Role for endothelial N-glycan mannose residues in monocyte recruitment during atherogenesis

Abstract

Objective

Up-regulated expression of endothelial adhesion molecules and subsequent binding to cognate monocytic receptors is an established paradigm in atherosclerosis. However, these proteins are the scaffolds with their post-translational modification with sugars providing the actual ligands. We showed recently that TNFα increased hypoglycosylated (mannose rich) N-glycans on the endothelial surface. In this study our aim was to determine if i) hypoglycosylated N-glycans are upregulated by pro-atherogenic stimuli (oscillatory flow) in vitro and in vivo, and ii) whether mannose residues on hypoglycosylated endothelial N-glycans mediate monocyte rolling and adhesion.

Methods and result

Staining with the mannose specific lectins ConA and LCA was increased in human aortic endothelial cells exposed to oscillatory shear or TNFα, and at sites of plaque development and progression in both mice and human vessels. Increasing surface N-linked mannose by inhibiting N-glycan processing, potentiated monocyte adhesion under flow during TNFα stimulation. Conversely, enzymatic removal of high-mannose N-glycans, or masking mannose residues with lectins, significantly decreased monocyte adhesion under flow. These effects occurred without altering induced expression of adhesion molecule proteins.

Conclusion

Hypoglycosylated (high mannose) N-glycans are present on the endothelial cell surface at sites of early human lesion development and are novel effectors of monocyte adhesion during atherogenesis.

Atherosclerosis is a progressive inflammatory disease of the artery walls that leads to the formation of vascular plaques which can occlude blood flow or rupture leading to stroke or myocardial infarction^{1, 2}. Early progression of the disease is characterized by infiltration of leukocytes, particularly monocytes, into the inflamed tissue³. To exit the circulation, monocytes follow a cascade of tethering, rolling, adhesion, and finally transmigration through endothelial cells lining the vessel wall ⁴. A common paradigm for interactions between endothelial cells and monocytes suggests that pro-inflammatory signals induce expression of adhesion molecules on the endothelial cell surface which directly interact with monocytes receptors leading to their rolling and adhesion^4–6. Interestingly, nearly all plasma membrane and secreted proteins are N-glycosylated. N-glycosylation is the co-/post-translational addition of oligosaccharides (glycans) onto the amide group of asparagine residues in an N-X-S/T motif ⁷. Protein N-glycosylation occurs via a multistep and sequential process, resulting in high-mannose, hybrid, and then complex N-glycans. The latter are the most diverse and branched family of N-glycans and it is thought that under basal conditions, most surface glycoproteins exist in this fully processed form, with hypoglycosylated N-glycans (high mannose/hybrid N-glycans) remaining at low levels⁸.

N-glycans are located at the interface between the endothelium and the circulation, positioning them for possible interactions with monocytes. In this manner, the protein adhesion molecules act as scaffolds for the N-glycans which serves as the actual ligands for monocytes, a concept that has been discussed previously^{9, 10}. In this paradigm, adhesion molecules alone are not sufficient to mediate adhesion, but require the appropriate complement of carbohydrate moieties to function, forming a molecular “zip code” for immune cell recognition¹¹. Indeed, a role for correct glycosylation of proteins as ligands of immune cells is well accepted. For example, proper N-glycosylation is required for CD44 to be recognized as a selectin ligand¹². At this time, little is known in regards to the role of endothelial mannose residues in inflammation and immune cell recruitment. It should be noted however that immune cells express various mannose specific receptors to mediate substrate recognition and signaling and some of these receptors have been implicated in atherosclerosis^{13, 14}. Recently, Matthijsen et al¹⁵ showed that macrophage expressed mannose binding lectin is enriched in early atherosclerotic lesions and DC-SIGN, which participates in cell adhesion via binding of high-mannose structures, has been found in early lesions also^{16, 17.}

