• cancer biomarker;
  • glycomics;
  • glycosphingolipidomics;
  • hepatocellular carcinoma


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Backgrounds & Aims

Glycosylation promoting or inhibiting tumour cell invasion and metastasis is of crucial importance in current cancer research. Tumour-associated carbohydrate antigens are predominantly expressed on the tumour cell surface. Glycosphingolipids (GSLs) are members of the family. To perform glycosphingolipidomic assays on neutral GSLs obtained from solid hepatocellular carcinoma (HCC) tissues and paired peritumoural tissues by linear ion trap quadrupole-electrospray ionization mass spectrometry.


Qualitative and quantitative analysis of fucosylated neutral GSLs was performed in the positive ion mode on the LTQ-XL mass spectrometer and MALDI-TOF-MS.


A group of fucosylated neutral GSLs in HCC was found to be expressed higher in the tumour tissues, as their proportion in total cellular GSLs was 3.3-fold higher in the tumour tissues than in the peritumoural tissues (P < 0.01). Moreover, qualitative analysis of the aberrant fucosylated GSLs were completed, and seven types of fucosylated GSLs that contained terminal Fuca2Gal- structure were identified by mass spectrometry.


Our results may lead to improved immunotherapy of HCC and contribute to understanding the role of aberrant fucosylated GSLs in the development and progress of HCC in following studies.


globotriaosylceramide (Galα4Galβ4Glcβ1Cer)


globoside (GalNAcβ3Galα4GlcNAcβ3Galβ4Glcβ1Cer)


glucosylceramide (Glcβ1Cer)

Globo H





hepatocellular carcinoma


lactosylceramide (Galβ4Glcβ1Cer)










linear ion trap quadrupole-electrospray ionization


matrix-assisted laser desorption/ionization-time-of-flight


mass spectrometry

Type I H antigen


Type II H antigen


Hepatocellular carcinoma (HCC) is the fifth most common malignant tumour in the world. About 0.5–1 million new cases occur around the world each year. HCC is one of the top causes of cancer death in China, and new cases in China alone account for more than half the global total. Treatments that target HCC specifically are scarce [1-6]. Searching for new therapeutic targets has become a major research focus for HCC.

Tumour-associated carbohydrate antigens are predominantly expressed on the tumour cell surface. Glycoconjugates are important information-bearing biomacromolecules of the cell membrane, and abnormal expression levels of glycoconjugates in tumours have been reported [7]. Previous studies with HCC patients have focused on the abnormal expression of glycoproteins [8-18]. Souady et al. recently reported that CD75s- and iso-CD75s-ganglioside expression in HCC cells is associated with poor histopathological differentiation [19]. However, knowledge about neutral GSLs in HCC is lacking. In one report, fucosylated GSLs were found to be present on HCC tissues [20]. Their structures are unknown.

Recently, mass spectrometry (MS) methods based on multistep fragmentation of permethylated glycans in ion traps have shown great promise for distinguishing isobaric glycan structures, such as globotriaosylceramide (Gb3; Galα4Galβ4Glcβ1Cer) and iso-Gb3 (Galα3Galβ4Glcβ1Cer) [21]. In this study, we employed ion trap MS to analyse fucosylated neutral glycolipids in HCC and paired peritumoral tissues and confirmed the presence of Globo H and glycosphingolipds with terminal Fucα2Gal- structures by comparison with a synthetic standard. The results revealed that Globo H and glycosphingolipids with terminal Fucα2Gal- structures show promise as a diagnostic indicator of HCC and as a target for treatment.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Cancer specimens

HCC and paired peritumoral tissues from 32 patients were collected from the First Affiliated Hospital of Soochow University and the Tongji Hospital of Shanghai Tongji University according to institutional guidelines. Patients, who were all ethnic Han Chinese, had been recently diagnosed and had not yet received any medical treatment for the cancer. Each HCC sample had been histologically verified. The characteristics of participants in this study are summarized in Table 1.

Table 1. Characteristics of participants in this study
Patient No.aAge/GenderbAetiologycT-BILdALTeAFPfGGTg
  1. a

    For each patient, HCC and peritumoral tissue was obtained.

  2. b

    Age and gender (male or female) of the patient.

  3. c

    Aetiology of the Hepatocellular carcinoma. NBNC means non-HBV and non-HCV liver disease.

  4. d

    Total bilirubin levels in μmol/L.

  5. e

    Alanine transaminase in IU/L.

  6. f

    AFP in μg/L.

  7. g

    Gamma-glutamyl transpeptidase in IU/L.


Total GSLs extraction from HCC and peritumoral tissues

Total GSL was extracted as previously described [21]. Briefly, 0.3 × 0.3 × 0.3 cm samples were extracted by sonication 4 times (1 ml each) with chloroform/methanol 1:1 (by volume) and another 4 times with isopropanol/hexane/water 60:25:20 (by volume). Supernatants removed after centrifugation were pooled and dried with centrifugal vacuum concentration. Total dried crude lipid was separated by anion-exchange chromatography of DEAE Sephadex A-25 (Sigma, St. Luis, MO, USA) on a glass column (Internal diameter × Length, 5 × 80 mm). The neutral lipid fraction was eluted with 5 column volumes of chloroform/methanol/water 30:60:8 (by volume), while the acidic lipid fraction was eluted with 0.8 M sodium acetate in methanol; both fractions were dried. The acidic fraction was also desalted by dialysis and dried by centrifugal vacuum concentration.

Per-N,O-methylation of neutral GSLs

A modification of the method of Ciukanu and Kerek [22, 23] was employed for per-N,O-methylation of GSLs. GSLs (1–20 μg) were dissolved with 100 μl of dimethyl sulphoxide in a glass tube. Powdered sodium hydroxide was ground with an agate pestle and mortar, dried at 100°C in an oven, measured out with a microspatula, and added to the sample solution. Iodomethane (80 μl) was added with a syringe, and the mixture was shaken for 1 h at room temperature. The methylation reaction was quenched with 2 ml of water. After the permethylated products were extracted by addition of 2 ml of dichloromethane, the dichloromethane extracts were washed 8 times with 2 ml of water. Following the final wash, the dichloromethane phase was transferred to a new tube and dried under vacuum.

