:0	1124.68	✓	✓
GM3	C16:1	1150.70	✓	✓
GM3	C16:0	1152.71	✓	✓
GM3	h16:0	1168.71	✓
GM3	C18:1	1178.73	✓	✓
GM3	C18:0	1180.74	✓	✓
GM3	C20:1	1206.76	✓
GM3	C20:0	1208.78	✓	✓
GM3	h20:0	1224.77		✓
GM3	C22:1	1234.79	✓	✓
GM3	C22:0	1236.81	✓	✓
GM3	h22:0	1252.80		✓
GM3	C23:1	1248.81	✓
GM3	C23:0	1250.82	✓
GM3	C24:2	1260.81	✓
GM3	C24:1	1262.82	✓	✓
GM3	h24:1	1278.82		✓
GM3	C24:0	1264.84	✓
GM3	C25:1	1276.84	✓
GM3	C25:0	1278.85	✓
GM2	C14:0	1327.76		✓
GM2	C16:1	1353.78		✓
GM2	C16:0	1355.79		✓
GM2	C17:0	1369.81		✓
GM2	C18:1	1381.81	✓	✓
GM2	C18:0	1383.82	✓	✓
GM2	C19:0	1397.84		✓
GM2	C20:0	1411.86		✓
GM2	C21:0	1425.87		✓
GM2	C22:1	1437.87		✓
GM2	C22:0	1439.89		✓
GM2	C24:1	1465.90		✓
GD3	C16:0	1443.81	✓	✓
GD3	C18:1	1469.82		✓
GD3	C18:0	1471.84	✓	✓
Neu5Ac-nLc4Cer	C14:0	1489.81	✓	✓ (c)
Neu5Ac-nLc4Cer	C15:0	1503.83		✓ (c)
Neu5Ac-nLc4Cer	C16:1	1515.83	✓	✓ (c)
Neu5Ac-nLc4Cer	C16:0	1517.85	✓	✓ (b,c)
Neu5Ac-nLc4Cer	C18:1	1543.86	✓	✓ (c)
Neu5Ac-nLc4Cer	C18:0	1545.88	✓	✓ (b,c)
Neu5Ac-nLc4Cer	C20:1	1571.89	✓	✓ (c)
Neu5Ac-nLc4Cer	C20:0	1573.91	✓	✓ (b,c)
Neu5Ac-nLc4Cer	C21:0	1587.92		✓ (c)
Neu5Ac-nLc4Cer	C22:1	1599.92	✓	✓ (c)
Neu5Ac-nLc4Cer	C22:0	1601.94	✓	✓ (c)
Neu5Ac-nLc4Cer	C23:1	1613.94		✓ (c)
Neu5Ac-nLc4Cer	C23:0	1615.95		✓ (c)
Neu5Ac-nLc4Cer	C24:2	1625.94	✓	✓ (c)
Neu5Ac-nLc4Cer	C24:1	1627.95	✓	✓ (c)
Neu5Ac-nLc4Cer	C25:1	1641.97	✓	✓ (c)
Neu5Ac-nLc4Cer	h24:1	1643.95		✓ (c)
Neu5Ac-nLc4Cer	C25:0	1643.99		✓ (c)
GD1	C16:0	1808.94	✓	✓
GD1	C18:0	1836.97	✓	✓
GD1	C20:0	1865.00	✓	✓
Neu5Ac-nLc6Cer	C16:1	1880.96	✓	✓ (c)
Neu5Ac-nLc6Cer	C16:0	1882.98	✓	✓ (b,c)
Neu5Ac-nLc6Cer	C18:0	1911.01		✓ (c)
Neu5Ac-nLc6Cer	C24:2	1967.07	✓	✓ (c)
Neu5Ac-nLc6Cer	C24:1	1993.09	✓	✓ (c)
Neu5Ac-Fuc-nLc6Cer	C16:0	2029.04	✓	✓
GT1	C16:0	2100.04	✓	✓
GT1	C18:0	2128.07	✓	✓
GT1	C20:0	2156.10	✓	✓
GT1	C22:0	2184.13		✓
GT1	C23:0	2198.15		✓
GT1	C24:2	2208.13		✓
GT1	C24:1	2210.15		✓
GT1	C25:1	2224.16		✓
Neu5Ac-nLc8Cer	C16:0	2248.11	✓	✓ (c)
Neu5Ac-nLc8Cer	C24:1	2358.22		✓ (c)
Neu5Ac-Fuc-nLc8Cer	C16:0	2394.17	✓	✓ (a,c)
Neu5Ac-Fuc2-nLc8Cer	C16:0	2540.23	✓	✓ (a)
Neu5Ac-nLc10Cer	C16:0	2613.24	✓	✓ (c)
Neu5Ac-Fuc-nLc10Cer	C16:0	2759.30	✓	✓ (a,c)
Neu5Ac-Fuc2-nLc10Cer	C16:0	2905.36	✓	✓ (a,c)
Neu5Ac-nLc12Cer	C16:0	2978.37	✓
Neu5Ac-Fuc-nLc12Cer	C16:0	3124.43	✓	✓ (c)
Neu5Ac-Fuc2-nLc12Cer	C16:0	3270.49	✓	✓ (c)
Neu5Ac-Fuc3-nLc12Cer	C16:0	3416.55		✓ (c)

*, ✓ indicates detection by the specified method; a, CD15s epitope; b, CD75s epitope; c, iso-CD75s epitope (assigned by overlay TLC assays).