Recently, we demonstrated that TNFα treatment increases hypoglycosylated, high-mannose and hybrid, N-glycans on the endothelial cell surface¹⁸. In vivo, atherosclerotic plaque progression is triggered by a variety of pro-inflammatory signals including TNFα, but a persistent pathogenic trigger of early endothelial cell dysfunction is the complex blood flow profiles at branch points, curvatures, and bifurcations in arterial vessels¹⁹. Herein we expand on our previous findings and demonstrate that disturbed flow increases hypoglycosylated N-glycans and further that these glycans are also enriched in early human atherosclerotic plaques. Tissue culture models demonstrate that inhibiting N-glycan maturation increases monocyte adhesion, but only under conditions of flow. Finally, enzymatic removal of high-mannose N-glycans, or masking mannose residues with targeted lectins attenuates monocytes adhesion to TNFα stimulated cells. Collectively, these data identify endothelial mannose as a novel ligand for monocyte adhesion during atherogenesis.

Materials and Methods

Detailed materials and methods are found in the online supplementary material. Detection of high mannose/hybrid N-glycans was determined by staining stimulated Human Aortic Endothelial Cells (HAECs) with the FITC-conjugated mannose specific lectins concanavalin A (ConA) and lens culinaris agglutinin (LCA)²⁰. Specificity of these lectins for mannose on tissue sections or cultured cells was demonstrated by loss of staining in the presence of α-methylmannose (500mM) as described¹⁸ and shown in supplemental figure 1. In vivo detection of mannose residues was determined by ConA and LCA staining of murine and human vessel sections at various stages of lesion progression. All animal protocols were approved by the Louisiana State University Health Sciences Center-Shreveport institutional animal care and use committee according to National Institute of Health guidelines. Experiments with human tissue were considered non-human due to the use of exclusively post-mortem samples. Adhesion molecule expression in HAECs was monitored by Western blot analysis, surface ELISA, and immunocytochemistry. Rolling and adhesion of THP-1 monocytes and primary human monocytes was determined using the Glycotech flow chamber as described¹⁸.

Results

Increase in hypoglycosylated N-glycans in response to atheroprone flow and at sites of plaque development in vivo

Our previous work demonstrated that TNFα stimulation increased hypoglycosylated N-glycans on the surface of endothelial cells¹⁸. While signaling through TNFα, and other pro-inflammatory cytokines, induces endothelial dysfunction, a persistent pathogenic trigger in vivo for atherogenesis is the complex shear profiles in regions of disturbed flow¹⁹. To determine if atheroprone shear also increases hypoglycosylated N-glycans, endothelial cells were kept static or exposed to either atheroprotective laminar shear stress (LSS) or atheroprone oscillatory shear stress (OSS) for 18 hours. Oscillatory shear stress induced increased levels of hypoglycosylated N-glycans as indicated by increased binding of the mannose specific lectins ConA and LCA compared to either static or LSS conditions (Figure 1A–B and supplementary figure 2)).

Figure 1. Increase in ConA and LCA staining by oscillatory flow and in regions of early plaque development.

HAECs were left static or exposed to laminar (LSS) or oscillatory shear (OSS) for 18 h before being labeled with FITC-conjugated LCA and ConA. Representative images are shown in (A) and quantitation in (B). n=4; *p<0.05 compared to static control and ^# p<0.05 compared to LSS by 1-way ANOVA with Tukey post – test. (C) ApoE −/− mice were fed a standard chow diet for 16 weeks. The innonimate artery was removed stained with ConA or anti-PECAM-1. Shown are represent images from serial sections from the same vessel. Photos from areas of the vessel where early stages of a lesion were visible are indicated by the arrow. EC surface ConA reactivity was only observed on lesions. Red-rhodamine conjugated ConA green- autofluorescence of basement membrane, blue-nuclei, n=4 mice.