Exoglycosidase digestions with α1,2 Fucosidase

Twenty micrograms of glycosphingolipids were digested by 5 μl α1-2 Fucosidase (P0724s, NEB, Ipswich, MA, USA) in 1 × G4 Reaction Buffer, supplemented with 100 μg/ml BSA, in a 50 μl reaction. The reaction mixture was incubated overnight at 37°C, and the glycosphingolipids were extracted by chloroform/methanol (1:1/volume) and permethylated as described above, followed by LTQ-ESI mass spectrometry analysis. Control was the GSLs treated with reaction buffer alone without enzyme.

Analysis of GSLs using LTQ-ESI-MS

After methylation, qualitative analysis of fucosylated neutral GSLs was performed in the positive ion mode on the LTQ-XL mass spectrometer (Thermo Fischer Scientific, Waltham, MA, USA) by using a metal needle for direct infusion of samples dissolved in methanol, with a flow rate of 5 μl/min and at ion spray voltage 3.5 kV, capillary voltage 35 V, capillary temperature 350°C, injection time 100 ms, activation time 30 ms, and isolation width m/z 1.5. All ions were detected as sodium adducts. To gather as much structure information as possible, MS3, MS4, and MS5 spectra of all the fucosylated neutral GSLs were obtained as well.

Parent ion scanning for Globo H

Parent ion scanning was performed in the positive ion mode by linear ion trap quadrupole-electrospray ionization mass spectrometry (LTQ-ESI-MS) at ion spray voltage 3.5 kV, capillary voltage 35 V, capillary temperature 350°C, injection time 100 ms, activation time 30 ms, and isolation width m/z 1.5. Scanning method was set as parent mass range m/z 1820–2000, parent mass step m/z 0.2, and product mass m/z 1290. The ion abundance of neutral GSLs analysed by parent ion scanning was between 2 × 105 and 8 × 105. The ion abundance for HCC and paired peritumoral samples was similar.

MALDI-TOF-MS analysis

Matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) MS and MALDI-TOF/TOF-MS/MS experiments were performed with a MALDI TOF-TOF mass spectrometer (4700; Applied Biosystems, Foster City, CA, USA). The matrix was prepared by mixing 50% volume of 10 mg/ml dihydroxybenzoic acid in acetonitrile and 50% volume of 0.1% trifluoroacetic acid in water. The dihydroxybenzoic acid matrix was spotted on the MALDI target (1 μl), which was followed by a 1-μl spot (containing 100 ng of GSLs) of sample dissolved in methanol, and allowed to dry before analysis. MS experiments were acquired using the reflectron settings in the positive mode. MS/MS data were obtained using the 1 kV mode with argon or air as collision gas (collision-induced dissociation cell gas pressure 3.5 × 10−6–9 × 10−7 Torr). MS spectra were summed from 1000 to 10 000 laser shots. For MS/MS, 5000 to 50 000 shots were summed. The Globo H standard used for MALDI-TOF assessment was kindly provided by Drs. Ouathek Ouerfelli and Samuel J. Danishefsky, Memorial Sloan-Kettering Cancer Center.

Real-time PCR

The PCR primers for FUT1 were GCAGGTTATGCCTCAGCG (forward), and TCCATCGCCAGC AAA CG (reverse). The primers for FUT2 were CGTTCAGATGCCTTTCTCCTTT (forward), and GGTCCCAGTGCCTTTGATGTTG (reverse). The primers for β-actin were CACCATTG GCAATGAGCGGTTCC (forward), and GTAGTTTCGTGGATGCCACAGG (reverse). Total RNA was extracted from the tumour tissues and paired peritumornal tissues respectively, and was reverse-transcribed to cDNA with oligo (dT) primers. Real-time PCR was performed on an ABI Prism 7500 Sequence Detection System. The program for real-time PCR reaction was 50°C for 2 min, 95°C for 10 min, followed by 40 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 1 min. The threshold cycle number (Ct value) of target gene was normalized by the reference gene β-actin. The expression level of enzyme for each patient was calculated by 2^−ΔCt. ΔCt = Ct (target gene)-Ct (β-actin gene).

Relative quantification of each type of GSL in total GSLs

Relative quantification was conducted with the formula (the sum of the intensity of the peaks represented the same type of GSL)/(the sum of the intensity of total cellular GSL peaks).

Statistical analysis

Statistical analysis was performed with the two-tailed Student t-test using GraphPad Prism software (La Jolla, CA, USA). A P value of less than 0.05 indicated statistical significance. Receiver operating characteristic curve (ROC) analysis was performed by spss software (Ver.17.0; Chicago, IL, USA) to assess the sensitivity and specificity of the potential diagnostic variables. AUC value means the area under the curve in ROC graph.


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Fucosylated pentosylceramides and hexosylceramides in HCC

To investigate whether cancer-associated changes in GSLs occur in HCC, we performed MS1 analysis on the 32 paired HCC and peritumoral tissues. The main neutral GSLs expressed by these tissues were glucosylceramide (GlcCer; Glcβ1Cer), lactosylceramide (LacCer), Gb3, and globoside (Gb4) (Table 2). Pentosylceramides and hexosylceramides are relatively low abundance GSLs (less than 5% in average).

Table 2. Major GSLs in HCC and paired peritumoral tissues
GSLHCCPeritumoral Tissue
Molecular ionsFatty AcidsaSphingosinebTotal GSL (%)Molecular ionsFatty AcidsaSphingosinebTotal GSL (%)
  1. Data were summarized as the average of 32 patients.

  2. a

    For fatty acids, the number before the colon refers to the carbon chain length. The number after the colon gives the total number of double bonds. Fatty acids with a 2-hydroxyl group are denoted by the prefix “h” before the abbreviation.

  3. b

    For the sphingoid base, the number before the colon refers to the carbon chain length. The number after the colon gives the total number of double bonds. “d” denotes dihydroxyl and “t” denotes trihydroxyl. For some ceramide forms, alternate fatty-N-acyl/sphingosine compositions are possible.