Figure 3.

Multiple-stained TLC run scanned by MS. Gangliosides derived from 150 µL serum of a healthy individual were applied for TLC, followed by the multiple overlay assay, namely immunostaining of CD75s and CD15s gangliosides, and then staining of iso-CD75s gangliosides with Maackia amurensis lectin. The lanes were analyzed by IR-MALDI-o-TOF MS by acquiring spectra in equidistant steps (Figure S2). The migration distance is plotted versus the m/z value, and the most abundant ganglioside ions are annotated. A synopsis of all ions detected is given in Table 1. (A) Overlay staining profiles showing CD15s (blue), iso-CD75s (black), and CD75s (red) signals along TLC plate. (B–F) Extracted ion chromatograms (migration distance versus intensity) of representative monosialogangliosides of neolacto series: (B) nLc4Cer species; (C) nLc6Cer species; (D) nLc8Cer species; (E) nLc10Cer species; and (F) nLc12Cer species.

Figure 4.

Multiple immuno- and lectin stainings of TLC-separated serum gangliosides. (A) Serum gangliosides of healthy individuals (Ctrl) and patients suffering from pancreatic (PC) or gastric cancer (GC) were separated by TLC and stained in three consecutive assays using anti-CD75s, anti-CD15s antibodies, and Maackia amurensis agglutinin to detect gangliosides with Neu5Acα2,6Galβ1,4GlcNAc (CD75s), Neu5Acα2,3Galβ1,4(Fucα1,3)GlcNAc (CD15s), and Neu5Acα2,3Galβ1,4GlcNAc epitopes (iso-CD75s), respectively. All samples were applied according to a constant serum volume of 60 µL. A mixture of highly polar gangliosides from human granulocytes was applied as reference (R). Components visualized were enumerated in order of their chromatographic mobility (see also Table S1). (B) CD75s (red) and CD15s (blue) antibodies, as well as iso-CD75s-signals (black), were quantified by densitometry. Proposed structures for the detected compounds are given in Table S1. In the TLC analysis of the reference sample (human granulocytes), CD15s-12 and iso-CD75s-12 were not detected. In contrast, these species were observed in the serum TLC profiles. Based on their chromatographic mobility, CD15s-12 and iso-CD75s-12 were inferred to carry longer glycan chains than CD15s-11 and iso-CD75s-11 containing an nLc12 backbone, but the exact mass and length were not determinable by MS; therefore, the number 12 here denotes any structure longer than nLc14.

Given the association of CD75s, CD15s, and iso-CD75s antigens with malignant development in the gastrointestinal tract [20,21,26,27], we hypothesized that nLc gangliosides carrying these immunologically significant epitopes at their termini are differentially expressed in GC and PC.

3.2. CD-15s-, CD-75s-, and Iso-CD75s-Containing Ganglioside Expression Determined by Multiple Overlay TLC Assay

The qualitative and relative quantitative analysis of ganglioside-rich silica fractions in serum samples of cancer patients and Ctrls was performed by multiple antibody/lectin overlayed TLC and subsequent densitometry analysis. A multiple overlay TLC assay previously developed for tissues [27] was adapted here for serum, specifically for the consecutive staining of the three ganglioside epitopes on a single TLC plate (Figure 3 and Figure 4). The volume of human serum, 60 µL, consumed for the multiple overlay TLC assay is hence three times lower compared to individual TLC analyses of each epitope. The improvement is serum volume and parallelization afforded by the combination of multiple TLC overlay assays with microscale GSL extraction, which increases the overall throughput and cost-effectiveness of the analysis. The stained TLC bands were then scanned by IR-MALDI-o-TOF MS for structural elucidation by co-localization and annotation of m/z signals to band position and differential staining (different dyes) of the epitopes (Figure 3).

In the first overlay TLC assay, gangliosides harboring terminal CD75s-epitope, Neu5Acα2,6Galβ1,4GlcNAc, were detected using the enzyme-amplified immunostaining technique and Fast Red dye (Figure 4A,B, left panels). Anti-CD75s overlay assay revealed red bands originating from molecules with high chromatographic mobility. Ganglioside bands of the anti-CD75s assay and the two subsequent assays were annotated according to the order of their migration distance from the start, with one being the fastest molecule. In the first assay, four ganglioside signals were visible: CD75s-1 to -4. An intense signal corresponding to CD75s-2 was observed in all samples, rarely accompanied by a faint signal of CD75s-4. In the second overlay TLC assay, gangliosides with the CD15s, sLe^x-epitope, and Neu5Acα2,3Galβ1,4(Fucα1,3)GlcNAc were detected, again, by the enzyme-amplified immunostaining with a substrate giving a blue precipitate (Figure 4A,B, middle panels). Using this overlay assay, we have evidenced at least twelve CD15s gangliosides, named CD15s-1 to -12, with a predominance of very complex gangliosides: CD15s-9 to -12. The latter could reflect either the preferential shedding and cellular expression of the long-chain CD15s gangliosides or a higher affinity for the anti-CD15s antibody, possibly because of the easier access of the antibody to sLe^x epitopes residing on longer glycan chains. In the third TLC overlay assay, iso-CD75s gangliosides, not recognized by anti-CD15s antibody, were detected using Maackia amurensis agglutinin I as a glycan binding probe. MAA I specifically reacts with trisaccharide Neu5Acα2,3Galβ1,4GlcNAc. Twelve bands could be differentiated: iso-CD75s-1 to -12 (Figure 4A,B, right panels). Since the doublet bands of iso-CD75s-1 and -2 were not fully resolved to be reliably quantified by densitometry (Figure 4B, the right panel), their intensities were summed up. Given the broad range of TLC mobilities and the high number of signals assigned using anti-CD15s antibody and MAA overlay assays, we evidence a higher structural diversity of α2,3-sialylated gangliosides in these human sera compared to only two α2,6-sialylated gangliosides: CD75s-2 (nLc4 species) and CD75s-4 (nLc6 species), that is, with a much simpler composition (Table 1).

