Human Erythrocyte Surface Fucose Expression Increases with Age and Hyperglycemia

Reactive oxygen species and other free radicals, together with glucose and its metabolites are believed to play important roles in the aging process. The carbohydrate components of glycosylated proteins are important in mediating cell-cell interactions and a role has been suggested for them in the aging process. Erythrocytes are critical cells in the human body, heavily glycosylated and relatively easily available and so are good candidates to yield insights into how patterns of glycosylation change with age and disease. It has been claimed, based on a periodic acid Schiff assay, that human aging is associated with a decline of erythrocyte surface sialic acids. Plant lectins allow for more specic assays for glycans, including determining the linkage of sialic acids and analysis of single cells by ow cytometry.

associated with a decline of erythrocyte surface sialic acids. Plant lectins allow for more speci c assays for glycans, including determining the linkage of sialic acids and analysis of single cells by ow cytometry.

Results
Sialic acid speci c plant lectins, Maackia amurensis lectin II (MAL), binding to α-2,3 linked sialic acids and Sambucus nigra (SNA), α-2,6 sialic acids, were used to show that erythrocyte surface sialic acids have no signi cant associations with donor age. A combination of storage and cellular aging produces a speci c loss of α-2,6 sialic acids. By contrast, erythrocyte surface terminal fucoses increase signi cantly with age of donor. In order to determine which aspects of aging are important in determining this change, we investigated whether this novel human aging biomarker is associated with higher plasma glucose values, assessed by glycated hemoglobin (HbA1c) and reactive oxygen species (ROS) generation. Fucose levels were associated with HbA1c levels, but not ROS generation.

Conclusion
Our study identi es novel glycan-based biomarkers for human aging and disease. The simplicity of lectins are presented here as an attractive tool to study cellular medication in aging and disease.

Background
The main function of erythrocytes is to carry oxygen from the lungs to the rest of the body and remove carbon dioxide as a waste product. Erythrocytes are highly glycosylated, with glycophorin as the main carrier (1), especially sialic acids. Sialic acids are negatively charged sugars and confer a bulk surface charge causing repulsion between cells, so preventing agglutination (2). They are also important signalling molecules as they are found at the periphery of cell surfaces (3). Siglecs, sialic acid-binding immunoglobulin-type lectins, are surface receptors on immune cells that recognize sialic acids. For instance, erythrocyte sialoproteins inhibit neutrophil activation through interaction with Siglec-9(4) and inhibit B cell activation through interaction with the lectin CD22 (unpublished data).