851.219:0d18:0851.2 19:0d18:0
863.4 20:0d18:1863.4 20:0d18:1
865.0 20:0d18:0865.0 20:0d18:0
877.2 21:0d18:1877.2 21:0d18:1
891.0 22:0d18:1891.0 22:0d18:1
893.2 22:0d18:0893.2 22:0d18:0
905.0 23:0d18:1905.0 23:0d18:1
916.8 24:1d18:1916.8 24:1d18:1
919.0 24:0d18:1919.0 24:0d18:1
935.0 h23:0d18:1935.0 h23:0d18:1
949.4 h24:0d18:1949.4 h24:0d18:1
952.5 h22:0t18:0952.5 h22:0t18:0
958.6 27:0d18:2958.6 27:0d18:2
966.2 h23:0t18:0966.2 h23:0t18:0
972.5 28:1d18:2972.5 28:1d18:2
981.2 h24:0t18:0981.2 h24:0t18:0
1008.5 h26:0t18:01008.5 h26:0t18:0
1037.5 h28:0t18:01037.5 h28:0t18:0
LacCer982.6 14:0d18:139.38 982.6 14:0d18:137.67
996.6 15:0d18:1996.6 15:0d18:1
1010.8 16:0d18:11010.8 16:0d18:1
1026.7 h15:0d18:11026.7 h15:0d18:1
1040.7 h16:0d18:11040.7 h16:0d18:1
1055.2 h17:0d18:11055.2 h17:0d18:1
1066.7 20:0d18:11066.7 20:0d18:1
1069.2 20:0d18:01069.2 20:0d18:0
1081.4 21:0d18:11081.4 21:0d18:1
1094.8 22:0d18:11094.8 22:0d18:1
1096.6 22:0d18:01096.6 22:0d18:0
1108.9 23:0d18:11108.9 23:0d18:1
1120.9 24:0d18:21120.9 24:0d18:2
1122.9 24:0d18:11122.9 24:0d18:1
1138.8 h23:0d18:11138.8 h23:0d18:1
1152.9 h24:0d18:11152.9 h24:0d18:1
1157.2 h22:0t18:01157.2 h22:0t18:0
1163.2 27:0d18:21163.2 27:0d18:2
1167.0 h25:0d18:11167.0 h25:0d18:1
1169.6 h23:0t18:01169.6 h23:0t18:0
1180.9 h26:0d18:11180.9 h26:0d18:1
1185.0 h24:0t18:01185.0 h24:0t18:0
1212.7 h26:0t18:01212.7 h26:0t18:0
1240.6 h28:0t18:01240.6 h28:0t18:0
Gb31187.8 14:0d18:116.66 1187.8 14:0d18:119.20
 1201.2 15:0d18:1 1201.2 15:0d18:1
 1214.9 16:0d18:1 1214.9 16:0d18:1
 1230.8 h15:0d18:1 1230.8 h15:0d18:1
 1244.9 h16:0d18:1 1244.9 h16:0d18:1
 1259.4 19:0d18:0 1259.4 19:0d18:0
 1270.7 20:0d18:1 1270.7 20:0d18:1
 1273.4 20:0d18:0 1273.4 20:0d18:0
 1285.4 21:0d18:1 1285.4 21:0d18:1
 1298.9 22:0d18:1 1298.9 22:0d18:1
 1300.8 22:0d18:0 1300.8 22:0d18:0
 1312.9 23:0d18:1 1312.9 23:0d18:1
 1325.0 24:1d18:1 1325.0 24:1d18:1
 1327.0 24:0d18:1 1327.0 24:0d18:1
 1342.9 h23:0d18:1 1342.9 h23:0d18:1
 1357.0 h24:0d18:1 1357.0 h24:0d18:1
 1360.8 h22:0t18:0 1360.8 h22:0t18:0
 1367.1 27:0d18:2 1367.1 27:0d18:2
 1374.8 h23:0t18:0 1374.8 h23:0t18:0
 1380.6 28:1d18:2 1380.6 28:1d18:2
 1388.9 h24:0t18:0 1388.9 h24:0t18:0
 1416.7 h26:0t18:0 1416.7 h26:0t18:0
 1444.6 h28:0t18:0 1445.0 h28:0t18:0
Gb41431.8 14:0d18:116.10 1431.8 14:0d18:114.97
1445.8 15:0d18:11445.8 15:0d18:1
1460.0 16:0d18:11460.0 16:0d18:1
1475.9 h15:0d18:11475.9 h15:0d18:1
1488.0 18:0d18:11488.0 18:0d18:1
1490.0 h16:0d18:11490.0 h16:0d18:1
1504.8 19:0d18:01504.8 19:0d18:0
1516.0 20:0d18:11516.0 20:0d18:1
1517.8 20:0d18:01517.8 20:0d18:0
1530.0 21:0d18:11530.0 21:0d18:1
1544.0 22:0d18:11544.0 22:0d18:1
1545.8 22:0d18:01545.8 22:0d18:0
1558.0 23:0d18:11558.0 23:0d18:1
1570.0 24:1d18:11570.0 24:1d18:1
1572.0 24:0d18:11572.0 24:0d18:1
1586.0 25:0d18:11586.0 25:0d18:1
1588.0 h23:0d18:11588.0 h23:0d18:1
1600.0 26:0d18:11600.0 26:0d18:1
1602.0 h24:0d18:11602.0 h24:0d18:1
1606.0 h22:0t18:01606.0 h22:0t18:0
1611.8 27:0d18:21611.8 27:0d18:2
1619.9 h23:0t18:01619.9 h23:0t18:0
1630.0 28:0d18:01630.0 28:0d18:0
m/z 10861634.0 16:0d18:11.841634.0 16:0d18:10.71
1664.0 18:0d18:01664.0 18:0d18:0
1718.0 22:0d18:11718.0 22:0d18:1
1746.0 24:0d18:11746.024:0d18:1
1748.2 24:0d18:01748.2 24:0d18:0
1774.126:0d18:11774.1 26:0d18:1
1776.016:0d18:01776.0 26:0d18:0
m/z 12601808.2 16:0d18:10.211808.216:0d18:10.12
1835.7 18:0d18:11835.7 18:0d18:1
m/z 12901838.0 16:0d18:10.711838.0 16:0d18:10.00
1868.1 18:0d18:01868.118:0d18:0
1922.5 22:0d18:11922.522:0d18:1
1936.3 23:0d18:11936.3 23:0d18:1
1948.624:1d18:11948.6 24:1d18:1
1950.8 24:0d18:11950.8 24:0d18:1

The MS1 spectrum showed that, compared with peritumoral tissues, HCC samples expressed three novel groups of ions (Fig. 1). Specifically, peaks at m/z 1634, 1664, 1718, 1745, and 1775 represent Fuc(Hex)3HexNAc-Cer; peaks m/z 1808, 1836, and 1892 represent Fuc2(Hex)3HexNAc-Cer; and peaks m/z 1838, 1868, 1922, 1936, and 1948 represent Fuc(Hex)4HexNAc-Cer (Table 2).