3.3. Structural Characterization of Multiple Overlay TLC Assay-Detected Serum Gangliosides by IR-MALDI-o-TOF MS

Structural elucidation of the immunostained and MAA-stained ganglioside species was complemented by a direct IR-MALDI-o-TOF MS analysis of the TLC lanes (Figure 3, Figure 4 and Figure S3) [26,27]. The measurements were performed in the scanning mode, i.e., the spectra were recorded in equidistant steps (step size 0.3 mm) along the lane. Results of such an experiment are depicted in Supplementary Figure S3, where the migration distance on TLC is plotted on the abscissa and the determined m/z values of the MS signals on the ordinate, with MS signal intensities expressed in greyscale. The structural assignments were carried out by combining the m/z values and immuno-/lectin staining signals obtained at the same position on the lane/TLC Plate, e.g., co-localized TLC/MALDI positions. The densitometric signals from the stained TLC chromatograms (Figure 3A and Figure 4A) concur with the extracted ion chromatograms from the IR-MALDI-o-TOF MS measurements (Figure 3B–F and Figure S3) and the nanoESI QTOF MS/MS data (Figure 2), depicting the highly heterogeneous ganglioside pattern. The recorded signals were attributed to GSLs of the ganglio- and nLc series (Supplementary Figure S3, Table S1). The ganglio-series was represented by GM3, GM2, GD3, GD1, and GT1, and the nLc series was represented by monosialylated gangliosides with up to dodecasaccharide core structures, including fucosylated species (Table 1 and Table S1).

We identified three GSLs that have an nLc6 core structure—two iso-CD75s-2 (VI³Neu5Ac-nLc6Cer) and one CD75s-4 (VI⁶Neu5Ac-nLc6Cer) species—with their heterogeneity stemming from the N-acyl group of ceramides, C16:0 or C24:1, and the type of sialylation being α2,3 or α2,6 (Figure 3C). Overall, we assigned 16 out of 19 overlay signals to gangliosides with nLc4 to nLc12 core structures (see Figure 3B–F, Figure 4, Table 1 and Table S1). The structures of overlay TLC assay-detected CD15s-12 and iso-CD75s-12 could not be elucidated, mainly because their signals, being so close to the application zone, could not be resolved by TLC (start in Figure 4). We assume that they represent multiple, highly polar, low-abundant gangliosides with an nLc backbone longer than nLc14. An overview of all detected gangliosides with their corresponding annotations is given in Table 1 (see also Table S1 for reference).

By combining TLC immuno-/lectin staining and IR-MALDI-o-TOF MS analysis, differences in the CD75s and iso-CD75s structural patterns were found in control human serum. CD75s epitopes, with α2,6-Sia, are preferably present on nLc4 and nLc6 species. Longer-chain gangliosides, with nLc8, nLc10, and nLc12 cores, with or without Fuc residues, were exclusively detected as α2,3-sialylated species (CD15s-6 to -11 and iso-CD75s-3 to -11 in Table 1). Furthermore, we observed that human serum gangliosides and nLc-species with cores longer than nLc8 preferentially carry C16:0 fatty acyl chains. This finding is consistent with our previous study [17], which also indicates a preferential shedding of gangliosides with C16 chains [28] in PC and GC. Interestingly, we observed different lipoform profiles of CD75s and iso-CD75s species: iso-CD75s gangliosides on nLc4 and nLc 6 backbone contained higher proportions of long-chain fatty acyls (up to C24) than the CD75s on nLc4 and nLc6chains. This fact is chiefly pronounced for the nLc4Cer and nLc6Cer species (Figure 3B,C) and may reflect different modes of shedding between CD75s and iso-CD75s gangliosides.

3.4. Serological Profile of CD15s- CD75s-, and Iso-CD75s-Containing Gangliosides in Pancreatic and Stomach Cancer Patients and Controls

To investigate the role and expression of CD15s-, CD75s-, and iso-CD75s-containing gangliosides with PC and GC cancer, 100 human serum samples were analyzed by multiple TLC overlay assay and IR-MALDI-o-TOF MS: 40 samples from PC patients, 40 samples from GC patients, and 20 samples from the cancer-free Ctrls (see Table S2 for detailed patient information). We identified 80 different CD15s-, CD75s-, and iso-CD75s-containing gangliosides in serum samples exhibiting a similar qualitative pattern in Ctrls and cancer patients. Again, no cancer-specific synthesis of such antigen-containing gangliosides was observed. We relatively quantified the expression levels of all detected ganglioside species to dissect possible differences among the three study groups and the association of ganglioside levels with the extent of the primary tumor and its metastatic behavior (Figure 5).