Page 3/15
The phrase aging can be interpreted in two ways: "cell-aging", referring to increasing lifetime or replication of cells, or "organismal-aging", referring to the age of the body from which cells are derived.
There are many biologically signi cant cell-aging speci c changes described for erythrocytes, including loss of potassium ions, organic phosphates and water (5), decrease in cell surface area and volume (6) and decrease in the activity of Na+/K + ATPase and Na+/H + exchanger (7). Old RBCs can also undergo a form of programmed cell death in a process known as eryptosis, which is triggered by calcium action on Ca 2+ /K + channels, scramblase, calpain and other proteases, and leads to cell clearance (8). Long term blood storage causes oxidative damage to RBCs and leads to dysfunction in glucose metabolism that contributes to decreased metabolism of glutathione (GSH) (9).
Erythrocytes are excellent "reporter cells"(10) for organismal-aging as they are readily available. However, only limited organismal-aging changes have been described for erythrocytes. In mice, their antioxidant capacity decreases from 6 to 26 months of life (11). In humans, a strong negative association between donor age and erythrocyte surface sialic acids has been reported (12). Furthermore, increased sialic acids in the plasma of corresponding donors suggested a cleavage process, which was attributed to increased oxidative damage to erythrocytes in older donors. In cell-aging of broblasts, Itakura and colleagues reported a similar decline in sialic acids: α-2,3 sialylation, α-2,6 sialylation for both N-and O-glycans (13).
Fucose is a sugar that can be attached close to the core or outer antennae of N-glycan chains. Enzymes that remove or add fucose have been studied with respect to aging in different cell types (14,15).
However, fucose on erythrocytes has not been studied with reference to either "cell-aging" or "organismalaging".
Changes in carbohydrates have been described in association with several diseases such as cancer, Alzheimer's disease and diabetes mellitus (16)(17)(18). Our understanding of glycosylation in cancer has improved as diagnostic biomarkers involving carbohydrate changes have been established (19). For example, the core-fucosylated form of α-fetoprotein (AFP) in serum is associated with hepatocellular carcinoma (HCC) and can be used as a marker for diagnosis (20). Functionally, abnormal glycosylation in cancer can dysregulate cell signalling and communication, cell-matrix interactions, tumour cell dissociation and invasion (19). For example, the over expression of one component of the Nglycosylation machinery, the β1,6-N-acetylglucosaminyl transferase (GnT-V) enzyme, in glioma cells leads to focal adhesions and increased tumour cell invasion (21). Protein glycosylation has been reported to be defective in Alzheimer's disease (22). Numerous connections between glycosylation and diabetes are starting to be recognized: for example, high levels of mannose binding lectin (MBL) in sera are associated with vascular complications (17) and nephropathy (18). In diabetes mellitus, in addition to N-and Oglycosylation, glycation, the non-enzymatic addition of sugars to proteins is particularly important (23,24) and measurement of the glycated N-terminal amino acid of hemoglobin, HbA1c, is an important clinical diagnostic and prognostic marker of diabetes (25).
Cellular aging in other cell types such as mesenchymal stem cells have been shown to be associated with the loss of sialic acids (26). Several studies have investigated the relationship between age of erythrocytes and cell surface sialic acids. Methods using periodic acid have been the commonest way of measuring total sialic acid content from erythrocyte membrane preparations, known as ghosts. It has been presumed, but not proven, that periodic acid Schiff (PAS) assay is speci c for sialic acids. A more recently developed glycobiological technique known as DNA sequencer-assisted uorophore-assisted carbohydrate electrophoresis (DSA-FACE) has been employed to study N-glycan pro le changes during human aging and found age-dependent alterations in the ratio of two key structures of N-glycans beyond the age of 40 in humans (27). N-and O-glycome pro les of cell surfaces can be full analysed my mass spectrometry (28) as well as state of art technology such as solid phase chemoenzymatic method (29). Human and plant lectins are being becoming increasingly popular for studies of cell surface changes, as they offer direct binding to speci c terminal glycans (30,31). The use of lectins with ow cytometry to determine cell surface glycosylation expression has the advantage of accessing all cell surface glycans available for binding, thus mimicking cell-cell interactions, contrasting with conventional techniques such as PAS staining. In this study, we use plant lectins to evaluate carbohydrate expression on red blood cells during human organismal-aging.