Figure 1. Fucosylated GSLs are overexpressed by HCC but not peritumoral tissues. Representative capillary ESI-MS1 profile analysis of permethylated GSLs obtained from HCC samples (A) and paired peritumoral tissues (B). NL, neutral loss.

Download figure to PowerPoint

The relative abundance of glycan signals in the MS results could reflect their mass ratio in tissue samples [24-26], allowing for comparative analyses of GSLs obtained from different samples. To measure the difference in GSLs between HCC and paired peritumoral tissues, we analysed the samples and calculated the proportion of all the GSLs ions detected in the MS1 spectra by comparing it with the ion abundance of total cellular GSLs. The results suggested that the proportion of these aberrant fucosylated GSLs were 3.3-fold more abundant in the HCC samples than in the peritumoral tissues (< 0.01) (Fig. S1 and Table S1).

To identify the structural identities of the three aberrantly fucosylated GSLs, we performed MS2 analysis. For Fuc(Hex)3HexNAc-Cer (profile of m/z 1634), loss of ceramide was represented by the product ion at m/z 1086, loss of terminal Fuc-OH by that at m/z 1446, and loss of terminal Fuc-Gal-OH by that at m/z 1224 (Fig. 2A). The corresponding ions for Fuc2(Hex)3HexNAc-Cer (m/z 1808) were at m/z 1260, 1602, and 1398 (Fig. 2B). The corresponding ions for Fuc(Hex)4HexNAc-Cer (m/z 1838) were at m/z 1290, 1632, and 1428 (Fig. 2C).


Figure 2. Fucosylated structures in HCC. Three groups of GSLs were identified. The groups contained product ions at m/z 1086, 1260, and 1290, representing Fuc(Hex)3HexNAc-Cer (A), Fuc2(Hex)3HexNAc-Cer (B), and Fuc(Hex)4HexNAc-Cer (C) respectively. NL, neutral loss.

Download figure to PowerPoint

Identification of Aberrant Pentosylceramides in HCC

Figure 3 (A and B) shows the MSn of Fuc(Hex)3HexNAc-Cer at m/z 1634, which followed the MSn pathway of m/z 1634 [RIGHTWARDS ARROW] 1086 [RIGHTWARDS ARROW] 660. In MS3 spectra (Fig. 3A), m/z 882 and 864 were B4 and C4 ions, respectively; m/z 660 and 678 were B3 and C3 ions, respectively, with the loss of -Gal-Glc-OH, which were at m/z 449 (Y2 ion) and m/z 431 (Z2 ion); m/z 415 and 433 were B2 and C2 ions, respectively, with the loss of -GlcNAc-Gal-Glc-OH, which were at m/z 694 (Y3 ion); m/z 676 (Z3 ion); m/z 880 and 898 were Z4 and Y4 ions, respectively, with the loss of terminal fucose, and m/z 850 and 868 were Z4 and Y4 ions, respectively, with the loss of terminal galactose (Fig. S2). These data are consistent with the decomposition of permethylated Fucα2Galβ3GlcNAcβ3Galα4Glcβ1Cer (type I H), Fucα2Galβ4GlcNAcβ3Galα4Glcβ1Cer (type II H), Fucα3(Galβ4)GlcNAcβ3Galβ4Glcβ1Cer (Lex), and Fucα4(Galβ3)GlcNAcβ3Galβ4Glcβ1Cer (Lea).


Figure 3. Identification of aberrant fucosylated GSLs in HCC tissues. Identification of pentosylceramide: MS3 (A) and MS4 (B) analyses of a representative ion (m/z 1634). NL, neutral loss. Identification of Leb and Ley: MS3 (C) and MS4 (D) analyses of a representative ion (m/z 1808). NL, neutral loss. Identification of Globo H: MS3 (E) and MS4 (F) analyses of a representative ion (m/z 1838). NL, neutral loss.

Download figure to PowerPoint

Figure 3B (MS4 analysis of m/z 660) showed that m/z 660 is a trisaccharide consisted of fucose, galactose, and glucosamine residues, which may represent the sequences Fuc-Gal-GlcNAc-, and Fuc-GlcNAc-Gal-. Terminal Fuc-Gal- B ions and C ions are represented by m/z 433 and 415, respectively, and internal -Gal-GlcNAc- or -GlcNAc-Gal- by m/z 472 (Fig. 3B). In addition, m/z 229 and 211 (or both) represent C1 ions and B1 ions of terminal fucose residue attached or not attached by the hydroxyl group respectively (Fig. S2).

Figure 3B also showed that m/z 558 was special MS4 fragment of Type I H antigen; m/z 503 was specific to Type II H antigen; m/z 299 was special fragment of Lea, and m/z 329 was specific to Lex. The main peaks m/z 433 revealed that the MS2 ions at m/z 1086 mainly contained type I/II H antigen (Fig. S2).

MS4 analysis showed that m/z 660 mainly represented the Fucα2Galβ3GlcNAc-ene B3 ion and Fucα2Galβ4GlcNAc-ene B3 ion, which could produce internal -Gal- Y ion at m/z 245 specifically in the MS4 spectra (Fig. 3B); and a little Fucα4(Galβ3)GlcNAc-ene B ion and Fucα3(Galβ4)GlcNAc-ene, which could produce terminal Gal- B and C ions at m/z 241 and m/z 259 specifically in the MS4 spectra (Fig. 3B and Fig. S2).

