Glycosylation Failure Extends to Glycoproteins in Gestational Diabetes Mellitus

Normal and diabetic amniotic fluid samples.

The study protocol was approved by the institutional review board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster. A total of 35 amniotic fluid samples (20 normal and 15 GDM) were collected from women at term pregnancy during cesarean delivery at the Queen Mary Hospital, Hong Kong. The diagnosis of GDM was according to the World Health Organization criteria using a 2-h 75-g oral glucose tolerance test (OGTT), as described (3). Blood glucose levels >7.8 mmol/L were defined as having GDM. Two women with GDM were treated by insulin and 13 women with GDM were treated through diet control. Maternal and infant demographic information of normal and GDM participants are shown in Table 1. The two groups of women were similar in gravidity, parity, age, BMI, gestational age at birth, fasting plasma glucose, and placental weight and fetal birth weight and differed only in 2-h plasma glucose during the OGTT. The amount of the GdA isolated from each amniotic fluid sample was limited. Therefore, NGdA and DGdA were randomly pooled into five groups for experimentation.

Cell culture.

Human female peripheral blood was obtained from the Hong Kong Red Cross. Peripheral blood mononuclear cells (PBMCs) were isolated by the Ficoll-Paque PLUS method (GE Healthcare, Uppsala, Sweden). Peripheral blood NK (pbNK) cells with a purity of >95% were isolated from PBMCs using a negative isolation kit (Dynal Biotech, Oslo, Norway). These cells were cultured in 10% FBS supplemented with RPMI 1640 medium containing 500 IU/mL recombinant interleukin (IL)-2 (Sigma-Aldrich, St. Louis, MO).

PMBCs, Jurkat (T-lymphoma cells), and OE-E6/E7 (oviductal cell line) were cultured in RPMI-1640 medium (Sigma-Aldrich). TEV-1 (trophoblast cell line) was cultured in DMEM medium (Sigma-Aldrich). All the culture media were supplemented with 10% FBS.

Determination of glycodelin concentration in amniotic fluid.

The glycodelin content from the amniotic fluid of normal and GDM pregnancies was assayed by enzyme-linked immunosorbent assay (ELISA). Polyclonal goat anti-human glycodelin antibody (1 μg in 100 μL; R&D Systems, Minneapolis, MN)-coated assay wells were incubated successively with the amniotic fluid sample or GdA standard (0–10 μg/mL), murine monoclonal antiglycodelin antibody (clone F43-7F9), and horseradish peroxidase–conjugated goat anti-mouse IgG (1:5000 vol/vol; Sigma-Aldrich). The ELISA signal was determined by absorption at 450 nm using a microplate reader (MR5000; Dynatech Laboratories, Chantilly, VA). GdA content was expressed as a ratio to the total protein content of the amniotic fluid.

Purification of GdA.

NGdA and DGdA were purified from amniotic fluid as described (21). Desialylation of NGdA was performed using sialidase-coated agarose beads (Sigma-Aldrich) at 37°C for 18 h. The success of desialylation was verified by the decreased binding to wheat germ agglutinin lectin (11). The concentration of purified GdA was estimated by a protein assay kit (Bio-Rad, Hercules, CA). The purity of the isolated GdA was checked by 12% SDS-PAGE, whereas the identity of the isolated GdA was confirmed by peptide mass fingerprinting of the GdA band in SDS-PAGE using matrix-assisted laser desorption/ionization–tandem time-of-flight mass spectrometer (4800 MALDI-TOF/TOF; Applied Biosystems, Warrington, U.K.).

Lectin binding assay.

Glycosylation of glycodelins was studied by a lectin-binding assay as described (11). Briefly, assay wells coated with various lectins (7.5 pmol per well) were blocked by 5% casein before successive incubation with GdA (250 ng per well), monoclonal antiglycodelin antibody (1 μg per well), and horseradish peroxidase–conjugated anti-mouse IgG (1:5000 vol/vol; Sigma-Aldrich). The signal was developed with o-phenylenediamine (Sigma-Aldrich) and determined as described above.

Glycan sequencing by MALDI-TOF and MALDI-TOF/TOF.