In vivo atherosclerotic lesion progression occurs in vessels subjected to disturbed flow such as the innominate artery, and the portion of the right common carotid prior to the brachiocephalic branchpoint. To determine if atheroprone regions of the vasculature express increased hypoglycosylated N-glycans, we examined the innominate artery from 16 week old ApoE −/− mice which had been maintained on a chow diet. At this age, early atherosclerotic plaque formation in the innominate artery has not developed past the early fatty streak stage of plaque progression. As seen in Figure 1C, healthy (no lesion) regions of the vasculature did not show luminal staining with rhodamine ConA. In contrast, early lesions expressed strong reactivity towards ConA demonstrating that hypoglycosylated N-glycans are present at sites of monocyte adhesion in vivo. Figure 1C also shows staining of PECAM-1 on adjacent sections illustrating that ConA reactive epitopes are present on the endothelium. This data demonstrates that hypoglycosylated N-glycans are positioned in the appropriate regions of the vascular, and expressed at the appropriate time of lesion development to act as mediators of monocyte adhesion.

We next determined if human atherosclerotic lesions contain hypoglycosylated N-glycans. A total of 24 vessels from 18 patients ranging in Stary score from 1–5^{21, 22} from the previously described collection were analyzed²³ (described in Supplemental Table I and II). Under this scoring system, stage 1 corresponds to the first appearance of monocyte-derived macrophages in the intima, stage 2 corresponds to fatty streak formation and significant accumulation of lipid-filled foam cells, and stage 3 refers to an early fatty streak lesion with extracellular lipid pools. Stage 4 plaques and above are recognized as advanced lesions containing a core of extracellular lipid and fibrotic changes and representing clinically symptomatic disease. When stained with Rhodamine ConA and LCA, luminal regions of the vasculature showed very low reactivity in plaque-free and stage 1 lesions (Figure 2A–D). Reactivity towards ConA and LCA dramatically increased in stage 2 and 3 lesions and importantly was localized to the luminal surface where they would encounter monocytes (Figure 2E–H). Interestingly, staining became variable in advanced stage 4 and 5 lesions with 5 of 9 sections examined showing strong luminal reactivity with ConA and LCA. Staining specificity was demonstrated by the ability to block ConA and LCA by including 500mM α-methylmannoside during staining (Supplemental Figure 1). These results show that hypoglycosylated N-glycans are present at the luminal surface during early atherosclerotic progression.

Figure 2. Increased hypoglycosylated N-glycans at site of early atherogensis in human vessels.

Serial sections of human coronary and carotid artery scored by Stary classification were stained with Rhodamine ConA and LCA. Red-lectin, green-autofluorescence of basement membrane, blue-nuclei. Images are representative of vessels stained: no plaque n=1, Stage 1 n=2, Stage 2 n=7, Stage 3 n=6, Stage4/5, n=9.

Alpha-mannosidase inhibitors increase surface mannose content and potentiate monocyte rolling and adhesion