Results
Neuraminidase sensitivity observed by lectin ow cytometry but not periodic acid Schiff assay Neuraminidase is an enzyme that speci cally removes terminal sialic acids from glycoproteins (32). Using this activity, we compared the classical PAS assay with the more modern approach of lectin ow cytometry for sialic acids on the cell surfaces of erythrocytes. Maackia amurensis lectin II (MAL) is a lectin speci c for α-2,3 linked sialic acids and Sambucus nigra agglutinin lectin (SNA) is speci c for α-2,6 linked sialic acids. Both lectins demonstrate high levels of binding by ow cytometry to untreated (sialylated) erythrocyte cell surfaces, but binding is lost once the erythrocytes have been treated with neuraminidase ( Figure 1a). However, while the PAS assay detected increasing amounts of sialic acids in the form of synthetic sialic acids (N-acetylneuraminic acid, NANA) and the glycoprotein fetuin, which contains high levels of sialic acid, the assay was not sensitive to neuraminidase treatment of whole erythrocytes ( Figure 1b). These data indicate that the PAS assay measures compounds in addition to surface accessible sialic acid residues, for which plant lectins MAL and SNA are more speci c.
Speci c cell surface α-2,6 sialic acid loss in storage plus cell-aging but no change to sialic acids in organismal-aging Using lectins MAL and SNA, we probed gravity Percoll fractionated stored erythrocytes from healthy donors to assess changes during cell-aging. It is well known that during cell-aging erythrocytes increase their density, become more rigid and fragile, and more deposition of C3 is observed. Our gravity based Percoll fractionation technique separates erythrocytes based on density into 5-7 fractions (Supplemental Figure 1a, 1b), with approximately 1% cells represented in the densest fractions (5, 6, and 7). We con rmed higher fragilities (Supplemental Figure 1c) and more C3 deposition (Supplemental Figure 1d) in the densest fractions. Although we failed to observe any differences in MAL binding to RBCs from any fraction (Figure 2a-d), we observed highly speci c SNA binding loss in the densest fractions (Figure 2a d). The data from Figure 2 are generated from erythrocytes stored for 7 or more days prior to fractionation: we do not observe MAL or SNA binding loss in fractionation of erythrocytes with no or reduced storage periods (data not shown). Western blotting identi es the SNA binding loss as arising from bands around 70kDa and 100kDa, corresponding to dimeric glycophorin A (GPA) and Band 3 respectively (Supplemental Figure 2). We also tried to reproduce these ndings in murine erythrocytes, but found they lacked SNA binding (Supplemental Figure 3a), suggesting poor lectin visibility of surface α-2,6 sialic acids. There were no changes in MAL or SNA binding when murine RBCs were fractionated by density (Supplemental Figure 3b-c).
With respect to organismal-aging, we observed that the binding of MAL and SNA to unfractionated erythrocytes showed no signi cant association with age of donors (n = 24) (Figure 3a, b), in contrast to the claims of Mehdi et al. (12). Overall, these data demonstrate, while there is a speci c loss of α-2,6 sialic acid under the combination of storage and cellular aging, there are no overall cellular or organismalaging related sialic acid changes to erythrocytes in humans.

Fucose speci c lectin binding increases with donor age
In contrast to sialic acid binding lectins, binding of those speci c for fucose on red blood cells increased signi cantly with age of donor (Figure 4a, b). Aleuria aurantia lectin (AAL) binds preferentially to fucose linked (α-1,6) to N-acetylglucosamine or fucose linked (α-1,3) to N-acetyllactosamine related structures (33). Ulex europaeus agglutinin I (U. Euros) binds to many glycoproteins and glycolipids containing αlinked fucose residues (Boyd, 1963). While U. Euros has a preference for blood group O erythrocytes, AAL has no speci city for blood groups (33). For both lectins, the associations between binding and age of donor were similar, which indicates that the underlying basis for the increase in binding with age is an increase in terminal fucoses.
Mass spectrometry based glycomic analysis identi es sialic acid and fucose containing structures on erythrocytes Binding of fucose-speci c lectin AAL is associated with hemoglobin HbA1c levels but not cellular ROS generation We investigated the underlying basis for the association between erythrocyte fucose expression and age by assessing whether it might be secondary to two aspects that have been linked with aging, glycation and oxidative stress (34)(35)(36)(37). Glycated hemoglobin (HbA1c) is a well-established clinical marker of longterm glycemic control (38), so we collected samples from patients who had had HbA1c levels determined, irrespective of age. Because of the age effect on fucose expression (Figure 3), we restricted our analysis to >45 years of age. We observed a positive association between HbA1C and AAL lectin binding levels (Figure 5a), which is the rst erythrocyte surface marker associated with HbA1c changes.
Oxidative stress is associated with diabetes (39,40) but no study has investigated the association between HbA1c levels and erythrocyte oxidation status. Here we show that the rate of internal reactive oxygen species production (ROS) is signi cantly elevated in red blood cells from patients with high HbA1c compared to a control HbA1c group (Figure 5b, c). Furthermore, there is a trend of positive association between HbA1c and ROS across both groups for patients of age 45 or above (Figure 5c). Yet, the correlation between ROS production and the binding of the fucose speci c lectin AAL (Figure 5d) does not appear signi cant. These results indicate that the relationship between HbA1c and AAL binding levels is likely to be independent of erythrocyte ROS generation capacity.

Discussion
Periodic acid Schiff assay versus lectin staining Traditionally, PAS assays have been used to stain for polysaccharides such as glycolipids and glycoproteins. They have also been used to measure sialic acids on red blood cells in a number of publications, despite no demonstration of speci city for the sialic acid sugar itself. We show here for the rst time that the PAS assay does not speci cally recognise sialic acids on the surface of erythrocytes by comparing staining of neuraminidase treated and untreated red blood cells, for which lectins such as MAL and SNA do demonstrate speci c binding.