Purified NGdA and DGdA (~15 μg each) were digested using trypsin (Sigma-Aldrich) and purified by a reverse-phase Sep-Pak C18 cartridge (Waters Corporation, Manchester, U.K.) as described (22). The N-glycans were then released by N-glycosidase F (Roche Applied Science, West Sussex, U.K.) and purified on a Sep-Pak C18 cartridge. The purified native N-glycans were permethylated as described (23), purified using a Sep-Pak C18 cartridge, dissolved in methanol, and mixed with 20 mg/mL 2,5-dihydrobenzoic acid in 70% methanol at a 1:1 ratio (vol/vol). The glycan-matrix mixture (1 μL) was spotted on a stainless-steel target plate and dried in a vacuum. MALDI-TOF and -TOF/TOF data were obtained using a 4800 MALDI-TOF/TOF mass spectrometer (AB Sciex UK Limited, Warrington, U.K.). Argon with a collision energy of 1 kV was used. The mass spectrometry (MS) and tandem MS (MS/MS) data obtained were analyzed using Data Explorer 4.9. The assignment of glycan sequence was done by manual annotation informed by knowledge of human biosynthetic pathways and aided by the glycobioinformatics tool, GlycoWorkBench (24). Relative quantification of the glycan abundance in a single spectrum was calculated relative to the total ion counts from all of the observed glycans (percentage of total abundance = [ion count of the glycan/total ion count of all glycans] × 100%).

Linkage analysis by gas chromatography–mass spectrometry.

The N-glycans were analyzed by gas chromatography–mass spectrometry (GC-MS) as described (23). The samples were dissolved in hexane before injection into the gas chromatographer–mass spectrometer (Clarus 500; PerkinElmer, Waltham, MA) fitted with an RTX-5 column (30 m × 0.32 mm internal diameter; Restek, Bellefonte, PA). The oven temperature was held at 90°C for 1 min and subsequently ramped to 290°C at a rate of 8°C per min. The acquired data were analyzed by TurboMass version 4.5.0.007 (Perkin Elmer Instruments, Shelton, CT).

XTT cell viability assay.

Cell viability was determined by the XTT assay (Roche Diagnostics, Basel, Switzerland), according to the manufacturer’s protocol. The absorbance of the resulting color product was measured at 450 nm with a λ correction at 595 nm. The changes in cell viability were expressed as the suppression index (S_I) using the following equation: suppression index (%) = (Abs GdA − Abs blank) × 100%.

Cell death analysis by flow cytometry.

Apoptotic and necrotic cell death were determined by flow cytometry using Yo-Pro-1 and propidium iodide dye (Invitrogen, Carlsbad, CA) and analyzed with a flow cytometer (Beckman Coulter, Fullerton, CA) equipped with a 488-nm argon laser. The fluorescence signal was measured using the 525-nm and 610-nm band pass filters and was analyzed by the Winlist software (Verity Software House, Topsham, ME).

Determination of cytokine production by ELISA.

The levels of IL-2 and IL-6 in the conditioned media were measured by ELISA (IL-2, BD Biosciences, Franklin Lakes, NJ; IL-6, Invitrogen, Carlsbad, CA), according to the manufacturer’s procedure. The absorbance derived from 3.3′,5.5′-tetramethylbenzidine was measured at 450 nm with a λ correction of 595 nm as above.

Glycodelin binding assay.

The binding of GdA was visualized by flow cytometric analysis. In brief, GdA was fluorescence labeled using the Alexa Fluor 488 microscale fluorescence labeling kit (Invitrogen) as described (25). The cells (5 × 10⁵) were incubated with 1 μg of Alexa Fluor 488–labeled GdA for 2 h, washed with PBS twice, and analyzed by flow cytometry as above.

Determination of placental sialidase activity.

Human placentae were obtained from 20 singleton pregnancies (10 normal pregnancies and 10 GDM pregnancies) after elective cesarean section at term before the onset of labor at the Queen Mary Hospital, Hong Kong. The tissues were dissected, washed with PBS, and homogenized in the presence of protease inhibitors and phosphatase inhibitor. The sialidase activity in the total cell lysates (50 μg) was determined by a fluorimetric assay using an artificial substrate, 4-methylumbelliferyl N-acetylneuraminic acid (Sigma-Aldrich) as described (11). Fluorescence emission was measured by a fluorometer with excitation at 340 nm and emission at 505 nm (Infinite F200; Tecan, Männedorf, Switzerland).

Data analysis.

All values were expressed as means ± SEM. The data were compared by ANOVA to discern differences between groups. Parametric Student t test or a nonparametric Mann-Whitney U test were used where appropriate as the posttest. A P value <0.05 was considered significant.

Purification and identification of NGdA and DGdA.

There was no significant difference in the amount of GdA in the amniotic fluid from normal (0.90 ± 0.32 μg/mg total protein; n = 20) and GDM (0.71 ± 0.56 μg/mg; n = 15) pregnancies. Purified NGdA and DGdA had a similar molecular size in SDS-PAGE (~30 kDa, Supplementary Fig. 1) and peptide mass fingerprinting in MS/MS (Supplementary Fig. 1). The peptide mass fingerprints of both NGdA and DGdA were significantly matched to the product of the progesterone-associated endometrial protein gene (protein score = 162 for NGdA, P < 0.001; protein score = 120 for DGdA, P < 0.001), a gene-encoding glycodelin.