Identification of Ley and Leb in HCC

Figure 3C and 3D shows the MS3 profile of (Fuc)2(Hex)4HexNAc-Cer at m/z 1808, which followed the Lewis b and/or y antigen-specific MSn pathway of m/z 1808 [RIGHTWARDS ARROW] 1260 [RIGHTWARDS ARROW] 834. In MS3 spectra (Fig. 3C), m/z 1038 and 1056 were B5 and C5 ions, respectively; m/z 834 and 852 were B4 and C4 ions, respectively, with the loss of -Gal-Glc-OH, which were at m/z 449 (Y2 ion) and m/z 431 (Z2 ion); m/z 628 and 646 were B3 and C3 ions, respectively, with the loss of terminal fucose and -Gal-Glc-OH, which were at m/z 449 (Y2 ion) and m/z 431 (Z2 ion); m/z 415 and 433 were B2 and C2 ions, respectively, with the loss of -(Fuc-)GalNAc-Gal-Glc-OH, which were at m/z 868 (Y4 ion) and m/z 850 (Z4 ion) respectively; m/z 1054 and m/z 1072 were B5 ion and C5 ion, with the loss of terminal fucose (Fig. S3). Besides, the ion m/z 530 was specific to Leb. These data are consistent with the decomposition of permethylated Leb:Fucα2Galβ3(Fucα4)GlcNAcβ3Galα4Glcβ1Cer and Ley: Fucα2Galβ4(Fucα3)GlcNAcβ3Galα4Glcβ1Cer (Fig. S3).

The MS3 ion fragments at m/z 834 represented the Fucα2Galβ3(Fucα4)GlcNAc-ene B4 ion and Fucα2Galβ4(Fucα3)GlcNAc-ene B4 ion. When subjected to MS4 analysis, terminal Fuc-Gal- B ions and C ions are observed by m/z 433 and 415, respectively, with the loss of Fuc-GlcNAc-, which were at m/z 442 (Y2 ion) and m/z 424 (Z2 ion); m/z 628 and 646 were B3 and C3 ions, respectively, with the loss of terminal fucose (Fig. 4D and Fig. S3).


Figure 4. Precursor ion scanning detection of Globo H in HCC and peritumoral tissues. A precursor ion mapping method was used to search for Globo H molecule by searching for ions that produced product ion at m/z 1290. (A), a representative Globo H expressed by HCC patients. Ions at m/z 1838, 1868, 1894, 1923, 1936, 1948, 1950, 1964, and 1980 represent Globo H with different fatty acyl lengths or modifications, sphingosine variations, or both. (B), no significant signals were found in peritumoral tissues.

Download figure to PowerPoint

Identification of Globo H in HCC

Figure 3E and 3F shows the multistep MSn of Fuc(Hex)4HexNAc-Cer at m/z 1838, which followed the Globo H-specific MSn pathway of m/z 1838 [RIGHTWARDS ARROW] 1290 [RIGHTWARDS ARROW] 660. In MS3 spectra (Fig. 3E), m/z 1068 and 1086 were B5 and C5 ions, respectively; m/z 864 and 882 were B4 and C4 ions, respectively, with the loss of -Gal-Glc-OH, which were at m/z 449 (Y2 ion) and m/z 431 (Z2 ion); m/z 660 and 678 were B3 and C3 ions, respectively, with the loss of -Gal-Gal-Glc-OH, which were at m/z 653 (Y3 ion) and m/z 635 (Z3 ion); m/z 415 and 433 were B2 and C2 ions, respectively, with the loss of -GalNAc-Gal-Gal-Glc-OH, which were at m/z 898 (Y4 ion) and m/z 880 (Z4 ion); and m/z 1084 and 1102 were Z5 and Y5 ions, respectively, with the loss of terminal fucose. These data are consistent with the decomposition of permethylated Fucα2Galβ3GalNAcβ3Galα4Galβ4Glcβ1Cer (Fig. S4).

The terminal trisaccharide at m/z 660 might represent the sequences Fuc-Gal-GalNAc-, and Fuc-GalNAc-Gal as above-mentioned. Terminal Fuc-Gal- B2 ions and C2 ions are represented by m/z 433 and 415, respectively, and internal -Gal-GalNAc- or -GalNAc-Gal- by m/z 472 (Fig. 3F). In addition, m/z 229 and 211 (or both) represent C1 ions and B1 ions of terminal fucose residue attached or not attached by the hydroxyl group respectively (Fig. S4). In conclusion, the ion fragments at m/z 660 represented the Fucα2Galβ3GalNAc-ene B ion, which could produce internal -Gal- Y ion at m/z 245 specifically in the MS4 spectra.

We also analysed a synthetic Globo H standard by MALDI-TOF MS-MS. The control structure was specific to Globo H at m/z 443 and m/z 660 (Fig. S5A and B). This information provided further evidence that the aberrant hexasaccharide Fuc(Hex)4HexNAc-Cer in the HCC specimens was Globo H.

Globo H overexpression in solid HCC tissues

We performed parent ion scanning at m/z 1290 to search for Globo H in the paired HCC and peritumoral tissue samples. The profile shown in Figure 4 shows parent GSLs for a representative HCC patient. Molecular ions at the m/z values shown (Fig. 4A) could generate the saccharide at m/z 1290 in the MS2 spectrum and were confirmed to be Globo H by the ion trap MSn method. The chromatographic spectrum of parent ion scanning demonstrated that Globo H was expressed in HCC tissues only and not in peritumoral tissues (Fig. 4B). Overall, the expression of Globo H was identified by parent ion scanning in 29 of the 32 HCC specimens (90.6%) but not in peritumoral tissue specimens (Table S2).

Overexpression of GSLs with Fucα1-2Gal- structures in solid HCC tissues

Qualitative analysis of three types of aberrant ions representing fucosylated GSLs in HCC was complex. It was difficult to measure the expression of single aberrant structure. However, from the perspective of glycosphingolipidomics, all these aberrant fucosylated GSLs subjected to MS2 analysis could produce a same major fragment ion at m/z 433 representing terminal Fuc-Gal-OH (Fig. 3). We thus performed MSn analysis of a standard GSL which contains terminal Fucα2Gal: Fucα2Galβ4Glcβ1Cer (provided by Dr. ChengFeng Xia, Kunming Institute of Botany, Chinese Academy of Sciences). Figure 5A showed the standard GSL and all the tumour GSLs containing terminal Fuc-Gal-OH (m/z 433 by MS2 analysis) generated MS3 ion fragments containing m/z 211 (Fuc- B ion), 229 (Fuc-OH C ion), and 245 (-Gal-OH Y ion).