Lack of age dependent sialic acid changes
In contrast to previous reports (12,41) of strong negative associations of sialic acids on the surface of erythrocytes with age of donor, we saw no signi cant associations between age and either α-2,3 or α-2,6 linked sialic acids. Lutz and Fehr(42) also reported, using PAS staining, a loss of sialic acids attached to glycophorins on aging erythrocytes. Such conclusions based on determining sialic acid changes through PAS staining may warrant re-examination. Nevertheless, upon storage and further fractionation based on density, we do observe a speci c loss of α-2,6 linked sialic acids, which appears to be associated with dimeric GPA and Band 3. This speci c glycan loss is only observed under both storage and fractionation, while α-2,3 linked sialic acids are consistently unchanged. Overall, our report indicates the stability of surface sialic acids during different types of aging of erythrocytes.

Fucose related changes
Unexpectedly, we showed a positive association of fucose expression on erythrocytes with age of donor. This has not been described before, although there are reports of decreases in plasma levels of N-glycans containing core fucose with aging (43). Perhaps both increased RBC surface fucose and reduced plasma fucose are explained by lower rates of enzymatic removal of RBC surface fucoses with age. However, since the lectins used in this study bind only to terminal fucose, we are not able to de ne the exact structure that is changing with age. N-glycomics mass spectrometry analysis provides an overview of possible fucose terminating structures, but further work would be needed to determine the structural changes with human organismal-aging.
Glycans as potential markers of aging Changes in glycosylation are caused by diabetes, which is more prevalent amongst older populations. Glycation affects proteins in two ways. Early glycation simply adds carbohydrates, non-enzymatically, to amino groups carried on lysine or arginine residues, as well as N terminal amino acids, such as HbA1c modi cation. Advanced glycation, which is associated with oxidation, causes much more signi cant structural damage by creating cross-links between and within proteins through formation of Advanced Glycation End Products (AGEs) (44,45). Our data show fucose expression on RBCs associated with the diabetes indicator, HbA1c, which is itself a product of glycation. Yet, fucose expression did not appear signi cantly related to oxidation capacity within RBC. These observations indicate that fucose levels are sensitive to the early phases of glycation.

Conclusions
Our study shows that by using more speci c reagents for sialic acids, cell-and organismal-aging do not produce signi cant changes to the expression of α-2,3 and α-2,6 sialic acids on erythrocytes. A combination of storage and cell-aging can cause speci c α-2,6 sialic acid loss, as shown by density fractionation. Surprisingly, human organismal-aging of erythrocytes appears to be associated with increased terminal fucose expression, which is also related to the early onset of diabetes, yet independent of erythrocyte oxidation capacity. These results illustrate the potential of using simple lectins to study the complex processes of ageing and disease.

Blood donors
Blood samples were retrieved from residual clinical samples obtained after clinical testing from the NHS Grampian Biorepository (15/WM/0430). For lectin-age studies (Figures 2 and 3), blood samples were collected irrespective of age, gender or HbA1c levels. For lectin-hyperglycaemia and oxidation studies ( Figure 4), blood samples were also collected irrespective of age or gender, selected by HbA1c levels in the normal (< 40 mmol/mol) and raised (>70 mmol/mol) ranges were collected at a 1:1 ratio.

Blood cell separation
Erythrocytes were separated from white blood cells in donor blood using Lymphoprep. Brie y, whole blood was diluted 1:1, layered over Lymphoprep and centrifuged at 400g for 35 minutes. The buffy coat was removed, while erythrocytes were isolated for analysis.

Neuraminidase treatment of erythrocytes
Puri ed erythrocytes were washed and treated with neuraminidase (Vibrio cholerae, Sigma) according to manufacturer's instructions for 30 minutes at 37°C in DMEM. Cells were washed with cold (4°C) PBS thoroughly to remove contaminating enzyme or free sialic acids.