With our findings that hypoglycosylated N-glycans are present during early lesion formation, we next investigated if these structures could provide monocyte binding epitopes. Several studies suggest that hypoglycosylated glycoproteins are defective in trafficking to the cell surface. To determine if adhesion molecules generated in the presence of N-glycan inhibitors could reach the cell surface, specific steps in the N-glycan processing were inhibited and surface expression and distribution of candidate adhesion molecules ICAM-1 and VCAM-1 were analyzed. Inhibitors used were tunicamycin (tunic) which blocks transfer of N-glycans by inhibiting oligosaccharyltransferase resulting in non-glycosylated proteins, kifunensine (kif), an alpha-mannosidase class 1 enzyme inhibitor, resulting in only high-mannose N-glycans being produced, and swainsonine (swain), an alpha-mannosidase class 2 inhibitor, resulting in high-mannose and minimally processed hybrid N-glycans. HAECs were pretreated with inhibitors before stimulation with TNFα. As seen in Figure 3A, each enzyme effectively inhibited the processing of glycoproteins as indicated by the molecular weight shifts in ICAM-1 and VCAM-1 by Western blot analysis. Importantly, complete inhibition of each enzyme was achieved as indicated by the appearance of only one band (different glycoforms would run at different molecular weights (MWt) due to the diversity of carbohydrate content). Tunicamycin decreased ICAM-1 and VCAM-1 MWt to ~52 and 76KDa (corresponding to predicted MWt from primary sequences) indicating that N-glycans constitute ~45% and 15% of the TNFα stimulated proteins respectively. Interestingly, kif and swain had similar effects on the decrease in MWt of VCAM-1, but resulted in the production of distinct ICAM-1 glycoforms, with a larger glycoforms being seen for swain, consistent with the prediction that kif would result in smaller high-mannose N-glycans and swain would allow for processing to the hybrid stage. Next, in order to determine if hypoglycosylated adhesion molecules reach the apical membrane, surface ELISAs were performed on cells treated with inhibitors, TNFα, or a combination of the two (Figure 3C–D). Neither kif nor swain altered surface expression levels of ICAM-1 or VCAM-1 but tunic blocked their transport to the cell surface. To test if kif and swain altered surface distribution of adhesion molecules, staining of ICAM-1 and VCAM-1 in unpermeabilized TNFα stimulated HAECs was performed. Surface staining patterns were similar between normal and hypoglycosylated ICAM-1 and VCAM-1 (Figure 3B). Finally, to confirm that the inhibitors were altering the monosaccharide content on the EC surface, cells exposed to inhibitors, or in combination with TNFα were stained for the presence of surface ConA and LCA reactive carbohydrates. As seen in Figure 3E, kif and swain both increased surface levels of ConA and, and in combination with TNFα further increased ConA levels above TNFα treatment alone. Similar results were observed with LCA (Fig 3F), except Kif did not enhance TNFα dependent effects. As expected, cells pretreated with tunicamycin did not show an increase in ConA or LCA with TNFα stimulation and also completely inhibited ICAM-1, VCAM-1 trafficking to the plasma membrane (Fig 3C–D). These data demonstrate that alpha-mannosidase inhibitors kif and swain effectively increase surface mannose content, restricting N-glycan processing to the complex types, without compromising the ability of adhesion molecules to reach the EC cell surface.

Figure 3. Alpha mannosidase inhibitors increase surface mannose residues without altering adhesion molecule expression.

HAECs were untreated or treated with 10 ng/ml TNFα for 4 hrs. Additionally, some cells were pretreated with tunicamycin (1 μM), kifusenenine (40 ng/ml), or swainsosnine (1 μM) for 2 hrs. A) Western blot analysis of whole cell lysates for ICAM-1, VCAM-1, and β-actin. B) Surface staining of ICAM-1 and VCAM-1 in control of TNFα stimulated cells with and without alpha-mannosidase inhibitors. C and D) Surface ELISA for ICAM-1 and VCAM-1 respectively. E) and F) Surface staining of cells with FITC-conjugated ConA or LCA respectively. All experiments were performed 4–6 times and statistical significance was calculated by one-way ANOVA with Tukey’s Post test *p<0.001 relative to control, ** p<0.001 relative to TNFα, ^# p<0.05 relative to TNFα.

Endothelial hypoglycosylated N-glycans regulate monocyte adhesion under flow

We next wanted to determine if hypoglycosylation of adhesion molecules on the EC surface would affect monocyte adhesion. Endothelial cells were left untreated, TNFα treated, or treated with alpha-mannosidase inhibitors alone or in combination with TNFα. Under static conditions, TNFα increased THP-1 and primary human monocyte adhesion, and this was not altered by pretreatment with kif or swain (Figure 4A–B). However, in the presence of flow (0.5 dyne/cm²) kif and swain potentiated the number of both THP-1 and primary human monocytes adhering to TNFα stimulated endothelial cells (Figure 4C–D). Neither kif nor swain alone affected THP-1 or primary human monocyte adhesion relative to control. These data demonstrate that adhesion molecule hypoglycosylation increases TNFα-dependent monocyte adhesion. Supplementary Figure 3 shows that kif and swain alone increase the number of rolling monocytes compared to control, but not to the degree of TNFα. This suggests that rolling of monocytes can be mediated at least in part, by hypoglycoyslation of constitutive proteins whose presence on the endothelial surface is not absolutely dependent on TNFα activation. However, this rolling is not transmitted to increased firm adhesion as indicated in Fig 4D.