Figure 5.

Box-plot diagrams showing alterations of serum ganglioside level associated with gastric cancer and pancreatic cancer. Each box represents the interquartile range (IQR, 25th–75th percentile), the central line indicates the median, and the whiskers (“Tukey whiskers”) extend to 1.5× IQR beyond the quartiles; data points outside this range are plotted individually as outliers. (A) Significant changes were assessed for CD15s-9 (p = 0.033) and iso-CD75s-1 (p = 0.010), comparing gastric cancer patients (GC, n = 40) with the control group (Ctrl, n = 20). Statistical significance was assessed using the Mann–Whitney U test; * p < 0.05, ** p < 0.01. (B) CD15s-12 is negatively correlated with the extent of the primary tumor in gastric cancer (p = 0.004/τ = −0.360): T1, tumor diameter is <2 cm; T2, tumor diameter >2 cm; T3, tumor penetrates the adjacent tissue; T4, the tumor extends into major veins or arteries. (C) Positive correlations of ganglioside levels with the occurrence and number of lymph node metastases in gastric cancer (N): iso-CD75s-1 (p = 0.003/τ = 0.372), iso-CD75s-3 (p = 0.044/τ = 0.252), and iso-CD75s-5 (p = 0.036/τ = 0.263). (D) Ganglioside levels positively correlated with the occurrence and number of lymph node metastases in gastric cancer (N): CD15s-10 (p = 0.043/τ = 0.256). N0, no metastases; N1, up to 6; N2, 7 to 15 metastases; and N3, more than 15 lymph nodes with metastases. (E) A significant increase was observed for iso-CD75s-5 (p = 0.044) in the serum of pancreatic cancer patients (PC, n = 40) compared to the control group (Ctrl, n = 20). (F) A negative correlation of ganglioside levels with the extent of the primary tumor in pancreatic cancer was found: iso-CD75s-5 (p = 0.006/τ = −0.344), iso-CD75s-9 (p = 0.018/τ = −0.297), iso-CD75s-10 (p = 0.046/τ = −0.251), and iso-CD75s-11 (p = 0.013/τ = −0.314). Statistical significance was assessed using the Mann–Whitney U test; * p < 0.05, ** p < 0.01. (G) Ganglioside levels negatively correlated with the extent of the primary tumor in pancreatic cancer: CD15s-10 (p = 0.021/τ = −0.291), CD15s-11 (p = 0.031/τ = −0.271), and CD15s-12 (p = 0.030/τ = −0.273). (H) Positive correlation between the iso-CD75s-4 level and the presence of metastases in pancreatic cancer (p = 0.024). (I) Significant changes were examined for CD15s-2 (p = 0.005), CD15s-6 (p = 0.026), CD15s-7 (p = 0.037), CD15s-9 (p = 0.007), CD15s-10 (p = 0.002), CD75s-2 (p = 0.044), and iso-CD75s-9 (p = 0.046), comparing gastric cancer patients (n = 40) with pancreatic cancer patients (n = 40). Statistical significance was assessed using the Mann–Whitney U test; * p < 0.05, ** p < 0.01. (J) Principal component analysis (PCA) of ganglioside levels. PC1, PC2, and PC3 accounted for 54.22%, 9.74%, and 9.02% of the total variance, respectively. PC1 separated pancreatic cancer (orange) and gastric cancer (pink) patients, driven mainly by iso-CD75s and CD15s species, while PC2 reflected variation in CD15s and CD75s species. Each point represents one patient sample.

3.5. Serum Ganglioside Levels in Gastric Cancer Patients

Significantly reduced expression of two gangliosides: CD15s-9 (p = 0.033) and iso-CD75s-1 (p = 0.010) was determined in the serum of GC patients (Figure 5A). Furthermore, a highly significant negative correlation between CD15s-12 levels and the extent of the primary gastric tumor (T2a and T2b) (p = 0.004/τ = 0.360) was evidenced (Figure 5B). According to the globally recognized TNM classification system for cancer staging, the penetration depth of the primary tumor is described by the T value: T1, the tumor invades lamina propria or submucosa; T2, the tumor invades muscularis propria and subserosa; T2a, the tumor has grown through the muscularis propria; T2b, the tumor has grown into the serosa; T3, the tumor has grown through the serosa; and T4, the tumor invades the adjacent structures (Figure 5B). Also, the occurrence of metastases correlated with serum levels of a few gangliosides. Patients were grouped according to the nodal staging, that is, the occurrence and number of metastases in the regional lymph nodes: N0, no metastases; N1, up to 6; N2, 7 to 15 metastases; and N3, more than 15 metastatic lymph nodes. The serum levels of CD15s-10 (p = 0.043/τ = 0.256), iso-CD75s-1 (p = 0.003/τ = 0.372), iso-CD75s-3 (p = 0.044/τ = 0.252), and iso-CD75s-5 gangliosides (p = 0.036/τ = 0.263) positively correlated with the presence and number of metastatic lymph nodes, i.e., the levels of gangliosides mentioned above increased with the number of metastatic nodes, e.g., N2 (Figure 5C,D). No significant correlations of serum ganglioside levels with distant metastases in GC patients were evidenced.

