Plasma High-Mannose and Complex/Hybrid N-Glycans Are Associated with Hypercholesterolemia in Humans and Rabbits

N-glycans play important roles in various pathophysiological processes and can be used as clinical diagnosis markers. However, plasma N-glycans change and their pathophysiological significance in the setting of hypercholesterolemia, a major risk factor for atherosclerosis, is unknown. Here, we collected plasma from both hypercholesterolemic patients and cholesterol-fed hypercholesterolemic rabbits, and determined the changes in the whole-plasma N-glycan profile by electrospray ionization mass spectrometry. We found that both the hypercholesterolemic patients and rabbits showed a dramatic change in their plasma glycan profile. Compared with healthy subjects, the hypercholesterolemic patients exhibited higher plasma levels of a cluster of high-mannose and complex/hybrid N-glycans (mainly including undecorated or sialylated glycans), whereas only a few fucosylated or fucosylated and sialylated N-glycans were increased. Additionally, cholesterol-fed hypercholesterolemic rabbits also displayed increased plasma levels of high-mannose in addition to high complex/hybrid N-glycan levels. The whole-plasma glycan profiles revealed that the plasma N-glycan levels were correlated with the plasma cholesterol levels, implying that N-glycans may be a target for treatment of hypercholesterolemia.


Introduction
N-glycans constitute a basic component of cell membrane and secreted proteins and play important roles in many physiological and pathological processes [1][2][3]. N-glycans are covalently attached to proteins at asparagine residues by an N-glycosidic bond [4]. In general, physiological functions of N-glycans can be classified into two categories: (1) the structural and Guidelines for Animal Experimentation of Xi'an Jiaotong University and the Guide for the Care and Use of Laboratory Animals Published by the US National Institutes of Health.

Biochemical assays
The blood samples obtained from either humans or rabbits were collected into a tube containing EDTA. The plasma was separated after centrifuged at 3,000 rpm/min for 10 min. Total cholesterol (TC), triglycerides (TG), low density lipoprotein-cholesterol (LDL-C), high density lipoprotein-cholesterol (HDL-C) and glucose (Biosino Bio-technology and Science Inc., Beijing, China) contents were determined as described previously [19]. For analysis of apoB-containing lipoproteins including VLDL, IDL, and LDL subfractions, the plasma lipoprotein profile was analyzed using gradient gel electrophoresis (Lipoprint LDL Subfraction System, Quantimetrix Corporation, Redondo Beach, CA, USA) according to the manufacturer's instructions [20]. The subfractions were then quantified using NIH image program version 1.62 (Bethesda, MD, USA).

Pretreatment of plasma samples
The plasma was collected and dialyzed against Milli-Q water at 4°C for 72 hours. The water was changed every 12 hours. The dialyzed plasma samples were finally lyophilized and stored at −20°C until use [21].

Enzymatic release and purification of N-glycans
The lyophilized plasma sample (3 mg) was dissolved in 300 μl of protein denaturation solution, containing 0.4 M DL-Dithiothreitol and 5% sodium dodecyl sulfate, and denatured at 100°C for 10 min. After the samples became cooled at room temperature, 30 μl of 1 M sodium phosphate buffer (pH 7.5), 30 μl of 10% aqueous NonidetP-40 (v/v), and 1 μl of Peptide -N-Glycosidase F solution (500 units) (New England BioLabs, Ipswich, MA, USA) were added and then incubated in 37°C water bath for 24 h, which was stopped by boiling at 100°C for 5 min. To remove protein, the samples were loaded onto a SepPak C18 solid phase extraction column (Waters, Milford, MA, USA) and the N-glycan fractions were desalted using a graphitized carbon SPE column (Alltech Associates, Deerfield, IL, USA). The target N-glycans were eluted with 25% acetonitrile (Fisher Scientific, Fairlawn, NJ, USA) containing 0.01% trifluoroacetic acid. The eluates were dried under a stream of nitrogen gas and stocked until use [22,23].