Figure 4. N-glycan synthesis inhibitors increase monocyte adhesion under flow.

HAECs were untreated or treated with 10 ng/ml TNFα for 4 hrs. Additionally, some cells were pretreated with kif (40 ng/ml) or swain (1 μM) for 2 hrs. Monocyte adhesion was determined using either THP-1 (panels A and C) or primary human monocytes (Panels B and D) under static (panel A and B) or flow (0.5 dyn/cm²) (panels C and D) conditions. Each experiment was performed 3–6 times. Statistical significance was calculated by one-way ANOVA with Tukeys post test. *p<0.001 relative to control, ** p<0.01 relative to control, ^# p<0.05 relative to TNFα.

Removal or blocking of mannose residues attenuates monocyte adhesion

To confirm a role for EC hypoglycosylated, high mannose N-glycans on EC-monocyte adhesion, cells were treated with TNFα and fixed before being exposed to the high-mannose N-glycan specific endoglycosidase H (Endo H). Figure 5A shows that Endo H treatment of cells resulted in ~70% decrease in ConA staining. No change in staining levels of the complex N-glycan lectin, SNA were observed (Fig 5A). Additionally, Endo H digestion did not change expression of ICAM-1 nor E-selectin on the cell surface (Fig 5A), thereby leaving the protein scaffolds for N-glycans unchanged. We next examined the effects of Endo H digestion on monocyte adhesion. Removing high-mannose N-glycans from the surface of endothelial cells did not alter monocyte adhesion under static conditions (Figure 5B) but resulted in an approximately 60% reduction in the number of adherent monocytes under flow (Figure 5C).

Figure 5. Removal of high-mannose N-glycans attenuated monocytes adhesion under flow.

HAECs were untreated or treated with TNFα (10 ng/ml, 4 hrs), fixed with 4% PFA, and digested with Endo H at 37°C. citrate buffer alone was included as vehicle control for EndoH A) Surface expression levels of mannose, sialic acid, ICAM-1 and E-selectin were determined by ConA, SNA staining and ELISA assays respectively. Data are expressed as fold relative to TNFα (n=3), *p<0.001 relative to citrate buffer control by 1-way ANOVA with Tukey post –test. No significant changes in SNA binding nor ICAM-1 or E-selectin expression was seen with EndoH treatment. Panel B and C respectively show THP-1 monocyte adhesion determined under static and flow (0.5 dynes/cm²) conditions. *p<0.05 or ^# p <0.01 compared to TNFα or TNFα + citrate groups by one-way ANOVA, (n=3–5) with Tukey’s post test.

To further test if endothelial mannose residues were participating in monocyte adhesion, a lectin blocking experiment was performed using three mannose specific lectins. HAECs were treated with TNFα and then exposed to either ConA, LCA or the α-1,3 mannose specific lectin galanthus navalis lectin (GNL) for the last 10 minutes of TNFα incubation. Cell were then washed and monocyte adhesion measured. This approach would allow for masking of mannose residues on the cell surface and block any potential interactions with hypoglycosylated N-glycan recognition receptors on monocytes. As seen in Figure 6A, blocking surface mannose residues with lectins did not change monocyte adhesion under static conditions. However, as seen in Figure 6B, in the presence of flow both ConA and GNL significantly reduced, while LCA modestly reduced, monocyte adhesion. The ability of lectins to inhibit monocyte adhesion was dose-dependent (Supplemental Figure 4). Collectively these data show that expression of the adhesion molecule protein scaffold is necessary, whereas their modification with N-glycan epitopes is conditional, but required, for monocyte adhesion under flow.