3.6. Serum Ganglioside Levels in Pancreatic Cancer Patients

Elevated iso-CD75s-5 serum levels were revealed in the PC group compared to the Ctrl group (p = 0.044, Figure 5E). The growth of the primary tumor, T, was negatively correlated with changes in serum levels of multiple α2,3-sialylated gangliosides: iso-CD75s-5 (p = 0.006/τ = −0.344), iso-CD75s-9 (p = 0.018/τ = −0.297), iso-CD75s-10 (p = 0.046/τ = −0.251), iso-CD75s-11 (p = 0.013/τ = -0.314), CD15s-10 (p = 0.021/τ = −0.291), CD15s-11 (p = 0.031/τ = −0.271), and CD15s-12 (p = 0.030/τ = −0.273). In the early-stage PC patients (T1 and T2, tumors with less or more than 2 cm in diameter, respectively), elevated levels of iso-CD75s-5 and iso-CD75s-9, -10, and -11, as well as CD15s-11 and -12 (Figure 5F,G), were found compared to Ctrls. However, these levels dropped during the disease progression (T3). Though a single patient with T4 PC had much higher levels of all seven aforementioned gangliosides than those in Ctrls, this was not significant. The negative correlations between ganglioside levels and the extent of primary pancreatic tumor are depicted in Figure 5F,G. Analogous to the GC group, no correlation could be established between the presence of distant metastases and the serum ganglioside levels in PC. However, when evaluating levels of individual gangliosides with regard to the lymphatic spread of the disease, iso-CD75s-4 was significantly increased in PC patients having regional lymph node metastases, N1, compared to those without, N0 (p = 0.024, Figure 5H).

4. Discussion

This targeted assay of specific antigen-containing GSLs, combining the microscale extraction of human serum GSLs, multiple overlay TLC assay, and direct IR-MALDI-o-TOF MS, enabled, for the first time to our knowledge, a comprehensive coverage of CD15s-, CD75s-, and iso-CD75s-containing gangliosides, amounting to a total of 80 different structures in Ctrls and PC/GC patients. This epitope-specific ganglioside assay is of particular value for immune system investigation in cancers and, in general, in diseases. Particularly, in combination with modern, more automated TLC technology and modern MALDI MS instruments capable of faster analysis time and better performance for MS detection and fragmentation, such a strategy can expedite profiling of specific epitope-containing gangliosides in human sera or other biological matrices. Out of 80 total serum gangliosides, 32 structures were attributed to the nLc series of core type 2 with complex gangliosides up to nLc12 chain length and various fucosylation degrees. This outcome attests to the advantage of this assay for the high-resolution structural profiling of both glycan chain and lipoforms of long-chain sialylated nLc gangliosides containing specific immune epitopes in human serum. The specificity of the glycan binding probes, anti-CD15s and anti-CD75s antibodies and plant lectin MAA, permitted the discerning of sialic acid- and Fuc-derived isomers, i.e., type 2 core structures with α2,3- and α2,6-sialylation, and proxy-terminal α1,3-fucosylation or internal fucosylation. Moreover, their lipoforms of diverse fatty acyls and the degree of unsaturation of the ceramide moiety can be concurrently determined by MALDI MS.

While the very low abundance of some of the long-chain nLc gangliosides can preclude unambiguous structural elucidation by sensitive techniques like nanoESI and/or nanoLC-ESI MS/MS, the high affinity of the glycan-binding probes used in the multiple overlay TLC assay for such complex gangliosides renders them amenable to structural elucidation by staining and IR-MALDI-o-TOF MS. Nevertheless, the location of fucose residue(s) in the backbone nLc chains by TLC-MS or nanoESI-QTOF MS had to remain speculative for some species. For instance, iso-CD75s-8 could be assigned to X³Neu5Ac-Fuc-nLc10Cer (d18:1, C16:0) (Figure 3E, Table 1), where the fucose is not bound to the penultimate GlcNAc (otherwise it would not be recognized by anti-iso-CD75s antibody), but the exact position along the oligosaccharide chain remains elusive. The signal of iso-CD75s-8 could have originated from several ganglioside species differing in their fucosylation pattern, where VIM-2 dodecasaccharide X³Neu5Ac-VII³Fuc-nLc10Cer possibly represented one of them. These structures could be interrogated in prospective overlay TLC assays utilizing probes that recognize the internal fucosylation in VIM-2 and difucosyl-sLe^x antigens and/or potentially with higher sensitive MS technologies.

The exact source of the highly polar complex nLc gangliosides in human serum or plasma remains to be elucidated. Apart from our previous studies [17,23] and this study, long-chain sialylated nLc species with variable extent of branching at Galβ1 were detected in tissues or cells: human erythrocytes and the placenta [29]; > nLc8 gangliosides in human granulocytes [25], which was also used as a reference mixture here (Supplementary Table S1); iso-CD75s-1, iso-CD75s-2, and iso-CD75s-3 and -4 in healthy human stomach tissue [30]; and α2,3/6-sialylated nLc species with up to nLc10Cer, with or without α1,3Fuc-GlcNAc moiety, were evidenced in colon adenocarcinoma [31,32,33], rectal cancer [34], GC [30], PC [18,26], and in colorectal cancer cell lines [35]. Still, compared to other lipidomics data, the comprehensive gangliosidome of human cancer tissue is rare, and on plasma/serum gangliosidome in cancer patients, even more so [36].