NGdA and DGdA have different lectin-binding affinities.

DGdA reacted weakly to concanavalin A (ConA), suggesting a low abundance of mannose/glucose in its glycans. DGdA also had reduced affinity to sialic acid (N-acetyl-5-neuraminic acid) and N-acetylglucosamine (GlcNAc)-binding lectin and wheat germ agglutinin (WGA) (Table 2). Because NGdA and DGdA had similar affinity to succinylated wheat germ agglutinin (S-WGA), a lectin that binds to N-acetylglucosamine, the reduced binding affinity of DGdA to WGA reflected a lower amount of sialylated glycans in the molecule.

Differential glycomics between NGdA and DGdA.

Glycomics analysis was performed using strategies previously optimized for GdA characterization (11). The permethylated N-glycans were subjected to MALDI-MS profiling and MALDI-MS/MS sequencing. Linkage analysis using GC-MS was subsequently carried out on the remaining samples. The complete MALDI spectra of NGdA and DGdA glycans are shown in single panels in Fig. 1 to facilitate visual comparison. For clarity, only the most informative molecular ions are annotated with glycan structures in this figure. Comprehensive annotations are shown on the magnified spectra, which are reproduced in Supplementary Fig. 2. Because the amounts of purified NGdA and DGdA were limited, only strong signals in the spectra could be sequenced by MALDI-TOF/TOF MS/MS. These components are flagged in Supplementary Fig. 2. Combining information on the glycan compositions, structure, and linkage, the structures were assigned manually, based on knowledge of human N-glycan biosynthetic pathways.

The N-glycans were highly complex, and most of the abundant glycans were biantennary and triantennary glycans. Some mass peaks of the same m/z value contained mixtures of glycans (e.g., glycans of m/z of 1,662, 1,836, 1,866, 2,070, 2,244, 2,285, 2,489, 2,693, 2,717, 2,850, and 3,253). The glycans of both NGdA and DGdA comprised high-mannose, hybrid, and abundant complex structures. NGdA from term pregnancy shared the same glycosylation pattern as that reported for GdA from midtrimester pregnancy (11); both had heavily sialylated glycans, Lewis X glycans, the Sda (NeuAcα2-3[GalNAcβ1-4]Gal) epitope in the high–molecular weight glycans and bisecting N-acetylglucosamine on some of the biantennary glycans. Some common characteristics of N-glycans in many glycoproteins were observed, such as lacNAc and lacdiNAc as antenna backbones, sialylated lacNAc or lacdiNAc antennae, fucosylated lacNAc or lacdiNAc, and fucosylated core GlcNAc. Linkage analysis by GC-MS (Supplementary Table 1) showed that DGdA contained terminal fucose, mannose, galactose, N-acetylglucosamine and N-acetylgalactosamine, 2-linked mannose, 6-linked galactose, 3,4-linked galactose, 2,4-linked mannose, 3,6-linked mannose, 3,4,6-linked mannose, 4-linked GlcNAc, and 4,6-linked GlcNAc. In addition, linkage analysis gave evidence for the specific glycan structures of GdA, such as Sda (3,4-linked galactose), bisecting GlcNAc (3,4,6-linked mannose), core fucose (4,6-linked GlcNAc), and terminal sialic acid (6-linked galactose). Comparison of these DGdA linkage data with published linkage data for NGdA (11) reveals substantially lower levels of 6-linked galactose and 6-linked N-acetylgalactosamine in the former compared with the latter. This is indicative of lower levels of α2-6 sialylation in DGdA compared with NGdA.