Figure 5. Identification of aberrant GSLs containing terminal Fucα1-2Gal- structure in solid HCC tissues. (A) MS4 analyses of the standard GSL (Fucα2Galβ4Glcβ1Cer) and representative ions from three groups of aberrant tumour GSLs (m/z 1324, m/z 1634, m/z 1808, and m/z 1838 respectively). (B) MS1 analysis of HCC-contained GSLs treated with or without α1,2 fucosidase.

Download figure to PowerPoint

We also treated the HCC GSLs with α1,2 fucosidase that catalyzes the hydrolysis of terminal fucose residues linked to galactose via α1-2 linkage, then permethylated the treated GSLs for mass spectrometry analysis. Samples treated with α1,2 fucosidase showed loss of most of the peaks in range m/z 1600–2000 as compared to untreated control (Fig. 5B), indicating these missing peaks were fucosylated GSLs as described above.

FUT1 and FUT2 are the fucosyltransferases responsible for the synthesis of Fucα1-2 Gal- structures. We have measured the levels of FUT1 and FUT2 by reverse transcription real-time PCR. The levels of FUT1 and FUT2 in tumour tissues were significantly higher than those in paired peritumornal tissues (Table S3).

Figure 6 showed the receiver operating characteristic (ROC) analysis of three groups of aberrant fucosylated GSLs (m/z 1086, m/z 1260 and m/z 1290) and Fucα1-2 Gal- structure, comparing the tumour tissues and paired peritumornal tissues. The area of m/z 1086 under the curve (AUC) value was 0.809, while the AUC of m/z 1260 was 0.722, and the AUC of m/z 1290 was 0.953. We also performed the ROC analysis of all GSL structures containing Fucα1-2 Gal- including m/z 1086, m/z 1260, and m/z1290, the AUC of all GSLs containing Fucα1-2 Gal- structure was 0.864. These results suggest that the GSLs containing terminal Fucα1-2 Gal- structure may serve as a potential marker for HCC.


Figure 6. ROC analysis for three novel groups fucosylated GSLs and the Fucα1-2 Gal- structure. (A) m/z 1086, (B) m/z 1260, (C) m/z 1290, (D) Fucα1-2 Gal- structure.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

New targets are urgently needed for diagnosing the highly malignant HCC and for treating it. Recently, monoclonal antibodies against vascular endothelial growth factor and its receptor, as well as small-molecule kinase inhibitors of the vascular endothelial growth factor receptor, have been shown to extend survival in patients with advanced HCC and has opened the door for other strategies targeting membrane proteins and receptors. If they are involved in cancer cell growth and metastasis, cell surface molecules expressed by hepatocarcinoma cells might be targets in the development of therapeutic antibodies. Animal studies have indicated that antifucosylated glycan antibodies have been effective in treating cancer. For example, anti-Ley monoclonal antibodies were found to inhibit RMG-1-H ovarian cancer cells in vitro [27].

In this study, we identified the cancer-related expression of a group of fucosylated GSLs in HCC firstly by glycosphingolipidomic methods. Three groups of fucosylated GSLs were identified, Fuc(Hex)3HexNAc-Cer, Fuc2(Hex)3HexNAc-Cer, and Fuc(Hex)4HexNAc-Cer. We found that Fuc(Hex)3HexNAc-Cer represents Fucα2Galβ3GlcNAcβ3Galβ4Glcβ1Cer, Fucα2Galβ4GlcNAcβ3Galβ4Glcβ1Cer, Fucα3(Galβ4)GlcNAcβ3Galβ4Glcβ1Cer, and Fucα4(Galβ3)GlcNAcβ3Galβ4Glcβ1Cer. Fuc2(Hex)3HexNAc-Cer represents Fucα2Galβ3(Fucα4)GlcNAcβ3Galβ4Glcβ1Cer and Fucα2Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer. Fuc(Hex)3HexNAc-Cer represents Fucα2Galβ3GalNAcβ3Galα4Galβ4Glcβ1Cer. Moreover, the ion m/z 433 was main fragment ion in corresponding MS2 spectra, indicating that all aberrant fucosylated GSLs contained terminal FucGal- structure in non-reducing end. The aberrant fucosylated GSLs displayed similar MS2 pattern with GSL standards containing Fucα2Gal-OH. The result of GSL standards containing Fucα2Gal- was consistent with a previous report [28].

Globo H was previously reported as a target for therapy in breast cancer [29]. In this study, we are the first to identify that Globo H is specifically expressed in HCC samples but completely absent in all peritumoral tissues. In addition to HCC, Globo H is expressed on the cell surface of breast, prostate, and lung carcinomas [30-32]. The restricted tissue distribution of Globo H suggests that it might be a suitable target for oncologic clinical applications.

More important, we found the terminal Fucα2Gal- was the dominant common structures of the HCC-associated GSL antigens. Thus, it might be feasible to produce antiterminal Fucα2Gal- monoclonal antibodies and vaccines that can recognize multiple types of HCC-associated GSL antigens containing terminal Fucα2Gal- residues for cancer treatment.

Fucosylation plays important roles in cell signalling, such as O-fucosylation for ligand mediated Notch signalling pathway [33, 34]. The protein O-fucosyltransferase-1 may serve as a possible therapeutic target for Notch-related human cancer [35]. Inactivation of FUT1 and FUT2 expression could reduce cell adhesion efficiency and inhibit cell growth [36]. FUT6, a member of the α1,3/4 fucosyltransferase subfamily, could regulate the P13K/AKT signalling pathway in HCC cells [37]. Aberrant fucosylation was reported in various tumours, especially in HCC [38, 39]. Fucosylated haptoglobin has been proposed as diagnostic marker for liver cancer and pancreatic cancer [40, 41]. Although the expression of FUT1 and FUT2 may be distinct in each individual patient, our results of real-time PCR suggest that the FUT1 and FUT2 may also serve as a potential target for HCC therapy.