The lack of unique ganglio-species with PC and/or GC cancer determined here suggests that (i) de novo tumor-specific ganglioside synthesis [37] or release into the circulation via shedding or vesicle transport does not occur and/or (ii) tumor cells remodel terminal epitopes of GSLs by the action of serological glycosidases (i.e., neuraminidase/sialidase) and glycosyltransferases [38]. Our data indicate serological level changes specific to cancer (cancer versus Ctrl) and to cancer type: GC versus PC. In fact, it is essential and of high utility for differential cancer mechanism elucidation that these epitope-containing GSLs not only differentiate each of the PC or GC from the Ctrl group but also exhibit potential discriminative power between PC and GC (Figure 5). The upregulation of CD15s-6, -7, -9, -10, and iso-CD75s-9 and downregulation of CD75s-2 and CD15s-2 with PC versus GC adds new value to the discriminative potential of C16 GD1a and C16 GM2 determined previously for PC/GC [17], expanding the set of differential biomarkers and potential enzymatic and/or gene regulatory mechanisms that are involved in differential cancer development. Generally, long-chain iso-CD75s and CD15s are predominant in PC, while shorter chains are found in GC, indicating a preferential chain elongation byglycosyltransferases in tandem with the antigen expression in PC (Figure 5J).

In general, it is of note that elevated sLe^x expression is a joint event in many carcinomas, including GC [39] and PC [40]. Our study unveiled several significant differences between cancer patient groups and cancer-free Ctrls: (i) a significant reduction in levels of iso-CD75s-1 and CD15s-9 in the GC group and (ii) a significant elevation of iso-CD75s-5 but no change in sLe^x content in the PC group. Interestingly, the reduced amounts of iso-CD75s-1 and sLe^x-containing ganglioside CD15s that we detected in sera of GC patients do not concur with the reported elevated levels of iso-CD75s gangliosides and/or sLe^x in tumor tissues [41,42]. One possible explanation for this discrepancy is that lectin histochemistry on the tumor cells does not distinguish between iso-CD75s or sLe^x epitopes attached to proteins (N- and O-linked) or to GSLs [41,42], while in our study, lipids and proteins were separated during extraction. The rise in serum levels of iso-CD75s-5 in PC detected here is in agreement with high expression in pancreatic adenocarcinoma tissues of 2,3-sialyltransferase—ST3Gal IV—that acts on type 2 Galβ1,4GlcNAcβ chain, forming iso-CD75s structures [43].

The cause of the altered ganglioside levels in serum remains to be elucidated, but few hypotheses are under active investigation, among which is that there is a greater extent of “shedding” by the tumor [44,45] to evade or inhibit the host immune response by uptake and masking tumor cells with these GSL species, a phenomenon contributing to tolerogenicity in cancer [46]. The level changes in plasma GSLs might not necessarily stem from the same origin as in tissues, given that blood cells can remodel their epitopes in immune responses, while the tumor microenvironment exhibits local metabolic remodeling processes to adapt to and/or accommodate migrating or resident cancer cells. Finally, the contribution of circulatory enzymes that are dynamically modulated by activated platelets is of interest in this context. The blood-borne sialyl-, galactosyl-, and FucT transferases alter termini of type 2 oligosaccharide chains and, hence, can create new glycan motifs such as sLe^x [38]. Human platelets gain a high α2,3-sialyltransferase activity with specificity for type 2 LacNAc upon activation [38]; hence, they could transform neutral nLc-GSL into respective iso-CD75s ganglioside. Analogously, desialylation by plasma membrane-bound sialidase(s), such as ganglioside-specific neuraminidase Neu3 [47], can occur in human cancers [48]. The low content of α2,3-sialo-type 2 glycolipids in the circulation of GC patients found in our study is concordant with a markedly higher expression of Neu3 in GC tissue than in non-tumoral tissues [49]. Finally, the role of anti-ganglioside antibodies in human diseases, including cancer, is starting to be unveiled [50], and autoantibodies to ganglio-series were detected in sera of GC [51]. Future studies interrogating the role of such enzymes and antibodies in GSL metabolism would help elucidate the origin and mechanisms of GSL changes in blood and tissues.

All statistically significant results we obtained were related to the expression of α2,3-sialo-gangliosides. Even though previous studies indicated the overexpression of α2,6-sialylated species and CD75s gangliosides in GC [24] and PC tissues [18], we did not observe this in the serum of our cohort. Distinctive GSL remodeling between local tissue and systemic GSL glycosylation and/or substantial inter-individual variability in circulating CD75s levels might explain this. The total amount of GSL-bound glycans in human serum can vary considerably among healthy individuals [52], and additional variability may arise from differences in ABO blood group and polymorphisms in FUT1-7 loci, which influence Lewis antigen expression profile [30,53].

Type 2 glycans are shared substrates for α2,3-sialyltransferases (such as ST3Gal III, IV, VI) and α2,6-sialyltransferases (such as ST6Gal I and II), representing a potential source of variability in serum gangliosidome. When these enzymes reside in the same cell, they compete for common substrates to form either CD75s or iso-CD75s structures [54]. Such enzymatic competition may influence a potential source of variability in the relative proportions of CD75s and iso-CD75s structures.

