**4. Discussion**

The atherosclerotic process in which we are interested is a combination of several factors, among which high cholesterol levels, driven by lipoprotein LDL, vasculature inflammation, and oxidative processes are the main components, deeply studied but still not easily correlated [1,3,43]. In particular, it is not yet clear whether the vessel inflammatory condition can influence the LDL passage to the subendothelial space and alter the activation of blood components through the interaction with the glycocalyx. In the in vitro model of inflammation and atherosclerosis using endothelial cells [44] and smooth muscle cells [3], we highlighted the role of the GAG HA in recruiting monocytes/macrophages from blood, promoting their adhesion to the endothelial layer, and altering the ECM composition and thickness of the intima layer.

We determine to use the pro-inflammatory cytokine TNF-α, since it is well known that is produced in the visceral fat of obese and overweight patients that are prone to CVD. Moreover, this chronic low-grade vascular inflammation is hypothesized to be the stimulus by which, within vasculature, reactive oxygen species (ROS) are generated through NAD(P)H oxidase activation and other sources, which in turn reduces NO availability, causing the local endothelial dysfunctions [2]. Therefore, TNF-α can contribute to vascular changes, favouring the development and acceleration of the atherothrombotic process in the clinical condition [2]. In this view, we used the pro-inflammatory cytokine TNF-α treatment on HUVEC cells as a model of vasculature inflammation, investigating the glycocalyx modifications and their effect of the endothelial barrier.

The effects of TNF-α on the NO production shown in Figure 1 confirm the effectiveness of the treatment, and even if the data are only the expression of the synthetic enzymes, this evidence is related to the NO levels [45]. The action of TNF-α on the HA synthesis is reported in Figure 2 and confirms the data previously found with a different cytokine, the IL-1β. Briefly, the synthesis of HA is increased by means of the overexpression of the membrane synthases HAS2, and the GAG is mainly localized around the cells as a pericellular coat that increases the distance between the cells. Among the synthetic HA enzymes, HAS2 is known to be the main producer of the polymer [3]. The HAS3 enzyme decrease in this phase is not yet well understood and deserves other investigations in order to unravel the exact role of the enzyme in the membrane. During the early stage of atherosclerotic lesion formation in Apolipoprotein E (Apoe)-deficient mice, the HAS3 expression is increased and controlled in vascular smooth muscle cells by the cytokine IL-1β [46], and even if data on endothelial cells are unavailable, there are clear indications that HAS3 might be a promising therapeutic target in atherosclerosis.

It is noteworthy that this pericellular coat can increase the adhesion and recruitment of the circulating monocytes [44] and protect cells from apoptotic events [47]. Since HA can also exert its effect on cell proliferation and migration depending on its dimensions [48–50] and organized sovramolecular architecture by soluble factors (e.g., TSG-6) [51], the mere synthesis of the polymer is not predictive of its effect. In our model, in fact, the accumulation of HA in the pericellular coat does not alter cell proliferation and migration as reported in Supplementary Figures S1 and S2, which was in line with the barrier role of the endothelial cells, but nevertheless can favour the adhesion and transmigration of blood cells to the sub-endothelium.

Inflammation is the cause of multiple effects on the endothelium, including changing the glycocalyx composition [10,19,52]. The heparan sulfate proteoglycan belonging to the Syndecans family is largely involved in the onset of different CVD in vasculature, and in particular is reported to be increased in patients with resistant hypertension [21], while

syndecan-3/-4 ectodomain fragments, produced by several stimuli, including heparinase or thrombin, decrease endothelial cell–cell adhesive barrier integrity [23] and are involved in the cell-extracellular matrix and cell–cell adhesion mechanisms [22]. HSPGs in endothelial cells include glypicans, located on the cell membrane, anchored by glycosylphosphatidylinositol (GPI), functioning mainly as modulators of growth factor signaling [53], but they seem more critically involved in developmental morphogenesis and positively correlate with the onset of certain types of cancers [54] and less with vascular inflammation.

The TNF-α treatment on HUVEC cells selectively increases the expression of Syndecan-4, and the expression remains high after the incubation for up to 48 h (Figure 3); the increase of syndecan-4 is also evident at the protein level, and is measurable with Western blot, even if the results are not significant. This finding can be either due to the low level of expression of the PG in the cells that do not consent to the change to be significant in the assay sensibility range, or to the turnover rate of Syndecan-4, which is formed by the protein core and the GAG moiety. In our data, in fact, we investigated the composition of the heparan sulfate chains of the membrane-bound PGs as well as those released in the medium and evidenced a difference in the percentage of various disaccharides. The different sulfation of the disaccharides can lead to a modification in the sulfation pattern of the GAG chains, and this can be the real important event for the Syndecan-4 related effects, while the protein core can be maintained at the same amount but continuously replaced.

The total amount of uronic containing GAGs remains invariable (Supplementary Figure S4), indicating a balance between the synthesis of the polymer chains that use the same UDP-sugar precursors as we underlined in a previous paper, involving, in particular HA and HS [20], but a control on the number and dimension of the various chains was impossible. As shown in Table 1 the main changes in disaccharides involve the NS and 6 sulfation. Unfortunately, the small size of the samples and high biological variability among them make it impossible to have statistically significant data, but the trend is very sharp. Together with the changes in the protein core of Syndecan-4 and in the NS sulfation, the synthetic enzymes EXT1 and EXT2, responsible for the polymerization, and NDST1 (N-deacetylase/N-sulfotransferase) are also increased in the TNF-α treated samples (Supplementary Figure S3). The sulfation pattern of the HS chains is frequently found altered in the inflammatory condition of tumours, as reported by several papers that indicate specific HS sulfotransferase as critical for the survival and invasiveness of those cells, for example, the 3-O-sulfotransferase [55] or the 2-O-sulfotransferase [56]. Moreover, the 6-O-sulfation of HS highly influences the polysaccharide structural diversity and is critically involved in the binding of many proteins, in particular growth factors [17].

The NDST1 enzyme is the component that catalyses both the N-deacetylation and the N-sulfation of glucosamine (GlcNAc) residue in the heparan sulfate. This enzyme modifies the GlcNAc-GlcA disaccharide-repeating sugar backbone to make N-sulfated heparosan, a prerequisite substrate for later modifications in heparin biosynthesis, such as 6-O-sulfation [15]. Therefore, the increase in the N-sulfation along the chain consequently also increases the sulfation in the C6, as summarized in Table 1. This modification seems more important in the chains bound to the membrane, i.e., carried on the proteoglycan core. For what concerns the CVD, it was reported that the protein PCSK9, an important drug target because of its crucial role in lipid metabolism, can interact with HS in N-sulfation rich domain [57], eventually linking the high cholesterol load with the vasculature inflammation at the onset of the pathology.

The mechanism of endothelial dysfunction involves the alteration of the layer permeability demonstrated in Figure 4, which is dependent upon the overexpression of Syndecan-4, while the silencing of the protein core of the PG does not change the barrier function of the membrane. The increase in Syndecan-4 after treatment with TNF-α can cause a rearrangement in the adhesion asset of the endothelial layer due to its involvement in the dynamics of focal adhesion and the formation of networks at gap junctions. The different integration of the cells with each other and with the basement membrane can be the molecular mechanism leading to lower permeability of the endothelial layer. Nevertheless, this mechanism needs further investigations to be clarified, in particular regarding the interaction with other membrane components.