The relative abundances of the glycans of NGdA and DGdA were compared (Fig. 1). Some important differences were found: 1) Of 147 glycoforms identified, >55% were found in both NGdA and DGdA (Supplementary Fig. 2). 2) The most notable difference between the two samples was that DGdA contained much lower levels of sialylated glycans (DGdA: 24.7%; N.GdA: 53.4%). From the MALDI-MS spectra, many of the strong peaks of NGdA shifted toward low molecular weight by 361 mass units in DGdA, corresponding to the molecular weight of permethylated sialic acid, indicating that these glycans lacked sialic acid in DGdA (see the orange, red, and blue panels in Fig. 1). 3) DGdA contained a lower abundance (1.9 vs. 10.3% of NGdA) of high-mannose glycans (m/z 1,580, 1,784, and 1,988). This is consistent with the reduced ConA binding affinity of DGdA in the lectin-binding assay. The low abundance of high-mannose glycans suggested that they were unlikely to occupy a glycosylation site fully, as previously described for the Asn28 site of GdS, which is exclusively occupied by high-mannose glycans (10). 4) DGdA contained more glycans capped with the Sda epitope (e.g., m/z 2,717, 2,850, 2,891, 3,253, 3,300, 3,341, 3,545, and 4,267; see the magenta panel in Fig. 1), which was a recently identified characteristic of the female glycodelins (11). The smaller amount of sialylated glycans in DGdA compared with NGdA was further confirmed by the decreased WGA-binding of DGdA in a second population of normal and GDM patients (Supplementary Table 2).

Reduced cytotoxicity of DGdA and desialylated NGdA on human lymphocytes.

Treatment with both NGdA and DGdA at concentrations of ≥0.01 μg/mL for 36 h significantly (P < 0.05) decreased the viability of PBMCs, the cytotoxicity of the latter was significantly (P < 0.05) lower than that of the former at concentrations of 0.01 and 0.1 μg/mL (Table 3). NGdA, but not DGdA, at a concentration of 0.1 μg/mL, significantly (P < 0.05) decreased the viability of Jurkat cells. At 1 μg/mL, the cytotoxic effect of NGdA on Jurkat cells was also significantly (P < 0.05) higher than that of DGdA (viability: NGdA, 41.7 ± 4.9%; DGdA, 79.4 ± 8.2%). At the tested concentrations, neither NGdA nor DGdA affected the viability of TEV-1 and OE-E6/E7 cells.

Compared with NGdA, the cytotoxic effect of desialylated NGdA could only be observed at higher concentrations (PBMCs: ≥0.1 μg/mL; Jurkat cells: ≥1 μg/mL). In addition, the suppression index of desialylated NGdA was higher than that of NGdA when tested at the same concentrations.

DGdA and desialylated NGdA have impaired ability to induce cell death of lymphocytes.

Treatment with 0.1 μg/mL NGdA significantly (P < 0.01) decreased the viable population of PBMCs from 87.0 ± 5.7% to 26.9 ± 2.9% in the YoPro-PI assay (Table 4 and Supplementary Fig. 3). The corresponding decrease by DGdA was significantly (P < 0.05) smaller (from 87.0 ± 5.7% to 42.4 ± 15.5%), indicating a lower cytotoxic activity of DGdA on PBMCs.

Treatment with 0.1 μg/mL NGdA for 48 h significantly (P < 0.05) decreased the percentage of viable Jurkat cells from 86.6 ± 2.4% to 77.2 ± 3.5% (Table 4 and Supplementary Fig. 3). By contrast, DGdA at the same concentration had no significant effect on the viability of these cells (82.9 ± 1.8%). A differential response (P < 0.05) of Jurkat cells to NGdA and DGdA was also observed at the concentration of 1 μg/mL.

After desialylation, the ability of NGdA to induce cell death of PBMCs and Jurkat cells was abolished (Table 4 and Supplementary Fig. 3). No significant difference was observed on the viable population after treatment with desialylated NGdA when compared with the control.

DGdA and desialylated NGdA have reduced ability in modulating the cytokine secretion by lymphocytes and NK cells.

NGdA dose-dependently inhibited IL-2 secretion by PBMCs and Jurkat cells (Table 5). DGdA and desialylated NGdA had a significantly (P < 0.05) lower suppressive effect on IL-2 secretion than that of NGdA at the same concentration in either cell types. Neither NGdA nor DGdA affected cell viability within the treatment period (data not shown). For IL-6, DGdA and desialylated NGdA had a significantly (P < 0.05) smaller stimulatory effect on pbNKs when compared with NGdA at the concentration of 1 μg/mL (Table 5).

DGdA and desialylated NGdA have reduced binding affinity to the lymphocytes.

The binding affinity of DGdA on Jurkat and pbNK cells was significantly lower (P < 0.05) than that of NGdA (Fig. 2). The binding of DGdA on PBMCs was also somewhat lower, though the difference did not reach statistical significance (NGdA: 35.7 ± 5.3%; DGdA: 27.6 ± 6.0%). Upon desialylation, the binding of desialylated NGdA became significantly reduced in all the cells tested.

Placental tissue of GDM has a higher sialidase activity.

The sialidase activity of the GDM placental tissue increased the production of the 4-methylumbelliferone in a time-dependent manner and was significantly (P < 0.05) higher than that of the normal placental tissue (Fig. 3).