Interestingly, the serum ganglioside levels of GC and PC showed similar trends in association with the histopathological parameters of tumor progression. First, several serum gangliosides with nLc8 to nLc12Cer cores were negatively correlated with the extent of the primary tumor, T, and this finding was characteristic for both cohorts. In GC, it was CD15s-12, whereas in PC, these were iso-CD75s-5, -9, -10, and -11, as well as CD15s-10, -11, and -12. Notably, all these species that altered during the primary tumor growth shared a common structural feature: fucosylation. Iso-CD75s-structures were mono- or difucosylated, while CD15s-structures carry three Fuc residues. The drop in these complex nLc gangliosides during tumor progression may be due to a decrease in the efficiency and energy expenditure to synthesize long glycan sequences in favor of neoplastic growth. Indeed, shorter glycans are very often a hallmark of cancer, including GM1, Neu5Gc-GM3, GD2, GD3, sLe^a, and sLe^x [55]. Change in the fucosylation pattern is a PC event [56] and frequently occurs in GC [57] with a lower expression of FUT3 and production of α1,3-fucosylated gangliosides in advanced stages [58]. Though serum levels of several long-chain fucosylated iso-CD75s and CD15s gangliosides decrease when the primary tumor mass increases, their levels remain significantly higher in T1 tumors than in Ctrls (Figure 6). Only when the critical tumor size exceeds 2 cm in diameter (T1 becomes T2) does the rate of ganglioside synthesis apparently slow down. It may be hypothesized that the role of these complex fucosylated ganglioside species is to prevent a humoral anti-tumor immune response in the early tumor stages, but they may not be needed when the tumor has grown enough and is resistant to the immune attack. This hypothesis aligns well with the fact that these highly polar nLc gangliosides are also present on various circulating cells, such as erythrocytes [59], lymphocytes [60,61], platelets [62], granulocytes [25,63], and on antigen-presenting cells, including human endothelial cells [64]. Therefore, these species may not be readily recognized as non-self, which may contribute to reduced immune recognition. Of note, the serum gangliosidome was in the highest concordance with that of human granulocytes, which express an extensive repertoire of monosialylated, fucosylated poly-LacNAc gangliosides (Table S1) [23,64]. The crucial role the granulocytes play as antigen-presenting cells in innate and adaptive immunity might serve as a target of the GC and PC aggressive tumors with high metastatic potentials to imitate and overproduce granulocyte-like cell surface gangliosides, potentially contributing to immune evasion.

Figure 6.

Biosynthesis pathway of neolacto-series glycosphingolipids (GSLs) and their cancer-associated alterations. (A) Neolacto-series GSL epitopes exhibiting downregulation in gastric cancer and positive correlation with the presence and number of metastatic lymph nodes. x denotes an undefined number of repeating Galβ1–4GlcNAc (LacNAc) units extending the neolacto-series glycan chain beyond the length resolved for CD15s-11. y indicates a variable number of fucose residues on the extended glycan chain. The exact number of repeating units and the precise fucosylation site could not be unambiguously determined by TLC-MS and are therefore represented schematically. (B) Neolacto-series GSL epitopes showing upregulation in pancreatic cancer and a negative correlation with the extent of the primary tumor (T stage).

The role of fucosylated Lewis glycans bound to lacto-/nLc-series GSLs, but not glycoproteins, in tumor rejection and cancer cell death has been evidenced [65]. However, we observed an inverse trend in association with the lymphogenic spread of the tumor: an increase in several serum ganglioside levels with progressing metastatic disease, which was, again, characteristic for both cancer cohorts. These were three iso-CD75s species (-1, -3, and -5) and CD15s-10 in GC and only one ganglioside, iso-CD75s-4, in PC. Only two out of five species positively correlate with the metastasis in GC and carry Fuc residue: iso-CD75s-5 and CD15s-10. Our results are concordant with those of Gomes et al., where a GC cell line transfected with ST3GAL4 that codes for ST3Gal IV sialyltransferase abundantly produces sLe^x epitope and displays an invasive phenotype [66]. In line with this, the extent of MAA I lectin binding to three different human GC cell lines was proportional to their metastatic capacity [67]. We did observe a rise in serum levels of iso-CD75s-4 ganglioside in association with the extent of lymphogenic dissemination, but in PC. A high expression of both ST3Gal III and ST3Gal IV was found in pancreatic adenocarcinoma tissues, and only PC cells overexpressing the latter show migratory and metastatic capabilities [43].

Taken together, the stage- and cancer type-associated alterations in CD15s-, CD75s-, and iso-CD75s-carrying gangliosides observed in this study are consistent with regulated remodeling of terminal glycosylation on type 2 nLc series chains rather than de novo synthesis of tumor-specific GSLs. Individual epitope-bearing gangliosides conversely dysregulate depending on cancer type and tumor stage, indicating dynamic regulation rather than uniform up- or downregulation. Due to the predominant dysregulation with PC/GC of α2,3-sialylated iso-CD75s and CD15s epitopes, generated primarily by ST3Gal sialyltransferases, particularly ST3Gal III and ST3Gal IV over α2,6-sialylated CD75s epitopes containing GSLS and catalyzed by ST6Gal I and ST6Gal II, we infer here a competition between these enzymes for common LacNAc substrates, shifting the balance between α2,3- and α2,6-sialylated motifs within the same biological context [28,43,54,66]. In parallel, altered expression or activity of the sialyl-Lewis X (CD15s) structures observed here concur with reportedly increased activity of α1,3-fucosyltransferases such as FUT3 and FUT7 in gastric and pancreatic cancers and are linked to tumor progression and metastatic behavior [57,58,60]. Beyond enzymatic regulation, changes in circulating levels of specific epitope-bearing gangliosides may reflect tumor-associated shedding, exosome release, or systemic remodeling of GSL termini by extracellular or blood-borne glycosidases and glycosyltransferases, mechanisms previously described for tumor-associated gangliosides [37,38,44]. Functionally, shedding and remodeling of GSLs have important immunological consequences for tolerogenicity and evasion of immune recognition [14,37,55]. In line with this, the presence of long-chain nLc-series gangliosides bearing self-like terminal motifs like CD15s resembling those on leukocytes may further facilitate immune camouflage during early tumor development [38,60]. The most recent view on the metastatic spread of tumors is that it may be a single illness arising from a common non-mutational event: the fusion of primary tumor cells with leukocytes [68]. Moreover, the adaptability of cancer cells to the new organs requires general metabolic flexibility and adaptability for metastatic development and spread. In this context, similarly expressed GSL epitopes on tumor cells, leukocytes, and resident cells could favor the spread. In light of these novel data, the importance of our results and method for prospective epitope-based profiling of GSLs in sera, cells, and tissues is obvious.