Figure 6. Blocking surface mannose attenuates monocyte adhesion under flow.

HAECs were untreated or treated with TNFα (10 ng/ml, 4 hrs). During the final 10 min of stimulation some cells were incubated with 5 μg of the mannose specific lectins LCA, ConA, or GNL. Adhesion of THP-1 monocytes was determined under static (A), and flow (0.5 dynes/cm2) (B) conditions. *p<0.01 or ^#p<0.05 compared to TNFα by one-way ANOVA with Tukey’s post test, n=4–6.

Discussion

The work presented herein represents a new paradigm for monocyte-EC interactions where hypoglycosylated high-mannose and hybrid N-glycans on adhesion molecules act as ligands for monocyte rolling and adhesion. Atheroprone shear and sites of early atherosclerotic lesions in murine and human tissues show elevated levels of mannose specific lectin staining (Figure 1 and 2). Induced hypoglycosylation does not alter TNFα-induced adhesion molecule trafficking or distribution (Figure 3), but does exacerbate monocyte adhesion (Figure 4), which can be attenuated by removing high-mannose N-glycans or masking mannose residues (Figure 5 and 6). Collectively, these data identify endothelial mannose as a novel ligand and regulator of monocyte adhesion under flow.

Protein N-glycosylation is a multi-step pathway whereby oligosaccharides are enzymatically added to asparagines residues in N-X-S/T motifs. A sequential process of carbohydrate trimming and addition gives rise to the three subtypes of N-glycans; high-mannose (5–9 mannose), hybrid (3–5 mannose), and complex (3 mannose). It is thought that most N-glycans are processed to the complex stage which helps maintain their surface localization by interaction with the galectin lattice⁸. However, our previous data¹⁸ and the current findings suggest that pro-atherogenic stimuli increase hypoglycosylated N-glycan content on the EC surface. These findings are supported by work showing that pro-inflammatory stimuli modifies glycan profiles on synoviocytes ²⁴ and endothelial cells^{25, 26}. Cancer, which has a recognized pro-inflammatory component, is associated with an increase in high-mannose N-glycans²⁷ and loss of N-glycan complexity of E-cadherin is associated with cancer progression²⁸. Also, hypoglycosylated N-glycans on various cell types have been implicated in the adhesion of neutrophils^29–31, demonstrating that they can be maintained on the surface of cells. Collectively, these data suggest that N-glycan processing is regulated during inflammatory stress.

Our current data extend these concepts to include EC-monocyte interactions in atherosclerosis development. We demonstrate that atheroprone oscillatory flow, which is an established stimulus to increase adhesion molecule expression^32–34, also increases expression of hypoglycosylated N-glycans on the surface of ECs. Of critical importance, early sites of plaque development in vivo, which represent focal regions of inflammation and monocyte adhesion, were highly decorated by hypoglycosylated N-glycans. The exact mechanism underlying hypoglycosylation of these N-glycans is unclear. One possible mechanism involves TNFα dependent down-regulation of alpha-mannosidase activity (which trims mannose residues from high-mannose structures thereby providing substrate for formation of complex N-glycans)¹⁸. This is also suggested by recent findings from array studies showing upregulated MAN1A1 in EC exposed to LSS compared to OSS ³⁵, which supports the concept that anti-inflammatory LSS promotes complex N-glycan production. A decrease in alpha-mannosidase activity has also been reported in rat aortas during drug-induced diabetes, which can be returned to normal levels with insulin administration³⁶, suggesting that inflammation is the underlying cause of the decrease in enzyme activity. Diabetes represents a significant risk factor for endothelial dysfunction and atherosclerosis³⁷ and dysregulation of N-glycan processing as an underlying modulator of disease demands future investigations.