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

We thank Elizabeth L Hess for critical reading of the manuscript. We thank First Affiliated Hospital of Soochow University and Tongji University hospital for providing samples.

Financial support: This work is supported by the National Natural Science Foundation of China [31000370], the Natural Science Foundation of Jiangsu province [BK2011288]; the National Science and Technology key projects during Twelfth Five-Year Plan Period of China [2012ZX09401-014, 2012ZX09502001-001], the Program for Changjiang Scholars and Innovative Research Team in University [IRT1075]; the Priority Academic Program Development of Jiangsu Higher Education Institutions. DZ is supported by MD Anderson Cancer Center and NIH grants AI079232. MD Anderson Cancer Center is supported in part by NIH grant CA16672.

Conflict of interest: DZ is President of NanoCruise Pharmaceutical Ltd., and an inventor involved in patents related to technologies mentioned in this article, issued or in application. The rest of the authors do not have any disclosures to report.


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information
  • 1
    Spangenberg HC, Thimme R, Blum HE. Targeted therapy for hepatocellular carcinoma. Nat Rev Gastroenterol Hepatol 2009; 6: 42332.
  • 2
    Huynh H. Tyrosine kinase inhibitors to treat liver cancer. Expert Opin Emerg Drugs 2010; 15: 1326.
  • 3
    Heo J, Breitbach CJ, Moon A, et al. Sequential therapy with JX-594, a targeted oncolytic poxvirus, followed by sorafenib in hepatocellular carcinoma: preclinical and clinical demonstration of combination efficacy. Mol Ther 2011; 19: 11709.
  • 4
    Höpfner M, Schuppan D, Scherübl H. Growth factor receptors and related signalling pathways as targets for novel treatment strategies of hepatocellular cancer. World J Gastroenterol 2008; 14: 114.
  • 5
    Xu C, Lee SA, Chen X. RNA interference as therapeutics for hepatocellular carcinoma. Recent Pat Anticancer Drug Discov 2011; 6: 10615.
  • 6
    Breous E, Thimme R. Potential of immunotherapy for hepatocellular carcinoma. J Hepatol 2011; 54: 8304.
  • 7
    Hakomori S. Glycosylation defining cancer malignancy: new wine in an old bottle. Proc Natl Acad Sci USA 2002; 99: 102313.
  • 8
    Comunale MA, Lowman M, Long RE, et al. Proteomic analysis of serum associated fucosylated glycoproteins in the development of primary hepatocellular carcinoma. J Proteome Res 2006; 5: 30815.
  • 9
    Goldman R, Ressom HW, Varghese RS, et al. Detection of hepatocellular carcinoma using glycomic analysis. Clin Cancer Res 2009; 15: 180813.
  • 10
    Liu XE, Desmyter L, Gao CF, et al. N-glycomic changes in hepatocellular carcinoma patients with liver cirrhosis induced by hepatitis B virus. Hepatology 2007; 46: 142635.
  • 11
    Ressom HW, Varghese RS, Goldman L, et al. Analysis of MALDI-TOF mass spectrometry data for discovery of peptide and glycan biomarkers of hepatocellular carcinoma. J Proteome Res 2008; 7: 60310.
  • 12
    Tang Z, Varghese RS, Bekesova S, et al. Identification of N-glycan serum markers associated with hepatocellular carcinoma from mass spectrometry data. J Proteome Res 2010; 9: 10412.
  • 13
    Yamashita K, Koide N, Endo T, Iwaki Y, Kobata A. Altered glycosylation of serum transferrin of patients with hepatocellular carcinoma. J Biol Chem 1989; 264: 241523.
  • 14
    Campion B, Léger D, Wieruszeski JM, Montreuil J, Spik G. Presence of fucosylated triantennary, tetraantennary and pentaantennary glycans in transferrin synthesized by the human hepatocarcinoma cell line Hep G2. Eur J Biochem 1989; 184: 40513.
  • 15
    Ohno M, Nishikawa A, Koketsu M, et al. Enzymatic basis of sugar structures of alpha-fetoprotein in hepatoma and hepatoblastoma cell lines: correlation with activities of alpha 1-6 fucosyltransferase and N-acetylglucosaminyltransferases III and V. Int J Cancer 1992; 51: 3157.
  • 16
    Nouso K, Kobayashi Y, Nakamura S, et al. Prognostic importance of fucosylated alpha-fetoprotein in hepatocellular carcinoma patients with low alpha-fetoprotein. J Gastroenterol Hepatol 2011; 26: 1195200.
  • 17
    Fang M, Dewaele S, Zhao YP, et al. Serum N-glycome biomarker for monitoring development of DENA-induced hepatocellular carcinoma in rat. Mol Cancer 2010; 9: 215.
  • 18
    Dai Z, Liu YK, Cui JF, et al. Identification and analysis of altered alpha1,6-fucosylated glycoproteins associated with hepatocellular carcinoma metastasis. Proteomics 2006; 6: 585767.
  • 19
    Souady J, Hülsewig M, Distler U, et al. Differences in CD75s- and iso-CD75s-ganglioside content and altered mRNA expression of sialyltransferases ST6GAL1 and ST3GAL6 in human hepatocellular carcinomas and nontumoral liver tissues. Glycobiology 2011; 21: 58494.
  • 20
    Zhao Y, Zhang JR, Fan SD, Li P. Isolation and purification of hepatoma associated fucosylated neutral glycosphingolipids and its effect on cytotoxicity of NK and LAK cells. J Dig Dis 1998; 6: 94850.
  • 21
    Li Y, Zhou D, Xia C, Wang PG, Levery SB. Sensitive quantitation of isoglobotriaosylceramide in the presence of isobaric components using electrospray ionization-ion trap mass spectrometry. Glycobiology 2008; 18: 16676.
  • 22
    Ciucanu I, Kerek F. A simple and rapid method for the permethylation of carbohydrates. Carbohydr Res 1984; 131: 20917.
  • 23
    Ciucanu I, Costello CE. Elimination of oxidative degradation during the per-O-methylation of carbohydrates. J Am Chem Soc 2003; 125: 162139.
  • 24
    Harvey DJ. Quantitative aspects of the matrix-assisted laser desorption mass spectrometry of complex oligosaccharides. Rapid Commun Mass Spectrom 1993; 7: 6149.
  • 25
    Naven TJ, Harvey DJ. Effect of structure on the signal strength of oligosaccharides in matrix-assisted laser desorption/ionization mass spectrometry on time-of-flight and magnetic sector instruments. Rapid Commun Mass Spectrom 1996; 10: 13616.
  • 26
    Papac DI, Wong A, Jones AJ. Analysis of acidic oligosaccharides and glycopeptides by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Anal Chem 1996; 68: 321523.
  • 27
    Li FF, Lin B, Hao YY, et al. Inhibitory effect of anti-Lewis y antibody on α1,2-fucosyltransferase gene transfected human ovarian cancer cells in vitro. Chin J Cell Mol Immunol 2008; 24: 43542.
  • 28
    Zhang H, Singh S, Reinhold VN. Congruent strategies for carbohydrate sequencing. 2. FragLib: an MSn spectral library. Anal Chem 2005; 77: 626370.
  • 29
    Chang WW, Lee CH, Lee P, et al. Expression of Globo H and SSEA3 in breast cancer stem cells and the involvement of fucosyl transferases 1 and 2 in Globo H synthesis. Proc Natl Acad Sci USA 2008; 105: 1166772.
  • 30
    Canevari S, Fossati G, Balsari A, Sonnino S, Colnaghi MI. Immunochemical analysis of the determinant recognized by a monoclonal antibody (MBr1) which specifically binds to human mammary epithelial cells. Cancer Res 1983; 43: 13015.
  • 31
    Mariani-Costantini R, Barbanti P, Colnaghi MI, et al. Reactivity of a monoclonal antibody with tissues and tumors from the human breast. Immunohistochemical localization of a new antigen and clinicopathologic correlations. Am J Pathol 1984; 115: 4756.
  • 32
    Zhang S, Cordon-Cardo C, Zhang HS, et al. Selection of tumor antigens as targets for immune attack using immunohistochemistry: i Focus on gangliosides. Int J Cancer 1997; 73: 429.
  • 33
    Stahl M, Uemura K, Ge C, et al. Roles of Pofut1 and O-fucose in mammalian Notch signaling. J Biol Chem 2008; 283: 1363851.
  • 34
    Linchao Lu, Stanley Pamela. Roles of O-Fucose Glycans in Notch Signaling Revealed by Mutant Mice. Methods Enzymol 2006; 417: 12736.
  • 35
    Yao D, Huang Y, Huang X, et al. Protein O-fucosyltransferase 1 (Pofut1) regulates lymphoid and myeloid homeostasis through modulation of Notch receptor ligand interactions. Blood 2011; 117: 565262.
  • 36
    Domenico P, Silvia A, Leonardo E, et al. RNA-Mediated gene silencing of FUT1 and FUT2 influences expression and activities of bovine and human fucosylated nucleolin and inhibits cell adhesion and proliferation. J Cell Biochem. 2010 111: 22938.
  • 37
    Guo Q, Guo B, Wang Y, et al. Functional analysis of a1,3/4-fucosyltransferase VI in human hepatocellular carcinoma cells. Biochem Biophys Res Commun 2012; 417: 3117.
  • 38
    Nakagawa T, Miyoshi E, Yakushijin T, et al. Glycomic analysis of alpha-fetoprotein L3 in hepatoma cell lines and hepatocellular carcinoma patients. J Proteome Res 2008; 7: 222233.
  • 39
    Ji J, Gu X, Fang M, et al. Expression of alpha 1,6-fucosyltransferase 8 in hepatitis B virus-related hepatocellular carcinoma influences tumour progression. Dig Liver Dis 2013; 45: 41421.
  • 40
    Miyoshi E, Shinzaki S, Moriwaki K, Matsumoto H. Identification of fucosylated haptoglobin as a novel tumor marker for pancreatic cancer and its possible application for a clinical diagnostic test. Methods Enzymol 2010; 478: 15364.
  • 41
    Pompach P, Brnakova Z, Sanda M, et al. Site-specific glycoforms of haptoglobin in liver cirrhosis and hepatocellular carcinoma. Mol Cell Proteomics 2013; 12: 128193.