Periodic acid Schiff assay
Neuraminidase treated erythrocytes and control erythrocytes, N-acetylneuraminic acid (NANA) and fetuin, were subjected to a PAS assay adapted from Mehdi and colleagues (12). Samples were oxidized using 0.04 M periodic acid (4 o C, 30 minutes, 1:5 ratio of acid to sample). Two volumes of resorcinol working solution (5 ml of 6.0% resorcinol solution, 0.125 ml of 0.1 M copper sulphate solution and 19.875 ml of distilled water, brought to a nal volume of 50 ml with 10 M HCl) was added, heated at 98 o C for 5 minutes and cooled in an ice bath for 2 minutes. 1.8 volumes of n-butanol was added, mixed vigorously and the mixture incubated for 3 minutes at 37 o C. Absorbance at 625nm was measured against a reagent blank.
For layering the fractions by gravity, the highest desired density of Percoll/SAH is added to an empty 15ml Falcon™ tube and 1 ml of erythrocyte/SAH mixture gently layered on top. Consecutive layers of lower density Percoll is applied directly on top of the erythrocyte/SAH mixture. No centrifugation is required. Fractions of erythrocytes rise to the interphases of various Percoll layers applied. Fractionated erythrocytes are extracted from each layer into fresh tubes and washed in PBS well before analysis.

Erythrocyte fragility assay
Washed erythrocyte samples were tested for fragility based on protocol provided in by Blasi and colleagues (46). Erythrocyte C3 deposition assay A sensitive, quantitative indirect enzyme-linked antiglobulin test (IELAT) was employed to test for C3 (Abcam) deposition on erythrocytes (47).

Reactive Oxygen Species (ROS) Production
Rate of ROS production was determined by loading puri ed oxidized or untreated RBC with oxidation sensitive dye CM-H2DCFDA (10 µM; C6827, Molecular Probes) in PBS and incubating for 60 minutes in the dark at 37 o C. RBC were then washed three times, resuspended in DMEM (Thermo Fisher) and uorescence determined immediately by spectro uorimeter (Fluostar Optima; BMG Labtech), with excitation of 485 nm and emission 530 nm. The rate of ROS formation was calculated for the linear portion of the uorescence/time curve generated over six hours, which typically lasted for three hours.

Glycomic analysis
Structural analysis was performed as described previously (48). Brie y, ghost membranes were subjected to sonication in the presence of CHAPS detergent, reduced in 4 M guanidine-HCl (Pierce), carboxymethylated, and digested with trypsin. The digested glycoproteins were then puri ed by Oasis Plus HLB Sep-Pak (186000132, Waters Corp.). N-glycans were released by peptide N-glycosidase F (E. C. 3.5.1.52; Roche Applied Science) digestion, while O-glycans were released with reductive elimination (55 mg of KBH 4 /mL in 0.1 M KOH). Released N-and O-glycans were permethylated using the sodium hydroxide procedure and puri ed by C 18 -Sep-Pak.
Matrix-assisted laser desorption ionization-time of ight MS (MALDI-TOF MS) and MALDI-TOF/TOF MS/MS were performed for the analysis of permethylated glycans. MS data were acquired using either a 4800 MALDI-TOF/TOF (Applied Biosystems) mass spectrometer. MS/MS data were acquired using a 4800 MALDI-TOF/TOF mass spectrometer. Permethylated samples were dissolved in 10 μl of methanol, and 1 μl of dissolved sample was premixed with 1 μl of matrix (10 mg/ml 3,4-diaminobenzophenone in 75% (v/v) aqueous MeCN), spotted onto a target plate, and dried under vacuum. For the MS/MS studies, the collision energy was set to 1 kV, and argon was used as collision gas. The 4700 Calibration standard kit, Calmix (Applied Biosystems), was used as the external calibrant for the MS mode, and [Glu1] brinopeptide B human (Sigma) was used as an external calibrant for the MS/MS mode.
The MS and MS/MS data were processed using Data Explorer 4.9 Software (Applied Biosystems). The processed spectra were subjected to manual assignment and annotation with the aid of a glycobioinformatics tool, GlycoWorkBench (49). The proposed assignments for the selected peaks were based on 12 C isotopic composition together with knowledge of the biosynthetic pathways. The proposed structures were then con rmed by data obtained from MS/MS experiments.