Our data show that pro-inflammatory stimuli increase mannose residues on the endothelial cell surface via enrichment of hypoglycosylated N-glycans. Immune cells express a variety of receptors which can bind to mannose, including some which prefer hypoglycosylated N-glycans. If immune cells can target hypoglycosylated N-glycans then a role of these N-glycans in autoimmune diseases would be expected. Indeed, multiple sclerosis, systemic lupus erythematosus, and congenital dyserthryopoetic anemia^38–43 are all associated with hypoglycosylated N-glycans. In this context, dysfunction in N-glycan processing has been postulated as a danger signal which can initiate immune responses⁴⁴ as many primitive organisms, such as viruses, fungi, and bacteria, express high levels of mannose-rich glycans on their surfaces. The appearance of mannose-rich N-glycans on endogenous proteins could be viewed as non-self and illicit an immune response²⁴. We note that increased ConA or LCA staining was not restricted to the endothelial monolayer in mouse or human atherosclerotic lesions, and although less intense than endothelial staining, these observations suggest hat hypoglycosylated epitopes may also be regulated in other cell types that comprise lesions. Further studies are required to determine precisely the location and cell type dependence for changes in hypoglycosylated epitopes in atherosclerotic lesions

The counter ligand on monocytes remains unclear but numerous candidate mannose recognizing proteins expressed on monocytes exist and include MBL, mannose receptor (MR) and DC-SIGN^15,45,46. MBL haplotype is associated with atherosclerotic risk levels^{21, 43} and MBL has been found in developing, but not advanced, murine and human lesions¹⁵. Apart from binding mannose residues, MBL also functions in complement activation⁴⁷. Numerous members of the complement cascade and receptors are found in early atherosclerotic lesions⁴⁸ suggesting a role for complement in atherosclerotic development. Another mannose recognizing protein found in atherosclerotic plaques is MR⁴⁵. MR receptor is expressed on a variety of monocyte derived cell populations including alternatively activated M2 macrophages and dendritic cells⁴⁹. There is also evidence that MR is present on monocytes as ricin toxin induced apoptosis of THP-1 monocytes can be attenuated with anti-MR antibodies⁵⁰. Lesion localized MR expressing macrophages do not associate with lipids and do not transition into foam cells⁵¹, suggesting they are serving a non-traditional function within the plaques, possibly recruited to the region by the mannose residues on the luminal surface. Another family of proteins associated with atherosclerosis development are the c-type lectins known as the selectins (CD62e, CD62l, and CD62l). Selectins have established roles in monocyte rolling and adhesion and are known to bind complex N-glycans and O-glycans via sialyl Lewis x (sLe^x) motifs⁶. However, evidence exist that they can interact with mannose⁵², thus a role for the selectins as receptors of endothelial mannose cannot be ruled out.

Another protein expressed on monocytes which is widely implicated in atherosclerosis and that can bind mannose is the integrin heterodimer Mac1 (CD11b/CD18). Mac1 has been shown to bind type-I fimbriated E. coli in a mannose dependent manner⁵³ and CD11b+ monocytes shows high binding affinity to mannosylated glycopolymers⁵⁴. Soluble CD11b demonstrates elevated binding to ICAM-1 produced in the presence of alpha-mannosidase 1 inhibitors⁵⁵ suggesting that increased mannose on ICAM-1 promotes binding. Additional work has shown that anti-CD18 antibodies inhibit neutrophil binding to endothelial cells cultured with alpha-mannosidase inhibitors³⁰. Collectively this work establishes Mac1 as a potential mannose receptor on the surface of monocytes that could recognize hypoglycosylated N-glcyans.