Supporting Information

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Fig. S1. Level of fucosylated GSLs expressed by HCC and peritumoral tissues. Ion abundance of fucosylated GSLs was quantified as the percentage of total cellular GSL. Fucosylated GSLs were 3.3-fold more abundant in HCC samples than in paired peritumoral tissues from 32 patients (*< 0.05 by Student t-test). Bars indicate means.

Fig. S2. Decomposition pathway of aberrant pentosaccharide. The decomposition pathway of a representative ion (m/z 1634). Type I H: Fucα2Galβ3GlcNAcβ3Galα4Glcβ1Cer; Type II H: Fucα2Galβ4GlcNAcβ3Galα4Glcβ1Cer; Lex: Fucα3(Galβ4)GlcNAcβ3Galβ4Glcβ1Cer; Lea: Fucα4(Galβ3)GlcNAcβ3Galβ4Glcβ1Cer.

Fig. S3. Decomposition pathway of Leb and Ley. The decomposition pathway of a representative ion (m/z 1808). Leb: Fucα2Galβ3(Fucα4)GlcNAcβ3Galα4Glcβ1Cer; Ley: Fucα2Galβ4(Fucα3)GlcNAcβ3Galα4Glcβ1Cer.

Fig. S4. Decomposition pathway of Globo H. The decomposition pathway of a representative ion (m/z 1838). Globo H: Fucα2Galβ3GalNAcβ3Galα4Galβ4Glcβ1Cer.

Fig. S5. Analysis of a standard Globo H. Analysis of a Globo H standard by MALDI-TOF MS-MS. (A), specific fragment ions at m/z 443 and 660 and (B), the decomposition pathway of Globo H standard.

liv12265-sup-0002-TableS1-S3.docWord document83K

Table S1. Expression level of three groups of fucosylated GSLs in HCC and peritumoral tissues.

Table S2. Globo H expression in HCC and peritumoral tissues.

Table S3. Expression of Fucosyltransferase 1 and 2 in HCC and paired peritumoral tissues.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.