ESI-MS
N-glycan analysis was performed with an LTQ XL linear ion trap electrospray ionization mass spectrometer (Thermo Scientific, Waltham, MA, USA). Briefly, N-glycan samples were directly infused via a Rheodyne loop injector with a volume of 2 μl and subsequently brought into the electrospray ion source by a stream of 50% methanol (v/v) at a flow rate of 200 μl/min. The molecular ions formed in the ion source were inspired and transferred to the ion trap via capillary and quadrupole. Therefore, molecular ions were detected. The spray voltage was set at 4 kV, with a sheath gas (nitrogen gas) flow rate of 30 arb., an auxiliary gas (nitrogen gas) flow rate of 5.0 arb., a capillary voltage of 37 V, a tube lens voltage of 250 V, and a capillary temperature of 37°C. For MS/MS analysis, N-glycans were subjected to fragmentation by collision induced decomposition, with helium as the collision gas. Collision parameters were left at default values with a normalized collision energy degree of 60 and an isotope width of m/z 3.00. Activation Q was set at 0.25, and activation time, at 30 ms. The MS and MS/MS data were recorded using LTQ Tune software (Thermo Scientific, Waltham, MA, USA) [22][23][24]. Glycan compositions and sequences were assigned manually. Then, we validated glycan structures either with reference to previous literature reports or by checking with GlycoWorkbench in databases such as CFG, CarbBank, and GLYCOSCIENCES.

Statistical Analysis
Results are presented as mean ± SEM. Statistical analysis was performed by two-tailed Student's t test using GraphPad Prism 5 software (GraphPad Software Inc., San Diego, CA, USA). A P value less than 0.05 was considered statistically significant.

Plasma lipids in hypercholesterolemic patients
Compared with healthy subjects, the hypercholesterolemic patients exhibited dramatically increased plasma lipid levels, including a 1.8-fold increase of TC, 1.6-fold increase of TG, and 2.4-fold increase of LDL-C, while no changes of HDL-C and glucose were observed ( Table 1). Analysis of plasma lipoprotein profiles revealed that all apoB-containing particles (VLDLs, IDLs, and LDLs) were markedly increased in hypercholesterolemic patients compared with the healthy controls (Fig 1A-1C). Furthermore, LDL subfractions, especially those with smaller size (fractions 3, 4 and 5) remarkably appeared in the hypercholesterolemic patients.
Plasma N-glycan profiling in the hypercholesterolemic patients Although 32 N-glycans were present in both healthy subjects and hypercholesterolemic patients, there was a difference in contents of these glycans between two groups. We found that 27 kinds of N-glycans were significantly higher in plasma of hypercholesterolemic patients than those of the control group (Fig 3). Heat map showed that three groups of N-glycans were apparently increased in the hypercholesterolemic patients (S2A

Plasma lipids in the cholesterol-fed rabbits
To determine whether the plasma N-glycan compositions can be induced experimentally, we compared plasma derived from cholesterol-fed rabbits with that from normal rabbits. Cholesterol-feeding dramatically increased plasma lipid levels in rabbits: 33-fold increase of TC, 3-fold increase of TG, 35-fold increase of LDL-C whereas HDL-C levels were decreased compared to the control rabbits (Fig 4A-4D). Analysis of lipoprotein profiles revealed that all apoB-containing particles were markedly increased in cholesterol-fed rabbits (Fig 4E-4G). Similar to human hypercholesterolemia, small-sized LDLs (designated as fractions 3-4) became prominent whereas they were not present in normal rabbits (Fig 4E-4G).