While our data now show that endothelial hypoglycosylated N-glycans are involved in mediating monocyte adhesion, the precise glycoproteins carrying these carbohydrates are currently unknown. Previous studies provide some evidence as to which proteins may be involved. As mentioned, when produced in the presence of alpha-mannosidase type I inhibitors⁵⁵, ICAM-1 binds more effectively to Mac1. Interestingly, the two N-glycan sites flanking the Mac1 binding domain of ICAM-1 have the highest degree of N-glycan complexity of all ICAM-1 N-glycosylation sites⁵⁶, so addition of hypoglycosylated N-glycans at these sites could potentially increase binding. Another major regulator of monocyte adhesion is VCAM-1. Interestingly, different glycoforms of VCAM1 are expressed in response to tumor cell conditioned media⁵⁷ suggesting that VCAM-1 might serve as a scaffold for pro-adhesive hypoglycosylated N-glycans. Supporting this concept, sialic acid, a sugar component of complex N-glycans, inhibits VCAM-1 dependent adhesion during flow⁵⁸. Finally, we note that effects of hypoglycosylated epitopes on increased monocyte adhesion were only revealed in the presence of flow suggesting that adhesion molecule N-glycosylation differentially effects rolling vs. firm adhesion. A traditional view is that rolling interactions are mediating by leukocyte selectins and endothelial PSGL-1, with ICAM-1 or VCAM-1 playing principal roles in firm adhesion. However, several studies utilizing adhesion molecule deficient mice or endothelial cells derived from these have shown that ICAM-1 and VCAM-1 can also regulate monocyte rolling and firm adhesion especially at lower shear rates ^59–67 which combined with data in Fig 3A, suggest that ICAM-1 and VCAM-1 may be important targets for regulation by hypoglycoyslation.

The fact that different stimuli induce different glycoforms of a protein is supportive of our previous work that demonstrated PPARγ controlled N-glycosylation independent of adhesion molecule expression¹⁸. It is known that different glycoforms of a protein perform different functions in cells. PSGL-1 is known to bind P-selectin through an O-glycan linked α-2,3 sialic acid (that is, a glycan linked to a serine or threonine residue). Yet, despite this requirement for this binding, only a small fraction of PSGL-1 contains sialic acid residues, implying that the majority of PSGL-1 does not serve as a selectin ligand⁹. Similarly, specific functions of CD44⁶⁸, GABAA⁶⁹, and IgG⁷⁰ are determined by their glycosylation status with different glycoforms of the proteins responsible for different functions. Thus an interesting possibility arises where only a small pool of any given protein scaffold, or structurally similar proteins, contains the hypoglycosylated N-glycans that are regulating monocyte adhesion. Conversely, the addition of a specific glycan to a protein does not make it a ligand for a particular receptor. As mentioned, the selectins are known to bind to the sLe^x motif. However, the majority of proteins that contain sLe^x motifs are not ligands for selectin proteins indicating that the saccharide structure itself is not sufficient to act as a ligand⁹. Thus, a combination of the proper protein scaffold and specific carbohydrate(s) is critical for receptor recognition. This concept is highlighted in the current work as kif and swain increase mannose content (Figure 3E–F) but do not induce protein adhesion without further stimulation (Figure 4C–D) indicating that only mannose residues on certain proteins can behave as ligands for monocyte adhesion. Additionally, the idea that different adhesion molecules undergo distinct N-glycosylation is demonstrated in Figure 3A. ICAM-1 undergoes distinct processing producing specific N-glycoforms in the presence of kif and swain. In contrast, VCAM-1 produces the same N-glycoforms when produced in the presence of kif and swain. These differences highlight how two major regulators of monocyte adhesion undergo distinct post-translation processing.

In conclusion, evidence is provided that hypoglycosylated N-glycans are upregulated by pro-inflammatory oscillatory shear stress and are enriched in the vessel lumen during early atherosclerotic plaque development. Blocking mannose residues associated with these hypoglycosylated N-glycans with lectins attenuates, while increasing their expression with pharmacological inhibitors of N-glycan maturation, potentiates monocyte adhesion thereby establishing endothelial mannose as a novel regulator of monocyte adhesion under flow.