Together, these mechanisms provide a biologically plausible framework linking the observed, stage-dependent alterations of CD15s, CD75s, and iso-CD75s gangliosides to tumor progression and immune modulation in gastric and pancreatic cancer.

Certainly, such findings require prospective independent validation in external, larger cohorts; hence, our findings are still limited in direct applicability in clinical translation. Nevertheless, they provide a solid basis for new hypotheses on signaling and metabolic pathways and lipid markers to be prioritized in future large cohort studies. Noteworthy, as well, is that the individual, person-specific immune “tone” or “profile” in health and disease can hinder the highly discriminative power of GC and PC based on immunometabolism. The impetus to establish lipid scores, more representative of systemic cumulative effects of multiple pathways, along with interval reference values for these antigens, can be prospectively valorized in individual patients to identify the risk of tumor development and monitor therapy response.

5. Conclusions

The differential immune-metabolism with and between pancreatic and stomach cancer and with cancer stage, underscored by the specific CD15s, CD75s, and isoCD75s epitope-carrying glycosphingolipids unraveled here, opens new directions of researching and understanding the immunoregulation and immune responses in cancer, as well as therapeutic strategies. Translation of such immune markers into clinical assays for stage monitoring and differential cancer diagnosis has, in our opinion, a high potential to advance the diagnostic management and precision medicine compared to markers of general metabolism, given that they underscore intrinsic cancer mechanisms of immune system evasion and cytotoxicity mechanisms.

Future studies of tumor tissue gangliosidome and serum/plasma from the same patients are urgently needed to infer the origin of the organ footprinting in circulating gangliosides, and scaled-up clinical cohort analyses are prospective prerequisites for the quest to discover cancer mechanisms and new therapy development.

Abbreviations

AP	alkaline phosphatase
BCIP	5-Bromo-4-chloro-3-indolyl phosphate
CD	cluster of differentiation
CD75s	Neu5Acα2,6Galβ1,4GlcNAc-
CD15s	Neu5Acα2,3Galβ1,4(Fucα1,3)GlcNAc-
CID	collision-induced dissociation
Ctrl	control group
CV	column volume
FucT	Fucosyltransferase
Gb	Globoside
GC	gastric cancer
GSL	Glycosphingolipid
GT	Glycosyltransferase
HILIC	hydrophilic interaction liquid chromatography
iso-CD75s	Neu5Acα2,3Galβ1,4GlcNAc
LacNAc	Galβ1,4GlcNAc
MAA	Maackia amurensis agglutinin
MWU	Mann–Whitney U test
Neu 3	Neuraminidase
NBT	nitro blue tetrazolium salt
nLc	neolacto-series
NP	normal phase
PC	pancreatic cancer patients
PLC	phospholipase C
R	Reference
RP	reverse phase
St	stomach cancer patients
VIM-2	Neu5Acα2,3Galβ1,4GlcNAcβ1,3Galβ1,4(Fucα1,3)GlcNAc

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cancers18040663/s1.

Author Contributions

J.S., S.K. and M.H. performed experiments, acquired, and processed data; J.S. interpreted the data and wrote the main part of the manuscript; R.M. and H.G.V. interpreted the data and contributed to writing the manuscript; J.P.-K. and L.B. acquired funds and interpreted the data; L.B. supervised the study and contributed to data interpretation and writing the manuscript. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The clinical samples were procured from a commercial R&D entity, RNTech company, which operated according to the required regulations. The authors are privy to all information pertaining to RNTech’s legal and logistic operations.

Informed Consent Statement

Research carried out on purchased/procured samples from company does not require local ethical approval. Companies including RNTECH have themselves the approvals and regulations in place to collect and market samples.

Data Availability Statement

All data described in this manuscript are provided as processed data in the manuscript and Supplementary files.

Conflicts of Interest

Author Jamal Souady was employed by the company QIAGEN GmbH. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding Statement

This work was supported by the European Commission, Seventh Framework Programme, project GLYFDIS to J. P.-K. [LSHB-CT-2006-037661]. S.K. was a fellow of the Studiumstiftung des Deutschen Volkes. Partial funding was provided by the Clinical Lipidomics Unit at University Medical Center of Mainz through its Lipidomic Core Facility.