Discussion
Hypercholesterolemia is a major risk factor for the development of atherosclerosis; therefore, to find a new diagnostic maker is essential for the prevention and the treatment for cardiovascular disease [25][26][27]. Global N-glycan profiling of human plasma based on mass spectrometry techniques has provided an alternative means to explore potentially promising markers for human diseases [28]. In the current study, we characterized the whole-plasma N-glycan profiling of both hypercholesterolemic patients and rabbits, and demonstrated for the first time that plasma levels of high-mannose and complex/hybrid N-glycans are associated with plasma cholesterol levels.
In human hypercholesterolemia, the major atherogenic lipoproteins are LDLs whereas in cholesterol-fed rabbits, the major atherogenic lipoproteins are those of intestinally and hepatically derived remnant lipoproteins called β-VLDLs. Although lipoprotein are quite different in compositions because of different pathogenesis between hypercholesterolemic patients and experimental rabbits, high plasma N-glycan levels are correlated with both human and rabbit plasma cholesterol levels, suggesting that N-glycans are simultaneously changed under the hypercholesterolemic state regardless of lipoprotein profiles (LDL vs β-VLDL). However, 32 N-glycans were detected in human plasma whereas only 24 N-glycans were observed in rabbit plasma, suggesting that the profiles of plasma lipoprotein may interacts with N-glycan compositions or there is a species difference between human and rabbit. In human plasma, the decorated complex and hybrid N-glycans accounted for main component, whereas high-mannose and undecorated complex N-glycans were major component in rabbits plasma, implying the different complexity of glycome in various species. Interestingly, either in hypercholesterolemic patients or high-cholesterol fed rabbits, the levels of N-glycan were higher than that of normal group. Pearson correlation coefficient showed that N-glycan levels were significantly positively correlated with cholesterol levels both in human (0.48 r 0.69) and rabbits (0.65 r 0.85) (P<0.05). However, it is currently unknown that how hypercholesterolemia affects N-glycans or how these two interacts with each other. It also remains unclear what the pathophysiological significance of plasma N-glycans is. Do N-glycans play any roles in the pathogenesis of atherosclerosis? Or can we use these N-glycans as a marker for diagnosis of atherosclerotic disease? Many plasma glycoproteins with N-glycosylation change are implicated in serious pathological conditions such as cancers, cardiovascular diseases and autoimmune diseases [29][30][31][32][33][34]. Their glycosylation pattern could be used as disease biomarkers for early diagnosis or anchor points for targeted treatments. Fox example, the fucosylation and sialylation of Alpha-1-acid glycoprotein are significantly increased in rheumatoid arthritis [34,35]. Changes in the glycosylation of Alpha-2-macroglobulin and Alpha-2-HS-glycoprotein have been associated with cancer and autoimmune diseases [34]. The N-glycosylation of lectin-like oxidized low density lipoprotein receptor-1 has been shown to modulate the pathogenesis of atherosclerosis [29,31]. N-glycosylation defect on scavenger receptor expressed by endothelial cells(SREC-I) could impair the uptake of modified LDL into macrophages and endothelial cells [32]. Treatment with tumor necrosis factor α (TNFα) in human umbilical vein endothelial cells (HUVECs) was shown to increase the surface expression of high mannose/hybrid N-glycans [36]. The increased high mannose/hybrid N-glycans acted as ligands further contribute to monocytes rolling and adhesion to endothelial cells, which process the development of atherosclerosis [36]. Macrophage-expressed mannose-binding lectin (MBL) was also found to be enriched in early stage of atherosclerotic lesions [37]. When exposed to oscillatory shear or under TNFα stimulation, mannose-specific lectin staining intensity was increased in human aortic endothelial cells and at sites of early atherosclerotic lesions in both mice and human arteries [38]. Conversely, enzymatic removal of high-mannose N-glycans, or masking mannose residues with lectins, strongly decreased monocyte adhesion under the flow [38]. Therefore, N-glycans profile is analogous to a molecular zip code that regulates leukocyte trafficking [39].
apoB-100 plays an important role in the assembly of VLDL and lipoproteins, and transports the majority of plasma cholesterol. Its N-glycosylation sites are occupied by high mannose, diantennary complex, and hybrid type structures [34,40]. Previous studies revealed that apoB-100 glycan modifications can induce atherogenic properties on LDL [40]. In our study, the abundant mannose and complex/hybrid N-glycans were detected in hypercholesterolemic patients and rabbits compared with their corresponding controls. Meanwhile, severe atherosclerotic plaque was also exhibited in the cholesterol-fed rabbits (data are not shown). N-glycan structure analyses revealed that the core-fucosylated bi-antennary is the common major structure at all glycosylation sites [32]. We also found that several of the fucosylated, or fucosylated and sialylated modified N-glycans in hypercholesterolemic patients showed much higher level than healthy controls. These results suggested that the abundance of Nglycans in plasma may be associated with glycosylation modifications of several lipoproteins, such as apoB-100 or other plasma secreted proteins. It remains to be verified, however, what the pathophysiological significance of high glycan abundance in hypercholesterolemia is in future.
In conclusion, we have shown for the first time that plasma N-glycans are associated with hypercholesterolemia. Although the molecular mechanisms and pathophysiological significance remain unclear, these results provide an important clue that plasma N-glycans can become a new target for diagnosis of hypercholesterolemia in future.